Training in progress, step 766, checkpoint
Browse files
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 101184122
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f5ca2d13a2d5c86e8d05ce17700618910b2596036ba51be793c5a9a0a244600b
|
| 3 |
size 101184122
|
last-checkpoint/rng_state_0.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14512
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:46fae2c30f4c8f4c1d015a0d34cf35c573f4f7bb737577048eae036e3b52402f
|
| 3 |
size 14512
|
last-checkpoint/rng_state_1.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14512
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:af7fc61f695a3c7b1e847889bacf200cd500602d65daf2881dae5d8e706098fd
|
| 3 |
size 14512
|
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:fc3adba656a0095bd6d7e2fff6d64eecd67bc391ad34d1bbda74959b48a5b807
|
| 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": 0.
|
| 5 |
"eval_steps": 383,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -2704,6 +2704,2695 @@
|
|
| 2704 |
"eval_samples_per_second": 16.076,
|
| 2705 |
"eval_steps_per_second": 4.022,
|
| 2706 |
"step": 383
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 2707 |
}
|
| 2708 |
],
|
| 2709 |
"logging_steps": 1,
|
|
@@ -2723,7 +5412,7 @@
|
|
| 2723 |
"attributes": {}
|
| 2724 |
}
|
| 2725 |
},
|
| 2726 |
-
"total_flos":
|
| 2727 |
"train_batch_size": 2,
|
| 2728 |
"trial_name": null,
|
| 2729 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.5004082965866405,
|
| 5 |
"eval_steps": 383,
|
| 6 |
+
"global_step": 766,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 2704 |
"eval_samples_per_second": 16.076,
|
| 2705 |
"eval_steps_per_second": 4.022,
|
| 2706 |
"step": 383
|
| 2707 |
+
},
|
| 2708 |
+
{
|
| 2709 |
+
"epoch": 0.2508574228319451,
|
| 2710 |
+
"grad_norm": Infinity,
|
| 2711 |
+
"learning_rate": 0.00017590256715962506,
|
| 2712 |
+
"loss": 1.4462,
|
| 2713 |
+
"step": 384
|
| 2714 |
+
},
|
| 2715 |
+
{
|
| 2716 |
+
"epoch": 0.25151069737056997,
|
| 2717 |
+
"grad_norm": Infinity,
|
| 2718 |
+
"learning_rate": 0.0001757641957541203,
|
| 2719 |
+
"loss": 1.5079,
|
| 2720 |
+
"step": 385
|
| 2721 |
+
},
|
| 2722 |
+
{
|
| 2723 |
+
"epoch": 0.2521639719091948,
|
| 2724 |
+
"grad_norm": Infinity,
|
| 2725 |
+
"learning_rate": 0.00017562548296696875,
|
| 2726 |
+
"loss": 1.536,
|
| 2727 |
+
"step": 386
|
| 2728 |
+
},
|
| 2729 |
+
{
|
| 2730 |
+
"epoch": 0.2528172464478197,
|
| 2731 |
+
"grad_norm": Infinity,
|
| 2732 |
+
"learning_rate": 0.0001754864294231886,
|
| 2733 |
+
"loss": 1.5538,
|
| 2734 |
+
"step": 387
|
| 2735 |
+
},
|
| 2736 |
+
{
|
| 2737 |
+
"epoch": 0.25347052098644457,
|
| 2738 |
+
"grad_norm": Infinity,
|
| 2739 |
+
"learning_rate": 0.00017534703574933347,
|
| 2740 |
+
"loss": 1.8808,
|
| 2741 |
+
"step": 388
|
| 2742 |
+
},
|
| 2743 |
+
{
|
| 2744 |
+
"epoch": 0.2541237955250694,
|
| 2745 |
+
"grad_norm": Infinity,
|
| 2746 |
+
"learning_rate": 0.00017520730257348946,
|
| 2747 |
+
"loss": 1.9979,
|
| 2748 |
+
"step": 389
|
| 2749 |
+
},
|
| 2750 |
+
{
|
| 2751 |
+
"epoch": 0.25477707006369427,
|
| 2752 |
+
"grad_norm": Infinity,
|
| 2753 |
+
"learning_rate": 0.00017506723052527242,
|
| 2754 |
+
"loss": 1.9746,
|
| 2755 |
+
"step": 390
|
| 2756 |
+
},
|
| 2757 |
+
{
|
| 2758 |
+
"epoch": 0.2554303446023191,
|
| 2759 |
+
"grad_norm": Infinity,
|
| 2760 |
+
"learning_rate": 0.00017492682023582528,
|
| 2761 |
+
"loss": 2.23,
|
| 2762 |
+
"step": 391
|
| 2763 |
+
},
|
| 2764 |
+
{
|
| 2765 |
+
"epoch": 0.25608361914094396,
|
| 2766 |
+
"grad_norm": Infinity,
|
| 2767 |
+
"learning_rate": 0.0001747860723378148,
|
| 2768 |
+
"loss": 2.2459,
|
| 2769 |
+
"step": 392
|
| 2770 |
+
},
|
| 2771 |
+
{
|
| 2772 |
+
"epoch": 0.2567368936795688,
|
| 2773 |
+
"grad_norm": Infinity,
|
| 2774 |
+
"learning_rate": 0.0001746449874654291,
|
| 2775 |
+
"loss": 2.1848,
|
| 2776 |
+
"step": 393
|
| 2777 |
+
},
|
| 2778 |
+
{
|
| 2779 |
+
"epoch": 0.2573901682181937,
|
| 2780 |
+
"grad_norm": Infinity,
|
| 2781 |
+
"learning_rate": 0.0001745035662543745,
|
| 2782 |
+
"loss": 2.5955,
|
| 2783 |
+
"step": 394
|
| 2784 |
+
},
|
| 2785 |
+
{
|
| 2786 |
+
"epoch": 0.25804344275681856,
|
| 2787 |
+
"grad_norm": Infinity,
|
| 2788 |
+
"learning_rate": 0.00017436180934187308,
|
| 2789 |
+
"loss": 2.6265,
|
| 2790 |
+
"step": 395
|
| 2791 |
+
},
|
| 2792 |
+
{
|
| 2793 |
+
"epoch": 0.2586967172954434,
|
| 2794 |
+
"grad_norm": Infinity,
|
| 2795 |
+
"learning_rate": 0.00017421971736665925,
|
| 2796 |
+
"loss": 2.9216,
|
| 2797 |
+
"step": 396
|
| 2798 |
+
},
|
| 2799 |
+
{
|
| 2800 |
+
"epoch": 0.25934999183406826,
|
| 2801 |
+
"grad_norm": Infinity,
|
| 2802 |
+
"learning_rate": 0.00017407729096897737,
|
| 2803 |
+
"loss": 2.6603,
|
| 2804 |
+
"step": 397
|
| 2805 |
+
},
|
| 2806 |
+
{
|
| 2807 |
+
"epoch": 0.2600032663726931,
|
| 2808 |
+
"grad_norm": Infinity,
|
| 2809 |
+
"learning_rate": 0.00017393453079057847,
|
| 2810 |
+
"loss": 3.1459,
|
| 2811 |
+
"step": 398
|
| 2812 |
+
},
|
| 2813 |
+
{
|
| 2814 |
+
"epoch": 0.26065654091131796,
|
| 2815 |
+
"grad_norm": Infinity,
|
| 2816 |
+
"learning_rate": 0.00017379143747471768,
|
| 2817 |
+
"loss": 3.2171,
|
| 2818 |
+
"step": 399
|
| 2819 |
+
},
|
| 2820 |
+
{
|
| 2821 |
+
"epoch": 0.26130981544994286,
|
| 2822 |
+
"grad_norm": Infinity,
|
| 2823 |
+
"learning_rate": 0.00017364801166615124,
|
| 2824 |
+
"loss": 4.4529,
|
| 2825 |
+
"step": 400
|
| 2826 |
+
},
|
| 2827 |
+
{
|
| 2828 |
+
"epoch": 0.2619630899885677,
|
| 2829 |
+
"grad_norm": Infinity,
|
| 2830 |
+
"learning_rate": 0.00017350425401113336,
|
| 2831 |
+
"loss": 1.3081,
|
| 2832 |
+
"step": 401
|
| 2833 |
+
},
|
| 2834 |
+
{
|
| 2835 |
+
"epoch": 0.26261636452719256,
|
| 2836 |
+
"grad_norm": Infinity,
|
| 2837 |
+
"learning_rate": 0.00017336016515741366,
|
| 2838 |
+
"loss": 1.4575,
|
| 2839 |
+
"step": 402
|
| 2840 |
+
},
|
| 2841 |
+
{
|
| 2842 |
+
"epoch": 0.2632696390658174,
|
| 2843 |
+
"grad_norm": Infinity,
|
| 2844 |
+
"learning_rate": 0.00017321574575423406,
|
| 2845 |
+
"loss": 1.4,
|
| 2846 |
+
"step": 403
|
| 2847 |
+
},
|
| 2848 |
+
{
|
| 2849 |
+
"epoch": 0.26392291360444226,
|
| 2850 |
+
"grad_norm": Infinity,
|
| 2851 |
+
"learning_rate": 0.00017307099645232578,
|
| 2852 |
+
"loss": 1.4742,
|
| 2853 |
+
"step": 404
|
| 2854 |
+
},
|
| 2855 |
+
{
|
| 2856 |
+
"epoch": 0.2645761881430671,
|
| 2857 |
+
"grad_norm": Infinity,
|
| 2858 |
+
"learning_rate": 0.00017292591790390665,
|
| 2859 |
+
"loss": 1.4765,
|
| 2860 |
+
"step": 405
|
| 2861 |
+
},
|
| 2862 |
+
{
|
| 2863 |
+
"epoch": 0.265229462681692,
|
| 2864 |
+
"grad_norm": Infinity,
|
| 2865 |
+
"learning_rate": 0.00017278051076267796,
|
| 2866 |
+
"loss": 1.5007,
|
| 2867 |
+
"step": 406
|
| 2868 |
+
},
|
| 2869 |
+
{
|
| 2870 |
+
"epoch": 0.26588273722031686,
|
| 2871 |
+
"grad_norm": Infinity,
|
| 2872 |
+
"learning_rate": 0.00017263477568382166,
|
| 2873 |
+
"loss": 1.5344,
|
| 2874 |
+
"step": 407
|
| 2875 |
+
},
|
| 2876 |
+
{
|
| 2877 |
+
"epoch": 0.2665360117589417,
|
| 2878 |
+
"grad_norm": Infinity,
|
| 2879 |
+
"learning_rate": 0.0001724887133239972,
|
| 2880 |
+
"loss": 1.5091,
|
| 2881 |
+
"step": 408
|
| 2882 |
+
},
|
| 2883 |
+
{
|
| 2884 |
+
"epoch": 0.26718928629756655,
|
| 2885 |
+
"grad_norm": Infinity,
|
| 2886 |
+
"learning_rate": 0.00017234232434133883,
|
| 2887 |
+
"loss": 1.6508,
|
| 2888 |
+
"step": 409
|
| 2889 |
+
},
|
| 2890 |
+
{
|
| 2891 |
+
"epoch": 0.2678425608361914,
|
| 2892 |
+
"grad_norm": Infinity,
|
| 2893 |
+
"learning_rate": 0.00017219560939545246,
|
| 2894 |
+
"loss": 1.7905,
|
| 2895 |
+
"step": 410
|
| 2896 |
+
},
|
| 2897 |
+
{
|
| 2898 |
+
"epoch": 0.26849583537481625,
|
| 2899 |
+
"grad_norm": Infinity,
|
| 2900 |
+
"learning_rate": 0.00017204856914741274,
|
| 2901 |
+
"loss": 1.686,
|
| 2902 |
+
"step": 411
|
| 2903 |
+
},
|
| 2904 |
+
{
|
| 2905 |
+
"epoch": 0.2691491099134411,
|
| 2906 |
+
"grad_norm": Infinity,
|
| 2907 |
+
"learning_rate": 0.0001719012042597601,
|
| 2908 |
+
"loss": 1.9091,
|
| 2909 |
+
"step": 412
|
| 2910 |
+
},
|
| 2911 |
+
{
|
| 2912 |
+
"epoch": 0.269802384452066,
|
| 2913 |
+
"grad_norm": Infinity,
|
| 2914 |
+
"learning_rate": 0.00017175351539649774,
|
| 2915 |
+
"loss": 1.9509,
|
| 2916 |
+
"step": 413
|
| 2917 |
+
},
|
| 2918 |
+
{
|
| 2919 |
+
"epoch": 0.27045565899069085,
|
| 2920 |
+
"grad_norm": Infinity,
|
| 2921 |
+
"learning_rate": 0.00017160550322308863,
|
| 2922 |
+
"loss": 1.9816,
|
| 2923 |
+
"step": 414
|
| 2924 |
+
},
|
| 2925 |
+
{
|
| 2926 |
+
"epoch": 0.2711089335293157,
|
| 2927 |
+
"grad_norm": Infinity,
|
| 2928 |
+
"learning_rate": 0.00017145716840645254,
|
| 2929 |
+
"loss": 2.3551,
|
| 2930 |
+
"step": 415
|
| 2931 |
+
},
|
| 2932 |
+
{
|
| 2933 |
+
"epoch": 0.27176220806794055,
|
| 2934 |
+
"grad_norm": Infinity,
|
| 2935 |
+
"learning_rate": 0.000171308511614963,
|
| 2936 |
+
"loss": 2.2993,
|
| 2937 |
+
"step": 416
|
| 2938 |
+
},
|
| 2939 |
+
{
|
| 2940 |
+
"epoch": 0.2724154826065654,
|
| 2941 |
+
"grad_norm": Infinity,
|
| 2942 |
+
"learning_rate": 0.00017115953351844434,
|
| 2943 |
+
"loss": 2.4097,
|
| 2944 |
+
"step": 417
|
| 2945 |
+
},
|
| 2946 |
+
{
|
| 2947 |
+
"epoch": 0.27306875714519024,
|
| 2948 |
+
"grad_norm": Infinity,
|
| 2949 |
+
"learning_rate": 0.00017101023478816857,
|
| 2950 |
+
"loss": 2.6394,
|
| 2951 |
+
"step": 418
|
| 2952 |
+
},
|
| 2953 |
+
{
|
| 2954 |
+
"epoch": 0.27372203168381515,
|
| 2955 |
+
"grad_norm": Infinity,
|
| 2956 |
+
"learning_rate": 0.00017086061609685257,
|
| 2957 |
+
"loss": 2.7541,
|
| 2958 |
+
"step": 419
|
| 2959 |
+
},
|
| 2960 |
+
{
|
| 2961 |
+
"epoch": 0.27437530622244,
|
| 2962 |
+
"grad_norm": Infinity,
|
| 2963 |
+
"learning_rate": 0.00017071067811865476,
|
| 2964 |
+
"loss": 2.6986,
|
| 2965 |
+
"step": 420
|
| 2966 |
+
},
|
| 2967 |
+
{
|
| 2968 |
+
"epoch": 0.27502858076106484,
|
| 2969 |
+
"grad_norm": Infinity,
|
| 2970 |
+
"learning_rate": 0.0001705604215291723,
|
| 2971 |
+
"loss": 2.8947,
|
| 2972 |
+
"step": 421
|
| 2973 |
+
},
|
| 2974 |
+
{
|
| 2975 |
+
"epoch": 0.2756818552996897,
|
| 2976 |
+
"grad_norm": Infinity,
|
| 2977 |
+
"learning_rate": 0.00017040984700543793,
|
| 2978 |
+
"loss": 2.5362,
|
| 2979 |
+
"step": 422
|
| 2980 |
+
},
|
| 2981 |
+
{
|
| 2982 |
+
"epoch": 0.27633512983831454,
|
| 2983 |
+
"grad_norm": Infinity,
|
| 2984 |
+
"learning_rate": 0.00017025895522591693,
|
| 2985 |
+
"loss": 2.8326,
|
| 2986 |
+
"step": 423
|
| 2987 |
+
},
|
| 2988 |
+
{
|
| 2989 |
+
"epoch": 0.2769884043769394,
|
| 2990 |
+
"grad_norm": Infinity,
|
| 2991 |
+
"learning_rate": 0.00017010774687050418,
|
| 2992 |
+
"loss": 3.813,
|
| 2993 |
+
"step": 424
|
| 2994 |
+
},
|
| 2995 |
+
{
|
| 2996 |
+
"epoch": 0.27764167891556424,
|
| 2997 |
