sparse-checkpoints / epfl_final /log_files /sparse_eval_dense.log
rubentium's picture
EPFL ALiBi and training provenance 551-600 of 1184
1e7c2c5 verified
Raw History Blame Contribute Delete
102 kB
nohup: ignoring input
*****************************************
Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed.
*****************************************
New run: Checkpoints will be saved to ./checkpoints/run_20260616_011729
Running config:
{
"model_path": "/mloscratch/homes/navasard/qwen3_06b_base",
"model_dtype": "bfloat16",
"init_from_scratch": true,
"train_path": "/mloscratch/homes/navasard/data/fineweb_large/tokenized",
"val_path": null,
"max_seq_length": 8192,
"num_workers": 8,
"output_dir": "./checkpoints/run_20260616_011729",
"resume_from_checkpoint": "checkpoints/run_20260523_215035/checkpoint-40000",
"force_new_run": true,
"num_epochs": 1,
"max_steps": null,
"per_device_batch_size": 2,
"gradient_accumulation_steps": 10,
"learning_rate": 0.0005,
"weight_decay": 0.01,
"warmup_steps": 750,
"lr_scheduler": "cosine",
"max_grad_norm": 1.0,
"save_steps": 1000,
"eval_steps": 250,
"eval_batches": 50,
"logging_steps": 10,
"seed": 42,
"verbose": true,
"checkpoint_activations": false,
"strategy": "ddp",
"mixed_precision": "bf16",
"wandb_enabled": true,
"wandb_project": "stochastic_sparsity",
"wandb_name": "sparse 1mask",
"wandb_tags": [
"qwen3",
"1.2b",
"sparse_stochastic"
],
"hidden_size": 1536,
"intermediate_size": 6144,
"num_attention_heads": 12,
"num_key_value_heads": 4,
"num_hidden_layers": null,
"head_dim": null,
"attention_type": "sparse_stochastic",
"sparse_block_size": 128,
"sparse_num_random_blocks": 31,
"sparse_mask_pool_size": 1,
"sparse_val_dense_compare": false,
"sparse_layer_indices": null,
"sparse_mask_cache_dir": null,
"dense_mode": false,
"dense_step_intervals": null
}
Loading tokenizer from /mloscratch/homes/navasard/qwen3_06b_base...
Resuming from checkpoint: checkpoints/run_20260523_215035/checkpoint-40000
Checkpoint activations: False
Checkpoint activations: False
Checkpoint activations: False
================================================================================
MODEL ARCHITECTURE
================================================================================
Model config:
vocab_size: 151936
hidden_size: 1536
num_hidden_layers: 28
num_attention_heads: 12
num_key_value_heads: 4
head_dim: 128
intermediate_size: 6144
max_position_embeddings: 32768
rope_theta: 1000000
attention_type: sparse_stochastic
sparse_block_size: 128
sparse_num_random_blocks: 31
sparse_mask_pool_size: 1
sparse_max_seq_length: 8192
Parameter count:
Total parameters: 1,202,352,640
Trainable parameters: 1,202,352,640
Non-trainable parameters: 0
Model size (BF16): 2.40 GB
--------------------------------------------------------------------------------
LAYER-BY-LAYER SUMMARY
--------------------------------------------------------------------------------
/home/navasard/.local/lib/python3.10/site-packages/torch/nn/attention/flex_attention.py:2055: UserWarning: flex_attention called without torch.compile() - this will use an unfused implementation that materializes the full scores matrix instead of generating a fused kernel.
SOLUTION: Use torch.compile(flex_attention)(...)
If you want to debug your score_mod/mask_mod, you can set:
torch.nn.attention.flex_attention._FLEX_ATTENTION_DISABLE_COMPILE_DEBUG = True
This will allow you to use print statements or breakpoints. Note: This doesn't work with the backwards pass and may produce incorrect results.
_warn_once(
================================================================================================================================================================
Layer (type:depth-idx) Input Shape Output Shape Param # Trainable
================================================================================================================================================================
Qwen3ForCausalLM [2, 8192] -- -- True
β”œβ”€Qwen3Model: 1-1 -- [2, 8192, 1536] -- True
β”‚ └─Embedding: 2-1 [2, 8192] [2, 8192, 1536] 233,373,696 True
β”‚ └─ModuleList: 2-2 -- -- -- True
β”‚ β”‚ └─Qwen3DecoderLayer: 3-1 [2, 8192, 1536] [2, 8192, 1536] 34,606,336 True
β”‚ β”‚ └─Qwen3DecoderLayer: 3-2 [2, 8192, 1536] [2, 8192, 1536] 34,606,336 True
β”‚ β”‚ └─Qwen3DecoderLayer: 3-3 [2, 8192, 1536] [2, 8192, 1536] 34,606,336 True
β”‚ β”‚ └─Qwen3DecoderLayer: 3-4 [2, 8192, 1536] [2, 8192, 1536] 34,606,336 True
β”‚ β”‚ └─Qwen3DecoderLayer: 3-5 [2, 8192, 1536] [2, 8192, 1536] 34,606,336 True
β”‚ β”‚ └─Qwen3DecoderLayer: 3-6 [2, 8192, 1536] [2, 8192, 1536] 34,606,336 True
β”‚ β”‚ └─Qwen3DecoderLayer: 3-7 [2, 8192, 1536] [2, 8192, 1536] 34,606,336 True
β”‚ β”‚ └─Qwen3DecoderLayer: 3-8 [2, 8192, 1536] [2, 8192, 1536] 34,606,336 True
β”‚ β”‚ └─Qwen3DecoderLayer: 3-9 [2, 8192, 1536] [2, 8192, 1536] 34,606,336 True
β”‚ β”‚ └─Qwen3DecoderLayer: 3-10 [2, 8192, 1536] [2, 8192, 1536] 34,606,336 True
β”‚ β”‚ └─Qwen3DecoderLayer: 3-11 [2, 8192, 1536] [2, 8192, 1536] 34,606,336 True
β”‚ β”‚ └─Qwen3DecoderLayer: 3-12 [2, 8192, 1536] [2, 8192, 1536] 34,606,336 True
β”‚ β”‚ └─Qwen3DecoderLayer: 3-13 [2, 8192, 1536] [2, 8192, 1536] 34,606,336 True
β”‚ β”‚ └─Qwen3DecoderLayer: 3-14 [2, 8192, 1536] [2, 8192, 1536] 34,606,336 True
β”‚ β”‚ └─Qwen3DecoderLayer: 3-15 [2, 8192, 1536] [2, 8192, 1536] 34,606,336 True
β”‚ β”‚ └─Qwen3DecoderLayer: 3-16 [2, 8192, 1536] [2, 8192, 1536] 34,606,336 True
β”‚ β”‚ └─Qwen3DecoderLayer: 3-17 [2, 8192, 1536] [2, 8192, 1536] 34,606,336 True
β”‚ β”‚ └─Qwen3DecoderLayer: 3-18 [2, 8192, 1536] [2, 8192, 1536] 34,606,336 True
β”‚ β”‚ └─Qwen3DecoderLayer: 3-19 [2, 8192, 1536] [2, 8192, 1536] 34,606,336 True
β”‚ β”‚ └─Qwen3DecoderLayer: 3-20 [2, 8192, 1536] [2, 8192, 1536] 34,606,336 True
β”‚ β”‚ └─Qwen3DecoderLayer: 3-21 [2, 8192, 1536] [2, 8192, 1536] 34,606,336 True
β”‚ β”‚ └─Qwen3DecoderLayer: 3-22 [2, 8192, 1536] [2, 8192, 1536] 34,606,336 True
β”‚ β”‚ └─Qwen3DecoderLayer: 3-23 [2, 8192, 1536] [2, 8192, 1536] 34,606,336 True
β”‚ β”‚ └─Qwen3DecoderLayer: 3-24 [2, 8192, 1536] [2, 8192, 1536] 34,606,336 True
β”‚ β”‚ └─Qwen3DecoderLayer: 3-25 [2, 8192, 1536] [2, 8192, 1536] 34,606,336 True
β”‚ β”‚ └─Qwen3DecoderLayer: 3-26 [2, 8192, 1536] [2, 8192, 1536] 34,606,336 True
β”‚ β”‚ └─Qwen3DecoderLayer: 3-27 [2, 8192, 1536] [2, 8192, 1536] 34,606,336 True
β”‚ β”‚ └─Qwen3DecoderLayer: 3-28 [2, 8192, 1536] [2, 8192, 1536] 34,606,336 True
β”‚ └─Qwen3RMSNorm: 2-3 [2, 8192, 1536] [2, 8192, 1536] 1,536 True
β”œβ”€Linear: 1-2 [2, 8192, 1536] [2, 8192, 151936] 233,373,696 True
================================================================================================================================================================
Total params: 1,435,726,336
Trainable params: 1,435,726,336
Non-trainable params: 0
Total mult-adds (G): 2.87
================================================================================================================================================================
Input size (MB): 0.13
Forward/backward pass size (MB): 52437.19
Params size (MB): 2871.45
Estimated Total Size (MB): 55308.77
================================================================================================================================================================
NOTE: torchinfo double-counts tied embed_tokens/lm_head weight (233,373,696 params). True unique parameter count: 1,202,352,640 (1202.4M)
================================================================================
Loading dataset from /mloscratch/homes/navasard/data/fineweb_large/tokenized...
Loading pre-tokenized HF dataset from /mloscratch/homes/navasard/data/fineweb_large/tokenized...Loading pre-tokenized HF dataset from /mloscratch/homes/navasard/data/fineweb_large/tokenized...
Using pre-tokenized dataset (no runtime tokenization)
Loading pre-tokenized HF dataset from /mloscratch/homes/navasard/data/fineweb_large/tokenized...
Loaded 12748481 pre-tokenized examples (stored_len=2048, pack_factor=4, packed_examples=3187120)
Loaded 12748481 pre-tokenized examples (stored_len=2048, pack_factor=4, packed_examples=3187120)
Loaded 12748481 pre-tokenized examples (stored_len=2048, pack_factor=4, packed_examples=3187120)
Dataset split: 3027764 train, 159356 validation
Training examples (packed, train split): 3027764
Validation examples (packed, val split): 159356
Per-device batch size: 2
World size (GPUs): 3
Gradient accumulation steps: 10
Effective batch size (seqs/step): 60
Sequence length: 8192
Tokens per optimizer step: 491,520
Micro-batches per epoch (per rank): 504627
Optimizer steps per epoch: 50462
Total optimizer steps: 50462
Loading checkpoint from checkpoints/run_20260523_215035/checkpoint-40000...
Overriding checkpoint LR with config LR: 0.0005
Resumed from checkpoint at step 40000 (epoch 0, micro_step 400000)
Resuming training from step 40000
wandb: [wandb.login()] Loaded credentials for https://api.wandb.ai from WANDB_API_KEY.
wandb: Currently logged in as: ruben-navasardyan4869 (ruben-navasardyan4869-epfl) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
wandb: Tracking run with wandb version 0.27.2
wandb: Run data is saved locally in /mloscratch/homes/navasard/sparse/wandb/run-20260616_012002-apvt4901
wandb: Run `wandb offline` to turn off syncing.
wandb: Syncing run sparse 1mask
wandb: ⭐️ View project at https://wandb.ai/ruben-navasardyan4869-epfl/stochastic_sparsity
wandb: πŸš€ View run at https://wandb.ai/ruben-navasardyan4869-epfl/stochastic_sparsity/runs/apvt4901
Starting training...
Validation at step 40000: Loss = 2.7035, Perplexity = 14.93
Switching attention mode for step 40001: sparse_stochastic (configured sparse attention)
Step 40010/50462 | Loss: 2.7082 | LR: 5.26e-05 | 58207.0ms/step | 1.0 samples/s
Step 40020/50462 | Loss: 2.7067 | LR: 5.25e-05 | 3114.1ms/step | 19.3 samples/s
Step 40030/50462 | Loss: 2.6470 | LR: 5.24e-05 | 3121.8ms/step | 19.2 samples/s
Step 40040/50462 | Loss: 2.6722 | LR: 5.23e-05 | 3122.8ms/step | 19.2 samples/s
Step 40050/50462 | Loss: 2.6606 | LR: 5.22e-05 | 3124.2ms/step | 19.2 samples/s
Step 40060/50462 | Loss: 2.6534 | LR: 5.21e-05 | 3125.4ms/step | 19.2 samples/s
Step 40070/50462 | Loss: 2.6650 | LR: 5.20e-05 | 3113.3ms/step | 19.3 samples/s
Step 40080/50462 | Loss: 2.6391 | LR: 5.19e-05 | 3115.6ms/step | 19.3 samples/s
Step 40090/50462 | Loss: 2.6721 | LR: 5.18e-05 | 3119.6ms/step | 19.2 samples/s
Step 40100/50462 | Loss: 2.6684 | LR: 5.17e-05 | 3127.1ms/step | 19.2 samples/s
Step 40110/50462 | Loss: 2.6566 | LR: 5.16e-05 | 3124.9ms/step | 19.2 samples/s
Step 40120/50462 | Loss: 2.6681 | LR: 5.15e-05 | 3114.0ms/step | 19.3 samples/s
Step 40130/50462 | Loss: 2.6465 | LR: 5.14e-05 | 3117.6ms/step | 19.2 samples/s
Step 40140/50462 | Loss: 2.6470 | LR: 5.13e-05 | 3114.3ms/step | 19.3 samples/s
Step 40150/50462 | Loss: 2.6224 | LR: 5.12e-05 | 3119.1ms/step | 19.2 samples/s
Step 40160/50462 | Loss: 2.6181 | LR: 5.11e-05 | 3193.4ms/step | 18.8 samples/s
Step 40170/50462 | Loss: 2.6592 | LR: 5.10e-05 | 3130.4ms/step | 19.2 samples/s
Step 40180/50462 | Loss: 2.6618 | LR: 5.09e-05 | 3120.6ms/step | 19.2 samples/s
Step 40190/50462 | Loss: 2.6587 | LR: 5.09e-05 | 3126.3ms/step | 19.2 samples/s
Step 40200/50462 | Loss: 2.6662 | LR: 5.08e-05 | 3124.4ms/step | 19.2 samples/s
Step 40210/50462 | Loss: 2.6719 | LR: 5.07e-05 | 3184.7ms/step | 18.8 samples/s
Step 40220/50462 | Loss: 2.6685 | LR: 5.06e-05 | 3122.8ms/step | 19.2 samples/s
Step 40230/50462 | Loss: 2.6634 | LR: 5.05e-05 | 3125.6ms/step | 19.2 samples/s
Step 40240/50462 | Loss: 2.6897 | LR: 5.04e-05 | 3127.3ms/step | 19.2 samples/s
Step 40250/50462 | Loss: 2.6498 | LR: 5.03e-05 | 3119.3ms/step | 19.2 samples/s
Validation at step 40250: Loss = 2.6608, Perplexity = 14.31
Step 40260/50462 | Loss: 2.6581 | LR: 5.02e-05 | 4781.8ms/step | 12.5 samples/s
Step 40270/50462 | Loss: 2.6324 | LR: 5.01e-05 | 3123.0ms/step | 19.2 samples/s
Step 40280/50462 | Loss: 2.6559 | LR: 5.00e-05 | 3122.6ms/step | 19.2 samples/s
Step 40290/50462 | Loss: 2.6511 | LR: 4.99e-05 | 3126.6ms/step | 19.2 samples/s
Step 40300/50462 | Loss: 2.6480 | LR: 4.98e-05 | 3123.7ms/step | 19.2 samples/s
Step 40310/50462 | Loss: 2.6356 | LR: 4.97e-05 | 3129.0ms/step | 19.2 samples/s
Step 40320/50462 | Loss: 2.6541 | LR: 4.96e-05 | 3121.3ms/step | 19.2 samples/s
Step 40330/50462 | Loss: 2.6754 | LR: 4.95e-05 | 3123.3ms/step | 19.2 samples/s
