Upload folder using huggingface_hub
Browse files- model.safetensors +1 -1
- optimizer.pt +1 -1
- rng_state_0.pth +1 -1
- rng_state_1.pth +1 -1
- scheduler.pt +1 -1
- trainer_state.json +1221 -5
model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1421709600
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0a4dc01a62180b671235df605bd949d91ee5e79a277541e196862797a12cb410
|
| 3 |
size 1421709600
|
optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 712762106
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:caa571c3b6266b8d7489d60262f29cb4d0642227fe03ea07f48f0ec877638e5f
|
| 3 |
size 712762106
|
rng_state_0.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14512
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:33cecfafe0d11e62c18063727af52cc3be05e3a270099438df514f675a702b94
|
| 3 |
size 14512
|
rng_state_1.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14512
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:988254c6cd2ce48fc90f3031675157ff19bb6013dae9b99c16a8a82f7dba3fd7
|
| 3 |
size 14512
|
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 sha256:1c34cc6321076d6d0f8c53d44b0a1cde5a1527d5c424cf5e2d4fffd6d91186ef
|
| 3 |
size 1064
|
trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
-
"best_metric": 2.
|
| 3 |
-
"best_model_checkpoint": "checkpoints-mistral-300M/checkpoint-
|
| 4 |
-
"epoch": 13.
|
| 5 |
"eval_steps": 1000,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -61415,13 +61415,1229 @@
|
|
| 61415 |
"eval_samples_per_second": 59.929,
|
| 61416 |
"eval_steps_per_second": 4.994,
|
| 61417 |
"step": 101000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 61418 |
}
|
| 61419 |
],
|
| 61420 |
"logging_steps": 10,
|
| 61421 |
"max_steps": 111060,
|
| 61422 |
"num_train_epochs": 15,
|
| 61423 |
"save_steps": 1000,
|
| 61424 |
-
"total_flos": 3.
|
| 61425 |
"trial_name": null,
|
| 61426 |
"trial_params": null
|
| 61427 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_metric": 2.39996337890625,
|
| 3 |
+
"best_model_checkpoint": "checkpoints-mistral-300M/checkpoint-103000",
|
| 4 |
+
"epoch": 13.910818886206721,
|
| 5 |
"eval_steps": 1000,
|
| 6 |
+
"global_step": 103000,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 61415 |
"eval_samples_per_second": 59.929,
|
| 61416 |
"eval_steps_per_second": 4.994,
|
| 61417 |
"step": 101000
|
| 61418 |
+
},
|
| 61419 |
+
{
|
| 61420 |
+
"epoch": 13.64,
|
| 61421 |
+
"learning_rate": 6.052167733471941e-06,
|
| 61422 |
+
"loss": 2.2055,
|
| 61423 |
+
"step": 101010
|
| 61424 |
+
},
|
| 61425 |
+
{
|
| 61426 |
+
"epoch": 13.64,
|
| 61427 |
+
"learning_rate": 6.0402418995578684e-06,
|
| 61428 |
+
"loss": 2.1925,
|
| 61429 |
+
"step": 101020
|
| 61430 |
+
},
|
| 61431 |
+
{
|
| 61432 |
+
"epoch": 13.64,
|
| 61433 |
+
"learning_rate": 6.028327585758325e-06,
|
| 61434 |
+
"loss": 2.1854,
|
| 61435 |
+
"step": 101030
|
| 61436 |
+
},
|
| 61437 |
+
{
|
| 61438 |
+
"epoch": 13.65,
|
| 61439 |
+
"learning_rate": 6.016424793026714e-06,
|
| 61440 |
+
"loss": 2.1818,
|
| 61441 |
+
"step": 101040
|
| 61442 |
+
},
|
| 61443 |
+
{
|
| 61444 |
+
"epoch": 13.65,
|
| 61445 |
+
"learning_rate": 6.004533522315558e-06,
|
| 61446 |
+
"loss": 2.2056,
|
| 61447 |
+
"step": 101050
|
| 61448 |
+
},
|
| 61449 |
+
{
|
| 61450 |
+
"epoch": 13.65,
|
| 61451 |
+
"learning_rate": 5.9926537745764115e-06,
|
| 61452 |
+
"loss": 2.2048,
|
| 61453 |
+
"step": 101060
|
| 61454 |
+
},
|
| 61455 |
+
{
|
| 61456 |
+
"epoch": 13.65,
|
| 61457 |
+
"learning_rate": 5.980785550759965e-06,
|
| 61458 |
+
"loss": 2.1758,
|
| 61459 |
+
"step": 101070
|
| 61460 |
+
},
|
| 61461 |
+
{
|
| 61462 |
+
"epoch": 13.65,
|
| 61463 |
+
"learning_rate": 5.968928851815907e-06,
|
| 61464 |
+
"loss": 2.2104,
|
| 61465 |
+
"step": 101080
|
| 61466 |
+
},
|
| 61467 |
+
{
|
| 61468 |
+
"epoch": 13.65,
|
| 61469 |
+
"learning_rate": 5.957083678693081e-06,
|
| 61470 |
+
"loss": 2.197,
|
| 61471 |
+
"step": 101090
|
| 61472 |
+
},
|
| 61473 |
+
{
|
| 61474 |
+
"epoch": 13.65,
|
| 61475 |
+
"learning_rate": 5.94525003233936e-06,
|
| 61476 |
+
"loss": 2.197,
|
| 61477 |
+
"step": 101100
|
| 61478 |
+
},
|
| 61479 |
+
{
|
| 61480 |
+
"epoch": 13.66,
|
| 61481 |
+
"learning_rate": 5.933427913701722e-06,
|
| 61482 |
+
"loss": 2.2029,
|
| 61483 |
+
"step": 101110
|
| 61484 |
+
},
|
| 61485 |
+
{
|
| 61486 |
+
"epoch": 13.66,
|
| 61487 |
+
"learning_rate": 5.9216173237262075e-06,
|
| 61488 |
+
"loss": 2.1813,
|
| 61489 |
+
"step": 101120
|
| 61490 |
+
},
|
| 61491 |
+
{
|
| 61492 |
+
"epoch": 13.66,
|
| 61493 |
+
"learning_rate": 5.9098182633579125e-06,
|
| 61494 |
+
"loss": 2.2079,
|
| 61495 |
+
"step": 101130
|
| 61496 |
+
},
|
| 61497 |
+
{
|
| 61498 |
+
"epoch": 13.66,
|
| 61499 |
+
"learning_rate": 5.898030733541065e-06,
|
| 61500 |
+
"loss": 2.233,
|
| 61501 |
+
"step": 101140
