Uploaded checkpoint-15000
Browse files- model.safetensors +1 -1
- optimizer.pt +1 -1
- rng_state.pth +1 -1
- scheduler.pt +1 -1
- trainer_state.json +3513 -5
model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 2692969128
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:383f4fae36db9da6980db3312b06ab3f4ce131795c57a92818ad3b4ba86b07ff
|
| 3 |
size 2692969128
|
optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 5386075202
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d370718701079151a6ee36cb10ead87c85c0451e71c56c5203d02a5abf8f5cd0
|
| 3 |
size 5386075202
|
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:ee3528bf0ace792176d57cac1ea8e325db1e81a8856e3e8a6e53688b51f9516e
|
| 3 |
size 14244
|
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 sha256:0cf4bd40b0e3062c56584d972e9743cac19669a0283ba7de8c76540e6d58df00
|
| 3 |
size 1064
|
trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
-
"best_metric": 1.
|
| 3 |
-
"best_model_checkpoint": "runs/deepseek_20240422-210351/checkpoint-
|
| 4 |
-
"epoch": 0.
|
| 5 |
"eval_steps": 5000,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -7023,6 +7023,3514 @@
|
|
| 7023 |
"eval_samples_per_second": 16.869,
|
| 7024 |
"eval_steps_per_second": 16.869,
|
| 7025 |
"step": 10000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 7026 |
}
|
| 7027 |
],
|
| 7028 |
"logging_steps": 10,
|
|
@@ -7030,7 +10538,7 @@
|
|
| 7030 |
"num_input_tokens_seen": 0,
|
| 7031 |
"num_train_epochs": 1,
|
| 7032 |
"save_steps": 5000,
|
| 7033 |
-
"total_flos":
|
| 7034 |
"train_batch_size": 1,
|
| 7035 |
"trial_name": null,
|
| 7036 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_metric": 1.3379485607147217,
|
| 3 |
+
"best_model_checkpoint": "runs/deepseek_20240422-210351/checkpoint-15000",
|
| 4 |
+
"epoch": 0.375,
|
| 5 |
"eval_steps": 5000,
|
| 6 |
+
"global_step": 15000,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 7023 |
"eval_samples_per_second": 16.869,
|
| 7024 |
"eval_steps_per_second": 16.869,
|
| 7025 |
"step": 10000
|
| 7026 |
+
},
|
| 7027 |
+
{
|
| 7028 |
+
"epoch": 0.25,
|
| 7029 |
+
"grad_norm": 20.75,
|
| 7030 |
+
"learning_rate": 1.3552542372881357e-05,
|
| 7031 |
+
"loss": 1.3801,
|
| 7032 |
+
"step": 10010
|
| 7033 |
+
},
|
| 7034 |
+
{
|
| 7035 |
+
"epoch": 0.25,
|
| 7036 |
+
"grad_norm": 12.875,
|
| 7037 |
+
"learning_rate": 1.3545762711864408e-05,
|
| 7038 |
+
"loss": 1.3604,
|
| 7039 |
+
"step": 10020
|
| 7040 |
+
},
|
| 7041 |
+
{
|
| 7042 |
+
"epoch": 0.25,
|
| 7043 |
+
"grad_norm": 10.9375,
|
| 7044 |
+
"learning_rate": 1.3538983050847458e-05,
|
| 7045 |
+
"loss": 1.489,
|
| 7046 |
+
"step": 10030
|
| 7047 |
+
},
|
| 7048 |
+
{
|
| 7049 |
+
"epoch": 0.25,
|
| 7050 |
+
"grad_norm": 10.6875,
|
| 7051 |
+
"learning_rate": 1.353220338983051e-05,
|
| 7052 |
+
"loss": 1.5012,
|
| 7053 |
+
"step": 10040
|
| 7054 |
+
},
|
| 7055 |
+
{
|
| 7056 |
+
"epoch": 0.25,
|
| 7057 |
+
"grad_norm": 21.625,
|
| 7058 |
+
"learning_rate": 1.352542372881356e-05,
|
| 7059 |
+
"loss": 1.4174,
|
| 7060 |
+
"step": 10050
|
| 7061 |
+
},
|
| 7062 |
+
{
|
| 7063 |
+
"epoch": 0.25,
|
| 7064 |
+
"grad_norm": 28.875,
|
| 7065 |
+
"learning_rate": 1.3518644067796612e-05,
|
| 7066 |
+
"loss": 1.2765,
|
| 7067 |
+
"step": 10060
|
| 7068 |
+
},
|
| 7069 |
+
{
|
| 7070 |
+
"epoch": 0.25,
|
| 7071 |
+
"grad_norm": 23.875,
|
| 7072 |
+
"learning_rate": 1.3511864406779662e-05,
|
| 7073 |
+
"loss": 1.4869,
|
| 7074 |
+
"step": 10070
|
| 7075 |
+
},
|
| 7076 |
+
{
|
| 7077 |
+
"epoch": 0.25,
|
| 7078 |
+
"grad_norm": 20.0,
|
| 7079 |
+
"learning_rate": 1.3505084745762713e-05,
|
| 7080 |
+
"loss": 1.2636,
|
| 7081 |
+
"step": 10080
|
| 7082 |
+
},
|
| 7083 |
+
{
|
| 7084 |
+
"epoch": 0.25,
|
| 7085 |
+
"grad_norm": 12.4375,
|
| 7086 |
+
"learning_rate": 1.3498305084745764e-05,
|
| 7087 |
+
"loss": 1.4241,
|
| 7088 |
+
"step": 10090
|
| 7089 |
+
},
|
| 7090 |
+
{
|
| 7091 |
+
"epoch": 0.25,
|
| 7092 |
+
"grad_norm": 25.5,
|
| 7093 |
+
"learning_rate": 1.3491525423728816e-05,
|
| 7094 |
+
"loss": 1.3722,
|
| 7095 |
+
"step": 10100
|
| 7096 |
+
},
|
| 7097 |
+
{
|
| 7098 |
+
"epoch": 0.25,
|
| 7099 |
+
"grad_norm": 21.5,
|
| 7100 |
+
"learning_rate": 1.3484745762711866e-05,
|
| 7101 |
+
"loss": 1.3611,
|
| 7102 |
+
"step": 10110
|
| 7103 |
+
},
|
| 7104 |
+
{
|
| 7105 |
+
"epoch": 0.25,
|
| 7106 |
+
"grad_norm": 14.375,
|
| 7107 |
+
"learning_rate": 1.3477966101694917e-05,
|
| 7108 |
+
"loss": 1.4016,
|
| 7109 |
+
"step": 10120
|
| 7110 |
+
},
|
| 7111 |
+
{
|
| 7112 |
+
"epoch": 0.25,
|
| 7113 |
+
"grad_norm": 11.3125,
|
| 7114 |
+
"learning_rate": 1.3471186440677968e-05,
|
| 7115 |
+
"loss": 1.3755,
|
| 7116 |
+
"step": 10130
|
| 7117 |
+
},
|
| 7118 |
+
{
|
| 7119 |
+
"epoch": 0.25,
|
| 7120 |
+
"grad_norm": 15.125,
|
| 7121 |
+
"learning_rate": 1.346440677966102e-05,
|
| 7122 |
+
"loss": 1.3698,
|
| 7123 |
+
"step": 10140
|
| 7124 |
+
},
|
| 7125 |
+
{
|
| 7126 |
+
"epoch": 0.25,
|
| 7127 |
+
"grad_norm": 23.875,
|
| 7128 |
+
"learning_rate": 1.345762711864407e-05,
|
| 7129 |
+
"loss": 1.3596,
|
| 7130 |
+
"step": 10150
|
| 7131 |
+
},
|
| 7132 |
+
{
|
| 7133 |
+
"epoch": 0.25,
|
| 7134 |
+
"grad_norm": 11.9375,
|
| 7135 |
+
"learning_rate": 1.3450847457627121e-05,
|
| 7136 |
+
"loss": 1.3765,
|
| 7137 |
+
"step": 10160
|
| 7138 |
+
},
|
| 7139 |
+
{
|
| 7140 |
+
"epoch": 0.25,
|
| 7141 |
+
"grad_norm": 33.0,
|
| 7142 |
+
"learning_rate": 1.3444067796610169e-05,
|
| 7143 |
+
"loss": 1.3891,
|
| 7144 |
+
"step": 10170
|
| 7145 |
+
},
|
| 7146 |
+
{
|
| 7147 |
+
"epoch": 0.25,
|
| 7148 |
+
"grad_norm": 17.25,
|
| 7149 |
+
"learning_rate": 1.343728813559322e-05,
|
| 7150 |
+
"loss": 1.3707,
|
| 7151 |
+
"step": 10180
|
| 7152 |
+
},
|
| 7153 |
+
{
|
| 7154 |
+
"epoch": 0.25,
|
| 7155 |
+
"grad_norm": 14.0625,
|
| 7156 |
+
"learning_rate": 1.3430508474576272e-05,
|
| 7157 |
+
"loss": 1.3348,
|
| 7158 |
+
"step": 10190
|
| 7159 |
+
},
|
| 7160 |
+
{
|
| 7161 |
+
"epoch": 0.26,
|
| 7162 |
+
"grad_norm": 30.625,
|
| 7163 |
+
"learning_rate": 1.3423728813559323e-05,
|
| 7164 |
+
"loss": 1.4954,
|
| 7165 |
+
"step": 10200
|
| 7166 |
+
},
|
| 7167 |
+
{
|
| 7168 |
+
"epoch": 0.26,
|
| 7169 |
+
"grad_norm": 7.8125,
|
| 7170 |
+
"learning_rate": 1.3416949152542373e-05,
|
| 7171 |
+
"loss": 1.4154,
|
| 7172 |
+
"step": 10210
|
| 7173 |
+
},
|
| 7174 |
+
{
|
| 7175 |
+
"epoch": 0.26,
|
| 7176 |
+
"grad_norm": 29.25,
|
| 7177 |
+
"learning_rate": 1.3410169491525424e-05,
|
| 7178 |
+
"loss": 1.5622,
|
| 7179 |
+
"step": 10220
|
| 7180 |
+
},
|
| 7181 |
+
{
|
| 7182 |
+
"epoch": 0.26,
|
| 7183 |
+
"grad_norm": 6.34375,
|
| 7184 |
+
"learning_rate": 1.3403389830508476e-05,
|
| 7185 |
+
"loss": 1.4377,
|
| 7186 |
+
"step": 10230
|
| 7187 |
+
},
|
| 7188 |
+
{
|
| 7189 |
+
"epoch": 0.26,
|
| 7190 |
+
"grad_norm": 31.875,
|
| 7191 |
+
"learning_rate": 1.3396610169491527e-05,
|
| 7192 |
+
"loss": 1.3888,
|
| 7193 |
+
"step": 10240
|
| 7194 |
+
},
|
| 7195 |
+
{
|
| 7196 |
+
"epoch": 0.26,
|
| 7197 |
+
"grad_norm": 41.5,
|
| 7198 |
+
"learning_rate": 1.3389830508474577e-05,
|
| 7199 |
+
"loss": 1.4047,
|
| 7200 |
+
"step": 10250
|
| 7201 |
+
},
|
| 7202 |
+
{
|
| 7203 |
+
"epoch": 0.26,
|
| 7204 |
+
"grad_norm": 20.0,
|
| 7205 |
+
"learning_rate": 1.3383050847457628e-05,
|
| 7206 |
+
"loss": 1.3593,
|
| 7207 |
+
"step": 10260
|
| 7208 |
+
},
|
| 7209 |
+
{
|
| 7210 |
+
"epoch": 0.26,
|
| 7211 |
+
"grad_norm": 15.6875,
|
| 7212 |
+
"learning_rate": 1.337627118644068e-05,
|
| 7213 |
+
"loss": 1.4452,
|
| 7214 |
+
"step": 10270
|
| 7215 |
+
},
|
| 7216 |
+
{
|
| 7217 |
+
"epoch": 0.26,
|
| 7218 |
+
"grad_norm": 8.8125,
|
| 7219 |
+
"learning_rate": 1.3369491525423731e-05,
|
| 7220 |
+
"loss": 1.4375,
|
| 7221 |
+
"step": 10280
|
| 7222 |
+
},
|
| 7223 |
+
{
|
| 7224 |
+
"epoch": 0.26,
|
| 7225 |
+
"grad_norm": 55.0,
|
| 7226 |
+
"learning_rate": 1.336271186440678e-05,
|
| 7227 |
+
"loss": 1.2869,
|
| 7228 |
+
"step": 10290
|
| 7229 |
+
},
|
| 7230 |
+
{
|
| 7231 |
+
"epoch": 0.26,
|
| 7232 |
+
"grad_norm": 17.0,
|
| 7233 |
+
"learning_rate": 1.3355932203389832e-05,
|
| 7234 |
+
"loss": 1.3051,
|
| 7235 |
+
"step": 10300
|
| 7236 |
+
},
|
| 7237 |
+
{
|
| 7238 |
+
"epoch": 0.26,
|
| 7239 |
+
"grad_norm": 6.71875,
|
| 7240 |
+
"learning_rate": 1.3349152542372883e-05,
|
| 7241 |
+
"loss": 1.4207,
|
| 7242 |
+
"step": 10310
|
| 7243 |
+
},
|
| 7244 |
+
{
|
| 7245 |
+
"epoch": 0.26,
|
| 7246 |
+
"grad_norm": 27.375,
|
| 7247 |
+
"learning_rate": 1.3342372881355933e-05,
|
| 7248 |
+
"loss": 1.5179,
|
| 7249 |
+
"step": 10320
|
| 7250 |
+
},
|
| 7251 |
+
{
|
| 7252 |
+
"epoch": 0.26,
|
| 7253 |
+
"grad_norm": 22.0,
|
| 7254 |
+
"learning_rate": 1.3335593220338985e-05,
|
| 7255 |
+
"loss": 1.3529,
|
| 7256 |
+
"step": 10330
|
| 7257 |
+
},
|
| 7258 |
+
{
|
| 7259 |
+
"epoch": 0.26,
|
| 7260 |
+
"grad_norm": 27.125,
|
| 7261 |
+
"learning_rate": 1.3328813559322036e-05,
|
| 7262 |
+
"loss": 1.2933,
|
| 7263 |
+
"step": 10340
|
| 7264 |
+
},
|
| 7265 |
+
{
|
| 7266 |
+
"epoch": 0.26,
|
| 7267 |
+
"grad_norm": 15.375,
|
| 7268 |
+
"learning_rate": 1.3322033898305087e-05,
|
| 7269 |
+
"loss": 1.4026,
|
| 7270 |
+
"step": 10350
|
| 7271 |
+
},
|
| 7272 |
+
{
|
| 7273 |
+
"epoch": 0.26,
|
| 7274 |
+
"grad_norm": 20.0,
|
| 7275 |
+
"learning_rate": 1.3315254237288137e-05,
|
| 7276 |
+
"loss": 1.455,
|
| 7277 |
+
"step": 10360
|
| 7278 |
+
},
|
| 7279 |
+
{
|
| 7280 |
+
"epoch": 0.26,
|
| 7281 |
+
"grad_norm": 25.75,
|
| 7282 |
+
"learning_rate": 1.3308474576271187e-05,
|
| 7283 |
+
"loss": 1.4478,
|
| 7284 |
+
"step": 10370
|
| 7285 |
+
},
|
| 7286 |
+
{
|
| 7287 |
+
"epoch": 0.26,
|
| 7288 |
+
"grad_norm": 17.875,
|
| 7289 |
+
"learning_rate": 1.3301694915254238e-05,
|
| 7290 |
+
"loss": 1.4811,
|
| 7291 |
+
"step": 10380
|
| 7292 |
+
},
|
| 7293 |
+
{
|
| 7294 |
+
"epoch": 0.26,
|
| 7295 |
+
"grad_norm": 23.0,
|
| 7296 |
+
"learning_rate": 1.3294915254237288e-05,
|
| 7297 |
+
"loss": 1.2626,
|
| 7298 |
+
"step": 10390
|
| 7299 |
+
},
|
| 7300 |
+
{
|
| 7301 |
+
"epoch": 0.26,
|
| 7302 |
+
"grad_norm": 24.375,
|
| 7303 |
+
"learning_rate": 1.328813559322034e-05,
|
| 7304 |
+
"loss": 1.4401,
|
| 7305 |
+
"step": 10400
|
| 7306 |
+
},
|
| 7307 |
+
{
|
| 7308 |
+
"epoch": 0.26,
|
| 7309 |
+
"grad_norm": 15.5625,
|
| 7310 |
+
"learning_rate": 1.328135593220339e-05,
|
| 7311 |
+
"loss": 1.3587,
|
| 7312 |
+
"step": 10410
|
| 7313 |
+
},
|
| 7314 |
+
{
|
| 7315 |
+
"epoch": 0.26,
|
| 7316 |
+
"grad_norm": 43.75,
|
| 7317 |
+
"learning_rate": 1.3274576271186442e-05,
|
| 7318 |
+
"loss": 1.3697,
|
| 7319 |
+
"step": 10420
|
| 7320 |
+
},
|
| 7321 |
+
{
|
| 7322 |
+
"epoch": 0.26,
|
| 7323 |
+
"grad_norm": 16.375,
|
| 7324 |
+
"learning_rate": 1.3267796610169492e-05,
|
| 7325 |
+
"loss": 1.5165,
|
| 7326 |
+
"step": 10430
|
| 7327 |
+
},
|
| 7328 |
+
{
|
| 7329 |
+
"epoch": 0.26,
|
| 7330 |
+
"grad_norm": 26.25,
|
| 7331 |
+
"learning_rate": 1.3261016949152543e-05,
|
| 7332 |
+
"loss": 1.3134,
|
| 7333 |
+
"step": 10440
|
| 7334 |
+
},
|
| 7335 |
+
{
|
| 7336 |
+
"epoch": 0.26,
|
| 7337 |
+
"grad_norm": 29.375,
|
| 7338 |
+
"learning_rate": 1.3254237288135595e-05,
|
| 7339 |
+
"loss": 1.5107,
|
| 7340 |
+
"step": 10450
|
| 7341 |
+
},
|
| 7342 |
+
{
|
| 7343 |
+
"epoch": 0.26,
|
| 7344 |
+
"grad_norm": 15.0625,
|
| 7345 |
+
"learning_rate": 1.3247457627118644e-05,
|
| 7346 |
+
"loss": 1.4568,
|
| 7347 |
+
"step": 10460
|
| 7348 |
+
},
|
| 7349 |
+
{
|
| 7350 |
+
"epoch": 0.26,
|
| 7351 |
+
"grad_norm": 19.625,
|
| 7352 |
+
"learning_rate": 1.3240677966101696e-05,
|
| 7353 |
+
"loss": 1.3214,
|
| 7354 |
+
"step": 10470
|
| 7355 |
+
},
|
| 7356 |
+
{
|
| 7357 |
+
"epoch": 0.26,
|
| 7358 |
+
"grad_norm": 8.8125,
|
| 7359 |
+
"learning_rate": 1.3233898305084747e-05,
|
| 7360 |
+
"loss": 1.1839,
|
| 7361 |
+
"step": 10480
|
| 7362 |
+
},
|
| 7363 |
+
{
|
| 7364 |
+
"epoch": 0.26,
|
| 7365 |
+
"grad_norm": 12.0,
|
| 7366 |
+
"learning_rate": 1.3227118644067798e-05,
|
| 7367 |
+
"loss": 1.4022,
|
| 7368 |
+
"step": 10490
|
| 7369 |
+
},
|
| 7370 |
+
{
|
| 7371 |
+
"epoch": 0.26,
|
| 7372 |
+
"grad_norm": 14.5,
|
| 7373 |
+
"learning_rate": 1.3220338983050848e-05,
|
| 7374 |
+
"loss": 1.3453,
|
| 7375 |
+
"step": 10500
|
| 7376 |
+
},
|
| 7377 |
+
{
|
| 7378 |
+
"epoch": 0.26,
|
| 7379 |
+
"grad_norm": 12.625,
|
| 7380 |
+
"learning_rate": 1.32135593220339e-05,
|
| 7381 |
+
"loss": 1.6298,
|
| 7382 |
+
"step": 10510
|
| 7383 |
+
},
|
| 7384 |
+
{
|
| 7385 |
+
"epoch": 0.26,
|
| 7386 |
+
"grad_norm": 13.625,
|
| 7387 |
+
"learning_rate": 1.3206779661016951e-05,
|
| 7388 |
+
"loss": 1.4567,
|
| 7389 |
+
"step": 10520
|
| 7390 |
+
},
|
| 7391 |
+
{
|
| 7392 |
+
"epoch": 0.26,
|
| 7393 |
+
"grad_norm": 23.75,
|
| 7394 |
+
"learning_rate": 1.3200000000000002e-05,
|
| 7395 |
+
"loss": 1.5548,
|
| 7396 |
+
"step": 10530
|
| 7397 |
+
},
|
| 7398 |
+
{
|
| 7399 |
+
"epoch": 0.26,
|
| 7400 |
+
"grad_norm": 54.5,
|
| 7401 |
+
"learning_rate": 1.3193220338983052e-05,
|
| 7402 |
+
"loss": 1.4111,
|
| 7403 |
+
"step": 10540
|
| 7404 |
+
},
|
| 7405 |
+
{
|
| 7406 |
+
"epoch": 0.26,
|
