Upload batch 15/24: sst2/mBERT/gated_multi_branch/seed_44 ... sst2/mBERT/hf_sequence_classifier/seed_44
Browse files- mbert_rtx6000_metrics/result_ablation_mbert/sst2/mBERT/gated_multi_branch/seed_44/all_results.json +15 -0
- mbert_rtx6000_metrics/result_ablation_mbert/sst2/mBERT/gated_multi_branch/seed_44/eval_results.json +9 -0
- mbert_rtx6000_metrics/result_ablation_mbert/sst2/mBERT/gated_multi_branch/seed_44/run_command.txt +1 -0
- mbert_rtx6000_metrics/result_ablation_mbert/sst2/mBERT/gated_multi_branch/seed_44/train_results.json +9 -0
- mbert_rtx6000_metrics/result_ablation_mbert/sst2/mBERT/gated_multi_branch/seed_44/trainer_state.json +1245 -0
- mbert_rtx6000_metrics/result_ablation_mbert/sst2/mBERT/hf_sequence_classifier/seed_42/all_results.json +15 -0
- mbert_rtx6000_metrics/result_ablation_mbert/sst2/mBERT/hf_sequence_classifier/seed_42/eval_results.json +9 -0
- mbert_rtx6000_metrics/result_ablation_mbert/sst2/mBERT/hf_sequence_classifier/seed_42/run_command.txt +1 -0
- mbert_rtx6000_metrics/result_ablation_mbert/sst2/mBERT/hf_sequence_classifier/seed_42/train_results.json +9 -0
- mbert_rtx6000_metrics/result_ablation_mbert/sst2/mBERT/hf_sequence_classifier/seed_42/trainer_state.json +1245 -0
- mbert_rtx6000_metrics/result_ablation_mbert/sst2/mBERT/hf_sequence_classifier/seed_43/all_results.json +15 -0
- mbert_rtx6000_metrics/result_ablation_mbert/sst2/mBERT/hf_sequence_classifier/seed_43/eval_results.json +9 -0
- mbert_rtx6000_metrics/result_ablation_mbert/sst2/mBERT/hf_sequence_classifier/seed_43/run_command.txt +1 -0
- mbert_rtx6000_metrics/result_ablation_mbert/sst2/mBERT/hf_sequence_classifier/seed_43/train_results.json +9 -0
- mbert_rtx6000_metrics/result_ablation_mbert/sst2/mBERT/hf_sequence_classifier/seed_43/trainer_state.json +1245 -0
- mbert_rtx6000_metrics/result_ablation_mbert/sst2/mBERT/hf_sequence_classifier/seed_44/all_results.json +15 -0
- mbert_rtx6000_metrics/result_ablation_mbert/sst2/mBERT/hf_sequence_classifier/seed_44/eval_results.json +9 -0
- mbert_rtx6000_metrics/result_ablation_mbert/sst2/mBERT/hf_sequence_classifier/seed_44/run_command.txt +1 -0
- mbert_rtx6000_metrics/result_ablation_mbert/sst2/mBERT/hf_sequence_classifier/seed_44/train_results.json +9 -0
- mbert_rtx6000_metrics/result_ablation_mbert/sst2/mBERT/hf_sequence_classifier/seed_44/trainer_state.json +1245 -0
mbert_rtx6000_metrics/result_ablation_mbert/sst2/mBERT/gated_multi_branch/seed_44/all_results.json
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"eval_accuracy": 0.8876146788990825,
|
| 4 |
+
"eval_loss": 0.41072866320610046,
|
| 5 |
+
"eval_runtime": 2.7528,
|
| 6 |
+
"eval_samples": 872,
|
| 7 |
+
"eval_samples_per_second": 316.769,
|
| 8 |
+
"eval_steps_per_second": 3.633,
|
| 9 |
+
"total_flos": 6385930423743024.0,
|
| 10 |
+
"train_loss": 0.2325162212280353,
|
| 11 |
+
"train_runtime": 539.9254,
|
| 12 |
+
"train_samples": 67349,
|
| 13 |
+
"train_samples_per_second": 374.213,
|
| 14 |
+
"train_steps_per_second": 7.801
|
| 15 |
+
}
|
mbert_rtx6000_metrics/result_ablation_mbert/sst2/mBERT/gated_multi_branch/seed_44/eval_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"eval_accuracy": 0.8876146788990825,
|
| 4 |
+
"eval_loss": 0.41072866320610046,
|
| 5 |
+
"eval_runtime": 2.7528,
|
| 6 |
+
"eval_samples": 872,
|
| 7 |
+
"eval_samples_per_second": 316.769,
|
| 8 |
+
"eval_steps_per_second": 3.633
|
| 9 |
+
}
|
mbert_rtx6000_metrics/result_ablation_mbert/sst2/mBERT/gated_multi_branch/seed_44/run_command.txt
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
/usr/local/bin/python /workspace/run_glue_MHA_gated.py --model_name_or_path bert-base-multilingual-cased --dataset_name tmnam20/ViGLUE --task_name sst2 --max_seq_length 128 --do_train --do_eval --evaluation_strategy epoch --save_strategy no --logging_strategy steps --logging_steps 25 --report_to none --num_train_epochs 3 --learning_rate 3e-05 --per_device_train_batch_size 48 --per_device_eval_batch_size 96 --gradient_accumulation_steps 1 --dataloader_num_workers 1 --seed 44 --output_dir result_ablation_mbert/sst2/mBERT/gated_multi_branch/seed_44 --overwrite_output_dir --trust_remote_code true --bf16 --tf32 true --skip_model_save --use_mha true --pooling_strategy gated_multi_branch --n_heads 4 --head_dim 32 --classifier_hidden_dim 512 --head_dropout 0.05
|
mbert_rtx6000_metrics/result_ablation_mbert/sst2/mBERT/gated_multi_branch/seed_44/train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"total_flos": 6385930423743024.0,
|
| 4 |
+
"train_loss": 0.2325162212280353,
|
| 5 |
+
"train_runtime": 539.9254,
|
| 6 |
+
"train_samples": 67349,
|
| 7 |
+
"train_samples_per_second": 374.213,
|
| 8 |
+
"train_steps_per_second": 7.801
|
| 9 |
+
}
|
mbert_rtx6000_metrics/result_ablation_mbert/sst2/mBERT/gated_multi_branch/seed_44/trainer_state.json
ADDED
|
@@ -0,0 +1,1245 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 3.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 4212,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.017806267806267807,
|
| 13 |
+
"grad_norm": 3.845249891281128,
|
| 14 |
+
"learning_rate": 2.9821937321937322e-05,
|
| 15 |
+
"loss": 0.6407,
|
| 16 |
+
"step": 25
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.03561253561253561,
|
| 20 |
+
"grad_norm": 10.928462028503418,
|
| 21 |
+
"learning_rate": 2.9643874643874644e-05,
|
| 22 |
+
"loss": 0.5261,
|
| 23 |
+
"step": 50
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.053418803418803416,
|
| 27 |
+
"grad_norm": 6.390876770019531,
|
| 28 |
+
"learning_rate": 2.9465811965811966e-05,
|
| 29 |
+
"loss": 0.5275,
|
| 30 |
+
"step": 75
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.07122507122507123,
|
| 34 |
+
"grad_norm": 5.588024616241455,
|
| 35 |
+
"learning_rate": 2.9287749287749288e-05,
|
| 36 |
+
"loss": 0.4763,
|
| 37 |
+
"step": 100
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.08903133903133903,
|
| 41 |
+
"grad_norm": 4.094708442687988,
|
| 42 |
+
"learning_rate": 2.910968660968661e-05,
|
| 43 |
+
"loss": 0.4665,
|
| 44 |
+
"step": 125
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.10683760683760683,
|
| 48 |
+
"grad_norm": 5.231828212738037,
|
| 49 |
+
"learning_rate": 2.893162393162393e-05,
|
| 50 |
+
"loss": 0.4244,
|
| 51 |
+
"step": 150
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.12464387464387465,
|
| 55 |
+
"grad_norm": 5.592471599578857,
|
| 56 |
+
"learning_rate": 2.8753561253561256e-05,
|
| 57 |
+
"loss": 0.4147,
|
| 58 |
+
"step": 175
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.14245014245014245,
|
| 62 |
+
"grad_norm": 5.000471115112305,
|
| 63 |
+
"learning_rate": 2.8575498575498578e-05,
|
| 64 |
+
"loss": 0.4154,
|
| 65 |
+
"step": 200
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.16025641025641027,
|
| 69 |
+
"grad_norm": 6.324474334716797,
|
| 70 |
+
"learning_rate": 2.8397435897435896e-05,
|
| 71 |
+
"loss": 0.4382,
|
| 72 |
+
"step": 225
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.17806267806267806,
|
| 76 |
+
"grad_norm": 6.329110622406006,
|
| 77 |
+
"learning_rate": 2.821937321937322e-05,
|
| 78 |
+
"loss": 0.4274,
|
| 79 |
+
"step": 250
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.19586894586894588,
|
| 83 |
+
"grad_norm": 7.549215793609619,
|
| 84 |
+
"learning_rate": 2.8041310541310543e-05,
|
| 85 |
+
"loss": 0.3923,
|
| 86 |
+
"step": 275
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.21367521367521367,
|
| 90 |
+
"grad_norm": 6.854594707489014,
|
| 91 |
+
"learning_rate": 2.786324786324786e-05,
|
| 92 |
+
"loss": 0.3977,
|
| 93 |
+
"step": 300
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.23148148148148148,
|
| 97 |
+
"grad_norm": 6.417609214782715,
|
| 98 |
+
"learning_rate": 2.7685185185185186e-05,
|
| 99 |
+
"loss": 0.3796,
|
| 100 |
+
"step": 325
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.2492877492877493,
|
| 104 |
+
"grad_norm": 4.953665733337402,
|
| 105 |
+
"learning_rate": 2.7507122507122508e-05,
|
| 106 |
+
"loss": 0.385,
|
| 107 |
+
"step": 350
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.2670940170940171,
|
| 111 |
+
"grad_norm": 5.475231647491455,
|
| 112 |
+
"learning_rate": 2.732905982905983e-05,
|
| 113 |
+
"loss": 0.3691,
|
| 114 |
+
"step": 375
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.2849002849002849,
|
| 118 |
+
"grad_norm": 4.771439075469971,
|
| 119 |
+
"learning_rate": 2.715099715099715e-05,
|
| 120 |
+
"loss": 0.3582,
|
| 121 |
+
"step": 400
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.3027065527065527,
|
| 125 |
+
"grad_norm": 5.15997838973999,
|
| 126 |
+
"learning_rate": 2.6972934472934473e-05,
|
| 127 |
+
"loss": 0.3674,
|
| 128 |
+
"step": 425
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.32051282051282054,
|
| 132 |
+
"grad_norm": 6.196379661560059,
|
| 133 |
+
"learning_rate": 2.67948717948718e-05,
|
| 134 |
+
"loss": 0.3471,
|
| 135 |
+
"step": 450
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.3383190883190883,
|
| 139 |
+
"grad_norm": 11.954277038574219,
|
| 140 |
+
"learning_rate": 2.6616809116809117e-05,
|
| 141 |
+
"loss": 0.3786,
|
| 142 |
+
"step": 475
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.3561253561253561,
|
| 146 |
+
"grad_norm": 6.4036784172058105,
|
| 147 |
+
"learning_rate": 2.643874643874644e-05,
|
| 148 |
+
"loss": 0.3527,
|
| 149 |
+
"step": 500
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.37393162393162394,
|
| 153 |
+
"grad_norm": 5.163303852081299,
|
| 154 |
+
"learning_rate": 2.6260683760683763e-05,
|
| 155 |
+
"loss": 0.3635,
|
| 156 |
+
"step": 525
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.39173789173789175,
|
| 160 |
+
"grad_norm": 6.5261101722717285,
|
| 161 |
+
"learning_rate": 2.6082621082621082e-05,
|
| 162 |
+
"loss": 0.3822,
|
| 163 |
+
"step": 550
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.40954415954415957,
|
| 167 |
+
"grad_norm": 6.685307025909424,
|
| 168 |
+
"learning_rate": 2.5904558404558404e-05,
|
| 169 |
+
"loss": 0.3208,
|
| 170 |
+
"step": 575
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.42735042735042733,
|
| 174 |
+
"grad_norm": 6.99289608001709,
|
| 175 |
+
"learning_rate": 2.572649572649573e-05,
|
| 176 |
+
"loss": 0.3447,
|
| 177 |
+
"step": 600
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.44515669515669515,
|
| 181 |
+
"grad_norm": 6.893125534057617,
|
| 182 |
+
"learning_rate": 2.5548433048433047e-05,
|
| 183 |
+
"loss": 0.3456,
|
| 184 |
+
"step": 625
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.46296296296296297,
|
| 188 |
+
"grad_norm": 9.95699405670166,
|
| 189 |
+
"learning_rate": 2.5370370370370372e-05,
|
| 190 |
+
"loss": 0.3369,
|
| 191 |
+
"step": 650
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.4807692307692308,
|
| 195 |
+
"grad_norm": 10.884265899658203,
|
| 196 |
+
"learning_rate": 2.5192307692307694e-05,
|
| 197 |
+
"loss": 0.3169,
|
| 198 |
+
"step": 675
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.4985754985754986,
|
| 202 |
+
"grad_norm": 5.716195106506348,
|
| 203 |
+
"learning_rate": 2.5014245014245015e-05,
|
| 204 |
+
"loss": 0.3317,
|
| 205 |
+
"step": 700
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.5163817663817664,
|
| 209 |
+
"grad_norm": 5.162127494812012,
|
| 210 |
+
"learning_rate": 2.4836182336182337e-05,
|
| 211 |
+
"loss": 0.3429,
|
| 212 |
+
"step": 725
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.5341880341880342,
|
| 216 |
+
"grad_norm": 6.0887370109558105,
|
| 217 |
+
"learning_rate": 2.465811965811966e-05,
|
| 218 |
+
"loss": 0.3071,
|
| 219 |
+
"step": 750
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.551994301994302,
|
| 223 |
+
"grad_norm": 5.057008743286133,
|
| 224 |
+
"learning_rate": 2.448005698005698e-05,
|
| 225 |
+
"loss": 0.3481,
|
| 226 |
+
"step": 775
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.5698005698005698,
|
| 230 |
+
"grad_norm": 4.378439903259277,
|
| 231 |
+
"learning_rate": 2.4301994301994302e-05,
|
| 232 |
+
"loss": 0.2881,
|
| 233 |
+
"step": 800
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.5876068376068376,
|
| 237 |
+
"grad_norm": 5.914629936218262,
|
| 238 |
+
"learning_rate": 2.4123931623931624e-05,
|
| 239 |
+
"loss": 0.3493,
|
| 240 |
+
"step": 825
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.6054131054131054,
|
| 244 |
+
"grad_norm": 5.119071006774902,
|
| 245 |
+
"learning_rate": 2.3945868945868946e-05,
|
| 246 |
+
"loss": 0.3229,
|
| 247 |
+
"step": 850
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.6232193732193733,
|
| 251 |
+
"grad_norm": 5.722135066986084,
|
| 252 |
+
"learning_rate": 2.3767806267806267e-05,
|
| 253 |
+
"loss": 0.3316,
|
| 254 |
+
"step": 875
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.6410256410256411,
|
| 258 |
+
"grad_norm": 5.101425647735596,
|
| 259 |
+
"learning_rate": 2.358974358974359e-05,
|
| 260 |
+
"loss": 0.2857,
|
| 261 |
+
"step": 900
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.6588319088319088,
|
| 265 |
+
"grad_norm": 7.980036735534668,
|
| 266 |
+
"learning_rate": 2.3411680911680914e-05,
|
| 267 |
+
"loss": 0.3169,
|
| 268 |
+
"step": 925
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.6766381766381766,
|
| 272 |
+
"grad_norm": 8.392170906066895,
|
| 273 |
+
"learning_rate": 2.3233618233618236e-05,
|
| 274 |
+
"loss": 0.3021,
|
| 275 |
+
"step": 950
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.6944444444444444,
|
| 279 |
+
"grad_norm": 4.117437839508057,
|
| 280 |
+
"learning_rate": 2.3055555555555554e-05,
|
| 281 |
+
"loss": 0.3053,
|
| 282 |
+
"step": 975
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.7122507122507122,
|
| 286 |
+
"grad_norm": 4.455058574676514,
|
| 287 |
+
"learning_rate": 2.287749287749288e-05,
|
| 288 |
+
"loss": 0.2874,
|
| 289 |
+
"step": 1000
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.73005698005698,
|
| 293 |
+
"grad_norm": 5.499744892120361,
|
| 294 |
+
"learning_rate": 2.26994301994302e-05,
|
| 295 |
+
"loss": 0.2852,
|
| 296 |
+
"step": 1025
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.7478632478632479,
|
| 300 |
+
"grad_norm": 7.632195949554443,
|
| 301 |
+
"learning_rate": 2.252136752136752e-05,
|
| 302 |
+
"loss": 0.3101,
|
| 303 |
+
"step": 1050
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.7656695156695157,
|
| 307 |
+
"grad_norm": 5.61528205871582,
|
| 308 |
+
"learning_rate": 2.2343304843304845e-05,
|
| 309 |
+
"loss": 0.3012,
|
| 310 |
+
"step": 1075
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.7834757834757835,
|
| 314 |
+
"grad_norm": 7.464178085327148,
|
| 315 |
+
"learning_rate": 2.2165242165242166e-05,
|
| 316 |
+
"loss": 0.3175,
|
| 317 |
+
"step": 1100
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 0.8012820512820513,
|
| 321 |
+
"grad_norm": 6.706811428070068,
|
| 322 |
+
"learning_rate": 2.1987179487179485e-05,
|
| 323 |
+
"loss": 0.2825,
|
| 324 |
+
"step": 1125
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.8190883190883191,
|
| 328 |
+
"grad_norm": 4.524678707122803,
|
| 329 |
+
"learning_rate": 2.180911680911681e-05,
|
| 330 |
+
"loss": 0.2935,
|
| 331 |
+
"step": 1150
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 0.8368945868945868,
|
| 335 |
+
"grad_norm": 6.103963851928711,
|
| 336 |
+
"learning_rate": 2.163105413105413e-05,
|
| 337 |
+
"loss": 0.289,
|
| 338 |
+
"step": 1175
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.8547008547008547,
|
| 342 |
+
"grad_norm": 3.6143007278442383,
|
| 343 |
+
"learning_rate": 2.1452991452991456e-05,
|
| 344 |
+
"loss": 0.2996,
|
| 345 |
+
"step": 1200
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.8725071225071225,
|
| 349 |
+
"grad_norm": 6.052546977996826,
|
| 350 |
+
"learning_rate": 2.1274928774928775e-05,
|
| 351 |
+
"loss": 0.2595,
|
| 352 |
+
"step": 1225
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 0.8903133903133903,
|
| 356 |
+
"grad_norm": 7.750393867492676,
|
| 357 |
+
"learning_rate": 2.1096866096866097e-05,
|
| 358 |
+
"loss": 0.288,
|
| 359 |
+
"step": 1250
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 0.9081196581196581,
|
| 363 |
+
"grad_norm": 6.529996871948242,
|
| 364 |
+
"learning_rate": 2.091880341880342e-05,
|
| 365 |
+
"loss": 0.2874,
|
| 366 |
+
"step": 1275
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 0.9259259259259259,
|
| 370 |
+
"grad_norm": 5.307050704956055,
|
| 371 |
+
"learning_rate": 2.074074074074074e-05,
|
| 372 |
+
"loss": 0.3164,
|
| 373 |
+
"step": 1300
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 0.9437321937321937,
|
| 377 |
+
"grad_norm": 3.603550672531128,
|
| 378 |
+
"learning_rate": 2.056267806267806e-05,
|
| 379 |
+
"loss": 0.3019,
|
| 380 |
+
"step": 1325
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 0.9615384615384616,
|
| 384 |
+
"grad_norm": 2.40566349029541,
|
| 385 |
+
"learning_rate": 2.0384615384615387e-05,
|
| 386 |
+
"loss": 0.2403,
|
| 387 |
+
"step": 1350
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 0.9793447293447294,
|
| 391 |
+
"grad_norm": 3.5955400466918945,
|
| 392 |
+
"learning_rate": 2.0206552706552705e-05,
|
| 393 |
+
"loss": 0.2857,
|
| 394 |
+
"step": 1375
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 0.9971509971509972,
|
| 398 |
+
"grad_norm": 5.2943854331970215,
|
| 399 |
+
"learning_rate": 2.002849002849003e-05,
|
| 400 |
+
"loss": 0.2904,
|
| 401 |
+
"step": 1400
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 1.0,
|
| 405 |
+
"eval_accuracy": 0.8727064220183486,
|
| 406 |
+
"eval_loss": 0.32926374673843384,
|
| 407 |
+
"eval_runtime": 3.385,
|
| 408 |
+
"eval_samples_per_second": 257.61,
|
| 409 |
+
"eval_steps_per_second": 2.954,
|
| 410 |
+
"step": 1404
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 1.014957264957265,
|
| 414 |
+
"grad_norm": 8.587410926818848,
|
| 415 |
+
"learning_rate": 1.9850427350427352e-05,
|
| 416 |
+
"loss": 0.2067,
|
| 417 |
+
"step": 1425
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"epoch": 1.0327635327635327,
|
| 421 |
+
"grad_norm": 9.450316429138184,
|
| 422 |
+
"learning_rate": 1.9672364672364674e-05,
|
| 423 |
+
"loss": 0.2291,
|
| 424 |
+
"step": 1450
|
| 425 |
+
},
|
| 426 |
+
{
|
| 427 |
+
"epoch": 1.0505698005698005,
|
| 428 |
+
"grad_norm": 9.076273918151855,
|
| 429 |
+
"learning_rate": 1.9494301994301995e-05,
|
| 430 |
+
"loss": 0.207,
|
| 431 |
+
"step": 1475
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 1.0683760683760684,
|
| 435 |
+
"grad_norm": 5.560942649841309,
|
| 436 |
+
"learning_rate": 1.9316239316239317e-05,
|
| 437 |
+
"loss": 0.2119,
|
| 438 |
+
"step": 1500
|
| 439 |
+
},
|
| 440 |
+
{
|
| 441 |
+
"epoch": 1.0861823361823362,
|
| 442 |
+
"grad_norm": 5.7185163497924805,
|
| 443 |
+
"learning_rate": 1.913817663817664e-05,
|
| 444 |
+
"loss": 0.2053,
|
| 445 |
+
"step": 1525
|
| 446 |
+
},
|
| 447 |
+
{
|
| 448 |
+
"epoch": 1.103988603988604,
|
| 449 |
+
"grad_norm": 5.721456050872803,
|
| 450 |
+
"learning_rate": 1.896011396011396e-05,
|
| 451 |
+
"loss": 0.2062,
|
| 452 |
+
"step": 1550
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 1.1217948717948718,
|
| 456 |
+
"grad_norm": 6.074375629425049,
|
| 457 |
+
"learning_rate": 1.8782051282051282e-05,
|
| 458 |
+
"loss": 0.1793,
|
| 459 |
+
"step": 1575
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 1.1396011396011396,
|
| 463 |
+
"grad_norm": 5.265392303466797,
|
| 464 |
+
"learning_rate": 1.8603988603988604e-05,
|
| 465 |
+
"loss": 0.2323,
|
| 466 |
+
"step": 1600
|
| 467 |
+
},
|
| 468 |
+
{
|
| 469 |
+
"epoch": 1.1574074074074074,
|
| 470 |
+
"grad_norm": 9.503076553344727,
|
| 471 |
+
"learning_rate": 1.8425925925925926e-05,
|
| 472 |
+
"loss": 0.2184,
|
| 473 |
+
"step": 1625
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 1.1752136752136753,
|
| 477 |
+
"grad_norm": 13.069637298583984,
|
| 478 |
+
"learning_rate": 1.8247863247863247e-05,
|
| 479 |
+
"loss": 0.2333,
|
| 480 |
+
"step": 1650
|
| 481 |
+
},
|
| 482 |
+
{
|
| 483 |
+
"epoch": 1.193019943019943,
|
| 484 |
+
"grad_norm": 5.102323055267334,
|
| 485 |
+
"learning_rate": 1.8069800569800572e-05,
|
| 486 |
+
"loss": 0.2249,
|
| 487 |
+
"step": 1675
|
| 488 |
+
},
|
| 489 |
+
{
|
| 490 |
+
"epoch": 1.210826210826211,
|
| 491 |
+
"grad_norm": 6.223441123962402,
|
| 492 |
+
"learning_rate": 1.7891737891737894e-05,
|
| 493 |
+
"loss": 0.2341,
|
| 494 |
+
"step": 1700
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 1.2286324786324787,
|
| 498 |
+
"grad_norm": 7.7142462730407715,
|
| 499 |
+
"learning_rate": 1.7713675213675212e-05,
|
| 500 |
+
"loss": 0.2234,
|
| 501 |
+
"step": 1725
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 1.2464387464387465,
|
| 505 |
+
"grad_norm": 8.117522239685059,
|
| 506 |
+
"learning_rate": 1.7535612535612538e-05,
|
| 507 |
+
"loss": 0.2052,
|
| 508 |
+
"step": 1750
|
| 509 |
+
},
|
| 510 |
+
{
|
| 511 |
+
"epoch": 1.2642450142450143,
|
| 512 |
+
"grad_norm": 4.956342697143555,
|
| 513 |
+
"learning_rate": 1.735754985754986e-05,
|
| 514 |
+
"loss": 0.2193,
|
| 515 |
+
"step": 1775
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 1.282051282051282,
|
| 519 |
+
"grad_norm": 7.980919361114502,
|
| 520 |
+
"learning_rate": 1.7179487179487178e-05,
|
| 521 |
+
"loss": 0.2098,
|
| 522 |
+
