Training in progress, step 738, checkpoint
Browse files
last-checkpoint/adapter_model.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 83945296
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:3a3937bb644e8be3db53724ff37a4cf4df56813bf6b4a93d38ec325890a084a4
|
3 |
size 83945296
|
last-checkpoint/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 43123028
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:bd82f4b1baa9236bf9259ffc67bbc17c841880be954604873e130ffce8157a86
|
3 |
size 43123028
|
last-checkpoint/rng_state.pth
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 14244
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:7bb2c147343e6526aaa232aefc3fc703d6b9ddcf5d44d0d12953eb06e064afe1
|
3 |
size 14244
|
last-checkpoint/scheduler.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 1064
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:3482c8bfdc1affb8191f8e392c3195155db922305404a13a0a905ef4dc9ec8d7
|
3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
@@ -1,9 +1,9 @@
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
-
"epoch": 0.
|
5 |
"eval_steps": 500,
|
6 |
-
"global_step":
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
@@ -3451,6 +3451,1728 @@
|
|
3451 |
"learning_rate": 5.032122403295977e-05,
|
3452 |
"loss": 2.3439,
|
3453 |
"step": 492
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
3454 |
}
|
3455 |
],
|
3456 |
"logging_steps": 1,
|
@@ -3470,7 +5192,7 @@
|
|
3470 |
"attributes": {}
|
3471 |
}
|
3472 |
},
|
3473 |
-
"total_flos":
|
3474 |
"train_batch_size": 4,
|
3475 |
"trial_name": null,
|
3476 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 0.7513362178671418,
|
5 |
"eval_steps": 500,
|
6 |
+
"global_step": 738,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
3451 |
"learning_rate": 5.032122403295977e-05,
|
3452 |
"loss": 2.3439,
|
3453 |
"step": 492
|
3454 |
+
},
|
3455 |
+
{
|
3456 |
+
"epoch": 0.5019088826673453,
|
3457 |
+
"grad_norm": 8.969354629516602,
|
3458 |
+
"learning_rate": 5.0160612845131414e-05,
|
3459 |
+
"loss": 2.5224,
|
3460 |
+
"step": 493
|
3461 |
+
},
|
3462 |
+
{
|
3463 |
+
"epoch": 0.5029269534232629,
|
3464 |
+
"grad_norm": 7.22929573059082,
|
3465 |
+
"learning_rate": 5e-05,
|
3466 |
+
"loss": 1.6864,
|
3467 |
+
"step": 494
|
3468 |
+
},
|
3469 |
+
{
|
3470 |
+
"epoch": 0.5039450241791804,
|
3471 |
+
"grad_norm": 7.851796627044678,
|
3472 |
+
"learning_rate": 4.9839387154868584e-05,
|
3473 |
+
"loss": 2.0388,
|
3474 |
+
"step": 495
|
3475 |
+
},
|
3476 |
+
{
|
3477 |
+
"epoch": 0.504963094935098,
|
3478 |
+
"grad_norm": 9.29887580871582,
|
3479 |
+
"learning_rate": 4.967877596704025e-05,
|
3480 |
+
"loss": 2.6501,
|
3481 |
+
"step": 496
|
3482 |
+
},
|
3483 |
+
{
|
3484 |
+
"epoch": 0.5059811656910155,
|
3485 |
+
"grad_norm": 8.79398250579834,
|
3486 |
+
"learning_rate": 4.951816809380097e-05,
|
3487 |
+
"loss": 2.5759,
|
3488 |
+
"step": 497
|
3489 |
+
},
|
3490 |
+
{
|
3491 |
+
"epoch": 0.506999236446933,
|
3492 |
+
"grad_norm": 10.286650657653809,
|
3493 |
+
"learning_rate": 4.9357565192402525e-05,
|
3494 |
+
"loss": 3.1672,
|
3495 |
+
"step": 498
|
3496 |
+
},
|
3497 |
+
{
|
3498 |
+
"epoch": 0.5080173072028505,
|
3499 |
+
"grad_norm": 8.799942016601562,
|
3500 |
+
"learning_rate": 4.919696892004539e-05,
|
3501 |
+
"loss": 1.9062,
|
3502 |
+
"step": 499
|
3503 |
+
},
|
3504 |
+
{
|
3505 |
+
"epoch": 0.5090353779587682,
|
3506 |
+
"grad_norm": 8.252535820007324,
|
3507 |
+
"learning_rate": 4.903638093386167e-05,
|
3508 |
+
"loss": 1.6945,
|
3509 |
+
"step": 500
|
3510 |
+
},
|
3511 |
+
{
|
3512 |
+
"epoch": 0.5100534487146857,
|
3513 |
+
"grad_norm": 5.307441234588623,
|
3514 |
+
"learning_rate": 4.887580289089787e-05,
|
3515 |
+
"loss": 1.7627,
|
3516 |
+
"step": 501
|
3517 |
+
},
|
3518 |
+
{
|
3519 |
+
"epoch": 0.5110715194706033,
|
3520 |
+
"grad_norm": 7.124861717224121,
|
3521 |
+
"learning_rate": 4.8715236448098016e-05,
|
3522 |
+
"loss": 2.5286,
|
3523 |
+
"step": 502
|
3524 |
+
},
|
3525 |
+
{
|
3526 |
+
"epoch": 0.5120895902265208,
|
3527 |
+
"grad_norm": 10.368796348571777,
|
3528 |
+
"learning_rate": 4.855468326228638e-05,
|
3529 |
+
"loss": 3.6972,
|
3530 |
+
"step": 503
|
3531 |
+
},
|
3532 |
+
{
|
3533 |
+
"epoch": 0.5131076609824383,
|
3534 |
+
"grad_norm": 10.741311073303223,
|
3535 |
+
"learning_rate": 4.8394144990150404e-05,
|
3536 |
+
"loss": 4.1789,
|
3537 |
+
"step": 504
|
3538 |
+
},
|
3539 |
+
{
|
3540 |
+
"epoch": 0.5141257317383559,
|
3541 |
+
"grad_norm": 10.585883140563965,
|
3542 |
+
"learning_rate": 4.8233623288223704e-05,
|
3543 |
+
"loss": 3.5806,
|
3544 |
+
"step": 505
|
3545 |
+
},
|
3546 |
+
{
|
3547 |
+
"epoch": 0.5151438024942734,
|
3548 |
+
"grad_norm": 13.6077880859375,
|
3549 |
+
"learning_rate": 4.807311981286888e-05,
|
3550 |
+
"loss": 3.5391,
|
3551 |
+
"step": 506
|
3552 |
+
},
|
3553 |
+
{
|
3554 |
+
"epoch": 0.5161618732501909,
|
3555 |
+
"grad_norm": 16.34162712097168,
|
3556 |
+
"learning_rate": 4.7912636220260473e-05,
|
3557 |
+
"loss": 3.6142,
|
3558 |
+
"step": 507
|
3559 |
+
},
|
3560 |
+
{
|
3561 |
+
"epoch": 0.5171799440061084,
|
3562 |
+
"grad_norm": 11.734464645385742,
|
3563 |
+
"learning_rate": 4.775217416636786e-05,
|
3564 |
+
"loss": 2.6898,
|
3565 |
+
"step": 508
|
3566 |
+
},
|
3567 |
+
{
|
3568 |
+
"epoch": 0.518198014762026,
|
3569 |
+
"grad_norm": 7.3565287590026855,
|
3570 |
+
"learning_rate": 4.759173530693814e-05,
|
3571 |
+
"loss": 2.0978,
|
3572 |
+
"step": 509
|
3573 |
+
},
|
3574 |
+
{
|
3575 |
+
"epoch": 0.5192160855179435,
|
3576 |
+
"grad_norm": 5.852792739868164,
|
3577 |
+
"learning_rate": 4.7431321297479135e-05,
|
3578 |
+
"loss": 1.4297,
|
3579 |
+
"step": 510
|
3580 |
+
},
|
3581 |
+
{
|
3582 |
+
"epoch": 0.520234156273861,
|
3583 |
+
"grad_norm": 6.781408786773682,
|
3584 |
+
"learning_rate": 4.727093379324222e-05,
|
3585 |
+
"loss": 1.5329,
|
3586 |
+
"step": 511
|
3587 |
+
},
|
3588 |
+
{
|
3589 |
+
"epoch": 0.5212522270297786,
|
3590 |
+
"grad_norm": 11.24429702758789,
|
3591 |
+
"learning_rate": 4.711057444920522e-05,
|
3592 |
+
"loss": 1.6744,
|
3593 |
+
"step": 512
|
3594 |
+
},
|
3595 |
+
{
|
3596 |
+
"epoch": 0.5222702977856961,
|
3597 |
+
"grad_norm": 7.679388046264648,
|
3598 |
+
"learning_rate": 4.695024492005548e-05,
|
3599 |
+
"loss": 2.2356,
|
3600 |
+
"step": 513
|
3601 |
+
},
|
3602 |
+
{
|
3603 |
+
"epoch": 0.5232883685416136,
|
3604 |
+
"grad_norm": 5.93134069442749,
|
3605 |
+
"learning_rate": 4.6789946860172634e-05,
|
3606 |
+
"loss": 1.2665,
|
3607 |
+
"step": 514
|
3608 |
+
},
|
3609 |
+
{
|
3610 |
+
"epoch": 0.5243064392975312,
|
3611 |
+
"grad_norm": 6.789477348327637,
|
3612 |
+
"learning_rate": 4.6629681923611603e-05,
|
3613 |
+
"loss": 1.6466,
|
3614 |
+
"step": 515
|
3615 |
+
},
|
3616 |
+
{
|
3617 |
+
"epoch": 0.5253245100534487,
