Training in progress, step 931, 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 14293800
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3da7b0f8965c3847c1eb3d447c49d4a2fefd69cb36beae15480bc6a81c95e32f
|
| 3 |
size 14293800
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 7580068
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:924b8dc63c310339560add3c51f706b25470c2fbae8609a05e92f2a16677a20e
|
| 3 |
size 7580068
|
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 sha256:364dd5e98eeed3542841a03e83c164bd5c8688a6a3b54580f2a6113076669add
|
| 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 sha256:72aeb0c6de0d18ab3aae0e2eab94b9cced4c1ed177eda63ddb050f2b052ee0e2
|
| 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":
|
| 5 |
"eval_steps": 233,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -4932,6 +4932,1630 @@
|
|
| 4932 |
"eval_samples_per_second": 70.026,
|
| 4933 |
"eval_steps_per_second": 35.013,
|
| 4934 |
"step": 699
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 4935 |
}
|
| 4936 |
],
|
| 4937 |
"logging_steps": 1,
|
|
@@ -4946,12 +6570,12 @@
|
|
| 4946 |
"should_evaluate": false,
|
| 4947 |
"should_log": false,
|
| 4948 |
"should_save": true,
|
| 4949 |
-
"should_training_stop":
|
| 4950 |
},
|
| 4951 |
"attributes": {}
|
| 4952 |
}
|
| 4953 |
},
|
| 4954 |
-
"total_flos":
|
| 4955 |
"train_batch_size": 2,
|
| 4956 |
"trial_name": null,
|
| 4957 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 1.000806234883096,
|
| 5 |
"eval_steps": 233,
|
| 6 |
+
"global_step": 931,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 4932 |
"eval_samples_per_second": 70.026,
|
| 4933 |
"eval_steps_per_second": 35.013,
|
| 4934 |
"step": 699
|
| 4935 |
+
},
|
| 4936 |
+
{
|
| 4937 |
+
"epoch": 0.7524858908895459,
|
| 4938 |
+
"grad_norm": 13.654646873474121,
|
| 4939 |
+
"learning_rate": 2.9470452223481204e-05,
|
| 4940 |
+
"loss": 2.9831,
|
| 4941 |
+
"step": 700
|
| 4942 |
+
},
|
| 4943 |
+
{
|
| 4944 |
+
"epoch": 0.7535608707336737,
|
| 4945 |
+
"grad_norm": 9.951108932495117,
|
| 4946 |
+
"learning_rate": 2.922904788530293e-05,
|
| 4947 |
+
"loss": 2.2996,
|
| 4948 |
+
"step": 701
|
| 4949 |
+
},
|
| 4950 |
+
{
|
| 4951 |
+
"epoch": 0.7546358505778017,
|
| 4952 |
+
"grad_norm": 9.867563247680664,
|
| 4953 |
+
"learning_rate": 2.8988466993083097e-05,
|
| 4954 |
+
"loss": 2.1596,
|
| 4955 |
+
"step": 702
|
| 4956 |
+
},
|
| 4957 |
+
{
|
| 4958 |
+
"epoch": 0.7557108304219295,
|
| 4959 |
+
"grad_norm": 14.348727226257324,
|
| 4960 |
+
"learning_rate": 2.8748712346068464e-05,
|
| 4961 |
+
"loss": 2.8225,
|
| 4962 |
+
"step": 703
|
| 4963 |
+
},
|
| 4964 |
+
{
|
| 4965 |
+
"epoch": 0.7567858102660575,
|
| 4966 |
+
"grad_norm": 14.35478687286377,
|
| 4967 |
+
"learning_rate": 2.8509786733892264e-05,
|
| 4968 |
+
"loss": 2.9542,
|
| 4969 |
+
"step": 704
|
| 4970 |
+
},
|
| 4971 |
+
{
|
| 4972 |
+
"epoch": 0.7578607901101855,
|
| 4973 |
+
"grad_norm": 23.562644958496094,
|
| 4974 |
+
"learning_rate": 2.827169293654147e-05,
|
| 4975 |
+
"loss": 4.984,
|
| 4976 |
+
"step": 705
|
| 4977 |
+
},
|
| 4978 |
+
{
|
| 4979 |
+
"epoch": 0.7589357699543133,
|
| 4980 |
+
"grad_norm": 11.014055252075195,
|
| 4981 |
+
"learning_rate": 2.8034433724324715e-05,
|
| 4982 |
+
"loss": 2.2337,
|
| 4983 |
+
"step": 706
|
| 4984 |
+
},
|
| 4985 |
+
{
|
| 4986 |
+
"epoch": 0.7600107497984413,
|
| 4987 |
+
"grad_norm": 14.4437837600708,
|
| 4988 |
+
"learning_rate": 2.77980118578398e-05,
|
| 4989 |
+
"loss": 2.8005,
|
| 4990 |
+
"step": 707
|
| 4991 |
+
},
|
| 4992 |
+
{
|
| 4993 |
+
"epoch": 0.7610857296425692,
|
| 4994 |
+
"grad_norm": 14.065092086791992,
|
| 4995 |
+
"learning_rate": 2.7562430087941814e-05,
|
| 4996 |
+
"loss": 3.0488,
|
| 4997 |
+
"step": 708
|
| 4998 |
+
},
|
| 4999 |
+
{
|
| 5000 |
+
"epoch": 0.7621607094866971,
|
| 5001 |
+
"grad_norm": 18.38102912902832,
|
| 5002 |
+
"learning_rate": 2.7327691155710976e-05,
|
| 5003 |
+
"loss": 2.857,
|
| 5004 |
+
"step": 709
|
| 5005 |
+
},
|
| 5006 |
+
{
|
| 5007 |
+
"epoch": 0.7632356893308251,
|
| 5008 |
+
"grad_norm": 15.708161354064941,
|
| 5009 |
+
"learning_rate": 2.7093797792420728e-05,
|
| 5010 |
+
"loss": 3.3492,
|
| 5011 |
+
"step": 710
|
| 5012 |
+
},
|
| 5013 |
+
{
|
| 5014 |
+
"epoch": 0.7643106691749529,
|
| 5015 |
+
"grad_norm": 13.264737129211426,
|
| 5016 |
+
"learning_rate": 2.68607527195061e-05,
|
| 5017 |
+
"loss": 2.2699,
|
| 5018 |
+
"step": 711
|
| 5019 |
+
},
|
| 5020 |
+
{
|
| 5021 |
+
"epoch": 0.7653856490190809,
|
| 5022 |
+
"grad_norm": 22.40721893310547,
|
| 5023 |
+
"learning_rate": 2.6628558648531843e-05,
|
| 5024 |
+
"loss": 3.6175,
|
| 5025 |
+
"step": 712
|
| 5026 |
+
},
|
| 5027 |
+
{
|
| 5028 |
+
"epoch": 0.7664606288632088,
|
| 5029 |
+
"grad_norm": 16.45822525024414,
|
| 5030 |
+
"learning_rate": 2.639721828116112e-05,
|
| 5031 |
+
"loss": 1.7208,
|
| 5032 |
+
"step": 713
|
| 5033 |
+
},
|
| 5034 |
+
{
|
| 5035 |
+
"epoch": 0.7675356087073367,
|
| 5036 |
+
"grad_norm": 15.185565948486328,
|
| 5037 |
+
"learning_rate": 2.6166734309123787e-05,
|
| 5038 |
+
"loss": 2.439,
|
| 5039 |
+
"step": 714
|
| 5040 |
+
},
|
| 5041 |
+
{
|
| 5042 |
+
"epoch": 0.7686105885514647,
|
| 5043 |
+
"grad_norm": 13.136150360107422,
|
| 5044 |
+
"learning_rate": 2.5937109414185366e-05,
|
| 5045 |
+
"loss": 2.9833,
|
| 5046 |
+
"step": 715
|
| 5047 |
+
},
|
| 5048 |
+
{
|
| 5049 |
+
"epoch": 0.7696855683955925,
|
| 5050 |
+
"grad_norm": 13.081589698791504,
|
| 5051 |
+
"learning_rate": 2.5708346268115647e-05,
|
| 5052 |
+
"loss": 2.4198,
|
| 5053 |
+
"step": 716
|
| 5054 |
+
},
|
| 5055 |
+
{
|
| 5056 |
+
"epoch": 0.7707605482397205,
|
| 5057 |
+
"grad_norm": 10.561399459838867,
|
| 5058 |
+
"learning_rate": 2.5480447532657624e-05,
|
| 5059 |
+
"loss": 2.3794,
|
| 5060 |
+
"step": 717
|
| 5061 |
+
},
|
| 5062 |
+
{
|
| 5063 |
+
"epoch": 0.7718355280838485,
|
| 5064 |
+
"grad_norm": 15.137680053710938,
|
| 5065 |
+
"learning_rate": 2.525341585949662e-05,
|
| 5066 |
+
"loss": 2.6444,
|
| 5067 |
+
"step": 718
|
| 5068 |
+
},
|
| 5069 |
+
{
|
| 5070 |
+
"epoch": 0.7729105079279763,
|
| 5071 |
+
"grad_norm": 18.32636070251465,
|
| 5072 |
+
"learning_rate": 2.5027253890229285e-05,
|
| 5073 |
+
"loss": 2.978,
|
| 5074 |
+
"step": 719
|
| 5075 |
+
},
|
| 5076 |
+
{
|
| 5077 |
+
"epoch": 0.7739854877721043,
|
| 5078 |
+
"grad_norm": 14.093949317932129,
|
| 5079 |
+
"learning_rate": 2.4801964256333053e-05,
|
| 5080 |
+
"loss": 2.4359,
|
| 5081 |
+
"step": 720
|
| 5082 |
+
},
|
| 5083 |
+
{
|
| 5084 |
+
"epoch": 0.7750604676162322,
|
| 5085 |
+
"grad_norm": 15.987081527709961,
|
| 5086 |
+
"learning_rate": 2.457754957913532e-05,
|
| 5087 |
+
"loss": 3.6221,
|
| 5088 |
+
"step": 721
|
| 5089 |
+
},
|
| 5090 |
+
{
|
| 5091 |
+
"epoch": 0.7761354474603601,
|
| 5092 |
+
"grad_norm": 14.7132568359375,
|
| 5093 |
+
