Upload folder using huggingface_hub
Browse files- model.safetensors +1 -1
- optimizer.pt +1 -1
- rng_state.pth +1 -1
- scaler.pt +1 -1
- scheduler.pt +1 -1
- trainer_state.json +1799 -8
- training_args.bin +1 -1
model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 378311620
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:90dec33ca25376d7fab8f5635e6bc543ed75144bb3ab10bcb665ac467671fcac
|
| 3 |
size 378311620
|
optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4771843
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:da58438b9f9a5189e0457a9df10c78a966d56673489be0f2fb4f6ff067b812d0
|
| 3 |
size 4771843
|
rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14308
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3777018ee5c1ad8c7475ebd82d76f78b442de355ff486eb51464ae8490de4e40
|
| 3 |
size 14308
|
scaler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 988
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:25a65a95fc061242cb0ca7b260e068dca633d7bc3d64ccc03eab3e798e52854b
|
| 3 |
size 988
|
scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1064
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cb692e54745ea187a8c3a1d90a6ad4458fb449ad0862cc16a6e69e0e4cdc10a8
|
| 3 |
size 1064
|
trainer_state.json
CHANGED
|
@@ -1,10 +1,10 @@
|
|
| 1 |
{
|
| 2 |
-
"best_global_step":
|
| 3 |
-
"best_metric": 0.
|
| 4 |
-
"best_model_checkpoint": "./wav2vec2_accent_classification_exp2/checkpoint-
|
| 5 |
-
"epoch":
|
| 6 |
"eval_steps": 500,
|
| 7 |
-
"global_step":
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
@@ -1799,12 +1799,1803 @@
|
|
| 1799 |
"eval_samples_per_second": 2.822,
|
| 1800 |
"eval_steps_per_second": 2.822,
|
| 1801 |
"step": 5049
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1802 |
}
|
| 1803 |
],
|
| 1804 |
"logging_steps": 20,
|
| 1805 |
-
"max_steps":
|
| 1806 |
"num_input_tokens_seen": 0,
|
| 1807 |
-
"num_train_epochs":
|
| 1808 |
"save_steps": 500,
|
| 1809 |
"stateful_callbacks": {
|
| 1810 |
"TrainerControl": {
|
|
@@ -1818,7 +3609,7 @@
|
|
| 1818 |
"attributes": {}
|
| 1819 |
}
|
| 1820 |
},
|
| 1821 |
-
"total_flos":
|
| 1822 |
"train_batch_size": 1,
|
| 1823 |
"trial_name": null,
|
| 1824 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_global_step": 6732,
|
| 3 |
+
"best_metric": 0.4074074074074074,
|
| 4 |
+
"best_model_checkpoint": "./wav2vec2_accent_classification_exp2/checkpoint-6732",
|
| 5 |
+
"epoch": 6.0,
|
| 6 |
"eval_steps": 500,
|
| 7 |
+
"global_step": 10098,
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
|
|
| 1799 |
"eval_samples_per_second": 2.822,
|
| 1800 |
"eval_steps_per_second": 2.822,
|
| 1801 |
"step": 5049
|
| 1802 |
+
},
|
| 1803 |
+
{
|
| 1804 |
+
"epoch": 3.0065359477124183,
|
| 1805 |
+
"grad_norm": 5.259908676147461,
|
| 1806 |
+
"learning_rate": 5.549075704225353e-05,
|
| 1807 |
+
"loss": 1.3437,
|
| 1808 |
+
"step": 5060
|
| 1809 |
+
},
|
| 1810 |
+
{
|
| 1811 |
+
"epoch": 3.0184194890077243,
|
| 1812 |
+
"grad_norm": 14.064613342285156,
|
| 1813 |
+
"learning_rate": 5.5270686619718314e-05,
|
| 1814 |
+
"loss": 1.43,
|
| 1815 |
+
"step": 5080
|
| 1816 |
+
},
|
| 1817 |
+
{
|
| 1818 |
+
"epoch": 3.0303030303030303,
|
| 1819 |
+
"grad_norm": 14.920954704284668,
|
| 1820 |
+
"learning_rate": 5.5050616197183105e-05,
|
| 1821 |
+
"loss": 1.8756,
|
| 1822 |
+
"step": 5100
|
| 1823 |
+
},
|
| 1824 |
+
{
|
| 1825 |
+
"epoch": 3.0421865715983363,
|
| 1826 |
+
"grad_norm": 17.76242446899414,
|
| 1827 |
+
"learning_rate": 5.483054577464789e-05,
|
| 1828 |
+
"loss": 1.3734,
|
| 1829 |
+
"step": 5120
|
| 1830 |
+
},
|
| 1831 |
+
{
|
| 1832 |
+
"epoch": 3.0540701128936423,
|
| 1833 |
+
"grad_norm": 9.35627269744873,
|
| 1834 |
+
"learning_rate": 5.461047535211268e-05,
|
| 1835 |
+
"loss": 1.4487,
|
| 1836 |
+
"step": 5140
|
| 1837 |
+
},
|
| 1838 |
+
{
|
| 1839 |
+
"epoch": 3.0659536541889483,
|
| 1840 |
+
"grad_norm": 5.479264259338379,
|
| 1841 |
+
"learning_rate": 5.4390404929577464e-05,
|
| 1842 |
+
"loss": 1.439,
|
| 1843 |
+
"step": 5160
|
| 1844 |
+
},
|
| 1845 |
+
{
|
| 1846 |
+
"epoch": 3.0778371954842543,
|
| 1847 |
+
"grad_norm": 5.219712257385254,
|
| 1848 |
+
"learning_rate": 5.4170334507042255e-05,
|
| 1849 |
+
"loss": 1.5694,
|
| 1850 |
+
"step": 5180
|
| 1851 |
+
},
|
| 1852 |
+
{
|
| 1853 |
+
"epoch": 3.0897207367795603,
|
| 1854 |
+
"grad_norm": 9.071990966796875,
|
| 1855 |
+
"learning_rate": 5.395026408450704e-05,
|
| 1856 |
+
"loss": 1.6675,
|
| 1857 |
+
"step": 5200
|
| 1858 |
+
},
|
| 1859 |
+
{
|
| 1860 |
+
"epoch": 3.1016042780748663,
|
| 1861 |
+
"grad_norm": 10.13392448425293,
|
| 1862 |
+
"learning_rate": 5.373019366197183e-05,
|
| 1863 |
+
"loss": 1.3387,
|
| 1864 |
+
"step": 5220
|
| 1865 |
+
},
|
| 1866 |
+
{
|
| 1867 |
+
"epoch": 3.1134878193701723,
|
| 1868 |
+
"grad_norm": 11.0034818649292,
|
| 1869 |
+
"learning_rate": 5.351012323943663e-05,
|
| 1870 |
+
"loss": 1.5826,
|
| 1871 |
+
"step": 5240
|
| 1872 |
+
},
|
| 1873 |
+
{
|
| 1874 |
+
"epoch": 3.1253713606654783,
|
| 1875 |
+
"grad_norm": 13.366768836975098,
|
| 1876 |
+
"learning_rate": 5.329005281690141e-05,
|
| 1877 |
+
"loss": 1.4869,
|
| 1878 |
+
"step": 5260
|
| 1879 |
+
},
|
| 1880 |
+
{
|
| 1881 |
+
"epoch": 3.1372549019607843,
|
| 1882 |
+
"grad_norm": 3.3045759201049805,
|
| 1883 |
+
"learning_rate": 5.30699823943662e-05,
|
| 1884 |
+
"loss": 1.0267,
|
| 1885 |
+
"step": 5280
|
| 1886 |
+
},
|
| 1887 |
+
{
|
| 1888 |
+
"epoch": 3.1491384432560903,
|
| 1889 |
+
"grad_norm": 7.309617042541504,
|
| 1890 |
+
"learning_rate": 5.284991197183099e-05,
|
| 1891 |
+
"loss": 1.408,
|
| 1892 |
+
"step": 5300
|
| 1893 |
+
},
|
| 1894 |
+
{
|
| 1895 |
+
"epoch": 3.1610219845513963,
|
| 1896 |
+
"grad_norm": 9.015344619750977,
|
| 1897 |
+
"learning_rate": 5.262984154929578e-05,
|
| 1898 |
+
"loss": 1.6,
|
| 1899 |
+
"step": 5320
|
| 1900 |
+
},
|
| 1901 |
+
{
|
| 1902 |
+
"epoch": 3.1729055258467023,
|
| 1903 |
+
"grad_norm": 0.17147333920001984,
|
| 1904 |
+
"learning_rate": 5.240977112676056e-05,
|
| 1905 |
+
"loss": 1.5795,
|
| 1906 |
+
"step": 5340
|
| 1907 |
+
},
|
| 1908 |
+
{
|
| 1909 |
+
"epoch": 3.1847890671420083,
|
| 1910 |
+
"grad_norm": 11.151992797851562,
|
| 1911 |
+
"learning_rate": 5.218970070422535e-05,
|
| 1912 |
+
"loss": 1.2323,
|
| 1913 |
+
"step": 5360
|
| 1914 |
+
},
|
| 1915 |
+
{
|
| 1916 |
+
"epoch": 3.1966726084373143,
|
| 1917 |
+
"grad_norm": 2.905815839767456,
|
| 1918 |
+
"learning_rate": 5.196963028169014e-05,
|
| 1919 |
+
"loss": 1.3181,
|
| 1920 |
+
"step": 5380
|
| 1921 |
+
},
|
| 1922 |
+
{
|
| 1923 |
+
"epoch": 3.2085561497326203,
|
| 1924 |
+
"grad_norm": 7.193173885345459,
|
| 1925 |
+
"learning_rate": 5.174955985915493e-05,
|
| 1926 |
+
"loss": 1.5094,
|
| 1927 |
+
"step": 5400
|
| 1928 |
+
},
|
| 1929 |
+
{
|
| 1930 |
+
"epoch": 3.2204396910279263,
|
| 1931 |
+
"grad_norm": 8.543306350708008,
|
| 1932 |
+
"learning_rate": 5.152948943661971e-05,
|
| 1933 |
+
"loss": 1.4647,
|
| 1934 |
+
"step": 5420
|
| 1935 |
+
},
|
| 1936 |
+
{
|
| 1937 |
+
"epoch": 3.2323232323232323,
|
| 1938 |
+
"grad_norm": 13.816205978393555,
|
| 1939 |
+
"learning_rate": 5.130941901408452e-05,
|
| 1940 |
+
"loss": 1.6882,
|
| 1941 |
+
"step": 5440
|
| 1942 |
+
},
|
| 1943 |
+
{
|
| 1944 |
+
"epoch": 3.2442067736185383,
|
| 1945 |
+
"grad_norm": 4.082400798797607,
|
| 1946 |
+
"learning_rate": 5.10893485915493e-05,
|
| 1947 |
+
"loss": 1.6905,
|
| 1948 |
+
"step": 5460
|
| 1949 |
+
},
|
| 1950 |
+
{
|
| 1951 |
+
"epoch": 3.2560903149138443,
|
| 1952 |
+
"grad_norm": 17.725006103515625,
|
| 1953 |
+
"learning_rate": 5.086927816901409e-05,
|
| 1954 |
+
"loss": 1.4803,
|
| 1955 |
+
"step": 5480
|
| 1956 |
+
},
|
| 1957 |
+
{
