Training in progress, epoch 9, checkpoint
Browse files
last-checkpoint/adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 111475752
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0d51ecd70bd6abe98049e3903569c3e9e0c225f8ca47ca6925715bbb4525a07f
|
| 3 |
size 111475752
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 223212738
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b1015cd357231c8969ba9b399558b4e4e768018a6be01973d55dd3f1a4a7117d
|
| 3 |
size 223212738
|
last-checkpoint/rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14244
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0a9f98afb0211ede3e469085bcd7ffb48239431e6596713b1e711384e5b986da
|
| 3 |
size 14244
|
last-checkpoint/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 sha256:709b689cdaa470cfc575e643ff39b4e6938bba001e0031274a8872524520efb6
|
| 3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": 40.896302250803856,
|
| 3 |
"best_model_checkpoint": "./exp/whisper-large-v3-turbo-zh-TW-clean-1/checkpoint-3397",
|
| 4 |
-
"epoch":
|
| 5 |
"eval_steps": 500,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -23896,6 +23896,2617 @@
|
|
| 23896 |
"eval_steps_per_second": 3.762,
|
| 23897 |
"eval_wer": 40.896302250803856,
|
| 23898 |
"step": 3397
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 23899 |
}
|
| 23900 |
],
|
| 23901 |
"logging_steps": 1,
|
|
@@ -23910,12 +26521,12 @@
|
|
| 23910 |
"should_evaluate": false,
|
| 23911 |
"should_log": false,
|
| 23912 |
"should_save": true,
|
| 23913 |
-
"should_training_stop":
|
| 23914 |
},
|
| 23915 |
"attributes": {}
|
| 23916 |
}
|
| 23917 |
},
|
| 23918 |
-
"total_flos":
|
| 23919 |
"train_batch_size": 4,
|
| 23920 |
"trial_name": null,
|
| 23921 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": 40.896302250803856,
|
| 3 |
"best_model_checkpoint": "./exp/whisper-large-v3-turbo-zh-TW-clean-1/checkpoint-3397",
|
| 4 |
+
"epoch": 9.986754966887418,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 3770,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 23896 |
"eval_steps_per_second": 3.762,
|
| 23897 |
"eval_wer": 40.896302250803856,
|
| 23898 |
"step": 3397
|
| 23899 |
+
},
|
| 23900 |
+
{
|
| 23901 |
+
"epoch": 9.001324503311258,
|
| 23902 |
+
"grad_norm": 1.3489973545074463,
|
| 23903 |
+
"learning_rate": 4.93368700265252e-05,
|
| 23904 |
+
"loss": 0.0367,
|
| 23905 |
+
"step": 3398
|
| 23906 |
+
},
|
| 23907 |
+
{
|
| 23908 |
+
"epoch": 9.003973509933775,
|
| 23909 |
+
"grad_norm": 0.38989582657814026,
|
| 23910 |
+
"learning_rate": 4.9204244031830244e-05,
|
| 23911 |
+
"loss": 0.0104,
|
| 23912 |
+
"step": 3399
|
| 23913 |
+
},
|
| 23914 |
+
{
|
| 23915 |
+
"epoch": 9.006622516556291,
|
| 23916 |
+
"grad_norm": 0.1691805124282837,
|
| 23917 |
+
"learning_rate": 4.907161803713528e-05,
|
| 23918 |
+
"loss": 0.0097,
|
| 23919 |
+
"step": 3400
|
| 23920 |
+
},
|
| 23921 |
+
{
|
| 23922 |
+
"epoch": 9.009271523178809,
|
| 23923 |
+
"grad_norm": 0.10428066551685333,
|
| 23924 |
+
"learning_rate": 4.893899204244032e-05,
|
| 23925 |
+
"loss": 0.0078,
|
| 23926 |
+
"step": 3401
|
| 23927 |
+
},
|
| 23928 |
+
{
|
| 23929 |
+
"epoch": 9.011920529801325,
|
| 23930 |
+
"grad_norm": 0.17672701179981232,
|
| 23931 |
+
"learning_rate": 4.880636604774536e-05,
|
| 23932 |
+
"loss": 0.008,
|
| 23933 |
+
"step": 3402
|
| 23934 |
+
},
|
| 23935 |
+
{
|
| 23936 |
+
"epoch": 9.01456953642384,
|
| 23937 |
+
"grad_norm": 0.31466829776763916,
|
| 23938 |
+
"learning_rate": 4.8673740053050396e-05,
|
| 23939 |
+
"loss": 0.0098,
|
| 23940 |
+
"step": 3403
|
| 23941 |
+
},
|
| 23942 |
+
{
|
| 23943 |
+
"epoch": 9.017218543046358,
|
| 23944 |
+
"grad_norm": 0.223367378115654,
|
| 23945 |
+
"learning_rate": 4.8541114058355434e-05,
|
| 23946 |
+
"loss": 0.0088,
|
| 23947 |
+
"step": 3404
|
| 23948 |
+
},
|
| 23949 |
+
{
|
| 23950 |
+
"epoch": 9.019867549668874,
|
| 23951 |
+
"grad_norm": 0.4342547357082367,
|
| 23952 |
+
"learning_rate": 4.840848806366048e-05,
|
| 23953 |
+
"loss": 0.0139,
|
| 23954 |
+
"step": 3405
|
| 23955 |
+
},
|
| 23956 |
+
{
|
| 23957 |
+
"epoch": 9.022516556291391,
|
| 23958 |
+
"grad_norm": 6.883068561553955,
|
| 23959 |
+
"learning_rate": 4.8275862068965517e-05,
|
| 23960 |
+
"loss": 0.2748,
|
| 23961 |
+
"step": 3406
|
| 23962 |
+
},
|
| 23963 |
+
{
|
| 23964 |
+
"epoch": 9.025165562913907,
|
| 23965 |
+
"grad_norm": 0.10145234316587448,
|
| 23966 |
+
"learning_rate": 4.8143236074270555e-05,
|
| 23967 |
+
"loss": 0.0066,
|
| 23968 |
+
"step": 3407
|
| 23969 |
+
},
|
| 23970 |
+
{
|
| 23971 |
+
"epoch": 9.027814569536424,
|
| 23972 |
+
"grad_norm": 0.10756523162126541,
|
| 23973 |
+
"learning_rate": 4.80106100795756e-05,
|
| 23974 |
+
"loss": 0.0075,
|
| 23975 |
+
"step": 3408
|
| 23976 |
+
},
|
| 23977 |
+
{
|
| 23978 |
+
"epoch": 9.03046357615894,
|
| 23979 |
+
"grad_norm": 0.12273053824901581,
|
| 23980 |
+
"learning_rate": 4.787798408488064e-05,
|
| 23981 |
+
"loss": 0.0067,
|
| 23982 |
+
"step": 3409
|
| 23983 |
+
},
|
| 23984 |
+
{
|
| 23985 |
+
"epoch": 9.033112582781458,
|
| 23986 |
+
"grad_norm": 0.07955110818147659,
|
| 23987 |
+
"learning_rate": 4.7745358090185675e-05,
|
| 23988 |
+
"loss": 0.0049,
|
| 23989 |
+
"step": 3410
|
| 23990 |
+
},
|
| 23991 |
+
{
|
| 23992 |
+
"epoch": 9.035761589403974,
|
| 23993 |
+
"grad_norm": 0.38133883476257324,
|
| 23994 |
+
"learning_rate": 4.761273209549072e-05,
|
| 23995 |
+
"loss": 0.0162,
|
| 23996 |
+
"step": 3411
|
| 23997 |
+
},
|
| 23998 |
+
{
|
| 23999 |
+
"epoch": 9.03841059602649,
|
| 24000 |
+
"grad_norm": 0.2501557171344757,
|
| 24001 |
+
"learning_rate": 4.748010610079576e-05,
|
| 24002 |
+
"loss": 0.0089,
|
| 24003 |
+
"step": 3412
|
| 24004 |
+
},
|
| 24005 |
+
{
|
| 24006 |
+
"epoch": 9.041059602649007,
|
| 24007 |
+
"grad_norm": 0.12085811793804169,
|
| 24008 |
+
"learning_rate": 4.73474801061008e-05,
|
| 24009 |
+
"loss": 0.0064,
|
| 24010 |
+
"step": 3413
|
| 24011 |
+
},
|
| 24012 |
+
{
|
| 24013 |
+
"epoch": 9.043708609271523,
|
| 24014 |
+
"grad_norm": 0.36201950907707214,
|
| 24015 |
+
"learning_rate": 4.721485411140584e-05,
|
| 24016 |
+
"loss": 0.0102,
|
| 24017 |
+
"step": 3414
|
| 24018 |
+
},
|
| 24019 |
+
{
|
| 24020 |
+
"epoch": 9.04635761589404,
|
| 24021 |
+
"grad_norm": 0.09906315058469772,
|
| 24022 |
+
"learning_rate": 4.708222811671087e-05,
|
| 24023 |
+
"loss": 0.0056,
|
| 24024 |
+
"step": 3415
|
| 24025 |
+
},
|
| 24026 |
+
{
|
| 24027 |
+
"epoch": 9.049006622516556,
|
| 24028 |
+
"grad_norm": 0.26838064193725586,
|
| 24029 |
+
"learning_rate": 4.694960212201592e-05,
|
| 24030 |
+
"loss": 0.0104,
|
| 24031 |
+
"step": 3416
|
| 24032 |
+
},
|
| 24033 |
+
{
|
| 24034 |
+
"epoch": 9.051655629139074,
|
| 24035 |
+
"grad_norm": 0.11671532690525055,
|
| 24036 |
+
"learning_rate": 4.6816976127320955e-05,
|
| 24037 |
+
"loss": 0.0071,
|
| 24038 |
+
"step": 3417
|
| 24039 |
+
},
|
| 24040 |
+
{
|
| 24041 |
+
"epoch": 9.05430463576159,
|
| 24042 |
+
"grad_norm": 0.11996608972549438,
|
| 24043 |
+
"learning_rate": 4.668435013262599e-05,
|
| 24044 |
+
"loss": 0.0064,
|
| 24045 |
+
"step": 3418
|
| 24046 |
+
},
|
| 24047 |
+
{
|
| 24048 |
+
"epoch": 9.056953642384107,
|
| 24049 |
+
"grad_norm": 0.37756839394569397,
|
| 24050 |
+
"learning_rate": 4.655172413793104e-05,
|
| 24051 |
+
"loss": 0.0128,
|
| 24052 |
+
"step": 3419
|
| 24053 |
+
},
|
| 24054 |
+
{
|
| 24055 |
+
"epoch": 9.059602649006623,
|
| 24056 |
+
"grad_norm": 0.1943897008895874,
|
| 24057 |
+
"learning_rate": 4.6419098143236075e-05,
|
| 24058 |
+
"loss": 0.0107,
|
| 24059 |
+
"step": 3420
|
| 24060 |
+
},
|
| 24061 |
+
{
|
| 24062 |
+
"epoch": 9.062251655629138,
|
| 24063 |
+
"grad_norm": 0.19497939944267273,
|
| 24064 |
+
"learning_rate": 4.628647214854111e-05,
|
| 24065 |
+
"loss": 0.0092,
|
| 24066 |
+
"step": 3421
|
| 24067 |
+
},
|
| 24068 |
+
{
|
| 24069 |
+
"epoch": 9.064900662251656,
|
| 24070 |
+
"grad_norm": 2.2906646728515625,
|
| 24071 |
+
"learning_rate": 4.615384615384616e-05,
|
| 24072 |
+
"loss": 0.0198,
|
| 24073 |
+
"step": 3422
|
| 24074 |
+
},
|
| 24075 |
+
{
|
| 24076 |
+
"epoch": 9.067549668874172,
|
| 24077 |
+
"grad_norm": 2.465967893600464,
|
| 24078 |
+
"learning_rate": 4.6021220159151196e-05,
|
| 24079 |
+
"loss": 0.0727,
|
| 24080 |
+
"step": 3423
|
| 24081 |
+
},
|
| 24082 |
+
{
|
| 24083 |
+
"epoch": 9.07019867549669,
|
| 24084 |
+
"grad_norm": 0.15241271257400513,
|
| 24085 |
+
"learning_rate": 4.5888594164456234e-05,
|
| 24086 |
+
"loss": 0.0072,
|
| 24087 |
+
"step": 3424
|
| 24088 |
+
},
|
| 24089 |
+
{
|
| 24090 |
+
"epoch": 9.072847682119205,
|
| 24091 |
+
"grad_norm": 0.15346211194992065,
|
| 24092 |
+
"learning_rate": 4.575596816976128e-05,
|
| 24093 |
+
"loss": 0.0088,
|
| 24094 |
+
"step": 3425
|
| 24095 |
+
},
|
| 24096 |
+
{
|
| 24097 |
+
"epoch": 9.075496688741723,
|
| 24098 |
+
"grad_norm": 0.3105780780315399,
|
| 24099 |
+
"learning_rate": 4.562334217506632e-05,
|
| 24100 |
+
"loss": 0.0074,
|
| 24101 |
+
"step": 3426
|
| 24102 |
+
},
|
| 24103 |
+
{
|
| 24104 |
+
"epoch": 9.078145695364238,
|
| 24105 |
+
"grad_norm": 0.28642329573631287,
|
| 24106 |
+
"learning_rate": 4.549071618037135e-05,
|
| 24107 |
+
"loss": 0.0071,
|
| 24108 |
+
"step": 3427
|
| 24109 |
+
},
|
| 24110 |
+
{
|
| 24111 |
+
"epoch": 9.080794701986754,
|
| 24112 |
+
"grad_norm": 0.17047229409217834,
|
| 24113 |
+
"learning_rate": 4.535809018567639e-05,
|
| 24114 |
+
"loss": 0.0074,
|
| 24115 |
+
"step": 3428
|
| 24116 |
+
},
|
| 24117 |
+
{
|
| 24118 |
+
"epoch": 9.083443708609272,
|
| 24119 |
+
"grad_norm": 1.974877953529358,
|
| 24120 |
+
"learning_rate": 4.522546419098143e-05,
|
| 24121 |
+
"loss": 0.1462,
|
| 24122 |
+
"step": 3429
|
| 24123 |
+
},
|
| 24124 |
+
{
|
| 24125 |
+
"epoch": 9.086092715231787,
|
| 24126 |
+
"grad_norm": 0.09645556658506393,
|
| 24127 |
+
"learning_rate": 4.509283819628647e-05,
|
| 24128 |
+
"loss": 0.0061,
|
| 24129 |
+
"step": 3430
|
| 24130 |
+
},
|
| 24131 |
+
{
|
| 24132 |
+
"epoch": 9.088741721854305,
|
| 24133 |
+
"grad_norm": 1.6515982151031494,
|
| 24134 |
+
"learning_rate": 4.4960212201591514e-05,
|
| 24135 |
+
"loss": 0.0304,
|
| 24136 |
+
"step": 3431
|
| 24137 |
+
},
|
| 24138 |
+
{
|
| 24139 |
+
"epoch": 9.09139072847682,
|
| 24140 |
+
"grad_norm": 0.12649770081043243,
|
| 24141 |
+
"learning_rate": 4.482758620689655e-05,
|
| 24142 |
+
"loss": 0.006,
|
| 24143 |
+
"step": 3432
|
| 24144 |
+
},
|
| 24145 |
+
{
|
| 24146 |
+
"epoch": 9.094039735099338,
|
| 24147 |
+
"grad_norm": 0.8409780263900757,
|
| 24148 |
+
"learning_rate": 4.469496021220159e-05,
|
| 24149 |
+
"loss": 0.0139,
|
| 24150 |
+
"step": 3433
|
| 24151 |
+
},
|
| 24152 |
+
{
|
| 24153 |
+
"epoch": 9.096688741721854,
|
| 24154 |
+
"grad_norm": 0.10516829788684845,
|
| 24155 |
+
"learning_rate": 4.4562334217506634e-05,
|
| 24156 |
+
"loss": 0.0068,
|
| 24157 |
+
"step": 3434
|
| 24158 |
+
},
|
| 24159 |
+
{
|
| 24160 |
+
"epoch": 9.099337748344372,
|
| 24161 |
+
"grad_norm": 0.12136232107877731,
|
| 24162 |
+
"learning_rate": 4.442970822281167e-05,
|
| 24163 |
+
"loss": 0.0067,
|
| 24164 |
+
"step": 3435
|
| 24165 |
+
},
|
| 24166 |
+
{
|
| 24167 |
+
"epoch": 9.101986754966887,
|
| 24168 |
+
"grad_norm": 0.1334480494260788,
|
| 24169 |
+
"learning_rate": 4.429708222811671e-05,
|
| 24170 |