"learning_rate": 2.4354012469783094e-05,
|
| 5094 |
+
"loss": 2.7112,
|
| 5095 |
+
"step": 722
|
| 5096 |
+
},
|
| 5097 |
+
{
|
| 5098 |
+
"epoch": 0.7772104273044881,
|
| 5099 |
+
"grad_norm": 15.25096321105957,
|
| 5100 |
+
"learning_rate": 2.4131355529212573e-05,
|
| 5101 |
+
"loss": 2.8029,
|
| 5102 |
+
"step": 723
|
| 5103 |
+
},
|
| 5104 |
+
{
|
| 5105 |
+
"epoch": 0.7782854071486159,
|
| 5106 |
+
"grad_norm": 11.852306365966797,
|
| 5107 |
+
"learning_rate": 2.3909581348118805e-05,
|
| 5108 |
+
"loss": 2.6288,
|
| 5109 |
+
"step": 724
|
| 5110 |
+
},
|
| 5111 |
+
{
|
| 5112 |
+
"epoch": 0.7793603869927439,
|
| 5113 |
+
"grad_norm": 14.22082805633545,
|
| 5114 |
+
"learning_rate": 2.368869250692567e-05,
|
| 5115 |
+
"loss": 2.6691,
|
| 5116 |
+
"step": 725
|
| 5117 |
+
},
|
| 5118 |
+
{
|
| 5119 |
+
"epoch": 0.7804353668368718,
|
| 5120 |
+
"grad_norm": 14.895218849182129,
|
| 5121 |
+
"learning_rate": 2.346869157575574e-05,
|
| 5122 |
+
"loss": 2.223,
|
| 5123 |
+
"step": 726
|
| 5124 |
+
},
|
| 5125 |
+
{
|
| 5126 |
+
"epoch": 0.7815103466809997,
|
| 5127 |
+
"grad_norm": 11.507866859436035,
|
| 5128 |
+
"learning_rate": 2.324958111440051e-05,
|
| 5129 |
+
"loss": 2.4171,
|
| 5130 |
+
"step": 727
|
| 5131 |
+
},
|
| 5132 |
+
{
|
| 5133 |
+
"epoch": 0.7825853265251277,
|
| 5134 |
+
"grad_norm": 10.818052291870117,
|
| 5135 |
+
"learning_rate": 2.3031363672290406e-05,
|
| 5136 |
+
"loss": 2.4578,
|
| 5137 |
+
"step": 728
|
| 5138 |
+
},
|
| 5139 |
+
{
|
| 5140 |
+
"epoch": 0.7836603063692555,
|
| 5141 |
+
"grad_norm": 14.49646282196045,
|
| 5142 |
+
"learning_rate": 2.28140417884654e-05,
|
| 5143 |
+
"loss": 3.3386,
|
| 5144 |
+
"step": 729
|
| 5145 |
+
},
|
| 5146 |
+
{
|
| 5147 |
+
"epoch": 0.7847352862133835,
|
| 5148 |
+
"grad_norm": 16.65618896484375,
|
| 5149 |
+
"learning_rate": 2.2597617991545162e-05,
|
| 5150 |
+
"loss": 2.4579,
|
| 5151 |
+
"step": 730
|
| 5152 |
+
},
|
| 5153 |
+
{
|
| 5154 |
+
"epoch": 0.7858102660575115,
|
| 5155 |
+
"grad_norm": 21.370105743408203,
|
| 5156 |
+
"learning_rate": 2.2382094799699917e-05,
|
| 5157 |
+
"loss": 2.0147,
|
| 5158 |
+
"step": 731
|
| 5159 |
+
},
|
| 5160 |
+
{
|
| 5161 |
+
"epoch": 0.7868852459016393,
|
| 5162 |
+
"grad_norm": 18.23256492614746,
|
| 5163 |
+
"learning_rate": 2.2167474720620974e-05,
|
| 5164 |
+
"loss": 2.8713,
|
| 5165 |
+
"step": 732
|
| 5166 |
+
},
|
| 5167 |
+
{
|
| 5168 |
+
"epoch": 0.7879602257457673,
|
| 5169 |
+
"grad_norm": 13.314383506774902,
|
| 5170 |
+
"learning_rate": 2.1953760251491563e-05,
|
| 5171 |
+
"loss": 2.5137,
|
| 5172 |
+
"step": 733
|
| 5173 |
+
},
|
| 5174 |
+
{
|
| 5175 |
+
"epoch": 0.7890352055898951,
|
| 5176 |
+
"grad_norm": 13.909300804138184,
|
| 5177 |
+
"learning_rate": 2.174095387895786e-05,
|
| 5178 |
+
"loss": 2.7543,
|
| 5179 |
+
"step": 734
|
| 5180 |
+
},
|
| 5181 |
+
{
|
| 5182 |
+
"epoch": 0.7901101854340231,
|
| 5183 |
+
"grad_norm": 12.180556297302246,
|
| 5184 |
+
"learning_rate": 2.152905807909995e-05,
|
| 5185 |
+
"loss": 2.0877,
|
| 5186 |
+
"step": 735
|
| 5187 |
+
},
|
| 5188 |
+
{
|
| 5189 |
+
"epoch": 0.7911851652781511,
|
| 5190 |
+
"grad_norm": 22.62627410888672,
|
| 5191 |
+
"learning_rate": 2.131807531740315e-05,
|
| 5192 |
+
"loss": 4.0689,
|
| 5193 |
+
"step": 736
|
| 5194 |
+
},
|
| 5195 |
+
{
|
| 5196 |
+
"epoch": 0.7922601451222789,
|
| 5197 |
+
"grad_norm": 15.548689842224121,
|
| 5198 |
+
"learning_rate": 2.1108008048729145e-05,
|
| 5199 |
+
"loss": 3.4416,
|
| 5200 |
+
"step": 737
|
| 5201 |
+
},
|
| 5202 |
+
{
|
| 5203 |
+
"epoch": 0.7933351249664069,
|
| 5204 |
+
"grad_norm": 17.072410583496094,
|
| 5205 |
+
"learning_rate": 2.0898858717287594e-05,
|
| 5206 |
+
"loss": 4.3289,
|
| 5207 |
+
"step": 738
|
| 5208 |
+
},
|
| 5209 |
+
{
|
| 5210 |
+
"epoch": 0.7944101048105348,
|
| 5211 |
+
"grad_norm": 12.466560363769531,
|
| 5212 |
+
"learning_rate": 2.0690629756607648e-05,
|
| 5213 |
+
"loss": 2.6264,
|
| 5214 |
+
"step": 739
|
| 5215 |
+
},
|
| 5216 |
+
{
|
| 5217 |
+
"epoch": 0.7954850846546627,
|
| 5218 |
+
"grad_norm": 10.822991371154785,
|
| 5219 |
+
"learning_rate": 2.0483323589509483e-05,
|
| 5220 |
+
"loss": 1.9285,
|
| 5221 |
+
"step": 740
|
| 5222 |
+
},
|
| 5223 |
+
{
|
| 5224 |
+
"epoch": 0.7965600644987907,
|
| 5225 |
+
"grad_norm": 16.706153869628906,
|
| 5226 |
+
"learning_rate": 2.0276942628076378e-05,
|
| 5227 |
+
"loss": 3.2333,
|
| 5228 |
+
"step": 741
|
| 5229 |
+
},
|
| 5230 |
+
{
|
| 5231 |
+
"epoch": 0.7976350443429185,
|
| 5232 |
+
"grad_norm": 13.810402870178223,
|
| 5233 |
+
"learning_rate": 2.0071489273626376e-05,
|
| 5234 |
+
"loss": 2.5904,
|
| 5235 |
+
"step": 742
|
| 5236 |
+
},
|
| 5237 |
+
{
|
| 5238 |
+
"epoch": 0.7987100241870465,
|
| 5239 |
+
"grad_norm": 15.366273880004883,
|
| 5240 |
+
"learning_rate": 1.9866965916684587e-05,
|
| 5241 |
+
"loss": 2.8742,
|
| 5242 |
+
"step": 743
|
| 5243 |
+
},
|
| 5244 |
+
{
|
| 5245 |
+
"epoch": 0.7997850040311744,
|
| 5246 |
+
"grad_norm": 11.5558500289917,
|
| 5247 |
+
"learning_rate": 1.966337493695516e-05,
|
| 5248 |
+
"loss": 2.3341,
|
| 5249 |
+
"step": 744
|
| 5250 |
+
},
|
| 5251 |
+
{
|
| 5252 |
+
"epoch": 0.8008599838753023,
|
| 5253 |
+
"grad_norm": 17.35418701171875,
|
| 5254 |
+
"learning_rate": 1.9460718703293768e-05,
|
| 5255 |
+
"loss": 2.6252,
|
| 5256 |
+
"step": 745
|
| 5257 |
+
},
|
| 5258 |
+
{
|
| 5259 |
+
"epoch": 0.8019349637194303,
|
| 5260 |
+
"grad_norm": 11.60128402709961,
|
| 5261 |
+
"learning_rate": 1.925899957367996e-05,
|
| 5262 |
+
"loss": 2.109,
|
| 5263 |
+
"step": 746
|
| 5264 |
+
},
|
| 5265 |
+
{
|
| 5266 |
+
"epoch": 0.8030099435635581,
|
| 5267 |
+
"grad_norm": 12.664255142211914,
|
| 5268 |
+
"learning_rate": 1.9058219895189666e-05,
|
| 5269 |
+
"loss": 2.339,
|
| 5270 |
+
"step": 747
|
| 5271 |
+
},
|
| 5272 |
+
{
|
| 5273 |
+
"epoch": 0.8040849234076861,
|
| 5274 |
+
"grad_norm": 16.977569580078125,
|
| 5275 |
+
"learning_rate": 1.8858382003968078e-05,
|
| 5276 |
+
"loss": 2.3301,
|
| 5277 |
+
"step": 748
|
| 5278 |
+
},
|
| 5279 |
+
{
|
| 5280 |
+
"epoch": 0.8051599032518141,
|
| 5281 |
+
"grad_norm": 12.833663940429688,
|
| 5282 |
+
"learning_rate": 1.8659488225202226e-05,
|
| 5283 |
+
"loss": 2.461,
|
| 5284 |
+
"step": 749
|
| 5285 |
+
},
|
| 5286 |
+
{
|
| 5287 |
+
"epoch": 0.8062348830959419,
|
| 5288 |
+
"grad_norm": 14.8052339553833,
|
| 5289 |
+
"learning_rate": 1.846154087309414e-05,
|
| 5290 |
+
"loss": 2.9448,
|
| 5291 |
+
"step": 750
|
| 5292 |
+
},
|
| 5293 |
+
{
|
| 5294 |
+
"epoch": 0.8073098629400699,
|
| 5295 |
+
"grad_norm": 16.044788360595703,
|
| 5296 |
+
"learning_rate": 1.826454225083375e-05,
|
| 5297 |
+
"loss": 2.9095,
|
| 5298 |
+
"step": 751
|
| 5299 |
+
},
|
| 5300 |
+
{
|
| 5301 |
+
"epoch": 0.8083848427841978,
|
| 5302 |
+
"grad_norm": 24.449602127075195,
|
| 5303 |
+
"learning_rate": 1.8068494650572243e-05,