|
| 61502 |
+
},
|
| 61503 |
+
{
|
| 61504 |
+
"epoch": 13.66,
|
| 61505 |
+
"learning_rate": 5.8862547352189424e-06,
|
| 61506 |
+
"loss": 2.1991,
|
| 61507 |
+
"step": 101150
|
| 61508 |
+
},
|
| 61509 |
+
{
|
| 61510 |
+
"epoch": 13.66,
|
| 61511 |
+
"learning_rate": 5.874490269333859e-06,
|
| 61512 |
+
"loss": 2.2204,
|
| 61513 |
+
"step": 101160
|
| 61514 |
+
},
|
| 61515 |
+
{
|
| 61516 |
+
"epoch": 13.66,
|
| 61517 |
+
"learning_rate": 5.862737336827294e-06,
|
| 61518 |
+
"loss": 2.1827,
|
| 61519 |
+
"step": 101170
|
| 61520 |
+
},
|
| 61521 |
+
{
|
| 61522 |
+
"epoch": 13.67,
|
| 61523 |
+
"learning_rate": 5.850995938639697e-06,
|
| 61524 |
+
"loss": 2.1891,
|
| 61525 |
+
"step": 101180
|
| 61526 |
+
},
|
| 61527 |
+
{
|
| 61528 |
+
"epoch": 13.67,
|
| 61529 |
+
"learning_rate": 5.839266075710747e-06,
|
| 61530 |
+
"loss": 2.2242,
|
| 61531 |
+
"step": 101190
|
| 61532 |
+
},
|
| 61533 |
+
{
|
| 61534 |
+
"epoch": 13.67,
|
| 61535 |
+
"learning_rate": 5.8275477489789944e-06,
|
| 61536 |
+
"loss": 2.1684,
|
| 61537 |
+
"step": 101200
|
| 61538 |
+
},
|
| 61539 |
+
{
|
| 61540 |
+
"epoch": 13.67,
|
| 61541 |
+
"learning_rate": 5.815840959382223e-06,
|
| 61542 |
+
"loss": 2.232,
|
| 61543 |
+
"step": 101210
|
| 61544 |
+
},
|
| 61545 |
+
{
|
| 61546 |
+
"epoch": 13.67,
|
| 61547 |
+
"learning_rate": 5.804145707857266e-06,
|
| 61548 |
+
"loss": 2.1859,
|
| 61549 |
+
"step": 101220
|
| 61550 |
+
},
|
| 61551 |
+
{
|
| 61552 |
+
"epoch": 13.67,
|
| 61553 |
+
"learning_rate": 5.792461995340009e-06,
|
| 61554 |
+
"loss": 2.1793,
|
| 61555 |
+
"step": 101230
|
| 61556 |
+
},
|
| 61557 |
+
{
|
| 61558 |
+
"epoch": 13.67,
|
| 61559 |
+
"learning_rate": 5.780789822765402e-06,
|
| 61560 |
+
"loss": 2.201,
|
| 61561 |
+
"step": 101240
|
| 61562 |
+
},
|
| 61563 |
+
{
|
| 61564 |
+
"epoch": 13.67,
|
| 61565 |
+
"learning_rate": 5.769129191067467e-06,
|
| 61566 |
+
"loss": 2.2156,
|
| 61567 |
+
"step": 101250
|
| 61568 |
+
},
|
| 61569 |
+
{
|
| 61570 |
+
"epoch": 13.68,
|
| 61571 |
+
"learning_rate": 5.757480101179391e-06,
|
| 61572 |
+
"loss": 2.1748,
|
| 61573 |
+
"step": 101260
|
| 61574 |
+
},
|
| 61575 |
+
{
|
| 61576 |
+
"epoch": 13.68,
|
| 61577 |
+
"learning_rate": 5.745842554033309e-06,
|
| 61578 |
+
"loss": 2.1969,
|
| 61579 |
+
"step": 101270
|
| 61580 |
+
},
|
| 61581 |
+
{
|
| 61582 |
+
"epoch": 13.68,
|
| 61583 |
+
"learning_rate": 5.734216550560528e-06,
|
| 61584 |
+
"loss": 2.2004,
|
| 61585 |
+
"step": 101280
|
| 61586 |
+
},
|
| 61587 |
+
{
|
| 61588 |
+
"epoch": 13.68,
|
| 61589 |
+
"learning_rate": 5.7226020916913865e-06,
|
| 61590 |
+
"loss": 2.2341,
|
| 61591 |
+
"step": 101290
|
| 61592 |
+
},
|
| 61593 |
+
{
|
| 61594 |
+
"epoch": 13.68,
|
| 61595 |
+
"learning_rate": 5.710999178355291e-06,
|
| 61596 |
+
"loss": 2.2203,
|
| 61597 |
+
"step": 101300
|
| 61598 |
+
},
|
| 61599 |
+
{
|
| 61600 |
+
"epoch": 13.68,
|
| 61601 |
+
"learning_rate": 5.699407811480799e-06,
|
| 61602 |
+
"loss": 2.2237,
|
| 61603 |
+
"step": 101310
|
| 61604 |
+
},
|
| 61605 |
+
{
|
| 61606 |
+
"epoch": 13.68,
|
| 61607 |
+
"learning_rate": 5.687827991995436e-06,
|
| 61608 |
+
"loss": 2.2007,
|
| 61609 |
+
"step": 101320
|
| 61610 |
+
},
|
| 61611 |
+
{
|
| 61612 |
+
"epoch": 13.69,
|
| 61613 |
+
"learning_rate": 5.676259720825876e-06,
|
| 61614 |
+
"loss": 2.2128,
|
| 61615 |
+
"step": 101330
|
| 61616 |
+
},
|
| 61617 |
+
{
|
| 61618 |
+
"epoch": 13.69,
|
| 61619 |
+
"learning_rate": 5.664702998897847e-06,
|
| 61620 |
+
"loss": 2.2224,
|
| 61621 |
+
"step": 101340
|
| 61622 |
+
},
|
| 61623 |
+
{
|
| 61624 |
+
"epoch": 13.69,
|
| 61625 |
+
"learning_rate": 5.653157827136156e-06,
|
| 61626 |
+
"loss": 2.1993,
|
| 61627 |
+
"step": 101350
|
| 61628 |
+
},
|
| 61629 |
+
{
|
| 61630 |
+
"epoch": 13.69,
|
| 61631 |
+
"learning_rate": 5.641624206464668e-06,
|
| 61632 |
+
"loss": 2.1912,
|
| 61633 |
+
"step": 101360
|
| 61634 |
+
},
|
| 61635 |
+
{
|
| 61636 |
+
"epoch": 13.69,
|
| 61637 |
+
"learning_rate": 5.6301021378063586e-06,
|
| 61638 |
+
"loss": 2.217,
|
| 61639 |
+
"step": 101370
|
| 61640 |
+
},
|
| 61641 |
+
{
|
| 61642 |
+
"epoch": 13.69,
|
| 61643 |
+
"learning_rate": 5.618591622083258e-06,
|
| 61644 |
+
"loss": 2.2013,
|
| 61645 |
+
"step": 101380
|
| 61646 |
+
},
|
| 61647 |
+
{
|
| 61648 |
+
"epoch": 13.69,
|
| 61649 |
+
"learning_rate": 5.607092660216478e-06,
|
| 61650 |
+
"loss": 2.2333,
|
| 61651 |
+
"step": 101390
|
| 61652 |
+
},
|
| 61653 |
+
{
|
| 61654 |
+
"epoch": 13.69,
|
| 61655 |
+
"learning_rate": 5.595605253126184e-06,
|
| 61656 |
+
"loss": 2.2003,
|
| 61657 |
+
"step": 101400
|
| 61658 |
+
},
|
| 61659 |
+
{
|
| 61660 |
+
"epoch": 13.7,
|
| 61661 |
+
"learning_rate": 5.584129401731641e-06,
|
| 61662 |
+
"loss": 2.2048,
|
| 61663 |
+
"step": 101410
|
| 61664 |
+
},
|
| 61665 |
+
{
|
