Upload training run run_20260408_151821

ParrotLLM end-of-training sync for local run runs/big_run/exp_c/run_20260408_151821.

.gitattributes

@@ -60,3 +60,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.webm filter=lfs diff=lfs merge=lfs -text
 runs/big_run/exp_c/run_20260408_124138/training_plots.pdf filter=lfs diff=lfs merge=lfs -text
 runs/big_run/exp_c/run_20260408_135803/training_plots.pdf filter=lfs diff=lfs merge=lfs -text
+runs/big_run/exp_c/run_20260408_151821/training_plots.pdf filter=lfs diff=lfs merge=lfs -text

runs/big_run/exp_c/run_20260408_151821/checkpoints/best_loss_3p6237_epoch_0001_step_0003000.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a23e12493e4ab6bfaa451e7eb772f6802f6a92716957e11a4cf4a44bd80d7349
+size 479869500

runs/big_run/exp_c/run_20260408_151821/checkpoints/best_loss_3p6458_epoch_0001_step_0002900.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d27bbad30a976e269530977b587f2533fd8d37ccdd55f3b4dae2d50aa4c5b7cb
+size 479869500

runs/big_run/exp_c/run_20260408_151821/checkpoints/best_loss_3p6690_epoch_0001_step_0002800.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:daa2f96b464a147540e41095ed3dbeac22921eba98f8520ca8d5fa690418de66
+size 479869500

runs/big_run/exp_c/run_20260408_151821/checkpoints/best_loss_3p6924_epoch_0001_step_0002700.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6d87ff0bc85e2294cf1d2db7a2664209983c0e68eef9e31112cfd2e1720d9063
+size 479869500

runs/big_run/exp_c/run_20260408_151821/checkpoints/best_loss_3p7192_epoch_0001_step_0002600.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:40845fd320997881dd198029f2979cd613820c72fb34708a05ae534633011f61
+size 479869500

runs/big_run/exp_c/run_20260408_151821/checkpoints/best_loss_3p7442_epoch_0001_step_0002500.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:96459810932dd958f42a06bd2bc9ba839f4fa71fed7c3d8cd662a362357f594f
+size 479869500

runs/big_run/exp_c/run_20260408_151821/checkpoints/best_loss_3p7509_epoch_0001_step_0002400.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5c2d22aaffd63a7c674045bff0afb3a8b54c3a81bedf8b427971e76b5de91907
+size 479869500

runs/big_run/exp_c/run_20260408_151821/checkpoints/best_loss_3p7645_epoch_0001_step_0002300.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:58d93289d0633e936040218df0a25efb91956038a12f792e2adf099bb9f1cac7
+size 479869500

runs/big_run/exp_c/run_20260408_151821/checkpoints/best_loss_3p7710_epoch_0001_step_0002200.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:078a52c67461d37c5ab2cc85309714acf102c5252b9abce56a1f6a7e54fda2a1
+size 479869500

runs/big_run/exp_c/run_20260408_151821/checkpoints/best_loss_3p7856_epoch_0001_step_0002100.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:03187d8f6bc1573a83d142861070e97410a092c9dffa746c30ba1771e3830d33
+size 479869500

runs/big_run/exp_c/run_20260408_151821/checkpoints/last_epoch_0000_step_0001000.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7f7927c8c1c8b05f3836196d5b62bbfe7a9afd8f3db8ede803caf72851ecd7b7
+size 479860488

runs/big_run/exp_c/run_20260408_151821/checkpoints/last_epoch_0000_step_0001250.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1939f50702e78c9817fa596ef7b0ca9f051985850a409b539dfbb45b54fe5dfb
+size 479860488

runs/big_run/exp_c/run_20260408_151821/checkpoints/last_epoch_0000_step_0001500.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:eacc3b0e0dbe322d03d78f6f931b9918d440795d5c97c8d95258fe8e79bd47c9
+size 479860488

runs/big_run/exp_c/run_20260408_151821/checkpoints/last_epoch_0000_step_0001750.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7ae59384476b182cd09a641bfc43c160295d075287b0a994b7fd2e5fc9e0b0f5
+size 479860488

runs/big_run/exp_c/run_20260408_151821/checkpoints/last_epoch_0001_step_0002000.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:564726384ccb2efc725002519d4105dc15ae186373d70b600929cd53898aac90
+size 479860488

runs/big_run/exp_c/run_20260408_151821/checkpoints/last_epoch_0001_step_0002250.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dbe5dccdc5b7a6c3a74726df728a82c693a1ac4ca4aff0b911d8d106283bd798
+size 479860488

runs/big_run/exp_c/run_20260408_151821/checkpoints/last_epoch_0001_step_0002500.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:027c76cb9bfb331fd7d5d3ea07c4769dfbba26ea2d13dd45e9f1b3334d09627e
+size 479860488

runs/big_run/exp_c/run_20260408_151821/checkpoints/last_epoch_0001_step_0002750.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:af1b0088cd06c5ffa0066b39dea2ccd2d3eee3a55a86a3c17cc62e460a697eaf
+size 479860488

runs/big_run/exp_c/run_20260408_151821/checkpoints/last_epoch_0001_step_0003000.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:540255e667835273b1c914238abb3ea2d21514b7d30ef3942d68f3acb3562d32
+size 479860488

runs/big_run/exp_c/run_20260408_151821/checkpoints/last_epoch_0001_step_0003052.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b69a18e8b30732ce6e773f2c92f16914cbfce51878ac6b72e9766ba40102c4ed
+size 479860488

runs/big_run/exp_c/run_20260408_151821/config.json

@@ -0,0 +1,88 @@
+{
+  "preprocess": null,
+  "model": {
+    "vocab_size": 50258,
+    "pad_token_id": 50257,
+    "bos_token_id": 50256,
+    "eos_token_id": 50256,
+    "d_model": 384,
+    "n_layers": 14,
+    "n_heads": 6,
+    "d_ff": 768,
+    "context_length": 1024,
+    "bias": false,
+    "dropout": 0.005,
+    "rope_theta": 10000.0,
+    "gradient_checkpointing": false
+  },
+  "training": {
+    "device": "auto",
+    "train_bin": "data/exp_c/train.bin",
+    "val_bin": "data/exp_a/val.bin",
+    "num_workers": 4,
+    "pin_memory": true,
+    "batch_size": 32,
+    "gradient_accumulation_steps": 1,
+    "train_samples_per_epoch": null,
+    "val_sequence_stride": null,
+    "learning_rate": 0.001305,
+    "min_lr": 0.0001305,
+    "weight_decay": 0.063,
+    "beta1": 0.9,
+    "beta2": 0.9978,
+    "grad_clip": 1.786,
+    "warmup_steps": 305,
+    "max_steps": 3052,
+    "lr_schedule": "wsd",
+    "lr_decay_ratio": 0.179,
+    "z_loss_coeff": 0.000275,
+    "save_every": 250,
+    "eval_every": 100,
+    "checkpoint_dir": "checkpoints",
+    "keep_last_checkpoints": 10,
+    "keep_best_checkpoints": 10,
+    "early_stopping_patience": 0,
+    "early_stopping_min_delta": 0.001,
+    "runs_dir": "runs/big_run/exp_c",
+    "log_every": 10,
+    "hf_upload": {
+      "repo_id": "ParrotLabs/Preprocessed",
+      "repo_type": "dataset",
+      "path_in_repo": "",
+      "private": null
+    },
+    "compile": true
+  },
+  "tune": null,
+  "logging": {
+    "console_level": "INFO",
+    "file_level": "DEBUG",
+    "components": {
+      "data_preprocessing": "DEBUG",
+      "model_initialization": "INFO",
+      "training": "INFO"
+    },
+    "profiler": null
+  },
+  "eval": {
+    "device": "auto",
+    "batch_size": 32,
+    "max_sequences": 512,
+    "datasets": [
+      {
+        "name": "wikitext",
+        "path": "wikitext",
+        "subset": "wikitext-103-raw-v1",
+        "split": "test"
+      },
+      {
+        "name": "exp_c_val",
+        "path": "data/exp_c/val.bin",
+        "subset": null,
+        "split": null
+      }
+    ]
+  },
+  "inference": null,
+  "chat": null
+}

