Uploaded checkpoint-30000
Browse files- model.safetensors +1 -1
- optimizer.pt +1 -1
- rng_state.pth +1 -1
- scheduler.pt +1 -1
- trainer_state.json +1793 -3
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 sha256:7407eb7257531b614b61c8d1d84d4fdb719b07a14e72e981f4319bd3b34e5f33
|
| 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 sha256:818910f673e75761c43fb2f9eda548ec1856808990050a5beae905ebe0e58a55
|
| 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 sha256:7f2917558eb213e594a8e2999ee1e91aeb27af45c09cae9c3a3a202b99cef622
|
| 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 sha256:992f92f28f32913b10be7c822cdebf0abc4183b682b29981995e05a2a28bcd4d
|
| 3 |
size 1064
|
trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": 0.6027244925498962,
|
| 3 |
"best_model_checkpoint": "runs/deepseek_CMU-AIR2/math-deepseek_FULL_HardArith_Interm_20240424-065814/checkpoint-5000",
|
| 4 |
-
"epoch": 0.
|
| 5 |
"eval_steps": 500,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -5377,6 +5377,1796 @@
|
|
| 5377 |
"eval_samples_per_second": 26.2,
|
| 5378 |
"eval_steps_per_second": 26.2,
|
| 5379 |
"step": 7500
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 5380 |
}
|
| 5381 |
],
|
| 5382 |
"logging_steps": 10,
|
|
@@ -5384,7 +7174,7 @@
|
|
| 5384 |
"num_input_tokens_seen": 0,
|
| 5385 |
"num_train_epochs": 1,
|
| 5386 |
"save_steps": 2500,
|
| 5387 |
-
"total_flos": 1.
|
| 5388 |
"train_batch_size": 1,
|
| 5389 |
"trial_name": null,
|
| 5390 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": 0.6027244925498962,
|
| 3 |
"best_model_checkpoint": "runs/deepseek_CMU-AIR2/math-deepseek_FULL_HardArith_Interm_20240424-065814/checkpoint-5000",
|
| 4 |
+
"epoch": 0.25,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 10000,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 5377 |
"eval_samples_per_second": 26.2,
|
| 5378 |
"eval_steps_per_second": 26.2,
|
| 5379 |
"step": 7500
|
| 5380 |
+
},
|
| 5381 |
+
{
|
| 5382 |
+
"epoch": 0.19,
|
| 5383 |
+
"grad_norm": 4.6875,
|
| 5384 |
+
"learning_rate": 5.242105263157895e-06,
|
| 5385 |
+
"loss": 0.5938,
|
| 5386 |
+
"step": 7510
|
| 5387 |
+
},
|
| 5388 |
+
{
|
| 5389 |
+
"epoch": 0.19,
|
| 5390 |
+
"grad_norm": 5.0625,
|
| 5391 |
+
"learning_rate": 5.2210526315789475e-06,
|
| 5392 |
+
"loss": 0.6286,
|
| 5393 |
+
"step": 7520
|
| 5394 |
+
},
|
| 5395 |
+
{
|
| 5396 |
+
"epoch": 0.19,
|
| 5397 |
+
"grad_norm": 4.53125,
|
| 5398 |
+
"learning_rate": 5.2e-06,
|
| 5399 |
+
"loss": 0.7065,
|
| 5400 |
+
"step": 7530
|
| 5401 |
+
},
|
| 5402 |
+
{
|
| 5403 |
+
"epoch": 0.19,
|
| 5404 |
+
"grad_norm": 3.28125,
|
| 5405 |
+
"learning_rate": 5.178947368421054e-06,
|
| 5406 |
+
"loss": 0.5862,
|
| 5407 |
+
"step": 7540
|
| 5408 |
+
},
|
| 5409 |
+
{
|
| 5410 |
+
"epoch": 0.19,
|
| 5411 |
+
"grad_norm": 4.5,
|
| 5412 |
+
"learning_rate": 5.157894736842106e-06,
|
| 5413 |
+
"loss": 0.5015,
|
| 5414 |
+
"step": 7550
|
| 5415 |
+
},
|
| 5416 |
+
{
|
| 5417 |
+
"epoch": 0.19,
|
| 5418 |
+
"grad_norm": 2.640625,
|
| 5419 |
+
"learning_rate": 5.136842105263158e-06,
|
| 5420 |
+
"loss": 0.6041,
|
| 5421 |
+
"step": 7560
|
| 5422 |
+
},
|
| 5423 |
+
{
|
| 5424 |
+
"epoch": 0.19,
|
| 5425 |
+
"grad_norm": 5.59375,
|
| 5426 |
+
"learning_rate": 5.115789473684211e-06,
|
| 5427 |
+
"loss": 0.6845,
|
| 5428 |
+
"step": 7570
|
| 5429 |
+
},
|
| 5430 |
+
{
|
| 5431 |
+
"epoch": 0.19,
|
| 5432 |
+
"grad_norm": 6.59375,
|
| 5433 |
+
"learning_rate": 5.0947368421052635e-06,
|
| 5434 |
+
"loss": 0.698,
|
| 5435 |
+
"step": 7580
|
| 5436 |
+
},
|
| 5437 |
+
{
|
| 5438 |
+
"epoch": 0.19,
|
| 5439 |
+
"grad_norm": 9.8125,
|
| 5440 |
+
"learning_rate": 5.073684210526316e-06,
|
| 5441 |
+
"loss": 0.4452,
|
| 5442 |
+
"step": 7590
|
| 5443 |
+
},
|
| 5444 |
+
{
|
| 5445 |
+
"epoch": 0.19,
|
| 5446 |
+
"grad_norm": 3.1875,
|
| 5447 |
+
"learning_rate": 5.052631578947369e-06,
|
| 5448 |
+
"loss": 0.6485,
|
| 5449 |
+
"step": 7600
|
| 5450 |
+
},
|
| 5451 |
+
{
|
| 5452 |
+
"epoch": 0.19,
|
| 5453 |
+
"grad_norm": 5.8125,
|
| 5454 |
+
"learning_rate": 5.0315789473684214e-06,
|
| 5455 |
+
"loss": 0.5703,
|
| 5456 |
+
"step": 7610
|
| 5457 |
+
},
|
| 5458 |
+
{
|
| 5459 |
+
"epoch": 0.19,
|
| 5460 |
+
"grad_norm": 3.953125,
|
| 5461 |
+
"learning_rate": 5.010526315789475e-06,
|
| 5462 |
+
"loss": 0.5871,
|
| 5463 |
+
"step": 7620
|
| 5464 |
+
},
|
| 5465 |
+
{
|
| 5466 |
+
"epoch": 0.19,
|
| 5467 |
+
"grad_norm": 5.5625,
|
| 5468 |
+
"learning_rate": 4.989473684210527e-06,
|
| 5469 |
+
"loss": 0.6024,
|
| 5470 |
+
"step": 7630
|
| 5471 |
+
},
|
| 5472 |
+
{
|
| 5473 |
+
"epoch": 0.19,
|
| 5474 |
+
"grad_norm": 2.9375,
|
| 5475 |
+
"learning_rate": 4.968421052631579e-06,
|
| 5476 |
+
"loss": 0.5723,
|
| 5477 |
+
"step": 7640
|
| 5478 |
+
},
|
| 5479 |
+
{
|
| 5480 |
+
"epoch": 0.19,
|
| 5481 |
+
"grad_norm": 3.15625,
|
| 5482 |
+
"learning_rate": 4.947368421052632e-06,
|
| 5483 |
+
"loss": 0.584,
|
| 5484 |
+
"step": 7650
|
| 5485 |
+
},
|
| 5486 |
+
{
|
| 5487 |
+
"epoch": 0.19,
|
| 5488 |
+
"grad_norm": 8.4375,
|
| 5489 |
+
"learning_rate": 4.926315789473685e-06,
|
| 5490 |
+
"loss": 0.6387,
|
| 5491 |
+
"step": 7660
|
| 5492 |
+
},
|
| 5493 |
+
{
|
| 5494 |
+
"epoch": 0.19,
|
| 5495 |
+
"grad_norm": 4.34375,
|
| 5496 |
+
"learning_rate": 4.905263157894737e-06,
|
| 5497 |
+
"loss": 0.6574,
|
| 5498 |
+
"step": 7670
|
| 5499 |
+
},
|
| 5500 |
+
{
|
| 5501 |
+
"epoch": 0.19,
|
| 5502 |
+
"grad_norm": 3.578125,
|
| 5503 |
+
"learning_rate": 4.88421052631579e-06,
|
| 5504 |
+
"loss": 0.5907,
|
| 5505 |
+
"step": 7680
|
| 5506 |
+
},
|
| 5507 |
+
{
|
| 5508 |
+
"epoch": 0.19,
|
| 5509 |
+
"grad_norm": 4.5,
|
| 5510 |
+
"learning_rate": 4.863157894736843e-06,
|
| 5511 |
+
"loss": 0.6319,
|
| 5512 |
+
"step": 7690
|
| 5513 |
+
},
|
| 5514 |
+
{
|
| 5515 |
+
"epoch": 0.19,
|
| 5516 |
+
"grad_norm": 3.234375,
|
| 5517 |
+
"learning_rate": 4.842105263157895e-06,
|
| 5518 |
+
"loss": 0.5053,
|
| 5519 |
+
"step": 7700
|
| 5520 |
+
},
|
| 5521 |
+
{
|
| 5522 |
+
"epoch": 0.19,
|
| 5523 |
+
"grad_norm": 1.7421875,
|
| 5524 |
+
"learning_rate": 4.821052631578948e-06,
|
| 5525 |
+
"loss": 0.5135,
|
| 5526 |
+
"step": 7710
|
| 5527 |
+
},
|
| 5528 |
+
{
|
| 5529 |
+
"epoch": 0.19,
|
| 5530 |
+
"grad_norm": 17.375,
|
| 5531 |
+
"learning_rate": 4.800000000000001e-06,
|
| 5532 |
+
"loss": 0.6893,
|
| 5533 |
+
"step": 7720
|
| 5534 |
+
},
|
| 5535 |
+
{
|
| 5536 |
+
"epoch": 0.19,
|
| 5537 |
+
"grad_norm": 3.609375,
|
| 5538 |
+
"learning_rate": 4.778947368421053e-06,
|
| 5539 |
+
"loss": 0.6257,
|
| 5540 |
+
"step": 7730
|
| 5541 |
+
},
|
| 5542 |
+
{
|
| 5543 |
+
"epoch": 0.19,
|
| 5544 |
+
"grad_norm": 4.46875,
|
| 5545 |
+
"learning_rate": 4.757894736842106e-06,
|
| 5546 |
+
"loss": 0.5808,
|
| 5547 |
+
