Commit
·
19aada7
1
Parent(s):
931ea52
End of training
Browse files- all_results.json +14 -0
- eval_results.json +9 -0
- pytorch_model.bin +1 -1
- train_results.json +8 -0
- trainer_state.json +1024 -0
all_results.json
ADDED
|
@@ -0,0 +1,14 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 100.0,
|
| 3 |
+
"eval_loss": 0.18918171525001526,
|
| 4 |
+
"eval_runtime": 100.9072,
|
| 5 |
+
"eval_samples": 2084,
|
| 6 |
+
"eval_samples_per_second": 20.653,
|
| 7 |
+
"eval_steps_per_second": 20.653,
|
| 8 |
+
"eval_wer": 0.16977498479627,
|
| 9 |
+
"train_loss": 1.1756210368718856,
|
| 10 |
+
"train_runtime": 18507.9964,
|
| 11 |
+
"train_samples": 4963,
|
| 12 |
+
"train_samples_per_second": 26.815,
|
| 13 |
+
"train_steps_per_second": 0.843
|
| 14 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 100.0,
|
| 3 |
+
"eval_loss": 0.18918171525001526,
|
| 4 |
+
"eval_runtime": 100.9072,
|
| 5 |
+
"eval_samples": 2084,
|
| 6 |
+
"eval_samples_per_second": 20.653,
|
| 7 |
+
"eval_steps_per_second": 20.653,
|
| 8 |
+
"eval_wer": 0.16977498479627
|
| 9 |
+
}
|
pytorch_model.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1262083569
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f5025805755cb647a7150a1cea815b7586094bde0cdcee51d48b2095bf500038
|
| 3 |
size 1262083569
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 100.0,
|
| 3 |
+
"train_loss": 1.1756210368718856,
|
| 4 |
+
"train_runtime": 18507.9964,
|
| 5 |
+
"train_samples": 4963,
|
| 6 |
+
"train_samples_per_second": 26.815,
|
| 7 |
+
"train_steps_per_second": 0.843
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1024 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 100.0,
|
| 5 |
+
"global_step": 15600,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.64,
|
| 12 |
+
"learning_rate": 3.3949999999999997e-06,
|
| 13 |
+
"loss": 18.8158,
|
| 14 |
+
"step": 100
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 1.28,
|
| 18 |
+
"learning_rate": 6.895e-06,
|
| 19 |
+
"loss": 9.3721,
|
| 20 |
+
"step": 200
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 1.92,
|
| 24 |
+
"learning_rate": 1.0394999999999998e-05,
|
| 25 |
+
"loss": 5.1453,
|
| 26 |
+
"step": 300
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 2.56,
|
| 30 |
+
"learning_rate": 1.3895e-05,
|
| 31 |
+
"loss": 4.2355,
|
| 32 |
+
"step": 400
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 3.21,
|
| 36 |
+
"learning_rate": 1.7395e-05,
|
| 37 |
+
"loss": 3.7107,
|
| 38 |
+
"step": 500
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 3.85,
|
| 42 |
+
"learning_rate": 2.0894999999999996e-05,
|
| 43 |
+
"loss": 3.3647,
|
| 44 |
+
"step": 600
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 4.49,
|
| 48 |
+
"learning_rate": 2.4394999999999996e-05,
|
| 49 |
+
"loss": 3.2191,
|
| 50 |
+
"step": 700
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 5.13,
|
| 54 |
+
"learning_rate": 2.7895e-05,
|
| 55 |
+
"loss": 3.1339,
|
| 56 |
+
"step": 800
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 5.77,
|
| 60 |
+
"learning_rate": 3.1395e-05,
|
| 61 |
+
"loss": 3.0484,
|
| 62 |
+
"step": 900
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 6.41,
|
| 66 |
+
"learning_rate": 3.4895e-05,
|
| 67 |
+
"loss": 3.0086,
|
| 68 |
+
"step": 1000
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 7.05,
|
| 72 |
+
"learning_rate": 3.8394999999999994e-05,
|
| 73 |
+
"loss": 2.9794,
|
| 74 |
+
"step": 1100
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 7.69,
|
| 78 |
+
"learning_rate": 4.1895e-05,
|
| 79 |
+
"loss": 2.9544,
|
| 80 |
+
"step": 1200
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 8.33,
|
| 84 |
+
"learning_rate": 4.5394999999999995e-05,
|
| 85 |
+
"loss": 2.901,
|
| 86 |
+
"step": 1300
