upload initial files

ck_report.json

@@ -0,0 +1,14 @@
+{
+  "training_examples (when pos_num=1 for ranking)": 130556,
+  "evaluation_steps": 200,
+  "train_batch_size": 16,
+  "epoch": 1,
+  "total_epochs": 5,
+  "steps": 8000,
+  "saved_at_total_steps": 8000,
+  "steps_per_epoch": 8160,
+  "eval_scores_on_dev": {
+    "loss": 1.1733529567718506,
+    "perplexity": 3.232813835144043
+  }
+}

hyperparams.json

@@ -0,0 +1,21 @@
+{
+  "model_select": "distilgpt2",
+  "dataset_name": "source_code",
+  "per_gpu_train_batch_size": 4,
+  "dev_batch_size": 8,
+  "num_epochs_train": 5,
+  "max_seq_length": 256,
+  "lr": 2e-05,
+  "warmup_ratio": 0.2,
+  "early_stop": 3,
+  "scheduler": "warmuplinear",
+  "seed": 122,
+  "accumulation_steps": 1,
+  "n_gpu": 4,
+  "visiable_device": "0",
+  "evaluation_steps": 200,
+  "wandb_project_name": "code_generate",
+  "restore_training": false,
+  "with_wandb": true,
+  "wandb_run_name": "model/distilgpt2_fine_tuned_coder"
+}

log.out

@@ -0,0 +1,889 @@
+2024-04-26 02:18:42,687 - trainer - INFO - Use pytorch device: cuda, with gpu_number=4
+2024-04-26 02:18:42,687 - trainer - INFO -    see seed for random, numpy and torch 122
+2024-04-26 02:18:43,540 - trainer - INFO - module.0.gpt.transformer.wte.weight	torch.Size([50259, 768])
+2024-04-26 02:18:43,540 - trainer - INFO - module.0.gpt.transformer.wpe.weight	torch.Size([1024, 768])
+2024-04-26 02:18:43,542 - trainer - INFO - module.0.gpt.transformer.h.0.ln_1.weight	torch.Size([768])
+2024-04-26 02:18:43,542 - trainer - INFO - module.0.gpt.transformer.h.0.ln_1.bias	torch.Size([768])
+2024-04-26 02:18:43,543 - trainer - INFO - module.0.gpt.transformer.h.0.attn.c_attn.weight	torch.Size([768, 2304])
+2024-04-26 02:18:43,543 - trainer - INFO - module.0.gpt.transformer.h.0.attn.c_attn.bias	torch.Size([2304])
+2024-04-26 02:18:43,544 - trainer - INFO - module.0.gpt.transformer.h.0.attn.c_proj.weight	torch.Size([768, 768])
+2024-04-26 02:18:43,544 - trainer - INFO - module.0.gpt.transformer.h.0.attn.c_proj.bias	torch.Size([768])
+2024-04-26 02:18:43,544 - trainer - INFO - module.0.gpt.transformer.h.0.ln_2.weight	torch.Size([768])
+2024-04-26 02:18:43,545 - trainer - INFO - module.0.gpt.transformer.h.0.ln_2.bias	torch.Size([768])
+2024-04-26 02:18:43,545 - trainer - INFO - module.0.gpt.transformer.h.0.mlp.c_fc.weight	torch.Size([768, 3072])
+2024-04-26 02:18:43,546 - trainer - INFO - module.0.gpt.transformer.h.0.mlp.c_fc.bias	torch.Size([3072])
+2024-04-26 02:18:43,546 - trainer - INFO - module.0.gpt.transformer.h.0.mlp.c_proj.weight	torch.Size([3072, 768])
+2024-04-26 02:18:43,547 - trainer - INFO - module.0.gpt.transformer.h.0.mlp.c_proj.bias	torch.Size([768])
+2024-04-26 02:18:43,547 - trainer - INFO - module.0.gpt.transformer.h.1.ln_1.weight	torch.Size([768])
+2024-04-26 02:18:43,547 - trainer - INFO - module.0.gpt.transformer.h.1.ln_1.bias	torch.Size([768])
+2024-04-26 02:18:43,548 - trainer - INFO - module.0.gpt.transformer.h.1.attn.c_attn.weight	torch.Size([768, 2304])
+2024-04-26 02:18:43,548 - trainer - INFO - module.0.gpt.transformer.h.1.attn.c_attn.bias	torch.Size([2304])
+2024-04-26 02:18:43,549 - trainer - INFO - module.0.gpt.transformer.h.1.attn.c_proj.weight	torch.Size([768, 768])
+2024-04-26 02:18:43,549 - trainer - INFO - module.0.gpt.transformer.h.1.attn.c_proj.bias	torch.Size([768])
+2024-04-26 02:18:43,549 - trainer - INFO - module.0.gpt.transformer.h.1.ln_2.weight	torch.Size([768])
+2024-04-26 02:18:43,550 - trainer - INFO - module.0.gpt.transformer.h.1.ln_2.bias	torch.Size([768])
+2024-04-26 02:18:43,550 - trainer - INFO - module.0.gpt.transformer.h.1.mlp.c_fc.weight	torch.Size([768, 3072])
+2024-04-26 02:18:43,551 - trainer - INFO - module.0.gpt.transformer.h.1.mlp.c_fc.bias	torch.Size([3072])
+2024-04-26 02:18:43,551 - trainer - INFO - module.0.gpt.transformer.h.1.mlp.c_proj.weight	torch.Size([3072, 768])
+2024-04-26 02:18:43,551 - trainer - INFO - module.0.gpt.transformer.h.1.mlp.c_proj.bias	torch.Size([768])
+2024-04-26 02:18:43,552 - trainer - INFO - module.0.gpt.transformer.h.2.ln_1.weight	torch.Size([768])
+2024-04-26 02:18:43,552 - trainer - INFO - module.0.gpt.transformer.h.2.ln_1.bias	torch.Size([768])
+2024-04-26 02:18:43,553 - trainer - INFO - module.0.gpt.transformer.h.2.attn.c_attn.weight	torch.Size([768, 2304])
+2024-04-26 02:18:43,553 - trainer - INFO - module.0.gpt.transformer.h.2.attn.c_attn.bias	torch.Size([2304])
+2024-04-26 02:18:43,554 - trainer - INFO - module.0.gpt.transformer.h.2.attn.c_proj.weight	torch.Size([768, 768])
+2024-04-26 02:18:43,554 - trainer - INFO - module.0.gpt.transformer.h.2.attn.c_proj.bias	torch.Size([768])
+2024-04-26 02:18:43,554 - trainer - INFO - module.0.gpt.transformer.h.2.ln_2.weight	torch.Size([768])
+2024-04-26 02:18:43,555 - trainer - INFO - module.0.gpt.transformer.h.2.ln_2.bias	torch.Size([768])
+2024-04-26 02:18:43,555 - trainer - INFO - module.0.gpt.transformer.h.2.mlp.c_fc.weight	torch.Size([768, 3072])
+2024-04-26 02:18:43,555 - trainer - INFO - module.0.gpt.transformer.h.2.mlp.c_fc.bias	torch.Size([3072])
+2024-04-26 02:18:43,556 - trainer - INFO - module.0.gpt.transformer.h.2.mlp.c_proj.weight	torch.Size([3072, 768])
+2024-04-26 02:18:43,556 - trainer - INFO - module.0.gpt.transformer.h.2.mlp.c_proj.bias	torch.Size([768])
+2024-04-26 02:18:43,557 - trainer - INFO - module.0.gpt.transformer.h.3.ln_1.weight	torch.Size([768])
+2024-04-26 02:18:43,557 - trainer - INFO - module.0.gpt.transformer.h.3.ln_1.bias	torch.Size([768])
+2024-04-26 02:18:43,558 - trainer - INFO - module.0.gpt.transformer.h.3.attn.c_attn.weight	torch.Size([768, 2304])
+2024-04-26 02:18:43,558 - trainer - INFO - module.0.gpt.transformer.h.3.attn.c_attn.bias	torch.Size([2304])
+2024-04-26 02:18:43,559 - trainer - INFO - module.0.gpt.transformer.h.3.attn.c_proj.weight	torch.Size([768, 768])
+2024-04-26 02:18:43,559 - trainer - INFO - module.0.gpt.transformer.h.3.attn.c_proj.bias	torch.Size([768])
+2024-04-26 02:18:43,559 - trainer - INFO - module.0.gpt.transformer.h.3.ln_2.weight	torch.Size([768])
+2024-04-26 02:18:43,560 - trainer - INFO - module.0.gpt.transformer.h.3.ln_2.bias	torch.Size([768])
+2024-04-26 02:18:43,560 - trainer - INFO - module.0.gpt.transformer.h.3.mlp.c_fc.weight	torch.Size([768, 3072])
+2024-04-26 02:18:43,561 - trainer - INFO - module.0.gpt.transformer.h.3.mlp.c_fc.bias	torch.Size([3072])
+2024-04-26 02:18:43,561 - trainer - INFO - module.0.gpt.transformer.h.3.mlp.c_proj.weight	torch.Size([3072, 768])
+2024-04-26 02:18:43,562 - trainer - INFO - module.0.gpt.transformer.h.3.mlp.c_proj.bias	torch.Size([768])
+2024-04-26 02:18:43,562 - trainer - INFO - module.0.gpt.transformer.h.4.ln_1.weight	torch.Size([768])
+2024-04-26 02:18:43,562 - trainer - INFO - module.0.gpt.transformer.h.4.ln_1.bias	torch.Size([768])
+2024-04-26 02:18:43,563 - trainer - INFO - module.0.gpt.transformer.h.4.attn.c_attn.weight	torch.Size([768, 2304])
+2024-04-26 02:18:43,563 - trainer - INFO - module.0.gpt.transformer.h.4.attn.c_attn.bias	torch.Size([2304])
+2024-04-26 02:18:43,564 - trainer - INFO - module.0.gpt.transformer.h.4.attn.c_proj.weight	torch.Size([768, 768])
+2024-04-26 02:18:43,564 - trainer - INFO - module.0.gpt.transformer.h.4.attn.c_proj.bias	torch.Size([768])
+2024-04-26 02:18:43,564 - trainer - INFO - module.0.gpt.transformer.h.4.ln_2.weight	torch.Size([768])
+2024-04-26 02:18:43,565 - trainer - INFO - module.0.gpt.transformer.h.4.ln_2.bias	torch.Size([768])
+2024-04-26 02:18:43,565 - trainer - INFO - module.0.gpt.transformer.h.4.mlp.c_fc.weight	torch.Size([768, 3072])
+2024-04-26 02:18:43,566 - trainer - INFO - module.0.gpt.transformer.h.4.mlp.c_fc.bias	torch.Size([3072])
+2024-04-26 02:18:43,566 - trainer - INFO - module.0.gpt.transformer.h.4.mlp.c_proj.weight	torch.Size([3072, 768])
+2024-04-26 02:18:43,567 - trainer - INFO - module.0.gpt.transformer.h.4.mlp.c_proj.bias	torch.Size([768])
+2024-04-26 02:18:43,567 - trainer - INFO - module.0.gpt.transformer.h.5.ln_1.weight	torch.Size([768])
+2024-04-26 02:18:43,567 - trainer - INFO - module.0.gpt.transformer.h.5.ln_1.bias	torch.Size([768])
+2024-04-26 02:18:43,568 - trainer - INFO - module.0.gpt.transformer.h.5.attn.c_attn.weight	torch.Size([768, 2304])
+2024-04-26 02:18:43,568 - trainer - INFO - module.0.gpt.transformer.h.5.attn.c_attn.bias	torch.Size([2304])
+2024-04-26 02:18:43,569 - trainer - INFO - module.0.gpt.transformer.h.5.attn.c_proj.weight	torch.Size([768, 768])
+2024-04-26 02:18:43,569 - trainer - INFO - module.0.gpt.transformer.h.5.attn.c_proj.bias	torch.Size([768])
+2024-04-26 02:18:43,570 - trainer - INFO - module.0.gpt.transformer.h.5.ln_2.weight	torch.Size([768])
+2024-04-26 02:18:43,570 - trainer - INFO - module.0.gpt.transformer.h.5.ln_2.bias	torch.Size([768])
+2024-04-26 02:18:43,570 - trainer - INFO - module.0.gpt.transformer.h.5.mlp.c_fc.weight	torch.Size([768, 3072])
+2024-04-26 02:18:43,571 - trainer - INFO - module.0.gpt.transformer.h.5.mlp.c_fc.bias	torch.Size([3072])
+2024-04-26 02:18:43,571 - trainer - INFO - module.0.gpt.transformer.h.5.mlp.c_proj.weight	torch.Size([3072, 768])
+2024-04-26 02:18:43,572 - trainer - INFO - module.0.gpt.transformer.h.5.mlp.c_proj.bias	torch.Size([768])
+2024-04-26 02:18:43,572 - trainer - INFO - module.0.gpt.transformer.ln_f.weight	torch.Size([768])
+2024-04-26 02:18:43,573 - trainer - INFO - module.0.gpt.transformer.ln_f.bias	torch.Size([768])
+2024-04-26 02:18:43,573 - trainer - INFO - module.0.gpt.lm_head.weight	torch.Size([50259, 768])
+2024-04-26 02:18:43,573 - trainer - INFO - DataParallel(
+  (module): Sequential(
+    (0): GPTSingleHead(
+      (gpt): GPT2LMHeadModel(
+        (transformer): GPT2Model(
+          (wte): Embedding(50259, 768)
+          (wpe): Embedding(1024, 768)
+          (drop): Dropout(p=0.1, inplace=False)
+          (h): ModuleList(
+            (0-5): 6 x GPT2Block(
+              (ln_1): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+              (attn): GPT2Attention(
