import os
import sys

with open(sys.argv[0]) as f:
    code = f.read()  # read the code of this file ASAP, for logging
import copy
import glob
import math
import threading
import time
import uuid
from dataclasses import dataclass
from collections import defaultdict
from itertools import accumulate
from pathlib import Path
import gc

os.environ["PYTORCH_ALLOC_CONF"] = "expandable_segments:True"
import torch

torch.empty(
    1, device=f"cuda:{os.environ['LOCAL_RANK']}", requires_grad=True
).backward()  # prevents a bug on some systems
import torch._dynamo as dynamo
import torch.distributed as dist
import torch.nn.functional as F

# torch._inductor.config.coordinate_descent_tuning = True # we have banned this flag for new records because it causes compilation to take 30min
import triton
import triton.language as tl
from kernels import get_kernel
from torch import Tensor, nn

dynamo.config.recompile_limit = 64


# -----------------------------------------------------------------------------
# Custom operators: FP8 matmul by @YouJiacheng

@torch.library.custom_op("nanogpt::mm", mutates_args=())
def mm_op(x: Tensor, w: Tensor, x_s: float, w_s: float, grad_s: float) -> tuple[Tensor, Tensor, Tensor]:
    @torch.compile
    def impl(x: Tensor, w: Tensor):
        assert x.is_contiguous() and w.is_contiguous()
        x_f8 = x.div(x_s).to(torch.float8_e4m3fn)
        w_f8 = w.div(w_s).to(torch.float8_e4m3fn)
        out = torch._scaled_mm(
            x_f8,
            w_f8.T,
            out_dtype=torch.bfloat16,
            scale_a=x.new_tensor(x_s, dtype=torch.float32),
            scale_b=x.new_tensor(w_s, dtype=torch.float32),
            use_fast_accum=True,
        )
        return out, x_f8, w_f8

    return impl(x, w)

@mm_op.register_fake
def _(x: Tensor, w: Tensor, *_):
    assert x.ndim == w.ndim == 2
    assert x.shape[1] == w.shape[1]
    assert x.device == w.device
    assert x.is_contiguous() and w.is_contiguous()
    return x @ w.T, x.to(torch.float8_e4m3fn), w.to(torch.float8_e4m3fn)

@torch.library.custom_op("nanogpt::mm_backward", mutates_args=())
def mm_backward_op(g: Tensor, x_f8: Tensor, w_f8: Tensor, x_s: float, w_s: float, grad_s: float) -> tuple[Tensor, Tensor]:
    @torch.compile
    def impl(grad: Tensor, x_f8: Tensor, w_f8: Tensor):
        assert grad.is_contiguous()
        x_inv_s = grad.new_tensor(x_s, dtype=torch.float32)
        w_inv_s = grad.new_tensor(w_s, dtype=torch.float32)
        grad_inv_s = grad.new_tensor(grad_s, dtype=torch.float32)
        grad_f8 = grad.div(grad_s).to(torch.float8_e5m2)
        grad_x = torch._scaled_mm(
            grad_f8,
            w_f8.T.contiguous().T,
            out_dtype=torch.bfloat16,
            scale_a=grad_inv_s,
            scale_b=w_inv_s,
            use_fast_accum=False,
        )
        # faster than grad_f8_t @ x_f8, for (d_out, d_in) == (50304, 768)
        grad_w = torch._scaled_mm(
            x_f8.T.contiguous(),
            grad_f8.T.contiguous().T,
            out_dtype=torch.float32,
            scale_a=x_inv_s,
            scale_b=grad_inv_s,
            use_fast_accum=False,
        ).T
        return grad_x, grad_w

    return impl(g, x_f8, w_f8)

@mm_backward_op.register_fake
def _(g: Tensor, x_f8: Tensor, w_f8: Tensor, *_):
    return x_f8.to(torch.bfloat16), w_f8.T.contiguous().T.to(torch.float32)

def backward(ctx, grad_out: Tensor, *_):
    x_f8, w_f8 = ctx.saved_tensors
    x_s, w_s, grad_s = ctx.scales
    grad_x, grad_w = torch.ops.nanogpt.mm_backward(
        grad_out, x_f8, w_f8, x_s, w_s, grad_s
    )
    return grad_x, grad_w, None, None, None

def setup_context(ctx: torch.autograd.function.FunctionCtx, inputs, output):
    *_, x_s, w_s, grad_s = inputs
    _, x_f8, w_f8 = output
    ctx.save_for_backward(x_f8, w_f8)
    ctx.scales = x_s, w_s, grad_s
    ctx.set_materialize_grads(False)

mm_op.register_autograd(backward, setup_context=setup_context)

# -----------------------------------------------------------------------------
# Triton kernel for symmetric matrix multiplication by @byronxu99

def _get_autotune_configs():
    return [
        triton.Config(
            {
                "BLOCK_SIZE_M": bm,
                "BLOCK_SIZE_N": bn,
                "BLOCK_SIZE_K": bk,
                "GROUP_SIZE_M": 8,
                "LOWER_UPPER": 1,
            },
            num_stages=stages,
            num_warps=warps,
        )
        for bm in [64, 128]
        for bn in [64, 128, 256]
        for bk in [64, 128]
        for stages, warps in [(3, 4), (3, 8), (4, 4)]
        if bm // bn <= 2 and bn // bm <= 2
    ]

@triton.jit
def _pid_to_block(
    pid,
    M,
    BLOCK_SIZE_M: tl.constexpr,
    BLOCK_SIZE_N: tl.constexpr,
    GROUP_SIZE_M: tl.constexpr,
):
    # Split output matrix into blocks of size (BLOCK_SIZE_M, BLOCK_SIZE_N)
    num_pid_m = tl.cdiv(M, BLOCK_SIZE_M)
    num_pid_n = tl.cdiv(M, BLOCK_SIZE_N)

    # Map PID to a single matrix in batch
    batch_idx = pid // (num_pid_m * num_pid_n)
    pid = pid % (num_pid_m * num_pid_n)

    # Map PID to 2D grid of blocks
    pid_m = pid // num_pid_n
    pid_n = pid % num_pid_n
    pid_m, pid_n = tl.swizzle2d(pid_m, pid_n, num_pid_m, num_pid_n, GROUP_SIZE_M)

    m_idx = pid_m * BLOCK_SIZE_M
    n_idx = pid_n * BLOCK_SIZE_N
    return batch_idx, m_idx, n_idx

@triton.autotune(
    configs=_get_autotune_configs(),
    key=["M", "K", "a_stride_r", "a_stride_c", "c_stride_r", "c_stride_c"],
)
@triton.jit
def XXT_kernel(
    A_ptr, C_ptr,
    M, K,
    a_stride_b, a_stride_r, a_stride_c,
    c_stride_b, c_stride_r, c_stride_c,
    BLOCK_SIZE_M: tl.constexpr,
    BLOCK_SIZE_N: tl.constexpr,
    BLOCK_SIZE_K: tl.constexpr,
    GROUP_SIZE_M: tl.constexpr,
    LOWER_UPPER: tl.constexpr,
):
    pid = tl.program_id(axis=0)
    batch_idx, m_idx, n_idx = _pid_to_block(
        pid, M, BLOCK_SIZE_M, BLOCK_SIZE_N, GROUP_SIZE_M
    )

    # Skip blocks that don't need to be computed
    skip_block_below_diag = (LOWER_UPPER == 0) and (n_idx + BLOCK_SIZE_N <= m_idx)
    skip_block_above_diag = (LOWER_UPPER != 0) and (m_idx + BLOCK_SIZE_M <= n_idx)
    if skip_block_below_diag or skip_block_above_diag:
        return

    # Index into one matrix of batch
    A_ptr += batch_idx * a_stride_b
    C_ptr += batch_idx * c_stride_b

    # Create pointer arrays for A and A.T
    offs_m = (m_idx + tl.arange(0, BLOCK_SIZE_M)) % M
    offs_n = (n_idx + tl.arange(0, BLOCK_SIZE_N)) % M
    offs_k = tl.arange(0, BLOCK_SIZE_K)
    a_ptrs = A_ptr + (offs_m[:, None] * a_stride_r + offs_k[None, :] * a_stride_c)
    at_ptrs = A_ptr + (offs_k[:, None] * a_stride_c + offs_n[None, :] * a_stride_r)

    accumulator = tl.zeros((BLOCK_SIZE_M, BLOCK_SIZE_N), dtype=tl.float32)

    # Accumulate over blocks of K
    for k in tl.range(0, tl.cdiv(K, BLOCK_SIZE_K)):
        a = tl.load(a_ptrs, mask=offs_k[None, :] < K - k * BLOCK_SIZE_K, other=0.0)
        at = tl.load(at_ptrs, mask=offs_k[:, None] < K - k * BLOCK_SIZE_K, other=0.0)
        accumulator = tl.dot(a, at, accumulator)
        a_ptrs += BLOCK_SIZE_K * a_stride_c
        at_ptrs += BLOCK_SIZE_K * a_stride_c

    out_dtype = C_ptr.dtype.element_ty
    output = accumulator.to(out_dtype)

    # Store block of C
    offs_cm = m_idx + tl.arange(0, BLOCK_SIZE_M)
    offs_cn = n_idx + tl.arange(0, BLOCK_SIZE_N)
    c_ptrs = C_ptr + (offs_cm[:, None] * c_stride_r + offs_cn[None, :] * c_stride_c)
    c_mask = (offs_cm[:, None] < M) & (offs_cn[None, :] < M)
    tl.store(c_ptrs, output, mask=c_mask)

    # Store block of C mirrored across the diagonal
    c_ptrs_t = C_ptr + (offs_cn[:, None] * c_stride_r + offs_cm[None, :] * c_stride_c)
    c_mask_t = (offs_cn[:, None] < M) & (offs_cm[None, :] < M)
    tl.store(c_ptrs_t, output.T, mask=c_mask_t)

def XXT(A: torch.Tensor, out: torch.Tensor):
    """
    Launch Triton kernel to compute C = A @ A.T
    """
    assert A.ndim == 2 or A.ndim == 3
    M, K = A.shape[-2:]
    assert out.size(-2) == M, "Output matrix has incorrect shape"
    assert out.size(-1) == M, "Output matrix has incorrect shape"

    batch_size = A.size(0) if A.ndim == 3 else 1
    input_batch_stride = A.stride(0) if A.ndim == 3 else 0
    output_batch_stride = out.stride(0) if out.ndim == 3 else 0

    grid = lambda meta: (
        batch_size * triton.cdiv(M, meta["BLOCK_SIZE_M"]) * triton.cdiv(M, meta["BLOCK_SIZE_N"]),
    )
    XXT_kernel[grid](
        A_ptr=A,
        C_ptr=out,
        M=M,
        K=K,
        a_stride_b=input_batch_stride,
        a_stride_r=A.stride(-2),
        a_stride_c=A.stride(-1),
        c_stride_b=output_batch_stride,
        c_stride_r=out.stride(-2),
        c_stride_c=out.stride(-1),
    )
    return out

@triton.autotune(
    configs=_get_autotune_configs(),
    key=["M", "a_stride_r", "a_stride_c", "c_stride_r", "c_stride_c"],
)
@triton.jit
def ba_plus_cAA_kernel(
    A_ptr, C_ptr,
    M,
    a_stride_b, a_stride_r, a_stride_c,
    c_stride_b, c_stride_r, c_stride_c,
    alpha, beta,
    BLOCK_SIZE_M: tl.constexpr,
    BLOCK_SIZE_N: tl.constexpr,
    BLOCK_SIZE_K: tl.constexpr,
    GROUP_SIZE_M: tl.constexpr,
    LOWER_UPPER: tl.constexpr,
):
    # This is mostly duplicated from XXT_kernel, but also loads and adds a block of A
    # Performance is slightly slower than XXT_kernel, so we use two separate kernels
    pid = tl.program_id(axis=0)
    batch_idx, m_idx, n_idx = _pid_to_block(
        pid, M, BLOCK_SIZE_M, BLOCK_SIZE_N, GROUP_SIZE_M
    )

    # Skip blocks that don't need to be computed
    skip_block_below_diag = (LOWER_UPPER == 0) and (n_idx + BLOCK_SIZE_N <= m_idx)
    skip_block_above_diag = (LOWER_UPPER != 0) and (m_idx + BLOCK_SIZE_M <= n_idx)
    if skip_block_below_diag or skip_block_above_diag:
        return

    # Index into one matrix of batch
    A_ptr += batch_idx * a_stride_b
    C_ptr += batch_idx * c_stride_b

    # Create pointer arrays for A and A.T
    offs_m = (m_idx + tl.arange(0, BLOCK_SIZE_M)) % M
    offs_n = (n_idx + tl.arange(0, BLOCK_SIZE_N)) % M
    offs_k = tl.arange(0, BLOCK_SIZE_K)
    a_ptrs = A_ptr + (offs_m[:, None] * a_stride_r + offs_k[None, :] * a_stride_c)
    at_ptrs = A_ptr + (offs_k[:, None] * a_stride_c + offs_n[None, :] * a_stride_r)

    accumulator = tl.zeros((BLOCK_SIZE_M, BLOCK_SIZE_N), dtype=tl.float32)

    # Accumulate over blocks of K
    for k in tl.range(0, tl.cdiv(M, BLOCK_SIZE_K)):
        a = tl.load(a_ptrs, mask=offs_k[None, :] < M - k * BLOCK_SIZE_K, other=0.0)
        at = tl.load(at_ptrs, mask=offs_k[:, None] < M - k * BLOCK_SIZE_K, other=0.0)
        accumulator = tl.dot(a, at, accumulator)
        a_ptrs += BLOCK_SIZE_K * a_stride_c
        at_ptrs += BLOCK_SIZE_K * a_stride_c

    # Load block of A to add (corresponds to the current block of C)
    offs_am = m_idx + tl.arange(0, BLOCK_SIZE_M)
    offs_an = n_idx + tl.arange(0, BLOCK_SIZE_N)
    a_add_ptrs = A_ptr + (offs_am[:, None] * a_stride_r + offs_an[None, :] * a_stride_c)
    a_add_mask = (offs_am[:, None] < M) & (offs_an[None, :] < M)
    a_add = tl.load(a_add_ptrs, mask=a_add_mask, other=0.0).to(tl.float32)

    # Apply alpha and beta
    accumulator *= alpha
    accumulator += a_add * beta

    out_dtype = C_ptr.dtype.element_ty
    output = accumulator.to(out_dtype)

    # Store block of C
    offs_cm = m_idx + tl.arange(0, BLOCK_SIZE_M)
    offs_cn = n_idx + tl.arange(0, BLOCK_SIZE_N)
    c_ptrs = C_ptr + (offs_cm[:, None] * c_stride_r + offs_cn[None, :] * c_stride_c)
    c_mask = (offs_cm[:, None] < M) & (offs_cn[None, :] < M)
    tl.store(c_ptrs, output, mask=c_mask)

    # Store block of C mirrored across the diagonal
    c_ptrs_t = C_ptr + (offs_cn[:, None] * c_stride_r + offs_cm[None, :] * c_stride_c)
    c_mask_t = (offs_cn[:, None] < M) & (offs_cm[None, :] < M)
    tl.store(c_ptrs_t, output.T, mask=c_mask_t)

def ba_plus_cAA(A: torch.Tensor, alpha: float, beta: float, out: torch.Tensor):
    """
    Launch Triton kernel to compute C = alpha * A @ A.T + beta * A
    """
    assert A.ndim == 2 or A.ndim == 3
    M, K = A.shape[-2:]
    assert M == K, "Input matrix must be square"
    assert out.size(-2) == M
    assert out.size(-1) == M

    batch_size = A.size(0) if A.ndim == 3 else 1
    input_batch_stride = A.stride(0) if A.ndim == 3 else 0
    output_batch_stride = out.stride(0) if out.ndim == 3 else 0

    grid = lambda meta: (
        batch_size * triton.cdiv(M, meta["BLOCK_SIZE_M"]) * triton.cdiv(M, meta["BLOCK_SIZE_N"]),
    )
    ba_plus_cAA_kernel[grid](
        A_ptr=A,
        C_ptr=out,
        M=M,
        a_stride_b=input_batch_stride,
        a_stride_r=A.stride(-2),
        a_stride_c=A.stride(-1),
        c_stride_b=output_batch_stride,
        c_stride_r=out.stride(-2),
        c_stride_c=out.stride(-1),
        alpha=alpha,
        beta=beta,
    )
    return out

# Computed for num_iters=5, safety_factor=2e-2, cushion=2
polar_express_coeffs = [
    (8.156554524902461, -22.48329292557795, 15.878769915207462),
    (4.042929935166739, -2.808917465908714, 0.5000178451051316),
    (3.8916678022926607, -2.772484153217685, 0.5060648178503393),
    (3.285753657755655, -2.3681294933425376, 0.46449024233003106),
    (2.3465413258596377, -1.7097828382687081, 0.42323551169305323)
]

@torch.compile(dynamic=False, fullgraph=True) # Must use dynamic=False or else it's much slower
def polar_express(G: torch.Tensor, split_baddbmm: bool = False):
    """
    Polar Express Sign Method: https://arxiv.org/pdf/2505.16932
    by Noah Amsel, David Persson, Christopher Musco, Robert M. Gower.
    """
    X = G.bfloat16()
    if G.size(-2) > G.size(-1):
        X = X.mT

    # Ensure spectral norm is at most 1
    X = X / (X.norm(dim=(-2, -1), keepdim=True) * (1 + 2e-2) + 1e-6)

    # Allocate buffers
    X = X.contiguous()
    A = torch.empty((*X.shape[:-1], X.size(-2)), device=X.device, dtype=X.dtype)
    B = torch.empty_like(A)
    C = torch.empty_like(X)

    # Select batched vs unbatched
    if split_baddbmm:
        BX_matmul = torch.bmm if X.ndim > 2 else torch.mm
    else:
        aX_plus_BX = torch.baddbmm if X.ndim > 2 else torch.addmm

    # Perform the iterations
    for a, b, c in polar_express_coeffs:
        XXT(X, out=A)  # A = X @ X.mT
        ba_plus_cAA(A, alpha=c, beta=b, out=B)  # B = b * A + c * A @ A

        # Referencing X twice causes pytorch to make a defensive copy,
        # resulting in a cudaMemcpyAsync in baddbmm.
        # For large matrices (i.e., the mlp weights), it's faster to split
        # the operation into two kernels to avoid this.
        if split_baddbmm:
            BX_matmul(B, X, out=C)  # C = B @ X
            C.add_(X, alpha=a)      # C = C + a*X  (in-place, X only read)
        else:
            aX_plus_BX(X, B, X, beta=a, out=C)  # C = a * X + B @ X

        X, C = C, X  # Swap references to avoid unnecessary copies

    if G.size(-2) > G.size(-1):
        X = X.mT
    return X


# -----------------------------------------------------------------------------
# Compiled helpers for NorMuon by @chrisjmccormick

@torch.compile(dynamic=False, fullgraph=True)
def cautious_wd_and_update_inplace(p, v, wd_tensor, lr_tensor):
    """Cautious weight decay + parameter update. wd_tensor and lr_tensor are 0-D CPU tensors."""
    mask = (v * p) >= 0
    wd_factor = wd_tensor.to(p.dtype)
    lr_factor = lr_tensor.to(p.dtype)
    p.copy_(p - (p * mask * wd_factor * lr_factor) - (v * lr_factor))


@torch.compile(dynamic=False, fullgraph=True)
def apply_normuon_variance_reduction(v_chunk, second_momentum_buffer, beta2, red_dim):
    """NorMuon variance reduction. Algebraically fuses the normalization steps to minimize memory ops."""
    v_mean = v_chunk.float().square().mean(dim=red_dim, keepdim=True)
    red_dim_size = v_chunk.size(red_dim)
    v_norm_sq = v_mean.sum(dim=(-2, -1), keepdim=True).mul_(red_dim_size)
    v_norm = v_norm_sq.sqrt_()
    second_momentum_buffer.lerp_(v_mean.to(dtype=second_momentum_buffer.dtype), 1 - beta2)
    step_size = second_momentum_buffer.clamp_min(1e-10).rsqrt_()
    scaled_sq_sum = (v_mean * red_dim_size) * step_size.float().square()
    v_norm_new = scaled_sq_sum.sum(dim=(-2, -1), keepdim=True).sqrt_()
    final_scale = step_size * (v_norm / v_norm_new.clamp_min_(1e-10))
    return v_chunk.mul_(final_scale.type_as(v_chunk))


# -----------------------------------------------------------------------------
# NorMuon optimizer

class NorMuon(torch.optim.Optimizer):
    """
    Muon - MomentUm Orthogonalized by Newton-schulz

    https://kellerjordan.github.io/posts/muon/

    Muon internally runs standard SGD-momentum, and then performs an orthogonalization post-
    processing step, in which each 2D parameter's update is replaced with the nearest orthogonal
    matrix. To efficiently orthogonalize each update, we use a Newton-Schulz iteration, which has
    the advantage that it can be stably run in bfloat16 on the GPU.

    Warning: This optimizer should not be used for the embedding layer, the final fully connected layer,
    or any {0,1}-D parameters; those should all be optimized by a standard method (e.g., AdamW).

    Differences from standard Muon:
    - Newton-Shulz is replaced with Polar Express for the orthogonalization step
    - NorMuon adds a low-rank variance estimator similar to Adafactor. https://arxiv.org/pdf/2510.05491
    - small 1D parameters handled here instead of in Adam
    - Cautious weight decay, a gated version of decoupled weight decay
    - Custom distributed sizing:
    The model stores all attn and mlp weights in the same shape, and then updates the view as
    needed on the forward pass. This enables attn and mlp weights to be contained within the same
    dist.reduce_scatter_tensor() call. The model architecture has been customized to enable
    (n_attn_layers+n_mlp_layers*2)%8==0 for batching across 8 GPUs with zero padding on mlp and attn.
    The scheduling is:
        1. reduce scatter attn_gate (10 params 6 padding params)
        2. reduce scatter attn/mlp round 1 (10 attn params 6 mlp params)
        3. reduce scatter attn/mlp round 2 (16 mlp params)
        4. wait on step 1, then compute update of 1 and schedule all gather
        5. wait on step 2, then compute update of 2 and schedule all gather
        6. wait on step 3, then compute update of 3 and schedule all gather
            GPUs receive [2 ATTN, 2 ATTN, 2 ATTN, 2 ATTN, 2 ATTN, 2 MLP, 2 MLP, 2 MLP]
            GPUs that receive params of type attn reshape before computing update
        7. wait on 4, then compute update of 4 and schedule all gather
        8. wait for each all gather to complete and update params
    Empirically, leading with small params provides an additional 0.2s improvement.
    """
    def __init__(self, params, lr=0.02, weight_decay=0.01, momentum=0.95, beta2=0.95, custom_sizing=True):
        defaults = dict(lr=lr, weight_decay=weight_decay, momentum=momentum, beta2=beta2)
        self.world_size = dist.get_world_size() if dist.is_initialized() else 1
        # custom sizing requires 8 GPUs
        if custom_sizing and dist.get_world_size()==8:
            param_groups = self.generate_custom_param_groups(params)
        else:
            param_groups = self.generate_standard_param_groups(params)
        super().__init__(param_groups, defaults)

    def reset(self):
        # expose a reset for clearing buffers
        for group in self.param_groups:
            if "momentum_buffer" in group:
                group["momentum_buffer"].zero_()
                group["second_momentum_buffer"].zero_()

    def generate_standard_param_groups(self, params):
        """
        Use this method if running on less than 8 GPU or experimenting with additional attn or mlp modules.
        Creates one param group per module.
        """
        groups = defaultdict(list)
        for param in params:
            groups[param.label].append(param)

        param_groups = []
        for module_name, group_params in groups.items():
            chunk_size = (len(group_params) + self.world_size - 1) // self.world_size
            param_groups.append(dict(params=group_params, chunk_size=chunk_size))

        return param_groups

    def generate_custom_param_groups(self, params):
        """
        Implementation requires that a single GPU does not receive both attn
        and mlp params when a param group is split across GPUs.
        """
        params_list = list(params)
        module_group_order = ['attn_gate', 'value_embed_gate', 'attn', 'mlp'] # 16, 10, 16, 32
        group_sizes = [16, 10, 16, 16, 16]
        # 72... its actually 74
        params_list.sort(key=lambda x: module_group_order.index(x.label))
        print0(len(params_list), console=True)
        idx = 0
        assert len(params_list) == sum(group_sizes)
        param_groups = []
        for size in group_sizes:
            chunk_size = (size + self.world_size - 1) // self.world_size
            group_params = params_list[idx: idx + size]
            param_groups.append(dict(params=group_params, chunk_size=chunk_size))
            idx += size

        return param_groups

    @torch.no_grad()
    def step(self):
        # Efficient distributed step by @YouJiacheng, @KonstantinWilleke, @alexrgilbert,
        # @adricarda, @tuttyfrutyee, @vdlad, @ryanyang0, @vagrawal, @varunneal, @chrisjmccormick
        rank = dist.get_rank()
        group_infos = []
        for group in self.param_groups:
            params: list[Tensor] = group["params"]
            if not params:
                continue

            chunk_size = group["chunk_size"]
            padded_num_params = chunk_size * self.world_size

            stacked_grads = torch.empty(
                (padded_num_params, *params[0].shape),
                dtype=params[0].dtype,
                device=params[0].device
            )
            for i, p in enumerate(params):
                stacked_grads[i].copy_(p.grad, non_blocking=True)
            if len(params) < padded_num_params:
                stacked_grads[len(params):].zero_()

            grad_chunk = torch.empty_like(stacked_grads[:chunk_size])

            reduce_future = dist.reduce_scatter_tensor(
                grad_chunk, stacked_grads, op=dist.ReduceOp.AVG, async_op=True
            ).get_future()

            group_infos.append(dict(grad_chunk=grad_chunk, reduce_future=reduce_future))

        all_gather_infos = []
        # Second pass: wait for gradients, compute updates for the local shard of parameters,
        # and launch all async all_gather operations.
        for group, info in zip(self.param_groups, group_infos):
            info["reduce_future"].wait()

            params = group["params"]
            grad_chunk = info["grad_chunk"]
            chunk_size = group["chunk_size"]
            padded_num_params = chunk_size * self.world_size

            start_idx = rank * chunk_size
            module_idx = start_idx if start_idx < len(params) else 0

            num_params = min(chunk_size, max(0, len(params) - start_idx))  # num params for this rank

            if "momentum_buffer" not in group:
                group["momentum_buffer"]  = torch.zeros_like(grad_chunk[:num_params])
            momentum_buffer = group["momentum_buffer"]
            # Apply momentum update to the persistent momentum buffer in-place
            momentum_buffer.lerp_(grad_chunk[:num_params], 1 - group["momentum"])
            updated_grads = grad_chunk[:num_params].lerp_(momentum_buffer, group["momentum"])

            grad_shape = updated_grads.shape
            if params[module_idx].label == 'attn':

                for p in params[module_idx:module_idx + num_params]:
                    assert p.label == 'attn'

                updated_grads = updated_grads.view(4 * grad_shape[0], grad_shape[1] // 4, grad_shape[2])

            ref_param = params[module_idx]
            param_shape = ref_param.shape

            # The below shape-based heuristic assumes that matrices have their input along the
            # row dimension and their output along the columns. Gates are an exception.
            is_gate = 'gate' in ref_param.label

            if "second_momentum_buffer" not in group:
                if is_gate:
                    group["second_momentum_buffer"] = torch.zeros_like(updated_grads[..., :, :1])
                else:
                    group["second_momentum_buffer"] = (torch.zeros_like(updated_grads[..., :, :1])
                        if param_shape[-2] >= param_shape[-1] else torch.zeros_like(updated_grads[..., :1, :])
                    )
            second_momentum_buffer = group["second_momentum_buffer"]

            if "param_lr_cpu" not in group:
                # Define multipliers for ALL params in this group (global, not per-shard)
                lr_mults = []
                wd_mults = []
                for p in params:
                    # Increase learning rate for modules with larger inputs than outputs.
                    # This shape check also assumes rows=input, columns=output, so take care
                    # when changing memory layouts. @chrisjmccormick
                    shape = p.shape
                    if len(shape) >= 2:
                        shape_mult = max(1.0, shape[-2] / shape[-1]) ** 0.5
                    else:
                        shape_mult = 1.0
                    lr_mults.append(shape_mult * getattr(p, "lr_mul", 1.0))
                    wd_mults.append(getattr(p, "wd_mul", 1.0))
                # Define as cpu tensors to enable Inductor constant folding
                group["param_lr_cpu"] = torch.tensor(lr_mults, dtype=torch.float32, device="cpu")
                group["param_wd_cpu"] = torch.tensor(wd_mults, dtype=torch.float32, device="cpu")

            eff_lr_all = group["param_lr_cpu"] * group["lr"]
            eff_wd_all = group["param_wd_cpu"] * group["weight_decay"] * group["lr"]

            # Slice the portion corresponding to this rank's shard
            eff_lr_cpu = eff_lr_all[module_idx:module_idx + num_params]
            eff_wd_cpu = eff_wd_all[module_idx:module_idx + num_params]

            # Compute zeropower for the entire chunk in a single, batched call.
            if num_params == 0:
                v_chunk = updated_grads
            else:
                v_chunk = polar_express(updated_grads, split_baddbmm=(ref_param.label == 'mlp'))

            # Note that the head orientation in O is transposed relative to QKV, so red_dim
            # is 'incorrect' for O. However, correcting this showed no improvement. @chrisjmccormick
            red_dim = -1 if (is_gate or param_shape[-2] >= param_shape[-1]) else -2

            v_chunk = apply_normuon_variance_reduction(
                v_chunk, second_momentum_buffer, group["beta2"], red_dim
            )

            v_chunk = v_chunk.view(grad_shape)

            # # "Cautious" weight decay (https://arxiv.org/abs/2510.12402)
            updated_params = torch.empty_like(grad_chunk)
            if num_params > 0:
                # Work on a stacked copy to avoid touching original params
                param_chunk = torch.stack(params[module_idx:module_idx + num_params])

                for local_idx in range(num_params):
                    cautious_wd_and_update_inplace(
                        param_chunk[local_idx],
                        v_chunk[local_idx],
                        eff_wd_cpu[local_idx],
                        eff_lr_cpu[local_idx],
                    )
            else:
                param_chunk = torch.zeros_like(v_chunk)

            updated_params[:num_params].copy_(param_chunk)
            if num_params < chunk_size:
                updated_params[num_params:].zero_()

            stacked_params = torch.empty(
                (padded_num_params, *param_shape),
                dtype=updated_params.dtype,
                device=updated_params.device,
            )

            gather_future = dist.all_gather_into_tensor(
                stacked_params, updated_params, async_op=True
            ).get_future()

            all_gather_infos.append(
                {
                    "gather_future": gather_future,
                    "stacked_params": stacked_params,
                    "orig_params": params,
                }
            )

        # Final pass: wait for all_gather to complete and copy results back into original parameter tensors.
        for info in all_gather_infos:
            info["gather_future"].wait()
            stacked_params = info["stacked_params"]
            orig_params = info["orig_params"]

            unstacked_params = torch.unbind(stacked_params)
            for i, p in enumerate(orig_params):
                p.copy_(unstacked_params[i], non_blocking=True)


class DistAdam(torch.optim.Optimizer):
    def __init__(self, params, label_order: list[str],lr: float = 1e-3, betas: tuple[float, float] = (0.9, 0.999), eps: float = 1e-8, weight_decay: float = 0.01):
        self.world_size = dist.get_world_size() if dist.is_initialized() else 1
        defaults = dict(lr=lr, betas=betas, eps=eps, weight_decay=weight_decay)
        params = list(params)
        # Group by label, with explicit ordering for execution control.
        params_by_label = defaultdict(list)
        for p in params:
            params_by_label[getattr(p, 'label', None)].append(p)
        param_groups = []
        for label in label_order:
            if label in params_by_label:
                param_groups.append(dict(params=params_by_label[label]))
        # include any unlabeled params at the end (processed last)
        if None in params_by_label:
            param_groups.append(dict(params=params_by_label[None]))
        super().__init__(param_groups, defaults)
        # init state: small params (numel < 1024) use full-sized state, others use sharded
        for p in params:
            chunk = p if p.numel() < 1024 else p[:p.size(0) // self.world_size]
            exp_avg = torch.zeros_like(chunk, dtype=torch.bfloat16, device=p.device)
            self.state[p] = dict(step=0, exp_avg=exp_avg, exp_avg_sq=torch.zeros_like(exp_avg))
        # DistributedAdam implementation by @vagrawal, @akash5474
        self.should_sync = False
        self._reduce_scatter_hooks = []
        self._reduce_scatter_futures = {}
        self.register_backward_hooks()

    def register_backward_hooks(self):
        for group in self.param_groups:
            for param in group["params"]:
                self._reduce_scatter_hooks.append(param.register_post_accumulate_grad_hook(self._sync_gradient))

    @torch.no_grad()
    def _sync_gradient(self, param):
        if not self.should_sync:
            return

        grad = param.grad
        if param.numel() < 1024:
            # Small params: use all_reduce (no scatter/gather needed)
            self._reduce_scatter_futures[param] = (
                dist.all_reduce(grad, op=dist.ReduceOp.AVG, async_op=True).get_future(),
                grad
            )
        else:
            rank_size = grad.shape[0] // self.world_size
            if grad is not None:
                grad_slice = torch.empty_like(grad[:rank_size])
                self._reduce_scatter_futures[param] = (
                    dist.reduce_scatter_tensor(grad_slice, grad, op=dist.ReduceOp.AVG, async_op=True).get_future(),
                    grad_slice
                )

    def copy_lm_to_embed(self):
        # run at 2/3 of training
        lm_head = self.param_groups[0]['params'][0]
        embed = self.param_groups[-1]['params'][0]
        lm_head_state = self.state[lm_head]
        embed_state = self.state[embed]
        embed_state['step'] = lm_head_state['step']
        embed_state['exp_avg'] = lm_head_state['exp_avg'].clone()
        embed_state['exp_avg_sq'] = lm_head_state['exp_avg_sq'].clone()
        embed.data.copy_(lm_head.data)

    @torch.compile
    @torch.no_grad()
    def step(self):
        rank = dist.get_rank()
        all_gather_futures: list[torch.Future] = []

        for group in self.param_groups:
            beta1, beta2 = group['betas']
            eps = group['eps']
            wd = group['weight_decay']
            for param in group['params']:
                if param not in self._reduce_scatter_futures:
                    continue

                fut, g_slice = self._reduce_scatter_futures[param]
                fut.wait()

                is_small = param.numel() < 1024
                if is_small:
                    # Small params: g_slice is actually full grad, p_slice is full param
                    p_slice = param
                else:
                    rank_size = param.shape[0] // self.world_size
                    p_slice = param[rank * rank_size:(rank + 1) * rank_size]

                lr = group['lr'] * getattr(param, "lr_mul", 1.0)
                state = self.state[param]

                exp_avg = state["exp_avg"]
                exp_avg_sq = state["exp_avg_sq"]
                state["step"] += 1
                t = state["step"]
                # update running averages
                exp_avg.mul_(beta1).add_(g_slice, alpha=1 - beta1)
                exp_avg_sq.mul_(beta2).addcmul_(g_slice, g_slice, value=1 - beta2)
                # bias corrections
                bias1 = 1 - beta1 ** t
                bias2 = 1 - beta2 ** t
                # compute step
                denom = exp_avg_sq.sqrt().add_(eps)
                step_size = lr * (bias2 ** 0.5 / bias1)
                update = exp_avg.div(denom).mul_(step_size)
                # cautious weight decay
                mask = (update * p_slice) > 0
                # lr as weight decay schedule
                eff_weight_decay = lr * wd * getattr(param, "wd_mul", 1.0)
                update.addcmul_(p_slice, mask, value=eff_weight_decay * lr)

                p_slice.add_(other=update, alpha=-1.0)

                if not is_small:
                    all_gather_futures.append(dist.all_gather_into_tensor(param, p_slice, async_op=True).get_future())

        self._reduce_scatter_futures.clear()
        torch.futures.collect_all(all_gather_futures).wait()

# -----------------------------------------------------------------------------
# PyTorch nn.Module definitions for the model

def norm(x: Tensor):
    return F.rms_norm(x, (x.size(-1),))

class CastedLinear(nn.Linear):
    def __init__(self, in_features: int, out_features: int, use_fp8=False, x_s=1.0, w_s=1.0, grad_s=1.0):
        super().__init__(in_features, out_features, bias=False)
        self.use_fp8 = False # use_fp8
        self.x_s = x_s
        self.w_s = w_s
        self.grad_s = grad_s

    def reset_parameters(self) -> None:
        with torch.no_grad():
            self.weight.zero_()  # @Grad62304977 and others

    def forward(self, x: Tensor):
        if self.use_fp8 and self.training:
            _x = x.flatten(0, -2)
            out: Tensor = torch.ops.nanogpt.mm(_x, self.weight, x_s=self.x_s, w_s=self.w_s, grad_s=self.grad_s)[0]
            return out.reshape(*x.shape[:-1], -1)
        else:
            return F.linear(x, self.weight.type_as(x))


# -----------------------------------------------------------------------------
# PyTorch nn.Module definitions for the model

# yarn implementation @classiclarryd
class Yarn(nn.Module):
    def __init__(self, head_dim, max_seq_len):
        super().__init__()
        self.head_dim = head_dim
        self.max_seq_len = max_seq_len
        self.reset()

    def reset(self):
        angular_freq = (1 / 1024) ** torch.linspace(0, 1, steps=self.head_dim//4, dtype=torch.float32, device=device)
        # half-truncate RoPE by @YouJiacheng (w/ base freq tuning)
        angular_freq = torch.cat([angular_freq, angular_freq.new_zeros(self.head_dim//4)])
        t = torch.arange(self.max_seq_len, dtype=torch.float32, device=device)
        theta = torch.outer(t, angular_freq)
        self.cos = nn.Buffer(
            theta.cos().to(torch.bfloat16), persistent=False
        )
        self.sin = nn.Buffer(
            theta.sin().to(torch.bfloat16), persistent=False
        )
        self.angular_freq = angular_freq
        # start with 0.1, inspired by 0.12 from @leloykun and learnable scalars used by @brendanh0gan https://x.com/hi_tysam/status/1879693583898591283
        self.attn_scale = 0.1

    def apply(self, old_window: int, new_window: int, alpha: int=1, beta: int=32):
        rotations = args.block_size * old_window * self.angular_freq / (2 * torch.pi)
        scaling_factor = old_window / new_window
        interpolation_weight = torch.clamp((rotations - alpha) / (beta - alpha), 0, 1)
        self.angular_freq *= scaling_factor + interpolation_weight * (1 - scaling_factor)
        t = torch.arange(self.max_seq_len, dtype=torch.float32, device=self.angular_freq.device)
        theta = torch.outer(t, self.angular_freq)
        self.cos.copy_(theta.cos())
        self.sin.copy_(theta.sin())
        self.attn_scale *= 0.2 * math.log(new_window / old_window) + 1

def rotary(x_BTHD: Tensor, cos: Tensor, sin: Tensor):
    assert cos.size(0) >= x_BTHD.size(-3)
    cos, sin = (
        cos[None, : x_BTHD.size(-3), None, :],
        sin[None, : x_BTHD.size(-3), None, :],
    )
    x1, x2 = x_BTHD.chunk(2, dim=-1)
    y1 = x1 * cos + x2 * sin
    y2 = x1 * (-sin) + x2 * cos
    return torch.cat((y1, y2), 3)

@dataclass
class AttnArgs:
    ve: torch.Tensor
    sa_lambdas: torch.Tensor
    seqlens: torch.Tensor
    bm_size: int
    cos: torch.Tensor
    sin: torch.Tensor
    attn_scale: float
    key_offset: bool

flash_attn_interface = get_kernel('varunneal/flash-attention-3').flash_attn_interface

class CausalSelfAttention(nn.Module):
    def __init__(self, dim: int, head_dim: int, num_heads: int, layer_idx: int):
        super().__init__()
        self.num_heads = num_heads
        self.head_dim = head_dim
        self.dim = dim
        self.hdim = num_heads * head_dim

        assert self.hdim == self.dim, "num_heads * head_dim must equal model_dim"
        std = self.dim ** -0.5
        bound = (3 ** 0.5) * std # improved init scale by @YouJiacheng
        # merged QKVO weights: suggested by many, implemented by @fernbear.bsky.social, and further improved by @YouJiacheng
        # https://x.com/hi_tysam/status/1879699187107033311
        # Simplified layout by @chrisjmccormick
        self.qkvo_w = nn.Parameter(torch.empty(self.dim * 4, self.hdim))
        # label all modules for explicit optimizer grouping
        self.qkvo_w.label = 'attn'

        with torch.no_grad():
            self.qkvo_w[:self.dim * 3].uniform_(-bound, bound)  # init QKV weights
            self.qkvo_w[self.dim * 3:].zero_()  # init O weights to zero

        # sparse gated attention to enable context based no-op by @classiclarryd
        self.attn_gate = CastedLinear(16, num_heads)
        self.attn_gate.weight.label = 'attn_gate'
        self.attn_gate.weight.lr_mul = 0.1

        # only include gates on layers with value embeds used on forward pass
        if layer_idx in [0,1,2,3,4,11, 12, 13, 14, 15]:
            self.value_embed_gate = CastedLinear(16, num_heads)
            self.value_embed_gate.weight.label = 'value_embed_gate'
            self.value_embed_gate.weight.lr_mul = 0.1

    def forward(self, x: Tensor, attn_args: AttnArgs):
        B, T = x.size(0), x.size(1) # batch size, sequence length
        assert B == 1, "varlen sequences requires B == 1"
        assert T % 16 == 0
        # unpack attention args
        cos, sin = attn_args.cos, attn_args.sin
        ve, sa_lambdas, key_offset = attn_args.ve, attn_args.sa_lambdas, attn_args.key_offset
        seqlens, attn_scale, bm_size = attn_args.seqlens, attn_args.attn_scale, attn_args.bm_size

        q, k, v = F.linear(x, sa_lambdas[0] * self.qkvo_w[:self.dim * 3].type_as(x)).view(B, T, 3 * self.num_heads, self.head_dim).chunk(3, dim=-2)
        q, k = norm(q), norm(k) # QK norm @Grad62304977
        q, k = rotary(q, cos, sin), rotary(k, cos, sin)
        if key_offset:
            # shift keys forward for the stationary head dims. Enables 1-layer induction.
            k[:, 1:, :, self.head_dim // 4:self.head_dim // 2] = k[:, :-1, :, self.head_dim // 4:self.head_dim // 2]
            k[:, 1:, :, 3 * self.head_dim // 4:] = k[:, :-1, :, 3 * self.head_dim // 4:]
        if ve is not None:
            ve_gate_out = 2 * torch.sigmoid(self.value_embed_gate(x[..., :self.value_embed_gate.weight.size(-1)])).view(B, T, self.num_heads, 1)
            v = v + ve_gate_out * ve.view_as(v) # @ KoszarskyB & @Grad62304977

        max_len = args.train_max_seq_len if self.training else (args.val_batch_size // (grad_accum_steps * world_size))

        # use flash_attn over flex_attn @varunneal. flash_attn_varlen suggested by @YouJiacheng
        y = flash_attn_interface.flash_attn_varlen_func(q[0], k[0], v[0], cu_seqlens_q=seqlens, cu_seqlens_k=seqlens,
                                                        max_seqlen_q=max_len, max_seqlen_k=max_len,
                                                        causal=True, softmax_scale=attn_scale, window_size=(bm_size, 0))
        y = y.view(B, T, self.num_heads, self.head_dim)
        y = y * torch.sigmoid(self.attn_gate(x[..., :self.attn_gate.weight.size(-1)])).view(B, T, self.num_heads, 1)
        y = y.contiguous().view(B, T, self.num_heads * self.head_dim) # re-assemble all head outputs side by side
        y = F.linear(y, sa_lambdas[1] * self.qkvo_w[self.dim * 3:].type_as(y))  # sa_lambdas[1] pre-multiplied to O @shenberg
        return y

class MLP(nn.Module):
    def __init__(self, dim: int):
        super().__init__()
        hdim = 4 * dim
        # Transposed layout to match attention weights
        self.c_fc = nn.Parameter(torch.empty(hdim, dim))
        self.c_proj = nn.Parameter(torch.empty(hdim, dim))
        # label all modules for explicit optimizer grouping
        self.c_fc.label = 'mlp'
        self.c_proj.label = 'mlp'
        self.c_proj.lr_mul = 2.

        std = 0.5 * (dim ** -0.5)
        bound = (3 ** 0.5) * std # improved init scale by @YouJiacheng
        with torch.no_grad():
            self.c_fc.uniform_(-bound, bound)
            self.c_proj.zero_() # zero init suggested by @Grad62304977

    def forward(self, x: Tensor):
        x = F.linear(x, self.c_fc.type_as(x))
        x = F.relu(x).square() # https://arxiv.org/abs/2109.08668v2; ~1-2% better than GELU; suggested by @SKYLINEZ007 and @Grad62304977
        x = F.linear(x, self.c_proj.T.type_as(x))
        return x

class Block(nn.Module):
    def __init__(self, dim: int, head_dim: int, num_heads: int, layer_idx: int):
        super().__init__()
        # skip attention of blocks.6 (the 7th layer) by @YouJiacheng
        self.attn = CausalSelfAttention(dim, head_dim, num_heads, layer_idx)
        # skip MLP blocks for first MLP layer by @EmelyanenkoK
        self.mlp = MLP(dim)

    def forward(self, x: Tensor, attn_args: AttnArgs):
        if self.attn is not None:
            x = x + self.attn(norm(x), attn_args)
        if self.mlp is not None:
            x = x + self.mlp(norm(x))
        return x

# -----------------------------------------------------------------------------
# The main model

def next_multiple_of_n(v: float | int, *, n: int):
    return next(x for x in range(n, int(v) + 1 + n, n) if x >= v)

@dataclass
class ForwardScheduleConfig:
    mtp_weights: torch.Tensor
    ws_short: int
    ws_long: int

class GPT(nn.Module):
    def __init__(self, vocab_size: int, num_layers: int, num_heads: int, head_dim: int, model_dim: int, max_seq_len: int):
        super().__init__()
        self.num_layers = num_layers
        vocab_size = next_multiple_of_n(vocab_size, n=128)

        self.smear_gate = CastedLinear(16, 1)
        self.smear_gate.weight.label = 'smear_gate'
        self.smear_gate.weight.lr_mul = 0.01
        self.smear_gate.weight.wd_mul = 0.0

        self.skip_gates = nn.ModuleList([CastedLinear(16, 1) for _ in range(3)])
        for sg in self.skip_gates:
            sg.weight.label = 'skip_gate'
            sg.weight.lr_mul = 0.01
            sg.weight.wd_mul = 0.0

        # token value embeddings by @KoszarskyB - inspired by @Grad62304977's value residual implementation following https://arxiv.org/abs/2410.17897
        # value embedding code simplification inspired by @ragulpr https://github.com/KellerJordan/modded-nanogpt/pull/78
        self.value_embeds = nn.ModuleList([nn.Embedding(vocab_size, model_dim) for _ in range(5)])
        for embed in self.value_embeds:
            nn.init.zeros_(embed.weight)
        for ve in self.value_embeds:
            ve.weight.label = 'value_embed'
        self.blocks = nn.ModuleList([Block(model_dim, head_dim, num_heads, i) for i in range(num_layers)])
        self.yarn = Yarn(head_dim, max_seq_len)
        # there are only 50257 unique GPT-2 tokens; we extend to nearest multiple of 128 for efficiency.
        # suggested to me by @Grad62304977. this originates from Karpathy's experiments.
        use_fp8 = not os.environ.get("DISABLE_FP8", False)

        self.lm_head = CastedLinear(model_dim, vocab_size, use_fp8=use_fp8, x_s=100/448, w_s=1.6/448, grad_s=0.75/448)
        nn.init.normal_(self.lm_head.weight, mean=0, std=0.005)
        self.lm_head.weight.label = 'lm_head'

        self.embed = nn.Embedding(vocab_size, model_dim)
        self.embed.weight.label = 'embed'

        self.embed2 = nn.Embedding(vocab_size, model_dim)
        self.embed2.weight.label = 'embed2'

        # x0_lambdas separated out for different optimizer treatment (no beta smoothing)
        self.x0_lambdas = nn.Parameter(torch.zeros(2*num_layers))
        self.x0_lambdas.label = 'x0_lambdas'
        self.x0_lambdas.lr_mul = 5.0
        self.x0_lambdas.wd_mul = 0.0

        pad = (-num_layers * 3 - 5) % dist.get_world_size()  # updated: 3*num_layers instead of 4*
        self.scalars = nn.Parameter(
            torch.cat(
                [
                    1.05 * torch.ones(num_layers),  # resid lambdas. 1.05 init such that layer i weight is i^(num_layers-i).
                    *[torch.tensor([0.5, 1.0]) for _ in range(num_layers)],  # SA lambdas
                    torch.zeros(1), # smear_lambda
                    0.5*torch.ones(1), # backout_lambda
                    -1.5 * torch.ones(3),  # skip_lambdas -> σ(-1.5) ≈ 0.18
                    torch.ones(pad),
                ]
            )
        )

        self.scalars.label = 'scalars'
        # set learning rates
        for param in self.value_embeds.parameters():
            param.lr_mul = 75.
            param.wd_mul = 5.
        for param in self.embed.parameters():
            param.wd_mul = 150.
        for param in self.embed2.parameters():
            param.lr_mul = 75.
            param.wd_mul = 5.
        for param in self.lm_head.parameters():
            param.wd_mul = 150.
        self.scalars.lr_mul = 5.0
        self.scalars.wd_mul = 0.0

        self.split_embed = False

    def forward(self, input_seq: Tensor, target_seq: Tensor, seqlens: Tensor, schedule_cfg: ForwardScheduleConfig):
        assert input_seq.ndim == 1