| 61666 |
+
"epoch": 13.7,
|
| 61667 |
+
"learning_rate": 5.572665106951179e-06,
|
| 61668 |
+
"loss": 2.1934,
|
| 61669 |
+
"step": 101420
|
| 61670 |
+
},
|
| 61671 |
+
{
|
| 61672 |
+
"epoch": 13.7,
|
| 61673 |
+
"learning_rate": 5.561212369702234e-06,
|
| 61674 |
+
"loss": 2.213,
|
| 61675 |
+
"step": 101430
|
| 61676 |
+
},
|
| 61677 |
+
{
|
| 61678 |
+
"epoch": 13.7,
|
| 61679 |
+
"learning_rate": 5.54977119090127e-06,
|
| 61680 |
+
"loss": 2.2165,
|
| 61681 |
+
"step": 101440
|
| 61682 |
+
},
|
| 61683 |
+
{
|
| 61684 |
+
"epoch": 13.7,
|
| 61685 |
+
"learning_rate": 5.538341571463839e-06,
|
| 61686 |
+
"loss": 2.2274,
|
| 61687 |
+
"step": 101450
|
| 61688 |
+
},
|
| 61689 |
+
{
|
| 61690 |
+
"epoch": 13.7,
|
| 61691 |
+
"learning_rate": 5.526923512304593e-06,
|
| 61692 |
+
"loss": 2.1762,
|
| 61693 |
+
"step": 101460
|
| 61694 |
+
},
|
| 61695 |
+
{
|
| 61696 |
+
"epoch": 13.7,
|
| 61697 |
+
"learning_rate": 5.5155170143372174e-06,
|
| 61698 |
+
"loss": 2.2149,
|
| 61699 |
+
"step": 101470
|
| 61700 |
+
},
|
| 61701 |
+
{
|
| 61702 |
+
"epoch": 13.71,
|
| 61703 |
+
"learning_rate": 5.504122078474516e-06,
|
| 61704 |
+
"loss": 2.2082,
|
| 61705 |
+
"step": 101480
|
| 61706 |
+
},
|
| 61707 |
+
{
|
| 61708 |
+
"epoch": 13.71,
|
| 61709 |
+
"learning_rate": 5.492738705628341e-06,
|
| 61710 |
+
"loss": 2.2212,
|
| 61711 |
+
"step": 101490
|
| 61712 |
+
},
|
| 61713 |
+
{
|
| 61714 |
+
"epoch": 13.71,
|
| 61715 |
+
"learning_rate": 5.481366896709616e-06,
|
| 61716 |
+
"loss": 2.1909,
|
| 61717 |
+
"step": 101500
|
| 61718 |
+
},
|
| 61719 |
+
{
|
| 61720 |
+
"epoch": 13.71,
|
| 61721 |
+
"learning_rate": 5.470006652628378e-06,
|
| 61722 |
+
"loss": 2.179,
|
| 61723 |
+
"step": 101510
|
| 61724 |
+
},
|
| 61725 |
+
{
|
| 61726 |
+
"epoch": 13.71,
|
| 61727 |
+
"learning_rate": 5.458657974293701e-06,
|
| 61728 |
+
"loss": 2.2055,
|
| 61729 |
+
"step": 101520
|
| 61730 |
+
},
|
| 61731 |
+
{
|
| 61732 |
+
"epoch": 13.71,
|
| 61733 |
+
"learning_rate": 5.4473208626137076e-06,
|
| 61734 |
+
"loss": 2.2205,
|
| 61735 |
+
"step": 101530
|
| 61736 |
+
},
|
| 61737 |
+
{
|
| 61738 |
+
"epoch": 13.71,
|
| 61739 |
+
"learning_rate": 5.435995318495656e-06,
|
| 61740 |
+
"loss": 2.1771,
|
| 61741 |
+
"step": 101540
|
| 61742 |
+
},
|
| 61743 |
+
{
|
| 61744 |
+
"epoch": 13.72,
|
| 61745 |
+
"learning_rate": 5.424681342845871e-06,
|
| 61746 |
+
"loss": 2.2074,
|
| 61747 |
+
"step": 101550
|
| 61748 |
+
},
|
| 61749 |
+
{
|
| 61750 |
+
"epoch": 13.72,
|
| 61751 |
+
"learning_rate": 5.413378936569695e-06,
|
| 61752 |
+
"loss": 2.1829,
|
| 61753 |
+
"step": 101560
|
| 61754 |
+
},
|
| 61755 |
+
{
|
| 61756 |
+
"epoch": 13.72,
|
| 61757 |
+
"learning_rate": 5.402088100571606e-06,
|
| 61758 |
+
"loss": 2.1937,
|
| 61759 |
+
"step": 101570
|
| 61760 |
+
},
|
| 61761 |
+
{
|
| 61762 |
+
"epoch": 13.72,
|
| 61763 |
+
"learning_rate": 5.390808835755111e-06,
|
| 61764 |
+
"loss": 2.2218,
|
| 61765 |
+
"step": 101580
|
| 61766 |
+
},
|
| 61767 |
+
{
|
| 61768 |
+
"epoch": 13.72,
|
| 61769 |
+
"learning_rate": 5.379541143022859e-06,
|
| 61770 |
+
"loss": 2.2359,
|
| 61771 |
+
"step": 101590
|
| 61772 |
+
},
|
| 61773 |
+
{
|
| 61774 |
+
"epoch": 13.72,
|
| 61775 |
+
"learning_rate": 5.368285023276458e-06,
|
| 61776 |
+
"loss": 2.1684,
|
| 61777 |
+
"step": 101600
|
| 61778 |
+
},
|
| 61779 |
+
{
|
| 61780 |
+
"epoch": 13.72,
|
| 61781 |
+
"learning_rate": 5.357040477416724e-06,
|
| 61782 |
+
"loss": 2.1749,
|
| 61783 |
+
"step": 101610
|
| 61784 |
+
},
|
| 61785 |
+
{
|
| 61786 |
+
"epoch": 13.72,
|
| 61787 |
+
"learning_rate": 5.345807506343419e-06,
|
| 61788 |
+
"loss": 2.1799,
|
| 61789 |
+
"step": 101620
|
| 61790 |
+
},
|
| 61791 |
+
{
|
| 61792 |
+
"epoch": 13.73,
|
| 61793 |
+
"learning_rate": 5.334586110955508e-06,
|
| 61794 |
+
"loss": 2.2011,
|
| 61795 |
+
"step": 101630
|
| 61796 |
+
},
|
| 61797 |
+
{
|
| 61798 |
+
"epoch": 13.73,
|
| 61799 |
+
"learning_rate": 5.3233762921509215e-06,
|
| 61800 |
+
"loss": 2.2168,
|
| 61801 |
+
"step": 101640
|
| 61802 |
+
},
|
| 61803 |
+
{
|
| 61804 |
+
"epoch": 13.73,
|
| 61805 |
+
"learning_rate": 5.3121780508266766e-06,
|
| 61806 |
+
"loss": 2.2133,
|
| 61807 |
+
"step": 101650
|
| 61808 |
+
},
|
| 61809 |
+
{
|
| 61810 |
+
"epoch": 13.73,
|
| 61811 |
+
"learning_rate": 5.300991387878972e-06,
|
| 61812 |
+
"loss": 2.22,
|
| 61813 |
+
"step": 101660
|
| 61814 |
+
},
|
| 61815 |
+
{
|
| 61816 |
+
"epoch": 13.73,
|
| 61817 |
+
"learning_rate": 5.289816304202927e-06,
|
| 61818 |
+
"loss": 2.2098,
|
| 61819 |
+
"step": 101670
|
| 61820 |
+
},
|
| 61821 |
+
{
|
| 61822 |
+
"epoch": 13.73,
|
| 61823 |
+
"learning_rate": 5.27865280069284e-06,
|
| 61824 |
+
"loss": 2.2029,
|
| 61825 |
+
"step": 101680
|
| 61826 |
+
},
|
| 61827 |
+
{
|
| 61828 |
+
"epoch": 13.73,
|
| 61829 |
+