+                (c_attn): Conv1D()
+                (c_proj): Conv1D()
+                (attn_dropout): Dropout(p=0.1, inplace=False)
+                (resid_dropout): Dropout(p=0.1, inplace=False)
+              )
+              (ln_2): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+              (mlp): GPT2MLP(
+                (c_fc): Conv1D()
+                (c_proj): Conv1D()
+                (act): NewGELUActivation()
+                (dropout): Dropout(p=0.1, inplace=False)
+              )
+            )
+          )
+          (ln_f): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
+        )
+        (lm_head): Linear(in_features=768, out_features=50259, bias=False)
+      )
+    )
+    (1): EmptyHeads()
+  )
+)
+2024-04-26 02:18:43,574 - trainer - INFO -   Total params: 81914112
+2024-04-26 02:18:43,574 - trainer - INFO -   Trainable params: 81914112
+2024-04-26 02:18:43,574 - trainer - INFO -   Non-trainable params: 0
+2024-04-26 02:18:43,590 - trainer - INFO -    Warmup-steps: 8160
+2024-04-26 02:18:43,594 - trainer - INFO - ***** Running training *****
+2024-04-26 02:18:43,594 - trainer - INFO -   Num of training examples (actually iterations per epoch for Iterable Dataset) = 130556
+2024-04-26 02:18:43,594 - trainer - INFO -   Output path (short): tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 02:18:43,594 - trainer - INFO -   Steps per Epoch = 8160 or iterations per epoch = 8160
+2024-04-26 02:18:43,594 - trainer - INFO -   Num of Epochs = 5
+2024-04-26 02:18:43,594 - trainer - INFO -   Best score (perplexity) = -inf
+2024-04-26 02:18:43,594 - trainer - INFO -   Eval every 200 steps or every 200 iterations
+2024-04-26 02:18:43,594 - trainer - INFO -   Early stop = 3
+2024-04-26 02:18:43,594 - trainer - INFO -   Gradient Accumulation steps = 1
+2024-04-26 02:18:43,594 - trainer - INFO -   Total optimization steps = 40800
+2024-04-26 02:18:43,594 - trainer - INFO -   Instantaneous batch size per GPU = 4 and n_gpu = 4 so the input batch size = 16
+2024-04-26 02:25:03,634 - trainer - INFO -    Save model to tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 02:25:04,787 - trainer - INFO -   Save check-point at epoch=0 step=200
+2024-04-26 02:25:04,788 - trainer - INFO -    ***** Evaluation report *****
+2024-04-26 02:25:04,788 - trainer - INFO -   Output path (short): tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 02:25:04,788 - trainer - INFO -   Early stop on: perplexity
+2024-04-26 02:25:04,788 - trainer - INFO -   Early stop count = 0/3
+2024-04-26 02:25:04,788 - trainer - INFO -   Eval steps = 200 or (iterations = 200)
+2024-04-26 02:25:04,788 - trainer - INFO -   Best score (perplexity) = -270.8600158691406
+2024-04-26 02:25:04,788 - trainer - INFO -   Gradient Accumulation steps = 1
+2024-04-26 02:25:04,788 - trainer - INFO -   Num of training examples (actually no. of iterations per epoch for Iterable Dataset)  = 130556
+2024-04-26 02:25:04,788 - trainer - INFO -   Num of development examples (actually no. of iterations per epoch for Iterable Dataset) = 14507
+2024-04-26 02:25:04,788 - trainer - INFO -   Time spent since last evaluation = 0h 6m 21s
+2024-04-26 02:25:04,788 - trainer - INFO -   Epoch = 1/5
+2024-04-26 02:25:04,788 - trainer - INFO -   Steps = 200/40800
+2024-04-26 02:25:04,788 - trainer - INFO -   Instantaneous batch size per GPU = 4 and n_gpu = 4 so the input batch size = 16
+2024-04-26 02:25:04,788 - trainer - INFO -   dev_loss = 5.601602	||	 dev_eval_scores = {'perplexity': 270.8600158691406}
+2024-04-26 02:25:04,789 - trainer - INFO -   train_loss = 14.094216346740723
+2024-04-26 02:25:04,789 - trainer - INFO - 
+********************************************
+2024-04-26 02:31:25,346 - trainer - INFO -    Save model to tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 02:31:31,186 - trainer - INFO -   Save check-point at epoch=0 step=400
+2024-04-26 02:31:31,187 - trainer - INFO -    ***** Evaluation report *****
+2024-04-26 02:31:31,187 - trainer - INFO -   Output path (short): tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 02:31:31,187 - trainer - INFO -   Early stop on: perplexity
+2024-04-26 02:31:31,187 - trainer - INFO -   Early stop count = 0/3
+2024-04-26 02:31:31,187 - trainer - INFO -   Eval steps = 200 or (iterations = 200)
+2024-04-26 02:31:31,187 - trainer - INFO -   Best score (perplexity) = -10.156302452087402
+2024-04-26 02:31:31,187 - trainer - INFO -   Gradient Accumulation steps = 1
+2024-04-26 02:31:31,187 - trainer - INFO -   Num of training examples (actually no. of iterations per epoch for Iterable Dataset)  = 130556
+2024-04-26 02:31:31,187 - trainer - INFO -   Num of development examples (actually no. of iterations per epoch for Iterable Dataset) = 14507
+2024-04-26 02:31:31,187 - trainer - INFO -   Time spent since last evaluation = 0h 6m 26s
+2024-04-26 02:31:31,187 - trainer - INFO -   Epoch = 1/5
+2024-04-26 02:31:31,187 - trainer - INFO -   Steps = 400/40800
+2024-04-26 02:31:31,187 - trainer - INFO -   Instantaneous batch size per GPU = 4 and n_gpu = 4 so the input batch size = 16
+2024-04-26 02:31:31,187 - trainer - INFO -   dev_loss = 2.318094	||	 dev_eval_scores = {'perplexity': 10.156302452087402}
+2024-04-26 02:31:31,220 - trainer - INFO -   train_loss = 8.5648775100708
+2024-04-26 02:31:31,220 - trainer - INFO - 
+********************************************
+2024-04-26 02:37:51,756 - trainer - INFO -    Save model to tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 02:37:57,711 - trainer - INFO -   Save check-point at epoch=0 step=600
+2024-04-26 02:37:57,711 - trainer - INFO -    ***** Evaluation report *****
+2024-04-26 02:37:57,711 - trainer - INFO -   Output path (short): tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 02:37:57,711 - trainer - INFO -   Early stop on: perplexity
+2024-04-26 02:37:57,711 - trainer - INFO -   Early stop count = 0/3
+2024-04-26 02:37:57,711 - trainer - INFO -   Eval steps = 200 or (iterations = 200)
+2024-04-26 02:37:57,711 - trainer - INFO -   Best score (perplexity) = -7.607259750366211
+2024-04-26 02:37:57,712 - trainer - INFO -   Gradient Accumulation steps = 1
+2024-04-26 02:37:57,712 - trainer - INFO -   Num of training examples (actually no. of iterations per epoch for Iterable Dataset)  = 130556
+2024-04-26 02:37:57,712 - trainer - INFO -   Num of development examples (actually no. of iterations per epoch for Iterable Dataset) = 14507
+2024-04-26 02:37:57,712 - trainer - INFO -   Time spent since last evaluation = 0h 6m 26s
+2024-04-26 02:37:57,712 - trainer - INFO -   Epoch = 1/5
+2024-04-26 02:37:57,712 - trainer - INFO -   Steps = 600/40800
+2024-04-26 02:37:57,712 - trainer - INFO -   Instantaneous batch size per GPU = 4 and n_gpu = 4 so the input batch size = 16
+2024-04-26 02:37:57,712 - trainer - INFO -   dev_loss = 2.029103	||	 dev_eval_scores = {'perplexity': 7.607259750366211}
+2024-04-26 02:37:57,712 - trainer - INFO -   train_loss = 6.4544525146484375
+2024-04-26 02:37:57,712 - trainer - INFO - 
+********************************************
+2024-04-26 02:44:17,920 - trainer - INFO -    Save model to tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 02:44:23,766 - trainer - INFO -   Save check-point at epoch=0 step=800
+2024-04-26 02:44:23,766 - trainer - INFO -    ***** Evaluation report *****
+2024-04-26 02:44:23,767 - trainer - INFO -   Output path (short): tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 02:44:23,767 - trainer - INFO -   Early stop on: perplexity
+2024-04-26 02:44:23,767 - trainer - INFO -   Early stop count = 0/3
+2024-04-26 02:44:23,767 - trainer - INFO -   Eval steps = 200 or (iterations = 200)
+2024-04-26 02:44:23,767 - trainer - INFO -   Best score (perplexity) = -6.791029453277588
+2024-04-26 02:44:23,767 - trainer - INFO -   Gradient Accumulation steps = 1
+2024-04-26 02:44:23,767 - trainer - INFO -   Num of training examples (actually no. of iterations per epoch for Iterable Dataset)  = 130556
+2024-04-26 02:44:23,767 - trainer - INFO -   Num of development examples (actually no. of iterations per epoch for Iterable Dataset) = 14507
+2024-04-26 02:44:23,767 - trainer - INFO -   Time spent since last evaluation = 0h 6m 26s
+2024-04-26 02:44:23,767 - trainer - INFO -   Epoch = 1/5
+2024-04-26 02:44:23,767 - trainer - INFO -   Steps = 800/40800
+2024-04-26 02:44:23,767 - trainer - INFO -   Instantaneous batch size per GPU = 4 and n_gpu = 4 so the input batch size = 16
+2024-04-26 02:44:23,767 - trainer - INFO -   dev_loss = 1.915603	||	 dev_eval_scores = {'perplexity': 6.791029453277588}
+2024-04-26 02:44:23,767 - trainer - INFO -   train_loss = 5.3493781089782715
+2024-04-26 02:44:23,768 - trainer - INFO - 
+********************************************
+2024-04-26 02:50:43,526 - trainer - INFO -    Save model to tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 02:50:49,464 - trainer - INFO -   Save check-point at epoch=0 step=1000
+2024-04-26 02:50:49,464 - trainer - INFO -    ***** Evaluation report *****
+2024-04-26 02:50:49,464 - trainer - INFO -   Output path (short): tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 02:50:49,464 - trainer - INFO -   Early stop on: perplexity
+2024-04-26 02:50:49,464 - trainer - INFO -   Early stop count = 0/3
+2024-04-26 02:50:49,464 - trainer - INFO -   Eval steps = 200 or (iterations = 200)
+2024-04-26 02:50:49,464 - trainer - INFO -   Best score (perplexity) = -6.073063373565674
+2024-04-26 02:50:49,464 - trainer - INFO -   Gradient Accumulation steps = 1
+2024-04-26 02:50:49,464 - trainer - INFO -   Num of training examples (actually no. of iterations per epoch for Iterable Dataset)  = 130556
+2024-04-26 02:50:49,464 - trainer - INFO -   Num of development examples (actually no. of iterations per epoch for Iterable Dataset) = 14507
+2024-04-26 02:50:49,464 - trainer - INFO -   Time spent since last evaluation = 0h 6m 25s
+2024-04-26 02:50:49,465 - trainer - INFO -   Epoch = 1/5
