2026-06-22 08:43:58,964 [INFO] Seed: 42 2026-06-22 08:43:58,980 [INFO] Using CUDA: NVIDIA A100 80GB PCIe 2026-06-22 08:43:58,980 [INFO] Enabled TF32 matmul/cudnn for CUDA 2026-06-22 08:44:51,887 [INFO] LLM2Seq Model Summary ================================================== Encoder: McGill-NLP/LLM2Vec-Sheared-LLaMA-mntp Trainable: False Adaptor: gated_residual Layer Fusion: True Fusion Type: tokenwise Salience Gate: True EncStack: True Global Tokens: 16 Decoder: Layers: 8 Hidden Size: 1024 Heads: 16 FFN Size: 4096 ================================================== Total params: 1,558,708,737 Trainable params: 263,830,017 Frozen params: 1,294,878,720 2026-06-22 08:44:52,571 [INFO] Hugging Face uploads enabled: repo=BienKieu/llm2seq-wikilingua path=checkpoints/h200_phase1_warmup 2026-06-22 08:44:52,571 [INFO] Train file: llm2seq_h200/data/processed/train.jsonl (13,999 examples) 2026-06-22 08:44:52,571 [INFO] Eval file: llm2seq_h200/data/processed/validation.jsonl (1,680 examples) 2026-06-22 08:44:52,572 [INFO] Starting training for 6 epochs (1,752 optimizer steps)... 2026-06-22 08:44:52,572 [INFO] Batch size: 16 2026-06-22 08:44:52,572 [INFO] Gradient accumulation: 3 2026-06-22 08:44:52,572 [INFO] Effective batch size: 48 2026-06-22 08:44:52,572 [INFO] Steps per epoch: 292 2026-06-22 08:44:52,572 [INFO] Warmup steps: 52 2026-06-22 08:44:52,572 [INFO] Eval every: 292 optimizer steps 2026-06-22 08:44:52,572 [INFO] Save every: 292 optimizer steps 2026-06-22 08:44:52,572 [INFO] Mixed precision: fp16=False, bf16=True 2026-06-22 08:45:25,164 [INFO] Step 5/1752 | Loss: 10.3101 | CE: 10.3101 | LR: 1.92e-05 | Epoch: 1/6 2026-06-22 08:45:56,556 [INFO] Step 10/1752 | Loss: 9.0068 | CE: 9.0068 | LR: 3.85e-05 | Epoch: 1/6 2026-06-22 08:46:27,563 [INFO] Step 15/1752 | Loss: 8.2175 | CE: 8.2175 | LR: 5.77e-05 | Epoch: 1/6 2026-06-22 08:46:58,968 [INFO] Step 20/1752 | Loss: 7.8386 | CE: 7.8386 | LR: 7.69e-05 | Epoch: 1/6 2026-06-22 08:47:30,426 [INFO] Step 25/1752 | Loss: 7.2799 | CE: 7.2799 | LR: 9.62e-05 | Epoch: 1/6 2026-06-22 08:48:02,294 [INFO] Step 30/1752 | Loss: 6.5960 | CE: 6.5960 | LR: 1.15e-04 | Epoch: 1/6 2026-06-22 08:48:33,507 [INFO] Step 35/1752 | Loss: 5.8359 | CE: 5.8359 | LR: 1.35e-04 | Epoch: 1/6 2026-06-22 08:49:05,518 [INFO] Step 40/1752 | Loss: 4.9562 | CE: 4.9562 | LR: 1.54e-04 | Epoch: 1/6 2026-06-22 08:49:37,517 [INFO] Step 45/1752 | Loss: 4.3904 | CE: 4.3904 | LR: 1.73e-04 | Epoch: 1/6 2026-06-22 08:50:09,209 [INFO] Step 50/1752 | Loss: 4.0451 | CE: 4.0451 | LR: 1.92e-04 | Epoch: 1/6 2026-06-22 08:50:40,739 [INFO] Step 55/1752 | Loss: 3.8212 | CE: 3.8212 | LR: 2.00e-04 | Epoch: 1/6 2026-06-22 08:51:12,570 [INFO] Step 60/1752 | Loss: 3.5860 | CE: 3.5860 | LR: 2.00e-04 | Epoch: 1/6 2026-06-22 08:51:43,935 [INFO] Step 65/1752 | Loss: 3.5093 | CE: 3.5093 | LR: 2.00e-04 | Epoch: 1/6 2026-06-22 08:52:16,381 [INFO] Step 70/1752 | Loss: 3.4100 | CE: 3.4100 | LR: 2.00e-04 | Epoch: 1/6 2026-06-22 08:52:47,316 [INFO] Step 75/1752 | Loss: 3.3261 | CE: 3.3261 | LR: 2.00e-04 | Epoch: 1/6 2026-06-22 08:53:19,668 [INFO] Step 80/1752 | Loss: 3.1444 | CE: 3.1444 | LR: 2.00e-04 | Epoch: 1/6 2026-06-22 08:53:51,542 [INFO] Step 85/1752 | Loss: 3.1159 | CE: 3.1159 | LR: 2.00e-04 | Epoch: 1/6 2026-06-22 08:54:27,005 [INFO] Seed: 42 2026-06-22 08:54:27,022 [INFO] Using CUDA: NVIDIA A100 80GB PCIe 2026-06-22 08:54:27,022 [INFO] Enabled TF32 matmul/cudnn for CUDA 2026-06-22 08:54:31,893 [INFO] LLM2Seq Model Summary ================================================== Encoder: McGill-NLP/LLM2Vec-Sheared-LLaMA-mntp Trainable: False Adaptor: gated_residual Layer Fusion: True Fusion Type: tokenwise Salience Gate: True EncStack: True Global Tokens: 16 Decoder: Layers: 8 Hidden Size: 1024 Heads: 16 FFN Size: 4096 ================================================== Total params: 1,558,708,737 Trainable params: 263,830,017 Frozen params: 1,294,878,720 2026-06-22 08:54:32,786 [INFO] Hugging Face uploads enabled: repo=BienKieu/llm2seq-wikilingua path=checkpoints/h200_phase1_warmup 2026-06-22 08:54:32,786 [INFO] Train file: llm2seq_h200/data/processed/train.jsonl (13,999 examples) 2026-06-22 08:54:32,786 [INFO] Eval file: llm2seq_h200/data/processed/validation.jsonl (1,680 examples) 2026-06-22 08:54:32,786 [INFO] Starting training for 6 epochs (1,752 optimizer steps)... 2026-06-22 08:54:32,786 [INFO] Batch size: 24 2026-06-22 08:54:32,786 [INFO] Gradient accumulation: 2 2026-06-22 08:54:32,786 [INFO] Effective batch size: 48 2026-06-22 08:54:32,786 [INFO] Steps per epoch: 292 2026-06-22 08:54:32,786 [INFO] Warmup steps: 52 2026-06-22 08:54:32,786 [INFO] Eval every: 292 optimizer steps 2026-06-22 08:54:32,786 [INFO] Save every: 292 optimizer steps 2026-06-22 08:54:32,786 [INFO] Mixed precision: fp16=False, bf16=True 2026-06-22 08:55:05,388 [INFO] Step 5/1752 | Loss: 10.3099 | CE: 10.3099 | LR: 1.92e-05 | Epoch: 1/6 2026-06-22 08:55:37,734 [INFO] Step 10/1752 | Loss: 9.0103 | CE: 9.0103 | LR: 3.85e-05 | Epoch: 1/6 2026-06-22 08:56:09,858 [INFO] Step 15/1752 | Loss: 8.2184 | CE: 8.2184 | LR: 5.77e-05 | Epoch: 1/6 2026-06-22 08:56:42,343 [INFO] Step 20/1752 | Loss: 7.8383 | CE: 7.8383 | LR: 7.69e-05 | Epoch: 1/6 2026-06-22 08:57:14,062 [INFO] Step 25/1752 | Loss: 7.2796 | CE: 7.2796 | LR: 9.62e-05 | Epoch: 1/6 2026-06-22 08:57:46,449 [INFO] Step 30/1752 | Loss: 6.5989 | CE: 6.5989 | LR: 1.15e-04 | Epoch: 1/6 2026-06-22 08:58:17,632 [INFO] Step 35/1752 | Loss: 5.8298 | CE: 5.8298 | LR: 1.35e-04 | Epoch: 1/6 2026-06-22 08:58:49,815 [INFO] Step 40/1752 | Loss: 4.9667 | CE: 4.9667 | LR: 1.54e-04 | Epoch: 1/6 2026-06-22 08:59:22,204 [INFO] Step 45/1752 | Loss: 4.4023 | CE: 4.4023 | LR: 1.73e-04 | Epoch: 1/6 2026-06-22 08:59:54,620 [INFO] Step 50/1752 | Loss: 4.0666 | CE: 4.0666 | LR: 1.92e-04 | Epoch: 1/6 2026-06-22 09:00:26,936 [INFO] Step 55/1752 | Loss: 3.8419 | CE: 3.8419 | LR: 2.00e-04 | Epoch: 1/6 2026-06-22 09:00:59,230 [INFO] Step 60/1752 | Loss: 3.6013 | CE: 3.6013 | LR: 2.00e-04 | Epoch: 1/6 2026-06-22 09:01:30,969 [INFO] Step 65/1752 | Loss: 3.5255 | CE: 3.5255 | LR: 2.00e-04 | Epoch: 1/6 2026-06-22 09:02:03,534 [INFO] Step 70/1752 | Loss: 3.4218 | CE: 3.4218 | LR: 2.00e-04 | Epoch: 1/6 2026-06-22 09:02:35,792 [INFO] Step 75/1752 | Loss: 3.3588 | CE: 3.3588 | LR: 2.00e-04 | Epoch: 1/6 2026-06-22 09:03:08,402 [INFO] Step 80/1752 | Loss: 3.1766 | CE: 3.1766 | LR: 2.00e-04 | Epoch: 1/6 2026-06-22 09:03:40,833 [INFO] Step 85/1752 | Loss: 3.1377 | CE: 3.1377 | LR: 2.00e-04 | Epoch: 1/6 2026-06-22 09:04:11,388 [INFO] Step 90/1752 | Loss: 3.0617 | CE: 