"step": 1800
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 1.29985754985755,
|
| 526 |
+
"grad_norm": 5.0420637130737305,
|
| 527 |
+
"learning_rate": 1.7001424501424503e-05,
|
| 528 |
+
"loss": 0.1807,
|
| 529 |
+
"step": 1825
|
| 530 |
+
},
|
| 531 |
+
{
|
| 532 |
+
"epoch": 1.3176638176638176,
|
| 533 |
+
"grad_norm": 3.8203840255737305,
|
| 534 |
+
"learning_rate": 1.6823361823361824e-05,
|
| 535 |
+
"loss": 0.2394,
|
| 536 |
+
"step": 1850
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 1.3354700854700854,
|
| 540 |
+
"grad_norm": 8.171833992004395,
|
| 541 |
+
"learning_rate": 1.6645299145299143e-05,
|
| 542 |
+
"loss": 0.2133,
|
| 543 |
+
"step": 1875
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 1.3532763532763532,
|
| 547 |
+
"grad_norm": 7.3192596435546875,
|
| 548 |
+
"learning_rate": 1.6467236467236468e-05,
|
| 549 |
+
"loss": 0.2297,
|
| 550 |
+
"step": 1900
|
| 551 |
+
},
|
| 552 |
+
{
|
| 553 |
+
"epoch": 1.371082621082621,
|
| 554 |
+
"grad_norm": 6.255660057067871,
|
| 555 |
+
"learning_rate": 1.628917378917379e-05,
|
| 556 |
+
"loss": 0.2006,
|
| 557 |
+
"step": 1925
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"epoch": 1.3888888888888888,
|
| 561 |
+
"grad_norm": 4.441915512084961,
|
| 562 |
+
"learning_rate": 1.6111111111111115e-05,
|
| 563 |
+
"loss": 0.2093,
|
| 564 |
+
"step": 1950
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 1.4066951566951567,
|
| 568 |
+
"grad_norm": 3.8488996028900146,
|
| 569 |
+
"learning_rate": 1.5933048433048433e-05,
|
| 570 |
+
"loss": 0.1901,
|
| 571 |
+
"step": 1975
|
| 572 |
+
},
|
| 573 |
+
{
|
| 574 |
+
"epoch": 1.4245014245014245,
|
| 575 |
+
"grad_norm": 5.333924293518066,
|
| 576 |
+
"learning_rate": 1.5754985754985755e-05,
|
| 577 |
+
"loss": 0.2421,
|
| 578 |
+
"step": 2000
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 1.4423076923076923,
|
| 582 |
+
"grad_norm": 7.383884906768799,
|
| 583 |
+
"learning_rate": 1.557692307692308e-05,
|
| 584 |
+
"loss": 0.2201,
|
| 585 |
+
"step": 2025
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 1.46011396011396,
|
| 589 |
+
"grad_norm": 6.1271443367004395,
|
| 590 |
+
"learning_rate": 1.5398860398860398e-05,
|
| 591 |
+
"loss": 0.2007,
|
| 592 |
+
"step": 2050
|
| 593 |
+
},
|
| 594 |
+
{
|
| 595 |
+
"epoch": 1.477920227920228,
|
| 596 |
+
"grad_norm": 7.160142421722412,
|
| 597 |
+
"learning_rate": 1.5220797720797721e-05,
|
| 598 |
+
"loss": 0.2182,
|
| 599 |
+
"step": 2075
|
| 600 |
+
},
|
| 601 |
+
{
|
| 602 |
+
"epoch": 1.4957264957264957,
|
| 603 |
+
"grad_norm": 3.9491279125213623,
|
| 604 |
+
"learning_rate": 1.5042735042735045e-05,
|
| 605 |
+
"loss": 0.2101,
|
| 606 |
+
"step": 2100
|
| 607 |
+
},
|
| 608 |
+
{
|
| 609 |
+
"epoch": 1.5135327635327636,
|
| 610 |
+
"grad_norm": 9.13621997833252,
|
| 611 |
+
"learning_rate": 1.4864672364672365e-05,
|
| 612 |
+
"loss": 0.1874,
|
| 613 |
+
"step": 2125
|
| 614 |
+
},
|
| 615 |
+
{
|
| 616 |
+
"epoch": 1.5313390313390314,
|
| 617 |
+
"grad_norm": 6.676900386810303,
|
| 618 |
+
"learning_rate": 1.4686609686609687e-05,
|
| 619 |
+
"loss": 0.2537,
|
| 620 |
+
"step": 2150
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 1.5491452991452992,
|
| 624 |
+
"grad_norm": 3.3525915145874023,
|
| 625 |
+
"learning_rate": 1.450854700854701e-05,
|
| 626 |
+
"loss": 0.2057,
|
| 627 |
+
"step": 2175
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"epoch": 1.566951566951567,
|
| 631 |
+
"grad_norm": 5.00832462310791,
|
| 632 |
+
"learning_rate": 1.433048433048433e-05,
|
| 633 |
+
"loss": 0.1971,
|
| 634 |
+
"step": 2200
|
| 635 |
+
},
|
| 636 |
+
{
|
| 637 |
+
"epoch": 1.5847578347578346,
|
| 638 |
+
"grad_norm": 8.427897453308105,
|
| 639 |
+
"learning_rate": 1.4152421652421652e-05,
|
| 640 |
+
"loss": 0.1981,
|
| 641 |
+
"step": 2225
|
| 642 |
+
},
|
| 643 |
+
{
|
| 644 |
+
"epoch": 1.6025641025641026,
|
| 645 |
+
"grad_norm": 3.4777560234069824,
|
| 646 |
+
"learning_rate": 1.3974358974358975e-05,
|
| 647 |
+
"loss": 0.1994,
|
| 648 |
+
"step": 2250
|
| 649 |
+
},
|
| 650 |
+
{
|
| 651 |
+
"epoch": 1.6203703703703702,
|
| 652 |
+
"grad_norm": 8.663381576538086,
|
| 653 |
+
"learning_rate": 1.3796296296296297e-05,
|
| 654 |
+
"loss": 0.1842,
|
| 655 |
+
"step": 2275
|
| 656 |
+
},
|
| 657 |
+
{
|
| 658 |
+
"epoch": 1.6381766381766383,
|
| 659 |
+
"grad_norm": 3.203263998031616,
|
| 660 |
+
"learning_rate": 1.3618233618233619e-05,
|
| 661 |
+
"loss": 0.195,
|
| 662 |
+
"step": 2300
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 1.6559829059829059,
|
| 666 |
+
"grad_norm": 4.679374694824219,
|
| 667 |
+
"learning_rate": 1.344017094017094e-05,
|
| 668 |
+
"loss": 0.1845,
|
| 669 |
+
"step": 2325
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 1.673789173789174,
|
| 673 |
+
"grad_norm": 6.6320037841796875,
|
| 674 |
+
"learning_rate": 1.3262108262108262e-05,
|
| 675 |
+
"loss": 0.1623,
|
| 676 |
+
"step": 2350
|
| 677 |
+
},
|
| 678 |
+
{
|
| 679 |
+
"epoch": 1.6915954415954415,
|
| 680 |
+
"grad_norm": 8.756708145141602,
|
| 681 |
+
"learning_rate": 1.3084045584045585e-05,
|
| 682 |
+
"loss": 0.2201,
|
| 683 |
+
"step": 2375
|
| 684 |
+
},
|
| 685 |
+
{
|
| 686 |
+
"epoch": 1.7094017094017095,
|
| 687 |
+
"grad_norm": 5.750958442687988,
|
| 688 |
+
"learning_rate": 1.2905982905982905e-05,
|
| 689 |
+
"loss": 0.2214,
|
| 690 |
+
"step": 2400
|
| 691 |
+
},
|
| 692 |
+
{
|
| 693 |
+
"epoch": 1.7272079772079771,
|
| 694 |
+
"grad_norm": 6.1745500564575195,
|
| 695 |
+
"learning_rate": 1.2727920227920229e-05,
|
| 696 |
+
"loss": 0.2118,
|
| 697 |
+
"step": 2425
|
| 698 |
+
},
|
| 699 |
+
{
|
| 700 |
+
"epoch": 1.7450142450142452,
|
| 701 |
+
"grad_norm": 1.9546324014663696,
|
| 702 |
+
"learning_rate": 1.254985754985755e-05,
|
| 703 |
+
"loss": 0.1743,
|
| 704 |
+
"step": 2450
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 1.7628205128205128,
|
| 708 |
+
"grad_norm": 10.336423873901367,
|
| 709 |
+
"learning_rate": 1.2371794871794872e-05,
|
| 710 |
+
"loss": 0.1948,
|
| 711 |
+
"step": 2475
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"epoch": 1.7806267806267806,
|
| 715 |
+
"grad_norm": 6.67774772644043,
|
| 716 |
+
"learning_rate": 1.2193732193732194e-05,
|
| 717 |
+
"loss": 0.2085,
|
| 718 |
+
"step": 2500
|
| 719 |
+
},
|
| 720 |
+
{
|
| 721 |
+
"epoch": 1.7984330484330484,
|
| 722 |
+
"grad_norm": 5.549098491668701,
|
| 723 |
+
"learning_rate": 1.2015669515669516e-05,
|
| 724 |
+
"loss": 0.201,
|
| 725 |
+
"step": 2525
|
| 726 |
+
},
|
| 727 |
+
{
|
| 728 |
+
"epoch": 1.8162393162393162,
|
| 729 |
+
"grad_norm": 3.348634719848633,
|
| 730 |
+
"learning_rate": 1.1837606837606839e-05,
|
| 731 |
+
"loss": 0.1941,
|
| 732 |
+
"step": 2550
|
| 733 |
+
},
|
| 734 |
+
{
|
| 735 |
+
"epoch": 1.834045584045584,
|
| 736 |
+
"grad_norm": 4.982662677764893,
|
| 737 |
+
"learning_rate": 1.1659544159544159e-05,
|
| 738 |
+
"loss": 0.2189,
|
| 739 |
+
"step": 2575
|
| 740 |
+
},
|
| 741 |
+
{
|
| 742 |
+
"epoch": 1.8518518518518519,
|
| 743 |
+
"grad_norm": 4.534923076629639,
|
| 744 |
+
"learning_rate": 1.148148148148148e-05,
|
| 745 |
+
"loss": 0.1907,
|
| 746 |
+
"step": 2600
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 1.8696581196581197,
|
| 750 |
+
"grad_norm": 8.87447452545166,
|
| 751 |
+
"learning_rate": 1.1303418803418804e-05,
|
| 752 |
+
"loss": 0.195,
|
| 753 |
+
"step": 2625
|
| 754 |
+
},
|
| 755 |
+
{
|
| 756 |
+
"epoch": 1.8874643874643875,
|
| 757 |
+
"grad_norm": 6.984755039215088,
|
| 758 |
+
"learning_rate": 1.1125356125356126e-05,
|
| 759 |
+
"loss": 0.1819,
|
| 760 |
+
"step": 2650
|
| 761 |
+
},
|
| 762 |
+
{
|
| 763 |
+
"epoch": 1.9052706552706553,
|
| 764 |
+
"grad_norm": 3.8289706707000732,
|
| 765 |
+
"learning_rate": 1.0947293447293448e-05,
|
| 766 |
+
"loss": 0.1908,
|
| 767 |
+
"step": 2675
|
| 768 |
+
},
|
| 769 |
+
{
|
| 770 |
+
"epoch": 1.9230769230769231,
|
| 771 |
+
"grad_norm": 7.2230916023254395,
|
| 772 |
+
"learning_rate": 1.076923076923077e-05,
|
| 773 |
+
"loss": 0.2139,
|
| 774 |
+
"step": 2700
|
| 775 |
+
},
|
| 776 |
+
{
|
| 777 |
+
"epoch": 1.940883190883191,
|
| 778 |
+
"grad_norm": 4.6364850997924805,
|
| 779 |
+
"learning_rate": 1.0591168091168091e-05,
|
| 780 |
+
"loss": 0.1882,
|
| 781 |
+
"step": 2725
|
| 782 |
+
},
|
| 783 |
+
{
|
| 784 |
+
"epoch": 1.9586894586894585,
|
| 785 |
+
"grad_norm": 4.1284894943237305,
|
| 786 |
+
"learning_rate": 1.0413105413105414e-05,
|
| 787 |
+
"loss": 0.2209,
|
| 788 |
+
"step": 2750
|
| 789 |
+
},
|
| 790 |
+
{
|
| 791 |
+
"epoch": 1.9764957264957266,
|
| 792 |
+
"grad_norm": 5.325596332550049,
|
| 793 |
+
"learning_rate": 1.0235042735042734e-05,
|
| 794 |
+
"loss": 0.2116,
|
| 795 |
+
"step": 2775
|
| 796 |
+
},
|
| 797 |
+
{
|
| 798 |
+
"epoch": 1.9943019943019942,
|
| 799 |
+
"grad_norm": 5.905292510986328,
|
| 800 |
+
"learning_rate": 1.0056980056980058e-05,
|
| 801 |
+
"loss": 0.1966,
|
| 802 |
+
"step": 2800
|
| 803 |
+
},
|
| 804 |
+
{
|
| 805 |
+
"epoch": 2.0,
|
| 806 |
+
"eval_accuracy": 0.8738532110091743,
|
| 807 |
+
"eval_loss": 0.36082279682159424,
|
| 808 |
+
"eval_runtime": 3.1146,
|
| 809 |
+
"eval_samples_per_second": 279.974,
|
| 810 |
+
"eval_steps_per_second": 3.211,
|
| 811 |
+
"step": 2808
|
| 812 |
+
},
|
| 813 |
+
{
|
| 814 |
+
"epoch": 2.012108262108262,
|
| 815 |
+
"grad_norm": 7.492818355560303,
|
| 816 |
+
"learning_rate": 9.87891737891738e-06,
|
| 817 |
+
"loss": 0.161,
|
| 818 |
+
"step": 2825
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 2.02991452991453,
|
| 822 |
+
"grad_norm": 5.991679668426514,
|
| 823 |
+
"learning_rate": 9.700854700854701e-06,
|
| 824 |
+
"loss": 0.1618,
|
| 825 |
+
"step": 2850
|
| 826 |
+
},
|
| 827 |
+
{
|
| 828 |
+
"epoch": 2.047720797720798,
|
| 829 |
+
"grad_norm": 7.299553394317627,
|
| 830 |
+
"learning_rate": 9.522792022792023e-06,
|
| 831 |
+
"loss": 0.1382,
|
| 832 |
+
"step": 2875
|
| 833 |
+
},
|
| 834 |
+
{
|
| 835 |
+
"epoch": 2.0655270655270654,
|
| 836 |
+
"grad_norm": 4.278791427612305,
|
| 837 |
+
"learning_rate": 9.344729344729345e-06,
|
| 838 |
+
"loss": 0.1562,
|
| 839 |
+
"step": 2900
|
| 840 |
+
},
|
| 841 |
+
{
|
| 842 |
+
"epoch": 2.0833333333333335,
|
| 843 |
+
"grad_norm": 10.374750137329102,
|
| 844 |
+
"learning_rate": 9.166666666666668e-06,
|
| 845 |
+
"loss": 0.1413,
|
| 846 |
+
"step": 2925
|
| 847 |
+
},
|
| 848 |
+
{
|
| 849 |
+
"epoch": 2.101139601139601,
|
| 850 |
+
"grad_norm": 3.6893279552459717,
|
| 851 |
+
"learning_rate": 8.988603988603988e-06,
|
| 852 |
+
"loss": 0.1243,
|
| 853 |
+
"step": 2950
|
| 854 |
+
},
|
| 855 |
+
{
|
| 856 |
+
"epoch": 2.118945868945869,
|
| 857 |
+
"grad_norm": 6.866631507873535,
|
| 858 |
+
"learning_rate": 8.81054131054131e-06,
|
| 859 |
+
"loss": 0.1328,
|
| 860 |
+
"step": 2975
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 2.1367521367521367,
|
| 864 |
+
"grad_norm": 11.11772632598877,
|
| 865 |
+
"learning_rate": 8.632478632478633e-06,
|
| 866 |
+
"loss": 0.1249,
|
| 867 |
+
"step": 3000
|
| 868 |
+
},
|
| 869 |
+
{
|
| 870 |
+
"epoch": 2.1545584045584047,
|
| 871 |
+
"grad_norm": 7.26266622543335,
|
| 872 |
+
"learning_rate": 8.454415954415955e-06,
|
| 873 |
+
"loss": 0.1291,
|
| 874 |
+
"step": 3025
|
| 875 |
+
},
|
| 876 |
+
{
|
| 877 |
+
"epoch": 2.1723646723646723,
|
| 878 |
+
"grad_norm": 5.019978046417236,
|
| 879 |
+
"learning_rate": 8.276353276353277e-06,
|
| 880 |
+
"loss": 0.1452,
|
| 881 |
+
"step": 3050
|
| 882 |
+
},
|
| 883 |
+
{
|
| 884 |
+
"epoch": 2.1901709401709404,
|
| 885 |
+
"grad_norm": 4.9190568923950195,
|
| 886 |
+
"learning_rate": 8.098290598290598e-06,
|
| 887 |
+
"loss": 0.1499,
|
| 888 |
+
"step": 3075
|
| 889 |
+
},
|
| 890 |
+
{
|
| 891 |
+
"epoch": 2.207977207977208,
|
| 892 |
+
"grad_norm": 5.169571876525879,
|
| 893 |
+
"learning_rate": 7.92022792022792e-06,
|
| 894 |
+
"loss": 0.1446,
|
| 895 |
+
"step": 3100
|
| 896 |
+
},
|
| 897 |
+
{
|
| 898 |
+
"epoch": 2.2257834757834756,
|
| 899 |
+
"grad_norm": 8.046879768371582,
|
| 900 |
+
"learning_rate": 7.742165242165244e-06,
|
| 901 |
+
"loss": 0.1314,
|
| 902 |
+
"step": 3125
|
| 903 |
+
},
|
| 904 |
+
{
|
| 905 |
+
"epoch": 2.2435897435897436,
|
| 906 |
+
"grad_norm": 7.1848859786987305,
|
| 907 |
+
"learning_rate": 7.564102564102564e-06,
|
| 908 |
+
"loss": 0.136,
|
| 909 |
+
"step": 3150
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 2.261396011396011,
|
| 913 |
+
"grad_norm": 3.3299129009246826,
|
| 914 |
+
"learning_rate": 7.386039886039886e-06,
|
| 915 |
+
"loss": 0.1446,
|
| 916 |
+
"step": 3175
|
| 917 |
+
},
|
| 918 |
+
{
|
| 919 |
+
"epoch": 2.2792022792022792,
|
| 920 |
+
"grad_norm": 4.5766401290893555,
|
| 921 |
+
"learning_rate": 7.207977207977209e-06,
|
| 922 |
+
"loss": 0.1386,
|
| 923 |
+
"step": 3200
|
| 924 |
+
},
|
| 925 |
+
{
|
| 926 |
+
"epoch": 2.297008547008547,
|
| 927 |
+
"grad_norm": 6.309956073760986,
|
| 928 |
+
"learning_rate": 7.02991452991453e-06,
|
| 929 |
+
"loss": 0.1409,
|
| 930 |
+
"step": 3225
|
| 931 |
+
},
|
| 932 |
+
{
|
| 933 |
+
"epoch": 2.314814814814815,
|
| 934 |
+
"grad_norm": 10.284469604492188,
|
| 935 |
+
"learning_rate": 6.851851851851852e-06,
|
| 936 |
+
"loss": 0.1454,
|
| 937 |
+
"step": 3250
|
| 938 |
+
},
|
| 939 |
+
{
|
| 940 |
+
"epoch": 2.3326210826210825,
|
| 941 |
+
"grad_norm": 4.707538604736328,
|
| 942 |
+
"learning_rate": 6.673789173789174e-06,
|
| 943 |
+
"loss": 0.1395,
|
| 944 |
+
"step": 3275
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 2.3504273504273505,
|
| 948 |
+
"grad_norm": 11.325845718383789,
|
| 949 |
+
"learning_rate": 6.4957264957264955e-06,
|
| 950 |
+
"loss": 0.1543,
|
| 951 |
+
"step": 3300
|
| 952 |
+
},
|
| 953 |
+
{
|
| 954 |
+
"epoch": 2.368233618233618,
|
| 955 |
+
"grad_norm": 5.101810455322266,
|
| 956 |
+
"learning_rate": 6.317663817663818e-06,
|
| 957 |
+
"loss": 0.1269,
|
| 958 |
+
"step": 3325
|
| 959 |
+
},
|
| 960 |
+
{
|
| 961 |
+
"epoch": 2.386039886039886,
|
| 962 |
+
"grad_norm": 6.664872646331787,
|
| 963 |
+
"learning_rate": 6.13960113960114e-06,
|
| 964 |
+
"loss": 0.1506,
|
| 965 |
+
"step": 3350
|
| 966 |
+
},
|
| 967 |
+
{
|
| 968 |
+
"epoch": 2.4038461538461537,
|
| 969 |
+
"grad_norm": 6.691899299621582,
|
| 970 |
+
"learning_rate": 5.9615384615384615e-06,
|
| 971 |
+
"loss": 0.1034,
|
| 972 |
+
"step": 3375
|
| 973 |
+
},
|
| 974 |
+
{
|
| 975 |
+
"epoch": 2.421652421652422,
|
| 976 |
+
"grad_norm": 7.85353946685791,
|
| 977 |
+
"learning_rate": 5.783475783475783e-06,
|
| 978 |
+
"loss": 0.1184,
|
| 979 |
+
"step": 3400
|
| 980 |
+
},
|
| 981 |
+
{
|
| 982 |
+
"epoch": 2.4394586894586894,
|
| 983 |
+
"grad_norm": 3.5622007846832275,
|
| 984 |
+
"learning_rate": 5.605413105413106e-06,
|
| 985 |
+
"loss": 0.1634,
|
| 986 |
+
"step": 3425
|
| 987 |
+
},
|
| 988 |
+
{
|
| 989 |
+
"epoch": 2.4572649572649574,
|
| 990 |
+
"grad_norm": 11.539738655090332,
|
| 991 |
+
"learning_rate": 5.4273504273504275e-06,
|
| 992 |
+
"loss": 0.1325,
|
| 993 |
+
"step": 3450
|
| 994 |
+
},
|
| 995 |
+
{
|
| 996 |
+
"epoch": 2.475071225071225,
|
| 997 |
+
"grad_norm": 5.857271671295166,
|
| 998 |
+
"learning_rate": 5.24928774928775e-06,
|
| 999 |
+
"loss": 0.1554,
|
| 1000 |
+
"step": 3475
|
| 1001 |
+
},
|
| 1002 |
+
{
|
| 1003 |
+
"epoch": 2.492877492877493,
|
| 1004 |
+
"grad_norm": 10.573086738586426,
|
| 1005 |
+
"learning_rate": 5.071225071225071e-06,
|
| 1006 |
+
"loss": 0.1348,
|
| 1007 |
+
"step": 3500
|
| 1008 |
+
},
|
| 1009 |
+
{
|
| 1010 |
+
"epoch": 2.5106837606837606,
|
| 1011 |
+
"grad_norm": 2.739942789077759,
|
| 1012 |
+
"learning_rate": 4.8931623931623934e-06,
|
| 1013 |
+
"loss": 0.1418,
|
| 1014 |
+
"step": 3525
|
| 1015 |
+
},
|
| 1016 |
+
{
|
| 1017 |
+
"epoch": 2.5284900284900287,
|
| 1018 |
+
"grad_norm": 6.804058074951172,
|
| 1019 |
+
"learning_rate": 4.715099715099715e-06,
|
| 1020 |
+
"loss": 0.1305,
|
| 1021 |
+
"step": 3550
|
| 1022 |
+
},
|
| 1023 |
+
{
|
| 1024 |
+
"epoch": 2.5462962962962963,
|
| 1025 |
+
"grad_norm": 5.385390758514404,
|
| 1026 |
+
"learning_rate": 4.537037037037038e-06,
|
| 1027 |
+
"loss": 0.1409,
|
| 1028 |
+
"step": 3575
|
| 1029 |
+
},
|
| 1030 |
+
{
|
| 1031 |
+
"epoch": 2.564102564102564,
|
| 1032 |
+
"grad_norm": 7.145414352416992,
|
| 1033 |
+
"learning_rate": 4.358974358974359e-06,
|
| 1034 |
+
"loss": 0.1509,
|
| 1035 |
+
"step": 3600
|
| 1036 |
+
},
|
| 1037 |
+
{
|
| 1038 |
+
"epoch": 2.581908831908832,
|
| 1039 |
+
"grad_norm": 7.240750789642334,
|
| 1040 |
+
"learning_rate": 4.18091168091168e-06,
|
| 1041 |
+
"loss": 0.1517,
|
| 1042 |
+
"step": 3625
|
| 1043 |
+
},
|
| 1044 |
+
{
|
| 1045 |
+
"epoch": 2.5997150997151,
|
| 1046 |
+
"grad_norm": 8.768129348754883,
|
| 1047 |
+
"learning_rate": 4.002849002849003e-06,
|
| 1048 |
+
"loss": 0.1307,
|
| 1049 |
+
"step": 3650
|
| 1050 |
+
},
|
| 1051 |
+
{
|
| 1052 |
+
"epoch": 2.6175213675213675,
|
| 1053 |
+
"grad_norm": 6.390810966491699,
|
| 1054 |
+
"learning_rate": 3.8247863247863246e-06,
|
| 1055 |
+
"loss": 0.1267,
|
| 1056 |
+
"step": 3675
|
| 1057 |
+
},
|
| 1058 |
+
{
|
| 1059 |
+
"epoch": 2.635327635327635,
|
| 1060 |
+
"grad_norm": 3.8830344676971436,
|
| 1061 |
+
"learning_rate": 3.646723646723647e-06,
|
| 1062 |
+
"loss": 0.1589,
|
| 1063 |
+
"step": 3700
|
| 1064 |
+
},
|
| 1065 |
+
{
|
| 1066 |
+
"epoch": 2.653133903133903,
|
| 1067 |
+
"grad_norm": 2.9310855865478516,
|
| 1068 |
+
"learning_rate": 3.468660968660969e-06,
|
| 1069 |
+
"loss": 0.1285,
|
| 1070 |
+
"step": 3725
|
| 1071 |
+
},
|
| 1072 |
+
{
|
| 1073 |
+
"epoch": 2.6709401709401708,
|
| 1074 |
+
"grad_norm": 4.946368217468262,
|
| 1075 |
+
"learning_rate": 3.290598290598291e-06,
|
| 1076 |
+
"loss": 0.1091,
|
| 1077 |
+
"step": 3750
|
| 1078 |
+
},
|
| 1079 |
+
{
|
| 1080 |
+
"epoch": 2.688746438746439,
|
| 1081 |
+
"grad_norm": 6.1895599365234375,
|
| 1082 |
+
"learning_rate": 3.1125356125356127e-06,
|
| 1083 |
+
"loss": 0.1516,
|
| 1084 |
+
"step": 3775
|
| 1085 |
+
},
|
| 1086 |
+
{
|
| 1087 |
+
"epoch": 2.7065527065527064,
|
| 1088 |
+
"grad_norm": 6.570561408996582,
|
| 1089 |
+
"learning_rate": 2.9344729344729344e-06,
|
| 1090 |
+
"loss": 0.1583,
|
| 1091 |
+
"step": 3800
|
| 1092 |
+
},
|
| 1093 |
+
{
|
| 1094 |
+
"epoch": 2.7243589743589745,
|
| 1095 |
+
"grad_norm": 3.769763946533203,
|
| 1096 |
+
"learning_rate": 2.7564102564102565e-06,
|
| 1097 |
+
"loss": 0.1331,
|
| 1098 |
+
"step": 3825
|
| 1099 |
+
},
|
| 1100 |
+
{
|
| 1101 |
+
"epoch": 2.742165242165242,
|
| 1102 |
+
"grad_norm": 6.275724411010742,
|
| 1103 |
+
"learning_rate": 2.5783475783475782e-06,
|
| 1104 |
+
"loss": 0.1296,
|
| 1105 |
+
"step": 3850
|
| 1106 |
+
},
|
| 1107 |
+
{
|
| 1108 |
+
"epoch": 2.75997150997151,
|
| 1109 |
+
"grad_norm": 5.346459865570068,
|
| 1110 |
+
"learning_rate": 2.4002849002849004e-06,
|
| 1111 |
+
"loss": 0.1351,
|
| 1112 |
+
"step": 3875
|
| 1113 |
+
},
|
| 1114 |
+
{
|
| 1115 |
+
"epoch": 2.7777777777777777,
|
| 1116 |
+
"grad_norm": 3.7127060890197754,
|
| 1117 |
+
"learning_rate": 2.222222222222222e-06,
|
| 1118 |
+
"loss": 0.1393,
|
| 1119 |
+
"step": 3900
|
| 1120 |
+
},
|
| 1121 |
+
{
|
| 1122 |
+
"epoch": 2.7955840455840457,
|
| 1123 |
+
"grad_norm": 12.39676570892334,
|
| 1124 |
+
"learning_rate": 2.044159544159544e-06,
|
| 1125 |
+
"loss": 0.1497,
|
| 1126 |
+
"step": 3925
|
| 1127 |
+
},
|
| 1128 |
+
{
|
| 1129 |
+
"epoch": 2.8133903133903133,
|
| 1130 |
+
"grad_norm": 13.283479690551758,
|
| 1131 |
+
"learning_rate": 1.8660968660968661e-06,
|
| 1132 |
+
"loss": 0.1282,
|
| 1133 |
+
"step": 3950
|
| 1134 |
+
},
|
| 1135 |
+
{
|
| 1136 |
+
"epoch": 2.8311965811965814,
|
| 1137 |
+
"grad_norm": 6.657817840576172,
|
| 1138 |
+
"learning_rate": 1.688034188034188e-06,
|
| 1139 |
+
"loss": 0.1427,
|
| 1140 |
+
"step": 3975
|
| 1141 |
+
},
|
| 1142 |
+
{
|
| 1143 |
+
"epoch": 2.849002849002849,
|
| 1144 |
+
"grad_norm": 9.8141450881958,
|
| 1145 |
+
"learning_rate": 1.50997150997151e-06,
|
| 1146 |
+
"loss": 0.1648,
|
| 1147 |
+
"step": 4000
|
| 1148 |
+
},
|
| 1149 |
+
{
|
| 1150 |
+
"epoch": 2.866809116809117,
|
| 1151 |
+
"grad_norm": 5.649667263031006,
|
| 1152 |
+
"learning_rate": 1.3319088319088321e-06,
|
| 1153 |
+
"loss": 0.1143,
|
| 1154 |
+
"step": 4025
|
| 1155 |
+
},
|
| 1156 |
+
{
|
| 1157 |
+
"epoch": 2.8846153846153846,
|
| 1158 |
+
"grad_norm": 8.374783515930176,
|
| 1159 |
+
"learning_rate": 1.153846153846154e-06,
|
| 1160 |
+
"loss": 0.1359,
|
| 1161 |
+
"step": 4050
|
| 1162 |
+
},
|
| 1163 |
+
{
|
| 1164 |
+
"epoch": 2.902421652421652,
|
| 1165 |
+
"grad_norm": 7.388200283050537,
|
| 1166 |
+
"learning_rate": 9.757834757834757e-07,
|
| 1167 |
+
"loss": 0.1559,
|
| 1168 |
+
"step": 4075
|
| 1169 |
+
},
|
| 1170 |
+
{
|
| 1171 |
+
"epoch": 2.92022792022792,
|
| 1172 |
+
"grad_norm": 4.6469855308532715,
|
| 1173 |
+
"learning_rate": 7.977207977207978e-07,
|
| 1174 |
+
"loss": 0.1333,
|
| 1175 |
+
"step": 4100
|
| 1176 |
+
},
|
| 1177 |
+
{
|
| 1178 |
+
"epoch": 2.9380341880341883,
|
| 1179 |
+
"grad_norm": 10.920737266540527,
|
| 1180 |
+
"learning_rate": 6.196581196581196e-07,
|
| 1181 |
+
"loss": 0.1501,
|
| 1182 |
+
"step": 4125
|
| 1183 |
+
},
|
| 1184 |
+
{
|
| 1185 |
+
"epoch": 2.955840455840456,
|
| 1186 |
+
"grad_norm": 7.988212585449219,
|
| 1187 |
+
"learning_rate": 4.415954415954416e-07,
|
| 1188 |