|
| 1958 |
+
"epoch": 3.2679738562091503,
|
| 1959 |
+
"grad_norm": 14.433451652526855,
|
| 1960 |
+
"learning_rate": 5.0649207746478876e-05,
|
| 1961 |
+
"loss": 1.3783,
|
| 1962 |
+
"step": 5500
|
| 1963 |
+
},
|
| 1964 |
+
{
|
| 1965 |
+
"epoch": 3.2798573975044563,
|
| 1966 |
+
"grad_norm": 13.463130950927734,
|
| 1967 |
+
"learning_rate": 5.042913732394367e-05,
|
| 1968 |
+
"loss": 1.5348,
|
| 1969 |
+
"step": 5520
|
| 1970 |
+
},
|
| 1971 |
+
{
|
| 1972 |
+
"epoch": 3.2917409387997623,
|
| 1973 |
+
"grad_norm": 6.446109771728516,
|
| 1974 |
+
"learning_rate": 5.020906690140845e-05,
|
| 1975 |
+
"loss": 1.805,
|
| 1976 |
+
"step": 5540
|
| 1977 |
+
},
|
| 1978 |
+
{
|
| 1979 |
+
"epoch": 3.3036244800950683,
|
| 1980 |
+
"grad_norm": 10.89792251586914,
|
| 1981 |
+
"learning_rate": 4.998899647887324e-05,
|
| 1982 |
+
"loss": 1.5201,
|
| 1983 |
+
"step": 5560
|
| 1984 |
+
},
|
| 1985 |
+
{
|
| 1986 |
+
"epoch": 3.3155080213903743,
|
| 1987 |
+
"grad_norm": 12.57170581817627,
|
| 1988 |
+
"learning_rate": 4.976892605633803e-05,
|
| 1989 |
+
"loss": 1.5706,
|
| 1990 |
+
"step": 5580
|
| 1991 |
+
},
|
| 1992 |
+
{
|
| 1993 |
+
"epoch": 3.3273915626856803,
|
| 1994 |
+
"grad_norm": 5.958362102508545,
|
| 1995 |
+
"learning_rate": 4.9548855633802824e-05,
|
| 1996 |
+
"loss": 1.4606,
|
| 1997 |
+
"step": 5600
|
| 1998 |
+
},
|
| 1999 |
+
{
|
| 2000 |
+
"epoch": 3.3392751039809863,
|
| 2001 |
+
"grad_norm": 12.095161437988281,
|
| 2002 |
+
"learning_rate": 4.932878521126761e-05,
|
| 2003 |
+
"loss": 1.5633,
|
| 2004 |
+
"step": 5620
|
| 2005 |
+
},
|
| 2006 |
+
{
|
| 2007 |
+
"epoch": 3.3511586452762923,
|
| 2008 |
+
"grad_norm": 17.618438720703125,
|
| 2009 |
+
"learning_rate": 4.91087147887324e-05,
|
| 2010 |
+
"loss": 1.5763,
|
| 2011 |
+
"step": 5640
|
| 2012 |
+
},
|
| 2013 |
+
{
|
| 2014 |
+
"epoch": 3.3630421865715983,
|
| 2015 |
+
"grad_norm": 5.972578048706055,
|
| 2016 |
+
"learning_rate": 4.888864436619718e-05,
|
| 2017 |
+
"loss": 1.0006,
|
| 2018 |
+
"step": 5660
|
| 2019 |
+
},
|
| 2020 |
+
{
|
| 2021 |
+
"epoch": 3.3749257278669043,
|
| 2022 |
+
"grad_norm": 11.534244537353516,
|
| 2023 |
+
"learning_rate": 4.8668573943661974e-05,
|
| 2024 |
+
"loss": 1.9326,
|
| 2025 |
+
"step": 5680
|
| 2026 |
+
},
|
| 2027 |
+
{
|
| 2028 |
+
"epoch": 3.3868092691622103,
|
| 2029 |
+
"grad_norm": 7.13762092590332,
|
| 2030 |
+
"learning_rate": 4.8448503521126765e-05,
|
| 2031 |
+
"loss": 1.369,
|
| 2032 |
+
"step": 5700
|
| 2033 |
+
},
|
| 2034 |
+
{
|
| 2035 |
+
"epoch": 3.3986928104575163,
|
| 2036 |
+
"grad_norm": 5.443957328796387,
|
| 2037 |
+
"learning_rate": 4.822843309859155e-05,
|
| 2038 |
+
"loss": 1.6741,
|
| 2039 |
+
"step": 5720
|
| 2040 |
+
},
|
| 2041 |
+
{
|
| 2042 |
+
"epoch": 3.4105763517528223,
|
| 2043 |
+
"grad_norm": 5.366059303283691,
|
| 2044 |
+
"learning_rate": 4.800836267605634e-05,
|
| 2045 |
+
"loss": 1.6731,
|
| 2046 |
+
"step": 5740
|
| 2047 |
+
},
|
| 2048 |
+
{
|
| 2049 |
+
"epoch": 3.4224598930481283,
|
| 2050 |
+
"grad_norm": 6.048867225646973,
|
| 2051 |
+
"learning_rate": 4.7788292253521124e-05,
|
| 2052 |
+
"loss": 1.3663,
|
| 2053 |
+
"step": 5760
|
| 2054 |
+
},
|
| 2055 |
+
{
|
| 2056 |
+
"epoch": 3.4343434343434343,
|
| 2057 |
+
"grad_norm": 14.81528377532959,
|
| 2058 |
+
"learning_rate": 4.7568221830985915e-05,
|
| 2059 |
+
"loss": 2.319,
|
| 2060 |
+
"step": 5780
|
| 2061 |
+
},
|
| 2062 |
+
{
|
| 2063 |
+
"epoch": 3.4462269756387403,
|
| 2064 |
+
"grad_norm": 6.7229814529418945,
|
| 2065 |
+
"learning_rate": 4.7348151408450706e-05,
|
| 2066 |
+
"loss": 1.5563,
|
| 2067 |
+
"step": 5800
|
| 2068 |
+
},
|
| 2069 |
+
{
|
| 2070 |
+
"epoch": 3.4581105169340463,
|
| 2071 |
+
"grad_norm": 7.397293567657471,
|
| 2072 |
+
"learning_rate": 4.71280809859155e-05,
|
| 2073 |
+
"loss": 1.6003,
|
| 2074 |
+
"step": 5820
|
| 2075 |
+
},
|
| 2076 |
+
{
|
| 2077 |
+
"epoch": 3.4699940582293523,
|
| 2078 |
+
"grad_norm": 6.6656718254089355,
|
| 2079 |
+
"learning_rate": 4.690801056338028e-05,
|
| 2080 |
+
"loss": 1.0575,
|
| 2081 |
+
"step": 5840
|
| 2082 |
+
},
|
| 2083 |
+
{
|
| 2084 |
+
"epoch": 3.4818775995246583,
|
| 2085 |
+
"grad_norm": 12.698165893554688,
|
| 2086 |
+
"learning_rate": 4.668794014084507e-05,
|
| 2087 |
+
"loss": 1.7236,
|
| 2088 |
+
"step": 5860
|
| 2089 |
+
},
|
| 2090 |
+
{
|
| 2091 |
+
"epoch": 3.4937611408199643,
|
| 2092 |
+
"grad_norm": 5.691989898681641,
|
| 2093 |
+
"learning_rate": 4.6467869718309856e-05,
|
| 2094 |
+
"loss": 1.3636,
|
| 2095 |
+
"step": 5880
|
| 2096 |
+
},
|
| 2097 |
+
{
|
| 2098 |
+
"epoch": 3.5056446821152702,
|
| 2099 |
+
"grad_norm": 5.1549153327941895,
|
| 2100 |
+
"learning_rate": 4.6247799295774654e-05,
|
| 2101 |
+
"loss": 1.0757,
|
| 2102 |
+
"step": 5900
|
| 2103 |
+
},
|
| 2104 |
+
{
|
| 2105 |
+
"epoch": 3.5175282234105762,
|
| 2106 |
+
"grad_norm": 6.807792663574219,
|
| 2107 |
+
"learning_rate": 4.602772887323944e-05,
|
| 2108 |
+
"loss": 1.4264,
|
| 2109 |
+
"step": 5920
|
| 2110 |
+
},
|
| 2111 |
+
{
|
| 2112 |
+
"epoch": 3.5294117647058822,
|
| 2113 |
+
"grad_norm": 11.486824035644531,
|
| 2114 |
+
"learning_rate": 4.580765845070423e-05,
|
| 2115 |
+
"loss": 1.5486,
|
| 2116 |
+
"step": 5940
|
| 2117 |
+
},
|
| 2118 |
+
{
|
| 2119 |
+
"epoch": 3.5412953060011882,
|
| 2120 |
+
"grad_norm": 19.278827667236328,
|
| 2121 |
+
"learning_rate": 4.558758802816901e-05,
|
| 2122 |
+
"loss": 1.8427,
|
| 2123 |
+
"step": 5960
|
| 2124 |
+
},
|
| 2125 |
+
{
|
| 2126 |
+
"epoch": 3.5531788472964942,
|
| 2127 |
+
"grad_norm": 8.828113555908203,
|
| 2128 |
+
"learning_rate": 4.5367517605633804e-05,
|
| 2129 |
+
"loss": 1.6961,
|
| 2130 |
+
"step": 5980
|
| 2131 |
+
},
|
| 2132 |
+
{
|
| 2133 |
+
"epoch": 3.5650623885918002,
|
| 2134 |
+
"grad_norm": 6.939361095428467,
|
| 2135 |
+
"learning_rate": 4.5147447183098595e-05,
|
| 2136 |
+
"loss": 1.4189,
|
| 2137 |
+
"step": 6000
|
| 2138 |
+
},
|
| 2139 |
+
{
|
| 2140 |
+
"epoch": 3.5769459298871062,
|
| 2141 |
+
"grad_norm": 8.03252124786377,
|
| 2142 |
+
"learning_rate": 4.4927376760563386e-05,
|
| 2143 |
+
"loss": 1.5236,
|
| 2144 |
+
"step": 6020
|
| 2145 |
+
},
|
| 2146 |
+
{
|
| 2147 |
+
"epoch": 3.5888294711824122,
|
| 2148 |
+
"grad_norm": 14.845807075500488,
|
| 2149 |
+
"learning_rate": 4.470730633802817e-05,
|
| 2150 |
+
"loss": 1.3323,
|
| 2151 |
+
"step": 6040
|
| 2152 |
+
},
|
| 2153 |
+
{
|
| 2154 |
+
"epoch": 3.6007130124777182,
|
| 2155 |
+
"grad_norm": 6.234992504119873,
|
| 2156 |
+
"learning_rate": 4.448723591549296e-05,
|
| 2157 |
+
"loss": 1.5554,
|
| 2158 |
+
"step": 6060
|
| 2159 |
+
},
|
| 2160 |
+
{
|
| 2161 |
+
"epoch": 3.6125965537730242,
|
| 2162 |
+
"grad_norm": 9.308655738830566,
|
| 2163 |
+
"learning_rate": 4.4278169014084505e-05,
|
| 2164 |
+
"loss": 1.7403,
|
| 2165 |
+
"step": 6080
|
| 2166 |
+
},
|
| 2167 |
+
{
|
| 2168 |
+
"epoch": 3.6244800950683302,
|
| 2169 |
+
"grad_norm": 6.452903747558594,
|
| 2170 |
+
"learning_rate": 4.4058098591549296e-05,
|
| 2171 |
+
"loss": 1.4449,
|
| 2172 |
+
"step": 6100
|
| 2173 |
+
},
|
| 2174 |
+
{
|
| 2175 |
+
"epoch": 3.6363636363636362,
|
| 2176 |
+
"grad_norm": 5.176253318786621,
|
| 2177 |
+
"learning_rate": 4.383802816901409e-05,
|
| 2178 |
+
"loss": 1.7978,
|
| 2179 |
+
"step": 6120
|
| 2180 |
+
},
|
| 2181 |
+
{
|
| 2182 |
+
"epoch": 3.6482471776589422,
|
| 2183 |
+
"grad_norm": 4.04074764251709,
|
| 2184 |
+
"learning_rate": 4.361795774647888e-05,
|
| 2185 |
+
"loss": 1.444,
|
| 2186 |
+
"step": 6140
|
| 2187 |
+
},
|
| 2188 |
+
{
|
| 2189 |
+
"epoch": 3.6601307189542482,
|
| 2190 |
+
"grad_norm": 0.018241392448544502,
|
| 2191 |
+
"learning_rate": 4.339788732394366e-05,
|
| 2192 |
+
"loss": 1.2793,
|
