Training in progress, step 1800, checkpoint
Browse files- last-checkpoint/model-00001-of-00004.safetensors +1 -1
- last-checkpoint/model-00002-of-00004.safetensors +1 -1
- last-checkpoint/model-00003-of-00004.safetensors +1 -1
- last-checkpoint/model-00004-of-00004.safetensors +1 -1
- last-checkpoint/optimizer.pt +1 -1
- last-checkpoint/scheduler.pt +1 -1
- last-checkpoint/trainer_state.json +2103 -3
last-checkpoint/model-00001-of-00004.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 4903351912
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:5859bd1404f892d35509df4eb9eb0f7737ea45e8f0d6979cb2b835aa7c512bc7
|
3 |
size 4903351912
|
last-checkpoint/model-00002-of-00004.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 4947570872
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:13e59e818fe01a75735fabb570d56fa405537d22f0e7c991d2ed42d79960ca24
|
3 |
size 4947570872
|
last-checkpoint/model-00003-of-00004.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 4962221464
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:87d6f7b783ba7bb115969fd3f2e645748ecdd1b8fe67ef98d138f57d9282fbf5
|
3 |
size 4962221464
|
last-checkpoint/model-00004-of-00004.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 3670322200
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:1ca0d89ee1ab87e72354decf620a8e0747d38d974878c367d3f4b984d8f027b2
|
3 |
size 3670322200
|
last-checkpoint/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 2216
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:b2160809fdf12c9c94b7b9d3bc115d1ba2a8669618c3440d0d106283ecbf1f93
|
3 |
size 2216
|
last-checkpoint/scheduler.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 1000
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:32d153f5bc2adb9c7a93379e862b4bc1d4648be9d3370182494e08b0e56d554f
|
3 |
size 1000
|
last-checkpoint/trainer_state.json
CHANGED
@@ -1,9 +1,9 @@
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
-
"epoch": 1.
|
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,
|
@@ -10507,6 +10507,2106 @@
|
|
10507 |
"learning_rate": 8.938532094069722e-06,
|
10508 |
"loss": 1.169,
|
10509 |
"step": 1500
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
10510 |
}
|
10511 |
],
|
10512 |
"logging_steps": 1,
|
@@ -10526,7 +12626,7 @@
|
|
10526 |
"attributes": {}
|
10527 |
}
|
10528 |
},
|
10529 |
-
"total_flos":
|
10530 |
"train_batch_size": 8,
|
10531 |
"trial_name": null,
|
10532 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 1.6853932584269664,
|
5 |
"eval_steps": 500,
|
6 |
+
"global_step": 1800,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
10507 |
"learning_rate": 8.938532094069722e-06,
|
10508 |
"loss": 1.169,
|
10509 |
"step": 1500
|
10510 |
+
},
|
10511 |
+
{
|
10512 |
+
"epoch": 1.405430711610487,
|
10513 |
+
"grad_norm": 0.0,
|
10514 |
+
"learning_rate": 8.91274582641178e-06,
|
10515 |
+
"loss": 1.1761,
|
10516 |
+
"step": 1501
|
10517 |
+
},
|
10518 |
+
{
|
10519 |
+
"epoch": 1.4063670411985019,
|
10520 |
+
"grad_norm": 0.0,
|
10521 |
+
"learning_rate": 8.886986139034863e-06,
|
10522 |
+
"loss": 1.1018,
|
10523 |
+
"step": 1502
|
10524 |
+
},
|
10525 |
+
{
|
10526 |
+
"epoch": 1.4073033707865168,
|
10527 |
+
"grad_norm": 0.0,
|
10528 |
+
"learning_rate": 8.86125309369454e-06,
|
10529 |
+
"loss": 1.1077,
|
10530 |
+
"step": 1503
|
10531 |
+
},
|
10532 |
+
{
|
10533 |
+
"epoch": 1.4082397003745317,
|
10534 |
+
"grad_norm": 0.0,
|
10535 |
+
"learning_rate": 8.835546752082527e-06,
|
10536 |
+
"loss": 1.1377,
|
10537 |
+
"step": 1504
|
10538 |
+
},
|
10539 |
+
{
|
10540 |
+
"epoch": 1.4091760299625469,
|
10541 |
+
"grad_norm": 0.0,
|
10542 |
+
"learning_rate": 8.809867175826496e-06,
|
10543 |
+
"loss": 1.1257,
|
10544 |
+
"step": 1505
|
10545 |
+
},
|
10546 |
+
{
|
10547 |
+
"epoch": 1.4101123595505618,
|
10548 |
+
"grad_norm": 0.0,
|
10549 |
+
"learning_rate": 8.784214426489964e-06,
|
10550 |
+
"loss": 1.1451,
|
10551 |
+
"step": 1506
|
10552 |
+
},
|
10553 |
+
{
|
10554 |
+
"epoch": 1.4110486891385767,
|
10555 |
+
"grad_norm": 0.0,
|
10556 |
+
"learning_rate": 8.758588565572137e-06,
|
10557 |
+
"loss": 1.1264,
|
10558 |
+
"step": 1507
|
10559 |
+
},
|
10560 |
+
{
|
10561 |
+
"epoch": 1.4119850187265919,
|
10562 |
+
"grad_norm": 0.0,
|
10563 |
+
"learning_rate": 8.732989654507758e-06,
|
10564 |
+
"loss": 1.1549,
|
10565 |
+
"step": 1508
|
10566 |
+
},
|
10567 |
+
{
|
10568 |
+
"epoch": 1.4129213483146068,
|
10569 |
+
"grad_norm": 0.0,
|
10570 |
+
"learning_rate": 8.707417754666944e-06,
|
10571 |
+
"loss": 1.1708,
|
10572 |
+
"step": 1509
|
10573 |
+
},
|
10574 |
+
{
|
10575 |
+
"epoch": 1.4138576779026217,
|
10576 |
+
"grad_norm": 0.0,
|
10577 |
+
"learning_rate": 8.681872927355083e-06,
|
10578 |
+
"loss": 1.1633,
|
10579 |
+
"step": 1510
|
10580 |
+
},
|
10581 |
+
{
|
10582 |
+
"epoch": 1.4147940074906367,
|
10583 |
+
"grad_norm": 0.0,
|
10584 |
+
"learning_rate": 8.656355233812656e-06,
|
10585 |
+
"loss": 1.1428,
|
10586 |
+
"step": 1511
|
10587 |
+
},
|
10588 |
+
{
|
10589 |
+
"epoch": 1.4157303370786516,
|
10590 |
+
"grad_norm": 0.0,
|
10591 |
+
"learning_rate": 8.630864735215068e-06,
|
10592 |
+
"loss": 1.1585,
|
10593 |
+
"step": 1512
|
10594 |
+
},
|
10595 |
+
{
|
10596 |
+
"epoch": 1.4166666666666667,
|
10597 |
+
"grad_norm": 0.0,
|
10598 |
+
"learning_rate": 8.60540149267256e-06,
|
10599 |
+
"loss": 1.1019,
|
10600 |
+
"step": 1513
|
10601 |
+
},
|
10602 |
+
{
|
10603 |
+
"epoch": 1.4176029962546817,
|
10604 |
+
"grad_norm": 0.0,
|
10605 |
+
"learning_rate": 8.579965567230013e-06,
|
10606 |
+
"loss": 1.1107,
|
10607 |
+
"step": 1514
|
10608 |
+
},
|
10609 |
+
{
|
10610 |
+
"epoch": 1.4185393258426966,
|
10611 |
+
"grad_norm": 0.0,
|
10612 |
+
"learning_rate": 8.554557019866812e-06,
|
10613 |
+
"loss": 1.088,
|
10614 |
+
"step": 1515
|
10615 |
+
},
|
10616 |
+
{
|
10617 |
+
"epoch": 1.4194756554307117,
|
10618 |
+
"grad_norm": 0.0,
|
10619 |
+
"learning_rate": 8.529175911496728e-06,
|
10620 |
+
"loss": 1.1172,
|
10621 |
+
"step": 1516
|
10622 |
+
},
|
10623 |
+
{
|
10624 |
+
"epoch": 1.4204119850187267,
|
10625 |
+
"grad_norm": 0.0,
|
10626 |
+
"learning_rate": 8.503822302967732e-06,
|
10627 |
+
"loss": 1.1868,
|
10628 |
+
"step": 1517
|
10629 |
+
},
|
10630 |
+
{
|
10631 |
+
"epoch": 1.4213483146067416,
|
10632 |
+
"grad_norm": 0.0,
|
10633 |
+
"learning_rate": 8.478496255061872e-06,
|
10634 |
+
"loss": 1.14,
|
10635 |
+
"step": 1518
|
10636 |
+
},
|
10637 |
+
{
|
10638 |
+
"epoch": 1.4222846441947565,
|
10639 |
+
"grad_norm": 0.0,
|
10640 |
+
"learning_rate": 8.453197828495132e-06,
|
10641 |
+
"loss": 1.1309,
|
10642 |
+
"step": 1519
|
10643 |
+
},
|
10644 |
+
{
|
10645 |
+
"epoch": 1.4232209737827715,
|
10646 |
+
"grad_norm": 0.0,
|
10647 |
+
"learning_rate": 8.427927083917262e-06,
|
10648 |
+
"loss": 1.1472,
|
10649 |
+
"step": 1520
|
10650 |
+
},
|
10651 |
+
{
|
10652 |
+
"epoch": 1.4241573033707864,
|
10653 |
+
"grad_norm": 0.0,
|
10654 |
+
"learning_rate": 8.40268408191167e-06,
|
10655 |
+
"loss": 1.205,
|
10656 |
+
"step": 1521
|
10657 |
+
},
|
10658 |
+
{
|
10659 |
+
"epoch": 1.4250936329588015,
|
10660 |
+
"grad_norm": 0.0,
|
10661 |
+
"learning_rate": 8.377468882995223e-06,
|
10662 |
+
"loss": 1.1165,
|
10663 |
+
"step": 1522
|
10664 |
+
},
|
10665 |
+
{
|
10666 |
+
"epoch": 1.4260299625468165,
|
10667 |
+
"grad_norm": 0.0,
|
10668 |
+
"learning_rate": 8.35228154761817e-06,
|
10669 |
+
"loss": 1.1351,
|
10670 |
+
"step": 1523
|
10671 |
+
},
|
10672 |
+
{
|
10673 |
+
"epoch": 1.4269662921348314,
|
10674 |
+
"grad_norm": 0.0,
|
10675 |
+
"learning_rate": 8.327122136163928e-06,
|
10676 |
+
"loss": 1.1118,
|
10677 |
+
"step": 1524
|
10678 |
+
},
|
10679 |
+
{
|
10680 |
+
"epoch": 1.4279026217228465,
|
10681 |
+
"grad_norm": 0.0,
|
10682 |
+
"learning_rate": 8.301990708948996e-06,
|
10683 |
+
"loss": 1.1238,
|
10684 |
+
"step": 1525
|
10685 |
+
},
|
10686 |
+
{
|
10687 |
+
"epoch": 1.4288389513108615,
|
10688 |
+
"grad_norm": 0.0,
|
10689 |
+
"learning_rate": 8.276887326222769e-06,
|
10690 |
+
"loss": 1.1302,
|
10691 |
+
"step": 1526
|
10692 |
+
},
|
10693 |
+
{
|
10694 |