        # unpack schedule_cfg
        mtp_weights, ws_short, ws_long = schedule_cfg.mtp_weights, schedule_cfg.ws_short, schedule_cfg.ws_long

        # set configs
        skip_connections = []
        skip_in = [2, 4, 6]
        skip_out = [9, 10, 11]
        x_backout = None
        backout_layer = 11

        # set lambdas
        resid_lambdas = self.scalars[: 1 * self.num_layers]
        x0_lambdas = self.x0_lambdas.view(-1, 2)
        sa_lambdas = self.scalars[1 * self.num_layers: 3 * self.num_layers].view(-1, 2)
        smear_lambda = self.scalars[3 * self.num_layers]
        backout_lambda = self.scalars[3 * self.num_layers+1]
        skip_lambdas = self.scalars[3 * self.num_layers+2: 3*self.num_layers+5]
        
        # set block masks and key shift
        short_bm = ws_short * args.block_size
        long_bm = ws_long * args.block_size
        bm_sizes = [long_bm, short_bm, short_bm, short_bm, long_bm, short_bm, short_bm, short_bm, 
            short_bm, short_bm, short_bm, long_bm, short_bm, short_bm, short_bm, long_bm]
        assert len(bm_sizes) == self.num_layers
        key_offset = [b==long_bm for b in bm_sizes] # apply partial key offset to long windows

        # weight-tied: use lm_head.weight for embedding lookup (or separate embed after split)
        if self.split_embed:
            x = self.embed(input_seq)
        else:
            x = F.embedding(input_seq, self.lm_head.weight)
        ve = [value_embed(input_seq) for value_embed in self.value_embeds]
        # 012 ... 012 structure on token value embeddings by @YouJiacheng, improved on @leloykun's U-net structure
        # dropping first layer updates this to .12 ... 012
        ve = [ve[0], ve[1], ve[2], ve[3], ve[4]] + [None] * (self.num_layers - 10) + [ve[0], ve[1], ve[2], ve[3], ve[4]]
        assert len(ve) == self.num_layers

        # smear token embed forward 1 position @classiclarryd
        smear_gate_out = smear_lambda * torch.sigmoid(self.smear_gate(x[1:, :self.smear_gate.weight.size(-1)]))
        x = torch.cat([x[:1], x[1:] + smear_gate_out * x[:-1]])
        x = x0 = norm(x[None])

        x02 = norm(self.embed2(input_seq)[None])

        skip_idx = 0
        for i in range(self.num_layers):
            attn_args = AttnArgs(
                ve=ve[i],
                sa_lambdas=sa_lambdas[i],
                seqlens=seqlens,
                bm_size=bm_sizes[i],
                cos=self.yarn.cos,
                sin=self.yarn.sin,
                attn_scale=self.yarn.attn_scale,
                key_offset = key_offset[i]
            )
            if i in skip_out:
                skip_gate_out = torch.sigmoid(skip_lambdas[skip_idx]) * 2 * torch.sigmoid(self.skip_gates[skip_idx](x0[..., :self.skip_gates[skip_idx].weight.size(-1)]))
                skip_idx += 1
                x = x + skip_gate_out * skip_connections.pop()
            if i == 0:
                x = (resid_lambdas[0] + x0_lambdas[0,0]) * x + x0_lambdas[0,1] * x02
            else:
                x = resid_lambdas[i] * x + x0_lambdas[i,0] * x0 + x0_lambdas[i,1] * x02
            x = self.blocks[i](x, attn_args)
            if i in skip_in:
                skip_connections.append(x)
            if i == backout_layer:
                x_backout = x

        # back out contributions from first 2/3 layers that are only required for downstream context and not direct prediction
        x -= backout_lambda * x_backout
        x = norm(x)

        if not self.training:
            loss = 0
            for i in range(4):
                logits: Tensor = F.linear(x.flatten(end_dim=1).chunk(4)[i], self.lm_head.weight.bfloat16()).float()
                logits = 23 * torch.sigmoid((logits + 5) / 7.5)
                loss += F.cross_entropy(logits.view(-1, logits.size(-1)), target_seq.chunk(4)[i], reduction="mean")/4
            return loss
        
        logits = self.lm_head(x)
        # @Grad62304977 added tanh softcapping following Gemma 2 paper, @KoszarskyB reduced it from 30 to 15
        # @YouJiacheng shifted it by +15 (2*sigmoid(2*x)=tanh(x)+1). @classiclarryd updated to 23*sigmoid((logits+5)/7.5)
        logits = 23 * torch.sigmoid((logits + 5) / 7.5)
        logits_for_loss = logits.float() if not self.training else logits

        n_predict = mtp_weights.size(0) if mtp_weights is not None else 1
        if self.training and n_predict > 1:
            # Multi-token prediction: take loss of the weighted average of next n_predict tokens
            logits_flat = logits_for_loss.view(-1, logits_for_loss.size(-1))
            idx = F.pad(target_seq, (0, n_predict - 1)).unfold(0, n_predict, 1)  # [T, n_predict] of shifted targets
            target_logits = logits_flat.gather(1, idx)
            cross_entropy = torch.logsumexp(logits_flat, dim=-1).unsqueeze(1) - target_logits
            for k in range(1, n_predict):  # zero out preds past end of sequence
                cross_entropy[-k:, k] = 0
            loss = (cross_entropy * mtp_weights).sum()
        elif self.training:
            loss = F.cross_entropy(logits_for_loss.view(-1, logits_for_loss.size(-1)), target_seq, reduction="sum")
        else:
            loss = F.cross_entropy(logits_for_loss.view(-1, logits_for_loss.size(-1)), target_seq, reduction="mean")
        return loss

# -----------------------------------------------------------------------------
# Distributed data loader

def _load_data_shard(file: Path):
    header = torch.from_file(str(file), False, 256, dtype=torch.int32) # header is 256 int32
    assert header[0] == 20240520, "magic number mismatch in the data .bin file"
    assert header[1] == 1, "unsupported version"
    num_tokens = int(header[2]) # number of tokens (claimed)
    with file.open("rb", buffering=0) as f:
        tokens = torch.empty(num_tokens, dtype=torch.uint16, pin_memory=True) # avoid pin_memory copy by @YouJiacheng
        f.seek(256 * 4)
        nbytes = f.readinto(tokens.numpy()) # avoid bytes->array copy by @YouJiacheng
        assert nbytes == 2 * num_tokens, "number of tokens read does not match header"
    return tokens

BOS_ID = 50256

class BOSFinder:
    # Helper for getting sequences that start at the beginning of documents by @varunneal based on work by @classiclarryd
    def __init__(self, tokens: Tensor, world_size: int = 1, quickload: bool = False):
        # Precompute BOS positions once per shard
        self.tokens=tokens
        self.size = tokens.numel()
        self.quickload = quickload
        if quickload:
            # only scan first 4 million tokens, then kickoff async thread to scan rest
            self.bos_idx = (tokens[:4_000_000] == BOS_ID).nonzero(as_tuple=True)[0].to(torch.int64).cpu().numpy()
            self.thread = None
            self.ready = threading.Event()
            self.start()
        else:
            self.bos_idx = (tokens == BOS_ID).nonzero(as_tuple=True)[0].to(torch.int64).cpu().numpy()
        self.i = 0
        self.world_size = world_size
        self.batch_iter = 0

    def _load(self):
        self.bos_idx_async = (self.tokens == BOS_ID).nonzero(as_tuple=True)[0].to(torch.int64).cpu().numpy()
        self.ready.set()

    def start(self):
        self.ready.clear()
        self.thread = threading.Thread(target=self._load)
        self.thread.start()

    def get(self):
        if self.thread:
            self.ready.wait()
            self.thread.join()
        self.bos_idx = self.bos_idx_async

    def next_batch(self, num_tokens_local: int, max_seq_len: int):
        # if quickload was used, repoint to the full dataset after 5 batches
        if self.quickload and self.batch_iter==5:
            self.get()
        n = len(self.bos_idx)
        starts = [[] for _ in range(self.world_size)]
        ends = [[] for _ in range(self.world_size)]

        idx = self.i
        for r in range(self.world_size):
            cur_len = 0
            while cur_len <= num_tokens_local:
                if idx >= n:
                    raise StopIteration(f"Insufficient BOS ahead; hit tail of shard.")
                cur = self.bos_idx[idx]
                starts[r].append(cur)
                end = min(self.bos_idx[idx + 1] if idx + 1 < n else self.size,
                          cur + max_seq_len,
                          cur + num_tokens_local - cur_len + 1)
                ends[r].append(end)
                cur_len += end - cur
                idx += 1

            assert cur_len == num_tokens_local + 1
        self.i = idx
        self.batch_iter+=1
        return starts, ends

class DataPreloader:
    # Helper for asynchronously loading next shard and indexing bos tokens
    def __init__(self, file_iter, world_size: int = 1):
        self.file_iter = file_iter
        self.world_size = world_size
        self.thread = None
        self.data = None
        self.ready = threading.Event()

    def _load(self):
        tokens = _load_data_shard(next(self.file_iter))
        self.data = (tokens, BOSFinder(tokens, self.world_size))
        self.ready.set()

    def start(self):
        self.ready.clear()
        self.thread = threading.Thread(target=self._load)
        self.thread.start()

    def get(self):
        if self.thread:
            self.ready.wait()
            self.thread.join()
        return self.data

def distributed_data_generator(filename_pattern: str, num_tokens: int, max_seq_len: int, grad_accum_steps: int = 1, align_to_bos: bool = True):
    # align_to_bos: each sequence begins with Beginning of Sequence token, sequences truncated to max_seq_len
    rank = dist.get_rank() if dist.is_initialized() else 0
    world_size = dist.get_world_size() if dist.is_initialized() else 1
    assert num_tokens % (world_size * grad_accum_steps) == 0, "Batch size must be divisible by world size"
    num_tokens = num_tokens // grad_accum_steps

    files = [Path(file) for file in sorted(glob.glob(filename_pattern))]
    if not files:
        raise FileNotFoundError(f"No files found for pattern: {filename_pattern}")

    file_iter = iter(files)  # Use itertools.cycle(files) for multi-epoch training
    tokens = _load_data_shard(next(file_iter))
    if align_to_bos:
        finder = BOSFinder(tokens, world_size=world_size, quickload=True)
        preloader = DataPreloader(file_iter, world_size)
        preloader.start()
    else:
        pos = 0  # for unaligned case

    while True:
        num_tokens_local = num_tokens // world_size
        max_num_docs = next_multiple_of_n(num_tokens_local // 300, n=128)  # median doc length is ~400

        if align_to_bos:
            try:
                seq_starts, seq_ends = finder.next_batch(num_tokens_local, max_seq_len)
                start_idxs, end_idxs = torch.tensor(seq_starts[rank]), torch.tensor(seq_ends[rank])
            except StopIteration:
                # This shard is exhausted, load the next one in the next loop iteration.
                tokens, finder = preloader.get()
                preloader.start()
                continue

            buf = torch.cat([tokens[i:j] for i, j in zip(start_idxs, end_idxs)])
            _inputs = buf[:-1]
            _targets = buf[1:]
            end_idxs[-1] -= 1  # last document was too long to account for _targets offset
            cum_lengths = (end_idxs - start_idxs).cumsum(0)

        else:
            if pos + num_tokens + 1 >= len(tokens):  # should not occur for val data
                tokens, pos = _load_data_shard(next(file_iter)), 0

            pos_local = pos + rank * num_tokens_local
            buf = tokens[pos_local: pos_local + num_tokens_local + 1]
            _inputs = buf[:-1].view(num_tokens_local, )
            _targets = buf[1:].view(num_tokens_local, )

            cum_lengths = torch.nonzero(_inputs == BOS_ID)[:, 0]
            pos += num_tokens


        _cum_lengths = torch.full((max_num_docs,), num_tokens_local)
        _cum_lengths[0] = 0
        _cum_lengths[1:len(cum_lengths) + 1] = cum_lengths

        # Cast to int32 on CPU before transfer to avoid dtype conversion during .to()
        _inputs = _inputs.to(dtype=torch.int32)
        _targets = _targets.to(dtype=torch.int64)
        _cum_lengths = _cum_lengths.to(dtype=torch.int32)

        new_params = yield (
            _inputs.to(device="cuda", non_blocking=True),
            _targets.to(device="cuda", non_blocking=True),
            _cum_lengths.to(device="cuda", non_blocking=True)
        )

        if new_params is not None:
            # makes it possible for generator to receive new (num_tokens, max_seq_len, grad_accum_steps) via .send()
            new_num_tokens, new_max_seq_len, new_grad_accum_steps = new_params
            assert new_num_tokens % (world_size * new_grad_accum_steps) == 0, "Num tokens must be divisible by world size"
            num_tokens = new_num_tokens // new_grad_accum_steps
            max_seq_len = new_max_seq_len

# -----------------------------------------------------------------------------
# Training Management

def get_bs(step: int):
    if step >= args.num_scheduled_iterations:
        return args.train_bs_extension
    x = step / args.num_scheduled_iterations
    bs_idx = int(len(args.train_bs_schedule) * x)
    return args.train_bs_schedule[bs_idx]

def get_ws(step: int):
    # set short window size to half of long window size
    # Higher ws on "extension" steps
    if step >= args.num_scheduled_iterations:
        return args.ws_final // 2, args.ws_final
    x = step / args.num_scheduled_iterations
    assert 0 <= x < 1
    ws_idx = int(len(args.ws_schedule) * x)
    return min(11,args.ws_schedule[ws_idx] // 2), args.ws_schedule[ws_idx]

# learning rate schedule: tied to batch size schedule, with cooldown at the end.
def get_lr(step: int):
    if step > args.num_scheduled_iterations:
        return 0.1
    lr_max = 1.0
    x = step / args.num_scheduled_iterations
    if x > 1/12:
       lr_max = 1.52  # (16/8)**0.6
    if x > 2/12:
        lr_max = 1.73  # (24/8)**0.5
    if x > 3/12:
        lr_max = 2.0
    if x >= 1 - args.cooldown_frac:
        w = (1 - x) / args.cooldown_frac
        lr = lr_max * w + (1 - w) * 0.1
        return lr
    return lr_max

def get_muon_momentum(step: int, muon_warmup_steps=300, muon_cooldown_steps=50, momentum_min=0.85, momentum_max=0.95):
    # warmup phase: linearly increase momentum from min to max
    # cooldown phase: linearly decrease momentum from max to min
    momentum_cd_start = args.num_iterations - muon_cooldown_steps
    if step < muon_warmup_steps:
        frac = step / muon_warmup_steps
        momentum = momentum_min + frac * (momentum_max - momentum_min)
    elif step > momentum_cd_start:
        frac = (step - momentum_cd_start) / muon_cooldown_steps
        momentum = momentum_max - frac * (momentum_max - momentum_min)
    else:
        momentum = momentum_max
    return momentum

class TrainingManager():
    """
    Manages three optimizers for Adam embed/lm_head, Adam scalars, and Muon weight matrices.
    Notable Features:
        1. Scalars are given higher momentum terms to smooth learning @ChrisJMcCormick
        2. Scalar weights are temporarily frozen during batch size or window size updates @ChrisJMcCormick
        3. Adam optimizers are only stepped on odd steps @classiclarryd
        4. Adam optimizers have hooks to start gradient communication during backwards pass @akash5474
        5. Muon has a linear momentum warmup and cooldown schedule
        6. Learning rates follow a linear decay schedule
        7. Embed/lm_head weights and optimizer state splits at 2/3 of training @classiclarryd

    Manages model architecture, data, and target that changes during training
    Notable Features:
        1. Multi Token Prediction schedule of [1, 0.5, 0.25->0] -> [1, 0.5->0] -> [1] @varunneal
        2. Sliding Attention window schedule of [1,3] -> [3,7] -> [5,11] -> [6,13]
        3. YaRN updates to RoPE on window changes
        4. Split embed and lm head at 2/3 of training
        5. Batch size schedule of 8 -> 16 -> 24
        6. Post training extension of long windows from 13 to 20
    """
    def __init__(self, model):
        self.mtp_weights_schedule = self._build_mtp_schedule()
        self.model = model

        adam_labels = ['lm_head', 'value_embed', 'smear_gate', 'skip_gate', 'x0_lambdas', 'embed2', 'embed']
        scalar_labels = ['scalars']
        muon_labels = ['attn_gate', 'value_embed_gate', 'attn', 'mlp']
        adam_params = [p for p in model.parameters() if getattr(p, 'label', None) in adam_labels]
        scalar_params = [p for p in model.parameters() if getattr(p, 'label', None) in scalar_labels]
        muon_params = [p for p in model.parameters() if getattr(p, 'label', None) in muon_labels]
        assert set(getattr(p, 'label', None) for p in model.parameters()) == set(adam_labels + scalar_labels + muon_labels), "All params must have label"

        self.adam_opt = DistAdam(adam_params, adam_labels, lr=0.004, betas=(0.8, 0.95), eps=1e-8, weight_decay=0.005)
        self.scalar_opt = DistAdam(scalar_params, scalar_labels, lr=0.008, betas=(0.9, 0.99), eps=1e-8, weight_decay=0.005)
        self.muon_opt = NorMuon(muon_params, lr=0.015, momentum=0.95, beta2=0.95, weight_decay=1.2)
        self.optimizers = [self.adam_opt, self.scalar_opt, self.muon_opt]
        # split after odd number step
        self.split_step = math.ceil(args.split_embed_frac * args.num_scheduled_iterations) | 1

        # set defaults
        for opt in self.optimizers:
            opt.freeze_timer = 0
            opt.odd_step_only = False 
            opt.should_sync = True
            for group in opt.param_groups:
                group["initial_lr"] = group["lr"]

        # on even steps, only step Muon params
        self.adam_opt.odd_step_only = True
        self.scalar_opt.odd_step_only = True

        self.reset()

    def _build_mtp_schedule(self):
        # Precompute MTP weights for all steps to avoid tensor allocation during training
        # Schedule: [1, 0.5, 0.25->0] -> [1, 0.5->0] -> [1]
        mtp_weights_schedule = []
        for s in range(args.num_iterations + 1):
            x = s / (args.num_scheduled_iterations/4)
            if x < 1/3:
                w = [1.0, 0.5, 0.25 * (1 - 3*x)]
            elif x < 2/3:
                w = [1.0, 0.5 * (1 - (3*x - 1))]
            else:
                w = [1.0]
            mtp_weights_schedule.append(torch.tensor(w, device=device))
        return mtp_weights_schedule

    def apply_final_ws_ext(self):
        self.ws_long = args.ws_validate_post_yarn_ext

    def get_forward_args(self):
        return ForwardScheduleConfig(
            mtp_weights = self.mtp_weights,
            ws_short = self.ws_short,
            ws_long = self.ws_long
        )
    
    def _is_active_step(self, opt, step: int):
        return (opt.odd_step_only and step%2==1) or not opt.odd_step_only

    def get_transition_steps(self):
        transition_steps = [0]
        ws_short, ws_long = get_ws(0)
        for step in range(1, args.num_iterations):
            ws_short, new_ws_long = get_ws(step)
            if new_ws_long != ws_long:
                transition_steps.append(step)
                ws_long = new_ws_long
        return transition_steps

    def advance_schedule(self, step: int):
        self.ws_short, new_ws_long = get_ws(step)
        # only apply yarn for first few
        if new_ws_long != self.ws_long and new_ws_long<=13:
            self.model.yarn.apply(self.ws_long, new_ws_long)

        new_batch_size = get_bs(step)
        if new_batch_size != self.batch_size:
            self.train_loader_send_args = (new_batch_size, args.train_max_seq_len, grad_accum_steps)
        else:
            self.train_loader_send_args = None

        self.ws_long = new_ws_long
        self.mtp_weights = self.mtp_weights_schedule[step]
    
    def step_optimizers(self, step: int):                
        step_lr = get_lr(step)
        muon_momentum = get_muon_momentum(step)
        for group in self.muon_opt.param_groups:
            group["momentum"] = muon_momentum

        for opt in self.optimizers:
            if opt.freeze_timer > 0:
                opt.freeze_timer -= 1
                opt.zero_grad(set_to_none=True)
            else:
                if self._is_active_step(opt, step):
                    for group in opt.param_groups:
                        group["lr"] = group["initial_lr"] * step_lr
                    opt.step()
                    opt.zero_grad(set_to_none=True)
                    if opt.odd_step_only:
                        opt.should_sync = False
        
        if step == self.split_step:
            self.adam_opt.copy_lm_to_embed()
            self.model.split_embed = True

    def start_transition(self, freeze_count=40):
        # freeze scalar weights during transition
        self.scalar_opt.freeze_timer = freeze_count
    
    def activate_hooks(self, step: int):
        for opt in self.optimizers:
            if self._is_active_step(opt, step):
                opt.should_sync = True

    def reset(self, state=None):
        if state is not None:
            for opt, opt_state in zip(self.optimizers, state):
                opt.should_sync = False
                opt.load_state_dict(opt_state)

        # muon momentum buffers not in state dict
        self.muon_opt.reset()
        self.model.split_embed = False

        self.ws_short, self.ws_long = get_ws(0)
        self.batch_size = get_bs(0)
        self.model.yarn.reset()

    def get_state(self):
        return [copy.deepcopy(opt.state_dict()) for opt in self.optimizers]

# -----------------------------------------------------------------------------
# int main

@dataclass
class Hyperparameters:
    # data
    train_files: str = "data/fineweb10B/fineweb_train_*.bin" # input .bin to train on
    val_files: str = "data/fineweb10B/fineweb_val_*.bin" # input .bin to eval validation loss on
    val_tokens: int = 10485760 # how many tokens of validation data? it's important to keep this fixed for consistent comparisons
    # batch sizes
    train_bs_schedule: tuple = (131072, 262144, 393216, 524288, 
                                524288, 524288, 524288, 524288,
                                524288, 524288, 524288, 524288
                               )
    train_bs_extension: int = 32 * 2048 * 8
    train_max_seq_len: int = 128 * 16 * 2 # doubled to enable longer window sizes
    val_batch_size: int = 4 * 64 * 1024 * 8
    # optimization
    num_scheduled_iterations: int = 4700  # number of steps to complete lr and ws schedule
    num_extension_iterations: int = 40  # number of steps to continue training at final lr and ws
    num_iterations: int = num_scheduled_iterations + num_extension_iterations
    cooldown_frac: float = 0.70  # fraction of num_scheduled_iterations spent cooling down the learning rate
    split_embed_frac: float = 2/3/4
    # evaluation and logging
    run_id: str = f"{uuid.uuid4()}"
    val_loss_every: int = 250  # every how many steps to evaluate val loss? 0 for only at the end
    save_checkpoint: bool = False
    # attention masking
    block_size: int = 128
    ws_schedule: tuple = (3, 7, 11, 13,
                          15, 17, 19, 21,
                          23, 23, 23, 23)
    ws_final: int = 23 # set final validation ws, used for YaRN extension and short window size
    ws_validate_post_yarn_ext: int = 27 # extend long windows out even further after applying YaRN

args = Hyperparameters()

data_path = os.environ.get("DATA_PATH", ".")
args.train_files = os.path.join(data_path, args.train_files)
args.val_files = os.path.join(data_path, args.val_files)

# torchrun sets these env variables
rank = int(os.environ["RANK"])
world_size = int(os.environ["WORLD_SIZE"])
assert 8 % world_size == 0, "world_size must be a divisor of 8"
grad_accum_steps = 8 // world_size
assert torch.cuda.is_available()
device = torch.device("cuda", int(os.environ["LOCAL_RANK"]))
torch.cuda.set_device(device)
dist.init_process_group(backend="nccl", device_id=device)
dist.barrier()
master_process = (rank == 0) # this process will do logging, checkpointing etc.

# begin logging
logfile = None
if master_process:
    run_id = args.run_id
    os.makedirs("logs", exist_ok=True)
    logfile = f"logs/{run_id}.txt"
    print(logfile)
def print0(s, console=False):
    if master_process:
        with open(logfile, "a") as f:
            if console:
                print(s)
            print(s, file=f)

# begin by printing this file (the Python code)
print0(code)
print0("="*100)
# log information about the hardware/software environment this is running on
print0(f"Running Python {sys.version}")
print0(f"Running PyTorch {torch.version.__version__} compiled for CUDA {torch.version.cuda}")
print0(f"Running Triton version {triton.__version__}")

def nvidia_smi():
    import subprocess  # avoid top level import
    return subprocess.run(["nvidia-smi"], stdout=subprocess.PIPE, stderr=subprocess.PIPE, text=True).stdout
print0(nvidia_smi())
print0("="*100)

model: nn.Module = GPT(
    vocab_size=50257,
    num_layers=16,
    num_heads=8,
    head_dim=128,
    model_dim=1024,
    max_seq_len=args.val_batch_size // (grad_accum_steps * world_size)
).cuda()
for m in model.modules():
    if isinstance(m, (nn.Embedding, nn.Linear)):
        m.weight.data = m.weight.data.bfloat16()
for param in model.parameters():
    dist.broadcast(param.detach(), 0)

model: nn.Module = torch.compile(model, dynamic=False, fullgraph=True)
training_manager = TrainingManager(model)

########################################
#            Warmup kernels            #
########################################
print0("Compiling model and warming up kernels (~7 minutes on first execution)", console=True)
# Warmup the training kernels, then re-initialize the state so we aren't cheating
initial_state = dict(model=copy.deepcopy(model.state_dict()),
                     optimizers=training_manager.get_state()) # save the initial state
train_loader = distributed_data_generator(args.train_files, args.train_bs_schedule[0], args.train_max_seq_len, grad_accum_steps=grad_accum_steps)
val_loader = distributed_data_generator(args.val_files, args.val_batch_size, -1, grad_accum_steps=grad_accum_steps, align_to_bos=False)

transition_steps = training_manager.get_transition_steps()
warmup_steps = sorted(set(s + offset for s in transition_steps for offset in [-1, 0, 1] if s + offset >= 0))
print0(f"Sampling steps {warmup_steps} for warmup", console=True)
for step in warmup_steps:
    training_manager.advance_schedule(step)
    model.eval()
    with torch.no_grad():
        inputs, targets, cum_seqlens = next(val_loader)
        model(inputs, targets, cum_seqlens, training_manager.get_forward_args())
    model.train()
    for idx in range(grad_accum_steps):
        # enable gradient sync for the DistAdam optimizers on the last iteration before we step them
        if idx == grad_accum_steps - 1:
            training_manager.activate_hooks(step)
        send_args = training_manager.train_loader_send_args
        inputs, targets, cum_seqlens = train_loader.send(send_args)
        (model(inputs, targets, cum_seqlens, training_manager.get_forward_args()) / grad_accum_steps).backward()
    training_manager.step_optimizers(step)
print0("Resetting Model", console=True)
model.zero_grad(set_to_none=True)
model.load_state_dict(initial_state["model"])
training_manager.reset(initial_state["optimizers"])
del val_loader, train_loader, initial_state
model.train()

########################################
#        Training and validation       #
########################################
train_loader = distributed_data_generator(args.train_files, args.train_bs_schedule[0], args.train_max_seq_len, grad_accum_steps=grad_accum_steps)

gc.collect()

training_time_ms = 0
# start the clock
torch.cuda.synchronize()
t0 = time.perf_counter()
# begin training
train_steps = args.num_iterations
for step in range(train_steps + 1):
    last_step = (step == train_steps)
    training_manager.advance_schedule(step)
    # --------------- VALIDATION SECTION -----------------
    if last_step or (args.val_loss_every > 0 and step % args.val_loss_every == 0):
        if last_step:
            training_manager.apply_final_ws_ext()
        # stop the clock
        torch.cuda.synchronize()
        training_time_ms += 1000 * (time.perf_counter() - t0)
        model.eval()
        assert args.val_tokens % args.val_batch_size == 0
        val_steps = grad_accum_steps * args.val_tokens // args.val_batch_size
        val_loader = distributed_data_generator(args.val_files, args.val_batch_size, -1, grad_accum_steps=grad_accum_steps, align_to_bos=False)
        val_loss = 0
        with torch.no_grad():
            for _ in range(val_steps):
                inputs, targets, cum_seqlens = next(val_loader)
                val_loss += model(inputs, targets, cum_seqlens, training_manager.get_forward_args())
        val_loss /= val_steps
        del val_loader
        dist.reduce(val_loss, 0, op=dist.ReduceOp.AVG)
        print0(f"step:{step}/{train_steps} val_loss:{val_loss:.4f} train_time:{training_time_ms:.0f}ms step_avg:{training_time_ms/max(step, 1):.2f}ms", console=True)
        model.train()
        # start the clock again
        torch.cuda.synchronize()
        t0 = time.perf_counter()

    if last_step:
        if master_process and args.save_checkpoint:
            log = dict(step=step, code=code, model=model.state_dict(), optimizers=[opt.state_dict() for opt in optimizers])
            os.makedirs(f"logs/{run_id}", exist_ok=True)
            torch.save(log, f"logs/{run_id}/state_step{step:06d}.pt")
        # the last step only has the validation loop, so break to avoid training
        break

    # --------------- TRAINING SECTION -----------------
    for idx in range(grad_accum_steps):
        # enable gradient sync for the DistAdam optimizers on the last iteration before we step them
        if idx == grad_accum_steps - 1:
            training_manager.activate_hooks(step)
        send_args = training_manager.train_loader_send_args
        inputs, targets, cum_seqlens = train_loader.send(send_args)
        (model(inputs, targets, cum_seqlens, training_manager.get_forward_args()) / grad_accum_steps).backward()
    training_manager.step_optimizers(step)

    # logging
    approx_training_time_ms = training_time_ms + 1000 * (time.perf_counter() - t0)
    print0(f"step:{step+1}/{train_steps} train_time:{approx_training_time_ms:.0f}ms step_avg:{approx_training_time_ms/(step + 1):.2f}ms", console=True)

print0(f"peak memory allocated: {torch.cuda.max_memory_allocated() // 1024 // 1024} MiB "
       f"reserved: {torch.cuda.max_memory_reserved() // 1024 // 1024} MiB", console=True)
dist.destroy_process_group()

====================================================================================================
Running Python 3.10.12 (main, Feb  4 2025, 14:57:36) [GCC 11.4.0]
Running PyTorch 2.10.0.dev20251210+cu126 compiled for CUDA 12.6
Running Triton version 3.6.0
Thu Jan  1 03:14:04 2026       
+-----------------------------------------------------------------------------------------+
| NVIDIA-SMI 550.127.08             Driver Version: 550.127.08     CUDA Version: 12.6     |
|-----------------------------------------+------------------------+----------------------+
| GPU  Name                 Persistence-M | Bus-Id          Disp.A | Volatile Uncorr. ECC |
| Fan  Temp   Perf          Pwr:Usage/Cap |           Memory-Usage | GPU-Util  Compute M. |
|                                         |                        |               MIG M. |
|=========================================+========================+======================|
|   0  NVIDIA H100 80GB HBM3          On  |   00000000:19:00.0 Off |                    0 |
| N/A   32C    P0            121W /  700W |    1520MiB /  81559MiB |      0%      Default |
|                                         |                        |             Disabled |
+-----------------------------------------+------------------------+----------------------+
|   1  NVIDIA H100 80GB HBM3          On  |   00000000:3B:00.0 Off |                    0 |
| N/A   30C    P0            124W /  700W |    1520MiB /  81559MiB |      0%      Default |
|                                         |                        |             Disabled |
+-----------------------------------------+------------------------+----------------------+
|   2  NVIDIA H100 80GB HBM3          On  |   00000000:4C:00.0 Off |                    0 |
| N/A   28C    P0            118W /  700W |    1520MiB /  81559MiB |      0%      Default |
|                                         |                        |             Disabled |
+-----------------------------------------+------------------------+----------------------+
|   3  NVIDIA H100 80GB HBM3          On  |   00000000:5D:00.0 Off |                    0 |
| N/A   30C    P0            119W /  700W |    1520MiB /  81559MiB |      0%      Default |
|                                         |                        |             Disabled |
+-----------------------------------------+------------------------+----------------------+
|   4  NVIDIA H100 80GB HBM3          On  |   00000000:9B:00.0 Off |                    0 |
| N/A   31C    P0            117W /  700W |    1520MiB /  81559MiB |      0%      Default |
|                                         |                        |             Disabled |
+-----------------------------------------+------------------------+----------------------+
|   5  NVIDIA H100 80GB HBM3          On  |   00000000:BB:00.0 Off |                    0 |
| N/A   29C    P0            118W /  700W |    1520MiB /  81559MiB |      0%      Default |
|                                         |                        |             Disabled |
+-----------------------------------------+------------------------+----------------------+
|   6  NVIDIA H100 80GB HBM3          On  |   00000000:CB:00.0 Off |                    0 |
| N/A   30C    P0            119W /  700W |    1520MiB /  81559MiB |      0%      Default |
|                                         |                        |             Disabled |
+-----------------------------------------+------------------------+----------------------+
|   7  NVIDIA H100 80GB HBM3          On  |   00000000:DB:00.0 Off |                    0 |
| N/A   27C    P0            112W /  700W |    1520MiB /  81559MiB |      0%      Default |
|                                         |                        |             Disabled |
+-----------------------------------------+------------------------+----------------------+
                                                                                         
+-----------------------------------------------------------------------------------------+
| Processes:                                                                              |
|  GPU   GI   CI        PID   Type   Process name                              GPU Memory |
|        ID   ID                                                               Usage      |
|=========================================================================================|
+-----------------------------------------------------------------------------------------+