|
3618 |
+
"grad_norm": 8.315037727355957,
|
3619 |
+
"learning_rate": 4.646945176408555e-05,
|
3620 |
+
"loss": 2.065,
|
3621 |
+
"step": 516
|
3622 |
+
},
|
3623 |
+
{
|
3624 |
+
"epoch": 0.5263425808093662,
|
3625 |
+
"grad_norm": 5.443754196166992,
|
3626 |
+
"learning_rate": 4.630925803494877e-05,
|
3627 |
+
"loss": 1.0138,
|
3628 |
+
"step": 517
|
3629 |
+
},
|
3630 |
+
{
|
3631 |
+
"epoch": 0.5273606515652838,
|
3632 |
+
"grad_norm": 6.596680641174316,
|
3633 |
+
"learning_rate": 4.6149102389179635e-05,
|
3634 |
+
"loss": 1.8229,
|
3635 |
+
"step": 518
|
3636 |
+
},
|
3637 |
+
{
|
3638 |
+
"epoch": 0.5283787223212013,
|
3639 |
+
"grad_norm": 5.735509872436523,
|
3640 |
+
"learning_rate": 4.598898647936354e-05,
|
3641 |
+
"loss": 1.5016,
|
3642 |
+
"step": 519
|
3643 |
+
},
|
3644 |
+
{
|
3645 |
+
"epoch": 0.5293967930771188,
|
3646 |
+
"grad_norm": 7.154899597167969,
|
3647 |
+
"learning_rate": 4.58289119576759e-05,
|
3648 |
+
"loss": 1.7249,
|
3649 |
+
"step": 520
|
3650 |
+
},
|
3651 |
+
{
|
3652 |
+
"epoch": 0.5304148638330364,
|
3653 |
+
"grad_norm": 5.887238502502441,
|
3654 |
+
"learning_rate": 4.566888047586507e-05,
|
3655 |
+
"loss": 1.3531,
|
3656 |
+
"step": 521
|
3657 |
+
},
|
3658 |
+
{
|
3659 |
+
"epoch": 0.5314329345889539,
|
3660 |
+
"grad_norm": 7.944952964782715,
|
3661 |
+
"learning_rate": 4.55088936852352e-05,
|
3662 |
+
"loss": 2.0604,
|
3663 |
+
"step": 522
|
3664 |
+
},
|
3665 |
+
{
|
3666 |
+
"epoch": 0.5324510053448714,
|
3667 |
+
"grad_norm": 8.235894203186035,
|
3668 |
+
"learning_rate": 4.5348953236629395e-05,
|
3669 |
+
"loss": 1.782,
|
3670 |
+
"step": 523
|
3671 |
+
},
|
3672 |
+
{
|
3673 |
+
"epoch": 0.533469076100789,
|
3674 |
+
"grad_norm": 9.824324607849121,
|
3675 |
+
"learning_rate": 4.518906078041252e-05,
|
3676 |
+
"loss": 3.1078,
|
3677 |
+
"step": 524
|
3678 |
+
},
|
3679 |
+
{
|
3680 |
+
"epoch": 0.5344871468567065,
|
3681 |
+
"grad_norm": 8.053499221801758,
|
3682 |
+
"learning_rate": 4.502921796645424e-05,
|
3683 |
+
"loss": 2.5225,
|
3684 |
+
"step": 525
|
3685 |
+
},
|
3686 |
+
{
|
3687 |
+
"epoch": 0.535505217612624,
|
3688 |
+
"grad_norm": 9.53549861907959,
|
3689 |
+
"learning_rate": 4.486942644411197e-05,
|
3690 |
+
"loss": 3.0847,
|
3691 |
+
"step": 526
|
3692 |
+
},
|
3693 |
+
{
|
3694 |
+
"epoch": 0.5365232883685416,
|
3695 |
+
"grad_norm": 8.427640914916992,
|
3696 |
+
"learning_rate": 4.4709687862213866e-05,
|
3697 |
+
"loss": 2.1704,
|
3698 |
+
"step": 527
|
3699 |
+
},
|
3700 |
+
{
|
3701 |
+
"epoch": 0.5375413591244591,
|
3702 |
+
"grad_norm": 7.989354610443115,
|
3703 |
+
"learning_rate": 4.4550003869041845e-05,
|
3704 |
+
"loss": 2.3719,
|
3705 |
+
"step": 528
|
3706 |
+
},
|
3707 |
+
{
|
3708 |
+
"epoch": 0.5385594298803766,
|
3709 |
+
"grad_norm": 7.53865909576416,
|
3710 |
+
"learning_rate": 4.439037611231448e-05,
|
3711 |
+
"loss": 2.4358,
|
3712 |
+
"step": 529
|
3713 |
+
},
|
3714 |
+
{
|
3715 |
+
"epoch": 0.5395775006362942,
|
3716 |
+
"grad_norm": 9.102818489074707,
|
3717 |
+
"learning_rate": 4.423080623917012e-05,
|
3718 |
+
"loss": 3.0774,
|
3719 |
+
"step": 530
|
3720 |
+
},
|
3721 |
+
{
|
3722 |
+
"epoch": 0.5405955713922117,
|
3723 |
+
"grad_norm": 10.17009162902832,
|
3724 |
+
"learning_rate": 4.407129589614979e-05,
|
3725 |
+
"loss": 2.719,
|
3726 |
+
"step": 531
|
3727 |
+
},
|
3728 |
+
{
|
3729 |
+
"epoch": 0.5416136421481293,
|
3730 |
+
"grad_norm": 8.132767677307129,
|
3731 |
+
"learning_rate": 4.3911846729180335e-05,
|
3732 |
+
"loss": 2.6276,
|
3733 |
+
"step": 532
|
3734 |
+
},
|
3735 |
+
{
|
3736 |
+
"epoch": 0.5426317129040469,
|
3737 |
+
"grad_norm": 8.669943809509277,
|
3738 |
+
"learning_rate": 4.3752460383557195e-05,
|
3739 |
+
"loss": 2.2211,
|
3740 |
+
"step": 533
|
3741 |
+
},
|
3742 |
+
{
|
3743 |
+
"epoch": 0.5436497836599644,
|
3744 |
+
"grad_norm": 8.190427780151367,
|
3745 |
+
"learning_rate": 4.359313850392772e-05,
|
3746 |
+
"loss": 2.2451,
|
3747 |
+
"step": 534
|
3748 |
+
},
|
3749 |
+
{
|
3750 |
+
"epoch": 0.544667854415882,
|
3751 |
+
"grad_norm": 7.185608386993408,
|
3752 |
+
"learning_rate": 4.3433882734274e-05,
|
3753 |
+
"loss": 1.938,
|
3754 |
+
"step": 535
|
3755 |
+
},
|
3756 |
+
{
|
3757 |
+
"epoch": 0.5456859251717995,
|
3758 |
+
"grad_norm": 9.735365867614746,
|
3759 |
+
"learning_rate": 4.327469471789597e-05,
|
3760 |
+
"loss": 3.3738,
|
3761 |
+
"step": 536
|
3762 |
+
},
|
3763 |
+
{
|
3764 |
+
"epoch": 0.546703995927717,
|
3765 |
+
"grad_norm": 9.06591796875,
|
3766 |
+
"learning_rate": 4.311557609739442e-05,
|
3767 |
+
"loss": 3.4894,
|
3768 |
+
"step": 537
|
3769 |
+
},
|
3770 |
+
{
|
3771 |
+
"epoch": 0.5477220666836345,
|
3772 |
+
"grad_norm": 8.038829803466797,
|
3773 |
+
"learning_rate": 4.295652851465412e-05,
|
3774 |
+
"loss": 2.6487,
|
3775 |
+
"step": 538
|
3776 |
+
},
|
3777 |
+
{
|
3778 |
+
"epoch": 0.5487401374395521,
|
3779 |
+
"grad_norm": 7.375051498413086,
|
3780 |
+
"learning_rate": 4.27975536108268e-05,
|
3781 |
+
"loss": 2.4853,
|
3782 |
+
"step": 539
|
3783 |
+
},
|
3784 |
+
{
|
3785 |
+
"epoch": 0.5497582081954696,
|
3786 |
+
"grad_norm": 9.910839080810547,
|
3787 |
+
"learning_rate": 4.2638653026314224e-05,
|
3788 |
+
"loss": 3.1606,
|
3789 |
+
"step": 540
|
3790 |
+
},
|
3791 |
+
{
|
3792 |
+
"epoch": 0.5507762789513871,
|
3793 |
+
"grad_norm": 7.77678918838501,
|
3794 |
+
"learning_rate": 4.24798284007513e-05,
|
3795 |
+
"loss": 2.33,
|
3796 |
+
"step": 541
|
3797 |
+
},
|
3798 |
+
{
|
3799 |
+
"epoch": 0.5517943497073047,
|
3800 |
+
"grad_norm": 7.377612113952637,
|
3801 |
+
"learning_rate": 4.232108137298919e-05,
|
3802 |
+
"loss": 2.299,
|
3803 |
+
"step": 542
|
3804 |
+
},
|
3805 |
+
{
|
3806 |
+
"epoch": 0.5528124204632222,
|
3807 |
+
"grad_norm": 9.510624885559082,
|
3808 |
+
"learning_rate": 4.216241358107831e-05,
|
3809 |
+
"loss": 2.8467,
|
3810 |
+
"step": 543
|
3811 |
+
},
|
3812 |
+
{
|
3813 |
+
"epoch": 0.5538304912191397,
|
3814 |
+
"grad_norm": 6.834048748016357,
|
3815 |
+
"learning_rate": 4.200382666225141e-05,
|
3816 |
+
"loss": 2.0166,
|
3817 |
+
"step": 544
|
3818 |
+
},
|
3819 |
+
{
|
3820 |
+
"epoch": 0.5548485619750573,
|
3821 |
+
"grad_norm": 8.245951652526855,
|
3822 |
+
"learning_rate": 4.1845322252906864e-05,
|
3823 |
+
"loss": 2.672,
|
3824 |
+
"step": 545
|
3825 |
+
},
|
3826 |
+
{
|
3827 |
+
"epoch": 0.5558666327309748,
|
3828 |
+
"grad_norm": 7.539649963378906,
|
3829 |
+
"learning_rate": 4.16869019885916e-05,
|
3830 |
+
"loss": 2.3618,
|
3831 |
+
"step": 546
|
3832 |
+
},
|
3833 |
+
{
|
3834 |
+
"epoch": 0.5568847034868923,
|
3835 |
+
"grad_norm": 7.983175754547119,
|
3836 |
+
"learning_rate": 4.152856750398426e-05,
|
3837 |
+
"loss": 2.2049,
|
3838 |
+
"step": 547
|
3839 |
+
},
|
3840 |
+
{
|
3841 |
+
"epoch": 0.5579027742428099,
|
3842 |
+
