Commit ·
a6b3be7
1
Parent(s): 452d75e
Upload trainer_state.json
Browse files- trainer_state.json +1080 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,1080 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 31.7626,
|
| 3 |
+
"best_model_checkpoint": "/content/drive/Shareddrives/UCM_SHARED/TFM_ESG/Ejemplo Oficial de T5-Summarization de HF/t5-small_adafactor/checkpoint-8100",
|
| 4 |
+
"epoch": 0.9880028228652082,
|
| 5 |
+
"global_step": 8400,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.02,
|
| 12 |
+
"learning_rate": 0.0009823570924488356,
|
| 13 |
+
"loss": 2.9361,
|
| 14 |
+
"step": 150
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.02,
|
| 18 |
+
"eval_gen_len": 18.8845,
|
| 19 |
+
"eval_loss": 2.6215693950653076,
|
| 20 |
+
"eval_rouge1": 26.8542,
|
| 21 |
+
"eval_rouge2": 6.8667,
|
| 22 |
+
"eval_rougeL": 21.1484,
|
| 23 |
+
"eval_rougeLsum": 21.1563,
|
| 24 |
+
"eval_runtime": 342.0879,
|
| 25 |
+
"eval_samples_per_second": 33.126,
|
| 26 |
+
"eval_steps_per_second": 1.383,
|
| 27 |
+
"step": 150
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 0.04,
|
| 31 |
+
"learning_rate": 0.0009648318042813455,
|
| 32 |
+
"loss": 2.8543,
|
| 33 |
+
"step": 300
|
| 34 |
+
},
|
| 35 |
+
{
|
| 36 |
+
"epoch": 0.04,
|
| 37 |
+
"eval_gen_len": 18.8097,
|
| 38 |
+
"eval_loss": 2.5835769176483154,
|
| 39 |
+
"eval_rouge1": 27.2234,
|
| 40 |
+
"eval_rouge2": 7.1753,
|
| 41 |
+
"eval_rougeL": 21.5276,
|
| 42 |
+
"eval_rougeLsum": 21.5308,
|
| 43 |
+
"eval_runtime": 342.5643,
|
| 44 |
+
"eval_samples_per_second": 33.08,
|
| 45 |
+
"eval_steps_per_second": 1.381,
|
| 46 |
+
"step": 300
|
| 47 |
+
},
|
| 48 |
+
{
|
| 49 |
+
"epoch": 0.05,
|
| 50 |
+
"learning_rate": 0.0009471888967301812,
|
| 51 |
+
"loss": 2.814,
|
| 52 |
+
"step": 450
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.05,
|
| 56 |
+
"eval_gen_len": 18.8294,
|
| 57 |
+
"eval_loss": 2.564615488052368,
|
| 58 |
+
"eval_rouge1": 28.1695,
|
| 59 |
+
"eval_rouge2": 7.7873,
|
| 60 |
+
"eval_rougeL": 22.2229,
|
| 61 |
+
"eval_rougeLsum": 22.2251,
|
| 62 |
+
"eval_runtime": 341.6846,
|
| 63 |
+
"eval_samples_per_second": 33.165,
|
| 64 |
+
"eval_steps_per_second": 1.384,
|
| 65 |
+
"step": 450
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.07,
|
| 69 |
+
"learning_rate": 0.0009295459891790166,
|
| 70 |
+
"loss": 2.7861,
|
| 71 |
+
"step": 600
|
| 72 |
+
},
|
| 73 |
+
{
|
| 74 |
+
"epoch": 0.07,
|
| 75 |
+
"eval_gen_len": 18.7867,
|
| 76 |
+
"eval_loss": 2.5409207344055176,
|
| 77 |
+
"eval_rouge1": 28.5349,
|
| 78 |
+
"eval_rouge2": 7.9732,
|
| 79 |
+
"eval_rougeL": 22.6959,
|
| 80 |
+
"eval_rougeLsum": 22.7078,
|
| 81 |
+
"eval_runtime": 341.6917,
|
| 82 |
+
"eval_samples_per_second": 33.164,
|
| 83 |
+
"eval_steps_per_second": 1.384,
|
| 84 |
+
"step": 600
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 0.09,
|
| 88 |
+
"learning_rate": 0.0009119030816278523,
|
| 89 |
+
"loss": 2.76,
|
| 90 |
+
"step": 750
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"epoch": 0.09,
|
| 94 |
+
"eval_gen_len": 18.7526,
|
| 95 |
+
"eval_loss": 2.5160539150238037,
|
| 96 |
+
"eval_rouge1": 28.5629,
|
| 97 |
+
"eval_rouge2": 7.9485,
|
| 98 |
+
"eval_rougeL": 22.6545,
|
| 99 |
+
"eval_rougeLsum": 22.6617,
|
| 100 |
+
"eval_runtime": 341.8755,
|
| 101 |
+
"eval_samples_per_second": 33.147,
|
| 102 |
+
"eval_steps_per_second": 1.384,
|
| 103 |
+
"step": 750
|
| 104 |
+
},
|
| 105 |
+
{
|
| 106 |
+
"epoch": 0.11,
|
| 107 |
+
"learning_rate": 0.0008942601740766878,
|
| 108 |
+
"loss": 2.7545,
|
| 109 |
+
"step": 900
|
| 110 |
+
},
|
| 111 |
+
{
|
| 112 |
+
"epoch": 0.11,
|
| 113 |
+
"eval_gen_len": 18.7481,
|
| 114 |
+
"eval_loss": 2.5028903484344482,
|
| 115 |
+
"eval_rouge1": 29.1851,
|
| 116 |
+
"eval_rouge2": 8.3586,
|
| 117 |
+
"eval_rougeL": 23.1781,
|
| 118 |
+
"eval_rougeLsum": 23.1815,
|
| 119 |
+
"eval_runtime": 341.7767,
|
| 120 |
+
"eval_samples_per_second": 33.156,
|
| 121 |
+
"eval_steps_per_second": 1.384,
|
| 122 |
+
"step": 900
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.12,
|
| 126 |
+
"learning_rate": 0.0008767348859091978,
|
| 127 |
+
"loss": 2.7294,
|
| 128 |
+
"step": 1050
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.12,
|
| 132 |
+
"eval_gen_len": 18.6948,
|
| 133 |
+
"eval_loss": 2.4874510765075684,
|
