Training in progress, step 1176, 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 83945296
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:89afbbbf8c70def5456cb4e1f5690a13c95cb980b6a3e688ce41f492f7781c4e
|
3 |
size 83945296
|
last-checkpoint/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 43123028
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:2ca550fc44e8ab5972cdb82e6cc2d2be54d9827de08e9c9e01edd80ed927285f
|
3 |
size 43123028
|
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 sha256:90a6362d85ed0c94e257d5285005ef2743fe40c91d7edfded8b6b3381de419c3
|
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 sha256:23cd479fc96a8f6e5bac35733fb4e3abbee1661f2eee36828b52baa4b0a5d4fd
|
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": 294,
|
6 |
-
"global_step":
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
@@ -6205,6 +6205,2072 @@
|
|
6205 |
"eval_samples_per_second": 12.298,
|
6206 |
"eval_steps_per_second": 6.149,
|
6207 |
"step": 882
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
6208 |
}
|
6209 |
],
|
6210 |
"logging_steps": 1,
|
@@ -6219,12 +8285,12 @@
|
|
6219 |
"should_evaluate": false,
|
6220 |
"should_log": false,
|
6221 |
"should_save": true,
|
6222 |
-
"should_training_stop":
|
6223 |
},
|
6224 |
"attributes": {}
|
6225 |
}
|
6226 |
},
|
6227 |
-
"total_flos":
|
6228 |
"train_batch_size": 2,
|
6229 |
"trial_name": null,
|
6230 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 1.0006379585326954,
|
5 |
"eval_steps": 294,
|
6 |
+
"global_step": 1176,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
6205 |
"eval_samples_per_second": 12.298,
|
6206 |
"eval_steps_per_second": 6.149,
|
6207 |
"step": 882
|
6208 |
+
},
|
6209 |
+
{
|
6210 |
+
"epoch": 0.7510898458266879,
|
6211 |
+
"grad_norm": 0.3817814588546753,
|
6212 |
+
"learning_rate": 2.9575675791302703e-05,
|
6213 |
+
"loss": 2.7678,
|
6214 |
+
"step": 883
|
6215 |
+
},
|
6216 |
+
{
|
6217 |
+
"epoch": 0.7519404572036151,
|
6218 |
+
"grad_norm": 0.4072461724281311,
|
6219 |
+
"learning_rate": 2.9384645085584663e-05,
|
6220 |
+
"loss": 2.7004,
|
6221 |
+
"step": 884
|
6222 |
+
},
|
6223 |
+
{
|
6224 |
+
"epoch": 0.7527910685805422,
|
6225 |
+
"grad_norm": 0.3934648036956787,
|
6226 |
+
"learning_rate": 2.9194127006927208e-05,
|
6227 |
+
"loss": 2.4883,
|
6228 |
+
"step": 885
|
6229 |
+
},
|
6230 |
+
{
|
6231 |
+
"epoch": 0.7536416799574694,
|
6232 |
+
"grad_norm": 0.43659910559654236,
|
6233 |
+
"learning_rate": 2.9004122938382617e-05,
|
6234 |
+
"loss": 2.7447,
|
6235 |
+
"step": 886
|
6236 |
+
},
|
6237 |
+
{
|
6238 |
+
"epoch": 0.7544922913343965,
|
6239 |
+
"grad_norm": 0.4242823123931885,
|
6240 |
+
"learning_rate": 2.881463425927161e-05,
|
6241 |
+
"loss": 2.7324,
|
6242 |
+
"step": 887
|
6243 |
+
},
|
6244 |
+
{
|
6245 |
+
"epoch": 0.7553429027113238,
|
6246 |
+
"grad_norm": 0.4546271860599518,
|
6247 |
+
"learning_rate": 2.86256623451736e-05,
|
6248 |
+
"loss": 2.5985,
|
6249 |
+
"step": 888
|
6250 |
+
},
|
6251 |
+
{
|
6252 |
+
"epoch": 0.756193514088251,
|
6253 |
+
"grad_norm": 0.462896466255188,
|
6254 |
+
"learning_rate": 2.8437208567916517e-05,
|
6255 |
+
"loss": 2.5522,
|
6256 |
+
"step": 889
|
6257 |
+
},
|
6258 |
+
{
|
6259 |
+
"epoch": 0.7570441254651781,
|
6260 |
+
"grad_norm": 0.47179552912712097,
|
6261 |
+
"learning_rate": 2.8249274295566864e-05,
|
6262 |
+
"loss": 2.7653,
|
6263 |
+
"step": 890
|
6264 |
+
},
|
6265 |
+
{
|
6266 |
+
"epoch": 0.7578947368421053,
|
6267 |
+
"grad_norm": 0.5069417357444763,
|
6268 |
+
"learning_rate": 2.8061860892420012e-05,
|
6269 |
+
"loss": 2.8404,
|
6270 |
+
"step": 891
|
6271 |
+
},
|
6272 |
+
{
|
6273 |
+
"epoch": 0.7587453482190324,
|
6274 |
+
"grad_norm": 0.5162703990936279,
|
6275 |
+
"learning_rate": 2.787496971898994e-05,
|
6276 |
+
"loss": 2.5679,
|
6277 |
+
"step": 892
|
6278 |
+
},
|
6279 |
+
{
|
6280 |
+
"epoch": 0.7595959595959596,
|
6281 |
+
"grad_norm": 0.5141138434410095,
|
6282 |
+
"learning_rate": 2.7688602131999565e-05,
|
6283 |
+
"loss": 2.7156,
|
6284 |
+
"step": 893
|
6285 |
+
},
|
6286 |
+
{
|
6287 |
+
"epoch": 0.7604465709728868,
|
6288 |
+
"grad_norm": 0.5267017483711243,
|
6289 |
+
"learning_rate": 2.7502759484370944e-05,
|
6290 |
+
"loss": 2.8324,
|
6291 |
+
"step": 894
|
6292 |
+
},
|
6293 |
+
{
|
6294 |
+
"epoch": 0.7612971823498139,
|
6295 |
+
"grad_norm": 0.5354244112968445,
|
6296 |
+
"learning_rate": 2.7317443125215357e-05,
|
6297 |
+
"loss": 2.7099,
|
6298 |
+
"step": 895
|
6299 |
+
},
|
6300 |
+
{
|
6301 |
+
"epoch": 0.7621477937267411,
|
6302 |
+
"grad_norm": 0.5597330331802368,
|
6303 |
+
"learning_rate": 2.7132654399823444e-05,
|
6304 |
+
"loss": 2.7128,
|
6305 |
+
"step": 896
|
6306 |
+
},
|
6307 |
+
{
|
6308 |
+
"epoch": 0.7629984051036682,
|
6309 |
+
"grad_norm": 0.6345269680023193,
|
6310 |
+
"learning_rate": 2.6948394649655627e-05,
|
6311 |
+
"loss": 2.8904,
|
6312 |
+
"step": 897
|
6313 |
+
},
|
6314 |
+
{
|
6315 |
+
"epoch": 0.7638490164805954,
|
6316 |
+
"grad_norm": 0.5837516784667969,
|
6317 |
+
"learning_rate": 2.6764665212332253e-05,
|
6318 |
+
"loss": 2.5579,
|
6319 |
+
"step": 898
|
6320 |
+
},
|
6321 |
+
{
|
6322 |
+
"epoch": 0.7646996278575225,
|
6323 |
+
"grad_norm": 0.7032825350761414,
|
6324 |
+
"learning_rate": 2.658146742162384e-05,
|
6325 |
+
"loss": 2.7115,
|
6326 |
+
"step": 899
|
6327 |
+
},
|
6328 |
+
{
|
6329 |
+
"epoch": 0.7655502392344498,
|
6330 |
+
"grad_norm": 0.6780116558074951,
|
6331 |
+
"learning_rate": 2.6398802607441507e-05,
|
6332 |
+
"loss": 2.6573,
|
6333 |
+
"step": 900
|
6334 |
+
},
|
6335 |
+
{
|
6336 |
+
"epoch": 0.766400850611377,
|
6337 |
+
"grad_norm": 0.2630327641963959,
|
6338 |
+
"learning_rate": 2.6216672095827266e-05,
|
6339 |
+
"loss": 2.5077,
|
6340 |
+
"step": 901
|
6341 |
+
},
|
6342 |
+
{
|
6343 |
+
"epoch": 0.7672514619883041,
|
6344 |
+
"grad_norm": 0.2665589153766632,
|
6345 |
+
"learning_rate": 2.6035077208944415e-05,
|
6346 |
+
"loss": 2.4456,
|
6347 |
+
"step": 902
|
6348 |
+
},
|
6349 |
+
{
|
6350 |
+
"epoch": 0.7681020733652313,
|
6351 |
+
"grad_norm": 0.2564515471458435,
|
6352 |
+
"learning_rate": 2.5854019265067853e-05,
|
6353 |
+
"loss": 2.2768,
|
6354 |
+
"step": 903
|
6355 |
+
},
|
6356 |
+
{
|
6357 |
+
"epoch": 0.7689526847421584,
|
6358 |
+
"grad_norm": 0.290310263633728,
|
6359 |
+
"learning_rate": 2.5673499578574645e-05,
|
6360 |
+
"loss": 2.6339,
|
6361 |
+
"step": 904
|
6362 |
+
},
|
6363 |
+
{
|
6364 |
+
"epoch": 0.7698032961190856,
|
6365 |
+
"grad_norm": 0.27893102169036865,
|
6366 |
+
"learning_rate": 2.5493519459934423e-05,
|
6367 |
+
"loss": 2.6156,
|
6368 |
+
"step": 905
|
6369 |
+
},
|
6370 |
+
{
|
6371 |
+
"epoch": 0.7706539074960128,
|
6372 |
+
"grad_norm": 0.27719345688819885,
|
6373 |
+
"learning_rate": 2.531408021569982e-05,
|
6374 |
+
"loss": 2.4212,
|
6375 |
+
"step": 906
|
6376 |
+
},
|
6377 |
+
{
|
6378 |
+
"epoch": 0.7715045188729399,
|
6379 |
+
"grad_norm": 0.2813560664653778,
|
6380 |
+
"learning_rate": 2.5135183148496978e-05,
|
6381 |
+
"loss": 2.5064,
|
6382 |
+
"step": 907
|
6383 |
+
},
|
6384 |
+
{
|
6385 |
+
"epoch": 0.7723551302498671,
|
6386 |
+
"grad_norm": 0.28871771693229675,
|
6387 |
+
"learning_rate": 2.4956829557016338e-05,
|
6388 |
+
"loss": 2.505,
|
6389 |
+
"step": 908
|
6390 |
+
},
|
6391 |
+
{
|
6392 |
+
"epoch": 0.7732057416267942,
|
6393 |
+
"grad_norm": 0.30079928040504456,
|
6394 |
+
"learning_rate": 2.4779020736002834e-05,
|
6395 |
+
"loss": 2.4488,
|
6396 |
+
"step": 909
|
6397 |
+
},
|
6398 |
+
{
|
6399 |
+
"epoch": 0.7740563530037214,
|
6400 |
+
"grad_norm": 0.27679336071014404,
|
6401 |
+
"learning_rate": 2.4601757976246686e-05,
|
6402 |
+
"loss": 2.6442,
|
6403 |
+
"step": 910
|
6404 |
+
},
|
6405 |
+
{
|
6406 |
+
"epoch": 0.7749069643806485,
|
6407 |
+
"grad_norm": 0.27239468693733215,
|
6408 |
+
"learning_rate": 2.4425042564574184e-05,
|
6409 |
+
"loss": 2.4256,
|
6410 |
+
"step": 911
|
6411 |
+
},
|
6412 |
+
{
|
6413 |
+
"epoch": 0.7757575757575758,
|
6414 |
+
"grad_norm": 0.2782379686832428,
|
6415 |
+
"learning_rate": 2.4248875783837987e-05,
|
6416 |
+
"loss": 2.5383,
|
6417 |
+
"step": 912
|
6418 |
+
},
|
6419 |
+
{
|
6420 |
+
"epoch": 0.776608187134503,