+2024-04-26 02:50:49,465 - trainer - INFO -   Steps = 1000/40800
+2024-04-26 02:50:49,465 - trainer - INFO -   Instantaneous batch size per GPU = 4 and n_gpu = 4 so the input batch size = 16
+2024-04-26 02:50:49,465 - trainer - INFO -   dev_loss = 1.803863	||	 dev_eval_scores = {'perplexity': 6.073063373565674}
+2024-04-26 02:50:49,465 - trainer - INFO -   train_loss = 4.66662073135376
+2024-04-26 02:50:49,465 - trainer - INFO - 
+********************************************
+2024-04-26 02:57:09,707 - trainer - INFO -    ***** Evaluation report *****
+2024-04-26 02:57:09,707 - trainer - INFO -   Output path (short): tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 02:57:09,707 - trainer - INFO -   Early stop on: perplexity
+2024-04-26 02:57:09,707 - trainer - INFO -   Early stop count = 1/3
+2024-04-26 02:57:09,707 - trainer - INFO -   Eval steps = 200 or (iterations = 200)
+2024-04-26 02:57:09,707 - trainer - INFO -   Best score (perplexity) = -6.073063373565674
+2024-04-26 02:57:09,708 - trainer - INFO -   Gradient Accumulation steps = 1
+2024-04-26 02:57:09,708 - trainer - INFO -   Num of training examples (actually no. of iterations per epoch for Iterable Dataset)  = 130556
+2024-04-26 02:57:09,708 - trainer - INFO -   Num of development examples (actually no. of iterations per epoch for Iterable Dataset) = 14507
+2024-04-26 02:57:09,708 - trainer - INFO -   Time spent since last evaluation = 0h 6m 20s
+2024-04-26 02:57:09,708 - trainer - INFO -   Epoch = 1/5
+2024-04-26 02:57:09,708 - trainer - INFO -   Steps = 1200/40800
+2024-04-26 02:57:09,708 - trainer - INFO -   Instantaneous batch size per GPU = 4 and n_gpu = 4 so the input batch size = 16
+2024-04-26 02:57:09,708 - trainer - INFO -   dev_loss = 1.808444	||	 dev_eval_scores = {'perplexity': 6.100945472717285}
+2024-04-26 02:57:09,708 - trainer - INFO -   train_loss = 4.205338001251221
+2024-04-26 02:57:09,708 - trainer - INFO - 
+********************************************
+2024-04-26 03:03:30,335 - trainer - INFO -    Save model to tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 03:03:36,292 - trainer - INFO -   Save check-point at epoch=0 step=1400
+2024-04-26 03:03:36,292 - trainer - INFO -    ***** Evaluation report *****
+2024-04-26 03:03:36,292 - trainer - INFO -   Output path (short): tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 03:03:36,292 - trainer - INFO -   Early stop on: perplexity
+2024-04-26 03:03:36,292 - trainer - INFO -   Early stop count = 0/3
+2024-04-26 03:03:36,292 - trainer - INFO -   Eval steps = 200 or (iterations = 200)
+2024-04-26 03:03:36,292 - trainer - INFO -   Best score (perplexity) = -5.51066780090332
+2024-04-26 03:03:36,292 - trainer - INFO -   Gradient Accumulation steps = 1
+2024-04-26 03:03:36,292 - trainer - INFO -   Num of training examples (actually no. of iterations per epoch for Iterable Dataset)  = 130556
+2024-04-26 03:03:36,292 - trainer - INFO -   Num of development examples (actually no. of iterations per epoch for Iterable Dataset) = 14507
+2024-04-26 03:03:36,292 - trainer - INFO -   Time spent since last evaluation = 0h 6m 26s
+2024-04-26 03:03:36,293 - trainer - INFO -   Epoch = 1/5
+2024-04-26 03:03:36,293 - trainer - INFO -   Steps = 1400/40800
+2024-04-26 03:03:36,293 - trainer - INFO -   Instantaneous batch size per GPU = 4 and n_gpu = 4 so the input batch size = 16
+2024-04-26 03:03:36,293 - trainer - INFO -   dev_loss = 1.706686	||	 dev_eval_scores = {'perplexity': 5.51066780090332}
+2024-04-26 03:03:36,293 - trainer - INFO -   train_loss = 3.8646857738494873
+2024-04-26 03:03:36,293 - trainer - INFO - 
+********************************************
+2024-04-26 03:09:56,141 - trainer - INFO -    Save model to tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 03:10:02,087 - trainer - INFO -   Save check-point at epoch=0 step=1600
+2024-04-26 03:10:02,087 - trainer - INFO -    ***** Evaluation report *****
+2024-04-26 03:10:02,088 - trainer - INFO -   Output path (short): tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 03:10:02,088 - trainer - INFO -   Early stop on: perplexity
+2024-04-26 03:10:02,088 - trainer - INFO -   Early stop count = 0/3
+2024-04-26 03:10:02,088 - trainer - INFO -   Eval steps = 200 or (iterations = 200)
+2024-04-26 03:10:02,088 - trainer - INFO -   Best score (perplexity) = -5.361582279205322
+2024-04-26 03:10:02,088 - trainer - INFO -   Gradient Accumulation steps = 1
+2024-04-26 03:10:02,088 - trainer - INFO -   Num of training examples (actually no. of iterations per epoch for Iterable Dataset)  = 130556
+2024-04-26 03:10:02,088 - trainer - INFO -   Num of development examples (actually no. of iterations per epoch for Iterable Dataset) = 14507
+2024-04-26 03:10:02,088 - trainer - INFO -   Time spent since last evaluation = 0h 6m 25s
+2024-04-26 03:10:02,088 - trainer - INFO -   Epoch = 1/5
+2024-04-26 03:10:02,088 - trainer - INFO -   Steps = 1600/40800
+2024-04-26 03:10:02,088 - trainer - INFO -   Instantaneous batch size per GPU = 4 and n_gpu = 4 so the input batch size = 16
+2024-04-26 03:10:02,088 - trainer - INFO -   dev_loss = 1.679259	||	 dev_eval_scores = {'perplexity': 5.361582279205322}
+2024-04-26 03:10:02,088 - trainer - INFO -   train_loss = 3.60662579536438
+2024-04-26 03:10:02,089 - trainer - INFO - 
+********************************************
+2024-04-26 03:16:22,224 - trainer - INFO -    Save model to tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 03:16:28,030 - trainer - INFO -   Save check-point at epoch=0 step=1800
+2024-04-26 03:16:28,030 - trainer - INFO -    ***** Evaluation report *****
+2024-04-26 03:16:28,030 - trainer - INFO -   Output path (short): tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 03:16:28,030 - trainer - INFO -   Early stop on: perplexity
+2024-04-26 03:16:28,030 - trainer - INFO -   Early stop count = 0/3
+2024-04-26 03:16:28,030 - trainer - INFO -   Eval steps = 200 or (iterations = 200)
+2024-04-26 03:16:28,030 - trainer - INFO -   Best score (perplexity) = -5.1808762550354
+2024-04-26 03:16:28,031 - trainer - INFO -   Gradient Accumulation steps = 1
+2024-04-26 03:16:28,031 - trainer - INFO -   Num of training examples (actually no. of iterations per epoch for Iterable Dataset)  = 130556
+2024-04-26 03:16:28,031 - trainer - INFO -   Num of development examples (actually no. of iterations per epoch for Iterable Dataset) = 14507
+2024-04-26 03:16:28,031 - trainer - INFO -   Time spent since last evaluation = 0h 6m 25s
+2024-04-26 03:16:28,031 - trainer - INFO -   Epoch = 1/5
+2024-04-26 03:16:28,031 - trainer - INFO -   Steps = 1800/40800
+2024-04-26 03:16:28,031 - trainer - INFO -   Instantaneous batch size per GPU = 4 and n_gpu = 4 so the input batch size = 16
+2024-04-26 03:16:28,031 - trainer - INFO -   dev_loss = 1.644974	||	 dev_eval_scores = {'perplexity': 5.1808762550354}
+2024-04-26 03:16:28,031 - trainer - INFO -   train_loss = 3.401608943939209
+2024-04-26 03:16:28,031 - trainer - INFO - 
+********************************************
+2024-04-26 03:22:47,629 - trainer - INFO -    Save model to tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 03:22:53,548 - trainer - INFO -   Save check-point at epoch=0 step=2000
+2024-04-26 03:22:53,549 - trainer - INFO -    ***** Evaluation report *****
+2024-04-26 03:22:53,549 - trainer - INFO -   Output path (short): tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 03:22:53,549 - trainer - INFO -   Early stop on: perplexity
+2024-04-26 03:22:53,549 - trainer - INFO -   Early stop count = 0/3
+2024-04-26 03:22:53,549 - trainer - INFO -   Eval steps = 200 or (iterations = 200)
+2024-04-26 03:22:53,549 - trainer - INFO -   Best score (perplexity) = -4.970845699310303
+2024-04-26 03:22:53,549 - trainer - INFO -   Gradient Accumulation steps = 1
+2024-04-26 03:22:53,549 - trainer - INFO -   Num of training examples (actually no. of iterations per epoch for Iterable Dataset)  = 130556
+2024-04-26 03:22:53,549 - trainer - INFO -   Num of development examples (actually no. of iterations per epoch for Iterable Dataset) = 14507
+2024-04-26 03:22:53,549 - trainer - INFO -   Time spent since last evaluation = 0h 6m 25s
+2024-04-26 03:22:53,549 - trainer - INFO -   Epoch = 1/5
+2024-04-26 03:22:53,549 - trainer - INFO -   Steps = 2000/40800
+2024-04-26 03:22:53,549 - trainer - INFO -   Instantaneous batch size per GPU = 4 and n_gpu = 4 so the input batch size = 16
+2024-04-26 03:22:53,549 - trainer - INFO -   dev_loss = 1.603590	||	 dev_eval_scores = {'perplexity': 4.970845699310303}
+2024-04-26 03:22:53,550 - trainer - INFO -   train_loss = 3.2337915897369385
+2024-04-26 03:22:53,550 - trainer - INFO - 
+********************************************
+2024-04-26 03:29:13,045 - trainer - INFO -    Save model to tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 03:29:18,974 - trainer - INFO -   Save check-point at epoch=0 step=2200
+2024-04-26 03:29:18,975 - trainer - INFO -    ***** Evaluation report *****
+2024-04-26 03:29:18,975 - trainer - INFO -   Output path (short): tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 03:29:18,975 - trainer - INFO -   Early stop on: perplexity
+2024-04-26 03:29:18,975 - trainer - INFO -   Early stop count = 0/3
+2024-04-26 03:29:18,975 - trainer - INFO -   Eval steps = 200 or (iterations = 200)
+2024-04-26 03:29:18,975 - trainer - INFO -   Best score (perplexity) = -4.858333587646484
+2024-04-26 03:29:18,975 - trainer - INFO -   Gradient Accumulation steps = 1
+2024-04-26 03:29:18,975 - trainer - INFO -   Num of training examples (actually no. of iterations per epoch for Iterable Dataset)  = 130556
+2024-04-26 03:29:18,975 - trainer - INFO -   Num of development examples (actually no. of iterations per epoch for Iterable Dataset) = 14507
+2024-04-26 03:29:18,976 - trainer - INFO -   Time spent since last evaluation = 0h 6m 25s
+2024-04-26 03:29:18,976 - trainer - INFO -   Epoch = 1/5
+2024-04-26 03:29:18,976 - trainer - INFO -   Steps = 2200/40800