|
| 5304 |
+
"loss": 3.3211,
|
| 5305 |
+
"step": 752
|
| 5306 |
+
},
|
| 5307 |
+
{
|
| 5308 |
+
"epoch": 0.8094598226283257,
|
| 5309 |
+
"grad_norm": 18.429533004760742,
|
| 5310 |
+
"learning_rate": 1.787340035339524e-05,
|
| 5311 |
+
"loss": 3.7572,
|
| 5312 |
+
"step": 753
|
| 5313 |
+
},
|
| 5314 |
+
{
|
| 5315 |
+
"epoch": 0.8105348024724537,
|
| 5316 |
+
"grad_norm": 18.038867950439453,
|
| 5317 |
+
"learning_rate": 1.7679261629296408e-05,
|
| 5318 |
+
"loss": 3.5666,
|
| 5319 |
+
"step": 754
|
| 5320 |
+
},
|
| 5321 |
+
{
|
| 5322 |
+
"epoch": 0.8116097823165815,
|
| 5323 |
+
"grad_norm": 17.882884979248047,
|
| 5324 |
+
"learning_rate": 1.7486080737150945e-05,
|
| 5325 |
+
"loss": 3.4553,
|
| 5326 |
+
"step": 755
|
| 5327 |
+
},
|
| 5328 |
+
{
|
| 5329 |
+
"epoch": 0.8126847621607095,
|
| 5330 |
+
"grad_norm": 18.42115592956543,
|
| 5331 |
+
"learning_rate": 1.7293859924689258e-05,
|
| 5332 |
+
"loss": 2.3743,
|
| 5333 |
+
"step": 756
|
| 5334 |
+
},
|
| 5335 |
+
{
|
| 5336 |
+
"epoch": 0.8137597420048374,
|
| 5337 |
+
"grad_norm": 10.389144897460938,
|
| 5338 |
+
"learning_rate": 1.7102601428470987e-05,
|
| 5339 |
+
"loss": 1.7921,
|
| 5340 |
+
"step": 757
|
| 5341 |
+
},
|
| 5342 |
+
{
|
| 5343 |
+
"epoch": 0.8148347218489653,
|
| 5344 |
+
"grad_norm": 19.917224884033203,
|
| 5345 |
+
"learning_rate": 1.691230747385878e-05,
|
| 5346 |
+
"loss": 4.1469,
|
| 5347 |
+
"step": 758
|
| 5348 |
+
},
|
| 5349 |
+
{
|
| 5350 |
+
"epoch": 0.8159097016930933,
|
| 5351 |
+
"grad_norm": 22.9906005859375,
|
| 5352 |
+
"learning_rate": 1.672298027499254e-05,
|
| 5353 |
+
"loss": 3.5433,
|
| 5354 |
+
"step": 759
|
| 5355 |
+
},
|
| 5356 |
+
{
|
| 5357 |
+
"epoch": 0.8169846815372211,
|
| 5358 |
+
"grad_norm": 9.970109939575195,
|
| 5359 |
+
"learning_rate": 1.653462203476356e-05,
|
| 5360 |
+
"loss": 1.9898,
|
| 5361 |
+
"step": 760
|
| 5362 |
+
},
|
| 5363 |
+
{
|
| 5364 |
+
"epoch": 0.8180596613813491,
|
| 5365 |
+
"grad_norm": 15.049894332885742,
|
| 5366 |
+
"learning_rate": 1.6347234944789014e-05,
|
| 5367 |
+
"loss": 3.014,
|
| 5368 |
+
"step": 761
|
| 5369 |
+
},
|
| 5370 |
+
{
|
| 5371 |
+
"epoch": 0.819134641225477,
|
| 5372 |
+
"grad_norm": 18.622976303100586,
|
| 5373 |
+
"learning_rate": 1.6160821185386364e-05,
|
| 5374 |
+
"loss": 2.6812,
|
| 5375 |
+
"step": 762
|
| 5376 |
+
},
|
| 5377 |
+
{
|
| 5378 |
+
"epoch": 0.8202096210696049,
|
| 5379 |
+
"grad_norm": 18.029285430908203,
|
| 5380 |
+
"learning_rate": 1.5975382925547965e-05,
|
| 5381 |
+
"loss": 2.8567,
|
| 5382 |
+
"step": 763
|
| 5383 |
+
},
|
| 5384 |
+
{
|
| 5385 |
+
"epoch": 0.8212846009137329,
|
| 5386 |
+
"grad_norm": 17.624483108520508,
|
| 5387 |
+
"learning_rate": 1.5790922322915958e-05,
|
| 5388 |
+
"loss": 3.039,
|
| 5389 |
+
"step": 764
|
| 5390 |
+
},
|
| 5391 |
+
{
|
| 5392 |
+
"epoch": 0.8223595807578608,
|
| 5393 |
+
"grad_norm": 21.21023178100586,
|
| 5394 |
+
"learning_rate": 1.5607441523756993e-05,
|
| 5395 |
+
"loss": 3.2504,
|
| 5396 |
+
"step": 765
|
| 5397 |
+
},
|
| 5398 |
+
{
|
| 5399 |
+
"epoch": 0.8234345606019887,
|
| 5400 |
+
"grad_norm": 15.159942626953125,
|
| 5401 |
+
"learning_rate": 1.5424942662937435e-05,
|
| 5402 |
+
"loss": 2.2915,
|
| 5403 |
+
"step": 766
|
| 5404 |
+
},
|
| 5405 |
+
{
|
| 5406 |
+
"epoch": 0.8245095404461167,
|
| 5407 |
+
"grad_norm": 16.94089126586914,
|
| 5408 |
+
"learning_rate": 1.5243427863898364e-05,
|
| 5409 |
+
"loss": 2.781,
|
| 5410 |
+
"step": 767
|
| 5411 |
+
},
|
| 5412 |
+
{
|
| 5413 |
+
"epoch": 0.8255845202902445,
|
| 5414 |
+
"grad_norm": 14.287386894226074,
|
| 5415 |
+
"learning_rate": 1.5062899238631e-05,
|
| 5416 |
+
"loss": 2.2052,
|
| 5417 |
+
"step": 768
|
| 5418 |
+
},
|
| 5419 |
+
{
|
| 5420 |
+
"epoch": 0.8266595001343725,
|
| 5421 |
+
"grad_norm": 23.06917953491211,
|
| 5422 |
+
"learning_rate": 1.4883358887652044e-05,
|
| 5423 |
+
"loss": 4.0856,
|
| 5424 |
+
"step": 769
|
| 5425 |
+
},
|
| 5426 |
+
{
|
| 5427 |
+
"epoch": 0.8277344799785004,
|
| 5428 |
+
"grad_norm": 14.637825012207031,
|
| 5429 |
+
"learning_rate": 1.4704808899979239e-05,
|
| 5430 |
+
"loss": 2.954,
|
| 5431 |
+
"step": 770
|
| 5432 |
+
},
|
| 5433 |
+
{
|
| 5434 |
+
"epoch": 0.8288094598226283,
|
| 5435 |
+
"grad_norm": 9.044286727905273,
|
| 5436 |
+
"learning_rate": 1.4527251353107163e-05,
|
| 5437 |
+
"loss": 1.8635,
|
| 5438 |
+
"step": 771
|
| 5439 |
+
},
|
| 5440 |
+
{
|
| 5441 |
+
"epoch": 0.8298844396667563,
|
| 5442 |
+
"grad_norm": 17.238901138305664,
|
| 5443 |
+
"learning_rate": 1.4350688312982864e-05,
|
| 5444 |
+
"loss": 2.5819,
|
| 5445 |
+
"step": 772
|
| 5446 |
+
},
|
| 5447 |
+
{
|
| 5448 |
+
"epoch": 0.8309594195108841,
|
| 5449 |
+
"grad_norm": 11.831186294555664,
|
| 5450 |
+
"learning_rate": 1.4175121833982052e-05,
|
| 5451 |
+
"loss": 2.4721,
|
| 5452 |
+
"step": 773
|
| 5453 |
+
},
|
| 5454 |
+
{
|
| 5455 |
+
"epoch": 0.8320343993550121,
|
| 5456 |
+
"grad_norm": 21.131061553955078,
|
| 5457 |
+
"learning_rate": 1.4000553958885021e-05,
|
| 5458 |
+
"loss": 4.1583,
|
| 5459 |
+
"step": 774
|
| 5460 |
+
},
|
| 5461 |
+
{
|
| 5462 |
+
"epoch": 0.83310937919914,
|
| 5463 |
+
"grad_norm": 15.86319637298584,
|
| 5464 |
+
"learning_rate": 1.3826986718852952e-05,
|
| 5465 |
+
"loss": 2.5534,
|
| 5466 |
+
"step": 775
|
| 5467 |
+
},
|
| 5468 |
+
{
|
| 5469 |
+
"epoch": 0.8341843590432679,
|
| 5470 |
+
"grad_norm": 16.642955780029297,
|
| 5471 |
+
"learning_rate": 1.365442213340432e-05,
|
| 5472 |
+
"loss": 2.3907,
|
| 5473 |
+
"step": 776
|
| 5474 |
+
},
|
| 5475 |
+
{
|
| 5476 |
+
"epoch": 0.8352593388873959,
|
| 5477 |
+
"grad_norm": 19.222917556762695,
|
| 5478 |
+
"learning_rate": 1.3482862210391245e-05,
|
| 5479 |
+
"loss": 3.667,
|
| 5480 |
+
"step": 777
|
| 5481 |
+
},
|
| 5482 |
+
{
|
| 5483 |
+
"epoch": 0.8363343187315238,
|
| 5484 |
+
"grad_norm": 18.172245025634766,
|
| 5485 |
+
"learning_rate": 1.3312308945976348e-05,
|
| 5486 |
+
"loss": 3.029,
|
| 5487 |
+
"step": 778
|
| 5488 |
+
},
|
| 5489 |
+
{
|
| 5490 |
+
"epoch": 0.8374092985756517,
|
| 5491 |
+
"grad_norm": 15.490896224975586,
|
| 5492 |
+
"learning_rate": 1.3142764324609303e-05,
|
| 5493 |
+
"loss": 3.208,
|
| 5494 |
+
"step": 779
|
| 5495 |
+
},
|
| 5496 |
+
{
|
| 5497 |
+
"epoch": 0.8384842784197797,
|
| 5498 |
+
"grad_norm": 18.96816062927246,
|
| 5499 |
+
"learning_rate": 1.2974230319003944e-05,
|
| 5500 |
+
"loss": 3.4618,
|
| 5501 |
+
"step": 780
|
| 5502 |
+
},
|
| 5503 |
+
{
|
| 5504 |
+
"epoch": 0.8395592582639075,
|
| 5505 |
+
"grad_norm": 21.89808464050293,
|
| 5506 |
+
"learning_rate": 1.2806708890115138e-05,
|
| 5507 |
+
"loss": 4.5078,
|
| 5508 |
+
"step": 781
|
| 5509 |
+
},
|
| 5510 |
+
{
|
| 5511 |
+
"epoch": 0.8406342381080355,
|
| 5512 |
+
"grad_norm": 14.64857006072998,
|
| 5513 |
+
"learning_rate": 1.2640201987116117e-05,
|
| 5514 |
+
"loss": 3.2866,
|
