Training in progress, step 1337, 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 51431872
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:72d65b08fa2983cd66a70d30a82ca8e09f6764ffa304c4e3ae8b38301a5f1bbc
|
| 3 |
size 51431872
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 26550644
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:98d48ec7ca434b1e9f4c4f61e1051da14f1ac514ce2a396d55a0746f4417ef2f
|
| 3 |
size 26550644
|
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 sha256:8d5939e7e158ad21d97a30b2b08f06383171448c8bb26d6523a8b0b1023422b3
|
| 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 sha256:49941719a7d531c5d88ff3914ae840c76014acca884550e614a9dcdd773ae93d
|
| 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": 335,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -7074,6 +7074,2330 @@
|
|
| 7074 |
"eval_samples_per_second": 38.95,
|
| 7075 |
"eval_steps_per_second": 19.475,
|
| 7076 |
"step": 1005
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 7077 |
}
|
| 7078 |
],
|
| 7079 |
"logging_steps": 1,
|
|
@@ -7088,12 +9412,12 @@
|
|
| 7088 |
"should_evaluate": false,
|
| 7089 |
"should_log": false,
|
| 7090 |
"should_save": true,
|
| 7091 |
-
"should_training_stop":
|
| 7092 |
},
|
| 7093 |
"attributes": {}
|
| 7094 |
}
|
| 7095 |
},
|
| 7096 |
-
"total_flos":
|
| 7097 |
"train_batch_size": 2,
|
| 7098 |
"trial_name": null,
|
| 7099 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.004443325581607889,
|
| 5 |
"eval_steps": 335,
|
| 6 |
+
"global_step": 1337,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 7074 |
"eval_samples_per_second": 38.95,
|
| 7075 |
"eval_steps_per_second": 19.475,
|
| 7076 |
"step": 1005
|
| 7077 |
+
},
|
| 7078 |
+
{
|
| 7079 |
+
"epoch": 0.0033432950898261305,
|
| 7080 |
+
"grad_norm": 1.9887334108352661,
|
| 7081 |
+
"learning_rate": 2.9163880954689228e-05,
|
| 7082 |
+
"loss": 0.8261,
|
| 7083 |
+
"step": 1006
|
| 7084 |
+
},
|
| 7085 |
+
{
|
| 7086 |
+
"epoch": 0.003346618444786196,
|
| 7087 |
+
"grad_norm": 2.993713140487671,
|
| 7088 |
+
"learning_rate": 2.8996973846694642e-05,
|
| 7089 |
+
"loss": 0.6206,
|
| 7090 |
+
"step": 1007
|
| 7091 |
+
},
|
| 7092 |
+
{
|
| 7093 |
+
"epoch": 0.0033499417997462618,
|
| 7094 |
+
"grad_norm": 1.2360352277755737,
|
| 7095 |
+
"learning_rate": 2.8830464694240634e-05,
|
| 7096 |
+
"loss": 1.2049,
|
| 7097 |
+
"step": 1008
|
| 7098 |
+
},
|
| 7099 |
+
{
|
| 7100 |
+
"epoch": 0.0033532651547063274,
|
| 7101 |
+
"grad_norm": 2.6683409214019775,
|
| 7102 |
+
"learning_rate": 2.8664354430572492e-05,
|
| 7103 |
+
"loss": 0.8874,
|
| 7104 |
+
"step": 1009
|
| 7105 |
+
},
|
| 7106 |
+
{
|
| 7107 |
+
"epoch": 0.0033565885096663935,
|
| 7108 |
+
"grad_norm": 1.849593162536621,
|
| 7109 |
+
"learning_rate": 2.8498643986699803e-05,
|
| 7110 |
+
"loss": 1.0419,
|
| 7111 |
+
"step": 1010
|
| 7112 |
+
},
|
| 7113 |
+
{
|
| 7114 |
+
"epoch": 0.003359911864626459,
|
| 7115 |
+
"grad_norm": 1.955989956855774,
|
| 7116 |
+
"learning_rate": 2.8333334291391323e-05,
|
| 7117 |
+
"loss": 1.3201,
|
| 7118 |
+
"step": 1011
|
| 7119 |
+
},
|
| 7120 |
+
{
|
| 7121 |
+
"epoch": 0.0033632352195865248,
|
| 7122 |
+
"grad_norm": 1.4712536334991455,
|
| 7123 |
+
"learning_rate": 2.8168426271169623e-05,
|
| 7124 |
+
"loss": 0.8732,
|
| 7125 |
+
"step": 1012
|
| 7126 |
+
},
|
| 7127 |
+
{
|
| 7128 |
+
"epoch": 0.0033665585745465904,
|
| 7129 |
+
"grad_norm": 1.8448526859283447,
|
| 7130 |
+
"learning_rate": 2.8003920850306085e-05,
|
| 7131 |
+
"loss": 1.313,
|
| 7132 |
+
"step": 1013
|
| 7133 |
+
},
|
| 7134 |
+
{
|
| 7135 |
+
"epoch": 0.0033698819295066565,
|
| 7136 |
+
"grad_norm": 2.3865597248077393,
|
| 7137 |
+
"learning_rate": 2.783981895081549e-05,
|
| 7138 |
+
"loss": 1.0349,
|
| 7139 |
+
"step": 1014
|
| 7140 |
+
},
|
| 7141 |
+
{
|
| 7142 |
+
"epoch": 0.003373205284466722,
|
| 7143 |
+
"grad_norm": 3.1011106967926025,
|
| 7144 |
+
"learning_rate": 2.767612149245099e-05,
|
| 7145 |
+
"loss": 1.1726,
|
| 7146 |
+
"step": 1015
|
| 7147 |
+
},
|
| 7148 |
+
{
|
| 7149 |
+
"epoch": 0.0033765286394267877,
|
| 7150 |
+
"grad_norm": 1.5871694087982178,
|
| 7151 |
+
"learning_rate": 2.751282939269908e-05,
|
| 7152 |
+
"loss": 0.7144,
|
| 7153 |
+
"step": 1016
|
| 7154 |
+
},
|
| 7155 |
+
{
|
| 7156 |
+
"epoch": 0.0033798519943868534,
|
| 7157 |
+
"grad_norm": 2.121737480163574,
|
| 7158 |
+
"learning_rate": 2.734994356677416e-05,
|
| 7159 |
+
"loss": 1.1891,
|
| 7160 |
+
"step": 1017
|
| 7161 |
+
},
|
| 7162 |
+
{
|
| 7163 |
+
"epoch": 0.003383175349346919,
|
| 7164 |
+
"grad_norm": 1.8265715837478638,
|
| 7165 |
+
"learning_rate": 2.7187464927613606e-05,
|
| 7166 |
+
"loss": 0.6414,
|
| 7167 |
+
"step": 1018
|
| 7168 |
+
},
|
| 7169 |
+
{
|
| 7170 |
+
"epoch": 0.003386498704306985,
|
| 7171 |
+
"grad_norm": 2.1911258697509766,
|
| 7172 |
+
"learning_rate": 2.702539438587267e-05,
|
| 7173 |
+
"loss": 0.9227,
|
| 7174 |
+
"step": 1019
|
| 7175 |
+
},
|
| 7176 |
+
{
|
| 7177 |
+
"epoch": 0.0033898220592670507,
|
| 7178 |
+
"grad_norm": 2.096592903137207,
|
| 7179 |
+
"learning_rate": 2.6863732849919164e-05,
|
| 7180 |
+
"loss": 0.268,
|
| 7181 |
+
"step": 1020
|
| 7182 |
+
},
|
| 7183 |
+
{
|
| 7184 |
+
"epoch": 0.0033931454142271163,
|
| 7185 |
+
"grad_norm": 1.7348281145095825,
|
| 7186 |
+
"learning_rate": 2.670248122582869e-05,
|
| 7187 |
+
"loss": 0.9092,
|
| 7188 |
+
"step": 1021
|
| 7189 |
+
},
|
| 7190 |
+
{
|
| 7191 |
+
"epoch": 0.003396468769187182,
|
| 7192 |
+
"grad_norm": 2.108294725418091,
|
| 7193 |
+
"learning_rate": 2.6541640417379276e-05,
|
| 7194 |
+
"loss": 0.98,
|
| 7195 |
+
"step": 1022
|
| 7196 |
+
},
|
| 7197 |
+
{
|
| 7198 |
+
"epoch": 0.003399792124147248,
|
| 7199 |
+
"grad_norm": 2.166422128677368,
|
| 7200 |
+
"learning_rate": 2.6381211326046428e-05,
|
| 7201 |
+
"loss": 1.0629,
|
| 7202 |
+
"step": 1023
|
| 7203 |
+
},
|
| 7204 |
+
{
|
| 7205 |
+
"epoch": 0.0034031154791073137,
|
| 7206 |
+
"grad_norm": 1.3211616277694702,
|
| 7207 |
+
"learning_rate": 2.6221194850998155e-05,
|
| 7208 |
+
"loss": 1.1247,
|
| 7209 |
+
"step": 1024
|
| 7210 |
+
},
|
| 7211 |
+
{
|
| 7212 |
+
"epoch": 0.0034064388340673793,
|
| 7213 |
+
"grad_norm": 1.4853568077087402,
|
| 7214 |
+
"learning_rate": 2.6061591889089776e-05,
|
| 7215 |
+
"loss": 0.3974,
|
| 7216 |
+
"step": 1025
|
| 7217 |
+
},
|
| 7218 |
+
{
|
| 7219 |
+
"epoch": 0.003409762189027445,
|
| 7220 |
+
"grad_norm": 1.7520192861557007,
|
| 7221 |
+
"learning_rate": 2.590240333485897e-05,
|
| 7222 |
+
"loss": 0.8057,
|
| 7223 |
+
"step": 1026
|
| 7224 |
+
},
|
| 7225 |
+
{
|
| 7226 |
+
"epoch": 0.003413085543987511,
|
| 7227 |
+
"grad_norm": 2.130795955657959,
|
| 7228 |
+
"learning_rate": 2.5743630080520765e-05,
|
| 7229 |
+
"loss": 0.8887,
|
| 7230 |
+
"step": 1027
|
| 7231 |
+
},
|
| 7232 |
+
{
|
| 7233 |
+
"epoch": 0.0034164088989475767,
|
| 7234 |
+
"grad_norm": 2.033766269683838,
|
| 7235 |
+
"learning_rate": 2.558527301596251e-05,
|
| 7236 |
+
"loss": 0.6583,
|
| 7237 |
+
"step": 1028
|
| 7238 |
+
},
|
| 7239 |
+
{
|
| 7240 |
+
"epoch": 0.0034197322539076423,
|
| 7241 |
+
"grad_norm": 1.8105857372283936,
|
| 7242 |
+
"learning_rate": 2.5427333028738988e-05,
|
| 7243 |
+
"loss": 0.4101,
|
| 7244 |
+
"step": 1029
|
| 7245 |
+
},
|
| 7246 |
+
{
|
| 7247 |
+
"epoch": 0.003423055608867708,
|
| 7248 |
+
"grad_norm": 2.780566930770874,
|
| 7249 |
+
"learning_rate": 2.5269811004067256e-05,
|
| 7250 |
+
"loss": 0.8233,
|
| 7251 |
+
"step": 1030
|
| 7252 |
+
},
|
| 7253 |
+
{
|
| 7254 |
+
"epoch": 0.003426378963827774,
|
| 7255 |
+
"grad_norm": 1.9028615951538086,
|
| 7256 |
+
"learning_rate": 2.5112707824821846e-05,
|
| 7257 |
+
"loss": 0.8525,
|
| 7258 |
+
"step": 1031
|
| 7259 |
+
},
|
| 7260 |
+
{
|
| 7261 |
+
"epoch": 0.0034297023187878396,
|
| 7262 |
+
"grad_norm": 3.4555652141571045,
|
| 7263 |
+
"learning_rate": 2.495602437152975e-05,
|
| 7264 |
+
"loss": 0.8126,
|
| 7265 |
+
"step": 1032
|
| 7266 |
+
},
|
| 7267 |
+
{
|
| 7268 |
+
"epoch": 0.0034330256737479053,
|
| 7269 |
+
"grad_norm": 1.3704328536987305,
|
| 7270 |
+
"learning_rate": 2.4799761522365438e-05,
|
| 7271 |
+
"loss": 0.7354,
|
| 7272 |
+
"step": 1033
|
| 7273 |
+
},
|
| 7274 |
+
{
|
| 7275 |
+
"epoch": 0.003436349028707971,
|
| 7276 |
+
"grad_norm": 2.7514312267303467,
|
| 7277 |
+
"learning_rate": 2.4643920153146116e-05,
|
| 7278 |
+
"loss": 1.19,
|
| 7279 |
+
"step": 1034
|
| 7280 |
+
},
|
| 7281 |
+
{
|
| 7282 |
+
"epoch": 0.0034396723836680365,
|
| 7283 |
+
"grad_norm": 2.007598638534546,
|
| 7284 |
+
"learning_rate": 2.448850113732658e-05,
|
| 7285 |
+
"loss": 1.0406,
|
| 7286 |
+
"step": 1035
|
| 7287 |
+
},
|
| 7288 |
+
{
|
| 7289 |
+
"epoch": 0.0034429957386281026,
|
| 7290 |
+
"grad_norm": 1.9599459171295166,
|
| 7291 |
+
"learning_rate": 2.4333505345994433e-05,
|
| 7292 |
+
"loss": 1.0742,
|
| 7293 |
+
"step": 1036
|
| 7294 |
+
},
|
| 7295 |
+
{
|
| 7296 |
+
"epoch": 0.0034463190935881682,
|
| 7297 |
+
"grad_norm": 1.525658130645752,
|
| 7298 |
+
"learning_rate": 2.4178933647865344e-05,
|
| 7299 |
+
"loss": 1.0423,
|
| 7300 |
+
"step": 1037
|
| 7301 |
+
},
|
| 7302 |
+
{
|
| 7303 |
+
"epoch": 0.003449642448548234,
|
| 7304 |
+
"grad_norm": 2.253969430923462,
|
| 7305 |
+
"learning_rate": 2.4024786909277806e-05,
|
| 7306 |
+
"loss": 0.7478,
|
| 7307 |
+
"step": 1038
|
| 7308 |
+
},
|
| 7309 |
+
{
|
| 7310 |
+
"epoch": 0.0034529658035082995,
|
| 7311 |
+
"grad_norm": 2.9593136310577393,
|
| 7312 |
+
"learning_rate": 2.3871065994188723e-05,
|
| 7313 |
+
"loss": 0.9221,
|
| 7314 |
+
"step": 1039
|
| 7315 |
+
},
|
| 7316 |
+
{
|
| 7317 |
+
"epoch": 0.0034562891584683656,
|
| 7318 |
+
"grad_norm": 1.6113935708999634,
|
| 7319 |
+
"learning_rate": 2.3717771764168262e-05,
|
| 7320 |
+
"loss": 1.0702,
|
| 7321 |
+
"step": 1040
|
| 7322 |
+
},
|
| 7323 |
+
{
|
| 7324 |
