Training in progress, step 1423, checkpoint
Browse files
last-checkpoint/adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 15182728
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d7c954034480e2c181641b5d82fdf37e66c016816a91398aefeae56cc54a0dc7
|
| 3 |
size 15182728
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 8076116
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:022c4eece60b498a82e0c90b1f98bcb19b2ff64c9f2867b071a0eecd6e33847e
|
| 3 |
size 8076116
|
last-checkpoint/rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14244
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5acea62cbaba34d5866ecd7f4d209876d05cf299e493d7e65b7e6be6a8d01aa7
|
| 3 |
size 14244
|
last-checkpoint/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1064
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a9d85461961714f3b571d1c5a9864a16b570870cca12155a77c936abb3216a83
|
| 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": 356,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -7515,6 +7515,2491 @@
|
|
| 7515 |
"eval_samples_per_second": 50.44,
|
| 7516 |
"eval_steps_per_second": 25.23,
|
| 7517 |
"step": 1068
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 7518 |
}
|
| 7519 |
],
|
| 7520 |
"logging_steps": 1,
|
|
@@ -7529,12 +10014,12 @@
|
|
| 7529 |
"should_evaluate": false,
|
| 7530 |
"should_log": false,
|
| 7531 |
"should_save": true,
|
| 7532 |
-
"should_training_stop":
|
| 7533 |
},
|
| 7534 |
"attributes": {}
|
| 7535 |
}
|
| 7536 |
},
|
| 7537 |
-
"total_flos":
|
| 7538 |
"train_batch_size": 2,
|
| 7539 |
"trial_name": null,
|
| 7540 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.24081908952445422,
|
| 5 |
"eval_steps": 356,
|
| 6 |
+
"global_step": 1423,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 7515 |
"eval_samples_per_second": 50.44,
|
| 7516 |
"eval_steps_per_second": 25.23,
|
| 7517 |
"step": 1068
|
| 7518 |
+
},
|
| 7519 |
+
{
|
| 7520 |
+
"epoch": 0.18091047554577763,
|
| 7521 |
+
"grad_norm": 4.463649749755859,
|
| 7522 |
+
"learning_rate": 2.9407331041607566e-05,
|
| 7523 |
+
"loss": 2.4297,
|
| 7524 |
+
"step": 1069
|
| 7525 |
+
},
|
| 7526 |
+
{
|
| 7527 |
+
"epoch": 0.18107970891859876,
|
| 7528 |
+
"grad_norm": 5.176490306854248,
|
| 7529 |
+
"learning_rate": 2.925002917124886e-05,
|
| 7530 |
+
"loss": 2.6072,
|
| 7531 |
+
"step": 1070
|
| 7532 |
+
},
|
| 7533 |
+
{
|
| 7534 |
+
"epoch": 0.18124894229141986,
|
| 7535 |
+
"grad_norm": 3.740882158279419,
|
| 7536 |
+
"learning_rate": 2.9093077037831827e-05,
|
| 7537 |
+
"loss": 2.7744,
|
| 7538 |
+
"step": 1071
|
| 7539 |
+
},
|
| 7540 |
+
{
|
| 7541 |
+
"epoch": 0.18141817566424098,
|
| 7542 |
+
"grad_norm": 4.830148220062256,
|
| 7543 |
+
"learning_rate": 2.8936475417214794e-05,
|
| 7544 |
+
"loss": 2.7888,
|
| 7545 |
+
"step": 1072
|
| 7546 |
+
},
|
| 7547 |
+
{
|
| 7548 |
+
"epoch": 0.1815874090370621,
|
| 7549 |
+
"grad_norm": 3.236553907394409,
|
| 7550 |
+
"learning_rate": 2.878022508352355e-05,
|
| 7551 |
+
"loss": 1.8341,
|
| 7552 |
+
"step": 1073
|
| 7553 |
+
},
|
| 7554 |
+
{
|
| 7555 |
+
"epoch": 0.18175664240988323,
|
| 7556 |
+
"grad_norm": 3.8409881591796875,
|
| 7557 |
+
"learning_rate": 2.862432680914725e-05,
|
| 7558 |
+
"loss": 2.2543,
|
| 7559 |
+
"step": 1074
|
| 7560 |
+
},
|
| 7561 |
+
{
|
| 7562 |
+
"epoch": 0.18192587578270436,
|
| 7563 |
+
"grad_norm": 4.8214898109436035,
|
| 7564 |
+
"learning_rate": 2.846878136473472e-05,
|
| 7565 |
+
"loss": 2.4273,
|
| 7566 |
+
"step": 1075
|
| 7567 |
+
},
|
| 7568 |
+
{
|
| 7569 |
+
"epoch": 0.18209510915552546,
|
| 7570 |
+
"grad_norm": 3.8864827156066895,
|
| 7571 |
+
"learning_rate": 2.8313589519190843e-05,
|
| 7572 |
+
"loss": 2.4231,
|
| 7573 |
+
"step": 1076
|
| 7574 |
+
},
|
| 7575 |
+
{
|
| 7576 |
+
"epoch": 0.18226434252834658,
|
| 7577 |
+
"grad_norm": 3.7336626052856445,
|
| 7578 |
+
"learning_rate": 2.815875203967233e-05,
|
| 7579 |
+
"loss": 2.1571,
|
| 7580 |
+
"step": 1077
|
| 7581 |
+
},
|
| 7582 |
+
{
|
| 7583 |
+
"epoch": 0.1824335759011677,
|
| 7584 |
+
"grad_norm": 4.008172988891602,
|
| 7585 |
+
"learning_rate": 2.8004269691584252e-05,
|
| 7586 |
+
"loss": 2.4853,
|
| 7587 |
+
"step": 1078
|
| 7588 |
+
},
|
| 7589 |
+
{
|
| 7590 |
+
"epoch": 0.18260280927398884,
|
| 7591 |
+
"grad_norm": 4.489894390106201,
|
| 7592 |
+
"learning_rate": 2.7850143238576198e-05,
|
| 7593 |
+
"loss": 2.6935,
|
| 7594 |
+
"step": 1079
|
| 7595 |
+
},
|
| 7596 |
+
{
|
| 7597 |
+
"epoch": 0.18277204264680996,
|
| 7598 |
+
"grad_norm": 3.847583770751953,
|
| 7599 |
+
"learning_rate": 2.7696373442538405e-05,
|
| 7600 |
+
"loss": 2.0218,
|
| 7601 |
+
"step": 1080
|
| 7602 |
+
},
|
| 7603 |
+
{
|
| 7604 |
+
"epoch": 0.18294127601963106,
|
| 7605 |
+
"grad_norm": 3.73066782951355,
|
| 7606 |
+
"learning_rate": 2.7542961063598104e-05,
|
| 7607 |
+
"loss": 2.0314,
|
| 7608 |
+
"step": 1081
|
| 7609 |
+
},
|
| 7610 |
+
{
|
| 7611 |
+
"epoch": 0.1831105093924522,
|
| 7612 |
+
"grad_norm": 4.3974928855896,
|
| 7613 |
+
"learning_rate": 2.738990686011572e-05,
|
| 7614 |
+
"loss": 2.4177,
|
| 7615 |
+
"step": 1082
|
| 7616 |
+
},
|
| 7617 |
+
{
|
| 7618 |
+
"epoch": 0.1832797427652733,
|
| 7619 |
+
"grad_norm": 5.359389781951904,
|
| 7620 |
+
"learning_rate": 2.7237211588681065e-05,
|
| 7621 |
+
"loss": 2.5658,
|
| 7622 |
+
"step": 1083
|
| 7623 |
+
},
|
| 7624 |
+
{
|
| 7625 |
+
"epoch": 0.18344897613809444,
|
| 7626 |
+
"grad_norm": 3.936758518218994,
|
| 7627 |
+
"learning_rate": 2.708487600410966e-05,
|
| 7628 |
+
"loss": 2.6159,
|
| 7629 |
+
"step": 1084
|
| 7630 |
+
},
|
| 7631 |
+
{
|
| 7632 |
+
"epoch": 0.18361820951091556,
|
| 7633 |
+
"grad_norm": 4.642032623291016,
|
| 7634 |
+
"learning_rate": 2.6932900859439047e-05,
|
| 7635 |
+
"loss": 2.444,
|
| 7636 |
+
"step": 1085
|
| 7637 |
+
},
|
| 7638 |
+
{
|
| 7639 |
+
"epoch": 0.18378744288373666,
|
| 7640 |
+
"grad_norm": 4.450744152069092,
|
| 7641 |
+
"learning_rate": 2.6781286905924928e-05,
|
| 7642 |
+
"loss": 2.3611,
|
| 7643 |
+
"step": 1086
|
| 7644 |
+
},
|
| 7645 |
+
{
|
| 7646 |
+
"epoch": 0.1839566762565578,
|
| 7647 |
+
"grad_norm": 3.1011619567871094,
|
| 7648 |
+
"learning_rate": 2.6630034893037614e-05,
|
| 7649 |
+
"loss": 1.9827,
|
| 7650 |
+
"step": 1087
|
| 7651 |
+
},
|
| 7652 |
+
{
|
| 7653 |
+
"epoch": 0.18412590962937891,
|
| 7654 |
+
"grad_norm": 3.8112988471984863,
|
| 7655 |
+
"learning_rate": 2.647914556845823e-05,
|
| 7656 |
+
"loss": 2.3059,
|
| 7657 |
+
"step": 1088
|
| 7658 |
+
},
|
| 7659 |
+
{
|
| 7660 |
+
"epoch": 0.18429514300220004,
|
| 7661 |
+
"grad_norm": 3.6921634674072266,
|
| 7662 |
+
"learning_rate": 2.6328619678074983e-05,
|
| 7663 |
+
"loss": 2.2578,
|
| 7664 |
+
"step": 1089
|
| 7665 |
+
},
|
| 7666 |
+
{
|
| 7667 |
+
"epoch": 0.18446437637502117,
|
| 7668 |
+
"grad_norm": 4.581288814544678,
|
| 7669 |
+
"learning_rate": 2.617845796597954e-05,
|
| 7670 |
+
"loss": 2.2475,
|
| 7671 |
+
"step": 1090
|
| 7672 |
+
},
|
| 7673 |
+
{
|
| 7674 |
+
"epoch": 0.18463360974784226,
|
| 7675 |
+
"grad_norm": 3.7660021781921387,
|
| 7676 |
+
"learning_rate": 2.6028661174463387e-05,
|
| 7677 |
+
"loss": 2.449,
|
| 7678 |
+
"step": 1091
|
| 7679 |
+
},
|
| 7680 |
+
{
|
| 7681 |
+
"epoch": 0.1848028431206634,
|
| 7682 |
+
"grad_norm": 3.9602880477905273,
|
| 7683 |
+
"learning_rate": 2.5879230044014004e-05,
|
| 7684 |
+
"loss": 2.8594,
|
| 7685 |
+
"step": 1092
|
| 7686 |
+
},
|
| 7687 |
+
{
|
| 7688 |
+
"epoch": 0.18497207649348452,
|
| 7689 |
+
"grad_norm": 3.1530981063842773,
|
| 7690 |
+
"learning_rate": 2.5730165313311393e-05,
|
| 7691 |
+
"loss": 1.9199,
|
| 7692 |
+
"step": 1093
|
| 7693 |
+
},
|
| 7694 |
+
{
|
| 7695 |
+
"epoch": 0.18514130986630564,
|
| 7696 |
+
"grad_norm": 5.039895057678223,
|
| 7697 |
+
"learning_rate": 2.5581467719224338e-05,
|
| 7698 |
+
"loss": 2.8326,
|
| 7699 |
+
"step": 1094
|
| 7700 |
+
},
|
| 7701 |
+
{
|
| 7702 |
+
"epoch": 0.18531054323912677,
|
| 7703 |
+
"grad_norm": 4.700978755950928,
|
| 7704 |
+
"learning_rate": 2.5433137996806722e-05,
|
| 7705 |
+
"loss": 2.4114,
|
| 7706 |
+
"step": 1095
|
| 7707 |
+
},
|
| 7708 |
+
{
|
| 7709 |
+
"epoch": 0.18547977661194787,
|
| 7710 |
+
"grad_norm": 3.5870587825775146,
|
| 7711 |
+
"learning_rate": 2.5285176879293902e-05,
|
| 7712 |
+
"loss": 2.1849,
|
| 7713 |
+
"step": 1096
|
| 7714 |
+
},
|
| 7715 |
+
{
|
| 7716 |
+
"epoch": 0.185649009984769,
|
| 7717 |
+
"grad_norm": 4.496454238891602,
|
| 7718 |
+
"learning_rate": 2.5137585098099236e-05,
|
| 7719 |
+
"loss": 2.905,
|
| 7720 |
+
"step": 1097
|
| 7721 |
+
},
|
| 7722 |
+
{
|
| 7723 |
+
"epoch": 0.18581824335759012,
|
| 7724 |
+
"grad_norm": 4.770108699798584,
|
| 7725 |
+
"learning_rate": 2.499036338281021e-05,
|
| 7726 |
+
"loss": 3.0991,
|
| 7727 |
+
"step": 1098
|
| 7728 |
+
},
|
| 7729 |
+
{
|
| 7730 |
+
"epoch": 0.18598747673041124,
|
| 7731 |
+
"grad_norm": 4.346565246582031,
|
| 7732 |
+
"learning_rate": 2.484351246118507e-05,
|
| 7733 |
+
"loss": 2.0213,
|
| 7734 |
+
"step": 1099
|
| 7735 |
+
},
|
| 7736 |
+
{
|
| 7737 |
+
"epoch": 0.18615671010323237,
|
| 7738 |
+
"grad_norm": 4.078486919403076,
|
| 7739 |
+
"learning_rate": 2.4697033059149132e-05,
|
| 7740 |
+
"loss": 2.7253,
|
| 7741 |
+
"step": 1100
|
| 7742 |
+
},
|
| 7743 |
+
{
|
| 7744 |
+
"epoch": 0.18632594347605347,
|
| 7745 |
+
"grad_norm": 5.557118892669678,
|
| 7746 |
+
"learning_rate": 2.45509259007911e-05,
|
| 7747 |
+
"loss": 3.4271,
|
| 7748 |
+
"step": 1101
|
| 7749 |
+
},
|
| 7750 |
+
{
|
| 7751 |
+
"epoch": 0.1864951768488746,
|
| 7752 |
+
"grad_norm": 4.36812686920166,
|
| 7753 |
+
"learning_rate": 2.440519170835962e-05,
|
| 7754 |
+
"loss": 2.8983,
|
| 7755 |
+
"step": 1102
|
| 7756 |
+
},
|
| 7757 |
+
{
|
| 7758 |
+
"epoch": 0.18666441022169572,
|
| 7759 |
+
"grad_norm": 3.7250893115997314,
|
| 7760 |
+
"learning_rate": 2.4259831202259697e-05,
|
| 7761 |
+
"loss": 2.417,
|
| 7762 |
+
"step": 1103
|
| 7763 |
+
},
|
| 7764 |
+
{
|
| 7765 |
+
"epoch": 0.18683364359451685,
|
| 7766 |
+
"grad_norm": 3.815370798110962,
|
| 7767 |
+
"learning_rate": 2.411484510104901e-05,
|
| 7768 |
+
"loss": 2.6491,
|
| 7769 |
+
"step": 1104
|
| 7770 |
+
},
|
| 7771 |
+
{
|
| 7772 |
+
"epoch": 0.18700287696733797,
|
| 7773 |
+
"grad_norm": 3.851109266281128,
|
| 7774 |
+
"learning_rate": 2.3970234121434553e-05,
|
| 7775 |
+
"loss": 2.5979,
|
| 7776 |
+
"step": 1105
|
| 7777 |
+
},
|
| 7778 |
+
{
|
| 7779 |
+
"epoch": 0.18717211034015907,
