Upload folder using huggingface_hub
Browse files- model-00001-of-00002.safetensors +1 -1
- model-00002-of-00002.safetensors +1 -1
- runs/Apr04_02-55-46_320d677d8fbf/events.out.tfevents.1743735349.320d677d8fbf.4053.0 +3 -0
- runs/Apr04_03-03-34_320d677d8fbf/events.out.tfevents.1743735817.320d677d8fbf.6876.0 +3 -0
- runs/Apr04_03-04-20_320d677d8fbf/events.out.tfevents.1743735863.320d677d8fbf.28090.0 +3 -0
- runs/Apr04_03-08-14_320d677d8fbf/events.out.tfevents.1743736097.320d677d8fbf.187095.0 +3 -0
- trainer_state.json +2463 -4
- training_args.bin +3 -0
model-00001-of-00002.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4938446392
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:62eaf2329eca44f22abaccd1f5f524b187c8f984c9aac7c1dc56eff75a82d671
|
| 3 |
size 4938446392
|
model-00002-of-00002.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 3821736024
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:41b049df65dea08166e7f563f856c60efd28c6710065880ca26c25fcda1ca4a5
|
| 3 |
size 3821736024
|
runs/Apr04_02-55-46_320d677d8fbf/events.out.tfevents.1743735349.320d677d8fbf.4053.0
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f2f002a72d324bed426996133723aa248fdbe8ff023c42c8a192f3d195032030
|
| 3 |
+
size 6158
|
runs/Apr04_03-03-34_320d677d8fbf/events.out.tfevents.1743735817.320d677d8fbf.6876.0
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9c305d00a9c0e40286408f92d6596a233011790bbe4659698ceb674b3c477cc3
|
| 3 |
+
size 6779
|
runs/Apr04_03-04-20_320d677d8fbf/events.out.tfevents.1743735863.320d677d8fbf.28090.0
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8e233659f46eb6a1707541f7355192dc0b0484455a8868b90ca82fb5aef09a3a
|
| 3 |
+
size 14338
|
runs/Apr04_03-08-14_320d677d8fbf/events.out.tfevents.1743736097.320d677d8fbf.187095.0
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c390d8c71846c2f48c04ce6d5accf75ef1d4d8dc8645e8ceae2f02c771f1fc0b
|
| 3 |
+
size 217463
|
trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch":
|
| 5 |
"eval_steps": 500,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -4557,6 +4557,2465 @@
|
|
| 4557 |
"learning_rate": 2.991522876735154e-05,
|
| 4558 |
"loss": 0.0065,
|
| 4559 |
"step": 6500
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 4560 |
}
|
| 4561 |
],
|
| 4562 |
"logging_steps": 10,
|
|
@@ -4571,12 +7030,12 @@
|
|
| 4571 |
"should_evaluate": false,
|
| 4572 |
"should_log": false,
|
| 4573 |
"should_save": true,
|
| 4574 |
-
"should_training_stop":
|
| 4575 |
},
|
| 4576 |
"attributes": {}
|
| 4577 |
}
|
| 4578 |
},
|
| 4579 |
-
"total_flos":
|
| 4580 |
"train_batch_size": 16,
|
| 4581 |
"trial_name": null,
|
| 4582 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 6.337135614702155,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 10000,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 4557 |
"learning_rate": 2.991522876735154e-05,
|
| 4558 |
"loss": 0.0065,
|
| 4559 |
"step": 6500
|
| 4560 |
+
},
|
| 4561 |
+
{
|
| 4562 |
+
"epoch": 4.1254752851711025,
|
| 4563 |
+
"grad_norm": 0.1556195318698883,
|
| 4564 |
+
"learning_rate": 2.976391850971065e-05,
|
| 4565 |
+
"loss": 0.0155,
|
| 4566 |
+
"step": 6510
|
| 4567 |
+
},
|
| 4568 |
+
{
|
| 4569 |
+
"epoch": 4.1318124207858045,
|
| 4570 |
+
"grad_norm": 0.24738621711730957,
|
| 4571 |
+
"learning_rate": 2.9612829550614836e-05,
|
| 4572 |
+
"loss": 0.0107,
|
| 4573 |
+
"step": 6520
|
| 4574 |
+
},
|
| 4575 |
+
{
|
| 4576 |
+
"epoch": 4.138149556400507,
|
| 4577 |
+
"grad_norm": 0.2733817398548126,
|
| 4578 |
+
"learning_rate": 2.9461963542348737e-05,
|
| 4579 |
+
"loss": 0.0083,
|
| 4580 |
+
"step": 6530
|
| 4581 |
+
},
|
| 4582 |
+
{
|
| 4583 |
+
"epoch": 4.144486692015209,
|
| 4584 |
+
"grad_norm": 0.4232415556907654,
|
| 4585 |
+
"learning_rate": 2.931132213475884e-05,
|
| 4586 |
+
"loss": 0.0154,
|
| 4587 |
+
"step": 6540
|
| 4588 |
+
},
|
| 4589 |
+
{
|
| 4590 |
+
"epoch": 4.150823827629911,
|
| 4591 |
+
"grad_norm": 0.16608037054538727,
|
| 4592 |
+
"learning_rate": 2.916090697523549e-05,
|
| 4593 |
+
"loss": 0.0117,
|
| 4594 |
+
"step": 6550
|
| 4595 |
+
},
|
| 4596 |
+
{
|
| 4597 |
+
"epoch": 4.157160963244613,
|
| 4598 |
+
"grad_norm": 0.226557657122612,
|
| 4599 |
+
"learning_rate": 2.9010719708694722e-05,
|
| 4600 |
+
"loss": 0.0087,
|
| 4601 |
+
"step": 6560
|
| 4602 |
+
},
|
| 4603 |
+
{
|
| 4604 |
+
"epoch": 4.163498098859316,
|
| 4605 |
+
"grad_norm": 0.34729430079460144,
|
| 4606 |
+
"learning_rate": 2.8860761977560436e-05,
|
| 4607 |
+
"loss": 0.012,
|
| 4608 |
+
"step": 6570
|
| 4609 |
+
},
|
| 4610 |
+
{
|
| 4611 |
+
"epoch": 4.169835234474018,
|
| 4612 |
+
"grad_norm": 0.2842123806476593,
|
| 4613 |
+
"learning_rate": 2.8711035421746367e-05,
|
| 4614 |
+
"loss": 0.0109,
|
| 4615 |
+
"step": 6580
|
| 4616 |
+
},
|
| 4617 |
+
{
|
| 4618 |
+
"epoch": 4.17617237008872,
|
| 4619 |
+
"grad_norm": 0.24015074968338013,
|
| 4620 |
+
"learning_rate": 2.8561541678638142e-05,
|
| 4621 |
+
"loss": 0.01,
|
| 4622 |
+
"step": 6590
|
| 4623 |
+
},
|
| 4624 |
+
{
|
| 4625 |
+
"epoch": 4.182509505703422,
|
| 4626 |
+
"grad_norm": 0.2127242535352707,
|
| 4627 |
+
"learning_rate": 2.8412282383075363e-05,
|
| 4628 |
+
"loss": 0.0092,
|
| 4629 |
+
"step": 6600
|
| 4630 |
+
},
|
| 4631 |
+
{
|
| 4632 |
+
"epoch": 4.188846641318124,
|
| 4633 |
+
"grad_norm": 0.1556386947631836,
|
| 4634 |
+
"learning_rate": 2.8263259167333777e-05,
|
| 4635 |
+
"loss": 0.0066,
|
| 4636 |
+
"step": 6610
|
| 4637 |
+
},
|
| 4638 |
+
{
|
| 4639 |
+
"epoch": 4.195183776932827,
|
| 4640 |
+
"grad_norm": 0.2527133524417877,
|
| 4641 |
+
"learning_rate": 2.811447366110741e-05,
|
| 4642 |
+
"loss": 0.0121,
|
| 4643 |
+
"step": 6620
|
| 4644 |
+
},
|
| 4645 |
+
{
|
| 4646 |
+
"epoch": 4.201520912547529,
|
| 4647 |
+
"grad_norm": 0.13160255551338196,
|
| 4648 |
+
"learning_rate": 2.7965927491490705e-05,
|
| 4649 |
+
"loss": 0.0069,
|
| 4650 |
+
"step": 6630
|
| 4651 |
+
},
|
| 4652 |
+
{
|
| 4653 |
+
"epoch": 4.2078580481622305,
|
| 4654 |
+
"grad_norm": 0.29007360339164734,
|
| 4655 |
+
"learning_rate": 2.7817622282960815e-05,
|
| 4656 |
+
"loss": 0.0107,
|
| 4657 |
+
"step": 6640
|
| 4658 |
+
},
|
| 4659 |
+
{
|
| 4660 |
+
"epoch": 4.2141951837769325,
|
| 4661 |
+
"grad_norm": 0.09741051495075226,
|
| 4662 |
+
"learning_rate": 2.766955965735968e-05,
|
| 4663 |
+
"loss": 0.0069,
|
| 4664 |
+
"step": 6650
|
| 4665 |
+
},
|
| 4666 |
+
{
|
| 4667 |
+
"epoch": 4.220532319391635,
|
| 4668 |
+
"grad_norm": 0.2465200573205948,
|
| 4669 |
+
"learning_rate": 2.7521741233876496e-05,
|
| 4670 |
+
"loss": 0.0105,
|
| 4671 |
+
"step": 6660
|
| 4672 |
+
},
|
| 4673 |
+
{
|
| 4674 |
+
"epoch": 4.226869455006337,
|
| 4675 |
+
"grad_norm": 0.18087448179721832,
|
| 4676 |
+
"learning_rate": 2.7374168629029813e-05,
|
| 4677 |
+
"loss": 0.0122,
|
| 4678 |
+
"step": 6670
|
| 4679 |
+
},
|
| 4680 |
+
{
|
| 4681 |
+
"epoch": 4.233206590621039,
|
| 4682 |
+
"grad_norm": 0.15381130576133728,
|
| 4683 |
+
"learning_rate": 2.7226843456650037e-05,
|
| 4684 |
+
"loss": 0.0091,
|
| 4685 |
+
"step": 6680
|
| 4686 |
+
},
|
| 4687 |
+
{
|
| 4688 |
+
"epoch": 4.239543726235741,
|
| 4689 |
+
"grad_norm": 0.16442646086215973,
|
| 4690 |
+
"learning_rate": 2.707976732786166e-05,
|
| 4691 |
+
"loss": 0.0094,
|
| 4692 |
+
"step": 6690
|
| 4693 |
+
},
|
| 4694 |
+
{
|
| 4695 |
+
"epoch": 4.245880861850444,
|
| 4696 |
+
"grad_norm": 0.14896883070468903,
|
| 4697 |
+
"learning_rate": 2.693294185106562e-05,
|
| 4698 |
+
"loss": 0.0158,
|
| 4699 |
+
"step": 6700
|
| 4700 |
+
},
|
| 4701 |
+
{
|
| 4702 |
+
"epoch": 4.252217997465146,
|
| 4703 |
+
"grad_norm": 0.1330922693014145,
|
| 4704 |
+
"learning_rate": 2.6786368631921836e-05,
|
| 4705 |
+
"loss": 0.0097,
|
| 4706 |
+
"step": 6710
|
| 4707 |
+
},
|
| 4708 |
+
{
|
| 4709 |
+
"epoch": 4.258555133079848,
|
| 4710 |
+
"grad_norm": 0.18497434258460999,
|
| 4711 |
+
"learning_rate": 2.6640049273331515e-05,
|
| 4712 |
+
"loss": 0.0082,
|
| 4713 |
+
"step": 6720
|
| 4714 |
+
},
|
| 4715 |
+
{
|
| 4716 |
+
"epoch": 4.26489226869455,
|
| 4717 |
+
"grad_norm": 0.12878283858299255,
|
| 4718 |
+
"learning_rate": 2.6493985375419778e-05,
|
| 4719 |
+
"loss": 0.0141,
|
| 4720 |
+
"step": 6730
|
| 4721 |
+
},
|
| 4722 |
+
{
|
| 4723 |
+
"epoch": 4.271229404309253,
|
| 4724 |
+
"grad_norm": 0.1790645867586136,
|
| 4725 |
+
"learning_rate": 2.6348178535517966e-05,
|
| 4726 |
+
"loss": 0.012,
|
| 4727 |
+
"step": 6740
|
| 4728 |
+
},
|
| 4729 |
+
{
|
| 4730 |
+
"epoch": 4.277566539923955,
|
| 4731 |
+
"grad_norm": 0.29659292101860046,
|
| 4732 |
+
"learning_rate": 2.6202630348146324e-05,
|
| 4733 |
+
"loss": 0.0141,
|
| 4734 |
+
"step": 6750
|
| 4735 |
+
},
|
| 4736 |
+
{
|
| 4737 |
+
"epoch": 4.283903675538657,
|
| 4738 |
+
"grad_norm": 0.2492123246192932,
|
| 4739 |
+
"learning_rate": 2.6057342404996522e-05,
|
| 4740 |
+
"loss": 0.0133,
|
| 4741 |
+
"step": 6760
|
| 4742 |
+
},
|
| 4743 |
+
{
|
| 4744 |
+
"epoch": 4.2902408111533585,
|
| 4745 |
+
"grad_norm": 0.11853443086147308,
|
| 4746 |
+
"learning_rate": 2.591231629491423e-05,
|
| 4747 |
+
"loss": 0.01,
|
| 4748 |
+
"step": 6770
|
| 4749 |
+
},
|
| 4750 |
+
{
|
| 4751 |
+
"epoch": 4.2965779467680605,
|
| 4752 |
+
"grad_norm": 0.24464382231235504,
|
| 4753 |
+
"learning_rate": 2.5767553603881767e-05,
|
| 4754 |
+
"loss": 0.0088,
|
| 4755 |
+
"step": 6780
|
| 4756 |
+
},
|
| 4757 |
+
{
|
| 4758 |
+
"epoch": 4.302915082382763,
|
| 4759 |
+
"grad_norm": 0.2709729075431824,
|
| 4760 |
+
"learning_rate": 2.562305591500069e-05,
|
| 4761 |
+
"loss": 0.0075,
|
| 4762 |
+
"step": 6790
|
| 4763 |
+
},
|
| 4764 |
+
{
|
| 4765 |
+
"epoch": 4.309252217997465,
|
| 4766 |
+
"grad_norm": 0.19660340249538422,
|
| 4767 |
+
"learning_rate": 2.547882480847461e-05,
|
| 4768 |
+
"loss": 0.0136,
|
| 4769 |
+
"step": 6800
|
| 4770 |
+
},
|
| 4771 |
+
{
|
| 4772 |
+
"epoch": 4.315589353612167,
|
| 4773 |
+
"grad_norm": 0.21766068041324615,
|
| 4774 |
+
"learning_rate": 2.5334861861591753e-05,
|
| 4775 |
+
"loss": 0.0073,
|
| 4776 |
+
"step": 6810
|
| 4777 |
+
},
|
| 4778 |
+
{
|
| 4779 |
+
"epoch": 4.321926489226869,
|
| 4780 |
+
"grad_norm": 0.27058249711990356,
|
| 4781 |
+
"learning_rate": 2.5191168648707887e-05,
|
| 4782 |
+
"loss": 0.0128,
|
| 4783 |
+
"step": 6820
|
| 4784 |
+
},
|
| 4785 |
+
{
|
| 4786 |
+
"epoch": 4.328263624841572,
|
| 4787 |
+
"grad_norm": 0.18688562512397766,
|
| 4788 |
+
"learning_rate": 2.5047746741228978e-05,
|
| 4789 |
+
"loss": 0.0146,
|
| 4790 |
+
