Training in progress, step 720, 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 48679352
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:607be725f52f077357c0dd6b617e4d0b6082e8caeb1b6dae914a0be70dc86f38
|
3 |
size 48679352
|
last-checkpoint/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 25152884
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:738b4d6e522c4d141b49b0fd83fec1621c7393c078ed22d6966243f70dca6d4f
|
3 |
size 25152884
|
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 sha256:3ec4f7cc50fb21bbb2ce05c58bfc343a8e8e934252ffd2ecec54dcd950a87a42
|
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 sha256:7c1861a03fc626556465a433fa3f619dd9f923bb33ff04b1959c4edb562d5130
|
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": 240,
|
6 |
-
"global_step":
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
@@ -3391,6 +3391,1694 @@
|
|
3391 |
"eval_samples_per_second": 19.865,
|
3392 |
"eval_steps_per_second": 9.932,
|
3393 |
"step": 480
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
3394 |
}
|
3395 |
],
|
3396 |
"logging_steps": 1,
|
@@ -3410,7 +5098,7 @@
|
|
3410 |
"attributes": {}
|
3411 |
}
|
3412 |
},
|
3413 |
-
"total_flos":
|
3414 |
"train_batch_size": 2,
|
3415 |
"trial_name": null,
|
3416 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 0.7515657620041754,
|
5 |
"eval_steps": 240,
|
6 |
+
"global_step": 720,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
3391 |
"eval_samples_per_second": 19.865,
|
3392 |
"eval_steps_per_second": 9.932,
|
3393 |
"step": 480
|
3394 |
+
},
|
3395 |
+
{
|
3396 |
+
"epoch": 0.5020876826722338,
|
3397 |
+
"grad_norm": 0.7070721387863159,
|
3398 |
+
"learning_rate": 0.00010099415851335299,
|
3399 |
+
"loss": 0.3174,
|
3400 |
+
"step": 481
|
3401 |
+
},
|
3402 |
+
{
|
3403 |
+
"epoch": 0.5031315240083507,
|
3404 |
+
"grad_norm": 0.7391024827957153,
|
3405 |
+
"learning_rate": 0.00010066277840778626,
|
3406 |
+
"loss": 0.3442,
|
3407 |
+
"step": 482
|
3408 |
+
},
|
3409 |
+
{
|
3410 |
+
"epoch": 0.5041753653444676,
|
3411 |
+
"grad_norm": 0.7629324793815613,
|
3412 |
+
"learning_rate": 0.00010033139102356642,
|
3413 |
+
"loss": 0.3439,
|
3414 |
+
"step": 483
|
3415 |
+
},
|
3416 |
+
{
|
3417 |
+
"epoch": 0.5052192066805845,
|
3418 |
+
"grad_norm": 0.7324389219284058,
|
3419 |
+
"learning_rate": 0.0001,
|
3420 |
+
"loss": 0.3063,
|
3421 |
+
"step": 484
|
3422 |
+
},
|
3423 |
+
{
|
3424 |
+
"epoch": 0.5062630480167014,
|
3425 |
+
"grad_norm": 0.6402798295021057,
|
3426 |
+
"learning_rate": 9.966860897643359e-05,
|
3427 |
+
"loss": 0.2383,
|
3428 |
+
"step": 485
|
3429 |
+
},
|
3430 |
+
{
|
3431 |
+
"epoch": 0.5073068893528184,
|
3432 |
+
"grad_norm": 0.7618774771690369,
|
3433 |
+
"learning_rate": 9.933722159221376e-05,
|
3434 |
+
"loss": 0.3004,
|
3435 |
+
"step": 486
|
3436 |
+
},
|
3437 |
+
{
|
3438 |
+
"epoch": 0.5083507306889353,
|
3439 |
+
"grad_norm": 0.8296042680740356,
|
3440 |
+
"learning_rate": 9.900584148664704e-05,
|
3441 |
+
"loss": 0.3208,
|
3442 |
+
"step": 487
|
3443 |
+
},
|
3444 |
+
{
|
3445 |
+
"epoch": 0.5093945720250522,
|
3446 |
+
"grad_norm": 0.7663673162460327,
|
3447 |
+
"learning_rate": 9.867447229896018e-05,
|
3448 |
+
"loss": 0.3204,
|
3449 |
+
"step": 488
|
3450 |
+
},
|
3451 |
+
{
|
3452 |
+
"epoch": 0.5104384133611691,
|
3453 |
+
"grad_norm": 0.7188003063201904,
|
3454 |
+
"learning_rate": 9.834311766825985e-05,
|
3455 |
+
"loss": 0.2645,
|
3456 |
+
"step": 489
|
3457 |
+
},
|
3458 |
+
{
|
3459 |
+
"epoch": 0.511482254697286,
|
3460 |
+
"grad_norm": 0.6017361879348755,
|
3461 |
+
"learning_rate": 9.801178123349298e-05,
|
3462 |
+
"loss": 0.2076,
|
3463 |
+
"step": 490
|
3464 |
+
},
|
3465 |
+
{
|
3466 |
+
"epoch": 0.5125260960334029,
|
3467 |
+
"grad_norm": 0.6702793836593628,
|
3468 |
+
"learning_rate": 9.768046663340669e-05,
|
3469 |
+
"loss": 0.2207,
|
3470 |
+
"step": 491
|
3471 |
+
},
|
3472 |
+
{
|
3473 |
+
"epoch": 0.5135699373695198,
|
3474 |
+
"grad_norm": 0.6283150911331177,
|
3475 |
+
"learning_rate": 9.734917750650816e-05,
|
3476 |
+
"loss": 0.2246,
|
3477 |
+
"step": 492
|
3478 |
+
},
|
3479 |
+
{
|
3480 |
+
"epoch": 0.5146137787056367,
|
3481 |
+
"grad_norm": 0.6348150968551636,
|
3482 |
+
"learning_rate": 9.701791749102495e-05,
|
3483 |
+
"loss": 0.1896,
|
3484 |
+
"step": 493
|
3485 |
+
},
|
3486 |
+
{
|
3487 |
+
"epoch": 0.5156576200417536,
|
3488 |
+
"grad_norm": 0.6684585213661194,
|
3489 |
+
"learning_rate": 9.668669022486494e-05,
|
3490 |
+
"loss": 0.2409,
|
3491 |
+
"step": 494
|
3492 |
+
},
|
3493 |
+
{
|
3494 |
+
"epoch": 0.5167014613778705,
|
3495 |
+
"grad_norm": 0.678677499294281,
|
3496 |
+
"learning_rate": 9.635549934557625e-05,
|
3497 |
+
"loss": 0.216,
|
3498 |
+
"step": 495
|
3499 |
+
},
|
3500 |
+
{
|
3501 |
+
"epoch": 0.5177453027139874,
|
3502 |
+
"grad_norm": 0.6523580551147461,
|
3503 |
+
"learning_rate": 9.602434849030745e-05,
|
3504 |
+
"loss": 0.1894,
|
3505 |
+
"step": 496
|
3506 |
+
},
|
3507 |
+
{
|
3508 |
+
"epoch": 0.5187891440501043,
|
3509 |
+
"grad_norm": 0.5122499465942383,
|
3510 |
+
"learning_rate": 9.569324129576757e-05,
|
3511 |
+
"loss": 0.1579,
|
3512 |
+
"step": 497
|
3513 |
+
},
|
3514 |
+
{
|
3515 |
+
"epoch": 0.5198329853862212,
|
3516 |
+
"grad_norm": 0.5820009112358093,
|
3517 |
+
"learning_rate": 9.536218139818614e-05,
|
3518 |
+
"loss": 0.1766,
|
3519 |
+
"step": 498
|
3520 |
+
},
|
3521 |
+
{
|
3522 |
+
"epoch": 0.5208768267223383,
|
3523 |
+
"grad_norm": 0.5032172203063965,
|
3524 |
+
"learning_rate": 9.503117243327337e-05,
|
3525 |
+
"loss": 0.1519,
|
3526 |
+
"step": 499
|
3527 |
+
},
|
3528 |
+
{
|
3529 |
+
"epoch": 0.5219206680584552,
|
3530 |
+
"grad_norm": 0.7425169944763184,
|
3531 |
+
"learning_rate": 9.47002180361799e-05,
|
3532 |
+
"loss": 0.1333,
|
3533 |
+
"step": 500
|
3534 |
+
},
|
3535 |
+
{
|
3536 |
+
"epoch": 0.5229645093945721,
|
3537 |
+
"grad_norm": 0.7141383290290833,
|
3538 |
+
"learning_rate": 9.436932184145737e-05,
|
3539 |
+
"loss": 0.4269,
|
3540 |
+
"step": 501
|
3541 |
+
},
|
3542 |
+
{
|
3543 |
+
"epoch": 0.524008350730689,
|
3544 |
+
"grad_norm": 0.662886917591095,
|
3545 |
+
"learning_rate": 9.403848748301802e-05,
|
3546 |
+
"loss": 0.2939,
|
3547 |
+
"step": 502
|
3548 |
+
},
|
3549 |
+
{
|
3550 |
+
"epoch": 0.5250521920668059,
|
3551 |
+
"grad_norm": 0.6695585250854492,
|
3552 |
+
"learning_rate": 9.370771859409513e-05,
|
3553 |
+
"loss": 0.3167,
|
3554 |
+
"step": 503
|
3555 |
+
},
|
3556 |
+
{
|
3557 |
+
"epoch": 0.5260960334029228,
|
3558 |
+
"grad_norm": 0.7355087399482727,
|
3559 |
+
"learning_rate": 9.337701880720303e-05,
|
3560 |
+
"loss": 0.3414,
|
3561 |
+
"step": 504