+
"loss": 0.1503,
|
| 1189 |
+
"step": 4150
|
| 1190 |
+
},
|
| 1191 |
+
{
|
| 1192 |
+
"epoch": 2.9736467236467234,
|
| 1193 |
+
"grad_norm": 12.452937126159668,
|
| 1194 |
+
"learning_rate": 2.635327635327635e-07,
|
| 1195 |
+
"loss": 0.1267,
|
| 1196 |
+
"step": 4175
|
| 1197 |
+
},
|
| 1198 |
+
{
|
| 1199 |
+
"epoch": 2.9914529914529915,
|
| 1200 |
+
"grad_norm": 5.0519232749938965,
|
| 1201 |
+
"learning_rate": 8.547008547008547e-08,
|
| 1202 |
+
"loss": 0.1321,
|
| 1203 |
+
"step": 4200
|
| 1204 |
+
},
|
| 1205 |
+
{
|
| 1206 |
+
"epoch": 3.0,
|
| 1207 |
+
"eval_accuracy": 0.8876146788990825,
|
| 1208 |
+
"eval_loss": 0.41072866320610046,
|
| 1209 |
+
"eval_runtime": 2.1519,
|
| 1210 |
+
"eval_samples_per_second": 405.223,
|
| 1211 |
+
"eval_steps_per_second": 4.647,
|
| 1212 |
+
"step": 4212
|
| 1213 |
+
},
|
| 1214 |
+
{
|
| 1215 |
+
"epoch": 3.0,
|
| 1216 |
+
"step": 4212,
|
| 1217 |
+
"total_flos": 6385930423743024.0,
|
| 1218 |
+
"train_loss": 0.2325162212280353,
|
| 1219 |
+
"train_runtime": 539.9254,
|
| 1220 |
+
"train_samples_per_second": 374.213,
|
| 1221 |
+
"train_steps_per_second": 7.801
|
| 1222 |
+
}
|
| 1223 |
+
],
|
| 1224 |
+
"logging_steps": 25,
|
| 1225 |
+
"max_steps": 4212,
|
| 1226 |
+
"num_input_tokens_seen": 0,
|
| 1227 |
+
"num_train_epochs": 3,
|
| 1228 |
+
"save_steps": 500,
|
| 1229 |
+
"stateful_callbacks": {
|
| 1230 |
+
"TrainerControl": {
|
| 1231 |
+
"args": {
|
| 1232 |
+
"should_epoch_stop": false,
|
| 1233 |
+
"should_evaluate": false,
|
| 1234 |
+
"should_log": false,
|
| 1235 |
+
"should_save": false,
|
| 1236 |
+
"should_training_stop": false
|
| 1237 |
+
},
|
| 1238 |
+
"attributes": {}
|
| 1239 |
+
}
|
| 1240 |
+
},
|
| 1241 |
+
"total_flos": 6385930423743024.0,
|
| 1242 |
+
"train_batch_size": 48,
|
| 1243 |
+
"trial_name": null,
|
| 1244 |
+
"trial_params": null
|
| 1245 |
+
}
|
mbert_rtx6000_metrics/result_ablation_mbert/sst2/mBERT/hf_sequence_classifier/seed_42/all_results.json
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"eval_accuracy": 0.8830275229357798,
|
| 4 |
+
"eval_loss": 0.40195930004119873,
|
| 5 |
+
"eval_runtime": 3.1242,
|
| 6 |
+
"eval_samples": 872,
|
| 7 |
+
"eval_samples_per_second": 279.108,
|
| 8 |
+
"eval_steps_per_second": 3.201,
|
| 9 |
+
"total_flos": 6167229095991660.0,
|
| 10 |
+
"train_loss": 0.2333266153747653,
|
| 11 |
+
"train_runtime": 605.5978,
|
| 12 |
+
"train_samples": 67349,
|
| 13 |
+
"train_samples_per_second": 333.632,
|
| 14 |
+
"train_steps_per_second": 6.955
|
| 15 |
+
}
|
mbert_rtx6000_metrics/result_ablation_mbert/sst2/mBERT/hf_sequence_classifier/seed_42/eval_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"eval_accuracy": 0.8830275229357798,
|
| 4 |
+
"eval_loss": 0.40195930004119873,
|
| 5 |
+
"eval_runtime": 3.1242,
|
| 6 |
+
"eval_samples": 872,
|
| 7 |
+
"eval_samples_per_second": 279.108,
|
| 8 |
+
"eval_steps_per_second": 3.201
|
| 9 |
+
}
|
mbert_rtx6000_metrics/result_ablation_mbert/sst2/mBERT/hf_sequence_classifier/seed_42/run_command.txt
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
/usr/local/bin/python /workspace/run_glue_MHA_gated.py --model_name_or_path bert-base-multilingual-cased --dataset_name tmnam20/ViGLUE --task_name sst2 --max_seq_length 128 --do_train --do_eval --evaluation_strategy epoch --save_strategy no --logging_strategy steps --logging_steps 25 --report_to none --num_train_epochs 3 --learning_rate 3e-05 --per_device_train_batch_size 48 --per_device_eval_batch_size 96 --gradient_accumulation_steps 1 --dataloader_num_workers 1 --seed 42 --output_dir result_ablation_mbert/sst2/mBERT/hf_sequence_classifier/seed_42 --overwrite_output_dir --trust_remote_code true --bf16 --tf32 true --skip_model_save
|
mbert_rtx6000_metrics/result_ablation_mbert/sst2/mBERT/hf_sequence_classifier/seed_42/train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"total_flos": 6167229095991660.0,
|
| 4 |
+
"train_loss": 0.2333266153747653,
|
| 5 |
+
"train_runtime": 605.5978,
|
| 6 |
+
"train_samples": 67349,
|
| 7 |
+
"train_samples_per_second": 333.632,
|
| 8 |
+
"train_steps_per_second": 6.955
|
| 9 |
+
}
|
mbert_rtx6000_metrics/result_ablation_mbert/sst2/mBERT/hf_sequence_classifier/seed_42/trainer_state.json
ADDED
|
@@ -0,0 +1,1245 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 3.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 4212,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.017806267806267807,
|
| 13 |
+
"grad_norm": 10.217129707336426,
|
| 14 |
+
"learning_rate": 2.9821937321937322e-05,
|
| 15 |
+
"loss": 0.644,
|
| 16 |
+
"step": 25
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.03561253561253561,
|
| 20 |
+
"grad_norm": 4.5217413902282715,
|
| 21 |
+
"learning_rate": 2.9643874643874644e-05,
|
| 22 |
+
"loss": 0.611,
|
| 23 |
+
"step": 50
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.053418803418803416,
|
| 27 |
+
"grad_norm": 5.5790863037109375,
|
| 28 |
+
"learning_rate": 2.9465811965811966e-05,
|
| 29 |
+
"loss": 0.5215,
|
| 30 |
+
"step": 75
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.07122507122507123,
|
| 34 |
+
"grad_norm": 6.627062797546387,
|
| 35 |
+
"learning_rate": 2.9287749287749288e-05,
|
| 36 |
+
"loss": 0.4722,
|
| 37 |
+
"step": 100
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.08903133903133903,
|
| 41 |
+
"grad_norm": 4.907867908477783,
|
| 42 |
+
"learning_rate": 2.910968660968661e-05,
|
| 43 |
+
"loss": 0.4753,
|
| 44 |
+
"step": 125
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.10683760683760683,
|
| 48 |
+
"grad_norm": 8.098604202270508,
|
| 49 |
+
"learning_rate": 2.893162393162393e-05,
|
| 50 |
+
"loss": 0.4478,
|
| 51 |
+
"step": 150
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.12464387464387465,
|
| 55 |
+
"grad_norm": 4.423886775970459,
|
| 56 |
+
"learning_rate": 2.8753561253561256e-05,
|
| 57 |
+
"loss": 0.444,
|
| 58 |
+
"step": 175
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.14245014245014245,
|
| 62 |
+
"grad_norm": 5.675281047821045,
|
| 63 |
+
"learning_rate": 2.8575498575498578e-05,
|
| 64 |
+
"loss": 0.4377,
|
| 65 |
+
"step": 200
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.16025641025641027,
|
| 69 |
+
"grad_norm": 5.216890811920166,
|
| 70 |
+
"learning_rate": 2.8397435897435896e-05,
|
| 71 |
+
"loss": 0.3952,
|
| 72 |
+
"step": 225
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.17806267806267806,
|
| 76 |
+
"grad_norm": 3.771064281463623,
|
| 77 |
+
"learning_rate": 2.821937321937322e-05,
|
| 78 |
+
"loss": 0.4133,
|
| 79 |
+
"step": 250
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.19586894586894588,
|
| 83 |
+
"grad_norm": 5.064457416534424,
|
| 84 |
+
"learning_rate": 2.8041310541310543e-05,
|
| 85 |
+
"loss": 0.4238,
|
| 86 |
+
"step": 275
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.21367521367521367,
|
| 90 |
+
"grad_norm": 3.4969229698181152,
|
| 91 |
+
"learning_rate": 2.786324786324786e-05,
|
| 92 |
+
"loss": 0.4117,
|
| 93 |
+
"step": 300
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.23148148148148148,
|
| 97 |
+
"grad_norm": 6.545503616333008,
|
| 98 |
+
"learning_rate": 2.7685185185185186e-05,
|
| 99 |
+
"loss": 0.3682,
|
| 100 |
+
"step": 325
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.2492877492877493,
|
| 104 |
+
"grad_norm": 4.645277976989746,
|
| 105 |
+
"learning_rate": 2.7507122507122508e-05,
|
| 106 |
+
"loss": 0.3718,
|
| 107 |
+
"step": 350
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.2670940170940171,
|
| 111 |
+
"grad_norm": 6.29575252532959,
|
| 112 |
+
"learning_rate": 2.732905982905983e-05,
|
| 113 |
+
"loss": 0.3886,
|
| 114 |
+
"step": 375
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.2849002849002849,
|
| 118 |
+
"grad_norm": 8.509537696838379,
|
| 119 |
+
"learning_rate": 2.715099715099715e-05,
|
| 120 |
+
"loss": 0.356,
|
| 121 |
+
"step": 400
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.3027065527065527,
|
| 125 |
+
"grad_norm": 4.903208255767822,
|
| 126 |
+
"learning_rate": 2.6972934472934473e-05,
|
| 127 |
+
"loss": 0.3594,
|
| 128 |
+
"step": 425
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.32051282051282054,
|
| 132 |
+
"grad_norm": 5.900155544281006,
|
| 133 |
+
"learning_rate": 2.67948717948718e-05,
|
| 134 |
+
"loss": 0.3986,
|
| 135 |
+
"step": 450
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.3383190883190883,
|
| 139 |
+
"grad_norm": 6.108882904052734,
|
| 140 |
+
"learning_rate": 2.6616809116809117e-05,
|
| 141 |
+
"loss": 0.3419,
|
| 142 |
+
"step": 475
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.3561253561253561,
|
| 146 |
+
"grad_norm": 3.5547218322753906,
|
| 147 |
+
"learning_rate": 2.643874643874644e-05,
|
| 148 |
+
"loss": 0.346,
|
| 149 |
+
"step": 500
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.37393162393162394,
|
| 153 |
+
"grad_norm": 5.8155717849731445,
|
| 154 |
+
"learning_rate": 2.6260683760683763e-05,
|
| 155 |
+
"loss": 0.3318,
|
| 156 |
+
"step": 525
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.39173789173789175,
|
| 160 |
+
"grad_norm": 4.526708126068115,
|
| 161 |
+
"learning_rate": 2.6082621082621082e-05,
|
| 162 |
+
"loss": 0.3663,
|
| 163 |
+
"step": 550
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.40954415954415957,
|
| 167 |
+
"grad_norm": 7.255255222320557,
|
| 168 |
+
"learning_rate": 2.5904558404558404e-05,
|
| 169 |
+
"loss": 0.3006,
|
| 170 |
+
"step": 575
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.42735042735042733,
|
| 174 |
+
"grad_norm": 6.101422309875488,
|
| 175 |
+
"learning_rate": 2.572649572649573e-05,
|
| 176 |
+
"loss": 0.3392,
|
| 177 |
+
"step": 600
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.44515669515669515,
|
| 181 |
+
"grad_norm": 6.404395580291748,
|
| 182 |
+
"learning_rate": 2.5548433048433047e-05,
|
| 183 |
+
"loss": 0.3527,
|
| 184 |
+
"step": 625
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.46296296296296297,
|
| 188 |
+
"grad_norm": 8.175993919372559,
|
| 189 |
+
"learning_rate": 2.5370370370370372e-05,
|
| 190 |
+
"loss": 0.3235,
|
| 191 |
+
"step": 650
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.4807692307692308,
|
| 195 |
+
"grad_norm": 3.758146286010742,
|
| 196 |
+
"learning_rate": 2.5192307692307694e-05,
|
| 197 |
+
"loss": 0.3632,
|
| 198 |
+
"step": 675
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.4985754985754986,
|
| 202 |
+
"grad_norm": 4.278959274291992,
|
| 203 |
+
"learning_rate": 2.5014245014245015e-05,
|
| 204 |
+
"loss": 0.3186,
|
| 205 |
+
"step": 700
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.5163817663817664,
|
| 209 |
+
"grad_norm": 6.148135185241699,
|
| 210 |
+
"learning_rate": 2.4836182336182337e-05,
|
| 211 |
+
"loss": 0.3218,
|
| 212 |
+
"step": 725
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.5341880341880342,
|
| 216 |
+
"grad_norm": 4.829958438873291,
|
| 217 |
+
"learning_rate": 2.465811965811966e-05,
|
| 218 |
+
"loss": 0.3111,
|
| 219 |
+
"step": 750
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.551994301994302,
|
| 223 |
+
"grad_norm": 6.673810005187988,
|
| 224 |
+
"learning_rate": 2.448005698005698e-05,
|
| 225 |
+
"loss": 0.3306,
|
| 226 |
+
"step": 775
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.5698005698005698,
|
| 230 |
+
"grad_norm": 5.420592784881592,
|
| 231 |
+
"learning_rate": 2.4301994301994302e-05,
|
| 232 |
+
"loss": 0.2815,
|
| 233 |
+
"step": 800
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.5876068376068376,
|
| 237 |
+
"grad_norm": 3.7341368198394775,
|
| 238 |
+
"learning_rate": 2.4123931623931624e-05,
|
| 239 |
+
"loss": 0.3361,
|
| 240 |
+
"step": 825
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.6054131054131054,
|
| 244 |
+
"grad_norm": 5.665746688842773,
|
| 245 |
+
"learning_rate": 2.3945868945868946e-05,
|
| 246 |
+
"loss": 0.3125,
|
| 247 |
+
"step": 850
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.6232193732193733,
|
| 251 |
+
"grad_norm": 6.519386291503906,
|
| 252 |
+
"learning_rate": 2.3767806267806267e-05,
|
| 253 |
+
"loss": 0.3007,
|
| 254 |
+
"step": 875
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.6410256410256411,
|
| 258 |
+
"grad_norm": 3.68295955657959,
|
| 259 |
+
"learning_rate": 2.358974358974359e-05,
|
| 260 |
+
"loss": 0.3465,
|
| 261 |
+
"step": 900
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.6588319088319088,
|
| 265 |
+
"grad_norm": 6.412512302398682,
|
| 266 |
+
"learning_rate": 2.3411680911680914e-05,
|
| 267 |
+
"loss": 0.2972,
|
| 268 |
+
"step": 925
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.6766381766381766,
|
| 272 |
+
"grad_norm": 3.934800386428833,
|
| 273 |
+
"learning_rate": 2.3233618233618236e-05,
|
| 274 |
+
"loss": 0.302,
|
| 275 |
+
"step": 950
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.6944444444444444,
|
| 279 |
+
"grad_norm": 7.739034175872803,
|
| 280 |
+
"learning_rate": 2.3055555555555554e-05,
|
| 281 |
+
"loss": 0.3025,
|
| 282 |
+
"step": 975
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.7122507122507122,
|
| 286 |
+
"grad_norm": 4.114175796508789,
|
| 287 |
+
"learning_rate": 2.287749287749288e-05,
|
| 288 |
+
"loss": 0.3327,
|
| 289 |
+
"step": 1000
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.73005698005698,
|
| 293 |
+
"grad_norm": 5.497580051422119,
|
| 294 |
+
"learning_rate": 2.26994301994302e-05,
|
| 295 |
+
"loss": 0.28,
|
| 296 |
+
"step": 1025
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.7478632478632479,
|
| 300 |
+
"grad_norm": 3.9182474613189697,
|
| 301 |
+
"learning_rate": 2.252136752136752e-05,
|
| 302 |
+
"loss": 0.3172,
|
| 303 |
+
"step": 1050
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.7656695156695157,
|
| 307 |
+
"grad_norm": 4.321267604827881,
|
| 308 |
+
"learning_rate": 2.2343304843304845e-05,
|
| 309 |
+
"loss": 0.2974,
|
| 310 |
+
"step": 1075
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.7834757834757835,
|
| 314 |
+
"grad_norm": 6.070116996765137,
|
| 315 |
+
"learning_rate": 2.2165242165242166e-05,
|
| 316 |
+
"loss": 0.2883,
|
| 317 |
+
"step": 1100
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 0.8012820512820513,
|
| 321 |
+
"grad_norm": 5.74845552444458,
|
| 322 |
+
"learning_rate": 2.1987179487179485e-05,
|
| 323 |
+
"loss": 0.2913,
|
| 324 |
+
"step": 1125
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.8190883190883191,
|
| 328 |
+
"grad_norm": 4.506502628326416,
|
| 329 |
+
"learning_rate": 2.180911680911681e-05,
|
| 330 |
+
"loss": 0.3104,
|
| 331 |
+
"step": 1150
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 0.8368945868945868,
|
| 335 |
+
"grad_norm": 3.877175807952881,
|
| 336 |
+
"learning_rate": 2.163105413105413e-05,
|
| 337 |
+
"loss": 0.2946,
|
| 338 |
+
"step": 1175
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.8547008547008547,
|
| 342 |
+
"grad_norm": 6.609568119049072,
|
| 343 |
+
"learning_rate": 2.1452991452991456e-05,
|
| 344 |
+
"loss": 0.2749,
|
| 345 |
+
"step": 1200
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.8725071225071225,
|
| 349 |
+
"grad_norm": 9.450400352478027,
|
| 350 |
+
"learning_rate": 2.1274928774928775e-05,
|
| 351 |
+
"loss": 0.2698,
|
| 352 |
+
"step": 1225
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 0.8903133903133903,
|
| 356 |
+
"grad_norm": 7.015203952789307,
|
| 357 |
+
"learning_rate": 2.1096866096866097e-05,
|
| 358 |
+
"loss": 0.3181,
|
| 359 |
+
"step": 1250
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 0.9081196581196581,
|
| 363 |
+
"grad_norm": 7.647506237030029,
|
| 364 |
+
"learning_rate": 2.091880341880342e-05,
|
| 365 |
+
"loss": 0.2925,
|
| 366 |
+
"step": 1275
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 0.9259259259259259,
|
| 370 |
+
"grad_norm": 5.632214069366455,
|
| 371 |
+
"learning_rate": 2.074074074074074e-05,
|
| 372 |
+
"loss": 0.2745,
|
| 373 |
+
"step": 1300
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 0.9437321937321937,
|
| 377 |
+
"grad_norm": 7.389995574951172,
|
| 378 |
+
"learning_rate": 2.056267806267806e-05,
|
| 379 |
+
"loss": 0.2534,
|
| 380 |
+
"step": 1325
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 0.9615384615384616,
|
| 384 |
+
"grad_norm": 4.803586483001709,
|
| 385 |
+
"learning_rate": 2.0384615384615387e-05,
|
| 386 |
+
"loss": 0.2768,
|
| 387 |
+
"step": 1350
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 0.9793447293447294,
|
| 391 |
+
"grad_norm": 4.420727252960205,
|
| 392 |
+
"learning_rate": 2.0206552706552705e-05,
|
| 393 |
+
"loss": 0.3008,
|
| 394 |
+
"step": 1375
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 0.9971509971509972,
|
| 398 |
+
"grad_norm": 3.7695438861846924,
|
| 399 |
+
"learning_rate": 2.002849002849003e-05,
|
| 400 |
+
"loss": 0.2639,
|
| 401 |
+
"step": 1400
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 1.0,
|
| 405 |
+
"eval_accuracy": 0.8692660550458715,
|
| 406 |
+
"eval_loss": 0.3410484492778778,
|
| 407 |
+
"eval_runtime": 3.3117,
|
| 408 |
+
"eval_samples_per_second": 263.305,
|
| 409 |
+
"eval_steps_per_second": 3.02,
|
| 410 |
+
"step": 1404
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 1.014957264957265,
|
| 414 |
+
"grad_norm": 4.10917329788208,
|
| 415 |
+
"learning_rate": 1.9850427350427352e-05,
|
| 416 |
+
"loss": 0.2324,
|
| 417 |
+
"step": 1425
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"epoch": 1.0327635327635327,
|
| 421 |
+
"grad_norm": 6.436245441436768,
|
| 422 |
+
"learning_rate": 1.9672364672364674e-05,
|
| 423 |
+
"loss": 0.2451,
|
| 424 |
+
"step": 1450
|
| 425 |
+
},
|
| 426 |
+
{
|
| 427 |
+
"epoch": 1.0505698005698005,
|
| 428 |
+
"grad_norm": 4.953073978424072,
|
| 429 |
+
"learning_rate": 1.9494301994301995e-05,
|
| 430 |
+
"loss": 0.2425,
|
| 431 |
+
"step": 1475
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 1.0683760683760684,
|
| 435 |
+
"grad_norm": 5.822015762329102,
|
| 436 |
+
"learning_rate": 1.9316239316239317e-05,
|
| 437 |
+
"loss": 0.2168,
|
| 438 |
+
"step": 1500
|
| 439 |
+
},
|
| 440 |
+
{
|
| 441 |
+
"epoch": 1.0861823361823362,
|
| 442 |
+
"grad_norm": 5.097800254821777,
|
| 443 |
+
"learning_rate": 1.913817663817664e-05,
|
| 444 |
+
"loss": 0.2334,
|
| 445 |
+
"step": 1525
|
| 446 |
+
},
|
| 447 |
+
{
|
| 448 |
+
"epoch": 1.103988603988604,
|
| 449 |
+
"grad_norm": 3.3635776042938232,
|
| 450 |
+
"learning_rate": 1.896011396011396e-05,
|
| 451 |
+
"loss": 0.2114,
|
| 452 |
+
"step": 1550
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 1.1217948717948718,
|
| 456 |
+
"grad_norm": 5.911916732788086,
|
| 457 |
+
"learning_rate": 1.8782051282051282e-05,
|
| 458 |
+
"loss": 0.2319,
|
| 459 |
+
"step": 1575
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 1.1396011396011396,
|
| 463 |
+
"grad_norm": 6.3679375648498535,
|
| 464 |
+
"learning_rate": 1.8603988603988604e-05,
|
| 465 |
+
"loss": 0.2175,
|
| 466 |
+
"step": 1600
|
| 467 |
+
},
|
| 468 |
+
{
|
| 469 |
+
"epoch": 1.1574074074074074,
|
| 470 |
+
"grad_norm": 9.391707420349121,
|
| 471 |
+
"learning_rate": 1.8425925925925926e-05,
|
| 472 |
+
"loss": 0.2137,
|
| 473 |
+
"step": 1625
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 1.1752136752136753,
|
| 477 |
+
"grad_norm": 5.359384059906006,
|
| 478 |
+
"learning_rate": 1.8247863247863247e-05,
|
| 479 |
+
"loss": 0.2092,
|
| 480 |
+
"step": 1650
|
| 481 |
+
},
|
| 482 |
+
{
|
| 483 |
+
"epoch": 1.193019943019943,
|
| 484 |
+
"grad_norm": 4.22092342376709,
|
| 485 |
+
"learning_rate": 1.8069800569800572e-05,
|
| 486 |
+
"loss": 0.2335,
|
| 487 |
+
"step": 1675
|
| 488 |
+
},
|
| 489 |
+
{
|
| 490 |
+
"epoch": 1.210826210826211,
|
| 491 |
+
"grad_norm": 7.763608455657959,
|
| 492 |
+
"learning_rate": 1.7891737891737894e-05,
|
| 493 |
+
"loss": 0.2282,
|
| 494 |
+
"step": 1700
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 1.2286324786324787,
|
| 498 |
+
"grad_norm": 3.6250720024108887,
|
| 499 |
+
"learning_rate": 1.7713675213675212e-05,
|
| 500 |
+
"loss": 0.2264,
|
| 501 |
+
"step": 1725
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 1.2464387464387465,
|
| 505 |
+
"grad_norm": 5.095157623291016,
|
| 506 |
+
"learning_rate": 1.7535612535612538e-05,
|
| 507 |
+
"loss": 0.234,
|
| 508 |
+
"step": 1750
|
| 509 |
+
},
|
| 510 |
+
{
|
| 511 |
+
"epoch": 1.2642450142450143,
|
| 512 |
+
"grad_norm": 5.401062488555908,
|
| 513 |
+
"learning_rate": 1.735754985754986e-05,
|
| 514 |
+
"loss": 0.2161,
|
| 515 |
+
"step": 1775
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 1.282051282051282,
|
| 519 |
+
"grad_norm": 6.845966339111328,
|
| 520 |
+
"learning_rate": 1.7179487179487178e-05,
|
| 521 |
+
"loss": 0.218,
|
| 522 |
+
"step": 1800
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 1.29985754985755,
|
| 526 |
+
"grad_norm": 5.141143321990967,
|
| 527 |
+
"learning_rate": 1.7001424501424503e-05,
|
| 528 |
+
"loss": 0.2304,
|
| 529 |
+
"step": 1825
|
| 530 |
+
},
|
| 531 |
+
{
|
| 532 |
+
"epoch": 1.3176638176638176,
|
| 533 |
+
"grad_norm": 4.181275844573975,
|
| 534 |
+
"learning_rate": 1.6823361823361824e-05,
|
| 535 |
+
"loss": 0.2163,
|
| 536 |
+
"step": 1850
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 1.3354700854700854,
|
| 540 |
+
"grad_norm": 3.245776653289795,
|
| 541 |
+