Step 40340/50462 | Loss: 2.6381 | LR: 4.94e-05 | 3122.1ms/step | 19.2 samples/s
Step 40350/50462 | Loss: 2.6280 | LR: 4.93e-05 | 3120.3ms/step | 19.2 samples/s
Step 40360/50462 | Loss: 2.6519 | LR: 4.92e-05 | 3125.3ms/step | 19.2 samples/s
Step 40370/50462 | Loss: 2.6669 | LR: 4.91e-05 | 3127.9ms/step | 19.2 samples/s
Step 40380/50462 | Loss: 2.6537 | LR: 4.90e-05 | 3120.6ms/step | 19.2 samples/s
Step 40390/50462 | Loss: 2.6498 | LR: 4.90e-05 | 3125.7ms/step | 19.2 samples/s
Step 40400/50462 | Loss: 2.6641 | LR: 4.89e-05 | 3123.8ms/step | 19.2 samples/s
Step 40410/50462 | Loss: 2.6764 | LR: 4.88e-05 | 3124.2ms/step | 19.2 samples/s
Step 40420/50462 | Loss: 2.6737 | LR: 4.87e-05 | 3125.2ms/step | 19.2 samples/s
Step 40430/50462 | Loss: 2.6439 | LR: 4.86e-05 | 3129.5ms/step | 19.2 samples/s
Step 40440/50462 | Loss: 2.6604 | LR: 4.85e-05 | 3128.8ms/step | 19.2 samples/s
Step 40450/50462 | Loss: 2.6494 | LR: 4.84e-05 | 3123.3ms/step | 19.2 samples/s
Step 40460/50462 | Loss: 2.6850 | LR: 4.83e-05 | 3122.2ms/step | 19.2 samples/s
Step 40470/50462 | Loss: 2.6370 | LR: 4.82e-05 | 3124.7ms/step | 19.2 samples/s
Step 40480/50462 | Loss: 2.6473 | LR: 4.81e-05 | 3187.6ms/step | 18.8 samples/s
Step 40490/50462 | Loss: 2.6615 | LR: 4.80e-05 | 3121.7ms/step | 19.2 samples/s
Step 40500/50462 | Loss: 2.6541 | LR: 4.79e-05 | 3120.5ms/step | 19.2 samples/s
Validation at step 40500: Loss = 2.6585, Perplexity = 14.27
Step 40510/50462 | Loss: 2.6518 | LR: 4.78e-05 | 4784.4ms/step | 12.5 samples/s
Step 40520/50462 | Loss: 2.6362 | LR: 4.77e-05 | 3128.3ms/step | 19.2 samples/s
Step 40530/50462 | Loss: 2.6407 | LR: 4.76e-05 | 3180.9ms/step | 18.9 samples/s
Step 40540/50462 | Loss: 2.6172 | LR: 4.76e-05 | 3126.6ms/step | 19.2 samples/s
Step 40550/50462 | Loss: 2.6673 | LR: 4.75e-05 | 3118.5ms/step | 19.2 samples/s
Step 40560/50462 | Loss: 2.6612 | LR: 4.74e-05 | 3129.7ms/step | 19.2 samples/s
Step 40570/50462 | Loss: 2.6362 | LR: 4.73e-05 | 3125.2ms/step | 19.2 samples/s
Step 40580/50462 | Loss: 2.6611 | LR: 4.72e-05 | 3123.4ms/step | 19.2 samples/s
Step 40590/50462 | Loss: 2.6473 | LR: 4.71e-05 | 3121.1ms/step | 19.2 samples/s
Step 40600/50462 | Loss: 2.6488 | LR: 4.70e-05 | 3122.1ms/step | 19.2 samples/s
Step 40610/50462 | Loss: 2.6682 | LR: 4.69e-05 | 3124.5ms/step | 19.2 samples/s
Step 40620/50462 | Loss: 2.6237 | LR: 4.68e-05 | 3126.5ms/step | 19.2 samples/s
Step 40630/50462 | Loss: 2.6426 | LR: 4.67e-05 | 3121.7ms/step | 19.2 samples/s
Step 40640/50462 | Loss: 2.6532 | LR: 4.66e-05 | 3125.7ms/step | 19.2 samples/s
Step 40650/50462 | Loss: 2.6554 | LR: 4.65e-05 | 3123.1ms/step | 19.2 samples/s
Step 40660/50462 | Loss: 2.6634 | LR: 4.64e-05 | 3127.2ms/step | 19.2 samples/s
Step 40670/50462 | Loss: 2.6435 | LR: 4.64e-05 | 3125.6ms/step | 19.2 samples/s
Step 40680/50462 | Loss: 2.6486 | LR: 4.63e-05 | 3120.1ms/step | 19.2 samples/s
Step 40690/50462 | Loss: 2.6423 | LR: 4.62e-05 | 3122.6ms/step | 19.2 samples/s
Step 40700/50462 | Loss: 2.6177 | LR: 4.61e-05 | 3121.9ms/step | 19.2 samples/s
Step 40710/50462 | Loss: 2.6586 | LR: 4.60e-05 | 3124.7ms/step | 19.2 samples/s
Step 40720/50462 | Loss: 2.6209 | LR: 4.59e-05 | 3127.2ms/step | 19.2 samples/s
Step 40730/50462 | Loss: 2.6326 | LR: 4.58e-05 | 3119.0ms/step | 19.2 samples/s
Step 40740/50462 | Loss: 2.6319 | LR: 4.57e-05 | 3121.8ms/step | 19.2 samples/s
Step 40750/50462 | Loss: 2.6411 | LR: 4.56e-05 | 3125.7ms/step | 19.2 samples/s
Validation at step 40750: Loss = 2.6570, Perplexity = 14.25
Step 40760/50462 | Loss: 2.6677 | LR: 4.55e-05 | 4788.3ms/step | 12.5 samples/s
Step 40770/50462 | Loss: 2.6458 | LR: 4.54e-05 | 3128.4ms/step | 19.2 samples/s
Step 40780/50462 | Loss: 2.6611 | LR: 4.54e-05 | 3127.1ms/step | 19.2 samples/s
Step 40790/50462 | Loss: 2.6447 | LR: 4.53e-05 | 3124.8ms/step | 19.2 samples/s
Step 40800/50462 | Loss: 2.6538 | LR: 4.52e-05 | 3185.1ms/step | 18.8 samples/s
Step 40810/50462 | Loss: 2.6106 | LR: 4.51e-05 | 3118.0ms/step | 19.2 samples/s
Step 40820/50462 | Loss: 2.6861 | LR: 4.50e-05 | 3114.1ms/step | 19.3 samples/s
Step 40830/50462 | Loss: 2.6020 | LR: 4.49e-05 | 3119.0ms/step | 19.2 samples/s
Step 40840/50462 | Loss: 2.6264 | LR: 4.48e-05 | 3120.5ms/step | 19.2 samples/s
Step 40850/50462 | Loss: 2.6688 | LR: 4.47e-05 | 3123.3ms/step | 19.2 samples/s
Step 40860/50462 | Loss: 2.6123 | LR: 4.46e-05 | 3180.5ms/step | 18.9 samples/s
Step 40870/50462 | Loss: 2.6717 | LR: 4.45e-05 | 3117.5ms/step | 19.2 samples/s
Step 40880/50462 | Loss: 2.6772 | LR: 4.45e-05 | 3116.1ms/step | 19.3 samples/s
Step 40890/50462 | Loss: 2.6077 | LR: 4.44e-05 | 3120.8ms/step | 19.2 samples/s
Step 40900/50462 | Loss: 2.6344 | LR: 4.43e-05 | 3118.5ms/step | 19.2 samples/s
Step 40910/50462 | Loss: 2.6552 | LR: 4.42e-05 | 3122.3ms/step | 19.2 samples/s
Step 40920/50462 | Loss: 2.6611 | LR: 4.41e-05 | 3119.1ms/step | 19.2 samples/s
Step 40930/50462 | Loss: 2.6526 | LR: 4.40e-05 | 3121.3ms/step | 19.2 samples/s
Step 40940/50462 | Loss: 2.6698 | LR: 4.39e-05 | 3119.1ms/step | 19.2 samples/s
Step 40950/50462 | Loss: 2.6247 | LR: 4.38e-05 | 3120.9ms/step | 19.2 samples/s
Step 40960/50462 | Loss: 2.6547 | LR: 4.37e-05 | 3120.2ms/step | 19.2 samples/s
Step 40970/50462 | Loss: 2.6436 | LR: 4.36e-05 | 3114.7ms/step | 19.3 samples/s
Step 40980/50462 | Loss: 2.6450 | LR: 4.36e-05 | 3120.9ms/step | 19.2 samples/s
Step 40990/50462 | Loss: 2.6142 | LR: 4.35e-05 | 3126.2ms/step | 19.2 samples/s
Step 41000/50462 | Loss: 2.6599 | LR: 4.34e-05 | 3127.9ms/step | 19.2 samples/s
Validation at step 41000: Loss = 2.6562, Perplexity = 14.24
Saved checkpoint to checkpoints/run_20260616_011729/checkpoint-41000
Step 41010/50462 | Loss: 2.6629 | LR: 4.33e-05 | 5643.9ms/step | 10.6 samples/s
Step 41020/50462 | Loss: 2.6598 | LR: 4.32e-05 | 3118.6ms/step | 19.2 samples/s
Step 41030/50462 | Loss: 2.6610 | LR: 4.31e-05 | 3123.6ms/step | 19.2 samples/s
Step 41040/50462 | Loss: 2.6267 | LR: 4.30e-05 | 3112.6ms/step | 19.3 samples/s
Step 41050/50462 | Loss: 2.6599 | LR: 4.29e-05 | 3117.9ms/step | 19.2 samples/s
Step 41060/50462 | Loss: 2.6648 | LR: 4.28e-05 | 3117.9ms/step | 19.2 samples/s
Step 41070/50462 | Loss: 2.6405 | LR: 4.28e-05 | 3120.3ms/step | 19.2 samples/s
Step 41080/50462 | Loss: 2.6324 | LR: 4.27e-05 | 3120.3ms/step | 19.2 samples/s
Step 41090/50462 | Loss: 2.6138 | LR: 4.26e-05 | 3115.3ms/step | 19.3 samples/s
Step 41100/50462 | Loss: 2.6457 | LR: 4.25e-05 | 3119.2ms/step | 19.2 samples/s
Step 41110/50462 | Loss: 2.6486 | LR: 4.24e-05 | 3122.7ms/step | 19.2 samples/s
Step 41120/50462 | Loss: 2.6409 | LR: 4.23e-05 | 3176.5ms/step | 18.9 samples/s
Step 41130/50462 | Loss: 2.6642 | LR: 4.22e-05 | 3118.5ms/step | 19.2 samples/s
Step 41140/50462 | Loss: 2.6368 | LR: 4.21e-05 | 3114.5ms/step | 19.3 samples/s
Step 41150/50462 | Loss: 2.6482 | LR: 4.21e-05 | 3118.2ms/step | 19.2 samples/s
Step 41160/50462 | Loss: 2.6491 | LR: 4.20e-05 | 3119.2ms/step | 19.2 samples/s
Step 41170/50462 | Loss: 2.6558 | LR: 4.19e-05 | 3119.3ms/step | 19.2 samples/s
Step 41180/50462 | Loss: 2.6378 | LR: 4.18e-05 | 3187.8ms/step | 18.8 samples/s
Step 41190/50462 | Loss: 2.5967 | LR: 4.17e-05 | 3124.1ms/step | 19.2 samples/s
Step 41200/50462 | Loss: 2.6352 | LR: 4.16e-05 | 3121.4ms/step | 19.2 samples/s
Step 41210/50462 | Loss: 2.6102 | LR: 4.15e-05 | 3120.3ms/step | 19.2 samples/s
Step 41220/50462 | Loss: 2.6255 | LR: 4.14e-05 | 3128.2ms/step | 19.2 samples/s
Step 41230/50462 | Loss: 2.6496 | LR: 4.14e-05 | 3119.9ms/step | 19.2 samples/s
Step 41240/50462 | Loss: 2.6420 | LR: 4.13e-05 | 3126.4ms/step | 19.2 samples/s
Step 41250/50462 | Loss: 2.6535 | LR: 4.12e-05 | 3125.3ms/step | 19.2 samples/s
Validation at step 41250: Loss = 2.6555, Perplexity = 14.23
Step 41260/50462 | Loss: 2.6492 | LR: 4.11e-05 | 4778.9ms/step | 12.6 samples/s
Step 41270/50462 | Loss: 2.6237 | LR: 4.10e-05 | 3120.8ms/step | 19.2 samples/s
Step 41280/50462 | Loss: 2.6025 | LR: 4.09e-05 | 3117.6ms/step | 19.2 samples/s
Step 41290/50462 | Loss: 2.6645 | LR: 4.08e-05 | 3118.0ms/step | 19.2 samples/s
Step 41300/50462 | Loss: 2.6537 | LR: 4.07e-05 | 3121.5ms/step | 19.2 samples/s
Step 41310/50462 | Loss: 2.6175 | LR: 4.07e-05 | 3119.6ms/step | 19.2 samples/s
Step 41320/50462 | Loss: 2.6966 | LR: 4.06e-05 | 3124.1ms/step | 19.2 samples/s
Step 41330/50462 | Loss: 2.6337 | LR: 4.05e-05 | 3118.4ms/step | 19.2 samples/s
Step 41340/50462 | Loss: 2.6287 | LR: 4.04e-05 | 3114.9ms/step | 19.3 samples/s
Step 41350/50462 | Loss: 2.6473 | LR: 4.03e-05 | 3114.1ms/step | 19.3 samples/s
Step 41360/50462 | Loss: 2.6744 | LR: 4.02e-05 | 3117.6ms/step | 19.2 samples/s
Step 41370/50462 | Loss: 2.6453 | LR: 4.01e-05 | 3123.1ms/step | 19.2 samples/s
Step 41380/50462 | Loss: 2.6421 | LR: 4.01e-05 | 3127.2ms/step | 19.2 samples/s
Step 41390/50462 | Loss: 2.6142 | LR: 4.00e-05 | 3125.8ms/step | 19.2 samples/s
Step 41400/50462 | Loss: 2.6553 | LR: 3.99e-05 | 3116.1ms/step | 19.3 samples/s
Step 41410/50462 | Loss: 2.6578 | LR: 3.98e-05 | 3115.0ms/step | 19.3 samples/s
Step 41420/50462 | Loss: 2.6219 | LR: 3.97e-05 | 3114.0ms/step | 19.3 samples/s
Step 41430/50462 | Loss: 2.6545 | LR: 3.96e-05 | 3119.6ms/step | 19.2 samples/s
Step 41440/50462 | Loss: 2.6243 | LR: 3.95e-05 | 3178.2ms/step | 18.9 samples/s
Step 41450/50462 | Loss: 2.6441 | LR: 3.95e-05 | 3125.2ms/step | 19.2 samples/s
Step 41460/50462 | Loss: 2.6277 | LR: 3.94e-05 | 3112.1ms/step | 19.3 samples/s
Step 41470/50462 | Loss: 2.6052 | LR: 3.93e-05 | 3118.4ms/step | 19.2 samples/s
Step 41480/50462 | Loss: 2.6624 | LR: 3.92e-05 | 3120.4ms/step | 19.2 samples/s
Step 41490/50462 | Loss: 2.6160 | LR: 3.91e-05 | 3112.6ms/step | 19.3 samples/s
Step 41500/50462 | Loss: 2.6604 | LR: 3.90e-05 | 3179.6ms/step | 18.9 samples/s
Validation at step 41500: Loss = 2.6550, Perplexity = 14.22
Step 41510/50462 | Loss: 2.6293 | LR: 3.90e-05 | 4768.7ms/step | 12.6 samples/s
Step 41520/50462 | Loss: 2.6439 | LR: 3.89e-05 | 3119.3ms/step | 19.2 samples/s
Step 41530/50462 | Loss: 2.6562 | LR: 3.88e-05 | 3114.5ms/step | 19.3 samples/s
Step 41540/50462 | Loss: 2.6264 | LR: 3.87e-05 | 3121.4ms/step | 19.2 samples/s
Step 41550/50462 | Loss: 2.6536 | LR: 3.86e-05 | 3120.9ms/step | 19.2 samples/s
Step 41560/50462 | Loss: 2.6630 | LR: 3.85e-05 | 3117.2ms/step | 19.2 samples/s
Step 41570/50462 | Loss: 2.6510 | LR: 3.84e-05 | 3113.5ms/step | 19.3 samples/s
Step 41580/50462 | Loss: 2.6558 | LR: 3.84e-05 | 3125.3ms/step | 19.2 samples/s
Step 41590/50462 | Loss: 2.6374 | LR: 3.83e-05 | 3120.8ms/step | 19.2 samples/s
Step 41600/50462 | Loss: 2.6522 | LR: 3.82e-05 | 3118.4ms/step | 19.2 samples/s
Step 41610/50462 | Loss: 2.6356 | LR: 3.81e-05 | 3114.3ms/step | 19.3 samples/s
Step 41620/50462 | Loss: 2.6379 | LR: 3.80e-05 | 3119.7ms/step | 19.2 samples/s
Step 41630/50462 | Loss: 2.6227 | LR: 3.79e-05 | 3115.6ms/step | 19.3 samples/s
Step 41640/50462 | Loss: 2.6370 | LR: 3.79e-05 | 3115.8ms/step | 19.3 samples/s
Step 41650/50462 | Loss: 2.6254 | LR: 3.78e-05 | 3121.4ms/step | 19.2 samples/s
Step 41660/50462 | Loss: 2.6524 | LR: 3.77e-05 | 3119.4ms/step | 19.2 samples/s
Step 41670/50462 | Loss: 2.6448 | LR: 3.76e-05 | 3120.1ms/step | 19.2 samples/s
Step 41680/50462 | Loss: 2.6290 | LR: 3.75e-05 | 3118.5ms/step | 19.2 samples/s
Step 41690/50462 | Loss: 2.6264 | LR: 3.74e-05 | 3119.1ms/step | 19.2 samples/s
Step 41700/50462 | Loss: 2.6406 | LR: 3.74e-05 | 3117.3ms/step | 19.2 samples/s
Step 41710/50462 | Loss: 2.6284 | LR: 3.73e-05 | 3123.0ms/step | 19.2 samples/s
Step 41720/50462 | Loss: 2.6261 | LR: 3.72e-05 | 3118.6ms/step | 19.2 samples/s
Step 41730/50462 | Loss: 2.6323 | LR: 3.71e-05 | 3120.3ms/step | 19.2 samples/s
Step 41740/50462 | Loss: 2.6262 | LR: 3.70e-05 | 3122.9ms/step | 19.2 samples/s
Step 41750/50462 | Loss: 2.6356 | LR: 3.69e-05 | 3123.3ms/step | 19.2 samples/s
Validation at step 41750: Loss = 2.6546, Perplexity = 14.22
Step 41760/50462 | Loss: 2.6432 | LR: 3.69e-05 | 4834.5ms/step | 12.4 samples/s
Step 41770/50462 | Loss: 2.6514 | LR: 3.68e-05 | 3118.8ms/step | 19.2 samples/s
Step 41780/50462 | Loss: 2.6389 | LR: 3.67e-05 | 3120.6ms/step | 19.2 samples/s
Step 41790/50462 | Loss: 2.6246 | LR: 3.66e-05 | 3123.6ms/step | 19.2 samples/s
Step 41800/50462 | Loss: 2.6269 | LR: 3.65e-05 | 3123.2ms/step | 19.2 samples/s
Step 41810/50462 | Loss: 2.6593 | LR: 3.64e-05 | 3118.7ms/step | 19.2 samples/s
