Training in progress, step 1108, checkpoint
Browse files
last-checkpoint/adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 25271744
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a69873baae6a9074b90daede1e89017f756e3658b21ab46e0a09b1325800b7d3
|
| 3 |
size 25271744
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 13685836
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:68d04871cc72386fe427ab7d1d1d39532b5862b52bd77d7f55bd35220007c172
|
| 3 |
size 13685836
|
last-checkpoint/rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14244
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9215b276ed498008f4b759f0ba01593d45ec470b6d10151dac9834a71408101d
|
| 3 |
size 14244
|
last-checkpoint/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1064
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3f13f55d1060c3c72ae6451af25abaa85fccc47c713075848b9ef644f620ac13
|
| 3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch":
|
| 5 |
"eval_steps": 277,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -5856,6 +5856,1953 @@
|
|
| 5856 |
"eval_samples_per_second": 56.135,
|
| 5857 |
"eval_steps_per_second": 28.127,
|
| 5858 |
"step": 831
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 5859 |
}
|
| 5860 |
],
|
| 5861 |
"logging_steps": 1,
|
|
@@ -5870,12 +7817,12 @@
|
|
| 5870 |
"should_evaluate": false,
|
| 5871 |
"should_log": false,
|
| 5872 |
"should_save": true,
|
| 5873 |
-
"should_training_stop":
|
| 5874 |
},
|
| 5875 |
"attributes": {}
|
| 5876 |
}
|
| 5877 |
},
|
| 5878 |
-
"total_flos": 2.
|
| 5879 |
"train_batch_size": 2,
|
| 5880 |
"trial_name": null,
|
| 5881 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 1.000451467268623,
|
| 5 |
"eval_steps": 277,
|
| 6 |
+
"global_step": 1108,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 5856 |
"eval_samples_per_second": 56.135,
|
| 5857 |
"eval_steps_per_second": 28.127,
|
| 5858 |
"step": 831
|
| 5859 |
+
},
|
| 5860 |
+
{
|
| 5861 |
+
"epoch": 0.7512415349887133,
|
| 5862 |
+
"grad_norm": 2.1794707775115967,
|
| 5863 |
+
"learning_rate": 2.9593447786519425e-05,
|
| 5864 |
+
"loss": 1.3264,
|
| 5865 |
+
"step": 832
|
| 5866 |
+
},
|
| 5867 |
+
{
|
| 5868 |
+
"epoch": 0.7521444695259594,
|
| 5869 |
+
"grad_norm": 1.6141659021377563,
|
| 5870 |
+
"learning_rate": 2.939055274225837e-05,
|
| 5871 |
+
"loss": 1.0558,
|
| 5872 |
+
"step": 833
|
| 5873 |
+
},
|
| 5874 |
+
{
|
| 5875 |
+
"epoch": 0.7530474040632055,
|
| 5876 |
+
"grad_norm": 1.9301906824111938,
|
| 5877 |
+
"learning_rate": 2.9188235737582415e-05,
|
| 5878 |
+
"loss": 1.5294,
|
| 5879 |
+
"step": 834
|
| 5880 |
+
},
|
| 5881 |
+
{
|
| 5882 |
+
"epoch": 0.7539503386004515,
|
| 5883 |
+
"grad_norm": 1.5749702453613281,
|
| 5884 |
+
"learning_rate": 2.8986498428746444e-05,
|
| 5885 |
+
"loss": 0.9791,
|
| 5886 |
+
"step": 835
|
| 5887 |
+
},
|
| 5888 |
+
{
|
| 5889 |
+
"epoch": 0.7548532731376976,
|
| 5890 |
+
"grad_norm": 1.8780022859573364,
|
| 5891 |
+
"learning_rate": 2.8785342467259568e-05,
|
| 5892 |
+
"loss": 1.0155,
|
| 5893 |
+
"step": 836
|
| 5894 |
+
},
|
| 5895 |
+
{
|
| 5896 |
+
"epoch": 0.7557562076749436,
|
| 5897 |
+
"grad_norm": 1.6968313455581665,
|
| 5898 |
+
"learning_rate": 2.8584769499871855e-05,
|
| 5899 |
+
"loss": 1.0465,
|
| 5900 |
+
"step": 837
|
| 5901 |
+
},
|
| 5902 |
+
{
|
| 5903 |
+
"epoch": 0.7566591422121897,
|
| 5904 |
+
"grad_norm": 1.3689275979995728,
|
| 5905 |
+
"learning_rate": 2.8384781168560693e-05,
|
| 5906 |
+
"loss": 1.0468,
|
| 5907 |
+
"step": 838
|
| 5908 |
+
},
|
| 5909 |
+
{
|
| 5910 |
+
"epoch": 0.7575620767494357,
|
| 5911 |
+
"grad_norm": 1.6762698888778687,
|
| 5912 |
+
"learning_rate": 2.8185379110517297e-05,
|
| 5913 |
+
"loss": 1.0407,
|
| 5914 |
+
"step": 839
|
| 5915 |
+
},
|
| 5916 |
+
{
|
| 5917 |
+
"epoch": 0.7584650112866818,
|
| 5918 |
+
"grad_norm": 1.3322452306747437,
|
| 5919 |
+
"learning_rate": 2.7986564958133565e-05,
|
| 5920 |
+
"loss": 0.9764,
|
| 5921 |
+
"step": 840
|
| 5922 |
+
},
|
| 5923 |
+
{
|
| 5924 |
+
"epoch": 0.7593679458239277,
|
| 5925 |
+
"grad_norm": 1.8644042015075684,
|
| 5926 |
+
"learning_rate": 2.7788340338988385e-05,
|
| 5927 |
+
"loss": 1.273,
|
| 5928 |
+
"step": 841
|
| 5929 |
+
},
|
| 5930 |
+
{
|
| 5931 |
+
"epoch": 0.7602708803611739,
|
| 5932 |
+
"grad_norm": 1.4388748407363892,
|
| 5933 |
+
"learning_rate": 2.759070687583456e-05,
|
| 5934 |
+
"loss": 0.8267,
|
| 5935 |
+
"step": 842
|
| 5936 |
+
},
|
| 5937 |
+
{
|
| 5938 |
+
"epoch": 0.7611738148984198,
|
| 5939 |
+
"grad_norm": 1.474744200706482,
|
| 5940 |
+
"learning_rate": 2.739366618658542e-05,
|
| 5941 |
+
"loss": 0.7948,
|
| 5942 |
+
"step": 843
|
| 5943 |
+
},
|
| 5944 |
+
{
|
| 5945 |
+
"epoch": 0.762076749435666,
|
| 5946 |
+
"grad_norm": 1.5191571712493896,
|
| 5947 |
+
"learning_rate": 2.719721988430153e-05,
|
| 5948 |
+
"loss": 0.7101,
|
| 5949 |
+
"step": 844
|
| 5950 |
+
},
|
| 5951 |
+
{
|
| 5952 |
+
"epoch": 0.7629796839729119,
|
| 5953 |
+
"grad_norm": 1.4666507244110107,
|
| 5954 |
+
"learning_rate": 2.7001369577177626e-05,
|
| 5955 |
+
"loss": 0.765,
|
| 5956 |
+
"step": 845
|
| 5957 |
+
},
|
| 5958 |
+
{
|
| 5959 |
+
"epoch": 0.763882618510158,
|
| 5960 |
+
"grad_norm": 1.5151119232177734,
|
| 5961 |
+
"learning_rate": 2.6806116868529364e-05,
|
| 5962 |
+
"loss": 0.9452,
|
| 5963 |
+
"step": 846
|
| 5964 |
+
},
|
| 5965 |
+
{
|
| 5966 |
+
"epoch": 0.764785553047404,
|
| 5967 |
+
"grad_norm": 1.1114673614501953,
|
| 5968 |
+
"learning_rate": 2.6611463356780096e-05,
|
| 5969 |
+
"loss": 0.5131,
|
| 5970 |
+
"step": 847
|
| 5971 |
+
},
|
| 5972 |
+
{
|
| 5973 |
+
"epoch": 0.7656884875846501,
|
| 5974 |
+
"grad_norm": 1.5805968046188354,
|
| 5975 |
+
"learning_rate": 2.6417410635448013e-05,
|
| 5976 |
+
"loss": 0.9875,
|
| 5977 |
+
"step": 848
|
| 5978 |
+
},
|
| 5979 |
+
{
|
| 5980 |
+
"epoch": 0.7665914221218961,
|
| 5981 |
+
"grad_norm": 1.464408278465271,
|
| 5982 |
+
"learning_rate": 2.622396029313293e-05,
|
| 5983 |
+
"loss": 1.0605,
|
| 5984 |
+
"step": 849
|
| 5985 |
+
},
|
| 5986 |
+
{
|
| 5987 |
+
"epoch": 0.7674943566591422,
|
| 5988 |
+
"grad_norm": 1.7082327604293823,
|
| 5989 |
+
"learning_rate": 2.6031113913503337e-05,
|
| 5990 |
+
"loss": 1.1644,
|
| 5991 |
+
"step": 850
|
| 5992 |
+
},
|
| 5993 |
+
{
|
| 5994 |
+
"epoch": 0.7683972911963882,
|
| 5995 |
+
"grad_norm": 1.4891247749328613,
|
| 5996 |
+
"learning_rate": 2.5838873075283343e-05,
|
| 5997 |
+
"loss": 0.934,
|
| 5998 |
+
"step": 851
|
| 5999 |
+
},
|
| 6000 |
+
{
|
| 6001 |
+
"epoch": 0.7693002257336343,
|
| 6002 |
+
"grad_norm": 1.7040550708770752,
|
| 6003 |
+
"learning_rate": 2.564723935223995e-05,
|
| 6004 |
+
"loss": 1.1936,
|
| 6005 |
+
"step": 852
|
| 6006 |
+
},
|
| 6007 |
+
{
|
| 6008 |
+
"epoch": 0.7702031602708803,
|
| 6009 |
+
"grad_norm": 1.9601792097091675,
|
| 6010 |
+
"learning_rate": 2.5456214313170002e-05,
|
| 6011 |
+
"loss": 1.5839,
|
| 6012 |
+
"step": 853
|
| 6013 |
+
},
|
| 6014 |
+
{
|
| 6015 |
+
"epoch": 0.7711060948081264,
|
| 6016 |
+
"grad_norm": 1.7376189231872559,
|
| 6017 |
+
"learning_rate": 2.5265799521887345e-05,
|
| 6018 |
+
"loss": 1.4663,
|
| 6019 |
+
"step": 854
|
| 6020 |
+
},
|
| 6021 |
+
{
|
| 6022 |
+
"epoch": 0.7720090293453724,
|
| 6023 |
+
"grad_norm": 1.6315890550613403,
|
| 6024 |
+
"learning_rate": 2.5075996537210133e-05,
|
| 6025 |
+
"loss": 1.2038,
|
| 6026 |
+
"step": 855
|
| 6027 |
+
},
|
| 6028 |
+
{
|
| 6029 |
+
"epoch": 0.7729119638826185,
|
| 6030 |
+
"grad_norm": 1.7750074863433838,
|
| 6031 |
+
"learning_rate": 2.4886806912948035e-05,
|
| 6032 |
+
"loss": 1.2161,
|
| 6033 |
+
"step": 856
|
| 6034 |
+
},
|
| 6035 |
+
{
|
| 6036 |
+
"epoch": 0.7738148984198645,
|
| 6037 |
+
"grad_norm": 1.5756460428237915,
|
| 6038 |
+
"learning_rate": 2.4698232197889337e-05,
|
| 6039 |
+
"loss": 0.9405,
|
| 6040 |
+
"step": 857
|
| 6041 |
+
},
|
| 6042 |
+
{
|
| 6043 |
+
"epoch": 0.7747178329571106,
|
| 6044 |
+
"grad_norm": 1.7079740762710571,
|
| 6045 |
+
"learning_rate": 2.451027393578863e-05,
|
| 6046 |
+
"loss": 1.3505,
|
| 6047 |
+
"step": 858
|
| 6048 |
+
},
|
| 6049 |
+
{
|
| 6050 |
+
"epoch": 0.7756207674943567,
|
| 6051 |
+
"grad_norm": 1.0900604724884033,
|
| 6052 |
+
