Uploaded checkpoint-5000
Browse files- adapter_model.safetensors +1 -1
- optimizer.pt +1 -1
- rng_state.pth +1 -1
- scheduler.pt +1 -1
- trainer_state.json +1793 -3
adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 119975656
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:59a43a50cdfe5cec56a43e50865db5614e4234af0a5044af87ecea2eb6174b21
|
| 3 |
size 119975656
|
optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 240145026
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b954c07030ca5f621b51b0e35175361ead8e8178f37d49b580a3f1ef92d97c55
|
| 3 |
size 240145026
|
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:af4fc59c67c352cae63c079b9681492ee5f0d6b4b95d053649ba31a9ffec9bc2
|
| 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 sha256:2be67f6aac8e482bb2022409709d8774ffb125292c0c9cf025c0ae747f3a6d57
|
| 3 |
size 1064
|
trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": 0.019110437482595444,
|
| 3 |
"best_model_checkpoint": "runs/deepseek_lora_20240422-151242/checkpoint-2500",
|
| 4 |
-
"epoch":
|
| 5 |
"eval_steps": 500,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -1797,6 +1797,1796 @@
|
|
| 1797 |
"eval_samples_per_second": 16.159,
|
| 1798 |
"eval_steps_per_second": 16.159,
|
| 1799 |
"step": 2500
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1800 |
}
|
| 1801 |
],
|
| 1802 |
"logging_steps": 10,
|
|
@@ -1804,7 +3594,7 @@
|
|
| 1804 |
"num_input_tokens_seen": 0,
|
| 1805 |
"num_train_epochs": 2,
|
| 1806 |
"save_steps": 2500,
|
| 1807 |
-
"total_flos":
|
| 1808 |
"train_batch_size": 1,
|
| 1809 |
"trial_name": null,
|
| 1810 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": 0.019110437482595444,
|
| 3 |
"best_model_checkpoint": "runs/deepseek_lora_20240422-151242/checkpoint-2500",
|
| 4 |
+
"epoch": 1.5665387326701654,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 5000,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 1797 |
"eval_samples_per_second": 16.159,
|
| 1798 |
"eval_steps_per_second": 16.159,
|
| 1799 |
"step": 2500
|
| 1800 |
+
},
|
| 1801 |
+
{
|
| 1802 |
+
"epoch": 0.79,
|
| 1803 |
+
"grad_norm": 0.0033025951124727726,
|
| 1804 |
+
"learning_rate": 1.1066666666666669e-05,
|
| 1805 |
+
"loss": 0.081,
|
| 1806 |
+
"step": 2510
|
| 1807 |
+
},
|
| 1808 |
+
{
|
| 1809 |
+
"epoch": 0.79,
|
| 1810 |
+
"grad_norm": 0.0032579610124230385,
|
| 1811 |
+
"learning_rate": 1.1022222222222224e-05,
|
| 1812 |
+
"loss": 0.0207,
|
| 1813 |
+
"step": 2520
|
| 1814 |
+
},
|
| 1815 |
+
{
|
| 1816 |
+
"epoch": 0.79,
|
| 1817 |
+
"grad_norm": 1.838714838027954,
|
| 1818 |
+
"learning_rate": 1.0977777777777779e-05,
|
| 1819 |
+
"loss": 0.1599,
|
| 1820 |
+
"step": 2530
|
| 1821 |
+
},
|
| 1822 |
+
{
|
| 1823 |
+
"epoch": 0.8,
|
| 1824 |
+
"grad_norm": 0.0018555274000391364,
|
| 1825 |
+
"learning_rate": 1.0933333333333334e-05,
|
| 1826 |
+
"loss": 0.002,
|
| 1827 |
+
"step": 2540
|
| 1828 |
+
},
|
| 1829 |
+
{
|
| 1830 |
+
"epoch": 0.8,
|
| 1831 |
+
"grad_norm": 0.0020195182878524065,
|
| 1832 |
+
"learning_rate": 1.088888888888889e-05,
|
| 1833 |
+
"loss": 0.008,
|
| 1834 |
+
"step": 2550
|
| 1835 |
+
},
|
| 1836 |
+
{
|
| 1837 |
+
"epoch": 0.8,
|
| 1838 |
+
"grad_norm": 0.0014295022701844573,
|
| 1839 |
+
"learning_rate": 1.0844444444444446e-05,
|
| 1840 |
+
"loss": 0.0276,
|
| 1841 |
+
"step": 2560
|
| 1842 |
+
},
|
| 1843 |
+
{
|
| 1844 |
+
"epoch": 0.81,
|
| 1845 |
+
"grad_norm": 2.145263433456421,
|
| 1846 |
+
"learning_rate": 1.0800000000000002e-05,
|
| 1847 |
+
"loss": 0.0248,
|
| 1848 |
+
"step": 2570
|
| 1849 |
+
},
|
| 1850 |
+
{
|
| 1851 |
+
"epoch": 0.81,
|
| 1852 |
+
"grad_norm": 2.27980375289917,
|
| 1853 |
+
"learning_rate": 1.0755555555555557e-05,
|
| 1854 |
+
"loss": 0.0358,
|
| 1855 |
+
"step": 2580
|
| 1856 |
+
},
|
| 1857 |
+
{
|
| 1858 |
+
"epoch": 0.81,
|
| 1859 |
+
"grad_norm": 2.667983293533325,
|
| 1860 |
+
"learning_rate": 1.0711111111111112e-05,
|
| 1861 |
+
"loss": 0.0651,
|
| 1862 |
+
"step": 2590
|
| 1863 |
+
},
|
| 1864 |
+
{
|
| 1865 |
+
"epoch": 0.81,
|
| 1866 |
+
"grad_norm": 0.0018497890559956431,
|
| 1867 |
+
"learning_rate": 1.0666666666666667e-05,
|
| 1868 |
+
"loss": 0.0387,
|
| 1869 |
+
"step": 2600
|
| 1870 |
+
},
|
| 1871 |
+
{
|
| 1872 |
+
"epoch": 0.82,
|
| 1873 |
+
"grad_norm": 0.0013170776655897498,
|
| 1874 |
+
"learning_rate": 1.0622222222222223e-05,
|
| 1875 |
+
"loss": 0.0173,
|
| 1876 |
+
"step": 2610
|
| 1877 |
+
},
|
| 1878 |
+
{
|
| 1879 |
+
"epoch": 0.82,
|
| 1880 |
+
"grad_norm": 0.004894789308309555,
|
| 1881 |
+
"learning_rate": 1.0577777777777778e-05,
|
| 1882 |
+
"loss": 0.043,
|
| 1883 |
+
"step": 2620
|
| 1884 |
+
},
|
| 1885 |
+
{
|
| 1886 |
+
"epoch": 0.82,
|
| 1887 |
+
"grad_norm": 0.04001269116997719,
|
| 1888 |
+
"learning_rate": 1.0533333333333333e-05,
|
| 1889 |
+
"loss": 0.0926,
|
| 1890 |
+
"step": 2630
|
| 1891 |
+
},
|
| 1892 |
+
{
|
| 1893 |
+
"epoch": 0.83,
|
| 1894 |
+
"grad_norm": 0.0019531013676896691,
|
| 1895 |
+
"learning_rate": 1.048888888888889e-05,
|
| 1896 |
+
"loss": 0.0511,
|
| 1897 |
+
"step": 2640
|
| 1898 |
+
},
|
| 1899 |
+
{
|
| 1900 |
+
"epoch": 0.83,
|
| 1901 |
+
"grad_norm": 2.29565167427063,
|
| 1902 |
+
"learning_rate": 1.0444444444444445e-05,
|
| 1903 |
+
"loss": 0.0803,
|
| 1904 |
+
"step": 2650
|
| 1905 |
+
},
|
| 1906 |
+
{
|
| 1907 |
+
"epoch": 0.83,
|
| 1908 |
+
"grad_norm": 2.7071011066436768,
|
| 1909 |
+
"learning_rate": 1.04e-05,
|
| 1910 |
+
"loss": 0.0604,
|
| 1911 |
+
"step": 2660
|
| 1912 |
+
},
|
| 1913 |
+
{
|
| 1914 |
+
"epoch": 0.84,
|
| 1915 |
+
"grad_norm": 0.0376460887491703,
|
| 1916 |
+
"learning_rate": 1.0355555555555557e-05,
|
| 1917 |
+
"loss": 0.0155,
|
| 1918 |
+
"step": 2670
|
| 1919 |
+
},
|
| 1920 |
+
{
|
| 1921 |
+
"epoch": 0.84,
|
| 1922 |
+
"grad_norm": 0.001858079107478261,
|
| 1923 |
+
"learning_rate": 1.0311111111111113e-05,
|
| 1924 |
+
"loss": 0.0665,
|
| 1925 |
+
"step": 2680
|
| 1926 |
+
},
|
| 1927 |
+
{
|
| 1928 |
+
"epoch": 0.84,
|
| 1929 |
+
"grad_norm": 0.0015587485395371914,
|
| 1930 |
+
"learning_rate": 1.0266666666666668e-05,
|
| 1931 |
+
"loss": 0.0613,
|
| 1932 |
+
"step": 2690
|
| 1933 |
+
},
|
| 1934 |
+
{
|
| 1935 |
+
"epoch": 0.85,
|
| 1936 |
+
"grad_norm": 0.047985997051000595,
|
| 1937 |
+
"learning_rate": 1.0222222222222223e-05,
|
| 1938 |
+
"loss": 0.0032,
|
| 1939 |
+
"step": 2700
|
| 1940 |
+
},
|
| 1941 |
+
{
|
| 1942 |
+
"epoch": 0.85,
|
| 1943 |
+
"grad_norm": 2.2135114669799805,
|
| 1944 |
+
"learning_rate": 1.0177777777777778e-05,
|
| 1945 |
+
"loss": 0.0508,
|
| 1946 |
+
"step": 2710
|
| 1947 |
+
},
|
| 1948 |
+
{
|
| 1949 |
+
"epoch": 0.85,
|
| 1950 |
+
"grad_norm": 1.1044480800628662,
|
| 1951 |
+
"learning_rate": 1.0133333333333335e-05,
|
| 1952 |
+
"loss": 0.0333,
|
| 1953 |
+
"step": 2720
|
| 1954 |
+
},
|
| 1955 |
+
{
|
| 1956 |
+
"epoch": 0.86,
|
| 1957 |
+
"grad_norm": 0.047782231122255325,
|
| 1958 |
+
"learning_rate": 1.008888888888889e-05,
|
| 1959 |
+
"loss": 0.0256,
|
| 1960 |
+
"step": 2730
|
| 1961 |
+
},
|
| 1962 |
+
{
|
| 1963 |
+
"epoch": 0.86,
|
| 1964 |
+
"grad_norm": 0.0029533940833061934,
|
| 1965 |
+
"learning_rate": 1.0044444444444446e-05,
|
| 1966 |
+
"loss": 0.0107,
|
| 1967 |
+