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 8.97,
|
| 90 |
+
"learning_rate": 4.8895e-05,
|
| 91 |
+
"loss": 2.6256,
|
| 92 |
+
"step": 1400
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 9.62,
|
| 96 |
+
"learning_rate": 5.2395e-05,
|
| 97 |
+
"loss": 2.0673,
|
| 98 |
+
"step": 1500
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 10.26,
|
| 102 |
+
"learning_rate": 5.589499999999999e-05,
|
| 103 |
+
"loss": 1.763,
|
| 104 |
+
"step": 1600
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 10.9,
|
| 108 |
+
"learning_rate": 5.9394999999999996e-05,
|
| 109 |
+
"loss": 1.6053,
|
| 110 |
+
"step": 1700
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 11.54,
|
| 114 |
+
"learning_rate": 6.289499999999999e-05,
|
| 115 |
+
"loss": 1.5242,
|
| 116 |
+
"step": 1800
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 12.18,
|
| 120 |
+
"learning_rate": 6.639499999999999e-05,
|
| 121 |
+
"loss": 1.4635,
|
| 122 |
+
"step": 1900
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 12.82,
|
| 126 |
+
"learning_rate": 6.9895e-05,
|
| 127 |
+
"loss": 1.4235,
|
| 128 |
+
"step": 2000
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 12.82,
|
| 132 |
+
"eval_loss": 0.4475284516811371,
|
| 133 |
+
"eval_runtime": 99.8092,
|
| 134 |
+
"eval_samples_per_second": 20.88,
|
| 135 |
+
"eval_steps_per_second": 20.88,
|
| 136 |
+
"eval_wer": 0.455098317453882,
|
| 137 |
+
"step": 2000
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 13.46,
|
| 141 |
+
"learning_rate": 6.950073529411765e-05,
|
| 142 |
+
"loss": 1.3615,
|
| 143 |
+
"step": 2100
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 14.1,
|
| 147 |
+
"learning_rate": 6.89860294117647e-05,
|
| 148 |
+
"loss": 1.2973,
|
| 149 |
+
"step": 2200
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 14.74,
|
| 153 |
+
"learning_rate": 6.847132352941176e-05,
|
| 154 |
+
"loss": 1.261,
|
| 155 |
+
"step": 2300
|
| 156 |
+
},
|
| 157 |
+
{
|
| 158 |
+
"epoch": 15.38,
|
| 159 |
+
"learning_rate": 6.795661764705882e-05,
|
| 160 |
+
"loss": 1.2356,
|
| 161 |
+
"step": 2400
|
| 162 |
+
},
|
| 163 |
+
{
|
| 164 |
+
"epoch": 16.03,
|
| 165 |
+
"learning_rate": 6.744191176470588e-05,
|
| 166 |
+
"loss": 1.1895,
|
| 167 |
+
"step": 2500
|
| 168 |
+
},
|
| 169 |
+
{
|
| 170 |
+
"epoch": 16.67,
|
| 171 |
+
"learning_rate": 6.692720588235293e-05,
|
| 172 |
+
"loss": 1.144,
|
| 173 |
+
"step": 2600
|
| 174 |
+
},
|
| 175 |
+
{
|
| 176 |
+
"epoch": 17.31,
|
| 177 |
+
"learning_rate": 6.64125e-05,
|
| 178 |
+
"loss": 1.1081,
|
| 179 |
+
"step": 2700
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 17.95,
|
| 183 |
+
"learning_rate": 6.589779411764706e-05,
|
| 184 |
+
"loss": 1.1043,
|
| 185 |
+
"step": 2800
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 18.59,
|
| 189 |
+
"learning_rate": 6.538308823529412e-05,
|
| 190 |
+
"loss": 1.074,
|
| 191 |
+
"step": 2900
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 19.23,
|
| 195 |
+
"learning_rate": 6.486838235294117e-05,
|
| 196 |
+
"loss": 1.0529,
|
| 197 |
+
"step": 3000
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 19.87,
|
| 201 |
+
"learning_rate": 6.435367647058823e-05,
|
| 202 |
+
"loss": 1.0452,
|
| 203 |
+
"step": 3100
|
| 204 |
+
},
|
| 205 |
+
{
|
| 206 |
+
"epoch": 20.51,
|
| 207 |
+
"learning_rate": 6.383897058823529e-05,
|
| 208 |
+
"loss": 1.0405,
|
| 209 |
+
"step": 3200
|
| 210 |
+
},
|
| 211 |
+
{
|
| 212 |
+
"epoch": 21.15,
|
| 213 |
+
"learning_rate": 6.332426470588235e-05,
|
| 214 |
+
"loss": 1.0131,
|
| 215 |
+
"step": 3300
|
| 216 |
+
},
|
| 217 |
+
{
|
| 218 |
+
"epoch": 21.79,
|
| 219 |
+
"learning_rate": 6.28095588235294e-05,
|
| 220 |
+
"loss": 1.0007,
|
| 221 |
+
"step": 3400
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 22.44,
|