| 2193 |
+
"step": 6160
|
| 2194 |
+
},
|
| 2195 |
+
{
|
| 2196 |
+
"epoch": 3.6720142602495542,
|
| 2197 |
+
"grad_norm": 25.371074676513672,
|
| 2198 |
+
"learning_rate": 4.317781690140845e-05,
|
| 2199 |
+
"loss": 1.6718,
|
| 2200 |
+
"step": 6180
|
| 2201 |
+
},
|
| 2202 |
+
{
|
| 2203 |
+
"epoch": 3.6838978015448602,
|
| 2204 |
+
"grad_norm": 9.880302429199219,
|
| 2205 |
+
"learning_rate": 4.295774647887324e-05,
|
| 2206 |
+
"loss": 1.7344,
|
| 2207 |
+
"step": 6200
|
| 2208 |
+
},
|
| 2209 |
+
{
|
| 2210 |
+
"epoch": 3.6957813428401662,
|
| 2211 |
+
"grad_norm": 17.729167938232422,
|
| 2212 |
+
"learning_rate": 4.2737676056338035e-05,
|
| 2213 |
+
"loss": 1.6367,
|
| 2214 |
+
"step": 6220
|
| 2215 |
+
},
|
| 2216 |
+
{
|
| 2217 |
+
"epoch": 3.7076648841354722,
|
| 2218 |
+
"grad_norm": 7.4893798828125,
|
| 2219 |
+
"learning_rate": 4.251760563380282e-05,
|
| 2220 |
+
"loss": 1.2116,
|
| 2221 |
+
"step": 6240
|
| 2222 |
+
},
|
| 2223 |
+
{
|
| 2224 |
+
"epoch": 3.7195484254307782,
|
| 2225 |
+
"grad_norm": 7.023115158081055,
|
| 2226 |
+
"learning_rate": 4.229753521126761e-05,
|
| 2227 |
+
"loss": 1.2528,
|
| 2228 |
+
"step": 6260
|
| 2229 |
+
},
|
| 2230 |
+
{
|
| 2231 |
+
"epoch": 3.7314319667260842,
|
| 2232 |
+
"grad_norm": 1.672008752822876,
|
| 2233 |
+
"learning_rate": 4.2077464788732394e-05,
|
| 2234 |
+
"loss": 1.3507,
|
| 2235 |
+
"step": 6280
|
| 2236 |
+
},
|
| 2237 |
+
{
|
| 2238 |
+
"epoch": 3.7433155080213902,
|
| 2239 |
+
"grad_norm": 7.472415924072266,
|
| 2240 |
+
"learning_rate": 4.1857394366197185e-05,
|
| 2241 |
+
"loss": 1.5015,
|
| 2242 |
+
"step": 6300
|
| 2243 |
+
},
|
| 2244 |
+
{
|
| 2245 |
+
"epoch": 3.7551990493166962,
|
| 2246 |
+
"grad_norm": 11.470372200012207,
|
| 2247 |
+
"learning_rate": 4.163732394366197e-05,
|
| 2248 |
+
"loss": 1.0308,
|
| 2249 |
+
"step": 6320
|
| 2250 |
+
},
|
| 2251 |
+
{
|
| 2252 |
+
"epoch": 3.7670825906120022,
|
| 2253 |
+
"grad_norm": 8.578750610351562,
|
| 2254 |
+
"learning_rate": 4.141725352112677e-05,
|
| 2255 |
+
"loss": 1.3318,
|
| 2256 |
+
"step": 6340
|
| 2257 |
+
},
|
| 2258 |
+
{
|
| 2259 |
+
"epoch": 3.7789661319073082,
|
| 2260 |
+
"grad_norm": 7.921478748321533,
|
| 2261 |
+
"learning_rate": 4.119718309859155e-05,
|
| 2262 |
+
"loss": 1.5459,
|
| 2263 |
+
"step": 6360
|
| 2264 |
+
},
|
| 2265 |
+
{
|
| 2266 |
+
"epoch": 3.7908496732026142,
|
| 2267 |
+
"grad_norm": 14.853532791137695,
|
| 2268 |
+
"learning_rate": 4.097711267605634e-05,
|
| 2269 |
+
"loss": 1.8492,
|
| 2270 |
+
"step": 6380
|
| 2271 |
+
},
|
| 2272 |
+
{
|
| 2273 |
+
"epoch": 3.8027332144979202,
|
| 2274 |
+
"grad_norm": 24.788694381713867,
|
| 2275 |
+
"learning_rate": 4.0757042253521126e-05,
|
| 2276 |
+
"loss": 1.8012,
|
| 2277 |
+
"step": 6400
|
| 2278 |
+
},
|
| 2279 |
+
{
|
| 2280 |
+
"epoch": 3.8146167557932262,
|
| 2281 |
+
"grad_norm": 2.027362108230591,
|
| 2282 |
+
"learning_rate": 4.053697183098592e-05,
|
| 2283 |
+
"loss": 1.7795,
|
| 2284 |
+
"step": 6420
|
| 2285 |
+
},
|
| 2286 |
+
{
|
| 2287 |
+
"epoch": 3.8265002970885322,
|
| 2288 |
+
"grad_norm": 5.183379650115967,
|
| 2289 |
+
"learning_rate": 4.031690140845071e-05,
|
| 2290 |
+
"loss": 1.4511,
|
| 2291 |
+
"step": 6440
|
| 2292 |
+
},
|
| 2293 |
+
{
|
| 2294 |
+
"epoch": 3.8383838383838382,
|
| 2295 |
+
"grad_norm": 5.597097873687744,
|
| 2296 |
+
"learning_rate": 4.009683098591549e-05,
|
| 2297 |
+
"loss": 1.2832,
|
| 2298 |
+
"step": 6460
|
| 2299 |
+
},
|
| 2300 |
+
{
|
| 2301 |
+
"epoch": 3.8502673796791442,
|
| 2302 |
+
"grad_norm": 9.230513572692871,
|
| 2303 |
+
"learning_rate": 3.987676056338028e-05,
|
| 2304 |
+
"loss": 1.9395,
|
| 2305 |
+
"step": 6480
|
| 2306 |
+
},
|
| 2307 |
+
{
|
| 2308 |
+
"epoch": 3.8621509209744502,
|
| 2309 |
+
"grad_norm": 7.705165386199951,
|
| 2310 |
+
"learning_rate": 3.9656690140845074e-05,
|
| 2311 |
+
"loss": 1.3734,
|
| 2312 |
+
"step": 6500
|
| 2313 |
+
},
|
| 2314 |
+
{
|
| 2315 |
+
"epoch": 3.8740344622697562,
|
| 2316 |
+
"grad_norm": 6.281782150268555,
|
| 2317 |
+
"learning_rate": 3.943661971830986e-05,
|
| 2318 |
+
"loss": 1.4061,
|
| 2319 |
+
"step": 6520
|
| 2320 |
+
},
|
| 2321 |
+
{
|
| 2322 |
+
"epoch": 3.8859180035650622,
|
| 2323 |
+
"grad_norm": 4.201906204223633,
|
| 2324 |
+
"learning_rate": 3.921654929577465e-05,
|
| 2325 |
+
"loss": 1.49,
|
| 2326 |
+
"step": 6540
|
| 2327 |
+
},
|
| 2328 |
+
{
|
| 2329 |
+
"epoch": 3.8978015448603682,
|
| 2330 |
+
"grad_norm": 8.452786445617676,
|
| 2331 |
+
"learning_rate": 3.899647887323944e-05,
|
| 2332 |
+
"loss": 1.3961,
|
| 2333 |
+
"step": 6560
|
| 2334 |
+
},
|
| 2335 |
+
{
|
| 2336 |
+
"epoch": 3.9096850861556742,
|
| 2337 |
+
"grad_norm": 10.373578071594238,
|
| 2338 |
+
"learning_rate": 3.8776408450704224e-05,
|
| 2339 |
+
"loss": 1.489,
|
| 2340 |
+
"step": 6580
|
| 2341 |
+
},
|
| 2342 |
+
{
|
| 2343 |
+
"epoch": 3.9215686274509802,
|
| 2344 |
+
"grad_norm": 7.816731929779053,
|
| 2345 |
+
"learning_rate": 3.8556338028169015e-05,
|
| 2346 |
+
"loss": 1.3659,
|
| 2347 |
+
"step": 6600
|
| 2348 |
+
},
|
| 2349 |
+
{
|
| 2350 |
+
"epoch": 3.9334521687462862,
|
| 2351 |
+
"grad_norm": 3.9288010597229004,
|
| 2352 |
+
"learning_rate": 3.83362676056338e-05,
|
| 2353 |
+
"loss": 1.7783,
|
| 2354 |
+
"step": 6620
|
| 2355 |
+
},
|
| 2356 |
+
{
|
| 2357 |
+
"epoch": 3.9453357100415927,
|
| 2358 |
+
"grad_norm": 16.12210464477539,
|
| 2359 |
+
"learning_rate": 3.81161971830986e-05,
|
| 2360 |
+
"loss": 1.8105,
|
| 2361 |
+
"step": 6640
|
| 2362 |
+
},
|
| 2363 |
+
{
|
| 2364 |
+
"epoch": 3.9572192513368982,
|
| 2365 |
+
"grad_norm": 3.387218713760376,
|
| 2366 |
+
"learning_rate": 3.789612676056338e-05,
|
| 2367 |
+
"loss": 1.6145,
|
| 2368 |
+
"step": 6660
|
| 2369 |
+
},
|
| 2370 |
+
{
|
| 2371 |
+
"epoch": 3.9691027926322047,
|
| 2372 |
+
"grad_norm": 9.880513191223145,
|
| 2373 |
+
"learning_rate": 3.767605633802817e-05,
|
| 2374 |
+
"loss": 1.7195,
|
| 2375 |
+
"step": 6680
|
| 2376 |
+
},
|
| 2377 |
+
{
|
| 2378 |
+
"epoch": 3.9809863339275102,
|
| 2379 |
+
"grad_norm": 16.724794387817383,
|
| 2380 |
+
"learning_rate": 3.7455985915492956e-05,
|
| 2381 |
+
"loss": 1.8867,
|
| 2382 |
+
"step": 6700
|
| 2383 |
+
},
|
| 2384 |
+
{
|
| 2385 |
+
"epoch": 3.9928698752228167,
|
| 2386 |
+
"grad_norm": 11.043305397033691,
|
| 2387 |
+
"learning_rate": 3.723591549295775e-05,
|
| 2388 |
+
"loss": 1.2336,
|
| 2389 |
+
"step": 6720
|
| 2390 |
+
},
|
| 2391 |
+
{
|
| 2392 |
+
"epoch": 4.0,
|
| 2393 |
+
"eval_accuracy": 0.4074074074074074,
|
| 2394 |
+
"eval_loss": 1.902851939201355,
|
| 2395 |
+
"eval_runtime": 105.2207,
|
| 2396 |
+
"eval_samples_per_second": 2.823,
|
| 2397 |
+
"eval_steps_per_second": 2.823,
|
| 2398 |
+
"step": 6732
|
| 2399 |
+
},
|
| 2400 |
+
{
|
| 2401 |
+
"epoch": 4.004753416518122,
|
| 2402 |
+
"grad_norm": 1.1523293256759644,
|
| 2403 |
+
"learning_rate": 3.701584507042254e-05,
|
| 2404 |
+
"loss": 1.6764,
|
| 2405 |
+
"step": 6740
|
| 2406 |
+
},
|
| 2407 |
+
{
|
| 2408 |
+
"epoch": 4.016636957813429,
|
| 2409 |
+
"grad_norm": 1.856052279472351,
|
| 2410 |
+
"learning_rate": 3.679577464788733e-05,
|
| 2411 |
+
"loss": 2.0626,
|
| 2412 |
+
"step": 6760
|
| 2413 |
+
},
|
| 2414 |
+
{
|
| 2415 |
+
"epoch": 4.028520499108734,
|
| 2416 |
+
"grad_norm": 4.87468147277832,
|
| 2417 |
+
"learning_rate": 3.657570422535211e-05,
|
| 2418 |
+
"loss": 1.3926,
|
| 2419 |
+
"step": 6780
|
| 2420 |
+
},
|
| 2421 |
+
{
|
| 2422 |
+
"epoch": 4.040404040404041,
|
| 2423 |
+
"grad_norm": 8.959086418151855,
|
| 2424 |
+
"learning_rate": 3.6355633802816904e-05,
|
| 2425 |
+
"loss": 1.2396,
|
| 2426 |
+
"step": 6800
|
| 2427 |
+
},
|
| 2428 |
+
{
|
| 2429 |
+
"epoch": 4.052287581699346,
|
| 2430 |