+
"loss": 0.0085,
|
| 24171 |
+
"step": 3436
|
| 24172 |
+
},
|
| 24173 |
+
{
|
| 24174 |
+
"epoch": 9.104635761589403,
|
| 24175 |
+
"grad_norm": 0.09830499440431595,
|
| 24176 |
+
"learning_rate": 4.4164456233421755e-05,
|
| 24177 |
+
"loss": 0.0072,
|
| 24178 |
+
"step": 3437
|
| 24179 |
+
},
|
| 24180 |
+
{
|
| 24181 |
+
"epoch": 9.10728476821192,
|
| 24182 |
+
"grad_norm": 0.07183544337749481,
|
| 24183 |
+
"learning_rate": 4.403183023872679e-05,
|
| 24184 |
+
"loss": 0.0059,
|
| 24185 |
+
"step": 3438
|
| 24186 |
+
},
|
| 24187 |
+
{
|
| 24188 |
+
"epoch": 9.109933774834436,
|
| 24189 |
+
"grad_norm": 0.9209287762641907,
|
| 24190 |
+
"learning_rate": 4.389920424403184e-05,
|
| 24191 |
+
"loss": 0.024,
|
| 24192 |
+
"step": 3439
|
| 24193 |
+
},
|
| 24194 |
+
{
|
| 24195 |
+
"epoch": 9.112582781456954,
|
| 24196 |
+
"grad_norm": 0.5843676328659058,
|
| 24197 |
+
"learning_rate": 4.376657824933687e-05,
|
| 24198 |
+
"loss": 0.0088,
|
| 24199 |
+
"step": 3440
|
| 24200 |
+
},
|
| 24201 |
+
{
|
| 24202 |
+
"epoch": 9.11523178807947,
|
| 24203 |
+
"grad_norm": 0.24383290112018585,
|
| 24204 |
+
"learning_rate": 4.363395225464191e-05,
|
| 24205 |
+
"loss": 0.0077,
|
| 24206 |
+
"step": 3441
|
| 24207 |
+
},
|
| 24208 |
+
{
|
| 24209 |
+
"epoch": 9.117880794701987,
|
| 24210 |
+
"grad_norm": 0.3678019344806671,
|
| 24211 |
+
"learning_rate": 4.350132625994695e-05,
|
| 24212 |
+
"loss": 0.0101,
|
| 24213 |
+
"step": 3442
|
| 24214 |
+
},
|
| 24215 |
+
{
|
| 24216 |
+
"epoch": 9.120529801324503,
|
| 24217 |
+
"grad_norm": 0.1334732174873352,
|
| 24218 |
+
"learning_rate": 4.336870026525199e-05,
|
| 24219 |
+
"loss": 0.0056,
|
| 24220 |
+
"step": 3443
|
| 24221 |
+
},
|
| 24222 |
+
{
|
| 24223 |
+
"epoch": 9.12317880794702,
|
| 24224 |
+
"grad_norm": 0.24420872330665588,
|
| 24225 |
+
"learning_rate": 4.323607427055703e-05,
|
| 24226 |
+
"loss": 0.0079,
|
| 24227 |
+
"step": 3444
|
| 24228 |
+
},
|
| 24229 |
+
{
|
| 24230 |
+
"epoch": 9.125827814569536,
|
| 24231 |
+
"grad_norm": 0.2850385308265686,
|
| 24232 |
+
"learning_rate": 4.310344827586207e-05,
|
| 24233 |
+
"loss": 0.0114,
|
| 24234 |
+
"step": 3445
|
| 24235 |
+
},
|
| 24236 |
+
{
|
| 24237 |
+
"epoch": 9.128476821192052,
|
| 24238 |
+
"grad_norm": 0.08591492474079132,
|
| 24239 |
+
"learning_rate": 4.297082228116711e-05,
|
| 24240 |
+
"loss": 0.0054,
|
| 24241 |
+
"step": 3446
|
| 24242 |
+
},
|
| 24243 |
+
{
|
| 24244 |
+
"epoch": 9.13112582781457,
|
| 24245 |
+
"grad_norm": 0.09644327312707901,
|
| 24246 |
+
"learning_rate": 4.283819628647215e-05,
|
| 24247 |
+
"loss": 0.006,
|
| 24248 |
+
"step": 3447
|
| 24249 |
+
},
|
| 24250 |
+
{
|
| 24251 |
+
"epoch": 9.133774834437085,
|
| 24252 |
+
"grad_norm": 0.2626118063926697,
|
| 24253 |
+
"learning_rate": 4.270557029177719e-05,
|
| 24254 |
+
"loss": 0.0091,
|
| 24255 |
+
"step": 3448
|
| 24256 |
+
},
|
| 24257 |
+
{
|
| 24258 |
+
"epoch": 9.136423841059603,
|
| 24259 |
+
"grad_norm": 0.3831780254840851,
|
| 24260 |
+
"learning_rate": 4.257294429708223e-05,
|
| 24261 |
+
"loss": 0.0089,
|
| 24262 |
+
"step": 3449
|
| 24263 |
+
},
|
| 24264 |
+
{
|
| 24265 |
+
"epoch": 9.139072847682119,
|
| 24266 |
+
"grad_norm": 1.390022873878479,
|
| 24267 |
+
"learning_rate": 4.244031830238727e-05,
|
| 24268 |
+
"loss": 0.0714,
|
| 24269 |
+
"step": 3450
|
| 24270 |
+
},
|
| 24271 |
+
{
|
| 24272 |
+
"epoch": 9.141721854304636,
|
| 24273 |
+
"grad_norm": 0.4582221508026123,
|
| 24274 |
+
"learning_rate": 4.2307692307692314e-05,
|
| 24275 |
+
"loss": 0.0113,
|
| 24276 |
+
"step": 3451
|
| 24277 |
+
},
|
| 24278 |
+
{
|
| 24279 |
+
"epoch": 9.144370860927152,
|
| 24280 |
+
"grad_norm": 0.11480309814214706,
|
| 24281 |
+
"learning_rate": 4.2175066312997345e-05,
|
| 24282 |
+
"loss": 0.0073,
|
| 24283 |
+
"step": 3452
|
| 24284 |
+
},
|
| 24285 |
+
{
|
| 24286 |
+
"epoch": 9.14701986754967,
|
| 24287 |
+
"grad_norm": 0.08496708422899246,
|
| 24288 |
+
"learning_rate": 4.204244031830238e-05,
|
| 24289 |
+
"loss": 0.0059,
|
| 24290 |
+
"step": 3453
|
| 24291 |
+
},
|
| 24292 |
+
{
|
| 24293 |
+
"epoch": 9.149668874172185,
|
| 24294 |
+
"grad_norm": 0.09760905802249908,
|
| 24295 |
+
"learning_rate": 4.190981432360743e-05,
|
| 24296 |
+
"loss": 0.0069,
|
| 24297 |
+
"step": 3454
|
| 24298 |
+
},
|
| 24299 |
+
{
|
| 24300 |
+
"epoch": 9.152317880794701,
|
| 24301 |
+
"grad_norm": 0.5061994194984436,
|
| 24302 |
+
"learning_rate": 4.1777188328912466e-05,
|
| 24303 |
+
"loss": 0.0104,
|
| 24304 |
+
"step": 3455
|
| 24305 |
+
},
|
| 24306 |
+
{
|
| 24307 |
+
"epoch": 9.154966887417219,
|
| 24308 |
+
"grad_norm": 0.2912754714488983,
|
| 24309 |
+
"learning_rate": 4.1644562334217504e-05,
|
| 24310 |
+
"loss": 0.0085,
|
| 24311 |
+
"step": 3456
|
| 24312 |
+
},
|
| 24313 |
+
{
|
| 24314 |
+
"epoch": 9.157615894039735,
|
| 24315 |
+
"grad_norm": 0.06553248316049576,
|
| 24316 |
+
"learning_rate": 4.151193633952255e-05,
|
| 24317 |
+
"loss": 0.0049,
|
| 24318 |
+
"step": 3457
|
| 24319 |
+
},
|
| 24320 |
+
{
|
| 24321 |
+
"epoch": 9.160264900662252,
|
| 24322 |
+
"grad_norm": 0.1199113130569458,
|
| 24323 |
+
"learning_rate": 4.1379310344827587e-05,
|
| 24324 |
+
"loss": 0.0071,
|
| 24325 |
+
"step": 3458
|
| 24326 |
+
},
|
| 24327 |
+
{
|
| 24328 |
+
"epoch": 9.162913907284768,
|
| 24329 |
+
"grad_norm": 0.13050176203250885,
|
| 24330 |
+
"learning_rate": 4.1246684350132625e-05,
|
| 24331 |
+
"loss": 0.0065,
|
| 24332 |
+
"step": 3459
|
| 24333 |
+
},
|
| 24334 |
+
{
|
| 24335 |
+
"epoch": 9.165562913907285,
|
| 24336 |
+
"grad_norm": 0.15488766133785248,
|
| 24337 |
+
"learning_rate": 4.111405835543767e-05,
|
| 24338 |
+
"loss": 0.0059,
|
| 24339 |
+
"step": 3460
|
| 24340 |
+
},
|
| 24341 |
+
{
|
| 24342 |
+
"epoch": 9.168211920529801,
|
| 24343 |
+
"grad_norm": 0.1280549168586731,
|
| 24344 |
+
"learning_rate": 4.098143236074271e-05,
|
| 24345 |
+
"loss": 0.0088,
|
| 24346 |
+
"step": 3461
|
| 24347 |
+
},
|
| 24348 |
+
{
|
| 24349 |
+
"epoch": 9.170860927152319,
|
| 24350 |
+
"grad_norm": 0.08911632001399994,
|
| 24351 |
+
"learning_rate": 4.0848806366047745e-05,
|
| 24352 |
+
"loss": 0.006,
|
| 24353 |
+
"step": 3462
|
| 24354 |
+
},
|
| 24355 |
+
{
|
| 24356 |
+
"epoch": 9.173509933774834,
|
| 24357 |
+
"grad_norm": 0.4611297845840454,
|
| 24358 |
+
"learning_rate": 4.071618037135279e-05,
|
| 24359 |
+
"loss": 0.0095,
|
| 24360 |
+
"step": 3463
|
| 24361 |
+
},
|
| 24362 |
+
{
|
| 24363 |
+
"epoch": 9.17615894039735,
|
| 24364 |
+
"grad_norm": 0.12303514778614044,
|
| 24365 |
+
"learning_rate": 4.058355437665783e-05,
|
| 24366 |
+
"loss": 0.0059,
|
| 24367 |
+
"step": 3464
|
| 24368 |
+
},
|
| 24369 |
+
{
|
| 24370 |
+
"epoch": 9.178807947019868,
|
| 24371 |
+
"grad_norm": 0.3407684862613678,
|
| 24372 |
+
"learning_rate": 4.0450928381962866e-05,
|
| 24373 |
+
"loss": 0.0072,
|
| 24374 |
+
"step": 3465
|
| 24375 |
+
},
|
| 24376 |
+
{
|
| 24377 |
+
"epoch": 9.181456953642384,
|
| 24378 |
+
"grad_norm": 0.7087451815605164,
|
| 24379 |
+
"learning_rate": 4.0318302387267904e-05,
|
| 24380 |
+
"loss": 0.0122,
|
| 24381 |
+
"step": 3466
|
| 24382 |
+
},
|
| 24383 |
+
{
|
| 24384 |
+
"epoch": 9.184105960264901,
|
| 24385 |
+
"grad_norm": 0.5841720700263977,
|
| 24386 |
+
"learning_rate": 4.018567639257294e-05,
|
| 24387 |
+
"loss": 0.0155,
|
| 24388 |
+
"step": 3467
|
| 24389 |
+
},
|
| 24390 |
+
{
|
| 24391 |
+
"epoch": 9.186754966887417,
|
| 24392 |
+
"grad_norm": 0.08858039975166321,
|
| 24393 |
+
"learning_rate": 4.005305039787799e-05,
|
| 24394 |
+
"loss": 0.0048,
|
| 24395 |
+
"step": 3468
|
| 24396 |
+
},
|
| 24397 |
+
{
|
| 24398 |
+
"epoch": 9.189403973509934,
|
| 24399 |
+
"grad_norm": 0.13346530497074127,
|
| 24400 |
+
"learning_rate": 3.9920424403183025e-05,
|
| 24401 |
+
"loss": 0.0065,
|
| 24402 |
+
"step": 3469
|
| 24403 |
+
},
|
| 24404 |
+
{
|
| 24405 |
+
"epoch": 9.19205298013245,
|
| 24406 |
+
"grad_norm": 0.09984955191612244,
|
| 24407 |
+
"learning_rate": 3.978779840848806e-05,
|
| 24408 |
+
"loss": 0.006,
|
| 24409 |
+
"step": 3470
|
| 24410 |
+
},
|
| 24411 |
+
{
|
| 24412 |
+
"epoch": 9.194701986754968,
|
| 24413 |
+
"grad_norm": 0.1724618822336197,
|
| 24414 |
+
"learning_rate": 3.965517241379311e-05,
|
| 24415 |
+
"loss": 0.0082,
|
| 24416 |
+
"step": 3471
|
| 24417 |
+
},
|
| 24418 |
+
{
|
| 24419 |
+
"epoch": 9.197350993377484,
|
| 24420 |
+
"grad_norm": 0.204783096909523,
|
| 24421 |
+
"learning_rate": 3.9522546419098145e-05,
|
| 24422 |
+
"loss": 0.007,
|
| 24423 |
+
"step": 3472
|
| 24424 |
+
},
|
| 24425 |
+
{
|
| 24426 |
+
"epoch": 9.2,
|
| 24427 |
+
"grad_norm": 0.6199121475219727,
|
| 24428 |
+
"learning_rate": 3.9389920424403183e-05,
|
| 24429 |
+
"loss": 0.0107,
|
| 24430 |
+
"step": 3473
|
| 24431 |
+
},
|
| 24432 |
+
{
|
| 24433 |
+
"epoch": 9.202649006622517,
|
| 24434 |
+
"grad_norm": 0.7197751998901367,
|
| 24435 |
+
"learning_rate": 3.925729442970823e-05,
|
| 24436 |
+
"loss": 0.0162,
|
| 24437 |
+
"step": 3474
|
| 24438 |
+
},
|
| 24439 |
+
{
|
| 24440 |
+
"epoch": 9.205298013245033,
|
| 24441 |
+
"grad_norm": 0.1388215869665146,
|
| 24442 |
+
"learning_rate": 3.9124668435013266e-05,
|
| 24443 |
+
"loss": 0.0077,
|
| 24444 |
+
"step": 3475
|
| 24445 |
+
},
|
| 24446 |
+
{
|
| 24447 |
+
"epoch": 9.20794701986755,
|
| 24448 |
+
"grad_norm": 0.1174258440732956,
|
| 24449 |
+
"learning_rate": 3.8992042440318304e-05,
|
| 24450 |
+
"loss": 0.0054,
|
| 24451 |
+
"step": 3476
|
| 24452 |
+
},
|
| 24453 |
+
{
|
| 24454 |
+
"epoch": 9.210596026490066,
|
| 24455 |
+
"grad_norm": 4.751803398132324,
|
| 24456 |
+
"learning_rate": 3.885941644562334e-05,
|
| 24457 |
+
"loss": 0.041,
|
| 24458 |
+
"step": 3477
|
| 24459 |
+
},
|
| 24460 |
+
{
|
| 24461 |
+
"epoch": 9.213245033112583,
|
| 24462 |
+
"grad_norm": 0.09883010387420654,
|
| 24463 |
+
"learning_rate": 3.872679045092838e-05,
|
| 24464 |
+
"loss": 0.0067,
|
| 24465 |
+
"step": 3478
|
| 24466 |
+
},
|
| 24467 |
+
{
|
| 24468 |
+
"epoch": 9.2158940397351,
|
| 24469 |
+
"grad_norm": 0.12656521797180176,
|
| 24470 |
+
"learning_rate": 3.859416445623342e-05,
|
| 24471 |
+
"loss": 0.0084,
|
| 24472 |
+
"step": 3479
|
| 24473 |
+
},
|
| 24474 |
+
{
|
| 24475 |
+
"epoch": 9.218543046357617,
|
| 24476 |
+
"grad_norm": 0.16607128083705902,
|
| 24477 |
+
"learning_rate": 3.846153846153846e-05,
|
| 24478 |
+
"loss": 0.0068,
|
| 24479 |
+
"step": 3480
|
| 24480 |
+
},
|
| 24481 |
+
{
|
| 24482 |
+
"epoch": 9.221192052980133,
|
| 24483 |
+
"grad_norm": 0.22199641168117523,
|
| 24484 |
+
"learning_rate": 3.83289124668435e-05,
|
| 24485 |
+
"loss": 0.0073,
|
| 24486 |
+
"step": 3481
|
| 24487 |
+
},
|
| 24488 |
+
{
|
| 24489 |
+
"epoch": 9.223841059602648,
|
| 24490 |
+
"grad_norm": 0.09226368367671967,
|
| 24491 |
+
"learning_rate": 3.819628647214854e-05,
|
| 24492 |
+
"loss": 0.0055,
|
| 24493 |
+
"step": 3482
|
| 24494 |
+
},
|
| 24495 |
+
{
|
| 24496 |
+
"epoch": 9.226490066225166,
|
| 24497 |
+
"grad_norm": 2.1479151248931885,
|
| 24498 |
+
"learning_rate": 3.8063660477453584e-05,
|
| 24499 |
+
"loss": 0.0295,
|
| 24500 |
+
"step": 3483
|
| 24501 |
+
},
|
| 24502 |
+
{
|
| 24503 |
+
"epoch": 9.229139072847682,
|
| 24504 |
+
"grad_norm": 0.11172802746295929,
|
| 24505 |
+
"learning_rate": 3.793103448275862e-05,
|
| 24506 |
+
"loss": 0.0053,
|