3.0617 | LR: 2.00e-04 | Epoch: 1/6 2026-06-22 09:04:43,741 [INFO] Step 95/1752 | Loss: 2.9882 | CE: 2.9882 | LR: 2.00e-04 | Epoch: 1/6 2026-06-22 09:05:16,185 [INFO] Step 100/1752 | Loss: 2.9936 | CE: 2.9936 | LR: 2.00e-04 | Epoch: 1/6 2026-06-22 09:05:48,407 [INFO] Step 105/1752 | Loss: 2.8929 | CE: 2.8929 | LR: 2.00e-04 | Epoch: 1/6 2026-06-22 09:06:20,663 [INFO] Step 110/1752 | Loss: 2.8342 | CE: 2.8342 | LR: 1.99e-04 | Epoch: 1/6 2026-06-22 09:06:51,975 [INFO] Step 115/1752 | Loss: 2.8710 | CE: 2.8710 | LR: 1.99e-04 | Epoch: 1/6 2026-06-22 09:07:24,229 [INFO] Step 120/1752 | Loss: 2.7763 | CE: 2.7763 | LR: 1.99e-04 | Epoch: 1/6 2026-06-22 09:07:55,519 [INFO] Step 125/1752 | Loss: 2.7521 | CE: 2.7521 | LR: 1.99e-04 | Epoch: 1/6 2026-06-22 09:08:27,725 [INFO] Step 130/1752 | Loss: 2.6914 | CE: 2.6914 | LR: 1.99e-04 | Epoch: 1/6 2026-06-22 09:08:59,985 [INFO] Step 135/1752 | Loss: 2.6851 | CE: 2.6851 | LR: 1.99e-04 | Epoch: 1/6 2026-06-22 09:09:31,663 [INFO] Step 140/1752 | Loss: 2.5694 | CE: 2.5694 | LR: 1.99e-04 | Epoch: 1/6 2026-06-22 09:10:04,003 [INFO] Step 145/1752 | Loss: 2.5791 | CE: 2.5791 | LR: 1.99e-04 | Epoch: 1/6 2026-06-22 09:10:36,411 [INFO] Step 150/1752 | Loss: 2.5329 | CE: 2.5329 | LR: 1.98e-04 | Epoch: 1/6 2026-06-22 09:11:07,805 [INFO] Step 155/1752 | Loss: 2.4948 | CE: 2.4948 | LR: 1.98e-04 | Epoch: 1/6 2026-06-22 09:11:40,007 [INFO] Step 160/1752 | Loss: 2.4423 | CE: 2.4423 | LR: 1.98e-04 | Epoch: 1/6 2026-06-22 09:12:12,359 [INFO] Step 165/1752 | Loss: 2.4318 | CE: 2.4318 | LR: 1.98e-04 | Epoch: 1/6 2026-06-22 09:12:45,043 [INFO] Step 170/1752 | Loss: 2.4162 | CE: 2.4162 | LR: 1.98e-04 | Epoch: 1/6 2026-06-22 09:13:17,628 [INFO] Step 175/1752 | Loss: 2.4034 | CE: 2.4034 | LR: 1.97e-04 | Epoch: 1/6 2026-06-22 09:13:49,958 [INFO] Step 180/1752 | Loss: 2.3306 | CE: 2.3306 | LR: 1.97e-04 | Epoch: 1/6 2026-06-22 09:14:21,971 [INFO] Step 185/1752 | Loss: 2.3402 | CE: 2.3402 | LR: 1.97e-04 | Epoch: 1/6 2026-06-22 09:14:54,538 [INFO] Step 190/1752 | Loss: 2.2873 | CE: 2.2873 | LR: 1.97e-04 | Epoch: 1/6 2026-06-22 09:15:26,798 [INFO] Step 195/1752 | Loss: 2.2685 | CE: 2.2685 | LR: 1.97e-04 | Epoch: 1/6 2026-06-22 09:15:58,582 [INFO] Step 200/1752 | Loss: 2.2830 | CE: 2.2830 | LR: 1.96e-04 | Epoch: 1/6 2026-06-22 09:16:30,100 [INFO] Step 205/1752 | Loss: 2.2615 | CE: 2.2615 | LR: 1.96e-04 | Epoch: 1/6 2026-06-22 09:17:01,572 [INFO] Step 210/1752 | Loss: 2.2175 | CE: 2.2175 | LR: 1.96e-04 | Epoch: 1/6 2026-06-22 09:17:32,953 [INFO] Step 215/1752 | Loss: 2.2886 | CE: 2.2886 | LR: 1.95e-04 | Epoch: 1/6 2026-06-22 09:18:05,600 [INFO] Step 220/1752 | Loss: 2.2106 | CE: 2.2106 | LR: 1.95e-04 | Epoch: 1/6 2026-06-22 09:18:37,767 [INFO] Step 225/1752 | Loss: 2.2364 | CE: 2.2364 | LR: 1.95e-04 | Epoch: 1/6 2026-06-22 09:19:10,119 [INFO] Step 230/1752 | Loss: 2.2020 | CE: 2.2020 | LR: 1.95e-04 | Epoch: 1/6 2026-06-22 09:19:42,312 [INFO] Step 235/1752 | Loss: 2.1621 | CE: 2.1621 | LR: 1.94e-04 | Epoch: 1/6 2026-06-22 09:20:14,263 [INFO] Step 240/1752 | Loss: 2.1642 | CE: 2.1642 | LR: 1.94e-04 | Epoch: 1/6 2026-06-22 09:20:44,688 [INFO] Step 245/1752 | Loss: 2.1670 | CE: 2.1670 | LR: 1.94e-04 | Epoch: 1/6 2026-06-22 09:21:16,779 [INFO] Step 250/1752 | Loss: 2.1058 | CE: 2.1058 | LR: 1.93e-04 | Epoch: 1/6 2026-06-22 09:21:49,306 [INFO] Step 255/1752 | Loss: 2.0787 | CE: 2.0787 | LR: 1.93e-04 | Epoch: 1/6 2026-06-22 09:22:21,410 [INFO] Step 260/1752 | Loss: 2.1216 | CE: 2.1216 | LR: 1.93e-04 | Epoch: 1/6 2026-06-22 09:22:53,898 [INFO] Step 265/1752 | Loss: 2.1117 | CE: 2.1117 | LR: 1.92e-04 | Epoch: 1/6 2026-06-22 09:23:26,232 [INFO] Step 270/1752 | Loss: 2.0928 | CE: 2.0928 | LR: 1.92e-04 | Epoch: 1/6 2026-06-22 09:23:58,515 [INFO] Step 275/1752 | Loss: 2.0381 | CE: 2.0381 | LR: 1.92e-04 | Epoch: 1/6 2026-06-22 09:24:30,421 [INFO] Step 280/1752 | Loss: 2.1006 | CE: 2.1006 | LR: 1.91e-04 | Epoch: 1/6 2026-06-22 09:25:02,602 [INFO] Step 285/1752 | Loss: 2.0804 | CE: 2.0804 | LR: 1.91e-04 | Epoch: 1/6 2026-06-22 09:25:35,051 [INFO] Step 290/1752 | Loss: 1.9947 | CE: 1.9947 | LR: 1.90e-04 | Epoch: 1/6 2026-06-22 09:28:37,494 [INFO] Step 292 | Eval Loss: 1.9995 2026-06-22 09:28:38,399 [INFO] Saved trainable-only checkpoint to runs/h200_llm2seq_phase1_warmup/best.pt (183 model tensors, stores_base_encoder_weights=False) 2026-06-22 09:28:38,926 [INFO] New best eval loss: 1.9995 2026-06-22 09:28:39,510 [INFO] Saved trainable-only checkpoint to runs/h200_llm2seq_phase1_warmup/epoch_002.pt (183 model tensors, stores_base_encoder_weights=False) 2026-06-22 09:28:40,184 [INFO] Uploading to HF: runs/h200_llm2seq_phase1_warmup/epoch_002.pt -> BienKieu/llm2seq-wikilingua/checkpoints/h200_phase1_warmup/epochs/epoch_002_step_292.pt 2026-06-22 09:29:04,516 [INFO] Uploading to HF: runs/h200_llm2seq_phase1_warmup/config.yaml -> BienKieu/llm2seq-wikilingua/checkpoints/h200_phase1_warmup/config.yaml 2026-06-22 09:29:05,024 [INFO] Uploading to HF: runs/h200_llm2seq_phase1_warmup/train.log -> BienKieu/llm2seq-wikilingua/checkpoints/h200_phase1_warmup/train.log 2026-06-22 09:29:05,543 [INFO] Uploading to HF: runs/h200_llm2seq_phase1_warmup/checkpoint_manifest.json -> BienKieu/llm2seq-wikilingua/checkpoints/h200_phase1_warmup/checkpoint_manifest.json 2026-06-22 09:29:06,220 [INFO] Uploading to HF: runs/h200_llm2seq_phase1_warmup/best.pt -> BienKieu/llm2seq-wikilingua/checkpoints/h200_phase1_warmup/best.pt 2026-06-22 09:29:13,361 [INFO] Saved trainable-only checkpoint to runs/h200_llm2seq_phase1_warmup/checkpoint_292.pt (183 model tensors, stores_base_encoder_weights=False) 2026-06-22 09:29:34,337 [INFO] Step 295/1752 | Loss: 1.9350 | CE: 1.9350 | LR: 1.90e-04 | Epoch: 2/6 2026-06-22 09:30:06,275 [INFO] Step 300/1752 | Loss: 1.9444 | CE: 1.9444 | LR: 1.90e-04 | Epoch: 2/6 2026-06-22 09:30:37,059 [INFO] Step 305/1752 | Loss: 1.9729 | CE: 1.9729 | LR: 1.89e-04 | Epoch: 2/6 2026-06-22 09:31:09,347 [INFO] Step 310/1752 | Loss: 1.9443 | CE: 1.9443 | LR: 1.89e-04 | Epoch: 2/6 2026-06-22 09:31:41,599 [INFO] Step 315/1752 | Loss: 1.9277 | CE: 1.9277 | LR: 1.88e-04 | Epoch: 2/6 2026-06-22 09:32:13,475 [INFO] Step 320/1752 | Loss: 1.9229 | CE: 1.9229 | LR: 1.88e-04 | Epoch: 2/6 2026-06-22 09:32:45,912 [INFO] Step 325/1752 | Loss: 1.9151 | CE: 1.9151 | LR: 1.88e-04 | Epoch: 2/6 2026-06-22 09:33:17,431 [INFO] Step 330/1752 | Loss: 1.9042 | CE: 1.9042 | LR: 1.87e-04 | Epoch: 2/6 2026-06-22 09:33:49,963 [INFO] Step 335/1752 | Loss: 1.8793 | CE: 1.8793 | LR: 1.87e-04 | Epoch: 2/6 2026-06-22 09:34:22,144 [INFO] Step 340/1752 | Loss: 1.8980 | CE: 1.8980 | LR: 1.86e-04 | Epoch: 2/6 2026-06-22 09:34:53,412 [INFO] Step 345/1752 | Loss: 1.9578 | CE: 1.9578 | LR: 1.86e-04 | Epoch: 2/6 2026-06-22 09:35:25,937 [INFO] Step 350/1752 | Loss: 1.8626 | CE: 1.8626 | LR: 1.85e-04 | Epoch: 2/6 2026-06-22 09:35:57,906 [INFO] Step 355/1752 | Loss: 1.8920 | CE: 1.8920 | LR: 1.85e-04 | Epoch: 2/6 2026-06-22 09:36:29,684 [INFO] Step 360/1752 | Loss: 1.8809 | CE: 1.8809 | LR: 1.84e-04 | Epoch: 2/6 2026-06-22 09:37:00,918 [INFO] Step 365/1752 | Loss: 1.8476 | CE: 1.8476 | LR: 1.84e-04 | Epoch: 2/6 2026-06-22 09:37:33,293 [INFO] Step 370/1752 | Loss: 1.8938 | CE: 1.8938 | LR: 1.83e-04 | Epoch: 2/6 2026-06-22 09:38:05,316 [INFO] Step 375/1752 | Loss: 1.8803 | CE: 1.8803 | LR: 1.83e-04 | Epoch: 2/6 2026-06-22 09:38:37,667 [INFO] Step 380/1752 | Loss: 1.8337 | CE: 1.8337 | LR: 1.82e-04 | Epoch: 2/6 2026-06-22 09:39:09,872 [INFO] Step 385/1752 | Loss: 1.8419 | CE: 1.8419 | LR: 1.82e-04 | Epoch: 2/6 2026-06-22 09:39:41,958 [INFO] Step 390/1752 | Loss: 1.8363 | CE: 1.8363 | LR: 1.81e-04 | Epoch: 2/6 2026-06-22 09:40:13,353 [INFO] Step 395/1752 | Loss: 1.8210 | CE: 1.8210 | LR: 1.81e-04 | Epoch: 2/6 2026-06-22 09:40:44,575 [INFO] Step 400/1752 | Loss: 1.8165 | CE: 1.8165 | LR: 1.80e-04 | Epoch: 2/6 2026-06-22 09:41:17,113 [INFO] Step 405/1752 | Loss: 1.8433 | CE: 1.8433 | LR: 1.79e-04 | Epoch: 2/6 2026-06-22 09:41:49,636 [INFO] Step 410/1752 | Loss: 1.7799 | CE: 1.7799 | LR: 1.79e-04 | Epoch: 2/6 2026-06-22 09:42:22,068 [INFO] Step 415/1752 | Loss: 1.7994 | CE: 1.7994 | LR: 1.78e-04 | Epoch: 2/6 2026-06-22 09:42:54,409 [INFO] Step 420/1752 | Loss: 1.7838 | CE: 1.7838 | LR: 1.78e-04 | Epoch: 2/6 2026-06-22 09:43:26,688 [INFO] Step 425/1752 | Loss: 1.8199 | CE: 1.8199 | LR: 1.77e-04 | Epoch: 2/6 2026-06-22 09:43:58,420 [INFO] Step 430/1752 | Loss: 1.8012 | CE: 1.8012 | LR: 1.77e-04 | Epoch: 2/6 2026-06-22 09:44:30,807 [INFO] Step 435/1752 | Loss: 1.8697 | CE: 1.8697 | LR: 1.76e-04 | Epoch: 2/6 2026-06-22 09:45:03,031 [INFO] Step 440/1752 | Loss: 1.8417 | CE: 1.8417 | LR: 1.75e-04 | Epoch: 2/6 2026-06-22 09:45:35,244 [INFO] Step 445/1752 | Loss: 1.7841 | CE: 1.7841 | LR: 1.75e-04 | Epoch: 2/6 2026-06-22 09:46:07,624 [INFO] Step 450/1752 | Loss: 1.7278 | CE: 1.7278 | LR: 1.74e-04 | Epoch: 2/6 2026-06-22 09:46:39,905 [INFO] Step 455/1752 | Loss: 1.7606 | CE: 1.7606 | LR: 1.74e-04 | Epoch: 2/6 2026-06-22 09:47:12,312 [INFO] Step 460/1752 | Loss: 1.7460 | CE: 1.7460 | LR: 1.73e-04 | Epoch: 2/6 2026-06-22 09:47:44,620 [INFO] Step 465/1752 | Loss: 1.7582 | CE: 1.7582 | LR: 1.72e-04 | Epoch: 2/6 2026-06-22 09:48:17,081 [INFO] Step 470/1752 | Loss: 1.7956 | CE: 1.7956 | LR: 1.72e-04 | Epoch: 2/6 2026-06-22 09:48:49,489 [INFO] Step 475/1752 | Loss: 1.7470 | CE: 1.7470 | LR: 1.71e-04 | Epoch: 2/6 2026-06-22 09:49:20,846 [INFO] Step 480/1752 | Loss: 1.8007 | CE: 1.8007 | LR: 1.70e-04 | Epoch: 2/6 2026-06-22 09:49:52,259 [INFO] Step 485/1752 | Loss: 1.7512 | CE: 1.7512 | LR: 1.70e-04 | Epoch: 2/6 2026-06-22 09:50:23,927 [INFO] Step 490/1752 | Loss: 1.6753 | CE: 1.6753 | LR: 1.69e-04 | Epoch: 2/6 2026-06-22 09:50:54,625 [INFO] Step 495/1752 | Loss: 1.7743 | CE: 1.7743 | LR: 1.68e-04 | Epoch: 2/6 2026-06-22 09:51:26,944 [INFO] Step 500/1752 | Loss: 1.7090 | CE: 1.7090 | LR: 1.68e-04 | Epoch: 2/6 2026-06-22 09:51:58,791 [INFO] Step 505/1752 | Loss: 1.7604 | CE: 1.7604 | LR: 1.67e-04 | Epoch: 2/6 2026-06-22 09:52:30,930 [INFO] Step 510/1752 | Loss: 1.7121 | CE: 1.7121 | LR: 1.66e-04 | Epoch: 2/6 2026-06-22 09:53:03,470 [INFO] Step 515/1752 | Loss: 1.7203 | CE: 1.7203 | LR: 1.66e-04 | Epoch: 2/6 2026-06-22 09:53:35,425 [INFO] Step 520/1752 | Loss: 1.6758 | CE: 1.6758 | LR: 1.65e-04 | Epoch: 2/6 2026-06-22 09:54:07,624 [INFO] Step 525/1752 | Loss: 1.6903 | CE: 1.6903 | LR: 1.64e-04 | Epoch: 2/6 2026-06-22 09:54:39,469 [INFO] Step 530/1752 | Loss: 1.7116 | CE: 1.7116 | LR: 1.63e-04 | Epoch: 2/6 2026-06-22 09:55:11,067 [INFO] Step 535/1752 | Loss: 1.7308 | CE: 1.7308 | LR: 1.63e-04 | Epoch: 2/6 2026-06-22 09:55:43,527 [INFO] Step 540/1752 | Loss: 1.6935 | CE: 1.6935 | LR: 1.62e-04 | Epoch: 2/6 2026-06-22 09:56:15,920 [INFO] Step 545/1752 | Loss: 1.6881 | CE: 1.6881 | LR: 1.61e-04 | Epoch: 2/6 2026-06-22 09:56:48,428 [INFO] Step 550/1752 | Loss: 1.7055 | CE: 1.7055 | LR: 1.61e-04 | Epoch: 2/6 2026-06-22 09:57:19,885 [INFO] Step 555/1752 | Loss: 1.6478 | CE: 1.6478 | LR: 1.60e-04 | Epoch: 2/6 2026-06-22 09:57:52,141 [INFO] Step 560/1752 | Loss: 1.7067 | CE: 1.7067 | LR: 1.59e-04 | Epoch: 2/6 2026-06-22 09:58:23,367 [INFO] Step 565/1752 | Loss: 1.6629 | CE: 1.6629 | LR: 1.58e-04 | Epoch: 2/6 2026-06-22 09:58:55,929 [INFO] Step 570/1752 | Loss: 1.6419 | CE: 1.6419 | LR: 1.58e-04 | Epoch: 2/6 2026-06-22 09:59:28,312 [INFO] Step 575/1752 | Loss: 1.6174 | CE: 1.6174 | LR: 1.57e-04 | Epoch: 2/6 2026-06-22 10:00:00,747 [INFO] Step 580/1752 | Loss: 1.6212 | CE: 1.6212 | LR: 1.56e-04 | Epoch: 2/6 2026-06-22 10:03:14,964 [INFO] Step 584 | Eval Loss: 1.6822 2026-06-22 10:03:16,296 [INFO] Saved trainable-only checkpoint to runs/h200_llm2seq_phase1_warmup/best.pt (183 model tensors, stores_base_encoder_weights=False) 2026-06-22 10:03:16,854 [INFO] New best eval loss: 1.6822 2026-06-22 10:03:17,373 [INFO] Saved trainable-only checkpoint to runs/h200_llm2seq_phase1_warmup/epoch_003.pt (183 model tensors, stores_base_encoder_weights=False) 2026-06-22 10:03:18,048 [INFO] Uploading to HF: runs/h200_llm2seq_phase1_warmup/epoch_003.pt -> BienKieu/llm2seq-wikilingua/checkpoints/h200_phase1_warmup/epochs/epoch_003_step_584.pt 2026-06-22 10:03:46,617 [INFO] Uploading to HF: runs/h200_llm2seq_phase1_warmup/config.yaml -> BienKieu/llm2seq-wikilingua/checkpoints/h200_phase1_warmup/config.yaml 2026-06-22 10:03:46,719 [WARNING] No files have been modified since last commit. Skipping to prevent empty commit. 2026-06-22 10:03:46,777 [INFO] Uploading to HF: runs/h200_llm2seq_phase1_warmup/train.log -> BienKieu/llm2seq-wikilingua/checkpoints/h200_phase1_warmup/train.log 2026-06-22 10:03:47,082 [INFO] Uploading to HF: runs/h200_llm2seq_phase1_warmup/checkpoint_manifest.json -> BienKieu/llm2seq-wikilingua/checkpoints/h200_phase1_warmup/checkpoint_manifest.json 2026-06-22 10:03:47,151 [WARNING] No files have been modified since last commit. Skipping to prevent empty commit. 