+
"grad_norm": 9.787593841552734,
|
| 2431 |
+
"learning_rate": 3.613556338028169e-05,
|
| 2432 |
+
"loss": 1.2043,
|
| 2433 |
+
"step": 6820
|
| 2434 |
+
},
|
| 2435 |
+
{
|
| 2436 |
+
"epoch": 4.064171122994653,
|
| 2437 |
+
"grad_norm": 20.61239242553711,
|
| 2438 |
+
"learning_rate": 3.5915492957746486e-05,
|
| 2439 |
+
"loss": 1.3016,
|
| 2440 |
+
"step": 6840
|
| 2441 |
+
},
|
| 2442 |
+
{
|
| 2443 |
+
"epoch": 4.076054664289958,
|
| 2444 |
+
"grad_norm": 8.819070816040039,
|
| 2445 |
+
"learning_rate": 3.569542253521127e-05,
|
| 2446 |
+
"loss": 1.3577,
|
| 2447 |
+
"step": 6860
|
| 2448 |
+
},
|
| 2449 |
+
{
|
| 2450 |
+
"epoch": 4.087938205585265,
|
| 2451 |
+
"grad_norm": 25.248939514160156,
|
| 2452 |
+
"learning_rate": 3.547535211267606e-05,
|
| 2453 |
+
"loss": 1.7398,
|
| 2454 |
+
"step": 6880
|
| 2455 |
+
},
|
| 2456 |
+
{
|
| 2457 |
+
"epoch": 4.09982174688057,
|
| 2458 |
+
"grad_norm": 9.243005752563477,
|
| 2459 |
+
"learning_rate": 3.5255281690140845e-05,
|
| 2460 |
+
"loss": 1.1347,
|
| 2461 |
+
"step": 6900
|
| 2462 |
+
},
|
| 2463 |
+
{
|
| 2464 |
+
"epoch": 4.111705288175877,
|
| 2465 |
+
"grad_norm": 9.166324615478516,
|
| 2466 |
+
"learning_rate": 3.5035211267605636e-05,
|
| 2467 |
+
"loss": 1.4357,
|
| 2468 |
+
"step": 6920
|
| 2469 |
+
},
|
| 2470 |
+
{
|
| 2471 |
+
"epoch": 4.123588829471182,
|
| 2472 |
+
"grad_norm": 15.312362670898438,
|
| 2473 |
+
"learning_rate": 3.481514084507043e-05,
|
| 2474 |
+
"loss": 1.0013,
|
| 2475 |
+
"step": 6940
|
| 2476 |
+
},
|
| 2477 |
+
{
|
| 2478 |
+
"epoch": 4.135472370766489,
|
| 2479 |
+
"grad_norm": 2.511606454849243,
|
| 2480 |
+
"learning_rate": 3.459507042253521e-05,
|
| 2481 |
+
"loss": 1.3722,
|
| 2482 |
+
"step": 6960
|
| 2483 |
+
},
|
| 2484 |
+
{
|
| 2485 |
+
"epoch": 4.147355912061794,
|
| 2486 |
+
"grad_norm": 5.319163799285889,
|
| 2487 |
+
"learning_rate": 3.4375e-05,
|
| 2488 |
+
"loss": 1.8293,
|
| 2489 |
+
"step": 6980
|
| 2490 |
+
},
|
| 2491 |
+
{
|
| 2492 |
+
"epoch": 4.159239453357101,
|
| 2493 |
+
"grad_norm": 5.499176979064941,
|
| 2494 |
+
"learning_rate": 3.4154929577464786e-05,
|
| 2495 |
+
"loss": 1.2195,
|
| 2496 |
+
"step": 7000
|
| 2497 |
+
},
|
| 2498 |
+
{
|
| 2499 |
+
"epoch": 4.171122994652406,
|
| 2500 |
+
"grad_norm": 5.21409797668457,
|
| 2501 |
+
"learning_rate": 3.393485915492958e-05,
|
| 2502 |
+
"loss": 1.6016,
|
| 2503 |
+
"step": 7020
|
| 2504 |
+
},
|
| 2505 |
+
{
|
| 2506 |
+
"epoch": 4.183006535947713,
|
| 2507 |
+
"grad_norm": 11.814042091369629,
|
| 2508 |
+
"learning_rate": 3.371478873239437e-05,
|
| 2509 |
+
"loss": 1.1857,
|
| 2510 |
+
"step": 7040
|
| 2511 |
+
},
|
| 2512 |
+
{
|
| 2513 |
+
"epoch": 4.194890077243018,
|
| 2514 |
+
"grad_norm": 10.077415466308594,
|
| 2515 |
+
"learning_rate": 3.349471830985916e-05,
|
| 2516 |
+
"loss": 1.5763,
|
| 2517 |
+
"step": 7060
|
| 2518 |
+
},
|
| 2519 |
+
{
|
| 2520 |
+
"epoch": 4.206773618538325,
|
| 2521 |
+
"grad_norm": 8.21226692199707,
|
| 2522 |
+
"learning_rate": 3.327464788732394e-05,
|
| 2523 |
+
"loss": 1.318,
|
| 2524 |
+
"step": 7080
|
| 2525 |
+
},
|
| 2526 |
+
{
|
| 2527 |
+
"epoch": 4.21865715983363,
|
| 2528 |
+
"grad_norm": 6.55881929397583,
|
| 2529 |
+
"learning_rate": 3.3054577464788734e-05,
|
| 2530 |
+
"loss": 1.5631,
|
| 2531 |
+
"step": 7100
|
| 2532 |
+
},
|
| 2533 |
+
{
|
| 2534 |
+
"epoch": 4.230540701128937,
|
| 2535 |
+
"grad_norm": 3.800264596939087,
|
| 2536 |
+
"learning_rate": 3.283450704225352e-05,
|
| 2537 |
+
"loss": 1.2667,
|
| 2538 |
+
"step": 7120
|
| 2539 |
+
},
|
| 2540 |
+
{
|
| 2541 |
+
"epoch": 4.242424242424242,
|
| 2542 |
+
"grad_norm": 24.056673049926758,
|
| 2543 |
+
"learning_rate": 3.2614436619718316e-05,
|
| 2544 |
+
"loss": 1.9465,
|
| 2545 |
+
"step": 7140
|
| 2546 |
+
},
|
| 2547 |
+
{
|
| 2548 |
+
"epoch": 4.254307783719549,
|
| 2549 |
+
"grad_norm": 15.87082576751709,
|
| 2550 |
+
"learning_rate": 3.23943661971831e-05,
|
| 2551 |
+
"loss": 1.2227,
|
| 2552 |
+
"step": 7160
|
| 2553 |
+
},
|
| 2554 |
+
{
|
| 2555 |
+
"epoch": 4.266191325014854,
|
| 2556 |
+
"grad_norm": 5.343487739562988,
|
| 2557 |
+
"learning_rate": 3.217429577464789e-05,
|
| 2558 |
+
"loss": 1.3855,
|
| 2559 |
+
"step": 7180
|
| 2560 |
+
},
|
| 2561 |
+
{
|
| 2562 |
+
"epoch": 4.278074866310161,
|
| 2563 |
+
"grad_norm": 11.594643592834473,
|
| 2564 |
+
"learning_rate": 3.1954225352112675e-05,
|
| 2565 |
+
"loss": 1.6792,
|
| 2566 |
+
"step": 7200
|
| 2567 |
+
},
|
| 2568 |
+
{
|
| 2569 |
+
"epoch": 4.289958407605466,
|
| 2570 |
+
"grad_norm": 3.1940860748291016,
|
| 2571 |
+
"learning_rate": 3.1734154929577466e-05,
|
| 2572 |
+
"loss": 1.0904,
|
| 2573 |
+
"step": 7220
|
| 2574 |
+
},
|
| 2575 |
+
{
|
| 2576 |
+
"epoch": 4.301841948900773,
|
| 2577 |
+
"grad_norm": 7.972840309143066,
|
| 2578 |
+
"learning_rate": 3.151408450704226e-05,
|
| 2579 |
+
"loss": 1.4883,
|
| 2580 |
+
"step": 7240
|
| 2581 |
+
},
|
| 2582 |
+
{
|
| 2583 |
+
"epoch": 4.313725490196078,
|
| 2584 |
+
"grad_norm": 7.334961891174316,
|
| 2585 |
+
"learning_rate": 3.129401408450705e-05,
|
| 2586 |
+
"loss": 1.8654,
|
| 2587 |
+
"step": 7260
|
| 2588 |
+
},
|
| 2589 |
+
{
|
| 2590 |
+
"epoch": 4.325609031491385,
|
| 2591 |
+
"grad_norm": 4.948212623596191,
|
| 2592 |
+
"learning_rate": 3.107394366197183e-05,
|
| 2593 |
+
"loss": 1.4228,
|
| 2594 |
+
"step": 7280
|
| 2595 |
+
},
|
| 2596 |
+
{
|
| 2597 |
+
"epoch": 4.33749257278669,
|
| 2598 |
+
"grad_norm": 18.957931518554688,
|
| 2599 |
+
"learning_rate": 3.085387323943662e-05,
|
| 2600 |
+
"loss": 1.4826,
|
| 2601 |
+
"step": 7300
|
| 2602 |
+
},
|
| 2603 |
+
{
|
| 2604 |
+
"epoch": 4.349376114081997,
|
| 2605 |
+
"grad_norm": 20.902833938598633,
|
| 2606 |
+
"learning_rate": 3.063380281690141e-05,
|
| 2607 |
+
"loss": 1.2242,
|
| 2608 |
+
"step": 7320
|
| 2609 |
+
},
|
| 2610 |
+
{
|
| 2611 |
+
"epoch": 4.361259655377302,
|
| 2612 |
+
"grad_norm": 25.95677375793457,
|
| 2613 |
+
"learning_rate": 3.04137323943662e-05,
|
| 2614 |
+
"loss": 0.779,
|
| 2615 |
+
"step": 7340
|
| 2616 |
+
},
|
| 2617 |
+
{
|
| 2618 |
+
"epoch": 4.373143196672609,
|
| 2619 |
+
"grad_norm": 10.477136611938477,
|
| 2620 |
+
"learning_rate": 3.019366197183099e-05,
|
| 2621 |
+
"loss": 1.1637,
|
| 2622 |
+
"step": 7360
|
| 2623 |
+
},
|
| 2624 |
+
{
|
| 2625 |
+
"epoch": 4.385026737967914,
|
| 2626 |
+
"grad_norm": 7.310263633728027,
|
| 2627 |
+
"learning_rate": 2.9973591549295776e-05,
|
| 2628 |
+
"loss": 1.6564,
|
| 2629 |
+
"step": 7380
|
| 2630 |
+
},
|
| 2631 |
+
{
|
| 2632 |
+
"epoch": 4.396910279263221,
|
| 2633 |
+
"grad_norm": 48.1497688293457,
|
| 2634 |
+
"learning_rate": 2.9753521126760564e-05,
|
| 2635 |
+
"loss": 2.1001,
|
| 2636 |
+
"step": 7400
|
| 2637 |
+
},
|
| 2638 |
+
{
|
| 2639 |
+
"epoch": 4.408793820558526,
|
| 2640 |
+
"grad_norm": 11.60807991027832,
|
| 2641 |
+
"learning_rate": 2.953345070422535e-05,
|
| 2642 |
+
"loss": 1.7998,
|
| 2643 |
+
"step": 7420
|
| 2644 |
+
},
|
| 2645 |
+
{
|
| 2646 |
+
"epoch": 4.420677361853833,
|
| 2647 |
+
"grad_norm": 3.019927501678467,
|
| 2648 |
+
"learning_rate": 2.9313380281690146e-05,
|
| 2649 |
+
"loss": 1.2465,
|
| 2650 |
+
"step": 7440
|
| 2651 |
+
},
|
| 2652 |
+
{
|
| 2653 |
+
"epoch": 4.432560903149138,
|
| 2654 |
+
"grad_norm": 10.465686798095703,
|
| 2655 |
+
"learning_rate": 2.9093309859154933e-05,
|
| 2656 |
+
"loss": 1.1023,
|
| 2657 |
+
"step": 7460
|
| 2658 |
+
},
|
| 2659 |
+
{
|
| 2660 |
+
"epoch": 4.444444444444445,
|
| 2661 |
+
"grad_norm": 6.366426467895508,
|
| 2662 |
+
"learning_rate": 2.887323943661972e-05,
|
| 2663 |
+
"loss": 1.0901,
|
| 2664 |
+
"step": 7480
|
| 2665 |
+
},
|
| 2666 |
+
{
|
| 2667 |
+
"epoch": 4.45632798573975,
|
| 2668 |
