Training in progress, step 3500, checkpoint
Browse files
last-checkpoint/model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 91951912
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:63ddd3219ba3caedfb25ce1bb02834533571ccfa44e0512a21c1a9e2d4a16b9a
|
| 3 |
size 91951912
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 183991627
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1ec199331024fb04b3247269ed56280dbaffad395c00fc492bab6daa70a1f23b
|
| 3 |
size 183991627
|
last-checkpoint/rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14645
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f196323d7423b60f8e4ceb7dbf8715ee326c0d068e5ff164f13c63b279b9f1a0
|
| 3 |
size 14645
|
last-checkpoint/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1465
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8f6e0d1005db7bc4c64bb41807eb5eb456971ed8619880b4ac58f03171a3fb34
|
| 3 |
size 1465
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -2,9 +2,9 @@
|
|
| 2 |
"best_global_step": null,
|
| 3 |
"best_metric": null,
|
| 4 |
"best_model_checkpoint": null,
|
| 5 |
-
"epoch": 0.
|
| 6 |
"eval_steps": 500,
|
| 7 |
-
"global_step":
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
@@ -21056,6 +21056,3514 @@
|
|
| 21056 |
"eval_samples_per_second": 24.533,
|
| 21057 |
"eval_steps_per_second": 1.533,
|
| 21058 |
"step": 3000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 21059 |
}
|
| 21060 |
],
|
| 21061 |
"logging_steps": 1,
|
|
|
|
| 2 |
"best_global_step": null,
|
| 3 |
"best_metric": null,
|
| 4 |
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 0.00035,
|
| 6 |
"eval_steps": 500,
|
| 7 |
+
"global_step": 3500,
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
|
|
| 21056 |
"eval_samples_per_second": 24.533,
|
| 21057 |
"eval_steps_per_second": 1.533,
|
| 21058 |
"step": 3000
|
| 21059 |
+
},
|
| 21060 |
+
{
|
| 21061 |
+
"epoch": 0.0003001,
|
| 21062 |
+
"grad_norm": 4.194701194763184,
|
| 21063 |
+
"learning_rate": 3e-05,
|
| 21064 |
+
"loss": 13.9688,
|
| 21065 |
+
"step": 3001
|
| 21066 |
+
},
|
| 21067 |
+
{
|
| 21068 |
+
"epoch": 0.0003002,
|
| 21069 |
+
"grad_norm": 4.1234259605407715,
|
| 21070 |
+
"learning_rate": 3.001e-05,
|
| 21071 |
+
"loss": 13.2422,
|
| 21072 |
+
"step": 3002
|
| 21073 |
+
},
|
| 21074 |
+
{
|
| 21075 |
+
"epoch": 0.0003003,
|
| 21076 |
+
"grad_norm": 4.26988410949707,
|
| 21077 |
+
"learning_rate": 3.002e-05,
|
| 21078 |
+
"loss": 13.5859,
|
| 21079 |
+
"step": 3003
|
| 21080 |
+
},
|
| 21081 |
+
{
|
| 21082 |
+
"epoch": 0.0003004,
|
| 21083 |
+
"grad_norm": 4.214878082275391,
|
| 21084 |
+
"learning_rate": 3.003e-05,
|
| 21085 |
+
"loss": 13.6094,
|
| 21086 |
+
"step": 3004
|
| 21087 |
+
},
|
| 21088 |
+
{
|
| 21089 |
+
"epoch": 0.0003005,
|
| 21090 |
+
"grad_norm": 4.335902214050293,
|
| 21091 |
+
"learning_rate": 3.0040000000000004e-05,
|
| 21092 |
+
"loss": 13.7266,
|
| 21093 |
+
"step": 3005
|
| 21094 |
+
},
|
| 21095 |
+
{
|
| 21096 |
+
"epoch": 0.0003006,
|
| 21097 |
+
"grad_norm": 4.131463050842285,
|
| 21098 |
+
"learning_rate": 3.0050000000000002e-05,
|
| 21099 |
+
"loss": 14.0234,
|
| 21100 |
+