Step 41820/50462 | Loss: 2.5862 | LR: 3.64e-05 | 3238.4ms/step | 18.5 samples/s
Step 41830/50462 | Loss: 2.6374 | LR: 3.63e-05 | 3123.9ms/step | 19.2 samples/s
Step 41840/50462 | Loss: 2.6123 | LR: 3.62e-05 | 3123.1ms/step | 19.2 samples/s
Step 41850/50462 | Loss: 2.6531 | LR: 3.61e-05 | 3114.7ms/step | 19.3 samples/s
Step 41860/50462 | Loss: 2.6394 | LR: 3.60e-05 | 3122.3ms/step | 19.2 samples/s
Step 41870/50462 | Loss: 2.6352 | LR: 3.60e-05 | 3130.3ms/step | 19.2 samples/s
Step 41880/50462 | Loss: 2.6490 | LR: 3.59e-05 | 3123.9ms/step | 19.2 samples/s
Step 41890/50462 | Loss: 2.6374 | LR: 3.58e-05 | 3122.2ms/step | 19.2 samples/s
Step 41900/50462 | Loss: 2.6636 | LR: 3.57e-05 | 3121.2ms/step | 19.2 samples/s
Step 41910/50462 | Loss: 2.6514 | LR: 3.56e-05 | 3117.5ms/step | 19.2 samples/s
Step 41920/50462 | Loss: 2.6478 | LR: 3.55e-05 | 3114.6ms/step | 19.3 samples/s
Step 41930/50462 | Loss: 2.6263 | LR: 3.55e-05 | 3123.8ms/step | 19.2 samples/s
Step 41940/50462 | Loss: 2.6554 | LR: 3.54e-05 | 3121.5ms/step | 19.2 samples/s
Step 41950/50462 | Loss: 2.6249 | LR: 3.53e-05 | 3116.5ms/step | 19.3 samples/s
Step 41960/50462 | Loss: 2.6470 | LR: 3.52e-05 | 3120.5ms/step | 19.2 samples/s
Step 41970/50462 | Loss: 2.6401 | LR: 3.51e-05 | 3123.1ms/step | 19.2 samples/s
Step 41980/50462 | Loss: 2.6362 | LR: 3.51e-05 | 3127.3ms/step | 19.2 samples/s
Step 41990/50462 | Loss: 2.6687 | LR: 3.50e-05 | 3127.9ms/step | 19.2 samples/s
Step 42000/50462 | Loss: 2.6465 | LR: 3.49e-05 | 3129.7ms/step | 19.2 samples/s
Validation at step 42000: Loss = 2.6542, Perplexity = 14.21
Saved checkpoint to checkpoints/run_20260616_011729/checkpoint-42000
Step 42010/50462 | Loss: 2.6162 | LR: 3.48e-05 | 5655.4ms/step | 10.6 samples/s
Step 42020/50462 | Loss: 2.6599 | LR: 3.47e-05 | 3125.4ms/step | 19.2 samples/s
Step 42030/50462 | Loss: 2.6789 | LR: 3.47e-05 | 3124.6ms/step | 19.2 samples/s
Step 42040/50462 | Loss: 2.6435 | LR: 3.46e-05 | 3120.9ms/step | 19.2 samples/s
Step 42050/50462 | Loss: 2.6428 | LR: 3.45e-05 | 3124.6ms/step | 19.2 samples/s
Step 42060/50462 | Loss: 2.6787 | LR: 3.44e-05 | 3117.7ms/step | 19.2 samples/s
Step 42070/50462 | Loss: 2.6605 | LR: 3.43e-05 | 3122.4ms/step | 19.2 samples/s
Step 42080/50462 | Loss: 2.6295 | LR: 3.43e-05 | 3176.1ms/step | 18.9 samples/s
Step 42090/50462 | Loss: 2.6584 | LR: 3.42e-05 | 3113.5ms/step | 19.3 samples/s
Step 42100/50462 | Loss: 2.6443 | LR: 3.41e-05 | 3112.2ms/step | 19.3 samples/s
Step 42110/50462 | Loss: 2.6121 | LR: 3.40e-05 | 3122.5ms/step | 19.2 samples/s
Step 42120/50462 | Loss: 2.6468 | LR: 3.39e-05 | 3119.8ms/step | 19.2 samples/s
Step 42130/50462 | Loss: 2.6568 | LR: 3.39e-05 | 3116.8ms/step | 19.3 samples/s
Step 42140/50462 | Loss: 2.6541 | LR: 3.38e-05 | 3245.7ms/step | 18.5 samples/s
Step 42150/50462 | Loss: 2.6477 | LR: 3.37e-05 | 3126.5ms/step | 19.2 samples/s
Step 42160/50462 | Loss: 2.6088 | LR: 3.36e-05 | 3130.1ms/step | 19.2 samples/s
Step 42170/50462 | Loss: 2.6365 | LR: 3.35e-05 | 3122.9ms/step | 19.2 samples/s
Step 42180/50462 | Loss: 2.6434 | LR: 3.35e-05 | 3124.2ms/step | 19.2 samples/s
Step 42190/50462 | Loss: 2.6717 | LR: 3.34e-05 | 3127.9ms/step | 19.2 samples/s
Step 42200/50462 | Loss: 2.6297 | LR: 3.33e-05 | 3123.8ms/step | 19.2 samples/s
Step 42210/50462 | Loss: 2.6343 | LR: 3.32e-05 | 3124.7ms/step | 19.2 samples/s
Step 42220/50462 | Loss: 2.6607 | LR: 3.32e-05 | 3121.9ms/step | 19.2 samples/s
Step 42230/50462 | Loss: 2.6485 | LR: 3.31e-05 | 3124.7ms/step | 19.2 samples/s
Step 42240/50462 | Loss: 2.6027 | LR: 3.30e-05 | 3131.8ms/step | 19.2 samples/s
Step 42250/50462 | Loss: 2.6541 | LR: 3.29e-05 | 3128.0ms/step | 19.2 samples/s
Validation at step 42250: Loss = 2.6539, Perplexity = 14.21
Step 42260/50462 | Loss: 2.6599 | LR: 3.28e-05 | 4787.6ms/step | 12.5 samples/s
Step 42270/50462 | Loss: 2.6695 | LR: 3.28e-05 | 3118.7ms/step | 19.2 samples/s
Step 42280/50462 | Loss: 2.6403 | LR: 3.27e-05 | 3127.8ms/step | 19.2 samples/s
Step 42290/50462 | Loss: 2.6557 | LR: 3.26e-05 | 3120.9ms/step | 19.2 samples/s
Step 42300/50462 | Loss: 2.6404 | LR: 3.25e-05 | 3126.7ms/step | 19.2 samples/s
Step 42310/50462 | Loss: 2.6336 | LR: 3.24e-05 | 3125.6ms/step | 19.2 samples/s
Step 42320/50462 | Loss: 2.6158 | LR: 3.24e-05 | 3122.8ms/step | 19.2 samples/s
Step 42330/50462 | Loss: 2.6625 | LR: 3.23e-05 | 3121.2ms/step | 19.2 samples/s
Step 42340/50462 | Loss: 2.6699 | LR: 3.22e-05 | 3124.7ms/step | 19.2 samples/s
Step 42350/50462 | Loss: 2.6189 | LR: 3.21e-05 | 3127.4ms/step | 19.2 samples/s
Step 42360/50462 | Loss: 2.6298 | LR: 3.21e-05 | 3125.2ms/step | 19.2 samples/s
Step 42370/50462 | Loss: 2.6398 | LR: 3.20e-05 | 3122.6ms/step | 19.2 samples/s
Step 42380/50462 | Loss: 2.6261 | LR: 3.19e-05 | 3119.1ms/step | 19.2 samples/s
Step 42390/50462 | Loss: 2.6554 | LR: 3.18e-05 | 3129.4ms/step | 19.2 samples/s
Step 42400/50462 | Loss: 2.6625 | LR: 3.18e-05 | 3189.9ms/step | 18.8 samples/s
Step 42410/50462 | Loss: 2.6530 | LR: 3.17e-05 | 3116.4ms/step | 19.3 samples/s
Step 42420/50462 | Loss: 2.6481 | LR: 3.16e-05 | 3119.4ms/step | 19.2 samples/s
Step 42430/50462 | Loss: 2.6348 | LR: 3.15e-05 | 3123.3ms/step | 19.2 samples/s
Step 42440/50462 | Loss: 2.6361 | LR: 3.14e-05 | 3124.4ms/step | 19.2 samples/s
Step 42450/50462 | Loss: 2.6534 | LR: 3.14e-05 | 3129.6ms/step | 19.2 samples/s
Step 42460/50462 | Loss: 2.6343 | LR: 3.13e-05 | 3186.6ms/step | 18.8 samples/s
Step 42470/50462 | Loss: 2.6440 | LR: 3.12e-05 | 3113.2ms/step | 19.3 samples/s
Step 42480/50462 | Loss: 2.6028 | LR: 3.11e-05 | 3115.4ms/step | 19.3 samples/s
Step 42490/50462 | Loss: 2.6471 | LR: 3.11e-05 | 3111.4ms/step | 19.3 samples/s
Step 42500/50462 | Loss: 2.6586 | LR: 3.10e-05 | 3117.4ms/step | 19.2 samples/s
Validation at step 42500: Loss = 2.6536, Perplexity = 14.20
Step 42510/50462 | Loss: 2.6556 | LR: 3.09e-05 | 4778.5ms/step | 12.6 samples/s
Step 42520/50462 | Loss: 2.6379 | LR: 3.08e-05 | 3120.4ms/step | 19.2 samples/s
Step 42530/50462 | Loss: 2.6319 | LR: 3.08e-05 | 3119.7ms/step | 19.2 samples/s
Step 42540/50462 | Loss: 2.6368 | LR: 3.07e-05 | 3120.7ms/step | 19.2 samples/s
Step 42550/50462 | Loss: 2.6644 | LR: 3.06e-05 | 3118.8ms/step | 19.2 samples/s
Step 42560/50462 | Loss: 2.7230 | LR: 3.05e-05 | 3120.6ms/step | 19.2 samples/s
Step 42570/50462 | Loss: 2.6533 | LR: 3.05e-05 | 3121.3ms/step | 19.2 samples/s
Step 42580/50462 | Loss: 2.6452 | LR: 3.04e-05 | 3120.2ms/step | 19.2 samples/s
Step 42590/50462 | Loss: 2.6361 | LR: 3.03e-05 | 3113.7ms/step | 19.3 samples/s
Step 42600/50462 | Loss: 2.6867 | LR: 3.02e-05 | 3119.7ms/step | 19.2 samples/s
Step 42610/50462 | Loss: 2.6361 | LR: 3.02e-05 | 3121.0ms/step | 19.2 samples/s
Step 42620/50462 | Loss: 2.6302 | LR: 3.01e-05 | 3122.2ms/step | 19.2 samples/s
Step 42630/50462 | Loss: 2.6608 | LR: 3.00e-05 | 3126.2ms/step | 19.2 samples/s
Step 42640/50462 | Loss: 2.6884 | LR: 2.99e-05 | 3124.8ms/step | 19.2 samples/s
Step 42650/50462 | Loss: 2.6558 | LR: 2.99e-05 | 3124.5ms/step | 19.2 samples/s
Step 42660/50462 | Loss: 2.6590 | LR: 2.98e-05 | 3120.1ms/step | 19.2 samples/s
Step 42670/50462 | Loss: 2.6311 | LR: 2.97e-05 | 3119.6ms/step | 19.2 samples/s
Step 42680/50462 | Loss: 2.6294 | LR: 2.96e-05 | 3118.4ms/step | 19.2 samples/s
Step 42690/50462 | Loss: 2.6489 | LR: 2.96e-05 | 3118.6ms/step | 19.2 samples/s
Step 42700/50462 | Loss: 2.6556 | LR: 2.95e-05 | 3124.8ms/step | 19.2 samples/s
Step 42710/50462 | Loss: 2.6357 | LR: 2.94e-05 | 3124.0ms/step | 19.2 samples/s
Step 42720/50462 | Loss: 2.6362 | LR: 2.93e-05 | 3174.9ms/step | 18.9 samples/s
Step 42730/50462 | Loss: 2.6372 | LR: 2.93e-05 | 3117.7ms/step | 19.2 samples/s
Step 42740/50462 | Loss: 2.6438 | LR: 2.92e-05 | 3119.0ms/step | 19.2 samples/s
Step 42750/50462 | Loss: 2.6380 | LR: 2.91e-05 | 3121.9ms/step | 19.2 samples/s
Validation at step 42750: Loss = 2.6534, Perplexity = 14.20
Step 42760/50462 | Loss: 2.6560 | LR: 2.90e-05 | 4773.0ms/step | 12.6 samples/s
Step 42770/50462 | Loss: 2.6430 | LR: 2.90e-05 | 3122.2ms/step | 19.2 samples/s
Step 42780/50462 | Loss: 2.6439 | LR: 2.89e-05 | 3240.5ms/step | 18.5 samples/s
Step 42790/50462 | Loss: 2.6602 | LR: 2.88e-05 | 3117.8ms/step | 19.2 samples/s
Step 42800/50462 | Loss: 2.6607 | LR: 2.87e-05 | 3120.6ms/step | 19.2 samples/s
Step 42810/50462 | Loss: 2.6652 | LR: 2.87e-05 | 3121.9ms/step | 19.2 samples/s
Step 42820/50462 | Loss: 2.6339 | LR: 2.86e-05 | 3125.1ms/step | 19.2 samples/s
Step 42830/50462 | Loss: 2.6290 | LR: 2.85e-05 | 3128.6ms/step | 19.2 samples/s
Step 42840/50462 | Loss: 2.6553 | LR: 2.84e-05 | 3127.9ms/step | 19.2 samples/s
Step 42850/50462 | Loss: 2.6450 | LR: 2.84e-05 | 3122.1ms/step | 19.2 samples/s
Step 42860/50462 | Loss: 2.6747 | LR: 2.83e-05 | 3130.7ms/step | 19.2 samples/s
Step 42870/50462 | Loss: 2.6556 | LR: 2.82e-05 | 3126.1ms/step | 19.2 samples/s
Step 42880/50462 | Loss: 2.6519 | LR: 2.82e-05 | 3125.8ms/step | 19.2 samples/s
Step 42890/50462 | Loss: 2.6357 | LR: 2.81e-05 | 3126.6ms/step | 19.2 samples/s
Step 42900/50462 | Loss: 2.6451 | LR: 2.80e-05 | 3124.3ms/step | 19.2 samples/s
Step 42910/50462 | Loss: 2.6729 | LR: 2.79e-05 | 3128.7ms/step | 19.2 samples/s
Step 42920/50462 | Loss: 2.6322 | LR: 2.79e-05 | 3124.5ms/step | 19.2 samples/s
Step 42930/50462 | Loss: 2.6310 | LR: 2.78e-05 | 3127.0ms/step | 19.2 samples/s
Step 42940/50462 | Loss: 2.6238 | LR: 2.77e-05 | 3123.4ms/step | 19.2 samples/s
Step 42950/50462 | Loss: 2.6717 | LR: 2.76e-05 | 3118.1ms/step | 19.2 samples/s
Step 42960/50462 | Loss: 2.6554 | LR: 2.76e-05 | 3117.1ms/step | 19.2 samples/s
Step 42970/50462 | Loss: 2.6263 | LR: 2.75e-05 | 3119.8ms/step | 19.2 samples/s
Step 42980/50462 | Loss: 2.6405 | LR: 2.74e-05 | 3119.2ms/step | 19.2 samples/s
Step 42990/50462 | Loss: 2.6477 | LR: 2.74e-05 | 3121.5ms/step | 19.2 samples/s
Step 43000/50462 | Loss: 2.6276 | LR: 2.73e-05 | 3123.3ms/step | 19.2 samples/s
Validation at step 43000: Loss = 2.6532, Perplexity = 14.20
Saved checkpoint to checkpoints/run_20260616_011729/checkpoint-43000
Step 43010/50462 | Loss: 2.6672 | LR: 2.72e-05 | 5625.8ms/step | 10.7 samples/s
Step 43020/50462 | Loss: 2.6384 | LR: 2.71e-05 | 3110.7ms/step | 19.3 samples/s
Step 43030/50462 | Loss: 2.6455 | LR: 2.71e-05 | 3116.9ms/step | 19.2 samples/s
Step 43040/50462 | Loss: 2.6295 | LR: 2.70e-05 | 3174.7ms/step | 18.9 samples/s
Step 43050/50462 | Loss: 2.6515 | LR: 2.69e-05 | 3122.2ms/step | 19.2 samples/s
Step 43060/50462 | Loss: 2.6290 | LR: 2.69e-05 | 3126.8ms/step | 19.2 samples/s
Step 43070/50462 | Loss: 2.6458 | LR: 2.68e-05 | 3123.8ms/step | 19.2 samples/s
Step 43080/50462 | Loss: 2.6249 | LR: 2.67e-05 | 3115.0ms/step | 19.3 samples/s
Step 43090/50462 | Loss: 2.6258 | LR: 2.66e-05 | 3119.0ms/step | 19.2 samples/s
Step 43100/50462 | Loss: 2.6190 | LR: 2.66e-05 | 3246.7ms/step | 18.5 samples/s
Step 43110/50462 | Loss: 2.6276 | LR: 2.65e-05 | 3119.3ms/step | 19.2 samples/s
Step 43120/50462 | Loss: 2.6418 | LR: 2.64e-05 | 3120.8ms/step | 19.2 samples/s
Step 43130/50462 | Loss: 2.6540 | LR: 2.64e-05 | 3118.8ms/step | 19.2 samples/s
Step 43140/50462 | Loss: 2.6232 | LR: 2.63e-05 | 3121.0ms/step | 19.2 samples/s
Step 43150/50462 | Loss: 2.6256 | LR: 2.62e-05 | 3122.7ms/step | 19.2 samples/s
Step 43160/50462 | Loss: 2.6627 | LR: 2.61e-05 | 3127.5ms/step | 19.2 samples/s
Step 43170/50462 | Loss: 2.6555 | LR: 2.61e-05 | 3119.3ms/step | 19.2 samples/s
Step 43180/50462 | Loss: 2.6102 | LR: 2.60e-05 | 3120.8ms/step | 19.2 samples/s
Step 43190/50462 | Loss: 2.6634 | LR: 2.59e-05 | 3125.8ms/step | 19.2 samples/s
Step 43200/50462 | Loss: 2.6426 | LR: 2.59e-05 | 3120.1ms/step | 19.2 samples/s
Step 43210/50462 | Loss: 2.6519 | LR: 2.58e-05 | 3123.0ms/step | 19.2 samples/s
Step 43220/50462 | Loss: 2.6552 | LR: 2.57e-05 | 3124.4ms/step | 19.2 samples/s
Step 43230/50462 | Loss: 2.6408 | LR: 2.57e-05 | 3124.8ms/step | 19.2 samples/s
Step 43240/50462 | Loss: 2.6663 | LR: 2.56e-05 | 3122.6ms/step | 19.2 samples/s
Step 43250/50462 | Loss: 2.6169 | LR: 2.55e-05 | 3122.2ms/step | 19.2 samples/s
Validation at step 43250: Loss = 2.6530, Perplexity = 14.20
Step 43260/50462 | Loss: 2.6163 | LR: 2.54e-05 | 4773.7ms/step | 12.6 samples/s
Step 43270/50462 | Loss: 2.6259 | LR: 2.54e-05 | 3122.7ms/step | 19.2 samples/s
Step 43280/50462 | Loss: 2.6470 | LR: 2.53e-05 | 3123.1ms/step | 19.2 samples/s
Step 43290/50462 | Loss: 2.6260 | LR: 2.52e-05 | 3124.9ms/step | 19.2 samples/s