"step": 7740
|
| 5548 |
+
},
|
| 5549 |
+
{
|
| 5550 |
+
"epoch": 0.19,
|
| 5551 |
+
"grad_norm": 3.984375,
|
| 5552 |
+
"learning_rate": 4.736842105263158e-06,
|
| 5553 |
+
"loss": 0.518,
|
| 5554 |
+
"step": 7750
|
| 5555 |
+
},
|
| 5556 |
+
{
|
| 5557 |
+
"epoch": 0.19,
|
| 5558 |
+
"grad_norm": 5.90625,
|
| 5559 |
+
"learning_rate": 4.71578947368421e-06,
|
| 5560 |
+
"loss": 0.5991,
|
| 5561 |
+
"step": 7760
|
| 5562 |
+
},
|
| 5563 |
+
{
|
| 5564 |
+
"epoch": 0.19,
|
| 5565 |
+
"grad_norm": 4.71875,
|
| 5566 |
+
"learning_rate": 4.694736842105264e-06,
|
| 5567 |
+
"loss": 0.6514,
|
| 5568 |
+
"step": 7770
|
| 5569 |
+
},
|
| 5570 |
+
{
|
| 5571 |
+
"epoch": 0.19,
|
| 5572 |
+
"grad_norm": 3.109375,
|
| 5573 |
+
"learning_rate": 4.6736842105263166e-06,
|
| 5574 |
+
"loss": 0.489,
|
| 5575 |
+
"step": 7780
|
| 5576 |
+
},
|
| 5577 |
+
{
|
| 5578 |
+
"epoch": 0.19,
|
| 5579 |
+
"grad_norm": 3.609375,
|
| 5580 |
+
"learning_rate": 4.652631578947368e-06,
|
| 5581 |
+
"loss": 0.5181,
|
| 5582 |
+
"step": 7790
|
| 5583 |
+
},
|
| 5584 |
+
{
|
| 5585 |
+
"epoch": 0.2,
|
| 5586 |
+
"grad_norm": 2.78125,
|
| 5587 |
+
"learning_rate": 4.631578947368421e-06,
|
| 5588 |
+
"loss": 0.5767,
|
| 5589 |
+
"step": 7800
|
| 5590 |
+
},
|
| 5591 |
+
{
|
| 5592 |
+
"epoch": 0.2,
|
| 5593 |
+
"grad_norm": 3.171875,
|
| 5594 |
+
"learning_rate": 4.6105263157894745e-06,
|
| 5595 |
+
"loss": 0.4968,
|
| 5596 |
+
"step": 7810
|
| 5597 |
+
},
|
| 5598 |
+
{
|
| 5599 |
+
"epoch": 0.2,
|
| 5600 |
+
"grad_norm": 4.8125,
|
| 5601 |
+
"learning_rate": 4.589473684210526e-06,
|
| 5602 |
+
"loss": 0.6562,
|
| 5603 |
+
"step": 7820
|
| 5604 |
+
},
|
| 5605 |
+
{
|
| 5606 |
+
"epoch": 0.2,
|
| 5607 |
+
"grad_norm": 2.703125,
|
| 5608 |
+
"learning_rate": 4.568421052631579e-06,
|
| 5609 |
+
"loss": 0.6273,
|
| 5610 |
+
"step": 7830
|
| 5611 |
+
},
|
| 5612 |
+
{
|
| 5613 |
+
"epoch": 0.2,
|
| 5614 |
+
"grad_norm": 4.75,
|
| 5615 |
+
"learning_rate": 4.547368421052632e-06,
|
| 5616 |
+
"loss": 0.6351,
|
| 5617 |
+
"step": 7840
|
| 5618 |
+
},
|
| 5619 |
+
{
|
| 5620 |
+
"epoch": 0.2,
|
| 5621 |
+
"grad_norm": 2.9375,
|
| 5622 |
+
"learning_rate": 4.526315789473685e-06,
|
| 5623 |
+
"loss": 0.5495,
|
| 5624 |
+
"step": 7850
|
| 5625 |
+
},
|
| 5626 |
+
{
|
| 5627 |
+
"epoch": 0.2,
|
| 5628 |
+
"grad_norm": 10.6875,
|
| 5629 |
+
"learning_rate": 4.505263157894737e-06,
|
| 5630 |
+
"loss": 0.6603,
|
| 5631 |
+
"step": 7860
|
| 5632 |
+
},
|
| 5633 |
+
{
|
| 5634 |
+
"epoch": 0.2,
|
| 5635 |
+
"grad_norm": 4.25,
|
| 5636 |
+
"learning_rate": 4.48421052631579e-06,
|
| 5637 |
+
"loss": 0.5525,
|
| 5638 |
+
"step": 7870
|
| 5639 |
+
},
|
| 5640 |
+
{
|
| 5641 |
+
"epoch": 0.2,
|
| 5642 |
+
"grad_norm": 3.171875,
|
| 5643 |
+
"learning_rate": 4.463157894736842e-06,
|
| 5644 |
+
"loss": 0.6219,
|
| 5645 |
+
"step": 7880
|
| 5646 |
+
},
|
| 5647 |
+
{
|
| 5648 |
+
"epoch": 0.2,
|
| 5649 |
+
"grad_norm": 2.859375,
|
| 5650 |
+
"learning_rate": 4.442105263157896e-06,
|
| 5651 |
+
"loss": 0.5518,
|
| 5652 |
+
"step": 7890
|
| 5653 |
+
},
|
| 5654 |
+
{
|
| 5655 |
+
"epoch": 0.2,
|
| 5656 |
+
"grad_norm": 3.109375,
|
| 5657 |
+
"learning_rate": 4.4210526315789476e-06,
|
| 5658 |
+
"loss": 0.5462,
|
| 5659 |
+
"step": 7900
|
| 5660 |
+
},
|
| 5661 |
+
{
|
| 5662 |
+
"epoch": 0.2,
|
| 5663 |
+
"grad_norm": 3.359375,
|
| 5664 |
+
"learning_rate": 4.4e-06,
|
| 5665 |
+
"loss": 0.5776,
|
| 5666 |
+
"step": 7910
|
| 5667 |
+
},
|
| 5668 |
+
{
|
| 5669 |
+
"epoch": 0.2,
|
| 5670 |
+
"grad_norm": 4.4375,
|
| 5671 |
+
"learning_rate": 4.378947368421053e-06,
|
| 5672 |
+
"loss": 0.4804,
|
| 5673 |
+
"step": 7920
|
| 5674 |
+
},
|
| 5675 |
+
{
|
| 5676 |
+
"epoch": 0.2,
|
| 5677 |
+
"grad_norm": 3.578125,
|
| 5678 |
+
"learning_rate": 4.3578947368421055e-06,
|
| 5679 |
+
"loss": 0.5858,
|
| 5680 |
+
"step": 7930
|
| 5681 |
+
},
|
| 5682 |
+
{
|
| 5683 |
+
"epoch": 0.2,
|
| 5684 |
+
"grad_norm": 3.71875,
|
| 5685 |
+
"learning_rate": 4.336842105263158e-06,
|
| 5686 |
+
"loss": 0.6847,
|
| 5687 |
+
"step": 7940
|
| 5688 |
+
},
|
| 5689 |
+
{
|
| 5690 |
+
"epoch": 0.2,
|
| 5691 |
+
"grad_norm": 3.984375,
|
| 5692 |
+
"learning_rate": 4.315789473684211e-06,
|
| 5693 |
+
"loss": 0.6626,
|
| 5694 |
+
"step": 7950
|
| 5695 |
+
},
|
| 5696 |
+
{
|
| 5697 |
+
"epoch": 0.2,
|
| 5698 |
+
"grad_norm": 9.4375,
|
| 5699 |
+
"learning_rate": 4.2947368421052635e-06,
|
| 5700 |
+
"loss": 0.5526,
|
| 5701 |
+
"step": 7960
|
| 5702 |
+
},
|
| 5703 |
+
{
|
| 5704 |
+
"epoch": 0.2,
|
| 5705 |
+
"grad_norm": 4.625,
|
| 5706 |
+
"learning_rate": 4.273684210526316e-06,
|
| 5707 |
+
"loss": 0.5661,
|
| 5708 |
+
"step": 7970
|
| 5709 |
+
},
|
| 5710 |
+
{
|
| 5711 |
+
"epoch": 0.2,
|
| 5712 |
+
"grad_norm": 4.28125,
|
| 5713 |
+
"learning_rate": 4.252631578947369e-06,
|
| 5714 |
+
"loss": 0.6182,
|
| 5715 |
+
"step": 7980
|
| 5716 |
+
},
|
| 5717 |
+
{
|
| 5718 |
+
"epoch": 0.2,
|
| 5719 |
+
"grad_norm": 6.3125,
|
| 5720 |
+
"learning_rate": 4.2315789473684215e-06,
|
| 5721 |
+
"loss": 0.629,
|
| 5722 |
+
"step": 7990
|
| 5723 |
+
},
|
| 5724 |
+
{
|
| 5725 |
+
"epoch": 0.2,
|
| 5726 |
+
"grad_norm": 3.59375,
|
| 5727 |
+
"learning_rate": 4.210526315789474e-06,
|
| 5728 |
+
"loss": 0.7233,
|
| 5729 |
+
"step": 8000
|
| 5730 |
+
},
|
| 5731 |
+
{
|
| 5732 |
+
"epoch": 0.2,
|
| 5733 |
+
"eval_loss": 0.6029936671257019,
|
| 5734 |
+
"eval_runtime": 38.1479,
|
| 5735 |
+
"eval_samples_per_second": 26.214,
|
| 5736 |
+
"eval_steps_per_second": 26.214,
|
| 5737 |
+
"step": 8000
|
| 5738 |
+
},
|
| 5739 |
+
{
|
| 5740 |
+
"epoch": 0.2,
|
| 5741 |
+
"grad_norm": 2.828125,
|
| 5742 |
+
"learning_rate": 4.189473684210527e-06,
|
| 5743 |
+
"loss": 0.5451,
|
| 5744 |
+
"step": 8010
|
| 5745 |
+
},
|
| 5746 |
+
{
|
| 5747 |
+
"epoch": 0.2,
|
| 5748 |
+
"grad_norm": 5.0,
|
| 5749 |
+
"learning_rate": 4.1684210526315794e-06,
|
| 5750 |
+
"loss": 0.5498,
|
| 5751 |
+
"step": 8020
|
| 5752 |
+
},
|
| 5753 |
+
{
|
| 5754 |
+
"epoch": 0.2,
|
| 5755 |
+
"grad_norm": 4.875,
|
| 5756 |
+
"learning_rate": 4.147368421052632e-06,
|
| 5757 |
+
"loss": 0.5518,
|
| 5758 |
+
"step": 8030
|
| 5759 |
+
},
|
| 5760 |
+
{
|
| 5761 |
+
"epoch": 0.2,
|
| 5762 |
+
"grad_norm": 3.609375,
|
| 5763 |
+
"learning_rate": 4.126315789473685e-06,
|
| 5764 |
+
"loss": 0.6088,
|
| 5765 |
+
"step": 8040
|
| 5766 |
+
},
|
| 5767 |
+
{
|
| 5768 |
+
"epoch": 0.2,
|
| 5769 |
+
"grad_norm": 41.0,
|
| 5770 |
+
"learning_rate": 4.105263157894737e-06,
|
| 5771 |
+
"loss": 0.6399,
|
| 5772 |
+
"step": 8050
|
| 5773 |
+
},
|
| 5774 |
+
{
|
| 5775 |
+
"epoch": 0.2,
|
| 5776 |
+
"grad_norm": 3.578125,
|
| 5777 |
+
"learning_rate": 4.08421052631579e-06,
|
| 5778 |
+
"loss": 0.5412,
|
| 5779 |
+
"step": 8060
|
| 5780 |
+
},
|
| 5781 |
+
{
|
| 5782 |
+