runs/big_run/exp_c/run_20260408_151821/train.log

@@ -0,0 +1,721 @@
+2026-04-08 15:18:22 - INFO - parrotllm - Logging initialised -> runs/big_run/exp_c/run_20260408_151821/train.log
+2026-04-08 15:18:22 - INFO - parrotllm.training - device=cuda:0 | rank=0 | world_size=8 | distributed=yes
+2026-04-08 15:18:22 - INFO - parrotllm.training - checkpoint directory=/home/user/code/ParrotLLM/runs/big_run/exp_c/run_20260408_151821/checkpoints | keep_last=10 | keep_best=10
+2026-04-08 15:18:22 - INFO - parrotllm.training - train windows=random-overlapping | tokens=508793600 | valid_starts=508792576 | sampled_windows_per_epoch=496384
+2026-04-08 15:18:22 - INFO - parrotllm.training - val windows=sequential | tokens=8646900 | stride=1024 | windows=8444
+2026-04-08 15:18:22 - INFO - parrotllm.training - compiling model with torch.compile...
+2026-04-08 15:18:24 - INFO - parrotllm.training - 
+============================================================
+MODEL ARCHITECTURE SUMMARY
+============================================================
+
+Configuration:
+  Vocab size: 50258
+  Block size (context): 1024
+  Layers: 14
+  Heads: 6
+  Embedding dim: 384
+  FFN hidden dim: 768
+  Dropout: 0.005
+  Bias: False
+
+Parameters (unique, weight-tied layers counted once):
+  Total: 39,966,592 (39.97M)
+  Trainable: 39,966,592 (39.97M)
+  Non-trainable: 0
+  Non-embedding: 39,966,592 (39.97M)
+  Position embeddings: 0 (RoPE: 0 learned params)
+  Size (MB): 159.87
+
+Layer-wise breakdown (torchinfo):
+  Note: torchinfo double-counts weight-tied layers (tok_emb/lm_head).
+============================================================================================================================================
+Layer (type:depth-idx)                   Input Shape               Output Shape              Param #                   Trainable
+============================================================================================================================================
+ParrotLLM                                [32, 1024]                [32, 1024, 50258]         --                        True
+├─Embedding: 1-1                         [32, 1024]                [32, 1024, 384]           19,299,072                True
+├─Dropout: 1-2                           [32, 1024, 384]           [32, 1024, 384]           --                        --
+├─ModuleList: 1-3                        --                        --                        --                        True
+│    └─TransformerBlock: 2-1             [32, 1024, 384]           [32, 1024, 384]           --                        True
+│    │    └─RMSNorm: 3-1                 [32, 1024, 384]           [32, 1024, 384]           384                       True
+│    │    └─MultiHeadAttention: 3-2      [32, 1024, 384]           [32, 1024, 384]           589,952                   True
+│    │    └─RMSNorm: 3-3                 [32, 1024, 384]           [32, 1024, 384]           384                       True
+│    │    └─RMSNorm: 3-4                 [32, 1024, 384]           [32, 1024, 384]           384                       True
+│    │    └─SwiGLUMLP: 3-5               [32, 1024, 384]           [32, 1024, 384]           884,736                   True
+│    │    └─RMSNorm: 3-6                 [32, 1024, 384]           [32, 1024, 384]           384                       True
+│    └─TransformerBlock: 2-2             [32, 1024, 384]           [32, 1024, 384]           --                        True
+│    │    └─RMSNorm: 3-7                 [32, 1024, 384]           [32, 1024, 384]           384                       True
+│    │    └─MultiHeadAttention: 3-8      [32, 1024, 384]           [32, 1024, 384]           589,952                   True
+│    │    └─RMSNorm: 3-9                 [32, 1024, 384]           [32, 1024, 384]           384                       True
+│    │    └─RMSNorm: 3-10                [32, 1024, 384]           [32, 1024, 384]           384                       True
+│    │    └─SwiGLUMLP: 3-11              [32, 1024, 384]           [32, 1024, 384]           884,736                   True
+│    │    └─RMSNorm: 3-12                [32, 1024, 384]           [32, 1024, 384]           384                       True
+│    └─TransformerBlock: 2-3             [32, 1024, 384]           [32, 1024, 384]           --                        True
+│    │    └─RMSNorm: 3-13                [32, 1024, 384]           [32, 1024, 384]           384                       True
+│    │    └─MultiHeadAttention: 3-14     [32, 1024, 384]           [32, 1024, 384]           589,952                   True
+│    │    └─RMSNorm: 3-15                [32, 1024, 384]           [32, 1024, 384]           384                       True
+│    │    └─RMSNorm: 3-16                [32, 1024, 384]           [32, 1024, 384]           384                       True
+│    │    └─SwiGLUMLP: 3-17              [32, 1024, 384]           [32, 1024, 384]           884,736                   True
+│    │    ���─RMSNorm: 3-18                [32, 1024, 384]           [32, 1024, 384]           384                       True
+│    └─TransformerBlock: 2-4             [32, 1024, 384]           [32, 1024, 384]           --                        True
+│    │    └─RMSNorm: 3-19                [32, 1024, 384]           [32, 1024, 384]           384                       True
+│    │    └─MultiHeadAttention: 3-20     [32, 1024, 384]           [32, 1024, 384]           589,952                   True
+│    │    └─RMSNorm: 3-21                [32, 1024, 384]           [32, 1024, 384]           384                       True
+│    │    └─RMSNorm: 3-22                [32, 1024, 384]           [32, 1024, 384]           384                       True
+│    │    └─SwiGLUMLP: 3-23              [32, 1024, 384]           [32, 1024, 384]           884,736                   True
+│    │    └─RMSNorm: 3-24                [32, 1024, 384]           [32, 1024, 384]           384                       True
+│    └─TransformerBlock: 2-5             [32, 1024, 384]           [32, 1024, 384]           --                        True
+│    │    └─RMSNorm: 3-25                [32, 1024, 384]           [32, 1024, 384]           384                       True
+│    │    └─MultiHeadAttention: 3-26     [32, 1024, 384]           [32, 1024, 384]           589,952                   True
+│    │    └─RMSNorm: 3-27                [32, 1024, 384]           [32, 1024, 384]           384                       True
+│    │    └─RMSNorm: 3-28                [32, 1024, 384]           [32, 1024, 384]           384                       True
+│    │    └─SwiGLUMLP: 3-29              [32, 1024, 384]           [32, 1024, 384]           884,736                   True
+│    │    └─RMSNorm: 3-30                [32, 1024, 384]           [32, 1024, 384]           384                       True
+│    └─TransformerBlock: 2-6             [32, 1024, 384]           [32, 1024, 384]           --                        True
+│    │    └─RMSNorm: 3-31                [32, 1024, 384]           [32, 1024, 384]           384                       True
+│    │    └─MultiHeadAttention: 3-32     [32, 1024, 384]           [32, 1024, 384]           589,952                   True
+│    │    └─RMSNorm: 3-33                [32, 1024, 384]           [32, 1024, 384]           384                       True
+│    │    └─RMSNorm: 3-34                [32, 1024, 384]           [32, 1024, 384]           384                       True
+│    │    └─SwiGLUMLP: 3-35              [32, 1024, 384]           [32, 1024, 384]           884,736                   True
+│    │    └─RMSNorm: 3-36                [32, 1024, 384]           [32, 1024, 384]           384                       True
+│    └─TransformerBlock: 2-7             [32, 1024, 384]           [32, 1024, 384]           --                        True
+│    │    └─RMSNorm: 3-37                [32, 1024, 384]           [32, 1024, 384]           384                       True
+│    │    └─MultiHeadAttention: 3-38     [32, 1024, 384]           [32, 1024, 384]           589,952                   True
+│    │    └─RMSNorm: 3-39                [32, 1024, 384]           [32, 1024, 384]           384                       True
+│    │    └─RMSNorm: 3-40                [32, 1024, 384]           [32, 1024, 384]           384                       True
+│    │    └─SwiGLUMLP: 3-41              [32, 1024, 384]           [32, 1024, 384]           884,736                   True
+│    │    └─RMSNorm: 3-42                [32, 1024, 384]           [32, 1024, 384]           384                       True
+│    └─TransformerBlock: 2-8             [32, 1024, 384]           [32, 1024, 384]           --                        True
+│    │    └─RMSNorm: 3-43                [32, 1024, 384]           [32, 1024, 384]           384                       True
+│    │    └─MultiHeadAttention: 3-44     [32, 1024, 384]           [32, 1024, 384]           589,952                   True
+│    │    └─RMSNorm: 3-45                [32, 1024, 384]           [32, 1024, 384]           384                       True
+│    │    └─RMSNorm: 3-46                [32, 1024, 384]           [32, 1024, 384]           384                       True
+│    │    └─SwiGLUMLP: 3-47              [32, 1024, 384]           [32, 1024, 384]           884,736                   True
+│    │    └─RMSNorm: 3-48                [32, 1024, 384]           [32, 1024, 384]           384                       True
+│    └─TransformerBlock: 2-9             [32, 1024, 384]           [32, 1024, 384]           --                        True
+│    │    └─RMSNorm: 3-49                [32, 1024, 384]           [32, 1024, 384]           384                       True
+│    │    └─MultiHeadAttention: 3-50     [32, 1024, 384]           [32, 1024, 384]           589,952                   True
+│    │    └─RMSNorm: 3-51                [32, 1024, 384]           [32, 1024, 384]           384                       True
+│    │    └─RMSNorm: 3-52                [32, 1024, 384]           [32, 1024, 384]           384                       True
+│    │    └─SwiGLUMLP: 3-53              [32, 1024, 384]           [32, 1024, 384]           884,736                   True
+│    │    └─RMSNorm: 3-54                [32, 1024, 384]           [32, 1024, 384]           384                       True
+│    └─TransformerBlock: 2-10            [32, 1024, 384]           [32, 1024, 384]           --                        True
+│    │    └─RMSNorm: 3-55                [32, 1024, 384]           [32, 1024, 384]           384                       True
+│    │    └─MultiHeadAttention: 3-56     [32, 1024, 384]           [32, 1024, 384]           589,952                   True