| 134 |
+
"eval_rouge1": 28.7611,
|
| 135 |
+
"eval_rouge2": 8.0309,
|
| 136 |
+
"eval_rougeL": 22.8289,
|
| 137 |
+
"eval_rougeLsum": 22.8364,
|
| 138 |
+
"eval_runtime": 341.7005,
|
| 139 |
+
"eval_samples_per_second": 33.164,
|
| 140 |
+
"eval_steps_per_second": 1.384,
|
| 141 |
+
"step": 1050
|
| 142 |
+
},
|
| 143 |
+
{
|
| 144 |
+
"epoch": 0.14,
|
| 145 |
+
"learning_rate": 0.0008590919783580335,
|
| 146 |
+
"loss": 2.7005,
|
| 147 |
+
"step": 1200
|
| 148 |
+
},
|
| 149 |
+
{
|
| 150 |
+
"epoch": 0.14,
|
| 151 |
+
"eval_gen_len": 18.7843,
|
| 152 |
+
"eval_loss": 2.4769885540008545,
|
| 153 |
+
"eval_rouge1": 29.1072,
|
| 154 |
+
"eval_rouge2": 8.2713,
|
| 155 |
+
"eval_rougeL": 23.0568,
|
| 156 |
+
"eval_rougeLsum": 23.0602,
|
| 157 |
+
"eval_runtime": 341.9693,
|
| 158 |
+
"eval_samples_per_second": 33.137,
|
| 159 |
+
"eval_steps_per_second": 1.383,
|
| 160 |
+
"step": 1200
|
| 161 |
+
},
|
| 162 |
+
{
|
| 163 |
+
"epoch": 0.16,
|
| 164 |
+
"learning_rate": 0.0008414490708068689,
|
| 165 |
+
"loss": 2.7057,
|
| 166 |
+
"step": 1350
|
| 167 |
+
},
|
| 168 |
+
{
|
| 169 |
+
"epoch": 0.16,
|
| 170 |
+
"eval_gen_len": 18.6973,
|
| 171 |
+
"eval_loss": 2.46754789352417,
|
| 172 |
+
"eval_rouge1": 29.0607,
|
| 173 |
+
"eval_rouge2": 8.4308,
|
| 174 |
+
"eval_rougeL": 23.1094,
|
| 175 |
+
"eval_rougeLsum": 23.1125,
|
| 176 |
+
"eval_runtime": 341.5228,
|
| 177 |
+
"eval_samples_per_second": 33.181,
|
| 178 |
+
"eval_steps_per_second": 1.385,
|
| 179 |
+
"step": 1350
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 0.18,
|
| 183 |
+
"learning_rate": 0.0008238061632557046,
|
| 184 |
+
"loss": 2.6779,
|
| 185 |
+
"step": 1500
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 0.18,
|
| 189 |
+
"eval_gen_len": 18.7927,
|
| 190 |
+
"eval_loss": 2.4461910724639893,
|
| 191 |
+
"eval_rouge1": 29.4815,
|
| 192 |
+
"eval_rouge2": 8.6203,
|
| 193 |
+
"eval_rougeL": 23.3742,
|
| 194 |
+
"eval_rougeLsum": 23.3745,
|
| 195 |
+
"eval_runtime": 341.6546,
|
| 196 |
+
"eval_samples_per_second": 33.168,
|
| 197 |
+
"eval_steps_per_second": 1.384,
|
| 198 |
+
"step": 1500
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.19,
|
| 202 |
+
"learning_rate": 0.00080616325570454,
|
| 203 |
+
"loss": 2.6944,
|
| 204 |
+
"step": 1650
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 0.19,
|
| 208 |
+
"eval_gen_len": 18.8219,
|
| 209 |
+
"eval_loss": 2.4397881031036377,
|
| 210 |
+
"eval_rouge1": 29.3817,
|
| 211 |
+
"eval_rouge2": 8.5247,
|
| 212 |
+
"eval_rougeL": 23.3088,
|
| 213 |
+
"eval_rougeLsum": 23.3137,
|
| 214 |
+
"eval_runtime": 341.8781,
|
| 215 |
+
"eval_samples_per_second": 33.146,
|
| 216 |
+
"eval_steps_per_second": 1.384,
|
| 217 |
+
"step": 1650
|
| 218 |
+
},
|
| 219 |
+
{
|
| 220 |
+
"epoch": 0.21,
|
| 221 |
+
"learning_rate": 0.0007885203481533757,
|
| 222 |
+
"loss": 2.6589,
|
| 223 |
+
"step": 1800
|
| 224 |
+
},
|
| 225 |
+
{
|
| 226 |
+
"epoch": 0.21,
|
| 227 |
+
"eval_gen_len": 18.7917,
|
| 228 |
+
"eval_loss": 2.429356813430786,
|
| 229 |
+
"eval_rouge1": 29.1914,
|
| 230 |
+
"eval_rouge2": 8.5565,
|
| 231 |
+
"eval_rougeL": 23.2017,
|
| 232 |
+
"eval_rougeLsum": 23.2039,
|
| 233 |
+
"eval_runtime": 341.5952,
|
| 234 |
+
"eval_samples_per_second": 33.174,
|
| 235 |
+
"eval_steps_per_second": 1.385,
|
| 236 |
+
"step": 1800
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 0.23,
|
| 240 |
+
"learning_rate": 0.0007708774406022112,
|
| 241 |
+
"loss": 2.6473,
|
| 242 |
+
"step": 1950
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 0.23,
|
| 246 |
+
"eval_gen_len": 18.7605,
|
| 247 |
+
"eval_loss": 2.417249917984009,
|
| 248 |
+
"eval_rouge1": 29.9482,
|
| 249 |
+
"eval_rouge2": 8.8321,
|
| 250 |
+
"eval_rougeL": 23.7848,
|
| 251 |
+
"eval_rougeLsum": 23.7954,
|
| 252 |
+
"eval_runtime": 341.816,
|
| 253 |
+
"eval_samples_per_second": 33.152,
|
| 254 |
+
"eval_steps_per_second": 1.384,
|
| 255 |
+
"step": 1950
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 0.25,
|
| 259 |
+
"learning_rate": 0.0007532345330510469,
|
| 260 |
+
"loss": 2.6498,
|
| 261 |
+
"step": 2100
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.25,
|
| 265 |
+
"eval_gen_len": 18.8079,
|
| 266 |
+
"eval_loss": 2.4158124923706055,
|
| 267 |
+
"eval_rouge1": 29.5843,
|
| 268 |
+
"eval_rouge2": 8.9229,
|
| 269 |
+