====================================================================================================
74
Compiling model and warming up kernels (~7 minutes on first execution)
Sampling steps [0, 1, 391, 392, 393, 783, 784, 785, 1174, 1175, 1176, 1566, 1567, 1568, 1958, 1959, 1960, 2349, 2350, 2351, 2741, 2742, 2743, 3133, 3134, 3135] for warmup
Resetting Model
step:0/4740 val_loss:10.8323 train_time:1ms step_avg:0.57ms
step:1/4740 train_time:102ms step_avg:102.49ms
step:2/4740 train_time:172ms step_avg:85.92ms
step:3/4740 train_time:229ms step_avg:76.42ms
step:4/4740 train_time:314ms step_avg:78.52ms
step:5/4740 train_time:372ms step_avg:74.43ms
step:6/4740 train_time:457ms step_avg:76.15ms
step:7/4740 train_time:515ms step_avg:73.51ms
step:8/4740 train_time:599ms step_avg:74.93ms
step:9/4740 train_time:658ms step_avg:73.07ms
step:10/4740 train_time:743ms step_avg:74.28ms
step:11/4740 train_time:801ms step_avg:72.82ms
step:12/4740 train_time:886ms step_avg:73.84ms
step:13/4740 train_time:944ms step_avg:72.61ms
step:14/4740 train_time:1030ms step_avg:73.55ms
step:15/4740 train_time:1093ms step_avg:72.89ms
step:16/4740 train_time:1180ms step_avg:73.76ms
step:17/4740 train_time:1238ms step_avg:72.85ms
step:18/4740 train_time:1324ms step_avg:73.53ms
step:19/4740 train_time:1382ms step_avg:72.75ms
step:20/4740 train_time:1468ms step_avg:73.38ms
step:21/4740 train_time:1527ms step_avg:72.70ms
step:22/4740 train_time:1612ms step_avg:73.26ms
step:23/4740 train_time:1671ms step_avg:72.63ms
step:24/4740 train_time:1756ms step_avg:73.18ms
step:25/4740 train_time:1814ms step_avg:72.57ms
step:26/4740 train_time:1899ms step_avg:73.05ms
step:27/4740 train_time:1958ms step_avg:72.51ms
step:28/4740 train_time:2043ms step_avg:72.98ms
step:29/4740 train_time:2103ms step_avg:72.52ms
step:30/4740 train_time:2191ms step_avg:73.04ms
step:31/4740 train_time:2250ms step_avg:72.60ms
step:32/4740 train_time:2336ms step_avg:73.00ms
step:33/4740 train_time:2395ms step_avg:72.58ms
step:34/4740 train_time:2481ms step_avg:72.98ms
step:35/4740 train_time:2539ms step_avg:72.54ms
step:36/4740 train_time:2625ms step_avg:72.91ms
step:37/4740 train_time:2683ms step_avg:72.51ms
step:38/4740 train_time:2769ms step_avg:72.87ms
step:39/4740 train_time:2828ms step_avg:72.51ms
step:40/4740 train_time:2914ms step_avg:72.84ms
step:41/4740 train_time:2973ms step_avg:72.51ms
step:42/4740 train_time:3059ms step_avg:72.84ms
step:43/4740 train_time:3117ms step_avg:72.49ms
step:44/4740 train_time:3204ms step_avg:72.81ms
step:45/4740 train_time:3262ms step_avg:72.50ms
step:46/4740 train_time:3349ms step_avg:72.80ms
step:47/4740 train_time:3408ms step_avg:72.51ms
step:48/4740 train_time:3494ms step_avg:72.79ms
step:49/4740 train_time:3553ms step_avg:72.50ms
step:50/4740 train_time:3638ms step_avg:72.76ms
step:51/4740 train_time:3697ms step_avg:72.49ms
step:52/4740 train_time:3782ms step_avg:72.73ms
step:53/4740 train_time:3840ms step_avg:72.46ms
step:54/4740 train_time:3925ms step_avg:72.69ms
step:55/4740 train_time:3985ms step_avg:72.45ms
step:56/4740 train_time:4072ms step_avg:72.71ms
step:57/4740 train_time:4130ms step_avg:72.46ms
step:58/4740 train_time:4217ms step_avg:72.70ms
step:59/4740 train_time:4276ms step_avg:72.47ms
step:60/4740 train_time:4362ms step_avg:72.70ms
step:61/4740 train_time:4420ms step_avg:72.47ms
step:62/4740 train_time:4507ms step_avg:72.69ms
step:63/4740 train_time:4565ms step_avg:72.46ms
step:64/4740 train_time:4651ms step_avg:72.66ms
step:65/4740 train_time:4709ms step_avg:72.45ms
step:66/4740 train_time:4795ms step_avg:72.65ms
step:67/4740 train_time:4853ms step_avg:72.44ms
step:68/4740 train_time:4939ms step_avg:72.63ms
step:69/4740 train_time:4998ms step_avg:72.43ms
step:70/4740 train_time:5083ms step_avg:72.62ms
step:71/4740 train_time:5142ms step_avg:72.42ms
step:72/4740 train_time:5228ms step_avg:72.61ms
step:73/4740 train_time:5287ms step_avg:72.43ms
step:74/4740 train_time:5372ms step_avg:72.60ms
step:75/4740 train_time:5432ms step_avg:72.42ms
step:76/4740 train_time:5518ms step_avg:72.60ms
step:77/4740 train_time:5577ms step_avg:72.43ms
step:78/4740 train_time:5662ms step_avg:72.59ms
step:79/4740 train_time:5721ms step_avg:72.42ms
step:80/4740 train_time:5806ms step_avg:72.58ms
step:81/4740 train_time:5865ms step_avg:72.40ms
step:82/4740 train_time:5951ms step_avg:72.58ms
step:83/4740 train_time:6011ms step_avg:72.42ms
step:84/4740 train_time:6097ms step_avg:72.58ms
step:85/4740 train_time:6157ms step_avg:72.43ms
step:86/4740 train_time:6243ms step_avg:72.59ms
step:87/4740 train_time:6302ms step_avg:72.44ms
step:88/4740 train_time:6388ms step_avg:72.59ms
step:89/4740 train_time:6448ms step_avg:72.45ms
step:90/4740 train_time:6533ms step_avg:72.59ms
step:91/4740 train_time:6594ms step_avg:72.46ms
step:92/4740 train_time:6681ms step_avg:72.61ms
step:93/4740 train_time:6739ms step_avg:72.46ms
step:94/4740 train_time:6824ms step_avg:72.59ms
step:95/4740 train_time:6882ms step_avg:72.45ms
step:96/4740 train_time:6968ms step_avg:72.58ms
step:97/4740 train_time:7027ms step_avg:72.44ms
step:98/4740 train_time:7112ms step_avg:72.57ms
step:99/4740 train_time:7171ms step_avg:72.43ms
step:100/4740 train_time:7257ms step_avg:72.57ms
step:101/4740 train_time:7315ms step_avg:72.43ms
step:102/4740 train_time:7402ms step_avg:72.57ms
step:103/4740 train_time:7461ms step_avg:72.43ms
step:104/4740 train_time:7546ms step_avg:72.56ms
step:105/4740 train_time:7606ms step_avg:72.44ms
step:106/4740 train_time:7691ms step_avg:72.56ms
step:107/4740 train_time:7751ms step_avg:72.44ms
step:108/4740 train_time:7836ms step_avg:72.56ms
step:109/4740 train_time:7895ms step_avg:72.43ms
step:110/4740 train_time:7980ms step_avg:72.55ms
step:111/4740 train_time:8039ms step_avg:72.43ms
step:112/4740 train_time:8126ms step_avg:72.55ms
step:113/4740 train_time:8184ms step_avg:72.43ms
step:114/4740 train_time:8270ms step_avg:72.55ms
step:115/4740 train_time:8330ms step_avg:72.43ms
step:116/4740 train_time:8415ms step_avg:72.55ms
step:117/4740 train_time:8474ms step_avg:72.42ms
step:118/4740 train_time:8560ms step_avg:72.54ms
step:119/4740 train_time:8619ms step_avg:72.43ms
step:120/4740 train_time:8704ms step_avg:72.54ms
step:121/4740 train_time:8763ms step_avg:72.42ms
step:122/4740 train_time:8849ms step_avg:72.53ms
step:123/4740 train_time:8907ms step_avg:72.42ms
step:124/4740 train_time:8993ms step_avg:72.52ms
step:125/4740 train_time:9052ms step_avg:72.41ms
step:126/4740 train_time:9138ms step_avg:72.52ms
step:127/4740 train_time:9196ms step_avg:72.41ms
step:128/4740 train_time:9282ms step_avg:72.51ms
step:129/4740 train_time:9340ms step_avg:72.41ms
step:130/4740 train_time:9426ms step_avg:72.51ms
step:131/4740 train_time:9485ms step_avg:72.40ms
step:132/4740 train_time:9571ms step_avg:72.51ms
step:133/4740 train_time:9630ms step_avg:72.41ms
step:134/4740 train_time:9716ms step_avg:72.51ms
step:135/4740 train_time:9774ms step_avg:72.40ms
step:136/4740 train_time:9860ms step_avg:72.50ms
step:137/4740 train_time:9918ms step_avg:72.39ms
step:138/4740 train_time:10003ms step_avg:72.49ms
step:139/4740 train_time:10062ms step_avg:72.39ms
step:140/4740 train_time:10149ms step_avg:72.49ms
step:141/4740 train_time:10208ms step_avg:72.40ms
step:142/4740 train_time:10294ms step_avg:72.49ms
step:143/4740 train_time:10353ms step_avg:72.40ms
step:144/4740 train_time:10438ms step_avg:72.49ms
step:145/4740 train_time:10497ms step_avg:72.39ms
step:146/4740 train_time:10583ms step_avg:72.49ms
step:147/4740 train_time:10641ms step_avg:72.39ms
step:148/4740 train_time:10726ms step_avg:72.47ms
step:149/4740 train_time:10786ms step_avg:72.39ms
step:150/4740 train_time:10871ms step_avg:72.48ms
step:151/4740 train_time:10929ms step_avg:72.38ms
step:152/4740 train_time:11015ms step_avg:72.47ms
step:153/4740 train_time:11073ms step_avg:72.37ms
step:154/4740 train_time:11159ms step_avg:72.46ms
step:155/4740 train_time:11218ms step_avg:72.37ms
step:156/4740 train_time:11304ms step_avg:72.46ms
step:157/4740 train_time:11363ms step_avg:72.37ms
step:158/4740 train_time:11448ms step_avg:72.46ms
step:159/4740 train_time:11507ms step_avg:72.37ms
step:160/4740 train_time:11593ms step_avg:72.46ms
step:161/4740 train_time:11652ms step_avg:72.37ms
step:162/4740 train_time:11737ms step_avg:72.45ms
step:163/4740 train_time:11795ms step_avg:72.36ms
step:164/4740 train_time:11881ms step_avg:72.44ms
step:165/4740 train_time:11939ms step_avg:72.36ms
step:166/4740 train_time:12024ms step_avg:72.43ms
step:167/4740 train_time:12083ms step_avg:72.35ms
step:168/4740 train_time:12168ms step_avg:72.43ms
step:169/4740 train_time:12228ms step_avg:72.35ms
step:170/4740 train_time:12313ms step_avg:72.43ms
step:171/4740 train_time:12372ms step_avg:72.35ms
step:172/4740 train_time:12457ms step_avg:72.43ms
step:173/4740 train_time:12516ms step_avg:72.35ms
step:174/4740 train_time:12602ms step_avg:72.42ms
step:175/4740 train_time:12660ms step_avg:72.34ms
step:176/4740 train_time:12745ms step_avg:72.42ms
step:177/4740 train_time:12804ms step_avg:72.34ms
step:178/4740 train_time:12890ms step_avg:72.41ms
step:179/4740 train_time:12949ms step_avg:72.34ms
step:180/4740 train_time:13035ms step_avg:72.42ms
step:181/4740 train_time:13093ms step_avg:72.34ms
step:182/4740 train_time:13179ms step_avg:72.41ms
step:183/4740 train_time:13237ms step_avg:72.33ms
step:184/4740 train_time:13322ms step_avg:72.40ms
step:185/4740 train_time:13381ms step_avg:72.33ms
step:186/4740 train_time:13467ms step_avg:72.40ms
step:187/4740 train_time:13526ms step_avg:72.33ms
step:188/4740 train_time:13612ms step_avg:72.40ms
step:189/4740 train_time:13671ms step_avg:72.33ms
step:190/4740 train_time:13756ms step_avg:72.40ms
step:191/4740 train_time:13813ms step_avg:72.32ms
step:192/4740 train_time:13898ms step_avg:72.39ms
step:193/4740 train_time:13957ms step_avg:72.32ms
step:194/4740 train_time:14043ms step_avg:72.39ms
step:195/4740 train_time:14102ms step_avg:72.32ms
step:196/4740 train_time:14188ms step_avg:72.39ms
step:197/4740 train_time:14246ms step_avg:72.32ms
step:198/4740 train_time:14331ms step_avg:72.38ms
step:199/4740 train_time:14390ms step_avg:72.31ms
step:200/4740 train_time:14475ms step_avg:72.38ms
step:201/4740 train_time:14534ms step_avg:72.31ms
step:202/4740 train_time:14620ms step_avg:72.38ms
step:203/4740 train_time:14679ms step_avg:72.31ms
step:204/4740 train_time:14764ms step_avg:72.37ms
step:205/4740 train_time:14822ms step_avg:72.30ms
step:206/4740 train_time:14907ms step_avg:72.37ms
step:207/4740 train_time:14965ms step_avg:72.30ms
step:208/4740 train_time:15051ms step_avg:72.36ms
step:209/4740 train_time:15111ms step_avg:72.30ms
step:210/4740 train_time:15197ms step_avg:72.36ms
step:211/4740 train_time:15254ms step_avg:72.30ms
step:212/4740 train_time:15340ms step_avg:72.36ms
step:213/4740 train_time:15398ms step_avg:72.29ms
step:214/4740 train_time:15484ms step_avg:72.35ms
step:215/4740 train_time:15542ms step_avg:72.29ms
step:216/4740 train_time:15627ms step_avg:72.35ms
step:217/4740 train_time:15687ms step_avg:72.29ms
step:218/4740 train_time:15772ms step_avg:72.35ms
step:219/4740 train_time:15830ms step_avg:72.28ms
step:220/4740 train_time:15915ms step_avg:72.34ms
step:221/4740 train_time:15975ms step_avg:72.28ms
step:222/4740 train_time:16061ms step_avg:72.35ms
step:223/4740 train_time:16119ms step_avg:72.28ms
step:224/4740 train_time:16206ms step_avg:72.35ms
step:225/4740 train_time:16264ms step_avg:72.29ms
step:226/4740 train_time:16349ms step_avg:72.34ms
step:227/4740 train_time:16408ms step_avg:72.28ms
step:228/4740 train_time:16494ms step_avg:72.34ms
step:229/4740 train_time:16552ms step_avg:72.28ms
step:230/4740 train_time:16637ms step_avg:72.34ms
step:231/4740 train_time:16695ms step_avg:72.27ms
step:232/4740 train_time:16781ms step_avg:72.33ms
step:233/4740 train_time:16839ms step_avg:72.27ms
step:234/4740 train_time:16924ms step_avg:72.32ms
step:235/4740 train_time:16982ms step_avg:72.26ms
step:236/4740 train_time:17067ms step_avg:72.32ms
step:237/4740 train_time:17127ms step_avg:72.26ms
step:238/4740 train_time:17212ms step_avg:72.32ms
step:239/4740 train_time:17271ms step_avg:72.26ms
step:240/4740 train_time:17356ms step_avg:72.32ms
step:241/4740 train_time:17415ms step_avg:72.26ms
step:242/4740 train_time:17501ms step_avg:72.32ms
step:243/4740 train_time:17559ms step_avg:72.26ms
step:244/4740 train_time:17644ms step_avg:72.31ms
step:245/4740 train_time:17703ms step_avg:72.26ms
step:246/4740 train_time:17789ms step_avg:72.31ms
step:247/4740 train_time:17847ms step_avg:72.25ms
step:248/4740 train_time:17933ms step_avg:72.31ms
step:249/4740 train_time:17991ms step_avg:72.25ms
step:250/4740 train_time:18076ms step_avg:72.30ms
step:250/4740 val_loss:4.5394 train_time:18120ms step_avg:72.48ms
step:251/4740 train_time:18144ms step_avg:72.29ms
step:252/4740 train_time:18221ms step_avg:72.31ms
step:253/4740 train_time:18280ms step_avg:72.25ms
step:254/4740 train_time:18366ms step_avg:72.31ms
step:255/4740 train_time:18426ms step_avg:72.26ms
step:256/4740 train_time:18514ms step_avg:72.32ms
step:257/4740 train_time:18573ms step_avg:72.27ms
step:258/4740 train_time:18659ms step_avg:72.32ms
step:259/4740 train_time:18716ms step_avg:72.26ms
step:260/4740 train_time:18801ms step_avg:72.31ms
step:261/4740 train_time:18859ms step_avg:72.26ms
step:262/4740 train_time:18943ms step_avg:72.30ms
step:263/4740 train_time:19001ms step_avg:72.25ms
step:264/4740 train_time:19087ms step_avg:72.30ms
step:265/4740 train_time:19146ms step_avg:72.25ms
step:266/4740 train_time:19232ms step_avg:72.30ms
step:267/4740 train_time:19291ms step_avg:72.25ms
step:268/4740 train_time:19376ms step_avg:72.30ms
step:269/4740 train_time:19434ms step_avg:72.25ms
step:270/4740 train_time:19519ms step_avg:72.29ms
step:271/4740 train_time:19579ms step_avg:72.25ms
step:272/4740 train_time:19665ms step_avg:72.30ms
step:273/4740 train_time:19723ms step_avg:72.25ms
step:274/4740 train_time:19809ms step_avg:72.30ms
step:275/4740 train_time:19868ms step_avg:72.25ms
step:276/4740 train_time:19953ms step_avg:72.29ms
step:277/4740 train_time:20011ms step_avg:72.24ms
step:278/4740 train_time:20096ms step_avg:72.29ms
step:279/4740 train_time:20154ms step_avg:72.24ms
step:280/4740 train_time:20239ms step_avg:72.28ms
step:281/4740 train_time:20298ms step_avg:72.23ms
step:282/4740 train_time:20384ms step_avg:72.28ms
step:283/4740 train_time:20443ms step_avg:72.24ms
step:284/4740 train_time:20528ms step_avg:72.28ms
step:285/4740 train_time:20588ms step_avg:72.24ms
step:286/4740 train_time:20673ms step_avg:72.28ms
step:287/4740 train_time:20731ms step_avg:72.23ms
step:288/4740 train_time:20816ms step_avg:72.28ms
step:289/4740 train_time:20874ms step_avg:72.23ms
step:290/4740 train_time:20959ms step_avg:72.27ms
step:291/4740 train_time:21017ms step_avg:72.22ms
step:292/4740 train_time:21102ms step_avg:72.27ms
step:293/4740 train_time:21160ms step_avg:72.22ms
step:294/4740 train_time:21245ms step_avg:72.26ms
step:295/4740 train_time:21304ms step_avg:72.22ms
step:296/4740 train_time:21389ms step_avg:72.26ms
step:297/4740 train_time:21447ms step_avg:72.21ms
step:298/4740 train_time:21534ms step_avg:72.26ms
step:299/4740 train_time:21592ms step_avg:72.21ms
step:300/4740 train_time:21678ms step_avg:72.26ms
step:301/4740 train_time:21737ms step_avg:72.22ms
step:302/4740 train_time:21822ms step_avg:72.26ms
step:303/4740 train_time:21881ms step_avg:72.21ms
step:304/4740 train_time:21966ms step_avg:72.26ms
step:305/4740 train_time:22025ms step_avg:72.21ms
step:306/4740 train_time:22110ms step_avg:72.25ms
step:307/4740 train_time:22168ms step_avg:72.21ms
step:308/4740 train_time:22254ms step_avg:72.25ms
step:309/4740 train_time:22312ms step_avg:72.21ms
step:310/4740 train_time:22398ms step_avg:72.25ms
step:311/4740 train_time:22457ms step_avg:72.21ms
step:312/4740 train_time:22542ms step_avg:72.25ms
step:313/4740 train_time:22600ms step_avg:72.21ms
step:314/4740 train_time:22686ms step_avg:72.25ms
step:315/4740 train_time:22746ms step_avg:72.21ms
step:316/4740 train_time:22830ms step_avg:72.25ms
step:317/4740 train_time:22888ms step_avg:72.20ms
step:318/4740 train_time:22974ms step_avg:72.24ms
step:319/4740 train_time:23031ms step_avg:72.20ms
step:320/4740 train_time:23116ms step_avg:72.24ms
step:321/4740 train_time:23174ms step_avg:72.19ms
step:322/4740 train_time:23259ms step_avg:72.23ms
step:323/4740 train_time:23317ms step_avg:72.19ms
step:324/4740 train_time:23403ms step_avg:72.23ms
step:325/4740 train_time:23461ms step_avg:72.19ms
step:326/4740 train_time:23546ms step_avg:72.23ms
step:327/4740 train_time:23604ms step_avg:72.18ms
step:328/4740 train_time:23690ms step_avg:72.23ms
step:329/4740 train_time:23749ms step_avg:72.18ms
step:330/4740 train_time:23834ms step_avg:72.22ms
step:331/4740 train_time:23892ms step_avg:72.18ms
step:332/4740 train_time:23977ms step_avg:72.22ms
step:333/4740 train_time:24035ms step_avg:72.18ms
step:334/4740 train_time:24121ms step_avg:72.22ms
step:335/4740 train_time:24179ms step_avg:72.18ms
step:336/4740 train_time:24263ms step_avg:72.21ms
step:337/4740 train_time:24322ms step_avg:72.17ms
step:338/4740 train_time:24407ms step_avg:72.21ms
step:339/4740 train_time:24465ms step_avg:72.17ms
step:340/4740 train_time:24551ms step_avg:72.21ms
step:341/4740 train_time:24610ms step_avg:72.17ms
step:342/4740 train_time:24694ms step_avg:72.21ms
step:343/4740 train_time:24753ms step_avg:72.17ms
step:344/4740 train_time:24838ms step_avg:72.20ms
step:345/4740 train_time:24896ms step_avg:72.16ms
step:346/4740 train_time:24981ms step_avg:72.20ms
step:347/4740 train_time:25040ms step_avg:72.16ms
step:348/4740 train_time:25124ms step_avg:72.20ms
step:349/4740 train_time:25182ms step_avg:72.16ms
step:350/4740 train_time:25267ms step_avg:72.19ms
step:351/4740 train_time:25326ms step_avg:72.15ms
step:352/4740 train_time:25412ms step_avg:72.19ms
step:353/4740 train_time:25469ms step_avg:72.15ms
step:354/4740 train_time:25555ms step_avg:72.19ms
step:355/4740 train_time:25612ms step_avg:72.15ms
step:356/4740 train_time:25698ms step_avg:72.18ms
step:357/4740 train_time:25756ms step_avg:72.15ms
step:358/4740 train_time:25841ms step_avg:72.18ms
step:359/4740 train_time:25900ms step_avg:72.14ms
step:360/4740 train_time:25986ms step_avg:72.18ms
step:361/4740 train_time:26044ms step_avg:72.15ms
step:362/4740 train_time:26129ms step_avg:72.18ms
step:363/4740 train_time:26188ms step_avg:72.14ms
step:364/4740 train_time:26274ms step_avg:72.18ms
step:365/4740 train_time:26331ms step_avg:72.14ms
step:366/4740 train_time:26417ms step_avg:72.18ms
step:367/4740 train_time:26476ms step_avg:72.14ms
step:368/4740 train_time:26562ms step_avg:72.18ms
step:369/4740 train_time:26621ms step_avg:72.14ms
step:370/4740 train_time:26707ms step_avg:72.18ms
step:371/4740 train_time:26766ms step_avg:72.14ms
step:372/4740 train_time:26851ms step_avg:72.18ms
step:373/4740 train_time:26909ms step_avg:72.14ms
step:374/4740 train_time:26994ms step_avg:72.18ms
step:375/4740 train_time:27052ms step_avg:72.14ms
step:376/4740 train_time:27138ms step_avg:72.18ms
step:377/4740 train_time:27196ms step_avg:72.14ms
step:378/4740 train_time:27281ms step_avg:72.17ms
step:379/4740 train_time:27339ms step_avg:72.13ms
step:380/4740 train_time:27425ms step_avg:72.17ms
step:381/4740 train_time:27483ms step_avg:72.13ms
step:382/4740 train_time:27568ms step_avg:72.17ms
step:383/4740 train_time:27627ms step_avg:72.13ms
step:384/4740 train_time:27712ms step_avg:72.17ms
step:385/4740 train_time:27770ms step_avg:72.13ms
step:386/4740 train_time:27855ms step_avg:72.16ms
step:387/4740 train_time:27913ms step_avg:72.13ms
step:388/4740 train_time:27998ms step_avg:72.16ms
step:389/4740 train_time:28057ms step_avg:72.13ms
step:390/4740 train_time:28142ms step_avg:72.16ms
step:391/4740 train_time:28201ms step_avg:72.13ms
step:392/4740 train_time:28286ms step_avg:72.16ms
step:393/4740 train_time:28358ms step_avg:72.16ms
step:394/4740 train_time:28515ms step_avg:72.37ms
step:395/4740 train_time:28616ms step_avg:72.44ms
step:396/4740 train_time:28773ms step_avg:72.66ms
step:397/4740 train_time:28872ms step_avg:72.73ms
step:398/4740 train_time:29032ms step_avg:72.94ms
step:399/4740 train_time:29130ms step_avg:73.01ms
step:400/4740 train_time:29288ms step_avg:73.22ms
step:401/4740 train_time:29386ms step_avg:73.28ms
step:402/4740 train_time:29544ms step_avg:73.49ms
step:403/4740 train_time:29644ms step_avg:73.56ms
step:404/4740 train_time:29803ms step_avg:73.77ms
step:405/4740 train_time:29903ms step_avg:73.83ms
step:406/4740 train_time:30063ms step_avg:74.05ms
step:407/4740 train_time:30161ms step_avg:74.11ms
step:408/4740 train_time:30319ms step_avg:74.31ms
step:409/4740 train_time:30419ms step_avg:74.37ms
step:410/4740 train_time:30577ms step_avg:74.58ms
step:411/4740 train_time:30678ms step_avg:74.64ms
step:412/4740 train_time:30838ms step_avg:74.85ms
step:413/4740 train_time:30937ms step_avg:74.91ms
step:414/4740 train_time:31095ms step_avg:75.11ms
step:415/4740 train_time:31196ms step_avg:75.17ms
step:416/4740 train_time:31353ms step_avg:75.37ms
step:417/4740 train_time:31454ms step_avg:75.43ms
step:418/4740 train_time:31612ms step_avg:75.63ms
step:419/4740 train_time:31711ms step_avg:75.68ms
step:420/4740 train_time:31871ms step_avg:75.88ms
step:421/4740 train_time:31969ms step_avg:75.94ms
step:422/4740 train_time:32127ms step_avg:76.13ms
step:423/4740 train_time:32226ms step_avg:76.18ms
step:424/4740 train_time:32387ms step_avg:76.38ms
step:425/4740 train_time:32484ms step_avg:76.43ms
step:426/4740 train_time:32644ms step_avg:76.63ms
step:427/4740 train_time:32742ms step_avg:76.68ms
step:428/4740 train_time:32903ms step_avg:76.88ms
step:429/4740 train_time:33001ms step_avg:76.93ms
step:430/4740 train_time:33160ms step_avg:77.12ms
step:431/4740 train_time:33260ms step_avg:77.17ms
step:432/4740 train_time:33419ms step_avg:77.36ms
step:433/4740 train_time:33517ms step_avg:77.41ms
step:434/4740 train_time:33676ms step_avg:77.59ms
step:435/4740 train_time:33777ms step_avg:77.65ms
step:436/4740 train_time:33935ms step_avg:77.83ms
step:437/4740 train_time:34034ms step_avg:77.88ms
step:438/4740 train_time:34193ms step_avg:78.07ms
step:439/4740 train_time:34292ms step_avg:78.11ms
step:440/4740 train_time:34451ms step_avg:78.30ms
step:441/4740 train_time:34549ms step_avg:78.34ms
step:442/4740 train_time:34709ms step_avg:78.53ms
step:443/4740 train_time:34808ms step_avg:78.57ms
step:444/4740 train_time:34967ms step_avg:78.75ms
step:445/4740 train_time:35066ms step_avg:78.80ms
step:446/4740 train_time:35225ms step_avg:78.98ms
step:447/4740 train_time:35324ms step_avg:79.02ms
step:448/4740 train_time:35484ms step_avg:79.21ms
step:449/4740 train_time:35583ms step_avg:79.25ms
step:450/4740 train_time:35742ms step_avg:79.43ms
step:451/4740 train_time:35843ms step_avg:79.47ms
step:452/4740 train_time:36002ms step_avg:79.65ms
step:453/4740 train_time:36101ms step_avg:79.69ms
step:454/4740 train_time:36258ms step_avg:79.86ms
step:455/4740 train_time:36360ms step_avg:79.91ms
step:456/4740 train_time:36519ms step_avg:80.09ms
step:457/4740 train_time:36619ms step_avg:80.13ms
step:458/4740 train_time:36778ms step_avg:80.30ms
step:459/4740 train_time:36880ms step_avg:80.35ms
step:460/4740 train_time:37038ms step_avg:80.52ms
step:461/4740 train_time:37137ms step_avg:80.56ms
step:462/4740 train_time:37295ms step_avg:80.73ms
step:463/4740 train_time:37396ms step_avg:80.77ms
step:464/4740 train_time:37554ms step_avg:80.94ms
step:465/4740 train_time:37655ms step_avg:80.98ms
step:466/4740 train_time:37813ms step_avg:81.14ms
step:467/4740 train_time:37913ms step_avg:81.18ms
step:468/4740 train_time:38071ms step_avg:81.35ms
step:469/4740 train_time:38170ms step_avg:81.39ms
step:470/4740 train_time:38329ms step_avg:81.55ms
step:471/4740 train_time:38427ms step_avg:81.59ms
step:472/4740 train_time:38588ms step_avg:81.75ms
step:473/4740 train_time:38687ms step_avg:81.79ms
step:474/4740 train_time:38844ms step_avg:81.95ms
step:475/4740 train_time:38944ms step_avg:81.99ms
step:476/4740 train_time:39104ms step_avg:82.15ms
step:477/4740 train_time:39202ms step_avg:82.18ms
step:478/4740 train_time:39361ms step_avg:82.34ms
step:479/4740 train_time:39460ms step_avg:82.38ms
step:480/4740 train_time:39619ms step_avg:82.54ms
step:481/4740 train_time:39719ms step_avg:82.58ms
step:482/4740 train_time:39876ms step_avg:82.73ms
step:483/4740 train_time:39978ms step_avg:82.77ms
step:484/4740 train_time:40136ms step_avg:82.93ms
step:485/4740 train_time:40235ms step_avg:82.96ms
step:486/4740 train_time:40393ms step_avg:83.11ms
step:487/4740 train_time:40491ms step_avg:83.14ms
step:488/4740 train_time:40651ms step_avg:83.30ms
step:489/4740 train_time:40749ms step_avg:83.33ms
step:490/4740 train_time:40909ms step_avg:83.49ms
step:491/4740 train_time:41009ms step_avg:83.52ms
step:492/4740 train_time:41168ms step_avg:83.68ms
step:493/4740 train_time:41267ms step_avg:83.71ms
step:494/4740 train_time:41426ms step_avg:83.86ms
step:495/4740 train_time:41524ms step_avg:83.89ms
step:496/4740 train_time:41686ms step_avg:84.04ms
step:497/4740 train_time:41783ms step_avg:84.07ms
step:498/4740 train_time:41943ms step_avg:84.22ms
step:499/4740 train_time:42043ms step_avg:84.26ms
step:500/4740 train_time:42202ms step_avg:84.40ms
step:500/4740 val_loss:4.0358 train_time:42274ms step_avg:84.55ms
step:501/4740 train_time:42304ms step_avg:84.44ms
step:502/4740 train_time:42463ms step_avg:84.59ms
step:503/4740 train_time:42563ms step_avg:84.62ms
step:504/4740 train_time:42728ms step_avg:84.78ms
step:505/4740 train_time:42825ms step_avg:84.80ms
step:506/4740 train_time:42982ms step_avg:84.95ms
step:507/4740 train_time:43081ms step_avg:84.97ms
step:508/4740 train_time:43239ms step_avg:85.12ms
step:509/4740 train_time:43343ms step_avg:85.15ms
step:510/4740 train_time:43502ms step_avg:85.30ms
step:511/4740 train_time:43603ms step_avg:85.33ms
step:512/4740 train_time:43766ms step_avg:85.48ms
step:513/4740 train_time:43865ms step_avg:85.51ms
step:514/4740 train_time:44022ms step_avg:85.65ms
step:515/4740 train_time:44121ms step_avg:85.67ms
step:516/4740 train_time:44278ms step_avg:85.81ms
step:517/4740 train_time:44381ms step_avg:85.84ms
step:518/4740 train_time:44539ms step_avg:85.98ms
step:519/4740 train_time:44641ms step_avg:86.01ms
step:520/4740 train_time:44800ms step_avg:86.15ms
step:521/4740 train_time:44899ms step_avg:86.18ms
step:522/4740 train_time:45056ms step_avg:86.31ms
step:523/4740 train_time:45154ms step_avg:86.34ms
step:524/4740 train_time:45313ms step_avg:86.48ms
step:525/4740 train_time:45412ms step_avg:86.50ms
step:526/4740 train_time:45572ms step_avg:86.64ms
step:527/4740 train_time:45672ms step_avg:86.66ms
step:528/4740 train_time:45831ms step_avg:86.80ms
step:529/4740 train_time:45930ms step_avg:86.82ms
step:530/4740 train_time:46087ms step_avg:86.96ms
step:531/4740 train_time:46186ms step_avg:86.98ms
step:532/4740 train_time:46346ms step_avg:87.12ms
step:533/4740 train_time:46445ms step_avg:87.14ms
step:534/4740 train_time:46605ms step_avg:87.28ms
step:535/4740 train_time:46705ms step_avg:87.30ms
step:536/4740 train_time:46864ms step_avg:87.43ms
step:537/4740 train_time:46962ms step_avg:87.45ms
step:538/4740 train_time:47119ms step_avg:87.58ms
step:539/4740 train_time:47223ms step_avg:87.61ms
step:540/4740 train_time:47381ms step_avg:87.74ms
step:541/4740 train_time:47480ms step_avg:87.76ms
step:542/4740 train_time:47639ms step_avg:87.89ms
step:543/4740 train_time:47742ms step_avg:87.92ms
step:544/4740 train_time:47902ms step_avg:88.05ms
step:545/4740 train_time:48001ms step_avg:88.07ms
step:546/4740 train_time:48159ms step_avg:88.20ms
step:547/4740 train_time:48259ms step_avg:88.22ms
step:548/4740 train_time:48417ms step_avg:88.35ms
step:549/4740 train_time:48518ms step_avg:88.37ms
step:550/4740 train_time:48675ms step_avg:88.50ms
step:551/4740 train_time:48774ms step_avg:88.52ms
step:552/4740 train_time:48935ms step_avg:88.65ms
step:553/4740 train_time:49033ms step_avg:88.67ms
step:554/4740 train_time:49192ms step_avg:88.79ms
step:555/4740 train_time:49291ms step_avg:88.81ms
step:556/4740 train_time:49450ms step_avg:88.94ms
step:557/4740 train_time:49548ms step_avg:88.96ms
step:558/4740 train_time:49707ms step_avg:89.08ms
step:559/4740 train_time:49807ms step_avg:89.10ms
step:560/4740 train_time:49968ms step_avg:89.23ms
step:561/4740 train_time:50067ms step_avg:89.25ms
step:562/4740 train_time:50226ms step_avg:89.37ms
step:563/4740 train_time:50326ms step_avg:89.39ms
step:564/4740 train_time:50487ms step_avg:89.52ms
step:565/4740 train_time:50584ms step_avg:89.53ms
step:566/4740 train_time:50742ms step_avg:89.65ms
step:567/4740 train_time:50843ms step_avg:89.67ms
step:568/4740 train_time:51003ms step_avg:89.79ms
step:569/4740 train_time:51101ms step_avg:89.81ms
step:570/4740 train_time:51260ms step_avg:89.93ms
step:571/4740 train_time:51361ms step_avg:89.95ms
step:572/4740 train_time:51518ms step_avg:90.07ms
step:573/4740 train_time:51620ms step_avg:90.09ms
step:574/4740 train_time:51776ms step_avg:90.20ms
step:575/4740 train_time:51876ms step_avg:90.22ms
step:576/4740 train_time:52036ms step_avg:90.34ms
step:577/4740 train_time:52134ms step_avg:90.35ms
step:578/4740 train_time:52293ms step_avg:90.47ms
step:579/4740 train_time:52392ms step_avg:90.49ms
step:580/4740 train_time:52551ms step_avg:90.61ms
step:581/4740 train_time:52649ms step_avg:90.62ms
step:582/4740 train_time:52808ms step_avg:90.74ms
step:583/4740 train_time:52908ms step_avg:90.75ms
step:584/4740 train_time:53068ms step_avg:90.87ms
step:585/4740 train_time:53166ms step_avg:90.88ms
step:586/4740 train_time:53325ms step_avg:91.00ms
step:587/4740 train_time:53424ms step_avg:91.01ms
step:588/4740 train_time:53583ms step_avg:91.13ms
step:589/4740 train_time:53682ms step_avg:91.14ms
step:590/4740 train_time:53839ms step_avg:91.25ms
step:591/4740 train_time:53940ms step_avg:91.27ms
step:592/4740 train_time:54098ms step_avg:91.38ms
step:593/4740 train_time:54199ms step_avg:91.40ms
step:594/4740 train_time:54356ms step_avg:91.51ms
step:595/4740 train_time:54457ms step_avg:91.52ms
step:596/4740 train_time:54615ms step_avg:91.64ms
step:597/4740 train_time:54714ms step_avg:91.65ms
step:598/4740 train_time:54873ms step_avg:91.76ms
step:599/4740 train_time:54973ms step_avg:91.78ms
step:600/4740 train_time:55132ms step_avg:91.89ms
step:601/4740 train_time:55231ms step_avg:91.90ms
step:602/4740 train_time:55390ms step_avg:92.01ms
step:603/4740 train_time:55489ms step_avg:92.02ms
step:604/4740 train_time:55648ms step_avg:92.13ms
step:605/4740 train_time:55747ms step_avg:92.14ms
step:606/4740 train_time:55907ms step_avg:92.26ms
step:607/4740 train_time:56005ms step_avg:92.27ms
step:608/4740 train_time:56166ms step_avg:92.38ms
step:609/4740 train_time:56265ms step_avg:92.39ms
step:610/4740 train_time:56423ms step_avg:92.50ms
step:611/4740 train_time:56523ms step_avg:92.51ms
step:612/4740 train_time:56682ms step_avg:92.62ms
step:613/4740 train_time:56783ms step_avg:92.63ms
step:614/4740 train_time:56943ms step_avg:92.74ms
step:615/4740 train_time:57044ms step_avg:92.75ms
step:616/4740 train_time:57202ms step_avg:92.86ms
step:617/4740 train_time:57302ms step_avg:92.87ms
step:618/4740 train_time:57460ms step_avg:92.98ms
step:619/4740 train_time:57559ms step_avg:92.99ms
step:620/4740 train_time:57717ms step_avg:93.09ms
step:621/4740 train_time:57818ms step_avg:93.11ms
step:622/4740 train_time:57976ms step_avg:93.21ms
step:623/4740 train_time:58074ms step_avg:93.22ms
step:624/4740 train_time:58235ms step_avg:93.32ms
step:625/4740 train_time:58334ms step_avg:93.33ms
step:626/4740 train_time:58493ms step_avg:93.44ms
step:627/4740 train_time:58591ms step_avg:93.45ms
step:628/4740 train_time:58751ms step_avg:93.55ms
step:629/4740 train_time:58849ms step_avg:93.56ms
step:630/4740 train_time:59008ms step_avg:93.66ms
step:631/4740 train_time:59107ms step_avg:93.67ms
step:632/4740 train_time:59266ms step_avg:93.78ms
step:633/4740 train_time:59365ms step_avg:93.78ms
step:634/4740 train_time:59525ms step_avg:93.89ms
step:635/4740 train_time:59624ms step_avg:93.90ms
step:636/4740 train_time:59782ms step_avg:94.00ms
step:637/4740 train_time:59883ms step_avg:94.01ms
step:638/4740 train_time:60041ms step_avg:94.11ms
step:639/4740 train_time:60143ms step_avg:94.12ms
step:640/4740 train_time:60302ms step_avg:94.22ms
step:641/4740 train_time:60401ms step_avg:94.23ms
step:642/4740 train_time:60559ms step_avg:94.33ms
step:643/4740 train_time:60659ms step_avg:94.34ms
step:644/4740 train_time:60816ms step_avg:94.44ms
step:645/4740 train_time:60917ms step_avg:94.44ms
step:646/4740 train_time:61075ms step_avg:94.54ms
step:647/4740 train_time:61174ms step_avg:94.55ms
step:648/4740 train_time:61332ms step_avg:94.65ms
step:649/4740 train_time:61431ms step_avg:94.66ms
step:650/4740 train_time:61590ms step_avg:94.75ms
step:651/4740 train_time:61688ms step_avg:94.76ms
step:652/4740 train_time:61848ms step_avg:94.86ms
step:653/4740 train_time:61948ms step_avg:94.87ms
step:654/4740 train_time:62106ms step_avg:94.96ms
step:655/4740 train_time:62205ms step_avg:94.97ms
step:656/4740 train_time:62364ms step_avg:95.07ms
step:657/4740 train_time:62464ms step_avg:95.07ms
step:658/4740 train_time:62622ms step_avg:95.17ms
step:659/4740 train_time:62722ms step_avg:95.18ms
step:660/4740 train_time:62882ms step_avg:95.28ms
step:661/4740 train_time:62980ms step_avg:95.28ms
step:662/4740 train_time:63140ms step_avg:95.38ms
step:663/4740 train_time:63241ms step_avg:95.39ms
step:664/4740 train_time:63399ms step_avg:95.48ms
step:665/4740 train_time:63498ms step_avg:95.49ms
step:666/4740 train_time:63656ms step_avg:95.58ms
step:667/4740 train_time:63755ms step_avg:95.58ms
step:668/4740 train_time:63915ms step_avg:95.68ms
step:669/4740 train_time:64013ms step_avg:95.69ms
step:670/4740 train_time:64173ms step_avg:95.78ms
step:671/4740 train_time:64272ms step_avg:95.79ms
step:672/4740 train_time:64431ms step_avg:95.88ms
step:673/4740 train_time:64530ms step_avg:95.88ms
step:674/4740 train_time:64690ms step_avg:95.98ms
step:675/4740 train_time:64789ms step_avg:95.98ms
step:676/4740 train_time:64949ms step_avg:96.08ms
step:677/4740 train_time:65048ms step_avg:96.08ms
step:678/4740 train_time:65207ms step_avg:96.17ms
step:679/4740 train_time:65306ms step_avg:96.18ms
step:680/4740 train_time:65466ms step_avg:96.27ms
step:681/4740 train_time:65565ms step_avg:96.28ms
step:682/4740 train_time:65726ms step_avg:96.37ms
step:683/4740 train_time:65825ms step_avg:96.38ms
step:684/4740 train_time:65984ms step_avg:96.47ms
step:685/4740 train_time:66084ms step_avg:96.47ms
step:686/4740 train_time:66243ms step_avg:96.56ms
step:687/4740 train_time:66342ms step_avg:96.57ms
step:688/4740 train_time:66500ms step_avg:96.66ms
step:689/4740 train_time:66600ms step_avg:96.66ms
step:690/4740 train_time:66759ms step_avg:96.75ms
step:691/4740 train_time:66859ms step_avg:96.76ms
step:692/4740 train_time:67017ms step_avg:96.84ms
step:693/4740 train_time:67118ms step_avg:96.85ms
step:694/4740 train_time:67275ms step_avg:96.94ms
step:695/4740 train_time:67375ms step_avg:96.94ms
step:696/4740 train_time:67534ms step_avg:97.03ms
step:697/4740 train_time:67632ms step_avg:97.03ms
step:698/4740 train_time:67791ms step_avg:97.12ms
step:699/4740 train_time:67889ms step_avg:97.12ms
step:700/4740 train_time:68049ms step_avg:97.21ms
step:701/4740 train_time:68147ms step_avg:97.21ms
step:702/4740 train_time:68307ms step_avg:97.30ms
step:703/4740 train_time:68405ms step_avg:97.30ms
step:704/4740 train_time:68566ms step_avg:97.39ms
step:705/4740 train_time:68665ms step_avg:97.40ms
step:706/4740 train_time:68824ms step_avg:97.48ms
step:707/4740 train_time:68924ms step_avg:97.49ms
step:708/4740 train_time:69083ms step_avg:97.57ms
step:709/4740 train_time:69182ms step_avg:97.58ms
step:710/4740 train_time:69341ms step_avg:97.66ms
step:711/4740 train_time:69443ms step_avg:97.67ms
step:712/4740 train_time:69601ms step_avg:97.75ms
step:713/4740 train_time:69701ms step_avg:97.76ms
step:714/4740 train_time:69859ms step_avg:97.84ms
step:715/4740 train_time:69961ms step_avg:97.85ms
step:716/4740 train_time:70119ms step_avg:97.93ms
step:717/4740 train_time:70222ms step_avg:97.94ms
step:718/4740 train_time:70379ms step_avg:98.02ms
step:719/4740 train_time:70480ms step_avg:98.02ms
step:720/4740 train_time:70637ms step_avg:98.11ms
step:721/4740 train_time:70737ms step_avg:98.11ms
step:722/4740 train_time:70895ms step_avg:98.19ms
step:723/4740 train_time:70995ms step_avg:98.19ms
step:724/4740 train_time:71155ms step_avg:98.28ms
step:725/4740 train_time:71254ms step_avg:98.28ms
step:726/4740 train_time:71414ms step_avg:98.37ms
step:727/4740 train_time:71513ms step_avg:98.37ms
step:728/4740 train_time:71672ms step_avg:98.45ms
step:729/4740 train_time:71771ms step_avg:98.45ms
step:730/4740 train_time:71931ms step_avg:98.54ms
step:731/4740 train_time:72029ms step_avg:98.53ms
step:732/4740 train_time:72188ms step_avg:98.62ms
step:733/4740 train_time:72287ms step_avg:98.62ms
step:734/4740 train_time:72447ms step_avg:98.70ms
step:735/4740 train_time:72545ms step_avg:98.70ms
step:736/4740 train_time:72704ms step_avg:98.78ms
step:737/4740 train_time:72803ms step_avg:98.78ms
step:738/4740 train_time:72963ms step_avg:98.87ms
step:739/4740 train_time:73063ms step_avg:98.87ms
step:740/4740 train_time:73222ms step_avg:98.95ms
step:741/4740 train_time:73323ms step_avg:98.95ms
step:742/4740 train_time:73481ms step_avg:99.03ms
step:743/4740 train_time:73582ms step_avg:99.03ms
step:744/4740 train_time:73740ms step_avg:99.11ms
step:745/4740 train_time:73841ms step_avg:99.12ms
step:746/4740 train_time:73999ms step_avg:99.19ms
step:747/4740 train_time:74099ms step_avg:99.20ms
step:748/4740 train_time:74257ms step_avg:99.27ms
step:749/4740 train_time:74358ms step_avg:99.28ms
step:750/4740 train_time:74515ms step_avg:99.35ms
step:750/4740 val_loss:3.6477 train_time:74587ms step_avg:99.45ms
step:751/4740 train_time:74617ms step_avg:99.36ms
step:752/4740 train_time:74776ms step_avg:99.44ms
step:753/4740 train_time:74878ms step_avg:99.44ms
step:754/4740 train_time:75042ms step_avg:99.53ms
step:755/4740 train_time:75140ms step_avg:99.52ms
step:756/4740 train_time:75298ms step_avg:99.60ms
step:757/4740 train_time:75395ms step_avg:99.60ms
step:758/4740 train_time:75553ms step_avg:99.67ms
step:759/4740 train_time:75652ms step_avg:99.67ms
step:760/4740 train_time:75812ms step_avg:99.75ms
step:761/4740 train_time:75915ms step_avg:99.76ms
step:762/4740 train_time:76076ms step_avg:99.84ms
step:763/4740 train_time:76175ms step_avg:99.84ms
step:764/4740 train_time:76334ms step_avg:99.91ms
step:765/4740 train_time:76432ms step_avg:99.91ms
step:766/4740 train_time:76589ms step_avg:99.99ms
step:767/4740 train_time:76689ms step_avg:99.99ms
step:768/4740 train_time:76848ms step_avg:100.06ms
step:769/4740 train_time:76949ms step_avg:100.06ms
step:770/4740 train_time:77107ms step_avg:100.14ms
step:771/4740 train_time:77208ms step_avg:100.14ms
step:772/4740 train_time:77365ms step_avg:100.21ms
step:773/4740 train_time:77463ms step_avg:100.21ms
step:774/4740 train_time:77621ms step_avg:100.29ms
step:775/4740 train_time:77719ms step_avg:100.28ms
step:776/4740 train_time:77881ms step_avg:100.36ms
step:777/4740 train_time:77980ms step_avg:100.36ms
step:778/4740 train_time:78140ms step_avg:100.44ms
step:779/4740 train_time:78239ms step_avg:100.44ms
step:780/4740 train_time:78399ms step_avg:100.51ms
step:781/4740 train_time:78498ms step_avg:100.51ms
step:782/4740 train_time:78656ms step_avg:100.58ms
step:783/4740 train_time:78755ms step_avg:100.58ms
step:784/4740 train_time:78917ms step_avg:100.66ms
step:785/4740 train_time:79029ms step_avg:100.67ms
step:786/4740 train_time:79265ms step_avg:100.85ms
step:787/4740 train_time:79406ms step_avg:100.90ms
step:788/4740 train_time:79647ms step_avg:101.07ms
step:789/4740 train_time:79785ms step_avg:101.12ms
step:790/4740 train_time:80027ms step_avg:101.30ms
step:791/4740 train_time:80163ms step_avg:101.34ms
step:792/4740 train_time:80402ms step_avg:101.52ms
step:793/4740 train_time:80542ms step_avg:101.57ms
step:794/4740 train_time:80783ms step_avg:101.74ms
step:795/4740 train_time:80920ms step_avg:101.79ms
step:796/4740 train_time:81164ms step_avg:101.96ms
step:797/4740 train_time:81301ms step_avg:102.01ms
step:798/4740 train_time:81539ms step_avg:102.18ms
step:799/4740 train_time:81678ms step_avg:102.22ms
step:800/4740 train_time:81917ms step_avg:102.40ms
step:801/4740 train_time:82056ms step_avg:102.44ms
step:802/4740 train_time:82298ms step_avg:102.62ms
step:803/4740 train_time:82433ms step_avg:102.66ms
step:804/4740 train_time:82673ms step_avg:102.83ms
step:805/4740 train_time:82810ms step_avg:102.87ms
step:806/4740 train_time:83056ms step_avg:103.05ms
step:807/4740 train_time:83191ms step_avg:103.09ms
step:808/4740 train_time:83434ms step_avg:103.26ms
step:809/4740 train_time:83570ms step_avg:103.30ms
step:810/4740 train_time:83814ms step_avg:103.47ms
step:811/4740 train_time:83949ms step_avg:103.51ms
step:812/4740 train_time:84194ms step_avg:103.69ms
step:813/4740 train_time:84329ms step_avg:103.73ms
step:814/4740 train_time:84572ms step_avg:103.90ms
step:815/4740 train_time:84707ms step_avg:103.93ms
step:816/4740 train_time:84950ms step_avg:104.10ms
step:817/4740 train_time:85084ms step_avg:104.14ms
step:818/4740 train_time:85327ms step_avg:104.31ms
step:819/4740 train_time:85463ms step_avg:104.35ms
step:820/4740 train_time:85702ms step_avg:104.51ms
step:821/4740 train_time:85839ms step_avg:104.55ms
step:822/4740 train_time:86078ms step_avg:104.72ms
step:823/4740 train_time:86216ms step_avg:104.76ms
step:824/4740 train_time:86457ms step_avg:104.92ms
step:825/4740 train_time:86593ms step_avg:104.96ms
step:826/4740 train_time:86834ms step_avg:105.13ms