+
"epoch": 1.4297752808988764,
|
10695 |
+
"grad_norm": 0.0,
|
10696 |
+
"learning_rate": 8.251812048167402e-06,
|
10697 |
+
"loss": 1.1591,
|
10698 |
+
"step": 1527
|
10699 |
+
},
|
10700 |
+
{
|
10701 |
+
"epoch": 1.4307116104868913,
|
10702 |
+
"grad_norm": 0.0,
|
10703 |
+
"learning_rate": 8.2267649348977e-06,
|
10704 |
+
"loss": 1.1453,
|
10705 |
+
"step": 1528
|
10706 |
+
},
|
10707 |
+
{
|
10708 |
+
"epoch": 1.4316479400749063,
|
10709 |
+
"grad_norm": 0.0,
|
10710 |
+
"learning_rate": 8.201746046460922e-06,
|
10711 |
+
"loss": 1.1393,
|
10712 |
+
"step": 1529
|
10713 |
+
},
|
10714 |
+
{
|
10715 |
+
"epoch": 1.4325842696629214,
|
10716 |
+
"grad_norm": 0.0,
|
10717 |
+
"learning_rate": 8.176755442836661e-06,
|
10718 |
+
"loss": 1.1307,
|
10719 |
+
"step": 1530
|
10720 |
+
},
|
10721 |
+
{
|
10722 |
+
"epoch": 1.4335205992509363,
|
10723 |
+
"grad_norm": 0.0,
|
10724 |
+
"learning_rate": 8.151793183936724e-06,
|
10725 |
+
"loss": 1.1203,
|
10726 |
+
"step": 1531
|
10727 |
+
},
|
10728 |
+
{
|
10729 |
+
"epoch": 1.4344569288389513,
|
10730 |
+
"grad_norm": 0.0,
|
10731 |
+
"learning_rate": 8.12685932960494e-06,
|
10732 |
+
"loss": 1.1445,
|
10733 |
+
"step": 1532
|
10734 |
+
},
|
10735 |
+
{
|
10736 |
+
"epoch": 1.4353932584269664,
|
10737 |
+
"grad_norm": 0.0,
|
10738 |
+
"learning_rate": 8.101953939617053e-06,
|
10739 |
+
"loss": 1.1466,
|
10740 |
+
"step": 1533
|
10741 |
+
},
|
10742 |
+
{
|
10743 |
+
"epoch": 1.4363295880149813,
|
10744 |
+
"grad_norm": 0.0,
|
10745 |
+
"learning_rate": 8.077077073680577e-06,
|
10746 |
+
"loss": 1.0897,
|
10747 |
+
"step": 1534
|
10748 |
+
},
|
10749 |
+
{
|
10750 |
+
"epoch": 1.4372659176029963,
|
10751 |
+
"grad_norm": 0.0,
|
10752 |
+
"learning_rate": 8.052228791434618e-06,
|
10753 |
+
"loss": 1.1645,
|
10754 |
+
"step": 1535
|
10755 |
+
},
|
10756 |
+
{
|
10757 |
+
"epoch": 1.4382022471910112,
|
10758 |
+
"grad_norm": 0.0,
|
10759 |
+
"learning_rate": 8.027409152449783e-06,
|
10760 |
+
"loss": 1.1831,
|
10761 |
+
"step": 1536
|
10762 |
+
},
|
10763 |
+
{
|
10764 |
+
"epoch": 1.4391385767790261,
|
10765 |
+
"grad_norm": 0.0,
|
10766 |
+
"learning_rate": 8.002618216227994e-06,
|
10767 |
+
"loss": 1.0751,
|
10768 |
+
"step": 1537
|
10769 |
+
},
|
10770 |
+
{
|
10771 |
+
"epoch": 1.4400749063670413,
|
10772 |
+
"grad_norm": 0.0,
|
10773 |
+
"learning_rate": 7.977856042202354e-06,
|
10774 |
+
"loss": 1.0924,
|
10775 |
+
"step": 1538
|
10776 |
+
},
|
10777 |
+
{
|
10778 |
+
"epoch": 1.4410112359550562,
|
10779 |
+
"grad_norm": 0.0,
|
10780 |
+
"learning_rate": 7.953122689737045e-06,
|
10781 |
+
"loss": 1.1164,
|
10782 |
+
"step": 1539
|
10783 |
+
},
|
10784 |
+
{
|
10785 |
+
"epoch": 1.4419475655430711,
|
10786 |
+
"grad_norm": 0.0,
|
10787 |
+
"learning_rate": 7.928418218127121e-06,
|
10788 |
+
"loss": 1.1018,
|
10789 |
+
"step": 1540
|
10790 |
+
},
|
10791 |
+
{
|
10792 |
+
"epoch": 1.4428838951310863,
|
10793 |
+
"grad_norm": 0.0,
|
10794 |
+
"learning_rate": 7.903742686598402e-06,
|
10795 |
+
"loss": 1.1509,
|
10796 |
+
"step": 1541
|
10797 |
+
},
|
10798 |
+
{
|
10799 |
+
"epoch": 1.4438202247191012,
|
10800 |
+
"grad_norm": 0.0,
|
10801 |
+
"learning_rate": 7.87909615430735e-06,
|
10802 |
+
"loss": 1.1266,
|
10803 |
+
"step": 1542
|
10804 |
+
},
|
10805 |
+
{
|
10806 |
+
"epoch": 1.4447565543071161,
|
10807 |
+
"grad_norm": 0.0,
|
10808 |
+
"learning_rate": 7.854478680340875e-06,
|
10809 |
+
"loss": 1.154,
|
10810 |
+
"step": 1543
|
10811 |
+
},
|
10812 |
+
{
|
10813 |
+
"epoch": 1.445692883895131,
|
10814 |
+
"grad_norm": 0.0,
|
10815 |
+
"learning_rate": 7.829890323716247e-06,
|
10816 |
+
"loss": 1.1338,
|
10817 |
+
"step": 1544
|
10818 |
+
},
|
10819 |
+
{
|
10820 |
+
"epoch": 1.446629213483146,
|
10821 |
+
"grad_norm": 0.0,
|
10822 |
+
"learning_rate": 7.805331143380926e-06,
|
10823 |
+
"loss": 1.1359,
|
10824 |
+
"step": 1545
|
10825 |
+
},
|
10826 |
+
{
|
10827 |
+
"epoch": 1.447565543071161,
|
10828 |
+
"grad_norm": 0.0,
|
10829 |
+
"learning_rate": 7.780801198212418e-06,
|
10830 |
+
"loss": 1.1192,
|
10831 |
+
"step": 1546
|
10832 |
+
},
|
10833 |
+
{
|
10834 |
+
"epoch": 1.448501872659176,
|
10835 |
+
"grad_norm": 0.0,
|
10836 |
+
"learning_rate": 7.756300547018137e-06,
|
10837 |
+
"loss": 1.1369,
|
10838 |
+
"step": 1547
|
10839 |
+
},
|
10840 |
+
{
|
10841 |
+
"epoch": 1.449438202247191,
|
10842 |
+
"grad_norm": 0.0,
|
10843 |
+
"learning_rate": 7.73182924853529e-06,
|
10844 |
+
"loss": 1.1355,
|
10845 |
+
"step": 1548
|
10846 |
+
},
|
10847 |
+
{
|
10848 |
+
"epoch": 1.450374531835206,
|
10849 |
+
"grad_norm": 0.0,
|
10850 |
+
"learning_rate": 7.707387361430696e-06,
|
10851 |
+
"loss": 1.1717,
|
10852 |
+
"step": 1549
|
10853 |
+
},
|
10854 |
+
{
|
10855 |
+
"epoch": 1.451310861423221,
|
10856 |
+
"grad_norm": 0.0,
|
10857 |
+
"learning_rate": 7.682974944300659e-06,
|
10858 |
+
"loss": 1.1523,
|
10859 |
+
"step": 1550
|
10860 |
+
},
|
10861 |
+
{
|
10862 |
+
"epoch": 1.452247191011236,
|
10863 |
+
"grad_norm": 0.0,
|
10864 |
+
"learning_rate": 7.658592055670857e-06,
|
10865 |
+
"loss": 1.1508,
|
10866 |
+
"step": 1551
|
10867 |
+
},
|
10868 |
+
{
|
10869 |
+
"epoch": 1.453183520599251,
|
10870 |
+
"grad_norm": 0.0,
|
10871 |
+
"learning_rate": 7.634238753996159e-06,
|
10872 |
+
"loss": 1.1318,
|
10873 |
+
"step": 1552
|
10874 |
+
},
|
10875 |
+
{
|
10876 |
+
"epoch": 1.4541198501872659,
|
10877 |
+
"grad_norm": 0.0,
|
10878 |
+
"learning_rate": 7.609915097660497e-06,
|
10879 |
+
"loss": 1.1081,
|
10880 |
+
"step": 1553
|
10881 |
+
},
|
10882 |
+
{
|
10883 |
+
"epoch": 1.4550561797752808,
|
10884 |
+
"grad_norm": 0.0,
|
10885 |
+
"learning_rate": 7.5856211449767555e-06,
|
10886 |
+
"loss": 1.142,
|
10887 |
+
"step": 1554
|
10888 |
+
},
|
10889 |
+
{
|
10890 |
+
"epoch": 1.455992509363296,
|
10891 |
+
"grad_norm": 0.0,
|
10892 |
+
"learning_rate": 7.561356954186582e-06,
|
10893 |
+
"loss": 1.1294,
|
10894 |
+
"step": 1555
|
10895 |
+
},
|
10896 |
+
{
|
10897 |
+
"epoch": 1.4569288389513109,
|
10898 |
+
"grad_norm": 0.0,
|
10899 |
+
"learning_rate": 7.537122583460294e-06,
|
10900 |
+
"loss": 1.1681,
|
10901 |
+
"step": 1556
|
10902 |
+
},
|
10903 |
+
{
|
10904 |
+
"epoch": 1.4578651685393258,
|
10905 |
+
"grad_norm": 0.0,
|
10906 |
+
"learning_rate": 7.512918090896719e-06,
|
10907 |
+
"loss": 1.126,
|
10908 |
+
"step": 1557
|
10909 |
+
},
|
10910 |
+
{
|
10911 |
+
"epoch": 1.458801498127341,
|
10912 |
+
"grad_norm": 0.0,
|
10913 |
+
"learning_rate": 7.4887435345230375e-06,
|
10914 |
+
"loss": 1.188,
|
10915 |
+
"step": 1558
|
10916 |
+
},
|
10917 |
+
{
|
10918 |
+
"epoch": 1.4597378277153559,
|
10919 |
+
"grad_norm": 0.0,
|
10920 |
+
"learning_rate": 7.464598972294668e-06,
|
10921 |
+
"loss": 1.1648,
|
10922 |
+
"step": 1559
|
10923 |
+
},
|
10924 |
+
{
|
10925 |
+
"epoch": 1.4606741573033708,
|
10926 |
+
"grad_norm": 0.0,
|
10927 |
+
"learning_rate": 7.440484462095137e-06,
|
10928 |
+
"loss": 1.0994,
|
10929 |
+
"step": 1560
|
10930 |
+
},
|
10931 |
+
{
|
10932 |
+
"epoch": 1.4616104868913857,
|
10933 |
+
"grad_norm": 0.0,
|
10934 |
+
"learning_rate": 7.4164000617359115e-06,
|
10935 |
+
"loss": 1.1331,
|
10936 |
+
"step": 1561
|
10937 |
+
},
|
10938 |
+
{
|
10939 |
+
"epoch": 1.4625468164794007,
|
10940 |
+
"grad_norm": 0.0,
|
10941 |
+
"learning_rate": 7.3923458289562665e-06,
|
10942 |
+
"loss": 1.1525,
|
10943 |
+
"step": 1562
|
10944 |
+
},
|
10945 |
+
{
|
10946 |
+
"epoch": 1.4634831460674158,
|
10947 |
+
"grad_norm": 0.0,
|
10948 |
+
"learning_rate": 7.368321821423179e-06,
|
10949 |
+
"loss": 1.1373,
|
10950 |
+
"step": 1563
|
10951 |
+
},
|
10952 |
+
{
|
10953 |
+
"epoch": 1.4644194756554307,
|
10954 |
+
"grad_norm": 0.0,
|
10955 |
+
"learning_rate": 7.344328096731144e-06,
|
10956 |
+
"loss": 1.1692,
|
10957 |
+
"step": 1564
|
10958 |
+
},
|
10959 |
+
{
|
10960 |
+
"epoch": 1.4653558052434457,
|
10961 |
+
"grad_norm": 0.0,
|
10962 |
+
"learning_rate": 7.320364712402055e-06,
|
10963 |
+
"loss": 1.1382,
|
10964 |
+
"step": 1565
|
10965 |
+
},
|
10966 |
+
{
|
10967 |
+
"epoch": 1.4662921348314606,
|
10968 |
+
"grad_norm": 0.0,
|