"step": 2740
|
| 1968 |
+
},
|
| 1969 |
+
{
|
| 1970 |
+
"epoch": 0.86,
|
| 1971 |
+
"grad_norm": 0.003864617319777608,
|
| 1972 |
+
"learning_rate": 1e-05,
|
| 1973 |
+
"loss": 0.0508,
|
| 1974 |
+
"step": 2750
|
| 1975 |
+
},
|
| 1976 |
+
{
|
| 1977 |
+
"epoch": 0.86,
|
| 1978 |
+
"grad_norm": 0.006308398675173521,
|
| 1979 |
+
"learning_rate": 9.955555555555556e-06,
|
| 1980 |
+
"loss": 0.0599,
|
| 1981 |
+
"step": 2760
|
| 1982 |
+
},
|
| 1983 |
+
{
|
| 1984 |
+
"epoch": 0.87,
|
| 1985 |
+
"grad_norm": 0.0014458250952884555,
|
| 1986 |
+
"learning_rate": 9.911111111111113e-06,
|
| 1987 |
+
"loss": 0.0457,
|
| 1988 |
+
"step": 2770
|
| 1989 |
+
},
|
| 1990 |
+
{
|
| 1991 |
+
"epoch": 0.87,
|
| 1992 |
+
"grad_norm": 0.0018096743151545525,
|
| 1993 |
+
"learning_rate": 9.866666666666668e-06,
|
| 1994 |
+
"loss": 0.0773,
|
| 1995 |
+
"step": 2780
|
| 1996 |
+
},
|
| 1997 |
+
{
|
| 1998 |
+
"epoch": 0.87,
|
| 1999 |
+
"grad_norm": 0.002803443931043148,
|
| 2000 |
+
"learning_rate": 9.822222222222223e-06,
|
| 2001 |
+
"loss": 0.0229,
|
| 2002 |
+
"step": 2790
|
| 2003 |
+
},
|
| 2004 |
+
{
|
| 2005 |
+
"epoch": 0.88,
|
| 2006 |
+
"grad_norm": 0.0039856089279055595,
|
| 2007 |
+
"learning_rate": 9.777777777777779e-06,
|
| 2008 |
+
"loss": 0.0738,
|
| 2009 |
+
"step": 2800
|
| 2010 |
+
},
|
| 2011 |
+
{
|
| 2012 |
+
"epoch": 0.88,
|
| 2013 |
+
"grad_norm": 0.0016119845677167177,
|
| 2014 |
+
"learning_rate": 9.733333333333334e-06,
|
| 2015 |
+
"loss": 0.1069,
|
| 2016 |
+
"step": 2810
|
| 2017 |
+
},
|
| 2018 |
+
{
|
| 2019 |
+
"epoch": 0.88,
|
| 2020 |
+
"grad_norm": 0.044685885310173035,
|
| 2021 |
+
"learning_rate": 9.688888888888889e-06,
|
| 2022 |
+
"loss": 0.0718,
|
| 2023 |
+
"step": 2820
|
| 2024 |
+
},
|
| 2025 |
+
{
|
| 2026 |
+
"epoch": 0.89,
|
| 2027 |
+
"grad_norm": 0.5644834041595459,
|
| 2028 |
+
"learning_rate": 9.644444444444444e-06,
|
| 2029 |
+
"loss": 0.0539,
|
| 2030 |
+
"step": 2830
|
| 2031 |
+
},
|
| 2032 |
+
{
|
| 2033 |
+
"epoch": 0.89,
|
| 2034 |
+
"grad_norm": 0.6698065400123596,
|
| 2035 |
+
"learning_rate": 9.600000000000001e-06,
|
| 2036 |
+
"loss": 0.0671,
|
| 2037 |
+
"step": 2840
|
| 2038 |
+
},
|
| 2039 |
+
{
|
| 2040 |
+
"epoch": 0.89,
|
| 2041 |
+
"grad_norm": 0.009338573552668095,
|
| 2042 |
+
"learning_rate": 9.555555555555556e-06,
|
| 2043 |
+
"loss": 0.0238,
|
| 2044 |
+
"step": 2850
|
| 2045 |
+
},
|
| 2046 |
+
{
|
| 2047 |
+
"epoch": 0.9,
|
| 2048 |
+
"grad_norm": 0.002303204732015729,
|
| 2049 |
+
"learning_rate": 9.511111111111112e-06,
|
| 2050 |
+
"loss": 0.0531,
|
| 2051 |
+
"step": 2860
|
| 2052 |
+
},
|
| 2053 |
+
{
|
| 2054 |
+
"epoch": 0.9,
|
| 2055 |
+
"grad_norm": 0.004544748924672604,
|
| 2056 |
+
"learning_rate": 9.466666666666667e-06,
|
| 2057 |
+
"loss": 0.133,
|
| 2058 |
+
"step": 2870
|
| 2059 |
+
},
|
| 2060 |
+
{
|
| 2061 |
+
"epoch": 0.9,
|
| 2062 |
+
"grad_norm": 0.004034551791846752,
|
| 2063 |
+
"learning_rate": 9.422222222222222e-06,
|
| 2064 |
+
"loss": 0.0174,
|
| 2065 |
+
"step": 2880
|
| 2066 |
+
},
|
| 2067 |
+
{
|
| 2068 |
+
"epoch": 0.91,
|
| 2069 |
+
"grad_norm": 0.0032248455099761486,
|
| 2070 |
+
"learning_rate": 9.377777777777779e-06,
|
| 2071 |
+
"loss": 0.0571,
|
| 2072 |
+
"step": 2890
|
| 2073 |
+
},
|
| 2074 |
+
{
|
| 2075 |
+
"epoch": 0.91,
|
| 2076 |
+
"grad_norm": 0.2451166957616806,
|
| 2077 |
+
"learning_rate": 9.333333333333334e-06,
|
| 2078 |
+
"loss": 0.0304,
|
| 2079 |
+
"step": 2900
|
| 2080 |
+
},
|
| 2081 |
+
{
|
| 2082 |
+
"epoch": 0.91,
|
| 2083 |
+
"grad_norm": 0.001918753026984632,
|
| 2084 |
+
"learning_rate": 9.28888888888889e-06,
|
| 2085 |
+
"loss": 0.0826,
|
| 2086 |
+
"step": 2910
|
| 2087 |
+
},
|
| 2088 |
+
{
|
| 2089 |
+
"epoch": 0.91,
|
| 2090 |
+
"grad_norm": 0.0014626473421230912,
|
| 2091 |
+
"learning_rate": 9.244444444444445e-06,
|
| 2092 |
+
"loss": 0.0288,
|
| 2093 |
+
"step": 2920
|
| 2094 |
+
},
|
| 2095 |
+
{
|
| 2096 |
+
"epoch": 0.92,
|
| 2097 |
+
"grad_norm": 0.003409222001209855,
|
| 2098 |
+
"learning_rate": 9.200000000000002e-06,
|
| 2099 |
+
"loss": 0.041,
|
| 2100 |
+
"step": 2930
|
| 2101 |
+
},
|
| 2102 |
+
{
|
| 2103 |
+
"epoch": 0.92,
|
| 2104 |
+
"grad_norm": 0.002315065125003457,
|
| 2105 |
+
"learning_rate": 9.155555555555557e-06,
|
| 2106 |
+
"loss": 0.0322,
|
| 2107 |
+
"step": 2940
|
| 2108 |
+
},
|
| 2109 |
+
{
|
| 2110 |
+
"epoch": 0.92,
|
| 2111 |
+
"grad_norm": 0.002485796343535185,
|
| 2112 |
+
"learning_rate": 9.111111111111112e-06,
|
| 2113 |
+
"loss": 0.0132,
|
| 2114 |
+
"step": 2950
|
| 2115 |
+
},
|
| 2116 |
+
{
|
| 2117 |
+
"epoch": 0.93,
|
| 2118 |
+
"grad_norm": 0.0027836193330585957,
|
| 2119 |
+
"learning_rate": 9.066666666666667e-06,
|
| 2120 |
+
"loss": 0.0424,
|
| 2121 |
+
"step": 2960
|
| 2122 |
+
},
|
| 2123 |
+
{
|
| 2124 |
+
"epoch": 0.93,
|
| 2125 |
+
"grad_norm": 1.2203580141067505,
|
| 2126 |
+
"learning_rate": 9.022222222222223e-06,
|
| 2127 |
+
"loss": 0.0607,
|
| 2128 |
+
"step": 2970
|
| 2129 |
+
},
|
| 2130 |
+
{
|
| 2131 |
+
"epoch": 0.93,
|
| 2132 |
+
"grad_norm": 0.0014286149526014924,
|
| 2133 |
+
"learning_rate": 8.977777777777778e-06,
|
| 2134 |
+
"loss": 0.0082,
|
| 2135 |
+
"step": 2980
|
| 2136 |
+
},
|
| 2137 |
+
{
|
| 2138 |
+
"epoch": 0.94,
|
| 2139 |
+
"grad_norm": 0.0011963992146775126,
|
| 2140 |
+
"learning_rate": 8.933333333333333e-06,
|
| 2141 |
+
"loss": 0.0447,
|
| 2142 |
+
"step": 2990
|
| 2143 |
+
},
|
| 2144 |
+
{
|
| 2145 |
+
"epoch": 0.94,
|
| 2146 |
+
"grad_norm": 0.0012861526338383555,
|
| 2147 |
+
"learning_rate": 8.888888888888888e-06,
|
| 2148 |
+
"loss": 0.0394,
|
| 2149 |
+
"step": 3000
|
| 2150 |
+
},
|
| 2151 |
+
{
|
| 2152 |
+
"epoch": 0.94,
|
| 2153 |
+
"eval_loss": 0.021464653313159943,
|
| 2154 |
+
"eval_runtime": 62.0041,
|
| 2155 |
+
"eval_samples_per_second": 16.128,
|
| 2156 |
+
"eval_steps_per_second": 16.128,
|
| 2157 |
+
"step": 3000
|
| 2158 |
+
},
|
| 2159 |
+
{
|
| 2160 |
+
"epoch": 0.94,
|
| 2161 |
+
"grad_norm": 2.6348252296447754,
|
| 2162 |
+
"learning_rate": 8.844444444444445e-06,
|
| 2163 |
+
"loss": 0.1294,
|
| 2164 |
+
"step": 3010
|
| 2165 |
+
},
|
| 2166 |
+
{
|
| 2167 |
+
"epoch": 0.95,
|
| 2168 |
+
"grad_norm": 0.0012073700781911612,
|
| 2169 |
+
"learning_rate": 8.8e-06,
|
| 2170 |
+
"loss": 0.0304,
|
| 2171 |
+
"step": 3020
|
| 2172 |
+
},
|
| 2173 |
+
{
|
| 2174 |
+
"epoch": 0.95,
|
| 2175 |
+
"grad_norm": 0.003195586148649454,
|
| 2176 |
+
"learning_rate": 8.755555555555556e-06,
|
| 2177 |
+
"loss": 0.0161,
|
| 2178 |
+
"step": 3030
|
| 2179 |
+
},
|
| 2180 |
+
{
|
| 2181 |
+
"epoch": 0.95,
|
| 2182 |
+
"grad_norm": 1.9521057605743408,
|
| 2183 |
+
"learning_rate": 8.711111111111111e-06,
|
| 2184 |
+
"loss": 0.1185,
|
| 2185 |
+
"step": 3040
|
| 2186 |
+
},
|
| 2187 |
+
{
|
| 2188 |
+
"epoch": 0.96,
|
| 2189 |
+
"grad_norm": 0.035520199686288834,
|
| 2190 |
+
"learning_rate": 8.666666666666668e-06,
|
| 2191 |
+
"loss": 0.0357,
|
| 2192 |
+
"step": 3050
|
| 2193 |
+
},
|
| 2194 |
+
{
|
| 2195 |
+
"epoch": 0.96,
|
| 2196 |
+
"grad_norm": 0.0031184712424874306,
|
| 2197 |
+
"learning_rate": 8.622222222222223e-06,
|
| 2198 |
+
"loss": 0.0505,
|
| 2199 |
+
"step": 3060
|
| 2200 |
+
},
|
| 2201 |
+
{
|