Step 43300/50462 | Loss: 2.6699 | LR: 2.52e-05 | 3118.3ms/step | 19.2 samples/s
Step 43310/50462 | Loss: 2.6348 | LR: 2.51e-05 | 3120.2ms/step | 19.2 samples/s
Step 43320/50462 | Loss: 2.6403 | LR: 2.50e-05 | 3119.1ms/step | 19.2 samples/s
Step 43330/50462 | Loss: 2.6286 | LR: 2.50e-05 | 3116.3ms/step | 19.3 samples/s
Step 43340/50462 | Loss: 2.6372 | LR: 2.49e-05 | 3118.0ms/step | 19.2 samples/s
Step 43350/50462 | Loss: 2.6228 | LR: 2.48e-05 | 3120.2ms/step | 19.2 samples/s
Step 43360/50462 | Loss: 2.6562 | LR: 2.48e-05 | 3177.8ms/step | 18.9 samples/s
Step 43370/50462 | Loss: 2.6469 | LR: 2.47e-05 | 3124.8ms/step | 19.2 samples/s
Step 43380/50462 | Loss: 2.6286 | LR: 2.46e-05 | 3125.2ms/step | 19.2 samples/s
Step 43390/50462 | Loss: 2.6575 | LR: 2.46e-05 | 3118.7ms/step | 19.2 samples/s
Step 43400/50462 | Loss: 2.6369 | LR: 2.45e-05 | 3130.0ms/step | 19.2 samples/s
Step 43410/50462 | Loss: 2.6576 | LR: 2.44e-05 | 3123.4ms/step | 19.2 samples/s
Step 43420/50462 | Loss: 2.6389 | LR: 2.44e-05 | 3186.9ms/step | 18.8 samples/s
Step 43430/50462 | Loss: 2.6412 | LR: 2.43e-05 | 3124.0ms/step | 19.2 samples/s
Step 43440/50462 | Loss: 2.6401 | LR: 2.42e-05 | 3128.7ms/step | 19.2 samples/s
Step 43450/50462 | Loss: 2.6636 | LR: 2.41e-05 | 3125.3ms/step | 19.2 samples/s
Step 43460/50462 | Loss: 2.6508 | LR: 2.41e-05 | 3123.0ms/step | 19.2 samples/s
Step 43470/50462 | Loss: 2.6629 | LR: 2.40e-05 | 3129.5ms/step | 19.2 samples/s
Step 43480/50462 | Loss: 2.6653 | LR: 2.39e-05 | 3131.8ms/step | 19.2 samples/s
Step 43490/50462 | Loss: 2.6339 | LR: 2.39e-05 | 3123.5ms/step | 19.2 samples/s
Step 43500/50462 | Loss: 2.6511 | LR: 2.38e-05 | 3126.3ms/step | 19.2 samples/s
Validation at step 43500: Loss = 2.6529, Perplexity = 14.20
Step 43510/50462 | Loss: 2.6538 | LR: 2.37e-05 | 4782.0ms/step | 12.5 samples/s
Step 43520/50462 | Loss: 2.6199 | LR: 2.37e-05 | 3123.8ms/step | 19.2 samples/s
Step 43530/50462 | Loss: 2.6418 | LR: 2.36e-05 | 3125.9ms/step | 19.2 samples/s
Step 43540/50462 | Loss: 2.6346 | LR: 2.35e-05 | 3118.6ms/step | 19.2 samples/s
Step 43550/50462 | Loss: 2.6438 | LR: 2.35e-05 | 3120.2ms/step | 19.2 samples/s
Step 43560/50462 | Loss: 2.6226 | LR: 2.34e-05 | 3112.5ms/step | 19.3 samples/s
Step 43570/50462 | Loss: 2.6433 | LR: 2.33e-05 | 3115.9ms/step | 19.3 samples/s
Step 43580/50462 | Loss: 2.6707 | LR: 2.33e-05 | 3121.3ms/step | 19.2 samples/s
Step 43590/50462 | Loss: 2.6320 | LR: 2.32e-05 | 3127.0ms/step | 19.2 samples/s
Step 43600/50462 | Loss: 2.6650 | LR: 2.31e-05 | 3117.4ms/step | 19.2 samples/s
Step 43610/50462 | Loss: 2.6471 | LR: 2.31e-05 | 3111.6ms/step | 19.3 samples/s
Step 43620/50462 | Loss: 2.6289 | LR: 2.30e-05 | 3121.6ms/step | 19.2 samples/s
Step 43630/50462 | Loss: 2.6680 | LR: 2.29e-05 | 3118.1ms/step | 19.2 samples/s
Step 43640/50462 | Loss: 2.6310 | LR: 2.29e-05 | 3119.2ms/step | 19.2 samples/s
Step 43650/50462 | Loss: 2.6486 | LR: 2.28e-05 | 3120.6ms/step | 19.2 samples/s
Step 43660/50462 | Loss: 2.6621 | LR: 2.27e-05 | 3124.0ms/step | 19.2 samples/s
Step 43670/50462 | Loss: 2.6513 | LR: 2.27e-05 | 3128.9ms/step | 19.2 samples/s
Step 43680/50462 | Loss: 2.6565 | LR: 2.26e-05 | 3187.4ms/step | 18.8 samples/s
Step 43690/50462 | Loss: 2.6491 | LR: 2.25e-05 | 3123.4ms/step | 19.2 samples/s
Step 43700/50462 | Loss: 2.6649 | LR: 2.25e-05 | 3124.6ms/step | 19.2 samples/s
Step 43710/50462 | Loss: 2.6682 | LR: 2.24e-05 | 3128.2ms/step | 19.2 samples/s
Step 43720/50462 | Loss: 2.6275 | LR: 2.24e-05 | 3126.5ms/step | 19.2 samples/s
Step 43730/50462 | Loss: 2.6170 | LR: 2.23e-05 | 3128.5ms/step | 19.2 samples/s
Step 43740/50462 | Loss: 2.6403 | LR: 2.22e-05 | 3230.5ms/step | 18.6 samples/s
Step 43750/50462 | Loss: 2.6419 | LR: 2.22e-05 | 3127.3ms/step | 19.2 samples/s
Validation at step 43750: Loss = 2.6528, Perplexity = 14.19
Step 43760/50462 | Loss: 2.6455 | LR: 2.21e-05 | 4796.0ms/step | 12.5 samples/s
Step 43770/50462 | Loss: 2.6334 | LR: 2.20e-05 | 3126.1ms/step | 19.2 samples/s
Step 43780/50462 | Loss: 2.6376 | LR: 2.20e-05 | 3126.3ms/step | 19.2 samples/s
Step 43790/50462 | Loss: 2.6444 | LR: 2.19e-05 | 3127.2ms/step | 19.2 samples/s
Step 43800/50462 | Loss: 2.6608 | LR: 2.18e-05 | 3128.1ms/step | 19.2 samples/s
Step 43810/50462 | Loss: 2.6069 | LR: 2.18e-05 | 3120.6ms/step | 19.2 samples/s
Step 43820/50462 | Loss: 2.6109 | LR: 2.17e-05 | 3125.0ms/step | 19.2 samples/s
Step 43830/50462 | Loss: 2.6521 | LR: 2.16e-05 | 3131.5ms/step | 19.2 samples/s
Step 43840/50462 | Loss: 2.6712 | LR: 2.16e-05 | 3118.5ms/step | 19.2 samples/s
Step 43850/50462 | Loss: 2.6629 | LR: 2.15e-05 | 3118.3ms/step | 19.2 samples/s
Step 43860/50462 | Loss: 2.6467 | LR: 2.14e-05 | 3121.2ms/step | 19.2 samples/s
Step 43870/50462 | Loss: 2.6314 | LR: 2.14e-05 | 3123.1ms/step | 19.2 samples/s
Step 43880/50462 | Loss: 2.6549 | LR: 2.13e-05 | 3130.1ms/step | 19.2 samples/s
Step 43890/50462 | Loss: 2.6471 | LR: 2.13e-05 | 3124.9ms/step | 19.2 samples/s
Step 43900/50462 | Loss: 2.7011 | LR: 2.12e-05 | 3116.7ms/step | 19.3 samples/s
Step 43910/50462 | Loss: 2.6651 | LR: 2.11e-05 | 3118.4ms/step | 19.2 samples/s
Step 43920/50462 | Loss: 2.6406 | LR: 2.11e-05 | 3127.1ms/step | 19.2 samples/s
Step 43930/50462 | Loss: 2.6687 | LR: 2.10e-05 | 3124.3ms/step | 19.2 samples/s
Step 43940/50462 | Loss: 2.6416 | LR: 2.09e-05 | 3129.3ms/step | 19.2 samples/s
Step 43950/50462 | Loss: 2.6566 | LR: 2.09e-05 | 3126.6ms/step | 19.2 samples/s
Step 43960/50462 | Loss: 2.6367 | LR: 2.08e-05 | 3128.0ms/step | 19.2 samples/s
Step 43970/50462 | Loss: 2.6326 | LR: 2.07e-05 | 3125.1ms/step | 19.2 samples/s
Step 43980/50462 | Loss: 2.6247 | LR: 2.07e-05 | 3129.2ms/step | 19.2 samples/s
Step 43990/50462 | Loss: 2.6598 | LR: 2.06e-05 | 3122.9ms/step | 19.2 samples/s
Step 44000/50462 | Loss: 2.6644 | LR: 2.06e-05 | 3177.1ms/step | 18.9 samples/s
Validation at step 44000: Loss = 2.6527, Perplexity = 14.19
Saved checkpoint to checkpoints/run_20260616_011729/checkpoint-44000
Step 44010/50462 | Loss: 2.6519 | LR: 2.05e-05 | 5664.1ms/step | 10.6 samples/s
Step 44020/50462 | Loss: 2.6434 | LR: 2.04e-05 | 3116.2ms/step | 19.3 samples/s
Step 44030/50462 | Loss: 2.6382 | LR: 2.04e-05 | 3117.4ms/step | 19.2 samples/s
Step 44040/50462 | Loss: 2.6327 | LR: 2.03e-05 | 3124.7ms/step | 19.2 samples/s
Step 44050/50462 | Loss: 2.6662 | LR: 2.02e-05 | 3119.5ms/step | 19.2 samples/s
Step 44060/50462 | Loss: 2.6640 | LR: 2.02e-05 | 3181.3ms/step | 18.9 samples/s
Step 44070/50462 | Loss: 2.6664 | LR: 2.01e-05 | 3180.0ms/step | 18.9 samples/s
Step 44080/50462 | Loss: 2.6473 | LR: 2.01e-05 | 3119.2ms/step | 19.2 samples/s
Step 44090/50462 | Loss: 2.6141 | LR: 2.00e-05 | 3117.4ms/step | 19.2 samples/s
Step 44100/50462 | Loss: 2.6346 | LR: 1.99e-05 | 3120.1ms/step | 19.2 samples/s
Step 44110/50462 | Loss: 2.6359 | LR: 1.99e-05 | 3123.8ms/step | 19.2 samples/s
Step 44120/50462 | Loss: 2.6536 | LR: 1.98e-05 | 3109.6ms/step | 19.3 samples/s
Step 44130/50462 | Loss: 2.6589 | LR: 1.97e-05 | 3118.6ms/step | 19.2 samples/s
Step 44140/50462 | Loss: 2.6311 | LR: 1.97e-05 | 3119.4ms/step | 19.2 samples/s
Step 44150/50462 | Loss: 2.6248 | LR: 1.96e-05 | 3120.3ms/step | 19.2 samples/s
Step 44160/50462 | Loss: 2.6519 | LR: 1.96e-05 | 3120.6ms/step | 19.2 samples/s
Step 44170/50462 | Loss: 2.6592 | LR: 1.95e-05 | 3118.6ms/step | 19.2 samples/s
Step 44180/50462 | Loss: 2.6265 | LR: 1.94e-05 | 3117.8ms/step | 19.2 samples/s
Step 44190/50462 | Loss: 2.6306 | LR: 1.94e-05 | 3119.9ms/step | 19.2 samples/s
Step 44200/50462 | Loss: 2.6529 | LR: 1.93e-05 | 3125.0ms/step | 19.2 samples/s
Step 44210/50462 | Loss: 2.6424 | LR: 1.93e-05 | 3114.0ms/step | 19.3 samples/s
Step 44220/50462 | Loss: 2.6446 | LR: 1.92e-05 | 3115.3ms/step | 19.3 samples/s
Step 44230/50462 | Loss: 2.6575 | LR: 1.91e-05 | 3111.2ms/step | 19.3 samples/s
Step 44240/50462 | Loss: 2.6467 | LR: 1.91e-05 | 3114.8ms/step | 19.3 samples/s
Step 44250/50462 | Loss: 2.6583 | LR: 1.90e-05 | 3120.8ms/step | 19.2 samples/s
Validation at step 44250: Loss = 2.6527, Perplexity = 14.19
Step 44260/50462 | Loss: 2.6532 | LR: 1.90e-05 | 4770.5ms/step | 12.6 samples/s
Step 44270/50462 | Loss: 2.6262 | LR: 1.89e-05 | 3119.7ms/step | 19.2 samples/s
Step 44280/50462 | Loss: 2.6449 | LR: 1.88e-05 | 3124.3ms/step | 19.2 samples/s
Step 44290/50462 | Loss: 2.6450 | LR: 1.88e-05 | 3123.9ms/step | 19.2 samples/s
Step 44300/50462 | Loss: 2.5956 | LR: 1.87e-05 | 3125.3ms/step | 19.2 samples/s
Step 44310/50462 | Loss: 2.6221 | LR: 1.87e-05 | 3122.9ms/step | 19.2 samples/s
Step 44320/50462 | Loss: 2.6575 | LR: 1.86e-05 | 3179.7ms/step | 18.9 samples/s
Step 44330/50462 | Loss: 2.6486 | LR: 1.85e-05 | 3134.6ms/step | 19.1 samples/s
Step 44340/50462 | Loss: 2.6429 | LR: 1.85e-05 | 3126.6ms/step | 19.2 samples/s
Step 44350/50462 | Loss: 2.6492 | LR: 1.84e-05 | 3127.4ms/step | 19.2 samples/s
Step 44360/50462 | Loss: 2.6212 | LR: 1.84e-05 | 3125.8ms/step | 19.2 samples/s
Step 44370/50462 | Loss: 2.6468 | LR: 1.83e-05 | 3133.7ms/step | 19.1 samples/s
Step 44380/50462 | Loss: 2.6375 | LR: 1.82e-05 | 3189.3ms/step | 18.8 samples/s
Step 44390/50462 | Loss: 2.6574 | LR: 1.82e-05 | 3179.5ms/step | 18.9 samples/s
Step 44400/50462 | Loss: 2.6267 | LR: 1.81e-05 | 3126.4ms/step | 19.2 samples/s
Step 44410/50462 | Loss: 2.6343 | LR: 1.81e-05 | 3129.2ms/step | 19.2 samples/s
Step 44420/50462 | Loss: 2.6435 | LR: 1.80e-05 | 3130.9ms/step | 19.2 samples/s
Step 44430/50462 | Loss: 2.6519 | LR: 1.79e-05 | 3129.9ms/step | 19.2 samples/s
Step 44440/50462 | Loss: 2.6403 | LR: 1.79e-05 | 3129.3ms/step | 19.2 samples/s
Step 44450/50462 | Loss: 2.6274 | LR: 1.78e-05 | 3124.2ms/step | 19.2 samples/s
Step 44460/50462 | Loss: 2.6445 | LR: 1.78e-05 | 3120.1ms/step | 19.2 samples/s
Step 44470/50462 | Loss: 2.6566 | LR: 1.77e-05 | 3119.6ms/step | 19.2 samples/s
Step 44480/50462 | Loss: 2.6734 | LR: 1.77e-05 | 3121.4ms/step | 19.2 samples/s
Step 44490/50462 | Loss: 2.6683 | LR: 1.76e-05 | 3122.9ms/step | 19.2 samples/s
Step 44500/50462 | Loss: 2.6299 | LR: 1.75e-05 | 3120.7ms/step | 19.2 samples/s
Validation at step 44500: Loss = 2.6526, Perplexity = 14.19
Step 44510/50462 | Loss: 2.6174 | LR: 1.75e-05 | 4782.3ms/step | 12.5 samples/s
Step 44520/50462 | Loss: 2.6185 | LR: 1.74e-05 | 3126.6ms/step | 19.2 samples/s
Step 44530/50462 | Loss: 2.6422 | LR: 1.74e-05 | 3126.6ms/step | 19.2 samples/s
Step 44540/50462 | Loss: 2.6312 | LR: 1.73e-05 | 3125.4ms/step | 19.2 samples/s
Step 44550/50462 | Loss: 2.6484 | LR: 1.72e-05 | 3115.5ms/step | 19.3 samples/s
Step 44560/50462 | Loss: 2.6500 | LR: 1.72e-05 | 3119.9ms/step | 19.2 samples/s
Step 44570/50462 | Loss: 2.6395 | LR: 1.71e-05 | 3123.4ms/step | 19.2 samples/s
Step 44580/50462 | Loss: 2.6176 | LR: 1.71e-05 | 3127.1ms/step | 19.2 samples/s
Step 44590/50462 | Loss: 2.6349 | LR: 1.70e-05 | 3125.2ms/step | 19.2 samples/s
Step 44600/50462 | Loss: 2.6571 | LR: 1.70e-05 | 3120.4ms/step | 19.2 samples/s
Step 44610/50462 | Loss: 2.6418 | LR: 1.69e-05 | 3124.6ms/step | 19.2 samples/s
Step 44620/50462 | Loss: 2.6567 | LR: 1.68e-05 | 3127.6ms/step | 19.2 samples/s
Step 44630/50462 | Loss: 2.6412 | LR: 1.68e-05 | 3121.5ms/step | 19.2 samples/s
Step 44640/50462 | Loss: 2.6331 | LR: 1.67e-05 | 3112.8ms/step | 19.3 samples/s
Step 44650/50462 | Loss: 2.6245 | LR: 1.67e-05 | 3174.1ms/step | 18.9 samples/s
Step 44660/50462 | Loss: 2.6578 | LR: 1.66e-05 | 3123.7ms/step | 19.2 samples/s
Step 44670/50462 | Loss: 2.6667 | LR: 1.66e-05 | 3128.1ms/step | 19.2 samples/s
Step 44680/50462 | Loss: 2.6539 | LR: 1.65e-05 | 3118.5ms/step | 19.2 samples/s
Step 44690/50462 | Loss: 2.6505 | LR: 1.64e-05 | 3119.2ms/step | 19.2 samples/s
Step 44700/50462 | Loss: 2.6579 | LR: 1.64e-05 | 3179.8ms/step | 18.9 samples/s
Step 44710/50462 | Loss: 2.6407 | LR: 1.63e-05 | 3179.5ms/step | 18.9 samples/s
Step 44720/50462 | Loss: 2.6393 | LR: 1.63e-05 | 3126.8ms/step | 19.2 samples/s
Step 44730/50462 | Loss: 2.6284 | LR: 1.62e-05 | 3121.6ms/step | 19.2 samples/s
Step 44740/50462 | Loss: 2.6266 | LR: 1.62e-05 | 3123.8ms/step | 19.2 samples/s
Step 44750/50462 | Loss: 2.6495 | LR: 1.61e-05 | 3126.3ms/step | 19.2 samples/s
Validation at step 44750: Loss = 2.6525, Perplexity = 14.19
Step 44760/50462 | Loss: 2.6655 | LR: 1.61e-05 | 4794.8ms/step | 12.5 samples/s
Step 44770/50462 | Loss: 2.6174 | LR: 1.60e-05 | 3128.0ms/step | 19.2 samples/s
Step 44780/50462 | Loss: 2.6460 | LR: 1.59e-05 | 3126.7ms/step | 19.2 samples/s