|
| 7780 |
+
"grad_norm": 3.8588759899139404,
|
| 7781 |
+
"learning_rate": 2.382599897826896e-05,
|
| 7782 |
+
"loss": 2.0042,
|
| 7783 |
+
"step": 1106
|
| 7784 |
+
},
|
| 7785 |
+
{
|
| 7786 |
+
"epoch": 0.1873413437129802,
|
| 7787 |
+
"grad_norm": 5.41898775100708,
|
| 7788 |
+
"learning_rate": 2.368214038454695e-05,
|
| 7789 |
+
"loss": 2.9259,
|
| 7790 |
+
"step": 1107
|
| 7791 |
+
},
|
| 7792 |
+
{
|
| 7793 |
+
"epoch": 0.18751057708580132,
|
| 7794 |
+
"grad_norm": 4.365620136260986,
|
| 7795 |
+
"learning_rate": 2.353865905140187e-05,
|
| 7796 |
+
"loss": 2.5574,
|
| 7797 |
+
"step": 1108
|
| 7798 |
+
},
|
| 7799 |
+
{
|
| 7800 |
+
"epoch": 0.18767981045862245,
|
| 7801 |
+
"grad_norm": 5.058574199676514,
|
| 7802 |
+
"learning_rate": 2.339555568810221e-05,
|
| 7803 |
+
"loss": 2.2964,
|
| 7804 |
+
"step": 1109
|
| 7805 |
+
},
|
| 7806 |
+
{
|
| 7807 |
+
"epoch": 0.18784904383144357,
|
| 7808 |
+
"grad_norm": 4.100536823272705,
|
| 7809 |
+
"learning_rate": 2.325283100204798e-05,
|
| 7810 |
+
"loss": 2.3641,
|
| 7811 |
+
"step": 1110
|
| 7812 |
+
},
|
| 7813 |
+
{
|
| 7814 |
+
"epoch": 0.18801827720426467,
|
| 7815 |
+
"grad_norm": 4.405045986175537,
|
| 7816 |
+
"learning_rate": 2.3110485698767335e-05,
|
| 7817 |
+
"loss": 2.3162,
|
| 7818 |
+
"step": 1111
|
| 7819 |
+
},
|
| 7820 |
+
{
|
| 7821 |
+
"epoch": 0.1881875105770858,
|
| 7822 |
+
"grad_norm": 4.183629989624023,
|
| 7823 |
+
"learning_rate": 2.296852048191306e-05,
|
| 7824 |
+
"loss": 2.2064,
|
| 7825 |
+
"step": 1112
|
| 7826 |
+
},
|
| 7827 |
+
{
|
| 7828 |
+
"epoch": 0.18835674394990692,
|
| 7829 |
+
"grad_norm": 3.983504295349121,
|
| 7830 |
+
"learning_rate": 2.2826936053258984e-05,
|
| 7831 |
+
"loss": 2.4288,
|
| 7832 |
+
"step": 1113
|
| 7833 |
+
},
|
| 7834 |
+
{
|
| 7835 |
+
"epoch": 0.18852597732272805,
|
| 7836 |
+
"grad_norm": 4.282350063323975,
|
| 7837 |
+
"learning_rate": 2.2685733112696604e-05,
|
| 7838 |
+
"loss": 2.2867,
|
| 7839 |
+
"step": 1114
|
| 7840 |
+
},
|
| 7841 |
+
{
|
| 7842 |
+
"epoch": 0.18869521069554918,
|
| 7843 |
+
"grad_norm": 5.274227142333984,
|
| 7844 |
+
"learning_rate": 2.2544912358231705e-05,
|
| 7845 |
+
"loss": 2.6285,
|
| 7846 |
+
"step": 1115
|
| 7847 |
+
},
|
| 7848 |
+
{
|
| 7849 |
+
"epoch": 0.18886444406837027,
|
| 7850 |
+
"grad_norm": 3.6848084926605225,
|
| 7851 |
+
"learning_rate": 2.24044744859807e-05,
|
| 7852 |
+
"loss": 2.1661,
|
| 7853 |
+
"step": 1116
|
| 7854 |
+
},
|
| 7855 |
+
{
|
| 7856 |
+
"epoch": 0.1890336774411914,
|
| 7857 |
+
"grad_norm": 3.2864272594451904,
|
| 7858 |
+
"learning_rate": 2.226442019016739e-05,
|
| 7859 |
+
"loss": 2.4459,
|
| 7860 |
+
"step": 1117
|
| 7861 |
+
},
|
| 7862 |
+
{
|
| 7863 |
+
"epoch": 0.18920291081401253,
|
| 7864 |
+
"grad_norm": 4.219311714172363,
|
| 7865 |
+
"learning_rate": 2.2124750163119455e-05,
|
| 7866 |
+
"loss": 2.9112,
|
| 7867 |
+
"step": 1118
|
| 7868 |
+
},
|
| 7869 |
+
{
|
| 7870 |
+
"epoch": 0.18937214418683365,
|
| 7871 |
+
"grad_norm": 4.823033332824707,
|
| 7872 |
+
"learning_rate": 2.1985465095264978e-05,
|
| 7873 |
+
"loss": 3.3097,
|
| 7874 |
+
"step": 1119
|
| 7875 |
+
},
|
| 7876 |
+
{
|
| 7877 |
+
"epoch": 0.18954137755965475,
|
| 7878 |
+
"grad_norm": 4.152816295623779,
|
| 7879 |
+
"learning_rate": 2.184656567512907e-05,
|
| 7880 |
+
"loss": 2.3346,
|
| 7881 |
+
"step": 1120
|
| 7882 |
+
},
|
| 7883 |
+
{
|
| 7884 |
+
"epoch": 0.18971061093247588,
|
| 7885 |
+
"grad_norm": 3.9774250984191895,
|
| 7886 |
+
"learning_rate": 2.170805258933055e-05,
|
| 7887 |
+
"loss": 2.7674,
|
| 7888 |
+
"step": 1121
|
| 7889 |
+
},
|
| 7890 |
+
{
|
| 7891 |
+
"epoch": 0.189879844305297,
|
| 7892 |
+
"grad_norm": 3.8187763690948486,
|
| 7893 |
+
"learning_rate": 2.156992652257839e-05,
|
| 7894 |
+
"loss": 1.9397,
|
| 7895 |
+
"step": 1122
|
| 7896 |
+
},
|
| 7897 |
+
{
|
| 7898 |
+
"epoch": 0.19004907767811813,
|
| 7899 |
+
"grad_norm": 5.369848251342773,
|
| 7900 |
+
"learning_rate": 2.1432188157668487e-05,
|
| 7901 |
+
"loss": 2.9257,
|
| 7902 |
+
"step": 1123
|
| 7903 |
+
},
|
| 7904 |
+
{
|
| 7905 |
+
"epoch": 0.19021831105093925,
|
| 7906 |
+
"grad_norm": 3.817648410797119,
|
| 7907 |
+
"learning_rate": 2.1294838175480235e-05,
|
| 7908 |
+
"loss": 2.4691,
|
| 7909 |
+
"step": 1124
|
| 7910 |
+
},
|
| 7911 |
+
{
|
| 7912 |
+
"epoch": 0.19038754442376035,
|
| 7913 |
+
"grad_norm": 3.840128183364868,
|
| 7914 |
+
"learning_rate": 2.1157877254973058e-05,
|
| 7915 |
+
"loss": 2.7187,
|
| 7916 |
+
"step": 1125
|
| 7917 |
+
},
|
| 7918 |
+
{
|
| 7919 |
+
"epoch": 0.19055677779658148,
|
| 7920 |
+
"grad_norm": 4.067428112030029,
|
| 7921 |
+
"learning_rate": 2.1021306073183167e-05,
|
| 7922 |
+
"loss": 1.8525,
|
| 7923 |
+
"step": 1126
|
| 7924 |
+
},
|
| 7925 |
+
{
|
| 7926 |
+
"epoch": 0.1907260111694026,
|
| 7927 |
+
"grad_norm": 3.3730556964874268,
|
| 7928 |
+
"learning_rate": 2.0885125305220265e-05,
|
| 7929 |
+
"loss": 2.7021,
|
| 7930 |
+
"step": 1127
|
| 7931 |
+
},
|
| 7932 |
+
{
|
| 7933 |
+
"epoch": 0.19089524454222373,
|
| 7934 |
+
"grad_norm": 3.6970086097717285,
|
| 7935 |
+
"learning_rate": 2.0749335624263997e-05,
|
| 7936 |
+
"loss": 2.3313,
|
| 7937 |
+
"step": 1128
|
| 7938 |
+
},
|
| 7939 |
+
{
|
| 7940 |
+
"epoch": 0.19106447791504486,
|
| 7941 |
+
"grad_norm": 4.548403739929199,
|
| 7942 |
+
"learning_rate": 2.061393770156088e-05,
|
| 7943 |
+
"loss": 2.9549,
|
| 7944 |
+
"step": 1129
|
| 7945 |
+
},
|
| 7946 |
+
{
|
| 7947 |
+
"epoch": 0.19123371128786595,
|
| 7948 |
+
"grad_norm": 3.9236278533935547,
|
| 7949 |
+
"learning_rate": 2.047893220642081e-05,
|
| 7950 |
+
"loss": 2.2832,
|
| 7951 |
+
"step": 1130
|
| 7952 |
+
},
|
| 7953 |
+
{
|
| 7954 |
+
"epoch": 0.19140294466068708,
|
| 7955 |
+
"grad_norm": 3.5762319564819336,
|
| 7956 |
+
"learning_rate": 2.0344319806213775e-05,
|
| 7957 |
+
"loss": 2.4004,
|
| 7958 |
+
"step": 1131
|
| 7959 |
+
},
|
| 7960 |
+
{
|
| 7961 |
+
"epoch": 0.1915721780335082,
|
| 7962 |
+
"grad_norm": 5.2948737144470215,
|
| 7963 |
+
"learning_rate": 2.0210101166366603e-05,
|
| 7964 |
+
"loss": 2.8598,
|
| 7965 |
+
"step": 1132
|
| 7966 |
+
},
|
| 7967 |
+
{
|
| 7968 |
+
"epoch": 0.19174141140632933,
|
| 7969 |
+
"grad_norm": 4.078364372253418,
|
| 7970 |
+
"learning_rate": 2.0076276950359685e-05,
|
| 7971 |
+
"loss": 2.0932,
|
| 7972 |
+
"step": 1133
|
| 7973 |
+
},
|
| 7974 |
+
{
|
| 7975 |
+
"epoch": 0.19191064477915046,
|
| 7976 |
+
"grad_norm": 3.4138169288635254,
|
| 7977 |
+
"learning_rate": 1.994284781972361e-05,
|
| 7978 |
+
"loss": 2.2644,
|
| 7979 |
+
"step": 1134
|
| 7980 |
+
},
|
| 7981 |
+
{
|
| 7982 |
+
"epoch": 0.19207987815197156,
|
| 7983 |
+
"grad_norm": 4.551220893859863,
|
| 7984 |
+
"learning_rate": 1.9809814434036e-05,
|
| 7985 |
+
"loss": 2.9021,
|
| 7986 |
+
"step": 1135
|
| 7987 |
+
},
|
| 7988 |
+
{
|
| 7989 |
+
"epoch": 0.19224911152479268,
|
| 7990 |
+
"grad_norm": 3.665961503982544,
|
| 7991 |
+
"learning_rate": 1.9677177450918205e-05,
|
| 7992 |
+
"loss": 2.451,
|
| 7993 |
+
"step": 1136
|
| 7994 |
+
},
|
| 7995 |
+
{
|
| 7996 |
+
"epoch": 0.1924183448976138,
|
| 7997 |
+
"grad_norm": 3.8647689819335938,
|
| 7998 |
+
"learning_rate": 1.954493752603199e-05,
|
| 7999 |
+
"loss": 2.4758,
|
| 8000 |
+
"step": 1137
|
| 8001 |
+
},
|
| 8002 |
+
{
|
| 8003 |
+
"epoch": 0.19258757827043493,
|
| 8004 |
+
"grad_norm": 4.684665679931641,
|
| 8005 |
+
"learning_rate": 1.9413095313076356e-05,
|
| 8006 |
+
"loss": 2.72,
|
| 8007 |
+
"step": 1138
|
| 8008 |
+
},
|
| 8009 |
+
{
|
| 8010 |
+
"epoch": 0.19275681164325606,
|
| 8011 |
+
"grad_norm": 3.8484325408935547,
|
| 8012 |
+
"learning_rate": 1.9281651463784367e-05,
|
| 8013 |
+
"loss": 2.6544,
|
| 8014 |
+
"step": 1139
|
| 8015 |
+
},
|
| 8016 |
+
{
|
| 8017 |
+
"epoch": 0.19292604501607716,
|
| 8018 |
+
"grad_norm": 4.144100189208984,
|
| 8019 |
+
"learning_rate": 1.9150606627919797e-05,
|
| 8020 |
+
"loss": 2.0265,
|
| 8021 |
+
"step": 1140
|
| 8022 |
+
},
|
| 8023 |
+
{
|
| 8024 |
+
"epoch": 0.19309527838889828,
|
| 8025 |
+
"grad_norm": 3.6583425998687744,
|
| 8026 |
+
"learning_rate": 1.9019961453274006e-05,
|
| 8027 |
+
"loss": 2.3777,
|
| 8028 |
+
"step": 1141
|
| 8029 |
+
},
|
| 8030 |
+
{
|
| 8031 |
+
"epoch": 0.1932645117617194,
|
| 8032 |
+
"grad_norm": 3.8792805671691895,
|
| 8033 |
+
"learning_rate": 1.8889716585662764e-05,
|
| 8034 |
+
"loss": 2.3128,
|
| 8035 |
+
"step": 1142
|
| 8036 |
+
},
|
| 8037 |
+
{
|
| 8038 |
+
"epoch": 0.19343374513454054,
|
| 8039 |
+
"grad_norm": 3.337419271469116,
|
| 8040 |
+
"learning_rate": 1.8759872668922938e-05,
|
| 8041 |
+
"loss": 2.0012,
|
| 8042 |
+
"step": 1143
|
| 8043 |
+
},
|
| 8044 |
+
{
|
| 8045 |
+
"epoch": 0.19360297850736166,
|
| 8046 |
+
"grad_norm": 3.3833539485931396,
|
| 8047 |
+
"learning_rate": 1.863043034490938e-05,
|
| 8048 |
+
"loss": 1.8814,
|
| 8049 |
+
"step": 1144
|
| 8050 |
+
},
|
| 8051 |
+
{
|
| 8052 |
+
"epoch": 0.19377221188018276,
|
| 8053 |
+
"grad_norm": 4.653625011444092,
|
| 8054 |
+
"learning_rate": 1.8501390253491856e-05,
|
| 8055 |
+
"loss": 2.7611,
|
| 8056 |
+
"step": 1145
|
| 8057 |
+
},
|
| 8058 |
+
{
|
| 8059 |
+
"epoch": 0.19394144525300389,
|
| 8060 |
+
"grad_norm": 4.226764678955078,
|
| 8061 |
+
"learning_rate": 1.837275303255165e-05,
|
| 8062 |
+
"loss": 2.4778,
|
| 8063 |
+
"step": 1146
|
| 8064 |
+
},
|
| 8065 |
+
{
|
| 8066 |
+
"epoch": 0.194110678625825,
|
| 8067 |
+
"grad_norm": 4.703558444976807,
|
| 8068 |
+
"learning_rate": 1.8244519317978638e-05,
|
| 8069 |
+
"loss": 2.9123,
|
| 8070 |
+
"step": 1147
|
| 8071 |
+
},
|
| 8072 |
+
{
|
| 8073 |
+
"epoch": 0.19427991199864614,
|
| 8074 |
+
"grad_norm": 3.336479425430298,
|
| 8075 |
+
"learning_rate": 1.811668974366805e-05,
|
| 8076 |
+
"loss": 2.5423,
|
| 8077 |
+
"step": 1148
|
| 8078 |
+
},
|
| 8079 |
+
{
|
| 8080 |
+
"epoch": 0.19444914537146726,
|
| 8081 |
+
"grad_norm": 4.1714582443237305,
|
| 8082 |
+
"learning_rate": 1.7989264941517293e-05,
|
| 8083 |
+
"loss": 2.6102,
|
| 8084 |
+
"step": 1149
|
| 8085 |
+
},
|
| 8086 |
+
{
|
| 8087 |
+
"epoch": 0.19461837874428836,
|
| 8088 |
+
"grad_norm": 4.257923126220703,
|
| 8089 |
+
"learning_rate": 1.786224554142285e-05,
|
| 8090 |
+
"loss": 2.8102,
|
| 8091 |
+
"step": 1150
|
| 8092 |
+
},
|
| 8093 |
+
{
|
| 8094 |
+
"epoch": 0.1947876121171095,
|
| 8095 |
+
"grad_norm": 3.027208089828491,
|
| 8096 |
+
"learning_rate": 1.7735632171277293e-05,
|
| 8097 |
+
"loss": 1.8859,