| 2202 |
+
"epoch": 0.96,
|
| 2203 |
+
"grad_norm": 1.8742607831954956,
|
| 2204 |
+
"learning_rate": 8.577777777777778e-06,
|
| 2205 |
+
"loss": 0.0607,
|
| 2206 |
+
"step": 3070
|
| 2207 |
+
},
|
| 2208 |
+
{
|
| 2209 |
+
"epoch": 0.96,
|
| 2210 |
+
"grad_norm": 0.001605386845767498,
|
| 2211 |
+
"learning_rate": 8.533333333333335e-06,
|
| 2212 |
+
"loss": 0.048,
|
| 2213 |
+
"step": 3080
|
| 2214 |
+
},
|
| 2215 |
+
{
|
| 2216 |
+
"epoch": 0.97,
|
| 2217 |
+
"grad_norm": 2.8198726177215576,
|
| 2218 |
+
"learning_rate": 8.48888888888889e-06,
|
| 2219 |
+
"loss": 0.029,
|
| 2220 |
+
"step": 3090
|
| 2221 |
+
},
|
| 2222 |
+
{
|
| 2223 |
+
"epoch": 0.97,
|
| 2224 |
+
"grad_norm": 0.0011747152311727405,
|
| 2225 |
+
"learning_rate": 8.444444444444446e-06,
|
| 2226 |
+
"loss": 0.0745,
|
| 2227 |
+
"step": 3100
|
| 2228 |
+
},
|
| 2229 |
+
{
|
| 2230 |
+
"epoch": 0.97,
|
| 2231 |
+
"grad_norm": 0.0037345418240875006,
|
| 2232 |
+
"learning_rate": 8.400000000000001e-06,
|
| 2233 |
+
"loss": 0.1246,
|
| 2234 |
+
"step": 3110
|
| 2235 |
+
},
|
| 2236 |
+
{
|
| 2237 |
+
"epoch": 0.98,
|
| 2238 |
+
"grad_norm": 0.03968101739883423,
|
| 2239 |
+
"learning_rate": 8.355555555555556e-06,
|
| 2240 |
+
"loss": 0.028,
|
| 2241 |
+
"step": 3120
|
| 2242 |
+
},
|
| 2243 |
+
{
|
| 2244 |
+
"epoch": 0.98,
|
| 2245 |
+
"grad_norm": 0.0017666455823928118,
|
| 2246 |
+
"learning_rate": 8.311111111111111e-06,
|
| 2247 |
+
"loss": 0.0398,
|
| 2248 |
+
"step": 3130
|
| 2249 |
+
},
|
| 2250 |
+
{
|
| 2251 |
+
"epoch": 0.98,
|
| 2252 |
+
"grad_norm": 0.0016725193709135056,
|
| 2253 |
+
"learning_rate": 8.266666666666667e-06,
|
| 2254 |
+
"loss": 0.0874,
|
| 2255 |
+
"step": 3140
|
| 2256 |
+
},
|
| 2257 |
+
{
|
| 2258 |
+
"epoch": 0.99,
|
| 2259 |
+
"grad_norm": 3.0152335166931152,
|
| 2260 |
+
"learning_rate": 8.222222222222222e-06,
|
| 2261 |
+
"loss": 0.0233,
|
| 2262 |
+
"step": 3150
|
| 2263 |
+
},
|
| 2264 |
+
{
|
| 2265 |
+
"epoch": 0.99,
|
| 2266 |
+
"grad_norm": 0.003967667929828167,
|
| 2267 |
+
"learning_rate": 8.177777777777779e-06,
|
| 2268 |
+
"loss": 0.0322,
|
| 2269 |
+
"step": 3160
|
| 2270 |
+
},
|
| 2271 |
+
{
|
| 2272 |
+
"epoch": 0.99,
|
| 2273 |
+
"grad_norm": 0.0012324820272624493,
|
| 2274 |
+
"learning_rate": 8.133333333333334e-06,
|
| 2275 |
+
"loss": 0.01,
|
| 2276 |
+
"step": 3170
|
| 2277 |
+
},
|
| 2278 |
+
{
|
| 2279 |
+
"epoch": 1.0,
|
| 2280 |
+
"grad_norm": 0.001890141749754548,
|
| 2281 |
+
"learning_rate": 8.08888888888889e-06,
|
| 2282 |
+
"loss": 0.0617,
|
| 2283 |
+
"step": 3180
|
| 2284 |
+
},
|
| 2285 |
+
{
|
| 2286 |
+
"epoch": 1.0,
|
| 2287 |
+
"grad_norm": 0.0017083596903830767,
|
| 2288 |
+
"learning_rate": 8.044444444444444e-06,
|
| 2289 |
+
"loss": 0.031,
|
| 2290 |
+
"step": 3190
|
| 2291 |
+
},
|
| 2292 |
+
{
|
| 2293 |
+
"epoch": 1.0,
|
| 2294 |
+
"grad_norm": 1.1974055767059326,
|
| 2295 |
+
"learning_rate": 8.000000000000001e-06,
|
| 2296 |
+
"loss": 0.0136,
|
| 2297 |
+
"step": 3200
|
| 2298 |
+
},
|
| 2299 |
+
{
|
| 2300 |
+
"epoch": 1.01,
|
| 2301 |
+
"grad_norm": 1.8210293054580688,
|
| 2302 |
+
"learning_rate": 7.955555555555557e-06,
|
| 2303 |
+
"loss": 0.0174,
|
| 2304 |
+
"step": 3210
|
| 2305 |
+
},
|
| 2306 |
+
{
|
| 2307 |
+
"epoch": 1.01,
|
| 2308 |
+
"grad_norm": 3.1489944458007812,
|
| 2309 |
+
"learning_rate": 7.911111111111112e-06,
|
| 2310 |
+
"loss": 0.0776,
|
| 2311 |
+
"step": 3220
|
| 2312 |
+
},
|
| 2313 |
+
{
|
| 2314 |
+
"epoch": 1.01,
|
| 2315 |
+
"grad_norm": 0.0014464000705629587,
|
| 2316 |
+
"learning_rate": 7.866666666666667e-06,
|
| 2317 |
+
"loss": 0.0164,
|
| 2318 |
+
"step": 3230
|
| 2319 |
+
},
|
| 2320 |
+
{
|
| 2321 |
+
"epoch": 1.02,
|
| 2322 |
+
"grad_norm": 0.5747243762016296,
|
| 2323 |
+
"learning_rate": 7.822222222222224e-06,
|
| 2324 |
+
"loss": 0.0093,
|
| 2325 |
+
"step": 3240
|
| 2326 |
+
},
|
| 2327 |
+
{
|
| 2328 |
+
"epoch": 1.02,
|
| 2329 |
+
"grad_norm": 0.009251836687326431,
|
| 2330 |
+
"learning_rate": 7.77777777777778e-06,
|
| 2331 |
+
"loss": 0.0171,
|
| 2332 |
+
"step": 3250
|
| 2333 |
+
},
|
| 2334 |
+
{
|
| 2335 |
+
"epoch": 1.02,
|
| 2336 |
+
"grad_norm": 0.2069794237613678,
|
| 2337 |
+
"learning_rate": 7.733333333333334e-06,
|
| 2338 |
+
"loss": 0.0301,
|
| 2339 |
+
"step": 3260
|
| 2340 |
+
},
|
| 2341 |
+
{
|
| 2342 |
+
"epoch": 1.02,
|
| 2343 |
+
"grad_norm": 2.251051664352417,
|
| 2344 |
+
"learning_rate": 7.68888888888889e-06,
|
| 2345 |
+
"loss": 0.0707,
|
| 2346 |
+
"step": 3270
|
| 2347 |
+
},
|
| 2348 |
+
{
|
| 2349 |
+
"epoch": 1.03,
|
| 2350 |
+
"grad_norm": 0.0010599648812785745,
|
| 2351 |
+
"learning_rate": 7.644444444444445e-06,
|
| 2352 |
+
"loss": 0.0053,
|
| 2353 |
+
"step": 3280
|
| 2354 |
+
},
|
| 2355 |
+
{
|
| 2356 |
+
"epoch": 1.03,
|
| 2357 |
+
"grad_norm": 0.0016899543115869164,
|
| 2358 |
+
"learning_rate": 7.600000000000001e-06,
|
| 2359 |
+
"loss": 0.0006,
|
| 2360 |
+
"step": 3290
|
| 2361 |
+
},
|
| 2362 |
+
{
|
| 2363 |
+
"epoch": 1.03,
|
| 2364 |
+
"grad_norm": 0.003940447699278593,
|
| 2365 |
+
"learning_rate": 7.555555555555556e-06,
|
| 2366 |
+
"loss": 0.014,
|
| 2367 |
+
"step": 3300
|
| 2368 |
+
},
|
| 2369 |
+
{
|
| 2370 |
+
"epoch": 1.04,
|
| 2371 |
+
"grad_norm": 0.7399486899375916,
|
| 2372 |
+
"learning_rate": 7.511111111111111e-06,
|
| 2373 |
+
"loss": 0.0309,
|
| 2374 |
+
"step": 3310
|
| 2375 |
+
},
|
| 2376 |
+
{
|
| 2377 |
+
"epoch": 1.04,
|
| 2378 |
+
"grad_norm": 0.0006520531023852527,
|
| 2379 |
+
"learning_rate": 7.4666666666666675e-06,
|
| 2380 |
+
"loss": 0.0218,
|
| 2381 |
+
"step": 3320
|
| 2382 |
+
},
|
| 2383 |
+
{
|
| 2384 |
+
"epoch": 1.04,
|
| 2385 |
+
"grad_norm": 2.1491100788116455,
|
| 2386 |
+
"learning_rate": 7.422222222222223e-06,
|
| 2387 |
+
"loss": 0.1215,
|
| 2388 |
+
"step": 3330
|
| 2389 |
+
},
|
| 2390 |
+
{
|
| 2391 |
+
"epoch": 1.05,
|
| 2392 |
+
"grad_norm": 3.6713123321533203,
|
| 2393 |
+
"learning_rate": 7.377777777777778e-06,
|
| 2394 |
+
"loss": 0.0429,
|
| 2395 |
+
"step": 3340
|
| 2396 |
+
},
|
| 2397 |
+
{
|
| 2398 |
+
"epoch": 1.05,
|
| 2399 |
+
"grad_norm": 0.03133914992213249,
|
| 2400 |
+
"learning_rate": 7.333333333333333e-06,
|
| 2401 |
+
"loss": 0.027,
|
| 2402 |
+
"step": 3350
|
| 2403 |
+
},
|
| 2404 |
+
{
|
| 2405 |
+
"epoch": 1.05,
|
| 2406 |
+
"grad_norm": 0.002692360896617174,
|
| 2407 |
+
"learning_rate": 7.28888888888889e-06,
|
| 2408 |
+
"loss": 0.0216,
|
| 2409 |
+
"step": 3360
|
| 2410 |
+
},
|
| 2411 |
+
{
|
| 2412 |
+
"epoch": 1.06,
|
| 2413 |
+
"grad_norm": 0.0034821536391973495,
|
| 2414 |
+
"learning_rate": 7.244444444444445e-06,
|
| 2415 |
+
"loss": 0.043,
|
| 2416 |
+
"step": 3370
|
| 2417 |
+
},
|
| 2418 |
+
{
|
| 2419 |
+
"epoch": 1.06,
|
| 2420 |
+
"grad_norm": 0.00455117505043745,
|
| 2421 |
+
"learning_rate": 7.2000000000000005e-06,
|
| 2422 |
+
"loss": 0.0688,
|
| 2423 |
+
"step": 3380
|
| 2424 |
+
},
|
| 2425 |
+
{
|
| 2426 |
+
"epoch": 1.06,
|
| 2427 |
+
"grad_norm": 3.5100715160369873,
|
| 2428 |
+
"learning_rate": 7.155555555555556e-06,
|
| 2429 |
+
"loss": 0.0136,
|
| 2430 |
+
"step": 3390
|
| 2431 |
+
},
|
| 2432 |
+
{
|
| 2433 |
+
"epoch": 1.07,
|
| 2434 |
+
"grad_norm": 0.0031513776630163193,
|
| 2435 |
+
