Training checkpoint at step 2000
Browse files- trainer_state.json +1886 -6
trainer_state.json
CHANGED
|
@@ -1,10 +1,10 @@
|
|
| 1 |
{
|
| 2 |
-
"best_global_step":
|
| 3 |
-
"best_metric": 2.
|
| 4 |
-
"best_model_checkpoint": "../runs/karpathy/fineweb-edu-100b-shuffle/meta-llama/Llama-3.2-1B/linear_adamw_wd1e-03_7x1024_mem32_bs64_hf_armt_dmem64/run_20/checkpoint-
|
| 5 |
-
"epoch": 0.
|
| 6 |
"eval_steps": 5,
|
| 7 |
-
"global_step":
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
@@ -1888,6 +1888,1886 @@
|
|
| 1888 |
"eval_samples_per_second": 3.464,
|
| 1889 |
"eval_steps_per_second": 1.747,
|
| 1890 |
"step": 1000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1891 |
}
|
| 1892 |
],
|
| 1893 |
"logging_steps": 25,
|
|
@@ -1907,7 +3787,7 @@
|
|
| 1907 |
"attributes": {}
|
| 1908 |
}
|
| 1909 |
},
|
| 1910 |
-
"total_flos":
|
| 1911 |
"train_batch_size": 1,
|
| 1912 |
"trial_name": null,
|
| 1913 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_global_step": 2000,
|
| 3 |
+
"best_metric": 2.449084520339966,
|
| 4 |
+
"best_model_checkpoint": "../runs/karpathy/fineweb-edu-100b-shuffle/meta-llama/Llama-3.2-1B/linear_adamw_wd1e-03_7x1024_mem32_bs64_hf_armt_dmem64/run_20/checkpoint-2000",
|
| 5 |
+
"epoch": 0.04,
|
| 6 |
"eval_steps": 5,
|
| 7 |
+
"global_step": 2000,
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
|
|
| 1888 |
"eval_samples_per_second": 3.464,
|
| 1889 |
"eval_steps_per_second": 1.747,
|
| 1890 |
"step": 1000
|
| 1891 |
+
},
|
| 1892 |
+
{
|
| 1893 |
+
"epoch": 0.0201,
|
| 1894 |
+
"eval_loss": 2.4881434440612793,
|
| 1895 |
+
"eval_runtime": 33.6922,
|
| 1896 |
+
"eval_samples_per_second": 3.473,
|
| 1897 |
+
"eval_steps_per_second": 1.751,
|
| 1898 |
+
"step": 1005
|
| 1899 |
+
},
|
| 1900 |
+
{
|
| 1901 |
+
"epoch": 0.0202,
|
| 1902 |
+
"eval_loss": 2.4879722595214844,
|
| 1903 |
+
"eval_runtime": 33.6857,
|
| 1904 |
+
"eval_samples_per_second": 3.473,
|
| 1905 |
+
"eval_steps_per_second": 1.751,
|
| 1906 |
+
"step": 1010
|
| 1907 |
+
},
|
| 1908 |
+
{
|
| 1909 |
+
"epoch": 0.0203,
|
| 1910 |
+
"eval_loss": 2.4876134395599365,
|
| 1911 |
+
"eval_runtime": 33.7945,
|
| 1912 |
+
"eval_samples_per_second": 3.462,
|
| 1913 |
+
"eval_steps_per_second": 1.746,
|
| 1914 |
+
"step": 1015
|
| 1915 |
+
},
|
| 1916 |
+
{
|
| 1917 |
+
"epoch": 0.0204,
|
| 1918 |
+
"eval_loss": 2.4872164726257324,
|
| 1919 |
+
"eval_runtime": 33.7811,
|
| 1920 |
+
"eval_samples_per_second": 3.463,
|
| 1921 |
+
"eval_steps_per_second": 1.747,
|
| 1922 |
+
"step": 1020
|
| 1923 |
+
},
|
| 1924 |
+
{
|
| 1925 |
+
"epoch": 0.0205,
|
| 1926 |
+
"grad_norm": 0.04204734602618554,
|
| 1927 |
+
"learning_rate": 2.048e-06,
|
| 1928 |
+
"loss": 2.4708,
|
| 1929 |
+
"step": 1025
|
| 1930 |
+
},
|
| 1931 |
+
{
|
| 1932 |
+
"epoch": 0.0205,
|
| 1933 |
+
"eval_loss": 2.48695707321167,
|
| 1934 |
+
"eval_runtime": 33.821,
|
| 1935 |
+
"eval_samples_per_second": 3.459,
|
| 1936 |
+
"eval_steps_per_second": 1.744,
|
| 1937 |
+
"step": 1025
|
| 1938 |
+
},
|
| 1939 |
+
{
|
| 1940 |
+
"epoch": 0.0206,
|
| 1941 |
+
"eval_loss": 2.486564874649048,
|
| 1942 |
+
"eval_runtime": 33.82,
|
| 1943 |
+
"eval_samples_per_second": 3.459,
|
| 1944 |
+
"eval_steps_per_second": 1.745,
|
| 1945 |
+
"step": 1030
|
| 1946 |
+
},
|
| 1947 |
+
{
|
| 1948 |
+
"epoch": 0.0207,
|
| 1949 |
+
"eval_loss": 2.486281633377075,
|
| 1950 |
+
"eval_runtime": 33.927,
|
| 1951 |
+
"eval_samples_per_second": 3.449,
|
| 1952 |
+
"eval_steps_per_second": 1.739,
|
| 1953 |
+
"step": 1035
|
| 1954 |
+
},
|
| 1955 |
+
{
|
| 1956 |
+
"epoch": 0.0208,
|
| 1957 |
+
"eval_loss": 2.4860103130340576,
|
| 1958 |
+
"eval_runtime": 33.9697,
|
| 1959 |
+
"eval_samples_per_second": 3.444,
|
| 1960 |
+
"eval_steps_per_second": 1.737,
|
| 1961 |
+
"step": 1040
|
| 1962 |
+
},
|
| 1963 |
+
{
|
| 1964 |
+
"epoch": 0.0209,
|
| 1965 |
+
"eval_loss": 2.4855759143829346,
|
| 1966 |
+
"eval_runtime": 33.9097,
|
| 1967 |
+
"eval_samples_per_second": 3.45,
|
| 1968 |
+
"eval_steps_per_second": 1.74,
|
| 1969 |
+
"step": 1045
|
| 1970 |
+
},
|
| 1971 |
+
{
|
| 1972 |
+
"epoch": 0.021,
|
| 1973 |
+
"grad_norm": 0.03813289834436041,
|
| 1974 |
+
"learning_rate": 2.098e-06,
|
| 1975 |
+
"loss": 2.4799,
|
| 1976 |
+
"step": 1050
|
| 1977 |
+
},
|
| 1978 |
+
{
|
| 1979 |
+
"epoch": 0.021,
|
| 1980 |
+
"eval_loss": 2.485349416732788,
|
| 1981 |
+
"eval_runtime": 34.0131,
|
| 1982 |
+
"eval_samples_per_second": 3.44,
|
| 1983 |
+
"eval_steps_per_second": 1.735,
|
| 1984 |
+
"step": 1050
|
| 1985 |
+
},
|
| 1986 |
+
{
|
| 1987 |
+
"epoch": 0.0211,
|
| 1988 |
+
"eval_loss": 2.48506498336792,
|
| 1989 |
+
"eval_runtime": 34.036,
|
| 1990 |
+
"eval_samples_per_second": 3.438,
|
| 1991 |
+
"eval_steps_per_second": 1.733,
|
| 1992 |
+
"step": 1055
|
| 1993 |
+
},
|
| 1994 |
+
{
|
| 1995 |
+
"epoch": 0.0212,
|
| 1996 |
+
"eval_loss": 2.484771966934204,
|
| 1997 |
+
"eval_runtime": 34.0842,
|
| 1998 |
+
"eval_samples_per_second": 3.433,
|
| 1999 |
+
"eval_steps_per_second": 1.731,
|
| 2000 |
+
"step": 1060
|
| 2001 |
+
},
|
| 2002 |
+
{
|
| 2003 |
+
"epoch": 0.0213,
|
| 2004 |
+
"eval_loss": 2.4846508502960205,
|
| 2005 |
+
"eval_runtime": 34.0289,
|
| 2006 |
+
"eval_samples_per_second": 3.438,
|
| 2007 |
+
"eval_steps_per_second": 1.734,
|
| 2008 |
+
"step": 1065
|
| 2009 |
+
},
|
| 2010 |
+
{
|
| 2011 |
+
"epoch": 0.0214,
|
| 2012 |
+
"eval_loss": 2.484158992767334,
|
| 2013 |
+
"eval_runtime": 34.0038,
|
| 2014 |
+
"eval_samples_per_second": 3.441,
|
| 2015 |
+
"eval_steps_per_second": 1.735,
|
| 2016 |
+
"step": 1070
|
| 2017 |
+
},
|
| 2018 |
+
{
|
| 2019 |
+
"epoch": 0.0215,
|
| 2020 |
+
"grad_norm": 0.04289680570208033,
|
| 2021 |
+
"learning_rate": 2.148e-06,
|
| 2022 |
+
"loss": 2.4822,
|
| 2023 |
+
"step": 1075
|
| 2024 |
+
},
|
| 2025 |
+
{
|
| 2026 |
+
"epoch": 0.0215,
|
| 2027 |
+
"eval_loss": 2.483947992324829,
|
| 2028 |
+
"eval_runtime": 33.9604,
|
| 2029 |
+
"eval_samples_per_second": 3.445,
|
| 2030 |
+
"eval_steps_per_second": 1.737,
|
| 2031 |
+
"step": 1075
|
| 2032 |
+
},
|
| 2033 |
+
{
|
| 2034 |
+
"epoch": 0.0216,
|
| 2035 |
+
"eval_loss": 2.4836008548736572,
|
| 2036 |
+
"eval_runtime": 33.9465,
|
| 2037 |
+
"eval_samples_per_second": 3.447,
|
| 2038 |
+
"eval_steps_per_second": 1.738,
|
| 2039 |
+
"step": 1080
|
| 2040 |
+
},
|
| 2041 |
+
{
|
| 2042 |
+
"epoch": 0.0217,
|
| 2043 |
+
"eval_loss": 2.483187675476074,
|
| 2044 |
+
"eval_runtime": 34.1344,
|
| 2045 |
+
"eval_samples_per_second": 3.428,
|
| 2046 |
+
"eval_steps_per_second": 1.728,
|
| 2047 |
+
"step": 1085
|
| 2048 |
+
},
|
| 2049 |
+
{
|
| 2050 |
+
"epoch": 0.0218,
|
| 2051 |
+
"eval_loss": 2.4829964637756348,
|
| 2052 |
+
"eval_runtime": 34.0915,
|
| 2053 |
+
"eval_samples_per_second": 3.432,
|
| 2054 |
+
"eval_steps_per_second": 1.731,
|
| 2055 |
+
"step": 1090
|
| 2056 |
+
},
|
| 2057 |
+
{
|
| 2058 |
+
"epoch": 0.0219,
|
| 2059 |
+
"eval_loss": 2.482805013656616,
|
| 2060 |
+
"eval_runtime": 33.9291,
|
| 2061 |
+
"eval_samples_per_second": 3.448,
|
| 2062 |
+
"eval_steps_per_second": 1.739,
|
| 2063 |
+
"step": 1095
|
| 2064 |
+
},
|
| 2065 |
+
{
|
| 2066 |
+
"epoch": 0.022,
|
| 2067 |
+
"grad_norm": 0.03972633299982532,
|
| 2068 |
+
"learning_rate": 2.198e-06,
|
| 2069 |
+
"loss": 2.4871,
|
| 2070 |
+
"step": 1100
|
| 2071 |
+
},
|
| 2072 |
+
{
|
| 2073 |
+
"epoch": 0.022,
|
| 2074 |
+
"eval_loss": 2.482428550720215,
|
| 2075 |
+
"eval_runtime": 33.7324,
|
| 2076 |
+
"eval_samples_per_second": 3.468,
|
| 2077 |
+
"eval_steps_per_second": 1.749,
|
| 2078 |
+
"step": 1100
|
| 2079 |
+
},
|
| 2080 |
+
{
|
| 2081 |
+
"epoch": 0.0221,
|
| 2082 |
+
"eval_loss": 2.4822213649749756,
|
| 2083 |
+
"eval_runtime": 33.7954,
|
| 2084 |
+
"eval_samples_per_second": 3.462,
|
| 2085 |
+
"eval_steps_per_second": 1.746,
|
| 2086 |
+
"step": 1105
|
| 2087 |
+
},
|
| 2088 |
+