| 24507 |
+
"step": 3484
|
| 24508 |
+
},
|
| 24509 |
+
{
|
| 24510 |
+
"epoch": 9.2317880794702,
|
| 24511 |
+
"grad_norm": 0.07896125316619873,
|
| 24512 |
+
"learning_rate": 3.779840848806366e-05,
|
| 24513 |
+
"loss": 0.0059,
|
| 24514 |
+
"step": 3485
|
| 24515 |
+
},
|
| 24516 |
+
{
|
| 24517 |
+
"epoch": 9.234437086092715,
|
| 24518 |
+
"grad_norm": 0.10223594307899475,
|
| 24519 |
+
"learning_rate": 3.7665782493368704e-05,
|
| 24520 |
+
"loss": 0.0052,
|
| 24521 |
+
"step": 3486
|
| 24522 |
+
},
|
| 24523 |
+
{
|
| 24524 |
+
"epoch": 9.237086092715233,
|
| 24525 |
+
"grad_norm": 0.17130863666534424,
|
| 24526 |
+
"learning_rate": 3.753315649867374e-05,
|
| 24527 |
+
"loss": 0.0074,
|
| 24528 |
+
"step": 3487
|
| 24529 |
+
},
|
| 24530 |
+
{
|
| 24531 |
+
"epoch": 9.239735099337748,
|
| 24532 |
+
"grad_norm": 0.1503896713256836,
|
| 24533 |
+
"learning_rate": 3.740053050397878e-05,
|
| 24534 |
+
"loss": 0.007,
|
| 24535 |
+
"step": 3488
|
| 24536 |
+
},
|
| 24537 |
+
{
|
| 24538 |
+
"epoch": 9.242384105960264,
|
| 24539 |
+
"grad_norm": 0.4115447998046875,
|
| 24540 |
+
"learning_rate": 3.7267904509283825e-05,
|
| 24541 |
+
"loss": 0.0081,
|
| 24542 |
+
"step": 3489
|
| 24543 |
+
},
|
| 24544 |
+
{
|
| 24545 |
+
"epoch": 9.245033112582782,
|
| 24546 |
+
"grad_norm": 1.1161892414093018,
|
| 24547 |
+
"learning_rate": 3.7135278514588856e-05,
|
| 24548 |
+
"loss": 0.0149,
|
| 24549 |
+
"step": 3490
|
| 24550 |
+
},
|
| 24551 |
+
{
|
| 24552 |
+
"epoch": 9.247682119205297,
|
| 24553 |
+
"grad_norm": 0.08194896578788757,
|
| 24554 |
+
"learning_rate": 3.7002652519893894e-05,
|
| 24555 |
+
"loss": 0.0048,
|
| 24556 |
+
"step": 3491
|
| 24557 |
+
},
|
| 24558 |
+
{
|
| 24559 |
+
"epoch": 9.250331125827815,
|
| 24560 |
+
"grad_norm": 0.28096774220466614,
|
| 24561 |
+
"learning_rate": 3.687002652519894e-05,
|
| 24562 |
+
"loss": 0.0086,
|
| 24563 |
+
"step": 3492
|
| 24564 |
+
},
|
| 24565 |
+
{
|
| 24566 |
+
"epoch": 9.25298013245033,
|
| 24567 |
+
"grad_norm": 0.25251340866088867,
|
| 24568 |
+
"learning_rate": 3.673740053050398e-05,
|
| 24569 |
+
"loss": 0.0098,
|
| 24570 |
+
"step": 3493
|
| 24571 |
+
},
|
| 24572 |
+
{
|
| 24573 |
+
"epoch": 9.255629139072848,
|
| 24574 |
+
"grad_norm": 0.11859967559576035,
|
| 24575 |
+
"learning_rate": 3.660477453580902e-05,
|
| 24576 |
+
"loss": 0.0074,
|
| 24577 |
+
"step": 3494
|
| 24578 |
+
},
|
| 24579 |
+
{
|
| 24580 |
+
"epoch": 9.258278145695364,
|
| 24581 |
+
"grad_norm": 0.17372949421405792,
|
| 24582 |
+
"learning_rate": 3.647214854111406e-05,
|
| 24583 |
+
"loss": 0.0065,
|
| 24584 |
+
"step": 3495
|
| 24585 |
+
},
|
| 24586 |
+
{
|
| 24587 |
+
"epoch": 9.260927152317882,
|
| 24588 |
+
"grad_norm": 0.1542670577764511,
|
| 24589 |
+
"learning_rate": 3.63395225464191e-05,
|
| 24590 |
+
"loss": 0.0071,
|
| 24591 |
+
"step": 3496
|
| 24592 |
+
},
|
| 24593 |
+
{
|
| 24594 |
+
"epoch": 9.263576158940397,
|
| 24595 |
+
"grad_norm": 1.8768460750579834,
|
| 24596 |
+
"learning_rate": 3.620689655172414e-05,
|
| 24597 |
+
"loss": 0.042,
|
| 24598 |
+
"step": 3497
|
| 24599 |
+
},
|
| 24600 |
+
{
|
| 24601 |
+
"epoch": 9.266225165562913,
|
| 24602 |
+
"grad_norm": 0.11673364043235779,
|
| 24603 |
+
"learning_rate": 3.607427055702918e-05,
|
| 24604 |
+
"loss": 0.0066,
|
| 24605 |
+
"step": 3498
|
| 24606 |
+
},
|
| 24607 |
+
{
|
| 24608 |
+
"epoch": 9.26887417218543,
|
| 24609 |
+
"grad_norm": 0.0539066381752491,
|
| 24610 |
+
"learning_rate": 3.594164456233422e-05,
|
| 24611 |
+
"loss": 0.0048,
|
| 24612 |
+
"step": 3499
|
| 24613 |
+
},
|
| 24614 |
+
{
|
| 24615 |
+
"epoch": 9.271523178807946,
|
| 24616 |
+
"grad_norm": 0.11874030530452728,
|
| 24617 |
+
"learning_rate": 3.580901856763926e-05,
|
| 24618 |
+
"loss": 0.0055,
|
| 24619 |
+
"step": 3500
|
| 24620 |
+
},
|
| 24621 |
+
{
|
| 24622 |
+
"epoch": 9.274172185430464,
|
| 24623 |
+
"grad_norm": 0.8672187328338623,
|
| 24624 |
+
"learning_rate": 3.56763925729443e-05,
|
| 24625 |
+
"loss": 0.0218,
|
| 24626 |
+
"step": 3501
|
| 24627 |
+
},
|
| 24628 |
+
{
|
| 24629 |
+
"epoch": 9.27682119205298,
|
| 24630 |
+
"grad_norm": 0.21986471116542816,
|
| 24631 |
+
"learning_rate": 3.554376657824933e-05,
|
| 24632 |
+
"loss": 0.0072,
|
| 24633 |
+
"step": 3502
|
| 24634 |
+
},
|
| 24635 |
+
{
|
| 24636 |
+
"epoch": 9.279470198675497,
|
| 24637 |
+
"grad_norm": 0.19156527519226074,
|
| 24638 |
+
"learning_rate": 3.541114058355438e-05,
|
| 24639 |
+
"loss": 0.0086,
|
| 24640 |
+
"step": 3503
|
| 24641 |
+
},
|
| 24642 |
+
{
|
| 24643 |
+
"epoch": 9.282119205298013,
|
| 24644 |
+
"grad_norm": 0.07234346866607666,
|
| 24645 |
+
"learning_rate": 3.5278514588859415e-05,
|
| 24646 |
+
"loss": 0.0045,
|
| 24647 |
+
"step": 3504
|
| 24648 |
+
},
|
| 24649 |
+
{
|
| 24650 |
+
"epoch": 9.28476821192053,
|
| 24651 |
+
"grad_norm": 0.10020241886377335,
|
| 24652 |
+
"learning_rate": 3.514588859416445e-05,
|
| 24653 |
+
"loss": 0.0053,
|
| 24654 |
+
"step": 3505
|
| 24655 |
+
},
|
| 24656 |
+
{
|
| 24657 |
+
"epoch": 9.287417218543046,
|
| 24658 |
+
"grad_norm": 0.11927499622106552,
|
| 24659 |
+
"learning_rate": 3.50132625994695e-05,
|
| 24660 |
+
"loss": 0.0068,
|
| 24661 |
+
"step": 3506
|
| 24662 |
+
},
|
| 24663 |
+
{
|
| 24664 |
+
"epoch": 9.290066225165562,
|
| 24665 |
+
"grad_norm": 0.3162318766117096,
|
| 24666 |
+
"learning_rate": 3.4880636604774536e-05,
|
| 24667 |
+
"loss": 0.0079,
|
| 24668 |
+
"step": 3507
|
| 24669 |
+
},
|
| 24670 |
+
{
|
| 24671 |
+
"epoch": 9.29271523178808,
|
| 24672 |
+
"grad_norm": 0.0676041841506958,
|
| 24673 |
+
"learning_rate": 3.4748010610079574e-05,
|
| 24674 |
+
"loss": 0.0052,
|
| 24675 |
+
"step": 3508
|
| 24676 |
+
},
|
| 24677 |
+
{
|
| 24678 |
+
"epoch": 9.295364238410595,
|
| 24679 |
+
"grad_norm": 0.1660802811384201,
|
| 24680 |
+
"learning_rate": 3.461538461538462e-05,
|
| 24681 |
+
"loss": 0.0063,
|
| 24682 |
+
"step": 3509
|
| 24683 |
+
},
|
| 24684 |
+
{
|
| 24685 |
+
"epoch": 9.298013245033113,
|
| 24686 |
+
"grad_norm": 0.22426308691501617,
|
| 24687 |
+
"learning_rate": 3.4482758620689657e-05,
|
| 24688 |
+
"loss": 0.0087,
|
| 24689 |
+
"step": 3510
|
| 24690 |
+
},
|
| 24691 |
+
{
|
| 24692 |
+
"epoch": 9.300662251655629,
|
| 24693 |
+
"grad_norm": 0.3508667051792145,
|
| 24694 |
+
"learning_rate": 3.4350132625994695e-05,
|
| 24695 |
+
"loss": 0.0063,
|
| 24696 |
+
"step": 3511
|
| 24697 |
+
},
|
| 24698 |
+
{
|
| 24699 |
+
"epoch": 9.303311258278146,
|
| 24700 |
+
"grad_norm": 0.08600914478302002,
|
| 24701 |
+
"learning_rate": 3.421750663129974e-05,
|
| 24702 |
+
"loss": 0.0059,
|
| 24703 |
+
"step": 3512
|
| 24704 |
+
},
|
| 24705 |
+
{
|
| 24706 |
+
"epoch": 9.305960264900662,
|
| 24707 |
+
"grad_norm": 0.19650253653526306,
|
| 24708 |
+
"learning_rate": 3.408488063660478e-05,
|
| 24709 |
+
"loss": 0.0072,
|
| 24710 |
+
"step": 3513
|
| 24711 |
+
},
|
| 24712 |
+
{
|
| 24713 |
+
"epoch": 9.30860927152318,
|
| 24714 |
+
"grad_norm": 1.2427210807800293,
|
| 24715 |
+
"learning_rate": 3.395225464190981e-05,
|
| 24716 |
+
"loss": 0.0161,
|
| 24717 |
+
"step": 3514
|
| 24718 |
+
},
|
| 24719 |
+
{
|
| 24720 |
+
"epoch": 9.311258278145695,
|
| 24721 |
+
"grad_norm": 0.1413085013628006,
|
| 24722 |
+
"learning_rate": 3.381962864721485e-05,
|
| 24723 |
+
"loss": 0.0061,
|
| 24724 |
+
"step": 3515
|
| 24725 |
+
},
|
| 24726 |
+
{
|
| 24727 |
+
"epoch": 9.313907284768211,
|
| 24728 |
+
"grad_norm": 0.08845118433237076,
|
| 24729 |
+
"learning_rate": 3.368700265251989e-05,
|
| 24730 |
+
"loss": 0.0056,
|
| 24731 |
+
"step": 3516
|
| 24732 |
+
},
|
| 24733 |
+
{
|
| 24734 |
+
"epoch": 9.316556291390729,
|
| 24735 |
+
"grad_norm": 0.09792495518922806,
|
| 24736 |
+
"learning_rate": 3.355437665782493e-05,
|
| 24737 |
+
"loss": 0.0061,
|
| 24738 |
+
"step": 3517
|
| 24739 |
+
},
|
| 24740 |
+
{
|
| 24741 |
+
"epoch": 9.319205298013244,
|
| 24742 |
+
"grad_norm": 0.10987572371959686,
|
| 24743 |
+
"learning_rate": 3.3421750663129974e-05,
|
| 24744 |
+
"loss": 0.0061,
|
| 24745 |
+
"step": 3518
|
| 24746 |
+
},
|
| 24747 |
+
{
|
| 24748 |
+
"epoch": 9.321854304635762,
|
| 24749 |
+
"grad_norm": 0.846443235874176,
|
| 24750 |
+
"learning_rate": 3.328912466843501e-05,
|
| 24751 |
+
"loss": 0.0147,
|
| 24752 |
+
"step": 3519
|
| 24753 |
+
},
|
| 24754 |
+
{
|
| 24755 |
+
"epoch": 9.324503311258278,
|
| 24756 |
+
"grad_norm": 0.07839330285787582,
|
| 24757 |
+
"learning_rate": 3.315649867374006e-05,
|
| 24758 |
+
"loss": 0.0047,
|
| 24759 |
+
"step": 3520
|
| 24760 |
+
},
|
| 24761 |
+
{
|
| 24762 |
+
"epoch": 9.327152317880795,
|
| 24763 |
+
"grad_norm": 0.11738426238298416,
|
| 24764 |
+
"learning_rate": 3.3023872679045095e-05,
|
| 24765 |
+
"loss": 0.0065,
|
| 24766 |
+
"step": 3521
|
| 24767 |
+
},
|
| 24768 |
+
{
|
| 24769 |
+
"epoch": 9.329801324503311,
|
| 24770 |
+
"grad_norm": 0.13772623240947723,
|
| 24771 |
+
"learning_rate": 3.289124668435013e-05,
|
| 24772 |
+
"loss": 0.0071,
|
| 24773 |
+
"step": 3522
|
| 24774 |
+
},
|
| 24775 |
+
{
|
| 24776 |
+
"epoch": 9.332450331125829,
|
| 24777 |
+
"grad_norm": 1.1634302139282227,
|
| 24778 |
+
"learning_rate": 3.275862068965518e-05,
|
| 24779 |
+
"loss": 0.0238,
|
| 24780 |
+
"step": 3523
|
| 24781 |
+
},
|
| 24782 |
+
{
|
| 24783 |
+
"epoch": 9.335099337748344,
|
| 24784 |
+
"grad_norm": 2.1623897552490234,
|
| 24785 |
+
"learning_rate": 3.2625994694960215e-05,
|
| 24786 |
+
"loss": 0.0999,
|
| 24787 |
+
"step": 3524
|
| 24788 |
+
},
|
| 24789 |
+
{
|
| 24790 |
+
"epoch": 9.33774834437086,
|
| 24791 |
+
"grad_norm": 0.19755753874778748,
|
| 24792 |
+
"learning_rate": 3.2493368700265253e-05,
|
| 24793 |
+
"loss": 0.0075,
|
| 24794 |
+
"step": 3525
|
| 24795 |
+
},
|
| 24796 |
+
{
|
| 24797 |
+
"epoch": 9.340397350993378,
|
| 24798 |
+
"grad_norm": 0.539291262626648,
|
| 24799 |
+
"learning_rate": 3.23607427055703e-05,
|
| 24800 |
+
"loss": 0.0131,
|
| 24801 |
+
"step": 3526
|
| 24802 |
+
},
|
| 24803 |
+
{
|
| 24804 |
+
"epoch": 9.343046357615894,
|
| 24805 |
+
"grad_norm": 0.1200319305062294,
|
| 24806 |
+
"learning_rate": 3.222811671087533e-05,
|
| 24807 |
+
"loss": 0.0064,
|
| 24808 |
+
"step": 3527
|
| 24809 |
+
},
|
| 24810 |
+
{
|
| 24811 |
+
"epoch": 9.345695364238411,
|
| 24812 |
+
"grad_norm": 0.10444632917642593,
|
| 24813 |
+
"learning_rate": 3.209549071618037e-05,
|
| 24814 |
+
"loss": 0.0061,
|
| 24815 |
+
"step": 3528
|
| 24816 |
+
},
|
| 24817 |
+
{
|
| 24818 |
+
"epoch": 9.348344370860927,
|
| 24819 |
+
"grad_norm": 0.09037845581769943,
|
| 24820 |
+
"learning_rate": 3.196286472148541e-05,
|
| 24821 |
+
"loss": 0.0054,
|
| 24822 |
+
"step": 3529
|
| 24823 |
+
},
|
| 24824 |
+
{
|
| 24825 |
+
"epoch": 9.350993377483444,
|
| 24826 |
+
"grad_norm": 0.09533439576625824,
|
| 24827 |
+
"learning_rate": 3.183023872679045e-05,
|
| 24828 |
+
"loss": 0.0066,
|
| 24829 |
+
"step": 3530
|
| 24830 |
+
},
|
| 24831 |
+
{
|
| 24832 |
+
"epoch": 9.35364238410596,
|
| 24833 |
+
"grad_norm": 0.07484545558691025,
|
| 24834 |
+
"learning_rate": 3.169761273209549e-05,
|
| 24835 |
+
"loss": 0.0052,
|
| 24836 |
+
"step": 3531
|
| 24837 |
+
},
|
| 24838 |
+
{
|
| 24839 |
+
"epoch": 9.356291390728476,
|
| 24840 |
+
"grad_norm": 0.13758543133735657,
|
| 24841 |
+
"learning_rate": 3.156498673740053e-05,
|
| 24842 |
+
"loss": 0.006,
|