10969 |
+
"learning_rate": 7.296431725885085e-06,
|
10970 |
+
"loss": 1.1261,
|
10971 |
+
"step": 1566
|
10972 |
+
},
|
10973 |
+
{
|
10974 |
+
"epoch": 1.4672284644194757,
|
10975 |
+
"grad_norm": 0.0,
|
10976 |
+
"learning_rate": 7.272529194556528e-06,
|
10977 |
+
"loss": 1.1731,
|
10978 |
+
"step": 1567
|
10979 |
+
},
|
10980 |
+
{
|
10981 |
+
"epoch": 1.4681647940074907,
|
10982 |
+
"grad_norm": 0.0,
|
10983 |
+
"learning_rate": 7.248657175719649e-06,
|
10984 |
+
"loss": 1.185,
|
10985 |
+
"step": 1568
|
10986 |
+
},
|
10987 |
+
{
|
10988 |
+
"epoch": 1.4691011235955056,
|
10989 |
+
"grad_norm": 0.0,
|
10990 |
+
"learning_rate": 7.2248157266045904e-06,
|
10991 |
+
"loss": 1.1735,
|
10992 |
+
"step": 1569
|
10993 |
+
},
|
10994 |
+
{
|
10995 |
+
"epoch": 1.4700374531835205,
|
10996 |
+
"grad_norm": 0.0,
|
10997 |
+
"learning_rate": 7.2010049043681825e-06,
|
10998 |
+
"loss": 1.1518,
|
10999 |
+
"step": 1570
|
11000 |
+
},
|
11001 |
+
{
|
11002 |
+
"epoch": 1.4709737827715355,
|
11003 |
+
"grad_norm": 0.0,
|
11004 |
+
"learning_rate": 7.17722476609384e-06,
|
11005 |
+
"loss": 1.1282,
|
11006 |
+
"step": 1571
|
11007 |
+
},
|
11008 |
+
{
|
11009 |
+
"epoch": 1.4719101123595506,
|
11010 |
+
"grad_norm": 0.0,
|
11011 |
+
"learning_rate": 7.153475368791427e-06,
|
11012 |
+
"loss": 1.1125,
|
11013 |
+
"step": 1572
|
11014 |
+
},
|
11015 |
+
{
|
11016 |
+
"epoch": 1.4728464419475655,
|
11017 |
+
"grad_norm": 0.0,
|
11018 |
+
"learning_rate": 7.129756769397096e-06,
|
11019 |
+
"loss": 1.1646,
|
11020 |
+
"step": 1573
|
11021 |
+
},
|
11022 |
+
{
|
11023 |
+
"epoch": 1.4737827715355805,
|
11024 |
+
"grad_norm": 0.0,
|
11025 |
+
"learning_rate": 7.106069024773166e-06,
|
11026 |
+
"loss": 1.1523,
|
11027 |
+
"step": 1574
|
11028 |
+
},
|
11029 |
+
{
|
11030 |
+
"epoch": 1.4747191011235956,
|
11031 |
+
"grad_norm": 0.0,
|
11032 |
+
"learning_rate": 7.082412191708001e-06,
|
11033 |
+
"loss": 1.1439,
|
11034 |
+
"step": 1575
|
11035 |
+
},
|
11036 |
+
{
|
11037 |
+
"epoch": 1.4756554307116105,
|
11038 |
+
"grad_norm": 0.0,
|
11039 |
+
"learning_rate": 7.058786326915839e-06,
|
11040 |
+
"loss": 1.1748,
|
11041 |
+
"step": 1576
|
11042 |
+
},
|
11043 |
+
{
|
11044 |
+
"epoch": 1.4765917602996255,
|
11045 |
+
"grad_norm": 0.0,
|
11046 |
+
"learning_rate": 7.035191487036685e-06,
|
11047 |
+
"loss": 1.13,
|
11048 |
+
"step": 1577
|
11049 |
+
},
|
11050 |
+
{
|
11051 |
+
"epoch": 1.4775280898876404,
|
11052 |
+
"grad_norm": 0.0,
|
11053 |
+
"learning_rate": 7.0116277286361765e-06,
|
11054 |
+
"loss": 1.118,
|
11055 |
+
"step": 1578
|
11056 |
+
},
|
11057 |
+
{
|
11058 |
+
"epoch": 1.4784644194756553,
|
11059 |
+
"grad_norm": 0.0,
|
11060 |
+
"learning_rate": 6.9880951082054196e-06,
|
11061 |
+
"loss": 1.1559,
|
11062 |
+
"step": 1579
|
11063 |
+
},
|
11064 |
+
{
|
11065 |
+
"epoch": 1.4794007490636705,
|
11066 |
+
"grad_norm": 0.0,
|
11067 |
+
"learning_rate": 6.964593682160868e-06,
|
11068 |
+
"loss": 1.1149,
|
11069 |
+
"step": 1580
|
11070 |
+
},
|
11071 |
+
{
|
11072 |
+
"epoch": 1.4803370786516854,
|
11073 |
+
"grad_norm": 0.0,
|
11074 |
+
"learning_rate": 6.941123506844212e-06,
|
11075 |
+
"loss": 1.1298,
|
11076 |
+
"step": 1581
|
11077 |
+
},
|
11078 |
+
{
|
11079 |
+
"epoch": 1.4812734082397003,
|
11080 |
+
"grad_norm": 0.0,
|
11081 |
+
"learning_rate": 6.917684638522206e-06,
|
11082 |
+
"loss": 1.1405,
|
11083 |
+
"step": 1582
|
11084 |
+
},
|
11085 |
+
{
|
11086 |
+
"epoch": 1.4822097378277155,
|
11087 |
+
"grad_norm": 0.0,
|
11088 |
+
"learning_rate": 6.894277133386549e-06,
|
11089 |
+
"loss": 1.1084,
|
11090 |
+
"step": 1583
|
11091 |
+
},
|
11092 |
+
{
|
11093 |
+
"epoch": 1.4831460674157304,
|
11094 |
+
"grad_norm": 0.0,
|
11095 |
+
"learning_rate": 6.870901047553769e-06,
|
11096 |
+
"loss": 1.1001,
|
11097 |
+
"step": 1584
|
11098 |
+
},
|
11099 |
+
{
|
11100 |
+
"epoch": 1.4840823970037453,
|
11101 |
+
"grad_norm": 0.0,
|
11102 |
+
"learning_rate": 6.847556437065046e-06,
|
11103 |
+
"loss": 1.1279,
|
11104 |
+
"step": 1585
|
11105 |
+
},
|
11106 |
+
{
|
11107 |
+
"epoch": 1.4850187265917603,
|
11108 |
+
"grad_norm": 0.0,
|
11109 |
+
"learning_rate": 6.824243357886111e-06,
|
11110 |
+
"loss": 1.1498,
|
11111 |
+
"step": 1586
|
11112 |
+
},
|
11113 |
+
{
|
11114 |
+
"epoch": 1.4859550561797752,
|
11115 |
+
"grad_norm": 0.0,
|
11116 |
+
"learning_rate": 6.800961865907109e-06,
|
11117 |
+
"loss": 1.1583,
|
11118 |
+
"step": 1587
|
11119 |
+
},
|
11120 |
+
{
|
11121 |
+
"epoch": 1.4868913857677903,
|
11122 |
+
"grad_norm": 0.0,
|
11123 |
+
"learning_rate": 6.777712016942464e-06,
|
11124 |
+
"loss": 1.157,
|
11125 |
+
"step": 1588
|
11126 |
+
},
|
11127 |
+
{
|
11128 |
+
"epoch": 1.4878277153558053,
|
11129 |
+
"grad_norm": 0.0,
|
11130 |
+
"learning_rate": 6.754493866730714e-06,
|
11131 |
+
"loss": 1.1747,
|
11132 |
+
"step": 1589
|
11133 |
+
},
|
11134 |
+
{
|
11135 |
+
"epoch": 1.4887640449438202,
|
11136 |
+
"grad_norm": 0.0,
|
11137 |
+
"learning_rate": 6.731307470934434e-06,
|
11138 |
+
"loss": 1.1574,
|
11139 |
+
"step": 1590
|
11140 |
+
},
|
11141 |
+
{
|
11142 |
+
"epoch": 1.4897003745318351,
|
11143 |
+
"grad_norm": 0.0,
|
11144 |
+
"learning_rate": 6.708152885140049e-06,
|
11145 |
+
"loss": 1.1248,
|
11146 |
+
"step": 1591
|
11147 |
+
},
|
11148 |
+
{
|
11149 |
+
"epoch": 1.4906367041198503,
|
11150 |
+
"grad_norm": 0.0,
|
11151 |
+
"learning_rate": 6.685030164857731e-06,
|
11152 |
+
"loss": 1.165,
|
11153 |
+
"step": 1592
|
11154 |
+
},
|
11155 |
+
{
|
11156 |
+
"epoch": 1.4915730337078652,
|
11157 |
+
"grad_norm": 0.0,
|
11158 |
+
"learning_rate": 6.661939365521266e-06,
|
11159 |
+
"loss": 1.1541,
|
11160 |
+
"step": 1593
|
11161 |
+
},
|
11162 |
+
{
|
11163 |
+
"epoch": 1.4925093632958801,
|
11164 |
+
"grad_norm": 0.0,
|
11165 |
+
"learning_rate": 6.6388805424879045e-06,
|
11166 |
+
"loss": 1.1691,
|
11167 |
+
"step": 1594
|
11168 |
+
},
|
11169 |
+
{
|
11170 |
+
"epoch": 1.493445692883895,
|
11171 |
+
"grad_norm": 0.0,
|
11172 |
+
"learning_rate": 6.615853751038233e-06,
|
11173 |
+
"loss": 1.1697,
|
11174 |
+
"step": 1595
|
11175 |
+
},
|
11176 |
+
{
|
11177 |
+
"epoch": 1.49438202247191,
|
11178 |
+
"grad_norm": 0.0,
|
11179 |
+
"learning_rate": 6.592859046376066e-06,
|
11180 |
+
"loss": 1.1568,
|
11181 |
+
"step": 1596
|
11182 |
+
},
|
11183 |
+
{
|
11184 |
+
"epoch": 1.4953183520599251,
|
11185 |
+
"grad_norm": 0.0,
|
11186 |
+
"learning_rate": 6.569896483628278e-06,
|
11187 |
+
"loss": 1.1656,
|
11188 |
+
"step": 1597
|
11189 |
+
},
|
11190 |
+
{
|
11191 |
+
"epoch": 1.49625468164794,
|
11192 |
+
"grad_norm": 0.0,
|
11193 |
+
"learning_rate": 6.54696611784468e-06,
|
11194 |
+
"loss": 1.151,
|
11195 |
+
"step": 1598
|
11196 |
+
},
|
11197 |
+
{
|
11198 |
+
"epoch": 1.497191011235955,
|
11199 |
+
"grad_norm": 0.0,
|
11200 |
+
"learning_rate": 6.524068003997917e-06,
|
11201 |
+
"loss": 1.1307,
|
11202 |
+
"step": 1599
|
11203 |
+
},
|
11204 |
+
{
|
11205 |
+
"epoch": 1.4981273408239701,
|
11206 |
+
"grad_norm": 0.0,
|
11207 |
+
"learning_rate": 6.501202196983301e-06,
|
11208 |
+
"loss": 1.1294,
|
11209 |
+
"step": 1600
|
11210 |
+
},
|
11211 |
+
{
|
11212 |
+
"epoch": 1.499063670411985,
|
11213 |
+
"grad_norm": 0.0,
|
11214 |
+
"learning_rate": 6.478368751618702e-06,
|
11215 |
+
"loss": 1.1632,
|
11216 |
+
"step": 1601
|
11217 |
+
},
|
11218 |
+
{
|
11219 |
+
"epoch": 1.5,
|
11220 |
+
"grad_norm": 0.0,
|
11221 |
+
"learning_rate": 6.455567722644394e-06,
|
11222 |
+
"loss": 1.1559,
|
11223 |
+
"step": 1602
|
11224 |
+
},
|
11225 |
+
{
|
11226 |
+
"epoch": 1.500936329588015,
|
11227 |
+
"grad_norm": 0.0,
|
11228 |
+
"learning_rate": 6.432799164722934e-06,
|
11229 |
+
"loss": 1.1024,
|
11230 |
+
"step": 1603
|
11231 |
+
},
|
11232 |
+
{
|
11233 |
+
"epoch": 1.5018726591760299,
|
11234 |
+
"grad_norm": 0.0,
|
11235 |
+
"learning_rate": 6.410063132439055e-06,
|
11236 |
+
"loss": 1.1236,
|
11237 |
+
"step": 1604
|
11238 |
+
},
|
11239 |
+
{
|
11240 |
+
"epoch": 1.5028089887640448,
|
11241 |
+
"grad_norm": 0.0,
|
11242 |
+
"learning_rate": 6.3873596802994985e-06,
|
11243 |
+
"loss": 1.1334,
|