{
|
| 2089 |
+
"epoch": 0.0222,
|
| 2090 |
+
"eval_loss": 2.481689214706421,
|
| 2091 |
+
"eval_runtime": 33.7787,
|
| 2092 |
+
"eval_samples_per_second": 3.464,
|
| 2093 |
+
"eval_steps_per_second": 1.747,
|
| 2094 |
+
"step": 1110
|
| 2095 |
+
},
|
| 2096 |
+
{
|
| 2097 |
+
"epoch": 0.0223,
|
| 2098 |
+
"eval_loss": 2.481731414794922,
|
| 2099 |
+
"eval_runtime": 33.6129,
|
| 2100 |
+
"eval_samples_per_second": 3.481,
|
| 2101 |
+
"eval_steps_per_second": 1.755,
|
| 2102 |
+
"step": 1115
|
| 2103 |
+
},
|
| 2104 |
+
{
|
| 2105 |
+
"epoch": 0.0224,
|
| 2106 |
+
"eval_loss": 2.4812448024749756,
|
| 2107 |
+
"eval_runtime": 33.511,
|
| 2108 |
+
"eval_samples_per_second": 3.491,
|
| 2109 |
+
"eval_steps_per_second": 1.761,
|
| 2110 |
+
"step": 1120
|
| 2111 |
+
},
|
| 2112 |
+
{
|
| 2113 |
+
"epoch": 0.0225,
|
| 2114 |
+
"grad_norm": 0.041792864961431496,
|
| 2115 |
+
"learning_rate": 2.2480000000000003e-06,
|
| 2116 |
+
"loss": 2.4766,
|
| 2117 |
+
"step": 1125
|
| 2118 |
+
},
|
| 2119 |
+
{
|
| 2120 |
+
"epoch": 0.0225,
|
| 2121 |
+
"eval_loss": 2.4809837341308594,
|
| 2122 |
+
"eval_runtime": 33.7009,
|
| 2123 |
+
"eval_samples_per_second": 3.472,
|
| 2124 |
+
"eval_steps_per_second": 1.751,
|
| 2125 |
+
"step": 1125
|
| 2126 |
+
},
|
| 2127 |
+
{
|
| 2128 |
+
"epoch": 0.0226,
|
| 2129 |
+
"eval_loss": 2.480768918991089,
|
| 2130 |
+
"eval_runtime": 33.6615,
|
| 2131 |
+
"eval_samples_per_second": 3.476,
|
| 2132 |
+
"eval_steps_per_second": 1.753,
|
| 2133 |
+
"step": 1130
|
| 2134 |
+
},
|
| 2135 |
+
{
|
| 2136 |
+
"epoch": 0.0227,
|
| 2137 |
+
"eval_loss": 2.480337381362915,
|
| 2138 |
+
"eval_runtime": 33.6203,
|
| 2139 |
+
"eval_samples_per_second": 3.48,
|
| 2140 |
+
"eval_steps_per_second": 1.755,
|
| 2141 |
+
"step": 1135
|
| 2142 |
+
},
|
| 2143 |
+
{
|
| 2144 |
+
"epoch": 0.0228,
|
| 2145 |
+
"eval_loss": 2.4803271293640137,
|
| 2146 |
+
"eval_runtime": 33.6559,
|
| 2147 |
+
"eval_samples_per_second": 3.476,
|
| 2148 |
+
"eval_steps_per_second": 1.753,
|
| 2149 |
+
"step": 1140
|
| 2150 |
+
},
|
| 2151 |
+
{
|
| 2152 |
+
"epoch": 0.0229,
|
| 2153 |
+
"eval_loss": 2.4799482822418213,
|
| 2154 |
+
"eval_runtime": 33.5023,
|
| 2155 |
+
"eval_samples_per_second": 3.492,
|
| 2156 |
+
"eval_steps_per_second": 1.761,
|
| 2157 |
+
"step": 1145
|
| 2158 |
+
},
|
| 2159 |
+
{
|
| 2160 |
+
"epoch": 0.023,
|
| 2161 |
+
"grad_norm": 0.035383899567194975,
|
| 2162 |
+
"learning_rate": 2.2980000000000003e-06,
|
| 2163 |
+
"loss": 2.4749,
|
| 2164 |
+
"step": 1150
|
| 2165 |
+
},
|
| 2166 |
+
{
|
| 2167 |
+
"epoch": 0.023,
|
| 2168 |
+
"eval_loss": 2.479668140411377,
|
| 2169 |
+
"eval_runtime": 33.4615,
|
| 2170 |
+
"eval_samples_per_second": 3.497,
|
| 2171 |
+
"eval_steps_per_second": 1.763,
|
| 2172 |
+
"step": 1150
|
| 2173 |
+
},
|
| 2174 |
+
{
|
| 2175 |
+
"epoch": 0.0231,
|
| 2176 |
+
"eval_loss": 2.4794092178344727,
|
| 2177 |
+
"eval_runtime": 33.4264,
|
| 2178 |
+
"eval_samples_per_second": 3.5,
|
| 2179 |
+
"eval_steps_per_second": 1.765,
|
| 2180 |
+
"step": 1155
|
| 2181 |
+
},
|
| 2182 |
+
{
|
| 2183 |
+
"epoch": 0.0232,
|
| 2184 |
+
"eval_loss": 2.4790964126586914,
|
| 2185 |
+
"eval_runtime": 33.4165,
|
| 2186 |
+
"eval_samples_per_second": 3.501,
|
| 2187 |
+
"eval_steps_per_second": 1.766,
|
| 2188 |
+
"step": 1160
|
| 2189 |
+
},
|
| 2190 |
+
{
|
| 2191 |
+
"epoch": 0.0233,
|
| 2192 |
+
"eval_loss": 2.4789323806762695,
|
| 2193 |
+
"eval_runtime": 33.2576,
|
| 2194 |
+
"eval_samples_per_second": 3.518,
|
| 2195 |
+
"eval_steps_per_second": 1.774,
|
| 2196 |
+
"step": 1165
|
| 2197 |
+
},
|
| 2198 |
+
{
|
| 2199 |
+
"epoch": 0.0234,
|
| 2200 |
+
"eval_loss": 2.4786429405212402,
|
| 2201 |
+
"eval_runtime": 33.3028,
|
| 2202 |
+
"eval_samples_per_second": 3.513,
|
| 2203 |
+
"eval_steps_per_second": 1.772,
|
| 2204 |
+
"step": 1170
|
| 2205 |
+
},
|
| 2206 |
+
{
|
| 2207 |
+
"epoch": 0.0235,
|
| 2208 |
+
"grad_norm": 0.034819138532107045,
|
| 2209 |
+
"learning_rate": 2.3480000000000002e-06,
|
| 2210 |
+
"loss": 2.4874,
|
| 2211 |
+
"step": 1175
|
| 2212 |
+
},
|
| 2213 |
+
{
|
| 2214 |
+
"epoch": 0.0235,
|
| 2215 |
+
"eval_loss": 2.4784486293792725,
|
| 2216 |
+
"eval_runtime": 33.3374,
|
| 2217 |
+
"eval_samples_per_second": 3.51,
|
| 2218 |
+
"eval_steps_per_second": 1.77,
|
| 2219 |
+
"step": 1175
|
| 2220 |
+
},
|
| 2221 |
+
{
|
| 2222 |
+
"epoch": 0.0236,
|
| 2223 |
+
"eval_loss": 2.478088855743408,
|
| 2224 |
+
"eval_runtime": 33.2864,
|
| 2225 |
+
"eval_samples_per_second": 3.515,
|
| 2226 |
+
"eval_steps_per_second": 1.772,
|
| 2227 |
+
"step": 1180
|
| 2228 |
+
},
|
| 2229 |
+
{
|
| 2230 |
+
"epoch": 0.0237,
|
| 2231 |
+
"eval_loss": 2.477979898452759,
|
| 2232 |
+
"eval_runtime": 33.4245,
|
| 2233 |
+
"eval_samples_per_second": 3.5,
|
| 2234 |
+
"eval_steps_per_second": 1.765,
|
| 2235 |
+
"step": 1185
|
| 2236 |
+
},
|
| 2237 |
+
{
|
| 2238 |
+
"epoch": 0.0238,
|
| 2239 |
+
"eval_loss": 2.4778709411621094,
|
| 2240 |
+
"eval_runtime": 33.2611,
|
| 2241 |
+
"eval_samples_per_second": 3.518,
|
| 2242 |
+
"eval_steps_per_second": 1.774,
|
| 2243 |
+
"step": 1190
|
| 2244 |
+
},
|
| 2245 |
+
{
|
| 2246 |
+
"epoch": 0.0239,
|
| 2247 |
+
"eval_loss": 2.477571487426758,
|
| 2248 |
+
"eval_runtime": 33.3418,
|
| 2249 |
+
"eval_samples_per_second": 3.509,
|
| 2250 |
+
"eval_steps_per_second": 1.77,
|
| 2251 |
+
"step": 1195
|
| 2252 |
+
},
|
| 2253 |
+
{
|
| 2254 |
+
"epoch": 0.024,
|
| 2255 |
+
"grad_norm": 0.037748109041694296,
|
| 2256 |
+
"learning_rate": 2.398e-06,
|
| 2257 |
+
"loss": 2.4666,
|
| 2258 |
+
"step": 1200
|
| 2259 |
+
},
|
| 2260 |
+
{
|
| 2261 |
+
"epoch": 0.024,
|
| 2262 |
+
"eval_loss": 2.4772226810455322,
|
| 2263 |
+
"eval_runtime": 33.3603,
|
| 2264 |
+
"eval_samples_per_second": 3.507,
|
| 2265 |
+
"eval_steps_per_second": 1.769,
|
| 2266 |
+
"step": 1200
|
| 2267 |
+
},
|
| 2268 |
+
{
|
| 2269 |
+
"epoch": 0.0241,
|
| 2270 |
+
"eval_loss": 2.4769959449768066,
|
| 2271 |
+
"eval_runtime": 33.21,
|
| 2272 |
+
"eval_samples_per_second": 3.523,
|
| 2273 |
+
"eval_steps_per_second": 1.777,
|
| 2274 |
+
"step": 1205
|
| 2275 |
+
},
|
| 2276 |
+
{
|
| 2277 |
+
"epoch": 0.0242,
|
| 2278 |
+
"eval_loss": 2.4768526554107666,
|
| 2279 |
+
"eval_runtime": 33.4359,
|
| 2280 |
+
"eval_samples_per_second": 3.499,
|
| 2281 |
+
"eval_steps_per_second": 1.765,
|
| 2282 |
+
"step": 1210
|
| 2283 |
+
},
|
| 2284 |
+
{
|
| 2285 |
+
"epoch": 0.0243,
|
| 2286 |
+
"eval_loss": 2.476616382598877,
|
| 2287 |
+
"eval_runtime": 33.3341,
|
| 2288 |
+
"eval_samples_per_second": 3.51,
|
| 2289 |
+
"eval_steps_per_second": 1.77,
|
| 2290 |
+
"step": 1215
|
| 2291 |
+
},
|
| 2292 |
+
{
|
| 2293 |
+
"epoch": 0.0244,
|
| 2294 |
+
"eval_loss": 2.476250171661377,
|
| 2295 |
+
"eval_runtime": 33.3422,
|
| 2296 |
+
"eval_samples_per_second": 3.509,
|
| 2297 |
+
"eval_steps_per_second": 1.77,
|
| 2298 |
+
"step": 1220
|
| 2299 |
+
},
|
| 2300 |
+
{
|
| 2301 |
+
"epoch": 0.0245,
|
| 2302 |
+
"grad_norm": 0.042904100843004035,
|
| 2303 |
+
"learning_rate": 2.448e-06,
|
| 2304 |
+
"loss": 2.4698,
|
| 2305 |
+
"step": 1225
|
| 2306 |
+
},
|
| 2307 |
+
{
|
| 2308 |
+
"epoch": 0.0245,
|
| 2309 |
+
"eval_loss": 2.475933790206909,
|
| 2310 |
+
"eval_runtime": 33.3238,
|
| 2311 |
+
"eval_samples_per_second": 3.511,
|
| 2312 |
+
"eval_steps_per_second": 1.771,
|
| 2313 |
+
"step": 1225
|
| 2314 |
+
},
|
| 2315 |
+
{
|
| 2316 |
+
"epoch": 0.0246,
|
| 2317 |
+
"eval_loss": 2.475733995437622,
|
| 2318 |
+
"eval_runtime": 33.337,
|
| 2319 |
+
"eval_samples_per_second": 3.51,
|
| 2320 |
+
"eval_steps_per_second": 1.77,
|
| 2321 |
+
"step": 1230
|
| 2322 |
+
},
|
| 2323 |
+
{
|
| 2324 |
+
"epoch": 0.0247,
|
| 2325 |
+
"eval_loss": 2.4756155014038086,
|
| 2326 |
+
"eval_runtime": 33.3642,
|
| 2327 |
+
"eval_samples_per_second": 3.507,
|
| 2328 |
+
"eval_steps_per_second": 1.768,
|