| 7407 |
+
"grad_norm": 13.0625,
|
| 7408 |
+
"learning_rate": 1.3186440677966103e-05,
|
| 7409 |
+
"loss": 1.433,
|
| 7410 |
+
"step": 10550
|
| 7411 |
+
},
|
| 7412 |
+
{
|
| 7413 |
+
"epoch": 0.26,
|
| 7414 |
+
"grad_norm": 17.375,
|
| 7415 |
+
"learning_rate": 1.3179661016949155e-05,
|
| 7416 |
+
"loss": 1.5689,
|
| 7417 |
+
"step": 10560
|
| 7418 |
+
},
|
| 7419 |
+
{
|
| 7420 |
+
"epoch": 0.26,
|
| 7421 |
+
"grad_norm": 20.25,
|
| 7422 |
+
"learning_rate": 1.3172881355932206e-05,
|
| 7423 |
+
"loss": 1.3946,
|
| 7424 |
+
"step": 10570
|
| 7425 |
+
},
|
| 7426 |
+
{
|
| 7427 |
+
"epoch": 0.26,
|
| 7428 |
+
"grad_norm": 15.5,
|
| 7429 |
+
"learning_rate": 1.3166101694915254e-05,
|
| 7430 |
+
"loss": 1.4482,
|
| 7431 |
+
"step": 10580
|
| 7432 |
+
},
|
| 7433 |
+
{
|
| 7434 |
+
"epoch": 0.26,
|
| 7435 |
+
"grad_norm": 9.875,
|
| 7436 |
+
"learning_rate": 1.3159322033898306e-05,
|
| 7437 |
+
"loss": 1.4798,
|
| 7438 |
+
"step": 10590
|
| 7439 |
+
},
|
| 7440 |
+
{
|
| 7441 |
+
"epoch": 0.27,
|
| 7442 |
+
"grad_norm": 24.875,
|
| 7443 |
+
"learning_rate": 1.3152542372881355e-05,
|
| 7444 |
+
"loss": 1.4339,
|
| 7445 |
+
"step": 10600
|
| 7446 |
+
},
|
| 7447 |
+
{
|
| 7448 |
+
"epoch": 0.27,
|
| 7449 |
+
"grad_norm": 27.875,
|
| 7450 |
+
"learning_rate": 1.3145762711864407e-05,
|
| 7451 |
+
"loss": 1.3277,
|
| 7452 |
+
"step": 10610
|
| 7453 |
+
},
|
| 7454 |
+
{
|
| 7455 |
+
"epoch": 0.27,
|
| 7456 |
+
"grad_norm": 15.4375,
|
| 7457 |
+
"learning_rate": 1.3138983050847458e-05,
|
| 7458 |
+
"loss": 1.362,
|
| 7459 |
+
"step": 10620
|
| 7460 |
+
},
|
| 7461 |
+
{
|
| 7462 |
+
"epoch": 0.27,
|
| 7463 |
+
"grad_norm": 26.875,
|
| 7464 |
+
"learning_rate": 1.313220338983051e-05,
|
| 7465 |
+
"loss": 1.3184,
|
| 7466 |
+
"step": 10630
|
| 7467 |
+
},
|
| 7468 |
+
{
|
| 7469 |
+
"epoch": 0.27,
|
| 7470 |
+
"grad_norm": 10.25,
|
| 7471 |
+
"learning_rate": 1.312542372881356e-05,
|
| 7472 |
+
"loss": 1.5107,
|
| 7473 |
+
"step": 10640
|
| 7474 |
+
},
|
| 7475 |
+
{
|
| 7476 |
+
"epoch": 0.27,
|
| 7477 |
+
"grad_norm": 15.0625,
|
| 7478 |
+
"learning_rate": 1.311864406779661e-05,
|
| 7479 |
+
"loss": 1.1496,
|
| 7480 |
+
"step": 10650
|
| 7481 |
+
},
|
| 7482 |
+
{
|
| 7483 |
+
"epoch": 0.27,
|
| 7484 |
+
"grad_norm": 14.5,
|
| 7485 |
+
"learning_rate": 1.3111864406779662e-05,
|
| 7486 |
+
"loss": 1.4018,
|
| 7487 |
+
"step": 10660
|
| 7488 |
+
},
|
| 7489 |
+
{
|
| 7490 |
+
"epoch": 0.27,
|
| 7491 |
+
"grad_norm": 38.25,
|
| 7492 |
+
"learning_rate": 1.3105084745762714e-05,
|
| 7493 |
+
"loss": 1.5046,
|
| 7494 |
+
"step": 10670
|
| 7495 |
+
},
|
| 7496 |
+
{
|
| 7497 |
+
"epoch": 0.27,
|
| 7498 |
+
"grad_norm": 16.25,
|
| 7499 |
+
"learning_rate": 1.3098305084745763e-05,
|
| 7500 |
+
"loss": 1.3867,
|
| 7501 |
+
"step": 10680
|
| 7502 |
+
},
|
| 7503 |
+
{
|
| 7504 |
+
"epoch": 0.27,
|
| 7505 |
+
"grad_norm": 9.125,
|
| 7506 |
+
"learning_rate": 1.3091525423728815e-05,
|
| 7507 |
+
"loss": 1.3284,
|
| 7508 |
+
"step": 10690
|
| 7509 |
+
},
|
| 7510 |
+
{
|
| 7511 |
+
"epoch": 0.27,
|
| 7512 |
+
"grad_norm": 27.75,
|
| 7513 |
+
"learning_rate": 1.3084745762711866e-05,
|
| 7514 |
+
"loss": 1.4511,
|
| 7515 |
+
"step": 10700
|
| 7516 |
+
},
|
| 7517 |
+
{
|
| 7518 |
+
"epoch": 0.27,
|
| 7519 |
+
"grad_norm": 11.0625,
|
| 7520 |
+
"learning_rate": 1.3077966101694917e-05,
|
| 7521 |
+
"loss": 1.4621,
|
| 7522 |
+
"step": 10710
|
| 7523 |
+
},
|
| 7524 |
+
{
|
| 7525 |
+
"epoch": 0.27,
|
| 7526 |
+
"grad_norm": 17.875,
|
| 7527 |
+
"learning_rate": 1.3071186440677967e-05,
|
| 7528 |
+
"loss": 1.3628,
|
| 7529 |
+
"step": 10720
|
| 7530 |
+
},
|
| 7531 |
+
{
|
| 7532 |
+
"epoch": 0.27,
|
| 7533 |
+
"grad_norm": 19.25,
|
| 7534 |
+
"learning_rate": 1.3064406779661019e-05,
|
| 7535 |
+
"loss": 1.3811,
|
| 7536 |
+
"step": 10730
|
| 7537 |
+
},
|
| 7538 |
+
{
|
| 7539 |
+
"epoch": 0.27,
|
| 7540 |
+
"grad_norm": 17.125,
|
| 7541 |
+
"learning_rate": 1.305762711864407e-05,
|
| 7542 |
+
"loss": 1.283,
|
| 7543 |
+
"step": 10740
|
| 7544 |
+
},
|
| 7545 |
+
{
|
| 7546 |
+
"epoch": 0.27,
|
| 7547 |
+
"grad_norm": 7.78125,
|
| 7548 |
+
"learning_rate": 1.305084745762712e-05,
|
| 7549 |
+
"loss": 1.3398,
|
| 7550 |
+
"step": 10750
|
| 7551 |
+
},
|
| 7552 |
+
{
|
| 7553 |
+
"epoch": 0.27,
|
| 7554 |
+
"grad_norm": 43.5,
|
| 7555 |
+
"learning_rate": 1.3044067796610171e-05,
|
| 7556 |
+
"loss": 1.3942,
|
| 7557 |
+
"step": 10760
|
| 7558 |
+
},
|
| 7559 |
+
{
|
| 7560 |
+
"epoch": 0.27,
|
| 7561 |
+
"grad_norm": 14.75,
|
| 7562 |
+
"learning_rate": 1.3037288135593222e-05,
|
| 7563 |
+
"loss": 1.407,
|
| 7564 |
+
"step": 10770
|
| 7565 |
+
},
|
| 7566 |
+
{
|
| 7567 |
+
"epoch": 0.27,
|
| 7568 |
+
"grad_norm": 19.625,
|
| 7569 |
+
"learning_rate": 1.3030508474576274e-05,
|
| 7570 |
+
"loss": 1.3883,
|
| 7571 |
+
"step": 10780
|
| 7572 |
+
},
|
| 7573 |
+
{
|
| 7574 |
+
"epoch": 0.27,
|
| 7575 |
+
"grad_norm": 22.125,
|
| 7576 |
+
"learning_rate": 1.3023728813559322e-05,
|
| 7577 |
+
"loss": 1.4614,
|
| 7578 |
+
"step": 10790
|
| 7579 |
+
},
|
| 7580 |
+
{
|
| 7581 |
+
"epoch": 0.27,
|
| 7582 |
+
"grad_norm": 9.75,
|
| 7583 |
+
"learning_rate": 1.3016949152542373e-05,
|
| 7584 |
+
"loss": 1.4048,
|
| 7585 |
+
"step": 10800
|
| 7586 |
+
},
|
| 7587 |
+
{
|
| 7588 |
+
"epoch": 0.27,
|
| 7589 |
+
"grad_norm": 16.625,
|
| 7590 |
+
"learning_rate": 1.3010169491525425e-05,
|
| 7591 |
+
"loss": 1.5016,
|
| 7592 |
+
"step": 10810
|
| 7593 |
+
},
|
| 7594 |
+
{
|
| 7595 |
+
"epoch": 0.27,
|
| 7596 |
+
"grad_norm": 16.75,
|
| 7597 |
+
"learning_rate": 1.3003389830508474e-05,
|
| 7598 |
+
"loss": 1.562,
|
| 7599 |
+
"step": 10820
|
| 7600 |
+
},
|
| 7601 |
+
{
|
| 7602 |
+
"epoch": 0.27,
|
| 7603 |
+
"grad_norm": 39.0,
|
| 7604 |
+
"learning_rate": 1.2996610169491526e-05,
|
| 7605 |
+
"loss": 1.3651,
|
| 7606 |
+
"step": 10830
|
| 7607 |
+
},
|
| 7608 |
+
{
|
| 7609 |
+
"epoch": 0.27,
|
| 7610 |
+
"grad_norm": 18.75,
|
| 7611 |
+
"learning_rate": 1.2989830508474577e-05,
|
| 7612 |
+
"loss": 1.4262,
|
| 7613 |
+
"step": 10840
|
| 7614 |
+
},
|
| 7615 |
+
{
|
| 7616 |
+
"epoch": 0.27,
|
| 7617 |
+
"grad_norm": 9.75,
|
| 7618 |
+
"learning_rate": 1.2983050847457629e-05,
|
| 7619 |
+
"loss": 1.3998,
|
| 7620 |
+
"step": 10850
|
| 7621 |
+
},
|
| 7622 |
+
{
|
| 7623 |
+
"epoch": 0.27,
|
| 7624 |
+
"grad_norm": 19.625,
|
| 7625 |
+
"learning_rate": 1.2976271186440678e-05,
|
| 7626 |
+
"loss": 1.3164,
|
| 7627 |
+
"step": 10860
|
| 7628 |
+
},
|
| 7629 |
+
{
|
| 7630 |
+
"epoch": 0.27,
|
| 7631 |
+
"grad_norm": 18.125,
|
| 7632 |
+
"learning_rate": 1.296949152542373e-05,
|
| 7633 |
+
"loss": 1.2888,
|
| 7634 |
+
"step": 10870
|
| 7635 |
+
},
|
| 7636 |
+
{
|
| 7637 |
+
"epoch": 0.27,
|
| 7638 |
+
"grad_norm": 8.5625,
|
| 7639 |
+
"learning_rate": 1.2962711864406781e-05,
|
| 7640 |
+
"loss": 1.3567,
|
| 7641 |
+
"step": 10880
|
| 7642 |
+
},
|
| 7643 |
+
{
|
| 7644 |
+
"epoch": 0.27,
|
| 7645 |
+
"grad_norm": 18.75,
|
| 7646 |
+
"learning_rate": 1.295593220338983e-05,
|
| 7647 |
+
"loss": 1.5165,
|
| 7648 |
+
"step": 10890
|
| 7649 |
+
},
|
| 7650 |
+
{
|
| 7651 |
+
"epoch": 0.27,
|
| 7652 |
+
"grad_norm": 11.125,
|
| 7653 |
+
"learning_rate": 1.2949152542372882e-05,
|
| 7654 |
+
"loss": 1.4366,
|
| 7655 |
+
"step": 10900
|
| 7656 |
+
},
|
| 7657 |
+
{
|
| 7658 |
+
"epoch": 0.27,
|
| 7659 |
+
"grad_norm": 14.875,
|
| 7660 |
+
"learning_rate": 1.2942372881355934e-05,
|
| 7661 |
+
"loss": 1.3693,
|
| 7662 |
+
"step": 10910
|
| 7663 |
+
},
|
| 7664 |
+
{
|
| 7665 |
+
"epoch": 0.27,
|
| 7666 |
+
"grad_norm": 12.625,
|
| 7667 |
+
"learning_rate": 1.2935593220338985e-05,
|
| 7668 |
+
"loss": 1.4838,
|
| 7669 |
+
"step": 10920
|
| 7670 |
+
},
|
| 7671 |
+
{
|
| 7672 |
+
"epoch": 0.27,
|
| 7673 |
+
"grad_norm": 11.375,
|
| 7674 |
+
"learning_rate": 1.2928813559322035e-05,
|
| 7675 |
+
"loss": 1.2853,
|
| 7676 |
+
"step": 10930
|
| 7677 |
+
},
|
| 7678 |
+
{
|
| 7679 |
+
"epoch": 0.27,
|
| 7680 |
+
"grad_norm": 17.375,
|
| 7681 |
+
"learning_rate": 1.2922033898305086e-05,
|
| 7682 |
+
"loss": 1.3973,
|
| 7683 |
+
"step": 10940
|
| 7684 |
+
},
|
| 7685 |
+
{
|
| 7686 |
+
"epoch": 0.27,
|
| 7687 |
+
"grad_norm": 26.75,
|
| 7688 |
+
"learning_rate": 1.2915254237288137e-05,
|
| 7689 |
+
"loss": 1.3605,
|
| 7690 |
+
"step": 10950
|
| 7691 |
+
},
|
| 7692 |
+
{
|
| 7693 |
+
"epoch": 0.27,
|
| 7694 |
+
"grad_norm": 23.5,
|
| 7695 |
+
"learning_rate": 1.2908474576271189e-05,
|
| 7696 |
+
"loss": 1.5397,
|
| 7697 |
+
"step": 10960
|
| 7698 |
+
},
|
| 7699 |
+
{
|
| 7700 |
+
"epoch": 0.27,
|
| 7701 |
+
"grad_norm": 34.5,
|
| 7702 |
+
"learning_rate": 1.2901694915254239e-05,
|
| 7703 |
+
"loss": 1.2128,
|
| 7704 |
+
"step": 10970
|
| 7705 |
+
},
|
| 7706 |
+
{
|
| 7707 |
+
"epoch": 0.27,
|
| 7708 |
+
"grad_norm": 15.0625,
|
| 7709 |
+
"learning_rate": 1.289491525423729e-05,
|
| 7710 |
+
"loss": 1.4024,
|
| 7711 |
+
"step": 10980
|
| 7712 |
+
},
|
| 7713 |
+
{
|
| 7714 |
+
"epoch": 0.27,
|
| 7715 |
+
"grad_norm": 13.75,
|
| 7716 |
+
"learning_rate": 1.2888135593220341e-05,
|
| 7717 |
+
"loss": 1.448,
|
| 7718 |
+
"step": 10990
|
| 7719 |
+
},
|
| 7720 |
+
{
|
| 7721 |
+
"epoch": 0.28,
|
| 7722 |
+
"grad_norm": 10.5,
|
| 7723 |
+
"learning_rate": 1.288135593220339e-05,
|
| 7724 |
+
"loss": 1.564,
|
| 7725 |
+
"step": 11000
|
| 7726 |
+
},
|
| 7727 |
+
{
|
| 7728 |
+
"epoch": 0.28,
|
| 7729 |
+
"grad_norm": 22.0,
|
| 7730 |
+
"learning_rate": 1.287457627118644e-05,
|
| 7731 |
+
"loss": 1.4604,
|
| 7732 |
+
"step": 11010
|
| 7733 |
+
},
|
| 7734 |
+
{
|
| 7735 |
+
"epoch": 0.28,
|
| 7736 |
+
"grad_norm": 33.25,
|
| 7737 |
+
"learning_rate": 1.2867796610169492e-05,
|
| 7738 |
+
"loss": 1.4426,
|
| 7739 |
+
"step": 11020
|
| 7740 |
+
},
|
| 7741 |
+
{
|
| 7742 |
+
"epoch": 0.28,
|
| 7743 |
+
"grad_norm": 18.25,
|
| 7744 |
+
"learning_rate": 1.2861016949152542e-05,
|
| 7745 |
+
"loss": 1.4848,
|
| 7746 |
+
"step": 11030
|
| 7747 |
+
},
|
| 7748 |
+
{
|
| 7749 |
+
"epoch": 0.28,
|
| 7750 |
+
"grad_norm": 28.75,
|
| 7751 |
+
"learning_rate": 1.2854237288135593e-05,
|
| 7752 |
+
"loss": 1.5427,
|
| 7753 |
+
"step": 11040
|
| 7754 |
+
},
|
| 7755 |
+
{
|
| 7756 |
+
"epoch": 0.28,
|
| 7757 |
+
"grad_norm": 28.25,
|
| 7758 |
+
"learning_rate": 1.2847457627118645e-05,
|
| 7759 |
+
"loss": 1.4344,
|
| 7760 |
+
"step": 11050
|
| 7761 |
+
},
|
| 7762 |
+
{
|
| 7763 |
+
"epoch": 0.28,
|
| 7764 |
+
"grad_norm": 12.5625,
|
| 7765 |
+
"learning_rate": 1.2840677966101696e-05,
|
| 7766 |
+
"loss": 1.3914,
|
| 7767 |
+
"step": 11060
|
| 7768 |
+
},
|
| 7769 |
+
{
|
| 7770 |
+
"epoch": 0.28,
|
| 7771 |
+
"grad_norm": 9.875,
|
| 7772 |
+
"learning_rate": 1.2833898305084746e-05,
|
| 7773 |
+
"loss": 1.3298,
|
| 7774 |
+
"step": 11070
|
| 7775 |
+
},
|
| 7776 |
+
{
|
| 7777 |
+
"epoch": 0.28,
|
| 7778 |
+
"grad_norm": 9.75,
|
| 7779 |
+
"learning_rate": 1.2827118644067797e-05,
|
| 7780 |
+
"loss": 1.5821,
|
| 7781 |
+
"step": 11080
|
| 7782 |
+
},
|
| 7783 |
+
{
|
| 7784 |
+
"epoch": 0.28,
|
| 7785 |
+
"grad_norm": 7.8125,
|
| 7786 |
+
"learning_rate": 1.2820338983050849e-05,
|
| 7787 |
+
"loss": 1.3951,
|
| 7788 |
+
"step": 11090
|
| 7789 |
+
},
|
| 7790 |
+
{
|
| 7791 |
+
"epoch": 0.28,
|
| 7792 |
+
"grad_norm": 9.4375,
|
| 7793 |
+
"learning_rate": 1.28135593220339e-05,
|
| 7794 |
+
"loss": 1.5711,
|
| 7795 |
+
"step": 11100
|
| 7796 |
+
},
|
| 7797 |
+
{
|
| 7798 |
+
"epoch": 0.28,
|
| 7799 |
+
"grad_norm": 16.5,
|
| 7800 |
+
"learning_rate": 1.280677966101695e-05,
|
| 7801 |
+
"loss": 1.3264,
|
| 7802 |
+
"step": 11110
|
| 7803 |
+
},
|
| 7804 |
+
{
|
| 7805 |
+
"epoch": 0.28,
|
| 7806 |
+
"grad_norm": 33.25,
|
| 7807 |
+
"learning_rate": 1.2800000000000001e-05,
|
| 7808 |
+
"loss": 1.2913,
|
| 7809 |
+
"step": 11120
|
| 7810 |
+
},
|
| 7811 |
+
{
|
| 7812 |
+
"epoch": 0.28,
|
| 7813 |
+
"grad_norm": 23.875,
|
| 7814 |
+
"learning_rate": 1.2793220338983053e-05,
|
| 7815 |
+
"loss": 1.4656,
|
| 7816 |
+
"step": 11130
|
| 7817 |
+
},
|
| 7818 |
+
{
|
| 7819 |
+
"epoch": 0.28,
|
| 7820 |
+
"grad_norm": 42.75,
|
| 7821 |
+
"learning_rate": 1.2786440677966104e-05,
|
| 7822 |
+
"loss": 1.3079,
|
| 7823 |
+
"step": 11140
|
| 7824 |
+
},
|
| 7825 |
+
{
|
| 7826 |
+
"epoch": 0.28,
|
| 7827 |
+
"grad_norm": 21.875,
|
| 7828 |
+
"learning_rate": 1.2779661016949154e-05,
|
| 7829 |
+
"loss": 1.3085,
|
| 7830 |
+
"step": 11150
|
| 7831 |
+
},
|
| 7832 |
+
{
|
| 7833 |
+
"epoch": 0.28,
|
| 7834 |
+
"grad_norm": 13.75,
|
| 7835 |
+
"learning_rate": 1.2772881355932205e-05,
|
| 7836 |
+
"loss": 1.2513,
|
| 7837 |
+
"step": 11160
|
| 7838 |
+
},
|
| 7839 |
+
{
|
| 7840 |
+
"epoch": 0.28,
|
| 7841 |
+
"grad_norm": 25.25,
|
| 7842 |
+
"learning_rate": 1.2766101694915256e-05,
|
| 7843 |
+
"loss": 1.2082,
|
| 7844 |
+
"step": 11170
|
| 7845 |
+
},
|
| 7846 |
+
{
|
| 7847 |
+
"epoch": 0.28,
|
| 7848 |
+
"grad_norm": 13.0625,
|
| 7849 |
+
"learning_rate": 1.2759322033898308e-05,
|
| 7850 |
+
"loss": 1.466,
|
| 7851 |
+
"step": 11180
|
| 7852 |
+
},
|
| 7853 |
+
{
|
| 7854 |
+
"epoch": 0.28,
|
| 7855 |
+
"grad_norm": 20.75,
|
| 7856 |
+
"learning_rate": 1.2752542372881358e-05,