step:827/4740 train_time:86970ms step_avg:105.16ms
step:828/4740 train_time:87213ms step_avg:105.33ms
step:829/4740 train_time:87349ms step_avg:105.37ms
step:830/4740 train_time:87593ms step_avg:105.53ms
step:831/4740 train_time:87727ms step_avg:105.57ms
step:832/4740 train_time:87971ms step_avg:105.73ms
step:833/4740 train_time:88109ms step_avg:105.77ms
step:834/4740 train_time:88352ms step_avg:105.94ms
step:835/4740 train_time:88488ms step_avg:105.97ms
step:836/4740 train_time:88729ms step_avg:106.14ms
step:837/4740 train_time:88867ms step_avg:106.17ms
step:838/4740 train_time:89108ms step_avg:106.33ms
step:839/4740 train_time:89246ms step_avg:106.37ms
step:840/4740 train_time:89489ms step_avg:106.53ms
step:841/4740 train_time:89626ms step_avg:106.57ms
step:842/4740 train_time:89872ms step_avg:106.74ms
step:843/4740 train_time:90008ms step_avg:106.77ms
step:844/4740 train_time:90250ms step_avg:106.93ms
step:845/4740 train_time:90385ms step_avg:106.96ms
step:846/4740 train_time:90626ms step_avg:107.12ms
step:847/4740 train_time:90764ms step_avg:107.16ms
step:848/4740 train_time:91007ms step_avg:107.32ms
step:849/4740 train_time:91145ms step_avg:107.36ms
step:850/4740 train_time:91389ms step_avg:107.52ms
step:851/4740 train_time:91524ms step_avg:107.55ms
step:852/4740 train_time:91769ms step_avg:107.71ms
step:853/4740 train_time:91907ms step_avg:107.75ms
step:854/4740 train_time:92151ms step_avg:107.90ms
step:855/4740 train_time:92286ms step_avg:107.94ms
step:856/4740 train_time:92525ms step_avg:108.09ms
step:857/4740 train_time:92664ms step_avg:108.13ms
step:858/4740 train_time:92907ms step_avg:108.28ms
step:859/4740 train_time:93044ms step_avg:108.32ms
step:860/4740 train_time:93285ms step_avg:108.47ms
step:861/4740 train_time:93424ms step_avg:108.51ms
step:862/4740 train_time:93667ms step_avg:108.66ms
step:863/4740 train_time:93803ms step_avg:108.69ms
step:864/4740 train_time:94044ms step_avg:108.85ms
step:865/4740 train_time:94183ms step_avg:108.88ms
step:866/4740 train_time:94423ms step_avg:109.03ms
step:867/4740 train_time:94559ms step_avg:109.06ms
step:868/4740 train_time:94801ms step_avg:109.22ms
step:869/4740 train_time:94937ms step_avg:109.25ms
step:870/4740 train_time:95177ms step_avg:109.40ms
step:871/4740 train_time:95314ms step_avg:109.43ms
step:872/4740 train_time:95554ms step_avg:109.58ms
step:873/4740 train_time:95688ms step_avg:109.61ms
step:874/4740 train_time:95933ms step_avg:109.76ms
step:875/4740 train_time:96068ms step_avg:109.79ms
step:876/4740 train_time:96310ms step_avg:109.94ms
step:877/4740 train_time:96449ms step_avg:109.98ms
step:878/4740 train_time:96690ms step_avg:110.13ms
step:879/4740 train_time:96825ms step_avg:110.15ms
step:880/4740 train_time:97070ms step_avg:110.31ms
step:881/4740 train_time:97205ms step_avg:110.33ms
step:882/4740 train_time:97444ms step_avg:110.48ms
step:883/4740 train_time:97582ms step_avg:110.51ms
step:884/4740 train_time:97821ms step_avg:110.66ms
step:885/4740 train_time:97962ms step_avg:110.69ms
step:886/4740 train_time:98204ms step_avg:110.84ms
step:887/4740 train_time:98341ms step_avg:110.87ms
step:888/4740 train_time:98578ms step_avg:111.01ms
step:889/4740 train_time:98718ms step_avg:111.04ms
step:890/4740 train_time:98956ms step_avg:111.19ms
step:891/4740 train_time:99093ms step_avg:111.22ms
step:892/4740 train_time:99335ms step_avg:111.36ms
step:893/4740 train_time:99472ms step_avg:111.39ms
step:894/4740 train_time:99713ms step_avg:111.54ms
step:895/4740 train_time:99849ms step_avg:111.56ms
step:896/4740 train_time:100093ms step_avg:111.71ms
step:897/4740 train_time:100228ms step_avg:111.74ms
step:898/4740 train_time:100469ms step_avg:111.88ms
step:899/4740 train_time:100605ms step_avg:111.91ms
step:900/4740 train_time:100846ms step_avg:112.05ms
step:901/4740 train_time:100985ms step_avg:112.08ms
step:902/4740 train_time:101227ms step_avg:112.23ms
step:903/4740 train_time:101364ms step_avg:112.25ms
step:904/4740 train_time:101604ms step_avg:112.39ms
step:905/4740 train_time:101744ms step_avg:112.42ms
step:906/4740 train_time:101984ms step_avg:112.57ms
step:907/4740 train_time:102121ms step_avg:112.59ms
step:908/4740 train_time:102361ms step_avg:112.73ms
step:909/4740 train_time:102499ms step_avg:112.76ms
step:910/4740 train_time:102738ms step_avg:112.90ms
step:911/4740 train_time:102874ms step_avg:112.92ms
step:912/4740 train_time:103114ms step_avg:113.06ms
step:913/4740 train_time:103251ms step_avg:113.09ms
step:914/4740 train_time:103495ms step_avg:113.23ms
step:915/4740 train_time:103630ms step_avg:113.26ms
step:916/4740 train_time:103872ms step_avg:113.40ms
step:917/4740 train_time:104007ms step_avg:113.42ms
step:918/4740 train_time:104248ms step_avg:113.56ms
step:919/4740 train_time:104384ms step_avg:113.58ms
step:920/4740 train_time:104626ms step_avg:113.72ms
step:921/4740 train_time:104770ms step_avg:113.76ms
step:922/4740 train_time:105011ms step_avg:113.89ms
step:923/4740 train_time:105146ms step_avg:113.92ms
step:924/4740 train_time:105388ms step_avg:114.06ms
step:925/4740 train_time:105525ms step_avg:114.08ms
step:926/4740 train_time:105770ms step_avg:114.22ms
step:927/4740 train_time:105906ms step_avg:114.25ms
step:928/4740 train_time:106146ms step_avg:114.38ms
step:929/4740 train_time:106283ms step_avg:114.41ms
step:930/4740 train_time:106527ms step_avg:114.54ms
step:931/4740 train_time:106664ms step_avg:114.57ms
step:932/4740 train_time:106906ms step_avg:114.71ms
step:933/4740 train_time:107043ms step_avg:114.73ms
step:934/4740 train_time:107282ms step_avg:114.86ms
step:935/4740 train_time:107420ms step_avg:114.89ms
step:936/4740 train_time:107660ms step_avg:115.02ms
step:937/4740 train_time:107799ms step_avg:115.05ms
step:938/4740 train_time:108039ms step_avg:115.18ms
step:939/4740 train_time:108177ms step_avg:115.20ms
step:940/4740 train_time:108415ms step_avg:115.34ms
step:941/4740 train_time:108553ms step_avg:115.36ms
step:942/4740 train_time:108795ms step_avg:115.49ms
step:943/4740 train_time:108930ms step_avg:115.51ms
step:944/4740 train_time:109171ms step_avg:115.65ms
step:945/4740 train_time:109308ms step_avg:115.67ms
step:946/4740 train_time:109551ms step_avg:115.80ms
step:947/4740 train_time:109686ms step_avg:115.82ms
step:948/4740 train_time:109931ms step_avg:115.96ms
step:949/4740 train_time:110065ms step_avg:115.98ms
step:950/4740 train_time:110305ms step_avg:116.11ms
step:951/4740 train_time:110442ms step_avg:116.13ms
step:952/4740 train_time:110683ms step_avg:116.26ms
step:953/4740 train_time:110821ms step_avg:116.29ms
step:954/4740 train_time:111063ms step_avg:116.42ms
step:955/4740 train_time:111199ms step_avg:116.44ms
step:956/4740 train_time:111437ms step_avg:116.57ms
step:957/4740 train_time:111574ms step_avg:116.59ms
step:958/4740 train_time:111815ms step_avg:116.72ms
step:959/4740 train_time:111954ms step_avg:116.74ms
step:960/4740 train_time:112195ms step_avg:116.87ms
step:961/4740 train_time:112330ms step_avg:116.89ms
step:962/4740 train_time:112573ms step_avg:117.02ms
step:963/4740 train_time:112708ms step_avg:117.04ms
step:964/4740 train_time:112952ms step_avg:117.17ms
step:965/4740 train_time:113089ms step_avg:117.19ms
step:966/4740 train_time:113332ms step_avg:117.32ms
step:967/4740 train_time:113467ms step_avg:117.34ms
step:968/4740 train_time:113710ms step_avg:117.47ms
step:969/4740 train_time:113847ms step_avg:117.49ms
step:970/4740 train_time:114092ms step_avg:117.62ms
step:971/4740 train_time:114228ms step_avg:117.64ms
step:972/4740 train_time:114468ms step_avg:117.77ms
step:973/4740 train_time:114606ms step_avg:117.79ms
step:974/4740 train_time:114846ms step_avg:117.91ms
step:975/4740 train_time:114984ms step_avg:117.93ms
step:976/4740 train_time:115226ms step_avg:118.06ms
step:977/4740 train_time:115363ms step_avg:118.08ms
step:978/4740 train_time:115604ms step_avg:118.20ms
step:979/4740 train_time:115742ms step_avg:118.22ms
step:980/4740 train_time:115983ms step_avg:118.35ms
step:981/4740 train_time:116122ms step_avg:118.37ms
step:982/4740 train_time:116363ms step_avg:118.50ms
step:983/4740 train_time:116500ms step_avg:118.52ms
step:984/4740 train_time:116741ms step_avg:118.64ms
step:985/4740 train_time:116878ms step_avg:118.66ms
step:986/4740 train_time:117118ms step_avg:118.78ms
step:987/4740 train_time:117255ms step_avg:118.80ms
step:988/4740 train_time:117495ms step_avg:118.92ms
step:989/4740 train_time:117633ms step_avg:118.94ms
step:990/4740 train_time:117873ms step_avg:119.06ms
step:991/4740 train_time:118009ms step_avg:119.08ms
step:992/4740 train_time:118250ms step_avg:119.20ms
step:993/4740 train_time:118386ms step_avg:119.22ms
step:994/4740 train_time:118630ms step_avg:119.35ms
step:995/4740 train_time:118766ms step_avg:119.36ms
step:996/4740 train_time:119008ms step_avg:119.49ms
step:997/4740 train_time:119145ms step_avg:119.50ms
step:998/4740 train_time:119391ms step_avg:119.63ms
step:999/4740 train_time:119527ms step_avg:119.65ms
step:1000/4740 train_time:119769ms step_avg:119.77ms
step:1000/4740 val_loss:3.4782 train_time:119866ms step_avg:119.87ms
step:1001/4740 train_time:119911ms step_avg:119.79ms
step:1002/4740 train_time:120151ms step_avg:119.91ms
step:1003/4740 train_time:120295ms step_avg:119.94ms
step:1004/4740 train_time:120532ms step_avg:120.05ms
step:1005/4740 train_time:120668ms step_avg:120.07ms
step:1006/4740 train_time:120907ms step_avg:120.19ms
step:1007/4740 train_time:121043ms step_avg:120.20ms
step:1008/4740 train_time:121290ms step_avg:120.33ms
step:1009/4740 train_time:121426ms step_avg:120.34ms
step:1010/4740 train_time:121667ms step_avg:120.46ms
step:1011/4740 train_time:121800ms step_avg:120.47ms
step:1012/4740 train_time:122045ms step_avg:120.60ms
step:1013/4740 train_time:122181ms step_avg:120.61ms
step:1014/4740 train_time:122425ms step_avg:120.73ms
step:1015/4740 train_time:122561ms step_avg:120.75ms
step:1016/4740 train_time:122801ms step_avg:120.87ms
step:1017/4740 train_time:122937ms step_avg:120.88ms
step:1018/4740 train_time:123178ms step_avg:121.00ms
step:1019/4740 train_time:123322ms step_avg:121.02ms
step:1020/4740 train_time:123565ms step_avg:121.14ms
step:1021/4740 train_time:123702ms step_avg:121.16ms
step:1022/4740 train_time:123943ms step_avg:121.27ms
step:1023/4740 train_time:124078ms step_avg:121.29ms
step:1024/4740 train_time:124326ms step_avg:121.41ms
step:1025/4740 train_time:124461ms step_avg:121.43ms
step:1026/4740 train_time:124704ms step_avg:121.54ms
step:1027/4740 train_time:124840ms step_avg:121.56ms
step:1028/4740 train_time:125079ms step_avg:121.67ms
step:1029/4740 train_time:125219ms step_avg:121.69ms
step:1030/4740 train_time:125462ms step_avg:121.81ms
step:1031/4740 train_time:125598ms step_avg:121.82ms
step:1032/4740 train_time:125837ms step_avg:121.93ms
step:1033/4740 train_time:125976ms step_avg:121.95ms
step:1034/4740 train_time:126215ms step_avg:122.06ms
step:1035/4740 train_time:126352ms step_avg:122.08ms
step:1036/4740 train_time:126594ms step_avg:122.20ms
step:1037/4740 train_time:126735ms step_avg:122.21ms
step:1038/4740 train_time:126976ms step_avg:122.33ms
step:1039/4740 train_time:127120ms step_avg:122.35ms
step:1040/4740 train_time:127362ms step_avg:122.46ms
step:1041/4740 train_time:127500ms step_avg:122.48ms
step:1042/4740 train_time:127744ms step_avg:122.59ms
step:1043/4740 train_time:127879ms step_avg:122.61ms
step:1044/4740 train_time:128122ms step_avg:122.72ms
step:1045/4740 train_time:128258ms step_avg:122.74ms
step:1046/4740 train_time:128500ms step_avg:122.85ms
step:1047/4740 train_time:128637ms step_avg:122.86ms
step:1048/4740 train_time:128876ms step_avg:122.97ms
step:1049/4740 train_time:129018ms step_avg:122.99ms
step:1050/4740 train_time:129260ms step_avg:123.10ms
step:1051/4740 train_time:129397ms step_avg:123.12ms
step:1052/4740 train_time:129637ms step_avg:123.23ms
step:1053/4740 train_time:129776ms step_avg:123.24ms
step:1054/4740 train_time:130015ms step_avg:123.35ms
step:1055/4740 train_time:130156ms step_avg:123.37ms
step:1056/4740 train_time:130396ms step_avg:123.48ms
step:1057/4740 train_time:130534ms step_avg:123.49ms
step:1058/4740 train_time:130772ms step_avg:123.60ms
step:1059/4740 train_time:130911ms step_avg:123.62ms
step:1060/4740 train_time:131150ms step_avg:123.73ms
step:1061/4740 train_time:131289ms step_avg:123.74ms
step:1062/4740 train_time:131528ms step_avg:123.85ms
step:1063/4740 train_time:131666ms step_avg:123.86ms
step:1064/4740 train_time:131906ms step_avg:123.97ms
step:1065/4740 train_time:132041ms step_avg:123.98ms
step:1066/4740 train_time:132285ms step_avg:124.09ms
step:1067/4740 train_time:132418ms step_avg:124.10ms
step:1068/4740 train_time:132662ms step_avg:124.22ms
step:1069/4740 train_time:132797ms step_avg:124.23ms
step:1070/4740 train_time:133040ms step_avg:124.34ms
step:1071/4740 train_time:133179ms step_avg:124.35ms
step:1072/4740 train_time:133424ms step_avg:124.46ms
step:1073/4740 train_time:133560ms step_avg:124.47ms
step:1074/4740 train_time:133802ms step_avg:124.58ms
step:1075/4740 train_time:133939ms step_avg:124.59ms
step:1076/4740 train_time:134177ms step_avg:124.70ms
step:1077/4740 train_time:134316ms step_avg:124.71ms
step:1078/4740 train_time:134555ms step_avg:124.82ms
step:1079/4740 train_time:134695ms step_avg:124.83ms
step:1080/4740 train_time:134936ms step_avg:124.94ms
step:1081/4740 train_time:135072ms step_avg:124.95ms
step:1082/4740 train_time:135314ms step_avg:125.06ms
step:1083/4740 train_time:135452ms step_avg:125.07ms
step:1084/4740 train_time:135694ms step_avg:125.18ms
step:1085/4740 train_time:135831ms step_avg:125.19ms
step:1086/4740 train_time:136069ms step_avg:125.29ms
step:1087/4740 train_time:136205ms step_avg:125.30ms
step:1088/4740 train_time:136448ms step_avg:125.41ms
step:1089/4740 train_time:136584ms step_avg:125.42ms
step:1090/4740 train_time:136826ms step_avg:125.53ms
step:1091/4740 train_time:136964ms step_avg:125.54ms
step:1092/4740 train_time:137207ms step_avg:125.65ms
step:1093/4740 train_time:137341ms step_avg:125.66ms
step:1094/4740 train_time:137585ms step_avg:125.76ms
step:1095/4740 train_time:137720ms step_avg:125.77ms
step:1096/4740 train_time:137962ms step_avg:125.88ms
step:1097/4740 train_time:138098ms step_avg:125.89ms
step:1098/4740 train_time:138340ms step_avg:125.99ms
step:1099/4740 train_time:138480ms step_avg:126.01ms
step:1100/4740 train_time:138725ms step_avg:126.11ms
step:1101/4740 train_time:138861ms step_avg:126.12ms
step:1102/4740 train_time:139105ms step_avg:126.23ms
step:1103/4740 train_time:139242ms step_avg:126.24ms
step:1104/4740 train_time:139486ms step_avg:126.35ms
step:1105/4740 train_time:139620ms step_avg:126.35ms
step:1106/4740 train_time:139865ms step_avg:126.46ms
step:1107/4740 train_time:140001ms step_avg:126.47ms
step:1108/4740 train_time:140246ms step_avg:126.58ms
step:1109/4740 train_time:140381ms step_avg:126.58ms
step:1110/4740 train_time:140625ms step_avg:126.69ms
step:1111/4740 train_time:140760ms step_avg:126.70ms
step:1112/4740 train_time:141005ms step_avg:126.80ms
step:1113/4740 train_time:141140ms step_avg:126.81ms
step:1114/4740 train_time:141382ms step_avg:126.91ms
step:1115/4740 train_time:141518ms step_avg:126.92ms
step:1116/4740 train_time:141758ms step_avg:127.02ms
step:1117/4740 train_time:141897ms step_avg:127.03ms
step:1118/4740 train_time:142136ms step_avg:127.13ms
step:1119/4740 train_time:142273ms step_avg:127.14ms
step:1120/4740 train_time:142514ms step_avg:127.25ms
step:1121/4740 train_time:142654ms step_avg:127.26ms
step:1122/4740 train_time:142893ms step_avg:127.36ms
step:1123/4740 train_time:143030ms step_avg:127.36ms
step:1124/4740 train_time:143269ms step_avg:127.46ms
step:1125/4740 train_time:143406ms step_avg:127.47ms
step:1126/4740 train_time:143648ms step_avg:127.57ms
step:1127/4740 train_time:143784ms step_avg:127.58ms
step:1128/4740 train_time:144026ms step_avg:127.68ms
step:1129/4740 train_time:144164ms step_avg:127.69ms
step:1130/4740 train_time:144406ms step_avg:127.79ms
step:1131/4740 train_time:144540ms step_avg:127.80ms
step:1132/4740 train_time:144785ms step_avg:127.90ms
step:1133/4740 train_time:144922ms step_avg:127.91ms
step:1134/4740 train_time:145167ms step_avg:128.01ms
step:1135/4740 train_time:145302ms step_avg:128.02ms
step:1136/4740 train_time:145546ms step_avg:128.12ms
step:1137/4740 train_time:145680ms step_avg:128.13ms
step:1138/4740 train_time:145923ms step_avg:128.23ms
step:1139/4740 train_time:146059ms step_avg:128.23ms
step:1140/4740 train_time:146301ms step_avg:128.33ms
step:1141/4740 train_time:146437ms step_avg:128.34ms
step:1142/4740 train_time:146679ms step_avg:128.44ms
step:1143/4740 train_time:146818ms step_avg:128.45ms
step:1144/4740 train_time:147059ms step_avg:128.55ms
step:1145/4740 train_time:147197ms step_avg:128.56ms
step:1146/4740 train_time:147435ms step_avg:128.65ms
step:1147/4740 train_time:147575ms step_avg:128.66ms
step:1148/4740 train_time:147816ms step_avg:128.76ms
step:1149/4740 train_time:147954ms step_avg:128.77ms
step:1150/4740 train_time:148192ms step_avg:128.86ms
step:1151/4740 train_time:148329ms step_avg:128.87ms
step:1152/4740 train_time:148568ms step_avg:128.97ms
step:1153/4740 train_time:148706ms step_avg:128.97ms
step:1154/4740 train_time:148948ms step_avg:129.07ms
step:1155/4740 train_time:149085ms step_avg:129.08ms
step:1156/4740 train_time:149326ms step_avg:129.17ms
step:1157/4740 train_time:149462ms step_avg:129.18ms
step:1158/4740 train_time:149704ms step_avg:129.28ms
step:1159/4740 train_time:149839ms step_avg:129.28ms
step:1160/4740 train_time:150085ms step_avg:129.38ms
step:1161/4740 train_time:150220ms step_avg:129.39ms
step:1162/4740 train_time:150464ms step_avg:129.49ms
step:1163/4740 train_time:150601ms step_avg:129.49ms
step:1164/4740 train_time:150844ms step_avg:129.59ms
step:1165/4740 train_time:150979ms step_avg:129.60ms
step:1166/4740 train_time:151223ms step_avg:129.69ms
step:1167/4740 train_time:151359ms step_avg:129.70ms
step:1168/4740 train_time:151602ms step_avg:129.80ms
step:1169/4740 train_time:151737ms step_avg:129.80ms
step:1170/4740 train_time:151978ms step_avg:129.90ms
step:1171/4740 train_time:152116ms step_avg:129.90ms
step:1172/4740 train_time:152356ms step_avg:130.00ms
step:1173/4740 train_time:152494ms step_avg:130.00ms
step:1174/4740 train_time:152734ms step_avg:130.10ms
step:1175/4740 train_time:152871ms step_avg:130.10ms
step:1176/4740 train_time:153143ms step_avg:130.22ms
step:1177/4740 train_time:153318ms step_avg:130.26ms
step:1178/4740 train_time:153637ms step_avg:130.42ms
step:1179/4740 train_time:153813ms step_avg:130.46ms
step:1180/4740 train_time:154128ms step_avg:130.62ms
step:1181/4740 train_time:154304ms step_avg:130.66ms
step:1182/4740 train_time:154621ms step_avg:130.81ms
step:1183/4740 train_time:154798ms step_avg:130.85ms
step:1184/4740 train_time:155117ms step_avg:131.01ms
step:1185/4740 train_time:155293ms step_avg:131.05ms
step:1186/4740 train_time:155610ms step_avg:131.21ms
step:1187/4740 train_time:155785ms step_avg:131.24ms
step:1188/4740 train_time:156103ms step_avg:131.40ms
step:1189/4740 train_time:156279ms step_avg:131.44ms
step:1190/4740 train_time:156598ms step_avg:131.60ms
step:1191/4740 train_time:156774ms step_avg:131.63ms
step:1192/4740 train_time:157091ms step_avg:131.79ms
step:1193/4740 train_time:157268ms step_avg:131.83ms
step:1194/4740 train_time:157584ms step_avg:131.98ms
step:1195/4740 train_time:157759ms step_avg:132.02ms
step:1196/4740 train_time:158077ms step_avg:132.17ms
step:1197/4740 train_time:158251ms step_avg:132.21ms
step:1198/4740 train_time:158567ms step_avg:132.36ms
step:1199/4740 train_time:158743ms step_avg:132.40ms
step:1200/4740 train_time:159062ms step_avg:132.55ms
step:1201/4740 train_time:159234ms step_avg:132.58ms
step:1202/4740 train_time:159552ms step_avg:132.74ms
step:1203/4740 train_time:159729ms step_avg:132.78ms
step:1204/4740 train_time:160042ms step_avg:132.93ms
step:1205/4740 train_time:160219ms step_avg:132.96ms
step:1206/4740 train_time:160537ms step_avg:133.12ms
step:1207/4740 train_time:160712ms step_avg:133.15ms
step:1208/4740 train_time:161027ms step_avg:133.30ms
step:1209/4740 train_time:161202ms step_avg:133.33ms
step:1210/4740 train_time:161520ms step_avg:133.49ms
step:1211/4740 train_time:161696ms step_avg:133.52ms
step:1212/4740 train_time:162013ms step_avg:133.67ms
step:1213/4740 train_time:162189ms step_avg:133.71ms
step:1214/4740 train_time:162506ms step_avg:133.86ms
step:1215/4740 train_time:162681ms step_avg:133.89ms
step:1216/4740 train_time:162998ms step_avg:134.04ms
step:1217/4740 train_time:163175ms step_avg:134.08ms
step:1218/4740 train_time:163490ms step_avg:134.23ms
step:1219/4740 train_time:163667ms step_avg:134.26ms
step:1220/4740 train_time:163986ms step_avg:134.41ms
step:1221/4740 train_time:164161ms step_avg:134.45ms
step:1222/4740 train_time:164481ms step_avg:134.60ms
step:1223/4740 train_time:164656ms step_avg:134.63ms
step:1224/4740 train_time:164974ms step_avg:134.78ms
step:1225/4740 train_time:165148ms step_avg:134.81ms
step:1226/4740 train_time:165465ms step_avg:134.96ms
step:1227/4740 train_time:165643ms step_avg:135.00ms
step:1228/4740 train_time:165961ms step_avg:135.15ms
step:1229/4740 train_time:166135ms step_avg:135.18ms
step:1230/4740 train_time:166454ms step_avg:135.33ms
step:1231/4740 train_time:166629ms step_avg:135.36ms
step:1232/4740 train_time:166943ms step_avg:135.51ms
step:1233/4740 train_time:167123ms step_avg:135.54ms
step:1234/4740 train_time:167439ms step_avg:135.69ms
step:1235/4740 train_time:167615ms step_avg:135.72ms
step:1236/4740 train_time:167929ms step_avg:135.86ms
step:1237/4740 train_time:168107ms step_avg:135.90ms
step:1238/4740 train_time:168423ms step_avg:136.04ms
step:1239/4740 train_time:168599ms step_avg:136.08ms
step:1240/4740 train_time:168919ms step_avg:136.22ms
step:1241/4740 train_time:169095ms step_avg:136.26ms
step:1242/4740 train_time:169414ms step_avg:136.40ms
step:1243/4740 train_time:169589ms step_avg:136.44ms
step:1244/4740 train_time:169905ms step_avg:136.58ms
step:1245/4740 train_time:170080ms step_avg:136.61ms
step:1246/4740 train_time:170400ms step_avg:136.76ms
step:1247/4740 train_time:170576ms step_avg:136.79ms
step:1248/4740 train_time:170892ms step_avg:136.93ms
step:1249/4740 train_time:171071ms step_avg:136.97ms
step:1250/4740 train_time:171387ms step_avg:137.11ms
step:1250/4740 val_loss:3.4030 train_time:171510ms step_avg:137.21ms
step:1251/4740 train_time:171567ms step_avg:137.14ms
step:1252/4740 train_time:171890ms step_avg:137.29ms
step:1253/4740 train_time:172066ms step_avg:137.32ms
step:1254/4740 train_time:172379ms step_avg:137.46ms
step:1255/4740 train_time:172553ms step_avg:137.49ms
step:1256/4740 train_time:172875ms step_avg:137.64ms
step:1257/4740 train_time:173054ms step_avg:137.67ms
step:1258/4740 train_time:173366ms step_avg:137.81ms
step:1259/4740 train_time:173541ms step_avg:137.84ms
step:1260/4740 train_time:173864ms step_avg:137.99ms
step:1261/4740 train_time:174040ms step_avg:138.02ms
step:1262/4740 train_time:174355ms step_avg:138.16ms
step:1263/4740 train_time:174533ms step_avg:138.19ms
step:1264/4740 train_time:174851ms step_avg:138.33ms
step:1265/4740 train_time:175026ms step_avg:138.36ms
step:1266/4740 train_time:175343ms step_avg:138.50ms
step:1267/4740 train_time:175521ms step_avg:138.53ms
step:1268/4740 train_time:175840ms step_avg:138.67ms
step:1269/4740 train_time:176014ms step_avg:138.70ms
step:1270/4740 train_time:176330ms step_avg:138.84ms
step:1271/4740 train_time:176506ms step_avg:138.87ms
step:1272/4740 train_time:176825ms step_avg:139.01ms
step:1273/4740 train_time:177000ms step_avg:139.04ms
step:1274/4740 train_time:177321ms step_avg:139.18ms
step:1275/4740 train_time:177496ms step_avg:139.21ms
step:1276/4740 train_time:177812ms step_avg:139.35ms
step:1277/4740 train_time:177987ms step_avg:139.38ms
step:1278/4740 train_time:178306ms step_avg:139.52ms
step:1279/4740 train_time:178479ms step_avg:139.55ms
step:1280/4740 train_time:178796ms step_avg:139.68ms
step:1281/4740 train_time:178974ms step_avg:139.71ms
step:1282/4740 train_time:179291ms step_avg:139.85ms
step:1283/4740 train_time:179469ms step_avg:139.88ms
step:1284/4740 train_time:179784ms step_avg:140.02ms
step:1285/4740 train_time:179961ms step_avg:140.05ms
step:1286/4740 train_time:180277ms step_avg:140.18ms
step:1287/4740 train_time:180453ms step_avg:140.21ms
step:1288/4740 train_time:180769ms step_avg:140.35ms
step:1289/4740 train_time:180943ms step_avg:140.37ms
step:1290/4740 train_time:181262ms step_avg:140.51ms
step:1291/4740 train_time:181438ms step_avg:140.54ms
step:1292/4740 train_time:181753ms step_avg:140.68ms
step:1293/4740 train_time:181930ms step_avg:140.70ms
step:1294/4740 train_time:182247ms step_avg:140.84ms
step:1295/4740 train_time:182424ms step_avg:140.87ms
step:1296/4740 train_time:182743ms step_avg:141.01ms
step:1297/4740 train_time:182920ms step_avg:141.03ms
step:1298/4740 train_time:183236ms step_avg:141.17ms
step:1299/4740 train_time:183413ms step_avg:141.20ms
step:1300/4740 train_time:183732ms step_avg:141.33ms
step:1301/4740 train_time:183908ms step_avg:141.36ms
step:1302/4740 train_time:184227ms step_avg:141.50ms
step:1303/4740 train_time:184402ms step_avg:141.52ms
step:1304/4740 train_time:184724ms step_avg:141.66ms
step:1305/4740 train_time:184898ms step_avg:141.68ms
step:1306/4740 train_time:185216ms step_avg:141.82ms
step:1307/4740 train_time:185393ms step_avg:141.85ms
step:1308/4740 train_time:185710ms step_avg:141.98ms
step:1309/4740 train_time:185886ms step_avg:142.01ms
step:1310/4740 train_time:186206ms step_avg:142.14ms
step:1311/4740 train_time:186380ms step_avg:142.17ms
step:1312/4740 train_time:186703ms step_avg:142.30ms
step:1313/4740 train_time:186879ms step_avg:142.33ms
step:1314/4740 train_time:187196ms step_avg:142.46ms
step:1315/4740 train_time:187373ms step_avg:142.49ms
step:1316/4740 train_time:187686ms step_avg:142.62ms
step:1317/4740 train_time:187864ms step_avg:142.65ms
step:1318/4740 train_time:188181ms step_avg:142.78ms
step:1319/4740 train_time:188355ms step_avg:142.80ms
step:1320/4740 train_time:188673ms step_avg:142.93ms
step:1321/4740 train_time:188848ms step_avg:142.96ms
step:1322/4740 train_time:189163ms step_avg:143.09ms
step:1323/4740 train_time:189338ms step_avg:143.11ms
step:1324/4740 train_time:189653ms step_avg:143.24ms
step:1325/4740 train_time:189833ms step_avg:143.27ms
step:1326/4740 train_time:190148ms step_avg:143.40ms
step:1327/4740 train_time:190330ms step_avg:143.43ms
step:1328/4740 train_time:190642ms step_avg:143.56ms
step:1329/4740 train_time:190824ms step_avg:143.58ms
step:1330/4740 train_time:191141ms step_avg:143.72ms
step:1331/4740 train_time:191317ms step_avg:143.74ms
step:1332/4740 train_time:191633ms step_avg:143.87ms
step:1333/4740 train_time:191810ms step_avg:143.89ms
step:1334/4740 train_time:192125ms step_avg:144.02ms
step:1335/4740 train_time:192304ms step_avg:144.05ms
step:1336/4740 train_time:192620ms step_avg:144.18ms
step:1337/4740 train_time:192796ms step_avg:144.20ms
step:1338/4740 train_time:193112ms step_avg:144.33ms
step:1339/4740 train_time:193291ms step_avg:144.35ms
step:1340/4740 train_time:193606ms step_avg:144.48ms
step:1341/4740 train_time:193781ms step_avg:144.50ms
step:1342/4740 train_time:194102ms step_avg:144.64ms
step:1343/4740 train_time:194277ms step_avg:144.66ms
step:1344/4740 train_time:194592ms step_avg:144.79ms
step:1345/4740 train_time:194769ms step_avg:144.81ms
step:1346/4740 train_time:195085ms step_avg:144.94ms
step:1347/4740 train_time:195261ms step_avg:144.96ms
step:1348/4740 train_time:195579ms step_avg:145.09ms
step:1349/4740 train_time:195757ms step_avg:145.11ms
step:1350/4740 train_time:196070ms step_avg:145.24ms
step:1351/4740 train_time:196248ms step_avg:145.26ms
step:1352/4740 train_time:196565ms step_avg:145.39ms
step:1353/4740 train_time:196743ms step_avg:145.41ms
step:1354/4740 train_time:197063ms step_avg:145.54ms
step:1355/4740 train_time:197237ms step_avg:145.56ms
step:1356/4740 train_time:197555ms step_avg:145.69ms
step:1357/4740 train_time:197735ms step_avg:145.71ms
step:1358/4740 train_time:198050ms step_avg:145.84ms
step:1359/4740 train_time:198225ms step_avg:145.86ms
step:1360/4740 train_time:198541ms step_avg:145.99ms
step:1361/4740 train_time:198720ms step_avg:146.01ms
step:1362/4740 train_time:199035ms step_avg:146.13ms
step:1363/4740 train_time:199210ms step_avg:146.16ms
step:1364/4740 train_time:199526ms step_avg:146.28ms
step:1365/4740 train_time:199706ms step_avg:146.30ms
step:1366/4740 train_time:200023ms step_avg:146.43ms
step:1367/4740 train_time:200198ms step_avg:146.45ms
step:1368/4740 train_time:200516ms step_avg:146.58ms
step:1369/4740 train_time:200694ms step_avg:146.60ms
step:1370/4740 train_time:201009ms step_avg:146.72ms
step:1371/4740 train_time:201187ms step_avg:146.74ms
step:1372/4740 train_time:201506ms step_avg:146.87ms
step:1373/4740 train_time:201680ms step_avg:146.89ms
step:1374/4740 train_time:202002ms step_avg:147.02ms
step:1375/4740 train_time:202175ms step_avg:147.04ms
step:1376/4740 train_time:202491ms step_avg:147.16ms
step:1377/4740 train_time:202667ms step_avg:147.18ms
step:1378/4740 train_time:202986ms step_avg:147.30ms
step:1379/4740 train_time:203160ms step_avg:147.32ms
step:1380/4740 train_time:203479ms step_avg:147.45ms
step:1381/4740 train_time:203656ms step_avg:147.47ms
step:1382/4740 train_time:203972ms step_avg:147.59ms
step:1383/4740 train_time:204148ms step_avg:147.61ms
step:1384/4740 train_time:204465ms step_avg:147.74ms
step:1385/4740 train_time:204642ms step_avg:147.76ms
step:1386/4740 train_time:204960ms step_avg:147.88ms
step:1387/4740 train_time:205134ms step_avg:147.90ms
step:1388/4740 train_time:205451ms step_avg:148.02ms
step:1389/4740 train_time:205627ms step_avg:148.04ms
step:1390/4740 train_time:205945ms step_avg:148.16ms
step:1391/4740 train_time:206121ms step_avg:148.18ms
step:1392/4740 train_time:206440ms step_avg:148.30ms
step:1393/4740 train_time:206615ms step_avg:148.32ms
step:1394/4740 train_time:206931ms step_avg:148.44ms
step:1395/4740 train_time:207105ms step_avg:148.46ms
step:1396/4740 train_time:207425ms step_avg:148.58ms
step:1397/4740 train_time:207600ms step_avg:148.60ms
step:1398/4740 train_time:207918ms step_avg:148.73ms
step:1399/4740 train_time:208093ms step_avg:148.74ms
step:1400/4740 train_time:208408ms step_avg:148.86ms
step:1401/4740 train_time:208584ms step_avg:148.88ms
step:1402/4740 train_time:208903ms step_avg:149.00ms
step:1403/4740 train_time:209079ms step_avg:149.02ms
step:1404/4740 train_time:209395ms step_avg:149.14ms
step:1405/4740 train_time:209571ms step_avg:149.16ms
step:1406/4740 train_time:209885ms step_avg:149.28ms
step:1407/4740 train_time:210063ms step_avg:149.30ms
step:1408/4740 train_time:210381ms step_avg:149.42ms
step:1409/4740 train_time:210557ms step_avg:149.44ms
step:1410/4740 train_time:210871ms step_avg:149.55ms
step:1411/4740 train_time:211048ms step_avg:149.57ms
step:1412/4740 train_time:211363ms step_avg:149.69ms
step:1413/4740 train_time:211542ms step_avg:149.71ms
step:1414/4740 train_time:211860ms step_avg:149.83ms
step:1415/4740 train_time:212037ms step_avg:149.85ms
step:1416/4740 train_time:212349ms step_avg:149.96ms
step:1417/4740 train_time:212530ms step_avg:149.99ms
step:1418/4740 train_time:212845ms step_avg:150.10ms
step:1419/4740 train_time:213025ms step_avg:150.12ms
step:1420/4740 train_time:213340ms step_avg:150.24ms
step:1421/4740 train_time:213516ms step_avg:150.26ms
step:1422/4740 train_time:213831ms step_avg:150.37ms
step:1423/4740 train_time:214009ms step_avg:150.39ms
step:1424/4740 train_time:214325ms step_avg:150.51ms
step:1425/4740 train_time:214501ms step_avg:150.53ms
step:1426/4740 train_time:214820ms step_avg:150.65ms
step:1427/4740 train_time:214998ms step_avg:150.66ms
step:1428/4740 train_time:215312ms step_avg:150.78ms
step:1429/4740 train_time:215488ms step_avg:150.80ms
step:1430/4740 train_time:215804ms step_avg:150.91ms
step:1431/4740 train_time:215980ms step_avg:150.93ms
step:1432/4740 train_time:216301ms step_avg:151.05ms
step:1433/4740 train_time:216477ms step_avg:151.07ms
step:1434/4740 train_time:216792ms step_avg:151.18ms
step:1435/4740 train_time:216970ms step_avg:151.20ms
step:1436/4740 train_time:217285ms step_avg:151.31ms
step:1437/4740 train_time:217460ms step_avg:151.33ms
step:1438/4740 train_time:217781ms step_avg:151.45ms
step:1439/4740 train_time:217955ms step_avg:151.46ms
step:1440/4740 train_time:218272ms step_avg:151.58ms
step:1441/4740 train_time:218448ms step_avg:151.59ms
step:1442/4740 train_time:218767ms step_avg:151.71ms
step:1443/4740 train_time:218941ms step_avg:151.73ms
step:1444/4740 train_time:219260ms step_avg:151.84ms
step:1445/4740 train_time:219438ms step_avg:151.86ms
step:1446/4740 train_time:219755ms step_avg:151.97ms
step:1447/4740 train_time:219932ms step_avg:151.99ms
step:1448/4740 train_time:220247ms step_avg:152.10ms
step:1449/4740 train_time:220425ms step_avg:152.12ms
step:1450/4740 train_time:220741ms step_avg:152.23ms
step:1451/4740 train_time:220919ms step_avg:152.25ms
step:1452/4740 train_time:221236ms step_avg:152.37ms
step:1453/4740 train_time:221411ms step_avg:152.38ms
step:1454/4740 train_time:221729ms step_avg:152.50ms
step:1455/4740 train_time:221903ms step_avg:152.51ms
step:1456/4740 train_time:222222ms step_avg:152.62ms
step:1457/4740 train_time:222399ms step_avg:152.64ms
step:1458/4740 train_time:222715ms step_avg:152.75ms
step:1459/4740 train_time:222891ms step_avg:152.77ms
step:1460/4740 train_time:223206ms step_avg:152.88ms
step:1461/4740 train_time:223382ms step_avg:152.90ms
step:1462/4740 train_time:223702ms step_avg:153.01ms
step:1463/4740 train_time:223878ms step_avg:153.03ms
step:1464/4740 train_time:224195ms step_avg:153.14ms
step:1465/4740 train_time:224374ms step_avg:153.16ms
step:1466/4740 train_time:224688ms step_avg:153.27ms
step:1467/4740 train_time:224865ms step_avg:153.28ms
step:1468/4740 train_time:225182ms step_avg:153.39ms
step:1469/4740 train_time:225359ms step_avg:153.41ms
step:1470/4740 train_time:225675ms step_avg:153.52ms
step:1471/4740 train_time:225853ms step_avg:153.54ms
step:1472/4740 train_time:226167ms step_avg:153.65ms
step:1473/4740 train_time:226344ms step_avg:153.66ms
step:1474/4740 train_time:226662ms step_avg:153.77ms
step:1475/4740 train_time:226839ms step_avg:153.79ms
step:1476/4740 train_time:227155ms step_avg:153.90ms
step:1477/4740 train_time:227331ms step_avg:153.91ms
step:1478/4740 train_time:227648ms step_avg:154.02ms
step:1479/4740 train_time:227824ms step_avg:154.04ms
step:1480/4740 train_time:228141ms step_avg:154.15ms
step:1481/4740 train_time:228316ms step_avg:154.16ms
step:1482/4740 train_time:228631ms step_avg:154.27ms
step:1483/4740 train_time:228805ms step_avg:154.29ms
step:1484/4740 train_time:229123ms step_avg:154.40ms
step:1485/4740 train_time:229298ms step_avg:154.41ms
step:1486/4740 train_time:229615ms step_avg:154.52ms
step:1487/4740 train_time:229791ms step_avg:154.53ms
step:1488/4740 train_time:230107ms step_avg:154.64ms
step:1489/4740 train_time:230282ms step_avg:154.66ms
step:1490/4740 train_time:230603ms step_avg:154.77ms
step:1491/4740 train_time:230777ms step_avg:154.78ms
step:1492/4740 train_time:231090ms step_avg:154.89ms
step:1493/4740 train_time:231265ms step_avg:154.90ms
step:1494/4740 train_time:231584ms step_avg:155.01ms
step:1495/4740 train_time:231762ms step_avg:155.03ms
step:1496/4740 train_time:232080ms step_avg:155.13ms
step:1497/4740 train_time:232255ms step_avg:155.15ms
step:1498/4740 train_time:232570ms step_avg:155.25ms
step:1499/4740 train_time:232748ms step_avg:155.27ms
step:1500/4740 train_time:233064ms step_avg:155.38ms
step:1500/4740 val_loss:3.3422 train_time:233187ms step_avg:155.46ms
step:1501/4740 train_time:233244ms step_avg:155.39ms
step:1502/4740 train_time:233570ms step_avg:155.51ms
step:1503/4740 train_time:233742ms step_avg:155.52ms
step:1504/4740 train_time:234057ms step_avg:155.62ms
step:1505/4740 train_time:234230ms step_avg:155.63ms
step:1506/4740 train_time:234554ms step_avg:155.75ms
step:1507/4740 train_time:234729ms step_avg:155.76ms
step:1508/4740 train_time:235046ms step_avg:155.87ms
step:1509/4740 train_time:235217ms step_avg:155.88ms
step:1510/4740 train_time:235543ms step_avg:155.99ms
step:1511/4740 train_time:235719ms step_avg:156.00ms
step:1512/4740 train_time:236035ms step_avg:156.11ms
step:1513/4740 train_time:236210ms step_avg:156.12ms
step:1514/4740 train_time:236528ms step_avg:156.23ms
step:1515/4740 train_time:236706ms step_avg:156.24ms
step:1516/4740 train_time:237022ms step_avg:156.35ms
step:1517/4740 train_time:237194ms step_avg:156.36ms
step:1518/4740 train_time:237509ms step_avg:156.46ms
step:1519/4740 train_time:237691ms step_avg:156.48ms
step:1520/4740 train_time:238004ms step_avg:156.58ms
step:1521/4740 train_time:238179ms step_avg:156.59ms
step:1522/4740 train_time:238495ms step_avg:156.70ms
step:1523/4740 train_time:238672ms step_avg:156.71ms
step:1524/4740 train_time:238989ms step_avg:156.82ms
step:1525/4740 train_time:239164ms step_avg:156.83ms
step:1526/4740 train_time:239483ms step_avg:156.94ms
step:1527/4740 train_time:239658ms step_avg:156.95ms
step:1528/4740 train_time:239974ms step_avg:157.05ms
step:1529/4740 train_time:240151ms step_avg:157.06ms
step:1530/4740 train_time:240467ms step_avg:157.17ms
step:1531/4740 train_time:240642ms step_avg:157.18ms
step:1532/4740 train_time:240962ms step_avg:157.29ms
step:1533/4740 train_time:241136ms step_avg:157.30ms
step:1534/4740 train_time:241453ms step_avg:157.40ms
step:1535/4740 train_time:241627ms step_avg:157.41ms
step:1536/4740 train_time:241946ms step_avg:157.52ms
step:1537/4740 train_time:242121ms step_avg:157.53ms
step:1538/4740 train_time:242441ms step_avg:157.63ms
step:1539/4740 train_time:242617ms step_avg:157.65ms
step:1540/4740 train_time:242934ms step_avg:157.75ms
step:1541/4740 train_time:243112ms step_avg:157.76ms
step:1542/4740 train_time:243429ms step_avg:157.87ms
step:1543/4740 train_time:243604ms step_avg:157.88ms
step:1544/4740 train_time:243922ms step_avg:157.98ms
step:1545/4740 train_time:244098ms step_avg:157.99ms
step:1546/4740 train_time:244415ms step_avg:158.10ms
step:1547/4740 train_time:244595ms step_avg:158.11ms
step:1548/4740 train_time:244911ms step_avg:158.21ms
step:1549/4740 train_time:245086ms step_avg:158.22ms
step:1550/4740 train_time:245403ms step_avg:158.32ms
step:1551/4740 train_time:245580ms step_avg:158.34ms
step:1552/4740 train_time:245899ms step_avg:158.44ms
step:1553/4740 train_time:246075ms step_avg:158.45ms
step:1554/4740 train_time:246392ms step_avg:158.55ms
step:1555/4740 train_time:246567ms step_avg:158.56ms
step:1556/4740 train_time:246885ms step_avg:158.67ms
step:1557/4740 train_time:247059ms step_avg:158.68ms
step:1558/4740 train_time:247374ms step_avg:158.78ms
step:1559/4740 train_time:247551ms step_avg:158.79ms
step:1560/4740 train_time:247868ms step_avg:158.89ms
step:1561/4740 train_time:248042ms step_avg:158.90ms
step:1562/4740 train_time:248360ms step_avg:159.00ms
step:1563/4740 train_time:248537ms step_avg:159.01ms
step:1564/4740 train_time:248853ms step_avg:159.11ms
step:1565/4740 train_time:249028ms step_avg:159.12ms
step:1566/4740 train_time:249348ms step_avg:159.23ms
step:1567/4740 train_time:249524ms step_avg:159.24ms
step:1568/4740 train_time:249845ms step_avg:159.34ms
step:1569/4740 train_time:250017ms step_avg:159.35ms
step:1570/4740 train_time:250337ms step_avg:159.45ms
step:1571/4740 train_time:250515ms step_avg:159.46ms
step:1572/4740 train_time:250833ms step_avg:159.56ms
step:1573/4740 train_time:251010ms step_avg:159.57ms
step:1574/4740 train_time:251326ms step_avg:159.67ms
step:1575/4740 train_time:251504ms step_avg:159.68ms
step:1576/4740 train_time:251822ms step_avg:159.79ms
step:1577/4740 train_time:251999ms step_avg:159.80ms
step:1578/4740 train_time:252318ms step_avg:159.90ms
step:1579/4740 train_time:252495ms step_avg:159.91ms
step:1580/4740 train_time:252813ms step_avg:160.01ms
step:1581/4740 train_time:252990ms step_avg:160.02ms
step:1582/4740 train_time:253309ms step_avg:160.12ms
step:1583/4740 train_time:253484ms step_avg:160.13ms
step:1584/4740 train_time:253803ms step_avg:160.23ms
step:1585/4740 train_time:253979ms step_avg:160.24ms
step:1586/4740 train_time:254302ms step_avg:160.34ms
step:1587/4740 train_time:254475ms step_avg:160.35ms
step:1588/4740 train_time:254790ms step_avg:160.45ms
step:1589/4740 train_time:254965ms step_avg:160.46ms
step:1590/4740 train_time:255285ms step_avg:160.56ms
step:1591/4740 train_time:255460ms step_avg:160.57ms
step:1592/4740 train_time:255780ms step_avg:160.67ms
step:1593/4740 train_time:255955ms step_avg:160.67ms
step:1594/4740 train_time:256275ms step_avg:160.77ms
step:1595/4740 train_time:256451ms step_avg:160.78ms
step:1596/4740 train_time:256769ms step_avg:160.88ms
step:1597/4740 train_time:256944ms step_avg:160.89ms
step:1598/4740 train_time:257263ms step_avg:160.99ms
step:1599/4740 train_time:257438ms step_avg:161.00ms
step:1600/4740 train_time:257757ms step_avg:161.10ms
step:1601/4740 train_time:257933ms step_avg:161.11ms
step:1602/4740 train_time:258252ms step_avg:161.21ms
step:1603/4740 train_time:258431ms step_avg:161.22ms
step:1604/4740 train_time:258752ms step_avg:161.32ms
step:1605/4740 train_time:258926ms step_avg:161.32ms
step:1606/4740 train_time:259245ms step_avg:161.42ms
step:1607/4740 train_time:259421ms step_avg:161.43ms
step:1608/4740 train_time:259741ms step_avg:161.53ms
step:1609/4740 train_time:259919ms step_avg:161.54ms
step:1610/4740 train_time:260234ms step_avg:161.64ms
step:1611/4740 train_time:260415ms step_avg:161.65ms
step:1612/4740 train_time:260730ms step_avg:161.74ms
step:1613/4740 train_time:260911ms step_avg:161.76ms