+
"grad_norm": Infinity,
|
| 2998 |
+
"learning_rate": 0.00016995622262052092,
|
| 2999 |
+
"loss": 3.9088,
|
| 3000 |
+
"step": 425
|
| 3001 |
+
},
|
| 3002 |
+
{
|
| 3003 |
+
"epoch": 0.27829495345418914,
|
| 3004 |
+
"grad_norm": Infinity,
|
| 3005 |
+
"learning_rate": 0.00016980438315871178,
|
| 3006 |
+
"loss": 1.4111,
|
| 3007 |
+
"step": 426
|
| 3008 |
+
},
|
| 3009 |
+
{
|
| 3010 |
+
"epoch": 0.278948227992814,
|
| 3011 |
+
"grad_norm": Infinity,
|
| 3012 |
+
"learning_rate": 0.00016965222916924167,
|
| 3013 |
+
"loss": 1.3758,
|
| 3014 |
+
"step": 427
|
| 3015 |
+
},
|
| 3016 |
+
{
|
| 3017 |
+
"epoch": 0.27960150253143884,
|
| 3018 |
+
"grad_norm": Infinity,
|
| 3019 |
+
"learning_rate": 0.0001694997613376928,
|
| 3020 |
+
"loss": 1.4541,
|
| 3021 |
+
"step": 428
|
| 3022 |
+
},
|
| 3023 |
+
{
|
| 3024 |
+
"epoch": 0.2802547770700637,
|
| 3025 |
+
"grad_norm": Infinity,
|
| 3026 |
+
"learning_rate": 0.00016934698035106133,
|
| 3027 |
+
"loss": 1.3758,
|
| 3028 |
+
"step": 429
|
| 3029 |
+
},
|
| 3030 |
+
{
|
| 3031 |
+
"epoch": 0.28090805160868854,
|
| 3032 |
+
"grad_norm": Infinity,
|
| 3033 |
+
"learning_rate": 0.00016919388689775464,
|
| 3034 |
+
"loss": 1.491,
|
| 3035 |
+
"step": 430
|
| 3036 |
+
},
|
| 3037 |
+
{
|
| 3038 |
+
"epoch": 0.2815613261473134,
|
| 3039 |
+
"grad_norm": Infinity,
|
| 3040 |
+
"learning_rate": 0.0001690404816675879,
|
| 3041 |
+
"loss": 1.5145,
|
| 3042 |
+
"step": 431
|
| 3043 |
+
},
|
| 3044 |
+
{
|
| 3045 |
+
"epoch": 0.2822146006859383,
|
| 3046 |
+
"grad_norm": Infinity,
|
| 3047 |
+
"learning_rate": 0.00016888676535178115,
|
| 3048 |
+
"loss": 1.5918,
|
| 3049 |
+
"step": 432
|
| 3050 |
+
},
|
| 3051 |
+
{
|
| 3052 |
+
"epoch": 0.28286787522456314,
|
| 3053 |
+
"grad_norm": Infinity,
|
| 3054 |
+
"learning_rate": 0.00016873273864295612,
|
| 3055 |
+
"loss": 1.5333,
|
| 3056 |
+
"step": 433
|
| 3057 |
+
},
|
| 3058 |
+
{
|
| 3059 |
+
"epoch": 0.283521149763188,
|
| 3060 |
+
"grad_norm": Infinity,
|
| 3061 |
+
"learning_rate": 0.00016857840223513315,
|
| 3062 |
+
"loss": 1.7168,
|
| 3063 |
+
"step": 434
|
| 3064 |
+
},
|
| 3065 |
+
{
|
| 3066 |
+
"epoch": 0.28417442430181283,
|
| 3067 |
+
"grad_norm": Infinity,
|
| 3068 |
+
"learning_rate": 0.00016842375682372805,
|
| 3069 |
+
"loss": 1.5902,
|
| 3070 |
+
"step": 435
|
| 3071 |
+
},
|
| 3072 |
+
{
|
| 3073 |
+
"epoch": 0.2848276988404377,
|
| 3074 |
+
"grad_norm": Infinity,
|
| 3075 |
+
"learning_rate": 0.00016826880310554887,
|
| 3076 |
+
"loss": 1.6932,
|
| 3077 |
+
"step": 436
|
| 3078 |
+
},
|
| 3079 |
+
{
|
| 3080 |
+
"epoch": 0.28548097337906253,
|
| 3081 |
+
"grad_norm": Infinity,
|
| 3082 |
+
"learning_rate": 0.00016811354177879287,
|
| 3083 |
+
"loss": 1.8251,
|
| 3084 |
+
"step": 437
|
| 3085 |
+
},
|
| 3086 |
+
{
|
| 3087 |
+
"epoch": 0.28613424791768743,
|
| 3088 |
+
"grad_norm": Infinity,
|
| 3089 |
+
"learning_rate": 0.00016795797354304345,
|
| 3090 |
+
"loss": 1.7183,
|
| 3091 |
+
"step": 438
|
| 3092 |
+
},
|
| 3093 |
+
{
|
| 3094 |
+
"epoch": 0.2867875224563123,
|
| 3095 |
+
"grad_norm": Infinity,
|
| 3096 |
+
"learning_rate": 0.00016780209909926676,
|
| 3097 |
+
"loss": 1.836,
|
| 3098 |
+
"step": 439
|
| 3099 |
+
},
|
| 3100 |
+
{
|
| 3101 |
+
"epoch": 0.28744079699493713,
|
| 3102 |
+
"grad_norm": Infinity,
|
| 3103 |
+
"learning_rate": 0.0001676459191498087,
|
| 3104 |
+
"loss": 2.0649,
|
| 3105 |
+
"step": 440
|
| 3106 |
+
},
|
| 3107 |
+
{
|
| 3108 |
+
"epoch": 0.288094071533562,
|
| 3109 |
+
"grad_norm": Infinity,
|
| 3110 |
+
"learning_rate": 0.0001674894343983918,
|
| 3111 |
+
"loss": 2.058,
|
| 3112 |
+
"step": 441
|
| 3113 |
+
},
|
| 3114 |
+
{
|
| 3115 |
+
"epoch": 0.28874734607218683,
|
| 3116 |
+
"grad_norm": Infinity,
|
| 3117 |
+
"learning_rate": 0.00016733264555011195,
|
| 3118 |
+
"loss": 2.1594,
|
| 3119 |
+
"step": 442
|
| 3120 |
+
},
|
| 3121 |
+
{
|
| 3122 |
+
"epoch": 0.2894006206108117,
|
| 3123 |
+
"grad_norm": Infinity,
|
| 3124 |
+
"learning_rate": 0.0001671755533114352,
|
| 3125 |
+
"loss": 2.3795,
|
| 3126 |
+
"step": 443
|
| 3127 |
+
},
|
| 3128 |
+
{
|
| 3129 |
+
"epoch": 0.2900538951494365,
|
| 3130 |
+
"grad_norm": Infinity,
|
| 3131 |
+
"learning_rate": 0.00016701815839019468,
|
| 3132 |
+
"loss": 2.6133,
|
| 3133 |
+
"step": 444
|
| 3134 |
+
},
|
| 3135 |
+
{
|
| 3136 |
+
"epoch": 0.29070716968806143,
|
| 3137 |
+
"grad_norm": Infinity,
|
| 3138 |
+
"learning_rate": 0.00016686046149558736,
|
| 3139 |
+
"loss": 2.5835,
|
| 3140 |
+
"step": 445
|
| 3141 |
+
},
|
| 3142 |
+
{
|
| 3143 |
+
"epoch": 0.2913604442266863,
|
| 3144 |
+
"grad_norm": Infinity,
|
| 3145 |
+
"learning_rate": 0.00016670246333817088,
|
| 3146 |
+
"loss": 2.9254,
|
| 3147 |
+
"step": 446
|
| 3148 |
+
},
|
| 3149 |
+
{
|
| 3150 |
+
"epoch": 0.2920137187653111,
|
| 3151 |
+
"grad_norm": Infinity,
|
| 3152 |
+
"learning_rate": 0.00016654416462986022,
|
| 3153 |
+
"loss": 2.9367,
|
| 3154 |
+
"step": 447
|
| 3155 |
+
},
|
| 3156 |
+
{
|
| 3157 |
+
"epoch": 0.292666993303936,
|
| 3158 |
+
"grad_norm": Infinity,
|
| 3159 |
+
"learning_rate": 0.00016638556608392473,
|
| 3160 |
+
"loss": 3.2409,
|
| 3161 |
+
"step": 448
|
| 3162 |
+
},
|
| 3163 |
+
{
|
| 3164 |
+
"epoch": 0.2933202678425608,
|
| 3165 |
+
"grad_norm": Infinity,
|
| 3166 |
+
"learning_rate": 0.00016622666841498463,
|
| 3167 |
+
"loss": 3.7771,
|
| 3168 |
+
"step": 449
|
| 3169 |
+
},
|
| 3170 |
+
{
|
| 3171 |
+
"epoch": 0.29397354238118567,
|
| 3172 |
+
"grad_norm": Infinity,
|
| 3173 |
+
"learning_rate": 0.00016606747233900815,
|
| 3174 |
+
"loss": 4.364,
|
| 3175 |
+
"step": 450
|
| 3176 |
+
},
|
| 3177 |
+
{
|
| 3178 |
+
"epoch": 0.2946268169198106,
|
| 3179 |
+
"grad_norm": Infinity,
|
| 3180 |
+
"learning_rate": 0.00016590797857330789,
|
| 3181 |
+
"loss": 1.2995,
|
| 3182 |
+
"step": 451
|
| 3183 |
+
},
|
| 3184 |
+
{
|
| 3185 |
+
"epoch": 0.2952800914584354,
|
| 3186 |
+
"grad_norm": Infinity,
|
| 3187 |
+
"learning_rate": 0.0001657481878365379,
|
| 3188 |
+
"loss": 1.4345,
|
| 3189 |
+
"step": 452
|
| 3190 |
+
},
|
| 3191 |
+
{
|
| 3192 |
+
"epoch": 0.29593336599706027,
|
| 3193 |
+
"grad_norm": Infinity,
|
| 3194 |
+
"learning_rate": 0.0001655881008486903,
|
| 3195 |
+
"loss": 1.421,
|
| 3196 |
+
"step": 453
|
| 3197 |
+
},
|
| 3198 |
+
{
|
| 3199 |
+
"epoch": 0.2965866405356851,
|
| 3200 |
+
"grad_norm": Infinity,
|
| 3201 |
+
"learning_rate": 0.0001654277183310921,
|
| 3202 |
+
"loss": 1.4081,
|
| 3203 |
+
"step": 454
|
| 3204 |
+
},
|
| 3205 |
+
{
|
| 3206 |
+
"epoch": 0.29723991507430997,
|
| 3207 |
+
"grad_norm": Infinity,
|
| 3208 |
+
"learning_rate": 0.0001652670410064019,
|
| 3209 |
+
"loss": 1.4488,
|
| 3210 |
+
"step": 455
|
| 3211 |
+
},
|
| 3212 |
+
{
|
| 3213 |
+
"epoch": 0.2978931896129348,
|
| 3214 |
+
"grad_norm": Infinity,
|
| 3215 |
+
"learning_rate": 0.00016510606959860665,
|
| 3216 |
+
"loss": 1.4637,
|
| 3217 |
+
"step": 456
|
| 3218 |
+
},
|
| 3219 |
+
{
|
| 3220 |
+
"epoch": 0.2985464641515597,
|
| 3221 |
+
"grad_norm": Infinity,
|
| 3222 |
+
"learning_rate": 0.00016494480483301836,
|
| 3223 |
+
"loss": 1.5709,
|
| 3224 |
+
"step": 457
|
| 3225 |
+
},
|
| 3226 |
+
{
|
| 3227 |
+
"epoch": 0.29919973869018457,
|
| 3228 |
+
"grad_norm": Infinity,
|
| 3229 |
+
"learning_rate": 0.00016478324743627101,
|
| 3230 |
+
"loss": 1.4784,
|
| 3231 |
+
"step": 458
|
| 3232 |
+
},
|
| 3233 |
+
{
|
| 3234 |
+
"epoch": 0.2998530132288094,
|
| 3235 |
+
"grad_norm": Infinity,
|
| 3236 |
+
"learning_rate": 0.00016462139813631693,
|
| 3237 |
+
"loss": 1.5429,
|
| 3238 |
+
"step": 459
|
| 3239 |
+
},
|
| 3240 |
+
{
|
| 3241 |
+
"epoch": 0.30050628776743427,
|
| 3242 |
+
"grad_norm": Infinity,
|
| 3243 |
+
"learning_rate": 0.00016445925766242391,
|
| 3244 |
+
"loss": 1.6066,
|
| 3245 |
+
"step": 460
|
| 3246 |
+
},
|
| 3247 |
+
{
|
| 3248 |
+
"epoch": 0.3011595623060591,
|
| 3249 |
+
"grad_norm": Infinity,
|
| 3250 |
+
"learning_rate": 0.00016429682674517156,
|
| 3251 |
+
"loss": 1.8577,
|
| 3252 |
+
"step": 461
|
| 3253 |
+
},
|
| 3254 |
+
{
|
| 3255 |
+
"epoch": 0.30181283684468396,
|
| 3256 |
+
"grad_norm": Infinity,
|
| 3257 |
+
"learning_rate": 0.00016413410611644825,
|
| 3258 |
+
"loss": 1.7408,
|
| 3259 |
+
"step": 462
|
| 3260 |
+
},
|
| 3261 |
+
{
|
| 3262 |
+
"epoch": 0.3024661113833088,
|
| 3263 |
+
"grad_norm": Infinity,
|
| 3264 |
+
"learning_rate": 0.0001639710965094478,
|
| 3265 |
+
"loss": 1.8884,
|
| 3266 |
+
"step": 463
|
| 3267 |
+
},
|
| 3268 |
+
{
|
| 3269 |
+
"epoch": 0.3031193859219337,
|
| 3270 |
+
"grad_norm": Infinity,
|
| 3271 |
+
"learning_rate": 0.00016380779865866603,
|
| 3272 |
+
"loss": 2.052,
|
| 3273 |
+
"step": 464
|
| 3274 |
+
},
|
| 3275 |
+
{
|
| 3276 |
+
"epoch": 0.30377266046055856,
|
| 3277 |
+
"grad_norm": Infinity,
|
| 3278 |
+
"learning_rate": 0.00016364421329989755,
|
| 3279 |
+
"loss": 1.9189,
|
| 3280 |
+
"step": 465
|
| 3281 |
+
},
|
| 3282 |
+
{
|
| 3283 |
+
"epoch": 0.3044259349991834,
|
| 3284 |
+
"grad_norm": Infinity,
|
| 3285 |
+
"learning_rate": 0.00016348034117023258,
|
| 3286 |
+
"loss": 2.2974,
|
| 3287 |
+
"step": 466
|
| 3288 |
+
},
|
| 3289 |
+
{
|
| 3290 |
+
"epoch": 0.30507920953780826,
|
| 3291 |
+
"grad_norm": Infinity,
|
| 3292 |
+
"learning_rate": 0.00016331618300805326,
|
| 3293 |
+
"loss": 2.187,
|
| 3294 |
+
"step": 467
|
| 3295 |
+
},
|
| 3296 |
+
{
|
| 3297 |
+
"epoch": 0.3057324840764331,
|
| 3298 |
+
"grad_norm": Infinity,
|
| 3299 |
+
"learning_rate": 0.00016315173955303068,
|
| 3300 |
+
"loss": 2.4096,
|
| 3301 |
+
"step": 468
|
| 3302 |
+
},
|
| 3303 |
+
{
|
| 3304 |
+
"epoch": 0.30638575861505796,
|
| 3305 |
+
"grad_norm": Infinity,
|
| 3306 |
+
"learning_rate": 0.00016298701154612147,
|
| 3307 |
+
"loss": 2.3334,
|
| 3308 |
+
"step": 469
|
| 3309 |
+
},
|
| 3310 |
+
{
|
| 3311 |
+
"epoch": 0.30703903315368286,
|
| 3312 |
+
"grad_norm": Infinity,
|
| 3313 |
+
"learning_rate": 0.00016282199972956425,
|
| 3314 |
+
"loss": 2.7411,
|
| 3315 |
+
"step": 470
|
| 3316 |
+
},
|
| 3317 |
+
{
|
| 3318 |
+
"epoch": 0.3076923076923077,
|
| 3319 |
+
"grad_norm": Infinity,
|
| 3320 |
+
"learning_rate": 0.00016265670484687654,
|
| 3321 |
+
"loss": 2.7242,
|
| 3322 |
+
"step": 471
|
| 3323 |
+
},
|
| 3324 |
+
{
|
| 3325 |
+
"epoch": 0.30834558223093256,
|
| 3326 |
+
"grad_norm": Infinity,
|
| 3327 |
+
"learning_rate": 0.0001624911276428513,
|
| 3328 |
+
"loss": 3.1537,
|
| 3329 |
+
"step": 472
|
| 3330 |
+
},
|
| 3331 |
+
{
|
| 3332 |
+
"epoch": 0.3089988567695574,
|
| 3333 |
+
"grad_norm": Infinity,
|
| 3334 |
+
"learning_rate": 0.0001623252688635536,
|
| 3335 |
+
"loss": 3.5625,
|
| 3336 |
+
"step": 473
|
| 3337 |
+
},
|
| 3338 |
+
{
|
| 3339 |
+
"epoch": 0.30965213130818225,
|
| 3340 |
+
"grad_norm": Infinity,
|
| 3341 |