| 24843 |
+
"step": 3532
|
| 24844 |
+
},
|
| 24845 |
+
{
|
| 24846 |
+
"epoch": 9.358940397350993,
|
| 24847 |
+
"grad_norm": 3.194049119949341,
|
| 24848 |
+
"learning_rate": 3.143236074270557e-05,
|
| 24849 |
+
"loss": 0.0414,
|
| 24850 |
+
"step": 3533
|
| 24851 |
+
},
|
| 24852 |
+
{
|
| 24853 |
+
"epoch": 9.36158940397351,
|
| 24854 |
+
"grad_norm": 0.4014771282672882,
|
| 24855 |
+
"learning_rate": 3.129973474801061e-05,
|
| 24856 |
+
"loss": 0.0117,
|
| 24857 |
+
"step": 3534
|
| 24858 |
+
},
|
| 24859 |
+
{
|
| 24860 |
+
"epoch": 9.364238410596027,
|
| 24861 |
+
"grad_norm": 1.8910866975784302,
|
| 24862 |
+
"learning_rate": 3.116710875331565e-05,
|
| 24863 |
+
"loss": 0.0437,
|
| 24864 |
+
"step": 3535
|
| 24865 |
+
},
|
| 24866 |
+
{
|
| 24867 |
+
"epoch": 9.366887417218543,
|
| 24868 |
+
"grad_norm": 0.1309318244457245,
|
| 24869 |
+
"learning_rate": 3.103448275862069e-05,
|
| 24870 |
+
"loss": 0.0067,
|
| 24871 |
+
"step": 3536
|
| 24872 |
+
},
|
| 24873 |
+
{
|
| 24874 |
+
"epoch": 9.36953642384106,
|
| 24875 |
+
"grad_norm": 0.2526959776878357,
|
| 24876 |
+
"learning_rate": 3.090185676392573e-05,
|
| 24877 |
+
"loss": 0.0094,
|
| 24878 |
+
"step": 3537
|
| 24879 |
+
},
|
| 24880 |
+
{
|
| 24881 |
+
"epoch": 9.372185430463576,
|
| 24882 |
+
"grad_norm": 0.17184001207351685,
|
| 24883 |
+
"learning_rate": 3.0769230769230774e-05,
|
| 24884 |
+
"loss": 0.0074,
|
| 24885 |
+
"step": 3538
|
| 24886 |
+
},
|
| 24887 |
+
{
|
| 24888 |
+
"epoch": 9.374834437086093,
|
| 24889 |
+
"grad_norm": 0.26467862725257874,
|
| 24890 |
+
"learning_rate": 3.063660477453581e-05,
|
| 24891 |
+
"loss": 0.0094,
|
| 24892 |
+
"step": 3539
|
| 24893 |
+
},
|
| 24894 |
+
{
|
| 24895 |
+
"epoch": 9.37748344370861,
|
| 24896 |
+
"grad_norm": 0.05459669604897499,
|
| 24897 |
+
"learning_rate": 3.050397877984085e-05,
|
| 24898 |
+
"loss": 0.0043,
|
| 24899 |
+
"step": 3540
|
| 24900 |
+
},
|
| 24901 |
+
{
|
| 24902 |
+
"epoch": 9.380132450331125,
|
| 24903 |
+
"grad_norm": 0.1171836405992508,
|
| 24904 |
+
"learning_rate": 3.0371352785145892e-05,
|
| 24905 |
+
"loss": 0.0067,
|
| 24906 |
+
"step": 3541
|
| 24907 |
+
},
|
| 24908 |
+
{
|
| 24909 |
+
"epoch": 9.382781456953643,
|
| 24910 |
+
"grad_norm": 0.10178942233324051,
|
| 24911 |
+
"learning_rate": 3.023872679045093e-05,
|
| 24912 |
+
"loss": 0.007,
|
| 24913 |
+
"step": 3542
|
| 24914 |
+
},
|
| 24915 |
+
{
|
| 24916 |
+
"epoch": 9.385430463576158,
|
| 24917 |
+
"grad_norm": 1.3025785684585571,
|
| 24918 |
+
"learning_rate": 3.0106100795755968e-05,
|
| 24919 |
+
"loss": 0.0344,
|
| 24920 |
+
"step": 3543
|
| 24921 |
+
},
|
| 24922 |
+
{
|
| 24923 |
+
"epoch": 9.388079470198676,
|
| 24924 |
+
"grad_norm": 1.7906361818313599,
|
| 24925 |
+
"learning_rate": 2.997347480106101e-05,
|
| 24926 |
+
"loss": 0.0824,
|
| 24927 |
+
"step": 3544
|
| 24928 |
+
},
|
| 24929 |
+
{
|
| 24930 |
+
"epoch": 9.390728476821192,
|
| 24931 |
+
"grad_norm": 0.10218120366334915,
|
| 24932 |
+
"learning_rate": 2.9840848806366047e-05,
|
| 24933 |
+
"loss": 0.0066,
|
| 24934 |
+
"step": 3545
|
| 24935 |
+
},
|
| 24936 |
+
{
|
| 24937 |
+
"epoch": 9.39337748344371,
|
| 24938 |
+
"grad_norm": 0.11341088265180588,
|
| 24939 |
+
"learning_rate": 2.970822281167109e-05,
|
| 24940 |
+
"loss": 0.0074,
|
| 24941 |
+
"step": 3546
|
| 24942 |
+
},
|
| 24943 |
+
{
|
| 24944 |
+
"epoch": 9.396026490066225,
|
| 24945 |
+
"grad_norm": 0.128928080201149,
|
| 24946 |
+
"learning_rate": 2.957559681697613e-05,
|
| 24947 |
+
"loss": 0.0063,
|
| 24948 |
+
"step": 3547
|
| 24949 |
+
},
|
| 24950 |
+
{
|
| 24951 |
+
"epoch": 9.398675496688742,
|
| 24952 |
+
"grad_norm": 0.09247614443302155,
|
| 24953 |
+
"learning_rate": 2.9442970822281168e-05,
|
| 24954 |
+
"loss": 0.0057,
|
| 24955 |
+
"step": 3548
|
| 24956 |
+
},
|
| 24957 |
+
{
|
| 24958 |
+
"epoch": 9.401324503311258,
|
| 24959 |
+
"grad_norm": 0.9803311824798584,
|
| 24960 |
+
"learning_rate": 2.9310344827586206e-05,
|
| 24961 |
+
"loss": 0.0182,
|
| 24962 |
+
"step": 3549
|
| 24963 |
+
},
|
| 24964 |
+
{
|
| 24965 |
+
"epoch": 9.403973509933774,
|
| 24966 |
+
"grad_norm": 0.23505902290344238,
|
| 24967 |
+
"learning_rate": 2.9177718832891247e-05,
|
| 24968 |
+
"loss": 0.0071,
|
| 24969 |
+
"step": 3550
|
| 24970 |
+
},
|
| 24971 |
+
{
|
| 24972 |
+
"epoch": 9.406622516556292,
|
| 24973 |
+
"grad_norm": 0.2330973744392395,
|
| 24974 |
+
"learning_rate": 2.904509283819629e-05,
|
| 24975 |
+
"loss": 0.011,
|
| 24976 |
+
"step": 3551
|
| 24977 |
+
},
|
| 24978 |
+
{
|
| 24979 |
+
"epoch": 9.409271523178807,
|
| 24980 |
+
"grad_norm": 0.11561492830514908,
|
| 24981 |
+
"learning_rate": 2.8912466843501326e-05,
|
| 24982 |
+
"loss": 0.0064,
|
| 24983 |
+
"step": 3552
|
| 24984 |
+
},
|
| 24985 |
+
{
|
| 24986 |
+
"epoch": 9.411920529801325,
|
| 24987 |
+
"grad_norm": 0.2049504816532135,
|
| 24988 |
+
"learning_rate": 2.8779840848806368e-05,
|
| 24989 |
+
"loss": 0.0082,
|
| 24990 |
+
"step": 3553
|
| 24991 |
+
},
|
| 24992 |
+
{
|
| 24993 |
+
"epoch": 9.41456953642384,
|
| 24994 |
+
"grad_norm": 0.08177812397480011,
|
| 24995 |
+
"learning_rate": 2.864721485411141e-05,
|
| 24996 |
+
"loss": 0.0052,
|
| 24997 |
+
"step": 3554
|
| 24998 |
+
},
|
| 24999 |
+
{
|
| 25000 |
+
"epoch": 9.417218543046358,
|
| 25001 |
+
"grad_norm": 0.12092561274766922,
|
| 25002 |
+
"learning_rate": 2.8514588859416444e-05,
|
| 25003 |
+
"loss": 0.0061,
|
| 25004 |
+
"step": 3555
|
| 25005 |
+
},
|
| 25006 |
+
{
|
| 25007 |
+
"epoch": 9.419867549668874,
|
| 25008 |
+
"grad_norm": 0.08898591995239258,
|
| 25009 |
+
"learning_rate": 2.8381962864721485e-05,
|
| 25010 |
+
"loss": 0.0061,
|
| 25011 |
+
"step": 3556
|
| 25012 |
+
},
|
| 25013 |
+
{
|
| 25014 |
+
"epoch": 9.422516556291392,
|
| 25015 |
+
"grad_norm": 0.15374480187892914,
|
| 25016 |
+
"learning_rate": 2.8249336870026527e-05,
|
| 25017 |
+
"loss": 0.0058,
|
| 25018 |
+
"step": 3557
|
| 25019 |
+
},
|
| 25020 |
+
{
|
| 25021 |
+
"epoch": 9.425165562913907,
|
| 25022 |
+
"grad_norm": 0.1372290402650833,
|
| 25023 |
+
"learning_rate": 2.8116710875331565e-05,
|
| 25024 |
+
"loss": 0.0071,
|
| 25025 |
+
"step": 3558
|
| 25026 |
+
},
|
| 25027 |
+
{
|
| 25028 |
+
"epoch": 9.427814569536423,
|
| 25029 |
+
"grad_norm": 0.11643172055482864,
|
| 25030 |
+
"learning_rate": 2.7984084880636606e-05,
|
| 25031 |
+
"loss": 0.0068,
|
| 25032 |
+
"step": 3559
|
| 25033 |
+
},
|
| 25034 |
+
{
|
| 25035 |
+
"epoch": 9.43046357615894,
|
| 25036 |
+
"grad_norm": 0.09352966398000717,
|
| 25037 |
+
"learning_rate": 2.7851458885941647e-05,
|
| 25038 |
+
"loss": 0.006,
|
| 25039 |
+
"step": 3560
|
| 25040 |
+
},
|
| 25041 |
+
{
|
| 25042 |
+
"epoch": 9.433112582781456,
|
| 25043 |
+
"grad_norm": 0.20532579720020294,
|
| 25044 |
+
"learning_rate": 2.7718832891246682e-05,
|
| 25045 |
+
"loss": 0.0071,
|
| 25046 |
+
"step": 3561
|
| 25047 |
+
},
|
| 25048 |
+
{
|
| 25049 |
+
"epoch": 9.435761589403974,
|
| 25050 |
+
"grad_norm": 1.9546130895614624,
|
| 25051 |
+
"learning_rate": 2.7586206896551723e-05,
|
| 25052 |
+
"loss": 0.0191,
|
| 25053 |
+
"step": 3562
|
| 25054 |
+
},
|
| 25055 |
+
{
|
| 25056 |
+
"epoch": 9.43841059602649,
|
| 25057 |
+
"grad_norm": 0.16163243353366852,
|
| 25058 |
+
"learning_rate": 2.7453580901856765e-05,
|
| 25059 |
+
"loss": 0.007,
|
| 25060 |
+
"step": 3563
|
| 25061 |
+
},
|
| 25062 |
+
{
|
| 25063 |
+
"epoch": 9.441059602649007,
|
| 25064 |
+
"grad_norm": 0.18948762118816376,
|
| 25065 |
+
"learning_rate": 2.7320954907161803e-05,
|
| 25066 |
+
"loss": 0.0079,
|
| 25067 |
+
"step": 3564
|
| 25068 |
+
},
|
| 25069 |
+
{
|
| 25070 |
+
"epoch": 9.443708609271523,
|
| 25071 |
+
"grad_norm": 0.2582177519798279,
|
| 25072 |
+
"learning_rate": 2.7188328912466844e-05,
|
| 25073 |
+
"loss": 0.0093,
|
| 25074 |
+
"step": 3565
|
| 25075 |
+
},
|
| 25076 |
+
{
|
| 25077 |
+
"epoch": 9.44635761589404,
|
| 25078 |
+
"grad_norm": 0.20868119597434998,
|
| 25079 |
+
"learning_rate": 2.7055702917771885e-05,
|
| 25080 |
+
"loss": 0.0081,
|
| 25081 |
+
"step": 3566
|
| 25082 |
+
},
|
| 25083 |
+
{
|
| 25084 |
+
"epoch": 9.449006622516556,
|
| 25085 |
+
"grad_norm": 1.0529993772506714,
|
| 25086 |
+
"learning_rate": 2.6923076923076927e-05,
|
| 25087 |
+
"loss": 0.0158,
|
| 25088 |
+
"step": 3567
|
| 25089 |
+
},
|
| 25090 |
+
{
|
| 25091 |
+
"epoch": 9.451655629139072,
|
| 25092 |
+
"grad_norm": 0.2311079204082489,
|
| 25093 |
+
"learning_rate": 2.679045092838196e-05,
|
| 25094 |
+
"loss": 0.0088,
|
| 25095 |
+
"step": 3568
|
| 25096 |
+
},
|
| 25097 |
+
{
|
| 25098 |
+
"epoch": 9.45430463576159,
|
| 25099 |
+
"grad_norm": 0.4316157400608063,
|
| 25100 |
+
"learning_rate": 2.6657824933687003e-05,
|
| 25101 |
+
"loss": 0.0106,
|
| 25102 |
+
"step": 3569
|
| 25103 |
+
},
|
| 25104 |
+
{
|
| 25105 |
+
"epoch": 9.456953642384105,
|
| 25106 |
+
"grad_norm": 0.2940143048763275,
|
| 25107 |
+
"learning_rate": 2.6525198938992044e-05,
|
| 25108 |
+
"loss": 0.0091,
|
| 25109 |
+
"step": 3570
|
| 25110 |
+
},
|
| 25111 |
+
{
|
| 25112 |
+
"epoch": 9.459602649006623,
|
| 25113 |
+
"grad_norm": 0.10199405252933502,
|
| 25114 |
+
"learning_rate": 2.6392572944297082e-05,
|
| 25115 |
+
"loss": 0.0061,
|
| 25116 |
+
"step": 3571
|
| 25117 |
+
},
|
| 25118 |
+
{
|
| 25119 |
+
"epoch": 9.462251655629139,
|
| 25120 |
+
"grad_norm": 0.09311787784099579,
|
| 25121 |
+
"learning_rate": 2.6259946949602123e-05,
|
| 25122 |
+
"loss": 0.0066,
|
| 25123 |
+
"step": 3572
|
| 25124 |
+
},
|
| 25125 |
+
{
|
| 25126 |
+
"epoch": 9.464900662251656,
|
| 25127 |
+
"grad_norm": 0.11943931877613068,
|
| 25128 |
+
"learning_rate": 2.6127320954907165e-05,
|
| 25129 |
+
"loss": 0.0067,
|
| 25130 |
+
"step": 3573
|
| 25131 |
+
},
|
| 25132 |
+
{
|
| 25133 |
+
"epoch": 9.467549668874172,
|
| 25134 |
+
"grad_norm": 0.08641272783279419,
|
| 25135 |
+
"learning_rate": 2.59946949602122e-05,
|
| 25136 |
+
"loss": 0.0064,
|
| 25137 |
+
"step": 3574
|
| 25138 |
+
},
|
| 25139 |
+
{
|
| 25140 |
+
"epoch": 9.47019867549669,
|
| 25141 |
+
"grad_norm": 2.144976854324341,
|
| 25142 |
+
"learning_rate": 2.586206896551724e-05,
|
| 25143 |
+
"loss": 0.0777,
|
| 25144 |
+
"step": 3575
|
| 25145 |
+
},
|
| 25146 |
+
{
|
| 25147 |
+
"epoch": 9.472847682119205,
|
| 25148 |
+
"grad_norm": 2.3750195503234863,
|
| 25149 |
+
"learning_rate": 2.5729442970822282e-05,
|
| 25150 |
+
"loss": 0.03,
|
| 25151 |
+
"step": 3576
|
| 25152 |
+
},
|
| 25153 |
+
{
|
| 25154 |
+
"epoch": 9.475496688741721,
|
| 25155 |
+
"grad_norm": 0.3863673210144043,
|
| 25156 |
+
"learning_rate": 2.559681697612732e-05,
|
| 25157 |
+
"loss": 0.0133,
|
| 25158 |
+
"step": 3577
|
| 25159 |
+
},
|
| 25160 |
+
{
|
| 25161 |
+
"epoch": 9.478145695364239,
|
| 25162 |
+
"grad_norm": 0.18488571047782898,
|
| 25163 |
+
"learning_rate": 2.546419098143236e-05,
|
| 25164 |
+
"loss": 0.0101,
|
| 25165 |
+
"step": 3578
|
| 25166 |
+
},
|
| 25167 |
+
{
|
| 25168 |
+
"epoch": 9.480794701986754,
|
| 25169 |
+
"grad_norm": 0.14614498615264893,
|
| 25170 |
+
"learning_rate": 2.5331564986737403e-05,
|
| 25171 |
+
"loss": 0.0078,
|
| 25172 |
+
"step": 3579
|
| 25173 |
+
},
|
| 25174 |
+
{
|
| 25175 |
+
"epoch": 9.483443708609272,
|
| 25176 |
+
"grad_norm": 0.502761721611023,
|
| 25177 |
+
"learning_rate": 2.519893899204244e-05,
|
| 25178 |
+
"loss": 0.0098,