|
| 8098 |
+
"step": 1151
|
| 8099 |
+
},
|
| 8100 |
+
{
|
| 8101 |
+
"epoch": 0.1949568454899306,
|
| 8102 |
+
"grad_norm": 4.274651050567627,
|
| 8103 |
+
"learning_rate": 1.7609425456965957e-05,
|
| 8104 |
+
"loss": 2.5545,
|
| 8105 |
+
"step": 1152
|
| 8106 |
+
},
|
| 8107 |
+
{
|
| 8108 |
+
"epoch": 0.19512607886275174,
|
| 8109 |
+
"grad_norm": 4.701289176940918,
|
| 8110 |
+
"learning_rate": 1.748362602236403e-05,
|
| 8111 |
+
"loss": 2.323,
|
| 8112 |
+
"step": 1153
|
| 8113 |
+
},
|
| 8114 |
+
{
|
| 8115 |
+
"epoch": 0.19529531223557287,
|
| 8116 |
+
"grad_norm": 4.134562015533447,
|
| 8117 |
+
"learning_rate": 1.7358234489333413e-05,
|
| 8118 |
+
"loss": 2.624,
|
| 8119 |
+
"step": 1154
|
| 8120 |
+
},
|
| 8121 |
+
{
|
| 8122 |
+
"epoch": 0.19546454560839396,
|
| 8123 |
+
"grad_norm": 5.8739471435546875,
|
| 8124 |
+
"learning_rate": 1.723325147771959e-05,
|
| 8125 |
+
"loss": 2.4788,
|
| 8126 |
+
"step": 1155
|
| 8127 |
+
},
|
| 8128 |
+
{
|
| 8129 |
+
"epoch": 0.1956337789812151,
|
| 8130 |
+
"grad_norm": 4.149096965789795,
|
| 8131 |
+
"learning_rate": 1.71086776053486e-05,
|
| 8132 |
+
"loss": 2.7216,
|
| 8133 |
+
"step": 1156
|
| 8134 |
+
},
|
| 8135 |
+
{
|
| 8136 |
+
"epoch": 0.19580301235403622,
|
| 8137 |
+
"grad_norm": 3.5903432369232178,
|
| 8138 |
+
"learning_rate": 1.698451348802407e-05,
|
| 8139 |
+
"loss": 2.5326,
|
| 8140 |
+
"step": 1157
|
| 8141 |
+
},
|
| 8142 |
+
{
|
| 8143 |
+
"epoch": 0.19597224572685734,
|
| 8144 |
+
"grad_norm": 4.321657657623291,
|
| 8145 |
+
"learning_rate": 1.6860759739524003e-05,
|
| 8146 |
+
"loss": 2.7635,
|
| 8147 |
+
"step": 1158
|
| 8148 |
+
},
|
| 8149 |
+
{
|
| 8150 |
+
"epoch": 0.19614147909967847,
|
| 8151 |
+
"grad_norm": 3.3195950984954834,
|
| 8152 |
+
"learning_rate": 1.6737416971597874e-05,
|
| 8153 |
+
"loss": 2.2309,
|
| 8154 |
+
"step": 1159
|
| 8155 |
+
},
|
| 8156 |
+
{
|
| 8157 |
+
"epoch": 0.19631071247249957,
|
| 8158 |
+
"grad_norm": 4.846458435058594,
|
| 8159 |
+
"learning_rate": 1.661448579396362e-05,
|
| 8160 |
+
"loss": 3.3152,
|
| 8161 |
+
"step": 1160
|
| 8162 |
+
},
|
| 8163 |
+
{
|
| 8164 |
+
"epoch": 0.1964799458453207,
|
| 8165 |
+
"grad_norm": 4.863012313842773,
|
| 8166 |
+
"learning_rate": 1.6491966814304483e-05,
|
| 8167 |
+
"loss": 2.2622,
|
| 8168 |
+
"step": 1161
|
| 8169 |
+
},
|
| 8170 |
+
{
|
| 8171 |
+
"epoch": 0.19664917921814182,
|
| 8172 |
+
"grad_norm": 4.729052543640137,
|
| 8173 |
+
"learning_rate": 1.63698606382661e-05,
|
| 8174 |
+
"loss": 3.1973,
|
| 8175 |
+
"step": 1162
|
| 8176 |
+
},
|
| 8177 |
+
{
|
| 8178 |
+
"epoch": 0.19681841259096294,
|
| 8179 |
+
"grad_norm": 3.6677439212799072,
|
| 8180 |
+
"learning_rate": 1.624816786945358e-05,
|
| 8181 |
+
"loss": 2.6858,
|
| 8182 |
+
"step": 1163
|
| 8183 |
+
},
|
| 8184 |
+
{
|
| 8185 |
+
"epoch": 0.19698764596378407,
|
| 8186 |
+
"grad_norm": 3.435737133026123,
|
| 8187 |
+
"learning_rate": 1.6126889109428345e-05,
|
| 8188 |
+
"loss": 2.2278,
|
| 8189 |
+
"step": 1164
|
| 8190 |
+
},
|
| 8191 |
+
{
|
| 8192 |
+
"epoch": 0.19715687933660517,
|
| 8193 |
+
"grad_norm": 3.5177526473999023,
|
| 8194 |
+
"learning_rate": 1.6006024957705355e-05,
|
| 8195 |
+
"loss": 2.4997,
|
| 8196 |
+
"step": 1165
|
| 8197 |
+
},
|
| 8198 |
+
{
|
| 8199 |
+
"epoch": 0.1973261127094263,
|
| 8200 |
+
"grad_norm": 4.816482067108154,
|
| 8201 |
+
"learning_rate": 1.5885576011750004e-05,
|
| 8202 |
+
"loss": 2.7894,
|
| 8203 |
+
"step": 1166
|
| 8204 |
+
},
|
| 8205 |
+
{
|
| 8206 |
+
"epoch": 0.19749534608224742,
|
| 8207 |
+
"grad_norm": 5.397835731506348,
|
| 8208 |
+
"learning_rate": 1.576554286697518e-05,
|
| 8209 |
+
"loss": 2.768,
|
| 8210 |
+
"step": 1167
|
| 8211 |
+
},
|
| 8212 |
+
{
|
| 8213 |
+
"epoch": 0.19766457945506855,
|
| 8214 |
+
"grad_norm": 3.875643491744995,
|
| 8215 |
+
"learning_rate": 1.5645926116738352e-05,
|
| 8216 |
+
"loss": 2.1851,
|
| 8217 |
+
"step": 1168
|
| 8218 |
+
},
|
| 8219 |
+
{
|
| 8220 |
+
"epoch": 0.19783381282788967,
|
| 8221 |
+
"grad_norm": 4.401752471923828,
|
| 8222 |
+
"learning_rate": 1.5526726352338695e-05,
|
| 8223 |
+
"loss": 2.5592,
|
| 8224 |
+
"step": 1169
|
| 8225 |
+
},
|
| 8226 |
+
{
|
| 8227 |
+
"epoch": 0.19800304620071077,
|
| 8228 |
+
"grad_norm": 3.9592509269714355,
|
| 8229 |
+
"learning_rate": 1.540794416301402e-05,
|
| 8230 |
+
"loss": 2.9438,
|
| 8231 |
+
"step": 1170
|
| 8232 |
+
},
|
| 8233 |
+
{
|
| 8234 |
+
"epoch": 0.1981722795735319,
|
| 8235 |
+
"grad_norm": 4.1664557456970215,
|
| 8236 |
+
"learning_rate": 1.528958013593801e-05,
|
| 8237 |
+
"loss": 3.0191,
|
| 8238 |
+
"step": 1171
|
| 8239 |
+
},
|
| 8240 |
+
{
|
| 8241 |
+
"epoch": 0.19834151294635302,
|
| 8242 |
+
"grad_norm": 3.8315508365631104,
|
| 8243 |
+
"learning_rate": 1.5171634856217265e-05,
|
| 8244 |
+
"loss": 2.4902,
|
| 8245 |
+
"step": 1172
|
| 8246 |
+
},
|
| 8247 |
+
{
|
| 8248 |
+
"epoch": 0.19851074631917415,
|
| 8249 |
+
"grad_norm": 4.379573822021484,
|
| 8250 |
+
"learning_rate": 1.5054108906888343e-05,
|
| 8251 |
+
"loss": 2.555,
|
| 8252 |
+
"step": 1173
|
| 8253 |
+
},
|
| 8254 |
+
{
|
| 8255 |
+
"epoch": 0.19867997969199527,
|
| 8256 |
+
"grad_norm": 4.705971717834473,
|
| 8257 |
+
"learning_rate": 1.493700286891494e-05,
|
| 8258 |
+
"loss": 3.048,
|
| 8259 |
+
"step": 1174
|
| 8260 |
+
},
|
| 8261 |
+
{
|
| 8262 |
+
"epoch": 0.19884921306481637,
|
| 8263 |
+
"grad_norm": 3.3631129264831543,
|
| 8264 |
+
"learning_rate": 1.4820317321185074e-05,
|
| 8265 |
+
"loss": 2.6068,
|
| 8266 |
+
"step": 1175
|
| 8267 |
+
},
|
| 8268 |
+
{
|
| 8269 |
+
"epoch": 0.1990184464376375,
|
| 8270 |
+
"grad_norm": 3.395667314529419,
|
| 8271 |
+
"learning_rate": 1.4704052840508087e-05,
|
| 8272 |
+
"loss": 1.7894,
|
| 8273 |
+
"step": 1176
|
| 8274 |
+
},
|
| 8275 |
+
{
|
| 8276 |
+
"epoch": 0.19918767981045862,
|
| 8277 |
+
"grad_norm": 5.260849952697754,
|
| 8278 |
+
"learning_rate": 1.4588210001611923e-05,
|
| 8279 |
+
"loss": 3.207,
|
| 8280 |
+
"step": 1177
|
| 8281 |
+
},
|
| 8282 |
+
{
|
| 8283 |
+
"epoch": 0.19935691318327975,
|
| 8284 |
+
"grad_norm": 4.87640380859375,
|
| 8285 |
+
"learning_rate": 1.4472789377140239e-05,
|
| 8286 |
+
"loss": 2.7079,
|
| 8287 |
+
"step": 1178
|
| 8288 |
+
},
|
| 8289 |
+
{
|
| 8290 |
+
"epoch": 0.19952614655610088,
|
| 8291 |
+
"grad_norm": 4.144646644592285,
|
| 8292 |
+
"learning_rate": 1.4357791537649512e-05,
|
| 8293 |
+
"loss": 2.8642,
|
| 8294 |
+
"step": 1179
|
| 8295 |
+
},
|
| 8296 |
+
{
|
| 8297 |
+
"epoch": 0.19969537992892197,
|
| 8298 |
+
"grad_norm": 4.221438407897949,
|
| 8299 |
+
"learning_rate": 1.4243217051606283e-05,
|
| 8300 |
+
"loss": 2.4766,
|
| 8301 |
+
"step": 1180
|
| 8302 |
+
},
|
| 8303 |
+
{
|
| 8304 |
+
"epoch": 0.1998646133017431,
|
| 8305 |
+
"grad_norm": 4.545912265777588,
|
| 8306 |
+
"learning_rate": 1.4129066485384413e-05,
|
| 8307 |
+
"loss": 2.8973,
|
| 8308 |
+
"step": 1181
|
| 8309 |
+
},
|
| 8310 |
+
{
|
| 8311 |
+
"epoch": 0.20003384667456423,
|
| 8312 |
+
"grad_norm": 3.545257806777954,
|
| 8313 |
+
"learning_rate": 1.401534040326209e-05,
|
| 8314 |
+
"loss": 1.86,
|
| 8315 |
+
"step": 1182
|
| 8316 |
+
},
|
| 8317 |
+
{
|
| 8318 |
+
"epoch": 0.20020308004738535,
|
| 8319 |
+
"grad_norm": 3.8931775093078613,
|
| 8320 |
+
"learning_rate": 1.3902039367419262e-05,
|
| 8321 |
+
"loss": 2.3143,
|
| 8322 |
+
"step": 1183
|
| 8323 |
+
},
|
| 8324 |
+
{
|
| 8325 |
+
"epoch": 0.20037231342020648,
|
| 8326 |
+
"grad_norm": 3.9221863746643066,
|
| 8327 |
+
"learning_rate": 1.3789163937934735e-05,
|
| 8328 |
+
"loss": 2.5844,
|
| 8329 |
+
"step": 1184
|
| 8330 |
+
},
|
| 8331 |
+
{
|
| 8332 |
+
"epoch": 0.20054154679302758,
|
| 8333 |
+
"grad_norm": 3.8118457794189453,
|
| 8334 |
+
"learning_rate": 1.3676714672783375e-05,
|
| 8335 |
+
"loss": 2.4459,
|
| 8336 |
+
"step": 1185
|
| 8337 |
+
},
|
| 8338 |
+
{
|
| 8339 |
+
"epoch": 0.2007107801658487,
|
| 8340 |
+
"grad_norm": 4.256392478942871,
|
| 8341 |
+
"learning_rate": 1.35646921278334e-05,
|
| 8342 |
+
"loss": 2.8082,
|
| 8343 |
+
"step": 1186
|
| 8344 |
+
},
|
| 8345 |
+
{
|
| 8346 |
+
"epoch": 0.20088001353866983,
|
| 8347 |
+
"grad_norm": 3.8734917640686035,
|
| 8348 |
+
"learning_rate": 1.3453096856843717e-05,
|
| 8349 |
+
"loss": 2.4939,
|
| 8350 |
+
"step": 1187
|
| 8351 |
+
},
|
| 8352 |
+
{
|
| 8353 |
+
"epoch": 0.20104924691149095,
|
| 8354 |
+
"grad_norm": 4.234657287597656,
|
| 8355 |
+
"learning_rate": 1.3341929411460985e-05,
|
| 8356 |
+
"loss": 2.4565,
|
| 8357 |
+
"step": 1188
|
| 8358 |
+
},
|
| 8359 |
+
{
|
| 8360 |
+
"epoch": 0.20121848028431208,
|
| 8361 |
+
"grad_norm": 5.375681400299072,
|
| 8362 |
+
"learning_rate": 1.3231190341217081e-05,
|
| 8363 |
+
"loss": 2.6613,
|
| 8364 |
+
"step": 1189
|
| 8365 |
+
},
|
| 8366 |
+
{
|
| 8367 |
+
"epoch": 0.20138771365713318,
|
| 8368 |
+
"grad_norm": 3.2132842540740967,
|
| 8369 |
+
"learning_rate": 1.3120880193526297e-05,
|
| 8370 |
+
"loss": 2.4232,
|
| 8371 |
+
"step": 1190
|
| 8372 |
+
},
|
| 8373 |
+
{
|
| 8374 |
+
"epoch": 0.2015569470299543,
|
| 8375 |
+
"grad_norm": 4.174328327178955,
|
| 8376 |
+
"learning_rate": 1.3010999513682587e-05,
|
| 8377 |
+
"loss": 2.9844,
|
| 8378 |
+
"step": 1191
|
| 8379 |
+
},
|
| 8380 |
+
{
|
| 8381 |
+
"epoch": 0.20172618040277543,
|
| 8382 |
+
"grad_norm": 4.462172985076904,
|
| 8383 |
+
"learning_rate": 1.2901548844856948e-05,
|
| 8384 |
+
"loss": 2.5642,
|
| 8385 |
+
"step": 1192
|
| 8386 |
+
},
|
| 8387 |
+
{
|
| 8388 |
+
"epoch": 0.20189541377559655,
|
| 8389 |
+
"grad_norm": 3.740525245666504,
|
| 8390 |
+
"learning_rate": 1.2792528728094754e-05,
|
| 8391 |
+
"loss": 1.8737,
|
| 8392 |
+
"step": 1193
|
| 8393 |
+
},
|
| 8394 |
+
{
|
| 8395 |
+
"epoch": 0.20206464714841768,
|
| 8396 |
+
"grad_norm": 4.4204511642456055,
|
| 8397 |
+
"learning_rate": 1.2683939702312986e-05,
|
| 8398 |
+
"loss": 2.8763,
|
| 8399 |
+
"step": 1194
|
| 8400 |
+
},
|
| 8401 |
+
{
|
| 8402 |
+
"epoch": 0.20223388052123878,
|
| 8403 |
+
"grad_norm": 3.847598075866699,
|
| 8404 |
+
"learning_rate": 1.2575782304297645e-05,
|
| 8405 |
+
"loss": 2.5697,
|
| 8406 |
+
"step": 1195
|
| 8407 |
+
},
|
| 8408 |
+
{
|
| 8409 |
+
"epoch": 0.2024031138940599,
|
| 8410 |
+
"grad_norm": 4.267530918121338,
|
| 8411 |
+
"learning_rate": 1.24680570687011e-05,
|
| 8412 |
+
"loss": 2.6124,
|
| 8413 |
+
"step": 1196
|
| 8414 |
+
},
|
| 8415 |
+
{
|
| 8416 |
+
"epoch": 0.20257234726688103,
|
| 8417 |
+
"grad_norm": 4.407855033874512,