11244 |
+
"step": 1605
|
11245 |
+
},
|
11246 |
+
{
|
11247 |
+
"epoch": 1.50374531835206,
|
11248 |
+
"grad_norm": 0.0,
|
11249 |
+
"learning_rate": 6.364688862732891e-06,
|
11250 |
+
"loss": 1.1242,
|
11251 |
+
"step": 1606
|
11252 |
+
},
|
11253 |
+
{
|
11254 |
+
"epoch": 1.5046816479400749,
|
11255 |
+
"grad_norm": 0.0,
|
11256 |
+
"learning_rate": 6.342050734089648e-06,
|
11257 |
+
"loss": 1.1469,
|
11258 |
+
"step": 1607
|
11259 |
+
},
|
11260 |
+
{
|
11261 |
+
"epoch": 1.50561797752809,
|
11262 |
+
"grad_norm": 0.0,
|
11263 |
+
"learning_rate": 6.3194453486417975e-06,
|
11264 |
+
"loss": 1.1377,
|
11265 |
+
"step": 1608
|
11266 |
+
},
|
11267 |
+
{
|
11268 |
+
"epoch": 1.506554307116105,
|
11269 |
+
"grad_norm": 0.0,
|
11270 |
+
"learning_rate": 6.296872760582864e-06,
|
11271 |
+
"loss": 1.1099,
|
11272 |
+
"step": 1609
|
11273 |
+
},
|
11274 |
+
{
|
11275 |
+
"epoch": 1.5074906367041199,
|
11276 |
+
"grad_norm": 0.0,
|
11277 |
+
"learning_rate": 6.274333024027763e-06,
|
11278 |
+
"loss": 1.1475,
|
11279 |
+
"step": 1610
|
11280 |
+
},
|
11281 |
+
{
|
11282 |
+
"epoch": 1.5084269662921348,
|
11283 |
+
"grad_norm": 0.0,
|
11284 |
+
"learning_rate": 6.2518261930126535e-06,
|
11285 |
+
"loss": 1.0754,
|
11286 |
+
"step": 1611
|
11287 |
+
},
|
11288 |
+
{
|
11289 |
+
"epoch": 1.5093632958801497,
|
11290 |
+
"grad_norm": 0.0,
|
11291 |
+
"learning_rate": 6.2293523214947835e-06,
|
11292 |
+
"loss": 1.1681,
|
11293 |
+
"step": 1612
|
11294 |
+
},
|
11295 |
+
{
|
11296 |
+
"epoch": 1.5102996254681647,
|
11297 |
+
"grad_norm": 0.0,
|
11298 |
+
"learning_rate": 6.206911463352414e-06,
|
11299 |
+
"loss": 1.1281,
|
11300 |
+
"step": 1613
|
11301 |
+
},
|
11302 |
+
{
|
11303 |
+
"epoch": 1.5112359550561798,
|
11304 |
+
"grad_norm": 0.0,
|
11305 |
+
"learning_rate": 6.184503672384641e-06,
|
11306 |
+
"loss": 1.0952,
|
11307 |
+
"step": 1614
|
11308 |
+
},
|
11309 |
+
{
|
11310 |
+
"epoch": 1.5121722846441947,
|
11311 |
+
"grad_norm": 0.0,
|
11312 |
+
"learning_rate": 6.1621290023112836e-06,
|
11313 |
+
"loss": 1.1379,
|
11314 |
+
"step": 1615
|
11315 |
+
},
|
11316 |
+
{
|
11317 |
+
"epoch": 1.5131086142322099,
|
11318 |
+
"grad_norm": 0.0,
|
11319 |
+
"learning_rate": 6.139787506772779e-06,
|
11320 |
+
"loss": 1.1544,
|
11321 |
+
"step": 1616
|
11322 |
+
},
|
11323 |
+
{
|
11324 |
+
"epoch": 1.5140449438202248,
|
11325 |
+
"grad_norm": 0.0,
|
11326 |
+
"learning_rate": 6.117479239330015e-06,
|
11327 |
+
"loss": 1.1538,
|
11328 |
+
"step": 1617
|
11329 |
+
},
|
11330 |
+
{
|
11331 |
+
"epoch": 1.5149812734082397,
|
11332 |
+
"grad_norm": 0.0,
|
11333 |
+
"learning_rate": 6.095204253464218e-06,
|
11334 |
+
"loss": 1.1352,
|
11335 |
+
"step": 1618
|
11336 |
+
},
|
11337 |
+
{
|
11338 |
+
"epoch": 1.5159176029962547,
|
11339 |
+
"grad_norm": 0.0,
|
11340 |
+
"learning_rate": 6.072962602576842e-06,
|
11341 |
+
"loss": 1.1392,
|
11342 |
+
"step": 1619
|
11343 |
+
},
|
11344 |
+
{
|
11345 |
+
"epoch": 1.5168539325842696,
|
11346 |
+
"grad_norm": 0.0,
|
11347 |
+
"learning_rate": 6.050754339989404e-06,
|
11348 |
+
"loss": 1.1455,
|
11349 |
+
"step": 1620
|
11350 |
+
},
|
11351 |
+
{
|
11352 |
+
"epoch": 1.5177902621722845,
|
11353 |
+
"grad_norm": 0.0,
|
11354 |
+
"learning_rate": 6.028579518943403e-06,
|
11355 |
+
"loss": 1.1365,
|
11356 |
+
"step": 1621
|
11357 |
+
},
|
11358 |
+
{
|
11359 |
+
"epoch": 1.5187265917602997,
|
11360 |
+
"grad_norm": 0.0,
|
11361 |
+
"learning_rate": 6.006438192600132e-06,
|
11362 |
+
"loss": 1.0375,
|
11363 |
+
"step": 1622
|
11364 |
+
},
|
11365 |
+
{
|
11366 |
+
"epoch": 1.5196629213483146,
|
11367 |
+
"grad_norm": 0.0,
|
11368 |
+
"learning_rate": 5.984330414040622e-06,
|
11369 |
+
"loss": 1.1491,
|
11370 |
+
"step": 1623
|
11371 |
+
},
|
11372 |
+
{
|
11373 |
+
"epoch": 1.5205992509363297,
|
11374 |
+
"grad_norm": 0.0,
|
11375 |
+
"learning_rate": 5.962256236265442e-06,
|
11376 |
+
"loss": 1.1085,
|
11377 |
+
"step": 1624
|
11378 |
+
},
|
11379 |
+
{
|
11380 |
+
"epoch": 1.5215355805243447,
|
11381 |
+
"grad_norm": 0.0,
|
11382 |
+
"learning_rate": 5.940215712194639e-06,
|
11383 |
+
"loss": 1.1701,
|
11384 |
+
"step": 1625
|
11385 |
+
},
|
11386 |
+
{
|
11387 |
+
"epoch": 1.5224719101123596,
|
11388 |
+
"grad_norm": 0.0,
|
11389 |
+
"learning_rate": 5.918208894667557e-06,
|
11390 |
+
"loss": 1.1008,
|
11391 |
+
"step": 1626
|
11392 |
+
},
|
11393 |
+
{
|
11394 |
+
"epoch": 1.5234082397003745,
|
11395 |
+
"grad_norm": 0.0,
|
11396 |
+
"learning_rate": 5.896235836442736e-06,
|
11397 |
+
"loss": 1.13,
|
11398 |
+
"step": 1627
|
11399 |
+
},
|
11400 |
+
{
|
11401 |
+
"epoch": 1.5243445692883895,
|
11402 |
+
"grad_norm": 0.0,
|
11403 |
+
"learning_rate": 5.874296590197799e-06,
|
11404 |
+
"loss": 1.0716,
|
11405 |
+
"step": 1628
|
11406 |
+
},
|
11407 |
+
{
|
11408 |
+
"epoch": 1.5252808988764044,
|
11409 |
+
"grad_norm": 0.0,
|
11410 |
+
"learning_rate": 5.852391208529291e-06,
|
11411 |
+
"loss": 1.0896,
|
11412 |
+
"step": 1629
|
11413 |
+
},
|
11414 |
+
{
|
11415 |
+
"epoch": 1.5262172284644193,
|
11416 |
+
"grad_norm": 0.0,
|
11417 |
+
"learning_rate": 5.8305197439525695e-06,
|
11418 |
+
"loss": 1.1592,
|
11419 |
+
"step": 1630
|
11420 |
+
},
|
11421 |
+
{
|
11422 |
+
"epoch": 1.5271535580524345,
|
11423 |
+
"grad_norm": 0.0,
|
11424 |
+
"learning_rate": 5.808682248901702e-06,
|
11425 |
+
"loss": 1.1458,
|
11426 |
+
"step": 1631
|
11427 |
+
},
|
11428 |
+
{
|
11429 |
+
"epoch": 1.5280898876404494,
|
11430 |
+
"grad_norm": 0.0,
|
11431 |
+
"learning_rate": 5.78687877572929e-06,
|
11432 |
+
"loss": 1.1197,
|
11433 |
+
"step": 1632
|
11434 |
+
},
|
11435 |
+
{
|
11436 |
+
"epoch": 1.5290262172284645,
|
11437 |
+
"grad_norm": 0.0,
|
11438 |
+
"learning_rate": 5.765109376706395e-06,
|
11439 |
+
"loss": 1.1197,
|
11440 |
+
"step": 1633
|
11441 |
+
},
|
11442 |
+
{
|
11443 |
+
"epoch": 1.5299625468164795,
|
11444 |
+
"grad_norm": 0.0,
|
11445 |
+
"learning_rate": 5.743374104022384e-06,
|
11446 |
+
"loss": 1.1329,
|
11447 |
+
"step": 1634
|
11448 |
+
},
|
11449 |
+
{
|
11450 |
+
"epoch": 1.5308988764044944,
|
11451 |
+
"grad_norm": 0.0,
|
11452 |
+
"learning_rate": 5.7216730097848e-06,
|
11453 |
+
"loss": 1.1374,
|
11454 |
+
"step": 1635
|
11455 |
+
},
|
11456 |
+
{
|
11457 |
+
"epoch": 1.5318352059925093,
|
11458 |
+
"grad_norm": 0.0,
|
11459 |
+
"learning_rate": 5.7000061460192544e-06,
|
11460 |
+
"loss": 1.0821,
|
11461 |
+
"step": 1636
|
11462 |
+
},
|
11463 |
+
{
|
11464 |
+
"epoch": 1.5327715355805243,
|
11465 |
+
"grad_norm": 0.0,
|
11466 |
+
"learning_rate": 5.678373564669304e-06,
|
11467 |
+
"loss": 1.141,
|
11468 |
+
"step": 1637
|
11469 |
+
},
|
11470 |
+
{
|
11471 |
+
"epoch": 1.5337078651685392,
|
11472 |
+
"grad_norm": 0.0,
|
11473 |
+
"learning_rate": 5.656775317596306e-06,
|
11474 |
+
"loss": 1.1377,
|
11475 |
+
"step": 1638
|
11476 |
+
},
|
11477 |
+
{
|
11478 |
+
"epoch": 1.5346441947565543,
|
11479 |
+
"grad_norm": 0.0,
|
11480 |
+
"learning_rate": 5.635211456579299e-06,
|
11481 |
+
"loss": 1.1862,
|
11482 |
+
"step": 1639
|
11483 |
+
},
|
11484 |
+
{
|
11485 |
+
"epoch": 1.5355805243445693,
|
11486 |
+
"grad_norm": 0.0,
|
11487 |
+
"learning_rate": 5.613682033314913e-06,
|
11488 |
+
"loss": 1.1611,
|
11489 |
+
"step": 1640
|
11490 |
+
},
|
11491 |
+
{
|
11492 |
+
"epoch": 1.5365168539325844,
|
11493 |
+
"grad_norm": 0.0,
|
11494 |
+
"learning_rate": 5.5921870994171945e-06,
|
11495 |
+
"loss": 1.1548,
|
11496 |
+
"step": 1641
|
11497 |
+
},
|
11498 |
+
{
|
11499 |
+
"epoch": 1.5374531835205993,
|
11500 |
+
"grad_norm": 0.0,
|
11501 |
+
"learning_rate": 5.570726706417502e-06,
|
11502 |
+
"loss": 1.1167,
|
11503 |
+
"step": 1642
|
11504 |
+
},
|
11505 |
+
{
|
11506 |
+
"epoch": 1.5383895131086143,
|
11507 |
+
"grad_norm": 0.0,
|
11508 |
+
"learning_rate": 5.5493009057644055e-06,
|
11509 |
+
"loss": 1.0833,
|
11510 |
+
"step": 1643
|
11511 |
+
},
|
11512 |
+
{
|
11513 |
+
"epoch": 1.5393258426966292,
|
11514 |
+
"grad_norm": 0.0,
|
11515 |
+
"learning_rate": 5.527909748823543e-06,
|
11516 |
+
"loss": 1.1003,
|
11517 |
+
"step": 1644
|
11518 |
+
},
|
11519 |
+
{
|