"grad_norm": 8.641951560974121,
|
3843 |
+
"learning_rate": 4.1370320432878404e-05,
|
3844 |
+
"loss": 2.2235,
|
3845 |
+
"step": 548
|
3846 |
+
},
|
3847 |
+
{
|
3848 |
+
"epoch": 0.5589208449987274,
|
3849 |
+
"grad_norm": 7.9181437492370605,
|
3850 |
+
"learning_rate": 4.1212162408165595e-05,
|
3851 |
+
"loss": 1.9295,
|
3852 |
+
"step": 549
|
3853 |
+
},
|
3854 |
+
{
|
3855 |
+
"epoch": 0.5599389157546449,
|
3856 |
+
"grad_norm": 10.45153522491455,
|
3857 |
+
"learning_rate": 4.105409506181854e-05,
|
3858 |
+
"loss": 2.1553,
|
3859 |
+
"step": 550
|
3860 |
+
},
|
3861 |
+
{
|
3862 |
+
"epoch": 0.5609569865105625,
|
3863 |
+
"grad_norm": 6.951171398162842,
|
3864 |
+
"learning_rate": 4.0896120024874286e-05,
|
3865 |
+
"loss": 2.5913,
|
3866 |
+
"step": 551
|
3867 |
+
},
|
3868 |
+
{
|
3869 |
+
"epoch": 0.56197505726648,
|
3870 |
+
"grad_norm": 9.614657402038574,
|
3871 |
+
"learning_rate": 4.073823892741735e-05,
|
3872 |
+
"loss": 4.2435,
|
3873 |
+
"step": 552
|
3874 |
+
},
|
3875 |
+
{
|
3876 |
+
"epoch": 0.5629931280223975,
|
3877 |
+
"grad_norm": 9.35623550415039,
|
3878 |
+
"learning_rate": 4.0580453398563e-05,
|
3879 |
+
"loss": 3.7123,
|
3880 |
+
"step": 553
|
3881 |
+
},
|
3882 |
+
{
|
3883 |
+
"epoch": 0.5640111987783151,
|
3884 |
+
"grad_norm": 10.756424903869629,
|
3885 |
+
"learning_rate": 4.042276506644024e-05,
|
3886 |
+
"loss": 3.6713,
|
3887 |
+
"step": 554
|
3888 |
+
},
|
3889 |
+
{
|
3890 |
+
"epoch": 0.5650292695342326,
|
3891 |
+
"grad_norm": 9.823023796081543,
|
3892 |
+
"learning_rate": 4.0265175558175265e-05,
|
3893 |
+
"loss": 3.7602,
|
3894 |
+
"step": 555
|
3895 |
+
},
|
3896 |
+
{
|
3897 |
+
"epoch": 0.5660473402901501,
|
3898 |
+
"grad_norm": 13.360715866088867,
|
3899 |
+
"learning_rate": 4.0107686499874465e-05,
|
3900 |
+
"loss": 3.269,
|
3901 |
+
"step": 556
|
3902 |
+
},
|
3903 |
+
{
|
3904 |
+
"epoch": 0.5670654110460677,
|
3905 |
+
"grad_norm": 14.194052696228027,
|
3906 |
+
"learning_rate": 3.9950299516607766e-05,
|
3907 |
+
"loss": 4.3906,
|
3908 |
+
"step": 557
|
3909 |
+
},
|
3910 |
+
{
|
3911 |
+
"epoch": 0.5680834818019852,
|
3912 |
+
"grad_norm": 16.591251373291016,
|
3913 |
+
"learning_rate": 3.979301623239177e-05,
|
3914 |
+
"loss": 4.4802,
|
3915 |
+
"step": 558
|
3916 |
+
},
|
3917 |
+
{
|
3918 |
+
"epoch": 0.5691015525579027,
|
3919 |
+
"grad_norm": 6.6096720695495605,
|
3920 |
+
"learning_rate": 3.9635838270173107e-05,
|
3921 |
+
"loss": 1.6842,
|
3922 |
+
"step": 559
|
3923 |
+
},
|
3924 |
+
{
|
3925 |
+
"epoch": 0.5701196233138203,
|
3926 |
+
"grad_norm": 6.252510070800781,
|
3927 |
+
"learning_rate": 3.94787672518116e-05,
|
3928 |
+
"loss": 1.6248,
|
3929 |
+
"step": 560
|
3930 |
+
},
|
3931 |
+
{
|
3932 |
+
"epoch": 0.5711376940697378,
|
3933 |
+
"grad_norm": 7.445550441741943,
|
3934 |
+
"learning_rate": 3.9321804798063565e-05,
|
3935 |
+
"loss": 1.7234,
|
3936 |
+
"step": 561
|
3937 |
+
},
|
3938 |
+
{
|
3939 |
+
"epoch": 0.5721557648256553,
|
3940 |
+
"grad_norm": 5.321173191070557,
|
3941 |
+
"learning_rate": 3.9164952528565057e-05,
|
3942 |
+
"loss": 1.2454,
|
3943 |
+
"step": 562
|
3944 |
+
},
|
3945 |
+
{
|
3946 |
+
"epoch": 0.5731738355815729,
|
3947 |
+
"grad_norm": 4.566540241241455,
|
3948 |
+
"learning_rate": 3.900821206181521e-05,
|
3949 |
+
"loss": 1.0588,
|
3950 |
+
"step": 563
|
3951 |
+
},
|
3952 |
+
{
|
3953 |
+
"epoch": 0.5741919063374905,
|
3954 |
+
"grad_norm": 8.349088668823242,
|
3955 |
+
"learning_rate": 3.8851585015159536e-05,
|
3956 |
+
"loss": 1.5751,
|
3957 |
+
"step": 564
|
3958 |
+
},
|
3959 |
+
{
|
3960 |
+
"epoch": 0.575209977093408,
|
3961 |
+
"grad_norm": 6.328129291534424,
|
3962 |
+
"learning_rate": 3.8695073004773106e-05,
|
3963 |
+
"loss": 2.025,
|
3964 |
+
"step": 565
|
3965 |
+
},
|
3966 |
+
{
|
3967 |
+
"epoch": 0.5762280478493256,
|
3968 |
+
"grad_norm": 8.211170196533203,
|
3969 |
+
"learning_rate": 3.8538677645644096e-05,
|
3970 |
+
"loss": 1.1548,
|
3971 |
+
"step": 566
|
3972 |
+
},
|
3973 |
+
{
|
3974 |
+
"epoch": 0.5772461186052431,
|
3975 |
+
"grad_norm": 5.518578052520752,
|
3976 |
+
"learning_rate": 3.838240055155692e-05,
|
3977 |
+
"loss": 1.2809,
|
3978 |
+
"step": 567
|
3979 |
+
},
|
3980 |
+
{
|
3981 |
+
"epoch": 0.5782641893611606,
|
3982 |
+
"grad_norm": 6.383520603179932,
|
3983 |
+
"learning_rate": 3.822624333507571e-05,
|
3984 |
+
"loss": 1.8485,
|
3985 |
+
"step": 568
|
3986 |
+
},
|
3987 |
+
{
|
3988 |
+
"epoch": 0.5792822601170782,
|
3989 |
+
"grad_norm": 5.425829887390137,
|
3990 |
+
"learning_rate": 3.8070207607527584e-05,
|
3991 |
+
"loss": 1.4567,
|
3992 |
+
"step": 569
|
3993 |
+
},
|
3994 |
+
{
|
3995 |
+
"epoch": 0.5803003308729957,
|
3996 |
+
"grad_norm": 8.478185653686523,
|
3997 |
+
"learning_rate": 3.791429497898608e-05,
|
3998 |
+
"loss": 2.0052,
|
3999 |
+
"step": 570
|
4000 |
+
},
|
4001 |
+
{
|
4002 |
+
"epoch": 0.5813184016289132,
|
4003 |
+
"grad_norm": 8.863068580627441,
|
4004 |
+
"learning_rate": 3.775850705825454e-05,
|
4005 |
+
"loss": 2.2554,
|
4006 |
+
"step": 571
|
4007 |
+
},
|
4008 |
+
{
|
4009 |
+
"epoch": 0.5823364723848308,
|
4010 |
+
"grad_norm": 5.8295183181762695,
|
4011 |
+
"learning_rate": 3.7602845452849463e-05,
|
4012 |
+
"loss": 1.2544,
|
4013 |
+
"step": 572
|
4014 |
+
},
|
4015 |
+
{
|
4016 |
+
"epoch": 0.5833545431407483,
|
4017 |
+
"grad_norm": 8.446788787841797,
|
4018 |
+
"learning_rate": 3.7447311768983964e-05,
|
4019 |
+
"loss": 2.4702,
|
4020 |
+
"step": 573
|
4021 |
+
},
|
4022 |
+
{
|
4023 |
+
"epoch": 0.5843726138966658,
|
4024 |
+
"grad_norm": 7.7443766593933105,
|
4025 |
+
"learning_rate": 3.7291907611551195e-05,
|
4026 |
+
"loss": 2.0707,
|
4027 |
+
"step": 574
|
4028 |
+
},
|
4029 |
+
{
|
4030 |
+
"epoch": 0.5853906846525834,
|
4031 |
+
"grad_norm": 8.347147941589355,
|
4032 |
+
"learning_rate": 3.713663458410779e-05,
|
4033 |
+
"loss": 1.659,
|
4034 |
+
"step": 575
|
4035 |
+
},
|
4036 |
+
{
|
4037 |
+
"epoch": 0.5864087554085009,
|
4038 |
+
"grad_norm": 7.487883567810059,
|
4039 |
+
"learning_rate": 3.69814942888572e-05,
|
4040 |
+
"loss": 2.0328,
|
4041 |
+
"step": 576
|
4042 |
+
},
|
4043 |
+
{
|
4044 |
+
"epoch": 0.5874268261644184,
|
4045 |
+
"grad_norm": 7.8575286865234375,
|
4046 |
+
"learning_rate": 3.682648832663339e-05,
|
4047 |
+
"loss": 1.8928,
|
4048 |
+
"step": 577
|
4049 |
+
},
|
4050 |
+
{
|
4051 |
+
"epoch": 0.588444896920336,
|
4052 |
+
"grad_norm": 8.947505950927734,
|
4053 |
+
"learning_rate": 3.6671618296884146e-05,
|
4054 |
+
"loss": 1.6774,
|
4055 |
+
"step": 578
|
4056 |
+
},
|
4057 |
+
{
|
4058 |
+
"epoch": 0.5894629676762535,
|
4059 |
+
"grad_norm": 5.097304821014404,
|
4060 |
+
"learning_rate": 3.6516885797654594e-05,
|
4061 |
+
"loss": 1.3306,
|
4062 |
+
"step": 579
|
4063 |
+
},
|
4064 |
+
{
|
4065 |
+
"epoch": 0.590481038432171,
|