Step 44790/50462 | Loss: 2.6323 | LR: 1.59e-05 | 3118.3ms/step | 19.2 samples/s
Step 44800/50462 | Loss: 2.6329 | LR: 1.58e-05 | 3120.8ms/step | 19.2 samples/s
Step 44810/50462 | Loss: 2.6607 | LR: 1.58e-05 | 3119.4ms/step | 19.2 samples/s
Step 44820/50462 | Loss: 2.6507 | LR: 1.57e-05 | 3119.4ms/step | 19.2 samples/s
Step 44830/50462 | Loss: 2.6731 | LR: 1.57e-05 | 3126.0ms/step | 19.2 samples/s
Step 44840/50462 | Loss: 2.6265 | LR: 1.56e-05 | 3120.3ms/step | 19.2 samples/s
Step 44850/50462 | Loss: 2.6352 | LR: 1.56e-05 | 3125.9ms/step | 19.2 samples/s
Step 44860/50462 | Loss: 2.6321 | LR: 1.55e-05 | 3126.2ms/step | 19.2 samples/s
Step 44870/50462 | Loss: 2.6226 | LR: 1.54e-05 | 3119.1ms/step | 19.2 samples/s
Step 44880/50462 | Loss: 2.6433 | LR: 1.54e-05 | 3127.0ms/step | 19.2 samples/s
Step 44890/50462 | Loss: 2.6033 | LR: 1.53e-05 | 3129.0ms/step | 19.2 samples/s
Step 44900/50462 | Loss: 2.6031 | LR: 1.53e-05 | 3122.8ms/step | 19.2 samples/s
Step 44910/50462 | Loss: 2.6266 | LR: 1.52e-05 | 3118.0ms/step | 19.2 samples/s
Step 44920/50462 | Loss: 2.6308 | LR: 1.52e-05 | 3123.0ms/step | 19.2 samples/s
Step 44930/50462 | Loss: 2.6344 | LR: 1.51e-05 | 3121.6ms/step | 19.2 samples/s
Step 44940/50462 | Loss: 2.6354 | LR: 1.51e-05 | 3123.6ms/step | 19.2 samples/s
Step 44950/50462 | Loss: 2.6002 | LR: 1.50e-05 | 3121.2ms/step | 19.2 samples/s
Step 44960/50462 | Loss: 2.6741 | LR: 1.50e-05 | 3178.2ms/step | 18.9 samples/s
Step 44970/50462 | Loss: 2.6249 | LR: 1.49e-05 | 3125.8ms/step | 19.2 samples/s
Step 44980/50462 | Loss: 2.6391 | LR: 1.49e-05 | 3130.0ms/step | 19.2 samples/s
Step 44990/50462 | Loss: 2.6632 | LR: 1.48e-05 | 3125.3ms/step | 19.2 samples/s
Step 45000/50462 | Loss: 2.6583 | LR: 1.47e-05 | 3119.1ms/step | 19.2 samples/s
Validation at step 45000: Loss = 2.6525, Perplexity = 14.19
Saved checkpoint to checkpoints/run_20260616_011729/checkpoint-45000
Step 45010/50462 | Loss: 2.6467 | LR: 1.47e-05 | 5661.0ms/step | 10.6 samples/s
Step 45020/50462 | Loss: 2.6336 | LR: 1.46e-05 | 3183.0ms/step | 18.9 samples/s
Step 45030/50462 | Loss: 2.6388 | LR: 1.46e-05 | 3189.5ms/step | 18.8 samples/s
Step 45040/50462 | Loss: 2.6527 | LR: 1.45e-05 | 3121.3ms/step | 19.2 samples/s
Step 45050/50462 | Loss: 2.6243 | LR: 1.45e-05 | 3121.3ms/step | 19.2 samples/s
Step 45060/50462 | Loss: 2.6482 | LR: 1.44e-05 | 3117.9ms/step | 19.2 samples/s
Step 45070/50462 | Loss: 2.6426 | LR: 1.44e-05 | 3125.8ms/step | 19.2 samples/s
Step 45080/50462 | Loss: 2.6334 | LR: 1.43e-05 | 3124.2ms/step | 19.2 samples/s
Step 45090/50462 | Loss: 2.6339 | LR: 1.43e-05 | 3125.9ms/step | 19.2 samples/s
Step 45100/50462 | Loss: 2.6682 | LR: 1.42e-05 | 3126.4ms/step | 19.2 samples/s
Step 45110/50462 | Loss: 2.6683 | LR: 1.42e-05 | 3125.6ms/step | 19.2 samples/s
Step 45120/50462 | Loss: 2.6508 | LR: 1.41e-05 | 3129.1ms/step | 19.2 samples/s
Step 45130/50462 | Loss: 2.6573 | LR: 1.41e-05 | 3125.2ms/step | 19.2 samples/s
Step 45140/50462 | Loss: 2.6273 | LR: 1.40e-05 | 3130.0ms/step | 19.2 samples/s
Step 45150/50462 | Loss: 2.6521 | LR: 1.40e-05 | 3121.6ms/step | 19.2 samples/s
Step 45160/50462 | Loss: 2.6577 | LR: 1.39e-05 | 3125.7ms/step | 19.2 samples/s
Step 45170/50462 | Loss: 2.6237 | LR: 1.39e-05 | 3117.7ms/step | 19.2 samples/s
Step 45180/50462 | Loss: 2.6498 | LR: 1.38e-05 | 3118.4ms/step | 19.2 samples/s
Step 45190/50462 | Loss: 2.6356 | LR: 1.37e-05 | 3114.7ms/step | 19.3 samples/s
Step 45200/50462 | Loss: 2.6613 | LR: 1.37e-05 | 3122.4ms/step | 19.2 samples/s
Step 45210/50462 | Loss: 2.6433 | LR: 1.36e-05 | 3124.6ms/step | 19.2 samples/s
Step 45220/50462 | Loss: 2.6317 | LR: 1.36e-05 | 3126.8ms/step | 19.2 samples/s
Step 45230/50462 | Loss: 2.6234 | LR: 1.35e-05 | 3123.8ms/step | 19.2 samples/s
Step 45240/50462 | Loss: 2.6574 | LR: 1.35e-05 | 3123.8ms/step | 19.2 samples/s
Step 45250/50462 | Loss: 2.6383 | LR: 1.34e-05 | 3125.4ms/step | 19.2 samples/s
Validation at step 45250: Loss = 2.6524, Perplexity = 14.19
Step 45260/50462 | Loss: 2.6510 | LR: 1.34e-05 | 4783.7ms/step | 12.5 samples/s
Step 45270/50462 | Loss: 2.6575 | LR: 1.33e-05 | 3123.7ms/step | 19.2 samples/s
Step 45280/50462 | Loss: 2.6163 | LR: 1.33e-05 | 3174.6ms/step | 18.9 samples/s
Step 45290/50462 | Loss: 2.6414 | LR: 1.32e-05 | 3129.1ms/step | 19.2 samples/s
Step 45300/50462 | Loss: 2.6563 | LR: 1.32e-05 | 3122.0ms/step | 19.2 samples/s
Step 45310/50462 | Loss: 2.6317 | LR: 1.31e-05 | 3125.6ms/step | 19.2 samples/s
Step 45320/50462 | Loss: 2.6364 | LR: 1.31e-05 | 3126.2ms/step | 19.2 samples/s
Step 45330/50462 | Loss: 2.6513 | LR: 1.30e-05 | 3121.0ms/step | 19.2 samples/s
Step 45340/50462 | Loss: 2.6415 | LR: 1.30e-05 | 3180.1ms/step | 18.9 samples/s
Step 45350/50462 | Loss: 2.6393 | LR: 1.29e-05 | 3187.9ms/step | 18.8 samples/s
Step 45360/50462 | Loss: 2.6647 | LR: 1.29e-05 | 3125.1ms/step | 19.2 samples/s
Step 45370/50462 | Loss: 2.6362 | LR: 1.28e-05 | 3124.1ms/step | 19.2 samples/s
Step 45380/50462 | Loss: 2.6293 | LR: 1.28e-05 | 3118.5ms/step | 19.2 samples/s
Step 45390/50462 | Loss: 2.6398 | LR: 1.27e-05 | 3113.1ms/step | 19.3 samples/s
Step 45400/50462 | Loss: 2.6205 | LR: 1.27e-05 | 3126.1ms/step | 19.2 samples/s
Step 45410/50462 | Loss: 2.6635 | LR: 1.26e-05 | 3122.6ms/step | 19.2 samples/s
Step 45420/50462 | Loss: 2.6353 | LR: 1.26e-05 | 3118.0ms/step | 19.2 samples/s
Step 45430/50462 | Loss: 2.6098 | LR: 1.25e-05 | 3127.1ms/step | 19.2 samples/s
Step 45440/50462 | Loss: 2.6527 | LR: 1.25e-05 | 3125.1ms/step | 19.2 samples/s
Step 45450/50462 | Loss: 2.6444 | LR: 1.24e-05 | 3118.1ms/step | 19.2 samples/s
Step 45460/50462 | Loss: 2.6295 | LR: 1.24e-05 | 3116.5ms/step | 19.3 samples/s
Step 45470/50462 | Loss: 2.6431 | LR: 1.23e-05 | 3116.4ms/step | 19.3 samples/s
Step 45480/50462 | Loss: 2.6320 | LR: 1.23e-05 | 3116.8ms/step | 19.3 samples/s
Step 45490/50462 | Loss: 2.6530 | LR: 1.22e-05 | 3127.2ms/step | 19.2 samples/s
Step 45500/50462 | Loss: 2.6543 | LR: 1.22e-05 | 3123.4ms/step | 19.2 samples/s
Validation at step 45500: Loss = 2.6524, Perplexity = 14.19
Step 45510/50462 | Loss: 2.6288 | LR: 1.21e-05 | 4773.0ms/step | 12.6 samples/s
Step 45520/50462 | Loss: 2.6721 | LR: 1.21e-05 | 3120.2ms/step | 19.2 samples/s
Step 45530/50462 | Loss: 2.6207 | LR: 1.20e-05 | 3116.2ms/step | 19.3 samples/s
Step 45540/50462 | Loss: 2.6691 | LR: 1.20e-05 | 3119.9ms/step | 19.2 samples/s
Step 45550/50462 | Loss: 2.6300 | LR: 1.19e-05 | 3123.3ms/step | 19.2 samples/s
Step 45560/50462 | Loss: 2.6519 | LR: 1.19e-05 | 3124.5ms/step | 19.2 samples/s
Step 45570/50462 | Loss: 2.6564 | LR: 1.19e-05 | 3127.5ms/step | 19.2 samples/s
Step 45580/50462 | Loss: 2.6345 | LR: 1.18e-05 | 3124.3ms/step | 19.2 samples/s
Step 45590/50462 | Loss: 2.6208 | LR: 1.18e-05 | 3123.9ms/step | 19.2 samples/s
Step 45600/50462 | Loss: 2.6310 | LR: 1.17e-05 | 3180.4ms/step | 18.9 samples/s
Step 45610/50462 | Loss: 2.6121 | LR: 1.17e-05 | 3130.3ms/step | 19.2 samples/s
Step 45620/50462 | Loss: 2.6426 | LR: 1.16e-05 | 3113.3ms/step | 19.3 samples/s
Step 45630/50462 | Loss: 2.6592 | LR: 1.16e-05 | 3120.6ms/step | 19.2 samples/s
Step 45640/50462 | Loss: 2.6359 | LR: 1.15e-05 | 3122.6ms/step | 19.2 samples/s
Step 45650/50462 | Loss: 2.6619 | LR: 1.15e-05 | 3125.6ms/step | 19.2 samples/s
Step 45660/50462 | Loss: 2.6447 | LR: 1.14e-05 | 3180.6ms/step | 18.9 samples/s
Step 45670/50462 | Loss: 2.6137 | LR: 1.14e-05 | 3175.0ms/step | 18.9 samples/s
Step 45680/50462 | Loss: 2.6313 | LR: 1.13e-05 | 3116.0ms/step | 19.3 samples/s
Step 45690/50462 | Loss: 2.6519 | LR: 1.13e-05 | 3116.4ms/step | 19.3 samples/s
Step 45700/50462 | Loss: 2.6495 | LR: 1.12e-05 | 3119.1ms/step | 19.2 samples/s
Step 45710/50462 | Loss: 2.6471 | LR: 1.12e-05 | 3124.7ms/step | 19.2 samples/s
Step 45720/50462 | Loss: 2.6389 | LR: 1.11e-05 | 3124.9ms/step | 19.2 samples/s
Step 45730/50462 | Loss: 2.6195 | LR: 1.11e-05 | 3126.3ms/step | 19.2 samples/s
Step 45740/50462 | Loss: 2.6357 | LR: 1.10e-05 | 3112.3ms/step | 19.3 samples/s
Step 45750/50462 | Loss: 2.6541 | LR: 1.10e-05 | 3116.4ms/step | 19.3 samples/s
Validation at step 45750: Loss = 2.6524, Perplexity = 14.19
Step 45760/50462 | Loss: 2.6054 | LR: 1.10e-05 | 4785.8ms/step | 12.5 samples/s
Step 45770/50462 | Loss: 2.6391 | LR: 1.09e-05 | 3122.6ms/step | 19.2 samples/s
Step 45780/50462 | Loss: 2.6321 | LR: 1.09e-05 | 3120.0ms/step | 19.2 samples/s
Step 45790/50462 | Loss: 2.6691 | LR: 1.08e-05 | 3117.1ms/step | 19.2 samples/s
Step 45800/50462 | Loss: 2.5859 | LR: 1.08e-05 | 3120.3ms/step | 19.2 samples/s
Step 45810/50462 | Loss: 2.6427 | LR: 1.07e-05 | 3118.1ms/step | 19.2 samples/s
Step 45820/50462 | Loss: 2.6408 | LR: 1.07e-05 | 3120.2ms/step | 19.2 samples/s
Step 45830/50462 | Loss: 2.6485 | LR: 1.06e-05 | 3122.3ms/step | 19.2 samples/s
Step 45840/50462 | Loss: 2.6618 | LR: 1.06e-05 | 3126.4ms/step | 19.2 samples/s
Step 45850/50462 | Loss: 2.6432 | LR: 1.05e-05 | 3120.4ms/step | 19.2 samples/s
Step 45860/50462 | Loss: 2.6647 | LR: 1.05e-05 | 3122.2ms/step | 19.2 samples/s
Step 45870/50462 | Loss: 2.6492 | LR: 1.05e-05 | 3123.9ms/step | 19.2 samples/s
Step 45880/50462 | Loss: 2.6528 | LR: 1.04e-05 | 3121.2ms/step | 19.2 samples/s
Step 45890/50462 | Loss: 2.6163 | LR: 1.04e-05 | 3121.0ms/step | 19.2 samples/s
Step 45900/50462 | Loss: 2.6391 | LR: 1.03e-05 | 3121.9ms/step | 19.2 samples/s
Step 45910/50462 | Loss: 2.6680 | LR: 1.03e-05 | 3116.5ms/step | 19.3 samples/s
Step 45920/50462 | Loss: 2.6210 | LR: 1.02e-05 | 3115.5ms/step | 19.3 samples/s
Step 45930/50462 | Loss: 2.6337 | LR: 1.02e-05 | 3177.4ms/step | 18.9 samples/s
Step 45940/50462 | Loss: 2.6526 | LR: 1.01e-05 | 3116.7ms/step | 19.3 samples/s
Step 45950/50462 | Loss: 2.6380 | LR: 1.01e-05 | 3118.8ms/step | 19.2 samples/s
Step 45960/50462 | Loss: 2.6236 | LR: 1.01e-05 | 3127.6ms/step | 19.2 samples/s
Step 45970/50462 | Loss: 2.6295 | LR: 1.00e-05 | 3117.6ms/step | 19.2 samples/s
Step 45980/50462 | Loss: 2.6749 | LR: 9.96e-06 | 3114.3ms/step | 19.3 samples/s
Step 45990/50462 | Loss: 2.6264 | LR: 9.92e-06 | 3230.6ms/step | 18.6 samples/s
Step 46000/50462 | Loss: 2.6434 | LR: 9.87e-06 | 3116.9ms/step | 19.2 samples/s
Validation at step 46000: Loss = 2.6524, Perplexity = 14.19
Saved checkpoint to checkpoints/run_20260616_011729/checkpoint-46000
Step 46010/50462 | Loss: 2.6455 | LR: 9.83e-06 | 5650.3ms/step | 10.6 samples/s
Step 46020/50462 | Loss: 2.6332 | LR: 9.79e-06 | 3125.7ms/step | 19.2 samples/s
Step 46030/50462 | Loss: 2.6555 | LR: 9.74e-06 | 3122.9ms/step | 19.2 samples/s
Step 46040/50462 | Loss: 2.6554 | LR: 9.70e-06 | 3123.1ms/step | 19.2 samples/s
Step 46050/50462 | Loss: 2.6350 | LR: 9.65e-06 | 3118.4ms/step | 19.2 samples/s
Step 46060/50462 | Loss: 2.6311 | LR: 9.61e-06 | 3117.5ms/step | 19.2 samples/s
Step 46070/50462 | Loss: 2.6547 | LR: 9.57e-06 | 3126.8ms/step | 19.2 samples/s
Step 46080/50462 | Loss: 2.6444 | LR: 9.52e-06 | 3120.1ms/step | 19.2 samples/s
Step 46090/50462 | Loss: 2.6151 | LR: 9.48e-06 | 3120.2ms/step | 19.2 samples/s
Step 46100/50462 | Loss: 2.6452 | LR: 9.44e-06 | 3127.6ms/step | 19.2 samples/s
Step 46110/50462 | Loss: 2.6400 | LR: 9.40e-06 | 3126.3ms/step | 19.2 samples/s
Step 46120/50462 | Loss: 2.6346 | LR: 9.35e-06 | 3119.9ms/step | 19.2 samples/s
Step 46130/50462 | Loss: 2.6356 | LR: 9.31e-06 | 3124.2ms/step | 19.2 samples/s
Step 46140/50462 | Loss: 2.6333 | LR: 9.27e-06 | 3128.1ms/step | 19.2 samples/s
Step 46150/50462 | Loss: 2.6333 | LR: 9.22e-06 | 3126.6ms/step | 19.2 samples/s
Step 46160/50462 | Loss: 2.6514 | LR: 9.18e-06 | 3122.4ms/step | 19.2 samples/s
Step 46170/50462 | Loss: 2.6526 | LR: 9.14e-06 | 3123.7ms/step | 19.2 samples/s
Step 46180/50462 | Loss: 2.6226 | LR: 9.10e-06 | 3121.0ms/step | 19.2 samples/s
Step 46190/50462 | Loss: 2.6235 | LR: 9.06e-06 | 3124.7ms/step | 19.2 samples/s
Step 46200/50462 | Loss: 2.6295 | LR: 9.01e-06 | 3126.3ms/step | 19.2 samples/s
Step 46210/50462 | Loss: 2.6495 | LR: 8.97e-06 | 3117.4ms/step | 19.2 samples/s
Step 46220/50462 | Loss: 2.6492 | LR: 8.93e-06 | 3120.4ms/step | 19.2 samples/s
Step 46230/50462 | Loss: 2.6560 | LR: 8.89e-06 | 3126.0ms/step | 19.2 samples/s
Step 46240/50462 | Loss: 2.6639 | LR: 8.85e-06 | 3178.0ms/step | 18.9 samples/s
Step 46250/50462 | Loss: 2.6592 | LR: 8.80e-06 | 3120.4ms/step | 19.2 samples/s
Validation at step 46250: Loss = 2.6524, Perplexity = 14.19
Step 46260/50462 | Loss: 2.6621 | LR: 8.76e-06 | 4778.3ms/step | 12.6 samples/s