|
| 7857 |
+
"loss": 1.2071,
|
| 7858 |
+
"step": 11190
|
| 7859 |
+
},
|
| 7860 |
+
{
|
| 7861 |
+
"epoch": 0.28,
|
| 7862 |
+
"grad_norm": 9.1875,
|
| 7863 |
+
"learning_rate": 1.2745762711864407e-05,
|
| 7864 |
+
"loss": 1.3974,
|
| 7865 |
+
"step": 11200
|
| 7866 |
+
},
|
| 7867 |
+
{
|
| 7868 |
+
"epoch": 0.28,
|
| 7869 |
+
"grad_norm": 33.0,
|
| 7870 |
+
"learning_rate": 1.2738983050847457e-05,
|
| 7871 |
+
"loss": 1.3897,
|
| 7872 |
+
"step": 11210
|
| 7873 |
+
},
|
| 7874 |
+
{
|
| 7875 |
+
"epoch": 0.28,
|
| 7876 |
+
"grad_norm": 25.625,
|
| 7877 |
+
"learning_rate": 1.2732203389830508e-05,
|
| 7878 |
+
"loss": 1.2995,
|
| 7879 |
+
"step": 11220
|
| 7880 |
+
},
|
| 7881 |
+
{
|
| 7882 |
+
"epoch": 0.28,
|
| 7883 |
+
"grad_norm": 24.0,
|
| 7884 |
+
"learning_rate": 1.272542372881356e-05,
|
| 7885 |
+
"loss": 1.5626,
|
| 7886 |
+
"step": 11230
|
| 7887 |
+
},
|
| 7888 |
+
{
|
| 7889 |
+
"epoch": 0.28,
|
| 7890 |
+
"grad_norm": 33.25,
|
| 7891 |
+
"learning_rate": 1.2718644067796611e-05,
|
| 7892 |
+
"loss": 1.4729,
|
| 7893 |
+
"step": 11240
|
| 7894 |
+
},
|
| 7895 |
+
{
|
| 7896 |
+
"epoch": 0.28,
|
| 7897 |
+
"grad_norm": 34.75,
|
| 7898 |
+
"learning_rate": 1.2711864406779661e-05,
|
| 7899 |
+
"loss": 1.3542,
|
| 7900 |
+
"step": 11250
|
| 7901 |
+
},
|
| 7902 |
+
{
|
| 7903 |
+
"epoch": 0.28,
|
| 7904 |
+
"grad_norm": 12.875,
|
| 7905 |
+
"learning_rate": 1.2705084745762712e-05,
|
| 7906 |
+
"loss": 1.3378,
|
| 7907 |
+
"step": 11260
|
| 7908 |
+
},
|
| 7909 |
+
{
|
| 7910 |
+
"epoch": 0.28,
|
| 7911 |
+
"grad_norm": 46.25,
|
| 7912 |
+
"learning_rate": 1.2698305084745764e-05,
|
| 7913 |
+
"loss": 1.6005,
|
| 7914 |
+
"step": 11270
|
| 7915 |
+
},
|
| 7916 |
+
{
|
| 7917 |
+
"epoch": 0.28,
|
| 7918 |
+
"grad_norm": 25.75,
|
| 7919 |
+
"learning_rate": 1.2691525423728815e-05,
|
| 7920 |
+
"loss": 1.3225,
|
| 7921 |
+
"step": 11280
|
| 7922 |
+
},
|
| 7923 |
+
{
|
| 7924 |
+
"epoch": 0.28,
|
| 7925 |
+
"grad_norm": 34.0,
|
| 7926 |
+
"learning_rate": 1.2684745762711865e-05,
|
| 7927 |
+
"loss": 1.251,
|
| 7928 |
+
"step": 11290
|
| 7929 |
+
},
|
| 7930 |
+
{
|
| 7931 |
+
"epoch": 0.28,
|
| 7932 |
+
"grad_norm": 10.625,
|
| 7933 |
+
"learning_rate": 1.2677966101694916e-05,
|
| 7934 |
+
"loss": 1.3167,
|
| 7935 |
+
"step": 11300
|
| 7936 |
+
},
|
| 7937 |
+
{
|
| 7938 |
+
"epoch": 0.28,
|
| 7939 |
+
"grad_norm": 15.375,
|
| 7940 |
+
"learning_rate": 1.2671186440677968e-05,
|
| 7941 |
+
"loss": 1.2932,
|
| 7942 |
+
"step": 11310
|
| 7943 |
+
},
|
| 7944 |
+
{
|
| 7945 |
+
"epoch": 0.28,
|
| 7946 |
+
"grad_norm": 16.375,
|
| 7947 |
+
"learning_rate": 1.2664406779661019e-05,
|
| 7948 |
+
"loss": 1.3602,
|
| 7949 |
+
"step": 11320
|
| 7950 |
+
},
|
| 7951 |
+
{
|
| 7952 |
+
"epoch": 0.28,
|
| 7953 |
+
"grad_norm": 26.5,
|
| 7954 |
+
"learning_rate": 1.2657627118644069e-05,
|
| 7955 |
+
"loss": 1.4041,
|
| 7956 |
+
"step": 11330
|
| 7957 |
+
},
|
| 7958 |
+
{
|
| 7959 |
+
"epoch": 0.28,
|
| 7960 |
+
"grad_norm": 61.5,
|
| 7961 |
+
"learning_rate": 1.265084745762712e-05,
|
| 7962 |
+
"loss": 1.4216,
|
| 7963 |
+
"step": 11340
|
| 7964 |
+
},
|
| 7965 |
+
{
|
| 7966 |
+
"epoch": 0.28,
|
| 7967 |
+
"grad_norm": 21.0,
|
| 7968 |
+
"learning_rate": 1.2644067796610171e-05,
|
| 7969 |
+
"loss": 1.3234,
|
| 7970 |
+
"step": 11350
|
| 7971 |
+
},
|
| 7972 |
+
{
|
| 7973 |
+
"epoch": 0.28,
|
| 7974 |
+
"grad_norm": 10.25,
|
| 7975 |
+
"learning_rate": 1.2637288135593221e-05,
|
| 7976 |
+
"loss": 1.3886,
|
| 7977 |
+
"step": 11360
|
| 7978 |
+
},
|
| 7979 |
+
{
|
| 7980 |
+
"epoch": 0.28,
|
| 7981 |
+
"grad_norm": 28.0,
|
| 7982 |
+
"learning_rate": 1.2630508474576273e-05,
|
| 7983 |
+
"loss": 1.4435,
|
| 7984 |
+
"step": 11370
|
| 7985 |
+
},
|
| 7986 |
+
{
|
| 7987 |
+
"epoch": 0.28,
|
| 7988 |
+
"grad_norm": 11.375,
|
| 7989 |
+
"learning_rate": 1.2623728813559324e-05,
|
| 7990 |
+
"loss": 1.429,
|
| 7991 |
+
"step": 11380
|
| 7992 |
+
},
|
| 7993 |
+
{
|
| 7994 |
+
"epoch": 0.28,
|
| 7995 |
+
"grad_norm": 20.25,
|
| 7996 |
+
"learning_rate": 1.2616949152542375e-05,
|
| 7997 |
+
"loss": 1.3044,
|
| 7998 |
+
"step": 11390
|
| 7999 |
+
},
|
| 8000 |
+
{
|
| 8001 |
+
"epoch": 0.28,
|
| 8002 |
+
"grad_norm": 35.25,
|
| 8003 |
+
"learning_rate": 1.2610169491525425e-05,
|
| 8004 |
+
"loss": 1.1607,
|
| 8005 |
+
"step": 11400
|
| 8006 |
+
},
|
| 8007 |
+
{
|
| 8008 |
+
"epoch": 0.29,
|
| 8009 |
+
"grad_norm": 18.75,
|
| 8010 |
+
"learning_rate": 1.2603389830508475e-05,
|
| 8011 |
+
"loss": 1.5203,
|
| 8012 |
+
"step": 11410
|
| 8013 |
+
},
|
| 8014 |
+
{
|
| 8015 |
+
"epoch": 0.29,
|
| 8016 |
+
"grad_norm": 19.0,
|
| 8017 |
+
"learning_rate": 1.2596610169491526e-05,
|
| 8018 |
+
"loss": 1.3805,
|
| 8019 |
+
"step": 11420
|
| 8020 |
+
},
|
| 8021 |
+
{
|
| 8022 |
+
"epoch": 0.29,
|
| 8023 |
+
"grad_norm": 27.625,
|
| 8024 |
+
"learning_rate": 1.2589830508474576e-05,
|
| 8025 |
+
"loss": 1.4146,
|
| 8026 |
+
"step": 11430
|
| 8027 |
+
},
|
| 8028 |
+
{
|
| 8029 |
+
"epoch": 0.29,
|
| 8030 |
+
"grad_norm": 7.6875,
|
| 8031 |
+
"learning_rate": 1.2583050847457627e-05,
|
| 8032 |
+
"loss": 1.3669,
|
| 8033 |
+
"step": 11440
|
| 8034 |
+
},
|
| 8035 |
+
{
|
| 8036 |
+
"epoch": 0.29,
|
| 8037 |
+
"grad_norm": 35.75,
|
| 8038 |
+
"learning_rate": 1.2576271186440679e-05,
|
| 8039 |
+
"loss": 1.3515,
|
| 8040 |
+
"step": 11450
|
| 8041 |
+
},
|
| 8042 |
+
{
|
| 8043 |
+
"epoch": 0.29,
|
| 8044 |
+
"grad_norm": 12.6875,
|
| 8045 |
+
"learning_rate": 1.256949152542373e-05,
|
| 8046 |
+
"loss": 1.2908,
|
| 8047 |
+
"step": 11460
|
| 8048 |
+
},
|
| 8049 |
+
{
|
| 8050 |
+
"epoch": 0.29,
|
| 8051 |
+
"grad_norm": 13.625,
|
| 8052 |
+
"learning_rate": 1.256271186440678e-05,
|
| 8053 |
+
"loss": 1.3585,
|
| 8054 |
+
"step": 11470
|
| 8055 |
+
},
|
| 8056 |
+
{
|
| 8057 |
+
"epoch": 0.29,
|
| 8058 |
+
"grad_norm": 19.125,
|
| 8059 |
+
"learning_rate": 1.2555932203389831e-05,
|
| 8060 |
+
"loss": 1.3873,
|
| 8061 |
+
"step": 11480
|
| 8062 |
+
},
|
| 8063 |
+
{
|
| 8064 |
+
"epoch": 0.29,
|
| 8065 |
+
"grad_norm": 18.5,
|
| 8066 |
+
"learning_rate": 1.2549152542372883e-05,
|
| 8067 |
+
"loss": 1.3207,
|
| 8068 |
+
"step": 11490
|
| 8069 |
+
},
|
| 8070 |
+
{
|
| 8071 |
+
"epoch": 0.29,
|
| 8072 |
+
"grad_norm": 21.125,
|
| 8073 |
+
"learning_rate": 1.2542372881355932e-05,
|
| 8074 |
+
"loss": 1.4772,
|
| 8075 |
+
"step": 11500
|
| 8076 |
+
},
|
| 8077 |
+
{
|
| 8078 |
+
"epoch": 0.29,
|
| 8079 |
+
"grad_norm": 7.4375,
|
| 8080 |
+
"learning_rate": 1.2535593220338984e-05,
|
| 8081 |
+
"loss": 1.496,
|
| 8082 |
+
"step": 11510
|
| 8083 |
+
},
|
| 8084 |
+
{
|
| 8085 |
+
"epoch": 0.29,
|
| 8086 |
+
"grad_norm": 15.6875,
|
| 8087 |
+
"learning_rate": 1.2528813559322035e-05,
|
| 8088 |
+
"loss": 1.2367,
|
| 8089 |
+
"step": 11520
|
| 8090 |
+
},
|
| 8091 |
+
{
|
| 8092 |
+
"epoch": 0.29,
|
| 8093 |
+
"grad_norm": 20.375,
|
| 8094 |
+
"learning_rate": 1.2522033898305087e-05,
|
| 8095 |
+
"loss": 1.4323,
|
| 8096 |
+
"step": 11530
|
| 8097 |
+
},
|
| 8098 |
+
{
|
| 8099 |
+
"epoch": 0.29,
|
| 8100 |
+
"grad_norm": 15.3125,
|
| 8101 |
+
"learning_rate": 1.2515254237288136e-05,
|
| 8102 |
+
"loss": 1.2777,
|
| 8103 |
+
"step": 11540
|
| 8104 |
+
},
|
| 8105 |
+
{
|
| 8106 |
+
"epoch": 0.29,
|
| 8107 |
+
"grad_norm": 17.25,
|
| 8108 |
+
"learning_rate": 1.2508474576271188e-05,
|
| 8109 |
+
"loss": 1.2263,
|
| 8110 |
+
"step": 11550
|
| 8111 |
+
},
|
| 8112 |
+
{
|
| 8113 |
+
"epoch": 0.29,
|
| 8114 |
+
"grad_norm": 10.1875,
|
| 8115 |
+
"learning_rate": 1.2501694915254239e-05,
|
| 8116 |
+
"loss": 1.5418,
|
| 8117 |
+
"step": 11560
|
| 8118 |
+
},
|
| 8119 |
+
{
|
| 8120 |
+
"epoch": 0.29,
|
| 8121 |
+
"grad_norm": 22.875,
|
| 8122 |
+
"learning_rate": 1.249491525423729e-05,
|
| 8123 |
+
"loss": 1.4742,
|
| 8124 |
+
"step": 11570
|
| 8125 |
+
},
|
| 8126 |
+
{
|
| 8127 |
+
"epoch": 0.29,
|
| 8128 |
+
"grad_norm": 37.5,
|
| 8129 |
+
"learning_rate": 1.248813559322034e-05,
|
| 8130 |
+
"loss": 1.38,
|
| 8131 |
+
"step": 11580
|
| 8132 |
+
},
|
| 8133 |
+
{
|
| 8134 |
+
"epoch": 0.29,
|
| 8135 |
+
"grad_norm": 7.15625,
|
| 8136 |
+
"learning_rate": 1.2481355932203392e-05,
|
| 8137 |
+
"loss": 1.2842,
|
| 8138 |
+
"step": 11590
|
| 8139 |
+
},
|
| 8140 |
+
{
|
| 8141 |
+
"epoch": 0.29,
|
| 8142 |
+
"grad_norm": 34.5,
|
| 8143 |
+
"learning_rate": 1.2474576271186443e-05,
|
| 8144 |
+
"loss": 1.2457,
|
| 8145 |
+
"step": 11600
|
| 8146 |
+
},
|
| 8147 |
+
{
|
| 8148 |
+
"epoch": 0.29,
|
| 8149 |
+
"grad_norm": 13.375,
|
| 8150 |
+
"learning_rate": 1.2467796610169494e-05,
|
| 8151 |
+
"loss": 1.3445,
|
| 8152 |
+
"step": 11610
|
| 8153 |
+
},
|
| 8154 |
+
{
|
| 8155 |
+
"epoch": 0.29,
|
| 8156 |
+
"grad_norm": 13.875,
|
| 8157 |
+
"learning_rate": 1.2461016949152542e-05,
|
| 8158 |
+
"loss": 1.3287,
|
| 8159 |
+
"step": 11620
|
| 8160 |
+
},
|
| 8161 |
+
{
|
| 8162 |
+
"epoch": 0.29,
|
| 8163 |
+
"grad_norm": 16.625,
|
| 8164 |
+
"learning_rate": 1.2454237288135594e-05,
|
| 8165 |
+
"loss": 1.4088,
|
| 8166 |
+
"step": 11630
|
| 8167 |
+
},
|
| 8168 |
+
{
|
| 8169 |
+
"epoch": 0.29,
|
| 8170 |
+
"grad_norm": 17.375,
|
| 8171 |
+
"learning_rate": 1.2447457627118643e-05,
|
| 8172 |
+
"loss": 1.2976,
|
| 8173 |
+
"step": 11640
|
| 8174 |
+
},
|
| 8175 |
+
{
|
| 8176 |
+
"epoch": 0.29,
|
| 8177 |
+
"grad_norm": 29.375,
|
| 8178 |
+
"learning_rate": 1.2440677966101695e-05,
|
| 8179 |
+
"loss": 1.257,
|
| 8180 |
+
"step": 11650
|
| 8181 |
+
},
|
| 8182 |
+
{
|
| 8183 |
+
"epoch": 0.29,
|
| 8184 |
+
"grad_norm": 15.4375,
|
| 8185 |
+
"learning_rate": 1.2433898305084746e-05,
|
| 8186 |
+
"loss": 1.4099,
|
| 8187 |
+
"step": 11660
|
| 8188 |
+
},
|
| 8189 |
+
{
|
| 8190 |
+
"epoch": 0.29,
|
| 8191 |
+
"grad_norm": 19.5,
|
| 8192 |
+
"learning_rate": 1.2427118644067798e-05,
|
| 8193 |
+
"loss": 1.3072,
|
| 8194 |
+
"step": 11670
|
| 8195 |
+
},
|
| 8196 |
+
{
|
| 8197 |
+
"epoch": 0.29,
|
| 8198 |
+
"grad_norm": 18.375,
|
| 8199 |
+
"learning_rate": 1.2420338983050847e-05,
|
| 8200 |
+
"loss": 1.3438,
|
| 8201 |
+
"step": 11680
|
| 8202 |
+
},
|
| 8203 |
+
{
|
| 8204 |
+
"epoch": 0.29,
|
| 8205 |
+
"grad_norm": 12.125,
|
| 8206 |
+
"learning_rate": 1.2413559322033899e-05,
|
| 8207 |
+
"loss": 1.4044,
|
| 8208 |
+
"step": 11690
|
| 8209 |
+
},
|
| 8210 |
+
{
|
| 8211 |
+
"epoch": 0.29,
|
| 8212 |
+
"grad_norm": 20.25,
|
| 8213 |
+
"learning_rate": 1.240677966101695e-05,
|
| 8214 |
+
"loss": 1.3873,
|
| 8215 |
+
"step": 11700
|
| 8216 |
+
},
|
| 8217 |
+
{
|
| 8218 |
+
"epoch": 0.29,
|
| 8219 |
+
"grad_norm": 20.25,
|
| 8220 |
+
"learning_rate": 1.2400000000000002e-05,
|
| 8221 |
+
"loss": 1.3976,
|
| 8222 |
+
"step": 11710
|
| 8223 |
+
},
|
| 8224 |
+
{
|
| 8225 |
+
"epoch": 0.29,
|
| 8226 |
+
"grad_norm": 25.125,
|
| 8227 |
+
"learning_rate": 1.2393220338983051e-05,
|
| 8228 |
+
"loss": 1.5033,
|
| 8229 |
+
"step": 11720
|
| 8230 |
+
},
|
| 8231 |
+
{
|
| 8232 |
+
"epoch": 0.29,
|
| 8233 |
+
"grad_norm": 37.0,
|
| 8234 |
+
"learning_rate": 1.2386440677966103e-05,
|
| 8235 |
+
"loss": 1.2516,
|
| 8236 |
+
"step": 11730
|
| 8237 |
+
},
|
| 8238 |
+
{
|
| 8239 |
+
"epoch": 0.29,
|
| 8240 |
+
"grad_norm": 18.625,
|
| 8241 |
+
"learning_rate": 1.2379661016949154e-05,
|
| 8242 |
+
"loss": 1.2863,
|
| 8243 |
+
"step": 11740
|
| 8244 |
+
},
|
| 8245 |
+
{
|
| 8246 |
+
"epoch": 0.29,
|
| 8247 |
+
"grad_norm": 29.5,
|
| 8248 |
+
"learning_rate": 1.2372881355932205e-05,
|
| 8249 |
+
"loss": 1.4199,
|
| 8250 |
+
"step": 11750
|
| 8251 |
+
},
|
| 8252 |
+
{
|
| 8253 |
+
"epoch": 0.29,
|
| 8254 |
+
"grad_norm": 14.8125,
|
| 8255 |
+
"learning_rate": 1.2366101694915255e-05,
|
| 8256 |
+
"loss": 1.5511,
|
| 8257 |
+
"step": 11760
|
| 8258 |
+
},
|
| 8259 |
+
{
|
| 8260 |
+
"epoch": 0.29,
|
| 8261 |
+
"grad_norm": 31.125,
|
| 8262 |
+
"learning_rate": 1.2359322033898307e-05,
|
| 8263 |
+
"loss": 1.5902,
|
| 8264 |
+
"step": 11770
|
| 8265 |
+
},
|
| 8266 |
+
{
|
| 8267 |
+
"epoch": 0.29,
|
| 8268 |
+
"grad_norm": 9.125,
|
| 8269 |
+
"learning_rate": 1.2352542372881358e-05,
|
| 8270 |
+
"loss": 1.5406,
|
| 8271 |
+
"step": 11780
|
| 8272 |
+
},
|
| 8273 |
+
{
|
| 8274 |
+
"epoch": 0.29,
|
| 8275 |
+
"grad_norm": 15.0625,
|
| 8276 |
+
"learning_rate": 1.234576271186441e-05,
|
| 8277 |
+
"loss": 1.4011,
|
| 8278 |
+
"step": 11790
|
| 8279 |
+
},
|
| 8280 |
+
{
|
| 8281 |
+
"epoch": 0.29,
|
| 8282 |
+
"grad_norm": 15.0625,
|
| 8283 |
+
"learning_rate": 1.2338983050847459e-05,
|
| 8284 |
+
"loss": 1.2501,
|
| 8285 |
+
"step": 11800
|
| 8286 |
+
},
|
| 8287 |
+
{
|
| 8288 |
+
"epoch": 0.3,
|
| 8289 |
+
"grad_norm": 15.9375,
|
| 8290 |
+
"learning_rate": 1.233220338983051e-05,
|
| 8291 |
+
"loss": 1.3155,
|
| 8292 |
+
"step": 11810
|
| 8293 |
+
},
|
| 8294 |
+
{
|
| 8295 |
+
"epoch": 0.3,
|
| 8296 |
+
"grad_norm": 7.28125,
|
| 8297 |
+
"learning_rate": 1.2325423728813562e-05,
|
| 8298 |
+
"loss": 1.3896,
|
| 8299 |
+
"step": 11820
|
| 8300 |
+
},
|
| 8301 |
+
{
|
| 8302 |
+
"epoch": 0.3,
|
| 8303 |
+
"grad_norm": 29.25,
|
| 8304 |
+
"learning_rate": 1.231864406779661e-05,
|
| 8305 |
+
"loss": 1.6491,
|
| 8306 |
+
"step": 11830
|
| 8307 |
+