|
3562 |
+
},
|
3563 |
+
{
|
3564 |
+
"epoch": 0.5271398747390397,
|
3565 |
+
"grad_norm": 0.6738423705101013,
|
3566 |
+
"learning_rate": 9.304639175409698e-05,
|
3567 |
+
"loss": 0.3176,
|
3568 |
+
"step": 505
|
3569 |
+
},
|
3570 |
+
{
|
3571 |
+
"epoch": 0.5281837160751566,
|
3572 |
+
"grad_norm": 0.7200036644935608,
|
3573 |
+
"learning_rate": 9.271584106573364e-05,
|
3574 |
+
"loss": 0.3276,
|
3575 |
+
"step": 506
|
3576 |
+
},
|
3577 |
+
{
|
3578 |
+
"epoch": 0.5292275574112735,
|
3579 |
+
"grad_norm": 0.6562217473983765,
|
3580 |
+
"learning_rate": 9.238537037223104e-05,
|
3581 |
+
"loss": 0.3353,
|
3582 |
+
"step": 507
|
3583 |
+
},
|
3584 |
+
{
|
3585 |
+
"epoch": 0.5302713987473904,
|
3586 |
+
"grad_norm": 0.6065823435783386,
|
3587 |
+
"learning_rate": 9.205498330282856e-05,
|
3588 |
+
"loss": 0.3062,
|
3589 |
+
"step": 508
|
3590 |
+
},
|
3591 |
+
{
|
3592 |
+
"epoch": 0.5313152400835073,
|
3593 |
+
"grad_norm": 0.6631349921226501,
|
3594 |
+
"learning_rate": 9.172468348584739e-05,
|
3595 |
+
"loss": 0.2497,
|
3596 |
+
"step": 509
|
3597 |
+
},
|
3598 |
+
{
|
3599 |
+
"epoch": 0.5323590814196242,
|
3600 |
+
"grad_norm": 0.613549530506134,
|
3601 |
+
"learning_rate": 9.139447454865033e-05,
|
3602 |
+
"loss": 0.2625,
|
3603 |
+
"step": 510
|
3604 |
+
},
|
3605 |
+
{
|
3606 |
+
"epoch": 0.5334029227557411,
|
3607 |
+
"grad_norm": 0.6613410711288452,
|
3608 |
+
"learning_rate": 9.106436011760229e-05,
|
3609 |
+
"loss": 0.244,
|
3610 |
+
"step": 511
|
3611 |
+
},
|
3612 |
+
{
|
3613 |
+
"epoch": 0.534446764091858,
|
3614 |
+
"grad_norm": 0.5914390087127686,
|
3615 |
+
"learning_rate": 9.073434381803024e-05,
|
3616 |
+
"loss": 0.2109,
|
3617 |
+
"step": 512
|
3618 |
+
},
|
3619 |
+
{
|
3620 |
+
"epoch": 0.535490605427975,
|
3621 |
+
"grad_norm": 0.5324224829673767,
|
3622 |
+
"learning_rate": 9.04044292741835e-05,
|
3623 |
+
"loss": 0.1926,
|
3624 |
+
"step": 513
|
3625 |
+
},
|
3626 |
+
{
|
3627 |
+
"epoch": 0.5365344467640919,
|
3628 |
+
"grad_norm": 0.652651846408844,
|
3629 |
+
"learning_rate": 9.007462010919386e-05,
|
3630 |
+
"loss": 0.2254,
|
3631 |
+
"step": 514
|
3632 |
+
},
|
3633 |
+
{
|
3634 |
+
"epoch": 0.5375782881002088,
|
3635 |
+
"grad_norm": 0.6112195253372192,
|
3636 |
+
"learning_rate": 8.974491994503584e-05,
|
3637 |
+
"loss": 0.205,
|
3638 |
+
"step": 515
|
3639 |
+
},
|
3640 |
+
{
|
3641 |
+
"epoch": 0.5386221294363257,
|
3642 |
+
"grad_norm": 0.5404685735702515,
|
3643 |
+
"learning_rate": 8.941533240248699e-05,
|
3644 |
+
"loss": 0.1928,
|
3645 |
+
"step": 516
|
3646 |
+
},
|
3647 |
+
{
|
3648 |
+
"epoch": 0.5396659707724426,
|
3649 |
+
"grad_norm": 0.47451335191726685,
|
3650 |
+
"learning_rate": 8.908586110108794e-05,
|
3651 |
+
"loss": 0.1307,
|
3652 |
+
"step": 517
|
3653 |
+
},
|
3654 |
+
{
|
3655 |
+
"epoch": 0.5407098121085595,
|
3656 |
+
"grad_norm": 0.8480343818664551,
|
3657 |
+
"learning_rate": 8.875650965910279e-05,
|
3658 |
+
"loss": 0.1533,
|
3659 |
+
"step": 518
|
3660 |
+
},
|
3661 |
+
{
|
3662 |
+
"epoch": 0.5417536534446764,
|
3663 |
+
"grad_norm": 0.6759589314460754,
|
3664 |
+
"learning_rate": 8.842728169347939e-05,
|
3665 |
+
"loss": 0.1468,
|
3666 |
+
"step": 519
|
3667 |
+
},
|
3668 |
+
{
|
3669 |
+
"epoch": 0.5427974947807933,
|
3670 |
+
"grad_norm": 0.5591132640838623,
|
3671 |
+
"learning_rate": 8.809818081980953e-05,
|
3672 |
+
"loss": 0.2246,
|
3673 |
+
"step": 520
|
3674 |
+
},
|
3675 |
+
{
|
3676 |
+
"epoch": 0.5438413361169102,
|
3677 |
+
"grad_norm": 0.6184394955635071,
|
3678 |
+
"learning_rate": 8.776921065228924e-05,
|
3679 |
+
"loss": 0.2189,
|
3680 |
+
"step": 521
|
3681 |
+
},
|
3682 |
+
{
|
3683 |
+
"epoch": 0.5448851774530271,
|
3684 |
+
"grad_norm": 0.5175319910049438,
|
3685 |
+
"learning_rate": 8.744037480367921e-05,
|
3686 |
+
"loss": 0.1886,
|
3687 |
+
"step": 522
|
3688 |
+
},
|
3689 |
+
{
|
3690 |
+
"epoch": 0.545929018789144,
|
3691 |
+
"grad_norm": 0.645250678062439,
|
3692 |
+
"learning_rate": 8.711167688526493e-05,
|
3693 |
+
"loss": 0.2297,
|
3694 |
+
"step": 523
|
3695 |
+
},
|
3696 |
+
{
|
3697 |
+
"epoch": 0.5469728601252609,
|
3698 |
+
"grad_norm": 0.6044825315475464,
|
3699 |
+
"learning_rate": 8.678312050681724e-05,
|
3700 |
+
"loss": 0.2029,
|
3701 |
+
"step": 524
|
3702 |
+
},
|
3703 |
+
{
|
3704 |
+
"epoch": 0.5480167014613778,
|
3705 |
+
"grad_norm": 0.5178519487380981,
|
3706 |
+
"learning_rate": 8.645470927655255e-05,
|
3707 |
+
"loss": 0.1447,
|
3708 |
+
"step": 525
|
3709 |
+
},
|
3710 |
+
{
|
3711 |
+
"epoch": 0.5490605427974948,
|
3712 |
+
"grad_norm": 0.8626076579093933,
|
3713 |
+
"learning_rate": 8.612644680109319e-05,
|
3714 |
+
"loss": 0.4495,
|
3715 |
+
"step": 526
|
3716 |
+
},
|
3717 |
+
{
|
3718 |
+
"epoch": 0.5501043841336117,
|
3719 |
+
"grad_norm": 0.8363009691238403,
|
3720 |
+
"learning_rate": 8.579833668542796e-05,
|
3721 |
+
"loss": 0.3709,
|
3722 |
+
"step": 527
|
3723 |
+
},
|
3724 |
+
{
|
3725 |
+
"epoch": 0.5511482254697286,
|
3726 |
+
"grad_norm": 0.872733473777771,
|
3727 |
+
"learning_rate": 8.547038253287233e-05,
|
3728 |
+
"loss": 0.3226,
|
3729 |
+
"step": 528
|
3730 |
+
},
|
3731 |
+
{
|
3732 |
+
"epoch": 0.5521920668058455,
|
3733 |
+
"grad_norm": 0.865210235118866,
|
3734 |
+
"learning_rate": 8.514258794502905e-05,
|
3735 |
+
"loss": 0.3579,
|
3736 |
+
"step": 529
|
3737 |
+
},
|
3738 |
+
{
|
3739 |
+
"epoch": 0.5532359081419624,
|
3740 |
+
"grad_norm": 0.7914073467254639,
|
3741 |
+
"learning_rate": 8.481495652174859e-05,
|
3742 |
+
"loss": 0.3216,
|
3743 |
+
"step": 530
|
3744 |
+
},
|
3745 |
+
{
|
3746 |
+
"epoch": 0.5542797494780793,
|
3747 |
+
"grad_norm": 0.8779425024986267,
|
3748 |
+
"learning_rate": 8.448749186108935e-05,
|
3749 |
+
"loss": 0.3532,
|
3750 |
+
"step": 531
|
3751 |
+
},
|
3752 |
+
{
|
3753 |
+
"epoch": 0.5553235908141962,
|
3754 |
+
"grad_norm": 0.8584082722663879,
|
3755 |
+
"learning_rate": 8.416019755927851e-05,
|
3756 |
+
"loss": 0.4009,
|
3757 |
+
"step": 532
|
3758 |
+
},
|
3759 |
+
{
|
3760 |
+
"epoch": 0.5563674321503131,
|
3761 |
+
"grad_norm": 0.8470184206962585,
|
3762 |
+
"learning_rate": 8.383307721067231e-05,
|
3763 |
+
"loss": 0.4198,
|
3764 |
+
"step": 533
|
3765 |
+
},
|
3766 |
+
{
|
3767 |
+
"epoch": 0.55741127348643,
|
3768 |
+
"grad_norm": 0.7338582277297974,
|
3769 |
+
"learning_rate": 8.35061344077166e-05,
|
3770 |
+
"loss": 0.2725,
|
3771 |
+
"step": 534
|
3772 |
+
},
|
3773 |
+
{
|
3774 |
+
"epoch": 0.558455114822547,
|
3775 |
+
"grad_norm": 0.7652982473373413,
|
3776 |
+
"learning_rate": 8.317937274090747e-05,
|
3777 |
+
"loss": 0.3007,
|
3778 |
+
"step": 535
|
3779 |
+
},
|
3780 |
+
{
|
3781 |
+