"step": 3006
|
| 21101 |
+
},
|
| 21102 |
+
{
|
| 21103 |
+
"epoch": 0.0003007,
|
| 21104 |
+
"grad_norm": 4.131214141845703,
|
| 21105 |
+
"learning_rate": 3.006e-05,
|
| 21106 |
+
"loss": 13.6484,
|
| 21107 |
+
"step": 3007
|
| 21108 |
+
},
|
| 21109 |
+
{
|
| 21110 |
+
"epoch": 0.0003008,
|
| 21111 |
+
"grad_norm": 4.261483669281006,
|
| 21112 |
+
"learning_rate": 3.0070000000000002e-05,
|
| 21113 |
+
"loss": 14.4219,
|
| 21114 |
+
"step": 3008
|
| 21115 |
+
},
|
| 21116 |
+
{
|
| 21117 |
+
"epoch": 0.0003009,
|
| 21118 |
+
"grad_norm": 4.358345031738281,
|
| 21119 |
+
"learning_rate": 3.008e-05,
|
| 21120 |
+
"loss": 13.8516,
|
| 21121 |
+
"step": 3009
|
| 21122 |
+
},
|
| 21123 |
+
{
|
| 21124 |
+
"epoch": 0.000301,
|
| 21125 |
+
"grad_norm": 4.224039077758789,
|
| 21126 |
+
"learning_rate": 3.009e-05,
|
| 21127 |
+
"loss": 14.2969,
|
| 21128 |
+
"step": 3010
|
| 21129 |
+
},
|
| 21130 |
+
{
|
| 21131 |
+
"epoch": 0.0003011,
|
| 21132 |
+
"grad_norm": 3.8865697383880615,
|
| 21133 |
+
"learning_rate": 3.01e-05,
|
| 21134 |
+
"loss": 13.7031,
|
| 21135 |
+
"step": 3011
|
| 21136 |
+
},
|
| 21137 |
+
{
|
| 21138 |
+
"epoch": 0.0003012,
|
| 21139 |
+
"grad_norm": 4.495300769805908,
|
| 21140 |
+
"learning_rate": 3.0109999999999998e-05,
|
| 21141 |
+
"loss": 14.375,
|
| 21142 |
+
"step": 3012
|
| 21143 |
+
},
|
| 21144 |
+
{
|
| 21145 |
+
"epoch": 0.0003013,
|
| 21146 |
+
"grad_norm": 4.00750732421875,
|
| 21147 |
+
"learning_rate": 3.0120000000000003e-05,
|
| 21148 |
+
"loss": 13.0625,
|
| 21149 |
+
"step": 3013
|
| 21150 |
+
},
|
| 21151 |
+
{
|
| 21152 |
+
"epoch": 0.0003014,
|
| 21153 |
+
"grad_norm": 4.163105487823486,
|
| 21154 |
+
"learning_rate": 3.0129999999999998e-05,
|
| 21155 |
+
"loss": 14.7656,
|
| 21156 |
+
"step": 3014
|
| 21157 |
+
},
|
| 21158 |
+
{
|
| 21159 |
+
"epoch": 0.0003015,
|
| 21160 |
+
"grad_norm": 4.139406204223633,
|
| 21161 |
+
"learning_rate": 3.0140000000000003e-05,
|
| 21162 |
+
"loss": 13.3359,
|
| 21163 |
+
"step": 3015
|
| 21164 |
+
},
|
| 21165 |
+
{
|
| 21166 |
+
"epoch": 0.0003016,
|
| 21167 |
+
"grad_norm": 4.277431011199951,
|
| 21168 |
+
"learning_rate": 3.015e-05,
|
| 21169 |
+
"loss": 13.5,
|
| 21170 |
+
"step": 3016
|
| 21171 |
+
},
|
| 21172 |
+
{
|
| 21173 |
+
"epoch": 0.0003017,
|
| 21174 |
+
"grad_norm": 4.73074197769165,
|
| 21175 |
+
"learning_rate": 3.016e-05,
|
| 21176 |
+
"loss": 13.8672,
|
| 21177 |
+
"step": 3017
|
| 21178 |
+
},
|
| 21179 |
+
{
|
| 21180 |
+
"epoch": 0.0003018,
|
| 21181 |
+
"grad_norm": 3.9893667697906494,
|
| 21182 |
+
"learning_rate": 3.017e-05,
|
| 21183 |
+
"loss": 13.5391,
|
| 21184 |
+
"step": 3018
|
| 21185 |
+
},
|
| 21186 |
+
{
|
| 21187 |
+
"epoch": 0.0003019,
|
| 21188 |
+
"grad_norm": 4.163100719451904,
|
| 21189 |
+
"learning_rate": 3.018e-05,
|
| 21190 |
+
"loss": 13.9609,
|
| 21191 |
+
"step": 3019
|
| 21192 |
+
},
|
| 21193 |
+
{
|
| 21194 |
+
"epoch": 0.000302,
|
| 21195 |
+
"grad_norm": 4.3695478439331055,
|
| 21196 |
+
"learning_rate": 3.019e-05,
|
| 21197 |
+
"loss": 13.7344,
|
| 21198 |
+
"step": 3020
|
| 21199 |
+
},
|
| 21200 |
+
{
|
| 21201 |
+
"epoch": 0.0003021,
|