|
6421 |
+
"grad_norm": 0.28101786971092224,
|
6422 |
+
"learning_rate": 2.407325891290817e-05,
|
6423 |
+
"loss": 2.5509,
|
6424 |
+
"step": 913
|
6425 |
+
},
|
6426 |
+
{
|
6427 |
+
"epoch": 0.7774587985114301,
|
6428 |
+
"grad_norm": 0.26222068071365356,
|
6429 |
+
"learning_rate": 2.3898193226662634e-05,
|
6430 |
+
"loss": 2.5718,
|
6431 |
+
"step": 914
|
6432 |
+
},
|
6433 |
+
{
|
6434 |
+
"epoch": 0.7783094098883573,
|
6435 |
+
"grad_norm": 0.28495025634765625,
|
6436 |
+
"learning_rate": 2.3723679995978088e-05,
|
6437 |
+
"loss": 2.7216,
|
6438 |
+
"step": 915
|
6439 |
+
},
|
6440 |
+
{
|
6441 |
+
"epoch": 0.7791600212652844,
|
6442 |
+
"grad_norm": 0.27799373865127563,
|
6443 |
+
"learning_rate": 2.3549720487720738e-05,
|
6444 |
+
"loss": 2.4825,
|
6445 |
+
"step": 916
|
6446 |
+
},
|
6447 |
+
{
|
6448 |
+
"epoch": 0.7800106326422116,
|
6449 |
+
"grad_norm": 0.2805662751197815,
|
6450 |
+
"learning_rate": 2.3376315964737004e-05,
|
6451 |
+
"loss": 2.5418,
|
6452 |
+
"step": 917
|
6453 |
+
},
|
6454 |
+
{
|
6455 |
+
"epoch": 0.7808612440191387,
|
6456 |
+
"grad_norm": 0.2894647717475891,
|
6457 |
+
"learning_rate": 2.3203467685844494e-05,
|
6458 |
+
"loss": 2.4872,
|
6459 |
+
"step": 918
|
6460 |
+
},
|
6461 |
+
{
|
6462 |
+
"epoch": 0.7817118553960659,
|
6463 |
+
"grad_norm": 0.2725624740123749,
|
6464 |
+
"learning_rate": 2.3031176905822805e-05,
|
6465 |
+
"loss": 2.6821,
|
6466 |
+
"step": 919
|
6467 |
+
},
|
6468 |
+
{
|
6469 |
+
"epoch": 0.7825624667729931,
|
6470 |
+
"grad_norm": 0.2858252227306366,
|
6471 |
+
"learning_rate": 2.2859444875404347e-05,
|
6472 |
+
"loss": 2.5256,
|
6473 |
+
"step": 920
|
6474 |
+
},
|
6475 |
+
{
|
6476 |
+
"epoch": 0.7834130781499202,
|
6477 |
+
"grad_norm": 0.27502021193504333,
|
6478 |
+
"learning_rate": 2.268827284126539e-05,
|
6479 |
+
"loss": 2.6097,
|
6480 |
+
"step": 921
|
6481 |
+
},
|
6482 |
+
{
|
6483 |
+
"epoch": 0.7842636895268474,
|
6484 |
+
"grad_norm": 0.30405184626579285,
|
6485 |
+
"learning_rate": 2.2517662046016975e-05,
|
6486 |
+
"loss": 2.5872,
|
6487 |
+
"step": 922
|
6488 |
+
},
|
6489 |
+
{
|
6490 |
+
"epoch": 0.7851143009037745,
|
6491 |
+
"grad_norm": 0.3032589256763458,
|
6492 |
+
"learning_rate": 2.234761372819577e-05,
|
6493 |
+
"loss": 2.6253,
|
6494 |
+
"step": 923
|
6495 |
+
},
|
6496 |
+
{
|
6497 |
+
"epoch": 0.7859649122807018,
|
6498 |
+
"grad_norm": 0.3149045705795288,
|
6499 |
+
"learning_rate": 2.2178129122255253e-05,
|
6500 |
+
"loss": 2.5631,
|
6501 |
+
"step": 924
|
6502 |
+
},
|
6503 |
+
{
|
6504 |
+
"epoch": 0.786815523657629,
|
6505 |
+
"grad_norm": 0.3490796387195587,
|
6506 |
+
"learning_rate": 2.200920945855669e-05,
|
6507 |
+
"loss": 2.7784,
|
6508 |
+
"step": 925
|
6509 |
+
},
|
6510 |
+
{
|
6511 |
+
"epoch": 0.7876661350345561,
|
6512 |
+
"grad_norm": 0.332727313041687,
|
6513 |
+
"learning_rate": 2.184085596336011e-05,
|
6514 |
+
"loss": 2.7369,
|
6515 |
+
"step": 926
|
6516 |
+
},
|
6517 |
+
{
|
6518 |
+
"epoch": 0.7885167464114833,
|
6519 |
+
"grad_norm": 0.345241516828537,
|
6520 |
+
"learning_rate": 2.1673069858815554e-05,
|
6521 |
+
"loss": 2.734,
|
6522 |
+
"step": 927
|
6523 |
+
},
|
6524 |
+
{
|
6525 |
+
"epoch": 0.7893673577884104,
|
6526 |
+
"grad_norm": 0.356146901845932,
|
6527 |
+
"learning_rate": 2.150585236295415e-05,
|
6528 |
+
"loss": 2.5664,
|
6529 |
+
"step": 928
|
6530 |
+
},
|
6531 |
+
{
|
6532 |
+
"epoch": 0.7902179691653376,
|
6533 |
+
"grad_norm": 0.34294942021369934,
|
6534 |
+
"learning_rate": 2.133920468967915e-05,
|
6535 |
+
"loss": 2.7841,
|
6536 |
+
"step": 929
|
6537 |
+
},
|
6538 |
+
{
|
6539 |
+
"epoch": 0.7910685805422647,
|
6540 |
+
"grad_norm": 0.3757612407207489,
|
6541 |
+
"learning_rate": 2.1173128048757306e-05,
|
6542 |
+
"loss": 2.8389,
|
6543 |
+
"step": 930
|
6544 |
+
},
|
6545 |
+
{
|
6546 |
+
"epoch": 0.7919191919191919,
|
6547 |
+
"grad_norm": 0.37201401591300964,
|
6548 |
+
"learning_rate": 2.1007623645810003e-05,
|
6549 |
+
"loss": 2.7424,
|
6550 |
+
"step": 931
|
6551 |
+
},
|
6552 |
+
{
|
6553 |
+
"epoch": 0.7927698032961191,
|
6554 |
+
"grad_norm": 0.37049031257629395,
|
6555 |
+
"learning_rate": 2.0842692682304444e-05,
|
6556 |
+
"loss": 2.7447,
|
6557 |
+
"step": 932
|
6558 |
+
},
|
6559 |
+
{
|
6560 |
+
"epoch": 0.7936204146730462,
|
6561 |
+
"grad_norm": 0.40525051951408386,
|
6562 |
+
"learning_rate": 2.0678336355545048e-05,
|
6563 |
+
"loss": 2.7354,
|
6564 |
+
"step": 933
|
6565 |
+
},
|
6566 |
+
{
|
6567 |
+
"epoch": 0.7944710260499734,
|
6568 |
+
"grad_norm": 0.37740442156791687,
|
6569 |
+
"learning_rate": 2.0514555858664663e-05,
|
6570 |
+
"loss": 2.5756,
|
6571 |
+
"step": 934
|
6572 |
+
},
|
6573 |
+
{
|
6574 |
+
"epoch": 0.7953216374269005,
|
6575 |
+
"grad_norm": 0.399179607629776,
|
6576 |
+
"learning_rate": 2.0351352380616008e-05,
|
6577 |
+
"loss": 2.606,
|
6578 |
+
"step": 935
|
6579 |
+
},
|
6580 |
+
{
|
6581 |
+
"epoch": 0.7961722488038278,
|
6582 |
+
"grad_norm": 0.4020417630672455,
|
6583 |
+
"learning_rate": 2.0188727106162874e-05,
|
6584 |
+
"loss": 2.5945,
|
6585 |
+
"step": 936
|
6586 |
+
},
|
6587 |
+
{
|
6588 |
+
"epoch": 0.797022860180755,
|
6589 |
+
"grad_norm": 0.4364849627017975,
|
6590 |
+
"learning_rate": 2.0026681215871656e-05,
|
6591 |
+
"loss": 2.7867,
|
6592 |
+
"step": 937
|
6593 |
+
},
|
6594 |
+
{
|
6595 |
+
"epoch": 0.7978734715576821,
|
6596 |
+
"grad_norm": 0.4335819482803345,
|
6597 |
+
"learning_rate": 1.986521588610285e-05,
|
6598 |
+
"loss": 2.6452,
|
6599 |
+
"step": 938
|
6600 |
+
},
|
6601 |
+
{
|
6602 |
+
"epoch": 0.7987240829346093,
|
6603 |
+
"grad_norm": 0.44533243775367737,
|
6604 |
+
"learning_rate": 1.9704332289002293e-05,
|
6605 |
+
"loss": 2.7037,
|
6606 |
+
"step": 939
|
6607 |
+
},
|
6608 |
+
{
|
6609 |
+
"epoch": 0.7995746943115364,
|
6610 |
+
"grad_norm": 0.46524372696876526,
|
6611 |
+
"learning_rate": 1.9544031592492763e-05,
|
6612 |
+
"loss": 2.7772,
|
6613 |
+
"step": 940
|
6614 |
+
},
|
6615 |
+
{
|
6616 |
+
"epoch": 0.8004253056884636,
|
6617 |
+
"grad_norm": 0.4814035892486572,
|
6618 |
+
"learning_rate": 1.9384314960265692e-05,
|
6619 |
+
"loss": 2.7603,
|
6620 |
+
"step": 941
|
6621 |
+
},
|
6622 |
+
{
|
6623 |
+
"epoch": 0.8012759170653907,
|
6624 |
+
"grad_norm": 0.5147087574005127,
|
6625 |
+
"learning_rate": 1.922518355177232e-05,
|
6626 |
+
"loss": 2.8118,
|
6627 |
+
"step": 942
|
6628 |
+
},
|
6629 |
+
{
|
6630 |
+
"epoch": 0.8021265284423179,
|
6631 |
+
"grad_norm": 0.5196573734283447,
|
6632 |
+
"learning_rate": 1.906663852221565e-05,
|
6633 |
+
"loss": 2.7226,
|
6634 |
+
"step": 943
|
6635 |
+
},
|
6636 |
+
{
|
6637 |
+
"epoch": 0.8029771398192451,
|
6638 |
+
"grad_norm": 0.538750171661377,
|
6639 |
+
"learning_rate": 1.890868102254182e-05,
|
6640 |
+
"loss": 2.8252,
|
6641 |
+
"step": 944
|
6642 |
+
},
|
6643 |
+
{
|
6644 |
+
"epoch": 0.8038277511961722,
|
6645 |
+
"grad_norm": 0.5554125308990479,
|
6646 |
+
"learning_rate": 1.875131219943187e-05,
|
6647 |
+
"loss": 2.7985,
|
6648 |
+
"step": 945
|
6649 |
+
},
|
6650 |
+
{
|
6651 |
+
"epoch": 0.8046783625730994,
|
6652 |
+
"grad_norm": 0.5610653162002563,
|
6653 |
+
"learning_rate": 1.8594533195293427e-05,
|
6654 |
+
"loss": 2.7006,
|
6655 |
+
"step": 946
|
6656 |
+
},
|
6657 |
+
{
|
6658 |
+
"epoch": 0.8055289739500265,
|
6659 |
+
"grad_norm": 0.5772664546966553,
|
6660 |
+
"learning_rate": 1.843834514825229e-05,
|
6661 |
+
"loss": 2.7453,
|
6662 |
+
"step": 947
|
6663 |
+
},
|
6664 |
+
{
|
6665 |
+
"epoch": 0.8063795853269538,
|
6666 |
+
"grad_norm": 0.7284188866615295,
|
6667 |
+
"learning_rate": 1.82827491921443e-05,
|
6668 |
+
"loss": 2.7826,
|
6669 |
+
"step": 948
|
6670 |
+
},
|
6671 |
+
{
|
6672 |
+
"epoch": 0.807230196703881,
|
6673 |
+
"grad_norm": 0.6359315514564514,
|
6674 |
+
"learning_rate": 1.8127746456507077e-05,
|
6675 |
+
"loss": 2.5094,
|
6676 |
+
"step": 949
|
6677 |
+
},
|
6678 |
+
{
|
6679 |
+
"epoch": 0.8080808080808081,
|
6680 |
+
"grad_norm": 0.7618624567985535,
|
6681 |
+
"learning_rate": 1.797333806657171e-05,
|
6682 |
+
"loss": 2.9762,
|
6683 |
+
"step": 950
|
6684 |
+
},
|
6685 |
+
{
|
6686 |
+
"epoch": 0.8089314194577353,
|
6687 |
+
"grad_norm": 0.2761584520339966,