+│    │    └─RMSNorm: 3-57                [32, 1024, 384]           [32, 1024, 384]           384                       True
+│    │    └─RMSNorm: 3-58                [32, 1024, 384]           [32, 1024, 384]           384                       True
+│    │    └─SwiGLUMLP: 3-59              [32, 1024, 384]           [32, 1024, 384]           884,736                   True
+│    │    └─RMSNorm: 3-60                [32, 1024, 384]           [32, 1024, 384]           384                       True
+│    └─TransformerBlock: 2-11            [32, 1024, 384]           [32, 1024, 384]           --                        True
+│    │    └─RMSNorm: 3-61                [32, 1024, 384]           [32, 1024, 384]           384                       True
+│    │    └─MultiHeadAttention: 3-62     [32, 1024, 384]           [32, 1024, 384]           589,952                   True
+│    │    └─RMSNorm: 3-63                [32, 1024, 384]           [32, 1024, 384]           384                       True
+│    │    └─RMSNorm: 3-64                [32, 1024, 384]           [32, 1024, 384]           384                       True
+│    │    └─SwiGLUMLP: 3-65              [32, 1024, 384]           [32, 1024, 384]           884,736                   True
+│    │    └─RMSNorm: 3-66                [32, 1024, 384]           [32, 1024, 384]           384                       True
+│    └─TransformerBlock: 2-12            [32, 1024, 384]           [32, 1024, 384]           --                        True
+│    │    └─RMSNorm: 3-67                [32, 1024, 384]           [32, 1024, 384]           384                       True
+│    │    └─MultiHeadAttention: 3-68     [32, 1024, 384]           [32, 1024, 384]           589,952                   True
+│    │    └─RMSNorm: 3-69                [32, 1024, 384]           [32, 1024, 384]           384                       True
+│    │    └─RMSNorm: 3-70                [32, 1024, 384]           [32, 1024, 384]           384                       True
+│    │    └─SwiGLUMLP: 3-71              [32, 1024, 384]           [32, 1024, 384]           884,736                   True
+│    │    └─RMSNorm: 3-72                [32, 1024, 384]           [32, 1024, 384]           384                       True
+│    └─TransformerBlock: 2-13            [32, 1024, 384]           [32, 1024, 384]           --                        True
+│    │    └─RMSNorm: 3-73                [32, 1024, 384]           [32, 1024, 384]           384                       True
+│    │    └─MultiHeadAttention: 3-74     [32, 1024, 384]           [32, 1024, 384]           589,952                   True
+│    │    └─RMSNorm: 3-75                [32, 1024, 384]           [32, 1024, 384]           384                       True
+│    │    └─RMSNorm: 3-76                [32, 1024, 384]           [32, 1024, 384]           384                       True
+│    │    └─SwiGLUMLP: 3-77              [32, 1024, 384]           [32, 1024, 384]           884,736                   True
+│    │    └─RMSNorm: 3-78                [32, 1024, 384]           [32, 1024, 384]           384                       True
+│    └─TransformerBlock: 2-14            [32, 1024, 384]           [32, 1024, 384]           --                        True
+│    │    └─RMSNorm: 3-79                [32, 1024, 384]           [32, 1024, 384]           384                       True
+│    │    └─MultiHeadAttention: 3-80     [32, 1024, 384]           [32, 1024, 384]           589,952                   True
+│    │    └─RMSNorm: 3-81                [32, 1024, 384]           [32, 1024, 384]           384                       True
+│    │    └─RMSNorm: 3-82                [32, 1024, 384]           [32, 1024, 384]           384                       True
+│    │    └─SwiGLUMLP: 3-83              [32, 1024, 384]           [32, 1024, 384]           884,736                   True
+│    │    └─RMSNorm: 3-84                [32, 1024, 384]           [32, 1024, 384]           384                       True
+├─RMSNorm: 1-4                           [32, 1024, 384]           [32, 1024, 384]           384                       True
+├─Linear: 1-5                            [32, 1024, 384]           [32, 1024, 50258]         19,299,072                True
+============================================================================================================================================
+Total params: 59,265,664
+Trainable params: 59,265,664
+Non-trainable params: 0
+Total mult-adds (Units.GIGABYTES): 1.90
+============================================================================================================================================
+Input size (MB): 0.26
+Forward/backward pass size (MB): 34515.45
+Params size (MB): 237.06
+Estimated Total Size (MB): 34752.78
+============================================================================================================================================
+============================================================
+2026-04-08 15:18:24 - INFO - parrotllm.training - Starting evaluation...
+2026-04-08 15:18:32 - INFO - parrotllm.training -   Initial val: loss=10.8922, ppl=53754.44
+2026-04-08 15:18:46 - INFO - parrotllm.training - 
+============================================================
+Starting training...
+  Steps per epoch (approx): 1939
+  Max steps: 3052
+============================================================
+2026-04-08 15:18:46 - INFO - parrotllm.training - LR schedule=wsd | decay_ratio=0.179 | z_loss_coeff=0.000275
+2026-04-08 15:27:38 - INFO - parrotllm.training - step      1 | epoch 0 | loss 10.9251 | lr 4.28e-06 | grad 7.1913
+2026-04-08 15:27:41 - INFO - parrotllm.training - step     10 | epoch 0 | loss 10.2404 | lr 4.28e-05 | grad 1.8256
+2026-04-08 15:27:44 - INFO - parrotllm.training - step     20 | epoch 0 | loss 9.8834 | lr 8.56e-05 | grad 1.7080
+2026-04-08 15:27:47 - INFO - parrotllm.training - step     30 | epoch 0 | loss 9.3318 | lr 1.28e-04 | grad 1.6015
+2026-04-08 15:27:50 - INFO - parrotllm.training - step     40 | epoch 0 | loss 8.6706 | lr 1.71e-04 | grad 1.6125
+2026-04-08 15:27:53 - INFO - parrotllm.training - step     50 | epoch 0 | loss 8.0446 | lr 2.14e-04 | grad 1.3510
+2026-04-08 15:27:56 - INFO - parrotllm.training - step     60 | epoch 0 | loss 7.5934 | lr 2.57e-04 | grad 0.9592
+2026-04-08 15:27:59 - INFO - parrotllm.training - step     70 | epoch 0 | loss 7.3250 | lr 3.00e-04 | grad 0.5899
+2026-04-08 15:28:02 - INFO - parrotllm.training - step     80 | epoch 0 | loss 7.0722 | lr 3.42e-04 | grad 0.6867
+2026-04-08 15:28:05 - INFO - parrotllm.training - step     90 | epoch 0 | loss 6.8537 | lr 3.85e-04 | grad 0.8117
+2026-04-08 15:28:08 - INFO - parrotllm.training - step    100 | epoch 0 | loss 6.7712 | lr 4.28e-04 | grad 1.0823
+2026-04-08 15:28:08 - INFO - parrotllm.training - Starting evaluation...
+2026-04-08 15:28:08 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:28:14 - INFO - parrotllm.training -   Train: loss=6.7712, ppl=872.39
+2026-04-08 15:28:14 - INFO - parrotllm.training -   Val:   loss=6.7855, ppl=884.90
+2026-04-08 15:28:14 - INFO - parrotllm.training -   ** New best validation loss! **
+2026-04-08 15:28:15 - INFO - parrotllm.training -   Saved best checkpoint: /home/user/code/ParrotLLM/runs/big_run/exp_c/run_20260408_151821/checkpoints/best_loss_6p7855_epoch_0000_step_0000100.pt
+2026-04-08 15:28:18 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:28:21 - INFO - parrotllm.training - step    110 | epoch 0 | loss 6.6444 | lr 4.71e-04 | grad 0.4445
+2026-04-08 15:28:24 - INFO - parrotllm.training - step    120 | epoch 0 | loss 6.5128 | lr 5.13e-04 | grad 0.8141
+2026-04-08 15:28:27 - INFO - parrotllm.training - step    130 | epoch 0 | loss 6.3627 | lr 5.56e-04 | grad 0.5265
+2026-04-08 15:28:30 - INFO - parrotllm.training - step    140 | epoch 0 | loss 6.3877 | lr 5.99e-04 | grad 1.0031
+2026-04-08 15:28:33 - INFO - parrotllm.training - step    150 | epoch 0 | loss 6.3363 | lr 6.42e-04 | grad 0.5299
+2026-04-08 15:28:36 - INFO - parrotllm.training - step    160 | epoch 0 | loss 6.0754 | lr 6.85e-04 | grad 0.4633
+2026-04-08 15:28:39 - INFO - parrotllm.training - step    170 | epoch 0 | loss 6.1721 | lr 7.27e-04 | grad 1.0633
+2026-04-08 15:28:43 - INFO - parrotllm.training - step    180 | epoch 0 | loss 5.9773 | lr 7.70e-04 | grad 0.6048
+2026-04-08 15:28:46 - INFO - parrotllm.training - step    190 | epoch 0 | loss 6.0687 | lr 8.13e-04 | grad 2.0068
+2026-04-08 15:28:49 - INFO - parrotllm.training - step    200 | epoch 0 | loss 5.9641 | lr 8.56e-04 | grad 0.3905
+2026-04-08 15:28:49 - INFO - parrotllm.training - Starting evaluation...
+2026-04-08 15:28:49 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:28:55 - INFO - parrotllm.training -   Train: loss=5.9641, ppl=389.20
+2026-04-08 15:28:55 - INFO - parrotllm.training -   Val:   loss=5.9415, ppl=380.49
+2026-04-08 15:28:55 - INFO - parrotllm.training -   ** New best validation loss! **
+2026-04-08 15:28:55 - INFO - parrotllm.training -   Saved best checkpoint: /home/user/code/ParrotLLM/runs/big_run/exp_c/run_20260408_151821/checkpoints/best_loss_5p9415_epoch_0000_step_0000200.pt
+2026-04-08 15:28:56 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:28:59 - INFO - parrotllm.training - step    210 | epoch 0 | loss 5.8993 | lr 8.99e-04 | grad 0.3702
+2026-04-08 15:29:02 - INFO - parrotllm.training - step    220 | epoch 0 | loss 5.7893 | lr 9.41e-04 | grad 0.3061
+2026-04-08 15:29:05 - INFO - parrotllm.training - step    230 | epoch 0 | loss 5.7677 | lr 9.84e-04 | grad 0.5724
+2026-04-08 15:29:09 - INFO - parrotllm.training - step    240 | epoch 0 | loss 5.8007 | lr 1.03e-03 | grad 0.7154
+2026-04-08 15:29:12 - INFO - parrotllm.training - step    250 | epoch 0 | loss 5.7076 | lr 1.07e-03 | grad 0.3399
+2026-04-08 15:29:13 - INFO - parrotllm.training - Saved checkpoint: /home/user/code/ParrotLLM/runs/big_run/exp_c/run_20260408_151821/checkpoints/last_epoch_0000_step_0000250.pt
+2026-04-08 15:29:16 - INFO - parrotllm.training - step    260 | epoch 0 | loss 5.6365 | lr 1.11e-03 | grad 0.9196
+2026-04-08 15:29:19 - INFO - parrotllm.training - step    270 | epoch 0 | loss 5.5430 | lr 1.16e-03 | grad 0.3849
+2026-04-08 15:29:23 - INFO - parrotllm.training - step    280 | epoch 0 | loss 5.5255 | lr 1.20e-03 | grad 0.4714
+2026-04-08 15:29:26 - INFO - parrotllm.training - step    290 | epoch 0 | loss 5.4422 | lr 1.24e-03 | grad 0.6009
+2026-04-08 15:29:29 - INFO - parrotllm.training - step    300 | epoch 0 | loss 5.4008 | lr 1.28e-03 | grad 0.4876
+2026-04-08 15:29:29 - INFO - parrotllm.training - Starting evaluation...
+2026-04-08 15:29:29 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:29:34 - INFO - parrotllm.training -   Train: loss=5.4008, ppl=221.59