|
| 8418 |
+
"learning_rate": 1.236076452803937e-05,
|
| 8419 |
+
"loss": 2.6668,
|
| 8420 |
+
"step": 1197
|
| 8421 |
+
},
|
| 8422 |
+
{
|
| 8423 |
+
"epoch": 0.20274158063970216,
|
| 8424 |
+
"grad_norm": 3.9101080894470215,
|
| 8425 |
+
"learning_rate": 1.2253905212689553e-05,
|
| 8426 |
+
"loss": 2.2668,
|
| 8427 |
+
"step": 1198
|
| 8428 |
+
},
|
| 8429 |
+
{
|
| 8430 |
+
"epoch": 0.20291081401252328,
|
| 8431 |
+
"grad_norm": 3.9558191299438477,
|
| 8432 |
+
"learning_rate": 1.2147479650887228e-05,
|
| 8433 |
+
"loss": 2.1533,
|
| 8434 |
+
"step": 1199
|
| 8435 |
+
},
|
| 8436 |
+
{
|
| 8437 |
+
"epoch": 0.20308004738534438,
|
| 8438 |
+
"grad_norm": 4.162629127502441,
|
| 8439 |
+
"learning_rate": 1.2041488368723764e-05,
|
| 8440 |
+
"loss": 2.6475,
|
| 8441 |
+
"step": 1200
|
| 8442 |
+
},
|
| 8443 |
+
{
|
| 8444 |
+
"epoch": 0.2032492807581655,
|
| 8445 |
+
"grad_norm": 4.642498970031738,
|
| 8446 |
+
"learning_rate": 1.1935931890143792e-05,
|
| 8447 |
+
"loss": 2.3502,
|
| 8448 |
+
"step": 1201
|
| 8449 |
+
},
|
| 8450 |
+
{
|
| 8451 |
+
"epoch": 0.20341851413098663,
|
| 8452 |
+
"grad_norm": 3.8693747520446777,
|
| 8453 |
+
"learning_rate": 1.183081073694261e-05,
|
| 8454 |
+
"loss": 2.6568,
|
| 8455 |
+
"step": 1202
|
| 8456 |
+
},
|
| 8457 |
+
{
|
| 8458 |
+
"epoch": 0.20358774750380776,
|
| 8459 |
+
"grad_norm": 4.497195243835449,
|
| 8460 |
+
"learning_rate": 1.1726125428763524e-05,
|
| 8461 |
+
"loss": 2.3555,
|
| 8462 |
+
"step": 1203
|
| 8463 |
+
},
|
| 8464 |
+
{
|
| 8465 |
+
"epoch": 0.20375698087662888,
|
| 8466 |
+
"grad_norm": 4.149742126464844,
|
| 8467 |
+
"learning_rate": 1.162187648309534e-05,
|
| 8468 |
+
"loss": 2.548,
|
| 8469 |
+
"step": 1204
|
| 8470 |
+
},
|
| 8471 |
+
{
|
| 8472 |
+
"epoch": 0.20392621424944998,
|
| 8473 |
+
"grad_norm": 3.7552905082702637,
|
| 8474 |
+
"learning_rate": 1.151806441526987e-05,
|
| 8475 |
+
"loss": 2.4119,
|
| 8476 |
+
"step": 1205
|
| 8477 |
+
},
|
| 8478 |
+
{
|
| 8479 |
+
"epoch": 0.2040954476222711,
|
| 8480 |
+
"grad_norm": 4.621352672576904,
|
| 8481 |
+
"learning_rate": 1.1414689738459228e-05,
|
| 8482 |
+
"loss": 2.7965,
|
| 8483 |
+
"step": 1206
|
| 8484 |
+
},
|
| 8485 |
+
{
|
| 8486 |
+
"epoch": 0.20426468099509223,
|
| 8487 |
+
"grad_norm": 4.034228801727295,
|
| 8488 |
+
"learning_rate": 1.1311752963673439e-05,
|
| 8489 |
+
"loss": 2.3959,
|
| 8490 |
+
"step": 1207
|
| 8491 |
+
},
|
| 8492 |
+
{
|
| 8493 |
+
"epoch": 0.20443391436791336,
|
| 8494 |
+
"grad_norm": 3.716707468032837,
|
| 8495 |
+
"learning_rate": 1.1209254599757867e-05,
|
| 8496 |
+
"loss": 2.3916,
|
| 8497 |
+
"step": 1208
|
| 8498 |
+
},
|
| 8499 |
+
{
|
| 8500 |
+
"epoch": 0.20460314774073446,
|
| 8501 |
+
"grad_norm": 4.484525680541992,
|
| 8502 |
+
"learning_rate": 1.1107195153390649e-05,
|
| 8503 |
+
"loss": 2.0843,
|
| 8504 |
+
"step": 1209
|
| 8505 |
+
},
|
| 8506 |
+
{
|
| 8507 |
+
"epoch": 0.20477238111355558,
|
| 8508 |
+
"grad_norm": 4.925705909729004,
|
| 8509 |
+
"learning_rate": 1.1005575129080203e-05,
|
| 8510 |
+
"loss": 2.7065,
|
| 8511 |
+
"step": 1210
|
| 8512 |
+
},
|
| 8513 |
+
{
|
| 8514 |
+
"epoch": 0.2049416144863767,
|
| 8515 |
+
"grad_norm": 3.711040735244751,
|
| 8516 |
+
"learning_rate": 1.0904395029162861e-05,
|
| 8517 |
+
"loss": 2.2043,
|
| 8518 |
+
"step": 1211
|
| 8519 |
+
},
|
| 8520 |
+
{
|
| 8521 |
+
"epoch": 0.20511084785919784,
|
| 8522 |
+
"grad_norm": 4.309438228607178,
|
| 8523 |
+
"learning_rate": 1.0803655353800168e-05,
|
| 8524 |
+
"loss": 2.8827,
|
| 8525 |
+
"step": 1212
|
| 8526 |
+
},
|
| 8527 |
+
{
|
| 8528 |
+
"epoch": 0.20528008123201896,
|
| 8529 |
+
"grad_norm": 4.561186790466309,
|
| 8530 |
+
"learning_rate": 1.0703356600976611e-05,
|
| 8531 |
+
"loss": 2.9096,
|
| 8532 |
+
"step": 1213
|
| 8533 |
+
},
|
| 8534 |
+
{
|
| 8535 |
+
"epoch": 0.20544931460484006,
|
| 8536 |
+
"grad_norm": 3.535050868988037,
|
| 8537 |
+
"learning_rate": 1.0603499266497063e-05,
|
| 8538 |
+
"loss": 2.0575,
|
| 8539 |
+
"step": 1214
|
| 8540 |
+
},
|
| 8541 |
+
{
|
| 8542 |
+
"epoch": 0.2056185479776612,
|
| 8543 |
+
"grad_norm": 4.217011451721191,
|
| 8544 |
+
"learning_rate": 1.0504083843984303e-05,
|
| 8545 |
+
"loss": 1.8126,
|
| 8546 |
+
"step": 1215
|
| 8547 |
+
},
|
| 8548 |
+
{
|
| 8549 |
+
"epoch": 0.2057877813504823,
|
| 8550 |
+
"grad_norm": 3.88653564453125,
|
| 8551 |
+
"learning_rate": 1.0405110824876619e-05,
|
| 8552 |
+
"loss": 2.4523,
|
| 8553 |
+
"step": 1216
|
| 8554 |
+
},
|
| 8555 |
+
{
|
| 8556 |
+
"epoch": 0.20595701472330344,
|
| 8557 |
+
"grad_norm": 3.8444836139678955,
|
| 8558 |
+
"learning_rate": 1.0306580698425439e-05,
|
| 8559 |
+
"loss": 2.4574,
|
| 8560 |
+
"step": 1217
|
| 8561 |
+
},
|
| 8562 |
+
{
|
| 8563 |
+
"epoch": 0.20612624809612456,
|
| 8564 |
+
"grad_norm": 4.170924663543701,
|
| 8565 |
+
"learning_rate": 1.0208493951692755e-05,
|
| 8566 |
+
"loss": 3.1727,
|
| 8567 |
+
"step": 1218
|
| 8568 |
+
},
|
| 8569 |
+
{
|
| 8570 |
+
"epoch": 0.20629548146894566,
|
| 8571 |
+
"grad_norm": 3.7632195949554443,
|
| 8572 |
+
"learning_rate": 1.0110851069548889e-05,
|
| 8573 |
+
"loss": 2.4326,
|
| 8574 |
+
"step": 1219
|
| 8575 |
+
},
|
| 8576 |
+
{
|
| 8577 |
+
"epoch": 0.2064647148417668,
|
| 8578 |
+
"grad_norm": 4.985208511352539,
|
| 8579 |
+
"learning_rate": 1.0013652534669982e-05,
|
| 8580 |
+
"loss": 2.8027,
|
| 8581 |
+
"step": 1220
|
| 8582 |
+
},
|
| 8583 |
+
{
|
| 8584 |
+
"epoch": 0.20663394821458791,
|
| 8585 |
+
"grad_norm": 4.505946636199951,
|
| 8586 |
+
"learning_rate": 9.916898827535625e-06,
|
| 8587 |
+
"loss": 2.2044,
|
| 8588 |
+
"step": 1221
|
| 8589 |
+
},
|
| 8590 |
+
{
|
| 8591 |
+
"epoch": 0.20680318158740904,
|
| 8592 |
+
"grad_norm": 4.337455749511719,
|
| 8593 |
+
"learning_rate": 9.820590426426523e-06,
|
| 8594 |
+
"loss": 2.3544,
|
| 8595 |
+
"step": 1222
|
| 8596 |
+
},
|
| 8597 |
+
{
|
| 8598 |
+
"epoch": 0.20697241496023017,
|
| 8599 |
+
"grad_norm": 5.021992206573486,
|
| 8600 |
+
"learning_rate": 9.724727807422118e-06,
|
| 8601 |
+
"loss": 3.0644,
|
| 8602 |
+
"step": 1223
|
| 8603 |
+
},
|
| 8604 |
+
{
|
| 8605 |
+
"epoch": 0.20714164833305126,
|
| 8606 |
+
"grad_norm": 4.401887893676758,
|
| 8607 |
+
"learning_rate": 9.629311444398192e-06,
|
| 8608 |
+
"loss": 2.3384,
|
| 8609 |
+
"step": 1224
|
| 8610 |
+
},
|
| 8611 |
+
{
|
| 8612 |
+
"epoch": 0.2073108817058724,
|
| 8613 |
+
"grad_norm": 4.148678779602051,
|
| 8614 |
+
"learning_rate": 9.534341809024583e-06,
|
| 8615 |
+
"loss": 2.6118,
|
| 8616 |
+
"step": 1225
|
| 8617 |
+
},
|
| 8618 |
+
{
|
| 8619 |
+
"epoch": 0.20748011507869352,
|
| 8620 |
+
"grad_norm": 3.7790815830230713,
|
| 8621 |
+
"learning_rate": 9.439819370762848e-06,
|
| 8622 |
+
"loss": 2.5443,
|
| 8623 |
+
"step": 1226
|
| 8624 |
+
},
|
| 8625 |
+
{
|
| 8626 |
+
"epoch": 0.20764934845151464,
|
| 8627 |
+
"grad_norm": 4.02851676940918,
|
| 8628 |
+
"learning_rate": 9.345744596863892e-06,
|
| 8629 |
+
"loss": 2.8159,
|
| 8630 |
+
"step": 1227
|
| 8631 |
+
},
|
| 8632 |
+
{
|
| 8633 |
+
"epoch": 0.20781858182433577,
|
| 8634 |
+
"grad_norm": 4.618631362915039,
|
| 8635 |
+
"learning_rate": 9.252117952365669e-06,
|
| 8636 |
+
"loss": 2.6413,
|
| 8637 |
+
"step": 1228
|
| 8638 |
+
},
|
| 8639 |
+
{
|
| 8640 |
+
"epoch": 0.20798781519715687,
|
| 8641 |
+
"grad_norm": 4.675975322723389,
|
| 8642 |
+
"learning_rate": 9.158939900090969e-06,
|
| 8643 |
+
"loss": 2.6299,
|
| 8644 |
+
"step": 1229
|
| 8645 |
+
},
|
| 8646 |
+
{
|
| 8647 |
+
"epoch": 0.208157048569978,
|
| 8648 |
+
"grad_norm": 4.649596691131592,
|
| 8649 |
+
"learning_rate": 9.066210900644988e-06,
|
| 8650 |
+
"loss": 2.5428,
|
| 8651 |
+
"step": 1230
|
| 8652 |
+
},
|
| 8653 |
+
{
|
| 8654 |
+
"epoch": 0.20832628194279912,
|
| 8655 |
+
"grad_norm": 4.847729206085205,
|
| 8656 |
+
"learning_rate": 8.973931412413195e-06,
|
| 8657 |
+
"loss": 2.7639,
|
| 8658 |
+
"step": 1231
|
| 8659 |
+
},
|
| 8660 |
+
{
|
| 8661 |
+
"epoch": 0.20849551531562024,
|
| 8662 |
+
"grad_norm": 4.831533908843994,
|
| 8663 |
+
"learning_rate": 8.882101891558991e-06,
|
| 8664 |
+
"loss": 2.3354,
|
| 8665 |
+
"step": 1232
|
| 8666 |
+
},
|
| 8667 |
+
{
|
| 8668 |
+
"epoch": 0.20866474868844137,
|
| 8669 |
+
"grad_norm": 4.117055416107178,
|
| 8670 |
+
"learning_rate": 8.790722792021422e-06,
|
| 8671 |
+
"loss": 2.3151,
|
| 8672 |
+
"step": 1233
|
| 8673 |
+
},
|
| 8674 |
+
{
|
| 8675 |
+
"epoch": 0.20883398206126247,
|
| 8676 |
+
"grad_norm": 4.020594596862793,
|
| 8677 |
+
"learning_rate": 8.699794565512975e-06,
|
| 8678 |
+
"loss": 2.1672,
|
| 8679 |
+
"step": 1234
|
| 8680 |
+
},
|
| 8681 |
+
{
|
| 8682 |
+
"epoch": 0.2090032154340836,
|
| 8683 |
+
"grad_norm": 3.865863084793091,
|
| 8684 |
+
"learning_rate": 8.609317661517413e-06,
|
| 8685 |
+
"loss": 2.4174,
|
| 8686 |
+
"step": 1235
|
| 8687 |
+
},
|
| 8688 |
+
{
|
| 8689 |
+
"epoch": 0.20917244880690472,
|
| 8690 |
+
"grad_norm": 3.537734031677246,
|
| 8691 |
+
"learning_rate": 8.519292527287393e-06,
|
| 8692 |
+
"loss": 2.3311,
|
| 8693 |
+
"step": 1236
|
| 8694 |
+
},
|
| 8695 |
+
{
|
| 8696 |
+
"epoch": 0.20934168217972585,
|
| 8697 |
+
"grad_norm": 4.704585552215576,
|
| 8698 |
+
"learning_rate": 8.429719607842412e-06,
|
| 8699 |
+
"loss": 2.6749,
|
| 8700 |
+
"step": 1237
|
| 8701 |
+
},
|
| 8702 |
+
{
|
| 8703 |
+
"epoch": 0.20951091555254697,
|
| 8704 |
+
"grad_norm": 4.5119218826293945,
|
| 8705 |
+
"learning_rate": 8.340599345966538e-06,
|
| 8706 |
+
"loss": 2.5822,
|
| 8707 |
+
"step": 1238
|
| 8708 |
+
},
|
| 8709 |
+
{
|
| 8710 |
+
"epoch": 0.20968014892536807,
|
| 8711 |
+
"grad_norm": 3.8941285610198975,
|
| 8712 |
+
"learning_rate": 8.251932182206212e-06,
|
| 8713 |
+
"loss": 2.6532,
|
| 8714 |
+
"step": 1239
|
| 8715 |
+
},
|
| 8716 |
+
{
|
| 8717 |
+
"epoch": 0.2098493822981892,
|
| 8718 |
+
"grad_norm": 2.9833123683929443,
|
| 8719 |
+
"learning_rate": 8.16371855486805e-06,
|
| 8720 |
+
"loss": 2.191,
|
| 8721 |
+
"step": 1240
|
| 8722 |
+
},
|
| 8723 |
+
{
|
| 8724 |
+
"epoch": 0.21001861567101032,
|
| 8725 |
+
"grad_norm": 4.344390392303467,
|
| 8726 |
+
"learning_rate": 8.075958900016778e-06,
|
| 8727 |
+
"loss": 2.3154,
|
| 8728 |
+
"step": 1241
|
| 8729 |
+
},
|
| 8730 |
+
{
|
| 8731 |
+
"epoch": 0.21018784904383145,
|
| 8732 |
+
"grad_norm": 3.952460765838623,
|
| 8733 |
+
"learning_rate": 7.988653651472955e-06,
|
| 8734 |
+
"loss": 2.3179,
|
| 8735 |
+
"step": 1242
|
| 8736 |
+
},
|