"learning_rate": 2.4322933665353776e-05,
|
| 6053 |
+
"loss": 0.8443,
|
| 6054 |
+
"step": 859
|
| 6055 |
+
},
|
| 6056 |
+
{
|
| 6057 |
+
"epoch": 0.7765237020316027,
|
| 6058 |
+
"grad_norm": 1.6753469705581665,
|
| 6059 |
+
"learning_rate": 2.413621292023349e-05,
|
| 6060 |
+
"loss": 1.2974,
|
| 6061 |
+
"step": 860
|
| 6062 |
+
},
|
| 6063 |
+
{
|
| 6064 |
+
"epoch": 0.7774266365688488,
|
| 6065 |
+
"grad_norm": 1.7383633852005005,
|
| 6066 |
+
"learning_rate": 2.3950113229004968e-05,
|
| 6067 |
+
"loss": 1.2505,
|
| 6068 |
+
"step": 861
|
| 6069 |
+
},
|
| 6070 |
+
{
|
| 6071 |
+
"epoch": 0.7783295711060948,
|
| 6072 |
+
"grad_norm": 1.237730860710144,
|
| 6073 |
+
"learning_rate": 2.3764636115160978e-05,
|
| 6074 |
+
"loss": 0.8125,
|
| 6075 |
+
"step": 862
|
| 6076 |
+
},
|
| 6077 |
+
{
|
| 6078 |
+
"epoch": 0.7792325056433409,
|
| 6079 |
+
"grad_norm": 1.43095064163208,
|
| 6080 |
+
"learning_rate": 2.3579783097097695e-05,
|
| 6081 |
+
"loss": 0.6666,
|
| 6082 |
+
"step": 863
|
| 6083 |
+
},
|
| 6084 |
+
{
|
| 6085 |
+
"epoch": 0.7801354401805869,
|
| 6086 |
+
"grad_norm": 1.8008005619049072,
|
| 6087 |
+
"learning_rate": 2.339555568810221e-05,
|
| 6088 |
+
"loss": 1.2575,
|
| 6089 |
+
"step": 864
|
| 6090 |
+
},
|
| 6091 |
+
{
|
| 6092 |
+
"epoch": 0.781038374717833,
|
| 6093 |
+
"grad_norm": 1.721174716949463,
|
| 6094 |
+
"learning_rate": 2.3211955396340002e-05,
|
| 6095 |
+
"loss": 1.4017,
|
| 6096 |
+
"step": 865
|
| 6097 |
+
},
|
| 6098 |
+
{
|
| 6099 |
+
"epoch": 0.781941309255079,
|
| 6100 |
+
"grad_norm": 1.8302357196807861,
|
| 6101 |
+
"learning_rate": 2.302898372484278e-05,
|
| 6102 |
+
"loss": 1.3493,
|
| 6103 |
+
"step": 866
|
| 6104 |
+
},
|
| 6105 |
+
{
|
| 6106 |
+
"epoch": 0.7828442437923251,
|
| 6107 |
+
"grad_norm": 1.9394512176513672,
|
| 6108 |
+
"learning_rate": 2.2846642171496102e-05,
|
| 6109 |
+
"loss": 1.5522,
|
| 6110 |
+
"step": 867
|
| 6111 |
+
},
|
| 6112 |
+
{
|
| 6113 |
+
"epoch": 0.7837471783295711,
|
| 6114 |
+
"grad_norm": 1.74718177318573,
|
| 6115 |
+
"learning_rate": 2.2664932229027024e-05,
|
| 6116 |
+
"loss": 1.3953,
|
| 6117 |
+
"step": 868
|
| 6118 |
+
},
|
| 6119 |
+
{
|
| 6120 |
+
"epoch": 0.7846501128668172,
|
| 6121 |
+
"grad_norm": 1.7172614336013794,
|
| 6122 |
+
"learning_rate": 2.2483855384992026e-05,
|
| 6123 |
+
"loss": 1.1969,
|
| 6124 |
+
"step": 869
|
| 6125 |
+
},
|
| 6126 |
+
{
|
| 6127 |
+
"epoch": 0.7855530474040632,
|
| 6128 |
+
"grad_norm": 1.8553889989852905,
|
| 6129 |
+
"learning_rate": 2.230341312176476e-05,
|
| 6130 |
+
"loss": 1.0929,
|
| 6131 |
+
"step": 870
|
| 6132 |
+
},
|
| 6133 |
+
{
|
| 6134 |
+
"epoch": 0.7864559819413093,
|
| 6135 |
+
"grad_norm": 1.6518503427505493,
|
| 6136 |
+
"learning_rate": 2.2123606916523953e-05,
|
| 6137 |
+
"loss": 1.1453,
|
| 6138 |
+
"step": 871
|
| 6139 |
+
},
|
| 6140 |
+
{
|
| 6141 |
+
"epoch": 0.7873589164785553,
|
| 6142 |
+
"grad_norm": 1.5877022743225098,
|
| 6143 |
+
"learning_rate": 2.1944438241241184e-05,
|
| 6144 |
+
"loss": 1.149,
|
| 6145 |
+
"step": 872
|
| 6146 |
+
},
|
| 6147 |
+
{
|
| 6148 |
+
"epoch": 0.7882618510158014,
|
| 6149 |
+
"grad_norm": 1.6167192459106445,
|
| 6150 |
+
"learning_rate": 2.1765908562669035e-05,
|
| 6151 |
+
"loss": 0.9507,
|
| 6152 |
+
"step": 873
|
| 6153 |
+
},
|
| 6154 |
+
{
|
| 6155 |
+
"epoch": 0.7891647855530474,
|
| 6156 |
+
"grad_norm": 1.4243950843811035,
|
| 6157 |
+
"learning_rate": 2.1588019342328968e-05,
|
| 6158 |
+
"loss": 0.9338,
|
| 6159 |
+
"step": 874
|
| 6160 |
+
},
|
| 6161 |
+
{
|
| 6162 |
+
"epoch": 0.7900677200902935,
|
| 6163 |
+
"grad_norm": 1.5820962190628052,
|
| 6164 |
+
"learning_rate": 2.1410772036499326e-05,
|
| 6165 |
+
"loss": 0.9402,
|
| 6166 |
+
"step": 875
|
| 6167 |
+
},
|
| 6168 |
+
{
|
| 6169 |
+
"epoch": 0.7909706546275395,
|
| 6170 |
+
"grad_norm": 1.3775403499603271,
|
| 6171 |
+
"learning_rate": 2.1234168096203512e-05,
|
| 6172 |
+
"loss": 0.6707,
|
| 6173 |
+
"step": 876
|
| 6174 |
+
},
|
| 6175 |
+
{
|
| 6176 |
+
"epoch": 0.7918735891647856,
|
| 6177 |
+
"grad_norm": 1.983003854751587,
|
| 6178 |
+
"learning_rate": 2.1058208967198045e-05,
|
| 6179 |
+
"loss": 1.4335,
|
| 6180 |
+
"step": 877
|
| 6181 |
+
},
|
| 6182 |
+
{
|
| 6183 |
+
"epoch": 0.7927765237020316,
|
| 6184 |
+
"grad_norm": 1.9397960901260376,
|
| 6185 |
+
"learning_rate": 2.088289608996071e-05,
|
| 6186 |
+
"loss": 1.1648,
|
| 6187 |
+
"step": 878
|
| 6188 |
+
},
|
| 6189 |
+
{
|
| 6190 |
+
"epoch": 0.7936794582392777,
|
| 6191 |
+
"grad_norm": 1.6338427066802979,
|
| 6192 |
+
"learning_rate": 2.0708230899678838e-05,
|
| 6193 |
+
"loss": 1.4529,
|
| 6194 |
+
"step": 879
|
| 6195 |
+
},
|
| 6196 |
+
{
|
| 6197 |
+
"epoch": 0.7945823927765236,
|
| 6198 |
+
"grad_norm": 1.9832733869552612,
|
| 6199 |
+
"learning_rate": 2.0534214826237484e-05,
|
| 6200 |
+
"loss": 1.7312,
|
| 6201 |
+
"step": 880
|
| 6202 |
+
},
|
| 6203 |
+
{
|
| 6204 |
+
"epoch": 0.7954853273137698,
|
| 6205 |
+
"grad_norm": 1.142454981803894,
|
| 6206 |
+
"learning_rate": 2.0360849294207808e-05,
|
| 6207 |
+
"loss": 0.6494,
|
| 6208 |
+
"step": 881
|
| 6209 |
+
},
|
| 6210 |
+
{
|
| 6211 |
+
"epoch": 0.7963882618510159,
|
| 6212 |
+
"grad_norm": 1.638196587562561,
|
| 6213 |
+
"learning_rate": 2.0188135722835233e-05,
|
| 6214 |
+
"loss": 1.3321,
|
| 6215 |
+
"step": 882
|
| 6216 |
+
},
|
| 6217 |
+
{
|
| 6218 |
+
"epoch": 0.7972911963882618,
|
| 6219 |
+
"grad_norm": 1.6472254991531372,
|
| 6220 |
+
"learning_rate": 2.0016075526028065e-05,
|
| 6221 |
+
"loss": 1.3264,
|
| 6222 |
+
"step": 883
|
| 6223 |
+
},
|
| 6224 |
+
{
|
| 6225 |
+
"epoch": 0.798194130925508,
|
| 6226 |
+
"grad_norm": 2.0473620891571045,
|
| 6227 |
+
"learning_rate": 1.9844670112345786e-05,
|
| 6228 |
+
"loss": 1.7047,
|
| 6229 |
+
"step": 884
|
| 6230 |
+
},
|
| 6231 |
+
{
|
| 6232 |
+
"epoch": 0.7990970654627539,
|
| 6233 |
+
"grad_norm": 1.5241625308990479,
|
| 6234 |
+
"learning_rate": 1.9673920884987463e-05,
|
| 6235 |
+
"loss": 1.1379,
|
| 6236 |
+
"step": 885
|
| 6237 |
+
},
|
| 6238 |
+
{
|
| 6239 |
+
"epoch": 0.8,
|
| 6240 |
+
"grad_norm": 1.3565895557403564,
|
| 6241 |
+
"learning_rate": 1.9503829241780412e-05,
|
| 6242 |
+
"loss": 0.8124,
|
| 6243 |
+
"step": 886
|
| 6244 |
+
},
|
| 6245 |
+
{
|
| 6246 |
+
"epoch": 0.800902934537246,
|
| 6247 |
+
"grad_norm": 1.800199270248413,
|
| 6248 |
+
"learning_rate": 1.9334396575168688e-05,
|
| 6249 |
+
"loss": 1.3687,
|
| 6250 |
+
"step": 887
|
| 6251 |
+
},
|
| 6252 |
+
{
|
| 6253 |
+
"epoch": 0.8018058690744921,
|
| 6254 |
+
"grad_norm": 1.655916452407837,
|
| 6255 |
+
"learning_rate": 1.9165624272201567e-05,
|
| 6256 |
+
"loss": 1.0146,
|
| 6257 |
+
"step": 888
|
| 6258 |
+
},
|
| 6259 |
+
{
|
| 6260 |
+
"epoch": 0.8027088036117381,
|
| 6261 |
+
"grad_norm": 1.7925070524215698,
|
| 6262 |
+
"learning_rate": 1.8997513714522487e-05,
|
| 6263 |
+
"loss": 1.0542,
|
| 6264 |
+
"step": 889
|
| 6265 |
+
},
|
| 6266 |
+
{
|
| 6267 |
+
"epoch": 0.8036117381489842,
|
| 6268 |
+
"grad_norm": 1.5180613994598389,
|
| 6269 |
+
"learning_rate": 1.8830066278357393e-05,
|
| 6270 |
+
"loss": 0.7936,
|
| 6271 |
+
"step": 890
|
| 6272 |
+
},
|
| 6273 |
+
{
|
| 6274 |
+
"epoch": 0.8045146726862302,
|
| 6275 |
+
"grad_norm": 1.3471791744232178,
|
| 6276 |
+
"learning_rate": 1.8663283334503633e-05,
|
| 6277 |
+
"loss": 1.0296,
|
| 6278 |
+
"step": 891
|
| 6279 |
+
},
|
| 6280 |
+
{
|
| 6281 |
+
"epoch": 0.8054176072234763,
|
| 6282 |
+
"grad_norm": 1.8048678636550903,
|
| 6283 |
+
"learning_rate": 1.8497166248318876e-05,
|
| 6284 |
+
"loss": 1.4417,
|
| 6285 |
+
"step": 892
|
| 6286 |
+
},
|
| 6287 |
+
{
|
| 6288 |
+
"epoch": 0.8063205417607223,
|
| 6289 |
+
"grad_norm": 1.628916621208191,
|
| 6290 |
+
"learning_rate": 1.833171637970963e-05,
|
| 6291 |
+
"loss": 1.1107,
|
| 6292 |
+
"step": 893
|
| 6293 |
+
},
|
| 6294 |
+
{
|
| 6295 |
+
"epoch": 0.8072234762979684,
|
| 6296 |
+
"grad_norm": 1.8453953266143799,
|
| 6297 |
+
"learning_rate": 1.8166935083120352e-05,
|
| 6298 |
+
"loss": 1.4268,
|
| 6299 |
+
"step": 894
|
| 6300 |
+
},
|
| 6301 |
+
{
|
| 6302 |
+
"epoch": 0.8081264108352144,
|
| 6303 |
+
"grad_norm": 1.3454231023788452,
|
| 6304 |
+
"learning_rate": 1.8002823707522297e-05,