"step": 6830
|
| 4791 |
+
},
|
| 4792 |
+
{
|
| 4793 |
+
"epoch": 4.334600760456274,
|
| 4794 |
+
"grad_norm": 0.23607704043388367,
|
| 4795 |
+
"learning_rate": 2.490459770759398e-05,
|
| 4796 |
+
"loss": 0.0089,
|
| 4797 |
+
"step": 6840
|
| 4798 |
+
},
|
| 4799 |
+
{
|
| 4800 |
+
"epoch": 4.340937896070976,
|
| 4801 |
+
"grad_norm": 0.1704394668340683,
|
| 4802 |
+
"learning_rate": 2.476172311325783e-05,
|
| 4803 |
+
"loss": 0.009,
|
| 4804 |
+
"step": 6850
|
| 4805 |
+
},
|
| 4806 |
+
{
|
| 4807 |
+
"epoch": 4.347275031685678,
|
| 4808 |
+
"grad_norm": 0.11735841631889343,
|
| 4809 |
+
"learning_rate": 2.4619124520674146e-05,
|
| 4810 |
+
"loss": 0.0099,
|
| 4811 |
+
"step": 6860
|
| 4812 |
+
},
|
| 4813 |
+
{
|
| 4814 |
+
"epoch": 4.35361216730038,
|
| 4815 |
+
"grad_norm": 0.17828071117401123,
|
| 4816 |
+
"learning_rate": 2.447680348927837e-05,
|
| 4817 |
+
"loss": 0.0069,
|
| 4818 |
+
"step": 6870
|
| 4819 |
+
},
|
| 4820 |
+
{
|
| 4821 |
+
"epoch": 4.359949302915083,
|
| 4822 |
+
"grad_norm": 0.09438079595565796,
|
| 4823 |
+
"learning_rate": 2.433476157547044e-05,
|
| 4824 |
+
"loss": 0.0077,
|
| 4825 |
+
"step": 6880
|
| 4826 |
+
},
|
| 4827 |
+
{
|
| 4828 |
+
"epoch": 4.366286438529785,
|
| 4829 |
+
"grad_norm": 0.14729665219783783,
|
| 4830 |
+
"learning_rate": 2.419300033259798e-05,
|
| 4831 |
+
"loss": 0.0131,
|
| 4832 |
+
"step": 6890
|
| 4833 |
+
},
|
| 4834 |
+
{
|
| 4835 |
+
"epoch": 4.3726235741444865,
|
| 4836 |
+
"grad_norm": 0.21107903122901917,
|
| 4837 |
+
"learning_rate": 2.405152131093926e-05,
|
| 4838 |
+
"loss": 0.0089,
|
| 4839 |
+
"step": 6900
|
| 4840 |
+
},
|
| 4841 |
+
{
|
| 4842 |
+
"epoch": 4.3789607097591885,
|
| 4843 |
+
"grad_norm": 0.26359254121780396,
|
| 4844 |
+
"learning_rate": 2.3910326057686127e-05,
|
| 4845 |
+
"loss": 0.012,
|
| 4846 |
+
"step": 6910
|
| 4847 |
+
},
|
| 4848 |
+
{
|
| 4849 |
+
"epoch": 4.385297845373891,
|
| 4850 |
+
"grad_norm": 0.22561860084533691,
|
| 4851 |
+
"learning_rate": 2.3769416116927335e-05,
|
| 4852 |
+
"loss": 0.0112,
|
| 4853 |
+
"step": 6920
|
| 4854 |
+
},
|
| 4855 |
+
{
|
| 4856 |
+
"epoch": 4.391634980988593,
|
| 4857 |
+
"grad_norm": 0.253752201795578,
|
| 4858 |
+
"learning_rate": 2.362879302963135e-05,
|
| 4859 |
+
"loss": 0.0087,
|
| 4860 |
+
"step": 6930
|
| 4861 |
+
},
|
| 4862 |
+
{
|
| 4863 |
+
"epoch": 4.397972116603295,
|
| 4864 |
+
"grad_norm": 0.1931048333644867,
|
| 4865 |
+
"learning_rate": 2.3488458333629777e-05,
|
| 4866 |
+
"loss": 0.0078,
|
| 4867 |
+
"step": 6940
|
| 4868 |
+
},
|
| 4869 |
+
{
|
| 4870 |
+
"epoch": 4.404309252217997,
|
| 4871 |
+
"grad_norm": 0.13094092905521393,
|
| 4872 |
+
"learning_rate": 2.3348413563600325e-05,
|
| 4873 |
+
"loss": 0.0088,
|
| 4874 |
+
"step": 6950
|
| 4875 |
+
},
|
| 4876 |
+
{
|
| 4877 |
+
"epoch": 4.4106463878327,
|
| 4878 |
+
"grad_norm": 0.14549599587917328,
|
| 4879 |
+
"learning_rate": 2.3208660251050158e-05,
|
| 4880 |
+
"loss": 0.0099,
|
| 4881 |
+
"step": 6960
|
| 4882 |
+
},
|
| 4883 |
+
{
|
| 4884 |
+
"epoch": 4.416983523447402,
|
| 4885 |
+
"grad_norm": 0.10897214710712433,
|
| 4886 |
+
"learning_rate": 2.3069199924299174e-05,
|
| 4887 |
+
"loss": 0.0084,
|
| 4888 |
+
"step": 6970
|
| 4889 |
+
},
|
| 4890 |
+
{
|
| 4891 |
+
"epoch": 4.423320659062104,
|
| 4892 |
+
"grad_norm": 0.13430652022361755,
|
| 4893 |
+
"learning_rate": 2.29300341084631e-05,
|
| 4894 |
+
"loss": 0.0076,
|
| 4895 |
+
"step": 6980
|
| 4896 |
+
},
|
| 4897 |
+
{
|
| 4898 |
+
"epoch": 4.429657794676806,
|
| 4899 |
+
"grad_norm": 0.3317708969116211,
|
| 4900 |
+
"learning_rate": 2.279116432543705e-05,
|
| 4901 |
+
"loss": 0.0178,
|
| 4902 |
+
"step": 6990
|
| 4903 |
+
},
|
| 4904 |
+
{
|
| 4905 |
+
"epoch": 4.435994930291509,
|
| 4906 |
+
"grad_norm": 0.29772767424583435,
|
| 4907 |
+
"learning_rate": 2.2652592093878666e-05,
|
| 4908 |
+
"loss": 0.012,
|
| 4909 |
+
"step": 7000
|
| 4910 |
+
},
|
| 4911 |
+
{
|
| 4912 |
+
"epoch": 4.442332065906211,
|
| 4913 |
+
"grad_norm": 0.287555068731308,
|
| 4914 |
+
"learning_rate": 2.251431892919171e-05,
|
| 4915 |
+
"loss": 0.0131,
|
| 4916 |
+
"step": 7010
|
| 4917 |
+
},
|
| 4918 |
+
{
|
| 4919 |
+
"epoch": 4.448669201520913,
|
| 4920 |
+
"grad_norm": 0.22690831124782562,
|
| 4921 |
+
"learning_rate": 2.237634634350934e-05,
|
| 4922 |
+
"loss": 0.0105,
|
| 4923 |
+
"step": 7020
|
| 4924 |
+
},
|
| 4925 |
+
{
|
| 4926 |
+
"epoch": 4.455006337135615,
|
| 4927 |
+
"grad_norm": 0.2328868806362152,
|
| 4928 |
+
"learning_rate": 2.2238675845677663e-05,
|
| 4929 |
+
"loss": 0.011,
|
| 4930 |
+
"step": 7030
|
| 4931 |
+
},
|
| 4932 |
+
{
|
| 4933 |
+
"epoch": 4.4613434727503165,
|
| 4934 |
+
"grad_norm": 0.17060483992099762,
|
| 4935 |
+
"learning_rate": 2.2101308941239203e-05,
|
| 4936 |
+
"loss": 0.0099,
|
| 4937 |
+
"step": 7040
|
| 4938 |
+
},
|
| 4939 |
+
{
|
| 4940 |
+
"epoch": 4.467680608365019,
|
| 4941 |
+
"grad_norm": 0.3242473006248474,
|
| 4942 |
+
"learning_rate": 2.196424713241637e-05,
|
| 4943 |
+
"loss": 0.0107,
|
| 4944 |
+
"step": 7050
|
| 4945 |
+
},
|
| 4946 |
+
{
|
| 4947 |
+
"epoch": 4.474017743979721,
|
| 4948 |
+
"grad_norm": 0.2539024353027344,
|
| 4949 |
+
"learning_rate": 2.182749191809518e-05,
|
| 4950 |
+
"loss": 0.0069,
|
| 4951 |
+
"step": 7060
|
| 4952 |
+
},
|
| 4953 |
+
{
|
| 4954 |
+
"epoch": 4.480354879594423,
|
| 4955 |
+
"grad_norm": 0.2088938057422638,
|
| 4956 |
+
"learning_rate": 2.1691044793808734e-05,
|
| 4957 |
+
"loss": 0.0118,
|
| 4958 |
+
"step": 7070
|
| 4959 |
+
},
|
| 4960 |
+
{
|
| 4961 |
+
"epoch": 4.486692015209125,
|
| 4962 |
+
"grad_norm": 0.24568264186382294,
|
| 4963 |
+
"learning_rate": 2.1554907251720945e-05,
|
| 4964 |
+
"loss": 0.0091,
|
| 4965 |
+
"step": 7080
|
| 4966 |
+
},
|
| 4967 |
+
{
|
| 4968 |
+
"epoch": 4.493029150823828,
|
| 4969 |
+
"grad_norm": 0.14442333579063416,
|
| 4970 |
+
"learning_rate": 2.1419080780610123e-05,
|
| 4971 |
+
"loss": 0.0088,
|
| 4972 |
+
"step": 7090
|
| 4973 |
+
},
|
| 4974 |
+
{
|
| 4975 |
+
"epoch": 4.49936628643853,
|
| 4976 |
+
"grad_norm": 0.22089751064777374,
|
| 4977 |
+
"learning_rate": 2.128356686585282e-05,
|
| 4978 |
+
"loss": 0.013,
|
| 4979 |
+
"step": 7100
|
| 4980 |
+
},
|
| 4981 |
+
{
|
| 4982 |
+
"epoch": 4.505703422053232,
|
| 4983 |
+
"grad_norm": 0.1270669400691986,
|
| 4984 |
+
"learning_rate": 2.1148366989407496e-05,
|
| 4985 |
+
"loss": 0.0113,
|
| 4986 |
+
"step": 7110
|
| 4987 |
+
},
|
| 4988 |
+
{
|
| 4989 |
+
"epoch": 4.512040557667934,
|
| 4990 |
+
"grad_norm": 0.11936408281326294,
|
| 4991 |
+
"learning_rate": 2.1013482629798333e-05,
|
| 4992 |
+
"loss": 0.0072,
|
| 4993 |
+
"step": 7120
|
| 4994 |
+
},
|
| 4995 |
+
{
|
| 4996 |
+
"epoch": 4.518377693282636,
|
| 4997 |
+
"grad_norm": 0.1418917030096054,
|
| 4998 |
+
"learning_rate": 2.0878915262099098e-05,
|
| 4999 |
+
"loss": 0.0097,
|
| 5000 |
+
"step": 7130
|
| 5001 |
+
},
|
| 5002 |
+
{
|
| 5003 |
+
"epoch": 4.524714828897339,
|
| 5004 |
+
"grad_norm": 0.22373194992542267,
|
| 5005 |
+
"learning_rate": 2.0744666357916925e-05,
|
| 5006 |
+
"loss": 0.0107,
|
| 5007 |
+
"step": 7140
|
| 5008 |
+
},
|
| 5009 |
+
{
|
| 5010 |
+
"epoch": 4.531051964512041,
|
| 5011 |
+
"grad_norm": 0.16027112305164337,
|
| 5012 |
+
"learning_rate": 2.061073738537635e-05,
|
| 5013 |
+
"loss": 0.0076,
|
| 5014 |
+
"step": 7150
|
| 5015 |
+
},
|
| 5016 |
+
{
|
| 5017 |
+
"epoch": 4.537389100126743,
|
| 5018 |
+
"grad_norm": 0.205812469124794,
|
| 5019 |
+
"learning_rate": 2.0477129809103147e-05,
|
| 5020 |
+
"loss": 0.0156,
|
| 5021 |
+
"step": 7160
|
| 5022 |
+
},
|
| 5023 |
+
{
|
| 5024 |
+
"epoch": 4.5437262357414445,
|
| 5025 |
+
"grad_norm": 0.07858346402645111,
|
| 5026 |
+
"learning_rate": 2.0343845090208368e-05,
|
| 5027 |
+
"loss": 0.0069,
|
| 5028 |
+
"step": 7170
|
| 5029 |
+
},
|
| 5030 |
+
{
|
| 5031 |
+
"epoch": 4.550063371356147,
|
| 5032 |
+
"grad_norm": 0.173725426197052,
|
| 5033 |
+
"learning_rate": 2.0210884686272368e-05,
|
| 5034 |
+
"loss": 0.0107,
|
| 5035 |
+
"step": 7180
|
| 5036 |
+
},
|
| 5037 |
+
{
|
| 5038 |
+
"epoch": 4.556400506970849,
|
| 5039 |
+
"grad_norm": 0.14513538777828217,
|
| 5040 |
+
"learning_rate": 2.0078250051328784e-05,
|
| 5041 |
+
"loss": 0.0096,
|
| 5042 |
+
"step": 7190
|
| 5043 |
+
},
|
| 5044 |
+
{
|
| 5045 |
+
"epoch": 4.562737642585551,
|
| 5046 |
+
"grad_norm": 0.14513027667999268,
|
| 5047 |
+
"learning_rate": 1.9945942635848748e-05,
|
| 5048 |
+
"loss": 0.0155,
|
| 5049 |
+
"step": 7200
|
| 5050 |
+
},
|
| 5051 |
+
{
|
| 5052 |
+
"epoch": 4.569074778200253,
|
| 5053 |
+
"grad_norm": 0.12547020614147186,
|
| 5054 |
+
"learning_rate": 1.981396388672496e-05,
|
| 5055 |
+
"loss": 0.007,
|
| 5056 |
+
"step": 7210
|
| 5057 |
+
},
|
| 5058 |
+
{
|
| 5059 |
+
"epoch": 4.575411913814955,
|
| 5060 |
+
"grad_norm": 0.09598460793495178,
|
| 5061 |
+
"learning_rate": 1.9682315247255894e-05,
|
| 5062 |
+
"loss": 0.0077,
|
| 5063 |
+
"step": 7220
|
| 5064 |
+
},
|
| 5065 |
+
{
|
| 5066 |
+
"epoch": 4.581749049429658,
|
| 5067 |
+
"grad_norm": 0.09776268899440765,
|
| 5068 |
+
"learning_rate": 1.9550998157129946e-05,
|
| 5069 |
+
"loss": 0.0088,
|
| 5070 |
+
"step": 7230
|
| 5071 |
+
},
|
| 5072 |
+
{
|
| 5073 |
+
"epoch": 4.58808618504436,
|
| 5074 |
+
"grad_norm": 0.14646144211292267,
|
| 5075 |
+
"learning_rate": 1.942001405240979e-05,
|
| 5076 |
+
"loss": 0.0066,
|
| 5077 |
+
"step": 7240
|
| 5078 |
+
},
|
| 5079 |
+
{
|
| 5080 |
+
"epoch": 4.594423320659062,
|
| 5081 |
+
"grad_norm": 0.09714064747095108,
|
| 5082 |
+
"learning_rate": 1.928936436551661e-05,
|
| 5083 |
+
"loss": 0.0068,
|
| 5084 |
+
"step": 7250
|
| 5085 |
+
},
|
| 5086 |
+
{
|
| 5087 |
+
"epoch": 4.600760456273765,
|
| 5088 |
+
"grad_norm": 0.09656299650669098,
|
| 5089 |
+
"learning_rate": 1.9159050525214452e-05,
|
| 5090 |
+
"loss": 0.0088,
|
| 5091 |
+
"step": 7260
|
| 5092 |
+
},
|
| 5093 |
+
{
|
| 5094 |
+
"epoch": 4.607097591888467,
|
| 5095 |
+
"grad_norm": 0.11101258546113968,
|
| 5096 |
+
"learning_rate": 1.9029073956594606e-05,
|
| 5097 |
+
"loss": 0.0125,
|
| 5098 |
+
"step": 7270
|
| 5099 |
+
},
|
| 5100 |
+
{
|
| 5101 |
+
"epoch": 4.613434727503169,
|
| 5102 |
+
"grad_norm": 0.24712498486042023,
|
| 5103 |
+
"learning_rate": 1.8899436081059975e-05,
|
| 5104 |
+
"loss": 0.0079,
|
| 5105 |
+
"step": 7280
|
| 5106 |
+
},
|
| 5107 |
+
{
|
| 5108 |
+
"epoch": 4.619771863117871,
|
| 5109 |
+
"grad_norm": 0.27654388546943665,
|
| 5110 |
+
"learning_rate": 1.877013831630961e-05,
|