| 8737 |
+
{
|
| 8738 |
+
"epoch": 0.21035708241665257,
|
| 8739 |
+
"grad_norm": 4.589304447174072,
|
| 8740 |
+
"learning_rate": 7.9018032408109e-06,
|
| 8741 |
+
"loss": 2.4392,
|
| 8742 |
+
"step": 1243
|
| 8743 |
+
},
|
| 8744 |
+
{
|
| 8745 |
+
"epoch": 0.21052631578947367,
|
| 8746 |
+
"grad_norm": 5.118439197540283,
|
| 8747 |
+
"learning_rate": 7.81540809735658e-06,
|
| 8748 |
+
"loss": 2.4878,
|
| 8749 |
+
"step": 1244
|
| 8750 |
+
},
|
| 8751 |
+
{
|
| 8752 |
+
"epoch": 0.2106955491622948,
|
| 8753 |
+
"grad_norm": 4.558487415313721,
|
| 8754 |
+
"learning_rate": 7.729468648185378e-06,
|
| 8755 |
+
"loss": 2.3178,
|
| 8756 |
+
"step": 1245
|
| 8757 |
+
},
|
| 8758 |
+
{
|
| 8759 |
+
"epoch": 0.21086478253511592,
|
| 8760 |
+
"grad_norm": 4.89497184753418,
|
| 8761 |
+
"learning_rate": 7.643985318120073e-06,
|
| 8762 |
+
"loss": 2.4273,
|
| 8763 |
+
"step": 1246
|
| 8764 |
+
},
|
| 8765 |
+
{
|
| 8766 |
+
"epoch": 0.21103401590793705,
|
| 8767 |
+
"grad_norm": 4.589718818664551,
|
| 8768 |
+
"learning_rate": 7.558958529728766e-06,
|
| 8769 |
+
"loss": 2.8548,
|
| 8770 |
+
"step": 1247
|
| 8771 |
+
},
|
| 8772 |
+
{
|
| 8773 |
+
"epoch": 0.21120324928075818,
|
| 8774 |
+
"grad_norm": 4.338314533233643,
|
| 8775 |
+
"learning_rate": 7.4743887033226765e-06,
|
| 8776 |
+
"loss": 2.4868,
|
| 8777 |
+
"step": 1248
|
| 8778 |
+
},
|
| 8779 |
+
{
|
| 8780 |
+
"epoch": 0.21137248265357927,
|
| 8781 |
+
"grad_norm": 4.816850185394287,
|
| 8782 |
+
"learning_rate": 7.390276256954187e-06,
|
| 8783 |
+
"loss": 3.0987,
|
| 8784 |
+
"step": 1249
|
| 8785 |
+
},
|
| 8786 |
+
{
|
| 8787 |
+
"epoch": 0.2115417160264004,
|
| 8788 |
+
"grad_norm": 4.53594970703125,
|
| 8789 |
+
"learning_rate": 7.306621606414721e-06,
|
| 8790 |
+
"loss": 2.4571,
|
| 8791 |
+
"step": 1250
|
| 8792 |
+
},
|
| 8793 |
+
{
|
| 8794 |
+
"epoch": 0.21171094939922153,
|
| 8795 |
+
"grad_norm": 4.075862407684326,
|
| 8796 |
+
"learning_rate": 7.223425165232644e-06,
|
| 8797 |
+
"loss": 3.1163,
|
| 8798 |
+
"step": 1251
|
| 8799 |
+
},
|
| 8800 |
+
{
|
| 8801 |
+
"epoch": 0.21188018277204265,
|
| 8802 |
+
"grad_norm": 4.89505672454834,
|
| 8803 |
+
"learning_rate": 7.140687344671282e-06,
|
| 8804 |
+
"loss": 3.2721,
|
| 8805 |
+
"step": 1252
|
| 8806 |
+
},
|
| 8807 |
+
{
|
| 8808 |
+
"epoch": 0.21204941614486378,
|
| 8809 |
+
"grad_norm": 3.91040301322937,
|
| 8810 |
+
"learning_rate": 7.0584085537268805e-06,
|
| 8811 |
+
"loss": 2.5361,
|
| 8812 |
+
"step": 1253
|
| 8813 |
+
},
|
| 8814 |
+
{
|
| 8815 |
+
"epoch": 0.21221864951768488,
|
| 8816 |
+
"grad_norm": 4.2942681312561035,
|
| 8817 |
+
"learning_rate": 6.976589199126615e-06,
|
| 8818 |
+
"loss": 2.8234,
|
| 8819 |
+
"step": 1254
|
| 8820 |
+
},
|
| 8821 |
+
{
|
| 8822 |
+
"epoch": 0.212387882890506,
|
| 8823 |
+
"grad_norm": 4.04157829284668,
|
| 8824 |
+
"learning_rate": 6.895229685326443e-06,
|
| 8825 |
+
"loss": 2.4417,
|
| 8826 |
+
"step": 1255
|
| 8827 |
+
},
|
| 8828 |
+
{
|
| 8829 |
+
"epoch": 0.21255711626332713,
|
| 8830 |
+
"grad_norm": 4.016750335693359,
|
| 8831 |
+
"learning_rate": 6.814330414509284e-06,
|
| 8832 |
+
"loss": 2.2354,
|
| 8833 |
+
"step": 1256
|
| 8834 |
+
},
|
| 8835 |
+
{
|
| 8836 |
+
"epoch": 0.21272634963614825,
|
| 8837 |
+
"grad_norm": 4.151035785675049,
|
| 8838 |
+
"learning_rate": 6.733891786582902e-06,
|
| 8839 |
+
"loss": 2.7173,
|
| 8840 |
+
"step": 1257
|
| 8841 |
+
},
|
| 8842 |
+
{
|
| 8843 |
+
"epoch": 0.21289558300896938,
|
| 8844 |
+
"grad_norm": 3.6798808574676514,
|
| 8845 |
+
"learning_rate": 6.653914199177969e-06,
|
| 8846 |
+
"loss": 2.4609,
|
| 8847 |
+
"step": 1258
|
| 8848 |
+
},
|
| 8849 |
+
{
|
| 8850 |
+
"epoch": 0.21306481638179048,
|
| 8851 |
+
"grad_norm": 4.583859920501709,
|
| 8852 |
+
"learning_rate": 6.574398047646147e-06,
|
| 8853 |
+
"loss": 2.5441,
|
| 8854 |
+
"step": 1259
|
| 8855 |
+
},
|
| 8856 |
+
{
|
| 8857 |
+
"epoch": 0.2132340497546116,
|
| 8858 |
+
"grad_norm": 3.824087619781494,
|
| 8859 |
+
"learning_rate": 6.495343725058067e-06,
|
| 8860 |
+
"loss": 3.1195,
|
| 8861 |
+
"step": 1260
|
| 8862 |
+
},
|
| 8863 |
+
{
|
| 8864 |
+
"epoch": 0.21340328312743273,
|
| 8865 |
+
"grad_norm": 4.636285305023193,
|
| 8866 |
+
"learning_rate": 6.416751622201389e-06,
|
| 8867 |
+
"loss": 2.7237,
|
| 8868 |
+
"step": 1261
|
| 8869 |
+
},
|
| 8870 |
+
{
|
| 8871 |
+
"epoch": 0.21357251650025386,
|
| 8872 |
+
"grad_norm": 3.757760763168335,
|
| 8873 |
+
"learning_rate": 6.338622127578941e-06,
|
| 8874 |
+
"loss": 2.2727,
|
| 8875 |
+
"step": 1262
|
| 8876 |
+
},
|
| 8877 |
+
{
|
| 8878 |
+
"epoch": 0.21374174987307498,
|
| 8879 |
+
"grad_norm": 4.643979072570801,
|
| 8880 |
+
"learning_rate": 6.260955627406706e-06,
|
| 8881 |
+
"loss": 2.9796,
|
| 8882 |
+
"step": 1263
|
| 8883 |
+
},
|
| 8884 |
+
{
|
| 8885 |
+
"epoch": 0.21391098324589608,
|
| 8886 |
+
"grad_norm": 3.206894874572754,
|
| 8887 |
+
"learning_rate": 6.183752505611962e-06,
|
| 8888 |
+
"loss": 1.7781,
|
| 8889 |
+
"step": 1264
|
| 8890 |
+
},
|
| 8891 |
+
{
|
| 8892 |
+
"epoch": 0.2140802166187172,
|
| 8893 |
+
"grad_norm": 3.4332616329193115,
|
| 8894 |
+
"learning_rate": 6.107013143831375e-06,
|
| 8895 |
+
"loss": 2.3096,
|
| 8896 |
+
"step": 1265
|
| 8897 |
+
},
|
| 8898 |
+
{
|
| 8899 |
+
"epoch": 0.21424944999153833,
|
| 8900 |
+
"grad_norm": 4.015029430389404,
|
| 8901 |
+
"learning_rate": 6.030737921409169e-06,
|
| 8902 |
+
"loss": 2.8321,
|
| 8903 |
+
"step": 1266
|
| 8904 |
+
},
|
| 8905 |
+
{
|
| 8906 |
+
"epoch": 0.21441868336435946,
|
| 8907 |
+
"grad_norm": 4.371044158935547,
|
| 8908 |
+
"learning_rate": 5.954927215395101e-06,
|
| 8909 |
+
"loss": 2.8319,
|
| 8910 |
+
"step": 1267
|
| 8911 |
+
},
|
| 8912 |
+
{
|
| 8913 |
+
"epoch": 0.21458791673718058,
|
| 8914 |
+
"grad_norm": 4.007357597351074,
|
| 8915 |
+
"learning_rate": 5.879581400542789e-06,
|
| 8916 |
+
"loss": 2.3559,
|
| 8917 |
+
"step": 1268
|
| 8918 |
+
},
|
| 8919 |
+
{
|
| 8920 |
+
"epoch": 0.21475715011000168,
|
| 8921 |
+
"grad_norm": 4.446166515350342,
|
| 8922 |
+
"learning_rate": 5.804700849307698e-06,
|
| 8923 |
+
"loss": 2.2894,
|
| 8924 |
+
"step": 1269
|
| 8925 |
+
},
|
| 8926 |
+
{
|
| 8927 |
+
"epoch": 0.2149263834828228,
|
| 8928 |
+
"grad_norm": 3.9980952739715576,
|
| 8929 |
+
"learning_rate": 5.73028593184538e-06,
|
| 8930 |
+
"loss": 2.5826,
|
| 8931 |
+
"step": 1270
|
| 8932 |
+
},
|
| 8933 |
+
{
|
| 8934 |
+
"epoch": 0.21509561685564393,
|
| 8935 |
+
"grad_norm": 3.8507676124572754,
|
| 8936 |
+
"learning_rate": 5.656337016009649e-06,
|
| 8937 |
+
"loss": 2.546,
|
| 8938 |
+
"step": 1271
|
| 8939 |
+
},
|
| 8940 |
+
{
|
| 8941 |
+
"epoch": 0.21526485022846506,
|
| 8942 |
+
"grad_norm": 3.3553953170776367,
|
| 8943 |
+
"learning_rate": 5.582854467350729e-06,
|
| 8944 |
+
"loss": 1.798,
|
| 8945 |
+
"step": 1272
|
| 8946 |
+
},
|
| 8947 |
+
{
|
| 8948 |
+
"epoch": 0.21543408360128619,
|
| 8949 |
+
"grad_norm": 4.618180751800537,
|
| 8950 |
+
"learning_rate": 5.5098386491134415e-06,
|
| 8951 |
+
"loss": 2.3345,
|
| 8952 |
+
"step": 1273
|
| 8953 |
+
},
|
| 8954 |
+
{
|
| 8955 |
+
"epoch": 0.21560331697410728,
|
| 8956 |
+
"grad_norm": 3.592133045196533,
|
| 8957 |
+
"learning_rate": 5.437289922235489e-06,
|
| 8958 |
+
"loss": 1.8888,
|
| 8959 |
+
"step": 1274
|
| 8960 |
+
},
|
| 8961 |
+
{
|
| 8962 |
+
"epoch": 0.2157725503469284,
|
| 8963 |
+
"grad_norm": 3.552612066268921,
|
| 8964 |
+
"learning_rate": 5.365208645345532e-06,
|
| 8965 |
+
"loss": 2.1888,
|
| 8966 |
+
"step": 1275
|
| 8967 |
+
},
|
| 8968 |
+
{
|
| 8969 |
+
"epoch": 0.21594178371974954,
|
| 8970 |
+
"grad_norm": 4.541576385498047,
|
| 8971 |
+
"learning_rate": 5.293595174761523e-06,
|
| 8972 |
+
"loss": 2.4499,
|
| 8973 |
+
"step": 1276
|
| 8974 |
+
},
|
| 8975 |
+
{
|
| 8976 |
+
"epoch": 0.21611101709257066,
|
| 8977 |
+
"grad_norm": 3.4159538745880127,
|
| 8978 |
+
"learning_rate": 5.222449864488943e-06,
|
| 8979 |
+
"loss": 2.2322,
|
| 8980 |
+
"step": 1277
|
| 8981 |
+
},
|
| 8982 |
+
{
|
| 8983 |
+
"epoch": 0.2162802504653918,
|
| 8984 |
+
"grad_norm": 3.6868090629577637,
|
| 8985 |
+
"learning_rate": 5.151773066219023e-06,
|
| 8986 |
+
"loss": 2.4456,
|
| 8987 |
+
"step": 1278
|
| 8988 |
+
},
|
| 8989 |
+
{
|
| 8990 |
+
"epoch": 0.21644948383821289,
|
| 8991 |
+
"grad_norm": 4.375200271606445,
|
| 8992 |
+
"learning_rate": 5.08156512932696e-06,
|
| 8993 |
+
"loss": 2.4408,
|
| 8994 |
+
"step": 1279
|
| 8995 |
+
},
|
| 8996 |
+
{
|
| 8997 |
+
"epoch": 0.216618717211034,
|
| 8998 |
+
"grad_norm": 3.336726665496826,
|
| 8999 |
+
"learning_rate": 5.011826400870301e-06,
|
| 9000 |
+
"loss": 2.1803,
|
| 9001 |
+
"step": 1280
|
| 9002 |
+
},
|
| 9003 |
+
{
|
| 9004 |
+
"epoch": 0.21678795058385514,
|
| 9005 |
+
"grad_norm": 5.18677282333374,
|
| 9006 |
+
"learning_rate": 4.942557225587119e-06,
|
| 9007 |
+
"loss": 2.5193,
|
| 9008 |
+
"step": 1281
|
| 9009 |
+
},
|
| 9010 |
+
{
|
| 9011 |
+
"epoch": 0.21695718395667626,
|
| 9012 |
+
"grad_norm": 4.183607578277588,
|
| 9013 |
+
"learning_rate": 4.8737579458943525e-06,
|
| 9014 |
+
"loss": 2.6889,
|
| 9015 |
+
"step": 1282
|
| 9016 |
+
},
|
| 9017 |
+
{
|
| 9018 |
+
"epoch": 0.2171264173294974,
|
| 9019 |
+
"grad_norm": 3.710606813430786,
|
| 9020 |
+
"learning_rate": 4.805428901886144e-06,
|
| 9021 |
+
"loss": 2.7766,
|
| 9022 |
+
"step": 1283
|
| 9023 |
+
},
|
| 9024 |
+
{
|
| 9025 |
+
"epoch": 0.2172956507023185,
|
| 9026 |
+
"grad_norm": 3.8886797428131104,
|
| 9027 |
+
"learning_rate": 4.737570431332128e-06,
|
| 9028 |
+
"loss": 2.2031,
|
| 9029 |
+
"step": 1284
|
| 9030 |
+
},
|
| 9031 |
+
{
|
| 9032 |
+
"epoch": 0.21746488407513961,
|
| 9033 |
+
"grad_norm": 3.714287281036377,
|
| 9034 |
+
"learning_rate": 4.670182869675721e-06,
|
| 9035 |
+
"loss": 2.5179,
|
| 9036 |
+
"step": 1285
|
| 9037 |
+
},
|
| 9038 |
+
{
|
| 9039 |
+
"epoch": 0.21763411744796074,
|
| 9040 |
+
"grad_norm": 3.832003116607666,
|
| 9041 |
+
"learning_rate": 4.60326655003257e-06,
|
| 9042 |
+
"loss": 1.9816,
|
| 9043 |
+
"step": 1286
|
| 9044 |
+
},
|
| 9045 |
+
{
|
| 9046 |
+
"epoch": 0.21780335082078187,
|
| 9047 |
+
"grad_norm": 4.320267677307129,
|
| 9048 |
+
"learning_rate": 4.536821803188773e-06,
|
| 9049 |
+
"loss": 2.4304,
|
| 9050 |
+
"step": 1287
|
| 9051 |
+
},
|
| 9052 |
+
{
|
| 9053 |
+
"epoch": 0.217972584193603,
|
| 9054 |
+
"grad_norm": 4.398102760314941,
|
| 9055 |
+
"learning_rate": 4.4708489575993496e-06,
|
| 9056 |
+