},
|
| 8308 |
+
{
|
| 8309 |
+
"epoch": 0.3,
|
| 8310 |
+
"grad_norm": 15.875,
|
| 8311 |
+
"learning_rate": 1.2311864406779661e-05,
|
| 8312 |
+
"loss": 1.3143,
|
| 8313 |
+
"step": 11840
|
| 8314 |
+
},
|
| 8315 |
+
{
|
| 8316 |
+
"epoch": 0.3,
|
| 8317 |
+
"grad_norm": 26.75,
|
| 8318 |
+
"learning_rate": 1.2305084745762713e-05,
|
| 8319 |
+
"loss": 1.3963,
|
| 8320 |
+
"step": 11850
|
| 8321 |
+
},
|
| 8322 |
+
{
|
| 8323 |
+
"epoch": 0.3,
|
| 8324 |
+
"grad_norm": 38.5,
|
| 8325 |
+
"learning_rate": 1.2298305084745762e-05,
|
| 8326 |
+
"loss": 1.3003,
|
| 8327 |
+
"step": 11860
|
| 8328 |
+
},
|
| 8329 |
+
{
|
| 8330 |
+
"epoch": 0.3,
|
| 8331 |
+
"grad_norm": 15.0,
|
| 8332 |
+
"learning_rate": 1.2291525423728814e-05,
|
| 8333 |
+
"loss": 1.51,
|
| 8334 |
+
"step": 11870
|
| 8335 |
+
},
|
| 8336 |
+
{
|
| 8337 |
+
"epoch": 0.3,
|
| 8338 |
+
"grad_norm": 23.5,
|
| 8339 |
+
"learning_rate": 1.2284745762711865e-05,
|
| 8340 |
+
"loss": 1.2961,
|
| 8341 |
+
"step": 11880
|
| 8342 |
+
},
|
| 8343 |
+
{
|
| 8344 |
+
"epoch": 0.3,
|
| 8345 |
+
"grad_norm": 12.5,
|
| 8346 |
+
"learning_rate": 1.2277966101694917e-05,
|
| 8347 |
+
"loss": 1.2989,
|
| 8348 |
+
"step": 11890
|
| 8349 |
+
},
|
| 8350 |
+
{
|
| 8351 |
+
"epoch": 0.3,
|
| 8352 |
+
"grad_norm": 32.0,
|
| 8353 |
+
"learning_rate": 1.2271186440677966e-05,
|
| 8354 |
+
"loss": 1.3863,
|
| 8355 |
+
"step": 11900
|
| 8356 |
+
},
|
| 8357 |
+
{
|
| 8358 |
+
"epoch": 0.3,
|
| 8359 |
+
"grad_norm": 22.625,
|
| 8360 |
+
"learning_rate": 1.2264406779661018e-05,
|
| 8361 |
+
"loss": 1.6525,
|
| 8362 |
+
"step": 11910
|
| 8363 |
+
},
|
| 8364 |
+
{
|
| 8365 |
+
"epoch": 0.3,
|
| 8366 |
+
"grad_norm": 64.0,
|
| 8367 |
+
"learning_rate": 1.2257627118644069e-05,
|
| 8368 |
+
"loss": 1.1806,
|
| 8369 |
+
"step": 11920
|
| 8370 |
+
},
|
| 8371 |
+
{
|
| 8372 |
+
"epoch": 0.3,
|
| 8373 |
+
"grad_norm": 12.0625,
|
| 8374 |
+
"learning_rate": 1.225084745762712e-05,
|
| 8375 |
+
"loss": 1.4508,
|
| 8376 |
+
"step": 11930
|
| 8377 |
+
},
|
| 8378 |
+
{
|
| 8379 |
+
"epoch": 0.3,
|
| 8380 |
+
"grad_norm": 29.125,
|
| 8381 |
+
"learning_rate": 1.224406779661017e-05,
|
| 8382 |
+
"loss": 1.5203,
|
| 8383 |
+
"step": 11940
|
| 8384 |
+
},
|
| 8385 |
+
{
|
| 8386 |
+
"epoch": 0.3,
|
| 8387 |
+
"grad_norm": 8.6875,
|
| 8388 |
+
"learning_rate": 1.2237288135593222e-05,
|
| 8389 |
+
"loss": 1.5215,
|
| 8390 |
+
"step": 11950
|
| 8391 |
+
},
|
| 8392 |
+
{
|
| 8393 |
+
"epoch": 0.3,
|
| 8394 |
+
"grad_norm": 7.40625,
|
| 8395 |
+
"learning_rate": 1.2230508474576273e-05,
|
| 8396 |
+
"loss": 1.2804,
|
| 8397 |
+
"step": 11960
|
| 8398 |
+
},
|
| 8399 |
+
{
|
| 8400 |
+
"epoch": 0.3,
|
| 8401 |
+
"grad_norm": 52.75,
|
| 8402 |
+
"learning_rate": 1.2223728813559323e-05,
|
| 8403 |
+
"loss": 1.3751,
|
| 8404 |
+
"step": 11970
|
| 8405 |
+
},
|
| 8406 |
+
{
|
| 8407 |
+
"epoch": 0.3,
|
| 8408 |
+
"grad_norm": 13.0,
|
| 8409 |
+
"learning_rate": 1.2216949152542374e-05,
|
| 8410 |
+
"loss": 1.2444,
|
| 8411 |
+
"step": 11980
|
| 8412 |
+
},
|
| 8413 |
+
{
|
| 8414 |
+
"epoch": 0.3,
|
| 8415 |
+
"grad_norm": 39.5,
|
| 8416 |
+
"learning_rate": 1.2210169491525426e-05,
|
| 8417 |
+
"loss": 1.2927,
|
| 8418 |
+
"step": 11990
|
| 8419 |
+
},
|
| 8420 |
+
{
|
| 8421 |
+
"epoch": 0.3,
|
| 8422 |
+
"grad_norm": 12.5625,
|
| 8423 |
+
"learning_rate": 1.2203389830508477e-05,
|
| 8424 |
+
"loss": 1.4383,
|
| 8425 |
+
"step": 12000
|
| 8426 |
+
},
|
| 8427 |
+
{
|
| 8428 |
+
"epoch": 0.3,
|
| 8429 |
+
"grad_norm": 41.5,
|
| 8430 |
+
"learning_rate": 1.2196610169491527e-05,
|
| 8431 |
+
"loss": 1.4106,
|
| 8432 |
+
"step": 12010
|
| 8433 |
+
},
|
| 8434 |
+
{
|
| 8435 |
+
"epoch": 0.3,
|
| 8436 |
+
"grad_norm": 16.5,
|
| 8437 |
+
"learning_rate": 1.2189830508474578e-05,
|
| 8438 |
+
"loss": 1.3155,
|
| 8439 |
+
"step": 12020
|
| 8440 |
+
},
|
| 8441 |
+
{
|
| 8442 |
+
"epoch": 0.3,
|
| 8443 |
+
"grad_norm": 13.9375,
|
| 8444 |
+
"learning_rate": 1.2183050847457628e-05,
|
| 8445 |
+
"loss": 1.3191,
|
| 8446 |
+
"step": 12030
|
| 8447 |
+
},
|
| 8448 |
+
{
|
| 8449 |
+
"epoch": 0.3,
|
| 8450 |
+
"grad_norm": 13.9375,
|
| 8451 |
+
"learning_rate": 1.2176271186440677e-05,
|
| 8452 |
+
"loss": 1.393,
|
| 8453 |
+
"step": 12040
|
| 8454 |
+
},
|
| 8455 |
+
{
|
| 8456 |
+
"epoch": 0.3,
|
| 8457 |
+
"grad_norm": 46.25,
|
| 8458 |
+
"learning_rate": 1.2169491525423729e-05,
|
| 8459 |
+
"loss": 1.3792,
|
| 8460 |
+
"step": 12050
|
| 8461 |
+
},
|
| 8462 |
+
{
|
| 8463 |
+
"epoch": 0.3,
|
| 8464 |
+
"grad_norm": 60.75,
|
| 8465 |
+
"learning_rate": 1.216271186440678e-05,
|
| 8466 |
+
"loss": 1.3197,
|
| 8467 |
+
"step": 12060
|
| 8468 |
+
},
|
| 8469 |
+
{
|
| 8470 |
+
"epoch": 0.3,
|
| 8471 |
+
"grad_norm": 27.875,
|
| 8472 |
+
"learning_rate": 1.2155932203389832e-05,
|
| 8473 |
+
"loss": 1.3858,
|
| 8474 |
+
"step": 12070
|
| 8475 |
+
},
|
| 8476 |
+
{
|
| 8477 |
+
"epoch": 0.3,
|
| 8478 |
+
"grad_norm": 42.25,
|
| 8479 |
+
"learning_rate": 1.2149152542372881e-05,
|
| 8480 |
+
"loss": 1.4581,
|
| 8481 |
+
"step": 12080
|
| 8482 |
+
},
|
| 8483 |
+
{
|
| 8484 |
+
"epoch": 0.3,
|
| 8485 |
+
"grad_norm": 16.75,
|
| 8486 |
+
"learning_rate": 1.2142372881355933e-05,
|
| 8487 |
+
"loss": 1.4428,
|
| 8488 |
+
"step": 12090
|
| 8489 |
+
},
|
| 8490 |
+
{
|
| 8491 |
+
"epoch": 0.3,
|
| 8492 |
+
"grad_norm": 40.75,
|
| 8493 |
+
"learning_rate": 1.2135593220338984e-05,
|
| 8494 |
+
"loss": 1.2876,
|
| 8495 |
+
"step": 12100
|
| 8496 |
+
},
|
| 8497 |
+
{
|
| 8498 |
+
"epoch": 0.3,
|
| 8499 |
+
"grad_norm": 34.75,
|
| 8500 |
+
"learning_rate": 1.2128813559322034e-05,
|
| 8501 |
+
"loss": 1.295,
|
| 8502 |
+
"step": 12110
|
| 8503 |
+
},
|
| 8504 |
+
{
|
| 8505 |
+
"epoch": 0.3,
|
| 8506 |
+
"grad_norm": 31.625,
|
| 8507 |
+
"learning_rate": 1.2122033898305085e-05,
|
| 8508 |
+
"loss": 1.4799,
|
| 8509 |
+
"step": 12120
|
| 8510 |
+
},
|
| 8511 |
+
{
|
| 8512 |
+
"epoch": 0.3,
|
| 8513 |
+
"grad_norm": 23.5,
|
| 8514 |
+
"learning_rate": 1.2115254237288137e-05,
|
| 8515 |
+
"loss": 1.3044,
|
| 8516 |
+
"step": 12130
|
| 8517 |
+
},
|
| 8518 |
+
{
|
| 8519 |
+
"epoch": 0.3,
|
| 8520 |
+
"grad_norm": 11.3125,
|
| 8521 |
+
"learning_rate": 1.2108474576271188e-05,
|
| 8522 |
+
"loss": 1.3421,
|
| 8523 |
+
"step": 12140
|
| 8524 |
+
},
|
| 8525 |
+
{
|
| 8526 |
+
"epoch": 0.3,
|
| 8527 |
+
"grad_norm": 17.875,
|
| 8528 |
+
"learning_rate": 1.2101694915254238e-05,
|
| 8529 |
+
"loss": 1.2581,
|
| 8530 |
+
"step": 12150
|
| 8531 |
+
},
|
| 8532 |
+
{
|
| 8533 |
+
"epoch": 0.3,
|
| 8534 |
+
"grad_norm": 32.75,
|
| 8535 |
+
"learning_rate": 1.209491525423729e-05,
|
| 8536 |
+
"loss": 1.4785,
|
| 8537 |
+
"step": 12160
|
| 8538 |
+
},
|
| 8539 |
+
{
|
| 8540 |
+
"epoch": 0.3,
|
| 8541 |
+
"grad_norm": 28.5,
|
| 8542 |
+
"learning_rate": 1.208813559322034e-05,
|
| 8543 |
+
"loss": 1.4118,
|
| 8544 |
+
"step": 12170
|
| 8545 |
+
},
|
| 8546 |
+
{
|
| 8547 |
+
"epoch": 0.3,
|
| 8548 |
+
"grad_norm": 11.875,
|
| 8549 |
+
"learning_rate": 1.2081355932203392e-05,
|
| 8550 |
+
"loss": 1.2336,
|
| 8551 |
+
"step": 12180
|
| 8552 |
+
},
|
| 8553 |
+
{
|
| 8554 |
+
"epoch": 0.3,
|
| 8555 |
+
"grad_norm": 23.75,
|
| 8556 |
+
"learning_rate": 1.2074576271186442e-05,
|
| 8557 |
+
"loss": 1.2431,
|
| 8558 |
+
"step": 12190
|
| 8559 |
+
},
|
| 8560 |
+
{
|
| 8561 |
+
"epoch": 0.3,
|
| 8562 |
+
"grad_norm": 8.4375,
|
| 8563 |
+
"learning_rate": 1.2067796610169493e-05,
|
| 8564 |
+
"loss": 1.3331,
|
| 8565 |
+
"step": 12200
|
| 8566 |
+
},
|
| 8567 |
+
{
|
| 8568 |
+
"epoch": 0.31,
|
| 8569 |
+
"grad_norm": 23.125,
|
| 8570 |
+
"learning_rate": 1.2061016949152544e-05,
|
| 8571 |
+
"loss": 1.5897,
|
| 8572 |
+
"step": 12210
|
| 8573 |
+
},
|
| 8574 |
+
{
|
| 8575 |
+
"epoch": 0.31,
|
| 8576 |
+
"grad_norm": 29.0,
|
| 8577 |
+
"learning_rate": 1.2054237288135596e-05,
|
| 8578 |
+
"loss": 1.4431,
|
| 8579 |
+
"step": 12220
|
| 8580 |
+
},
|
| 8581 |
+
{
|
| 8582 |
+
"epoch": 0.31,
|
| 8583 |
+
"grad_norm": 12.75,
|
| 8584 |
+
"learning_rate": 1.2047457627118646e-05,
|
| 8585 |
+
"loss": 1.207,
|
| 8586 |
+
"step": 12230
|
| 8587 |
+
},
|
| 8588 |
+
{
|
| 8589 |
+
"epoch": 0.31,
|
| 8590 |
+
"grad_norm": 17.75,
|
| 8591 |
+
"learning_rate": 1.2040677966101695e-05,
|
| 8592 |
+
"loss": 1.318,
|
| 8593 |
+
"step": 12240
|
| 8594 |
+
},
|
| 8595 |
+
{
|
| 8596 |
+
"epoch": 0.31,
|
| 8597 |
+
"grad_norm": 14.6875,
|
| 8598 |
+
"learning_rate": 1.2033898305084745e-05,
|
| 8599 |
+
"loss": 1.3912,
|
| 8600 |
+
"step": 12250
|
| 8601 |
+
},
|
| 8602 |
+
{
|
| 8603 |
+
"epoch": 0.31,
|
| 8604 |
+
"grad_norm": 26.0,
|
| 8605 |
+
"learning_rate": 1.2027118644067796e-05,
|
| 8606 |
+
"loss": 1.356,
|
| 8607 |
+
"step": 12260
|
| 8608 |
+
},
|
| 8609 |
+
{
|
| 8610 |
+
"epoch": 0.31,
|
| 8611 |
+
"grad_norm": 28.875,
|
| 8612 |
+
"learning_rate": 1.2020338983050848e-05,
|
| 8613 |
+
"loss": 1.4116,
|
| 8614 |
+
"step": 12270
|
| 8615 |
+
},
|
| 8616 |
+
{
|
| 8617 |
+
"epoch": 0.31,
|
| 8618 |
+
"grad_norm": 8.875,
|
| 8619 |
+
"learning_rate": 1.20135593220339e-05,
|
| 8620 |
+
"loss": 1.3749,
|
| 8621 |
+
"step": 12280
|
| 8622 |
+
},
|
| 8623 |
+
{
|
| 8624 |
+
"epoch": 0.31,
|
| 8625 |
+
"grad_norm": 14.1875,
|
| 8626 |
+
"learning_rate": 1.2006779661016949e-05,
|
| 8627 |
+
"loss": 1.4265,
|
| 8628 |
+
"step": 12290
|
| 8629 |
+
},
|
| 8630 |
+
{
|
| 8631 |
+
"epoch": 0.31,
|
| 8632 |
+
"grad_norm": 9.9375,
|
| 8633 |
+
"learning_rate": 1.2e-05,
|
| 8634 |
+
"loss": 1.4168,
|
| 8635 |
+
"step": 12300
|
| 8636 |
+
},
|
| 8637 |
+
{
|
| 8638 |
+
"epoch": 0.31,
|
| 8639 |
+
"grad_norm": 30.375,
|
| 8640 |
+
"learning_rate": 1.1993220338983052e-05,
|
| 8641 |
+
"loss": 1.3941,
|
| 8642 |
+
"step": 12310
|
| 8643 |
+
},
|
| 8644 |
+
{
|
| 8645 |
+
"epoch": 0.31,
|
| 8646 |
+
"grad_norm": 21.375,
|
| 8647 |
+
"learning_rate": 1.1986440677966103e-05,
|
| 8648 |
+
"loss": 1.3206,
|
| 8649 |
+
"step": 12320
|
| 8650 |
+
},
|
| 8651 |
+
{
|
| 8652 |
+
"epoch": 0.31,
|
| 8653 |
+
"grad_norm": 30.5,
|
| 8654 |
+
"learning_rate": 1.1979661016949153e-05,
|
| 8655 |
+
"loss": 1.4332,
|
| 8656 |
+
"step": 12330
|
| 8657 |
+
},
|
| 8658 |
+
{
|
| 8659 |
+
"epoch": 0.31,
|
| 8660 |
+
"grad_norm": 11.9375,
|
| 8661 |
+
"learning_rate": 1.1972881355932204e-05,
|
| 8662 |
+
"loss": 1.3988,
|
| 8663 |
+
"step": 12340
|
| 8664 |
+
},
|
| 8665 |
+
{
|
| 8666 |
+
"epoch": 0.31,
|
| 8667 |
+
"grad_norm": 19.625,
|
| 8668 |
+
"learning_rate": 1.1966101694915256e-05,
|
| 8669 |
+
"loss": 1.5356,
|
| 8670 |
+
"step": 12350
|
| 8671 |
+
},
|
| 8672 |
+
{
|
| 8673 |
+
"epoch": 0.31,
|
| 8674 |
+
"grad_norm": 24.375,
|
| 8675 |
+
"learning_rate": 1.1959322033898307e-05,
|
| 8676 |
+
"loss": 1.371,
|
| 8677 |
+
"step": 12360
|
| 8678 |
+
},
|
| 8679 |
+
{
|
| 8680 |
+
"epoch": 0.31,
|
| 8681 |
+
"grad_norm": 21.75,
|
| 8682 |
+
"learning_rate": 1.1952542372881357e-05,
|
| 8683 |
+
"loss": 1.6957,
|
| 8684 |
+
"step": 12370
|
| 8685 |
+
},
|
| 8686 |
+
{
|
| 8687 |
+
"epoch": 0.31,
|
| 8688 |
+
"grad_norm": 28.375,
|
| 8689 |
+
"learning_rate": 1.1945762711864408e-05,
|
| 8690 |
+
"loss": 1.4933,
|
| 8691 |
+
"step": 12380
|
| 8692 |
+
},
|
| 8693 |
+
{
|
| 8694 |
+
"epoch": 0.31,
|
| 8695 |
+
"grad_norm": 16.5,
|
| 8696 |
+
"learning_rate": 1.193898305084746e-05,
|
| 8697 |
+
"loss": 1.3463,
|
| 8698 |
+
"step": 12390
|
| 8699 |
+
},
|
| 8700 |
+
{
|
| 8701 |
+
"epoch": 0.31,
|
| 8702 |
+
"grad_norm": 15.4375,
|
| 8703 |
+
"learning_rate": 1.1932203389830511e-05,
|
| 8704 |
+
"loss": 1.431,
|
| 8705 |
+
"step": 12400
|
| 8706 |
+
},
|
| 8707 |
+
{
|
| 8708 |
+
"epoch": 0.31,
|
| 8709 |
+
"grad_norm": 21.875,
|
| 8710 |
+
"learning_rate": 1.192542372881356e-05,
|
| 8711 |
+
"loss": 1.3883,
|
| 8712 |
+
"step": 12410
|
| 8713 |
+
},
|
| 8714 |
+
{
|
| 8715 |
+
"epoch": 0.31,
|
| 8716 |
+
"grad_norm": 16.375,
|
| 8717 |
+
"learning_rate": 1.1918644067796612e-05,
|
| 8718 |
+
"loss": 1.4184,
|
| 8719 |
+
"step": 12420
|
| 8720 |
+
},
|
| 8721 |
+
{
|
| 8722 |
+
"epoch": 0.31,
|
| 8723 |
+
"grad_norm": 18.0,
|
| 8724 |
+
"learning_rate": 1.1911864406779663e-05,
|
| 8725 |
+
"loss": 1.3175,
|
| 8726 |
+
"step": 12430
|
| 8727 |
+
},
|
| 8728 |
+
{
|
| 8729 |
+
"epoch": 0.31,
|
| 8730 |
+
"grad_norm": 17.625,
|
| 8731 |
+
"learning_rate": 1.1905084745762713e-05,
|
| 8732 |
+
"loss": 1.3652,
|
| 8733 |
+
"step": 12440
|
| 8734 |
+
},
|
| 8735 |
+
{
|
| 8736 |
+
"epoch": 0.31,
|
| 8737 |
+
"grad_norm": 19.0,
|
| 8738 |
+
"learning_rate": 1.1898305084745763e-05,
|
| 8739 |
+
"loss": 1.494,
|
| 8740 |
+
"step": 12450
|
| 8741 |
+
},
|
| 8742 |
+
{
|
| 8743 |
+
"epoch": 0.31,
|
| 8744 |
+
"grad_norm": 31.875,
|
| 8745 |
+
"learning_rate": 1.1891525423728814e-05,
|
| 8746 |
+
"loss": 1.4232,
|
| 8747 |
+
"step": 12460
|
| 8748 |
+
},
|
| 8749 |
+
{
|
| 8750 |
+
"epoch": 0.31,
|
| 8751 |
+
"grad_norm": 39.5,
|
| 8752 |
+
"learning_rate": 1.1884745762711864e-05,
|
| 8753 |
+
"loss": 1.3255,
|
| 8754 |
+
"step": 12470
|
| 8755 |
+
},
|
| 8756 |
+
{
|
| 8757 |
+
"epoch": 0.31,
|
| 8758 |
+
"grad_norm": 53.75,
|
| 8759 |
+
"learning_rate": 1.1877966101694915e-05,
|