| 5111 |
+
"loss": 0.0086,
|
| 5112 |
+
"step": 7290
|
| 5113 |
+
},
|
| 5114 |
+
{
|
| 5115 |
+
"epoch": 4.6261089987325725,
|
| 5116 |
+
"grad_norm": 0.30745381116867065,
|
| 5117 |
+
"learning_rate": 1.8641182076323148e-05,
|
| 5118 |
+
"loss": 0.0115,
|
| 5119 |
+
"step": 7300
|
| 5120 |
+
},
|
| 5121 |
+
{
|
| 5122 |
+
"epoch": 4.632446134347275,
|
| 5123 |
+
"grad_norm": 0.21805402636528015,
|
| 5124 |
+
"learning_rate": 1.851256877134538e-05,
|
| 5125 |
+
"loss": 0.0086,
|
| 5126 |
+
"step": 7310
|
| 5127 |
+
},
|
| 5128 |
+
{
|
| 5129 |
+
"epoch": 4.638783269961977,
|
| 5130 |
+
"grad_norm": 0.14376069605350494,
|
| 5131 |
+
"learning_rate": 1.838429980787081e-05,
|
| 5132 |
+
"loss": 0.0078,
|
| 5133 |
+
"step": 7320
|
| 5134 |
+
},
|
| 5135 |
+
{
|
| 5136 |
+
"epoch": 4.645120405576679,
|
| 5137 |
+
"grad_norm": 0.15503360331058502,
|
| 5138 |
+
"learning_rate": 1.8256376588628238e-05,
|
| 5139 |
+
"loss": 0.0061,
|
| 5140 |
+
"step": 7330
|
| 5141 |
+
},
|
| 5142 |
+
{
|
| 5143 |
+
"epoch": 4.651457541191381,
|
| 5144 |
+
"grad_norm": 0.21702007949352264,
|
| 5145 |
+
"learning_rate": 1.8128800512565513e-05,
|
| 5146 |
+
"loss": 0.0078,
|
| 5147 |
+
"step": 7340
|
| 5148 |
+
},
|
| 5149 |
+
{
|
| 5150 |
+
"epoch": 4.657794676806084,
|
| 5151 |
+
"grad_norm": 0.16204503178596497,
|
| 5152 |
+
"learning_rate": 1.800157297483417e-05,
|
| 5153 |
+
"loss": 0.0084,
|
| 5154 |
+
"step": 7350
|
| 5155 |
+
},
|
| 5156 |
+
{
|
| 5157 |
+
"epoch": 4.664131812420786,
|
| 5158 |
+
"grad_norm": 0.16800753772258759,
|
| 5159 |
+
"learning_rate": 1.787469536677419e-05,
|
| 5160 |
+
"loss": 0.0123,
|
| 5161 |
+
"step": 7360
|
| 5162 |
+
},
|
| 5163 |
+
{
|
| 5164 |
+
"epoch": 4.670468948035488,
|
| 5165 |
+
"grad_norm": 0.26554012298583984,
|
| 5166 |
+
"learning_rate": 1.774816907589873e-05,
|
| 5167 |
+
"loss": 0.0155,
|
| 5168 |
+
"step": 7370
|
| 5169 |
+
},
|
| 5170 |
+
{
|
| 5171 |
+
"epoch": 4.67680608365019,
|
| 5172 |
+
"grad_norm": 0.18996752798557281,
|
| 5173 |
+
"learning_rate": 1.7621995485879062e-05,
|
| 5174 |
+
"loss": 0.0093,
|
| 5175 |
+
"step": 7380
|
| 5176 |
+
},
|
| 5177 |
+
{
|
| 5178 |
+
"epoch": 4.683143219264892,
|
| 5179 |
+
"grad_norm": 0.08091560751199722,
|
| 5180 |
+
"learning_rate": 1.749617597652934e-05,
|
| 5181 |
+
"loss": 0.0087,
|
| 5182 |
+
"step": 7390
|
| 5183 |
+
},
|
| 5184 |
+
{
|
| 5185 |
+
"epoch": 4.689480354879595,
|
| 5186 |
+
"grad_norm": 0.16652986407279968,
|
| 5187 |
+
"learning_rate": 1.7370711923791567e-05,
|
| 5188 |
+
"loss": 0.0068,
|
| 5189 |
+
"step": 7400
|
| 5190 |
+
},
|
| 5191 |
+
{
|
| 5192 |
+
"epoch": 4.695817490494297,
|
| 5193 |
+
"grad_norm": 0.06975802779197693,
|
| 5194 |
+
"learning_rate": 1.7245604699720535e-05,
|
| 5195 |
+
"loss": 0.0098,
|
| 5196 |
+
"step": 7410
|
| 5197 |
+
},
|
| 5198 |
+
{
|
| 5199 |
+
"epoch": 4.702154626108999,
|
| 5200 |
+
"grad_norm": 0.2281133234500885,
|
| 5201 |
+
"learning_rate": 1.712085567246878e-05,
|
| 5202 |
+
"loss": 0.0213,
|
| 5203 |
+
"step": 7420
|
| 5204 |
+
},
|
| 5205 |
+
{
|
| 5206 |
+
"epoch": 4.7084917617237005,
|
| 5207 |
+
"grad_norm": 0.12594124674797058,
|
| 5208 |
+
"learning_rate": 1.699646620627168e-05,
|
| 5209 |
+
"loss": 0.0082,
|
| 5210 |
+
"step": 7430
|
| 5211 |
+
},
|
| 5212 |
+
{
|
| 5213 |
+
"epoch": 4.714828897338403,
|
| 5214 |
+
"grad_norm": 0.18425187468528748,
|
| 5215 |
+
"learning_rate": 1.6872437661432517e-05,
|
| 5216 |
+
"loss": 0.0068,
|
| 5217 |
+
"step": 7440
|
| 5218 |
+
},
|
| 5219 |
+
{
|
| 5220 |
+
"epoch": 4.721166032953105,
|
| 5221 |
+
"grad_norm": 0.3101319670677185,
|
| 5222 |
+
"learning_rate": 1.6748771394307585e-05,
|
| 5223 |
+
"loss": 0.0076,
|
| 5224 |
+
"step": 7450
|
| 5225 |
+
},
|
| 5226 |
+
{
|
| 5227 |
+
"epoch": 4.727503168567807,
|
| 5228 |
+
"grad_norm": 0.1615813672542572,
|
| 5229 |
+
"learning_rate": 1.662546875729138e-05,
|
| 5230 |
+
"loss": 0.0078,
|
| 5231 |
+
"step": 7460
|
| 5232 |
+
},
|
| 5233 |
+
{
|
| 5234 |
+
"epoch": 4.733840304182509,
|
| 5235 |
+
"grad_norm": 0.22787398099899292,
|
| 5236 |
+
"learning_rate": 1.6502531098801753e-05,
|
| 5237 |
+
"loss": 0.0112,
|
| 5238 |
+
"step": 7470
|
| 5239 |
+
},
|
| 5240 |
+
{
|
| 5241 |
+
"epoch": 4.740177439797211,
|
| 5242 |
+
"grad_norm": 0.08283796161413193,
|
| 5243 |
+
"learning_rate": 1.637995976326527e-05,
|
| 5244 |
+
"loss": 0.0157,
|
| 5245 |
+
"step": 7480
|
| 5246 |
+
},
|
| 5247 |
+
{
|
| 5248 |
+
"epoch": 4.746514575411914,
|
| 5249 |
+
"grad_norm": 0.20619574189186096,
|
| 5250 |
+
"learning_rate": 1.62577560911024e-05,
|
| 5251 |
+
"loss": 0.0119,
|
| 5252 |
+
"step": 7490
|
| 5253 |
+
},
|
| 5254 |
+
{
|
| 5255 |
+
"epoch": 4.752851711026616,
|
| 5256 |
+
"grad_norm": 0.35674479603767395,
|
| 5257 |
+
"learning_rate": 1.6135921418712956e-05,
|
| 5258 |
+
"loss": 0.0087,
|
| 5259 |
+
"step": 7500
|
| 5260 |
+
},
|
| 5261 |
+
{
|
| 5262 |
+
"epoch": 4.759188846641318,
|
| 5263 |
+
"grad_norm": 0.2025870382785797,
|
| 5264 |
+
"learning_rate": 1.6014457078461353e-05,
|
| 5265 |
+
"loss": 0.0062,
|
| 5266 |
+
"step": 7510
|
| 5267 |
+
},
|
| 5268 |
+
{
|
| 5269 |
+
"epoch": 4.765525982256021,
|
| 5270 |
+
"grad_norm": 0.13968420028686523,
|
| 5271 |
+
"learning_rate": 1.5893364398662176e-05,
|
| 5272 |
+
"loss": 0.0132,
|
| 5273 |
+
"step": 7520
|
| 5274 |
+
},
|
| 5275 |
+
{
|
| 5276 |
+
"epoch": 4.771863117870723,
|
| 5277 |
+
"grad_norm": 0.18326647579669952,
|
| 5278 |
+
"learning_rate": 1.5772644703565565e-05,
|
| 5279 |
+
"loss": 0.0096,
|
| 5280 |
+
"step": 7530
|
| 5281 |
+
},
|
| 5282 |
+
{
|
| 5283 |
+
"epoch": 4.778200253485425,
|
| 5284 |
+
"grad_norm": 0.2547348141670227,
|
| 5285 |
+
"learning_rate": 1.5652299313342773e-05,
|
| 5286 |
+
"loss": 0.0084,
|
| 5287 |
+
"step": 7540
|
| 5288 |
+
},
|
| 5289 |
+
{
|
| 5290 |
+
"epoch": 4.784537389100127,
|
| 5291 |
+
"grad_norm": 0.12518715858459473,
|
| 5292 |
+
"learning_rate": 1.553232954407171e-05,
|
| 5293 |
+
"loss": 0.01,
|
| 5294 |
+
"step": 7550
|
| 5295 |
+
},
|
| 5296 |
+
{
|
| 5297 |
+
"epoch": 4.7908745247148286,
|
| 5298 |
+
"grad_norm": 0.12570436298847198,
|
| 5299 |
+
"learning_rate": 1.5412736707722537e-05,
|
| 5300 |
+
"loss": 0.0094,
|
| 5301 |
+
"step": 7560
|
| 5302 |
+
},
|
| 5303 |
+
{
|
| 5304 |
+
"epoch": 4.797211660329531,
|
| 5305 |
+
"grad_norm": 0.15311287343502045,
|
| 5306 |
+
"learning_rate": 1.5293522112143373e-05,
|
| 5307 |
+
"loss": 0.0057,
|
| 5308 |
+
"step": 7570
|
| 5309 |
+
},
|
| 5310 |
+
{
|
| 5311 |
+
"epoch": 4.803548795944233,
|
| 5312 |
+
"grad_norm": 0.08851443231105804,
|
| 5313 |
+
"learning_rate": 1.517468706104589e-05,
|
| 5314 |
+
"loss": 0.0083,
|
| 5315 |
+
"step": 7580
|
| 5316 |
+
},
|
| 5317 |
+
{
|
| 5318 |
+
"epoch": 4.809885931558935,
|
| 5319 |
+
"grad_norm": 0.14807093143463135,
|
| 5320 |
+
"learning_rate": 1.5056232853991209e-05,
|
| 5321 |
+
"loss": 0.0097,
|
| 5322 |
+
"step": 7590
|
| 5323 |
+
},
|
| 5324 |
+
{
|
| 5325 |
+
"epoch": 4.816223067173637,
|
| 5326 |
+
"grad_norm": 0.20004534721374512,
|
| 5327 |
+
"learning_rate": 1.4938160786375572e-05,
|
| 5328 |
+
"loss": 0.0101,
|
| 5329 |
+
"step": 7600
|
| 5330 |
+
},
|
| 5331 |
+
{
|
| 5332 |
+
"epoch": 4.82256020278834,
|
| 5333 |
+
"grad_norm": 0.2074967622756958,
|
| 5334 |
+
"learning_rate": 1.4820472149416154e-05,
|
| 5335 |
+
"loss": 0.0073,
|
| 5336 |
+
"step": 7610
|
| 5337 |
+
},
|
| 5338 |
+
{
|
| 5339 |
+
"epoch": 4.828897338403042,
|
| 5340 |
+
"grad_norm": 0.1131831631064415,
|
| 5341 |
+
"learning_rate": 1.470316823013707e-05,
|
| 5342 |
+
"loss": 0.0067,
|
| 5343 |
+
"step": 7620
|
| 5344 |
+
},
|
| 5345 |
+
{
|
| 5346 |
+
"epoch": 4.835234474017744,
|
| 5347 |
+
"grad_norm": 0.21209746599197388,
|
| 5348 |
+
"learning_rate": 1.4586250311355132e-05,
|
| 5349 |
+
"loss": 0.014,
|
| 5350 |
+
"step": 7630
|
| 5351 |
+
},
|
| 5352 |
+
{
|
| 5353 |
+
"epoch": 4.841571609632446,
|
| 5354 |
+
"grad_norm": 0.14775687456130981,
|
| 5355 |
+
"learning_rate": 1.4469719671666043e-05,
|
| 5356 |
+
"loss": 0.0081,
|
| 5357 |
+
"step": 7640
|
| 5358 |
+
},
|
| 5359 |
+
{
|
| 5360 |
+
"epoch": 4.847908745247148,
|
| 5361 |
+
"grad_norm": 0.20488443970680237,
|
| 5362 |
+
"learning_rate": 1.435357758543015e-05,
|
| 5363 |
+
"loss": 0.0073,
|
| 5364 |
+
"step": 7650
|
| 5365 |
+
},
|
| 5366 |
+
{
|
| 5367 |
+
"epoch": 4.854245880861851,
|
| 5368 |
+
"grad_norm": 0.06777448952198029,
|
| 5369 |
+
"learning_rate": 1.4237825322758736e-05,
|
| 5370 |
+
"loss": 0.0131,
|
| 5371 |
+
"step": 7660
|
| 5372 |
+
},
|
| 5373 |
+
{
|
| 5374 |
+
"epoch": 4.860583016476553,
|
| 5375 |
+
"grad_norm": 0.13396242260932922,
|
| 5376 |
+
"learning_rate": 1.412246414949997e-05,
|
| 5377 |
+
"loss": 0.005,
|
| 5378 |
+
"step": 7670
|
| 5379 |
+
},
|
| 5380 |
+
{
|
| 5381 |
+
"epoch": 4.866920152091255,
|
| 5382 |
+
"grad_norm": 0.21779701113700867,
|
| 5383 |
+
"learning_rate": 1.4007495327225162e-05,
|
| 5384 |
+
"loss": 0.0076,
|
| 5385 |
+
"step": 7680
|
| 5386 |
+
},
|
| 5387 |
+
{
|
| 5388 |
+
"epoch": 4.873257287705957,
|
| 5389 |
+
"grad_norm": 0.2239043265581131,
|
| 5390 |
+
"learning_rate": 1.389292011321498e-05,
|
| 5391 |
+
"loss": 0.0091,
|
| 5392 |
+
"step": 7690
|
| 5393 |
+
},
|
| 5394 |
+
{
|
| 5395 |
+
"epoch": 4.879594423320659,
|
| 5396 |
+
"grad_norm": 0.06960437446832657,
|
| 5397 |
+
"learning_rate": 1.3778739760445552e-05,
|
| 5398 |
+
"loss": 0.007,
|
| 5399 |
+
"step": 7700
|
| 5400 |
+
},
|
| 5401 |
+
{
|
| 5402 |
+
"epoch": 4.885931558935361,
|
| 5403 |
+
"grad_norm": 0.14277929067611694,
|
| 5404 |
+
"learning_rate": 1.3664955517574968e-05,
|
| 5405 |
+
"loss": 0.0078,
|
| 5406 |
+
"step": 7710
|
| 5407 |
+
},
|
| 5408 |
+
{
|
| 5409 |
+
"epoch": 4.892268694550063,
|
| 5410 |
+
"grad_norm": 0.17942270636558533,
|
| 5411 |
+
"learning_rate": 1.3551568628929434e-05,
|
| 5412 |
+
"loss": 0.0093,
|
| 5413 |
+
"step": 7720
|
| 5414 |
+
},
|
| 5415 |
+
{
|
| 5416 |
+
"epoch": 4.898605830164765,
|
| 5417 |
+
"grad_norm": 0.23902519047260284,
|
| 5418 |
+
"learning_rate": 1.343858033448982e-05,
|
| 5419 |
+
"loss": 0.0087,
|
| 5420 |
+
"step": 7730
|
| 5421 |
+
},
|
| 5422 |
+
{
|
| 5423 |
+
"epoch": 4.904942965779467,
|
| 5424 |
+
"grad_norm": 0.33525460958480835,
|
| 5425 |
+
"learning_rate": 1.3325991869878013e-05,
|
| 5426 |
+
"loss": 0.0064,
|
| 5427 |
+
"step": 7740
|
| 5428 |
+
},
|
| 5429 |
+
{
|
| 5430 |
+
"epoch": 4.91128010139417,
|
| 5431 |
+
"grad_norm": 0.11698520183563232,
|
| 5432 |