11520 |
+
"epoch": 1.5402621722846441,
|
11521 |
+
"grad_norm": 0.0,
|
11522 |
+
"learning_rate": 5.506553286877476e-06,
|
11523 |
+
"loss": 1.0904,
|
11524 |
+
"step": 1645
|
11525 |
+
},
|
11526 |
+
{
|
11527 |
+
"epoch": 1.541198501872659,
|
11528 |
+
"grad_norm": 0.0,
|
11529 |
+
"learning_rate": 5.485231571125617e-06,
|
11530 |
+
"loss": 1.1111,
|
11531 |
+
"step": 1646
|
11532 |
+
},
|
11533 |
+
{
|
11534 |
+
"epoch": 1.5421348314606742,
|
11535 |
+
"grad_norm": 0.0,
|
11536 |
+
"learning_rate": 5.463944652684066e-06,
|
11537 |
+
"loss": 1.1194,
|
11538 |
+
"step": 1647
|
11539 |
+
},
|
11540 |
+
{
|
11541 |
+
"epoch": 1.5430711610486891,
|
11542 |
+
"grad_norm": 0.0,
|
11543 |
+
"learning_rate": 5.442692582585487e-06,
|
11544 |
+
"loss": 1.1249,
|
11545 |
+
"step": 1648
|
11546 |
+
},
|
11547 |
+
{
|
11548 |
+
"epoch": 1.5440074906367043,
|
11549 |
+
"grad_norm": 0.0,
|
11550 |
+
"learning_rate": 5.421475411779029e-06,
|
11551 |
+
"loss": 1.1436,
|
11552 |
+
"step": 1649
|
11553 |
+
},
|
11554 |
+
{
|
11555 |
+
"epoch": 1.5449438202247192,
|
11556 |
+
"grad_norm": 0.0,
|
11557 |
+
"learning_rate": 5.400293191130154e-06,
|
11558 |
+
"loss": 1.1295,
|
11559 |
+
"step": 1650
|
11560 |
+
},
|
11561 |
+
{
|
11562 |
+
"epoch": 1.5458801498127341,
|
11563 |
+
"grad_norm": 0.0,
|
11564 |
+
"learning_rate": 5.3791459714205365e-06,
|
11565 |
+
"loss": 1.1103,
|
11566 |
+
"step": 1651
|
11567 |
+
},
|
11568 |
+
{
|
11569 |
+
"epoch": 1.546816479400749,
|
11570 |
+
"grad_norm": 0.0,
|
11571 |
+
"learning_rate": 5.35803380334795e-06,
|
11572 |
+
"loss": 1.1181,
|
11573 |
+
"step": 1652
|
11574 |
+
},
|
11575 |
+
{
|
11576 |
+
"epoch": 1.547752808988764,
|
11577 |
+
"grad_norm": 0.0,
|
11578 |
+
"learning_rate": 5.336956737526127e-06,
|
11579 |
+
"loss": 1.1575,
|
11580 |
+
"step": 1653
|
11581 |
+
},
|
11582 |
+
{
|
11583 |
+
"epoch": 1.548689138576779,
|
11584 |
+
"grad_norm": 0.0,
|
11585 |
+
"learning_rate": 5.315914824484656e-06,
|
11586 |
+
"loss": 1.1069,
|
11587 |
+
"step": 1654
|
11588 |
+
},
|
11589 |
+
{
|
11590 |
+
"epoch": 1.5496254681647939,
|
11591 |
+
"grad_norm": 0.0,
|
11592 |
+
"learning_rate": 5.2949081146688396e-06,
|
11593 |
+
"loss": 1.1024,
|
11594 |
+
"step": 1655
|
11595 |
+
},
|
11596 |
+
{
|
11597 |
+
"epoch": 1.550561797752809,
|
11598 |
+
"grad_norm": 0.0,
|
11599 |
+
"learning_rate": 5.2739366584396e-06,
|
11600 |
+
"loss": 1.1304,
|
11601 |
+
"step": 1656
|
11602 |
+
},
|
11603 |
+
{
|
11604 |
+
"epoch": 1.551498127340824,
|
11605 |
+
"grad_norm": 0.0,
|
11606 |
+
"learning_rate": 5.253000506073323e-06,
|
11607 |
+
"loss": 1.1182,
|
11608 |
+
"step": 1657
|
11609 |
+
},
|
11610 |
+
{
|
11611 |
+
"epoch": 1.552434456928839,
|
11612 |
+
"grad_norm": 0.0,
|
11613 |
+
"learning_rate": 5.232099707761782e-06,
|
11614 |
+
"loss": 1.1937,
|
11615 |
+
"step": 1658
|
11616 |
+
},
|
11617 |
+
{
|
11618 |
+
"epoch": 1.553370786516854,
|
11619 |
+
"grad_norm": 0.0,
|
11620 |
+
"learning_rate": 5.211234313611977e-06,
|
11621 |
+
"loss": 1.1658,
|
11622 |
+
"step": 1659
|
11623 |
+
},
|
11624 |
+
{
|
11625 |
+
"epoch": 1.554307116104869,
|
11626 |
+
"grad_norm": 0.0,
|
11627 |
+
"learning_rate": 5.190404373646028e-06,
|
11628 |
+
"loss": 1.1923,
|
11629 |
+
"step": 1660
|
11630 |
+
},
|
11631 |
+
{
|
11632 |
+
"epoch": 1.5552434456928839,
|
11633 |
+
"grad_norm": 0.0,
|
11634 |
+
"learning_rate": 5.169609937801077e-06,
|
11635 |
+
"loss": 1.1418,
|
11636 |
+
"step": 1661
|
11637 |
+
},
|
11638 |
+
{
|
11639 |
+
"epoch": 1.5561797752808988,
|
11640 |
+
"grad_norm": 0.0,
|
11641 |
+
"learning_rate": 5.148851055929127e-06,
|
11642 |
+
"loss": 1.1935,
|
11643 |
+
"step": 1662
|
11644 |
+
},
|
11645 |
+
{
|
11646 |
+
"epoch": 1.5571161048689137,
|
11647 |
+
"grad_norm": 0.0,
|
11648 |
+
"learning_rate": 5.128127777796965e-06,
|
11649 |
+
"loss": 1.1214,
|
11650 |
+
"step": 1663
|
11651 |
+
},
|
11652 |
+
{
|
11653 |
+
"epoch": 1.5580524344569289,
|
11654 |
+
"grad_norm": 0.0,
|
11655 |
+
"learning_rate": 5.107440153086012e-06,
|
11656 |
+
"loss": 1.1117,
|
11657 |
+
"step": 1664
|
11658 |
+
},
|
11659 |
+
{
|
11660 |
+
"epoch": 1.5589887640449438,
|
11661 |
+
"grad_norm": 0.0,
|
11662 |
+
"learning_rate": 5.086788231392205e-06,
|
11663 |
+
"loss": 1.1638,
|
11664 |
+
"step": 1665
|
11665 |
+
},
|
11666 |
+
{
|
11667 |
+
"epoch": 1.559925093632959,
|
11668 |
+
"grad_norm": 0.0,
|
11669 |
+
"learning_rate": 5.066172062225909e-06,
|
11670 |
+
"loss": 1.1493,
|
11671 |
+
"step": 1666
|
11672 |
+
},
|
11673 |
+
{
|
11674 |
+
"epoch": 1.5608614232209739,
|
11675 |
+
"grad_norm": 0.0,
|
11676 |
+
"learning_rate": 5.045591695011769e-06,
|
11677 |
+
"loss": 1.0885,
|
11678 |
+
"step": 1667
|
11679 |
+
},
|
11680 |
+
{
|
11681 |
+
"epoch": 1.5617977528089888,
|
11682 |
+
"grad_norm": 0.0,
|
11683 |
+
"learning_rate": 5.025047179088585e-06,
|
11684 |
+
"loss": 1.1453,
|
11685 |
+
"step": 1668
|
11686 |
+
},
|
11687 |
+
{
|
11688 |
+
"epoch": 1.5627340823970037,
|
11689 |
+
"grad_norm": 0.0,
|
11690 |
+
"learning_rate": 5.004538563709231e-06,
|
11691 |
+
"loss": 1.1053,
|
11692 |
+
"step": 1669
|
11693 |
+
},
|
11694 |
+
{
|
11695 |
+
"epoch": 1.5636704119850187,
|
11696 |
+
"grad_norm": 0.0,
|
11697 |
+
"learning_rate": 4.9840658980404975e-06,
|
11698 |
+
"loss": 1.1227,
|
11699 |
+
"step": 1670
|
11700 |
+
},
|
11701 |
+
{
|
11702 |
+
"epoch": 1.5646067415730336,
|
11703 |
+
"grad_norm": 0.0,
|
11704 |
+
"learning_rate": 4.9636292311629895e-06,
|
11705 |
+
"loss": 1.1435,
|
11706 |
+
"step": 1671
|
11707 |
+
},
|
11708 |
+
{
|
11709 |
+
"epoch": 1.5655430711610487,
|
11710 |
+
"grad_norm": 0.0,
|
11711 |
+
"learning_rate": 4.943228612071029e-06,
|
11712 |
+
"loss": 1.1342,
|
11713 |
+
"step": 1672
|
11714 |
+
},
|
11715 |
+
{
|
11716 |
+
"epoch": 1.5664794007490637,
|
11717 |
+
"grad_norm": 0.0,
|
11718 |
+
"learning_rate": 4.922864089672495e-06,
|
11719 |
+
"loss": 1.1126,
|
11720 |
+
"step": 1673
|
11721 |
+
},
|
11722 |
+
{
|
11723 |
+
"epoch": 1.5674157303370788,
|
11724 |
+
"grad_norm": 0.0,
|
11725 |
+
"learning_rate": 4.902535712788732e-06,
|
11726 |
+
"loss": 1.1245,
|
11727 |
+
"step": 1674
|
11728 |
+
},
|
11729 |
+
{
|
11730 |
+
"epoch": 1.5683520599250937,
|
11731 |
+
"grad_norm": 0.0,
|
11732 |
+
"learning_rate": 4.882243530154454e-06,
|
11733 |
+
"loss": 1.1784,
|
11734 |
+
"step": 1675
|
11735 |
+
},
|
11736 |
+
{
|
11737 |
+
"epoch": 1.5692883895131087,
|
11738 |
+
"grad_norm": 0.0,
|
11739 |
+
"learning_rate": 4.86198759041757e-06,
|
11740 |
+
"loss": 1.1366,
|
11741 |
+
"step": 1676
|
11742 |
+
},
|
11743 |
+
{
|
11744 |
+
"epoch": 1.5702247191011236,
|
11745 |
+
"grad_norm": 0.0,
|
11746 |
+
"learning_rate": 4.841767942139124e-06,
|
11747 |
+
"loss": 1.1074,
|
11748 |
+
"step": 1677
|
11749 |
+
},
|
11750 |
+
{
|
11751 |
+
"epoch": 1.5711610486891385,
|
11752 |
+
"grad_norm": 0.0,
|
11753 |
+
"learning_rate": 4.821584633793157e-06,
|
11754 |
+
"loss": 1.0981,
|
11755 |
+
"step": 1678
|
11756 |
+
},
|
11757 |
+
{
|
11758 |
+
"epoch": 1.5720973782771535,
|
11759 |
+
"grad_norm": 0.0,
|
11760 |
+
"learning_rate": 4.8014377137665794e-06,
|
11761 |
+
"loss": 1.1212,
|
11762 |
+
"step": 1679
|
11763 |
+
},
|
11764 |
+
{
|
11765 |
+
"epoch": 1.5730337078651684,
|
11766 |
+
"grad_norm": 0.0,
|
11767 |
+
"learning_rate": 4.781327230359057e-06,
|
11768 |
+
"loss": 1.1368,
|
11769 |
+
"step": 1680
|
11770 |
+
},
|
11771 |
+
{
|
11772 |
+
"epoch": 1.5739700374531835,
|
11773 |
+
"grad_norm": 0.0,
|
11774 |
+
"learning_rate": 4.761253231782929e-06,
|
11775 |
+
"loss": 1.1538,
|
11776 |
+
"step": 1681
|
11777 |
+
},
|
11778 |
+
{
|
11779 |
+
"epoch": 1.5749063670411985,
|
11780 |
+
"grad_norm": 0.0,
|
11781 |
+
"learning_rate": 4.741215766163047e-06,
|
11782 |
+
"loss": 1.1577,
|
11783 |
+
"step": 1682
|
11784 |
+
},
|
11785 |
+
{
|
11786 |
+
"epoch": 1.5758426966292136,
|
11787 |
+
"grad_norm": 0.0,
|
11788 |
+
"learning_rate": 4.72121488153668e-06,
|
11789 |
+
"loss": 1.1337,
|
11790 |
+
"step": 1683
|
11791 |
+
},
|
11792 |
+
{
|
11793 |
+
"epoch": 1.5767790262172285,
|
11794 |
+
"grad_norm": 0.0,