4066 |
+
"grad_norm": 6.418907642364502,
|
4067 |
+
"learning_rate": 3.636229242557075e-05,
|
4068 |
+
"loss": 1.9186,
|
4069 |
+
"step": 580
|
4070 |
+
},
|
4071 |
+
{
|
4072 |
+
"epoch": 0.5914991091880886,
|
4073 |
+
"grad_norm": 7.3138346672058105,
|
4074 |
+
"learning_rate": 3.620783977582305e-05,
|
4075 |
+
"loss": 2.4993,
|
4076 |
+
"step": 581
|
4077 |
+
},
|
4078 |
+
{
|
4079 |
+
"epoch": 0.5925171799440061,
|
4080 |
+
"grad_norm": 7.914095878601074,
|
4081 |
+
"learning_rate": 3.605352944214986e-05,
|
4082 |
+
"loss": 2.078,
|
4083 |
+
"step": 582
|
4084 |
+
},
|
4085 |
+
{
|
4086 |
+
"epoch": 0.5935352506999236,
|
4087 |
+
"grad_norm": 10.451981544494629,
|
4088 |
+
"learning_rate": 3.5899363016821e-05,
|
4089 |
+
"loss": 2.5348,
|
4090 |
+
"step": 583
|
4091 |
+
},
|
4092 |
+
{
|
4093 |
+
"epoch": 0.5945533214558412,
|
4094 |
+
"grad_norm": 6.191624164581299,
|
4095 |
+
"learning_rate": 3.5745342090621405e-05,
|
4096 |
+
"loss": 1.6607,
|
4097 |
+
"step": 584
|
4098 |
+
},
|
4099 |
+
{
|
4100 |
+
"epoch": 0.5955713922117587,
|
4101 |
+
"grad_norm": 7.947683811187744,
|
4102 |
+
"learning_rate": 3.559146825283465e-05,
|
4103 |
+
"loss": 2.4664,
|
4104 |
+
"step": 585
|
4105 |
+
},
|
4106 |
+
{
|
4107 |
+
"epoch": 0.5965894629676762,
|
4108 |
+
"grad_norm": 7.410199165344238,
|
4109 |
+
"learning_rate": 3.5437743091226565e-05,
|
4110 |
+
"loss": 2.0212,
|
4111 |
+
"step": 586
|
4112 |
+
},
|
4113 |
+
{
|
4114 |
+
"epoch": 0.5976075337235938,
|
4115 |
+
"grad_norm": 8.705409049987793,
|
4116 |
+
"learning_rate": 3.528416819202881e-05,
|
4117 |
+
"loss": 2.2274,
|
4118 |
+
"step": 587
|
4119 |
+
},
|
4120 |
+
{
|
4121 |
+
"epoch": 0.5986256044795113,
|
4122 |
+
"grad_norm": 7.487548351287842,
|
4123 |
+
"learning_rate": 3.5130745139922574e-05,
|
4124 |
+
"loss": 2.104,
|
4125 |
+
"step": 588
|
4126 |
+
},
|
4127 |
+
{
|
4128 |
+
"epoch": 0.5996436752354288,
|
4129 |
+
"grad_norm": 8.788456916809082,
|
4130 |
+
"learning_rate": 3.497747551802221e-05,
|
4131 |
+
"loss": 2.5106,
|
4132 |
+
"step": 589
|
4133 |
+
},
|
4134 |
+
{
|
4135 |
+
"epoch": 0.6006617459913464,
|
4136 |
+
"grad_norm": 7.41387939453125,
|
4137 |
+
"learning_rate": 3.482436090785882e-05,
|
4138 |
+
"loss": 2.1219,
|
4139 |
+
"step": 590
|
4140 |
+
},
|
4141 |
+
{
|
4142 |
+
"epoch": 0.6016798167472639,
|
4143 |
+
"grad_norm": 6.481340408325195,
|
4144 |
+
"learning_rate": 3.467140288936407e-05,
|
4145 |
+
"loss": 1.9451,
|
4146 |
+
"step": 591
|
4147 |
+
},
|
4148 |
+
{
|
4149 |
+
"epoch": 0.6026978875031814,
|
4150 |
+
"grad_norm": 7.278069496154785,
|
4151 |
+
"learning_rate": 3.451860304085378e-05,
|
4152 |
+
"loss": 1.8661,
|
4153 |
+
"step": 592
|
4154 |
+
},
|
4155 |
+
{
|
4156 |
+
"epoch": 0.603715958259099,
|
4157 |
+
"grad_norm": 8.016121864318848,
|
4158 |
+
"learning_rate": 3.43659629390117e-05,
|
4159 |
+
"loss": 1.9884,
|
4160 |
+
"step": 593
|
4161 |
+
},
|
4162 |
+
{
|
4163 |
+
"epoch": 0.6047340290150165,
|
4164 |
+
"grad_norm": 8.917866706848145,
|
4165 |
+
"learning_rate": 3.421348415887315e-05,
|
4166 |
+
"loss": 2.6266,
|
4167 |
+
"step": 594
|
4168 |
+
},
|
4169 |
+
{
|
4170 |
+
"epoch": 0.605752099770934,
|
4171 |
+
"grad_norm": 9.271273612976074,
|
4172 |
+
"learning_rate": 3.406116827380889e-05,
|
4173 |
+
"loss": 2.6668,
|
4174 |
+
"step": 595
|
4175 |
+
},
|
4176 |
+
{
|
4177 |
+
"epoch": 0.6067701705268517,
|
4178 |
+
"grad_norm": 7.660860061645508,
|
4179 |
+
"learning_rate": 3.390901685550887e-05,
|
4180 |
+
"loss": 2.373,
|
4181 |
+
"step": 596
|
4182 |
+
},
|
4183 |
+
{
|
4184 |
+
"epoch": 0.6077882412827692,
|
4185 |
+
"grad_norm": 7.496829032897949,
|
4186 |
+
"learning_rate": 3.375703147396583e-05,
|
4187 |
+
"loss": 2.137,
|
4188 |
+
"step": 597
|
4189 |
+
},
|
4190 |
+
{
|
4191 |
+
"epoch": 0.6088063120386867,
|
4192 |
+
"grad_norm": 10.63588809967041,
|
4193 |
+
"learning_rate": 3.360521369745937e-05,
|
4194 |
+
"loss": 2.0113,
|
4195 |
+
"step": 598
|
4196 |
+
},
|
4197 |
+
{
|
4198 |
+
"epoch": 0.6098243827946043,
|
4199 |
+
"grad_norm": 8.661003112792969,
|
4200 |
+
"learning_rate": 3.345356509253959e-05,
|
4201 |
+
"loss": 2.202,
|
4202 |
+
"step": 599
|
4203 |
+
},
|
4204 |
+
{
|
4205 |
+
"epoch": 0.6108424535505218,
|
4206 |
+
"grad_norm": 6.928518295288086,
|
4207 |
+
"learning_rate": 3.330208722401097e-05,
|
4208 |
+
"loss": 1.6603,
|
4209 |
+
"step": 600
|
4210 |
+
},
|
4211 |
+
{
|
4212 |
+
"epoch": 0.6118605243064393,
|
4213 |
+
"grad_norm": 5.956086158752441,
|
4214 |
+
"learning_rate": 3.315078165491622e-05,
|
4215 |
+
"loss": 2.2319,
|
4216 |
+
"step": 601
|
4217 |
+
},
|
4218 |
+
{
|
4219 |
+
"epoch": 0.6128785950623569,
|
4220 |
+
"grad_norm": 9.131757736206055,
|
4221 |
+
"learning_rate": 3.2999649946520174e-05,
|
4222 |
+
"loss": 3.3601,
|
4223 |
+
"step": 602
|
4224 |
+
},
|
4225 |
+
{
|
4226 |
+
"epoch": 0.6138966658182744,
|
4227 |
+
"grad_norm": 8.110289573669434,
|
4228 |
+
"learning_rate": 3.2848693658293675e-05,
|
4229 |
+
"loss": 2.8758,
|
4230 |
+
"step": 603
|
4231 |
+
},
|
4232 |
+
{
|
4233 |
+
"epoch": 0.6149147365741919,
|
4234 |
+
"grad_norm": 11.287444114685059,
|
4235 |
+
"learning_rate": 3.2697914347897406e-05,
|
4236 |
+
"loss": 4.129,
|
4237 |
+
"step": 604
|
4238 |
+
},
|
4239 |
+
{
|
4240 |
+
"epoch": 0.6159328073301095,
|
4241 |
+
"grad_norm": 10.69924259185791,
|
4242 |
+
"learning_rate": 3.254731357116597e-05,
|
4243 |
+
"loss": 4.2776,
|
4244 |
+
"step": 605
|
4245 |
+
},
|
4246 |
+
{
|
4247 |
+
"epoch": 0.616950878086027,
|
4248 |
+
"grad_norm": 9.89280891418457,
|
4249 |
+
"learning_rate": 3.239689288209168e-05,
|
4250 |
+
"loss": 3.1346,
|
4251 |
+
"step": 606
|
4252 |
+
},
|
4253 |
+
{
|
4254 |
+
"epoch": 0.6179689488419445,
|
4255 |
+
"grad_norm": 11.832335472106934,
|
4256 |
+
"learning_rate": 3.224665383280867e-05,
|
4257 |
+
"loss": 3.4148,
|
4258 |
+
"step": 607
|
4259 |
+
},
|
4260 |
+
{
|
4261 |
+
"epoch": 0.6189870195978621,
|
4262 |
+
"grad_norm": 13.277129173278809,
|
4263 |
+
"learning_rate": 3.2096597973576694e-05,
|
4264 |
+
"loss": 3.4906,
|
4265 |
+
"step": 608
|
4266 |
+
},
|
4267 |
+
{
|
4268 |
+
"epoch": 0.6200050903537796,
|
4269 |
+
"grad_norm": 6.8787994384765625,
|
4270 |
+
"learning_rate": 3.194672685276532e-05,
|
4271 |
+
"loss": 1.4383,
|
4272 |
+
"step": 609
|
4273 |
+
},
|
4274 |
+
{
|
4275 |
+
"epoch": 0.6210231611096971,
|
4276 |
+
"grad_norm": 5.783747673034668,
|
4277 |
+
"learning_rate": 3.179704201683786e-05,
|
4278 |
+
"loss": 1.3518,
|
4279 |
+
"step": 610
|
4280 |
+
},
|
4281 |
+
{
|
4282 |
+
"epoch": 0.6220412318656147,
|
4283 |
+
"grad_norm": 5.462782859802246,
|
4284 |
+
"learning_rate": 3.16475450103354e-05,
|
4285 |
+
"loss": 1.249,
|
4286 |
+
"step": 611
|
4287 |
+
},
|
4288 |
+
{
|
4289 |
+
"epoch": 0.6230593026215322,
|
4290 |