"epoch": 0.5594989561586639,
|
3782 |
+
"grad_norm": 0.7415357232093811,
|
3783 |
+
"learning_rate": 8.28527957987517e-05,
|
3784 |
+
"loss": 0.3274,
|
3785 |
+
"step": 536
|
3786 |
+
},
|
3787 |
+
{
|
3788 |
+
"epoch": 0.5605427974947808,
|
3789 |
+
"grad_norm": 0.6662179231643677,
|
3790 |
+
"learning_rate": 8.252640716772749e-05,
|
3791 |
+
"loss": 0.2606,
|
3792 |
+
"step": 537
|
3793 |
+
},
|
3794 |
+
{
|
3795 |
+
"epoch": 0.5615866388308977,
|
3796 |
+
"grad_norm": 0.6139498353004456,
|
3797 |
+
"learning_rate": 8.2200210432245e-05,
|
3798 |
+
"loss": 0.2382,
|
3799 |
+
"step": 538
|
3800 |
+
},
|
3801 |
+
{
|
3802 |
+
"epoch": 0.5626304801670147,
|
3803 |
+
"grad_norm": 0.7014831900596619,
|
3804 |
+
"learning_rate": 8.187420917460686e-05,
|
3805 |
+
"loss": 0.2542,
|
3806 |
+
"step": 539
|
3807 |
+
},
|
3808 |
+
{
|
3809 |
+
"epoch": 0.5636743215031316,
|
3810 |
+
"grad_norm": 0.7138461470603943,
|
3811 |
+
"learning_rate": 8.154840697496917e-05,
|
3812 |
+
"loss": 0.2392,
|
3813 |
+
"step": 540
|
3814 |
+
},
|
3815 |
+
{
|
3816 |
+
"epoch": 0.5647181628392485,
|
3817 |
+
"grad_norm": 0.7507902979850769,
|
3818 |
+
"learning_rate": 8.122280741130176e-05,
|
3819 |
+
"loss": 0.2554,
|
3820 |
+
"step": 541
|
3821 |
+
},
|
3822 |
+
{
|
3823 |
+
"epoch": 0.5657620041753654,
|
3824 |
+
"grad_norm": 0.8535422086715698,
|
3825 |
+
"learning_rate": 8.089741405934922e-05,
|
3826 |
+
"loss": 0.2433,
|
3827 |
+
"step": 542
|
3828 |
+
},
|
3829 |
+
{
|
3830 |
+
"epoch": 0.5668058455114823,
|
3831 |
+
"grad_norm": 0.660111129283905,
|
3832 |
+
"learning_rate": 8.057223049259155e-05,
|
3833 |
+
"loss": 0.1921,
|
3834 |
+
"step": 543
|
3835 |
+
},
|
3836 |
+
{
|
3837 |
+
"epoch": 0.5678496868475992,
|
3838 |
+
"grad_norm": 0.49220120906829834,
|
3839 |
+
"learning_rate": 8.024726028220474e-05,
|
3840 |
+
"loss": 0.1793,
|
3841 |
+
"step": 544
|
3842 |
+
},
|
3843 |
+
{
|
3844 |
+
"epoch": 0.5688935281837161,
|
3845 |
+
"grad_norm": 0.5934033393859863,
|
3846 |
+
"learning_rate": 7.992250699702182e-05,
|
3847 |
+
"loss": 0.1686,
|
3848 |
+
"step": 545
|
3849 |
+
},
|
3850 |
+
{
|
3851 |
+
"epoch": 0.569937369519833,
|
3852 |
+
"grad_norm": 0.6598916053771973,
|
3853 |
+
"learning_rate": 7.959797420349355e-05,
|
3854 |
+
"loss": 0.1981,
|
3855 |
+
"step": 546
|
3856 |
+
},
|
3857 |
+
{
|
3858 |
+
"epoch": 0.5709812108559499,
|
3859 |
+
"grad_norm": 0.6087566018104553,
|
3860 |
+
"learning_rate": 7.927366546564911e-05,
|
3861 |
+
"loss": 0.1845,
|
3862 |
+
"step": 547
|
3863 |
+
},
|
3864 |
+
{
|
3865 |
+
"epoch": 0.5720250521920668,
|
3866 |
+
"grad_norm": 0.4998890459537506,
|
3867 |
+
"learning_rate": 7.894958434505725e-05,
|
3868 |
+
"loss": 0.1524,
|
3869 |
+
"step": 548
|
3870 |
+
},
|
3871 |
+
{
|
3872 |
+
"epoch": 0.5730688935281837,
|
3873 |
+
"grad_norm": 0.5460024476051331,
|
3874 |
+
"learning_rate": 7.862573440078686e-05,
|
3875 |
+
"loss": 0.1808,
|
3876 |
+
"step": 549
|
3877 |
+
},
|
3878 |
+
{
|
3879 |
+
"epoch": 0.5741127348643006,
|
3880 |
+
"grad_norm": 0.7462297677993774,
|
3881 |
+
"learning_rate": 7.83021191893682e-05,
|
3882 |
+
"loss": 0.1723,
|
3883 |
+
"step": 550
|
3884 |
+
},
|
3885 |
+
{
|
3886 |
+
"epoch": 0.5751565762004175,
|
3887 |
+
"grad_norm": 0.5173273086547852,
|
3888 |
+
"learning_rate": 7.797874226475361e-05,
|
3889 |
+
"loss": 0.3054,
|
3890 |
+
"step": 551
|
3891 |
+
},
|
3892 |
+
{
|
3893 |
+
"epoch": 0.5762004175365344,
|
3894 |
+
"grad_norm": 0.6547046303749084,
|
3895 |
+
"learning_rate": 7.765560717827858e-05,
|
3896 |
+
"loss": 0.3101,
|
3897 |
+
"step": 552
|
3898 |
+
},
|
3899 |
+
{
|
3900 |
+
"epoch": 0.5772442588726514,
|
3901 |
+
"grad_norm": 0.676986575126648,
|
3902 |
+
"learning_rate": 7.733271747862265e-05,
|
3903 |
+
"loss": 0.3376,
|
3904 |
+
"step": 553
|
3905 |
+
},
|
3906 |
+
{
|
3907 |
+
"epoch": 0.5782881002087683,
|
3908 |
+
"grad_norm": 0.7368578910827637,
|
3909 |
+
"learning_rate": 7.701007671177067e-05,
|
3910 |
+
"loss": 0.3517,
|
3911 |
+
"step": 554
|
3912 |
+
},
|
3913 |
+
{
|
3914 |
+
"epoch": 0.5793319415448852,
|
3915 |
+
"grad_norm": 0.7136873006820679,
|
3916 |
+
"learning_rate": 7.668768842097353e-05,
|
3917 |
+
"loss": 0.3077,
|
3918 |
+
"step": 555
|
3919 |
+
},
|
3920 |
+
{
|
3921 |
+
"epoch": 0.5803757828810021,
|
3922 |
+
"grad_norm": 0.7121712565422058,
|
3923 |
+
"learning_rate": 7.636555614670953e-05,
|
3924 |
+
"loss": 0.3271,
|
3925 |
+
"step": 556
|
3926 |
+
},
|
3927 |
+
{
|
3928 |
+
"epoch": 0.581419624217119,
|
3929 |
+
"grad_norm": 0.7123695611953735,
|
3930 |
+
"learning_rate": 7.604368342664533e-05,
|
3931 |
+
"loss": 0.3356,
|
3932 |
+
"step": 557
|
3933 |
+
},
|
3934 |
+
{
|
3935 |
+
"epoch": 0.5824634655532359,
|
3936 |
+
"grad_norm": 0.7206712961196899,
|
3937 |
+
"learning_rate": 7.572207379559721e-05,
|
3938 |
+
"loss": 0.2915,
|
3939 |
+
"step": 558
|
3940 |
+
},
|
3941 |
+
{
|
3942 |
+
"epoch": 0.5835073068893528,
|
3943 |
+
"grad_norm": 0.6520224809646606,
|
3944 |
+
"learning_rate": 7.540073078549221e-05,
|
3945 |
+
"loss": 0.2657,
|
3946 |
+
"step": 559
|
3947 |
+
},
|
3948 |
+
{
|
3949 |
+
"epoch": 0.5845511482254697,
|
3950 |
+
"grad_norm": 0.5960420370101929,
|
3951 |
+
"learning_rate": 7.507965792532921e-05,
|
3952 |
+
"loss": 0.1952,
|
3953 |
+
"step": 560
|
3954 |
+
},
|
3955 |
+
{
|
3956 |
+
"epoch": 0.5855949895615866,
|
3957 |
+
"grad_norm": 0.5834378004074097,
|
3958 |
+
"learning_rate": 7.475885874114047e-05,
|
3959 |
+
"loss": 0.1878,
|
3960 |
+
"step": 561
|
3961 |
+
},
|
3962 |
+
{
|
3963 |
+
"epoch": 0.5866388308977035,
|
3964 |
+
"grad_norm": 0.6201406121253967,
|
3965 |
+
"learning_rate": 7.443833675595255e-05,
|
3966 |
+
"loss": 0.1927,
|
3967 |
+
"step": 562
|
3968 |
+
},
|
3969 |
+
{
|
3970 |
+
"epoch": 0.5876826722338204,
|
3971 |
+
"grad_norm": 0.5904473662376404,
|
3972 |
+
"learning_rate": 7.411809548974792e-05,
|
3973 |
+
"loss": 0.1804,
|
3974 |
+
"step": 563
|
3975 |
+
},
|
3976 |
+
{
|
3977 |
+
"epoch": 0.5887265135699373,
|
3978 |
+
"grad_norm": 0.5292779803276062,
|
3979 |
+
"learning_rate": 7.379813845942623e-05,
|
3980 |
+
"loss": 0.1379,
|
3981 |
+
"step": 564
|
3982 |
+
},
|
3983 |
+
{
|
3984 |
+
"epoch": 0.5897703549060542,
|
3985 |
+
"grad_norm": 0.618929922580719,
|
3986 |
+
"learning_rate": 7.347846917876544e-05,
|
3987 |
+
"loss": 0.1926,
|
3988 |
+
"step": 565
|
3989 |
+
},
|
3990 |
+
{
|
3991 |
+
"epoch": 0.5908141962421712,
|
3992 |
+
"grad_norm": 0.5607888698577881,
|
3993 |
+
"learning_rate": 7.315909115838367e-05,
|
3994 |
+
"loss": 0.1845,
|
3995 |
+
"step": 566
|
3996 |
+
},
|
3997 |
+
{
|
3998 |
+
"epoch": 0.5918580375782881,
|
3999 |
+