| 225 |
+
"learning_rate": 6.229485294117647e-05,
|
| 226 |
+
"loss": 0.9744,
|
| 227 |
+
"step": 3500
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 23.08,
|
| 231 |
+
"learning_rate": 6.178014705882353e-05,
|
| 232 |
+
"loss": 0.9773,
|
| 233 |
+
"step": 3600
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 23.72,
|
| 237 |
+
"learning_rate": 6.126544117647059e-05,
|
| 238 |
+
"loss": 0.9767,
|
| 239 |
+
"step": 3700
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 24.36,
|
| 243 |
+
"learning_rate": 6.0750735294117646e-05,
|
| 244 |
+
"loss": 0.9495,
|
| 245 |
+
"step": 3800
|
| 246 |
+
},
|
| 247 |
+
{
|
| 248 |
+
"epoch": 25.0,
|
| 249 |
+
"learning_rate": 6.024117647058823e-05,
|
| 250 |
+
"loss": 0.9667,
|
| 251 |
+
"step": 3900
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 25.64,
|
| 255 |
+
"learning_rate": 5.9726470588235294e-05,
|
| 256 |
+
"loss": 0.9383,
|
| 257 |
+
"step": 4000
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 25.64,
|
| 261 |
+
"eval_loss": 0.223493292927742,
|
| 262 |
+
"eval_runtime": 99.8226,
|
| 263 |
+
"eval_samples_per_second": 20.877,
|
| 264 |
+
"eval_steps_per_second": 20.877,
|
| 265 |
+
"eval_wer": 0.23281978512061627,
|
| 266 |
+
"step": 4000
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 26.28,
|
| 270 |
+
"learning_rate": 5.921176470588234e-05,
|
| 271 |
+
"loss": 0.9534,
|
| 272 |
+
"step": 4100
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 26.92,
|
| 276 |
+
"learning_rate": 5.8697058823529405e-05,
|
| 277 |
+
"loss": 0.9294,
|
| 278 |
+
"step": 4200
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 27.56,
|
| 282 |
+
"learning_rate": 5.818235294117647e-05,
|
| 283 |
+
"loss": 0.9348,
|
| 284 |
+
"step": 4300
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 28.21,
|
| 288 |
+
"learning_rate": 5.766764705882352e-05,
|
| 289 |
+
"loss": 0.9209,
|
| 290 |
+
"step": 4400
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 28.85,
|
| 294 |
+
"learning_rate": 5.7152941176470585e-05,
|
| 295 |
+
"loss": 0.9104,
|
| 296 |
+
"step": 4500
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 29.49,
|
| 300 |
+
"learning_rate": 5.663823529411764e-05,
|
| 301 |
+
"loss": 0.9009,
|
| 302 |
+
"step": 4600
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 30.13,
|
| 306 |
+
"learning_rate": 5.61235294117647e-05,
|
| 307 |
+
"loss": 0.9109,
|
| 308 |
+
"step": 4700
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 30.77,
|
| 312 |
+
"learning_rate": 5.560882352941176e-05,
|
| 313 |
+
"loss": 0.8973,
|
| 314 |
+
"step": 4800
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 31.41,
|
| 318 |
+
"learning_rate": 5.509411764705882e-05,
|
| 319 |
+
"loss": 0.864,
|
| 320 |
+
"step": 4900
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 32.05,
|
| 324 |
+
"learning_rate": 5.4579411764705876e-05,
|
| 325 |
+
"loss": 0.8654,
|
| 326 |
+
"step": 5000
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 32.69,
|
| 330 |
+
"learning_rate": 5.406470588235294e-05,
|
| 331 |
+
"loss": 0.8523,
|
| 332 |
+
"step": 5100
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 33.33,
|
| 336 |
+
"learning_rate": 5.3549999999999994e-05,
|
| 337 |
+
"loss": 0.8794,
|
| 338 |
+
"step": 5200
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 33.97,
|
| 342 |
+
"learning_rate": 5.3035294117647056e-05,
|
| 343 |
+
"loss": 0.8577,
|
| 344 |
+
"step": 5300
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 34.62,
|
| 348 |
+
"learning_rate": 5.252058823529411e-05,
|
| 349 |
+
"loss": 0.8585,
|
| 350 |
+
"step": 5400
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 35.26,
|
| 354 |
+
"learning_rate": 5.2005882352941174e-05,
|
| 355 |
+
"loss": 0.8473,
|
| 356 |
+
"step": 5500