|
| 6305 |
+
"loss": 1.0191,
|
| 6306 |
+
"step": 895
|
| 6307 |
+
},
|
| 6308 |
+
{
|
| 6309 |
+
"epoch": 0.8090293453724605,
|
| 6310 |
+
"grad_norm": 1.2696763277053833,
|
| 6311 |
+
"learning_rate": 1.783938359640238e-05,
|
| 6312 |
+
"loss": 0.6779,
|
| 6313 |
+
"step": 896
|
| 6314 |
+
},
|
| 6315 |
+
{
|
| 6316 |
+
"epoch": 0.8099322799097065,
|
| 6317 |
+
"grad_norm": 1.6980637311935425,
|
| 6318 |
+
"learning_rate": 1.7676616087752308e-05,
|
| 6319 |
+
"loss": 0.841,
|
| 6320 |
+
"step": 897
|
| 6321 |
+
},
|
| 6322 |
+
{
|
| 6323 |
+
"epoch": 0.8108352144469526,
|
| 6324 |
+
"grad_norm": 1.556020975112915,
|
| 6325 |
+
"learning_rate": 1.7514522514057553e-05,
|
| 6326 |
+
"loss": 1.0395,
|
| 6327 |
+
"step": 898
|
| 6328 |
+
},
|
| 6329 |
+
{
|
| 6330 |
+
"epoch": 0.8117381489841986,
|
| 6331 |
+
"grad_norm": 1.463173747062683,
|
| 6332 |
+
"learning_rate": 1.7353104202286517e-05,
|
| 6333 |
+
"loss": 0.9776,
|
| 6334 |
+
"step": 899
|
| 6335 |
+
},
|
| 6336 |
+
{
|
| 6337 |
+
"epoch": 0.8126410835214447,
|
| 6338 |
+
"grad_norm": 1.660776972770691,
|
| 6339 |
+
"learning_rate": 1.719236247387951e-05,
|
| 6340 |
+
"loss": 0.9951,
|
| 6341 |
+
"step": 900
|
| 6342 |
+
},
|
| 6343 |
+
{
|
| 6344 |
+
"epoch": 0.8135440180586907,
|
| 6345 |
+
"grad_norm": 1.5283406972885132,
|
| 6346 |
+
"learning_rate": 1.703229864473811e-05,
|
| 6347 |
+
"loss": 1.1435,
|
| 6348 |
+
"step": 901
|
| 6349 |
+
},
|
| 6350 |
+
{
|
| 6351 |
+
"epoch": 0.8144469525959368,
|
| 6352 |
+
"grad_norm": 1.4751852750778198,
|
| 6353 |
+
"learning_rate": 1.687291402521434e-05,
|
| 6354 |
+
"loss": 0.9997,
|
| 6355 |
+
"step": 902
|
| 6356 |
+
},
|
| 6357 |
+
{
|
| 6358 |
+
"epoch": 0.8153498871331828,
|
| 6359 |
+
"grad_norm": 1.2015782594680786,
|
| 6360 |
+
"learning_rate": 1.671420992009981e-05,
|
| 6361 |
+
"loss": 0.6096,
|
| 6362 |
+
"step": 903
|
| 6363 |
+
},
|
| 6364 |
+
{
|
| 6365 |
+
"epoch": 0.8162528216704289,
|
| 6366 |
+
"grad_norm": 1.6067993640899658,
|
| 6367 |
+
"learning_rate": 1.6556187628615273e-05,
|
| 6368 |
+
"loss": 1.0961,
|
| 6369 |
+
"step": 904
|
| 6370 |
+
},
|
| 6371 |
+
{
|
| 6372 |
+
"epoch": 0.8171557562076749,
|
| 6373 |
+
"grad_norm": 1.8478807210922241,
|
| 6374 |
+
"learning_rate": 1.6398848444399794e-05,
|
| 6375 |
+
"loss": 1.2024,
|
| 6376 |
+
"step": 905
|
| 6377 |
+
},
|
| 6378 |
+
{
|
| 6379 |
+
"epoch": 0.818058690744921,
|
| 6380 |
+
"grad_norm": 1.8460716009140015,
|
| 6381 |
+
"learning_rate": 1.6242193655500183e-05,
|
| 6382 |
+
"loss": 1.631,
|
| 6383 |
+
"step": 906
|
| 6384 |
+
},
|
| 6385 |
+
{
|
| 6386 |
+
"epoch": 0.8189616252821671,
|
| 6387 |
+
"grad_norm": 1.6105166673660278,
|
| 6388 |
+
"learning_rate": 1.608622454436062e-05,
|
| 6389 |
+
"loss": 1.194,
|
| 6390 |
+
"step": 907
|
| 6391 |
+
},
|
| 6392 |
+
{
|
| 6393 |
+
"epoch": 0.8198645598194131,
|
| 6394 |
+
"grad_norm": 1.4500304460525513,
|
| 6395 |
+
"learning_rate": 1.593094238781191e-05,
|
| 6396 |
+
"loss": 1.0055,
|
| 6397 |
+
"step": 908
|
| 6398 |
+
},
|
| 6399 |
+
{
|
| 6400 |
+
"epoch": 0.8207674943566592,
|
| 6401 |
+
"grad_norm": 1.8839186429977417,
|
| 6402 |
+
"learning_rate": 1.577634845706116e-05,
|
| 6403 |
+
"loss": 1.4827,
|
| 6404 |
+
"step": 909
|
| 6405 |
+
},
|
| 6406 |
+
{
|
| 6407 |
+
"epoch": 0.8216704288939052,
|
| 6408 |
+
"grad_norm": 1.7479982376098633,
|
| 6409 |
+
"learning_rate": 1.562244401768144e-05,
|
| 6410 |
+
"loss": 1.284,
|
| 6411 |
+
"step": 910
|
| 6412 |
+
},
|
| 6413 |
+
{
|
| 6414 |
+
"epoch": 0.8225733634311513,
|
| 6415 |
+
"grad_norm": 1.308475375175476,
|
| 6416 |
+
"learning_rate": 1.546923032960127e-05,
|
| 6417 |
+
"loss": 0.8433,
|
| 6418 |
+
"step": 911
|
| 6419 |
+
},
|
| 6420 |
+
{
|
| 6421 |
+
"epoch": 0.8234762979683973,
|
| 6422 |
+
"grad_norm": 1.3568767309188843,
|
| 6423 |
+
"learning_rate": 1.5316708647094446e-05,
|
| 6424 |
+
"loss": 0.9064,
|
| 6425 |
+
"step": 912
|
| 6426 |
+
},
|
| 6427 |
+
{
|
| 6428 |
+
"epoch": 0.8243792325056434,
|
| 6429 |
+
"grad_norm": 1.556696891784668,
|
| 6430 |
+
"learning_rate": 1.5164880218769618e-05,
|
| 6431 |
+
"loss": 0.8663,
|
| 6432 |
+
"step": 913
|
| 6433 |
+
},
|
| 6434 |
+
{
|
| 6435 |
+
"epoch": 0.8252821670428894,
|
| 6436 |
+
"grad_norm": 1.2990261316299438,
|
| 6437 |
+
"learning_rate": 1.5013746287560226e-05,
|
| 6438 |
+
"loss": 1.0119,
|
| 6439 |
+
"step": 914
|
| 6440 |
+
},
|
| 6441 |
+
{
|
| 6442 |
+
"epoch": 0.8261851015801355,
|
| 6443 |
+
"grad_norm": 1.4363588094711304,
|
| 6444 |
+
"learning_rate": 1.4863308090714256e-05,
|
| 6445 |
+
"loss": 0.8974,
|
| 6446 |
+
"step": 915
|
| 6447 |
+
},
|
| 6448 |
+
{
|
| 6449 |
+
"epoch": 0.8270880361173815,
|
| 6450 |
+
"grad_norm": 1.665542721748352,
|
| 6451 |
+
"learning_rate": 1.4713566859784045e-05,
|
| 6452 |
+
"loss": 1.2443,
|
| 6453 |
+
"step": 916
|
| 6454 |
+
},
|
| 6455 |
+
{
|
| 6456 |
+
"epoch": 0.8279909706546276,
|
| 6457 |
+
"grad_norm": 1.580439805984497,
|
| 6458 |
+
"learning_rate": 1.4564523820616326e-05,
|
| 6459 |
+
"loss": 0.7228,
|
| 6460 |
+
"step": 917
|
| 6461 |
+
},
|
| 6462 |
+
{
|
| 6463 |
+
"epoch": 0.8288939051918736,
|
| 6464 |
+
"grad_norm": 1.664021611213684,
|
| 6465 |
+
"learning_rate": 1.4416180193342144e-05,
|
| 6466 |
+
"loss": 1.1174,
|
| 6467 |
+
"step": 918
|
| 6468 |
+
},
|
| 6469 |
+
{
|
| 6470 |
+
"epoch": 0.8297968397291197,
|
| 6471 |
+
"grad_norm": 1.0893442630767822,
|
| 6472 |
+
"learning_rate": 1.426853719236676e-05,
|
| 6473 |
+
"loss": 0.488,
|
| 6474 |
+
"step": 919
|
| 6475 |
+
},
|
| 6476 |
+
{
|
| 6477 |
+
"epoch": 0.8306997742663657,
|
| 6478 |
+
"grad_norm": 1.5953645706176758,
|
| 6479 |
+
"learning_rate": 1.4121596026359951e-05,
|
| 6480 |
+
"loss": 1.3008,
|
| 6481 |
+
"step": 920
|
| 6482 |
+
},
|
| 6483 |
+
{
|
| 6484 |
+
"epoch": 0.8316027088036118,
|
| 6485 |
+
"grad_norm": 1.5674923658370972,
|
| 6486 |
+
"learning_rate": 1.3975357898245812e-05,
|
| 6487 |
+
"loss": 0.9816,
|
| 6488 |
+
"step": 921
|
| 6489 |
+
},
|
| 6490 |
+
{
|
| 6491 |
+
"epoch": 0.8325056433408577,
|
| 6492 |
+
"grad_norm": 1.0914887189865112,
|
| 6493 |
+
"learning_rate": 1.3829824005193181e-05,
|
| 6494 |
+
"loss": 0.4505,
|
| 6495 |
+
"step": 922
|
| 6496 |
+
},
|
| 6497 |
+
{
|
| 6498 |
+
"epoch": 0.8334085778781039,
|
| 6499 |
+
"grad_norm": 1.6878334283828735,
|
| 6500 |
+
"learning_rate": 1.3684995538605683e-05,
|
| 6501 |
+
"loss": 1.1872,
|
| 6502 |
+
"step": 923
|
| 6503 |
+
},
|
| 6504 |
+
{
|
| 6505 |
+
"epoch": 0.8343115124153498,
|
| 6506 |
+
"grad_norm": 1.5240914821624756,
|
| 6507 |
+
"learning_rate": 1.3540873684111976e-05,
|
| 6508 |
+
"loss": 0.8582,
|
| 6509 |
+
"step": 924
|
| 6510 |
+
},
|
| 6511 |
+
{
|
| 6512 |
+
"epoch": 0.835214446952596,
|
| 6513 |
+
"grad_norm": 1.0279604196548462,
|
| 6514 |
+
"learning_rate": 1.339745962155613e-05,
|
| 6515 |
+
"loss": 0.2322,
|
| 6516 |
+
"step": 925
|
| 6517 |
+
},
|
| 6518 |
+
{
|
| 6519 |
+
"epoch": 0.8361173814898419,
|
| 6520 |
+
"grad_norm": 1.7582857608795166,
|
| 6521 |
+
"learning_rate": 1.3254754524987956e-05,
|
| 6522 |
+
"loss": 1.4453,
|
| 6523 |
+
"step": 926
|
| 6524 |
+
},
|
| 6525 |
+
{
|
| 6526 |
+
"epoch": 0.837020316027088,
|
| 6527 |
+
"grad_norm": 1.7181614637374878,
|
| 6528 |
+
"learning_rate": 1.3112759562653265e-05,
|
| 6529 |
+
"loss": 1.1686,
|
| 6530 |
+
"step": 927
|
| 6531 |
+
},
|
| 6532 |
+
{
|
| 6533 |
+
"epoch": 0.837923250564334,
|
| 6534 |
+
"grad_norm": 1.7236977815628052,
|
| 6535 |
+
"learning_rate": 1.2971475896984475e-05,
|
| 6536 |
+
"loss": 1.0686,
|
| 6537 |
+
"step": 928
|
| 6538 |
+
},
|
| 6539 |
+
{
|
| 6540 |
+
"epoch": 0.8388261851015801,
|
| 6541 |
+
"grad_norm": 1.8403332233428955,
|
| 6542 |
+
"learning_rate": 1.2830904684591017e-05,
|
| 6543 |
+
"loss": 1.5042,
|
| 6544 |
+
"step": 929
|
| 6545 |
+
},
|
| 6546 |
+
{
|
| 6547 |
+
"epoch": 0.8397291196388262,
|
| 6548 |
+
"grad_norm": 1.3492463827133179,
|
| 6549 |
+
"learning_rate": 1.2691047076249851e-05,
|
| 6550 |
+
"loss": 0.599,
|
| 6551 |
+
"step": 930
|
| 6552 |
+
},
|
| 6553 |
+
{
|
| 6554 |
+
"epoch": 0.8406320541760722,
|
| 6555 |
+
"grad_norm": 1.0862035751342773,
|
| 6556 |
+
"learning_rate": 1.2551904216896037e-05,
|
| 6557 |
+