|
11795 |
+
"learning_rate": 4.7012506258534155e-06,
|
11796 |
+
"loss": 1.1163,
|
11797 |
+
"step": 1684
|
11798 |
+
},
|
11799 |
+
{
|
11800 |
+
"epoch": 1.5777153558052435,
|
11801 |
+
"grad_norm": 0.0,
|
11802 |
+
"learning_rate": 4.6813230469750056e-06,
|
11803 |
+
"loss": 1.1293,
|
11804 |
+
"step": 1685
|
11805 |
+
},
|
11806 |
+
{
|
11807 |
+
"epoch": 1.5786516853932584,
|
11808 |
+
"grad_norm": 0.0,
|
11809 |
+
"learning_rate": 4.661432192675283e-06,
|
11810 |
+
"loss": 1.1452,
|
11811 |
+
"step": 1686
|
11812 |
+
},
|
11813 |
+
{
|
11814 |
+
"epoch": 1.5795880149812733,
|
11815 |
+
"grad_norm": 0.0,
|
11816 |
+
"learning_rate": 4.641578110640055e-06,
|
11817 |
+
"loss": 1.1115,
|
11818 |
+
"step": 1687
|
11819 |
+
},
|
11820 |
+
{
|
11821 |
+
"epoch": 1.5805243445692883,
|
11822 |
+
"grad_norm": 0.0,
|
11823 |
+
"learning_rate": 4.621760848466936e-06,
|
11824 |
+
"loss": 1.1763,
|
11825 |
+
"step": 1688
|
11826 |
+
},
|
11827 |
+
{
|
11828 |
+
"epoch": 1.5814606741573034,
|
11829 |
+
"grad_norm": 0.0,
|
11830 |
+
"learning_rate": 4.601980453665305e-06,
|
11831 |
+
"loss": 1.1725,
|
11832 |
+
"step": 1689
|
11833 |
+
},
|
11834 |
+
{
|
11835 |
+
"epoch": 1.5823970037453183,
|
11836 |
+
"grad_norm": 0.0,
|
11837 |
+
"learning_rate": 4.582236973656142e-06,
|
11838 |
+
"loss": 1.1052,
|
11839 |
+
"step": 1690
|
11840 |
+
},
|
11841 |
+
{
|
11842 |
+
"epoch": 1.5833333333333335,
|
11843 |
+
"grad_norm": 0.0,
|
11844 |
+
"learning_rate": 4.562530455771925e-06,
|
11845 |
+
"loss": 1.1333,
|
11846 |
+
"step": 1691
|
11847 |
+
},
|
11848 |
+
{
|
11849 |
+
"epoch": 1.5842696629213484,
|
11850 |
+
"grad_norm": 0.0,
|
11851 |
+
"learning_rate": 4.5428609472565155e-06,
|
11852 |
+
"loss": 1.1603,
|
11853 |
+
"step": 1692
|
11854 |
+
},
|
11855 |
+
{
|
11856 |
+
"epoch": 1.5852059925093633,
|
11857 |
+
"grad_norm": 0.0,
|
11858 |
+
"learning_rate": 4.523228495265069e-06,
|
11859 |
+
"loss": 1.1142,
|
11860 |
+
"step": 1693
|
11861 |
+
},
|
11862 |
+
{
|
11863 |
+
"epoch": 1.5861423220973783,
|
11864 |
+
"grad_norm": 0.0,
|
11865 |
+
"learning_rate": 4.503633146863884e-06,
|
11866 |
+
"loss": 1.0894,
|
11867 |
+
"step": 1694
|
11868 |
+
},
|
11869 |
+
{
|
11870 |
+
"epoch": 1.5870786516853932,
|
11871 |
+
"grad_norm": 0.0,
|
11872 |
+
"learning_rate": 4.484074949030311e-06,
|
11873 |
+
"loss": 1.1276,
|
11874 |
+
"step": 1695
|
11875 |
+
},
|
11876 |
+
{
|
11877 |
+
"epoch": 1.5880149812734081,
|
11878 |
+
"grad_norm": 0.0,
|
11879 |
+
"learning_rate": 4.464553948652652e-06,
|
11880 |
+
"loss": 1.1378,
|
11881 |
+
"step": 1696
|
11882 |
+
},
|
11883 |
+
{
|
11884 |
+
"epoch": 1.5889513108614233,
|
11885 |
+
"grad_norm": 0.0,
|
11886 |
+
"learning_rate": 4.445070192530014e-06,
|
11887 |
+
"loss": 1.1139,
|
11888 |
+
"step": 1697
|
11889 |
+
},
|
11890 |
+
{
|
11891 |
+
"epoch": 1.5898876404494382,
|
11892 |
+
"grad_norm": 0.0,
|
11893 |
+
"learning_rate": 4.425623727372219e-06,
|
11894 |
+
"loss": 1.1721,
|
11895 |
+
"step": 1698
|
11896 |
+
},
|
11897 |
+
{
|
11898 |
+
"epoch": 1.5908239700374533,
|
11899 |
+
"grad_norm": 0.0,
|
11900 |
+
"learning_rate": 4.406214599799699e-06,
|
11901 |
+
"loss": 1.1481,
|
11902 |
+
"step": 1699
|
11903 |
+
},
|
11904 |
+
{
|
11905 |
+
"epoch": 1.5917602996254683,
|
11906 |
+
"grad_norm": 0.0,
|
11907 |
+
"learning_rate": 4.38684285634337e-06,
|
11908 |
+
"loss": 1.1395,
|
11909 |
+
"step": 1700
|
11910 |
+
},
|
11911 |
+
{
|
11912 |
+
"epoch": 1.5926966292134832,
|
11913 |
+
"grad_norm": 0.0,
|
11914 |
+
"learning_rate": 4.367508543444514e-06,
|
11915 |
+
"loss": 1.1295,
|
11916 |
+
"step": 1701
|
11917 |
+
},
|
11918 |
+
{
|
11919 |
+
"epoch": 1.5936329588014981,
|
11920 |
+
"grad_norm": 0.0,
|
11921 |
+
"learning_rate": 4.3482117074546944e-06,
|
11922 |
+
"loss": 1.1242,
|
11923 |
+
"step": 1702
|
11924 |
+
},
|
11925 |
+
{
|
11926 |
+
"epoch": 1.594569288389513,
|
11927 |
+
"grad_norm": 0.0,
|
11928 |
+
"learning_rate": 4.328952394635615e-06,
|
11929 |
+
"loss": 1.1241,
|
11930 |
+
"step": 1703
|
11931 |
+
},
|
11932 |
+
{
|
11933 |
+
"epoch": 1.595505617977528,
|
11934 |
+
"grad_norm": 0.0,
|
11935 |
+
"learning_rate": 4.309730651159023e-06,
|
11936 |
+
"loss": 1.1349,
|
11937 |
+
"step": 1704
|
11938 |
+
},
|
11939 |
+
{
|
11940 |
+
"epoch": 1.596441947565543,
|
11941 |
+
"grad_norm": 0.0,
|
11942 |
+
"learning_rate": 4.290546523106614e-06,
|
11943 |
+
"loss": 1.1303,
|
11944 |
+
"step": 1705
|
11945 |
+
},
|
11946 |
+
{
|
11947 |
+
"epoch": 1.597378277153558,
|
11948 |
+
"grad_norm": 0.0,
|
11949 |
+
"learning_rate": 4.271400056469885e-06,
|
11950 |
+
"loss": 1.1152,
|
11951 |
+
"step": 1706
|
11952 |
+
},
|
11953 |
+
{
|
11954 |
+
"epoch": 1.598314606741573,
|
11955 |
+
"grad_norm": 0.0,
|
11956 |
+
"learning_rate": 4.252291297150053e-06,
|
11957 |
+
"loss": 1.1145,
|
11958 |
+
"step": 1707
|
11959 |
+
},
|
11960 |
+
{
|
11961 |
+
"epoch": 1.5992509363295881,
|
11962 |
+
"grad_norm": 0.0,
|
11963 |
+
"learning_rate": 4.233220290957944e-06,
|
11964 |
+
"loss": 1.1283,
|
11965 |
+
"step": 1708
|
11966 |
+
},
|
11967 |
+
{
|
11968 |
+
"epoch": 1.600187265917603,
|
11969 |
+
"grad_norm": 0.0,
|
11970 |
+
"learning_rate": 4.214187083613858e-06,
|
11971 |
+
"loss": 1.1321,
|
11972 |
+
"step": 1709
|
11973 |
+
},
|
11974 |
+
{
|
11975 |
+
"epoch": 1.601123595505618,
|
11976 |
+
"grad_norm": 0.0,
|
11977 |
+
"learning_rate": 4.195191720747496e-06,
|
11978 |
+
"loss": 1.1436,
|
11979 |
+
"step": 1710
|
11980 |
+
},
|
11981 |
+
{
|
11982 |
+
"epoch": 1.602059925093633,
|
11983 |
+
"grad_norm": 0.0,
|
11984 |
+
"learning_rate": 4.176234247897824e-06,
|
11985 |
+
"loss": 1.1157,
|
11986 |
+
"step": 1711
|
11987 |
+
},
|
11988 |
+
{
|
11989 |
+
"epoch": 1.6029962546816479,
|
11990 |
+
"grad_norm": 0.0,
|
11991 |
+
"learning_rate": 4.157314710512968e-06,
|
11992 |
+
"loss": 1.1178,
|
11993 |
+
"step": 1712
|
11994 |
+
},
|
11995 |
+
{
|
11996 |
+
"epoch": 1.6039325842696628,
|
11997 |
+
"grad_norm": 0.0,
|
11998 |
+
"learning_rate": 4.138433153950108e-06,
|
11999 |
+
"loss": 1.1501,
|
12000 |
+
"step": 1713
|
12001 |
+
},
|
12002 |
+
{
|
12003 |
+
"epoch": 1.604868913857678,
|
12004 |
+
"grad_norm": 0.0,
|
12005 |
+
"learning_rate": 4.11958962347538e-06,
|
12006 |
+
"loss": 1.1095,
|
12007 |
+
"step": 1714
|
12008 |
+
},
|
12009 |
+
{
|
12010 |
+
"epoch": 1.6058052434456929,
|
12011 |
+
"grad_norm": 0.0,
|
12012 |
+
"learning_rate": 4.100784164263747e-06,
|
12013 |
+
"loss": 1.172,
|
12014 |
+
"step": 1715
|
12015 |
+
},
|
12016 |
+
{
|
12017 |
+
"epoch": 1.606741573033708,
|
12018 |
+
"grad_norm": 0.0,
|
12019 |
+
"learning_rate": 4.0820168213989e-06,
|
12020 |
+
"loss": 1.1468,
|
12021 |
+
"step": 1716
|
12022 |
+
},
|
12023 |
+
{
|
12024 |
+
"epoch": 1.607677902621723,
|
12025 |
+
"grad_norm": 0.0,
|
12026 |
+
"learning_rate": 4.063287639873163e-06,
|
12027 |
+
"loss": 1.1513,
|
12028 |
+
"step": 1717
|
12029 |
+
},
|
12030 |
+
{
|
12031 |
+
"epoch": 1.6086142322097379,
|
12032 |
+
"grad_norm": 0.0,
|
12033 |
+
"learning_rate": 4.044596664587364e-06,
|
12034 |
+
"loss": 1.1497,
|
12035 |
+
"step": 1718
|
12036 |
+
},
|
12037 |
+
{
|
12038 |
+
"epoch": 1.6095505617977528,
|
12039 |
+
"grad_norm": 0.0,
|
12040 |
+
"learning_rate": 4.02594394035073e-06,
|
12041 |
+
"loss": 1.1375,
|
12042 |
+
"step": 1719
|
12043 |
+
},
|
12044 |
+
{
|
12045 |
+
"epoch": 1.6104868913857677,
|
12046 |
+
"grad_norm": 0.0,
|
12047 |
+
"learning_rate": 4.0073295118808e-06,
|
12048 |
+
"loss": 1.1103,
|
12049 |
+
"step": 1720
|
12050 |
+
},
|
12051 |
+
{
|
12052 |
+
"epoch": 1.6114232209737827,
|
12053 |
+
"grad_norm": 0.0,
|
12054 |
+
"learning_rate": 3.988753423803302e-06,
|
12055 |
+
"loss": 1.1027,
|
12056 |
+
"step": 1721
|
12057 |
+
},
|
12058 |
+
{
|
12059 |
+
"epoch": 1.6123595505617978,
|
12060 |
+
"grad_norm": 0.0,
|
12061 |
+
"learning_rate": 3.970215720652035e-06,
|
12062 |
+
"loss": 1.1495,
|
12063 |
+
"step": 1722
|
12064 |
+
},
|
12065 |
+
{
|
12066 |
+
"epoch": 1.6132958801498127,
|
12067 |
+
"grad_norm": 0.0,
|
12068 |
+
"learning_rate": 3.951716446868792e-06,
|
12069 |
+