+
"grad_norm": 5.050539016723633,
|
4291 |
+
"learning_rate": 3.1498237375860886e-05,
|
4292 |
+
"loss": 1.1348,
|
4293 |
+
"step": 612
|
4294 |
+
},
|
4295 |
+
{
|
4296 |
+
"epoch": 0.6240773733774497,
|
4297 |
+
"grad_norm": 8.341720581054688,
|
4298 |
+
"learning_rate": 3.1349120654063225e-05,
|
4299 |
+
"loss": 1.7345,
|
4300 |
+
"step": 613
|
4301 |
+
},
|
4302 |
+
{
|
4303 |
+
"epoch": 0.6250954441333673,
|
4304 |
+
"grad_norm": 4.832444190979004,
|
4305 |
+
"learning_rate": 3.120019638362136e-05,
|
4306 |
+
"loss": 1.0501,
|
4307 |
+
"step": 614
|
4308 |
+
},
|
4309 |
+
{
|
4310 |
+
"epoch": 0.6261135148892848,
|
4311 |
+
"grad_norm": 7.373495578765869,
|
4312 |
+
"learning_rate": 3.1051466101228385e-05,
|
4313 |
+
"loss": 1.7428,
|
4314 |
+
"step": 615
|
4315 |
+
},
|
4316 |
+
{
|
4317 |
+
"epoch": 0.6271315856452023,
|
4318 |
+
"grad_norm": 5.6345319747924805,
|
4319 |
+
"learning_rate": 3.090293134157572e-05,
|
4320 |
+
"loss": 1.2435,
|
4321 |
+
"step": 616
|
4322 |
+
},
|
4323 |
+
{
|
4324 |
+
"epoch": 0.6281496564011199,
|
4325 |
+
"grad_norm": 6.5224609375,
|
4326 |
+
"learning_rate": 3.0754593637337276e-05,
|
4327 |
+
"loss": 1.4176,
|
4328 |
+
"step": 617
|
4329 |
+
},
|
4330 |
+
{
|
4331 |
+
"epoch": 0.6291677271570374,
|
4332 |
+
"grad_norm": 8.80791187286377,
|
4333 |
+
"learning_rate": 3.06064545191536e-05,
|
4334 |
+
"loss": 2.4285,
|
4335 |
+
"step": 618
|
4336 |
+
},
|
4337 |
+
{
|
4338 |
+
"epoch": 0.6301857979129549,
|
4339 |
+
"grad_norm": 9.331201553344727,
|
4340 |
+
"learning_rate": 3.0458515515616115e-05,
|
4341 |
+
"loss": 2.7192,
|
4342 |
+
"step": 619
|
4343 |
+
},
|
4344 |
+
{
|
4345 |
+
"epoch": 0.6312038686688725,
|
4346 |
+
"grad_norm": 9.033586502075195,
|
4347 |
+
"learning_rate": 3.0310778153251324e-05,
|
4348 |
+
"loss": 1.8652,
|
4349 |
+
"step": 620
|
4350 |
+
},
|
4351 |
+
{
|
4352 |
+
"epoch": 0.63222193942479,
|
4353 |
+
"grad_norm": 6.689144134521484,
|
4354 |
+
"learning_rate": 3.0163243956505095e-05,
|
4355 |
+
"loss": 1.5773,
|
4356 |
+
"step": 621
|
4357 |
+
},
|
4358 |
+
{
|
4359 |
+
"epoch": 0.6332400101807075,
|
4360 |
+
"grad_norm": 8.037043571472168,
|
4361 |
+
"learning_rate": 3.0015914447726867e-05,
|
4362 |
+
"loss": 2.3296,
|
4363 |
+
"step": 622
|
4364 |
+
},
|
4365 |
+
{
|
4366 |
+
"epoch": 0.6342580809366251,
|
4367 |
+
"grad_norm": 7.927774906158447,
|
4368 |
+
"learning_rate": 2.986879114715403e-05,
|
4369 |
+
"loss": 2.2707,
|
4370 |
+
"step": 623
|
4371 |
+
},
|
4372 |
+
{
|
4373 |
+
"epoch": 0.6352761516925426,
|
4374 |
+
"grad_norm": 5.514461994171143,
|
4375 |
+
"learning_rate": 2.9721875572896157e-05,
|
4376 |
+
"loss": 1.7974,
|
4377 |
+
"step": 624
|
4378 |
+
},
|
4379 |
+
{
|
4380 |
+
"epoch": 0.6362942224484601,
|
4381 |
+
"grad_norm": 7.439801216125488,
|
4382 |
+
"learning_rate": 2.95751692409194e-05,
|
4383 |
+
"loss": 2.1823,
|
4384 |
+
"step": 625
|
4385 |
+
},
|
4386 |
+
{
|
4387 |
+
"epoch": 0.6373122932043777,
|
4388 |
+
"grad_norm": 7.419183731079102,
|
4389 |
+
"learning_rate": 2.942867366503077e-05,
|
4390 |
+
"loss": 2.1965,
|
4391 |
+
"step": 626
|
4392 |
+
},
|
4393 |
+
{
|
4394 |
+
"epoch": 0.6383303639602952,
|
4395 |
+
"grad_norm": 5.545042037963867,
|
4396 |
+
"learning_rate": 2.9282390356862606e-05,
|
4397 |
+
"loss": 1.4957,
|
4398 |
+
"step": 627
|
4399 |
+
},
|
4400 |
+
{
|
4401 |
+
"epoch": 0.6393484347162128,
|
4402 |
+
"grad_norm": 11.62447738647461,
|
4403 |
+
"learning_rate": 2.9136320825856967e-05,
|
4404 |
+
"loss": 3.3109,
|
4405 |
+
"step": 628
|
4406 |
+
},
|
4407 |
+
{
|
4408 |
+
"epoch": 0.6403665054721304,
|
4409 |
+
"grad_norm": 8.367134094238281,
|
4410 |
+
"learning_rate": 2.899046657924992e-05,
|
4411 |
+
"loss": 2.2194,
|
4412 |
+
"step": 629
|
4413 |
+
},
|
4414 |
+
{
|
4415 |
+
"epoch": 0.6413845762280479,
|
4416 |
+
"grad_norm": 10.391725540161133,
|
4417 |
+
"learning_rate": 2.884482912205621e-05,
|
4418 |
+
"loss": 2.0195,
|
4419 |
+
"step": 630
|
4420 |
+
},
|
4421 |
+
{
|
4422 |
+
"epoch": 0.6424026469839654,
|
4423 |
+
"grad_norm": 8.217406272888184,
|
4424 |
+
"learning_rate": 2.8699409957053535e-05,
|
4425 |
+
"loss": 2.4132,
|
4426 |
+
"step": 631
|
4427 |
+
},
|
4428 |
+
{
|
4429 |
+
"epoch": 0.643420717739883,
|
4430 |
+
"grad_norm": 8.29297161102295,
|
4431 |
+
"learning_rate": 2.855421058476719e-05,
|
4432 |
+
"loss": 2.4454,
|
4433 |
+
"step": 632
|
4434 |
+
},
|
4435 |
+
{
|
4436 |
+
"epoch": 0.6444387884958005,
|
4437 |
+
"grad_norm": 8.815670013427734,
|
4438 |
+
"learning_rate": 2.840923250345442e-05,
|
4439 |
+
"loss": 2.5413,
|
4440 |
+
"step": 633
|
4441 |
+
},
|
4442 |
+
{
|
4443 |
+
"epoch": 0.645456859251718,
|
4444 |
+
"grad_norm": 8.5559720993042,
|
4445 |
+
"learning_rate": 2.8264477209089145e-05,
|
4446 |
+
"loss": 2.7664,
|
4447 |
+
"step": 634
|
4448 |
+
},
|
4449 |
+
{
|
4450 |
+
"epoch": 0.6464749300076356,
|
4451 |
+
"grad_norm": 8.682782173156738,
|
4452 |
+
"learning_rate": 2.8119946195346375e-05,
|
4453 |
+
"loss": 2.5312,
|
4454 |
+
"step": 635
|
4455 |
+
},
|
4456 |
+
{
|
4457 |
+
"epoch": 0.6474930007635531,
|
4458 |
+
"grad_norm": 11.519887924194336,
|
4459 |
+
"learning_rate": 2.7975640953586846e-05,
|
4460 |
+
"loss": 2.9688,
|
4461 |
+
"step": 636
|
4462 |
+
},
|
4463 |
+
{
|
4464 |
+
"epoch": 0.6485110715194706,
|
4465 |
+
"grad_norm": 8.966607093811035,
|
4466 |
+
"learning_rate": 2.7831562972841696e-05,
|
4467 |
+
"loss": 2.7022,
|
4468 |
+
"step": 637
|
4469 |
+
},
|
4470 |
+
{
|
4471 |
+
"epoch": 0.6495291422753882,
|
4472 |
+
"grad_norm": 8.183965682983398,
|
4473 |
+
"learning_rate": 2.768771373979697e-05,
|
4474 |
+
"loss": 2.3317,
|
4475 |
+
"step": 638
|
4476 |
+
},
|
4477 |
+
{
|
4478 |
+
"epoch": 0.6505472130313057,
|
4479 |
+
"grad_norm": 8.993667602539062,
|
4480 |
+
"learning_rate": 2.7544094738778436e-05,
|
4481 |
+
"loss": 2.7296,
|
4482 |
+
"step": 639
|
4483 |
+
},
|
4484 |
+
{
|
4485 |
+
"epoch": 0.6515652837872232,
|
4486 |
+
"grad_norm": 7.731354713439941,
|
4487 |
+
"learning_rate": 2.74007074517361e-05,
|
4488 |
+
"loss": 1.9501,
|
4489 |
+
"step": 640
|
4490 |
+
},
|
4491 |
+
{
|
4492 |
+
"epoch": 0.6525833545431408,
|
4493 |
+
"grad_norm": 6.967146396636963,
|
4494 |
+
"learning_rate": 2.7257553358229034e-05,
|
4495 |
+
"loss": 1.8838,
|
4496 |
+
"step": 641
|
4497 |
+
},
|
4498 |
+
{
|
4499 |
+
"epoch": 0.6536014252990583,
|
4500 |
+
"grad_norm": 6.557554244995117,
|
4501 |
+
"learning_rate": 2.7114633935410085e-05,
|
4502 |
+
"loss": 1.7431,
|
4503 |
+
"step": 642
|
4504 |
+
},
|
4505 |
+
{
|
4506 |
+
"epoch": 0.6546194960549758,
|
4507 |
+
"grad_norm": 10.207218170166016,
|
4508 |
+
"learning_rate": 2.6971950658010666e-05,
|
4509 |
+
"loss": 2.4966,
|
4510 |
+
"step": 643
|
4511 |
+
},
|
4512 |
+
{
|
4513 |
+
"epoch": 0.6556375668108934,
|