+2024-04-26 03:29:18,976 - trainer - INFO -   Instantaneous batch size per GPU = 4 and n_gpu = 4 so the input batch size = 16
+2024-04-26 03:29:18,976 - trainer - INFO -   dev_loss = 1.580696	||	 dev_eval_scores = {'perplexity': 4.858333587646484}
+2024-04-26 03:29:18,976 - trainer - INFO -   train_loss = 3.092155694961548
+2024-04-26 03:29:18,976 - trainer - INFO - 
+********************************************
+2024-04-26 03:35:38,899 - trainer - INFO -    Save model to tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 03:35:44,832 - trainer - INFO -   Save check-point at epoch=0 step=2400
+2024-04-26 03:35:44,832 - trainer - INFO -    ***** Evaluation report *****
+2024-04-26 03:35:44,832 - trainer - INFO -   Output path (short): tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 03:35:44,832 - trainer - INFO -   Early stop on: perplexity
+2024-04-26 03:35:44,832 - trainer - INFO -   Early stop count = 0/3
+2024-04-26 03:35:44,832 - trainer - INFO -   Eval steps = 200 or (iterations = 200)
+2024-04-26 03:35:44,832 - trainer - INFO -   Best score (perplexity) = -4.7346601486206055
+2024-04-26 03:35:44,833 - trainer - INFO -   Gradient Accumulation steps = 1
+2024-04-26 03:35:44,833 - trainer - INFO -   Num of training examples (actually no. of iterations per epoch for Iterable Dataset)  = 130556
+2024-04-26 03:35:44,833 - trainer - INFO -   Num of development examples (actually no. of iterations per epoch for Iterable Dataset) = 14507
+2024-04-26 03:35:44,833 - trainer - INFO -   Time spent since last evaluation = 0h 6m 25s
+2024-04-26 03:35:44,833 - trainer - INFO -   Epoch = 1/5
+2024-04-26 03:35:44,833 - trainer - INFO -   Steps = 2400/40800
+2024-04-26 03:35:44,833 - trainer - INFO -   Instantaneous batch size per GPU = 4 and n_gpu = 4 so the input batch size = 16
+2024-04-26 03:35:44,833 - trainer - INFO -   dev_loss = 1.554910	||	 dev_eval_scores = {'perplexity': 4.7346601486206055}
+2024-04-26 03:35:44,833 - trainer - INFO -   train_loss = 2.974703311920166
+2024-04-26 03:35:44,833 - trainer - INFO - 
+********************************************
+2024-04-26 03:42:04,939 - trainer - INFO -    Save model to tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 03:42:10,876 - trainer - INFO -   Save check-point at epoch=0 step=2600
+2024-04-26 03:42:10,877 - trainer - INFO -    ***** Evaluation report *****
+2024-04-26 03:42:10,877 - trainer - INFO -   Output path (short): tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 03:42:10,877 - trainer - INFO -   Early stop on: perplexity
+2024-04-26 03:42:10,877 - trainer - INFO -   Early stop count = 0/3
+2024-04-26 03:42:10,877 - trainer - INFO -   Eval steps = 200 or (iterations = 200)
+2024-04-26 03:42:10,877 - trainer - INFO -   Best score (perplexity) = -4.624922275543213
+2024-04-26 03:42:10,877 - trainer - INFO -   Gradient Accumulation steps = 1
+2024-04-26 03:42:10,877 - trainer - INFO -   Num of training examples (actually no. of iterations per epoch for Iterable Dataset)  = 130556
+2024-04-26 03:42:10,877 - trainer - INFO -   Num of development examples (actually no. of iterations per epoch for Iterable Dataset) = 14507
+2024-04-26 03:42:10,877 - trainer - INFO -   Time spent since last evaluation = 0h 6m 26s
+2024-04-26 03:42:10,877 - trainer - INFO -   Epoch = 1/5
+2024-04-26 03:42:10,877 - trainer - INFO -   Steps = 2600/40800
+2024-04-26 03:42:10,877 - trainer - INFO -   Instantaneous batch size per GPU = 4 and n_gpu = 4 so the input batch size = 16
+2024-04-26 03:42:10,877 - trainer - INFO -   dev_loss = 1.531460	||	 dev_eval_scores = {'perplexity': 4.624922275543213}
+2024-04-26 03:42:10,878 - trainer - INFO -   train_loss = 2.8716752529144287
+2024-04-26 03:42:10,878 - trainer - INFO - 
+********************************************
+2024-04-26 03:48:30,754 - trainer - INFO -    Save model to tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 03:48:36,689 - trainer - INFO -   Save check-point at epoch=0 step=2800
+2024-04-26 03:48:36,690 - trainer - INFO -    ***** Evaluation report *****
+2024-04-26 03:48:36,690 - trainer - INFO -   Output path (short): tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 03:48:36,690 - trainer - INFO -   Early stop on: perplexity
+2024-04-26 03:48:36,690 - trainer - INFO -   Early stop count = 0/3
+2024-04-26 03:48:36,690 - trainer - INFO -   Eval steps = 200 or (iterations = 200)
+2024-04-26 03:48:36,690 - trainer - INFO -   Best score (perplexity) = -4.533045291900635
+2024-04-26 03:48:36,690 - trainer - INFO -   Gradient Accumulation steps = 1
+2024-04-26 03:48:36,690 - trainer - INFO -   Num of training examples (actually no. of iterations per epoch for Iterable Dataset)  = 130556
+2024-04-26 03:48:36,690 - trainer - INFO -   Num of development examples (actually no. of iterations per epoch for Iterable Dataset) = 14507
+2024-04-26 03:48:36,690 - trainer - INFO -   Time spent since last evaluation = 0h 6m 25s
+2024-04-26 03:48:36,690 - trainer - INFO -   Epoch = 1/5
+2024-04-26 03:48:36,690 - trainer - INFO -   Steps = 2800/40800
+2024-04-26 03:48:36,690 - trainer - INFO -   Instantaneous batch size per GPU = 4 and n_gpu = 4 so the input batch size = 16
+2024-04-26 03:48:36,691 - trainer - INFO -   dev_loss = 1.511394	||	 dev_eval_scores = {'perplexity': 4.533045291900635}
+2024-04-26 03:48:36,691 - trainer - INFO -   train_loss = 2.781400680541992
+2024-04-26 03:48:36,691 - trainer - INFO - 
+********************************************
+2024-04-26 03:54:56,573 - trainer - INFO -    Save model to tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 03:55:02,481 - trainer - INFO -   Save check-point at epoch=0 step=3000
+2024-04-26 03:55:02,482 - trainer - INFO -    ***** Evaluation report *****
+2024-04-26 03:55:02,482 - trainer - INFO -   Output path (short): tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 03:55:02,482 - trainer - INFO -   Early stop on: perplexity
+2024-04-26 03:55:02,482 - trainer - INFO -   Early stop count = 0/3
+2024-04-26 03:55:02,482 - trainer - INFO -   Eval steps = 200 or (iterations = 200)
+2024-04-26 03:55:02,482 - trainer - INFO -   Best score (perplexity) = -4.453883647918701
+2024-04-26 03:55:02,482 - trainer - INFO -   Gradient Accumulation steps = 1
+2024-04-26 03:55:02,482 - trainer - INFO -   Num of training examples (actually no. of iterations per epoch for Iterable Dataset)  = 130556
+2024-04-26 03:55:02,482 - trainer - INFO -   Num of development examples (actually no. of iterations per epoch for Iterable Dataset) = 14507
+2024-04-26 03:55:02,482 - trainer - INFO -   Time spent since last evaluation = 0h 6m 25s
+2024-04-26 03:55:02,482 - trainer - INFO -   Epoch = 1/5
+2024-04-26 03:55:02,482 - trainer - INFO -   Steps = 3000/40800
+2024-04-26 03:55:02,482 - trainer - INFO -   Instantaneous batch size per GPU = 4 and n_gpu = 4 so the input batch size = 16
+2024-04-26 03:55:02,482 - trainer - INFO -   dev_loss = 1.493776	||	 dev_eval_scores = {'perplexity': 4.453883647918701}
+2024-04-26 03:55:02,482 - trainer - INFO -   train_loss = 2.702195167541504
+2024-04-26 03:55:02,483 - trainer - INFO - 
+********************************************
+2024-04-26 04:01:21,916 - trainer - INFO -    Save model to tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 04:01:27,748 - trainer - INFO -   Save check-point at epoch=0 step=3200
+2024-04-26 04:01:27,748 - trainer - INFO -    ***** Evaluation report *****
+2024-04-26 04:01:27,749 - trainer - INFO -   Output path (short): tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 04:01:27,749 - trainer - INFO -   Early stop on: perplexity
+2024-04-26 04:01:27,749 - trainer - INFO -   Early stop count = 0/3
+2024-04-26 04:01:27,749 - trainer - INFO -   Eval steps = 200 or (iterations = 200)
+2024-04-26 04:01:27,749 - trainer - INFO -   Best score (perplexity) = -4.359768867492676
+2024-04-26 04:01:27,749 - trainer - INFO -   Gradient Accumulation steps = 1
+2024-04-26 04:01:27,749 - trainer - INFO -   Num of training examples (actually no. of iterations per epoch for Iterable Dataset)  = 130556
+2024-04-26 04:01:27,749 - trainer - INFO -   Num of development examples (actually no. of iterations per epoch for Iterable Dataset) = 14507
+2024-04-26 04:01:27,749 - trainer - INFO -   Time spent since last evaluation = 0h 6m 25s
+2024-04-26 04:01:27,749 - trainer - INFO -   Epoch = 1/5
+2024-04-26 04:01:27,749 - trainer - INFO -   Steps = 3200/40800
+2024-04-26 04:01:27,749 - trainer - INFO -   Instantaneous batch size per GPU = 4 and n_gpu = 4 so the input batch size = 16
+2024-04-26 04:01:27,749 - trainer - INFO -   dev_loss = 1.472419	||	 dev_eval_scores = {'perplexity': 4.359768867492676}
+2024-04-26 04:01:27,749 - trainer - INFO -   train_loss = 2.6316800117492676
+2024-04-26 04:01:27,749 - trainer - INFO - 
+********************************************
+2024-04-26 04:07:47,186 - trainer - INFO -    Save model to tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 04:07:52,996 - trainer - INFO -   Save check-point at epoch=0 step=3400
+2024-04-26 04:07:52,996 - trainer - INFO -    ***** Evaluation report *****
+2024-04-26 04:07:52,996 - trainer - INFO -   Output path (short): tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 04:07:52,996 - trainer - INFO -   Early stop on: perplexity
+2024-04-26 04:07:52,996 - trainer - INFO -   Early stop count = 0/3
+2024-04-26 04:07:52,996 - trainer - INFO -   Eval steps = 200 or (iterations = 200)
+2024-04-26 04:07:52,996 - trainer - INFO -   Best score (perplexity) = -4.2930779457092285
+2024-04-26 04:07:52,996 - trainer - INFO -   Gradient Accumulation steps = 1
+2024-04-26 04:07:52,996 - trainer - INFO -   Num of training examples (actually no. of iterations per epoch for Iterable Dataset)  = 130556
+2024-04-26 04:07:52,997 - trainer - INFO -   Num of development examples (actually no. of iterations per epoch for Iterable Dataset) = 14507