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 35.9,
|
| 360 |
+
"learning_rate": 5.149117647058823e-05,
|
| 361 |
+
"loss": 0.8465,
|
| 362 |
+
"step": 5600
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 36.54,
|
| 366 |
+
"learning_rate": 5.097647058823529e-05,
|
| 367 |
+
"loss": 0.838,
|
| 368 |
+
"step": 5700
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 37.18,
|
| 372 |
+
"learning_rate": 5.046176470588235e-05,
|
| 373 |
+
"loss": 0.8402,
|
| 374 |
+
"step": 5800
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 37.82,
|
| 378 |
+
"learning_rate": 4.99470588235294e-05,
|
| 379 |
+
"loss": 0.8432,
|
| 380 |
+
"step": 5900
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 38.46,
|
| 384 |
+
"learning_rate": 4.9432352941176465e-05,
|
| 385 |
+
"loss": 0.8359,
|
| 386 |
+
"step": 6000
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 38.46,
|
| 390 |
+
"eval_loss": 0.20040307939052582,
|
| 391 |
+
"eval_runtime": 101.1606,
|
| 392 |
+
"eval_samples_per_second": 20.601,
|
| 393 |
+
"eval_steps_per_second": 20.601,
|
| 394 |
+
"eval_wer": 0.20981147374822623,
|
| 395 |
+
"step": 6000
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 39.1,
|
| 399 |
+
"learning_rate": 4.891764705882353e-05,
|
| 400 |
+
"loss": 0.8456,
|
| 401 |
+
"step": 6100
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 39.74,
|
| 405 |
+
"learning_rate": 4.840294117647059e-05,
|
| 406 |
+
"loss": 0.8285,
|
| 407 |
+
"step": 6200
|
| 408 |
+
},
|
| 409 |
+
{
|
| 410 |
+
"epoch": 40.38,
|
| 411 |
+
"learning_rate": 4.788823529411764e-05,
|
| 412 |
+
"loss": 0.8054,
|
| 413 |
+
"step": 6300
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 41.03,
|
| 417 |
+
"learning_rate": 4.73735294117647e-05,
|
| 418 |
+
"loss": 0.82,
|
| 419 |
+
"step": 6400
|
| 420 |
+
},
|
| 421 |
+
{
|
| 422 |
+
"epoch": 41.67,
|
| 423 |
+
"learning_rate": 4.685882352941176e-05,
|
| 424 |
+
"loss": 0.8082,
|
| 425 |
+
"step": 6500
|
| 426 |
+
},
|
| 427 |
+
{
|
| 428 |
+
"epoch": 42.31,
|
| 429 |
+
"learning_rate": 4.6344117647058825e-05,
|
| 430 |
+
"loss": 0.8049,
|
| 431 |
+
"step": 6600
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 42.95,
|
| 435 |
+
"learning_rate": 4.5829411764705873e-05,
|
| 436 |
+
"loss": 0.8088,
|
| 437 |
+
"step": 6700
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 43.59,
|
| 441 |
+
"learning_rate": 4.5314705882352936e-05,
|
| 442 |
+
"loss": 0.8092,
|
| 443 |
+
"step": 6800
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 44.23,
|
| 447 |
+
"learning_rate": 4.48e-05,
|
| 448 |
+
"loss": 0.8212,
|
| 449 |
+
"step": 6900
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 44.87,
|
| 453 |
+
"learning_rate": 4.428529411764706e-05,
|
| 454 |
+
"loss": 0.7847,
|
| 455 |
+
"step": 7000
|
| 456 |
+
},
|
| 457 |
+
{
|
| 458 |
+
"epoch": 45.51,
|
| 459 |
+
"learning_rate": 4.3775735294117646e-05,
|
| 460 |
+
"loss": 0.7846,
|
| 461 |
+
"step": 7100
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 46.15,
|
| 465 |
+
"learning_rate": 4.32610294117647e-05,
|
| 466 |
+
"loss": 0.7966,
|
| 467 |
+
"step": 7200
|
| 468 |
+
},
|
| 469 |
+
{
|
| 470 |
+
"epoch": 46.79,
|
| 471 |
+
"learning_rate": 4.2746323529411764e-05,
|
| 472 |
+
"loss": 0.7811,
|
| 473 |
+
"step": 7300
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 47.44,
|
| 477 |
+
"learning_rate": 4.223161764705882e-05,
|
| 478 |
+
"loss": 0.7706,
|
| 479 |
+
"step": 7400
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 48.08,
|
| 483 |
+
"learning_rate": 4.1716911764705875e-05,
|
| 484 |
+
"loss": 0.7958,
|
| 485 |
+
"step": 7500
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 48.72,
|
| 489 |
+
"learning_rate": 4.120220588235294e-05,
|
| 490 |
+
"loss": 0.7617,
|
| 491 |
+
"step": 7600