"loss": 0.6616,
|
| 6558 |
+
"step": 931
|
| 6559 |
+
},
|
| 6560 |
+
{
|
| 6561 |
+
"epoch": 0.8415349887133183,
|
| 6562 |
+
"grad_norm": 1.1652106046676636,
|
| 6563 |
+
"learning_rate": 1.2413477245613437e-05,
|
| 6564 |
+
"loss": 0.6117,
|
| 6565 |
+
"step": 932
|
| 6566 |
+
},
|
| 6567 |
+
{
|
| 6568 |
+
"epoch": 0.8424379232505643,
|
| 6569 |
+
"grad_norm": 1.392179250717163,
|
| 6570 |
+
"learning_rate": 1.2275767295625363e-05,
|
| 6571 |
+
"loss": 1.2104,
|
| 6572 |
+
"step": 933
|
| 6573 |
+
},
|
| 6574 |
+
{
|
| 6575 |
+
"epoch": 0.8433408577878104,
|
| 6576 |
+
"grad_norm": 1.479121208190918,
|
| 6577 |
+
"learning_rate": 1.2138775494285182e-05,
|
| 6578 |
+
"loss": 0.6797,
|
| 6579 |
+
"step": 934
|
| 6580 |
+
},
|
| 6581 |
+
{
|
| 6582 |
+
"epoch": 0.8442437923250564,
|
| 6583 |
+
"grad_norm": 1.5013456344604492,
|
| 6584 |
+
"learning_rate": 1.2002502963067275e-05,
|
| 6585 |
+
"loss": 0.8688,
|
| 6586 |
+
"step": 935
|
| 6587 |
+
},
|
| 6588 |
+
{
|
| 6589 |
+
"epoch": 0.8451467268623025,
|
| 6590 |
+
"grad_norm": 1.5734107494354248,
|
| 6591 |
+
"learning_rate": 1.1866950817557742e-05,
|
| 6592 |
+
"loss": 1.0534,
|
| 6593 |
+
"step": 936
|
| 6594 |
+
},
|
| 6595 |
+
{
|
| 6596 |
+
"epoch": 0.8460496613995485,
|
| 6597 |
+
"grad_norm": 1.5367547273635864,
|
| 6598 |
+
"learning_rate": 1.1732120167445248e-05,
|
| 6599 |
+
"loss": 0.6457,
|
| 6600 |
+
"step": 937
|
| 6601 |
+
},
|
| 6602 |
+
{
|
| 6603 |
+
"epoch": 0.8469525959367946,
|
| 6604 |
+
"grad_norm": 1.51642906665802,
|
| 6605 |
+
"learning_rate": 1.1598012116512092e-05,
|
| 6606 |
+
"loss": 1.3725,
|
| 6607 |
+
"step": 938
|
| 6608 |
+
},
|
| 6609 |
+
{
|
| 6610 |
+
"epoch": 0.8478555304740406,
|
| 6611 |
+
"grad_norm": 2.061697244644165,
|
| 6612 |
+
"learning_rate": 1.1464627762624946e-05,
|
| 6613 |
+
"loss": 1.3527,
|
| 6614 |
+
"step": 939
|
| 6615 |
+
},
|
| 6616 |
+
{
|
| 6617 |
+
"epoch": 0.8487584650112867,
|
| 6618 |
+
"grad_norm": 1.8436322212219238,
|
| 6619 |
+
"learning_rate": 1.1331968197725984e-05,
|
| 6620 |
+
"loss": 0.9779,
|
| 6621 |
+
"step": 940
|
| 6622 |
+
},
|
| 6623 |
+
{
|
| 6624 |
+
"epoch": 0.8496613995485327,
|
| 6625 |
+
"grad_norm": 1.7328407764434814,
|
| 6626 |
+
"learning_rate": 1.1200034507824086e-05,
|
| 6627 |
+
"loss": 1.0868,
|
| 6628 |
+
"step": 941
|
| 6629 |
+
},
|
| 6630 |
+
{
|
| 6631 |
+
"epoch": 0.8505643340857788,
|
| 6632 |
+
"grad_norm": 1.7387075424194336,
|
| 6633 |
+
"learning_rate": 1.1068827772985646e-05,
|
| 6634 |
+
"loss": 1.1025,
|
| 6635 |
+
"step": 942
|
| 6636 |
+
},
|
| 6637 |
+
{
|
| 6638 |
+
"epoch": 0.8514672686230248,
|
| 6639 |
+
"grad_norm": 1.6153181791305542,
|
| 6640 |
+
"learning_rate": 1.0938349067325959e-05,
|
| 6641 |
+
"loss": 0.9804,
|
| 6642 |
+
"step": 943
|
| 6643 |
+
},
|
| 6644 |
+
{
|
| 6645 |
+
"epoch": 0.8523702031602709,
|
| 6646 |
+
"grad_norm": 1.6639015674591064,
|
| 6647 |
+
"learning_rate": 1.0808599459000368e-05,
|
| 6648 |
+
"loss": 1.108,
|
| 6649 |
+
"step": 944
|
| 6650 |
+
},
|
| 6651 |
+
{
|
| 6652 |
+
"epoch": 0.8532731376975169,
|
| 6653 |
+
"grad_norm": 1.188910961151123,
|
| 6654 |
+
"learning_rate": 1.0679580010195445e-05,
|
| 6655 |
+
"loss": 0.5745,
|
| 6656 |
+
"step": 945
|
| 6657 |
+
},
|
| 6658 |
+
{
|
| 6659 |
+
"epoch": 0.854176072234763,
|
| 6660 |
+
"grad_norm": 1.5239681005477905,
|
| 6661 |
+
"learning_rate": 1.0551291777120464e-05,
|
| 6662 |
+
"loss": 0.9202,
|
| 6663 |
+
"step": 946
|
| 6664 |
+
},
|
| 6665 |
+
{
|
| 6666 |
+
"epoch": 0.855079006772009,
|
| 6667 |
+
"grad_norm": 2.2678442001342773,
|
| 6668 |
+
"learning_rate": 1.0423735809998514e-05,
|
| 6669 |
+
"loss": 0.84,
|
| 6670 |
+
"step": 947
|
| 6671 |
+
},
|
| 6672 |
+
{
|
| 6673 |
+
"epoch": 0.8559819413092551,
|
| 6674 |
+
"grad_norm": 1.5488260984420776,
|
| 6675 |
+
"learning_rate": 1.0296913153058163e-05,
|
| 6676 |
+
"loss": 0.9499,
|
| 6677 |
+
"step": 948
|
| 6678 |
+
},
|
| 6679 |
+
{
|
| 6680 |
+
"epoch": 0.8568848758465011,
|
| 6681 |
+
"grad_norm": 1.705390214920044,
|
| 6682 |
+
"learning_rate": 1.0170824844524728e-05,
|
| 6683 |
+
"loss": 1.3049,
|
| 6684 |
+
"step": 949
|
| 6685 |
+
},
|
| 6686 |
+
{
|
| 6687 |
+
"epoch": 0.8577878103837472,
|
| 6688 |
+
"grad_norm": 1.4453303813934326,
|
| 6689 |
+
"learning_rate": 1.0045471916611782e-05,
|
| 6690 |
+
"loss": 0.8621,
|
| 6691 |
+
"step": 950
|
| 6692 |
+
},
|
| 6693 |
+
{
|
| 6694 |
+
"epoch": 0.8586907449209932,
|
| 6695 |
+
"grad_norm": 1.8204010725021362,
|
| 6696 |
+
"learning_rate": 9.920855395512896e-06,
|
| 6697 |
+
"loss": 1.2233,
|
| 6698 |
+
"step": 951
|
| 6699 |
+
},
|
| 6700 |
+
{
|
| 6701 |
+
"epoch": 0.8595936794582393,
|
| 6702 |
+
"grad_norm": 1.5969047546386719,
|
| 6703 |
+
"learning_rate": 9.796976301392934e-06,
|
| 6704 |
+
"loss": 1.0275,
|
| 6705 |
+
"step": 952
|
| 6706 |
+
},
|
| 6707 |
+
{
|
| 6708 |
+
"epoch": 0.8604966139954854,
|
| 6709 |
+
"grad_norm": 1.5068354606628418,
|
| 6710 |
+
"learning_rate": 9.673835648379948e-06,
|
| 6711 |
+
"loss": 0.9911,
|
| 6712 |
+
"step": 953
|
| 6713 |
+
},
|
| 6714 |
+
{
|
| 6715 |
+
"epoch": 0.8613995485327314,
|
| 6716 |
+
"grad_norm": 1.7460449934005737,
|
| 6717 |
+
"learning_rate": 9.551434444556805e-06,
|
| 6718 |
+
"loss": 1.0814,
|
| 6719 |
+
"step": 954
|
| 6720 |
+
},
|
| 6721 |
+
{
|
| 6722 |
+
"epoch": 0.8623024830699775,
|
| 6723 |
+
"grad_norm": 1.3490803241729736,
|
| 6724 |
+
"learning_rate": 9.429773691952858e-06,
|
| 6725 |
+
"loss": 0.9293,
|
| 6726 |
+
"step": 955
|
| 6727 |
+
},
|
| 6728 |
+
{
|
| 6729 |
+
"epoch": 0.8632054176072235,
|
| 6730 |
+
"grad_norm": 1.359166145324707,
|
| 6731 |
+
"learning_rate": 9.308854386535848e-06,
|
| 6732 |
+
"loss": 0.4809,
|
| 6733 |
+
"step": 956
|
| 6734 |
+
},
|
| 6735 |
+
{
|
| 6736 |
+
"epoch": 0.8641083521444696,
|
| 6737 |
+
"grad_norm": 1.6106394529342651,
|
| 6738 |
+
"learning_rate": 9.188677518203703e-06,
|
| 6739 |
+
"loss": 1.251,
|
| 6740 |
+
"step": 957
|
| 6741 |
+
},
|
| 6742 |
+
{
|
| 6743 |
+
"epoch": 0.8650112866817156,
|
| 6744 |
+
"grad_norm": 1.3939250707626343,
|
| 6745 |
+
"learning_rate": 9.069244070776428e-06,
|
| 6746 |
+
"loss": 0.8553,
|
| 6747 |
+
"step": 958
|
| 6748 |
+
},
|
| 6749 |
+
{
|
| 6750 |
+
"epoch": 0.8659142212189617,
|
| 6751 |
+
"grad_norm": 1.9870485067367554,
|
| 6752 |
+
"learning_rate": 8.950555021988061e-06,
|
| 6753 |
+
"loss": 1.4127,
|
| 6754 |
+
"step": 959
|
| 6755 |
+
},
|
| 6756 |
+
{
|
| 6757 |
+
"epoch": 0.8668171557562077,
|
| 6758 |
+
"grad_norm": 1.3053696155548096,
|
| 6759 |
+
"learning_rate": 8.832611343478681e-06,
|
| 6760 |
+
"loss": 0.7796,
|
| 6761 |
+
"step": 960
|
| 6762 |
+
},
|
| 6763 |
+
{
|
| 6764 |
+
"epoch": 0.8677200902934538,
|
| 6765 |
+
"grad_norm": 1.4121004343032837,
|
| 6766 |
+
"learning_rate": 8.715414000786448e-06,
|
| 6767 |
+
"loss": 1.0717,
|
| 6768 |
+
"step": 961
|
| 6769 |
+
},
|
| 6770 |
+
{
|
| 6771 |
+
"epoch": 0.8686230248306998,
|
| 6772 |
+
"grad_norm": 2.2081167697906494,
|
| 6773 |
+
"learning_rate": 8.598963953339667e-06,
|
| 6774 |
+
"loss": 1.6921,
|
| 6775 |
+
"step": 962
|
| 6776 |
+
},
|
| 6777 |
+
{
|
| 6778 |
+
"epoch": 0.8695259593679459,
|
| 6779 |
+
"grad_norm": 2.1040449142456055,
|
| 6780 |
+
"learning_rate": 8.483262154448968e-06,
|
| 6781 |
+
"loss": 1.713,
|
| 6782 |
+
"step": 963
|
| 6783 |
+
},
|
| 6784 |
+
{
|
| 6785 |
+
"epoch": 0.8704288939051918,
|
| 6786 |
+
"grad_norm": 2.0439398288726807,
|
| 6787 |
+
"learning_rate": 8.368309551299536e-06,
|
| 6788 |
+
"loss": 1.7621,
|
| 6789 |
+
"step": 964
|
| 6790 |
+
},
|
| 6791 |
+
{
|
| 6792 |
+
"epoch": 0.871331828442438,
|
| 6793 |
+
"grad_norm": 1.6524083614349365,
|
| 6794 |
+
"learning_rate": 8.25410708494324e-06,
|
| 6795 |
+
"loss": 0.8079,
|
| 6796 |
+
"step": 965
|
| 6797 |
+
},
|
| 6798 |
+
{
|
| 6799 |
+
"epoch": 0.8722347629796839,
|
| 6800 |
+
"grad_norm": 1.563395380973816,
|
| 6801 |
+
"learning_rate": 8.140655690291044e-06,
|
| 6802 |
+
"loss": 0.9102,
|
| 6803 |
+
"step": 966
|
| 6804 |
+
},
|
| 6805 |
+
{
|
| 6806 |
+
"epoch": 0.87313769751693,
|
| 6807 |
+
"grad_norm": 1.7537543773651123,
|
| 6808 |
+
"learning_rate": 8.027956296105354e-06,
|
| 6809 |
+
"loss": 1.1403,
|
| 6810 |