4514 |
+
"grad_norm": 7.477417469024658,
|
4515 |
+
"learning_rate": 2.682950499832535e-05,
|
4516 |
+
"loss": 2.1944,
|
4517 |
+
"step": 644
|
4518 |
+
},
|
4519 |
+
{
|
4520 |
+
"epoch": 0.6566556375668109,
|
4521 |
+
"grad_norm": 10.127610206604004,
|
4522 |
+
"learning_rate": 2.6687298426196973e-05,
|
4523 |
+
"loss": 2.6473,
|
4524 |
+
"step": 645
|
4525 |
+
},
|
4526 |
+
{
|
4527 |
+
"epoch": 0.6576737083227284,
|
4528 |
+
"grad_norm": 6.374731540679932,
|
4529 |
+
"learning_rate": 2.6545332409001265e-05,
|
4530 |
+
"loss": 1.8528,
|
4531 |
+
"step": 646
|
4532 |
+
},
|
4533 |
+
{
|
4534 |
+
"epoch": 0.658691779078646,
|
4535 |
+
"grad_norm": 6.7048444747924805,
|
4536 |
+
"learning_rate": 2.6403608411631742e-05,
|
4537 |
+
"loss": 1.7493,
|
4538 |
+
"step": 647
|
4539 |
+
},
|
4540 |
+
{
|
4541 |
+
"epoch": 0.6597098498345635,
|
4542 |
+
"grad_norm": 7.112037181854248,
|
4543 |
+
"learning_rate": 2.6262127896484602e-05,
|
4544 |
+
"loss": 2.0421,
|
4545 |
+
"step": 648
|
4546 |
+
},
|
4547 |
+
{
|
4548 |
+
"epoch": 0.660727920590481,
|
4549 |
+
"grad_norm": 8.483193397521973,
|
4550 |
+
"learning_rate": 2.612089232344371e-05,
|
4551 |
+
"loss": 1.91,
|
4552 |
+
"step": 649
|
4553 |
+
},
|
4554 |
+
{
|
4555 |
+
"epoch": 0.6617459913463986,
|
4556 |
+
"grad_norm": 10.052485466003418,
|
4557 |
+
"learning_rate": 2.5979903149865387e-05,
|
4558 |
+
"loss": 2.0998,
|
4559 |
+
"step": 650
|
4560 |
+
},
|
4561 |
+
{
|
4562 |
+
"epoch": 0.6627640621023161,
|
4563 |
+
"grad_norm": 8.01032543182373,
|
4564 |
+
"learning_rate": 2.5839161830563474e-05,
|
4565 |
+
"loss": 2.5145,
|
4566 |
+
"step": 651
|
4567 |
+
},
|
4568 |
+
{
|
4569 |
+
"epoch": 0.6637821328582336,
|
4570 |
+
"grad_norm": 9.746928215026855,
|
4571 |
+
"learning_rate": 2.569866981779433e-05,
|
4572 |
+
"loss": 3.3683,
|
4573 |
+
"step": 652
|
4574 |
+
},
|
4575 |
+
{
|
4576 |
+
"epoch": 0.6648002036141512,
|
4577 |
+
"grad_norm": 8.607123374938965,
|
4578 |
+
"learning_rate": 2.555842856124182e-05,
|
4579 |
+
"loss": 2.9144,
|
4580 |
+
"step": 653
|
4581 |
+
},
|
4582 |
+
{
|
4583 |
+
"epoch": 0.6658182743700687,
|
4584 |
+
"grad_norm": 10.463346481323242,
|
4585 |
+
"learning_rate": 2.5418439508002258e-05,
|
4586 |
+
"loss": 3.9062,
|
4587 |
+
"step": 654
|
4588 |
+
},
|
4589 |
+
{
|
4590 |
+
"epoch": 0.6668363451259862,
|
4591 |
+
"grad_norm": 9.336942672729492,
|
4592 |
+
"learning_rate": 2.5278704102569662e-05,
|
4593 |
+
"loss": 3.3966,
|
4594 |
+
"step": 655
|
4595 |
+
},
|
4596 |
+
{
|
4597 |
+
"epoch": 0.6678544158819038,
|
4598 |
+
"grad_norm": 10.415209770202637,
|
4599 |
+
"learning_rate": 2.5139223786820747e-05,
|
4600 |
+
"loss": 3.7271,
|
4601 |
+
"step": 656
|
4602 |
+
},
|
4603 |
+
{
|
4604 |
+
"epoch": 0.6688724866378213,
|
4605 |
+
"grad_norm": 14.631210327148438,
|
4606 |
+
"learning_rate": 2.500000000000001e-05,
|
4607 |
+
"loss": 3.7071,
|
4608 |
+
"step": 657
|
4609 |
+
},
|
4610 |
+
{
|
4611 |
+
"epoch": 0.6698905573937388,
|
4612 |
+
"grad_norm": 13.001562118530273,
|
4613 |
+
"learning_rate": 2.486103417870493e-05,
|
4614 |
+
"loss": 3.214,
|
4615 |
+
"step": 658
|
4616 |
+
},
|
4617 |
+
{
|
4618 |
+
"epoch": 0.6709086281496563,
|
4619 |
+
"grad_norm": 11.307893753051758,
|
4620 |
+
"learning_rate": 2.472232775687119e-05,
|
4621 |
+
"loss": 2.8893,
|
4622 |
+
"step": 659
|
4623 |
+
},
|
4624 |
+
{
|
4625 |
+
"epoch": 0.6719266989055739,
|
4626 |
+
"grad_norm": 7.8647379875183105,
|
4627 |
+
"learning_rate": 2.4583882165757766e-05,
|
4628 |
+
"loss": 2.0442,
|
4629 |
+
"step": 660
|
4630 |
+
},
|
4631 |
+
{
|
4632 |
+
"epoch": 0.6729447696614915,
|
4633 |
+
"grad_norm": 5.790807247161865,
|
4634 |
+
"learning_rate": 2.4445698833932234e-05,
|
4635 |
+
"loss": 1.3228,
|
4636 |
+
"step": 661
|
4637 |
+
},
|
4638 |
+
{
|
4639 |
+
"epoch": 0.6739628404174091,
|
4640 |
+
"grad_norm": 5.694929599761963,
|
4641 |
+
"learning_rate": 2.4307779187256064e-05,
|
4642 |
+
"loss": 1.3618,
|
4643 |
+
"step": 662
|
4644 |
+
},
|
4645 |
+
{
|
4646 |
+
"epoch": 0.6749809111733266,
|
4647 |
+
"grad_norm": 5.114007949829102,
|
4648 |
+
"learning_rate": 2.417012464886978e-05,
|
4649 |
+
"loss": 1.2137,
|
4650 |
+
"step": 663
|
4651 |
+
},
|
4652 |
+
{
|
4653 |
+
"epoch": 0.6759989819292441,
|
4654 |
+
"grad_norm": 7.429940223693848,
|
4655 |
+
"learning_rate": 2.4032736639178444e-05,
|
4656 |
+
"loss": 1.8593,
|
4657 |
+
"step": 664
|
4658 |
+
},
|
4659 |
+
{
|
4660 |
+
"epoch": 0.6770170526851617,
|
4661 |
+
"grad_norm": 5.101173400878906,
|
4662 |
+
"learning_rate": 2.389561657583681e-05,
|
4663 |
+
"loss": 0.9669,
|
4664 |
+
"step": 665
|
4665 |
+
},
|
4666 |
+
{
|
4667 |
+
"epoch": 0.6780351234410792,
|
4668 |
+
"grad_norm": 7.89351224899292,
|
4669 |
+
"learning_rate": 2.3758765873734896e-05,
|
4670 |
+
"loss": 1.8615,
|
4671 |
+
"step": 666
|
4672 |
+
},
|
4673 |
+
{
|
4674 |
+
"epoch": 0.6790531941969967,
|
4675 |
+
"grad_norm": 7.043496608734131,
|
4676 |
+
"learning_rate": 2.3622185944983188e-05,
|
4677 |
+
"loss": 1.7828,
|
4678 |
+
"step": 667
|
4679 |
+
},
|
4680 |
+
{
|
4681 |
+
"epoch": 0.6800712649529143,
|
4682 |
+
"grad_norm": 7.9154510498046875,
|
4683 |
+
"learning_rate": 2.3485878198898252e-05,
|
4684 |
+
"loss": 2.2469,
|
4685 |
+
"step": 668
|
4686 |
+
},
|
4687 |
+
{
|
4688 |
+
"epoch": 0.6810893357088318,
|
4689 |
+
"grad_norm": 6.627047061920166,
|
4690 |
+
"learning_rate": 2.3349844041988045e-05,
|
4691 |
+
"loss": 1.5789,
|
4692 |
+
"step": 669
|
4693 |
+
},
|
4694 |
+
{
|
4695 |
+
"epoch": 0.6821074064647493,
|
4696 |
+
"grad_norm": 5.884915828704834,
|
4697 |
+
"learning_rate": 2.3214084877937464e-05,
|
4698 |
+
"loss": 1.5281,
|
4699 |
+
"step": 670
|
4700 |
+
},
|
4701 |
+
{
|
4702 |
+
"epoch": 0.6831254772206669,
|
4703 |
+
"grad_norm": 6.640014171600342,
|
4704 |
+
"learning_rate": 2.30786021075939e-05,
|
4705 |
+
"loss": 1.4942,
|
4706 |
+
"step": 671
|
4707 |
+
},
|
4708 |
+
{
|
4709 |
+
"epoch": 0.6841435479765844,
|
4710 |
+
"grad_norm": 6.866456985473633,
|
4711 |
+
"learning_rate": 2.294339712895271e-05,
|
4712 |
+
"loss": 1.674,
|
4713 |
+
"step": 672
|
4714 |
+
},
|
4715 |
+
{
|
4716 |
+
"epoch": 0.6851616187325019,
|
4717 |
+
"grad_norm": 6.7534990310668945,
|
4718 |
+
"learning_rate": 2.28084713371428e-05,
|
4719 |
+
"loss": 1.3313,
|
4720 |
+
"step": 673
|
4721 |
+
},
|
4722 |
+
{
|
4723 |
+
"epoch": 0.6861796894884195,
|
4724 |
+
"grad_norm": 6.38292121887207,
|
4725 |
+
"learning_rate": 2.2673826124412312e-05,
|
4726 |
+
"loss": 1.6016,
|
4727 |
+
"step": 674
|
4728 |
+
},
|
4729 |
+
{
|
4730 |
+
"epoch": 0.687197760244337,
|
4731 |
+
"grad_norm": 7.129096031188965,
|
4732 |
+
"learning_rate": 2.2539462880114194e-05,
|
4733 |
+
"loss": 1.8662,
|
4734 |
+
"step": 675
|
4735 |
+
},
|
4736 |
+
{
|
4737 |
+
"epoch": 0.6882158310002545,