"grad_norm": 0.56803297996521,
|
4000 |
+
"learning_rate": 7.284000790570029e-05,
|
4001 |
+
"loss": 0.1762,
|
4002 |
+
"step": 567
|
4003 |
+
},
|
4004 |
+
{
|
4005 |
+
"epoch": 0.592901878914405,
|
4006 |
+
"grad_norm": 0.5143932700157166,
|
4007 |
+
"learning_rate": 7.252122292489747e-05,
|
4008 |
+
"loss": 0.1514,
|
4009 |
+
"step": 568
|
4010 |
+
},
|
4011 |
+
{
|
4012 |
+
"epoch": 0.593945720250522,
|
4013 |
+
"grad_norm": 0.6080281734466553,
|
4014 |
+
"learning_rate": 7.220273971688192e-05,
|
4015 |
+
"loss": 0.2516,
|
4016 |
+
"step": 569
|
4017 |
+
},
|
4018 |
+
{
|
4019 |
+
"epoch": 0.5949895615866388,
|
4020 |
+
"grad_norm": 0.6721866130828857,
|
4021 |
+
"learning_rate": 7.188456177924605e-05,
|
4022 |
+
"loss": 0.2707,
|
4023 |
+
"step": 570
|
4024 |
+
},
|
4025 |
+
{
|
4026 |
+
"epoch": 0.5960334029227558,
|
4027 |
+
"grad_norm": 0.5116624236106873,
|
4028 |
+
"learning_rate": 7.156669260622996e-05,
|
4029 |
+
"loss": 0.2083,
|
4030 |
+
"step": 571
|
4031 |
+
},
|
4032 |
+
{
|
4033 |
+
"epoch": 0.5970772442588727,
|
4034 |
+
"grad_norm": 0.5874140858650208,
|
4035 |
+
"learning_rate": 7.124913568868287e-05,
|
4036 |
+
"loss": 0.1598,
|
4037 |
+
"step": 572
|
4038 |
+
},
|
4039 |
+
{
|
4040 |
+
"epoch": 0.5981210855949896,
|
4041 |
+
"grad_norm": 0.506820559501648,
|
4042 |
+
"learning_rate": 7.093189451402469e-05,
|
4043 |
+
"loss": 0.1572,
|
4044 |
+
"step": 573
|
4045 |
+
},
|
4046 |
+
{
|
4047 |
+
"epoch": 0.5991649269311065,
|
4048 |
+
"grad_norm": 0.7033873796463013,
|
4049 |
+
"learning_rate": 7.061497256620793e-05,
|
4050 |
+
"loss": 0.2867,
|
4051 |
+
"step": 574
|
4052 |
+
},
|
4053 |
+
{
|
4054 |
+
"epoch": 0.6002087682672234,
|
4055 |
+
"grad_norm": 0.7932460904121399,
|
4056 |
+
"learning_rate": 7.029837332567927e-05,
|
4057 |
+
"loss": 0.2998,
|
4058 |
+
"step": 575
|
4059 |
+
},
|
4060 |
+
{
|
4061 |
+
"epoch": 0.6012526096033403,
|
4062 |
+
"grad_norm": 0.8427619934082031,
|
4063 |
+
"learning_rate": 6.998210026934148e-05,
|
4064 |
+
"loss": 0.426,
|
4065 |
+
"step": 576
|
4066 |
+
},
|
4067 |
+
{
|
4068 |
+
"epoch": 0.6022964509394572,
|
4069 |
+
"grad_norm": 0.7818666100502014,
|
4070 |
+
"learning_rate": 6.966615687051516e-05,
|
4071 |
+
"loss": 0.3559,
|
4072 |
+
"step": 577
|
4073 |
+
},
|
4074 |
+
{
|
4075 |
+
"epoch": 0.6033402922755741,
|
4076 |
+
"grad_norm": 0.821897566318512,
|
4077 |
+
"learning_rate": 6.935054659890052e-05,
|
4078 |
+
"loss": 0.3928,
|
4079 |
+
"step": 578
|
4080 |
+
},
|
4081 |
+
{
|
4082 |
+
"epoch": 0.6043841336116911,
|
4083 |
+
"grad_norm": 0.7375624179840088,
|
4084 |
+
"learning_rate": 6.903527292053942e-05,
|
4085 |
+
"loss": 0.3203,
|
4086 |
+
"step": 579
|
4087 |
+
},
|
4088 |
+
{
|
4089 |
+
"epoch": 0.605427974947808,
|
4090 |
+
"grad_norm": 0.7323412299156189,
|
4091 |
+
"learning_rate": 6.87203392977773e-05,
|
4092 |
+
"loss": 0.3014,
|
4093 |
+
"step": 580
|
4094 |
+
},
|
4095 |
+
{
|
4096 |
+
"epoch": 0.6064718162839249,
|
4097 |
+
"grad_norm": 0.8238475322723389,
|
4098 |
+
"learning_rate": 6.840574918922493e-05,
|
4099 |
+
"loss": 0.3447,
|
4100 |
+
"step": 581
|
4101 |
+
},
|
4102 |
+
{
|
4103 |
+
"epoch": 0.6075156576200418,
|
4104 |
+
"grad_norm": 0.7970190644264221,
|
4105 |
+
"learning_rate": 6.809150604972079e-05,
|
4106 |
+
"loss": 0.3556,
|
4107 |
+
"step": 582
|
4108 |
+
},
|
4109 |
+
{
|
4110 |
+
"epoch": 0.6085594989561587,
|
4111 |
+
"grad_norm": 0.718948483467102,
|
4112 |
+
"learning_rate": 6.777761333029275e-05,
|
4113 |
+
"loss": 0.318,
|
4114 |
+
"step": 583
|
4115 |
+
},
|
4116 |
+
{
|
4117 |
+
"epoch": 0.6096033402922756,
|
4118 |
+
"grad_norm": 0.8113385438919067,
|
4119 |
+
"learning_rate": 6.746407447812049e-05,
|
4120 |
+
"loss": 0.2928,
|
4121 |
+
"step": 584
|
4122 |
+
},
|
4123 |
+
{
|
4124 |
+
"epoch": 0.6106471816283925,
|
4125 |
+
"grad_norm": 0.732028603553772,
|
4126 |
+
"learning_rate": 6.715089293649752e-05,
|
4127 |
+
"loss": 0.2688,
|
4128 |
+
"step": 585
|
4129 |
+
},
|
4130 |
+
{
|
4131 |
+
"epoch": 0.6116910229645094,
|
4132 |
+
"grad_norm": 0.693304181098938,
|
4133 |
+
"learning_rate": 6.683807214479323e-05,
|
4134 |
+
"loss": 0.2421,
|
4135 |
+
"step": 586
|
4136 |
+
},
|
4137 |
+
{
|
4138 |
+
"epoch": 0.6127348643006263,
|
4139 |
+
"grad_norm": 0.8302125334739685,
|
4140 |
+
"learning_rate": 6.652561553841537e-05,
|
4141 |
+
"loss": 0.2625,
|
4142 |
+
"step": 587
|
4143 |
+
},
|
4144 |
+
{
|
4145 |
+
"epoch": 0.6137787056367432,
|
4146 |
+
"grad_norm": 0.6446481943130493,
|
4147 |
+
"learning_rate": 6.621352654877207e-05,
|
4148 |
+
"loss": 0.2261,
|
4149 |
+
"step": 588
|
4150 |
+
},
|
4151 |
+
{
|
4152 |
+
"epoch": 0.6148225469728601,
|
4153 |
+
"grad_norm": 0.7838292717933655,
|
4154 |
+
"learning_rate": 6.59018086032344e-05,
|
4155 |
+
"loss": 0.2619,
|
4156 |
+
"step": 589
|
4157 |
+
},
|
4158 |
+
{
|
4159 |
+
"epoch": 0.615866388308977,
|
4160 |
+
"grad_norm": 0.6317050457000732,
|
4161 |
+
"learning_rate": 6.55904651250986e-05,
|
4162 |
+
"loss": 0.1996,
|
4163 |
+
"step": 590
|
4164 |
+
},
|
4165 |
+
{
|
4166 |
+
"epoch": 0.6169102296450939,
|
4167 |
+
"grad_norm": 0.6110920310020447,
|
4168 |
+
"learning_rate": 6.527949953354835e-05,
|
4169 |
+
"loss": 0.2078,
|
4170 |
+
"step": 591
|
4171 |
+
},
|
4172 |
+
{
|
4173 |
+
"epoch": 0.6179540709812108,
|
4174 |
+
"grad_norm": 0.5425273180007935,
|
4175 |
+
"learning_rate": 6.496891524361757e-05,
|
4176 |
+
"loss": 0.1852,
|
4177 |
+
"step": 592
|
4178 |
+
},
|
4179 |
+
{
|
4180 |
+
"epoch": 0.6189979123173278,
|
4181 |
+
"grad_norm": 0.7897228002548218,
|
4182 |
+
"learning_rate": 6.465871566615263e-05,
|
4183 |
+
"loss": 0.2217,
|
4184 |
+
"step": 593
|
4185 |
+
},
|
4186 |
+
{
|
4187 |
+
"epoch": 0.6200417536534447,
|
4188 |
+
"grad_norm": 0.6448274254798889,
|
4189 |
+
"learning_rate": 6.434890420777491e-05,
|
4190 |
+
"loss": 0.1918,
|
4191 |
+
"step": 594
|
4192 |
+
},
|
4193 |
+
{
|
4194 |
+
"epoch": 0.6210855949895616,
|
4195 |
+
"grad_norm": 0.690799355506897,
|
4196 |
+
"learning_rate": 6.403948427084356e-05,
|
4197 |
+
"loss": 0.2129,
|
4198 |
+
"step": 595
|
4199 |
+
},
|
4200 |
+
{
|
4201 |
+
"epoch": 0.6221294363256785,
|
4202 |
+
"grad_norm": 0.6128472685813904,
|
4203 |
+
"learning_rate": 6.373045925341794e-05,
|
4204 |
+
"loss": 0.196,
|
4205 |
+
"step": 596
|
4206 |
+
},
|
4207 |
+
{
|
4208 |
+
"epoch": 0.6231732776617954,
|
4209 |
+
"grad_norm": 0.4894169270992279,
|
4210 |
+
"learning_rate": 6.342183254922046e-05,
|
4211 |
+
"loss": 0.1341,
|
4212 |
+
"step": 597
|
4213 |
+
},
|
4214 |
+
{
|
4215 |
+
"epoch": 0.6242171189979123,
|
4216 |
+
"grad_norm": 0.5475450754165649,
|
4217 |
+
"learning_rate": 6.311360754759923e-05,