"loss": 2.147,
|
| 9057 |
+
"step": 1288
|
| 9058 |
+
},
|
| 9059 |
+
{
|
| 9060 |
+
"epoch": 0.2181418175664241,
|
| 9061 |
+
"grad_norm": 4.570779323577881,
|
| 9062 |
+
"learning_rate": 4.405348339386572e-06,
|
| 9063 |
+
"loss": 2.7773,
|
| 9064 |
+
"step": 1289
|
| 9065 |
+
},
|
| 9066 |
+
{
|
| 9067 |
+
"epoch": 0.21831105093924522,
|
| 9068 |
+
"grad_norm": 4.003580570220947,
|
| 9069 |
+
"learning_rate": 4.3403202723383715e-06,
|
| 9070 |
+
"loss": 3.0046,
|
| 9071 |
+
"step": 1290
|
| 9072 |
+
},
|
| 9073 |
+
{
|
| 9074 |
+
"epoch": 0.21848028431206634,
|
| 9075 |
+
"grad_norm": 3.8217952251434326,
|
| 9076 |
+
"learning_rate": 4.275765077906701e-06,
|
| 9077 |
+
"loss": 1.964,
|
| 9078 |
+
"step": 1291
|
| 9079 |
+
},
|
| 9080 |
+
{
|
| 9081 |
+
"epoch": 0.21864951768488747,
|
| 9082 |
+
"grad_norm": 3.9775161743164062,
|
| 9083 |
+
"learning_rate": 4.211683075206007e-06,
|
| 9084 |
+
"loss": 2.1878,
|
| 9085 |
+
"step": 1292
|
| 9086 |
+
},
|
| 9087 |
+
{
|
| 9088 |
+
"epoch": 0.2188187510577086,
|
| 9089 |
+
"grad_norm": 5.138439655303955,
|
| 9090 |
+
"learning_rate": 4.148074581011574e-06,
|
| 9091 |
+
"loss": 2.2576,
|
| 9092 |
+
"step": 1293
|
| 9093 |
+
},
|
| 9094 |
+
{
|
| 9095 |
+
"epoch": 0.2189879844305297,
|
| 9096 |
+
"grad_norm": 3.420557975769043,
|
| 9097 |
+
"learning_rate": 4.084939909758012e-06,
|
| 9098 |
+
"loss": 1.8567,
|
| 9099 |
+
"step": 1294
|
| 9100 |
+
},
|
| 9101 |
+
{
|
| 9102 |
+
"epoch": 0.21915721780335082,
|
| 9103 |
+
"grad_norm": 3.8210952281951904,
|
| 9104 |
+
"learning_rate": 4.02227937353773e-06,
|
| 9105 |
+
"loss": 2.2086,
|
| 9106 |
+
"step": 1295
|
| 9107 |
+
},
|
| 9108 |
+
{
|
| 9109 |
+
"epoch": 0.21932645117617194,
|
| 9110 |
+
"grad_norm": 5.87318754196167,
|
| 9111 |
+
"learning_rate": 3.96009328209932e-06,
|
| 9112 |
+
"loss": 2.6338,
|
| 9113 |
+
"step": 1296
|
| 9114 |
+
},
|
| 9115 |
+
{
|
| 9116 |
+
"epoch": 0.21949568454899307,
|
| 9117 |
+
"grad_norm": 4.616683483123779,
|
| 9118 |
+
"learning_rate": 3.898381942846041e-06,
|
| 9119 |
+
"loss": 3.199,
|
| 9120 |
+
"step": 1297
|
| 9121 |
+
},
|
| 9122 |
+
{
|
| 9123 |
+
"epoch": 0.2196649179218142,
|
| 9124 |
+
"grad_norm": 3.7832658290863037,
|
| 9125 |
+
"learning_rate": 3.837145660834385e-06,
|
| 9126 |
+
"loss": 2.6104,
|
| 9127 |
+
"step": 1298
|
| 9128 |
+
},
|
| 9129 |
+
{
|
| 9130 |
+
"epoch": 0.2198341512946353,
|
| 9131 |
+
"grad_norm": 4.564416408538818,
|
| 9132 |
+
"learning_rate": 3.7763847387724427e-06,
|
| 9133 |
+
"loss": 2.2879,
|
| 9134 |
+
"step": 1299
|
| 9135 |
+
},
|
| 9136 |
+
{
|
| 9137 |
+
"epoch": 0.22000338466745642,
|
| 9138 |
+
"grad_norm": 3.89300274848938,
|
| 9139 |
+
"learning_rate": 3.7160994770184755e-06,
|
| 9140 |
+
"loss": 2.2949,
|
| 9141 |
+
"step": 1300
|
| 9142 |
+
},
|
| 9143 |
+
{
|
| 9144 |
+
"epoch": 0.22017261804027755,
|
| 9145 |
+
"grad_norm": 6.857992649078369,
|
| 9146 |
+
"learning_rate": 3.656290173579446e-06,
|
| 9147 |
+
"loss": 2.2883,
|
| 9148 |
+
"step": 1301
|
| 9149 |
+
},
|
| 9150 |
+
{
|
| 9151 |
+
"epoch": 0.22034185141309867,
|
| 9152 |
+
"grad_norm": 4.126461505889893,
|
| 9153 |
+
"learning_rate": 3.5969571241095124e-06,
|
| 9154 |
+
"loss": 2.3038,
|
| 9155 |
+
"step": 1302
|
| 9156 |
+
},
|
| 9157 |
+
{
|
| 9158 |
+
"epoch": 0.22051108478591977,
|
| 9159 |
+
"grad_norm": 4.268237113952637,
|
| 9160 |
+
"learning_rate": 3.5381006219085687e-06,
|
| 9161 |
+
"loss": 2.7391,
|
| 9162 |
+
"step": 1303
|
| 9163 |
+
},
|
| 9164 |
+
{
|
| 9165 |
+
"epoch": 0.2206803181587409,
|
| 9166 |
+
"grad_norm": 5.277632713317871,
|
| 9167 |
+
"learning_rate": 3.4797209579208066e-06,
|
| 9168 |
+
"loss": 2.8033,
|
| 9169 |
+
"step": 1304
|
| 9170 |
+
},
|
| 9171 |
+
{
|
| 9172 |
+
"epoch": 0.22084955153156202,
|
| 9173 |
+
"grad_norm": 4.43790864944458,
|
| 9174 |
+
"learning_rate": 3.4218184207332806e-06,
|
| 9175 |
+
"loss": 2.6966,
|
| 9176 |
+
"step": 1305
|
| 9177 |
+
},
|
| 9178 |
+
{
|
| 9179 |
+
"epoch": 0.22101878490438315,
|
| 9180 |
+
"grad_norm": 3.7487828731536865,
|
| 9181 |
+
"learning_rate": 3.364393296574453e-06,
|
| 9182 |
+
"loss": 2.3024,
|
| 9183 |
+
"step": 1306
|
| 9184 |
+
},
|
| 9185 |
+
{
|
| 9186 |
+
"epoch": 0.22118801827720427,
|
| 9187 |
+
"grad_norm": 3.9089081287384033,
|
| 9188 |
+
"learning_rate": 3.3074458693128086e-06,
|
| 9189 |
+
"loss": 2.0895,
|
| 9190 |
+
"step": 1307
|
| 9191 |
+
},
|
| 9192 |
+
{
|
| 9193 |
+
"epoch": 0.22135725165002537,
|
| 9194 |
+
"grad_norm": 3.9897654056549072,
|
| 9195 |
+
"learning_rate": 3.2509764204554648e-06,
|
| 9196 |
+
"loss": 2.0479,
|
| 9197 |
+
"step": 1308
|
| 9198 |
+
},
|
| 9199 |
+
{
|
| 9200 |
+
"epoch": 0.2215264850228465,
|
| 9201 |
+
"grad_norm": 4.850733280181885,
|
| 9202 |
+
"learning_rate": 3.194985229146741e-06,
|
| 9203 |
+
"loss": 2.8468,
|
| 9204 |
+
"step": 1309
|
| 9205 |
+
},
|
| 9206 |
+
{
|
| 9207 |
+
"epoch": 0.22169571839566762,
|
| 9208 |
+
"grad_norm": 3.6863033771514893,
|
| 9209 |
+
"learning_rate": 3.1394725721667793e-06,
|
| 9210 |
+
"loss": 2.4135,
|
| 9211 |
+
"step": 1310
|
| 9212 |
+
},
|
| 9213 |
+
{
|
| 9214 |
+
"epoch": 0.22186495176848875,
|
| 9215 |
+
"grad_norm": 4.675195693969727,
|
| 9216 |
+
"learning_rate": 3.0844387239302053e-06,
|
| 9217 |
+
"loss": 2.7795,
|
| 9218 |
+
"step": 1311
|
| 9219 |
+
},
|
| 9220 |
+
{
|
| 9221 |
+
"epoch": 0.22203418514130988,
|
| 9222 |
+
"grad_norm": 4.724118709564209,
|
| 9223 |
+
"learning_rate": 3.0298839564847805e-06,
|
| 9224 |
+
"loss": 2.9463,
|
| 9225 |
+
"step": 1312
|
| 9226 |
+
},
|
| 9227 |
+
{
|
| 9228 |
+
"epoch": 0.22220341851413097,
|
| 9229 |
+
"grad_norm": 4.622027397155762,
|
| 9230 |
+
"learning_rate": 2.9758085395099834e-06,
|
| 9231 |
+
"loss": 2.6941,
|
| 9232 |
+
"step": 1313
|
| 9233 |
+
},
|
| 9234 |
+
{
|
| 9235 |
+
"epoch": 0.2223726518869521,
|
| 9236 |
+
"grad_norm": 4.488037109375,
|
| 9237 |
+
"learning_rate": 2.9222127403157775e-06,
|
| 9238 |
+
"loss": 2.747,
|
| 9239 |
+
"step": 1314
|
| 9240 |
+
},
|
| 9241 |
+
{
|
| 9242 |
+
"epoch": 0.22254188525977323,
|
| 9243 |
+
"grad_norm": 4.043283939361572,
|
| 9244 |
+
"learning_rate": 2.869096823841244e-06,
|
| 9245 |
+
"loss": 2.3501,
|
| 9246 |
+
"step": 1315
|
| 9247 |
+
},
|
| 9248 |
+
{
|
| 9249 |
+
"epoch": 0.22271111863259435,
|
| 9250 |
+
"grad_norm": 5.293920993804932,
|
| 9251 |
+
"learning_rate": 2.81646105265323e-06,
|
| 9252 |
+
"loss": 2.7319,
|
| 9253 |
+
"step": 1316
|
| 9254 |
+
},
|
| 9255 |
+
{
|
| 9256 |
+
"epoch": 0.22288035200541548,
|
| 9257 |
+
"grad_norm": 3.8888704776763916,
|
| 9258 |
+
"learning_rate": 2.764305686945101e-06,
|
| 9259 |
+
"loss": 2.7473,
|
| 9260 |
+
"step": 1317
|
| 9261 |
+
},
|
| 9262 |
+
{
|
| 9263 |
+
"epoch": 0.22304958537823658,
|
| 9264 |
+
"grad_norm": 4.733285427093506,
|
| 9265 |
+
"learning_rate": 2.712630984535469e-06,
|
| 9266 |
+
"loss": 3.0542,
|
| 9267 |
+
"step": 1318
|
| 9268 |
+
},
|
| 9269 |
+
{
|
| 9270 |
+
"epoch": 0.2232188187510577,
|
| 9271 |
+
"grad_norm": 3.7613630294799805,
|
| 9272 |
+
"learning_rate": 2.6614372008668674e-06,
|
| 9273 |
+
"loss": 2.5238,
|
| 9274 |
+
"step": 1319
|
| 9275 |
+
},
|
| 9276 |
+
{
|
| 9277 |
+
"epoch": 0.22338805212387883,
|
| 9278 |
+
"grad_norm": 5.252293109893799,
|
| 9279 |
+
"learning_rate": 2.6107245890045206e-06,
|
| 9280 |
+
"loss": 2.4133,
|
| 9281 |
+
"step": 1320
|
| 9282 |
+
},
|
| 9283 |
+
{
|
| 9284 |
+
"epoch": 0.22355728549669995,
|
| 9285 |
+
"grad_norm": 4.533067226409912,
|
| 9286 |
+
"learning_rate": 2.5604933996351e-06,
|
| 9287 |
+
"loss": 2.7163,
|
| 9288 |
+
"step": 1321
|
| 9289 |
+
},
|
| 9290 |
+
{
|
| 9291 |
+
"epoch": 0.22372651886952108,
|
| 9292 |
+
"grad_norm": 3.7899880409240723,
|
| 9293 |
+
"learning_rate": 2.510743881065447e-06,
|
| 9294 |
+
"loss": 2.3336,
|
| 9295 |
+
"step": 1322
|
| 9296 |
+
},
|
| 9297 |
+
{
|
| 9298 |
+
"epoch": 0.22389575224234218,
|
| 9299 |
+
"grad_norm": 4.550700664520264,
|
| 9300 |
+
"learning_rate": 2.461476279221375e-06,
|
| 9301 |
+
"loss": 2.1424,
|
| 9302 |
+
"step": 1323
|
| 9303 |
+
},
|
| 9304 |
+
{
|
| 9305 |
+
"epoch": 0.2240649856151633,
|
| 9306 |
+
"grad_norm": 4.408599853515625,
|
| 9307 |
+
"learning_rate": 2.41269083764647e-06,
|
| 9308 |
+
"loss": 2.6417,
|
| 9309 |
+
"step": 1324
|
| 9310 |
+
},
|
| 9311 |
+
{
|
| 9312 |
+
"epoch": 0.22423421898798443,
|
| 9313 |
+
"grad_norm": 4.652854919433594,
|
| 9314 |
+
"learning_rate": 2.364387797500822e-06,
|
| 9315 |
+
"loss": 2.2149,
|
| 9316 |
+
"step": 1325
|
| 9317 |
+
},
|
| 9318 |
+
{
|
| 9319 |
+
"epoch": 0.22440345236080556,
|
| 9320 |
+
"grad_norm": 4.0078864097595215,
|
| 9321 |
+
"learning_rate": 2.316567397559921e-06,
|
| 9322 |
+
"loss": 2.3853,
|
| 9323 |
+
"step": 1326
|
| 9324 |
+
},
|
| 9325 |
+
{
|
| 9326 |
+
"epoch": 0.22457268573362668,
|
| 9327 |
+
"grad_norm": 4.239938735961914,
|
| 9328 |
+
"learning_rate": 2.269229874213419e-06,
|
| 9329 |
+
"loss": 2.2991,
|
| 9330 |
+
"step": 1327
|
| 9331 |
+
},
|
| 9332 |
+
{
|
| 9333 |
+
"epoch": 0.22474191910644778,
|
| 9334 |
+
"grad_norm": 4.008525371551514,
|
| 9335 |
+
"learning_rate": 2.2223754614639436e-06,
|
| 9336 |
+
"loss": 2.9246,
|
| 9337 |
+
"step": 1328
|
| 9338 |
+
},
|
| 9339 |
+
{
|
| 9340 |
+
"epoch": 0.2249111524792689,
|
| 9341 |
+
"grad_norm": 4.444522380828857,
|
| 9342 |
+
"learning_rate": 2.176004390926001e-06,
|
| 9343 |
+
"loss": 2.714,
|
| 9344 |
+
"step": 1329
|
| 9345 |
+
},
|
| 9346 |
+
{
|
| 9347 |
+
"epoch": 0.22508038585209003,
|
| 9348 |
+
"grad_norm": 4.701035499572754,
|
| 9349 |
+
"learning_rate": 2.130116891824796e-06,
|
| 9350 |
+
"loss": 2.8857,
|
| 9351 |
+
"step": 1330
|
| 9352 |
+
},
|
| 9353 |
+
{
|
| 9354 |
+
"epoch": 0.22524961922491116,
|
| 9355 |
+
"grad_norm": 3.9119105339050293,
|
| 9356 |
+
"learning_rate": 2.0847131909950913e-06,
|
| 9357 |
+
"loss": 2.1843,
|
| 9358 |
+
"step": 1331
|
| 9359 |
+
},
|
| 9360 |
+
{
|
| 9361 |
+
"epoch": 0.22541885259773228,
|
| 9362 |
+
"grad_norm": 3.347607135772705,
|
| 9363 |
+
"learning_rate": 2.0397935128801283e-06,
|
| 9364 |
+
"loss": 1.9973,
|
| 9365 |
+
"step": 1332
|
| 9366 |
+
},
|
| 9367 |
+
{
|
| 9368 |
+
"epoch": 0.22558808597055338,
|
| 9369 |
+
"grad_norm": 4.204796314239502,
|
| 9370 |
+
"learning_rate": 1.995358079530463e-06,
|
| 9371 |
+
"loss": 2.7891,
|
| 9372 |
+
"step": 1333
|
| 9373 |
+
},
|
| 9374 |
+
{
|
| 9375 |
+
"epoch": 0.2257573193433745,
|
| 9376 |
+
"grad_norm": 3.666780471801758,