| 2329 |
+
"step": 1235
|
| 2330 |
+
},
|
| 2331 |
+
{
|
| 2332 |
+
"epoch": 0.0248,
|
| 2333 |
+
"eval_loss": 2.475208044052124,
|
| 2334 |
+
"eval_runtime": 33.3567,
|
| 2335 |
+
"eval_samples_per_second": 3.508,
|
| 2336 |
+
"eval_steps_per_second": 1.769,
|
| 2337 |
+
"step": 1240
|
| 2338 |
+
},
|
| 2339 |
+
{
|
| 2340 |
+
"epoch": 0.0249,
|
| 2341 |
+
"eval_loss": 2.4751882553100586,
|
| 2342 |
+
"eval_runtime": 33.2409,
|
| 2343 |
+
"eval_samples_per_second": 3.52,
|
| 2344 |
+
"eval_steps_per_second": 1.775,
|
| 2345 |
+
"step": 1245
|
| 2346 |
+
},
|
| 2347 |
+
{
|
| 2348 |
+
"epoch": 0.025,
|
| 2349 |
+
"grad_norm": 0.04198064762114288,
|
| 2350 |
+
"learning_rate": 2.498e-06,
|
| 2351 |
+
"loss": 2.4544,
|
| 2352 |
+
"step": 1250
|
| 2353 |
+
},
|
| 2354 |
+
{
|
| 2355 |
+
"epoch": 0.025,
|
| 2356 |
+
"eval_loss": 2.4749433994293213,
|
| 2357 |
+
"eval_runtime": 33.219,
|
| 2358 |
+
"eval_samples_per_second": 3.522,
|
| 2359 |
+
"eval_steps_per_second": 1.776,
|
| 2360 |
+
"step": 1250
|
| 2361 |
+
},
|
| 2362 |
+
{
|
| 2363 |
+
"epoch": 0.0251,
|
| 2364 |
+
"eval_loss": 2.475109577178955,
|
| 2365 |
+
"eval_runtime": 33.293,
|
| 2366 |
+
"eval_samples_per_second": 3.514,
|
| 2367 |
+
"eval_steps_per_second": 1.772,
|
| 2368 |
+
"step": 1255
|
| 2369 |
+
},
|
| 2370 |
+
{
|
| 2371 |
+
"epoch": 0.0252,
|
| 2372 |
+
"eval_loss": 2.474750280380249,
|
| 2373 |
+
"eval_runtime": 33.5388,
|
| 2374 |
+
"eval_samples_per_second": 3.488,
|
| 2375 |
+
"eval_steps_per_second": 1.759,
|
| 2376 |
+
"step": 1260
|
| 2377 |
+
},
|
| 2378 |
+
{
|
| 2379 |
+
"epoch": 0.0253,
|
| 2380 |
+
"eval_loss": 2.4743547439575195,
|
| 2381 |
+
"eval_runtime": 33.3597,
|
| 2382 |
+
"eval_samples_per_second": 3.507,
|
| 2383 |
+
"eval_steps_per_second": 1.769,
|
| 2384 |
+
"step": 1265
|
| 2385 |
+
},
|
| 2386 |
+
{
|
| 2387 |
+
"epoch": 0.0254,
|
| 2388 |
+
"eval_loss": 2.4740777015686035,
|
| 2389 |
+
"eval_runtime": 33.3283,
|
| 2390 |
+
"eval_samples_per_second": 3.511,
|
| 2391 |
+
"eval_steps_per_second": 1.77,
|
| 2392 |
+
"step": 1270
|
| 2393 |
+
},
|
| 2394 |
+
{
|
| 2395 |
+
"epoch": 0.0255,
|
| 2396 |
+
"grad_norm": 0.03252077443949688,
|
| 2397 |
+
"learning_rate": 2.5480000000000004e-06,
|
| 2398 |
+
"loss": 2.4647,
|
| 2399 |
+
"step": 1275
|
| 2400 |
+
},
|
| 2401 |
+
{
|
| 2402 |
+
"epoch": 0.0255,
|
| 2403 |
+
"eval_loss": 2.473674774169922,
|
| 2404 |
+
"eval_runtime": 33.2492,
|
| 2405 |
+
"eval_samples_per_second": 3.519,
|
| 2406 |
+
"eval_steps_per_second": 1.774,
|
| 2407 |
+
"step": 1275
|
| 2408 |
+
},
|
| 2409 |
+
{
|
| 2410 |
+
"epoch": 0.0256,
|
| 2411 |
+
"eval_loss": 2.4734930992126465,
|
| 2412 |
+
"eval_runtime": 33.2934,
|
| 2413 |
+
"eval_samples_per_second": 3.514,
|
| 2414 |
+
"eval_steps_per_second": 1.772,
|
| 2415 |
+
"step": 1280
|
| 2416 |
+
},
|
| 2417 |
+
{
|
| 2418 |
+
"epoch": 0.0257,
|
| 2419 |
+
"eval_loss": 2.4735071659088135,
|
| 2420 |
+
"eval_runtime": 33.466,
|
| 2421 |
+
"eval_samples_per_second": 3.496,
|
| 2422 |
+
"eval_steps_per_second": 1.763,
|
| 2423 |
+
"step": 1285
|
| 2424 |
+
},
|
| 2425 |
+
{
|
| 2426 |
+
"epoch": 0.0258,
|
| 2427 |
+
"eval_loss": 2.4733572006225586,
|
| 2428 |
+
"eval_runtime": 33.248,
|
| 2429 |
+
"eval_samples_per_second": 3.519,
|
| 2430 |
+
"eval_steps_per_second": 1.775,
|
| 2431 |
+
"step": 1290
|
| 2432 |
+
},
|
| 2433 |
+
{
|
| 2434 |
+
"epoch": 0.0259,
|
| 2435 |
+
"eval_loss": 2.4730312824249268,
|
| 2436 |
+
"eval_runtime": 33.3551,
|
| 2437 |
+
"eval_samples_per_second": 3.508,
|
| 2438 |
+
"eval_steps_per_second": 1.769,
|
| 2439 |
+
"step": 1295
|
| 2440 |
+
},
|
| 2441 |
+
{
|
| 2442 |
+
"epoch": 0.026,
|
| 2443 |
+
"grad_norm": 0.034740776600877266,
|
| 2444 |
+
"learning_rate": 2.598e-06,
|
| 2445 |
+
"loss": 2.4625,
|
| 2446 |
+
"step": 1300
|
| 2447 |
+
},
|
| 2448 |
+
{
|
| 2449 |
+
"epoch": 0.026,
|
| 2450 |
+
"eval_loss": 2.4726204872131348,
|
| 2451 |
+
"eval_runtime": 33.3147,
|
| 2452 |
+
"eval_samples_per_second": 3.512,
|
| 2453 |
+
"eval_steps_per_second": 1.771,
|
| 2454 |
+
"step": 1300
|
| 2455 |
+
},
|
| 2456 |
+
{
|
| 2457 |
+
"epoch": 0.0261,
|
| 2458 |
+
"eval_loss": 2.4729621410369873,
|
| 2459 |
+
"eval_runtime": 33.3118,
|
| 2460 |
+
"eval_samples_per_second": 3.512,
|
| 2461 |
+
"eval_steps_per_second": 1.771,
|
| 2462 |
+
"step": 1305
|
| 2463 |
+
},
|
| 2464 |
+
{
|
| 2465 |
+
"epoch": 0.0262,
|
| 2466 |
+
"eval_loss": 2.4726085662841797,
|
| 2467 |
+
"eval_runtime": 33.4111,
|
| 2468 |
+
"eval_samples_per_second": 3.502,
|
| 2469 |
+
"eval_steps_per_second": 1.766,
|
| 2470 |
+
"step": 1310
|
| 2471 |
+
},
|
| 2472 |
+
{
|
| 2473 |
+
"epoch": 0.0263,
|
| 2474 |
+
"eval_loss": 2.4724133014678955,
|
| 2475 |
+
"eval_runtime": 33.3144,
|
| 2476 |
+
"eval_samples_per_second": 3.512,
|
| 2477 |
+
"eval_steps_per_second": 1.771,
|
| 2478 |
+
"step": 1315
|
| 2479 |
+
},
|
| 2480 |
+
{
|
| 2481 |
+
"epoch": 0.0264,
|
| 2482 |
+
"eval_loss": 2.471963405609131,
|
| 2483 |
+
"eval_runtime": 33.3272,
|
| 2484 |
+
"eval_samples_per_second": 3.511,
|
| 2485 |
+
"eval_steps_per_second": 1.77,
|
| 2486 |
+
"step": 1320
|
| 2487 |
+
},
|
| 2488 |
+
{
|
| 2489 |
+
"epoch": 0.0265,
|
| 2490 |
+
"grad_norm": 0.039738232523319775,
|
| 2491 |
+
"learning_rate": 2.648e-06,
|
| 2492 |
+
"loss": 2.4734,
|
| 2493 |
+
"step": 1325
|
| 2494 |
+
},
|
| 2495 |
+
{
|
| 2496 |
+
"epoch": 0.0265,
|
| 2497 |
+
"eval_loss": 2.4717814922332764,
|
| 2498 |
+
"eval_runtime": 33.2395,
|
| 2499 |
+
"eval_samples_per_second": 3.52,
|
| 2500 |
+
"eval_steps_per_second": 1.775,
|
| 2501 |
+
"step": 1325
|
| 2502 |
+
},
|
| 2503 |
+
{
|
| 2504 |
+
"epoch": 0.0266,
|
| 2505 |
+
"eval_loss": 2.471389055252075,
|
| 2506 |
+
"eval_runtime": 33.2159,
|
| 2507 |
+
"eval_samples_per_second": 3.522,
|
| 2508 |
+
"eval_steps_per_second": 1.776,
|
| 2509 |
+
"step": 1330
|
| 2510 |
+
},
|
| 2511 |
+
{
|
| 2512 |
+
"epoch": 0.0267,
|
| 2513 |
+
"eval_loss": 2.4711251258850098,
|
| 2514 |
+
"eval_runtime": 33.4193,
|
| 2515 |
+
"eval_samples_per_second": 3.501,
|
| 2516 |
+
"eval_steps_per_second": 1.765,
|
| 2517 |
+
"step": 1335
|
| 2518 |
+
},
|
| 2519 |
+
{
|
| 2520 |
+
"epoch": 0.0268,
|
| 2521 |
+
"eval_loss": 2.470979928970337,
|
| 2522 |
+
"eval_runtime": 33.2748,
|
| 2523 |
+
"eval_samples_per_second": 3.516,
|
| 2524 |
+
"eval_steps_per_second": 1.773,
|
| 2525 |
+
"step": 1340
|
| 2526 |
+
},
|
| 2527 |
+
{
|
| 2528 |
+
"epoch": 0.0269,
|
| 2529 |
+
"eval_loss": 2.4706759452819824,
|
| 2530 |
+
"eval_runtime": 33.3367,
|
| 2531 |
+
"eval_samples_per_second": 3.51,
|
| 2532 |
+
"eval_steps_per_second": 1.77,
|
| 2533 |
+
"step": 1345
|
| 2534 |
+
},
|
| 2535 |
+
{
|
| 2536 |
+
"epoch": 0.027,
|
| 2537 |
+
"grad_norm": 0.036968596903604725,
|
| 2538 |
+
"learning_rate": 2.6980000000000003e-06,
|
| 2539 |
+
"loss": 2.4642,
|
| 2540 |
+
"step": 1350
|
| 2541 |
+
},
|
| 2542 |
+
{
|
| 2543 |
+
"epoch": 0.027,
|
| 2544 |
+
"eval_loss": 2.470658302307129,
|
| 2545 |
+
"eval_runtime": 33.3288,
|
| 2546 |
+
"eval_samples_per_second": 3.51,
|
| 2547 |
+
"eval_steps_per_second": 1.77,
|
| 2548 |
+
"step": 1350
|
| 2549 |
+
},
|
| 2550 |
+
{
|
| 2551 |
+
"epoch": 0.0271,
|
| 2552 |
+
"eval_loss": 2.4704952239990234,
|
| 2553 |
+
"eval_runtime": 33.3162,
|
| 2554 |
+
"eval_samples_per_second": 3.512,
|
| 2555 |
+
"eval_steps_per_second": 1.771,
|
| 2556 |
+
"step": 1355
|
| 2557 |
+
},
|
| 2558 |
+
{
|
| 2559 |
+
"epoch": 0.0272,
|
| 2560 |
+
"eval_loss": 2.470270872116089,
|
| 2561 |
+
"eval_runtime": 33.35,
|
| 2562 |
+
"eval_samples_per_second": 3.508,
|
| 2563 |
+
"eval_steps_per_second": 1.769,
|
| 2564 |
+
"step": 1360
|
| 2565 |
+
},
|
| 2566 |
+
{
|
| 2567 |
+
"epoch": 0.0273,
|
| 2568 |
+
"eval_loss": 2.4699764251708984,
|
| 2569 |
+
"eval_runtime": 33.3696,
|
| 2570 |
+
"eval_samples_per_second": 3.506,
|