"learning_rate": 1.6645299145299143e-05,
|
| 542 |
+
"loss": 0.1972,
|
| 543 |
+
"step": 1875
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 1.3532763532763532,
|
| 547 |
+
"grad_norm": 6.467990875244141,
|
| 548 |
+
"learning_rate": 1.6467236467236468e-05,
|
| 549 |
+
"loss": 0.2178,
|
| 550 |
+
"step": 1900
|
| 551 |
+
},
|
| 552 |
+
{
|
| 553 |
+
"epoch": 1.371082621082621,
|
| 554 |
+
"grad_norm": 6.5718841552734375,
|
| 555 |
+
"learning_rate": 1.628917378917379e-05,
|
| 556 |
+
"loss": 0.212,
|
| 557 |
+
"step": 1925
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"epoch": 1.3888888888888888,
|
| 561 |
+
"grad_norm": 6.254003524780273,
|
| 562 |
+
"learning_rate": 1.6111111111111115e-05,
|
| 563 |
+
"loss": 0.1943,
|
| 564 |
+
"step": 1950
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 1.4066951566951567,
|
| 568 |
+
"grad_norm": 4.2838053703308105,
|
| 569 |
+
"learning_rate": 1.5933048433048433e-05,
|
| 570 |
+
"loss": 0.1871,
|
| 571 |
+
"step": 1975
|
| 572 |
+
},
|
| 573 |
+
{
|
| 574 |
+
"epoch": 1.4245014245014245,
|
| 575 |
+
"grad_norm": 4.545979022979736,
|
| 576 |
+
"learning_rate": 1.5754985754985755e-05,
|
| 577 |
+
"loss": 0.2047,
|
| 578 |
+
"step": 2000
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 1.4423076923076923,
|
| 582 |
+
"grad_norm": 4.526988983154297,
|
| 583 |
+
"learning_rate": 1.557692307692308e-05,
|
| 584 |
+
"loss": 0.1953,
|
| 585 |
+
"step": 2025
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 1.46011396011396,
|
| 589 |
+
"grad_norm": 5.631552219390869,
|
| 590 |
+
"learning_rate": 1.5398860398860398e-05,
|
| 591 |
+
"loss": 0.2126,
|
| 592 |
+
"step": 2050
|
| 593 |
+
},
|
| 594 |
+
{
|
| 595 |
+
"epoch": 1.477920227920228,
|
| 596 |
+
"grad_norm": 9.568936347961426,
|
| 597 |
+
"learning_rate": 1.5220797720797721e-05,
|
| 598 |
+
"loss": 0.2136,
|
| 599 |
+
"step": 2075
|
| 600 |
+
},
|
| 601 |
+
{
|
| 602 |
+
"epoch": 1.4957264957264957,
|
| 603 |
+
"grad_norm": 5.368617534637451,
|
| 604 |
+
"learning_rate": 1.5042735042735045e-05,
|
| 605 |
+
"loss": 0.2138,
|
| 606 |
+
"step": 2100
|
| 607 |
+
},
|
| 608 |
+
{
|
| 609 |
+
"epoch": 1.5135327635327636,
|
| 610 |
+
"grad_norm": 5.690488815307617,
|
| 611 |
+
"learning_rate": 1.4864672364672365e-05,
|
| 612 |
+
"loss": 0.1846,
|
| 613 |
+
"step": 2125
|
| 614 |
+
},
|
| 615 |
+
{
|
| 616 |
+
"epoch": 1.5313390313390314,
|
| 617 |
+
"grad_norm": 4.5680766105651855,
|
| 618 |
+
"learning_rate": 1.4686609686609687e-05,
|
| 619 |
+
"loss": 0.1981,
|
| 620 |
+
"step": 2150
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 1.5491452991452992,
|
| 624 |
+
"grad_norm": 6.659402847290039,
|
| 625 |
+
"learning_rate": 1.450854700854701e-05,
|
| 626 |
+
"loss": 0.2254,
|
| 627 |
+
"step": 2175
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"epoch": 1.566951566951567,
|
| 631 |
+
"grad_norm": 7.409360885620117,
|
| 632 |
+
"learning_rate": 1.433048433048433e-05,
|
| 633 |
+
"loss": 0.1889,
|
| 634 |
+
"step": 2200
|
| 635 |
+
},
|
| 636 |
+
{
|
| 637 |
+
"epoch": 1.5847578347578346,
|
| 638 |
+
"grad_norm": 6.811413288116455,
|
| 639 |
+
"learning_rate": 1.4152421652421652e-05,
|
| 640 |
+
"loss": 0.1924,
|
| 641 |
+
"step": 2225
|
| 642 |
+
},
|
| 643 |
+
{
|
| 644 |
+
"epoch": 1.6025641025641026,
|
| 645 |
+
"grad_norm": 6.660957336425781,
|
| 646 |
+
"learning_rate": 1.3974358974358975e-05,
|
| 647 |
+
"loss": 0.1852,
|
| 648 |
+
"step": 2250
|
| 649 |
+
},
|
| 650 |
+
{
|
| 651 |
+
"epoch": 1.6203703703703702,
|
| 652 |
+
"grad_norm": 4.420598030090332,
|
| 653 |
+
"learning_rate": 1.3796296296296297e-05,
|
| 654 |
+
"loss": 0.2059,
|
| 655 |
+
"step": 2275
|
| 656 |
+
},
|
| 657 |
+
{
|
| 658 |
+
"epoch": 1.6381766381766383,
|
| 659 |
+
"grad_norm": 5.48427677154541,
|
| 660 |
+
"learning_rate": 1.3618233618233619e-05,
|
| 661 |
+
"loss": 0.2442,
|
| 662 |
+
"step": 2300
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 1.6559829059829059,
|
| 666 |
+
"grad_norm": 4.802734375,
|
| 667 |
+
"learning_rate": 1.344017094017094e-05,
|
| 668 |
+
"loss": 0.1997,
|
| 669 |
+
"step": 2325
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 1.673789173789174,
|
| 673 |
+
"grad_norm": 5.091231822967529,
|
| 674 |
+
"learning_rate": 1.3262108262108262e-05,
|
| 675 |
+
"loss": 0.2162,
|
| 676 |
+
"step": 2350
|
| 677 |
+
},
|
| 678 |
+
{
|
| 679 |
+
"epoch": 1.6915954415954415,
|
| 680 |
+
"grad_norm": 4.21741247177124,
|
| 681 |
+
"learning_rate": 1.3084045584045585e-05,
|
| 682 |
+
"loss": 0.1746,
|
| 683 |
+
"step": 2375
|
| 684 |
+
},
|
| 685 |
+
{
|
| 686 |
+
"epoch": 1.7094017094017095,
|
| 687 |
+
"grad_norm": 4.4245734214782715,
|
| 688 |
+
"learning_rate": 1.2905982905982905e-05,
|
| 689 |
+
"loss": 0.2044,
|
| 690 |
+
"step": 2400
|
| 691 |
+
},
|
| 692 |
+
{
|
| 693 |
+
"epoch": 1.7272079772079771,
|
| 694 |
+
"grad_norm": 4.303463935852051,
|
| 695 |
+
"learning_rate": 1.2727920227920229e-05,
|
| 696 |
+
"loss": 0.1981,
|
| 697 |
+
"step": 2425
|
| 698 |
+
},
|
| 699 |
+
{
|
| 700 |
+
"epoch": 1.7450142450142452,
|
| 701 |
+
"grad_norm": 5.587222099304199,
|
| 702 |
+
"learning_rate": 1.254985754985755e-05,
|
| 703 |
+
"loss": 0.2146,
|
| 704 |
+
"step": 2450
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 1.7628205128205128,
|
| 708 |
+
"grad_norm": 7.019397735595703,
|
| 709 |
+
"learning_rate": 1.2371794871794872e-05,
|
| 710 |
+
"loss": 0.1769,
|
| 711 |
+
"step": 2475
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"epoch": 1.7806267806267806,
|
| 715 |
+
"grad_norm": 8.05568790435791,
|
| 716 |
+
"learning_rate": 1.2193732193732194e-05,
|
| 717 |
+
"loss": 0.2143,
|
| 718 |
+
"step": 2500
|
| 719 |
+
},
|
| 720 |
+
{
|
| 721 |
+
"epoch": 1.7984330484330484,
|
| 722 |
+
"grad_norm": 4.52636194229126,
|
| 723 |
+
"learning_rate": 1.2015669515669516e-05,
|
| 724 |
+
"loss": 0.2195,
|
| 725 |
+
"step": 2525
|
| 726 |
+
},
|
| 727 |
+
{
|
| 728 |
+
"epoch": 1.8162393162393162,
|
| 729 |
+
"grad_norm": 6.613002777099609,
|
| 730 |
+
"learning_rate": 1.1837606837606839e-05,
|
| 731 |
+
"loss": 0.2079,
|
| 732 |
+
"step": 2550
|
| 733 |
+
},
|
| 734 |
+
{
|
| 735 |
+
"epoch": 1.834045584045584,
|
| 736 |
+
"grad_norm": 5.012784004211426,
|
| 737 |
+
"learning_rate": 1.1659544159544159e-05,
|
| 738 |
+
"loss": 0.1967,
|
| 739 |
+
"step": 2575
|
| 740 |
+
},
|
| 741 |
+
{
|
| 742 |
+
"epoch": 1.8518518518518519,
|
| 743 |
+
"grad_norm": 6.7266459465026855,
|
| 744 |
+
"learning_rate": 1.148148148148148e-05,
|
| 745 |
+
"loss": 0.2029,
|
| 746 |
+
"step": 2600
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 1.8696581196581197,
|
| 750 |
+
"grad_norm": 6.2435526847839355,
|
| 751 |
+
"learning_rate": 1.1303418803418804e-05,
|
| 752 |
+
"loss": 0.2009,
|
| 753 |
+
"step": 2625
|
| 754 |
+
},
|
| 755 |
+
{
|
| 756 |
+
"epoch": 1.8874643874643875,
|
| 757 |
+
"grad_norm": 6.99838924407959,
|
| 758 |
+
"learning_rate": 1.1125356125356126e-05,
|
| 759 |
+
"loss": 0.2178,
|
| 760 |
+
"step": 2650
|
| 761 |
+
},
|
| 762 |
+
{
|
| 763 |
+
"epoch": 1.9052706552706553,
|
| 764 |
+
"grad_norm": 7.170559883117676,
|
| 765 |
+
"learning_rate": 1.0947293447293448e-05,
|
| 766 |
+
"loss": 0.204,
|
| 767 |
+
"step": 2675
|
| 768 |
+
},
|
| 769 |
+
{
|
| 770 |
+
"epoch": 1.9230769230769231,
|
| 771 |
+
"grad_norm": 4.116168022155762,
|
| 772 |
+
"learning_rate": 1.076923076923077e-05,
|
| 773 |
+
"loss": 0.2,
|
| 774 |
+
"step": 2700
|
| 775 |
+
},
|
| 776 |
+
{
|
| 777 |
+
"epoch": 1.940883190883191,
|
| 778 |
+
"grad_norm": 4.995635509490967,
|
| 779 |
+
"learning_rate": 1.0591168091168091e-05,
|
| 780 |
+
"loss": 0.1887,
|
| 781 |
+
"step": 2725
|
| 782 |
+
},
|
| 783 |
+
{
|
| 784 |
+
"epoch": 1.9586894586894585,
|
| 785 |
+
"grad_norm": 7.485500812530518,
|
| 786 |
+
"learning_rate": 1.0413105413105414e-05,
|
| 787 |
+
"loss": 0.1699,
|
| 788 |
+
"step": 2750
|
| 789 |
+
},
|
| 790 |
+
{
|
| 791 |
+
"epoch": 1.9764957264957266,
|
| 792 |
+
"grad_norm": 4.375154972076416,
|
| 793 |
+
"learning_rate": 1.0235042735042734e-05,
|
| 794 |
+
"loss": 0.1961,
|
| 795 |
+
"step": 2775
|
| 796 |
+
},
|
| 797 |
+
{
|
| 798 |
+
"epoch": 1.9943019943019942,
|
| 799 |
+
"grad_norm": 4.151944637298584,
|
| 800 |
+
"learning_rate": 1.0056980056980058e-05,
|
| 801 |
+
"loss": 0.1891,
|
| 802 |
+
"step": 2800
|
| 803 |
+
},
|
| 804 |
+
{
|
| 805 |
+
"epoch": 2.0,
|
| 806 |
+
"eval_accuracy": 0.8795871559633027,
|
| 807 |
+
"eval_loss": 0.3776057958602905,
|
| 808 |
+
"eval_runtime": 3.3461,
|
| 809 |
+
"eval_samples_per_second": 260.602,
|
| 810 |
+
"eval_steps_per_second": 2.989,
|
| 811 |
+
"step": 2808
|
| 812 |
+
},
|
| 813 |
+
{
|
| 814 |
+
"epoch": 2.012108262108262,
|
| 815 |
+
"grad_norm": 6.2388176918029785,
|
| 816 |
+
"learning_rate": 9.87891737891738e-06,
|
| 817 |
+
"loss": 0.1453,
|
| 818 |
+
"step": 2825
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 2.02991452991453,
|
| 822 |
+
"grad_norm": 2.0722806453704834,
|
| 823 |
+
"learning_rate": 9.700854700854701e-06,
|
| 824 |
+
"loss": 0.158,
|
| 825 |
+
"step": 2850
|
| 826 |
+
},
|
| 827 |
+
{
|
| 828 |
+
"epoch": 2.047720797720798,
|
| 829 |
+
"grad_norm": 5.726858139038086,
|
| 830 |
+
"learning_rate": 9.522792022792023e-06,
|
| 831 |
+
"loss": 0.1514,
|
| 832 |
+
"step": 2875
|
| 833 |
+
},
|
| 834 |
+
{
|
| 835 |
+
"epoch": 2.0655270655270654,
|
| 836 |
+
"grad_norm": 4.155969619750977,
|
| 837 |
+
"learning_rate": 9.344729344729345e-06,
|
| 838 |
+
"loss": 0.1406,
|
| 839 |
+
"step": 2900
|
| 840 |
+
},
|
| 841 |
+
{
|
| 842 |
+
"epoch": 2.0833333333333335,
|
| 843 |
+
"grad_norm": 2.9551212787628174,
|
| 844 |
+
"learning_rate": 9.166666666666668e-06,
|
| 845 |
+
"loss": 0.1495,
|
| 846 |
+
"step": 2925
|
| 847 |
+
},
|
| 848 |
+
{
|
| 849 |
+
"epoch": 2.101139601139601,
|
| 850 |
+
"grad_norm": 5.706986904144287,
|
| 851 |
+
"learning_rate": 8.988603988603988e-06,
|
| 852 |
+
"loss": 0.1475,
|
| 853 |
+
"step": 2950
|
| 854 |
+
},
|
| 855 |
+
{
|
| 856 |
+
"epoch": 2.118945868945869,
|
| 857 |
+
"grad_norm": 4.989862442016602,
|
| 858 |
+
"learning_rate": 8.81054131054131e-06,
|
| 859 |
+
"loss": 0.1335,
|
| 860 |
+
"step": 2975
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 2.1367521367521367,
|
| 864 |
+
"grad_norm": 9.048837661743164,
|
| 865 |
+
"learning_rate": 8.632478632478633e-06,
|
| 866 |
+
"loss": 0.148,
|
| 867 |
+
"step": 3000
|
| 868 |
+
},
|
| 869 |
+
{
|
| 870 |
+
"epoch": 2.1545584045584047,
|
| 871 |
+
"grad_norm": 3.5153439044952393,
|
| 872 |
+
"learning_rate": 8.454415954415955e-06,
|
| 873 |
+
"loss": 0.142,
|
| 874 |
+
"step": 3025
|
| 875 |
+
},
|
| 876 |
+
{
|
| 877 |
+
"epoch": 2.1723646723646723,
|
| 878 |
+
"grad_norm": 6.949209213256836,
|
| 879 |
+
"learning_rate": 8.276353276353277e-06,
|
| 880 |
+
"loss": 0.1507,
|
| 881 |
+
"step": 3050
|
| 882 |
+
},
|
| 883 |
+
{
|
| 884 |
+
"epoch": 2.1901709401709404,
|
| 885 |
+
"grad_norm": 7.0229172706604,
|
| 886 |
+
"learning_rate": 8.098290598290598e-06,
|
| 887 |
+
"loss": 0.1431,
|
| 888 |
+
"step": 3075
|
| 889 |
+
},
|
| 890 |
+
{
|
| 891 |
+
"epoch": 2.207977207977208,
|
| 892 |
+
"grad_norm": 2.5189661979675293,
|
| 893 |
+
"learning_rate": 7.92022792022792e-06,
|
| 894 |
+
"loss": 0.1585,
|
| 895 |
+
"step": 3100
|
| 896 |
+
},
|
| 897 |
+
{
|
| 898 |
+
"epoch": 2.2257834757834756,
|
| 899 |
+
"grad_norm": 8.046323776245117,
|
| 900 |
+
"learning_rate": 7.742165242165244e-06,
|
| 901 |
+
"loss": 0.1379,
|
| 902 |
+
"step": 3125
|
| 903 |
+
},
|
| 904 |
+
{
|
| 905 |
+
"epoch": 2.2435897435897436,
|
| 906 |
+
"grad_norm": 4.326846599578857,
|
| 907 |
+
"learning_rate": 7.564102564102564e-06,
|
| 908 |
+
"loss": 0.1274,
|
| 909 |
+
"step": 3150
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 2.261396011396011,
|
| 913 |
+
"grad_norm": 5.842421531677246,
|
| 914 |
+
"learning_rate": 7.386039886039886e-06,
|
| 915 |
+
"loss": 0.1484,
|
| 916 |
+
"step": 3175
|
| 917 |
+
},
|
| 918 |
+
{
|
| 919 |
+
"epoch": 2.2792022792022792,
|
| 920 |
+
"grad_norm": 6.523318290710449,
|
| 921 |
+
"learning_rate": 7.207977207977209e-06,
|
| 922 |
+
"loss": 0.1528,
|
| 923 |
+
"step": 3200
|
| 924 |
+
},
|
| 925 |
+
{
|
| 926 |
+
"epoch": 2.297008547008547,
|
| 927 |
+
"grad_norm": 11.860811233520508,
|
| 928 |
+
"learning_rate": 7.02991452991453e-06,
|
| 929 |
+
"loss": 0.1183,
|
| 930 |
+
"step": 3225
|
| 931 |
+
},
|
| 932 |
+
{
|
| 933 |
+
"epoch": 2.314814814814815,
|
| 934 |
+
"grad_norm": 10.215193748474121,
|
| 935 |
+
"learning_rate": 6.851851851851852e-06,
|
| 936 |
+
"loss": 0.1497,
|
| 937 |
+
"step": 3250
|
| 938 |
+
},
|
| 939 |
+
{
|
| 940 |
+
"epoch": 2.3326210826210825,
|
| 941 |
+
"grad_norm": 10.560525894165039,
|
| 942 |
+
"learning_rate": 6.673789173789174e-06,
|
| 943 |
+
"loss": 0.166,
|
| 944 |
+
"step": 3275
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 2.3504273504273505,
|
| 948 |
+
"grad_norm": 4.075424671173096,
|
| 949 |
+
"learning_rate": 6.4957264957264955e-06,
|
| 950 |
+
"loss": 0.1297,
|
| 951 |
+
"step": 3300
|
| 952 |
+
},
|
| 953 |
+
{
|
| 954 |
+
"epoch": 2.368233618233618,
|
| 955 |
+
"grad_norm": 11.69821548461914,
|
| 956 |
+
"learning_rate": 6.317663817663818e-06,
|
| 957 |
+
"loss": 0.1341,
|
| 958 |
+
"step": 3325
|
| 959 |
+
},
|
| 960 |
+
{
|
| 961 |
+
"epoch": 2.386039886039886,
|
| 962 |
+
"grad_norm": 5.4101409912109375,
|
| 963 |
+
"learning_rate": 6.13960113960114e-06,
|
| 964 |
+
"loss": 0.1532,
|
| 965 |
+
"step": 3350
|
| 966 |
+
},
|
| 967 |
+
{
|
| 968 |
+
"epoch": 2.4038461538461537,
|
| 969 |
+
"grad_norm": 5.306498050689697,
|
| 970 |
+
"learning_rate": 5.9615384615384615e-06,
|
| 971 |
+
"loss": 0.1501,
|
| 972 |
+
"step": 3375
|
| 973 |
+
},
|
| 974 |
+
{
|
| 975 |
+
"epoch": 2.421652421652422,
|
| 976 |
+
"grad_norm": 8.793139457702637,
|
| 977 |
+
"learning_rate": 5.783475783475783e-06,
|
| 978 |
+
"loss": 0.1431,
|
| 979 |
+
"step": 3400
|
| 980 |
+
},
|
| 981 |
+
{
|
| 982 |
+
"epoch": 2.4394586894586894,
|
| 983 |
+
"grad_norm": 2.030932664871216,
|
| 984 |
+
"learning_rate": 5.605413105413106e-06,
|
| 985 |
+
"loss": 0.1552,
|
| 986 |
+
"step": 3425
|
| 987 |
+
},
|
| 988 |
+
{
|
| 989 |
+
"epoch": 2.4572649572649574,
|
| 990 |
+
"grad_norm": 8.993446350097656,
|
| 991 |
+
"learning_rate": 5.4273504273504275e-06,
|
| 992 |
+
"loss": 0.1272,
|
| 993 |
+
"step": 3450
|
| 994 |
+
},
|
| 995 |
+
{
|
| 996 |
+
"epoch": 2.475071225071225,
|
| 997 |
+
"grad_norm": 5.066373825073242,
|
| 998 |
+
"learning_rate": 5.24928774928775e-06,
|
| 999 |
+
"loss": 0.1219,
|
| 1000 |
+
"step": 3475
|
| 1001 |
+
},
|
| 1002 |
+
{
|
| 1003 |
+
"epoch": 2.492877492877493,
|
| 1004 |
+
"grad_norm": 4.286602973937988,
|
| 1005 |
+
"learning_rate": 5.071225071225071e-06,
|
| 1006 |
+
"loss": 0.1016,
|
| 1007 |
+
"step": 3500
|
| 1008 |
+
},
|
| 1009 |
+
{
|
| 1010 |
+
"epoch": 2.5106837606837606,
|
| 1011 |
+
"grad_norm": 12.636241912841797,
|
| 1012 |
+
"learning_rate": 4.8931623931623934e-06,
|
| 1013 |
+
"loss": 0.1414,
|
| 1014 |
+
"step": 3525
|
| 1015 |
+
},
|
| 1016 |
+
{
|
| 1017 |
+
"epoch": 2.5284900284900287,
|
| 1018 |
+
"grad_norm": 7.780368804931641,
|
| 1019 |
+
"learning_rate": 4.715099715099715e-06,
|
| 1020 |
+
"loss": 0.155,
|
| 1021 |
+
"step": 3550
|
| 1022 |
+
},
|
| 1023 |
+
{
|
| 1024 |
+
"epoch": 2.5462962962962963,
|
| 1025 |
+
"grad_norm": 7.307154178619385,
|
| 1026 |
+
"learning_rate": 4.537037037037038e-06,
|
| 1027 |
+
"loss": 0.1643,
|
| 1028 |
+
"step": 3575
|
| 1029 |
+
},
|
| 1030 |
+
{
|
| 1031 |
+
"epoch": 2.564102564102564,
|
| 1032 |
+
"grad_norm": 4.874617576599121,
|
| 1033 |
+
"learning_rate": 4.358974358974359e-06,
|
| 1034 |
+
"loss": 0.1407,
|
| 1035 |
+
"step": 3600
|
| 1036 |
+
},
|
| 1037 |
+
{
|
| 1038 |
+
"epoch": 2.581908831908832,
|
| 1039 |
+
"grad_norm": 3.491570472717285,
|
| 1040 |
+
"learning_rate": 4.18091168091168e-06,
|
| 1041 |
+
"loss": 0.126,
|
| 1042 |
+
"step": 3625
|
| 1043 |
+
},
|
| 1044 |
+
{
|
| 1045 |
+
"epoch": 2.5997150997151,
|
| 1046 |
+
"grad_norm": 7.297542572021484,
|
| 1047 |
+
"learning_rate": 4.002849002849003e-06,
|
| 1048 |
+
"loss": 0.1268,
|
| 1049 |
+
"step": 3650
|
| 1050 |
+
},
|
| 1051 |
+
{
|
| 1052 |
+
"epoch": 2.6175213675213675,
|
| 1053 |
+
"grad_norm": 6.328291893005371,
|
| 1054 |
+
"learning_rate": 3.8247863247863246e-06,
|
| 1055 |
+
"loss": 0.1691,
|
| 1056 |
+
"step": 3675
|
| 1057 |
+
},
|
| 1058 |
+
{
|
| 1059 |
+
"epoch": 2.635327635327635,
|
| 1060 |
+
"grad_norm": 7.41910457611084,
|
| 1061 |
+
"learning_rate": 3.646723646723647e-06,
|
| 1062 |
+
"loss": 0.1288,
|
| 1063 |
+
"step": 3700
|
| 1064 |
+
},
|
| 1065 |
+
{
|
| 1066 |
+
"epoch": 2.653133903133903,
|
| 1067 |
+
"grad_norm": 4.908723831176758,
|
| 1068 |
+
"learning_rate": 3.468660968660969e-06,
|
| 1069 |
+
"loss": 0.1566,
|
| 1070 |
+
"step": 3725
|
| 1071 |
+
},
|
| 1072 |
+
{
|
| 1073 |
+
"epoch": 2.6709401709401708,
|
| 1074 |
+
"grad_norm": 6.918929100036621,
|
| 1075 |
+
"learning_rate": 3.290598290598291e-06,
|
| 1076 |
+
"loss": 0.1252,
|
| 1077 |
+
"step": 3750
|
| 1078 |
+
},
|
| 1079 |
+
{
|
| 1080 |
+
"epoch": 2.688746438746439,
|
| 1081 |
+
"grad_norm": 1.5778158903121948,
|
| 1082 |
+
"learning_rate": 3.1125356125356127e-06,
|
| 1083 |
+
"loss": 0.1225,
|
| 1084 |
+
"step": 3775
|
| 1085 |
+
},
|
| 1086 |
+
{
|
| 1087 |
+
"epoch": 2.7065527065527064,
|
| 1088 |
+
"grad_norm": 7.621183395385742,
|
| 1089 |
+
"learning_rate": 2.9344729344729344e-06,
|
| 1090 |
+
"loss": 0.1511,
|
| 1091 |
+
"step": 3800
|
| 1092 |
+
},
|
| 1093 |
+
{
|
| 1094 |
+
"epoch": 2.7243589743589745,
|
| 1095 |
+
"grad_norm": 5.350505352020264,
|
| 1096 |
+
"learning_rate": 2.7564102564102565e-06,
|
| 1097 |
+
"loss": 0.1153,
|
| 1098 |
+
"step": 3825
|
| 1099 |
+
},
|
| 1100 |
+
{
|
| 1101 |
+
"epoch": 2.742165242165242,
|
| 1102 |
+
"grad_norm": 8.659831047058105,
|
| 1103 |
+
"learning_rate": 2.5783475783475782e-06,
|
| 1104 |
+
"loss": 0.1214,
|
| 1105 |
+
"step": 3850
|
| 1106 |
+
},
|
| 1107 |
+
{
|
| 1108 |
+
"epoch": 2.75997150997151,
|
| 1109 |
+
"grad_norm": 8.685633659362793,
|
| 1110 |
+
"learning_rate": 2.4002849002849004e-06,
|
| 1111 |
+
"loss": 0.1344,
|
| 1112 |
+
"step": 3875
|
| 1113 |
+
},
|
| 1114 |
+
{
|
| 1115 |
+
"epoch": 2.7777777777777777,
|
| 1116 |
+
"grad_norm": 7.774617671966553,
|
| 1117 |
+
"learning_rate": 2.222222222222222e-06,
|
| 1118 |
+
"loss": 0.1419,
|
| 1119 |
+
"step": 3900
|
| 1120 |
+
},
|
| 1121 |
+
{
|
| 1122 |
+
"epoch": 2.7955840455840457,
|
| 1123 |
+
"grad_norm": 8.621310234069824,
|
| 1124 |
+
"learning_rate": 2.044159544159544e-06,
|
| 1125 |
+
"loss": 0.1414,
|
| 1126 |
+
"step": 3925
|
| 1127 |
+
},
|
| 1128 |
+
{
|
| 1129 |
+
"epoch": 2.8133903133903133,
|
| 1130 |
+
"grad_norm": 3.968473434448242,
|
| 1131 |
+
"learning_rate": 1.8660968660968661e-06,
|
| 1132 |
+
"loss": 0.1401,
|
| 1133 |
+
"step": 3950
|
| 1134 |
+
},
|
| 1135 |
+
{
|
| 1136 |
+
"epoch": 2.8311965811965814,
|
| 1137 |
+
"grad_norm": 3.0339131355285645,
|
| 1138 |
+
"learning_rate": 1.688034188034188e-06,
|
| 1139 |
+
"loss": 0.1434,
|
| 1140 |
+
"step": 3975
|
| 1141 |
+
},
|
| 1142 |
+
{
|
| 1143 |
+
"epoch": 2.849002849002849,
|
| 1144 |
+
"grad_norm": 3.1119580268859863,
|
| 1145 |
+
"learning_rate": 1.50997150997151e-06,
|
| 1146 |
+
"loss": 0.1359,
|
| 1147 |
+
"step": 4000
|
| 1148 |
+
},
|
| 1149 |
+
{
|
| 1150 |
+
"epoch": 2.866809116809117,
|
| 1151 |
+
"grad_norm": 7.304025173187256,
|
| 1152 |
+
"learning_rate": 1.3319088319088321e-06,
|
| 1153 |
+
"loss": 0.1321,
|
| 1154 |
+
"step": 4025
|
| 1155 |
+
},
|
| 1156 |
+
{
|
| 1157 |
+
"epoch": 2.8846153846153846,
|
| 1158 |
+
"grad_norm": 4.910098075866699,
|
| 1159 |
+
"learning_rate": 1.153846153846154e-06,
|
| 1160 |
+
"loss": 0.1128,
|
| 1161 |
+
"step": 4050
|
| 1162 |
+
},
|
| 1163 |
+
{
|
| 1164 |
+
"epoch": 2.902421652421652,
|
| 1165 |
+
"grad_norm": 6.606889724731445,
|
| 1166 |
+
"learning_rate": 9.757834757834757e-07,
|
| 1167 |
+
"loss": 0.1395,
|
| 1168 |
+
"step": 4075
|
| 1169 |
+
},
|
| 1170 |
+
{
|
| 1171 |
+
"epoch": 2.92022792022792,
|
| 1172 |
+
"grad_norm": 6.8984479904174805,
|
| 1173 |
+
"learning_rate": 7.977207977207978e-07,
|
| 1174 |
+
"loss": 0.1254,
|
| 1175 |
+
"step": 4100
|
| 1176 |
+
},
|
| 1177 |
+
{
|
| 1178 |
+
"epoch": 2.9380341880341883,
|
| 1179 |
+
"grad_norm": 3.285421848297119,
|
| 1180 |
+
"learning_rate": 6.196581196581196e-07,
|
| 1181 |
+
"loss": 0.132,
|
| 1182 |
+
"step": 4125
|
| 1183 |
+
},
|
| 1184 |
+
{
|
| 1185 |
+
"epoch": 2.955840455840456,
|
| 1186 |
+
"grad_norm": 6.842528820037842,
|
| 1187 |
+
"learning_rate": 4.415954415954416e-07,
|
| 1188 |
+
"loss": 0.1414,
|
| 1189 |
+
"step": 4150
|
| 1190 |
+
},
|
| 1191 |
+
{
|
| 1192 |
+
"epoch": 2.9736467236467234,
|
| 1193 |
+
"grad_norm": 6.8176774978637695,
|
| 1194 |
+
"learning_rate": 2.635327635327635e-07,
|
| 1195 |
+
"loss": 0.1236,
|
| 1196 |
+
"step": 4175
|
| 1197 |
+
},
|
| 1198 |
+
{
|
| 1199 |
+
"epoch": 2.9914529914529915,
|
| 1200 |
+
"grad_norm": 7.452104091644287,
|
| 1201 |
+
"learning_rate": 8.547008547008547e-08,
|
| 1202 |
+
"loss": 0.1464,
|
| 1203 |
+
"step": 4200
|
| 1204 |
+
},
|
| 1205 |
+
{
|
| 1206 |
+
"epoch": 3.0,
|
| 1207 |
+
"eval_accuracy": 0.8830275229357798,
|
| 1208 |