"epoch": 0.2,
|
| 5783 |
+
"grad_norm": 4.21875,
|
| 5784 |
+
"learning_rate": 4.063157894736842e-06,
|
| 5785 |
+
"loss": 0.4933,
|
| 5786 |
+
"step": 8070
|
| 5787 |
+
},
|
| 5788 |
+
{
|
| 5789 |
+
"epoch": 0.2,
|
| 5790 |
+
"grad_norm": 3.234375,
|
| 5791 |
+
"learning_rate": 4.042105263157895e-06,
|
| 5792 |
+
"loss": 0.6182,
|
| 5793 |
+
"step": 8080
|
| 5794 |
+
},
|
| 5795 |
+
{
|
| 5796 |
+
"epoch": 0.2,
|
| 5797 |
+
"grad_norm": 2.953125,
|
| 5798 |
+
"learning_rate": 4.021052631578948e-06,
|
| 5799 |
+
"loss": 0.5342,
|
| 5800 |
+
"step": 8090
|
| 5801 |
+
},
|
| 5802 |
+
{
|
| 5803 |
+
"epoch": 0.2,
|
| 5804 |
+
"grad_norm": 1.6328125,
|
| 5805 |
+
"learning_rate": 4.000000000000001e-06,
|
| 5806 |
+
"loss": 0.5838,
|
| 5807 |
+
"step": 8100
|
| 5808 |
+
},
|
| 5809 |
+
{
|
| 5810 |
+
"epoch": 0.2,
|
| 5811 |
+
"grad_norm": 2.40625,
|
| 5812 |
+
"learning_rate": 3.9789473684210525e-06,
|
| 5813 |
+
"loss": 0.6717,
|
| 5814 |
+
"step": 8110
|
| 5815 |
+
},
|
| 5816 |
+
{
|
| 5817 |
+
"epoch": 0.2,
|
| 5818 |
+
"grad_norm": 7.0,
|
| 5819 |
+
"learning_rate": 3.957894736842106e-06,
|
| 5820 |
+
"loss": 0.6484,
|
| 5821 |
+
"step": 8120
|
| 5822 |
+
},
|
| 5823 |
+
{
|
| 5824 |
+
"epoch": 0.2,
|
| 5825 |
+
"grad_norm": 10.4375,
|
| 5826 |
+
"learning_rate": 3.936842105263159e-06,
|
| 5827 |
+
"loss": 0.6253,
|
| 5828 |
+
"step": 8130
|
| 5829 |
+
},
|
| 5830 |
+
{
|
| 5831 |
+
"epoch": 0.2,
|
| 5832 |
+
"grad_norm": 3.078125,
|
| 5833 |
+
"learning_rate": 3.9157894736842104e-06,
|
| 5834 |
+
"loss": 0.5562,
|
| 5835 |
+
"step": 8140
|
| 5836 |
+
},
|
| 5837 |
+
{
|
| 5838 |
+
"epoch": 0.2,
|
| 5839 |
+
"grad_norm": 5.4375,
|
| 5840 |
+
"learning_rate": 3.894736842105263e-06,
|
| 5841 |
+
"loss": 0.6162,
|
| 5842 |
+
"step": 8150
|
| 5843 |
+
},
|
| 5844 |
+
{
|
| 5845 |
+
"epoch": 0.2,
|
| 5846 |
+
"grad_norm": 2.328125,
|
| 5847 |
+
"learning_rate": 3.873684210526316e-06,
|
| 5848 |
+
"loss": 0.5602,
|
| 5849 |
+
"step": 8160
|
| 5850 |
+
},
|
| 5851 |
+
{
|
| 5852 |
+
"epoch": 0.2,
|
| 5853 |
+
"grad_norm": 6.78125,
|
| 5854 |
+
"learning_rate": 3.852631578947369e-06,
|
| 5855 |
+
"loss": 0.5414,
|
| 5856 |
+
"step": 8170
|
| 5857 |
+
},
|
| 5858 |
+
{
|
| 5859 |
+
"epoch": 0.2,
|
| 5860 |
+
"grad_norm": 4.40625,
|
| 5861 |
+
"learning_rate": 3.831578947368421e-06,
|
| 5862 |
+
"loss": 0.5783,
|
| 5863 |
+
"step": 8180
|
| 5864 |
+
},
|
| 5865 |
+
{
|
| 5866 |
+
"epoch": 0.2,
|
| 5867 |
+
"grad_norm": 3.46875,
|
| 5868 |
+
"learning_rate": 3.810526315789474e-06,
|
| 5869 |
+
"loss": 0.583,
|
| 5870 |
+
"step": 8190
|
| 5871 |
+
},
|
| 5872 |
+
{
|
| 5873 |
+
"epoch": 0.2,
|
| 5874 |
+
"grad_norm": 6.5625,
|
| 5875 |
+
"learning_rate": 3.789473684210527e-06,
|
| 5876 |
+
"loss": 0.632,
|
| 5877 |
+
"step": 8200
|
| 5878 |
+
},
|
| 5879 |
+
{
|
| 5880 |
+
"epoch": 0.21,
|
| 5881 |
+
"grad_norm": 3.625,
|
| 5882 |
+
"learning_rate": 3.768421052631579e-06,
|
| 5883 |
+
"loss": 0.5573,
|
| 5884 |
+
"step": 8210
|
| 5885 |
+
},
|
| 5886 |
+
{
|
| 5887 |
+
"epoch": 0.21,
|
| 5888 |
+
"grad_norm": 3.578125,
|
| 5889 |
+
"learning_rate": 3.7473684210526317e-06,
|
| 5890 |
+
"loss": 0.5176,
|
| 5891 |
+
"step": 8220
|
| 5892 |
+
},
|
| 5893 |
+
{
|
| 5894 |
+
"epoch": 0.21,
|
| 5895 |
+
"grad_norm": 2.796875,
|
| 5896 |
+
"learning_rate": 3.7263157894736848e-06,
|
| 5897 |
+
"loss": 0.4684,
|
| 5898 |
+
"step": 8230
|
| 5899 |
+
},
|
| 5900 |
+
{
|
| 5901 |
+
"epoch": 0.21,
|
| 5902 |
+
"grad_norm": 6.71875,
|
| 5903 |
+
"learning_rate": 3.7052631578947374e-06,
|
| 5904 |
+
"loss": 0.6114,
|
| 5905 |
+
"step": 8240
|
| 5906 |
+
},
|
| 5907 |
+
{
|
| 5908 |
+
"epoch": 0.21,
|
| 5909 |
+
"grad_norm": 5.5625,
|
| 5910 |
+
"learning_rate": 3.6842105263157896e-06,
|
| 5911 |
+
"loss": 0.5437,
|
| 5912 |
+
"step": 8250
|
| 5913 |
+
},
|
| 5914 |
+
{
|
| 5915 |
+
"epoch": 0.21,
|
| 5916 |
+
"grad_norm": 4.53125,
|
| 5917 |
+
"learning_rate": 3.6631578947368423e-06,
|
| 5918 |
+
"loss": 0.5119,
|
| 5919 |
+
"step": 8260
|
| 5920 |
+
},
|
| 5921 |
+
{
|
| 5922 |
+
"epoch": 0.21,
|
| 5923 |
+
"grad_norm": 4.21875,
|
| 5924 |
+
"learning_rate": 3.642105263157895e-06,
|
| 5925 |
+
"loss": 0.5742,
|
| 5926 |
+
"step": 8270
|
| 5927 |
+
},
|
| 5928 |
+
{
|
| 5929 |
+
"epoch": 0.21,
|
| 5930 |
+
"grad_norm": 5.46875,
|
| 5931 |
+
"learning_rate": 3.621052631578948e-06,
|
| 5932 |
+
"loss": 0.6704,
|
| 5933 |
+
"step": 8280
|
| 5934 |
+
},
|
| 5935 |
+
{
|
| 5936 |
+
"epoch": 0.21,
|
| 5937 |
+
"grad_norm": 3.25,
|
| 5938 |
+
"learning_rate": 3.6000000000000003e-06,
|
| 5939 |
+
"loss": 0.5655,
|
| 5940 |
+
"step": 8290
|
| 5941 |
+
},
|
| 5942 |
+
{
|
| 5943 |
+
"epoch": 0.21,
|
| 5944 |
+
"grad_norm": 3.15625,
|
| 5945 |
+
"learning_rate": 3.578947368421053e-06,
|
| 5946 |
+
"loss": 0.5382,
|
| 5947 |
+
"step": 8300
|
| 5948 |
+
},
|
| 5949 |
+
{
|
| 5950 |
+
"epoch": 0.21,
|
| 5951 |
+
"grad_norm": 6.875,
|
| 5952 |
+
"learning_rate": 3.5578947368421056e-06,
|
| 5953 |
+
"loss": 0.5485,
|
| 5954 |
+
"step": 8310
|
| 5955 |
+
},
|
| 5956 |
+
{
|
| 5957 |
+
"epoch": 0.21,
|
| 5958 |
+
"grad_norm": 7.3125,
|
| 5959 |
+
"learning_rate": 3.536842105263158e-06,
|
| 5960 |
+
"loss": 0.5124,
|
| 5961 |
+
"step": 8320
|
| 5962 |
+
},
|
| 5963 |
+
{
|
| 5964 |
+
"epoch": 0.21,
|
| 5965 |
+
"grad_norm": 4.625,
|
| 5966 |
+
"learning_rate": 3.515789473684211e-06,
|
| 5967 |
+
"loss": 0.4819,
|
| 5968 |
+
"step": 8330
|
| 5969 |
+
},
|
| 5970 |
+
{
|
| 5971 |
+
"epoch": 0.21,
|
| 5972 |
+
"grad_norm": 6.53125,
|
| 5973 |
+
"learning_rate": 3.4947368421052635e-06,
|
| 5974 |
+
"loss": 0.5999,
|
| 5975 |
+
"step": 8340
|
| 5976 |
+
},
|
| 5977 |
+
{
|
| 5978 |
+
"epoch": 0.21,
|
| 5979 |
+
"grad_norm": 4.3125,
|
| 5980 |
+
"learning_rate": 3.473684210526316e-06,
|
| 5981 |
+
"loss": 0.5241,
|
| 5982 |
+
"step": 8350
|
| 5983 |
+
},
|
| 5984 |
+
{
|
| 5985 |
+
"epoch": 0.21,
|
| 5986 |
+
"grad_norm": 7.78125,
|
| 5987 |
+
"learning_rate": 3.4526315789473684e-06,
|
| 5988 |
+
"loss": 0.6193,
|
| 5989 |
+
"step": 8360
|
| 5990 |
+
},
|
| 5991 |
+
{
|
| 5992 |
+
"epoch": 0.21,
|
| 5993 |
+
"grad_norm": 5.09375,
|
| 5994 |
+
"learning_rate": 3.4315789473684215e-06,
|
| 5995 |
+
"loss": 0.6701,
|
| 5996 |
+
"step": 8370
|
| 5997 |
+
},
|
| 5998 |
+
{
|
| 5999 |
+
"epoch": 0.21,
|
| 6000 |
+
"grad_norm": 2.453125,
|
| 6001 |
+
"learning_rate": 3.410526315789474e-06,
|
| 6002 |
+
"loss": 0.4458,
|
| 6003 |
+
"step": 8380
|
| 6004 |
+
},
|
| 6005 |
+
{
|
| 6006 |
+
"epoch": 0.21,
|
| 6007 |
+
"grad_norm": 2.53125,
|
| 6008 |
+
"learning_rate": 3.3894736842105264e-06,
|
| 6009 |
+
"loss": 0.5441,
|
| 6010 |
+
"step": 8390
|
| 6011 |
+
},
|
| 6012 |
+
{
|
| 6013 |
+
"epoch": 0.21,
|
| 6014 |
+
"grad_norm": 3.71875,
|
| 6015 |
+
"learning_rate": 3.368421052631579e-06,