|
4218 |
+
"loss": 0.1655,
|
4219 |
+
"step": 598
|
4220 |
+
},
|
4221 |
+
{
|
4222 |
+
"epoch": 0.6252609603340292,
|
4223 |
+
"grad_norm": 0.5742066502571106,
|
4224 |
+
"learning_rate": 6.280578763349078e-05,
|
4225 |
+
"loss": 0.2158,
|
4226 |
+
"step": 599
|
4227 |
+
},
|
4228 |
+
{
|
4229 |
+
"epoch": 0.6263048016701461,
|
4230 |
+
"grad_norm": 0.6302378177642822,
|
4231 |
+
"learning_rate": 6.249837618738311e-05,
|
4232 |
+
"loss": 0.1211,
|
4233 |
+
"step": 600
|
4234 |
+
},
|
4235 |
+
{
|
4236 |
+
"epoch": 0.627348643006263,
|
4237 |
+
"grad_norm": 0.6498920321464539,
|
4238 |
+
"learning_rate": 6.219137658527818e-05,
|
4239 |
+
"loss": 0.3415,
|
4240 |
+
"step": 601
|
4241 |
+
},
|
4242 |
+
{
|
4243 |
+
"epoch": 0.6283924843423799,
|
4244 |
+
"grad_norm": 0.5803071856498718,
|
4245 |
+
"learning_rate": 6.188479219865529e-05,
|
4246 |
+
"loss": 0.3446,
|
4247 |
+
"step": 602
|
4248 |
+
},
|
4249 |
+
{
|
4250 |
+
"epoch": 0.6294363256784968,
|
4251 |
+
"grad_norm": 0.7242146134376526,
|
4252 |
+
"learning_rate": 6.157862639443374e-05,
|
4253 |
+
"loss": 0.3406,
|
4254 |
+
"step": 603
|
4255 |
+
},
|
4256 |
+
{
|
4257 |
+
"epoch": 0.6304801670146137,
|
4258 |
+
"grad_norm": 0.741543710231781,
|
4259 |
+
"learning_rate": 6.127288253493591e-05,
|
4260 |
+
"loss": 0.2851,
|
4261 |
+
"step": 604
|
4262 |
+
},
|
4263 |
+
{
|
4264 |
+
"epoch": 0.6315240083507306,
|
4265 |
+
"grad_norm": 0.6710807681083679,
|
4266 |
+
"learning_rate": 6.09675639778504e-05,
|
4267 |
+
"loss": 0.2975,
|
4268 |
+
"step": 605
|
4269 |
+
},
|
4270 |
+
{
|
4271 |
+
"epoch": 0.6325678496868476,
|
4272 |
+
"grad_norm": 0.6401992440223694,
|
4273 |
+
"learning_rate": 6.0662674076195194e-05,
|
4274 |
+
"loss": 0.2934,
|
4275 |
+
"step": 606
|
4276 |
+
},
|
4277 |
+
{
|
4278 |
+
"epoch": 0.6336116910229646,
|
4279 |
+
"grad_norm": 0.7866775393486023,
|
4280 |
+
"learning_rate": 6.03582161782806e-05,
|
4281 |
+
"loss": 0.3303,
|
4282 |
+
"step": 607
|
4283 |
+
},
|
4284 |
+
{
|
4285 |
+
"epoch": 0.6346555323590815,
|
4286 |
+
"grad_norm": 0.6878888607025146,
|
4287 |
+
"learning_rate": 6.005419362767286e-05,
|
4288 |
+
"loss": 0.2885,
|
4289 |
+
"step": 608
|
4290 |
+
},
|
4291 |
+
{
|
4292 |
+
"epoch": 0.6356993736951984,
|
4293 |
+
"grad_norm": 0.667226254940033,
|
4294 |
+
"learning_rate": 5.975060976315703e-05,
|
4295 |
+
"loss": 0.2663,
|
4296 |
+
"step": 609
|
4297 |
+
},
|
4298 |
+
{
|
4299 |
+
"epoch": 0.6367432150313153,
|
4300 |
+
"grad_norm": 0.5909189581871033,
|
4301 |
+
"learning_rate": 5.9447467918700614e-05,
|
4302 |
+
"loss": 0.2023,
|
4303 |
+
"step": 610
|
4304 |
+
},
|
4305 |
+
{
|
4306 |
+
"epoch": 0.6377870563674322,
|
4307 |
+
"grad_norm": 0.6986932158470154,
|
4308 |
+
"learning_rate": 5.9144771423416826e-05,
|
4309 |
+
"loss": 0.2354,
|
4310 |
+
"step": 611
|
4311 |
+
},
|
4312 |
+
{
|
4313 |
+
"epoch": 0.6388308977035491,
|
4314 |
+
"grad_norm": 0.5562401413917542,
|
4315 |
+
"learning_rate": 5.8842523601528e-05,
|
4316 |
+
"loss": 0.1928,
|
4317 |
+
"step": 612
|
4318 |
+
},
|
4319 |
+
{
|
4320 |
+
"epoch": 0.639874739039666,
|
4321 |
+
"grad_norm": 0.5309166312217712,
|
4322 |
+
"learning_rate": 5.854072777232914e-05,
|
4323 |
+
"loss": 0.1611,
|
4324 |
+
"step": 613
|
4325 |
+
},
|
4326 |
+
{
|
4327 |
+
"epoch": 0.6409185803757829,
|
4328 |
+
"grad_norm": 0.4029114842414856,
|
4329 |
+
"learning_rate": 5.823938725015148e-05,
|
4330 |
+
"loss": 0.14,
|
4331 |
+
"step": 614
|
4332 |
+
},
|
4333 |
+
{
|
4334 |
+
"epoch": 0.6419624217118998,
|
4335 |
+
"grad_norm": 0.4246949851512909,
|
4336 |
+
"learning_rate": 5.793850534432599e-05,
|
4337 |
+
"loss": 0.1194,
|
4338 |
+
"step": 615
|
4339 |
+
},
|
4340 |
+
{
|
4341 |
+
"epoch": 0.6430062630480167,
|
4342 |
+
"grad_norm": 0.44221794605255127,
|
4343 |
+
"learning_rate": 5.763808535914723e-05,
|
4344 |
+
"loss": 0.121,
|
4345 |
+
"step": 616
|
4346 |
+
},
|
4347 |
+
{
|
4348 |
+
"epoch": 0.6440501043841336,
|
4349 |
+
"grad_norm": 0.40379002690315247,
|
4350 |
+
"learning_rate": 5.7338130593836755e-05,
|
4351 |
+
"loss": 0.1215,
|
4352 |
+
"step": 617
|
4353 |
+
},
|
4354 |
+
{
|
4355 |
+
"epoch": 0.6450939457202505,
|
4356 |
+
"grad_norm": 0.6201443076133728,
|
4357 |
+
"learning_rate": 5.7038644342507205e-05,
|
4358 |
+
"loss": 0.1686,
|
4359 |
+
"step": 618
|
4360 |
+
},
|
4361 |
+
{
|
4362 |
+
"epoch": 0.6461377870563675,
|
4363 |
+
"grad_norm": 0.7134044766426086,
|
4364 |
+
"learning_rate": 5.673962989412599e-05,
|
4365 |
+
"loss": 0.3048,
|
4366 |
+
"step": 619
|
4367 |
+
},
|
4368 |
+
{
|
4369 |
+
"epoch": 0.6471816283924844,
|
4370 |
+
"grad_norm": 0.5085525512695312,
|
4371 |
+
"learning_rate": 5.644109053247901e-05,
|
4372 |
+
"loss": 0.2137,
|
4373 |
+
"step": 620
|
4374 |
+
},
|
4375 |
+
{
|
4376 |
+
"epoch": 0.6482254697286013,
|
4377 |
+
"grad_norm": 0.5813112854957581,
|
4378 |
+
"learning_rate": 5.614302953613489e-05,
|
4379 |
+
"loss": 0.2164,
|
4380 |
+
"step": 621
|
4381 |
+
},
|
4382 |
+
{
|
4383 |
+
"epoch": 0.6492693110647182,
|
4384 |
+
"grad_norm": 0.5314549803733826,
|
4385 |
+
"learning_rate": 5.584545017840885e-05,
|
4386 |
+
"loss": 0.1781,
|
4387 |
+
"step": 622
|
4388 |
+
},
|
4389 |
+
{
|
4390 |
+
"epoch": 0.6503131524008351,
|
4391 |
+
"grad_norm": 0.4466283619403839,
|
4392 |
+
"learning_rate": 5.5548355727326574e-05,
|
4393 |
+
"loss": 0.1564,
|
4394 |
+
"step": 623
|
4395 |
+
},
|
4396 |
+
{
|
4397 |
+
"epoch": 0.651356993736952,
|
4398 |
+
"grad_norm": 0.7003150582313538,
|
4399 |
+
"learning_rate": 5.525174944558866e-05,
|
4400 |
+
"loss": 0.3604,
|
4401 |
+
"step": 624
|
4402 |
+
},
|
4403 |
+
{
|
4404 |
+
"epoch": 0.6524008350730689,
|
4405 |
+
"grad_norm": 0.7904312610626221,
|
4406 |
+
"learning_rate": 5.4955634590534545e-05,
|
4407 |
+
"loss": 0.3483,
|
4408 |
+
"step": 625
|
4409 |
+
},
|
4410 |
+
{
|
4411 |
+
"epoch": 0.6534446764091858,
|
4412 |
+
"grad_norm": 0.7673099637031555,
|
4413 |
+
"learning_rate": 5.466001441410682e-05,
|
4414 |
+
"loss": 0.3912,
|
4415 |
+
"step": 626
|
4416 |
+
},
|
4417 |
+
{
|
4418 |
+
"epoch": 0.6544885177453027,
|
4419 |
+
"grad_norm": 0.823867678642273,
|
4420 |
+
"learning_rate": 5.4364892162815436e-05,
|
4421 |
+
"loss": 0.3618,
|
4422 |
+
"step": 627
|
4423 |
+
},
|
4424 |
+
{
|
4425 |
+
"epoch": 0.6555323590814196,
|
4426 |
+
"grad_norm": 0.6855948567390442,
|
4427 |
+
"learning_rate": 5.407027107770219e-05,
|
4428 |
+
"loss": 0.2816,
|
4429 |
+
"step": 628
|
4430 |
+
},
|
4431 |
+
{
|
4432 |
+
"epoch": 0.6565762004175365,
|
4433 |
+
"grad_norm": 0.7723731994628906,
|
4434 |
+
"learning_rate": 5.377615439430508e-05,
|
4435 |
+
"loss": 0.3292,