|
| 25179 |
+
"step": 3580
|
| 25180 |
+
},
|
| 25181 |
+
{
|
| 25182 |
+
"epoch": 9.486092715231788,
|
| 25183 |
+
"grad_norm": 0.12372557818889618,
|
| 25184 |
+
"learning_rate": 2.506631299734748e-05,
|
| 25185 |
+
"loss": 0.0064,
|
| 25186 |
+
"step": 3581
|
| 25187 |
+
},
|
| 25188 |
+
{
|
| 25189 |
+
"epoch": 9.488741721854305,
|
| 25190 |
+
"grad_norm": 0.06779627501964569,
|
| 25191 |
+
"learning_rate": 2.493368700265252e-05,
|
| 25192 |
+
"loss": 0.0045,
|
| 25193 |
+
"step": 3582
|
| 25194 |
+
},
|
| 25195 |
+
{
|
| 25196 |
+
"epoch": 9.491390728476821,
|
| 25197 |
+
"grad_norm": 0.20525391399860382,
|
| 25198 |
+
"learning_rate": 2.480106100795756e-05,
|
| 25199 |
+
"loss": 0.009,
|
| 25200 |
+
"step": 3583
|
| 25201 |
+
},
|
| 25202 |
+
{
|
| 25203 |
+
"epoch": 9.494039735099339,
|
| 25204 |
+
"grad_norm": 0.12155789136886597,
|
| 25205 |
+
"learning_rate": 2.46684350132626e-05,
|
| 25206 |
+
"loss": 0.0063,
|
| 25207 |
+
"step": 3584
|
| 25208 |
+
},
|
| 25209 |
+
{
|
| 25210 |
+
"epoch": 9.496688741721854,
|
| 25211 |
+
"grad_norm": 0.10192214697599411,
|
| 25212 |
+
"learning_rate": 2.453580901856764e-05,
|
| 25213 |
+
"loss": 0.0057,
|
| 25214 |
+
"step": 3585
|
| 25215 |
+
},
|
| 25216 |
+
{
|
| 25217 |
+
"epoch": 9.49933774834437,
|
| 25218 |
+
"grad_norm": 0.987741231918335,
|
| 25219 |
+
"learning_rate": 2.440318302387268e-05,
|
| 25220 |
+
"loss": 0.0233,
|
| 25221 |
+
"step": 3586
|
| 25222 |
+
},
|
| 25223 |
+
{
|
| 25224 |
+
"epoch": 9.501986754966888,
|
| 25225 |
+
"grad_norm": 0.39217236638069153,
|
| 25226 |
+
"learning_rate": 2.4270557029177717e-05,
|
| 25227 |
+
"loss": 0.0098,
|
| 25228 |
+
"step": 3587
|
| 25229 |
+
},
|
| 25230 |
+
{
|
| 25231 |
+
"epoch": 9.504635761589403,
|
| 25232 |
+
"grad_norm": 0.11593826860189438,
|
| 25233 |
+
"learning_rate": 2.4137931034482758e-05,
|
| 25234 |
+
"loss": 0.0055,
|
| 25235 |
+
"step": 3588
|
| 25236 |
+
},
|
| 25237 |
+
{
|
| 25238 |
+
"epoch": 9.507284768211921,
|
| 25239 |
+
"grad_norm": 4.071221828460693,
|
| 25240 |
+
"learning_rate": 2.40053050397878e-05,
|
| 25241 |
+
"loss": 0.0562,
|
| 25242 |
+
"step": 3589
|
| 25243 |
+
},
|
| 25244 |
+
{
|
| 25245 |
+
"epoch": 9.509933774834437,
|
| 25246 |
+
"grad_norm": 2.431532382965088,
|
| 25247 |
+
"learning_rate": 2.3872679045092838e-05,
|
| 25248 |
+
"loss": 0.2236,
|
| 25249 |
+
"step": 3590
|
| 25250 |
+
},
|
| 25251 |
+
{
|
| 25252 |
+
"epoch": 9.512582781456954,
|
| 25253 |
+
"grad_norm": 0.9646773934364319,
|
| 25254 |
+
"learning_rate": 2.374005305039788e-05,
|
| 25255 |
+
"loss": 0.0297,
|
| 25256 |
+
"step": 3591
|
| 25257 |
+
},
|
| 25258 |
+
{
|
| 25259 |
+
"epoch": 9.51523178807947,
|
| 25260 |
+
"grad_norm": 0.9836496710777283,
|
| 25261 |
+
"learning_rate": 2.360742705570292e-05,
|
| 25262 |
+
"loss": 0.0152,
|
| 25263 |
+
"step": 3592
|
| 25264 |
+
},
|
| 25265 |
+
{
|
| 25266 |
+
"epoch": 9.517880794701988,
|
| 25267 |
+
"grad_norm": 0.08301722258329391,
|
| 25268 |
+
"learning_rate": 2.347480106100796e-05,
|
| 25269 |
+
"loss": 0.0054,
|
| 25270 |
+
"step": 3593
|
| 25271 |
+
},
|
| 25272 |
+
{
|
| 25273 |
+
"epoch": 9.520529801324503,
|
| 25274 |
+
"grad_norm": 0.3062174618244171,
|
| 25275 |
+
"learning_rate": 2.3342175066312996e-05,
|
| 25276 |
+
"loss": 0.0075,
|
| 25277 |
+
"step": 3594
|
| 25278 |
+
},
|
| 25279 |
+
{
|
| 25280 |
+
"epoch": 9.52317880794702,
|
| 25281 |
+
"grad_norm": 1.9968500137329102,
|
| 25282 |
+
"learning_rate": 2.3209549071618038e-05,
|
| 25283 |
+
"loss": 0.022,
|
| 25284 |
+
"step": 3595
|
| 25285 |
+
},
|
| 25286 |
+
{
|
| 25287 |
+
"epoch": 9.525827814569537,
|
| 25288 |
+
"grad_norm": 0.047811876982450485,
|
| 25289 |
+
"learning_rate": 2.307692307692308e-05,
|
| 25290 |
+
"loss": 0.0038,
|
| 25291 |
+
"step": 3596
|
| 25292 |
+
},
|
| 25293 |
+
{
|
| 25294 |
+
"epoch": 9.528476821192053,
|
| 25295 |
+
"grad_norm": 0.08924803137779236,
|
| 25296 |
+
"learning_rate": 2.2944297082228117e-05,
|
| 25297 |
+
"loss": 0.0063,
|
| 25298 |
+
"step": 3597
|
| 25299 |
+
},
|
| 25300 |
+
{
|
| 25301 |
+
"epoch": 9.53112582781457,
|
| 25302 |
+
"grad_norm": 0.16977474093437195,
|
| 25303 |
+
"learning_rate": 2.281167108753316e-05,
|
| 25304 |
+
"loss": 0.0068,
|
| 25305 |
+
"step": 3598
|
| 25306 |
+
},
|
| 25307 |
+
{
|
| 25308 |
+
"epoch": 9.533774834437086,
|
| 25309 |
+
"grad_norm": 0.06967270374298096,
|
| 25310 |
+
"learning_rate": 2.2679045092838196e-05,
|
| 25311 |
+
"loss": 0.0038,
|
| 25312 |
+
"step": 3599
|
| 25313 |
+
},
|
| 25314 |
+
{
|
| 25315 |
+
"epoch": 9.536423841059603,
|
| 25316 |
+
"grad_norm": 0.10326126962900162,
|
| 25317 |
+
"learning_rate": 2.2546419098143234e-05,
|
| 25318 |
+
"loss": 0.007,
|
| 25319 |
+
"step": 3600
|
| 25320 |
+
},
|
| 25321 |
+
{
|
| 25322 |
+
"epoch": 9.53907284768212,
|
| 25323 |
+
"grad_norm": 0.0956234410405159,
|
| 25324 |
+
"learning_rate": 2.2413793103448276e-05,
|
| 25325 |
+
"loss": 0.0048,
|
| 25326 |
+
"step": 3601
|
| 25327 |
+
},
|
| 25328 |
+
{
|
| 25329 |
+
"epoch": 9.541721854304635,
|
| 25330 |
+
"grad_norm": 0.11128674447536469,
|
| 25331 |
+
"learning_rate": 2.2281167108753317e-05,
|
| 25332 |
+
"loss": 0.0073,
|
| 25333 |
+
"step": 3602
|
| 25334 |
+
},
|
| 25335 |
+
{
|
| 25336 |
+
"epoch": 9.544370860927152,
|
| 25337 |
+
"grad_norm": 0.34343084692955017,
|
| 25338 |
+
"learning_rate": 2.2148541114058355e-05,
|
| 25339 |
+
"loss": 0.0094,
|
| 25340 |
+
"step": 3603
|
| 25341 |
+
},
|
| 25342 |
+
{
|
| 25343 |
+
"epoch": 9.547019867549668,
|
| 25344 |
+
"grad_norm": 0.10045126080513,
|
| 25345 |
+
"learning_rate": 2.2015915119363396e-05,
|
| 25346 |
+
"loss": 0.0058,
|
| 25347 |
+
"step": 3604
|
| 25348 |
+
},
|
| 25349 |
+
{
|
| 25350 |
+
"epoch": 9.549668874172186,
|
| 25351 |
+
"grad_norm": 0.20021073520183563,
|
| 25352 |
+
"learning_rate": 2.1883289124668434e-05,
|
| 25353 |
+
"loss": 0.0067,
|
| 25354 |
+
"step": 3605
|
| 25355 |
+
},
|
| 25356 |
+
{
|
| 25357 |
+
"epoch": 9.552317880794702,
|
| 25358 |
+
"grad_norm": 0.16547444462776184,
|
| 25359 |
+
"learning_rate": 2.1750663129973476e-05,
|
| 25360 |
+
"loss": 0.0066,
|
| 25361 |
+
"step": 3606
|
| 25362 |
+
},
|
| 25363 |
+
{
|
| 25364 |
+
"epoch": 9.55496688741722,
|
| 25365 |
+
"grad_norm": 0.6814205050468445,
|
| 25366 |
+
"learning_rate": 2.1618037135278514e-05,
|
| 25367 |
+
"loss": 0.0097,
|
| 25368 |
+
"step": 3607
|
| 25369 |
+
},
|
| 25370 |
+
{
|
| 25371 |
+
"epoch": 9.557615894039735,
|
| 25372 |
+
"grad_norm": 0.15485073626041412,
|
| 25373 |
+
"learning_rate": 2.1485411140583555e-05,
|
| 25374 |
+
"loss": 0.0066,
|
| 25375 |
+
"step": 3608
|
| 25376 |
+
},
|
| 25377 |
+
{
|
| 25378 |
+
"epoch": 9.560264900662252,
|
| 25379 |
+
"grad_norm": 0.3462410867214203,
|
| 25380 |
+
"learning_rate": 2.1352785145888597e-05,
|
| 25381 |
+
"loss": 0.0099,
|
| 25382 |
+
"step": 3609
|
| 25383 |
+
},
|
| 25384 |
+
{
|
| 25385 |
+
"epoch": 9.562913907284768,
|
| 25386 |
+
"grad_norm": 1.7099510431289673,
|
| 25387 |
+
"learning_rate": 2.1220159151193635e-05,
|
| 25388 |
+
"loss": 0.0234,
|
| 25389 |
+
"step": 3610
|
| 25390 |
+
},
|
| 25391 |
+
{
|
| 25392 |
+
"epoch": 9.565562913907284,
|
| 25393 |
+
"grad_norm": 0.18412642180919647,
|
| 25394 |
+
"learning_rate": 2.1087533156498673e-05,
|
| 25395 |
+
"loss": 0.0065,
|
| 25396 |
+
"step": 3611
|
| 25397 |
+
},
|
| 25398 |
+
{
|
| 25399 |
+
"epoch": 9.568211920529802,
|
| 25400 |
+
"grad_norm": 0.10271494835615158,
|
| 25401 |
+
"learning_rate": 2.0954907161803714e-05,
|
| 25402 |
+
"loss": 0.0054,
|
| 25403 |
+
"step": 3612
|
| 25404 |
+
},
|
| 25405 |
+
{
|
| 25406 |
+
"epoch": 9.570860927152317,
|
| 25407 |
+
"grad_norm": 0.1996956765651703,
|
| 25408 |
+
"learning_rate": 2.0822281167108752e-05,
|
| 25409 |
+
"loss": 0.0094,
|
| 25410 |
+
"step": 3613
|
| 25411 |
+
},
|
| 25412 |
+
{
|
| 25413 |
+
"epoch": 9.573509933774835,
|
| 25414 |
+
"grad_norm": 0.215884268283844,
|
| 25415 |
+
"learning_rate": 2.0689655172413793e-05,
|
| 25416 |
+
"loss": 0.008,
|
| 25417 |
+
"step": 3614
|
| 25418 |
+
},
|
| 25419 |
+
{
|
| 25420 |
+
"epoch": 9.57615894039735,
|
| 25421 |
+
"grad_norm": 0.40959951281547546,
|
| 25422 |
+
"learning_rate": 2.0557029177718835e-05,
|
| 25423 |
+
"loss": 0.0104,
|
| 25424 |
+
"step": 3615
|
| 25425 |
+
},
|
| 25426 |
+
{
|
| 25427 |
+
"epoch": 9.578807947019868,
|
| 25428 |
+
"grad_norm": 0.2678709626197815,
|
| 25429 |
+
"learning_rate": 2.0424403183023873e-05,
|
| 25430 |
+
"loss": 0.0095,
|
| 25431 |
+
"step": 3616
|
| 25432 |
+
},
|
| 25433 |
+
{
|
| 25434 |
+
"epoch": 9.581456953642384,
|
| 25435 |
+
"grad_norm": 0.6541901230812073,
|
| 25436 |
+
"learning_rate": 2.0291777188328914e-05,
|
| 25437 |
+
"loss": 0.0136,
|
| 25438 |
+
"step": 3617
|
| 25439 |
+
},
|
| 25440 |
+
{
|
| 25441 |
+
"epoch": 9.584105960264901,
|
| 25442 |
+
"grad_norm": 0.14791490137577057,
|
| 25443 |
+
"learning_rate": 2.0159151193633952e-05,
|
| 25444 |
+
"loss": 0.0065,
|
| 25445 |
+
"step": 3618
|
| 25446 |
+
},
|
| 25447 |
+
{
|
| 25448 |
+
"epoch": 9.586754966887417,
|
| 25449 |
+
"grad_norm": 0.2216922789812088,
|
| 25450 |
+
"learning_rate": 2.0026525198938993e-05,
|
| 25451 |
+
"loss": 0.0071,
|
| 25452 |
+
"step": 3619
|
| 25453 |
+
},
|
| 25454 |
+
{
|
| 25455 |
+
"epoch": 9.589403973509933,
|
| 25456 |
+
"grad_norm": 0.08360443264245987,
|
| 25457 |
+
"learning_rate": 1.989389920424403e-05,
|
| 25458 |
+
"loss": 0.0057,
|
| 25459 |
+
"step": 3620
|
| 25460 |
+
},
|
| 25461 |
+
{
|
| 25462 |
+
"epoch": 9.59205298013245,
|
| 25463 |
+
"grad_norm": 0.21307526528835297,
|
| 25464 |
+
"learning_rate": 1.9761273209549073e-05,
|
| 25465 |
+
"loss": 0.0078,
|
| 25466 |
+
"step": 3621
|
| 25467 |
+
},
|
| 25468 |
+
{
|
| 25469 |
+
"epoch": 9.594701986754966,
|
| 25470 |
+
"grad_norm": 0.16253237426280975,
|
| 25471 |
+
"learning_rate": 1.9628647214854114e-05,
|
| 25472 |
+
"loss": 0.0067,
|
| 25473 |
+
"step": 3622
|
| 25474 |
+
},
|
| 25475 |
+
{
|
| 25476 |
+
"epoch": 9.597350993377484,
|
| 25477 |
+
"grad_norm": 0.07736147940158844,
|
| 25478 |
+
"learning_rate": 1.9496021220159152e-05,
|
| 25479 |
+
"loss": 0.0049,
|
| 25480 |
+
"step": 3623
|
| 25481 |
+
},
|
| 25482 |
+
{
|
| 25483 |
+
"epoch": 9.6,
|
| 25484 |
+
"grad_norm": 0.18379540741443634,
|
| 25485 |
+
"learning_rate": 1.936339522546419e-05,
|
| 25486 |
+
"loss": 0.0089,
|
| 25487 |
+
"step": 3624
|
| 25488 |
+
},
|
| 25489 |
+
{
|
| 25490 |
+
"epoch": 9.602649006622517,
|
| 25491 |
+
"grad_norm": 0.07767052948474884,
|
| 25492 |
+
"learning_rate": 1.923076923076923e-05,
|
| 25493 |
+
"loss": 0.0053,
|
| 25494 |
+
"step": 3625
|
| 25495 |
+
},
|
| 25496 |
+
{
|
| 25497 |
+
"epoch": 9.605298013245033,
|
| 25498 |
+
"grad_norm": 0.15335260331630707,
|
| 25499 |
+
"learning_rate": 1.909814323607427e-05,
|
| 25500 |
+
"loss": 0.0066,
|
| 25501 |
+
"step": 3626
|
| 25502 |
+
},
|
| 25503 |
+
{
|
| 25504 |
+
"epoch": 9.607947019867549,
|
| 25505 |
+
"grad_norm": 0.1968509703874588,
|
| 25506 |
+
"learning_rate": 1.896551724137931e-05,
|
| 25507 |
+
"loss": 0.0076,
|
| 25508 |
+
"step": 3627
|
| 25509 |
+
},
|
| 25510 |
+
{
|
| 25511 |
+
"epoch": 9.610596026490066,
|
| 25512 |
+
"grad_norm": 0.06846405565738678,
|
| 25513 |
+
"learning_rate": 1.8832891246684352e-05,
|
| 25514 |
+
"loss": 0.0048,