+
"epoch": 0.003459612513428431,
|
| 7325 |
+
"grad_norm": 1.477838397026062,
|
| 7326 |
+
"learning_rate": 2.3564905078395073e-05,
|
| 7327 |
+
"loss": 0.8423,
|
| 7328 |
+
"step": 1041
|
| 7329 |
+
},
|
| 7330 |
+
{
|
| 7331 |
+
"epoch": 0.003462935868388497,
|
| 7332 |
+
"grad_norm": 1.8534512519836426,
|
| 7333 |
+
"learning_rate": 2.3412466793651654e-05,
|
| 7334 |
+
"loss": 0.6136,
|
| 7335 |
+
"step": 1042
|
| 7336 |
+
},
|
| 7337 |
+
{
|
| 7338 |
+
"epoch": 0.0034662592233485625,
|
| 7339 |
+
"grad_norm": 1.2977598905563354,
|
| 7340 |
+
"learning_rate": 2.32604577643193e-05,
|
| 7341 |
+
"loss": 1.0533,
|
| 7342 |
+
"step": 1043
|
| 7343 |
+
},
|
| 7344 |
+
{
|
| 7345 |
+
"epoch": 0.0034695825783086286,
|
| 7346 |
+
"grad_norm": 1.402443289756775,
|
| 7347 |
+
"learning_rate": 2.310887884237346e-05,
|
| 7348 |
+
"loss": 0.6865,
|
| 7349 |
+
"step": 1044
|
| 7350 |
+
},
|
| 7351 |
+
{
|
| 7352 |
+
"epoch": 0.003472905933268694,
|
| 7353 |
+
"grad_norm": 1.5745853185653687,
|
| 7354 |
+
"learning_rate": 2.2957730877378947e-05,
|
| 7355 |
+
"loss": 0.8526,
|
| 7356 |
+
"step": 1045
|
| 7357 |
+
},
|
| 7358 |
+
{
|
| 7359 |
+
"epoch": 0.00347622928822876,
|
| 7360 |
+
"grad_norm": 2.4579102993011475,
|
| 7361 |
+
"learning_rate": 2.280701471648512e-05,
|
| 7362 |
+
"loss": 0.5886,
|
| 7363 |
+
"step": 1046
|
| 7364 |
+
},
|
| 7365 |
+
{
|
| 7366 |
+
"epoch": 0.0034795526431888255,
|
| 7367 |
+
"grad_norm": 2.511986017227173,
|
| 7368 |
+
"learning_rate": 2.2656731204421255e-05,
|
| 7369 |
+
"loss": 1.0093,
|
| 7370 |
+
"step": 1047
|
| 7371 |
+
},
|
| 7372 |
+
{
|
| 7373 |
+
"epoch": 0.003482875998148891,
|
| 7374 |
+
"grad_norm": 1.5310604572296143,
|
| 7375 |
+
"learning_rate": 2.2506881183491647e-05,
|
| 7376 |
+
"loss": 0.5451,
|
| 7377 |
+
"step": 1048
|
| 7378 |
+
},
|
| 7379 |
+
{
|
| 7380 |
+
"epoch": 0.003486199353108957,
|
| 7381 |
+
"grad_norm": 2.594639539718628,
|
| 7382 |
+
"learning_rate": 2.2357465493571016e-05,
|
| 7383 |
+
"loss": 1.3524,
|
| 7384 |
+
"step": 1049
|
| 7385 |
+
},
|
| 7386 |
+
{
|
| 7387 |
+
"epoch": 0.003489522708069023,
|
| 7388 |
+
"grad_norm": 2.8036839962005615,
|
| 7389 |
+
"learning_rate": 2.2208484972099743e-05,
|
| 7390 |
+
"loss": 0.8427,
|
| 7391 |
+
"step": 1050
|
| 7392 |
+
},
|
| 7393 |
+
{
|
| 7394 |
+
"epoch": 0.0034928460630290884,
|
| 7395 |
+
"grad_norm": 1.729481816291809,
|
| 7396 |
+
"learning_rate": 2.2059940454079175e-05,
|
| 7397 |
+
"loss": 0.5532,
|
| 7398 |
+
"step": 1051
|
| 7399 |
+
},
|
| 7400 |
+
{
|
| 7401 |
+
"epoch": 0.003496169417989154,
|
| 7402 |
+
"grad_norm": 1.573091745376587,
|
| 7403 |
+
"learning_rate": 2.191183277206703e-05,
|
| 7404 |
+
"loss": 0.5843,
|
| 7405 |
+
"step": 1052
|
| 7406 |
+
},
|
| 7407 |
+
{
|
| 7408 |
+
"epoch": 0.00349949277294922,
|
| 7409 |
+
"grad_norm": 2.2913410663604736,
|
| 7410 |
+
"learning_rate": 2.176416275617259e-05,
|
| 7411 |
+
"loss": 0.6285,
|
| 7412 |
+
"step": 1053
|
| 7413 |
+
},
|
| 7414 |
+
{
|
| 7415 |
+
"epoch": 0.0035028161279092858,
|
| 7416 |
+
"grad_norm": 2.0742692947387695,
|
| 7417 |
+
"learning_rate": 2.1616931234052108e-05,
|
| 7418 |
+
"loss": 1.4509,
|
| 7419 |
+
"step": 1054
|
| 7420 |
+
},
|
| 7421 |
+
{
|
| 7422 |
+
"epoch": 0.0035061394828693514,
|
| 7423 |
+
"grad_norm": 2.3040852546691895,
|
| 7424 |
+
"learning_rate": 2.1470139030904312e-05,
|
| 7425 |
+
"loss": 0.8464,
|
| 7426 |
+
"step": 1055
|
| 7427 |
+
},
|
| 7428 |
+
{
|
| 7429 |
+
"epoch": 0.003509462837829417,
|
| 7430 |
+
"grad_norm": 2.0199368000030518,
|
| 7431 |
+
"learning_rate": 2.132378696946542e-05,
|
| 7432 |
+
"loss": 1.0534,
|
| 7433 |
+
"step": 1056
|
| 7434 |
+
},
|
| 7435 |
+
{
|
| 7436 |
+
"epoch": 0.003512786192789483,
|
| 7437 |
+
"grad_norm": 2.384861469268799,
|
| 7438 |
+
"learning_rate": 2.1177875870004993e-05,
|
| 7439 |
+
"loss": 1.1861,
|
| 7440 |
+
"step": 1057
|
| 7441 |
+
},
|
| 7442 |
+
{
|
| 7443 |
+
"epoch": 0.0035161095477495487,
|
| 7444 |
+
"grad_norm": 1.7507251501083374,
|
| 7445 |
+
"learning_rate": 2.103240655032095e-05,
|
| 7446 |
+
"loss": 0.6772,
|
| 7447 |
+
"step": 1058
|
| 7448 |
+
},
|
| 7449 |
+
{
|
| 7450 |
+
"epoch": 0.0035194329027096144,
|
| 7451 |
+
"grad_norm": 1.9110218286514282,
|
| 7452 |
+
"learning_rate": 2.0887379825735176e-05,
|
| 7453 |
+
"loss": 0.4735,
|
| 7454 |
+
"step": 1059
|
| 7455 |
+
},
|
| 7456 |
+
{
|
| 7457 |
+
"epoch": 0.00352275625766968,
|
| 7458 |
+
"grad_norm": 2.0105197429656982,
|
| 7459 |
+
"learning_rate": 2.074279650908897e-05,
|
| 7460 |
+
"loss": 0.6666,
|
| 7461 |
+
"step": 1060
|
| 7462 |
+
},
|
| 7463 |
+
{
|
| 7464 |
+
"epoch": 0.0035260796126297457,
|
| 7465 |
+
"grad_norm": 2.125974416732788,
|
| 7466 |
+
"learning_rate": 2.0598657410738343e-05,
|
| 7467 |
+
"loss": 1.151,
|
| 7468 |
+
"step": 1061
|
| 7469 |
+
},
|
| 7470 |
+
{
|
| 7471 |
+
"epoch": 0.0035294029675898117,
|
| 7472 |
+
"grad_norm": 1.5914584398269653,
|
| 7473 |
+
"learning_rate": 2.0454963338549625e-05,
|
| 7474 |
+
"loss": 0.7659,
|
| 7475 |
+
"step": 1062
|
| 7476 |
+
},
|
| 7477 |
+
{
|
| 7478 |
+
"epoch": 0.0035327263225498774,
|
| 7479 |
+
"grad_norm": 2.65944504737854,
|
| 7480 |
+
"learning_rate": 2.0311715097894855e-05,
|
| 7481 |
+
"loss": 0.6658,
|
| 7482 |
+
"step": 1063
|
| 7483 |
+
},
|
| 7484 |
+
{
|
| 7485 |
+
"epoch": 0.003536049677509943,
|
| 7486 |
+
"grad_norm": 1.8550827503204346,
|
| 7487 |
+
"learning_rate": 2.0168913491647255e-05,
|
| 7488 |
+
"loss": 1.1008,
|
| 7489 |
+
"step": 1064
|
| 7490 |
+
},
|
| 7491 |
+
{
|
| 7492 |
+
"epoch": 0.0035393730324700086,
|
| 7493 |
+
"grad_norm": 1.9186201095581055,
|
| 7494 |
+
"learning_rate": 2.0026559320176875e-05,
|
| 7495 |
+
"loss": 1.0043,
|
| 7496 |
+
"step": 1065
|
| 7497 |
+
},
|
| 7498 |
+
{
|
| 7499 |
+
"epoch": 0.0035426963874300747,
|
| 7500 |
+
"grad_norm": 2.0259082317352295,
|
| 7501 |
+
"learning_rate": 1.9884653381345875e-05,
|
| 7502 |
+
"loss": 1.2005,
|
| 7503 |
+
"step": 1066
|
| 7504 |
+
},
|
| 7505 |
+
{
|
| 7506 |
+
"epoch": 0.0035460197423901403,
|
| 7507 |
+
"grad_norm": 2.626133680343628,
|
| 7508 |
+
"learning_rate": 1.9743196470504234e-05,
|
| 7509 |
+
"loss": 0.9398,
|
| 7510 |
+
"step": 1067
|
| 7511 |
+
},
|
| 7512 |
+
{
|
| 7513 |
+
"epoch": 0.003549343097350206,
|
| 7514 |
+
"grad_norm": 2.803363561630249,
|
| 7515 |
+
"learning_rate": 1.9602189380485214e-05,
|
| 7516 |
+
"loss": 0.7764,
|
| 7517 |
+
"step": 1068
|
| 7518 |
+
},
|
| 7519 |
+
{
|
| 7520 |
+
"epoch": 0.0035526664523102716,
|
| 7521 |
+
"grad_norm": 2.6616218090057373,
|
| 7522 |
+
"learning_rate": 1.9461632901600935e-05,
|
| 7523 |
+
"loss": 0.867,
|
| 7524 |
+
"step": 1069
|
| 7525 |
+
},
|
| 7526 |
+
{
|
| 7527 |
+
"epoch": 0.0035559898072703377,
|
| 7528 |
+
"grad_norm": 1.805355191230774,
|
| 7529 |
+
"learning_rate": 1.9321527821637987e-05,
|
| 7530 |
+
"loss": 0.9621,
|
| 7531 |
+
"step": 1070
|
| 7532 |
+
},
|
| 7533 |
+
{
|
| 7534 |
+
"epoch": 0.0035593131622304033,
|
| 7535 |
+
"grad_norm": 1.9301918745040894,
|
| 7536 |
+
"learning_rate": 1.9181874925852926e-05,
|
| 7537 |
+
"loss": 1.1156,
|
| 7538 |
+
"step": 1071
|
| 7539 |
+
},
|
| 7540 |
+
{
|
| 7541 |
+
"epoch": 0.003562636517190469,
|
| 7542 |
+
"grad_norm": 1.2671960592269897,
|
| 7543 |
+
"learning_rate": 1.904267499696791e-05,
|
| 7544 |
+
"loss": 1.0542,
|
| 7545 |
+
"step": 1072
|
| 7546 |
+
},
|
| 7547 |
+
{
|
| 7548 |
+
"epoch": 0.0035659598721505346,
|
| 7549 |
+
"grad_norm": 1.832224726676941,
|
| 7550 |
+
"learning_rate": 1.8903928815166426e-05,
|
| 7551 |
+
"loss": 0.9923,
|
| 7552 |
+
"step": 1073
|
| 7553 |
+
},
|
| 7554 |
+
{
|
| 7555 |
+
"epoch": 0.0035692832271106,
|
| 7556 |
+
"grad_norm": 2.3667545318603516,
|
| 7557 |
+
"learning_rate": 1.8765637158088623e-05,
|
| 7558 |
+
"loss": 0.7825,
|
| 7559 |
+
"step": 1074
|
| 7560 |
+
},
|
| 7561 |
+
{
|
| 7562 |
+
"epoch": 0.0035726065820706663,
|
| 7563 |
+
"grad_norm": 1.8500324487686157,
|
| 7564 |
+
"learning_rate": 1.8627800800827332e-05,
|
| 7565 |
+
"loss": 0.4897,
|
| 7566 |
+
"step": 1075
|
| 7567 |
+
},
|
| 7568 |
+
{
|
| 7569 |
+
"epoch": 0.003575929937030732,
|
| 7570 |
+
"grad_norm": 2.8491454124450684,
|
| 7571 |
+
"learning_rate": 1.8490420515923445e-05,
|
| 7572 |
+
"loss": 1.0215,
|
| 7573 |
+
"step": 1076
|
| 7574 |
+
},
|
| 7575 |
+
{
|
| 7576 |
+
"epoch": 0.0035792532919907975,
|
| 7577 |
+
"grad_norm": 2.6926255226135254,
|
| 7578 |
+
"learning_rate": 1.8353497073361647e-05,
|
| 7579 |
+
"loss": 0.8692,
|
| 7580 |
+
"step": 1077
|
| 7581 |
+
},
|
| 7582 |
+
{
|
| 7583 |
+
"epoch": 0.003582576646950863,
|
| 7584 |
+
"grad_norm": 3.2963078022003174,
|
| 7585 |
+
"learning_rate": 1.821703124056623e-05,
|
| 7586 |
+
"loss": 1.2669,
|
| 7587 |
+
"step": 1078
|
| 7588 |
+
},
|
| 7589 |
+
{
|
| 7590 |
+
"epoch": 0.0035859000019109293,
|
| 7591 |
+
"grad_norm": 1.8386890888214111,
|
| 7592 |
+
"learning_rate": 1.808102378239659e-05,
|
| 7593 |
+
"loss": 1.0086,
|
| 7594 |
+
"step": 1079
|
| 7595 |
+
},
|
| 7596 |
+
{
|
| 7597 |
+
"epoch": 0.003589223356870995,
|
| 7598 |
+
"grad_norm": 2.384105920791626,
|
| 7599 |
+
"learning_rate": 1.794547546114308e-05,
|
| 7600 |
+
"loss": 1.6574,
|
| 7601 |
+
"step": 1080
|
| 7602 |
+
},
|
| 7603 |
+
{
|
| 7604 |
+
"epoch": 0.0035925467118310605,
|
| 7605 |
+
"grad_norm": 2.209704875946045,
|
| 7606 |
+
"learning_rate": 1.78103870365227e-05,
|
| 7607 |
+
"loss": 1.2232,
|
| 7608 |
+
"step": 1081
|
| 7609 |
+
},
|
| 7610 |
+
{
|
| 7611 |
+
"epoch": 0.003595870066791126,
|
| 7612 |
+
"grad_norm": 2.2206103801727295,
|
| 7613 |
+
"learning_rate": 1.7675759265674797e-05,
|
| 7614 |
+
"loss": 1.0329,
|
| 7615 |
+
"step": 1082
|
| 7616 |
+
},
|
| 7617 |
+
{
|
| 7618 |
+
"epoch": 0.0035991934217511922,
|
| 7619 |
+
"grad_norm": 1.4707410335540771,
|
| 7620 |
+
"learning_rate": 1.7541592903156933e-05,