| 8760 |
+
"loss": 1.3136,
|
| 8761 |
+
"step": 12480
|
| 8762 |
+
},
|
| 8763 |
+
{
|
| 8764 |
+
"epoch": 0.31,
|
| 8765 |
+
"grad_norm": 31.625,
|
| 8766 |
+
"learning_rate": 1.1871186440677967e-05,
|
| 8767 |
+
"loss": 1.3727,
|
| 8768 |
+
"step": 12490
|
| 8769 |
+
},
|
| 8770 |
+
{
|
| 8771 |
+
"epoch": 0.31,
|
| 8772 |
+
"grad_norm": 11.25,
|
| 8773 |
+
"learning_rate": 1.1864406779661018e-05,
|
| 8774 |
+
"loss": 1.1973,
|
| 8775 |
+
"step": 12500
|
| 8776 |
+
},
|
| 8777 |
+
{
|
| 8778 |
+
"epoch": 0.31,
|
| 8779 |
+
"grad_norm": 19.125,
|
| 8780 |
+
"learning_rate": 1.1857627118644068e-05,
|
| 8781 |
+
"loss": 1.5978,
|
| 8782 |
+
"step": 12510
|
| 8783 |
+
},
|
| 8784 |
+
{
|
| 8785 |
+
"epoch": 0.31,
|
| 8786 |
+
"grad_norm": 10.8125,
|
| 8787 |
+
"learning_rate": 1.185084745762712e-05,
|
| 8788 |
+
"loss": 1.3858,
|
| 8789 |
+
"step": 12520
|
| 8790 |
+
},
|
| 8791 |
+
{
|
| 8792 |
+
"epoch": 0.31,
|
| 8793 |
+
"grad_norm": 16.875,
|
| 8794 |
+
"learning_rate": 1.184406779661017e-05,
|
| 8795 |
+
"loss": 1.4182,
|
| 8796 |
+
"step": 12530
|
| 8797 |
+
},
|
| 8798 |
+
{
|
| 8799 |
+
"epoch": 0.31,
|
| 8800 |
+
"grad_norm": 25.0,
|
| 8801 |
+
"learning_rate": 1.183728813559322e-05,
|
| 8802 |
+
"loss": 1.1735,
|
| 8803 |
+
"step": 12540
|
| 8804 |
+
},
|
| 8805 |
+
{
|
| 8806 |
+
"epoch": 0.31,
|
| 8807 |
+
"grad_norm": 30.375,
|
| 8808 |
+
"learning_rate": 1.1830508474576272e-05,
|
| 8809 |
+
"loss": 1.2487,
|
| 8810 |
+
"step": 12550
|
| 8811 |
+
},
|
| 8812 |
+
{
|
| 8813 |
+
"epoch": 0.31,
|
| 8814 |
+
"grad_norm": 40.75,
|
| 8815 |
+
"learning_rate": 1.1823728813559323e-05,
|
| 8816 |
+
"loss": 1.3385,
|
| 8817 |
+
"step": 12560
|
| 8818 |
+
},
|
| 8819 |
+
{
|
| 8820 |
+
"epoch": 0.31,
|
| 8821 |
+
"grad_norm": 11.25,
|
| 8822 |
+
"learning_rate": 1.1816949152542375e-05,
|
| 8823 |
+
"loss": 1.3932,
|
| 8824 |
+
"step": 12570
|
| 8825 |
+
},
|
| 8826 |
+
{
|
| 8827 |
+
"epoch": 0.31,
|
| 8828 |
+
"grad_norm": 12.875,
|
| 8829 |
+
"learning_rate": 1.1810169491525424e-05,
|
| 8830 |
+
"loss": 1.3053,
|
| 8831 |
+
"step": 12580
|
| 8832 |
+
},
|
| 8833 |
+
{
|
| 8834 |
+
"epoch": 0.31,
|
| 8835 |
+
"grad_norm": 16.0,
|
| 8836 |
+
"learning_rate": 1.1803389830508476e-05,
|
| 8837 |
+
"loss": 1.1677,
|
| 8838 |
+
"step": 12590
|
| 8839 |
+
},
|
| 8840 |
+
{
|
| 8841 |
+
"epoch": 0.32,
|
| 8842 |
+
"grad_norm": 33.0,
|
| 8843 |
+
"learning_rate": 1.1796610169491527e-05,
|
| 8844 |
+
"loss": 1.3547,
|
| 8845 |
+
"step": 12600
|
| 8846 |
+
},
|
| 8847 |
+
{
|
| 8848 |
+
"epoch": 0.32,
|
| 8849 |
+
"grad_norm": 13.25,
|
| 8850 |
+
"learning_rate": 1.1789830508474578e-05,
|
| 8851 |
+
"loss": 1.4628,
|
| 8852 |
+
"step": 12610
|
| 8853 |
+
},
|
| 8854 |
+
{
|
| 8855 |
+
"epoch": 0.32,
|
| 8856 |
+
"grad_norm": 36.5,
|
| 8857 |
+
"learning_rate": 1.1783050847457628e-05,
|
| 8858 |
+
"loss": 1.3457,
|
| 8859 |
+
"step": 12620
|
| 8860 |
+
},
|
| 8861 |
+
{
|
| 8862 |
+
"epoch": 0.32,
|
| 8863 |
+
"grad_norm": 77.0,
|
| 8864 |
+
"learning_rate": 1.177627118644068e-05,
|
| 8865 |
+
"loss": 1.4649,
|
| 8866 |
+
"step": 12630
|
| 8867 |
+
},
|
| 8868 |
+
{
|
| 8869 |
+
"epoch": 0.32,
|
| 8870 |
+
"grad_norm": 26.25,
|
| 8871 |
+
"learning_rate": 1.1769491525423731e-05,
|
| 8872 |
+
"loss": 1.6308,
|
| 8873 |
+
"step": 12640
|
| 8874 |
+
},
|
| 8875 |
+
{
|
| 8876 |
+
"epoch": 0.32,
|
| 8877 |
+
"grad_norm": 21.125,
|
| 8878 |
+
"learning_rate": 1.1762711864406782e-05,
|
| 8879 |
+
"loss": 1.315,
|
| 8880 |
+
"step": 12650
|
| 8881 |
+
},
|
| 8882 |
+
{
|
| 8883 |
+
"epoch": 0.32,
|
| 8884 |
+
"grad_norm": 18.5,
|
| 8885 |
+
"learning_rate": 1.175593220338983e-05,
|
| 8886 |
+
"loss": 1.3999,
|
| 8887 |
+
"step": 12660
|
| 8888 |
+
},
|
| 8889 |
+
{
|
| 8890 |
+
"epoch": 0.32,
|
| 8891 |
+
"grad_norm": 12.4375,
|
| 8892 |
+
"learning_rate": 1.1749152542372882e-05,
|
| 8893 |
+
"loss": 1.3569,
|
| 8894 |
+
"step": 12670
|
| 8895 |
+
},
|
| 8896 |
+
{
|
| 8897 |
+
"epoch": 0.32,
|
| 8898 |
+
"grad_norm": 50.0,
|
| 8899 |
+
"learning_rate": 1.1742372881355931e-05,
|
| 8900 |
+
"loss": 1.2881,
|
| 8901 |
+
"step": 12680
|
| 8902 |
+
},
|
| 8903 |
+
{
|
| 8904 |
+
"epoch": 0.32,
|
| 8905 |
+
"grad_norm": 14.3125,
|
| 8906 |
+
"learning_rate": 1.1735593220338983e-05,
|
| 8907 |
+
"loss": 1.5333,
|
| 8908 |
+
"step": 12690
|
| 8909 |
+
},
|
| 8910 |
+
{
|
| 8911 |
+
"epoch": 0.32,
|
| 8912 |
+
"grad_norm": 32.0,
|
| 8913 |
+
"learning_rate": 1.1728813559322034e-05,
|
| 8914 |
+
"loss": 1.4207,
|
| 8915 |
+
"step": 12700
|
| 8916 |
+
},
|
| 8917 |
+
{
|
| 8918 |
+
"epoch": 0.32,
|
| 8919 |
+
"grad_norm": 19.5,
|
| 8920 |
+
"learning_rate": 1.1722033898305086e-05,
|
| 8921 |
+
"loss": 1.3351,
|
| 8922 |
+
"step": 12710
|
| 8923 |
+
},
|
| 8924 |
+
{
|
| 8925 |
+
"epoch": 0.32,
|
| 8926 |
+
"grad_norm": 66.0,
|
| 8927 |
+
"learning_rate": 1.1715254237288135e-05,
|
| 8928 |
+
"loss": 1.3974,
|
| 8929 |
+
"step": 12720
|
| 8930 |
+
},
|
| 8931 |
+
{
|
| 8932 |
+
"epoch": 0.32,
|
| 8933 |
+
"grad_norm": 17.375,
|
| 8934 |
+
"learning_rate": 1.1708474576271187e-05,
|
| 8935 |
+
"loss": 1.3413,
|
| 8936 |
+
"step": 12730
|
| 8937 |
+
},
|
| 8938 |
+
{
|
| 8939 |
+
"epoch": 0.32,
|
| 8940 |
+
"grad_norm": 17.5,
|
| 8941 |
+
"learning_rate": 1.1701694915254238e-05,
|
| 8942 |
+
"loss": 1.4573,
|
| 8943 |
+
"step": 12740
|
| 8944 |
+
},
|
| 8945 |
+
{
|
| 8946 |
+
"epoch": 0.32,
|
| 8947 |
+
"grad_norm": 17.25,
|
| 8948 |
+
"learning_rate": 1.169491525423729e-05,
|
| 8949 |
+
"loss": 1.4507,
|
| 8950 |
+
"step": 12750
|
| 8951 |
+
},
|
| 8952 |
+
{
|
| 8953 |
+
"epoch": 0.32,
|
| 8954 |
+
"grad_norm": 23.5,
|
| 8955 |
+
"learning_rate": 1.168813559322034e-05,
|
| 8956 |
+
"loss": 1.4594,
|
| 8957 |
+
"step": 12760
|
| 8958 |
+
},
|
| 8959 |
+
{
|
| 8960 |
+
"epoch": 0.32,
|
| 8961 |
+
"grad_norm": 27.25,
|
| 8962 |
+
"learning_rate": 1.168135593220339e-05,
|
| 8963 |
+
"loss": 1.3051,
|
| 8964 |
+
"step": 12770
|
| 8965 |
+
},
|
| 8966 |
+
{
|
| 8967 |
+
"epoch": 0.32,
|
| 8968 |
+
"grad_norm": 14.8125,
|
| 8969 |
+
"learning_rate": 1.1674576271186442e-05,
|
| 8970 |
+
"loss": 1.4516,
|
| 8971 |
+
"step": 12780
|
| 8972 |
+
},
|
| 8973 |
+
{
|
| 8974 |
+
"epoch": 0.32,
|
| 8975 |
+
"grad_norm": 12.1875,
|
| 8976 |
+
"learning_rate": 1.1667796610169494e-05,
|
| 8977 |
+
"loss": 1.3318,
|
| 8978 |
+
"step": 12790
|
| 8979 |
+
},
|
| 8980 |
+
{
|
| 8981 |
+
"epoch": 0.32,
|
| 8982 |
+
"grad_norm": 12.0625,
|
| 8983 |
+
"learning_rate": 1.1661016949152543e-05,
|
| 8984 |
+
"loss": 1.5477,
|
| 8985 |
+
"step": 12800
|
| 8986 |
+
},
|
| 8987 |
+
{
|
| 8988 |
+
"epoch": 0.32,
|
| 8989 |
+
"grad_norm": 40.75,
|
| 8990 |
+
"learning_rate": 1.1654237288135595e-05,
|
| 8991 |
+
"loss": 1.2766,
|
| 8992 |
+
"step": 12810
|
| 8993 |
+
},
|
| 8994 |
+
{
|
| 8995 |
+
"epoch": 0.32,
|
| 8996 |
+
"grad_norm": 18.0,
|
| 8997 |
+
"learning_rate": 1.1647457627118646e-05,
|
| 8998 |
+
"loss": 1.2983,
|
| 8999 |
+
"step": 12820
|
| 9000 |
+
},
|
| 9001 |
+
{
|
| 9002 |
+
"epoch": 0.32,
|
| 9003 |
+
"grad_norm": 18.25,
|
| 9004 |
+
"learning_rate": 1.1640677966101697e-05,
|
| 9005 |
+
"loss": 1.2756,
|
| 9006 |
+
"step": 12830
|
| 9007 |
+
},
|
| 9008 |
+
{
|
| 9009 |
+
"epoch": 0.32,
|
| 9010 |
+
"grad_norm": 20.625,
|
| 9011 |
+
"learning_rate": 1.1633898305084747e-05,
|
| 9012 |
+
"loss": 1.4396,
|
| 9013 |
+
"step": 12840
|
| 9014 |
+
},
|
| 9015 |
+
{
|
| 9016 |
+
"epoch": 0.32,
|
| 9017 |
+
"grad_norm": 28.0,
|
| 9018 |
+
"learning_rate": 1.1627118644067799e-05,
|
| 9019 |
+
"loss": 1.3779,
|
| 9020 |
+
"step": 12850
|
| 9021 |
+
},
|
| 9022 |
+
{
|
| 9023 |
+
"epoch": 0.32,
|
| 9024 |
+
"grad_norm": 13.1875,
|
| 9025 |
+
"learning_rate": 1.162033898305085e-05,
|
| 9026 |
+
"loss": 1.4271,
|
| 9027 |
+
"step": 12860
|
| 9028 |
+
},
|
| 9029 |
+
{
|
| 9030 |
+
"epoch": 0.32,
|
| 9031 |
+
"grad_norm": 21.875,
|
| 9032 |
+
"learning_rate": 1.1613559322033898e-05,
|
| 9033 |
+
"loss": 1.2491,
|
| 9034 |
+
"step": 12870
|
| 9035 |
+
},
|
| 9036 |
+
{
|
| 9037 |
+
"epoch": 0.32,
|
| 9038 |
+
"grad_norm": 43.25,
|
| 9039 |
+
"learning_rate": 1.160677966101695e-05,
|
| 9040 |
+
"loss": 1.4746,
|
| 9041 |
+
"step": 12880
|
| 9042 |
+
},
|
| 9043 |
+
{
|
| 9044 |
+
"epoch": 0.32,
|
| 9045 |
+
"grad_norm": 51.75,
|
| 9046 |
+
"learning_rate": 1.16e-05,
|
| 9047 |
+
"loss": 1.4252,
|
| 9048 |
+
"step": 12890
|
| 9049 |
+
},
|
| 9050 |
+
{
|
| 9051 |
+
"epoch": 0.32,
|
| 9052 |
+
"grad_norm": 19.25,
|
| 9053 |
+
"learning_rate": 1.159322033898305e-05,
|
| 9054 |
+
"loss": 1.2963,
|
| 9055 |
+
"step": 12900
|
| 9056 |
+
},
|
| 9057 |
+
{
|
| 9058 |
+
"epoch": 0.32,
|
| 9059 |
+
"grad_norm": 14.4375,
|
| 9060 |
+
"learning_rate": 1.1586440677966102e-05,
|
| 9061 |
+
"loss": 1.3186,
|
| 9062 |
+
"step": 12910
|
| 9063 |
+
},
|
| 9064 |
+
{
|
| 9065 |
+
"epoch": 0.32,
|
| 9066 |
+
"grad_norm": 13.0625,
|
| 9067 |
+
"learning_rate": 1.1579661016949153e-05,
|
| 9068 |
+
"loss": 1.316,
|
| 9069 |
+
"step": 12920
|
| 9070 |
+
},
|
| 9071 |
+
{
|
| 9072 |
+
"epoch": 0.32,
|
| 9073 |
+
"grad_norm": 18.875,
|
| 9074 |
+
"learning_rate": 1.1572881355932205e-05,
|
| 9075 |
+
"loss": 1.3342,
|
| 9076 |
+
"step": 12930
|
| 9077 |
+
},
|
| 9078 |
+
{
|
| 9079 |
+
"epoch": 0.32,
|
| 9080 |
+
"grad_norm": 10.125,
|
| 9081 |
+
"learning_rate": 1.1566101694915254e-05,
|
| 9082 |
+
"loss": 1.1205,
|
| 9083 |
+
"step": 12940
|
| 9084 |
+
},
|
| 9085 |
+
{
|
| 9086 |
+
"epoch": 0.32,
|
| 9087 |
+
"grad_norm": 23.875,
|
| 9088 |
+
"learning_rate": 1.1559322033898306e-05,
|
| 9089 |
+
"loss": 1.4295,
|
| 9090 |
+
"step": 12950
|
| 9091 |
+
},
|
| 9092 |
+
{
|
| 9093 |
+
"epoch": 0.32,
|
| 9094 |
+
"grad_norm": 29.5,
|
| 9095 |
+
"learning_rate": 1.1552542372881357e-05,
|
| 9096 |
+
"loss": 1.3836,
|
| 9097 |
+
"step": 12960
|
| 9098 |
+
},
|
| 9099 |
+
{
|
| 9100 |
+
"epoch": 0.32,
|
| 9101 |
+
"grad_norm": 24.5,
|
| 9102 |
+
"learning_rate": 1.1545762711864409e-05,
|
| 9103 |
+
"loss": 1.2477,
|
| 9104 |
+
"step": 12970
|
| 9105 |
+
},
|
| 9106 |
+
{
|
| 9107 |
+
"epoch": 0.32,
|
| 9108 |
+
"grad_norm": 15.4375,
|
| 9109 |
+
"learning_rate": 1.1538983050847458e-05,
|
| 9110 |
+
"loss": 1.3418,
|
| 9111 |
+
"step": 12980
|
| 9112 |
+
},
|
| 9113 |
+
{
|
| 9114 |
+
"epoch": 0.32,
|
| 9115 |
+
"grad_norm": 30.375,
|
| 9116 |
+
"learning_rate": 1.153220338983051e-05,
|
| 9117 |
+
"loss": 1.3886,
|
| 9118 |
+
"step": 12990
|
| 9119 |
+
},
|
| 9120 |
+
{
|
| 9121 |
+
"epoch": 0.33,
|
| 9122 |
+
"grad_norm": 29.5,
|
| 9123 |
+
"learning_rate": 1.1525423728813561e-05,
|
| 9124 |
+
"loss": 1.4092,
|
| 9125 |
+
"step": 13000
|
| 9126 |
+
},
|
| 9127 |
+
{
|
| 9128 |
+
"epoch": 0.33,
|
| 9129 |
+
"grad_norm": 29.125,
|
| 9130 |
+
"learning_rate": 1.151864406779661e-05,
|
| 9131 |
+
"loss": 1.5852,
|
| 9132 |
+
"step": 13010
|
| 9133 |
+
},
|
| 9134 |
+
{
|
| 9135 |
+
"epoch": 0.33,
|
| 9136 |
+
"grad_norm": 18.875,
|
| 9137 |
+
"learning_rate": 1.1511864406779662e-05,
|
| 9138 |
+
"loss": 1.3838,
|
| 9139 |
+
"step": 13020
|
| 9140 |
+
},
|
| 9141 |
+
{
|
| 9142 |
+
"epoch": 0.33,
|
| 9143 |
+
"grad_norm": 17.0,
|
| 9144 |
+
"learning_rate": 1.1505084745762714e-05,
|
| 9145 |
+
"loss": 1.3919,
|
| 9146 |
+
"step": 13030
|
| 9147 |
+
},
|
| 9148 |
+
{
|
| 9149 |
+
"epoch": 0.33,
|
| 9150 |
+
"grad_norm": 14.125,
|
| 9151 |
+
"learning_rate": 1.1498305084745765e-05,
|
| 9152 |
+
"loss": 1.3076,
|
| 9153 |
+
"step": 13040
|
| 9154 |
+
},
|
| 9155 |
+
{
|
| 9156 |
+
"epoch": 0.33,
|
| 9157 |
+
"grad_norm": 10.0625,
|
| 9158 |
+
"learning_rate": 1.1491525423728815e-05,
|
| 9159 |
+
"loss": 1.339,
|
| 9160 |
+
"step": 13050
|
| 9161 |
+
},
|
| 9162 |
+
{
|
| 9163 |
+
"epoch": 0.33,
|
| 9164 |
+
"grad_norm": 14.25,
|
| 9165 |
+
"learning_rate": 1.1484745762711866e-05,
|
| 9166 |
+
"loss": 1.4315,
|
| 9167 |
+
"step": 13060
|
| 9168 |
+
},
|
| 9169 |
+
{
|
| 9170 |
+
"epoch": 0.33,
|
| 9171 |
+
"grad_norm": 8.6875,
|
| 9172 |
+
"learning_rate": 1.1477966101694916e-05,
|
| 9173 |
+
"loss": 1.3238,
|
| 9174 |
+
"step": 13070
|
| 9175 |
+
},
|
| 9176 |
+
{
|
| 9177 |
+
"epoch": 0.33,
|
| 9178 |
+
"grad_norm": 26.0,
|
| 9179 |
+
"learning_rate": 1.1471186440677965e-05,
|
| 9180 |
+
"loss": 1.3567,
|
| 9181 |
+
"step": 13080
|
| 9182 |
+
},
|
| 9183 |
+
{
|
| 9184 |
+
"epoch": 0.33,
|
| 9185 |
+
"grad_norm": 44.0,
|
| 9186 |
+
"learning_rate": 1.1464406779661017e-05,
|
| 9187 |
+
"loss": 1.2877,
|
| 9188 |
+
"step": 13090
|
| 9189 |
+
},
|
| 9190 |
+
{
|
| 9191 |
+
"epoch": 0.33,
|
| 9192 |
+
"grad_norm": 30.375,
|
| 9193 |
+
"learning_rate": 1.1457627118644068e-05,
|
| 9194 |
+
"loss": 1.3856,
|
| 9195 |
+
"step": 13100
|
| 9196 |
+
},
|
| 9197 |
+
{
|
| 9198 |
+
"epoch": 0.33,
|
| 9199 |
+
"grad_norm": 20.875,
|
| 9200 |
+
"learning_rate": 1.145084745762712e-05,
|
| 9201 |
+
"loss": 1.3871,
|
| 9202 |
+
"step": 13110
|
| 9203 |
+
},
|
| 9204 |
+
{
|
| 9205 |
+
"epoch": 0.33,
|
| 9206 |
+
"grad_norm": 22.625,
|
| 9207 |
+
"learning_rate": 1.144406779661017e-05,
|
| 9208 |
+
"loss": 1.3937,
|
| 9209 |
+
"step": 13120
|
| 9210 |
+
},
|
| 9211 |
+
{
|
| 9212 |