+
"learning_rate": 1.3213804466343421e-05,
|
| 5433 |
+
"loss": 0.0088,
|
| 5434 |
+
"step": 7750
|
| 5435 |
+
},
|
| 5436 |
+
{
|
| 5437 |
+
"epoch": 4.917617237008872,
|
| 5438 |
+
"grad_norm": 0.1842937171459198,
|
| 5439 |
+
"learning_rate": 1.3102019350749528e-05,
|
| 5440 |
+
"loss": 0.0092,
|
| 5441 |
+
"step": 7760
|
| 5442 |
+
},
|
| 5443 |
+
{
|
| 5444 |
+
"epoch": 4.923954372623574,
|
| 5445 |
+
"grad_norm": 0.09481093287467957,
|
| 5446 |
+
"learning_rate": 1.299063774556042e-05,
|
| 5447 |
+
"loss": 0.0059,
|
| 5448 |
+
"step": 7770
|
| 5449 |
+
},
|
| 5450 |
+
{
|
| 5451 |
+
"epoch": 4.930291508238277,
|
| 5452 |
+
"grad_norm": 0.17098549008369446,
|
| 5453 |
+
"learning_rate": 1.2879660868827508e-05,
|
| 5454 |
+
"loss": 0.0084,
|
| 5455 |
+
"step": 7780
|
| 5456 |
+
},
|
| 5457 |
+
{
|
| 5458 |
+
"epoch": 4.936628643852979,
|
| 5459 |
+
"grad_norm": 0.3938882648944855,
|
| 5460 |
+
"learning_rate": 1.2769089934176126e-05,
|
| 5461 |
+
"loss": 0.0113,
|
| 5462 |
+
"step": 7790
|
| 5463 |
+
},
|
| 5464 |
+
{
|
| 5465 |
+
"epoch": 4.942965779467681,
|
| 5466 |
+
"grad_norm": 0.21974030137062073,
|
| 5467 |
+
"learning_rate": 1.2658926150792322e-05,
|
| 5468 |
+
"loss": 0.0074,
|
| 5469 |
+
"step": 7800
|
| 5470 |
+
},
|
| 5471 |
+
{
|
| 5472 |
+
"epoch": 4.949302915082383,
|
| 5473 |
+
"grad_norm": 0.1278766393661499,
|
| 5474 |
+
"learning_rate": 1.2549170723409549e-05,
|
| 5475 |
+
"loss": 0.01,
|
| 5476 |
+
"step": 7810
|
| 5477 |
+
},
|
| 5478 |
+
{
|
| 5479 |
+
"epoch": 4.955640050697085,
|
| 5480 |
+
"grad_norm": 0.18058639764785767,
|
| 5481 |
+
"learning_rate": 1.243982485229559e-05,
|
| 5482 |
+
"loss": 0.0075,
|
| 5483 |
+
"step": 7820
|
| 5484 |
+
},
|
| 5485 |
+
{
|
| 5486 |
+
"epoch": 4.961977186311787,
|
| 5487 |
+
"grad_norm": 0.20735391974449158,
|
| 5488 |
+
"learning_rate": 1.233088973323937e-05,
|
| 5489 |
+
"loss": 0.0074,
|
| 5490 |
+
"step": 7830
|
| 5491 |
+
},
|
| 5492 |
+
{
|
| 5493 |
+
"epoch": 4.968314321926489,
|
| 5494 |
+
"grad_norm": 0.12043243646621704,
|
| 5495 |
+
"learning_rate": 1.2222366557537911e-05,
|
| 5496 |
+
"loss": 0.0093,
|
| 5497 |
+
"step": 7840
|
| 5498 |
+
},
|
| 5499 |
+
{
|
| 5500 |
+
"epoch": 4.974651457541191,
|
| 5501 |
+
"grad_norm": 0.28951793909072876,
|
| 5502 |
+
"learning_rate": 1.2114256511983274e-05,
|
| 5503 |
+
"loss": 0.007,
|
| 5504 |
+
"step": 7850
|
| 5505 |
+
},
|
| 5506 |
+
{
|
| 5507 |
+
"epoch": 4.980988593155893,
|
| 5508 |
+
"grad_norm": 0.1455940455198288,
|
| 5509 |
+
"learning_rate": 1.2006560778849578e-05,
|
| 5510 |
+
"loss": 0.0067,
|
| 5511 |
+
"step": 7860
|
| 5512 |
+
},
|
| 5513 |
+
{
|
| 5514 |
+
"epoch": 4.987325728770596,
|
| 5515 |
+
"grad_norm": 0.1419738233089447,
|
| 5516 |
+
"learning_rate": 1.1899280535880119e-05,
|
| 5517 |
+
"loss": 0.0115,
|
| 5518 |
+
"step": 7870
|
| 5519 |
+
},
|
| 5520 |
+
{
|
| 5521 |
+
"epoch": 4.993662864385298,
|
| 5522 |
+
"grad_norm": 0.3839736878871918,
|
| 5523 |
+
"learning_rate": 1.1792416956274444e-05,
|
| 5524 |
+
"loss": 0.0103,
|
| 5525 |
+
"step": 7880
|
| 5526 |
+
},
|
| 5527 |
+
{
|
| 5528 |
+
"epoch": 5.0,
|
| 5529 |
+
"grad_norm": 0.12225103378295898,
|
| 5530 |
+
"learning_rate": 1.1685971208675539e-05,
|
| 5531 |
+
"loss": 0.0083,
|
| 5532 |
+
"step": 7890
|
| 5533 |
+
},
|
| 5534 |
+
{
|
| 5535 |
+
"epoch": 5.006337135614702,
|
| 5536 |
+
"grad_norm": 0.22479389607906342,
|
| 5537 |
+
"learning_rate": 1.157994445715706e-05,
|
| 5538 |
+
"loss": 0.0082,
|
| 5539 |
+
"step": 7900
|
| 5540 |
+
},
|
| 5541 |
+
{
|
| 5542 |
+
"epoch": 5.012674271229404,
|
| 5543 |
+
"grad_norm": 0.2502928674221039,
|
| 5544 |
+
"learning_rate": 1.1474337861210543e-05,
|
| 5545 |
+
"loss": 0.0073,
|
| 5546 |
+
"step": 7910
|
| 5547 |
+
},
|
| 5548 |
+
{
|
| 5549 |
+
"epoch": 5.019011406844107,
|
| 5550 |
+
"grad_norm": 0.11566631495952606,
|
| 5551 |
+
"learning_rate": 1.1369152575732822e-05,
|
| 5552 |
+
"loss": 0.0058,
|
| 5553 |
+
"step": 7920
|
| 5554 |
+
},
|
| 5555 |
+
{
|
| 5556 |
+
"epoch": 5.025348542458809,
|
| 5557 |
+
"grad_norm": 0.11840217560529709,
|
| 5558 |
+
"learning_rate": 1.1264389751013326e-05,
|
| 5559 |
+
"loss": 0.007,
|
| 5560 |
+
"step": 7930
|
| 5561 |
+
},
|
| 5562 |
+
{
|
| 5563 |
+
"epoch": 5.031685678073511,
|
| 5564 |
+
"grad_norm": 0.30334770679473877,
|
| 5565 |
+
"learning_rate": 1.1160050532721528e-05,
|
| 5566 |
+
"loss": 0.0116,
|
| 5567 |
+
"step": 7940
|
| 5568 |
+
},
|
| 5569 |
+
{
|
| 5570 |
+
"epoch": 5.038022813688213,
|
| 5571 |
+
"grad_norm": 0.17277342081069946,
|
| 5572 |
+
"learning_rate": 1.1056136061894384e-05,
|
| 5573 |
+
"loss": 0.0069,
|
| 5574 |
+
"step": 7950
|
| 5575 |
+
},
|
| 5576 |
+
{
|
| 5577 |
+
"epoch": 5.044359949302915,
|
| 5578 |
+
"grad_norm": 0.09852743148803711,
|
| 5579 |
+
"learning_rate": 1.095264747492391e-05,
|
| 5580 |
+
"loss": 0.0056,
|
| 5581 |
+
"step": 7960
|
| 5582 |
+
},
|
| 5583 |
+
{
|
| 5584 |
+
"epoch": 5.050697084917617,
|
| 5585 |
+
"grad_norm": 0.10726805031299591,
|
| 5586 |
+
"learning_rate": 1.0849585903544706e-05,
|
| 5587 |
+
"loss": 0.0148,
|
| 5588 |
+
"step": 7970
|
| 5589 |
+
},
|
| 5590 |
+
{
|
| 5591 |
+
"epoch": 5.057034220532319,
|
| 5592 |
+
"grad_norm": 0.08306515961885452,
|
| 5593 |
+
"learning_rate": 1.0746952474821614e-05,
|
| 5594 |
+
"loss": 0.005,
|
| 5595 |
+
"step": 7980
|
| 5596 |
+
},
|
| 5597 |
+
{
|
| 5598 |
+
"epoch": 5.063371356147021,
|
| 5599 |
+
"grad_norm": 0.13456346094608307,
|
| 5600 |
+
"learning_rate": 1.0644748311137376e-05,
|
| 5601 |
+
"loss": 0.0071,
|
| 5602 |
+
"step": 7990
|
| 5603 |
+
},
|
| 5604 |
+
{
|
| 5605 |
+
"epoch": 5.069708491761724,
|
| 5606 |
+
"grad_norm": 0.2890625,
|
| 5607 |
+
"learning_rate": 1.0542974530180327e-05,
|
| 5608 |
+
"loss": 0.0063,
|
| 5609 |
+
"step": 8000
|
| 5610 |
+
},
|
| 5611 |
+
{
|
| 5612 |
+
"epoch": 5.076045627376426,
|
| 5613 |
+
"grad_norm": 0.15506230294704437,
|
| 5614 |
+
"learning_rate": 1.0441632244932237e-05,
|
| 5615 |
+
"loss": 0.0077,
|
| 5616 |
+
"step": 8010
|
| 5617 |
+
},
|
| 5618 |
+
{
|
| 5619 |
+
"epoch": 5.082382762991128,
|
| 5620 |
+
"grad_norm": 0.17413881421089172,
|
| 5621 |
+
"learning_rate": 1.0340722563656107e-05,
|
| 5622 |
+
"loss": 0.0066,
|
| 5623 |
+
"step": 8020
|
| 5624 |
+
},
|
| 5625 |
+
{
|
| 5626 |
+
"epoch": 5.08871989860583,
|
| 5627 |
+
"grad_norm": 0.15269486606121063,
|
| 5628 |
+
"learning_rate": 1.0240246589884044e-05,
|
| 5629 |
+
"loss": 0.0056,
|
| 5630 |
+
"step": 8030
|
| 5631 |
+
},
|
| 5632 |
+
{
|
| 5633 |
+
"epoch": 5.095057034220532,
|
| 5634 |
+
"grad_norm": 0.17267774045467377,
|
| 5635 |
+
"learning_rate": 1.0140205422405214e-05,
|
| 5636 |
+
"loss": 0.0062,
|
| 5637 |
+
"step": 8040
|
| 5638 |
+
},
|
| 5639 |
+
{
|
| 5640 |
+
"epoch": 5.101394169835235,
|
| 5641 |
+
"grad_norm": 0.20491677522659302,
|
| 5642 |
+
"learning_rate": 1.0040600155253765e-05,
|
| 5643 |
+
"loss": 0.0082,
|
| 5644 |
+
"step": 8050
|
| 5645 |
+
},
|
| 5646 |
+
{
|
| 5647 |
+
"epoch": 5.107731305449937,
|
| 5648 |
+
"grad_norm": 0.15973028540611267,
|
| 5649 |
+
"learning_rate": 9.941431877696955e-06,
|
| 5650 |
+
"loss": 0.0095,
|
| 5651 |
+
"step": 8060
|
| 5652 |
+
},
|
| 5653 |
+
{
|
| 5654 |
+
"epoch": 5.114068441064639,
|
| 5655 |
+
"grad_norm": 0.13956964015960693,
|
| 5656 |
+
"learning_rate": 9.842701674223187e-06,
|
| 5657 |
+
"loss": 0.008,
|
| 5658 |
+
"step": 8070
|
| 5659 |
+
},
|
| 5660 |
+
{
|
| 5661 |
+
"epoch": 5.120405576679341,
|
| 5662 |
+
"grad_norm": 0.29668980836868286,
|
| 5663 |
+
"learning_rate": 9.744410624530148e-06,
|
| 5664 |
+
"loss": 0.009,
|
| 5665 |
+
"step": 8080
|
| 5666 |
+
},
|
| 5667 |
+
{
|
| 5668 |
+
"epoch": 5.126742712294043,
|
| 5669 |
+
"grad_norm": 0.1549568921327591,
|
| 5670 |
+
"learning_rate": 9.646559803512994e-06,
|
| 5671 |
+
"loss": 0.0073,
|
| 5672 |
+
"step": 8090
|
| 5673 |
+
},
|
| 5674 |
+
{
|
| 5675 |
+
"epoch": 5.133079847908745,
|
| 5676 |
+
"grad_norm": 0.15987129509449005,
|
| 5677 |
+
"learning_rate": 9.549150281252633e-06,
|
| 5678 |
+
"loss": 0.0078,
|
| 5679 |
+
"step": 8100
|
| 5680 |
+
},
|
| 5681 |
+
{
|
| 5682 |
+
"epoch": 5.139416983523447,
|
| 5683 |
+
"grad_norm": 0.0577310174703598,
|
| 5684 |
+
"learning_rate": 9.452183123004e-06,
|
| 5685 |
+
"loss": 0.0098,
|
| 5686 |
+
"step": 8110
|
| 5687 |
+
},
|
| 5688 |
+
{
|
| 5689 |
+
"epoch": 5.145754119138149,
|
| 5690 |
+
"grad_norm": 0.1426042914390564,
|
| 5691 |
+
"learning_rate": 9.355659389184396e-06,
|
| 5692 |
+
"loss": 0.0078,
|
| 5693 |
+
"step": 8120
|
| 5694 |
+
},
|
| 5695 |
+
{
|
| 5696 |
+
"epoch": 5.152091254752852,
|
| 5697 |
+
"grad_norm": 0.13074810802936554,
|
| 5698 |
+
"learning_rate": 9.259580135361929e-06,
|
| 5699 |
+
"loss": 0.0081,
|
| 5700 |
+
"step": 8130
|
| 5701 |
+
},
|
| 5702 |
+
{
|
| 5703 |
+
"epoch": 5.158428390367554,
|
| 5704 |
+
"grad_norm": 0.19135090708732605,
|
| 5705 |
+
"learning_rate": 9.163946412243896e-06,
|
| 5706 |
+
"loss": 0.0098,
|
| 5707 |
+
"step": 8140
|
| 5708 |
+
},
|
| 5709 |
+
{
|
| 5710 |
+
"epoch": 5.164765525982256,
|
| 5711 |
+
"grad_norm": 0.15256141126155853,
|
| 5712 |
+
"learning_rate": 9.068759265665384e-06,
|
| 5713 |
+
"loss": 0.0053,
|
| 5714 |
+
"step": 8150
|
| 5715 |
+
},
|
| 5716 |
+
{
|
| 5717 |
+
"epoch": 5.171102661596958,
|
| 5718 |
+
"grad_norm": 0.15117891132831573,
|
| 5719 |
+
"learning_rate": 8.974019736577777e-06,
|
| 5720 |
+
"loss": 0.0079,
|
| 5721 |
+
"step": 8160
|
| 5722 |
+
},
|
| 5723 |
+
{
|
| 5724 |
+
"epoch": 5.17743979721166,
|
| 5725 |
+
"grad_norm": 0.09071781486272812,
|
| 5726 |
+
"learning_rate": 8.879728861037384e-06,
|
| 5727 |
+
"loss": 0.0099,
|
| 5728 |
+
"step": 8170
|
| 5729 |
+
},
|
| 5730 |
+
{
|
| 5731 |
+
"epoch": 5.183776932826363,
|
| 5732 |
+
"grad_norm": 0.24567152559757233,
|
| 5733 |
+
"learning_rate": 8.785887670194138e-06,
|
| 5734 |
+
"loss": 0.0055,
|
| 5735 |
+
"step": 8180
|
| 5736 |
+
},
|
| 5737 |
+
{
|
| 5738 |
+
"epoch": 5.190114068441065,
|
| 5739 |
+
"grad_norm": 0.19762946665287018,
|
| 5740 |
+
"learning_rate": 8.692497190280224e-06,
|
| 5741 |
+
"loss": 0.0092,
|
| 5742 |
+
"step": 8190
|
| 5743 |
+
},
|
| 5744 |
+
{
|
| 5745 |
+
"epoch": 5.196451204055767,
|
| 5746 |
+
"grad_norm": 0.24934445321559906,
|
| 5747 |
+
"learning_rate": 8.599558442598998e-06,
|
| 5748 |
+
"loss": 0.0094,
|
| 5749 |
+
"step": 8200
|
| 5750 |
+
},
|
| 5751 |
+
{
|
| 5752 |
+
"epoch": 5.202788339670469,
|
| 5753 |
+
"grad_norm": 0.1780957281589508,