+
"eval_loss": 0.40195930004119873,
|
| 1209 |
+
"eval_runtime": 3.3408,
|
| 1210 |
+
"eval_samples_per_second": 261.014,
|
| 1211 |
+
"eval_steps_per_second": 2.993,
|
| 1212 |
+
"step": 4212
|
| 1213 |
+
},
|
| 1214 |
+
{
|
| 1215 |
+
"epoch": 3.0,
|
| 1216 |
+
"step": 4212,
|
| 1217 |
+
"total_flos": 6167229095991660.0,
|
| 1218 |
+
"train_loss": 0.2333266153747653,
|
| 1219 |
+
"train_runtime": 605.5978,
|
| 1220 |
+
"train_samples_per_second": 333.632,
|
| 1221 |
+
"train_steps_per_second": 6.955
|
| 1222 |
+
}
|
| 1223 |
+
],
|
| 1224 |
+
"logging_steps": 25,
|
| 1225 |
+
"max_steps": 4212,
|
| 1226 |
+
"num_input_tokens_seen": 0,
|
| 1227 |
+
"num_train_epochs": 3,
|
| 1228 |
+
"save_steps": 500,
|
| 1229 |
+
"stateful_callbacks": {
|
| 1230 |
+
"TrainerControl": {
|
| 1231 |
+
"args": {
|
| 1232 |
+
"should_epoch_stop": false,
|
| 1233 |
+
"should_evaluate": false,
|
| 1234 |
+
"should_log": false,
|
| 1235 |
+
"should_save": false,
|
| 1236 |
+
"should_training_stop": false
|
| 1237 |
+
},
|
| 1238 |
+
"attributes": {}
|
| 1239 |
+
}
|
| 1240 |
+
},
|
| 1241 |
+
"total_flos": 6167229095991660.0,
|
| 1242 |
+
"train_batch_size": 48,
|
| 1243 |
+
"trial_name": null,
|
| 1244 |
+
"trial_params": null
|
| 1245 |
+
}
|
mbert_rtx6000_metrics/result_ablation_mbert/sst2/mBERT/hf_sequence_classifier/seed_43/all_results.json
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"eval_accuracy": 0.8864678899082569,
|
| 4 |
+
"eval_loss": 0.4080387055873871,
|
| 5 |
+
"eval_runtime": 1.8265,
|
| 6 |
+
"eval_samples": 872,
|
| 7 |
+
"eval_samples_per_second": 477.422,
|
| 8 |
+
"eval_steps_per_second": 5.475,
|
| 9 |
+
"total_flos": 6166879651621260.0,
|
| 10 |
+
"train_loss": 0.2339267143723179,
|
| 11 |
+
"train_runtime": 614.1221,
|
| 12 |
+
"train_samples": 67349,
|
| 13 |
+
"train_samples_per_second": 329.001,
|
| 14 |
+
"train_steps_per_second": 6.859
|
| 15 |
+
}
|
mbert_rtx6000_metrics/result_ablation_mbert/sst2/mBERT/hf_sequence_classifier/seed_43/eval_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"eval_accuracy": 0.8864678899082569,
|
| 4 |
+
"eval_loss": 0.4080387055873871,
|
| 5 |
+
"eval_runtime": 1.8265,
|
| 6 |
+
"eval_samples": 872,
|
| 7 |
+
"eval_samples_per_second": 477.422,
|
| 8 |
+
"eval_steps_per_second": 5.475
|
| 9 |
+
}
|
mbert_rtx6000_metrics/result_ablation_mbert/sst2/mBERT/hf_sequence_classifier/seed_43/run_command.txt
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
/usr/local/bin/python /workspace/run_glue_MHA_gated.py --model_name_or_path bert-base-multilingual-cased --dataset_name tmnam20/ViGLUE --task_name sst2 --max_seq_length 128 --do_train --do_eval --evaluation_strategy epoch --save_strategy no --logging_strategy steps --logging_steps 25 --report_to none --num_train_epochs 3 --learning_rate 3e-05 --per_device_train_batch_size 48 --per_device_eval_batch_size 96 --gradient_accumulation_steps 1 --dataloader_num_workers 1 --seed 43 --output_dir result_ablation_mbert/sst2/mBERT/hf_sequence_classifier/seed_43 --overwrite_output_dir --trust_remote_code true --bf16 --tf32 true --skip_model_save
|
mbert_rtx6000_metrics/result_ablation_mbert/sst2/mBERT/hf_sequence_classifier/seed_43/train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"total_flos": 6166879651621260.0,
|
| 4 |
+
"train_loss": 0.2339267143723179,
|
| 5 |
+
"train_runtime": 614.1221,
|
| 6 |
+
"train_samples": 67349,
|
| 7 |
+
"train_samples_per_second": 329.001,
|
| 8 |
+
"train_steps_per_second": 6.859
|
| 9 |
+
}
|
mbert_rtx6000_metrics/result_ablation_mbert/sst2/mBERT/hf_sequence_classifier/seed_43/trainer_state.json
ADDED
|
@@ -0,0 +1,1245 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 3.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 4212,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.017806267806267807,
|
| 13 |
+
"grad_norm": 9.142781257629395,
|
| 14 |
+
"learning_rate": 2.9821937321937322e-05,
|
| 15 |
+
"loss": 0.6182,
|
| 16 |
+
"step": 25
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.03561253561253561,
|
| 20 |
+
"grad_norm": 5.316717624664307,
|
| 21 |
+
"learning_rate": 2.9643874643874644e-05,
|
| 22 |
+
"loss": 0.5684,
|
| 23 |
+
"step": 50
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.053418803418803416,
|
| 27 |
+
"grad_norm": 5.904895305633545,
|
| 28 |
+
"learning_rate": 2.9465811965811966e-05,
|
| 29 |
+
"loss": 0.476,
|
| 30 |
+
"step": 75
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.07122507122507123,
|
| 34 |
+
"grad_norm": 5.940618991851807,
|
| 35 |
+
"learning_rate": 2.9287749287749288e-05,
|
| 36 |
+
"loss": 0.4622,
|
| 37 |
+
"step": 100
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.08903133903133903,
|
| 41 |
+
"grad_norm": 8.635560035705566,
|
| 42 |
+
"learning_rate": 2.910968660968661e-05,
|
| 43 |
+
"loss": 0.4882,
|
| 44 |
+
"step": 125
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.10683760683760683,
|
| 48 |
+
"grad_norm": 13.317228317260742,
|
| 49 |
+
"learning_rate": 2.893162393162393e-05,
|
| 50 |
+
"loss": 0.4654,
|
| 51 |
+
"step": 150
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.12464387464387465,
|
| 55 |
+
"grad_norm": 4.6408467292785645,
|
| 56 |
+
"learning_rate": 2.8753561253561256e-05,
|
| 57 |
+
"loss": 0.4155,
|
| 58 |
+
"step": 175
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.14245014245014245,
|
| 62 |
+
"grad_norm": 5.532208442687988,
|
| 63 |
+
"learning_rate": 2.8575498575498578e-05,
|
| 64 |
+
"loss": 0.3975,
|
| 65 |
+
"step": 200
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.16025641025641027,
|
| 69 |
+
"grad_norm": 4.817173957824707,
|
| 70 |
+
"learning_rate": 2.8397435897435896e-05,
|
| 71 |
+
"loss": 0.4456,
|
| 72 |
+
"step": 225
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.17806267806267806,
|
| 76 |
+
"grad_norm": 3.21955943107605,
|
| 77 |
+
"learning_rate": 2.821937321937322e-05,
|
| 78 |
+
"loss": 0.3906,
|
| 79 |
+
"step": 250
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.19586894586894588,
|
| 83 |
+
"grad_norm": 4.784478664398193,
|
| 84 |
+
"learning_rate": 2.8041310541310543e-05,
|
| 85 |
+
"loss": 0.4222,
|
| 86 |
+
"step": 275
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.21367521367521367,
|
| 90 |
+
"grad_norm": 3.998802900314331,
|
| 91 |
+
"learning_rate": 2.786324786324786e-05,
|
| 92 |
+
"loss": 0.4033,
|
| 93 |
+
"step": 300
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.23148148148148148,
|
| 97 |
+
"grad_norm": 5.3973798751831055,
|
| 98 |
+
"learning_rate": 2.7685185185185186e-05,
|
| 99 |
+
"loss": 0.3978,
|
| 100 |
+
"step": 325
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.2492877492877493,
|
| 104 |
+
"grad_norm": 4.920936107635498,
|
| 105 |
+
"learning_rate": 2.7507122507122508e-05,
|
| 106 |
+
"loss": 0.375,
|
| 107 |
+
"step": 350
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.2670940170940171,
|
| 111 |
+
"grad_norm": 12.355620384216309,
|
| 112 |
+
"learning_rate": 2.732905982905983e-05,
|
| 113 |
+
"loss": 0.3617,
|
| 114 |
+
"step": 375
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.2849002849002849,
|
| 118 |
+
"grad_norm": 9.220647811889648,
|
| 119 |
+
"learning_rate": 2.715099715099715e-05,
|
| 120 |
+
"loss": 0.3945,
|
| 121 |
+
"step": 400
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.3027065527065527,
|
| 125 |
+
"grad_norm": 5.939899921417236,
|
| 126 |
+
"learning_rate": 2.6972934472934473e-05,
|
| 127 |
+
"loss": 0.3868,
|
| 128 |
+
"step": 425
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.32051282051282054,
|
| 132 |
+
"grad_norm": 4.926178455352783,
|
| 133 |
+
"learning_rate": 2.67948717948718e-05,
|
| 134 |
+
"loss": 0.3732,
|
| 135 |
+
"step": 450
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.3383190883190883,
|
| 139 |
+
"grad_norm": 5.1928253173828125,
|
| 140 |
+
"learning_rate": 2.6616809116809117e-05,
|
| 141 |
+
"loss": 0.3663,
|
| 142 |
+
"step": 475
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.3561253561253561,
|
| 146 |
+
"grad_norm": 4.939578533172607,
|
| 147 |
+
"learning_rate": 2.643874643874644e-05,
|
| 148 |
+
"loss": 0.3513,
|
| 149 |
+
"step": 500
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.37393162393162394,
|
| 153 |
+
"grad_norm": 4.331312656402588,
|
| 154 |
+
"learning_rate": 2.6260683760683763e-05,
|
| 155 |
+
"loss": 0.3356,
|
| 156 |
+
"step": 525
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.39173789173789175,
|
| 160 |
+
"grad_norm": 4.112658500671387,
|
| 161 |
+
"learning_rate": 2.6082621082621082e-05,
|
| 162 |
+
"loss": 0.3476,
|
| 163 |
+
"step": 550
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.40954415954415957,
|
| 167 |
+
"grad_norm": 9.450350761413574,
|
| 168 |
+
"learning_rate": 2.5904558404558404e-05,
|
| 169 |
+
"loss": 0.3555,
|
| 170 |
+
"step": 575
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.42735042735042733,
|
| 174 |
+
"grad_norm": 5.540179252624512,
|
| 175 |
+
"learning_rate": 2.572649572649573e-05,
|
| 176 |
+
"loss": 0.3258,
|
| 177 |
+
"step": 600
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.44515669515669515,
|
| 181 |
+
"grad_norm": 4.87565279006958,
|
| 182 |
+
"learning_rate": 2.5548433048433047e-05,
|
| 183 |
+
"loss": 0.3484,
|
| 184 |
+
"step": 625
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.46296296296296297,
|
| 188 |
+
"grad_norm": 3.5047225952148438,
|
| 189 |
+
"learning_rate": 2.5370370370370372e-05,
|
| 190 |
+
"loss": 0.3218,
|
| 191 |
+
"step": 650
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.4807692307692308,
|
| 195 |
+
"grad_norm": 5.873397350311279,
|
| 196 |
+
"learning_rate": 2.5192307692307694e-05,
|
| 197 |
+
"loss": 0.3358,
|
| 198 |
+
"step": 675
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.4985754985754986,
|
| 202 |
+
"grad_norm": 3.2325706481933594,
|
| 203 |
+
"learning_rate": 2.5014245014245015e-05,
|
| 204 |
+
"loss": 0.3209,
|
| 205 |
+
"step": 700
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.5163817663817664,
|
| 209 |
+
"grad_norm": 3.2658989429473877,
|
| 210 |
+
"learning_rate": 2.4836182336182337e-05,
|
| 211 |
+
"loss": 0.3245,
|
| 212 |
+
"step": 725
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.5341880341880342,
|
| 216 |
+
"grad_norm": 5.441763877868652,
|
| 217 |
+
"learning_rate": 2.465811965811966e-05,
|
| 218 |
+
"loss": 0.359,
|
| 219 |
+
"step": 750
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.551994301994302,
|
| 223 |
+
"grad_norm": 3.5071098804473877,
|
| 224 |
+
"learning_rate": 2.448005698005698e-05,
|
| 225 |
+
"loss": 0.341,
|
| 226 |
+
"step": 775
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.5698005698005698,
|
| 230 |
+
"grad_norm": 4.158889293670654,
|
| 231 |
+
"learning_rate": 2.4301994301994302e-05,
|
| 232 |
+
"loss": 0.3304,
|
| 233 |
+
"step": 800
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.5876068376068376,
|
| 237 |
+
"grad_norm": 6.116085052490234,
|
| 238 |
+
"learning_rate": 2.4123931623931624e-05,
|
| 239 |
+
"loss": 0.3072,
|
| 240 |
+
"step": 825
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.6054131054131054,
|
| 244 |
+
"grad_norm": 2.7430808544158936,
|
| 245 |
+
"learning_rate": 2.3945868945868946e-05,
|
| 246 |
+
"loss": 0.2823,
|
| 247 |
+
"step": 850
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.6232193732193733,
|
| 251 |
+
"grad_norm": 5.130170822143555,
|
| 252 |
+
"learning_rate": 2.3767806267806267e-05,
|
| 253 |
+
"loss": 0.3415,
|
| 254 |
+
"step": 875
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.6410256410256411,
|
| 258 |
+
"grad_norm": 7.36715030670166,
|
| 259 |
+
"learning_rate": 2.358974358974359e-05,
|
| 260 |
+
"loss": 0.3226,
|
| 261 |
+
"step": 900
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.6588319088319088,
|
| 265 |
+
"grad_norm": 4.875875949859619,
|
| 266 |
+
"learning_rate": 2.3411680911680914e-05,
|
| 267 |
+
"loss": 0.3059,
|
| 268 |
+
"step": 925
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.6766381766381766,
|
| 272 |
+
"grad_norm": 9.965974807739258,
|
| 273 |
+
"learning_rate": 2.3233618233618236e-05,
|
| 274 |
+
"loss": 0.2966,
|
| 275 |
+
"step": 950
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.6944444444444444,
|
| 279 |
+
"grad_norm": 4.627007961273193,
|
| 280 |
+
"learning_rate": 2.3055555555555554e-05,
|
| 281 |
+
"loss": 0.2778,
|
| 282 |
+
"step": 975
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.7122507122507122,
|
| 286 |
+
"grad_norm": 6.086953163146973,
|
| 287 |
+
"learning_rate": 2.287749287749288e-05,
|
| 288 |
+
"loss": 0.3049,
|
| 289 |
+
"step": 1000
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.73005698005698,
|
| 293 |
+
"grad_norm": 5.65731143951416,
|
| 294 |
+
"learning_rate": 2.26994301994302e-05,
|
| 295 |
+
"loss": 0.3077,
|
| 296 |
+
"step": 1025
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.7478632478632479,
|
| 300 |
+
"grad_norm": 3.337294816970825,
|
| 301 |
+
"learning_rate": 2.252136752136752e-05,
|
| 302 |
+
"loss": 0.3029,
|
| 303 |
+
"step": 1050
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.7656695156695157,
|
| 307 |
+
"grad_norm": 3.77978515625,
|
| 308 |
+
"learning_rate": 2.2343304843304845e-05,
|
| 309 |
+
"loss": 0.2582,
|
| 310 |
+
"step": 1075
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.7834757834757835,
|
| 314 |
+
"grad_norm": 5.06842565536499,
|
| 315 |
+
"learning_rate": 2.2165242165242166e-05,
|
| 316 |
+
"loss": 0.3316,
|
| 317 |
+
"step": 1100
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 0.8012820512820513,
|
| 321 |
+
"grad_norm": 8.506872177124023,
|
| 322 |
+
"learning_rate": 2.1987179487179485e-05,
|
| 323 |
+
"loss": 0.2847,
|
| 324 |
+
"step": 1125
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.8190883190883191,
|
| 328 |
+
"grad_norm": 4.712424278259277,
|
| 329 |
+
"learning_rate": 2.180911680911681e-05,
|
| 330 |
+
"loss": 0.2999,
|
| 331 |
+
"step": 1150
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 0.8368945868945868,
|
| 335 |
+
"grad_norm": 6.640626907348633,
|
| 336 |
+
"learning_rate": 2.163105413105413e-05,
|
| 337 |
+
"loss": 0.312,
|
| 338 |
+
"step": 1175
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.8547008547008547,
|
| 342 |
+
"grad_norm": 4.37123441696167,
|
| 343 |
+
"learning_rate": 2.1452991452991456e-05,
|
| 344 |
+
"loss": 0.267,
|
| 345 |
+
"step": 1200
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.8725071225071225,
|
| 349 |
+
"grad_norm": 6.221196174621582,
|
| 350 |
+
"learning_rate": 2.1274928774928775e-05,
|
| 351 |
+
"loss": 0.3166,
|
| 352 |
+
"step": 1225
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 0.8903133903133903,
|
| 356 |
+
"grad_norm": 4.169477462768555,
|
| 357 |
+
"learning_rate": 2.1096866096866097e-05,
|
| 358 |
+
"loss": 0.3089,
|
| 359 |
+
"step": 1250
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 0.9081196581196581,
|
| 363 |
+
"grad_norm": 3.6809983253479004,
|
| 364 |
+
"learning_rate": 2.091880341880342e-05,
|
| 365 |
+
"loss": 0.2869,
|
| 366 |
+
"step": 1275
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 0.9259259259259259,
|
| 370 |
+
"grad_norm": 8.10815143585205,
|
| 371 |
+
"learning_rate": 2.074074074074074e-05,
|
| 372 |
+
"loss": 0.2823,
|
| 373 |
+
"step": 1300
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 0.9437321937321937,
|
| 377 |
+
"grad_norm": 7.873946189880371,
|
| 378 |
+
"learning_rate": 2.056267806267806e-05,
|
| 379 |
+
"loss": 0.2514,
|
| 380 |
+
"step": 1325
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 0.9615384615384616,
|
| 384 |
+
"grad_norm": 5.184146404266357,
|
| 385 |
+
"learning_rate": 2.0384615384615387e-05,
|
| 386 |
+
"loss": 0.2653,
|
| 387 |
+
"step": 1350
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 0.9793447293447294,
|
| 391 |
+
"grad_norm": 5.263430118560791,
|
| 392 |
+
"learning_rate": 2.0206552706552705e-05,
|
| 393 |
+
"loss": 0.3002,
|
| 394 |
+
"step": 1375
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 0.9971509971509972,
|
| 398 |
+
"grad_norm": 5.290490627288818,
|
| 399 |
+
"learning_rate": 2.002849002849003e-05,
|
| 400 |
+
"loss": 0.2663,
|
| 401 |
+
"step": 1400
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 1.0,
|
| 405 |
+
"eval_accuracy": 0.8623853211009175,
|
| 406 |
+
"eval_loss": 0.3829037845134735,
|
| 407 |
+
"eval_runtime": 3.3549,
|
| 408 |
+
"eval_samples_per_second": 259.916,
|
| 409 |
+
"eval_steps_per_second": 2.981,
|
| 410 |
+
"step": 1404
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 1.014957264957265,
|
| 414 |
+
"grad_norm": 7.713810443878174,
|
| 415 |
+
"learning_rate": 1.9850427350427352e-05,
|
| 416 |
+
"loss": 0.212,
|
| 417 |
+
"step": 1425
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"epoch": 1.0327635327635327,
|
| 421 |
+
"grad_norm": 7.390405178070068,
|
| 422 |
+
"learning_rate": 1.9672364672364674e-05,
|
| 423 |
+
"loss": 0.2371,
|
| 424 |
+
"step": 1450
|
| 425 |
+
},
|
| 426 |
+
{
|
| 427 |
+
"epoch": 1.0505698005698005,
|
| 428 |
+
"grad_norm": 6.012370586395264,
|
| 429 |
+
"learning_rate": 1.9494301994301995e-05,
|
| 430 |
+
"loss": 0.234,
|
| 431 |
+
"step": 1475
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 1.0683760683760684,
|
| 435 |
+
"grad_norm": 4.798014163970947,
|
| 436 |
+
"learning_rate": 1.9316239316239317e-05,
|
| 437 |
+
"loss": 0.2246,
|
| 438 |
+
"step": 1500
|
| 439 |
+
},
|
| 440 |
+
{
|
| 441 |
+
"epoch": 1.0861823361823362,
|
| 442 |
+
"grad_norm": 5.562344551086426,
|
| 443 |
+
"learning_rate": 1.913817663817664e-05,
|
| 444 |
+
"loss": 0.2338,
|
| 445 |
+
"step": 1525
|
| 446 |
+
},
|
| 447 |
+
{
|
| 448 |
+
"epoch": 1.103988603988604,
|
| 449 |
+
"grad_norm": 7.6661176681518555,
|
| 450 |
+
"learning_rate": 1.896011396011396e-05,
|
| 451 |
+
"loss": 0.221,
|
| 452 |
+
"step": 1550
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 1.1217948717948718,
|
| 456 |
+
"grad_norm": 4.017697334289551,
|
| 457 |
+
"learning_rate": 1.8782051282051282e-05,
|
| 458 |
+
"loss": 0.1961,
|
| 459 |
+
"step": 1575
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 1.1396011396011396,
|
| 463 |
+
"grad_norm": 10.011595726013184,
|
| 464 |
+
"learning_rate": 1.8603988603988604e-05,
|
| 465 |
+
"loss": 0.225,
|
| 466 |
+
"step": 1600
|
| 467 |
+
},
|
| 468 |
+
{
|
| 469 |
+
"epoch": 1.1574074074074074,
|
| 470 |
+
"grad_norm": 7.570844650268555,
|
| 471 |
+
"learning_rate": 1.8425925925925926e-05,
|
| 472 |
+
"loss": 0.203,
|
| 473 |
+
"step": 1625
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 1.1752136752136753,
|
| 477 |
+
"grad_norm": 4.3218488693237305,
|
| 478 |
+
"learning_rate": 1.8247863247863247e-05,
|
| 479 |
+
"loss": 0.2415,
|
| 480 |
+
"step": 1650
|
| 481 |
+
},
|
| 482 |
+
{
|
| 483 |
+
"epoch": 1.193019943019943,
|
| 484 |
+
"grad_norm": 8.753271102905273,
|
| 485 |
+
"learning_rate": 1.8069800569800572e-05,
|
| 486 |
+
"loss": 0.2226,
|
| 487 |
+
"step": 1675
|
| 488 |
+
},
|
| 489 |
+
{
|
| 490 |
+
"epoch": 1.210826210826211,
|
| 491 |
+
"grad_norm": 4.950158596038818,
|
| 492 |
+
"learning_rate": 1.7891737891737894e-05,
|
| 493 |
+
"loss": 0.2421,
|
| 494 |
+
"step": 1700
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 1.2286324786324787,
|
| 498 |
+
"grad_norm": 5.160200119018555,
|
| 499 |
+
"learning_rate": 1.7713675213675212e-05,
|
| 500 |
+
"loss": 0.2106,
|
| 501 |
+
"step": 1725
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 1.2464387464387465,
|
| 505 |
+
"grad_norm": 5.516897678375244,
|
| 506 |
+
"learning_rate": 1.7535612535612538e-05,
|
| 507 |
+
"loss": 0.2096,
|
| 508 |
+
"step": 1750
|
| 509 |
+
},
|
| 510 |
+
{
|
| 511 |
+
"epoch": 1.2642450142450143,
|
| 512 |
+
"grad_norm": 2.6887731552124023,
|
| 513 |
+
"learning_rate": 1.735754985754986e-05,
|
| 514 |
+
"loss": 0.2245,
|
| 515 |
+
"step": 1775
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 1.282051282051282,
|
| 519 |
+
"grad_norm": 5.131377220153809,
|
| 520 |
+
"learning_rate": 1.7179487179487178e-05,
|
| 521 |
+
"loss": 0.2186,
|
| 522 |
+
"step": 1800
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 1.29985754985755,
|
| 526 |
+
"grad_norm": 8.250959396362305,
|
| 527 |
+
"learning_rate": 1.7001424501424503e-05,
|
| 528 |
+
"loss": 0.2139,
|
| 529 |
+
"step": 1825
|
| 530 |
+
},
|
| 531 |
+
{
|
| 532 |
+
"epoch": 1.3176638176638176,
|
| 533 |
+
"grad_norm": 8.145303726196289,
|
| 534 |
+
"learning_rate": 1.6823361823361824e-05,
|
| 535 |
+
"loss": 0.2144,
|
| 536 |
+
"step": 1850
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 1.3354700854700854,
|
| 540 |
+
"grad_norm": 7.607888221740723,
|
| 541 |
+
"learning_rate": 1.6645299145299143e-05,
|
| 542 |
+
"loss": 0.2195,
|
| 543 |
+
"step": 1875
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 1.3532763532763532,
|
| 547 |
+
"grad_norm": 6.065781116485596,
|
| 548 |
+
"learning_rate": 1.6467236467236468e-05,
|
| 549 |
+
"loss": 0.2047,
|
| 550 |
+
"step": 1900
|
| 551 |
+
},
|
| 552 |
+
{
|
| 553 |
+
"epoch": 1.371082621082621,
|
| 554 |
+
"grad_norm": 4.67059326171875,
|
| 555 |
+
"learning_rate": 1.628917378917379e-05,
|
| 556 |
+
"loss": 0.2096,
|
| 557 |
+
"step": 1925
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"epoch": 1.3888888888888888,
|
| 561 |
+