+
"learning_rate": 0.00016215912925631723,
|
| 3342 |
+
"loss": 3.6483,
|
| 3343 |
+
"step": 474
|
| 3344 |
+
},
|
| 3345 |
+
{
|
| 3346 |
+
"epoch": 0.3103054058468071,
|
| 3347 |
+
"grad_norm": Infinity,
|
| 3348 |
+
"learning_rate": 0.00016199270956974128,
|
| 3349 |
+
"loss": 4.2782,
|
| 3350 |
+
"step": 475
|
| 3351 |
+
},
|
| 3352 |
+
{
|
| 3353 |
+
"epoch": 0.31095868038543195,
|
| 3354 |
+
"grad_norm": Infinity,
|
| 3355 |
+
"learning_rate": 0.00016182601055368697,
|
| 3356 |
+
"loss": 1.3157,
|
| 3357 |
+
"step": 476
|
| 3358 |
+
},
|
| 3359 |
+
{
|
| 3360 |
+
"epoch": 0.31161195492405686,
|
| 3361 |
+
"grad_norm": Infinity,
|
| 3362 |
+
"learning_rate": 0.00016165903295927401,
|
| 3363 |
+
"loss": 1.4153,
|
| 3364 |
+
"step": 477
|
| 3365 |
+
},
|
| 3366 |
+
{
|
| 3367 |
+
"epoch": 0.3122652294626817,
|
| 3368 |
+
"grad_norm": Infinity,
|
| 3369 |
+
"learning_rate": 0.00016149177753887746,
|
| 3370 |
+
"loss": 1.4591,
|
| 3371 |
+
"step": 478
|
| 3372 |
+
},
|
| 3373 |
+
{
|
| 3374 |
+
"epoch": 0.31291850400130655,
|
| 3375 |
+
"grad_norm": Infinity,
|
| 3376 |
+
"learning_rate": 0.00016132424504612406,
|
| 3377 |
+
"loss": 1.4796,
|
| 3378 |
+
"step": 479
|
| 3379 |
+
},
|
| 3380 |
+
{
|
| 3381 |
+
"epoch": 0.3135717785399314,
|
| 3382 |
+
"grad_norm": Infinity,
|
| 3383 |
+
"learning_rate": 0.00016115643623588915,
|
| 3384 |
+
"loss": 1.5419,
|
| 3385 |
+
"step": 480
|
| 3386 |
+
},
|
| 3387 |
+
{
|
| 3388 |
+
"epoch": 0.31422505307855625,
|
| 3389 |
+
"grad_norm": Infinity,
|
| 3390 |
+
"learning_rate": 0.00016098835186429303,
|
| 3391 |
+
"loss": 1.4612,
|
| 3392 |
+
"step": 481
|
| 3393 |
+
},
|
| 3394 |
+
{
|
| 3395 |
+
"epoch": 0.3148783276171811,
|
| 3396 |
+
"grad_norm": Infinity,
|
| 3397 |
+
"learning_rate": 0.00016081999268869766,
|
| 3398 |
+
"loss": 1.6256,
|
| 3399 |
+
"step": 482
|
| 3400 |
+
},
|
| 3401 |
+
{
|
| 3402 |
+
"epoch": 0.315531602155806,
|
| 3403 |
+
"grad_norm": Infinity,
|
| 3404 |
+
"learning_rate": 0.00016065135946770323,
|
| 3405 |
+
"loss": 1.6535,
|
| 3406 |
+
"step": 483
|
| 3407 |
+
},
|
| 3408 |
+
{
|
| 3409 |
+
"epoch": 0.31618487669443085,
|
| 3410 |
+
"grad_norm": Infinity,
|
| 3411 |
+
"learning_rate": 0.00016048245296114468,
|
| 3412 |
+
"loss": 1.6035,
|
| 3413 |
+
"step": 484
|
| 3414 |
+
},
|
| 3415 |
+
{
|
| 3416 |
+
"epoch": 0.3168381512330557,
|
| 3417 |
+
"grad_norm": Infinity,
|
| 3418 |
+
"learning_rate": 0.00016031327393008845,
|
| 3419 |
+
"loss": 1.575,
|
| 3420 |
+
"step": 485
|
| 3421 |
+
},
|
| 3422 |
+
{
|
| 3423 |
+
"epoch": 0.31749142577168055,
|
| 3424 |
+
"grad_norm": Infinity,
|
| 3425 |
+
"learning_rate": 0.00016014382313682881,
|
| 3426 |
+
"loss": 1.792,
|
| 3427 |
+
"step": 486
|
| 3428 |
+
},
|
| 3429 |
+
{
|
| 3430 |
+
"epoch": 0.3181447003103054,
|
| 3431 |
+
"grad_norm": Infinity,
|
| 3432 |
+
"learning_rate": 0.00015997410134488464,
|
| 3433 |
+
"loss": 1.7314,
|
| 3434 |
+
"step": 487
|
| 3435 |
+
},
|
| 3436 |
+
{
|
| 3437 |
+
"epoch": 0.31879797484893024,
|
| 3438 |
+
"grad_norm": Infinity,
|
| 3439 |
+
"learning_rate": 0.00015980410931899582,
|
| 3440 |
+
"loss": 1.8793,
|
| 3441 |
+
"step": 488
|
| 3442 |
+
},
|
| 3443 |
+
{
|
| 3444 |
+
"epoch": 0.31945124938755515,
|
| 3445 |
+
"grad_norm": Infinity,
|
| 3446 |
+
"learning_rate": 0.00015963384782511993,
|
| 3447 |
+
"loss": 1.9586,
|
| 3448 |
+
"step": 489
|
| 3449 |
+
},
|
| 3450 |
+
{
|
| 3451 |
+
"epoch": 0.32010452392618,
|
| 3452 |
+
"grad_norm": Infinity,
|
| 3453 |
+
"learning_rate": 0.00015946331763042867,
|
| 3454 |
+
"loss": 1.9951,
|
| 3455 |
+
"step": 490
|
| 3456 |
+
},
|
| 3457 |
+
{
|
| 3458 |
+
"epoch": 0.32075779846480484,
|
| 3459 |
+
"grad_norm": Infinity,
|
| 3460 |
+
"learning_rate": 0.0001592925195033045,
|
| 3461 |
+
"loss": 2.1583,
|
| 3462 |
+
"step": 491
|
| 3463 |
+
},
|
| 3464 |
+
{
|
| 3465 |
+
"epoch": 0.3214110730034297,
|
| 3466 |
+
"grad_norm": Infinity,
|
| 3467 |
+
"learning_rate": 0.00015912145421333719,
|
| 3468 |
+
"loss": 2.3153,
|
| 3469 |
+
"step": 492
|
| 3470 |
+
},
|
| 3471 |
+
{
|
| 3472 |
+
"epoch": 0.32206434754205454,
|
| 3473 |
+
"grad_norm": Infinity,
|
| 3474 |
+
"learning_rate": 0.00015895012253132017,
|
| 3475 |
+
"loss": 2.3933,
|
| 3476 |
+
"step": 493
|
| 3477 |
+
},
|
| 3478 |
+
{
|
| 3479 |
+
"epoch": 0.3227176220806794,
|
| 3480 |
+
"grad_norm": Infinity,
|
| 3481 |
+
"learning_rate": 0.00015877852522924732,
|
| 3482 |
+
"loss": 2.3513,
|
| 3483 |
+
"step": 494
|
| 3484 |
+
},
|
| 3485 |
+
{
|
| 3486 |
+
"epoch": 0.32337089661930424,
|
| 3487 |
+
"grad_norm": Infinity,
|
| 3488 |
+
"learning_rate": 0.00015860666308030932,
|
| 3489 |
+
"loss": 2.6687,
|
| 3490 |
+
"step": 495
|
| 3491 |
+
},
|
| 3492 |
+
{
|
| 3493 |
+
"epoch": 0.32402417115792914,
|
| 3494 |
+
"grad_norm": Infinity,
|
| 3495 |
+
"learning_rate": 0.00015843453685889016,
|
| 3496 |
+
"loss": 2.8344,
|
| 3497 |
+
"step": 496
|
| 3498 |
+
},
|
| 3499 |
+
{
|
| 3500 |
+
"epoch": 0.324677445696554,
|
| 3501 |
+
"grad_norm": Infinity,
|
| 3502 |
+
"learning_rate": 0.00015826214734056374,
|
| 3503 |
+
"loss": 2.883,
|
| 3504 |
+
"step": 497
|
| 3505 |
+
},
|
| 3506 |
+
{
|
| 3507 |
+
"epoch": 0.32533072023517884,
|
| 3508 |
+
"grad_norm": Infinity,
|
| 3509 |
+
"learning_rate": 0.0001580894953020904,
|
| 3510 |
+
"loss": 3.1467,
|
| 3511 |
+
"step": 498
|
| 3512 |
+
},
|
| 3513 |
+
{
|
| 3514 |
+
"epoch": 0.3259839947738037,
|
| 3515 |
+
"grad_norm": Infinity,
|
| 3516 |
+
"learning_rate": 0.00015791658152141327,
|
| 3517 |
+
"loss": 3.3732,
|
| 3518 |
+
"step": 499
|
| 3519 |
+
},
|
| 3520 |
+
{
|
| 3521 |
+
"epoch": 0.32663726931242854,
|
| 3522 |
+
"grad_norm": Infinity,
|
| 3523 |
+
"learning_rate": 0.0001577434067776548,
|
| 3524 |
+
"loss": 4.2269,
|
| 3525 |
+
"step": 500
|
| 3526 |
+
},
|
| 3527 |
+
{
|
| 3528 |
+
"epoch": 0.3272905438510534,
|
| 3529 |
+
"grad_norm": Infinity,
|
| 3530 |
+
"learning_rate": 0.00015756997185111348,
|
| 3531 |
+
"loss": 1.2486,
|
| 3532 |
+
"step": 501
|
| 3533 |
+
},
|
| 3534 |
+
{
|
| 3535 |
+
"epoch": 0.3279438183896783,
|
| 3536 |
+
"grad_norm": Infinity,
|
| 3537 |
+
"learning_rate": 0.00015739627752325996,
|
| 3538 |
+
"loss": 1.3895,
|
| 3539 |
+
"step": 502
|
| 3540 |
+
},
|
| 3541 |
+
{
|
| 3542 |
+
"epoch": 0.32859709292830314,
|
| 3543 |
+
"grad_norm": Infinity,
|
| 3544 |
+
"learning_rate": 0.0001572223245767338,
|
| 3545 |
+
"loss": 1.422,
|
| 3546 |
+
"step": 503
|
| 3547 |
+
},
|
| 3548 |
+
{
|
| 3549 |
+
"epoch": 0.329250367466928,
|
| 3550 |
+
"grad_norm": Infinity,
|
| 3551 |
+
"learning_rate": 0.00015704811379533987,
|
| 3552 |
+
"loss": 1.4185,
|
| 3553 |
+
"step": 504
|
| 3554 |
+
},
|
| 3555 |
+
{
|
| 3556 |
+
"epoch": 0.32990364200555283,
|
| 3557 |
+
"grad_norm": Infinity,
|
| 3558 |
+
"learning_rate": 0.0001568736459640447,
|
| 3559 |
+
"loss": 1.4656,
|
| 3560 |
+
"step": 505
|
| 3561 |
+
},
|
| 3562 |
+
{
|
| 3563 |
+
"epoch": 0.3305569165441777,
|
| 3564 |
+
"grad_norm": Infinity,
|
| 3565 |
+
"learning_rate": 0.00015669892186897318,
|
| 3566 |
+
"loss": 1.4505,
|
| 3567 |
+
"step": 506
|
| 3568 |
+
},
|
| 3569 |
+
{
|
| 3570 |
+
"epoch": 0.33121019108280253,
|
| 3571 |
+
"grad_norm": Infinity,
|
| 3572 |
+
"learning_rate": 0.00015652394229740484,
|
| 3573 |
+
"loss": 1.6007,
|
| 3574 |
+
"step": 507
|
| 3575 |
+
},
|
| 3576 |
+
{
|
| 3577 |
+
"epoch": 0.3318634656214274,
|
| 3578 |
+
"grad_norm": Infinity,
|
| 3579 |
+
"learning_rate": 0.00015634870803777026,
|
| 3580 |
+
"loss": 1.6112,
|
| 3581 |
+
"step": 508
|
| 3582 |
+
},
|
| 3583 |
+
{
|
| 3584 |
+
"epoch": 0.3325167401600523,
|
| 3585 |
+
"grad_norm": Infinity,
|
| 3586 |
+
"learning_rate": 0.00015617321987964776,
|
| 3587 |
+
"loss": 1.7202,
|
| 3588 |
+
"step": 509
|
| 3589 |
+
},
|
| 3590 |
+
{
|
| 3591 |
+
"epoch": 0.33317001469867713,
|
| 3592 |
+
"grad_norm": Infinity,
|
| 3593 |
+
"learning_rate": 0.00015599747861375955,
|
| 3594 |
+
"loss": 1.6825,
|
| 3595 |
+
"step": 510
|
| 3596 |
+
},
|
| 3597 |
+
{
|
| 3598 |
+
"epoch": 0.333823289237302,
|
| 3599 |
+
"grad_norm": Infinity,
|
| 3600 |
+
"learning_rate": 0.0001558214850319684,
|
| 3601 |
+
"loss": 1.8399,
|
| 3602 |
+
"step": 511
|
| 3603 |
+
},
|
| 3604 |
+
{
|
| 3605 |
+
"epoch": 0.3344765637759268,
|
| 3606 |
+
"grad_norm": Infinity,
|
| 3607 |
+
"learning_rate": 0.00015564523992727387,
|
| 3608 |
+
"loss": 1.7425,
|
| 3609 |
+
"step": 512
|
| 3610 |
+
},
|
| 3611 |
+
{
|
| 3612 |
+
"epoch": 0.3351298383145517,
|
| 3613 |
+
"grad_norm": Infinity,
|
| 3614 |
+
"learning_rate": 0.00015546874409380897,
|
| 3615 |
+
"loss": 1.7817,
|
| 3616 |
+
"step": 513
|
| 3617 |
+
},
|
| 3618 |
+
{
|
| 3619 |
+
"epoch": 0.3357831128531765,
|
| 3620 |
+
"grad_norm": Infinity,
|
| 3621 |
+
"learning_rate": 0.00015529199832683635,
|
| 3622 |
+
"loss": 2.0647,
|
| 3623 |
+
"step": 514
|
| 3624 |
+
},
|
| 3625 |
+
{
|
| 3626 |
+
"epoch": 0.33643638739180143,
|
| 3627 |
+
"grad_norm": Infinity,
|
| 3628 |
+
"learning_rate": 0.0001551150034227449,
|
| 3629 |
+
"loss": 1.9344,
|
| 3630 |
+
"step": 515
|
| 3631 |
+
},
|
| 3632 |
+
{
|
| 3633 |
+
"epoch": 0.3370896619304263,
|
| 3634 |
+
"grad_norm": Infinity,
|
| 3635 |
+
"learning_rate": 0.00015493776017904602,
|
| 3636 |
+
"loss": 2.2739,
|
| 3637 |
+
"step": 516
|
| 3638 |
+
},
|
| 3639 |
+
{
|
| 3640 |
+
"epoch": 0.3377429364690511,
|
| 3641 |
+
"grad_norm": Infinity,
|
| 3642 |
+
"learning_rate": 0.00015476026939437014,
|
| 3643 |
+
"loss": 2.4646,
|
| 3644 |
+
"step": 517
|
| 3645 |
+
},
|
| 3646 |
+
{
|
| 3647 |
+
"epoch": 0.338396211007676,
|
| 3648 |
+
"grad_norm": Infinity,
|
| 3649 |
+
"learning_rate": 0.00015458253186846301,
|
| 3650 |
+
"loss": 2.5544,
|
| 3651 |
+
"step": 518
|
| 3652 |
+
},
|
| 3653 |
+
{
|
| 3654 |
+
"epoch": 0.3390494855463008,
|
| 3655 |
+
"grad_norm": Infinity,
|
| 3656 |
+
"learning_rate": 0.00015440454840218225,
|
| 3657 |
+
"loss": 2.4784,
|
| 3658 |
+
"step": 519
|
| 3659 |
+
},
|
| 3660 |
+
{
|
| 3661 |
+
"epoch": 0.33970276008492567,
|
| 3662 |
+
"grad_norm": Infinity,
|
| 3663 |
+
"learning_rate": 0.00015422631979749354,
|
| 3664 |
+
"loss": 2.8688,
|
| 3665 |
+
"step": 520
|
| 3666 |
+
},
|
| 3667 |
+
{
|
| 3668 |
+
"epoch": 0.3403560346235506,
|
| 3669 |
+
"grad_norm": Infinity,
|
| 3670 |
+
"learning_rate": 0.00015404784685746716,
|
| 3671 |
+
"loss": 2.8859,
|
| 3672 |
+
"step": 521
|
| 3673 |
+
},
|
| 3674 |
+
{
|
| 3675 |
+
"epoch": 0.3410093091621754,
|
| 3676 |
+
"grad_norm": Infinity,
|
| 3677 |
+
"learning_rate": 0.0001538691303862744,
|
| 3678 |
+
"loss": 3.362,
|
| 3679 |
+
"step": 522
|
| 3680 |
+
},
|
| 3681 |
+
{
|
| 3682 |
+
"epoch": 0.34166258370080027,
|
| 3683 |
+
"grad_norm": Infinity,
|
| 3684 |
+
"learning_rate": 0.00015369017118918373,