"learning_rate": 5.267500878242048e-06,
|
| 61830 |
+
"loss": 2.1947,
|
| 61831 |
+
"step": 101690
|
| 61832 |
+
},
|
| 61833 |
+
{
|
| 61834 |
+
"epoch": 13.74,
|
| 61835 |
+
"learning_rate": 5.256360537742937e-06,
|
| 61836 |
+
"loss": 2.2075,
|
| 61837 |
+
"step": 101700
|
| 61838 |
+
},
|
| 61839 |
+
{
|
| 61840 |
+
"epoch": 13.74,
|
| 61841 |
+
"learning_rate": 5.245231780087044e-06,
|
| 61842 |
+
"loss": 2.201,
|
| 61843 |
+
"step": 101710
|
| 61844 |
+
},
|
| 61845 |
+
{
|
| 61846 |
+
"epoch": 13.74,
|
| 61847 |
+
"learning_rate": 5.2341146061648905e-06,
|
| 61848 |
+
"loss": 2.184,
|
| 61849 |
+
"step": 101720
|
| 61850 |
+
},
|
| 61851 |
+
{
|
| 61852 |
+
"epoch": 13.74,
|
| 61853 |
+
"learning_rate": 5.223009016866114e-06,
|
| 61854 |
+
"loss": 2.2246,
|
| 61855 |
+
"step": 101730
|
| 61856 |
+
},
|
| 61857 |
+
{
|
| 61858 |
+
"epoch": 13.74,
|
| 61859 |
+
"learning_rate": 5.2119150130794205e-06,
|
| 61860 |
+
"loss": 2.2135,
|
| 61861 |
+
"step": 101740
|
| 61862 |
+
},
|
| 61863 |
+
{
|
| 61864 |
+
"epoch": 13.74,
|
| 61865 |
+
"learning_rate": 5.200832595692583e-06,
|
| 61866 |
+
"loss": 2.2184,
|
| 61867 |
+
"step": 101750
|
| 61868 |
+
},
|
| 61869 |
+
{
|
| 61870 |
+
"epoch": 13.74,
|
| 61871 |
+
"learning_rate": 5.189761765592443e-06,
|
| 61872 |
+
"loss": 2.2135,
|
| 61873 |
+
"step": 101760
|
| 61874 |
+
},
|
| 61875 |
+
{
|
| 61876 |
+
"epoch": 13.74,
|
| 61877 |
+
"learning_rate": 5.178702523664957e-06,
|
| 61878 |
+
"loss": 2.1781,
|
| 61879 |
+
"step": 101770
|
| 61880 |
+
},
|
| 61881 |
+
{
|
| 61882 |
+
"epoch": 13.75,
|
| 61883 |
+
"learning_rate": 5.167654870795085e-06,
|
| 61884 |
+
"loss": 2.2232,
|
| 61885 |
+
"step": 101780
|
| 61886 |
+
},
|
| 61887 |
+
{
|
| 61888 |
+
"epoch": 13.75,
|
| 61889 |
+
"learning_rate": 5.156618807866936e-06,
|
| 61890 |
+
"loss": 2.2125,
|
| 61891 |
+
"step": 101790
|
| 61892 |
+
},
|
| 61893 |
+
{
|
| 61894 |
+
"epoch": 13.75,
|
| 61895 |
+
"learning_rate": 5.1455943357635864e-06,
|
| 61896 |
+
"loss": 2.2199,
|
| 61897 |
+
"step": 101800
|
| 61898 |
+
},
|
| 61899 |
+
{
|
| 61900 |
+
"epoch": 13.75,
|
| 61901 |
+
"learning_rate": 5.134581455367315e-06,
|
| 61902 |
+
"loss": 2.209,
|
| 61903 |
+
"step": 101810
|
| 61904 |
+
},
|
| 61905 |
+
{
|
| 61906 |
+
"epoch": 13.75,
|
| 61907 |
+
"learning_rate": 5.123580167559349e-06,
|
| 61908 |
+
"loss": 2.2137,
|
| 61909 |
+
"step": 101820
|
| 61910 |
+
},
|
| 61911 |
+
{
|
| 61912 |
+
"epoch": 13.75,
|
| 61913 |
+
"learning_rate": 5.112590473220102e-06,
|
| 61914 |
+
"loss": 2.203,
|
| 61915 |
+
"step": 101830
|
| 61916 |
+
},
|
| 61917 |
+
{
|
| 61918 |
+
"epoch": 13.75,
|
| 61919 |
+
"learning_rate": 5.101612373228969e-06,
|
| 61920 |
+
"loss": 2.1788,
|
| 61921 |
+
"step": 101840
|
| 61922 |
+
},
|
| 61923 |
+
{
|
| 61924 |
+
"epoch": 13.76,
|
| 61925 |
+
"learning_rate": 5.090645868464449e-06,
|
| 61926 |
+
"loss": 2.2026,
|
| 61927 |
+
"step": 101850
|
| 61928 |
+
},
|
| 61929 |
+
{
|
| 61930 |
+
"epoch": 13.76,
|
| 61931 |
+
"learning_rate": 5.079690959804156e-06,
|
| 61932 |
+
"loss": 2.2249,
|
| 61933 |
+
"step": 101860
|
| 61934 |
+
},
|
| 61935 |
+
{
|
| 61936 |
+
"epoch": 13.76,
|
| 61937 |
+
"learning_rate": 5.069841457403523e-06,
|
| 61938 |
+
"loss": 2.1968,
|
| 61939 |
+
"step": 101870
|
| 61940 |
+
},
|
| 61941 |
+
{
|
| 61942 |
+
"epoch": 13.76,
|
| 61943 |
+
"learning_rate": 5.058908583755594e-06,
|
| 61944 |
+
"loss": 2.2105,
|
| 61945 |
+
"step": 101880
|
| 61946 |
+
},
|
| 61947 |
+
{
|
| 61948 |
+
"epoch": 13.76,
|
| 61949 |
+
"learning_rate": 5.047987308751561e-06,
|
| 61950 |
+
"loss": 2.1889,
|
| 61951 |
+
"step": 101890
|
| 61952 |
+
},
|
| 61953 |
+
{
|
| 61954 |
+
"epoch": 13.76,
|
| 61955 |
+
"learning_rate": 5.037077633265424e-06,
|
| 61956 |
+
"loss": 2.2238,
|
| 61957 |
+
"step": 101900
|
| 61958 |
+
},
|
| 61959 |
+
{
|
| 61960 |
+
"epoch": 13.76,
|
| 61961 |
+
"learning_rate": 5.026179558170185e-06,
|
| 61962 |
+
"loss": 2.203,
|
| 61963 |
+
"step": 101910
|
| 61964 |
+
},
|
| 61965 |
+
{
|
| 61966 |
+
"epoch": 13.76,
|
| 61967 |
+
"learning_rate": 5.015293084337929e-06,
|
| 61968 |
+
"loss": 2.2099,
|
| 61969 |
+
"step": 101920
|
| 61970 |
+
},
|
| 61971 |
+
{
|
| 61972 |
+
"epoch": 13.77,
|
| 61973 |
+
"learning_rate": 5.004418212639844e-06,
|
| 61974 |
+
"loss": 2.2041,
|
| 61975 |
+
"step": 101930
|
| 61976 |
+
},
|
| 61977 |
+
{
|
| 61978 |
+
"epoch": 13.77,
|
| 61979 |
+
"learning_rate": 4.993554943946182e-06,
|
| 61980 |
+
"loss": 2.219,
|
| 61981 |
+
"step": 101940
|
| 61982 |
+
},
|
| 61983 |
+
{
|
| 61984 |
+
"epoch": 13.77,
|
| 61985 |
+
"learning_rate": 4.982703279126232e-06,
|
| 61986 |
+
"loss": 2.1971,
|
| 61987 |
+
"step": 101950
|
| 61988 |
+
},
|
| 61989 |
+
{
|
| 61990 |
+
"epoch": 13.77,
|
| 61991 |
+
"learning_rate": 4.971863219048383e-06,
|
| 61992 |
+