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 49.36,
|
| 495 |
+
"learning_rate": 4.06875e-05,
|
| 496 |
+
"loss": 0.7662,
|
| 497 |
+
"step": 7700
|
| 498 |
+
},
|
| 499 |
+
{
|
| 500 |
+
"epoch": 50.0,
|
| 501 |
+
"learning_rate": 4.0172794117647055e-05,
|
| 502 |
+
"loss": 0.7545,
|
| 503 |
+
"step": 7800
|
| 504 |
+
},
|
| 505 |
+
{
|
| 506 |
+
"epoch": 50.64,
|
| 507 |
+
"learning_rate": 3.965808823529411e-05,
|
| 508 |
+
"loss": 0.7482,
|
| 509 |
+
"step": 7900
|
| 510 |
+
},
|
| 511 |
+
{
|
| 512 |
+
"epoch": 51.28,
|
| 513 |
+
"learning_rate": 3.914338235294117e-05,
|
| 514 |
+
"loss": 0.7633,
|
| 515 |
+
"step": 8000
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 51.28,
|
| 519 |
+
"eval_loss": 0.19603605568408966,
|
| 520 |
+
"eval_runtime": 98.7836,
|
| 521 |
+
"eval_samples_per_second": 21.097,
|
| 522 |
+
"eval_steps_per_second": 21.097,
|
| 523 |
+
"eval_wer": 0.1882221771741334,
|
| 524 |
+
"step": 8000
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 51.92,
|
| 528 |
+
"learning_rate": 3.8628676470588235e-05,
|
| 529 |
+
"loss": 0.7522,
|
| 530 |
+
"step": 8100
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 52.56,
|
| 534 |
+
"learning_rate": 3.811397058823529e-05,
|
| 535 |
+
"loss": 0.7411,
|
| 536 |
+
"step": 8200
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 53.21,
|
| 540 |
+
"learning_rate": 3.7599264705882346e-05,
|
| 541 |
+
"loss": 0.7443,
|
| 542 |
+
"step": 8300
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 53.85,
|
| 546 |
+
"learning_rate": 3.708455882352941e-05,
|
| 547 |
+
"loss": 0.7583,
|
| 548 |
+
"step": 8400
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 54.49,
|
| 552 |
+
"learning_rate": 3.656985294117647e-05,
|
| 553 |
+
"loss": 0.7254,
|
| 554 |
+
"step": 8500
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 55.13,
|
| 558 |
+
"learning_rate": 3.605514705882352e-05,
|
| 559 |
+
"loss": 0.7329,
|
| 560 |
+
"step": 8600
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 55.77,
|
| 564 |
+
"learning_rate": 3.554044117647058e-05,
|
| 565 |
+
"loss": 0.7216,
|
| 566 |
+
"step": 8700
|
| 567 |
+
},
|
| 568 |
+
{
|
| 569 |
+
"epoch": 56.41,
|
| 570 |
+
"learning_rate": 3.5025735294117644e-05,
|
| 571 |
+
"loss": 0.7322,
|
| 572 |
+
"step": 8800
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 57.05,
|
| 576 |
+
"learning_rate": 3.4511029411764706e-05,
|
| 577 |
+
"loss": 0.7405,
|
| 578 |
+
"step": 8900
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 57.69,
|
| 582 |
+
"learning_rate": 3.399632352941176e-05,
|
| 583 |
+
"loss": 0.7194,
|
| 584 |
+
"step": 9000
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 58.33,
|
| 588 |
+
"learning_rate": 3.3481617647058824e-05,
|
| 589 |
+
"loss": 0.7146,
|
| 590 |
+
"step": 9100
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 58.97,
|
| 594 |
+
"learning_rate": 3.297205882352941e-05,
|
| 595 |
+
"loss": 0.7128,
|
| 596 |
+
"step": 9200
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"epoch": 59.62,
|
| 600 |
+
"learning_rate": 3.2457352941176466e-05,
|
| 601 |
+
"loss": 0.7186,
|
| 602 |
+
"step": 9300
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 60.26,
|
| 606 |
+
"learning_rate": 3.194264705882353e-05,
|
| 607 |
+
"loss": 0.7038,
|
| 608 |
+
"step": 9400
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 60.9,
|
| 612 |
+
"learning_rate": 3.142794117647058e-05,
|
| 613 |
+
"loss": 0.696,
|
| 614 |
+
"step": 9500
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 61.54,
|
| 618 |
+
"learning_rate": 3.0913235294117646e-05,
|
| 619 |
+
"loss": 0.707,
|
| 620 |
+
"step": 9600
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 62.18,
|
| 624 |
+
"learning_rate": 3.0398529411764704e-05,
|
| 625 |
+
"loss": 0.7113,