| 5515 |
+
"step": 782
|
| 5516 |
+
},
|
| 5517 |
+
{
|
| 5518 |
+
"epoch": 0.8417092179521634,
|
| 5519 |
+
"grad_norm": 15.451770782470703,
|
| 5520 |
+
"learning_rate": 1.2474711547375683e-05,
|
| 5521 |
+
"loss": 2.3176,
|
| 5522 |
+
"step": 783
|
| 5523 |
+
},
|
| 5524 |
+
{
|
| 5525 |
+
"epoch": 0.8427841977962913,
|
| 5526 |
+
"grad_norm": 12.625808715820312,
|
| 5527 |
+
"learning_rate": 1.2310239496435749e-05,
|
| 5528 |
+
"loss": 2.557,
|
| 5529 |
+
"step": 784
|
| 5530 |
+
},
|
| 5531 |
+
{
|
| 5532 |
+
"epoch": 0.8438591776404193,
|
| 5533 |
+
"grad_norm": 16.126930236816406,
|
| 5534 |
+
"learning_rate": 1.2146787747988919e-05,
|
| 5535 |
+
"loss": 2.9173,
|
| 5536 |
+
"step": 785
|
| 5537 |
+
},
|
| 5538 |
+
{
|
| 5539 |
+
"epoch": 0.8449341574845471,
|
| 5540 |
+
"grad_norm": 16.86185646057129,
|
| 5541 |
+
"learning_rate": 1.1984358203856116e-05,
|
| 5542 |
+
"loss": 2.4147,
|
| 5543 |
+
"step": 786
|
| 5544 |
+
},
|
| 5545 |
+
{
|
| 5546 |
+
"epoch": 0.8460091373286751,
|
| 5547 |
+
"grad_norm": 17.082571029663086,
|
| 5548 |
+
"learning_rate": 1.1822952753964667e-05,
|
| 5549 |
+
"loss": 2.9913,
|
| 5550 |
+
"step": 787
|
| 5551 |
+
},
|
| 5552 |
+
{
|
| 5553 |
+
"epoch": 0.847084117172803,
|
| 5554 |
+
"grad_norm": 15.173123359680176,
|
| 5555 |
+
"learning_rate": 1.1662573276326061e-05,
|
| 5556 |
+
"loss": 2.8246,
|
| 5557 |
+
"step": 788
|
| 5558 |
+
},
|
| 5559 |
+
{
|
| 5560 |
+
"epoch": 0.8481590970169309,
|
| 5561 |
+
"grad_norm": 17.791603088378906,
|
| 5562 |
+
"learning_rate": 1.1503221637014327e-05,
|
| 5563 |
+
"loss": 3.7321,
|
| 5564 |
+
"step": 789
|
| 5565 |
+
},
|
| 5566 |
+
{
|
| 5567 |
+
"epoch": 0.8492340768610589,
|
| 5568 |
+
"grad_norm": 16.170446395874023,
|
| 5569 |
+
"learning_rate": 1.134489969014414e-05,
|
| 5570 |
+
"loss": 3.1235,
|
| 5571 |
+
"step": 790
|
| 5572 |
+
},
|
| 5573 |
+
{
|
| 5574 |
+
"epoch": 0.8503090567051867,
|
| 5575 |
+
"grad_norm": 13.791912078857422,
|
| 5576 |
+
"learning_rate": 1.1187609277849376e-05,
|
| 5577 |
+
"loss": 2.7921,
|
| 5578 |
+
"step": 791
|
| 5579 |
+
},
|
| 5580 |
+
{
|
| 5581 |
+
"epoch": 0.8513840365493147,
|
| 5582 |
+
"grad_norm": 12.047784805297852,
|
| 5583 |
+
"learning_rate": 1.1031352230261637e-05,
|
| 5584 |
+
"loss": 1.8449,
|
| 5585 |
+
"step": 792
|
| 5586 |
+
},
|
| 5587 |
+
{
|
| 5588 |
+
"epoch": 0.8524590163934426,
|
| 5589 |
+
"grad_norm": 21.286800384521484,
|
| 5590 |
+
"learning_rate": 1.0876130365488878e-05,
|
| 5591 |
+
"loss": 3.6973,
|
| 5592 |
+
"step": 793
|
| 5593 |
+
},
|
| 5594 |
+
{
|
| 5595 |
+
"epoch": 0.8535339962375705,
|
| 5596 |
+
"grad_norm": 15.110943794250488,
|
| 5597 |
+
"learning_rate": 1.072194548959442e-05,
|
| 5598 |
+
"loss": 3.3045,
|
| 5599 |
+
"step": 794
|
| 5600 |
+
},
|
| 5601 |
+
{
|
| 5602 |
+
"epoch": 0.8546089760816985,
|
| 5603 |
+
"grad_norm": 11.98826789855957,
|
| 5604 |
+
"learning_rate": 1.0568799396575746e-05,
|
| 5605 |
+
"loss": 2.831,
|
| 5606 |
+
"step": 795
|
| 5607 |
+
},
|
| 5608 |
+
{
|
| 5609 |
+
"epoch": 0.8556839559258264,
|
| 5610 |
+
"grad_norm": 18.719282150268555,
|
| 5611 |
+
"learning_rate": 1.0416693868343797e-05,
|
| 5612 |
+
"loss": 3.2608,
|
| 5613 |
+
"step": 796
|
| 5614 |
+
},
|
| 5615 |
+
{
|
| 5616 |
+
"epoch": 0.8567589357699543,
|
| 5617 |
+
"grad_norm": 14.92679500579834,
|
| 5618 |
+
"learning_rate": 1.0265630674702076e-05,
|
| 5619 |
+
"loss": 2.494,
|
| 5620 |
+
"step": 797
|
| 5621 |
+
},
|
| 5622 |
+
{
|
| 5623 |
+
"epoch": 0.8578339156140823,
|
| 5624 |
+
"grad_norm": 12.818703651428223,
|
| 5625 |
+
"learning_rate": 1.0115611573326233e-05,
|
| 5626 |
+
"loss": 2.6008,
|
| 5627 |
+
"step": 798
|
| 5628 |
+
},
|
| 5629 |
+
{
|
| 5630 |
+
"epoch": 0.8589088954582101,
|
| 5631 |
+
"grad_norm": 17.74081039428711,
|
| 5632 |
+
"learning_rate": 9.966638309743482e-06,
|
| 5633 |
+
"loss": 3.2788,
|
| 5634 |
+
"step": 799
|
| 5635 |
+
},
|
| 5636 |
+
{
|
| 5637 |
+
"epoch": 0.8599838753023381,
|
| 5638 |
+
"grad_norm": 13.859757423400879,
|
| 5639 |
+
"learning_rate": 9.818712617312287e-06,
|
| 5640 |
+
"loss": 2.3176,
|
| 5641 |
+
"step": 800
|
| 5642 |
+
},
|
| 5643 |
+
{
|
| 5644 |
+
"epoch": 0.861058855146466,
|
| 5645 |
+
"grad_norm": 26.295276641845703,
|
| 5646 |
+
"learning_rate": 9.671836217202334e-06,
|
| 5647 |
+
"loss": 3.3959,
|
| 5648 |
+
"step": 801
|
| 5649 |
+
},
|
| 5650 |
+
{
|
| 5651 |
+
"epoch": 0.8621338349905939,
|
| 5652 |
+
"grad_norm": 15.147491455078125,
|
| 5653 |
+
"learning_rate": 9.52601081837431e-06,
|
| 5654 |
+
"loss": 3.1437,
|
| 5655 |
+
"step": 802
|
| 5656 |
+
},
|
| 5657 |
+
{
|
| 5658 |
+
"epoch": 0.8632088148347219,
|
| 5659 |
+
"grad_norm": 12.152910232543945,
|
| 5660 |
+
"learning_rate": 9.381238117560187e-06,
|
| 5661 |
+
"loss": 2.4838,
|
| 5662 |
+
"step": 803
|
| 5663 |
+
},
|
| 5664 |
+
{
|
| 5665 |
+
"epoch": 0.8642837946788497,
|
| 5666 |
+
"grad_norm": 28.314746856689453,
|
| 5667 |
+
"learning_rate": 9.237519799243355e-06,
|
| 5668 |
+
"loss": 3.2608,
|
| 5669 |
+
"step": 804
|
| 5670 |
+
},
|
| 5671 |
+
{
|
| 5672 |
+
"epoch": 0.8653587745229777,
|
| 5673 |
+
"grad_norm": 14.662104606628418,
|
| 5674 |
+
"learning_rate": 9.094857535639156e-06,
|
| 5675 |
+
"loss": 2.9242,
|
| 5676 |
+
"step": 805
|
| 5677 |
+
},
|
| 5678 |
+
{
|
| 5679 |
+
"epoch": 0.8664337543671056,
|
| 5680 |
+
"grad_norm": 13.321383476257324,
|
| 5681 |
+
"learning_rate": 8.95325298667523e-06,
|
| 5682 |
+
"loss": 2.8041,
|
| 5683 |
+
"step": 806
|
| 5684 |
+
},
|
| 5685 |
+
{
|
| 5686 |
+
"epoch": 0.8675087342112335,
|
| 5687 |
+
"grad_norm": 16.60413360595703,
|
| 5688 |
+
"learning_rate": 8.812707799972442e-06,
|
| 5689 |
+
"loss": 2.5522,
|
| 5690 |
+
"step": 807
|
| 5691 |
+
},
|
| 5692 |
+
{
|
| 5693 |
+
"epoch": 0.8685837140553615,
|
| 5694 |
+
"grad_norm": 16.389780044555664,
|
| 5695 |
+
"learning_rate": 8.673223610825531e-06,
|
| 5696 |
+
"loss": 2.647,
|
| 5697 |
+
"step": 808
|
| 5698 |
+
},
|
| 5699 |
+
{
|
| 5700 |
+
"epoch": 0.8696586938994894,
|
| 5701 |
+
"grad_norm": 10.362971305847168,
|
| 5702 |
+
"learning_rate": 8.53480204218412e-06,
|
| 5703 |
+
"loss": 2.1054,
|
| 5704 |
+
"step": 809
|
| 5705 |
+
},
|
| 5706 |
+
{
|
| 5707 |
+
"epoch": 0.8707336737436173,
|
| 5708 |
+
"grad_norm": 18.633100509643555,
|
| 5709 |
+
"learning_rate": 8.397444704633906e-06,
|
| 5710 |
+
"loss": 3.8698,
|
| 5711 |
+
"step": 810
|
| 5712 |
+
},
|
| 5713 |
+
{
|
| 5714 |
+
"epoch": 0.8718086535877452,
|
| 5715 |
+
"grad_norm": 14.801193237304688,
|
| 5716 |
+
"learning_rate": 8.261153196377814e-06,
|
| 5717 |
+
"loss": 2.1947,
|
| 5718 |
+
"step": 811
|
| 5719 |
+
},
|
| 5720 |
+
{
|
| 5721 |
+
"epoch": 0.8728836334318731,
|
| 5722 |
+
"grad_norm": 16.298856735229492,
|
| 5723 |
+
"learning_rate": 8.1259291032175e-06,
|
| 5724 |
+
"loss": 3.4508,
|
| 5725 |
+
"step": 812
|
| 5726 |
+
},
|