+2024-04-26 04:07:52,997 - trainer - INFO -   Time spent since last evaluation = 0h 6m 25s
+2024-04-26 04:07:52,997 - trainer - INFO -   Epoch = 1/5
+2024-04-26 04:07:52,997 - trainer - INFO -   Steps = 3400/40800
+2024-04-26 04:07:52,997 - trainer - INFO -   Instantaneous batch size per GPU = 4 and n_gpu = 4 so the input batch size = 16
+2024-04-26 04:07:52,997 - trainer - INFO -   dev_loss = 1.457004	||	 dev_eval_scores = {'perplexity': 4.2930779457092285}
+2024-04-26 04:07:52,997 - trainer - INFO -   train_loss = 2.5693111419677734
+2024-04-26 04:07:52,997 - trainer - INFO - 
+********************************************
+2024-04-26 04:14:12,633 - trainer - INFO -    Save model to tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 04:14:18,310 - trainer - INFO -   Save check-point at epoch=0 step=3600
+2024-04-26 04:14:18,310 - trainer - INFO -    ***** Evaluation report *****
+2024-04-26 04:14:18,310 - trainer - INFO -   Output path (short): tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 04:14:18,310 - trainer - INFO -   Early stop on: perplexity
+2024-04-26 04:14:18,310 - trainer - INFO -   Early stop count = 0/3
+2024-04-26 04:14:18,310 - trainer - INFO -   Eval steps = 200 or (iterations = 200)
+2024-04-26 04:14:18,310 - trainer - INFO -   Best score (perplexity) = -4.221639633178711
+2024-04-26 04:14:18,310 - trainer - INFO -   Gradient Accumulation steps = 1
+2024-04-26 04:14:18,310 - trainer - INFO -   Num of training examples (actually no. of iterations per epoch for Iterable Dataset)  = 130556
+2024-04-26 04:14:18,310 - trainer - INFO -   Num of development examples (actually no. of iterations per epoch for Iterable Dataset) = 14507
+2024-04-26 04:14:18,311 - trainer - INFO -   Time spent since last evaluation = 0h 6m 25s
+2024-04-26 04:14:18,311 - trainer - INFO -   Epoch = 1/5
+2024-04-26 04:14:18,311 - trainer - INFO -   Steps = 3600/40800
+2024-04-26 04:14:18,311 - trainer - INFO -   Instantaneous batch size per GPU = 4 and n_gpu = 4 so the input batch size = 16
+2024-04-26 04:14:18,311 - trainer - INFO -   dev_loss = 1.440224	||	 dev_eval_scores = {'perplexity': 4.221639633178711}
+2024-04-26 04:14:18,311 - trainer - INFO -   train_loss = 2.5129594802856445
+2024-04-26 04:14:18,311 - trainer - INFO - 
+********************************************
+2024-04-26 04:20:38,684 - trainer - INFO -    Save model to tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 04:20:44,511 - trainer - INFO -   Save check-point at epoch=0 step=3800
+2024-04-26 04:20:44,512 - trainer - INFO -    ***** Evaluation report *****
+2024-04-26 04:20:44,512 - trainer - INFO -   Output path (short): tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 04:20:44,512 - trainer - INFO -   Early stop on: perplexity
+2024-04-26 04:20:44,512 - trainer - INFO -   Early stop count = 0/3
+2024-04-26 04:20:44,512 - trainer - INFO -   Eval steps = 200 or (iterations = 200)
+2024-04-26 04:20:44,512 - trainer - INFO -   Best score (perplexity) = -4.147531986236572
+2024-04-26 04:20:44,512 - trainer - INFO -   Gradient Accumulation steps = 1
+2024-04-26 04:20:44,512 - trainer - INFO -   Num of training examples (actually no. of iterations per epoch for Iterable Dataset)  = 130556
+2024-04-26 04:20:44,512 - trainer - INFO -   Num of development examples (actually no. of iterations per epoch for Iterable Dataset) = 14507
+2024-04-26 04:20:44,512 - trainer - INFO -   Time spent since last evaluation = 0h 6m 26s
+2024-04-26 04:20:44,512 - trainer - INFO -   Epoch = 1/5
+2024-04-26 04:20:44,512 - trainer - INFO -   Steps = 3800/40800
+2024-04-26 04:20:44,512 - trainer - INFO -   Instantaneous batch size per GPU = 4 and n_gpu = 4 so the input batch size = 16
+2024-04-26 04:20:44,513 - trainer - INFO -   dev_loss = 1.422513	||	 dev_eval_scores = {'perplexity': 4.147531986236572}
+2024-04-26 04:20:44,513 - trainer - INFO -   train_loss = 2.460688829421997
+2024-04-26 04:20:44,513 - trainer - INFO - 
+********************************************
+2024-04-26 04:27:04,526 - trainer - INFO -    Save model to tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 04:27:10,385 - trainer - INFO -   Save check-point at epoch=0 step=4000
+2024-04-26 04:27:10,386 - trainer - INFO -    ***** Evaluation report *****
+2024-04-26 04:27:10,386 - trainer - INFO -   Output path (short): tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 04:27:10,386 - trainer - INFO -   Early stop on: perplexity
+2024-04-26 04:27:10,386 - trainer - INFO -   Early stop count = 0/3
+2024-04-26 04:27:10,386 - trainer - INFO -   Eval steps = 200 or (iterations = 200)
+2024-04-26 04:27:10,386 - trainer - INFO -   Best score (perplexity) = -4.087435722351074
+2024-04-26 04:27:10,386 - trainer - INFO -   Gradient Accumulation steps = 1
+2024-04-26 04:27:10,386 - trainer - INFO -   Num of training examples (actually no. of iterations per epoch for Iterable Dataset)  = 130556
+2024-04-26 04:27:10,386 - trainer - INFO -   Num of development examples (actually no. of iterations per epoch for Iterable Dataset) = 14507
+2024-04-26 04:27:10,386 - trainer - INFO -   Time spent since last evaluation = 0h 6m 25s
+2024-04-26 04:27:10,386 - trainer - INFO -   Epoch = 1/5
+2024-04-26 04:27:10,386 - trainer - INFO -   Steps = 4000/40800
+2024-04-26 04:27:10,386 - trainer - INFO -   Instantaneous batch size per GPU = 4 and n_gpu = 4 so the input batch size = 16
+2024-04-26 04:27:10,386 - trainer - INFO -   dev_loss = 1.407918	||	 dev_eval_scores = {'perplexity': 4.087435722351074}
+2024-04-26 04:27:10,387 - trainer - INFO -   train_loss = 2.4136698246002197
+2024-04-26 04:27:10,387 - trainer - INFO - 
+********************************************
+2024-04-26 04:33:30,165 - trainer - INFO -    Save model to tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 04:33:36,001 - trainer - INFO -   Save check-point at epoch=0 step=4200
+2024-04-26 04:33:36,001 - trainer - INFO -    ***** Evaluation report *****
+2024-04-26 04:33:36,001 - trainer - INFO -   Output path (short): tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 04:33:36,001 - trainer - INFO -   Early stop on: perplexity
+2024-04-26 04:33:36,001 - trainer - INFO -   Early stop count = 0/3
+2024-04-26 04:33:36,001 - trainer - INFO -   Eval steps = 200 or (iterations = 200)
+2024-04-26 04:33:36,001 - trainer - INFO -   Best score (perplexity) = -4.028451442718506
+2024-04-26 04:33:36,002 - trainer - INFO -   Gradient Accumulation steps = 1
+2024-04-26 04:33:36,002 - trainer - INFO -   Num of training examples (actually no. of iterations per epoch for Iterable Dataset)  = 130556
+2024-04-26 04:33:36,002 - trainer - INFO -   Num of development examples (actually no. of iterations per epoch for Iterable Dataset) = 14507
+2024-04-26 04:33:36,002 - trainer - INFO -   Time spent since last evaluation = 0h 6m 25s
+2024-04-26 04:33:36,002 - trainer - INFO -   Epoch = 1/5
+2024-04-26 04:33:36,002 - trainer - INFO -   Steps = 4200/40800
+2024-04-26 04:33:36,002 - trainer - INFO -   Instantaneous batch size per GPU = 4 and n_gpu = 4 so the input batch size = 16
+2024-04-26 04:33:36,002 - trainer - INFO -   dev_loss = 1.393382	||	 dev_eval_scores = {'perplexity': 4.028451442718506}
+2024-04-26 04:33:36,002 - trainer - INFO -   train_loss = 2.3706307411193848
+2024-04-26 04:33:36,002 - trainer - INFO - 
+********************************************
+2024-04-26 04:39:55,706 - trainer - INFO -    Save model to tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 04:40:01,545 - trainer - INFO -   Save check-point at epoch=0 step=4400
+2024-04-26 04:40:01,545 - trainer - INFO -    ***** Evaluation report *****
+2024-04-26 04:40:01,545 - trainer - INFO -   Output path (short): tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 04:40:01,545 - trainer - INFO -   Early stop on: perplexity
+2024-04-26 04:40:01,545 - trainer - INFO -   Early stop count = 0/3
+2024-04-26 04:40:01,545 - trainer - INFO -   Eval steps = 200 or (iterations = 200)
+2024-04-26 04:40:01,545 - trainer - INFO -   Best score (perplexity) = -3.976846694946289
+2024-04-26 04:40:01,546 - trainer - INFO -   Gradient Accumulation steps = 1
+2024-04-26 04:40:01,546 - trainer - INFO -   Num of training examples (actually no. of iterations per epoch for Iterable Dataset)  = 130556
+2024-04-26 04:40:01,546 - trainer - INFO -   Num of development examples (actually no. of iterations per epoch for Iterable Dataset) = 14507
+2024-04-26 04:40:01,546 - trainer - INFO -   Time spent since last evaluation = 0h 6m 25s
+2024-04-26 04:40:01,546 - trainer - INFO -   Epoch = 1/5
+2024-04-26 04:40:01,546 - trainer - INFO -   Steps = 4400/40800
+2024-04-26 04:40:01,546 - trainer - INFO -   Instantaneous batch size per GPU = 4 and n_gpu = 4 so the input batch size = 16
+2024-04-26 04:40:01,546 - trainer - INFO -   dev_loss = 1.380489	||	 dev_eval_scores = {'perplexity': 3.976846694946289}
+2024-04-26 04:40:01,546 - trainer - INFO -   train_loss = 2.330047369003296
+2024-04-26 04:40:01,546 - trainer - INFO - 
+********************************************
+2024-04-26 04:46:21,905 - trainer - INFO -    Save model to tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 04:46:27,763 - trainer - INFO -   Save check-point at epoch=0 step=4600
+2024-04-26 04:46:27,764 - trainer - INFO -    ***** Evaluation report *****
+2024-04-26 04:46:27,764 - trainer - INFO -   Output path (short): tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 04:46:27,764 - trainer - INFO -   Early stop on: perplexity
+2024-04-26 04:46:27,764 - trainer - INFO -   Early stop count = 0/3
+2024-04-26 04:46:27,764 - trainer - INFO -   Eval steps = 200 or (iterations = 200)
+2024-04-26 04:46:27,764 - trainer - INFO -   Best score (perplexity) = -3.920635461807251
+2024-04-26 04:46:27,764 - trainer - INFO -   Gradient Accumulation steps = 1
+2024-04-26 04:46:27,764 - trainer - INFO -   Num of training examples (actually no. of iterations per epoch for Iterable Dataset)  = 130556