|
| 5754 |
+
"learning_rate": 8.507072443513702e-06,
|
| 5755 |
+
"loss": 0.0083,
|
| 5756 |
+
"step": 8210
|
| 5757 |
+
},
|
| 5758 |
+
{
|
| 5759 |
+
"epoch": 5.2091254752851714,
|
| 5760 |
+
"grad_norm": 0.23371122777462006,
|
| 5761 |
+
"learning_rate": 8.415040204436426e-06,
|
| 5762 |
+
"loss": 0.0082,
|
| 5763 |
+
"step": 8220
|
| 5764 |
+
},
|
| 5765 |
+
{
|
| 5766 |
+
"epoch": 5.215462610899873,
|
| 5767 |
+
"grad_norm": 0.06640614569187164,
|
| 5768 |
+
"learning_rate": 8.323462731816961e-06,
|
| 5769 |
+
"loss": 0.0052,
|
| 5770 |
+
"step": 8230
|
| 5771 |
+
},
|
| 5772 |
+
{
|
| 5773 |
+
"epoch": 5.221799746514575,
|
| 5774 |
+
"grad_norm": 0.118456169962883,
|
| 5775 |
+
"learning_rate": 8.232341027131885e-06,
|
| 5776 |
+
"loss": 0.0081,
|
| 5777 |
+
"step": 8240
|
| 5778 |
+
},
|
| 5779 |
+
{
|
| 5780 |
+
"epoch": 5.228136882129277,
|
| 5781 |
+
"grad_norm": 0.1477072387933731,
|
| 5782 |
+
"learning_rate": 8.141676086873572e-06,
|
| 5783 |
+
"loss": 0.0081,
|
| 5784 |
+
"step": 8250
|
| 5785 |
+
},
|
| 5786 |
+
{
|
| 5787 |
+
"epoch": 5.23447401774398,
|
| 5788 |
+
"grad_norm": 0.1423831284046173,
|
| 5789 |
+
"learning_rate": 8.051468902539272e-06,
|
| 5790 |
+
"loss": 0.0058,
|
| 5791 |
+
"step": 8260
|
| 5792 |
+
},
|
| 5793 |
+
{
|
| 5794 |
+
"epoch": 5.240811153358682,
|
| 5795 |
+
"grad_norm": 0.14859850704669952,
|
| 5796 |
+
"learning_rate": 7.96172046062032e-06,
|
| 5797 |
+
"loss": 0.0094,
|
| 5798 |
+
"step": 8270
|
| 5799 |
+
},
|
| 5800 |
+
{
|
| 5801 |
+
"epoch": 5.247148288973384,
|
| 5802 |
+
"grad_norm": 0.22598117589950562,
|
| 5803 |
+
"learning_rate": 7.872431742591268e-06,
|
| 5804 |
+
"loss": 0.0079,
|
| 5805 |
+
"step": 8280
|
| 5806 |
+
},
|
| 5807 |
+
{
|
| 5808 |
+
"epoch": 5.253485424588086,
|
| 5809 |
+
"grad_norm": 0.20482905209064484,
|
| 5810 |
+
"learning_rate": 7.783603724899257e-06,
|
| 5811 |
+
"loss": 0.0079,
|
| 5812 |
+
"step": 8290
|
| 5813 |
+
},
|
| 5814 |
+
{
|
| 5815 |
+
"epoch": 5.259822560202788,
|
| 5816 |
+
"grad_norm": 0.21788014471530914,
|
| 5817 |
+
"learning_rate": 7.695237378953223e-06,
|
| 5818 |
+
"loss": 0.0055,
|
| 5819 |
+
"step": 8300
|
| 5820 |
+
},
|
| 5821 |
+
{
|
| 5822 |
+
"epoch": 5.266159695817491,
|
| 5823 |
+
"grad_norm": 0.09685339778661728,
|
| 5824 |
+
"learning_rate": 7.607333671113409e-06,
|
| 5825 |
+
"loss": 0.0038,
|
| 5826 |
+
"step": 8310
|
| 5827 |
+
},
|
| 5828 |
+
{
|
| 5829 |
+
"epoch": 5.272496831432193,
|
| 5830 |
+
"grad_norm": 0.175571009516716,
|
| 5831 |
+
"learning_rate": 7.519893562680663e-06,
|
| 5832 |
+
"loss": 0.0062,
|
| 5833 |
+
"step": 8320
|
| 5834 |
+
},
|
| 5835 |
+
{
|
| 5836 |
+
"epoch": 5.278833967046895,
|
| 5837 |
+
"grad_norm": 0.20628724992275238,
|
| 5838 |
+
"learning_rate": 7.432918009885997e-06,
|
| 5839 |
+
"loss": 0.0088,
|
| 5840 |
+
"step": 8330
|
| 5841 |
+
},
|
| 5842 |
+
{
|
| 5843 |
+
"epoch": 5.285171102661597,
|
| 5844 |
+
"grad_norm": 0.0737687200307846,
|
| 5845 |
+
"learning_rate": 7.3464079638801365e-06,
|
| 5846 |
+
"loss": 0.0088,
|
| 5847 |
+
"step": 8340
|
| 5848 |
+
},
|
| 5849 |
+
{
|
| 5850 |
+
"epoch": 5.2915082382762995,
|
| 5851 |
+
"grad_norm": 0.10191112756729126,
|
| 5852 |
+
"learning_rate": 7.260364370723044e-06,
|
| 5853 |
+
"loss": 0.0081,
|
| 5854 |
+
"step": 8350
|
| 5855 |
+
},
|
| 5856 |
+
{
|
| 5857 |
+
"epoch": 5.297845373891001,
|
| 5858 |
+
"grad_norm": 0.35432806611061096,
|
| 5859 |
+
"learning_rate": 7.174788171373731e-06,
|
| 5860 |
+
"loss": 0.0096,
|
| 5861 |
+
"step": 8360
|
| 5862 |
+
},
|
| 5863 |
+
{
|
| 5864 |
+
"epoch": 5.304182509505703,
|
| 5865 |
+
"grad_norm": 0.11743071675300598,
|
| 5866 |
+
"learning_rate": 7.089680301679752e-06,
|
| 5867 |
+
"loss": 0.0132,
|
| 5868 |
+
"step": 8370
|
| 5869 |
+
},
|
| 5870 |
+
{
|
| 5871 |
+
"epoch": 5.310519645120405,
|
| 5872 |
+
"grad_norm": 0.15060439705848694,
|
| 5873 |
+
"learning_rate": 7.005041692367154e-06,
|
| 5874 |
+
"loss": 0.008,
|
| 5875 |
+
"step": 8380
|
| 5876 |
+
},
|
| 5877 |
+
{
|
| 5878 |
+
"epoch": 5.316856780735108,
|
| 5879 |
+
"grad_norm": 0.19431732594966888,
|
| 5880 |
+
"learning_rate": 6.92087326903022e-06,
|
| 5881 |
+
"loss": 0.0062,
|
| 5882 |
+
"step": 8390
|
| 5883 |
+
},
|
| 5884 |
+
{
|
| 5885 |
+
"epoch": 5.32319391634981,
|
| 5886 |
+
"grad_norm": 0.10982516407966614,
|
| 5887 |
+
"learning_rate": 6.837175952121306e-06,
|
| 5888 |
+
"loss": 0.0057,
|
| 5889 |
+
"step": 8400
|
| 5890 |
+
},
|
| 5891 |
+
{
|
| 5892 |
+
"epoch": 5.329531051964512,
|
| 5893 |
+
"grad_norm": 0.09768649935722351,
|
| 5894 |
+
"learning_rate": 6.753950656940905e-06,
|
| 5895 |
+
"loss": 0.0047,
|
| 5896 |
+
"step": 8410
|
| 5897 |
+
},
|
| 5898 |
+
{
|
| 5899 |
+
"epoch": 5.335868187579214,
|
| 5900 |
+
"grad_norm": 0.2184268981218338,
|
| 5901 |
+
"learning_rate": 6.671198293627479e-06,
|
| 5902 |
+
"loss": 0.0061,
|
| 5903 |
+
"step": 8420
|
| 5904 |
+
},
|
| 5905 |
+
{
|
| 5906 |
+
"epoch": 5.342205323193916,
|
| 5907 |
+
"grad_norm": 0.11367691308259964,
|
| 5908 |
+
"learning_rate": 6.588919767147639e-06,
|
| 5909 |
+
"loss": 0.0048,
|
| 5910 |
+
"step": 8430
|
| 5911 |
+
},
|
| 5912 |
+
{
|
| 5913 |
+
"epoch": 5.348542458808619,
|
| 5914 |
+
"grad_norm": 0.11176607757806778,
|
| 5915 |
+
"learning_rate": 6.5071159772861436e-06,
|
| 5916 |
+
"loss": 0.0092,
|
| 5917 |
+
"step": 8440
|
| 5918 |
+
},
|
| 5919 |
+
{
|
| 5920 |
+
"epoch": 5.354879594423321,
|
| 5921 |
+
"grad_norm": 0.12881191074848175,
|
| 5922 |
+
"learning_rate": 6.425787818636131e-06,
|
| 5923 |
+
"loss": 0.0056,
|
| 5924 |
+
"step": 8450
|
| 5925 |
+
},
|
| 5926 |
+
{
|
| 5927 |
+
"epoch": 5.361216730038023,
|
| 5928 |
+
"grad_norm": 0.19846558570861816,
|
| 5929 |
+
"learning_rate": 6.344936180589351e-06,
|
| 5930 |
+
"loss": 0.0059,
|
| 5931 |
+
"step": 8460
|
| 5932 |
+
},
|
| 5933 |
+
{
|
| 5934 |
+
"epoch": 5.367553865652725,
|
| 5935 |
+
"grad_norm": 0.2730921506881714,
|
| 5936 |
+
"learning_rate": 6.264561947326331e-06,
|
| 5937 |
+
"loss": 0.0074,
|
| 5938 |
+
"step": 8470
|
| 5939 |
+
},
|
| 5940 |
+
{
|
| 5941 |
+
"epoch": 5.3738910012674275,
|
| 5942 |
+
"grad_norm": 0.1230594590306282,
|
| 5943 |
+
"learning_rate": 6.184665997806832e-06,
|
| 5944 |
+
"loss": 0.0076,
|
| 5945 |
+
"step": 8480
|
| 5946 |
+
},
|
| 5947 |
+
{
|
| 5948 |
+
"epoch": 5.380228136882129,
|
| 5949 |
+
"grad_norm": 0.12817886471748352,
|
| 5950 |
+
"learning_rate": 6.1052492057601275e-06,
|
| 5951 |
+
"loss": 0.0069,
|
| 5952 |
+
"step": 8490
|
| 5953 |
+
},
|
| 5954 |
+
{
|
| 5955 |
+
"epoch": 5.386565272496831,
|
| 5956 |
+
"grad_norm": 0.08666050434112549,
|
| 5957 |
+
"learning_rate": 6.026312439675552e-06,
|
| 5958 |
+
"loss": 0.0059,
|
| 5959 |
+
"step": 8500
|
| 5960 |
+
},
|
| 5961 |
+
{
|
| 5962 |
+
"epoch": 5.392902408111533,
|
| 5963 |
+
"grad_norm": 0.2247517853975296,
|
| 5964 |
+
"learning_rate": 5.947856562792925e-06,
|
| 5965 |
+
"loss": 0.0147,
|
| 5966 |
+
"step": 8510
|
| 5967 |
+
},
|
| 5968 |
+
{
|
| 5969 |
+
"epoch": 5.399239543726236,
|
| 5970 |
+
"grad_norm": 0.08660010248422623,
|
| 5971 |
+
"learning_rate": 5.869882433093155e-06,
|
| 5972 |
+
"loss": 0.0073,
|
| 5973 |
+
"step": 8520
|
| 5974 |
+
},
|
| 5975 |
+
{
|
| 5976 |
+
"epoch": 5.405576679340938,
|
| 5977 |
+
"grad_norm": 0.0791560560464859,
|
| 5978 |
+
"learning_rate": 5.79239090328883e-06,
|
| 5979 |
+
"loss": 0.0061,
|
| 5980 |
+
"step": 8530
|
| 5981 |
+
},
|
| 5982 |
+
{
|
| 5983 |
+
"epoch": 5.41191381495564,
|
| 5984 |
+
"grad_norm": 0.07316572964191437,
|
| 5985 |
+
"learning_rate": 5.715382820814885e-06,
|
| 5986 |
+
"loss": 0.0084,
|
| 5987 |
+
"step": 8540
|
| 5988 |
+
},
|
| 5989 |
+
{
|
| 5990 |
+
"epoch": 5.418250950570342,
|
| 5991 |
+
"grad_norm": 0.12857115268707275,
|
| 5992 |
+
"learning_rate": 5.6388590278194096e-06,
|
| 5993 |
+
"loss": 0.0068,
|
| 5994 |
+
"step": 8550
|
| 5995 |
+
},
|
| 5996 |
+
{
|
| 5997 |
+
"epoch": 5.424588086185044,
|
| 5998 |
+
"grad_norm": 0.21100053191184998,
|
| 5999 |
+
"learning_rate": 5.562820361154314e-06,
|
| 6000 |
+
"loss": 0.0075,
|
| 6001 |
+
"step": 8560
|
| 6002 |
+
},
|
| 6003 |
+
{
|
| 6004 |
+
"epoch": 5.430925221799747,
|
| 6005 |
+
"grad_norm": 0.08549866825342178,
|
| 6006 |
+
"learning_rate": 5.48726765236629e-06,
|
| 6007 |
+
"loss": 0.006,
|
| 6008 |
+
"step": 8570
|
| 6009 |
+
},
|
| 6010 |
+
{
|
| 6011 |
+
"epoch": 5.437262357414449,
|
| 6012 |
+
"grad_norm": 0.15550892055034637,
|
| 6013 |
+
"learning_rate": 5.412201727687644e-06,
|
| 6014 |
+
"loss": 0.0063,
|
| 6015 |
+
"step": 8580
|
| 6016 |
+
},
|
| 6017 |
+
{
|
| 6018 |
+
"epoch": 5.443599493029151,
|
| 6019 |
+
"grad_norm": 0.18836095929145813,
|
| 6020 |
+
"learning_rate": 5.337623408027293e-06,
|
| 6021 |
+
"loss": 0.0087,
|
| 6022 |
+
"step": 8590
|
| 6023 |
+
},
|
| 6024 |
+
{
|
| 6025 |
+
"epoch": 5.449936628643853,
|
| 6026 |
+
"grad_norm": 0.08456358313560486,
|
| 6027 |
+
"learning_rate": 5.263533508961827e-06,
|
| 6028 |
+
"loss": 0.006,
|
| 6029 |
+
"step": 8600
|
| 6030 |
+
},
|
| 6031 |
+
{
|
| 6032 |
+
"epoch": 5.4562737642585555,
|
| 6033 |
+
"grad_norm": 0.054700568318367004,
|
| 6034 |
+
"learning_rate": 5.1899328407264855e-06,
|
| 6035 |
+
"loss": 0.0053,
|
| 6036 |
+
"step": 8610
|
| 6037 |
+
},
|
| 6038 |
+
{
|
| 6039 |
+
"epoch": 5.462610899873257,
|
| 6040 |
+
"grad_norm": 0.13627506792545319,
|
| 6041 |
+
"learning_rate": 5.116822208206396e-06,
|
| 6042 |
+
"loss": 0.006,
|
| 6043 |
+
"step": 8620
|
| 6044 |
+
},
|
| 6045 |
+
{
|
| 6046 |
+
"epoch": 5.468948035487959,
|
| 6047 |
+
"grad_norm": 0.1252082735300064,
|
| 6048 |
+
"learning_rate": 5.044202410927706e-06,
|
| 6049 |
+
"loss": 0.0125,
|
| 6050 |
+
"step": 8630
|
| 6051 |
+
},
|
| 6052 |
+
{
|
| 6053 |
+
"epoch": 5.475285171102661,
|
| 6054 |
+
"grad_norm": 0.2162148505449295,
|
| 6055 |
+
"learning_rate": 4.972074243048897e-06,
|
| 6056 |
+
"loss": 0.0065,
|
| 6057 |
+
"step": 8640
|
| 6058 |
+
},
|
| 6059 |
+
{
|
| 6060 |
+
"epoch": 5.481622306717364,
|
| 6061 |
+
"grad_norm": 0.12487296015024185,
|
| 6062 |
+
"learning_rate": 4.900438493352055e-06,
|
| 6063 |
+
"loss": 0.0063,
|
| 6064 |
+
"step": 8650
|
| 6065 |
+
},
|
| 6066 |
+
{
|
| 6067 |
+
"epoch": 5.487959442332066,
|
| 6068 |
+
"grad_norm": 0.14202018082141876,
|
| 6069 |
+
"learning_rate": 4.829295945234258e-06,
|
| 6070 |
+
"loss": 0.0063,
|
| 6071 |
+
"step": 8660
|
| 6072 |
+
},
|
| 6073 |
+
{
|
| 6074 |
+
"epoch": 5.494296577946768,
|
| 6075 |
+