| 2571 |
+
"eval_steps_per_second": 1.768,
|
| 2572 |
+
"step": 1365
|
| 2573 |
+
},
|
| 2574 |
+
{
|
| 2575 |
+
"epoch": 0.0274,
|
| 2576 |
+
"eval_loss": 2.469688653945923,
|
| 2577 |
+
"eval_runtime": 33.4143,
|
| 2578 |
+
"eval_samples_per_second": 3.501,
|
| 2579 |
+
"eval_steps_per_second": 1.766,
|
| 2580 |
+
"step": 1370
|
| 2581 |
+
},
|
| 2582 |
+
{
|
| 2583 |
+
"epoch": 0.0275,
|
| 2584 |
+
"grad_norm": 0.03899590922475157,
|
| 2585 |
+
"learning_rate": 2.748e-06,
|
| 2586 |
+
"loss": 2.4579,
|
| 2587 |
+
"step": 1375
|
| 2588 |
+
},
|
| 2589 |
+
{
|
| 2590 |
+
"epoch": 0.0275,
|
| 2591 |
+
"eval_loss": 2.469435691833496,
|
| 2592 |
+
"eval_runtime": 33.34,
|
| 2593 |
+
"eval_samples_per_second": 3.509,
|
| 2594 |
+
"eval_steps_per_second": 1.77,
|
| 2595 |
+
"step": 1375
|
| 2596 |
+
},
|
| 2597 |
+
{
|
| 2598 |
+
"epoch": 0.0276,
|
| 2599 |
+
"eval_loss": 2.469395160675049,
|
| 2600 |
+
"eval_runtime": 33.2655,
|
| 2601 |
+
"eval_samples_per_second": 3.517,
|
| 2602 |
+
"eval_steps_per_second": 1.774,
|
| 2603 |
+
"step": 1380
|
| 2604 |
+
},
|
| 2605 |
+
{
|
| 2606 |
+
"epoch": 0.0277,
|
| 2607 |
+
"eval_loss": 2.46889328956604,
|
| 2608 |
+
"eval_runtime": 33.3344,
|
| 2609 |
+
"eval_samples_per_second": 3.51,
|
| 2610 |
+
"eval_steps_per_second": 1.77,
|
| 2611 |
+
"step": 1385
|
| 2612 |
+
},
|
| 2613 |
+
{
|
| 2614 |
+
"epoch": 0.0278,
|
| 2615 |
+
"eval_loss": 2.468695640563965,
|
| 2616 |
+
"eval_runtime": 33.4003,
|
| 2617 |
+
"eval_samples_per_second": 3.503,
|
| 2618 |
+
"eval_steps_per_second": 1.766,
|
| 2619 |
+
"step": 1390
|
| 2620 |
+
},
|
| 2621 |
+
{
|
| 2622 |
+
"epoch": 0.0279,
|
| 2623 |
+
"eval_loss": 2.4685797691345215,
|
| 2624 |
+
"eval_runtime": 33.252,
|
| 2625 |
+
"eval_samples_per_second": 3.519,
|
| 2626 |
+
"eval_steps_per_second": 1.774,
|
| 2627 |
+
"step": 1395
|
| 2628 |
+
},
|
| 2629 |
+
{
|
| 2630 |
+
"epoch": 0.028,
|
| 2631 |
+
"grad_norm": 0.03498385470366268,
|
| 2632 |
+
"learning_rate": 2.798e-06,
|
| 2633 |
+
"loss": 2.472,
|
| 2634 |
+
"step": 1400
|
| 2635 |
+
},
|
| 2636 |
+
{
|
| 2637 |
+
"epoch": 0.028,
|
| 2638 |
+
"eval_loss": 2.468594789505005,
|
| 2639 |
+
"eval_runtime": 33.5555,
|
| 2640 |
+
"eval_samples_per_second": 3.487,
|
| 2641 |
+
"eval_steps_per_second": 1.758,
|
| 2642 |
+
"step": 1400
|
| 2643 |
+
},
|
| 2644 |
+
{
|
| 2645 |
+
"epoch": 0.0281,
|
| 2646 |
+
"eval_loss": 2.4685287475585938,
|
| 2647 |
+
"eval_runtime": 33.3147,
|
| 2648 |
+
"eval_samples_per_second": 3.512,
|
| 2649 |
+
"eval_steps_per_second": 1.771,
|
| 2650 |
+
"step": 1405
|
| 2651 |
+
},
|
| 2652 |
+
{
|
| 2653 |
+
"epoch": 0.0282,
|
| 2654 |
+
"eval_loss": 2.467956304550171,
|
| 2655 |
+
"eval_runtime": 33.3679,
|
| 2656 |
+
"eval_samples_per_second": 3.506,
|
| 2657 |
+
"eval_steps_per_second": 1.768,
|
| 2658 |
+
"step": 1410
|
| 2659 |
+
},
|
| 2660 |
+
{
|
| 2661 |
+
"epoch": 0.0283,
|
| 2662 |
+
"eval_loss": 2.467761993408203,
|
| 2663 |
+
"eval_runtime": 33.3242,
|
| 2664 |
+
"eval_samples_per_second": 3.511,
|
| 2665 |
+
"eval_steps_per_second": 1.77,
|
| 2666 |
+
"step": 1415
|
| 2667 |
+
},
|
| 2668 |
+
{
|
| 2669 |
+
"epoch": 0.0284,
|
| 2670 |
+
"eval_loss": 2.467660903930664,
|
| 2671 |
+
"eval_runtime": 33.3677,
|
| 2672 |
+
"eval_samples_per_second": 3.506,
|
| 2673 |
+
"eval_steps_per_second": 1.768,
|
| 2674 |
+
"step": 1420
|
| 2675 |
+
},
|
| 2676 |
+
{
|
| 2677 |
+
"epoch": 0.0285,
|
| 2678 |
+
"grad_norm": 0.03333480906358989,
|
| 2679 |
+
"learning_rate": 2.848e-06,
|
| 2680 |
+
"loss": 2.4676,
|
| 2681 |
+
"step": 1425
|
| 2682 |
+
},
|
| 2683 |
+
{
|
| 2684 |
+
"epoch": 0.0285,
|
| 2685 |
+
"eval_loss": 2.4673027992248535,
|
| 2686 |
+
"eval_runtime": 33.3388,
|
| 2687 |
+
"eval_samples_per_second": 3.509,
|
| 2688 |
+
"eval_steps_per_second": 1.77,
|
| 2689 |
+
"step": 1425
|
| 2690 |
+
},
|
| 2691 |
+
{
|
| 2692 |
+
"epoch": 0.0286,
|
| 2693 |
+
"eval_loss": 2.467072010040283,
|
| 2694 |
+
"eval_runtime": 33.3596,
|
| 2695 |
+
"eval_samples_per_second": 3.507,
|
| 2696 |
+
"eval_steps_per_second": 1.769,
|
| 2697 |
+
"step": 1430
|
| 2698 |
+
},
|
| 2699 |
+
{
|
| 2700 |
+
"epoch": 0.0287,
|
| 2701 |
+
"eval_loss": 2.4668517112731934,
|
| 2702 |
+
"eval_runtime": 33.5136,
|
| 2703 |
+
"eval_samples_per_second": 3.491,
|
| 2704 |
+
"eval_steps_per_second": 1.76,
|
| 2705 |
+
"step": 1435
|
| 2706 |
+
},
|
| 2707 |
+
{
|
| 2708 |
+
"epoch": 0.0288,
|
| 2709 |
+
"eval_loss": 2.4666786193847656,
|
| 2710 |
+
"eval_runtime": 33.3405,
|
| 2711 |
+
"eval_samples_per_second": 3.509,
|
| 2712 |
+
"eval_steps_per_second": 1.77,
|
| 2713 |
+
"step": 1440
|
| 2714 |
+
},
|
| 2715 |
+
{
|
| 2716 |
+
"epoch": 0.0289,
|
| 2717 |
+
"eval_loss": 2.4667794704437256,
|
| 2718 |
+
"eval_runtime": 33.3333,
|
| 2719 |
+
"eval_samples_per_second": 3.51,
|
| 2720 |
+
"eval_steps_per_second": 1.77,
|
| 2721 |
+
"step": 1445
|
| 2722 |
+
},
|
| 2723 |
+
{
|
| 2724 |
+
"epoch": 0.029,
|
| 2725 |
+
"grad_norm": 0.03480548121480933,
|
| 2726 |
+
"learning_rate": 2.8980000000000005e-06,
|
| 2727 |
+
"loss": 2.4524,
|
| 2728 |
+
"step": 1450
|
| 2729 |
+
},
|
| 2730 |
+
{
|
| 2731 |
+
"epoch": 0.029,
|
| 2732 |
+
"eval_loss": 2.466280460357666,
|
| 2733 |
+
"eval_runtime": 33.4727,
|
| 2734 |
+
"eval_samples_per_second": 3.495,
|
| 2735 |
+
"eval_steps_per_second": 1.763,
|
| 2736 |
+
"step": 1450
|
| 2737 |
+
},
|
| 2738 |
+
{
|
| 2739 |
+
"epoch": 0.0291,
|
| 2740 |
+
"eval_loss": 2.4659922122955322,
|
| 2741 |
+
"eval_runtime": 33.3309,
|
| 2742 |
+
"eval_samples_per_second": 3.51,
|
| 2743 |
+
"eval_steps_per_second": 1.77,
|
| 2744 |
+
"step": 1455
|
| 2745 |
+
},
|
| 2746 |
+
{
|
| 2747 |
+
"epoch": 0.0292,
|
| 2748 |
+
"eval_loss": 2.4657278060913086,
|
| 2749 |
+
"eval_runtime": 33.326,
|
| 2750 |
+
"eval_samples_per_second": 3.511,
|
| 2751 |
+
"eval_steps_per_second": 1.77,
|
| 2752 |
+
"step": 1460
|
| 2753 |
+
},
|
| 2754 |
+
{
|
| 2755 |
+
"epoch": 0.0293,
|
| 2756 |
+
"eval_loss": 2.4654440879821777,
|
| 2757 |
+
"eval_runtime": 33.3457,
|
| 2758 |
+
"eval_samples_per_second": 3.509,
|
| 2759 |
+
"eval_steps_per_second": 1.769,
|
| 2760 |
+
"step": 1465
|
| 2761 |
+
},
|
| 2762 |
+
{
|
| 2763 |
+
"epoch": 0.0294,
|
| 2764 |
+
"eval_loss": 2.465367317199707,
|
| 2765 |
+
"eval_runtime": 33.2824,
|
| 2766 |
+
"eval_samples_per_second": 3.515,
|
| 2767 |
+
"eval_steps_per_second": 1.773,
|
| 2768 |
+
"step": 1470
|
| 2769 |
+
},
|
| 2770 |
+
{
|
| 2771 |
+
"epoch": 0.0295,
|
| 2772 |
+
"grad_norm": 0.03652712436191979,
|
| 2773 |
+
"learning_rate": 2.9480000000000004e-06,
|
| 2774 |
+
"loss": 2.466,
|
| 2775 |
+
"step": 1475
|
| 2776 |
+
},
|
| 2777 |
+
{
|
| 2778 |
+
"epoch": 0.0295,
|
| 2779 |
+
"eval_loss": 2.465318202972412,
|
| 2780 |
+
"eval_runtime": 33.3264,
|
| 2781 |
+
"eval_samples_per_second": 3.511,
|
| 2782 |
+
"eval_steps_per_second": 1.77,
|
| 2783 |
+
"step": 1475
|
| 2784 |
+
},
|
| 2785 |
+
{
|
| 2786 |
+
"epoch": 0.0296,
|
| 2787 |
+
"eval_loss": 2.465156316757202,
|
| 2788 |
+
"eval_runtime": 33.2661,
|
| 2789 |
+
"eval_samples_per_second": 3.517,
|
| 2790 |
+
"eval_steps_per_second": 1.774,
|
| 2791 |
+
"step": 1480
|
| 2792 |
+
},
|
| 2793 |
+
{
|
| 2794 |
+
"epoch": 0.0297,
|
| 2795 |
+
"eval_loss": 2.4648799896240234,
|
| 2796 |
+
"eval_runtime": 33.4782,
|
| 2797 |
+
"eval_samples_per_second": 3.495,
|
| 2798 |
+
"eval_steps_per_second": 1.762,
|
| 2799 |
+
"step": 1485
|
| 2800 |
+
},
|
| 2801 |
+
{
|
| 2802 |
+
"epoch": 0.0298,
|
| 2803 |
+
"eval_loss": 2.4646074771881104,
|
| 2804 |
+
"eval_runtime": 33.3194,
|
| 2805 |
+
"eval_samples_per_second": 3.511,
|
| 2806 |
+
"eval_steps_per_second": 1.771,
|
| 2807 |
+
"step": 1490
|
| 2808 |
+
},
|
| 2809 |
+
{
|
| 2810 |
+
"epoch": 0.0299,
|
| 2811 |
+
"eval_loss": 2.464465856552124,
|
| 2812 |
+
"eval_runtime": 33.3466,
|
| 2813 |
+