"grad_norm": 6.151140213012695,
|
| 562 |
+
"learning_rate": 1.6111111111111115e-05,
|
| 563 |
+
"loss": 0.2177,
|
| 564 |
+
"step": 1950
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 1.4066951566951567,
|
| 568 |
+
"grad_norm": 3.8113534450531006,
|
| 569 |
+
"learning_rate": 1.5933048433048433e-05,
|
| 570 |
+
"loss": 0.2106,
|
| 571 |
+
"step": 1975
|
| 572 |
+
},
|
| 573 |
+
{
|
| 574 |
+
"epoch": 1.4245014245014245,
|
| 575 |
+
"grad_norm": 7.541844367980957,
|
| 576 |
+
"learning_rate": 1.5754985754985755e-05,
|
| 577 |
+
"loss": 0.2115,
|
| 578 |
+
"step": 2000
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 1.4423076923076923,
|
| 582 |
+
"grad_norm": 8.138694763183594,
|
| 583 |
+
"learning_rate": 1.557692307692308e-05,
|
| 584 |
+
"loss": 0.2343,
|
| 585 |
+
"step": 2025
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 1.46011396011396,
|
| 589 |
+
"grad_norm": 7.199629306793213,
|
| 590 |
+
"learning_rate": 1.5398860398860398e-05,
|
| 591 |
+
"loss": 0.2072,
|
| 592 |
+
"step": 2050
|
| 593 |
+
},
|
| 594 |
+
{
|
| 595 |
+
"epoch": 1.477920227920228,
|
| 596 |
+
"grad_norm": 4.225950717926025,
|
| 597 |
+
"learning_rate": 1.5220797720797721e-05,
|
| 598 |
+
"loss": 0.1929,
|
| 599 |
+
"step": 2075
|
| 600 |
+
},
|
| 601 |
+
{
|
| 602 |
+
"epoch": 1.4957264957264957,
|
| 603 |
+
"grad_norm": 4.128393173217773,
|
| 604 |
+
"learning_rate": 1.5042735042735045e-05,
|
| 605 |
+
"loss": 0.1967,
|
| 606 |
+
"step": 2100
|
| 607 |
+
},
|
| 608 |
+
{
|
| 609 |
+
"epoch": 1.5135327635327636,
|
| 610 |
+
"grad_norm": 6.81208610534668,
|
| 611 |
+
"learning_rate": 1.4864672364672365e-05,
|
| 612 |
+
"loss": 0.2178,
|
| 613 |
+
"step": 2125
|
| 614 |
+
},
|
| 615 |
+
{
|
| 616 |
+
"epoch": 1.5313390313390314,
|
| 617 |
+
"grad_norm": 4.572250843048096,
|
| 618 |
+
"learning_rate": 1.4686609686609687e-05,
|
| 619 |
+
"loss": 0.2058,
|
| 620 |
+
"step": 2150
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 1.5491452991452992,
|
| 624 |
+
"grad_norm": 7.7006025314331055,
|
| 625 |
+
"learning_rate": 1.450854700854701e-05,
|
| 626 |
+
"loss": 0.2086,
|
| 627 |
+
"step": 2175
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"epoch": 1.566951566951567,
|
| 631 |
+
"grad_norm": 5.852411270141602,
|
| 632 |
+
"learning_rate": 1.433048433048433e-05,
|
| 633 |
+
"loss": 0.1943,
|
| 634 |
+
"step": 2200
|
| 635 |
+
},
|
| 636 |
+
{
|
| 637 |
+
"epoch": 1.5847578347578346,
|
| 638 |
+
"grad_norm": 9.573450088500977,
|
| 639 |
+
"learning_rate": 1.4152421652421652e-05,
|
| 640 |
+
"loss": 0.2241,
|
| 641 |
+
"step": 2225
|
| 642 |
+
},
|
| 643 |
+
{
|
| 644 |
+
"epoch": 1.6025641025641026,
|
| 645 |
+
"grad_norm": 10.747583389282227,
|
| 646 |
+
"learning_rate": 1.3974358974358975e-05,
|
| 647 |
+
"loss": 0.2005,
|
| 648 |
+
"step": 2250
|
| 649 |
+
},
|
| 650 |
+
{
|
| 651 |
+
"epoch": 1.6203703703703702,
|
| 652 |
+
"grad_norm": 9.476792335510254,
|
| 653 |
+
"learning_rate": 1.3796296296296297e-05,
|
| 654 |
+
"loss": 0.2266,
|
| 655 |
+
"step": 2275
|
| 656 |
+
},
|
| 657 |
+
{
|
| 658 |
+
"epoch": 1.6381766381766383,
|
| 659 |
+
"grad_norm": 4.619626045227051,
|
| 660 |
+
"learning_rate": 1.3618233618233619e-05,
|
| 661 |
+
"loss": 0.2053,
|
| 662 |
+
"step": 2300
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 1.6559829059829059,
|
| 666 |
+
"grad_norm": 7.528178691864014,
|
| 667 |
+
"learning_rate": 1.344017094017094e-05,
|
| 668 |
+
"loss": 0.2405,
|
| 669 |
+
"step": 2325
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 1.673789173789174,
|
| 673 |
+
"grad_norm": 5.723983287811279,
|
| 674 |
+
"learning_rate": 1.3262108262108262e-05,
|
| 675 |
+
"loss": 0.2043,
|
| 676 |
+
"step": 2350
|
| 677 |
+
},
|
| 678 |
+
{
|
| 679 |
+
"epoch": 1.6915954415954415,
|
| 680 |
+
"grad_norm": 12.369087219238281,
|
| 681 |
+
"learning_rate": 1.3084045584045585e-05,
|
| 682 |
+
"loss": 0.1988,
|
| 683 |
+
"step": 2375
|
| 684 |
+
},
|
| 685 |
+
{
|
| 686 |
+
"epoch": 1.7094017094017095,
|
| 687 |
+
"grad_norm": 6.811823844909668,
|
| 688 |
+
"learning_rate": 1.2905982905982905e-05,
|
| 689 |
+
"loss": 0.2012,
|
| 690 |
+
"step": 2400
|
| 691 |
+
},
|
| 692 |
+
{
|
| 693 |
+
"epoch": 1.7272079772079771,
|
| 694 |
+
"grad_norm": 4.423567771911621,
|
| 695 |
+
"learning_rate": 1.2727920227920229e-05,
|
| 696 |
+
"loss": 0.1963,
|
| 697 |
+
"step": 2425
|
| 698 |
+
},
|
| 699 |
+
{
|
| 700 |
+
"epoch": 1.7450142450142452,
|
| 701 |
+
"grad_norm": 5.570356369018555,
|
| 702 |
+
"learning_rate": 1.254985754985755e-05,
|
| 703 |
+
"loss": 0.2074,
|
| 704 |
+
"step": 2450
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 1.7628205128205128,
|
| 708 |
+
"grad_norm": 4.269906997680664,
|
| 709 |
+
"learning_rate": 1.2371794871794872e-05,
|
| 710 |
+
"loss": 0.1903,
|
| 711 |
+
"step": 2475
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"epoch": 1.7806267806267806,
|
| 715 |
+
"grad_norm": 6.301336288452148,
|
| 716 |
+
"learning_rate": 1.2193732193732194e-05,
|
| 717 |
+
"loss": 0.23,
|
| 718 |
+
"step": 2500
|
| 719 |
+
},
|
| 720 |
+
{
|
| 721 |
+
"epoch": 1.7984330484330484,
|
| 722 |
+
"grad_norm": 7.00869083404541,
|
| 723 |
+
"learning_rate": 1.2015669515669516e-05,
|
| 724 |
+
"loss": 0.1992,
|
| 725 |
+
"step": 2525
|
| 726 |
+
},
|
| 727 |
+
{
|
| 728 |
+
"epoch": 1.8162393162393162,
|
| 729 |
+
"grad_norm": 8.060930252075195,
|
| 730 |
+
"learning_rate": 1.1837606837606839e-05,
|
| 731 |
+
"loss": 0.1908,
|
| 732 |
+
"step": 2550
|
| 733 |
+
},
|
| 734 |
+
{
|
| 735 |
+
"epoch": 1.834045584045584,
|
| 736 |
+
"grad_norm": 10.444920539855957,
|
| 737 |
+
"learning_rate": 1.1659544159544159e-05,
|
| 738 |
+
"loss": 0.2009,
|
| 739 |
+
"step": 2575
|
| 740 |
+
},
|
| 741 |
+
{
|
| 742 |
+
"epoch": 1.8518518518518519,
|
| 743 |
+
"grad_norm": 6.913994312286377,
|
| 744 |
+
"learning_rate": 1.148148148148148e-05,
|
| 745 |
+
"loss": 0.2016,
|
| 746 |
+
"step": 2600
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 1.8696581196581197,
|
| 750 |
+
"grad_norm": 5.606993198394775,
|
| 751 |
+
"learning_rate": 1.1303418803418804e-05,
|
| 752 |
+
"loss": 0.1952,
|
| 753 |
+
"step": 2625
|
| 754 |
+
},
|
| 755 |
+
{
|
| 756 |
+
"epoch": 1.8874643874643875,
|
| 757 |
+
"grad_norm": 7.157936096191406,
|
| 758 |
+
"learning_rate": 1.1125356125356126e-05,
|
| 759 |
+
"loss": 0.1896,
|
| 760 |
+
"step": 2650
|
| 761 |
+
},
|
| 762 |
+
{
|
| 763 |
+
"epoch": 1.9052706552706553,
|
| 764 |
+
"grad_norm": 7.919309616088867,
|
| 765 |
+
"learning_rate": 1.0947293447293448e-05,
|
| 766 |
+
"loss": 0.2017,
|
| 767 |
+
"step": 2675
|
| 768 |
+
},
|
| 769 |
+
{
|
| 770 |
+
"epoch": 1.9230769230769231,
|
| 771 |
+
"grad_norm": 5.88226842880249,
|
| 772 |
+
"learning_rate": 1.076923076923077e-05,
|
| 773 |
+
"loss": 0.2043,
|
| 774 |
+
"step": 2700
|
| 775 |
+
},
|
| 776 |
+
{
|
| 777 |
+
"epoch": 1.940883190883191,
|
| 778 |
+
"grad_norm": 7.708987236022949,
|
| 779 |
+
"learning_rate": 1.0591168091168091e-05,
|
| 780 |
+
"loss": 0.2022,
|
| 781 |
+
"step": 2725
|
| 782 |
+
},
|
| 783 |
+
{
|
| 784 |
+
"epoch": 1.9586894586894585,
|
| 785 |
+
"grad_norm": 5.221554279327393,
|
| 786 |
+
"learning_rate": 1.0413105413105414e-05,
|
| 787 |
+
"loss": 0.1896,
|
| 788 |
+
"step": 2750
|
| 789 |
+
},
|
| 790 |
+
{
|
| 791 |
+
"epoch": 1.9764957264957266,
|
| 792 |
+
"grad_norm": 6.299310684204102,
|
| 793 |
+
"learning_rate": 1.0235042735042734e-05,
|
| 794 |
+
"loss": 0.1815,
|
| 795 |
+
"step": 2775
|
| 796 |
+
},
|
| 797 |
+
{
|
| 798 |
+
"epoch": 1.9943019943019942,
|
| 799 |
+
"grad_norm": 6.227346420288086,
|
| 800 |
+
"learning_rate": 1.0056980056980058e-05,
|
| 801 |
+
"loss": 0.1893,
|
| 802 |
+
"step": 2800
|
| 803 |
+
},
|
| 804 |
+
{
|
| 805 |
+
"epoch": 2.0,
|
| 806 |
+
"eval_accuracy": 0.8761467889908257,
|
| 807 |
+
"eval_loss": 0.3532339632511139,
|
| 808 |
+
"eval_runtime": 3.2873,
|
| 809 |
+
"eval_samples_per_second": 265.267,
|
| 810 |
+
"eval_steps_per_second": 3.042,
|
| 811 |
+
"step": 2808
|
| 812 |
+
},
|
| 813 |
+
{
|
| 814 |
+
"epoch": 2.012108262108262,
|
| 815 |
+
"grad_norm": 6.901354789733887,
|
| 816 |
+
"learning_rate": 9.87891737891738e-06,
|
| 817 |
+
"loss": 0.1462,
|
| 818 |
+
"step": 2825
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 2.02991452991453,
|
| 822 |
+
"grad_norm": 5.1674885749816895,
|
| 823 |
+
"learning_rate": 9.700854700854701e-06,
|
| 824 |
+
"loss": 0.1411,
|
| 825 |
+
"step": 2850
|
| 826 |
+
},
|
| 827 |
+
{
|
| 828 |
+
"epoch": 2.047720797720798,
|
| 829 |
+
"grad_norm": 5.998133659362793,
|
| 830 |
+
"learning_rate": 9.522792022792023e-06,
|
| 831 |
+
"loss": 0.1287,
|
| 832 |
+
"step": 2875
|
| 833 |
+
},
|
| 834 |
+
{
|
| 835 |
+
"epoch": 2.0655270655270654,
|
| 836 |
+
"grad_norm": 9.355203628540039,
|
| 837 |
+
"learning_rate": 9.344729344729345e-06,
|
| 838 |
+
"loss": 0.1345,
|
| 839 |
+
"step": 2900
|
| 840 |
+
},
|
| 841 |
+
{
|
| 842 |
+
"epoch": 2.0833333333333335,
|
| 843 |
+
"grad_norm": 3.6849324703216553,
|
| 844 |
+
"learning_rate": 9.166666666666668e-06,
|
| 845 |
+
"loss": 0.1446,
|
| 846 |
+
"step": 2925
|
| 847 |
+
},
|
| 848 |
+
{
|
| 849 |
+
"epoch": 2.101139601139601,
|
| 850 |
+
"grad_norm": 7.546777248382568,
|
| 851 |
+
"learning_rate": 8.988603988603988e-06,
|
| 852 |
+
"loss": 0.1416,
|
| 853 |
+
"step": 2950
|
| 854 |
+
},
|
| 855 |
+
{
|
| 856 |
+
"epoch": 2.118945868945869,
|
| 857 |
+
"grad_norm": 6.241270542144775,
|
| 858 |
+
"learning_rate": 8.81054131054131e-06,
|
| 859 |
+
"loss": 0.1111,
|
| 860 |
+
"step": 2975
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 2.1367521367521367,
|
| 864 |
+
"grad_norm": 6.003485202789307,
|
| 865 |
+
"learning_rate": 8.632478632478633e-06,
|
| 866 |
+
"loss": 0.173,
|
| 867 |
+
"step": 3000
|
| 868 |
+
},
|
| 869 |
+
{
|
| 870 |
+
"epoch": 2.1545584045584047,
|
| 871 |
+
"grad_norm": 7.710633754730225,
|
| 872 |
+
"learning_rate": 8.454415954415955e-06,
|
| 873 |
+
"loss": 0.1401,
|
| 874 |
+
"step": 3025
|
| 875 |
+
},
|
| 876 |
+
{
|
| 877 |
+
"epoch": 2.1723646723646723,
|
| 878 |
+
"grad_norm": 7.690117835998535,
|
| 879 |
+
"learning_rate": 8.276353276353277e-06,
|
| 880 |
+
"loss": 0.1533,
|
| 881 |
+
"step": 3050
|
| 882 |
+
},
|
| 883 |
+
{
|
| 884 |
+
"epoch": 2.1901709401709404,
|
| 885 |
+
"grad_norm": 5.578719615936279,
|
| 886 |
+
"learning_rate": 8.098290598290598e-06,
|
| 887 |
+
"loss": 0.1381,
|
| 888 |
+
"step": 3075
|
| 889 |
+
},
|
| 890 |
+
{
|
| 891 |
+
"epoch": 2.207977207977208,
|
| 892 |
+
"grad_norm": 6.358754634857178,
|
| 893 |
+
"learning_rate": 7.92022792022792e-06,
|
| 894 |
+
"loss": 0.1719,
|
| 895 |
+
"step": 3100
|
| 896 |
+
},
|
| 897 |
+
{
|
| 898 |
+
"epoch": 2.2257834757834756,
|
| 899 |
+
"grad_norm": 6.356100559234619,
|
| 900 |
+
"learning_rate": 7.742165242165244e-06,
|
| 901 |
+
"loss": 0.1501,
|
| 902 |
+
"step": 3125
|
| 903 |
+
},
|
| 904 |
+
{
|
| 905 |
+
"epoch": 2.2435897435897436,
|
| 906 |
+
"grad_norm": 4.342570781707764,
|
| 907 |
+
"learning_rate": 7.564102564102564e-06,
|
| 908 |
+
"loss": 0.1325,
|
| 909 |
+
"step": 3150
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 2.261396011396011,
|
| 913 |
+
"grad_norm": 9.004411697387695,
|
| 914 |
+
"learning_rate": 7.386039886039886e-06,
|
| 915 |
+
"loss": 0.1669,
|
| 916 |
+
"step": 3175
|
| 917 |
+
},
|
| 918 |
+
{
|
| 919 |
+
"epoch": 2.2792022792022792,
|
| 920 |
+
"grad_norm": 4.607677936553955,
|
| 921 |
+
"learning_rate": 7.207977207977209e-06,
|
| 922 |
+
"loss": 0.125,
|
| 923 |
+
"step": 3200
|
| 924 |
+
},
|
| 925 |
+
{
|
| 926 |
+
"epoch": 2.297008547008547,
|
| 927 |
+
"grad_norm": 1.611449122428894,
|
| 928 |
+
"learning_rate": 7.02991452991453e-06,
|
| 929 |
+
"loss": 0.1257,
|
| 930 |
+
"step": 3225
|
| 931 |
+
},
|
| 932 |
+
{
|
| 933 |
+
"epoch": 2.314814814814815,
|
| 934 |
+
"grad_norm": 4.159427165985107,
|
| 935 |
+
"learning_rate": 6.851851851851852e-06,
|
| 936 |
+
"loss": 0.1772,
|
| 937 |
+
"step": 3250
|
| 938 |
+
},
|
| 939 |
+
{
|
| 940 |
+
"epoch": 2.3326210826210825,
|
| 941 |
+
"grad_norm": 10.165175437927246,
|
| 942 |
+
"learning_rate": 6.673789173789174e-06,
|
| 943 |
+
"loss": 0.1581,
|
| 944 |
+
"step": 3275
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 2.3504273504273505,
|
| 948 |
+
"grad_norm": 10.160175323486328,
|
| 949 |
+
"learning_rate": 6.4957264957264955e-06,
|
| 950 |
+
"loss": 0.146,
|
| 951 |
+
"step": 3300
|
| 952 |
+
},
|
| 953 |
+
{
|
| 954 |
+
"epoch": 2.368233618233618,
|
| 955 |
+
"grad_norm": 7.565778732299805,
|
| 956 |
+
"learning_rate": 6.317663817663818e-06,
|
| 957 |
+
"loss": 0.1324,
|
| 958 |
+
"step": 3325
|
| 959 |
+
},
|
| 960 |
+
{
|
| 961 |
+
"epoch": 2.386039886039886,
|
| 962 |
+
"grad_norm": 2.8253228664398193,
|
| 963 |
+
"learning_rate": 6.13960113960114e-06,
|
| 964 |
+
"loss": 0.1239,
|
| 965 |
+
"step": 3350
|
| 966 |
+
},
|
| 967 |
+
{
|
| 968 |
+
"epoch": 2.4038461538461537,
|
| 969 |
+
"grad_norm": 10.886297225952148,
|
| 970 |
+
"learning_rate": 5.9615384615384615e-06,
|
| 971 |
+
"loss": 0.1332,
|
| 972 |
+
"step": 3375
|
| 973 |
+
},
|
| 974 |
+
{
|
| 975 |
+
"epoch": 2.421652421652422,
|
| 976 |
+
"grad_norm": 6.231756210327148,
|
| 977 |
+
"learning_rate": 5.783475783475783e-06,
|
| 978 |
+
"loss": 0.128,
|
| 979 |
+
"step": 3400
|
| 980 |
+
},
|
| 981 |
+
{
|
| 982 |
+
"epoch": 2.4394586894586894,
|
| 983 |
+
"grad_norm": 6.863856315612793,
|
| 984 |
+
"learning_rate": 5.605413105413106e-06,
|
| 985 |
+
"loss": 0.1417,
|
| 986 |
+
"step": 3425
|
| 987 |
+
},
|
| 988 |
+
{
|
| 989 |
+
"epoch": 2.4572649572649574,
|
| 990 |
+
"grad_norm": 6.738685607910156,
|
| 991 |
+
"learning_rate": 5.4273504273504275e-06,
|
| 992 |
+
"loss": 0.1232,
|
| 993 |
+
"step": 3450
|
| 994 |
+
},
|
| 995 |
+
{
|
| 996 |
+
"epoch": 2.475071225071225,
|
| 997 |
+
"grad_norm": 3.517829656600952,
|
| 998 |
+
"learning_rate": 5.24928774928775e-06,
|
| 999 |
+
"loss": 0.1289,
|
| 1000 |
+
"step": 3475
|
| 1001 |
+
},
|
| 1002 |
+
{
|
| 1003 |
+
"epoch": 2.492877492877493,
|
| 1004 |
+
"grad_norm": 6.8838300704956055,
|
| 1005 |
+
"learning_rate": 5.071225071225071e-06,
|
| 1006 |
+
"loss": 0.1379,
|
| 1007 |
+
"step": 3500
|
| 1008 |
+
},
|
| 1009 |
+
{
|
| 1010 |
+
"epoch": 2.5106837606837606,
|
| 1011 |
+
"grad_norm": 8.647956848144531,
|
| 1012 |
+
"learning_rate": 4.8931623931623934e-06,
|
| 1013 |
+
"loss": 0.1413,
|
| 1014 |
+
"step": 3525
|
| 1015 |
+
},
|
| 1016 |
+
{
|
| 1017 |
+
"epoch": 2.5284900284900287,
|
| 1018 |
+
"grad_norm": 7.859837055206299,
|
| 1019 |
+
"learning_rate": 4.715099715099715e-06,
|
| 1020 |
+
"loss": 0.1708,
|
| 1021 |
+
"step": 3550
|
| 1022 |
+
},
|
| 1023 |
+
{
|
| 1024 |
+
"epoch": 2.5462962962962963,
|
| 1025 |
+
"grad_norm": 8.834534645080566,
|
| 1026 |
+
"learning_rate": 4.537037037037038e-06,
|
| 1027 |
+
"loss": 0.156,
|
| 1028 |
+
"step": 3575
|
| 1029 |
+
},
|
| 1030 |
+
{
|
| 1031 |
+
"epoch": 2.564102564102564,
|
| 1032 |
+
"grad_norm": 9.083367347717285,
|
| 1033 |
+
"learning_rate": 4.358974358974359e-06,
|
| 1034 |
+
"loss": 0.1305,
|
| 1035 |
+
"step": 3600
|
| 1036 |
+
},
|
| 1037 |
+
{
|
| 1038 |
+
"epoch": 2.581908831908832,
|
| 1039 |
+
"grad_norm": 2.169090986251831,
|
| 1040 |
+
"learning_rate": 4.18091168091168e-06,
|
| 1041 |
+
"loss": 0.1411,
|
| 1042 |
+
"step": 3625
|
| 1043 |
+
},
|
| 1044 |
+
{
|
| 1045 |
+
"epoch": 2.5997150997151,
|
| 1046 |
+
"grad_norm": 5.438903331756592,
|
| 1047 |
+
"learning_rate": 4.002849002849003e-06,
|
| 1048 |
+
"loss": 0.1552,
|
| 1049 |
+
"step": 3650
|
| 1050 |
+
},
|
| 1051 |
+
{
|
| 1052 |
+
"epoch": 2.6175213675213675,
|
| 1053 |
+
"grad_norm": 4.352532386779785,
|
| 1054 |
+
"learning_rate": 3.8247863247863246e-06,
|
| 1055 |
+
"loss": 0.1325,
|
| 1056 |
+
"step": 3675
|
| 1057 |
+
},
|
| 1058 |
+
{
|
| 1059 |
+
"epoch": 2.635327635327635,
|
| 1060 |
+
"grad_norm": 6.723628997802734,
|
| 1061 |
+
"learning_rate": 3.646723646723647e-06,
|
| 1062 |
+
"loss": 0.1205,
|
| 1063 |
+
"step": 3700
|
| 1064 |
+
},
|
| 1065 |
+
{
|
| 1066 |
+
"epoch": 2.653133903133903,
|
| 1067 |
+
"grad_norm": 4.988849639892578,
|
| 1068 |
+
"learning_rate": 3.468660968660969e-06,
|
| 1069 |
+
"loss": 0.1441,
|
| 1070 |
+
"step": 3725
|
| 1071 |
+
},
|
| 1072 |
+
{
|
| 1073 |
+
"epoch": 2.6709401709401708,
|
| 1074 |
+
"grad_norm": 4.654062271118164,
|
| 1075 |
+
"learning_rate": 3.290598290598291e-06,
|
| 1076 |
+
"loss": 0.1257,
|
| 1077 |
+
"step": 3750
|
| 1078 |
+
},
|
| 1079 |
+
{
|
| 1080 |
+
"epoch": 2.688746438746439,
|
| 1081 |
+
"grad_norm": 7.567593097686768,
|
| 1082 |
+
"learning_rate": 3.1125356125356127e-06,
|
| 1083 |
+
"loss": 0.1521,
|
| 1084 |
+
"step": 3775
|
| 1085 |
+
},
|
| 1086 |
+
{
|
| 1087 |
+
"epoch": 2.7065527065527064,
|
| 1088 |
+
"grad_norm": 9.861932754516602,
|
| 1089 |
+
"learning_rate": 2.9344729344729344e-06,
|
| 1090 |
+
"loss": 0.1355,
|
| 1091 |
+
"step": 3800
|
| 1092 |
+
},
|
| 1093 |
+
{
|
| 1094 |
+
"epoch": 2.7243589743589745,
|
| 1095 |
+
"grad_norm": 3.0837855339050293,
|
| 1096 |
+
"learning_rate": 2.7564102564102565e-06,
|
| 1097 |
+
"loss": 0.1591,
|
| 1098 |
+
"step": 3825
|
| 1099 |
+
},
|
| 1100 |
+
{
|
| 1101 |
+
"epoch": 2.742165242165242,
|
| 1102 |
+
"grad_norm": 8.532748222351074,
|
| 1103 |
+
"learning_rate": 2.5783475783475782e-06,
|
| 1104 |
+
"loss": 0.1154,
|
| 1105 |
+
"step": 3850
|
| 1106 |
+
},
|
| 1107 |
+
{
|
| 1108 |
+
"epoch": 2.75997150997151,
|
| 1109 |
+
"grad_norm": 5.763181209564209,
|
| 1110 |
+
"learning_rate": 2.4002849002849004e-06,
|
| 1111 |
+
"loss": 0.1341,
|
| 1112 |
+
"step": 3875
|
| 1113 |
+
},
|
| 1114 |
+
{
|
| 1115 |
+
"epoch": 2.7777777777777777,
|
| 1116 |
+
"grad_norm": 6.507135391235352,
|
| 1117 |
+
"learning_rate": 2.222222222222222e-06,
|
| 1118 |
+
"loss": 0.1532,
|
| 1119 |
+
"step": 3900
|
| 1120 |
+
},
|
| 1121 |
+
{
|
| 1122 |
+
"epoch": 2.7955840455840457,
|
| 1123 |
+
"grad_norm": 8.4640474319458,
|
| 1124 |
+
"learning_rate": 2.044159544159544e-06,
|
| 1125 |
+
"loss": 0.1421,
|
| 1126 |
+
"step": 3925
|
| 1127 |
+
},
|
| 1128 |
+
{
|
| 1129 |
+
"epoch": 2.8133903133903133,
|
| 1130 |
+
"grad_norm": 8.161584854125977,
|
| 1131 |
+
"learning_rate": 1.8660968660968661e-06,
|
| 1132 |
+
"loss": 0.1303,
|
| 1133 |
+
"step": 3950
|
| 1134 |
+
},
|
| 1135 |
+
{
|
| 1136 |
+
"epoch": 2.8311965811965814,
|
| 1137 |
+
"grad_norm": 7.714197158813477,
|
| 1138 |
+
"learning_rate": 1.688034188034188e-06,
|
| 1139 |
+
"loss": 0.1455,
|
| 1140 |
+
"step": 3975
|
| 1141 |
+
},
|
| 1142 |
+
{
|
| 1143 |
+
"epoch": 2.849002849002849,
|
| 1144 |
+
"grad_norm": 7.299932479858398,
|
| 1145 |
+
"learning_rate": 1.50997150997151e-06,
|
| 1146 |
+
"loss": 0.1434,
|
| 1147 |
+
"step": 4000
|
| 1148 |
+
},
|
| 1149 |
+
{
|
| 1150 |
+
"epoch": 2.866809116809117,
|
| 1151 |
+
"grad_norm": 4.749539375305176,
|
| 1152 |
+
"learning_rate": 1.3319088319088321e-06,
|
| 1153 |
+
"loss": 0.134,
|
| 1154 |
+
"step": 4025
|
| 1155 |
+
},
|
| 1156 |
+
{
|
| 1157 |
+
"epoch": 2.8846153846153846,
|
| 1158 |
+
"grad_norm": 4.350413799285889,
|
| 1159 |
+
"learning_rate": 1.153846153846154e-06,
|
| 1160 |
+
"loss": 0.139,
|
| 1161 |
+
"step": 4050
|
| 1162 |
+
},
|
| 1163 |
+
{
|
| 1164 |
+
"epoch": 2.902421652421652,
|
| 1165 |
+
"grad_norm": 8.422185897827148,
|
| 1166 |
+
"learning_rate": 9.757834757834757e-07,
|
| 1167 |
+
"loss": 0.1343,
|
| 1168 |
+
"step": 4075
|
| 1169 |
+
},
|
| 1170 |
+
{
|
| 1171 |
+
"epoch": 2.92022792022792,
|
| 1172 |
+
"grad_norm": 6.347653388977051,
|
| 1173 |
+
"learning_rate": 7.977207977207978e-07,
|
| 1174 |
+
"loss": 0.1431,
|
| 1175 |
+
"step": 4100
|
| 1176 |
+
},
|
| 1177 |
+
{
|
| 1178 |
+
"epoch": 2.9380341880341883,
|
| 1179 |
+
"grad_norm": 3.6505072116851807,
|
| 1180 |
+
"learning_rate": 6.196581196581196e-07,
|
| 1181 |
+
"loss": 0.1304,
|
| 1182 |
+
"step": 4125
|
| 1183 |
+
},
|
| 1184 |
+
{
|
| 1185 |
+
"epoch": 2.955840455840456,
|
| 1186 |
+
"grad_norm": 6.734041690826416,
|
| 1187 |
+
"learning_rate": 4.415954415954416e-07,
|
| 1188 |
+
"loss": 0.1464,
|
| 1189 |
+
"step": 4150
|
| 1190 |
+
},
|
| 1191 |
+
{
|
| 1192 |
+
"epoch": 2.9736467236467234,
|
| 1193 |
+
"grad_norm": 0.9986034035682678,
|
| 1194 |
+
"learning_rate": 2.635327635327635e-07,
|
| 1195 |
+
"loss": 0.1148,
|
| 1196 |
+
"step": 4175
|
| 1197 |
+
},
|
| 1198 |
+
{
|
| 1199 |
+
"epoch": 2.9914529914529915,
|
| 1200 |
+
"grad_norm": 11.330714225769043,
|
| 1201 |
+
"learning_rate": 8.547008547008547e-08,
|
| 1202 |
+
"loss": 0.1407,
|
| 1203 |
+
"step": 4200
|
| 1204 |
+
},
|
| 1205 |
+
{
|
| 1206 |
+
"epoch": 3.0,
|
| 1207 |
+
"eval_accuracy": 0.8864678899082569,
|
| 1208 |
+
"eval_loss": 0.4080387055873871,
|
| 1209 |
+
"eval_runtime": 1.9307,
|
| 1210 |
+
"eval_samples_per_second": 451.643,
|
| 1211 |
+
"eval_steps_per_second": 5.179,
|
| 1212 |
+
"step": 4212
|
| 1213 |
+
},
|
| 1214 |
+
{
|
| 1215 |
+
"epoch": 3.0,
|
| 1216 |
+
"step": 4212,
|
| 1217 |
+
"total_flos": 6166879651621260.0,
|
| 1218 |
+
"train_loss": 0.2339267143723179,
|
| 1219 |
+
"train_runtime": 614.1221,
|
| 1220 |
+
"train_samples_per_second": 329.001,
|
| 1221 |
+
"train_steps_per_second": 6.859
|
| 1222 |
+
}
|
| 1223 |
+
],
|
| 1224 |
+
"logging_steps": 25,
|
| 1225 |
+
"max_steps": 4212,
|
| 1226 |
+