"eval_rougeL": 23.6193,
|
| 270 |
+
"eval_rougeLsum": 23.6255,
|
| 271 |
+
"eval_runtime": 341.8291,
|
| 272 |
+
"eval_samples_per_second": 33.151,
|
| 273 |
+
"eval_steps_per_second": 1.384,
|
| 274 |
+
"step": 2100
|
| 275 |
+
},
|
| 276 |
+
{
|
| 277 |
+
"epoch": 0.26,
|
| 278 |
+
"learning_rate": 0.0007355916254998823,
|
| 279 |
+
"loss": 2.6321,
|
| 280 |
+
"step": 2250
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 0.26,
|
| 284 |
+
"eval_gen_len": 18.7639,
|
| 285 |
+
"eval_loss": 2.4022669792175293,
|
| 286 |
+
"eval_rouge1": 29.9147,
|
| 287 |
+
"eval_rouge2": 9.0085,
|
| 288 |
+
"eval_rougeL": 23.8569,
|
| 289 |
+
"eval_rougeLsum": 23.8709,
|
| 290 |
+
"eval_runtime": 341.59,
|
| 291 |
+
"eval_samples_per_second": 33.174,
|
| 292 |
+
"eval_steps_per_second": 1.385,
|
| 293 |
+
"step": 2250
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 0.28,
|
| 297 |
+
"learning_rate": 0.000717948717948718,
|
| 298 |
+
"loss": 2.614,
|
| 299 |
+
"step": 2400
|
| 300 |
+
},
|
| 301 |
+
{
|
| 302 |
+
"epoch": 0.28,
|
| 303 |
+
"eval_gen_len": 18.6922,
|
| 304 |
+
"eval_loss": 2.3971035480499268,
|
| 305 |
+
"eval_rouge1": 29.711,
|
| 306 |
+
"eval_rouge2": 9.0017,
|
| 307 |
+
"eval_rougeL": 23.761,
|
| 308 |
+
"eval_rougeLsum": 23.7742,
|
| 309 |
+
"eval_runtime": 341.4293,
|
| 310 |
+
"eval_samples_per_second": 33.19,
|
| 311 |
+
"eval_steps_per_second": 1.385,
|
| 312 |
+
"step": 2400
|
| 313 |
+
},
|
| 314 |
+
{
|
| 315 |
+
"epoch": 0.3,
|
| 316 |
+
"learning_rate": 0.0007003058103975535,
|
| 317 |
+
"loss": 2.6212,
|
| 318 |
+
"step": 2550
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 0.3,
|
| 322 |
+
"eval_gen_len": 18.7082,
|
| 323 |
+
"eval_loss": 2.390623092651367,
|
| 324 |
+
"eval_rouge1": 29.498,
|
| 325 |
+
"eval_rouge2": 8.6483,
|
| 326 |
+
"eval_rougeL": 23.5006,
|
| 327 |
+
"eval_rougeLsum": 23.5112,
|
| 328 |
+
"eval_runtime": 341.8905,
|
| 329 |
+
"eval_samples_per_second": 33.145,
|
| 330 |
+
"eval_steps_per_second": 1.383,
|
| 331 |
+
"step": 2550
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 0.32,
|
| 335 |
+
"learning_rate": 0.0006826629028463891,
|
| 336 |
+
"loss": 2.6151,
|
| 337 |
+
"step": 2700
|
| 338 |
+
},
|
| 339 |
+
{
|
| 340 |
+
"epoch": 0.32,
|
| 341 |
+
"eval_gen_len": 18.8923,
|
| 342 |
+
"eval_loss": 2.3796188831329346,
|
| 343 |
+
"eval_rouge1": 29.8502,
|
| 344 |
+
"eval_rouge2": 8.9427,
|
| 345 |
+
"eval_rougeL": 23.6665,
|
| 346 |
+
"eval_rougeLsum": 23.6781,
|
| 347 |
+
"eval_runtime": 341.6707,
|
| 348 |
+
"eval_samples_per_second": 33.166,
|
| 349 |
+
"eval_steps_per_second": 1.384,
|
| 350 |
+
"step": 2700
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 0.34,
|
| 354 |
+
"learning_rate": 0.0006650199952952247,
|
| 355 |
+
"loss": 2.6123,
|
| 356 |
+
"step": 2850
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 0.34,
|
| 360 |
+
"eval_gen_len": 18.7541,
|
| 361 |
+
"eval_loss": 2.371220588684082,
|
| 362 |
+
"eval_rouge1": 30.0211,
|
| 363 |
+
"eval_rouge2": 9.2412,
|
| 364 |
+
"eval_rougeL": 24.0388,
|
| 365 |
+
"eval_rougeLsum": 24.045,
|
| 366 |
+
"eval_runtime": 341.6315,
|
| 367 |
+
"eval_samples_per_second": 33.17,
|
| 368 |
+
"eval_steps_per_second": 1.385,
|
| 369 |
+
"step": 2850
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 0.35,
|
| 373 |
+
"learning_rate": 0.0006473770877440603,
|
| 374 |
+
"loss": 2.5928,
|
| 375 |
+
"step": 3000
|
| 376 |
+
},
|
| 377 |
+
{
|
| 378 |
+
"epoch": 0.35,
|
| 379 |
+
"eval_gen_len": 18.7963,
|
| 380 |
+
"eval_loss": 2.359456777572632,
|
| 381 |
+
"eval_rouge1": 30.3185,
|
| 382 |
+
"eval_rouge2": 9.2796,
|
| 383 |
+
"eval_rougeL": 24.2004,
|
| 384 |
+
"eval_rougeLsum": 24.2101,
|
| 385 |
+
"eval_runtime": 341.8833,
|
| 386 |
+
"eval_samples_per_second": 33.146,
|
| 387 |
+
"eval_steps_per_second": 1.384,
|
| 388 |
+
"step": 3000
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 0.37,
|
| 392 |
+
"learning_rate": 0.0006297341801928958,
|
| 393 |
+
"loss": 2.5746,
|
| 394 |
+
"step": 3150
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 0.37,
|
| 398 |
+
"eval_gen_len": 18.8229,
|
| 399 |
+
"eval_loss": 2.366935968399048,
|
| 400 |
+
"eval_rouge1": 29.8303,
|
| 401 |
+
"eval_rouge2": 9.1665,
|
| 402 |
+
"eval_rougeL": 23.8553,
|
| 403 |
+
"eval_rougeLsum": 23.8672,
|
| 404 |
+