+
"step": 967
|
| 6811 |
+
},
|
| 6812 |
+
{
|
| 6813 |
+
"epoch": 0.874040632054176,
|
| 6814 |
+
"grad_norm": 1.7073965072631836,
|
| 6815 |
+
"learning_rate": 7.916009824992254e-06,
|
| 6816 |
+
"loss": 1.0226,
|
| 6817 |
+
"step": 968
|
| 6818 |
+
},
|
| 6819 |
+
{
|
| 6820 |
+
"epoch": 0.8749435665914221,
|
| 6821 |
+
"grad_norm": 1.5671749114990234,
|
| 6822 |
+
"learning_rate": 7.804817193394242e-06,
|
| 6823 |
+
"loss": 0.8421,
|
| 6824 |
+
"step": 969
|
| 6825 |
+
},
|
| 6826 |
+
{
|
| 6827 |
+
"epoch": 0.8758465011286681,
|
| 6828 |
+
"grad_norm": 1.9673864841461182,
|
| 6829 |
+
"learning_rate": 7.6943793115824e-06,
|
| 6830 |
+
"loss": 1.5373,
|
| 6831 |
+
"step": 970
|
| 6832 |
+
},
|
| 6833 |
+
{
|
| 6834 |
+
"epoch": 0.8767494356659142,
|
| 6835 |
+
"grad_norm": 2.0507729053497314,
|
| 6836 |
+
"learning_rate": 7.584697083649184e-06,
|
| 6837 |
+
"loss": 1.3602,
|
| 6838 |
+
"step": 971
|
| 6839 |
+
},
|
| 6840 |
+
{
|
| 6841 |
+
"epoch": 0.8776523702031602,
|
| 6842 |
+
"grad_norm": 1.852461814880371,
|
| 6843 |
+
"learning_rate": 7.475771407500942e-06,
|
| 6844 |
+
"loss": 1.3545,
|
| 6845 |
+
"step": 972
|
| 6846 |
+
},
|
| 6847 |
+
{
|
| 6848 |
+
"epoch": 0.8785553047404063,
|
| 6849 |
+
"grad_norm": 1.611844778060913,
|
| 6850 |
+
"learning_rate": 7.367603174850502e-06,
|
| 6851 |
+
"loss": 1.0021,
|
| 6852 |
+
"step": 973
|
| 6853 |
+
},
|
| 6854 |
+
{
|
| 6855 |
+
"epoch": 0.8794582392776523,
|
| 6856 |
+
"grad_norm": 1.618187665939331,
|
| 6857 |
+
"learning_rate": 7.2601932712100026e-06,
|
| 6858 |
+
"loss": 1.1898,
|
| 6859 |
+
"step": 974
|
| 6860 |
+
},
|
| 6861 |
+
{
|
| 6862 |
+
"epoch": 0.8803611738148984,
|
| 6863 |
+
"grad_norm": 1.1232011318206787,
|
| 6864 |
+
"learning_rate": 7.153542575883543e-06,
|
| 6865 |
+
"loss": 0.7507,
|
| 6866 |
+
"step": 975
|
| 6867 |
+
},
|
| 6868 |
+
{
|
| 6869 |
+
"epoch": 0.8812641083521444,
|
| 6870 |
+
"grad_norm": 1.6856050491333008,
|
| 6871 |
+
"learning_rate": 7.047651961959978e-06,
|
| 6872 |
+
"loss": 1.2039,
|
| 6873 |
+
"step": 976
|
| 6874 |
+
},
|
| 6875 |
+
{
|
| 6876 |
+
"epoch": 0.8821670428893905,
|
| 6877 |
+
"grad_norm": 1.8394546508789062,
|
| 6878 |
+
"learning_rate": 6.942522296305853e-06,
|
| 6879 |
+
"loss": 0.7429,
|
| 6880 |
+
"step": 977
|
| 6881 |
+
},
|
| 6882 |
+
{
|
| 6883 |
+
"epoch": 0.8830699774266366,
|
| 6884 |
+
"grad_norm": 2.1657228469848633,
|
| 6885 |
+
"learning_rate": 6.838154439558253e-06,
|
| 6886 |
+
"loss": 1.4468,
|
| 6887 |
+
"step": 978
|
| 6888 |
+
},
|
| 6889 |
+
{
|
| 6890 |
+
"epoch": 0.8839729119638826,
|
| 6891 |
+
"grad_norm": 1.8799837827682495,
|
| 6892 |
+
"learning_rate": 6.73454924611776e-06,
|
| 6893 |
+
"loss": 1.3512,
|
| 6894 |
+
"step": 979
|
| 6895 |
+
},
|
| 6896 |
+
{
|
| 6897 |
+
"epoch": 0.8848758465011287,
|
| 6898 |
+
"grad_norm": 1.911229133605957,
|
| 6899 |
+
"learning_rate": 6.631707564141454e-06,
|
| 6900 |
+
"loss": 1.3261,
|
| 6901 |
+
"step": 980
|
| 6902 |
+
},
|
| 6903 |
+
{
|
| 6904 |
+
"epoch": 0.8857787810383747,
|
| 6905 |
+
"grad_norm": 1.6039128303527832,
|
| 6906 |
+
"learning_rate": 6.529630235535955e-06,
|
| 6907 |
+
"loss": 1.099,
|
| 6908 |
+
"step": 981
|
| 6909 |
+
},
|
| 6910 |
+
{
|
| 6911 |
+
"epoch": 0.8866817155756208,
|
| 6912 |
+
"grad_norm": 2.1180193424224854,
|
| 6913 |
+
"learning_rate": 6.428318095950647e-06,
|
| 6914 |
+
"loss": 1.5815,
|
| 6915 |
+
"step": 982
|
| 6916 |
+
},
|
| 6917 |
+
{
|
| 6918 |
+
"epoch": 0.8875846501128668,
|
| 6919 |
+
"grad_norm": 1.5415163040161133,
|
| 6920 |
+
"learning_rate": 6.327771974770625e-06,
|
| 6921 |
+
"loss": 1.2605,
|
| 6922 |
+
"step": 983
|
| 6923 |
+
},
|
| 6924 |
+
{
|
| 6925 |
+
"epoch": 0.8884875846501129,
|
| 6926 |
+
"grad_norm": 1.9346685409545898,
|
| 6927 |
+
"learning_rate": 6.227992695110096e-06,
|
| 6928 |
+
"loss": 1.2316,
|
| 6929 |
+
"step": 984
|
| 6930 |
+
},
|
| 6931 |
+
{
|
| 6932 |
+
"epoch": 0.8893905191873589,
|
| 6933 |
+
"grad_norm": 1.5397899150848389,
|
| 6934 |
+
"learning_rate": 6.128981073805584e-06,
|
| 6935 |
+
"loss": 0.9698,
|
| 6936 |
+
"step": 985
|
| 6937 |
+
},
|
| 6938 |
+
{
|
| 6939 |
+
"epoch": 0.890293453724605,
|
| 6940 |
+
"grad_norm": 1.7011609077453613,
|
| 6941 |
+
"learning_rate": 6.030737921409169e-06,
|
| 6942 |
+
"loss": 1.1717,
|
| 6943 |
+
"step": 986
|
| 6944 |
+
},
|
| 6945 |
+
{
|
| 6946 |
+
"epoch": 0.891196388261851,
|
| 6947 |
+
"grad_norm": 1.8230894804000854,
|
| 6948 |
+
"learning_rate": 5.93326404218194e-06,
|
| 6949 |
+
"loss": 1.1845,
|
| 6950 |
+
"step": 987
|
| 6951 |
+
},
|
| 6952 |
+
{
|
| 6953 |
+
"epoch": 0.8920993227990971,
|
| 6954 |
+
"grad_norm": 1.4834116697311401,
|
| 6955 |
+
"learning_rate": 5.836560234087418e-06,
|
| 6956 |
+
"loss": 1.4324,
|
| 6957 |
+
"step": 988
|
| 6958 |
+
},
|
| 6959 |
+
{
|
| 6960 |
+
"epoch": 0.8930022573363431,
|
| 6961 |
+
"grad_norm": 1.7495137453079224,
|
| 6962 |
+
"learning_rate": 5.740627288784905e-06,
|
| 6963 |
+
"loss": 1.33,
|
| 6964 |
+
"step": 989
|
| 6965 |
+
},
|
| 6966 |
+
{
|
| 6967 |
+
"epoch": 0.8939051918735892,
|
| 6968 |
+
"grad_norm": 1.7764803171157837,
|
| 6969 |
+
"learning_rate": 5.645465991623167e-06,
|
| 6970 |
+
"loss": 1.1222,
|
| 6971 |
+
"step": 990
|
| 6972 |
+
},
|
| 6973 |
+
{
|
| 6974 |
+
"epoch": 0.8948081264108352,
|
| 6975 |
+
"grad_norm": 1.9144630432128906,
|
| 6976 |
+
"learning_rate": 5.551077121633874e-06,
|
| 6977 |
+
"loss": 1.6709,
|
| 6978 |
+
"step": 991
|
| 6979 |
+
},
|
| 6980 |
+
{
|
| 6981 |
+
"epoch": 0.8957110609480813,
|
| 6982 |
+
"grad_norm": 1.673215627670288,
|
| 6983 |
+
"learning_rate": 5.4574614515253145e-06,
|
| 6984 |
+
"loss": 1.3176,
|
| 6985 |
+
"step": 992
|
| 6986 |
+
},
|
| 6987 |
+
{
|
| 6988 |
+
"epoch": 0.8966139954853273,
|
| 6989 |
+
"grad_norm": 1.0927332639694214,
|
| 6990 |
+
"learning_rate": 5.3646197476759586e-06,
|
| 6991 |
+
"loss": 0.5053,
|
| 6992 |
+
"step": 993
|
| 6993 |
+
},
|
| 6994 |
+
{
|
| 6995 |
+
"epoch": 0.8975169300225734,
|
| 6996 |
+
"grad_norm": 1.4994996786117554,
|
| 6997 |
+
"learning_rate": 5.272552770128314e-06,
|
| 6998 |
+
"loss": 0.9512,
|
| 6999 |
+
"step": 994
|
| 7000 |
+
},
|
| 7001 |
+
{
|
| 7002 |
+
"epoch": 0.8984198645598194,
|
| 7003 |
+
"grad_norm": 1.7333306074142456,
|
| 7004 |
+
"learning_rate": 5.181261272582638e-06,
|
| 7005 |
+
"loss": 1.088,
|
| 7006 |
+
"step": 995
|
| 7007 |
+
},
|
| 7008 |
+
{
|
| 7009 |
+
"epoch": 0.8993227990970655,
|
| 7010 |
+
"grad_norm": 1.976701021194458,
|
| 7011 |
+
"learning_rate": 5.090746002390734e-06,
|
| 7012 |
+
"loss": 1.4127,
|
| 7013 |
+
"step": 996
|
| 7014 |
+
},
|
| 7015 |
+
{
|
| 7016 |
+
"epoch": 0.9002257336343115,
|
| 7017 |
+
"grad_norm": 1.407848596572876,
|
| 7018 |
+
"learning_rate": 5.001007700549898e-06,
|
| 7019 |
+
"loss": 0.6292,
|
| 7020 |
+
"step": 997
|
| 7021 |
+
},
|
| 7022 |
+
{
|
| 7023 |
+
"epoch": 0.9011286681715576,
|
| 7024 |
+
"grad_norm": 1.795478343963623,
|
| 7025 |
+
"learning_rate": 4.912047101696848e-06,
|
| 7026 |
+
"loss": 1.4059,
|
| 7027 |
+
"step": 998
|
| 7028 |
+
},
|
| 7029 |
+
{
|
| 7030 |
+
"epoch": 0.9020316027088036,
|
| 7031 |
+
"grad_norm": 1.258590579032898,
|
| 7032 |
+
"learning_rate": 4.823864934101618e-06,
|
| 7033 |
+
"loss": 0.6135,
|
| 7034 |
+
"step": 999
|
| 7035 |
+
},
|
| 7036 |
+
{
|
| 7037 |
+
"epoch": 0.9029345372460497,
|
| 7038 |
+
"grad_norm": 1.7146711349487305,
|
| 7039 |
+
"learning_rate": 4.7364619196617495e-06,
|
| 7040 |
+
"loss": 1.4644,
|
| 7041 |
+
"step": 1000
|
| 7042 |
+
},
|
| 7043 |
+
{
|
| 7044 |
+
"epoch": 0.9038374717832958,
|
| 7045 |
+
"grad_norm": 1.9694057703018188,
|
| 7046 |
+
"learning_rate": 4.649838773896231e-06,
|
| 7047 |
+
"loss": 1.3474,
|
| 7048 |
+
"step": 1001
|
| 7049 |
+
},
|
| 7050 |
+
{
|
| 7051 |
+
"epoch": 0.9047404063205418,
|
| 7052 |
+
"grad_norm": 1.1898510456085205,
|
| 7053 |
+
"learning_rate": 4.563996205939747e-06,
|
| 7054 |
+
"loss": 0.4143,
|
| 7055 |
+
"step": 1002
|
| 7056 |
+
},
|
| 7057 |
+
{
|
| 7058 |
+
"epoch": 0.9056433408577879,
|
| 7059 |
+
"grad_norm": 1.5058674812316895,
|
| 7060 |
+
"learning_rate": 4.478934918536837e-06,
|
| 7061 |
+
"loss": 0.6336,
|
| 7062 |
+
"step": 1003
|
| 7063 |
+
},
|
| 7064 |