+2026-04-08 15:29:34 - INFO - parrotllm.training -   Val:   loss=5.3978, ppl=220.91
+2026-04-08 15:29:34 - INFO - parrotllm.training -   ** New best validation loss! **
+2026-04-08 15:29:35 - INFO - parrotllm.training -   Saved best checkpoint: /home/user/code/ParrotLLM/runs/big_run/exp_c/run_20260408_151821/checkpoints/best_loss_5p3978_epoch_0000_step_0000300.pt
+2026-04-08 15:29:36 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:29:39 - INFO - parrotllm.training - step    310 | epoch 0 | loss 5.3590 | lr 1.30e-03 | grad 0.5017
+2026-04-08 15:29:42 - INFO - parrotllm.training - step    320 | epoch 0 | loss 5.3750 | lr 1.30e-03 | grad 0.3000
+2026-04-08 15:29:45 - INFO - parrotllm.training - step    330 | epoch 0 | loss 5.2987 | lr 1.30e-03 | grad 0.7844
+2026-04-08 15:29:48 - INFO - parrotllm.training - step    340 | epoch 0 | loss 5.1440 | lr 1.30e-03 | grad 0.5360
+2026-04-08 15:29:51 - INFO - parrotllm.training - step    350 | epoch 0 | loss 5.2244 | lr 1.30e-03 | grad 0.6217
+2026-04-08 15:29:54 - INFO - parrotllm.training - step    360 | epoch 0 | loss 5.2049 | lr 1.30e-03 | grad 0.3205
+2026-04-08 15:29:57 - INFO - parrotllm.training - step    370 | epoch 0 | loss 5.2625 | lr 1.30e-03 | grad 0.2888
+2026-04-08 15:30:00 - INFO - parrotllm.training - step    380 | epoch 0 | loss 5.0553 | lr 1.30e-03 | grad 0.6176
+2026-04-08 15:30:03 - INFO - parrotllm.training - step    390 | epoch 0 | loss 4.9976 | lr 1.30e-03 | grad 0.5151
+2026-04-08 15:30:07 - INFO - parrotllm.training - step    400 | epoch 0 | loss 5.0366 | lr 1.30e-03 | grad 0.3490
+2026-04-08 15:30:07 - INFO - parrotllm.training - Starting evaluation...
+2026-04-08 15:30:07 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:30:12 - INFO - parrotllm.training -   Train: loss=5.0366, ppl=153.95
+2026-04-08 15:30:12 - INFO - parrotllm.training -   Val:   loss=5.0045, ppl=149.08
+2026-04-08 15:30:12 - INFO - parrotllm.training -   ** New best validation loss! **
+2026-04-08 15:30:13 - INFO - parrotllm.training -   Saved best checkpoint: /home/user/code/ParrotLLM/runs/big_run/exp_c/run_20260408_151821/checkpoints/best_loss_5p0045_epoch_0000_step_0000400.pt
+2026-04-08 15:30:14 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:30:17 - INFO - parrotllm.training - step    410 | epoch 0 | loss 4.9822 | lr 1.30e-03 | grad 0.6036
+2026-04-08 15:30:20 - INFO - parrotllm.training - step    420 | epoch 0 | loss 5.0989 | lr 1.30e-03 | grad 0.5228
+2026-04-08 15:30:23 - INFO - parrotllm.training - step    430 | epoch 0 | loss 4.9068 | lr 1.30e-03 | grad 0.3168
+2026-04-08 15:30:26 - INFO - parrotllm.training - step    440 | epoch 0 | loss 4.9066 | lr 1.30e-03 | grad 0.4042
+2026-04-08 15:30:29 - INFO - parrotllm.training - step    450 | epoch 0 | loss 4.8326 | lr 1.30e-03 | grad 0.3326
+2026-04-08 15:30:32 - INFO - parrotllm.training - step    460 | epoch 0 | loss 4.8046 | lr 1.30e-03 | grad 0.3633
+2026-04-08 15:30:35 - INFO - parrotllm.training - step    470 | epoch 0 | loss 4.7192 | lr 1.30e-03 | grad 0.3735
+2026-04-08 15:30:38 - INFO - parrotllm.training - step    480 | epoch 0 | loss 4.7584 | lr 1.30e-03 | grad 0.3602
+2026-04-08 15:30:41 - INFO - parrotllm.training - step    490 | epoch 0 | loss 4.8350 | lr 1.30e-03 | grad 0.4553
+2026-04-08 15:30:44 - INFO - parrotllm.training - step    500 | epoch 0 | loss 4.6601 | lr 1.30e-03 | grad 0.3017
+2026-04-08 15:30:44 - INFO - parrotllm.training - Starting evaluation...
+2026-04-08 15:30:44 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:30:50 - INFO - parrotllm.training -   Train: loss=4.6601, ppl=105.65
+2026-04-08 15:30:50 - INFO - parrotllm.training -   Val:   loss=4.6890, ppl=108.75
+2026-04-08 15:30:50 - INFO - parrotllm.training -   ** New best validation loss! **
+2026-04-08 15:30:51 - INFO - parrotllm.training -   Saved best checkpoint: /home/user/code/ParrotLLM/runs/big_run/exp_c/run_20260408_151821/checkpoints/best_loss_4p6890_epoch_0000_step_0000500.pt
+2026-04-08 15:30:52 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:30:53 - INFO - parrotllm.training - Saved checkpoint: /home/user/code/ParrotLLM/runs/big_run/exp_c/run_20260408_151821/checkpoints/last_epoch_0000_step_0000500.pt
+2026-04-08 15:30:57 - INFO - parrotllm.training - step    510 | epoch 0 | loss 4.6302 | lr 1.30e-03 | grad 0.4871
+2026-04-08 15:31:00 - INFO - parrotllm.training - step    520 | epoch 0 | loss 4.7525 | lr 1.30e-03 | grad 0.3075
+2026-04-08 15:31:03 - INFO - parrotllm.training - step    530 | epoch 0 | loss 4.6357 | lr 1.30e-03 | grad 0.2917
+2026-04-08 15:31:06 - INFO - parrotllm.training - step    540 | epoch 0 | loss 4.6641 | lr 1.30e-03 | grad 0.4476
+2026-04-08 15:31:09 - INFO - parrotllm.training - step    550 | epoch 0 | loss 4.5351 | lr 1.30e-03 | grad 0.2815
+2026-04-08 15:31:12 - INFO - parrotllm.training - step    560 | epoch 0 | loss 4.4852 | lr 1.30e-03 | grad 0.5069
+2026-04-08 15:31:15 - INFO - parrotllm.training - step    570 | epoch 0 | loss 4.5373 | lr 1.30e-03 | grad 0.3964
+2026-04-08 15:31:18 - INFO - parrotllm.training - step    580 | epoch 0 | loss 4.3484 | lr 1.30e-03 | grad 0.3406
+2026-04-08 15:31:21 - INFO - parrotllm.training - step    590 | epoch 0 | loss 4.4593 | lr 1.30e-03 | grad 0.4430
+2026-04-08 15:31:24 - INFO - parrotllm.training - step    600 | epoch 0 | loss 4.4929 | lr 1.30e-03 | grad 0.3649
+2026-04-08 15:31:24 - INFO - parrotllm.training - Starting evaluation...
+2026-04-08 15:31:24 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:31:30 - INFO - parrotllm.training -   Train: loss=4.4929, ppl=89.38
+2026-04-08 15:31:30 - INFO - parrotllm.training -   Val:   loss=4.4547, ppl=86.03
+2026-04-08 15:31:30 - INFO - parrotllm.training -   ** New best validation loss! **
+2026-04-08 15:31:31 - INFO - parrotllm.training -   Saved best checkpoint: /home/user/code/ParrotLLM/runs/big_run/exp_c/run_20260408_151821/checkpoints/best_loss_4p4547_epoch_0000_step_0000600.pt
+2026-04-08 15:31:32 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:31:35 - INFO - parrotllm.training - step    610 | epoch 0 | loss 4.4223 | lr 1.30e-03 | grad 0.2856
+2026-04-08 15:31:38 - INFO - parrotllm.training - step    620 | epoch 0 | loss 4.4594 | lr 1.30e-03 | grad 0.4465
+2026-04-08 15:31:41 - INFO - parrotllm.training - step    630 | epoch 0 | loss 4.3711 | lr 1.30e-03 | grad 0.2910
+2026-04-08 15:31:44 - INFO - parrotllm.training - step    640 | epoch 0 | loss 4.3741 | lr 1.30e-03 | grad 0.4039
+2026-04-08 15:31:47 - INFO - parrotllm.training - step    650 | epoch 0 | loss 4.4339 | lr 1.30e-03 | grad 0.3592
+2026-04-08 15:31:50 - INFO - parrotllm.training - step    660 | epoch 0 | loss 4.2392 | lr 1.30e-03 | grad 0.2919
+2026-04-08 15:31:54 - INFO - parrotllm.training - step    670 | epoch 0 | loss 4.3642 | lr 1.30e-03 | grad 0.3359
+2026-04-08 15:31:57 - INFO - parrotllm.training - step    680 | epoch 0 | loss 4.3686 | lr 1.30e-03 | grad 0.3131
+2026-04-08 15:32:00 - INFO - parrotllm.training - step    690 | epoch 0 | loss 4.2038 | lr 1.30e-03 | grad 0.4343
+2026-04-08 15:32:03 - INFO - parrotllm.training - step    700 | epoch 0 | loss 4.2976 | lr 1.30e-03 | grad 0.2925
+2026-04-08 15:32:03 - INFO - parrotllm.training - Starting evaluation...
+2026-04-08 15:32:03 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:32:08 - INFO - parrotllm.training -   Train: loss=4.2976, ppl=73.52
+2026-04-08 15:32:08 - INFO - parrotllm.training -   Val:   loss=4.2959, ppl=73.40
+2026-04-08 15:32:08 - INFO - parrotllm.training -   ** New best validation loss! **
+2026-04-08 15:32:09 - INFO - parrotllm.training -   Saved best checkpoint: /home/user/code/ParrotLLM/runs/big_run/exp_c/run_20260408_151821/checkpoints/best_loss_4p2959_epoch_0000_step_0000700.pt
+2026-04-08 15:32:10 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:32:13 - INFO - parrotllm.training - step    710 | epoch 0 | loss 4.2350 | lr 1.30e-03 | grad 0.3885
+2026-04-08 15:32:16 - INFO - parrotllm.training - step    720 | epoch 0 | loss 4.3226 | lr 1.30e-03 | grad 0.3863
+2026-04-08 15:32:19 - INFO - parrotllm.training - step    730 | epoch 0 | loss 4.1857 | lr 1.30e-03 | grad 0.2947
+2026-04-08 15:32:23 - INFO - parrotllm.training - step    740 | epoch 0 | loss 4.1703 | lr 1.30e-03 | grad 0.2679
+2026-04-08 15:32:26 - INFO - parrotllm.training - step    750 | epoch 0 | loss 4.2227 | lr 1.30e-03 | grad 0.3085
+2026-04-08 15:32:26 - INFO - parrotllm.training - Saved checkpoint: /home/user/code/ParrotLLM/runs/big_run/exp_c/run_20260408_151821/checkpoints/last_epoch_0000_step_0000750.pt
+2026-04-08 15:32:30 - INFO - parrotllm.training - step    760 | epoch 0 | loss 4.2147 | lr 1.30e-03 | grad 0.4243
+2026-04-08 15:32:33 - INFO - parrotllm.training - step    770 | epoch 0 | loss 4.1278 | lr 1.30e-03 | grad 0.2800
+2026-04-08 15:32:36 - INFO - parrotllm.training - step    780 | epoch 0 | loss 4.2099 | lr 1.30e-03 | grad 0.3031
+2026-04-08 15:32:39 - INFO - parrotllm.training - step    790 | epoch 0 | loss 4.2144 | lr 1.30e-03 | grad 0.2910
+2026-04-08 15:32:43 - INFO - parrotllm.training - step    800 | epoch 0 | loss 4.2627 | lr 1.30e-03 | grad 0.3624
+2026-04-08 15:32:43 - INFO - parrotllm.training - Starting evaluation...
+2026-04-08 15:32:43 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:32:48 - INFO - parrotllm.training -   Train: loss=4.2627, ppl=71.00
+2026-04-08 15:32:48 - INFO - parrotllm.training -   Val:   loss=4.1968, ppl=66.47
+2026-04-08 15:32:48 - INFO - parrotllm.training -   ** New best validation loss! **
+2026-04-08 15:32:49 - INFO - parrotllm.training -   Saved best checkpoint: /home/user/code/ParrotLLM/runs/big_run/exp_c/run_20260408_151821/checkpoints/best_loss_4p1968_epoch_0000_step_0000800.pt
+2026-04-08 15:32:50 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:32:53 - INFO - parrotllm.training - step    810 | epoch 0 | loss 4.0694 | lr 1.30e-03 | grad 0.2921
+2026-04-08 15:32:56 - INFO - parrotllm.training - step    820 | epoch 0 | loss 4.1191 | lr 1.30e-03 | grad 0.2928
+2026-04-08 15:32:59 - INFO - parrotllm.training - step    830 | epoch 0 | loss 4.1893 | lr 1.30e-03 | grad 0.3178
+2026-04-08 15:33:02 - INFO - parrotllm.training - step    840 | epoch 0 | loss 4.1510 | lr 1.30e-03 | grad 0.3229
+2026-04-08 15:33:05 - INFO - parrotllm.training - step    850 | epoch 0 | loss 4.0849 | lr 1.30e-03 | grad 0.3319
+2026-04-08 15:33:08 - INFO - parrotllm.training - step    860 | epoch 0 | loss 4.0477 | lr 1.30e-03 | grad 0.2827
+2026-04-08 15:33:11 - INFO - parrotllm.training - step    870 | epoch 0 | loss 4.0942 | lr 1.30e-03 | grad 0.3390
+2026-04-08 15:33:14 - INFO - parrotllm.training - step    880 | epoch 0 | loss 4.1632 | lr 1.30e-03 | grad 0.3094
+2026-04-08 15:33:17 - INFO - parrotllm.training - step    890 | epoch 0 | loss 4.1429 | lr 1.30e-03 | grad 0.3378