|
| 9377 |
+
"learning_rate": 1.951407110602899e-06,
|
| 9378 |
+
"loss": 2.076,
|
| 9379 |
+
"step": 1334
|
| 9380 |
+
},
|
| 9381 |
+
{
|
| 9382 |
+
"epoch": 0.22592655271619563,
|
| 9383 |
+
"grad_norm": 4.629287242889404,
|
| 9384 |
+
"learning_rate": 1.9079408233594e-06,
|
| 9385 |
+
"loss": 2.9586,
|
| 9386 |
+
"step": 1335
|
| 9387 |
+
},
|
| 9388 |
+
{
|
| 9389 |
+
"epoch": 0.22609578608901676,
|
| 9390 |
+
"grad_norm": 4.100510597229004,
|
| 9391 |
+
"learning_rate": 1.8649594326660025e-06,
|
| 9392 |
+
"loss": 2.4321,
|
| 9393 |
+
"step": 1336
|
| 9394 |
+
},
|
| 9395 |
+
{
|
| 9396 |
+
"epoch": 0.22626501946183789,
|
| 9397 |
+
"grad_norm": 4.083179950714111,
|
| 9398 |
+
"learning_rate": 1.822463150991771e-06,
|
| 9399 |
+
"loss": 2.4957,
|
| 9400 |
+
"step": 1337
|
| 9401 |
+
},
|
| 9402 |
+
{
|
| 9403 |
+
"epoch": 0.22643425283465898,
|
| 9404 |
+
"grad_norm": 5.191402435302734,
|
| 9405 |
+
"learning_rate": 1.780452188407744e-06,
|
| 9406 |
+
"loss": 2.417,
|
| 9407 |
+
"step": 1338
|
| 9408 |
+
},
|
| 9409 |
+
{
|
| 9410 |
+
"epoch": 0.2266034862074801,
|
| 9411 |
+
"grad_norm": 3.7358388900756836,
|
| 9412 |
+
"learning_rate": 1.738926752585901e-06,
|
| 9413 |
+
"loss": 2.3845,
|
| 9414 |
+
"step": 1339
|
| 9415 |
+
},
|
| 9416 |
+
{
|
| 9417 |
+
"epoch": 0.22677271958030124,
|
| 9418 |
+
"grad_norm": 5.2599310874938965,
|
| 9419 |
+
"learning_rate": 1.697887048798108e-06,
|
| 9420 |
+
"loss": 2.1591,
|
| 9421 |
+
"step": 1340
|
| 9422 |
+
},
|
| 9423 |
+
{
|
| 9424 |
+
"epoch": 0.22694195295312236,
|
| 9425 |
+
"grad_norm": 4.448916912078857,
|
| 9426 |
+
"learning_rate": 1.6573332799151076e-06,
|
| 9427 |
+
"loss": 2.65,
|
| 9428 |
+
"step": 1341
|
| 9429 |
+
},
|
| 9430 |
+
{
|
| 9431 |
+
"epoch": 0.2271111863259435,
|
| 9432 |
+
"grad_norm": 4.82081937789917,
|
| 9433 |
+
"learning_rate": 1.6172656464055747e-06,
|
| 9434 |
+
"loss": 2.4846,
|
| 9435 |
+
"step": 1342
|
| 9436 |
+
},
|
| 9437 |
+
{
|
| 9438 |
+
"epoch": 0.22728041969876459,
|
| 9439 |
+
"grad_norm": 4.220665454864502,
|
| 9440 |
+
"learning_rate": 1.5776843463350288e-06,
|
| 9441 |
+
"loss": 2.1787,
|
| 9442 |
+
"step": 1343
|
| 9443 |
+
},
|
| 9444 |
+
{
|
| 9445 |
+
"epoch": 0.2274496530715857,
|
| 9446 |
+
"grad_norm": 5.071519374847412,
|
| 9447 |
+
"learning_rate": 1.538589575364946e-06,
|
| 9448 |
+
"loss": 2.5224,
|
| 9449 |
+
"step": 1344
|
| 9450 |
+
},
|
| 9451 |
+
{
|
| 9452 |
+
"epoch": 0.22761888644440684,
|
| 9453 |
+
"grad_norm": 4.073892116546631,
|
| 9454 |
+
"learning_rate": 1.4999815267517593e-06,
|
| 9455 |
+
"loss": 2.2127,
|
| 9456 |
+
"step": 1345
|
| 9457 |
+
},
|
| 9458 |
+
{
|
| 9459 |
+
"epoch": 0.22778811981722796,
|
| 9460 |
+
"grad_norm": 3.7308478355407715,
|
| 9461 |
+
"learning_rate": 1.4618603913458596e-06,
|
| 9462 |
+
"loss": 2.5417,
|
| 9463 |
+
"step": 1346
|
| 9464 |
+
},
|
| 9465 |
+
{
|
| 9466 |
+
"epoch": 0.2279573531900491,
|
| 9467 |
+
"grad_norm": 4.07778263092041,
|
| 9468 |
+
"learning_rate": 1.4242263575906967e-06,
|
| 9469 |
+
"loss": 2.789,
|
| 9470 |
+
"step": 1347
|
| 9471 |
+
},
|
| 9472 |
+
{
|
| 9473 |
+
"epoch": 0.2281265865628702,
|
| 9474 |
+
"grad_norm": 5.16205358505249,
|
| 9475 |
+
"learning_rate": 1.3870796115218687e-06,
|
| 9476 |
+
"loss": 2.9348,
|
| 9477 |
+
"step": 1348
|
| 9478 |
+
},
|
| 9479 |
+
{
|
| 9480 |
+
"epoch": 0.2282958199356913,
|
| 9481 |
+
"grad_norm": 4.10355281829834,
|
| 9482 |
+
"learning_rate": 1.350420336766134e-06,
|
| 9483 |
+
"loss": 2.5149,
|
| 9484 |
+
"step": 1349
|
| 9485 |
+
},
|
| 9486 |
+
{
|
| 9487 |
+
"epoch": 0.22846505330851244,
|
| 9488 |
+
"grad_norm": 3.0065155029296875,
|
| 9489 |
+
"learning_rate": 1.314248714540589e-06,
|
| 9490 |
+
"loss": 1.4556,
|
| 9491 |
+
"step": 1350
|
| 9492 |
+
},
|
| 9493 |
+
{
|
| 9494 |
+
"epoch": 0.22863428668133357,
|
| 9495 |
+
"grad_norm": 3.5119566917419434,
|
| 9496 |
+
"learning_rate": 1.2785649236517038e-06,
|
| 9497 |
+
"loss": 2.1689,
|
| 9498 |
+
"step": 1351
|
| 9499 |
+
},
|
| 9500 |
+
{
|
| 9501 |
+
"epoch": 0.2288035200541547,
|
| 9502 |
+
"grad_norm": 3.5558135509490967,
|
| 9503 |
+
"learning_rate": 1.2433691404944547e-06,
|
| 9504 |
+
"loss": 2.5565,
|
| 9505 |
+
"step": 1352
|
| 9506 |
+
},
|
| 9507 |
+
{
|
| 9508 |
+
"epoch": 0.2289727534269758,
|
| 9509 |
+
"grad_norm": 4.180010795593262,
|
| 9510 |
+
"learning_rate": 1.2086615390514478e-06,
|
| 9511 |
+
"loss": 2.4111,
|
| 9512 |
+
"step": 1353
|
| 9513 |
+
},
|
| 9514 |
+
{
|
| 9515 |
+
"epoch": 0.22914198679979692,
|
| 9516 |
+
"grad_norm": 3.945111036300659,
|
| 9517 |
+
"learning_rate": 1.1744422908921193e-06,
|
| 9518 |
+
"loss": 2.7189,
|
| 9519 |
+
"step": 1354
|
| 9520 |
+
},
|
| 9521 |
+
{
|
| 9522 |
+
"epoch": 0.22931122017261804,
|
| 9523 |
+
"grad_norm": 3.6358213424682617,
|
| 9524 |
+
"learning_rate": 1.1407115651717703e-06,
|
| 9525 |
+
"loss": 1.8899,
|
| 9526 |
+
"step": 1355
|
| 9527 |
+
},
|
| 9528 |
+
{
|
| 9529 |
+
"epoch": 0.22948045354543917,
|
| 9530 |
+
"grad_norm": 4.161632061004639,
|
| 9531 |
+
"learning_rate": 1.1074695286308667e-06,
|
| 9532 |
+
"loss": 1.9383,
|
| 9533 |
+
"step": 1356
|
| 9534 |
+
},
|
| 9535 |
+
{
|
| 9536 |
+
"epoch": 0.2296496869182603,
|
| 9537 |
+
"grad_norm": 4.46504020690918,
|
| 9538 |
+
"learning_rate": 1.0747163455941178e-06,
|
| 9539 |
+
"loss": 2.687,
|
| 9540 |
+
"step": 1357
|
| 9541 |
+
},
|
| 9542 |
+
{
|
| 9543 |
+
"epoch": 0.2298189202910814,
|
| 9544 |
+
"grad_norm": 3.580151081085205,
|
| 9545 |
+
"learning_rate": 1.0424521779696662e-06,
|
| 9546 |
+
"loss": 2.3252,
|
| 9547 |
+
"step": 1358
|
| 9548 |
+
},
|
| 9549 |
+
{
|
| 9550 |
+
"epoch": 0.22998815366390252,
|
| 9551 |
+
"grad_norm": 4.707612037658691,
|
| 9552 |
+
"learning_rate": 1.0106771852483431e-06,
|
| 9553 |
+
"loss": 2.9648,
|
| 9554 |
+
"step": 1359
|
| 9555 |
+
},
|
| 9556 |
+
{
|
| 9557 |
+
"epoch": 0.23015738703672364,
|
| 9558 |
+
"grad_norm": 4.678205966949463,
|
| 9559 |
+
"learning_rate": 9.793915245028596e-07,
|
| 9560 |
+
"loss": 2.7101,
|
| 9561 |
+
"step": 1360
|
| 9562 |
+
},
|
| 9563 |
+
{
|
| 9564 |
+
"epoch": 0.23032662040954477,
|
| 9565 |
+
"grad_norm": 4.214313983917236,
|
| 9566 |
+
"learning_rate": 9.485953503869826e-07,
|
| 9567 |
+
"loss": 2.2753,
|
| 9568 |
+
"step": 1361
|
| 9569 |
+
},
|
| 9570 |
+
{
|
| 9571 |
+
"epoch": 0.2304958537823659,
|
| 9572 |
+
"grad_norm": 4.550292015075684,
|
| 9573 |
+
"learning_rate": 9.182888151348712e-07,
|
| 9574 |
+
"loss": 3.1793,
|
| 9575 |
+
"step": 1362
|
| 9576 |
+
},
|
| 9577 |
+
{
|
| 9578 |
+
"epoch": 0.230665087155187,
|
| 9579 |
+
"grad_norm": 4.984288692474365,
|
| 9580 |
+
"learning_rate": 8.884720685601977e-07,
|
| 9581 |
+
"loss": 2.7728,
|
| 9582 |
+
"step": 1363
|
| 9583 |
+
},
|
| 9584 |
+
{
|
| 9585 |
+
"epoch": 0.23083432052800812,
|
| 9586 |
+
"grad_norm": 4.659316539764404,
|
| 9587 |
+
"learning_rate": 8.591452580555159e-07,
|
| 9588 |
+
"loss": 2.3956,
|
| 9589 |
+
"step": 1364
|
| 9590 |
+
},
|
| 9591 |
+
{
|
| 9592 |
+
"epoch": 0.23100355390082924,
|
| 9593 |
+
"grad_norm": 4.046222686767578,
|
| 9594 |
+
"learning_rate": 8.303085285914613e-07,
|
| 9595 |
+
"loss": 2.6888,
|
| 9596 |
+
"step": 1365
|
| 9597 |
+
},
|
| 9598 |
+
{
|
| 9599 |
+
"epoch": 0.23117278727365037,
|
| 9600 |
+
"grad_norm": 3.9164886474609375,
|
| 9601 |
+
"learning_rate": 8.019620227160851e-07,
|
| 9602 |
+
"loss": 2.3108,
|
| 9603 |
+
"step": 1366
|
| 9604 |
+
},
|
| 9605 |
+
{
|
| 9606 |
+
"epoch": 0.2313420206464715,
|
| 9607 |
+
"grad_norm": 4.377208709716797,
|
| 9608 |
+
"learning_rate": 7.741058805540991e-07,
|
| 9609 |
+
"loss": 2.2299,
|
| 9610 |
+
"step": 1367
|
| 9611 |
+
},
|
| 9612 |
+
{
|
| 9613 |
+
"epoch": 0.2315112540192926,
|
| 9614 |
+
"grad_norm": 4.208359241485596,
|
| 9615 |
+
"learning_rate": 7.467402398062206e-07,
|
| 9616 |
+
"loss": 2.4627,
|
| 9617 |
+
"step": 1368
|
| 9618 |
+
},
|
| 9619 |
+
{
|
| 9620 |
+
"epoch": 0.23168048739211372,
|
| 9621 |
+
"grad_norm": 4.220484733581543,
|
| 9622 |
+
"learning_rate": 7.19865235748507e-07,
|
| 9623 |
+
"loss": 2.1133,
|
| 9624 |
+
"step": 1369
|
| 9625 |
+
},
|
| 9626 |
+
{
|
| 9627 |
+
"epoch": 0.23184972076493485,
|
| 9628 |
+
"grad_norm": 3.961956024169922,
|
| 9629 |
+
"learning_rate": 6.934810012316217e-07,
|
| 9630 |
+
"loss": 2.2017,
|
| 9631 |
+
"step": 1370
|
| 9632 |
+
},
|
| 9633 |
+
{
|
| 9634 |
+
"epoch": 0.23201895413775597,
|
| 9635 |
+
"grad_norm": 4.142365455627441,
|
| 9636 |
+
"learning_rate": 6.675876666802361e-07,
|
| 9637 |
+
"loss": 2.5549,
|
| 9638 |
+
"step": 1371
|
| 9639 |
+
},
|
| 9640 |
+
{
|
| 9641 |
+
"epoch": 0.2321881875105771,
|
| 9642 |
+
"grad_norm": 3.670931577682495,
|
| 9643 |
+
"learning_rate": 6.421853600923622e-07,
|
| 9644 |
+
"loss": 1.9114,
|
| 9645 |
+
"step": 1372
|
| 9646 |
+
},
|
| 9647 |
+
{
|
| 9648 |
+
"epoch": 0.2323574208833982,
|
| 9649 |
+
"grad_norm": 3.944248914718628,
|
| 9650 |
+
"learning_rate": 6.172742070387206e-07,
|
| 9651 |
+
"loss": 2.3554,
|
| 9652 |
+
"step": 1373
|
| 9653 |
+
},
|
| 9654 |
+
{
|
| 9655 |
+
"epoch": 0.23252665425621932,
|
| 9656 |
+
"grad_norm": 4.574911117553711,
|
| 9657 |
+
"learning_rate": 5.928543306621293e-07,
|
| 9658 |
+
"loss": 2.6813,
|
| 9659 |
+
"step": 1374
|
| 9660 |
+
},
|
| 9661 |
+
{
|
| 9662 |
+
"epoch": 0.23269588762904045,
|
| 9663 |
+
"grad_norm": 4.949753761291504,
|
| 9664 |
+
"learning_rate": 5.689258516768825e-07,
|
| 9665 |
+
"loss": 3.0825,
|
| 9666 |
+
"step": 1375
|
| 9667 |
+
},
|
| 9668 |
+
{
|
| 9669 |
+
"epoch": 0.23286512100186157,
|
| 9670 |
+
"grad_norm": 4.875388145446777,
|
| 9671 |
+
"learning_rate": 5.454888883681731e-07,
|
| 9672 |
+
"loss": 2.7601,
|
| 9673 |
+
"step": 1376
|
| 9674 |
+
},
|
| 9675 |
+
{
|
| 9676 |
+
"epoch": 0.2330343543746827,
|
| 9677 |
+
"grad_norm": 4.345504283905029,
|
| 9678 |
+
"learning_rate": 5.225435565914483e-07,
|
| 9679 |
+
"loss": 2.5763,
|
| 9680 |
+
"step": 1377
|
| 9681 |
+
},
|
| 9682 |
+
{
|
| 9683 |
+
"epoch": 0.2332035877475038,
|
| 9684 |
+
"grad_norm": 3.981414318084717,
|
| 9685 |
+
"learning_rate": 5.000899697719552e-07,
|
| 9686 |
+
"loss": 2.3318,
|
| 9687 |
+
"step": 1378
|
| 9688 |
+
},
|
| 9689 |
+
{
|
| 9690 |
+
"epoch": 0.23337282112032492,
|
| 9691 |
+
"grad_norm": 4.636585235595703,
|
| 9692 |
+
"learning_rate": 4.781282389040409e-07,
|
| 9693 |
+
"loss": 2.6304,
|
| 9694 |
+
"step": 1379
|
| 9695 |
+
},
|
| 9696 |
+
{
|
| 9697 |
+