|
4436 |
+
"step": 629
|
4437 |
+
},
|
4438 |
+
{
|
4439 |
+
"epoch": 0.6576200417536534,
|
4440 |
+
"grad_norm": 0.7081869840621948,
|
4441 |
+
"learning_rate": 5.348254534262262e-05,
|
4442 |
+
"loss": 0.3232,
|
4443 |
+
"step": 630
|
4444 |
+
},
|
4445 |
+
{
|
4446 |
+
"epoch": 0.6586638830897703,
|
4447 |
+
"grad_norm": 0.7101826071739197,
|
4448 |
+
"learning_rate": 5.318944714707861e-05,
|
4449 |
+
"loss": 0.3557,
|
4450 |
+
"step": 631
|
4451 |
+
},
|
4452 |
+
{
|
4453 |
+
"epoch": 0.6597077244258872,
|
4454 |
+
"grad_norm": 0.7043560147285461,
|
4455 |
+
"learning_rate": 5.289686302648661e-05,
|
4456 |
+
"loss": 0.3251,
|
4457 |
+
"step": 632
|
4458 |
+
},
|
4459 |
+
{
|
4460 |
+
"epoch": 0.6607515657620042,
|
4461 |
+
"grad_norm": 0.8865169286727905,
|
4462 |
+
"learning_rate": 5.2604796194014507e-05,
|
4463 |
+
"loss": 0.3514,
|
4464 |
+
"step": 633
|
4465 |
+
},
|
4466 |
+
{
|
4467 |
+
"epoch": 0.6617954070981211,
|
4468 |
+
"grad_norm": 0.8106626868247986,
|
4469 |
+
"learning_rate": 5.2313249857149414e-05,
|
4470 |
+
"loss": 0.3226,
|
4471 |
+
"step": 634
|
4472 |
+
},
|
4473 |
+
{
|
4474 |
+
"epoch": 0.662839248434238,
|
4475 |
+
"grad_norm": 0.7511535286903381,
|
4476 |
+
"learning_rate": 5.202222721766226e-05,
|
4477 |
+
"loss": 0.3186,
|
4478 |
+
"step": 635
|
4479 |
+
},
|
4480 |
+
{
|
4481 |
+
"epoch": 0.6638830897703549,
|
4482 |
+
"grad_norm": 0.7110910415649414,
|
4483 |
+
"learning_rate": 5.1731731471572755e-05,
|
4484 |
+
"loss": 0.29,
|
4485 |
+
"step": 636
|
4486 |
+
},
|
4487 |
+
{
|
4488 |
+
"epoch": 0.6649269311064718,
|
4489 |
+
"grad_norm": 0.7598642110824585,
|
4490 |
+
"learning_rate": 5.144176580911431e-05,
|
4491 |
+
"loss": 0.2552,
|
4492 |
+
"step": 637
|
4493 |
+
},
|
4494 |
+
{
|
4495 |
+
"epoch": 0.6659707724425887,
|
4496 |
+
"grad_norm": 0.5399108529090881,
|
4497 |
+
"learning_rate": 5.115233341469877e-05,
|
4498 |
+
"loss": 0.2105,
|
4499 |
+
"step": 638
|
4500 |
+
},
|
4501 |
+
{
|
4502 |
+
"epoch": 0.6670146137787056,
|
4503 |
+
"grad_norm": 0.6335327625274658,
|
4504 |
+
"learning_rate": 5.0863437466881836e-05,
|
4505 |
+
"loss": 0.2272,
|
4506 |
+
"step": 639
|
4507 |
+
},
|
4508 |
+
{
|
4509 |
+
"epoch": 0.6680584551148225,
|
4510 |
+
"grad_norm": 0.6470877528190613,
|
4511 |
+
"learning_rate": 5.0575081138327715e-05,
|
4512 |
+
"loss": 0.2329,
|
4513 |
+
"step": 640
|
4514 |
+
},
|
4515 |
+
{
|
4516 |
+
"epoch": 0.6691022964509394,
|
4517 |
+
"grad_norm": 0.5501940250396729,
|
4518 |
+
"learning_rate": 5.028726759577467e-05,
|
4519 |
+
"loss": 0.2057,
|
4520 |
+
"step": 641
|
4521 |
+
},
|
4522 |
+
{
|
4523 |
+
"epoch": 0.6701461377870563,
|
4524 |
+
"grad_norm": 0.48475509881973267,
|
4525 |
+
"learning_rate": 5.000000000000002e-05,
|
4526 |
+
"loss": 0.1643,
|
4527 |
+
"step": 642
|
4528 |
+
},
|
4529 |
+
{
|
4530 |
+
"epoch": 0.6711899791231732,
|
4531 |
+
"grad_norm": 0.8148300051689148,
|
4532 |
+
"learning_rate": 4.97132815057854e-05,
|
4533 |
+
"loss": 0.2454,
|
4534 |
+
"step": 643
|
4535 |
+
},
|
4536 |
+
{
|
4537 |
+
"epoch": 0.6722338204592901,
|
4538 |
+
"grad_norm": 0.47534969449043274,
|
4539 |
+
"learning_rate": 4.942711526188229e-05,
|
4540 |
+
"loss": 0.1713,
|
4541 |
+
"step": 644
|
4542 |
+
},
|
4543 |
+
{
|
4544 |
+
"epoch": 0.673277661795407,
|
4545 |
+
"grad_norm": 0.5151733160018921,
|
4546 |
+
"learning_rate": 4.914150441097736e-05,
|
4547 |
+
"loss": 0.1701,
|
4548 |
+
"step": 645
|
4549 |
+
},
|
4550 |
+
{
|
4551 |
+
"epoch": 0.6743215031315241,
|
4552 |
+
"grad_norm": 0.5037069916725159,
|
4553 |
+
"learning_rate": 4.885645208965779e-05,
|
4554 |
+
"loss": 0.1814,
|
4555 |
+
"step": 646
|
4556 |
+
},
|
4557 |
+
{
|
4558 |
+
"epoch": 0.675365344467641,
|
4559 |
+
"grad_norm": 0.4882695972919464,
|
4560 |
+
"learning_rate": 4.857196142837716e-05,
|
4561 |
+
"loss": 0.1685,
|
4562 |
+
"step": 647
|
4563 |
+
},
|
4564 |
+
{
|
4565 |
+
"epoch": 0.6764091858037579,
|
4566 |
+
"grad_norm": 0.614020586013794,
|
4567 |
+
"learning_rate": 4.8288035551420697e-05,
|
4568 |
+
"loss": 0.1878,
|
4569 |
+
"step": 648
|
4570 |
+
},
|
4571 |
+
{
|
4572 |
+
"epoch": 0.6774530271398748,
|
4573 |
+
"grad_norm": 0.44983476400375366,
|
4574 |
+
"learning_rate": 4.80046775768713e-05,
|
4575 |
+
"loss": 0.1455,
|
4576 |
+
"step": 649
|
4577 |
+
},
|
4578 |
+
{
|
4579 |
+
"epoch": 0.6784968684759917,
|
4580 |
+
"grad_norm": 0.7257928252220154,
|
4581 |
+
"learning_rate": 4.7721890616575103e-05,
|
4582 |
+
"loss": 0.1417,
|
4583 |
+
"step": 650
|
4584 |
+
},
|
4585 |
+
{
|
4586 |
+
"epoch": 0.6795407098121086,
|
4587 |
+
"grad_norm": 0.5455628633499146,
|
4588 |
+
"learning_rate": 4.743967777610727e-05,
|
4589 |
+
"loss": 0.298,
|
4590 |
+
"step": 651
|
4591 |
+
},
|
4592 |
+
{
|
4593 |
+
"epoch": 0.6805845511482255,
|
4594 |
+
"grad_norm": 0.6038815379142761,
|
4595 |
+
"learning_rate": 4.715804215473809e-05,
|
4596 |
+
"loss": 0.322,
|
4597 |
+
"step": 652
|
4598 |
+
},
|
4599 |
+
{
|
4600 |
+
"epoch": 0.6816283924843424,
|
4601 |
+
"grad_norm": 0.6264936327934265,
|
4602 |
+
"learning_rate": 4.687698684539866e-05,
|
4603 |
+
"loss": 0.3185,
|
4604 |
+
"step": 653
|
4605 |
+
},
|
4606 |
+
{
|
4607 |
+
"epoch": 0.6826722338204593,
|
4608 |
+
"grad_norm": 0.6083415150642395,
|
4609 |
+
"learning_rate": 4.659651493464721e-05,
|
4610 |
+
"loss": 0.2614,
|
4611 |
+
"step": 654
|
4612 |
+
},
|
4613 |
+
{
|
4614 |
+
"epoch": 0.6837160751565762,
|
4615 |
+
"grad_norm": 0.616062343120575,
|
4616 |
+
"learning_rate": 4.6316629502635025e-05,
|
4617 |
+
"loss": 0.2663,
|
4618 |
+
"step": 655
|
4619 |
+
},
|
4620 |
+
{
|
4621 |
+
"epoch": 0.6847599164926931,
|
4622 |
+
"grad_norm": 0.6474526524543762,
|
4623 |
+
"learning_rate": 4.603733362307261e-05,
|
4624 |
+
"loss": 0.3216,
|
4625 |
+
"step": 656
|
4626 |
+
},
|
4627 |
+
{
|
4628 |
+
"epoch": 0.68580375782881,
|
4629 |
+
"grad_norm": 0.6374465227127075,
|
4630 |
+
"learning_rate": 4.575863036319604e-05,
|
4631 |
+
"loss": 0.2995,
|
4632 |
+
"step": 657
|
4633 |
+
},
|
4634 |
+
{
|
4635 |
+
"epoch": 0.6868475991649269,
|
4636 |
+
"grad_norm": 0.627227246761322,
|
4637 |
+
"learning_rate": 4.548052278373327e-05,
|
4638 |
+
"loss": 0.2399,
|
4639 |
+
"step": 658
|
4640 |
+
},
|
4641 |
+
{
|
4642 |
+
"epoch": 0.6878914405010439,
|
4643 |
+
"grad_norm": 0.6535525321960449,
|
4644 |
+
"learning_rate": 4.520301393887032e-05,
|
4645 |
+
"loss": 0.2309,
|
4646 |
+
"step": 659
|
4647 |
+
},
|
4648 |
+
{
|
4649 |
+
"epoch": 0.6889352818371608,
|
4650 |
+
"grad_norm": 0.5703950524330139,
|
4651 |
+
"learning_rate": 4.492610687621804e-05,
|
4652 |
+
"loss": 0.2128,
|
4653 |
+
"step": 660
|
4654 |
+