"learning_rate": 7.111111111111112e-06,
|
| 2436 |
+
"loss": 0.0155,
|
| 2437 |
+
"step": 3400
|
| 2438 |
+
},
|
| 2439 |
+
{
|
| 2440 |
+
"epoch": 1.07,
|
| 2441 |
+
"grad_norm": 1.3787556886672974,
|
| 2442 |
+
"learning_rate": 7.066666666666667e-06,
|
| 2443 |
+
"loss": 0.0254,
|
| 2444 |
+
"step": 3410
|
| 2445 |
+
},
|
| 2446 |
+
{
|
| 2447 |
+
"epoch": 1.07,
|
| 2448 |
+
"grad_norm": 0.000773483538068831,
|
| 2449 |
+
"learning_rate": 7.022222222222222e-06,
|
| 2450 |
+
"loss": 0.0031,
|
| 2451 |
+
"step": 3420
|
| 2452 |
+
},
|
| 2453 |
+
{
|
| 2454 |
+
"epoch": 1.07,
|
| 2455 |
+
"grad_norm": 0.0014327614335343242,
|
| 2456 |
+
"learning_rate": 6.977777777777779e-06,
|
| 2457 |
+
"loss": 0.0376,
|
| 2458 |
+
"step": 3430
|
| 2459 |
+
},
|
| 2460 |
+
{
|
| 2461 |
+
"epoch": 1.08,
|
| 2462 |
+
"grad_norm": 0.0031394639518111944,
|
| 2463 |
+
"learning_rate": 6.9333333333333344e-06,
|
| 2464 |
+
"loss": 0.0596,
|
| 2465 |
+
"step": 3440
|
| 2466 |
+
},
|
| 2467 |
+
{
|
| 2468 |
+
"epoch": 1.08,
|
| 2469 |
+
"grad_norm": 0.0024705962277948856,
|
| 2470 |
+
"learning_rate": 6.88888888888889e-06,
|
| 2471 |
+
"loss": 0.0095,
|
| 2472 |
+
"step": 3450
|
| 2473 |
+
},
|
| 2474 |
+
{
|
| 2475 |
+
"epoch": 1.08,
|
| 2476 |
+
"grad_norm": 0.0005248151137493551,
|
| 2477 |
+
"learning_rate": 6.844444444444445e-06,
|
| 2478 |
+
"loss": 0.0199,
|
| 2479 |
+
"step": 3460
|
| 2480 |
+
},
|
| 2481 |
+
{
|
| 2482 |
+
"epoch": 1.09,
|
| 2483 |
+
"grad_norm": 0.49637269973754883,
|
| 2484 |
+
"learning_rate": 6.800000000000001e-06,
|
| 2485 |
+
"loss": 0.0154,
|
| 2486 |
+
"step": 3470
|
| 2487 |
+
},
|
| 2488 |
+
{
|
| 2489 |
+
"epoch": 1.09,
|
| 2490 |
+
"grad_norm": 1.3430148363113403,
|
| 2491 |
+
"learning_rate": 6.755555555555556e-06,
|
| 2492 |
+
"loss": 0.0395,
|
| 2493 |
+
"step": 3480
|
| 2494 |
+
},
|
| 2495 |
+
{
|
| 2496 |
+
"epoch": 1.09,
|
| 2497 |
+
"grad_norm": 1.885108232498169,
|
| 2498 |
+
"learning_rate": 6.711111111111111e-06,
|
| 2499 |
+
"loss": 0.0472,
|
| 2500 |
+
"step": 3490
|
| 2501 |
+
},
|
| 2502 |
+
{
|
| 2503 |
+
"epoch": 1.1,
|
| 2504 |
+
"grad_norm": 0.0007279856945388019,
|
| 2505 |
+
"learning_rate": 6.666666666666667e-06,
|
| 2506 |
+
"loss": 0.0369,
|
| 2507 |
+
"step": 3500
|
| 2508 |
+
},
|
| 2509 |
+
{
|
| 2510 |
+
"epoch": 1.1,
|
| 2511 |
+
"eval_loss": 0.022176509723067284,
|
| 2512 |
+
"eval_runtime": 61.8847,
|
| 2513 |
+
"eval_samples_per_second": 16.159,
|
| 2514 |
+
"eval_steps_per_second": 16.159,
|
| 2515 |
+
"step": 3500
|
| 2516 |
+
},
|
| 2517 |
+
{
|
| 2518 |
+
"epoch": 1.1,
|
| 2519 |
+
"grad_norm": 0.003986152354627848,
|
| 2520 |
+
"learning_rate": 6.6222222222222236e-06,
|
| 2521 |
+
"loss": 0.0583,
|
| 2522 |
+
"step": 3510
|
| 2523 |
+
},
|
| 2524 |
+
{
|
| 2525 |
+
"epoch": 1.1,
|
| 2526 |
+
"grad_norm": 0.0014766375534236431,
|
| 2527 |
+
"learning_rate": 6.577777777777779e-06,
|
| 2528 |
+
"loss": 0.0153,
|
| 2529 |
+
"step": 3520
|
| 2530 |
+
},
|
| 2531 |
+
{
|
| 2532 |
+
"epoch": 1.11,
|
| 2533 |
+
"grad_norm": 0.791298508644104,
|
| 2534 |
+
"learning_rate": 6.533333333333334e-06,
|
| 2535 |
+
"loss": 0.0508,
|
| 2536 |
+
"step": 3530
|
| 2537 |
+
},
|
| 2538 |
+
{
|
| 2539 |
+
"epoch": 1.11,
|
| 2540 |
+
"grad_norm": 0.02302105911076069,
|
| 2541 |
+
"learning_rate": 6.488888888888889e-06,
|
| 2542 |
+
"loss": 0.0275,
|
| 2543 |
+
"step": 3540
|
| 2544 |
+
},
|
| 2545 |
+
{
|
| 2546 |
+
"epoch": 1.11,
|
| 2547 |
+
"grad_norm": 0.00240346509963274,
|
| 2548 |
+
"learning_rate": 6.444444444444445e-06,
|
| 2549 |
+
"loss": 0.0062,
|
| 2550 |
+
"step": 3550
|
| 2551 |
+
},
|
| 2552 |
+
{
|
| 2553 |
+
"epoch": 1.12,
|
| 2554 |
+
"grad_norm": 0.0011688686208799481,
|
| 2555 |
+
"learning_rate": 6.4000000000000006e-06,
|
| 2556 |
+
"loss": 0.0458,
|
| 2557 |
+
"step": 3560
|
| 2558 |
+
},
|
| 2559 |
+
{
|
| 2560 |
+
"epoch": 1.12,
|
| 2561 |
+
"grad_norm": 0.0007077022455632687,
|
| 2562 |
+
"learning_rate": 6.355555555555556e-06,
|
| 2563 |
+
"loss": 0.0274,
|
| 2564 |
+
"step": 3570
|
| 2565 |
+
},
|
| 2566 |
+
{
|
| 2567 |
+
"epoch": 1.12,
|
| 2568 |
+
"grad_norm": 0.0014391880249604583,
|
| 2569 |
+
"learning_rate": 6.311111111111111e-06,
|
| 2570 |
+
"loss": 0.0178,
|
| 2571 |
+
"step": 3580
|
| 2572 |
+
},
|
| 2573 |
+
{
|
| 2574 |
+
"epoch": 1.12,
|
| 2575 |
+
"grad_norm": 0.002971325535327196,
|
| 2576 |
+
"learning_rate": 6.266666666666668e-06,
|
| 2577 |
+
"loss": 0.0099,
|
| 2578 |
+
"step": 3590
|
| 2579 |
+
},
|
| 2580 |
+
{
|
| 2581 |
+
"epoch": 1.13,
|
| 2582 |
+
"grad_norm": 1.5885258913040161,
|
| 2583 |
+
"learning_rate": 6.222222222222223e-06,
|
| 2584 |
+
"loss": 0.0277,
|
| 2585 |
+
"step": 3600
|
| 2586 |
+
},
|
| 2587 |
+
{
|
| 2588 |
+
"epoch": 1.13,
|
| 2589 |
+
"grad_norm": 0.0023686839267611504,
|
| 2590 |
+
"learning_rate": 6.177777777777778e-06,
|
| 2591 |
+
"loss": 0.0001,
|
| 2592 |
+
"step": 3610
|
| 2593 |
+
},
|
| 2594 |
+
{
|
| 2595 |
+
"epoch": 1.13,
|
| 2596 |
+
"grad_norm": 0.0018330584280192852,
|
| 2597 |
+
"learning_rate": 6.133333333333334e-06,
|
| 2598 |
+
"loss": 0.034,
|
| 2599 |
+
"step": 3620
|
| 2600 |
+
},
|
| 2601 |
+
{
|
| 2602 |
+
"epoch": 1.14,
|
| 2603 |
+
"grad_norm": 0.5174291133880615,
|
| 2604 |
+
"learning_rate": 6.08888888888889e-06,
|
| 2605 |
+
"loss": 0.0095,
|
| 2606 |
+
"step": 3630
|
| 2607 |
+
},
|
| 2608 |
+
{
|
| 2609 |
+
"epoch": 1.14,
|
| 2610 |
+
"grad_norm": 0.6314389705657959,
|
| 2611 |
+
"learning_rate": 6.044444444444445e-06,
|
| 2612 |
+
"loss": 0.035,
|
| 2613 |
+
"step": 3640
|
| 2614 |
+
},
|
| 2615 |
+
{
|
| 2616 |
+
"epoch": 1.14,
|
| 2617 |
+
"grad_norm": 4.273723125457764,
|
| 2618 |
+
"learning_rate": 6e-06,
|
| 2619 |
+
"loss": 0.0332,
|
| 2620 |
+
"step": 3650
|
| 2621 |
+
},
|
| 2622 |
+
{
|
| 2623 |
+
"epoch": 1.15,
|
| 2624 |
+
"grad_norm": 1.590868592262268,
|
| 2625 |
+
"learning_rate": 5.955555555555555e-06,
|
| 2626 |
+
"loss": 0.0398,
|
| 2627 |
+
"step": 3660
|
| 2628 |
+
},
|
| 2629 |
+
{
|
| 2630 |
+
"epoch": 1.15,
|
| 2631 |
+
"grad_norm": 1.3291077613830566,
|
| 2632 |
+
"learning_rate": 5.911111111111112e-06,
|
| 2633 |
+
"loss": 0.0306,
|
| 2634 |
+
"step": 3670
|
| 2635 |
+
},
|
| 2636 |
+
{
|
| 2637 |
+
"epoch": 1.15,
|
| 2638 |
+
"grad_norm": 0.0011222765315324068,
|
| 2639 |
+
"learning_rate": 5.8666666666666675e-06,
|
| 2640 |
+
"loss": 0.0194,
|
| 2641 |
+
"step": 3680
|
| 2642 |
+
},
|
| 2643 |
+
{
|
| 2644 |
+
"epoch": 1.16,
|
| 2645 |
+
"grad_norm": 1.7731515169143677,
|
| 2646 |
+
"learning_rate": 5.822222222222223e-06,
|
| 2647 |
+
"loss": 0.0747,
|
| 2648 |
+
"step": 3690
|
| 2649 |
+
},
|
| 2650 |
+
{
|
| 2651 |
+
"epoch": 1.16,
|
| 2652 |
+
"grad_norm": 3.05898380279541,
|
| 2653 |
+
"learning_rate": 5.777777777777778e-06,
|
| 2654 |
+
"loss": 0.0239,
|
| 2655 |
+
"step": 3700
|
| 2656 |
+
},
|
| 2657 |
+
{
|
| 2658 |
+
"epoch": 1.16,
|
| 2659 |
+
"grad_norm": 0.0017392480513080955,
|
| 2660 |
+
"learning_rate": 5.733333333333334e-06,
|
| 2661 |
+
"loss": 0.0415,
|
| 2662 |
+
"step": 3710
|
| 2663 |
+
},
|
| 2664 |
+
{
|
| 2665 |
+
"epoch": 1.17,
|
| 2666 |
+
"grad_norm": 0.03126866742968559,
|
| 2667 |
+
"learning_rate": 5.688888888888889e-06,
|
| 2668 |
+