+
"grad_norm": 20.92388343811035,
|
| 2669 |
+
"learning_rate": 2.865316901408451e-05,
|
| 2670 |
+
"loss": 1.2325,
|
| 2671 |
+
"step": 7500
|
| 2672 |
+
},
|
| 2673 |
+
{
|
| 2674 |
+
"epoch": 4.468211527035057,
|
| 2675 |
+
"grad_norm": 17.193758010864258,
|
| 2676 |
+
"learning_rate": 2.8433098591549296e-05,
|
| 2677 |
+
"loss": 1.3298,
|
| 2678 |
+
"step": 7520
|
| 2679 |
+
},
|
| 2680 |
+
{
|
| 2681 |
+
"epoch": 4.480095068330362,
|
| 2682 |
+
"grad_norm": 5.884913921356201,
|
| 2683 |
+
"learning_rate": 2.8213028169014087e-05,
|
| 2684 |
+
"loss": 1.2775,
|
| 2685 |
+
"step": 7540
|
| 2686 |
+
},
|
| 2687 |
+
{
|
| 2688 |
+
"epoch": 4.491978609625669,
|
| 2689 |
+
"grad_norm": 6.033341407775879,
|
| 2690 |
+
"learning_rate": 2.7992957746478874e-05,
|
| 2691 |
+
"loss": 1.4184,
|
| 2692 |
+
"step": 7560
|
| 2693 |
+
},
|
| 2694 |
+
{
|
| 2695 |
+
"epoch": 4.503862150920974,
|
| 2696 |
+
"grad_norm": 44.68092727661133,
|
| 2697 |
+
"learning_rate": 2.7772887323943665e-05,
|
| 2698 |
+
"loss": 1.4361,
|
| 2699 |
+
"step": 7580
|
| 2700 |
+
},
|
| 2701 |
+
{
|
| 2702 |
+
"epoch": 4.515745692216281,
|
| 2703 |
+
"grad_norm": 11.81906795501709,
|
| 2704 |
+
"learning_rate": 2.7552816901408453e-05,
|
| 2705 |
+
"loss": 1.578,
|
| 2706 |
+
"step": 7600
|
| 2707 |
+
},
|
| 2708 |
+
{
|
| 2709 |
+
"epoch": 4.527629233511586,
|
| 2710 |
+
"grad_norm": 12.85550594329834,
|
| 2711 |
+
"learning_rate": 2.733274647887324e-05,
|
| 2712 |
+
"loss": 1.439,
|
| 2713 |
+
"step": 7620
|
| 2714 |
+
},
|
| 2715 |
+
{
|
| 2716 |
+
"epoch": 4.539512774806893,
|
| 2717 |
+
"grad_norm": 5.915256977081299,
|
| 2718 |
+
"learning_rate": 2.711267605633803e-05,
|
| 2719 |
+
"loss": 1.4547,
|
| 2720 |
+
"step": 7640
|
| 2721 |
+
},
|
| 2722 |
+
{
|
| 2723 |
+
"epoch": 4.551396316102198,
|
| 2724 |
+
"grad_norm": 12.035588264465332,
|
| 2725 |
+
"learning_rate": 2.689260563380282e-05,
|
| 2726 |
+
"loss": 1.3403,
|
| 2727 |
+
"step": 7660
|
| 2728 |
+
},
|
| 2729 |
+
{
|
| 2730 |
+
"epoch": 4.563279857397505,
|
| 2731 |
+
"grad_norm": 6.767269611358643,
|
| 2732 |
+
"learning_rate": 2.6672535211267606e-05,
|
| 2733 |
+
"loss": 1.1761,
|
| 2734 |
+
"step": 7680
|
| 2735 |
+
},
|
| 2736 |
+
{
|
| 2737 |
+
"epoch": 4.57516339869281,
|
| 2738 |
+
"grad_norm": 10.033326148986816,
|
| 2739 |
+
"learning_rate": 2.6452464788732394e-05,
|
| 2740 |
+
"loss": 1.352,
|
| 2741 |
+
"step": 7700
|
| 2742 |
+
},
|
| 2743 |
+
{
|
| 2744 |
+
"epoch": 4.587046939988117,
|
| 2745 |
+
"grad_norm": 11.945486068725586,
|
| 2746 |
+
"learning_rate": 2.623239436619718e-05,
|
| 2747 |
+
"loss": 1.5271,
|
| 2748 |
+
"step": 7720
|
| 2749 |
+
},
|
| 2750 |
+
{
|
| 2751 |
+
"epoch": 4.598930481283422,
|
| 2752 |
+
"grad_norm": 15.092402458190918,
|
| 2753 |
+
"learning_rate": 2.601232394366197e-05,
|
| 2754 |
+
"loss": 1.3031,
|
| 2755 |
+
"step": 7740
|
| 2756 |
+
},
|
| 2757 |
+
{
|
| 2758 |
+
"epoch": 4.610814022578729,
|
| 2759 |
+
"grad_norm": 9.422025680541992,
|
| 2760 |
+
"learning_rate": 2.5792253521126763e-05,
|
| 2761 |
+
"loss": 1.2642,
|
| 2762 |
+
"step": 7760
|
| 2763 |
+
},
|
| 2764 |
+
{
|
| 2765 |
+
"epoch": 4.622697563874034,
|
| 2766 |
+
"grad_norm": 6.706630706787109,
|
| 2767 |
+
"learning_rate": 2.557218309859155e-05,
|
| 2768 |
+
"loss": 1.4985,
|
| 2769 |
+
"step": 7780
|
| 2770 |
+
},
|
| 2771 |
+
{
|
| 2772 |
+
"epoch": 4.634581105169341,
|
| 2773 |
+
"grad_norm": 11.070060729980469,
|
| 2774 |
+
"learning_rate": 2.535211267605634e-05,
|
| 2775 |
+
"loss": 1.3693,
|
| 2776 |
+
"step": 7800
|
| 2777 |
+
},
|
| 2778 |
+
{
|
| 2779 |
+
"epoch": 4.646464646464646,
|
| 2780 |
+
"grad_norm": 21.464950561523438,
|
| 2781 |
+
"learning_rate": 2.5132042253521126e-05,
|
| 2782 |
+
"loss": 1.2615,
|
| 2783 |
+
"step": 7820
|
| 2784 |
+
},
|
| 2785 |
+
{
|
| 2786 |
+
"epoch": 4.658348187759953,
|
| 2787 |
+
"grad_norm": 0.3174152076244354,
|
| 2788 |
+
"learning_rate": 2.4911971830985917e-05,
|
| 2789 |
+
"loss": 1.7061,
|
| 2790 |
+
"step": 7840
|
| 2791 |
+
},
|
| 2792 |
+
{
|
| 2793 |
+
"epoch": 4.670231729055258,
|
| 2794 |
+
"grad_norm": 10.314111709594727,
|
| 2795 |
+
"learning_rate": 2.4691901408450704e-05,
|
| 2796 |
+
"loss": 1.2515,
|
| 2797 |
+
"step": 7860
|
| 2798 |
+
},
|
| 2799 |
+
{
|
| 2800 |
+
"epoch": 4.682115270350565,
|
| 2801 |
+
"grad_norm": 1.4397400617599487,
|
| 2802 |
+
"learning_rate": 2.4471830985915495e-05,
|
| 2803 |
+
"loss": 1.4254,
|
| 2804 |
+
"step": 7880
|
| 2805 |
+
},
|
| 2806 |
+
{
|
| 2807 |
+
"epoch": 4.69399881164587,
|
| 2808 |
+
"grad_norm": 11.450037002563477,
|
| 2809 |
+
"learning_rate": 2.4251760563380283e-05,
|
| 2810 |
+
"loss": 1.1362,
|
| 2811 |
+
"step": 7900
|
| 2812 |
+
},
|
| 2813 |
+
{
|
| 2814 |
+
"epoch": 4.705882352941177,
|
| 2815 |
+
"grad_norm": 17.582212448120117,
|
| 2816 |
+
"learning_rate": 2.4031690140845074e-05,
|
| 2817 |
+
"loss": 1.6171,
|
| 2818 |
+
"step": 7920
|
| 2819 |
+
},
|
| 2820 |
+
{
|
| 2821 |
+
"epoch": 4.717765894236482,
|
| 2822 |
+
"grad_norm": 0.15067483484745026,
|
| 2823 |
+
"learning_rate": 2.381161971830986e-05,
|
| 2824 |
+
"loss": 1.0266,
|
| 2825 |
+
"step": 7940
|
| 2826 |
+
},
|
| 2827 |
+
{
|
| 2828 |
+
"epoch": 4.729649435531789,
|
| 2829 |
+
"grad_norm": 0.35611966252326965,
|
| 2830 |
+
"learning_rate": 2.359154929577465e-05,
|
| 2831 |
+
"loss": 1.8023,
|
| 2832 |
+
"step": 7960
|
| 2833 |
+
},
|
| 2834 |
+
{
|
| 2835 |
+
"epoch": 4.741532976827094,
|
| 2836 |
+
"grad_norm": 10.370597839355469,
|
| 2837 |
+
"learning_rate": 2.337147887323944e-05,
|
| 2838 |
+
"loss": 1.8925,
|
| 2839 |
+
"step": 7980
|
| 2840 |
+
},
|
| 2841 |
+
{
|
| 2842 |
+
"epoch": 4.753416518122401,
|
| 2843 |
+
"grad_norm": 5.118237495422363,
|
| 2844 |
+
"learning_rate": 2.3151408450704227e-05,
|
| 2845 |
+
"loss": 1.2258,
|
| 2846 |
+
"step": 8000
|
| 2847 |
+
},
|
| 2848 |
+
{
|
| 2849 |
+
"epoch": 4.765300059417706,
|
| 2850 |
+
"grad_norm": 29.316865921020508,
|
| 2851 |
+
"learning_rate": 2.2931338028169015e-05,
|
| 2852 |
+
"loss": 1.0854,
|
| 2853 |
+
"step": 8020
|
| 2854 |
+
},
|
| 2855 |
+
{
|
| 2856 |
+
"epoch": 4.777183600713013,
|
| 2857 |
+
"grad_norm": 12.539017677307129,
|
| 2858 |
+
"learning_rate": 2.2711267605633806e-05,
|
| 2859 |
+
"loss": 1.529,
|
| 2860 |
+
"step": 8040
|
| 2861 |
+
},
|
| 2862 |
+
{
|
| 2863 |
+
"epoch": 4.789067142008318,
|
| 2864 |
+
"grad_norm": 16.720726013183594,
|
| 2865 |
+
"learning_rate": 2.2491197183098593e-05,
|
| 2866 |
+
"loss": 1.8251,
|
| 2867 |
+
"step": 8060
|
| 2868 |
+
},
|
| 2869 |
+
{
|
| 2870 |
+
"epoch": 4.800950683303625,
|
| 2871 |
+
"grad_norm": 16.409805297851562,
|
| 2872 |
+
"learning_rate": 2.227112676056338e-05,
|
| 2873 |
+
"loss": 1.0862,
|
| 2874 |
+
"step": 8080
|
| 2875 |
+
},
|
| 2876 |
+
{
|
| 2877 |
+
"epoch": 4.81283422459893,
|
| 2878 |
+
"grad_norm": 18.887989044189453,
|
| 2879 |
+
"learning_rate": 2.205105633802817e-05,
|
| 2880 |
+
"loss": 1.2557,
|
| 2881 |
+
"step": 8100
|
| 2882 |
+
},
|
| 2883 |
+
{
|
| 2884 |
+
"epoch": 4.824717765894237,
|
| 2885 |
+
"grad_norm": 2.1975255012512207,
|
| 2886 |
+
"learning_rate": 2.1830985915492956e-05,
|
| 2887 |
+
"loss": 1.2347,
|
| 2888 |
+
"step": 8120
|
| 2889 |
+
},
|
| 2890 |
+
{
|
| 2891 |
+
"epoch": 4.836601307189542,
|
| 2892 |
+
"grad_norm": 0.16356885433197021,
|
| 2893 |
+
"learning_rate": 2.1610915492957747e-05,
|
| 2894 |
+
"loss": 1.2155,
|
| 2895 |
+
"step": 8140
|
| 2896 |
+
},
|
| 2897 |
+
{
|
| 2898 |
+
"epoch": 4.848484848484849,
|
| 2899 |
+
"grad_norm": 27.676633834838867,
|
| 2900 |
+
"learning_rate": 2.1390845070422534e-05,
|
| 2901 |
+
"loss": 2.5601,
|
| 2902 |
+
"step": 8160
|
| 2903 |
+
},
|
| 2904 |
+
{
|
| 2905 |
+