},
|
4655 |
+
{
|
4656 |
+
"epoch": 0.6899791231732777,
|
4657 |
+
"grad_norm": 0.5039522647857666,
|
4658 |
+
"learning_rate": 4.4649804636778456e-05,
|
4659 |
+
"loss": 0.1812,
|
4660 |
+
"step": 661
|
4661 |
+
},
|
4662 |
+
{
|
4663 |
+
"epoch": 0.6910229645093946,
|
4664 |
+
"grad_norm": 0.6118776798248291,
|
4665 |
+
"learning_rate": 4.4374110254911306e-05,
|
4666 |
+
"loss": 0.225,
|
4667 |
+
"step": 662
|
4668 |
+
},
|
4669 |
+
{
|
4670 |
+
"epoch": 0.6920668058455115,
|
4671 |
+
"grad_norm": 0.4194796085357666,
|
4672 |
+
"learning_rate": 4.4099026758300944e-05,
|
4673 |
+
"loss": 0.1415,
|
4674 |
+
"step": 663
|
4675 |
+
},
|
4676 |
+
{
|
4677 |
+
"epoch": 0.6931106471816284,
|
4678 |
+
"grad_norm": 0.5777245163917542,
|
4679 |
+
"learning_rate": 4.382455716792291e-05,
|
4680 |
+
"loss": 0.2032,
|
4681 |
+
"step": 664
|
4682 |
+
},
|
4683 |
+
{
|
4684 |
+
"epoch": 0.6941544885177453,
|
4685 |
+
"grad_norm": 0.41527435183525085,
|
4686 |
+
"learning_rate": 4.355070449801083e-05,
|
4687 |
+
"loss": 0.1205,
|
4688 |
+
"step": 665
|
4689 |
+
},
|
4690 |
+
{
|
4691 |
+
"epoch": 0.6951983298538622,
|
4692 |
+
"grad_norm": 0.502178430557251,
|
4693 |
+
"learning_rate": 4.32774717560232e-05,
|
4694 |
+
"loss": 0.1514,
|
4695 |
+
"step": 666
|
4696 |
+
},
|
4697 |
+
{
|
4698 |
+
"epoch": 0.6962421711899791,
|
4699 |
+
"grad_norm": 0.5909983515739441,
|
4700 |
+
"learning_rate": 4.300486194261057e-05,
|
4701 |
+
"loss": 0.176,
|
4702 |
+
"step": 667
|
4703 |
+
},
|
4704 |
+
{
|
4705 |
+
"epoch": 0.697286012526096,
|
4706 |
+
"grad_norm": 0.48378539085388184,
|
4707 |
+
"learning_rate": 4.273287805158245e-05,
|
4708 |
+
"loss": 0.1627,
|
4709 |
+
"step": 668
|
4710 |
+
},
|
4711 |
+
{
|
4712 |
+
"epoch": 0.6983298538622129,
|
4713 |
+
"grad_norm": 0.6273384094238281,
|
4714 |
+
"learning_rate": 4.2461523069874346e-05,
|
4715 |
+
"loss": 0.2523,
|
4716 |
+
"step": 669
|
4717 |
+
},
|
4718 |
+
{
|
4719 |
+
"epoch": 0.6993736951983298,
|
4720 |
+
"grad_norm": 0.5280055403709412,
|
4721 |
+
"learning_rate": 4.219079997751515e-05,
|
4722 |
+
"loss": 0.2035,
|
4723 |
+
"step": 670
|
4724 |
+
},
|
4725 |
+
{
|
4726 |
+
"epoch": 0.7004175365344467,
|
4727 |
+
"grad_norm": 0.539364755153656,
|
4728 |
+
"learning_rate": 4.192071174759435e-05,
|
4729 |
+
"loss": 0.1896,
|
4730 |
+
"step": 671
|
4731 |
+
},
|
4732 |
+
{
|
4733 |
+
"epoch": 0.7014613778705637,
|
4734 |
+
"grad_norm": 0.5403777956962585,
|
4735 |
+
"learning_rate": 4.165126134622926e-05,
|
4736 |
+
"loss": 0.1624,
|
4737 |
+
"step": 672
|
4738 |
+
},
|
4739 |
+
{
|
4740 |
+
"epoch": 0.7025052192066806,
|
4741 |
+
"grad_norm": 0.448177695274353,
|
4742 |
+
"learning_rate": 4.1382451732532665e-05,
|
4743 |
+
"loss": 0.1308,
|
4744 |
+
"step": 673
|
4745 |
+
},
|
4746 |
+
{
|
4747 |
+
"epoch": 0.7035490605427975,
|
4748 |
+
"grad_norm": 0.6774344444274902,
|
4749 |
+
"learning_rate": 4.1114285858580045e-05,
|
4750 |
+
"loss": 0.3161,
|
4751 |
+
"step": 674
|
4752 |
+
},
|
4753 |
+
{
|
4754 |
+
"epoch": 0.7045929018789144,
|
4755 |
+
"grad_norm": 0.7080472707748413,
|
4756 |
+
"learning_rate": 4.0846766669377446e-05,
|
4757 |
+
"loss": 0.3357,
|
4758 |
+
"step": 675
|
4759 |
+
},
|
4760 |
+
{
|
4761 |
+
"epoch": 0.7056367432150313,
|
4762 |
+
"grad_norm": 0.6589325666427612,
|
4763 |
+
"learning_rate": 4.0579897102828966e-05,
|
4764 |
+
"loss": 0.2815,
|
4765 |
+
"step": 676
|
4766 |
+
},
|
4767 |
+
{
|
4768 |
+
"epoch": 0.7066805845511482,
|
4769 |
+
"grad_norm": 0.7149707078933716,
|
4770 |
+
"learning_rate": 4.0313680089704454e-05,
|
4771 |
+
"loss": 0.3612,
|
4772 |
+
"step": 677
|
4773 |
+
},
|
4774 |
+
{
|
4775 |
+
"epoch": 0.7077244258872651,
|
4776 |
+
"grad_norm": 0.6252415776252747,
|
4777 |
+
"learning_rate": 4.004811855360748e-05,
|
4778 |
+
"loss": 0.2726,
|
4779 |
+
"step": 678
|
4780 |
+
},
|
4781 |
+
{
|
4782 |
+
"epoch": 0.708768267223382,
|
4783 |
+
"grad_norm": 0.7726844549179077,
|
4784 |
+
"learning_rate": 3.9783215410943174e-05,
|
4785 |
+
"loss": 0.3229,
|
4786 |
+
"step": 679
|
4787 |
+
},
|
4788 |
+
{
|
4789 |
+
"epoch": 0.7098121085594989,
|
4790 |
+
"grad_norm": 0.7369757890701294,
|
4791 |
+
"learning_rate": 3.951897357088602e-05,
|
4792 |
+
"loss": 0.3436,
|
4793 |
+
"step": 680
|
4794 |
+
},
|
4795 |
+
{
|
4796 |
+
"epoch": 0.7108559498956158,
|
4797 |
+
"grad_norm": 0.788517951965332,
|
4798 |
+
"learning_rate": 3.925539593534824e-05,
|
4799 |
+
"loss": 0.312,
|
4800 |
+
"step": 681
|
4801 |
+
},
|
4802 |
+
{
|
4803 |
+
"epoch": 0.7118997912317327,
|
4804 |
+
"grad_norm": 0.7983633875846863,
|
4805 |
+
"learning_rate": 3.899248539894757e-05,
|
4806 |
+
"loss": 0.361,
|
4807 |
+
"step": 682
|
4808 |
+
},
|
4809 |
+
{
|
4810 |
+
"epoch": 0.7129436325678496,
|
4811 |
+
"grad_norm": 0.7084015607833862,
|
4812 |
+
"learning_rate": 3.873024484897576e-05,
|
4813 |
+
"loss": 0.2836,
|
4814 |
+
"step": 683
|
4815 |
+
},
|
4816 |
+
{
|
4817 |
+
"epoch": 0.7139874739039666,
|
4818 |
+
"grad_norm": 0.7584156394004822,
|
4819 |
+
"learning_rate": 3.8468677165366754e-05,
|
4820 |
+
"loss": 0.2955,
|
4821 |
+
"step": 684
|
4822 |
+
},
|
4823 |
+
{
|
4824 |
+
"epoch": 0.7150313152400835,
|
4825 |
+
"grad_norm": 0.7465482354164124,
|
4826 |
+
"learning_rate": 3.820778522066494e-05,
|
4827 |
+
"loss": 0.2564,
|
4828 |
+
"step": 685
|
4829 |
+
},
|
4830 |
+
{
|
4831 |
+
"epoch": 0.7160751565762005,
|
4832 |
+
"grad_norm": 0.6189156770706177,
|
4833 |
+
"learning_rate": 3.794757187999386e-05,
|
4834 |
+
"loss": 0.221,
|
4835 |
+
"step": 686
|
4836 |
+
},
|
4837 |
+
{
|
4838 |
+
"epoch": 0.7171189979123174,
|
4839 |
+
"grad_norm": 0.6960480809211731,
|
4840 |
+
"learning_rate": 3.7688040001024475e-05,
|
4841 |
+
"loss": 0.2522,
|
4842 |
+
"step": 687
|
4843 |
+
},
|
4844 |
+
{
|
4845 |
+
"epoch": 0.7181628392484343,
|
4846 |
+
"grad_norm": 0.572296679019928,
|
4847 |
+
"learning_rate": 3.7429192433944014e-05,
|
4848 |
+
"loss": 0.1997,
|
4849 |
+
"step": 688
|
4850 |
+
},
|
4851 |
+
{
|
4852 |
+
"epoch": 0.7192066805845512,
|
4853 |
+
"grad_norm": 0.6478104591369629,
|
4854 |
+
"learning_rate": 3.717103202142457e-05,
|
4855 |
+
"loss": 0.2126,
|
4856 |
+
"step": 689
|
4857 |
+
},
|
4858 |
+
{
|
4859 |
+
"epoch": 0.7202505219206681,
|
4860 |
+
"grad_norm": 0.7389695644378662,
|
4861 |
+
"learning_rate": 3.691356159859177e-05,
|
4862 |
+
"loss": 0.2333,
|
4863 |
+
"step": 690
|
4864 |
+
},
|
4865 |
+
{
|
4866 |
+
"epoch": 0.721294363256785,
|
4867 |
+
"grad_norm": 0.5489468574523926,
|
4868 |
+
"learning_rate": 3.665678399299388e-05,
|
4869 |
+
"loss": 0.1701,
|
4870 |
+
"step": 691
|
4871 |
+
},
|
4872 |
+
{
|
4873 |
+