+
{
|
| 7065 |
+
"epoch": 0.9065462753950339,
|
| 7066 |
+
"grad_norm": 1.483232855796814,
|
| 7067 |
+
"learning_rate": 4.394655608036092e-06,
|
| 7068 |
+
"loss": 1.2108,
|
| 7069 |
+
"step": 1004
|
| 7070 |
+
},
|
| 7071 |
+
{
|
| 7072 |
+
"epoch": 0.90744920993228,
|
| 7073 |
+
"grad_norm": 1.6962145566940308,
|
| 7074 |
+
"learning_rate": 4.311158964384543e-06,
|
| 7075 |
+
"loss": 1.1302,
|
| 7076 |
+
"step": 1005
|
| 7077 |
+
},
|
| 7078 |
+
{
|
| 7079 |
+
"epoch": 0.908352144469526,
|
| 7080 |
+
"grad_norm": 1.6561692953109741,
|
| 7081 |
+
"learning_rate": 4.228445671121972e-06,
|
| 7082 |
+
"loss": 1.0076,
|
| 7083 |
+
"step": 1006
|
| 7084 |
+
},
|
| 7085 |
+
{
|
| 7086 |
+
"epoch": 0.909255079006772,
|
| 7087 |
+
"grad_norm": 1.3989051580429077,
|
| 7088 |
+
"learning_rate": 4.146516405375289e-06,
|
| 7089 |
+
"loss": 0.885,
|
| 7090 |
+
"step": 1007
|
| 7091 |
+
},
|
| 7092 |
+
{
|
| 7093 |
+
"epoch": 0.910158013544018,
|
| 7094 |
+
"grad_norm": 2.143188953399658,
|
| 7095 |
+
"learning_rate": 4.065371837853016e-06,
|
| 7096 |
+
"loss": 1.506,
|
| 7097 |
+
"step": 1008
|
| 7098 |
+
},
|
| 7099 |
+
{
|
| 7100 |
+
"epoch": 0.9110609480812641,
|
| 7101 |
+
"grad_norm": 1.4973962306976318,
|
| 7102 |
+
"learning_rate": 3.985012632839824e-06,
|
| 7103 |
+
"loss": 0.9991,
|
| 7104 |
+
"step": 1009
|
| 7105 |
+
},
|
| 7106 |
+
{
|
| 7107 |
+
"epoch": 0.9119638826185101,
|
| 7108 |
+
"grad_norm": 1.6684536933898926,
|
| 7109 |
+
"learning_rate": 3.905439448191051e-06,
|
| 7110 |
+
"loss": 0.9142,
|
| 7111 |
+
"step": 1010
|
| 7112 |
+
},
|
| 7113 |
+
{
|
| 7114 |
+
"epoch": 0.9128668171557562,
|
| 7115 |
+
"grad_norm": 1.2150386571884155,
|
| 7116 |
+
"learning_rate": 3.826652935327301e-06,
|
| 7117 |
+
"loss": 0.5438,
|
| 7118 |
+
"step": 1011
|
| 7119 |
+
},
|
| 7120 |
+
{
|
| 7121 |
+
"epoch": 0.9137697516930022,
|
| 7122 |
+
"grad_norm": 1.4311710596084595,
|
| 7123 |
+
"learning_rate": 3.748653739229191e-06,
|
| 7124 |
+
"loss": 1.0845,
|
| 7125 |
+
"step": 1012
|
| 7126 |
+
},
|
| 7127 |
+
{
|
| 7128 |
+
"epoch": 0.9146726862302483,
|
| 7129 |
+
"grad_norm": 2.0449724197387695,
|
| 7130 |
+
"learning_rate": 3.6714424984319986e-06,
|
| 7131 |
+
"loss": 1.2579,
|
| 7132 |
+
"step": 1013
|
| 7133 |
+
},
|
| 7134 |
+
{
|
| 7135 |
+
"epoch": 0.9155756207674943,
|
| 7136 |
+
"grad_norm": 1.6338882446289062,
|
| 7137 |
+
"learning_rate": 3.595019845020442e-06,
|
| 7138 |
+
"loss": 1.0042,
|
| 7139 |
+
"step": 1014
|
| 7140 |
+
},
|
| 7141 |
+
{
|
| 7142 |
+
"epoch": 0.9164785553047404,
|
| 7143 |
+
"grad_norm": 1.2822327613830566,
|
| 7144 |
+
"learning_rate": 3.519386404623537e-06,
|
| 7145 |
+
"loss": 0.6456,
|
| 7146 |
+
"step": 1015
|
| 7147 |
+
},
|
| 7148 |
+
{
|
| 7149 |
+
"epoch": 0.9173814898419864,
|
| 7150 |
+
"grad_norm": 1.741418480873108,
|
| 7151 |
+
"learning_rate": 3.4445427964094778e-06,
|
| 7152 |
+
"loss": 1.0898,
|
| 7153 |
+
"step": 1016
|
| 7154 |
+
},
|
| 7155 |
+
{
|
| 7156 |
+
"epoch": 0.9182844243792325,
|
| 7157 |
+
"grad_norm": 1.4085890054702759,
|
| 7158 |
+
"learning_rate": 3.370489633080509e-06,
|
| 7159 |
+
"loss": 0.9378,
|
| 7160 |
+
"step": 1017
|
| 7161 |
+
},
|
| 7162 |
+
{
|
| 7163 |
+
"epoch": 0.9191873589164785,
|
| 7164 |
+
"grad_norm": 1.8852719068527222,
|
| 7165 |
+
"learning_rate": 3.2972275208679625e-06,
|
| 7166 |
+
"loss": 1.1135,
|
| 7167 |
+
"step": 1018
|
| 7168 |
+
},
|
| 7169 |
+
{
|
| 7170 |
+
"epoch": 0.9200902934537246,
|
| 7171 |
+
"grad_norm": 1.8236089944839478,
|
| 7172 |
+
"learning_rate": 3.2247570595273168e-06,
|
| 7173 |
+
"loss": 0.8719,
|
| 7174 |
+
"step": 1019
|
| 7175 |
+
},
|
| 7176 |
+
{
|
| 7177 |
+
"epoch": 0.9209932279909706,
|
| 7178 |
+
"grad_norm": 1.8349026441574097,
|
| 7179 |
+
"learning_rate": 3.1530788423332126e-06,
|
| 7180 |
+
"loss": 1.224,
|
| 7181 |
+
"step": 1020
|
| 7182 |
+
},
|
| 7183 |
+
{
|
| 7184 |
+
"epoch": 0.9218961625282167,
|
| 7185 |
+
"grad_norm": 1.607959508895874,
|
| 7186 |
+
"learning_rate": 3.0821934560746447e-06,
|
| 7187 |
+
"loss": 1.3352,
|
| 7188 |
+
"step": 1021
|
| 7189 |
+
},
|
| 7190 |
+
{
|
| 7191 |
+
"epoch": 0.9227990970654627,
|
| 7192 |
+
"grad_norm": 1.503532886505127,
|
| 7193 |
+
"learning_rate": 3.012101481050156e-06,
|
| 7194 |
+
"loss": 0.9005,
|
| 7195 |
+
"step": 1022
|
| 7196 |
+
},
|
| 7197 |
+
{
|
| 7198 |
+
"epoch": 0.9237020316027088,
|
| 7199 |
+
"grad_norm": 1.654677152633667,
|
| 7200 |
+
"learning_rate": 2.942803491063095e-06,
|
| 7201 |
+
"loss": 0.7169,
|
| 7202 |
+
"step": 1023
|
| 7203 |
+
},
|
| 7204 |
+
{
|
| 7205 |
+
"epoch": 0.9246049661399548,
|
| 7206 |
+
"grad_norm": 1.8319858312606812,
|
| 7207 |
+
"learning_rate": 2.8743000534168675e-06,
|
| 7208 |
+
"loss": 1.2094,
|
| 7209 |
+
"step": 1024
|
| 7210 |
+
},
|
| 7211 |
+
{
|
| 7212 |
+
"epoch": 0.9255079006772009,
|
| 7213 |
+
"grad_norm": 1.5955770015716553,
|
| 7214 |
+
"learning_rate": 2.806591728910357e-06,
|
| 7215 |
+
"loss": 1.0488,
|
| 7216 |
+
"step": 1025
|
| 7217 |
+
},
|
| 7218 |
+
{
|
| 7219 |
+
"epoch": 0.926410835214447,
|
| 7220 |
+
"grad_norm": 1.5378631353378296,
|
| 7221 |
+
"learning_rate": 2.739679071833301e-06,
|
| 7222 |
+
"loss": 0.9005,
|
| 7223 |
+
"step": 1026
|
| 7224 |
+
},
|
| 7225 |
+
{
|
| 7226 |
+
"epoch": 0.927313769751693,
|
| 7227 |
+
"grad_norm": 1.4716532230377197,
|
| 7228 |
+
"learning_rate": 2.6735626299617457e-06,
|
| 7229 |
+
"loss": 0.8866,
|
| 7230 |
+
"step": 1027
|
| 7231 |
+
},
|
| 7232 |
+
{
|
| 7233 |
+
"epoch": 0.9282167042889391,
|
| 7234 |
+
"grad_norm": 1.300451636314392,
|
| 7235 |
+
"learning_rate": 2.608242944553607e-06,
|
| 7236 |
+
"loss": 1.0212,
|
| 7237 |
+
"step": 1028
|
| 7238 |
+
},
|
| 7239 |
+
{
|
| 7240 |
+
"epoch": 0.9291196388261851,
|
| 7241 |
+
"grad_norm": 1.453568696975708,
|
| 7242 |
+
"learning_rate": 2.5437205503441754e-06,
|
| 7243 |
+
"loss": 0.8845,
|
| 7244 |
+
"step": 1029
|
| 7245 |
+
},
|
| 7246 |
+
{
|
| 7247 |
+
"epoch": 0.9300225733634312,
|
| 7248 |
+
"grad_norm": 1.7968647480010986,
|
| 7249 |
+
"learning_rate": 2.479995975541749e-06,
|
| 7250 |
+
"loss": 1.1663,
|
| 7251 |
+
"step": 1030
|
| 7252 |
+
},
|
| 7253 |
+
{
|
| 7254 |
+
"epoch": 0.9309255079006772,
|
| 7255 |
+
"grad_norm": 1.079252004623413,
|
| 7256 |
+
"learning_rate": 2.417069741823397e-06,
|
| 7257 |
+
"loss": 0.4307,
|
| 7258 |
+
"step": 1031
|
| 7259 |
+
},
|
| 7260 |
+
{
|
| 7261 |
+
"epoch": 0.9318284424379233,
|
| 7262 |
+
"grad_norm": 1.5583209991455078,
|
| 7263 |
+
"learning_rate": 2.354942364330559e-06,
|
| 7264 |
+
"loss": 0.8638,
|
| 7265 |
+
"step": 1032
|
| 7266 |
+
},
|
| 7267 |
+
{
|
| 7268 |
+
"epoch": 0.9327313769751693,
|
| 7269 |
+
"grad_norm": 1.7534904479980469,
|
| 7270 |
+
"learning_rate": 2.2936143516649188e-06,
|
| 7271 |
+
"loss": 1.2616,
|
| 7272 |
+
"step": 1033
|
| 7273 |
+
},
|
| 7274 |
+
{
|
| 7275 |
+
"epoch": 0.9336343115124154,
|
| 7276 |
+
"grad_norm": 1.5467944145202637,
|
| 7277 |
+
"learning_rate": 2.233086205884227e-06,
|
| 7278 |
+
"loss": 0.7197,
|
| 7279 |
+
"step": 1034
|
| 7280 |
+
},
|
| 7281 |
+
{
|
| 7282 |
+
"epoch": 0.9345372460496614,
|
| 7283 |
+
"grad_norm": 1.2515900135040283,
|
| 7284 |
+
"learning_rate": 2.1733584224981396e-06,
|
| 7285 |
+
"loss": 0.6519,
|
| 7286 |
+
"step": 1035
|
| 7287 |
+
},
|
| 7288 |
+
{
|
| 7289 |
+
"epoch": 0.9354401805869075,
|
| 7290 |
+
"grad_norm": 1.4876164197921753,
|
| 7291 |
+
"learning_rate": 2.1144314904642195e-06,
|
| 7292 |
+
"loss": 0.7608,
|
| 7293 |
+
"step": 1036
|
| 7294 |
+
},
|
| 7295 |
+
{
|
| 7296 |
+
"epoch": 0.9363431151241535,
|
| 7297 |
+
"grad_norm": 1.707992434501648,
|
| 7298 |
+
"learning_rate": 2.0563058921839183e-06,
|
| 7299 |
+
"loss": 1.022,
|
| 7300 |
+
"step": 1037
|
| 7301 |
+
},
|
| 7302 |
+
{
|
| 7303 |
+
"epoch": 0.9372460496613996,
|
| 7304 |
+
"grad_norm": 1.9715466499328613,
|
| 7305 |
+
"learning_rate": 1.9989821034986032e-06,
|
| 7306 |
+
"loss": 1.3803,
|
| 7307 |
+
"step": 1038
|
| 7308 |
+
},
|
| 7309 |
+
{
|
| 7310 |
+
"epoch": 0.9381489841986456,
|
| 7311 |
+
"grad_norm": 1.510226845741272,
|
| 7312 |
+
"learning_rate": 1.942460593685713e-06,
|
| 7313 |
+
"loss": 0.9968,
|
| 7314 |
+
"step": 1039
|
| 7315 |
+
},
|
| 7316 |