|
| 7621 |
+
"loss": 0.5543,
|
| 7622 |
+
"step": 1083
|
| 7623 |
+
},
|
| 7624 |
+
{
|
| 7625 |
+
"epoch": 0.003602516776711258,
|
| 7626 |
+
"grad_norm": 1.7666308879852295,
|
| 7627 |
+
"learning_rate": 1.7407888700940523e-05,
|
| 7628 |
+
"loss": 0.6173,
|
| 7629 |
+
"step": 1084
|
| 7630 |
+
},
|
| 7631 |
+
{
|
| 7632 |
+
"epoch": 0.0036058401316713235,
|
| 7633 |
+
"grad_norm": 3.515272378921509,
|
| 7634 |
+
"learning_rate": 1.7274647408406698e-05,
|
| 7635 |
+
"loss": 0.7079,
|
| 7636 |
+
"step": 1085
|
| 7637 |
+
},
|
| 7638 |
+
{
|
| 7639 |
+
"epoch": 0.003609163486631389,
|
| 7640 |
+
"grad_norm": 1.8757985830307007,
|
| 7641 |
+
"learning_rate": 1.7141869772342088e-05,
|
| 7642 |
+
"loss": 0.6742,
|
| 7643 |
+
"step": 1086
|
| 7644 |
+
},
|
| 7645 |
+
{
|
| 7646 |
+
"epoch": 0.003612486841591455,
|
| 7647 |
+
"grad_norm": 1.7394914627075195,
|
| 7648 |
+
"learning_rate": 1.7009556536934602e-05,
|
| 7649 |
+
"loss": 0.9227,
|
| 7650 |
+
"step": 1087
|
| 7651 |
+
},
|
| 7652 |
+
{
|
| 7653 |
+
"epoch": 0.003615810196551521,
|
| 7654 |
+
"grad_norm": 1.7292208671569824,
|
| 7655 |
+
"learning_rate": 1.6877708443769392e-05,
|
| 7656 |
+
"loss": 0.6633,
|
| 7657 |
+
"step": 1088
|
| 7658 |
+
},
|
| 7659 |
+
{
|
| 7660 |
+
"epoch": 0.0036191335515115865,
|
| 7661 |
+
"grad_norm": 2.7111852169036865,
|
| 7662 |
+
"learning_rate": 1.6746326231824495e-05,
|
| 7663 |
+
"loss": 1.4088,
|
| 7664 |
+
"step": 1089
|
| 7665 |
+
},
|
| 7666 |
+
{
|
| 7667 |
+
"epoch": 0.003622456906471652,
|
| 7668 |
+
"grad_norm": 2.8773698806762695,
|
| 7669 |
+
"learning_rate": 1.661541063746679e-05,
|
| 7670 |
+
"loss": 0.9911,
|
| 7671 |
+
"step": 1090
|
| 7672 |
+
},
|
| 7673 |
+
{
|
| 7674 |
+
"epoch": 0.0036257802614317177,
|
| 7675 |
+
"grad_norm": 2.6539015769958496,
|
| 7676 |
+
"learning_rate": 1.648496239444799e-05,
|
| 7677 |
+
"loss": 0.8291,
|
| 7678 |
+
"step": 1091
|
| 7679 |
+
},
|
| 7680 |
+
{
|
| 7681 |
+
"epoch": 0.003629103616391784,
|
| 7682 |
+
"grad_norm": 1.9189116954803467,
|
| 7683 |
+
"learning_rate": 1.635498223390022e-05,
|
| 7684 |
+
"loss": 1.0237,
|
| 7685 |
+
"step": 1092
|
| 7686 |
+
},
|
| 7687 |
+
{
|
| 7688 |
+
"epoch": 0.0036324269713518494,
|
| 7689 |
+
"grad_norm": 1.969369888305664,
|
| 7690 |
+
"learning_rate": 1.6225470884332304e-05,
|
| 7691 |
+
"loss": 1.0304,
|
| 7692 |
+
"step": 1093
|
| 7693 |
+
},
|
| 7694 |
+
{
|
| 7695 |
+
"epoch": 0.003635750326311915,
|
| 7696 |
+
"grad_norm": 2.1338069438934326,
|
| 7697 |
+
"learning_rate": 1.6096429071625374e-05,
|
| 7698 |
+
"loss": 1.0386,
|
| 7699 |
+
"step": 1094
|
| 7700 |
+
},
|
| 7701 |
+
{
|
| 7702 |
+
"epoch": 0.0036390736812719807,
|
| 7703 |
+
"grad_norm": 2.693995714187622,
|
| 7704 |
+
"learning_rate": 1.5967857519028928e-05,
|
| 7705 |
+
"loss": 0.6724,
|
| 7706 |
+
"step": 1095
|
| 7707 |
+
},
|
| 7708 |
+
{
|
| 7709 |
+
"epoch": 0.003642397036232047,
|
| 7710 |
+
"grad_norm": 1.5616436004638672,
|
| 7711 |
+
"learning_rate": 1.5839756947156846e-05,
|
| 7712 |
+
"loss": 1.0765,
|
| 7713 |
+
"step": 1096
|
| 7714 |
+
},
|
| 7715 |
+
{
|
| 7716 |
+
"epoch": 0.0036457203911921124,
|
| 7717 |
+
"grad_norm": 2.3733086585998535,
|
| 7718 |
+
"learning_rate": 1.5712128073983146e-05,
|
| 7719 |
+
"loss": 0.6867,
|
| 7720 |
+
"step": 1097
|
| 7721 |
+
},
|
| 7722 |
+
{
|
| 7723 |
+
"epoch": 0.003649043746152178,
|
| 7724 |
+
"grad_norm": 1.7389389276504517,
|
| 7725 |
+
"learning_rate": 1.5584971614838128e-05,
|
| 7726 |
+
"loss": 0.9531,
|
| 7727 |
+
"step": 1098
|
| 7728 |
+
},
|
| 7729 |
+
{
|
| 7730 |
+
"epoch": 0.0036523671011122437,
|
| 7731 |
+
"grad_norm": 2.9344327449798584,
|
| 7732 |
+
"learning_rate": 1.5458288282404398e-05,
|
| 7733 |
+
"loss": 0.6522,
|
| 7734 |
+
"step": 1099
|
| 7735 |
+
},
|
| 7736 |
+
{
|
| 7737 |
+
"epoch": 0.0036556904560723098,
|
| 7738 |
+
"grad_norm": 2.1339237689971924,
|
| 7739 |
+
"learning_rate": 1.533207878671269e-05,
|
| 7740 |
+
"loss": 1.0749,
|
| 7741 |
+
"step": 1100
|
| 7742 |
+
},
|
| 7743 |
+
{
|
| 7744 |
+
"epoch": 0.0036590138110323754,
|
| 7745 |
+
"grad_norm": 2.975815534591675,
|
| 7746 |
+
"learning_rate": 1.5206343835138092e-05,
|
| 7747 |
+
"loss": 0.4487,
|
| 7748 |
+
"step": 1101
|
| 7749 |
+
},
|
| 7750 |
+
{
|
| 7751 |
+
"epoch": 0.003662337165992441,
|
| 7752 |
+
"grad_norm": 1.2044059038162231,
|
| 7753 |
+
"learning_rate": 1.5081084132395907e-05,
|
| 7754 |
+
"loss": 0.6741,
|
| 7755 |
+
"step": 1102
|
| 7756 |
+
},
|
| 7757 |
+
{
|
| 7758 |
+
"epoch": 0.0036656605209525067,
|
| 7759 |
+
"grad_norm": 1.7217930555343628,
|
| 7760 |
+
"learning_rate": 1.4956300380537747e-05,
|
| 7761 |
+
"loss": 1.0937,
|
| 7762 |
+
"step": 1103
|
| 7763 |
+
},
|
| 7764 |
+
{
|
| 7765 |
+
"epoch": 0.0036689838759125723,
|
| 7766 |
+
"grad_norm": 1.3124263286590576,
|
| 7767 |
+
"learning_rate": 1.4831993278947742e-05,
|
| 7768 |
+
"loss": 0.4858,
|
| 7769 |
+
"step": 1104
|
| 7770 |
+
},
|
| 7771 |
+
{
|
| 7772 |
+
"epoch": 0.0036723072308726384,
|
| 7773 |
+
"grad_norm": 2.2098429203033447,
|
| 7774 |
+
"learning_rate": 1.4708163524338436e-05,
|
| 7775 |
+
"loss": 1.0121,
|
| 7776 |
+
"step": 1105
|
| 7777 |
+
},
|
| 7778 |
+
{
|
| 7779 |
+
"epoch": 0.003675630585832704,
|
| 7780 |
+
"grad_norm": 1.8337208032608032,
|
| 7781 |
+
"learning_rate": 1.4584811810746935e-05,
|
| 7782 |
+
"loss": 1.2821,
|
| 7783 |
+
"step": 1106
|
| 7784 |
+
},
|
| 7785 |
+
{
|
| 7786 |
+
"epoch": 0.0036789539407927696,
|
| 7787 |
+
"grad_norm": 1.871624231338501,
|
| 7788 |
+
"learning_rate": 1.4461938829531107e-05,
|
| 7789 |
+
"loss": 0.6309,
|
| 7790 |
+
"step": 1107
|
| 7791 |
+
},
|
| 7792 |
+
{
|
| 7793 |
+
"epoch": 0.0036822772957528353,
|
| 7794 |
+
"grad_norm": 2.333550453186035,
|
| 7795 |
+
"learning_rate": 1.4339545269365585e-05,
|
| 7796 |
+
"loss": 0.9257,
|
| 7797 |
+
"step": 1108
|
| 7798 |
+
},
|
| 7799 |
+
{
|
| 7800 |
+
"epoch": 0.0036856006507129013,
|
| 7801 |
+
"grad_norm": 1.6626560688018799,
|
| 7802 |
+
"learning_rate": 1.4217631816237952e-05,
|
| 7803 |
+
"loss": 0.5538,
|
| 7804 |
+
"step": 1109
|
| 7805 |
+
},
|
| 7806 |
+
{
|
| 7807 |
+
"epoch": 0.003688924005672967,
|
| 7808 |
+
"grad_norm": 1.7753632068634033,
|
| 7809 |
+
"learning_rate": 1.4096199153444934e-05,
|
| 7810 |
+
"loss": 0.6881,
|
| 7811 |
+
"step": 1110
|
| 7812 |
+
},
|
| 7813 |
+
{
|
| 7814 |
+
"epoch": 0.0036922473606330326,
|
| 7815 |
+
"grad_norm": 2.1906261444091797,
|
| 7816 |
+
"learning_rate": 1.3975247961588478e-05,
|
| 7817 |
+
"loss": 1.2149,
|
| 7818 |
+
"step": 1111
|
| 7819 |
+
},
|
| 7820 |
+
{
|
| 7821 |
+
"epoch": 0.0036955707155930983,
|
| 7822 |
+
"grad_norm": 1.9990803003311157,
|
| 7823 |
+
"learning_rate": 1.385477891857211e-05,
|
| 7824 |
+
"loss": 0.9241,
|
| 7825 |
+
"step": 1112
|
| 7826 |
+
},
|
| 7827 |
+
{
|
| 7828 |
+
"epoch": 0.0036988940705531643,
|
| 7829 |
+
"grad_norm": 1.9631175994873047,
|
| 7830 |
+
"learning_rate": 1.3734792699596921e-05,
|
| 7831 |
+
"loss": 0.9384,
|
| 7832 |
+
"step": 1113
|
| 7833 |
+
},
|
| 7834 |
+
{
|
| 7835 |
+
"epoch": 0.00370221742551323,
|
| 7836 |
+
"grad_norm": 1.601910948753357,
|
| 7837 |
+
"learning_rate": 1.361528997715792e-05,
|
| 7838 |
+
"loss": 1.0611,
|
| 7839 |
+
"step": 1114
|
| 7840 |
+
},
|
| 7841 |
+
{
|
| 7842 |
+
"epoch": 0.0037055407804732956,
|
| 7843 |
+
"grad_norm": 1.8392131328582764,
|
| 7844 |
+
"learning_rate": 1.3496271421040219e-05,
|
| 7845 |
+
"loss": 0.9894,
|
| 7846 |
+
"step": 1115
|
| 7847 |
+
},
|
| 7848 |
+
{
|
| 7849 |
+
"epoch": 0.0037088641354333612,
|
| 7850 |
+
"grad_norm": 3.042781352996826,
|
| 7851 |
+
"learning_rate": 1.3377737698315295e-05,
|
| 7852 |
+
"loss": 0.9245,
|
| 7853 |
+
"step": 1116
|
| 7854 |
+
},
|
| 7855 |
+
{
|
| 7856 |
+
"epoch": 0.003712187490393427,
|
| 7857 |
+
"grad_norm": 2.590090751647949,
|
| 7858 |
+
"learning_rate": 1.3259689473337289e-05,
|
| 7859 |
+
"loss": 1.0378,
|
| 7860 |
+
"step": 1117
|
| 7861 |
+
},
|
| 7862 |
+
{
|
| 7863 |
+
"epoch": 0.003715510845353493,
|
| 7864 |
+
"grad_norm": 2.383646011352539,
|
| 7865 |
+
"learning_rate": 1.3142127407739168e-05,
|
| 7866 |
+
"loss": 0.5865,
|
| 7867 |
+
"step": 1118
|
| 7868 |
+
},
|
| 7869 |
+
{
|
| 7870 |
+
"epoch": 0.0037188342003135586,
|
| 7871 |
+
"grad_norm": 1.675218939781189,
|
| 7872 |
+
"learning_rate": 1.3025052160429118e-05,
|
| 7873 |
+
"loss": 0.442,
|
| 7874 |
+
"step": 1119
|
| 7875 |
+
},
|
| 7876 |
+
{
|
| 7877 |
+
"epoch": 0.003722157555273624,
|
| 7878 |
+
"grad_norm": 2.6803250312805176,
|
| 7879 |
+
"learning_rate": 1.290846438758687e-05,
|
| 7880 |
+
"loss": 1.1026,
|
| 7881 |
+
"step": 1120
|
| 7882 |
+
},
|
| 7883 |
+
{
|
| 7884 |
+
"epoch": 0.00372548091023369,
|
| 7885 |
+
"grad_norm": 2.088174343109131,
|
| 7886 |
+
"learning_rate": 1.2792364742659856e-05,
|
| 7887 |
+
"loss": 1.0888,
|
| 7888 |
+
"step": 1121
|
| 7889 |
+
},
|
| 7890 |
+
{
|
| 7891 |
+
"epoch": 0.003728804265193756,
|
| 7892 |
+
"grad_norm": 2.361299514770508,
|
| 7893 |
+
"learning_rate": 1.2676753876359781e-05,
|
| 7894 |
+
"loss": 0.9856,
|
| 7895 |
+
"step": 1122
|
| 7896 |
+
},
|
| 7897 |
+
{
|
| 7898 |
+
"epoch": 0.0037321276201538215,
|
| 7899 |
+
"grad_norm": 1.9991151094436646,
|
| 7900 |
+
"learning_rate": 1.2561632436658811e-05,
|
| 7901 |
+
"loss": 0.7982,
|
| 7902 |
+
"step": 1123
|
| 7903 |
+
},
|
| 7904 |
+
{
|
| 7905 |
+
"epoch": 0.003735450975113887,
|
| 7906 |
+
"grad_norm": 3.7411093711853027,
|
| 7907 |
+
"learning_rate": 1.2447001068785969e-05,
|
| 7908 |
+
"loss": 1.2467,
|
| 7909 |
+
"step": 1124
|
| 7910 |
+
},
|
| 7911 |
+
{
|
| 7912 |
+
"epoch": 0.003738774330073953,
|
| 7913 |
+
"grad_norm": 2.4900856018066406,
|
| 7914 |
+
"learning_rate": 1.233286041522358e-05,
|
| 7915 |
+
"loss": 0.707,
|
| 7916 |
+
"step": 1125
|
| 7917 |
+
},
|
| 7918 |
+
{
|
| 7919 |
+
"epoch": 0.003742097685034019,