"epoch": 0.7223382045929019,
|
4874 |
+
"grad_norm": 0.7258986830711365,
|
4875 |
+
"learning_rate": 3.64007020245706e-05,
|
4876 |
+
"loss": 0.2457,
|
4877 |
+
"step": 692
|
4878 |
+
},
|
4879 |
+
{
|
4880 |
+
"epoch": 0.7233820459290188,
|
4881 |
+
"grad_norm": 0.6785321235656738,
|
4882 |
+
"learning_rate": 3.614531850562203e-05,
|
4883 |
+
"loss": 0.1936,
|
4884 |
+
"step": 693
|
4885 |
+
},
|
4886 |
+
{
|
4887 |
+
"epoch": 0.7244258872651357,
|
4888 |
+
"grad_norm": 0.612426221370697,
|
4889 |
+
"learning_rate": 3.589063624077802e-05,
|
4890 |
+
"loss": 0.2403,
|
4891 |
+
"step": 694
|
4892 |
+
},
|
4893 |
+
{
|
4894 |
+
"epoch": 0.7254697286012526,
|
4895 |
+
"grad_norm": 0.5947864651679993,
|
4896 |
+
"learning_rate": 3.563665802696707e-05,
|
4897 |
+
"loss": 0.1743,
|
4898 |
+
"step": 695
|
4899 |
+
},
|
4900 |
+
{
|
4901 |
+
"epoch": 0.7265135699373695,
|
4902 |
+
"grad_norm": 0.6579543352127075,
|
4903 |
+
"learning_rate": 3.538338665338589e-05,
|
4904 |
+
"loss": 0.1928,
|
4905 |
+
"step": 696
|
4906 |
+
},
|
4907 |
+
{
|
4908 |
+
"epoch": 0.7275574112734864,
|
4909 |
+
"grad_norm": 0.5460782647132874,
|
4910 |
+
"learning_rate": 3.513082490146864e-05,
|
4911 |
+
"loss": 0.1655,
|
4912 |
+
"step": 697
|
4913 |
+
},
|
4914 |
+
{
|
4915 |
+
"epoch": 0.7286012526096033,
|
4916 |
+
"grad_norm": 0.7640422582626343,
|
4917 |
+
"learning_rate": 3.487897554485628e-05,
|
4918 |
+
"loss": 0.1659,
|
4919 |
+
"step": 698
|
4920 |
+
},
|
4921 |
+
{
|
4922 |
+
"epoch": 0.7296450939457203,
|
4923 |
+
"grad_norm": 0.7361250519752502,
|
4924 |
+
"learning_rate": 3.462784134936636e-05,
|
4925 |
+
"loss": 0.1992,
|
4926 |
+
"step": 699
|
4927 |
+
},
|
4928 |
+
{
|
4929 |
+
"epoch": 0.7306889352818372,
|
4930 |
+
"grad_norm": 1.2653623819351196,
|
4931 |
+
"learning_rate": 3.4377425072962465e-05,
|
4932 |
+
"loss": 0.1184,
|
4933 |
+
"step": 700
|
4934 |
+
},
|
4935 |
+
{
|
4936 |
+
"epoch": 0.7317327766179541,
|
4937 |
+
"grad_norm": 0.6173591613769531,
|
4938 |
+
"learning_rate": 3.412772946572389e-05,
|
4939 |
+
"loss": 0.4121,
|
4940 |
+
"step": 701
|
4941 |
+
},
|
4942 |
+
{
|
4943 |
+
"epoch": 0.732776617954071,
|
4944 |
+
"grad_norm": 0.5525224208831787,
|
4945 |
+
"learning_rate": 3.387875726981563e-05,
|
4946 |
+
"loss": 0.2601,
|
4947 |
+
"step": 702
|
4948 |
+
},
|
4949 |
+
{
|
4950 |
+
"epoch": 0.7338204592901879,
|
4951 |
+
"grad_norm": 0.6985558867454529,
|
4952 |
+
"learning_rate": 3.363051121945809e-05,
|
4953 |
+
"loss": 0.3448,
|
4954 |
+
"step": 703
|
4955 |
+
},
|
4956 |
+
{
|
4957 |
+
"epoch": 0.7348643006263048,
|
4958 |
+
"grad_norm": 0.580680251121521,
|
4959 |
+
"learning_rate": 3.3382994040897196e-05,
|
4960 |
+
"loss": 0.2642,
|
4961 |
+
"step": 704
|
4962 |
+
},
|
4963 |
+
{
|
4964 |
+
"epoch": 0.7359081419624217,
|
4965 |
+
"grad_norm": 0.5876568555831909,
|
4966 |
+
"learning_rate": 3.3136208452374254e-05,
|
4967 |
+
"loss": 0.271,
|
4968 |
+
"step": 705
|
4969 |
+
},
|
4970 |
+
{
|
4971 |
+
"epoch": 0.7369519832985386,
|
4972 |
+
"grad_norm": 0.6181269884109497,
|
4973 |
+
"learning_rate": 3.289015716409631e-05,
|
4974 |
+
"loss": 0.2522,
|
4975 |
+
"step": 706
|
4976 |
+
},
|
4977 |
+
{
|
4978 |
+
"epoch": 0.7379958246346555,
|
4979 |
+
"grad_norm": 0.6312392354011536,
|
4980 |
+
"learning_rate": 3.264484287820634e-05,
|
4981 |
+
"loss": 0.2735,
|
4982 |
+
"step": 707
|
4983 |
+
},
|
4984 |
+
{
|
4985 |
+
"epoch": 0.7390396659707724,
|
4986 |
+
"grad_norm": 0.6163091063499451,
|
4987 |
+
"learning_rate": 3.2400268288753425e-05,
|
4988 |
+
"loss": 0.2415,
|
4989 |
+
"step": 708
|
4990 |
+
},
|
4991 |
+
{
|
4992 |
+
"epoch": 0.7400835073068893,
|
4993 |
+
"grad_norm": 0.6303150057792664,
|
4994 |
+
"learning_rate": 3.2156436081663356e-05,
|
4995 |
+
"loss": 0.2495,
|
4996 |
+
"step": 709
|
4997 |
+
},
|
4998 |
+
{
|
4999 |
+
"epoch": 0.7411273486430062,
|
5000 |
+
"grad_norm": 0.6544148325920105,
|
5001 |
+
"learning_rate": 3.191334893470907e-05,
|
5002 |
+
"loss": 0.2445,
|
5003 |
+
"step": 710
|
5004 |
+
},
|
5005 |
+
{
|
5006 |
+
"epoch": 0.7421711899791231,
|
5007 |
+
"grad_norm": 0.468227744102478,
|
5008 |
+
"learning_rate": 3.167100951748115e-05,
|
5009 |
+
"loss": 0.1481,
|
5010 |
+
"step": 711
|
5011 |
+
},
|
5012 |
+
{
|
5013 |
+
"epoch": 0.7432150313152401,
|
5014 |
+
"grad_norm": 0.5340932607650757,
|
5015 |
+
"learning_rate": 3.14294204913587e-05,
|
5016 |
+
"loss": 0.1431,
|
5017 |
+
"step": 712
|
5018 |
+
},
|
5019 |
+
{
|
5020 |
+
"epoch": 0.744258872651357,
|
5021 |
+
"grad_norm": 0.5851957201957703,
|
5022 |
+
"learning_rate": 3.1188584509479866e-05,
|
5023 |
+
"loss": 0.1737,
|
5024 |
+
"step": 713
|
5025 |
+
},
|
5026 |
+
{
|
5027 |
+
"epoch": 0.7453027139874739,
|
5028 |
+
"grad_norm": 0.5840248465538025,
|
5029 |
+
"learning_rate": 3.094850421671295e-05,
|
5030 |
+
"loss": 0.1752,
|
5031 |
+
"step": 714
|
5032 |
+
},
|
5033 |
+
{
|
5034 |
+
"epoch": 0.7463465553235908,
|
5035 |
+
"grad_norm": 0.5659369826316833,
|
5036 |
+
"learning_rate": 3.0709182249627255e-05,
|
5037 |
+
"loss": 0.1967,
|
5038 |
+
"step": 715
|
5039 |
+
},
|
5040 |
+
{
|
5041 |
+
"epoch": 0.7473903966597077,
|
5042 |
+
"grad_norm": 0.457015722990036,
|
5043 |
+
"learning_rate": 3.0470621236464036e-05,
|
5044 |
+
"loss": 0.1544,
|
5045 |
+
"step": 716
|
5046 |
+
},
|
5047 |
+
{
|
5048 |
+
"epoch": 0.7484342379958246,
|
5049 |
+
"grad_norm": 0.5182324647903442,
|
5050 |
+
"learning_rate": 3.023282379710779e-05,
|
5051 |
+
"loss": 0.1414,
|
5052 |
+
"step": 717
|
5053 |
+
},
|
5054 |
+
{
|
5055 |
+
"epoch": 0.7494780793319415,
|
5056 |
+
"grad_norm": 0.5334721207618713,
|
5057 |
+
"learning_rate": 2.9995792543057478e-05,
|
5058 |
+
"loss": 0.1615,
|
5059 |
+
"step": 718
|
5060 |
+
},
|
5061 |
+
{
|
5062 |
+
"epoch": 0.7505219206680585,
|
5063 |
+
"grad_norm": 0.6061464548110962,
|
5064 |
+
"learning_rate": 2.9759530077397636e-05,
|
5065 |
+
"loss": 0.2452,
|
5066 |
+
"step": 719
|
5067 |
+
},
|
5068 |
+
{
|
5069 |
+
"epoch": 0.7515657620041754,
|
5070 |
+
"grad_norm": 0.5774762630462646,
|
5071 |
+
"learning_rate": 2.9524038994770107e-05,
|
5072 |
+
"loss": 0.2234,
|
5073 |
+
"step": 720
|
5074 |
+
},
|
5075 |
+
{
|
5076 |
+
"epoch": 0.7515657620041754,
|
5077 |
+
"eval_loss": 0.21803000569343567,
|
5078 |
+
"eval_runtime": 81.2533,
|
5079 |
+
"eval_samples_per_second": 19.864,
|
5080 |
+
"eval_steps_per_second": 9.932,
|
5081 |
+
"step": 720
|
5082 |
}
|
5083 |
],
|
5084 |
"logging_steps": 1,
|
|
|
5098 |
"attributes": {}
|
5099 |
}
|
5100 |
},
|
5101 |
+
"total_flos": 4.025453542322995e+17,
|
5102 |
"train_batch_size": 2,
|
5103 |
"trial_name": null,
|
5104 |
"trial_params": null
|