|
| 626 |
+
"step": 9700
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 62.82,
|
| 630 |
+
"learning_rate": 2.9883823529411763e-05,
|
| 631 |
+
"loss": 0.6997,
|
| 632 |
+
"step": 9800
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 63.46,
|
| 636 |
+
"learning_rate": 2.9369117647058822e-05,
|
| 637 |
+
"loss": 0.6815,
|
| 638 |
+
"step": 9900
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 64.1,
|
| 642 |
+
"learning_rate": 2.885441176470588e-05,
|
| 643 |
+
"loss": 0.7001,
|
| 644 |
+
"step": 10000
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 64.1,
|
| 648 |
+
"eval_loss": 0.19019167125225067,
|
| 649 |
+
"eval_runtime": 99.1402,
|
| 650 |
+
"eval_samples_per_second": 21.021,
|
| 651 |
+
"eval_steps_per_second": 21.021,
|
| 652 |
+
"eval_wer": 0.18092438678289074,
|
| 653 |
+
"step": 10000
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 64.74,
|
| 657 |
+
"learning_rate": 2.833970588235294e-05,
|
| 658 |
+
"loss": 0.7032,
|
| 659 |
+
"step": 10100
|
| 660 |
+
},
|
| 661 |
+
{
|
| 662 |
+
"epoch": 65.38,
|
| 663 |
+
"learning_rate": 2.7825e-05,
|
| 664 |
+
"loss": 0.7102,
|
| 665 |
+
"step": 10200
|
| 666 |
+
},
|
| 667 |
+
{
|
| 668 |
+
"epoch": 66.03,
|
| 669 |
+
"learning_rate": 2.7310294117647054e-05,
|
| 670 |
+
"loss": 0.6916,
|
| 671 |
+
"step": 10300
|
| 672 |
+
},
|
| 673 |
+
{
|
| 674 |
+
"epoch": 66.67,
|
| 675 |
+
"learning_rate": 2.6795588235294116e-05,
|
| 676 |
+
"loss": 0.6907,
|
| 677 |
+
"step": 10400
|
| 678 |
+
},
|
| 679 |
+
{
|
| 680 |
+
"epoch": 67.31,
|
| 681 |
+
"learning_rate": 2.6280882352941172e-05,
|
| 682 |
+
"loss": 0.6711,
|
| 683 |
+
"step": 10500
|
| 684 |
+
},
|
| 685 |
+
{
|
| 686 |
+
"epoch": 67.95,
|
| 687 |
+
"learning_rate": 2.5766176470588234e-05,
|
| 688 |
+
"loss": 0.6843,
|
| 689 |
+
"step": 10600
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 68.59,
|
| 693 |
+
"learning_rate": 2.525147058823529e-05,
|
| 694 |
+
"loss": 0.6847,
|
| 695 |
+
"step": 10700
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"epoch": 69.23,
|
| 699 |
+
"learning_rate": 2.4736764705882352e-05,
|
| 700 |
+
"loss": 0.6533,
|
| 701 |
+
"step": 10800
|
| 702 |
+
},
|
| 703 |
+
{
|
| 704 |
+
"epoch": 69.87,
|
| 705 |
+
"learning_rate": 2.4222058823529407e-05,
|
| 706 |
+
"loss": 0.6883,
|
| 707 |
+
"step": 10900
|
| 708 |
+
},
|
| 709 |
+
{
|
| 710 |
+
"epoch": 70.51,
|
| 711 |
+
"learning_rate": 2.3712499999999997e-05,
|
| 712 |
+
"loss": 0.6779,
|
| 713 |
+
"step": 11000
|
| 714 |
+
},
|
| 715 |
+
{
|
| 716 |
+
"epoch": 71.15,
|
| 717 |
+
"learning_rate": 2.319779411764706e-05,
|
| 718 |
+
"loss": 0.6739,
|
| 719 |
+
"step": 11100
|
| 720 |
+
},
|
| 721 |
+
{
|
| 722 |
+
"epoch": 71.79,
|
| 723 |
+
"learning_rate": 2.2683088235294115e-05,
|
| 724 |
+
"loss": 0.6662,
|
| 725 |
+
"step": 11200
|
| 726 |
+
},
|
| 727 |
+
{
|
| 728 |
+
"epoch": 72.44,
|
| 729 |
+
"learning_rate": 2.2168382352941177e-05,
|
| 730 |
+
"loss": 0.6539,
|
| 731 |
+
"step": 11300
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 73.08,
|
| 735 |
+
"learning_rate": 2.1653676470588232e-05,
|
| 736 |
+
"loss": 0.6589,
|
| 737 |
+
"step": 11400
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 73.72,
|
| 741 |
+
"learning_rate": 2.1138970588235295e-05,
|
| 742 |
+
"loss": 0.6545,
|
| 743 |
+
"step": 11500
|
| 744 |
+
},
|
| 745 |
+
{
|
| 746 |
+
"epoch": 74.36,
|
| 747 |
+
"learning_rate": 2.062426470588235e-05,
|
| 748 |
+
"loss": 0.6585,
|
| 749 |
+
"step": 11600
|
| 750 |
+
},
|
| 751 |
+
{
|
| 752 |
+
"epoch": 75.0,
|
| 753 |
+
"learning_rate": 2.0109558823529412e-05,
|
| 754 |
+
"loss": 0.6472,
|
| 755 |
+
"step": 11700
|
| 756 |
+
},
|
| 757 |
+
{
|
| 758 |
+
"epoch": 75.64,
|
| 759 |
+
"learning_rate": 1.9594852941176468e-05,
|