|
| 7920 |
+
"grad_norm": 2.1829333305358887,
|
| 7921 |
+
"learning_rate": 1.2219211115703621e-05,
|
| 7922 |
+
"loss": 0.6891,
|
| 7923 |
+
"step": 1126
|
| 7924 |
+
},
|
| 7925 |
+
{
|
| 7926 |
+
"epoch": 0.0037454210399940845,
|
| 7927 |
+
"grad_norm": 1.6782306432724,
|
| 7928 |
+
"learning_rate": 1.210605380720411e-05,
|
| 7929 |
+
"loss": 1.1077,
|
| 7930 |
+
"step": 1127
|
| 7931 |
+
},
|
| 7932 |
+
{
|
| 7933 |
+
"epoch": 0.00374874439495415,
|
| 7934 |
+
"grad_norm": 2.4409914016723633,
|
| 7935 |
+
"learning_rate": 1.1993389123945597e-05,
|
| 7936 |
+
"loss": 0.8018,
|
| 7937 |
+
"step": 1128
|
| 7938 |
+
},
|
| 7939 |
+
{
|
| 7940 |
+
"epoch": 0.003752067749914216,
|
| 7941 |
+
"grad_norm": 1.8187549114227295,
|
| 7942 |
+
"learning_rate": 1.1881217697387547e-05,
|
| 7943 |
+
"loss": 1.0921,
|
| 7944 |
+
"step": 1129
|
| 7945 |
+
},
|
| 7946 |
+
{
|
| 7947 |
+
"epoch": 0.003755391104874282,
|
| 7948 |
+
"grad_norm": 1.4356580972671509,
|
| 7949 |
+
"learning_rate": 1.1769540156224923e-05,
|
| 7950 |
+
"loss": 0.7464,
|
| 7951 |
+
"step": 1130
|
| 7952 |
+
},
|
| 7953 |
+
{
|
| 7954 |
+
"epoch": 0.0037587144598343475,
|
| 7955 |
+
"grad_norm": 2.073961019515991,
|
| 7956 |
+
"learning_rate": 1.1658357126384479e-05,
|
| 7957 |
+
"loss": 0.6749,
|
| 7958 |
+
"step": 1131
|
| 7959 |
+
},
|
| 7960 |
+
{
|
| 7961 |
+
"epoch": 0.003762037814794413,
|
| 7962 |
+
"grad_norm": 1.284589409828186,
|
| 7963 |
+
"learning_rate": 1.1547669231021395e-05,
|
| 7964 |
+
"loss": 0.4601,
|
| 7965 |
+
"step": 1132
|
| 7966 |
+
},
|
| 7967 |
+
{
|
| 7968 |
+
"epoch": 0.0037653611697544788,
|
| 7969 |
+
"grad_norm": 2.1579086780548096,
|
| 7970 |
+
"learning_rate": 1.1437477090515713e-05,
|
| 7971 |
+
"loss": 1.0621,
|
| 7972 |
+
"step": 1133
|
| 7973 |
+
},
|
| 7974 |
+
{
|
| 7975 |
+
"epoch": 0.0037686845247145444,
|
| 7976 |
+
"grad_norm": 1.9478920698165894,
|
| 7977 |
+
"learning_rate": 1.1327781322468889e-05,
|
| 7978 |
+
"loss": 0.8049,
|
| 7979 |
+
"step": 1134
|
| 7980 |
+
},
|
| 7981 |
+
{
|
| 7982 |
+
"epoch": 0.0037720078796746105,
|
| 7983 |
+
"grad_norm": 1.604015827178955,
|
| 7984 |
+
"learning_rate": 1.1218582541700362e-05,
|
| 7985 |
+
"loss": 1.1651,
|
| 7986 |
+
"step": 1135
|
| 7987 |
+
},
|
| 7988 |
+
{
|
| 7989 |
+
"epoch": 0.003775331234634676,
|
| 7990 |
+
"grad_norm": 2.576106309890747,
|
| 7991 |
+
"learning_rate": 1.1109881360244034e-05,
|
| 7992 |
+
"loss": 1.2095,
|
| 7993 |
+
"step": 1136
|
| 7994 |
+
},
|
| 7995 |
+
{
|
| 7996 |
+
"epoch": 0.0037786545895947417,
|
| 7997 |
+
"grad_norm": 1.4734787940979004,
|
| 7998 |
+
"learning_rate": 1.100167838734486e-05,
|
| 7999 |
+
"loss": 0.8627,
|
| 8000 |
+
"step": 1137
|
| 8001 |
+
},
|
| 8002 |
+
{
|
| 8003 |
+
"epoch": 0.0037819779445548074,
|
| 8004 |
+
"grad_norm": 1.465071439743042,
|
| 8005 |
+
"learning_rate": 1.0893974229455538e-05,
|
| 8006 |
+
"loss": 1.0974,
|
| 8007 |
+
"step": 1138
|
| 8008 |
+
},
|
| 8009 |
+
{
|
| 8010 |
+
"epoch": 0.0037853012995148734,
|
| 8011 |
+
"grad_norm": 1.63118577003479,
|
| 8012 |
+
"learning_rate": 1.0786769490232895e-05,
|
| 8013 |
+
"loss": 1.0472,
|
| 8014 |
+
"step": 1139
|
| 8015 |
+
},
|
| 8016 |
+
{
|
| 8017 |
+
"epoch": 0.003788624654474939,
|
| 8018 |
+
"grad_norm": 1.9454184770584106,
|
| 8019 |
+
"learning_rate": 1.0680064770534748e-05,
|
| 8020 |
+
"loss": 1.0103,
|
| 8021 |
+
"step": 1140
|
| 8022 |
+
},
|
| 8023 |
+
{
|
| 8024 |
+
"epoch": 0.0037919480094350047,
|
| 8025 |
+
"grad_norm": 1.2677627801895142,
|
| 8026 |
+
"learning_rate": 1.0573860668416358e-05,
|
| 8027 |
+
"loss": 0.7591,
|
| 8028 |
+
"step": 1141
|
| 8029 |
+
},
|
| 8030 |
+
{
|
| 8031 |
+
"epoch": 0.0037952713643950703,
|
| 8032 |
+
"grad_norm": 1.8168962001800537,
|
| 8033 |
+
"learning_rate": 1.046815777912713e-05,
|
| 8034 |
+
"loss": 0.3897,
|
| 8035 |
+
"step": 1142
|
| 8036 |
+
},
|
| 8037 |
+
{
|
| 8038 |
+
"epoch": 0.0037985947193551364,
|
| 8039 |
+
"grad_norm": 2.0347893238067627,
|
| 8040 |
+
"learning_rate": 1.0362956695107361e-05,
|
| 8041 |
+
"loss": 0.9761,
|
| 8042 |
+
"step": 1143
|
| 8043 |
+
},
|
| 8044 |
+
{
|
| 8045 |
+
"epoch": 0.003801918074315202,
|
| 8046 |
+
"grad_norm": 1.2987099885940552,
|
| 8047 |
+
"learning_rate": 1.0258258005984777e-05,
|
| 8048 |
+
"loss": 0.7144,
|
| 8049 |
+
"step": 1144
|
| 8050 |
+
},
|
| 8051 |
+
{
|
| 8052 |
+
"epoch": 0.0038052414292752677,
|
| 8053 |
+
"grad_norm": 2.079082489013672,
|
| 8054 |
+
"learning_rate": 1.015406229857132e-05,
|
| 8055 |
+
"loss": 0.8902,
|
| 8056 |
+
"step": 1145
|
| 8057 |
+
},
|
| 8058 |
+
{
|
| 8059 |
+
"epoch": 0.0038085647842353333,
|
| 8060 |
+
"grad_norm": 1.4583004713058472,
|
| 8061 |
+
"learning_rate": 1.0050370156859811e-05,
|
| 8062 |
+
"loss": 0.8563,
|
| 8063 |
+
"step": 1146
|
| 8064 |
+
},
|
| 8065 |
+
{
|
| 8066 |
+
"epoch": 0.003811888139195399,
|
| 8067 |
+
"grad_norm": 2.9463703632354736,
|
| 8068 |
+
"learning_rate": 9.947182162020729e-06,
|
| 8069 |
+
"loss": 0.5976,
|
| 8070 |
+
"step": 1147
|
| 8071 |
+
},
|
| 8072 |
+
{
|
| 8073 |
+
"epoch": 0.003815211494155465,
|
| 8074 |
+
"grad_norm": 1.276196002960205,
|
| 8075 |
+
"learning_rate": 9.844498892398946e-06,
|
| 8076 |
+
"loss": 1.2324,
|
| 8077 |
+
"step": 1148
|
| 8078 |
+
},
|
| 8079 |
+
{
|
| 8080 |
+
"epoch": 0.0038185348491155307,
|
| 8081 |
+
"grad_norm": 2.1213057041168213,
|
| 8082 |
+
"learning_rate": 9.742320923510428e-06,
|
| 8083 |
+
"loss": 1.2369,
|
| 8084 |
+
"step": 1149
|
| 8085 |
+
},
|
| 8086 |
+
{
|
| 8087 |
+
"epoch": 0.0038218582040755963,
|
| 8088 |
+
"grad_norm": 2.704526424407959,
|
| 8089 |
+
"learning_rate": 9.640648828039045e-06,
|
| 8090 |
+
"loss": 0.9242,
|
| 8091 |
+
"step": 1150
|
| 8092 |
+
},
|
| 8093 |
+
{
|
| 8094 |
+
"epoch": 0.003825181559035662,
|
| 8095 |
+
"grad_norm": 1.3100472688674927,
|
| 8096 |
+
"learning_rate": 9.539483175833397e-06,
|
| 8097 |
+
"loss": 0.6016,
|
| 8098 |
+
"step": 1151
|
| 8099 |
+
},
|
| 8100 |
+
{
|
| 8101 |
+
"epoch": 0.003828504913995728,
|
| 8102 |
+
"grad_norm": 2.6044392585754395,
|
| 8103 |
+
"learning_rate": 9.438824533903545e-06,
|
| 8104 |
+
"loss": 0.6004,
|
| 8105 |
+
"step": 1152
|
| 8106 |
+
},
|
| 8107 |
+
{
|
| 8108 |
+
"epoch": 0.0038318282689557936,
|
| 8109 |
+
"grad_norm": 1.9593836069107056,
|
| 8110 |
+
"learning_rate": 9.338673466417924e-06,
|
| 8111 |
+
"loss": 1.0782,
|
| 8112 |
+
"step": 1153
|
| 8113 |
+
},
|
| 8114 |
+
{
|
| 8115 |
+
"epoch": 0.0038351516239158593,
|
| 8116 |
+
"grad_norm": 2.4404966831207275,
|
| 8117 |
+
"learning_rate": 9.2390305347001e-06,
|
| 8118 |
+
"loss": 0.9982,
|
| 8119 |
+
"step": 1154
|
| 8120 |
+
},
|
| 8121 |
+
{
|
| 8122 |
+
"epoch": 0.003838474978875925,
|
| 8123 |
+
"grad_norm": 1.8588268756866455,
|
| 8124 |
+
"learning_rate": 9.139896297225626e-06,
|
| 8125 |
+
"loss": 0.65,
|
| 8126 |
+
"step": 1155
|
| 8127 |
+
},
|
| 8128 |
+
{
|
| 8129 |
+
"epoch": 0.003841798333835991,
|
| 8130 |
+
"grad_norm": 1.6673916578292847,
|
| 8131 |
+
"learning_rate": 9.041271309619048e-06,
|
| 8132 |
+
"loss": 1.3098,
|
| 8133 |
+
"step": 1156
|
| 8134 |
+
},
|
| 8135 |
+
{
|
| 8136 |
+
"epoch": 0.0038451216887960566,
|
| 8137 |
+
"grad_norm": 2.00077223777771,
|
| 8138 |
+
"learning_rate": 8.943156124650531e-06,
|
| 8139 |
+
"loss": 1.1602,
|
| 8140 |
+
"step": 1157
|
| 8141 |
+
},
|
| 8142 |
+
{
|
| 8143 |
+
"epoch": 0.0038484450437561222,
|
| 8144 |
+
"grad_norm": 3.087786912918091,
|
| 8145 |
+
"learning_rate": 8.845551292233045e-06,
|
| 8146 |
+
"loss": 0.9981,
|
| 8147 |
+
"step": 1158
|
| 8148 |
+
},
|
| 8149 |
+
{
|
| 8150 |
+
"epoch": 0.003851768398716188,
|
| 8151 |
+
"grad_norm": 1.7945916652679443,
|
| 8152 |
+
"learning_rate": 8.748457359419093e-06,
|
| 8153 |
+
"loss": 0.9974,
|
| 8154 |
+
"step": 1159
|
| 8155 |
+
},
|
| 8156 |
+
{
|
| 8157 |
+
"epoch": 0.0038550917536762535,
|
| 8158 |
+
"grad_norm": 2.521268129348755,
|
| 8159 |
+
"learning_rate": 8.651874870397692e-06,
|
| 8160 |
+
"loss": 0.9116,
|
| 8161 |
+
"step": 1160
|
| 8162 |
+
},
|
| 8163 |
+
{
|
| 8164 |
+
"epoch": 0.0038584151086363196,
|
| 8165 |
+
"grad_norm": 2.505439281463623,
|
| 8166 |
+
"learning_rate": 8.555804366491405e-06,
|
| 8167 |
+
"loss": 0.9118,
|
| 8168 |
+
"step": 1161
|
| 8169 |
+
},
|
| 8170 |
+
{
|
| 8171 |
+
"epoch": 0.003861738463596385,
|
| 8172 |
+
"grad_norm": 2.009557008743286,
|
| 8173 |
+
"learning_rate": 8.460246386153147e-06,
|
| 8174 |
+
"loss": 0.6388,
|
| 8175 |
+
"step": 1162
|
| 8176 |
+
},
|
| 8177 |
+
{
|
| 8178 |
+
"epoch": 0.003865061818556451,
|
| 8179 |
+
"grad_norm": 3.128729820251465,
|
| 8180 |
+
"learning_rate": 8.365201464963302e-06,
|
| 8181 |
+
"loss": 1.0031,
|
| 8182 |
+
"step": 1163
|
| 8183 |
+
},
|
| 8184 |
+
{
|
| 8185 |
+
"epoch": 0.0038683851735165165,
|
| 8186 |
+
"grad_norm": 1.6357756853103638,
|
| 8187 |
+
"learning_rate": 8.27067013562669e-06,
|
| 8188 |
+
"loss": 0.8037,
|
| 8189 |
+
"step": 1164
|
| 8190 |
+
},
|
| 8191 |
+
{
|
| 8192 |
+
"epoch": 0.0038717085284765826,
|
| 8193 |
+
"grad_norm": 2.719797134399414,
|
| 8194 |
+
"learning_rate": 8.17665292796952e-06,
|
| 8195 |
+
"loss": 1.0075,
|
| 8196 |
+
"step": 1165
|
| 8197 |
+
},
|
| 8198 |
+
{
|
| 8199 |
+
"epoch": 0.003875031883436648,
|
| 8200 |
+
"grad_norm": 2.046483278274536,
|
| 8201 |
+
"learning_rate": 8.083150368936532e-06,
|
| 8202 |
+
"loss": 0.2509,
|
| 8203 |
+
"step": 1166
|
| 8204 |
+
},
|
| 8205 |
+
{
|
| 8206 |
+
"epoch": 0.003878355238396714,
|
| 8207 |
+
"grad_norm": 1.5247260332107544,
|
| 8208 |
+
"learning_rate": 7.990162982587924e-06,
|
| 8209 |
+
"loss": 1.1599,
|
| 8210 |
+
"step": 1167
|
| 8211 |
+
},
|
| 8212 |
+
{
|
| 8213 |
+
"epoch": 0.0038816785933567795,
|
| 8214 |
+
"grad_norm": 1.7768243551254272,
|
| 8215 |
+
"learning_rate": 7.897691290096498e-06,
|
| 8216 |
+
"loss": 0.6034,
|
| 8217 |