2026-06-22 10:03:47,613 [INFO] Uploading to HF: runs/h200_llm2seq_phase1_warmup/best.pt -> BienKieu/llm2seq-wikilingua/checkpoints/h200_phase1_warmup/best.pt 2026-06-22 10:03:53,073 [INFO] Deleted old local epoch checkpoint: runs/h200_llm2seq_phase1_warmup/epoch_002.pt 2026-06-22 10:03:54,710 [INFO] Saved trainable-only checkpoint to runs/h200_llm2seq_phase1_warmup/checkpoint_584.pt (183 model tensors, stores_base_encoder_weights=False) 2026-06-22 10:03:56,219 [INFO] Deleted old checkpoint: runs/h200_llm2seq_phase1_warmup/checkpoint_292.pt 2026-06-22 10:04:02,622 [INFO] Step 585/1752 | Loss: 1.5807 | CE: 1.5807 | LR: 1.55e-04 | Epoch: 3/6 2026-06-22 10:04:35,035 [INFO] Step 590/1752 | Loss: 1.5907 | CE: 1.5907 | LR: 1.55e-04 | Epoch: 3/6 2026-06-22 10:05:07,317 [INFO] Step 595/1752 | Loss: 1.5715 | CE: 1.5715 | LR: 1.54e-04 | Epoch: 3/6 2026-06-22 10:05:39,907 [INFO] Step 600/1752 | Loss: 1.4803 | CE: 1.4803 | LR: 1.53e-04 | Epoch: 3/6 2026-06-22 10:06:12,311 [INFO] Step 605/1752 | Loss: 1.5707 | CE: 1.5707 | LR: 1.52e-04 | Epoch: 3/6 2026-06-22 10:06:44,246 [INFO] Step 610/1752 | Loss: 1.5219 | CE: 1.5219 | LR: 1.51e-04 | Epoch: 3/6 2026-06-22 10:07:16,230 [INFO] Step 615/1752 | Loss: 1.5256 | CE: 1.5256 | LR: 1.51e-04 | Epoch: 3/6 2026-06-22 10:07:48,506 [INFO] Step 620/1752 | Loss: 1.5528 | CE: 1.5528 | LR: 1.50e-04 | Epoch: 3/6 2026-06-22 10:08:20,796 [INFO] Step 625/1752 | Loss: 1.4953 | CE: 1.4953 | LR: 1.49e-04 | Epoch: 3/6 2026-06-22 10:08:52,342 [INFO] Step 630/1752 | Loss: 1.5569 | CE: 1.5569 | LR: 1.48e-04 | Epoch: 3/6 2026-06-22 10:09:24,665 [INFO] Step 635/1752 | Loss: 1.5379 | CE: 1.5379 | LR: 1.47e-04 | Epoch: 3/6 2026-06-22 10:09:56,373 [INFO] Step 640/1752 | Loss: 1.5133 | CE: 1.5133 | LR: 1.47e-04 | Epoch: 3/6 2026-06-22 10:10:28,650 [INFO] Step 645/1752 | Loss: 1.5475 | CE: 1.5475 | LR: 1.46e-04 | Epoch: 3/6 2026-06-22 10:11:00,108 [INFO] Step 650/1752 | Loss: 1.5682 | CE: 1.5682 | LR: 1.45e-04 | Epoch: 3/6 2026-06-22 10:11:32,118 [INFO] Step 655/1752 | Loss: 1.4982 | CE: 1.4982 | LR: 1.44e-04 | Epoch: 3/6 2026-06-22 10:12:04,417 [INFO] Step 660/1752 | Loss: 1.5415 | CE: 1.5415 | LR: 1.43e-04 | Epoch: 3/6 2026-06-22 10:12:36,903 [INFO] Step 665/1752 | Loss: 1.5503 | CE: 1.5503 | LR: 1.42e-04 | Epoch: 3/6 2026-06-22 10:13:07,998 [INFO] Step 670/1752 | Loss: 1.5028 | CE: 1.5028 | LR: 1.42e-04 | Epoch: 3/6 2026-06-22 10:13:39,492 [INFO] Step 675/1752 | Loss: 1.5061 | CE: 1.5061 | LR: 1.41e-04 | Epoch: 3/6 2026-06-22 10:14:10,996 [INFO] Step 680/1752 | Loss: 1.5236 | CE: 1.5236 | LR: 1.40e-04 | Epoch: 3/6 2026-06-22 10:14:42,427 [INFO] Step 685/1752 | Loss: 1.5076 | CE: 1.5076 | LR: 1.39e-04 | Epoch: 3/6 2026-06-22 10:15:14,928 [INFO] Step 690/1752 | Loss: 1.5132 | CE: 1.5132 | LR: 1.38e-04 | Epoch: 3/6 2026-06-22 10:15:47,412 [INFO] Step 695/1752 | Loss: 1.4917 | CE: 1.4917 | LR: 1.37e-04 | Epoch: 3/6 2026-06-22 10:16:18,951 [INFO] Step 700/1752 | Loss: 1.4909 | CE: 1.4909 | LR: 1.36e-04 | Epoch: 3/6 2026-06-22 10:16:50,494 [INFO] Step 705/1752 | Loss: 1.5360 | CE: 1.5360 | LR: 1.36e-04 | Epoch: 3/6 2026-06-22 10:17:22,841 [INFO] Step 710/1752 | Loss: 1.5068 | CE: 1.5068 | LR: 1.35e-04 | Epoch: 3/6 2026-06-22 10:17:54,847 [INFO] Step 715/1752 | Loss: 1.4550 | CE: 1.4550 | LR: 1.34e-04 | Epoch: 3/6 2026-06-22 10:18:26,295 [INFO] Step 720/1752 | Loss: 1.5080 | CE: 1.5080 | LR: 1.33e-04 | Epoch: 3/6 2026-06-22 10:18:58,540 [INFO] Step 725/1752 | Loss: 1.4953 | CE: 1.4953 | LR: 1.32e-04 | Epoch: 3/6 2026-06-22 10:19:30,956 [INFO] Step 730/1752 | Loss: 1.5135 | CE: 1.5135 | LR: 1.31e-04 | Epoch: 3/6 2026-06-22 10:20:03,658 [INFO] Step 735/1752 | Loss: 1.4821 | CE: 1.4821 | LR: 1.30e-04 | Epoch: 3/6 2026-06-22 10:20:35,399 [INFO] Step 740/1752 | Loss: 1.4714 | CE: 1.4714 | LR: 1.29e-04 | Epoch: 3/6 2026-06-22 10:21:07,326 [INFO] Step 745/1752 | Loss: 1.4887 | CE: 1.4887 | LR: 1.29e-04 | Epoch: 3/6 2026-06-22 10:21:39,842 [INFO] Step 750/1752 | Loss: 1.3961 | CE: 1.3961 | LR: 1.28e-04 | Epoch: 3/6 2026-06-22 10:22:12,360 [INFO] Step 755/1752 | Loss: 1.4710 | CE: 1.4710 | LR: 1.27e-04 | Epoch: 3/6 2026-06-22 10:22:44,820 [INFO] Step 760/1752 | Loss: 1.4404 | CE: 1.4404 | LR: 1.26e-04 | Epoch: 3/6 2026-06-22 10:23:17,227 [INFO] Step 765/1752 | Loss: 1.4232 | CE: 1.4232 | LR: 1.25e-04 | Epoch: 3/6 2026-06-22 10:23:49,312 [INFO] Step 770/1752 | Loss: 1.4611 | CE: 1.4611 | LR: 1.24e-04 | Epoch: 3/6 2026-06-22 10:24:21,265 [INFO] Step 775/1752 | Loss: 1.4715 | CE: 1.4715 | LR: 1.23e-04 | Epoch: 3/6 2026-06-22 10:24:53,754 [INFO] Step 780/1752 | Loss: 1.4922 | CE: 1.4922 | LR: 1.22e-04 | Epoch: 3/6 2026-06-22 10:25:26,221 [INFO] Step 785/1752 | Loss: 1.4915 | CE: 1.4915 | LR: 1.21e-04 | Epoch: 3/6 2026-06-22 10:25:58,864 [INFO] Step 790/1752 | Loss: 1.4889 | CE: 1.4889 | LR: 1.21e-04 | Epoch: 3/6 2026-06-22 10:26:30,873 [INFO] Step 795/1752 | Loss: 1.4994 | CE: 1.4994 | LR: 1.20e-04 | Epoch: 3/6 2026-06-22 10:27:03,215 [INFO] Step 800/1752 | Loss: 1.4964 | CE: 1.4964 | LR: 1.19e-04 | Epoch: 3/6 2026-06-22 10:27:35,516 [INFO] Step 805/1752 | Loss: 1.4611 | CE: 1.4611 | LR: 1.18e-04 | Epoch: 3/6 2026-06-22 10:28:07,639 [INFO] Step 810/1752 | Loss: 1.5168 | CE: 1.5168 | LR: 1.17e-04 | Epoch: 3/6 2026-06-22 10:28:39,676 [INFO] Step 815/1752 | Loss: 1.4443 | CE: 1.4443 | LR: 1.16e-04 | Epoch: 3/6 2026-06-22 10:29:12,283 [INFO] Step 820/1752 | Loss: 1.4426 | CE: 1.4426 | LR: 1.15e-04 | Epoch: 3/6 2026-06-22 10:29:44,657 [INFO] Step 825/1752 | Loss: 1.5093 | CE: 1.5093 | LR: 1.14e-04 | Epoch: 3/6 2026-06-22 10:30:16,680 [INFO] Step 830/1752 | Loss: 1.4769 | CE: 1.4769 | LR: 1.13e-04 | Epoch: 3/6 2026-06-22 10:30:49,148 [INFO] Step 835/1752 | Loss: 1.4459 | CE: 1.4459 | LR: 1.12e-04 | Epoch: 3/6 2026-06-22 10:31:20,832 [INFO] Step 840/1752 | Loss: 1.4468 | CE: 1.4468 | LR: 1.11e-04 | Epoch: 3/6 2026-06-22 10:31:52,929 [INFO] Step 845/1752 | Loss: 1.4317 | CE: 1.4317 | LR: 1.11e-04 | Epoch: 3/6 2026-06-22 10:32:25,292 [INFO] Step 850/1752 | Loss: 1.4316 | CE: 1.4316 | LR: 1.10e-04 | Epoch: 3/6 2026-06-22 10:32:57,401 [INFO] Step 855/1752 | Loss: 1.4066 | CE: 1.4066 | LR: 1.09e-04 | Epoch: 3/6 2026-06-22 10:33:29,117 [INFO] Step 860/1752 | Loss: 1.4366 | CE: 1.4366 | LR: 1.08e-04 | Epoch: 3/6 2026-06-22 10:34:01,705 [INFO] Step 865/1752 | Loss: 1.4767 | CE: 1.4767 | LR: 1.07e-04 | Epoch: 3/6 2026-06-22 10:34:33,585 [INFO] Step 870/1752 | Loss: 1.4281 | CE: 1.4281 | LR: 1.06e-04 | Epoch: 3/6 2026-06-22 10:35:05,687 [INFO] Step 875/1752 | Loss: 1.4390 | CE: 1.4390 | LR: 1.05e-04 | Epoch: 4/6 2026-06-22 10:38:01,288 [INFO] Step 876 | Eval Loss: 1.4991 2026-06-22 10:38:02,589 [INFO] Saved trainable-only checkpoint to runs/h200_llm2seq_phase1_warmup/best.pt (183 model tensors, stores_base_encoder_weights=False) 2026-06-22 10:38:03,244 [INFO] New best eval loss: 1.4991 2026-06-22 10:38:04,001 [INFO] Saved trainable-only checkpoint to runs/h200_llm2seq_phase1_warmup/epoch_004.pt (183 model tensors, stores_base_encoder_weights=False) 2026-06-22 10:38:05,000 [INFO] Uploading to HF: runs/h200_llm2seq_phase1_warmup/epoch_004.pt -> BienKieu/llm2seq-wikilingua/checkpoints/h200_phase1_warmup/epochs/epoch_004_step_876.pt 2026-06-22 10:38:19,779 [INFO] Uploading to HF: runs/h200_llm2seq_phase1_warmup/config.yaml -> BienKieu/llm2seq-wikilingua/checkpoints/h200_phase1_warmup/config.yaml 2026-06-22 10:38:19,852 [WARNING] No files have been modified since last commit. Skipping to prevent empty commit. 