|
| 3685 |
+
"loss": 3.297,
|
| 3686 |
+
"step": 523
|
| 3687 |
+
},
|
| 3688 |
+
{
|
| 3689 |
+
"epoch": 0.3423158582394251,
|
| 3690 |
+
"grad_norm": Infinity,
|
| 3691 |
+
"learning_rate": 0.00015351097007255742,
|
| 3692 |
+
"loss": 3.6869,
|
| 3693 |
+
"step": 524
|
| 3694 |
+
},
|
| 3695 |
+
{
|
| 3696 |
+
"epoch": 0.34296913277804997,
|
| 3697 |
+
"grad_norm": Infinity,
|
| 3698 |
+
"learning_rate": 0.00015333152784384777,
|
| 3699 |
+
"loss": 4.2406,
|
| 3700 |
+
"step": 525
|
| 3701 |
+
},
|
| 3702 |
+
{
|
| 3703 |
+
"epoch": 0.3436224073166748,
|
| 3704 |
+
"grad_norm": Infinity,
|
| 3705 |
+
"learning_rate": 0.0001531518453115934,
|
| 3706 |
+
"loss": 1.3543,
|
| 3707 |
+
"step": 526
|
| 3708 |
+
},
|
| 3709 |
+
{
|
| 3710 |
+
"epoch": 0.34427568185529966,
|
| 3711 |
+
"grad_norm": Infinity,
|
| 3712 |
+
"learning_rate": 0.00015297192328541582,
|
| 3713 |
+
"loss": 1.4295,
|
| 3714 |
+
"step": 527
|
| 3715 |
+
},
|
| 3716 |
+
{
|
| 3717 |
+
"epoch": 0.34492895639392457,
|
| 3718 |
+
"grad_norm": Infinity,
|
| 3719 |
+
"learning_rate": 0.00015279176257601557,
|
| 3720 |
+
"loss": 1.4386,
|
| 3721 |
+
"step": 528
|
| 3722 |
+
},
|
| 3723 |
+
{
|
| 3724 |
+
"epoch": 0.3455822309325494,
|
| 3725 |
+
"grad_norm": Infinity,
|
| 3726 |
+
"learning_rate": 0.00015261136399516873,
|
| 3727 |
+
"loss": 1.4315,
|
| 3728 |
+
"step": 529
|
| 3729 |
+
},
|
| 3730 |
+
{
|
| 3731 |
+
"epoch": 0.34623550547117427,
|
| 3732 |
+
"grad_norm": Infinity,
|
| 3733 |
+
"learning_rate": 0.00015243072835572318,
|
| 3734 |
+
"loss": 1.4927,
|
| 3735 |
+
"step": 530
|
| 3736 |
+
},
|
| 3737 |
+
{
|
| 3738 |
+
"epoch": 0.3468887800097991,
|
| 3739 |
+
"grad_norm": Infinity,
|
| 3740 |
+
"learning_rate": 0.0001522498564715949,
|
| 3741 |
+
"loss": 1.4866,
|
| 3742 |
+
"step": 531
|
| 3743 |
+
},
|
| 3744 |
+
{
|
| 3745 |
+
"epoch": 0.34754205454842396,
|
| 3746 |
+
"grad_norm": Infinity,
|
| 3747 |
+
"learning_rate": 0.0001520687491577644,
|
| 3748 |
+
"loss": 1.5698,
|
| 3749 |
+
"step": 532
|
| 3750 |
+
},
|
| 3751 |
+
{
|
| 3752 |
+
"epoch": 0.3481953290870488,
|
| 3753 |
+
"grad_norm": Infinity,
|
| 3754 |
+
"learning_rate": 0.00015188740723027296,
|
| 3755 |
+
"loss": 1.5634,
|
| 3756 |
+
"step": 533
|
| 3757 |
+
},
|
| 3758 |
+
{
|
| 3759 |
+
"epoch": 0.3488486036256737,
|
| 3760 |
+
"grad_norm": Infinity,
|
| 3761 |
+
"learning_rate": 0.00015170583150621905,
|
| 3762 |
+
"loss": 1.6559,
|
| 3763 |
+
"step": 534
|
| 3764 |
+
},
|
| 3765 |
+
{
|
| 3766 |
+
"epoch": 0.34950187816429856,
|
| 3767 |
+
"grad_norm": Infinity,
|
| 3768 |
+
"learning_rate": 0.00015152402280375454,
|
| 3769 |
+
"loss": 1.726,
|
| 3770 |
+
"step": 535
|
| 3771 |
+
},
|
| 3772 |
+
{
|
| 3773 |
+
"epoch": 0.3501551527029234,
|
| 3774 |
+
"grad_norm": Infinity,
|
| 3775 |
+
"learning_rate": 0.00015134198194208107,
|
| 3776 |
+
"loss": 1.752,
|
| 3777 |
+
"step": 536
|
| 3778 |
+
},
|
| 3779 |
+
{
|
| 3780 |
+
"epoch": 0.35080842724154826,
|
| 3781 |
+
"grad_norm": Infinity,
|
| 3782 |
+
"learning_rate": 0.00015115970974144635,
|
| 3783 |
+
"loss": 1.7894,
|
| 3784 |
+
"step": 537
|
| 3785 |
+
},
|
| 3786 |
+
{
|
| 3787 |
+
"epoch": 0.3514617017801731,
|
| 3788 |
+
"grad_norm": Infinity,
|
| 3789 |
+
"learning_rate": 0.00015097720702314055,
|
| 3790 |
+
"loss": 1.8187,
|
| 3791 |
+
"step": 538
|
| 3792 |
+
},
|
| 3793 |
+
{
|
| 3794 |
+
"epoch": 0.35211497631879796,
|
| 3795 |
+
"grad_norm": Infinity,
|
| 3796 |
+
"learning_rate": 0.00015079447460949238,
|
| 3797 |
+
"loss": 1.8663,
|
| 3798 |
+
"step": 539
|
| 3799 |
+
},
|
| 3800 |
+
{
|
| 3801 |
+
"epoch": 0.3527682508574228,
|
| 3802 |
+
"grad_norm": Infinity,
|
| 3803 |
+
"learning_rate": 0.00015061151332386566,
|
| 3804 |
+
"loss": 1.8478,
|
| 3805 |
+
"step": 540
|
| 3806 |
+
},
|
| 3807 |
+
{
|
| 3808 |
+
"epoch": 0.3534215253960477,
|
| 3809 |
+
"grad_norm": Infinity,
|
| 3810 |
+
"learning_rate": 0.0001504283239906553,
|
| 3811 |
+
"loss": 1.8881,
|
| 3812 |
+
"step": 541
|
| 3813 |
+
},
|
| 3814 |
+
{
|
| 3815 |
+
"epoch": 0.35407479993467256,
|
| 3816 |
+
"grad_norm": Infinity,
|
| 3817 |
+
"learning_rate": 0.00015024490743528393,
|
| 3818 |
+
"loss": 2.1571,
|
| 3819 |
+
"step": 542
|
| 3820 |
+
},
|
| 3821 |
+
{
|
| 3822 |
+
"epoch": 0.3547280744732974,
|
| 3823 |
+
"grad_norm": Infinity,
|
| 3824 |
+
"learning_rate": 0.00015006126448419791,
|
| 3825 |
+
"loss": 2.4004,
|
| 3826 |
+
"step": 543
|
| 3827 |
+
},
|
| 3828 |
+
{
|
| 3829 |
+
"epoch": 0.35538134901192225,
|
| 3830 |
+
"grad_norm": Infinity,
|
| 3831 |
+
"learning_rate": 0.00014987739596486374,
|
| 3832 |
+
"loss": 2.4008,
|
| 3833 |
+
"step": 544
|
| 3834 |
+
},
|
| 3835 |
+
{
|
| 3836 |
+
"epoch": 0.3560346235505471,
|
| 3837 |
+
"grad_norm": Infinity,
|
| 3838 |
+
"learning_rate": 0.00014969330270576427,
|
| 3839 |
+
"loss": 2.7406,
|
| 3840 |
+
"step": 545
|
| 3841 |
+
},
|
| 3842 |
+
{
|
| 3843 |
+
"epoch": 0.35668789808917195,
|
| 3844 |
+
"grad_norm": Infinity,
|
| 3845 |
+
"learning_rate": 0.00014950898553639505,
|
| 3846 |
+
"loss": 2.6133,
|
| 3847 |
+
"step": 546
|
| 3848 |
+
},
|
| 3849 |
+
{
|
| 3850 |
+
"epoch": 0.35734117262779685,
|
| 3851 |
+
"grad_norm": Infinity,
|
| 3852 |
+
"learning_rate": 0.00014932444528726043,
|
| 3853 |
+
"loss": 2.9329,
|
| 3854 |
+
"step": 547
|
| 3855 |
+
},
|
| 3856 |
+
{
|
| 3857 |
+
"epoch": 0.3579944471664217,
|
| 3858 |
+
"grad_norm": Infinity,
|
| 3859 |
+
"learning_rate": 0.00014913968278987004,
|
| 3860 |
+
"loss": 3.1703,
|
| 3861 |
+
"step": 548
|
| 3862 |
+
},
|
| 3863 |
+
{
|
| 3864 |
+
"epoch": 0.35864772170504655,
|
| 3865 |
+
"grad_norm": Infinity,
|
| 3866 |
+
"learning_rate": 0.00014895469887673483,
|
| 3867 |
+
"loss": 2.9231,
|
| 3868 |
+
"step": 549
|
| 3869 |
+
},
|
| 3870 |
+
{
|
| 3871 |
+
"epoch": 0.3593009962436714,
|
| 3872 |
+
"grad_norm": Infinity,
|
| 3873 |
+
"learning_rate": 0.00014876949438136347,
|
| 3874 |
+
"loss": 4.0923,
|
| 3875 |
+
"step": 550
|
| 3876 |
+
},
|
| 3877 |
+
{
|
| 3878 |
+
"epoch": 0.35995427078229625,
|
| 3879 |
+
"grad_norm": Infinity,
|
| 3880 |
+
"learning_rate": 0.00014858407013825854,
|
| 3881 |
+
"loss": 1.3462,
|
| 3882 |
+
"step": 551
|
| 3883 |
+
},
|
| 3884 |
+
{
|
| 3885 |
+
"epoch": 0.3606075453209211,
|
| 3886 |
+
"grad_norm": Infinity,
|
| 3887 |
+
"learning_rate": 0.00014839842698291267,
|
| 3888 |
+
"loss": 1.4769,
|
| 3889 |
+
"step": 552
|
| 3890 |
+
},
|
| 3891 |
+
{
|
| 3892 |
+
"epoch": 0.361260819859546,
|
| 3893 |
+
"grad_norm": Infinity,
|
| 3894 |
+
"learning_rate": 0.00014821256575180507,
|
| 3895 |
+
"loss": 1.3811,
|
| 3896 |
+
"step": 553
|
| 3897 |
+
},
|
| 3898 |
+
{
|
| 3899 |
+
"epoch": 0.36191409439817085,
|
| 3900 |
+
"grad_norm": Infinity,
|
| 3901 |
+
"learning_rate": 0.00014802648728239742,
|
| 3902 |
+
"loss": 1.3628,
|
| 3903 |
+
"step": 554
|
| 3904 |
+
},
|
| 3905 |
+
{
|
| 3906 |
+
"epoch": 0.3625673689367957,
|
| 3907 |
+
"grad_norm": Infinity,
|
| 3908 |
+
"learning_rate": 0.00014784019241313026,
|
| 3909 |
+
"loss": 1.4687,
|
| 3910 |
+
"step": 555
|
| 3911 |
+
},
|
| 3912 |
+
{
|
| 3913 |
+
"epoch": 0.36322064347542055,
|
| 3914 |
+
"grad_norm": Infinity,
|
| 3915 |
+
"learning_rate": 0.00014765368198341918,
|
| 3916 |
+
"loss": 1.4745,
|
| 3917 |
+
"step": 556
|
| 3918 |
+
},
|
| 3919 |
+
{
|
| 3920 |
+
"epoch": 0.3638739180140454,
|
| 3921 |
+
"grad_norm": Infinity,
|
| 3922 |
+
"learning_rate": 0.00014746695683365112,
|
| 3923 |
+
"loss": 1.4466,
|
| 3924 |
+
"step": 557
|
| 3925 |
+
},
|
| 3926 |
+
{
|
| 3927 |
+
"epoch": 0.36452719255267024,
|
| 3928 |
+
"grad_norm": Infinity,
|
| 3929 |
+
"learning_rate": 0.0001472800178051805,
|
| 3930 |
+
"loss": 1.6015,
|
| 3931 |
+
"step": 558
|
| 3932 |
+
},
|
| 3933 |
+
{
|
| 3934 |
+
"epoch": 0.3651804670912951,
|
| 3935 |
+
"grad_norm": Infinity,
|
| 3936 |
+
"learning_rate": 0.00014709286574032536,
|
| 3937 |
+
"loss": 1.5199,
|
| 3938 |
+
"step": 559
|
| 3939 |
+
},
|
| 3940 |
+
{
|
| 3941 |
+
"epoch": 0.36583374162992,
|
| 3942 |
+
"grad_norm": Infinity,
|
| 3943 |
+
"learning_rate": 0.0001469055014823637,
|
| 3944 |
+
"loss": 1.5306,
|
| 3945 |
+
"step": 560
|
| 3946 |
+
},
|
| 3947 |
+
{
|
| 3948 |
+
"epoch": 0.36648701616854484,
|
| 3949 |
+
"grad_norm": Infinity,
|
| 3950 |
+
"learning_rate": 0.0001467179258755297,
|
| 3951 |
+
"loss": 1.745,
|
| 3952 |
+
"step": 561
|
| 3953 |
+
},
|
| 3954 |
+
{
|
| 3955 |
+
"epoch": 0.3671402907071697,
|
| 3956 |
+
"grad_norm": Infinity,
|
| 3957 |
+
"learning_rate": 0.00014653013976500975,
|
| 3958 |
+
"loss": 1.8516,
|
| 3959 |
+
"step": 562
|
| 3960 |
+
},
|
| 3961 |
+
{
|
| 3962 |
+
"epoch": 0.36779356524579454,
|
| 3963 |
+
"grad_norm": Infinity,
|
| 3964 |
+
"learning_rate": 0.0001463421439969388,
|
| 3965 |
+
"loss": 1.8714,
|
| 3966 |
+
"step": 563
|
| 3967 |
+
},
|
| 3968 |
+
{
|
| 3969 |
+
"epoch": 0.3684468397844194,
|
| 3970 |
+
"grad_norm": Infinity,
|
| 3971 |
+
"learning_rate": 0.00014615393941839637,
|
| 3972 |
+
"loss": 1.8113,
|
| 3973 |
+
"step": 564
|
| 3974 |
+
},
|
| 3975 |
+
{
|
| 3976 |
+
"epoch": 0.36910011432304424,
|
| 3977 |
+
"grad_norm": Infinity,
|
| 3978 |
+
"learning_rate": 0.00014596552687740302,
|
| 3979 |
+
"loss": 2.0245,
|
| 3980 |
+
"step": 565
|
| 3981 |
+
},
|
| 3982 |
+
{
|
| 3983 |
+
"epoch": 0.36975338886166914,
|
| 3984 |
+
"grad_norm": Infinity,
|
| 3985 |
+
"learning_rate": 0.00014577690722291622,
|
| 3986 |
+
"loss": 1.963,
|
| 3987 |
+
"step": 566
|
| 3988 |
+
},
|
| 3989 |
+
{
|
| 3990 |
+
"epoch": 0.370406663400294,
|
| 3991 |
+
"grad_norm": Infinity,
|
| 3992 |
+
"learning_rate": 0.00014558808130482674,
|
| 3993 |
+
"loss": 2.3101,
|
| 3994 |
+
"step": 567
|
| 3995 |
+
},
|
| 3996 |
+
{
|
| 3997 |
+
"epoch": 0.37105993793891884,
|
| 3998 |
+
"grad_norm": Infinity,
|
| 3999 |
+
"learning_rate": 0.00014539904997395468,
|
| 4000 |
+
"loss": 2.2081,
|
| 4001 |
+
"step": 568
|
| 4002 |
+
},
|
| 4003 |
+
{
|
| 4004 |
+
"epoch": 0.3717132124775437,
|
| 4005 |
+
"grad_norm": Infinity,
|
| 4006 |
+
"learning_rate": 0.00014520981408204574,
|
| 4007 |
+
"loss": 2.5591,
|
| 4008 |
+
"step": 569
|
| 4009 |
+
},
|
| 4010 |
+
{
|
| 4011 |
+
"epoch": 0.37236648701616853,
|
| 4012 |
+
"grad_norm": Infinity,
|
| 4013 |
+
"learning_rate": 0.00014502037448176734,
|
| 4014 |
+
"loss": 2.301,
|
| 4015 |
+
"step": 570
|
| 4016 |
+
},
|
| 4017 |
+
{
|
| 4018 |
+
"epoch": 0.3730197615547934,
|
| 4019 |
+
"grad_norm": Infinity,
|
| 4020 |
+
"learning_rate": 0.00014483073202670475,
|
| 4021 |
+
"loss": 2.8042,
|
| 4022 |
+
"step": 571
|
| 4023 |
+
},
|
| 4024 |
+
{
|
| 4025 |
+
"epoch": 0.37367303609341823,
|
| 4026 |
+
"grad_norm": Infinity,
|
| 4027 |
+
"learning_rate": 0.00014464088757135728,
|
| 4028 |
+
"loss": 2.9381,