"loss": 1.1401,
|
12070 |
+
"step": 1723
|
12071 |
+
},
|
12072 |
+
{
|
12073 |
+
"epoch": 1.6142322097378277,
|
12074 |
+
"grad_norm": 0.0,
|
12075 |
+
"learning_rate": 3.9332556468032154e-06,
|
12076 |
+
"loss": 1.1324,
|
12077 |
+
"step": 1724
|
12078 |
+
},
|
12079 |
+
{
|
12080 |
+
"epoch": 1.6151685393258428,
|
12081 |
+
"grad_norm": 0.0,
|
12082 |
+
"learning_rate": 3.91483336471274e-06,
|
12083 |
+
"loss": 1.1533,
|
12084 |
+
"step": 1725
|
12085 |
+
},
|
12086 |
+
{
|
12087 |
+
"epoch": 1.6161048689138577,
|
12088 |
+
"grad_norm": 0.0,
|
12089 |
+
"learning_rate": 3.896449644762432e-06,
|
12090 |
+
"loss": 1.1735,
|
12091 |
+
"step": 1726
|
12092 |
+
},
|
12093 |
+
{
|
12094 |
+
"epoch": 1.6170411985018727,
|
12095 |
+
"grad_norm": 0.0,
|
12096 |
+
"learning_rate": 3.8781045310249195e-06,
|
12097 |
+
"loss": 1.1851,
|
12098 |
+
"step": 1727
|
12099 |
+
},
|
12100 |
+
{
|
12101 |
+
"epoch": 1.6179775280898876,
|
12102 |
+
"grad_norm": 0.0,
|
12103 |
+
"learning_rate": 3.859798067480287e-06,
|
12104 |
+
"loss": 1.122,
|
12105 |
+
"step": 1728
|
12106 |
+
},
|
12107 |
+
{
|
12108 |
+
"epoch": 1.6189138576779025,
|
12109 |
+
"grad_norm": 0.0,
|
12110 |
+
"learning_rate": 3.841530298015947e-06,
|
12111 |
+
"loss": 1.1203,
|
12112 |
+
"step": 1729
|
12113 |
+
},
|
12114 |
+
{
|
12115 |
+
"epoch": 1.6198501872659175,
|
12116 |
+
"grad_norm": 0.0,
|
12117 |
+
"learning_rate": 3.823301266426547e-06,
|
12118 |
+
"loss": 1.157,
|
12119 |
+
"step": 1730
|
12120 |
+
},
|
12121 |
+
{
|
12122 |
+
"epoch": 1.6207865168539326,
|
12123 |
+
"grad_norm": 0.0,
|
12124 |
+
"learning_rate": 3.8051110164138784e-06,
|
12125 |
+
"loss": 1.1388,
|
12126 |
+
"step": 1731
|
12127 |
+
},
|
12128 |
+
{
|
12129 |
+
"epoch": 1.6217228464419475,
|
12130 |
+
"grad_norm": 0.0,
|
12131 |
+
"learning_rate": 3.786959591586743e-06,
|
12132 |
+
"loss": 1.1765,
|
12133 |
+
"step": 1732
|
12134 |
+
},
|
12135 |
+
{
|
12136 |
+
"epoch": 1.6226591760299627,
|
12137 |
+
"grad_norm": 0.0,
|
12138 |
+
"learning_rate": 3.7688470354608785e-06,
|
12139 |
+
"loss": 1.1289,
|
12140 |
+
"step": 1733
|
12141 |
+
},
|
12142 |
+
{
|
12143 |
+
"epoch": 1.6235955056179776,
|
12144 |
+
"grad_norm": 0.0,
|
12145 |
+
"learning_rate": 3.750773391458835e-06,
|
12146 |
+
"loss": 1.1855,
|
12147 |
+
"step": 1734
|
12148 |
+
},
|
12149 |
+
{
|
12150 |
+
"epoch": 1.6245318352059925,
|
12151 |
+
"grad_norm": 0.0,
|
12152 |
+
"learning_rate": 3.732738702909873e-06,
|
12153 |
+
"loss": 1.112,
|
12154 |
+
"step": 1735
|
12155 |
+
},
|
12156 |
+
{
|
12157 |
+
"epoch": 1.6254681647940075,
|
12158 |
+
"grad_norm": 0.0,
|
12159 |
+
"learning_rate": 3.714743013049853e-06,
|
12160 |
+
"loss": 1.147,
|
12161 |
+
"step": 1736
|
12162 |
+
},
|
12163 |
+
{
|
12164 |
+
"epoch": 1.6264044943820224,
|
12165 |
+
"grad_norm": 0.0,
|
12166 |
+
"learning_rate": 3.6967863650211677e-06,
|
12167 |
+
"loss": 1.0953,
|
12168 |
+
"step": 1737
|
12169 |
+
},
|
12170 |
+
{
|
12171 |
+
"epoch": 1.6273408239700373,
|
12172 |
+
"grad_norm": 0.0,
|
12173 |
+
"learning_rate": 3.678868801872586e-06,
|
12174 |
+
"loss": 1.119,
|
12175 |
+
"step": 1738
|
12176 |
+
},
|
12177 |
+
{
|
12178 |
+
"epoch": 1.6282771535580525,
|
12179 |
+
"grad_norm": 0.0,
|
12180 |
+
"learning_rate": 3.6609903665591804e-06,
|
12181 |
+
"loss": 1.1757,
|
12182 |
+
"step": 1739
|
12183 |
+
},
|
12184 |
+
{
|
12185 |
+
"epoch": 1.6292134831460674,
|
12186 |
+
"grad_norm": 0.0,
|
12187 |
+
"learning_rate": 3.6431511019422395e-06,
|
12188 |
+
"loss": 1.1023,
|
12189 |
+
"step": 1740
|
12190 |
+
},
|
12191 |
+
{
|
12192 |
+
"epoch": 1.6301498127340825,
|
12193 |
+
"grad_norm": 0.0,
|
12194 |
+
"learning_rate": 3.625351050789121e-06,
|
12195 |
+
"loss": 1.1254,
|
12196 |
+
"step": 1741
|
12197 |
+
},
|
12198 |
+
{
|
12199 |
+
"epoch": 1.6310861423220975,
|
12200 |
+
"grad_norm": 0.0,
|
12201 |
+
"learning_rate": 3.607590255773179e-06,
|
12202 |
+
"loss": 1.1341,
|
12203 |
+
"step": 1742
|
12204 |
+
},
|
12205 |
+
{
|
12206 |
+
"epoch": 1.6320224719101124,
|
12207 |
+
"grad_norm": 0.0,
|
12208 |
+
"learning_rate": 3.5898687594736625e-06,
|
12209 |
+
"loss": 1.1491,
|
12210 |
+
"step": 1743
|
12211 |
+
},
|
12212 |
+
{
|
12213 |
+
"epoch": 1.6329588014981273,
|
12214 |
+
"grad_norm": 0.0,
|
12215 |
+
"learning_rate": 3.5721866043756117e-06,
|
12216 |
+
"loss": 1.1443,
|
12217 |
+
"step": 1744
|
12218 |
+
},
|
12219 |
+
{
|
12220 |
+
"epoch": 1.6338951310861423,
|
12221 |
+
"grad_norm": 0.0,
|
12222 |
+
"learning_rate": 3.5545438328697324e-06,
|
12223 |
+
"loss": 1.1176,
|
12224 |
+
"step": 1745
|
12225 |
+
},
|
12226 |
+
{
|
12227 |
+
"epoch": 1.6348314606741572,
|
12228 |
+
"grad_norm": 0.0,
|
12229 |
+
"learning_rate": 3.536940487252334e-06,
|
12230 |
+
"loss": 1.1327,
|
12231 |
+
"step": 1746
|
12232 |
+
},
|
12233 |
+
{
|
12234 |
+
"epoch": 1.6357677902621723,
|
12235 |
+
"grad_norm": 0.0,
|
12236 |
+
"learning_rate": 3.5193766097251914e-06,
|
12237 |
+
"loss": 1.1626,
|
12238 |
+
"step": 1747
|
12239 |
+
},
|
12240 |
+
{
|
12241 |
+
"epoch": 1.6367041198501873,
|
12242 |
+
"grad_norm": 0.0,
|
12243 |
+
"learning_rate": 3.501852242395463e-06,
|
12244 |
+
"loss": 1.1547,
|
12245 |
+
"step": 1748
|
12246 |
+
},
|
12247 |
+
{
|
12248 |
+
"epoch": 1.6376404494382022,
|
12249 |
+
"grad_norm": 0.0,
|
12250 |
+
"learning_rate": 3.4843674272755967e-06,
|
12251 |
+
"loss": 1.2044,
|
12252 |
+
"step": 1749
|
12253 |
+
},
|
12254 |
+
{
|
12255 |
+
"epoch": 1.6385767790262173,
|
12256 |
+
"grad_norm": 0.0,
|
12257 |
+
"learning_rate": 3.4669222062832098e-06,
|
12258 |
+
"loss": 1.1689,
|
12259 |
+
"step": 1750
|
12260 |
+
},
|
12261 |
+
{
|
12262 |
+
"epoch": 1.6395131086142323,
|
12263 |
+
"grad_norm": 0.0,
|
12264 |
+
"learning_rate": 3.449516621240994e-06,
|
12265 |
+
"loss": 1.1815,
|
12266 |
+
"step": 1751
|
12267 |
+
},
|
12268 |
+
{
|
12269 |
+
"epoch": 1.6404494382022472,
|
12270 |
+
"grad_norm": 0.0,
|
12271 |
+
"learning_rate": 3.432150713876634e-06,
|
12272 |
+
"loss": 1.1351,
|
12273 |
+
"step": 1752
|
12274 |
+
},
|
12275 |
+
{
|
12276 |
+
"epoch": 1.6413857677902621,
|
12277 |
+
"grad_norm": 0.0,
|
12278 |
+
"learning_rate": 3.414824525822673e-06,
|
12279 |
+
"loss": 1.1487,
|
12280 |
+
"step": 1753
|
12281 |
+
},
|
12282 |
+
{
|
12283 |
+
"epoch": 1.642322097378277,
|
12284 |
+
"grad_norm": 0.0,
|
12285 |
+
"learning_rate": 3.397538098616453e-06,
|
12286 |
+
"loss": 1.1323,
|
12287 |
+
"step": 1754
|
12288 |
+
},
|
12289 |
+
{
|
12290 |
+
"epoch": 1.643258426966292,
|
12291 |
+
"grad_norm": 0.0,
|
12292 |
+
"learning_rate": 3.38029147369997e-06,
|
12293 |
+
"loss": 1.1588,
|
12294 |
+
"step": 1755
|
12295 |
+
},
|
12296 |
+
{
|
12297 |
+
"epoch": 1.6441947565543071,
|
12298 |
+
"grad_norm": 0.0,
|
12299 |
+
"learning_rate": 3.3630846924198268e-06,
|
12300 |
+
"loss": 1.1534,
|
12301 |
+
"step": 1756
|
12302 |
+
},
|
12303 |
+
{
|
12304 |
+
"epoch": 1.645131086142322,
|
12305 |
+
"grad_norm": 0.0,
|
12306 |
+
"learning_rate": 3.3459177960270807e-06,
|
12307 |
+
"loss": 1.1498,
|
12308 |
+
"step": 1757
|
12309 |
+
},
|
12310 |
+
{
|
12311 |
+
"epoch": 1.6460674157303372,
|
12312 |
+
"grad_norm": 0.0,
|
12313 |
+
"learning_rate": 3.3287908256771904e-06,
|
12314 |
+
"loss": 1.0942,
|
12315 |
+
"step": 1758
|
12316 |
+
},
|
12317 |
+
{
|
12318 |
+
"epoch": 1.6470037453183521,
|
12319 |
+
"grad_norm": 0.0,
|
12320 |
+
"learning_rate": 3.311703822429886e-06,
|
12321 |
+
"loss": 1.1544,
|
12322 |
+
"step": 1759
|
12323 |
+
},
|
12324 |
+
{
|
12325 |
+
"epoch": 1.647940074906367,
|
12326 |
+
"grad_norm": 0.0,
|
12327 |
+
"learning_rate": 3.294656827249074e-06,
|
12328 |
+
"loss": 1.1519,
|
12329 |
+
"step": 1760
|
12330 |
+
},
|
12331 |
+
{
|
12332 |
+
"epoch": 1.648876404494382,
|
12333 |
+
"grad_norm": 0.0,
|
12334 |
+
"learning_rate": 3.277649881002767e-06,
|
12335 |
+
"loss": 1.2016,
|
12336 |
+
"step": 1761
|
12337 |
+
},
|
12338 |
+
{
|
12339 |
+
"epoch": 1.649812734082397,
|
12340 |
+
"grad_norm": 0.0,
|
12341 |
+
"learning_rate": 3.2606830244629516e-06,
|
12342 |
+
"loss": 1.1253,
|
12343 |
+
"step": 1762
|
12344 |