2026-06-22 10:38:19,905 [INFO] Uploading to HF: runs/h200_llm2seq_phase1_warmup/train.log -> BienKieu/llm2seq-wikilingua/checkpoints/h200_phase1_warmup/train.log 2026-06-22 10:38:20,357 [INFO] Uploading to HF: runs/h200_llm2seq_phase1_warmup/checkpoint_manifest.json -> BienKieu/llm2seq-wikilingua/checkpoints/h200_phase1_warmup/checkpoint_manifest.json 2026-06-22 10:38:20,430 [WARNING] No files have been modified since last commit. Skipping to prevent empty commit. 2026-06-22 10:38:20,909 [INFO] Uploading to HF: runs/h200_llm2seq_phase1_warmup/best.pt -> BienKieu/llm2seq-wikilingua/checkpoints/h200_phase1_warmup/best.pt 2026-06-22 10:38:26,349 [INFO] Deleted old local epoch checkpoint: runs/h200_llm2seq_phase1_warmup/epoch_003.pt 2026-06-22 10:38:28,006 [INFO] Saved trainable-only checkpoint to runs/h200_llm2seq_phase1_warmup/checkpoint_876.pt (183 model tensors, stores_base_encoder_weights=False) 2026-06-22 10:38:29,526 [INFO] Deleted old checkpoint: runs/h200_llm2seq_phase1_warmup/checkpoint_584.pt 2026-06-22 10:38:55,202 [INFO] Step 880/1752 | Loss: 1.3208 | CE: 1.3208 | LR: 1.04e-04 | Epoch: 4/6 2026-06-22 10:39:27,384 [INFO] Step 885/1752 | Loss: 1.3451 | CE: 1.3451 | LR: 1.03e-04 | Epoch: 4/6 2026-06-22 10:39:59,517 [INFO] Step 890/1752 | Loss: 1.3093 | CE: 1.3093 | LR: 1.02e-04 | Epoch: 4/6 2026-06-22 10:40:31,225 [INFO] Step 895/1752 | Loss: 1.3153 | CE: 1.3153 | LR: 1.01e-04 | Epoch: 4/6 2026-06-22 10:41:03,651 [INFO] Step 900/1752 | Loss: 1.3002 | CE: 1.3002 | LR: 1.00e-04 | Epoch: 4/6 2026-06-22 10:41:36,213 [INFO] Step 905/1752 | Loss: 1.2925 | CE: 1.2925 | LR: 9.94e-05 | Epoch: 4/6 2026-06-22 10:42:07,557 [INFO] Step 910/1752 | Loss: 1.2964 | CE: 1.2964 | LR: 9.85e-05 | Epoch: 4/6 2026-06-22 10:42:39,617 [INFO] Step 915/1752 | Loss: 1.3178 | CE: 1.3178 | LR: 9.76e-05 | Epoch: 4/6 2026-06-22 10:43:12,059 [INFO] Step 920/1752 | Loss: 1.2906 | CE: 1.2906 | LR: 9.67e-05 | Epoch: 4/6 2026-06-22 10:43:44,047 [INFO] Step 925/1752 | Loss: 1.2991 | CE: 1.2991 | LR: 9.58e-05 | Epoch: 4/6 2026-06-22 10:44:16,664 [INFO] Step 930/1752 | Loss: 1.2975 | CE: 1.2975 | LR: 9.48e-05 | Epoch: 4/6 2026-06-22 10:44:48,410 [INFO] Step 935/1752 | Loss: 1.3179 | CE: 1.3179 | LR: 9.39e-05 | Epoch: 4/6 2026-06-22 10:45:20,821 [INFO] Step 940/1752 | Loss: 1.2666 | CE: 1.2666 | LR: 9.30e-05 | Epoch: 4/6 2026-06-22 10:45:52,927 [INFO] Step 945/1752 | Loss: 1.2895 | CE: 1.2895 | LR: 9.21e-05 | Epoch: 4/6 2026-06-22 10:46:25,139 [INFO] Step 950/1752 | Loss: 1.2928 | CE: 1.2928 | LR: 9.11e-05 | Epoch: 4/6 2026-06-22 10:46:56,462 [INFO] Step 955/1752 | Loss: 1.2796 | CE: 1.2796 | LR: 9.02e-05 | Epoch: 4/6 2026-06-22 10:47:28,984 [INFO] Step 960/1752 | Loss: 1.3273 | CE: 1.3273 | LR: 8.93e-05 | Epoch: 4/6 2026-06-22 10:48:01,525 [INFO] Step 965/1752 | Loss: 1.2906 | CE: 1.2906 | LR: 8.84e-05 | Epoch: 4/6 2026-06-22 10:48:33,918 [INFO] Step 970/1752 | Loss: 1.2841 | CE: 1.2841 | LR: 8.75e-05 | Epoch: 4/6 2026-06-22 10:49:05,905 [INFO] Step 975/1752 | Loss: 1.3040 | CE: 1.3040 | LR: 8.66e-05 | Epoch: 4/6 2026-06-22 10:49:38,320 [INFO] Step 980/1752 | Loss: 1.2384 | CE: 1.2384 | LR: 8.56e-05 | Epoch: 4/6 2026-06-22 10:50:10,730 [INFO] Step 985/1752 | Loss: 1.2552 | CE: 1.2552 | LR: 8.47e-05 | Epoch: 4/6 2026-06-22 10:50:42,624 [INFO] Step 990/1752 | Loss: 1.2306 | CE: 1.2306 | LR: 8.38e-05 | Epoch: 4/6 2026-06-22 10:51:14,740 [INFO] Step 995/1752 | Loss: 1.2751 | CE: 1.2751 | LR: 8.29e-05 | Epoch: 4/6 2026-06-22 10:51:47,002 [INFO] Step 1000/1752 | Loss: 1.3002 | CE: 1.3002 | LR: 8.20e-05 | Epoch: 4/6 2026-06-22 10:52:19,152 [INFO] Step 1005/1752 | Loss: 1.2430 | CE: 1.2430 | LR: 8.11e-05 | Epoch: 4/6 2026-06-22 10:52:51,784 [INFO] Step 1010/1752 | Loss: 1.2553 | CE: 1.2553 | LR: 8.02e-05 | Epoch: 4/6 2026-06-22 10:53:22,901 [INFO] Step 1015/1752 | Loss: 1.2665 | CE: 1.2665 | LR: 7.93e-05 | Epoch: 4/6 2026-06-22 10:53:55,310 [INFO] Step 1020/1752 | Loss: 1.2786 | CE: 1.2786 | LR: 7.84e-05 | Epoch: 4/6 2026-06-22 10:54:27,471 [INFO] Step 1025/1752 | Loss: 1.2790 | CE: 1.2790 | LR: 7.75e-05 | Epoch: 4/6 2026-06-22 10:54:59,826 [INFO] Step 1030/1752 | Loss: 1.2880 | CE: 1.2880 | LR: 7.66e-05 | Epoch: 4/6 2026-06-22 10:55:30,889 [INFO] Step 1035/1752 | Loss: 1.2758 | CE: 1.2758 | LR: 7.57e-05 | Epoch: 4/6 2026-06-22 10:56:03,380 [INFO] Step 1040/1752 | Loss: 1.3089 | CE: 1.3089 | LR: 7.48e-05 | Epoch: 4/6 2026-06-22 10:56:35,646 [INFO] Step 1045/1752 | Loss: 1.3001 | CE: 1.3001 | LR: 7.39e-05 | Epoch: 4/6 2026-06-22 10:57:08,136 [INFO] Step 1050/1752 | Loss: 1.2781 | CE: 1.2781 | LR: 7.30e-05 | Epoch: 4/6 2026-06-22 10:57:39,927 [INFO] Step 1055/1752 | Loss: 1.2994 | CE: 1.2994 | LR: 7.21e-05 | Epoch: 4/6 2026-06-22 10:58:11,875 [INFO] Step 1060/1752 | Loss: 1.2744 | CE: 1.2744 | LR: 7.12e-05 | Epoch: 4/6 2026-06-22 10:58:44,379 [INFO] Step 1065/1752 | Loss: 1.2803 | CE: 1.2803 | LR: 7.03e-05 | Epoch: 4/6 2026-06-22 10:59:17,136 [INFO] Step 1070/1752 | Loss: 1.2653 | CE: 1.2653 | LR: 6.95e-05 | Epoch: 4/6 2026-06-22 10:59:49,293 [INFO] Step 1075/1752 | Loss: 1.2764 | CE: 1.2764 | LR: 6.86e-05 | Epoch: 4/6 2026-06-22 11:00:21,722 [INFO] Step 1080/1752 | Loss: 1.2354 | CE: 1.2354 | LR: 6.77e-05 | Epoch: 4/6 2026-06-22 11:00:54,002 [INFO] Step 1085/1752 | Loss: 1.2708 | CE: 1.2708 | LR: 6.68e-05 | Epoch: 4/6 2026-06-22 11:01:26,375 [INFO] Step 1090/1752 | Loss: 1.2767 | CE: 1.2767 | LR: 6.60e-05 | Epoch: 4/6 2026-06-22 11:01:57,753 [INFO] Step 1095/1752 | Loss: 1.2249 | CE: 1.2249 | LR: 6.51e-05 | Epoch: 4/6 2026-06-22 11:02:30,044 [INFO] Step 1100/1752 | Loss: 1.2974 | CE: 1.2974 | LR: 6.42e-05 | Epoch: 4/6 2026-06-22 11:03:02,314 [INFO] Step 1105/1752 | Loss: 1.3114 | CE: 1.3114 | LR: 6.34e-05 | Epoch: 4/6 2026-06-22 11:03:34,175 [INFO] Step 1110/1752 | Loss: 1.2851 | CE: 1.2851 | LR: 6.25e-05 | Epoch: 4/6 2026-06-22 11:04:05,095 [INFO] Step 1115/1752 | Loss: 1.2511 | CE: 1.2511 | LR: 6.16e-05 | Epoch: 4/6 2026-06-22 11:04:35,700 [INFO] Step 1120/1752 | Loss: 1.2305 | CE: 1.2305 | LR: 6.08e-05 | Epoch: 4/6 2026-06-22 11:05:07,856 [INFO] Step 1125/1752 | Loss: 1.2823 | CE: 1.2823 | LR: 5.99e-05 | Epoch: 4/6 2026-06-22 11:05:39,853 [INFO] Step 1130/1752 | Loss: 1.2886 | CE: 1.2886 | LR: 5.91e-05 | Epoch: 4/6 2026-06-22 11:06:12,311 [INFO] Step 1135/1752 | Loss: 1.2709 | CE: 1.2709 | LR: 5.83e-05 | Epoch: 4/6 2026-06-22 11:06:43,557 [INFO] Step 1140/1752 | Loss: 1.2964 | CE: 1.2964 | LR: 5.74e-05 | Epoch: 4/6 2026-06-22 11:07:15,730 [INFO] Step 1145/1752 | Loss: 1.3247 | CE: 1.3247 | LR: 5.66e-05 | Epoch: 4/6 2026-06-22 11:07:48,247 [INFO] Step 1150/1752 | Loss: 1.2479 | CE: 1.2479 | LR: 5.58e-05 | Epoch: 4/6 2026-06-22 11:08:20,661 [INFO] Step 1155/1752 | Loss: 1.2405 | CE: 1.2405 | LR: 5.49e-05 | Epoch: 4/6 2026-06-22 11:08:52,225 [INFO] Step 1160/1752 | Loss: 1.2836 | CE: 1.2836 | LR: 5.41e-05 | Epoch: 4/6 2026-06-22 11:09:24,229 [INFO] Step 1165/1752 | Loss: 1.2984 | CE: 1.2984 | LR: 5.33e-05 | Epoch: 4/6 2026-06-22 11:12:33,133 [INFO] Step 1168 | Eval Loss: 1.4205 2026-06-22 11:12:34,412 [INFO] Saved trainable-only checkpoint to runs/h200_llm2seq_phase1_warmup/best.pt (183 model tensors, stores_base_encoder_weights=False) 2026-06-22 11:12:35,002 [INFO] New best eval loss: 1.4205 2026-06-22 11:12:35,759 [INFO] Saved trainable-only checkpoint to runs/h200_llm2seq_phase1_warmup/epoch_005.pt (183 model tensors, stores_base_encoder_weights=False) 2026-06-22 11:12:36,680 [INFO] Uploading to HF: runs/h200_llm2seq_phase1_warmup/epoch_005.pt -> BienKieu/llm2seq-wikilingua/checkpoints/h200_phase1_warmup/epochs/epoch_005_step_1168.pt 2026-06-22 11:12:55,585 [INFO] Uploading to HF: runs/h200_llm2seq_phase1_warmup/config.yaml -> BienKieu/llm2seq-wikilingua/checkpoints/h200_phase1_warmup/config.yaml 2026-06-22 11:12:55,657 [WARNING] No files have been modified since last commit. Skipping to prevent empty commit. 2026-06-22 11:12:55,706 [INFO] Uploading to HF: runs/h200_llm2seq_phase1_warmup/train.log -> BienKieu/llm2seq-wikilingua/checkpoints/h200_phase1_warmup/train.log 2026-06-22 11:12:56,121 [INFO] Uploading to HF: runs/h200_llm2seq_phase1_warmup/checkpoint_manifest.json -> BienKieu/llm2seq-wikilingua/checkpoints/h200_phase1_warmup/checkpoint_manifest.json 2026-06-22 11:12:56,194 [WARNING] No files have been modified since last commit. Skipping to prevent empty commit. 2026-06-22 11:12:56,665 [INFO] Uploading to HF: runs/h200_llm2seq_phase1_warmup/best.pt -> BienKieu/llm2seq-wikilingua/checkpoints/h200_phase1_warmup/best.pt 2026-06-22 11:13:01,849 [INFO] Deleted old local epoch checkpoint: runs/h200_llm2seq_phase1_warmup/epoch_004.pt 2026-06-22 11:13:03,528 [INFO] Saved trainable-only checkpoint to runs/h200_llm2seq_phase1_warmup/checkpoint_1168.pt (183 model tensors, stores_base_encoder_weights=False) 2026-06-22 11:13:05,031 [INFO] Deleted old checkpoint: runs/h200_llm2seq_phase1_warmup/checkpoint_876.pt 2026-06-22 11:13:17,552 [INFO] Step 1170/1752 | Loss: 1.1710 | CE: 1.1710 | LR: 5.25e-05 | Epoch: 5/6 2026-06-22 11:13:49,954 [INFO] Step 1175/1752 | Loss: 1.1079 | CE: 1.1079 | LR: 5.17e-05 | Epoch: 5/6 2026-06-22 11:14:22,231 [INFO] Step 1180/1752 | Loss: 1.1380 | CE: 1.1380 | LR: 5.09e-05 | Epoch: 5/6 2026-06-22 11:14:54,036 [INFO] Step 1185/1752 | Loss: 1.1377 | CE: 1.1377 | LR: 5.01e-05 | Epoch: 5/6 2026-06-22 11:15:26,334 [INFO] Step 1190/1752 | Loss: 1.1404 | CE: 1.1404 | LR: 4.93e-05 | Epoch: 5/6 2026-06-22 11:15:58,785 [INFO] Step 1195/1752 | Loss: 1.1272 | CE: 1.1272 | LR: 4.85e-05 | Epoch: 5/6 2026-06-22 11:16:31,217 [INFO] Step 1200/1752 | Loss: 1.1078 | CE: 1.1078 | LR: 4.77e-05 | Epoch: 5/6 2026-06-22 11:17:03,478 [INFO] Step 1205/1752 | Loss: 1.1711 | CE: 1.1711 | LR: 4.69e-05 | Epoch: 5/6 2026-06-22 11:17:35,240 [INFO] Step 1210/1752 | Loss: 1.1036 | CE: 1.1036 | LR: 4.61e-05 | Epoch: 5/6 2026-06-22 11:18:07,850 [INFO] Step 1215/1752 | Loss: 1.1526 | CE: 1.1526 | LR: 4.53e-05 | Epoch: 5/6 2026-06-22 11:18:40,284 [INFO] Step 1220/1752 | Loss: 1.1535 | CE: 1.1535 | LR: 4.46e-05 | Epoch: 5/6 2026-06-22 11:19:11,725 [INFO] Step 1225/1752 | Loss: 1.1374 | CE: 1.1374 | LR: 4.38e-05 | Epoch: 5/6 2026-06-22 11:19:43,985 [INFO] Step 1230/1752 | Loss: 1.1303 | CE: 1.1303 | LR: 4.30e-05 | Epoch: 5/6 2026-06-22 11:20:15,615 [INFO] Step 1235/1752 | Loss: 1.1377 | CE: 1.1377 | LR: 4.23e-05 | Epoch: 5/6 2026-06-22 11:20:48,069 [INFO] Step 1240/1752 | Loss: 1.1162 | CE: 1.1162 | LR: 4.15e-05 | Epoch: 5/6 2026-06-22 11:21:20,398 [INFO] Step 1245/1752 | Loss: 1.0996 | CE: 1.0996 | LR: 4.08e-05 | Epoch: 5/6 2026-06-22 11:21:53,156 [INFO] Step 1250/1752 | Loss: 1.1006 | CE: 1.1006 | LR: 4.00e-05 | Epoch: 5/6 2026-06-22 11:22:25,651 [INFO] Step 1255/1752 | Loss: 1.0847 | CE: 1.0847 | LR: 3.93e-05 | Epoch: 5/6 2026-06-22 11:22:58,066 [INFO] Step 1260/1752 | Loss: 1.1466 | CE: 1.1466 | LR: 3.86e-05 | Epoch: 5/6 2026-06-22 11:23:30,569 [INFO] Step 1265/1752 | Loss: 1.1018 | CE: 1.1018 | LR: 3.78e-05 | Epoch: 5/6 2026-06-22 11:24:02,781 [INFO] Step 1270/1752 | Loss: 1.1411 | CE: 1.1411 | LR: 3.71e-05 | Epoch: 5/6 2026-06-22 11:24:35,325 [INFO] Step 1275/1752 | Loss: 1.1057 | CE: 1.1057 | LR: 3.64e-05 | Epoch: 5/6 2026-06-22 11:25:07,872 [INFO] Step 1280/1752 | Loss: 1.0734 | CE: 1.0734 | LR: 3.57e-05 | Epoch: 5/6 2026-06-22 11:25:39,576 [INFO] Step 1285/1752 | Loss: 1.1409 | CE: 1.1409 | LR: 3.50e-05 | Epoch: 5/6 2026-06-22 11:26:11,408 [INFO] Step 1290/1752 | Loss: 1.1218 | CE: 1.1218 | LR: 3.43e-05 | Epoch: 5/6 2026-06-22 11:26:43,173 [INFO] Step 1295/1752 | Loss: 1.1195 | CE: 1.1195 | LR: 3.36e-05 | Epoch: 5/6 2026-06-22 11:27:15,002 [INFO] Step 1300/1752 | Loss: 1.0929 | CE: 1.0929 | LR: 3.29e-05 | Epoch: 5/6 2026-06-22 11:27:47,231 [INFO] Step 1305/1752 | Loss: 1.1359 | CE: 1.1359 | LR: 3.22e-05 | Epoch: 5/6 2026-06-22 11:28:19,372 [INFO] Step 1310/1752 | Loss: 1.1028 | CE: 1.1028 | LR: 3.15e-05 | Epoch: 5/6 2026-06-22 11:28:51,506 [INFO] Step 1315/1752 | Loss: 1.1616 | CE: 1.1616 | LR: 3.09e-05 | Epoch: 5/6 2026-06-22 11:29:24,146 [INFO] Step 1320/1752 | Loss: 1.0910 | CE: 1.0910 | LR: 3.02e-05 | Epoch: 5/6 2026-06-22 11:29:56,279 [INFO] Step 1325/1752 | Loss: 1.1270 | CE: 1.1270 | LR: 2.96e-05 | Epoch: 5/6 2026-06-22 11:30:28,347 [INFO] Step 1330/1752 | Loss: 1.1062 | CE: 1.1062 | LR: 2.89e-05 | Epoch: 5/6 2026-06-22 11:31:00,068 [INFO] Step 1335/1752 | Loss: 1.1213 | CE: 1.1213 | LR: 2.83e-05 | Epoch: 5/6 2026-06-22 11:31:31,829 [INFO] Step 1340/1752 | Loss: 1.1268 | CE: 1.1268 | LR: 2.76e-05 | Epoch: 5/6 2026-06-22 11:32:04,052 [INFO] Step 1345/1752 | Loss: 1.0880 | CE: 1.0880 | LR: 