"grad_norm": 0.18836723268032074,
|
| 6076 |
+
"learning_rate": 4.758647376699032e-06,
|
| 6077 |
+
"loss": 0.0069,
|
| 6078 |
+
"step": 8670
|
| 6079 |
+
},
|
| 6080 |
+
{
|
| 6081 |
+
"epoch": 5.50063371356147,
|
| 6082 |
+
"grad_norm": 0.2227470427751541,
|
| 6083 |
+
"learning_rate": 4.688493560347773e-06,
|
| 6084 |
+
"loss": 0.0111,
|
| 6085 |
+
"step": 8680
|
| 6086 |
+
},
|
| 6087 |
+
{
|
| 6088 |
+
"epoch": 5.506970849176172,
|
| 6089 |
+
"grad_norm": 0.1131884977221489,
|
| 6090 |
+
"learning_rate": 4.618835263371396e-06,
|
| 6091 |
+
"loss": 0.0054,
|
| 6092 |
+
"step": 8690
|
| 6093 |
+
},
|
| 6094 |
+
{
|
| 6095 |
+
"epoch": 5.513307984790875,
|
| 6096 |
+
"grad_norm": 0.08110499382019043,
|
| 6097 |
+
"learning_rate": 4.549673247541875e-06,
|
| 6098 |
+
"loss": 0.0039,
|
| 6099 |
+
"step": 8700
|
| 6100 |
+
},
|
| 6101 |
+
{
|
| 6102 |
+
"epoch": 5.519645120405577,
|
| 6103 |
+
"grad_norm": 0.2664946913719177,
|
| 6104 |
+
"learning_rate": 4.48100826920394e-06,
|
| 6105 |
+
"loss": 0.0055,
|
| 6106 |
+
"step": 8710
|
| 6107 |
+
},
|
| 6108 |
+
{
|
| 6109 |
+
"epoch": 5.525982256020279,
|
| 6110 |
+
"grad_norm": 0.09621760994195938,
|
| 6111 |
+
"learning_rate": 4.412841079266777e-06,
|
| 6112 |
+
"loss": 0.0066,
|
| 6113 |
+
"step": 8720
|
| 6114 |
+
},
|
| 6115 |
+
{
|
| 6116 |
+
"epoch": 5.532319391634981,
|
| 6117 |
+
"grad_norm": 0.07459976524114609,
|
| 6118 |
+
"learning_rate": 4.3451724231958644e-06,
|
| 6119 |
+
"loss": 0.0103,
|
| 6120 |
+
"step": 8730
|
| 6121 |
+
},
|
| 6122 |
+
{
|
| 6123 |
+
"epoch": 5.5386565272496835,
|
| 6124 |
+
"grad_norm": 0.252323180437088,
|
| 6125 |
+
"learning_rate": 4.27800304100478e-06,
|
| 6126 |
+
"loss": 0.0041,
|
| 6127 |
+
"step": 8740
|
| 6128 |
+
},
|
| 6129 |
+
{
|
| 6130 |
+
"epoch": 5.544993662864385,
|
| 6131 |
+
"grad_norm": 0.11829400062561035,
|
| 6132 |
+
"learning_rate": 4.2113336672471245e-06,
|
| 6133 |
+
"loss": 0.0042,
|
| 6134 |
+
"step": 8750
|
| 6135 |
+
},
|
| 6136 |
+
{
|
| 6137 |
+
"epoch": 5.551330798479087,
|
| 6138 |
+
"grad_norm": 0.07013269513845444,
|
| 6139 |
+
"learning_rate": 4.145165031008508e-06,
|
| 6140 |
+
"loss": 0.0047,
|
| 6141 |
+
"step": 8760
|
| 6142 |
+
},
|
| 6143 |
+
{
|
| 6144 |
+
"epoch": 5.557667934093789,
|
| 6145 |
+
"grad_norm": 0.2011323720216751,
|
| 6146 |
+
"learning_rate": 4.079497855898501e-06,
|
| 6147 |
+
"loss": 0.0067,
|
| 6148 |
+
"step": 8770
|
| 6149 |
+
},
|
| 6150 |
+
{
|
| 6151 |
+
"epoch": 5.564005069708491,
|
| 6152 |
+
"grad_norm": 0.06639205664396286,
|
| 6153 |
+
"learning_rate": 4.01433286004283e-06,
|
| 6154 |
+
"loss": 0.0086,
|
| 6155 |
+
"step": 8780
|
| 6156 |
+
},
|
| 6157 |
+
{
|
| 6158 |
+
"epoch": 5.570342205323194,
|
| 6159 |
+
"grad_norm": 0.24140506982803345,
|
| 6160 |
+
"learning_rate": 3.949670756075447e-06,
|
| 6161 |
+
"loss": 0.0101,
|
| 6162 |
+
"step": 8790
|
| 6163 |
+
},
|
| 6164 |
+
{
|
| 6165 |
+
"epoch": 5.576679340937896,
|
| 6166 |
+
"grad_norm": 0.1788935363292694,
|
| 6167 |
+
"learning_rate": 3.885512251130763e-06,
|
| 6168 |
+
"loss": 0.0068,
|
| 6169 |
+
"step": 8800
|
| 6170 |
+
},
|
| 6171 |
+
{
|
| 6172 |
+
"epoch": 5.583016476552598,
|
| 6173 |
+
"grad_norm": 0.10148416459560394,
|
| 6174 |
+
"learning_rate": 3.821858046835913e-06,
|
| 6175 |
+
"loss": 0.011,
|
| 6176 |
+
"step": 8810
|
| 6177 |
+
},
|
| 6178 |
+
{
|
| 6179 |
+
"epoch": 5.589353612167301,
|
| 6180 |
+
"grad_norm": 0.25111889839172363,
|
| 6181 |
+
"learning_rate": 3.75870883930306e-06,
|
| 6182 |
+
"loss": 0.0052,
|
| 6183 |
+
"step": 8820
|
| 6184 |
+
},
|
| 6185 |
+
{
|
| 6186 |
+
"epoch": 5.595690747782003,
|
| 6187 |
+
"grad_norm": 0.1619209200143814,
|
| 6188 |
+
"learning_rate": 3.696065319121833e-06,
|
| 6189 |
+
"loss": 0.0061,
|
| 6190 |
+
"step": 8830
|
| 6191 |
+
},
|
| 6192 |
+
{
|
| 6193 |
+
"epoch": 5.602027883396705,
|
| 6194 |
+
"grad_norm": 0.1424710899591446,
|
| 6195 |
+
"learning_rate": 3.6339281713517303e-06,
|
| 6196 |
+
"loss": 0.0147,
|
| 6197 |
+
"step": 8840
|
| 6198 |
+
},
|
| 6199 |
+
{
|
| 6200 |
+
"epoch": 5.608365019011407,
|
| 6201 |
+
"grad_norm": 0.093357153236866,
|
| 6202 |
+
"learning_rate": 3.5722980755146517e-06,
|
| 6203 |
+
"loss": 0.01,
|
| 6204 |
+
"step": 8850
|
| 6205 |
+
},
|
| 6206 |
+
{
|
| 6207 |
+
"epoch": 5.614702154626109,
|
| 6208 |
+
"grad_norm": 0.09066253155469894,
|
| 6209 |
+
"learning_rate": 3.511175705587433e-06,
|
| 6210 |
+
"loss": 0.0064,
|
| 6211 |
+
"step": 8860
|
| 6212 |
+
},
|
| 6213 |
+
{
|
| 6214 |
+
"epoch": 5.6210392902408115,
|
| 6215 |
+
"grad_norm": 0.24886588752269745,
|
| 6216 |
+
"learning_rate": 3.4505617299945336e-06,
|
| 6217 |
+
"loss": 0.0072,
|
| 6218 |
+
"step": 8870
|
| 6219 |
+
},
|
| 6220 |
+
{
|
| 6221 |
+
"epoch": 5.6273764258555135,
|
| 6222 |
+
"grad_norm": 0.22231824696063995,
|
| 6223 |
+
"learning_rate": 3.390456811600673e-06,
|
| 6224 |
+
"loss": 0.0111,
|
| 6225 |
+
"step": 8880
|
| 6226 |
+
},
|
| 6227 |
+
{
|
| 6228 |
+
"epoch": 5.633713561470215,
|
| 6229 |
+
"grad_norm": 0.11994162201881409,
|
| 6230 |
+
"learning_rate": 3.3308616077036115e-06,
|
| 6231 |
+
"loss": 0.0088,
|
| 6232 |
+
"step": 8890
|
| 6233 |
+
},
|
| 6234 |
+
{
|
| 6235 |
+
"epoch": 5.640050697084917,
|
| 6236 |
+
"grad_norm": 0.11728494614362717,
|
| 6237 |
+
"learning_rate": 3.271776770026963e-06,
|
| 6238 |
+
"loss": 0.0095,
|
| 6239 |
+
"step": 8900
|
| 6240 |
+
},
|
| 6241 |
+
{
|
| 6242 |
+
"epoch": 5.64638783269962,
|
| 6243 |
+
"grad_norm": 0.13038089871406555,
|
| 6244 |
+
"learning_rate": 3.213202944713023e-06,
|
| 6245 |
+
"loss": 0.0053,
|
| 6246 |
+
"step": 8910
|
| 6247 |
+
},
|
| 6248 |
+
{
|
| 6249 |
+
"epoch": 5.652724968314322,
|
| 6250 |
+
"grad_norm": 0.09141723066568375,
|
| 6251 |
+
"learning_rate": 3.155140772315773e-06,
|
| 6252 |
+
"loss": 0.0062,
|
| 6253 |
+
"step": 8920
|
| 6254 |
+
},
|
| 6255 |
+
{
|
| 6256 |
+
"epoch": 5.659062103929024,
|
| 6257 |
+
"grad_norm": 0.07316479086875916,
|
| 6258 |
+
"learning_rate": 3.0975908877938277e-06,
|
| 6259 |
+
"loss": 0.0062,
|
| 6260 |
+
"step": 8930
|
| 6261 |
+
},
|
| 6262 |
+
{
|
| 6263 |
+
"epoch": 5.665399239543726,
|
| 6264 |
+
"grad_norm": 0.18643398582935333,
|
| 6265 |
+
"learning_rate": 3.040553920503503e-06,
|
| 6266 |
+
"loss": 0.0065,
|
| 6267 |
+
"step": 8940
|
| 6268 |
+
},
|
| 6269 |
+
{
|
| 6270 |
+
"epoch": 5.671736375158428,
|
| 6271 |
+
"grad_norm": 0.151958167552948,
|
| 6272 |
+
"learning_rate": 2.9840304941919415e-06,
|
| 6273 |
+
"loss": 0.0055,
|
| 6274 |
+
"step": 8950
|
| 6275 |
+
},
|
| 6276 |
+
{
|
| 6277 |
+
"epoch": 5.678073510773131,
|
| 6278 |
+
"grad_norm": 0.2611534297466278,
|
| 6279 |
+
"learning_rate": 2.928021226990263e-06,
|
| 6280 |
+
"loss": 0.0075,
|
| 6281 |
+
"step": 8960
|
| 6282 |
+
},
|
| 6283 |
+
{
|
| 6284 |
+
"epoch": 5.684410646387833,
|
| 6285 |
+
"grad_norm": 0.16426794230937958,
|
| 6286 |
+
"learning_rate": 2.8725267314068495e-06,
|
| 6287 |
+
"loss": 0.0046,
|
| 6288 |
+
"step": 8970
|
| 6289 |
+
},
|
| 6290 |
+
{
|
| 6291 |
+
"epoch": 5.690747782002535,
|
| 6292 |
+
"grad_norm": 0.11258076876401901,
|
| 6293 |
+
"learning_rate": 2.817547614320615e-06,
|
| 6294 |
+
"loss": 0.0095,
|
| 6295 |
+
"step": 8980
|
| 6296 |
+
},
|
| 6297 |
+
{
|
| 6298 |
+
"epoch": 5.697084917617237,
|
| 6299 |
+
"grad_norm": 0.15174216032028198,
|
| 6300 |
+
"learning_rate": 2.7630844769743757e-06,
|
| 6301 |
+
"loss": 0.0069,
|
| 6302 |
+
"step": 8990
|
| 6303 |
+
},
|
| 6304 |
+
{
|
| 6305 |
+
"epoch": 5.7034220532319395,
|
| 6306 |
+
"grad_norm": 0.15580610930919647,
|
| 6307 |
+
"learning_rate": 2.7091379149682685e-06,
|
| 6308 |
+
"loss": 0.0079,
|
| 6309 |
+
"step": 9000
|
| 6310 |
+
},
|
| 6311 |
+
{
|
| 6312 |
+
"epoch": 5.7097591888466415,
|
| 6313 |
+
"grad_norm": 0.12744936347007751,
|
| 6314 |
+
"learning_rate": 2.6557085182532582e-06,
|
| 6315 |
+
"loss": 0.0086,
|
| 6316 |
+
"step": 9010
|
| 6317 |
+
},
|
| 6318 |
+
{
|
| 6319 |
+
"epoch": 5.716096324461343,
|
| 6320 |
+
"grad_norm": 0.1542847603559494,
|
| 6321 |
+
"learning_rate": 2.602796871124663e-06,
|
| 6322 |
+
"loss": 0.0096,
|
| 6323 |
+
"step": 9020
|
| 6324 |
+
},
|
| 6325 |
+
{
|
| 6326 |
+
"epoch": 5.722433460076045,
|
| 6327 |
+
"grad_norm": 0.13958968222141266,
|
| 6328 |
+
"learning_rate": 2.5504035522157854e-06,
|
| 6329 |
+
"loss": 0.0089,
|
| 6330 |
+
"step": 9030
|
| 6331 |
+
},
|
| 6332 |
+
{
|
| 6333 |
+
"epoch": 5.728770595690747,
|
| 6334 |
+
"grad_norm": 0.20519228279590607,
|
| 6335 |
+
"learning_rate": 2.4985291344915674e-06,
|
| 6336 |
+
"loss": 0.0061,
|
| 6337 |
+
"step": 9040
|
| 6338 |
+
},
|
| 6339 |
+
{
|
| 6340 |
+
"epoch": 5.73510773130545,
|
| 6341 |
+
"grad_norm": 0.11365640163421631,
|
| 6342 |
+
"learning_rate": 2.4471741852423237e-06,
|
| 6343 |
+
"loss": 0.0046,
|
| 6344 |
+
"step": 9050
|
| 6345 |
+
},
|
| 6346 |
+
{
|
| 6347 |
+
"epoch": 5.741444866920152,
|
| 6348 |
+
"grad_norm": 0.0870782658457756,
|
| 6349 |
+
"learning_rate": 2.3963392660775575e-06,
|
| 6350 |
+
"loss": 0.0079,
|
| 6351 |
+
"step": 9060
|
| 6352 |
+
},
|
| 6353 |
+
{
|
| 6354 |
+
"epoch": 5.747782002534854,
|
| 6355 |
+
"grad_norm": 0.07872149348258972,
|
| 6356 |
+
"learning_rate": 2.3460249329197824e-06,
|
| 6357 |
+
"loss": 0.0055,
|
| 6358 |
+
"step": 9070
|
| 6359 |
+
},
|
| 6360 |
+
{
|
| 6361 |
+
"epoch": 5.754119138149557,
|
| 6362 |
+
"grad_norm": 0.08804658055305481,
|
| 6363 |
+
"learning_rate": 2.296231735998511e-06,
|
| 6364 |
+
"loss": 0.007,
|
| 6365 |
+
"step": 9080
|
| 6366 |
+
},
|
| 6367 |
+
{
|
| 6368 |
+
"epoch": 5.760456273764259,
|
| 6369 |
+
"grad_norm": 0.05982697755098343,
|
| 6370 |
+
"learning_rate": 2.2469602198441573e-06,
|
| 6371 |
+
"loss": 0.0044,
|
| 6372 |
+
"step": 9090
|
| 6373 |
+
},
|
| 6374 |
+
{
|
| 6375 |
+
"epoch": 5.766793409378961,
|
| 6376 |
+
"grad_norm": 0.10259458422660828,
|
| 6377 |
+
"learning_rate": 2.1982109232821178e-06,
|
| 6378 |
+
"loss": 0.006,
|
| 6379 |
+
"step": 9100
|
| 6380 |
+
},
|
| 6381 |
+
{
|
| 6382 |
+
"epoch": 5.773130544993663,
|
| 6383 |
+
"grad_norm": 0.1546311378479004,
|
| 6384 |
+
"learning_rate": 2.149984379426906e-06,
|
| 6385 |
+
"loss": 0.0067,
|
| 6386 |
+
"step": 9110
|
| 6387 |
+
},
|
| 6388 |
+
{
|
| 6389 |
+
"epoch": 5.779467680608365,
|
| 6390 |
+
"grad_norm": 0.14351686835289001,
|
| 6391 |
+
"learning_rate": 2.102281115676258e-06,
|
| 6392 |
+
"loss": 0.0056,
|
| 6393 |
+
"step": 9120
|
| 6394 |
+
},
|
| 6395 |
+
{
|
| 6396 |
+