"loss": 0.0814,
|
| 2669 |
+
"step": 3720
|
| 2670 |
+
},
|
| 2671 |
+
{
|
| 2672 |
+
"epoch": 1.17,
|
| 2673 |
+
"grad_norm": 0.0008881581015884876,
|
| 2674 |
+
"learning_rate": 5.6444444444444445e-06,
|
| 2675 |
+
"loss": 0.0211,
|
| 2676 |
+
"step": 3730
|
| 2677 |
+
},
|
| 2678 |
+
{
|
| 2679 |
+
"epoch": 1.17,
|
| 2680 |
+
"grad_norm": 0.0007252222276292741,
|
| 2681 |
+
"learning_rate": 5.600000000000001e-06,
|
| 2682 |
+
"loss": 0.0441,
|
| 2683 |
+
"step": 3740
|
| 2684 |
+
},
|
| 2685 |
+
{
|
| 2686 |
+
"epoch": 1.17,
|
| 2687 |
+
"grad_norm": 0.002276132581755519,
|
| 2688 |
+
"learning_rate": 5.555555555555557e-06,
|
| 2689 |
+
"loss": 0.0433,
|
| 2690 |
+
"step": 3750
|
| 2691 |
+
},
|
| 2692 |
+
{
|
| 2693 |
+
"epoch": 1.18,
|
| 2694 |
+
"grad_norm": 0.0014605377800762653,
|
| 2695 |
+
"learning_rate": 5.511111111111112e-06,
|
| 2696 |
+
"loss": 0.0204,
|
| 2697 |
+
"step": 3760
|
| 2698 |
+
},
|
| 2699 |
+
{
|
| 2700 |
+
"epoch": 1.18,
|
| 2701 |
+
"grad_norm": 0.0017992451321333647,
|
| 2702 |
+
"learning_rate": 5.466666666666667e-06,
|
| 2703 |
+
"loss": 0.0992,
|
| 2704 |
+
"step": 3770
|
| 2705 |
+
},
|
| 2706 |
+
{
|
| 2707 |
+
"epoch": 1.18,
|
| 2708 |
+
"grad_norm": 0.0018204891821369529,
|
| 2709 |
+
"learning_rate": 5.422222222222223e-06,
|
| 2710 |
+
"loss": 0.0132,
|
| 2711 |
+
"step": 3780
|
| 2712 |
+
},
|
| 2713 |
+
{
|
| 2714 |
+
"epoch": 1.19,
|
| 2715 |
+
"grad_norm": 0.16287018358707428,
|
| 2716 |
+
"learning_rate": 5.3777777777777784e-06,
|
| 2717 |
+
"loss": 0.0291,
|
| 2718 |
+
"step": 3790
|
| 2719 |
+
},
|
| 2720 |
+
{
|
| 2721 |
+
"epoch": 1.19,
|
| 2722 |
+
"grad_norm": 0.21604575216770172,
|
| 2723 |
+
"learning_rate": 5.333333333333334e-06,
|
| 2724 |
+
"loss": 0.0158,
|
| 2725 |
+
"step": 3800
|
| 2726 |
+
},
|
| 2727 |
+
{
|
| 2728 |
+
"epoch": 1.19,
|
| 2729 |
+
"grad_norm": 0.0013267812319099903,
|
| 2730 |
+
"learning_rate": 5.288888888888889e-06,
|
| 2731 |
+
"loss": 0.0053,
|
| 2732 |
+
"step": 3810
|
| 2733 |
+
},
|
| 2734 |
+
{
|
| 2735 |
+
"epoch": 1.2,
|
| 2736 |
+
"grad_norm": 0.0006841329741291702,
|
| 2737 |
+
"learning_rate": 5.244444444444445e-06,
|
| 2738 |
+
"loss": 0.0212,
|
| 2739 |
+
"step": 3820
|
| 2740 |
+
},
|
| 2741 |
+
{
|
| 2742 |
+
"epoch": 1.2,
|
| 2743 |
+
"grad_norm": 0.19196736812591553,
|
| 2744 |
+
"learning_rate": 5.2e-06,
|
| 2745 |
+
"loss": 0.0206,
|
| 2746 |
+
"step": 3830
|
| 2747 |
+
},
|
| 2748 |
+
{
|
| 2749 |
+
"epoch": 1.2,
|
| 2750 |
+
"grad_norm": 0.001533058239147067,
|
| 2751 |
+
"learning_rate": 5.155555555555556e-06,
|
| 2752 |
+
"loss": 0.0338,
|
| 2753 |
+
"step": 3840
|
| 2754 |
+
},
|
| 2755 |
+
{
|
| 2756 |
+
"epoch": 1.21,
|
| 2757 |
+
"grad_norm": 0.001141805900260806,
|
| 2758 |
+
"learning_rate": 5.1111111111111115e-06,
|
| 2759 |
+
"loss": 0.0006,
|
| 2760 |
+
"step": 3850
|
| 2761 |
+
},
|
| 2762 |
+
{
|
| 2763 |
+
"epoch": 1.21,
|
| 2764 |
+
"grad_norm": 1.453786849975586,
|
| 2765 |
+
"learning_rate": 5.0666666666666676e-06,
|
| 2766 |
+
"loss": 0.0449,
|
| 2767 |
+
"step": 3860
|
| 2768 |
+
},
|
| 2769 |
+
{
|
| 2770 |
+
"epoch": 1.21,
|
| 2771 |
+
"grad_norm": 0.0016863587079569697,
|
| 2772 |
+
"learning_rate": 5.022222222222223e-06,
|
| 2773 |
+
"loss": 0.0379,
|
| 2774 |
+
"step": 3870
|
| 2775 |
+
},
|
| 2776 |
+
{
|
| 2777 |
+
"epoch": 1.22,
|
| 2778 |
+
"grad_norm": 0.0005151178920641541,
|
| 2779 |
+
"learning_rate": 4.977777777777778e-06,
|
| 2780 |
+
"loss": 0.0318,
|
| 2781 |
+
"step": 3880
|
| 2782 |
+
},
|
| 2783 |
+
{
|
| 2784 |
+
"epoch": 1.22,
|
| 2785 |
+
"grad_norm": 2.0469911098480225,
|
| 2786 |
+
"learning_rate": 4.933333333333334e-06,
|
| 2787 |
+
"loss": 0.0185,
|
| 2788 |
+
"step": 3890
|
| 2789 |
+
},
|
| 2790 |
+
{
|
| 2791 |
+
"epoch": 1.22,
|
| 2792 |
+
"grad_norm": 0.7001366019248962,
|
| 2793 |
+
"learning_rate": 4.888888888888889e-06,
|
| 2794 |
+
"loss": 0.0566,
|
| 2795 |
+
"step": 3900
|
| 2796 |
+
},
|
| 2797 |
+
{
|
| 2798 |
+
"epoch": 1.23,
|
| 2799 |
+
"grad_norm": 0.0007412733975797892,
|
| 2800 |
+
"learning_rate": 4.8444444444444446e-06,
|
| 2801 |
+
"loss": 0.0003,
|
| 2802 |
+
"step": 3910
|
| 2803 |
+
},
|
| 2804 |
+
{
|
| 2805 |
+
"epoch": 1.23,
|
| 2806 |
+
"grad_norm": 0.0008465189021080732,
|
| 2807 |
+
"learning_rate": 4.800000000000001e-06,
|
| 2808 |
+
"loss": 0.0493,
|
| 2809 |
+
"step": 3920
|
| 2810 |
+
},
|
| 2811 |
+
{
|
| 2812 |
+
"epoch": 1.23,
|
| 2813 |
+
"grad_norm": 2.3275091648101807,
|
| 2814 |
+
"learning_rate": 4.755555555555556e-06,
|
| 2815 |
+
"loss": 0.0109,
|
| 2816 |
+
"step": 3930
|
| 2817 |
+
},
|
| 2818 |
+
{
|
| 2819 |
+
"epoch": 1.23,
|
| 2820 |
+
"grad_norm": 0.0014587095938622952,
|
| 2821 |
+
"learning_rate": 4.711111111111111e-06,
|
| 2822 |
+
"loss": 0.0,
|
| 2823 |
+
"step": 3940
|
| 2824 |
+
},
|
| 2825 |
+
{
|
| 2826 |
+
"epoch": 1.24,
|
| 2827 |
+
"grad_norm": 0.0013576536439359188,
|
| 2828 |
+
"learning_rate": 4.666666666666667e-06,
|
| 2829 |
+
"loss": 0.0255,
|
| 2830 |
+
"step": 3950
|
| 2831 |
+
},
|
| 2832 |
+
{
|
| 2833 |
+
"epoch": 1.24,
|
| 2834 |
+
"grad_norm": 0.008543136529624462,
|
| 2835 |
+
"learning_rate": 4.622222222222222e-06,
|
| 2836 |
+
"loss": 0.0352,
|
| 2837 |
+
"step": 3960
|
| 2838 |
+
},
|
| 2839 |
+
{
|
| 2840 |
+
"epoch": 1.24,
|
| 2841 |
+
"grad_norm": 0.0009798618266358972,
|
| 2842 |
+
"learning_rate": 4.5777777777777785e-06,
|
| 2843 |
+
"loss": 0.0556,
|
| 2844 |
+
"step": 3970
|
| 2845 |
+
},
|
| 2846 |
+
{
|
| 2847 |
+
"epoch": 1.25,
|
| 2848 |
+
"grad_norm": 0.0005635471316054463,
|
| 2849 |
+
"learning_rate": 4.533333333333334e-06,
|
| 2850 |
+
"loss": 0.0325,
|
| 2851 |
+
"step": 3980
|
| 2852 |
+
},
|
| 2853 |
+
{
|
| 2854 |
+
"epoch": 1.25,
|
| 2855 |
+
"grad_norm": 0.8276817202568054,
|
| 2856 |
+
"learning_rate": 4.488888888888889e-06,
|
| 2857 |
+
"loss": 0.0129,
|
| 2858 |
+
"step": 3990
|
| 2859 |
+
},
|
| 2860 |
+
{
|
| 2861 |
+
"epoch": 1.25,
|
| 2862 |
+
"grad_norm": 0.013259605504572392,
|
| 2863 |
+
"learning_rate": 4.444444444444444e-06,
|
| 2864 |
+
"loss": 0.0193,
|
| 2865 |
+
"step": 4000
|
| 2866 |
+
},
|
| 2867 |
+
{
|
| 2868 |
+
"epoch": 1.25,
|
| 2869 |
+
"eval_loss": 0.022383807227015495,
|
| 2870 |
+
"eval_runtime": 61.879,
|
| 2871 |
+
"eval_samples_per_second": 16.161,
|
| 2872 |
+
"eval_steps_per_second": 16.161,
|
| 2873 |
+
"step": 4000
|
| 2874 |
+
},
|
| 2875 |
+
{
|
| 2876 |
+
"epoch": 1.26,
|
| 2877 |
+
"grad_norm": 0.0017350054113194346,
|
| 2878 |
+
"learning_rate": 4.4e-06,
|
| 2879 |
+
"loss": 0.0464,
|
| 2880 |
+
"step": 4010
|
| 2881 |
+
},
|
| 2882 |
+
{
|
| 2883 |
+
"epoch": 1.26,
|
| 2884 |
+
"grad_norm": 0.8484432697296143,
|
| 2885 |
+
"learning_rate": 4.3555555555555555e-06,
|
| 2886 |
+
"loss": 0.0215,
|
| 2887 |
+
"step": 4020
|
| 2888 |
+
},
|
| 2889 |
+
{
|
| 2890 |
+
"epoch": 1.26,
|
| 2891 |
+
"grad_norm": 8.376138687133789,
|
| 2892 |
+
"learning_rate": 4.3111111111111115e-06,
|
| 2893 |
+
"loss": 0.0781,
|
| 2894 |
+
"step": 4030
|
| 2895 |
+
},
|
| 2896 |
+
{
|
| 2897 |
+
"epoch": 1.27,
|
| 2898 |
+
"grad_norm": 0.0006440070574171841,
|
| 2899 |
+
"learning_rate": 4.266666666666668e-06,
|
| 2900 |
+
"loss": 0.01,
|
| 2901 |
+
"step": 4040
|