+
},
|
12345 |
+
{
|
12346 |
+
"epoch": 1.6507490636704119,
|
12347 |
+
"grad_norm": 0.0,
|
12348 |
+
"learning_rate": 3.2437562983055026e-06,
|
12349 |
+
"loss": 1.1326,
|
12350 |
+
"step": 1763
|
12351 |
+
},
|
12352 |
+
{
|
12353 |
+
"epoch": 1.651685393258427,
|
12354 |
+
"grad_norm": 0.0,
|
12355 |
+
"learning_rate": 3.2268697431100995e-06,
|
12356 |
+
"loss": 1.1575,
|
12357 |
+
"step": 1764
|
12358 |
+
},
|
12359 |
+
{
|
12360 |
+
"epoch": 1.652621722846442,
|
12361 |
+
"grad_norm": 0.0,
|
12362 |
+
"learning_rate": 3.2100233993601004e-06,
|
12363 |
+
"loss": 1.1771,
|
12364 |
+
"step": 1765
|
12365 |
+
},
|
12366 |
+
{
|
12367 |
+
"epoch": 1.653558052434457,
|
12368 |
+
"grad_norm": 0.0,
|
12369 |
+
"learning_rate": 3.1932173074424777e-06,
|
12370 |
+
"loss": 1.1624,
|
12371 |
+
"step": 1766
|
12372 |
+
},
|
12373 |
+
{
|
12374 |
+
"epoch": 1.654494382022472,
|
12375 |
+
"grad_norm": 0.0,
|
12376 |
+
"learning_rate": 3.1764515076477e-06,
|
12377 |
+
"loss": 1.122,
|
12378 |
+
"step": 1767
|
12379 |
+
},
|
12380 |
+
{
|
12381 |
+
"epoch": 1.655430711610487,
|
12382 |
+
"grad_norm": 0.0,
|
12383 |
+
"learning_rate": 3.1597260401696374e-06,
|
12384 |
+
"loss": 1.1395,
|
12385 |
+
"step": 1768
|
12386 |
+
},
|
12387 |
+
{
|
12388 |
+
"epoch": 1.6563670411985019,
|
12389 |
+
"grad_norm": 0.0,
|
12390 |
+
"learning_rate": 3.1430409451054624e-06,
|
12391 |
+
"loss": 1.0973,
|
12392 |
+
"step": 1769
|
12393 |
+
},
|
12394 |
+
{
|
12395 |
+
"epoch": 1.6573033707865168,
|
12396 |
+
"grad_norm": 0.0,
|
12397 |
+
"learning_rate": 3.1263962624555776e-06,
|
12398 |
+
"loss": 1.114,
|
12399 |
+
"step": 1770
|
12400 |
+
},
|
12401 |
+
{
|
12402 |
+
"epoch": 1.6582397003745317,
|
12403 |
+
"grad_norm": 0.0,
|
12404 |
+
"learning_rate": 3.1097920321234866e-06,
|
12405 |
+
"loss": 1.1375,
|
12406 |
+
"step": 1771
|
12407 |
+
},
|
12408 |
+
{
|
12409 |
+
"epoch": 1.6591760299625467,
|
12410 |
+
"grad_norm": 0.0,
|
12411 |
+
"learning_rate": 3.0932282939157133e-06,
|
12412 |
+
"loss": 1.1368,
|
12413 |
+
"step": 1772
|
12414 |
+
},
|
12415 |
+
{
|
12416 |
+
"epoch": 1.6601123595505618,
|
12417 |
+
"grad_norm": 0.0,
|
12418 |
+
"learning_rate": 3.0767050875417246e-06,
|
12419 |
+
"loss": 1.1765,
|
12420 |
+
"step": 1773
|
12421 |
+
},
|
12422 |
+
{
|
12423 |
+
"epoch": 1.6610486891385767,
|
12424 |
+
"grad_norm": 0.0,
|
12425 |
+
"learning_rate": 3.060222452613797e-06,
|
12426 |
+
"loss": 1.1724,
|
12427 |
+
"step": 1774
|
12428 |
+
},
|
12429 |
+
{
|
12430 |
+
"epoch": 1.6619850187265919,
|
12431 |
+
"grad_norm": 0.0,
|
12432 |
+
"learning_rate": 3.0437804286469474e-06,
|
12433 |
+
"loss": 1.101,
|
12434 |
+
"step": 1775
|
12435 |
+
},
|
12436 |
+
{
|
12437 |
+
"epoch": 1.6629213483146068,
|
12438 |
+
"grad_norm": 0.0,
|
12439 |
+
"learning_rate": 3.0273790550588412e-06,
|
12440 |
+
"loss": 1.177,
|
12441 |
+
"step": 1776
|
12442 |
+
},
|
12443 |
+
{
|
12444 |
+
"epoch": 1.6638576779026217,
|
12445 |
+
"grad_norm": 0.0,
|
12446 |
+
"learning_rate": 3.0110183711696894e-06,
|
12447 |
+
"loss": 1.1187,
|
12448 |
+
"step": 1777
|
12449 |
+
},
|
12450 |
+
{
|
12451 |
+
"epoch": 1.6647940074906367,
|
12452 |
+
"grad_norm": 0.0,
|
12453 |
+
"learning_rate": 2.9946984162021377e-06,
|
12454 |
+
"loss": 1.1256,
|
12455 |
+
"step": 1778
|
12456 |
+
},
|
12457 |
+
{
|
12458 |
+
"epoch": 1.6657303370786516,
|
12459 |
+
"grad_norm": 0.0,
|
12460 |
+
"learning_rate": 2.978419229281213e-06,
|
12461 |
+
"loss": 1.1114,
|
12462 |
+
"step": 1779
|
12463 |
+
},
|
12464 |
+
{
|
12465 |
+
"epoch": 1.6666666666666665,
|
12466 |
+
"grad_norm": 0.0,
|
12467 |
+
"learning_rate": 2.962180849434191e-06,
|
12468 |
+
"loss": 1.1562,
|
12469 |
+
"step": 1780
|
12470 |
+
},
|
12471 |
+
{
|
12472 |
+
"epoch": 1.6676029962546817,
|
12473 |
+
"grad_norm": 0.0,
|
12474 |
+
"learning_rate": 2.945983315590513e-06,
|
12475 |
+
"loss": 1.1561,
|
12476 |
+
"step": 1781
|
12477 |
+
},
|
12478 |
+
{
|
12479 |
+
"epoch": 1.6685393258426966,
|
12480 |
+
"grad_norm": 0.0,
|
12481 |
+
"learning_rate": 2.9298266665817167e-06,
|
12482 |
+
"loss": 1.101,
|
12483 |
+
"step": 1782
|
12484 |
+
},
|
12485 |
+
{
|
12486 |
+
"epoch": 1.6694756554307117,
|
12487 |
+
"grad_norm": 0.0,
|
12488 |
+
"learning_rate": 2.9137109411413058e-06,
|
12489 |
+
"loss": 1.1677,
|
12490 |
+
"step": 1783
|
12491 |
+
},
|
12492 |
+
{
|
12493 |
+
"epoch": 1.6704119850187267,
|
12494 |
+
"grad_norm": 0.0,
|
12495 |
+
"learning_rate": 2.8976361779046793e-06,
|
12496 |
+
"loss": 1.1426,
|
12497 |
+
"step": 1784
|
12498 |
+
},
|
12499 |
+
{
|
12500 |
+
"epoch": 1.6713483146067416,
|
12501 |
+
"grad_norm": 0.0,
|
12502 |
+
"learning_rate": 2.8816024154090437e-06,
|
12503 |
+
"loss": 1.1294,
|
12504 |
+
"step": 1785
|
12505 |
+
},
|
12506 |
+
{
|
12507 |
+
"epoch": 1.6722846441947565,
|
12508 |
+
"grad_norm": 0.0,
|
12509 |
+
"learning_rate": 2.865609692093296e-06,
|
12510 |
+
"loss": 1.1024,
|
12511 |
+
"step": 1786
|
12512 |
+
},
|
12513 |
+
{
|
12514 |
+
"epoch": 1.6732209737827715,
|
12515 |
+
"grad_norm": 0.0,
|
12516 |
+
"learning_rate": 2.849658046297965e-06,
|
12517 |
+
"loss": 1.1059,
|
12518 |
+
"step": 1787
|
12519 |
+
},
|
12520 |
+
{
|
12521 |
+
"epoch": 1.6741573033707864,
|
12522 |
+
"grad_norm": 0.0,
|
12523 |
+
"learning_rate": 2.833747516265082e-06,
|
12524 |
+
"loss": 1.1023,
|
12525 |
+
"step": 1788
|
12526 |
+
},
|
12527 |
+
{
|
12528 |
+
"epoch": 1.6750936329588015,
|
12529 |
+
"grad_norm": 0.0,
|
12530 |
+
"learning_rate": 2.8178781401381215e-06,
|
12531 |
+
"loss": 1.1259,
|
12532 |
+
"step": 1789
|
12533 |
+
},
|
12534 |
+
{
|
12535 |
+
"epoch": 1.6760299625468165,
|
12536 |
+
"grad_norm": 0.0,
|
12537 |
+
"learning_rate": 2.802049955961901e-06,
|
12538 |
+
"loss": 1.0863,
|
12539 |
+
"step": 1790
|
12540 |
+
},
|
12541 |
+
{
|
12542 |
+
"epoch": 1.6769662921348316,
|
12543 |
+
"grad_norm": 0.0,
|
12544 |
+
"learning_rate": 2.78626300168247e-06,
|
12545 |
+
"loss": 1.1555,
|
12546 |
+
"step": 1791
|
12547 |
+
},
|
12548 |
+
{
|
12549 |
+
"epoch": 1.6779026217228465,
|
12550 |
+
"grad_norm": 0.0,
|
12551 |
+
"learning_rate": 2.770517315147039e-06,
|
12552 |
+
"loss": 1.0907,
|
12553 |
+
"step": 1792
|
12554 |
+
},
|
12555 |
+
{
|
12556 |
+
"epoch": 1.6788389513108615,
|
12557 |
+
"grad_norm": 0.0,
|
12558 |
+
"learning_rate": 2.7548129341038966e-06,
|
12559 |
+
"loss": 1.0777,
|
12560 |
+
"step": 1793
|
12561 |
+
},
|
12562 |
+
{
|
12563 |
+
"epoch": 1.6797752808988764,
|
12564 |
+
"grad_norm": 0.0,
|
12565 |
+
"learning_rate": 2.7391498962022934e-06,
|
12566 |
+
"loss": 1.1522,
|
12567 |
+
"step": 1794
|
12568 |
+
},
|
12569 |
+
{
|
12570 |
+
"epoch": 1.6807116104868913,
|
12571 |
+
"grad_norm": 0.0,
|
12572 |
+
"learning_rate": 2.7235282389923677e-06,
|
12573 |
+
"loss": 1.1227,
|
12574 |
+
"step": 1795
|
12575 |
+
},
|
12576 |
+
{
|
12577 |
+
"epoch": 1.6816479400749063,
|
12578 |
+
"grad_norm": 0.0,
|
12579 |
+
"learning_rate": 2.70794799992506e-06,
|
12580 |
+
"loss": 1.1192,
|
12581 |
+
"step": 1796
|
12582 |
+
},
|
12583 |
+
{
|
12584 |
+
"epoch": 1.6825842696629212,
|
12585 |
+
"grad_norm": 0.0,
|
12586 |
+
"learning_rate": 2.6924092163520098e-06,
|
12587 |
+
"loss": 1.0938,
|
12588 |
+
"step": 1797
|
12589 |
+
},
|
12590 |
+
{
|
12591 |
+
"epoch": 1.6835205992509363,
|
12592 |
+
"grad_norm": 0.0,
|
12593 |
+
"learning_rate": 2.676911925525467e-06,
|
12594 |
+
"loss": 1.1367,
|
12595 |
+
"step": 1798
|
12596 |
+
},
|
12597 |
+
{
|
12598 |
+
"epoch": 1.6844569288389513,
|
12599 |
+
"grad_norm": 0.0,
|
12600 |
+
"learning_rate": 2.6614561645982196e-06,
|
12601 |
+
"loss": 1.0976,
|
12602 |
+
"step": 1799
|
12603 |
+
},
|
12604 |
+
{
|
12605 |
+
"epoch": 1.6853932584269664,
|
12606 |
+
"grad_norm": 0.0,
|
12607 |
+
"learning_rate": 2.6460419706234897e-06,
|
12608 |
+
"loss": 1.1064,
|
12609 |
+
"step": 1800
|
12610 |
}
|
12611 |
],
|
12612 |
"logging_steps": 1,
|
|
|
12626 |
"attributes": {}
|
12627 |
}
|
12628 |
},
|
12629 |
+
"total_flos": 5.889731177333391e+18,
|
12630 |
"train_batch_size": 8,
|
12631 |
"trial_name": null,
|
12632 |
"trial_params": null
|