"epoch": 5.7858048162230675,
|
| 6397 |
+
"grad_norm": 0.05316438898444176,
|
| 6398 |
+
"learning_rate": 2.0551016537054493e-06,
|
| 6399 |
+
"loss": 0.0061,
|
| 6400 |
+
"step": 9130
|
| 6401 |
+
},
|
| 6402 |
+
{
|
| 6403 |
+
"epoch": 5.7921419518377695,
|
| 6404 |
+
"grad_norm": 0.0845993310213089,
|
| 6405 |
+
"learning_rate": 2.008446509461498e-06,
|
| 6406 |
+
"loss": 0.006,
|
| 6407 |
+
"step": 9140
|
| 6408 |
+
},
|
| 6409 |
+
{
|
| 6410 |
+
"epoch": 5.798479087452471,
|
| 6411 |
+
"grad_norm": 0.156995952129364,
|
| 6412 |
+
"learning_rate": 1.962316193157593e-06,
|
| 6413 |
+
"loss": 0.0054,
|
| 6414 |
+
"step": 9150
|
| 6415 |
+
},
|
| 6416 |
+
{
|
| 6417 |
+
"epoch": 5.804816223067173,
|
| 6418 |
+
"grad_norm": 0.20765309035778046,
|
| 6419 |
+
"learning_rate": 1.91671120926748e-06,
|
| 6420 |
+
"loss": 0.0049,
|
| 6421 |
+
"step": 9160
|
| 6422 |
+
},
|
| 6423 |
+
{
|
| 6424 |
+
"epoch": 5.811153358681876,
|
| 6425 |
+
"grad_norm": 0.15031087398529053,
|
| 6426 |
+
"learning_rate": 1.8716320565199618e-06,
|
| 6427 |
+
"loss": 0.0069,
|
| 6428 |
+
"step": 9170
|
| 6429 |
+
},
|
| 6430 |
+
{
|
| 6431 |
+
"epoch": 5.817490494296578,
|
| 6432 |
+
"grad_norm": 0.08898857980966568,
|
| 6433 |
+
"learning_rate": 1.8270792278934302e-06,
|
| 6434 |
+
"loss": 0.0052,
|
| 6435 |
+
"step": 9180
|
| 6436 |
+
},
|
| 6437 |
+
{
|
| 6438 |
+
"epoch": 5.82382762991128,
|
| 6439 |
+
"grad_norm": 0.16047555208206177,
|
| 6440 |
+
"learning_rate": 1.7830532106104747e-06,
|
| 6441 |
+
"loss": 0.0061,
|
| 6442 |
+
"step": 9190
|
| 6443 |
+
},
|
| 6444 |
+
{
|
| 6445 |
+
"epoch": 5.830164765525982,
|
| 6446 |
+
"grad_norm": 0.24277397990226746,
|
| 6447 |
+
"learning_rate": 1.7395544861325718e-06,
|
| 6448 |
+
"loss": 0.0061,
|
| 6449 |
+
"step": 9200
|
| 6450 |
+
},
|
| 6451 |
+
{
|
| 6452 |
+
"epoch": 5.836501901140684,
|
| 6453 |
+
"grad_norm": 0.1472398340702057,
|
| 6454 |
+
"learning_rate": 1.696583530154794e-06,
|
| 6455 |
+
"loss": 0.0123,
|
| 6456 |
+
"step": 9210
|
| 6457 |
+
},
|
| 6458 |
+
{
|
| 6459 |
+
"epoch": 5.842839036755387,
|
| 6460 |
+
"grad_norm": 0.15417790412902832,
|
| 6461 |
+
"learning_rate": 1.6541408126006463e-06,
|
| 6462 |
+
"loss": 0.0058,
|
| 6463 |
+
"step": 9220
|
| 6464 |
+
},
|
| 6465 |
+
{
|
| 6466 |
+
"epoch": 5.849176172370089,
|
| 6467 |
+
"grad_norm": 0.10276266187429428,
|
| 6468 |
+
"learning_rate": 1.6122267976168781e-06,
|
| 6469 |
+
"loss": 0.0076,
|
| 6470 |
+
"step": 9230
|
| 6471 |
+
},
|
| 6472 |
+
{
|
| 6473 |
+
"epoch": 5.855513307984791,
|
| 6474 |
+
"grad_norm": 0.22714407742023468,
|
| 6475 |
+
"learning_rate": 1.5708419435684462e-06,
|
| 6476 |
+
"loss": 0.0079,
|
| 6477 |
+
"step": 9240
|
| 6478 |
+
},
|
| 6479 |
+
{
|
| 6480 |
+
"epoch": 5.861850443599493,
|
| 6481 |
+
"grad_norm": 0.08724337071180344,
|
| 6482 |
+
"learning_rate": 1.5299867030334814e-06,
|
| 6483 |
+
"loss": 0.006,
|
| 6484 |
+
"step": 9250
|
| 6485 |
+
},
|
| 6486 |
+
{
|
| 6487 |
+
"epoch": 5.8681875792141955,
|
| 6488 |
+
"grad_norm": 0.1354474574327469,
|
| 6489 |
+
"learning_rate": 1.4896615227983468e-06,
|
| 6490 |
+
"loss": 0.0063,
|
| 6491 |
+
"step": 9260
|
| 6492 |
+
},
|
| 6493 |
+
{
|
| 6494 |
+
"epoch": 5.8745247148288975,
|
| 6495 |
+
"grad_norm": 0.09567315876483917,
|
| 6496 |
+
"learning_rate": 1.4498668438527597e-06,
|
| 6497 |
+
"loss": 0.0073,
|
| 6498 |
+
"step": 9270
|
| 6499 |
+
},
|
| 6500 |
+
{
|
| 6501 |
+
"epoch": 5.880861850443599,
|
| 6502 |
+
"grad_norm": 0.25440752506256104,
|
| 6503 |
+
"learning_rate": 1.4106031013849496e-06,
|
| 6504 |
+
"loss": 0.0059,
|
| 6505 |
+
"step": 9280
|
| 6506 |
+
},
|
| 6507 |
+
{
|
| 6508 |
+
"epoch": 5.887198986058301,
|
| 6509 |
+
"grad_norm": 0.046442631632089615,
|
| 6510 |
+
"learning_rate": 1.3718707247769135e-06,
|
| 6511 |
+
"loss": 0.0085,
|
| 6512 |
+
"step": 9290
|
| 6513 |
+
},
|
| 6514 |
+
{
|
| 6515 |
+
"epoch": 5.893536121673003,
|
| 6516 |
+
"grad_norm": 0.202633798122406,
|
| 6517 |
+
"learning_rate": 1.333670137599713e-06,
|
| 6518 |
+
"loss": 0.0078,
|
| 6519 |
+
"step": 9300
|
| 6520 |
+
},
|
| 6521 |
+
{
|
| 6522 |
+
"epoch": 5.899873257287706,
|
| 6523 |
+
"grad_norm": 0.11459842324256897,
|
| 6524 |
+
"learning_rate": 1.2960017576088446e-06,
|
| 6525 |
+
"loss": 0.0049,
|
| 6526 |
+
"step": 9310
|
| 6527 |
+
},
|
| 6528 |
+
{
|
| 6529 |
+
"epoch": 5.906210392902408,
|
| 6530 |
+
"grad_norm": 0.06873945146799088,
|
| 6531 |
+
"learning_rate": 1.2588659967397e-06,
|
| 6532 |
+
"loss": 0.0172,
|
| 6533 |
+
"step": 9320
|
| 6534 |
+
},
|
| 6535 |
+
{
|
| 6536 |
+
"epoch": 5.91254752851711,
|
| 6537 |
+
"grad_norm": 0.17459562420845032,
|
| 6538 |
+
"learning_rate": 1.222263261102985e-06,
|
| 6539 |
+
"loss": 0.0071,
|
| 6540 |
+
"step": 9330
|
| 6541 |
+
},
|
| 6542 |
+
{
|
| 6543 |
+
"epoch": 5.918884664131813,
|
| 6544 |
+
"grad_norm": 0.15463189780712128,
|
| 6545 |
+
"learning_rate": 1.1861939509803687e-06,
|
| 6546 |
+
"loss": 0.0068,
|
| 6547 |
+
"step": 9340
|
| 6548 |
+
},
|
| 6549 |
+
{
|
| 6550 |
+
"epoch": 5.925221799746515,
|
| 6551 |
+
"grad_norm": 0.052368469536304474,
|
| 6552 |
+
"learning_rate": 1.1506584608200367e-06,
|
| 6553 |
+
"loss": 0.0063,
|
| 6554 |
+
"step": 9350
|
| 6555 |
+
},
|
| 6556 |
+
{
|
| 6557 |
+
"epoch": 5.931558935361217,
|
| 6558 |
+
"grad_norm": 0.08080845326185226,
|
| 6559 |
+
"learning_rate": 1.1156571792324211e-06,
|
| 6560 |
+
"loss": 0.0037,
|
| 6561 |
+
"step": 9360
|
| 6562 |
+
},
|
| 6563 |
+
{
|
| 6564 |
+
"epoch": 5.937896070975919,
|
| 6565 |
+
"grad_norm": 0.1452123522758484,
|
| 6566 |
+
"learning_rate": 1.0811904889859336e-06,
|
| 6567 |
+
"loss": 0.0106,
|
| 6568 |
+
"step": 9370
|
| 6569 |
+
},
|
| 6570 |
+
{
|
| 6571 |
+
"epoch": 5.944233206590621,
|
| 6572 |
+
"grad_norm": 0.18937888741493225,
|
| 6573 |
+
"learning_rate": 1.0472587670027678e-06,
|
| 6574 |
+
"loss": 0.0045,
|
| 6575 |
+
"step": 9380
|
| 6576 |
+
},
|
| 6577 |
+
{
|
| 6578 |
+
"epoch": 5.9505703422053235,
|
| 6579 |
+
"grad_norm": 0.1064092218875885,
|
| 6580 |
+
"learning_rate": 1.0138623843548078e-06,
|
| 6581 |
+
"loss": 0.01,
|
| 6582 |
+
"step": 9390
|
| 6583 |
+
},
|
| 6584 |
+
{
|
| 6585 |
+
"epoch": 5.9569074778200255,
|
| 6586 |
+
"grad_norm": 0.15949054062366486,
|
| 6587 |
+
"learning_rate": 9.810017062595322e-07,
|
| 6588 |
+
"loss": 0.0093,
|
| 6589 |
+
"step": 9400
|
| 6590 |
+
},
|
| 6591 |
+
{
|
| 6592 |
+
"epoch": 5.9632446134347274,
|
| 6593 |
+
"grad_norm": 0.11651349812746048,
|
| 6594 |
+
"learning_rate": 9.486770920760668e-07,
|
| 6595 |
+
"loss": 0.0062,
|
| 6596 |
+
"step": 9410
|
| 6597 |
+
},
|
| 6598 |
+
{
|
| 6599 |
+
"epoch": 5.969581749049429,
|
| 6600 |
+
"grad_norm": 0.16830092668533325,
|
| 6601 |
+
"learning_rate": 9.168888953011989e-07,
|
| 6602 |
+
"loss": 0.005,
|
| 6603 |
+
"step": 9420
|
| 6604 |
+
},
|
| 6605 |
+
{
|
| 6606 |
+
"epoch": 5.975918884664132,
|
| 6607 |
+
"grad_norm": 0.08310183882713318,
|
| 6608 |
+
"learning_rate": 8.856374635655695e-07,
|
| 6609 |
+
"loss": 0.0059,
|
| 6610 |
+
"step": 9430
|
| 6611 |
+
},
|
| 6612 |
+
{
|
| 6613 |
+
"epoch": 5.982256020278834,
|
| 6614 |
+
"grad_norm": 0.12785309553146362,
|
| 6615 |
+
"learning_rate": 8.549231386298151e-07,
|
| 6616 |
+
"loss": 0.0097,
|
| 6617 |
+
"step": 9440
|
| 6618 |
+
},
|
| 6619 |
+
{
|
| 6620 |
+
"epoch": 5.988593155893536,
|
| 6621 |
+
"grad_norm": 0.2574361264705658,
|
| 6622 |
+
"learning_rate": 8.247462563808817e-07,
|
| 6623 |
+
"loss": 0.0071,
|
| 6624 |
+
"step": 9450
|
| 6625 |
+
},
|
| 6626 |
+
{
|
| 6627 |
+
"epoch": 5.994930291508238,
|
| 6628 |
+
"grad_norm": 0.31402966380119324,
|
| 6629 |
+
"learning_rate": 7.951071468283167e-07,
|
| 6630 |
+
"loss": 0.0065,
|
| 6631 |
+
"step": 9460
|
| 6632 |
+
},
|
| 6633 |
+
{
|
| 6634 |
+
"epoch": 6.00126742712294,
|
| 6635 |
+
"grad_norm": 0.1356462836265564,
|
| 6636 |
+
"learning_rate": 7.66006134100672e-07,
|
| 6637 |
+
"loss": 0.0058,
|
| 6638 |
+
"step": 9470
|
| 6639 |
+
},
|
| 6640 |
+
{
|
| 6641 |
+
"epoch": 6.007604562737643,
|
| 6642 |
+
"grad_norm": 0.08312007039785385,
|
| 6643 |
+
"learning_rate": 7.374435364419674e-07,
|
| 6644 |
+
"loss": 0.004,
|
| 6645 |
+
"step": 9480
|
| 6646 |
+
},
|
| 6647 |
+
{
|
| 6648 |
+
"epoch": 6.013941698352345,
|
| 6649 |
+
"grad_norm": 0.25194162130355835,
|
| 6650 |
+
"learning_rate": 7.094196662081831e-07,
|
| 6651 |
+
"loss": 0.0105,
|
| 6652 |
+
"step": 9490
|
| 6653 |
+
},
|
| 6654 |
+
{
|
| 6655 |
+
"epoch": 6.020278833967047,
|
| 6656 |
+
"grad_norm": 0.11876359581947327,
|
| 6657 |
+
"learning_rate": 6.819348298638839e-07,
|
| 6658 |
+
"loss": 0.005,
|
| 6659 |
+
"step": 9500
|
| 6660 |
+
},
|
| 6661 |
+
{
|
| 6662 |
+
"epoch": 6.026615969581749,
|
| 6663 |
+
"grad_norm": 0.11712250113487244,
|
| 6664 |
+
"learning_rate": 6.549893279788277e-07,
|
| 6665 |
+
"loss": 0.0087,
|
| 6666 |
+
"step": 9510
|
| 6667 |
+
},
|
| 6668 |
+
{
|
| 6669 |
+
"epoch": 6.032953105196452,
|
| 6670 |
+
"grad_norm": 0.24556323885917664,
|
| 6671 |
+
"learning_rate": 6.285834552247128e-07,
|
| 6672 |
+
"loss": 0.0059,
|
| 6673 |
+
"step": 9520
|
| 6674 |
+
},
|
| 6675 |
+
{
|
| 6676 |
+
"epoch": 6.0392902408111535,
|
| 6677 |
+
"grad_norm": 0.12040778994560242,
|
| 6678 |
+
"learning_rate": 6.027175003719354e-07,
|
| 6679 |
+
"loss": 0.0062,
|
| 6680 |
+
"step": 9530
|
| 6681 |
+
},
|
| 6682 |
+
{
|
| 6683 |
+
"epoch": 6.0456273764258555,
|
| 6684 |
+
"grad_norm": 0.07460421323776245,
|
| 6685 |
+
"learning_rate": 5.773917462864264e-07,
|
| 6686 |
+
"loss": 0.0091,
|
| 6687 |
+
"step": 9540
|
| 6688 |
+
},
|
| 6689 |
+
{
|
| 6690 |
+
"epoch": 6.051964512040557,
|
| 6691 |
+
"grad_norm": 0.18476122617721558,
|
| 6692 |
+
"learning_rate": 5.526064699265753e-07,
|
| 6693 |
+
"loss": 0.01,
|
| 6694 |
+
"step": 9550
|
| 6695 |
+
},
|
| 6696 |
+
{
|
| 6697 |
+
"epoch": 6.05830164765526,
|
| 6698 |
+
"grad_norm": 0.06253950297832489,
|
| 6699 |
+
"learning_rate": 5.283619423401998e-07,
|
| 6700 |
+
"loss": 0.0076,
|
| 6701 |
+
"step": 9560
|
| 6702 |
+
},
|
| 6703 |
+
{
|
| 6704 |
+
"epoch": 6.064638783269962,
|
| 6705 |
+
"grad_norm": 0.07986325025558472,
|
| 6706 |
+
"learning_rate": 5.046584286615697e-07,
|
| 6707 |
+
"loss": 0.0111,
|
| 6708 |
+
"step": 9570
|
| 6709 |
+
},
|
| 6710 |
+
{
|
| 6711 |
+
"epoch": 6.070975918884664,
|
| 6712 |
+
"grad_norm": 0.25652700662612915,
|
| 6713 |
+
"learning_rate": 4.814961881085045e-07,
|
| 6714 |
+
"loss": 0.004,
|
| 6715 |
+
"step": 9580
|
| 6716 |
+
},