"num_input_tokens_seen": 0,
|
| 1227 |
+
"num_train_epochs": 3,
|
| 1228 |
+
"save_steps": 500,
|
| 1229 |
+
"stateful_callbacks": {
|
| 1230 |
+
"TrainerControl": {
|
| 1231 |
+
"args": {
|
| 1232 |
+
"should_epoch_stop": false,
|
| 1233 |
+
"should_evaluate": false,
|
| 1234 |
+
"should_log": false,
|
| 1235 |
+
"should_save": false,
|
| 1236 |
+
"should_training_stop": false
|
| 1237 |
+
},
|
| 1238 |
+
"attributes": {}
|
| 1239 |
+
}
|
| 1240 |
+
},
|
| 1241 |
+
"total_flos": 6166879651621260.0,
|
| 1242 |
+
"train_batch_size": 48,
|
| 1243 |
+
"trial_name": null,
|
| 1244 |
+
"trial_params": null
|
| 1245 |
+
}
|
mbert_rtx6000_metrics/result_ablation_mbert/sst2/mBERT/hf_sequence_classifier/seed_44/all_results.json
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"eval_accuracy": 0.8692660550458715,
|
| 4 |
+
"eval_loss": 0.42636725306510925,
|
| 5 |
+
"eval_runtime": 1.8276,
|
| 6 |
+
"eval_samples": 872,
|
| 7 |
+
"eval_samples_per_second": 477.133,
|
| 8 |
+
"eval_steps_per_second": 5.472,
|
| 9 |
+
"total_flos": 6159320861556240.0,
|
| 10 |
+
"train_loss": 0.23417754752909922,
|
| 11 |
+
"train_runtime": 612.7269,
|
| 12 |
+
"train_samples": 67349,
|
| 13 |
+
"train_samples_per_second": 329.75,
|
| 14 |
+
"train_steps_per_second": 6.874
|
| 15 |
+
}
|
mbert_rtx6000_metrics/result_ablation_mbert/sst2/mBERT/hf_sequence_classifier/seed_44/eval_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"eval_accuracy": 0.8692660550458715,
|
| 4 |
+
"eval_loss": 0.42636725306510925,
|
| 5 |
+
"eval_runtime": 1.8276,
|
| 6 |
+
"eval_samples": 872,
|
| 7 |
+
"eval_samples_per_second": 477.133,
|
| 8 |
+
"eval_steps_per_second": 5.472
|
| 9 |
+
}
|
mbert_rtx6000_metrics/result_ablation_mbert/sst2/mBERT/hf_sequence_classifier/seed_44/run_command.txt
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
/usr/local/bin/python /workspace/run_glue_MHA_gated.py --model_name_or_path bert-base-multilingual-cased --dataset_name tmnam20/ViGLUE --task_name sst2 --max_seq_length 128 --do_train --do_eval --evaluation_strategy epoch --save_strategy no --logging_strategy steps --logging_steps 25 --report_to none --num_train_epochs 3 --learning_rate 3e-05 --per_device_train_batch_size 48 --per_device_eval_batch_size 96 --gradient_accumulation_steps 1 --dataloader_num_workers 1 --seed 44 --output_dir result_ablation_mbert/sst2/mBERT/hf_sequence_classifier/seed_44 --overwrite_output_dir --trust_remote_code true --bf16 --tf32 true --skip_model_save
|
mbert_rtx6000_metrics/result_ablation_mbert/sst2/mBERT/hf_sequence_classifier/seed_44/train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"total_flos": 6159320861556240.0,
|
| 4 |
+
"train_loss": 0.23417754752909922,
|
| 5 |
+
"train_runtime": 612.7269,
|
| 6 |
+
"train_samples": 67349,
|
| 7 |
+
"train_samples_per_second": 329.75,
|
| 8 |
+
"train_steps_per_second": 6.874
|
| 9 |
+
}
|
mbert_rtx6000_metrics/result_ablation_mbert/sst2/mBERT/hf_sequence_classifier/seed_44/trainer_state.json
ADDED
|
@@ -0,0 +1,1245 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 3.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 4212,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.017806267806267807,
|
| 13 |
+
"grad_norm": 4.799685001373291,
|
| 14 |
+
"learning_rate": 2.9821937321937322e-05,
|
| 15 |
+
"loss": 0.6357,
|
| 16 |
+
"step": 25
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.03561253561253561,
|
| 20 |
+
"grad_norm": 10.083998680114746,
|
| 21 |
+
"learning_rate": 2.9643874643874644e-05,
|
| 22 |
+
"loss": 0.5448,
|
| 23 |
+
"step": 50
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.053418803418803416,
|
| 27 |
+
"grad_norm": 9.5925874710083,
|
| 28 |
+
"learning_rate": 2.9465811965811966e-05,
|
| 29 |
+
"loss": 0.5227,
|
| 30 |
+
"step": 75
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.07122507122507123,
|
| 34 |
+
"grad_norm": 5.56010103225708,
|
| 35 |
+
"learning_rate": 2.9287749287749288e-05,
|
| 36 |
+
"loss": 0.4906,
|
| 37 |
+
"step": 100
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.08903133903133903,
|
| 41 |
+
"grad_norm": 6.422469139099121,
|
| 42 |
+
"learning_rate": 2.910968660968661e-05,
|
| 43 |
+
"loss": 0.462,
|
| 44 |
+
"step": 125
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.10683760683760683,
|
| 48 |
+
"grad_norm": 5.7629475593566895,
|
| 49 |
+
"learning_rate": 2.893162393162393e-05,
|
| 50 |
+
"loss": 0.4122,
|
| 51 |
+
"step": 150
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.12464387464387465,
|
| 55 |
+
"grad_norm": 5.757438659667969,
|
| 56 |
+
"learning_rate": 2.8753561253561256e-05,
|
| 57 |
+
"loss": 0.4388,
|
| 58 |
+
"step": 175
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.14245014245014245,
|
| 62 |
+
"grad_norm": 5.925106525421143,
|
| 63 |
+
"learning_rate": 2.8575498575498578e-05,
|
| 64 |
+
"loss": 0.4154,
|
| 65 |
+
"step": 200
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.16025641025641027,
|
| 69 |
+
"grad_norm": 7.59193754196167,
|
| 70 |
+
"learning_rate": 2.8397435897435896e-05,
|
| 71 |
+
"loss": 0.4253,
|
| 72 |
+
"step": 225
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.17806267806267806,
|
| 76 |
+
"grad_norm": 9.617377281188965,
|
| 77 |
+
"learning_rate": 2.821937321937322e-05,
|
| 78 |
+
"loss": 0.4131,
|
| 79 |
+
"step": 250
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.19586894586894588,
|
| 83 |
+
"grad_norm": 5.792850017547607,
|
| 84 |
+
"learning_rate": 2.8041310541310543e-05,
|
| 85 |
+
"loss": 0.3923,
|
| 86 |
+
"step": 275
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.21367521367521367,
|
| 90 |
+
"grad_norm": 6.110167503356934,
|
| 91 |
+
"learning_rate": 2.786324786324786e-05,
|
| 92 |
+
"loss": 0.4082,
|
| 93 |
+
"step": 300
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.23148148148148148,
|
| 97 |
+
"grad_norm": 8.889596939086914,
|
| 98 |
+
"learning_rate": 2.7685185185185186e-05,
|
| 99 |
+
"loss": 0.3614,
|
| 100 |
+
"step": 325
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.2492877492877493,
|
| 104 |
+
"grad_norm": 5.16226863861084,
|
| 105 |
+
"learning_rate": 2.7507122507122508e-05,
|
| 106 |
+
"loss": 0.3748,
|
| 107 |
+
"step": 350
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.2670940170940171,
|
| 111 |
+
"grad_norm": 7.106471538543701,
|
| 112 |
+
"learning_rate": 2.732905982905983e-05,
|
| 113 |
+
"loss": 0.3799,
|
| 114 |
+
"step": 375
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.2849002849002849,
|
| 118 |
+
"grad_norm": 4.715440273284912,
|
| 119 |
+
"learning_rate": 2.715099715099715e-05,
|
| 120 |
+
"loss": 0.3615,
|
| 121 |
+
"step": 400
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.3027065527065527,
|
| 125 |
+
"grad_norm": 5.503936767578125,
|
| 126 |
+
"learning_rate": 2.6972934472934473e-05,
|
| 127 |
+
"loss": 0.3736,
|
| 128 |
+
"step": 425
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.32051282051282054,
|
| 132 |
+
"grad_norm": 4.505261421203613,
|
| 133 |
+
"learning_rate": 2.67948717948718e-05,
|
| 134 |
+
"loss": 0.3648,
|
| 135 |
+
"step": 450
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.3383190883190883,
|
| 139 |
+
"grad_norm": 10.338563919067383,
|
| 140 |
+
"learning_rate": 2.6616809116809117e-05,
|
| 141 |
+
"loss": 0.363,
|
| 142 |
+
"step": 475
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.3561253561253561,
|
| 146 |
+
"grad_norm": 6.767603874206543,
|
| 147 |
+
"learning_rate": 2.643874643874644e-05,
|
| 148 |
+
"loss": 0.3484,
|
| 149 |
+
"step": 500
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.37393162393162394,
|
| 153 |
+
"grad_norm": 5.122454643249512,
|
| 154 |
+
"learning_rate": 2.6260683760683763e-05,
|
| 155 |
+
"loss": 0.3683,
|
| 156 |
+
"step": 525
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.39173789173789175,
|
| 160 |
+
"grad_norm": 5.724037170410156,
|
| 161 |
+
"learning_rate": 2.6082621082621082e-05,
|
| 162 |
+
"loss": 0.3767,
|
| 163 |
+
"step": 550
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.40954415954415957,
|
| 167 |
+
"grad_norm": 5.562438488006592,
|
| 168 |
+
"learning_rate": 2.5904558404558404e-05,
|
| 169 |
+
"loss": 0.3298,
|
| 170 |
+
"step": 575
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.42735042735042733,
|
| 174 |
+
"grad_norm": 4.359918594360352,
|
| 175 |
+
"learning_rate": 2.572649572649573e-05,
|
| 176 |
+
"loss": 0.3478,
|
| 177 |
+
"step": 600
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.44515669515669515,
|
| 181 |
+
"grad_norm": 8.671807289123535,
|
| 182 |
+
"learning_rate": 2.5548433048433047e-05,
|
| 183 |
+
"loss": 0.3687,
|
| 184 |
+
"step": 625
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.46296296296296297,
|
| 188 |
+
"grad_norm": 7.157382488250732,
|
| 189 |
+
"learning_rate": 2.5370370370370372e-05,
|
| 190 |
+
"loss": 0.3227,
|
| 191 |
+
"step": 650
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.4807692307692308,
|
| 195 |
+
"grad_norm": 13.883027076721191,
|
| 196 |
+
"learning_rate": 2.5192307692307694e-05,
|
| 197 |
+
"loss": 0.3201,
|
| 198 |
+
"step": 675
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.4985754985754986,
|
| 202 |
+
"grad_norm": 4.598976135253906,
|
| 203 |
+
"learning_rate": 2.5014245014245015e-05,
|
| 204 |
+
"loss": 0.3103,
|
| 205 |
+
"step": 700
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.5163817663817664,
|
| 209 |
+
"grad_norm": 5.2030816078186035,
|
| 210 |
+
"learning_rate": 2.4836182336182337e-05,
|
| 211 |
+
"loss": 0.3326,
|
| 212 |
+
"step": 725
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.5341880341880342,
|
| 216 |
+
"grad_norm": 6.345943450927734,
|
| 217 |
+
"learning_rate": 2.465811965811966e-05,
|
| 218 |
+
"loss": 0.3309,
|
| 219 |
+
"step": 750
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.551994301994302,
|
| 223 |
+
"grad_norm": 5.7266411781311035,
|
| 224 |
+
"learning_rate": 2.448005698005698e-05,
|
| 225 |
+
"loss": 0.3459,
|
| 226 |
+
"step": 775
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.5698005698005698,
|
| 230 |
+
"grad_norm": 3.579359531402588,
|
| 231 |
+
"learning_rate": 2.4301994301994302e-05,
|
| 232 |
+
"loss": 0.2899,
|
| 233 |
+
"step": 800
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.5876068376068376,
|
| 237 |
+
"grad_norm": 7.04976749420166,
|
| 238 |
+
"learning_rate": 2.4123931623931624e-05,
|
| 239 |
+
"loss": 0.345,
|
| 240 |
+
"step": 825
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.6054131054131054,
|
| 244 |
+
"grad_norm": 6.2196149826049805,
|
| 245 |
+
"learning_rate": 2.3945868945868946e-05,
|
| 246 |
+
"loss": 0.3194,
|
| 247 |
+
"step": 850
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.6232193732193733,
|
| 251 |
+
"grad_norm": 10.938206672668457,
|
| 252 |
+
"learning_rate": 2.3767806267806267e-05,
|
| 253 |
+
"loss": 0.3406,
|
| 254 |
+
"step": 875
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.6410256410256411,
|
| 258 |
+
"grad_norm": 4.341131210327148,
|
| 259 |
+
"learning_rate": 2.358974358974359e-05,
|
| 260 |
+
"loss": 0.3107,
|
| 261 |
+
"step": 900
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.6588319088319088,
|
| 265 |
+
"grad_norm": 4.434405326843262,
|
| 266 |
+
"learning_rate": 2.3411680911680914e-05,
|
| 267 |
+
"loss": 0.3342,
|
| 268 |
+
"step": 925
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.6766381766381766,
|
| 272 |
+
"grad_norm": 5.367786884307861,
|
| 273 |
+
"learning_rate": 2.3233618233618236e-05,
|
| 274 |
+
"loss": 0.3028,
|
| 275 |
+
"step": 950
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.6944444444444444,
|
| 279 |
+
"grad_norm": 4.740630626678467,
|
| 280 |
+
"learning_rate": 2.3055555555555554e-05,
|
| 281 |
+
"loss": 0.3057,
|
| 282 |
+
"step": 975
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.7122507122507122,
|
| 286 |
+
"grad_norm": 6.123144149780273,
|
| 287 |
+
"learning_rate": 2.287749287749288e-05,
|
| 288 |
+
"loss": 0.3023,
|
| 289 |
+
"step": 1000
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.73005698005698,
|
| 293 |
+
"grad_norm": 5.624679088592529,
|
| 294 |
+
"learning_rate": 2.26994301994302e-05,
|
| 295 |
+
"loss": 0.2814,
|
| 296 |
+
"step": 1025
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.7478632478632479,
|
| 300 |
+
"grad_norm": 4.4744086265563965,
|
| 301 |
+
"learning_rate": 2.252136752136752e-05,
|
| 302 |
+
"loss": 0.3135,
|
| 303 |
+
"step": 1050
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.7656695156695157,
|
| 307 |
+
"grad_norm": 3.447956085205078,
|
| 308 |
+
"learning_rate": 2.2343304843304845e-05,
|
| 309 |
+
"loss": 0.3048,
|
| 310 |
+
"step": 1075
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.7834757834757835,
|
| 314 |
+
"grad_norm": 5.444612503051758,
|
| 315 |
+
"learning_rate": 2.2165242165242166e-05,
|
| 316 |
+
"loss": 0.338,
|
| 317 |
+
"step": 1100
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 0.8012820512820513,
|
| 321 |
+
"grad_norm": 6.376465320587158,
|
| 322 |
+
"learning_rate": 2.1987179487179485e-05,
|
| 323 |
+
"loss": 0.2819,
|
| 324 |
+
"step": 1125
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.8190883190883191,
|
| 328 |
+
"grad_norm": 4.561492919921875,
|
| 329 |
+
"learning_rate": 2.180911680911681e-05,
|
| 330 |
+
"loss": 0.2862,
|
| 331 |
+
"step": 1150
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 0.8368945868945868,
|
| 335 |
+
"grad_norm": 5.228828430175781,
|
| 336 |
+
"learning_rate": 2.163105413105413e-05,
|
| 337 |
+
"loss": 0.2905,
|
| 338 |
+
"step": 1175
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.8547008547008547,
|
| 342 |
+
"grad_norm": 3.6635024547576904,
|
| 343 |
+
"learning_rate": 2.1452991452991456e-05,
|
| 344 |
+
"loss": 0.2875,
|
| 345 |
+
"step": 1200
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.8725071225071225,
|
| 349 |
+
"grad_norm": 8.664366722106934,
|
| 350 |
+
"learning_rate": 2.1274928774928775e-05,
|
| 351 |
+
"loss": 0.258,
|
| 352 |
+
"step": 1225
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 0.8903133903133903,
|
| 356 |
+
"grad_norm": 5.095256805419922,
|
| 357 |
+
"learning_rate": 2.1096866096866097e-05,
|
| 358 |
+
"loss": 0.3011,
|
| 359 |
+
"step": 1250
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 0.9081196581196581,
|
| 363 |
+
"grad_norm": 4.570535182952881,
|
| 364 |
+
"learning_rate": 2.091880341880342e-05,
|
| 365 |
+
"loss": 0.281,
|
| 366 |
+
"step": 1275
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 0.9259259259259259,
|
| 370 |
+
"grad_norm": 4.990248203277588,
|
| 371 |
+
"learning_rate": 2.074074074074074e-05,
|
| 372 |
+
"loss": 0.3078,
|
| 373 |
+
"step": 1300
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 0.9437321937321937,
|
| 377 |
+
"grad_norm": 4.986328125,
|
| 378 |
+
"learning_rate": 2.056267806267806e-05,
|
| 379 |
+
"loss": 0.2861,
|
| 380 |
+
"step": 1325
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 0.9615384615384616,
|
| 384 |
+
"grad_norm": 4.004408836364746,
|
| 385 |
+
"learning_rate": 2.0384615384615387e-05,
|
| 386 |
+
"loss": 0.2339,
|
| 387 |
+
"step": 1350
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 0.9793447293447294,
|
| 391 |
+
"grad_norm": 7.177834510803223,
|
| 392 |
+
"learning_rate": 2.0206552706552705e-05,
|
| 393 |
+
"loss": 0.2729,
|
| 394 |
+
"step": 1375
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 0.9971509971509972,
|
| 398 |
+
"grad_norm": 3.175143241882324,
|
| 399 |
+
"learning_rate": 2.002849002849003e-05,
|
| 400 |
+
"loss": 0.2788,
|
| 401 |
+
"step": 1400
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 1.0,
|
| 405 |
+
"eval_accuracy": 0.8704128440366973,
|
| 406 |
+
"eval_loss": 0.34563368558883667,
|
| 407 |
+
"eval_runtime": 3.2611,
|
| 408 |
+
"eval_samples_per_second": 267.398,
|
| 409 |
+
"eval_steps_per_second": 3.066,
|
| 410 |
+
"step": 1404
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 1.014957264957265,
|
| 414 |
+
"grad_norm": 4.211446762084961,
|
| 415 |
+
"learning_rate": 1.9850427350427352e-05,
|
| 416 |
+
"loss": 0.2315,
|
| 417 |
+
"step": 1425
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"epoch": 1.0327635327635327,
|
| 421 |
+
"grad_norm": 10.738426208496094,
|
| 422 |
+
"learning_rate": 1.9672364672364674e-05,
|
| 423 |
+
"loss": 0.2337,
|
| 424 |
+
"step": 1450
|
| 425 |
+
},
|
| 426 |
+
{
|
| 427 |
+
"epoch": 1.0505698005698005,
|
| 428 |
+
"grad_norm": 5.899084091186523,
|
| 429 |
+
"learning_rate": 1.9494301994301995e-05,
|
| 430 |
+
"loss": 0.219,
|
| 431 |
+
"step": 1475
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 1.0683760683760684,
|
| 435 |
+
"grad_norm": 7.09181022644043,
|
| 436 |
+
"learning_rate": 1.9316239316239317e-05,
|
| 437 |
+
"loss": 0.2194,
|
| 438 |
+
"step": 1500
|
| 439 |
+
},
|
| 440 |
+
{
|
| 441 |
+
"epoch": 1.0861823361823362,
|
| 442 |
+
"grad_norm": 4.535739421844482,
|
| 443 |
+
"learning_rate": 1.913817663817664e-05,
|
| 444 |
+
"loss": 0.2006,
|
| 445 |
+
"step": 1525
|
| 446 |
+
},
|
| 447 |
+
{
|
| 448 |
+
"epoch": 1.103988603988604,
|
| 449 |
+
"grad_norm": 3.318687915802002,
|
| 450 |
+
"learning_rate": 1.896011396011396e-05,
|
| 451 |
+
"loss": 0.215,
|
| 452 |
+
"step": 1550
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 1.1217948717948718,
|
| 456 |
+
"grad_norm": 6.167133331298828,
|
| 457 |
+
"learning_rate": 1.8782051282051282e-05,
|
| 458 |
+
"loss": 0.1954,
|
| 459 |
+
"step": 1575
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 1.1396011396011396,
|
| 463 |
+
"grad_norm": 5.016181468963623,
|
| 464 |
+
"learning_rate": 1.8603988603988604e-05,
|
| 465 |
+
"loss": 0.2152,
|
| 466 |
+
"step": 1600
|
| 467 |
+
},
|
| 468 |
+
{
|
| 469 |
+
"epoch": 1.1574074074074074,
|
| 470 |
+
"grad_norm": 3.163026809692383,
|
| 471 |
+
"learning_rate": 1.8425925925925926e-05,
|
| 472 |
+
"loss": 0.2162,
|
| 473 |
+
"step": 1625
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 1.1752136752136753,
|
| 477 |
+
"grad_norm": 7.48093318939209,
|
| 478 |
+
"learning_rate": 1.8247863247863247e-05,
|
| 479 |
+
"loss": 0.2168,
|
| 480 |
+
"step": 1650
|
| 481 |
+
},
|
| 482 |
+
{
|
| 483 |
+
"epoch": 1.193019943019943,
|
| 484 |
+
"grad_norm": 3.8616273403167725,
|
| 485 |
+
"learning_rate": 1.8069800569800572e-05,
|
| 486 |
+
"loss": 0.2309,
|
| 487 |
+
"step": 1675
|
| 488 |
+
},
|
| 489 |
+
{
|
| 490 |
+
"epoch": 1.210826210826211,
|
| 491 |
+
"grad_norm": 5.980706214904785,
|
| 492 |
+
"learning_rate": 1.7891737891737894e-05,
|
| 493 |
+
"loss": 0.2362,
|
| 494 |
+
"step": 1700
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 1.2286324786324787,
|
| 498 |
+
"grad_norm": 8.343663215637207,
|
| 499 |
+
"learning_rate": 1.7713675213675212e-05,
|
| 500 |
+
"loss": 0.2119,
|
| 501 |
+
"step": 1725
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 1.2464387464387465,
|
| 505 |
+
"grad_norm": 6.118024826049805,
|
| 506 |
+
"learning_rate": 1.7535612535612538e-05,
|
| 507 |
+
"loss": 0.2013,
|
| 508 |
+
"step": 1750
|
| 509 |
+
},
|
| 510 |
+
{
|
| 511 |
+
"epoch": 1.2642450142450143,
|
| 512 |
+
"grad_norm": 3.189924716949463,
|
| 513 |
+
"learning_rate": 1.735754985754986e-05,
|
| 514 |
+
"loss": 0.2242,
|
| 515 |
+
"step": 1775
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 1.282051282051282,
|
| 519 |
+
"grad_norm": 13.744040489196777,
|
| 520 |
+
"learning_rate": 1.7179487179487178e-05,
|
| 521 |
+
"loss": 0.2257,
|
| 522 |
+
"step": 1800
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 1.29985754985755,
|
| 526 |
+
"grad_norm": 4.990574359893799,
|
| 527 |
+
"learning_rate": 1.7001424501424503e-05,
|
| 528 |
+
"loss": 0.1947,
|
| 529 |
+
"step": 1825
|
| 530 |
+
},
|
| 531 |
+
{
|
| 532 |
+
"epoch": 1.3176638176638176,
|
| 533 |
+
"grad_norm": 5.4855217933654785,
|
| 534 |
+
"learning_rate": 1.6823361823361824e-05,
|
| 535 |
+
"loss": 0.2457,
|
| 536 |
+
"step": 1850
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 1.3354700854700854,
|
| 540 |
+
"grad_norm": 8.326170921325684,
|
| 541 |
+
"learning_rate": 1.6645299145299143e-05,
|
| 542 |
+
"loss": 0.2256,
|
| 543 |
+
"step": 1875
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 1.3532763532763532,
|
| 547 |
+
"grad_norm": 6.513815402984619,
|
| 548 |
+
"learning_rate": 1.6467236467236468e-05,
|
| 549 |
+
"loss": 0.2308,
|
| 550 |
+
"step": 1900
|
| 551 |
+
},
|
| 552 |
+
{
|
| 553 |
+
"epoch": 1.371082621082621,
|
| 554 |
+
"grad_norm": 6.100254058837891,
|
| 555 |
+
"learning_rate": 1.628917378917379e-05,
|
| 556 |
+
"loss": 0.2089,
|
| 557 |
+
"step": 1925
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"epoch": 1.3888888888888888,
|
| 561 |
+
"grad_norm": 5.418431758880615,
|
| 562 |
+
"learning_rate": 1.6111111111111115e-05,
|
| 563 |
+
"loss": 0.216,
|
| 564 |
+
"step": 1950
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 1.4066951566951567,
|
| 568 |
+
"grad_norm": 6.286018371582031,
|
| 569 |
+
"learning_rate": 1.5933048433048433e-05,
|
| 570 |
+
"loss": 0.2084,
|
| 571 |
+
"step": 1975
|
| 572 |
+
},
|
| 573 |
+
{
|
| 574 |
+
"epoch": 1.4245014245014245,
|
| 575 |
+
"grad_norm": 10.062414169311523,
|
| 576 |
+
"learning_rate": 1.5754985754985755e-05,
|
| 577 |
+
"loss": 0.2214,
|
| 578 |
+
"step": 2000
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 1.4423076923076923,