+2026-04-08 15:33:21 - INFO - parrotllm.training - step    900 | epoch 0 | loss 4.0734 | lr 1.30e-03 | grad 0.2768
+2026-04-08 15:33:21 - INFO - parrotllm.training - Starting evaluation...
+2026-04-08 15:33:21 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:33:26 - INFO - parrotllm.training -   Train: loss=4.0734, ppl=58.75
+2026-04-08 15:33:26 - INFO - parrotllm.training -   Val:   loss=4.1133, ppl=61.15
+2026-04-08 15:33:26 - INFO - parrotllm.training -   ** New best validation loss! **
+2026-04-08 15:33:27 - INFO - parrotllm.training -   Saved best checkpoint: /home/user/code/ParrotLLM/runs/big_run/exp_c/run_20260408_151821/checkpoints/best_loss_4p1133_epoch_0000_step_0000900.pt
+2026-04-08 15:33:28 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:33:31 - INFO - parrotllm.training - step    910 | epoch 0 | loss 4.1199 | lr 1.30e-03 | grad 0.3350
+2026-04-08 15:33:34 - INFO - parrotllm.training - step    920 | epoch 0 | loss 4.1784 | lr 1.30e-03 | grad 0.4393
+2026-04-08 15:33:37 - INFO - parrotllm.training - step    930 | epoch 0 | loss 4.1188 | lr 1.30e-03 | grad 0.3162
+2026-04-08 15:33:40 - INFO - parrotllm.training - step    940 | epoch 0 | loss 4.1430 | lr 1.30e-03 | grad 0.2769
+2026-04-08 15:33:43 - INFO - parrotllm.training - step    950 | epoch 0 | loss 4.1601 | lr 1.30e-03 | grad 0.3317
+2026-04-08 15:33:46 - INFO - parrotllm.training - step    960 | epoch 0 | loss 3.9971 | lr 1.30e-03 | grad 0.2652
+2026-04-08 15:33:49 - INFO - parrotllm.training - step    970 | epoch 0 | loss 4.1860 | lr 1.30e-03 | grad 0.3124
+2026-04-08 15:33:52 - INFO - parrotllm.training - step    980 | epoch 0 | loss 4.0343 | lr 1.30e-03 | grad 0.2936
+2026-04-08 15:33:55 - INFO - parrotllm.training - step    990 | epoch 0 | loss 3.9024 | lr 1.30e-03 | grad 0.3808
+2026-04-08 15:33:58 - INFO - parrotllm.training - step   1000 | epoch 0 | loss 4.0872 | lr 1.30e-03 | grad 0.2896
+2026-04-08 15:33:58 - INFO - parrotllm.training - Starting evaluation...
+2026-04-08 15:33:58 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:34:04 - INFO - parrotllm.training -   Train: loss=4.0872, ppl=59.57
+2026-04-08 15:34:04 - INFO - parrotllm.training -   Val:   loss=4.0582, ppl=57.87
+2026-04-08 15:34:04 - INFO - parrotllm.training -   ** New best validation loss! **
+2026-04-08 15:34:05 - INFO - parrotllm.training -   Saved best checkpoint: /home/user/code/ParrotLLM/runs/big_run/exp_c/run_20260408_151821/checkpoints/best_loss_4p0582_epoch_0000_step_0001000.pt
+2026-04-08 15:34:06 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:34:07 - INFO - parrotllm.training - Saved checkpoint: /home/user/code/ParrotLLM/runs/big_run/exp_c/run_20260408_151821/checkpoints/last_epoch_0000_step_0001000.pt
+2026-04-08 15:34:10 - INFO - parrotllm.training - step   1010 | epoch 0 | loss 3.9443 | lr 1.30e-03 | grad 0.2616
+2026-04-08 15:34:14 - INFO - parrotllm.training - step   1020 | epoch 0 | loss 3.9606 | lr 1.30e-03 | grad 0.2707
+2026-04-08 15:34:17 - INFO - parrotllm.training - step   1030 | epoch 0 | loss 4.0572 | lr 1.30e-03 | grad 0.2839
+2026-04-08 15:34:20 - INFO - parrotllm.training - step   1040 | epoch 0 | loss 4.0803 | lr 1.30e-03 | grad 0.3363
+2026-04-08 15:34:23 - INFO - parrotllm.training - step   1050 | epoch 0 | loss 4.0010 | lr 1.30e-03 | grad 0.2776
+2026-04-08 15:34:26 - INFO - parrotllm.training - step   1060 | epoch 0 | loss 4.0288 | lr 1.30e-03 | grad 0.2486
+2026-04-08 15:34:29 - INFO - parrotllm.training - step   1070 | epoch 0 | loss 4.0880 | lr 1.30e-03 | grad 0.3123
+2026-04-08 15:34:32 - INFO - parrotllm.training - step   1080 | epoch 0 | loss 4.1342 | lr 1.30e-03 | grad 0.2622
+2026-04-08 15:34:35 - INFO - parrotllm.training - step   1090 | epoch 0 | loss 4.0376 | lr 1.30e-03 | grad 0.3454
+2026-04-08 15:34:38 - INFO - parrotllm.training - step   1100 | epoch 0 | loss 3.9568 | lr 1.30e-03 | grad 0.2603
+2026-04-08 15:34:38 - INFO - parrotllm.training - Starting evaluation...
+2026-04-08 15:34:38 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:34:44 - INFO - parrotllm.training -   Train: loss=3.9568, ppl=52.29
+2026-04-08 15:34:44 - INFO - parrotllm.training -   Val:   loss=4.0076, ppl=55.01
+2026-04-08 15:34:44 - INFO - parrotllm.training -   ** New best validation loss! **
+2026-04-08 15:34:45 - INFO - parrotllm.training -   Saved best checkpoint: /home/user/code/ParrotLLM/runs/big_run/exp_c/run_20260408_151821/checkpoints/best_loss_4p0076_epoch_0000_step_0001100.pt
+2026-04-08 15:34:46 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:34:49 - INFO - parrotllm.training - step   1110 | epoch 0 | loss 3.9544 | lr 1.30e-03 | grad 0.2787
+2026-04-08 15:34:52 - INFO - parrotllm.training - step   1120 | epoch 0 | loss 3.9563 | lr 1.30e-03 | grad 0.2916
+2026-04-08 15:34:55 - INFO - parrotllm.training - step   1130 | epoch 0 | loss 4.0133 | lr 1.30e-03 | grad 0.3392
+2026-04-08 15:34:58 - INFO - parrotllm.training - step   1140 | epoch 0 | loss 3.9636 | lr 1.30e-03 | grad 0.2606
+2026-04-08 15:35:01 - INFO - parrotllm.training - step   1150 | epoch 0 | loss 4.0171 | lr 1.30e-03 | grad 0.2491
+2026-04-08 15:35:04 - INFO - parrotllm.training - step   1160 | epoch 0 | loss 3.9510 | lr 1.30e-03 | grad 0.2897
+2026-04-08 15:35:07 - INFO - parrotllm.training - step   1170 | epoch 0 | loss 4.0261 | lr 1.30e-03 | grad 0.2287
+2026-04-08 15:35:10 - INFO - parrotllm.training - step   1180 | epoch 0 | loss 3.9100 | lr 1.30e-03 | grad 0.3025
+2026-04-08 15:35:13 - INFO - parrotllm.training - step   1190 | epoch 0 | loss 3.9441 | lr 1.30e-03 | grad 0.2564
+2026-04-08 15:35:17 - INFO - parrotllm.training - step   1200 | epoch 0 | loss 3.9958 | lr 1.30e-03 | grad 0.3061
+2026-04-08 15:35:17 - INFO - parrotllm.training - Starting evaluation...
+2026-04-08 15:35:17 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:35:22 - INFO - parrotllm.training -   Train: loss=3.9958, ppl=54.37
+2026-04-08 15:35:22 - INFO - parrotllm.training -   Val:   loss=3.9682, ppl=52.89
+2026-04-08 15:35:22 - INFO - parrotllm.training -   ** New best validation loss! **
+2026-04-08 15:35:23 - INFO - parrotllm.training -   Saved best checkpoint: /home/user/code/ParrotLLM/runs/big_run/exp_c/run_20260408_151821/checkpoints/best_loss_3p9682_epoch_0000_step_0001200.pt
+2026-04-08 15:35:24 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:35:27 - INFO - parrotllm.training - step   1210 | epoch 0 | loss 3.9472 | lr 1.30e-03 | grad 0.2657
+2026-04-08 15:35:30 - INFO - parrotllm.training - step   1220 | epoch 0 | loss 3.9451 | lr 1.30e-03 | grad 0.2892
+2026-04-08 15:35:33 - INFO - parrotllm.training - step   1230 | epoch 0 | loss 3.9132 | lr 1.30e-03 | grad 0.2481
+2026-04-08 15:35:36 - INFO - parrotllm.training - step   1240 | epoch 0 | loss 3.8677 | lr 1.30e-03 | grad 0.3440
+2026-04-08 15:35:39 - INFO - parrotllm.training - step   1250 | epoch 0 | loss 3.8879 | lr 1.30e-03 | grad 0.2571
+2026-04-08 15:35:40 - INFO - parrotllm.training - Saved checkpoint: /home/user/code/ParrotLLM/runs/big_run/exp_c/run_20260408_151821/checkpoints/last_epoch_0000_step_0001250.pt
+2026-04-08 15:35:44 - INFO - parrotllm.training - step   1260 | epoch 0 | loss 3.7287 | lr 1.30e-03 | grad 0.2418
+2026-04-08 15:35:47 - INFO - parrotllm.training - step   1270 | epoch 0 | loss 3.8966 | lr 1.30e-03 | grad 0.3168
+2026-04-08 15:35:50 - INFO - parrotllm.training - step   1280 | epoch 0 | loss 3.8584 | lr 1.30e-03 | grad 0.2344
+2026-04-08 15:35:53 - INFO - parrotllm.training - step   1290 | epoch 0 | loss 3.9525 | lr 1.30e-03 | grad 0.2687
+2026-04-08 15:35:56 - INFO - parrotllm.training - step   1300 | epoch 0 | loss 3.8434 | lr 1.30e-03 | grad 0.2807
+2026-04-08 15:35:56 - INFO - parrotllm.training - Starting evaluation...
+2026-04-08 15:35:56 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:36:02 - INFO - parrotllm.training -   Train: loss=3.8434, ppl=46.68
+2026-04-08 15:36:02 - INFO - parrotllm.training -   Val:   loss=3.9362, ppl=51.22
+2026-04-08 15:36:02 - INFO - parrotllm.training -   ** New best validation loss! **
+2026-04-08 15:36:03 - INFO - parrotllm.training -   Saved best checkpoint: /home/user/code/ParrotLLM/runs/big_run/exp_c/run_20260408_151821/checkpoints/best_loss_3p9362_epoch_0000_step_0001300.pt
+2026-04-08 15:36:04 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:36:07 - INFO - parrotllm.training - step   1310 | epoch 0 | loss 3.8655 | lr 1.30e-03 | grad 0.2795
+2026-04-08 15:36:10 - INFO - parrotllm.training - step   1320 | epoch 0 | loss 3.9084 | lr 1.30e-03 | grad 0.2482
+2026-04-08 15:36:13 - INFO - parrotllm.training - step   1330 | epoch 0 | loss 3.8100 | lr 1.30e-03 | grad 0.2668
+2026-04-08 15:36:16 - INFO - parrotllm.training - step   1340 | epoch 0 | loss 3.9457 | lr 1.30e-03 | grad 0.2666
+2026-04-08 15:36:19 - INFO - parrotllm.training - step   1350 | epoch 0 | loss 3.8187 | lr 1.30e-03 | grad 0.2760
+2026-04-08 15:36:23 - INFO - parrotllm.training - step   1360 | epoch 0 | loss 3.8922 | lr 1.30e-03 | grad 0.3163
+2026-04-08 15:36:26 - INFO - parrotllm.training - step   1370 | epoch 0 | loss 3.9750 | lr 1.30e-03 | grad 0.3074
+2026-04-08 15:36:29 - INFO - parrotllm.training - step   1380 | epoch 0 | loss 3.7766 | lr 1.30e-03 | grad 0.2382
+2026-04-08 15:36:32 - INFO - parrotllm.training - step   1390 | epoch 0 | loss 4.0230 | lr 1.30e-03 | grad 0.2868
+2026-04-08 15:36:35 - INFO - parrotllm.training - step   1400 | epoch 0 | loss 3.7552 | lr 1.30e-03 | grad 0.2877
+2026-04-08 15:36:35 - INFO - parrotllm.training - Starting evaluation...
+2026-04-08 15:36:35 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:36:41 - INFO - parrotllm.training -   Train: loss=3.7552, ppl=42.74
+2026-04-08 15:36:41 - INFO - parrotllm.training -   Val:   loss=3.9075, ppl=49.77
+2026-04-08 15:36:41 - INFO - parrotllm.training -   ** New best validation loss! **
+2026-04-08 15:36:42 - INFO - parrotllm.training -   Saved best checkpoint: /home/user/code/ParrotLLM/runs/big_run/exp_c/run_20260408_151821/checkpoints/best_loss_3p9075_epoch_0000_step_0001400.pt
+2026-04-08 15:36:43 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:36:46 - INFO - parrotllm.training - step   1410 | epoch 0 | loss 3.7790 | lr 1.30e-03 | grad 0.2597
+2026-04-08 15:36:49 - INFO - parrotllm.training - step   1420 | epoch 0 | loss 3.9175 | lr 1.30e-03 | grad 0.2811
+2026-04-08 15:36:52 - INFO - parrotllm.training - step   1430 | epoch 0 | loss 3.8659 | lr 1.30e-03 | grad 0.2873
+2026-04-08 15:36:55 - INFO - parrotllm.training - step   1440 | epoch 0 | loss 3.8926 | lr 1.30e-03 | grad 0.2997
+2026-04-08 15:36:58 - INFO - parrotllm.training - step   1450 | epoch 0 | loss 3.8397 | lr 1.30e-03 | grad 0.2625
