End of training
Browse files- all_results.json +14 -0
- eval_results.json +9 -0
- train_results.json +8 -0
- trainer_state.json +76 -0
all_results.json
ADDED
|
@@ -0,0 +1,14 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 5.0,
|
| 3 |
+
"eval_loss": 0.160756453871727,
|
| 4 |
+
"eval_runtime": 90.9822,
|
| 5 |
+
"eval_samples": 1647,
|
| 6 |
+
"eval_samples_per_second": 18.102,
|
| 7 |
+
"eval_steps_per_second": 9.057,
|
| 8 |
+
"eval_wer": 0.1395158819323869,
|
| 9 |
+
"train_loss": 1.1678272247314454,
|
| 10 |
+
"train_runtime": 1371.4783,
|
| 11 |
+
"train_samples": 3478,
|
| 12 |
+
"train_samples_per_second": 12.68,
|
| 13 |
+
"train_steps_per_second": 0.529
|
| 14 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 5.0,
|
| 3 |
+
"eval_loss": 0.160756453871727,
|
| 4 |
+
"eval_runtime": 90.9822,
|
| 5 |
+
"eval_samples": 1647,
|
| 6 |
+
"eval_samples_per_second": 18.102,
|
| 7 |
+
"eval_steps_per_second": 9.057,
|
| 8 |
+
"eval_wer": 0.1395158819323869
|
| 9 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 5.0,
|
| 3 |
+
"train_loss": 1.1678272247314454,
|
| 4 |
+
"train_runtime": 1371.4783,
|
| 5 |
+
"train_samples": 3478,
|
| 6 |
+
"train_samples_per_second": 12.68,
|
| 7 |
+
"train_steps_per_second": 0.529
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,76 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 5.0,
|
| 5 |
+
"global_step": 725,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.69,
|
| 12 |
+
"learning_rate": 3.75e-07,
|
| 13 |
+
"loss": 1.2459,
|
| 14 |
+
"step": 100
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 1.38,
|
| 18 |
+
"learning_rate": 7.5e-07,
|
| 19 |
+
"loss": 1.2298,
|
| 20 |
+
"step": 200
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 2.07,
|
| 24 |
+
"learning_rate": 1.125e-06,
|
| 25 |
+
"loss": 1.19,
|
| 26 |
+
"step": 300
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 2.76,
|
| 30 |
+
"learning_rate": 1.5e-06,
|
| 31 |
+
"loss": 1.1671,
|
| 32 |
+
"step": 400
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 3.45,
|
| 36 |
+
"learning_rate": 1.875e-06,
|
| 37 |
+
"loss": 1.1224,
|
| 38 |
+
"step": 500
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 3.45,
|
| 42 |
+
"eval_loss": 0.1640530526638031,
|
| 43 |
+
"eval_runtime": 91.0313,
|
| 44 |
+
"eval_samples_per_second": 18.093,
|
| 45 |
+
"eval_steps_per_second": 9.052,
|
| 46 |
+
"eval_wer": 0.13961801654580738,
|
| 47 |
+
"step": 500
|
| 48 |
+
},
|
| 49 |
+
{
|
| 50 |
+
"epoch": 4.14,
|
| 51 |
+
"learning_rate": 2.25e-06,
|
| 52 |
+
"loss": 1.1193,
|
| 53 |
+
"step": 600
|
| 54 |
+
},
|
| 55 |
+
{
|
| 56 |
+
"epoch": 4.83,
|
| 57 |
+
"learning_rate": 2.625e-06,
|
| 58 |
+
"loss": 1.1034,
|
| 59 |
+
"step": 700
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 5.0,
|
| 63 |
+
"step": 725,
|
| 64 |
+
"total_flos": 6.315878094644183e+17,
|
| 65 |
+
"train_loss": 1.1678272247314454,
|
| 66 |
+
"train_runtime": 1371.4783,
|
| 67 |
+
"train_samples_per_second": 12.68,
|
| 68 |
+
"train_steps_per_second": 0.529
|
| 69 |
+
}
|
| 70 |
+
],
|
| 71 |
+
"max_steps": 725,
|
| 72 |
+
"num_train_epochs": 5,
|
| 73 |
+
"total_flos": 6.315878094644183e+17,
|
| 74 |
+
"trial_name": null,
|
| 75 |
+
"trial_params": null
|
| 76 |
+
}
|