|
| 6016 |
+
"loss": 0.5787,
|
| 6017 |
+
"step": 8400
|
| 6018 |
+
},
|
| 6019 |
+
{
|
| 6020 |
+
"epoch": 0.21,
|
| 6021 |
+
"grad_norm": 5.1875,
|
| 6022 |
+
"learning_rate": 3.347368421052632e-06,
|
| 6023 |
+
"loss": 0.5972,
|
| 6024 |
+
"step": 8410
|
| 6025 |
+
},
|
| 6026 |
+
{
|
| 6027 |
+
"epoch": 0.21,
|
| 6028 |
+
"grad_norm": 3.828125,
|
| 6029 |
+
"learning_rate": 3.3263157894736848e-06,
|
| 6030 |
+
"loss": 0.5784,
|
| 6031 |
+
"step": 8420
|
| 6032 |
+
},
|
| 6033 |
+
{
|
| 6034 |
+
"epoch": 0.21,
|
| 6035 |
+
"grad_norm": 2.40625,
|
| 6036 |
+
"learning_rate": 3.305263157894737e-06,
|
| 6037 |
+
"loss": 0.6005,
|
| 6038 |
+
"step": 8430
|
| 6039 |
+
},
|
| 6040 |
+
{
|
| 6041 |
+
"epoch": 0.21,
|
| 6042 |
+
"grad_norm": 4.875,
|
| 6043 |
+
"learning_rate": 3.2842105263157897e-06,
|
| 6044 |
+
"loss": 0.5343,
|
| 6045 |
+
"step": 8440
|
| 6046 |
+
},
|
| 6047 |
+
{
|
| 6048 |
+
"epoch": 0.21,
|
| 6049 |
+
"grad_norm": 3.0,
|
| 6050 |
+
"learning_rate": 3.2631578947368423e-06,
|
| 6051 |
+
"loss": 0.5407,
|
| 6052 |
+
"step": 8450
|
| 6053 |
+
},
|
| 6054 |
+
{
|
| 6055 |
+
"epoch": 0.21,
|
| 6056 |
+
"grad_norm": 3.234375,
|
| 6057 |
+
"learning_rate": 3.2421052631578945e-06,
|
| 6058 |
+
"loss": 0.5047,
|
| 6059 |
+
"step": 8460
|
| 6060 |
+
},
|
| 6061 |
+
{
|
| 6062 |
+
"epoch": 0.21,
|
| 6063 |
+
"grad_norm": 5.6875,
|
| 6064 |
+
"learning_rate": 3.2210526315789476e-06,
|
| 6065 |
+
"loss": 0.4705,
|
| 6066 |
+
"step": 8470
|
| 6067 |
+
},
|
| 6068 |
+
{
|
| 6069 |
+
"epoch": 0.21,
|
| 6070 |
+
"grad_norm": 2.515625,
|
| 6071 |
+
"learning_rate": 3.2000000000000003e-06,
|
| 6072 |
+
"loss": 0.5838,
|
| 6073 |
+
"step": 8480
|
| 6074 |
+
},
|
| 6075 |
+
{
|
| 6076 |
+
"epoch": 0.21,
|
| 6077 |
+
"grad_norm": 3.453125,
|
| 6078 |
+
"learning_rate": 3.178947368421053e-06,
|
| 6079 |
+
"loss": 0.547,
|
| 6080 |
+
"step": 8490
|
| 6081 |
+
},
|
| 6082 |
+
{
|
| 6083 |
+
"epoch": 0.21,
|
| 6084 |
+
"grad_norm": 5.0625,
|
| 6085 |
+
"learning_rate": 3.157894736842105e-06,
|
| 6086 |
+
"loss": 0.6311,
|
| 6087 |
+
"step": 8500
|
| 6088 |
+
},
|
| 6089 |
+
{
|
| 6090 |
+
"epoch": 0.21,
|
| 6091 |
+
"eval_loss": 0.6170388460159302,
|
| 6092 |
+
"eval_runtime": 38.1693,
|
| 6093 |
+
"eval_samples_per_second": 26.199,
|
| 6094 |
+
"eval_steps_per_second": 26.199,
|
| 6095 |
+
"step": 8500
|
| 6096 |
+
},
|
| 6097 |
+
{
|
| 6098 |
+
"epoch": 0.21,
|
| 6099 |
+
"grad_norm": 5.96875,
|
| 6100 |
+
"learning_rate": 3.1368421052631582e-06,
|
| 6101 |
+
"loss": 0.6415,
|
| 6102 |
+
"step": 8510
|
| 6103 |
+
},
|
| 6104 |
+
{
|
| 6105 |
+
"epoch": 0.21,
|
| 6106 |
+
"grad_norm": 3.484375,
|
| 6107 |
+
"learning_rate": 3.115789473684211e-06,
|
| 6108 |
+
"loss": 0.5877,
|
| 6109 |
+
"step": 8520
|
| 6110 |
+
},
|
| 6111 |
+
{
|
| 6112 |
+
"epoch": 0.21,
|
| 6113 |
+
"grad_norm": 3.8125,
|
| 6114 |
+
"learning_rate": 3.094736842105263e-06,
|
| 6115 |
+
"loss": 0.5951,
|
| 6116 |
+
"step": 8530
|
| 6117 |
+
},
|
| 6118 |
+
{
|
| 6119 |
+
"epoch": 0.21,
|
| 6120 |
+
"grad_norm": 4.0625,
|
| 6121 |
+
"learning_rate": 3.0736842105263158e-06,
|
| 6122 |
+
"loss": 0.6047,
|
| 6123 |
+
"step": 8540
|
| 6124 |
+
},
|
| 6125 |
+
{
|
| 6126 |
+
"epoch": 0.21,
|
| 6127 |
+
"grad_norm": 2.609375,
|
| 6128 |
+
"learning_rate": 3.052631578947369e-06,
|
| 6129 |
+
"loss": 0.5694,
|
| 6130 |
+
"step": 8550
|
| 6131 |
+
},
|
| 6132 |
+
{
|
| 6133 |
+
"epoch": 0.21,
|
| 6134 |
+
"grad_norm": 4.0625,
|
| 6135 |
+
"learning_rate": 3.0315789473684215e-06,
|
| 6136 |
+
"loss": 0.6126,
|
| 6137 |
+
"step": 8560
|
| 6138 |
+
},
|
| 6139 |
+
{
|
| 6140 |
+
"epoch": 0.21,
|
| 6141 |
+
"grad_norm": 2.75,
|
| 6142 |
+
"learning_rate": 3.0105263157894737e-06,
|
| 6143 |
+
"loss": 0.521,
|
| 6144 |
+
"step": 8570
|
| 6145 |
+
},
|
| 6146 |
+
{
|
| 6147 |
+
"epoch": 0.21,
|
| 6148 |
+
"grad_norm": 4.46875,
|
| 6149 |
+
"learning_rate": 2.9894736842105264e-06,
|
| 6150 |
+
"loss": 0.6389,
|
| 6151 |
+
"step": 8580
|
| 6152 |
+
},
|
| 6153 |
+
{
|
| 6154 |
+
"epoch": 0.21,
|
| 6155 |
+
"grad_norm": 4.75,
|
| 6156 |
+
"learning_rate": 2.9684210526315795e-06,
|
| 6157 |
+
"loss": 0.5909,
|
| 6158 |
+
"step": 8590
|
| 6159 |
+
},
|
| 6160 |
+
{
|
| 6161 |
+
"epoch": 0.21,
|
| 6162 |
+
"grad_norm": 4.5625,
|
| 6163 |
+
"learning_rate": 2.9473684210526317e-06,
|
| 6164 |
+
"loss": 0.6578,
|
| 6165 |
+
"step": 8600
|
| 6166 |
+
},
|
| 6167 |
+
{
|
| 6168 |
+
"epoch": 0.22,
|
| 6169 |
+
"grad_norm": 6.0,
|
| 6170 |
+
"learning_rate": 2.9263157894736844e-06,
|
| 6171 |
+
"loss": 0.5277,
|
| 6172 |
+
"step": 8610
|
| 6173 |
+
},
|
| 6174 |
+
{
|
| 6175 |
+
"epoch": 0.22,
|
| 6176 |
+
"grad_norm": 2.71875,
|
| 6177 |
+
"learning_rate": 2.905263157894737e-06,
|
| 6178 |
+
"loss": 0.5313,
|
| 6179 |
+
"step": 8620
|
| 6180 |
+
},
|
| 6181 |
+
{
|
| 6182 |
+
"epoch": 0.22,
|
| 6183 |
+
"grad_norm": 3.34375,
|
| 6184 |
+
"learning_rate": 2.88421052631579e-06,
|
| 6185 |
+
"loss": 0.5201,
|
| 6186 |
+
"step": 8630
|
| 6187 |
+
},
|
| 6188 |
+
{
|
| 6189 |
+
"epoch": 0.22,
|
| 6190 |
+
"grad_norm": 4.0,
|
| 6191 |
+
"learning_rate": 2.8631578947368423e-06,
|
| 6192 |
+
"loss": 0.7751,
|
| 6193 |
+
"step": 8640
|
| 6194 |
+
},
|
| 6195 |
+
{
|
| 6196 |
+
"epoch": 0.22,
|
| 6197 |
+
"grad_norm": 2.546875,
|
| 6198 |
+
"learning_rate": 2.842105263157895e-06,
|
| 6199 |
+
"loss": 0.6536,
|
| 6200 |
+
"step": 8650
|
| 6201 |
+
},
|
| 6202 |
+
{
|
| 6203 |
+
"epoch": 0.22,
|
| 6204 |
+
"grad_norm": 6.34375,
|
| 6205 |
+
"learning_rate": 2.8210526315789476e-06,
|
| 6206 |
+
"loss": 0.6661,
|
| 6207 |
+
"step": 8660
|
| 6208 |
+
},
|
| 6209 |
+
{
|
| 6210 |
+
"epoch": 0.22,
|
| 6211 |
+
"grad_norm": 4.65625,
|
| 6212 |
+
"learning_rate": 2.8000000000000003e-06,
|
| 6213 |
+
"loss": 0.6284,
|
| 6214 |
+
"step": 8670
|
| 6215 |
+
},
|
| 6216 |
+
{
|
| 6217 |
+
"epoch": 0.22,
|
| 6218 |
+
"grad_norm": 5.5625,
|
| 6219 |
+
"learning_rate": 2.7789473684210525e-06,
|
| 6220 |
+
"loss": 0.673,
|
| 6221 |
+
"step": 8680
|
| 6222 |
+
},
|
| 6223 |
+
{
|
| 6224 |
+
"epoch": 0.22,
|
| 6225 |
+
"grad_norm": 3.03125,
|
| 6226 |
+
"learning_rate": 2.7578947368421056e-06,
|
| 6227 |
+
"loss": 0.646,
|
| 6228 |
+
"step": 8690
|
| 6229 |
+
},
|
| 6230 |
+
{
|
| 6231 |
+
"epoch": 0.22,
|
| 6232 |
+
"grad_norm": 2.765625,
|
| 6233 |
+
"learning_rate": 2.7368421052631583e-06,
|
| 6234 |
+
"loss": 0.5917,
|
| 6235 |
+
"step": 8700
|
| 6236 |
+
},
|
| 6237 |
+
{
|
| 6238 |
+
"epoch": 0.22,
|
| 6239 |
+
"grad_norm": 6.03125,
|
| 6240 |
+
"learning_rate": 2.7157894736842105e-06,
|
| 6241 |
+
"loss": 0.5706,
|
| 6242 |
+
"step": 8710
|
| 6243 |
+
},
|
| 6244 |
+
{
|
| 6245 |
+
"epoch": 0.22,
|
| 6246 |
+
"grad_norm": 3.171875,
|
| 6247 |
+
"learning_rate": 2.694736842105263e-06,
|