+2024-04-26 04:46:27,764 - trainer - INFO -   Num of development examples (actually no. of iterations per epoch for Iterable Dataset) = 14507
+2024-04-26 04:46:27,764 - trainer - INFO -   Time spent since last evaluation = 0h 6m 26s
+2024-04-26 04:46:27,764 - trainer - INFO -   Epoch = 1/5
+2024-04-26 04:46:27,764 - trainer - INFO -   Steps = 4600/40800
+2024-04-26 04:46:27,764 - trainer - INFO -   Instantaneous batch size per GPU = 4 and n_gpu = 4 so the input batch size = 16
+2024-04-26 04:46:27,764 - trainer - INFO -   dev_loss = 1.366254	||	 dev_eval_scores = {'perplexity': 3.920635461807251}
+2024-04-26 04:46:27,765 - trainer - INFO -   train_loss = 2.2929983139038086
+2024-04-26 04:46:27,765 - trainer - INFO - 
+********************************************
+2024-04-26 04:52:47,547 - trainer - INFO -    Save model to tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 04:52:53,122 - trainer - INFO -   Save check-point at epoch=0 step=4800
+2024-04-26 04:52:53,122 - trainer - INFO -    ***** Evaluation report *****
+2024-04-26 04:52:53,122 - trainer - INFO -   Output path (short): tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 04:52:53,122 - trainer - INFO -   Early stop on: perplexity
+2024-04-26 04:52:53,122 - trainer - INFO -   Early stop count = 0/3
+2024-04-26 04:52:53,122 - trainer - INFO -   Eval steps = 200 or (iterations = 200)
+2024-04-26 04:52:53,122 - trainer - INFO -   Best score (perplexity) = -3.866814613342285
+2024-04-26 04:52:53,122 - trainer - INFO -   Gradient Accumulation steps = 1
+2024-04-26 04:52:53,122 - trainer - INFO -   Num of training examples (actually no. of iterations per epoch for Iterable Dataset)  = 130556
+2024-04-26 04:52:53,122 - trainer - INFO -   Num of development examples (actually no. of iterations per epoch for Iterable Dataset) = 14507
+2024-04-26 04:52:53,123 - trainer - INFO -   Time spent since last evaluation = 0h 6m 25s
+2024-04-26 04:52:53,123 - trainer - INFO -   Epoch = 1/5
+2024-04-26 04:52:53,123 - trainer - INFO -   Steps = 4800/40800
+2024-04-26 04:52:53,123 - trainer - INFO -   Instantaneous batch size per GPU = 4 and n_gpu = 4 so the input batch size = 16
+2024-04-26 04:52:53,123 - trainer - INFO -   dev_loss = 1.352431	||	 dev_eval_scores = {'perplexity': 3.866814613342285}
+2024-04-26 04:52:53,123 - trainer - INFO -   train_loss = 2.2579383850097656
+2024-04-26 04:52:53,123 - trainer - INFO - 
+********************************************
+2024-04-26 04:59:12,707 - trainer - INFO -    Save model to tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 04:59:18,585 - trainer - INFO -   Save check-point at epoch=0 step=5000
+2024-04-26 04:59:18,586 - trainer - INFO -    ***** Evaluation report *****
+2024-04-26 04:59:18,586 - trainer - INFO -   Output path (short): tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 04:59:18,586 - trainer - INFO -   Early stop on: perplexity
+2024-04-26 04:59:18,586 - trainer - INFO -   Early stop count = 0/3
+2024-04-26 04:59:18,586 - trainer - INFO -   Eval steps = 200 or (iterations = 200)
+2024-04-26 04:59:18,586 - trainer - INFO -   Best score (perplexity) = -3.827284574508667
+2024-04-26 04:59:18,586 - trainer - INFO -   Gradient Accumulation steps = 1
+2024-04-26 04:59:18,586 - trainer - INFO -   Num of training examples (actually no. of iterations per epoch for Iterable Dataset)  = 130556
+2024-04-26 04:59:18,586 - trainer - INFO -   Num of development examples (actually no. of iterations per epoch for Iterable Dataset) = 14507
+2024-04-26 04:59:18,586 - trainer - INFO -   Time spent since last evaluation = 0h 6m 25s
+2024-04-26 04:59:18,586 - trainer - INFO -   Epoch = 1/5
+2024-04-26 04:59:18,586 - trainer - INFO -   Steps = 5000/40800
+2024-04-26 04:59:18,586 - trainer - INFO -   Instantaneous batch size per GPU = 4 and n_gpu = 4 so the input batch size = 16
+2024-04-26 04:59:18,586 - trainer - INFO -   dev_loss = 1.342156	||	 dev_eval_scores = {'perplexity': 3.827284574508667}
+2024-04-26 04:59:18,587 - trainer - INFO -   train_loss = 2.225395679473877
+2024-04-26 04:59:18,587 - trainer - INFO - 
+********************************************
+2024-04-26 05:05:39,819 - trainer - INFO -    Save model to tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 05:05:45,329 - trainer - INFO -   Save check-point at epoch=0 step=5200
+2024-04-26 05:05:45,330 - trainer - INFO -    ***** Evaluation report *****
+2024-04-26 05:05:45,330 - trainer - INFO -   Output path (short): tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 05:05:45,330 - trainer - INFO -   Early stop on: perplexity
+2024-04-26 05:05:45,330 - trainer - INFO -   Early stop count = 0/3
+2024-04-26 05:05:45,330 - trainer - INFO -   Eval steps = 200 or (iterations = 200)
+2024-04-26 05:05:45,330 - trainer - INFO -   Best score (perplexity) = -3.7697432041168213
+2024-04-26 05:05:45,330 - trainer - INFO -   Gradient Accumulation steps = 1
+2024-04-26 05:05:45,330 - trainer - INFO -   Num of training examples (actually no. of iterations per epoch for Iterable Dataset)  = 130556
+2024-04-26 05:05:45,330 - trainer - INFO -   Num of development examples (actually no. of iterations per epoch for Iterable Dataset) = 14507
+2024-04-26 05:05:45,330 - trainer - INFO -   Time spent since last evaluation = 0h 6m 26s
+2024-04-26 05:05:45,330 - trainer - INFO -   Epoch = 1/5
+2024-04-26 05:05:45,330 - trainer - INFO -   Steps = 5200/40800
+2024-04-26 05:05:45,330 - trainer - INFO -   Instantaneous batch size per GPU = 4 and n_gpu = 4 so the input batch size = 16
+2024-04-26 05:05:45,330 - trainer - INFO -   dev_loss = 1.327007	||	 dev_eval_scores = {'perplexity': 3.7697432041168213}
+2024-04-26 05:05:45,331 - trainer - INFO -   train_loss = 2.194683790206909
+2024-04-26 05:05:45,331 - trainer - INFO - 
+********************************************
+2024-04-26 05:12:05,019 - trainer - INFO -    Save model to tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 05:12:10,879 - trainer - INFO -   Save check-point at epoch=0 step=5400
+2024-04-26 05:12:10,879 - trainer - INFO -    ***** Evaluation report *****
+2024-04-26 05:12:10,879 - trainer - INFO -   Output path (short): tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 05:12:10,879 - trainer - INFO -   Early stop on: perplexity
+2024-04-26 05:12:10,879 - trainer - INFO -   Early stop count = 0/3
+2024-04-26 05:12:10,880 - trainer - INFO -   Eval steps = 200 or (iterations = 200)
+2024-04-26 05:12:10,880 - trainer - INFO -   Best score (perplexity) = -3.732077121734619
+2024-04-26 05:12:10,880 - trainer - INFO -   Gradient Accumulation steps = 1
+2024-04-26 05:12:10,880 - trainer - INFO -   Num of training examples (actually no. of iterations per epoch for Iterable Dataset)  = 130556
+2024-04-26 05:12:10,880 - trainer - INFO -   Num of development examples (actually no. of iterations per epoch for Iterable Dataset) = 14507
+2024-04-26 05:12:10,880 - trainer - INFO -   Time spent since last evaluation = 0h 6m 25s
+2024-04-26 05:12:10,880 - trainer - INFO -   Epoch = 1/5
+2024-04-26 05:12:10,880 - trainer - INFO -   Steps = 5400/40800
+2024-04-26 05:12:10,880 - trainer - INFO -   Instantaneous batch size per GPU = 4 and n_gpu = 4 so the input batch size = 16
+2024-04-26 05:12:10,880 - trainer - INFO -   dev_loss = 1.316965	||	 dev_eval_scores = {'perplexity': 3.732077121734619}
+2024-04-26 05:12:10,880 - trainer - INFO -   train_loss = 2.16521954536438
+2024-04-26 05:12:10,880 - trainer - INFO - 
+********************************************
+2024-04-26 05:18:31,741 - trainer - INFO -    Save model to tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 05:18:37,604 - trainer - INFO -   Save check-point at epoch=0 step=5600
+2024-04-26 05:18:37,605 - trainer - INFO -    ***** Evaluation report *****
+2024-04-26 05:18:37,605 - trainer - INFO -   Output path (short): tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 05:18:37,605 - trainer - INFO -   Early stop on: perplexity
+2024-04-26 05:18:37,605 - trainer - INFO -   Early stop count = 0/3
+2024-04-26 05:18:37,605 - trainer - INFO -   Eval steps = 200 or (iterations = 200)
+2024-04-26 05:18:37,605 - trainer - INFO -   Best score (perplexity) = -3.6822173595428467
+2024-04-26 05:18:37,605 - trainer - INFO -   Gradient Accumulation steps = 1
+2024-04-26 05:18:37,605 - trainer - INFO -   Num of training examples (actually no. of iterations per epoch for Iterable Dataset)  = 130556
+2024-04-26 05:18:37,605 - trainer - INFO -   Num of development examples (actually no. of iterations per epoch for Iterable Dataset) = 14507
+2024-04-26 05:18:37,605 - trainer - INFO -   Time spent since last evaluation = 0h 6m 26s
+2024-04-26 05:18:37,605 - trainer - INFO -   Epoch = 1/5
+2024-04-26 05:18:37,605 - trainer - INFO -   Steps = 5600/40800
+2024-04-26 05:18:37,605 - trainer - INFO -   Instantaneous batch size per GPU = 4 and n_gpu = 4 so the input batch size = 16
+2024-04-26 05:18:37,605 - trainer - INFO -   dev_loss = 1.303515	||	 dev_eval_scores = {'perplexity': 3.6822173595428467}
+2024-04-26 05:18:37,606 - trainer - INFO -   train_loss = 2.1381325721740723
+2024-04-26 05:18:37,606 - trainer - INFO - 
+********************************************
+2024-04-26 05:24:57,533 - trainer - INFO -    Save model to tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 05:25:03,410 - trainer - INFO -   Save check-point at epoch=0 step=5800
+2024-04-26 05:25:03,411 - trainer - INFO -    ***** Evaluation report *****
+2024-04-26 05:25:03,411 - trainer - INFO -   Output path (short): tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 05:25:03,411 - trainer - INFO -   Early stop on: perplexity
+2024-04-26 05:25:03,411 - trainer - INFO -   Early stop count = 0/3
+2024-04-26 05:25:03,411 - trainer - INFO -   Eval steps = 200 or (iterations = 200)
+2024-04-26 05:25:03,411 - trainer - INFO -   Best score (perplexity) = -3.641592264175415
+2024-04-26 05:25:03,411 - trainer - INFO -   Gradient Accumulation steps = 1