|
6688 |
+
"learning_rate": 1.7819525143254755e-05,
|
6689 |
+
"loss": 2.3495,
|
6690 |
+
"step": 951
|
6691 |
+
},
|
6692 |
+
{
|
6693 |
+
"epoch": 0.8097820308346624,
|
6694 |
+
"grad_norm": 0.2787644565105438,
|
6695 |
+
"learning_rate": 1.7666308803150043e-05,
|
6696 |
+
"loss": 2.6593,
|
6697 |
+
"step": 952
|
6698 |
+
},
|
6699 |
+
{
|
6700 |
+
"epoch": 0.8106326422115896,
|
6701 |
+
"grad_norm": 0.2583043873310089,
|
6702 |
+
"learning_rate": 1.751369015852046e-05,
|
6703 |
+
"loss": 2.3744,
|
6704 |
+
"step": 953
|
6705 |
+
},
|
6706 |
+
{
|
6707 |
+
"epoch": 0.8114832535885167,
|
6708 |
+
"grad_norm": 0.27526992559432983,
|
6709 |
+
"learning_rate": 1.7361670317290012e-05,
|
6710 |
+
"loss": 2.424,
|
6711 |
+
"step": 954
|
6712 |
+
},
|
6713 |
+
{
|
6714 |
+
"epoch": 0.8123338649654439,
|
6715 |
+
"grad_norm": 0.2713683843612671,
|
6716 |
+
"learning_rate": 1.7210250383035807e-05,
|
6717 |
+
"loss": 2.5069,
|
6718 |
+
"step": 955
|
6719 |
+
},
|
6720 |
+
{
|
6721 |
+
"epoch": 0.8131844763423711,
|
6722 |
+
"grad_norm": 0.28076109290122986,
|
6723 |
+
"learning_rate": 1.7059431454979824e-05,
|
6724 |
+
"loss": 2.4204,
|
6725 |
+
"step": 956
|
6726 |
+
},
|
6727 |
+
{
|
6728 |
+
"epoch": 0.8140350877192982,
|
6729 |
+
"grad_norm": 0.2972472012042999,
|
6730 |
+
"learning_rate": 1.6909214627981197e-05,
|
6731 |
+
"loss": 2.6109,
|
6732 |
+
"step": 957
|
6733 |
+
},
|
6734 |
+
{
|
6735 |
+
"epoch": 0.8148856990962254,
|
6736 |
+
"grad_norm": 0.31532159447669983,
|
6737 |
+
"learning_rate": 1.6759600992528147e-05,
|
6738 |
+
"loss": 2.6416,
|
6739 |
+
"step": 958
|
6740 |
+
},
|
6741 |
+
{
|
6742 |
+
"epoch": 0.8157363104731525,
|
6743 |
+
"grad_norm": 0.2699948847293854,
|
6744 |
+
"learning_rate": 1.6610591634729965e-05,
|
6745 |
+
"loss": 2.4951,
|
6746 |
+
"step": 959
|
6747 |
+
},
|
6748 |
+
{
|
6749 |
+
"epoch": 0.8165869218500797,
|
6750 |
+
"grad_norm": 0.27630969882011414,
|
6751 |
+
"learning_rate": 1.6462187636309345e-05,
|
6752 |
+
"loss": 2.6336,
|
6753 |
+
"step": 960
|
6754 |
+
},
|
6755 |
+
{
|
6756 |
+
"epoch": 0.817437533227007,
|
6757 |
+
"grad_norm": 0.27848508954048157,
|
6758 |
+
"learning_rate": 1.631439007459441e-05,
|
6759 |
+
"loss": 2.6848,
|
6760 |
+
"step": 961
|
6761 |
+
},
|
6762 |
+
{
|
6763 |
+
"epoch": 0.8182881446039341,
|
6764 |
+
"grad_norm": 0.2589823007583618,
|
6765 |
+
"learning_rate": 1.61672000225108e-05,
|
6766 |
+
"loss": 2.6077,
|
6767 |
+
"step": 962
|
6768 |
+
},
|
6769 |
+
{
|
6770 |
+
"epoch": 0.8191387559808613,
|
6771 |
+
"grad_norm": 0.27338507771492004,
|
6772 |
+
"learning_rate": 1.6020618548574108e-05,
|
6773 |
+
"loss": 2.4618,
|
6774 |
+
"step": 963
|
6775 |
+
},
|
6776 |
+
{
|
6777 |
+
"epoch": 0.8199893673577884,
|
6778 |
+
"grad_norm": 0.26967307925224304,
|
6779 |
+
"learning_rate": 1.587464671688187e-05,
|
6780 |
+
"loss": 2.4733,
|
6781 |
+
"step": 964
|
6782 |
+
},
|
6783 |
+
{
|
6784 |
+
"epoch": 0.8208399787347156,
|
6785 |
+
"grad_norm": 0.27979278564453125,
|
6786 |
+
"learning_rate": 1.5729285587106136e-05,
|
6787 |
+
"loss": 2.6789,
|
6788 |
+
"step": 965
|
6789 |
+
},
|
6790 |
+
{
|
6791 |
+
"epoch": 0.8216905901116427,
|
6792 |
+
"grad_norm": 0.263261616230011,
|
6793 |
+
"learning_rate": 1.5584536214485457e-05,
|
6794 |
+
"loss": 2.5471,
|
6795 |
+
"step": 966
|
6796 |
+
},
|
6797 |
+
{
|
6798 |
+
"epoch": 0.8225412014885699,
|
6799 |
+
"grad_norm": 0.27126452326774597,
|
6800 |
+
"learning_rate": 1.5440399649817385e-05,
|
6801 |
+
"loss": 2.6185,
|
6802 |
+
"step": 967
|
6803 |
+
},
|
6804 |
+
{
|
6805 |
+
"epoch": 0.8233918128654971,
|
6806 |
+
"grad_norm": 0.2681058943271637,
|
6807 |
+
"learning_rate": 1.5296876939450978e-05,
|
6808 |
+
"loss": 2.567,
|
6809 |
+
"step": 968
|
6810 |
+
},
|
6811 |
+
{
|
6812 |
+
"epoch": 0.8242424242424242,
|
6813 |
+
"grad_norm": 0.25772204995155334,
|
6814 |
+
"learning_rate": 1.5153969125278934e-05,
|
6815 |
+
"loss": 2.4961,
|
6816 |
+
"step": 969
|
6817 |
+
},
|
6818 |
+
{
|
6819 |
+
"epoch": 0.8250930356193514,
|
6820 |
+
"grad_norm": 0.27491864562034607,
|
6821 |
+
"learning_rate": 1.5011677244730161e-05,
|
6822 |
+
"loss": 2.6572,
|
6823 |
+
"step": 970
|
6824 |
+
},
|
6825 |
+
{
|
6826 |
+
"epoch": 0.8259436469962785,
|
6827 |
+
"grad_norm": 0.3069652318954468,
|
6828 |
+
"learning_rate": 1.4870002330762289e-05,
|
6829 |
+
"loss": 2.6633,
|
6830 |
+
"step": 971
|
6831 |
+
},
|
6832 |
+
{
|
6833 |
+
"epoch": 0.8267942583732057,
|
6834 |
+
"grad_norm": 0.29441124200820923,
|
6835 |
+
"learning_rate": 1.4728945411854133e-05,
|
6836 |
+
"loss": 2.5631,
|
6837 |
+
"step": 972
|
6838 |
+
},
|
6839 |
+
{
|
6840 |
+
"epoch": 0.827644869750133,
|
6841 |
+
"grad_norm": 0.29318681359291077,
|
6842 |
+
"learning_rate": 1.4588507511998162e-05,
|
6843 |
+
"loss": 2.7143,
|
6844 |
+
"step": 973
|
6845 |
+
},
|
6846 |
+
{
|
6847 |
+
"epoch": 0.8284954811270601,
|
6848 |
+
"grad_norm": 0.3085382282733917,
|
6849 |
+
"learning_rate": 1.4448689650693147e-05,
|
6850 |
+
"loss": 2.6399,
|
6851 |
+
"step": 974
|
6852 |
+
},
|
6853 |
+
{
|
6854 |
+
"epoch": 0.8293460925039873,
|
6855 |
+
"grad_norm": 0.3353934586048126,
|
6856 |
+
"learning_rate": 1.4309492842936768e-05,
|
6857 |
+
"loss": 2.7392,
|
6858 |
+
"step": 975
|
6859 |
+
},
|
6860 |
+
{
|
6861 |
+
"epoch": 0.8301967038809144,
|
6862 |
+
"grad_norm": 0.35874679684638977,
|
6863 |
+
"learning_rate": 1.4170918099218166e-05,
|
6864 |
+
"loss": 2.6478,
|
6865 |
+
"step": 976
|
6866 |
+
},
|
6867 |
+
{
|
6868 |
+
"epoch": 0.8310473152578416,
|
6869 |
+
"grad_norm": 0.3231167793273926,
|
6870 |
+
"learning_rate": 1.4032966425510663e-05,
|
6871 |
+
"loss": 2.5808,
|
6872 |
+
"step": 977
|
6873 |
+
},
|
6874 |
+
{
|
6875 |
+
"epoch": 0.8318979266347687,
|
6876 |
+
"grad_norm": 0.3558177351951599,
|
6877 |
+
"learning_rate": 1.3895638823264446e-05,
|
6878 |
+
"loss": 2.8192,
|
6879 |
+
"step": 978
|
6880 |
+
},
|
6881 |
+
{
|
6882 |
+
"epoch": 0.8327485380116959,
|
6883 |
+
"grad_norm": 0.3447312116622925,
|
6884 |
+
"learning_rate": 1.3758936289399348e-05,
|
6885 |
+
"loss": 2.439,
|
6886 |
+
"step": 979
|
6887 |
+
},
|
6888 |
+
{
|
6889 |
+
"epoch": 0.8335991493886231,
|
6890 |
+
"grad_norm": 0.35401877760887146,
|
6891 |
+
"learning_rate": 1.3622859816297473e-05,
|
6892 |
+
"loss": 2.6874,
|
6893 |
+
"step": 980
|
6894 |
+
},
|
6895 |
+
{
|
6896 |
+
"epoch": 0.8344497607655502,
|
6897 |
+
"grad_norm": 0.3697657585144043,
|
6898 |
+
"learning_rate": 1.3487410391796162e-05,
|
6899 |
+
"loss": 2.6217,
|
6900 |
+
"step": 981
|
6901 |
+
},
|
6902 |
+
{
|
6903 |
+
"epoch": 0.8353003721424774,
|
6904 |
+
"grad_norm": 0.39387422800064087,
|
6905 |
+
"learning_rate": 1.3352588999180726e-05,
|
6906 |
+
"loss": 2.9086,
|
6907 |
+
"step": 982
|
6908 |
+
},
|
6909 |
+
{
|
6910 |
+
"epoch": 0.8361509835194045,
|
6911 |
+
"grad_norm": 0.4256942570209503,
|
6912 |
+
"learning_rate": 1.3218396617177287e-05,
|
6913 |
+
"loss": 2.6578,
|
6914 |
+
"step": 983
|
6915 |
+
},
|
6916 |
+
{
|
6917 |
+
"epoch": 0.8370015948963317,
|
6918 |
+
"grad_norm": 0.40965893864631653,
|
6919 |
+
"learning_rate": 1.308483421994573e-05,
|
6920 |
+
"loss": 2.6781,
|
6921 |
+
"step": 984
|
6922 |
+
},
|
6923 |
+
{
|
6924 |
+
"epoch": 0.837852206273259,
|
6925 |
+
"grad_norm": 0.4041212797164917,
|
6926 |
+
"learning_rate": 1.2951902777072655e-05,
|
6927 |
+
"loss": 2.8038,
|
6928 |
+
"step": 985
|
6929 |
+
},
|
6930 |
+
{
|
6931 |
+
"epoch": 0.8387028176501861,
|
6932 |
+
"grad_norm": 0.416715532541275,
|
6933 |
+
"learning_rate": 1.2819603253564205e-05,
|
6934 |
+
"loss": 2.5885,
|
6935 |
+
"step": 986
|
6936 |
+
},
|
6937 |
+
{
|
6938 |
+
"epoch": 0.8395534290271133,
|
6939 |
+
"grad_norm": 0.4417300522327423,
|
6940 |
+
"learning_rate": 1.2687936609839235e-05,
|
6941 |
+
"loss": 2.8096,
|
6942 |
+
"step": 987
|
6943 |
+
},
|
6944 |
+
{
|
6945 |
+
"epoch": 0.8404040404040404,
|
6946 |
+
"grad_norm": 0.47883424162864685,
|
6947 |
+
"learning_rate": 1.2556903801722219e-05,
|
6948 |
+
"loss": 2.8092,
|
6949 |
+
"step": 988
|
6950 |
+
},
|
6951 |
+
{
|
6952 |
+
"epoch": 0.8412546517809676,
|
6953 |
+
"grad_norm": 0.4706938862800598,
|
6954 |
+