+
"epoch": 0.33,
|
| 9213 |
+
"grad_norm": 19.625,
|
| 9214 |
+
"learning_rate": 1.143728813559322e-05,
|
| 9215 |
+
"loss": 1.4812,
|
| 9216 |
+
"step": 13130
|
| 9217 |
+
},
|
| 9218 |
+
{
|
| 9219 |
+
"epoch": 0.33,
|
| 9220 |
+
"grad_norm": 68.5,
|
| 9221 |
+
"learning_rate": 1.1430508474576272e-05,
|
| 9222 |
+
"loss": 1.6372,
|
| 9223 |
+
"step": 13140
|
| 9224 |
+
},
|
| 9225 |
+
{
|
| 9226 |
+
"epoch": 0.33,
|
| 9227 |
+
"grad_norm": 29.75,
|
| 9228 |
+
"learning_rate": 1.1423728813559322e-05,
|
| 9229 |
+
"loss": 1.3724,
|
| 9230 |
+
"step": 13150
|
| 9231 |
+
},
|
| 9232 |
+
{
|
| 9233 |
+
"epoch": 0.33,
|
| 9234 |
+
"grad_norm": 14.25,
|
| 9235 |
+
"learning_rate": 1.1416949152542373e-05,
|
| 9236 |
+
"loss": 1.2524,
|
| 9237 |
+
"step": 13160
|
| 9238 |
+
},
|
| 9239 |
+
{
|
| 9240 |
+
"epoch": 0.33,
|
| 9241 |
+
"grad_norm": 12.375,
|
| 9242 |
+
"learning_rate": 1.1410169491525425e-05,
|
| 9243 |
+
"loss": 1.5147,
|
| 9244 |
+
"step": 13170
|
| 9245 |
+
},
|
| 9246 |
+
{
|
| 9247 |
+
"epoch": 0.33,
|
| 9248 |
+
"grad_norm": 59.75,
|
| 9249 |
+
"learning_rate": 1.1403389830508476e-05,
|
| 9250 |
+
"loss": 1.4632,
|
| 9251 |
+
"step": 13180
|
| 9252 |
+
},
|
| 9253 |
+
{
|
| 9254 |
+
"epoch": 0.33,
|
| 9255 |
+
"grad_norm": 22.125,
|
| 9256 |
+
"learning_rate": 1.1396610169491526e-05,
|
| 9257 |
+
"loss": 1.3722,
|
| 9258 |
+
"step": 13190
|
| 9259 |
+
},
|
| 9260 |
+
{
|
| 9261 |
+
"epoch": 0.33,
|
| 9262 |
+
"grad_norm": 20.5,
|
| 9263 |
+
"learning_rate": 1.1389830508474577e-05,
|
| 9264 |
+
"loss": 1.349,
|
| 9265 |
+
"step": 13200
|
| 9266 |
+
},
|
| 9267 |
+
{
|
| 9268 |
+
"epoch": 0.33,
|
| 9269 |
+
"grad_norm": 17.125,
|
| 9270 |
+
"learning_rate": 1.1383050847457629e-05,
|
| 9271 |
+
"loss": 1.3975,
|
| 9272 |
+
"step": 13210
|
| 9273 |
+
},
|
| 9274 |
+
{
|
| 9275 |
+
"epoch": 0.33,
|
| 9276 |
+
"grad_norm": 32.5,
|
| 9277 |
+
"learning_rate": 1.137627118644068e-05,
|
| 9278 |
+
"loss": 1.4174,
|
| 9279 |
+
"step": 13220
|
| 9280 |
+
},
|
| 9281 |
+
{
|
| 9282 |
+
"epoch": 0.33,
|
| 9283 |
+
"grad_norm": 11.3125,
|
| 9284 |
+
"learning_rate": 1.136949152542373e-05,
|
| 9285 |
+
"loss": 1.2684,
|
| 9286 |
+
"step": 13230
|
| 9287 |
+
},
|
| 9288 |
+
{
|
| 9289 |
+
"epoch": 0.33,
|
| 9290 |
+
"grad_norm": 22.375,
|
| 9291 |
+
"learning_rate": 1.1362711864406781e-05,
|
| 9292 |
+
"loss": 1.3821,
|
| 9293 |
+
"step": 13240
|
| 9294 |
+
},
|
| 9295 |
+
{
|
| 9296 |
+
"epoch": 0.33,
|
| 9297 |
+
"grad_norm": 15.6875,
|
| 9298 |
+
"learning_rate": 1.1355932203389833e-05,
|
| 9299 |
+
"loss": 1.3647,
|
| 9300 |
+
"step": 13250
|
| 9301 |
+
},
|
| 9302 |
+
{
|
| 9303 |
+
"epoch": 0.33,
|
| 9304 |
+
"grad_norm": 9.1875,
|
| 9305 |
+
"learning_rate": 1.1349152542372884e-05,
|
| 9306 |
+
"loss": 1.3544,
|
| 9307 |
+
"step": 13260
|
| 9308 |
+
},
|
| 9309 |
+
{
|
| 9310 |
+
"epoch": 0.33,
|
| 9311 |
+
"grad_norm": 27.375,
|
| 9312 |
+
"learning_rate": 1.1342372881355934e-05,
|
| 9313 |
+
"loss": 1.2971,
|
| 9314 |
+
"step": 13270
|
| 9315 |
+
},
|
| 9316 |
+
{
|
| 9317 |
+
"epoch": 0.33,
|
| 9318 |
+
"grad_norm": 38.75,
|
| 9319 |
+
"learning_rate": 1.1335593220338983e-05,
|
| 9320 |
+
"loss": 1.5632,
|
| 9321 |
+
"step": 13280
|
| 9322 |
+
},
|
| 9323 |
+
{
|
| 9324 |
+
"epoch": 0.33,
|
| 9325 |
+
"grad_norm": 36.0,
|
| 9326 |
+
"learning_rate": 1.1328813559322033e-05,
|
| 9327 |
+
"loss": 1.5077,
|
| 9328 |
+
"step": 13290
|
| 9329 |
+
},
|
| 9330 |
+
{
|
| 9331 |
+
"epoch": 0.33,
|
| 9332 |
+
"grad_norm": 31.0,
|
| 9333 |
+
"learning_rate": 1.1322033898305084e-05,
|
| 9334 |
+
"loss": 1.1637,
|
| 9335 |
+
"step": 13300
|
| 9336 |
+
},
|
| 9337 |
+
{
|
| 9338 |
+
"epoch": 0.33,
|
| 9339 |
+
"grad_norm": 29.25,
|
| 9340 |
+
"learning_rate": 1.1315254237288136e-05,
|
| 9341 |
+
"loss": 1.4482,
|
| 9342 |
+
"step": 13310
|
| 9343 |
+
},
|
| 9344 |
+
{
|
| 9345 |
+
"epoch": 0.33,
|
| 9346 |
+
"grad_norm": 14.0,
|
| 9347 |
+
"learning_rate": 1.1308474576271187e-05,
|
| 9348 |
+
"loss": 1.3316,
|
| 9349 |
+
"step": 13320
|
| 9350 |
+
},
|
| 9351 |
+
{
|
| 9352 |
+
"epoch": 0.33,
|
| 9353 |
+
"grad_norm": 26.375,
|
| 9354 |
+
"learning_rate": 1.1301694915254237e-05,
|
| 9355 |
+
"loss": 1.3351,
|
| 9356 |
+
"step": 13330
|
| 9357 |
+
},
|
| 9358 |
+
{
|
| 9359 |
+
"epoch": 0.33,
|
| 9360 |
+
"grad_norm": 27.0,
|
| 9361 |
+
"learning_rate": 1.1294915254237288e-05,
|
| 9362 |
+
"loss": 1.4783,
|
| 9363 |
+
"step": 13340
|
| 9364 |
+
},
|
| 9365 |
+
{
|
| 9366 |
+
"epoch": 0.33,
|
| 9367 |
+
"grad_norm": 28.5,
|
| 9368 |
+
"learning_rate": 1.128813559322034e-05,
|
| 9369 |
+
"loss": 1.4635,
|
| 9370 |
+
"step": 13350
|
| 9371 |
+
},
|
| 9372 |
+
{
|
| 9373 |
+
"epoch": 0.33,
|
| 9374 |
+
"grad_norm": 28.625,
|
| 9375 |
+
"learning_rate": 1.1281355932203391e-05,
|
| 9376 |
+
"loss": 1.2941,
|
| 9377 |
+
"step": 13360
|
| 9378 |
+
},
|
| 9379 |
+
{
|
| 9380 |
+
"epoch": 0.33,
|
| 9381 |
+
"grad_norm": 21.25,
|
| 9382 |
+
"learning_rate": 1.1274576271186441e-05,
|
| 9383 |
+
"loss": 1.4337,
|
| 9384 |
+
"step": 13370
|
| 9385 |
+
},
|
| 9386 |
+
{
|
| 9387 |
+
"epoch": 0.33,
|
| 9388 |
+
"grad_norm": 9.25,
|
| 9389 |
+
"learning_rate": 1.1267796610169492e-05,
|
| 9390 |
+
"loss": 1.3615,
|
| 9391 |
+
"step": 13380
|
| 9392 |
+
},
|
| 9393 |
+
{
|
| 9394 |
+
"epoch": 0.33,
|
| 9395 |
+
"grad_norm": 30.5,
|
| 9396 |
+
"learning_rate": 1.1261016949152544e-05,
|
| 9397 |
+
"loss": 1.4474,
|
| 9398 |
+
"step": 13390
|
| 9399 |
+
},
|
| 9400 |
+
{
|
| 9401 |
+
"epoch": 0.34,
|
| 9402 |
+
"grad_norm": 29.5,
|
| 9403 |
+
"learning_rate": 1.1254237288135595e-05,
|
| 9404 |
+
"loss": 1.512,
|
| 9405 |
+
"step": 13400
|
| 9406 |
+
},
|
| 9407 |
+
{
|
| 9408 |
+
"epoch": 0.34,
|
| 9409 |
+
"grad_norm": 16.375,
|
| 9410 |
+
"learning_rate": 1.1247457627118645e-05,
|
| 9411 |
+
"loss": 1.3367,
|
| 9412 |
+
"step": 13410
|
| 9413 |
+
},
|
| 9414 |
+
{
|
| 9415 |
+
"epoch": 0.34,
|
| 9416 |
+
"grad_norm": 17.25,
|
| 9417 |
+
"learning_rate": 1.1240677966101696e-05,
|
| 9418 |
+
"loss": 1.3797,
|
| 9419 |
+
"step": 13420
|
| 9420 |
+
},
|
| 9421 |
+
{
|
| 9422 |
+
"epoch": 0.34,
|
| 9423 |
+
"grad_norm": 21.75,
|
| 9424 |
+
"learning_rate": 1.1233898305084748e-05,
|
| 9425 |
+
"loss": 1.3935,
|
| 9426 |
+
"step": 13430
|
| 9427 |
+
},
|
| 9428 |
+
{
|
| 9429 |
+
"epoch": 0.34,
|
| 9430 |
+
"grad_norm": 48.0,
|
| 9431 |
+
"learning_rate": 1.1227118644067799e-05,
|
| 9432 |
+
"loss": 1.2878,
|
| 9433 |
+
"step": 13440
|
| 9434 |
+
},
|
| 9435 |
+
{
|
| 9436 |
+
"epoch": 0.34,
|
| 9437 |
+
"grad_norm": 12.3125,
|
| 9438 |
+
"learning_rate": 1.1220338983050849e-05,
|
| 9439 |
+
"loss": 1.1581,
|
| 9440 |
+
"step": 13450
|
| 9441 |
+
},
|
| 9442 |
+
{
|
| 9443 |
+
"epoch": 0.34,
|
| 9444 |
+
"grad_norm": 17.75,
|
| 9445 |
+
"learning_rate": 1.12135593220339e-05,
|
| 9446 |
+
"loss": 1.4249,
|
| 9447 |
+
"step": 13460
|
| 9448 |
+
},
|
| 9449 |
+
{
|
| 9450 |
+
"epoch": 0.34,
|
| 9451 |
+
"grad_norm": 10.0,
|
| 9452 |
+
"learning_rate": 1.1206779661016951e-05,
|
| 9453 |
+
"loss": 1.3671,
|
| 9454 |
+
"step": 13470
|
| 9455 |
+
},
|
| 9456 |
+
{
|
| 9457 |
+
"epoch": 0.34,
|
| 9458 |
+
"grad_norm": 9.6875,
|
| 9459 |
+
"learning_rate": 1.1200000000000001e-05,
|
| 9460 |
+
"loss": 1.4198,
|
| 9461 |
+
"step": 13480
|
| 9462 |
+
},
|
| 9463 |
+
{
|
| 9464 |
+
"epoch": 0.34,
|
| 9465 |
+
"grad_norm": 25.5,
|
| 9466 |
+
"learning_rate": 1.1193220338983051e-05,
|
| 9467 |
+
"loss": 1.3878,
|
| 9468 |
+
"step": 13490
|
| 9469 |
+
},
|
| 9470 |
+
{
|
| 9471 |
+
"epoch": 0.34,
|
| 9472 |
+
"grad_norm": 26.75,
|
| 9473 |
+
"learning_rate": 1.1186440677966102e-05,
|
| 9474 |
+
"loss": 1.158,
|
| 9475 |
+
"step": 13500
|
| 9476 |
+
},
|
| 9477 |
+
{
|
| 9478 |
+
"epoch": 0.34,
|
| 9479 |
+
"grad_norm": 31.375,
|
| 9480 |
+
"learning_rate": 1.1179661016949152e-05,
|
| 9481 |
+
"loss": 1.3105,
|
| 9482 |
+
"step": 13510
|
| 9483 |
+
},
|
| 9484 |
+
{
|
| 9485 |
+
"epoch": 0.34,
|
| 9486 |
+
"grad_norm": 15.75,
|
| 9487 |
+
"learning_rate": 1.1172881355932203e-05,
|
| 9488 |
+
"loss": 1.5006,
|
| 9489 |
+
"step": 13520
|
| 9490 |
+
},
|
| 9491 |
+
{
|
| 9492 |
+
"epoch": 0.34,
|
| 9493 |
+
"grad_norm": 72.0,
|
| 9494 |
+
"learning_rate": 1.1166101694915255e-05,
|
| 9495 |
+
"loss": 1.4191,
|
| 9496 |
+
"step": 13530
|
| 9497 |
+
},
|
| 9498 |
+
{
|
| 9499 |
+
"epoch": 0.34,
|
| 9500 |
+
"grad_norm": 20.875,
|
| 9501 |
+
"learning_rate": 1.1159322033898306e-05,
|
| 9502 |
+
"loss": 1.3523,
|
| 9503 |
+
"step": 13540
|
| 9504 |
+
},
|
| 9505 |
+
{
|
| 9506 |
+
"epoch": 0.34,
|
| 9507 |
+
"grad_norm": 22.25,
|
| 9508 |
+
"learning_rate": 1.1152542372881356e-05,
|
| 9509 |
+
"loss": 1.3726,
|
| 9510 |
+
"step": 13550
|
| 9511 |
+
},
|
| 9512 |
+
{
|
| 9513 |
+
"epoch": 0.34,
|
| 9514 |
+
"grad_norm": 48.25,
|
| 9515 |
+
"learning_rate": 1.1145762711864407e-05,
|
| 9516 |
+
"loss": 1.3706,
|
| 9517 |
+
"step": 13560
|
| 9518 |
+
},
|
| 9519 |
+
{
|
| 9520 |
+
"epoch": 0.34,
|
| 9521 |
+
"grad_norm": 39.25,
|
| 9522 |
+
"learning_rate": 1.1138983050847459e-05,
|
| 9523 |
+
"loss": 1.3988,
|
| 9524 |
+
"step": 13570
|
| 9525 |
+
},
|
| 9526 |
+
{
|
| 9527 |
+
"epoch": 0.34,
|
| 9528 |
+
"grad_norm": 34.0,
|
| 9529 |
+
"learning_rate": 1.113220338983051e-05,
|
| 9530 |
+
"loss": 1.36,
|
| 9531 |
+
"step": 13580
|
| 9532 |
+
},
|
| 9533 |
+
{
|
| 9534 |
+
"epoch": 0.34,
|
| 9535 |
+
"grad_norm": 18.25,
|
| 9536 |
+
"learning_rate": 1.112542372881356e-05,
|
| 9537 |
+
"loss": 1.5547,
|
| 9538 |
+
"step": 13590
|
| 9539 |
+
},
|
| 9540 |
+
{
|
| 9541 |
+
"epoch": 0.34,
|
| 9542 |
+
"grad_norm": 27.875,
|
| 9543 |
+
"learning_rate": 1.1118644067796611e-05,
|
| 9544 |
+
"loss": 1.2315,
|
| 9545 |
+
"step": 13600
|
| 9546 |
+
},
|
| 9547 |
+
{
|
| 9548 |
+
"epoch": 0.34,
|
| 9549 |
+
"grad_norm": 26.375,
|
| 9550 |
+
"learning_rate": 1.1111864406779663e-05,
|
| 9551 |
+
"loss": 1.1956,
|
| 9552 |
+
"step": 13610
|
| 9553 |
+
},
|
| 9554 |
+
{
|
| 9555 |
+
"epoch": 0.34,
|
| 9556 |
+
"grad_norm": 13.375,
|
| 9557 |
+
"learning_rate": 1.1105084745762712e-05,
|
| 9558 |
+
"loss": 1.2934,
|
| 9559 |
+
"step": 13620
|
| 9560 |
+
},
|
| 9561 |
+
{
|
| 9562 |
+
"epoch": 0.34,
|
| 9563 |
+
"grad_norm": 26.625,
|
| 9564 |
+
"learning_rate": 1.1098305084745764e-05,
|
| 9565 |
+
"loss": 1.416,
|
| 9566 |
+
"step": 13630
|
| 9567 |
+
},
|
| 9568 |
+
{
|
| 9569 |
+
"epoch": 0.34,
|
| 9570 |
+
"grad_norm": 17.875,
|
| 9571 |
+
"learning_rate": 1.1091525423728815e-05,
|
| 9572 |
+
"loss": 1.2588,
|
| 9573 |
+
"step": 13640
|
| 9574 |
+
},
|
| 9575 |
+
{
|
| 9576 |
+
"epoch": 0.34,
|
| 9577 |
+
"grad_norm": 10.5625,
|
| 9578 |
+
"learning_rate": 1.1084745762711867e-05,
|
| 9579 |
+
"loss": 1.3671,
|
| 9580 |
+
"step": 13650
|
| 9581 |
+
},
|
| 9582 |
+
{
|
| 9583 |
+
"epoch": 0.34,
|
| 9584 |
+
"grad_norm": 14.3125,
|
| 9585 |
+
"learning_rate": 1.1077966101694916e-05,
|
| 9586 |
+
"loss": 1.4697,
|
| 9587 |
+
"step": 13660
|
| 9588 |
+
},
|
| 9589 |
+
{
|
| 9590 |
+
"epoch": 0.34,
|
| 9591 |
+
"grad_norm": 18.625,
|
| 9592 |
+
"learning_rate": 1.1071186440677968e-05,
|
| 9593 |
+
"loss": 1.2121,
|
| 9594 |
+
"step": 13670
|
| 9595 |
+
},
|
| 9596 |
+
{
|
| 9597 |
+
"epoch": 0.34,
|
| 9598 |
+
"grad_norm": 21.25,
|
| 9599 |
+
"learning_rate": 1.1064406779661019e-05,
|
| 9600 |
+
"loss": 1.2724,
|
| 9601 |
+
"step": 13680
|
| 9602 |
+
},
|
| 9603 |
+
{
|
| 9604 |
+
"epoch": 0.34,
|
| 9605 |
+
"grad_norm": 21.75,
|
| 9606 |
+
"learning_rate": 1.105762711864407e-05,
|
| 9607 |
+
"loss": 1.3785,
|
| 9608 |
+
"step": 13690
|
| 9609 |
+
},
|
| 9610 |
+
{
|
| 9611 |
+
"epoch": 0.34,
|
| 9612 |
+
"grad_norm": 14.75,
|
| 9613 |
+
"learning_rate": 1.1050847457627118e-05,
|
| 9614 |
+
"loss": 1.5127,
|
| 9615 |
+
"step": 13700
|
| 9616 |
+
},
|
| 9617 |
+
{
|
| 9618 |
+
"epoch": 0.34,
|
| 9619 |
+
"grad_norm": 26.375,
|
| 9620 |
+
"learning_rate": 1.104406779661017e-05,
|
| 9621 |
+
"loss": 1.5337,
|
| 9622 |
+
"step": 13710
|
| 9623 |
+
},
|
| 9624 |
+
{
|
| 9625 |
+
"epoch": 0.34,
|
| 9626 |
+
"grad_norm": 18.125,
|
| 9627 |
+
"learning_rate": 1.1037288135593221e-05,
|
| 9628 |
+
"loss": 1.4343,
|
| 9629 |
+
"step": 13720
|
| 9630 |
+
},
|
| 9631 |
+
{
|
| 9632 |
+
"epoch": 0.34,
|
| 9633 |
+
"grad_norm": 31.0,
|
| 9634 |
+
"learning_rate": 1.1030508474576271e-05,
|
| 9635 |
+
"loss": 1.4872,
|
| 9636 |
+
"step": 13730
|
| 9637 |
+
},
|
| 9638 |
+
{
|
| 9639 |
+
"epoch": 0.34,
|
| 9640 |
+
"grad_norm": 24.125,
|
| 9641 |
+
"learning_rate": 1.1023728813559322e-05,
|
| 9642 |
+
"loss": 1.3495,
|
| 9643 |
+
"step": 13740
|
| 9644 |
+
},
|
| 9645 |
+
{
|
| 9646 |
+
"epoch": 0.34,
|
| 9647 |
+
"grad_norm": 12.125,
|
| 9648 |
+
"learning_rate": 1.1016949152542374e-05,
|
| 9649 |
+
"loss": 1.2934,
|
| 9650 |
+
"step": 13750
|
| 9651 |
+
},
|
| 9652 |
+
{
|
| 9653 |
+
"epoch": 0.34,
|
| 9654 |
+
"grad_norm": 34.25,
|
| 9655 |
+
"learning_rate": 1.1010169491525423e-05,
|
| 9656 |
+
"loss": 1.4117,
|
| 9657 |
+
"step": 13760
|
| 9658 |
+
},
|
| 9659 |
+
{
|
| 9660 |
+
"epoch": 0.34,
|
| 9661 |
+
"grad_norm": 19.625,
|
| 9662 |
+
"learning_rate": 1.1003389830508475e-05,