| 760 |
+
"loss": 0.6657,
|
| 761 |
+
"step": 11800
|
| 762 |
+
},
|
| 763 |
+
{
|
| 764 |
+
"epoch": 76.28,
|
| 765 |
+
"learning_rate": 1.9080147058823527e-05,
|
| 766 |
+
"loss": 0.65,
|
| 767 |
+
"step": 11900
|
| 768 |
+
},
|
| 769 |
+
{
|
| 770 |
+
"epoch": 76.92,
|
| 771 |
+
"learning_rate": 1.8565441176470586e-05,
|
| 772 |
+
"loss": 0.652,
|
| 773 |
+
"step": 12000
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 76.92,
|
| 777 |
+
"eval_loss": 0.197893887758255,
|
| 778 |
+
"eval_runtime": 98.6246,
|
| 779 |
+
"eval_samples_per_second": 21.131,
|
| 780 |
+
"eval_steps_per_second": 21.131,
|
| 781 |
+
"eval_wer": 0.17747820798702615,
|
| 782 |
+
"step": 12000
|
| 783 |
+
},
|
| 784 |
+
{
|
| 785 |
+
"epoch": 77.56,
|
| 786 |
+
"learning_rate": 1.8050735294117645e-05,
|
| 787 |
+
"loss": 0.6559,
|
| 788 |
+
"step": 12100
|
| 789 |
+
},
|
| 790 |
+
{
|
| 791 |
+
"epoch": 78.21,
|
| 792 |
+
"learning_rate": 1.7536029411764703e-05,
|
| 793 |
+
"loss": 0.6282,
|
| 794 |
+
"step": 12200
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 78.85,
|
| 798 |
+
"learning_rate": 1.7021323529411762e-05,
|
| 799 |
+
"loss": 0.6251,
|
| 800 |
+
"step": 12300
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 79.49,
|
| 804 |
+
"learning_rate": 1.650661764705882e-05,
|
| 805 |
+
"loss": 0.6147,
|
| 806 |
+
"step": 12400
|
| 807 |
+
},
|
| 808 |
+
{
|
| 809 |
+
"epoch": 80.13,
|
| 810 |
+
"learning_rate": 1.599191176470588e-05,
|
| 811 |
+
"loss": 0.6412,
|
| 812 |
+
"step": 12500
|
| 813 |
+
},
|
| 814 |
+
{
|
| 815 |
+
"epoch": 80.77,
|
| 816 |
+
"learning_rate": 1.547720588235294e-05,
|
| 817 |
+
"loss": 0.6309,
|
| 818 |
+
"step": 12600
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 81.41,
|
| 822 |
+
"learning_rate": 1.4962499999999998e-05,
|
| 823 |
+
"loss": 0.6197,
|
| 824 |
+
"step": 12700
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 82.05,
|
| 828 |
+
"learning_rate": 1.4447794117647057e-05,
|
| 829 |
+
"loss": 0.6238,
|
| 830 |
+
"step": 12800
|
| 831 |
+
},
|
| 832 |
+
{
|
| 833 |
+
"epoch": 82.69,
|
| 834 |
+
"learning_rate": 1.3933088235294116e-05,
|
| 835 |
+
"loss": 0.635,
|
| 836 |
+
"step": 12900
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 83.33,
|
| 840 |
+
"learning_rate": 1.3418382352941174e-05,
|
| 841 |
+
"loss": 0.6294,
|
| 842 |
+
"step": 13000
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 83.97,
|
| 846 |
+
"learning_rate": 1.2903676470588233e-05,
|
| 847 |
+
"loss": 0.6171,
|
| 848 |
+
"step": 13100
|
| 849 |
+
},
|
| 850 |
+
{
|
| 851 |
+
"epoch": 84.62,
|
| 852 |
+
"learning_rate": 1.2388970588235292e-05,
|
| 853 |
+
"loss": 0.6255,
|
| 854 |
+
"step": 13200
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 85.26,
|
| 858 |
+
"learning_rate": 1.1874264705882353e-05,
|
| 859 |
+
"loss": 0.6238,
|
| 860 |
+
"step": 13300
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 85.9,
|
| 864 |
+
"learning_rate": 1.1359558823529412e-05,
|
| 865 |
+
"loss": 0.5923,
|
| 866 |
+
"step": 13400
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 86.54,
|
| 870 |
+
"learning_rate": 1.084485294117647e-05,
|
| 871 |
+
"loss": 0.6066,
|
| 872 |
+
"step": 13500
|
| 873 |
+
},
|
| 874 |
+
{
|
| 875 |
+
"epoch": 87.18,
|
| 876 |
+
"learning_rate": 1.033014705882353e-05,
|
| 877 |
+
"loss": 0.6143,
|
| 878 |
+
"step": 13600
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 87.82,
|
| 882 |
+
"learning_rate": 9.815441176470588e-06,
|
| 883 |
+
"loss": 0.6146,
|
| 884 |
+
"step": 13700
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 88.46,
|
| 888 |
+
"learning_rate": 9.300735294117647e-06,
|
| 889 |
+
"loss": 0.6102,
|
| 890 |
+
"step": 13800
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 89.1,
|
| 894 |
+