| 2902 |
+
},
|
| 2903 |
+
{
|
| 2904 |
+
"epoch": 1.27,
|
| 2905 |
+
"grad_norm": 0.001261880504898727,
|
| 2906 |
+
"learning_rate": 4.222222222222223e-06,
|
| 2907 |
+
"loss": 0.1061,
|
| 2908 |
+
"step": 4050
|
| 2909 |
+
},
|
| 2910 |
+
{
|
| 2911 |
+
"epoch": 1.27,
|
| 2912 |
+
"grad_norm": 0.9878569841384888,
|
| 2913 |
+
"learning_rate": 4.177777777777778e-06,
|
| 2914 |
+
"loss": 0.0254,
|
| 2915 |
+
"step": 4060
|
| 2916 |
+
},
|
| 2917 |
+
{
|
| 2918 |
+
"epoch": 1.28,
|
| 2919 |
+
"grad_norm": 0.05665569379925728,
|
| 2920 |
+
"learning_rate": 4.133333333333333e-06,
|
| 2921 |
+
"loss": 0.0044,
|
| 2922 |
+
"step": 4070
|
| 2923 |
+
},
|
| 2924 |
+
{
|
| 2925 |
+
"epoch": 1.28,
|
| 2926 |
+
"grad_norm": 0.6482369303703308,
|
| 2927 |
+
"learning_rate": 4.088888888888889e-06,
|
| 2928 |
+
"loss": 0.0259,
|
| 2929 |
+
"step": 4080
|
| 2930 |
+
},
|
| 2931 |
+
{
|
| 2932 |
+
"epoch": 1.28,
|
| 2933 |
+
"grad_norm": 0.0012170624686405063,
|
| 2934 |
+
"learning_rate": 4.044444444444445e-06,
|
| 2935 |
+
"loss": 0.0438,
|
| 2936 |
+
"step": 4090
|
| 2937 |
+
},
|
| 2938 |
+
{
|
| 2939 |
+
"epoch": 1.28,
|
| 2940 |
+
"grad_norm": 0.062455616891384125,
|
| 2941 |
+
"learning_rate": 4.000000000000001e-06,
|
| 2942 |
+
"loss": 0.0284,
|
| 2943 |
+
"step": 4100
|
| 2944 |
+
},
|
| 2945 |
+
{
|
| 2946 |
+
"epoch": 1.29,
|
| 2947 |
+
"grad_norm": 0.0014791043940931559,
|
| 2948 |
+
"learning_rate": 3.955555555555556e-06,
|
| 2949 |
+
"loss": 0.0212,
|
| 2950 |
+
"step": 4110
|
| 2951 |
+
},
|
| 2952 |
+
{
|
| 2953 |
+
"epoch": 1.29,
|
| 2954 |
+
"grad_norm": 3.445824384689331,
|
| 2955 |
+
"learning_rate": 3.911111111111112e-06,
|
| 2956 |
+
"loss": 0.0511,
|
| 2957 |
+
"step": 4120
|
| 2958 |
+
},
|
| 2959 |
+
{
|
| 2960 |
+
"epoch": 1.29,
|
| 2961 |
+
"grad_norm": 3.981981039047241,
|
| 2962 |
+
"learning_rate": 3.866666666666667e-06,
|
| 2963 |
+
"loss": 0.0482,
|
| 2964 |
+
"step": 4130
|
| 2965 |
+
},
|
| 2966 |
+
{
|
| 2967 |
+
"epoch": 1.3,
|
| 2968 |
+
"grad_norm": 0.04528600350022316,
|
| 2969 |
+
"learning_rate": 3.8222222222222224e-06,
|
| 2970 |
+
"loss": 0.0224,
|
| 2971 |
+
"step": 4140
|
| 2972 |
+
},
|
| 2973 |
+
{
|
| 2974 |
+
"epoch": 1.3,
|
| 2975 |
+
"grad_norm": 0.0005888556479476392,
|
| 2976 |
+
"learning_rate": 3.777777777777778e-06,
|
| 2977 |
+
"loss": 0.0517,
|
| 2978 |
+
"step": 4150
|
| 2979 |
+
},
|
| 2980 |
+
{
|
| 2981 |
+
"epoch": 1.3,
|
| 2982 |
+
"grad_norm": 2.5039355754852295,
|
| 2983 |
+
"learning_rate": 3.7333333333333337e-06,
|
| 2984 |
+
"loss": 0.037,
|
| 2985 |
+
"step": 4160
|
| 2986 |
+
},
|
| 2987 |
+
{
|
| 2988 |
+
"epoch": 1.31,
|
| 2989 |
+
"grad_norm": 2.8256263732910156,
|
| 2990 |
+
"learning_rate": 3.688888888888889e-06,
|
| 2991 |
+
"loss": 0.0261,
|
| 2992 |
+
"step": 4170
|
| 2993 |
+
},
|
| 2994 |
+
{
|
| 2995 |
+
"epoch": 1.31,
|
| 2996 |
+
"grad_norm": 0.0010332756210118532,
|
| 2997 |
+
"learning_rate": 3.644444444444445e-06,
|
| 2998 |
+
"loss": 0.0269,
|
| 2999 |
+
"step": 4180
|
| 3000 |
+
},
|
| 3001 |
+
{
|
| 3002 |
+
"epoch": 1.31,
|
| 3003 |
+
"grad_norm": 0.0005628996295854449,
|
| 3004 |
+
"learning_rate": 3.6000000000000003e-06,
|
| 3005 |
+
"loss": 0.0053,
|
| 3006 |
+
"step": 4190
|
| 3007 |
+
},
|
| 3008 |
+
{
|
| 3009 |
+
"epoch": 1.32,
|
| 3010 |
+
"grad_norm": 0.0006191189750097692,
|
| 3011 |
+
"learning_rate": 3.555555555555556e-06,
|
| 3012 |
+
"loss": 0.0056,
|
| 3013 |
+
"step": 4200
|
| 3014 |
+
},
|
| 3015 |
+
{
|
| 3016 |
+
"epoch": 1.32,
|
| 3017 |
+
"grad_norm": 2.178837537765503,
|
| 3018 |
+
"learning_rate": 3.511111111111111e-06,
|
| 3019 |
+
"loss": 0.0136,
|
| 3020 |
+
"step": 4210
|
| 3021 |
+
},
|
| 3022 |
+
{
|
| 3023 |
+
"epoch": 1.32,
|
| 3024 |
+
"grad_norm": 0.00813736766576767,
|
| 3025 |
+
"learning_rate": 3.4666666666666672e-06,
|
| 3026 |
+
"loss": 0.0249,
|
| 3027 |
+
"step": 4220
|
| 3028 |
+
},
|
| 3029 |
+
{
|
| 3030 |
+
"epoch": 1.33,
|
| 3031 |
+
"grad_norm": 0.0015987426741048694,
|
| 3032 |
+
"learning_rate": 3.4222222222222224e-06,
|
| 3033 |
+
"loss": 0.0424,
|
| 3034 |
+
"step": 4230
|
| 3035 |
+
},
|
| 3036 |
+
{
|
| 3037 |
+
"epoch": 1.33,
|
| 3038 |
+
"grad_norm": 0.0005763465305790305,
|
| 3039 |
+
"learning_rate": 3.377777777777778e-06,
|
| 3040 |
+
"loss": 0.0149,
|
| 3041 |
+
"step": 4240
|
| 3042 |
+
},
|
| 3043 |
+
{
|
| 3044 |
+
"epoch": 1.33,
|
| 3045 |
+
"grad_norm": 0.0005146890762262046,
|
| 3046 |
+
"learning_rate": 3.3333333333333333e-06,
|
| 3047 |
+
"loss": 0.0207,
|
| 3048 |
+
"step": 4250
|
| 3049 |
+
},
|
| 3050 |
+
{
|
| 3051 |
+
"epoch": 1.33,
|
| 3052 |
+
"grad_norm": 0.0012806435115635395,
|
| 3053 |
+
"learning_rate": 3.2888888888888894e-06,
|
| 3054 |
+
"loss": 0.0157,
|
| 3055 |
+
"step": 4260
|
| 3056 |
+
},
|
| 3057 |
+
{
|
| 3058 |
+
"epoch": 1.34,
|
| 3059 |
+
"grad_norm": 0.0015921818558126688,
|
| 3060 |
+
"learning_rate": 3.2444444444444446e-06,
|
| 3061 |
+
"loss": 0.029,
|
| 3062 |
+
"step": 4270
|
| 3063 |
+
},
|
| 3064 |
+
{
|
| 3065 |
+
"epoch": 1.34,
|
| 3066 |
+
"grad_norm": 3.0711112022399902,
|
| 3067 |
+
"learning_rate": 3.2000000000000003e-06,
|
| 3068 |
+
"loss": 0.1302,
|
| 3069 |
+
"step": 4280
|
| 3070 |
+
},
|
| 3071 |
+
{
|
| 3072 |
+
"epoch": 1.34,
|
| 3073 |
+
"grad_norm": 0.0007020393386483192,
|
| 3074 |
+
"learning_rate": 3.1555555555555555e-06,
|
| 3075 |
+
"loss": 0.0076,
|
| 3076 |
+
"step": 4290
|
| 3077 |
+
},
|
| 3078 |
+
{
|
| 3079 |
+
"epoch": 1.35,
|
| 3080 |
+
"grad_norm": 0.0005048629245720804,
|
| 3081 |
+
"learning_rate": 3.1111111111111116e-06,
|
| 3082 |
+
"loss": 0.0086,
|
| 3083 |
+
"step": 4300
|
| 3084 |
+
},
|
| 3085 |
+
{
|
| 3086 |
+
"epoch": 1.35,
|
| 3087 |
+
"grad_norm": 0.0023996131494641304,
|
| 3088 |
+
"learning_rate": 3.066666666666667e-06,
|
| 3089 |
+
"loss": 0.0196,
|
| 3090 |
+
"step": 4310
|
| 3091 |
+
},
|
| 3092 |
+
{
|
| 3093 |
+
"epoch": 1.35,
|
| 3094 |
+
"grad_norm": 0.0025332309305667877,
|
| 3095 |
+
"learning_rate": 3.0222222222222225e-06,
|
| 3096 |
+
"loss": 0.0461,
|
| 3097 |
+
"step": 4320
|
| 3098 |
+
},
|
| 3099 |
+
{
|
| 3100 |
+
"epoch": 1.36,
|
| 3101 |
+
"grad_norm": 0.0007468872354365885,
|
| 3102 |
+
"learning_rate": 2.9777777777777777e-06,
|
| 3103 |
+
"loss": 0.0282,
|
| 3104 |
+
"step": 4330
|
| 3105 |
+
},
|
| 3106 |
+
{
|
| 3107 |
+
"epoch": 1.36,
|
| 3108 |
+
"grad_norm": 0.0005446248687803745,
|
| 3109 |
+
"learning_rate": 2.9333333333333338e-06,
|
| 3110 |
+
"loss": 0.0469,
|
| 3111 |
+
"step": 4340
|
| 3112 |
+
},
|
| 3113 |
+
{
|
| 3114 |
+
"epoch": 1.36,
|
| 3115 |
+
"grad_norm": 0.0011565398890525103,
|
| 3116 |
+
"learning_rate": 2.888888888888889e-06,
|
| 3117 |
+
"loss": 0.0335,
|
| 3118 |
+
"step": 4350
|
| 3119 |
+
},
|
| 3120 |
+
{
|
| 3121 |
+
"epoch": 1.37,
|
| 3122 |
+
"grad_norm": 1.8544560670852661,
|
| 3123 |
+
"learning_rate": 2.8444444444444446e-06,
|
| 3124 |
+
"loss": 0.0703,
|
| 3125 |
+
"step": 4360
|
| 3126 |
+
},
|
| 3127 |
+
{
|
| 3128 |
+
"epoch": 1.37,
|
| 3129 |
+
"grad_norm": 3.215923547744751,
|
| 3130 |
+
"learning_rate": 2.8000000000000003e-06,
|
| 3131 |
+
"loss": 0.0396,
|
| 3132 |
+
"step": 4370
|
| 3133 |
+
},
|
| 3134 |
+
{
|
| 3135 |
+
"epoch": 1.37,
|