|
| 25515 |
+
"step": 3628
|
| 25516 |
+
},
|
| 25517 |
+
{
|
| 25518 |
+
"epoch": 9.613245033112582,
|
| 25519 |
+
"grad_norm": 0.12656621634960175,
|
| 25520 |
+
"learning_rate": 1.870026525198939e-05,
|
| 25521 |
+
"loss": 0.0079,
|
| 25522 |
+
"step": 3629
|
| 25523 |
+
},
|
| 25524 |
+
{
|
| 25525 |
+
"epoch": 9.6158940397351,
|
| 25526 |
+
"grad_norm": 0.0959506556391716,
|
| 25527 |
+
"learning_rate": 1.8567639257294428e-05,
|
| 25528 |
+
"loss": 0.0058,
|
| 25529 |
+
"step": 3630
|
| 25530 |
+
},
|
| 25531 |
+
{
|
| 25532 |
+
"epoch": 9.618543046357615,
|
| 25533 |
+
"grad_norm": 0.2598751187324524,
|
| 25534 |
+
"learning_rate": 1.843501326259947e-05,
|
| 25535 |
+
"loss": 0.0072,
|
| 25536 |
+
"step": 3631
|
| 25537 |
+
},
|
| 25538 |
+
{
|
| 25539 |
+
"epoch": 9.621192052980133,
|
| 25540 |
+
"grad_norm": 0.16255703568458557,
|
| 25541 |
+
"learning_rate": 1.830238726790451e-05,
|
| 25542 |
+
"loss": 0.0068,
|
| 25543 |
+
"step": 3632
|
| 25544 |
+
},
|
| 25545 |
+
{
|
| 25546 |
+
"epoch": 9.623841059602649,
|
| 25547 |
+
"grad_norm": 0.14356957376003265,
|
| 25548 |
+
"learning_rate": 1.816976127320955e-05,
|
| 25549 |
+
"loss": 0.0071,
|
| 25550 |
+
"step": 3633
|
| 25551 |
+
},
|
| 25552 |
+
{
|
| 25553 |
+
"epoch": 9.626490066225166,
|
| 25554 |
+
"grad_norm": 0.11708579212427139,
|
| 25555 |
+
"learning_rate": 1.803713527851459e-05,
|
| 25556 |
+
"loss": 0.0067,
|
| 25557 |
+
"step": 3634
|
| 25558 |
+
},
|
| 25559 |
+
{
|
| 25560 |
+
"epoch": 9.629139072847682,
|
| 25561 |
+
"grad_norm": 0.11293361335992813,
|
| 25562 |
+
"learning_rate": 1.790450928381963e-05,
|
| 25563 |
+
"loss": 0.0063,
|
| 25564 |
+
"step": 3635
|
| 25565 |
+
},
|
| 25566 |
+
{
|
| 25567 |
+
"epoch": 9.631788079470198,
|
| 25568 |
+
"grad_norm": 0.38313353061676025,
|
| 25569 |
+
"learning_rate": 1.7771883289124666e-05,
|
| 25570 |
+
"loss": 0.0118,
|
| 25571 |
+
"step": 3636
|
| 25572 |
+
},
|
| 25573 |
+
{
|
| 25574 |
+
"epoch": 9.634437086092715,
|
| 25575 |
+
"grad_norm": 0.13401620090007782,
|
| 25576 |
+
"learning_rate": 1.7639257294429708e-05,
|
| 25577 |
+
"loss": 0.0067,
|
| 25578 |
+
"step": 3637
|
| 25579 |
+
},
|
| 25580 |
+
{
|
| 25581 |
+
"epoch": 9.637086092715231,
|
| 25582 |
+
"grad_norm": 0.23206515610218048,
|
| 25583 |
+
"learning_rate": 1.750663129973475e-05,
|
| 25584 |
+
"loss": 0.0069,
|
| 25585 |
+
"step": 3638
|
| 25586 |
+
},
|
| 25587 |
+
{
|
| 25588 |
+
"epoch": 9.639735099337749,
|
| 25589 |
+
"grad_norm": 0.19798971712589264,
|
| 25590 |
+
"learning_rate": 1.7374005305039787e-05,
|
| 25591 |
+
"loss": 0.0062,
|
| 25592 |
+
"step": 3639
|
| 25593 |
+
},
|
| 25594 |
+
{
|
| 25595 |
+
"epoch": 9.642384105960264,
|
| 25596 |
+
"grad_norm": 0.09165431559085846,
|
| 25597 |
+
"learning_rate": 1.7241379310344828e-05,
|
| 25598 |
+
"loss": 0.0058,
|
| 25599 |
+
"step": 3640
|
| 25600 |
+
},
|
| 25601 |
+
{
|
| 25602 |
+
"epoch": 9.645033112582782,
|
| 25603 |
+
"grad_norm": 0.616093099117279,
|
| 25604 |
+
"learning_rate": 1.710875331564987e-05,
|
| 25605 |
+
"loss": 0.0127,
|
| 25606 |
+
"step": 3641
|
| 25607 |
+
},
|
| 25608 |
+
{
|
| 25609 |
+
"epoch": 9.647682119205298,
|
| 25610 |
+
"grad_norm": 1.1477009057998657,
|
| 25611 |
+
"learning_rate": 1.6976127320954904e-05,
|
| 25612 |
+
"loss": 0.0116,
|
| 25613 |
+
"step": 3642
|
| 25614 |
+
},
|
| 25615 |
+
{
|
| 25616 |
+
"epoch": 9.650331125827815,
|
| 25617 |
+
"grad_norm": 0.13470661640167236,
|
| 25618 |
+
"learning_rate": 1.6843501326259946e-05,
|
| 25619 |
+
"loss": 0.0062,
|
| 25620 |
+
"step": 3643
|
| 25621 |
+
},
|
| 25622 |
+
{
|
| 25623 |
+
"epoch": 9.652980132450331,
|
| 25624 |
+
"grad_norm": 0.11247550696134567,
|
| 25625 |
+
"learning_rate": 1.6710875331564987e-05,
|
| 25626 |
+
"loss": 0.0052,
|
| 25627 |
+
"step": 3644
|
| 25628 |
+
},
|
| 25629 |
+
{
|
| 25630 |
+
"epoch": 9.655629139072847,
|
| 25631 |
+
"grad_norm": 0.5074613094329834,
|
| 25632 |
+
"learning_rate": 1.657824933687003e-05,
|
| 25633 |
+
"loss": 0.0084,
|
| 25634 |
+
"step": 3645
|
| 25635 |
+
},
|
| 25636 |
+
{
|
| 25637 |
+
"epoch": 9.658278145695364,
|
| 25638 |
+
"grad_norm": 0.19383744895458221,
|
| 25639 |
+
"learning_rate": 1.6445623342175066e-05,
|
| 25640 |
+
"loss": 0.0078,
|
| 25641 |
+
"step": 3646
|
| 25642 |
+
},
|
| 25643 |
+
{
|
| 25644 |
+
"epoch": 9.66092715231788,
|
| 25645 |
+
"grad_norm": 1.3290162086486816,
|
| 25646 |
+
"learning_rate": 1.6312997347480108e-05,
|
| 25647 |
+
"loss": 0.0304,
|
| 25648 |
+
"step": 3647
|
| 25649 |
+
},
|
| 25650 |
+
{
|
| 25651 |
+
"epoch": 9.663576158940398,
|
| 25652 |
+
"grad_norm": 0.3068723678588867,
|
| 25653 |
+
"learning_rate": 1.618037135278515e-05,
|
| 25654 |
+
"loss": 0.0083,
|
| 25655 |
+
"step": 3648
|
| 25656 |
+
},
|
| 25657 |
+
{
|
| 25658 |
+
"epoch": 9.666225165562913,
|
| 25659 |
+
"grad_norm": 0.16725531220436096,
|
| 25660 |
+
"learning_rate": 1.6047745358090184e-05,
|
| 25661 |
+
"loss": 0.0082,
|
| 25662 |
+
"step": 3649
|
| 25663 |
+
},
|
| 25664 |
+
{
|
| 25665 |
+
"epoch": 9.668874172185431,
|
| 25666 |
+
"grad_norm": 0.09185387194156647,
|
| 25667 |
+
"learning_rate": 1.5915119363395225e-05,
|
| 25668 |
+
"loss": 0.0058,
|
| 25669 |
+
"step": 3650
|
| 25670 |
+
},
|
| 25671 |
+
{
|
| 25672 |
+
"epoch": 9.671523178807947,
|
| 25673 |
+
"grad_norm": 0.20803192257881165,
|
| 25674 |
+
"learning_rate": 1.5782493368700266e-05,
|
| 25675 |
+
"loss": 0.0062,
|
| 25676 |
+
"step": 3651
|
| 25677 |
+
},
|
| 25678 |
+
{
|
| 25679 |
+
"epoch": 9.674172185430464,
|
| 25680 |
+
"grad_norm": 0.11356912553310394,
|
| 25681 |
+
"learning_rate": 1.5649867374005304e-05,
|
| 25682 |
+
"loss": 0.007,
|
| 25683 |
+
"step": 3652
|
| 25684 |
+
},
|
| 25685 |
+
{
|
| 25686 |
+
"epoch": 9.67682119205298,
|
| 25687 |
+
"grad_norm": 0.12919805943965912,
|
| 25688 |
+
"learning_rate": 1.5517241379310346e-05,
|
| 25689 |
+
"loss": 0.0063,
|
| 25690 |
+
"step": 3653
|
| 25691 |
+
},
|
| 25692 |
+
{
|
| 25693 |
+
"epoch": 9.679470198675496,
|
| 25694 |
+
"grad_norm": 0.08865955471992493,
|
| 25695 |
+
"learning_rate": 1.5384615384615387e-05,
|
| 25696 |
+
"loss": 0.0057,
|
| 25697 |
+
"step": 3654
|
| 25698 |
+
},
|
| 25699 |
+
{
|
| 25700 |
+
"epoch": 9.682119205298013,
|
| 25701 |
+
"grad_norm": 0.21951282024383545,
|
| 25702 |
+
"learning_rate": 1.5251989389920425e-05,
|
| 25703 |
+
"loss": 0.0076,
|
| 25704 |
+
"step": 3655
|
| 25705 |
+
},
|
| 25706 |
+
{
|
| 25707 |
+
"epoch": 9.68476821192053,
|
| 25708 |
+
"grad_norm": 0.2228957563638687,
|
| 25709 |
+
"learning_rate": 1.5119363395225465e-05,
|
| 25710 |
+
"loss": 0.0071,
|
| 25711 |
+
"step": 3656
|
| 25712 |
+
},
|
| 25713 |
+
{
|
| 25714 |
+
"epoch": 9.687417218543047,
|
| 25715 |
+
"grad_norm": 0.49982765316963196,
|
| 25716 |
+
"learning_rate": 1.4986737400530505e-05,
|
| 25717 |
+
"loss": 0.0086,
|
| 25718 |
+
"step": 3657
|
| 25719 |
+
},
|
| 25720 |
+
{
|
| 25721 |
+
"epoch": 9.690066225165562,
|
| 25722 |
+
"grad_norm": 0.17828427255153656,
|
| 25723 |
+
"learning_rate": 1.4854111405835544e-05,
|
| 25724 |
+
"loss": 0.007,
|
| 25725 |
+
"step": 3658
|
| 25726 |
+
},
|
| 25727 |
+
{
|
| 25728 |
+
"epoch": 9.69271523178808,
|
| 25729 |
+
"grad_norm": 0.07209718227386475,
|
| 25730 |
+
"learning_rate": 1.4721485411140584e-05,
|
| 25731 |
+
"loss": 0.0051,
|
| 25732 |
+
"step": 3659
|
| 25733 |
+
},
|
| 25734 |
+
{
|
| 25735 |
+
"epoch": 9.695364238410596,
|
| 25736 |
+
"grad_norm": 0.9999616146087646,
|
| 25737 |
+
"learning_rate": 1.4588859416445624e-05,
|
| 25738 |
+
"loss": 0.0124,
|
| 25739 |
+
"step": 3660
|
| 25740 |
+
},
|
| 25741 |
+
{
|
| 25742 |
+
"epoch": 9.698013245033113,
|
| 25743 |
+
"grad_norm": 0.4745708703994751,
|
| 25744 |
+
"learning_rate": 1.4456233421750663e-05,
|
| 25745 |
+
"loss": 0.0128,
|
| 25746 |
+
"step": 3661
|
| 25747 |
+
},
|
| 25748 |
+
{
|
| 25749 |
+
"epoch": 9.70066225165563,
|
| 25750 |
+
"grad_norm": 0.05109310522675514,
|
| 25751 |
+
"learning_rate": 1.4323607427055705e-05,
|
| 25752 |
+
"loss": 0.0033,
|
| 25753 |
+
"step": 3662
|
| 25754 |
+
},
|
| 25755 |
+
{
|
| 25756 |
+
"epoch": 9.703311258278145,
|
| 25757 |
+
"grad_norm": 0.05594147741794586,
|
| 25758 |
+
"learning_rate": 1.4190981432360743e-05,
|
| 25759 |
+
"loss": 0.0042,
|
| 25760 |
+
"step": 3663
|
| 25761 |
+
},
|
| 25762 |
+
{
|
| 25763 |
+
"epoch": 9.705960264900662,
|
| 25764 |
+
"grad_norm": 0.11659505218267441,
|
| 25765 |
+
"learning_rate": 1.4058355437665782e-05,
|
| 25766 |
+
"loss": 0.0064,
|
| 25767 |
+
"step": 3664
|
| 25768 |
+
},
|
| 25769 |
+
{
|
| 25770 |
+
"epoch": 9.708609271523178,
|
| 25771 |
+
"grad_norm": 0.1137813925743103,
|
| 25772 |
+
"learning_rate": 1.3925729442970824e-05,
|
| 25773 |
+
"loss": 0.0051,
|
| 25774 |
+
"step": 3665
|
| 25775 |
+
},
|
| 25776 |
+
{
|
| 25777 |
+
"epoch": 9.711258278145696,
|
| 25778 |
+
"grad_norm": 0.48468494415283203,
|
| 25779 |
+
"learning_rate": 1.3793103448275862e-05,
|
| 25780 |
+
"loss": 0.0081,
|
| 25781 |
+
"step": 3666
|
| 25782 |
+
},
|
| 25783 |
+
{
|
| 25784 |
+
"epoch": 9.713907284768212,
|
| 25785 |
+
"grad_norm": 0.08497241139411926,
|
| 25786 |
+
"learning_rate": 1.3660477453580901e-05,
|
| 25787 |
+
"loss": 0.0047,
|
| 25788 |
+
"step": 3667
|
| 25789 |
+
},
|
| 25790 |
+
{
|
| 25791 |
+
"epoch": 9.716556291390729,
|
| 25792 |
+
"grad_norm": 0.923937976360321,
|
| 25793 |
+
"learning_rate": 1.3527851458885943e-05,
|
| 25794 |
+
"loss": 0.0116,
|
| 25795 |
+
"step": 3668
|
| 25796 |
+
},
|
| 25797 |
+
{
|
| 25798 |
+
"epoch": 9.719205298013245,
|
| 25799 |
+
"grad_norm": 0.09511914104223251,
|
| 25800 |
+
"learning_rate": 1.339522546419098e-05,
|
| 25801 |
+
"loss": 0.0051,
|
| 25802 |
+
"step": 3669
|
| 25803 |
+
},
|
| 25804 |
+
{
|
| 25805 |
+
"epoch": 9.721854304635762,
|
| 25806 |
+
"grad_norm": 0.1370173543691635,
|
| 25807 |
+
"learning_rate": 1.3262599469496022e-05,
|
| 25808 |
+
"loss": 0.0069,
|
| 25809 |
+
"step": 3670
|
| 25810 |
+
},
|
| 25811 |
+
{
|
| 25812 |
+
"epoch": 9.724503311258278,
|
| 25813 |
+
"grad_norm": 0.7186203002929688,
|
| 25814 |
+
"learning_rate": 1.3129973474801062e-05,
|
| 25815 |
+
"loss": 0.0094,
|
| 25816 |
+
"step": 3671
|
| 25817 |
+
},
|
| 25818 |
+
{
|
| 25819 |
+
"epoch": 9.727152317880794,
|
| 25820 |
+
"grad_norm": 0.11266762018203735,
|
| 25821 |
+
"learning_rate": 1.29973474801061e-05,
|
| 25822 |
+
"loss": 0.0067,
|
| 25823 |
+
"step": 3672
|
| 25824 |
+
},
|
| 25825 |
+
{
|
| 25826 |
+
"epoch": 9.729801324503311,
|
| 25827 |
+
"grad_norm": 0.08314207941293716,
|
| 25828 |
+
"learning_rate": 1.2864721485411141e-05,
|
| 25829 |
+
"loss": 0.004,
|
| 25830 |
+
"step": 3673
|
| 25831 |
+
},
|
| 25832 |
+
{
|
| 25833 |
+
"epoch": 9.732450331125827,
|
| 25834 |
+
"grad_norm": 0.18078316748142242,
|
| 25835 |
+
"learning_rate": 1.273209549071618e-05,
|
| 25836 |
+
"loss": 0.0062,
|
| 25837 |
+
"step": 3674
|
| 25838 |
+
},
|
| 25839 |
+
{
|
| 25840 |
+
"epoch": 9.735099337748345,
|
| 25841 |
+
"grad_norm": 0.3765209913253784,
|
| 25842 |
+
"learning_rate": 1.259946949602122e-05,
|
| 25843 |
+
"loss": 0.0091,
|
| 25844 |
+
"step": 3675
|
| 25845 |
+
},
|
| 25846 |
+
{
|
| 25847 |
+
"epoch": 9.73774834437086,
|
| 25848 |
+
"grad_norm": 0.12933093309402466,
|
| 25849 |
+
"learning_rate": 1.246684350132626e-05,