| 6248 |
+
"loss": 0.477,
|
| 6249 |
+
"step": 8720
|
| 6250 |
+
},
|
| 6251 |
+
{
|
| 6252 |
+
"epoch": 0.22,
|
| 6253 |
+
"grad_norm": 5.875,
|
| 6254 |
+
"learning_rate": 2.6736842105263162e-06,
|
| 6255 |
+
"loss": 0.646,
|
| 6256 |
+
"step": 8730
|
| 6257 |
+
},
|
| 6258 |
+
{
|
| 6259 |
+
"epoch": 0.22,
|
| 6260 |
+
"grad_norm": 4.75,
|
| 6261 |
+
"learning_rate": 2.652631578947369e-06,
|
| 6262 |
+
"loss": 0.638,
|
| 6263 |
+
"step": 8740
|
| 6264 |
+
},
|
| 6265 |
+
{
|
| 6266 |
+
"epoch": 0.22,
|
| 6267 |
+
"grad_norm": 3.4375,
|
| 6268 |
+
"learning_rate": 2.631578947368421e-06,
|
| 6269 |
+
"loss": 0.6282,
|
| 6270 |
+
"step": 8750
|
| 6271 |
+
},
|
| 6272 |
+
{
|
| 6273 |
+
"epoch": 0.22,
|
| 6274 |
+
"grad_norm": 3.6875,
|
| 6275 |
+
"learning_rate": 2.6105263157894738e-06,
|
| 6276 |
+
"loss": 0.6418,
|
| 6277 |
+
"step": 8760
|
| 6278 |
+
},
|
| 6279 |
+
{
|
| 6280 |
+
"epoch": 0.22,
|
| 6281 |
+
"grad_norm": 2.828125,
|
| 6282 |
+
"learning_rate": 2.589473684210527e-06,
|
| 6283 |
+
"loss": 0.6228,
|
| 6284 |
+
"step": 8770
|
| 6285 |
+
},
|
| 6286 |
+
{
|
| 6287 |
+
"epoch": 0.22,
|
| 6288 |
+
"grad_norm": 3.765625,
|
| 6289 |
+
"learning_rate": 2.568421052631579e-06,
|
| 6290 |
+
"loss": 0.5794,
|
| 6291 |
+
"step": 8780
|
| 6292 |
+
},
|
| 6293 |
+
{
|
| 6294 |
+
"epoch": 0.22,
|
| 6295 |
+
"grad_norm": 4.9375,
|
| 6296 |
+
"learning_rate": 2.5473684210526317e-06,
|
| 6297 |
+
"loss": 0.5207,
|
| 6298 |
+
"step": 8790
|
| 6299 |
+
},
|
| 6300 |
+
{
|
| 6301 |
+
"epoch": 0.22,
|
| 6302 |
+
"grad_norm": 4.75,
|
| 6303 |
+
"learning_rate": 2.5263157894736844e-06,
|
| 6304 |
+
"loss": 0.623,
|
| 6305 |
+
"step": 8800
|
| 6306 |
+
},
|
| 6307 |
+
{
|
| 6308 |
+
"epoch": 0.22,
|
| 6309 |
+
"grad_norm": 4.40625,
|
| 6310 |
+
"learning_rate": 2.5052631578947375e-06,
|
| 6311 |
+
"loss": 0.6086,
|
| 6312 |
+
"step": 8810
|
| 6313 |
+
},
|
| 6314 |
+
{
|
| 6315 |
+
"epoch": 0.22,
|
| 6316 |
+
"grad_norm": 3.109375,
|
| 6317 |
+
"learning_rate": 2.4842105263157897e-06,
|
| 6318 |
+
"loss": 0.579,
|
| 6319 |
+
"step": 8820
|
| 6320 |
+
},
|
| 6321 |
+
{
|
| 6322 |
+
"epoch": 0.22,
|
| 6323 |
+
"grad_norm": 3.65625,
|
| 6324 |
+
"learning_rate": 2.4631578947368424e-06,
|
| 6325 |
+
"loss": 0.4531,
|
| 6326 |
+
"step": 8830
|
| 6327 |
+
},
|
| 6328 |
+
{
|
| 6329 |
+
"epoch": 0.22,
|
| 6330 |
+
"grad_norm": 4.75,
|
| 6331 |
+
"learning_rate": 2.442105263157895e-06,
|
| 6332 |
+
"loss": 0.5167,
|
| 6333 |
+
"step": 8840
|
| 6334 |
+
},
|
| 6335 |
+
{
|
| 6336 |
+
"epoch": 0.22,
|
| 6337 |
+
"grad_norm": 7.78125,
|
| 6338 |
+
"learning_rate": 2.4210526315789477e-06,
|
| 6339 |
+
"loss": 0.6362,
|
| 6340 |
+
"step": 8850
|
| 6341 |
+
},
|
| 6342 |
+
{
|
| 6343 |
+
"epoch": 0.22,
|
| 6344 |
+
"grad_norm": 5.53125,
|
| 6345 |
+
"learning_rate": 2.4000000000000003e-06,
|
| 6346 |
+
"loss": 0.6628,
|
| 6347 |
+
"step": 8860
|
| 6348 |
+
},
|
| 6349 |
+
{
|
| 6350 |
+
"epoch": 0.22,
|
| 6351 |
+
"grad_norm": 2.453125,
|
| 6352 |
+
"learning_rate": 2.378947368421053e-06,
|
| 6353 |
+
"loss": 0.722,
|
| 6354 |
+
"step": 8870
|
| 6355 |
+
},
|
| 6356 |
+
{
|
| 6357 |
+
"epoch": 0.22,
|
| 6358 |
+
"grad_norm": 2.921875,
|
| 6359 |
+
"learning_rate": 2.357894736842105e-06,
|
| 6360 |
+
"loss": 0.5313,
|
| 6361 |
+
"step": 8880
|
| 6362 |
+
},
|
| 6363 |
+
{
|
| 6364 |
+
"epoch": 0.22,
|
| 6365 |
+
"grad_norm": 5.25,
|
| 6366 |
+
"learning_rate": 2.3368421052631583e-06,
|
| 6367 |
+
"loss": 0.6277,
|
| 6368 |
+
"step": 8890
|
| 6369 |
+
},
|
| 6370 |
+
{
|
| 6371 |
+
"epoch": 0.22,
|
| 6372 |
+
"grad_norm": 4.0,
|
| 6373 |
+
"learning_rate": 2.3157894736842105e-06,
|
| 6374 |
+
"loss": 0.5938,
|
| 6375 |
+
"step": 8900
|
| 6376 |
+
},
|
| 6377 |
+
{
|
| 6378 |
+
"epoch": 0.22,
|
| 6379 |
+
"grad_norm": 3.546875,
|
| 6380 |
+
"learning_rate": 2.294736842105263e-06,
|
| 6381 |
+
"loss": 0.6851,
|
| 6382 |
+
"step": 8910
|
| 6383 |
+
},
|
| 6384 |
+
{
|
| 6385 |
+
"epoch": 0.22,
|
| 6386 |
+
"grad_norm": 6.0,
|
| 6387 |
+
"learning_rate": 2.273684210526316e-06,
|
| 6388 |
+
"loss": 0.4746,
|
| 6389 |
+
"step": 8920
|
| 6390 |
+
},
|
| 6391 |
+
{
|
| 6392 |
+
"epoch": 0.22,
|
| 6393 |
+
"grad_norm": 4.0,
|
| 6394 |
+
"learning_rate": 2.2526315789473685e-06,
|
| 6395 |
+
"loss": 0.5109,
|
| 6396 |
+
"step": 8930
|
| 6397 |
+
},
|
| 6398 |
+
{
|
| 6399 |
+
"epoch": 0.22,
|
| 6400 |
+
"grad_norm": 2.421875,
|
| 6401 |
+
"learning_rate": 2.231578947368421e-06,
|
| 6402 |
+
"loss": 0.4732,
|
| 6403 |
+
"step": 8940
|
| 6404 |
+
},
|
| 6405 |
+
{
|
| 6406 |
+
"epoch": 0.22,
|
| 6407 |
+
"grad_norm": 3.234375,
|
| 6408 |
+
"learning_rate": 2.2105263157894738e-06,
|
| 6409 |
+
"loss": 0.7736,
|
| 6410 |
+
"step": 8950
|
| 6411 |
+
},
|
| 6412 |
+
{
|
| 6413 |
+
"epoch": 0.22,
|
| 6414 |
+
"grad_norm": 4.34375,
|
| 6415 |
+
"learning_rate": 2.1894736842105264e-06,
|
| 6416 |
+
"loss": 0.6251,
|
| 6417 |
+
"step": 8960
|
| 6418 |
+
},
|
| 6419 |
+
{
|
| 6420 |
+
"epoch": 0.22,
|
| 6421 |
+
"grad_norm": 2.75,
|
| 6422 |
+
"learning_rate": 2.168421052631579e-06,
|
| 6423 |
+
"loss": 0.6428,
|
| 6424 |
+
"step": 8970
|
| 6425 |
+
},
|
| 6426 |
+
{
|
| 6427 |
+
"epoch": 0.22,
|
| 6428 |
+
"grad_norm": 11.5625,
|
| 6429 |
+
"learning_rate": 2.1473684210526317e-06,
|
| 6430 |
+
"loss": 0.5886,
|
| 6431 |
+
"step": 8980
|
| 6432 |
+
},
|
| 6433 |
+
{
|
| 6434 |
+
"epoch": 0.22,
|
| 6435 |
+
"grad_norm": 3.8125,
|
| 6436 |
+
"learning_rate": 2.1263157894736844e-06,
|
| 6437 |
+
"loss": 0.597,
|
| 6438 |
+
"step": 8990
|
| 6439 |
+
},
|
| 6440 |
+
{
|
| 6441 |
+
"epoch": 0.23,
|
| 6442 |
+
"grad_norm": 1.8125,
|
| 6443 |
+
"learning_rate": 2.105263157894737e-06,
|
| 6444 |
+
"loss": 0.527,
|
| 6445 |
+
"step": 9000
|
| 6446 |
+
},
|
| 6447 |
+
{
|
| 6448 |
+
"epoch": 0.23,
|
| 6449 |
+
"eval_loss": 0.6106051802635193,
|
| 6450 |
+
"eval_runtime": 38.1285,
|
| 6451 |
+
"eval_samples_per_second": 26.227,
|
| 6452 |
+
"eval_steps_per_second": 26.227,
|
| 6453 |
+
"step": 9000
|
| 6454 |
+
},
|
| 6455 |
+
{
|
| 6456 |
+
"epoch": 0.23,
|
| 6457 |
+
"grad_norm": 4.375,
|
| 6458 |
+
"learning_rate": 2.0842105263157897e-06,
|
| 6459 |
+
"loss": 0.5752,
|
| 6460 |
+
"step": 9010
|
| 6461 |
+
},
|
| 6462 |
+
{
|
| 6463 |
+
"epoch": 0.23,
|
| 6464 |
+
"grad_norm": 4.9375,
|
| 6465 |
+
"learning_rate": 2.0631578947368424e-06,
|
| 6466 |
+
"loss": 0.5578,
|
| 6467 |
+
"step": 9020
|
| 6468 |
+
},
|
| 6469 |
+
{
|
| 6470 |
+
"epoch": 0.23,
|
| 6471 |
+
"grad_norm": 3.9375,
|
| 6472 |
+
"learning_rate": 2.042105263157895e-06,
|
| 6473 |
+
"loss": 0.6715,
|
| 6474 |
+
"step": 9030
|
| 6475 |
+
},
|
| 6476 |
+
{
|
| 6477 |
+
"epoch": 0.23,
|
| 6478 |
+
"grad_norm": 6.28125,
|
| 6479 |
+
"learning_rate": 2.0210526315789477e-06,
|
| 6480 |
+
"loss": 0.5559,
|
| 6481 |