"learning_rate": 1.2426505780436326e-05,
|
6955 |
+
"loss": 2.6285,
|
6956 |
+
"step": 989
|
6957 |
+
},
|
6958 |
+
{
|
6959 |
+
"epoch": 0.8421052631578947,
|
6960 |
+
"grad_norm": 0.4550638496875763,
|
6961 |
+
"learning_rate": 1.2296743492596586e-05,
|
6962 |
+
"loss": 2.5213,
|
6963 |
+
"step": 990
|
6964 |
+
},
|
6965 |
+
{
|
6966 |
+
"epoch": 0.8429558745348219,
|
6967 |
+
"grad_norm": 0.48749181628227234,
|
6968 |
+
"learning_rate": 1.2167617880202908e-05,
|
6969 |
+
"loss": 2.791,
|
6970 |
+
"step": 991
|
6971 |
+
},
|
6972 |
+
{
|
6973 |
+
"epoch": 0.8438064859117491,
|
6974 |
+
"grad_norm": 0.46483585238456726,
|
6975 |
+
"learning_rate": 1.2039129880633349e-05,
|
6976 |
+
"loss": 2.7672,
|
6977 |
+
"step": 992
|
6978 |
+
},
|
6979 |
+
{
|
6980 |
+
"epoch": 0.8446570972886762,
|
6981 |
+
"grad_norm": 0.510725200176239,
|
6982 |
+
"learning_rate": 1.1911280426637273e-05,
|
6983 |
+
"loss": 2.9072,
|
6984 |
+
"step": 993
|
6985 |
+
},
|
6986 |
+
{
|
6987 |
+
"epoch": 0.8455077086656034,
|
6988 |
+
"grad_norm": 0.5274655222892761,
|
6989 |
+
"learning_rate": 1.1784070446328476e-05,
|
6990 |
+
"loss": 2.8219,
|
6991 |
+
"step": 994
|
6992 |
+
},
|
6993 |
+
{
|
6994 |
+
"epoch": 0.8463583200425305,
|
6995 |
+
"grad_norm": 0.5416675806045532,
|
6996 |
+
"learning_rate": 1.1657500863178694e-05,
|
6997 |
+
"loss": 2.7237,
|
6998 |
+
"step": 995
|
6999 |
+
},
|
7000 |
+
{
|
7001 |
+
"epoch": 0.8472089314194577,
|
7002 |
+
"grad_norm": 0.5472414493560791,
|
7003 |
+
"learning_rate": 1.153157259601062e-05,
|
7004 |
+
"loss": 2.68,
|
7005 |
+
"step": 996
|
7006 |
+
},
|
7007 |
+
{
|
7008 |
+
"epoch": 0.8480595427963848,
|
7009 |
+
"grad_norm": 0.5868435502052307,
|
7010 |
+
"learning_rate": 1.1406286558991375e-05,
|
7011 |
+
"loss": 2.7562,
|
7012 |
+
"step": 997
|
7013 |
+
},
|
7014 |
+
{
|
7015 |
+
"epoch": 0.8489101541733121,
|
7016 |
+
"grad_norm": 0.6191973090171814,
|
7017 |
+
"learning_rate": 1.1281643661625895e-05,
|
7018 |
+
"loss": 2.7655,
|
7019 |
+
"step": 998
|
7020 |
+
},
|
7021 |
+
{
|
7022 |
+
"epoch": 0.8497607655502393,
|
7023 |
+
"grad_norm": 0.6745545268058777,
|
7024 |
+
"learning_rate": 1.1157644808750312e-05,
|
7025 |
+
"loss": 2.8605,
|
7026 |
+
"step": 999
|
7027 |
+
},
|
7028 |
+
{
|
7029 |
+
"epoch": 0.8506113769271664,
|
7030 |
+
"grad_norm": 0.697556734085083,
|
7031 |
+
"learning_rate": 1.103429090052528e-05,
|
7032 |
+
"loss": 2.6958,
|
7033 |
+
"step": 1000
|
7034 |
+
},
|
7035 |
+
{
|
7036 |
+
"epoch": 0.8514619883040936,
|
7037 |
+
"grad_norm": 0.26258981227874756,
|
7038 |
+
"learning_rate": 1.0911582832429589e-05,
|
7039 |
+
"loss": 2.4457,
|
7040 |
+
"step": 1001
|
7041 |
+
},
|
7042 |
+
{
|
7043 |
+
"epoch": 0.8523125996810207,
|
7044 |
+
"grad_norm": 0.26820749044418335,
|
7045 |
+
"learning_rate": 1.0789521495253618e-05,
|
7046 |
+
"loss": 2.6249,
|
7047 |
+
"step": 1002
|
7048 |
+
},
|
7049 |
+
{
|
7050 |
+
"epoch": 0.8531632110579479,
|
7051 |
+
"grad_norm": 0.26697129011154175,
|
7052 |
+
"learning_rate": 1.0668107775092751e-05,
|
7053 |
+
"loss": 2.4464,
|
7054 |
+
"step": 1003
|
7055 |
+
},
|
7056 |
+
{
|
7057 |
+
"epoch": 0.8540138224348751,
|
7058 |
+
"grad_norm": 0.2953515350818634,
|
7059 |
+
"learning_rate": 1.0547342553341144e-05,
|
7060 |
+
"loss": 2.5413,
|
7061 |
+
"step": 1004
|
7062 |
+
},
|
7063 |
+
{
|
7064 |
+
"epoch": 0.8548644338118022,
|
7065 |
+
"grad_norm": 0.264387309551239,
|
7066 |
+
"learning_rate": 1.0427226706685178e-05,
|
7067 |
+
"loss": 2.6889,
|
7068 |
+
"step": 1005
|
7069 |
+
},
|
7070 |
+
{
|
7071 |
+
"epoch": 0.8557150451887294,
|
7072 |
+
"grad_norm": 0.2656628489494324,
|
7073 |
+
"learning_rate": 1.030776110709718e-05,
|
7074 |
+
"loss": 2.5115,
|
7075 |
+
"step": 1006
|
7076 |
+
},
|
7077 |
+
{
|
7078 |
+
"epoch": 0.8565656565656565,
|
7079 |
+
"grad_norm": 0.25700512528419495,
|
7080 |
+
"learning_rate": 1.0188946621828976e-05,
|
7081 |
+
"loss": 2.5474,
|
7082 |
+
"step": 1007
|
7083 |
+
},
|
7084 |
+
{
|
7085 |
+
"epoch": 0.8574162679425837,
|
7086 |
+
"grad_norm": 0.28423216938972473,
|
7087 |
+
"learning_rate": 1.0070784113405763e-05,
|
7088 |
+
"loss": 2.5926,
|
7089 |
+
"step": 1008
|
7090 |
+
},
|
7091 |
+
{
|
7092 |
+
"epoch": 0.8582668793195108,
|
7093 |
+
"grad_norm": 0.2653867304325104,
|
7094 |
+
"learning_rate": 9.953274439619741e-06,
|
7095 |
+
"loss": 2.4022,
|
7096 |
+
"step": 1009
|
7097 |
+
},
|
7098 |
+
{
|
7099 |
+
"epoch": 0.8591174906964381,
|
7100 |
+
"grad_norm": 0.27558040618896484,
|
7101 |
+
"learning_rate": 9.836418453523833e-06,
|
7102 |
+
"loss": 2.6529,
|
7103 |
+
"step": 1010
|
7104 |
+
},
|
7105 |
+
{
|
7106 |
+
"epoch": 0.8599681020733653,
|
7107 |
+
"grad_norm": 0.25471609830856323,
|
7108 |
+
"learning_rate": 9.720217003425647e-06,
|
7109 |
+
"loss": 2.5865,
|
7110 |
+
"step": 1011
|
7111 |
+
},
|
7112 |
+
{
|
7113 |
+
"epoch": 0.8608187134502924,
|
7114 |
+
"grad_norm": 0.2609470784664154,
|
7115 |
+
"learning_rate": 9.60467093288121e-06,
|
7116 |
+
"loss": 2.5492,
|
7117 |
+
"step": 1012
|
7118 |
+
},
|
7119 |
+
{
|
7120 |
+
"epoch": 0.8616693248272196,
|
7121 |
+
"grad_norm": 0.29608237743377686,
|
7122 |
+
"learning_rate": 9.489781080688865e-06,
|
7123 |
+
"loss": 2.5104,
|
7124 |
+
"step": 1013
|
7125 |
+
},
|
7126 |
+
{
|
7127 |
+
"epoch": 0.8625199362041467,
|
7128 |
+
"grad_norm": 0.29410475492477417,
|
7129 |
+
"learning_rate": 9.375548280883128e-06,
|
7130 |
+
"loss": 2.6142,
|
7131 |
+
"step": 1014
|
7132 |
+
},
|
7133 |
+
{
|
7134 |
+
"epoch": 0.8633705475810739,
|
7135 |
+
"grad_norm": 0.269397497177124,
|
7136 |
+
"learning_rate": 9.261973362728827e-06,
|
7137 |
+
"loss": 2.5652,
|
7138 |
+
"step": 1015
|
7139 |
+
},
|
7140 |
+
{
|
7141 |
+
"epoch": 0.8642211589580011,
|
7142 |
+
"grad_norm": 0.28159505128860474,
|
7143 |
+
"learning_rate": 9.149057150714801e-06,
|
7144 |
+
"loss": 2.7499,
|
7145 |
+
"step": 1016
|
7146 |
+
},
|
7147 |
+
{
|
7148 |
+
"epoch": 0.8650717703349282,
|
7149 |
+
"grad_norm": 0.27257442474365234,
|
7150 |
+
"learning_rate": 9.036800464548157e-06,
|
7151 |
+
"loss": 2.6305,
|
7152 |
+
"step": 1017
|
7153 |
+
},
|
7154 |
+
{
|
7155 |
+
"epoch": 0.8659223817118554,
|
7156 |
+
"grad_norm": 0.26392441987991333,
|
7157 |
+
"learning_rate": 8.92520411914819e-06,
|
7158 |
+
"loss": 2.4767,
|
7159 |
+
"step": 1018
|
7160 |
+
},
|
7161 |
+
{
|
7162 |
+
"epoch": 0.8667729930887825,
|
7163 |
+
"grad_norm": 0.2779422998428345,
|
7164 |
+
"learning_rate": 8.814268924640468e-06,
|
7165 |
+
"loss": 2.6056,
|
7166 |
+
"step": 1019
|
7167 |
+
},
|
7168 |
+
{
|
7169 |
+
"epoch": 0.8676236044657097,
|
7170 |
+
"grad_norm": 0.27914363145828247,
|
7171 |
+
"learning_rate": 8.70399568635104e-06,
|
7172 |
+
"loss": 2.5612,
|
7173 |
+
"step": 1020
|
7174 |
+
},
|
7175 |
+
{
|
7176 |
+
"epoch": 0.8684742158426368,
|
7177 |
+
"grad_norm": 0.28274431824684143,
|
7178 |
+
"learning_rate": 8.594385204800482e-06,
|
7179 |
+
"loss": 2.5673,
|
7180 |
+
"step": 1021
|
7181 |
+
},
|
7182 |
+
{
|
7183 |
+
"epoch": 0.8693248272195641,
|
7184 |
+
"grad_norm": 0.30289795994758606,
|
7185 |
+
"learning_rate": 8.485438275698154e-06,
|
7186 |
+
"loss": 2.6624,
|
7187 |
+
"step": 1022
|
7188 |
+
},
|
7189 |
+
{
|
7190 |
+
"epoch": 0.8701754385964913,
|
7191 |
+
"grad_norm": 0.3084239363670349,
|
7192 |
+
"learning_rate": 8.377155689936434e-06,
|
7193 |
+
"loss": 2.5199,
|
7194 |
+
"step": 1023
|
7195 |
+
},
|
7196 |
+
{
|
7197 |
+
"epoch": 0.8710260499734184,
|
7198 |
+
"grad_norm": 0.32038623094558716,
|
7199 |
+
"learning_rate": 8.269538233584883e-06,
|
7200 |
+
"loss": 2.5174,
|
7201 |
+
"step": 1024
|
7202 |
+
},
|
7203 |
+
{
|
7204 |
+
"epoch": 0.8718766613503456,
|
7205 |
+
"grad_norm": 0.32201388478279114,
|
7206 |
+
"learning_rate": 8.162586687884654e-06,
|
7207 |
+
"loss": 2.5764,
|
7208 |
+
"step": 1025
|
7209 |
+
},
|
7210 |
+
{
|
7211 |
+
"epoch": 0.8727272727272727,
|
7212 |
+
"grad_norm": 0.31848224997520447,
|
7213 |
+
"learning_rate": 8.056301829242784e-06,
|
7214 |
+
"loss": 2.6327,
|
7215 |
+
"step": 1026
|
7216 |
+
},
|
7217 |
+
{
|
7218 |
+
"epoch": 0.8735778841041999,
|
7219 |
+
"grad_norm": 0.34866663813591003,
|
7220 |
+