Step 46270/50462 | Loss: 2.6485 | LR: 8.72e-06 | 3122.0ms/step | 19.2 samples/s
Step 46280/50462 | Loss: 2.6795 | LR: 8.68e-06 | 3121.4ms/step | 19.2 samples/s
Step 46290/50462 | Loss: 2.6684 | LR: 8.64e-06 | 3114.9ms/step | 19.3 samples/s
Step 46300/50462 | Loss: 2.6358 | LR: 8.60e-06 | 3118.1ms/step | 19.2 samples/s
Step 46310/50462 | Loss: 2.6428 | LR: 8.56e-06 | 3234.6ms/step | 18.5 samples/s
Step 46320/50462 | Loss: 2.6702 | LR: 8.52e-06 | 3126.7ms/step | 19.2 samples/s
Step 46330/50462 | Loss: 2.6513 | LR: 8.47e-06 | 3118.8ms/step | 19.2 samples/s
Step 46340/50462 | Loss: 2.6489 | LR: 8.43e-06 | 3122.9ms/step | 19.2 samples/s
Step 46350/50462 | Loss: 2.6704 | LR: 8.39e-06 | 3122.8ms/step | 19.2 samples/s
Step 46360/50462 | Loss: 2.6356 | LR: 8.35e-06 | 3123.0ms/step | 19.2 samples/s
Step 46370/50462 | Loss: 2.6178 | LR: 8.31e-06 | 3123.5ms/step | 19.2 samples/s
Step 46380/50462 | Loss: 2.6456 | LR: 8.27e-06 | 3111.1ms/step | 19.3 samples/s
Step 46390/50462 | Loss: 2.6341 | LR: 8.23e-06 | 3120.5ms/step | 19.2 samples/s
Step 46400/50462 | Loss: 2.6258 | LR: 8.19e-06 | 3123.6ms/step | 19.2 samples/s
Step 46410/50462 | Loss: 2.6369 | LR: 8.15e-06 | 3121.9ms/step | 19.2 samples/s
Step 46420/50462 | Loss: 2.6336 | LR: 8.11e-06 | 3123.3ms/step | 19.2 samples/s
Step 46430/50462 | Loss: 2.6595 | LR: 8.07e-06 | 3118.5ms/step | 19.2 samples/s
Step 46440/50462 | Loss: 2.6394 | LR: 8.03e-06 | 3116.4ms/step | 19.3 samples/s
Step 46450/50462 | Loss: 2.6675 | LR: 7.99e-06 | 3126.8ms/step | 19.2 samples/s
Step 46460/50462 | Loss: 2.6435 | LR: 7.95e-06 | 3125.2ms/step | 19.2 samples/s
Step 46470/50462 | Loss: 2.6352 | LR: 7.91e-06 | 3128.2ms/step | 19.2 samples/s
Step 46480/50462 | Loss: 2.6317 | LR: 7.87e-06 | 3125.6ms/step | 19.2 samples/s
Step 46490/50462 | Loss: 2.6266 | LR: 7.83e-06 | 3122.3ms/step | 19.2 samples/s
Step 46500/50462 | Loss: 2.6463 | LR: 7.80e-06 | 3129.4ms/step | 19.2 samples/s
Validation at step 46500: Loss = 2.6524, Perplexity = 14.19
Step 46510/50462 | Loss: 2.6369 | LR: 7.76e-06 | 4797.7ms/step | 12.5 samples/s
Step 46520/50462 | Loss: 2.6611 | LR: 7.72e-06 | 3126.9ms/step | 19.2 samples/s
Step 46530/50462 | Loss: 2.6400 | LR: 7.68e-06 | 3127.7ms/step | 19.2 samples/s
Step 46540/50462 | Loss: 2.6575 | LR: 7.64e-06 | 3121.9ms/step | 19.2 samples/s
Step 46550/50462 | Loss: 2.6174 | LR: 7.60e-06 | 3118.2ms/step | 19.2 samples/s
Step 46560/50462 | Loss: 2.6320 | LR: 7.56e-06 | 3183.0ms/step | 18.8 samples/s
Step 46570/50462 | Loss: 2.6443 | LR: 7.52e-06 | 3117.4ms/step | 19.2 samples/s
Step 46580/50462 | Loss: 2.6345 | LR: 7.49e-06 | 3118.1ms/step | 19.2 samples/s
Step 46590/50462 | Loss: 2.6274 | LR: 7.45e-06 | 3120.2ms/step | 19.2 samples/s
Step 46600/50462 | Loss: 2.6452 | LR: 7.41e-06 | 3126.6ms/step | 19.2 samples/s
Step 46610/50462 | Loss: 2.6431 | LR: 7.37e-06 | 3123.4ms/step | 19.2 samples/s
Step 46620/50462 | Loss: 2.6227 | LR: 7.33e-06 | 3121.2ms/step | 19.2 samples/s
Step 46630/50462 | Loss: 2.6396 | LR: 7.29e-06 | 3238.6ms/step | 18.5 samples/s
Step 46640/50462 | Loss: 2.6502 | LR: 7.26e-06 | 3116.0ms/step | 19.3 samples/s
Step 46650/50462 | Loss: 2.6440 | LR: 7.22e-06 | 3123.1ms/step | 19.2 samples/s
Step 46660/50462 | Loss: 2.6330 | LR: 7.18e-06 | 3120.6ms/step | 19.2 samples/s
Step 46670/50462 | Loss: 2.6502 | LR: 7.14e-06 | 3120.4ms/step | 19.2 samples/s
Step 46680/50462 | Loss: 2.6561 | LR: 7.11e-06 | 3126.3ms/step | 19.2 samples/s
Step 46690/50462 | Loss: 2.6729 | LR: 7.07e-06 | 3121.2ms/step | 19.2 samples/s
Step 46700/50462 | Loss: 2.6335 | LR: 7.03e-06 | 3122.3ms/step | 19.2 samples/s
Step 46710/50462 | Loss: 2.6667 | LR: 6.99e-06 | 3125.8ms/step | 19.2 samples/s
Step 46720/50462 | Loss: 2.6506 | LR: 6.96e-06 | 3118.0ms/step | 19.2 samples/s
Step 46730/50462 | Loss: 2.6304 | LR: 6.92e-06 | 3125.7ms/step | 19.2 samples/s
Step 46740/50462 | Loss: 2.6540 | LR: 6.88e-06 | 3128.3ms/step | 19.2 samples/s
Step 46750/50462 | Loss: 2.6371 | LR: 6.85e-06 | 3116.0ms/step | 19.3 samples/s
Validation at step 46750: Loss = 2.6524, Perplexity = 14.19
Step 46760/50462 | Loss: 2.6531 | LR: 6.81e-06 | 4785.0ms/step | 12.5 samples/s
Step 46770/50462 | Loss: 2.6433 | LR: 6.77e-06 | 3113.2ms/step | 19.3 samples/s
Step 46780/50462 | Loss: 2.6712 | LR: 6.74e-06 | 3116.6ms/step | 19.3 samples/s
Step 46790/50462 | Loss: 2.6242 | LR: 6.70e-06 | 3126.1ms/step | 19.2 samples/s
Step 46800/50462 | Loss: 2.6625 | LR: 6.66e-06 | 3122.0ms/step | 19.2 samples/s
Step 46810/50462 | Loss: 2.6563 | LR: 6.63e-06 | 3112.3ms/step | 19.3 samples/s
Step 46820/50462 | Loss: 2.6226 | LR: 6.59e-06 | 3116.1ms/step | 19.3 samples/s
Step 46830/50462 | Loss: 2.6232 | LR: 6.56e-06 | 3118.1ms/step | 19.2 samples/s
Step 46840/50462 | Loss: 2.6475 | LR: 6.52e-06 | 3122.7ms/step | 19.2 samples/s
Step 46850/50462 | Loss: 2.6402 | LR: 6.48e-06 | 3119.4ms/step | 19.2 samples/s
Step 46860/50462 | Loss: 2.6408 | LR: 6.45e-06 | 3115.9ms/step | 19.3 samples/s
Step 46870/50462 | Loss: 2.6002 | LR: 6.41e-06 | 3110.4ms/step | 19.3 samples/s
Step 46880/50462 | Loss: 2.6365 | LR: 6.38e-06 | 3173.7ms/step | 18.9 samples/s
Step 46890/50462 | Loss: 2.6423 | LR: 6.34e-06 | 3120.6ms/step | 19.2 samples/s
Step 46900/50462 | Loss: 2.6334 | LR: 6.31e-06 | 3115.6ms/step | 19.3 samples/s
Step 46910/50462 | Loss: 2.6248 | LR: 6.27e-06 | 3125.8ms/step | 19.2 samples/s
Step 46920/50462 | Loss: 2.6648 | LR: 6.24e-06 | 3121.7ms/step | 19.2 samples/s
Step 46930/50462 | Loss: 2.6395 | LR: 6.20e-06 | 3118.9ms/step | 19.2 samples/s
Step 46940/50462 | Loss: 2.6443 | LR: 6.17e-06 | 3123.9ms/step | 19.2 samples/s
Step 46950/50462 | Loss: 2.6299 | LR: 6.13e-06 | 3229.3ms/step | 18.6 samples/s
Step 46960/50462 | Loss: 2.6550 | LR: 6.10e-06 | 3118.3ms/step | 19.2 samples/s
Step 46970/50462 | Loss: 2.6536 | LR: 6.06e-06 | 3115.8ms/step | 19.3 samples/s
Step 46980/50462 | Loss: 2.6497 | LR: 6.03e-06 | 3111.4ms/step | 19.3 samples/s
Step 46990/50462 | Loss: 2.6708 | LR: 5.99e-06 | 3118.8ms/step | 19.2 samples/s
Step 47000/50462 | Loss: 2.6534 | LR: 5.96e-06 | 3115.1ms/step | 19.3 samples/s
Validation at step 47000: Loss = 2.6523, Perplexity = 14.19
Saved checkpoint to checkpoints/run_20260616_011729/checkpoint-47000
Step 47010/50462 | Loss: 2.6653 | LR: 5.93e-06 | 5637.8ms/step | 10.6 samples/s
Step 47020/50462 | Loss: 2.6405 | LR: 5.89e-06 | 3119.9ms/step | 19.2 samples/s
Step 47030/50462 | Loss: 2.6616 | LR: 5.86e-06 | 3123.1ms/step | 19.2 samples/s
Step 47040/50462 | Loss: 2.6074 | LR: 5.82e-06 | 3125.0ms/step | 19.2 samples/s
Step 47050/50462 | Loss: 2.6474 | LR: 5.79e-06 | 3130.5ms/step | 19.2 samples/s
Step 47060/50462 | Loss: 2.6437 | LR: 5.76e-06 | 3116.5ms/step | 19.3 samples/s
Step 47070/50462 | Loss: 2.6326 | LR: 5.72e-06 | 3123.9ms/step | 19.2 samples/s
Step 47080/50462 | Loss: 2.6425 | LR: 5.69e-06 | 3118.8ms/step | 19.2 samples/s
Step 47090/50462 | Loss: 2.6362 | LR: 5.65e-06 | 3120.2ms/step | 19.2 samples/s
Step 47100/50462 | Loss: 2.6426 | LR: 5.62e-06 | 3116.4ms/step | 19.3 samples/s
Step 47110/50462 | Loss: 2.6460 | LR: 5.59e-06 | 3123.6ms/step | 19.2 samples/s
Step 47120/50462 | Loss: 2.5977 | LR: 5.56e-06 | 3120.8ms/step | 19.2 samples/s
Step 47130/50462 | Loss: 2.6408 | LR: 5.52e-06 | 3122.3ms/step | 19.2 samples/s
Step 47140/50462 | Loss: 2.6546 | LR: 5.49e-06 | 3117.1ms/step | 19.2 samples/s
Step 47150/50462 | Loss: 2.6147 | LR: 5.46e-06 | 3121.3ms/step | 19.2 samples/s
Step 47160/50462 | Loss: 2.6537 | LR: 5.42e-06 | 3123.7ms/step | 19.2 samples/s
Step 47170/50462 | Loss: 2.6370 | LR: 5.39e-06 | 3123.0ms/step | 19.2 samples/s
Step 47180/50462 | Loss: 2.6466 | LR: 5.36e-06 | 3120.7ms/step | 19.2 samples/s
Step 47190/50462 | Loss: 2.6405 | LR: 5.33e-06 | 3119.6ms/step | 19.2 samples/s
Step 47200/50462 | Loss: 2.6355 | LR: 5.29e-06 | 3183.6ms/step | 18.8 samples/s
Step 47210/50462 | Loss: 2.6123 | LR: 5.26e-06 | 3130.9ms/step | 19.2 samples/s
Step 47220/50462 | Loss: 2.6538 | LR: 5.23e-06 | 3128.4ms/step | 19.2 samples/s
Step 47230/50462 | Loss: 2.6482 | LR: 5.20e-06 | 3122.0ms/step | 19.2 samples/s
Step 47240/50462 | Loss: 2.6194 | LR: 5.16e-06 | 3114.7ms/step | 19.3 samples/s
Step 47250/50462 | Loss: 2.6486 | LR: 5.13e-06 | 3115.5ms/step | 19.3 samples/s
Validation at step 47250: Loss = 2.6523, Perplexity = 14.19
Step 47260/50462 | Loss: 2.6418 | LR: 5.10e-06 | 4774.2ms/step | 12.6 samples/s
Step 47270/50462 | Loss: 2.6571 | LR: 5.07e-06 | 3181.2ms/step | 18.9 samples/s
Step 47280/50462 | Loss: 2.6312 | LR: 5.04e-06 | 3176.0ms/step | 18.9 samples/s
Step 47290/50462 | Loss: 2.6577 | LR: 5.01e-06 | 3113.8ms/step | 19.3 samples/s
Step 47300/50462 | Loss: 2.6564 | LR: 4.97e-06 | 3120.7ms/step | 19.2 samples/s
Step 47310/50462 | Loss: 2.6197 | LR: 4.94e-06 | 3118.1ms/step | 19.2 samples/s
Step 47320/50462 | Loss: 2.6349 | LR: 4.91e-06 | 3114.5ms/step | 19.3 samples/s
Step 47330/50462 | Loss: 2.6379 | LR: 4.88e-06 | 3125.2ms/step | 19.2 samples/s
Step 47340/50462 | Loss: 2.6043 | LR: 4.85e-06 | 3125.8ms/step | 19.2 samples/s
Step 47350/50462 | Loss: 2.6512 | LR: 4.82e-06 | 3125.1ms/step | 19.2 samples/s
Step 47360/50462 | Loss: 2.6176 | LR: 4.79e-06 | 3119.3ms/step | 19.2 samples/s
Step 47370/50462 | Loss: 2.6601 | LR: 4.76e-06 | 3116.5ms/step | 19.3 samples/s
Step 47380/50462 | Loss: 2.6517 | LR: 4.73e-06 | 3121.2ms/step | 19.2 samples/s
Step 47390/50462 | Loss: 2.6373 | LR: 4.70e-06 | 3119.0ms/step | 19.2 samples/s
Step 47400/50462 | Loss: 2.6358 | LR: 4.67e-06 | 3119.4ms/step | 19.2 samples/s
Step 47410/50462 | Loss: 2.6520 | LR: 4.64e-06 | 3119.0ms/step | 19.2 samples/s
Step 47420/50462 | Loss: 2.5991 | LR: 4.61e-06 | 3111.4ms/step | 19.3 samples/s
Step 47430/50462 | Loss: 2.6304 | LR: 4.58e-06 | 3113.5ms/step | 19.3 samples/s
Step 47440/50462 | Loss: 2.6393 | LR: 4.55e-06 | 3122.6ms/step | 19.2 samples/s
Step 47450/50462 | Loss: 2.6111 | LR: 4.52e-06 | 3110.5ms/step | 19.3 samples/s
Step 47460/50462 | Loss: 2.6550 | LR: 4.49e-06 | 3113.9ms/step | 19.3 samples/s
Step 47470/50462 | Loss: 2.6559 | LR: 4.46e-06 | 3106.3ms/step | 19.3 samples/s
Step 47480/50462 | Loss: 2.6510 | LR: 4.43e-06 | 3110.5ms/step | 19.3 samples/s
Step 47490/50462 | Loss: 2.6459 | LR: 4.40e-06 | 3113.3ms/step | 19.3 samples/s
Step 47500/50462 | Loss: 2.6387 | LR: 4.37e-06 | 3120.4ms/step | 19.2 samples/s
Validation at step 47500: Loss = 2.6523, Perplexity = 14.19
Step 47510/50462 | Loss: 2.6080 | LR: 4.34e-06 | 4777.6ms/step | 12.6 samples/s
Step 47520/50462 | Loss: 2.6558 | LR: 4.31e-06 | 3172.9ms/step | 18.9 samples/s
Step 47530/50462 | Loss: 2.6521 | LR: 4.28e-06 | 3118.2ms/step | 19.2 samples/s
Step 47540/50462 | Loss: 2.6229 | LR: 4.25e-06 | 3114.4ms/step | 19.3 samples/s
Step 47550/50462 | Loss: 2.6509 | LR: 4.22e-06 | 3122.1ms/step | 19.2 samples/s
Step 47560/50462 | Loss: 2.6323 | LR: 4.19e-06 | 3111.9ms/step | 19.3 samples/s
Step 47570/50462 | Loss: 2.6466 | LR: 4.16e-06 | 3120.6ms/step | 19.2 samples/s
Step 47580/50462 | Loss: 2.6808 | LR: 4.13e-06 | 3121.8ms/step | 19.2 samples/s
Step 47590/50462 | Loss: 2.6511 | LR: 4.11e-06 | 3172.5ms/step | 18.9 samples/s
Step 47600/50462 | Loss: 2.6418 | LR: 4.08e-06 | 3174.7ms/step | 18.9 samples/s
Step 47610/50462 | Loss: 2.6203 | LR: 4.05e-06 | 3121.4ms/step | 19.2 samples/s
Step 47620/50462 | Loss: 2.6321 | LR: 4.02e-06 | 3111.2ms/step | 19.3 samples/s
Step 47630/50462 | Loss: 2.6639 | LR: 3.99e-06 | 3113.2ms/step | 19.3 samples/s
Step 47640/50462 | Loss: 2.6168 | LR: 3.97e-06 | 3122.2ms/step | 19.2 samples/s
Step 47650/50462 | Loss: 2.6395 | LR: 3.94e-06 | 3122.0ms/step | 19.2 samples/s
Step 47660/50462 | Loss: 2.6388 | LR: 3.91e-06 | 3121.2ms/step | 19.2 samples/s
Step 47670/50462 | Loss: 2.6278 | LR: 3.88e-06 | 3119.7ms/step | 19.2 samples/s
Step 47680/50462 | Loss: 2.6469 | LR: 3.85e-06 | 3124.7ms/step | 19.2 samples/s
Step 47690/50462 | Loss: 2.6388 | LR: 3.83e-06 | 3120.5ms/step | 19.2 samples/s
Step 47700/50462 | Loss: 2.6508 | LR: 3.80e-06 | 3123.4ms/step | 19.2 samples/s
Step 47710/50462 | Loss: 2.6168 | LR: 3.77e-06 | 3121.9ms/step | 19.2 samples/s
Step 47720/50462 | Loss: 2.6423 | LR: 3.74e-06 | 3115.3ms/step | 19.3 samples/s
Step 47730/50462 | Loss: 2.6421 | LR: 3.72e-06 | 3120.2ms/step | 19.2 samples/s
Step 47740/50462 | Loss: 2.6275 | LR: 3.69e-06 | 3121.5ms/step | 19.2 samples/s
Step 47750/50462 | Loss: 2.6524 | LR: 3.66e-06 | 3122.5ms/step | 19.2 samples/s