|
| 9663 |
+
"loss": 1.5345,
|
| 9664 |
+
"step": 13770
|
| 9665 |
+
},
|
| 9666 |
+
{
|
| 9667 |
+
"epoch": 0.34,
|
| 9668 |
+
"grad_norm": 21.125,
|
| 9669 |
+
"learning_rate": 1.0996610169491526e-05,
|
| 9670 |
+
"loss": 1.4747,
|
| 9671 |
+
"step": 13780
|
| 9672 |
+
},
|
| 9673 |
+
{
|
| 9674 |
+
"epoch": 0.34,
|
| 9675 |
+
"grad_norm": 28.5,
|
| 9676 |
+
"learning_rate": 1.0989830508474578e-05,
|
| 9677 |
+
"loss": 1.3429,
|
| 9678 |
+
"step": 13790
|
| 9679 |
+
},
|
| 9680 |
+
{
|
| 9681 |
+
"epoch": 0.34,
|
| 9682 |
+
"grad_norm": 15.4375,
|
| 9683 |
+
"learning_rate": 1.0983050847457627e-05,
|
| 9684 |
+
"loss": 1.0869,
|
| 9685 |
+
"step": 13800
|
| 9686 |
+
},
|
| 9687 |
+
{
|
| 9688 |
+
"epoch": 0.35,
|
| 9689 |
+
"grad_norm": 35.25,
|
| 9690 |
+
"learning_rate": 1.0976271186440679e-05,
|
| 9691 |
+
"loss": 1.5203,
|
| 9692 |
+
"step": 13810
|
| 9693 |
+
},
|
| 9694 |
+
{
|
| 9695 |
+
"epoch": 0.35,
|
| 9696 |
+
"grad_norm": 55.25,
|
| 9697 |
+
"learning_rate": 1.096949152542373e-05,
|
| 9698 |
+
"loss": 1.3117,
|
| 9699 |
+
"step": 13820
|
| 9700 |
+
},
|
| 9701 |
+
{
|
| 9702 |
+
"epoch": 0.35,
|
| 9703 |
+
"grad_norm": 14.3125,
|
| 9704 |
+
"learning_rate": 1.0962711864406782e-05,
|
| 9705 |
+
"loss": 1.4044,
|
| 9706 |
+
"step": 13830
|
| 9707 |
+
},
|
| 9708 |
+
{
|
| 9709 |
+
"epoch": 0.35,
|
| 9710 |
+
"grad_norm": 15.125,
|
| 9711 |
+
"learning_rate": 1.0955932203389831e-05,
|
| 9712 |
+
"loss": 1.5186,
|
| 9713 |
+
"step": 13840
|
| 9714 |
+
},
|
| 9715 |
+
{
|
| 9716 |
+
"epoch": 0.35,
|
| 9717 |
+
"grad_norm": 20.5,
|
| 9718 |
+
"learning_rate": 1.0949152542372883e-05,
|
| 9719 |
+
"loss": 1.251,
|
| 9720 |
+
"step": 13850
|
| 9721 |
+
},
|
| 9722 |
+
{
|
| 9723 |
+
"epoch": 0.35,
|
| 9724 |
+
"grad_norm": 42.75,
|
| 9725 |
+
"learning_rate": 1.0942372881355934e-05,
|
| 9726 |
+
"loss": 1.3674,
|
| 9727 |
+
"step": 13860
|
| 9728 |
+
},
|
| 9729 |
+
{
|
| 9730 |
+
"epoch": 0.35,
|
| 9731 |
+
"grad_norm": 12.0,
|
| 9732 |
+
"learning_rate": 1.0935593220338985e-05,
|
| 9733 |
+
"loss": 1.4548,
|
| 9734 |
+
"step": 13870
|
| 9735 |
+
},
|
| 9736 |
+
{
|
| 9737 |
+
"epoch": 0.35,
|
| 9738 |
+
"grad_norm": 18.5,
|
| 9739 |
+
"learning_rate": 1.0928813559322035e-05,
|
| 9740 |
+
"loss": 1.2881,
|
| 9741 |
+
"step": 13880
|
| 9742 |
+
},
|
| 9743 |
+
{
|
| 9744 |
+
"epoch": 0.35,
|
| 9745 |
+
"grad_norm": 47.0,
|
| 9746 |
+
"learning_rate": 1.0922033898305087e-05,
|
| 9747 |
+
"loss": 1.4235,
|
| 9748 |
+
"step": 13890
|
| 9749 |
+
},
|
| 9750 |
+
{
|
| 9751 |
+
"epoch": 0.35,
|
| 9752 |
+
"grad_norm": 15.9375,
|
| 9753 |
+
"learning_rate": 1.0915254237288135e-05,
|
| 9754 |
+
"loss": 1.267,
|
| 9755 |
+
"step": 13900
|
| 9756 |
+
},
|
| 9757 |
+
{
|
| 9758 |
+
"epoch": 0.35,
|
| 9759 |
+
"grad_norm": 30.75,
|
| 9760 |
+
"learning_rate": 1.0908474576271186e-05,
|
| 9761 |
+
"loss": 1.2051,
|
| 9762 |
+
"step": 13910
|
| 9763 |
+
},
|
| 9764 |
+
{
|
| 9765 |
+
"epoch": 0.35,
|
| 9766 |
+
"grad_norm": 21.625,
|
| 9767 |
+
"learning_rate": 1.0901694915254237e-05,
|
| 9768 |
+
"loss": 1.4902,
|
| 9769 |
+
"step": 13920
|
| 9770 |
+
},
|
| 9771 |
+
{
|
| 9772 |
+
"epoch": 0.35,
|
| 9773 |
+
"grad_norm": 19.0,
|
| 9774 |
+
"learning_rate": 1.0894915254237289e-05,
|
| 9775 |
+
"loss": 1.0872,
|
| 9776 |
+
"step": 13930
|
| 9777 |
+
},
|
| 9778 |
+
{
|
| 9779 |
+
"epoch": 0.35,
|
| 9780 |
+
"grad_norm": 48.25,
|
| 9781 |
+
"learning_rate": 1.0888135593220339e-05,
|
| 9782 |
+
"loss": 1.3268,
|
| 9783 |
+
"step": 13940
|
| 9784 |
+
},
|
| 9785 |
+
{
|
| 9786 |
+
"epoch": 0.35,
|
| 9787 |
+
"grad_norm": 20.0,
|
| 9788 |
+
"learning_rate": 1.088135593220339e-05,
|
| 9789 |
+
"loss": 1.3066,
|
| 9790 |
+
"step": 13950
|
| 9791 |
+
},
|
| 9792 |
+
{
|
| 9793 |
+
"epoch": 0.35,
|
| 9794 |
+
"grad_norm": 12.1875,
|
| 9795 |
+
"learning_rate": 1.0874576271186441e-05,
|
| 9796 |
+
"loss": 1.3713,
|
| 9797 |
+
"step": 13960
|
| 9798 |
+
},
|
| 9799 |
+
{
|
| 9800 |
+
"epoch": 0.35,
|
| 9801 |
+
"grad_norm": 21.375,
|
| 9802 |
+
"learning_rate": 1.0867796610169493e-05,
|
| 9803 |
+
"loss": 1.4468,
|
| 9804 |
+
"step": 13970
|
| 9805 |
+
},
|
| 9806 |
+
{
|
| 9807 |
+
"epoch": 0.35,
|
| 9808 |
+
"grad_norm": 17.0,
|
| 9809 |
+
"learning_rate": 1.0861016949152542e-05,
|
| 9810 |
+
"loss": 1.4413,
|
| 9811 |
+
"step": 13980
|
| 9812 |
+
},
|
| 9813 |
+
{
|
| 9814 |
+
"epoch": 0.35,
|
| 9815 |
+
"grad_norm": 21.0,
|
| 9816 |
+
"learning_rate": 1.0854237288135594e-05,
|
| 9817 |
+
"loss": 1.3061,
|
| 9818 |
+
"step": 13990
|
| 9819 |
+
},
|
| 9820 |
+
{
|
| 9821 |
+
"epoch": 0.35,
|
| 9822 |
+
"grad_norm": 20.875,
|
| 9823 |
+
"learning_rate": 1.0847457627118645e-05,
|
| 9824 |
+
"loss": 1.4503,
|
| 9825 |
+
"step": 14000
|
| 9826 |
+
},
|
| 9827 |
+
{
|
| 9828 |
+
"epoch": 0.35,
|
| 9829 |
+
"grad_norm": 15.125,
|
| 9830 |
+
"learning_rate": 1.0840677966101697e-05,
|
| 9831 |
+
"loss": 1.2215,
|
| 9832 |
+
"step": 14010
|
| 9833 |
+
},
|
| 9834 |
+
{
|
| 9835 |
+
"epoch": 0.35,
|
| 9836 |
+
"grad_norm": 24.625,
|
| 9837 |
+
"learning_rate": 1.0833898305084746e-05,
|
| 9838 |
+
"loss": 1.4355,
|
| 9839 |
+
"step": 14020
|
| 9840 |
+
},
|
| 9841 |
+
{
|
| 9842 |
+
"epoch": 0.35,
|
| 9843 |
+
"grad_norm": 13.8125,
|
| 9844 |
+
"learning_rate": 1.0827118644067798e-05,
|
| 9845 |
+
"loss": 1.4657,
|
| 9846 |
+
"step": 14030
|
| 9847 |
+
},
|
| 9848 |
+
{
|
| 9849 |
+
"epoch": 0.35,
|
| 9850 |
+
"grad_norm": 26.25,
|
| 9851 |
+
"learning_rate": 1.0820338983050849e-05,
|
| 9852 |
+
"loss": 1.268,
|
| 9853 |
+
"step": 14040
|
| 9854 |
+
},
|
| 9855 |
+
{
|
| 9856 |
+
"epoch": 0.35,
|
| 9857 |
+
"grad_norm": 20.5,
|
| 9858 |
+
"learning_rate": 1.08135593220339e-05,
|
| 9859 |
+
"loss": 1.306,
|
| 9860 |
+
"step": 14050
|
| 9861 |
+
},
|
| 9862 |
+
{
|
| 9863 |
+
"epoch": 0.35,
|
| 9864 |
+
"grad_norm": 12.8125,
|
| 9865 |
+
"learning_rate": 1.080677966101695e-05,
|
| 9866 |
+
"loss": 1.4965,
|
| 9867 |
+
"step": 14060
|
| 9868 |
+
},
|
| 9869 |
+
{
|
| 9870 |
+
"epoch": 0.35,
|
| 9871 |
+
"grad_norm": 18.5,
|
| 9872 |
+
"learning_rate": 1.0800000000000002e-05,
|
| 9873 |
+
"loss": 1.3629,
|
| 9874 |
+
"step": 14070
|
| 9875 |
+
},
|
| 9876 |
+
{
|
| 9877 |
+
"epoch": 0.35,
|
| 9878 |
+
"grad_norm": 32.0,
|
| 9879 |
+
"learning_rate": 1.0793220338983053e-05,
|
| 9880 |
+
"loss": 1.2364,
|
| 9881 |
+
"step": 14080
|
| 9882 |
+
},
|
| 9883 |
+
{
|
| 9884 |
+
"epoch": 0.35,
|
| 9885 |
+
"grad_norm": 14.4375,
|
| 9886 |
+
"learning_rate": 1.0786440677966103e-05,
|
| 9887 |
+
"loss": 1.334,
|
| 9888 |
+
"step": 14090
|
| 9889 |
+
},
|
| 9890 |
+
{
|
| 9891 |
+
"epoch": 0.35,
|
| 9892 |
+
"grad_norm": 11.875,
|
| 9893 |
+
"learning_rate": 1.0779661016949154e-05,
|
| 9894 |
+
"loss": 1.4108,
|
| 9895 |
+
"step": 14100
|
| 9896 |
+
},
|
| 9897 |
+
{
|
| 9898 |
+
"epoch": 0.35,
|
| 9899 |
+
"grad_norm": 35.5,
|
| 9900 |
+
"learning_rate": 1.0772881355932204e-05,
|
| 9901 |
+
"loss": 1.4491,
|
| 9902 |
+
"step": 14110
|
| 9903 |
+
},
|
| 9904 |
+
{
|
| 9905 |
+
"epoch": 0.35,
|
| 9906 |
+
"grad_norm": 16.25,
|
| 9907 |
+
"learning_rate": 1.0766101694915254e-05,
|
| 9908 |
+
"loss": 1.4305,
|
| 9909 |
+
"step": 14120
|
| 9910 |
+
},
|
| 9911 |
+
{
|
| 9912 |
+
"epoch": 0.35,
|
| 9913 |
+
"grad_norm": 22.75,
|
| 9914 |
+
"learning_rate": 1.0759322033898305e-05,
|
| 9915 |
+
"loss": 1.324,
|
| 9916 |
+
"step": 14130
|
| 9917 |
+
},
|
| 9918 |
+
{
|
| 9919 |
+
"epoch": 0.35,
|
| 9920 |
+
"grad_norm": 16.625,
|
| 9921 |
+
"learning_rate": 1.0752542372881356e-05,
|
| 9922 |
+
"loss": 1.1912,
|
| 9923 |
+
"step": 14140
|
| 9924 |
+
},
|
| 9925 |
+
{
|
| 9926 |
+
"epoch": 0.35,
|
| 9927 |
+
"grad_norm": 24.875,
|
| 9928 |
+
"learning_rate": 1.0745762711864408e-05,
|
| 9929 |
+
"loss": 1.3567,
|
| 9930 |
+
"step": 14150
|
| 9931 |
+
},
|
| 9932 |
+
{
|
| 9933 |
+
"epoch": 0.35,
|
| 9934 |
+
"grad_norm": 15.25,
|
| 9935 |
+
"learning_rate": 1.0738983050847457e-05,
|
| 9936 |
+
"loss": 1.3639,
|
| 9937 |
+
"step": 14160
|
| 9938 |
+
},
|
| 9939 |
+
{
|
| 9940 |
+
"epoch": 0.35,
|
| 9941 |
+
"grad_norm": 11.5625,
|
| 9942 |
+
"learning_rate": 1.0732203389830509e-05,
|
| 9943 |
+
"loss": 1.4455,
|
| 9944 |
+
"step": 14170
|
| 9945 |
+
},
|
| 9946 |
+
{
|
| 9947 |
+
"epoch": 0.35,
|
| 9948 |
+
"grad_norm": 13.4375,
|
| 9949 |
+
"learning_rate": 1.072542372881356e-05,
|
| 9950 |
+
"loss": 1.4245,
|
| 9951 |
+
"step": 14180
|
| 9952 |
+
},
|
| 9953 |
+
{
|
| 9954 |
+
"epoch": 0.35,
|
| 9955 |
+
"grad_norm": 12.9375,
|
| 9956 |
+
"learning_rate": 1.0718644067796612e-05,
|
| 9957 |
+
"loss": 1.303,
|
| 9958 |
+
"step": 14190
|
| 9959 |
+
},
|
| 9960 |
+
{
|
| 9961 |
+
"epoch": 0.35,
|
| 9962 |
+
"grad_norm": 22.125,
|
| 9963 |
+
"learning_rate": 1.0711864406779661e-05,
|
| 9964 |
+
"loss": 1.3533,
|
| 9965 |
+
"step": 14200
|
| 9966 |
+
},
|
| 9967 |
+
{
|
| 9968 |
+
"epoch": 0.36,
|
| 9969 |
+
"grad_norm": 13.9375,
|
| 9970 |
+
"learning_rate": 1.0705084745762713e-05,
|
| 9971 |
+
"loss": 1.3825,
|
| 9972 |
+
"step": 14210
|
| 9973 |
+
},
|
| 9974 |
+
{
|
| 9975 |
+
"epoch": 0.36,
|
| 9976 |
+
"grad_norm": 30.875,
|
| 9977 |
+
"learning_rate": 1.0698305084745764e-05,
|
| 9978 |
+
"loss": 1.3004,
|
| 9979 |
+
"step": 14220
|
| 9980 |
+
},
|
| 9981 |
+
{
|
| 9982 |
+
"epoch": 0.36,
|
| 9983 |
+
"grad_norm": 22.625,
|
| 9984 |
+
"learning_rate": 1.0691525423728814e-05,
|
| 9985 |
+
"loss": 1.3967,
|
| 9986 |
+
"step": 14230
|
| 9987 |
+
},
|
| 9988 |
+
{
|
| 9989 |
+
"epoch": 0.36,
|
| 9990 |
+
"grad_norm": 13.125,
|
| 9991 |
+
"learning_rate": 1.0684745762711865e-05,
|
| 9992 |
+
"loss": 1.2226,
|
| 9993 |
+
"step": 14240
|
| 9994 |
+
},
|
| 9995 |
+
{
|
| 9996 |
+
"epoch": 0.36,
|
| 9997 |
+
"grad_norm": 33.25,
|
| 9998 |
+
"learning_rate": 1.0677966101694917e-05,
|
| 9999 |
+
"loss": 1.3159,
|
| 10000 |
+
"step": 14250
|
| 10001 |
+
},
|
| 10002 |
+
{
|
| 10003 |
+
"epoch": 0.36,
|
| 10004 |
+
"grad_norm": 20.0,
|
| 10005 |
+
"learning_rate": 1.0671186440677968e-05,
|
| 10006 |
+
"loss": 1.239,
|
| 10007 |
+
"step": 14260
|
| 10008 |
+
},
|
| 10009 |
+
{
|
| 10010 |
+
"epoch": 0.36,
|
| 10011 |
+
"grad_norm": 16.25,
|
| 10012 |
+
"learning_rate": 1.0664406779661018e-05,
|
| 10013 |
+
"loss": 1.3726,
|
| 10014 |
+
"step": 14270
|
| 10015 |
+
},
|
| 10016 |
+
{
|
| 10017 |
+
"epoch": 0.36,
|
| 10018 |
+
"grad_norm": 16.625,
|
| 10019 |
+
"learning_rate": 1.065762711864407e-05,
|
| 10020 |
+
"loss": 1.3982,
|
| 10021 |
+
"step": 14280
|
| 10022 |
+
},
|
| 10023 |
+
{
|
| 10024 |
+
"epoch": 0.36,
|
| 10025 |
+
"grad_norm": 23.875,
|
| 10026 |
+
"learning_rate": 1.065084745762712e-05,
|
| 10027 |
+
"loss": 1.4308,
|
| 10028 |
+
"step": 14290
|
| 10029 |
+
},
|
| 10030 |
+
{
|
| 10031 |
+
"epoch": 0.36,
|
| 10032 |
+
"grad_norm": 15.0625,
|
| 10033 |
+
"learning_rate": 1.0644067796610172e-05,
|
| 10034 |
+
"loss": 1.299,
|
| 10035 |
+
"step": 14300
|
| 10036 |
+
},
|
| 10037 |
+
{
|
| 10038 |
+
"epoch": 0.36,
|
| 10039 |
+
"grad_norm": 40.5,
|
| 10040 |
+
"learning_rate": 1.0637288135593222e-05,
|
| 10041 |
+
"loss": 1.3387,
|
| 10042 |
+
"step": 14310
|
| 10043 |
+
},
|
| 10044 |
+
{
|
| 10045 |
+
"epoch": 0.36,
|
| 10046 |
+
"grad_norm": 29.375,
|
| 10047 |
+
"learning_rate": 1.0630508474576271e-05,
|
| 10048 |
+
"loss": 1.5603,
|
| 10049 |
+
"step": 14320
|
| 10050 |
+
},
|
| 10051 |
+
{
|
| 10052 |
+
"epoch": 0.36,
|
| 10053 |
+
"grad_norm": 28.0,
|
| 10054 |
+
"learning_rate": 1.0623728813559323e-05,
|
| 10055 |
+
"loss": 1.4149,
|
| 10056 |
+
"step": 14330
|
| 10057 |
+
},
|
| 10058 |
+
{
|
| 10059 |
+
"epoch": 0.36,
|
| 10060 |
+
"grad_norm": 51.5,
|
| 10061 |
+
"learning_rate": 1.0616949152542373e-05,
|
| 10062 |
+
"loss": 1.4769,
|
| 10063 |
+
"step": 14340
|
| 10064 |
+
},
|
| 10065 |
+
{
|
| 10066 |
+
"epoch": 0.36,
|
| 10067 |
+
"grad_norm": 18.125,
|
| 10068 |
+
"learning_rate": 1.0610169491525424e-05,
|
| 10069 |
+
"loss": 1.5026,
|
| 10070 |
+
"step": 14350
|
| 10071 |
+
},
|
| 10072 |
+
{
|
| 10073 |
+
"epoch": 0.36,
|
| 10074 |
+
"grad_norm": 30.5,
|
| 10075 |
+
"learning_rate": 1.0603389830508475e-05,
|
| 10076 |
+
"loss": 1.2405,
|
| 10077 |
+
"step": 14360
|
| 10078 |
+
},
|
| 10079 |
+
{
|
| 10080 |
+
"epoch": 0.36,
|
| 10081 |
+
"grad_norm": 19.625,
|
| 10082 |
+
"learning_rate": 1.0596610169491525e-05,
|
| 10083 |
+
"loss": 1.3363,
|
| 10084 |
+
"step": 14370
|
| 10085 |
+
},
|
| 10086 |
+
{
|
| 10087 |
+
"epoch": 0.36,
|
| 10088 |
+
"grad_norm": 20.75,
|
| 10089 |
+
"learning_rate": 1.0589830508474576e-05,
|
| 10090 |
+
"loss": 1.4854,
|
| 10091 |
+
"step": 14380
|
| 10092 |
+
},
|
| 10093 |
+
{
|
| 10094 |
+
"epoch": 0.36,
|
| 10095 |
+
"grad_norm": 17.875,
|
| 10096 |
+
"learning_rate": 1.0583050847457628e-05,
|
| 10097 |
+
"loss": 1.5275,
|
| 10098 |
+
"step": 14390
|
| 10099 |
+
},
|
| 10100 |
+
{
|
| 10101 |
+
"epoch": 0.36,
|
| 10102 |
+
"grad_norm": 20.625,
|
| 10103 |
+
"learning_rate": 1.057627118644068e-05,
|
| 10104 |
+
"loss": 1.3683,
|
| 10105 |
+
"step": 14400
|
| 10106 |
+
},
|
| 10107 |
+
{
|
| 10108 |
+
"epoch": 0.36,
|
| 10109 |
+
"grad_norm": 27.0,
|