+
"step": 1168
|
| 8218 |
+
},
|
| 8219 |
+
{
|
| 8220 |
+
"epoch": 0.0038850019483168455,
|
| 8221 |
+
"grad_norm": 3.438926935195923,
|
| 8222 |
+
"learning_rate": 7.805735809744696e-06,
|
| 8223 |
+
"loss": 0.6389,
|
| 8224 |
+
"step": 1169
|
| 8225 |
+
},
|
| 8226 |
+
{
|
| 8227 |
+
"epoch": 0.003888325303276911,
|
| 8228 |
+
"grad_norm": 1.7144073247909546,
|
| 8229 |
+
"learning_rate": 7.714297056921715e-06,
|
| 8230 |
+
"loss": 1.159,
|
| 8231 |
+
"step": 1170
|
| 8232 |
+
},
|
| 8233 |
+
{
|
| 8234 |
+
"epoch": 0.003891648658236977,
|
| 8235 |
+
"grad_norm": 1.8290139436721802,
|
| 8236 |
+
"learning_rate": 7.623375544120648e-06,
|
| 8237 |
+
"loss": 0.9152,
|
| 8238 |
+
"step": 1171
|
| 8239 |
+
},
|
| 8240 |
+
{
|
| 8241 |
+
"epoch": 0.0038949720131970424,
|
| 8242 |
+
"grad_norm": 2.1877503395080566,
|
| 8243 |
+
"learning_rate": 7.532971780935516e-06,
|
| 8244 |
+
"loss": 1.0348,
|
| 8245 |
+
"step": 1172
|
| 8246 |
+
},
|
| 8247 |
+
{
|
| 8248 |
+
"epoch": 0.003898295368157108,
|
| 8249 |
+
"grad_norm": 1.77268648147583,
|
| 8250 |
+
"learning_rate": 7.443086274058497e-06,
|
| 8251 |
+
"loss": 0.7937,
|
| 8252 |
+
"step": 1173
|
| 8253 |
+
},
|
| 8254 |
+
{
|
| 8255 |
+
"epoch": 0.003901618723117174,
|
| 8256 |
+
"grad_norm": 1.3810285329818726,
|
| 8257 |
+
"learning_rate": 7.353719527277125e-06,
|
| 8258 |
+
"loss": 0.9105,
|
| 8259 |
+
"step": 1174
|
| 8260 |
+
},
|
| 8261 |
+
{
|
| 8262 |
+
"epoch": 0.0039049420780772398,
|
| 8263 |
+
"grad_norm": 1.8284751176834106,
|
| 8264 |
+
"learning_rate": 7.2648720414712716e-06,
|
| 8265 |
+
"loss": 1.1995,
|
| 8266 |
+
"step": 1175
|
| 8267 |
+
},
|
| 8268 |
+
{
|
| 8269 |
+
"epoch": 0.003908265433037306,
|
| 8270 |
+
"grad_norm": 1.7051944732666016,
|
| 8271 |
+
"learning_rate": 7.1765443146106e-06,
|
| 8272 |
+
"loss": 0.7849,
|
| 8273 |
+
"step": 1176
|
| 8274 |
+
},
|
| 8275 |
+
{
|
| 8276 |
+
"epoch": 0.003911588787997371,
|
| 8277 |
+
"grad_norm": 1.6046769618988037,
|
| 8278 |
+
"learning_rate": 7.088736841751575e-06,
|
| 8279 |
+
"loss": 0.8096,
|
| 8280 |
+
"step": 1177
|
| 8281 |
+
},
|
| 8282 |
+
{
|
| 8283 |
+
"epoch": 0.003914912142957437,
|
| 8284 |
+
"grad_norm": 2.0210366249084473,
|
| 8285 |
+
"learning_rate": 7.001450115034758e-06,
|
| 8286 |
+
"loss": 0.6858,
|
| 8287 |
+
"step": 1178
|
| 8288 |
+
},
|
| 8289 |
+
{
|
| 8290 |
+
"epoch": 0.003918235497917502,
|
| 8291 |
+
"grad_norm": 1.8352065086364746,
|
| 8292 |
+
"learning_rate": 6.914684623682099e-06,
|
| 8293 |
+
"loss": 0.8634,
|
| 8294 |
+
"step": 1179
|
| 8295 |
+
},
|
| 8296 |
+
{
|
| 8297 |
+
"epoch": 0.003921558852877568,
|
| 8298 |
+
"grad_norm": 1.6586220264434814,
|
| 8299 |
+
"learning_rate": 6.828440853994089e-06,
|
| 8300 |
+
"loss": 1.0516,
|
| 8301 |
+
"step": 1180
|
| 8302 |
+
},
|
| 8303 |
+
{
|
| 8304 |
+
"epoch": 0.0039248822078376345,
|
| 8305 |
+
"grad_norm": 1.8991930484771729,
|
| 8306 |
+
"learning_rate": 6.7427192893471105e-06,
|
| 8307 |
+
"loss": 1.3651,
|
| 8308 |
+
"step": 1181
|
| 8309 |
+
},
|
| 8310 |
+
{
|
| 8311 |
+
"epoch": 0.0039282055627977,
|
| 8312 |
+
"grad_norm": 2.0748491287231445,
|
| 8313 |
+
"learning_rate": 6.657520410190721e-06,
|
| 8314 |
+
"loss": 0.9304,
|
| 8315 |
+
"step": 1182
|
| 8316 |
+
},
|
| 8317 |
+
{
|
| 8318 |
+
"epoch": 0.003931528917757766,
|
| 8319 |
+
"grad_norm": 2.1835572719573975,
|
| 8320 |
+
"learning_rate": 6.572844694044911e-06,
|
| 8321 |
+
"loss": 0.7364,
|
| 8322 |
+
"step": 1183
|
| 8323 |
+
},
|
| 8324 |
+
{
|
| 8325 |
+
"epoch": 0.003934852272717832,
|
| 8326 |
+
"grad_norm": 1.8947280645370483,
|
| 8327 |
+
"learning_rate": 6.488692615497516e-06,
|
| 8328 |
+
"loss": 0.9096,
|
| 8329 |
+
"step": 1184
|
| 8330 |
+
},
|
| 8331 |
+
{
|
| 8332 |
+
"epoch": 0.003938175627677897,
|
| 8333 |
+
"grad_norm": 1.584531307220459,
|
| 8334 |
+
"learning_rate": 6.405064646201464e-06,
|
| 8335 |
+
"loss": 0.8091,
|
| 8336 |
+
"step": 1185
|
| 8337 |
+
},
|
| 8338 |
+
{
|
| 8339 |
+
"epoch": 0.003941498982637963,
|
| 8340 |
+
"grad_norm": 2.568542718887329,
|
| 8341 |
+
"learning_rate": 6.321961254872166e-06,
|
| 8342 |
+
"loss": 0.8995,
|
| 8343 |
+
"step": 1186
|
| 8344 |
+
},
|
| 8345 |
+
{
|
| 8346 |
+
"epoch": 0.003944822337598028,
|
| 8347 |
+
"grad_norm": 1.460935354232788,
|
| 8348 |
+
"learning_rate": 6.239382907284941e-06,
|
| 8349 |
+
"loss": 0.6506,
|
| 8350 |
+
"step": 1187
|
| 8351 |
+
},
|
| 8352 |
+
{
|
| 8353 |
+
"epoch": 0.003948145692558094,
|
| 8354 |
+
"grad_norm": 2.127983570098877,
|
| 8355 |
+
"learning_rate": 6.157330066272282e-06,
|
| 8356 |
+
"loss": 0.8965,
|
| 8357 |
+
"step": 1188
|
| 8358 |
+
},
|
| 8359 |
+
{
|
| 8360 |
+
"epoch": 0.00395146904751816,
|
| 8361 |
+
"grad_norm": 1.676879644393921,
|
| 8362 |
+
"learning_rate": 6.0758031917214296e-06,
|
| 8363 |
+
"loss": 1.0078,
|
| 8364 |
+
"step": 1189
|
| 8365 |
+
},
|
| 8366 |
+
{
|
| 8367 |
+
"epoch": 0.003954792402478226,
|
| 8368 |
+
"grad_norm": 1.7694756984710693,
|
| 8369 |
+
"learning_rate": 5.994802740571659e-06,
|
| 8370 |
+
"loss": 0.5985,
|
| 8371 |
+
"step": 1190
|
| 8372 |
+
},
|
| 8373 |
+
{
|
| 8374 |
+
"epoch": 0.003958115757438292,
|
| 8375 |
+
"grad_norm": 2.5703816413879395,
|
| 8376 |
+
"learning_rate": 5.914329166811727e-06,
|
| 8377 |
+
"loss": 0.9092,
|
| 8378 |
+
"step": 1191
|
| 8379 |
+
},
|
| 8380 |
+
{
|
| 8381 |
+
"epoch": 0.003961439112398357,
|
| 8382 |
+
"grad_norm": 1.5351899862289429,
|
| 8383 |
+
"learning_rate": 5.834382921477466e-06,
|
| 8384 |
+
"loss": 1.1089,
|
| 8385 |
+
"step": 1192
|
| 8386 |
+
},
|
| 8387 |
+
{
|
| 8388 |
+
"epoch": 0.003964762467358423,
|
| 8389 |
+
"grad_norm": 2.5164711475372314,
|
| 8390 |
+
"learning_rate": 5.754964452649037e-06,
|
| 8391 |
+
"loss": 0.9535,
|
| 8392 |
+
"step": 1193
|
| 8393 |
+
},
|
| 8394 |
+
{
|
| 8395 |
+
"epoch": 0.003968085822318489,
|
| 8396 |
+
"grad_norm": 2.1561896800994873,
|
| 8397 |
+
"learning_rate": 5.67607420544859e-06,
|
| 8398 |
+
"loss": 1.2628,
|
| 8399 |
+
"step": 1194
|
| 8400 |
+
},
|
| 8401 |
+
{
|
| 8402 |
+
"epoch": 0.003971409177278554,
|
| 8403 |
+
"grad_norm": 2.350405693054199,
|
| 8404 |
+
"learning_rate": 5.597712622037754e-06,
|
| 8405 |
+
"loss": 0.826,
|
| 8406 |
+
"step": 1195
|
| 8407 |
+
},
|
| 8408 |
+
{
|
| 8409 |
+
"epoch": 0.00397473253223862,
|
| 8410 |
+
"grad_norm": 2.584944009780884,
|
| 8411 |
+
"learning_rate": 5.519880141615042e-06,
|
| 8412 |
+
"loss": 0.9049,
|
| 8413 |
+
"step": 1196
|
| 8414 |
+
},
|
| 8415 |
+
{
|
| 8416 |
+
"epoch": 0.003978055887198686,
|
| 8417 |
+
"grad_norm": 1.9516925811767578,
|
| 8418 |
+
"learning_rate": 5.442577200413546e-06,
|
| 8419 |
+
"loss": 1.0605,
|
| 8420 |
+
"step": 1197
|
| 8421 |
+
},
|
| 8422 |
+
{
|
| 8423 |
+
"epoch": 0.0039813792421587516,
|
| 8424 |
+
"grad_norm": 3.750558853149414,
|
| 8425 |
+
"learning_rate": 5.365804231698368e-06,
|
| 8426 |
+
"loss": 0.9567,
|
| 8427 |
+
"step": 1198
|
| 8428 |
+
},
|
| 8429 |
+
{
|
| 8430 |
+
"epoch": 0.003984702597118818,
|
| 8431 |
+
"grad_norm": 1.9897552728652954,
|
| 8432 |
+
"learning_rate": 5.289561665764198e-06,
|
| 8433 |
+
"loss": 0.6373,
|
| 8434 |
+
"step": 1199
|
| 8435 |
+
},
|
| 8436 |
+
{
|
| 8437 |
+
"epoch": 0.003988025952078883,
|
| 8438 |
+
"grad_norm": 2.0267670154571533,
|
| 8439 |
+
"learning_rate": 5.213849929933024e-06,
|
| 8440 |
+
"loss": 0.7988,
|
| 8441 |
+
"step": 1200
|
| 8442 |
+
},
|
| 8443 |
+
{
|
| 8444 |
+
"epoch": 0.003991349307038949,
|
| 8445 |
+
"grad_norm": 1.9218477010726929,
|
| 8446 |
+
"learning_rate": 5.138669448551614e-06,
|
| 8447 |
+
"loss": 0.8301,
|
| 8448 |
+
"step": 1201
|
| 8449 |
+
},
|
| 8450 |
+
{
|
| 8451 |
+
"epoch": 0.003994672661999015,
|
| 8452 |
+
"grad_norm": 2.785154104232788,
|
| 8453 |
+
"learning_rate": 5.0640206429891535e-06,
|
| 8454 |
+
"loss": 0.8819,
|
| 8455 |
+
"step": 1202
|
| 8456 |
+
},
|
| 8457 |
+
{
|
| 8458 |
+
"epoch": 0.00399799601695908,
|
| 8459 |
+
"grad_norm": 1.7020105123519897,
|
| 8460 |
+
"learning_rate": 4.989903931634965e-06,
|
| 8461 |
+
"loss": 0.9052,
|
| 8462 |
+
"step": 1203
|
| 8463 |
+
},
|
| 8464 |
+
{
|
| 8465 |
+
"epoch": 0.004001319371919146,
|
| 8466 |
+
"grad_norm": 1.9669575691223145,
|
| 8467 |
+
"learning_rate": 4.916319729896057e-06,
|
| 8468 |
+
"loss": 0.8468,
|
| 8469 |
+
"step": 1204
|
| 8470 |
+
},
|
| 8471 |
+
{
|
| 8472 |
+
"epoch": 0.004004642726879211,
|
| 8473 |
+
"grad_norm": 2.421363353729248,
|
| 8474 |
+
"learning_rate": 4.843268450194871e-06,
|
| 8475 |
+
"loss": 0.7444,
|
| 8476 |
+
"step": 1205
|
| 8477 |
+
},
|
| 8478 |
+
{
|
| 8479 |
+
"epoch": 0.0040079660818392775,
|
| 8480 |
+
"grad_norm": 1.5673800706863403,
|
| 8481 |
+
"learning_rate": 4.770750501966925e-06,
|
| 8482 |
+
"loss": 0.9701,
|
| 8483 |
+
"step": 1206
|
| 8484 |
+
},
|
| 8485 |
+
{
|
| 8486 |
+
"epoch": 0.004011289436799344,
|
| 8487 |
+
"grad_norm": 1.9752105474472046,
|
| 8488 |
+
"learning_rate": 4.698766291658552e-06,
|
| 8489 |
+
"loss": 1.0881,
|
| 8490 |
+
"step": 1207
|
| 8491 |
+
},
|
| 8492 |
+
{
|
| 8493 |
+
"epoch": 0.004014612791759409,
|
| 8494 |
+
"grad_norm": 1.9096660614013672,
|
| 8495 |
+
"learning_rate": 4.627316222724598e-06,
|
| 8496 |
+
"loss": 0.9221,
|
| 8497 |
+
"step": 1208
|
| 8498 |
+
},
|
| 8499 |
+
{
|
| 8500 |
+
"epoch": 0.004017936146719475,
|
| 8501 |
+
"grad_norm": 3.038297653198242,
|
| 8502 |
+
"learning_rate": 4.556400695626173e-06,
|
| 8503 |
+
"loss": 0.7838,
|
| 8504 |
+
"step": 1209
|
| 8505 |
+
},
|
| 8506 |
+
{
|
| 8507 |
+
"epoch": 0.004021259501679541,
|
| 8508 |
+
"grad_norm": 2.406052350997925,
|
| 8509 |
+
"learning_rate": 4.486020107828448e-06,
|
| 8510 |
+
"loss": 1.0768,
|
| 8511 |
+
"step": 1210
|
| 8512 |
+
},
|
| 8513 |
+
{
|
| 8514 |
+
"epoch": 0.004024582856639606,
|
| 8515 |
+
"grad_norm": 2.2107632160186768,
|
| 8516 |
+
"learning_rate": 4.416174853798283e-06,