"epoch": 4.860368389780154,
|
| 2906 |
+
"grad_norm": 6.721032619476318,
|
| 2907 |
+
"learning_rate": 2.1170774647887325e-05,
|
| 2908 |
+
"loss": 1.0707,
|
| 2909 |
+
"step": 8180
|
| 2910 |
+
},
|
| 2911 |
+
{
|
| 2912 |
+
"epoch": 4.872251931075461,
|
| 2913 |
+
"grad_norm": 17.30257225036621,
|
| 2914 |
+
"learning_rate": 2.0950704225352113e-05,
|
| 2915 |
+
"loss": 1.305,
|
| 2916 |
+
"step": 8200
|
| 2917 |
+
},
|
| 2918 |
+
{
|
| 2919 |
+
"epoch": 4.884135472370766,
|
| 2920 |
+
"grad_norm": 12.832918167114258,
|
| 2921 |
+
"learning_rate": 2.07306338028169e-05,
|
| 2922 |
+
"loss": 1.2079,
|
| 2923 |
+
"step": 8220
|
| 2924 |
+
},
|
| 2925 |
+
{
|
| 2926 |
+
"epoch": 4.896019013666073,
|
| 2927 |
+
"grad_norm": 25.992534637451172,
|
| 2928 |
+
"learning_rate": 2.051056338028169e-05,
|
| 2929 |
+
"loss": 1.6578,
|
| 2930 |
+
"step": 8240
|
| 2931 |
+
},
|
| 2932 |
+
{
|
| 2933 |
+
"epoch": 4.907902554961378,
|
| 2934 |
+
"grad_norm": 21.847524642944336,
|
| 2935 |
+
"learning_rate": 2.029049295774648e-05,
|
| 2936 |
+
"loss": 1.221,
|
| 2937 |
+
"step": 8260
|
| 2938 |
+
},
|
| 2939 |
+
{
|
| 2940 |
+
"epoch": 4.919786096256685,
|
| 2941 |
+
"grad_norm": 15.508858680725098,
|
| 2942 |
+
"learning_rate": 2.007042253521127e-05,
|
| 2943 |
+
"loss": 1.4104,
|
| 2944 |
+
"step": 8280
|
| 2945 |
+
},
|
| 2946 |
+
{
|
| 2947 |
+
"epoch": 4.93166963755199,
|
| 2948 |
+
"grad_norm": 16.669252395629883,
|
| 2949 |
+
"learning_rate": 1.9850352112676057e-05,
|
| 2950 |
+
"loss": 1.6439,
|
| 2951 |
+
"step": 8300
|
| 2952 |
+
},
|
| 2953 |
+
{
|
| 2954 |
+
"epoch": 4.943553178847297,
|
| 2955 |
+
"grad_norm": 11.705305099487305,
|
| 2956 |
+
"learning_rate": 1.9630281690140845e-05,
|
| 2957 |
+
"loss": 1.1035,
|
| 2958 |
+
"step": 8320
|
| 2959 |
+
},
|
| 2960 |
+
{
|
| 2961 |
+
"epoch": 4.955436720142602,
|
| 2962 |
+
"grad_norm": 6.209210395812988,
|
| 2963 |
+
"learning_rate": 1.9410211267605636e-05,
|
| 2964 |
+
"loss": 1.0942,
|
| 2965 |
+
"step": 8340
|
| 2966 |
+
},
|
| 2967 |
+
{
|
| 2968 |
+
"epoch": 4.967320261437909,
|
| 2969 |
+
"grad_norm": 28.53778839111328,
|
| 2970 |
+
"learning_rate": 1.9190140845070423e-05,
|
| 2971 |
+
"loss": 1.3909,
|
| 2972 |
+
"step": 8360
|
| 2973 |
+
},
|
| 2974 |
+
{
|
| 2975 |
+
"epoch": 4.979203802733214,
|
| 2976 |
+
"grad_norm": 16.432416915893555,
|
| 2977 |
+
"learning_rate": 1.8970070422535214e-05,
|
| 2978 |
+
"loss": 1.7145,
|
| 2979 |
+
"step": 8380
|
| 2980 |
+
},
|
| 2981 |
+
{
|
| 2982 |
+
"epoch": 4.991087344028521,
|
| 2983 |
+
"grad_norm": 50.011844635009766,
|
| 2984 |
+
"learning_rate": 1.8750000000000002e-05,
|
| 2985 |
+
"loss": 1.5421,
|
| 2986 |
+
"step": 8400
|
| 2987 |
+
},
|
| 2988 |
+
{
|
| 2989 |
+
"epoch": 5.0,
|
| 2990 |
+
"eval_accuracy": 0.39730639730639733,
|
| 2991 |
+
"eval_loss": 2.0971696376800537,
|
| 2992 |
+
"eval_runtime": 105.4359,
|
| 2993 |
+
"eval_samples_per_second": 2.817,
|
| 2994 |
+
"eval_steps_per_second": 2.817,
|
| 2995 |
+
"step": 8415
|
| 2996 |
+
},
|
| 2997 |
+
{
|
| 2998 |
+
"epoch": 5.002970885323826,
|
| 2999 |
+
"grad_norm": 0.27923718094825745,
|
| 3000 |
+
"learning_rate": 1.852992957746479e-05,
|
| 3001 |
+
"loss": 0.9886,
|
| 3002 |
+
"step": 8420
|
| 3003 |
+
},
|
| 3004 |
+
{
|
| 3005 |
+
"epoch": 5.014854426619133,
|
| 3006 |
+
"grad_norm": 14.41185188293457,
|
| 3007 |
+
"learning_rate": 1.830985915492958e-05,
|
| 3008 |
+
"loss": 1.1069,
|
| 3009 |
+
"step": 8440
|
| 3010 |
+
},
|
| 3011 |
+
{
|
| 3012 |
+
"epoch": 5.026737967914438,
|
| 3013 |
+
"grad_norm": 10.032986640930176,
|
| 3014 |
+
"learning_rate": 1.8089788732394368e-05,
|
| 3015 |
+
"loss": 1.3234,
|
| 3016 |
+
"step": 8460
|
| 3017 |
+
},
|
| 3018 |
+
{
|
| 3019 |
+
"epoch": 5.038621509209745,
|
| 3020 |
+
"grad_norm": 0.9974661469459534,
|
| 3021 |
+
"learning_rate": 1.7869718309859155e-05,
|
| 3022 |
+
"loss": 0.8732,
|
| 3023 |
+
"step": 8480
|
| 3024 |
+
},
|
| 3025 |
+
{
|
| 3026 |
+
"epoch": 5.05050505050505,
|
| 3027 |
+
"grad_norm": 8.552937507629395,
|
| 3028 |
+
"learning_rate": 1.7649647887323946e-05,
|
| 3029 |
+
"loss": 1.3454,
|
| 3030 |
+
"step": 8500
|
| 3031 |
+
},
|
| 3032 |
+
{
|
| 3033 |
+
"epoch": 5.062388591800357,
|
| 3034 |
+
"grad_norm": 1.974118709564209,
|
| 3035 |
+
"learning_rate": 1.7440580985915494e-05,
|
| 3036 |
+
"loss": 1.6707,
|
| 3037 |
+
"step": 8520
|
| 3038 |
+
},
|
| 3039 |
+
{
|
| 3040 |
+
"epoch": 5.074272133095662,
|
| 3041 |
+
"grad_norm": 3.7020468711853027,
|
| 3042 |
+
"learning_rate": 1.722051056338028e-05,
|
| 3043 |
+
"loss": 1.6472,
|
| 3044 |
+
"step": 8540
|
| 3045 |
+
},
|
| 3046 |
+
{
|
| 3047 |
+
"epoch": 5.086155674390969,
|
| 3048 |
+
"grad_norm": 19.44769287109375,
|
| 3049 |
+
"learning_rate": 1.7000440140845072e-05,
|
| 3050 |
+
"loss": 1.2964,
|
| 3051 |
+
"step": 8560
|
| 3052 |
+
},
|
| 3053 |
+
{
|
| 3054 |
+
"epoch": 5.098039215686274,
|
| 3055 |
+
"grad_norm": 9.980234146118164,
|
| 3056 |
+
"learning_rate": 1.678036971830986e-05,
|
| 3057 |
+
"loss": 1.2214,
|
| 3058 |
+
"step": 8580
|
| 3059 |
+
},
|
| 3060 |
+
{
|
| 3061 |
+
"epoch": 5.109922756981581,
|
| 3062 |
+
"grad_norm": 12.875494003295898,
|
| 3063 |
+
"learning_rate": 1.6560299295774647e-05,
|
| 3064 |
+
"loss": 1.2813,
|
| 3065 |
+
"step": 8600
|
| 3066 |
+
},
|
| 3067 |
+
{
|
| 3068 |
+
"epoch": 5.121806298276886,
|
| 3069 |
+
"grad_norm": 10.326569557189941,
|
| 3070 |
+
"learning_rate": 1.6340228873239438e-05,
|
| 3071 |
+
"loss": 1.6121,
|
| 3072 |
+
"step": 8620
|
| 3073 |
+
},
|
| 3074 |
+
{
|
| 3075 |
+
"epoch": 5.133689839572193,
|
| 3076 |
+
"grad_norm": 4.770064353942871,
|
| 3077 |
+
"learning_rate": 1.6120158450704226e-05,
|
| 3078 |
+
"loss": 1.3623,
|
| 3079 |
+
"step": 8640
|
| 3080 |
+
},
|
| 3081 |
+
{
|
| 3082 |
+
"epoch": 5.145573380867498,
|
| 3083 |
+
"grad_norm": 7.65989351272583,
|
| 3084 |
+
"learning_rate": 1.5911091549295777e-05,
|
| 3085 |
+
"loss": 1.4138,
|
| 3086 |
+
"step": 8660
|
| 3087 |
+
},
|
| 3088 |
+
{
|
| 3089 |
+
"epoch": 5.157456922162805,
|
| 3090 |
+
"grad_norm": 15.983546257019043,
|
| 3091 |
+
"learning_rate": 1.5691021126760564e-05,
|
| 3092 |
+
"loss": 1.9227,
|
| 3093 |
+
"step": 8680
|
| 3094 |
+
},
|
| 3095 |
+
{
|
| 3096 |
+
"epoch": 5.16934046345811,
|
| 3097 |
+
"grad_norm": 12.394526481628418,
|
| 3098 |
+
"learning_rate": 1.5470950704225355e-05,
|
| 3099 |
+
"loss": 1.3912,
|
| 3100 |
+
"step": 8700
|
| 3101 |
+
},
|
| 3102 |
+
{
|
| 3103 |
+
"epoch": 5.181224004753417,
|
| 3104 |
+
"grad_norm": 9.449789047241211,
|
| 3105 |
+
"learning_rate": 1.5250880281690141e-05,
|
| 3106 |
+
"loss": 1.0565,
|
| 3107 |
+
"step": 8720
|
| 3108 |
+
},
|
| 3109 |
+
{
|
| 3110 |
+
"epoch": 5.193107546048722,
|
| 3111 |
+
"grad_norm": 2.6865243911743164,
|
| 3112 |
+
"learning_rate": 1.5030809859154929e-05,
|
| 3113 |
+
"loss": 1.3442,
|
| 3114 |
+
"step": 8740
|
| 3115 |
+
},
|
| 3116 |
+
{
|
| 3117 |
+
"epoch": 5.204991087344029,
|
| 3118 |
+
"grad_norm": 4.256808280944824,
|
| 3119 |
+
"learning_rate": 1.481073943661972e-05,
|
| 3120 |
+
"loss": 1.1088,
|
| 3121 |
+
"step": 8760
|
| 3122 |
+
},
|
| 3123 |
+
{
|
| 3124 |
+
"epoch": 5.216874628639334,
|
| 3125 |
+
"grad_norm": 25.540273666381836,
|
| 3126 |
+
"learning_rate": 1.4590669014084507e-05,
|
| 3127 |
+
"loss": 1.4719,
|
| 3128 |
+
"step": 8780
|
| 3129 |
+
},
|
| 3130 |
+
{
|
| 3131 |
+
"epoch": 5.228758169934641,
|
| 3132 |
+
"grad_norm": 16.564638137817383,
|
| 3133 |
+
"learning_rate": 1.4370598591549298e-05,
|
| 3134 |
+
"loss": 1.5104,
|
| 3135 |
+
"step": 8800
|
| 3136 |
+
},
|
| 3137 |
+
{
|
| 3138 |
+
"epoch": 5.240641711229946,
|
| 3139 |
+
"grad_norm": 16.423961639404297,
|
| 3140 |
+