"learning_rate": 7.950684429226463e-06,
|
7221 |
+
"loss": 2.8354,
|
7222 |
+
"step": 1027
|
7223 |
+
},
|
7224 |
+
{
|
7225 |
+
"epoch": 0.8744284954811271,
|
7226 |
+
"grad_norm": 0.3639064431190491,
|
7227 |
+
"learning_rate": 7.845735254557606e-06,
|
7228 |
+
"loss": 2.7546,
|
7229 |
+
"step": 1028
|
7230 |
+
},
|
7231 |
+
{
|
7232 |
+
"epoch": 0.8752791068580542,
|
7233 |
+
"grad_norm": 0.35733523964881897,
|
7234 |
+
"learning_rate": 7.741455067107162e-06,
|
7235 |
+
"loss": 2.69,
|
7236 |
+
"step": 1029
|
7237 |
+
},
|
7238 |
+
{
|
7239 |
+
"epoch": 0.8761297182349814,
|
7240 |
+
"grad_norm": 0.35862669348716736,
|
7241 |
+
"learning_rate": 7.637844623889556e-06,
|
7242 |
+
"loss": 2.5966,
|
7243 |
+
"step": 1030
|
7244 |
+
},
|
7245 |
+
{
|
7246 |
+
"epoch": 0.8769803296119085,
|
7247 |
+
"grad_norm": 0.36205005645751953,
|
7248 |
+
"learning_rate": 7.534904677057353e-06,
|
7249 |
+
"loss": 2.4966,
|
7250 |
+
"step": 1031
|
7251 |
+
},
|
7252 |
+
{
|
7253 |
+
"epoch": 0.8778309409888357,
|
7254 |
+
"grad_norm": 0.3843689560890198,
|
7255 |
+
"learning_rate": 7.4326359738956515e-06,
|
7256 |
+
"loss": 2.7942,
|
7257 |
+
"step": 1032
|
7258 |
+
},
|
7259 |
+
{
|
7260 |
+
"epoch": 0.8786815523657628,
|
7261 |
+
"grad_norm": 0.40694114565849304,
|
7262 |
+
"learning_rate": 7.331039256816663e-06,
|
7263 |
+
"loss": 2.739,
|
7264 |
+
"step": 1033
|
7265 |
+
},
|
7266 |
+
{
|
7267 |
+
"epoch": 0.87953216374269,
|
7268 |
+
"grad_norm": 0.40859633684158325,
|
7269 |
+
"learning_rate": 7.230115263354431e-06,
|
7270 |
+
"loss": 2.6879,
|
7271 |
+
"step": 1034
|
7272 |
+
},
|
7273 |
+
{
|
7274 |
+
"epoch": 0.8803827751196173,
|
7275 |
+
"grad_norm": 0.41907498240470886,
|
7276 |
+
"learning_rate": 7.129864726159408e-06,
|
7277 |
+
"loss": 2.5804,
|
7278 |
+
"step": 1035
|
7279 |
+
},
|
7280 |
+
{
|
7281 |
+
"epoch": 0.8812333864965444,
|
7282 |
+
"grad_norm": 0.4525774121284485,
|
7283 |
+
"learning_rate": 7.030288372993066e-06,
|
7284 |
+
"loss": 2.6814,
|
7285 |
+
"step": 1036
|
7286 |
+
},
|
7287 |
+
{
|
7288 |
+
"epoch": 0.8820839978734716,
|
7289 |
+
"grad_norm": 0.45679178833961487,
|
7290 |
+
"learning_rate": 6.931386926722772e-06,
|
7291 |
+
"loss": 2.7909,
|
7292 |
+
"step": 1037
|
7293 |
+
},
|
7294 |
+
{
|
7295 |
+
"epoch": 0.8829346092503987,
|
7296 |
+
"grad_norm": 0.47028636932373047,
|
7297 |
+
"learning_rate": 6.833161105316421e-06,
|
7298 |
+
"loss": 2.8758,
|
7299 |
+
"step": 1038
|
7300 |
+
},
|
7301 |
+
{
|
7302 |
+
"epoch": 0.8837852206273259,
|
7303 |
+
"grad_norm": 0.4830760359764099,
|
7304 |
+
"learning_rate": 6.7356116218372566e-06,
|
7305 |
+
"loss": 2.6979,
|
7306 |
+
"step": 1039
|
7307 |
+
},
|
7308 |
+
{
|
7309 |
+
"epoch": 0.8846358320042531,
|
7310 |
+
"grad_norm": 0.49996817111968994,
|
7311 |
+
"learning_rate": 6.63873918443868e-06,
|
7312 |
+
"loss": 2.8513,
|
7313 |
+
"step": 1040
|
7314 |
+
},
|
7315 |
+
{
|
7316 |
+
"epoch": 0.8854864433811802,
|
7317 |
+
"grad_norm": 0.4843011200428009,
|
7318 |
+
"learning_rate": 6.542544496359138e-06,
|
7319 |
+
"loss": 2.553,
|
7320 |
+
"step": 1041
|
7321 |
+
},
|
7322 |
+
{
|
7323 |
+
"epoch": 0.8863370547581074,
|
7324 |
+
"grad_norm": 0.5989816188812256,
|
7325 |
+
"learning_rate": 6.447028255917054e-06,
|
7326 |
+
"loss": 2.5985,
|
7327 |
+
"step": 1042
|
7328 |
+
},
|
7329 |
+
{
|
7330 |
+
"epoch": 0.8871876661350345,
|
7331 |
+
"grad_norm": 0.5418705940246582,
|
7332 |
+
"learning_rate": 6.352191156505627e-06,
|
7333 |
+
"loss": 2.7277,
|
7334 |
+
"step": 1043
|
7335 |
+
},
|
7336 |
+
{
|
7337 |
+
"epoch": 0.8880382775119617,
|
7338 |
+
"grad_norm": 0.5338729619979858,
|
7339 |
+
"learning_rate": 6.258033886587911e-06,
|
7340 |
+
"loss": 2.9324,
|
7341 |
+
"step": 1044
|
7342 |
+
},
|
7343 |
+
{
|
7344 |
+
"epoch": 0.8888888888888888,
|
7345 |
+
"grad_norm": 0.5353463292121887,
|
7346 |
+
"learning_rate": 6.164557129691828e-06,
|
7347 |
+
"loss": 2.6506,
|
7348 |
+
"step": 1045
|
7349 |
+
},
|
7350 |
+
{
|
7351 |
+
"epoch": 0.889739500265816,
|
7352 |
+
"grad_norm": 0.6170979142189026,
|
7353 |
+
"learning_rate": 6.0717615644051206e-06,
|
7354 |
+
"loss": 2.7699,
|
7355 |
+
"step": 1046
|
7356 |
+
},
|
7357 |
+
{
|
7358 |
+
"epoch": 0.8905901116427433,
|
7359 |
+
"grad_norm": 0.6048348546028137,
|
7360 |
+
"learning_rate": 5.979647864370486e-06,
|
7361 |
+
"loss": 2.7457,
|
7362 |
+
"step": 1047
|
7363 |
+
},
|
7364 |
+
{
|
7365 |
+
"epoch": 0.8914407230196704,
|
7366 |
+
"grad_norm": 0.6213808059692383,
|
7367 |
+
"learning_rate": 5.888216698280647e-06,
|
7368 |
+
"loss": 2.671,
|
7369 |
+
"step": 1048
|
7370 |
+
},
|
7371 |
+
{
|
7372 |
+
"epoch": 0.8922913343965976,
|
7373 |
+
"grad_norm": 0.6980977654457092,
|
7374 |
+
"learning_rate": 5.7974687298735454e-06,
|
7375 |
+
"loss": 2.9322,
|
7376 |
+
"step": 1049
|
7377 |
+
},
|
7378 |
+
{
|
7379 |
+
"epoch": 0.8931419457735247,
|
7380 |
+
"grad_norm": 0.762441873550415,
|
7381 |
+
"learning_rate": 5.7074046179275255e-06,
|
7382 |
+
"loss": 2.9242,
|
7383 |
+
"step": 1050
|
7384 |
+
},
|
7385 |
+
{
|
7386 |
+
"epoch": 0.8939925571504519,
|
7387 |
+
"grad_norm": 0.2550997734069824,
|
7388 |
+
"learning_rate": 5.6180250162564455e-06,
|
7389 |
+
"loss": 2.3309,
|
7390 |
+
"step": 1051
|
7391 |
+
},
|
7392 |
+
{
|
7393 |
+
"epoch": 0.8948431685273791,
|
7394 |
+
"grad_norm": 0.2799069285392761,
|
7395 |
+
"learning_rate": 5.5293305737050825e-06,
|
7396 |
+
"loss": 2.6428,
|
7397 |
+
"step": 1052
|
7398 |
+
},
|
7399 |
+
{
|
7400 |
+
"epoch": 0.8956937799043062,
|
7401 |
+
"grad_norm": 0.25860702991485596,
|
7402 |
+
"learning_rate": 5.441321934144339e-06,
|
7403 |
+
"loss": 2.2591,
|
7404 |
+
"step": 1053
|
7405 |
+
},
|
7406 |
+
{
|
7407 |
+
"epoch": 0.8965443912812334,
|
7408 |
+
"grad_norm": 0.262991726398468,
|
7409 |
+
"learning_rate": 5.35399973646653e-06,
|
7410 |
+
"loss": 2.3974,
|
7411 |
+
"step": 1054
|
7412 |
+
},
|
7413 |
+
{
|
7414 |
+
"epoch": 0.8973950026581605,
|
7415 |
+
"grad_norm": 0.25796017050743103,
|
7416 |
+
"learning_rate": 5.267364614580861e-06,
|
7417 |
+
"loss": 2.4487,
|
7418 |
+
"step": 1055
|
7419 |
+
},
|
7420 |
+
{
|
7421 |
+
"epoch": 0.8982456140350877,
|
7422 |
+
"grad_norm": 0.26405441761016846,
|
7423 |
+
"learning_rate": 5.181417197408734e-06,
|
7424 |
+
"loss": 2.3894,
|
7425 |
+
"step": 1056
|
7426 |
+
},
|
7427 |
+
{
|
7428 |
+
"epoch": 0.8990962254120148,
|
7429 |
+
"grad_norm": 0.27741700410842896,
|
7430 |
+
"learning_rate": 5.09615810887919e-06,
|
7431 |
+
"loss": 2.5089,
|
7432 |
+
"step": 1057
|
7433 |
+
},
|
7434 |
+
{
|
7435 |
+
"epoch": 0.899946836788942,
|
7436 |
+
"grad_norm": 0.27698126435279846,
|
7437 |
+
"learning_rate": 5.011587967924414e-06,
|
7438 |
+
"loss": 2.4868,
|
7439 |
+
"step": 1058
|
7440 |
+
},
|
7441 |
+
{
|
7442 |
+
"epoch": 0.9007974481658693,
|
7443 |
+
"grad_norm": 0.2699805796146393,
|
7444 |
+
"learning_rate": 4.927707388475255e-06,
|
7445 |
+
"loss": 2.4822,
|
7446 |
+
"step": 1059
|
7447 |
+
},
|
7448 |
+
{
|
7449 |
+
"epoch": 0.9016480595427964,
|
7450 |
+
"grad_norm": 0.2688996493816376,
|
7451 |
+
"learning_rate": 4.84451697945667e-06,
|
7452 |
+
"loss": 2.574,
|
7453 |
+
"step": 1060
|
7454 |
+
},
|
7455 |
+
{
|
7456 |
+
"epoch": 0.9024986709197236,
|
7457 |
+
"grad_norm": 0.2651573419570923,
|
7458 |
+
"learning_rate": 4.7620173447834425e-06,
|
7459 |
+
"loss": 2.5219,
|
7460 |
+
"step": 1061
|
7461 |
+
},
|
7462 |
+
{
|
7463 |
+
"epoch": 0.9033492822966507,
|
7464 |
+
"grad_norm": 0.27622607350349426,
|
7465 |
+
"learning_rate": 4.680209083355713e-06,
|
7466 |
+
"loss": 2.6289,
|
7467 |
+
"step": 1062
|
7468 |
+
},
|
7469 |
+
{
|
7470 |
+
"epoch": 0.9041998936735779,
|
7471 |
+
"grad_norm": 0.27151504158973694,
|
7472 |
+
"learning_rate": 4.5990927890545935e-06,
|
7473 |
+
"loss": 2.625,
|
7474 |
+
"step": 1063
|
7475 |
+
},
|
7476 |
+
{
|
7477 |
+
"epoch": 0.9050505050505051,
|
7478 |
+
"grad_norm": 0.2963936924934387,
|
7479 |
+
"learning_rate": 4.518669050737989e-06,
|
7480 |
+
"loss": 2.4124,
|
7481 |
+
"step": 1064
|
7482 |
+
},
|
7483 |
+
{
|
7484 |
+
"epoch": 0.9059011164274322,
|
7485 |
+
"grad_norm": 0.2703079879283905,
|
7486 |
+
"learning_rate": 4.438938452236219e-06,