"loss": 2.1991,
|
| 61993 |
+
"step": 101960
|
| 61994 |
+
},
|
| 61995 |
+
{
|
| 61996 |
+
"epoch": 13.77,
|
| 61997 |
+
"learning_rate": 4.9610347645800904e-06,
|
| 61998 |
+
"loss": 2.2002,
|
| 61999 |
+
"step": 101970
|
| 62000 |
+
},
|
| 62001 |
+
{
|
| 62002 |
+
"epoch": 13.77,
|
| 62003 |
+
"learning_rate": 4.950217916587912e-06,
|
| 62004 |
+
"loss": 2.1817,
|
| 62005 |
+
"step": 101980
|
| 62006 |
+
},
|
| 62007 |
+
{
|
| 62008 |
+
"epoch": 13.77,
|
| 62009 |
+
"learning_rate": 4.939412675937405e-06,
|
| 62010 |
+
"loss": 2.2307,
|
| 62011 |
+
"step": 101990
|
| 62012 |
+
},
|
| 62013 |
+
{
|
| 62014 |
+
"epoch": 13.78,
|
| 62015 |
+
"learning_rate": 4.928619043493259e-06,
|
| 62016 |
+
"loss": 2.2212,
|
| 62017 |
+
"step": 102000
|
| 62018 |
+
},
|
| 62019 |
+
{
|
| 62020 |
+
"epoch": 13.78,
|
| 62021 |
+
"eval_loss": 2.400679111480713,
|
| 62022 |
+
"eval_runtime": 1953.3809,
|
| 62023 |
+
"eval_samples_per_second": 38.976,
|
| 62024 |
+
"eval_steps_per_second": 3.248,
|
| 62025 |
+
"step": 102000
|
| 62026 |
+
},
|
| 62027 |
+
{
|
| 62028 |
+
"epoch": 13.78,
|
| 62029 |
+
"learning_rate": 4.917837020119203e-06,
|
| 62030 |
+
"loss": 2.2046,
|
| 62031 |
+
"step": 102010
|
| 62032 |
+
},
|
| 62033 |
+
{
|
| 62034 |
+
"epoch": 13.78,
|
| 62035 |
+
"learning_rate": 4.907066606678062e-06,
|
| 62036 |
+
"loss": 2.2169,
|
| 62037 |
+
"step": 102020
|
| 62038 |
+
},
|
| 62039 |
+
{
|
| 62040 |
+
"epoch": 13.78,
|
| 62041 |
+
"learning_rate": 4.896307804031713e-06,
|
| 62042 |
+
"loss": 2.2054,
|
| 62043 |
+
"step": 102030
|
| 62044 |
+
},
|
| 62045 |
+
{
|
| 62046 |
+
"epoch": 13.78,
|
| 62047 |
+
"learning_rate": 4.885560613041101e-06,
|
| 62048 |
+
"loss": 2.2078,
|
| 62049 |
+
"step": 102040
|
| 62050 |
+
},
|
| 62051 |
+
{
|
| 62052 |
+
"epoch": 13.78,
|
| 62053 |
+
"learning_rate": 4.874825034566271e-06,
|
| 62054 |
+
"loss": 2.2067,
|
| 62055 |
+
"step": 102050
|
| 62056 |
+
},
|
| 62057 |
+
{
|
| 62058 |
+
"epoch": 13.78,
|
| 62059 |
+
"learning_rate": 4.86410106946632e-06,
|
| 62060 |
+
"loss": 2.234,
|
| 62061 |
+
"step": 102060
|
| 62062 |
+
},
|
| 62063 |
+
{
|
| 62064 |
+
"epoch": 13.79,
|
| 62065 |
+
"learning_rate": 4.8533887185993604e-06,
|
| 62066 |
+
"loss": 2.2144,
|
| 62067 |
+
"step": 102070
|
| 62068 |
+
},
|
| 62069 |
+
{
|
| 62070 |
+
"epoch": 13.79,
|
| 62071 |
+
"learning_rate": 4.842687982822691e-06,
|
| 62072 |
+
"loss": 2.2134,
|
| 62073 |
+
"step": 102080
|
| 62074 |
+
},
|
| 62075 |
+
{
|
| 62076 |
+
"epoch": 13.79,
|
| 62077 |
+
"learning_rate": 4.831998862992592e-06,
|
| 62078 |
+
"loss": 2.2034,
|
| 62079 |
+
"step": 102090
|
| 62080 |
+
},
|
| 62081 |
+
{
|
| 62082 |
+
"epoch": 13.79,
|
| 62083 |
+
"learning_rate": 4.821321359964447e-06,
|
| 62084 |
+
"loss": 2.1982,
|
| 62085 |
+
"step": 102100
|
| 62086 |
+
},
|
| 62087 |
+
{
|
| 62088 |
+
"epoch": 13.79,
|
| 62089 |
+
"learning_rate": 4.81065547459269e-06,
|
| 62090 |
+
"loss": 2.2427,
|
| 62091 |
+
"step": 102110
|
| 62092 |
+
},
|
| 62093 |
+
{
|
| 62094 |
+
"epoch": 13.79,
|
| 62095 |
+
"learning_rate": 4.800001207730836e-06,
|
| 62096 |
+
"loss": 2.234,
|
| 62097 |
+
"step": 102120
|
| 62098 |
+
},
|
| 62099 |
+
{
|
| 62100 |
+
"epoch": 13.79,
|
| 62101 |
+
"learning_rate": 4.789358560231521e-06,
|
| 62102 |
+
"loss": 2.2076,
|
| 62103 |
+
"step": 102130
|
| 62104 |
+
},
|
| 62105 |
+
{
|
| 62106 |
+
"epoch": 13.79,
|
| 62107 |
+
"learning_rate": 4.778727532946347e-06,
|
| 62108 |
+
"loss": 2.2022,
|
| 62109 |
+
"step": 102140
|
| 62110 |
+
},
|
| 62111 |
+
{
|
| 62112 |
+
"epoch": 13.8,
|
| 62113 |
+
"learning_rate": 4.768108126726067e-06,
|
| 62114 |
+
"loss": 2.2359,
|
| 62115 |
+
"step": 102150
|
| 62116 |
+
},
|
| 62117 |
+
{
|
| 62118 |
+
"epoch": 13.8,
|
| 62119 |
+
"learning_rate": 4.7575003424204675e-06,
|
| 62120 |
+
"loss": 2.2241,
|
| 62121 |
+
"step": 102160
|
| 62122 |
+
},
|
| 62123 |
+
{
|
| 62124 |
+
"epoch": 13.8,
|
| 62125 |
+
"learning_rate": 4.7469041808784195e-06,
|
| 62126 |
+
"loss": 2.211,
|
| 62127 |
+
"step": 102170
|
| 62128 |
+
},
|
| 62129 |
+
{
|
| 62130 |
+
"epoch": 13.8,
|
| 62131 |
+
"learning_rate": 4.736319642947895e-06,
|
| 62132 |
+
"loss": 2.2132,
|
| 62133 |
+
"step": 102180
|
| 62134 |
+
},
|
| 62135 |
+
{
|
| 62136 |
+
"epoch": 13.8,
|
| 62137 |
+
"learning_rate": 4.725746729475849e-06,
|
| 62138 |
+
"loss": 2.1927,
|
| 62139 |
+
"step": 102190
|
| 62140 |
+
},
|
| 62141 |
+
{
|
| 62142 |
+
"epoch": 13.8,
|
| 62143 |
+
"learning_rate": 4.715185441308389e-06,
|
| 62144 |
+
"loss": 2.215,
|
| 62145 |
+
"step": 102200
|
| 62146 |
+
},
|
| 62147 |
+
{
|
| 62148 |
+
"epoch": 13.8,
|
| 62149 |
+
"learning_rate": 4.704635779290655e-06,
|
| 62150 |
+
"loss": 2.1973,
|
| 62151 |
+
"step": 102210
|
| 62152 |
+
},
|
| 62153 |
+
{
|
| 62154 |
+
"epoch": 13.81,
|
| 62155 |
+