+2024-04-26 05:25:03,411 - trainer - INFO -   Num of training examples (actually no. of iterations per epoch for Iterable Dataset)  = 130556
+2024-04-26 05:25:03,411 - trainer - INFO -   Num of development examples (actually no. of iterations per epoch for Iterable Dataset) = 14507
+2024-04-26 05:25:03,411 - trainer - INFO -   Time spent since last evaluation = 0h 6m 25s
+2024-04-26 05:25:03,411 - trainer - INFO -   Epoch = 1/5
+2024-04-26 05:25:03,411 - trainer - INFO -   Steps = 5800/40800
+2024-04-26 05:25:03,411 - trainer - INFO -   Instantaneous batch size per GPU = 4 and n_gpu = 4 so the input batch size = 16
+2024-04-26 05:25:03,411 - trainer - INFO -   dev_loss = 1.292421	||	 dev_eval_scores = {'perplexity': 3.641592264175415}
+2024-04-26 05:25:03,412 - trainer - INFO -   train_loss = 2.113192319869995
+2024-04-26 05:25:03,412 - trainer - INFO - 
+********************************************
+2024-04-26 05:31:23,054 - trainer - INFO -    Save model to tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 05:31:28,936 - trainer - INFO -   Save check-point at epoch=0 step=6000
+2024-04-26 05:31:28,937 - trainer - INFO -    ***** Evaluation report *****
+2024-04-26 05:31:28,937 - trainer - INFO -   Output path (short): tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 05:31:28,937 - trainer - INFO -   Early stop on: perplexity
+2024-04-26 05:31:28,937 - trainer - INFO -   Early stop count = 0/3
+2024-04-26 05:31:28,937 - trainer - INFO -   Eval steps = 200 or (iterations = 200)
+2024-04-26 05:31:28,937 - trainer - INFO -   Best score (perplexity) = -3.602872133255005
+2024-04-26 05:31:28,937 - trainer - INFO -   Gradient Accumulation steps = 1
+2024-04-26 05:31:28,937 - trainer - INFO -   Num of training examples (actually no. of iterations per epoch for Iterable Dataset)  = 130556
+2024-04-26 05:31:28,937 - trainer - INFO -   Num of development examples (actually no. of iterations per epoch for Iterable Dataset) = 14507
+2024-04-26 05:31:28,937 - trainer - INFO -   Time spent since last evaluation = 0h 6m 25s
+2024-04-26 05:31:28,937 - trainer - INFO -   Epoch = 1/5
+2024-04-26 05:31:28,937 - trainer - INFO -   Steps = 6000/40800
+2024-04-26 05:31:28,937 - trainer - INFO -   Instantaneous batch size per GPU = 4 and n_gpu = 4 so the input batch size = 16
+2024-04-26 05:31:28,938 - trainer - INFO -   dev_loss = 1.281731	||	 dev_eval_scores = {'perplexity': 3.602872133255005}
+2024-04-26 05:31:28,938 - trainer - INFO -   train_loss = 2.0891873836517334
+2024-04-26 05:31:28,938 - trainer - INFO - 
+********************************************
+2024-04-26 05:37:49,590 - trainer - INFO -    Save model to tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 05:37:55,449 - trainer - INFO -   Save check-point at epoch=0 step=6200
+2024-04-26 05:37:55,449 - trainer - INFO -    ***** Evaluation report *****
+2024-04-26 05:37:55,450 - trainer - INFO -   Output path (short): tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 05:37:55,450 - trainer - INFO -   Early stop on: perplexity
+2024-04-26 05:37:55,450 - trainer - INFO -   Early stop count = 0/3
+2024-04-26 05:37:55,450 - trainer - INFO -   Eval steps = 200 or (iterations = 200)
+2024-04-26 05:37:55,450 - trainer - INFO -   Best score (perplexity) = -3.5650696754455566
+2024-04-26 05:37:55,450 - trainer - INFO -   Gradient Accumulation steps = 1
+2024-04-26 05:37:55,450 - trainer - INFO -   Num of training examples (actually no. of iterations per epoch for Iterable Dataset)  = 130556
+2024-04-26 05:37:55,450 - trainer - INFO -   Num of development examples (actually no. of iterations per epoch for Iterable Dataset) = 14507
+2024-04-26 05:37:55,450 - trainer - INFO -   Time spent since last evaluation = 0h 6m 26s
+2024-04-26 05:37:55,450 - trainer - INFO -   Epoch = 1/5
+2024-04-26 05:37:55,450 - trainer - INFO -   Steps = 6200/40800
+2024-04-26 05:37:55,450 - trainer - INFO -   Instantaneous batch size per GPU = 4 and n_gpu = 4 so the input batch size = 16
+2024-04-26 05:37:55,450 - trainer - INFO -   dev_loss = 1.271184	||	 dev_eval_scores = {'perplexity': 3.5650696754455566}
+2024-04-26 05:37:55,450 - trainer - INFO -   train_loss = 2.066126585006714
+2024-04-26 05:37:55,451 - trainer - INFO - 
+********************************************
+2024-04-26 05:44:15,283 - trainer - INFO -    Save model to tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 05:44:21,132 - trainer - INFO -   Save check-point at epoch=0 step=6400
+2024-04-26 05:44:21,133 - trainer - INFO -    ***** Evaluation report *****
+2024-04-26 05:44:21,133 - trainer - INFO -   Output path (short): tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 05:44:21,133 - trainer - INFO -   Early stop on: perplexity
+2024-04-26 05:44:21,133 - trainer - INFO -   Early stop count = 0/3
+2024-04-26 05:44:21,133 - trainer - INFO -   Eval steps = 200 or (iterations = 200)
+2024-04-26 05:44:21,133 - trainer - INFO -   Best score (perplexity) = -3.517021894454956
+2024-04-26 05:44:21,133 - trainer - INFO -   Gradient Accumulation steps = 1
+2024-04-26 05:44:21,133 - trainer - INFO -   Num of training examples (actually no. of iterations per epoch for Iterable Dataset)  = 130556
+2024-04-26 05:44:21,133 - trainer - INFO -   Num of development examples (actually no. of iterations per epoch for Iterable Dataset) = 14507
+2024-04-26 05:44:21,133 - trainer - INFO -   Time spent since last evaluation = 0h 6m 25s
+2024-04-26 05:44:21,133 - trainer - INFO -   Epoch = 1/5
+2024-04-26 05:44:21,133 - trainer - INFO -   Steps = 6400/40800
+2024-04-26 05:44:21,133 - trainer - INFO -   Instantaneous batch size per GPU = 4 and n_gpu = 4 so the input batch size = 16
+2024-04-26 05:44:21,133 - trainer - INFO -   dev_loss = 1.257615	||	 dev_eval_scores = {'perplexity': 3.517021894454956}
+2024-04-26 05:44:21,134 - trainer - INFO -   train_loss = 2.0438156127929688
+2024-04-26 05:44:21,134 - trainer - INFO - 
+********************************************
+2024-04-26 05:50:41,206 - trainer - INFO -    Save model to tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 05:50:47,135 - trainer - INFO -   Save check-point at epoch=0 step=6600
+2024-04-26 05:50:47,135 - trainer - INFO -    ***** Evaluation report *****
+2024-04-26 05:50:47,135 - trainer - INFO -   Output path (short): tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 05:50:47,135 - trainer - INFO -   Early stop on: perplexity
+2024-04-26 05:50:47,135 - trainer - INFO -   Early stop count = 0/3
+2024-04-26 05:50:47,135 - trainer - INFO -   Eval steps = 200 or (iterations = 200)
+2024-04-26 05:50:47,135 - trainer - INFO -   Best score (perplexity) = -3.4847798347473145
+2024-04-26 05:50:47,135 - trainer - INFO -   Gradient Accumulation steps = 1
+2024-04-26 05:50:47,135 - trainer - INFO -   Num of training examples (actually no. of iterations per epoch for Iterable Dataset)  = 130556
+2024-04-26 05:50:47,136 - trainer - INFO -   Num of development examples (actually no. of iterations per epoch for Iterable Dataset) = 14507
+2024-04-26 05:50:47,136 - trainer - INFO -   Time spent since last evaluation = 0h 6m 26s
+2024-04-26 05:50:47,136 - trainer - INFO -   Epoch = 1/5
+2024-04-26 05:50:47,136 - trainer - INFO -   Steps = 6600/40800
+2024-04-26 05:50:47,136 - trainer - INFO -   Instantaneous batch size per GPU = 4 and n_gpu = 4 so the input batch size = 16
+2024-04-26 05:50:47,136 - trainer - INFO -   dev_loss = 1.248405	||	 dev_eval_scores = {'perplexity': 3.4847798347473145}
+2024-04-26 05:50:47,136 - trainer - INFO -   train_loss = 2.022505283355713
+2024-04-26 05:50:47,136 - trainer - INFO - 
+********************************************
+2024-04-26 05:57:06,888 - trainer - INFO -    Save model to tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 05:57:12,755 - trainer - INFO -   Save check-point at epoch=0 step=6800
+2024-04-26 05:57:12,755 - trainer - INFO -    ***** Evaluation report *****
+2024-04-26 05:57:12,756 - trainer - INFO -   Output path (short): tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 05:57:12,756 - trainer - INFO -   Early stop on: perplexity
+2024-04-26 05:57:12,756 - trainer - INFO -   Early stop count = 0/3
+2024-04-26 05:57:12,756 - trainer - INFO -   Eval steps = 200 or (iterations = 200)
+2024-04-26 05:57:12,756 - trainer - INFO -   Best score (perplexity) = -3.441448450088501
+2024-04-26 05:57:12,756 - trainer - INFO -   Gradient Accumulation steps = 1
+2024-04-26 05:57:12,756 - trainer - INFO -   Num of training examples (actually no. of iterations per epoch for Iterable Dataset)  = 130556
+2024-04-26 05:57:12,756 - trainer - INFO -   Num of development examples (actually no. of iterations per epoch for Iterable Dataset) = 14507
+2024-04-26 05:57:12,756 - trainer - INFO -   Time spent since last evaluation = 0h 6m 25s
+2024-04-26 05:57:12,756 - trainer - INFO -   Epoch = 1/5
+2024-04-26 05:57:12,756 - trainer - INFO -   Steps = 6800/40800
+2024-04-26 05:57:12,756 - trainer - INFO -   Instantaneous batch size per GPU = 4 and n_gpu = 4 so the input batch size = 16
+2024-04-26 05:57:12,756 - trainer - INFO -   dev_loss = 1.235892	||	 dev_eval_scores = {'perplexity': 3.441448450088501}
+2024-04-26 05:57:12,756 - trainer - INFO -   train_loss = 2.0026967525482178
+2024-04-26 05:57:12,757 - trainer - INFO - 
+********************************************
+2024-04-26 06:03:32,820 - trainer - INFO -    Save model to tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 06:03:38,713 - trainer - INFO -   Save check-point at epoch=0 step=7000
+2024-04-26 06:03:38,714 - trainer - INFO -    ***** Evaluation report *****
+2024-04-26 06:03:38,714 - trainer - INFO -   Output path (short): tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 06:03:38,714 - trainer - INFO -   Early stop on: perplexity
+2024-04-26 06:03:38,714 - trainer - INFO -   Early stop count = 0/3
+2024-04-26 06:03:38,714 - trainer - INFO -   Eval steps = 200 or (iterations = 200)