|
7487 |
+
"loss": 2.482,
|
7488 |
+
"step": 1065
|
7489 |
+
},
|
7490 |
+
{
|
7491 |
+
"epoch": 0.9067517278043594,
|
7492 |
+
"grad_norm": 0.25973039865493774,
|
7493 |
+
"learning_rate": 4.359901572347758e-06,
|
7494 |
+
"loss": 2.6443,
|
7495 |
+
"step": 1066
|
7496 |
+
},
|
7497 |
+
{
|
7498 |
+
"epoch": 0.9076023391812865,
|
7499 |
+
"grad_norm": 0.27102747559547424,
|
7500 |
+
"learning_rate": 4.281558984835143e-06,
|
7501 |
+
"loss": 2.6452,
|
7502 |
+
"step": 1067
|
7503 |
+
},
|
7504 |
+
{
|
7505 |
+
"epoch": 0.9084529505582137,
|
7506 |
+
"grad_norm": 0.2813124656677246,
|
7507 |
+
"learning_rate": 4.203911258420712e-06,
|
7508 |
+
"loss": 2.6012,
|
7509 |
+
"step": 1068
|
7510 |
+
},
|
7511 |
+
{
|
7512 |
+
"epoch": 0.9093035619351408,
|
7513 |
+
"grad_norm": 0.274431437253952,
|
7514 |
+
"learning_rate": 4.126958956782545e-06,
|
7515 |
+
"loss": 2.6887,
|
7516 |
+
"step": 1069
|
7517 |
+
},
|
7518 |
+
{
|
7519 |
+
"epoch": 0.910154173312068,
|
7520 |
+
"grad_norm": 0.2668261229991913,
|
7521 |
+
"learning_rate": 4.050702638550275e-06,
|
7522 |
+
"loss": 2.5839,
|
7523 |
+
"step": 1070
|
7524 |
+
},
|
7525 |
+
{
|
7526 |
+
"epoch": 0.9110047846889953,
|
7527 |
+
"grad_norm": 0.2745119631290436,
|
7528 |
+
"learning_rate": 3.975142857301117e-06,
|
7529 |
+
"loss": 2.4371,
|
7530 |
+
"step": 1071
|
7531 |
+
},
|
7532 |
+
{
|
7533 |
+
"epoch": 0.9118553960659224,
|
7534 |
+
"grad_norm": 0.31567510962486267,
|
7535 |
+
"learning_rate": 3.900280161555881e-06,
|
7536 |
+
"loss": 2.4209,
|
7537 |
+
"step": 1072
|
7538 |
+
},
|
7539 |
+
{
|
7540 |
+
"epoch": 0.9127060074428496,
|
7541 |
+
"grad_norm": 0.28338539600372314,
|
7542 |
+
"learning_rate": 3.826115094774863e-06,
|
7543 |
+
"loss": 2.47,
|
7544 |
+
"step": 1073
|
7545 |
+
},
|
7546 |
+
{
|
7547 |
+
"epoch": 0.9135566188197767,
|
7548 |
+
"grad_norm": 0.3005933463573456,
|
7549 |
+
"learning_rate": 3.7526481953539915e-06,
|
7550 |
+
"loss": 2.5939,
|
7551 |
+
"step": 1074
|
7552 |
+
},
|
7553 |
+
{
|
7554 |
+
"epoch": 0.9144072301967039,
|
7555 |
+
"grad_norm": 0.31587204337120056,
|
7556 |
+
"learning_rate": 3.6798799966209497e-06,
|
7557 |
+
"loss": 2.6447,
|
7558 |
+
"step": 1075
|
7559 |
+
},
|
7560 |
+
{
|
7561 |
+
"epoch": 0.915257841573631,
|
7562 |
+
"grad_norm": 0.3279857635498047,
|
7563 |
+
"learning_rate": 3.607811026831176e-06,
|
7564 |
+
"loss": 2.5845,
|
7565 |
+
"step": 1076
|
7566 |
+
},
|
7567 |
+
{
|
7568 |
+
"epoch": 0.9161084529505582,
|
7569 |
+
"grad_norm": 0.3389846980571747,
|
7570 |
+
"learning_rate": 3.5364418091641373e-06,
|
7571 |
+
"loss": 2.7449,
|
7572 |
+
"step": 1077
|
7573 |
+
},
|
7574 |
+
{
|
7575 |
+
"epoch": 0.9169590643274854,
|
7576 |
+
"grad_norm": 0.3381056487560272,
|
7577 |
+
"learning_rate": 3.4657728617195295e-06,
|
7578 |
+
"loss": 2.407,
|
7579 |
+
"step": 1078
|
7580 |
+
},
|
7581 |
+
{
|
7582 |
+
"epoch": 0.9178096757044125,
|
7583 |
+
"grad_norm": 0.3725632429122925,
|
7584 |
+
"learning_rate": 3.3958046975134495e-06,
|
7585 |
+
"loss": 2.8884,
|
7586 |
+
"step": 1079
|
7587 |
+
},
|
7588 |
+
{
|
7589 |
+
"epoch": 0.9186602870813397,
|
7590 |
+
"grad_norm": 0.3629099726676941,
|
7591 |
+
"learning_rate": 3.32653782447474e-06,
|
7592 |
+
"loss": 2.6872,
|
7593 |
+
"step": 1080
|
7594 |
+
},
|
7595 |
+
{
|
7596 |
+
"epoch": 0.9195108984582668,
|
7597 |
+
"grad_norm": 0.37425747513771057,
|
7598 |
+
"learning_rate": 3.25797274544124e-06,
|
7599 |
+
"loss": 2.659,
|
7600 |
+
"step": 1081
|
7601 |
+
},
|
7602 |
+
{
|
7603 |
+
"epoch": 0.920361509835194,
|
7604 |
+
"grad_norm": 0.3709467649459839,
|
7605 |
+
"learning_rate": 3.1901099581561845e-06,
|
7606 |
+
"loss": 2.5593,
|
7607 |
+
"step": 1082
|
7608 |
+
},
|
7609 |
+
{
|
7610 |
+
"epoch": 0.9212121212121213,
|
7611 |
+
"grad_norm": 0.37324512004852295,
|
7612 |
+
"learning_rate": 3.122949955264587e-06,
|
7613 |
+
"loss": 2.5461,
|
7614 |
+
"step": 1083
|
7615 |
+
},
|
7616 |
+
{
|
7617 |
+
"epoch": 0.9220627325890484,
|
7618 |
+
"grad_norm": 0.39722326397895813,
|
7619 |
+
"learning_rate": 3.0564932243095866e-06,
|
7620 |
+
"loss": 2.8043,
|
7621 |
+
"step": 1084
|
7622 |
+
},
|
7623 |
+
{
|
7624 |
+
"epoch": 0.9229133439659756,
|
7625 |
+
"grad_norm": 0.40863049030303955,
|
7626 |
+
"learning_rate": 2.9907402477290514e-06,
|
7627 |
+
"loss": 2.5307,
|
7628 |
+
"step": 1085
|
7629 |
+
},
|
7630 |
+
{
|
7631 |
+
"epoch": 0.9237639553429027,
|
7632 |
+
"grad_norm": 0.4645911157131195,
|
7633 |
+
"learning_rate": 2.9256915028519573e-06,
|
7634 |
+
"loss": 2.3838,
|
7635 |
+
"step": 1086
|
7636 |
+
},
|
7637 |
+
{
|
7638 |
+
"epoch": 0.9246145667198299,
|
7639 |
+
"grad_norm": 0.42394137382507324,
|
7640 |
+
"learning_rate": 2.8613474618949366e-06,
|
7641 |
+
"loss": 2.7368,
|
7642 |
+
"step": 1087
|
7643 |
+
},
|
7644 |
+
{
|
7645 |
+
"epoch": 0.925465178096757,
|
7646 |
+
"grad_norm": 0.44541221857070923,
|
7647 |
+
"learning_rate": 2.7977085919589254e-06,
|
7648 |
+
"loss": 2.6973,
|
7649 |
+
"step": 1088
|
7650 |
+
},
|
7651 |
+
{
|
7652 |
+
"epoch": 0.9263157894736842,
|
7653 |
+
"grad_norm": 0.45734384655952454,
|
7654 |
+
"learning_rate": 2.7347753550256872e-06,
|
7655 |
+
"loss": 2.8454,
|
7656 |
+
"step": 1089
|
7657 |
+
},
|
7658 |
+
{
|
7659 |
+
"epoch": 0.9271664008506114,
|
7660 |
+
"grad_norm": 0.4581248164176941,
|
7661 |
+
"learning_rate": 2.672548207954495e-06,
|
7662 |
+
"loss": 2.7356,
|
7663 |
+
"step": 1090
|
7664 |
+
},
|
7665 |
+
{
|
7666 |
+
"epoch": 0.9280170122275385,
|
7667 |
+
"grad_norm": 0.45908254384994507,
|
7668 |
+
"learning_rate": 2.6110276024788214e-06,
|
7669 |
+
"loss": 2.5562,
|
7670 |
+
"step": 1091
|
7671 |
+
},
|
7672 |
+
{
|
7673 |
+
"epoch": 0.9288676236044657,
|
7674 |
+
"grad_norm": 0.4975565969944,
|
7675 |
+
"learning_rate": 2.550213985203076e-06,
|
7676 |
+
"loss": 2.8292,
|
7677 |
+
"step": 1092
|
7678 |
+
},
|
7679 |
+
{
|
7680 |
+
"epoch": 0.9297182349813928,
|
7681 |
+
"grad_norm": 0.5445611476898193,
|
7682 |
+
"learning_rate": 2.4901077975992838e-06,
|
7683 |
+
"loss": 2.9136,
|
7684 |
+
"step": 1093
|
7685 |
+
},
|
7686 |
+
{
|
7687 |
+
"epoch": 0.93056884635832,
|
7688 |
+
"grad_norm": 0.5516471862792969,
|
7689 |
+
"learning_rate": 2.4307094760039785e-06,
|
7690 |
+
"loss": 3.0927,
|
7691 |
+
"step": 1094
|
7692 |
+
},
|
7693 |
+
{
|
7694 |
+
"epoch": 0.9314194577352473,
|
7695 |
+
"grad_norm": 0.5213038921356201,
|
7696 |
+
"learning_rate": 2.3720194516149818e-06,
|
7697 |
+
"loss": 2.705,
|
7698 |
+
"step": 1095
|
7699 |
+
},
|
7700 |
+
{
|
7701 |
+
"epoch": 0.9322700691121744,
|
7702 |
+
"grad_norm": 0.5873907208442688,
|
7703 |
+
"learning_rate": 2.3140381504882737e-06,
|
7704 |
+
"loss": 2.8415,
|
7705 |
+
"step": 1096
|
7706 |
+
},
|
7707 |
+
{
|
7708 |
+
"epoch": 0.9331206804891016,
|
7709 |
+
"grad_norm": 0.6056773662567139,
|
7710 |
+
"learning_rate": 2.2567659935349372e-06,
|
7711 |
+
"loss": 2.6933,
|
7712 |
+
"step": 1097
|
7713 |
+
},
|
7714 |
+
{
|
7715 |
+
"epoch": 0.9339712918660287,
|
7716 |
+
"grad_norm": 0.6297914981842041,
|
7717 |
+
"learning_rate": 2.200203396517997e-06,
|
7718 |
+
"loss": 2.8197,
|
7719 |
+
"step": 1098
|
7720 |
+
},
|
7721 |
+
{
|
7722 |
+
"epoch": 0.9348219032429559,
|
7723 |
+
"grad_norm": 0.6759224534034729,
|
7724 |
+
"learning_rate": 2.144350770049597e-06,
|
7725 |
+
"loss": 2.7496,
|
7726 |
+
"step": 1099
|
7727 |
+
},
|
7728 |
+
{
|
7729 |
+
"epoch": 0.935672514619883,
|
7730 |
+
"grad_norm": 0.7379116415977478,
|
7731 |
+
"learning_rate": 2.0892085195878154e-06,
|
7732 |
+
"loss": 2.926,
|
7733 |
+
"step": 1100
|
7734 |
+
},
|
7735 |
+
{
|
7736 |
+
"epoch": 0.9365231259968102,
|
7737 |
+
"grad_norm": 0.25760582089424133,
|
7738 |
+
"learning_rate": 2.034777045433811e-06,
|
7739 |
+
"loss": 2.3794,
|
7740 |
+
"step": 1101
|
7741 |
+
},
|
7742 |
+
{
|
7743 |
+
"epoch": 0.9373737373737374,
|
7744 |
+
"grad_norm": 0.2775615155696869,
|
7745 |
+
"learning_rate": 1.9810567427289595e-06,
|
7746 |
+
"loss": 2.5169,
|
7747 |
+
"step": 1102
|
7748 |
+
},
|
7749 |
+
{
|
7750 |
+
"epoch": 0.9382243487506645,
|
7751 |
+
"grad_norm": 0.26485174894332886,
|
7752 |
+
"learning_rate": 1.92804800145191e-06,
|
7753 |
+