"learning_rate": 1.4150528169014086e-05,
|
| 3141 |
+
"loss": 1.324,
|
| 3142 |
+
"step": 8820
|
| 3143 |
+
},
|
| 3144 |
+
{
|
| 3145 |
+
"epoch": 5.252525252525253,
|
| 3146 |
+
"grad_norm": 7.4870805740356445,
|
| 3147 |
+
"learning_rate": 1.3930457746478873e-05,
|
| 3148 |
+
"loss": 1.4307,
|
| 3149 |
+
"step": 8840
|
| 3150 |
+
},
|
| 3151 |
+
{
|
| 3152 |
+
"epoch": 5.264408793820558,
|
| 3153 |
+
"grad_norm": 1.6392219066619873,
|
| 3154 |
+
"learning_rate": 1.3710387323943662e-05,
|
| 3155 |
+
"loss": 1.4722,
|
| 3156 |
+
"step": 8860
|
| 3157 |
+
},
|
| 3158 |
+
{
|
| 3159 |
+
"epoch": 5.276292335115865,
|
| 3160 |
+
"grad_norm": 10.110002517700195,
|
| 3161 |
+
"learning_rate": 1.349031690140845e-05,
|
| 3162 |
+
"loss": 1.9203,
|
| 3163 |
+
"step": 8880
|
| 3164 |
+
},
|
| 3165 |
+
{
|
| 3166 |
+
"epoch": 5.28817587641117,
|
| 3167 |
+
"grad_norm": 0.26278719305992126,
|
| 3168 |
+
"learning_rate": 1.327024647887324e-05,
|
| 3169 |
+
"loss": 1.4466,
|
| 3170 |
+
"step": 8900
|
| 3171 |
+
},
|
| 3172 |
+
{
|
| 3173 |
+
"epoch": 5.300059417706477,
|
| 3174 |
+
"grad_norm": 13.663880348205566,
|
| 3175 |
+
"learning_rate": 1.3050176056338028e-05,
|
| 3176 |
+
"loss": 1.1327,
|
| 3177 |
+
"step": 8920
|
| 3178 |
+
},
|
| 3179 |
+
{
|
| 3180 |
+
"epoch": 5.311942959001782,
|
| 3181 |
+
"grad_norm": 17.244321823120117,
|
| 3182 |
+
"learning_rate": 1.2830105633802816e-05,
|
| 3183 |
+
"loss": 1.1933,
|
| 3184 |
+
"step": 8940
|
| 3185 |
+
},
|
| 3186 |
+
{
|
| 3187 |
+
"epoch": 5.323826500297089,
|
| 3188 |
+
"grad_norm": 6.0645341873168945,
|
| 3189 |
+
"learning_rate": 1.2610035211267607e-05,
|
| 3190 |
+
"loss": 1.1798,
|
| 3191 |
+
"step": 8960
|
| 3192 |
+
},
|
| 3193 |
+
{
|
| 3194 |
+
"epoch": 5.335710041592394,
|
| 3195 |
+
"grad_norm": 2.7921664714813232,
|
| 3196 |
+
"learning_rate": 1.2389964788732394e-05,
|
| 3197 |
+
"loss": 1.1191,
|
| 3198 |
+
"step": 8980
|
| 3199 |
+
},
|
| 3200 |
+
{
|
| 3201 |
+
"epoch": 5.347593582887701,
|
| 3202 |
+
"grad_norm": 3.2458791732788086,
|
| 3203 |
+
"learning_rate": 1.2169894366197184e-05,
|
| 3204 |
+
"loss": 1.2999,
|
| 3205 |
+
"step": 9000
|
| 3206 |
+
},
|
| 3207 |
+
{
|
| 3208 |
+
"epoch": 5.359477124183006,
|
| 3209 |
+
"grad_norm": 0.3115676939487457,
|
| 3210 |
+
"learning_rate": 1.1949823943661973e-05,
|
| 3211 |
+
"loss": 1.4403,
|
| 3212 |
+
"step": 9020
|
| 3213 |
+
},
|
| 3214 |
+
{
|
| 3215 |
+
"epoch": 5.371360665478313,
|
| 3216 |
+
"grad_norm": 17.281970977783203,
|
| 3217 |
+
"learning_rate": 1.1729753521126762e-05,
|
| 3218 |
+
"loss": 1.4465,
|
| 3219 |
+
"step": 9040
|
| 3220 |
+
},
|
| 3221 |
+
{
|
| 3222 |
+
"epoch": 5.383244206773618,
|
| 3223 |
+
"grad_norm": 5.0808539390563965,
|
| 3224 |
+
"learning_rate": 1.1509683098591551e-05,
|
| 3225 |
+
"loss": 1.2003,
|
| 3226 |
+
"step": 9060
|
| 3227 |
+
},
|
| 3228 |
+
{
|
| 3229 |
+
"epoch": 5.395127748068925,
|
| 3230 |
+
"grad_norm": 0.5730569958686829,
|
| 3231 |
+
"learning_rate": 1.1289612676056339e-05,
|
| 3232 |
+
"loss": 1.3958,
|
| 3233 |
+
"step": 9080
|
| 3234 |
+
},
|
| 3235 |
+
{
|
| 3236 |
+
"epoch": 5.40701128936423,
|
| 3237 |
+
"grad_norm": 5.8343424797058105,
|
| 3238 |
+
"learning_rate": 1.1069542253521126e-05,
|
| 3239 |
+
"loss": 1.4439,
|
| 3240 |
+
"step": 9100
|
| 3241 |
+
},
|
| 3242 |
+
{
|
| 3243 |
+
"epoch": 5.418894830659537,
|
| 3244 |
+
"grad_norm": 14.066546440124512,
|
| 3245 |
+
"learning_rate": 1.0849471830985916e-05,
|
| 3246 |
+
"loss": 1.4107,
|
| 3247 |
+
"step": 9120
|
| 3248 |
+
},
|
| 3249 |
+
{
|
| 3250 |
+
"epoch": 5.430778371954842,
|
| 3251 |
+
"grad_norm": 2.0192813873291016,
|
| 3252 |
+
"learning_rate": 1.0629401408450705e-05,
|
| 3253 |
+
"loss": 1.2033,
|
| 3254 |
+
"step": 9140
|
| 3255 |
+
},
|
| 3256 |
+
{
|
| 3257 |
+
"epoch": 5.442661913250149,
|
| 3258 |
+
"grad_norm": 24.278059005737305,
|
| 3259 |
+
"learning_rate": 1.0409330985915492e-05,
|
| 3260 |
+
"loss": 1.4415,
|
| 3261 |
+
"step": 9160
|
| 3262 |
+
},
|
| 3263 |
+
{
|
| 3264 |
+
"epoch": 5.454545454545454,
|
| 3265 |
+
"grad_norm": 45.362178802490234,
|
| 3266 |
+
"learning_rate": 1.0189260563380282e-05,
|
| 3267 |
+
"loss": 1.6362,
|
| 3268 |
+
"step": 9180
|
| 3269 |
+
},
|
| 3270 |
+
{
|
| 3271 |
+
"epoch": 5.466428995840761,
|
| 3272 |
+
"grad_norm": 0.005078072659671307,
|
| 3273 |
+
"learning_rate": 9.96919014084507e-06,
|
| 3274 |
+
"loss": 1.4426,
|
| 3275 |
+
"step": 9200
|
| 3276 |
+
},
|
| 3277 |
+
{
|
| 3278 |
+
"epoch": 5.478312537136066,
|
| 3279 |
+
"grad_norm": 1.114734172821045,
|
| 3280 |
+
"learning_rate": 9.74911971830986e-06,
|
| 3281 |
+
"loss": 1.2417,
|
| 3282 |
+
"step": 9220
|
| 3283 |
+
},
|
| 3284 |
+
{
|
| 3285 |
+
"epoch": 5.490196078431373,
|
| 3286 |
+
"grad_norm": 5.923354625701904,
|
| 3287 |
+
"learning_rate": 9.52904929577465e-06,
|
| 3288 |
+
"loss": 1.2631,
|
| 3289 |
+
"step": 9240
|
| 3290 |
+
},
|
| 3291 |
+
{
|
| 3292 |
+
"epoch": 5.502079619726679,
|
| 3293 |
+
"grad_norm": 19.652738571166992,
|
| 3294 |
+
"learning_rate": 9.308978873239437e-06,
|
| 3295 |
+
"loss": 0.9863,
|
| 3296 |
+
"step": 9260
|
| 3297 |
+
},
|
| 3298 |
+
{
|
| 3299 |
+
"epoch": 5.513963161021985,
|
| 3300 |
+
"grad_norm": 7.192227363586426,
|
| 3301 |
+
"learning_rate": 9.088908450704226e-06,
|
| 3302 |
+
"loss": 0.7249,
|
| 3303 |
+
"step": 9280
|
| 3304 |
+
},
|
| 3305 |
+
{
|
| 3306 |
+
"epoch": 5.52584670231729,
|
| 3307 |
+
"grad_norm": 0.43900468945503235,
|
| 3308 |
+
"learning_rate": 8.868838028169015e-06,
|
| 3309 |
+
"loss": 1.2102,
|
| 3310 |
+
"step": 9300
|
| 3311 |
+
},
|
| 3312 |
+
{
|
| 3313 |
+
"epoch": 5.537730243612597,
|
| 3314 |
+
"grad_norm": 4.792228698730469,
|
| 3315 |
+
"learning_rate": 8.648767605633803e-06,
|
| 3316 |
+
"loss": 1.4583,
|
| 3317 |
+
"step": 9320
|
| 3318 |
+
},
|
| 3319 |
+
{
|
| 3320 |
+
"epoch": 5.549613784907903,
|
| 3321 |
+
"grad_norm": 4.418882846832275,
|
| 3322 |
+
"learning_rate": 8.428697183098592e-06,
|
| 3323 |
+
"loss": 1.427,
|
| 3324 |
+
"step": 9340
|
| 3325 |
+
},
|
| 3326 |
+
{
|
| 3327 |
+
"epoch": 5.561497326203209,
|
| 3328 |
+
"grad_norm": 6.352531909942627,
|
| 3329 |
+
"learning_rate": 8.20862676056338e-06,
|
| 3330 |
+
"loss": 1.4876,
|
| 3331 |
+
"step": 9360
|
| 3332 |
+
},
|
| 3333 |
+
{
|
| 3334 |
+
"epoch": 5.573380867498514,
|
| 3335 |
+
"grad_norm": 5.210249900817871,
|
| 3336 |
+
"learning_rate": 7.988556338028169e-06,
|
| 3337 |
+
"loss": 0.7589,
|
| 3338 |
+
"step": 9380
|
| 3339 |
+
},
|
| 3340 |
+
{
|
| 3341 |
+
"epoch": 5.585264408793821,
|
| 3342 |
+
"grad_norm": 5.8901543617248535,
|
| 3343 |
+
"learning_rate": 7.768485915492958e-06,
|
| 3344 |
+
"loss": 0.8805,
|
| 3345 |
+
"step": 9400
|
| 3346 |
+
},
|
| 3347 |
+
{
|
| 3348 |
+
"epoch": 5.597147950089127,
|
| 3349 |
+
"grad_norm": 24.47085189819336,
|
| 3350 |
+
"learning_rate": 7.548415492957747e-06,
|
| 3351 |
+
"loss": 1.5747,
|
| 3352 |
+
"step": 9420
|
| 3353 |
+
},
|
| 3354 |
+
{
|
| 3355 |
+
"epoch": 5.609031491384433,
|
| 3356 |
+
"grad_norm": 16.86648178100586,
|
| 3357 |
+
"learning_rate": 7.3283450704225365e-06,
|
| 3358 |
+
"loss": 1.3989,
|
| 3359 |
+
"step": 9440
|
| 3360 |
+
},
|
| 3361 |
+
{
|
| 3362 |
+
"epoch": 5.620915032679738,
|
| 3363 |
+
"grad_norm": 6.325050354003906,
|
| 3364 |
+
"learning_rate": 7.108274647887324e-06,
|
| 3365 |
+
"loss": 1.2151,
|
| 3366 |
+
"step": 9460
|
| 3367 |
+
},
|
| 3368 |
+
{
|
| 3369 |
+
"epoch": 5.632798573975045,
|
| 3370 |
+
"grad_norm": 8.903736114501953,
|
| 3371 |
+
"learning_rate": 6.888204225352113e-06,
|
| 3372 |
+
"loss": 1.2345,
|
| 3373 |
+
"step": 9480
|
| 3374 |
+
},
|
| 3375 |
+
{
|
| 3376 |
+
"epoch": 5.644682115270351,
|
| 3377 |
+
"grad_norm": 9.477288246154785,