| 5727 |
+
{
|
| 5728 |
+
"epoch": 0.8739586132760011,
|
| 5729 |
+
"grad_norm": 17.139835357666016,
|
| 5730 |
+
"learning_rate": 7.991773998534802e-06,
|
| 5731 |
+
"loss": 2.6625,
|
| 5732 |
+
"step": 813
|
| 5733 |
+
},
|
| 5734 |
+
{
|
| 5735 |
+
"epoch": 0.875033593120129,
|
| 5736 |
+
"grad_norm": 14.451316833496094,
|
| 5737 |
+
"learning_rate": 7.858689443273547e-06,
|
| 5738 |
+
"loss": 2.5462,
|
| 5739 |
+
"step": 814
|
| 5740 |
+
},
|
| 5741 |
+
{
|
| 5742 |
+
"epoch": 0.8761085729642569,
|
| 5743 |
+
"grad_norm": 19.52531623840332,
|
| 5744 |
+
"learning_rate": 7.72667698592131e-06,
|
| 5745 |
+
"loss": 2.7614,
|
| 5746 |
+
"step": 815
|
| 5747 |
+
},
|
| 5748 |
+
{
|
| 5749 |
+
"epoch": 0.8771835528083849,
|
| 5750 |
+
"grad_norm": 23.297826766967773,
|
| 5751 |
+
"learning_rate": 7.595738162491383e-06,
|
| 5752 |
+
"loss": 2.0079,
|
| 5753 |
+
"step": 816
|
| 5754 |
+
},
|
| 5755 |
+
{
|
| 5756 |
+
"epoch": 0.8782585326525127,
|
| 5757 |
+
"grad_norm": 16.850067138671875,
|
| 5758 |
+
"learning_rate": 7.465874496504943e-06,
|
| 5759 |
+
"loss": 3.0355,
|
| 5760 |
+
"step": 817
|
| 5761 |
+
},
|
| 5762 |
+
{
|
| 5763 |
+
"epoch": 0.8793335124966407,
|
| 5764 |
+
"grad_norm": 19.124881744384766,
|
| 5765 |
+
"learning_rate": 7.337087498973327e-06,
|
| 5766 |
+
"loss": 3.1917,
|
| 5767 |
+
"step": 818
|
| 5768 |
+
},
|
| 5769 |
+
{
|
| 5770 |
+
"epoch": 0.8804084923407686,
|
| 5771 |
+
"grad_norm": 16.559707641601562,
|
| 5772 |
+
"learning_rate": 7.209378668380451e-06,
|
| 5773 |
+
"loss": 2.6744,
|
| 5774 |
+
"step": 819
|
| 5775 |
+
},
|
| 5776 |
+
{
|
| 5777 |
+
"epoch": 0.8814834721848965,
|
| 5778 |
+
"grad_norm": 11.441034317016602,
|
| 5779 |
+
"learning_rate": 7.0827494906653526e-06,
|
| 5780 |
+
"loss": 2.2681,
|
| 5781 |
+
"step": 820
|
| 5782 |
+
},
|
| 5783 |
+
{
|
| 5784 |
+
"epoch": 0.8825584520290245,
|
| 5785 |
+
"grad_norm": 18.97810935974121,
|
| 5786 |
+
"learning_rate": 6.957201439204897e-06,
|
| 5787 |
+
"loss": 3.6872,
|
| 5788 |
+
"step": 821
|
| 5789 |
+
},
|
| 5790 |
+
{
|
| 5791 |
+
"epoch": 0.8836334318731524,
|
| 5792 |
+
"grad_norm": 15.341508865356445,
|
| 5793 |
+
"learning_rate": 6.832735974796689e-06,
|
| 5794 |
+
"loss": 2.0165,
|
| 5795 |
+
"step": 822
|
| 5796 |
+
},
|
| 5797 |
+
{
|
| 5798 |
+
"epoch": 0.8847084117172803,
|
| 5799 |
+
"grad_norm": 16.359432220458984,
|
| 5800 |
+
"learning_rate": 6.7093545456419886e-06,
|
| 5801 |
+
"loss": 3.3588,
|
| 5802 |
+
"step": 823
|
| 5803 |
+
},
|
| 5804 |
+
{
|
| 5805 |
+
"epoch": 0.8857833915614082,
|
| 5806 |
+
"grad_norm": 16.616304397583008,
|
| 5807 |
+
"learning_rate": 6.5870585873289425e-06,
|
| 5808 |
+
"loss": 3.643,
|
| 5809 |
+
"step": 824
|
| 5810 |
+
},
|
| 5811 |
+
{
|
| 5812 |
+
"epoch": 0.8868583714055361,
|
| 5813 |
+
"grad_norm": 14.876137733459473,
|
| 5814 |
+
"learning_rate": 6.4658495228158146e-06,
|
| 5815 |
+
"loss": 2.8213,
|
| 5816 |
+
"step": 825
|
| 5817 |
+
},
|
| 5818 |
+
{
|
| 5819 |
+
"epoch": 0.8879333512496641,
|
| 5820 |
+
"grad_norm": 11.03225326538086,
|
| 5821 |
+
"learning_rate": 6.345728762414505e-06,
|
| 5822 |
+
"loss": 2.1051,
|
| 5823 |
+
"step": 826
|
| 5824 |
+
},
|
| 5825 |
+
{
|
| 5826 |
+
"epoch": 0.889008331093792,
|
| 5827 |
+
"grad_norm": 17.549360275268555,
|
| 5828 |
+
"learning_rate": 6.226697703774076e-06,
|
| 5829 |
+
"loss": 3.0634,
|
| 5830 |
+
"step": 827
|
| 5831 |
+
},
|
| 5832 |
+
{
|
| 5833 |
+
"epoch": 0.8900833109379199,
|
| 5834 |
+
"grad_norm": 12.46760368347168,
|
| 5835 |
+
"learning_rate": 6.108757731864489e-06,
|
| 5836 |
+
"loss": 2.7482,
|
| 5837 |
+
"step": 828
|
| 5838 |
+
},
|
| 5839 |
+
{
|
| 5840 |
+
"epoch": 0.8911582907820479,
|
| 5841 |
+
"grad_norm": 11.985771179199219,
|
| 5842 |
+
"learning_rate": 5.99191021896055e-06,
|
| 5843 |
+
"loss": 2.4109,
|
| 5844 |
+
"step": 829
|
| 5845 |
+
},
|
| 5846 |
+
{
|
| 5847 |
+
"epoch": 0.8922332706261757,
|
| 5848 |
+
"grad_norm": 15.262474060058594,
|
| 5849 |
+
"learning_rate": 5.876156524625864e-06,
|
| 5850 |
+
"loss": 3.0215,
|
| 5851 |
+
"step": 830
|
| 5852 |
+
},
|
| 5853 |
+
{
|
| 5854 |
+
"epoch": 0.8933082504703037,
|
| 5855 |
+
"grad_norm": 12.595239639282227,
|
| 5856 |
+
"learning_rate": 5.7614979956971075e-06,
|
| 5857 |
+
"loss": 2.726,
|
| 5858 |
+
"step": 831
|
| 5859 |
+
},
|
| 5860 |
+
{
|
| 5861 |
+
"epoch": 0.8943832303144316,
|
| 5862 |
+
"grad_norm": 18.540124893188477,
|
| 5863 |
+
"learning_rate": 5.647935966268225e-06,
|
| 5864 |
+
"loss": 1.9021,
|
| 5865 |
+
"step": 832
|
| 5866 |
+
},
|
| 5867 |
+
{
|
| 5868 |
+
"epoch": 0.8954582101585595,
|
| 5869 |
+
"grad_norm": 12.652937889099121,
|
| 5870 |
+
"learning_rate": 5.5354717576750816e-06,
|
| 5871 |
+
"loss": 2.366,
|
| 5872 |
+
"step": 833
|
| 5873 |
+
},
|
| 5874 |
+
{
|
| 5875 |
+
"epoch": 0.8965331900026875,
|
| 5876 |
+
"grad_norm": 13.983992576599121,
|
| 5877 |
+
"learning_rate": 5.424106678479945e-06,
|
| 5878 |
+
"loss": 2.7013,
|
| 5879 |
+
"step": 834
|
| 5880 |
+
},
|
| 5881 |
+
{
|
| 5882 |
+
"epoch": 0.8976081698468154,
|
| 5883 |
+
"grad_norm": 15.04990291595459,
|
| 5884 |
+
"learning_rate": 5.313842024456306e-06,
|
| 5885 |
+
"loss": 2.3414,
|
| 5886 |
+
"step": 835
|
| 5887 |
+
},
|
| 5888 |
+
{
|
| 5889 |
+
"epoch": 0.8986831496909433,
|
| 5890 |
+
"grad_norm": 18.419647216796875,
|
| 5891 |
+
"learning_rate": 5.204679078573827e-06,
|
| 5892 |
+
"loss": 3.8336,
|
| 5893 |
+
"step": 836
|
| 5894 |
+
},
|
| 5895 |
+
{
|
| 5896 |
+
"epoch": 0.8997581295350712,
|
| 5897 |
+
"grad_norm": 17.847749710083008,
|
| 5898 |
+
"learning_rate": 5.096619110983347e-06,
|
| 5899 |
+
"loss": 2.9503,
|
| 5900 |
+
"step": 837
|
| 5901 |
+
},
|
| 5902 |
+
{
|
| 5903 |
+
"epoch": 0.9008331093791991,
|
| 5904 |
+
"grad_norm": 15.034934043884277,
|
| 5905 |
+
"learning_rate": 4.9896633790022405e-06,
|
| 5906 |
+
"loss": 2.6579,
|
| 5907 |
+
"step": 838
|
| 5908 |
+
},
|
| 5909 |
+
{
|
| 5910 |
+
"epoch": 0.9019080892233271,
|
| 5911 |
+
"grad_norm": 16.45047950744629,
|
| 5912 |
+
"learning_rate": 4.883813127099579e-06,
|
| 5913 |
+
"loss": 3.8603,
|
| 5914 |
+
"step": 839
|
| 5915 |
+
},
|
| 5916 |
+
{
|
| 5917 |
+
"epoch": 0.902983069067455,
|
| 5918 |
+
"grad_norm": 14.44510269165039,
|
| 5919 |
+
"learning_rate": 4.779069586881857e-06,
|
| 5920 |
+
"loss": 2.7965,
|
| 5921 |
+
"step": 840
|
| 5922 |
+
},
|
| 5923 |
+
{
|
| 5924 |
+
"epoch": 0.9040580489115829,
|
| 5925 |
+
"grad_norm": 14.151533126831055,
|
| 5926 |
+
"learning_rate": 4.675433977078547e-06,
|
| 5927 |
+
"loss": 2.3626,
|
| 5928 |
+
"step": 841
|
| 5929 |
+
},
|
| 5930 |
+
{
|
| 5931 |
+
"epoch": 0.9051330287557108,
|
| 5932 |
+
"grad_norm": 20.559478759765625,
|
| 5933 |
+
"learning_rate": 4.572907503527923e-06,
|
| 5934 |
+
"loss": 2.3899,
|
| 5935 |
+
"step": 842
|
| 5936 |
+
},
|
| 5937 |
+
{
|
| 5938 |
+
"epoch": 0.9062080085998387,