|
| 8517 |
+
"loss": 0.9042,
|
| 8518 |
+
"step": 1211
|
| 8519 |
+
},
|
| 8520 |
+
{
|
| 8521 |
+
"epoch": 0.004027906211599672,
|
| 8522 |
+
"grad_norm": 1.4963706731796265,
|
| 8523 |
+
"learning_rate": 4.34686532500218e-06,
|
| 8524 |
+
"loss": 0.7624,
|
| 8525 |
+
"step": 1212
|
| 8526 |
+
},
|
| 8527 |
+
{
|
| 8528 |
+
"epoch": 0.004031229566559737,
|
| 8529 |
+
"grad_norm": 1.8679277896881104,
|
| 8530 |
+
"learning_rate": 4.2780919099040585e-06,
|
| 8531 |
+
"loss": 0.9459,
|
| 8532 |
+
"step": 1213
|
| 8533 |
+
},
|
| 8534 |
+
{
|
| 8535 |
+
"epoch": 0.0040345529215198035,
|
| 8536 |
+
"grad_norm": 2.617995023727417,
|
| 8537 |
+
"learning_rate": 4.2098549939629696e-06,
|
| 8538 |
+
"loss": 1.009,
|
| 8539 |
+
"step": 1214
|
| 8540 |
+
},
|
| 8541 |
+
{
|
| 8542 |
+
"epoch": 0.0040378762764798695,
|
| 8543 |
+
"grad_norm": 1.7749000787734985,
|
| 8544 |
+
"learning_rate": 4.14215495963105e-06,
|
| 8545 |
+
"loss": 0.8112,
|
| 8546 |
+
"step": 1215
|
| 8547 |
+
},
|
| 8548 |
+
{
|
| 8549 |
+
"epoch": 0.004041199631439935,
|
| 8550 |
+
"grad_norm": 1.2091432809829712,
|
| 8551 |
+
"learning_rate": 4.074992186351367e-06,
|
| 8552 |
+
"loss": 0.6515,
|
| 8553 |
+
"step": 1216
|
| 8554 |
+
},
|
| 8555 |
+
{
|
| 8556 |
+
"epoch": 0.004044522986400001,
|
| 8557 |
+
"grad_norm": 2.0824778079986572,
|
| 8558 |
+
"learning_rate": 4.008367050555683e-06,
|
| 8559 |
+
"loss": 1.0088,
|
| 8560 |
+
"step": 1217
|
| 8561 |
+
},
|
| 8562 |
+
{
|
| 8563 |
+
"epoch": 0.004047846341360066,
|
| 8564 |
+
"grad_norm": 1.503562569618225,
|
| 8565 |
+
"learning_rate": 3.942279925662506e-06,
|
| 8566 |
+
"loss": 0.8859,
|
| 8567 |
+
"step": 1218
|
| 8568 |
+
},
|
| 8569 |
+
{
|
| 8570 |
+
"epoch": 0.004051169696320132,
|
| 8571 |
+
"grad_norm": 2.2736172676086426,
|
| 8572 |
+
"learning_rate": 3.876731182074888e-06,
|
| 8573 |
+
"loss": 0.9602,
|
| 8574 |
+
"step": 1219
|
| 8575 |
+
},
|
| 8576 |
+
{
|
| 8577 |
+
"epoch": 0.004054493051280198,
|
| 8578 |
+
"grad_norm": 1.9392001628875732,
|
| 8579 |
+
"learning_rate": 3.811721187178352e-06,
|
| 8580 |
+
"loss": 1.2851,
|
| 8581 |
+
"step": 1220
|
| 8582 |
+
},
|
| 8583 |
+
{
|
| 8584 |
+
"epoch": 0.004057816406240263,
|
| 8585 |
+
"grad_norm": 1.8600444793701172,
|
| 8586 |
+
"learning_rate": 3.747250305338934e-06,
|
| 8587 |
+
"loss": 0.3464,
|
| 8588 |
+
"step": 1221
|
| 8589 |
+
},
|
| 8590 |
+
{
|
| 8591 |
+
"epoch": 0.004061139761200329,
|
| 8592 |
+
"grad_norm": 3.741283655166626,
|
| 8593 |
+
"learning_rate": 3.6833188979009447e-06,
|
| 8594 |
+
"loss": 0.9047,
|
| 8595 |
+
"step": 1222
|
| 8596 |
+
},
|
| 8597 |
+
{
|
| 8598 |
+
"epoch": 0.0040644631161603955,
|
| 8599 |
+
"grad_norm": 2.066908359527588,
|
| 8600 |
+
"learning_rate": 3.6199273231852016e-06,
|
| 8601 |
+
"loss": 0.7508,
|
| 8602 |
+
"step": 1223
|
| 8603 |
+
},
|
| 8604 |
+
{
|
| 8605 |
+
"epoch": 0.004067786471120461,
|
| 8606 |
+
"grad_norm": 1.472381591796875,
|
| 8607 |
+
"learning_rate": 3.5570759364867976e-06,
|
| 8608 |
+
"loss": 1.1071,
|
| 8609 |
+
"step": 1224
|
| 8610 |
+
},
|
| 8611 |
+
{
|
| 8612 |
+
"epoch": 0.004071109826080527,
|
| 8613 |
+
"grad_norm": 2.138390302658081,
|
| 8614 |
+
"learning_rate": 3.494765090073193e-06,
|
| 8615 |
+
"loss": 0.7563,
|
| 8616 |
+
"step": 1225
|
| 8617 |
+
},
|
| 8618 |
+
{
|
| 8619 |
+
"epoch": 0.004074433181040592,
|
| 8620 |
+
"grad_norm": 1.5176939964294434,
|
| 8621 |
+
"learning_rate": 3.432995133182315e-06,
|
| 8622 |
+
"loss": 1.1034,
|
| 8623 |
+
"step": 1226
|
| 8624 |
+
},
|
| 8625 |
+
{
|
| 8626 |
+
"epoch": 0.004077756536000658,
|
| 8627 |
+
"grad_norm": 1.9056531190872192,
|
| 8628 |
+
"learning_rate": 3.3717664120204717e-06,
|
| 8629 |
+
"loss": 0.7667,
|
| 8630 |
+
"step": 1227
|
| 8631 |
+
},
|
| 8632 |
+
{
|
| 8633 |
+
"epoch": 0.004081079890960724,
|
| 8634 |
+
"grad_norm": 3.191318988800049,
|
| 8635 |
+
"learning_rate": 3.3110792697604755e-06,
|
| 8636 |
+
"loss": 0.8358,
|
| 8637 |
+
"step": 1228
|
| 8638 |
+
},
|
| 8639 |
+
{
|
| 8640 |
+
"epoch": 0.004084403245920789,
|
| 8641 |
+
"grad_norm": 1.8594859838485718,
|
| 8642 |
+
"learning_rate": 3.250934046539722e-06,
|
| 8643 |
+
"loss": 0.2147,
|
| 8644 |
+
"step": 1229
|
| 8645 |
+
},
|
| 8646 |
+
{
|
| 8647 |
+
"epoch": 0.004087726600880855,
|
| 8648 |
+
"grad_norm": 1.8989686965942383,
|
| 8649 |
+
"learning_rate": 3.1913310794582817e-06,
|
| 8650 |
+
"loss": 0.7407,
|
| 8651 |
+
"step": 1230
|
| 8652 |
+
},
|
| 8653 |
+
{
|
| 8654 |
+
"epoch": 0.0040910499558409206,
|
| 8655 |
+
"grad_norm": 2.4154255390167236,
|
| 8656 |
+
"learning_rate": 3.1322707025770114e-06,
|
| 8657 |
+
"loss": 0.7876,
|
| 8658 |
+
"step": 1231
|
| 8659 |
+
},
|
| 8660 |
+
{
|
| 8661 |
+
"epoch": 0.004094373310800987,
|
| 8662 |
+
"grad_norm": 2.14886736869812,
|
| 8663 |
+
"learning_rate": 3.073753246915656e-06,
|
| 8664 |
+
"loss": 0.8468,
|
| 8665 |
+
"step": 1232
|
| 8666 |
+
},
|
| 8667 |
+
{
|
| 8668 |
+
"epoch": 0.004097696665761053,
|
| 8669 |
+
"grad_norm": 2.599923849105835,
|
| 8670 |
+
"learning_rate": 3.015779040451017e-06,
|
| 8671 |
+
"loss": 0.8232,
|
| 8672 |
+
"step": 1233
|
| 8673 |
+
},
|
| 8674 |
+
{
|
| 8675 |
+
"epoch": 0.004101020020721118,
|
| 8676 |
+
"grad_norm": 1.8031104803085327,
|
| 8677 |
+
"learning_rate": 2.958348408115108e-06,
|
| 8678 |
+
"loss": 0.9074,
|
| 8679 |
+
"step": 1234
|
| 8680 |
+
},
|
| 8681 |
+
{
|
| 8682 |
+
"epoch": 0.004104343375681184,
|
| 8683 |
+
"grad_norm": 1.2037352323532104,
|
| 8684 |
+
"learning_rate": 2.9014616717933595e-06,
|
| 8685 |
+
"loss": 0.494,
|
| 8686 |
+
"step": 1235
|
| 8687 |
+
},
|
| 8688 |
+
{
|
| 8689 |
+
"epoch": 0.00410766673064125,
|
| 8690 |
+
"grad_norm": 1.8279937505722046,
|
| 8691 |
+
"learning_rate": 2.8451191503227614e-06,
|
| 8692 |
+
"loss": 0.7566,
|
| 8693 |
+
"step": 1236
|
| 8694 |
+
},
|
| 8695 |
+
{
|
| 8696 |
+
"epoch": 0.004110990085601315,
|
| 8697 |
+
"grad_norm": 1.9116507768630981,
|
| 8698 |
+
"learning_rate": 2.7893211594901215e-06,
|
| 8699 |
+
"loss": 0.937,
|
| 8700 |
+
"step": 1237
|
| 8701 |
+
},
|
| 8702 |
+
{
|
| 8703 |
+
"epoch": 0.004114313440561381,
|
| 8704 |
+
"grad_norm": 1.6439839601516724,
|
| 8705 |
+
"learning_rate": 2.7340680120302554e-06,
|
| 8706 |
+
"loss": 0.7953,
|
| 8707 |
+
"step": 1238
|
| 8708 |
+
},
|
| 8709 |
+
{
|
| 8710 |
+
"epoch": 0.0041176367955214465,
|
| 8711 |
+
"grad_norm": 2.841966390609741,
|
| 8712 |
+
"learning_rate": 2.6793600176243105e-06,
|
| 8713 |
+
"loss": 0.7789,
|
| 8714 |
+
"step": 1239
|
| 8715 |
+
},
|
| 8716 |
+
{
|
| 8717 |
+
"epoch": 0.004120960150481513,
|
| 8718 |
+
"grad_norm": 1.5409972667694092,
|
| 8719 |
+
"learning_rate": 2.625197482897912e-06,
|
| 8720 |
+
"loss": 0.4756,
|
| 8721 |
+
"step": 1240
|
| 8722 |
+
},
|
| 8723 |
+
{
|
| 8724 |
+
"epoch": 0.004124283505441579,
|
| 8725 |
+
"grad_norm": 1.8044242858886719,
|
| 8726 |
+
"learning_rate": 2.5715807114195525e-06,
|
| 8727 |
+
"loss": 0.7198,
|
| 8728 |
+
"step": 1241
|
| 8729 |
+
},
|
| 8730 |
+
{
|
| 8731 |
+
"epoch": 0.004127606860401644,
|
| 8732 |
+
"grad_norm": 2.6371781826019287,
|
| 8733 |
+
"learning_rate": 2.5185100036988484e-06,
|
| 8734 |
+
"loss": 0.6385,
|
| 8735 |
+
"step": 1242
|
| 8736 |
+
},
|
| 8737 |
+
{
|
| 8738 |
+
"epoch": 0.00413093021536171,
|
| 8739 |
+
"grad_norm": 1.3817503452301025,
|
| 8740 |
+
"learning_rate": 2.4659856571848105e-06,
|
| 8741 |
+
"loss": 0.9401,
|
| 8742 |
+
"step": 1243
|
| 8743 |
+
},
|
| 8744 |
+
{
|
| 8745 |
+
"epoch": 0.004134253570321775,
|
| 8746 |
+
"grad_norm": 1.8377162218093872,
|
| 8747 |
+
"learning_rate": 2.4140079662642865e-06,
|
| 8748 |
+
"loss": 0.8917,
|
| 8749 |
+
"step": 1244
|
| 8750 |
+
},
|
| 8751 |
+
{
|
| 8752 |
+
"epoch": 0.004137576925281841,
|
| 8753 |
+
"grad_norm": 1.7400139570236206,
|
| 8754 |
+
"learning_rate": 2.362577222260198e-06,
|
| 8755 |
+
"loss": 1.108,
|
| 8756 |
+
"step": 1245
|
| 8757 |
+
},
|
| 8758 |
+
{
|
| 8759 |
+
"epoch": 0.004140900280241907,
|
| 8760 |
+
"grad_norm": 1.966605544090271,
|
| 8761 |
+
"learning_rate": 2.3116937134299745e-06,
|
| 8762 |
+
"loss": 0.6442,
|
| 8763 |
+
"step": 1246
|
| 8764 |
+
},
|
| 8765 |
+
{
|
| 8766 |
+
"epoch": 0.0041442236352019724,
|
| 8767 |
+
"grad_norm": 1.370110273361206,
|
| 8768 |
+
"learning_rate": 2.2613577249639083e-06,
|
| 8769 |
+
"loss": 0.777,
|
| 8770 |
+
"step": 1247
|
| 8771 |
+
},
|
| 8772 |
+
{
|
| 8773 |
+
"epoch": 0.0041475469901620385,
|
| 8774 |
+
"grad_norm": 1.5771291255950928,
|
| 8775 |
+
"learning_rate": 2.2115695389835712e-06,
|
| 8776 |
+
"loss": 0.6103,
|
| 8777 |
+
"step": 1248
|
| 8778 |
+
},
|
| 8779 |
+
{
|
| 8780 |
+
"epoch": 0.004150870345122105,
|
| 8781 |
+
"grad_norm": 2.013594388961792,
|
| 8782 |
+
"learning_rate": 2.1623294345402447e-06,
|
| 8783 |
+
"loss": 1.0112,
|
| 8784 |
+
"step": 1249
|
| 8785 |
+
},
|
| 8786 |
+
{
|
| 8787 |
+
"epoch": 0.00415419370008217,
|
| 8788 |
+
"grad_norm": 2.0027196407318115,
|
| 8789 |
+
"learning_rate": 2.1136376876133234e-06,
|
| 8790 |
+
"loss": 0.8395,
|
| 8791 |
+
"step": 1250
|
| 8792 |
+
},
|
| 8793 |
+
{
|
| 8794 |
+
"epoch": 0.004157517055042236,
|
| 8795 |
+
"grad_norm": 2.6950972080230713,
|
| 8796 |
+
"learning_rate": 2.0654945711087834e-06,
|
| 8797 |
+
"loss": 1.2351,
|
| 8798 |
+
"step": 1251
|
| 8799 |
+
},
|
| 8800 |
+
{
|
| 8801 |
+
"epoch": 0.004160840410002301,
|
| 8802 |
+
"grad_norm": 2.4111666679382324,
|
| 8803 |
+
"learning_rate": 2.0179003548576602e-06,
|
| 8804 |
+
"loss": 1.0297,
|
| 8805 |
+
"step": 1252
|
| 8806 |
+
},
|
| 8807 |
+
{
|
| 8808 |
+
"epoch": 0.004164163764962367,
|
| 8809 |
+
"grad_norm": 1.8260390758514404,
|
| 8810 |
+
"learning_rate": 1.970855305614516e-06,
|
| 8811 |
+
"loss": 1.159,
|
| 8812 |
+
"step": 1253
|
| 8813 |
+
},
|
| 8814 |
+
{
|
| 8815 |
+
"epoch": 0.004167487119922433,
|
| 8816 |