+2024-04-26 06:03:38,714 - trainer - INFO -   Best score (perplexity) = -3.3976998329162598
+2024-04-26 06:03:38,714 - trainer - INFO -   Gradient Accumulation steps = 1
+2024-04-26 06:03:38,714 - trainer - INFO -   Num of training examples (actually no. of iterations per epoch for Iterable Dataset)  = 130556
+2024-04-26 06:03:38,714 - trainer - INFO -   Num of development examples (actually no. of iterations per epoch for Iterable Dataset) = 14507
+2024-04-26 06:03:38,714 - trainer - INFO -   Time spent since last evaluation = 0h 6m 25s
+2024-04-26 06:03:38,714 - trainer - INFO -   Epoch = 1/5
+2024-04-26 06:03:38,714 - trainer - INFO -   Steps = 7000/40800
+2024-04-26 06:03:38,714 - trainer - INFO -   Instantaneous batch size per GPU = 4 and n_gpu = 4 so the input batch size = 16
+2024-04-26 06:03:38,715 - trainer - INFO -   dev_loss = 1.223099	||	 dev_eval_scores = {'perplexity': 3.3976998329162598}
+2024-04-26 06:03:38,715 - trainer - INFO -   train_loss = 1.983184576034546
+2024-04-26 06:03:38,715 - trainer - INFO - 
+********************************************
+2024-04-26 06:09:59,334 - trainer - INFO -    Save model to tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 06:10:05,217 - trainer - INFO -   Save check-point at epoch=0 step=7200
+2024-04-26 06:10:05,217 - trainer - INFO -    ***** Evaluation report *****
+2024-04-26 06:10:05,217 - trainer - INFO -   Output path (short): tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 06:10:05,217 - trainer - INFO -   Early stop on: perplexity
+2024-04-26 06:10:05,217 - trainer - INFO -   Early stop count = 0/3
+2024-04-26 06:10:05,217 - trainer - INFO -   Eval steps = 200 or (iterations = 200)
+2024-04-26 06:10:05,217 - trainer - INFO -   Best score (perplexity) = -3.3713600635528564
+2024-04-26 06:10:05,217 - trainer - INFO -   Gradient Accumulation steps = 1
+2024-04-26 06:10:05,217 - trainer - INFO -   Num of training examples (actually no. of iterations per epoch for Iterable Dataset)  = 130556
+2024-04-26 06:10:05,218 - trainer - INFO -   Num of development examples (actually no. of iterations per epoch for Iterable Dataset) = 14507
+2024-04-26 06:10:05,218 - trainer - INFO -   Time spent since last evaluation = 0h 6m 26s
+2024-04-26 06:10:05,218 - trainer - INFO -   Epoch = 1/5
+2024-04-26 06:10:05,218 - trainer - INFO -   Steps = 7200/40800
+2024-04-26 06:10:05,218 - trainer - INFO -   Instantaneous batch size per GPU = 4 and n_gpu = 4 so the input batch size = 16
+2024-04-26 06:10:05,218 - trainer - INFO -   dev_loss = 1.215316	||	 dev_eval_scores = {'perplexity': 3.3713600635528564}
+2024-04-26 06:10:05,218 - trainer - INFO -   train_loss = 1.9642337560653687
+2024-04-26 06:10:05,218 - trainer - INFO - 
+********************************************
+2024-04-26 06:16:24,736 - trainer - INFO -    Save model to tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 06:16:30,613 - trainer - INFO -   Save check-point at epoch=0 step=7400
+2024-04-26 06:16:30,613 - trainer - INFO -    ***** Evaluation report *****
+2024-04-26 06:16:30,613 - trainer - INFO -   Output path (short): tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 06:16:30,613 - trainer - INFO -   Early stop on: perplexity
+2024-04-26 06:16:30,613 - trainer - INFO -   Early stop count = 0/3
+2024-04-26 06:16:30,613 - trainer - INFO -   Eval steps = 200 or (iterations = 200)
+2024-04-26 06:16:30,613 - trainer - INFO -   Best score (perplexity) = -3.334381341934204
+2024-04-26 06:16:30,613 - trainer - INFO -   Gradient Accumulation steps = 1
+2024-04-26 06:16:30,613 - trainer - INFO -   Num of training examples (actually no. of iterations per epoch for Iterable Dataset)  = 130556
+2024-04-26 06:16:30,613 - trainer - INFO -   Num of development examples (actually no. of iterations per epoch for Iterable Dataset) = 14507
+2024-04-26 06:16:30,613 - trainer - INFO -   Time spent since last evaluation = 0h 6m 25s
+2024-04-26 06:16:30,613 - trainer - INFO -   Epoch = 1/5
+2024-04-26 06:16:30,614 - trainer - INFO -   Steps = 7400/40800
+2024-04-26 06:16:30,614 - trainer - INFO -   Instantaneous batch size per GPU = 4 and n_gpu = 4 so the input batch size = 16
+2024-04-26 06:16:30,614 - trainer - INFO -   dev_loss = 1.204287	||	 dev_eval_scores = {'perplexity': 3.334381341934204}
+2024-04-26 06:16:30,614 - trainer - INFO -   train_loss = 1.9464004039764404
+2024-04-26 06:16:30,614 - trainer - INFO - 
+********************************************
+2024-04-26 06:22:50,090 - trainer - INFO -    Save model to tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 06:22:55,903 - trainer - INFO -   Save check-point at epoch=0 step=7600
+2024-04-26 06:22:55,903 - trainer - INFO -    ***** Evaluation report *****
+2024-04-26 06:22:55,903 - trainer - INFO -   Output path (short): tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 06:22:55,903 - trainer - INFO -   Early stop on: perplexity
+2024-04-26 06:22:55,903 - trainer - INFO -   Early stop count = 0/3
+2024-04-26 06:22:55,903 - trainer - INFO -   Eval steps = 200 or (iterations = 200)
+2024-04-26 06:22:55,903 - trainer - INFO -   Best score (perplexity) = -3.299593448638916
+2024-04-26 06:22:55,903 - trainer - INFO -   Gradient Accumulation steps = 1
+2024-04-26 06:22:55,903 - trainer - INFO -   Num of training examples (actually no. of iterations per epoch for Iterable Dataset)  = 130556
+2024-04-26 06:22:55,903 - trainer - INFO -   Num of development examples (actually no. of iterations per epoch for Iterable Dataset) = 14507
+2024-04-26 06:22:55,903 - trainer - INFO -   Time spent since last evaluation = 0h 6m 25s
+2024-04-26 06:22:55,903 - trainer - INFO -   Epoch = 1/5
+2024-04-26 06:22:55,904 - trainer - INFO -   Steps = 7600/40800
+2024-04-26 06:22:55,904 - trainer - INFO -   Instantaneous batch size per GPU = 4 and n_gpu = 4 so the input batch size = 16
+2024-04-26 06:22:55,904 - trainer - INFO -   dev_loss = 1.193799	||	 dev_eval_scores = {'perplexity': 3.299593448638916}
+2024-04-26 06:22:55,904 - trainer - INFO -   train_loss = 1.9291884899139404
+2024-04-26 06:22:55,904 - trainer - INFO - 
+********************************************
+2024-04-26 06:29:16,451 - trainer - INFO -    Save model to tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 06:29:22,361 - trainer - INFO -   Save check-point at epoch=0 step=7800
+2024-04-26 06:29:22,361 - trainer - INFO -    ***** Evaluation report *****
+2024-04-26 06:29:22,361 - trainer - INFO -   Output path (short): tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 06:29:22,361 - trainer - INFO -   Early stop on: perplexity
+2024-04-26 06:29:22,361 - trainer - INFO -   Early stop count = 0/3
+2024-04-26 06:29:22,362 - trainer - INFO -   Eval steps = 200 or (iterations = 200)
+2024-04-26 06:29:22,362 - trainer - INFO -   Best score (perplexity) = -3.2615699768066406
+2024-04-26 06:29:22,362 - trainer - INFO -   Gradient Accumulation steps = 1
+2024-04-26 06:29:22,362 - trainer - INFO -   Num of training examples (actually no. of iterations per epoch for Iterable Dataset)  = 130556
+2024-04-26 06:29:22,362 - trainer - INFO -   Num of development examples (actually no. of iterations per epoch for Iterable Dataset) = 14507
+2024-04-26 06:29:22,362 - trainer - INFO -   Time spent since last evaluation = 0h 6m 26s
+2024-04-26 06:29:22,362 - trainer - INFO -   Epoch = 1/5
+2024-04-26 06:29:22,362 - trainer - INFO -   Steps = 7800/40800
+2024-04-26 06:29:22,362 - trainer - INFO -   Instantaneous batch size per GPU = 4 and n_gpu = 4 so the input batch size = 16
+2024-04-26 06:29:22,362 - trainer - INFO -   dev_loss = 1.182209	||	 dev_eval_scores = {'perplexity': 3.2615699768066406}
+2024-04-26 06:29:22,362 - trainer - INFO -   train_loss = 1.9121544361114502
+2024-04-26 06:29:22,362 - trainer - INFO - 
+********************************************
+2024-04-26 06:35:42,699 - trainer - INFO -    Save model to tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 06:35:48,617 - trainer - INFO -   Save check-point at epoch=0 step=8000
+2024-04-26 06:35:48,617 - trainer - INFO -    ***** Evaluation report *****
+2024-04-26 06:35:48,618 - trainer - INFO -   Output path (short): tmp/model/distilgpt2_fine_tuned_coder
+2024-04-26 06:35:48,618 - trainer - INFO -   Early stop on: perplexity
+2024-04-26 06:35:48,618 - trainer - INFO -   Early stop count = 0/3
+2024-04-26 06:35:48,618 - trainer - INFO -   Eval steps = 200 or (iterations = 200)
+2024-04-26 06:35:48,618 - trainer - INFO -   Best score (perplexity) = -3.232813835144043
+2024-04-26 06:35:48,618 - trainer - INFO -   Gradient Accumulation steps = 1
+2024-04-26 06:35:48,618 - trainer - INFO -   Num of training examples (actually no. of iterations per epoch for Iterable Dataset)  = 130556
+2024-04-26 06:35:48,618 - trainer - INFO -   Num of development examples (actually no. of iterations per epoch for Iterable Dataset) = 14507
+2024-04-26 06:35:48,618 - trainer - INFO -   Time spent since last evaluation = 0h 6m 26s
+2024-04-26 06:35:48,618 - trainer - INFO -   Epoch = 1/5
+2024-04-26 06:35:48,618 - trainer - INFO -   Steps = 8000/40800
+2024-04-26 06:35:48,618 - trainer - INFO -   Instantaneous batch size per GPU = 4 and n_gpu = 4 so the input batch size = 16
+2024-04-26 06:35:48,618 - trainer - INFO -   dev_loss = 1.173353	||	 dev_eval_scores = {'perplexity': 3.232813835144043}
+2024-04-26 06:35:48,618 - trainer - INFO -   train_loss = 1.8961435556411743
+2024-04-26 06:35:48,619 - trainer - INFO - 
+********************************************
+2024-04-26 06:37:54,340 - trainer - INFO - epoch 1 ends, 4 epoches left
+2024-04-26 06:37:54,862 - trainer - INFO - 
+global_average_loss=1.8839226961135864,global_steps=8160 on training set

modules.json

@@ -0,0 +1,14 @@
+[
+  {
+    "idx": 0,
+    "name": "0",
+    "path": "0_GPTSingleHead",
+    "type": "model"
+  },
+  {
+    "idx": 1,
+    "name": "1",
+    "path": "1_EmptyHeads",
+    "type": "model"
+  }
+]