step:1614/4740 train_time:261222ms step_avg:161.85ms
step:1615/4740 train_time:261404ms step_avg:161.86ms
step:1616/4740 train_time:261721ms step_avg:161.96ms
step:1617/4740 train_time:261901ms step_avg:161.97ms
step:1618/4740 train_time:262217ms step_avg:162.06ms
step:1619/4740 train_time:262393ms step_avg:162.07ms
step:1620/4740 train_time:262710ms step_avg:162.17ms
step:1621/4740 train_time:262890ms step_avg:162.18ms
step:1622/4740 train_time:263207ms step_avg:162.27ms
step:1623/4740 train_time:263388ms step_avg:162.28ms
step:1624/4740 train_time:263704ms step_avg:162.38ms
step:1625/4740 train_time:263883ms step_avg:162.39ms
step:1626/4740 train_time:264204ms step_avg:162.49ms
step:1627/4740 train_time:264378ms step_avg:162.49ms
step:1628/4740 train_time:264695ms step_avg:162.59ms
step:1629/4740 train_time:264872ms step_avg:162.60ms
step:1630/4740 train_time:265190ms step_avg:162.69ms
step:1631/4740 train_time:265369ms step_avg:162.70ms
step:1632/4740 train_time:265686ms step_avg:162.80ms
step:1633/4740 train_time:265862ms step_avg:162.81ms
step:1634/4740 train_time:266183ms step_avg:162.90ms
step:1635/4740 train_time:266360ms step_avg:162.91ms
step:1636/4740 train_time:266678ms step_avg:163.01ms
step:1637/4740 train_time:266855ms step_avg:163.01ms
step:1638/4740 train_time:267171ms step_avg:163.11ms
step:1639/4740 train_time:267349ms step_avg:163.12ms
step:1640/4740 train_time:267668ms step_avg:163.21ms
step:1641/4740 train_time:267843ms step_avg:163.22ms
step:1642/4740 train_time:268164ms step_avg:163.32ms
step:1643/4740 train_time:268342ms step_avg:163.32ms
step:1644/4740 train_time:268661ms step_avg:163.42ms
step:1645/4740 train_time:268836ms step_avg:163.43ms
step:1646/4740 train_time:269154ms step_avg:163.52ms
step:1647/4740 train_time:269332ms step_avg:163.53ms
step:1648/4740 train_time:269648ms step_avg:163.62ms
step:1649/4740 train_time:269821ms step_avg:163.63ms
step:1650/4740 train_time:270140ms step_avg:163.72ms
step:1651/4740 train_time:270319ms step_avg:163.73ms
step:1652/4740 train_time:270638ms step_avg:163.82ms
step:1653/4740 train_time:270816ms step_avg:163.83ms
step:1654/4740 train_time:271133ms step_avg:163.93ms
step:1655/4740 train_time:271311ms step_avg:163.93ms
step:1656/4740 train_time:271627ms step_avg:164.03ms
step:1657/4740 train_time:271803ms step_avg:164.03ms
step:1658/4740 train_time:272122ms step_avg:164.13ms
step:1659/4740 train_time:272299ms step_avg:164.13ms
step:1660/4740 train_time:272620ms step_avg:164.23ms
step:1661/4740 train_time:272797ms step_avg:164.24ms
step:1662/4740 train_time:273109ms step_avg:164.33ms
step:1663/4740 train_time:273294ms step_avg:164.34ms
step:1664/4740 train_time:273611ms step_avg:164.43ms
step:1665/4740 train_time:273790ms step_avg:164.44ms
step:1666/4740 train_time:274107ms step_avg:164.53ms
step:1667/4740 train_time:274285ms step_avg:164.54ms
step:1668/4740 train_time:274605ms step_avg:164.63ms
step:1669/4740 train_time:274780ms step_avg:164.64ms
step:1670/4740 train_time:275095ms step_avg:164.73ms
step:1671/4740 train_time:275274ms step_avg:164.74ms
step:1672/4740 train_time:275596ms step_avg:164.83ms
step:1673/4740 train_time:275773ms step_avg:164.84ms
step:1674/4740 train_time:276089ms step_avg:164.93ms
step:1675/4740 train_time:276267ms step_avg:164.94ms
step:1676/4740 train_time:276586ms step_avg:165.03ms
step:1677/4740 train_time:276764ms step_avg:165.03ms
step:1678/4740 train_time:277081ms step_avg:165.13ms
step:1679/4740 train_time:277257ms step_avg:165.13ms
step:1680/4740 train_time:277574ms step_avg:165.22ms
step:1681/4740 train_time:277751ms step_avg:165.23ms
step:1682/4740 train_time:278074ms step_avg:165.32ms
step:1683/4740 train_time:278252ms step_avg:165.33ms
step:1684/4740 train_time:278570ms step_avg:165.42ms
step:1685/4740 train_time:278744ms step_avg:165.43ms
step:1686/4740 train_time:279065ms step_avg:165.52ms
step:1687/4740 train_time:279239ms step_avg:165.52ms
step:1688/4740 train_time:279559ms step_avg:165.62ms
step:1689/4740 train_time:279736ms step_avg:165.62ms
step:1690/4740 train_time:280053ms step_avg:165.71ms
step:1691/4740 train_time:280228ms step_avg:165.72ms
step:1692/4740 train_time:280546ms step_avg:165.81ms
step:1693/4740 train_time:280722ms step_avg:165.81ms
step:1694/4740 train_time:281045ms step_avg:165.91ms
step:1695/4740 train_time:281219ms step_avg:165.91ms
step:1696/4740 train_time:281540ms step_avg:166.00ms
step:1697/4740 train_time:281716ms step_avg:166.01ms
step:1698/4740 train_time:282032ms step_avg:166.10ms
step:1699/4740 train_time:282211ms step_avg:166.10ms
step:1700/4740 train_time:282526ms step_avg:166.19ms
step:1701/4740 train_time:282704ms step_avg:166.20ms
step:1702/4740 train_time:283022ms step_avg:166.29ms
step:1703/4740 train_time:283200ms step_avg:166.29ms
step:1704/4740 train_time:283518ms step_avg:166.38ms
step:1705/4740 train_time:283695ms step_avg:166.39ms
step:1706/4740 train_time:284010ms step_avg:166.48ms
step:1707/4740 train_time:284190ms step_avg:166.49ms
step:1708/4740 train_time:284506ms step_avg:166.57ms
step:1709/4740 train_time:284683ms step_avg:166.58ms
step:1710/4740 train_time:285002ms step_avg:166.67ms
step:1711/4740 train_time:285177ms step_avg:166.67ms
step:1712/4740 train_time:285494ms step_avg:166.76ms
step:1713/4740 train_time:285674ms step_avg:166.77ms
step:1714/4740 train_time:285991ms step_avg:166.86ms
step:1715/4740 train_time:286166ms step_avg:166.86ms
step:1716/4740 train_time:286487ms step_avg:166.95ms
step:1717/4740 train_time:286664ms step_avg:166.96ms
step:1718/4740 train_time:286984ms step_avg:167.05ms
step:1719/4740 train_time:287158ms step_avg:167.05ms
step:1720/4740 train_time:287475ms step_avg:167.14ms
step:1721/4740 train_time:287653ms step_avg:167.14ms
step:1722/4740 train_time:287971ms step_avg:167.23ms
step:1723/4740 train_time:288146ms step_avg:167.23ms
step:1724/4740 train_time:288462ms step_avg:167.32ms
step:1725/4740 train_time:288641ms step_avg:167.33ms
step:1726/4740 train_time:288963ms step_avg:167.42ms
step:1727/4740 train_time:289140ms step_avg:167.42ms
step:1728/4740 train_time:289459ms step_avg:167.51ms
step:1729/4740 train_time:289635ms step_avg:167.52ms
step:1730/4740 train_time:289953ms step_avg:167.60ms
step:1731/4740 train_time:290133ms step_avg:167.61ms
step:1732/4740 train_time:290450ms step_avg:167.70ms
step:1733/4740 train_time:290627ms step_avg:167.70ms
step:1734/4740 train_time:290945ms step_avg:167.79ms
step:1735/4740 train_time:291120ms step_avg:167.79ms
step:1736/4740 train_time:291441ms step_avg:167.88ms
step:1737/4740 train_time:291615ms step_avg:167.88ms
step:1738/4740 train_time:291932ms step_avg:167.97ms
step:1739/4740 train_time:292109ms step_avg:167.98ms
step:1740/4740 train_time:292426ms step_avg:168.06ms
step:1741/4740 train_time:292600ms step_avg:168.06ms
step:1742/4740 train_time:292915ms step_avg:168.15ms
step:1743/4740 train_time:293096ms step_avg:168.16ms
step:1744/4740 train_time:293412ms step_avg:168.24ms
step:1745/4740 train_time:293588ms step_avg:168.25ms
step:1746/4740 train_time:293907ms step_avg:168.33ms
step:1747/4740 train_time:294082ms step_avg:168.34ms
step:1748/4740 train_time:294403ms step_avg:168.42ms
step:1749/4740 train_time:294579ms step_avg:168.43ms
step:1750/4740 train_time:294898ms step_avg:168.51ms
step:1750/4740 val_loss:3.2800 train_time:295022ms step_avg:168.58ms
step:1751/4740 train_time:295080ms step_avg:168.52ms
step:1752/4740 train_time:295405ms step_avg:168.61ms
step:1753/4740 train_time:295582ms step_avg:168.62ms
step:1754/4740 train_time:295897ms step_avg:168.70ms
step:1755/4740 train_time:296071ms step_avg:168.70ms
step:1756/4740 train_time:296395ms step_avg:168.79ms
step:1757/4740 train_time:296572ms step_avg:168.79ms
step:1758/4740 train_time:296886ms step_avg:168.88ms
step:1759/4740 train_time:297062ms step_avg:168.88ms
step:1760/4740 train_time:297383ms step_avg:168.97ms
step:1761/4740 train_time:297562ms step_avg:168.97ms
step:1762/4740 train_time:297878ms step_avg:169.06ms
step:1763/4740 train_time:298055ms step_avg:169.06ms
step:1764/4740 train_time:298378ms step_avg:169.15ms
step:1765/4740 train_time:298552ms step_avg:169.15ms
step:1766/4740 train_time:298874ms step_avg:169.24ms
step:1767/4740 train_time:299047ms step_avg:169.24ms
step:1768/4740 train_time:299364ms step_avg:169.32ms
step:1769/4740 train_time:299543ms step_avg:169.33ms
step:1770/4740 train_time:299859ms step_avg:169.41ms
step:1771/4740 train_time:300039ms step_avg:169.42ms
step:1772/4740 train_time:300358ms step_avg:169.50ms
step:1773/4740 train_time:300534ms step_avg:169.51ms
step:1774/4740 train_time:300855ms step_avg:169.59ms
step:1775/4740 train_time:301031ms step_avg:169.59ms
step:1776/4740 train_time:301347ms step_avg:169.68ms
step:1777/4740 train_time:301525ms step_avg:169.68ms
step:1778/4740 train_time:301844ms step_avg:169.77ms
step:1779/4740 train_time:302020ms step_avg:169.77ms
step:1780/4740 train_time:302336ms step_avg:169.85ms
step:1781/4740 train_time:302512ms step_avg:169.86ms
step:1782/4740 train_time:302829ms step_avg:169.94ms
step:1783/4740 train_time:303006ms step_avg:169.94ms
step:1784/4740 train_time:303325ms step_avg:170.02ms
step:1785/4740 train_time:303501ms step_avg:170.03ms
step:1786/4740 train_time:303819ms step_avg:170.11ms
step:1787/4740 train_time:303994ms step_avg:170.11ms
step:1788/4740 train_time:304315ms step_avg:170.20ms
step:1789/4740 train_time:304492ms step_avg:170.20ms
step:1790/4740 train_time:304812ms step_avg:170.29ms
step:1791/4740 train_time:304988ms step_avg:170.29ms
step:1792/4740 train_time:305305ms step_avg:170.37ms
step:1793/4740 train_time:305484ms step_avg:170.38ms
step:1794/4740 train_time:305801ms step_avg:170.46ms
step:1795/4740 train_time:305976ms step_avg:170.46ms
step:1796/4740 train_time:306295ms step_avg:170.54ms
step:1797/4740 train_time:306470ms step_avg:170.55ms
step:1798/4740 train_time:306790ms step_avg:170.63ms
step:1799/4740 train_time:306967ms step_avg:170.63ms
step:1800/4740 train_time:307284ms step_avg:170.71ms
step:1801/4740 train_time:307460ms step_avg:170.72ms
step:1802/4740 train_time:307778ms step_avg:170.80ms
step:1803/4740 train_time:307954ms step_avg:170.80ms
step:1804/4740 train_time:308276ms step_avg:170.88ms
step:1805/4740 train_time:308452ms step_avg:170.89ms
step:1806/4740 train_time:308774ms step_avg:170.97ms
step:1807/4740 train_time:308948ms step_avg:170.97ms
step:1808/4740 train_time:309268ms step_avg:171.06ms
step:1809/4740 train_time:309450ms step_avg:171.06ms
step:1810/4740 train_time:309767ms step_avg:171.14ms
step:1811/4740 train_time:309946ms step_avg:171.15ms
step:1812/4740 train_time:310261ms step_avg:171.23ms
step:1813/4740 train_time:310443ms step_avg:171.23ms
step:1814/4740 train_time:310757ms step_avg:171.31ms
step:1815/4740 train_time:310935ms step_avg:171.31ms
step:1816/4740 train_time:311255ms step_avg:171.40ms
step:1817/4740 train_time:311432ms step_avg:171.40ms
step:1818/4740 train_time:311751ms step_avg:171.48ms
step:1819/4740 train_time:311927ms step_avg:171.48ms
step:1820/4740 train_time:312244ms step_avg:171.56ms
step:1821/4740 train_time:312421ms step_avg:171.57ms
step:1822/4740 train_time:312741ms step_avg:171.65ms
step:1823/4740 train_time:312920ms step_avg:171.65ms
step:1824/4740 train_time:313236ms step_avg:171.73ms
step:1825/4740 train_time:313413ms step_avg:171.73ms
step:1826/4740 train_time:313733ms step_avg:171.81ms
step:1827/4740 train_time:313911ms step_avg:171.82ms
step:1828/4740 train_time:314229ms step_avg:171.90ms
step:1829/4740 train_time:314404ms step_avg:171.90ms
step:1830/4740 train_time:314722ms step_avg:171.98ms
step:1831/4740 train_time:314897ms step_avg:171.98ms
step:1832/4740 train_time:315217ms step_avg:172.06ms
step:1833/4740 train_time:315392ms step_avg:172.06ms
step:1834/4740 train_time:315712ms step_avg:172.14ms
step:1835/4740 train_time:315888ms step_avg:172.15ms
step:1836/4740 train_time:316204ms step_avg:172.22ms
step:1837/4740 train_time:316379ms step_avg:172.23ms
step:1838/4740 train_time:316698ms step_avg:172.31ms
step:1839/4740 train_time:316875ms step_avg:172.31ms
step:1840/4740 train_time:317196ms step_avg:172.39ms
step:1841/4740 train_time:317372ms step_avg:172.39ms
step:1842/4740 train_time:317689ms step_avg:172.47ms
step:1843/4740 train_time:317868ms step_avg:172.47ms
step:1844/4740 train_time:318187ms step_avg:172.55ms
step:1845/4740 train_time:318361ms step_avg:172.55ms
step:1846/4740 train_time:318681ms step_avg:172.63ms
step:1847/4740 train_time:318857ms step_avg:172.64ms
step:1848/4740 train_time:319177ms step_avg:172.71ms
step:1849/4740 train_time:319351ms step_avg:172.72ms
step:1850/4740 train_time:319672ms step_avg:172.80ms
step:1851/4740 train_time:319846ms step_avg:172.80ms
step:1852/4740 train_time:320165ms step_avg:172.88ms
step:1853/4740 train_time:320345ms step_avg:172.88ms
step:1854/4740 train_time:320661ms step_avg:172.96ms
step:1855/4740 train_time:320837ms step_avg:172.96ms
step:1856/4740 train_time:321157ms step_avg:173.04ms
step:1857/4740 train_time:321332ms step_avg:173.04ms
step:1858/4740 train_time:321654ms step_avg:173.12ms
step:1859/4740 train_time:321834ms step_avg:173.12ms
step:1860/4740 train_time:322152ms step_avg:173.20ms
step:1861/4740 train_time:322328ms step_avg:173.20ms
step:1862/4740 train_time:322646ms step_avg:173.28ms
step:1863/4740 train_time:322823ms step_avg:173.28ms
step:1864/4740 train_time:323141ms step_avg:173.36ms
step:1865/4740 train_time:323321ms step_avg:173.36ms
step:1866/4740 train_time:323638ms step_avg:173.44ms
step:1867/4740 train_time:323813ms step_avg:173.44ms
step:1868/4740 train_time:324134ms step_avg:173.52ms
step:1869/4740 train_time:324311ms step_avg:173.52ms
step:1870/4740 train_time:324631ms step_avg:173.60ms
step:1871/4740 train_time:324807ms step_avg:173.60ms
step:1872/4740 train_time:325125ms step_avg:173.68ms
step:1873/4740 train_time:325300ms step_avg:173.68ms
step:1874/4740 train_time:325617ms step_avg:173.75ms
step:1875/4740 train_time:325792ms step_avg:173.76ms
step:1876/4740 train_time:326110ms step_avg:173.83ms
step:1877/4740 train_time:326286ms step_avg:173.83ms
step:1878/4740 train_time:326606ms step_avg:173.91ms
step:1879/4740 train_time:326781ms step_avg:173.91ms
step:1880/4740 train_time:327100ms step_avg:173.99ms
step:1881/4740 train_time:327278ms step_avg:173.99ms
step:1882/4740 train_time:327597ms step_avg:174.07ms
step:1883/4740 train_time:327771ms step_avg:174.07ms
step:1884/4740 train_time:328087ms step_avg:174.14ms
step:1885/4740 train_time:328264ms step_avg:174.15ms
step:1886/4740 train_time:328582ms step_avg:174.22ms
step:1887/4740 train_time:328758ms step_avg:174.22ms
step:1888/4740 train_time:329077ms step_avg:174.30ms
step:1889/4740 train_time:329251ms step_avg:174.30ms
step:1890/4740 train_time:329569ms step_avg:174.38ms
step:1891/4740 train_time:329745ms step_avg:174.38ms
step:1892/4740 train_time:330062ms step_avg:174.45ms
step:1893/4740 train_time:330244ms step_avg:174.46ms
step:1894/4740 train_time:330561ms step_avg:174.53ms
step:1895/4740 train_time:330739ms step_avg:174.53ms
step:1896/4740 train_time:331056ms step_avg:174.61ms
step:1897/4740 train_time:331233ms step_avg:174.61ms
step:1898/4740 train_time:331551ms step_avg:174.68ms
step:1899/4740 train_time:331728ms step_avg:174.69ms
step:1900/4740 train_time:332044ms step_avg:174.76ms
step:1901/4740 train_time:332223ms step_avg:174.76ms
step:1902/4740 train_time:332539ms step_avg:174.84ms
step:1903/4740 train_time:332716ms step_avg:174.84ms
step:1904/4740 train_time:333037ms step_avg:174.91ms
step:1905/4740 train_time:333212ms step_avg:174.91ms
step:1906/4740 train_time:333531ms step_avg:174.99ms
step:1907/4740 train_time:333710ms step_avg:174.99ms
step:1908/4740 train_time:334024ms step_avg:175.06ms
step:1909/4740 train_time:334202ms step_avg:175.07ms
step:1910/4740 train_time:334518ms step_avg:175.14ms
step:1911/4740 train_time:334700ms step_avg:175.14ms
step:1912/4740 train_time:335017ms step_avg:175.22ms
step:1913/4740 train_time:335197ms step_avg:175.22ms
step:1914/4740 train_time:335515ms step_avg:175.30ms
step:1915/4740 train_time:335691ms step_avg:175.30ms
step:1916/4740 train_time:336011ms step_avg:175.37ms
step:1917/4740 train_time:336187ms step_avg:175.37ms
step:1918/4740 train_time:336503ms step_avg:175.44ms
step:1919/4740 train_time:336681ms step_avg:175.45ms
step:1920/4740 train_time:336999ms step_avg:175.52ms
step:1921/4740 train_time:337175ms step_avg:175.52ms
step:1922/4740 train_time:337495ms step_avg:175.60ms
step:1923/4740 train_time:337670ms step_avg:175.60ms
step:1924/4740 train_time:337988ms step_avg:175.67ms
step:1925/4740 train_time:338164ms step_avg:175.67ms
step:1926/4740 train_time:338483ms step_avg:175.74ms
step:1927/4740 train_time:338658ms step_avg:175.74ms
step:1928/4740 train_time:338978ms step_avg:175.82ms
step:1929/4740 train_time:339155ms step_avg:175.82ms
step:1930/4740 train_time:339476ms step_avg:175.89ms
step:1931/4740 train_time:339652ms step_avg:175.89ms
step:1932/4740 train_time:339973ms step_avg:175.97ms
step:1933/4740 train_time:340149ms step_avg:175.97ms
step:1934/4740 train_time:340465ms step_avg:176.04ms
step:1935/4740 train_time:340647ms step_avg:176.05ms
step:1936/4740 train_time:340964ms step_avg:176.12ms
step:1937/4740 train_time:341143ms step_avg:176.12ms
step:1938/4740 train_time:341457ms step_avg:176.19ms
step:1939/4740 train_time:341642ms step_avg:176.19ms
step:1940/4740 train_time:341956ms step_avg:176.27ms
step:1941/4740 train_time:342133ms step_avg:176.27ms
step:1942/4740 train_time:342454ms step_avg:176.34ms
step:1943/4740 train_time:342629ms step_avg:176.34ms
step:1944/4740 train_time:342944ms step_avg:176.41ms
step:1945/4740 train_time:343122ms step_avg:176.41ms
step:1946/4740 train_time:343441ms step_avg:176.49ms
step:1947/4740 train_time:343618ms step_avg:176.49ms
step:1948/4740 train_time:343937ms step_avg:176.56ms
step:1949/4740 train_time:344112ms step_avg:176.56ms
step:1950/4740 train_time:344430ms step_avg:176.63ms
step:1951/4740 train_time:344609ms step_avg:176.63ms
step:1952/4740 train_time:344925ms step_avg:176.70ms
step:1953/4740 train_time:345103ms step_avg:176.70ms
step:1954/4740 train_time:345423ms step_avg:176.78ms
step:1955/4740 train_time:345606ms step_avg:176.78ms
step:1956/4740 train_time:345923ms step_avg:176.85ms
step:1957/4740 train_time:346101ms step_avg:176.85ms
step:1958/4740 train_time:346419ms step_avg:176.93ms
step:1959/4740 train_time:346595ms step_avg:176.92ms
step:1960/4740 train_time:346914ms step_avg:177.00ms
step:1961/4740 train_time:347089ms step_avg:177.00ms
step:1962/4740 train_time:347410ms step_avg:177.07ms
step:1963/4740 train_time:347588ms step_avg:177.07ms
step:1964/4740 train_time:347905ms step_avg:177.14ms
step:1965/4740 train_time:348082ms step_avg:177.14ms
step:1966/4740 train_time:348399ms step_avg:177.21ms
step:1967/4740 train_time:348577ms step_avg:177.21ms
step:1968/4740 train_time:348899ms step_avg:177.29ms
step:1969/4740 train_time:349074ms step_avg:177.28ms
step:1970/4740 train_time:349394ms step_avg:177.36ms
step:1971/4740 train_time:349568ms step_avg:177.36ms
step:1972/4740 train_time:349889ms step_avg:177.43ms
step:1973/4740 train_time:350067ms step_avg:177.43ms
step:1974/4740 train_time:350383ms step_avg:177.50ms
step:1975/4740 train_time:350562ms step_avg:177.50ms
step:1976/4740 train_time:350880ms step_avg:177.57ms
step:1977/4740 train_time:351058ms step_avg:177.57ms
step:1978/4740 train_time:351379ms step_avg:177.64ms
step:1979/4740 train_time:351554ms step_avg:177.64ms
step:1980/4740 train_time:351876ms step_avg:177.72ms
step:1981/4740 train_time:352052ms step_avg:177.71ms
step:1982/4740 train_time:352375ms step_avg:177.79ms
step:1983/4740 train_time:352551ms step_avg:177.79ms
step:1984/4740 train_time:352872ms step_avg:177.86ms
step:1985/4740 train_time:353049ms step_avg:177.86ms
step:1986/4740 train_time:353371ms step_avg:177.93ms
step:1987/4740 train_time:353547ms step_avg:177.93ms
step:1988/4740 train_time:353866ms step_avg:178.00ms
step:1989/4740 train_time:354043ms step_avg:178.00ms
step:1990/4740 train_time:354364ms step_avg:178.07ms
step:1991/4740 train_time:354542ms step_avg:178.07ms
step:1992/4740 train_time:354860ms step_avg:178.14ms
step:1993/4740 train_time:355035ms step_avg:178.14ms
step:1994/4740 train_time:355355ms step_avg:178.21ms
step:1995/4740 train_time:355533ms step_avg:178.21ms
step:1996/4740 train_time:355851ms step_avg:178.28ms
step:1997/4740 train_time:356031ms step_avg:178.28ms
step:1998/4740 train_time:356350ms step_avg:178.35ms
step:1999/4740 train_time:356530ms step_avg:178.35ms
step:2000/4740 train_time:356848ms step_avg:178.42ms
step:2000/4740 val_loss:3.2332 train_time:356972ms step_avg:178.49ms
step:2001/4740 train_time:357031ms step_avg:178.43ms
step:2002/4740 train_time:357356ms step_avg:178.50ms
step:2003/4740 train_time:357530ms step_avg:178.50ms
step:2004/4740 train_time:357842ms step_avg:178.56ms
step:2005/4740 train_time:358020ms step_avg:178.56ms
step:2006/4740 train_time:358348ms step_avg:178.64ms
step:2007/4740 train_time:358524ms step_avg:178.64ms
step:2008/4740 train_time:358841ms step_avg:178.71ms
step:2009/4740 train_time:359017ms step_avg:178.70ms
step:2010/4740 train_time:359343ms step_avg:178.78ms
step:2011/4740 train_time:359517ms step_avg:178.78ms
step:2012/4740 train_time:359833ms step_avg:178.84ms
step:2013/4740 train_time:360012ms step_avg:178.84ms
step:2014/4740 train_time:360331ms step_avg:178.91ms
step:2015/4740 train_time:360507ms step_avg:178.91ms
step:2016/4740 train_time:360827ms step_avg:178.98ms
step:2017/4740 train_time:361004ms step_avg:178.98ms
step:2018/4740 train_time:361326ms step_avg:179.05ms
step:2019/4740 train_time:361501ms step_avg:179.05ms
step:2020/4740 train_time:361819ms step_avg:179.12ms
step:2021/4740 train_time:361998ms step_avg:179.12ms
step:2022/4740 train_time:362319ms step_avg:179.19ms
step:2023/4740 train_time:362494ms step_avg:179.19ms
step:2024/4740 train_time:362812ms step_avg:179.25ms
step:2025/4740 train_time:362990ms step_avg:179.25ms
step:2026/4740 train_time:363309ms step_avg:179.32ms
step:2027/4740 train_time:363487ms step_avg:179.32ms
step:2028/4740 train_time:363806ms step_avg:179.39ms
step:2029/4740 train_time:363981ms step_avg:179.39ms
step:2030/4740 train_time:364304ms step_avg:179.46ms
step:2031/4740 train_time:364477ms step_avg:179.46ms
step:2032/4740 train_time:364797ms step_avg:179.53ms
step:2033/4740 train_time:364972ms step_avg:179.52ms
step:2034/4740 train_time:365291ms step_avg:179.59ms
step:2035/4740 train_time:365468ms step_avg:179.59ms
step:2036/4740 train_time:365785ms step_avg:179.66ms
step:2037/4740 train_time:365963ms step_avg:179.66ms
step:2038/4740 train_time:366284ms step_avg:179.73ms
step:2039/4740 train_time:366459ms step_avg:179.72ms
step:2040/4740 train_time:366778ms step_avg:179.79ms
step:2041/4740 train_time:366956ms step_avg:179.79ms
step:2042/4740 train_time:367271ms step_avg:179.86ms
step:2043/4740 train_time:367451ms step_avg:179.86ms
step:2044/4740 train_time:367768ms step_avg:179.93ms
step:2045/4740 train_time:367945ms step_avg:179.92ms
step:2046/4740 train_time:368265ms step_avg:179.99ms
step:2047/4740 train_time:368441ms step_avg:179.99ms
step:2048/4740 train_time:368757ms step_avg:180.06ms
step:2049/4740 train_time:368935ms step_avg:180.06ms
step:2050/4740 train_time:369254ms step_avg:180.12ms
step:2051/4740 train_time:369432ms step_avg:180.12ms
step:2052/4740 train_time:369751ms step_avg:180.19ms
step:2053/4740 train_time:369931ms step_avg:180.19ms
step:2054/4740 train_time:370249ms step_avg:180.26ms
step:2055/4740 train_time:370428ms step_avg:180.26ms
step:2056/4740 train_time:370748ms step_avg:180.32ms
step:2057/4740 train_time:370923ms step_avg:180.32ms
step:2058/4740 train_time:371244ms step_avg:180.39ms
step:2059/4740 train_time:371423ms step_avg:180.39ms
step:2060/4740 train_time:371744ms step_avg:180.46ms
step:2061/4740 train_time:371919ms step_avg:180.46ms
step:2062/4740 train_time:372242ms step_avg:180.52ms
step:2063/4740 train_time:372417ms step_avg:180.52ms
step:2064/4740 train_time:372737ms step_avg:180.59ms
step:2065/4740 train_time:372915ms step_avg:180.59ms
step:2066/4740 train_time:373234ms step_avg:180.66ms
step:2067/4740 train_time:373409ms step_avg:180.65ms
step:2068/4740 train_time:373732ms step_avg:180.72ms
step:2069/4740 train_time:373909ms step_avg:180.72ms
step:2070/4740 train_time:374231ms step_avg:180.79ms
step:2071/4740 train_time:374407ms step_avg:180.79ms
step:2072/4740 train_time:374728ms step_avg:180.85ms
step:2073/4740 train_time:374908ms step_avg:180.85ms
step:2074/4740 train_time:375228ms step_avg:180.92ms
step:2075/4740 train_time:375405ms step_avg:180.92ms
step:2076/4740 train_time:375726ms step_avg:180.99ms
step:2077/4740 train_time:375902ms step_avg:180.98ms
step:2078/4740 train_time:376222ms step_avg:181.05ms
step:2079/4740 train_time:376398ms step_avg:181.05ms
step:2080/4740 train_time:376716ms step_avg:181.11ms
step:2081/4740 train_time:376895ms step_avg:181.11ms
step:2082/4740 train_time:377209ms step_avg:181.18ms
step:2083/4740 train_time:377387ms step_avg:181.17ms
step:2084/4740 train_time:377708ms step_avg:181.24ms
step:2085/4740 train_time:377890ms step_avg:181.24ms
step:2086/4740 train_time:378210ms step_avg:181.31ms
step:2087/4740 train_time:378384ms step_avg:181.31ms
step:2088/4740 train_time:378705ms step_avg:181.37ms
step:2089/4740 train_time:378882ms step_avg:181.37ms
step:2090/4740 train_time:379201ms step_avg:181.44ms
step:2091/4740 train_time:379379ms step_avg:181.43ms
step:2092/4740 train_time:379703ms step_avg:181.50ms
step:2093/4740 train_time:379879ms step_avg:181.50ms
step:2094/4740 train_time:380193ms step_avg:181.56ms
step:2095/4740 train_time:380374ms step_avg:181.56ms
step:2096/4740 train_time:380694ms step_avg:181.63ms
step:2097/4740 train_time:380873ms step_avg:181.63ms
step:2098/4740 train_time:381190ms step_avg:181.69ms
step:2099/4740 train_time:381371ms step_avg:181.69ms
step:2100/4740 train_time:381689ms step_avg:181.76ms
step:2101/4740 train_time:381873ms step_avg:181.76ms
step:2102/4740 train_time:382189ms step_avg:181.82ms
step:2103/4740 train_time:382369ms step_avg:181.82ms
step:2104/4740 train_time:382690ms step_avg:181.89ms
step:2105/4740 train_time:382867ms step_avg:181.88ms
step:2106/4740 train_time:383184ms step_avg:181.95ms
step:2107/4740 train_time:383361ms step_avg:181.95ms
step:2108/4740 train_time:383682ms step_avg:182.01ms
step:2109/4740 train_time:383858ms step_avg:182.01ms
step:2110/4740 train_time:384177ms step_avg:182.07ms
step:2111/4740 train_time:384353ms step_avg:182.07ms
step:2112/4740 train_time:384675ms step_avg:182.14ms
step:2113/4740 train_time:384852ms step_avg:182.14ms
step:2114/4740 train_time:385170ms step_avg:182.20ms
step:2115/4740 train_time:385348ms step_avg:182.20ms
step:2116/4740 train_time:385667ms step_avg:182.26ms
step:2117/4740 train_time:385844ms step_avg:182.26ms
step:2118/4740 train_time:386163ms step_avg:182.32ms
step:2119/4740 train_time:386339ms step_avg:182.32ms
step:2120/4740 train_time:386656ms step_avg:182.38ms
step:2121/4740 train_time:386834ms step_avg:182.38ms
step:2122/4740 train_time:387154ms step_avg:182.45ms
step:2123/4740 train_time:387333ms step_avg:182.45ms
step:2124/4740 train_time:387652ms step_avg:182.51ms
step:2125/4740 train_time:387831ms step_avg:182.51ms
step:2126/4740 train_time:388145ms step_avg:182.57ms
step:2127/4740 train_time:388327ms step_avg:182.57ms
step:2128/4740 train_time:388645ms step_avg:182.63ms
step:2129/4740 train_time:388822ms step_avg:182.63ms
step:2130/4740 train_time:389144ms step_avg:182.70ms
step:2131/4740 train_time:389319ms step_avg:182.69ms
step:2132/4740 train_time:389636ms step_avg:182.76ms
step:2133/4740 train_time:389815ms step_avg:182.75ms
step:2134/4740 train_time:390133ms step_avg:182.82ms
step:2135/4740 train_time:390311ms step_avg:182.82ms
step:2136/4740 train_time:390631ms step_avg:182.88ms
step:2137/4740 train_time:390810ms step_avg:182.88ms
step:2138/4740 train_time:391131ms step_avg:182.94ms
step:2139/4740 train_time:391312ms step_avg:182.94ms
step:2140/4740 train_time:391630ms step_avg:183.00ms
step:2141/4740 train_time:391805ms step_avg:183.00ms
step:2142/4740 train_time:392126ms step_avg:183.07ms
step:2143/4740 train_time:392300ms step_avg:183.06ms
step:2144/4740 train_time:392624ms step_avg:183.13ms
step:2145/4740 train_time:392800ms step_avg:183.12ms
step:2146/4740 train_time:393118ms step_avg:183.19ms
step:2147/4740 train_time:393296ms step_avg:183.18ms
step:2148/4740 train_time:393615ms step_avg:183.25ms
step:2149/4740 train_time:393791ms step_avg:183.24ms
step:2150/4740 train_time:394109ms step_avg:183.31ms
step:2151/4740 train_time:394286ms step_avg:183.30ms
step:2152/4740 train_time:394605ms step_avg:183.37ms
step:2153/4740 train_time:394783ms step_avg:183.36ms
step:2154/4740 train_time:395105ms step_avg:183.43ms
step:2155/4740 train_time:395281ms step_avg:183.43ms
step:2156/4740 train_time:395603ms step_avg:183.49ms
step:2157/4740 train_time:395781ms step_avg:183.49ms
step:2158/4740 train_time:396101ms step_avg:183.55ms
step:2159/4740 train_time:396276ms step_avg:183.55ms
step:2160/4740 train_time:396594ms step_avg:183.61ms
step:2161/4740 train_time:396776ms step_avg:183.61ms
step:2162/4740 train_time:397093ms step_avg:183.67ms
step:2163/4740 train_time:397271ms step_avg:183.67ms
step:2164/4740 train_time:397589ms step_avg:183.73ms
step:2165/4740 train_time:397771ms step_avg:183.73ms
step:2166/4740 train_time:398087ms step_avg:183.79ms
step:2167/4740 train_time:398265ms step_avg:183.79ms
step:2168/4740 train_time:398585ms step_avg:183.85ms
step:2169/4740 train_time:398762ms step_avg:183.85ms
step:2170/4740 train_time:399082ms step_avg:183.91ms
step:2171/4740 train_time:399258ms step_avg:183.90ms
step:2172/4740 train_time:399575ms step_avg:183.97ms
step:2173/4740 train_time:399752ms step_avg:183.96ms
step:2174/4740 train_time:400071ms step_avg:184.03ms
step:2175/4740 train_time:400250ms step_avg:184.02ms
step:2176/4740 train_time:400570ms step_avg:184.09ms
step:2177/4740 train_time:400746ms step_avg:184.08ms
step:2178/4740 train_time:401066ms step_avg:184.14ms
step:2179/4740 train_time:401240ms step_avg:184.14ms
step:2180/4740 train_time:401560ms step_avg:184.20ms
step:2181/4740 train_time:401735ms step_avg:184.20ms
step:2182/4740 train_time:402055ms step_avg:184.26ms
step:2183/4740 train_time:402232ms step_avg:184.26ms
step:2184/4740 train_time:402548ms step_avg:184.32ms
step:2185/4740 train_time:402724ms step_avg:184.31ms
step:2186/4740 train_time:403046ms step_avg:184.38ms
step:2187/4740 train_time:403223ms step_avg:184.37ms
step:2188/4740 train_time:403542ms step_avg:184.43ms
step:2189/4740 train_time:403718ms step_avg:184.43ms
step:2190/4740 train_time:404034ms step_avg:184.49ms
step:2191/4740 train_time:404212ms step_avg:184.49ms
step:2192/4740 train_time:404529ms step_avg:184.55ms
step:2193/4740 train_time:404707ms step_avg:184.55ms
step:2194/4740 train_time:405027ms step_avg:184.61ms
step:2195/4740 train_time:405205ms step_avg:184.60ms
step:2196/4740 train_time:405528ms step_avg:184.67ms
step:2197/4740 train_time:405703ms step_avg:184.66ms
step:2198/4740 train_time:406026ms step_avg:184.73ms
step:2199/4740 train_time:406201ms step_avg:184.72ms
step:2200/4740 train_time:406521ms step_avg:184.78ms
step:2201/4740 train_time:406697ms step_avg:184.78ms
step:2202/4740 train_time:407020ms step_avg:184.84ms
step:2203/4740 train_time:407195ms step_avg:184.84ms
step:2204/4740 train_time:407514ms step_avg:184.90ms
step:2205/4740 train_time:407692ms step_avg:184.89ms
step:2206/4740 train_time:408010ms step_avg:184.95ms
step:2207/4740 train_time:408189ms step_avg:184.95ms
step:2208/4740 train_time:408510ms step_avg:185.01ms
step:2209/4740 train_time:408684ms step_avg:185.01ms
step:2210/4740 train_time:409006ms step_avg:185.07ms
step:2211/4740 train_time:409185ms step_avg:185.07ms
step:2212/4740 train_time:409505ms step_avg:185.13ms
step:2213/4740 train_time:409679ms step_avg:185.12ms
step:2214/4740 train_time:409995ms step_avg:185.18ms
step:2215/4740 train_time:410174ms step_avg:185.18ms
step:2216/4740 train_time:410491ms step_avg:185.24ms
step:2217/4740 train_time:410675ms step_avg:185.24ms
step:2218/4740 train_time:410991ms step_avg:185.30ms
step:2219/4740 train_time:411170ms step_avg:185.30ms
step:2220/4740 train_time:411493ms step_avg:185.36ms
step:2221/4740 train_time:411673ms step_avg:185.35ms
step:2222/4740 train_time:411989ms step_avg:185.41ms
step:2223/4740 train_time:412171ms step_avg:185.41ms
step:2224/4740 train_time:412490ms step_avg:185.47ms
step:2225/4740 train_time:412667ms step_avg:185.47ms
step:2226/4740 train_time:412984ms step_avg:185.53ms
step:2227/4740 train_time:413164ms step_avg:185.52ms
step:2228/4740 train_time:413484ms step_avg:185.59ms
step:2229/4740 train_time:413660ms step_avg:185.58ms
step:2230/4740 train_time:413982ms step_avg:185.64ms
step:2231/4740 train_time:414158ms step_avg:185.64ms
step:2232/4740 train_time:414477ms step_avg:185.70ms
step:2233/4740 train_time:414656ms step_avg:185.69ms
step:2234/4740 train_time:414973ms step_avg:185.75ms
step:2235/4740 train_time:415153ms step_avg:185.75ms
step:2236/4740 train_time:415469ms step_avg:185.81ms
step:2237/4740 train_time:415649ms step_avg:185.81ms
step:2238/4740 train_time:415967ms step_avg:185.87ms
step:2239/4740 train_time:416147ms step_avg:185.86ms
step:2240/4740 train_time:416467ms step_avg:185.92ms
step:2241/4740 train_time:416643ms step_avg:185.92ms
step:2242/4740 train_time:416966ms step_avg:185.98ms
step:2243/4740 train_time:417142ms step_avg:185.98ms
step:2244/4740 train_time:417464ms step_avg:186.04ms
step:2245/4740 train_time:417639ms step_avg:186.03ms
step:2246/4740 train_time:417960ms step_avg:186.09ms
step:2247/4740 train_time:418137ms step_avg:186.09ms
step:2248/4740 train_time:418454ms step_avg:186.15ms
step:2249/4740 train_time:418630ms step_avg:186.14ms
step:2250/4740 train_time:418950ms step_avg:186.20ms
step:2250/4740 val_loss:3.1965 train_time:419074ms step_avg:186.26ms
step:2251/4740 train_time:419130ms step_avg:186.20ms
step:2252/4740 train_time:419458ms step_avg:186.26ms
step:2253/4740 train_time:419633ms step_avg:186.26ms
step:2254/4740 train_time:419946ms step_avg:186.31ms
step:2255/4740 train_time:420123ms step_avg:186.31ms
step:2256/4740 train_time:420447ms step_avg:186.37ms
step:2257/4740 train_time:420624ms step_avg:186.36ms
step:2258/4740 train_time:420941ms step_avg:186.42ms
step:2259/4740 train_time:421117ms step_avg:186.42ms
step:2260/4740 train_time:421442ms step_avg:186.48ms
step:2261/4740 train_time:421617ms step_avg:186.47ms
step:2262/4740 train_time:421938ms step_avg:186.53ms
step:2263/4740 train_time:422112ms step_avg:186.53ms
step:2264/4740 train_time:422431ms step_avg:186.59ms
step:2265/4740 train_time:422607ms step_avg:186.58ms
step:2266/4740 train_time:422925ms step_avg:186.64ms
step:2267/4740 train_time:423100ms step_avg:186.63ms
step:2268/4740 train_time:423419ms step_avg:186.69ms
step:2269/4740 train_time:423599ms step_avg:186.69ms
step:2270/4740 train_time:423918ms step_avg:186.75ms
step:2271/4740 train_time:424096ms step_avg:186.74ms
step:2272/4740 train_time:424414ms step_avg:186.80ms
step:2273/4740 train_time:424593ms step_avg:186.80ms
step:2274/4740 train_time:424908ms step_avg:186.86ms
step:2275/4740 train_time:425086ms step_avg:186.85ms
step:2276/4740 train_time:425406ms step_avg:186.91ms
step:2277/4740 train_time:425586ms step_avg:186.91ms
step:2278/4740 train_time:425907ms step_avg:186.97ms
step:2279/4740 train_time:426084ms step_avg:186.96ms
step:2280/4740 train_time:426404ms step_avg:187.02ms
step:2281/4740 train_time:426583ms step_avg:187.02ms
step:2282/4740 train_time:426905ms step_avg:187.08ms
step:2283/4740 train_time:427081ms step_avg:187.07ms
step:2284/4740 train_time:427401ms step_avg:187.13ms
step:2285/4740 train_time:427576ms step_avg:187.12ms
step:2286/4740 train_time:427897ms step_avg:187.18ms
step:2287/4740 train_time:428073ms step_avg:187.18ms
step:2288/4740 train_time:428389ms step_avg:187.23ms
step:2289/4740 train_time:428566ms step_avg:187.23ms
step:2290/4740 train_time:428886ms step_avg:187.29ms
step:2291/4740 train_time:429065ms step_avg:187.28ms
step:2292/4740 train_time:429383ms step_avg:187.34ms
step:2293/4740 train_time:429559ms step_avg:187.33ms
step:2294/4740 train_time:429880ms step_avg:187.39ms
step:2295/4740 train_time:430057ms step_avg:187.39ms
step:2296/4740 train_time:430379ms step_avg:187.45ms
step:2297/4740 train_time:430553ms step_avg:187.44ms
step:2298/4740 train_time:430871ms step_avg:187.50ms
step:2299/4740 train_time:431050ms step_avg:187.49ms
step:2300/4740 train_time:431368ms step_avg:187.55ms
step:2301/4740 train_time:431546ms step_avg:187.55ms
step:2302/4740 train_time:431867ms step_avg:187.61ms
step:2303/4740 train_time:432044ms step_avg:187.60ms
step:2304/4740 train_time:432366ms step_avg:187.66ms
step:2305/4740 train_time:432541ms step_avg:187.65ms
step:2306/4740 train_time:432860ms step_avg:187.71ms
step:2307/4740 train_time:433038ms step_avg:187.71ms
step:2308/4740 train_time:433358ms step_avg:187.76ms
step:2309/4740 train_time:433534ms step_avg:187.76ms
step:2310/4740 train_time:433851ms step_avg:187.81ms
step:2311/4740 train_time:434030ms step_avg:187.81ms
step:2312/4740 train_time:434348ms step_avg:187.87ms
step:2313/4740 train_time:434528ms step_avg:187.86ms
step:2314/4740 train_time:434845ms step_avg:187.92ms
step:2315/4740 train_time:435021ms step_avg:187.91ms
step:2316/4740 train_time:435343ms step_avg:187.97ms
step:2317/4740 train_time:435518ms step_avg:187.97ms
step:2318/4740 train_time:435842ms step_avg:188.02ms
step:2319/4740 train_time:436017ms step_avg:188.02ms
step:2320/4740 train_time:436340ms step_avg:188.08ms
step:2321/4740 train_time:436516ms step_avg:188.07ms
step:2322/4740 train_time:436831ms step_avg:188.13ms
step:2323/4740 train_time:437008ms step_avg:188.12ms
step:2324/4740 train_time:437328ms step_avg:188.18ms
step:2325/4740 train_time:437506ms step_avg:188.17ms
step:2326/4740 train_time:437822ms step_avg:188.23ms
step:2327/4740 train_time:438001ms step_avg:188.23ms
step:2328/4740 train_time:438321ms step_avg:188.28ms
step:2329/4740 train_time:438498ms step_avg:188.28ms
step:2330/4740 train_time:438819ms step_avg:188.33ms
step:2331/4740 train_time:438994ms step_avg:188.33ms
step:2332/4740 train_time:439315ms step_avg:188.39ms
step:2333/4740 train_time:439492ms step_avg:188.38ms
step:2334/4740 train_time:439810ms step_avg:188.44ms
step:2335/4740 train_time:439987ms step_avg:188.43ms
step:2336/4740 train_time:440306ms step_avg:188.49ms
step:2337/4740 train_time:440483ms step_avg:188.48ms
step:2338/4740 train_time:440801ms step_avg:188.54ms
step:2339/4740 train_time:440978ms step_avg:188.53ms
step:2340/4740 train_time:441299ms step_avg:188.59ms
step:2341/4740 train_time:441476ms step_avg:188.58ms
step:2342/4740 train_time:441797ms step_avg:188.64ms
step:2343/4740 train_time:441974ms step_avg:188.64ms
step:2344/4740 train_time:442293ms step_avg:188.69ms
step:2345/4740 train_time:442472ms step_avg:188.69ms
step:2346/4740 train_time:442792ms step_avg:188.74ms
step:2347/4740 train_time:442967ms step_avg:188.74ms
step:2348/4740 train_time:443283ms step_avg:188.79ms
step:2349/4740 train_time:443462ms step_avg:188.79ms
step:2350/4740 train_time:443779ms step_avg:188.84ms
step:2351/4740 train_time:443956ms step_avg:188.84ms
step:2352/4740 train_time:444276ms step_avg:188.89ms
step:2353/4740 train_time:444454ms step_avg:188.89ms
step:2354/4740 train_time:444773ms step_avg:188.94ms
step:2355/4740 train_time:444952ms step_avg:188.94ms
step:2356/4740 train_time:445270ms step_avg:188.99ms
step:2357/4740 train_time:445448ms step_avg:188.99ms
step:2358/4740 train_time:445768ms step_avg:189.04ms
step:2359/4740 train_time:445949ms step_avg:189.04ms
step:2360/4740 train_time:446268ms step_avg:189.10ms
step:2361/4740 train_time:446448ms step_avg:189.09ms
step:2362/4740 train_time:446767ms step_avg:189.15ms
step:2363/4740 train_time:446944ms step_avg:189.14ms
step:2364/4740 train_time:447264ms step_avg:189.20ms
step:2365/4740 train_time:447440ms step_avg:189.19ms
step:2366/4740 train_time:447764ms step_avg:189.25ms
step:2367/4740 train_time:447940ms step_avg:189.24ms
step:2368/4740 train_time:448260ms step_avg:189.30ms
step:2369/4740 train_time:448437ms step_avg:189.29ms
step:2370/4740 train_time:448761ms step_avg:189.35ms
step:2371/4740 train_time:448938ms step_avg:189.35ms
step:2372/4740 train_time:449256ms step_avg:189.40ms
step:2373/4740 train_time:449434ms step_avg:189.39ms
step:2374/4740 train_time:449756ms step_avg:189.45ms
step:2375/4740 train_time:449934ms step_avg:189.45ms
step:2376/4740 train_time:450252ms step_avg:189.50ms
step:2377/4740 train_time:450431ms step_avg:189.50ms
step:2378/4740 train_time:450750ms step_avg:189.55ms
step:2379/4740 train_time:450930ms step_avg:189.55ms
step:2380/4740 train_time:451247ms step_avg:189.60ms
step:2381/4740 train_time:451428ms step_avg:189.60ms
step:2382/4740 train_time:451748ms step_avg:189.65ms
step:2383/4740 train_time:451925ms step_avg:189.65ms
step:2384/4740 train_time:452246ms step_avg:189.70ms
step:2385/4740 train_time:452424ms step_avg:189.70ms
step:2386/4740 train_time:452745ms step_avg:189.75ms
step:2387/4740 train_time:452921ms step_avg:189.74ms
step:2388/4740 train_time:453240ms step_avg:189.80ms
step:2389/4740 train_time:453417ms step_avg:189.79ms
step:2390/4740 train_time:453740ms step_avg:189.85ms
step:2391/4740 train_time:453916ms step_avg:189.84ms
step:2392/4740 train_time:454241ms step_avg:189.90ms
step:2393/4740 train_time:454418ms step_avg:189.89ms
step:2394/4740 train_time:454741ms step_avg:189.95ms
step:2395/4740 train_time:454916ms step_avg:189.94ms
step:2396/4740 train_time:455239ms step_avg:190.00ms