|
| 5939 |
+
"grad_norm": 11.681385040283203,
|
| 5940 |
+
"learning_rate": 4.471491359163094e-06,
|
| 5941 |
+
"loss": 2.3098,
|
| 5942 |
+
"step": 843
|
| 5943 |
+
},
|
| 5944 |
+
{
|
| 5945 |
+
"epoch": 0.9072829884439667,
|
| 5946 |
+
"grad_norm": 16.471866607666016,
|
| 5947 |
+
"learning_rate": 4.3711867239980335e-06,
|
| 5948 |
+
"loss": 2.6948,
|
| 5949 |
+
"step": 844
|
| 5950 |
+
},
|
| 5951 |
+
{
|
| 5952 |
+
"epoch": 0.9083579682880946,
|
| 5953 |
+
"grad_norm": 10.209484100341797,
|
| 5954 |
+
"learning_rate": 4.271994765113952e-06,
|
| 5955 |
+
"loss": 2.1588,
|
| 5956 |
+
"step": 845
|
| 5957 |
+
},
|
| 5958 |
+
{
|
| 5959 |
+
"epoch": 0.9094329481322225,
|
| 5960 |
+
"grad_norm": 15.725363731384277,
|
| 5961 |
+
"learning_rate": 4.173916636645591e-06,
|
| 5962 |
+
"loss": 2.8127,
|
| 5963 |
+
"step": 846
|
| 5964 |
+
},
|
| 5965 |
+
{
|
| 5966 |
+
"epoch": 0.9105079279763505,
|
| 5967 |
+
"grad_norm": 15.666772842407227,
|
| 5968 |
+
"learning_rate": 4.0769534797679645e-06,
|
| 5969 |
+
"loss": 2.667,
|
| 5970 |
+
"step": 847
|
| 5971 |
+
},
|
| 5972 |
+
{
|
| 5973 |
+
"epoch": 0.9115829078204783,
|
| 5974 |
+
"grad_norm": 15.909188270568848,
|
| 5975 |
+
"learning_rate": 3.9811064226828895e-06,
|
| 5976 |
+
"loss": 3.317,
|
| 5977 |
+
"step": 848
|
| 5978 |
+
},
|
| 5979 |
+
{
|
| 5980 |
+
"epoch": 0.9126578876646063,
|
| 5981 |
+
"grad_norm": 11.588101387023926,
|
| 5982 |
+
"learning_rate": 3.8863765806060105e-06,
|
| 5983 |
+
"loss": 1.9476,
|
| 5984 |
+
"step": 849
|
| 5985 |
+
},
|
| 5986 |
+
{
|
| 5987 |
+
"epoch": 0.9137328675087342,
|
| 5988 |
+
"grad_norm": 10.225213050842285,
|
| 5989 |
+
"learning_rate": 3.7927650557537555e-06,
|
| 5990 |
+
"loss": 2.1618,
|
| 5991 |
+
"step": 850
|
| 5992 |
+
},
|
| 5993 |
+
{
|
| 5994 |
+
"epoch": 0.9148078473528621,
|
| 5995 |
+
"grad_norm": 19.92328643798828,
|
| 5996 |
+
"learning_rate": 3.7002729373304957e-06,
|
| 5997 |
+
"loss": 2.6487,
|
| 5998 |
+
"step": 851
|
| 5999 |
+
},
|
| 6000 |
+
{
|
| 6001 |
+
"epoch": 0.9158828271969901,
|
| 6002 |
+
"grad_norm": 21.881338119506836,
|
| 6003 |
+
"learning_rate": 3.6089013015159433e-06,
|
| 6004 |
+
"loss": 4.0831,
|
| 6005 |
+
"step": 852
|
| 6006 |
+
},
|
| 6007 |
+
{
|
| 6008 |
+
"epoch": 0.916957807041118,
|
| 6009 |
+
"grad_norm": 16.8553524017334,
|
| 6010 |
+
"learning_rate": 3.5186512114525282e-06,
|
| 6011 |
+
"loss": 3.4859,
|
| 6012 |
+
"step": 853
|
| 6013 |
+
},
|
| 6014 |
+
{
|
| 6015 |
+
"epoch": 0.9180327868852459,
|
| 6016 |
+
"grad_norm": 15.485822677612305,
|
| 6017 |
+
"learning_rate": 3.4295237172331516e-06,
|
| 6018 |
+
"loss": 2.7058,
|
| 6019 |
+
"step": 854
|
| 6020 |
+
},
|
| 6021 |
+
{
|
| 6022 |
+
"epoch": 0.9191077667293738,
|
| 6023 |
+
"grad_norm": 11.821995735168457,
|
| 6024 |
+
"learning_rate": 3.3415198558888305e-06,
|
| 6025 |
+
"loss": 2.4467,
|
| 6026 |
+
"step": 855
|
| 6027 |
+
},
|
| 6028 |
+
{
|
| 6029 |
+
"epoch": 0.9201827465735017,
|
| 6030 |
+
"grad_norm": 10.078314781188965,
|
| 6031 |
+
"learning_rate": 3.2546406513767504e-06,
|
| 6032 |
+
"loss": 1.897,
|
| 6033 |
+
"step": 856
|
| 6034 |
+
},
|
| 6035 |
+
{
|
| 6036 |
+
"epoch": 0.9212577264176297,
|
| 6037 |
+
"grad_norm": 13.78282356262207,
|
| 6038 |
+
"learning_rate": 3.1688871145683086e-06,
|
| 6039 |
+
"loss": 2.9327,
|
| 6040 |
+
"step": 857
|
| 6041 |
+
},
|
| 6042 |
+
{
|
| 6043 |
+
"epoch": 0.9223327062617576,
|
| 6044 |
+
"grad_norm": 17.97776222229004,
|
| 6045 |
+
"learning_rate": 3.0842602432373024e-06,
|
| 6046 |
+
"loss": 2.6016,
|
| 6047 |
+
"step": 858
|
| 6048 |
+
},
|
| 6049 |
+
{
|
| 6050 |
+
"epoch": 0.9234076861058855,
|
| 6051 |
+
"grad_norm": 17.566068649291992,
|
| 6052 |
+
"learning_rate": 3.0007610220483927e-06,
|
| 6053 |
+
"loss": 2.7243,
|
| 6054 |
+
"step": 859
|
| 6055 |
+
},
|
| 6056 |
+
{
|
| 6057 |
+
"epoch": 0.9244826659500134,
|
| 6058 |
+
"grad_norm": 12.550614356994629,
|
| 6059 |
+
"learning_rate": 2.918390422545614e-06,
|
| 6060 |
+
"loss": 2.2397,
|
| 6061 |
+
"step": 860
|
| 6062 |
+
},
|
| 6063 |
+
{
|
| 6064 |
+
"epoch": 0.9255576457941413,
|
| 6065 |
+
"grad_norm": 26.070829391479492,
|
| 6066 |
+
"learning_rate": 2.8371494031410704e-06,
|
| 6067 |
+
"loss": 3.6046,
|
| 6068 |
+
"step": 861
|
| 6069 |
+
},
|
| 6070 |
+
{
|
| 6071 |
+
"epoch": 0.9266326256382693,
|
| 6072 |
+
"grad_norm": 20.401790618896484,
|
| 6073 |
+
"learning_rate": 2.757038909103793e-06,
|
| 6074 |
+
"loss": 2.7354,
|
| 6075 |
+
"step": 862
|
| 6076 |
+
},
|
| 6077 |
+
{
|
| 6078 |
+
"epoch": 0.9277076054823972,
|
| 6079 |
+
"grad_norm": 10.504318237304688,
|
| 6080 |
+
"learning_rate": 2.6780598725487214e-06,
|
| 6081 |
+
"loss": 2.4165,
|
| 6082 |
+
"step": 863
|
| 6083 |
+
},
|
| 6084 |
+
{
|
| 6085 |
+
"epoch": 0.9287825853265251,
|
| 6086 |
+
"grad_norm": 12.440912246704102,
|
| 6087 |
+
"learning_rate": 2.6002132124258947e-06,
|
| 6088 |
+
"loss": 2.0556,
|
| 6089 |
+
"step": 864
|
| 6090 |
+
},
|
| 6091 |
+
{
|
| 6092 |
+
"epoch": 0.9298575651706531,
|
| 6093 |
+
"grad_norm": 15.771154403686523,
|
| 6094 |
+
"learning_rate": 2.5234998345097238e-06,
|
| 6095 |
+
"loss": 3.0948,
|
| 6096 |
+
"step": 865
|
| 6097 |
+
},
|
| 6098 |
+
{
|
| 6099 |
+
"epoch": 0.930932545014781,
|
| 6100 |
+
"grad_norm": 13.663370132446289,
|
| 6101 |
+
"learning_rate": 2.4479206313884784e-06,
|
| 6102 |
+
"loss": 2.6857,
|
| 6103 |
+
"step": 866
|
| 6104 |
+
},
|
| 6105 |
+
{
|
| 6106 |
+
"epoch": 0.9320075248589089,
|
| 6107 |
+
"grad_norm": 12.99375057220459,
|
| 6108 |
+
"learning_rate": 2.3734764824538515e-06,
|
| 6109 |
+
"loss": 2.46,
|
| 6110 |
+
"step": 867
|
| 6111 |
+
},
|
| 6112 |
+
{
|
| 6113 |
+
"epoch": 0.9330825047030368,
|
| 6114 |
+
"grad_norm": 10.756906509399414,
|
| 6115 |
+
"learning_rate": 2.300168253890833e-06,
|
| 6116 |
+
"loss": 2.2951,
|
| 6117 |
+
"step": 868
|
| 6118 |
+
},
|
| 6119 |
+
{
|
| 6120 |
+
"epoch": 0.9341574845471647,
|
| 6121 |
+
"grad_norm": 15.61550521850586,
|
| 6122 |
+
"learning_rate": 2.2279967986674756e-06,
|
| 6123 |
+
"loss": 2.947,
|
| 6124 |
+
"step": 869
|
| 6125 |
+
},
|
| 6126 |
+
{
|
| 6127 |
+
"epoch": 0.9352324643912927,
|
| 6128 |
+
"grad_norm": 17.821706771850586,
|
| 6129 |
+
"learning_rate": 2.1569629565251546e-06,
|
| 6130 |
+
"loss": 2.9358,
|
| 6131 |
+
"step": 870
|
| 6132 |
+
},
|
| 6133 |
+
{
|
| 6134 |
+
"epoch": 0.9363074442354206,
|
| 6135 |
+
"grad_norm": 13.073902130126953,
|
| 6136 |
+
"learning_rate": 2.0870675539686023e-06,
|
| 6137 |
+
"loss": 2.665,
|
| 6138 |
+
"step": 871
|
| 6139 |
+
},
|
| 6140 |
+
{
|
| 6141 |
+
"epoch": 0.9373824240795485,
|
| 6142 |
+
"grad_norm": 16.576457977294922,
|
| 6143 |
+
"learning_rate": 2.0183114042564567e-06,
|
| 6144 |
+
"loss": 3.3876,
|
| 6145 |
+
"step": 872
|
| 6146 |
+
},
|
| 6147 |
+
{
|
| 6148 |
+
"epoch": 0.9384574039236764,
|
| 6149 |
+