| 10110 |
+
"learning_rate": 1.0569491525423729e-05,
|
| 10111 |
+
"loss": 1.3453,
|
| 10112 |
+
"step": 14410
|
| 10113 |
+
},
|
| 10114 |
+
{
|
| 10115 |
+
"epoch": 0.36,
|
| 10116 |
+
"grad_norm": 15.9375,
|
| 10117 |
+
"learning_rate": 1.056271186440678e-05,
|
| 10118 |
+
"loss": 1.4344,
|
| 10119 |
+
"step": 14420
|
| 10120 |
+
},
|
| 10121 |
+
{
|
| 10122 |
+
"epoch": 0.36,
|
| 10123 |
+
"grad_norm": 20.5,
|
| 10124 |
+
"learning_rate": 1.0555932203389832e-05,
|
| 10125 |
+
"loss": 1.4687,
|
| 10126 |
+
"step": 14430
|
| 10127 |
+
},
|
| 10128 |
+
{
|
| 10129 |
+
"epoch": 0.36,
|
| 10130 |
+
"grad_norm": 43.5,
|
| 10131 |
+
"learning_rate": 1.0549152542372883e-05,
|
| 10132 |
+
"loss": 1.6406,
|
| 10133 |
+
"step": 14440
|
| 10134 |
+
},
|
| 10135 |
+
{
|
| 10136 |
+
"epoch": 0.36,
|
| 10137 |
+
"grad_norm": 21.25,
|
| 10138 |
+
"learning_rate": 1.0542372881355933e-05,
|
| 10139 |
+
"loss": 1.4751,
|
| 10140 |
+
"step": 14450
|
| 10141 |
+
},
|
| 10142 |
+
{
|
| 10143 |
+
"epoch": 0.36,
|
| 10144 |
+
"grad_norm": 11.5625,
|
| 10145 |
+
"learning_rate": 1.0535593220338984e-05,
|
| 10146 |
+
"loss": 1.3908,
|
| 10147 |
+
"step": 14460
|
| 10148 |
+
},
|
| 10149 |
+
{
|
| 10150 |
+
"epoch": 0.36,
|
| 10151 |
+
"grad_norm": 33.25,
|
| 10152 |
+
"learning_rate": 1.0528813559322036e-05,
|
| 10153 |
+
"loss": 1.3654,
|
| 10154 |
+
"step": 14470
|
| 10155 |
+
},
|
| 10156 |
+
{
|
| 10157 |
+
"epoch": 0.36,
|
| 10158 |
+
"grad_norm": 12.5625,
|
| 10159 |
+
"learning_rate": 1.0522033898305087e-05,
|
| 10160 |
+
"loss": 1.3358,
|
| 10161 |
+
"step": 14480
|
| 10162 |
+
},
|
| 10163 |
+
{
|
| 10164 |
+
"epoch": 0.36,
|
| 10165 |
+
"grad_norm": 12.5,
|
| 10166 |
+
"learning_rate": 1.0515254237288137e-05,
|
| 10167 |
+
"loss": 1.2108,
|
| 10168 |
+
"step": 14490
|
| 10169 |
+
},
|
| 10170 |
+
{
|
| 10171 |
+
"epoch": 0.36,
|
| 10172 |
+
"grad_norm": 8.625,
|
| 10173 |
+
"learning_rate": 1.0508474576271188e-05,
|
| 10174 |
+
"loss": 1.483,
|
| 10175 |
+
"step": 14500
|
| 10176 |
+
},
|
| 10177 |
+
{
|
| 10178 |
+
"epoch": 0.36,
|
| 10179 |
+
"grad_norm": 25.75,
|
| 10180 |
+
"learning_rate": 1.050169491525424e-05,
|
| 10181 |
+
"loss": 1.4243,
|
| 10182 |
+
"step": 14510
|
| 10183 |
+
},
|
| 10184 |
+
{
|
| 10185 |
+
"epoch": 0.36,
|
| 10186 |
+
"grad_norm": 12.75,
|
| 10187 |
+
"learning_rate": 1.049491525423729e-05,
|
| 10188 |
+
"loss": 1.4495,
|
| 10189 |
+
"step": 14520
|
| 10190 |
+
},
|
| 10191 |
+
{
|
| 10192 |
+
"epoch": 0.36,
|
| 10193 |
+
"grad_norm": 26.375,
|
| 10194 |
+
"learning_rate": 1.0488135593220339e-05,
|
| 10195 |
+
"loss": 1.3261,
|
| 10196 |
+
"step": 14530
|
| 10197 |
+
},
|
| 10198 |
+
{
|
| 10199 |
+
"epoch": 0.36,
|
| 10200 |
+
"grad_norm": 17.0,
|
| 10201 |
+
"learning_rate": 1.048135593220339e-05,
|
| 10202 |
+
"loss": 1.3303,
|
| 10203 |
+
"step": 14540
|
| 10204 |
+
},
|
| 10205 |
+
{
|
| 10206 |
+
"epoch": 0.36,
|
| 10207 |
+
"grad_norm": 16.625,
|
| 10208 |
+
"learning_rate": 1.047457627118644e-05,
|
| 10209 |
+
"loss": 1.2757,
|
| 10210 |
+
"step": 14550
|
| 10211 |
+
},
|
| 10212 |
+
{
|
| 10213 |
+
"epoch": 0.36,
|
| 10214 |
+
"grad_norm": 41.0,
|
| 10215 |
+
"learning_rate": 1.0467796610169491e-05,
|
| 10216 |
+
"loss": 1.2599,
|
| 10217 |
+
"step": 14560
|
| 10218 |
+
},
|
| 10219 |
+
{
|
| 10220 |
+
"epoch": 0.36,
|
| 10221 |
+
"grad_norm": 18.25,
|
| 10222 |
+
"learning_rate": 1.0461016949152543e-05,
|
| 10223 |
+
"loss": 1.3105,
|
| 10224 |
+
"step": 14570
|
| 10225 |
+
},
|
| 10226 |
+
{
|
| 10227 |
+
"epoch": 0.36,
|
| 10228 |
+
"grad_norm": 29.375,
|
| 10229 |
+
"learning_rate": 1.0454237288135594e-05,
|
| 10230 |
+
"loss": 1.4942,
|
| 10231 |
+
"step": 14580
|
| 10232 |
+
},
|
| 10233 |
+
{
|
| 10234 |
+
"epoch": 0.36,
|
| 10235 |
+
"grad_norm": 15.8125,
|
| 10236 |
+
"learning_rate": 1.0447457627118644e-05,
|
| 10237 |
+
"loss": 1.5753,
|
| 10238 |
+
"step": 14590
|
| 10239 |
+
},
|
| 10240 |
+
{
|
| 10241 |
+
"epoch": 0.36,
|
| 10242 |
+
"grad_norm": 27.0,
|
| 10243 |
+
"learning_rate": 1.0440677966101695e-05,
|
| 10244 |
+
"loss": 1.4115,
|
| 10245 |
+
"step": 14600
|
| 10246 |
+
},
|
| 10247 |
+
{
|
| 10248 |
+
"epoch": 0.37,
|
| 10249 |
+
"grad_norm": 32.25,
|
| 10250 |
+
"learning_rate": 1.0433898305084747e-05,
|
| 10251 |
+
"loss": 1.3518,
|
| 10252 |
+
"step": 14610
|
| 10253 |
+
},
|
| 10254 |
+
{
|
| 10255 |
+
"epoch": 0.37,
|
| 10256 |
+
"grad_norm": 29.625,
|
| 10257 |
+
"learning_rate": 1.0427118644067798e-05,
|
| 10258 |
+
"loss": 1.522,
|
| 10259 |
+
"step": 14620
|
| 10260 |
+
},
|
| 10261 |
+
{
|
| 10262 |
+
"epoch": 0.37,
|
| 10263 |
+
"grad_norm": 25.75,
|
| 10264 |
+
"learning_rate": 1.0420338983050848e-05,
|
| 10265 |
+
"loss": 1.341,
|
| 10266 |
+
"step": 14630
|
| 10267 |
+
},
|
| 10268 |
+
{
|
| 10269 |
+
"epoch": 0.37,
|
| 10270 |
+
"grad_norm": 35.25,
|
| 10271 |
+
"learning_rate": 1.04135593220339e-05,
|
| 10272 |
+
"loss": 1.458,
|
| 10273 |
+
"step": 14640
|
| 10274 |
+
},
|
| 10275 |
+
{
|
| 10276 |
+
"epoch": 0.37,
|
| 10277 |
+
"grad_norm": 9.5625,
|
| 10278 |
+
"learning_rate": 1.040677966101695e-05,
|
| 10279 |
+
"loss": 1.4959,
|
| 10280 |
+
"step": 14650
|
| 10281 |
+
},
|
| 10282 |
+
{
|
| 10283 |
+
"epoch": 0.37,
|
| 10284 |
+
"grad_norm": 30.0,
|
| 10285 |
+
"learning_rate": 1.04e-05,
|
| 10286 |
+
"loss": 1.3038,
|
| 10287 |
+
"step": 14660
|
| 10288 |
+
},
|
| 10289 |
+
{
|
| 10290 |
+
"epoch": 0.37,
|
| 10291 |
+
"grad_norm": 29.375,
|
| 10292 |
+
"learning_rate": 1.0393220338983052e-05,
|
| 10293 |
+
"loss": 1.1989,
|
| 10294 |
+
"step": 14670
|
| 10295 |
+
},
|
| 10296 |
+
{
|
| 10297 |
+
"epoch": 0.37,
|
| 10298 |
+
"grad_norm": 11.3125,
|
| 10299 |
+
"learning_rate": 1.0386440677966103e-05,
|
| 10300 |
+
"loss": 1.4619,
|
| 10301 |
+
"step": 14680
|
| 10302 |
+
},
|
| 10303 |
+
{
|
| 10304 |
+
"epoch": 0.37,
|
| 10305 |
+
"grad_norm": 29.75,
|
| 10306 |
+
"learning_rate": 1.0379661016949155e-05,
|
| 10307 |
+
"loss": 1.386,
|
| 10308 |
+
"step": 14690
|
| 10309 |
+
},
|
| 10310 |
+
{
|
| 10311 |
+
"epoch": 0.37,
|
| 10312 |
+
"grad_norm": 20.0,
|
| 10313 |
+
"learning_rate": 1.0372881355932204e-05,
|
| 10314 |
+
"loss": 1.339,
|
| 10315 |
+
"step": 14700
|
| 10316 |
+
},
|
| 10317 |
+
{
|
| 10318 |
+
"epoch": 0.37,
|
| 10319 |
+
"grad_norm": 25.0,
|
| 10320 |
+
"learning_rate": 1.0366101694915256e-05,
|
| 10321 |
+
"loss": 1.3886,
|
| 10322 |
+
"step": 14710
|
| 10323 |
+
},
|
| 10324 |
+
{
|
| 10325 |
+
"epoch": 0.37,
|
| 10326 |
+
"grad_norm": 24.875,
|
| 10327 |
+
"learning_rate": 1.0359322033898307e-05,
|
| 10328 |
+
"loss": 1.364,
|
| 10329 |
+
"step": 14720
|
| 10330 |
+
},
|
| 10331 |
+
{
|
| 10332 |
+
"epoch": 0.37,
|
| 10333 |
+
"grad_norm": 26.0,
|
| 10334 |
+
"learning_rate": 1.0352542372881358e-05,
|
| 10335 |
+
"loss": 1.416,
|
| 10336 |
+
"step": 14730
|
| 10337 |
+
},
|
| 10338 |
+
{
|
| 10339 |
+
"epoch": 0.37,
|
| 10340 |
+
"grad_norm": 118.0,
|
| 10341 |
+
"learning_rate": 1.0345762711864406e-05,
|
| 10342 |
+
"loss": 1.3624,
|
| 10343 |
+
"step": 14740
|
| 10344 |
+
},
|
| 10345 |
+
{
|
| 10346 |
+
"epoch": 0.37,
|
| 10347 |
+
"grad_norm": 14.375,
|
| 10348 |
+
"learning_rate": 1.0338983050847458e-05,
|
| 10349 |
+
"loss": 1.346,
|
| 10350 |
+
"step": 14750
|
| 10351 |
+
},
|
| 10352 |
+
{
|
| 10353 |
+
"epoch": 0.37,
|
| 10354 |
+
"grad_norm": 19.5,
|
| 10355 |
+
"learning_rate": 1.033220338983051e-05,
|
| 10356 |
+
"loss": 1.4017,
|
| 10357 |
+
"step": 14760
|
| 10358 |
+
},
|
| 10359 |
+
{
|
| 10360 |
+
"epoch": 0.37,
|
| 10361 |
+
"grad_norm": 14.625,
|
| 10362 |
+
"learning_rate": 1.0325423728813559e-05,
|
| 10363 |
+
"loss": 1.4782,
|
| 10364 |
+
"step": 14770
|
| 10365 |
+
},
|
| 10366 |
+
{
|
| 10367 |
+
"epoch": 0.37,
|
| 10368 |
+
"grad_norm": 18.125,
|
| 10369 |
+
"learning_rate": 1.031864406779661e-05,
|
| 10370 |
+
"loss": 1.3452,
|
| 10371 |
+
"step": 14780
|
| 10372 |
+
},
|
| 10373 |
+
{
|
| 10374 |
+
"epoch": 0.37,
|
| 10375 |
+
"grad_norm": 8.6875,
|
| 10376 |
+
"learning_rate": 1.0311864406779662e-05,
|
| 10377 |
+
"loss": 1.3968,
|
| 10378 |
+
"step": 14790
|
| 10379 |
+
},
|
| 10380 |
+
{
|
| 10381 |
+
"epoch": 0.37,
|
| 10382 |
+
"grad_norm": 29.25,
|
| 10383 |
+
"learning_rate": 1.0305084745762712e-05,
|
| 10384 |
+
"loss": 1.3137,
|
| 10385 |
+
"step": 14800
|
| 10386 |
+
},
|
| 10387 |
+
{
|
| 10388 |
+
"epoch": 0.37,
|
| 10389 |
+
"grad_norm": 20.75,
|
| 10390 |
+
"learning_rate": 1.0298305084745763e-05,
|
| 10391 |
+
"loss": 1.4651,
|
| 10392 |
+
"step": 14810
|
| 10393 |
+
},
|
| 10394 |
+
{
|
| 10395 |
+
"epoch": 0.37,
|
| 10396 |
+
"grad_norm": 12.4375,
|
| 10397 |
+
"learning_rate": 1.0291525423728814e-05,
|
| 10398 |
+
"loss": 1.468,
|
| 10399 |
+
"step": 14820
|
| 10400 |
+
},
|
| 10401 |
+
{
|
| 10402 |
+
"epoch": 0.37,
|
| 10403 |
+
"grad_norm": 19.375,
|
| 10404 |
+
"learning_rate": 1.0284745762711866e-05,
|
| 10405 |
+
"loss": 1.5157,
|
| 10406 |
+
"step": 14830
|
| 10407 |
+
},
|
| 10408 |
+
{
|
| 10409 |
+
"epoch": 0.37,
|
| 10410 |
+
"grad_norm": 39.25,
|
| 10411 |
+
"learning_rate": 1.0277966101694915e-05,
|
| 10412 |
+
"loss": 1.4106,
|
| 10413 |
+
"step": 14840
|
| 10414 |
+
},
|
| 10415 |
+
{
|
| 10416 |
+
"epoch": 0.37,
|
| 10417 |
+
"grad_norm": 24.25,
|
| 10418 |
+
"learning_rate": 1.0271186440677967e-05,
|
| 10419 |
+
"loss": 1.3044,
|
| 10420 |
+
"step": 14850
|
| 10421 |
+
},
|
| 10422 |
+
{
|
| 10423 |
+
"epoch": 0.37,
|
| 10424 |
+
"grad_norm": 7.25,
|
| 10425 |
+
"learning_rate": 1.0264406779661018e-05,
|
| 10426 |
+
"loss": 1.3231,
|
| 10427 |
+
"step": 14860
|
| 10428 |
+
},
|
| 10429 |
+
{
|
| 10430 |
+
"epoch": 0.37,
|
| 10431 |
+
"grad_norm": 26.0,
|
| 10432 |
+
"learning_rate": 1.025762711864407e-05,
|
| 10433 |
+
"loss": 1.2322,
|
| 10434 |
+
"step": 14870
|
| 10435 |
+
},
|
| 10436 |
+
{
|
| 10437 |
+
"epoch": 0.37,
|
| 10438 |
+
"grad_norm": 16.625,
|
| 10439 |
+
"learning_rate": 1.025084745762712e-05,
|
| 10440 |
+
"loss": 1.1767,
|
| 10441 |
+
"step": 14880
|
| 10442 |
+
},
|
| 10443 |
+
{
|
| 10444 |
+
"epoch": 0.37,
|
| 10445 |
+
"grad_norm": 11.625,
|
| 10446 |
+
"learning_rate": 1.024406779661017e-05,
|
| 10447 |
+
"loss": 1.3503,
|
| 10448 |
+
"step": 14890
|
| 10449 |
+
},
|
| 10450 |
+
{
|
| 10451 |
+
"epoch": 0.37,
|
| 10452 |
+
"grad_norm": 9.6875,
|
| 10453 |
+
"learning_rate": 1.0237288135593222e-05,
|
| 10454 |
+
"loss": 1.3777,
|
| 10455 |
+
"step": 14900
|
| 10456 |
+
},
|
| 10457 |
+
{
|
| 10458 |
+
"epoch": 0.37,
|
| 10459 |
+
"grad_norm": 46.75,
|
| 10460 |
+
"learning_rate": 1.0230508474576274e-05,
|
| 10461 |
+
"loss": 1.3334,
|
| 10462 |
+
"step": 14910
|
| 10463 |
+
},
|
| 10464 |
+
{
|
| 10465 |
+
"epoch": 0.37,
|
| 10466 |
+
"grad_norm": 23.25,
|
| 10467 |
+
"learning_rate": 1.0223728813559323e-05,
|
| 10468 |
+
"loss": 1.2998,
|
| 10469 |
+
"step": 14920
|
| 10470 |
+
},
|
| 10471 |
+
{
|
| 10472 |
+
"epoch": 0.37,
|
| 10473 |
+
"grad_norm": 45.75,
|
| 10474 |
+
"learning_rate": 1.0216949152542375e-05,
|
| 10475 |
+
"loss": 1.3287,
|
| 10476 |
+
"step": 14930
|
| 10477 |
+
},
|
| 10478 |
+
{
|
| 10479 |
+
"epoch": 0.37,
|
| 10480 |
+
"grad_norm": 20.0,
|
| 10481 |
+
"learning_rate": 1.0210169491525423e-05,
|
| 10482 |
+
"loss": 1.4629,
|
| 10483 |
+
"step": 14940
|
| 10484 |
+
},
|
| 10485 |
+
{
|
| 10486 |
+
"epoch": 0.37,
|
| 10487 |
+
"grad_norm": 9.375,
|
| 10488 |
+
"learning_rate": 1.0203389830508474e-05,
|
| 10489 |
+
"loss": 1.2114,
|
| 10490 |
+
"step": 14950
|
| 10491 |
+
},
|
| 10492 |
+
{
|
| 10493 |
+
"epoch": 0.37,
|
| 10494 |
+
"grad_norm": 35.75,
|
| 10495 |
+
"learning_rate": 1.0196610169491525e-05,
|
| 10496 |
+
"loss": 1.4157,
|
| 10497 |
+
"step": 14960
|
| 10498 |
+
},
|
| 10499 |
+
{
|
| 10500 |
+
"epoch": 0.37,
|
| 10501 |
+
"grad_norm": 19.375,
|
| 10502 |
+
"learning_rate": 1.0189830508474577e-05,
|
| 10503 |
+
"loss": 1.3757,
|
| 10504 |
+
"step": 14970
|
| 10505 |
+
},
|
| 10506 |
+
{
|
| 10507 |
+
"epoch": 0.37,
|
| 10508 |
+
"grad_norm": 16.75,
|
| 10509 |
+
"learning_rate": 1.0183050847457627e-05,
|
| 10510 |
+
"loss": 1.4091,
|
| 10511 |
+
"step": 14980
|
| 10512 |
+
},
|
| 10513 |
+
{
|
| 10514 |
+
"epoch": 0.37,
|
| 10515 |
+
"grad_norm": 12.0,
|
| 10516 |
+
"learning_rate": 1.0176271186440678e-05,
|
| 10517 |
+
"loss": 1.4074,
|
| 10518 |
+
"step": 14990
|
| 10519 |
+
},
|
| 10520 |
+
{
|
| 10521 |
+
"epoch": 0.38,
|
| 10522 |
+
"grad_norm": 20.25,
|
| 10523 |
+
"learning_rate": 1.016949152542373e-05,
|
| 10524 |
+
"loss": 1.2817,
|
| 10525 |
+
"step": 15000
|
| 10526 |
+
},
|
| 10527 |
+
{
|
| 10528 |
+
"epoch": 0.38,
|
| 10529 |
+
"eval_loss": 1.3379485607147217,
|
| 10530 |
+
"eval_runtime": 59.3062,
|
| 10531 |
+
"eval_samples_per_second": 16.862,
|
| 10532 |
+
"eval_steps_per_second": 16.862,
|
| 10533 |
+
"step": 15000
|
| 10534 |
}
|
| 10535 |
],
|
| 10536 |
"logging_steps": 10,
|
|
|
|
| 10538 |
"num_input_tokens_seen": 0,
|
| 10539 |
"num_train_epochs": 1,
|
| 10540 |
"save_steps": 5000,
|
| 10541 |
+
"total_flos": 2.3600547495936e+17,
|
| 10542 |
"train_batch_size": 1,
|
| 10543 |
"trial_name": null,
|
| 10544 |
"trial_params": null
|