|
| 4029 |
+
"step": 572
|
| 4030 |
+
},
|
| 4031 |
+
{
|
| 4032 |
+
"epoch": 0.37432631063204314,
|
| 4033 |
+
"grad_norm": Infinity,
|
| 4034 |
+
"learning_rate": 0.00014445084197113443,
|
| 4035 |
+
"loss": 2.9872,
|
| 4036 |
+
"step": 573
|
| 4037 |
+
},
|
| 4038 |
+
{
|
| 4039 |
+
"epoch": 0.374979585170668,
|
| 4040 |
+
"grad_norm": Infinity,
|
| 4041 |
+
"learning_rate": 0.00014426059608235208,
|
| 4042 |
+
"loss": 3.4324,
|
| 4043 |
+
"step": 574
|
| 4044 |
+
},
|
| 4045 |
+
{
|
| 4046 |
+
"epoch": 0.37563285970929283,
|
| 4047 |
+
"grad_norm": Infinity,
|
| 4048 |
+
"learning_rate": 0.00014407015076222846,
|
| 4049 |
+
"loss": 4.1841,
|
| 4050 |
+
"step": 575
|
| 4051 |
+
},
|
| 4052 |
+
{
|
| 4053 |
+
"epoch": 0.3762861342479177,
|
| 4054 |
+
"grad_norm": Infinity,
|
| 4055 |
+
"learning_rate": 0.00014387950686888047,
|
| 4056 |
+
"loss": 1.1931,
|
| 4057 |
+
"step": 576
|
| 4058 |
+
},
|
| 4059 |
+
{
|
| 4060 |
+
"epoch": 0.37693940878654253,
|
| 4061 |
+
"grad_norm": Infinity,
|
| 4062 |
+
"learning_rate": 0.0001436886652613198,
|
| 4063 |
+
"loss": 1.3665,
|
| 4064 |
+
"step": 577
|
| 4065 |
+
},
|
| 4066 |
+
{
|
| 4067 |
+
"epoch": 0.3775926833251674,
|
| 4068 |
+
"grad_norm": Infinity,
|
| 4069 |
+
"learning_rate": 0.00014349762679944896,
|
| 4070 |
+
"loss": 1.402,
|
| 4071 |
+
"step": 578
|
| 4072 |
+
},
|
| 4073 |
+
{
|
| 4074 |
+
"epoch": 0.3782459578637923,
|
| 4075 |
+
"grad_norm": Infinity,
|
| 4076 |
+
"learning_rate": 0.00014330639234405742,
|
| 4077 |
+
"loss": 1.4069,
|
| 4078 |
+
"step": 579
|
| 4079 |
+
},
|
| 4080 |
+
{
|
| 4081 |
+
"epoch": 0.37889923240241713,
|
| 4082 |
+
"grad_norm": Infinity,
|
| 4083 |
+
"learning_rate": 0.00014311496275681783,
|
| 4084 |
+
"loss": 1.4134,
|
| 4085 |
+
"step": 580
|
| 4086 |
+
},
|
| 4087 |
+
{
|
| 4088 |
+
"epoch": 0.379552506941042,
|
| 4089 |
+
"grad_norm": Infinity,
|
| 4090 |
+
"learning_rate": 0.00014292333890028204,
|
| 4091 |
+
"loss": 1.5248,
|
| 4092 |
+
"step": 581
|
| 4093 |
+
},
|
| 4094 |
+
{
|
| 4095 |
+
"epoch": 0.3802057814796668,
|
| 4096 |
+
"grad_norm": Infinity,
|
| 4097 |
+
"learning_rate": 0.00014273152163787726,
|
| 4098 |
+
"loss": 1.5282,
|
| 4099 |
+
"step": 582
|
| 4100 |
+
},
|
| 4101 |
+
{
|
| 4102 |
+
"epoch": 0.3808590560182917,
|
| 4103 |
+
"grad_norm": Infinity,
|
| 4104 |
+
"learning_rate": 0.00014253951183390215,
|
| 4105 |
+
"loss": 1.6725,
|
| 4106 |
+
"step": 583
|
| 4107 |
+
},
|
| 4108 |
+
{
|
| 4109 |
+
"epoch": 0.3815123305569165,
|
| 4110 |
+
"grad_norm": Infinity,
|
| 4111 |
+
"learning_rate": 0.0001423473103535229,
|
| 4112 |
+
"loss": 1.5775,
|
| 4113 |
+
"step": 584
|
| 4114 |
+
},
|
| 4115 |
+
{
|
| 4116 |
+
"epoch": 0.3821656050955414,
|
| 4117 |
+
"grad_norm": Infinity,
|
| 4118 |
+
"learning_rate": 0.00014215491806276944,
|
| 4119 |
+
"loss": 1.7086,
|
| 4120 |
+
"step": 585
|
| 4121 |
+
},
|
| 4122 |
+
{
|
| 4123 |
+
"epoch": 0.3828188796341663,
|
| 4124 |
+
"grad_norm": Infinity,
|
| 4125 |
+
"learning_rate": 0.0001419623358285314,
|
| 4126 |
+
"loss": 1.7671,
|
| 4127 |
+
"step": 586
|
| 4128 |
+
},
|
| 4129 |
+
{
|
| 4130 |
+
"epoch": 0.3834721541727911,
|
| 4131 |
+
"grad_norm": Infinity,
|
| 4132 |
+
"learning_rate": 0.0001417695645185543,
|
| 4133 |
+
"loss": 1.7763,
|
| 4134 |
+
"step": 587
|
| 4135 |
+
},
|
| 4136 |
+
{
|
| 4137 |
+
"epoch": 0.384125428711416,
|
| 4138 |
+
"grad_norm": Infinity,
|
| 4139 |
+
"learning_rate": 0.00014157660500143553,
|
| 4140 |
+
"loss": 1.9369,
|
| 4141 |
+
"step": 588
|
| 4142 |
+
},
|
| 4143 |
+
{
|
| 4144 |
+
"epoch": 0.3847787032500408,
|
| 4145 |
+
"grad_norm": Infinity,
|
| 4146 |
+
"learning_rate": 0.00014138345814662068,
|
| 4147 |
+
"loss": 1.8933,
|
| 4148 |
+
"step": 589
|
| 4149 |
+
},
|
| 4150 |
+
{
|
| 4151 |
+
"epoch": 0.38543197778866567,
|
| 4152 |
+
"grad_norm": Infinity,
|
| 4153 |
+
"learning_rate": 0.0001411901248243993,
|
| 4154 |
+
"loss": 1.9815,
|
| 4155 |
+
"step": 590
|
| 4156 |
+
},
|
| 4157 |
+
{
|
| 4158 |
+
"epoch": 0.3860852523272905,
|
| 4159 |
+
"grad_norm": Infinity,
|
| 4160 |
+
"learning_rate": 0.00014099660590590114,
|
| 4161 |
+
"loss": 2.2974,
|
| 4162 |
+
"step": 591
|
| 4163 |
+
},
|
| 4164 |
+
{
|
| 4165 |
+
"epoch": 0.3867385268659154,
|
| 4166 |
+
"grad_norm": Infinity,
|
| 4167 |
+
"learning_rate": 0.00014080290226309224,
|
| 4168 |
+
"loss": 2.2827,
|
| 4169 |
+
"step": 592
|
| 4170 |
+
},
|
| 4171 |
+
{
|
| 4172 |
+
"epoch": 0.38739180140454027,
|
| 4173 |
+
"grad_norm": Infinity,
|
| 4174 |
+
"learning_rate": 0.00014060901476877107,
|
| 4175 |
+
"loss": 2.8547,
|
| 4176 |
+
"step": 593
|
| 4177 |
+
},
|
| 4178 |
+
{
|
| 4179 |
+
"epoch": 0.3880450759431651,
|
| 4180 |
+
"grad_norm": Infinity,
|
| 4181 |
+
"learning_rate": 0.00014041494429656442,
|
| 4182 |
+
"loss": 2.4635,
|
| 4183 |
+
"step": 594
|
| 4184 |
+
},
|
| 4185 |
+
{
|
| 4186 |
+
"epoch": 0.38869835048178997,
|
| 4187 |
+
"grad_norm": Infinity,
|
| 4188 |
+
"learning_rate": 0.00014022069172092352,
|
| 4189 |
+
"loss": 2.9595,
|
| 4190 |
+
"step": 595
|
| 4191 |
+
},
|
| 4192 |
+
{
|
| 4193 |
+
"epoch": 0.3893516250204148,
|
| 4194 |
+
"grad_norm": Infinity,
|
| 4195 |
+
"learning_rate": 0.00014002625791712021,
|
| 4196 |
+
"loss": 2.7913,
|
| 4197 |
+
"step": 596
|
| 4198 |
+
},
|
| 4199 |
+
{
|
| 4200 |
+
"epoch": 0.39000489955903966,
|
| 4201 |
+
"grad_norm": Infinity,
|
| 4202 |
+
"learning_rate": 0.00013983164376124286,
|
| 4203 |
+
"loss": 2.8025,
|
| 4204 |
+
"step": 597
|
| 4205 |
+
},
|
| 4206 |
+
{
|
| 4207 |
+
"epoch": 0.39065817409766457,
|
| 4208 |
+
"grad_norm": Infinity,
|
| 4209 |
+
"learning_rate": 0.0001396368501301925,
|
| 4210 |
+
"loss": 3.4075,
|
| 4211 |
+
"step": 598
|
| 4212 |
+
},
|
| 4213 |
+
{
|
| 4214 |
+
"epoch": 0.3913114486362894,
|
| 4215 |
+
"grad_norm": Infinity,
|
| 4216 |
+
"learning_rate": 0.0001394418779016789,
|
| 4217 |
+
"loss": 3.6667,
|
| 4218 |
+
"step": 599
|
| 4219 |
+
},
|
| 4220 |
+
{
|
| 4221 |
+
"epoch": 0.39196472317491426,
|
| 4222 |
+
"grad_norm": Infinity,
|
| 4223 |
+
"learning_rate": 0.00013924672795421637,
|
| 4224 |
+
"loss": 4.7289,
|
| 4225 |
+
"step": 600
|
| 4226 |
+
},
|
| 4227 |
+
{
|
| 4228 |
+
"epoch": 0.3926179977135391,
|
| 4229 |
+
"grad_norm": Infinity,
|
| 4230 |
+
"learning_rate": 0.00013905140116712026,
|
| 4231 |
+
"loss": 1.3319,
|
| 4232 |
+
"step": 601
|
| 4233 |
+
},
|
| 4234 |
+
{
|
| 4235 |
+
"epoch": 0.39327127225216396,
|
| 4236 |
+
"grad_norm": Infinity,
|
| 4237 |
+
"learning_rate": 0.00013885589842050253,
|
| 4238 |
+
"loss": 1.4609,
|
| 4239 |
+
"step": 602
|
| 4240 |
+
},
|
| 4241 |
+
{
|
| 4242 |
+
"epoch": 0.3939245467907888,
|
| 4243 |
+
"grad_norm": Infinity,
|
| 4244 |
+
"learning_rate": 0.0001386602205952681,
|
| 4245 |
+
"loss": 1.4258,
|
| 4246 |
+
"step": 603
|
| 4247 |
+
},
|
| 4248 |
+
{
|
| 4249 |
+
"epoch": 0.3945778213294137,
|
| 4250 |
+
"grad_norm": Infinity,
|
| 4251 |
+
"learning_rate": 0.00013846436857311068,
|
| 4252 |
+
"loss": 1.4458,
|
| 4253 |
+
"step": 604
|
| 4254 |
+
},
|
| 4255 |
+
{
|
| 4256 |
+
"epoch": 0.39523109586803856,
|
| 4257 |
+
"grad_norm": Infinity,
|
| 4258 |
+
"learning_rate": 0.000138268343236509,
|
| 4259 |
+
"loss": 1.2812,
|
| 4260 |
+
"step": 605
|
| 4261 |
+
},
|
| 4262 |
+
{
|
| 4263 |
+
"epoch": 0.3958843704066634,
|
| 4264 |
+
"grad_norm": Infinity,
|
| 4265 |
+
"learning_rate": 0.00013807214546872256,
|
| 4266 |
+
"loss": 1.5763,
|
| 4267 |
+
"step": 606
|
| 4268 |
+
},
|
| 4269 |
+
{
|
| 4270 |
+
"epoch": 0.39653764494528826,
|
| 4271 |
+
"grad_norm": Infinity,
|
| 4272 |
+
"learning_rate": 0.00013787577615378792,
|
| 4273 |
+
"loss": 1.5288,
|
| 4274 |
+
"step": 607
|
| 4275 |
+
},
|
| 4276 |
+
{
|
| 4277 |
+
"epoch": 0.3971909194839131,
|
| 4278 |
+
"grad_norm": Infinity,
|
| 4279 |
+
"learning_rate": 0.00013767923617651463,
|
| 4280 |
+
"loss": 1.5984,
|
| 4281 |
+
"step": 608
|
| 4282 |
+
},
|
| 4283 |
+
{
|
| 4284 |
+
"epoch": 0.39784419402253796,
|
| 4285 |
+
"grad_norm": Infinity,
|
| 4286 |
+
"learning_rate": 0.00013748252642248115,
|
| 4287 |
+
"loss": 1.5448,
|
| 4288 |
+
"step": 609
|
| 4289 |
+
},
|
| 4290 |
+
{
|
| 4291 |
+
"epoch": 0.3984974685611628,
|
| 4292 |
+
"grad_norm": Infinity,
|
| 4293 |
+
"learning_rate": 0.00013728564777803088,
|
| 4294 |
+
"loss": 1.745,
|
| 4295 |
+
"step": 610
|
| 4296 |
+
},
|
| 4297 |
+
{
|
| 4298 |
+
"epoch": 0.3991507430997877,
|
| 4299 |
+
"grad_norm": Infinity,
|
| 4300 |
+
"learning_rate": 0.00013708860113026834,
|
| 4301 |
+
"loss": 1.7803,
|
| 4302 |
+
"step": 611
|
| 4303 |
+
},
|
| 4304 |
+
{
|
| 4305 |
+
"epoch": 0.39980401763841256,
|
| 4306 |
+
"grad_norm": Infinity,
|
| 4307 |
+
"learning_rate": 0.00013689138736705495,
|
| 4308 |
+
"loss": 1.7888,
|
| 4309 |
+
"step": 612
|
| 4310 |
+
},
|
| 4311 |
+
{
|
| 4312 |
+
"epoch": 0.4004572921770374,
|
| 4313 |
+
"grad_norm": Infinity,
|
| 4314 |
+
"learning_rate": 0.0001366940073770052,
|
| 4315 |
+
"loss": 1.8004,
|
| 4316 |
+
"step": 613
|
| 4317 |
+
},
|
| 4318 |
+
{
|
| 4319 |
+
"epoch": 0.40111056671566225,
|
| 4320 |
+
"grad_norm": Infinity,
|
| 4321 |
+
"learning_rate": 0.00013649646204948255,
|
| 4322 |
+
"loss": 1.7229,
|
| 4323 |
+
"step": 614
|
| 4324 |
+
},
|
| 4325 |
+
{
|
| 4326 |
+
"epoch": 0.4017638412542871,
|
| 4327 |
+
"grad_norm": Infinity,
|
| 4328 |
+
"learning_rate": 0.00013629875227459532,
|
| 4329 |
+
"loss": 2.0557,
|
| 4330 |
+
"step": 615
|
| 4331 |
+
},
|
| 4332 |
+
{
|
| 4333 |
+
"epoch": 0.40241711579291195,
|
| 4334 |
+
"grad_norm": Infinity,
|
| 4335 |
+
"learning_rate": 0.00013610087894319302,
|
| 4336 |
+
"loss": 1.9848,
|
| 4337 |
+
"step": 616
|
| 4338 |
+
},
|
| 4339 |
+
{
|
| 4340 |
+
"epoch": 0.40307039033153685,
|
| 4341 |
+
"grad_norm": Infinity,
|
| 4342 |
+
"learning_rate": 0.00013590284294686203,
|
| 4343 |
+
"loss": 2.3338,
|
| 4344 |
+
"step": 617
|
| 4345 |
+
},
|
| 4346 |
+
{
|
| 4347 |
+
"epoch": 0.4037236648701617,
|
| 4348 |
+
"grad_norm": Infinity,
|
| 4349 |
+
"learning_rate": 0.00013570464517792153,
|
| 4350 |
+
"loss": 2.2821,
|
| 4351 |
+
"step": 618
|
| 4352 |
+
},
|
| 4353 |
+
{
|
| 4354 |
+
"epoch": 0.40437693940878655,
|
| 4355 |
+
"grad_norm": Infinity,
|
| 4356 |
+
"learning_rate": 0.00013550628652941985,
|
| 4357 |
+
"loss": 2.5272,
|
| 4358 |
+
"step": 619
|
| 4359 |
+
},
|
| 4360 |
+
{
|
| 4361 |
+
"epoch": 0.4050302139474114,
|
| 4362 |
+
"grad_norm": Infinity,
|
| 4363 |
+
"learning_rate": 0.0001353077678951301,
|
| 4364 |
+
"loss": 2.847,
|
| 4365 |
+
"step": 620
|
| 4366 |
+
},
|
| 4367 |
+
{
|
| 4368 |
+
"epoch": 0.40568348848603625,
|
| 4369 |
+
"grad_norm": Infinity,
|
| 4370 |
+
"learning_rate": 0.00013510909016954624,
|
| 4371 |
+
"loss": 3.0231,
|
| 4372 |
+
"step": 621
|
| 4373 |