|
| 25850 |
+
"loss": 0.007,
|
| 25851 |
+
"step": 3676
|
| 25852 |
+
},
|
| 25853 |
+
{
|
| 25854 |
+
"epoch": 9.740397350993378,
|
| 25855 |
+
"grad_norm": 0.08696829527616501,
|
| 25856 |
+
"learning_rate": 1.23342175066313e-05,
|
| 25857 |
+
"loss": 0.0056,
|
| 25858 |
+
"step": 3677
|
| 25859 |
+
},
|
| 25860 |
+
{
|
| 25861 |
+
"epoch": 9.743046357615894,
|
| 25862 |
+
"grad_norm": 0.11974237114191055,
|
| 25863 |
+
"learning_rate": 1.220159151193634e-05,
|
| 25864 |
+
"loss": 0.0066,
|
| 25865 |
+
"step": 3678
|
| 25866 |
+
},
|
| 25867 |
+
{
|
| 25868 |
+
"epoch": 9.745695364238411,
|
| 25869 |
+
"grad_norm": 0.0988985225558281,
|
| 25870 |
+
"learning_rate": 1.2068965517241379e-05,
|
| 25871 |
+
"loss": 0.0062,
|
| 25872 |
+
"step": 3679
|
| 25873 |
+
},
|
| 25874 |
+
{
|
| 25875 |
+
"epoch": 9.748344370860927,
|
| 25876 |
+
"grad_norm": 0.08422379940748215,
|
| 25877 |
+
"learning_rate": 1.1936339522546419e-05,
|
| 25878 |
+
"loss": 0.0056,
|
| 25879 |
+
"step": 3680
|
| 25880 |
+
},
|
| 25881 |
+
{
|
| 25882 |
+
"epoch": 9.750993377483443,
|
| 25883 |
+
"grad_norm": 0.10629460215568542,
|
| 25884 |
+
"learning_rate": 1.180371352785146e-05,
|
| 25885 |
+
"loss": 0.0063,
|
| 25886 |
+
"step": 3681
|
| 25887 |
+
},
|
| 25888 |
+
{
|
| 25889 |
+
"epoch": 9.75364238410596,
|
| 25890 |
+
"grad_norm": 0.08831437677145004,
|
| 25891 |
+
"learning_rate": 1.1671087533156498e-05,
|
| 25892 |
+
"loss": 0.0054,
|
| 25893 |
+
"step": 3682
|
| 25894 |
+
},
|
| 25895 |
+
{
|
| 25896 |
+
"epoch": 9.756291390728476,
|
| 25897 |
+
"grad_norm": 0.06336377561092377,
|
| 25898 |
+
"learning_rate": 1.153846153846154e-05,
|
| 25899 |
+
"loss": 0.0048,
|
| 25900 |
+
"step": 3683
|
| 25901 |
+
},
|
| 25902 |
+
{
|
| 25903 |
+
"epoch": 9.758940397350994,
|
| 25904 |
+
"grad_norm": 0.08550837635993958,
|
| 25905 |
+
"learning_rate": 1.140583554376658e-05,
|
| 25906 |
+
"loss": 0.0059,
|
| 25907 |
+
"step": 3684
|
| 25908 |
+
},
|
| 25909 |
+
{
|
| 25910 |
+
"epoch": 9.76158940397351,
|
| 25911 |
+
"grad_norm": 0.20795460045337677,
|
| 25912 |
+
"learning_rate": 1.1273209549071617e-05,
|
| 25913 |
+
"loss": 0.0083,
|
| 25914 |
+
"step": 3685
|
| 25915 |
+
},
|
| 25916 |
+
{
|
| 25917 |
+
"epoch": 9.764238410596027,
|
| 25918 |
+
"grad_norm": 0.06397215276956558,
|
| 25919 |
+
"learning_rate": 1.1140583554376659e-05,
|
| 25920 |
+
"loss": 0.0045,
|
| 25921 |
+
"step": 3686
|
| 25922 |
+
},
|
| 25923 |
+
{
|
| 25924 |
+
"epoch": 9.766887417218543,
|
| 25925 |
+
"grad_norm": 0.2030840516090393,
|
| 25926 |
+
"learning_rate": 1.1007957559681698e-05,
|
| 25927 |
+
"loss": 0.008,
|
| 25928 |
+
"step": 3687
|
| 25929 |
+
},
|
| 25930 |
+
{
|
| 25931 |
+
"epoch": 9.76953642384106,
|
| 25932 |
+
"grad_norm": 0.09845580160617828,
|
| 25933 |
+
"learning_rate": 1.0875331564986738e-05,
|
| 25934 |
+
"loss": 0.0056,
|
| 25935 |
+
"step": 3688
|
| 25936 |
+
},
|
| 25937 |
+
{
|
| 25938 |
+
"epoch": 9.772185430463576,
|
| 25939 |
+
"grad_norm": 0.0654703751206398,
|
| 25940 |
+
"learning_rate": 1.0742705570291778e-05,
|
| 25941 |
+
"loss": 0.0054,
|
| 25942 |
+
"step": 3689
|
| 25943 |
+
},
|
| 25944 |
+
{
|
| 25945 |
+
"epoch": 9.774834437086092,
|
| 25946 |
+
"grad_norm": 0.08135940134525299,
|
| 25947 |
+
"learning_rate": 1.0610079575596817e-05,
|
| 25948 |
+
"loss": 0.0048,
|
| 25949 |
+
"step": 3690
|
| 25950 |
+
},
|
| 25951 |
+
{
|
| 25952 |
+
"epoch": 9.77748344370861,
|
| 25953 |
+
"grad_norm": 0.13394837081432343,
|
| 25954 |
+
"learning_rate": 1.0477453580901857e-05,
|
| 25955 |
+
"loss": 0.0069,
|
| 25956 |
+
"step": 3691
|
| 25957 |
+
},
|
| 25958 |
+
{
|
| 25959 |
+
"epoch": 9.780132450331125,
|
| 25960 |
+
"grad_norm": 0.2942659258842468,
|
| 25961 |
+
"learning_rate": 1.0344827586206897e-05,
|
| 25962 |
+
"loss": 0.009,
|
| 25963 |
+
"step": 3692
|
| 25964 |
+
},
|
| 25965 |
+
{
|
| 25966 |
+
"epoch": 9.782781456953643,
|
| 25967 |
+
"grad_norm": 1.7077428102493286,
|
| 25968 |
+
"learning_rate": 1.0212201591511936e-05,
|
| 25969 |
+
"loss": 0.024,
|
| 25970 |
+
"step": 3693
|
| 25971 |
+
},
|
| 25972 |
+
{
|
| 25973 |
+
"epoch": 9.785430463576159,
|
| 25974 |
+
"grad_norm": 0.05311325937509537,
|
| 25975 |
+
"learning_rate": 1.0079575596816976e-05,
|
| 25976 |
+
"loss": 0.004,
|
| 25977 |
+
"step": 3694
|
| 25978 |
+
},
|
| 25979 |
+
{
|
| 25980 |
+
"epoch": 9.788079470198676,
|
| 25981 |
+
"grad_norm": 0.15471872687339783,
|
| 25982 |
+
"learning_rate": 9.946949602122016e-06,
|
| 25983 |
+
"loss": 0.0077,
|
| 25984 |
+
"step": 3695
|
| 25985 |
+
},
|
| 25986 |
+
{
|
| 25987 |
+
"epoch": 9.790728476821192,
|
| 25988 |
+
"grad_norm": 0.11402935534715652,
|
| 25989 |
+
"learning_rate": 9.814323607427057e-06,
|
| 25990 |
+
"loss": 0.0063,
|
| 25991 |
+
"step": 3696
|
| 25992 |
+
},
|
| 25993 |
+
{
|
| 25994 |
+
"epoch": 9.79337748344371,
|
| 25995 |
+
"grad_norm": 0.07030440121889114,
|
| 25996 |
+
"learning_rate": 9.681697612732095e-06,
|
| 25997 |
+
"loss": 0.0049,
|
| 25998 |
+
"step": 3697
|
| 25999 |
+
},
|
| 26000 |
+
{
|
| 26001 |
+
"epoch": 9.796026490066225,
|
| 26002 |
+
"grad_norm": 0.10101749002933502,
|
| 26003 |
+
"learning_rate": 9.549071618037135e-06,
|
| 26004 |
+
"loss": 0.0053,
|
| 26005 |
+
"step": 3698
|
| 26006 |
+
},
|
| 26007 |
+
{
|
| 26008 |
+
"epoch": 9.798675496688741,
|
| 26009 |
+
"grad_norm": 0.46287235617637634,
|
| 26010 |
+
"learning_rate": 9.416445623342176e-06,
|
| 26011 |
+
"loss": 0.0124,
|
| 26012 |
+
"step": 3699
|
| 26013 |
+
},
|
| 26014 |
+
{
|
| 26015 |
+
"epoch": 9.801324503311259,
|
| 26016 |
+
"grad_norm": 0.2550599277019501,
|
| 26017 |
+
"learning_rate": 9.283819628647214e-06,
|
| 26018 |
+
"loss": 0.0069,
|
| 26019 |
+
"step": 3700
|
| 26020 |
+
},
|
| 26021 |
+
{
|
| 26022 |
+
"epoch": 9.803973509933774,
|
| 26023 |
+
"grad_norm": 0.24267533421516418,
|
| 26024 |
+
"learning_rate": 9.151193633952255e-06,
|
| 26025 |
+
"loss": 0.0084,
|
| 26026 |
+
"step": 3701
|
| 26027 |
+
},
|
| 26028 |
+
{
|
| 26029 |
+
"epoch": 9.806622516556292,
|
| 26030 |
+
"grad_norm": 0.08424295485019684,
|
| 26031 |
+
"learning_rate": 9.018567639257295e-06,
|
| 26032 |
+
"loss": 0.0043,
|
| 26033 |
+
"step": 3702
|
| 26034 |
+
},
|
| 26035 |
+
{
|
| 26036 |
+
"epoch": 9.809271523178808,
|
| 26037 |
+
"grad_norm": 0.11411385238170624,
|
| 26038 |
+
"learning_rate": 8.885941644562333e-06,
|
| 26039 |
+
"loss": 0.0047,
|
| 26040 |
+
"step": 3703
|
| 26041 |
+
},
|
| 26042 |
+
{
|
| 26043 |
+
"epoch": 9.811920529801325,
|
| 26044 |
+
"grad_norm": 0.08485867083072662,
|
| 26045 |
+
"learning_rate": 8.753315649867374e-06,
|
| 26046 |
+
"loss": 0.0049,
|
| 26047 |
+
"step": 3704
|
| 26048 |
+
},
|
| 26049 |
+
{
|
| 26050 |
+
"epoch": 9.814569536423841,
|
| 26051 |
+
"grad_norm": 0.3491503596305847,
|
| 26052 |
+
"learning_rate": 8.620689655172414e-06,
|
| 26053 |
+
"loss": 0.0098,
|
| 26054 |
+
"step": 3705
|
| 26055 |
+
},
|
| 26056 |
+
{
|
| 26057 |
+
"epoch": 9.817218543046359,
|
| 26058 |
+
"grad_norm": 0.0885666087269783,
|
| 26059 |
+
"learning_rate": 8.488063660477452e-06,
|
| 26060 |
+
"loss": 0.0056,
|
| 26061 |
+
"step": 3706
|
| 26062 |
+
},
|
| 26063 |
+
{
|
| 26064 |
+
"epoch": 9.819867549668874,
|
| 26065 |
+
"grad_norm": 0.6458684206008911,
|
| 26066 |
+
"learning_rate": 8.355437665782494e-06,
|
| 26067 |
+
"loss": 0.0129,
|
| 26068 |
+
"step": 3707
|
| 26069 |
+
},
|
| 26070 |
+
{
|
| 26071 |
+
"epoch": 9.82251655629139,
|
| 26072 |
+
"grad_norm": 0.10896582156419754,
|
| 26073 |
+
"learning_rate": 8.222811671087533e-06,
|
| 26074 |
+
"loss": 0.0058,
|
| 26075 |
+
"step": 3708
|
| 26076 |
+
},
|
| 26077 |
+
{
|
| 26078 |
+
"epoch": 9.825165562913908,
|
| 26079 |
+
"grad_norm": 0.06155227869749069,
|
| 26080 |
+
"learning_rate": 8.090185676392575e-06,
|
| 26081 |
+
"loss": 0.0041,
|
| 26082 |
+
"step": 3709
|
| 26083 |
+
},
|
| 26084 |
+
{
|
| 26085 |
+
"epoch": 9.827814569536423,
|
| 26086 |
+
"grad_norm": 0.12445490807294846,
|
| 26087 |
+
"learning_rate": 7.957559681697613e-06,
|
| 26088 |
+
"loss": 0.0076,
|
| 26089 |
+
"step": 3710
|
| 26090 |
+
},
|
| 26091 |
+
{
|
| 26092 |
+
"epoch": 9.830463576158941,
|
| 26093 |
+
"grad_norm": 0.1290195882320404,
|
| 26094 |
+
"learning_rate": 7.824933687002652e-06,
|
| 26095 |
+
"loss": 0.0052,
|
| 26096 |
+
"step": 3711
|
| 26097 |
+
},
|
| 26098 |
+
{
|
| 26099 |
+
"epoch": 9.833112582781457,
|
| 26100 |
+
"grad_norm": 0.26849737763404846,
|
| 26101 |
+
"learning_rate": 7.692307692307694e-06,
|
| 26102 |
+
"loss": 0.0091,
|
| 26103 |
+
"step": 3712
|
| 26104 |
+
},
|
| 26105 |
+
{
|
| 26106 |
+
"epoch": 9.835761589403974,
|
| 26107 |
+
"grad_norm": 0.08209025114774704,
|
| 26108 |
+
"learning_rate": 7.559681697612732e-06,
|
| 26109 |
+
"loss": 0.0054,
|
| 26110 |
+
"step": 3713
|
| 26111 |
+
},
|
| 26112 |
+
{
|
| 26113 |
+
"epoch": 9.83841059602649,
|
| 26114 |
+
"grad_norm": 0.06356347352266312,
|
| 26115 |
+
"learning_rate": 7.427055702917772e-06,
|
| 26116 |
+
"loss": 0.0038,
|
| 26117 |
+
"step": 3714
|
| 26118 |
+
},
|
| 26119 |
+
{
|
| 26120 |
+
"epoch": 9.841059602649006,
|
| 26121 |
+
"grad_norm": 0.16247440874576569,
|
| 26122 |
+
"learning_rate": 7.294429708222812e-06,
|
| 26123 |
+
"loss": 0.0084,
|
| 26124 |
+
"step": 3715
|
| 26125 |
+
},
|
| 26126 |
+
{
|
| 26127 |
+
"epoch": 9.843708609271523,
|
| 26128 |
+
"grad_norm": 0.175484761595726,
|
| 26129 |
+
"learning_rate": 7.161803713527852e-06,
|
| 26130 |
+
"loss": 0.0071,
|
| 26131 |
+
"step": 3716
|
| 26132 |
+
},
|
| 26133 |
+
{
|
| 26134 |
+
"epoch": 9.84635761589404,
|
| 26135 |
+
"grad_norm": 0.16009952127933502,
|
| 26136 |
+
"learning_rate": 7.029177718832891e-06,
|
| 26137 |
+
"loss": 0.0054,
|
| 26138 |
+
"step": 3717
|
| 26139 |
+
},
|
| 26140 |
+
{
|
| 26141 |
+
"epoch": 9.849006622516557,
|
| 26142 |
+
"grad_norm": 0.824354887008667,
|
| 26143 |
+
"learning_rate": 6.896551724137931e-06,
|
| 26144 |
+
"loss": 0.0225,
|
| 26145 |
+
"step": 3718
|
| 26146 |
+
},
|
| 26147 |
+
{
|
| 26148 |
+
"epoch": 9.851655629139072,
|
| 26149 |
+
"grad_norm": 1.2063053846359253,
|
| 26150 |
+
"learning_rate": 6.763925729442971e-06,
|
| 26151 |
+
"loss": 0.0186,
|
| 26152 |
+
"step": 3719
|
| 26153 |
+
},
|
| 26154 |
+
{
|
| 26155 |
+
"epoch": 9.85430463576159,
|
| 26156 |
+
"grad_norm": 0.1102658361196518,
|
| 26157 |
+
"learning_rate": 6.631299734748011e-06,
|
| 26158 |
+
"loss": 0.0068,
|
| 26159 |
+
"step": 3720
|
| 26160 |
+
},
|
| 26161 |
+
{
|
| 26162 |
+
"epoch": 9.856953642384106,
|
| 26163 |
+
"grad_norm": 0.6607574820518494,
|
| 26164 |
+
"learning_rate": 6.49867374005305e-06,
|
| 26165 |
+
"loss": 0.0163,
|
| 26166 |
+
"step": 3721
|
| 26167 |
+
},
|
| 26168 |
+
{
|
| 26169 |
+
"epoch": 9.859602649006623,
|
| 26170 |
+
"grad_norm": 0.05733625218272209,
|
| 26171 |
+
"learning_rate": 6.36604774535809e-06,
|
| 26172 |
+
"loss": 0.0043,
|
| 26173 |
+
"step": 3722
|
| 26174 |
+
},
|
| 26175 |
+
{
|
| 26176 |
+
"epoch": 9.862251655629139,
|
| 26177 |
+
"grad_norm": 1.854579210281372,
|
| 26178 |
+
"learning_rate": 6.23342175066313e-06,
|
| 26179 |
+
"loss": 0.0532,
|
| 26180 |
+
"step": 3723
|
| 26181 |
+
},
|
| 26182 |
+
{
|
| 26183 |
+
"epoch": 9.864900662251655,
|
| 26184 |
+
"grad_norm": 0.4233485460281372,
|
| 26185 |
+
"learning_rate": 6.10079575596817e-06,
|