+
{
|
| 7317 |
+
"epoch": 0.9390519187358917,
|
| 7318 |
+
"grad_norm": 1.6002311706542969,
|
| 7319 |
+
"learning_rate": 1.8867418254548297e-06,
|
| 7320 |
+
"loss": 1.1801,
|
| 7321 |
+
"step": 1040
|
| 7322 |
+
},
|
| 7323 |
+
{
|
| 7324 |
+
"epoch": 0.9399548532731377,
|
| 7325 |
+
"grad_norm": 1.6373189687728882,
|
| 7326 |
+
"learning_rate": 1.8318262549440135e-06,
|
| 7327 |
+
"loss": 0.9959,
|
| 7328 |
+
"step": 1041
|
| 7329 |
+
},
|
| 7330 |
+
{
|
| 7331 |
+
"epoch": 0.9408577878103838,
|
| 7332 |
+
"grad_norm": 1.6004449129104614,
|
| 7333 |
+
"learning_rate": 1.7777143317159406e-06,
|
| 7334 |
+
"loss": 1.1855,
|
| 7335 |
+
"step": 1042
|
| 7336 |
+
},
|
| 7337 |
+
{
|
| 7338 |
+
"epoch": 0.9417607223476298,
|
| 7339 |
+
"grad_norm": 1.7518302202224731,
|
| 7340 |
+
"learning_rate": 1.724406498754294e-06,
|
| 7341 |
+
"loss": 1.2286,
|
| 7342 |
+
"step": 1043
|
| 7343 |
+
},
|
| 7344 |
+
{
|
| 7345 |
+
"epoch": 0.9426636568848759,
|
| 7346 |
+
"grad_norm": 1.3142588138580322,
|
| 7347 |
+
"learning_rate": 1.671903192460156e-06,
|
| 7348 |
+
"loss": 0.8243,
|
| 7349 |
+
"step": 1044
|
| 7350 |
+
},
|
| 7351 |
+
{
|
| 7352 |
+
"epoch": 0.9435665914221218,
|
| 7353 |
+
"grad_norm": 1.5260287523269653,
|
| 7354 |
+
"learning_rate": 1.6202048426483651e-06,
|
| 7355 |
+
"loss": 0.6199,
|
| 7356 |
+
"step": 1045
|
| 7357 |
+
},
|
| 7358 |
+
{
|
| 7359 |
+
"epoch": 0.944469525959368,
|
| 7360 |
+
"grad_norm": 1.7571525573730469,
|
| 7361 |
+
"learning_rate": 1.5693118725440658e-06,
|
| 7362 |
+
"loss": 1.0413,
|
| 7363 |
+
"step": 1046
|
| 7364 |
+
},
|
| 7365 |
+
{
|
| 7366 |
+
"epoch": 0.9453724604966139,
|
| 7367 |
+
"grad_norm": 1.945106029510498,
|
| 7368 |
+
"learning_rate": 1.5192246987791981e-06,
|
| 7369 |
+
"loss": 2.0071,
|
| 7370 |
+
"step": 1047
|
| 7371 |
+
},
|
| 7372 |
+
{
|
| 7373 |
+
"epoch": 0.94627539503386,
|
| 7374 |
+
"grad_norm": 2.676851272583008,
|
| 7375 |
+
"learning_rate": 1.4699437313891007e-06,
|
| 7376 |
+
"loss": 1.9852,
|
| 7377 |
+
"step": 1048
|
| 7378 |
+
},
|
| 7379 |
+
{
|
| 7380 |
+
"epoch": 0.9471783295711061,
|
| 7381 |
+
"grad_norm": 1.967848300933838,
|
| 7382 |
+
"learning_rate": 1.4214693738091812e-06,
|
| 7383 |
+
"loss": 1.345,
|
| 7384 |
+
"step": 1049
|
| 7385 |
+
},
|
| 7386 |
+
{
|
| 7387 |
+
"epoch": 0.9480812641083521,
|
| 7388 |
+
"grad_norm": 1.4572781324386597,
|
| 7389 |
+
"learning_rate": 1.373802022871551e-06,
|
| 7390 |
+
"loss": 0.5701,
|
| 7391 |
+
"step": 1050
|
| 7392 |
+
},
|
| 7393 |
+
{
|
| 7394 |
+
"epoch": 0.9489841986455982,
|
| 7395 |
+
"grad_norm": 1.3797591924667358,
|
| 7396 |
+
"learning_rate": 1.3269420688018508e-06,
|
| 7397 |
+
"loss": 0.8554,
|
| 7398 |
+
"step": 1051
|
| 7399 |
+
},
|
| 7400 |
+
{
|
| 7401 |
+
"epoch": 0.9498871331828442,
|
| 7402 |
+
"grad_norm": 1.8382567167282104,
|
| 7403 |
+
"learning_rate": 1.2808898952160198e-06,
|
| 7404 |
+
"loss": 1.1622,
|
| 7405 |
+
"step": 1052
|
| 7406 |
+
},
|
| 7407 |
+
{
|
| 7408 |
+
"epoch": 0.9507900677200903,
|
| 7409 |
+
"grad_norm": 1.3922321796417236,
|
| 7410 |
+
"learning_rate": 1.2356458791171088e-06,
|
| 7411 |
+
"loss": 1.0533,
|
| 7412 |
+
"step": 1053
|
| 7413 |
+
},
|
| 7414 |
+
{
|
| 7415 |
+
"epoch": 0.9516930022573363,
|
| 7416 |
+
"grad_norm": 1.9144047498703003,
|
| 7417 |
+
"learning_rate": 1.1912103908922945e-06,
|
| 7418 |
+
"loss": 1.1513,
|
| 7419 |
+
"step": 1054
|
| 7420 |
+
},
|
| 7421 |
+
{
|
| 7422 |
+
"epoch": 0.9525959367945824,
|
| 7423 |
+
"grad_norm": 1.6635743379592896,
|
| 7424 |
+
"learning_rate": 1.147583794309759e-06,
|
| 7425 |
+
"loss": 1.3228,
|
| 7426 |
+
"step": 1055
|
| 7427 |
+
},
|
| 7428 |
+
{
|
| 7429 |
+
"epoch": 0.9534988713318284,
|
| 7430 |
+
"grad_norm": 1.768678903579712,
|
| 7431 |
+
"learning_rate": 1.1047664465157592e-06,
|
| 7432 |
+
"loss": 1.1004,
|
| 7433 |
+
"step": 1056
|
| 7434 |
+
},
|
| 7435 |
+
{
|
| 7436 |
+
"epoch": 0.9544018058690745,
|
| 7437 |
+
"grad_norm": 1.643366813659668,
|
| 7438 |
+
"learning_rate": 1.0627586980317073e-06,
|
| 7439 |
+
"loss": 0.9478,
|
| 7440 |
+
"step": 1057
|
| 7441 |
+
},
|
| 7442 |
+
{
|
| 7443 |
+
"epoch": 0.9553047404063205,
|
| 7444 |
+
"grad_norm": 1.3332719802856445,
|
| 7445 |
+
"learning_rate": 1.02156089275125e-06,
|
| 7446 |
+
"loss": 0.7794,
|
| 7447 |
+
"step": 1058
|
| 7448 |
+
},
|
| 7449 |
+
{
|
| 7450 |
+
"epoch": 0.9562076749435666,
|
| 7451 |
+
"grad_norm": 2.097630262374878,
|
| 7452 |
+
"learning_rate": 9.811733679375156e-07,
|
| 7453 |
+
"loss": 1.7037,
|
| 7454 |
+
"step": 1059
|
| 7455 |
+
},
|
| 7456 |
+
{
|
| 7457 |
+
"epoch": 0.9571106094808126,
|
| 7458 |
+
"grad_norm": 1.8974721431732178,
|
| 7459 |
+
"learning_rate": 9.415964542203059e-07,
|
| 7460 |
+
"loss": 1.433,
|
| 7461 |
+
"step": 1060
|
| 7462 |
+
},
|
| 7463 |
+
{
|
| 7464 |
+
"epoch": 0.9580135440180587,
|
| 7465 |
+
"grad_norm": 1.405699610710144,
|
| 7466 |
+
"learning_rate": 9.028304755934081e-07,
|
| 7467 |
+
"loss": 0.7409,
|
| 7468 |
+
"step": 1061
|
| 7469 |
+
},
|
| 7470 |
+
{
|
| 7471 |
+
"epoch": 0.9589164785553047,
|
| 7472 |
+
"grad_norm": 1.6016145944595337,
|
| 7473 |
+
"learning_rate": 8.648757494119752e-07,
|
| 7474 |
+
"loss": 1.0633,
|
| 7475 |
+
"step": 1062
|
| 7476 |
+
},
|
| 7477 |
+
{
|
| 7478 |
+
"epoch": 0.9598194130925508,
|
| 7479 |
+
"grad_norm": 1.3242273330688477,
|
| 7480 |
+
"learning_rate": 8.277325863898511e-07,
|
| 7481 |
+
"loss": 0.6265,
|
| 7482 |
+
"step": 1063
|
| 7483 |
+
},
|
| 7484 |
+
{
|
| 7485 |
+
"epoch": 0.9607223476297968,
|
| 7486 |
+
"grad_norm": 1.8740465641021729,
|
| 7487 |
+
"learning_rate": 7.914012905970935e-07,
|
| 7488 |
+
"loss": 1.4374,
|
| 7489 |
+
"step": 1064
|
| 7490 |
+
},
|
| 7491 |
+
{
|
| 7492 |
+
"epoch": 0.9616252821670429,
|
| 7493 |
+
"grad_norm": 1.297940969467163,
|
| 7494 |
+
"learning_rate": 7.558821594574772e-07,
|
| 7495 |
+
"loss": 0.8479,
|
| 7496 |
+
"step": 1065
|
| 7497 |
+
},
|
| 7498 |
+
{
|
| 7499 |
+
"epoch": 0.9625282167042889,
|
| 7500 |
+
"grad_norm": 1.7031662464141846,
|
| 7501 |
+
"learning_rate": 7.21175483745995e-07,
|
| 7502 |
+
"loss": 1.2352,
|
| 7503 |
+
"step": 1066
|
| 7504 |
+
},
|
| 7505 |
+
{
|
| 7506 |
+
"epoch": 0.963431151241535,
|
| 7507 |
+
"grad_norm": 1.7558894157409668,
|
| 7508 |
+
"learning_rate": 6.872815475865934e-07,
|
| 7509 |
+
"loss": 1.1158,
|
| 7510 |
+
"step": 1067
|
| 7511 |
+
},
|
| 7512 |
+
{
|
| 7513 |
+
"epoch": 0.964334085778781,
|
| 7514 |
+
"grad_norm": 1.4458980560302734,
|
| 7515 |
+
"learning_rate": 6.542006284497304e-07,
|
| 7516 |
+
"loss": 0.9124,
|
| 7517 |
+
"step": 1068
|
| 7518 |
+
},
|
| 7519 |
+
{
|
| 7520 |
+
"epoch": 0.9652370203160271,
|
| 7521 |
+
"grad_norm": 1.6152722835540771,
|
| 7522 |
+
"learning_rate": 6.219329971501653e-07,
|
| 7523 |
+
"loss": 0.7919,
|
| 7524 |
+
"step": 1069
|
| 7525 |
+
},
|
| 7526 |
+
{
|
| 7527 |
+
"epoch": 0.9661399548532731,
|
| 7528 |
+
"grad_norm": 1.3795706033706665,
|
| 7529 |
+
"learning_rate": 5.904789178447389e-07,
|
| 7530 |
+
"loss": 0.8727,
|
| 7531 |
+
"step": 1070
|
| 7532 |
+
},
|
| 7533 |
+
{
|
| 7534 |
+
"epoch": 0.9670428893905192,
|
| 7535 |
+
"grad_norm": 1.5099116563796997,
|
| 7536 |
+
"learning_rate": 5.598386480301976e-07,
|
| 7537 |
+
"loss": 0.9675,
|
| 7538 |
+
"step": 1071
|
| 7539 |
+
},
|
| 7540 |
+
{
|
| 7541 |
+
"epoch": 0.9679458239277653,
|
| 7542 |
+
"grad_norm": 1.8839224576950073,
|
| 7543 |
+
"learning_rate": 5.300124385410943e-07,
|
| 7544 |
+
"loss": 1.5235,
|
| 7545 |
+
"step": 1072
|
| 7546 |
+
},
|
| 7547 |
+
{
|
| 7548 |
+
"epoch": 0.9688487584650113,
|
| 7549 |
+
"grad_norm": 2.0309793949127197,
|
| 7550 |
+
"learning_rate": 5.010005335477241e-07,
|
| 7551 |
+
"loss": 1.2777,
|
| 7552 |
+
"step": 1073
|
| 7553 |
+
},
|
| 7554 |
+
{
|
| 7555 |
+
"epoch": 0.9697516930022574,
|
| 7556 |
+
"grad_norm": 1.4652881622314453,
|
| 7557 |
+
"learning_rate": 4.728031705541369e-07,
|
| 7558 |
+
"loss": 0.7781,
|
| 7559 |
+
"step": 1074
|
| 7560 |
+
},
|
| 7561 |
+
{
|
| 7562 |
+
"epoch": 0.9706546275395034,
|
| 7563 |
+
"grad_norm": 1.8330317735671997,
|
| 7564 |
+
"learning_rate": 4.4542058039619417e-07,
|
| 7565 |
+
"loss": 1.2341,
|
| 7566 |
+
"step": 1075
|
| 7567 |
+
},