"eval_runtime": 341.7148,
|
| 405 |
+
"eval_samples_per_second": 33.162,
|
| 406 |
+
"eval_steps_per_second": 1.384,
|
| 407 |
+
"step": 3150
|
| 408 |
+
},
|
| 409 |
+
{
|
| 410 |
+
"epoch": 0.39,
|
| 411 |
+
"learning_rate": 0.0006120912726417314,
|
| 412 |
+
"loss": 2.5619,
|
| 413 |
+
"step": 3300
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 0.39,
|
| 417 |
+
"eval_gen_len": 18.8087,
|
| 418 |
+
"eval_loss": 2.3527944087982178,
|
| 419 |
+
"eval_rouge1": 30.4684,
|
| 420 |
+
"eval_rouge2": 9.4451,
|
| 421 |
+
"eval_rougeL": 24.2867,
|
| 422 |
+
"eval_rougeLsum": 24.2939,
|
| 423 |
+
"eval_runtime": 341.8038,
|
| 424 |
+
"eval_samples_per_second": 33.154,
|
| 425 |
+
"eval_steps_per_second": 1.384,
|
| 426 |
+
"step": 3300
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"epoch": 0.41,
|
| 430 |
+
"learning_rate": 0.000594448365090567,
|
| 431 |
+
"loss": 2.5677,
|
| 432 |
+
"step": 3450
|
| 433 |
+
},
|
| 434 |
+
{
|
| 435 |
+
"epoch": 0.41,
|
| 436 |
+
"eval_gen_len": 18.8491,
|
| 437 |
+
"eval_loss": 2.343312978744507,
|
| 438 |
+
"eval_rouge1": 30.496,
|
| 439 |
+
"eval_rouge2": 9.4356,
|
| 440 |
+
"eval_rougeL": 24.31,
|
| 441 |
+
"eval_rougeLsum": 24.3125,
|
| 442 |
+
"eval_runtime": 342.0548,
|
| 443 |
+
"eval_samples_per_second": 33.129,
|
| 444 |
+
"eval_steps_per_second": 1.383,
|
| 445 |
+
"step": 3450
|
| 446 |
+
},
|
| 447 |
+
{
|
| 448 |
+
"epoch": 0.42,
|
| 449 |
+
"learning_rate": 0.0005768054575394025,
|
| 450 |
+
"loss": 2.5456,
|
| 451 |
+
"step": 3600
|
| 452 |
+
},
|
| 453 |
+
{
|
| 454 |
+
"epoch": 0.42,
|
| 455 |
+
"eval_gen_len": 18.809,
|
| 456 |
+
"eval_loss": 2.3360698223114014,
|
| 457 |
+
"eval_rouge1": 30.5752,
|
| 458 |
+
"eval_rouge2": 9.4803,
|
| 459 |
+
"eval_rougeL": 24.3462,
|
| 460 |
+
"eval_rougeLsum": 24.3559,
|
| 461 |
+
"eval_runtime": 341.7402,
|
| 462 |
+
"eval_samples_per_second": 33.16,
|
| 463 |
+
"eval_steps_per_second": 1.384,
|
| 464 |
+
"step": 3600
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 0.44,
|
| 468 |
+
"learning_rate": 0.0005591625499882381,
|
| 469 |
+
"loss": 2.5669,
|
| 470 |
+
"step": 3750
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 0.44,
|
| 474 |
+
"eval_gen_len": 18.8287,
|
| 475 |
+
"eval_loss": 2.327768325805664,
|
| 476 |
+
"eval_rouge1": 30.0561,
|
| 477 |
+
"eval_rouge2": 9.2402,
|
| 478 |
+
"eval_rougeL": 24.0535,
|
| 479 |
+
"eval_rougeLsum": 24.0713,
|
| 480 |
+
"eval_runtime": 341.71,
|
| 481 |
+
"eval_samples_per_second": 33.163,
|
| 482 |
+
"eval_steps_per_second": 1.384,
|
| 483 |
+
"step": 3750
|
| 484 |
+
},
|
| 485 |
+
{
|
| 486 |
+
"epoch": 0.46,
|
| 487 |
+
"learning_rate": 0.0005415196424370737,
|
| 488 |
+
"loss": 2.5446,
|
| 489 |
+
"step": 3900
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 0.46,
|
| 493 |
+
"eval_gen_len": 18.8118,
|
| 494 |
+
"eval_loss": 2.3254404067993164,
|
| 495 |
+
"eval_rouge1": 30.3142,
|
| 496 |
+
"eval_rouge2": 9.4929,
|
| 497 |
+
"eval_rougeL": 24.335,
|
| 498 |
+
"eval_rougeLsum": 24.3432,
|
| 499 |
+
"eval_runtime": 341.8271,
|
| 500 |
+
"eval_samples_per_second": 33.151,
|
| 501 |
+
"eval_steps_per_second": 1.384,
|
| 502 |
+
"step": 3900
|
| 503 |
+
},
|
| 504 |
+
{
|
| 505 |
+
"epoch": 0.48,
|
| 506 |
+
"learning_rate": 0.0005238767348859092,
|
| 507 |
+
"loss": 2.5536,
|
| 508 |
+
"step": 4050
|
| 509 |
+
},
|
| 510 |
+
{
|
| 511 |
+
"epoch": 0.48,
|
| 512 |
+
"eval_gen_len": 18.9035,
|
| 513 |
+
"eval_loss": 2.3251070976257324,
|
| 514 |
+
"eval_rouge1": 30.2523,
|
| 515 |
+
"eval_rouge2": 9.3005,
|
| 516 |
+
"eval_rougeL": 24.163,
|
| 517 |
+
"eval_rougeLsum": 24.1711,
|
| 518 |
+
"eval_runtime": 341.8151,
|
| 519 |
+
"eval_samples_per_second": 33.152,
|
| 520 |
+
"eval_steps_per_second": 1.384,
|
| 521 |
+
"step": 4050
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 0.49,
|
| 525 |
+
"learning_rate": 0.0005062338273347448,
|
| 526 |
+
"loss": 2.5154,
|
| 527 |
+
"step": 4200
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 0.49,
|
| 531 |
+
"eval_gen_len": 18.7458,
|
| 532 |
+
"eval_loss": 2.313904047012329,
|
| 533 |
+
"eval_rouge1": 30.7718,
|
| 534 |
+
"eval_rouge2": 9.8676,
|
| 535 |
+
"eval_rougeL": 24.7111,
|
| 536 |
+
"eval_rougeLsum": 24.7266,
|
| 537 |
+
"eval_runtime": 341.6852,
|
| 538 |
+
"eval_samples_per_second": 33.165,
|
| 539 |
+