"learning_rate": 8.786029411764706e-06,
|
| 895 |
+
"loss": 0.6112,
|
| 896 |
+
"step": 13900
|
| 897 |
+
},
|
| 898 |
+
{
|
| 899 |
+
"epoch": 89.74,
|
| 900 |
+
"learning_rate": 8.271323529411765e-06,
|
| 901 |
+
"loss": 0.6025,
|
| 902 |
+
"step": 14000
|
| 903 |
+
},
|
| 904 |
+
{
|
| 905 |
+
"epoch": 89.74,
|
| 906 |
+
"eval_loss": 0.18655425310134888,
|
| 907 |
+
"eval_runtime": 99.4533,
|
| 908 |
+
"eval_samples_per_second": 20.955,
|
| 909 |
+
"eval_steps_per_second": 20.955,
|
| 910 |
+
"eval_wer": 0.16957226839651326,
|
| 911 |
+
"step": 14000
|
| 912 |
+
},
|
| 913 |
+
{
|
| 914 |
+
"epoch": 90.38,
|
| 915 |
+
"learning_rate": 7.756617647058824e-06,
|
| 916 |
+
"loss": 0.6147,
|
| 917 |
+
"step": 14100
|
| 918 |
+
},
|
| 919 |
+
{
|
| 920 |
+
"epoch": 91.03,
|
| 921 |
+
"learning_rate": 7.241911764705882e-06,
|
| 922 |
+
"loss": 0.5954,
|
| 923 |
+
"step": 14200
|
| 924 |
+
},
|
| 925 |
+
{
|
| 926 |
+
"epoch": 91.67,
|
| 927 |
+
"learning_rate": 6.7272058823529405e-06,
|
| 928 |
+
"loss": 0.5998,
|
| 929 |
+
"step": 14300
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"epoch": 92.31,
|
| 933 |
+
"learning_rate": 6.212499999999999e-06,
|
| 934 |
+
"loss": 0.6078,
|
| 935 |
+
"step": 14400
|
| 936 |
+
},
|
| 937 |
+
{
|
| 938 |
+
"epoch": 92.95,
|
| 939 |
+
"learning_rate": 5.697794117647058e-06,
|
| 940 |
+
"loss": 0.6068,
|
| 941 |
+
"step": 14500
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 93.59,
|
| 945 |
+
"learning_rate": 5.183088235294118e-06,
|
| 946 |
+
"loss": 0.6015,
|
| 947 |
+
"step": 14600
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 94.23,
|
| 951 |
+
"learning_rate": 4.668382352941176e-06,
|
| 952 |
+
"loss": 0.6013,
|
| 953 |
+
"step": 14700
|
| 954 |
+
},
|
| 955 |
+
{
|
| 956 |
+
"epoch": 94.87,
|
| 957 |
+
"learning_rate": 4.153676470588235e-06,
|
| 958 |
+
"loss": 0.5948,
|
| 959 |
+
"step": 14800
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 95.51,
|
| 963 |
+
"learning_rate": 3.6389705882352937e-06,
|
| 964 |
+
"loss": 0.5909,
|
| 965 |
+
"step": 14900
|
| 966 |
+
},
|
| 967 |
+
{
|
| 968 |
+
"epoch": 96.15,
|
| 969 |
+
"learning_rate": 3.1242647058823526e-06,
|
| 970 |
+
"loss": 0.6129,
|
| 971 |
+
"step": 15000
|
| 972 |
+
},
|
| 973 |
+
{
|
| 974 |
+
"epoch": 96.79,
|
| 975 |
+
"learning_rate": 2.614705882352941e-06,
|
| 976 |
+
"loss": 0.5795,
|
| 977 |
+
"step": 15100
|
| 978 |
+
},
|
| 979 |
+
{
|
| 980 |
+
"epoch": 97.44,
|
| 981 |
+
"learning_rate": 2.1e-06,
|
| 982 |
+
"loss": 0.5967,
|
| 983 |
+
"step": 15200
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 98.08,
|
| 987 |
+
"learning_rate": 1.5852941176470587e-06,
|
| 988 |
+
"loss": 0.6013,
|
| 989 |
+
"step": 15300
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 98.72,
|
| 993 |
+
"learning_rate": 1.0705882352941175e-06,
|
| 994 |
+
"loss": 0.5981,
|
| 995 |
+
"step": 15400
|
| 996 |
+
},
|
| 997 |
+
{
|
| 998 |
+
"epoch": 99.36,
|
| 999 |
+
"learning_rate": 5.558823529411764e-07,
|
| 1000 |
+
"loss": 0.6133,
|
| 1001 |
+
"step": 15500
|
| 1002 |
+
},
|
| 1003 |
+
{
|
| 1004 |
+
"epoch": 100.0,
|
| 1005 |
+
"learning_rate": 4.117647058823529e-08,
|
| 1006 |
+
"loss": 0.5932,
|
| 1007 |
+
"step": 15600
|
| 1008 |
+
},
|
| 1009 |
+
{
|
| 1010 |
+
"epoch": 100.0,
|
| 1011 |
+
"step": 15600,
|
| 1012 |
+
"total_flos": 5.043365508466233e+19,
|
| 1013 |
+
"train_loss": 1.1756210368718856,
|
| 1014 |
+
"train_runtime": 18507.9964,
|
| 1015 |
+
"train_samples_per_second": 26.815,
|
| 1016 |
+
"train_steps_per_second": 0.843
|
| 1017 |
+
}
|
| 1018 |
+
],
|
| 1019 |
+
"max_steps": 15600,
|
| 1020 |
+
"num_train_epochs": 100,
|
| 1021 |
+
"total_flos": 5.043365508466233e+19,
|
| 1022 |
+
"trial_name": null,
|
| 1023 |
+
"trial_params": null
|
| 1024 |
+
}
|