|
| 7568 |
+
{
|
| 7569 |
+
"epoch": 0.9715575620767495,
|
| 7570 |
+
"grad_norm": 2.0610013008117676,
|
| 7571 |
+
"learning_rate": 4.188529872396374e-07,
|
| 7572 |
+
"loss": 0.8753,
|
| 7573 |
+
"step": 1076
|
| 7574 |
+
},
|
| 7575 |
+
{
|
| 7576 |
+
"epoch": 0.9724604966139955,
|
| 7577 |
+
"grad_norm": 1.3807718753814697,
|
| 7578 |
+
"learning_rate": 3.9310060857834507e-07,
|
| 7579 |
+
"loss": 0.9572,
|
| 7580 |
+
"step": 1077
|
| 7581 |
+
},
|
| 7582 |
+
{
|
| 7583 |
+
"epoch": 0.9733634311512416,
|
| 7584 |
+
"grad_norm": 1.1987665891647339,
|
| 7585 |
+
"learning_rate": 3.681636552324452e-07,
|
| 7586 |
+
"loss": 0.5451,
|
| 7587 |
+
"step": 1078
|
| 7588 |
+
},
|
| 7589 |
+
{
|
| 7590 |
+
"epoch": 0.9742663656884876,
|
| 7591 |
+
"grad_norm": 1.537322998046875,
|
| 7592 |
+
"learning_rate": 3.440423313466834e-07,
|
| 7593 |
+
"loss": 0.7694,
|
| 7594 |
+
"step": 1079
|
| 7595 |
+
},
|
| 7596 |
+
{
|
| 7597 |
+
"epoch": 0.9751693002257337,
|
| 7598 |
+
"grad_norm": 1.7335373163223267,
|
| 7599 |
+
"learning_rate": 3.2073683438866853e-07,
|
| 7600 |
+
"loss": 0.8225,
|
| 7601 |
+
"step": 1080
|
| 7602 |
+
},
|
| 7603 |
+
{
|
| 7604 |
+
"epoch": 0.9760722347629797,
|
| 7605 |
+
"grad_norm": 1.9911307096481323,
|
| 7606 |
+
"learning_rate": 2.9824735514732974e-07,
|
| 7607 |
+
"loss": 1.4599,
|
| 7608 |
+
"step": 1081
|
| 7609 |
+
},
|
| 7610 |
+
{
|
| 7611 |
+
"epoch": 0.9769751693002258,
|
| 7612 |
+
"grad_norm": 1.5399749279022217,
|
| 7613 |
+
"learning_rate": 2.765740777313064e-07,
|
| 7614 |
+
"loss": 1.2908,
|
| 7615 |
+
"step": 1082
|
| 7616 |
+
},
|
| 7617 |
+
{
|
| 7618 |
+
"epoch": 0.9778781038374718,
|
| 7619 |
+
"grad_norm": 1.6297422647476196,
|
| 7620 |
+
"learning_rate": 2.5571717956743844e-07,
|
| 7621 |
+
"loss": 1.3364,
|
| 7622 |
+
"step": 1083
|
| 7623 |
+
},
|
| 7624 |
+
{
|
| 7625 |
+
"epoch": 0.9787810383747179,
|
| 7626 |
+
"grad_norm": 2.4442360401153564,
|
| 7627 |
+
"learning_rate": 2.3567683139936735e-07,
|
| 7628 |
+
"loss": 0.7743,
|
| 7629 |
+
"step": 1084
|
| 7630 |
+
},
|
| 7631 |
+
{
|
| 7632 |
+
"epoch": 0.9796839729119639,
|
| 7633 |
+
"grad_norm": 1.7662993669509888,
|
| 7634 |
+
"learning_rate": 2.1645319728607062e-07,
|
| 7635 |
+
"loss": 1.1924,
|
| 7636 |
+
"step": 1085
|
| 7637 |
+
},
|
| 7638 |
+
{
|
| 7639 |
+
"epoch": 0.98058690744921,
|
| 7640 |
+
"grad_norm": 1.6082686185836792,
|
| 7641 |
+
"learning_rate": 1.9804643460056284e-07,
|
| 7642 |
+
"loss": 1.2603,
|
| 7643 |
+
"step": 1086
|
| 7644 |
+
},
|
| 7645 |
+
{
|
| 7646 |
+
"epoch": 0.981489841986456,
|
| 7647 |
+
"grad_norm": 1.8181642293930054,
|
| 7648 |
+
"learning_rate": 1.8045669402859677e-07,
|
| 7649 |
+
"loss": 1.1942,
|
| 7650 |
+
"step": 1087
|
| 7651 |
+
},
|
| 7652 |
+
{
|
| 7653 |
+
"epoch": 0.982392776523702,
|
| 7654 |
+
"grad_norm": 1.5067050457000732,
|
| 7655 |
+
"learning_rate": 1.6368411956741991e-07,
|
| 7656 |
+
"loss": 0.9393,
|
| 7657 |
+
"step": 1088
|
| 7658 |
+
},
|
| 7659 |
+
{
|
| 7660 |
+
"epoch": 0.983295711060948,
|
| 7661 |
+
"grad_norm": 1.5651663541793823,
|
| 7662 |
+
"learning_rate": 1.4772884852461977e-07,
|
| 7663 |
+
"loss": 0.9082,
|
| 7664 |
+
"step": 1089
|
| 7665 |
+
},
|
| 7666 |
+
{
|
| 7667 |
+
"epoch": 0.9841986455981941,
|
| 7668 |
+
"grad_norm": 1.480655312538147,
|
| 7669 |
+
"learning_rate": 1.3259101151694708e-07,
|
| 7670 |
+
"loss": 0.8793,
|
| 7671 |
+
"step": 1090
|
| 7672 |
+
},
|
| 7673 |
+
{
|
| 7674 |
+
"epoch": 0.9851015801354401,
|
| 7675 |
+
"grad_norm": 2.0336530208587646,
|
| 7676 |
+
"learning_rate": 1.1827073246932774e-07,
|
| 7677 |
+
"loss": 1.455,
|
| 7678 |
+
"step": 1091
|
| 7679 |
+
},
|
| 7680 |
+
{
|
| 7681 |
+
"epoch": 0.9860045146726862,
|
| 7682 |
+
"grad_norm": 1.600927472114563,
|
| 7683 |
+
"learning_rate": 1.0476812861377471e-07,
|
| 7684 |
+
"loss": 0.9614,
|
| 7685 |
+
"step": 1092
|
| 7686 |
+
},
|
| 7687 |
+
{
|
| 7688 |
+
"epoch": 0.9869074492099322,
|
| 7689 |
+
"grad_norm": 1.7774394750595093,
|
| 7690 |
+
"learning_rate": 9.208331048846663e-08,
|
| 7691 |
+
"loss": 1.3924,
|
| 7692 |
+
"step": 1093
|
| 7693 |
+
},
|
| 7694 |
+
{
|
| 7695 |
+
"epoch": 0.9878103837471783,
|
| 7696 |
+
"grad_norm": 2.1832051277160645,
|
| 7697 |
+
"learning_rate": 8.021638193682624e-08,
|
| 7698 |
+
"loss": 1.5662,
|
| 7699 |
+
"step": 1094
|
| 7700 |
+
},
|
| 7701 |
+
{
|
| 7702 |
+
"epoch": 0.9887133182844243,
|
| 7703 |
+
"grad_norm": 1.5761157274246216,
|
| 7704 |
+
"learning_rate": 6.916744010667663e-08,
|
| 7705 |
+
"loss": 1.3133,
|
| 7706 |
+
"step": 1095
|
| 7707 |
+
},
|
| 7708 |
+
{
|
| 7709 |
+
"epoch": 0.9896162528216704,
|
| 7710 |
+
"grad_norm": 1.2132411003112793,
|
| 7711 |
+
"learning_rate": 5.893657544947528e-08,
|
| 7712 |
+
"loss": 0.569,
|
| 7713 |
+
"step": 1096
|
| 7714 |
+
},
|
| 7715 |
+
{
|
| 7716 |
+
"epoch": 0.9905191873589165,
|
| 7717 |
+
"grad_norm": 1.2810653448104858,
|
| 7718 |
+
"learning_rate": 4.9523871719492353e-08,
|
| 7719 |
+
"loss": 0.6272,
|
| 7720 |
+
"step": 1097
|
| 7721 |
+
},
|
| 7722 |
+
{
|
| 7723 |
+
"epoch": 0.9914221218961625,
|
| 7724 |
+
"grad_norm": 1.5991464853286743,
|
| 7725 |
+
"learning_rate": 4.092940597322237e-08,
|
| 7726 |
+
"loss": 0.9652,
|
| 7727 |
+
"step": 1098
|
| 7728 |
+
},
|
| 7729 |
+
{
|
| 7730 |
+
"epoch": 0.9923250564334086,
|
| 7731 |
+
"grad_norm": 1.7079530954360962,
|
| 7732 |
+
"learning_rate": 3.3153248568695835e-08,
|
| 7733 |
+
"loss": 1.6534,
|
| 7734 |
+
"step": 1099
|
| 7735 |
+
},
|
| 7736 |
+
{
|
| 7737 |
+
"epoch": 0.9932279909706546,
|
| 7738 |
+
"grad_norm": 1.7328083515167236,
|
| 7739 |
+
"learning_rate": 2.6195463164901955e-08,
|
| 7740 |
+
"loss": 0.8277,
|
| 7741 |
+
"step": 1100
|
| 7742 |
+
},
|
| 7743 |
+
{
|
| 7744 |
+
"epoch": 0.9941309255079007,
|
| 7745 |
+
"grad_norm": 1.4812557697296143,
|
| 7746 |
+
"learning_rate": 2.0056106721288992e-08,
|
| 7747 |
+
"loss": 1.0553,
|
| 7748 |
+
"step": 1101
|
| 7749 |
+
},
|
| 7750 |
+
{
|
| 7751 |
+
"epoch": 0.9950338600451467,
|
| 7752 |
+
"grad_norm": 1.4631847143173218,
|
| 7753 |
+
"learning_rate": 1.47352294973091e-08,
|
| 7754 |
+
"loss": 0.8053,
|
| 7755 |
+
"step": 1102
|
| 7756 |
+
},
|
| 7757 |
+
{
|
| 7758 |
+
"epoch": 0.9959367945823928,
|
| 7759 |
+
"grad_norm": 1.5872337818145752,
|
| 7760 |
+
"learning_rate": 1.0232875051963132e-08,
|
| 7761 |
+
"loss": 1.0591,
|
| 7762 |
+
"step": 1103
|
| 7763 |
+
},
|
| 7764 |
+
{
|
| 7765 |
+
"epoch": 0.9968397291196388,
|
| 7766 |
+
"grad_norm": 1.5717228651046753,
|
| 7767 |
+
"learning_rate": 6.549080243489769e-09,
|
| 7768 |
+
"loss": 1.0753,
|
| 7769 |
+
"step": 1104
|
| 7770 |
+
},
|
| 7771 |
+
{
|
| 7772 |
+
"epoch": 0.9977426636568849,
|
| 7773 |
+
"grad_norm": 1.4837099313735962,
|
| 7774 |
+
"learning_rate": 3.6838752290102585e-09,
|
| 7775 |
+
"loss": 1.1738,
|
| 7776 |
+
"step": 1105
|
| 7777 |
+
},
|
| 7778 |
+
{
|
| 7779 |
+
"epoch": 0.9986455981941309,
|
| 7780 |
+
"grad_norm": 1.5750898122787476,
|
| 7781 |
+
"learning_rate": 1.6372834643618717e-09,
|
| 7782 |
+
"loss": 1.0258,
|
| 7783 |
+
"step": 1106
|
| 7784 |
+
},
|
| 7785 |
+
{
|
| 7786 |
+
"epoch": 0.999548532731377,
|
| 7787 |
+
"grad_norm": 1.7029435634613037,
|
| 7788 |
+
"learning_rate": 4.0932170380925116e-10,
|
| 7789 |
+
"loss": 1.0436,
|
| 7790 |
+
"step": 1107
|
| 7791 |
+
},
|
| 7792 |
+
{
|
| 7793 |
+
"epoch": 1.000451467268623,
|
| 7794 |
+
"grad_norm": 2.608523368835449,
|
| 7795 |
+
"learning_rate": 0.0,
|
| 7796 |
+
"loss": 1.6067,
|
| 7797 |
+
"step": 1108
|
| 7798 |
+
},
|
| 7799 |
+
{
|
| 7800 |
+
"epoch": 1.000451467268623,
|
| 7801 |
+
"eval_loss": 1.0790674686431885,
|
| 7802 |
+
"eval_runtime": 8.2901,
|
| 7803 |
+
"eval_samples_per_second": 56.332,
|
| 7804 |
+
"eval_steps_per_second": 28.226,
|
| 7805 |
+
"step": 1108
|
| 7806 |
}
|
| 7807 |
],
|
| 7808 |
"logging_steps": 1,
|
|
|
|
| 7817 |
"should_evaluate": false,
|
| 7818 |
"should_log": false,
|
| 7819 |
"should_save": true,
|
| 7820 |
+
"should_training_stop": true
|
| 7821 |
},
|
| 7822 |
"attributes": {}
|
| 7823 |
}
|
| 7824 |
},
|
| 7825 |
+
"total_flos": 2.834175119012659e+16,
|
| 7826 |
"train_batch_size": 2,
|
| 7827 |
"trial_name": null,
|
| 7828 |
"trial_params": null
|