+
},
|
| 4374 |
+
{
|
| 4375 |
+
"epoch": 0.4063367630246611,
|
| 4376 |
+
"grad_norm": Infinity,
|
| 4377 |
+
"learning_rate": 0.00013491025424787915,
|
| 4378 |
+
"loss": 2.7372,
|
| 4379 |
+
"step": 622
|
| 4380 |
+
},
|
| 4381 |
+
{
|
| 4382 |
+
"epoch": 0.40699003756328594,
|
| 4383 |
+
"grad_norm": Infinity,
|
| 4384 |
+
"learning_rate": 0.00013471126102605245,
|
| 4385 |
+
"loss": 3.2824,
|
| 4386 |
+
"step": 623
|
| 4387 |
+
},
|
| 4388 |
+
{
|
| 4389 |
+
"epoch": 0.40764331210191085,
|
| 4390 |
+
"grad_norm": Infinity,
|
| 4391 |
+
"learning_rate": 0.00013451211140069858,
|
| 4392 |
+
"loss": 3.2836,
|
| 4393 |
+
"step": 624
|
| 4394 |
+
},
|
| 4395 |
+
{
|
| 4396 |
+
"epoch": 0.4082965866405357,
|
| 4397 |
+
"grad_norm": Infinity,
|
| 4398 |
+
"learning_rate": 0.00013431280626915467,
|
| 4399 |
+
"loss": 4.2905,
|
| 4400 |
+
"step": 625
|
| 4401 |
+
},
|
| 4402 |
+
{
|
| 4403 |
+
"epoch": 0.40894986117916055,
|
| 4404 |
+
"grad_norm": Infinity,
|
| 4405 |
+
"learning_rate": 0.0001341133465294585,
|
| 4406 |
+
"loss": 1.1817,
|
| 4407 |
+
"step": 626
|
| 4408 |
+
},
|
| 4409 |
+
{
|
| 4410 |
+
"epoch": 0.4096031357177854,
|
| 4411 |
+
"grad_norm": Infinity,
|
| 4412 |
+
"learning_rate": 0.0001339137330803446,
|
| 4413 |
+
"loss": 1.3825,
|
| 4414 |
+
"step": 627
|
| 4415 |
+
},
|
| 4416 |
+
{
|
| 4417 |
+
"epoch": 0.41025641025641024,
|
| 4418 |
+
"grad_norm": Infinity,
|
| 4419 |
+
"learning_rate": 0.00013371396682124005,
|
| 4420 |
+
"loss": 1.4202,
|
| 4421 |
+
"step": 628
|
| 4422 |
+
},
|
| 4423 |
+
{
|
| 4424 |
+
"epoch": 0.4109096847950351,
|
| 4425 |
+
"grad_norm": Infinity,
|
| 4426 |
+
"learning_rate": 0.0001335140486522604,
|
| 4427 |
+
"loss": 1.4288,
|
| 4428 |
+
"step": 629
|
| 4429 |
+
},
|
| 4430 |
+
{
|
| 4431 |
+
"epoch": 0.41156295933366,
|
| 4432 |
+
"grad_norm": Infinity,
|
| 4433 |
+
"learning_rate": 0.00013331397947420576,
|
| 4434 |
+
"loss": 1.4444,
|
| 4435 |
+
"step": 630
|
| 4436 |
+
},
|
| 4437 |
+
{
|
| 4438 |
+
"epoch": 0.41221623387228484,
|
| 4439 |
+
"grad_norm": Infinity,
|
| 4440 |
+
"learning_rate": 0.00013311376018855663,
|
| 4441 |
+
"loss": 1.5484,
|
| 4442 |
+
"step": 631
|
| 4443 |
+
},
|
| 4444 |
+
{
|
| 4445 |
+
"epoch": 0.4128695084109097,
|
| 4446 |
+
"grad_norm": Infinity,
|
| 4447 |
+
"learning_rate": 0.0001329133916974699,
|
| 4448 |
+
"loss": 1.5702,
|
| 4449 |
+
"step": 632
|
| 4450 |
+
},
|
| 4451 |
+
{
|
| 4452 |
+
"epoch": 0.41352278294953454,
|
| 4453 |
+
"grad_norm": Infinity,
|
| 4454 |
+
"learning_rate": 0.00013271287490377467,
|
| 4455 |
+
"loss": 1.6131,
|
| 4456 |
+
"step": 633
|
| 4457 |
+
},
|
| 4458 |
+
{
|
| 4459 |
+
"epoch": 0.4141760574881594,
|
| 4460 |
+
"grad_norm": Infinity,
|
| 4461 |
+
"learning_rate": 0.00013251221071096836,
|
| 4462 |
+
"loss": 1.6361,
|
| 4463 |
+
"step": 634
|
| 4464 |
+
},
|
| 4465 |
+
{
|
| 4466 |
+
"epoch": 0.41482933202678424,
|
| 4467 |
+
"grad_norm": Infinity,
|
| 4468 |
+
"learning_rate": 0.00013231140002321253,
|
| 4469 |
+
"loss": 1.6556,
|
| 4470 |
+
"step": 635
|
| 4471 |
+
},
|
| 4472 |
+
{
|
| 4473 |
+
"epoch": 0.41548260656540914,
|
| 4474 |
+
"grad_norm": Infinity,
|
| 4475 |
+
"learning_rate": 0.00013211044374532882,
|
| 4476 |
+
"loss": 1.785,
|
| 4477 |
+
"step": 636
|
| 4478 |
+
},
|
| 4479 |
+
{
|
| 4480 |
+
"epoch": 0.416135881104034,
|
| 4481 |
+
"grad_norm": Infinity,
|
| 4482 |
+
"learning_rate": 0.00013190934278279487,
|
| 4483 |
+
"loss": 1.7989,
|
| 4484 |
+
"step": 637
|
| 4485 |
+
},
|
| 4486 |
+
{
|
| 4487 |
+
"epoch": 0.41678915564265884,
|
| 4488 |
+
"grad_norm": Infinity,
|
| 4489 |
+
"learning_rate": 0.00013170809804174022,
|
| 4490 |
+
"loss": 1.8345,
|
| 4491 |
+
"step": 638
|
| 4492 |
+
},
|
| 4493 |
+
{
|
| 4494 |
+
"epoch": 0.4174424301812837,
|
| 4495 |
+
"grad_norm": Infinity,
|
| 4496 |
+
"learning_rate": 0.00013150671042894228,
|
| 4497 |
+
"loss": 1.8782,
|
| 4498 |
+
"step": 639
|
| 4499 |
+
},
|
| 4500 |
+
{
|
| 4501 |
+
"epoch": 0.41809570471990853,
|
| 4502 |
+
"grad_norm": Infinity,
|
| 4503 |
+
"learning_rate": 0.00013130518085182225,
|
| 4504 |
+
"loss": 1.9949,
|
| 4505 |
+
"step": 640
|
| 4506 |
+
},
|
| 4507 |
+
{
|
| 4508 |
+
"epoch": 0.4187489792585334,
|
| 4509 |
+
"grad_norm": Infinity,
|
| 4510 |
+
"learning_rate": 0.00013110351021844094,
|
| 4511 |
+
"loss": 2.1345,
|
| 4512 |
+
"step": 641
|
| 4513 |
+
},
|
| 4514 |
+
{
|
| 4515 |
+
"epoch": 0.41940225379715823,
|
| 4516 |
+
"grad_norm": Infinity,
|
| 4517 |
+
"learning_rate": 0.00013090169943749476,
|
| 4518 |
+
"loss": 2.5038,
|
| 4519 |
+
"step": 642
|
| 4520 |
+
},
|
| 4521 |
+
{
|
| 4522 |
+
"epoch": 0.42005552833578313,
|
| 4523 |
+
"grad_norm": Infinity,
|
| 4524 |
+
"learning_rate": 0.0001306997494183116,
|
| 4525 |
+
"loss": 2.7144,
|
| 4526 |
+
"step": 643
|
| 4527 |
+
},
|
| 4528 |
+
{
|
| 4529 |
+
"epoch": 0.420708802874408,
|
| 4530 |
+
"grad_norm": Infinity,
|
| 4531 |
+
"learning_rate": 0.00013049766107084678,
|
| 4532 |
+
"loss": 2.7682,
|
| 4533 |
+
"step": 644
|
| 4534 |
+
},
|
| 4535 |
+
{
|
| 4536 |
+
"epoch": 0.42136207741303283,
|
| 4537 |
+
"grad_norm": Infinity,
|
| 4538 |
+
"learning_rate": 0.00013029543530567884,
|
| 4539 |
+
"loss": 2.5713,
|
| 4540 |
+
"step": 645
|
| 4541 |
+
},
|
| 4542 |
+
{
|
| 4543 |
+
"epoch": 0.4220153519516577,
|
| 4544 |
+
"grad_norm": Infinity,
|
| 4545 |
+
"learning_rate": 0.00013009307303400556,
|
| 4546 |
+
"loss": 2.7619,
|
| 4547 |
+
"step": 646
|
| 4548 |
+
},
|
| 4549 |
+
{
|
| 4550 |
+
"epoch": 0.42266862649028253,
|
| 4551 |
+
"grad_norm": Infinity,
|
| 4552 |
+
"learning_rate": 0.0001298905751676397,
|
| 4553 |
+
"loss": 2.9728,
|
| 4554 |
+
"step": 647
|
| 4555 |
+
},
|
| 4556 |
+
{
|
| 4557 |
+
"epoch": 0.4233219010289074,
|
| 4558 |
+
"grad_norm": Infinity,
|
| 4559 |
+
"learning_rate": 0.0001296879426190051,
|
| 4560 |
+
"loss": 2.9242,
|
| 4561 |
+
"step": 648
|
| 4562 |
+
},
|
| 4563 |
+
{
|
| 4564 |
+
"epoch": 0.4239751755675323,
|
| 4565 |
+
"grad_norm": Infinity,
|
| 4566 |
+
"learning_rate": 0.00012948517630113245,
|
| 4567 |
+
"loss": 3.7497,
|
| 4568 |
+
"step": 649
|
| 4569 |
+
},
|
| 4570 |
+
{
|
| 4571 |
+
"epoch": 0.42462845010615713,
|
| 4572 |
+
"grad_norm": Infinity,
|
| 4573 |
+
"learning_rate": 0.00012928227712765504,
|
| 4574 |
+
"loss": 3.9534,
|
| 4575 |
+
"step": 650
|
| 4576 |
+
},
|
| 4577 |
+
{
|
| 4578 |
+
"epoch": 0.425281724644782,
|
| 4579 |
+
"grad_norm": Infinity,
|
| 4580 |
+
"learning_rate": 0.00012907924601280498,
|
| 4581 |
+
"loss": 1.3583,
|
| 4582 |
+
"step": 651
|
| 4583 |
+
},
|
| 4584 |
+
{
|
| 4585 |
+
"epoch": 0.4259349991834068,
|
| 4586 |
+
"grad_norm": Infinity,
|
| 4587 |
+
"learning_rate": 0.0001288760838714088,
|
| 4588 |
+
"loss": 1.3782,
|
| 4589 |
+
"step": 652
|
| 4590 |
+
},
|
| 4591 |
+
{
|
| 4592 |
+
"epoch": 0.4265882737220317,
|
| 4593 |
+
"grad_norm": Infinity,
|
| 4594 |
+
"learning_rate": 0.0001286727916188834,
|
| 4595 |
+
"loss": 1.3568,
|
| 4596 |
+
"step": 653
|
| 4597 |
+
},
|
| 4598 |
+
{
|
| 4599 |
+
"epoch": 0.4272415482606565,
|
| 4600 |
+
"grad_norm": Infinity,
|
| 4601 |
+
"learning_rate": 0.00012846937017123197,
|
| 4602 |
+
"loss": 1.3555,
|
| 4603 |
+
"step": 654
|
| 4604 |
+
},
|
| 4605 |
+
{
|
| 4606 |
+
"epoch": 0.42789482279928137,
|
| 4607 |
+
"grad_norm": Infinity,
|
| 4608 |
+
"learning_rate": 0.00012826582044503978,
|
| 4609 |
+
"loss": 1.5263,
|
| 4610 |
+
"step": 655
|
| 4611 |
+
},
|
| 4612 |
+
{
|
| 4613 |
+
"epoch": 0.4285480973379063,
|
| 4614 |
+
"grad_norm": Infinity,
|
| 4615 |
+
"learning_rate": 0.00012806214335747018,
|
| 4616 |
+
"loss": 1.502,
|
| 4617 |
+
"step": 656
|
| 4618 |
+
},
|
| 4619 |
+
{
|
| 4620 |
+
"epoch": 0.4292013718765311,
|
| 4621 |
+
"grad_norm": Infinity,
|
| 4622 |
+
"learning_rate": 0.00012785833982626032,
|
| 4623 |
+
"loss": 1.4846,
|
| 4624 |
+
"step": 657
|
| 4625 |
+
},
|
| 4626 |
+
{
|
| 4627 |
+
"epoch": 0.42985464641515597,
|
| 4628 |
+
"grad_norm": Infinity,
|
| 4629 |
+
"learning_rate": 0.00012765441076971712,
|
| 4630 |
+
"loss": 1.5687,
|
| 4631 |
+
"step": 658
|
| 4632 |
+
},
|
| 4633 |
+
{
|
| 4634 |
+
"epoch": 0.4305079209537808,
|
| 4635 |
+
"grad_norm": Infinity,
|
| 4636 |
+
"learning_rate": 0.0001274503571067131,
|
| 4637 |
+
"loss": 1.6626,
|
| 4638 |
+
"step": 659
|
| 4639 |
+
},
|
| 4640 |
+
{
|
| 4641 |
+
"epoch": 0.43116119549240567,
|
| 4642 |
+
"grad_norm": Infinity,
|
| 4643 |
+
"learning_rate": 0.0001272461797566823,
|
| 4644 |
+
"loss": 1.7197,
|
| 4645 |
+
"step": 660
|
| 4646 |
+
},
|
| 4647 |
+
{
|
| 4648 |
+
"epoch": 0.4318144700310305,
|
| 4649 |
+
"grad_norm": Infinity,
|
| 4650 |
+
"learning_rate": 0.0001270418796396159,
|
| 4651 |
+
"loss": 1.7319,
|
| 4652 |
+
"step": 661
|
| 4653 |
+
},
|
| 4654 |
+
{
|
| 4655 |
+
"epoch": 0.4324677445696554,
|
| 4656 |
+
"grad_norm": Infinity,
|
| 4657 |
+
"learning_rate": 0.00012683745767605846,
|
| 4658 |
+
"loss": 1.8091,
|
| 4659 |
+
"step": 662
|
| 4660 |
+
},
|
| 4661 |
+
{
|
| 4662 |
+
"epoch": 0.43312101910828027,
|
| 4663 |
+
"grad_norm": Infinity,
|
| 4664 |
+
"learning_rate": 0.00012663291478710336,
|
| 4665 |
+
"loss": 1.8593,
|
| 4666 |
+
"step": 663
|
| 4667 |
+
},
|
| 4668 |
+
{
|
| 4669 |
+
"epoch": 0.4337742936469051,
|
| 4670 |
+
"grad_norm": Infinity,
|
| 4671 |
+
"learning_rate": 0.00012642825189438902,
|
| 4672 |
+
"loss": 1.9488,
|
| 4673 |
+
"step": 664
|
| 4674 |
+
},
|
| 4675 |
+
{
|
| 4676 |
+
"epoch": 0.43442756818552997,
|
| 4677 |
+
"grad_norm": Infinity,
|
| 4678 |
+
"learning_rate": 0.00012622346992009447,
|
| 4679 |
+
"loss": 2.1199,
|
| 4680 |
+
"step": 665
|
| 4681 |
+
},
|
| 4682 |
+
{
|
| 4683 |
+
"epoch": 0.4350808427241548,
|
| 4684 |
+
"grad_norm": Infinity,
|
| 4685 |
+
"learning_rate": 0.0001260185697869353,
|
| 4686 |
+
"loss": 2.0434,
|
| 4687 |
+
"step": 666
|
| 4688 |
+
},
|
| 4689 |
+
{
|
| 4690 |
+
"epoch": 0.43573411726277966,
|
| 4691 |
+
"grad_norm": Infinity,
|
| 4692 |
+
"learning_rate": 0.00012581355241815965,
|
| 4693 |
+
"loss": 2.3479,
|
| 4694 |
+
"step": 667
|
| 4695 |
+
},
|
| 4696 |
+
{
|
| 4697 |
+
"epoch": 0.43638739180140457,
|
| 4698 |
+
"grad_norm": Infinity,
|
| 4699 |
+
"learning_rate": 0.00012560841873754375,
|
| 4700 |
+
"loss": 2.2125,
|
| 4701 |
+
"step": 668
|
| 4702 |
+
},
|
| 4703 |
+
{
|
| 4704 |
+
"epoch": 0.4370406663400294,
|
| 4705 |
+
"grad_norm": Infinity,
|
| 4706 |
+
"learning_rate": 0.00012540316966938795,
|
| 4707 |
+
"loss": 2.3785,
|
| 4708 |
+
"step": 669
|
| 4709 |
+
},
|
| 4710 |
+
{
|
| 4711 |
+
"epoch": 0.43769394087865426,
|
| 4712 |
+
"grad_norm": Infinity,
|
| 4713 |
+
"learning_rate": 0.00012519780613851254,
|
| 4714 |
+
"loss": 2.6029,
|
| 4715 |
+
"step": 670
|
| 4716 |
+
},
|
| 4717 |
+
{
|
| 4718 |
+