+
"step": 9040
|
| 6482 |
+
},
|
| 6483 |
+
{
|
| 6484 |
+
"epoch": 0.23,
|
| 6485 |
+
"grad_norm": 2.734375,
|
| 6486 |
+
"learning_rate": 2.0000000000000003e-06,
|
| 6487 |
+
"loss": 0.5839,
|
| 6488 |
+
"step": 9050
|
| 6489 |
+
},
|
| 6490 |
+
{
|
| 6491 |
+
"epoch": 0.23,
|
| 6492 |
+
"grad_norm": 4.375,
|
| 6493 |
+
"learning_rate": 1.978947368421053e-06,
|
| 6494 |
+
"loss": 0.6546,
|
| 6495 |
+
"step": 9060
|
| 6496 |
+
},
|
| 6497 |
+
{
|
| 6498 |
+
"epoch": 0.23,
|
| 6499 |
+
"grad_norm": 5.1875,
|
| 6500 |
+
"learning_rate": 1.9578947368421052e-06,
|
| 6501 |
+
"loss": 0.4581,
|
| 6502 |
+
"step": 9070
|
| 6503 |
+
},
|
| 6504 |
+
{
|
| 6505 |
+
"epoch": 0.23,
|
| 6506 |
+
"grad_norm": 1.8046875,
|
| 6507 |
+
"learning_rate": 1.936842105263158e-06,
|
| 6508 |
+
"loss": 0.6111,
|
| 6509 |
+
"step": 9080
|
| 6510 |
+
},
|
| 6511 |
+
{
|
| 6512 |
+
"epoch": 0.23,
|
| 6513 |
+
"grad_norm": 4.0625,
|
| 6514 |
+
"learning_rate": 1.9157894736842105e-06,
|
| 6515 |
+
"loss": 0.569,
|
| 6516 |
+
"step": 9090
|
| 6517 |
+
},
|
| 6518 |
+
{
|
| 6519 |
+
"epoch": 0.23,
|
| 6520 |
+
"grad_norm": 3.25,
|
| 6521 |
+
"learning_rate": 1.8947368421052634e-06,
|
| 6522 |
+
"loss": 0.6089,
|
| 6523 |
+
"step": 9100
|
| 6524 |
+
},
|
| 6525 |
+
{
|
| 6526 |
+
"epoch": 0.23,
|
| 6527 |
+
"grad_norm": 8.0625,
|
| 6528 |
+
"learning_rate": 1.8736842105263158e-06,
|
| 6529 |
+
"loss": 0.5164,
|
| 6530 |
+
"step": 9110
|
| 6531 |
+
},
|
| 6532 |
+
{
|
| 6533 |
+
"epoch": 0.23,
|
| 6534 |
+
"grad_norm": 2.875,
|
| 6535 |
+
"learning_rate": 1.8526315789473687e-06,
|
| 6536 |
+
"loss": 0.5259,
|
| 6537 |
+
"step": 9120
|
| 6538 |
+
},
|
| 6539 |
+
{
|
| 6540 |
+
"epoch": 0.23,
|
| 6541 |
+
"grad_norm": 4.0625,
|
| 6542 |
+
"learning_rate": 1.8315789473684211e-06,
|
| 6543 |
+
"loss": 0.6318,
|
| 6544 |
+
"step": 9130
|
| 6545 |
+
},
|
| 6546 |
+
{
|
| 6547 |
+
"epoch": 0.23,
|
| 6548 |
+
"grad_norm": 3.625,
|
| 6549 |
+
"learning_rate": 1.810526315789474e-06,
|
| 6550 |
+
"loss": 0.5954,
|
| 6551 |
+
"step": 9140
|
| 6552 |
+
},
|
| 6553 |
+
{
|
| 6554 |
+
"epoch": 0.23,
|
| 6555 |
+
"grad_norm": 4.25,
|
| 6556 |
+
"learning_rate": 1.7894736842105265e-06,
|
| 6557 |
+
"loss": 0.5146,
|
| 6558 |
+
"step": 9150
|
| 6559 |
+
},
|
| 6560 |
+
{
|
| 6561 |
+
"epoch": 0.23,
|
| 6562 |
+
"grad_norm": 5.28125,
|
| 6563 |
+
"learning_rate": 1.768421052631579e-06,
|
| 6564 |
+
"loss": 0.6194,
|
| 6565 |
+
"step": 9160
|
| 6566 |
+
},
|
| 6567 |
+
{
|
| 6568 |
+
"epoch": 0.23,
|
| 6569 |
+
"grad_norm": 3.3125,
|
| 6570 |
+
"learning_rate": 1.7473684210526318e-06,
|
| 6571 |
+
"loss": 0.5031,
|
| 6572 |
+
"step": 9170
|
| 6573 |
+
},
|
| 6574 |
+
{
|
| 6575 |
+
"epoch": 0.23,
|
| 6576 |
+
"grad_norm": 3.59375,
|
| 6577 |
+
"learning_rate": 1.7263157894736842e-06,
|
| 6578 |
+
"loss": 0.5199,
|
| 6579 |
+
"step": 9180
|
| 6580 |
+
},
|
| 6581 |
+
{
|
| 6582 |
+
"epoch": 0.23,
|
| 6583 |
+
"grad_norm": 12.75,
|
| 6584 |
+
"learning_rate": 1.705263157894737e-06,
|
| 6585 |
+
"loss": 0.6261,
|
| 6586 |
+
"step": 9190
|
| 6587 |
+
},
|
| 6588 |
+
{
|
| 6589 |
+
"epoch": 0.23,
|
| 6590 |
+
"grad_norm": 3.640625,
|
| 6591 |
+
"learning_rate": 1.6842105263157895e-06,
|
| 6592 |
+
"loss": 0.6349,
|
| 6593 |
+
"step": 9200
|
| 6594 |
+
},
|
| 6595 |
+
{
|
| 6596 |
+
"epoch": 0.23,
|
| 6597 |
+
"grad_norm": 3.59375,
|
| 6598 |
+
"learning_rate": 1.6631578947368424e-06,
|
| 6599 |
+
"loss": 0.5244,
|
| 6600 |
+
"step": 9210
|
| 6601 |
+
},
|
| 6602 |
+
{
|
| 6603 |
+
"epoch": 0.23,
|
| 6604 |
+
"grad_norm": 5.15625,
|
| 6605 |
+
"learning_rate": 1.6421052631578948e-06,
|
| 6606 |
+
"loss": 0.5428,
|
| 6607 |
+
"step": 9220
|
| 6608 |
+
},
|
| 6609 |
+
{
|
| 6610 |
+
"epoch": 0.23,
|
| 6611 |
+
"grad_norm": 2.625,
|
| 6612 |
+
"learning_rate": 1.6210526315789473e-06,
|
| 6613 |
+
"loss": 0.6296,
|
| 6614 |
+
"step": 9230
|
| 6615 |
+
},
|
| 6616 |
+
{
|
| 6617 |
+
"epoch": 0.23,
|
| 6618 |
+
"grad_norm": 4.125,
|
| 6619 |
+
"learning_rate": 1.6000000000000001e-06,
|
| 6620 |
+
"loss": 0.5923,
|
| 6621 |
+
"step": 9240
|
| 6622 |
+
},
|
| 6623 |
+
{
|
| 6624 |
+
"epoch": 0.23,
|
| 6625 |
+
"grad_norm": 3.875,
|
| 6626 |
+
"learning_rate": 1.5789473684210526e-06,
|
| 6627 |
+
"loss": 0.6146,
|
| 6628 |
+
"step": 9250
|
| 6629 |
+
},
|
| 6630 |
+
{
|
| 6631 |
+
"epoch": 0.23,
|
| 6632 |
+
"grad_norm": 6.4375,
|
| 6633 |
+
"learning_rate": 1.5578947368421054e-06,
|
| 6634 |
+
"loss": 0.6558,
|
| 6635 |
+
"step": 9260
|
| 6636 |
+
},
|
| 6637 |
+
{
|
| 6638 |
+
"epoch": 0.23,
|
| 6639 |
+
"grad_norm": 5.28125,
|
| 6640 |
+
"learning_rate": 1.5368421052631579e-06,
|
| 6641 |
+
"loss": 0.5426,
|
| 6642 |
+
"step": 9270
|
| 6643 |
+
},
|
| 6644 |
+
{
|
| 6645 |
+
"epoch": 0.23,
|
| 6646 |
+
"grad_norm": 3.046875,
|
| 6647 |
+
"learning_rate": 1.5157894736842108e-06,
|
| 6648 |
+
"loss": 0.611,
|
| 6649 |
+
"step": 9280
|
| 6650 |
+
},
|
| 6651 |
+
{
|
| 6652 |
+
"epoch": 0.23,
|
| 6653 |
+
"grad_norm": 4.5625,
|
| 6654 |
+
"learning_rate": 1.4947368421052632e-06,
|
| 6655 |
+
"loss": 0.5013,
|
| 6656 |
+
"step": 9290
|
| 6657 |
+
},
|
| 6658 |
+
{
|
| 6659 |
+
"epoch": 0.23,
|
| 6660 |
+
"grad_norm": 3.34375,
|
| 6661 |
+
"learning_rate": 1.4736842105263159e-06,
|
| 6662 |
+
"loss": 0.599,
|
| 6663 |
+
"step": 9300
|
| 6664 |
+
},
|
| 6665 |
+
{
|
| 6666 |
+
"epoch": 0.23,
|
| 6667 |
+
"grad_norm": 4.3125,
|
| 6668 |
+
"learning_rate": 1.4526315789473685e-06,
|
| 6669 |
+
"loss": 0.5549,
|
| 6670 |
+
"step": 9310
|
| 6671 |
+
},
|
| 6672 |
+
{
|
| 6673 |
+
"epoch": 0.23,
|
| 6674 |
+
"grad_norm": 7.375,
|
| 6675 |
+
"learning_rate": 1.4315789473684212e-06,
|
| 6676 |
+
"loss": 0.5597,
|
| 6677 |
+
"step": 9320
|
| 6678 |
+
},
|
| 6679 |
+
{
|
| 6680 |
+
"epoch": 0.23,
|
| 6681 |
+
"grad_norm": 4.9375,
|
| 6682 |
+
"learning_rate": 1.4105263157894738e-06,
|
| 6683 |
+
"loss": 0.5402,
|
| 6684 |
+
"step": 9330
|
| 6685 |
+
},
|
| 6686 |
+
{
|
| 6687 |
+
"epoch": 0.23,
|
| 6688 |
+
"grad_norm": 3.34375,
|
| 6689 |
+
"learning_rate": 1.3894736842105263e-06,
|
| 6690 |
+
"loss": 0.56,
|
| 6691 |
+
"step": 9340
|
| 6692 |
+
},
|
| 6693 |
+
{
|
| 6694 |
+
"epoch": 0.23,
|
| 6695 |
+
"grad_norm": 3.5,
|
| 6696 |
+
"learning_rate": 1.3684210526315791e-06,
|
| 6697 |
+
"loss": 0.5991,
|
| 6698 |
+
"step": 9350
|
| 6699 |
+
},
|
| 6700 |
+
{
|
| 6701 |
+
"epoch": 0.23,
|
| 6702 |
+
"grad_norm": 3.625,
|
| 6703 |
+
"learning_rate": 1.3473684210526316e-06,
|
| 6704 |
+
"loss": 0.499,
|
| 6705 |
+
"step": 9360
|
| 6706 |
+
},
|
| 6707 |
+
{
|
| 6708 |
+
"epoch": 0.23,
|
| 6709 |
+
"grad_norm": 4.125,
|
| 6710 |
+
"learning_rate": 1.3263157894736844e-06,
|
| 6711 |
+
"loss": 0.5946,
|
| 6712 |
+
"step": 9370
|
| 6713 |
+
},
|
| 6714 |
+
{
|
| 6715 |
+
"epoch": 0.23,
|
| 6716 |