"epoch": 0.23354205449314605,
|
| 9698 |
+
"grad_norm": 4.593474864959717,
|
| 9699 |
+
"learning_rate": 4.5665847255070835e-07,
|
| 9700 |
+
"loss": 2.2621,
|
| 9701 |
+
"step": 1380
|
| 9702 |
+
},
|
| 9703 |
+
{
|
| 9704 |
+
"epoch": 0.23371128786596718,
|
| 9705 |
+
"grad_norm": 4.252377986907959,
|
| 9706 |
+
"learning_rate": 4.3568077684302823e-07,
|
| 9707 |
+
"loss": 2.372,
|
| 9708 |
+
"step": 1381
|
| 9709 |
+
},
|
| 9710 |
+
{
|
| 9711 |
+
"epoch": 0.2338805212387883,
|
| 9712 |
+
"grad_norm": 3.252465009689331,
|
| 9713 |
+
"learning_rate": 4.15195255479639e-07,
|
| 9714 |
+
"loss": 2.0566,
|
| 9715 |
+
"step": 1382
|
| 9716 |
+
},
|
| 9717 |
+
{
|
| 9718 |
+
"epoch": 0.2340497546116094,
|
| 9719 |
+
"grad_norm": 4.524016857147217,
|
| 9720 |
+
"learning_rate": 3.9520200972621436e-07,
|
| 9721 |
+
"loss": 2.2689,
|
| 9722 |
+
"step": 1383
|
| 9723 |
+
},
|
| 9724 |
+
{
|
| 9725 |
+
"epoch": 0.23421898798443053,
|
| 9726 |
+
"grad_norm": 4.80737829208374,
|
| 9727 |
+
"learning_rate": 3.7570113841497447e-07,
|
| 9728 |
+
"loss": 2.312,
|
| 9729 |
+
"step": 1384
|
| 9730 |
+
},
|
| 9731 |
+
{
|
| 9732 |
+
"epoch": 0.23438822135725165,
|
| 9733 |
+
"grad_norm": 3.7588346004486084,
|
| 9734 |
+
"learning_rate": 3.5669273794418644e-07,
|
| 9735 |
+
"loss": 1.9949,
|
| 9736 |
+
"step": 1385
|
| 9737 |
+
},
|
| 9738 |
+
{
|
| 9739 |
+
"epoch": 0.23455745473007278,
|
| 9740 |
+
"grad_norm": 3.397432804107666,
|
| 9741 |
+
"learning_rate": 3.381769022776982e-07,
|
| 9742 |
+
"loss": 2.0444,
|
| 9743 |
+
"step": 1386
|
| 9744 |
+
},
|
| 9745 |
+
{
|
| 9746 |
+
"epoch": 0.2347266881028939,
|
| 9747 |
+
"grad_norm": 3.9758918285369873,
|
| 9748 |
+
"learning_rate": 3.2015372294450507e-07,
|
| 9749 |
+
"loss": 2.7044,
|
| 9750 |
+
"step": 1387
|
| 9751 |
+
},
|
| 9752 |
+
{
|
| 9753 |
+
"epoch": 0.234895921475715,
|
| 9754 |
+
"grad_norm": 3.689208745956421,
|
| 9755 |
+
"learning_rate": 3.0262328903820633e-07,
|
| 9756 |
+
"loss": 1.3957,
|
| 9757 |
+
"step": 1388
|
| 9758 |
+
},
|
| 9759 |
+
{
|
| 9760 |
+
"epoch": 0.23506515484853613,
|
| 9761 |
+
"grad_norm": 3.5054447650909424,
|
| 9762 |
+
"learning_rate": 2.855856872166607e-07,
|
| 9763 |
+
"loss": 2.096,
|
| 9764 |
+
"step": 1389
|
| 9765 |
+
},
|
| 9766 |
+
{
|
| 9767 |
+
"epoch": 0.23523438822135725,
|
| 9768 |
+
"grad_norm": 4.941432476043701,
|
| 9769 |
+
"learning_rate": 2.6904100170150883e-07,
|
| 9770 |
+
"loss": 2.3338,
|
| 9771 |
+
"step": 1390
|
| 9772 |
+
},
|
| 9773 |
+
{
|
| 9774 |
+
"epoch": 0.23540362159417838,
|
| 9775 |
+
"grad_norm": 3.514775514602661,
|
| 9776 |
+
"learning_rate": 2.5298931427777396e-07,
|
| 9777 |
+
"loss": 2.1124,
|
| 9778 |
+
"step": 1391
|
| 9779 |
+
},
|
| 9780 |
+
{
|
| 9781 |
+
"epoch": 0.23557285496699948,
|
| 9782 |
+
"grad_norm": 3.9897100925445557,
|
| 9783 |
+
"learning_rate": 2.3743070429345093e-07,
|
| 9784 |
+
"loss": 2.0383,
|
| 9785 |
+
"step": 1392
|
| 9786 |
+
},
|
| 9787 |
+
{
|
| 9788 |
+
"epoch": 0.2357420883398206,
|
| 9789 |
+
"grad_norm": 4.227541923522949,
|
| 9790 |
+
"learning_rate": 2.2236524865910658e-07,
|
| 9791 |
+
"loss": 2.1775,
|
| 9792 |
+
"step": 1393
|
| 9793 |
+
},
|
| 9794 |
+
{
|
| 9795 |
+
"epoch": 0.23591132171264173,
|
| 9796 |
+
"grad_norm": 3.1946425437927246,
|
| 9797 |
+
"learning_rate": 2.0779302184751325e-07,
|
| 9798 |
+
"loss": 2.0118,
|
| 9799 |
+
"step": 1394
|
| 9800 |
+
},
|
| 9801 |
+
{
|
| 9802 |
+
"epoch": 0.23608055508546286,
|
| 9803 |
+
"grad_norm": 3.867051601409912,
|
| 9804 |
+
"learning_rate": 1.937140958932604e-07,
|
| 9805 |
+
"loss": 2.4111,
|
| 9806 |
+
"step": 1395
|
| 9807 |
+
},
|
| 9808 |
+
{
|
| 9809 |
+
"epoch": 0.23624978845828398,
|
| 9810 |
+
"grad_norm": 4.440875053405762,
|
| 9811 |
+
"learning_rate": 1.8012854039244353e-07,
|
| 9812 |
+
"loss": 2.065,
|
| 9813 |
+
"step": 1396
|
| 9814 |
+
},
|
| 9815 |
+
{
|
| 9816 |
+
"epoch": 0.23641902183110508,
|
| 9817 |
+
"grad_norm": 5.504641056060791,
|
| 9818 |
+
"learning_rate": 1.670364225022758e-07,
|
| 9819 |
+
"loss": 2.3649,
|
| 9820 |
+
"step": 1397
|
| 9821 |
+
},
|
| 9822 |
+
{
|
| 9823 |
+
"epoch": 0.2365882552039262,
|
| 9824 |
+
"grad_norm": 5.530081748962402,
|
| 9825 |
+
"learning_rate": 1.5443780694074373e-07,
|
| 9826 |
+
"loss": 2.8206,
|
| 9827 |
+
"step": 1398
|
| 9828 |
+
},
|
| 9829 |
+
{
|
| 9830 |
+
"epoch": 0.23675748857674733,
|
| 9831 |
+
"grad_norm": 4.322780609130859,
|
| 9832 |
+
"learning_rate": 1.4233275598635186e-07,
|
| 9833 |
+
"loss": 3.3659,
|
| 9834 |
+
"step": 1399
|
| 9835 |
+
},
|
| 9836 |
+
{
|
| 9837 |
+
"epoch": 0.23692672194956846,
|
| 9838 |
+
"grad_norm": 4.434557914733887,
|
| 9839 |
+
"learning_rate": 1.307213294777676e-07,
|
| 9840 |
+
"loss": 2.6113,
|
| 9841 |
+
"step": 1400
|
| 9842 |
+
},
|
| 9843 |
+
{
|
| 9844 |
+
"epoch": 0.23709595532238958,
|
| 9845 |
+
"grad_norm": 3.878546953201294,
|
| 9846 |
+
"learning_rate": 1.196035848135102e-07,
|
| 9847 |
+
"loss": 2.454,
|
| 9848 |
+
"step": 1401
|
| 9849 |
+
},
|
| 9850 |
+
{
|
| 9851 |
+
"epoch": 0.23726518869521068,
|
| 9852 |
+
"grad_norm": 4.334795951843262,
|
| 9853 |
+
"learning_rate": 1.0897957695171768e-07,
|
| 9854 |
+
"loss": 2.5534,
|
| 9855 |
+
"step": 1402
|
| 9856 |
+
},
|
| 9857 |
+
{
|
| 9858 |
+
"epoch": 0.2374344220680318,
|
| 9859 |
+
"grad_norm": 5.5782060623168945,
|
| 9860 |
+
"learning_rate": 9.88493584098471e-08,
|
| 9861 |
+
"loss": 2.6021,
|
| 9862 |
+
"step": 1403
|
| 9863 |
+
},
|
| 9864 |
+
{
|
| 9865 |
+
"epoch": 0.23760365544085293,
|
| 9866 |
+
"grad_norm": 3.7061827182769775,
|
| 9867 |
+
"learning_rate": 8.921297926439698e-08,
|
| 9868 |
+
"loss": 2.634,
|
| 9869 |
+
"step": 1404
|
| 9870 |
+
},
|
| 9871 |
+
{
|
| 9872 |
+
"epoch": 0.23777288881367406,
|
| 9873 |
+
"grad_norm": 3.850057601928711,
|
| 9874 |
+
"learning_rate": 8.007048715068521e-08,
|
| 9875 |
+
"loss": 2.484,
|
| 9876 |
+
"step": 1405
|
| 9877 |
+
},
|
| 9878 |
+
{
|
| 9879 |
+
"epoch": 0.2379421221864952,
|
| 9880 |
+
"grad_norm": 4.672886371612549,
|
| 9881 |
+
"learning_rate": 7.142192726263818e-08,
|
| 9882 |
+
"loss": 2.3213,
|
| 9883 |
+
"step": 1406
|
| 9884 |
+
},
|
| 9885 |
+
{
|
| 9886 |
+
"epoch": 0.23811135555931628,
|
| 9887 |
+
"grad_norm": 4.367487907409668,
|
| 9888 |
+
"learning_rate": 6.326734235249099e-08,
|
| 9889 |
+
"loss": 2.7815,
|
| 9890 |
+
"step": 1407
|
| 9891 |
+
},
|
| 9892 |
+
{
|
| 9893 |
+
"epoch": 0.2382805889321374,
|
| 9894 |
+
"grad_norm": 4.483260154724121,
|
| 9895 |
+
"learning_rate": 5.560677273064307e-08,
|
| 9896 |
+
"loss": 2.539,
|
| 9897 |
+
"step": 1408
|
| 9898 |
+
},
|
| 9899 |
+
{
|
| 9900 |
+
"epoch": 0.23844982230495854,
|
| 9901 |
+
"grad_norm": 6.026078224182129,
|
| 9902 |
+
"learning_rate": 4.8440256265447345e-08,
|
| 9903 |
+
"loss": 2.7139,
|
| 9904 |
+
"step": 1409
|
| 9905 |
+
},
|
| 9906 |
+
{
|
| 9907 |
+
"epoch": 0.23861905567777966,
|
| 9908 |
+
"grad_norm": 4.309380531311035,
|
| 9909 |
+
"learning_rate": 4.17678283830103e-08,
|
| 9910 |
+
"loss": 2.7043,
|
| 9911 |
+
"step": 1410
|
| 9912 |
+
},
|
| 9913 |
+
{
|
| 9914 |
+
"epoch": 0.2387882890506008,
|
| 9915 |
+
"grad_norm": 3.0925745964050293,
|
| 9916 |
+
"learning_rate": 3.5589522066992176e-08,
|
| 9917 |
+
"loss": 1.8575,
|
| 9918 |
+
"step": 1411
|
| 9919 |
+
},
|
| 9920 |
+
{
|
| 9921 |
+
"epoch": 0.2389575224234219,
|
| 9922 |
+
"grad_norm": 4.351414680480957,
|
| 9923 |
+
"learning_rate": 2.9905367858507064e-08,
|
| 9924 |
+
"loss": 2.3953,
|
| 9925 |
+
"step": 1412
|
| 9926 |
+
},
|
| 9927 |
+
{
|
| 9928 |
+
"epoch": 0.239126755796243,
|
| 9929 |
+
"grad_norm": 5.115963935852051,
|
| 9930 |
+
"learning_rate": 2.471539385592303e-08,
|
| 9931 |
+
"loss": 2.4598,
|
| 9932 |
+
"step": 1413
|
| 9933 |
+
},
|
| 9934 |
+
{
|
| 9935 |
+
"epoch": 0.23929598916906414,
|
| 9936 |
+
"grad_norm": 4.487088680267334,
|
| 9937 |
+
"learning_rate": 2.0019625714740032e-08,
|
| 9938 |
+
"loss": 2.7845,
|
| 9939 |
+
"step": 1414
|
| 9940 |
+
},
|
| 9941 |
+
{
|
| 9942 |
+
"epoch": 0.23946522254188526,
|
| 9943 |
+
"grad_norm": 3.7317934036254883,
|
| 9944 |
+
"learning_rate": 1.5818086647445552e-08,
|
| 9945 |
+
"loss": 2.6197,
|
| 9946 |
+
"step": 1415
|
| 9947 |
+
},
|
| 9948 |
+
{
|
| 9949 |
+
"epoch": 0.2396344559147064,
|
| 9950 |
+
"grad_norm": 3.9608500003814697,
|
| 9951 |
+
"learning_rate": 1.2110797423436904e-08,
|
| 9952 |
+
"loss": 1.7485,
|
| 9953 |
+
"step": 1416
|
| 9954 |
+
},
|
| 9955 |
+
{
|
| 9956 |
+
"epoch": 0.2398036892875275,
|
| 9957 |
+
"grad_norm": 3.5881524085998535,
|
| 9958 |
+
"learning_rate": 8.897776368865795e-09,
|
| 9959 |
+
"loss": 2.072,
|
| 9960 |
+
"step": 1417
|
| 9961 |
+
},
|
| 9962 |
+
{
|
| 9963 |
+
"epoch": 0.23997292266034861,
|
| 9964 |
+
"grad_norm": 4.391948699951172,
|
| 9965 |
+
"learning_rate": 6.179039366616124e-09,
|
| 9966 |
+
"loss": 2.0867,
|
| 9967 |
+
"step": 1418
|
| 9968 |
+
},
|
| 9969 |
+
{
|
| 9970 |
+
"epoch": 0.24014215603316974,
|
| 9971 |
+
"grad_norm": 4.468619346618652,
|
| 9972 |
+
"learning_rate": 3.954599856159646e-09,
|
| 9973 |
+
"loss": 2.5233,
|
| 9974 |
+
"step": 1419
|
| 9975 |
+
},
|
| 9976 |
+
{
|
| 9977 |
+
"epoch": 0.24031138940599087,
|
| 9978 |
+
"grad_norm": 3.5762696266174316,
|
| 9979 |
+
"learning_rate": 2.224468833522675e-09,
|
| 9980 |
+
"loss": 2.0478,
|
| 9981 |
+
"step": 1420
|
| 9982 |
+
},
|
| 9983 |
+
{
|
| 9984 |
+
"epoch": 0.240480622778812,
|
| 9985 |
+
"grad_norm": 3.846388339996338,
|
| 9986 |
+
"learning_rate": 9.886548512305637e-10,
|
| 9987 |
+
"loss": 2.5218,
|
| 9988 |
+
"step": 1421
|
| 9989 |
+
},
|
| 9990 |
+
{
|
| 9991 |
+
"epoch": 0.2406498561516331,
|
| 9992 |
+
"grad_norm": 3.55391263961792,
|
| 9993 |
+
"learning_rate": 2.471640182633017e-10,
|
| 9994 |
+
"loss": 2.2687,
|
| 9995 |
+
"step": 1422
|
| 9996 |
+
},
|
| 9997 |
+
{
|
| 9998 |
+
"epoch": 0.24081908952445422,
|
| 9999 |
+
"grad_norm": 4.181069850921631,
|
| 10000 |
+
"learning_rate": 0.0,
|
| 10001 |
+
"loss": 2.4863,
|
| 10002 |
+
"step": 1423
|
| 10003 |
}
|
| 10004 |
],
|
| 10005 |
"logging_steps": 1,
|
|
|
|
| 10014 |
"should_evaluate": false,
|
| 10015 |
"should_log": false,
|
| 10016 |
"should_save": true,
|
| 10017 |
+
"should_training_stop": true
|
| 10018 |
},
|
| 10019 |
"attributes": {}
|
| 10020 |
}
|
| 10021 |
},
|
| 10022 |
+
"total_flos": 1.0917797110480896e+16,
|
| 10023 |
"train_batch_size": 2,
|
| 10024 |
"trial_name": null,
|
| 10025 |
"trial_params": null
|