| 3136 |
+
"grad_norm": 1.535111427307129,
|
| 3137 |
+
"learning_rate": 2.755555555555556e-06,
|
| 3138 |
+
"loss": 0.0719,
|
| 3139 |
+
"step": 4380
|
| 3140 |
+
},
|
| 3141 |
+
{
|
| 3142 |
+
"epoch": 1.38,
|
| 3143 |
+
"grad_norm": 0.01463981345295906,
|
| 3144 |
+
"learning_rate": 2.7111111111111116e-06,
|
| 3145 |
+
"loss": 0.0246,
|
| 3146 |
+
"step": 4390
|
| 3147 |
+
},
|
| 3148 |
+
{
|
| 3149 |
+
"epoch": 1.38,
|
| 3150 |
+
"grad_norm": 4.059357643127441,
|
| 3151 |
+
"learning_rate": 2.666666666666667e-06,
|
| 3152 |
+
"loss": 0.0367,
|
| 3153 |
+
"step": 4400
|
| 3154 |
+
},
|
| 3155 |
+
{
|
| 3156 |
+
"epoch": 1.38,
|
| 3157 |
+
"grad_norm": 0.0006797238602302969,
|
| 3158 |
+
"learning_rate": 2.6222222222222225e-06,
|
| 3159 |
+
"loss": 0.0489,
|
| 3160 |
+
"step": 4410
|
| 3161 |
+
},
|
| 3162 |
+
{
|
| 3163 |
+
"epoch": 1.38,
|
| 3164 |
+
"grad_norm": 1.5824676752090454,
|
| 3165 |
+
"learning_rate": 2.577777777777778e-06,
|
| 3166 |
+
"loss": 0.0285,
|
| 3167 |
+
"step": 4420
|
| 3168 |
+
},
|
| 3169 |
+
{
|
| 3170 |
+
"epoch": 1.39,
|
| 3171 |
+
"grad_norm": 0.00045088992919772863,
|
| 3172 |
+
"learning_rate": 2.5333333333333338e-06,
|
| 3173 |
+
"loss": 0.0328,
|
| 3174 |
+
"step": 4430
|
| 3175 |
+
},
|
| 3176 |
+
{
|
| 3177 |
+
"epoch": 1.39,
|
| 3178 |
+
"grad_norm": 1.129044771194458,
|
| 3179 |
+
"learning_rate": 2.488888888888889e-06,
|
| 3180 |
+
"loss": 0.058,
|
| 3181 |
+
"step": 4440
|
| 3182 |
+
},
|
| 3183 |
+
{
|
| 3184 |
+
"epoch": 1.39,
|
| 3185 |
+
"grad_norm": 0.00038907071575522423,
|
| 3186 |
+
"learning_rate": 2.4444444444444447e-06,
|
| 3187 |
+
"loss": 0.0504,
|
| 3188 |
+
"step": 4450
|
| 3189 |
+
},
|
| 3190 |
+
{
|
| 3191 |
+
"epoch": 1.4,
|
| 3192 |
+
"grad_norm": 0.0011422157986089587,
|
| 3193 |
+
"learning_rate": 2.4000000000000003e-06,
|
| 3194 |
+
"loss": 0.022,
|
| 3195 |
+
"step": 4460
|
| 3196 |
+
},
|
| 3197 |
+
{
|
| 3198 |
+
"epoch": 1.4,
|
| 3199 |
+
"grad_norm": 1.4815700054168701,
|
| 3200 |
+
"learning_rate": 2.3555555555555555e-06,
|
| 3201 |
+
"loss": 0.0416,
|
| 3202 |
+
"step": 4470
|
| 3203 |
+
},
|
| 3204 |
+
{
|
| 3205 |
+
"epoch": 1.4,
|
| 3206 |
+
"grad_norm": 2.4049506187438965,
|
| 3207 |
+
"learning_rate": 2.311111111111111e-06,
|
| 3208 |
+
"loss": 0.0286,
|
| 3209 |
+
"step": 4480
|
| 3210 |
+
},
|
| 3211 |
+
{
|
| 3212 |
+
"epoch": 1.41,
|
| 3213 |
+
"grad_norm": 0.32649314403533936,
|
| 3214 |
+
"learning_rate": 2.266666666666667e-06,
|
| 3215 |
+
"loss": 0.0191,
|
| 3216 |
+
"step": 4490
|
| 3217 |
+
},
|
| 3218 |
+
{
|
| 3219 |
+
"epoch": 1.41,
|
| 3220 |
+
"grad_norm": 0.001270271954126656,
|
| 3221 |
+
"learning_rate": 2.222222222222222e-06,
|
| 3222 |
+
"loss": 0.0208,
|
| 3223 |
+
"step": 4500
|
| 3224 |
+
},
|
| 3225 |
+
{
|
| 3226 |
+
"epoch": 1.41,
|
| 3227 |
+
"eval_loss": 0.018216300755739212,
|
| 3228 |
+
"eval_runtime": 61.8518,
|
| 3229 |
+
"eval_samples_per_second": 16.168,
|
| 3230 |
+
"eval_steps_per_second": 16.168,
|
| 3231 |
+
"step": 4500
|
| 3232 |
+
},
|
| 3233 |
+
{
|
| 3234 |
+
"epoch": 1.41,
|
| 3235 |
+
"grad_norm": 0.0010293573141098022,
|
| 3236 |
+
"learning_rate": 2.1777777777777777e-06,
|
| 3237 |
+
"loss": 0.0186,
|
| 3238 |
+
"step": 4510
|
| 3239 |
+
},
|
| 3240 |
+
{
|
| 3241 |
+
"epoch": 1.42,
|
| 3242 |
+
"grad_norm": 0.47490406036376953,
|
| 3243 |
+
"learning_rate": 2.133333333333334e-06,
|
| 3244 |
+
"loss": 0.0181,
|
| 3245 |
+
"step": 4520
|
| 3246 |
+
},
|
| 3247 |
+
{
|
| 3248 |
+
"epoch": 1.42,
|
| 3249 |
+
"grad_norm": 0.000406662467867136,
|
| 3250 |
+
"learning_rate": 2.088888888888889e-06,
|
| 3251 |
+
"loss": 0.0383,
|
| 3252 |
+
"step": 4530
|
| 3253 |
+
},
|
| 3254 |
+
{
|
| 3255 |
+
"epoch": 1.42,
|
| 3256 |
+
"grad_norm": 0.0009539751335978508,
|
| 3257 |
+
"learning_rate": 2.0444444444444447e-06,
|
| 3258 |
+
"loss": 0.0139,
|
| 3259 |
+
"step": 4540
|
| 3260 |
+
},
|
| 3261 |
+
{
|
| 3262 |
+
"epoch": 1.43,
|
| 3263 |
+
"grad_norm": 0.0005080964183434844,
|
| 3264 |
+
"learning_rate": 2.0000000000000003e-06,
|
| 3265 |
+
"loss": 0.0084,
|
| 3266 |
+
"step": 4550
|
| 3267 |
+
},
|
| 3268 |
+
{
|
| 3269 |
+
"epoch": 1.43,
|
| 3270 |
+
"grad_norm": 0.0007780414307489991,
|
| 3271 |
+
"learning_rate": 1.955555555555556e-06,
|
| 3272 |
+
"loss": 0.0133,
|
| 3273 |
+
"step": 4560
|
| 3274 |
+
},
|
| 3275 |
+
{
|
| 3276 |
+
"epoch": 1.43,
|
| 3277 |
+
"grad_norm": 0.0003762560954783112,
|
| 3278 |
+
"learning_rate": 1.9111111111111112e-06,
|
| 3279 |
+
"loss": 0.0149,
|
| 3280 |
+
"step": 4570
|
| 3281 |
+
},
|
| 3282 |
+
{
|
| 3283 |
+
"epoch": 1.43,
|
| 3284 |
+
"grad_norm": 0.0010716378455981612,
|
| 3285 |
+
"learning_rate": 1.8666666666666669e-06,
|
| 3286 |
+
"loss": 0.0096,
|
| 3287 |
+
"step": 4580
|
| 3288 |
+
},
|
| 3289 |
+
{
|
| 3290 |
+
"epoch": 1.44,
|
| 3291 |
+
"grad_norm": 0.0012602531351149082,
|
| 3292 |
+
"learning_rate": 1.8222222222222225e-06,
|
| 3293 |
+
"loss": 0.0501,
|
| 3294 |
+
"step": 4590
|
| 3295 |
+
},
|
| 3296 |
+
{
|
| 3297 |
+
"epoch": 1.44,
|
| 3298 |
+
"grad_norm": 0.0011576686520129442,
|
| 3299 |
+
"learning_rate": 1.777777777777778e-06,
|
| 3300 |
+
"loss": 0.0367,
|
| 3301 |
+
"step": 4600
|
| 3302 |
+
},
|
| 3303 |
+
{
|
| 3304 |
+
"epoch": 1.44,
|
| 3305 |
+
"grad_norm": 0.00044630825868807733,
|
| 3306 |
+
"learning_rate": 1.7333333333333336e-06,
|
| 3307 |
+
"loss": 0.0335,
|
| 3308 |
+
"step": 4610
|
| 3309 |
+
},
|
| 3310 |
+
{
|
| 3311 |
+
"epoch": 1.45,
|
| 3312 |
+
"grad_norm": 0.0005398432840593159,
|
| 3313 |
+
"learning_rate": 1.688888888888889e-06,
|
| 3314 |
+
"loss": 0.0359,
|
| 3315 |
+
"step": 4620
|
| 3316 |
+
},
|
| 3317 |
+
{
|
| 3318 |
+
"epoch": 1.45,
|
| 3319 |
+
"grad_norm": 0.7338486909866333,
|
| 3320 |
+
"learning_rate": 1.6444444444444447e-06,
|
| 3321 |
+
"loss": 0.0557,
|
| 3322 |
+
"step": 4630
|
| 3323 |
+
},
|
| 3324 |
+
{
|
| 3325 |
+
"epoch": 1.45,
|
| 3326 |
+
"grad_norm": 0.0004819166788365692,
|
| 3327 |
+
"learning_rate": 1.6000000000000001e-06,
|
| 3328 |
+
"loss": 0.0,
|
| 3329 |
+
"step": 4640
|
| 3330 |
+
},
|
| 3331 |
+
{
|
| 3332 |
+
"epoch": 1.46,
|
| 3333 |
+
"grad_norm": 1.6127287149429321,
|
| 3334 |
+
"learning_rate": 1.5555555555555558e-06,
|
| 3335 |
+
"loss": 0.0276,
|
| 3336 |
+
"step": 4650
|
| 3337 |
+
},
|
| 3338 |
+
{
|
| 3339 |
+
"epoch": 1.46,
|
| 3340 |
+
"grad_norm": 0.07772258669137955,
|
| 3341 |
+
"learning_rate": 1.5111111111111112e-06,
|
| 3342 |
+
"loss": 0.088,
|
| 3343 |
+
"step": 4660
|
| 3344 |
+
},
|
| 3345 |
+
{
|
| 3346 |
+
"epoch": 1.46,
|
| 3347 |
+
"grad_norm": 0.0005863177357241511,
|
| 3348 |
+
"learning_rate": 1.4666666666666669e-06,
|
| 3349 |
+
"loss": 0.0132,
|
| 3350 |
+
"step": 4670
|
| 3351 |
+
},
|
| 3352 |
+
{
|
| 3353 |
+
"epoch": 1.47,
|
| 3354 |
+
"grad_norm": 0.0006563079077750444,
|
| 3355 |
+
"learning_rate": 1.4222222222222223e-06,
|
| 3356 |
+
"loss": 0.0028,
|
| 3357 |
+
"step": 4680
|
| 3358 |
+
},
|
| 3359 |
+
{
|
| 3360 |
+
"epoch": 1.47,
|
| 3361 |
+
"grad_norm": 2.803096294403076,
|
| 3362 |
+
"learning_rate": 1.377777777777778e-06,
|
| 3363 |
+
"loss": 0.0694,
|
| 3364 |
+
"step": 4690
|
| 3365 |
+
},
|
| 3366 |
+
{
|
| 3367 |
+
"epoch": 1.47,
|
| 3368 |