|
| 582 |
+
"grad_norm": 7.4862380027771,
|
| 583 |
+
"learning_rate": 1.557692307692308e-05,
|
| 584 |
+
"loss": 0.1854,
|
| 585 |
+
"step": 2025
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 1.46011396011396,
|
| 589 |
+
"grad_norm": 3.7670817375183105,
|
| 590 |
+
"learning_rate": 1.5398860398860398e-05,
|
| 591 |
+
"loss": 0.2015,
|
| 592 |
+
"step": 2050
|
| 593 |
+
},
|
| 594 |
+
{
|
| 595 |
+
"epoch": 1.477920227920228,
|
| 596 |
+
"grad_norm": 6.391080379486084,
|
| 597 |
+
"learning_rate": 1.5220797720797721e-05,
|
| 598 |
+
"loss": 0.2264,
|
| 599 |
+
"step": 2075
|
| 600 |
+
},
|
| 601 |
+
{
|
| 602 |
+
"epoch": 1.4957264957264957,
|
| 603 |
+
"grad_norm": 3.947739362716675,
|
| 604 |
+
"learning_rate": 1.5042735042735045e-05,
|
| 605 |
+
"loss": 0.225,
|
| 606 |
+
"step": 2100
|
| 607 |
+
},
|
| 608 |
+
{
|
| 609 |
+
"epoch": 1.5135327635327636,
|
| 610 |
+
"grad_norm": 7.628139495849609,
|
| 611 |
+
"learning_rate": 1.4864672364672365e-05,
|
| 612 |
+
"loss": 0.2019,
|
| 613 |
+
"step": 2125
|
| 614 |
+
},
|
| 615 |
+
{
|
| 616 |
+
"epoch": 1.5313390313390314,
|
| 617 |
+
"grad_norm": 6.7792816162109375,
|
| 618 |
+
"learning_rate": 1.4686609686609687e-05,
|
| 619 |
+
"loss": 0.2277,
|
| 620 |
+
"step": 2150
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 1.5491452991452992,
|
| 624 |
+
"grad_norm": 4.773514747619629,
|
| 625 |
+
"learning_rate": 1.450854700854701e-05,
|
| 626 |
+
"loss": 0.21,
|
| 627 |
+
"step": 2175
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"epoch": 1.566951566951567,
|
| 631 |
+
"grad_norm": 5.480628490447998,
|
| 632 |
+
"learning_rate": 1.433048433048433e-05,
|
| 633 |
+
"loss": 0.2103,
|
| 634 |
+
"step": 2200
|
| 635 |
+
},
|
| 636 |
+
{
|
| 637 |
+
"epoch": 1.5847578347578346,
|
| 638 |
+
"grad_norm": 5.608632564544678,
|
| 639 |
+
"learning_rate": 1.4152421652421652e-05,
|
| 640 |
+
"loss": 0.2114,
|
| 641 |
+
"step": 2225
|
| 642 |
+
},
|
| 643 |
+
{
|
| 644 |
+
"epoch": 1.6025641025641026,
|
| 645 |
+
"grad_norm": 5.65090799331665,
|
| 646 |
+
"learning_rate": 1.3974358974358975e-05,
|
| 647 |
+
"loss": 0.2085,
|
| 648 |
+
"step": 2250
|
| 649 |
+
},
|
| 650 |
+
{
|
| 651 |
+
"epoch": 1.6203703703703702,
|
| 652 |
+
"grad_norm": 5.235284805297852,
|
| 653 |
+
"learning_rate": 1.3796296296296297e-05,
|
| 654 |
+
"loss": 0.195,
|
| 655 |
+
"step": 2275
|
| 656 |
+
},
|
| 657 |
+
{
|
| 658 |
+
"epoch": 1.6381766381766383,
|
| 659 |
+
"grad_norm": 3.7024197578430176,
|
| 660 |
+
"learning_rate": 1.3618233618233619e-05,
|
| 661 |
+
"loss": 0.1913,
|
| 662 |
+
"step": 2300
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 1.6559829059829059,
|
| 666 |
+
"grad_norm": 7.2997307777404785,
|
| 667 |
+
"learning_rate": 1.344017094017094e-05,
|
| 668 |
+
"loss": 0.2009,
|
| 669 |
+
"step": 2325
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 1.673789173789174,
|
| 673 |
+
"grad_norm": 4.518612384796143,
|
| 674 |
+
"learning_rate": 1.3262108262108262e-05,
|
| 675 |
+
"loss": 0.1721,
|
| 676 |
+
"step": 2350
|
| 677 |
+
},
|
| 678 |
+
{
|
| 679 |
+
"epoch": 1.6915954415954415,
|
| 680 |
+
"grad_norm": 6.771218776702881,
|
| 681 |
+
"learning_rate": 1.3084045584045585e-05,
|
| 682 |
+
"loss": 0.2233,
|
| 683 |
+
"step": 2375
|
| 684 |
+
},
|
| 685 |
+
{
|
| 686 |
+
"epoch": 1.7094017094017095,
|
| 687 |
+
"grad_norm": 6.508902549743652,
|
| 688 |
+
"learning_rate": 1.2905982905982905e-05,
|
| 689 |
+
"loss": 0.2226,
|
| 690 |
+
"step": 2400
|
| 691 |
+
},
|
| 692 |
+
{
|
| 693 |
+
"epoch": 1.7272079772079771,
|
| 694 |
+
"grad_norm": 7.198089122772217,
|
| 695 |
+
"learning_rate": 1.2727920227920229e-05,
|
| 696 |
+
"loss": 0.2065,
|
| 697 |
+
"step": 2425
|
| 698 |
+
},
|
| 699 |
+
{
|
| 700 |
+
"epoch": 1.7450142450142452,
|
| 701 |
+
"grad_norm": 5.113999843597412,
|
| 702 |
+
"learning_rate": 1.254985754985755e-05,
|
| 703 |
+
"loss": 0.175,
|
| 704 |
+
"step": 2450
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 1.7628205128205128,
|
| 708 |
+
"grad_norm": 6.887495994567871,
|
| 709 |
+
"learning_rate": 1.2371794871794872e-05,
|
| 710 |
+
"loss": 0.1982,
|
| 711 |
+
"step": 2475
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"epoch": 1.7806267806267806,
|
| 715 |
+
"grad_norm": 5.467659950256348,
|
| 716 |
+
"learning_rate": 1.2193732193732194e-05,
|
| 717 |
+
"loss": 0.2304,
|
| 718 |
+
"step": 2500
|
| 719 |
+
},
|
| 720 |
+
{
|
| 721 |
+
"epoch": 1.7984330484330484,
|
| 722 |
+
"grad_norm": 7.171382904052734,
|
| 723 |
+
"learning_rate": 1.2015669515669516e-05,
|
| 724 |
+
"loss": 0.1897,
|
| 725 |
+
"step": 2525
|
| 726 |
+
},
|
| 727 |
+
{
|
| 728 |
+
"epoch": 1.8162393162393162,
|
| 729 |
+
"grad_norm": 3.2795259952545166,
|
| 730 |
+
"learning_rate": 1.1837606837606839e-05,
|
| 731 |
+
"loss": 0.1931,
|
| 732 |
+
"step": 2550
|
| 733 |
+
},
|
| 734 |
+
{
|
| 735 |
+
"epoch": 1.834045584045584,
|
| 736 |
+
"grad_norm": 2.7242727279663086,
|
| 737 |
+
"learning_rate": 1.1659544159544159e-05,
|
| 738 |
+
"loss": 0.2247,
|
| 739 |
+
"step": 2575
|
| 740 |
+
},
|
| 741 |
+
{
|
| 742 |
+
"epoch": 1.8518518518518519,
|
| 743 |
+
"grad_norm": 4.571079254150391,
|
| 744 |
+
"learning_rate": 1.148148148148148e-05,
|
| 745 |
+
"loss": 0.1953,
|
| 746 |
+
"step": 2600
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 1.8696581196581197,
|
| 750 |
+
"grad_norm": 5.116637706756592,
|
| 751 |
+
"learning_rate": 1.1303418803418804e-05,
|
| 752 |
+
"loss": 0.1801,
|
| 753 |
+
"step": 2625
|
| 754 |
+
},
|
| 755 |
+
{
|
| 756 |
+
"epoch": 1.8874643874643875,
|
| 757 |
+
"grad_norm": 3.9028632640838623,
|
| 758 |
+
"learning_rate": 1.1125356125356126e-05,
|
| 759 |
+
"loss": 0.1945,
|
| 760 |
+
"step": 2650
|
| 761 |
+
},
|
| 762 |
+
{
|
| 763 |
+
"epoch": 1.9052706552706553,
|
| 764 |
+
"grad_norm": 7.042113304138184,
|
| 765 |
+
"learning_rate": 1.0947293447293448e-05,
|
| 766 |
+
"loss": 0.1864,
|
| 767 |
+
"step": 2675
|
| 768 |
+
},
|
| 769 |
+
{
|
| 770 |
+
"epoch": 1.9230769230769231,
|
| 771 |
+
"grad_norm": 7.586813926696777,
|
| 772 |
+
"learning_rate": 1.076923076923077e-05,
|
| 773 |
+
"loss": 0.2153,
|
| 774 |
+
"step": 2700
|
| 775 |
+
},
|
| 776 |
+
{
|
| 777 |
+
"epoch": 1.940883190883191,
|
| 778 |
+
"grad_norm": 5.323196887969971,
|
| 779 |
+
"learning_rate": 1.0591168091168091e-05,
|
| 780 |
+
"loss": 0.1852,
|
| 781 |
+
"step": 2725
|
| 782 |
+
},
|
| 783 |
+
{
|
| 784 |
+
"epoch": 1.9586894586894585,
|
| 785 |
+
"grad_norm": 4.046555042266846,
|
| 786 |
+
"learning_rate": 1.0413105413105414e-05,
|
| 787 |
+
"loss": 0.2118,
|
| 788 |
+
"step": 2750
|
| 789 |
+
},
|
| 790 |
+
{
|
| 791 |
+
"epoch": 1.9764957264957266,
|
| 792 |
+
"grad_norm": 4.3773956298828125,
|
| 793 |
+
"learning_rate": 1.0235042735042734e-05,
|
| 794 |
+
"loss": 0.211,
|
| 795 |
+
"step": 2775
|
| 796 |
+
},
|
| 797 |
+
{
|
| 798 |
+
"epoch": 1.9943019943019942,
|
| 799 |
+
"grad_norm": 7.254610538482666,
|
| 800 |
+
"learning_rate": 1.0056980056980058e-05,
|
| 801 |
+
"loss": 0.2092,
|
| 802 |
+
"step": 2800
|
| 803 |
+
},
|
| 804 |
+
{
|
| 805 |
+
"epoch": 2.0,
|
| 806 |
+
"eval_accuracy": 0.8704128440366973,
|
| 807 |
+
"eval_loss": 0.36860382556915283,
|
| 808 |
+
"eval_runtime": 3.287,
|
| 809 |
+
"eval_samples_per_second": 265.287,
|
| 810 |
+
"eval_steps_per_second": 3.042,
|
| 811 |
+
"step": 2808
|
| 812 |
+
},
|
| 813 |
+
{
|
| 814 |
+
"epoch": 2.012108262108262,
|
| 815 |
+
"grad_norm": 6.591655731201172,
|
| 816 |
+
"learning_rate": 9.87891737891738e-06,
|
| 817 |
+
"loss": 0.159,
|
| 818 |
+
"step": 2825
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 2.02991452991453,
|
| 822 |
+
"grad_norm": 5.955344200134277,
|
| 823 |
+
"learning_rate": 9.700854700854701e-06,
|
| 824 |
+
"loss": 0.1606,
|
| 825 |
+
"step": 2850
|
| 826 |
+
},
|
| 827 |
+
{
|
| 828 |
+
"epoch": 2.047720797720798,
|
| 829 |
+
"grad_norm": 5.710184574127197,
|
| 830 |
+
"learning_rate": 9.522792022792023e-06,
|
| 831 |
+
"loss": 0.1396,
|
| 832 |
+
"step": 2875
|
| 833 |
+
},
|
| 834 |
+
{
|
| 835 |
+
"epoch": 2.0655270655270654,
|
| 836 |
+
"grad_norm": 10.179372787475586,
|
| 837 |
+
"learning_rate": 9.344729344729345e-06,
|
| 838 |
+
"loss": 0.1386,
|
| 839 |
+
"step": 2900
|
| 840 |
+
},
|
| 841 |
+
{
|
| 842 |
+
"epoch": 2.0833333333333335,
|
| 843 |
+
"grad_norm": 6.0492753982543945,
|
| 844 |
+
"learning_rate": 9.166666666666668e-06,
|
| 845 |
+
"loss": 0.1413,
|
| 846 |
+
"step": 2925
|
| 847 |
+
},
|
| 848 |
+
{
|
| 849 |
+
"epoch": 2.101139601139601,
|
| 850 |
+
"grad_norm": 2.60746431350708,
|
| 851 |
+
"learning_rate": 8.988603988603988e-06,
|
| 852 |
+
"loss": 0.1352,
|
| 853 |
+
"step": 2950
|
| 854 |
+
},
|
| 855 |
+
{
|
| 856 |
+
"epoch": 2.118945868945869,
|
| 857 |
+
"grad_norm": 7.517121315002441,
|
| 858 |
+
"learning_rate": 8.81054131054131e-06,
|
| 859 |
+
"loss": 0.1407,
|
| 860 |
+
"step": 2975
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 2.1367521367521367,
|
| 864 |
+
"grad_norm": 8.477869033813477,
|
| 865 |
+
"learning_rate": 8.632478632478633e-06,
|
| 866 |
+
"loss": 0.1217,
|
| 867 |
+
"step": 3000
|
| 868 |
+
},
|
| 869 |
+
{
|
| 870 |
+
"epoch": 2.1545584045584047,
|
| 871 |
+
"grad_norm": 7.134650707244873,
|
| 872 |
+
"learning_rate": 8.454415954415955e-06,
|
| 873 |
+
"loss": 0.1311,
|
| 874 |
+
"step": 3025
|
| 875 |
+
},
|
| 876 |
+
{
|
| 877 |
+
"epoch": 2.1723646723646723,
|
| 878 |
+
"grad_norm": 3.7906930446624756,
|
| 879 |
+
"learning_rate": 8.276353276353277e-06,
|
| 880 |
+
"loss": 0.1415,
|
| 881 |
+
"step": 3050
|
| 882 |
+
},
|
| 883 |
+
{
|
| 884 |
+
"epoch": 2.1901709401709404,
|
| 885 |
+
"grad_norm": 5.15700626373291,
|
| 886 |
+
"learning_rate": 8.098290598290598e-06,
|
| 887 |
+
"loss": 0.1554,
|
| 888 |
+
"step": 3075
|
| 889 |
+
},
|
| 890 |
+
{
|
| 891 |
+
"epoch": 2.207977207977208,
|
| 892 |
+
"grad_norm": 4.09053897857666,
|
| 893 |
+
"learning_rate": 7.92022792022792e-06,
|
| 894 |
+
"loss": 0.1742,
|
| 895 |
+
"step": 3100
|
| 896 |
+
},
|
| 897 |
+
{
|
| 898 |
+
"epoch": 2.2257834757834756,
|
| 899 |
+
"grad_norm": 3.2536919116973877,
|
| 900 |
+
"learning_rate": 7.742165242165244e-06,
|
| 901 |
+
"loss": 0.1194,
|
| 902 |
+
"step": 3125
|
| 903 |
+
},
|
| 904 |
+
{
|
| 905 |
+
"epoch": 2.2435897435897436,
|
| 906 |
+
"grad_norm": 5.1549787521362305,
|
| 907 |
+
"learning_rate": 7.564102564102564e-06,
|
| 908 |
+
"loss": 0.1189,
|
| 909 |
+
"step": 3150
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 2.261396011396011,
|
| 913 |
+
"grad_norm": 3.462674379348755,
|
| 914 |
+
"learning_rate": 7.386039886039886e-06,
|
| 915 |
+
"loss": 0.1543,
|
| 916 |
+
"step": 3175
|
| 917 |
+
},
|
| 918 |
+
{
|
| 919 |
+
"epoch": 2.2792022792022792,
|
| 920 |
+
"grad_norm": 7.184431552886963,
|
| 921 |
+
"learning_rate": 7.207977207977209e-06,
|
| 922 |
+
"loss": 0.1384,
|
| 923 |
+
"step": 3200
|
| 924 |
+
},
|
| 925 |
+
{
|
| 926 |
+
"epoch": 2.297008547008547,
|
| 927 |
+
"grad_norm": 7.542169094085693,
|
| 928 |
+
"learning_rate": 7.02991452991453e-06,
|
| 929 |
+
"loss": 0.1503,
|
| 930 |
+
"step": 3225
|
| 931 |
+
},
|
| 932 |
+
{
|
| 933 |
+
"epoch": 2.314814814814815,
|
| 934 |
+
"grad_norm": 5.781777858734131,
|
| 935 |
+
"learning_rate": 6.851851851851852e-06,
|
| 936 |
+
"loss": 0.1505,
|
| 937 |
+
"step": 3250
|
| 938 |
+
},
|
| 939 |
+
{
|
| 940 |
+
"epoch": 2.3326210826210825,
|
| 941 |
+
"grad_norm": 6.463603496551514,
|
| 942 |
+
"learning_rate": 6.673789173789174e-06,
|
| 943 |
+
"loss": 0.1273,
|
| 944 |
+
"step": 3275
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 2.3504273504273505,
|
| 948 |
+
"grad_norm": 8.750436782836914,
|
| 949 |
+
"learning_rate": 6.4957264957264955e-06,
|
| 950 |
+
"loss": 0.1599,
|
| 951 |
+
"step": 3300
|
| 952 |
+
},
|
| 953 |
+
{
|
| 954 |
+
"epoch": 2.368233618233618,
|
| 955 |
+
"grad_norm": 4.049947738647461,
|
| 956 |
+
"learning_rate": 6.317663817663818e-06,
|
| 957 |
+
"loss": 0.1247,
|
| 958 |
+
"step": 3325
|
| 959 |
+
},
|
| 960 |
+
{
|
| 961 |
+
"epoch": 2.386039886039886,
|
| 962 |
+
"grad_norm": 8.16646671295166,
|
| 963 |
+
"learning_rate": 6.13960113960114e-06,
|
| 964 |
+
"loss": 0.1263,
|
| 965 |
+
"step": 3350
|
| 966 |
+
},
|
| 967 |
+
{
|
| 968 |
+
"epoch": 2.4038461538461537,
|
| 969 |
+
"grad_norm": 7.965654373168945,
|
| 970 |
+
"learning_rate": 5.9615384615384615e-06,
|
| 971 |
+
"loss": 0.1091,
|
| 972 |
+
"step": 3375
|
| 973 |
+
},
|
| 974 |
+
{
|
| 975 |
+
"epoch": 2.421652421652422,
|
| 976 |
+
"grad_norm": 9.628610610961914,
|
| 977 |
+
"learning_rate": 5.783475783475783e-06,
|
| 978 |
+
"loss": 0.1328,
|
| 979 |
+
"step": 3400
|
| 980 |
+
},
|
| 981 |
+
{
|
| 982 |
+
"epoch": 2.4394586894586894,
|
| 983 |
+
"grad_norm": 3.573007345199585,
|
| 984 |
+
"learning_rate": 5.605413105413106e-06,
|
| 985 |
+
"loss": 0.1679,
|
| 986 |
+
"step": 3425
|
| 987 |
+
},
|
| 988 |
+
{
|
| 989 |
+
"epoch": 2.4572649572649574,
|
| 990 |
+
"grad_norm": 3.6096930503845215,
|
| 991 |
+
"learning_rate": 5.4273504273504275e-06,
|
| 992 |
+
"loss": 0.1362,
|
| 993 |
+
"step": 3450
|
| 994 |
+
},
|
| 995 |
+
{
|
| 996 |
+
"epoch": 2.475071225071225,
|
| 997 |
+
"grad_norm": 8.211902618408203,
|
| 998 |
+
"learning_rate": 5.24928774928775e-06,
|
| 999 |
+
"loss": 0.1382,
|
| 1000 |
+
"step": 3475
|
| 1001 |
+
},
|
| 1002 |
+
{
|
| 1003 |
+
"epoch": 2.492877492877493,
|
| 1004 |
+
"grad_norm": 7.721240997314453,
|
| 1005 |
+
"learning_rate": 5.071225071225071e-06,
|
| 1006 |
+
"loss": 0.1388,
|
| 1007 |
+
"step": 3500
|
| 1008 |
+
},
|
| 1009 |
+
{
|
| 1010 |
+
"epoch": 2.5106837606837606,
|
| 1011 |
+
"grad_norm": 1.8307784795761108,
|
| 1012 |
+
"learning_rate": 4.8931623931623934e-06,
|
| 1013 |
+
"loss": 0.1508,
|
| 1014 |
+
"step": 3525
|
| 1015 |
+
},
|
| 1016 |
+
{
|
| 1017 |
+
"epoch": 2.5284900284900287,
|
| 1018 |
+
"grad_norm": 2.01850962638855,
|
| 1019 |
+
"learning_rate": 4.715099715099715e-06,
|
| 1020 |
+
"loss": 0.1212,
|
| 1021 |
+
"step": 3550
|
| 1022 |
+
},
|
| 1023 |
+
{
|
| 1024 |
+
"epoch": 2.5462962962962963,
|
| 1025 |
+
"grad_norm": 4.245933532714844,
|
| 1026 |
+
"learning_rate": 4.537037037037038e-06,
|
| 1027 |
+
"loss": 0.1367,
|
| 1028 |
+
"step": 3575
|
| 1029 |
+
},
|
| 1030 |
+
{
|
| 1031 |
+
"epoch": 2.564102564102564,
|
| 1032 |
+
"grad_norm": 4.798195838928223,
|
| 1033 |
+
"learning_rate": 4.358974358974359e-06,
|
| 1034 |
+
"loss": 0.1594,
|
| 1035 |
+
"step": 3600
|
| 1036 |
+
},
|
| 1037 |
+
{
|
| 1038 |
+
"epoch": 2.581908831908832,
|
| 1039 |
+
"grad_norm": 6.1055521965026855,
|
| 1040 |
+
"learning_rate": 4.18091168091168e-06,
|
| 1041 |
+
"loss": 0.1462,
|
| 1042 |
+
"step": 3625
|
| 1043 |
+
},
|
| 1044 |
+
{
|
| 1045 |
+
"epoch": 2.5997150997151,
|
| 1046 |
+
"grad_norm": 6.343952178955078,
|
| 1047 |
+
"learning_rate": 4.002849002849003e-06,
|
| 1048 |
+
"loss": 0.1462,
|
| 1049 |
+
"step": 3650
|
| 1050 |
+
},
|
| 1051 |
+
{
|
| 1052 |
+
"epoch": 2.6175213675213675,
|
| 1053 |
+
"grad_norm": 8.123350143432617,
|
| 1054 |
+
"learning_rate": 3.8247863247863246e-06,
|
| 1055 |
+
"loss": 0.1332,
|
| 1056 |
+
"step": 3675
|
| 1057 |
+
},
|
| 1058 |
+
{
|
| 1059 |
+
"epoch": 2.635327635327635,
|
| 1060 |
+
"grad_norm": 5.625495910644531,
|
| 1061 |
+
"learning_rate": 3.646723646723647e-06,
|
| 1062 |
+
"loss": 0.1555,
|
| 1063 |
+
"step": 3700
|
| 1064 |
+
},
|
| 1065 |
+
{
|
| 1066 |
+
"epoch": 2.653133903133903,
|
| 1067 |
+
"grad_norm": 6.5732526779174805,
|
| 1068 |
+
"learning_rate": 3.468660968660969e-06,
|
| 1069 |
+
"loss": 0.1458,
|
| 1070 |
+
"step": 3725
|
| 1071 |
+
},
|
| 1072 |
+
{
|
| 1073 |
+
"epoch": 2.6709401709401708,
|
| 1074 |
+
"grad_norm": 5.222440242767334,
|
| 1075 |
+
"learning_rate": 3.290598290598291e-06,
|
| 1076 |
+
"loss": 0.1242,
|
| 1077 |
+
"step": 3750
|
| 1078 |
+
},
|
| 1079 |
+
{
|
| 1080 |
+
"epoch": 2.688746438746439,
|
| 1081 |
+
"grad_norm": 5.1020426750183105,
|
| 1082 |
+
"learning_rate": 3.1125356125356127e-06,
|
| 1083 |
+
"loss": 0.1407,
|
| 1084 |
+
"step": 3775
|
| 1085 |
+
},
|
| 1086 |
+
{
|
| 1087 |
+
"epoch": 2.7065527065527064,
|
| 1088 |
+
"grad_norm": 7.731722831726074,
|
| 1089 |
+
"learning_rate": 2.9344729344729344e-06,
|
| 1090 |
+
"loss": 0.1463,
|
| 1091 |
+
"step": 3800
|
| 1092 |
+
},
|
| 1093 |
+
{
|
| 1094 |
+
"epoch": 2.7243589743589745,
|
| 1095 |
+
"grad_norm": 4.011707782745361,
|
| 1096 |
+
"learning_rate": 2.7564102564102565e-06,
|
| 1097 |
+
"loss": 0.1314,
|
| 1098 |
+
"step": 3825
|
| 1099 |
+
},
|
| 1100 |
+
{
|
| 1101 |
+
"epoch": 2.742165242165242,
|
| 1102 |
+
"grad_norm": 6.288882732391357,
|
| 1103 |
+
"learning_rate": 2.5783475783475782e-06,
|
| 1104 |
+
"loss": 0.1211,
|
| 1105 |
+
"step": 3850
|
| 1106 |
+
},
|
| 1107 |
+
{
|
| 1108 |
+
"epoch": 2.75997150997151,
|
| 1109 |
+
"grad_norm": 5.277860164642334,
|
| 1110 |
+
"learning_rate": 2.4002849002849004e-06,
|
| 1111 |
+
"loss": 0.1565,
|
| 1112 |
+
"step": 3875
|
| 1113 |
+
},
|
| 1114 |
+
{
|
| 1115 |
+
"epoch": 2.7777777777777777,
|
| 1116 |
+
"grad_norm": 4.074717998504639,
|
| 1117 |
+
"learning_rate": 2.222222222222222e-06,
|
| 1118 |
+
"loss": 0.1376,
|
| 1119 |
+
"step": 3900
|
| 1120 |
+
},
|
| 1121 |
+
{
|
| 1122 |
+
"epoch": 2.7955840455840457,
|
| 1123 |
+
"grad_norm": 8.561663627624512,
|
| 1124 |
+
"learning_rate": 2.044159544159544e-06,
|
| 1125 |
+
"loss": 0.1464,
|
| 1126 |
+
"step": 3925
|
| 1127 |
+
},
|
| 1128 |
+
{
|
| 1129 |
+
"epoch": 2.8133903133903133,
|
| 1130 |
+
"grad_norm": 7.566883563995361,
|
| 1131 |
+
"learning_rate": 1.8660968660968661e-06,
|
| 1132 |
+
"loss": 0.1296,
|
| 1133 |
+
"step": 3950
|
| 1134 |
+
},
|
| 1135 |
+
{
|
| 1136 |
+
"epoch": 2.8311965811965814,
|
| 1137 |
+
"grad_norm": 5.515147686004639,
|
| 1138 |
+
"learning_rate": 1.688034188034188e-06,
|
| 1139 |
+
"loss": 0.1385,
|
| 1140 |
+
"step": 3975
|
| 1141 |
+
},
|
| 1142 |
+
{
|
| 1143 |
+
"epoch": 2.849002849002849,
|
| 1144 |
+
"grad_norm": 8.940464973449707,
|
| 1145 |
+
"learning_rate": 1.50997150997151e-06,
|
| 1146 |
+
"loss": 0.1678,
|
| 1147 |
+
"step": 4000
|
| 1148 |
+
},
|
| 1149 |
+
{
|
| 1150 |
+
"epoch": 2.866809116809117,
|
| 1151 |
+
"grad_norm": 8.116877555847168,
|
| 1152 |
+
"learning_rate": 1.3319088319088321e-06,
|
| 1153 |
+
"loss": 0.1202,
|
| 1154 |
+
"step": 4025
|
| 1155 |
+
},
|
| 1156 |
+
{
|
| 1157 |
+
"epoch": 2.8846153846153846,
|
| 1158 |
+
"grad_norm": 3.568586826324463,
|
| 1159 |
+
"learning_rate": 1.153846153846154e-06,
|
| 1160 |
+
"loss": 0.1306,
|
| 1161 |
+
"step": 4050
|
| 1162 |
+
},
|
| 1163 |
+
{
|
| 1164 |
+
"epoch": 2.902421652421652,
|
| 1165 |
+
"grad_norm": 5.682431697845459,
|
| 1166 |
+
"learning_rate": 9.757834757834757e-07,
|
| 1167 |
+
"loss": 0.169,
|
| 1168 |
+
"step": 4075
|
| 1169 |
+
},
|
| 1170 |
+
{
|
| 1171 |
+
"epoch": 2.92022792022792,
|
| 1172 |
+
"grad_norm": 5.389318466186523,
|
| 1173 |
+
"learning_rate": 7.977207977207978e-07,
|
| 1174 |
+
"loss": 0.148,
|
| 1175 |
+
"step": 4100
|
| 1176 |
+
},
|
| 1177 |
+
{
|
| 1178 |
+
"epoch": 2.9380341880341883,
|
| 1179 |
+
"grad_norm": 7.532285213470459,
|
| 1180 |
+
"learning_rate": 6.196581196581196e-07,
|
| 1181 |
+
"loss": 0.1587,
|
| 1182 |
+
"step": 4125
|
| 1183 |
+
},
|
| 1184 |
+
{
|
| 1185 |
+
"epoch": 2.955840455840456,
|
| 1186 |
+
"grad_norm": 8.695751190185547,
|
| 1187 |
+
"learning_rate": 4.415954415954416e-07,
|
| 1188 |
+
"loss": 0.1467,
|
| 1189 |
+
"step": 4150
|
| 1190 |
+
},
|
| 1191 |
+
{
|
| 1192 |
+
"epoch": 2.9736467236467234,
|
| 1193 |
+
"grad_norm": 9.546716690063477,
|
| 1194 |
+
"learning_rate": 2.635327635327635e-07,
|
| 1195 |
+
"loss": 0.1347,
|
| 1196 |
+
"step": 4175
|
| 1197 |
+
},
|
| 1198 |
+
{
|
| 1199 |
+
"epoch": 2.9914529914529915,
|
| 1200 |
+
"grad_norm": 2.956402063369751,
|
| 1201 |
+
"learning_rate": 8.547008547008547e-08,
|
| 1202 |
+
"loss": 0.1293,
|
| 1203 |
+
"step": 4200
|
| 1204 |
+
},
|
| 1205 |
+
{
|
| 1206 |
+
"epoch": 3.0,
|
| 1207 |
+
"eval_accuracy": 0.8692660550458715,
|
| 1208 |
+
"eval_loss": 0.42636725306510925,
|
| 1209 |
+
"eval_runtime": 1.8575,
|
| 1210 |
+
"eval_samples_per_second": 469.444,
|
| 1211 |
+
"eval_steps_per_second": 5.384,
|
| 1212 |
+
"step": 4212
|
| 1213 |
+
},
|
| 1214 |
+
{
|
| 1215 |
+
"epoch": 3.0,
|
| 1216 |
+
"step": 4212,
|
| 1217 |
+
"total_flos": 6159320861556240.0,
|
| 1218 |
+
"train_loss": 0.23417754752909922,
|
| 1219 |
+
"train_runtime": 612.7269,
|
| 1220 |
+
"train_samples_per_second": 329.75,
|
| 1221 |
+
"train_steps_per_second": 6.874
|
| 1222 |
+
}
|
| 1223 |
+
],
|
| 1224 |
+
"logging_steps": 25,
|
| 1225 |
+
"max_steps": 4212,
|
| 1226 |
+
"num_input_tokens_seen": 0,
|
| 1227 |
+
"num_train_epochs": 3,
|
| 1228 |
+
"save_steps": 500,
|
| 1229 |
+
"stateful_callbacks": {
|
| 1230 |
+
"TrainerControl": {
|
| 1231 |
+
"args": {
|
| 1232 |
+
"should_epoch_stop": false,
|
| 1233 |
+
"should_evaluate": false,
|
| 1234 |
+
"should_log": false,
|
| 1235 |
+
"should_save": false,
|
| 1236 |
+
"should_training_stop": false
|
| 1237 |
+
},
|
| 1238 |
+
"attributes": {}
|
| 1239 |
+
}
|
| 1240 |
+
},
|
| 1241 |
+
"total_flos": 6159320861556240.0,
|
| 1242 |
+
"train_batch_size": 48,
|
| 1243 |
+
"trial_name": null,
|
| 1244 |
+
"trial_params": null
|
| 1245 |
+
}
|