Validation at step 47750: Loss = 2.6523, Perplexity = 14.19
Step 47760/50462 | Loss: 2.6145 | LR: 3.64e-06 | 4775.0ms/step | 12.6 samples/s
Step 47770/50462 | Loss: 2.6788 | LR: 3.61e-06 | 3120.4ms/step | 19.2 samples/s
Step 47780/50462 | Loss: 2.6365 | LR: 3.58e-06 | 3119.7ms/step | 19.2 samples/s
Step 47790/50462 | Loss: 2.6256 | LR: 3.56e-06 | 3119.3ms/step | 19.2 samples/s
Step 47800/50462 | Loss: 2.6191 | LR: 3.53e-06 | 3117.9ms/step | 19.2 samples/s
Step 47810/50462 | Loss: 2.6502 | LR: 3.50e-06 | 3120.1ms/step | 19.2 samples/s
Step 47820/50462 | Loss: 2.6470 | LR: 3.48e-06 | 3118.6ms/step | 19.2 samples/s
Step 47830/50462 | Loss: 2.6396 | LR: 3.45e-06 | 3119.1ms/step | 19.2 samples/s
Step 47840/50462 | Loss: 2.6600 | LR: 3.42e-06 | 3171.4ms/step | 18.9 samples/s
Step 47850/50462 | Loss: 2.6521 | LR: 3.40e-06 | 3110.1ms/step | 19.3 samples/s
Step 47860/50462 | Loss: 2.6568 | LR: 3.37e-06 | 3111.3ms/step | 19.3 samples/s
Step 47870/50462 | Loss: 2.6479 | LR: 3.35e-06 | 3110.2ms/step | 19.3 samples/s
Step 47880/50462 | Loss: 2.6040 | LR: 3.32e-06 | 3112.8ms/step | 19.3 samples/s
Step 47890/50462 | Loss: 2.6606 | LR: 3.30e-06 | 3123.7ms/step | 19.2 samples/s
Step 47900/50462 | Loss: 2.6517 | LR: 3.27e-06 | 3114.0ms/step | 19.3 samples/s
Step 47910/50462 | Loss: 2.6486 | LR: 3.24e-06 | 3170.9ms/step | 18.9 samples/s
Step 47920/50462 | Loss: 2.6228 | LR: 3.22e-06 | 3177.2ms/step | 18.9 samples/s
Step 47930/50462 | Loss: 2.6619 | LR: 3.19e-06 | 3120.7ms/step | 19.2 samples/s
Step 47940/50462 | Loss: 2.6349 | LR: 3.17e-06 | 3116.4ms/step | 19.3 samples/s
Step 47950/50462 | Loss: 2.6430 | LR: 3.14e-06 | 3122.1ms/step | 19.2 samples/s
Step 47960/50462 | Loss: 2.6095 | LR: 3.12e-06 | 3116.7ms/step | 19.3 samples/s
Step 47970/50462 | Loss: 2.6455 | LR: 3.09e-06 | 3114.3ms/step | 19.3 samples/s
Step 47980/50462 | Loss: 2.6451 | LR: 3.07e-06 | 3121.4ms/step | 19.2 samples/s
Step 47990/50462 | Loss: 2.6265 | LR: 3.04e-06 | 3113.8ms/step | 19.3 samples/s
Step 48000/50462 | Loss: 2.6342 | LR: 3.02e-06 | 3117.0ms/step | 19.2 samples/s
Validation at step 48000: Loss = 2.6523, Perplexity = 14.19
Saved checkpoint to checkpoints/run_20260616_011729/checkpoint-48000
Step 48010/50462 | Loss: 2.6480 | LR: 3.00e-06 | 5645.9ms/step | 10.6 samples/s
Step 48020/50462 | Loss: 2.6806 | LR: 2.97e-06 | 3114.5ms/step | 19.3 samples/s
Step 48030/50462 | Loss: 2.6194 | LR: 2.95e-06 | 3118.7ms/step | 19.2 samples/s
Step 48040/50462 | Loss: 2.6310 | LR: 2.92e-06 | 3114.1ms/step | 19.3 samples/s
Step 48050/50462 | Loss: 2.6274 | LR: 2.90e-06 | 3122.7ms/step | 19.2 samples/s
Step 48060/50462 | Loss: 2.6418 | LR: 2.87e-06 | 3128.3ms/step | 19.2 samples/s
Step 48070/50462 | Loss: 2.6465 | LR: 2.85e-06 | 3119.0ms/step | 19.2 samples/s
Step 48080/50462 | Loss: 2.6679 | LR: 2.83e-06 | 3120.1ms/step | 19.2 samples/s
Step 48090/50462 | Loss: 2.6455 | LR: 2.80e-06 | 3123.8ms/step | 19.2 samples/s
Step 48100/50462 | Loss: 2.6506 | LR: 2.78e-06 | 3119.7ms/step | 19.2 samples/s
Step 48110/50462 | Loss: 2.6378 | LR: 2.76e-06 | 3121.4ms/step | 19.2 samples/s
Step 48120/50462 | Loss: 2.6355 | LR: 2.73e-06 | 3114.9ms/step | 19.3 samples/s
Step 48130/50462 | Loss: 2.6366 | LR: 2.71e-06 | 3122.4ms/step | 19.2 samples/s
Step 48140/50462 | Loss: 2.6167 | LR: 2.69e-06 | 3121.0ms/step | 19.2 samples/s
Step 48150/50462 | Loss: 2.6306 | LR: 2.66e-06 | 3135.3ms/step | 19.1 samples/s
Step 48160/50462 | Loss: 2.6534 | LR: 2.64e-06 | 3179.5ms/step | 18.9 samples/s
Step 48170/50462 | Loss: 2.6327 | LR: 2.62e-06 | 3120.6ms/step | 19.2 samples/s
Step 48180/50462 | Loss: 2.6445 | LR: 2.60e-06 | 3136.1ms/step | 19.1 samples/s
Step 48190/50462 | Loss: 2.6398 | LR: 2.57e-06 | 3132.0ms/step | 19.2 samples/s
Step 48200/50462 | Loss: 2.6519 | LR: 2.55e-06 | 3117.3ms/step | 19.2 samples/s
Step 48210/50462 | Loss: 2.6254 | LR: 2.53e-06 | 3111.4ms/step | 19.3 samples/s
Step 48220/50462 | Loss: 2.6330 | LR: 2.51e-06 | 3114.7ms/step | 19.3 samples/s
Step 48230/50462 | Loss: 2.6414 | LR: 2.48e-06 | 3118.9ms/step | 19.2 samples/s
Step 48240/50462 | Loss: 2.6382 | LR: 2.46e-06 | 3232.7ms/step | 18.6 samples/s
Step 48250/50462 | Loss: 2.6743 | LR: 2.44e-06 | 3123.5ms/step | 19.2 samples/s
Validation at step 48250: Loss = 2.6523, Perplexity = 14.19
Step 48260/50462 | Loss: 2.6108 | LR: 2.42e-06 | 4787.2ms/step | 12.5 samples/s
Step 48270/50462 | Loss: 2.6390 | LR: 2.39e-06 | 3122.3ms/step | 19.2 samples/s
Step 48280/50462 | Loss: 2.6491 | LR: 2.37e-06 | 3119.7ms/step | 19.2 samples/s
Step 48290/50462 | Loss: 2.6340 | LR: 2.35e-06 | 3119.4ms/step | 19.2 samples/s
Step 48300/50462 | Loss: 2.6260 | LR: 2.33e-06 | 3117.7ms/step | 19.2 samples/s
Step 48310/50462 | Loss: 2.6507 | LR: 2.31e-06 | 3121.7ms/step | 19.2 samples/s
Step 48320/50462 | Loss: 2.6454 | LR: 2.29e-06 | 3122.4ms/step | 19.2 samples/s
Step 48330/50462 | Loss: 2.6343 | LR: 2.27e-06 | 3126.8ms/step | 19.2 samples/s
Step 48340/50462 | Loss: 2.6522 | LR: 2.24e-06 | 3118.9ms/step | 19.2 samples/s
Step 48350/50462 | Loss: 2.6664 | LR: 2.22e-06 | 3130.5ms/step | 19.2 samples/s
Step 48360/50462 | Loss: 2.6257 | LR: 2.20e-06 | 3127.1ms/step | 19.2 samples/s
Step 48370/50462 | Loss: 2.6615 | LR: 2.18e-06 | 3129.2ms/step | 19.2 samples/s
Step 48380/50462 | Loss: 2.6281 | LR: 2.16e-06 | 3129.0ms/step | 19.2 samples/s
Step 48390/50462 | Loss: 2.6724 | LR: 2.14e-06 | 3131.4ms/step | 19.2 samples/s
Step 48400/50462 | Loss: 2.6312 | LR: 2.12e-06 | 3129.3ms/step | 19.2 samples/s
Step 48410/50462 | Loss: 2.6582 | LR: 2.10e-06 | 3124.5ms/step | 19.2 samples/s
Step 48420/50462 | Loss: 2.6338 | LR: 2.08e-06 | 3127.1ms/step | 19.2 samples/s
Step 48430/50462 | Loss: 2.6318 | LR: 2.06e-06 | 3132.0ms/step | 19.2 samples/s
Step 48440/50462 | Loss: 2.6503 | LR: 2.04e-06 | 3123.5ms/step | 19.2 samples/s
Step 48450/50462 | Loss: 2.6135 | LR: 2.02e-06 | 3122.5ms/step | 19.2 samples/s
Step 48460/50462 | Loss: 2.6638 | LR: 2.00e-06 | 3130.0ms/step | 19.2 samples/s
Step 48470/50462 | Loss: 2.6523 | LR: 1.98e-06 | 3127.1ms/step | 19.2 samples/s
Step 48480/50462 | Loss: 2.6337 | LR: 1.96e-06 | 3187.0ms/step | 18.8 samples/s
Step 48490/50462 | Loss: 2.6403 | LR: 1.94e-06 | 3125.0ms/step | 19.2 samples/s
Step 48500/50462 | Loss: 2.6083 | LR: 1.92e-06 | 3129.4ms/step | 19.2 samples/s
Validation at step 48500: Loss = 2.6523, Perplexity = 14.19
Step 48510/50462 | Loss: 2.6432 | LR: 1.90e-06 | 4784.6ms/step | 12.5 samples/s
Step 48520/50462 | Loss: 2.6429 | LR: 1.88e-06 | 3127.9ms/step | 19.2 samples/s
Step 48530/50462 | Loss: 2.6614 | LR: 1.86e-06 | 3131.5ms/step | 19.2 samples/s
Step 48540/50462 | Loss: 2.6170 | LR: 1.84e-06 | 3132.8ms/step | 19.2 samples/s
Step 48550/50462 | Loss: 2.6260 | LR: 1.82e-06 | 3127.2ms/step | 19.2 samples/s
Step 48560/50462 | Loss: 2.6367 | LR: 1.80e-06 | 3240.2ms/step | 18.5 samples/s
Step 48570/50462 | Loss: 2.6389 | LR: 1.78e-06 | 3141.1ms/step | 19.1 samples/s
Step 48580/50462 | Loss: 2.6311 | LR: 1.77e-06 | 3146.8ms/step | 19.1 samples/s
Step 48590/50462 | Loss: 2.6570 | LR: 1.75e-06 | 3126.7ms/step | 19.2 samples/s
Step 48600/50462 | Loss: 2.6448 | LR: 1.73e-06 | 3129.2ms/step | 19.2 samples/s
Step 48610/50462 | Loss: 2.6306 | LR: 1.71e-06 | 3124.5ms/step | 19.2 samples/s
Step 48620/50462 | Loss: 2.6389 | LR: 1.69e-06 | 3120.5ms/step | 19.2 samples/s
Step 48630/50462 | Loss: 2.6608 | LR: 1.67e-06 | 3124.2ms/step | 19.2 samples/s
Step 48640/50462 | Loss: 2.6422 | LR: 1.66e-06 | 3127.5ms/step | 19.2 samples/s
Step 48650/50462 | Loss: 2.6574 | LR: 1.64e-06 | 3118.2ms/step | 19.2 samples/s
Step 48660/50462 | Loss: 2.6542 | LR: 1.62e-06 | 3117.6ms/step | 19.2 samples/s
Step 48670/50462 | Loss: 2.6545 | LR: 1.60e-06 | 3122.9ms/step | 19.2 samples/s
Step 48680/50462 | Loss: 2.6374 | LR: 1.58e-06 | 3115.2ms/step | 19.3 samples/s
Step 48690/50462 | Loss: 2.6510 | LR: 1.57e-06 | 3115.2ms/step | 19.3 samples/s
Step 48700/50462 | Loss: 2.6147 | LR: 1.55e-06 | 3118.5ms/step | 19.2 samples/s
Step 48710/50462 | Loss: 2.6366 | LR: 1.53e-06 | 3117.5ms/step | 19.2 samples/s
Step 48720/50462 | Loss: 2.6569 | LR: 1.51e-06 | 3114.4ms/step | 19.3 samples/s
Step 48730/50462 | Loss: 2.6399 | LR: 1.50e-06 | 3116.9ms/step | 19.2 samples/s
Step 48740/50462 | Loss: 2.6190 | LR: 1.48e-06 | 3115.2ms/step | 19.3 samples/s
Step 48750/50462 | Loss: 2.6620 | LR: 1.46e-06 | 3129.1ms/step | 19.2 samples/s
Validation at step 48750: Loss = 2.6523, Perplexity = 14.19
Step 48760/50462 | Loss: 2.6731 | LR: 1.44e-06 | 4803.6ms/step | 12.5 samples/s
Step 48770/50462 | Loss: 2.6131 | LR: 1.43e-06 | 3135.4ms/step | 19.1 samples/s
Step 48780/50462 | Loss: 2.6367 | LR: 1.41e-06 | 3128.4ms/step | 19.2 samples/s
Step 48790/50462 | Loss: 2.6470 | LR: 1.39e-06 | 3132.9ms/step | 19.2 samples/s
Step 48800/50462 | Loss: 2.6238 | LR: 1.38e-06 | 3186.9ms/step | 18.8 samples/s
Step 48810/50462 | Loss: 2.6301 | LR: 1.36e-06 | 3143.4ms/step | 19.1 samples/s
Step 48820/50462 | Loss: 2.6411 | LR: 1.34e-06 | 3133.1ms/step | 19.2 samples/s
Step 48830/50462 | Loss: 2.6525 | LR: 1.33e-06 | 3140.2ms/step | 19.1 samples/s
Step 48840/50462 | Loss: 2.6681 | LR: 1.31e-06 | 3131.4ms/step | 19.2 samples/s
Step 48850/50462 | Loss: 2.6377 | LR: 1.30e-06 | 3144.9ms/step | 19.1 samples/s
Step 48860/50462 | Loss: 2.6064 | LR: 1.28e-06 | 3133.4ms/step | 19.1 samples/s
Step 48870/50462 | Loss: 2.6508 | LR: 1.26e-06 | 3139.7ms/step | 19.1 samples/s
Step 48880/50462 | Loss: 2.6319 | LR: 1.25e-06 | 3235.1ms/step | 18.5 samples/s
Step 48890/50462 | Loss: 2.6605 | LR: 1.23e-06 | 3128.4ms/step | 19.2 samples/s
Step 48900/50462 | Loss: 2.6412 | LR: 1.22e-06 | 3128.0ms/step | 19.2 samples/s
Step 48910/50462 | Loss: 2.6548 | LR: 1.20e-06 | 3139.1ms/step | 19.1 samples/s
Step 48920/50462 | Loss: 2.6401 | LR: 1.19e-06 | 3134.9ms/step | 19.1 samples/s
Step 48930/50462 | Loss: 2.6366 | LR: 1.17e-06 | 3136.3ms/step | 19.1 samples/s
Step 48940/50462 | Loss: 2.6168 | LR: 1.16e-06 | 3138.8ms/step | 19.1 samples/s
Step 48950/50462 | Loss: 2.6181 | LR: 1.14e-06 | 3128.2ms/step | 19.2 samples/s
Step 48960/50462 | Loss: 2.6492 | LR: 1.13e-06 | 3135.7ms/step | 19.1 samples/s
Step 48970/50462 | Loss: 2.6148 | LR: 1.11e-06 | 3137.5ms/step | 19.1 samples/s
Step 48980/50462 | Loss: 2.6797 | LR: 1.10e-06 | 3132.5ms/step | 19.2 samples/s
Step 48990/50462 | Loss: 2.6659 | LR: 1.08e-06 | 3140.4ms/step | 19.1 samples/s
Step 49000/50462 | Loss: 2.6508 | LR: 1.07e-06 | 3124.7ms/step | 19.2 samples/s
Validation at step 49000: Loss = 2.6523, Perplexity = 14.19
Saved checkpoint to checkpoints/run_20260616_011729/checkpoint-49000
Step 49010/50462 | Loss: 2.6364 | LR: 1.05e-06 | 5665.9ms/step | 10.6 samples/s
Step 49020/50462 | Loss: 2.6454 | LR: 1.04e-06 | 3123.0ms/step | 19.2 samples/s
Step 49030/50462 | Loss: 2.6483 | LR: 1.02e-06 | 3131.6ms/step | 19.2 samples/s
Step 49040/50462 | Loss: 2.6438 | LR: 1.01e-06 | 3135.7ms/step | 19.1 samples/s
Step 49050/50462 | Loss: 2.6468 | LR: 9.95e-07 | 3136.0ms/step | 19.1 samples/s
Step 49060/50462 | Loss: 2.6646 | LR: 9.81e-07 | 3137.6ms/step | 19.1 samples/s
Step 49070/50462 | Loss: 2.6621 | LR: 9.67e-07 | 3133.6ms/step | 19.1 samples/s
Step 49080/50462 | Loss: 2.6672 | LR: 9.53e-07 | 3132.1ms/step | 19.2 samples/s
Step 49090/50462 | Loss: 2.6163 | LR: 9.39e-07 | 3130.9ms/step | 19.2 samples/s
Step 49100/50462 | Loss: 2.6666 | LR: 9.25e-07 | 3133.5ms/step | 19.1 samples/s
Step 49110/50462 | Loss: 2.6509 | LR: 9.12e-07 | 3135.7ms/step | 19.1 samples/s
Step 49120/50462 | Loss: 2.6218 | LR: 8.99e-07 | 3190.8ms/step | 18.8 samples/s
Step 49130/50462 | Loss: 2.6496 | LR: 8.85e-07 | 3120.6ms/step | 19.2 samples/s
Step 49140/50462 | Loss: 2.6220 | LR: 8.72e-07 | 3124.7ms/step | 19.2 samples/s
Step 49150/50462 | Loss: 2.6525 | LR: 8.59e-07 | 3130.8ms/step | 19.2 samples/s
Step 49160/50462 | Loss: 2.6446 | LR: 8.46e-07 | 3139.6ms/step | 19.1 samples/s
Step 49170/50462 | Loss: 2.6399 | LR: 8.33e-07 | 3133.1ms/step | 19.2 samples/s
Step 49180/50462 | Loss: 2.6310 | LR: 8.20e-07 | 3135.4ms/step | 19.1 samples/s
Step 49190/50462 | Loss: 2.6192 | LR: 8.07e-07 | 3131.4ms/step | 19.2 samples/s
Step 49200/50462 | Loss: 2.6753 | LR: 7.95e-07 | 3246.2ms/step | 18.5 samples/s
Step 49210/50462 | Loss: 2.6427 | LR: 7.82e-07 | 3128.1ms/step | 19.2 samples/s
Step 49220/50462 | Loss: 2.6532 | LR: 7.70e-07 | 3142.5ms/step | 19.1 samples/s
Step 49230/50462 | Loss: 2.6531 | LR: 7.57e-07 | 3142.5ms/step | 19.1 samples/s