"eval_steps_per_second": 1.384,
|
| 540 |
+
"step": 4200
|
| 541 |
+
},
|
| 542 |
+
{
|
| 543 |
+
"epoch": 0.51,
|
| 544 |
+
"learning_rate": 0.0004885909197835804,
|
| 545 |
+
"loss": 2.5176,
|
| 546 |
+
"step": 4350
|
| 547 |
+
},
|
| 548 |
+
{
|
| 549 |
+
"epoch": 0.51,
|
| 550 |
+
"eval_gen_len": 18.7983,
|
| 551 |
+
"eval_loss": 2.309589385986328,
|
| 552 |
+
"eval_rouge1": 30.5617,
|
| 553 |
+
"eval_rouge2": 9.6562,
|
| 554 |
+
"eval_rougeL": 24.4754,
|
| 555 |
+
"eval_rougeLsum": 24.4862,
|
| 556 |
+
"eval_runtime": 341.8284,
|
| 557 |
+
"eval_samples_per_second": 33.151,
|
| 558 |
+
"eval_steps_per_second": 1.384,
|
| 559 |
+
"step": 4350
|
| 560 |
+
},
|
| 561 |
+
{
|
| 562 |
+
"epoch": 0.53,
|
| 563 |
+
"learning_rate": 0.0004709480122324159,
|
| 564 |
+
"loss": 2.5307,
|
| 565 |
+
"step": 4500
|
| 566 |
+
},
|
| 567 |
+
{
|
| 568 |
+
"epoch": 0.53,
|
| 569 |
+
"eval_gen_len": 18.8217,
|
| 570 |
+
"eval_loss": 2.3089160919189453,
|
| 571 |
+
"eval_rouge1": 30.601,
|
| 572 |
+
"eval_rouge2": 9.6672,
|
| 573 |
+
"eval_rougeL": 24.4465,
|
| 574 |
+
"eval_rougeLsum": 24.4567,
|
| 575 |
+
"eval_runtime": 341.6213,
|
| 576 |
+
"eval_samples_per_second": 33.171,
|
| 577 |
+
"eval_steps_per_second": 1.385,
|
| 578 |
+
"step": 4500
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 0.55,
|
| 582 |
+
"learning_rate": 0.00045330510468125144,
|
| 583 |
+
"loss": 2.515,
|
| 584 |
+
"step": 4650
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 0.55,
|
| 588 |
+
"eval_gen_len": 18.8096,
|
| 589 |
+
"eval_loss": 2.2991859912872314,
|
| 590 |
+
"eval_rouge1": 31.1807,
|
| 591 |
+
"eval_rouge2": 10.0105,
|
| 592 |
+
"eval_rougeL": 24.9947,
|
| 593 |
+
"eval_rougeLsum": 25.0096,
|
| 594 |
+
"eval_runtime": 341.7452,
|
| 595 |
+
"eval_samples_per_second": 33.159,
|
| 596 |
+
"eval_steps_per_second": 1.384,
|
| 597 |
+
"step": 4650
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 0.56,
|
| 601 |
+
"learning_rate": 0.000435662197130087,
|
| 602 |
+
"loss": 2.5168,
|
| 603 |
+
"step": 4800
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 0.56,
|
| 607 |
+
"eval_gen_len": 18.8254,
|
| 608 |
+
"eval_loss": 2.292003631591797,
|
| 609 |
+
"eval_rouge1": 30.8999,
|
| 610 |
+
"eval_rouge2": 9.8832,
|
| 611 |
+
"eval_rougeL": 24.7713,
|
| 612 |
+
"eval_rougeLsum": 24.7823,
|
| 613 |
+
"eval_runtime": 341.8542,
|
| 614 |
+
"eval_samples_per_second": 33.149,
|
| 615 |
+
"eval_steps_per_second": 1.384,
|
| 616 |
+
"step": 4800
|
| 617 |
+
},
|
| 618 |
+
{
|
| 619 |
+
"epoch": 0.58,
|
| 620 |
+
"learning_rate": 0.0004180192895789226,
|
| 621 |
+
"loss": 2.5021,
|
| 622 |
+
"step": 4950
|
| 623 |
+
},
|
| 624 |
+
{
|
| 625 |
+
"epoch": 0.58,
|
| 626 |
+
"eval_gen_len": 18.8787,
|
| 627 |
+
"eval_loss": 2.285507917404175,
|
| 628 |
+
"eval_rouge1": 31.0569,
|
| 629 |
+
"eval_rouge2": 9.8752,
|
| 630 |
+
"eval_rougeL": 24.791,
|
| 631 |
+
"eval_rougeLsum": 24.8033,
|
| 632 |
+
"eval_runtime": 341.7747,
|
| 633 |
+
"eval_samples_per_second": 33.156,
|
| 634 |
+
"eval_steps_per_second": 1.384,
|
| 635 |
+
"step": 4950
|
| 636 |
+
},
|
| 637 |
+
{
|
| 638 |
+
"epoch": 0.6,
|
| 639 |
+
"learning_rate": 0.00040037638202775815,
|
| 640 |
+
"loss": 2.501,
|
| 641 |
+
"step": 5100
|
| 642 |
+
},
|
| 643 |
+
{
|
| 644 |
+
"epoch": 0.6,
|
| 645 |
+
"eval_gen_len": 18.8315,
|
| 646 |
+
"eval_loss": 2.2868235111236572,
|
| 647 |
+
"eval_rouge1": 30.7132,
|
| 648 |
+
"eval_rouge2": 9.8654,
|
| 649 |
+
"eval_rougeL": 24.6084,
|
| 650 |
+
"eval_rougeLsum": 24.6244,
|
| 651 |
+
"eval_runtime": 341.9154,
|
| 652 |
+
"eval_samples_per_second": 33.143,
|
| 653 |
+
"eval_steps_per_second": 1.383,
|
| 654 |
+
"step": 5100
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 0.62,
|
| 658 |
+
"learning_rate": 0.0003827334744765937,
|
| 659 |
+
"loss": 2.4849,
|
| 660 |
+
"step": 5250
|
| 661 |
+
},
|
| 662 |
+
{
|
| 663 |
+
"epoch": 0.62,
|
| 664 |
+
"eval_gen_len": 18.8196,
|
| 665 |
+
"eval_loss": 2.2783043384552,
|
| 666 |
+
"eval_rouge1": 31.3434,
|
| 667 |
+
"eval_rouge2": 10.206,
|
| 668 |
+
"eval_rougeL": 25.0954,
|
| 669 |
+
"eval_rougeLsum": 25.1114,
|
| 670 |
+
"eval_runtime": 341.9076,
|
| 671 |
+
"eval_samples_per_second": 33.143,
|
| 672 |
+
"eval_steps_per_second": 1.383,
|
| 673 |
+
"step": 5250
|
| 674 |
+
},
|
| 675 |
+
{
|
| 676 |
+