"learning_rate": 4.694097744266873e-06,
|
| 62156 |
+
"loss": 2.1824,
|
| 62157 |
+
"step": 102220
|
| 62158 |
+
},
|
| 62159 |
+
{
|
| 62160 |
+
"epoch": 13.81,
|
| 62161 |
+
"learning_rate": 4.683571337080333e-06,
|
| 62162 |
+
"loss": 2.2286,
|
| 62163 |
+
"step": 102230
|
| 62164 |
+
},
|
| 62165 |
+
{
|
| 62166 |
+
"epoch": 13.81,
|
| 62167 |
+
"learning_rate": 4.673056558573346e-06,
|
| 62168 |
+
"loss": 2.2037,
|
| 62169 |
+
"step": 102240
|
| 62170 |
+
},
|
| 62171 |
+
{
|
| 62172 |
+
"epoch": 13.81,
|
| 62173 |
+
"learning_rate": 4.662553409587405e-06,
|
| 62174 |
+
"loss": 2.2007,
|
| 62175 |
+
"step": 102250
|
| 62176 |
+
},
|
| 62177 |
+
{
|
| 62178 |
+
"epoch": 13.81,
|
| 62179 |
+
"learning_rate": 4.652061890962988e-06,
|
| 62180 |
+
"loss": 2.2184,
|
| 62181 |
+
"step": 102260
|
| 62182 |
+
},
|
| 62183 |
+
{
|
| 62184 |
+
"epoch": 13.81,
|
| 62185 |
+
"learning_rate": 4.641582003539606e-06,
|
| 62186 |
+
"loss": 2.197,
|
| 62187 |
+
"step": 102270
|
| 62188 |
+
},
|
| 62189 |
+
{
|
| 62190 |
+
"epoch": 13.81,
|
| 62191 |
+
"learning_rate": 4.6311137481559565e-06,
|
| 62192 |
+
"loss": 2.2,
|
| 62193 |
+
"step": 102280
|
| 62194 |
+
},
|
| 62195 |
+
{
|
| 62196 |
+
"epoch": 13.81,
|
| 62197 |
+
"learning_rate": 4.620657125649701e-06,
|
| 62198 |
+
"loss": 2.2164,
|
| 62199 |
+
"step": 102290
|
| 62200 |
+
},
|
| 62201 |
+
{
|
| 62202 |
+
"epoch": 13.82,
|
| 62203 |
+
"learning_rate": 4.610212136857672e-06,
|
| 62204 |
+
"loss": 2.1976,
|
| 62205 |
+
"step": 102300
|
| 62206 |
+
},
|
| 62207 |
+
{
|
| 62208 |
+
"epoch": 13.82,
|
| 62209 |
+
"learning_rate": 4.599778782615632e-06,
|
| 62210 |
+
"loss": 2.2099,
|
| 62211 |
+
"step": 102310
|
| 62212 |
+
},
|
| 62213 |
+
{
|
| 62214 |
+
"epoch": 13.82,
|
| 62215 |
+
"learning_rate": 4.589357063758514e-06,
|
| 62216 |
+
"loss": 2.2143,
|
| 62217 |
+
"step": 102320
|
| 62218 |
+
},
|
| 62219 |
+
{
|
| 62220 |
+
"epoch": 13.82,
|
| 62221 |
+
"learning_rate": 4.578946981120335e-06,
|
| 62222 |
+
"loss": 2.1881,
|
| 62223 |
+
"step": 102330
|
| 62224 |
+
},
|
| 62225 |
+
{
|
| 62226 |
+
"epoch": 13.82,
|
| 62227 |
+
"learning_rate": 4.568548535534111e-06,
|
| 62228 |
+
"loss": 2.2009,
|
| 62229 |
+
"step": 102340
|
| 62230 |
+
},
|
| 62231 |
+
{
|
| 62232 |
+
"epoch": 13.82,
|
| 62233 |
+
"learning_rate": 4.5581617278319604e-06,
|
| 62234 |
+
"loss": 2.2198,
|
| 62235 |
+
"step": 102350
|
| 62236 |
+
},
|
| 62237 |
+
{
|
| 62238 |
+
"epoch": 13.82,
|
| 62239 |
+
"learning_rate": 4.547786558845068e-06,
|
| 62240 |
+
"loss": 2.2031,
|
| 62241 |
+
"step": 102360
|
| 62242 |
+
},
|
| 62243 |
+
{
|
| 62244 |
+
"epoch": 13.83,
|
| 62245 |
+
"learning_rate": 4.537423029403686e-06,
|
| 62246 |
+
"loss": 2.1882,
|
| 62247 |
+
"step": 102370
|
| 62248 |
+
},
|
| 62249 |
+
{
|
| 62250 |
+
"epoch": 13.83,
|
| 62251 |
+
"learning_rate": 4.527071140337185e-06,
|
| 62252 |
+
"loss": 2.195,
|
| 62253 |
+
"step": 102380
|
| 62254 |
+
},
|
| 62255 |
+
{
|
| 62256 |
+
"epoch": 13.83,
|
| 62257 |
+
"learning_rate": 4.5167308924738685e-06,
|
| 62258 |
+
"loss": 2.205,
|
| 62259 |
+
"step": 102390
|
| 62260 |
+
},
|
| 62261 |
+
{
|
| 62262 |
+
"epoch": 13.83,
|
| 62263 |
+
"learning_rate": 4.506402286641259e-06,
|
| 62264 |
+
"loss": 2.1698,
|
| 62265 |
+
"step": 102400
|
| 62266 |
+
},
|
| 62267 |
+
{
|
| 62268 |
+
"epoch": 13.83,
|
| 62269 |
+
"learning_rate": 4.4960853236658435e-06,
|
| 62270 |
+
"loss": 2.1919,
|
| 62271 |
+
"step": 102410
|
| 62272 |
+
},
|
| 62273 |
+
{
|
| 62274 |
+
"epoch": 13.83,
|
| 62275 |
+
"learning_rate": 4.485780004373263e-06,
|
| 62276 |
+
"loss": 2.2087,
|
| 62277 |
+
"step": 102420
|
| 62278 |
+
},
|
| 62279 |
+
{
|
| 62280 |
+
"epoch": 13.83,
|
| 62281 |
+
"learning_rate": 4.475486329588157e-06,
|
| 62282 |
+
"loss": 2.202,
|
| 62283 |
+
"step": 102430
|
| 62284 |
+
},
|
| 62285 |
+
{
|
| 62286 |
+
"epoch": 13.84,
|
| 62287 |
+
"learning_rate": 4.4652043001342505e-06,
|
| 62288 |
+
"loss": 2.2202,
|
| 62289 |
+
"step": 102440
|
| 62290 |
+
},
|
| 62291 |
+
{
|
| 62292 |
+
"epoch": 13.84,
|
| 62293 |
+
"learning_rate": 4.454933916834369e-06,
|
| 62294 |
+
"loss": 2.172,
|
| 62295 |
+
"step": 102450
|
| 62296 |
+
},
|
| 62297 |
+
{
|
| 62298 |
+
"epoch": 13.84,
|
| 62299 |
+
"learning_rate": 4.4446751805103544e-06,
|
| 62300 |
+
"loss": 2.1928,
|
| 62301 |
+
"step": 102460
|
| 62302 |
+
},
|
| 62303 |
+
{
|
| 62304 |
+
"epoch": 13.84,
|
| 62305 |
+
"learning_rate": 4.43442809198315e-06,
|
| 62306 |
+
"loss": 2.2236,
|
| 62307 |
+
"step": 102470
|
| 62308 |
+
},
|
| 62309 |
+
{
|
| 62310 |
+
"epoch": 13.84,
|
| 62311 |
+
"learning_rate": 4.424192652072767e-06,
|
| 62312 |
+
"loss": 2.2181,
|
| 62313 |
+
"step": 102480
|
| 62314 |
+
},
|
| 62315 |
+
{
|
| 62316 |
+
"epoch": 13.84,
|
| 62317 |
+
"learning_rate": 4.413968861598266e-06,
|
| 62318 |