+
"grad_norm": 4.59375,
|
| 6717 |
+
"learning_rate": 1.3052631578947369e-06,
|
| 6718 |
+
"loss": 0.678,
|
| 6719 |
+
"step": 9380
|
| 6720 |
+
},
|
| 6721 |
+
{
|
| 6722 |
+
"epoch": 0.23,
|
| 6723 |
+
"grad_norm": 2.9375,
|
| 6724 |
+
"learning_rate": 1.2842105263157895e-06,
|
| 6725 |
+
"loss": 0.7289,
|
| 6726 |
+
"step": 9390
|
| 6727 |
+
},
|
| 6728 |
+
{
|
| 6729 |
+
"epoch": 0.23,
|
| 6730 |
+
"grad_norm": 3.953125,
|
| 6731 |
+
"learning_rate": 1.2631578947368422e-06,
|
| 6732 |
+
"loss": 0.6926,
|
| 6733 |
+
"step": 9400
|
| 6734 |
+
},
|
| 6735 |
+
{
|
| 6736 |
+
"epoch": 0.24,
|
| 6737 |
+
"grad_norm": 9.1875,
|
| 6738 |
+
"learning_rate": 1.2421052631578948e-06,
|
| 6739 |
+
"loss": 0.6079,
|
| 6740 |
+
"step": 9410
|
| 6741 |
+
},
|
| 6742 |
+
{
|
| 6743 |
+
"epoch": 0.24,
|
| 6744 |
+
"grad_norm": 9.4375,
|
| 6745 |
+
"learning_rate": 1.2210526315789475e-06,
|
| 6746 |
+
"loss": 0.6642,
|
| 6747 |
+
"step": 9420
|
| 6748 |
+
},
|
| 6749 |
+
{
|
| 6750 |
+
"epoch": 0.24,
|
| 6751 |
+
"grad_norm": 4.28125,
|
| 6752 |
+
"learning_rate": 1.2000000000000002e-06,
|
| 6753 |
+
"loss": 0.5805,
|
| 6754 |
+
"step": 9430
|
| 6755 |
+
},
|
| 6756 |
+
{
|
| 6757 |
+
"epoch": 0.24,
|
| 6758 |
+
"grad_norm": 4.6875,
|
| 6759 |
+
"learning_rate": 1.1789473684210526e-06,
|
| 6760 |
+
"loss": 0.564,
|
| 6761 |
+
"step": 9440
|
| 6762 |
+
},
|
| 6763 |
+
{
|
| 6764 |
+
"epoch": 0.24,
|
| 6765 |
+
"grad_norm": 3.34375,
|
| 6766 |
+
"learning_rate": 1.1578947368421053e-06,
|
| 6767 |
+
"loss": 0.571,
|
| 6768 |
+
"step": 9450
|
| 6769 |
+
},
|
| 6770 |
+
{
|
| 6771 |
+
"epoch": 0.24,
|
| 6772 |
+
"grad_norm": 4.625,
|
| 6773 |
+
"learning_rate": 1.136842105263158e-06,
|
| 6774 |
+
"loss": 0.538,
|
| 6775 |
+
"step": 9460
|
| 6776 |
+
},
|
| 6777 |
+
{
|
| 6778 |
+
"epoch": 0.24,
|
| 6779 |
+
"grad_norm": 7.84375,
|
| 6780 |
+
"learning_rate": 1.1157894736842106e-06,
|
| 6781 |
+
"loss": 0.5777,
|
| 6782 |
+
"step": 9470
|
| 6783 |
+
},
|
| 6784 |
+
{
|
| 6785 |
+
"epoch": 0.24,
|
| 6786 |
+
"grad_norm": 3.65625,
|
| 6787 |
+
"learning_rate": 1.0947368421052632e-06,
|
| 6788 |
+
"loss": 0.683,
|
| 6789 |
+
"step": 9480
|
| 6790 |
+
},
|
| 6791 |
+
{
|
| 6792 |
+
"epoch": 0.24,
|
| 6793 |
+
"grad_norm": 3.3125,
|
| 6794 |
+
"learning_rate": 1.0736842105263159e-06,
|
| 6795 |
+
"loss": 0.5143,
|
| 6796 |
+
"step": 9490
|
| 6797 |
+
},
|
| 6798 |
+
{
|
| 6799 |
+
"epoch": 0.24,
|
| 6800 |
+
"grad_norm": 3.109375,
|
| 6801 |
+
"learning_rate": 1.0526315789473685e-06,
|
| 6802 |
+
"loss": 0.4899,
|
| 6803 |
+
"step": 9500
|
| 6804 |
+
},
|
| 6805 |
+
{
|
| 6806 |
+
"epoch": 0.24,
|
| 6807 |
+
"eval_loss": 0.6154074668884277,
|
| 6808 |
+
"eval_runtime": 38.1761,
|
| 6809 |
+
"eval_samples_per_second": 26.194,
|
| 6810 |
+
"eval_steps_per_second": 26.194,
|
| 6811 |
+
"step": 9500
|
| 6812 |
+
},
|
| 6813 |
+
{
|
| 6814 |
+
"epoch": 0.24,
|
| 6815 |
+
"grad_norm": 7.125,
|
| 6816 |
+
"learning_rate": 1.0315789473684212e-06,
|
| 6817 |
+
"loss": 0.7389,
|
| 6818 |
+
"step": 9510
|
| 6819 |
+
},
|
| 6820 |
+
{
|
| 6821 |
+
"epoch": 0.24,
|
| 6822 |
+
"grad_norm": 3.734375,
|
| 6823 |
+
"learning_rate": 1.0105263157894738e-06,
|
| 6824 |
+
"loss": 0.5922,
|
| 6825 |
+
"step": 9520
|
| 6826 |
+
},
|
| 6827 |
+
{
|
| 6828 |
+
"epoch": 0.24,
|
| 6829 |
+
"grad_norm": 4.46875,
|
| 6830 |
+
"learning_rate": 9.894736842105265e-07,
|
| 6831 |
+
"loss": 0.5658,
|
| 6832 |
+
"step": 9530
|
| 6833 |
+
},
|
| 6834 |
+
{
|
| 6835 |
+
"epoch": 0.24,
|
| 6836 |
+
"grad_norm": 3.765625,
|
| 6837 |
+
"learning_rate": 9.68421052631579e-07,
|
| 6838 |
+
"loss": 0.608,
|
| 6839 |
+
"step": 9540
|
| 6840 |
+
},
|
| 6841 |
+
{
|
| 6842 |
+
"epoch": 0.24,
|
| 6843 |
+
"grad_norm": 3.78125,
|
| 6844 |
+
"learning_rate": 9.473684210526317e-07,
|
| 6845 |
+
"loss": 0.5337,
|
| 6846 |
+
"step": 9550
|
| 6847 |
+
},
|
| 6848 |
+
{
|
| 6849 |
+
"epoch": 0.24,
|
| 6850 |
+
"grad_norm": 3.515625,
|
| 6851 |
+
"learning_rate": 9.263157894736844e-07,
|
| 6852 |
+
"loss": 0.5736,
|
| 6853 |
+
"step": 9560
|
| 6854 |
+
},
|
| 6855 |
+
{
|
| 6856 |
+
"epoch": 0.24,
|
| 6857 |
+
"grad_norm": 4.03125,
|
| 6858 |
+
"learning_rate": 9.05263157894737e-07,
|
| 6859 |
+
"loss": 0.5319,
|
| 6860 |
+
"step": 9570
|
| 6861 |
+
},
|
| 6862 |
+
{
|
| 6863 |
+
"epoch": 0.24,
|
| 6864 |
+
"grad_norm": 3.859375,
|
| 6865 |
+
"learning_rate": 8.842105263157895e-07,
|
| 6866 |
+
"loss": 0.5011,
|
| 6867 |
+
"step": 9580
|
| 6868 |
+
},
|
| 6869 |
+
{
|
| 6870 |
+
"epoch": 0.24,
|
| 6871 |
+
"grad_norm": 2.828125,
|
| 6872 |
+
"learning_rate": 8.631578947368421e-07,
|
| 6873 |
+
"loss": 0.6734,
|
| 6874 |
+
"step": 9590
|
| 6875 |
+
},
|
| 6876 |
+
{
|
| 6877 |
+
"epoch": 0.24,
|
| 6878 |
+
"grad_norm": 2.625,
|
| 6879 |
+
"learning_rate": 8.421052631578948e-07,
|
| 6880 |
+
"loss": 0.5045,
|
| 6881 |
+
"step": 9600
|
| 6882 |
+
},
|
| 6883 |
+
{
|
| 6884 |
+
"epoch": 0.24,
|
| 6885 |
+
"grad_norm": 4.78125,
|
| 6886 |
+
"learning_rate": 8.210526315789474e-07,
|
| 6887 |
+
"loss": 0.5957,
|
| 6888 |
+
"step": 9610
|
| 6889 |
+
},
|
| 6890 |
+
{
|
| 6891 |
+
"epoch": 0.24,
|
| 6892 |
+
"grad_norm": 2.984375,
|
| 6893 |
+
"learning_rate": 8.000000000000001e-07,
|
| 6894 |
+
"loss": 0.6944,
|
| 6895 |
+
"step": 9620
|
| 6896 |
+
},
|
| 6897 |
+
{
|
| 6898 |
+
"epoch": 0.24,
|
| 6899 |
+
"grad_norm": 2.546875,
|
| 6900 |
+
"learning_rate": 7.789473684210527e-07,
|
| 6901 |
+
"loss": 0.5249,
|
| 6902 |
+
"step": 9630
|
| 6903 |
+
},
|
| 6904 |
+
{
|
| 6905 |
+
"epoch": 0.24,
|
| 6906 |
+
"grad_norm": 12.9375,
|
| 6907 |
+
"learning_rate": 7.578947368421054e-07,
|
| 6908 |
+
"loss": 0.5981,
|
| 6909 |
+
"step": 9640
|
| 6910 |
+
},
|
| 6911 |
+
{
|
| 6912 |
+
"epoch": 0.24,
|
| 6913 |
+
"grad_norm": 7.25,
|
| 6914 |
+
"learning_rate": 7.368421052631579e-07,
|
| 6915 |
+
"loss": 0.633,
|
| 6916 |
+
"step": 9650
|
| 6917 |
+
},
|
| 6918 |
+
{
|
| 6919 |
+
"epoch": 0.24,
|
| 6920 |
+
"grad_norm": 3.59375,
|
| 6921 |
+
"learning_rate": 7.157894736842106e-07,
|
| 6922 |
+
"loss": 0.5535,
|
| 6923 |
+
"step": 9660
|
| 6924 |
+
},
|
| 6925 |
+
{
|
| 6926 |
+
"epoch": 0.24,
|
| 6927 |
+
"grad_norm": 5.8125,
|
| 6928 |
+
"learning_rate": 6.947368421052631e-07,
|
| 6929 |
+
"loss": 0.6233,
|
| 6930 |
+
"step": 9670
|
| 6931 |
+
},
|
| 6932 |
+
{
|
| 6933 |
+
"epoch": 0.24,
|
| 6934 |
+
"grad_norm": 11.4375,
|
| 6935 |
+
"learning_rate": 6.736842105263158e-07,
|
| 6936 |
+
"loss": 0.5603,
|
| 6937 |
+
"step": 9680
|
| 6938 |
+
},
|
| 6939 |
+
{
|
| 6940 |
+
"epoch": 0.24,
|
| 6941 |
+
"grad_norm": 3.890625,
|
| 6942 |
+
"learning_rate": 6.526315789473684e-07,
|
| 6943 |
+
"loss": 0.5977,
|
| 6944 |
+
"step": 9690
|
| 6945 |
+
},
|
| 6946 |
+
{
|
| 6947 |
+
"epoch": 0.24,
|
| 6948 |
+