"epoch": 0.64,
|
| 677 |
+
"learning_rate": 0.0003650905669254293,
|
| 678 |
+
"loss": 2.4939,
|
| 679 |
+
"step": 5400
|
| 680 |
+
},
|
| 681 |
+
{
|
| 682 |
+
"epoch": 0.64,
|
| 683 |
+
"eval_gen_len": 18.8579,
|
| 684 |
+
"eval_loss": 2.275907039642334,
|
| 685 |
+
"eval_rouge1": 31.1467,
|
| 686 |
+
"eval_rouge2": 10.0457,
|
| 687 |
+
"eval_rougeL": 24.964,
|
| 688 |
+
"eval_rougeLsum": 24.9793,
|
| 689 |
+
"eval_runtime": 341.6475,
|
| 690 |
+
"eval_samples_per_second": 33.169,
|
| 691 |
+
"eval_steps_per_second": 1.384,
|
| 692 |
+
"step": 5400
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 0.65,
|
| 696 |
+
"learning_rate": 0.00034744765937426485,
|
| 697 |
+
"loss": 2.4624,
|
| 698 |
+
"step": 5550
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 0.65,
|
| 702 |
+
"eval_gen_len": 18.8098,
|
| 703 |
+
"eval_loss": 2.2712931632995605,
|
| 704 |
+
"eval_rouge1": 31.4288,
|
| 705 |
+
"eval_rouge2": 10.1719,
|
| 706 |
+
"eval_rougeL": 25.1096,
|
| 707 |
+
"eval_rougeLsum": 25.1276,
|
| 708 |
+
"eval_runtime": 341.723,
|
| 709 |
+
"eval_samples_per_second": 33.161,
|
| 710 |
+
"eval_steps_per_second": 1.384,
|
| 711 |
+
"step": 5550
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"epoch": 0.67,
|
| 715 |
+
"learning_rate": 0.0003298047518231004,
|
| 716 |
+
"loss": 2.456,
|
| 717 |
+
"step": 5700
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 0.67,
|
| 721 |
+
"eval_gen_len": 18.8262,
|
| 722 |
+
"eval_loss": 2.2674217224121094,
|
| 723 |
+
"eval_rouge1": 31.1515,
|
| 724 |
+
"eval_rouge2": 10.1208,
|
| 725 |
+
"eval_rougeL": 25.0456,
|
| 726 |
+
"eval_rougeLsum": 25.0532,
|
| 727 |
+
"eval_runtime": 341.9008,
|
| 728 |
+
"eval_samples_per_second": 33.144,
|
| 729 |
+
"eval_steps_per_second": 1.383,
|
| 730 |
+
"step": 5700
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 0.69,
|
| 734 |
+
"learning_rate": 0.000312161844271936,
|
| 735 |
+
"loss": 2.4667,
|
| 736 |
+
"step": 5850
|
| 737 |
+
},
|
| 738 |
+
{
|
| 739 |
+
"epoch": 0.69,
|
| 740 |
+
"eval_gen_len": 18.7816,
|
| 741 |
+
"eval_loss": 2.262035608291626,
|
| 742 |
+
"eval_rouge1": 31.3741,
|
| 743 |
+
"eval_rouge2": 10.1733,
|
| 744 |
+
"eval_rougeL": 25.1421,
|
| 745 |
+
"eval_rougeLsum": 25.159,
|
| 746 |
+
"eval_runtime": 341.6066,
|
| 747 |
+
"eval_samples_per_second": 33.173,
|
| 748 |
+
"eval_steps_per_second": 1.385,
|
| 749 |
+
"step": 5850
|
| 750 |
+
},
|
| 751 |
+
{
|
| 752 |
+
"epoch": 0.71,
|
| 753 |
+
"learning_rate": 0.000294636556104446,
|
| 754 |
+
"loss": 2.4658,
|
| 755 |
+
"step": 6000
|
| 756 |
+
},
|
| 757 |
+
{
|
| 758 |
+
"epoch": 0.71,
|
| 759 |
+
"eval_gen_len": 18.8,
|
| 760 |
+
"eval_loss": 2.258843183517456,
|
| 761 |
+
"eval_rouge1": 31.3913,
|
| 762 |
+
"eval_rouge2": 10.2645,
|
| 763 |
+
"eval_rougeL": 25.1746,
|
| 764 |
+
"eval_rougeLsum": 25.188,
|
| 765 |
+
"eval_runtime": 341.9247,
|
| 766 |
+
"eval_samples_per_second": 33.142,
|
| 767 |
+
"eval_steps_per_second": 1.383,
|
| 768 |
+
"step": 6000
|
| 769 |
+
},
|
| 770 |
+
{
|
| 771 |
+
"epoch": 0.72,
|
| 772 |
+
"learning_rate": 0.00027699364855328156,
|
| 773 |
+
"loss": 2.4943,
|
| 774 |
+
"step": 6150
|
| 775 |
+
},
|
| 776 |
+
{
|
| 777 |
+
"epoch": 0.72,
|
| 778 |
+
"eval_gen_len": 18.8074,
|
| 779 |
+
"eval_loss": 2.2533156871795654,
|
| 780 |
+
"eval_rouge1": 31.3905,
|
| 781 |
+
"eval_rouge2": 10.164,
|
| 782 |
+
"eval_rougeL": 25.093,
|
| 783 |
+
"eval_rougeLsum": 25.107,
|
| 784 |
+
"eval_runtime": 341.8317,
|
| 785 |
+
"eval_samples_per_second": 33.151,
|
| 786 |
+
"eval_steps_per_second": 1.384,
|
| 787 |
+
"step": 6150
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 0.74,
|
| 791 |
+
"learning_rate": 0.00025935074100211713,
|
| 792 |
+
"loss": 2.473,
|
| 793 |
+
"step": 6300
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 0.74,
|
| 797 |
+
"eval_gen_len": 18.7439,
|
| 798 |
+
"eval_loss": 2.2551848888397217,
|
| 799 |
+
"eval_rouge1": 31.1105,
|
| 800 |
+
"eval_rouge2": 10.1939,
|
| 801 |
+
"eval_rougeL": 24.9214,
|
| 802 |
+
"eval_rougeLsum": 24.9321,
|
| 803 |
+
"eval_runtime": 341.6034,
|
| 804 |
+
"eval_samples_per_second": 33.173,
|
| 805 |
+
"eval_steps_per_second": 1.385,
|
| 806 |
+
"step": 6300
|
| 807 |
+
},
|
| 808 |
+
{
|
| 809 |
+
"epoch": 0.76,
|
| 810 |
+
"learning_rate": 0.00024170783345095273,
|
| 811 |