"epoch": 0.4383472154172791,
|
| 4719 |
+
"grad_norm": Infinity,
|
| 4720 |
+
"learning_rate": 0.00012499232907025353,
|
| 4721 |
+
"loss": 2.5433,
|
| 4722 |
+
"step": 671
|
| 4723 |
+
},
|
| 4724 |
+
{
|
| 4725 |
+
"epoch": 0.43900048995590396,
|
| 4726 |
+
"grad_norm": Infinity,
|
| 4727 |
+
"learning_rate": 0.00012478673939045863,
|
| 4728 |
+
"loss": 2.7926,
|
| 4729 |
+
"step": 672
|
| 4730 |
+
},
|
| 4731 |
+
{
|
| 4732 |
+
"epoch": 0.4396537644945288,
|
| 4733 |
+
"grad_norm": Infinity,
|
| 4734 |
+
"learning_rate": 0.00012458103802548274,
|
| 4735 |
+
"loss": 3.0005,
|
| 4736 |
+
"step": 673
|
| 4737 |
+
},
|
| 4738 |
+
{
|
| 4739 |
+
"epoch": 0.44030703903315366,
|
| 4740 |
+
"grad_norm": Infinity,
|
| 4741 |
+
"learning_rate": 0.00012437522590218417,
|
| 4742 |
+
"loss": 3.1366,
|
| 4743 |
+
"step": 674
|
| 4744 |
+
},
|
| 4745 |
+
{
|
| 4746 |
+
"epoch": 0.44096031357177856,
|
| 4747 |
+
"grad_norm": Infinity,
|
| 4748 |
+
"learning_rate": 0.00012416930394792026,
|
| 4749 |
+
"loss": 4.2293,
|
| 4750 |
+
"step": 675
|
| 4751 |
+
},
|
| 4752 |
+
{
|
| 4753 |
+
"epoch": 0.4416135881104034,
|
| 4754 |
+
"grad_norm": Infinity,
|
| 4755 |
+
"learning_rate": 0.0001239632730905432,
|
| 4756 |
+
"loss": 1.2318,
|
| 4757 |
+
"step": 676
|
| 4758 |
+
},
|
| 4759 |
+
{
|
| 4760 |
+
"epoch": 0.44226686264902826,
|
| 4761 |
+
"grad_norm": Infinity,
|
| 4762 |
+
"learning_rate": 0.0001237571342583959,
|
| 4763 |
+
"loss": 1.3872,
|
| 4764 |
+
"step": 677
|
| 4765 |
+
},
|
| 4766 |
+
{
|
| 4767 |
+
"epoch": 0.4429201371876531,
|
| 4768 |
+
"grad_norm": Infinity,
|
| 4769 |
+
"learning_rate": 0.00012355088838030776,
|
| 4770 |
+
"loss": 1.4916,
|
| 4771 |
+
"step": 678
|
| 4772 |
+
},
|
| 4773 |
+
{
|
| 4774 |
+
"epoch": 0.44357341172627796,
|
| 4775 |
+
"grad_norm": Infinity,
|
| 4776 |
+
"learning_rate": 0.00012334453638559057,
|
| 4777 |
+
"loss": 1.434,
|
| 4778 |
+
"step": 679
|
| 4779 |
+
},
|
| 4780 |
+
{
|
| 4781 |
+
"epoch": 0.4442266862649028,
|
| 4782 |
+
"grad_norm": Infinity,
|
| 4783 |
+
"learning_rate": 0.00012313807920403419,
|
| 4784 |
+
"loss": 1.457,
|
| 4785 |
+
"step": 680
|
| 4786 |
+
},
|
| 4787 |
+
{
|
| 4788 |
+
"epoch": 0.4448799608035277,
|
| 4789 |
+
"grad_norm": Infinity,
|
| 4790 |
+
"learning_rate": 0.00012293151776590245,
|
| 4791 |
+
"loss": 1.4525,
|
| 4792 |
+
"step": 681
|
| 4793 |
+
},
|
| 4794 |
+
{
|
| 4795 |
+
"epoch": 0.44553323534215256,
|
| 4796 |
+
"grad_norm": Infinity,
|
| 4797 |
+
"learning_rate": 0.00012272485300192902,
|
| 4798 |
+
"loss": 1.4693,
|
| 4799 |
+
"step": 682
|
| 4800 |
+
},
|
| 4801 |
+
{
|
| 4802 |
+
"epoch": 0.4461865098807774,
|
| 4803 |
+
"grad_norm": Infinity,
|
| 4804 |
+
"learning_rate": 0.00012251808584331304,
|
| 4805 |
+
"loss": 1.493,
|
| 4806 |
+
"step": 683
|
| 4807 |
+
},
|
| 4808 |
+
{
|
| 4809 |
+
"epoch": 0.44683978441940225,
|
| 4810 |
+
"grad_norm": Infinity,
|
| 4811 |
+
"learning_rate": 0.00012231121722171512,
|
| 4812 |
+
"loss": 1.5844,
|
| 4813 |
+
"step": 684
|
| 4814 |
+
},
|
| 4815 |
+
{
|
| 4816 |
+
"epoch": 0.4474930589580271,
|
| 4817 |
+
"grad_norm": Infinity,
|
| 4818 |
+
"learning_rate": 0.00012210424806925301,
|
| 4819 |
+
"loss": 1.6257,
|
| 4820 |
+
"step": 685
|
| 4821 |
+
},
|
| 4822 |
+
{
|
| 4823 |
+
"epoch": 0.44814633349665195,
|
| 4824 |
+
"grad_norm": Infinity,
|
| 4825 |
+
"learning_rate": 0.00012189717931849731,
|
| 4826 |
+
"loss": 1.6306,
|
| 4827 |
+
"step": 686
|
| 4828 |
+
},
|
| 4829 |
+
{
|
| 4830 |
+
"epoch": 0.4487996080352768,
|
| 4831 |
+
"grad_norm": Infinity,
|
| 4832 |
+
"learning_rate": 0.00012169001190246765,
|
| 4833 |
+
"loss": 1.7162,
|
| 4834 |
+
"step": 687
|
| 4835 |
+
},
|
| 4836 |
+
{
|
| 4837 |
+
"epoch": 0.4494528825739017,
|
| 4838 |
+
"grad_norm": Infinity,
|
| 4839 |
+
"learning_rate": 0.00012148274675462801,
|
| 4840 |
+
"loss": 1.8626,
|
| 4841 |
+
"step": 688
|
| 4842 |
+
},
|
| 4843 |
+
{
|
| 4844 |
+
"epoch": 0.45010615711252655,
|
| 4845 |
+
"grad_norm": Infinity,
|
| 4846 |
+
"learning_rate": 0.00012127538480888283,
|
| 4847 |
+
"loss": 1.857,
|
| 4848 |
+
"step": 689
|
| 4849 |
+
},
|
| 4850 |
+
{
|
| 4851 |
+
"epoch": 0.4507594316511514,
|
| 4852 |
+
"grad_norm": Infinity,
|
| 4853 |
+
"learning_rate": 0.00012106792699957263,
|
| 4854 |
+
"loss": 1.9262,
|
| 4855 |
+
"step": 690
|
| 4856 |
+
},
|
| 4857 |
+
{
|
| 4858 |
+
"epoch": 0.45141270618977625,
|
| 4859 |
+
"grad_norm": Infinity,
|
| 4860 |
+
"learning_rate": 0.00012086037426147003,
|
| 4861 |
+
"loss": 2.0395,
|
| 4862 |
+
"step": 691
|
| 4863 |
+
},
|
| 4864 |
+
{
|
| 4865 |
+
"epoch": 0.4520659807284011,
|
| 4866 |
+
"grad_norm": Infinity,
|
| 4867 |
+
"learning_rate": 0.00012065272752977526,
|
| 4868 |
+
"loss": 2.0832,
|
| 4869 |
+
"step": 692
|
| 4870 |
+
},
|
| 4871 |
+
{
|
| 4872 |
+
"epoch": 0.45271925526702594,
|
| 4873 |
+
"grad_norm": Infinity,
|
| 4874 |
+
"learning_rate": 0.00012044498774011209,
|
| 4875 |
+
"loss": 2.5049,
|
| 4876 |
+
"step": 693
|
| 4877 |
+
},
|
| 4878 |
+
{
|
| 4879 |
+
"epoch": 0.45337252980565085,
|
| 4880 |
+
"grad_norm": Infinity,
|
| 4881 |
+
"learning_rate": 0.00012023715582852357,
|
| 4882 |
+
"loss": 2.4738,
|
| 4883 |
+
"step": 694
|
| 4884 |
+
},
|
| 4885 |
+
{
|
| 4886 |
+
"epoch": 0.4540258043442757,
|
| 4887 |
+
"grad_norm": Infinity,
|
| 4888 |
+
"learning_rate": 0.00012002923273146794,
|
| 4889 |
+
"loss": 2.2563,
|
| 4890 |
+
"step": 695
|
| 4891 |
+
},
|
| 4892 |
+
{
|
| 4893 |
+
"epoch": 0.45467907888290054,
|
| 4894 |
+
"grad_norm": Infinity,
|
| 4895 |
+
"learning_rate": 0.0001198212193858142,
|
| 4896 |
+
"loss": 2.936,
|
| 4897 |
+
"step": 696
|
| 4898 |
+
},
|
| 4899 |
+
{
|
| 4900 |
+
"epoch": 0.4553323534215254,
|
| 4901 |
+
"grad_norm": Infinity,
|
| 4902 |
+
"learning_rate": 0.00011961311672883804,
|
| 4903 |
+
"loss": 3.3187,
|
| 4904 |
+
"step": 697
|
| 4905 |
+
},
|
| 4906 |
+
{
|
| 4907 |
+
"epoch": 0.45598562796015024,
|
| 4908 |
+
"grad_norm": Infinity,
|
| 4909 |
+
"learning_rate": 0.00011940492569821753,
|
| 4910 |
+
"loss": 3.1426,
|
| 4911 |
+
"step": 698
|
| 4912 |
+
},
|
| 4913 |
+
{
|
| 4914 |
+
"epoch": 0.4566389024987751,
|
| 4915 |
+
"grad_norm": Infinity,
|
| 4916 |
+
"learning_rate": 0.00011919664723202906,
|
| 4917 |
+
"loss": 3.4786,
|
| 4918 |
+
"step": 699
|
| 4919 |
+
},
|
| 4920 |
+
{
|
| 4921 |
+
"epoch": 0.4572921770374,
|
| 4922 |
+
"grad_norm": Infinity,
|
| 4923 |
+
"learning_rate": 0.00011898828226874284,
|
| 4924 |
+
"loss": 4.6187,
|
| 4925 |
+
"step": 700
|
| 4926 |
+
},
|
| 4927 |
+
{
|
| 4928 |
+
"epoch": 0.45794545157602484,
|
| 4929 |
+
"grad_norm": Infinity,
|
| 4930 |
+
"learning_rate": 0.00011877983174721892,
|
| 4931 |
+
"loss": 1.315,
|
| 4932 |
+
"step": 701
|
| 4933 |
+
},
|
| 4934 |
+
{
|
| 4935 |
+
"epoch": 0.4585987261146497,
|
| 4936 |
+
"grad_norm": Infinity,
|
| 4937 |
+
"learning_rate": 0.00011857129660670281,
|
| 4938 |
+
"loss": 1.3934,
|
| 4939 |
+
"step": 702
|
| 4940 |
+
},
|
| 4941 |
+
{
|
| 4942 |
+
"epoch": 0.45925200065327454,
|
| 4943 |
+
"grad_norm": Infinity,
|
| 4944 |
+
"learning_rate": 0.00011836267778682133,
|
| 4945 |
+
"loss": 1.4157,
|
| 4946 |
+
"step": 703
|
| 4947 |
+
},
|
| 4948 |
+
{
|
| 4949 |
+
"epoch": 0.4599052751918994,
|
| 4950 |
+
"grad_norm": Infinity,
|
| 4951 |
+
"learning_rate": 0.00011815397622757838,
|
| 4952 |
+
"loss": 1.3945,
|
| 4953 |
+
"step": 704
|
| 4954 |
+
},
|
| 4955 |
+
{
|
| 4956 |
+
"epoch": 0.46055854973052424,
|
| 4957 |
+
"grad_norm": Infinity,
|
| 4958 |
+
"learning_rate": 0.00011794519286935055,
|
| 4959 |
+
"loss": 1.4258,
|
| 4960 |
+
"step": 705
|
| 4961 |
+
},
|
| 4962 |
+
{
|
| 4963 |
+
"epoch": 0.4612118242691491,
|
| 4964 |
+
"grad_norm": Infinity,
|
| 4965 |
+
"learning_rate": 0.00011773632865288309,
|
| 4966 |
+
"loss": 1.4199,
|
| 4967 |
+
"step": 706
|
| 4968 |
+
},
|
| 4969 |
+
{
|
| 4970 |
+
"epoch": 0.461865098807774,
|
| 4971 |
+
"grad_norm": Infinity,
|
| 4972 |
+
"learning_rate": 0.00011752738451928566,
|
| 4973 |
+
"loss": 1.5205,
|
| 4974 |
+
"step": 707
|
| 4975 |
+
},
|
| 4976 |
+
{
|
| 4977 |
+
"epoch": 0.46251837334639884,
|
| 4978 |
+
"grad_norm": Infinity,
|
| 4979 |
+
"learning_rate": 0.00011731836141002787,
|
| 4980 |
+
"loss": 1.5761,
|
| 4981 |
+
"step": 708
|
| 4982 |
+
},
|
| 4983 |
+
{
|
| 4984 |
+
"epoch": 0.4631716478850237,
|
| 4985 |
+
"grad_norm": Infinity,
|
| 4986 |
+
"learning_rate": 0.00011710926026693525,
|
| 4987 |
+
"loss": 1.6457,
|
| 4988 |
+
"step": 709
|
| 4989 |
+
},
|
| 4990 |
+
{
|
| 4991 |
+
"epoch": 0.46382492242364853,
|
| 4992 |
+
"grad_norm": Infinity,
|
| 4993 |
+
"learning_rate": 0.00011690008203218493,
|
| 4994 |
+
"loss": 1.8223,
|
| 4995 |
+
"step": 710
|
| 4996 |
+
},
|
| 4997 |
+
{
|
| 4998 |
+
"epoch": 0.4644781969622734,
|
| 4999 |
+
"grad_norm": Infinity,
|
| 5000 |
+
"learning_rate": 0.0001166908276483014,
|
| 5001 |
+
"loss": 1.7146,
|
| 5002 |
+
"step": 711
|
| 5003 |
+
},
|
| 5004 |
+
{
|
| 5005 |
+
"epoch": 0.46513147150089823,
|
| 5006 |
+
"grad_norm": Infinity,
|
| 5007 |
+
"learning_rate": 0.00011648149805815227,
|
| 5008 |
+
"loss": 1.6361,
|
| 5009 |
+
"step": 712
|
| 5010 |
+
},
|
| 5011 |
+
{
|
| 5012 |
+
"epoch": 0.46578474603952313,
|
| 5013 |
+
"grad_norm": Infinity,
|
| 5014 |
+
"learning_rate": 0.000116272094204944,
|
| 5015 |
+
"loss": 1.7237,
|
| 5016 |
+
"step": 713
|
| 5017 |
+
},
|
| 5018 |
+
{
|
| 5019 |
+
"epoch": 0.466438020578148,
|
| 5020 |
+
"grad_norm": Infinity,
|
| 5021 |
+
"learning_rate": 0.00011606261703221772,
|
| 5022 |
+
"loss": 1.8705,
|
| 5023 |
+
"step": 714
|
| 5024 |
+
},
|
| 5025 |
+
{
|
| 5026 |
+
"epoch": 0.46709129511677283,
|
| 5027 |
+
"grad_norm": Infinity,
|
| 5028 |
+
"learning_rate": 0.0001158530674838449,
|
| 5029 |
+
"loss": 1.9701,
|
| 5030 |
+
"step": 715
|
| 5031 |
+
},
|
| 5032 |
+
{
|
| 5033 |
+
"epoch": 0.4677445696553977,
|
| 5034 |
+
"grad_norm": Infinity,
|
| 5035 |
+
"learning_rate": 0.0001156434465040231,
|
| 5036 |
+
"loss": 2.136,
|
| 5037 |
+
"step": 716
|
| 5038 |
+
},
|
| 5039 |
+
{
|
| 5040 |
+
"epoch": 0.46839784419402253,
|
| 5041 |
+
"grad_norm": Infinity,
|
| 5042 |
+
"learning_rate": 0.00011543375503727174,
|
| 5043 |
+
"loss": 2.0878,
|
| 5044 |
+
"step": 717
|
| 5045 |
+
},
|
| 5046 |
+
{
|
| 5047 |
+
"epoch": 0.4690511187326474,
|
| 5048 |
+
"grad_norm": Infinity,
|
| 5049 |
+
"learning_rate": 0.00011522399402842783,
|
| 5050 |
+
"loss": 2.4607,
|
| 5051 |
+
"step": 718
|
| 5052 |
+
},
|
| 5053 |
+
{
|
| 5054 |
+
"epoch": 0.4697043932712722,
|
| 5055 |
+
"grad_norm": Infinity,
|
| 5056 |
+
"learning_rate": 0.00011501416442264184,
|
| 5057 |
+
"loss": 2.4705,
|
| 5058 |
+
"step": 719
|
| 5059 |
+
},
|
| 5060 |
+
{
|
| 5061 |
+
"epoch": 0.47035766780989713,
|