+2026-04-08 15:37:01 - INFO - parrotllm.training - step   1460 | epoch 0 | loss 3.9603 | lr 1.30e-03 | grad 0.2715
+2026-04-08 15:37:04 - INFO - parrotllm.training - step   1470 | epoch 0 | loss 3.8937 | lr 1.30e-03 | grad 0.2517
+2026-04-08 15:37:07 - INFO - parrotllm.training - step   1480 | epoch 0 | loss 3.8239 | lr 1.30e-03 | grad 0.2721
+2026-04-08 15:37:10 - INFO - parrotllm.training - step   1490 | epoch 0 | loss 3.8118 | lr 1.30e-03 | grad 0.2447
+2026-04-08 15:37:13 - INFO - parrotllm.training - step   1500 | epoch 0 | loss 3.8199 | lr 1.30e-03 | grad 0.3141
+2026-04-08 15:37:13 - INFO - parrotllm.training - Starting evaluation...
+2026-04-08 15:37:13 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:37:19 - INFO - parrotllm.training -   Train: loss=3.8199, ppl=45.60
+2026-04-08 15:37:19 - INFO - parrotllm.training -   Val:   loss=3.8827, ppl=48.55
+2026-04-08 15:37:19 - INFO - parrotllm.training -   ** New best validation loss! **
+2026-04-08 15:37:20 - INFO - parrotllm.training -   Saved best checkpoint: /home/user/code/ParrotLLM/runs/big_run/exp_c/run_20260408_151821/checkpoints/best_loss_3p8827_epoch_0000_step_0001500.pt
+2026-04-08 15:37:21 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:37:22 - INFO - parrotllm.training - Saved checkpoint: /home/user/code/ParrotLLM/runs/big_run/exp_c/run_20260408_151821/checkpoints/last_epoch_0000_step_0001500.pt
+2026-04-08 15:37:25 - INFO - parrotllm.training - step   1510 | epoch 0 | loss 3.9160 | lr 1.30e-03 | grad 0.3018
+2026-04-08 15:37:29 - INFO - parrotllm.training - step   1520 | epoch 0 | loss 3.7643 | lr 1.30e-03 | grad 0.2798
+2026-04-08 15:37:32 - INFO - parrotllm.training - step   1530 | epoch 0 | loss 3.8525 | lr 1.30e-03 | grad 0.2756
+2026-04-08 15:37:35 - INFO - parrotllm.training - step   1540 | epoch 0 | loss 3.8289 | lr 1.30e-03 | grad 0.3411
+2026-04-08 15:37:38 - INFO - parrotllm.training - step   1550 | epoch 0 | loss 3.8377 | lr 1.30e-03 | grad 0.2297
+2026-04-08 15:37:41 - INFO - parrotllm.training - step   1560 | epoch 0 | loss 3.8661 | lr 1.30e-03 | grad 0.2548
+2026-04-08 15:37:44 - INFO - parrotllm.training - step   1570 | epoch 0 | loss 3.8610 | lr 1.30e-03 | grad 0.2449
+2026-04-08 15:37:47 - INFO - parrotllm.training - step   1580 | epoch 0 | loss 3.8051 | lr 1.30e-03 | grad 0.2975
+2026-04-08 15:37:50 - INFO - parrotllm.training - step   1590 | epoch 0 | loss 3.7952 | lr 1.30e-03 | grad 0.2435
+2026-04-08 15:37:53 - INFO - parrotllm.training - step   1600 | epoch 0 | loss 3.9102 | lr 1.30e-03 | grad 0.2366
+2026-04-08 15:37:53 - INFO - parrotllm.training - Starting evaluation...
+2026-04-08 15:37:53 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:37:59 - INFO - parrotllm.training -   Train: loss=3.9102, ppl=49.91
+2026-04-08 15:37:59 - INFO - parrotllm.training -   Val:   loss=3.8590, ppl=47.42
+2026-04-08 15:37:59 - INFO - parrotllm.training -   ** New best validation loss! **
+2026-04-08 15:38:00 - INFO - parrotllm.training -   Saved best checkpoint: /home/user/code/ParrotLLM/runs/big_run/exp_c/run_20260408_151821/checkpoints/best_loss_3p8590_epoch_0000_step_0001600.pt
+2026-04-08 15:38:01 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:38:04 - INFO - parrotllm.training - step   1610 | epoch 0 | loss 3.7410 | lr 1.30e-03 | grad 0.3190
+2026-04-08 15:38:07 - INFO - parrotllm.training - step   1620 | epoch 0 | loss 3.7748 | lr 1.30e-03 | grad 0.2509
+2026-04-08 15:38:10 - INFO - parrotllm.training - step   1630 | epoch 0 | loss 3.8266 | lr 1.30e-03 | grad 0.2503
+2026-04-08 15:38:13 - INFO - parrotllm.training - step   1640 | epoch 0 | loss 3.7936 | lr 1.30e-03 | grad 0.2822
+2026-04-08 15:38:16 - INFO - parrotllm.training - step   1650 | epoch 0 | loss 3.7314 | lr 1.30e-03 | grad 0.2861
+2026-04-08 15:38:19 - INFO - parrotllm.training - step   1660 | epoch 0 | loss 3.8403 | lr 1.30e-03 | grad 0.2941
+2026-04-08 15:38:22 - INFO - parrotllm.training - step   1670 | epoch 0 | loss 3.9259 | lr 1.30e-03 | grad 0.2414
+2026-04-08 15:38:25 - INFO - parrotllm.training - step   1680 | epoch 0 | loss 3.8358 | lr 1.30e-03 | grad 0.2492
+2026-04-08 15:38:29 - INFO - parrotllm.training - step   1690 | epoch 0 | loss 3.7906 | lr 1.30e-03 | grad 0.2543
+2026-04-08 15:38:32 - INFO - parrotllm.training - step   1700 | epoch 0 | loss 3.8325 | lr 1.30e-03 | grad 0.2767
+2026-04-08 15:38:32 - INFO - parrotllm.training - Starting evaluation...
+2026-04-08 15:38:32 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:38:37 - INFO - parrotllm.training -   Train: loss=3.8325, ppl=46.18
+2026-04-08 15:38:37 - INFO - parrotllm.training -   Val:   loss=3.8459, ppl=46.80
+2026-04-08 15:38:37 - INFO - parrotllm.training -   ** New best validation loss! **
+2026-04-08 15:38:38 - INFO - parrotllm.training -   Saved best checkpoint: /home/user/code/ParrotLLM/runs/big_run/exp_c/run_20260408_151821/checkpoints/best_loss_3p8459_epoch_0000_step_0001700.pt
+2026-04-08 15:38:39 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:38:42 - INFO - parrotllm.training - step   1710 | epoch 0 | loss 3.8248 | lr 1.30e-03 | grad 0.2720
+2026-04-08 15:38:45 - INFO - parrotllm.training - step   1720 | epoch 0 | loss 3.7627 | lr 1.30e-03 | grad 0.2342
+2026-04-08 15:38:48 - INFO - parrotllm.training - step   1730 | epoch 0 | loss 3.7739 | lr 1.30e-03 | grad 0.2218
+2026-04-08 15:38:51 - INFO - parrotllm.training - step   1740 | epoch 0 | loss 3.7250 | lr 1.30e-03 | grad 0.2545
+2026-04-08 15:38:55 - INFO - parrotllm.training - step   1750 | epoch 0 | loss 3.7428 | lr 1.30e-03 | grad 0.2287
+2026-04-08 15:38:55 - INFO - parrotllm.training - Saved checkpoint: /home/user/code/ParrotLLM/runs/big_run/exp_c/run_20260408_151821/checkpoints/last_epoch_0000_step_0001750.pt
+2026-04-08 15:38:59 - INFO - parrotllm.training - step   1760 | epoch 0 | loss 3.9441 | lr 1.30e-03 | grad 0.2419
+2026-04-08 15:39:02 - INFO - parrotllm.training - step   1770 | epoch 0 | loss 3.8131 | lr 1.30e-03 | grad 0.3044
+2026-04-08 15:39:05 - INFO - parrotllm.training - step   1780 | epoch 0 | loss 3.7074 | lr 1.30e-03 | grad 0.2370
+2026-04-08 15:39:08 - INFO - parrotllm.training - step   1790 | epoch 0 | loss 3.7951 | lr 1.30e-03 | grad 0.2714
+2026-04-08 15:39:12 - INFO - parrotllm.training - step   1800 | epoch 0 | loss 3.7393 | lr 1.30e-03 | grad 0.2444
+2026-04-08 15:39:12 - INFO - parrotllm.training - Starting evaluation...
+2026-04-08 15:39:12 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:39:17 - INFO - parrotllm.training -   Train: loss=3.7393, ppl=42.07
+2026-04-08 15:39:17 - INFO - parrotllm.training -   Val:   loss=3.8229, ppl=45.73
+2026-04-08 15:39:17 - INFO - parrotllm.training -   ** New best validation loss! **
+2026-04-08 15:39:18 - INFO - parrotllm.training -   Saved best checkpoint: /home/user/code/ParrotLLM/runs/big_run/exp_c/run_20260408_151821/checkpoints/best_loss_3p8229_epoch_0000_step_0001800.pt
+2026-04-08 15:39:19 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:39:22 - INFO - parrotllm.training - step   1810 | epoch 0 | loss 3.8659 | lr 1.30e-03 | grad 0.3400
+2026-04-08 15:39:25 - INFO - parrotllm.training - step   1820 | epoch 0 | loss 3.7745 | lr 1.30e-03 | grad 0.2969
+2026-04-08 15:39:28 - INFO - parrotllm.training - step   1830 | epoch 0 | loss 3.7995 | lr 1.30e-03 | grad 0.2690
+2026-04-08 15:39:31 - INFO - parrotllm.training - step   1840 | epoch 0 | loss 3.7271 | lr 1.30e-03 | grad 0.2528
+2026-04-08 15:39:34 - INFO - parrotllm.training - step   1850 | epoch 0 | loss 3.7232 | lr 1.30e-03 | grad 0.2525
+2026-04-08 15:39:37 - INFO - parrotllm.training - step   1860 | epoch 0 | loss 3.6941 | lr 1.30e-03 | grad 0.2962
+2026-04-08 15:39:40 - INFO - parrotllm.training - step   1870 | epoch 0 | loss 3.9053 | lr 1.30e-03 | grad 0.2322
+2026-04-08 15:39:43 - INFO - parrotllm.training - step   1880 | epoch 0 | loss 3.8358 | lr 1.30e-03 | grad 0.2268
+2026-04-08 15:39:46 - INFO - parrotllm.training - step   1890 | epoch 0 | loss 3.7932 | lr 1.30e-03 | grad 0.2670
+2026-04-08 15:39:49 - INFO - parrotllm.training - step   1900 | epoch 0 | loss 3.7916 | lr 1.30e-03 | grad 0.3776
+2026-04-08 15:39:49 - INFO - parrotllm.training - Starting evaluation...
+2026-04-08 15:39:49 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:39:55 - INFO - parrotllm.training -   Train: loss=3.7916, ppl=44.33
+2026-04-08 15:39:55 - INFO - parrotllm.training -   Val:   loss=3.8155, ppl=45.40
+2026-04-08 15:39:55 - INFO - parrotllm.training -   ** New best validation loss! **
+2026-04-08 15:39:56 - INFO - parrotllm.training -   Saved best checkpoint: /home/user/code/ParrotLLM/runs/big_run/exp_c/run_20260408_151821/checkpoints/best_loss_3p8155_epoch_0000_step_0001900.pt
+2026-04-08 15:39:57 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:40:00 - INFO - parrotllm.training - step   1910 | epoch 0 | loss 3.7547 | lr 1.30e-03 | grad 0.2405
+2026-04-08 15:40:03 - INFO - parrotllm.training - step   1920 | epoch 0 | loss 3.8596 | lr 1.30e-03 | grad 0.2561
+2026-04-08 15:40:06 - INFO - parrotllm.training - step   1930 | epoch 0 | loss 3.8044 | lr 1.30e-03 | grad 0.2849
+2026-04-08 15:40:09 - INFO - parrotllm.training - Starting evaluation...
+2026-04-08 15:40:09 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:40:14 - INFO - parrotllm.training - Epoch 1 complete:
+2026-04-08 15:40:14 - INFO - parrotllm.training -   Train: loss=3.6964, ppl=40.30
+2026-04-08 15:40:14 - INFO - parrotllm.training -   Val:   loss=3.8045, ppl=44.90
+2026-04-08 15:40:14 - INFO - parrotllm.training -   ** New best validation loss! **
+2026-04-08 15:40:15 - INFO - parrotllm.training -   Saved best checkpoint: /home/user/code/ParrotLLM/runs/big_run/exp_c/run_20260408_151821/checkpoints/best_loss_3p8045_epoch_0001_step_0001939.pt
+2026-04-08 15:40:16 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:40:16 - INFO - parrotllm.training - step   1940 | epoch 1 | loss 3.7313 | lr 1.30e-03 | grad 0.2490
+2026-04-08 15:40:20 - INFO - parrotllm.training - step   1950 | epoch 1 | loss 3.6261 | lr 1.30e-03 | grad 0.2731
+2026-04-08 15:40:23 - INFO - parrotllm.training - step   1960 | epoch 1 | loss 3.8195 | lr 1.30e-03 | grad 0.2481
+2026-04-08 15:40:26 - INFO - parrotllm.training - step   1970 | epoch 1 | loss 3.6477 | lr 1.30e-03 | grad 0.2665
+2026-04-08 15:40:29 - INFO - parrotllm.training - step   1980 | epoch 1 | loss 3.7149 | lr 1.30e-03 | grad 0.2287
+2026-04-08 15:40:32 - INFO - parrotllm.training - step   1990 | epoch 1 | loss 3.7834 | lr 1.30e-03 | grad 0.2910
+2026-04-08 15:40:35 - INFO - parrotllm.training - step   2000 | epoch 1 | loss 3.6749 | lr 1.30e-03 | grad 0.2534
+2026-04-08 15:40:35 - INFO - parrotllm.training - Starting evaluation...
+2026-04-08 15:40:35 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:40:40 - INFO - parrotllm.training -   Train: loss=3.6749, ppl=39.44