+
"loss": 2.4687,
|
| 812 |
+
"step": 6450
|
| 813 |
+
},
|
| 814 |
+
{
|
| 815 |
+
"epoch": 0.76,
|
| 816 |
+
"eval_gen_len": 18.7709,
|
| 817 |
+
"eval_loss": 2.247098684310913,
|
| 818 |
+
"eval_rouge1": 31.3387,
|
| 819 |
+
"eval_rouge2": 10.2446,
|
| 820 |
+
"eval_rougeL": 25.0746,
|
| 821 |
+
"eval_rougeLsum": 25.0936,
|
| 822 |
+
"eval_runtime": 341.7347,
|
| 823 |
+
"eval_samples_per_second": 33.16,
|
| 824 |
+
"eval_steps_per_second": 1.384,
|
| 825 |
+
"step": 6450
|
| 826 |
+
},
|
| 827 |
+
{
|
| 828 |
+
"epoch": 0.78,
|
| 829 |
+
"learning_rate": 0.0002240649258997883,
|
| 830 |
+
"loss": 2.4286,
|
| 831 |
+
"step": 6600
|
| 832 |
+
},
|
| 833 |
+
{
|
| 834 |
+
"epoch": 0.78,
|
| 835 |
+
"eval_gen_len": 18.8313,
|
| 836 |
+
"eval_loss": 2.2467916011810303,
|
| 837 |
+
"eval_rouge1": 31.1953,
|
| 838 |
+
"eval_rouge2": 10.1277,
|
| 839 |
+
"eval_rougeL": 24.9535,
|
| 840 |
+
"eval_rougeLsum": 24.9714,
|
| 841 |
+
"eval_runtime": 341.8485,
|
| 842 |
+
"eval_samples_per_second": 33.149,
|
| 843 |
+
"eval_steps_per_second": 1.384,
|
| 844 |
+
"step": 6600
|
| 845 |
+
},
|
| 846 |
+
{
|
| 847 |
+
"epoch": 0.79,
|
| 848 |
+
"learning_rate": 0.00020642201834862386,
|
| 849 |
+
"loss": 2.4492,
|
| 850 |
+
"step": 6750
|
| 851 |
+
},
|
| 852 |
+
{
|
| 853 |
+
"epoch": 0.79,
|
| 854 |
+
"eval_gen_len": 18.7977,
|
| 855 |
+
"eval_loss": 2.2422139644622803,
|
| 856 |
+
"eval_rouge1": 31.7203,
|
| 857 |
+
"eval_rouge2": 10.4934,
|
| 858 |
+
"eval_rougeL": 25.3862,
|
| 859 |
+
"eval_rougeLsum": 25.3946,
|
| 860 |
+
"eval_runtime": 341.6909,
|
| 861 |
+
"eval_samples_per_second": 33.164,
|
| 862 |
+
"eval_steps_per_second": 1.384,
|
| 863 |
+
"step": 6750
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 0.81,
|
| 867 |
+
"learning_rate": 0.00018877911079745943,
|
| 868 |
+
"loss": 2.4427,
|
| 869 |
+
"step": 6900
|
| 870 |
+
},
|
| 871 |
+
{
|
| 872 |
+
"epoch": 0.81,
|
| 873 |
+
"eval_gen_len": 18.8144,
|
| 874 |
+
"eval_loss": 2.2402756214141846,
|
| 875 |
+
"eval_rouge1": 31.5498,
|
| 876 |
+
"eval_rouge2": 10.4086,
|
| 877 |
+
"eval_rougeL": 25.2384,
|
| 878 |
+
"eval_rougeLsum": 25.2593,
|
| 879 |
+
"eval_runtime": 341.7269,
|
| 880 |
+
"eval_samples_per_second": 33.161,
|
| 881 |
+
"eval_steps_per_second": 1.384,
|
| 882 |
+
"step": 6900
|
| 883 |
+
},
|
| 884 |
+
{
|
| 885 |
+
"epoch": 0.83,
|
| 886 |
+
"learning_rate": 0.000171136203246295,
|
| 887 |
+
"loss": 2.4641,
|
| 888 |
+
"step": 7050
|
| 889 |
+
},
|
| 890 |
+
{
|
| 891 |
+
"epoch": 0.83,
|
| 892 |
+
"eval_gen_len": 18.7966,
|
| 893 |
+
"eval_loss": 2.2365546226501465,
|
| 894 |
+
"eval_rouge1": 31.4038,
|
| 895 |
+
"eval_rouge2": 10.3691,
|
| 896 |
+
"eval_rougeL": 25.1786,
|
| 897 |
+
"eval_rougeLsum": 25.195,
|
| 898 |
+
"eval_runtime": 341.8284,
|
| 899 |
+
"eval_samples_per_second": 33.151,
|
| 900 |
+
"eval_steps_per_second": 1.384,
|
| 901 |
+
"step": 7050
|
| 902 |
+
},
|
| 903 |
+
{
|
| 904 |
+
"epoch": 0.85,
|
| 905 |
+
"learning_rate": 0.00015349329569513056,
|
| 906 |
+
"loss": 2.4276,
|
| 907 |
+
"step": 7200
|
| 908 |
+
},
|
| 909 |
+
{
|
| 910 |
+
"epoch": 0.85,
|
| 911 |
+
"eval_gen_len": 18.7789,
|
| 912 |
+
"eval_loss": 2.2344412803649902,
|
| 913 |
+
"eval_rouge1": 31.5076,
|
| 914 |
+
"eval_rouge2": 10.4403,
|
| 915 |
+
"eval_rougeL": 25.299,
|
| 916 |
+
"eval_rougeLsum": 25.3201,
|
| 917 |
+
"eval_runtime": 342.6233,
|
| 918 |
+
"eval_samples_per_second": 33.074,
|
| 919 |
+
"eval_steps_per_second": 1.381,
|
| 920 |
+
"step": 7200
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 0.86,
|
| 924 |
+
"learning_rate": 0.00013585038814396613,
|
| 925 |
+
"loss": 2.4402,
|
| 926 |
+
"step": 7350
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 0.86,
|
| 930 |
+
"eval_gen_len": 18.7713,
|
| 931 |
+
"eval_loss": 2.2317440509796143,
|
| 932 |
+
"eval_rouge1": 31.5216,
|
| 933 |
+
"eval_rouge2": 10.3901,
|
| 934 |
+
"eval_rougeL": 25.276,
|
| 935 |
+
"eval_rougeLsum": 25.2943,
|
| 936 |
+
"eval_runtime": 345.4971,
|
| 937 |
+
"eval_samples_per_second": 32.799,
|
| 938 |
+
"eval_steps_per_second": 1.369,
|
| 939 |
+
"step": 7350
|
| 940 |
+
},
|
| 941 |
+
{
|
| 942 |
+
"epoch": 0.88,
|
| 943 |
+
"learning_rate": 0.00011820748059280171,
|
| 944 |
+
"loss": 2.44,
|
| 945 |
+
"step": 7500
|
| 946 |
+
},
|
| 947 |
+
{
|
| 948 |
+