"eval_samples_per_second": 3.509,
|
| 2814 |
+
"eval_steps_per_second": 1.769,
|
| 2815 |
+
"step": 1495
|
| 2816 |
+
},
|
| 2817 |
+
{
|
| 2818 |
+
"epoch": 0.03,
|
| 2819 |
+
"grad_norm": 0.03778721361564108,
|
| 2820 |
+
"learning_rate": 2.9980000000000003e-06,
|
| 2821 |
+
"loss": 2.4684,
|
| 2822 |
+
"step": 1500
|
| 2823 |
+
},
|
| 2824 |
+
{
|
| 2825 |
+
"epoch": 0.03,
|
| 2826 |
+
"eval_loss": 2.464305877685547,
|
| 2827 |
+
"eval_runtime": 33.25,
|
| 2828 |
+
"eval_samples_per_second": 3.519,
|
| 2829 |
+
"eval_steps_per_second": 1.774,
|
| 2830 |
+
"step": 1500
|
| 2831 |
+
},
|
| 2832 |
+
{
|
| 2833 |
+
"epoch": 0.0301,
|
| 2834 |
+
"eval_loss": 2.464261531829834,
|
| 2835 |
+
"eval_runtime": 33.3761,
|
| 2836 |
+
"eval_samples_per_second": 3.505,
|
| 2837 |
+
"eval_steps_per_second": 1.768,
|
| 2838 |
+
"step": 1505
|
| 2839 |
+
},
|
| 2840 |
+
{
|
| 2841 |
+
"epoch": 0.0302,
|
| 2842 |
+
"eval_loss": 2.464185953140259,
|
| 2843 |
+
"eval_runtime": 33.4957,
|
| 2844 |
+
"eval_samples_per_second": 3.493,
|
| 2845 |
+
"eval_steps_per_second": 1.761,
|
| 2846 |
+
"step": 1510
|
| 2847 |
+
},
|
| 2848 |
+
{
|
| 2849 |
+
"epoch": 0.0303,
|
| 2850 |
+
"eval_loss": 2.4639229774475098,
|
| 2851 |
+
"eval_runtime": 33.2475,
|
| 2852 |
+
"eval_samples_per_second": 3.519,
|
| 2853 |
+
"eval_steps_per_second": 1.775,
|
| 2854 |
+
"step": 1515
|
| 2855 |
+
},
|
| 2856 |
+
{
|
| 2857 |
+
"epoch": 0.0304,
|
| 2858 |
+
"eval_loss": 2.4636595249176025,
|
| 2859 |
+
"eval_runtime": 33.3124,
|
| 2860 |
+
"eval_samples_per_second": 3.512,
|
| 2861 |
+
"eval_steps_per_second": 1.771,
|
| 2862 |
+
"step": 1520
|
| 2863 |
+
},
|
| 2864 |
+
{
|
| 2865 |
+
"epoch": 0.0305,
|
| 2866 |
+
"grad_norm": 0.035809836530372154,
|
| 2867 |
+
"learning_rate": 3.0480000000000003e-06,
|
| 2868 |
+
"loss": 2.4631,
|
| 2869 |
+
"step": 1525
|
| 2870 |
+
},
|
| 2871 |
+
{
|
| 2872 |
+
"epoch": 0.0305,
|
| 2873 |
+
"eval_loss": 2.46356201171875,
|
| 2874 |
+
"eval_runtime": 33.3423,
|
| 2875 |
+
"eval_samples_per_second": 3.509,
|
| 2876 |
+
"eval_steps_per_second": 1.77,
|
| 2877 |
+
"step": 1525
|
| 2878 |
+
},
|
| 2879 |
+
{
|
| 2880 |
+
"epoch": 0.0306,
|
| 2881 |
+
"eval_loss": 2.463318347930908,
|
| 2882 |
+
"eval_runtime": 33.3917,
|
| 2883 |
+
"eval_samples_per_second": 3.504,
|
| 2884 |
+
"eval_steps_per_second": 1.767,
|
| 2885 |
+
"step": 1530
|
| 2886 |
+
},
|
| 2887 |
+
{
|
| 2888 |
+
"epoch": 0.0307,
|
| 2889 |
+
"eval_loss": 2.4631264209747314,
|
| 2890 |
+
"eval_runtime": 33.4053,
|
| 2891 |
+
"eval_samples_per_second": 3.502,
|
| 2892 |
+
"eval_steps_per_second": 1.766,
|
| 2893 |
+
"step": 1535
|
| 2894 |
+
},
|
| 2895 |
+
{
|
| 2896 |
+
"epoch": 0.0308,
|
| 2897 |
+
"eval_loss": 2.462981700897217,
|
| 2898 |
+
"eval_runtime": 33.2608,
|
| 2899 |
+
"eval_samples_per_second": 3.518,
|
| 2900 |
+
"eval_steps_per_second": 1.774,
|
| 2901 |
+
"step": 1540
|
| 2902 |
+
},
|
| 2903 |
+
{
|
| 2904 |
+
"epoch": 0.0309,
|
| 2905 |
+
"eval_loss": 2.462719202041626,
|
| 2906 |
+
"eval_runtime": 33.3259,
|
| 2907 |
+
"eval_samples_per_second": 3.511,
|
| 2908 |
+
"eval_steps_per_second": 1.77,
|
| 2909 |
+
"step": 1545
|
| 2910 |
+
},
|
| 2911 |
+
{
|
| 2912 |
+
"epoch": 0.031,
|
| 2913 |
+
"grad_norm": 0.05979367258550731,
|
| 2914 |
+
"learning_rate": 3.0980000000000007e-06,
|
| 2915 |
+
"loss": 2.46,
|
| 2916 |
+
"step": 1550
|
| 2917 |
+
},
|
| 2918 |
+
{
|
| 2919 |
+
"epoch": 0.031,
|
| 2920 |
+
"eval_loss": 2.462733268737793,
|
| 2921 |
+
"eval_runtime": 33.3195,
|
| 2922 |
+
"eval_samples_per_second": 3.511,
|
| 2923 |
+
"eval_steps_per_second": 1.771,
|
| 2924 |
+
"step": 1550
|
| 2925 |
+
},
|
| 2926 |
+
{
|
| 2927 |
+
"epoch": 0.0311,
|
| 2928 |
+
"eval_loss": 2.4625959396362305,
|
| 2929 |
+
"eval_runtime": 33.3704,
|
| 2930 |
+
"eval_samples_per_second": 3.506,
|
| 2931 |
+
"eval_steps_per_second": 1.768,
|
| 2932 |
+
"step": 1555
|
| 2933 |
+
},
|
| 2934 |
+
{
|
| 2935 |
+
"epoch": 0.0312,
|
| 2936 |
+
"eval_loss": 2.462366819381714,
|
| 2937 |
+
"eval_runtime": 33.4047,
|
| 2938 |
+
"eval_samples_per_second": 3.503,
|
| 2939 |
+
"eval_steps_per_second": 1.766,
|
| 2940 |
+
"step": 1560
|
| 2941 |
+
},
|
| 2942 |
+
{
|
| 2943 |
+
"epoch": 0.0313,
|
| 2944 |
+
"eval_loss": 2.4618427753448486,
|
| 2945 |
+
"eval_runtime": 33.3896,
|
| 2946 |
+
"eval_samples_per_second": 3.504,
|
| 2947 |
+
"eval_steps_per_second": 1.767,
|
| 2948 |
+
"step": 1565
|
| 2949 |
+
},
|
| 2950 |
+
{
|
| 2951 |
+
"epoch": 0.0314,
|
| 2952 |
+
"eval_loss": 2.4616317749023438,
|
| 2953 |
+
"eval_runtime": 33.3414,
|
| 2954 |
+
"eval_samples_per_second": 3.509,
|
| 2955 |
+
"eval_steps_per_second": 1.77,
|
| 2956 |
+
"step": 1570
|
| 2957 |
+
},
|
| 2958 |
+
{
|
| 2959 |
+
"epoch": 0.0315,
|
| 2960 |
+
"grad_norm": 0.031804244667956116,
|
| 2961 |
+
"learning_rate": 3.1480000000000006e-06,
|
| 2962 |
+
"loss": 2.4477,
|
| 2963 |
+
"step": 1575
|
| 2964 |
+
},
|
| 2965 |
+
{
|
| 2966 |
+
"epoch": 0.0315,
|
| 2967 |
+
"eval_loss": 2.4615368843078613,
|
| 2968 |
+
"eval_runtime": 33.3548,
|
| 2969 |
+
"eval_samples_per_second": 3.508,
|
| 2970 |
+
"eval_steps_per_second": 1.769,
|
| 2971 |
+
"step": 1575
|
| 2972 |
+
},
|
| 2973 |
+
{
|
| 2974 |
+
"epoch": 0.0316,
|
| 2975 |
+
"eval_loss": 2.461198091506958,
|
| 2976 |
+
"eval_runtime": 33.2416,
|
| 2977 |
+
"eval_samples_per_second": 3.52,
|
| 2978 |
+
"eval_steps_per_second": 1.775,
|
| 2979 |
+
"step": 1580
|
| 2980 |
+
},
|
| 2981 |
+
{
|
| 2982 |
+
"epoch": 0.0317,
|
| 2983 |
+
"eval_loss": 2.4611523151397705,
|
| 2984 |
+
"eval_runtime": 33.3445,
|
| 2985 |
+
"eval_samples_per_second": 3.509,
|
| 2986 |
+
"eval_steps_per_second": 1.769,
|
| 2987 |
+
"step": 1585
|
| 2988 |
+
},
|
| 2989 |
+
{
|
| 2990 |
+
"epoch": 0.0318,
|
| 2991 |
+
"eval_loss": 2.4609127044677734,
|
| 2992 |
+
"eval_runtime": 33.3175,
|
| 2993 |
+
"eval_samples_per_second": 3.512,
|
| 2994 |
+
"eval_steps_per_second": 1.771,
|
| 2995 |
+
"step": 1590
|
| 2996 |
+
},
|
| 2997 |
+
{
|
| 2998 |
+
"epoch": 0.0319,
|
| 2999 |
+
"eval_loss": 2.4608800411224365,
|
| 3000 |
+
"eval_runtime": 33.3052,
|
| 3001 |
+
"eval_samples_per_second": 3.513,
|
| 3002 |
+
"eval_steps_per_second": 1.771,
|
| 3003 |
+
"step": 1595
|
| 3004 |
+
},
|
| 3005 |
+
{
|
| 3006 |
+
"epoch": 0.032,
|
| 3007 |
+
"grad_norm": 0.03365841309984822,
|
| 3008 |
+
"learning_rate": 3.198e-06,
|
| 3009 |
+
"loss": 2.4523,
|
| 3010 |
+
"step": 1600
|
| 3011 |
+
},
|
| 3012 |
+
{
|
| 3013 |
+
"epoch": 0.032,
|
| 3014 |
+
"eval_loss": 2.460757255554199,
|
| 3015 |
+
"eval_runtime": 33.2636,
|
| 3016 |
+
"eval_samples_per_second": 3.517,
|
| 3017 |
+
"eval_steps_per_second": 1.774,
|
| 3018 |
+
"step": 1600
|
| 3019 |
+
},
|
| 3020 |
+
{
|
| 3021 |
+
"epoch": 0.0321,
|
| 3022 |
+
"eval_loss": 2.4605917930603027,
|
| 3023 |
+
"eval_runtime": 33.4595,
|
| 3024 |
+
"eval_samples_per_second": 3.497,
|
| 3025 |
+
"eval_steps_per_second": 1.763,
|
| 3026 |
+
"step": 1605
|
| 3027 |
+
},
|
| 3028 |
+
{
|
| 3029 |
+
"epoch": 0.0322,
|
| 3030 |
+
"eval_loss": 2.4604575634002686,
|
| 3031 |
+
"eval_runtime": 33.2706,
|
| 3032 |
+
"eval_samples_per_second": 3.517,
|
| 3033 |
+
"eval_steps_per_second": 1.773,
|
| 3034 |
+
"step": 1610
|
| 3035 |
+
},
|
| 3036 |
+
{
|
| 3037 |
+
"epoch": 0.0323,
|
| 3038 |
+
"eval_loss": 2.4603111743927,
|
| 3039 |
+
"eval_runtime": 33.405,
|
| 3040 |
+
"eval_samples_per_second": 3.502,
|
| 3041 |
+
"eval_steps_per_second": 1.766,
|
| 3042 |
+
"step": 1615
|
| 3043 |
+
},
|
| 3044 |
+
{
|
| 3045 |
+
"epoch": 0.0324,
|
| 3046 |
+
"eval_loss": 2.460045337677002,
|
| 3047 |
+
"eval_runtime": 33.2598,
|
| 3048 |
+
"eval_samples_per_second": 3.518,
|
| 3049 |
+
"eval_steps_per_second": 1.774,
|
| 3050 |
+
"step": 1620
|
| 3051 |
+
},
|
| 3052 |
+
{
|
| 3053 |
+
"epoch": 0.0325,
|
| 3054 |
+
"grad_norm": 0.03534600587541967,