step:2397/4740 train_time:455418ms step_avg:190.00ms
step:2398/4740 train_time:455738ms step_avg:190.05ms
step:2399/4740 train_time:455918ms step_avg:190.04ms
step:2400/4740 train_time:456242ms step_avg:190.10ms
step:2401/4740 train_time:456418ms step_avg:190.09ms
step:2402/4740 train_time:456739ms step_avg:190.15ms
step:2403/4740 train_time:456920ms step_avg:190.15ms
step:2404/4740 train_time:457242ms step_avg:190.20ms
step:2405/4740 train_time:457420ms step_avg:190.20ms
step:2406/4740 train_time:457741ms step_avg:190.25ms
step:2407/4740 train_time:457918ms step_avg:190.24ms
step:2408/4740 train_time:458242ms step_avg:190.30ms
step:2409/4740 train_time:458418ms step_avg:190.29ms
step:2410/4740 train_time:458740ms step_avg:190.35ms
step:2411/4740 train_time:458917ms step_avg:190.34ms
step:2412/4740 train_time:459242ms step_avg:190.40ms
step:2413/4740 train_time:459419ms step_avg:190.39ms
step:2414/4740 train_time:459740ms step_avg:190.45ms
step:2415/4740 train_time:459917ms step_avg:190.44ms
step:2416/4740 train_time:460238ms step_avg:190.50ms
step:2417/4740 train_time:460415ms step_avg:190.49ms
step:2418/4740 train_time:460734ms step_avg:190.54ms
step:2419/4740 train_time:460911ms step_avg:190.54ms
step:2420/4740 train_time:461229ms step_avg:190.59ms
step:2421/4740 train_time:461408ms step_avg:190.59ms
step:2422/4740 train_time:461724ms step_avg:190.64ms
step:2423/4740 train_time:461906ms step_avg:190.63ms
step:2424/4740 train_time:462228ms step_avg:190.69ms
step:2425/4740 train_time:462409ms step_avg:190.68ms
step:2426/4740 train_time:462728ms step_avg:190.74ms
step:2427/4740 train_time:462907ms step_avg:190.73ms
step:2428/4740 train_time:463228ms step_avg:190.79ms
step:2429/4740 train_time:463406ms step_avg:190.78ms
step:2430/4740 train_time:463723ms step_avg:190.83ms
step:2431/4740 train_time:463900ms step_avg:190.83ms
step:2432/4740 train_time:464221ms step_avg:190.88ms
step:2433/4740 train_time:464399ms step_avg:190.88ms
step:2434/4740 train_time:464719ms step_avg:190.93ms
step:2435/4740 train_time:464897ms step_avg:190.92ms
step:2436/4740 train_time:465217ms step_avg:190.98ms
step:2437/4740 train_time:465394ms step_avg:190.97ms
step:2438/4740 train_time:465714ms step_avg:191.02ms
step:2439/4740 train_time:465896ms step_avg:191.02ms
step:2440/4740 train_time:466212ms step_avg:191.07ms
step:2441/4740 train_time:466389ms step_avg:191.06ms
step:2442/4740 train_time:466706ms step_avg:191.12ms
step:2443/4740 train_time:466890ms step_avg:191.11ms
step:2444/4740 train_time:467205ms step_avg:191.16ms
step:2445/4740 train_time:467390ms step_avg:191.16ms
step:2446/4740 train_time:467707ms step_avg:191.21ms
step:2447/4740 train_time:467885ms step_avg:191.21ms
step:2448/4740 train_time:468205ms step_avg:191.26ms
step:2449/4740 train_time:468385ms step_avg:191.26ms
step:2450/4740 train_time:468703ms step_avg:191.31ms
step:2451/4740 train_time:468880ms step_avg:191.30ms
step:2452/4740 train_time:469202ms step_avg:191.35ms
step:2453/4740 train_time:469379ms step_avg:191.35ms
step:2454/4740 train_time:469703ms step_avg:191.40ms
step:2455/4740 train_time:469879ms step_avg:191.40ms
step:2456/4740 train_time:470200ms step_avg:191.45ms
step:2457/4740 train_time:470378ms step_avg:191.44ms
step:2458/4740 train_time:470701ms step_avg:191.50ms
step:2459/4740 train_time:470875ms step_avg:191.49ms
step:2460/4740 train_time:471195ms step_avg:191.54ms
step:2461/4740 train_time:471372ms step_avg:191.54ms
step:2462/4740 train_time:471691ms step_avg:191.59ms
step:2463/4740 train_time:471870ms step_avg:191.58ms
step:2464/4740 train_time:472188ms step_avg:191.63ms
step:2465/4740 train_time:472368ms step_avg:191.63ms
step:2466/4740 train_time:472689ms step_avg:191.68ms
step:2467/4740 train_time:472869ms step_avg:191.68ms
step:2468/4740 train_time:473189ms step_avg:191.73ms
step:2469/4740 train_time:473365ms step_avg:191.72ms
step:2470/4740 train_time:473683ms step_avg:191.77ms
step:2471/4740 train_time:473861ms step_avg:191.77ms
step:2472/4740 train_time:474181ms step_avg:191.82ms
step:2473/4740 train_time:474360ms step_avg:191.82ms
step:2474/4740 train_time:474684ms step_avg:191.87ms
step:2475/4740 train_time:474861ms step_avg:191.86ms
step:2476/4740 train_time:475179ms step_avg:191.91ms
step:2477/4740 train_time:475356ms step_avg:191.91ms
step:2478/4740 train_time:475679ms step_avg:191.96ms
step:2479/4740 train_time:475855ms step_avg:191.95ms
step:2480/4740 train_time:476173ms step_avg:192.01ms
step:2481/4740 train_time:476350ms step_avg:192.00ms
step:2482/4740 train_time:476671ms step_avg:192.05ms
step:2483/4740 train_time:476850ms step_avg:192.05ms
step:2484/4740 train_time:477167ms step_avg:192.10ms
step:2485/4740 train_time:477347ms step_avg:192.09ms
step:2486/4740 train_time:477666ms step_avg:192.14ms
step:2487/4740 train_time:477842ms step_avg:192.14ms
step:2488/4740 train_time:478163ms step_avg:192.19ms
step:2489/4740 train_time:478338ms step_avg:192.18ms
step:2490/4740 train_time:478662ms step_avg:192.23ms
step:2491/4740 train_time:478838ms step_avg:192.23ms
step:2492/4740 train_time:479158ms step_avg:192.28ms
step:2493/4740 train_time:479340ms step_avg:192.27ms
step:2494/4740 train_time:479659ms step_avg:192.33ms
step:2495/4740 train_time:479837ms step_avg:192.32ms
step:2496/4740 train_time:480155ms step_avg:192.37ms
step:2497/4740 train_time:480335ms step_avg:192.36ms
step:2498/4740 train_time:480657ms step_avg:192.42ms
step:2499/4740 train_time:480834ms step_avg:192.41ms
step:2500/4740 train_time:481154ms step_avg:192.46ms
step:2500/4740 val_loss:3.1620 train_time:481279ms step_avg:192.51ms
step:2501/4740 train_time:481336ms step_avg:192.46ms
step:2502/4740 train_time:481666ms step_avg:192.51ms
step:2503/4740 train_time:481839ms step_avg:192.50ms
step:2504/4740 train_time:482151ms step_avg:192.55ms
step:2505/4740 train_time:482328ms step_avg:192.55ms
step:2506/4740 train_time:482654ms step_avg:192.60ms
step:2507/4740 train_time:482834ms step_avg:192.59ms
step:2508/4740 train_time:483152ms step_avg:192.64ms
step:2509/4740 train_time:483328ms step_avg:192.64ms
step:2510/4740 train_time:483650ms step_avg:192.69ms
step:2511/4740 train_time:483831ms step_avg:192.68ms
step:2512/4740 train_time:484148ms step_avg:192.73ms
step:2513/4740 train_time:484322ms step_avg:192.73ms
step:2514/4740 train_time:484644ms step_avg:192.78ms
step:2515/4740 train_time:484822ms step_avg:192.77ms
step:2516/4740 train_time:485141ms step_avg:192.82ms
step:2517/4740 train_time:485317ms step_avg:192.82ms
step:2518/4740 train_time:485639ms step_avg:192.87ms
step:2519/4740 train_time:485817ms step_avg:192.86ms
step:2520/4740 train_time:486140ms step_avg:192.91ms
step:2521/4740 train_time:486318ms step_avg:192.91ms
step:2522/4740 train_time:486639ms step_avg:192.96ms
step:2523/4740 train_time:486818ms step_avg:192.95ms
step:2524/4740 train_time:487138ms step_avg:193.00ms
step:2525/4740 train_time:487319ms step_avg:193.00ms
step:2526/4740 train_time:487635ms step_avg:193.05ms
step:2527/4740 train_time:487817ms step_avg:193.04ms
step:2528/4740 train_time:488132ms step_avg:193.09ms
step:2529/4740 train_time:488313ms step_avg:193.09ms
step:2530/4740 train_time:488628ms step_avg:193.13ms
step:2531/4740 train_time:488814ms step_avg:193.13ms
step:2532/4740 train_time:489134ms step_avg:193.18ms
step:2533/4740 train_time:489314ms step_avg:193.18ms
step:2534/4740 train_time:489630ms step_avg:193.22ms
step:2535/4740 train_time:489817ms step_avg:193.22ms
step:2536/4740 train_time:490137ms step_avg:193.27ms
step:2537/4740 train_time:490316ms step_avg:193.27ms
step:2538/4740 train_time:490633ms step_avg:193.31ms
step:2539/4740 train_time:490814ms step_avg:193.31ms
step:2540/4740 train_time:491133ms step_avg:193.36ms
step:2541/4740 train_time:491316ms step_avg:193.36ms
step:2542/4740 train_time:491632ms step_avg:193.40ms
step:2543/4740 train_time:491814ms step_avg:193.40ms
step:2544/4740 train_time:492128ms step_avg:193.45ms
step:2545/4740 train_time:492312ms step_avg:193.44ms
step:2546/4740 train_time:492630ms step_avg:193.49ms
step:2547/4740 train_time:492809ms step_avg:193.49ms
step:2548/4740 train_time:493128ms step_avg:193.54ms
step:2549/4740 train_time:493306ms step_avg:193.53ms
step:2550/4740 train_time:493629ms step_avg:193.58ms
step:2551/4740 train_time:493804ms step_avg:193.57ms
step:2552/4740 train_time:494123ms step_avg:193.62ms
step:2553/4740 train_time:494301ms step_avg:193.62ms
step:2554/4740 train_time:494623ms step_avg:193.67ms
step:2555/4740 train_time:494799ms step_avg:193.66ms
step:2556/4740 train_time:495121ms step_avg:193.71ms
step:2557/4740 train_time:495301ms step_avg:193.70ms
step:2558/4740 train_time:495626ms step_avg:193.76ms
step:2559/4740 train_time:495804ms step_avg:193.75ms
step:2560/4740 train_time:496123ms step_avg:193.80ms
step:2561/4740 train_time:496298ms step_avg:193.79ms
step:2562/4740 train_time:496617ms step_avg:193.84ms
step:2563/4740 train_time:496796ms step_avg:193.83ms
step:2564/4740 train_time:497116ms step_avg:193.88ms
step:2565/4740 train_time:497296ms step_avg:193.88ms
step:2566/4740 train_time:497614ms step_avg:193.93ms
step:2567/4740 train_time:497797ms step_avg:193.92ms
step:2568/4740 train_time:498115ms step_avg:193.97ms
step:2569/4740 train_time:498294ms step_avg:193.96ms
step:2570/4740 train_time:498611ms step_avg:194.01ms
step:2571/4740 train_time:498792ms step_avg:194.01ms
step:2572/4740 train_time:499112ms step_avg:194.06ms
step:2573/4740 train_time:499289ms step_avg:194.05ms
step:2574/4740 train_time:499609ms step_avg:194.10ms
step:2575/4740 train_time:499785ms step_avg:194.09ms
step:2576/4740 train_time:500105ms step_avg:194.14ms
step:2577/4740 train_time:500288ms step_avg:194.14ms
step:2578/4740 train_time:500608ms step_avg:194.18ms
step:2579/4740 train_time:500786ms step_avg:194.18ms
step:2580/4740 train_time:501106ms step_avg:194.23ms
step:2581/4740 train_time:501282ms step_avg:194.22ms
step:2582/4740 train_time:501603ms step_avg:194.27ms
step:2583/4740 train_time:501778ms step_avg:194.26ms
step:2584/4740 train_time:502098ms step_avg:194.31ms
step:2585/4740 train_time:502277ms step_avg:194.30ms
step:2586/4740 train_time:502595ms step_avg:194.35ms
step:2587/4740 train_time:502773ms step_avg:194.35ms
step:2588/4740 train_time:503095ms step_avg:194.40ms
step:2589/4740 train_time:503274ms step_avg:194.39ms
step:2590/4740 train_time:503589ms step_avg:194.44ms
step:2591/4740 train_time:503769ms step_avg:194.43ms
step:2592/4740 train_time:504086ms step_avg:194.48ms
step:2593/4740 train_time:504269ms step_avg:194.47ms
step:2594/4740 train_time:504585ms step_avg:194.52ms
step:2595/4740 train_time:504770ms step_avg:194.52ms
step:2596/4740 train_time:505087ms step_avg:194.56ms
step:2597/4740 train_time:505268ms step_avg:194.56ms
step:2598/4740 train_time:505589ms step_avg:194.61ms
step:2599/4740 train_time:505769ms step_avg:194.60ms
step:2600/4740 train_time:506092ms step_avg:194.65ms
step:2601/4740 train_time:506269ms step_avg:194.64ms
step:2602/4740 train_time:506587ms step_avg:194.69ms
step:2603/4740 train_time:506766ms step_avg:194.69ms
step:2604/4740 train_time:507089ms step_avg:194.73ms
step:2605/4740 train_time:507266ms step_avg:194.73ms
step:2606/4740 train_time:507585ms step_avg:194.78ms
step:2607/4740 train_time:507765ms step_avg:194.77ms
step:2608/4740 train_time:508083ms step_avg:194.82ms
step:2609/4740 train_time:508261ms step_avg:194.81ms
step:2610/4740 train_time:508582ms step_avg:194.86ms
step:2611/4740 train_time:508757ms step_avg:194.85ms
step:2612/4740 train_time:509076ms step_avg:194.90ms
step:2613/4740 train_time:509257ms step_avg:194.89ms
step:2614/4740 train_time:509576ms step_avg:194.94ms
step:2615/4740 train_time:509754ms step_avg:194.93ms
step:2616/4740 train_time:510075ms step_avg:194.98ms
step:2617/4740 train_time:510256ms step_avg:194.98ms
step:2618/4740 train_time:510576ms step_avg:195.03ms
step:2619/4740 train_time:510755ms step_avg:195.02ms
step:2620/4740 train_time:511073ms step_avg:195.07ms
step:2621/4740 train_time:511252ms step_avg:195.06ms
step:2622/4740 train_time:511572ms step_avg:195.11ms
step:2623/4740 train_time:511752ms step_avg:195.10ms
step:2624/4740 train_time:512073ms step_avg:195.15ms
step:2625/4740 train_time:512250ms step_avg:195.14ms
step:2626/4740 train_time:512564ms step_avg:195.19ms
step:2627/4740 train_time:512745ms step_avg:195.18ms
step:2628/4740 train_time:513066ms step_avg:195.23ms
step:2629/4740 train_time:513241ms step_avg:195.22ms
step:2630/4740 train_time:513563ms step_avg:195.27ms
step:2631/4740 train_time:513739ms step_avg:195.26ms
step:2632/4740 train_time:514057ms step_avg:195.31ms
step:2633/4740 train_time:514234ms step_avg:195.30ms
step:2634/4740 train_time:514552ms step_avg:195.35ms
step:2635/4740 train_time:514734ms step_avg:195.34ms
step:2636/4740 train_time:515051ms step_avg:195.39ms
step:2637/4740 train_time:515229ms step_avg:195.38ms
step:2638/4740 train_time:515548ms step_avg:195.43ms
step:2639/4740 train_time:515729ms step_avg:195.43ms
step:2640/4740 train_time:516044ms step_avg:195.47ms
step:2641/4740 train_time:516225ms step_avg:195.47ms
step:2642/4740 train_time:516545ms step_avg:195.51ms
step:2643/4740 train_time:516721ms step_avg:195.51ms
step:2644/4740 train_time:517040ms step_avg:195.55ms
step:2645/4740 train_time:517218ms step_avg:195.55ms
step:2646/4740 train_time:517535ms step_avg:195.59ms
step:2647/4740 train_time:517717ms step_avg:195.59ms
step:2648/4740 train_time:518035ms step_avg:195.63ms
step:2649/4740 train_time:518215ms step_avg:195.63ms
step:2650/4740 train_time:518533ms step_avg:195.67ms
step:2651/4740 train_time:518713ms step_avg:195.67ms
step:2652/4740 train_time:519032ms step_avg:195.71ms
step:2653/4740 train_time:519209ms step_avg:195.71ms
step:2654/4740 train_time:519527ms step_avg:195.75ms
step:2655/4740 train_time:519705ms step_avg:195.75ms
step:2656/4740 train_time:520027ms step_avg:195.79ms
step:2657/4740 train_time:520204ms step_avg:195.79ms
step:2658/4740 train_time:520523ms step_avg:195.83ms
step:2659/4740 train_time:520703ms step_avg:195.83ms
step:2660/4740 train_time:521025ms step_avg:195.87ms
step:2661/4740 train_time:521200ms step_avg:195.87ms
step:2662/4740 train_time:521523ms step_avg:195.91ms
step:2663/4740 train_time:521698ms step_avg:195.91ms
step:2664/4740 train_time:522017ms step_avg:195.95ms
step:2665/4740 train_time:522196ms step_avg:195.95ms
step:2666/4740 train_time:522516ms step_avg:195.99ms
step:2667/4740 train_time:522696ms step_avg:195.99ms
step:2668/4740 train_time:523013ms step_avg:196.03ms
step:2669/4740 train_time:523191ms step_avg:196.02ms
step:2670/4740 train_time:523510ms step_avg:196.07ms
step:2671/4740 train_time:523687ms step_avg:196.06ms
step:2672/4740 train_time:524005ms step_avg:196.11ms
step:2673/4740 train_time:524183ms step_avg:196.10ms
step:2674/4740 train_time:524504ms step_avg:196.15ms
step:2675/4740 train_time:524682ms step_avg:196.14ms
step:2676/4740 train_time:525002ms step_avg:196.19ms
step:2677/4740 train_time:525183ms step_avg:196.18ms
step:2678/4740 train_time:525505ms step_avg:196.23ms
step:2679/4740 train_time:525684ms step_avg:196.22ms
step:2680/4740 train_time:526004ms step_avg:196.27ms
step:2681/4740 train_time:526181ms step_avg:196.26ms
step:2682/4740 train_time:526501ms step_avg:196.31ms
step:2683/4740 train_time:526678ms step_avg:196.30ms
step:2684/4740 train_time:527000ms step_avg:196.35ms
step:2685/4740 train_time:527178ms step_avg:196.34ms
step:2686/4740 train_time:527495ms step_avg:196.39ms
step:2687/4740 train_time:527673ms step_avg:196.38ms
step:2688/4740 train_time:527995ms step_avg:196.43ms
step:2689/4740 train_time:528173ms step_avg:196.42ms
step:2690/4740 train_time:528491ms step_avg:196.46ms
step:2691/4740 train_time:528665ms step_avg:196.46ms
step:2692/4740 train_time:528986ms step_avg:196.50ms
step:2693/4740 train_time:529163ms step_avg:196.50ms
step:2694/4740 train_time:529485ms step_avg:196.54ms
step:2695/4740 train_time:529663ms step_avg:196.54ms
step:2696/4740 train_time:529985ms step_avg:196.58ms
step:2697/4740 train_time:530160ms step_avg:196.57ms
step:2698/4740 train_time:530483ms step_avg:196.62ms
step:2699/4740 train_time:530658ms step_avg:196.61ms
step:2700/4740 train_time:530978ms step_avg:196.66ms
step:2701/4740 train_time:531157ms step_avg:196.65ms
step:2702/4740 train_time:531476ms step_avg:196.70ms
step:2703/4740 train_time:531654ms step_avg:196.69ms
step:2704/4740 train_time:531971ms step_avg:196.73ms
step:2705/4740 train_time:532154ms step_avg:196.73ms
step:2706/4740 train_time:532472ms step_avg:196.77ms
step:2707/4740 train_time:532649ms step_avg:196.77ms
step:2708/4740 train_time:532969ms step_avg:196.81ms
step:2709/4740 train_time:533146ms step_avg:196.81ms
step:2710/4740 train_time:533468ms step_avg:196.85ms
step:2711/4740 train_time:533644ms step_avg:196.84ms
step:2712/4740 train_time:533968ms step_avg:196.89ms
step:2713/4740 train_time:534141ms step_avg:196.88ms
step:2714/4740 train_time:534463ms step_avg:196.93ms
step:2715/4740 train_time:534639ms step_avg:196.92ms
step:2716/4740 train_time:534963ms step_avg:196.97ms
step:2717/4740 train_time:535140ms step_avg:196.96ms
step:2718/4740 train_time:535464ms step_avg:197.01ms
step:2719/4740 train_time:535640ms step_avg:197.00ms
step:2720/4740 train_time:535961ms step_avg:197.04ms
step:2721/4740 train_time:536136ms step_avg:197.04ms
step:2722/4740 train_time:536452ms step_avg:197.08ms
step:2723/4740 train_time:536635ms step_avg:197.07ms
step:2724/4740 train_time:536957ms step_avg:197.12ms
step:2725/4740 train_time:537133ms step_avg:197.11ms
step:2726/4740 train_time:537447ms step_avg:197.16ms
step:2727/4740 train_time:537628ms step_avg:197.15ms
step:2728/4740 train_time:537945ms step_avg:197.19ms
step:2729/4740 train_time:538125ms step_avg:197.19ms
step:2730/4740 train_time:538444ms step_avg:197.23ms
step:2731/4740 train_time:538626ms step_avg:197.23ms
step:2732/4740 train_time:538946ms step_avg:197.27ms
step:2733/4740 train_time:539124ms step_avg:197.26ms
step:2734/4740 train_time:539445ms step_avg:197.31ms
step:2735/4740 train_time:539622ms step_avg:197.30ms
step:2736/4740 train_time:539939ms step_avg:197.35ms
step:2737/4740 train_time:540118ms step_avg:197.34ms
step:2738/4740 train_time:540441ms step_avg:197.39ms
step:2739/4740 train_time:540619ms step_avg:197.38ms
step:2740/4740 train_time:540937ms step_avg:197.42ms
step:2741/4740 train_time:541115ms step_avg:197.42ms
step:2742/4740 train_time:541431ms step_avg:197.46ms
step:2743/4740 train_time:541613ms step_avg:197.45ms
step:2744/4740 train_time:541931ms step_avg:197.50ms
step:2745/4740 train_time:542108ms step_avg:197.49ms
step:2746/4740 train_time:542429ms step_avg:197.53ms
step:2747/4740 train_time:542605ms step_avg:197.53ms
step:2748/4740 train_time:542924ms step_avg:197.57ms
step:2749/4740 train_time:543102ms step_avg:197.56ms
step:2750/4740 train_time:543425ms step_avg:197.61ms
step:2750/4740 val_loss:3.1286 train_time:543547ms step_avg:197.65ms
step:2751/4740 train_time:543604ms step_avg:197.60ms
step:2752/4740 train_time:543932ms step_avg:197.65ms
step:2753/4740 train_time:544109ms step_avg:197.64ms
step:2754/4740 train_time:544424ms step_avg:197.68ms
step:2755/4740 train_time:544601ms step_avg:197.68ms
step:2756/4740 train_time:544930ms step_avg:197.73ms
step:2757/4740 train_time:545111ms step_avg:197.72ms
step:2758/4740 train_time:545429ms step_avg:197.76ms
step:2759/4740 train_time:545606ms step_avg:197.75ms
step:2760/4740 train_time:545930ms step_avg:197.80ms
step:2761/4740 train_time:546108ms step_avg:197.79ms
step:2762/4740 train_time:546429ms step_avg:197.84ms
step:2763/4740 train_time:546610ms step_avg:197.83ms
step:2764/4740 train_time:546935ms step_avg:197.88ms
step:2765/4740 train_time:547111ms step_avg:197.87ms
step:2766/4740 train_time:547432ms step_avg:197.91ms
step:2767/4740 train_time:547609ms step_avg:197.91ms
step:2768/4740 train_time:547933ms step_avg:197.95ms
step:2769/4740 train_time:548108ms step_avg:197.94ms
step:2770/4740 train_time:548429ms step_avg:197.99ms
step:2771/4740 train_time:548607ms step_avg:197.98ms
step:2772/4740 train_time:548930ms step_avg:198.03ms
step:2773/4740 train_time:549111ms step_avg:198.02ms
step:2774/4740 train_time:549434ms step_avg:198.07ms
step:2775/4740 train_time:549609ms step_avg:198.06ms
step:2776/4740 train_time:549933ms step_avg:198.10ms
step:2777/4740 train_time:550111ms step_avg:198.10ms
step:2778/4740 train_time:550432ms step_avg:198.14ms
step:2779/4740 train_time:550607ms step_avg:198.13ms
step:2780/4740 train_time:550930ms step_avg:198.18ms
step:2781/4740 train_time:551107ms step_avg:198.17ms
step:2782/4740 train_time:551427ms step_avg:198.21ms
step:2783/4740 train_time:551608ms step_avg:198.21ms
step:2784/4740 train_time:551929ms step_avg:198.25ms
step:2785/4740 train_time:552104ms step_avg:198.24ms
step:2786/4740 train_time:552426ms step_avg:198.29ms
step:2787/4740 train_time:552608ms step_avg:198.28ms
step:2788/4740 train_time:552927ms step_avg:198.32ms
step:2789/4740 train_time:553106ms step_avg:198.32ms
step:2790/4740 train_time:553427ms step_avg:198.36ms
step:2791/4740 train_time:553606ms step_avg:198.35ms
step:2792/4740 train_time:553928ms step_avg:198.40ms
step:2793/4740 train_time:554108ms step_avg:198.39ms
step:2794/4740 train_time:554429ms step_avg:198.44ms
step:2795/4740 train_time:554606ms step_avg:198.43ms
step:2796/4740 train_time:554928ms step_avg:198.47ms
step:2797/4740 train_time:555105ms step_avg:198.46ms
step:2798/4740 train_time:555426ms step_avg:198.51ms
step:2799/4740 train_time:555606ms step_avg:198.50ms
step:2800/4740 train_time:555928ms step_avg:198.55ms
step:2801/4740 train_time:556106ms step_avg:198.54ms
step:2802/4740 train_time:556428ms step_avg:198.58ms
step:2803/4740 train_time:556604ms step_avg:198.57ms
step:2804/4740 train_time:556925ms step_avg:198.62ms
step:2805/4740 train_time:557104ms step_avg:198.61ms
step:2806/4740 train_time:557422ms step_avg:198.65ms
step:2807/4740 train_time:557605ms step_avg:198.65ms
step:2808/4740 train_time:557927ms step_avg:198.69ms
step:2809/4740 train_time:558107ms step_avg:198.69ms
step:2810/4740 train_time:558429ms step_avg:198.73ms
step:2811/4740 train_time:558607ms step_avg:198.72ms
step:2812/4740 train_time:558928ms step_avg:198.77ms
step:2813/4740 train_time:559105ms step_avg:198.76ms
step:2814/4740 train_time:559428ms step_avg:198.80ms
step:2815/4740 train_time:559604ms step_avg:198.79ms
step:2816/4740 train_time:559925ms step_avg:198.84ms
step:2817/4740 train_time:560103ms step_avg:198.83ms
step:2818/4740 train_time:560425ms step_avg:198.87ms
step:2819/4740 train_time:560601ms step_avg:198.87ms
step:2820/4740 train_time:560919ms step_avg:198.91ms
step:2821/4740 train_time:561098ms step_avg:198.90ms
step:2822/4740 train_time:561421ms step_avg:198.94ms
step:2823/4740 train_time:561599ms step_avg:198.94ms
step:2824/4740 train_time:561915ms step_avg:198.98ms
step:2825/4740 train_time:562094ms step_avg:198.97ms
step:2826/4740 train_time:562413ms step_avg:199.01ms
step:2827/4740 train_time:562592ms step_avg:199.01ms
step:2828/4740 train_time:562910ms step_avg:199.05ms
step:2829/4740 train_time:563091ms step_avg:199.04ms
step:2830/4740 train_time:563411ms step_avg:199.09ms
step:2831/4740 train_time:563591ms step_avg:199.08ms
step:2832/4740 train_time:563910ms step_avg:199.12ms
step:2833/4740 train_time:564089ms step_avg:199.11ms
step:2834/4740 train_time:564411ms step_avg:199.16ms
step:2835/4740 train_time:564588ms step_avg:199.15ms
step:2836/4740 train_time:564910ms step_avg:199.19ms
step:2837/4740 train_time:565087ms step_avg:199.18ms
step:2838/4740 train_time:565408ms step_avg:199.23ms
step:2839/4740 train_time:565584ms step_avg:199.22ms
step:2840/4740 train_time:565903ms step_avg:199.26ms
step:2841/4740 train_time:566084ms step_avg:199.26ms
step:2842/4740 train_time:566404ms step_avg:199.30ms
step:2843/4740 train_time:566583ms step_avg:199.29ms
step:2844/4740 train_time:566901ms step_avg:199.33ms
step:2845/4740 train_time:567080ms step_avg:199.33ms
step:2846/4740 train_time:567400ms step_avg:199.37ms
step:2847/4740 train_time:567579ms step_avg:199.36ms
step:2848/4740 train_time:567898ms step_avg:199.40ms
step:2849/4740 train_time:568081ms step_avg:199.40ms
step:2850/4740 train_time:568401ms step_avg:199.44ms
step:2851/4740 train_time:568584ms step_avg:199.43ms
step:2852/4740 train_time:568904ms step_avg:199.48ms
step:2853/4740 train_time:569084ms step_avg:199.47ms
step:2854/4740 train_time:569407ms step_avg:199.51ms
step:2855/4740 train_time:569584ms step_avg:199.50ms
step:2856/4740 train_time:569897ms step_avg:199.54ms
step:2857/4740 train_time:570079ms step_avg:199.54ms
step:2858/4740 train_time:570402ms step_avg:199.58ms
step:2859/4740 train_time:570581ms step_avg:199.57ms
step:2860/4740 train_time:570899ms step_avg:199.62ms
step:2861/4740 train_time:571083ms step_avg:199.61ms
step:2862/4740 train_time:571400ms step_avg:199.65ms
step:2863/4740 train_time:571579ms step_avg:199.64ms
step:2864/4740 train_time:571900ms step_avg:199.69ms
step:2865/4740 train_time:572078ms step_avg:199.68ms
step:2866/4740 train_time:572399ms step_avg:199.72ms
step:2867/4740 train_time:572578ms step_avg:199.71ms
step:2868/4740 train_time:572897ms step_avg:199.75ms
step:2869/4740 train_time:573077ms step_avg:199.75ms
step:2870/4740 train_time:573393ms step_avg:199.79ms
step:2871/4740 train_time:573575ms step_avg:199.78ms
step:2872/4740 train_time:573897ms step_avg:199.82ms
step:2873/4740 train_time:574078ms step_avg:199.82ms
step:2874/4740 train_time:574395ms step_avg:199.86ms
step:2875/4740 train_time:574574ms step_avg:199.85ms
step:2876/4740 train_time:574891ms step_avg:199.89ms
step:2877/4740 train_time:575070ms step_avg:199.89ms
step:2878/4740 train_time:575392ms step_avg:199.93ms
step:2879/4740 train_time:575568ms step_avg:199.92ms
step:2880/4740 train_time:575892ms step_avg:199.96ms
step:2881/4740 train_time:576068ms step_avg:199.95ms
step:2882/4740 train_time:576392ms step_avg:200.00ms
step:2883/4740 train_time:576567ms step_avg:199.99ms
step:2884/4740 train_time:576890ms step_avg:200.03ms
step:2885/4740 train_time:577069ms step_avg:200.02ms
step:2886/4740 train_time:577393ms step_avg:200.07ms
step:2887/4740 train_time:577569ms step_avg:200.06ms
step:2888/4740 train_time:577892ms step_avg:200.10ms
step:2889/4740 train_time:578069ms step_avg:200.09ms
step:2890/4740 train_time:578393ms step_avg:200.14ms
step:2891/4740 train_time:578569ms step_avg:200.13ms
step:2892/4740 train_time:578892ms step_avg:200.17ms
step:2893/4740 train_time:579070ms step_avg:200.16ms
step:2894/4740 train_time:579392ms step_avg:200.20ms
step:2895/4740 train_time:579569ms step_avg:200.20ms
step:2896/4740 train_time:579890ms step_avg:200.24ms
step:2897/4740 train_time:580066ms step_avg:200.23ms
step:2898/4740 train_time:580389ms step_avg:200.27ms
step:2899/4740 train_time:580567ms step_avg:200.26ms
step:2900/4740 train_time:580888ms step_avg:200.31ms
step:2901/4740 train_time:581066ms step_avg:200.30ms
step:2902/4740 train_time:581390ms step_avg:200.34ms
step:2903/4740 train_time:581565ms step_avg:200.33ms
step:2904/4740 train_time:581889ms step_avg:200.37ms
step:2905/4740 train_time:582066ms step_avg:200.37ms
step:2906/4740 train_time:582389ms step_avg:200.41ms
step:2907/4740 train_time:582566ms step_avg:200.40ms
step:2908/4740 train_time:582889ms step_avg:200.44ms
step:2909/4740 train_time:583065ms step_avg:200.43ms
step:2910/4740 train_time:583388ms step_avg:200.48ms
step:2911/4740 train_time:583566ms step_avg:200.47ms
step:2912/4740 train_time:583884ms step_avg:200.51ms
step:2913/4740 train_time:584063ms step_avg:200.50ms
step:2914/4740 train_time:584387ms step_avg:200.54ms
step:2915/4740 train_time:584563ms step_avg:200.54ms
step:2916/4740 train_time:584879ms step_avg:200.58ms
step:2917/4740 train_time:585062ms step_avg:200.57ms
step:2918/4740 train_time:585381ms step_avg:200.61ms
step:2919/4740 train_time:585563ms step_avg:200.60ms
step:2920/4740 train_time:585880ms step_avg:200.64ms
step:2921/4740 train_time:586060ms step_avg:200.64ms
step:2922/4740 train_time:586376ms step_avg:200.68ms
step:2923/4740 train_time:586560ms step_avg:200.67ms
step:2924/4740 train_time:586878ms step_avg:200.71ms
step:2925/4740 train_time:587059ms step_avg:200.70ms
step:2926/4740 train_time:587380ms step_avg:200.75ms
step:2927/4740 train_time:587560ms step_avg:200.74ms
step:2928/4740 train_time:587880ms step_avg:200.78ms
step:2929/4740 train_time:588063ms step_avg:200.77ms
step:2930/4740 train_time:588384ms step_avg:200.81ms
step:2931/4740 train_time:588565ms step_avg:200.81ms
step:2932/4740 train_time:588885ms step_avg:200.85ms
step:2933/4740 train_time:589065ms step_avg:200.84ms
step:2934/4740 train_time:589387ms step_avg:200.88ms
step:2935/4740 train_time:589566ms step_avg:200.87ms
step:2936/4740 train_time:589887ms step_avg:200.92ms
step:2937/4740 train_time:590063ms step_avg:200.91ms
step:2938/4740 train_time:590379ms step_avg:200.95ms
step:2939/4740 train_time:590562ms step_avg:200.94ms
step:2940/4740 train_time:590884ms step_avg:200.98ms
step:2941/4740 train_time:591060ms step_avg:200.97ms
step:2942/4740 train_time:591381ms step_avg:201.01ms
step:2943/4740 train_time:591563ms step_avg:201.01ms
step:2944/4740 train_time:591881ms step_avg:201.05ms
step:2945/4740 train_time:592061ms step_avg:201.04ms
step:2946/4740 train_time:592379ms step_avg:201.08ms
step:2947/4740 train_time:592561ms step_avg:201.07ms
step:2948/4740 train_time:592882ms step_avg:201.11ms
step:2949/4740 train_time:593061ms step_avg:201.11ms
step:2950/4740 train_time:593380ms step_avg:201.15ms
step:2951/4740 train_time:593561ms step_avg:201.14ms
step:2952/4740 train_time:593882ms step_avg:201.18ms
step:2953/4740 train_time:594057ms step_avg:201.17ms
step:2954/4740 train_time:594376ms step_avg:201.21ms
step:2955/4740 train_time:594555ms step_avg:201.20ms
step:2956/4740 train_time:594872ms step_avg:201.24ms
step:2957/4740 train_time:595050ms step_avg:201.23ms
step:2958/4740 train_time:595371ms step_avg:201.27ms
step:2959/4740 train_time:595548ms step_avg:201.27ms
step:2960/4740 train_time:595869ms step_avg:201.31ms
step:2961/4740 train_time:596047ms step_avg:201.30ms
step:2962/4740 train_time:596370ms step_avg:201.34ms
step:2963/4740 train_time:596546ms step_avg:201.33ms
step:2964/4740 train_time:596866ms step_avg:201.37ms
step:2965/4740 train_time:597047ms step_avg:201.36ms
step:2966/4740 train_time:597368ms step_avg:201.41ms
step:2967/4740 train_time:597547ms step_avg:201.40ms
step:2968/4740 train_time:597868ms step_avg:201.44ms
step:2969/4740 train_time:598045ms step_avg:201.43ms
step:2970/4740 train_time:598367ms step_avg:201.47ms
step:2971/4740 train_time:598548ms step_avg:201.46ms
step:2972/4740 train_time:598868ms step_avg:201.50ms
step:2973/4740 train_time:599048ms step_avg:201.50ms
step:2974/4740 train_time:599366ms step_avg:201.54ms
step:2975/4740 train_time:599546ms step_avg:201.53ms
step:2976/4740 train_time:599867ms step_avg:201.57ms
step:2977/4740 train_time:600048ms step_avg:201.56ms
step:2978/4740 train_time:600366ms step_avg:201.60ms
step:2979/4740 train_time:600549ms step_avg:201.59ms
step:2980/4740 train_time:600872ms step_avg:201.63ms
step:2981/4740 train_time:601049ms step_avg:201.63ms
step:2982/4740 train_time:601369ms step_avg:201.67ms
step:2983/4740 train_time:601549ms step_avg:201.66ms
step:2984/4740 train_time:601870ms step_avg:201.70ms
step:2985/4740 train_time:602050ms step_avg:201.69ms
step:2986/4740 train_time:602370ms step_avg:201.73ms
step:2987/4740 train_time:602550ms step_avg:201.72ms
step:2988/4740 train_time:602868ms step_avg:201.76ms
step:2989/4740 train_time:603048ms step_avg:201.76ms
step:2990/4740 train_time:603370ms step_avg:201.80ms
step:2991/4740 train_time:603549ms step_avg:201.79ms
step:2992/4740 train_time:603868ms step_avg:201.83ms
step:2993/4740 train_time:604044ms step_avg:201.82ms
step:2994/4740 train_time:604364ms step_avg:201.86ms
step:2995/4740 train_time:604545ms step_avg:201.85ms
step:2996/4740 train_time:604863ms step_avg:201.89ms
step:2997/4740 train_time:605041ms step_avg:201.88ms
step:2998/4740 train_time:605360ms step_avg:201.92ms
step:2999/4740 train_time:605538ms step_avg:201.91ms
step:3000/4740 train_time:605857ms step_avg:201.95ms
step:3000/4740 val_loss:3.0978 train_time:605982ms step_avg:201.99ms
step:3001/4740 train_time:606040ms step_avg:201.95ms
step:3002/4740 train_time:606367ms step_avg:201.99ms
step:3003/4740 train_time:606544ms step_avg:201.98ms
step:3004/4740 train_time:606856ms step_avg:202.02ms
step:3005/4740 train_time:607040ms step_avg:202.01ms
step:3006/4740 train_time:607365ms step_avg:202.05ms
step:3007/4740 train_time:607543ms step_avg:202.04ms
step:3008/4740 train_time:607863ms step_avg:202.08ms
step:3009/4740 train_time:608042ms step_avg:202.07ms
step:3010/4740 train_time:608361ms step_avg:202.11ms
step:3011/4740 train_time:608542ms step_avg:202.11ms
step:3012/4740 train_time:608862ms step_avg:202.15ms
step:3013/4740 train_time:609040ms step_avg:202.14ms
step:3014/4740 train_time:609362ms step_avg:202.18ms
step:3015/4740 train_time:609539ms step_avg:202.17ms
step:3016/4740 train_time:609861ms step_avg:202.21ms
step:3017/4740 train_time:610039ms step_avg:202.20ms
step:3018/4740 train_time:610361ms step_avg:202.24ms
step:3019/4740 train_time:610539ms step_avg:202.23ms
step:3020/4740 train_time:610864ms step_avg:202.27ms
step:3021/4740 train_time:611039ms step_avg:202.26ms
step:3022/4740 train_time:611360ms step_avg:202.30ms
step:3023/4740 train_time:611538ms step_avg:202.30ms
step:3024/4740 train_time:611863ms step_avg:202.34ms
step:3025/4740 train_time:612040ms step_avg:202.33ms
step:3026/4740 train_time:612363ms step_avg:202.37ms
step:3027/4740 train_time:612542ms step_avg:202.36ms
step:3028/4740 train_time:612862ms step_avg:202.40ms
step:3029/4740 train_time:613040ms step_avg:202.39ms
step:3030/4740 train_time:613363ms step_avg:202.43ms
step:3031/4740 train_time:613540ms step_avg:202.42ms
step:3032/4740 train_time:613861ms step_avg:202.46ms
step:3033/4740 train_time:614038ms step_avg:202.45ms
step:3034/4740 train_time:614360ms step_avg:202.49ms
step:3035/4740 train_time:614537ms step_avg:202.48ms
step:3036/4740 train_time:614862ms step_avg:202.52ms
step:3037/4740 train_time:615041ms step_avg:202.52ms
step:3038/4740 train_time:615359ms step_avg:202.55ms
step:3039/4740 train_time:615539ms step_avg:202.55ms
step:3040/4740 train_time:615859ms step_avg:202.59ms
step:3041/4740 train_time:616036ms step_avg:202.58ms
step:3042/4740 train_time:616357ms step_avg:202.62ms
step:3043/4740 train_time:616535ms step_avg:202.61ms
step:3044/4740 train_time:616856ms step_avg:202.65ms
step:3045/4740 train_time:617034ms step_avg:202.64ms
step:3046/4740 train_time:617357ms step_avg:202.68ms
step:3047/4740 train_time:617533ms step_avg:202.67ms
step:3048/4740 train_time:617855ms step_avg:202.71ms
step:3049/4740 train_time:618029ms step_avg:202.70ms
step:3050/4740 train_time:618353ms step_avg:202.74ms
step:3051/4740 train_time:618530ms step_avg:202.73ms
step:3052/4740 train_time:618852ms step_avg:202.77ms
step:3053/4740 train_time:619029ms step_avg:202.76ms
step:3054/4740 train_time:619350ms step_avg:202.80ms
step:3055/4740 train_time:619530ms step_avg:202.79ms
step:3056/4740 train_time:619852ms step_avg:202.83ms
step:3057/4740 train_time:620028ms step_avg:202.82ms
step:3058/4740 train_time:620349ms step_avg:202.86ms
step:3059/4740 train_time:620527ms step_avg:202.85ms
step:3060/4740 train_time:620847ms step_avg:202.89ms
step:3061/4740 train_time:621026ms step_avg:202.88ms
step:3062/4740 train_time:621345ms step_avg:202.92ms
step:3063/4740 train_time:621525ms step_avg:202.91ms
step:3064/4740 train_time:621841ms step_avg:202.95ms
step:3065/4740 train_time:622023ms step_avg:202.94ms
step:3066/4740 train_time:622343ms step_avg:202.98ms
step:3067/4740 train_time:622522ms step_avg:202.97ms
step:3068/4740 train_time:622840ms step_avg:203.01ms
step:3069/4740 train_time:623024ms step_avg:203.01ms
step:3070/4740 train_time:623342ms step_avg:203.04ms
step:3071/4740 train_time:623520ms step_avg:203.03ms
step:3072/4740 train_time:623843ms step_avg:203.07ms
step:3073/4740 train_time:624022ms step_avg:203.07ms
step:3074/4740 train_time:624341ms step_avg:203.10ms
step:3075/4740 train_time:624518ms step_avg:203.10ms
step:3076/4740 train_time:624838ms step_avg:203.13ms
step:3077/4740 train_time:625016ms step_avg:203.13ms
step:3078/4740 train_time:625341ms step_avg:203.16ms
step:3079/4740 train_time:625517ms step_avg:203.16ms
step:3080/4740 train_time:625840ms step_avg:203.19ms
step:3081/4740 train_time:626016ms step_avg:203.19ms
step:3082/4740 train_time:626340ms step_avg:203.23ms
step:3083/4740 train_time:626514ms step_avg:203.22ms
step:3084/4740 train_time:626839ms step_avg:203.26ms
step:3085/4740 train_time:627015ms step_avg:203.25ms
step:3086/4740 train_time:627339ms step_avg:203.29ms
step:3087/4740 train_time:627514ms step_avg:203.28ms
step:3088/4740 train_time:627837ms step_avg:203.32ms
step:3089/4740 train_time:628013ms step_avg:203.31ms
step:3090/4740 train_time:628338ms step_avg:203.35ms
step:3091/4740 train_time:628514ms step_avg:203.34ms
step:3092/4740 train_time:628838ms step_avg:203.38ms
step:3093/4740 train_time:629012ms step_avg:203.37ms
step:3094/4740 train_time:629334ms step_avg:203.40ms
step:3095/4740 train_time:629510ms step_avg:203.40ms
step:3096/4740 train_time:629830ms step_avg:203.43ms
step:3097/4740 train_time:630007ms step_avg:203.43ms
step:3098/4740 train_time:630325ms step_avg:203.46ms
step:3099/4740 train_time:630509ms step_avg:203.46ms
step:3100/4740 train_time:630833ms step_avg:203.49ms
step:3101/4740 train_time:631011ms step_avg:203.49ms
step:3102/4740 train_time:631331ms step_avg:203.52ms
step:3103/4740 train_time:631507ms step_avg:203.52ms
step:3104/4740 train_time:631827ms step_avg:203.55ms
step:3105/4740 train_time:632010ms step_avg:203.55ms
step:3106/4740 train_time:632327ms step_avg:203.58ms
step:3107/4740 train_time:632508ms step_avg:203.58ms
step:3108/4740 train_time:632830ms step_avg:203.61ms
step:3109/4740 train_time:633009ms step_avg:203.61ms
step:3110/4740 train_time:633327ms step_avg:203.64ms
step:3111/4740 train_time:633505ms step_avg:203.63ms
step:3112/4740 train_time:633826ms step_avg:203.67ms
step:3113/4740 train_time:634003ms step_avg:203.66ms
step:3114/4740 train_time:634319ms step_avg:203.70ms
step:3115/4740 train_time:634503ms step_avg:203.69ms
step:3116/4740 train_time:634822ms step_avg:203.73ms
step:3117/4740 train_time:635000ms step_avg:203.72ms
step:3118/4740 train_time:635319ms step_avg:203.76ms
step:3119/4740 train_time:635496ms step_avg:203.75ms
step:3120/4740 train_time:635817ms step_avg:203.79ms
step:3121/4740 train_time:635992ms step_avg:203.78ms
step:3122/4740 train_time:636317ms step_avg:203.82ms
step:3123/4740 train_time:636493ms step_avg:203.81ms
step:3124/4740 train_time:636815ms step_avg:203.85ms
step:3125/4740 train_time:636990ms step_avg:203.84ms
step:3126/4740 train_time:637311ms step_avg:203.87ms
step:3127/4740 train_time:637489ms step_avg:203.87ms
step:3128/4740 train_time:637811ms step_avg:203.90ms
step:3129/4740 train_time:637990ms step_avg:203.90ms
step:3130/4740 train_time:638314ms step_avg:203.93ms
step:3131/4740 train_time:638492ms step_avg:203.93ms
step:3132/4740 train_time:638810ms step_avg:203.96ms
step:3133/4740 train_time:638991ms step_avg:203.96ms
step:3134/4740 train_time:639316ms step_avg:203.99ms
step:3135/4740 train_time:639492ms step_avg:203.98ms
step:3136/4740 train_time:639814ms step_avg:204.02ms
step:3137/4740 train_time:639991ms step_avg:204.01ms
step:3138/4740 train_time:640315ms step_avg:204.05ms
step:3139/4740 train_time:640490ms step_avg:204.04ms
step:3140/4740 train_time:640811ms step_avg:204.08ms
step:3141/4740 train_time:640987ms step_avg:204.07ms
step:3142/4740 train_time:641312ms step_avg:204.11ms
step:3143/4740 train_time:641489ms step_avg:204.10ms
step:3144/4740 train_time:641810ms step_avg:204.14ms
step:3145/4740 train_time:641989ms step_avg:204.13ms
step:3146/4740 train_time:642312ms step_avg:204.17ms
step:3147/4740 train_time:642489ms step_avg:204.16ms
step:3148/4740 train_time:642813ms step_avg:204.20ms
step:3149/4740 train_time:642990ms step_avg:204.19ms
step:3150/4740 train_time:643312ms step_avg:204.23ms
step:3151/4740 train_time:643489ms step_avg:204.22ms
step:3152/4740 train_time:643812ms step_avg:204.26ms
step:3153/4740 train_time:643990ms step_avg:204.25ms
step:3154/4740 train_time:644312ms step_avg:204.28ms
step:3155/4740 train_time:644491ms step_avg:204.28ms
step:3156/4740 train_time:644812ms step_avg:204.31ms
step:3157/4740 train_time:644991ms step_avg:204.30ms
step:3158/4740 train_time:645312ms step_avg:204.34ms
step:3159/4740 train_time:645489ms step_avg:204.33ms
step:3160/4740 train_time:645809ms step_avg:204.37ms
step:3161/4740 train_time:645991ms step_avg:204.36ms
step:3162/4740 train_time:646315ms step_avg:204.40ms
step:3163/4740 train_time:646491ms step_avg:204.39ms
step:3164/4740 train_time:646816ms step_avg:204.43ms
step:3165/4740 train_time:646992ms step_avg:204.42ms
step:3166/4740 train_time:647315ms step_avg:204.46ms
step:3167/4740 train_time:647492ms step_avg:204.45ms
step:3168/4740 train_time:647815ms step_avg:204.49ms
step:3169/4740 train_time:647990ms step_avg:204.48ms
step:3170/4740 train_time:648312ms step_avg:204.51ms
step:3171/4740 train_time:648489ms step_avg:204.51ms
step:3172/4740 train_time:648810ms step_avg:204.54ms
step:3173/4740 train_time:648991ms step_avg:204.54ms
step:3174/4740 train_time:649312ms step_avg:204.57ms
step:3175/4740 train_time:649487ms step_avg:204.56ms
step:3176/4740 train_time:649806ms step_avg:204.60ms
step:3177/4740 train_time:649987ms step_avg:204.59ms
step:3178/4740 train_time:650314ms step_avg:204.63ms
step:3179/4740 train_time:650491ms step_avg:204.62ms