+2026-04-08 15:40:40 - INFO - parrotllm.training -   Val:   loss=3.7984, ppl=44.63
+2026-04-08 15:40:40 - INFO - parrotllm.training -   ** New best validation loss! **
+2026-04-08 15:40:41 - INFO - parrotllm.training -   Saved best checkpoint: /home/user/code/ParrotLLM/runs/big_run/exp_c/run_20260408_151821/checkpoints/best_loss_3p7984_epoch_0001_step_0002000.pt
+2026-04-08 15:40:42 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:40:43 - INFO - parrotllm.training - Saved checkpoint: /home/user/code/ParrotLLM/runs/big_run/exp_c/run_20260408_151821/checkpoints/last_epoch_0001_step_0002000.pt
+2026-04-08 15:40:47 - INFO - parrotllm.training - step   2010 | epoch 1 | loss 3.7938 | lr 1.30e-03 | grad 0.2423
+2026-04-08 15:40:50 - INFO - parrotllm.training - step   2020 | epoch 1 | loss 3.7500 | lr 1.30e-03 | grad 0.2718
+2026-04-08 15:40:53 - INFO - parrotllm.training - step   2030 | epoch 1 | loss 3.8121 | lr 1.30e-03 | grad 0.2747
+2026-04-08 15:40:56 - INFO - parrotllm.training - step   2040 | epoch 1 | loss 3.7522 | lr 1.30e-03 | grad 0.2585
+2026-04-08 15:40:59 - INFO - parrotllm.training - step   2050 | epoch 1 | loss 3.6133 | lr 1.30e-03 | grad 0.3089
+2026-04-08 15:41:03 - INFO - parrotllm.training - step   2060 | epoch 1 | loss 3.6841 | lr 1.30e-03 | grad 0.2435
+2026-04-08 15:41:06 - INFO - parrotllm.training - step   2070 | epoch 1 | loss 3.6880 | lr 1.30e-03 | grad 0.3035
+2026-04-08 15:41:09 - INFO - parrotllm.training - step   2080 | epoch 1 | loss 3.6949 | lr 1.30e-03 | grad 0.2477
+2026-04-08 15:41:12 - INFO - parrotllm.training - step   2090 | epoch 1 | loss 3.7941 | lr 1.30e-03 | grad 0.2476
+2026-04-08 15:41:15 - INFO - parrotllm.training - step   2100 | epoch 1 | loss 3.7240 | lr 1.30e-03 | grad 0.2397
+2026-04-08 15:41:15 - INFO - parrotllm.training - Starting evaluation...
+2026-04-08 15:41:15 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:41:21 - INFO - parrotllm.training -   Train: loss=3.7240, ppl=41.43
+2026-04-08 15:41:21 - INFO - parrotllm.training -   Val:   loss=3.7856, ppl=44.06
+2026-04-08 15:41:21 - INFO - parrotllm.training -   ** New best validation loss! **
+2026-04-08 15:41:21 - INFO - parrotllm.training -   Saved best checkpoint: /home/user/code/ParrotLLM/runs/big_run/exp_c/run_20260408_151821/checkpoints/best_loss_3p7856_epoch_0001_step_0002100.pt
+2026-04-08 15:41:22 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:41:25 - INFO - parrotllm.training - step   2110 | epoch 1 | loss 3.6192 | lr 1.30e-03 | grad 0.2636
+2026-04-08 15:41:28 - INFO - parrotllm.training - step   2120 | epoch 1 | loss 3.7707 | lr 1.30e-03 | grad 0.2388
+2026-04-08 15:41:31 - INFO - parrotllm.training - step   2130 | epoch 1 | loss 3.8116 | lr 1.30e-03 | grad 0.2888
+2026-04-08 15:41:34 - INFO - parrotllm.training - step   2140 | epoch 1 | loss 3.8087 | lr 1.30e-03 | grad 0.2367
+2026-04-08 15:41:38 - INFO - parrotllm.training - step   2150 | epoch 1 | loss 3.6403 | lr 1.30e-03 | grad 0.2128
+2026-04-08 15:41:41 - INFO - parrotllm.training - step   2160 | epoch 1 | loss 3.7132 | lr 1.30e-03 | grad 0.3412
+2026-04-08 15:41:44 - INFO - parrotllm.training - step   2170 | epoch 1 | loss 3.7299 | lr 1.30e-03 | grad 0.2787
+2026-04-08 15:41:47 - INFO - parrotllm.training - step   2180 | epoch 1 | loss 3.7728 | lr 1.30e-03 | grad 0.2814
+2026-04-08 15:41:50 - INFO - parrotllm.training - step   2190 | epoch 1 | loss 3.7121 | lr 1.30e-03 | grad 0.2459
+2026-04-08 15:41:53 - INFO - parrotllm.training - step   2200 | epoch 1 | loss 3.7121 | lr 1.30e-03 | grad 0.2179
+2026-04-08 15:41:53 - INFO - parrotllm.training - Starting evaluation...
+2026-04-08 15:41:53 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:41:59 - INFO - parrotllm.training -   Train: loss=3.7121, ppl=40.94
+2026-04-08 15:41:59 - INFO - parrotllm.training -   Val:   loss=3.7710, ppl=43.42
+2026-04-08 15:41:59 - INFO - parrotllm.training -   ** New best validation loss! **
+2026-04-08 15:42:00 - INFO - parrotllm.training -   Saved best checkpoint: /home/user/code/ParrotLLM/runs/big_run/exp_c/run_20260408_151821/checkpoints/best_loss_3p7710_epoch_0001_step_0002200.pt
+2026-04-08 15:42:00 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:42:03 - INFO - parrotllm.training - step   2210 | epoch 1 | loss 3.7015 | lr 1.30e-03 | grad 0.2415
+2026-04-08 15:42:07 - INFO - parrotllm.training - step   2220 | epoch 1 | loss 3.7428 | lr 1.30e-03 | grad 0.2481
+2026-04-08 15:42:10 - INFO - parrotllm.training - step   2230 | epoch 1 | loss 3.7161 | lr 1.30e-03 | grad 0.2212
+2026-04-08 15:42:13 - INFO - parrotllm.training - step   2240 | epoch 1 | loss 3.7382 | lr 1.30e-03 | grad 0.2476
+2026-04-08 15:42:16 - INFO - parrotllm.training - step   2250 | epoch 1 | loss 3.8764 | lr 1.30e-03 | grad 0.2595
+2026-04-08 15:42:17 - INFO - parrotllm.training - Saved checkpoint: /home/user/code/ParrotLLM/runs/big_run/exp_c/run_20260408_151821/checkpoints/last_epoch_0001_step_0002250.pt
+2026-04-08 15:42:20 - INFO - parrotllm.training - step   2260 | epoch 1 | loss 3.7767 | lr 1.30e-03 | grad 0.2349
+2026-04-08 15:42:23 - INFO - parrotllm.training - step   2270 | epoch 1 | loss 3.6889 | lr 1.30e-03 | grad 0.3087
+2026-04-08 15:42:26 - INFO - parrotllm.training - step   2280 | epoch 1 | loss 3.5946 | lr 1.30e-03 | grad 0.2607
+2026-04-08 15:42:30 - INFO - parrotllm.training - step   2290 | epoch 1 | loss 3.7708 | lr 1.30e-03 | grad 0.2122
+2026-04-08 15:42:33 - INFO - parrotllm.training - step   2300 | epoch 1 | loss 3.6699 | lr 1.30e-03 | grad 0.2939
+2026-04-08 15:42:33 - INFO - parrotllm.training - Starting evaluation...
+2026-04-08 15:42:33 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:42:38 - INFO - parrotllm.training -   Train: loss=3.6699, ppl=39.25
+2026-04-08 15:42:38 - INFO - parrotllm.training -   Val:   loss=3.7645, ppl=43.14
+2026-04-08 15:42:38 - INFO - parrotllm.training -   ** New best validation loss! **
+2026-04-08 15:42:39 - INFO - parrotllm.training -   Saved best checkpoint: /home/user/code/ParrotLLM/runs/big_run/exp_c/run_20260408_151821/checkpoints/best_loss_3p7645_epoch_0001_step_0002300.pt
+2026-04-08 15:42:40 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:42:43 - INFO - parrotllm.training - step   2310 | epoch 1 | loss 3.6026 | lr 1.30e-03 | grad 0.2336
+2026-04-08 15:42:46 - INFO - parrotllm.training - step   2320 | epoch 1 | loss 3.7249 | lr 1.30e-03 | grad 0.2755
+2026-04-08 15:42:49 - INFO - parrotllm.training - step   2330 | epoch 1 | loss 3.7662 | lr 1.30e-03 | grad 0.2843
+2026-04-08 15:42:52 - INFO - parrotllm.training - step   2340 | epoch 1 | loss 3.6889 | lr 1.30e-03 | grad 0.2827
+2026-04-08 15:42:55 - INFO - parrotllm.training - step   2350 | epoch 1 | loss 3.7208 | lr 1.30e-03 | grad 0.2560
+2026-04-08 15:42:59 - INFO - parrotllm.training - step   2360 | epoch 1 | loss 3.6366 | lr 1.30e-03 | grad 0.2363
+2026-04-08 15:43:02 - INFO - parrotllm.training - step   2370 | epoch 1 | loss 3.6698 | lr 1.30e-03 | grad 0.2555
+2026-04-08 15:43:05 - INFO - parrotllm.training - step   2380 | epoch 1 | loss 3.5933 | lr 1.30e-03 | grad 0.2343
+2026-04-08 15:43:08 - INFO - parrotllm.training - step   2390 | epoch 1 | loss 3.8069 | lr 1.30e-03 | grad 0.2477
+2026-04-08 15:43:11 - INFO - parrotllm.training - step   2400 | epoch 1 | loss 3.7073 | lr 1.30e-03 | grad 0.2314
+2026-04-08 15:43:11 - INFO - parrotllm.training - Starting evaluation...
+2026-04-08 15:43:11 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:43:17 - INFO - parrotllm.training -   Train: loss=3.7073, ppl=40.74
+2026-04-08 15:43:17 - INFO - parrotllm.training -   Val:   loss=3.7509, ppl=42.56
+2026-04-08 15:43:17 - INFO - parrotllm.training -   ** New best validation loss! **
+2026-04-08 15:43:17 - INFO - parrotllm.training -   Saved best checkpoint: /home/user/code/ParrotLLM/runs/big_run/exp_c/run_20260408_151821/checkpoints/best_loss_3p7509_epoch_0001_step_0002400.pt
+2026-04-08 15:43:18 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:43:21 - INFO - parrotllm.training - step   2410 | epoch 1 | loss 3.7404 | lr 1.30e-03 | grad 0.2637
+2026-04-08 15:43:24 - INFO - parrotllm.training - step   2420 | epoch 1 | loss 3.6777 | lr 1.30e-03 | grad 0.2572
+2026-04-08 15:43:27 - INFO - parrotllm.training - step   2430 | epoch 1 | loss 3.7775 | lr 1.30e-03 | grad 0.2693
+2026-04-08 15:43:30 - INFO - parrotllm.training - step   2440 | epoch 1 | loss 3.6974 | lr 1.30e-03 | grad 0.2119
+2026-04-08 15:43:34 - INFO - parrotllm.training - step   2450 | epoch 1 | loss 3.6649 | lr 1.30e-03 | grad 0.2627
+2026-04-08 15:43:37 - INFO - parrotllm.training - step   2460 | epoch 1 | loss 3.7536 | lr 1.30e-03 | grad 0.2344
+2026-04-08 15:43:40 - INFO - parrotllm.training - step   2470 | epoch 1 | loss 3.7454 | lr 1.30e-03 | grad 0.3035
+2026-04-08 15:43:43 - INFO - parrotllm.training - step   2480 | epoch 1 | loss 3.6178 | lr 1.30e-03 | grad 0.2845
+2026-04-08 15:43:46 - INFO - parrotllm.training - step   2490 | epoch 1 | loss 3.7363 | lr 1.30e-03 | grad 0.2596
+2026-04-08 15:43:49 - INFO - parrotllm.training - step   2500 | epoch 1 | loss 3.7079 | lr 1.30e-03 | grad 0.2670
+2026-04-08 15:43:49 - INFO - parrotllm.training - Starting evaluation...
+2026-04-08 15:43:49 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:43:54 - INFO - parrotllm.training -   Train: loss=3.7079, ppl=40.77
+2026-04-08 15:43:54 - INFO - parrotllm.training -   Val:   loss=3.7442, ppl=42.28
+2026-04-08 15:43:54 - INFO - parrotllm.training -   ** New best validation loss! **
+2026-04-08 15:43:55 - INFO - parrotllm.training -   Saved best checkpoint: /home/user/code/ParrotLLM/runs/big_run/exp_c/run_20260408_151821/checkpoints/best_loss_3p7442_epoch_0001_step_0002500.pt
+2026-04-08 15:43:56 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:43:57 - INFO - parrotllm.training - Saved checkpoint: /home/user/code/ParrotLLM/runs/big_run/exp_c/run_20260408_151821/checkpoints/last_epoch_0001_step_0002500.pt
+2026-04-08 15:44:01 - INFO - parrotllm.training - step   2510 | epoch 1 | loss 3.7756 | lr 1.30e-03 | grad 0.2613
+2026-04-08 15:44:04 - INFO - parrotllm.training - step   2520 | epoch 1 | loss 3.7104 | lr 1.27e-03 | grad 0.2520
+2026-04-08 15:44:07 - INFO - parrotllm.training - step   2530 | epoch 1 | loss 3.7595 | lr 1.25e-03 | grad 0.2585
+2026-04-08 15:44:10 - INFO - parrotllm.training - step   2540 | epoch 1 | loss 3.5595 | lr 1.23e-03 | grad 0.2308
+2026-04-08 15:44:13 - INFO - parrotllm.training - step   2550 | epoch 1 | loss 3.6637 | lr 1.21e-03 | grad 0.2207
+2026-04-08 15:44:16 - INFO - parrotllm.training - step   2560 | epoch 1 | loss 3.6794 | lr 1.19e-03 | grad 0.2110
+2026-04-08 15:44:20 - INFO - parrotllm.training - step   2570 | epoch 1 | loss 3.6872 | lr 1.17e-03 | grad 0.2318