|
| 3055 |
+
"learning_rate": 3.248e-06,
|
| 3056 |
+
"loss": 2.45,
|
| 3057 |
+
"step": 1625
|
| 3058 |
+
},
|
| 3059 |
+
{
|
| 3060 |
+
"epoch": 0.0325,
|
| 3061 |
+
"eval_loss": 2.460045099258423,
|
| 3062 |
+
"eval_runtime": 33.2663,
|
| 3063 |
+
"eval_samples_per_second": 3.517,
|
| 3064 |
+
"eval_steps_per_second": 1.774,
|
| 3065 |
+
"step": 1625
|
| 3066 |
+
},
|
| 3067 |
+
{
|
| 3068 |
+
"epoch": 0.0326,
|
| 3069 |
+
"eval_loss": 2.4599287509918213,
|
| 3070 |
+
"eval_runtime": 33.2545,
|
| 3071 |
+
"eval_samples_per_second": 3.518,
|
| 3072 |
+
"eval_steps_per_second": 1.774,
|
| 3073 |
+
"step": 1630
|
| 3074 |
+
},
|
| 3075 |
+
{
|
| 3076 |
+
"epoch": 0.0327,
|
| 3077 |
+
"eval_loss": 2.459611654281616,
|
| 3078 |
+
"eval_runtime": 33.4189,
|
| 3079 |
+
"eval_samples_per_second": 3.501,
|
| 3080 |
+
"eval_steps_per_second": 1.765,
|
| 3081 |
+
"step": 1635
|
| 3082 |
+
},
|
| 3083 |
+
{
|
| 3084 |
+
"epoch": 0.0328,
|
| 3085 |
+
"eval_loss": 2.4594151973724365,
|
| 3086 |
+
"eval_runtime": 33.284,
|
| 3087 |
+
"eval_samples_per_second": 3.515,
|
| 3088 |
+
"eval_steps_per_second": 1.773,
|
| 3089 |
+
"step": 1640
|
| 3090 |
+
},
|
| 3091 |
+
{
|
| 3092 |
+
"epoch": 0.0329,
|
| 3093 |
+
"eval_loss": 2.4589221477508545,
|
| 3094 |
+
"eval_runtime": 33.4033,
|
| 3095 |
+
"eval_samples_per_second": 3.503,
|
| 3096 |
+
"eval_steps_per_second": 1.766,
|
| 3097 |
+
"step": 1645
|
| 3098 |
+
},
|
| 3099 |
+
{
|
| 3100 |
+
"epoch": 0.033,
|
| 3101 |
+
"grad_norm": 0.032596527761614855,
|
| 3102 |
+
"learning_rate": 3.298e-06,
|
| 3103 |
+
"loss": 2.4422,
|
| 3104 |
+
"step": 1650
|
| 3105 |
+
},
|
| 3106 |
+
{
|
| 3107 |
+
"epoch": 0.033,
|
| 3108 |
+
"eval_loss": 2.4589502811431885,
|
| 3109 |
+
"eval_runtime": 33.2986,
|
| 3110 |
+
"eval_samples_per_second": 3.514,
|
| 3111 |
+
"eval_steps_per_second": 1.772,
|
| 3112 |
+
"step": 1650
|
| 3113 |
+
},
|
| 3114 |
+
{
|
| 3115 |
+
"epoch": 0.0331,
|
| 3116 |
+
"eval_loss": 2.4588239192962646,
|
| 3117 |
+
"eval_runtime": 33.4046,
|
| 3118 |
+
"eval_samples_per_second": 3.503,
|
| 3119 |
+
"eval_steps_per_second": 1.766,
|
| 3120 |
+
"step": 1655
|
| 3121 |
+
},
|
| 3122 |
+
{
|
| 3123 |
+
"epoch": 0.0332,
|
| 3124 |
+
"eval_loss": 2.458603620529175,
|
| 3125 |
+
"eval_runtime": 33.3448,
|
| 3126 |
+
"eval_samples_per_second": 3.509,
|
| 3127 |
+
"eval_steps_per_second": 1.769,
|
| 3128 |
+
"step": 1660
|
| 3129 |
+
},
|
| 3130 |
+
{
|
| 3131 |
+
"epoch": 0.0333,
|
| 3132 |
+
"eval_loss": 2.458559513092041,
|
| 3133 |
+
"eval_runtime": 33.368,
|
| 3134 |
+
"eval_samples_per_second": 3.506,
|
| 3135 |
+
"eval_steps_per_second": 1.768,
|
| 3136 |
+
"step": 1665
|
| 3137 |
+
},
|
| 3138 |
+
{
|
| 3139 |
+
"epoch": 0.0334,
|
| 3140 |
+
"eval_loss": 2.458500862121582,
|
| 3141 |
+
"eval_runtime": 33.2335,
|
| 3142 |
+
"eval_samples_per_second": 3.521,
|
| 3143 |
+
"eval_steps_per_second": 1.775,
|
| 3144 |
+
"step": 1670
|
| 3145 |
+
},
|
| 3146 |
+
{
|
| 3147 |
+
"epoch": 0.0335,
|
| 3148 |
+
"grad_norm": 0.03339611698643194,
|
| 3149 |
+
"learning_rate": 3.348e-06,
|
| 3150 |
+
"loss": 2.447,
|
| 3151 |
+
"step": 1675
|
| 3152 |
+
},
|
| 3153 |
+
{
|
| 3154 |
+
"epoch": 0.0335,
|
| 3155 |
+
"eval_loss": 2.458252191543579,
|
| 3156 |
+
"eval_runtime": 33.3623,
|
| 3157 |
+
"eval_samples_per_second": 3.507,
|
| 3158 |
+
"eval_steps_per_second": 1.768,
|
| 3159 |
+
"step": 1675
|
| 3160 |
+
},
|
| 3161 |
+
{
|
| 3162 |
+
"epoch": 0.0336,
|
| 3163 |
+
"eval_loss": 2.4580931663513184,
|
| 3164 |
+
"eval_runtime": 33.2532,
|
| 3165 |
+
"eval_samples_per_second": 3.518,
|
| 3166 |
+
"eval_steps_per_second": 1.774,
|
| 3167 |
+
"step": 1680
|
| 3168 |
+
},
|
| 3169 |
+
{
|
| 3170 |
+
"epoch": 0.0337,
|
| 3171 |
+
"eval_loss": 2.4578795433044434,
|
| 3172 |
+
"eval_runtime": 33.3214,
|
| 3173 |
+
"eval_samples_per_second": 3.511,
|
| 3174 |
+
"eval_steps_per_second": 1.771,
|
| 3175 |
+
"step": 1685
|
| 3176 |
+
},
|
| 3177 |
+
{
|
| 3178 |
+
"epoch": 0.0338,
|
| 3179 |
+
"eval_loss": 2.4576218128204346,
|
| 3180 |
+
"eval_runtime": 33.248,
|
| 3181 |
+
"eval_samples_per_second": 3.519,
|
| 3182 |
+
"eval_steps_per_second": 1.775,
|
| 3183 |
+
"step": 1690
|
| 3184 |
+
},
|
| 3185 |
+
{
|
| 3186 |
+
"epoch": 0.0339,
|
| 3187 |
+
"eval_loss": 2.4576828479766846,
|
| 3188 |
+
"eval_runtime": 33.3499,
|
| 3189 |
+
"eval_samples_per_second": 3.508,
|
| 3190 |
+
"eval_steps_per_second": 1.769,
|
| 3191 |
+
"step": 1695
|
| 3192 |
+
},
|
| 3193 |
+
{
|
| 3194 |
+
"epoch": 0.034,
|
| 3195 |
+
"grad_norm": 0.03028181865357742,
|
| 3196 |
+
"learning_rate": 3.3980000000000003e-06,
|
| 3197 |
+
"loss": 2.4582,
|
| 3198 |
+
"step": 1700
|
| 3199 |
+
},
|
| 3200 |
+
{
|
| 3201 |
+
"epoch": 0.034,
|
| 3202 |
+
"eval_loss": 2.457383155822754,
|
| 3203 |
+
"eval_runtime": 33.2574,
|
| 3204 |
+
"eval_samples_per_second": 3.518,
|
| 3205 |
+
"eval_steps_per_second": 1.774,
|
| 3206 |
+
"step": 1700
|
| 3207 |
+
},
|
| 3208 |
+
{
|
| 3209 |
+
"epoch": 0.0341,
|
| 3210 |
+
"eval_loss": 2.4572579860687256,
|
| 3211 |
+
"eval_runtime": 33.2947,
|
| 3212 |
+
"eval_samples_per_second": 3.514,
|
| 3213 |
+
"eval_steps_per_second": 1.772,
|
| 3214 |
+
"step": 1705
|
| 3215 |
+
},
|
| 3216 |
+
{
|
| 3217 |
+
"epoch": 0.0342,
|
| 3218 |
+
"eval_loss": 2.4584450721740723,
|
| 3219 |
+
"eval_runtime": 33.3296,
|
| 3220 |
+
"eval_samples_per_second": 3.51,
|
| 3221 |
+
"eval_steps_per_second": 1.77,
|
| 3222 |
+
"step": 1710
|
| 3223 |
+
},
|
| 3224 |
+
{
|
| 3225 |
+
"epoch": 0.0343,
|
| 3226 |
+
"eval_loss": 2.458603858947754,
|
| 3227 |
+
"eval_runtime": 33.3017,
|
| 3228 |
+
"eval_samples_per_second": 3.513,
|
| 3229 |
+
"eval_steps_per_second": 1.772,
|
| 3230 |
+
"step": 1715
|
| 3231 |
+
},
|
| 3232 |
+
{
|
| 3233 |
+
"epoch": 0.0344,
|
| 3234 |
+
"eval_loss": 2.4579555988311768,
|
| 3235 |
+
"eval_runtime": 33.292,
|
| 3236 |
+
"eval_samples_per_second": 3.514,
|
| 3237 |
+
"eval_steps_per_second": 1.772,
|
| 3238 |
+
"step": 1720
|
| 3239 |
+
},
|
| 3240 |
+
{
|
| 3241 |
+
"epoch": 0.0345,
|
| 3242 |
+
"grad_norm": 0.03734241446236971,
|
| 3243 |
+
"learning_rate": 3.4480000000000003e-06,
|
| 3244 |
+
"loss": 2.4501,
|
| 3245 |
+
"step": 1725
|
| 3246 |
+
},
|
| 3247 |
+
{
|
| 3248 |
+
"epoch": 0.0345,
|
| 3249 |
+
"eval_loss": 2.4574153423309326,
|
| 3250 |
+
"eval_runtime": 33.4313,
|
| 3251 |
+
"eval_samples_per_second": 3.5,
|
| 3252 |
+
"eval_steps_per_second": 1.765,
|
| 3253 |
+
"step": 1725
|
| 3254 |
+
},
|
| 3255 |
+
{
|
| 3256 |
+
"epoch": 0.0346,
|
| 3257 |
+
"eval_loss": 2.456867218017578,
|
| 3258 |
+
"eval_runtime": 33.2833,
|
| 3259 |
+
"eval_samples_per_second": 3.515,
|
| 3260 |
+
"eval_steps_per_second": 1.773,
|
| 3261 |
+
"step": 1730
|
| 3262 |
+
},
|
| 3263 |
+
{
|
| 3264 |
+
"epoch": 0.0347,
|
| 3265 |
+
"eval_loss": 2.4567270278930664,
|
| 3266 |
+
"eval_runtime": 33.3694,
|
| 3267 |
+
"eval_samples_per_second": 3.506,
|
| 3268 |
+
"eval_steps_per_second": 1.768,
|
| 3269 |
+
"step": 1735
|
| 3270 |
+
},
|
| 3271 |
+
{
|
| 3272 |
+
"epoch": 0.0348,
|
| 3273 |
+
"eval_loss": 2.456348180770874,
|
| 3274 |
+
"eval_runtime": 33.3416,
|
| 3275 |
+
"eval_samples_per_second": 3.509,
|
| 3276 |
+
"eval_steps_per_second": 1.77,
|
| 3277 |
+
"step": 1740
|
| 3278 |
+
},
|
| 3279 |
+
{
|
| 3280 |
+
"epoch": 0.0349,
|
| 3281 |
+
"eval_loss": 2.4563136100769043,
|
| 3282 |
+
"eval_runtime": 33.3531,
|
| 3283 |
+
"eval_samples_per_second": 3.508,
|
| 3284 |
+
"eval_steps_per_second": 1.769,
|
| 3285 |
+
"step": 1745
|
| 3286 |
+
},
|
| 3287 |
+
{
|
| 3288 |
+
"epoch": 0.035,
|
| 3289 |
+
"grad_norm": 0.030782538004837847,
|
| 3290 |
+
"learning_rate": 3.4980000000000002e-06,
|
| 3291 |
+
"loss": 2.4509,
|
| 3292 |
+
"step": 1750
|
| 3293 |
+
},
|
| 3294 |
+
{
|
| 3295 |
+
"epoch": 0.035,
|
| 3296 |
+
"eval_loss": 2.455827236175537,