+
"grad_norm": 2.396212577819824,
|
| 8817 |
+
"learning_rate": 1.9243596870559988e-06,
|
| 8818 |
+
"loss": 0.9679,
|
| 8819 |
+
"step": 1254
|
| 8820 |
+
},
|
| 8821 |
+
{
|
| 8822 |
+
"epoch": 0.004170810474882498,
|
| 8823 |
+
"grad_norm": 2.532935380935669,
|
| 8824 |
+
"learning_rate": 1.8784137597792738e-06,
|
| 8825 |
+
"loss": 0.9187,
|
| 8826 |
+
"step": 1255
|
| 8827 |
+
},
|
| 8828 |
+
{
|
| 8829 |
+
"epoch": 0.0041741338298425645,
|
| 8830 |
+
"grad_norm": 1.9746872186660767,
|
| 8831 |
+
"learning_rate": 1.8330177813006388e-06,
|
| 8832 |
+
"loss": 0.9043,
|
| 8833 |
+
"step": 1256
|
| 8834 |
+
},
|
| 8835 |
+
{
|
| 8836 |
+
"epoch": 0.0041774571848026305,
|
| 8837 |
+
"grad_norm": 1.886202335357666,
|
| 8838 |
+
"learning_rate": 1.7881720060540786e-06,
|
| 8839 |
+
"loss": 0.8272,
|
| 8840 |
+
"step": 1257
|
| 8841 |
+
},
|
| 8842 |
+
{
|
| 8843 |
+
"epoch": 0.004180780539762696,
|
| 8844 |
+
"grad_norm": 1.5669914484024048,
|
| 8845 |
+
"learning_rate": 1.7438766853897558e-06,
|
| 8846 |
+
"loss": 0.798,
|
| 8847 |
+
"step": 1258
|
| 8848 |
+
},
|
| 8849 |
+
{
|
| 8850 |
+
"epoch": 0.004184103894722762,
|
| 8851 |
+
"grad_norm": 1.6605589389801025,
|
| 8852 |
+
"learning_rate": 1.7001320675727239e-06,
|
| 8853 |
+
"loss": 1.1453,
|
| 8854 |
+
"step": 1259
|
| 8855 |
+
},
|
| 8856 |
+
{
|
| 8857 |
+
"epoch": 0.004187427249682827,
|
| 8858 |
+
"grad_norm": 2.4264414310455322,
|
| 8859 |
+
"learning_rate": 1.656938397781449e-06,
|
| 8860 |
+
"loss": 0.727,
|
| 8861 |
+
"step": 1260
|
| 8862 |
+
},
|
| 8863 |
+
{
|
| 8864 |
+
"epoch": 0.004190750604642893,
|
| 8865 |
+
"grad_norm": 2.188891887664795,
|
| 8866 |
+
"learning_rate": 1.6142959181064343e-06,
|
| 8867 |
+
"loss": 0.9834,
|
| 8868 |
+
"step": 1261
|
| 8869 |
+
},
|
| 8870 |
+
{
|
| 8871 |
+
"epoch": 0.004194073959602959,
|
| 8872 |
+
"grad_norm": 1.7915462255477905,
|
| 8873 |
+
"learning_rate": 1.5722048675489541e-06,
|
| 8874 |
+
"loss": 0.8014,
|
| 8875 |
+
"step": 1262
|
| 8876 |
+
},
|
| 8877 |
+
{
|
| 8878 |
+
"epoch": 0.004197397314563024,
|
| 8879 |
+
"grad_norm": 2.2112960815429688,
|
| 8880 |
+
"learning_rate": 1.5306654820195886e-06,
|
| 8881 |
+
"loss": 0.669,
|
| 8882 |
+
"step": 1263
|
| 8883 |
+
},
|
| 8884 |
+
{
|
| 8885 |
+
"epoch": 0.00420072066952309,
|
| 8886 |
+
"grad_norm": 1.741981029510498,
|
| 8887 |
+
"learning_rate": 1.489677994336991e-06,
|
| 8888 |
+
"loss": 0.9008,
|
| 8889 |
+
"step": 1264
|
| 8890 |
+
},
|
| 8891 |
+
{
|
| 8892 |
+
"epoch": 0.004204044024483156,
|
| 8893 |
+
"grad_norm": 1.3700710535049438,
|
| 8894 |
+
"learning_rate": 1.4492426342265552e-06,
|
| 8895 |
+
"loss": 0.7386,
|
| 8896 |
+
"step": 1265
|
| 8897 |
+
},
|
| 8898 |
+
{
|
| 8899 |
+
"epoch": 0.004207367379443222,
|
| 8900 |
+
"grad_norm": 2.1248631477355957,
|
| 8901 |
+
"learning_rate": 1.4093596283191179e-06,
|
| 8902 |
+
"loss": 0.977,
|
| 8903 |
+
"step": 1266
|
| 8904 |
+
},
|
| 8905 |
+
{
|
| 8906 |
+
"epoch": 0.004210690734403288,
|
| 8907 |
+
"grad_norm": 2.1686811447143555,
|
| 8908 |
+
"learning_rate": 1.3700292001497028e-06,
|
| 8909 |
+
"loss": 0.8883,
|
| 8910 |
+
"step": 1267
|
| 8911 |
+
},
|
| 8912 |
+
{
|
| 8913 |
+
"epoch": 0.004214014089363353,
|
| 8914 |
+
"grad_norm": 1.8819714784622192,
|
| 8915 |
+
"learning_rate": 1.3312515701562667e-06,
|
| 8916 |
+
"loss": 0.5279,
|
| 8917 |
+
"step": 1268
|
| 8918 |
+
},
|
| 8919 |
+
{
|
| 8920 |
+
"epoch": 0.004217337444323419,
|
| 8921 |
+
"grad_norm": 1.7714366912841797,
|
| 8922 |
+
"learning_rate": 1.2930269556784558e-06,
|
| 8923 |
+
"loss": 1.1745,
|
| 8924 |
+
"step": 1269
|
| 8925 |
+
},
|
| 8926 |
+
{
|
| 8927 |
+
"epoch": 0.004220660799283485,
|
| 8928 |
+
"grad_norm": 2.3303427696228027,
|
| 8929 |
+
"learning_rate": 1.2553555709564068e-06,
|
| 8930 |
+
"loss": 1.0176,
|
| 8931 |
+
"step": 1270
|
| 8932 |
+
},
|
| 8933 |
+
{
|
| 8934 |
+
"epoch": 0.00422398415424355,
|
| 8935 |
+
"grad_norm": 1.934847354888916,
|
| 8936 |
+
"learning_rate": 1.2182376271294815e-06,
|
| 8937 |
+
"loss": 1.0643,
|
| 8938 |
+
"step": 1271
|
| 8939 |
+
},
|
| 8940 |
+
{
|
| 8941 |
+
"epoch": 0.004227307509203616,
|
| 8942 |
+
"grad_norm": 1.9978758096694946,
|
| 8943 |
+
"learning_rate": 1.1816733322352114e-06,
|
| 8944 |
+
"loss": 1.093,
|
| 8945 |
+
"step": 1272
|
| 8946 |
+
},
|
| 8947 |
+
{
|
| 8948 |
+
"epoch": 0.0042306308641636816,
|
| 8949 |
+
"grad_norm": 1.9631693363189697,
|
| 8950 |
+
"learning_rate": 1.1456628912079992e-06,
|
| 8951 |
+
"loss": 1.091,
|
| 8952 |
+
"step": 1273
|
| 8953 |
+
},
|
| 8954 |
+
{
|
| 8955 |
+
"epoch": 0.004233954219123748,
|
| 8956 |
+
"grad_norm": 1.8431519269943237,
|
| 8957 |
+
"learning_rate": 1.110206505878031e-06,
|
| 8958 |
+
"loss": 0.8671,
|
| 8959 |
+
"step": 1274
|
| 8960 |
+
},
|
| 8961 |
+
{
|
| 8962 |
+
"epoch": 0.004237277574083814,
|
| 8963 |
+
"grad_norm": 2.2053067684173584,
|
| 8964 |
+
"learning_rate": 1.0753043749701652e-06,
|
| 8965 |
+
"loss": 0.8372,
|
| 8966 |
+
"step": 1275
|
| 8967 |
+
},
|
| 8968 |
+
{
|
| 8969 |
+
"epoch": 0.004240600929043879,
|
| 8970 |
+
"grad_norm": 2.2077174186706543,
|
| 8971 |
+
"learning_rate": 1.040956694102746e-06,
|
| 8972 |
+
"loss": 1.0352,
|
| 8973 |
+
"step": 1276
|
| 8974 |
+
},
|
| 8975 |
+
{
|
| 8976 |
+
"epoch": 0.004243924284003945,
|
| 8977 |
+
"grad_norm": 1.8683050870895386,
|
| 8978 |
+
"learning_rate": 1.0071636557866137e-06,
|
| 8979 |
+
"loss": 0.9413,
|
| 8980 |
+
"step": 1277
|
| 8981 |
+
},
|
| 8982 |
+
{
|
| 8983 |
+
"epoch": 0.00424724763896401,
|
| 8984 |
+
"grad_norm": 1.6869767904281616,
|
| 8985 |
+
"learning_rate": 9.73925449423896e-07,
|
| 8986 |
+
"loss": 0.8921,
|
| 8987 |
+
"step": 1278
|
| 8988 |
+
},
|
| 8989 |
+
{
|
| 8990 |
+
"epoch": 0.004250570993924076,
|
| 8991 |
+
"grad_norm": 2.4182016849517822,
|
| 8992 |
+
"learning_rate": 9.412422613070848e-07,
|
| 8993 |
+
"loss": 0.9755,
|
| 8994 |
+
"step": 1279
|
| 8995 |
+
},
|
| 8996 |
+
{
|
| 8997 |
+
"epoch": 0.004253894348884142,
|
| 8998 |
+
"grad_norm": 1.7648463249206543,
|
| 8999 |
+
"learning_rate": 9.091142746178726e-07,
|
| 9000 |
+
"loss": 1.1577,
|
| 9001 |
+
"step": 1280
|
| 9002 |
+
},
|
| 9003 |
+
{
|
| 9004 |
+
"epoch": 0.0042572177038442075,
|
| 9005 |
+
"grad_norm": 1.8752923011779785,
|
| 9006 |
+
"learning_rate": 8.775416694262073e-07,
|
| 9007 |
+
"loss": 0.96,
|
| 9008 |
+
"step": 1281
|
| 9009 |
+
},
|
| 9010 |
+
{
|
| 9011 |
+
"epoch": 0.004260541058804274,
|
| 9012 |
+
"grad_norm": 1.7734326124191284,
|
| 9013 |
+
"learning_rate": 8.465246226892154e-07,
|
| 9014 |
+
"loss": 0.6681,
|
| 9015 |
+
"step": 1282
|
| 9016 |
+
},
|
| 9017 |
+
{
|
| 9018 |
+
"epoch": 0.00426386441376434,
|
| 9019 |
+
"grad_norm": 2.1282567977905273,
|
| 9020 |
+
"learning_rate": 8.160633082502922e-07,
|
| 9021 |
+
"loss": 0.8363,
|
| 9022 |
+
"step": 1283
|
| 9023 |
+
},
|
| 9024 |
+
{
|
| 9025 |
+
"epoch": 0.004267187768724405,
|
| 9026 |
+
"grad_norm": 1.7843672037124634,
|
| 9027 |
+
"learning_rate": 7.861578968380578e-07,
|
| 9028 |
+
"loss": 1.0503,
|
| 9029 |
+
"step": 1284
|
| 9030 |
+
},
|
| 9031 |
+
{
|
| 9032 |
+
"epoch": 0.004270511123684471,
|
| 9033 |
+
"grad_norm": 1.9426231384277344,
|
| 9034 |
+
"learning_rate": 7.568085560654247e-07,
|
| 9035 |
+
"loss": 1.0447,
|
| 9036 |
+
"step": 1285
|
| 9037 |
+
},
|
| 9038 |
+
{
|
| 9039 |
+
"epoch": 0.004273834478644536,
|
| 9040 |
+
"grad_norm": 2.154215097427368,
|
| 9041 |
+
"learning_rate": 7.28015450428654e-07,
|
| 9042 |
+
"loss": 1.0945,
|
| 9043 |
+
"step": 1286
|
| 9044 |
+
},
|
| 9045 |
+
{
|
| 9046 |
+
"epoch": 0.004277157833604602,
|
| 9047 |
+
"grad_norm": 1.881326675415039,
|
| 9048 |
+
"learning_rate": 6.99778741306456e-07,
|
| 9049 |
+
"loss": 0.8126,
|
| 9050 |
+
"step": 1287
|
| 9051 |
+
},
|
| 9052 |
+
{
|
| 9053 |
+
"epoch": 0.004280481188564668,
|
| 9054 |
+
"grad_norm": 2.788193941116333,
|
| 9055 |
+
"learning_rate": 6.720985869590357e-07,
|
| 9056 |
+
"loss": 0.6972,
|
| 9057 |
+
"step": 1288
|
| 9058 |
+
},
|
| 9059 |
+
{
|
| 9060 |
+
"epoch": 0.0042838045435247335,
|
| 9061 |
+
"grad_norm": 2.3980252742767334,
|
| 9062 |
+
"learning_rate": 6.44975142527282e-07,
|
| 9063 |
+
"loss": 1.0364,
|
| 9064 |
+
"step": 1289
|
| 9065 |
+
},
|
| 9066 |
+
{
|
| 9067 |
+
"epoch": 0.0042871278984847995,
|
| 9068 |
+
"grad_norm": 2.5638821125030518,
|
| 9069 |
+
"learning_rate": 6.184085600318024e-07,
|
| 9070 |
+
"loss": 0.9393,
|
| 9071 |
+
"step": 1290
|
| 9072 |
+
},
|
| 9073 |
+
{
|
| 9074 |
+
"epoch": 0.004290451253444865,
|
| 9075 |
+
"grad_norm": 3.0566511154174805,
|
| 9076 |
+
"learning_rate": 5.92398988372167e-07,
|
| 9077 |
+
"loss": 1.0036,
|
| 9078 |
+
"step": 1291
|
| 9079 |
+
},
|
| 9080 |
+
{
|
| 9081 |
+
"epoch": 0.004293774608404931,
|
| 9082 |
+
"grad_norm": 2.413839101791382,
|
| 9083 |
+
"learning_rate": 5.669465733259994e-07,
|
| 9084 |
+
"loss": 1.3389,
|
| 9085 |
+
"step": 1292
|
| 9086 |
+
},
|
| 9087 |
+
{
|
| 9088 |
+
"epoch": 0.004297097963364997,
|
| 9089 |
+
"grad_norm": 2.7977092266082764,
|
| 9090 |
+
"learning_rate": 5.420514575482205e-07,
|
| 9091 |
+
"loss": 1.091,
|
| 9092 |
+
"step": 1293
|
| 9093 |
+
},
|
| 9094 |
+
{
|
| 9095 |
+
"epoch": 0.004300421318325062,
|
| 9096 |
+
"grad_norm": 3.4734342098236084,
|
| 9097 |
+
"learning_rate": 5.177137805701615e-07,
|
| 9098 |
+
"loss": 0.7818,
|
| 9099 |
+
"step": 1294
|
| 9100 |
+
},
|
| 9101 |
+
{
|
| 9102 |
+
"epoch": 0.004303744673285128,
|
| 9103 |
+
"grad_norm": 2.988787889480591,
|
| 9104 |
+
"learning_rate": 4.939336787988857e-07,
|
| 9105 |
+
"loss": 0.7218,
|
| 9106 |
+
"step": 1295
|
| 9107 |
+
},
|
| 9108 |
+
{
|
| 9109 |
+
"epoch": 0.004307068028245194,
|
| 9110 |
+
"grad_norm": 1.6737980842590332,
|
| 9111 |
+
"learning_rate": 4.7071128551637867e-07,
|
| 9112 |
+
"loss": 1.1564,
|
| 9113 |
+
"step": 1296
|
| 9114 |
+
},