"grad_norm": 12.26412296295166,
|
| 6150 |
+
"learning_rate": 1.9506953073917365e-06,
|
| 6151 |
+
"loss": 2.3773,
|
| 6152 |
+
"step": 873
|
| 6153 |
+
},
|
| 6154 |
+
{
|
| 6155 |
+
"epoch": 0.9395323837678043,
|
| 6156 |
+
"grad_norm": 15.137334823608398,
|
| 6157 |
+
"learning_rate": 1.8842200501124618e-06,
|
| 6158 |
+
"loss": 2.353,
|
| 6159 |
+
"step": 874
|
| 6160 |
+
},
|
| 6161 |
+
{
|
| 6162 |
+
"epoch": 0.9406073636119323,
|
| 6163 |
+
"grad_norm": 11.213092803955078,
|
| 6164 |
+
"learning_rate": 1.818886405882636e-06,
|
| 6165 |
+
"loss": 1.8275,
|
| 6166 |
+
"step": 875
|
| 6167 |
+
},
|
| 6168 |
+
{
|
| 6169 |
+
"epoch": 0.9416823434560602,
|
| 6170 |
+
"grad_norm": 12.951726913452148,
|
| 6171 |
+
"learning_rate": 1.7546951348831441e-06,
|
| 6172 |
+
"loss": 2.4801,
|
| 6173 |
+
"step": 876
|
| 6174 |
+
},
|
| 6175 |
+
{
|
| 6176 |
+
"epoch": 0.9427573233001881,
|
| 6177 |
+
"grad_norm": 9.286628723144531,
|
| 6178 |
+
"learning_rate": 1.6916469840029369e-06,
|
| 6179 |
+
"loss": 1.8095,
|
| 6180 |
+
"step": 877
|
| 6181 |
+
},
|
| 6182 |
+
{
|
| 6183 |
+
"epoch": 0.9438323031443161,
|
| 6184 |
+
"grad_norm": 16.789724349975586,
|
| 6185 |
+
"learning_rate": 1.6297426868303378e-06,
|
| 6186 |
+
"loss": 2.2362,
|
| 6187 |
+
"step": 878
|
| 6188 |
+
},
|
| 6189 |
+
{
|
| 6190 |
+
"epoch": 0.944907282988444,
|
| 6191 |
+
"grad_norm": 9.946615219116211,
|
| 6192 |
+
"learning_rate": 1.5689829636445496e-06,
|
| 6193 |
+
"loss": 2.0806,
|
| 6194 |
+
"step": 879
|
| 6195 |
+
},
|
| 6196 |
+
{
|
| 6197 |
+
"epoch": 0.9459822628325719,
|
| 6198 |
+
"grad_norm": 10.376627922058105,
|
| 6199 |
+
"learning_rate": 1.5093685214072174e-06,
|
| 6200 |
+
"loss": 2.7637,
|
| 6201 |
+
"step": 880
|
| 6202 |
+
},
|
| 6203 |
+
{
|
| 6204 |
+
"epoch": 0.9470572426766998,
|
| 6205 |
+
"grad_norm": 10.574774742126465,
|
| 6206 |
+
"learning_rate": 1.4509000537541895e-06,
|
| 6207 |
+
"loss": 2.6576,
|
| 6208 |
+
"step": 881
|
| 6209 |
+
},
|
| 6210 |
+
{
|
| 6211 |
+
"epoch": 0.9481322225208277,
|
| 6212 |
+
"grad_norm": 19.413841247558594,
|
| 6213 |
+
"learning_rate": 1.3935782409875476e-06,
|
| 6214 |
+
"loss": 3.1755,
|
| 6215 |
+
"step": 882
|
| 6216 |
+
},
|
| 6217 |
+
{
|
| 6218 |
+
"epoch": 0.9492072023649557,
|
| 6219 |
+
"grad_norm": 24.996675491333008,
|
| 6220 |
+
"learning_rate": 1.337403750067545e-06,
|
| 6221 |
+
"loss": 4.3345,
|
| 6222 |
+
"step": 883
|
| 6223 |
+
},
|
| 6224 |
+
{
|
| 6225 |
+
"epoch": 0.9502821822090836,
|
| 6226 |
+
"grad_norm": 15.001642227172852,
|
| 6227 |
+
"learning_rate": 1.2823772346050034e-06,
|
| 6228 |
+
"loss": 3.8014,
|
| 6229 |
+
"step": 884
|
| 6230 |
+
},
|
| 6231 |
+
{
|
| 6232 |
+
"epoch": 0.9513571620532115,
|
| 6233 |
+
"grad_norm": 20.97924041748047,
|
| 6234 |
+
"learning_rate": 1.2284993348535723e-06,
|
| 6235 |
+
"loss": 3.1809,
|
| 6236 |
+
"step": 885
|
| 6237 |
+
},
|
| 6238 |
+
{
|
| 6239 |
+
"epoch": 0.9524321418973394,
|
| 6240 |
+
"grad_norm": 12.326496124267578,
|
| 6241 |
+
"learning_rate": 1.1757706777023592e-06,
|
| 6242 |
+
"loss": 2.0848,
|
| 6243 |
+
"step": 886
|
| 6244 |
+
},
|
| 6245 |
+
{
|
| 6246 |
+
"epoch": 0.9535071217414673,
|
| 6247 |
+
"grad_norm": 16.77161979675293,
|
| 6248 |
+
"learning_rate": 1.1241918766686122e-06,
|
| 6249 |
+
"loss": 2.5849,
|
| 6250 |
+
"step": 887
|
| 6251 |
+
},
|
| 6252 |
+
{
|
| 6253 |
+
"epoch": 0.9545821015855953,
|
| 6254 |
+
"grad_norm": 18.30064582824707,
|
| 6255 |
+
"learning_rate": 1.0737635318905704e-06,
|
| 6256 |
+
"loss": 2.8427,
|
| 6257 |
+
"step": 888
|
| 6258 |
+
},
|
| 6259 |
+
{
|
| 6260 |
+
"epoch": 0.9556570814297232,
|
| 6261 |
+
"grad_norm": 16.708187103271484,
|
| 6262 |
+
"learning_rate": 1.0244862301205249e-06,
|
| 6263 |
+
"loss": 3.3412,
|
| 6264 |
+
"step": 889
|
| 6265 |
+
},
|
| 6266 |
+
{
|
| 6267 |
+
"epoch": 0.9567320612738511,
|
| 6268 |
+
"grad_norm": 9.98150634765625,
|
| 6269 |
+
"learning_rate": 9.763605447179137e-07,
|
| 6270 |
+
"loss": 1.9673,
|
| 6271 |
+
"step": 890
|
| 6272 |
+
},
|
| 6273 |
+
{
|
| 6274 |
+
"epoch": 0.957807041117979,
|
| 6275 |
+
"grad_norm": 16.982563018798828,
|
| 6276 |
+
"learning_rate": 9.293870356427259e-07,
|
| 6277 |
+
"loss": 2.7158,
|
| 6278 |
+
"step": 891
|
| 6279 |
+
},
|
| 6280 |
+
{
|
| 6281 |
+
"epoch": 0.958882020962107,
|
| 6282 |
+
"grad_norm": 14.91385555267334,
|
| 6283 |
+
"learning_rate": 8.835662494489638e-07,
|
| 6284 |
+
"loss": 2.4724,
|
| 6285 |
+
"step": 892
|
| 6286 |
+
},
|
| 6287 |
+
{
|
| 6288 |
+
"epoch": 0.9599570008062349,
|
| 6289 |
+
"grad_norm": 11.67471981048584,
|
| 6290 |
+
"learning_rate": 8.388987192782472e-07,
|
| 6291 |
+
"loss": 2.0277,
|
| 6292 |
+
"step": 893
|
| 6293 |
+
},
|
| 6294 |
+
{
|
| 6295 |
+
"epoch": 0.9610319806503628,
|
| 6296 |
+
"grad_norm": 17.66205596923828,
|
| 6297 |
+
"learning_rate": 7.953849648536516e-07,
|
| 6298 |
+
"loss": 3.0247,
|
| 6299 |
+
"step": 894
|
| 6300 |
+
},
|
| 6301 |
+
{
|
| 6302 |
+
"epoch": 0.9621069604944907,
|
| 6303 |
+
"grad_norm": 12.348430633544922,
|
| 6304 |
+
"learning_rate": 7.53025492473669e-07,
|
| 6305 |
+
"loss": 2.1941,
|
| 6306 |
+
"step": 895
|
| 6307 |
+
},
|
| 6308 |
+
{
|
| 6309 |
+
"epoch": 0.9631819403386187,
|
| 6310 |
+
"grad_norm": 11.82780647277832,
|
| 6311 |
+
"learning_rate": 7.118207950062905e-07,
|
| 6312 |
+
"loss": 1.9668,
|
| 6313 |
+
"step": 896
|
| 6314 |
+
},
|
| 6315 |
+
{
|
| 6316 |
+
"epoch": 0.9642569201827466,
|
| 6317 |
+
"grad_norm": 11.388262748718262,
|
| 6318 |
+
"learning_rate": 6.717713518832325e-07,
|
| 6319 |
+
"loss": 1.7028,
|
| 6320 |
+
"step": 897
|
| 6321 |
+
},
|
| 6322 |
+
{
|
| 6323 |
+
"epoch": 0.9653319000268745,
|
| 6324 |
+
"grad_norm": 20.69382095336914,
|
| 6325 |
+
"learning_rate": 6.328776290944749e-07,
|
| 6326 |
+
"loss": 2.4699,
|
| 6327 |
+
"step": 898
|
| 6328 |
+
},
|
| 6329 |
+
{
|
| 6330 |
+
"epoch": 0.9664068798710024,
|
| 6331 |
+
"grad_norm": 16.286813735961914,
|
| 6332 |
+
"learning_rate": 5.9514007918271e-07,
|
| 6333 |
+
"loss": 2.5732,
|
| 6334 |
+
"step": 899
|
| 6335 |
+
},
|
| 6336 |
+
{
|
| 6337 |
+
"epoch": 0.9674818597151303,
|
| 6338 |
+
"grad_norm": 16.142581939697266,
|
| 6339 |
+
"learning_rate": 5.585591412381797e-07,
|
| 6340 |
+
"loss": 3.1468,
|
| 6341 |
+
"step": 900
|
| 6342 |
+
},
|
| 6343 |
+
{
|
| 6344 |
+
"epoch": 0.9685568395592583,
|
| 6345 |
+
"grad_norm": 21.596254348754883,
|
| 6346 |
+
"learning_rate": 5.231352408934686e-07,
|
| 6347 |
+
"loss": 3.7028,
|
| 6348 |
+
"step": 901
|
| 6349 |
+
},
|
| 6350 |
+
{
|
| 6351 |
+
"epoch": 0.9696318194033862,
|
| 6352 |
+
"grad_norm": 10.433320999145508,
|
| 6353 |
+
"learning_rate": 4.88868790318675e-07,
|
| 6354 |
+
"loss": 2.0005,
|
| 6355 |
+
"step": 902
|
| 6356 |
+
},
|
| 6357 |
+
{
|
| 6358 |
+
"epoch": 0.9707067992475141,
|
| 6359 |
+
"grad_norm": 20.765460968017578,
|
| 6360 |
+