|
| 3297 |
+
"eval_runtime": 33.3143,
|
| 3298 |
+
"eval_samples_per_second": 3.512,
|
| 3299 |
+
"eval_steps_per_second": 1.771,
|
| 3300 |
+
"step": 1750
|
| 3301 |
+
},
|
| 3302 |
+
{
|
| 3303 |
+
"epoch": 0.0351,
|
| 3304 |
+
"eval_loss": 2.4558639526367188,
|
| 3305 |
+
"eval_runtime": 33.3716,
|
| 3306 |
+
"eval_samples_per_second": 3.506,
|
| 3307 |
+
"eval_steps_per_second": 1.768,
|
| 3308 |
+
"step": 1755
|
| 3309 |
+
},
|
| 3310 |
+
{
|
| 3311 |
+
"epoch": 0.0352,
|
| 3312 |
+
"eval_loss": 2.4555938243865967,
|
| 3313 |
+
"eval_runtime": 33.2966,
|
| 3314 |
+
"eval_samples_per_second": 3.514,
|
| 3315 |
+
"eval_steps_per_second": 1.772,
|
| 3316 |
+
"step": 1760
|
| 3317 |
+
},
|
| 3318 |
+
{
|
| 3319 |
+
"epoch": 0.0353,
|
| 3320 |
+
"eval_loss": 2.4551546573638916,
|
| 3321 |
+
"eval_runtime": 33.3145,
|
| 3322 |
+
"eval_samples_per_second": 3.512,
|
| 3323 |
+
"eval_steps_per_second": 1.771,
|
| 3324 |
+
"step": 1765
|
| 3325 |
+
},
|
| 3326 |
+
{
|
| 3327 |
+
"epoch": 0.0354,
|
| 3328 |
+
"eval_loss": 2.454957962036133,
|
| 3329 |
+
"eval_runtime": 33.3201,
|
| 3330 |
+
"eval_samples_per_second": 3.511,
|
| 3331 |
+
"eval_steps_per_second": 1.771,
|
| 3332 |
+
"step": 1770
|
| 3333 |
+
},
|
| 3334 |
+
{
|
| 3335 |
+
"epoch": 0.0355,
|
| 3336 |
+
"grad_norm": 0.03281862515471333,
|
| 3337 |
+
"learning_rate": 3.548e-06,
|
| 3338 |
+
"loss": 2.4439,
|
| 3339 |
+
"step": 1775
|
| 3340 |
+
},
|
| 3341 |
+
{
|
| 3342 |
+
"epoch": 0.0355,
|
| 3343 |
+
"eval_loss": 2.455031394958496,
|
| 3344 |
+
"eval_runtime": 33.264,
|
| 3345 |
+
"eval_samples_per_second": 3.517,
|
| 3346 |
+
"eval_steps_per_second": 1.774,
|
| 3347 |
+
"step": 1775
|
| 3348 |
+
},
|
| 3349 |
+
{
|
| 3350 |
+
"epoch": 0.0356,
|
| 3351 |
+
"eval_loss": 2.4550724029541016,
|
| 3352 |
+
"eval_runtime": 33.3734,
|
| 3353 |
+
"eval_samples_per_second": 3.506,
|
| 3354 |
+
"eval_steps_per_second": 1.768,
|
| 3355 |
+
"step": 1780
|
| 3356 |
+
},
|
| 3357 |
+
{
|
| 3358 |
+
"epoch": 0.0357,
|
| 3359 |
+
"eval_loss": 2.454719305038452,
|
| 3360 |
+
"eval_runtime": 33.3267,
|
| 3361 |
+
"eval_samples_per_second": 3.511,
|
| 3362 |
+
"eval_steps_per_second": 1.77,
|
| 3363 |
+
"step": 1785
|
| 3364 |
+
},
|
| 3365 |
+
{
|
| 3366 |
+
"epoch": 0.0358,
|
| 3367 |
+
"eval_loss": 2.4547033309936523,
|
| 3368 |
+
"eval_runtime": 33.2651,
|
| 3369 |
+
"eval_samples_per_second": 3.517,
|
| 3370 |
+
"eval_steps_per_second": 1.774,
|
| 3371 |
+
"step": 1790
|
| 3372 |
+
},
|
| 3373 |
+
{
|
| 3374 |
+
"epoch": 0.0359,
|
| 3375 |
+
"eval_loss": 2.454416275024414,
|
| 3376 |
+
"eval_runtime": 33.3612,
|
| 3377 |
+
"eval_samples_per_second": 3.507,
|
| 3378 |
+
"eval_steps_per_second": 1.769,
|
| 3379 |
+
"step": 1795
|
| 3380 |
+
},
|
| 3381 |
+
{
|
| 3382 |
+
"epoch": 0.036,
|
| 3383 |
+
"grad_norm": 0.031756006482001914,
|
| 3384 |
+
"learning_rate": 3.5980000000000005e-06,
|
| 3385 |
+
"loss": 2.4493,
|
| 3386 |
+
"step": 1800
|
| 3387 |
+
},
|
| 3388 |
+
{
|
| 3389 |
+
"epoch": 0.036,
|
| 3390 |
+
"eval_loss": 2.454286813735962,
|
| 3391 |
+
"eval_runtime": 33.326,
|
| 3392 |
+
"eval_samples_per_second": 3.511,
|
| 3393 |
+
"eval_steps_per_second": 1.77,
|
| 3394 |
+
"step": 1800
|
| 3395 |
+
},
|
| 3396 |
+
{
|
| 3397 |
+
"epoch": 0.0361,
|
| 3398 |
+
"eval_loss": 2.4541101455688477,
|
| 3399 |
+
"eval_runtime": 33.2597,
|
| 3400 |
+
"eval_samples_per_second": 3.518,
|
| 3401 |
+
"eval_steps_per_second": 1.774,
|
| 3402 |
+
"step": 1805
|
| 3403 |
+
},
|
| 3404 |
+
{
|
| 3405 |
+
"epoch": 0.0362,
|
| 3406 |
+
"eval_loss": 2.4541351795196533,
|
| 3407 |
+
"eval_runtime": 33.2421,
|
| 3408 |
+
"eval_samples_per_second": 3.52,
|
| 3409 |
+
"eval_steps_per_second": 1.775,
|
| 3410 |
+
"step": 1810
|
| 3411 |
+
},
|
| 3412 |
+
{
|
| 3413 |
+
"epoch": 0.0363,
|
| 3414 |
+
"eval_loss": 2.4537973403930664,
|
| 3415 |
+
"eval_runtime": 33.3201,
|
| 3416 |
+
"eval_samples_per_second": 3.511,
|
| 3417 |
+
"eval_steps_per_second": 1.771,
|
| 3418 |
+
"step": 1815
|
| 3419 |
+
},
|
| 3420 |
+
{
|
| 3421 |
+
"epoch": 0.0364,
|
| 3422 |
+
"eval_loss": 2.4534847736358643,
|
| 3423 |
+
"eval_runtime": 33.2973,
|
| 3424 |
+
"eval_samples_per_second": 3.514,
|
| 3425 |
+
"eval_steps_per_second": 1.772,
|
| 3426 |
+
"step": 1820
|
| 3427 |
+
},
|
| 3428 |
+
{
|
| 3429 |
+
"epoch": 0.0365,
|
| 3430 |
+
"grad_norm": 0.03128096989289917,
|
| 3431 |
+
"learning_rate": 3.6480000000000005e-06,
|
| 3432 |
+
"loss": 2.4526,
|
| 3433 |
+
"step": 1825
|
| 3434 |
+
},
|
| 3435 |
+
{
|
| 3436 |
+
"epoch": 0.0365,
|
| 3437 |
+
"eval_loss": 2.453655481338501,
|
| 3438 |
+
"eval_runtime": 33.3755,
|
| 3439 |
+
"eval_samples_per_second": 3.506,
|
| 3440 |
+
"eval_steps_per_second": 1.768,
|
| 3441 |
+
"step": 1825
|
| 3442 |
+
},
|
| 3443 |
+
{
|
| 3444 |
+
"epoch": 0.0366,
|
| 3445 |
+
"eval_loss": 2.4534049034118652,
|
| 3446 |
+
"eval_runtime": 33.332,
|
| 3447 |
+
"eval_samples_per_second": 3.51,
|
| 3448 |
+
"eval_steps_per_second": 1.77,
|
| 3449 |
+
"step": 1830
|
| 3450 |
+
},
|
| 3451 |
+
{
|
| 3452 |
+
"epoch": 0.0367,
|
| 3453 |
+
"eval_loss": 2.4529781341552734,
|
| 3454 |
+
"eval_runtime": 33.3325,
|
| 3455 |
+
"eval_samples_per_second": 3.51,
|
| 3456 |
+
"eval_steps_per_second": 1.77,
|
| 3457 |
+
"step": 1835
|
| 3458 |
+
},
|
| 3459 |
+
{
|
| 3460 |
+
"epoch": 0.0368,
|
| 3461 |
+
"eval_loss": 2.454005241394043,
|
| 3462 |
+
"eval_runtime": 33.3975,
|
| 3463 |
+
"eval_samples_per_second": 3.503,
|
| 3464 |
+
"eval_steps_per_second": 1.767,
|
| 3465 |
+
"step": 1840
|
| 3466 |
+
},
|
| 3467 |
+
{
|
| 3468 |
+
"epoch": 0.0369,
|
| 3469 |
+
"eval_loss": 2.4538745880126953,
|
| 3470 |
+
"eval_runtime": 33.3,
|
| 3471 |
+
"eval_samples_per_second": 3.514,
|
| 3472 |
+
"eval_steps_per_second": 1.772,
|
| 3473 |
+
"step": 1845
|
| 3474 |
+
},
|
| 3475 |
+
{
|
| 3476 |
+
"epoch": 0.037,
|
| 3477 |
+
"grad_norm": 0.02999582338402207,
|
| 3478 |
+
"learning_rate": 3.6980000000000004e-06,
|
| 3479 |
+
"loss": 2.4309,
|
| 3480 |
+
"step": 1850
|
| 3481 |
+
},
|
| 3482 |
+
{
|
| 3483 |
+
"epoch": 0.037,
|
| 3484 |
+
"eval_loss": 2.4534404277801514,
|
| 3485 |
+
"eval_runtime": 33.2825,
|
| 3486 |
+
"eval_samples_per_second": 3.515,
|
| 3487 |
+
"eval_steps_per_second": 1.773,
|
| 3488 |
+
"step": 1850
|
| 3489 |
+
},
|
| 3490 |
+
{
|
| 3491 |
+
"epoch": 0.0371,
|
| 3492 |
+
"eval_loss": 2.4529800415039062,
|
| 3493 |
+
"eval_runtime": 33.513,
|
| 3494 |
+
"eval_samples_per_second": 3.491,
|
| 3495 |
+
"eval_steps_per_second": 1.761,
|
| 3496 |
+
"step": 1855
|
| 3497 |
+
},
|
| 3498 |
+
{
|
| 3499 |
+
"epoch": 0.0372,
|
| 3500 |
+
"eval_loss": 2.453007221221924,
|
| 3501 |
+
"eval_runtime": 33.3414,
|
| 3502 |
+
"eval_samples_per_second": 3.509,
|
| 3503 |
+
"eval_steps_per_second": 1.77,
|
| 3504 |
+
"step": 1860
|
| 3505 |
+
},
|
| 3506 |
+
{
|
| 3507 |
+
"epoch": 0.0373,
|
| 3508 |
+
"eval_loss": 2.452350616455078,
|
| 3509 |
+
"eval_runtime": 33.3625,
|
| 3510 |
+
"eval_samples_per_second": 3.507,
|
| 3511 |
+
"eval_steps_per_second": 1.768,
|
| 3512 |
+
"step": 1865
|
| 3513 |
+
},
|
| 3514 |
+
{
|
| 3515 |
+
"epoch": 0.0374,
|
| 3516 |
+
"eval_loss": 2.4522666931152344,
|
| 3517 |
+
"eval_runtime": 33.3116,
|
| 3518 |
+
"eval_samples_per_second": 3.512,
|
| 3519 |
+
"eval_steps_per_second": 1.771,
|
| 3520 |
+
"step": 1870
|
| 3521 |
+
},
|
| 3522 |
+
{
|
| 3523 |
+
"epoch": 0.0375,
|
| 3524 |
+
"grad_norm": 0.0409025592520596,
|
| 3525 |
+
"learning_rate": 3.7480000000000004e-06,
|
| 3526 |
+
"loss": 2.442,
|
| 3527 |
+
"step": 1875
|
| 3528 |
+
},
|
| 3529 |
+
{
|
| 3530 |
+
"epoch": 0.0375,
|
| 3531 |
+
"eval_loss": 2.4521546363830566,
|
| 3532 |
+
"eval_runtime": 33.3782,
|
| 3533 |
+
"eval_samples_per_second": 3.505,
|
| 3534 |
+
"eval_steps_per_second": 1.768,
|
| 3535 |
+
"step": 1875
|
| 3536 |
+
},
|
| 3537 |
+
{
|
| 3538 |