"loss": 2.4342,
|
7754 |
+
"step": 1103
|
7755 |
+
},
|
7756 |
+
{
|
7757 |
+
"epoch": 0.9390749601275917,
|
7758 |
+
"grad_norm": 0.24917316436767578,
|
7759 |
+
"learning_rate": 1.8757512064157656e-06,
|
7760 |
+
"loss": 2.4177,
|
7761 |
+
"step": 1104
|
7762 |
+
},
|
7763 |
+
{
|
7764 |
+
"epoch": 0.9399255715045188,
|
7765 |
+
"grad_norm": 0.25125351548194885,
|
7766 |
+
"learning_rate": 1.8241667372653316e-06,
|
7767 |
+
"loss": 2.2352,
|
7768 |
+
"step": 1105
|
7769 |
+
},
|
7770 |
+
{
|
7771 |
+
"epoch": 0.940776182881446,
|
7772 |
+
"grad_norm": 0.2972431182861328,
|
7773 |
+
"learning_rate": 1.7732949684743594e-06,
|
7774 |
+
"loss": 2.4803,
|
7775 |
+
"step": 1106
|
7776 |
+
},
|
7777 |
+
{
|
7778 |
+
"epoch": 0.9416267942583733,
|
7779 |
+
"grad_norm": 0.2540951073169708,
|
7780 |
+
"learning_rate": 1.7231362693427288e-06,
|
7781 |
+
"loss": 2.5117,
|
7782 |
+
"step": 1107
|
7783 |
+
},
|
7784 |
+
{
|
7785 |
+
"epoch": 0.9424774056353004,
|
7786 |
+
"grad_norm": 0.26120516657829285,
|
7787 |
+
"learning_rate": 1.6736910039939157e-06,
|
7788 |
+
"loss": 2.3744,
|
7789 |
+
"step": 1108
|
7790 |
+
},
|
7791 |
+
{
|
7792 |
+
"epoch": 0.9433280170122276,
|
7793 |
+
"grad_norm": 0.2617899477481842,
|
7794 |
+
"learning_rate": 1.62495953137225e-06,
|
7795 |
+
"loss": 2.4657,
|
7796 |
+
"step": 1109
|
7797 |
+
},
|
7798 |
+
{
|
7799 |
+
"epoch": 0.9441786283891547,
|
7800 |
+
"grad_norm": 0.29227423667907715,
|
7801 |
+
"learning_rate": 1.576942205240317e-06,
|
7802 |
+
"loss": 2.4297,
|
7803 |
+
"step": 1110
|
7804 |
+
},
|
7805 |
+
{
|
7806 |
+
"epoch": 0.9450292397660819,
|
7807 |
+
"grad_norm": 0.2886480689048767,
|
7808 |
+
"learning_rate": 1.5296393741764391e-06,
|
7809 |
+
"loss": 2.5686,
|
7810 |
+
"step": 1111
|
7811 |
+
},
|
7812 |
+
{
|
7813 |
+
"epoch": 0.945879851143009,
|
7814 |
+
"grad_norm": 0.27778005599975586,
|
7815 |
+
"learning_rate": 1.4830513815720759e-06,
|
7816 |
+
"loss": 2.5925,
|
7817 |
+
"step": 1112
|
7818 |
+
},
|
7819 |
+
{
|
7820 |
+
"epoch": 0.9467304625199362,
|
7821 |
+
"grad_norm": 0.2743171155452728,
|
7822 |
+
"learning_rate": 1.4371785656294046e-06,
|
7823 |
+
"loss": 2.6989,
|
7824 |
+
"step": 1113
|
7825 |
+
},
|
7826 |
+
{
|
7827 |
+
"epoch": 0.9475810738968634,
|
7828 |
+
"grad_norm": 0.27193981409072876,
|
7829 |
+
"learning_rate": 1.392021259358811e-06,
|
7830 |
+
"loss": 2.457,
|
7831 |
+
"step": 1114
|
7832 |
+
},
|
7833 |
+
{
|
7834 |
+
"epoch": 0.9484316852737905,
|
7835 |
+
"grad_norm": 0.2713122069835663,
|
7836 |
+
"learning_rate": 1.3475797905764809e-06,
|
7837 |
+
"loss": 2.6029,
|
7838 |
+
"step": 1115
|
7839 |
+
},
|
7840 |
+
{
|
7841 |
+
"epoch": 0.9492822966507177,
|
7842 |
+
"grad_norm": 0.25593551993370056,
|
7843 |
+
"learning_rate": 1.303854481902067e-06,
|
7844 |
+
"loss": 2.6974,
|
7845 |
+
"step": 1116
|
7846 |
+
},
|
7847 |
+
{
|
7848 |
+
"epoch": 0.9501329080276448,
|
7849 |
+
"grad_norm": 0.26953616738319397,
|
7850 |
+
"learning_rate": 1.2608456507562705e-06,
|
7851 |
+
"loss": 2.6234,
|
7852 |
+
"step": 1117
|
7853 |
+
},
|
7854 |
+
{
|
7855 |
+
"epoch": 0.950983519404572,
|
7856 |
+
"grad_norm": 0.28958868980407715,
|
7857 |
+
"learning_rate": 1.2185536093585747e-06,
|
7858 |
+
"loss": 2.6315,
|
7859 |
+
"step": 1118
|
7860 |
+
},
|
7861 |
+
{
|
7862 |
+
"epoch": 0.9518341307814993,
|
7863 |
+
"grad_norm": 0.263473242521286,
|
7864 |
+
"learning_rate": 1.1769786647250147e-06,
|
7865 |
+
"loss": 2.5595,
|
7866 |
+
"step": 1119
|
7867 |
+
},
|
7868 |
+
{
|
7869 |
+
"epoch": 0.9526847421584264,
|
7870 |
+
"grad_norm": 0.26081758737564087,
|
7871 |
+
"learning_rate": 1.1361211186658894e-06,
|
7872 |
+
"loss": 2.4885,
|
7873 |
+
"step": 1120
|
7874 |
+
},
|
7875 |
+
{
|
7876 |
+
"epoch": 0.9535353535353536,
|
7877 |
+
"grad_norm": 0.275305837392807,
|
7878 |
+
"learning_rate": 1.0959812677835968e-06,
|
7879 |
+
"loss": 2.794,
|
7880 |
+
"step": 1121
|
7881 |
+
},
|
7882 |
+
{
|
7883 |
+
"epoch": 0.9543859649122807,
|
7884 |
+
"grad_norm": 0.28377169370651245,
|
7885 |
+
"learning_rate": 1.0565594034704918e-06,
|
7886 |
+
"loss": 2.7399,
|
7887 |
+
"step": 1122
|
7888 |
+
},
|
7889 |
+
{
|
7890 |
+
"epoch": 0.9552365762892079,
|
7891 |
+
"grad_norm": 0.282060831785202,
|
7892 |
+
"learning_rate": 1.0178558119067315e-06,
|
7893 |
+
"loss": 2.4815,
|
7894 |
+
"step": 1123
|
7895 |
+
},
|
7896 |
+
{
|
7897 |
+
"epoch": 0.956087187666135,
|
7898 |
+
"grad_norm": 0.2975423038005829,
|
7899 |
+
"learning_rate": 9.798707740582447e-07,
|
7900 |
+
"loss": 2.5839,
|
7901 |
+
"step": 1124
|
7902 |
+
},
|
7903 |
+
{
|
7904 |
+
"epoch": 0.9569377990430622,
|
7905 |
+
"grad_norm": 0.3122364580631256,
|
7906 |
+
"learning_rate": 9.42604565674654e-07,
|
7907 |
+
"loss": 2.6381,
|
7908 |
+
"step": 1125
|
7909 |
+
},
|
7910 |
+
{
|
7911 |
+
"epoch": 0.9577884104199894,
|
7912 |
+
"grad_norm": 0.3378431797027588,
|
7913 |
+
"learning_rate": 9.060574572873237e-07,
|
7914 |
+
"loss": 2.5897,
|
7915 |
+
"step": 1126
|
7916 |
+
},
|
7917 |
+
{
|
7918 |
+
"epoch": 0.9586390217969165,
|
7919 |
+
"grad_norm": 0.33718162775039673,
|
7920 |
+
"learning_rate": 8.702297142073379e-07,
|
7921 |
+
"loss": 2.689,
|
7922 |
+
"step": 1127
|
7923 |
+
},
|
7924 |
+
{
|
7925 |
+
"epoch": 0.9594896331738437,
|
7926 |
+
"grad_norm": 0.34144076704978943,
|
7927 |
+
"learning_rate": 8.351215965235915e-07,
|
7928 |
+
"loss": 2.4692,
|
7929 |
+
"step": 1128
|
7930 |
+
},
|
7931 |
+
{
|
7932 |
+
"epoch": 0.9603402445507708,
|
7933 |
+
"grad_norm": 0.3324032425880432,
|
7934 |
+
"learning_rate": 8.007333591009358e-07,
|
7935 |
+
"loss": 2.653,
|
7936 |
+
"step": 1129
|
7937 |
+
},
|
7938 |
+
{
|
7939 |
+
"epoch": 0.961190855927698,
|
7940 |
+
"grad_norm": 0.3601853847503662,
|
7941 |
+
"learning_rate": 7.670652515782917e-07,
|
7942 |
+
"loss": 2.5995,
|
7943 |
+
"step": 1130
|
7944 |
+
},
|
7945 |
+
{
|
7946 |
+
"epoch": 0.9620414673046253,
|
7947 |
+
"grad_norm": 0.3701609969139099,
|
7948 |
+
"learning_rate": 7.341175183668503e-07,
|
7949 |
+
"loss": 2.6668,
|
7950 |
+
"step": 1131
|
7951 |
+
},
|
7952 |
+
{
|
7953 |
+
"epoch": 0.9628920786815524,
|
7954 |
+
"grad_norm": 0.37852516770362854,
|
7955 |
+
"learning_rate": 7.018903986483083e-07,
|
7956 |
+
"loss": 2.7223,
|
7957 |
+
"step": 1132
|
7958 |
+
},
|
7959 |
+
{
|
7960 |
+
"epoch": 0.9637426900584796,
|
7961 |
+
"grad_norm": 0.391939640045166,
|
7962 |
+
"learning_rate": 6.703841263730914e-07,
|
7963 |
+
"loss": 2.7923,
|
7964 |
+
"step": 1133
|
7965 |
+
},
|
7966 |
+
{
|
7967 |
+
"epoch": 0.9645933014354067,
|
7968 |
+
"grad_norm": 0.38707369565963745,
|
7969 |
+
"learning_rate": 6.395989302587113e-07,
|
7970 |
+
"loss": 2.6445,
|
7971 |
+
"step": 1134
|
7972 |
+
},
|
7973 |
+
{
|
7974 |
+
"epoch": 0.9654439128123339,
|
7975 |
+
"grad_norm": 0.4049709141254425,
|
7976 |
+
"learning_rate": 6.095350337880667e-07,
|
7977 |
+
"loss": 2.7038,
|
7978 |
+
"step": 1135
|
7979 |
+
},
|
7980 |
+
{
|
7981 |
+
"epoch": 0.966294524189261,
|
7982 |
+
"grad_norm": 0.41350454092025757,
|
7983 |
+
"learning_rate": 5.801926552078563e-07,
|
7984 |
+
"loss": 2.6472,
|
7985 |
+
"step": 1136
|
7986 |
+
},
|
7987 |
+
{
|
7988 |
+
"epoch": 0.9671451355661882,
|
7989 |
+
"grad_norm": 0.42945706844329834,
|
7990 |
+
"learning_rate": 5.515720075269348e-07,
|
7991 |
+
"loss": 2.7802,
|
7992 |
+
"step": 1137
|
7993 |
+
},
|
7994 |
+
{
|
7995 |
+
"epoch": 0.9679957469431154,
|
7996 |
+
"grad_norm": 0.4451653063297272,
|
7997 |
+
"learning_rate": 5.236732985148374e-07,
|
7998 |
+
"loss": 2.7633,
|
7999 |
+
"step": 1138
|
8000 |
+
},
|
8001 |
+
{
|
8002 |
+
"epoch": 0.9688463583200425,
|
8003 |
+
"grad_norm": 0.46683037281036377,
|
8004 |
+
"learning_rate": 4.964967307002244e-07,
|
8005 |
+
"loss": 2.8216,
|
8006 |
+
"step": 1139
|
8007 |
+
},
|
8008 |
+
{
|
8009 |
+
"epoch": 0.9696969696969697,
|
8010 |
+
"grad_norm": 0.4562697112560272,
|
8011 |
+
"learning_rate": 4.7004250136940543e-07,
|
8012 |
+
"loss": 2.7924,
|
8013 |
+
"step": 1140
|
8014 |
+
},
|
8015 |
+
{
|
8016 |
+
"epoch": 0.9705475810738968,
|
8017 |
+
"grad_norm": 0.48352983593940735,
|
8018 |
+
"learning_rate": 4.443108025649623e-07,
|
8019 |
+
"loss": 2.5207,