"grad_norm": 3.25,
|
| 6949 |
+
"learning_rate": 6.315789473684211e-07,
|
| 6950 |
+
"loss": 0.5737,
|
| 6951 |
+
"step": 9700
|
| 6952 |
+
},
|
| 6953 |
+
{
|
| 6954 |
+
"epoch": 0.24,
|
| 6955 |
+
"grad_norm": 8.0625,
|
| 6956 |
+
"learning_rate": 6.105263157894738e-07,
|
| 6957 |
+
"loss": 0.5797,
|
| 6958 |
+
"step": 9710
|
| 6959 |
+
},
|
| 6960 |
+
{
|
| 6961 |
+
"epoch": 0.24,
|
| 6962 |
+
"grad_norm": 6.4375,
|
| 6963 |
+
"learning_rate": 5.894736842105263e-07,
|
| 6964 |
+
"loss": 0.6256,
|
| 6965 |
+
"step": 9720
|
| 6966 |
+
},
|
| 6967 |
+
{
|
| 6968 |
+
"epoch": 0.24,
|
| 6969 |
+
"grad_norm": 2.484375,
|
| 6970 |
+
"learning_rate": 5.68421052631579e-07,
|
| 6971 |
+
"loss": 0.4984,
|
| 6972 |
+
"step": 9730
|
| 6973 |
+
},
|
| 6974 |
+
{
|
| 6975 |
+
"epoch": 0.24,
|
| 6976 |
+
"grad_norm": 4.28125,
|
| 6977 |
+
"learning_rate": 5.473684210526316e-07,
|
| 6978 |
+
"loss": 0.6489,
|
| 6979 |
+
"step": 9740
|
| 6980 |
+
},
|
| 6981 |
+
{
|
| 6982 |
+
"epoch": 0.24,
|
| 6983 |
+
"grad_norm": 2.328125,
|
| 6984 |
+
"learning_rate": 5.263157894736843e-07,
|
| 6985 |
+
"loss": 0.5204,
|
| 6986 |
+
"step": 9750
|
| 6987 |
+
},
|
| 6988 |
+
{
|
| 6989 |
+
"epoch": 0.24,
|
| 6990 |
+
"grad_norm": 3.09375,
|
| 6991 |
+
"learning_rate": 5.052631578947369e-07,
|
| 6992 |
+
"loss": 0.6044,
|
| 6993 |
+
"step": 9760
|
| 6994 |
+
},
|
| 6995 |
+
{
|
| 6996 |
+
"epoch": 0.24,
|
| 6997 |
+
"grad_norm": 3.078125,
|
| 6998 |
+
"learning_rate": 4.842105263157895e-07,
|
| 6999 |
+
"loss": 0.6061,
|
| 7000 |
+
"step": 9770
|
| 7001 |
+
},
|
| 7002 |
+
{
|
| 7003 |
+
"epoch": 0.24,
|
| 7004 |
+
"grad_norm": 5.59375,
|
| 7005 |
+
"learning_rate": 4.631578947368422e-07,
|
| 7006 |
+
"loss": 0.5491,
|
| 7007 |
+
"step": 9780
|
| 7008 |
+
},
|
| 7009 |
+
{
|
| 7010 |
+
"epoch": 0.24,
|
| 7011 |
+
"grad_norm": 3.46875,
|
| 7012 |
+
"learning_rate": 4.421052631578947e-07,
|
| 7013 |
+
"loss": 0.5567,
|
| 7014 |
+
"step": 9790
|
| 7015 |
+
},
|
| 7016 |
+
{
|
| 7017 |
+
"epoch": 0.24,
|
| 7018 |
+
"grad_norm": 3.671875,
|
| 7019 |
+
"learning_rate": 4.210526315789474e-07,
|
| 7020 |
+
"loss": 0.6661,
|
| 7021 |
+
"step": 9800
|
| 7022 |
+
},
|
| 7023 |
+
{
|
| 7024 |
+
"epoch": 0.25,
|
| 7025 |
+
"grad_norm": 2.875,
|
| 7026 |
+
"learning_rate": 4.0000000000000003e-07,
|
| 7027 |
+
"loss": 0.672,
|
| 7028 |
+
"step": 9810
|
| 7029 |
+
},
|
| 7030 |
+
{
|
| 7031 |
+
"epoch": 0.25,
|
| 7032 |
+
"grad_norm": 7.15625,
|
| 7033 |
+
"learning_rate": 3.789473684210527e-07,
|
| 7034 |
+
"loss": 0.5887,
|
| 7035 |
+
"step": 9820
|
| 7036 |
+
},
|
| 7037 |
+
{
|
| 7038 |
+
"epoch": 0.25,
|
| 7039 |
+
"grad_norm": 2.453125,
|
| 7040 |
+
"learning_rate": 3.578947368421053e-07,
|
| 7041 |
+
"loss": 0.4978,
|
| 7042 |
+
"step": 9830
|
| 7043 |
+
},
|
| 7044 |
+
{
|
| 7045 |
+
"epoch": 0.25,
|
| 7046 |
+
"grad_norm": 2.859375,
|
| 7047 |
+
"learning_rate": 3.368421052631579e-07,
|
| 7048 |
+
"loss": 0.5254,
|
| 7049 |
+
"step": 9840
|
| 7050 |
+
},
|
| 7051 |
+
{
|
| 7052 |
+
"epoch": 0.25,
|
| 7053 |
+
"grad_norm": 3.515625,
|
| 7054 |
+
"learning_rate": 3.1578947368421055e-07,
|
| 7055 |
+
"loss": 0.5738,
|
| 7056 |
+
"step": 9850
|
| 7057 |
+
},
|
| 7058 |
+
{
|
| 7059 |
+
"epoch": 0.25,
|
| 7060 |
+
"grad_norm": 3.828125,
|
| 7061 |
+
"learning_rate": 2.9473684210526315e-07,
|
| 7062 |
+
"loss": 0.5846,
|
| 7063 |
+
"step": 9860
|
| 7064 |
+
},
|
| 7065 |
+
{
|
| 7066 |
+
"epoch": 0.25,
|
| 7067 |
+
"grad_norm": 6.0625,
|
| 7068 |
+
"learning_rate": 2.736842105263158e-07,
|
| 7069 |
+
"loss": 0.6591,
|
| 7070 |
+
"step": 9870
|
| 7071 |
+
},
|
| 7072 |
+
{
|
| 7073 |
+
"epoch": 0.25,
|
| 7074 |
+
"grad_norm": 5.0,
|
| 7075 |
+
"learning_rate": 2.5263157894736846e-07,
|
| 7076 |
+
"loss": 0.6697,
|
| 7077 |
+
"step": 9880
|
| 7078 |
+
},
|
| 7079 |
+
{
|
| 7080 |
+
"epoch": 0.25,
|
| 7081 |
+
"grad_norm": 4.25,
|
| 7082 |
+
"learning_rate": 2.315789473684211e-07,
|
| 7083 |
+
"loss": 0.4831,
|
| 7084 |
+
"step": 9890
|
| 7085 |
+
},
|
| 7086 |
+
{
|
| 7087 |
+
"epoch": 0.25,
|
| 7088 |
+
"grad_norm": 2.65625,
|
| 7089 |
+
"learning_rate": 2.105263157894737e-07,
|
| 7090 |
+
"loss": 0.5504,
|
| 7091 |
+
"step": 9900
|
| 7092 |
+
},
|
| 7093 |
+
{
|
| 7094 |
+
"epoch": 0.25,
|
| 7095 |
+
"grad_norm": 2.65625,
|
| 7096 |
+
"learning_rate": 1.8947368421052634e-07,
|
| 7097 |
+
"loss": 0.5576,
|
| 7098 |
+
"step": 9910
|
| 7099 |
+
},
|
| 7100 |
+
{
|
| 7101 |
+
"epoch": 0.25,
|
| 7102 |
+
"grad_norm": 3.875,
|
| 7103 |
+
"learning_rate": 1.6842105263157895e-07,
|
| 7104 |
+
"loss": 0.4733,
|
| 7105 |
+
"step": 9920
|
| 7106 |
+
},
|
| 7107 |
+
{
|
| 7108 |
+
"epoch": 0.25,
|
| 7109 |
+
"grad_norm": 3.03125,
|
| 7110 |
+
"learning_rate": 1.4736842105263158e-07,
|
| 7111 |
+
"loss": 0.5077,
|
| 7112 |
+
"step": 9930
|
| 7113 |
+
},
|
| 7114 |
+
{
|
| 7115 |
+
"epoch": 0.25,
|
| 7116 |
+
"grad_norm": 4.5,
|
| 7117 |
+
"learning_rate": 1.2631578947368423e-07,
|
| 7118 |
+
"loss": 0.5068,
|
| 7119 |
+
"step": 9940
|
| 7120 |
+
},
|
| 7121 |
+
{
|
| 7122 |
+
"epoch": 0.25,
|
| 7123 |
+
"grad_norm": 2.453125,
|
| 7124 |
+
"learning_rate": 1.0526315789473685e-07,
|
| 7125 |
+
"loss": 0.4754,
|
| 7126 |
+
"step": 9950
|
| 7127 |
+
},
|
| 7128 |
+
{
|
| 7129 |
+
"epoch": 0.25,
|
| 7130 |
+
"grad_norm": 6.09375,
|
| 7131 |
+
"learning_rate": 8.421052631578947e-08,
|
| 7132 |
+
"loss": 0.5751,
|
| 7133 |
+
"step": 9960
|
| 7134 |
+
},
|
| 7135 |
+
{
|
| 7136 |
+
"epoch": 0.25,
|
| 7137 |
+
"grad_norm": 3.65625,
|
| 7138 |
+
"learning_rate": 6.315789473684211e-08,
|
| 7139 |
+
"loss": 0.5133,
|
| 7140 |
+
"step": 9970
|
| 7141 |
+
},
|
| 7142 |
+
{
|
| 7143 |
+
"epoch": 0.25,
|
| 7144 |
+
"grad_norm": 3.171875,
|
| 7145 |
+
"learning_rate": 4.2105263157894737e-08,
|
| 7146 |
+
"loss": 0.6084,
|
| 7147 |
+
"step": 9980
|
| 7148 |
+
},
|
| 7149 |
+
{
|
| 7150 |
+
"epoch": 0.25,
|
| 7151 |
+
"grad_norm": 16.5,
|
| 7152 |
+
"learning_rate": 2.1052631578947368e-08,
|
| 7153 |
+
"loss": 0.6159,
|
| 7154 |
+
"step": 9990
|
| 7155 |
+
},
|
| 7156 |
+
{
|
| 7157 |
+
"epoch": 0.25,
|
| 7158 |
+
"grad_norm": 4.46875,
|
| 7159 |
+
"learning_rate": 0.0,
|
| 7160 |
+
"loss": 0.6108,
|
| 7161 |
+
"step": 10000
|
| 7162 |
+
},
|
| 7163 |
+
{
|
| 7164 |
+
"epoch": 0.25,
|
| 7165 |
+
"eval_loss": 0.6137469410896301,
|
| 7166 |
+
"eval_runtime": 38.1278,
|
| 7167 |
+
"eval_samples_per_second": 26.228,
|
| 7168 |
+
"eval_steps_per_second": 26.228,
|
| 7169 |
+
"step": 10000
|
| 7170 |
}
|
| 7171 |
],
|
| 7172 |
"logging_steps": 10,
|
|
|
|
| 7174 |
"num_input_tokens_seen": 0,
|
| 7175 |
"num_train_epochs": 1,
|
| 7176 |
"save_steps": 2500,
|
| 7177 |
+
"total_flos": 1.5733698330624e+17,
|
| 7178 |
"train_batch_size": 1,
|
| 7179 |
"trial_name": null,
|
| 7180 |
"trial_params": null
|