+
"epoch": 0.0376,
|
| 3539 |
+
"eval_loss": 2.4520437717437744,
|
| 3540 |
+
"eval_runtime": 33.2887,
|
| 3541 |
+
"eval_samples_per_second": 3.515,
|
| 3542 |
+
"eval_steps_per_second": 1.772,
|
| 3543 |
+
"step": 1880
|
| 3544 |
+
},
|
| 3545 |
+
{
|
| 3546 |
+
"epoch": 0.0377,
|
| 3547 |
+
"eval_loss": 2.4519331455230713,
|
| 3548 |
+
"eval_runtime": 33.3746,
|
| 3549 |
+
"eval_samples_per_second": 3.506,
|
| 3550 |
+
"eval_steps_per_second": 1.768,
|
| 3551 |
+
"step": 1885
|
| 3552 |
+
},
|
| 3553 |
+
{
|
| 3554 |
+
"epoch": 0.0378,
|
| 3555 |
+
"eval_loss": 2.451744556427002,
|
| 3556 |
+
"eval_runtime": 33.3214,
|
| 3557 |
+
"eval_samples_per_second": 3.511,
|
| 3558 |
+
"eval_steps_per_second": 1.771,
|
| 3559 |
+
"step": 1890
|
| 3560 |
+
},
|
| 3561 |
+
{
|
| 3562 |
+
"epoch": 0.0379,
|
| 3563 |
+
"eval_loss": 2.451737642288208,
|
| 3564 |
+
"eval_runtime": 33.3457,
|
| 3565 |
+
"eval_samples_per_second": 3.509,
|
| 3566 |
+
"eval_steps_per_second": 1.769,
|
| 3567 |
+
"step": 1895
|
| 3568 |
+
},
|
| 3569 |
+
{
|
| 3570 |
+
"epoch": 0.038,
|
| 3571 |
+
"grad_norm": 0.03431980647954774,
|
| 3572 |
+
"learning_rate": 3.7980000000000007e-06,
|
| 3573 |
+
"loss": 2.4477,
|
| 3574 |
+
"step": 1900
|
| 3575 |
+
},
|
| 3576 |
+
{
|
| 3577 |
+
"epoch": 0.038,
|
| 3578 |
+
"eval_loss": 2.4515624046325684,
|
| 3579 |
+
"eval_runtime": 33.312,
|
| 3580 |
+
"eval_samples_per_second": 3.512,
|
| 3581 |
+
"eval_steps_per_second": 1.771,
|
| 3582 |
+
"step": 1900
|
| 3583 |
+
},
|
| 3584 |
+
{
|
| 3585 |
+
"epoch": 0.0381,
|
| 3586 |
+
"eval_loss": 2.4512295722961426,
|
| 3587 |
+
"eval_runtime": 33.3607,
|
| 3588 |
+
"eval_samples_per_second": 3.507,
|
| 3589 |
+
"eval_steps_per_second": 1.769,
|
| 3590 |
+
"step": 1905
|
| 3591 |
+
},
|
| 3592 |
+
{
|
| 3593 |
+
"epoch": 0.0382,
|
| 3594 |
+
"eval_loss": 2.4510445594787598,
|
| 3595 |
+
"eval_runtime": 33.339,
|
| 3596 |
+
"eval_samples_per_second": 3.509,
|
| 3597 |
+
"eval_steps_per_second": 1.77,
|
| 3598 |
+
"step": 1910
|
| 3599 |
+
},
|
| 3600 |
+
{
|
| 3601 |
+
"epoch": 0.0383,
|
| 3602 |
+
"eval_loss": 2.4508397579193115,
|
| 3603 |
+
"eval_runtime": 33.3996,
|
| 3604 |
+
"eval_samples_per_second": 3.503,
|
| 3605 |
+
"eval_steps_per_second": 1.766,
|
| 3606 |
+
"step": 1915
|
| 3607 |
+
},
|
| 3608 |
+
{
|
| 3609 |
+
"epoch": 0.0384,
|
| 3610 |
+
"eval_loss": 2.4510440826416016,
|
| 3611 |
+
"eval_runtime": 33.2905,
|
| 3612 |
+
"eval_samples_per_second": 3.515,
|
| 3613 |
+
"eval_steps_per_second": 1.772,
|
| 3614 |
+
"step": 1920
|
| 3615 |
+
},
|
| 3616 |
+
{
|
| 3617 |
+
"epoch": 0.0385,
|
| 3618 |
+
"grad_norm": 0.03587224652231601,
|
| 3619 |
+
"learning_rate": 3.848e-06,
|
| 3620 |
+
"loss": 2.4433,
|
| 3621 |
+
"step": 1925
|
| 3622 |
+
},
|
| 3623 |
+
{
|
| 3624 |
+
"epoch": 0.0385,
|
| 3625 |
+
"eval_loss": 2.450984239578247,
|
| 3626 |
+
"eval_runtime": 33.3263,
|
| 3627 |
+
"eval_samples_per_second": 3.511,
|
| 3628 |
+
"eval_steps_per_second": 1.77,
|
| 3629 |
+
"step": 1925
|
| 3630 |
+
},
|
| 3631 |
+
{
|
| 3632 |
+
"epoch": 0.0386,
|
| 3633 |
+
"eval_loss": 2.45090651512146,
|
| 3634 |
+
"eval_runtime": 33.3244,
|
| 3635 |
+
"eval_samples_per_second": 3.511,
|
| 3636 |
+
"eval_steps_per_second": 1.77,
|
| 3637 |
+
"step": 1930
|
| 3638 |
+
},
|
| 3639 |
+
{
|
| 3640 |
+
"epoch": 0.0387,
|
| 3641 |
+
"eval_loss": 2.450443983078003,
|
| 3642 |
+
"eval_runtime": 33.3023,
|
| 3643 |
+
"eval_samples_per_second": 3.513,
|
| 3644 |
+
"eval_steps_per_second": 1.772,
|
| 3645 |
+
"step": 1935
|
| 3646 |
+
},
|
| 3647 |
+
{
|
| 3648 |
+
"epoch": 0.0388,
|
| 3649 |
+
"eval_loss": 2.450309991836548,
|
| 3650 |
+
"eval_runtime": 33.4354,
|
| 3651 |
+
"eval_samples_per_second": 3.499,
|
| 3652 |
+
"eval_steps_per_second": 1.765,
|
| 3653 |
+
"step": 1940
|
| 3654 |
+
},
|
| 3655 |
+
{
|
| 3656 |
+
"epoch": 0.0389,
|
| 3657 |
+
"eval_loss": 2.4500510692596436,
|
| 3658 |
+
"eval_runtime": 33.3238,
|
| 3659 |
+
"eval_samples_per_second": 3.511,
|
| 3660 |
+
"eval_steps_per_second": 1.771,
|
| 3661 |
+
"step": 1945
|
| 3662 |
+
},
|
| 3663 |
+
{
|
| 3664 |
+
"epoch": 0.039,
|
| 3665 |
+
"grad_norm": 0.027239293031380653,
|
| 3666 |
+
"learning_rate": 3.898e-06,
|
| 3667 |
+
"loss": 2.4347,
|
| 3668 |
+
"step": 1950
|
| 3669 |
+
},
|
| 3670 |
+
{
|
| 3671 |
+
"epoch": 0.039,
|
| 3672 |
+
"eval_loss": 2.4498231410980225,
|
| 3673 |
+
"eval_runtime": 33.3306,
|
| 3674 |
+
"eval_samples_per_second": 3.51,
|
| 3675 |
+
"eval_steps_per_second": 1.77,
|
| 3676 |
+
"step": 1950
|
| 3677 |
+
},
|
| 3678 |
+
{
|
| 3679 |
+
"epoch": 0.0391,
|
| 3680 |
+
"eval_loss": 2.449704170227051,
|
| 3681 |
+
"eval_runtime": 33.3865,
|
| 3682 |
+
"eval_samples_per_second": 3.504,
|
| 3683 |
+
"eval_steps_per_second": 1.767,
|
| 3684 |
+
"step": 1955
|
| 3685 |
+
},
|
| 3686 |
+
{
|
| 3687 |
+
"epoch": 0.0392,
|
| 3688 |
+
"eval_loss": 2.44974684715271,
|
| 3689 |
+
"eval_runtime": 33.419,
|
| 3690 |
+
"eval_samples_per_second": 3.501,
|
| 3691 |
+
"eval_steps_per_second": 1.765,
|
| 3692 |
+
"step": 1960
|
| 3693 |
+
},
|
| 3694 |
+
{
|
| 3695 |
+
"epoch": 0.0393,
|
| 3696 |
+
"eval_loss": 2.450090169906616,
|
| 3697 |
+
"eval_runtime": 33.5315,
|
| 3698 |
+
"eval_samples_per_second": 3.489,
|
| 3699 |
+
"eval_steps_per_second": 1.76,
|
| 3700 |
+
"step": 1965
|
| 3701 |
+
},
|
| 3702 |
+
{
|
| 3703 |
+
"epoch": 0.0394,
|
| 3704 |
+
"eval_loss": 2.4494845867156982,
|
| 3705 |
+
"eval_runtime": 33.4607,
|
| 3706 |
+
"eval_samples_per_second": 3.497,
|
| 3707 |
+
"eval_steps_per_second": 1.763,
|
| 3708 |
+
"step": 1970
|
| 3709 |
+
},
|
| 3710 |
+
{
|
| 3711 |
+
"epoch": 0.0395,
|
| 3712 |
+
"grad_norm": 0.031553482039351585,
|
| 3713 |
+
"learning_rate": 3.948e-06,
|
| 3714 |
+
"loss": 2.4466,
|
| 3715 |
+
"step": 1975
|
| 3716 |
+
},
|
| 3717 |
+
{
|
| 3718 |
+
"epoch": 0.0395,
|
| 3719 |
+
"eval_loss": 2.449598550796509,
|
| 3720 |
+
"eval_runtime": 33.4853,
|
| 3721 |
+
"eval_samples_per_second": 3.494,
|
| 3722 |
+
"eval_steps_per_second": 1.762,
|
| 3723 |
+
"step": 1975
|
| 3724 |
+
},
|
| 3725 |
+
{
|
| 3726 |
+
"epoch": 0.0396,
|
| 3727 |
+
"eval_loss": 2.449420213699341,
|
| 3728 |
+
"eval_runtime": 33.4626,
|
| 3729 |
+
"eval_samples_per_second": 3.496,
|
| 3730 |
+
"eval_steps_per_second": 1.763,
|
| 3731 |
+
"step": 1980
|
| 3732 |
+
},
|
| 3733 |
+
{
|
| 3734 |
+
"epoch": 0.0397,
|
| 3735 |
+
"eval_loss": 2.449462890625,
|
| 3736 |
+
"eval_runtime": 33.4049,
|
| 3737 |
+
"eval_samples_per_second": 3.502,
|
| 3738 |
+
"eval_steps_per_second": 1.766,
|
| 3739 |
+
"step": 1985
|
| 3740 |
+
},
|
| 3741 |
+
{
|
| 3742 |
+
"epoch": 0.0398,
|
| 3743 |
+
"eval_loss": 2.449423313140869,
|
| 3744 |
+
"eval_runtime": 33.5823,
|
| 3745 |
+
"eval_samples_per_second": 3.484,
|
| 3746 |
+
"eval_steps_per_second": 1.757,
|
| 3747 |
+
"step": 1990
|
| 3748 |
+
},
|
| 3749 |
+
{
|
| 3750 |
+
"epoch": 0.0399,
|
| 3751 |
+
"eval_loss": 2.4491324424743652,
|
| 3752 |
+
"eval_runtime": 33.662,
|
| 3753 |
+
"eval_samples_per_second": 3.476,
|
| 3754 |
+
"eval_steps_per_second": 1.753,
|
| 3755 |
+
"step": 1995
|
| 3756 |
+
},
|
| 3757 |
+
{
|
| 3758 |
+
"epoch": 0.04,
|
| 3759 |
+
"grad_norm": 0.03314009226524554,
|
| 3760 |
+
"learning_rate": 3.9980000000000005e-06,
|
| 3761 |
+
"loss": 2.4391,
|
| 3762 |
+
"step": 2000
|
| 3763 |
+
},
|
| 3764 |
+
{
|
| 3765 |
+
"epoch": 0.04,
|
| 3766 |
+
"eval_loss": 2.449084520339966,
|
| 3767 |
+
"eval_runtime": 33.5872,
|
| 3768 |
+
"eval_samples_per_second": 3.483,
|
| 3769 |
+
"eval_steps_per_second": 1.757,
|
| 3770 |
+
"step": 2000
|
| 3771 |
}
|
| 3772 |
],
|
| 3773 |
"logging_steps": 25,
|
|
|
|
| 3787 |
"attributes": {}
|
| 3788 |
}
|
| 3789 |
},
|
| 3790 |
+
"total_flos": 5.570603510971498e+18,
|
| 3791 |
"train_batch_size": 1,
|
| 3792 |
"trial_name": null,
|
| 3793 |
"trial_params": null
|