|
4738 |
+
"grad_norm": 6.555764675140381,
|
4739 |
+
"learning_rate": 2.240538299069178e-05,
|
4740 |
+
"loss": 1.9315,
|
4741 |
+
"step": 676
|
4742 |
+
},
|
4743 |
+
{
|
4744 |
+
"epoch": 0.689233901756172,
|
4745 |
+
"grad_norm": 5.772182941436768,
|
4746 |
+
"learning_rate": 2.2271587839664672e-05,
|
4747 |
+
"loss": 1.3156,
|
4748 |
+
"step": 677
|
4749 |
+
},
|
4750 |
+
{
|
4751 |
+
"epoch": 0.6902519725120896,
|
4752 |
+
"grad_norm": 7.608791351318359,
|
4753 |
+
"learning_rate": 2.213807880761434e-05,
|
4754 |
+
"loss": 1.9463,
|
4755 |
+
"step": 678
|
4756 |
+
},
|
4757 |
+
{
|
4758 |
+
"epoch": 0.6912700432680071,
|
4759 |
+
"grad_norm": 7.279063701629639,
|
4760 |
+
"learning_rate": 2.2004857272169876e-05,
|
4761 |
+
"loss": 1.9304,
|
4762 |
+
"step": 679
|
4763 |
+
},
|
4764 |
+
{
|
4765 |
+
"epoch": 0.6922881140239247,
|
4766 |
+
"grad_norm": 9.676162719726562,
|
4767 |
+
"learning_rate": 2.1871924607993797e-05,
|
4768 |
+
"loss": 2.3767,
|
4769 |
+
"step": 680
|
4770 |
+
},
|
4771 |
+
{
|
4772 |
+
"epoch": 0.6933061847798422,
|
4773 |
+
"grad_norm": 7.1779093742370605,
|
4774 |
+
"learning_rate": 2.1739282186767923e-05,
|
4775 |
+
"loss": 1.6381,
|
4776 |
+
"step": 681
|
4777 |
+
},
|
4778 |
+
{
|
4779 |
+
"epoch": 0.6943242555357597,
|
4780 |
+
"grad_norm": 6.892930030822754,
|
4781 |
+
"learning_rate": 2.160693137717912e-05,
|
4782 |
+
"loss": 2.134,
|
4783 |
+
"step": 682
|
4784 |
+
},
|
4785 |
+
{
|
4786 |
+
"epoch": 0.6953423262916772,
|
4787 |
+
"grad_norm": 9.403331756591797,
|
4788 |
+
"learning_rate": 2.1474873544905205e-05,
|
4789 |
+
"loss": 2.2294,
|
4790 |
+
"step": 683
|
4791 |
+
},
|
4792 |
+
{
|
4793 |
+
"epoch": 0.6963603970475948,
|
4794 |
+
"grad_norm": 7.7654595375061035,
|
4795 |
+
"learning_rate": 2.134311005260093e-05,
|
4796 |
+
"loss": 2.0953,
|
4797 |
+
"step": 684
|
4798 |
+
},
|
4799 |
+
{
|
4800 |
+
"epoch": 0.6973784678035123,
|
4801 |
+
"grad_norm": 10.087757110595703,
|
4802 |
+
"learning_rate": 2.1211642259883867e-05,
|
4803 |
+
"loss": 2.9221,
|
4804 |
+
"step": 685
|
4805 |
+
},
|
4806 |
+
{
|
4807 |
+
"epoch": 0.6983965385594298,
|
4808 |
+
"grad_norm": 8.816588401794434,
|
4809 |
+
"learning_rate": 2.108047152332028e-05,
|
4810 |
+
"loss": 2.6949,
|
4811 |
+
"step": 686
|
4812 |
+
},
|
4813 |
+
{
|
4814 |
+
"epoch": 0.6994146093153474,
|
4815 |
+
"grad_norm": 8.12427043914795,
|
4816 |
+
"learning_rate": 2.0949599196411325e-05,
|
4817 |
+
"loss": 1.7944,
|
4818 |
+
"step": 687
|
4819 |
+
},
|
4820 |
+
{
|
4821 |
+
"epoch": 0.7004326800712649,
|
4822 |
+
"grad_norm": 7.3718461990356445,
|
4823 |
+
"learning_rate": 2.0819026629578952e-05,
|
4824 |
+
"loss": 2.1142,
|
4825 |
+
"step": 688
|
4826 |
+
},
|
4827 |
+
{
|
4828 |
+
"epoch": 0.7014507508271824,
|
4829 |
+
"grad_norm": 7.3536577224731445,
|
4830 |
+
"learning_rate": 2.0688755170151996e-05,
|
4831 |
+
"loss": 2.0029,
|
4832 |
+
"step": 689
|
4833 |
+
},
|
4834 |
+
{
|
4835 |
+
"epoch": 0.7024688215831,
|
4836 |
+
"grad_norm": 8.220134735107422,
|
4837 |
+
"learning_rate": 2.0558786162352244e-05,
|
4838 |
+
"loss": 2.2986,
|
4839 |
+
"step": 690
|
4840 |
+
},
|
4841 |
+
{
|
4842 |
+
"epoch": 0.7034868923390175,
|
4843 |
+
"grad_norm": 9.169322967529297,
|
4844 |
+
"learning_rate": 2.0429120947280678e-05,
|
4845 |
+
"loss": 2.3455,
|
4846 |
+
"step": 691
|
4847 |
+
},
|
4848 |
+
{
|
4849 |
+
"epoch": 0.704504963094935,
|
4850 |
+
"grad_norm": 8.935730934143066,
|
4851 |
+
"learning_rate": 2.029976086290347e-05,
|
4852 |
+
"loss": 2.1588,
|
4853 |
+
"step": 692
|
4854 |
+
},
|
4855 |
+
{
|
4856 |
+
"epoch": 0.7055230338508527,
|
4857 |
+
"grad_norm": 7.555604934692383,
|
4858 |
+
"learning_rate": 2.017070724403835e-05,
|
4859 |
+
"loss": 2.2783,
|
4860 |
+
"step": 693
|
4861 |
+
},
|
4862 |
+
{
|
4863 |
+
"epoch": 0.7065411046067702,
|
4864 |
+
"grad_norm": 7.896771430969238,
|
4865 |
+
"learning_rate": 2.0041961422340676e-05,
|
4866 |
+
"loss": 1.8964,
|
4867 |
+
"step": 694
|
4868 |
+
},
|
4869 |
+
{
|
4870 |
+
"epoch": 0.7075591753626878,
|
4871 |
+
"grad_norm": 8.242528915405273,
|
4872 |
+
"learning_rate": 1.9913524726289784e-05,
|
4873 |
+
"loss": 1.9936,
|
4874 |
+
"step": 695
|
4875 |
+
},
|
4876 |
+
{
|
4877 |
+
"epoch": 0.7085772461186053,
|
4878 |
+
"grad_norm": 7.946272373199463,
|
4879 |
+
"learning_rate": 1.9785398481175294e-05,
|
4880 |
+
"loss": 2.1526,
|
4881 |
+
"step": 696
|
4882 |
+
},
|
4883 |
+
{
|
4884 |
+
"epoch": 0.7095953168745228,
|
4885 |
+
"grad_norm": 8.382307052612305,
|
4886 |
+
"learning_rate": 1.965758400908334e-05,
|
4887 |
+
"loss": 2.4691,
|
4888 |
+
"step": 697
|
4889 |
+
},
|
4890 |
+
{
|
4891 |
+
"epoch": 0.7106133876304404,
|
4892 |
+
"grad_norm": 6.839285373687744,
|
4893 |
+
"learning_rate": 1.9530082628883056e-05,
|
4894 |
+
"loss": 1.7924,
|
4895 |
+
"step": 698
|
4896 |
+
},
|
4897 |
+
{
|
4898 |
+
"epoch": 0.7116314583863579,
|
4899 |
+
"grad_norm": 12.65297794342041,
|
4900 |
+
"learning_rate": 1.9402895656212833e-05,
|
4901 |
+
"loss": 2.0093,
|
4902 |
+
"step": 699
|
4903 |
+
},
|
4904 |
+
{
|
4905 |
+
"epoch": 0.7126495291422754,
|
4906 |
+
"grad_norm": 11.35102653503418,
|
4907 |
+
"learning_rate": 1.927602440346687e-05,
|
4908 |
+
"loss": 1.7963,
|
4909 |
+
"step": 700
|
4910 |
+
},
|
4911 |
+
{
|
4912 |
+
"epoch": 0.713667599898193,
|
4913 |
+
"grad_norm": 7.479799747467041,
|
4914 |
+
"learning_rate": 1.914947017978153e-05,
|
4915 |
+
"loss": 3.4169,
|
4916 |
+
"step": 701
|
4917 |
+
},
|
4918 |
+
{
|
4919 |
+
"epoch": 0.7146856706541105,
|
4920 |
+
"grad_norm": 9.703947067260742,
|
4921 |
+
"learning_rate": 1.9023234291021873e-05,
|
4922 |
+
"loss": 2.8178,
|
4923 |
+
"step": 702
|
4924 |
+
},
|
4925 |
+
{
|
4926 |
+
"epoch": 0.715703741410028,
|
4927 |
+
"grad_norm": 10.218291282653809,
|
4928 |
+
"learning_rate": 1.889731803976822e-05,
|
4929 |
+
"loss": 2.841,
|
4930 |
+
"step": 703
|
4931 |
+
},
|
4932 |
+
{
|
4933 |
+
"epoch": 0.7167218121659455,
|
4934 |
+
"grad_norm": 12.210125923156738,
|
4935 |
+
"learning_rate": 1.8771722725302643e-05,
|
4936 |
+
"loss": 3.9947,
|
4937 |
+
"step": 704
|
4938 |
+
},
|
4939 |
+
{
|
4940 |
+
"epoch": 0.7177398829218631,
|
4941 |
+
"grad_norm": 9.851053237915039,
|
4942 |
+
"learning_rate": 1.8646449643595565e-05,
|
4943 |
+
"loss": 2.8836,
|
4944 |
+
"step": 705
|
4945 |
+
},
|
4946 |
+
{
|
4947 |
+
"epoch": 0.7187579536777806,
|
4948 |
+
"grad_norm": 11.182621955871582,
|
4949 |
+
"learning_rate": 1.8521500087292467e-05,
|
4950 |
+
"loss": 3.2881,
|
4951 |
+
"step": 706
|
4952 |
+
},
|
4953 |
+
{
|
4954 |
+
"epoch": 0.7197760244336981,
|
4955 |
+
"grad_norm": 16.472837448120117,
|
4956 |
+
"learning_rate": 1.8396875345700497e-05,
|
4957 |
+
"loss": 3.6782,
|
4958 |
+
"step": 707
|
4959 |
+
},
|
4960 |
+
{
|
4961 |
+