"learning_rate": 4.557601882164808e-07,
|
| 6361 |
+
"loss": 3.7903,
|
| 6362 |
+
"step": 903
|
| 6363 |
+
},
|
| 6364 |
+
{
|
| 6365 |
+
"epoch": 0.971781779091642,
|
| 6366 |
+
"grad_norm": 18.080036163330078,
|
| 6367 |
+
"learning_rate": 4.2380981981759994e-07,
|
| 6368 |
+
"loss": 3.3676,
|
| 6369 |
+
"step": 904
|
| 6370 |
+
},
|
| 6371 |
+
{
|
| 6372 |
+
"epoch": 0.97285675893577,
|
| 6373 |
+
"grad_norm": 14.852317810058594,
|
| 6374 |
+
"learning_rate": 3.930180568762931e-07,
|
| 6375 |
+
"loss": 2.6543,
|
| 6376 |
+
"step": 905
|
| 6377 |
+
},
|
| 6378 |
+
{
|
| 6379 |
+
"epoch": 0.9739317387798979,
|
| 6380 |
+
"grad_norm": 18.983341217041016,
|
| 6381 |
+
"learning_rate": 3.633852576659935e-07,
|
| 6382 |
+
"loss": 3.433,
|
| 6383 |
+
"step": 906
|
| 6384 |
+
},
|
| 6385 |
+
{
|
| 6386 |
+
"epoch": 0.9750067186240258,
|
| 6387 |
+
"grad_norm": 14.545281410217285,
|
| 6388 |
+
"learning_rate": 3.3491176697517667e-07,
|
| 6389 |
+
"loss": 2.89,
|
| 6390 |
+
"step": 907
|
| 6391 |
+
},
|
| 6392 |
+
{
|
| 6393 |
+
"epoch": 0.9760816984681537,
|
| 6394 |
+
"grad_norm": 10.29315185546875,
|
| 6395 |
+
"learning_rate": 3.0759791610335267e-07,
|
| 6396 |
+
"loss": 2.0145,
|
| 6397 |
+
"step": 908
|
| 6398 |
+
},
|
| 6399 |
+
{
|
| 6400 |
+
"epoch": 0.9771566783122816,
|
| 6401 |
+
"grad_norm": 21.322124481201172,
|
| 6402 |
+
"learning_rate": 2.81444022857158e-07,
|
| 6403 |
+
"loss": 3.5667,
|
| 6404 |
+
"step": 909
|
| 6405 |
+
},
|
| 6406 |
+
{
|
| 6407 |
+
"epoch": 0.9782316581564096,
|
| 6408 |
+
"grad_norm": 13.829008102416992,
|
| 6409 |
+
"learning_rate": 2.5645039154675864e-07,
|
| 6410 |
+
"loss": 2.6689,
|
| 6411 |
+
"step": 910
|
| 6412 |
+
},
|
| 6413 |
+
{
|
| 6414 |
+
"epoch": 0.9793066380005375,
|
| 6415 |
+
"grad_norm": 15.563042640686035,
|
| 6416 |
+
"learning_rate": 2.3261731298217514e-07,
|
| 6417 |
+
"loss": 2.4102,
|
| 6418 |
+
"step": 911
|
| 6419 |
+
},
|
| 6420 |
+
{
|
| 6421 |
+
"epoch": 0.9803816178446654,
|
| 6422 |
+
"grad_norm": 17.429529190063477,
|
| 6423 |
+
"learning_rate": 2.099450644700407e-07,
|
| 6424 |
+
"loss": 2.4791,
|
| 6425 |
+
"step": 912
|
| 6426 |
+
},
|
| 6427 |
+
{
|
| 6428 |
+
"epoch": 0.9814565976887933,
|
| 6429 |
+
"grad_norm": 15.122339248657227,
|
| 6430 |
+
"learning_rate": 1.8843390981024834e-07,
|
| 6431 |
+
"loss": 2.1212,
|
| 6432 |
+
"step": 913
|
| 6433 |
+
},
|
| 6434 |
+
{
|
| 6435 |
+
"epoch": 0.9825315775329213,
|
| 6436 |
+
"grad_norm": 17.688457489013672,
|
| 6437 |
+
"learning_rate": 1.6808409929298663e-07,
|
| 6438 |
+
"loss": 2.279,
|
| 6439 |
+
"step": 914
|
| 6440 |
+
},
|
| 6441 |
+
{
|
| 6442 |
+
"epoch": 0.9836065573770492,
|
| 6443 |
+
"grad_norm": 17.063894271850586,
|
| 6444 |
+
"learning_rate": 1.488958696957421e-07,
|
| 6445 |
+
"loss": 3.7765,
|
| 6446 |
+
"step": 915
|
| 6447 |
+
},
|
| 6448 |
+
{
|
| 6449 |
+
"epoch": 0.9846815372211771,
|
| 6450 |
+
"grad_norm": 10.769682884216309,
|
| 6451 |
+
"learning_rate": 1.3086944428060132e-07,
|
| 6452 |
+
"loss": 2.1393,
|
| 6453 |
+
"step": 916
|
| 6454 |
+
},
|
| 6455 |
+
{
|
| 6456 |
+
"epoch": 0.985756517065305,
|
| 6457 |
+
"grad_norm": 10.24843692779541,
|
| 6458 |
+
"learning_rate": 1.1400503279163088e-07,
|
| 6459 |
+
"loss": 2.2815,
|
| 6460 |
+
"step": 917
|
| 6461 |
+
},
|
| 6462 |
+
{
|
| 6463 |
+
"epoch": 0.986831496909433,
|
| 6464 |
+
"grad_norm": 13.591650009155273,
|
| 6465 |
+
"learning_rate": 9.83028314524348e-08,
|
| 6466 |
+
"loss": 2.809,
|
| 6467 |
+
"step": 918
|
| 6468 |
+
},
|
| 6469 |
+
{
|
| 6470 |
+
"epoch": 0.9879064767535609,
|
| 6471 |
+
"grad_norm": 13.300793647766113,
|
| 6472 |
+
"learning_rate": 8.376302296387861e-08,
|
| 6473 |
+
"loss": 2.9568,
|
| 6474 |
+
"step": 919
|
| 6475 |
+
},
|
| 6476 |
+
{
|
| 6477 |
+
"epoch": 0.9889814565976888,
|
| 6478 |
+
"grad_norm": 16.980257034301758,
|
| 6479 |
+
"learning_rate": 7.038577650195777e-08,
|
| 6480 |
+
"loss": 2.0696,
|
| 6481 |
+
"step": 920
|
| 6482 |
+
},
|
| 6483 |
+
{
|
| 6484 |
+
"epoch": 0.9900564364418167,
|
| 6485 |
+
"grad_norm": 17.34079360961914,
|
| 6486 |
+
"learning_rate": 5.8171247715854696e-08,
|
| 6487 |
+
"loss": 3.1045,
|
| 6488 |
+
"step": 921
|
| 6489 |
+
},
|
| 6490 |
+
{
|
| 6491 |
+
"epoch": 0.9911314162859446,
|
| 6492 |
+
"grad_norm": 18.76333999633789,
|
| 6493 |
+
"learning_rate": 4.711957872606254e-08,
|
| 6494 |
+
"loss": 2.967,
|
| 6495 |
+
"step": 922
|
| 6496 |
+
},
|
| 6497 |
+
{
|
| 6498 |
+
"epoch": 0.9922063961300726,
|
| 6499 |
+
"grad_norm": 15.434213638305664,
|
| 6500 |
+
"learning_rate": 3.7230898122808665e-08,
|
| 6501 |
+
"loss": 3.0553,
|
| 6502 |
+
"step": 923
|
| 6503 |
+
},
|
| 6504 |
+
{
|
| 6505 |
+
"epoch": 0.9932813759742005,
|
| 6506 |
+
"grad_norm": 20.012191772460938,
|
| 6507 |
+
"learning_rate": 2.850532096452252e-08,
|
| 6508 |
+
"loss": 3.6372,
|
| 6509 |
+
"step": 924
|
| 6510 |
+
},
|
| 6511 |
+
{
|
| 6512 |
+
"epoch": 0.9943563558183284,
|
| 6513 |
+
"grad_norm": 14.739432334899902,
|
| 6514 |
+
"learning_rate": 2.0942948776481175e-08,
|
| 6515 |
+
"loss": 1.9846,
|
| 6516 |
+
"step": 925
|
| 6517 |
+
},
|
| 6518 |
+
{
|
| 6519 |
+
"epoch": 0.9954313356624563,
|
| 6520 |
+
"grad_norm": 14.258848190307617,
|
| 6521 |
+
"learning_rate": 1.4543869549665801e-08,
|
| 6522 |
+
"loss": 2.9188,
|
| 6523 |
+
"step": 926
|
| 6524 |
+
},
|
| 6525 |
+
{
|
| 6526 |
+
"epoch": 0.9965063155065843,
|
| 6527 |
+
"grad_norm": 14.79078483581543,
|
| 6528 |
+
"learning_rate": 9.308157739706946e-09,
|
| 6529 |
+
"loss": 2.4791,
|
| 6530 |
+
"step": 927
|
| 6531 |
+
},
|
| 6532 |
+
{
|
| 6533 |
+
"epoch": 0.9975812953507122,
|
| 6534 |
+
"grad_norm": 17.466236114501953,
|
| 6535 |
+
"learning_rate": 5.23587426601857e-09,
|
| 6536 |
+
"loss": 2.1995,
|
| 6537 |
+
"step": 928
|
| 6538 |
+
},
|
| 6539 |
+
{
|
| 6540 |
+
"epoch": 0.9986562751948401,
|
| 6541 |
+
"grad_norm": 19.64586067199707,
|
| 6542 |
+
"learning_rate": 2.327066511120801e-09,
|
| 6543 |
+
"loss": 2.872,
|
| 6544 |
+
"step": 929
|
| 6545 |
+
},
|
| 6546 |
+
{
|
| 6547 |
+
"epoch": 0.999731255038968,
|
| 6548 |
+
"grad_norm": 12.937053680419922,
|
| 6549 |
+
"learning_rate": 5.817683200515233e-10,
|
| 6550 |
+
"loss": 2.6175,
|
| 6551 |
+
"step": 930
|
| 6552 |
+
},
|
| 6553 |
+
{
|
| 6554 |
+
"epoch": 1.000806234883096,
|
| 6555 |
+
"grad_norm": 16.165599822998047,
|
| 6556 |
+
"learning_rate": 0.0,
|
| 6557 |
+
"loss": 2.2581,
|
| 6558 |
+
"step": 931
|
| 6559 |
}
|
| 6560 |
],
|
| 6561 |
"logging_steps": 1,
|
|
|
|
| 6570 |
"should_evaluate": false,
|
| 6571 |
"should_log": false,
|
| 6572 |
"should_save": true,
|
| 6573 |
+
"should_training_stop": true
|
| 6574 |
},
|
| 6575 |
"attributes": {}
|
| 6576 |
}
|
| 6577 |
},
|
| 6578 |
+
"total_flos": 7022441851256832.0,
|
| 6579 |
"train_batch_size": 2,
|
| 6580 |
"trial_name": null,
|
| 6581 |
"trial_params": null
|