step:3180/4740 train_time:650815ms step_avg:204.66ms
step:3181/4740 train_time:650991ms step_avg:204.65ms
step:3182/4740 train_time:651313ms step_avg:204.69ms
step:3183/4740 train_time:651491ms step_avg:204.68ms
step:3184/4740 train_time:651815ms step_avg:204.72ms
step:3185/4740 train_time:651993ms step_avg:204.71ms
step:3186/4740 train_time:652318ms step_avg:204.74ms
step:3187/4740 train_time:652493ms step_avg:204.74ms
step:3188/4740 train_time:652816ms step_avg:204.77ms
step:3189/4740 train_time:652997ms step_avg:204.77ms
step:3190/4740 train_time:653321ms step_avg:204.80ms
step:3191/4740 train_time:653500ms step_avg:204.79ms
step:3192/4740 train_time:653825ms step_avg:204.83ms
step:3193/4740 train_time:654000ms step_avg:204.82ms
step:3194/4740 train_time:654319ms step_avg:204.86ms
step:3195/4740 train_time:654501ms step_avg:204.85ms
step:3196/4740 train_time:654824ms step_avg:204.89ms
step:3197/4740 train_time:655003ms step_avg:204.88ms
step:3198/4740 train_time:655325ms step_avg:204.92ms
step:3199/4740 train_time:655503ms step_avg:204.91ms
step:3200/4740 train_time:655824ms step_avg:204.95ms
step:3201/4740 train_time:656002ms step_avg:204.94ms
step:3202/4740 train_time:656321ms step_avg:204.97ms
step:3203/4740 train_time:656506ms step_avg:204.97ms
step:3204/4740 train_time:656824ms step_avg:205.00ms
step:3205/4740 train_time:657004ms step_avg:204.99ms
step:3206/4740 train_time:657325ms step_avg:205.03ms
step:3207/4740 train_time:657507ms step_avg:205.02ms
step:3208/4740 train_time:657827ms step_avg:205.06ms
step:3209/4740 train_time:658004ms step_avg:205.05ms
step:3210/4740 train_time:658328ms step_avg:205.09ms
step:3211/4740 train_time:658508ms step_avg:205.08ms
step:3212/4740 train_time:658829ms step_avg:205.12ms
step:3213/4740 train_time:659008ms step_avg:205.11ms
step:3214/4740 train_time:659327ms step_avg:205.14ms
step:3215/4740 train_time:659505ms step_avg:205.13ms
step:3216/4740 train_time:659825ms step_avg:205.17ms
step:3217/4740 train_time:660006ms step_avg:205.16ms
step:3218/4740 train_time:660325ms step_avg:205.20ms
step:3219/4740 train_time:660506ms step_avg:205.19ms
step:3220/4740 train_time:660827ms step_avg:205.23ms
step:3221/4740 train_time:661008ms step_avg:205.22ms
step:3222/4740 train_time:661332ms step_avg:205.25ms
step:3223/4740 train_time:661514ms step_avg:205.25ms
step:3224/4740 train_time:661838ms step_avg:205.28ms
step:3225/4740 train_time:662016ms step_avg:205.28ms
step:3226/4740 train_time:662340ms step_avg:205.31ms
step:3227/4740 train_time:662517ms step_avg:205.30ms
step:3228/4740 train_time:662843ms step_avg:205.34ms
step:3229/4740 train_time:663021ms step_avg:205.33ms
step:3230/4740 train_time:663343ms step_avg:205.37ms
step:3231/4740 train_time:663522ms step_avg:205.36ms
step:3232/4740 train_time:663843ms step_avg:205.40ms
step:3233/4740 train_time:664021ms step_avg:205.39ms
step:3234/4740 train_time:664340ms step_avg:205.42ms
step:3235/4740 train_time:664522ms step_avg:205.42ms
step:3236/4740 train_time:664843ms step_avg:205.45ms
step:3237/4740 train_time:665021ms step_avg:205.44ms
step:3238/4740 train_time:665338ms step_avg:205.48ms
step:3239/4740 train_time:665519ms step_avg:205.47ms
step:3240/4740 train_time:665841ms step_avg:205.51ms
step:3241/4740 train_time:666020ms step_avg:205.50ms
step:3242/4740 train_time:666342ms step_avg:205.53ms
step:3243/4740 train_time:666521ms step_avg:205.53ms
step:3244/4740 train_time:666844ms step_avg:205.56ms
step:3245/4740 train_time:667022ms step_avg:205.55ms
step:3246/4740 train_time:667339ms step_avg:205.59ms
step:3247/4740 train_time:667517ms step_avg:205.58ms
step:3248/4740 train_time:667840ms step_avg:205.62ms
step:3249/4740 train_time:668017ms step_avg:205.61ms
step:3250/4740 train_time:668340ms step_avg:205.64ms
step:3250/4740 val_loss:3.0672 train_time:668464ms step_avg:205.68ms
step:3251/4740 train_time:668521ms step_avg:205.64ms
step:3252/4740 train_time:668850ms step_avg:205.67ms
step:3253/4740 train_time:669025ms step_avg:205.66ms
step:3254/4740 train_time:669340ms step_avg:205.70ms
step:3255/4740 train_time:669520ms step_avg:205.69ms
step:3256/4740 train_time:669848ms step_avg:205.73ms
step:3257/4740 train_time:670031ms step_avg:205.72ms
step:3258/4740 train_time:670347ms step_avg:205.75ms
step:3259/4740 train_time:670525ms step_avg:205.75ms
step:3260/4740 train_time:670848ms step_avg:205.78ms
step:3261/4740 train_time:671030ms step_avg:205.77ms
step:3262/4740 train_time:671351ms step_avg:205.81ms
step:3263/4740 train_time:671529ms step_avg:205.80ms
step:3264/4740 train_time:671851ms step_avg:205.84ms
step:3265/4740 train_time:672031ms step_avg:205.83ms
step:3266/4740 train_time:672348ms step_avg:205.86ms
step:3267/4740 train_time:672527ms step_avg:205.85ms
step:3268/4740 train_time:672846ms step_avg:205.89ms
step:3269/4740 train_time:673026ms step_avg:205.88ms
step:3270/4740 train_time:673344ms step_avg:205.92ms
step:3271/4740 train_time:673522ms step_avg:205.91ms
step:3272/4740 train_time:673841ms step_avg:205.94ms
step:3273/4740 train_time:674022ms step_avg:205.93ms
step:3274/4740 train_time:674341ms step_avg:205.97ms
step:3275/4740 train_time:674521ms step_avg:205.96ms
step:3276/4740 train_time:674843ms step_avg:206.00ms
step:3277/4740 train_time:675025ms step_avg:205.99ms
step:3278/4740 train_time:675344ms step_avg:206.02ms
step:3279/4740 train_time:675525ms step_avg:206.02ms
step:3280/4740 train_time:675846ms step_avg:206.05ms
step:3281/4740 train_time:676026ms step_avg:206.04ms
step:3282/4740 train_time:676346ms step_avg:206.08ms
step:3283/4740 train_time:676528ms step_avg:206.07ms
step:3284/4740 train_time:676848ms step_avg:206.10ms
step:3285/4740 train_time:677029ms step_avg:206.10ms
step:3286/4740 train_time:677345ms step_avg:206.13ms
step:3287/4740 train_time:677526ms step_avg:206.12ms
step:3288/4740 train_time:677851ms step_avg:206.16ms
step:3289/4740 train_time:678030ms step_avg:206.15ms
step:3290/4740 train_time:678351ms step_avg:206.19ms
step:3291/4740 train_time:678530ms step_avg:206.18ms
step:3292/4740 train_time:678850ms step_avg:206.21ms
step:3293/4740 train_time:679027ms step_avg:206.20ms
step:3294/4740 train_time:679346ms step_avg:206.24ms
step:3295/4740 train_time:679525ms step_avg:206.23ms
step:3296/4740 train_time:679845ms step_avg:206.26ms
step:3297/4740 train_time:680025ms step_avg:206.26ms
step:3298/4740 train_time:680342ms step_avg:206.29ms
step:3299/4740 train_time:680521ms step_avg:206.28ms
step:3300/4740 train_time:680842ms step_avg:206.32ms
step:3301/4740 train_time:681021ms step_avg:206.31ms
step:3302/4740 train_time:681340ms step_avg:206.34ms
step:3303/4740 train_time:681520ms step_avg:206.33ms
step:3304/4740 train_time:681840ms step_avg:206.37ms
step:3305/4740 train_time:682018ms step_avg:206.36ms
step:3306/4740 train_time:682338ms step_avg:206.39ms
step:3307/4740 train_time:682521ms step_avg:206.39ms
step:3308/4740 train_time:682843ms step_avg:206.42ms
step:3309/4740 train_time:683024ms step_avg:206.41ms
step:3310/4740 train_time:683345ms step_avg:206.45ms
step:3311/4740 train_time:683525ms step_avg:206.44ms
step:3312/4740 train_time:683846ms step_avg:206.48ms
step:3313/4740 train_time:684024ms step_avg:206.47ms
step:3314/4740 train_time:684341ms step_avg:206.50ms
step:3315/4740 train_time:684525ms step_avg:206.49ms
step:3316/4740 train_time:684848ms step_avg:206.53ms
step:3317/4740 train_time:685029ms step_avg:206.52ms
step:3318/4740 train_time:685354ms step_avg:206.56ms
step:3319/4740 train_time:685531ms step_avg:206.55ms
step:3320/4740 train_time:685853ms step_avg:206.58ms
step:3321/4740 train_time:686032ms step_avg:206.57ms
step:3322/4740 train_time:686354ms step_avg:206.61ms
step:3323/4740 train_time:686534ms step_avg:206.60ms
step:3324/4740 train_time:686854ms step_avg:206.63ms
step:3325/4740 train_time:687032ms step_avg:206.63ms
step:3326/4740 train_time:687354ms step_avg:206.66ms
step:3327/4740 train_time:687532ms step_avg:206.65ms
step:3328/4740 train_time:687854ms step_avg:206.69ms
step:3329/4740 train_time:688031ms step_avg:206.68ms
step:3330/4740 train_time:688353ms step_avg:206.71ms
step:3331/4740 train_time:688531ms step_avg:206.70ms
step:3332/4740 train_time:688851ms step_avg:206.74ms
step:3333/4740 train_time:689030ms step_avg:206.73ms
step:3334/4740 train_time:689352ms step_avg:206.76ms
step:3335/4740 train_time:689530ms step_avg:206.76ms
step:3336/4740 train_time:689852ms step_avg:206.79ms
step:3337/4740 train_time:690032ms step_avg:206.78ms
step:3338/4740 train_time:690352ms step_avg:206.82ms
step:3339/4740 train_time:690530ms step_avg:206.81ms
step:3340/4740 train_time:690850ms step_avg:206.84ms
step:3341/4740 train_time:691031ms step_avg:206.83ms
step:3342/4740 train_time:691355ms step_avg:206.87ms
step:3343/4740 train_time:691533ms step_avg:206.86ms
step:3344/4740 train_time:691854ms step_avg:206.89ms
step:3345/4740 train_time:692029ms step_avg:206.88ms
step:3346/4740 train_time:692353ms step_avg:206.92ms
step:3347/4740 train_time:692532ms step_avg:206.91ms
step:3348/4740 train_time:692855ms step_avg:206.95ms
step:3349/4740 train_time:693033ms step_avg:206.94ms
step:3350/4740 train_time:693354ms step_avg:206.97ms
step:3351/4740 train_time:693537ms step_avg:206.96ms
step:3352/4740 train_time:693862ms step_avg:207.00ms
step:3353/4740 train_time:694040ms step_avg:206.99ms
step:3354/4740 train_time:694363ms step_avg:207.03ms
step:3355/4740 train_time:694542ms step_avg:207.02ms
step:3356/4740 train_time:694862ms step_avg:207.05ms
step:3357/4740 train_time:695042ms step_avg:207.04ms
step:3358/4740 train_time:695363ms step_avg:207.08ms
step:3359/4740 train_time:695539ms step_avg:207.07ms
step:3360/4740 train_time:695861ms step_avg:207.10ms
step:3361/4740 train_time:696039ms step_avg:207.09ms
step:3362/4740 train_time:696361ms step_avg:207.13ms
step:3363/4740 train_time:696538ms step_avg:207.12ms
step:3364/4740 train_time:696862ms step_avg:207.15ms
step:3365/4740 train_time:697040ms step_avg:207.14ms
step:3366/4740 train_time:697360ms step_avg:207.18ms
step:3367/4740 train_time:697535ms step_avg:207.17ms
step:3368/4740 train_time:697861ms step_avg:207.20ms
step:3369/4740 train_time:698039ms step_avg:207.19ms
step:3370/4740 train_time:698361ms step_avg:207.23ms
step:3371/4740 train_time:698540ms step_avg:207.22ms
step:3372/4740 train_time:698864ms step_avg:207.25ms
step:3373/4740 train_time:699044ms step_avg:207.25ms
step:3374/4740 train_time:699364ms step_avg:207.28ms
step:3375/4740 train_time:699542ms step_avg:207.27ms
step:3376/4740 train_time:699863ms step_avg:207.31ms
step:3377/4740 train_time:700042ms step_avg:207.30ms
step:3378/4740 train_time:700361ms step_avg:207.33ms
step:3379/4740 train_time:700539ms step_avg:207.32ms
step:3380/4740 train_time:700860ms step_avg:207.35ms
step:3381/4740 train_time:701038ms step_avg:207.35ms
step:3382/4740 train_time:701359ms step_avg:207.38ms
step:3383/4740 train_time:701540ms step_avg:207.37ms
step:3384/4740 train_time:701863ms step_avg:207.41ms
step:3385/4740 train_time:702041ms step_avg:207.40ms
step:3386/4740 train_time:702362ms step_avg:207.43ms
step:3387/4740 train_time:702540ms step_avg:207.42ms
step:3388/4740 train_time:702864ms step_avg:207.46ms
step:3389/4740 train_time:703049ms step_avg:207.45ms
step:3390/4740 train_time:703370ms step_avg:207.48ms
step:3391/4740 train_time:703548ms step_avg:207.48ms
step:3392/4740 train_time:703869ms step_avg:207.51ms
step:3393/4740 train_time:704050ms step_avg:207.50ms
step:3394/4740 train_time:704369ms step_avg:207.53ms
step:3395/4740 train_time:704549ms step_avg:207.53ms
step:3396/4740 train_time:704873ms step_avg:207.56ms
step:3397/4740 train_time:705051ms step_avg:207.55ms
step:3398/4740 train_time:705374ms step_avg:207.59ms
step:3399/4740 train_time:705551ms step_avg:207.58ms
step:3400/4740 train_time:705876ms step_avg:207.61ms
step:3401/4740 train_time:706054ms step_avg:207.60ms
step:3402/4740 train_time:706377ms step_avg:207.64ms
step:3403/4740 train_time:706554ms step_avg:207.63ms
step:3404/4740 train_time:706880ms step_avg:207.66ms
step:3405/4740 train_time:707056ms step_avg:207.65ms
step:3406/4740 train_time:707376ms step_avg:207.69ms
step:3407/4740 train_time:707556ms step_avg:207.68ms
step:3408/4740 train_time:707875ms step_avg:207.71ms
step:3409/4740 train_time:708056ms step_avg:207.70ms
step:3410/4740 train_time:708380ms step_avg:207.74ms
step:3411/4740 train_time:708558ms step_avg:207.73ms
step:3412/4740 train_time:708879ms step_avg:207.76ms
step:3413/4740 train_time:709061ms step_avg:207.75ms
step:3414/4740 train_time:709383ms step_avg:207.79ms
step:3415/4740 train_time:709560ms step_avg:207.78ms
step:3416/4740 train_time:709883ms step_avg:207.81ms
step:3417/4740 train_time:710063ms step_avg:207.80ms
step:3418/4740 train_time:710383ms step_avg:207.84ms
step:3419/4740 train_time:710561ms step_avg:207.83ms
step:3420/4740 train_time:710884ms step_avg:207.86ms
step:3421/4740 train_time:711065ms step_avg:207.85ms
step:3422/4740 train_time:711383ms step_avg:207.89ms
step:3423/4740 train_time:711564ms step_avg:207.88ms
step:3424/4740 train_time:711883ms step_avg:207.91ms
step:3425/4740 train_time:712063ms step_avg:207.90ms
step:3426/4740 train_time:712384ms step_avg:207.93ms
step:3427/4740 train_time:712563ms step_avg:207.93ms
step:3428/4740 train_time:712884ms step_avg:207.96ms
step:3429/4740 train_time:713064ms step_avg:207.95ms
step:3430/4740 train_time:713385ms step_avg:207.98ms
step:3431/4740 train_time:713565ms step_avg:207.98ms
step:3432/4740 train_time:713883ms step_avg:208.01ms
step:3433/4740 train_time:714065ms step_avg:208.00ms
step:3434/4740 train_time:714388ms step_avg:208.03ms
step:3435/4740 train_time:714568ms step_avg:208.03ms
step:3436/4740 train_time:714886ms step_avg:208.06ms
step:3437/4740 train_time:715066ms step_avg:208.05ms
step:3438/4740 train_time:715383ms step_avg:208.08ms
step:3439/4740 train_time:715566ms step_avg:208.07ms
step:3440/4740 train_time:715885ms step_avg:208.11ms
step:3441/4740 train_time:716068ms step_avg:208.10ms
step:3442/4740 train_time:716389ms step_avg:208.13ms
step:3443/4740 train_time:716567ms step_avg:208.12ms
step:3444/4740 train_time:716886ms step_avg:208.15ms
step:3445/4740 train_time:717071ms step_avg:208.15ms
step:3446/4740 train_time:717391ms step_avg:208.18ms
step:3447/4740 train_time:717569ms step_avg:208.17ms
step:3448/4740 train_time:717891ms step_avg:208.20ms
step:3449/4740 train_time:718071ms step_avg:208.20ms
step:3450/4740 train_time:718394ms step_avg:208.23ms
step:3451/4740 train_time:718571ms step_avg:208.22ms
step:3452/4740 train_time:718891ms step_avg:208.25ms
step:3453/4740 train_time:719072ms step_avg:208.25ms
step:3454/4740 train_time:719392ms step_avg:208.28ms
step:3455/4740 train_time:719571ms step_avg:208.27ms
step:3456/4740 train_time:719895ms step_avg:208.30ms
step:3457/4740 train_time:720070ms step_avg:208.29ms
step:3458/4740 train_time:720394ms step_avg:208.33ms
step:3459/4740 train_time:720571ms step_avg:208.32ms
step:3460/4740 train_time:720893ms step_avg:208.35ms
step:3461/4740 train_time:721074ms step_avg:208.34ms
step:3462/4740 train_time:721398ms step_avg:208.38ms
step:3463/4740 train_time:721575ms step_avg:208.37ms
step:3464/4740 train_time:721898ms step_avg:208.40ms
step:3465/4740 train_time:722075ms step_avg:208.39ms
step:3466/4740 train_time:722402ms step_avg:208.43ms
step:3467/4740 train_time:722576ms step_avg:208.42ms
step:3468/4740 train_time:722899ms step_avg:208.45ms
step:3469/4740 train_time:723076ms step_avg:208.44ms
step:3470/4740 train_time:723399ms step_avg:208.47ms
step:3471/4740 train_time:723577ms step_avg:208.46ms
step:3472/4740 train_time:723900ms step_avg:208.50ms
step:3473/4740 train_time:724077ms step_avg:208.49ms
step:3474/4740 train_time:724400ms step_avg:208.52ms
step:3475/4740 train_time:724576ms step_avg:208.51ms
step:3476/4740 train_time:724897ms step_avg:208.54ms
step:3477/4740 train_time:725072ms step_avg:208.53ms
step:3478/4740 train_time:725396ms step_avg:208.57ms
step:3479/4740 train_time:725573ms step_avg:208.56ms
step:3480/4740 train_time:725898ms step_avg:208.59ms
step:3481/4740 train_time:726075ms step_avg:208.58ms
step:3482/4740 train_time:726401ms step_avg:208.62ms
step:3483/4740 train_time:726577ms step_avg:208.61ms
step:3484/4740 train_time:726900ms step_avg:208.64ms
step:3485/4740 train_time:727078ms step_avg:208.63ms
step:3486/4740 train_time:727401ms step_avg:208.66ms
step:3487/4740 train_time:727582ms step_avg:208.66ms
step:3488/4740 train_time:727903ms step_avg:208.69ms
step:3489/4740 train_time:728084ms step_avg:208.68ms
step:3490/4740 train_time:728405ms step_avg:208.71ms
step:3491/4740 train_time:728587ms step_avg:208.70ms
step:3492/4740 train_time:728910ms step_avg:208.74ms
step:3493/4740 train_time:729089ms step_avg:208.73ms
step:3494/4740 train_time:729412ms step_avg:208.76ms
step:3495/4740 train_time:729590ms step_avg:208.75ms
step:3496/4740 train_time:729911ms step_avg:208.78ms
step:3497/4740 train_time:730091ms step_avg:208.78ms
step:3498/4740 train_time:730410ms step_avg:208.81ms
step:3499/4740 train_time:730590ms step_avg:208.80ms
step:3500/4740 train_time:730913ms step_avg:208.83ms
step:3500/4740 val_loss:3.0368 train_time:731037ms step_avg:208.87ms
step:3501/4740 train_time:731096ms step_avg:208.82ms
step:3502/4740 train_time:731427ms step_avg:208.86ms
step:3503/4740 train_time:731600ms step_avg:208.85ms
step:3504/4740 train_time:731915ms step_avg:208.88ms
step:3505/4740 train_time:732090ms step_avg:208.87ms
step:3506/4740 train_time:732416ms step_avg:208.90ms
step:3507/4740 train_time:732595ms step_avg:208.90ms
step:3508/4740 train_time:732916ms step_avg:208.93ms
step:3509/4740 train_time:733094ms step_avg:208.92ms
step:3510/4740 train_time:733420ms step_avg:208.95ms
step:3511/4740 train_time:733597ms step_avg:208.94ms
step:3512/4740 train_time:733917ms step_avg:208.97ms
step:3513/4740 train_time:734092ms step_avg:208.96ms
step:3514/4740 train_time:734415ms step_avg:209.00ms
step:3515/4740 train_time:734596ms step_avg:208.99ms
step:3516/4740 train_time:734915ms step_avg:209.02ms
step:3517/4740 train_time:735092ms step_avg:209.01ms
step:3518/4740 train_time:735412ms step_avg:209.04ms
step:3519/4740 train_time:735592ms step_avg:209.03ms
step:3520/4740 train_time:735914ms step_avg:209.07ms
step:3521/4740 train_time:736091ms step_avg:209.06ms
step:3522/4740 train_time:736414ms step_avg:209.09ms
step:3523/4740 train_time:736591ms step_avg:209.08ms
step:3524/4740 train_time:736914ms step_avg:209.11ms
step:3525/4740 train_time:737091ms step_avg:209.10ms
step:3526/4740 train_time:737414ms step_avg:209.14ms
step:3527/4740 train_time:737591ms step_avg:209.13ms
step:3528/4740 train_time:737913ms step_avg:209.16ms
step:3529/4740 train_time:738091ms step_avg:209.15ms
step:3530/4740 train_time:738414ms step_avg:209.18ms
step:3531/4740 train_time:738590ms step_avg:209.17ms
step:3532/4740 train_time:738913ms step_avg:209.21ms
step:3533/4740 train_time:739088ms step_avg:209.20ms
step:3534/4740 train_time:739410ms step_avg:209.23ms
step:3535/4740 train_time:739587ms step_avg:209.22ms
step:3536/4740 train_time:739912ms step_avg:209.25ms
step:3537/4740 train_time:740088ms step_avg:209.24ms
step:3538/4740 train_time:740412ms step_avg:209.27ms
step:3539/4740 train_time:740589ms step_avg:209.26ms
step:3540/4740 train_time:740911ms step_avg:209.30ms
step:3541/4740 train_time:741091ms step_avg:209.29ms
step:3542/4740 train_time:741415ms step_avg:209.32ms
step:3543/4740 train_time:741592ms step_avg:209.31ms
step:3544/4740 train_time:741913ms step_avg:209.34ms
step:3545/4740 train_time:742092ms step_avg:209.33ms
step:3546/4740 train_time:742414ms step_avg:209.37ms
step:3547/4740 train_time:742594ms step_avg:209.36ms
step:3548/4740 train_time:742918ms step_avg:209.39ms
step:3549/4740 train_time:743096ms step_avg:209.38ms
step:3550/4740 train_time:743420ms step_avg:209.41ms
step:3551/4740 train_time:743596ms step_avg:209.40ms
step:3552/4740 train_time:743917ms step_avg:209.44ms
step:3553/4740 train_time:744093ms step_avg:209.43ms
step:3554/4740 train_time:744414ms step_avg:209.46ms
step:3555/4740 train_time:744594ms step_avg:209.45ms
step:3556/4740 train_time:744921ms step_avg:209.48ms
step:3557/4740 train_time:745098ms step_avg:209.47ms
step:3558/4740 train_time:745419ms step_avg:209.51ms
step:3559/4740 train_time:745598ms step_avg:209.50ms
step:3560/4740 train_time:745918ms step_avg:209.53ms
step:3561/4740 train_time:746098ms step_avg:209.52ms
step:3562/4740 train_time:746418ms step_avg:209.55ms
step:3563/4740 train_time:746597ms step_avg:209.54ms
step:3564/4740 train_time:746917ms step_avg:209.57ms
step:3565/4740 train_time:747094ms step_avg:209.56ms
step:3566/4740 train_time:747416ms step_avg:209.59ms
step:3567/4740 train_time:747594ms step_avg:209.59ms
step:3568/4740 train_time:747914ms step_avg:209.62ms
step:3569/4740 train_time:748089ms step_avg:209.61ms
step:3570/4740 train_time:748413ms step_avg:209.64ms
step:3571/4740 train_time:748588ms step_avg:209.63ms
step:3572/4740 train_time:748913ms step_avg:209.66ms
step:3573/4740 train_time:749089ms step_avg:209.65ms
step:3574/4740 train_time:749411ms step_avg:209.68ms
step:3575/4740 train_time:749590ms step_avg:209.68ms
step:3576/4740 train_time:749914ms step_avg:209.71ms
step:3577/4740 train_time:750091ms step_avg:209.70ms
step:3578/4740 train_time:750417ms step_avg:209.73ms
step:3579/4740 train_time:750595ms step_avg:209.72ms
step:3580/4740 train_time:750916ms step_avg:209.75ms
step:3581/4740 train_time:751095ms step_avg:209.74ms
step:3582/4740 train_time:751415ms step_avg:209.78ms
step:3583/4740 train_time:751597ms step_avg:209.77ms
step:3584/4740 train_time:751918ms step_avg:209.80ms
step:3585/4740 train_time:752095ms step_avg:209.79ms
step:3586/4740 train_time:752419ms step_avg:209.82ms
step:3587/4740 train_time:752597ms step_avg:209.81ms
step:3588/4740 train_time:752921ms step_avg:209.84ms
step:3589/4740 train_time:753098ms step_avg:209.83ms
step:3590/4740 train_time:753418ms step_avg:209.87ms
step:3591/4740 train_time:753598ms step_avg:209.86ms
step:3592/4740 train_time:753919ms step_avg:209.89ms
step:3593/4740 train_time:754103ms step_avg:209.88ms
step:3594/4740 train_time:754425ms step_avg:209.91ms
step:3595/4740 train_time:754602ms step_avg:209.90ms
step:3596/4740 train_time:754923ms step_avg:209.93ms
step:3597/4740 train_time:755100ms step_avg:209.92ms
step:3598/4740 train_time:755421ms step_avg:209.96ms
step:3599/4740 train_time:755600ms step_avg:209.95ms
step:3600/4740 train_time:755919ms step_avg:209.98ms
step:3601/4740 train_time:756098ms step_avg:209.97ms
step:3602/4740 train_time:756417ms step_avg:210.00ms
step:3603/4740 train_time:756597ms step_avg:209.99ms
step:3604/4740 train_time:756919ms step_avg:210.02ms
step:3605/4740 train_time:757096ms step_avg:210.01ms
step:3606/4740 train_time:757416ms step_avg:210.04ms
step:3607/4740 train_time:757595ms step_avg:210.03ms
step:3608/4740 train_time:757917ms step_avg:210.07ms
step:3609/4740 train_time:758094ms step_avg:210.06ms
step:3610/4740 train_time:758418ms step_avg:210.09ms
step:3611/4740 train_time:758596ms step_avg:210.08ms
step:3612/4740 train_time:758920ms step_avg:210.11ms
step:3613/4740 train_time:759097ms step_avg:210.10ms
step:3614/4740 train_time:759419ms step_avg:210.13ms
step:3615/4740 train_time:759595ms step_avg:210.12ms
step:3616/4740 train_time:759916ms step_avg:210.15ms
step:3617/4740 train_time:760098ms step_avg:210.15ms
step:3618/4740 train_time:760418ms step_avg:210.18ms
step:3619/4740 train_time:760594ms step_avg:210.17ms
step:3620/4740 train_time:760916ms step_avg:210.20ms
step:3621/4740 train_time:761091ms step_avg:210.19ms
step:3622/4740 train_time:761414ms step_avg:210.22ms
step:3623/4740 train_time:761592ms step_avg:210.21ms
step:3624/4740 train_time:761917ms step_avg:210.24ms
step:3625/4740 train_time:762096ms step_avg:210.23ms
step:3626/4740 train_time:762416ms step_avg:210.26ms
step:3627/4740 train_time:762598ms step_avg:210.26ms
step:3628/4740 train_time:762922ms step_avg:210.29ms
step:3629/4740 train_time:763099ms step_avg:210.28ms
step:3630/4740 train_time:763421ms step_avg:210.31ms
step:3631/4740 train_time:763599ms step_avg:210.30ms
step:3632/4740 train_time:763921ms step_avg:210.33ms
step:3633/4740 train_time:764099ms step_avg:210.32ms
step:3634/4740 train_time:764420ms step_avg:210.35ms
step:3635/4740 train_time:764600ms step_avg:210.34ms
step:3636/4740 train_time:764922ms step_avg:210.37ms
step:3637/4740 train_time:765100ms step_avg:210.37ms
step:3638/4740 train_time:765421ms step_avg:210.40ms
step:3639/4740 train_time:765601ms step_avg:210.39ms
step:3640/4740 train_time:765921ms step_avg:210.42ms
step:3641/4740 train_time:766103ms step_avg:210.41ms
step:3642/4740 train_time:766420ms step_avg:210.44ms
step:3643/4740 train_time:766596ms step_avg:210.43ms
step:3644/4740 train_time:766917ms step_avg:210.46ms
step:3645/4740 train_time:767094ms step_avg:210.45ms
step:3646/4740 train_time:767416ms step_avg:210.48ms
step:3647/4740 train_time:767595ms step_avg:210.47ms
step:3648/4740 train_time:767921ms step_avg:210.50ms
step:3649/4740 train_time:768097ms step_avg:210.50ms
step:3650/4740 train_time:768422ms step_avg:210.53ms
step:3651/4740 train_time:768599ms step_avg:210.52ms
step:3652/4740 train_time:768920ms step_avg:210.55ms
step:3653/4740 train_time:769097ms step_avg:210.54ms
step:3654/4740 train_time:769419ms step_avg:210.57ms
step:3655/4740 train_time:769598ms step_avg:210.56ms
step:3656/4740 train_time:769918ms step_avg:210.59ms
step:3657/4740 train_time:770093ms step_avg:210.58ms
step:3658/4740 train_time:770414ms step_avg:210.61ms
step:3659/4740 train_time:770594ms step_avg:210.60ms
step:3660/4740 train_time:770915ms step_avg:210.63ms
step:3661/4740 train_time:771095ms step_avg:210.62ms
step:3662/4740 train_time:771416ms step_avg:210.65ms
step:3663/4740 train_time:771594ms step_avg:210.65ms
step:3664/4740 train_time:771915ms step_avg:210.68ms
step:3665/4740 train_time:772098ms step_avg:210.67ms
step:3666/4740 train_time:772419ms step_avg:210.70ms
step:3667/4740 train_time:772601ms step_avg:210.69ms
step:3668/4740 train_time:772920ms step_avg:210.72ms
step:3669/4740 train_time:773098ms step_avg:210.71ms
step:3670/4740 train_time:773418ms step_avg:210.74ms
step:3671/4740 train_time:773597ms step_avg:210.73ms
step:3672/4740 train_time:773920ms step_avg:210.76ms
step:3673/4740 train_time:774098ms step_avg:210.75ms
step:3674/4740 train_time:774419ms step_avg:210.78ms
step:3675/4740 train_time:774596ms step_avg:210.77ms
step:3676/4740 train_time:774917ms step_avg:210.80ms
step:3677/4740 train_time:775093ms step_avg:210.79ms
step:3678/4740 train_time:775421ms step_avg:210.83ms
step:3679/4740 train_time:775601ms step_avg:210.82ms
step:3680/4740 train_time:775922ms step_avg:210.85ms
step:3681/4740 train_time:776101ms step_avg:210.84ms
step:3682/4740 train_time:776424ms step_avg:210.87ms
step:3683/4740 train_time:776603ms step_avg:210.86ms
step:3684/4740 train_time:776924ms step_avg:210.89ms
step:3685/4740 train_time:777100ms step_avg:210.88ms
step:3686/4740 train_time:777419ms step_avg:210.91ms
step:3687/4740 train_time:777601ms step_avg:210.90ms
step:3688/4740 train_time:777919ms step_avg:210.93ms
step:3689/4740 train_time:778101ms step_avg:210.92ms
step:3690/4740 train_time:778424ms step_avg:210.96ms
step:3691/4740 train_time:778603ms step_avg:210.95ms
step:3692/4740 train_time:778925ms step_avg:210.98ms
step:3693/4740 train_time:779104ms step_avg:210.97ms
step:3694/4740 train_time:779427ms step_avg:211.00ms
step:3695/4740 train_time:779611ms step_avg:210.99ms
step:3696/4740 train_time:779930ms step_avg:211.02ms
step:3697/4740 train_time:780108ms step_avg:211.01ms
step:3698/4740 train_time:780430ms step_avg:211.04ms
step:3699/4740 train_time:780608ms step_avg:211.03ms
step:3700/4740 train_time:780929ms step_avg:211.06ms
step:3701/4740 train_time:781109ms step_avg:211.05ms
step:3702/4740 train_time:781426ms step_avg:211.08ms
step:3703/4740 train_time:781607ms step_avg:211.07ms
step:3704/4740 train_time:781924ms step_avg:211.10ms
step:3705/4740 train_time:782104ms step_avg:211.09ms
step:3706/4740 train_time:782423ms step_avg:211.12ms
step:3707/4740 train_time:782602ms step_avg:211.11ms
step:3708/4740 train_time:782923ms step_avg:211.14ms
step:3709/4740 train_time:783103ms step_avg:211.14ms
step:3710/4740 train_time:783425ms step_avg:211.17ms
step:3711/4740 train_time:783605ms step_avg:211.16ms
step:3712/4740 train_time:783923ms step_avg:211.19ms
step:3713/4740 train_time:784103ms step_avg:211.18ms
step:3714/4740 train_time:784427ms step_avg:211.21ms
step:3715/4740 train_time:784609ms step_avg:211.20ms
step:3716/4740 train_time:784930ms step_avg:211.23ms
step:3717/4740 train_time:785112ms step_avg:211.22ms
step:3718/4740 train_time:785431ms step_avg:211.25ms
step:3719/4740 train_time:785608ms step_avg:211.24ms
step:3720/4740 train_time:785931ms step_avg:211.27ms
step:3721/4740 train_time:786111ms step_avg:211.26ms
step:3722/4740 train_time:786432ms step_avg:211.29ms
step:3723/4740 train_time:786610ms step_avg:211.28ms
step:3724/4740 train_time:786932ms step_avg:211.31ms
step:3725/4740 train_time:787111ms step_avg:211.31ms
step:3726/4740 train_time:787433ms step_avg:211.33ms
step:3727/4740 train_time:787611ms step_avg:211.33ms
step:3728/4740 train_time:787932ms step_avg:211.36ms
step:3729/4740 train_time:788109ms step_avg:211.35ms
step:3730/4740 train_time:788430ms step_avg:211.38ms
step:3731/4740 train_time:788608ms step_avg:211.37ms
step:3732/4740 train_time:788928ms step_avg:211.40ms
step:3733/4740 train_time:789108ms step_avg:211.39ms
step:3734/4740 train_time:789430ms step_avg:211.42ms
step:3735/4740 train_time:789606ms step_avg:211.41ms
step:3736/4740 train_time:789929ms step_avg:211.44ms
step:3737/4740 train_time:790108ms step_avg:211.43ms
step:3738/4740 train_time:790429ms step_avg:211.46ms
step:3739/4740 train_time:790610ms step_avg:211.45ms
step:3740/4740 train_time:790931ms step_avg:211.48ms
step:3741/4740 train_time:791109ms step_avg:211.47ms
step:3742/4740 train_time:791429ms step_avg:211.50ms
step:3743/4740 train_time:791609ms step_avg:211.49ms
step:3744/4740 train_time:791931ms step_avg:211.52ms
step:3745/4740 train_time:792108ms step_avg:211.51ms
step:3746/4740 train_time:792430ms step_avg:211.54ms
step:3747/4740 train_time:792610ms step_avg:211.53ms
step:3748/4740 train_time:792933ms step_avg:211.56ms
step:3749/4740 train_time:793113ms step_avg:211.55ms
step:3750/4740 train_time:793433ms step_avg:211.58ms
step:3750/4740 val_loss:3.0085 train_time:793556ms step_avg:211.61ms
step:3751/4740 train_time:793613ms step_avg:211.57ms
step:3752/4740 train_time:793942ms step_avg:211.61ms
step:3753/4740 train_time:794119ms step_avg:211.60ms
step:3754/4740 train_time:794436ms step_avg:211.62ms
step:3755/4740 train_time:794617ms step_avg:211.62ms
step:3756/4740 train_time:794942ms step_avg:211.65ms
step:3757/4740 train_time:795121ms step_avg:211.64ms
step:3758/4740 train_time:795437ms step_avg:211.67ms
step:3759/4740 train_time:795619ms step_avg:211.66ms
step:3760/4740 train_time:795942ms step_avg:211.69ms
step:3761/4740 train_time:796123ms step_avg:211.68ms
step:3762/4740 train_time:796439ms step_avg:211.71ms
step:3763/4740 train_time:796618ms step_avg:211.70ms
step:3764/4740 train_time:796938ms step_avg:211.73ms
step:3765/4740 train_time:797115ms step_avg:211.72ms
step:3766/4740 train_time:797436ms step_avg:211.75ms
step:3767/4740 train_time:797612ms step_avg:211.74ms
step:3768/4740 train_time:797936ms step_avg:211.77ms
step:3769/4740 train_time:798113ms step_avg:211.76ms
step:3770/4740 train_time:798436ms step_avg:211.79ms
step:3771/4740 train_time:798612ms step_avg:211.78ms
step:3772/4740 train_time:798935ms step_avg:211.81ms
step:3773/4740 train_time:799111ms step_avg:211.80ms
step:3774/4740 train_time:799434ms step_avg:211.83ms
step:3775/4740 train_time:799611ms step_avg:211.82ms
step:3776/4740 train_time:799935ms step_avg:211.85ms
step:3777/4740 train_time:800111ms step_avg:211.84ms
step:3778/4740 train_time:800434ms step_avg:211.87ms
step:3779/4740 train_time:800610ms step_avg:211.86ms
step:3780/4740 train_time:800932ms step_avg:211.89ms
step:3781/4740 train_time:801110ms step_avg:211.88ms
step:3782/4740 train_time:801435ms step_avg:211.91ms
step:3783/4740 train_time:801610ms step_avg:211.90ms
step:3784/4740 train_time:801927ms step_avg:211.93ms
step:3785/4740 train_time:802107ms step_avg:211.92ms
step:3786/4740 train_time:802431ms step_avg:211.95ms
step:3787/4740 train_time:802612ms step_avg:211.94ms
step:3788/4740 train_time:802930ms step_avg:211.97ms
step:3789/4740 train_time:803111ms step_avg:211.96ms
step:3790/4740 train_time:803435ms step_avg:211.99ms
step:3791/4740 train_time:803613ms step_avg:211.98ms
step:3792/4740 train_time:803934ms step_avg:212.01ms
step:3793/4740 train_time:804112ms step_avg:212.00ms
step:3794/4740 train_time:804435ms step_avg:212.03ms
step:3795/4740 train_time:804613ms step_avg:212.02ms
step:3796/4740 train_time:804933ms step_avg:212.05ms
step:3797/4740 train_time:805111ms step_avg:212.04ms
step:3798/4740 train_time:805435ms step_avg:212.07ms
step:3799/4740 train_time:805612ms step_avg:212.06ms
step:3800/4740 train_time:805934ms step_avg:212.09ms
step:3801/4740 train_time:806113ms step_avg:212.08ms
step:3802/4740 train_time:806433ms step_avg:212.11ms
step:3803/4740 train_time:806610ms step_avg:212.10ms
step:3804/4740 train_time:806934ms step_avg:212.13ms
step:3805/4740 train_time:807112ms step_avg:212.12ms
step:3806/4740 train_time:807433ms step_avg:212.15ms
step:3807/4740 train_time:807609ms step_avg:212.14ms
step:3808/4740 train_time:807931ms step_avg:212.17ms
step:3809/4740 train_time:808110ms step_avg:212.16ms
step:3810/4740 train_time:808433ms step_avg:212.19ms
step:3811/4740 train_time:808611ms step_avg:212.18ms
step:3812/4740 train_time:808934ms step_avg:212.21ms
step:3813/4740 train_time:809111ms step_avg:212.20ms
step:3814/4740 train_time:809433ms step_avg:212.23ms
step:3815/4740 train_time:809612ms step_avg:212.22ms
step:3816/4740 train_time:809932ms step_avg:212.25ms
step:3817/4740 train_time:810111ms step_avg:212.24ms
step:3818/4740 train_time:810433ms step_avg:212.27ms
step:3819/4740 train_time:810610ms step_avg:212.26ms
step:3820/4740 train_time:810935ms step_avg:212.29ms
step:3821/4740 train_time:811111ms step_avg:212.28ms
step:3822/4740 train_time:811431ms step_avg:212.31ms
step:3823/4740 train_time:811608ms step_avg:212.30ms
step:3824/4740 train_time:811928ms step_avg:212.32ms
step:3825/4740 train_time:812110ms step_avg:212.32ms
step:3826/4740 train_time:812432ms step_avg:212.35ms
step:3827/4740 train_time:812611ms step_avg:212.34ms
step:3828/4740 train_time:812934ms step_avg:212.37ms
step:3829/4740 train_time:813114ms step_avg:212.36ms
step:3830/4740 train_time:813432ms step_avg:212.38ms
step:3831/4740 train_time:813614ms step_avg:212.38ms
step:3832/4740 train_time:813936ms step_avg:212.41ms
step:3833/4740 train_time:814112ms step_avg:212.40ms
step:3834/4740 train_time:814432ms step_avg:212.42ms
step:3835/4740 train_time:814610ms step_avg:212.41ms
step:3836/4740 train_time:814933ms step_avg:212.44ms
step:3837/4740 train_time:815111ms step_avg:212.43ms
step:3838/4740 train_time:815435ms step_avg:212.46ms
step:3839/4740 train_time:815610ms step_avg:212.45ms
step:3840/4740 train_time:815931ms step_avg:212.48ms
step:3841/4740 train_time:816113ms step_avg:212.47ms
step:3842/4740 train_time:816436ms step_avg:212.50ms
step:3843/4740 train_time:816612ms step_avg:212.49ms
step:3844/4740 train_time:816934ms step_avg:212.52ms
step:3845/4740 train_time:817111ms step_avg:212.51ms
step:3846/4740 train_time:817435ms step_avg:212.54ms
step:3847/4740 train_time:817613ms step_avg:212.53ms
step:3848/4740 train_time:817938ms step_avg:212.56ms
step:3849/4740 train_time:818115ms step_avg:212.55ms
step:3850/4740 train_time:818435ms step_avg:212.58ms
step:3851/4740 train_time:818613ms step_avg:212.57ms
step:3852/4740 train_time:818936ms step_avg:212.60ms
step:3853/4740 train_time:819110ms step_avg:212.59ms
step:3854/4740 train_time:819434ms step_avg:212.62ms
step:3855/4740 train_time:819613ms step_avg:212.61ms
step:3856/4740 train_time:819937ms step_avg:212.64ms
step:3857/4740 train_time:820112ms step_avg:212.63ms
step:3858/4740 train_time:820434ms step_avg:212.66ms
step:3859/4740 train_time:820615ms step_avg:212.65ms
step:3860/4740 train_time:820938ms step_avg:212.68ms
step:3861/4740 train_time:821118ms step_avg:212.67ms
step:3862/4740 train_time:821437ms step_avg:212.70ms
step:3863/4740 train_time:821618ms step_avg:212.69ms
step:3864/4740 train_time:821940ms step_avg:212.72ms
step:3865/4740 train_time:822120ms step_avg:212.71ms
step:3866/4740 train_time:822441ms step_avg:212.74ms
step:3867/4740 train_time:822620ms step_avg:212.73ms
step:3868/4740 train_time:822939ms step_avg:212.76ms
step:3869/4740 train_time:823115ms step_avg:212.75ms
step:3870/4740 train_time:823438ms step_avg:212.77ms
step:3871/4740 train_time:823614ms step_avg:212.77ms
step:3872/4740 train_time:823940ms step_avg:212.79ms
step:3873/4740 train_time:824118ms step_avg:212.79ms
step:3874/4740 train_time:824439ms step_avg:212.81ms
step:3875/4740 train_time:824621ms step_avg:212.81ms
step:3876/4740 train_time:824943ms step_avg:212.83ms
step:3877/4740 train_time:825123ms step_avg:212.83ms
step:3878/4740 train_time:825441ms step_avg:212.85ms
step:3879/4740 train_time:825620ms step_avg:212.84ms
step:3880/4740 train_time:825941ms step_avg:212.87ms
step:3881/4740 train_time:826120ms step_avg:212.86ms
step:3882/4740 train_time:826439ms step_avg:212.89ms
step:3883/4740 train_time:826617ms step_avg:212.88ms
step:3884/4740 train_time:826940ms step_avg:212.91ms
step:3885/4740 train_time:827116ms step_avg:212.90ms
step:3886/4740 train_time:827440ms step_avg:212.93ms
step:3887/4740 train_time:827615ms step_avg:212.92ms
step:3888/4740 train_time:827939ms step_avg:212.95ms
step:3889/4740 train_time:828116ms step_avg:212.94ms
step:3890/4740 train_time:828439ms step_avg:212.97ms
step:3891/4740 train_time:828615ms step_avg:212.96ms
step:3892/4740 train_time:828939ms step_avg:212.99ms
step:3893/4740 train_time:829117ms step_avg:212.98ms
step:3894/4740 train_time:829441ms step_avg:213.00ms
step:3895/4740 train_time:829618ms step_avg:213.00ms
step:3896/4740 train_time:829942ms step_avg:213.02ms
step:3897/4740 train_time:830119ms step_avg:213.01ms
step:3898/4740 train_time:830440ms step_avg:213.04ms
step:3899/4740 train_time:830619ms step_avg:213.03ms
step:3900/4740 train_time:830943ms step_avg:213.06ms
step:3901/4740 train_time:831121ms step_avg:213.05ms
step:3902/4740 train_time:831448ms step_avg:213.08ms
step:3903/4740 train_time:831626ms step_avg:213.07ms
step:3904/4740 train_time:831948ms step_avg:213.10ms
step:3905/4740 train_time:832126ms step_avg:213.09ms
step:3906/4740 train_time:832447ms step_avg:213.12ms
step:3907/4740 train_time:832630ms step_avg:213.11ms
step:3908/4740 train_time:832953ms step_avg:213.14ms
step:3909/4740 train_time:833130ms step_avg:213.13ms
step:3910/4740 train_time:833453ms step_avg:213.16ms
step:3911/4740 train_time:833629ms step_avg:213.15ms
step:3912/4740 train_time:833946ms step_avg:213.18ms
step:3913/4740 train_time:834126ms step_avg:213.17ms
step:3914/4740 train_time:834446ms step_avg:213.20ms
step:3915/4740 train_time:834628ms step_avg:213.19ms
step:3916/4740 train_time:834952ms step_avg:213.22ms
step:3917/4740 train_time:835132ms step_avg:213.21ms
step:3918/4740 train_time:835454ms step_avg:213.23ms
step:3919/4740 train_time:835634ms step_avg:213.23ms
step:3920/4740 train_time:835953ms step_avg:213.25ms
step:3921/4740 train_time:836132ms step_avg:213.24ms
step:3922/4740 train_time:836454ms step_avg:213.27ms
step:3923/4740 train_time:836634ms step_avg:213.26ms
step:3924/4740 train_time:836956ms step_avg:213.29ms
step:3925/4740 train_time:837136ms step_avg:213.28ms
step:3926/4740 train_time:837457ms step_avg:213.31ms
step:3927/4740 train_time:837636ms step_avg:213.30ms
step:3928/4740 train_time:837958ms step_avg:213.33ms
step:3929/4740 train_time:838135ms step_avg:213.32ms
step:3930/4740 train_time:838453ms step_avg:213.35ms
step:3931/4740 train_time:838635ms step_avg:213.34ms
step:3932/4740 train_time:838955ms step_avg:213.37ms
step:3933/4740 train_time:839136ms step_avg:213.36ms
step:3934/4740 train_time:839457ms step_avg:213.38ms
step:3935/4740 train_time:839642ms step_avg:213.38ms
step:3936/4740 train_time:839965ms step_avg:213.41ms
step:3937/4740 train_time:840143ms step_avg:213.40ms
step:3938/4740 train_time:840466ms step_avg:213.42ms
step:3939/4740 train_time:840645ms step_avg:213.42ms
step:3940/4740 train_time:840968ms step_avg:213.44ms
step:3941/4740 train_time:841149ms step_avg:213.44ms
step:3942/4740 train_time:841471ms step_avg:213.46ms
step:3943/4740 train_time:841648ms step_avg:213.45ms
step:3944/4740 train_time:841967ms step_avg:213.48ms
step:3945/4740 train_time:842147ms step_avg:213.47ms
step:3946/4740 train_time:842468ms step_avg:213.50ms
step:3947/4740 train_time:842651ms step_avg:213.49ms
step:3948/4740 train_time:842973ms step_avg:213.52ms
step:3949/4740 train_time:843150ms step_avg:213.51ms
step:3950/4740 train_time:843474ms step_avg:213.54ms
step:3951/4740 train_time:843650ms step_avg:213.53ms
step:3952/4740 train_time:843970ms step_avg:213.56ms
step:3953/4740 train_time:844150ms step_avg:213.55ms
step:3954/4740 train_time:844472ms step_avg:213.57ms
step:3955/4740 train_time:844650ms step_avg:213.57ms
step:3956/4740 train_time:844973ms step_avg:213.59ms
step:3957/4740 train_time:845151ms step_avg:213.58ms
step:3958/4740 train_time:845474ms step_avg:213.61ms
step:3959/4740 train_time:845655ms step_avg:213.60ms
step:3960/4740 train_time:845979ms step_avg:213.63ms
step:3961/4740 train_time:846160ms step_avg:213.62ms
step:3962/4740 train_time:846482ms step_avg:213.65ms