+
"loss": 2.2153,
|
| 62319 |
+
"step": 102490
|
| 62320 |
+
},
|
| 62321 |
+
{
|
| 62322 |
+
"epoch": 13.84,
|
| 62323 |
+
"learning_rate": 4.403756721377827e-06,
|
| 62324 |
+
"loss": 2.2056,
|
| 62325 |
+
"step": 102500
|
| 62326 |
+
},
|
| 62327 |
+
{
|
| 62328 |
+
"epoch": 13.84,
|
| 62329 |
+
"learning_rate": 4.393556232228596e-06,
|
| 62330 |
+
"loss": 2.1954,
|
| 62331 |
+
"step": 102510
|
| 62332 |
+
},
|
| 62333 |
+
{
|
| 62334 |
+
"epoch": 13.85,
|
| 62335 |
+
"learning_rate": 4.383367394966886e-06,
|
| 62336 |
+
"loss": 2.1726,
|
| 62337 |
+
"step": 102520
|
| 62338 |
+
},
|
| 62339 |
+
{
|
| 62340 |
+
"epoch": 13.85,
|
| 62341 |
+
"learning_rate": 4.373190210408045e-06,
|
| 62342 |
+
"loss": 2.1764,
|
| 62343 |
+
"step": 102530
|
| 62344 |
+
},
|
| 62345 |
+
{
|
| 62346 |
+
"epoch": 13.85,
|
| 62347 |
+
"learning_rate": 4.363024679366456e-06,
|
| 62348 |
+
"loss": 2.2037,
|
| 62349 |
+
"step": 102540
|
| 62350 |
+
},
|
| 62351 |
+
{
|
| 62352 |
+
"epoch": 13.85,
|
| 62353 |
+
"learning_rate": 4.352870802655633e-06,
|
| 62354 |
+
"loss": 2.1961,
|
| 62355 |
+
"step": 102550
|
| 62356 |
+
},
|
| 62357 |
+
{
|
| 62358 |
+
"epoch": 13.85,
|
| 62359 |
+
"learning_rate": 4.342728581088079e-06,
|
| 62360 |
+
"loss": 2.1842,
|
| 62361 |
+
"step": 102560
|
| 62362 |
+
},
|
| 62363 |
+
{
|
| 62364 |
+
"epoch": 13.85,
|
| 62365 |
+
"learning_rate": 4.332598015475425e-06,
|
| 62366 |
+
"loss": 2.1952,
|
| 62367 |
+
"step": 102570
|
| 62368 |
+
},
|
| 62369 |
+
{
|
| 62370 |
+
"epoch": 13.85,
|
| 62371 |
+
"learning_rate": 4.322479106628374e-06,
|
| 62372 |
+
"loss": 2.1948,
|
| 62373 |
+
"step": 102580
|
| 62374 |
+
},
|
| 62375 |
+
{
|
| 62376 |
+
"epoch": 13.86,
|
| 62377 |
+
"learning_rate": 4.3123718553566446e-06,
|
| 62378 |
+
"loss": 2.1408,
|
| 62379 |
+
"step": 102590
|
| 62380 |
+
},
|
| 62381 |
+
{
|
| 62382 |
+
"epoch": 13.86,
|
| 62383 |
+
"learning_rate": 4.302276262469056e-06,
|
| 62384 |
+
"loss": 2.2133,
|
| 62385 |
+
"step": 102600
|
| 62386 |
+
},
|
| 62387 |
+
{
|
| 62388 |
+
"epoch": 13.86,
|
| 62389 |
+
"learning_rate": 4.29219232877348e-06,
|
| 62390 |
+
"loss": 2.1878,
|
| 62391 |
+
"step": 102610
|
| 62392 |
+
},
|
| 62393 |
+
{
|
| 62394 |
+
"epoch": 13.86,
|
| 62395 |
+
"learning_rate": 4.282120055076904e-06,
|
| 62396 |
+
"loss": 2.2225,
|
| 62397 |
+
"step": 102620
|
| 62398 |
+
},
|
| 62399 |
+
{
|
| 62400 |
+
"epoch": 13.86,
|
| 62401 |
+
"learning_rate": 4.272059442185316e-06,
|
| 62402 |
+
"loss": 2.2082,
|
| 62403 |
+
"step": 102630
|
| 62404 |
+
},
|
| 62405 |
+
{
|
| 62406 |
+
"epoch": 13.86,
|
| 62407 |
+
"learning_rate": 4.262010490903772e-06,
|
| 62408 |
+
"loss": 2.1941,
|
| 62409 |
+
"step": 102640
|
| 62410 |
+
},
|
| 62411 |
+
{
|
| 62412 |
+
"epoch": 13.86,
|
| 62413 |
+
"learning_rate": 4.251973202036462e-06,
|
| 62414 |
+
"loss": 2.2062,
|
| 62415 |
+
"step": 102650
|
| 62416 |
+
},
|
| 62417 |
+
{
|
| 62418 |
+
"epoch": 13.86,
|
| 62419 |
+
"learning_rate": 4.241947576386595e-06,
|
| 62420 |
+
"loss": 2.1949,
|
| 62421 |
+
"step": 102660
|
| 62422 |
+
},
|
| 62423 |
+
{
|
| 62424 |
+
"epoch": 13.87,
|
| 62425 |
+
"learning_rate": 4.231933614756444e-06,
|
| 62426 |
+
"loss": 2.2056,
|
| 62427 |
+
"step": 102670
|
| 62428 |
+
},
|
| 62429 |
+
{
|
| 62430 |
+
"epoch": 13.87,
|
| 62431 |
+
"learning_rate": 4.221931317947352e-06,
|
| 62432 |
+
"loss": 2.2287,
|
| 62433 |
+
"step": 102680
|
| 62434 |
+
},
|
| 62435 |
+
{
|
| 62436 |
+
"epoch": 13.87,
|
| 62437 |
+
"learning_rate": 4.211940686759729e-06,
|
| 62438 |
+
"loss": 2.1932,
|
| 62439 |
+
"step": 102690
|
| 62440 |
+
},
|
| 62441 |
+
{
|
| 62442 |
+
"epoch": 13.87,
|
| 62443 |
+
"learning_rate": 4.201961721993119e-06,
|
| 62444 |
+
"loss": 2.196,
|
| 62445 |
+
"step": 102700
|
| 62446 |
+
},
|
| 62447 |
+
{
|
| 62448 |
+
"epoch": 13.87,
|
| 62449 |
+
"learning_rate": 4.191994424445966e-06,
|
| 62450 |
+
"loss": 2.2051,
|
| 62451 |
+
"step": 102710
|
| 62452 |
+
},
|
| 62453 |
+
{
|
| 62454 |
+
"epoch": 13.87,
|
| 62455 |
+
"learning_rate": 4.182038794915948e-06,
|
| 62456 |
+
"loss": 2.2149,
|
| 62457 |
+
"step": 102720
|
| 62458 |
+
},
|
| 62459 |
+
{
|
| 62460 |
+
"epoch": 13.87,
|
| 62461 |
+
"learning_rate": 4.1720948341997625e-06,
|
| 62462 |
+
"loss": 2.1769,
|
| 62463 |
+
"step": 102730
|
| 62464 |
+
},
|
| 62465 |
+
{
|
| 62466 |
+
"epoch": 13.88,
|
| 62467 |
+
"learning_rate": 4.162162543093123e-06,
|
| 62468 |
+
"loss": 2.2089,
|
| 62469 |
+
"step": 102740
|
| 62470 |
+
},
|
| 62471 |
+
{
|
| 62472 |
+
"epoch": 13.88,
|
| 62473 |
+
"learning_rate": 4.152241922390858e-06,
|
| 62474 |
+
"loss": 2.2152,
|
| 62475 |
+
"step": 102750
|
| 62476 |
+
},
|
| 62477 |
+
{
|
| 62478 |
+
"epoch": 13.88,
|
| 62479 |
+
"learning_rate": 4.14233297288682e-06,
|
| 62480 |
+
"loss": 2.1864,
|
| 62481 |
+