"epoch": 0.88,
|
| 949 |
+
"eval_gen_len": 18.7671,
|
| 950 |
+
"eval_loss": 2.2292771339416504,
|
| 951 |
+
"eval_rouge1": 31.4244,
|
| 952 |
+
"eval_rouge2": 10.4211,
|
| 953 |
+
"eval_rougeL": 25.2592,
|
| 954 |
+
"eval_rougeLsum": 25.2735,
|
| 955 |
+
"eval_runtime": 342.0133,
|
| 956 |
+
"eval_samples_per_second": 33.133,
|
| 957 |
+
"eval_steps_per_second": 1.383,
|
| 958 |
+
"step": 7500
|
| 959 |
+
},
|
| 960 |
+
{
|
| 961 |
+
"epoch": 0.9,
|
| 962 |
+
"learning_rate": 0.00010056457304163728,
|
| 963 |
+
"loss": 2.4251,
|
| 964 |
+
"step": 7650
|
| 965 |
+
},
|
| 966 |
+
{
|
| 967 |
+
"epoch": 0.9,
|
| 968 |
+
"eval_gen_len": 18.7972,
|
| 969 |
+
"eval_loss": 2.226907968521118,
|
| 970 |
+
"eval_rouge1": 31.4887,
|
| 971 |
+
"eval_rouge2": 10.3959,
|
| 972 |
+
"eval_rougeL": 25.2335,
|
| 973 |
+
"eval_rougeLsum": 25.2545,
|
| 974 |
+
"eval_runtime": 341.7402,
|
| 975 |
+
"eval_samples_per_second": 33.16,
|
| 976 |
+
"eval_steps_per_second": 1.384,
|
| 977 |
+
"step": 7650
|
| 978 |
+
},
|
| 979 |
+
{
|
| 980 |
+
"epoch": 0.92,
|
| 981 |
+
"learning_rate": 8.292166549047284e-05,
|
| 982 |
+
"loss": 2.456,
|
| 983 |
+
"step": 7800
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 0.92,
|
| 987 |
+
"eval_gen_len": 18.786,
|
| 988 |
+
"eval_loss": 2.224229097366333,
|
| 989 |
+
"eval_rouge1": 31.4508,
|
| 990 |
+
"eval_rouge2": 10.4079,
|
| 991 |
+
"eval_rougeL": 25.2328,
|
| 992 |
+
"eval_rougeLsum": 25.2564,
|
| 993 |
+
"eval_runtime": 341.8379,
|
| 994 |
+
"eval_samples_per_second": 33.15,
|
| 995 |
+
"eval_steps_per_second": 1.384,
|
| 996 |
+
"step": 7800
|
| 997 |
+
},
|
| 998 |
+
{
|
| 999 |
+
"epoch": 0.94,
|
| 1000 |
+
"learning_rate": 6.527875793930841e-05,
|
| 1001 |
+
"loss": 2.4181,
|
| 1002 |
+
"step": 7950
|
| 1003 |
+
},
|
| 1004 |
+
{
|
| 1005 |
+
"epoch": 0.94,
|
| 1006 |
+
"eval_gen_len": 18.8012,
|
| 1007 |
+
"eval_loss": 2.2223522663116455,
|
| 1008 |
+
"eval_rouge1": 31.6181,
|
| 1009 |
+
"eval_rouge2": 10.5558,
|
| 1010 |
+
"eval_rougeL": 25.3867,
|
| 1011 |
+
"eval_rougeLsum": 25.4042,
|
| 1012 |
+
"eval_runtime": 342.1259,
|
| 1013 |
+
"eval_samples_per_second": 33.122,
|
| 1014 |
+
"eval_steps_per_second": 1.383,
|
| 1015 |
+
"step": 7950
|
| 1016 |
+
},
|
| 1017 |
+
{
|
| 1018 |
+
"epoch": 0.95,
|
| 1019 |
+
"learning_rate": 4.763585038814397e-05,
|
| 1020 |
+
"loss": 2.4288,
|
| 1021 |
+
"step": 8100
|
| 1022 |
+
},
|
| 1023 |
+
{
|
| 1024 |
+
"epoch": 0.95,
|
| 1025 |
+
"eval_gen_len": 18.7953,
|
| 1026 |
+
"eval_loss": 2.22170352935791,
|
| 1027 |
+
"eval_rouge1": 31.7626,
|
| 1028 |
+
"eval_rouge2": 10.6059,
|
| 1029 |
+
"eval_rougeL": 25.4827,
|
| 1030 |
+
"eval_rougeLsum": 25.4958,
|
| 1031 |
+
"eval_runtime": 341.8728,
|
| 1032 |
+
"eval_samples_per_second": 33.147,
|
| 1033 |
+
"eval_steps_per_second": 1.384,
|
| 1034 |
+
"step": 8100
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 0.97,
|
| 1038 |
+
"learning_rate": 2.9992942836979537e-05,
|
| 1039 |
+
"loss": 2.4327,
|
| 1040 |
+
"step": 8250
|
| 1041 |
+
},
|
| 1042 |
+
{
|
| 1043 |
+
"epoch": 0.97,
|
| 1044 |
+
"eval_gen_len": 18.7827,
|
| 1045 |
+
"eval_loss": 2.220174789428711,
|
| 1046 |
+
"eval_rouge1": 31.6839,
|
| 1047 |
+
"eval_rouge2": 10.5615,
|
| 1048 |
+
"eval_rougeL": 25.4137,
|
| 1049 |
+
"eval_rougeLsum": 25.433,
|
| 1050 |
+
"eval_runtime": 342.1089,
|
| 1051 |
+
"eval_samples_per_second": 33.124,
|
| 1052 |
+
"eval_steps_per_second": 1.383,
|
| 1053 |
+
"step": 8250
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 0.99,
|
| 1057 |
+
"learning_rate": 1.2350035285815103e-05,
|
| 1058 |
+
"loss": 2.4118,
|
| 1059 |
+
"step": 8400
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 0.99,
|
| 1063 |
+
"eval_gen_len": 18.7979,
|
| 1064 |
+
"eval_loss": 2.2196593284606934,
|
| 1065 |
+
"eval_rouge1": 31.6519,
|
| 1066 |
+
"eval_rouge2": 10.4949,
|
| 1067 |
+
"eval_rougeL": 25.3751,
|
| 1068 |
+
"eval_rougeLsum": 25.3984,
|
| 1069 |
+
"eval_runtime": 342.1132,
|
| 1070 |
+
"eval_samples_per_second": 33.124,
|
| 1071 |
+
"eval_steps_per_second": 1.383,
|
| 1072 |
+
"step": 8400
|
| 1073 |
+
}
|
| 1074 |
+
],
|
| 1075 |
+
"max_steps": 8502,
|
| 1076 |
+
"num_train_epochs": 1,
|
| 1077 |
+
"total_flos": 4.261943991730176e+16,
|
| 1078 |
+
"trial_name": null,
|
| 1079 |
+
"trial_params": null
|
| 1080 |
+
}
|