Step 49240/50462 | Loss: 2.6469 | LR: 7.45e-07 | 3141.4ms/step | 19.1 samples/s
Step 49250/50462 | Loss: 2.6385 | LR: 7.33e-07 | 3138.3ms/step | 19.1 samples/s
Validation at step 49250: Loss = 2.6523, Perplexity = 14.19
Step 49260/50462 | Loss: 2.6136 | LR: 7.21e-07 | 4808.9ms/step | 12.5 samples/s
Step 49270/50462 | Loss: 2.6324 | LR: 7.09e-07 | 3139.8ms/step | 19.1 samples/s
Step 49280/50462 | Loss: 2.6143 | LR: 6.97e-07 | 3117.8ms/step | 19.2 samples/s
Step 49290/50462 | Loss: 2.6369 | LR: 6.85e-07 | 3127.9ms/step | 19.2 samples/s
Step 49300/50462 | Loss: 2.6367 | LR: 6.74e-07 | 3126.8ms/step | 19.2 samples/s
Step 49310/50462 | Loss: 2.6406 | LR: 6.62e-07 | 3133.9ms/step | 19.1 samples/s
Step 49320/50462 | Loss: 2.6290 | LR: 6.51e-07 | 3130.7ms/step | 19.2 samples/s
Step 49330/50462 | Loss: 2.6392 | LR: 6.39e-07 | 3134.2ms/step | 19.1 samples/s
Step 49340/50462 | Loss: 2.6482 | LR: 6.28e-07 | 3132.1ms/step | 19.2 samples/s
Step 49350/50462 | Loss: 2.6516 | LR: 6.17e-07 | 3130.4ms/step | 19.2 samples/s
Step 49360/50462 | Loss: 2.6623 | LR: 6.06e-07 | 3123.2ms/step | 19.2 samples/s
Step 49370/50462 | Loss: 2.6421 | LR: 5.95e-07 | 3129.0ms/step | 19.2 samples/s
Step 49380/50462 | Loss: 2.6582 | LR: 5.84e-07 | 3132.6ms/step | 19.2 samples/s
Step 49390/50462 | Loss: 2.6546 | LR: 5.73e-07 | 3135.6ms/step | 19.1 samples/s
Step 49400/50462 | Loss: 2.6524 | LR: 5.63e-07 | 3140.0ms/step | 19.1 samples/s
Step 49410/50462 | Loss: 2.6462 | LR: 5.52e-07 | 3132.6ms/step | 19.2 samples/s
Step 49420/50462 | Loss: 2.6431 | LR: 5.42e-07 | 3134.7ms/step | 19.1 samples/s
Step 49430/50462 | Loss: 2.6236 | LR: 5.31e-07 | 3126.1ms/step | 19.2 samples/s
Step 49440/50462 | Loss: 2.6314 | LR: 5.21e-07 | 3191.9ms/step | 18.8 samples/s
Step 49450/50462 | Loss: 2.6196 | LR: 5.11e-07 | 3124.8ms/step | 19.2 samples/s
Step 49460/50462 | Loss: 2.6511 | LR: 5.01e-07 | 3121.1ms/step | 19.2 samples/s
Step 49470/50462 | Loss: 2.6542 | LR: 4.91e-07 | 3123.7ms/step | 19.2 samples/s
Step 49480/50462 | Loss: 2.6438 | LR: 4.81e-07 | 3133.1ms/step | 19.2 samples/s
Step 49490/50462 | Loss: 2.6287 | LR: 4.72e-07 | 3124.8ms/step | 19.2 samples/s
Step 49500/50462 | Loss: 2.6497 | LR: 4.62e-07 | 3128.2ms/step | 19.2 samples/s
Validation at step 49500: Loss = 2.6523, Perplexity = 14.19
Step 49510/50462 | Loss: 2.6558 | LR: 4.52e-07 | 4809.3ms/step | 12.5 samples/s
Step 49520/50462 | Loss: 2.6382 | LR: 4.43e-07 | 3193.1ms/step | 18.8 samples/s
Step 49530/50462 | Loss: 2.6433 | LR: 4.34e-07 | 3184.2ms/step | 18.8 samples/s
Step 49540/50462 | Loss: 2.6334 | LR: 4.24e-07 | 3138.4ms/step | 19.1 samples/s
Step 49550/50462 | Loss: 2.6429 | LR: 4.15e-07 | 3145.6ms/step | 19.1 samples/s
Step 49560/50462 | Loss: 2.6582 | LR: 4.06e-07 | 3142.4ms/step | 19.1 samples/s
Step 49570/50462 | Loss: 2.6565 | LR: 3.97e-07 | 3119.7ms/step | 19.2 samples/s
Step 49580/50462 | Loss: 2.6552 | LR: 3.88e-07 | 3130.3ms/step | 19.2 samples/s
Step 49590/50462 | Loss: 2.6288 | LR: 3.79e-07 | 3133.1ms/step | 19.2 samples/s
Step 49600/50462 | Loss: 2.6643 | LR: 3.71e-07 | 3138.8ms/step | 19.1 samples/s
Step 49610/50462 | Loss: 2.6421 | LR: 3.62e-07 | 3139.4ms/step | 19.1 samples/s
Step 49620/50462 | Loss: 2.6415 | LR: 3.54e-07 | 3122.1ms/step | 19.2 samples/s
Step 49630/50462 | Loss: 2.6422 | LR: 3.45e-07 | 3128.4ms/step | 19.2 samples/s
Step 49640/50462 | Loss: 2.6431 | LR: 3.37e-07 | 3129.2ms/step | 19.2 samples/s
Step 49650/50462 | Loss: 2.6487 | LR: 3.29e-07 | 3140.3ms/step | 19.1 samples/s
Step 49660/50462 | Loss: 2.6300 | LR: 3.21e-07 | 3132.0ms/step | 19.2 samples/s
Step 49670/50462 | Loss: 2.6176 | LR: 3.13e-07 | 3129.7ms/step | 19.2 samples/s
Step 49680/50462 | Loss: 2.6613 | LR: 3.05e-07 | 3133.9ms/step | 19.1 samples/s
Step 49690/50462 | Loss: 2.6764 | LR: 2.97e-07 | 3136.0ms/step | 19.1 samples/s
Step 49700/50462 | Loss: 2.6506 | LR: 2.90e-07 | 3137.4ms/step | 19.1 samples/s
Step 49710/50462 | Loss: 2.6473 | LR: 2.82e-07 | 3128.4ms/step | 19.2 samples/s
Step 49720/50462 | Loss: 2.6337 | LR: 2.75e-07 | 3117.8ms/step | 19.2 samples/s
Step 49730/50462 | Loss: 2.6492 | LR: 2.67e-07 | 3131.5ms/step | 19.2 samples/s
Step 49740/50462 | Loss: 2.6723 | LR: 2.60e-07 | 3136.6ms/step | 19.1 samples/s
Step 49750/50462 | Loss: 2.6420 | LR: 2.53e-07 | 3130.2ms/step | 19.2 samples/s
Validation at step 49750: Loss = 2.6523, Perplexity = 14.19
Step 49760/50462 | Loss: 2.6594 | LR: 2.46e-07 | 4862.1ms/step | 12.3 samples/s
Step 49770/50462 | Loss: 2.6490 | LR: 2.39e-07 | 3126.2ms/step | 19.2 samples/s
Step 49780/50462 | Loss: 2.6339 | LR: 2.32e-07 | 3132.9ms/step | 19.2 samples/s
Step 49790/50462 | Loss: 2.6268 | LR: 2.25e-07 | 3140.6ms/step | 19.1 samples/s
Step 49800/50462 | Loss: 2.6680 | LR: 2.19e-07 | 3140.4ms/step | 19.1 samples/s
Step 49810/50462 | Loss: 2.6596 | LR: 2.12e-07 | 3128.4ms/step | 19.2 samples/s
Step 49820/50462 | Loss: 2.6532 | LR: 2.06e-07 | 3121.5ms/step | 19.2 samples/s
Step 49830/50462 | Loss: 2.6438 | LR: 1.99e-07 | 3128.7ms/step | 19.2 samples/s
Step 49840/50462 | Loss: 2.6568 | LR: 1.93e-07 | 3189.1ms/step | 18.8 samples/s
Step 49850/50462 | Loss: 2.6366 | LR: 1.87e-07 | 3191.5ms/step | 18.8 samples/s
Step 49860/50462 | Loss: 2.6358 | LR: 1.81e-07 | 3132.2ms/step | 19.2 samples/s
Step 49870/50462 | Loss: 2.6514 | LR: 1.75e-07 | 3140.8ms/step | 19.1 samples/s
Step 49880/50462 | Loss: 2.6378 | LR: 1.69e-07 | 3125.5ms/step | 19.2 samples/s
Step 49890/50462 | Loss: 2.6452 | LR: 1.63e-07 | 3126.7ms/step | 19.2 samples/s
Step 49900/50462 | Loss: 2.6582 | LR: 1.58e-07 | 3128.8ms/step | 19.2 samples/s
Step 49910/50462 | Loss: 2.6448 | LR: 1.52e-07 | 3122.8ms/step | 19.2 samples/s
Step 49920/50462 | Loss: 2.6564 | LR: 1.47e-07 | 3118.7ms/step | 19.2 samples/s
Step 49930/50462 | Loss: 2.6286 | LR: 1.41e-07 | 3132.0ms/step | 19.2 samples/s
Step 49940/50462 | Loss: 2.6390 | LR: 1.36e-07 | 3133.6ms/step | 19.1 samples/s
Step 49950/50462 | Loss: 2.6016 | LR: 1.31e-07 | 3132.1ms/step | 19.2 samples/s
Step 49960/50462 | Loss: 2.6337 | LR: 1.26e-07 | 3137.0ms/step | 19.1 samples/s
Step 49970/50462 | Loss: 2.6243 | LR: 1.21e-07 | 3133.5ms/step | 19.1 samples/s
Step 49980/50462 | Loss: 2.6393 | LR: 1.16e-07 | 3128.7ms/step | 19.2 samples/s
Step 49990/50462 | Loss: 2.6154 | LR: 1.11e-07 | 3134.8ms/step | 19.1 samples/s
Step 50000/50462 | Loss: 2.6229 | LR: 1.07e-07 | 3131.3ms/step | 19.2 samples/s
Validation at step 50000: Loss = 2.6523, Perplexity = 14.19
Saved checkpoint to checkpoints/run_20260616_011729/checkpoint-50000
Step 50010/50462 | Loss: 2.6196 | LR: 1.02e-07 | 5684.8ms/step | 10.6 samples/s
Step 50020/50462 | Loss: 2.6302 | LR: 9.75e-08 | 3133.6ms/step | 19.1 samples/s
Step 50030/50462 | Loss: 2.6589 | LR: 9.32e-08 | 3130.3ms/step | 19.2 samples/s
Step 50040/50462 | Loss: 2.6888 | LR: 8.89e-08 | 3136.8ms/step | 19.1 samples/s
Step 50050/50462 | Loss: 2.6419 | LR: 8.47e-08 | 3133.8ms/step | 19.1 samples/s
Step 50060/50462 | Loss: 2.6373 | LR: 8.07e-08 | 3127.6ms/step | 19.2 samples/s
Step 50070/50462 | Loss: 2.6557 | LR: 7.67e-08 | 3122.9ms/step | 19.2 samples/s
Step 50080/50462 | Loss: 2.6276 | LR: 7.28e-08 | 3186.8ms/step | 18.8 samples/s
Step 50090/50462 | Loss: 2.6211 | LR: 6.91e-08 | 3138.0ms/step | 19.1 samples/s
Step 50100/50462 | Loss: 2.6282 | LR: 6.54e-08 | 3124.0ms/step | 19.2 samples/s
Step 50110/50462 | Loss: 2.6667 | LR: 6.19e-08 | 3130.8ms/step | 19.2 samples/s
Step 50120/50462 | Loss: 2.6510 | LR: 5.84e-08 | 3138.7ms/step | 19.1 samples/s
Step 50130/50462 | Loss: 2.6420 | LR: 5.50e-08 | 3131.9ms/step | 19.2 samples/s
Step 50140/50462 | Loss: 2.6642 | LR: 5.18e-08 | 3132.6ms/step | 19.2 samples/s
Step 50150/50462 | Loss: 2.6204 | LR: 4.86e-08 | 3134.7ms/step | 19.1 samples/s
Step 50160/50462 | Loss: 2.6376 | LR: 4.55e-08 | 3195.8ms/step | 18.8 samples/s
Step 50170/50462 | Loss: 2.6048 | LR: 4.26e-08 | 3179.8ms/step | 18.9 samples/s
Step 50180/50462 | Loss: 2.6089 | LR: 3.97e-08 | 3126.2ms/step | 19.2 samples/s
Step 50190/50462 | Loss: 2.6590 | LR: 3.69e-08 | 3126.9ms/step | 19.2 samples/s
Step 50200/50462 | Loss: 2.6269 | LR: 3.43e-08 | 3137.4ms/step | 19.1 samples/s
Step 50210/50462 | Loss: 2.6562 | LR: 3.17e-08 | 3148.9ms/step | 19.1 samples/s
Step 50220/50462 | Loss: 2.6462 | LR: 2.92e-08 | 3140.9ms/step | 19.1 samples/s
Step 50230/50462 | Loss: 2.6657 | LR: 2.69e-08 | 3140.3ms/step | 19.1 samples/s
Step 50240/50462 | Loss: 2.6311 | LR: 2.46e-08 | 3129.2ms/step | 19.2 samples/s
Step 50250/50462 | Loss: 2.6219 | LR: 2.24e-08 | 3126.1ms/step | 19.2 samples/s
Validation at step 50250: Loss = 2.6523, Perplexity = 14.19
Step 50260/50462 | Loss: 2.6716 | LR: 2.04e-08 | 4799.8ms/step | 12.5 samples/s
Step 50270/50462 | Loss: 2.6808 | LR: 1.84e-08 | 3129.7ms/step | 19.2 samples/s
Step 50280/50462 | Loss: 2.6529 | LR: 1.65e-08 | 3142.4ms/step | 19.1 samples/s
Step 50290/50462 | Loss: 2.6686 | LR: 1.48e-08 | 3138.6ms/step | 19.1 samples/s
Step 50300/50462 | Loss: 2.6209 | LR: 1.31e-08 | 3142.0ms/step | 19.1 samples/s
Step 50310/50462 | Loss: 2.6156 | LR: 1.15e-08 | 3129.9ms/step | 19.2 samples/s
Step 50320/50462 | Loss: 2.6395 | LR: 1.01e-08 | 3130.9ms/step | 19.2 samples/s
Step 50330/50462 | Loss: 2.6627 | LR: 8.70e-09 | 3132.4ms/step | 19.2 samples/s
Step 50340/50462 | Loss: 2.6483 | LR: 7.43e-09 | 3140.9ms/step | 19.1 samples/s
Step 50350/50462 | Loss: 2.6276 | LR: 6.26e-09 | 3136.2ms/step | 19.1 samples/s
Step 50360/50462 | Loss: 2.6498 | LR: 5.19e-09 | 3129.6ms/step | 19.2 samples/s
Step 50370/50462 | Loss: 2.6263 | LR: 4.23e-09 | 3133.8ms/step | 19.1 samples/s
Step 50380/50462 | Loss: 2.6210 | LR: 3.36e-09 | 3125.8ms/step | 19.2 samples/s
Step 50390/50462 | Loss: 2.6584 | LR: 2.59e-09 | 3136.7ms/step | 19.1 samples/s
Step 50400/50462 | Loss: 2.6425 | LR: 1.92e-09 | 3197.8ms/step | 18.8 samples/s
Step 50410/50462 | Loss: 2.6343 | LR: 1.35e-09 | 3127.8ms/step | 19.2 samples/s
Step 50420/50462 | Loss: 2.6569 | LR: 8.81e-10 | 3129.9ms/step | 19.2 samples/s
Step 50430/50462 | Loss: 2.6486 | LR: 5.11e-10 | 3132.9ms/step | 19.2 samples/s
Step 50440/50462 | Loss: 2.6173 | LR: 2.42e-10 | 3138.6ms/step | 19.1 samples/s
Step 50450/50462 | Loss: 2.6578 | LR: 7.19e-11 | 3141.1ms/step | 19.1 samples/s
Step 50460/50462 | Loss: 2.6329 | LR: 2.00e-12 | 3130.1ms/step | 19.2 samples/s
Saved checkpoint to checkpoints/run_20260616_011729/checkpoint-50462
wandb: updating run metadata
wandb: uploading output.log; uploading wandb-summary.json; uploading config.yaml
wandb: uploading history steps 1046-1046, summary, console lines 1100-1100
wandb:
wandb: Run history:
wandb: eval/dense_interval_active ▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁
wandb: eval/dense_interval_end ▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁
wandb: eval/dense_interval_start ▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁
wandb: eval/loss β–ˆβ–‚β–‚β–‚β–‚β–β–β–β–β–β–β–β–β–β–β–β–β–β–β–β–β–β–β–β–β–β–β–β–β–β–β–β–β–β–β–β–β–β–β–
wandb: eval/perplexity β–ˆβ–‚β–‚β–‚β–‚β–β–β–β–β–β–β–β–β–β–β–β–β–β–β–β–β–β–β–β–β–β–β–β–β–β–β–β–β–β–β–β–β–β–β–
wandb: eval/uses_dense_attention ▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁
wandb: eval/uses_sparse_attention ▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁
wandb: speed/ms_per_step β–‚β–‚β–‚β–ƒβ–‚β–‚β–ƒβ–‚β–‚β–‚β–ƒβ–‚β–ƒβ–‚β–ˆβ–‚β–‚β–‚β–ƒβ–ƒβ–‚β–‚β–β–β–ƒβ–‚β–‚β–‚β–ƒβ–‚β–‚β–‚β–‚β–ƒβ–…β–ƒβ–„β–‚β–„β–„
wandb: speed/samples_per_sec β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆ
wandb: speed/steps_per_sec β–‡β–‡β–‡β–‡β–‡β–ˆβ–ˆβ–ˆβ–ˆβ–‡β–ˆβ–ˆβ–‡β–‡β–ˆβ–‡β–β–†β–‡β–‡β–‡β–‡β–‡β–‡β–ˆβ–ˆβ–ˆβ–ˆβ–‡β–ˆβ–†β–β–†β–†β–‡β–‡β–β–‡β–‚β–‡
wandb: +4 ...
wandb:
wandb: Run summary:
wandb: eval/dense_interval_active 0
wandb: eval/dense_interval_end -1
wandb: eval/dense_interval_start -1
wandb: eval/loss 2.65233
wandb: eval/perplexity 14.18704
wandb: eval/uses_dense_attention 0
wandb: eval/uses_sparse_attention 1
wandb: speed/ms_per_step 3130.08585
wandb: speed/samples_per_sec 19.1688
wandb: speed/steps_per_sec 0.31948
wandb: +4 ...
wandb:
wandb: πŸš€ View run sparse 1mask at: https://wandb.ai/ruben-navasardyan4869-epfl/stochastic_sparsity/runs/apvt4901
wandb: ⭐️ View project at: https://wandb.ai/ruben-navasardyan4869-epfl/stochastic_sparsity
wandb: Synced 5 W&B file(s), 0 media file(s), 0 artifact file(s) and 0 other file(s)
wandb: Find logs at: ./wandb/run-20260616_012002-apvt4901/logs
Training complete!