| 26186 |
+
"loss": 0.0096,
|
| 26187 |
+
"step": 3724
|
| 26188 |
+
},
|
| 26189 |
+
{
|
| 26190 |
+
"epoch": 9.867549668874172,
|
| 26191 |
+
"grad_norm": 0.20195017755031586,
|
| 26192 |
+
"learning_rate": 5.968169761273209e-06,
|
| 26193 |
+
"loss": 0.0076,
|
| 26194 |
+
"step": 3725
|
| 26195 |
+
},
|
| 26196 |
+
{
|
| 26197 |
+
"epoch": 9.870198675496688,
|
| 26198 |
+
"grad_norm": 0.07918886095285416,
|
| 26199 |
+
"learning_rate": 5.835543766578249e-06,
|
| 26200 |
+
"loss": 0.0055,
|
| 26201 |
+
"step": 3726
|
| 26202 |
+
},
|
| 26203 |
+
{
|
| 26204 |
+
"epoch": 9.872847682119206,
|
| 26205 |
+
"grad_norm": 0.13845008611679077,
|
| 26206 |
+
"learning_rate": 5.70291777188329e-06,
|
| 26207 |
+
"loss": 0.0085,
|
| 26208 |
+
"step": 3727
|
| 26209 |
+
},
|
| 26210 |
+
{
|
| 26211 |
+
"epoch": 9.875496688741721,
|
| 26212 |
+
"grad_norm": 3.525977611541748,
|
| 26213 |
+
"learning_rate": 5.570291777188329e-06,
|
| 26214 |
+
"loss": 0.074,
|
| 26215 |
+
"step": 3728
|
| 26216 |
+
},
|
| 26217 |
+
{
|
| 26218 |
+
"epoch": 9.878145695364239,
|
| 26219 |
+
"grad_norm": 1.5265491008758545,
|
| 26220 |
+
"learning_rate": 5.437665782493369e-06,
|
| 26221 |
+
"loss": 0.048,
|
| 26222 |
+
"step": 3729
|
| 26223 |
+
},
|
| 26224 |
+
{
|
| 26225 |
+
"epoch": 9.880794701986755,
|
| 26226 |
+
"grad_norm": 1.900270700454712,
|
| 26227 |
+
"learning_rate": 5.305039787798409e-06,
|
| 26228 |
+
"loss": 0.0921,
|
| 26229 |
+
"step": 3730
|
| 26230 |
+
},
|
| 26231 |
+
{
|
| 26232 |
+
"epoch": 9.883443708609272,
|
| 26233 |
+
"grad_norm": 0.14628323912620544,
|
| 26234 |
+
"learning_rate": 5.172413793103448e-06,
|
| 26235 |
+
"loss": 0.0082,
|
| 26236 |
+
"step": 3731
|
| 26237 |
+
},
|
| 26238 |
+
{
|
| 26239 |
+
"epoch": 9.886092715231788,
|
| 26240 |
+
"grad_norm": 0.0840618833899498,
|
| 26241 |
+
"learning_rate": 5.039787798408488e-06,
|
| 26242 |
+
"loss": 0.005,
|
| 26243 |
+
"step": 3732
|
| 26244 |
+
},
|
| 26245 |
+
{
|
| 26246 |
+
"epoch": 9.888741721854304,
|
| 26247 |
+
"grad_norm": 0.0762125700712204,
|
| 26248 |
+
"learning_rate": 4.9071618037135285e-06,
|
| 26249 |
+
"loss": 0.0047,
|
| 26250 |
+
"step": 3733
|
| 26251 |
+
},
|
| 26252 |
+
{
|
| 26253 |
+
"epoch": 9.891390728476821,
|
| 26254 |
+
"grad_norm": 0.11019699275493622,
|
| 26255 |
+
"learning_rate": 4.774535809018567e-06,
|
| 26256 |
+
"loss": 0.0062,
|
| 26257 |
+
"step": 3734
|
| 26258 |
+
},
|
| 26259 |
+
{
|
| 26260 |
+
"epoch": 9.894039735099337,
|
| 26261 |
+
"grad_norm": 0.12956224381923676,
|
| 26262 |
+
"learning_rate": 4.641909814323607e-06,
|
| 26263 |
+
"loss": 0.0065,
|
| 26264 |
+
"step": 3735
|
| 26265 |
+
},
|
| 26266 |
+
{
|
| 26267 |
+
"epoch": 9.896688741721855,
|
| 26268 |
+
"grad_norm": 0.14278213679790497,
|
| 26269 |
+
"learning_rate": 4.5092838196286476e-06,
|
| 26270 |
+
"loss": 0.01,
|
| 26271 |
+
"step": 3736
|
| 26272 |
+
},
|
| 26273 |
+
{
|
| 26274 |
+
"epoch": 9.89933774834437,
|
| 26275 |
+
"grad_norm": 0.14741727709770203,
|
| 26276 |
+
"learning_rate": 4.376657824933687e-06,
|
| 26277 |
+
"loss": 0.0059,
|
| 26278 |
+
"step": 3737
|
| 26279 |
+
},
|
| 26280 |
+
{
|
| 26281 |
+
"epoch": 9.901986754966888,
|
| 26282 |
+
"grad_norm": 0.11628686636686325,
|
| 26283 |
+
"learning_rate": 4.244031830238726e-06,
|
| 26284 |
+
"loss": 0.007,
|
| 26285 |
+
"step": 3738
|
| 26286 |
+
},
|
| 26287 |
+
{
|
| 26288 |
+
"epoch": 9.904635761589404,
|
| 26289 |
+
"grad_norm": 0.080344058573246,
|
| 26290 |
+
"learning_rate": 4.111405835543767e-06,
|
| 26291 |
+
"loss": 0.0059,
|
| 26292 |
+
"step": 3739
|
| 26293 |
+
},
|
| 26294 |
+
{
|
| 26295 |
+
"epoch": 9.90728476821192,
|
| 26296 |
+
"grad_norm": 0.18567059934139252,
|
| 26297 |
+
"learning_rate": 3.978779840848806e-06,
|
| 26298 |
+
"loss": 0.0078,
|
| 26299 |
+
"step": 3740
|
| 26300 |
+
},
|
| 26301 |
+
{
|
| 26302 |
+
"epoch": 9.909933774834437,
|
| 26303 |
+
"grad_norm": 0.6573330163955688,
|
| 26304 |
+
"learning_rate": 3.846153846153847e-06,
|
| 26305 |
+
"loss": 0.0089,
|
| 26306 |
+
"step": 3741
|
| 26307 |
+
},
|
| 26308 |
+
{
|
| 26309 |
+
"epoch": 9.912582781456953,
|
| 26310 |
+
"grad_norm": 0.0857321098446846,
|
| 26311 |
+
"learning_rate": 3.713527851458886e-06,
|
| 26312 |
+
"loss": 0.0054,
|
| 26313 |
+
"step": 3742
|
| 26314 |
+
},
|
| 26315 |
+
{
|
| 26316 |
+
"epoch": 9.91523178807947,
|
| 26317 |
+
"grad_norm": 0.16797839105129242,
|
| 26318 |
+
"learning_rate": 3.580901856763926e-06,
|
| 26319 |
+
"loss": 0.0093,
|
| 26320 |
+
"step": 3743
|
| 26321 |
+
},
|
| 26322 |
+
{
|
| 26323 |
+
"epoch": 9.917880794701986,
|
| 26324 |
+
"grad_norm": 0.06998272985219955,
|
| 26325 |
+
"learning_rate": 3.4482758620689654e-06,
|
| 26326 |
+
"loss": 0.0047,
|
| 26327 |
+
"step": 3744
|
| 26328 |
+
},
|
| 26329 |
+
{
|
| 26330 |
+
"epoch": 9.920529801324504,
|
| 26331 |
+
"grad_norm": 0.0836014598608017,
|
| 26332 |
+
"learning_rate": 3.3156498673740055e-06,
|
| 26333 |
+
"loss": 0.0046,
|
| 26334 |
+
"step": 3745
|
| 26335 |
+
},
|
| 26336 |
+
{
|
| 26337 |
+
"epoch": 9.92317880794702,
|
| 26338 |
+
"grad_norm": 0.141579270362854,
|
| 26339 |
+
"learning_rate": 3.183023872679045e-06,
|
| 26340 |
+
"loss": 0.0083,
|
| 26341 |
+
"step": 3746
|
| 26342 |
+
},
|
| 26343 |
+
{
|
| 26344 |
+
"epoch": 9.925827814569537,
|
| 26345 |
+
"grad_norm": 0.0892159640789032,
|
| 26346 |
+
"learning_rate": 3.050397877984085e-06,
|
| 26347 |
+
"loss": 0.0048,
|
| 26348 |
+
"step": 3747
|
| 26349 |
+
},
|
| 26350 |
+
{
|
| 26351 |
+
"epoch": 9.928476821192053,
|
| 26352 |
+
"grad_norm": 0.09989528357982635,
|
| 26353 |
+
"learning_rate": 2.9177718832891245e-06,
|
| 26354 |
+
"loss": 0.005,
|
| 26355 |
+
"step": 3748
|
| 26356 |
+
},
|
| 26357 |
+
{
|
| 26358 |
+
"epoch": 9.931125827814569,
|
| 26359 |
+
"grad_norm": 0.13360846042633057,
|
| 26360 |
+
"learning_rate": 2.7851458885941646e-06,
|
| 26361 |
+
"loss": 0.0064,
|
| 26362 |
+
"step": 3749
|
| 26363 |
+
},
|
| 26364 |
+
{
|
| 26365 |
+
"epoch": 9.933774834437086,
|
| 26366 |
+
"grad_norm": 0.19383981823921204,
|
| 26367 |
+
"learning_rate": 2.6525198938992043e-06,
|
| 26368 |
+
"loss": 0.007,
|
| 26369 |
+
"step": 3750
|
| 26370 |
+
},
|
| 26371 |
+
{
|
| 26372 |
+
"epoch": 9.936423841059602,
|
| 26373 |
+
"grad_norm": 0.19157814979553223,
|
| 26374 |
+
"learning_rate": 2.519893899204244e-06,
|
| 26375 |
+
"loss": 0.0055,
|
| 26376 |
+
"step": 3751
|
| 26377 |
+
},
|
| 26378 |
+
{
|
| 26379 |
+
"epoch": 9.93907284768212,
|
| 26380 |
+
"grad_norm": 0.06832128018140793,
|
| 26381 |
+
"learning_rate": 2.3872679045092837e-06,
|
| 26382 |
+
"loss": 0.0053,
|
| 26383 |
+
"step": 3752
|
| 26384 |
+
},
|
| 26385 |
+
{
|
| 26386 |
+
"epoch": 9.941721854304635,
|
| 26387 |
+
"grad_norm": 0.5007702708244324,
|
| 26388 |
+
"learning_rate": 2.2546419098143238e-06,
|
| 26389 |
+
"loss": 0.0115,
|
| 26390 |
+
"step": 3753
|
| 26391 |
+
},
|
| 26392 |
+
{
|
| 26393 |
+
"epoch": 9.944370860927153,
|
| 26394 |
+
"grad_norm": 0.18187777698040009,
|
| 26395 |
+
"learning_rate": 2.122015915119363e-06,
|
| 26396 |
+
"loss": 0.0075,
|
| 26397 |
+
"step": 3754
|
| 26398 |
+
},
|
| 26399 |
+
{
|
| 26400 |
+
"epoch": 9.947019867549669,
|
| 26401 |
+
"grad_norm": 0.1455528289079666,
|
| 26402 |
+
"learning_rate": 1.989389920424403e-06,
|
| 26403 |
+
"loss": 0.0075,
|
| 26404 |
+
"step": 3755
|
| 26405 |
+
},
|
| 26406 |
+
{
|
| 26407 |
+
"epoch": 9.949668874172186,
|
| 26408 |
+
"grad_norm": 0.11611378937959671,
|
| 26409 |
+
"learning_rate": 1.856763925729443e-06,
|
| 26410 |
+
"loss": 0.0058,
|
| 26411 |
+
"step": 3756
|
| 26412 |
+
},
|
| 26413 |
+
{
|
| 26414 |
+
"epoch": 9.952317880794702,
|
| 26415 |
+
"grad_norm": 0.08630287647247314,
|
| 26416 |
+
"learning_rate": 1.7241379310344827e-06,
|
| 26417 |
+
"loss": 0.0054,
|
| 26418 |
+
"step": 3757
|
| 26419 |
+
},
|
| 26420 |
+
{
|
| 26421 |
+
"epoch": 9.954966887417218,
|
| 26422 |
+
"grad_norm": 0.37754523754119873,
|
| 26423 |
+
"learning_rate": 1.5915119363395226e-06,
|
| 26424 |
+
"loss": 0.0111,
|
| 26425 |
+
"step": 3758
|
| 26426 |
+
},
|
| 26427 |
+
{
|
| 26428 |
+
"epoch": 9.957615894039735,
|
| 26429 |
+
"grad_norm": 2.650604248046875,
|
| 26430 |
+
"learning_rate": 1.4588859416445623e-06,
|
| 26431 |
+
"loss": 0.0598,
|
| 26432 |
+
"step": 3759
|
| 26433 |
+
},
|
| 26434 |
+
{
|
| 26435 |
+
"epoch": 9.960264900662251,
|
| 26436 |
+
"grad_norm": 1.2378737926483154,
|
| 26437 |
+
"learning_rate": 1.3262599469496022e-06,
|
| 26438 |
+
"loss": 0.0196,
|
| 26439 |
+
"step": 3760
|
| 26440 |
+
},
|
| 26441 |
+
{
|
| 26442 |
+
"epoch": 9.962913907284769,
|
| 26443 |
+
"grad_norm": 0.4178251624107361,
|
| 26444 |
+
"learning_rate": 1.1936339522546418e-06,
|
| 26445 |
+
"loss": 0.0102,
|
| 26446 |
+
"step": 3761
|
| 26447 |
+
},
|
| 26448 |
+
{
|
| 26449 |
+
"epoch": 9.965562913907284,
|
| 26450 |
+
"grad_norm": 0.5158873796463013,
|
| 26451 |
+
"learning_rate": 1.0610079575596815e-06,
|
| 26452 |
+
"loss": 0.0058,
|
| 26453 |
+
"step": 3762
|
| 26454 |
+
},
|
| 26455 |
+
{
|
| 26456 |
+
"epoch": 9.968211920529802,
|
| 26457 |
+
"grad_norm": 0.0933571457862854,
|
| 26458 |
+
"learning_rate": 9.283819628647215e-07,
|
| 26459 |
+
"loss": 0.006,
|
| 26460 |
+
"step": 3763
|
| 26461 |
+
},
|
| 26462 |
+
{
|
| 26463 |
+
"epoch": 9.970860927152318,
|
| 26464 |
+
"grad_norm": 0.207793727517128,
|
| 26465 |
+
"learning_rate": 7.957559681697613e-07,
|
| 26466 |
+
"loss": 0.0098,
|
| 26467 |
+
"step": 3764
|
| 26468 |
+
},
|
| 26469 |
+
{
|
| 26470 |
+
"epoch": 9.973509933774835,
|
| 26471 |
+
"grad_norm": 0.11516987532377243,
|
| 26472 |
+
"learning_rate": 6.631299734748011e-07,
|
| 26473 |
+
"loss": 0.0063,
|
| 26474 |
+
"step": 3765
|
| 26475 |
+
},
|
| 26476 |
+
{
|
| 26477 |
+
"epoch": 9.976158940397351,
|
| 26478 |
+
"grad_norm": 4.050484657287598,
|
| 26479 |
+
"learning_rate": 5.305039787798408e-07,
|
| 26480 |
+
"loss": 0.0562,
|
| 26481 |
+
"step": 3766
|
| 26482 |
+
},
|
| 26483 |
+
{
|
| 26484 |
+
"epoch": 9.978807947019867,
|
| 26485 |
+
"grad_norm": 0.06329333037137985,
|
| 26486 |
+
"learning_rate": 3.9787798408488065e-07,
|
| 26487 |
+
"loss": 0.0044,
|
| 26488 |
+
"step": 3767
|
| 26489 |
+
},
|
| 26490 |
+
{
|
| 26491 |
+
"epoch": 9.981456953642384,
|
| 26492 |
+
"grad_norm": 0.5716582536697388,
|
| 26493 |
+
"learning_rate": 2.652519893899204e-07,
|
| 26494 |
+
"loss": 0.0068,
|
| 26495 |
+
"step": 3768
|
| 26496 |
+
},
|
| 26497 |
+
{
|
| 26498 |
+
"epoch": 9.9841059602649,
|
| 26499 |
+
"grad_norm": 0.15470373630523682,
|
| 26500 |
+
"learning_rate": 1.326259946949602e-07,
|
| 26501 |
+
"loss": 0.0072,
|
| 26502 |
+
"step": 3769
|
| 26503 |
+
},
|
| 26504 |
+
{
|
| 26505 |
+
"epoch": 9.986754966887418,
|
| 26506 |
+
"grad_norm": 0.05631321296095848,
|
| 26507 |
+
"learning_rate": 0.0,
|
| 26508 |
+
"loss": 0.004,
|
| 26509 |
+
"step": 3770
|
| 26510 |
}
|
| 26511 |
],
|
| 26512 |
"logging_steps": 1,
|
|
|
|
| 26521 |
"should_evaluate": false,
|
| 26522 |
"should_log": false,
|
| 26523 |
"should_save": true,
|
| 26524 |
+
"should_training_stop": true
|
| 26525 |
},
|
| 26526 |
"attributes": {}
|
| 26527 |
}
|
| 26528 |
},
|
| 26529 |
+
"total_flos": 2.1337935347515392e+20,
|
| 26530 |
"train_batch_size": 4,
|
| 26531 |
"trial_name": null,
|
| 26532 |
"trial_params": null
|