|
| 9115 |
+
{
|
| 9116 |
+
"epoch": 0.004310391383205259,
|
| 9117 |
+
"grad_norm": 2.376605749130249,
|
| 9118 |
+
"learning_rate": 4.4804673087877057e-07,
|
| 9119 |
+
"loss": 0.3682,
|
| 9120 |
+
"step": 1297
|
| 9121 |
+
},
|
| 9122 |
+
{
|
| 9123 |
+
"epoch": 0.0043137147381653255,
|
| 9124 |
+
"grad_norm": 2.320552110671997,
|
| 9125 |
+
"learning_rate": 4.2594014191565945e-07,
|
| 9126 |
+
"loss": 0.9988,
|
| 9127 |
+
"step": 1298
|
| 9128 |
+
},
|
| 9129 |
+
{
|
| 9130 |
+
"epoch": 0.004317038093125391,
|
| 9131 |
+
"grad_norm": 2.8528194427490234,
|
| 9132 |
+
"learning_rate": 4.043916425293559e-07,
|
| 9133 |
+
"loss": 1.3724,
|
| 9134 |
+
"step": 1299
|
| 9135 |
+
},
|
| 9136 |
+
{
|
| 9137 |
+
"epoch": 0.004320361448085457,
|
| 9138 |
+
"grad_norm": 3.585374593734741,
|
| 9139 |
+
"learning_rate": 3.83401353494206e-07,
|
| 9140 |
+
"loss": 0.9507,
|
| 9141 |
+
"step": 1300
|
| 9142 |
+
},
|
| 9143 |
+
{
|
| 9144 |
+
"epoch": 0.004323684803045523,
|
| 9145 |
+
"grad_norm": 2.1745338439941406,
|
| 9146 |
+
"learning_rate": 3.6296939245593633e-07,
|
| 9147 |
+
"loss": 0.6232,
|
| 9148 |
+
"step": 1301
|
| 9149 |
+
},
|
| 9150 |
+
{
|
| 9151 |
+
"epoch": 0.004327008158005588,
|
| 9152 |
+
"grad_norm": 1.2730436325073242,
|
| 9153 |
+
"learning_rate": 3.4309587393096534e-07,
|
| 9154 |
+
"loss": 0.6022,
|
| 9155 |
+
"step": 1302
|
| 9156 |
+
},
|
| 9157 |
+
{
|
| 9158 |
+
"epoch": 0.004330331512965654,
|
| 9159 |
+
"grad_norm": 2.007568120956421,
|
| 9160 |
+
"learning_rate": 3.2378090930577087e-07,
|
| 9161 |
+
"loss": 1.1188,
|
| 9162 |
+
"step": 1303
|
| 9163 |
+
},
|
| 9164 |
+
{
|
| 9165 |
+
"epoch": 0.004333654867925719,
|
| 9166 |
+
"grad_norm": 3.0893633365631104,
|
| 9167 |
+
"learning_rate": 3.0502460683624613e-07,
|
| 9168 |
+
"loss": 0.5077,
|
| 9169 |
+
"step": 1304
|
| 9170 |
+
},
|
| 9171 |
+
{
|
| 9172 |
+
"epoch": 0.004336978222885785,
|
| 9173 |
+
"grad_norm": 1.57036554813385,
|
| 9174 |
+
"learning_rate": 2.868270716471444e-07,
|
| 9175 |
+
"loss": 0.5061,
|
| 9176 |
+
"step": 1305
|
| 9177 |
+
},
|
| 9178 |
+
{
|
| 9179 |
+
"epoch": 0.004340301577845851,
|
| 9180 |
+
"grad_norm": 1.7737019062042236,
|
| 9181 |
+
"learning_rate": 2.6918840573144644e-07,
|
| 9182 |
+
"loss": 1.0445,
|
| 9183 |
+
"step": 1306
|
| 9184 |
+
},
|
| 9185 |
+
{
|
| 9186 |
+
"epoch": 0.004343624932805917,
|
| 9187 |
+
"grad_norm": 2.042353391647339,
|
| 9188 |
+
"learning_rate": 2.521087079497719e-07,
|
| 9189 |
+
"loss": 0.9395,
|
| 9190 |
+
"step": 1307
|
| 9191 |
+
},
|
| 9192 |
+
{
|
| 9193 |
+
"epoch": 0.004346948287765983,
|
| 9194 |
+
"grad_norm": 2.071798086166382,
|
| 9195 |
+
"learning_rate": 2.3558807402989103e-07,
|
| 9196 |
+
"loss": 1.0835,
|
| 9197 |
+
"step": 1308
|
| 9198 |
+
},
|
| 9199 |
+
{
|
| 9200 |
+
"epoch": 0.004350271642726049,
|
| 9201 |
+
"grad_norm": 1.6557015180587769,
|
| 9202 |
+
"learning_rate": 2.1962659656614703e-07,
|
| 9203 |
+
"loss": 1.0824,
|
| 9204 |
+
"step": 1309
|
| 9205 |
+
},
|
| 9206 |
+
{
|
| 9207 |
+
"epoch": 0.004353594997686114,
|
| 9208 |
+
"grad_norm": 1.98556649684906,
|
| 9209 |
+
"learning_rate": 2.042243650189124e-07,
|
| 9210 |
+
"loss": 1.0463,
|
| 9211 |
+
"step": 1310
|
| 9212 |
+
},
|
| 9213 |
+
{
|
| 9214 |
+
"epoch": 0.00435691835264618,
|
| 9215 |
+
"grad_norm": 1.9155919551849365,
|
| 9216 |
+
"learning_rate": 1.8938146571413352e-07,
|
| 9217 |
+
"loss": 0.8835,
|
| 9218 |
+
"step": 1311
|
| 9219 |
+
},
|
| 9220 |
+
{
|
| 9221 |
+
"epoch": 0.004360241707606245,
|
| 9222 |
+
"grad_norm": 2.178790807723999,
|
| 9223 |
+
"learning_rate": 1.750979818428422e-07,
|
| 9224 |
+
"loss": 0.6642,
|
| 9225 |
+
"step": 1312
|
| 9226 |
+
},
|
| 9227 |
+
{
|
| 9228 |
+
"epoch": 0.004363565062566311,
|
| 9229 |
+
"grad_norm": 1.4291096925735474,
|
| 9230 |
+
"learning_rate": 1.6137399346064508e-07,
|
| 9231 |
+
"loss": 0.7367,
|
| 9232 |
+
"step": 1313
|
| 9233 |
+
},
|
| 9234 |
+
{
|
| 9235 |
+
"epoch": 0.004366888417526377,
|
| 9236 |
+
"grad_norm": 3.2246644496917725,
|
| 9237 |
+
"learning_rate": 1.4820957748733488e-07,
|
| 9238 |
+
"loss": 0.8849,
|
| 9239 |
+
"step": 1314
|
| 9240 |
+
},
|
| 9241 |
+
{
|
| 9242 |
+
"epoch": 0.004370211772486443,
|
| 9243 |
+
"grad_norm": 2.7322123050689697,
|
| 9244 |
+
"learning_rate": 1.3560480770642425e-07,
|
| 9245 |
+
"loss": 1.2786,
|
| 9246 |
+
"step": 1315
|
| 9247 |
+
},
|
| 9248 |
+
{
|
| 9249 |
+
"epoch": 0.004373535127446509,
|
| 9250 |
+
"grad_norm": 2.1076912879943848,
|
| 9251 |
+
"learning_rate": 1.235597547647127e-07,
|
| 9252 |
+
"loss": 1.0382,
|
| 9253 |
+
"step": 1316
|
| 9254 |
+
},
|
| 9255 |
+
{
|
| 9256 |
+
"epoch": 0.004376858482406574,
|
| 9257 |
+
"grad_norm": 2.021298885345459,
|
| 9258 |
+
"learning_rate": 1.120744861719536e-07,
|
| 9259 |
+
"loss": 0.7905,
|
| 9260 |
+
"step": 1317
|
| 9261 |
+
},
|
| 9262 |
+
{
|
| 9263 |
+
"epoch": 0.00438018183736664,
|
| 9264 |
+
"grad_norm": 2.3875844478607178,
|
| 9265 |
+
"learning_rate": 1.0114906630040999e-07,
|
| 9266 |
+
"loss": 0.7163,
|
| 9267 |
+
"step": 1318
|
| 9268 |
+
},
|
| 9269 |
+
{
|
| 9270 |
+
"epoch": 0.004383505192326706,
|
| 9271 |
+
"grad_norm": 1.4649884700775146,
|
| 9272 |
+
"learning_rate": 9.078355638453273e-08,
|
| 9273 |
+
"loss": 1.1184,
|
| 9274 |
+
"step": 1319
|
| 9275 |
+
},
|
| 9276 |
+
{
|
| 9277 |
+
"epoch": 0.004386828547286771,
|
| 9278 |
+
"grad_norm": 1.641730546951294,
|
| 9279 |
+
"learning_rate": 8.097801452061626e-08,
|
| 9280 |
+
"loss": 0.7022,
|
| 9281 |
+
"step": 1320
|
| 9282 |
+
},
|
| 9283 |
+
{
|
| 9284 |
+
"epoch": 0.004390151902246837,
|
| 9285 |
+
"grad_norm": 2.3859140872955322,
|
| 9286 |
+
"learning_rate": 7.173249566645446e-08,
|
| 9287 |
+
"loss": 1.2353,
|
| 9288 |
+
"step": 1321
|
| 9289 |
+
},
|
| 9290 |
+
{
|
| 9291 |
+
"epoch": 0.004393475257206903,
|
| 9292 |
+
"grad_norm": 1.9294756650924683,
|
| 9293 |
+
"learning_rate": 6.304705164104086e-08,
|
| 9294 |
+
"loss": 0.857,
|
| 9295 |
+
"step": 1322
|
| 9296 |
+
},
|
| 9297 |
+
{
|
| 9298 |
+
"epoch": 0.0043967986121669685,
|
| 9299 |
+
"grad_norm": 2.262242078781128,
|
| 9300 |
+
"learning_rate": 5.4921731124280007e-08,
|
| 9301 |
+
"loss": 0.8556,
|
| 9302 |
+
"step": 1323
|
| 9303 |
+
},
|
| 9304 |
+
{
|
| 9305 |
+
"epoch": 0.004400121967127035,
|
| 9306 |
+
"grad_norm": 1.5056575536727905,
|
| 9307 |
+
"learning_rate": 4.735657965672102e-08,
|
| 9308 |
+
"loss": 0.8407,
|
| 9309 |
+
"step": 1324
|
| 9310 |
+
},
|
| 9311 |
+
{
|
| 9312 |
+
"epoch": 0.0044034453220871,
|
| 9313 |
+
"grad_norm": 2.3266441822052,
|
| 9314 |
+
"learning_rate": 4.035163963926891e-08,
|
| 9315 |
+
"loss": 0.6501,
|
| 9316 |
+
"step": 1325
|
| 9317 |
+
},
|
| 9318 |
+
{
|
| 9319 |
+
"epoch": 0.004406768677047166,
|
| 9320 |
+
"grad_norm": 2.55777907371521,
|
| 9321 |
+
"learning_rate": 3.390695033300695e-08,
|
| 9322 |
+
"loss": 1.2284,
|
| 9323 |
+
"step": 1326
|
| 9324 |
+
},
|
| 9325 |
+
{
|
| 9326 |
+
"epoch": 0.004410092032007232,
|
| 9327 |
+
"grad_norm": 2.0342135429382324,
|
| 9328 |
+
"learning_rate": 2.8022547858930213e-08,
|
| 9329 |
+
"loss": 0.7973,
|
| 9330 |
+
"step": 1327
|
| 9331 |
+
},
|
| 9332 |
+
{
|
| 9333 |
+
"epoch": 0.004413415386967297,
|
| 9334 |
+
"grad_norm": 1.9018234014511108,
|
| 9335 |
+
"learning_rate": 2.2698465197745766e-08,
|
| 9336 |
+
"loss": 0.7389,
|
| 9337 |
+
"step": 1328
|
| 9338 |
+
},
|
| 9339 |
+
{
|
| 9340 |
+
"epoch": 0.004416738741927363,
|
| 9341 |
+
"grad_norm": 1.6029261350631714,
|
| 9342 |
+
"learning_rate": 1.7934732189706093e-08,
|
| 9343 |
+
"loss": 0.796,
|
| 9344 |
+
"step": 1329
|
| 9345 |
+
},
|
| 9346 |
+
{
|
| 9347 |
+
"epoch": 0.004420062096887428,
|
| 9348 |
+
"grad_norm": 1.9555233716964722,
|
| 9349 |
+
"learning_rate": 1.3731375534442593e-08,
|
| 9350 |
+
"loss": 0.5044,
|
| 9351 |
+
"step": 1330
|
| 9352 |
+
},
|
| 9353 |
+
{
|
| 9354 |
+
"epoch": 0.0044233854518474945,
|
| 9355 |
+
"grad_norm": 1.6823757886886597,
|
| 9356 |
+
"learning_rate": 1.0088418790787923e-08,
|
| 9357 |
+
"loss": 0.5384,
|
| 9358 |
+
"step": 1331
|
| 9359 |
+
},
|
| 9360 |
+
{
|
| 9361 |
+
"epoch": 0.0044267088068075605,
|
| 9362 |
+
"grad_norm": 1.7445098161697388,
|
| 9363 |
+
"learning_rate": 7.005882376676098e-09,
|
| 9364 |
+
"loss": 0.8662,
|
| 9365 |
+
"step": 1332
|
| 9366 |
+
},
|
| 9367 |
+
{
|
| 9368 |
+
"epoch": 0.004430032161767626,
|
| 9369 |
+
"grad_norm": 1.3536102771759033,
|
| 9370 |
+
"learning_rate": 4.483783568998145e-09,
|
| 9371 |
+
"loss": 0.6267,
|
| 9372 |
+
"step": 1333
|
| 9373 |
+
},
|
| 9374 |
+
{
|
| 9375 |
+
"epoch": 0.004433355516727692,
|
| 9376 |
+
"grad_norm": 2.434354543685913,
|
| 9377 |
+
"learning_rate": 2.5221365035577037e-09,
|
| 9378 |
+
"loss": 0.9421,
|
| 9379 |
+
"step": 1334
|
| 9380 |
+
},
|
| 9381 |
+
{
|
| 9382 |
+
"epoch": 0.004436678871687758,
|
| 9383 |
+
"grad_norm": 3.5411949157714844,
|
| 9384 |
+
"learning_rate": 1.1209521749155904e-09,
|
| 9385 |
+
"loss": 0.6256,
|
| 9386 |
+
"step": 1335
|
| 9387 |
+
},
|
| 9388 |
+
{
|
| 9389 |
+
"epoch": 0.004440002226647823,
|
| 9390 |
+
"grad_norm": 1.8648262023925781,
|
| 9391 |
+
"learning_rate": 2.8023843640090364e-10,
|
| 9392 |
+
"loss": 0.8315,
|
| 9393 |
+
"step": 1336
|
| 9394 |
+
},
|
| 9395 |
+
{
|
| 9396 |
+
"epoch": 0.004443325581607889,
|
| 9397 |
+
"grad_norm": 3.0049996376037598,
|
| 9398 |
+
"learning_rate": 0.0,
|
| 9399 |
+
"loss": 0.7712,
|
| 9400 |
+
"step": 1337
|
| 9401 |
}
|
| 9402 |
],
|
| 9403 |
"logging_steps": 1,
|
|
|
|
| 9412 |
"should_evaluate": false,
|
| 9413 |
"should_log": false,
|
| 9414 |
"should_save": true,
|
| 9415 |
+
"should_training_stop": true
|
| 9416 |
},
|
| 9417 |
"attributes": {}
|
| 9418 |
}
|
| 9419 |
},
|
| 9420 |
+
"total_flos": 1.018183511703552e+17,
|
| 9421 |
"train_batch_size": 2,
|
| 9422 |
"trial_name": null,
|
| 9423 |
"trial_params": null
|