"step": 102760
|
| 62482 |
+
},
|
| 62483 |
+
{
|
| 62484 |
+
"epoch": 13.88,
|
| 62485 |
+
"learning_rate": 4.13243569537397e-06,
|
| 62486 |
+
"loss": 2.2047,
|
| 62487 |
+
"step": 102770
|
| 62488 |
+
},
|
| 62489 |
+
{
|
| 62490 |
+
"epoch": 13.88,
|
| 62491 |
+
"learning_rate": 4.122550090644361e-06,
|
| 62492 |
+
"loss": 2.2095,
|
| 62493 |
+
"step": 102780
|
| 62494 |
+
},
|
| 62495 |
+
{
|
| 62496 |
+
"epoch": 13.88,
|
| 62497 |
+
"learning_rate": 4.112676159488992e-06,
|
| 62498 |
+
"loss": 2.1928,
|
| 62499 |
+
"step": 102790
|
| 62500 |
+
},
|
| 62501 |
+
{
|
| 62502 |
+
"epoch": 13.88,
|
| 62503 |
+
"learning_rate": 4.102813902698082e-06,
|
| 62504 |
+
"loss": 2.2041,
|
| 62505 |
+
"step": 102800
|
| 62506 |
+
},
|
| 62507 |
+
{
|
| 62508 |
+
"epoch": 13.89,
|
| 62509 |
+
"learning_rate": 4.092963321060766e-06,
|
| 62510 |
+
"loss": 2.2078,
|
| 62511 |
+
"step": 102810
|
| 62512 |
+
},
|
| 62513 |
+
{
|
| 62514 |
+
"epoch": 13.89,
|
| 62515 |
+
"learning_rate": 4.0831244153653965e-06,
|
| 62516 |
+
"loss": 2.1855,
|
| 62517 |
+
"step": 102820
|
| 62518 |
+
},
|
| 62519 |
+
{
|
| 62520 |
+
"epoch": 13.89,
|
| 62521 |
+
"learning_rate": 4.07329718639926e-06,
|
| 62522 |
+
"loss": 2.2114,
|
| 62523 |
+
"step": 102830
|
| 62524 |
+
},
|
| 62525 |
+
{
|
| 62526 |
+
"epoch": 13.89,
|
| 62527 |
+
"learning_rate": 4.063481634948779e-06,
|
| 62528 |
+
"loss": 2.2252,
|
| 62529 |
+
"step": 102840
|
| 62530 |
+
},
|
| 62531 |
+
{
|
| 62532 |
+
"epoch": 13.89,
|
| 62533 |
+
"learning_rate": 4.053677761799423e-06,
|
| 62534 |
+
"loss": 2.1712,
|
| 62535 |
+
"step": 102850
|
| 62536 |
+
},
|
| 62537 |
+
{
|
| 62538 |
+
"epoch": 13.89,
|
| 62539 |
+
"learning_rate": 4.043885567735733e-06,
|
| 62540 |
+
"loss": 2.1821,
|
| 62541 |
+
"step": 102860
|
| 62542 |
+
},
|
| 62543 |
+
{
|
| 62544 |
+
"epoch": 13.89,
|
| 62545 |
+
"learning_rate": 4.034105053541298e-06,
|
| 62546 |
+
"loss": 2.1961,
|
| 62547 |
+
"step": 102870
|
| 62548 |
+
},
|
| 62549 |
+
{
|
| 62550 |
+
"epoch": 13.89,
|
| 62551 |
+
"learning_rate": 4.0243362199988085e-06,
|
| 62552 |
+
"loss": 2.2207,
|
| 62553 |
+
"step": 102880
|
| 62554 |
+
},
|
| 62555 |
+
{
|
| 62556 |
+
"epoch": 13.9,
|
| 62557 |
+
"learning_rate": 4.0145790678899555e-06,
|
| 62558 |
+
"loss": 2.2167,
|
| 62559 |
+
"step": 102890
|
| 62560 |
+
},
|
| 62561 |
+
{
|
| 62562 |
+
"epoch": 13.9,
|
| 62563 |
+
"learning_rate": 4.004833597995616e-06,
|
| 62564 |
+
"loss": 2.2107,
|
| 62565 |
+
"step": 102900
|
| 62566 |
+
},
|
| 62567 |
+
{
|
| 62568 |
+
"epoch": 13.9,
|
| 62569 |
+
"learning_rate": 3.995099811095564e-06,
|
| 62570 |
+
"loss": 2.1945,
|
| 62571 |
+
"step": 102910
|
| 62572 |
+
},
|
| 62573 |
+
{
|
| 62574 |
+
"epoch": 13.9,
|
| 62575 |
+
"learning_rate": 3.985377707968778e-06,
|
| 62576 |
+
"loss": 2.2154,
|
| 62577 |
+
"step": 102920
|
| 62578 |
+
},
|
| 62579 |
+
{
|
| 62580 |
+
"epoch": 13.9,
|
| 62581 |
+
"learning_rate": 3.975667289393236e-06,
|
| 62582 |
+
"loss": 2.2188,
|
| 62583 |
+
"step": 102930
|
| 62584 |
+
},
|
| 62585 |
+
{
|
| 62586 |
+
"epoch": 13.9,
|
| 62587 |
+
"learning_rate": 3.9659685561460154e-06,
|
| 62588 |
+
"loss": 2.2029,
|
| 62589 |
+
"step": 102940
|
| 62590 |
+
},
|
| 62591 |
+
{
|
| 62592 |
+
"epoch": 13.9,
|
| 62593 |
+
"learning_rate": 3.956281509003228e-06,
|
| 62594 |
+
"loss": 2.1711,
|
| 62595 |
+
"step": 102950
|
| 62596 |
+
},
|
| 62597 |
+
{
|
| 62598 |
+
"epoch": 13.91,
|
| 62599 |
+
"learning_rate": 3.9466061487400395e-06,
|
| 62600 |
+
"loss": 2.1978,
|
| 62601 |
+
"step": 102960
|
| 62602 |
+
},
|
| 62603 |
+
{
|
| 62604 |
+
"epoch": 13.91,
|
| 62605 |
+
"learning_rate": 3.9369424761307115e-06,
|
| 62606 |
+
"loss": 2.2257,
|
| 62607 |
+
"step": 102970
|
| 62608 |
+
},
|
| 62609 |
+
{
|
| 62610 |
+
"epoch": 13.91,
|
| 62611 |
+
"learning_rate": 3.927290491948609e-06,
|
| 62612 |
+
"loss": 2.2007,
|
| 62613 |
+
"step": 102980
|
| 62614 |
+
},
|
| 62615 |
+
{
|
| 62616 |
+
"epoch": 13.91,
|
| 62617 |
+
"learning_rate": 3.917650196966049e-06,
|
| 62618 |
+
"loss": 2.1983,
|
| 62619 |
+
"step": 102990
|
| 62620 |
+
},
|
| 62621 |
+
{
|
| 62622 |
+
"epoch": 13.91,
|
| 62623 |
+
"learning_rate": 3.90802159195453e-06,
|
| 62624 |
+
"loss": 2.2049,
|
| 62625 |
+
"step": 103000
|
| 62626 |
+
},
|
| 62627 |
+
{
|
| 62628 |
+
"epoch": 13.91,
|
| 62629 |
+
"eval_loss": 2.39996337890625,
|
| 62630 |
+
"eval_runtime": 1269.0782,
|
| 62631 |
+
"eval_samples_per_second": 59.992,
|
| 62632 |
+
"eval_steps_per_second": 5.0,
|
| 62633 |
+
"step": 103000
|
| 62634 |
}
|
| 62635 |
],
|
| 62636 |
"logging_steps": 10,
|
| 62637 |
"max_steps": 111060,
|
| 62638 |
"num_train_epochs": 15,
|
| 62639 |
"save_steps": 1000,
|
| 62640 |
+
"total_flos": 3.250975421969203e+19,
|
| 62641 |
"trial_name": null,
|
| 62642 |
"trial_params": null
|
| 62643 |
}
|