|
| 6717 |
+
{
|
| 6718 |
+
"epoch": 6.077313054499366,
|
| 6719 |
+
"grad_norm": 0.10732295364141464,
|
| 6720 |
+
"learning_rate": 4.5887547397955864e-07,
|
| 6721 |
+
"loss": 0.0052,
|
| 6722 |
+
"step": 9590
|
| 6723 |
+
},
|
| 6724 |
+
{
|
| 6725 |
+
"epoch": 6.083650190114068,
|
| 6726 |
+
"grad_norm": 0.09151678532361984,
|
| 6727 |
+
"learning_rate": 4.367965336512403e-07,
|
| 6728 |
+
"loss": 0.0109,
|
| 6729 |
+
"step": 9600
|
| 6730 |
+
},
|
| 6731 |
+
{
|
| 6732 |
+
"epoch": 6.089987325728771,
|
| 6733 |
+
"grad_norm": 0.057842839509248734,
|
| 6734 |
+
"learning_rate": 4.1525960857530243e-07,
|
| 6735 |
+
"loss": 0.0059,
|
| 6736 |
+
"step": 9610
|
| 6737 |
+
},
|
| 6738 |
+
{
|
| 6739 |
+
"epoch": 6.096324461343473,
|
| 6740 |
+
"grad_norm": 0.1317504197359085,
|
| 6741 |
+
"learning_rate": 3.9426493427611177e-07,
|
| 6742 |
+
"loss": 0.0067,
|
| 6743 |
+
"step": 9620
|
| 6744 |
+
},
|
| 6745 |
+
{
|
| 6746 |
+
"epoch": 6.102661596958175,
|
| 6747 |
+
"grad_norm": 0.0697597935795784,
|
| 6748 |
+
"learning_rate": 3.738127403480507e-07,
|
| 6749 |
+
"loss": 0.0064,
|
| 6750 |
+
"step": 9630
|
| 6751 |
+
},
|
| 6752 |
+
{
|
| 6753 |
+
"epoch": 6.108998732572877,
|
| 6754 |
+
"grad_norm": 0.3189285695552826,
|
| 6755 |
+
"learning_rate": 3.5390325045304706e-07,
|
| 6756 |
+
"loss": 0.0066,
|
| 6757 |
+
"step": 9640
|
| 6758 |
+
},
|
| 6759 |
+
{
|
| 6760 |
+
"epoch": 6.11533586818758,
|
| 6761 |
+
"grad_norm": 0.07586175948381424,
|
| 6762 |
+
"learning_rate": 3.3453668231809286e-07,
|
| 6763 |
+
"loss": 0.0061,
|
| 6764 |
+
"step": 9650
|
| 6765 |
+
},
|
| 6766 |
+
{
|
| 6767 |
+
"epoch": 6.1216730038022815,
|
| 6768 |
+
"grad_norm": 0.28144219517707825,
|
| 6769 |
+
"learning_rate": 3.157132477328628e-07,
|
| 6770 |
+
"loss": 0.0061,
|
| 6771 |
+
"step": 9660
|
| 6772 |
+
},
|
| 6773 |
+
{
|
| 6774 |
+
"epoch": 6.1280101394169835,
|
| 6775 |
+
"grad_norm": 0.21867600083351135,
|
| 6776 |
+
"learning_rate": 2.9743315254743833e-07,
|
| 6777 |
+
"loss": 0.016,
|
| 6778 |
+
"step": 9670
|
| 6779 |
+
},
|
| 6780 |
+
{
|
| 6781 |
+
"epoch": 6.134347275031685,
|
| 6782 |
+
"grad_norm": 0.08317314088344574,
|
| 6783 |
+
"learning_rate": 2.796965966699927e-07,
|
| 6784 |
+
"loss": 0.0143,
|
| 6785 |
+
"step": 9680
|
| 6786 |
+
},
|
| 6787 |
+
{
|
| 6788 |
+
"epoch": 6.140684410646388,
|
| 6789 |
+
"grad_norm": 0.13135367631912231,
|
| 6790 |
+
"learning_rate": 2.625037740646763e-07,
|
| 6791 |
+
"loss": 0.0058,
|
| 6792 |
+
"step": 9690
|
| 6793 |
+
},
|
| 6794 |
+
{
|
| 6795 |
+
"epoch": 6.14702154626109,
|
| 6796 |
+
"grad_norm": 0.06645505130290985,
|
| 6797 |
+
"learning_rate": 2.458548727494292e-07,
|
| 6798 |
+
"loss": 0.0109,
|
| 6799 |
+
"step": 9700
|
| 6800 |
+
},
|
| 6801 |
+
{
|
| 6802 |
+
"epoch": 6.153358681875792,
|
| 6803 |
+
"grad_norm": 0.09750212728977203,
|
| 6804 |
+
"learning_rate": 2.2975007479397738e-07,
|
| 6805 |
+
"loss": 0.0071,
|
| 6806 |
+
"step": 9710
|
| 6807 |
+
},
|
| 6808 |
+
{
|
| 6809 |
+
"epoch": 6.159695817490494,
|
| 6810 |
+
"grad_norm": 0.09561249613761902,
|
| 6811 |
+
"learning_rate": 2.1418955631781202e-07,
|
| 6812 |
+
"loss": 0.0076,
|
| 6813 |
+
"step": 9720
|
| 6814 |
+
},
|
| 6815 |
+
{
|
| 6816 |
+
"epoch": 6.166032953105196,
|
| 6817 |
+
"grad_norm": 0.1740742325782776,
|
| 6818 |
+
"learning_rate": 1.9917348748826335e-07,
|
| 6819 |
+
"loss": 0.006,
|
| 6820 |
+
"step": 9730
|
| 6821 |
+
},
|
| 6822 |
+
{
|
| 6823 |
+
"epoch": 6.172370088719899,
|
| 6824 |
+
"grad_norm": 0.18339525163173676,
|
| 6825 |
+
"learning_rate": 1.847020325186577e-07,
|
| 6826 |
+
"loss": 0.006,
|
| 6827 |
+
"step": 9740
|
| 6828 |
+
},
|
| 6829 |
+
{
|
| 6830 |
+
"epoch": 6.178707224334601,
|
| 6831 |
+
"grad_norm": 0.06436607986688614,
|
| 6832 |
+
"learning_rate": 1.7077534966650766e-07,
|
| 6833 |
+
"loss": 0.0122,
|
| 6834 |
+
"step": 9750
|
| 6835 |
+
},
|
| 6836 |
+
{
|
| 6837 |
+
"epoch": 6.185044359949303,
|
| 6838 |
+
"grad_norm": 0.05160180851817131,
|
| 6839 |
+
"learning_rate": 1.5739359123178587e-07,
|
| 6840 |
+
"loss": 0.008,
|
| 6841 |
+
"step": 9760
|
| 6842 |
+
},
|
| 6843 |
+
{
|
| 6844 |
+
"epoch": 6.191381495564005,
|
| 6845 |
+
"grad_norm": 0.0478476956486702,
|
| 6846 |
+
"learning_rate": 1.4455690355525964e-07,
|
| 6847 |
+
"loss": 0.0054,
|
| 6848 |
+
"step": 9770
|
| 6849 |
+
},
|
| 6850 |
+
{
|
| 6851 |
+
"epoch": 6.197718631178708,
|
| 6852 |
+
"grad_norm": 0.13458269834518433,
|
| 6853 |
+
"learning_rate": 1.3226542701689215e-07,
|
| 6854 |
+
"loss": 0.0051,
|
| 6855 |
+
"step": 9780
|
| 6856 |
+
},
|
| 6857 |
+
{
|
| 6858 |
+
"epoch": 6.2040557667934095,
|
| 6859 |
+
"grad_norm": 0.14054587483406067,
|
| 6860 |
+
"learning_rate": 1.2051929603428825e-07,
|
| 6861 |
+
"loss": 0.0066,
|
| 6862 |
+
"step": 9790
|
| 6863 |
+
},
|
| 6864 |
+
{
|
| 6865 |
+
"epoch": 6.2103929024081115,
|
| 6866 |
+
"grad_norm": 0.3035231828689575,
|
| 6867 |
+
"learning_rate": 1.0931863906127327e-07,
|
| 6868 |
+
"loss": 0.0057,
|
| 6869 |
+
"step": 9800
|
| 6870 |
+
},
|
| 6871 |
+
{
|
| 6872 |
+
"epoch": 6.216730038022813,
|
| 6873 |
+
"grad_norm": 0.3359600901603699,
|
| 6874 |
+
"learning_rate": 9.866357858642205e-08,
|
| 6875 |
+
"loss": 0.0088,
|
| 6876 |
+
"step": 9810
|
| 6877 |
+
},
|
| 6878 |
+
{
|
| 6879 |
+
"epoch": 6.223067173637516,
|
| 6880 |
+
"grad_norm": 0.13854598999023438,
|
| 6881 |
+
"learning_rate": 8.855423113177664e-08,
|
| 6882 |
+
"loss": 0.0057,
|
| 6883 |
+
"step": 9820
|
| 6884 |
+
},
|
| 6885 |
+
{
|
| 6886 |
+
"epoch": 6.229404309252218,
|
| 6887 |
+
"grad_norm": 0.061560165137052536,
|
| 6888 |
+
"learning_rate": 7.899070725153613e-08,
|
| 6889 |
+
"loss": 0.0047,
|
| 6890 |
+
"step": 9830
|
| 6891 |
+
},
|
| 6892 |
+
{
|
| 6893 |
+
"epoch": 6.23574144486692,
|
| 6894 |
+
"grad_norm": 0.1461103856563568,
|
| 6895 |
+
"learning_rate": 6.997311153086883e-08,
|
| 6896 |
+
"loss": 0.004,
|
| 6897 |
+
"step": 9840
|
| 6898 |
+
},
|
| 6899 |
+
{
|
| 6900 |
+
"epoch": 6.242078580481622,
|
| 6901 |
+
"grad_norm": 0.3121045231819153,
|
| 6902 |
+
"learning_rate": 6.150154258476315e-08,
|
| 6903 |
+
"loss": 0.0073,
|
| 6904 |
+
"step": 9850
|
| 6905 |
+
},
|
| 6906 |
+
{
|
| 6907 |
+
"epoch": 6.248415716096324,
|
| 6908 |
+
"grad_norm": 0.18830816447734833,
|
| 6909 |
+
"learning_rate": 5.3576093056922906e-08,
|
| 6910 |
+
"loss": 0.0067,
|
| 6911 |
+
"step": 9860
|
| 6912 |
+
},
|
| 6913 |
+
{
|
| 6914 |
+
"epoch": 6.254752851711027,
|
| 6915 |
+
"grad_norm": 0.060759589076042175,
|
| 6916 |
+
"learning_rate": 4.619684961881254e-08,
|
| 6917 |
+
"loss": 0.005,
|
| 6918 |
+
"step": 9870
|
| 6919 |
+
},
|
| 6920 |
+
{
|
| 6921 |
+
"epoch": 6.261089987325729,
|
| 6922 |
+
"grad_norm": 0.1718645989894867,
|
| 6923 |
+
"learning_rate": 3.936389296864129e-08,
|
| 6924 |
+
"loss": 0.0049,
|
| 6925 |
+
"step": 9880
|
| 6926 |
+
},
|
| 6927 |
+
{
|
| 6928 |
+
"epoch": 6.267427122940431,
|
| 6929 |
+
"grad_norm": 0.09834865480661392,
|
| 6930 |
+
"learning_rate": 3.3077297830541584e-08,
|
| 6931 |
+
"loss": 0.0066,
|
| 6932 |
+
"step": 9890
|
| 6933 |
+
},
|
| 6934 |
+
{
|
| 6935 |
+
"epoch": 6.273764258555133,
|
| 6936 |
+
"grad_norm": 0.11674518138170242,
|
| 6937 |
+
"learning_rate": 2.7337132953697554e-08,
|
| 6938 |
+
"loss": 0.0086,
|
| 6939 |
+
"step": 9900
|
| 6940 |
+
},
|
| 6941 |
+
{
|
| 6942 |
+
"epoch": 6.280101394169836,
|
| 6943 |
+
"grad_norm": 0.17682595551013947,
|
| 6944 |
+
"learning_rate": 2.214346111164556e-08,
|
| 6945 |
+
"loss": 0.0065,
|
| 6946 |
+
"step": 9910
|
| 6947 |
+
},
|
| 6948 |
+
{
|
| 6949 |
+
"epoch": 6.2864385297845375,
|
| 6950 |
+
"grad_norm": 0.07229286432266235,
|
| 6951 |
+
"learning_rate": 1.749633910153592e-08,
|
| 6952 |
+
"loss": 0.0038,
|
| 6953 |
+
"step": 9920
|
| 6954 |
+
},
|
| 6955 |
+
{
|
| 6956 |
+
"epoch": 6.2927756653992395,
|
| 6957 |
+
"grad_norm": 0.083204485476017,
|
| 6958 |
+
"learning_rate": 1.3395817743561134e-08,
|
| 6959 |
+
"loss": 0.0057,
|
| 6960 |
+
"step": 9930
|
| 6961 |
+
},
|
| 6962 |
+
{
|
| 6963 |
+
"epoch": 6.299112801013941,
|
| 6964 |
+
"grad_norm": 0.2328864485025406,
|
| 6965 |
+
"learning_rate": 9.841941880361916e-09,
|
| 6966 |
+
"loss": 0.0108,
|
| 6967 |
+
"step": 9940
|
| 6968 |
+
},
|
| 6969 |
+
{
|
| 6970 |
+
"epoch": 6.305449936628644,
|
| 6971 |
+
"grad_norm": 0.1641470342874527,
|
| 6972 |
+
"learning_rate": 6.834750376549792e-09,
|
| 6973 |
+
"loss": 0.007,
|
| 6974 |
+
"step": 9950
|
| 6975 |
+
},
|
| 6976 |
+
{
|
| 6977 |
+
"epoch": 6.311787072243346,
|
| 6978 |
+
"grad_norm": 0.08585844933986664,
|
| 6979 |
+
"learning_rate": 4.3742761183018784e-09,
|
| 6980 |
+
"loss": 0.012,
|
| 6981 |
+
"step": 9960
|
| 6982 |
+
},
|
| 6983 |
+
{
|
| 6984 |
+
"epoch": 6.318124207858048,
|
| 6985 |
+
"grad_norm": 0.10597538948059082,
|
| 6986 |
+
"learning_rate": 2.4605460129556445e-09,
|
| 6987 |
+
"loss": 0.004,
|
| 6988 |
+
"step": 9970
|
| 6989 |
+
},
|
| 6990 |
+
{
|
| 6991 |
+
"epoch": 6.32446134347275,
|
| 6992 |
+
"grad_norm": 0.21842822432518005,
|
| 6993 |
+
"learning_rate": 1.0935809887702154e-09,
|
| 6994 |
+
"loss": 0.006,
|
| 6995 |
+
"step": 9980
|
| 6996 |
+
},
|
| 6997 |
+
{
|
| 6998 |
+
"epoch": 6.330798479087452,
|
| 6999 |
+
"grad_norm": 0.0765758827328682,
|
| 7000 |
+
"learning_rate": 2.7339599464326627e-10,
|
| 7001 |
+
"loss": 0.0059,
|
| 7002 |
+
"step": 9990
|
| 7003 |
+
},
|
| 7004 |
+
{
|
| 7005 |
+
"epoch": 6.337135614702155,
|
| 7006 |
+
"grad_norm": 0.09744177758693695,
|
| 7007 |
+
"learning_rate": 0.0,
|
| 7008 |
+
"loss": 0.0057,
|
| 7009 |
+
"step": 10000
|
| 7010 |
+
},
|
| 7011 |
+
{
|
| 7012 |
+
"epoch": 6.337135614702155,
|
| 7013 |
+
"step": 10000,
|
| 7014 |
+
"total_flos": 3.5060140843731366e+17,
|
| 7015 |
+
"train_loss": 0.02414526521936059,
|
| 7016 |
+
"train_runtime": 5814.7906,
|
| 7017 |
+
"train_samples_per_second": 27.516,
|
| 7018 |
+
"train_steps_per_second": 1.72
|
| 7019 |
}
|
| 7020 |
],
|
| 7021 |
"logging_steps": 10,
|
|
|
|
| 7030 |
"should_evaluate": false,
|
| 7031 |
"should_log": false,
|
| 7032 |
"should_save": true,
|
| 7033 |
+
"should_training_stop": true
|
| 7034 |
},
|
| 7035 |
"attributes": {}
|
| 7036 |
}
|
| 7037 |
},
|
| 7038 |
+
"total_flos": 3.5060140843731366e+17,
|
| 7039 |
"train_batch_size": 16,
|
| 7040 |
"trial_name": null,
|
| 7041 |
"trial_params": null
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:84923e4733ccef3490ee1b765c64cf4119517e18ac78a7caf63ab1bc0ad41ca9
|
| 3 |
+
size 5240
|