+2026-04-08 15:44:23 - INFO - parrotllm.training - step   2580 | epoch 1 | loss 3.6743 | lr 1.15e-03 | grad 0.2062
+2026-04-08 15:44:26 - INFO - parrotllm.training - step   2590 | epoch 1 | loss 3.5646 | lr 1.12e-03 | grad 0.2120
+2026-04-08 15:44:29 - INFO - parrotllm.training - step   2600 | epoch 1 | loss 3.6925 | lr 1.10e-03 | grad 0.1942
+2026-04-08 15:44:29 - INFO - parrotllm.training - Starting evaluation...
+2026-04-08 15:44:29 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:44:34 - INFO - parrotllm.training -   Train: loss=3.6925, ppl=40.14
+2026-04-08 15:44:34 - INFO - parrotllm.training -   Val:   loss=3.7192, ppl=41.23
+2026-04-08 15:44:34 - INFO - parrotllm.training -   ** New best validation loss! **
+2026-04-08 15:44:35 - INFO - parrotllm.training -   Saved best checkpoint: /home/user/code/ParrotLLM/runs/big_run/exp_c/run_20260408_151821/checkpoints/best_loss_3p7192_epoch_0001_step_0002600.pt
+2026-04-08 15:44:36 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:44:39 - INFO - parrotllm.training - step   2610 | epoch 1 | loss 3.6840 | lr 1.08e-03 | grad 0.2228
+2026-04-08 15:44:42 - INFO - parrotllm.training - step   2620 | epoch 1 | loss 3.5813 | lr 1.06e-03 | grad 0.2199
+2026-04-08 15:44:45 - INFO - parrotllm.training - step   2630 | epoch 1 | loss 3.6373 | lr 1.04e-03 | grad 0.2208
+2026-04-08 15:44:48 - INFO - parrotllm.training - step   2640 | epoch 1 | loss 3.6257 | lr 1.02e-03 | grad 0.2077
+2026-04-08 15:44:51 - INFO - parrotllm.training - step   2650 | epoch 1 | loss 3.6178 | lr 9.95e-04 | grad 0.2063
+2026-04-08 15:44:54 - INFO - parrotllm.training - step   2660 | epoch 1 | loss 3.6992 | lr 9.74e-04 | grad 0.2146
+2026-04-08 15:44:58 - INFO - parrotllm.training - step   2670 | epoch 1 | loss 3.6155 | lr 9.52e-04 | grad 0.1953
+2026-04-08 15:45:01 - INFO - parrotllm.training - step   2680 | epoch 1 | loss 3.6776 | lr 9.31e-04 | grad 0.2371
+2026-04-08 15:45:04 - INFO - parrotllm.training - step   2690 | epoch 1 | loss 3.5047 | lr 9.09e-04 | grad 0.2456
+2026-04-08 15:45:07 - INFO - parrotllm.training - step   2700 | epoch 1 | loss 3.6258 | lr 8.88e-04 | grad 0.2154
+2026-04-08 15:45:07 - INFO - parrotllm.training - Starting evaluation...
+2026-04-08 15:45:07 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:45:12 - INFO - parrotllm.training -   Train: loss=3.6258, ppl=37.56
+2026-04-08 15:45:12 - INFO - parrotllm.training -   Val:   loss=3.6924, ppl=40.14
+2026-04-08 15:45:12 - INFO - parrotllm.training -   ** New best validation loss! **
+2026-04-08 15:45:13 - INFO - parrotllm.training -   Saved best checkpoint: /home/user/code/ParrotLLM/runs/big_run/exp_c/run_20260408_151821/checkpoints/best_loss_3p6924_epoch_0001_step_0002700.pt
+2026-04-08 15:45:14 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:45:17 - INFO - parrotllm.training - step   2710 | epoch 1 | loss 3.6489 | lr 8.66e-04 | grad 0.2351
+2026-04-08 15:45:20 - INFO - parrotllm.training - step   2720 | epoch 1 | loss 3.5517 | lr 8.45e-04 | grad 0.2217
+2026-04-08 15:45:23 - INFO - parrotllm.training - step   2730 | epoch 1 | loss 3.6574 | lr 8.23e-04 | grad 0.1996
+2026-04-08 15:45:26 - INFO - parrotllm.training - step   2740 | epoch 1 | loss 3.6018 | lr 8.02e-04 | grad 0.2048
+2026-04-08 15:45:29 - INFO - parrotllm.training - step   2750 | epoch 1 | loss 3.6005 | lr 7.80e-04 | grad 0.2141
+2026-04-08 15:45:30 - INFO - parrotllm.training - Saved checkpoint: /home/user/code/ParrotLLM/runs/big_run/exp_c/run_20260408_151821/checkpoints/last_epoch_0001_step_0002750.pt
+2026-04-08 15:45:34 - INFO - parrotllm.training - step   2760 | epoch 1 | loss 3.6176 | lr 7.59e-04 | grad 0.2111
+2026-04-08 15:45:37 - INFO - parrotllm.training - step   2770 | epoch 1 | loss 3.5393 | lr 7.37e-04 | grad 0.1982
+2026-04-08 15:45:40 - INFO - parrotllm.training - step   2780 | epoch 1 | loss 3.6184 | lr 7.16e-04 | grad 0.1871
+2026-04-08 15:45:44 - INFO - parrotllm.training - step   2790 | epoch 1 | loss 3.7416 | lr 6.94e-04 | grad 0.1905
+2026-04-08 15:45:47 - INFO - parrotllm.training - step   2800 | epoch 1 | loss 3.5801 | lr 6.73e-04 | grad 0.1969
+2026-04-08 15:45:47 - INFO - parrotllm.training - Starting evaluation...
+2026-04-08 15:45:47 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:45:52 - INFO - parrotllm.training -   Train: loss=3.5801, ppl=35.88
+2026-04-08 15:45:52 - INFO - parrotllm.training -   Val:   loss=3.6690, ppl=39.21
+2026-04-08 15:45:52 - INFO - parrotllm.training -   ** New best validation loss! **
+2026-04-08 15:45:53 - INFO - parrotllm.training -   Saved best checkpoint: /home/user/code/ParrotLLM/runs/big_run/exp_c/run_20260408_151821/checkpoints/best_loss_3p6690_epoch_0001_step_0002800.pt
+2026-04-08 15:45:54 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:45:57 - INFO - parrotllm.training - step   2810 | epoch 1 | loss 3.5645 | lr 6.51e-04 | grad 0.1819
+2026-04-08 15:46:00 - INFO - parrotllm.training - step   2820 | epoch 1 | loss 3.5655 | lr 6.30e-04 | grad 0.1871
+2026-04-08 15:46:03 - INFO - parrotllm.training - step   2830 | epoch 1 | loss 3.5574 | lr 6.08e-04 | grad 0.1823
+2026-04-08 15:46:06 - INFO - parrotllm.training - step   2840 | epoch 1 | loss 3.6654 | lr 5.87e-04 | grad 0.2037
+2026-04-08 15:46:09 - INFO - parrotllm.training - step   2850 | epoch 1 | loss 3.5174 | lr 5.65e-04 | grad 0.1844
+2026-04-08 15:46:12 - INFO - parrotllm.training - step   2860 | epoch 1 | loss 3.5785 | lr 5.44e-04 | grad 0.1926
+2026-04-08 15:46:15 - INFO - parrotllm.training - step   2870 | epoch 1 | loss 3.7967 | lr 5.22e-04 | grad 0.1913
+2026-04-08 15:46:19 - INFO - parrotllm.training - step   2880 | epoch 1 | loss 3.5314 | lr 5.00e-04 | grad 0.1864
+2026-04-08 15:46:22 - INFO - parrotllm.training - step   2890 | epoch 1 | loss 3.5295 | lr 4.79e-04 | grad 0.1902
+2026-04-08 15:46:25 - INFO - parrotllm.training - step   2900 | epoch 1 | loss 3.5369 | lr 4.57e-04 | grad 0.1818
+2026-04-08 15:46:25 - INFO - parrotllm.training - Starting evaluation...
+2026-04-08 15:46:25 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:46:30 - INFO - parrotllm.training -   Train: loss=3.5369, ppl=34.36
+2026-04-08 15:46:30 - INFO - parrotllm.training -   Val:   loss=3.6458, ppl=38.31
+2026-04-08 15:46:30 - INFO - parrotllm.training -   ** New best validation loss! **
+2026-04-08 15:46:31 - INFO - parrotllm.training -   Saved best checkpoint: /home/user/code/ParrotLLM/runs/big_run/exp_c/run_20260408_151821/checkpoints/best_loss_3p6458_epoch_0001_step_0002900.pt
+2026-04-08 15:46:32 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:46:35 - INFO - parrotllm.training - step   2910 | epoch 1 | loss 3.4898 | lr 4.36e-04 | grad 0.1758
+2026-04-08 15:46:38 - INFO - parrotllm.training - step   2920 | epoch 1 | loss 3.5773 | lr 4.14e-04 | grad 0.1747
+2026-04-08 15:46:41 - INFO - parrotllm.training - step   2930 | epoch 1 | loss 3.5991 | lr 3.93e-04 | grad 0.1673
+2026-04-08 15:46:44 - INFO - parrotllm.training - step   2940 | epoch 1 | loss 3.6129 | lr 3.71e-04 | grad 0.1826
+2026-04-08 15:46:47 - INFO - parrotllm.training - step   2950 | epoch 1 | loss 3.5709 | lr 3.50e-04 | grad 0.1754
+2026-04-08 15:46:50 - INFO - parrotllm.training - step   2960 | epoch 1 | loss 3.5649 | lr 3.28e-04 | grad 0.1891
+2026-04-08 15:46:53 - INFO - parrotllm.training - step   2970 | epoch 1 | loss 3.5021 | lr 3.07e-04 | grad 0.1828
+2026-04-08 15:46:56 - INFO - parrotllm.training - step   2980 | epoch 1 | loss 3.5723 | lr 2.85e-04 | grad 0.1765
+2026-04-08 15:47:00 - INFO - parrotllm.training - step   2990 | epoch 1 | loss 3.5628 | lr 2.64e-04 | grad 0.1583
+2026-04-08 15:47:03 - INFO - parrotllm.training - step   3000 | epoch 1 | loss 3.5245 | lr 2.42e-04 | grad 0.1615
+2026-04-08 15:47:03 - INFO - parrotllm.training - Starting evaluation...
+2026-04-08 15:47:03 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:47:08 - INFO - parrotllm.training -   Train: loss=3.5245, ppl=33.94
+2026-04-08 15:47:08 - INFO - parrotllm.training -   Val:   loss=3.6237, ppl=37.47
+2026-04-08 15:47:08 - INFO - parrotllm.training -   ** New best validation loss! **
+2026-04-08 15:47:09 - INFO - parrotllm.training -   Saved best checkpoint: /home/user/code/ParrotLLM/runs/big_run/exp_c/run_20260408_151821/checkpoints/best_loss_3p6237_epoch_0001_step_0003000.pt
+2026-04-08 15:47:10 - INFO - parrotllm.training - ------------------------------------------------------------
+2026-04-08 15:47:11 - INFO - parrotllm.training - Saved checkpoint: /home/user/code/ParrotLLM/runs/big_run/exp_c/run_20260408_151821/checkpoints/last_epoch_0001_step_0003000.pt
+2026-04-08 15:47:15 - INFO - parrotllm.training - step   3010 | epoch 1 | loss 3.5011 | lr 2.21e-04 | grad 0.1660
+2026-04-08 15:47:18 - INFO - parrotllm.training - step   3020 | epoch 1 | loss 3.5879 | lr 1.99e-04 | grad 0.1656
+2026-04-08 15:47:21 - INFO - parrotllm.training - step   3030 | epoch 1 | loss 3.5520 | lr 1.78e-04 | grad 0.1592
+2026-04-08 15:47:24 - INFO - parrotllm.training - step   3040 | epoch 1 | loss 3.4962 | lr 1.56e-04 | grad 0.1554
+2026-04-08 15:47:27 - INFO - parrotllm.training - step   3050 | epoch 1 | loss 3.5037 | lr 1.35e-04 | grad 0.1556
+2026-04-08 15:47:28 - INFO - parrotllm.training - Saved final checkpoint: /home/user/code/ParrotLLM/runs/big_run/exp_c/run_20260408_151821/checkpoints/last_epoch_0001_step_0003052.pt
+2026-04-08 15:47:28 - INFO - parrotllm.training - 
+Loss Curve:
+  Loss
+  ^
+   10.93 |*                                                 
+         |*                                                 
+         |*                                                 
+         |*                                                 
+         |**                                                
+         | ****                                             
+         |    *******                                       
+    3.43 |        ******************************************
+         +--------------------------------------------------> step
+         1                                             3052
+2026-04-08 15:47:28 - INFO - parrotllm.training - 
+============================================================
+TRAINING COMPLETE
+============================================================
+  Epochs: 2
+  Total steps: 3052
+  Total time: 0.48 hours
+  Best validation loss: 3.6237
+  Run directory: runs/big_run/exp_c/run_20260408_151821
+============================================================
+2026-04-08 15:47:28 - INFO - parrotllm.training - Uploading completed run directory to Hugging Face: repo=ParrotLabs/Preprocessed

runs/big_run/exp_c/run_20260408_151821/training_plots.pdf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:168fab2899bb14835aa7bdd5780d42086f4cabac72224a9e6b25fa0b77c3c08f
+size 112744