"epoch": 0.7207940951896157,
|
4962 |
+
"grad_norm": 13.632477760314941,
|
4963 |
+
"learning_rate": 1.8272576704775074e-05,
|
4964 |
+
"loss": 3.5599,
|
4965 |
+
"step": 708
|
4966 |
+
},
|
4967 |
+
{
|
4968 |
+
"epoch": 0.7218121659455332,
|
4969 |
+
"grad_norm": 8.531991958618164,
|
4970 |
+
"learning_rate": 1.8148605447106797e-05,
|
4971 |
+
"loss": 1.815,
|
4972 |
+
"step": 709
|
4973 |
+
},
|
4974 |
+
{
|
4975 |
+
"epoch": 0.7228302367014507,
|
4976 |
+
"grad_norm": 6.116468906402588,
|
4977 |
+
"learning_rate": 1.8024962851908107e-05,
|
4978 |
+
"loss": 1.3279,
|
4979 |
+
"step": 710
|
4980 |
+
},
|
4981 |
+
{
|
4982 |
+
"epoch": 0.7238483074573683,
|
4983 |
+
"grad_norm": 6.058359622955322,
|
4984 |
+
"learning_rate": 1.7901650195000068e-05,
|
4985 |
+
"loss": 1.1209,
|
4986 |
+
"step": 711
|
4987 |
+
},
|
4988 |
+
{
|
4989 |
+
"epoch": 0.7248663782132858,
|
4990 |
+
"grad_norm": 7.301308631896973,
|
4991 |
+
"learning_rate": 1.7778668748799242e-05,
|
4992 |
+
"loss": 1.6941,
|
4993 |
+
"step": 712
|
4994 |
+
},
|
4995 |
+
{
|
4996 |
+
"epoch": 0.7258844489692033,
|
4997 |
+
"grad_norm": 6.059625148773193,
|
4998 |
+
"learning_rate": 1.76560197823046e-05,
|
4999 |
+
"loss": 1.4134,
|
5000 |
+
"step": 713
|
5001 |
+
},
|
5002 |
+
{
|
5003 |
+
"epoch": 0.7269025197251209,
|
5004 |
+
"grad_norm": 5.40415620803833,
|
5005 |
+
"learning_rate": 1.753370456108433e-05,
|
5006 |
+
"loss": 1.5117,
|
5007 |
+
"step": 714
|
5008 |
+
},
|
5009 |
+
{
|
5010 |
+
"epoch": 0.7279205904810384,
|
5011 |
+
"grad_norm": 6.5403008460998535,
|
5012 |
+
"learning_rate": 1.7411724347262824e-05,
|
5013 |
+
"loss": 1.397,
|
5014 |
+
"step": 715
|
5015 |
+
},
|
5016 |
+
{
|
5017 |
+
"epoch": 0.7289386612369559,
|
5018 |
+
"grad_norm": 8.339217185974121,
|
5019 |
+
"learning_rate": 1.729008039950772e-05,
|
5020 |
+
"loss": 1.5315,
|
5021 |
+
"step": 716
|
5022 |
+
},
|
5023 |
+
{
|
5024 |
+
"epoch": 0.7299567319928735,
|
5025 |
+
"grad_norm": 5.882655620574951,
|
5026 |
+
"learning_rate": 1.7168773973016776e-05,
|
5027 |
+
"loss": 1.1574,
|
5028 |
+
"step": 717
|
5029 |
+
},
|
5030 |
+
{
|
5031 |
+
"epoch": 0.730974802748791,
|
5032 |
+
"grad_norm": 6.183307647705078,
|
5033 |
+
"learning_rate": 1.7047806319505076e-05,
|
5034 |
+
"loss": 1.3367,
|
5035 |
+
"step": 718
|
5036 |
+
},
|
5037 |
+
{
|
5038 |
+
"epoch": 0.7319928735047085,
|
5039 |
+
"grad_norm": 6.28183126449585,
|
5040 |
+
"learning_rate": 1.692717868719195e-05,
|
5041 |
+
"loss": 1.5637,
|
5042 |
+
"step": 719
|
5043 |
+
},
|
5044 |
+
{
|
5045 |
+
"epoch": 0.7330109442606261,
|
5046 |
+
"grad_norm": 4.728903293609619,
|
5047 |
+
"learning_rate": 1.680689232078827e-05,
|
5048 |
+
"loss": 1.4179,
|
5049 |
+
"step": 720
|
5050 |
+
},
|
5051 |
+
{
|
5052 |
+
"epoch": 0.7340290150165436,
|
5053 |
+
"grad_norm": 6.95587158203125,
|
5054 |
+
"learning_rate": 1.668694846148343e-05,
|
5055 |
+
"loss": 1.7837,
|
5056 |
+
"step": 721
|
5057 |
+
},
|
5058 |
+
{
|
5059 |
+
"epoch": 0.7350470857724611,
|
5060 |
+
"grad_norm": 5.531774997711182,
|
5061 |
+
"learning_rate": 1.6567348346932658e-05,
|
5062 |
+
"loss": 1.2069,
|
5063 |
+
"step": 722
|
5064 |
+
},
|
5065 |
+
{
|
5066 |
+
"epoch": 0.7360651565283787,
|
5067 |
+
"grad_norm": 5.498968601226807,
|
5068 |
+
"learning_rate": 1.644809321124423e-05,
|
5069 |
+
"loss": 1.1316,
|
5070 |
+
"step": 723
|
5071 |
+
},
|
5072 |
+
{
|
5073 |
+
"epoch": 0.7370832272842962,
|
5074 |
+
"grad_norm": 7.1133809089660645,
|
5075 |
+
"learning_rate": 1.6329184284966677e-05,
|
5076 |
+
"loss": 2.0335,
|
5077 |
+
"step": 724
|
5078 |
+
},
|
5079 |
+
{
|
5080 |
+
"epoch": 0.7381012980402138,
|
5081 |
+
"grad_norm": 6.765145301818848,
|
5082 |
+
"learning_rate": 1.621062279507617e-05,
|
5083 |
+
"loss": 2.0067,
|
5084 |
+
"step": 725
|
5085 |
+
},
|
5086 |
+
{
|
5087 |
+
"epoch": 0.7391193687961314,
|
5088 |
+
"grad_norm": 7.21923828125,
|
5089 |
+
"learning_rate": 1.609240996496378e-05,
|
5090 |
+
"loss": 2.2922,
|
5091 |
+
"step": 726
|
5092 |
+
},
|
5093 |
+
{
|
5094 |
+
"epoch": 0.7401374395520489,
|
5095 |
+
"grad_norm": 5.8889360427856445,
|
5096 |
+
"learning_rate": 1.597454701442288e-05,
|
5097 |
+
"loss": 1.6363,
|
5098 |
+
"step": 727
|
5099 |
+
},
|
5100 |
+
{
|
5101 |
+
"epoch": 0.7411555103079664,
|
5102 |
+
"grad_norm": 8.041604042053223,
|
5103 |
+
"learning_rate": 1.5857035159636623e-05,
|
5104 |
+
"loss": 1.6933,
|
5105 |
+
"step": 728
|
5106 |
+
},
|
5107 |
+
{
|
5108 |
+
"epoch": 0.742173581063884,
|
5109 |
+
"grad_norm": 7.711045742034912,
|
5110 |
+
"learning_rate": 1.5739875613165283e-05,
|
5111 |
+
"loss": 1.9258,
|
5112 |
+
"step": 729
|
5113 |
+
},
|
5114 |
+
{
|
5115 |
+
"epoch": 0.7431916518198015,
|
5116 |
+
"grad_norm": 7.747977256774902,
|
5117 |
+
"learning_rate": 1.5623069583933836e-05,
|
5118 |
+
"loss": 2.5273,
|
5119 |
+
"step": 730
|
5120 |
+
},
|
5121 |
+
{
|
5122 |
+
"epoch": 0.744209722575719,
|
5123 |
+
"grad_norm": 8.055684089660645,
|
5124 |
+
"learning_rate": 1.550661827721941e-05,
|
5125 |
+
"loss": 2.0398,
|
5126 |
+
"step": 731
|
5127 |
+
},
|
5128 |
+
{
|
5129 |
+
"epoch": 0.7452277933316366,
|
5130 |
+
"grad_norm": 8.75759220123291,
|
5131 |
+
"learning_rate": 1.5390522894638938e-05,
|
5132 |
+
"loss": 2.5372,
|
5133 |
+
"step": 732
|
5134 |
+
},
|
5135 |
+
{
|
5136 |
+
"epoch": 0.7462458640875541,
|
5137 |
+
"grad_norm": 6.629666805267334,
|
5138 |
+
"learning_rate": 1.527478463413666e-05,
|
5139 |
+
"loss": 1.8586,
|
5140 |
+
"step": 733
|
5141 |
+
},
|
5142 |
+
{
|
5143 |
+
"epoch": 0.7472639348434716,
|
5144 |
+
"grad_norm": 7.634647369384766,
|
5145 |
+
"learning_rate": 1.5159404689971795e-05,
|
5146 |
+
"loss": 1.7609,
|
5147 |
+
"step": 734
|
5148 |
+
},
|
5149 |
+
{
|
5150 |
+
"epoch": 0.7482820055993892,
|
5151 |
+
"grad_norm": 8.821757316589355,
|
5152 |
+
"learning_rate": 1.5044384252706312e-05,
|
5153 |
+
"loss": 2.5073,
|
5154 |
+
"step": 735
|
5155 |
+
},
|
5156 |
+
{
|
5157 |
+
"epoch": 0.7493000763553067,
|
5158 |
+
"grad_norm": 7.940456867218018,
|
5159 |
+
"learning_rate": 1.4929724509192488e-05,
|
5160 |
+
"loss": 2.6403,
|
5161 |
+
"step": 736
|
5162 |
+
},
|
5163 |
+
{
|
5164 |
+
"epoch": 0.7503181471112242,
|
5165 |
+
"grad_norm": 7.819153308868408,
|
5166 |
+
"learning_rate": 1.4815426642560754e-05,
|
5167 |
+
"loss": 2.3173,
|
5168 |
+
"step": 737
|
5169 |
+
},
|
5170 |
+
{
|
5171 |
+
"epoch": 0.7513362178671418,
|
5172 |
+
"grad_norm": 7.586490154266357,
|
5173 |
+
"learning_rate": 1.470149183220748e-05,
|
5174 |
+
"loss": 2.0191,
|
5175 |
+
"step": 738
|
5176 |
}
|
5177 |
],
|
5178 |
"logging_steps": 1,
|
|
|
5192 |
"attributes": {}
|
5193 |
}
|
5194 |
},
|
5195 |
+
"total_flos": 6.046717707288576e+17,
|
5196 |
"train_batch_size": 4,
|
5197 |
"trial_name": null,
|
5198 |
"trial_params": null
|