2.70e-05 | Epoch: 5/6 2026-06-22 11:32:35,500 [INFO] Step 1350/1752 | Loss: 1.0946 | CE: 1.0946 | LR: 2.63e-05 | Epoch: 5/6 2026-06-22 11:33:07,493 [INFO] Step 1355/1752 | Loss: 1.1400 | CE: 1.1400 | LR: 2.57e-05 | Epoch: 5/6 2026-06-22 11:33:39,915 [INFO] Step 1360/1752 | Loss: 1.1039 | CE: 1.1039 | LR: 2.51e-05 | Epoch: 5/6 2026-06-22 11:34:12,349 [INFO] Step 1365/1752 | Loss: 1.1223 | CE: 1.1223 | LR: 2.45e-05 | Epoch: 5/6 2026-06-22 11:34:44,334 [INFO] Step 1370/1752 | Loss: 1.1656 | CE: 1.1656 | LR: 2.39e-05 | Epoch: 5/6 2026-06-22 11:35:16,775 [INFO] Step 1375/1752 | Loss: 1.0992 | CE: 1.0992 | LR: 2.33e-05 | Epoch: 5/6 2026-06-22 11:35:48,317 [INFO] Step 1380/1752 | Loss: 1.1532 | CE: 1.1532 | LR: 2.27e-05 | Epoch: 5/6 2026-06-22 11:36:20,717 [INFO] Step 1385/1752 | Loss: 1.1202 | CE: 1.1202 | LR: 2.21e-05 | Epoch: 5/6 2026-06-22 11:36:53,023 [INFO] Step 1390/1752 | Loss: 1.1053 | CE: 1.1053 | LR: 2.16e-05 | Epoch: 5/6 2026-06-22 11:37:25,051 [INFO] Step 1395/1752 | Loss: 1.1010 | CE: 1.1010 | LR: 2.10e-05 | Epoch: 5/6 2026-06-22 11:37:57,315 [INFO] Step 1400/1752 | Loss: 1.1209 | CE: 1.1209 | LR: 2.04e-05 | Epoch: 5/6 2026-06-22 11:38:29,676 [INFO] Step 1405/1752 | Loss: 1.1421 | CE: 1.1421 | LR: 2.00e-05 | Epoch: 5/6 2026-06-22 11:39:01,454 [INFO] Step 1410/1752 | Loss: 1.1297 | CE: 1.1297 | LR: 2.00e-05 | Epoch: 5/6 2026-06-22 11:39:33,832 [INFO] Step 1415/1752 | Loss: 1.0863 | CE: 1.0863 | LR: 2.00e-05 | Epoch: 5/6 2026-06-22 11:40:06,413 [INFO] Step 1420/1752 | Loss: 1.0922 | CE: 1.0922 | LR: 2.00e-05 | Epoch: 5/6 2026-06-22 11:40:38,762 [INFO] Step 1425/1752 | Loss: 1.0800 | CE: 1.0800 | LR: 2.00e-05 | Epoch: 5/6 2026-06-22 11:41:10,779 [INFO] Step 1430/1752 | Loss: 1.1099 | CE: 1.1099 | LR: 2.00e-05 | Epoch: 5/6 2026-06-22 11:41:42,977 [INFO] Step 1435/1752 | Loss: 1.1217 | CE: 1.1217 | LR: 2.00e-05 | Epoch: 5/6 2026-06-22 11:42:15,317 [INFO] Step 1440/1752 | Loss: 1.1200 | CE: 1.1200 | LR: 2.00e-05 | Epoch: 5/6 2026-06-22 11:42:47,501 [INFO] Step 1445/1752 | Loss: 1.0869 | CE: 1.0869 | LR: 2.00e-05 | Epoch: 5/6 2026-06-22 11:43:19,883 [INFO] Step 1450/1752 | Loss: 1.0718 | CE: 1.0718 | LR: 2.00e-05 | Epoch: 5/6 2026-06-22 11:43:51,710 [INFO] Step 1455/1752 | Loss: 1.1315 | CE: 1.1315 | LR: 2.00e-05 | Epoch: 5/6 2026-06-22 11:44:24,884 [INFO] Step 1460/1752 | Loss: 1.0639 | CE: 1.0639 | LR: 2.00e-05 | Epoch: 6/6 2026-06-22 11:47:14,015 [INFO] Step 1460 | Eval Loss: 1.4004 2026-06-22 11:47:15,318 [INFO] Saved trainable-only checkpoint to runs/h200_llm2seq_phase1_warmup/best.pt (183 model tensors, stores_base_encoder_weights=False) 2026-06-22 11:47:15,972 [INFO] New best eval loss: 1.4004 2026-06-22 11:47:16,723 [INFO] Saved trainable-only checkpoint to runs/h200_llm2seq_phase1_warmup/epoch_006.pt (183 model tensors, stores_base_encoder_weights=False) 2026-06-22 11:47:17,723 [INFO] Uploading to HF: runs/h200_llm2seq_phase1_warmup/epoch_006.pt -> BienKieu/llm2seq-wikilingua/checkpoints/h200_phase1_warmup/epochs/epoch_006_step_1460.pt 2026-06-22 11:47:34,463 [INFO] Uploading to HF: runs/h200_llm2seq_phase1_warmup/config.yaml -> BienKieu/llm2seq-wikilingua/checkpoints/h200_phase1_warmup/config.yaml 2026-06-22 11:47:34,539 [WARNING] No files have been modified since last commit. Skipping to prevent empty commit. 2026-06-22 11:47:34,596 [INFO] Uploading to HF: runs/h200_llm2seq_phase1_warmup/train.log -> BienKieu/llm2seq-wikilingua/checkpoints/h200_phase1_warmup/train.log 2026-06-22 11:47:34,980 [INFO] Uploading to HF: runs/h200_llm2seq_phase1_warmup/checkpoint_manifest.json -> BienKieu/llm2seq-wikilingua/checkpoints/h200_phase1_warmup/checkpoint_manifest.json 2026-06-22 11:47:35,048 [WARNING] No files have been modified since last commit. Skipping to prevent empty commit. 2026-06-22 11:47:35,550 [INFO] Uploading to HF: runs/h200_llm2seq_phase1_warmup/best.pt -> BienKieu/llm2seq-wikilingua/checkpoints/h200_phase1_warmup/best.pt 2026-06-22 11:47:41,409 [INFO] Deleted old local epoch checkpoint: runs/h200_llm2seq_phase1_warmup/epoch_005.pt 2026-06-22 11:47:43,085 [INFO] Saved trainable-only checkpoint to runs/h200_llm2seq_phase1_warmup/checkpoint_1460.pt (183 model tensors, stores_base_encoder_weights=False) 2026-06-22 11:47:44,569 [INFO] Deleted old checkpoint: runs/h200_llm2seq_phase1_warmup/checkpoint_1168.pt 2026-06-22 11:48:17,377 [INFO] Step 1465/1752 | Loss: 1.0548 | CE: 1.0548 | LR: 2.00e-05 | Epoch: 6/6 2026-06-22 11:48:49,681 [INFO] Step 1470/1752 | Loss: 1.0112 | CE: 1.0112 | LR: 2.00e-05 | Epoch: 6/6 2026-06-22 11:49:22,074 [INFO] Step 1475/1752 | Loss: 1.0250 | CE: 1.0250 | LR: 2.00e-05 | Epoch: 6/6 2026-06-22 11:49:53,267 [INFO] Step 1480/1752 | Loss: 1.0555 | CE: 1.0555 | LR: 2.00e-05 | Epoch: 6/6 2026-06-22 11:50:25,655 [INFO] Step 1485/1752 | Loss: 1.0306 | CE: 1.0306 | LR: 2.00e-05 | Epoch: 6/6 2026-06-22 11:50:58,032 [INFO] Step 1490/1752 | Loss: 1.0392 | CE: 1.0392 | LR: 2.00e-05 | Epoch: 6/6 2026-06-22 11:51:30,409 [INFO] Step 1495/1752 | Loss: 1.0266 | CE: 1.0266 | LR: 2.00e-05 | Epoch: 6/6 2026-06-22 11:52:02,746 [INFO] Step 1500/1752 | Loss: 1.0639 | CE: 1.0639 | LR: 2.00e-05 | Epoch: 6/6 2026-06-22 11:52:35,279 [INFO] Step 1505/1752 | Loss: 1.0579 | CE: 1.0579 | LR: 2.00e-05 | Epoch: 6/6 2026-06-22 11:53:06,920 [INFO] Step 1510/1752 | Loss: 1.0667 | CE: 1.0667 | LR: 2.00e-05 | Epoch: 6/6 2026-06-22 11:53:37,967 [INFO] Step 1515/1752 | Loss: 1.0225 | CE: 1.0225 | LR: 2.00e-05 | Epoch: 6/6 2026-06-22 11:54:09,620 [INFO] Step 1520/1752 | Loss: 1.0392 | CE: 1.0392 | LR: 2.00e-05 | Epoch: 6/6 2026-06-22 11:54:41,949 [INFO] Step 1525/1752 | Loss: 1.0473 | CE: 1.0473 | LR: 2.00e-05 | Epoch: 6/6 2026-06-22 11:55:14,498 [INFO] Step 1530/1752 | Loss: 1.0463 | CE: 1.0463 | LR: 2.00e-05 | Epoch: 6/6 2026-06-22 11:55:46,835 [INFO] Step 1535/1752 | Loss: 1.0403 | CE: 1.0403 | LR: 2.00e-05 | Epoch: 6/6 2026-06-22 11:56:19,478 [INFO] Step 1540/1752 | Loss: 0.9778 | CE: 0.9778 | LR: 2.00e-05 | Epoch: 6/6 2026-06-22 11:56:51,114 [INFO] Step 1545/1752 | Loss: 1.0135 | CE: 1.0135 | LR: 2.00e-05 | Epoch: 6/6 2026-06-22 11:57:23,674 [INFO] Step 1550/1752 | Loss: 0.9878 | CE: 0.9878 | LR: 2.00e-05 | Epoch: 6/6 2026-06-22 11:57:55,618 [INFO] Step 1555/1752 | Loss: 1.0698 | CE: 1.0698 | LR: 2.00e-05 | Epoch: 6/6 2026-06-22 11:58:28,001 [INFO] Step 1560/1752 | Loss: 1.0374 | CE: 1.0374 | LR: 2.00e-05 | Epoch: 6/6 2026-06-22 11:58:59,759 [INFO] Step 1565/1752 | Loss: 1.0289 | CE: 1.0289 | LR: 2.00e-05 | Epoch: 6/6 2026-06-22 11:59:31,995 [INFO] Step 1570/1752 | Loss: 1.0166 | CE: 1.0166 | LR: 2.00e-05 | Epoch: 