+
"grad_norm": 0.7182486653327942,
|
| 3369 |
+
"learning_rate": 1.3333333333333334e-06,
|
| 3370 |
+
"loss": 0.0065,
|
| 3371 |
+
"step": 4700
|
| 3372 |
+
},
|
| 3373 |
+
{
|
| 3374 |
+
"epoch": 1.48,
|
| 3375 |
+
"grad_norm": 0.0009489316144026816,
|
| 3376 |
+
"learning_rate": 1.288888888888889e-06,
|
| 3377 |
+
"loss": 0.0136,
|
| 3378 |
+
"step": 4710
|
| 3379 |
+
},
|
| 3380 |
+
{
|
| 3381 |
+
"epoch": 1.48,
|
| 3382 |
+
"grad_norm": 0.001954195322468877,
|
| 3383 |
+
"learning_rate": 1.2444444444444445e-06,
|
| 3384 |
+
"loss": 0.0142,
|
| 3385 |
+
"step": 4720
|
| 3386 |
+
},
|
| 3387 |
+
{
|
| 3388 |
+
"epoch": 1.48,
|
| 3389 |
+
"grad_norm": 0.5916552543640137,
|
| 3390 |
+
"learning_rate": 1.2000000000000002e-06,
|
| 3391 |
+
"loss": 0.0469,
|
| 3392 |
+
"step": 4730
|
| 3393 |
+
},
|
| 3394 |
+
{
|
| 3395 |
+
"epoch": 1.49,
|
| 3396 |
+
"grad_norm": 2.119339942932129,
|
| 3397 |
+
"learning_rate": 1.1555555555555556e-06,
|
| 3398 |
+
"loss": 0.0551,
|
| 3399 |
+
"step": 4740
|
| 3400 |
+
},
|
| 3401 |
+
{
|
| 3402 |
+
"epoch": 1.49,
|
| 3403 |
+
"grad_norm": 0.0009017220581881702,
|
| 3404 |
+
"learning_rate": 1.111111111111111e-06,
|
| 3405 |
+
"loss": 0.0486,
|
| 3406 |
+
"step": 4750
|
| 3407 |
+
},
|
| 3408 |
+
{
|
| 3409 |
+
"epoch": 1.49,
|
| 3410 |
+
"grad_norm": 0.0012501273304224014,
|
| 3411 |
+
"learning_rate": 1.066666666666667e-06,
|
| 3412 |
+
"loss": 0.0048,
|
| 3413 |
+
"step": 4760
|
| 3414 |
+
},
|
| 3415 |
+
{
|
| 3416 |
+
"epoch": 1.49,
|
| 3417 |
+
"grad_norm": 0.8751838803291321,
|
| 3418 |
+
"learning_rate": 1.0222222222222223e-06,
|
| 3419 |
+
"loss": 0.0409,
|
| 3420 |
+
"step": 4770
|
| 3421 |
+
},
|
| 3422 |
+
{
|
| 3423 |
+
"epoch": 1.5,
|
| 3424 |
+
"grad_norm": 0.0013982664095237851,
|
| 3425 |
+
"learning_rate": 9.77777777777778e-07,
|
| 3426 |
+
"loss": 0.0577,
|
| 3427 |
+
"step": 4780
|
| 3428 |
+
},
|
| 3429 |
+
{
|
| 3430 |
+
"epoch": 1.5,
|
| 3431 |
+
"grad_norm": 0.0014040243113413453,
|
| 3432 |
+
"learning_rate": 9.333333333333334e-07,
|
| 3433 |
+
"loss": 0.031,
|
| 3434 |
+
"step": 4790
|
| 3435 |
+
},
|
| 3436 |
+
{
|
| 3437 |
+
"epoch": 1.5,
|
| 3438 |
+
"grad_norm": 3.977519989013672,
|
| 3439 |
+
"learning_rate": 8.88888888888889e-07,
|
| 3440 |
+
"loss": 0.1057,
|
| 3441 |
+
"step": 4800
|
| 3442 |
+
},
|
| 3443 |
+
{
|
| 3444 |
+
"epoch": 1.51,
|
| 3445 |
+
"grad_norm": 0.0011434502666816115,
|
| 3446 |
+
"learning_rate": 8.444444444444445e-07,
|
| 3447 |
+
"loss": 0.0066,
|
| 3448 |
+
"step": 4810
|
| 3449 |
+
},
|
| 3450 |
+
{
|
| 3451 |
+
"epoch": 1.51,
|
| 3452 |
+
"grad_norm": 0.0013947113184258342,
|
| 3453 |
+
"learning_rate": 8.000000000000001e-07,
|
| 3454 |
+
"loss": 0.0337,
|
| 3455 |
+
"step": 4820
|
| 3456 |
+
},
|
| 3457 |
+
{
|
| 3458 |
+
"epoch": 1.51,
|
| 3459 |
+
"grad_norm": 0.3012702167034149,
|
| 3460 |
+
"learning_rate": 7.555555555555556e-07,
|
| 3461 |
+
"loss": 0.0554,
|
| 3462 |
+
"step": 4830
|
| 3463 |
+
},
|
| 3464 |
+
{
|
| 3465 |
+
"epoch": 1.52,
|
| 3466 |
+
"grad_norm": 0.1272149384021759,
|
| 3467 |
+
"learning_rate": 7.111111111111112e-07,
|
| 3468 |
+
"loss": 0.0208,
|
| 3469 |
+
"step": 4840
|
| 3470 |
+
},
|
| 3471 |
+
{
|
| 3472 |
+
"epoch": 1.52,
|
| 3473 |
+
"grad_norm": 0.004117014352232218,
|
| 3474 |
+
"learning_rate": 6.666666666666667e-07,
|
| 3475 |
+
"loss": 0.0357,
|
| 3476 |
+
"step": 4850
|
| 3477 |
+
},
|
| 3478 |
+
{
|
| 3479 |
+
"epoch": 1.52,
|
| 3480 |
+
"grad_norm": 0.0005114726372994483,
|
| 3481 |
+
"learning_rate": 6.222222222222223e-07,
|
| 3482 |
+
"loss": 0.0189,
|
| 3483 |
+
"step": 4860
|
| 3484 |
+
},
|
| 3485 |
+
{
|
| 3486 |
+
"epoch": 1.53,
|
| 3487 |
+
"grad_norm": 0.000449322717031464,
|
| 3488 |
+
"learning_rate": 5.777777777777778e-07,
|
| 3489 |
+
"loss": 0.079,
|
| 3490 |
+
"step": 4870
|
| 3491 |
+
},
|
| 3492 |
+
{
|
| 3493 |
+
"epoch": 1.53,
|
| 3494 |
+
"grad_norm": 1.3229221105575562,
|
| 3495 |
+
"learning_rate": 5.333333333333335e-07,
|
| 3496 |
+
"loss": 0.0197,
|
| 3497 |
+
"step": 4880
|
| 3498 |
+
},
|
| 3499 |
+
{
|
| 3500 |
+
"epoch": 1.53,
|
| 3501 |
+
"grad_norm": 4.149044990539551,
|
| 3502 |
+
"learning_rate": 4.88888888888889e-07,
|
| 3503 |
+
"loss": 0.0383,
|
| 3504 |
+
"step": 4890
|
| 3505 |
+
},
|
| 3506 |
+
{
|
| 3507 |
+
"epoch": 1.54,
|
| 3508 |
+
"grad_norm": 0.00048808971769176424,
|
| 3509 |
+
"learning_rate": 4.444444444444445e-07,
|
| 3510 |
+
"loss": 0.0473,
|
| 3511 |
+
"step": 4900
|
| 3512 |
+
},
|
| 3513 |
+
{
|
| 3514 |
+
"epoch": 1.54,
|
| 3515 |
+
"grad_norm": 0.0010669564362615347,
|
| 3516 |
+
"learning_rate": 4.0000000000000003e-07,
|
| 3517 |
+
"loss": 0.0838,
|
| 3518 |
+
"step": 4910
|
| 3519 |
+
},
|
| 3520 |
+
{
|
| 3521 |
+
"epoch": 1.54,
|
| 3522 |
+
"grad_norm": 0.030236389487981796,
|
| 3523 |
+
"learning_rate": 3.555555555555556e-07,
|
| 3524 |
+
"loss": 0.0108,
|
| 3525 |
+
"step": 4920
|
| 3526 |
+
},
|
| 3527 |
+
{
|
| 3528 |
+
"epoch": 1.54,
|
| 3529 |
+
"grad_norm": 0.0004108142456971109,
|
| 3530 |
+
"learning_rate": 3.111111111111111e-07,
|
| 3531 |
+
"loss": 0.0301,
|
| 3532 |
+
"step": 4930
|
| 3533 |
+
},
|
| 3534 |
+
{
|
| 3535 |
+
"epoch": 1.55,
|
| 3536 |
+
"grad_norm": 0.0008436466450802982,
|
| 3537 |
+
"learning_rate": 2.666666666666667e-07,
|
| 3538 |
+
"loss": 0.0101,
|
| 3539 |
+
"step": 4940
|
| 3540 |
+
},
|
| 3541 |
+
{
|
| 3542 |
+
"epoch": 1.55,
|
| 3543 |
+
"grad_norm": 0.0007567039574496448,
|
| 3544 |
+
"learning_rate": 2.2222222222222224e-07,
|
| 3545 |
+
"loss": 0.0237,
|
| 3546 |
+
"step": 4950
|
| 3547 |
+
},
|
| 3548 |
+
{
|
| 3549 |
+
"epoch": 1.55,
|
| 3550 |
+
"grad_norm": 0.001360149122774601,
|
| 3551 |
+
"learning_rate": 1.777777777777778e-07,
|
| 3552 |
+
"loss": 0.0352,
|
| 3553 |
+
"step": 4960
|
| 3554 |
+
},
|
| 3555 |
+
{
|
| 3556 |
+
"epoch": 1.56,
|
| 3557 |
+
"grad_norm": 0.0012200737837702036,
|
| 3558 |
+
"learning_rate": 1.3333333333333336e-07,
|
| 3559 |
+
"loss": 0.0482,
|
| 3560 |
+
"step": 4970
|
| 3561 |
+
},
|
| 3562 |
+
{
|
| 3563 |
+
"epoch": 1.56,
|
| 3564 |
+
"grad_norm": 0.0006987242959439754,
|
| 3565 |
+
"learning_rate": 8.88888888888889e-08,
|
| 3566 |
+
"loss": 0.0164,
|
| 3567 |
+
"step": 4980
|
| 3568 |
+
},
|
| 3569 |
+
{
|
| 3570 |
+
"epoch": 1.56,
|
| 3571 |
+
"grad_norm": 0.005137372761964798,
|
| 3572 |
+
"learning_rate": 4.444444444444445e-08,
|
| 3573 |
+
"loss": 0.0424,
|
| 3574 |
+
"step": 4990
|
| 3575 |
+
},
|
| 3576 |
+
{
|
| 3577 |
+
"epoch": 1.57,
|
| 3578 |
+
"grad_norm": 0.0009189796401187778,
|
| 3579 |
+
"learning_rate": 0.0,
|
| 3580 |
+
"loss": 0.0353,
|
| 3581 |
+
"step": 5000
|
| 3582 |
+
},
|
| 3583 |
+
{
|
| 3584 |
+
"epoch": 1.57,
|
| 3585 |
+
"eval_loss": 0.021896515041589737,
|
| 3586 |
+
"eval_runtime": 61.7867,
|
| 3587 |
+
"eval_samples_per_second": 16.185,
|
| 3588 |
+
"eval_steps_per_second": 16.185,
|
| 3589 |
+
"step": 5000
|
| 3590 |
}
|
| 3591 |
],
|
| 3592 |
"logging_steps": 10,
|
|
|
|
| 3594 |
"num_input_tokens_seen": 0,
|
| 3595 |
"num_train_epochs": 2,
|
| 3596 |
"save_steps": 2500,
|
| 3597 |
+
"total_flos": 8.051062996992e+16,
|
| 3598 |
"train_batch_size": 1,
|
| 3599 |
"trial_name": null,
|
| 3600 |
"trial_params": null
|