| 5062 |
+
"grad_norm": Infinity,
|
| 5063 |
+
"learning_rate": 0.00011480426716537315,
|
| 5064 |
+
"loss": 2.6055,
|
| 5065 |
+
"step": 720
|
| 5066 |
+
},
|
| 5067 |
+
{
|
| 5068 |
+
"epoch": 0.471010942348522,
|
| 5069 |
+
"grad_norm": Infinity,
|
| 5070 |
+
"learning_rate": 0.00011459430320238611,
|
| 5071 |
+
"loss": 2.7481,
|
| 5072 |
+
"step": 721
|
| 5073 |
+
},
|
| 5074 |
+
{
|
| 5075 |
+
"epoch": 0.4716642168871468,
|
| 5076 |
+
"grad_norm": Infinity,
|
| 5077 |
+
"learning_rate": 0.00011438427347974554,
|
| 5078 |
+
"loss": 2.9075,
|
| 5079 |
+
"step": 722
|
| 5080 |
+
},
|
| 5081 |
+
{
|
| 5082 |
+
"epoch": 0.4723174914257717,
|
| 5083 |
+
"grad_norm": Infinity,
|
| 5084 |
+
"learning_rate": 0.00011417417894381268,
|
| 5085 |
+
"loss": 3.147,
|
| 5086 |
+
"step": 723
|
| 5087 |
+
},
|
| 5088 |
+
{
|
| 5089 |
+
"epoch": 0.4729707659643965,
|
| 5090 |
+
"grad_norm": Infinity,
|
| 5091 |
+
"learning_rate": 0.00011396402054124067,
|
| 5092 |
+
"loss": 3.0189,
|
| 5093 |
+
"step": 724
|
| 5094 |
+
},
|
| 5095 |
+
{
|
| 5096 |
+
"epoch": 0.47362404050302137,
|
| 5097 |
+
"grad_norm": Infinity,
|
| 5098 |
+
"learning_rate": 0.00011375379921897051,
|
| 5099 |
+
"loss": 4.2631,
|
| 5100 |
+
"step": 725
|
| 5101 |
+
},
|
| 5102 |
+
{
|
| 5103 |
+
"epoch": 0.4742773150416463,
|
| 5104 |
+
"grad_norm": Infinity,
|
| 5105 |
+
"learning_rate": 0.00011354351592422665,
|
| 5106 |
+
"loss": 1.4092,
|
| 5107 |
+
"step": 726
|
| 5108 |
+
},
|
| 5109 |
+
{
|
| 5110 |
+
"epoch": 0.4749305895802711,
|
| 5111 |
+
"grad_norm": Infinity,
|
| 5112 |
+
"learning_rate": 0.00011333317160451286,
|
| 5113 |
+
"loss": 1.4541,
|
| 5114 |
+
"step": 727
|
| 5115 |
+
},
|
| 5116 |
+
{
|
| 5117 |
+
"epoch": 0.47558386411889597,
|
| 5118 |
+
"grad_norm": Infinity,
|
| 5119 |
+
"learning_rate": 0.00011312276720760782,
|
| 5120 |
+
"loss": 1.4517,
|
| 5121 |
+
"step": 728
|
| 5122 |
+
},
|
| 5123 |
+
{
|
| 5124 |
+
"epoch": 0.4762371386575208,
|
| 5125 |
+
"grad_norm": Infinity,
|
| 5126 |
+
"learning_rate": 0.00011291230368156087,
|
| 5127 |
+
"loss": 1.3585,
|
| 5128 |
+
"step": 729
|
| 5129 |
+
},
|
| 5130 |
+
{
|
| 5131 |
+
"epoch": 0.47689041319614567,
|
| 5132 |
+
"grad_norm": Infinity,
|
| 5133 |
+
"learning_rate": 0.00011270178197468789,
|
| 5134 |
+
"loss": 1.4466,
|
| 5135 |
+
"step": 730
|
| 5136 |
+
},
|
| 5137 |
+
{
|
| 5138 |
+
"epoch": 0.4775436877347705,
|
| 5139 |
+
"grad_norm": Infinity,
|
| 5140 |
+
"learning_rate": 0.00011249120303556681,
|
| 5141 |
+
"loss": 1.6079,
|
| 5142 |
+
"step": 731
|
| 5143 |
+
},
|
| 5144 |
+
{
|
| 5145 |
+
"epoch": 0.4781969622733954,
|
| 5146 |
+
"grad_norm": Infinity,
|
| 5147 |
+
"learning_rate": 0.0001122805678130335,
|
| 5148 |
+
"loss": 1.5759,
|
| 5149 |
+
"step": 732
|
| 5150 |
+
},
|
| 5151 |
+
{
|
| 5152 |
+
"epoch": 0.47885023681202027,
|
| 5153 |
+
"grad_norm": Infinity,
|
| 5154 |
+
"learning_rate": 0.00011206987725617741,
|
| 5155 |
+
"loss": 1.4571,
|
| 5156 |
+
"step": 733
|
| 5157 |
+
},
|
| 5158 |
+
{
|
| 5159 |
+
"epoch": 0.4795035113506451,
|
| 5160 |
+
"grad_norm": Infinity,
|
| 5161 |
+
"learning_rate": 0.00011185913231433733,
|
| 5162 |
+
"loss": 1.6658,
|
| 5163 |
+
"step": 734
|
| 5164 |
+
},
|
| 5165 |
+
{
|
| 5166 |
+
"epoch": 0.48015678588926997,
|
| 5167 |
+
"grad_norm": Infinity,
|
| 5168 |
+
"learning_rate": 0.00011164833393709706,
|
| 5169 |
+
"loss": 1.7304,
|
| 5170 |
+
"step": 735
|
| 5171 |
+
},
|
| 5172 |
+
{
|
| 5173 |
+
"epoch": 0.4808100604278948,
|
| 5174 |
+
"grad_norm": Infinity,
|
| 5175 |
+
"learning_rate": 0.00011143748307428126,
|
| 5176 |
+
"loss": 1.6036,
|
| 5177 |
+
"step": 736
|
| 5178 |
+
},
|
| 5179 |
+
{
|
| 5180 |
+
"epoch": 0.48146333496651966,
|
| 5181 |
+
"grad_norm": Infinity,
|
| 5182 |
+
"learning_rate": 0.00011122658067595094,
|
| 5183 |
+
"loss": 1.7269,
|
| 5184 |
+
"step": 737
|
| 5185 |
+
},
|
| 5186 |
+
{
|
| 5187 |
+
"epoch": 0.4821166095051445,
|
| 5188 |
+
"grad_norm": Infinity,
|
| 5189 |
+
"learning_rate": 0.00011101562769239946,
|
| 5190 |
+
"loss": 1.7812,
|
| 5191 |
+
"step": 738
|
| 5192 |
+
},
|
| 5193 |
+
{
|
| 5194 |
+
"epoch": 0.4827698840437694,
|
| 5195 |
+
"grad_norm": Infinity,
|
| 5196 |
+
"learning_rate": 0.00011080462507414806,
|
| 5197 |
+
"loss": 1.9052,
|
| 5198 |
+
"step": 739
|
| 5199 |
+
},
|
| 5200 |
+
{
|
| 5201 |
+
"epoch": 0.48342315858239426,
|
| 5202 |
+
"grad_norm": Infinity,
|
| 5203 |
+
"learning_rate": 0.00011059357377194161,
|
| 5204 |
+
"loss": 2.0699,
|
| 5205 |
+
"step": 740
|
| 5206 |
+
},
|
| 5207 |
+
{
|
| 5208 |
+
"epoch": 0.4840764331210191,
|
| 5209 |
+
"grad_norm": Infinity,
|
| 5210 |
+
"learning_rate": 0.00011038247473674434,
|
| 5211 |
+
"loss": 2.2282,
|
| 5212 |
+
"step": 741
|
| 5213 |
+
},
|
| 5214 |
+
{
|
| 5215 |
+
"epoch": 0.48472970765964396,
|
| 5216 |
+
"grad_norm": Infinity,
|
| 5217 |
+
"learning_rate": 0.0001101713289197356,
|
| 5218 |
+
"loss": 2.5375,
|
| 5219 |
+
"step": 742
|
| 5220 |
+
},
|
| 5221 |
+
{
|
| 5222 |
+
"epoch": 0.4853829821982688,
|
| 5223 |
+
"grad_norm": Infinity,
|
| 5224 |
+
"learning_rate": 0.0001099601372723055,
|
| 5225 |
+
"loss": 2.5702,
|
| 5226 |
+
"step": 743
|
| 5227 |
+
},
|
| 5228 |
+
{
|
| 5229 |
+
"epoch": 0.48603625673689366,
|
| 5230 |
+
"grad_norm": Infinity,
|
| 5231 |
+
"learning_rate": 0.00010974890074605062,
|
| 5232 |
+
"loss": 2.7113,
|
| 5233 |
+
"step": 744
|
| 5234 |
+
},
|
| 5235 |
+
{
|
| 5236 |
+
"epoch": 0.48668953127551856,
|
| 5237 |
+
"grad_norm": Infinity,
|
| 5238 |
+
"learning_rate": 0.00010953762029276982,
|
| 5239 |
+
"loss": 3.0058,
|
| 5240 |
+
"step": 745
|
| 5241 |
+
},
|
| 5242 |
+
{
|
| 5243 |
+
"epoch": 0.4873428058141434,
|
| 5244 |
+
"grad_norm": Infinity,
|
| 5245 |
+
"learning_rate": 0.00010932629686445986,
|
| 5246 |
+
"loss": 2.9393,
|
| 5247 |
+
"step": 746
|
| 5248 |
+
},
|
| 5249 |
+
{
|
| 5250 |
+
"epoch": 0.48799608035276826,
|
| 5251 |
+
"grad_norm": Infinity,
|
| 5252 |
+
"learning_rate": 0.00010911493141331113,
|
| 5253 |
+
"loss": 3.0283,
|
| 5254 |
+
"step": 747
|
| 5255 |
+
},
|
| 5256 |
+
{
|
| 5257 |
+
"epoch": 0.4886493548913931,
|
| 5258 |
+
"grad_norm": Infinity,
|
| 5259 |
+
"learning_rate": 0.00010890352489170341,
|
| 5260 |
+
"loss": 3.3168,
|
| 5261 |
+
"step": 748
|
| 5262 |
+
},
|
| 5263 |
+
{
|
| 5264 |
+
"epoch": 0.48930262943001795,
|
| 5265 |
+
"grad_norm": Infinity,
|
| 5266 |
+
"learning_rate": 0.00010869207825220147,
|
| 5267 |
+
"loss": 3.5297,
|
| 5268 |
+
"step": 749
|
| 5269 |
+
},
|
| 5270 |
+
{
|
| 5271 |
+
"epoch": 0.4899559039686428,
|
| 5272 |
+
"grad_norm": Infinity,
|
| 5273 |
+
"learning_rate": 0.00010848059244755093,
|
| 5274 |
+
"loss": 4.1073,
|
| 5275 |
+
"step": 750
|
| 5276 |
+
},
|
| 5277 |
+
{
|
| 5278 |
+
"epoch": 0.4906091785072677,
|
| 5279 |
+
"grad_norm": Infinity,
|
| 5280 |
+
"learning_rate": 0.0001082690684306738,
|
| 5281 |
+
"loss": 1.3489,
|
| 5282 |
+
"step": 751
|
| 5283 |
+
},
|
| 5284 |
+
{
|
| 5285 |
+
"epoch": 0.49126245304589256,
|
| 5286 |
+
"grad_norm": Infinity,
|
| 5287 |
+
"learning_rate": 0.00010805750715466429,
|
| 5288 |
+
"loss": 1.3607,
|
| 5289 |
+
"step": 752
|
| 5290 |
+
},
|
| 5291 |
+
{
|
| 5292 |
+
"epoch": 0.4919157275845174,
|
| 5293 |
+
"grad_norm": Infinity,
|
| 5294 |
+
"learning_rate": 0.0001078459095727845,
|
| 5295 |
+
"loss": 1.4329,
|
| 5296 |
+
"step": 753
|
| 5297 |
+
},
|
| 5298 |
+
{
|
| 5299 |
+
"epoch": 0.49256900212314225,
|
| 5300 |
+
"grad_norm": Infinity,
|
| 5301 |
+
"learning_rate": 0.00010763427663846015,
|
| 5302 |
+
"loss": 1.4151,
|
| 5303 |
+
"step": 754
|
| 5304 |
+
},
|
| 5305 |
+
{
|
| 5306 |
+
"epoch": 0.4932222766617671,
|
| 5307 |
+
"grad_norm": Infinity,
|
| 5308 |
+
"learning_rate": 0.00010742260930527625,
|
| 5309 |
+
"loss": 1.3808,
|
| 5310 |
+
"step": 755
|
| 5311 |
+
},
|
| 5312 |
+
{
|
| 5313 |
+
"epoch": 0.49387555120039195,
|
| 5314 |
+
"grad_norm": Infinity,
|
| 5315 |
+
"learning_rate": 0.00010721090852697275,
|
| 5316 |
+
"loss": 1.5492,
|
| 5317 |
+
"step": 756
|
| 5318 |
+
},
|
| 5319 |
+
{
|
| 5320 |
+
"epoch": 0.4945288257390168,
|
| 5321 |
+
"grad_norm": Infinity,
|
| 5322 |
+
"learning_rate": 0.00010699917525744032,
|
| 5323 |
+
"loss": 1.4695,
|
| 5324 |
+
"step": 757
|
| 5325 |
+
},
|
| 5326 |
+
{
|
| 5327 |
+
"epoch": 0.4951821002776417,
|
| 5328 |
+
"grad_norm": Infinity,
|
| 5329 |
+
"learning_rate": 0.00010678741045071609,
|
| 5330 |
+
"loss": 1.5456,
|
| 5331 |
+
"step": 758
|
| 5332 |
+
},
|
| 5333 |
+
{
|
| 5334 |
+
"epoch": 0.49583537481626655,
|
| 5335 |
+
"grad_norm": Infinity,
|
| 5336 |
+
"learning_rate": 0.0001065756150609792,
|
| 5337 |
+
"loss": 1.5857,
|
| 5338 |
+
"step": 759
|
| 5339 |
+
},
|
| 5340 |
+
{
|
| 5341 |
+
"epoch": 0.4964886493548914,
|
| 5342 |
+
"grad_norm": Infinity,
|
| 5343 |
+
"learning_rate": 0.00010636379004254664,
|
| 5344 |
+
"loss": 1.4613,
|
| 5345 |
+
"step": 760
|
| 5346 |
+
},
|
| 5347 |
+
{
|
| 5348 |
+
"epoch": 0.49714192389351625,
|
| 5349 |
+
"grad_norm": Infinity,
|
| 5350 |
+
"learning_rate": 0.0001061519363498689,
|
| 5351 |
+
"loss": 1.704,
|
| 5352 |
+
"step": 761
|
| 5353 |
+
},
|
| 5354 |
+
{
|
| 5355 |
+
"epoch": 0.4977951984321411,
|
| 5356 |
+
"grad_norm": Infinity,
|
| 5357 |
+
"learning_rate": 0.00010594005493752568,
|
| 5358 |
+
"loss": 1.6429,
|
| 5359 |
+
"step": 762
|
| 5360 |
+
},
|
| 5361 |
+
{
|
| 5362 |
+
"epoch": 0.49844847297076594,
|
| 5363 |
+
"grad_norm": Infinity,
|
| 5364 |
+
"learning_rate": 0.00010572814676022158,
|
| 5365 |
+
"loss": 1.7032,
|
| 5366 |
+
"step": 763
|
| 5367 |
+
},
|
| 5368 |
+
{
|
| 5369 |
+
"epoch": 0.49910174750939085,
|
| 5370 |
+
"grad_norm": Infinity,
|
| 5371 |
+
"learning_rate": 0.00010551621277278176,
|
| 5372 |
+
"loss": 1.6818,
|
| 5373 |
+
"step": 764
|
| 5374 |
+
},
|
| 5375 |
+
{
|
| 5376 |
+
"epoch": 0.4997550220480157,
|
| 5377 |
+
"grad_norm": Infinity,
|
| 5378 |
+
"learning_rate": 0.00010530425393014774,
|
| 5379 |
+
"loss": 2.0321,
|
| 5380 |
+
"step": 765
|
| 5381 |
+
},
|
| 5382 |
+
{
|
| 5383 |
+
"epoch": 0.5004082965866405,
|
| 5384 |
+
"grad_norm": Infinity,
|
| 5385 |
+
"learning_rate": 0.00010509227118737298,
|
| 5386 |
+
"loss": 1.8955,
|
| 5387 |
+
"step": 766
|
| 5388 |
+
},
|
| 5389 |
+
{
|
| 5390 |
+
"epoch": 0.5004082965866405,
|
| 5391 |
+
"eval_loss": 2.0970346927642822,
|
| 5392 |
+
"eval_runtime": 161.1777,
|
| 5393 |
+
"eval_samples_per_second": 15.995,
|
| 5394 |
+
"eval_steps_per_second": 4.002,
|
| 5395 |
+
"step": 766
|
| 5396 |
}
|
| 5397 |
],
|
| 5398 |
"logging_steps": 1,
|
|
|
|
| 5412 |
"attributes": {}
|
| 5413 |
}
|
| 5414 |
},
|
| 5415 |
+
"total_flos": 3.153985917513892e+17,
|
| 5416 |
"train_batch_size": 2,
|
| 5417 |
"trial_name": null,
|
| 5418 |
"trial_params": null
|