6/6 2026-06-22 12:00:04,652 [INFO] Step 1575/1752 | Loss: 0.9962 | CE: 0.9962 | LR: 2.00e-05 | Epoch: 6/6 2026-06-22 12:00:36,632 [INFO] Step 1580/1752 | Loss: 1.0050 | CE: 1.0050 | LR: 2.00e-05 | Epoch: 6/6 2026-06-22 12:01:09,079 [INFO] Step 1585/1752 | Loss: 1.0555 | CE: 1.0555 | LR: 2.00e-05 | Epoch: 6/6 2026-06-22 12:01:41,115 [INFO] Step 1590/1752 | Loss: 1.0473 | CE: 1.0473 | LR: 2.00e-05 | Epoch: 6/6 2026-06-22 12:02:13,460 [INFO] Step 1595/1752 | Loss: 1.0337 | CE: 1.0337 | LR: 2.00e-05 | Epoch: 6/6 2026-06-22 12:02:46,105 [INFO] Step 1600/1752 | Loss: 1.0341 | CE: 1.0341 | LR: 2.00e-05 | Epoch: 6/6 2026-06-22 12:03:18,605 [INFO] Step 1605/1752 | Loss: 1.0351 | CE: 1.0351 | LR: 2.00e-05 | Epoch: 6/6 2026-06-22 12:03:51,122 [INFO] Step 1610/1752 | Loss: 1.0601 | CE: 1.0601 | LR: 2.00e-05 | Epoch: 6/6 2026-06-22 12:04:23,437 [INFO] Step 1615/1752 | Loss: 1.0112 | CE: 1.0112 | LR: 2.00e-05 | Epoch: 6/6 2026-06-22 12:04:54,204 [INFO] Step 1620/1752 | Loss: 0.9945 | CE: 0.9945 | LR: 2.00e-05 | Epoch: 6/6 2026-06-22 12:05:26,018 [INFO] Step 1625/1752 | Loss: 0.9954 | CE: 0.9954 | LR: 2.00e-05 | Epoch: 6/6 2026-06-22 12:05:58,257 [INFO] Step 1630/1752 | Loss: 1.0626 | CE: 1.0626 | LR: 2.00e-05 | Epoch: 6/6 2026-06-22 12:06:30,219 [INFO] Step 1635/1752 | Loss: 1.0575 | CE: 1.0575 | LR: 2.00e-05 | Epoch: 6/6 2026-06-22 12:07:02,244 [INFO] Step 1640/1752 | Loss: 1.0771 | CE: 1.0771 | LR: 2.00e-05 | Epoch: 6/6 2026-06-22 12:07:34,356 [INFO] Step 1645/1752 | Loss: 1.0308 | CE: 1.0308 | LR: 2.00e-05 | Epoch: 6/6 2026-06-22 12:08:05,826 [INFO] Step 1650/1752 | Loss: 1.0528 | CE: 1.0528 | LR: 2.00e-05 | Epoch: 6/6 2026-06-22 12:08:37,302 [INFO] Step 1655/1752 | Loss: 1.0564 | CE: 1.0564 | LR: 2.00e-05 | Epoch: 6/6 2026-06-22 12:09:09,126 [INFO] Step 1660/1752 | Loss: 1.0474 | CE: 1.0474 | LR: 2.00e-05 | Epoch: 6/6 2026-06-22 12:09:41,214 [INFO] Step 1665/1752 | Loss: 1.0290 | CE: 1.0290 | LR: 2.00e-05 | Epoch: 6/6 2026-06-22 12:10:12,589 [INFO] Step 1670/1752 | Loss: 1.0312 | CE: 1.0312 | LR: 2.00e-05 | Epoch: 6/6 2026-06-22 12:10:44,157 [INFO] Step 1675/1752 | Loss: 1.0436 | CE: 1.0436 | LR: 2.00e-05 | Epoch: 6/6 2026-06-22 12:11:16,588 [INFO] Step 1680/1752 | Loss: 1.0308 | CE: 1.0308 | LR: 2.00e-05 | Epoch: 6/6 2026-06-22 12:11:48,689 [INFO] Step 1685/1752 | Loss: 1.0372 | CE: 1.0372 | LR: 2.00e-05 | Epoch: 6/6 2026-06-22 12:12:20,543 [INFO] Step 1690/1752 | Loss: 1.0499 | CE: 1.0499 | LR: 2.00e-05 | Epoch: 6/6 2026-06-22 12:12:52,898 [INFO] Step 1695/1752 | Loss: 1.0536 | CE: 1.0536 | LR: 2.00e-05 | Epoch: 6/6 2026-06-22 12:13:25,667 [INFO] Step 1700/1752 | Loss: 1.0238 | CE: 1.0238 | LR: 2.00e-05 | Epoch: 6/6 2026-06-22 12:13:57,810 [INFO] Step 1705/1752 | Loss: 1.0288 | CE: 1.0288 | LR: 2.00e-05 | Epoch: 6/6 2026-06-22 12:14:30,176 [INFO] Step 1710/1752 | Loss: 1.0280 | CE: 1.0280 | LR: 2.00e-05 | Epoch: 6/6 2026-06-22 12:15:01,775 [INFO] Step 1715/1752 | Loss: 1.0046 | CE: 1.0046 | LR: 2.00e-05 | Epoch: 6/6 2026-06-22 12:15:33,897 [INFO] Step 1720/1752 | Loss: 1.0090 | CE: 1.0090 | LR: 2.00e-05 | Epoch: 6/6 2026-06-22 12:16:06,197 [INFO] Step 1725/1752 | Loss: 1.0284 | CE: 1.0284 | LR: 2.00e-05 | Epoch: 6/6 2026-06-22 12:16:38,518 [INFO] Step 1730/1752 | Loss: 1.0160 | CE: 1.0160 | LR: 2.00e-05 | Epoch: 6/6 2026-06-22 12:17:10,636 [INFO] Step 1735/1752 | Loss: 1.0689 | CE: 1.0689 | LR: 2.00e-05 | Epoch: 6/6 2026-06-22 12:17:42,747 [INFO] Step 1740/1752 | Loss: 1.0231 | CE: 1.0231 | LR: 2.00e-05 | Epoch: 6/6 2026-06-22 12:18:14,361 [INFO] Step 1745/1752 | Loss: 1.0283 | CE: 1.0283 | LR: 2.00e-05 | Epoch: 6/6 2026-06-22 12:18:47,178 [INFO] Step 1750/1752 | Loss: 0.9860 | CE: 0.9860 | LR: 2.00e-05 | Epoch: 7/6 2026-06-22 12:21:48,613 [INFO] Step 1752 | Eval Loss: 1.4091 2026-06-22 12:21:49,617 [INFO] Saved trainable-only checkpoint to runs/h200_llm2seq_phase1_warmup/epoch_007.pt (183 model tensors, stores_base_encoder_weights=False) 2026-06-22 12:21:50,710 [INFO] Uploading to HF: runs/h200_llm2seq_phase1_warmup/epoch_007.pt -> BienKieu/llm2seq-wikilingua/checkpoints/h200_phase1_warmup/epochs/epoch_007_step_1752.pt 2026-06-22 12:22:04,972 [INFO] Uploading to HF: runs/h200_llm2seq_phase1_warmup/config.yaml -> BienKieu/llm2seq-wikilingua/checkpoints/h200_phase1_warmup/config.yaml 2026-06-22 12:22:05,053 [WARNING] No files have been modified since last commit. Skipping to prevent empty commit. 2026-06-22 12:22:05,107 [INFO] Uploading to HF: runs/h200_llm2seq_phase1_warmup/train.log -> BienKieu/llm2seq-wikilingua/checkpoints/h200_phase1_warmup/train.log 2026-06-22 12:22:05,615 [INFO] Uploading to HF: runs/h200_llm2seq_phase1_warmup/checkpoint_manifest.json -> BienKieu/llm2seq-wikilingua/checkpoints/h200_phase1_warmup/checkpoint_manifest.json 2026-06-22 12:22:05,697 [WARNING] No files have been modified since last commit. Skipping to prevent empty commit. 2026-06-22 12:22:06,189 [INFO] Uploading to HF: runs/h200_llm2seq_phase1_warmup/best.pt -> BienKieu/llm2seq-wikilingua/checkpoints/h200_phase1_warmup/best.pt 2026-06-22 12:22:10,284 [WARNING] No files have been modified since last commit. Skipping to prevent empty commit. 2026-06-22 12:22:10,368 [INFO] Deleted old local epoch checkpoint: runs/h200_llm2seq_phase1_warmup/epoch_006.pt 2026-06-22 12:22:12,049 [INFO] Saved trainable-only checkpoint to runs/h200_llm2seq_phase1_warmup/checkpoint_1752.pt (183 model tensors, stores_base_encoder_weights=False) 2026-06-22 12:22:13,668 [INFO] Deleted old checkpoint: runs/h200_llm2seq_phase1_warmup/checkpoint_1460.pt 2026-06-22 12:22:14,476 [INFO] Saved trainable-only checkpoint to runs/h200_llm2seq_phase1_warmup/final.pt (183 model tensors, stores_base_encoder_weights=False) 2026-06-22 12:22:15,363 [INFO] Uploading to HF: runs/h200_llm2seq_phase1_warmup/best.pt -> BienKieu/llm2seq-wikilingua/checkpoints/h200_phase1_warmup/best.pt 2026-06-22 12:22:19,518 [WARNING] No files have been modified since last commit. Skipping to prevent empty commit. 2026-06-22 12:22:20,009 [INFO] Uploading to HF: runs/h200_llm2seq_phase1_warmup/final.pt -> BienKieu/llm2seq-wikilingua/checkpoints/h200_phase1_warmup/final.pt 2026-06-22 12:22:28,145 [INFO] Uploading to HF: runs/h200_llm2seq_phase1_warmup/config.yaml -> BienKieu/llm2seq-wikilingua/checkpoints/h200_phase1_warmup/config.yaml 2026-06-22 12:22:28,213 [WARNING] No files have been modified since last commit. Skipping to prevent empty commit. 2026-06-22 12:22:28,279 [INFO] Uploading to HF: runs/h200_llm2seq_phase1_warmup/train.log -> BienKieu/llm2seq-wikilingua/checkpoints/h200_phase1_warmup/train.log