step:3963/4740 train_time:846666ms step_avg:213.64ms
step:3964/4740 train_time:846984ms step_avg:213.67ms
step:3965/4740 train_time:847166ms step_avg:213.66ms
step:3966/4740 train_time:847487ms step_avg:213.69ms
step:3967/4740 train_time:847668ms step_avg:213.68ms
step:3968/4740 train_time:847986ms step_avg:213.71ms
step:3969/4740 train_time:848167ms step_avg:213.70ms
step:3970/4740 train_time:848490ms step_avg:213.73ms
step:3971/4740 train_time:848670ms step_avg:213.72ms
step:3972/4740 train_time:848991ms step_avg:213.74ms
step:3973/4740 train_time:849168ms step_avg:213.73ms
step:3974/4740 train_time:849485ms step_avg:213.76ms
step:3975/4740 train_time:849664ms step_avg:213.75ms
step:3976/4740 train_time:849984ms step_avg:213.78ms
step:3977/4740 train_time:850163ms step_avg:213.77ms
step:3978/4740 train_time:850483ms step_avg:213.80ms
step:3979/4740 train_time:850664ms step_avg:213.79ms
step:3980/4740 train_time:850984ms step_avg:213.82ms
step:3981/4740 train_time:851167ms step_avg:213.81ms
step:3982/4740 train_time:851487ms step_avg:213.83ms
step:3983/4740 train_time:851666ms step_avg:213.83ms
step:3984/4740 train_time:851987ms step_avg:213.85ms
step:3985/4740 train_time:852166ms step_avg:213.84ms
step:3986/4740 train_time:852487ms step_avg:213.87ms
step:3987/4740 train_time:852668ms step_avg:213.86ms
step:3988/4740 train_time:852991ms step_avg:213.89ms
step:3989/4740 train_time:853167ms step_avg:213.88ms
step:3990/4740 train_time:853485ms step_avg:213.91ms
step:3991/4740 train_time:853667ms step_avg:213.90ms
step:3992/4740 train_time:853991ms step_avg:213.93ms
step:3993/4740 train_time:854168ms step_avg:213.92ms
step:3994/4740 train_time:854486ms step_avg:213.94ms
step:3995/4740 train_time:854670ms step_avg:213.94ms
step:3996/4740 train_time:854996ms step_avg:213.96ms
step:3997/4740 train_time:855173ms step_avg:213.95ms
step:3998/4740 train_time:855493ms step_avg:213.98ms
step:3999/4740 train_time:855672ms step_avg:213.97ms
step:4000/4740 train_time:855995ms step_avg:214.00ms
step:4000/4740 val_loss:2.9817 train_time:856117ms step_avg:214.03ms
step:4001/4740 train_time:856175ms step_avg:213.99ms
step:4002/4740 train_time:856507ms step_avg:214.02ms
step:4003/4740 train_time:856680ms step_avg:214.01ms
step:4004/4740 train_time:856996ms step_avg:214.03ms
step:4005/4740 train_time:857172ms step_avg:214.03ms
step:4006/4740 train_time:857501ms step_avg:214.05ms
step:4007/4740 train_time:857679ms step_avg:214.05ms
step:4008/4740 train_time:858001ms step_avg:214.07ms
step:4009/4740 train_time:858176ms step_avg:214.06ms
step:4010/4740 train_time:858502ms step_avg:214.09ms
step:4011/4740 train_time:858677ms step_avg:214.08ms
step:4012/4740 train_time:858997ms step_avg:214.11ms
step:4013/4740 train_time:859177ms step_avg:214.10ms
step:4014/4740 train_time:859498ms step_avg:214.13ms
step:4015/4740 train_time:859684ms step_avg:214.12ms
step:4016/4740 train_time:860003ms step_avg:214.14ms
step:4017/4740 train_time:860180ms step_avg:214.14ms
step:4018/4740 train_time:860502ms step_avg:214.16ms
step:4019/4740 train_time:860682ms step_avg:214.15ms
step:4020/4740 train_time:861004ms step_avg:214.18ms
step:4021/4740 train_time:861182ms step_avg:214.17ms
step:4022/4740 train_time:861503ms step_avg:214.20ms
step:4023/4740 train_time:861683ms step_avg:214.19ms
step:4024/4740 train_time:862000ms step_avg:214.21ms
step:4025/4740 train_time:862184ms step_avg:214.21ms
step:4026/4740 train_time:862500ms step_avg:214.23ms
step:4027/4740 train_time:862686ms step_avg:214.23ms
step:4028/4740 train_time:863002ms step_avg:214.25ms
step:4029/4740 train_time:863185ms step_avg:214.24ms
step:4030/4740 train_time:863507ms step_avg:214.27ms
step:4031/4740 train_time:863687ms step_avg:214.26ms
step:4032/4740 train_time:864008ms step_avg:214.29ms
step:4033/4740 train_time:864191ms step_avg:214.28ms
step:4034/4740 train_time:864512ms step_avg:214.31ms
step:4035/4740 train_time:864690ms step_avg:214.30ms
step:4036/4740 train_time:865012ms step_avg:214.32ms
step:4037/4740 train_time:865191ms step_avg:214.32ms
step:4038/4740 train_time:865512ms step_avg:214.34ms
step:4039/4740 train_time:865694ms step_avg:214.33ms
step:4040/4740 train_time:866014ms step_avg:214.36ms
step:4041/4740 train_time:866195ms step_avg:214.35ms
step:4042/4740 train_time:866515ms step_avg:214.38ms
step:4043/4740 train_time:866693ms step_avg:214.37ms
step:4044/4740 train_time:867014ms step_avg:214.40ms
step:4045/4740 train_time:867194ms step_avg:214.39ms
step:4046/4740 train_time:867516ms step_avg:214.41ms
step:4047/4740 train_time:867693ms step_avg:214.40ms
step:4048/4740 train_time:868014ms step_avg:214.43ms
step:4049/4740 train_time:868193ms step_avg:214.42ms
step:4050/4740 train_time:868516ms step_avg:214.45ms
step:4051/4740 train_time:868693ms step_avg:214.44ms
step:4052/4740 train_time:869013ms step_avg:214.47ms
step:4053/4740 train_time:869195ms step_avg:214.46ms
step:4054/4740 train_time:869515ms step_avg:214.48ms
step:4055/4740 train_time:869692ms step_avg:214.47ms
step:4056/4740 train_time:870013ms step_avg:214.50ms
step:4057/4740 train_time:870194ms step_avg:214.49ms
step:4058/4740 train_time:870517ms step_avg:214.52ms
step:4059/4740 train_time:870693ms step_avg:214.51ms
step:4060/4740 train_time:871016ms step_avg:214.54ms
step:4061/4740 train_time:871194ms step_avg:214.53ms
step:4062/4740 train_time:871519ms step_avg:214.55ms
step:4063/4740 train_time:871696ms step_avg:214.54ms
step:4064/4740 train_time:872019ms step_avg:214.57ms
step:4065/4740 train_time:872196ms step_avg:214.56ms
step:4066/4740 train_time:872520ms step_avg:214.59ms
step:4067/4740 train_time:872697ms step_avg:214.58ms
step:4068/4740 train_time:873020ms step_avg:214.61ms
step:4069/4740 train_time:873196ms step_avg:214.60ms
step:4070/4740 train_time:873524ms step_avg:214.63ms
step:4071/4740 train_time:873701ms step_avg:214.62ms
step:4072/4740 train_time:874029ms step_avg:214.64ms
step:4073/4740 train_time:874206ms step_avg:214.63ms
step:4074/4740 train_time:874528ms step_avg:214.66ms
step:4075/4740 train_time:874707ms step_avg:214.65ms
step:4076/4740 train_time:875029ms step_avg:214.68ms
step:4077/4740 train_time:875208ms step_avg:214.67ms
step:4078/4740 train_time:875528ms step_avg:214.70ms
step:4079/4740 train_time:875708ms step_avg:214.69ms
step:4080/4740 train_time:876028ms step_avg:214.71ms
step:4081/4740 train_time:876209ms step_avg:214.70ms
step:4082/4740 train_time:876528ms step_avg:214.73ms
step:4083/4740 train_time:876708ms step_avg:214.72ms
step:4084/4740 train_time:877029ms step_avg:214.75ms
step:4085/4740 train_time:877210ms step_avg:214.74ms
step:4086/4740 train_time:877533ms step_avg:214.77ms
step:4087/4740 train_time:877711ms step_avg:214.76ms
step:4088/4740 train_time:878031ms step_avg:214.78ms
step:4089/4740 train_time:878210ms step_avg:214.77ms
step:4090/4740 train_time:878532ms step_avg:214.80ms
step:4091/4740 train_time:878709ms step_avg:214.79ms
step:4092/4740 train_time:879032ms step_avg:214.82ms
step:4093/4740 train_time:879211ms step_avg:214.81ms
step:4094/4740 train_time:879530ms step_avg:214.83ms
step:4095/4740 train_time:879711ms step_avg:214.83ms
step:4096/4740 train_time:880036ms step_avg:214.85ms
step:4097/4740 train_time:880215ms step_avg:214.84ms
step:4098/4740 train_time:880537ms step_avg:214.87ms
step:4099/4740 train_time:880713ms step_avg:214.86ms
step:4100/4740 train_time:881036ms step_avg:214.89ms
step:4101/4740 train_time:881213ms step_avg:214.88ms
step:4102/4740 train_time:881536ms step_avg:214.90ms
step:4103/4740 train_time:881713ms step_avg:214.89ms
step:4104/4740 train_time:882035ms step_avg:214.92ms
step:4105/4740 train_time:882215ms step_avg:214.91ms
step:4106/4740 train_time:882536ms step_avg:214.94ms
step:4107/4740 train_time:882712ms step_avg:214.93ms
step:4108/4740 train_time:883036ms step_avg:214.96ms
step:4109/4740 train_time:883217ms step_avg:214.95ms
step:4110/4740 train_time:883538ms step_avg:214.97ms
step:4111/4740 train_time:883714ms step_avg:214.96ms
step:4112/4740 train_time:884040ms step_avg:214.99ms
step:4113/4740 train_time:884218ms step_avg:214.98ms
step:4114/4740 train_time:884538ms step_avg:215.01ms
step:4115/4740 train_time:884718ms step_avg:215.00ms
step:4116/4740 train_time:885039ms step_avg:215.02ms
step:4117/4740 train_time:885220ms step_avg:215.02ms
step:4118/4740 train_time:885542ms step_avg:215.04ms
step:4119/4740 train_time:885724ms step_avg:215.03ms
step:4120/4740 train_time:886044ms step_avg:215.06ms
step:4121/4740 train_time:886222ms step_avg:215.05ms
step:4122/4740 train_time:886545ms step_avg:215.08ms
step:4123/4740 train_time:886727ms step_avg:215.07ms
step:4124/4740 train_time:887048ms step_avg:215.09ms
step:4125/4740 train_time:887225ms step_avg:215.08ms
step:4126/4740 train_time:887544ms step_avg:215.11ms
step:4127/4740 train_time:887725ms step_avg:215.10ms
step:4128/4740 train_time:888046ms step_avg:215.13ms
step:4129/4740 train_time:888231ms step_avg:215.12ms
step:4130/4740 train_time:888555ms step_avg:215.15ms
step:4131/4740 train_time:888732ms step_avg:215.14ms
step:4132/4740 train_time:889054ms step_avg:215.16ms
step:4133/4740 train_time:889236ms step_avg:215.16ms
step:4134/4740 train_time:889558ms step_avg:215.18ms
step:4135/4740 train_time:889740ms step_avg:215.17ms
step:4136/4740 train_time:890060ms step_avg:215.20ms
step:4137/4740 train_time:890240ms step_avg:215.19ms
step:4138/4740 train_time:890563ms step_avg:215.22ms
step:4139/4740 train_time:890741ms step_avg:215.21ms
step:4140/4740 train_time:891061ms step_avg:215.23ms
step:4141/4740 train_time:891241ms step_avg:215.22ms
step:4142/4740 train_time:891561ms step_avg:215.25ms
step:4143/4740 train_time:891741ms step_avg:215.24ms
step:4144/4740 train_time:892063ms step_avg:215.27ms
step:4145/4740 train_time:892241ms step_avg:215.26ms
step:4146/4740 train_time:892564ms step_avg:215.28ms
step:4147/4740 train_time:892744ms step_avg:215.27ms
step:4148/4740 train_time:893065ms step_avg:215.30ms
step:4149/4740 train_time:893242ms step_avg:215.29ms
step:4150/4740 train_time:893563ms step_avg:215.32ms
step:4151/4740 train_time:893744ms step_avg:215.31ms
step:4152/4740 train_time:894065ms step_avg:215.33ms
step:4153/4740 train_time:894244ms step_avg:215.32ms
step:4154/4740 train_time:894566ms step_avg:215.35ms
step:4155/4740 train_time:894747ms step_avg:215.34ms
step:4156/4740 train_time:895068ms step_avg:215.37ms
step:4157/4740 train_time:895250ms step_avg:215.36ms
step:4158/4740 train_time:895569ms step_avg:215.38ms
step:4159/4740 train_time:895748ms step_avg:215.38ms
step:4160/4740 train_time:896074ms step_avg:215.40ms
step:4161/4740 train_time:896251ms step_avg:215.39ms
step:4162/4740 train_time:896573ms step_avg:215.42ms
step:4163/4740 train_time:896753ms step_avg:215.41ms
step:4164/4740 train_time:897076ms step_avg:215.44ms
step:4165/4740 train_time:897258ms step_avg:215.43ms
step:4166/4740 train_time:897577ms step_avg:215.45ms
step:4167/4740 train_time:897755ms step_avg:215.44ms
step:4168/4740 train_time:898076ms step_avg:215.47ms
step:4169/4740 train_time:898256ms step_avg:215.46ms
step:4170/4740 train_time:898576ms step_avg:215.49ms
step:4171/4740 train_time:898754ms step_avg:215.48ms
step:4172/4740 train_time:899077ms step_avg:215.50ms
step:4173/4740 train_time:899259ms step_avg:215.49ms
step:4174/4740 train_time:899580ms step_avg:215.52ms
step:4175/4740 train_time:899757ms step_avg:215.51ms
step:4176/4740 train_time:900078ms step_avg:215.54ms
step:4177/4740 train_time:900259ms step_avg:215.53ms
step:4178/4740 train_time:900580ms step_avg:215.55ms
step:4179/4740 train_time:900757ms step_avg:215.54ms
step:4180/4740 train_time:901076ms step_avg:215.57ms
step:4181/4740 train_time:901258ms step_avg:215.56ms
step:4182/4740 train_time:901579ms step_avg:215.59ms
step:4183/4740 train_time:901757ms step_avg:215.58ms
step:4184/4740 train_time:902079ms step_avg:215.60ms
step:4185/4740 train_time:902257ms step_avg:215.59ms
step:4186/4740 train_time:902579ms step_avg:215.62ms
step:4187/4740 train_time:902761ms step_avg:215.61ms
step:4188/4740 train_time:903082ms step_avg:215.64ms
step:4189/4740 train_time:903260ms step_avg:215.63ms
step:4190/4740 train_time:903583ms step_avg:215.65ms
step:4191/4740 train_time:903762ms step_avg:215.64ms
step:4192/4740 train_time:904080ms step_avg:215.67ms
step:4193/4740 train_time:904260ms step_avg:215.66ms
step:4194/4740 train_time:904579ms step_avg:215.68ms
step:4195/4740 train_time:904761ms step_avg:215.68ms
step:4196/4740 train_time:905079ms step_avg:215.70ms
step:4197/4740 train_time:905260ms step_avg:215.69ms
step:4198/4740 train_time:905580ms step_avg:215.72ms
step:4199/4740 train_time:905760ms step_avg:215.71ms
step:4200/4740 train_time:906082ms step_avg:215.73ms
step:4201/4740 train_time:906264ms step_avg:215.73ms
step:4202/4740 train_time:906586ms step_avg:215.75ms
step:4203/4740 train_time:906766ms step_avg:215.74ms
step:4204/4740 train_time:907093ms step_avg:215.77ms
step:4205/4740 train_time:907268ms step_avg:215.76ms
step:4206/4740 train_time:907581ms step_avg:215.78ms
step:4207/4740 train_time:907765ms step_avg:215.77ms
step:4208/4740 train_time:908085ms step_avg:215.80ms
step:4209/4740 train_time:908267ms step_avg:215.79ms
step:4210/4740 train_time:908589ms step_avg:215.82ms
step:4211/4740 train_time:908770ms step_avg:215.81ms
step:4212/4740 train_time:909090ms step_avg:215.83ms
step:4213/4740 train_time:909270ms step_avg:215.82ms
step:4214/4740 train_time:909590ms step_avg:215.85ms
step:4215/4740 train_time:909769ms step_avg:215.84ms
step:4216/4740 train_time:910086ms step_avg:215.86ms
step:4217/4740 train_time:910267ms step_avg:215.86ms
step:4218/4740 train_time:910588ms step_avg:215.88ms
step:4219/4740 train_time:910770ms step_avg:215.87ms
step:4220/4740 train_time:911090ms step_avg:215.90ms
step:4221/4740 train_time:911270ms step_avg:215.89ms
step:4222/4740 train_time:911588ms step_avg:215.91ms
step:4223/4740 train_time:911771ms step_avg:215.91ms
step:4224/4740 train_time:912098ms step_avg:215.93ms
step:4225/4740 train_time:912274ms step_avg:215.92ms
step:4226/4740 train_time:912596ms step_avg:215.95ms
step:4227/4740 train_time:912779ms step_avg:215.94ms
step:4228/4740 train_time:913100ms step_avg:215.96ms
step:4229/4740 train_time:913276ms step_avg:215.96ms
step:4230/4740 train_time:913598ms step_avg:215.98ms
step:4231/4740 train_time:913778ms step_avg:215.97ms
step:4232/4740 train_time:914099ms step_avg:216.00ms
step:4233/4740 train_time:914279ms step_avg:215.99ms
step:4234/4740 train_time:914601ms step_avg:216.01ms
step:4235/4740 train_time:914779ms step_avg:216.00ms
step:4236/4740 train_time:915101ms step_avg:216.03ms
step:4237/4740 train_time:915277ms step_avg:216.02ms
step:4238/4740 train_time:915600ms step_avg:216.05ms
step:4239/4740 train_time:915779ms step_avg:216.04ms
step:4240/4740 train_time:916103ms step_avg:216.06ms
step:4241/4740 train_time:916280ms step_avg:216.05ms
step:4242/4740 train_time:916604ms step_avg:216.08ms
step:4243/4740 train_time:916778ms step_avg:216.07ms
step:4244/4740 train_time:917097ms step_avg:216.09ms
step:4245/4740 train_time:917281ms step_avg:216.09ms
step:4246/4740 train_time:917603ms step_avg:216.11ms
step:4247/4740 train_time:917781ms step_avg:216.10ms
step:4248/4740 train_time:918101ms step_avg:216.13ms
step:4249/4740 train_time:918280ms step_avg:216.12ms
step:4250/4740 train_time:918602ms step_avg:216.14ms
step:4250/4740 val_loss:2.9563 train_time:918730ms step_avg:216.17ms
step:4251/4740 train_time:918788ms step_avg:216.13ms
step:4252/4740 train_time:919116ms step_avg:216.16ms
step:4253/4740 train_time:919296ms step_avg:216.15ms
step:4254/4740 train_time:919611ms step_avg:216.18ms
step:4255/4740 train_time:919789ms step_avg:216.17ms
step:4256/4740 train_time:920116ms step_avg:216.19ms
step:4257/4740 train_time:920295ms step_avg:216.18ms
step:4258/4740 train_time:920615ms step_avg:216.21ms
step:4259/4740 train_time:920790ms step_avg:216.20ms
step:4260/4740 train_time:921118ms step_avg:216.22ms
step:4261/4740 train_time:921298ms step_avg:216.22ms
step:4262/4740 train_time:921616ms step_avg:216.24ms
step:4263/4740 train_time:921796ms step_avg:216.23ms
step:4264/4740 train_time:922118ms step_avg:216.26ms
step:4265/4740 train_time:922297ms step_avg:216.25ms
step:4266/4740 train_time:922619ms step_avg:216.27ms
step:4267/4740 train_time:922796ms step_avg:216.26ms
step:4268/4740 train_time:923119ms step_avg:216.29ms
step:4269/4740 train_time:923298ms step_avg:216.28ms
step:4270/4740 train_time:923620ms step_avg:216.30ms
step:4271/4740 train_time:923796ms step_avg:216.30ms
step:4272/4740 train_time:924122ms step_avg:216.32ms
step:4273/4740 train_time:924301ms step_avg:216.31ms
step:4274/4740 train_time:924624ms step_avg:216.34ms
step:4275/4740 train_time:924804ms step_avg:216.33ms
step:4276/4740 train_time:925127ms step_avg:216.35ms
step:4277/4740 train_time:925304ms step_avg:216.34ms
step:4278/4740 train_time:925625ms step_avg:216.37ms
step:4279/4740 train_time:925803ms step_avg:216.36ms
step:4280/4740 train_time:926125ms step_avg:216.38ms
step:4281/4740 train_time:926303ms step_avg:216.38ms
step:4282/4740 train_time:926622ms step_avg:216.40ms
step:4283/4740 train_time:926800ms step_avg:216.39ms
step:4284/4740 train_time:927123ms step_avg:216.42ms
step:4285/4740 train_time:927303ms step_avg:216.41ms
step:4286/4740 train_time:927623ms step_avg:216.43ms
step:4287/4740 train_time:927803ms step_avg:216.42ms
step:4288/4740 train_time:928123ms step_avg:216.45ms
step:4289/4740 train_time:928300ms step_avg:216.44ms
step:4290/4740 train_time:928620ms step_avg:216.46ms
step:4291/4740 train_time:928799ms step_avg:216.45ms
step:4292/4740 train_time:929121ms step_avg:216.48ms
step:4293/4740 train_time:929299ms step_avg:216.47ms
step:4294/4740 train_time:929622ms step_avg:216.49ms
step:4295/4740 train_time:929802ms step_avg:216.48ms
step:4296/4740 train_time:930126ms step_avg:216.51ms
step:4297/4740 train_time:930303ms step_avg:216.50ms
step:4298/4740 train_time:930625ms step_avg:216.53ms
step:4299/4740 train_time:930807ms step_avg:216.52ms
step:4300/4740 train_time:931131ms step_avg:216.54ms
step:4301/4740 train_time:931311ms step_avg:216.53ms
step:4302/4740 train_time:931634ms step_avg:216.56ms
step:4303/4740 train_time:931812ms step_avg:216.55ms
step:4304/4740 train_time:932136ms step_avg:216.57ms
step:4305/4740 train_time:932315ms step_avg:216.57ms
step:4306/4740 train_time:932638ms step_avg:216.59ms
step:4307/4740 train_time:932817ms step_avg:216.58ms
step:4308/4740 train_time:933138ms step_avg:216.61ms
step:4309/4740 train_time:933316ms step_avg:216.60ms
step:4310/4740 train_time:933639ms step_avg:216.62ms
step:4311/4740 train_time:933817ms step_avg:216.61ms
step:4312/4740 train_time:934143ms step_avg:216.64ms
step:4313/4740 train_time:934320ms step_avg:216.63ms
step:4314/4740 train_time:934644ms step_avg:216.65ms
step:4315/4740 train_time:934821ms step_avg:216.64ms
step:4316/4740 train_time:935149ms step_avg:216.67ms
step:4317/4740 train_time:935328ms step_avg:216.66ms
step:4318/4740 train_time:935648ms step_avg:216.69ms
step:4319/4740 train_time:935827ms step_avg:216.68ms
step:4320/4740 train_time:936147ms step_avg:216.70ms
step:4321/4740 train_time:936329ms step_avg:216.69ms
step:4322/4740 train_time:936652ms step_avg:216.72ms
step:4323/4740 train_time:936830ms step_avg:216.71ms
step:4324/4740 train_time:937152ms step_avg:216.73ms
step:4325/4740 train_time:937329ms step_avg:216.72ms
step:4326/4740 train_time:937652ms step_avg:216.75ms
step:4327/4740 train_time:937832ms step_avg:216.74ms
step:4328/4740 train_time:938156ms step_avg:216.76ms
step:4329/4740 train_time:938333ms step_avg:216.76ms
step:4330/4740 train_time:938654ms step_avg:216.78ms
step:4331/4740 train_time:938831ms step_avg:216.77ms
step:4332/4740 train_time:939154ms step_avg:216.79ms
step:4333/4740 train_time:939331ms step_avg:216.79ms
step:4334/4740 train_time:939652ms step_avg:216.81ms
step:4335/4740 train_time:939830ms step_avg:216.80ms
step:4336/4740 train_time:940153ms step_avg:216.83ms
step:4337/4740 train_time:940333ms step_avg:216.82ms
step:4338/4740 train_time:940654ms step_avg:216.84ms
step:4339/4740 train_time:940835ms step_avg:216.83ms
step:4340/4740 train_time:941158ms step_avg:216.86ms
step:4341/4740 train_time:941334ms step_avg:216.85ms
step:4342/4740 train_time:941659ms step_avg:216.87ms
step:4343/4740 train_time:941841ms step_avg:216.86ms
step:4344/4740 train_time:942165ms step_avg:216.89ms
step:4345/4740 train_time:942341ms step_avg:216.88ms
step:4346/4740 train_time:942661ms step_avg:216.90ms
step:4347/4740 train_time:942841ms step_avg:216.89ms
step:4348/4740 train_time:943162ms step_avg:216.92ms
step:4349/4740 train_time:943342ms step_avg:216.91ms
step:4350/4740 train_time:943664ms step_avg:216.93ms
step:4351/4740 train_time:943843ms step_avg:216.93ms
step:4352/4740 train_time:944169ms step_avg:216.95ms
step:4353/4740 train_time:944346ms step_avg:216.94ms
step:4354/4740 train_time:944665ms step_avg:216.96ms
step:4355/4740 train_time:944845ms step_avg:216.96ms
step:4356/4740 train_time:945167ms step_avg:216.98ms
step:4357/4740 train_time:945350ms step_avg:216.97ms
step:4358/4740 train_time:945674ms step_avg:217.00ms
step:4359/4740 train_time:945855ms step_avg:216.99ms
step:4360/4740 train_time:946176ms step_avg:217.01ms
step:4361/4740 train_time:946352ms step_avg:217.00ms
step:4362/4740 train_time:946676ms step_avg:217.03ms
step:4363/4740 train_time:946854ms step_avg:217.02ms
step:4364/4740 train_time:947179ms step_avg:217.04ms
step:4365/4740 train_time:947356ms step_avg:217.03ms
step:4366/4740 train_time:947677ms step_avg:217.06ms
step:4367/4740 train_time:947854ms step_avg:217.05ms
step:4368/4740 train_time:948181ms step_avg:217.07ms
step:4369/4740 train_time:948357ms step_avg:217.06ms
step:4370/4740 train_time:948681ms step_avg:217.09ms
step:4371/4740 train_time:948856ms step_avg:217.08ms
step:4372/4740 train_time:949181ms step_avg:217.10ms
step:4373/4740 train_time:949356ms step_avg:217.09ms
step:4374/4740 train_time:949680ms step_avg:217.12ms
step:4375/4740 train_time:949855ms step_avg:217.11ms
step:4376/4740 train_time:950179ms step_avg:217.13ms
step:4377/4740 train_time:950356ms step_avg:217.12ms
step:4378/4740 train_time:950677ms step_avg:217.15ms
step:4379/4740 train_time:950856ms step_avg:217.14ms
step:4380/4740 train_time:951179ms step_avg:217.16ms
step:4381/4740 train_time:951356ms step_avg:217.15ms
step:4382/4740 train_time:951680ms step_avg:217.18ms
step:4383/4740 train_time:951855ms step_avg:217.17ms
step:4384/4740 train_time:952178ms step_avg:217.19ms
step:4385/4740 train_time:952360ms step_avg:217.19ms
step:4386/4740 train_time:952682ms step_avg:217.21ms
step:4387/4740 train_time:952862ms step_avg:217.20ms
step:4388/4740 train_time:953180ms step_avg:217.22ms
step:4389/4740 train_time:953359ms step_avg:217.22ms
step:4390/4740 train_time:953681ms step_avg:217.24ms
step:4391/4740 train_time:953863ms step_avg:217.23ms
step:4392/4740 train_time:954184ms step_avg:217.26ms
step:4393/4740 train_time:954361ms step_avg:217.25ms
step:4394/4740 train_time:954681ms step_avg:217.27ms
step:4395/4740 train_time:954860ms step_avg:217.26ms
step:4396/4740 train_time:955182ms step_avg:217.28ms
step:4397/4740 train_time:955361ms step_avg:217.28ms
step:4398/4740 train_time:955681ms step_avg:217.30ms
step:4399/4740 train_time:955861ms step_avg:217.29ms
step:4400/4740 train_time:956183ms step_avg:217.31ms
step:4401/4740 train_time:956363ms step_avg:217.31ms
step:4402/4740 train_time:956687ms step_avg:217.33ms
step:4403/4740 train_time:956866ms step_avg:217.32ms
step:4404/4740 train_time:957192ms step_avg:217.35ms
step:4405/4740 train_time:957368ms step_avg:217.34ms
step:4406/4740 train_time:957688ms step_avg:217.36ms
step:4407/4740 train_time:957868ms step_avg:217.35ms
step:4408/4740 train_time:958191ms step_avg:217.38ms
step:4409/4740 train_time:958371ms step_avg:217.37ms
step:4410/4740 train_time:958693ms step_avg:217.39ms
step:4411/4740 train_time:958870ms step_avg:217.38ms
step:4412/4740 train_time:959192ms step_avg:217.41ms
step:4413/4740 train_time:959371ms step_avg:217.40ms
step:4414/4740 train_time:959692ms step_avg:217.42ms
step:4415/4740 train_time:959867ms step_avg:217.41ms
step:4416/4740 train_time:960187ms step_avg:217.43ms
step:4417/4740 train_time:960372ms step_avg:217.43ms
step:4418/4740 train_time:960691ms step_avg:217.45ms
step:4419/4740 train_time:960870ms step_avg:217.44ms
step:4420/4740 train_time:961193ms step_avg:217.46ms
step:4421/4740 train_time:961372ms step_avg:217.46ms
step:4422/4740 train_time:961692ms step_avg:217.48ms
step:4423/4740 train_time:961871ms step_avg:217.47ms
step:4424/4740 train_time:962194ms step_avg:217.49ms
step:4425/4740 train_time:962373ms step_avg:217.49ms
step:4426/4740 train_time:962694ms step_avg:217.51ms
step:4427/4740 train_time:962872ms step_avg:217.50ms
step:4428/4740 train_time:963194ms step_avg:217.52ms
step:4429/4740 train_time:963372ms step_avg:217.51ms
step:4430/4740 train_time:963695ms step_avg:217.54ms
step:4431/4740 train_time:963875ms step_avg:217.53ms
step:4432/4740 train_time:964198ms step_avg:217.55ms
step:4433/4740 train_time:964375ms step_avg:217.54ms
step:4434/4740 train_time:964695ms step_avg:217.57ms
step:4435/4740 train_time:964875ms step_avg:217.56ms
step:4436/4740 train_time:965195ms step_avg:217.58ms
step:4437/4740 train_time:965373ms step_avg:217.57ms
step:4438/4740 train_time:965698ms step_avg:217.60ms
step:4439/4740 train_time:965877ms step_avg:217.59ms
step:4440/4740 train_time:966198ms step_avg:217.61ms
step:4441/4740 train_time:966375ms step_avg:217.60ms
step:4442/4740 train_time:966699ms step_avg:217.63ms
step:4443/4740 train_time:966877ms step_avg:217.62ms
step:4444/4740 train_time:967197ms step_avg:217.64ms
step:4445/4740 train_time:967375ms step_avg:217.63ms
step:4446/4740 train_time:967697ms step_avg:217.66ms
step:4447/4740 train_time:967882ms step_avg:217.65ms
step:4448/4740 train_time:968200ms step_avg:217.67ms
step:4449/4740 train_time:968379ms step_avg:217.66ms
step:4450/4740 train_time:968701ms step_avg:217.69ms
step:4451/4740 train_time:968881ms step_avg:217.68ms
step:4452/4740 train_time:969200ms step_avg:217.70ms
step:4453/4740 train_time:969378ms step_avg:217.69ms
step:4454/4740 train_time:969701ms step_avg:217.71ms
step:4455/4740 train_time:969881ms step_avg:217.71ms
step:4456/4740 train_time:970200ms step_avg:217.73ms
step:4457/4740 train_time:970381ms step_avg:217.72ms
step:4458/4740 train_time:970699ms step_avg:217.74ms
step:4459/4740 train_time:970881ms step_avg:217.74ms
step:4460/4740 train_time:971202ms step_avg:217.76ms
step:4461/4740 train_time:971383ms step_avg:217.75ms
step:4462/4740 train_time:971702ms step_avg:217.77ms
step:4463/4740 train_time:971887ms step_avg:217.77ms
step:4464/4740 train_time:972212ms step_avg:217.79ms
step:4465/4740 train_time:972389ms step_avg:217.78ms
step:4466/4740 train_time:972709ms step_avg:217.80ms
step:4467/4740 train_time:972888ms step_avg:217.79ms
step:4468/4740 train_time:973214ms step_avg:217.82ms
step:4469/4740 train_time:973393ms step_avg:217.81ms
step:4470/4740 train_time:973717ms step_avg:217.83ms
step:4471/4740 train_time:973894ms step_avg:217.82ms
step:4472/4740 train_time:974216ms step_avg:217.85ms
step:4473/4740 train_time:974395ms step_avg:217.84ms
step:4474/4740 train_time:974715ms step_avg:217.86ms
step:4475/4740 train_time:974895ms step_avg:217.85ms
step:4476/4740 train_time:975218ms step_avg:217.88ms
step:4477/4740 train_time:975392ms step_avg:217.87ms
step:4478/4740 train_time:975715ms step_avg:217.89ms
step:4479/4740 train_time:975891ms step_avg:217.88ms
step:4480/4740 train_time:976220ms step_avg:217.91ms
step:4481/4740 train_time:976396ms step_avg:217.90ms
step:4482/4740 train_time:976719ms step_avg:217.92ms
step:4483/4740 train_time:976896ms step_avg:217.91ms
step:4484/4740 train_time:977219ms step_avg:217.93ms
step:4485/4740 train_time:977398ms step_avg:217.93ms
step:4486/4740 train_time:977717ms step_avg:217.95ms
step:4487/4740 train_time:977899ms step_avg:217.94ms
step:4488/4740 train_time:978218ms step_avg:217.96ms
step:4489/4740 train_time:978395ms step_avg:217.95ms
step:4490/4740 train_time:978717ms step_avg:217.98ms
step:4491/4740 train_time:978895ms step_avg:217.97ms
step:4492/4740 train_time:979216ms step_avg:217.99ms
step:4493/4740 train_time:979394ms step_avg:217.98ms
step:4494/4740 train_time:979718ms step_avg:218.01ms
step:4495/4740 train_time:979897ms step_avg:218.00ms
step:4496/4740 train_time:980219ms step_avg:218.02ms
step:4497/4740 train_time:980398ms step_avg:218.01ms
step:4498/4740 train_time:980725ms step_avg:218.04ms
step:4499/4740 train_time:980903ms step_avg:218.03ms
step:4500/4740 train_time:981220ms step_avg:218.05ms
step:4500/4740 val_loss:2.9340 train_time:981347ms step_avg:218.08ms
step:4501/4740 train_time:981405ms step_avg:218.04ms
step:4502/4740 train_time:981739ms step_avg:218.07ms
step:4503/4740 train_time:981913ms step_avg:218.06ms
step:4504/4740 train_time:982230ms step_avg:218.08ms
step:4505/4740 train_time:982404ms step_avg:218.07ms
step:4506/4740 train_time:982733ms step_avg:218.09ms
step:4507/4740 train_time:982913ms step_avg:218.09ms
step:4508/4740 train_time:983233ms step_avg:218.11ms
step:4509/4740 train_time:983410ms step_avg:218.10ms
step:4510/4740 train_time:983736ms step_avg:218.12ms
step:4511/4740 train_time:983915ms step_avg:218.11ms
step:4512/4740 train_time:984234ms step_avg:218.14ms
step:4513/4740 train_time:984412ms step_avg:218.13ms
step:4514/4740 train_time:984738ms step_avg:218.15ms
step:4515/4740 train_time:984917ms step_avg:218.14ms
step:4516/4740 train_time:985239ms step_avg:218.17ms
step:4517/4740 train_time:985418ms step_avg:218.16ms
step:4518/4740 train_time:985744ms step_avg:218.18ms
step:4519/4740 train_time:985924ms step_avg:218.17ms
step:4520/4740 train_time:986244ms step_avg:218.20ms
step:4521/4740 train_time:986420ms step_avg:218.19ms
step:4522/4740 train_time:986743ms step_avg:218.21ms
step:4523/4740 train_time:986923ms step_avg:218.20ms
step:4524/4740 train_time:987243ms step_avg:218.22ms
step:4525/4740 train_time:987425ms step_avg:218.22ms
step:4526/4740 train_time:987744ms step_avg:218.24ms
step:4527/4740 train_time:987925ms step_avg:218.23ms
step:4528/4740 train_time:988244ms step_avg:218.25ms
step:4529/4740 train_time:988424ms step_avg:218.24ms
step:4530/4740 train_time:988745ms step_avg:218.27ms
step:4531/4740 train_time:988923ms step_avg:218.26ms
step:4532/4740 train_time:989243ms step_avg:218.28ms
step:4533/4740 train_time:989425ms step_avg:218.27ms
step:4534/4740 train_time:989745ms step_avg:218.29ms
step:4535/4740 train_time:989925ms step_avg:218.29ms
step:4536/4740 train_time:990248ms step_avg:218.31ms
step:4537/4740 train_time:990428ms step_avg:218.30ms
step:4538/4740 train_time:990751ms step_avg:218.32ms
step:4539/4740 train_time:990929ms step_avg:218.31ms
step:4540/4740 train_time:991252ms step_avg:218.34ms
step:4541/4740 train_time:991430ms step_avg:218.33ms
step:4542/4740 train_time:991750ms step_avg:218.35ms
step:4543/4740 train_time:991932ms step_avg:218.34ms
step:4544/4740 train_time:992256ms step_avg:218.37ms
step:4545/4740 train_time:992434ms step_avg:218.36ms
step:4546/4740 train_time:992754ms step_avg:218.38ms
step:4547/4740 train_time:992932ms step_avg:218.37ms
step:4548/4740 train_time:993257ms step_avg:218.39ms
step:4549/4740 train_time:993438ms step_avg:218.39ms
step:4550/4740 train_time:993761ms step_avg:218.41ms
step:4551/4740 train_time:993938ms step_avg:218.40ms
step:4552/4740 train_time:994262ms step_avg:218.42ms
step:4553/4740 train_time:994441ms step_avg:218.41ms
step:4554/4740 train_time:994764ms step_avg:218.44ms
step:4555/4740 train_time:994940ms step_avg:218.43ms
step:4556/4740 train_time:995262ms step_avg:218.45ms
step:4557/4740 train_time:995441ms step_avg:218.44ms
step:4558/4740 train_time:995760ms step_avg:218.46ms
step:4559/4740 train_time:995945ms step_avg:218.46ms
step:4560/4740 train_time:996270ms step_avg:218.48ms
step:4561/4740 train_time:996452ms step_avg:218.47ms
step:4562/4740 train_time:996773ms step_avg:218.49ms
step:4563/4740 train_time:996951ms step_avg:218.49ms
step:4564/4740 train_time:997274ms step_avg:218.51ms
step:4565/4740 train_time:997456ms step_avg:218.50ms
step:4566/4740 train_time:997777ms step_avg:218.52ms
step:4567/4740 train_time:997956ms step_avg:218.51ms
step:4568/4740 train_time:998281ms step_avg:218.54ms
step:4569/4740 train_time:998463ms step_avg:218.53ms
step:4570/4740 train_time:998789ms step_avg:218.55ms
step:4571/4740 train_time:998966ms step_avg:218.54ms
step:4572/4740 train_time:999287ms step_avg:218.57ms
step:4573/4740 train_time:999467ms step_avg:218.56ms
step:4574/4740 train_time:999793ms step_avg:218.58ms
step:4575/4740 train_time:999970ms step_avg:218.57ms
step:4576/4740 train_time:1000293ms step_avg:218.60ms
step:4577/4740 train_time:1000468ms step_avg:218.59ms
step:4578/4740 train_time:1000789ms step_avg:218.61ms
step:4579/4740 train_time:1000971ms step_avg:218.60ms
step:4580/4740 train_time:1001293ms step_avg:218.62ms
step:4581/4740 train_time:1001471ms step_avg:218.61ms
step:4582/4740 train_time:1001793ms step_avg:218.64ms
step:4583/4740 train_time:1001971ms step_avg:218.63ms
step:4584/4740 train_time:1002294ms step_avg:218.65ms
step:4585/4740 train_time:1002473ms step_avg:218.64ms
step:4586/4740 train_time:1002796ms step_avg:218.66ms
step:4587/4740 train_time:1002971ms step_avg:218.66ms
step:4588/4740 train_time:1003294ms step_avg:218.68ms
step:4589/4740 train_time:1003472ms step_avg:218.67ms
step:4590/4740 train_time:1003796ms step_avg:218.69ms
step:4591/4740 train_time:1003973ms step_avg:218.68ms
step:4592/4740 train_time:1004296ms step_avg:218.71ms
step:4593/4740 train_time:1004475ms step_avg:218.70ms
step:4594/4740 train_time:1004798ms step_avg:218.72ms
step:4595/4740 train_time:1004974ms step_avg:218.71ms
step:4596/4740 train_time:1005296ms step_avg:218.73ms
step:4597/4740 train_time:1005474ms step_avg:218.72ms
step:4598/4740 train_time:1005797ms step_avg:218.75ms
step:4599/4740 train_time:1005975ms step_avg:218.74ms
step:4600/4740 train_time:1006299ms step_avg:218.76ms
step:4601/4740 train_time:1006477ms step_avg:218.75ms
step:4602/4740 train_time:1006801ms step_avg:218.77ms
step:4603/4740 train_time:1006978ms step_avg:218.77ms
step:4604/4740 train_time:1007303ms step_avg:218.79ms
step:4605/4740 train_time:1007481ms step_avg:218.78ms
step:4606/4740 train_time:1007805ms step_avg:218.80ms
step:4607/4740 train_time:1007980ms step_avg:218.79ms
step:4608/4740 train_time:1008300ms step_avg:218.82ms
step:4609/4740 train_time:1008477ms step_avg:218.81ms
step:4610/4740 train_time:1008803ms step_avg:218.83ms
step:4611/4740 train_time:1008982ms step_avg:218.82ms
step:4612/4740 train_time:1009303ms step_avg:218.84ms
step:4613/4740 train_time:1009483ms step_avg:218.83ms
step:4614/4740 train_time:1009809ms step_avg:218.86ms
step:4615/4740 train_time:1009989ms step_avg:218.85ms
step:4616/4740 train_time:1010311ms step_avg:218.87ms
step:4617/4740 train_time:1010491ms step_avg:218.86ms
step:4618/4740 train_time:1010812ms step_avg:218.89ms
step:4619/4740 train_time:1010992ms step_avg:218.88ms
step:4620/4740 train_time:1011314ms step_avg:218.90ms
step:4621/4740 train_time:1011494ms step_avg:218.89ms
step:4622/4740 train_time:1011813ms step_avg:218.91ms
step:4623/4740 train_time:1011994ms step_avg:218.90ms
step:4624/4740 train_time:1012313ms step_avg:218.93ms
step:4625/4740 train_time:1012493ms step_avg:218.92ms
step:4626/4740 train_time:1012815ms step_avg:218.94ms
step:4627/4740 train_time:1012997ms step_avg:218.93ms
step:4628/4740 train_time:1013323ms step_avg:218.95ms
step:4629/4740 train_time:1013500ms step_avg:218.95ms
step:4630/4740 train_time:1013820ms step_avg:218.97ms
step:4631/4740 train_time:1013999ms step_avg:218.96ms
step:4632/4740 train_time:1014318ms step_avg:218.98ms
step:4633/4740 train_time:1014501ms step_avg:218.97ms
step:4634/4740 train_time:1014821ms step_avg:218.99ms
step:4635/4740 train_time:1015003ms step_avg:218.99ms
step:4636/4740 train_time:1015321ms step_avg:219.01ms
step:4637/4740 train_time:1015504ms step_avg:219.00ms
step:4638/4740 train_time:1015820ms step_avg:219.02ms
step:4639/4740 train_time:1016004ms step_avg:219.01ms
step:4640/4740 train_time:1016324ms step_avg:219.04ms
step:4641/4740 train_time:1016504ms step_avg:219.03ms
step:4642/4740 train_time:1016821ms step_avg:219.05ms
step:4643/4740 train_time:1017002ms step_avg:219.04ms
step:4644/4740 train_time:1017320ms step_avg:219.06ms
step:4645/4740 train_time:1017502ms step_avg:219.05ms
step:4646/4740 train_time:1017822ms step_avg:219.07ms
step:4647/4740 train_time:1018004ms step_avg:219.07ms
step:4648/4740 train_time:1018325ms step_avg:219.09ms
step:4649/4740 train_time:1018506ms step_avg:219.08ms
step:4650/4740 train_time:1018824ms step_avg:219.10ms
step:4651/4740 train_time:1019006ms step_avg:219.09ms
step:4652/4740 train_time:1019332ms step_avg:219.12ms
step:4653/4740 train_time:1019512ms step_avg:219.11ms
step:4654/4740 train_time:1019830ms step_avg:219.13ms
step:4655/4740 train_time:1020009ms step_avg:219.12ms
step:4656/4740 train_time:1020336ms step_avg:219.14ms
step:4657/4740 train_time:1020514ms step_avg:219.14ms
step:4658/4740 train_time:1020838ms step_avg:219.16ms
step:4659/4740 train_time:1021015ms step_avg:219.15ms
step:4660/4740 train_time:1021340ms step_avg:219.17ms
step:4661/4740 train_time:1021519ms step_avg:219.16ms
step:4662/4740 train_time:1021838ms step_avg:219.18ms
step:4663/4740 train_time:1022016ms step_avg:219.18ms
step:4664/4740 train_time:1022339ms step_avg:219.20ms
step:4665/4740 train_time:1022516ms step_avg:219.19ms
step:4666/4740 train_time:1022838ms step_avg:219.21ms
step:4667/4740 train_time:1023016ms step_avg:219.20ms
step:4668/4740 train_time:1023339ms step_avg:219.22ms
step:4669/4740 train_time:1023519ms step_avg:219.22ms
step:4670/4740 train_time:1023842ms step_avg:219.24ms
step:4671/4740 train_time:1024020ms step_avg:219.23ms
step:4672/4740 train_time:1024340ms step_avg:219.25ms
step:4673/4740 train_time:1024522ms step_avg:219.24ms
step:4674/4740 train_time:1024841ms step_avg:219.26ms
step:4675/4740 train_time:1025022ms step_avg:219.26ms
step:4676/4740 train_time:1025343ms step_avg:219.28ms
step:4677/4740 train_time:1025520ms step_avg:219.27ms
step:4678/4740 train_time:1025842ms step_avg:219.29ms
step:4679/4740 train_time:1026024ms step_avg:219.28ms
step:4680/4740 train_time:1026346ms step_avg:219.30ms
step:4681/4740 train_time:1026523ms step_avg:219.30ms
step:4682/4740 train_time:1026845ms step_avg:219.32ms
step:4683/4740 train_time:1027024ms step_avg:219.31ms
step:4684/4740 train_time:1027344ms step_avg:219.33ms
step:4685/4740 train_time:1027526ms step_avg:219.32ms
step:4686/4740 train_time:1027852ms step_avg:219.35ms
step:4687/4740 train_time:1028030ms step_avg:219.34ms
step:4688/4740 train_time:1028350ms step_avg:219.36ms
step:4689/4740 train_time:1028529ms step_avg:219.35ms
step:4690/4740 train_time:1028852ms step_avg:219.37ms
step:4691/4740 train_time:1029028ms step_avg:219.36ms
step:4692/4740 train_time:1029344ms step_avg:219.38ms
step:4693/4740 train_time:1029524ms step_avg:219.37ms
step:4694/4740 train_time:1029847ms step_avg:219.40ms
step:4695/4740 train_time:1030024ms step_avg:219.39ms
step:4696/4740 train_time:1030345ms step_avg:219.41ms
step:4697/4740 train_time:1030523ms step_avg:219.40ms
step:4698/4740 train_time:1030846ms step_avg:219.42ms
step:4699/4740 train_time:1031023ms step_avg:219.41ms
step:4700/4740 train_time:1031344ms step_avg:219.43ms
step:4701/4740 train_time:1031523ms step_avg:219.43ms
step:4702/4740 train_time:1031844ms step_avg:219.45ms
step:4703/4740 train_time:1032024ms step_avg:219.44ms
step:4704/4740 train_time:1032345ms step_avg:219.46ms
step:4705/4740 train_time:1032528ms step_avg:219.45ms
step:4706/4740 train_time:1032851ms step_avg:219.48ms
step:4707/4740 train_time:1033028ms step_avg:219.47ms
step:4708/4740 train_time:1033347ms step_avg:219.49ms
step:4709/4740 train_time:1033531ms step_avg:219.48ms
step:4710/4740 train_time:1033853ms step_avg:219.50ms
step:4711/4740 train_time:1034033ms step_avg:219.49ms
step:4712/4740 train_time:1034358ms step_avg:219.52ms
step:4713/4740 train_time:1034534ms step_avg:219.51ms
step:4714/4740 train_time:1034858ms step_avg:219.53ms
step:4715/4740 train_time:1035033ms step_avg:219.52ms
step:4716/4740 train_time:1035356ms step_avg:219.54ms
step:4717/4740 train_time:1035533ms step_avg:219.53ms
step:4718/4740 train_time:1035861ms step_avg:219.56ms
step:4719/4740 train_time:1036037ms step_avg:219.55ms
step:4720/4740 train_time:1036361ms step_avg:219.57ms
step:4721/4740 train_time:1036537ms step_avg:219.56ms
step:4722/4740 train_time:1036860ms step_avg:219.58ms
step:4723/4740 train_time:1037040ms step_avg:219.57ms
step:4724/4740 train_time:1037356ms step_avg:219.59ms
step:4725/4740 train_time:1037537ms step_avg:219.58ms
step:4726/4740 train_time:1037860ms step_avg:219.61ms
step:4727/4740 train_time:1038038ms step_avg:219.60ms
step:4728/4740 train_time:1038356ms step_avg:219.62ms
step:4729/4740 train_time:1038535ms step_avg:219.61ms
step:4730/4740 train_time:1038859ms step_avg:219.63ms
step:4731/4740 train_time:1039037ms step_avg:219.62ms
step:4732/4740 train_time:1039359ms step_avg:219.64ms
step:4733/4740 train_time:1039536ms step_avg:219.64ms
step:4734/4740 train_time:1039860ms step_avg:219.66ms
step:4735/4740 train_time:1040039ms step_avg:219.65ms
step:4736/4740 train_time:1040361ms step_avg:219.67ms
step:4737/4740 train_time:1040537ms step_avg:219.66ms
step:4738/4740 train_time:1040863ms step_avg:219.68ms
step:4739/4740 train_time:1041040ms step_avg:219.68ms
step:4740/4740 train_time:1041362ms step_avg:219.70ms
step:4740/4740 val_loss:2.9166 train_time:1041483ms step_avg:219.72ms
peak memory allocated: 57397 MiB reserved: 62992 MiB