|
| 3378 |
+
"learning_rate": 6.668133802816902e-06,
|
| 3379 |
+
"loss": 1.6694,
|
| 3380 |
+
"step": 9500
|
| 3381 |
+
},
|
| 3382 |
+
{
|
| 3383 |
+
"epoch": 5.656565656565657,
|
| 3384 |
+
"grad_norm": 79.3915023803711,
|
| 3385 |
+
"learning_rate": 6.448063380281691e-06,
|
| 3386 |
+
"loss": 1.4245,
|
| 3387 |
+
"step": 9520
|
| 3388 |
+
},
|
| 3389 |
+
{
|
| 3390 |
+
"epoch": 5.668449197860962,
|
| 3391 |
+
"grad_norm": 12.133055686950684,
|
| 3392 |
+
"learning_rate": 6.227992957746479e-06,
|
| 3393 |
+
"loss": 1.2977,
|
| 3394 |
+
"step": 9540
|
| 3395 |
+
},
|
| 3396 |
+
{
|
| 3397 |
+
"epoch": 5.680332739156269,
|
| 3398 |
+
"grad_norm": 4.151798248291016,
|
| 3399 |
+
"learning_rate": 6.0079225352112685e-06,
|
| 3400 |
+
"loss": 1.2221,
|
| 3401 |
+
"step": 9560
|
| 3402 |
+
},
|
| 3403 |
+
{
|
| 3404 |
+
"epoch": 5.692216280451575,
|
| 3405 |
+
"grad_norm": 4.959856033325195,
|
| 3406 |
+
"learning_rate": 5.787852112676057e-06,
|
| 3407 |
+
"loss": 1.1806,
|
| 3408 |
+
"step": 9580
|
| 3409 |
+
},
|
| 3410 |
+
{
|
| 3411 |
+
"epoch": 5.704099821746881,
|
| 3412 |
+
"grad_norm": 5.8293633460998535,
|
| 3413 |
+
"learning_rate": 5.567781690140845e-06,
|
| 3414 |
+
"loss": 1.8405,
|
| 3415 |
+
"step": 9600
|
| 3416 |
+
},
|
| 3417 |
+
{
|
| 3418 |
+
"epoch": 5.715983363042186,
|
| 3419 |
+
"grad_norm": 0.5208277106285095,
|
| 3420 |
+
"learning_rate": 5.347711267605634e-06,
|
| 3421 |
+
"loss": 1.0997,
|
| 3422 |
+
"step": 9620
|
| 3423 |
+
},
|
| 3424 |
+
{
|
| 3425 |
+
"epoch": 5.727866904337493,
|
| 3426 |
+
"grad_norm": 0.5993483066558838,
|
| 3427 |
+
"learning_rate": 5.127640845070423e-06,
|
| 3428 |
+
"loss": 1.3571,
|
| 3429 |
+
"step": 9640
|
| 3430 |
+
},
|
| 3431 |
+
{
|
| 3432 |
+
"epoch": 5.739750445632799,
|
| 3433 |
+
"grad_norm": 6.959007263183594,
|
| 3434 |
+
"learning_rate": 4.907570422535211e-06,
|
| 3435 |
+
"loss": 1.5781,
|
| 3436 |
+
"step": 9660
|
| 3437 |
+
},
|
| 3438 |
+
{
|
| 3439 |
+
"epoch": 5.751633986928105,
|
| 3440 |
+
"grad_norm": 11.25479507446289,
|
| 3441 |
+
"learning_rate": 4.6875000000000004e-06,
|
| 3442 |
+
"loss": 1.4159,
|
| 3443 |
+
"step": 9680
|
| 3444 |
+
},
|
| 3445 |
+
{
|
| 3446 |
+
"epoch": 5.76351752822341,
|
| 3447 |
+
"grad_norm": 24.656896591186523,
|
| 3448 |
+
"learning_rate": 4.467429577464789e-06,
|
| 3449 |
+
"loss": 0.9139,
|
| 3450 |
+
"step": 9700
|
| 3451 |
+
},
|
| 3452 |
+
{
|
| 3453 |
+
"epoch": 5.775401069518717,
|
| 3454 |
+
"grad_norm": 10.791423797607422,
|
| 3455 |
+
"learning_rate": 4.247359154929577e-06,
|
| 3456 |
+
"loss": 1.4309,
|
| 3457 |
+
"step": 9720
|
| 3458 |
+
},
|
| 3459 |
+
{
|
| 3460 |
+
"epoch": 5.787284610814023,
|
| 3461 |
+
"grad_norm": 5.377380847930908,
|
| 3462 |
+
"learning_rate": 4.0272887323943664e-06,
|
| 3463 |
+
"loss": 1.6518,
|
| 3464 |
+
"step": 9740
|
| 3465 |
+
},
|
| 3466 |
+
{
|
| 3467 |
+
"epoch": 5.799168152109329,
|
| 3468 |
+
"grad_norm": 17.60517120361328,
|
| 3469 |
+
"learning_rate": 3.807218309859155e-06,
|
| 3470 |
+
"loss": 1.4515,
|
| 3471 |
+
"step": 9760
|
| 3472 |
+
},
|
| 3473 |
+
{
|
| 3474 |
+
"epoch": 5.811051693404634,
|
| 3475 |
+
"grad_norm": 10.215800285339355,
|
| 3476 |
+
"learning_rate": 3.587147887323944e-06,
|
| 3477 |
+
"loss": 0.8738,
|
| 3478 |
+
"step": 9780
|
| 3479 |
+
},
|
| 3480 |
+
{
|
| 3481 |
+
"epoch": 5.822935234699941,
|
| 3482 |
+
"grad_norm": 19.97536277770996,
|
| 3483 |
+
"learning_rate": 3.3670774647887324e-06,
|
| 3484 |
+
"loss": 1.5413,
|
| 3485 |
+
"step": 9800
|
| 3486 |
+
},
|
| 3487 |
+
{
|
| 3488 |
+
"epoch": 5.834818775995247,
|
| 3489 |
+
"grad_norm": 6.600990295410156,
|
| 3490 |
+
"learning_rate": 3.1470070422535212e-06,
|
| 3491 |
+
"loss": 1.3498,
|
| 3492 |
+
"step": 9820
|
| 3493 |
+
},
|
| 3494 |
+
{
|
| 3495 |
+
"epoch": 5.846702317290553,
|
| 3496 |
+
"grad_norm": 6.998388767242432,
|
| 3497 |
+
"learning_rate": 2.92693661971831e-06,
|
| 3498 |
+
"loss": 1.1162,
|
| 3499 |
+
"step": 9840
|
| 3500 |
+
},
|
| 3501 |
+
{
|
| 3502 |
+
"epoch": 5.858585858585858,
|
| 3503 |
+
"grad_norm": 9.804807662963867,
|
| 3504 |
+
"learning_rate": 2.7068661971830984e-06,
|
| 3505 |
+
"loss": 1.5677,
|
| 3506 |
+
"step": 9860
|
| 3507 |
+
},
|
| 3508 |
+
{
|
| 3509 |
+
"epoch": 5.870469399881165,
|
| 3510 |
+
"grad_norm": 7.46671724319458,
|
| 3511 |
+
"learning_rate": 2.4867957746478876e-06,
|
| 3512 |
+
"loss": 1.1665,
|
| 3513 |
+
"step": 9880
|
| 3514 |
+
},
|
| 3515 |
+
{
|
| 3516 |
+
"epoch": 5.882352941176471,
|
| 3517 |
+
"grad_norm": 10.575834274291992,
|
| 3518 |
+
"learning_rate": 2.2667253521126764e-06,
|
| 3519 |
+
"loss": 1.2868,
|
| 3520 |
+
"step": 9900
|
| 3521 |
+
},
|
| 3522 |
+
{
|
| 3523 |
+
"epoch": 5.894236482471777,
|
| 3524 |
+
"grad_norm": 15.221187591552734,
|
| 3525 |
+
"learning_rate": 2.046654929577465e-06,
|
| 3526 |
+
"loss": 1.1183,
|
| 3527 |
+
"step": 9920
|
| 3528 |
+
},
|
| 3529 |
+
{
|
| 3530 |
+
"epoch": 5.906120023767082,
|
| 3531 |
+
"grad_norm": 6.444254398345947,
|
| 3532 |
+
"learning_rate": 1.8265845070422536e-06,
|
| 3533 |
+
"loss": 1.6049,
|
| 3534 |
+
"step": 9940
|
| 3535 |
+
},
|
| 3536 |
+
{
|
| 3537 |
+
"epoch": 5.918003565062389,
|
| 3538 |
+
"grad_norm": 5.689063549041748,
|
| 3539 |
+
"learning_rate": 1.6065140845070422e-06,
|
| 3540 |
+
"loss": 1.0063,
|
| 3541 |
+
"step": 9960
|
| 3542 |
+
},
|
| 3543 |
+
{
|
| 3544 |
+
"epoch": 5.929887106357695,
|
| 3545 |
+
"grad_norm": 7.111664295196533,
|
| 3546 |
+
"learning_rate": 1.386443661971831e-06,
|
| 3547 |
+
"loss": 0.9414,
|
| 3548 |
+
"step": 9980
|
| 3549 |
+
},
|
| 3550 |
+
{
|
| 3551 |
+
"epoch": 5.941770647653001,
|
| 3552 |
+
"grad_norm": 37.21208953857422,
|
| 3553 |
+
"learning_rate": 1.1663732394366198e-06,
|
| 3554 |
+
"loss": 1.4099,
|
| 3555 |
+
"step": 10000
|
| 3556 |
+
},
|
| 3557 |
+
{
|
| 3558 |
+
"epoch": 5.953654188948306,
|
| 3559 |
+
"grad_norm": 3.9379427433013916,
|
| 3560 |
+
"learning_rate": 9.463028169014084e-07,
|
| 3561 |
+
"loss": 1.1238,
|
| 3562 |
+
"step": 10020
|
| 3563 |
+
},
|
| 3564 |
+
{
|
| 3565 |
+
"epoch": 5.965537730243613,
|
| 3566 |
+
"grad_norm": 2.1166274547576904,
|
| 3567 |
+
"learning_rate": 7.262323943661972e-07,
|
| 3568 |
+
"loss": 1.2788,
|
| 3569 |
+
"step": 10040
|
| 3570 |
+
},
|
| 3571 |
+
{
|
| 3572 |
+
"epoch": 5.977421271538919,
|
| 3573 |
+
"grad_norm": 7.178572654724121,
|
| 3574 |
+
"learning_rate": 5.061619718309859e-07,
|
| 3575 |
+
"loss": 1.0986,
|
| 3576 |
+
"step": 10060
|
| 3577 |
+
},
|
| 3578 |
+
{
|
| 3579 |
+
"epoch": 5.989304812834225,
|
| 3580 |
+
"grad_norm": 0.30666860938072205,
|
| 3581 |
+
"learning_rate": 2.860915492957747e-07,
|
| 3582 |
+
"loss": 1.3072,
|
| 3583 |
+
"step": 10080
|
| 3584 |
+
},
|
| 3585 |
+
{
|
| 3586 |
+
"epoch": 6.0,
|
| 3587 |
+
"eval_accuracy": 0.4074074074074074,
|
| 3588 |
+
"eval_loss": 2.2214395999908447,
|
| 3589 |
+
"eval_runtime": 105.6981,
|
| 3590 |
+
"eval_samples_per_second": 2.81,
|
| 3591 |
+
"eval_steps_per_second": 2.81,
|
| 3592 |
+
"step": 10098
|
| 3593 |
}
|
| 3594 |
],
|
| 3595 |
"logging_steps": 20,
|
| 3596 |
+
"max_steps": 10098,
|
| 3597 |
"num_input_tokens_seen": 0,
|
| 3598 |
+
"num_train_epochs": 6,
|
| 3599 |
"save_steps": 500,
|
| 3600 |
"stateful_callbacks": {
|
| 3601 |
"TrainerControl": {
|
|
|
|
| 3609 |
"attributes": {}
|
| 3610 |
}
|
| 3611 |
},
|
| 3612 |
+
"total_flos": 2.5433536104702945e+18,
|
| 3613 |
"train_batch_size": 1,
|
| 3614 |
"trial_name": null,
|
| 3615 |
"trial_params": null
|
training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 5368
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a06cc0c81e74876caeafbe898f65be97fa973b1744e977206810876d496bb4da
|
| 3 |
size 5368
|