|
8020 |
+
"step": 1141
|
8021 |
+
},
|
8022 |
+
{
|
8023 |
+
"epoch": 0.971398192450824,
|
8024 |
+
"grad_norm": 0.4667017459869385,
|
8025 |
+
"learning_rate": 4.193018210843058e-07,
|
8026 |
+
"loss": 2.6552,
|
8027 |
+
"step": 1142
|
8028 |
+
},
|
8029 |
+
{
|
8030 |
+
"epoch": 0.9722488038277513,
|
8031 |
+
"grad_norm": 0.49360570311546326,
|
8032 |
+
"learning_rate": 3.950157384783104e-07,
|
8033 |
+
"loss": 2.7523,
|
8034 |
+
"step": 1143
|
8035 |
+
},
|
8036 |
+
{
|
8037 |
+
"epoch": 0.9730994152046784,
|
8038 |
+
"grad_norm": 0.526669442653656,
|
8039 |
+
"learning_rate": 3.714527310500371e-07,
|
8040 |
+
"loss": 2.7332,
|
8041 |
+
"step": 1144
|
8042 |
+
},
|
8043 |
+
{
|
8044 |
+
"epoch": 0.9739500265816056,
|
8045 |
+
"grad_norm": 0.5592300891876221,
|
8046 |
+
"learning_rate": 3.486129698534457e-07,
|
8047 |
+
"loss": 2.7264,
|
8048 |
+
"step": 1145
|
8049 |
+
},
|
8050 |
+
{
|
8051 |
+
"epoch": 0.9748006379585327,
|
8052 |
+
"grad_norm": 0.5670251250267029,
|
8053 |
+
"learning_rate": 3.264966206921294e-07,
|
8054 |
+
"loss": 2.8538,
|
8055 |
+
"step": 1146
|
8056 |
+
},
|
8057 |
+
{
|
8058 |
+
"epoch": 0.9756512493354599,
|
8059 |
+
"grad_norm": 0.5744861960411072,
|
8060 |
+
"learning_rate": 3.0510384411812644e-07,
|
8061 |
+
"loss": 2.6884,
|
8062 |
+
"step": 1147
|
8063 |
+
},
|
8064 |
+
{
|
8065 |
+
"epoch": 0.976501860712387,
|
8066 |
+
"grad_norm": 0.6117346882820129,
|
8067 |
+
"learning_rate": 2.844347954307325e-07,
|
8068 |
+
"loss": 2.8192,
|
8069 |
+
"step": 1148
|
8070 |
+
},
|
8071 |
+
{
|
8072 |
+
"epoch": 0.9773524720893142,
|
8073 |
+
"grad_norm": 0.7534440159797668,
|
8074 |
+
"learning_rate": 2.644896246754236e-07,
|
8075 |
+
"loss": 2.7148,
|
8076 |
+
"step": 1149
|
8077 |
+
},
|
8078 |
+
{
|
8079 |
+
"epoch": 0.9782030834662414,
|
8080 |
+
"grad_norm": 0.7108234167098999,
|
8081 |
+
"learning_rate": 2.452684766427349e-07,
|
8082 |
+
"loss": 2.8751,
|
8083 |
+
"step": 1150
|
8084 |
+
},
|
8085 |
+
{
|
8086 |
+
"epoch": 0.9790536948431685,
|
8087 |
+
"grad_norm": 0.264021635055542,
|
8088 |
+
"learning_rate": 2.2677149086718364e-07,
|
8089 |
+
"loss": 2.6035,
|
8090 |
+
"step": 1151
|
8091 |
+
},
|
8092 |
+
{
|
8093 |
+
"epoch": 0.9799043062200957,
|
8094 |
+
"grad_norm": 0.2746794819831848,
|
8095 |
+
"learning_rate": 2.0899880162630336e-07,
|
8096 |
+
"loss": 2.5935,
|
8097 |
+
"step": 1152
|
8098 |
+
},
|
8099 |
+
{
|
8100 |
+
"epoch": 0.9807549175970228,
|
8101 |
+
"grad_norm": 0.2742723226547241,
|
8102 |
+
"learning_rate": 1.9195053793964468e-07,
|
8103 |
+
"loss": 2.4199,
|
8104 |
+
"step": 1153
|
8105 |
+
},
|
8106 |
+
{
|
8107 |
+
"epoch": 0.98160552897395,
|
8108 |
+
"grad_norm": 0.2900630235671997,
|
8109 |
+
"learning_rate": 1.7562682356786487e-07,
|
8110 |
+
"loss": 2.5906,
|
8111 |
+
"step": 1154
|
8112 |
+
},
|
8113 |
+
{
|
8114 |
+
"epoch": 0.9824561403508771,
|
8115 |
+
"grad_norm": 0.26273003220558167,
|
8116 |
+
"learning_rate": 1.6002777701175086e-07,
|
8117 |
+
"loss": 2.4287,
|
8118 |
+
"step": 1155
|
8119 |
+
},
|
8120 |
+
{
|
8121 |
+
"epoch": 0.9833067517278044,
|
8122 |
+
"grad_norm": 0.28278976678848267,
|
8123 |
+
"learning_rate": 1.451535115114866e-07,
|
8124 |
+
"loss": 2.6875,
|
8125 |
+
"step": 1156
|
8126 |
+
},
|
8127 |
+
{
|
8128 |
+
"epoch": 0.9841573631047316,
|
8129 |
+
"grad_norm": 0.2644931674003601,
|
8130 |
+
"learning_rate": 1.310041350457092e-07,
|
8131 |
+
"loss": 2.5806,
|
8132 |
+
"step": 1157
|
8133 |
+
},
|
8134 |
+
{
|
8135 |
+
"epoch": 0.9850079744816587,
|
8136 |
+
"grad_norm": 0.26039785146713257,
|
8137 |
+
"learning_rate": 1.1757975033078739e-07,
|
8138 |
+
"loss": 2.2813,
|
8139 |
+
"step": 1158
|
8140 |
+
},
|
8141 |
+
{
|
8142 |
+
"epoch": 0.9858585858585859,
|
8143 |
+
"grad_norm": 0.27614063024520874,
|
8144 |
+
"learning_rate": 1.0488045482008879e-07,
|
8145 |
+
"loss": 2.6836,
|
8146 |
+
"step": 1159
|
8147 |
+
},
|
8148 |
+
{
|
8149 |
+
"epoch": 0.986709197235513,
|
8150 |
+
"grad_norm": 0.27048394083976746,
|
8151 |
+
"learning_rate": 9.29063407032249e-08,
|
8152 |
+
"loss": 2.5862,
|
8153 |
+
"step": 1160
|
8154 |
+
},
|
8155 |
+
{
|
8156 |
+
"epoch": 0.9875598086124402,
|
8157 |
+
"grad_norm": 0.28808289766311646,
|
8158 |
+
"learning_rate": 8.16574949054072e-08,
|
8159 |
+
"loss": 2.765,
|
8160 |
+
"step": 1161
|
8161 |
+
},
|
8162 |
+
{
|
8163 |
+
"epoch": 0.9884104199893674,
|
8164 |
+
"grad_norm": 0.323574423789978,
|
8165 |
+
"learning_rate": 7.113399908681429e-08,
|
8166 |
+
"loss": 2.6432,
|
8167 |
+
"step": 1162
|
8168 |
+
},
|
8169 |
+
{
|
8170 |
+
"epoch": 0.9892610313662945,
|
8171 |
+
"grad_norm": 0.3291761875152588,
|
8172 |
+
"learning_rate": 6.133592964201463e-08,
|
8173 |
+
"loss": 2.6037,
|
8174 |
+
"step": 1163
|
8175 |
+
},
|
8176 |
+
{
|
8177 |
+
"epoch": 0.9901116427432217,
|
8178 |
+
"grad_norm": 0.345300555229187,
|
8179 |
+
"learning_rate": 5.226335769936697e-08,
|
8180 |
+
"loss": 2.6076,
|
8181 |
+
"step": 1164
|
8182 |
+
},
|
8183 |
+
{
|
8184 |
+
"epoch": 0.9909622541201488,
|
8185 |
+
"grad_norm": 0.37476396560668945,
|
8186 |
+
"learning_rate": 4.391634912056519e-08,
|
8187 |
+
"loss": 2.8919,
|
8188 |
+
"step": 1165
|
8189 |
+
},
|
8190 |
+
{
|
8191 |
+
"epoch": 0.991812865497076,
|
8192 |
+
"grad_norm": 0.36915716528892517,
|
8193 |
+
"learning_rate": 3.629496450011649e-08,
|
8194 |
+
"loss": 2.5514,
|
8195 |
+
"step": 1166
|
8196 |
+
},
|
8197 |
+
{
|
8198 |
+
"epoch": 0.9926634768740031,
|
8199 |
+
"grad_norm": 0.3973197937011719,
|
8200 |
+
"learning_rate": 2.9399259164897274e-08,
|
8201 |
+
"loss": 2.8279,
|
8202 |
+
"step": 1167
|
8203 |
+
},
|
8204 |
+
{
|
8205 |
+
"epoch": 0.9935140882509303,
|
8206 |
+
"grad_norm": 0.42460885643959045,
|
8207 |
+
"learning_rate": 2.322928317378681e-08,
|
8208 |
+
"loss": 2.5969,
|
8209 |
+
"step": 1168
|
8210 |
+
},
|
8211 |
+
{
|
8212 |
+
"epoch": 0.9943646996278576,
|
8213 |
+
"grad_norm": 0.4290263056755066,
|
8214 |
+
"learning_rate": 1.778508131728973e-08,
|
8215 |
+
"loss": 2.7255,
|
8216 |
+
"step": 1169
|
8217 |
+
},
|
8218 |
+
{
|
8219 |
+
"epoch": 0.9952153110047847,
|
8220 |
+
"grad_norm": 0.4642498195171356,
|
8221 |
+
"learning_rate": 1.3066693117191886e-08,
|
8222 |
+
"loss": 2.7869,
|
8223 |
+
"step": 1170
|
8224 |
+
},
|
8225 |
+
{
|
8226 |
+
"epoch": 0.9960659223817119,
|
8227 |
+
"grad_norm": 0.5059999823570251,
|
8228 |
+
"learning_rate": 9.074152826271665e-09,
|
8229 |
+
"loss": 2.582,
|
8230 |
+
"step": 1171
|
8231 |
+
},
|
8232 |
+
{
|
8233 |
+
"epoch": 0.996916533758639,
|
8234 |
+
"grad_norm": 0.49888238310813904,
|
8235 |
+
"learning_rate": 5.807489428111268e-09,
|
8236 |
+
"loss": 2.8126,
|
8237 |
+
"step": 1172
|
8238 |
+
},
|
8239 |
+
{
|
8240 |
+
"epoch": 0.9977671451355662,
|
8241 |
+
"grad_norm": 0.5143745541572571,
|
8242 |
+
"learning_rate": 3.2667266368080484e-09,
|
8243 |
+
"loss": 2.7024,
|
8244 |
+
"step": 1173
|
8245 |
+
},
|
8246 |
+
{
|
8247 |
+
"epoch": 0.9986177565124934,
|
8248 |
+
"grad_norm": 0.5905907154083252,
|
8249 |
+
"learning_rate": 1.4518828968523857e-09,
|
8250 |
+
"loss": 2.7198,
|
8251 |
+
"step": 1174
|
8252 |
+
},
|
8253 |
+
{
|
8254 |
+
"epoch": 0.9994683678894205,
|
8255 |
+
"grad_norm": 0.6912564635276794,
|
8256 |
+
"learning_rate": 3.629713829500503e-10,
|
8257 |
+
"loss": 2.6901,
|
8258 |
+
"step": 1175
|
8259 |
+
},
|
8260 |
+
{
|
8261 |
+
"epoch": 1.0006379585326954,
|
8262 |
+
"grad_norm": 1.383741021156311,
|
8263 |
+
"learning_rate": 0.0,
|
8264 |
+
"loss": 4.1919,
|
8265 |
+
"step": 1176
|
8266 |
+
},
|
8267 |
+
{
|
8268 |
+
"epoch": 1.0006379585326954,
|
8269 |
+
"eval_loss": 2.6547293663024902,
|
8270 |
+
"eval_runtime": 80.6568,
|
8271 |
+
"eval_samples_per_second": 12.274,
|
8272 |
+
"eval_steps_per_second": 6.137,
|
8273 |
+
"step": 1176
|
8274 |
}
|
8275 |
],
|
8276 |
"logging_steps": 1,
|
|
|
8285 |
"should_evaluate": false,
|
8286 |
"should_log": false,
|
8287 |
"should_save": true,
|
8288 |
+
"should_training_stop": true
|
8289 |
},
|
8290 |
"attributes": {}
|
8291 |
}
|
8292 |
},
|
8293 |
+
"total_flos": 1.199442414922629e+18,
|
8294 |
"train_batch_size": 2,
|
8295 |
"trial_name": null,
|
8296 |
"trial_params": null
|