| 21202 |
+
"grad_norm": 4.551717758178711,
|
| 21203 |
+
"learning_rate": 3.02e-05,
|
| 21204 |
+
"loss": 14.1797,
|
| 21205 |
+
"step": 3021
|
| 21206 |
+
},
|
| 21207 |
+
{
|
| 21208 |
+
"epoch": 0.0003022,
|
| 21209 |
+
"grad_norm": 3.9851553440093994,
|
| 21210 |
+
"learning_rate": 3.0210000000000004e-05,
|
| 21211 |
+
"loss": 12.9688,
|
| 21212 |
+
"step": 3022
|
| 21213 |
+
},
|
| 21214 |
+
{
|
| 21215 |
+
"epoch": 0.0003023,
|
| 21216 |
+
"grad_norm": 4.359283924102783,
|
| 21217 |
+
"learning_rate": 3.0220000000000002e-05,
|
| 21218 |
+
"loss": 14.0625,
|
| 21219 |
+
"step": 3023
|
| 21220 |
+
},
|
| 21221 |
+
{
|
| 21222 |
+
"epoch": 0.0003024,
|
| 21223 |
+
"grad_norm": 5.082442760467529,
|
| 21224 |
+
"learning_rate": 3.023e-05,
|
| 21225 |
+
"loss": 14.6484,
|
| 21226 |
+
"step": 3024
|
| 21227 |
+
},
|
| 21228 |
+
{
|
| 21229 |
+
"epoch": 0.0003025,
|
| 21230 |
+
"grad_norm": 4.216992378234863,
|
| 21231 |
+
"learning_rate": 3.0240000000000002e-05,
|
| 21232 |
+
"loss": 13.9453,
|
| 21233 |
+
"step": 3025
|
| 21234 |
+
},
|
| 21235 |
+
{
|
| 21236 |
+
"epoch": 0.0003026,
|
| 21237 |
+
"grad_norm": 4.368180751800537,
|
| 21238 |
+
"learning_rate": 3.025e-05,
|
| 21239 |
+
"loss": 14.3828,
|
| 21240 |
+
"step": 3026
|
| 21241 |
+
},
|
| 21242 |
+
{
|
| 21243 |
+
"epoch": 0.0003027,
|
| 21244 |
+
"grad_norm": 4.183067798614502,
|
| 21245 |
+
"learning_rate": 3.026e-05,
|
| 21246 |
+
"loss": 12.9141,
|
| 21247 |
+
"step": 3027
|
| 21248 |
+
},
|
| 21249 |
+
{
|
| 21250 |
+
"epoch": 0.0003028,
|
| 21251 |
+
"grad_norm": 4.025392532348633,
|
| 21252 |
+
"learning_rate": 3.027e-05,
|
| 21253 |
+
"loss": 14.0156,
|
| 21254 |
+
"step": 3028
|
| 21255 |
+
},
|
| 21256 |
+
{
|
| 21257 |
+
"epoch": 0.0003029,
|
| 21258 |
+
"grad_norm": 4.078479290008545,
|
| 21259 |
+
"learning_rate": 3.0279999999999998e-05,
|
| 21260 |
+
"loss": 13.4688,
|
| 21261 |
+
"step": 3029
|
| 21262 |
+
},
|
| 21263 |
+
{
|
| 21264 |
+
"epoch": 0.000303,
|
| 21265 |
+
"grad_norm": 4.384459972381592,
|
| 21266 |
+
"learning_rate": 3.0290000000000003e-05,
|
| 21267 |
+
"loss": 14.8984,
|
| 21268 |
+
"step": 3030
|
| 21269 |
+
},
|
| 21270 |
+
{
|
| 21271 |
+
"epoch": 0.0003031,
|
| 21272 |
+
"grad_norm": 4.016053199768066,
|
| 21273 |
+
"learning_rate": 3.0299999999999998e-05,
|
| 21274 |
+
"loss": 13.5391,
|
| 21275 |
+
"step": 3031
|
| 21276 |
+
},
|
| 21277 |
+
{
|
| 21278 |
+
"epoch": 0.0003032,
|
| 21279 |
+
"grad_norm": 4.216038227081299,
|
| 21280 |
+
"learning_rate": 3.0310000000000003e-05,
|
| 21281 |
+
"loss": 13.7422,
|
| 21282 |
+
"step": 3032
|
| 21283 |
+
},
|
| 21284 |
+
{
|
| 21285 |
+
"epoch": 0.0003033,
|
| 21286 |
+
"grad_norm": 4.121150493621826,
|
| 21287 |
+
"learning_rate": 3.032e-05,
|
| 21288 |
+
"loss": 13.5156,
|
| 21289 |
+
"step": 3033
|
| 21290 |
+
},
|
| 21291 |
+
{
|
| 21292 |
+
"epoch": 0.0003034,
|
| 21293 |
+
"grad_norm": 4.356866836547852,
|
| 21294 |
+
"learning_rate": 3.033e-05,
|
| 21295 |
+
"loss": 13.9062,
|
| 21296 |
+
"step": 3034
|
| 21297 |
+
},
|
| 21298 |
+
{
|
| 21299 |
+
"epoch": 0.0003035,
|
| 21300 |
+
"grad_norm": 4.39754581451416,
|
| 21301 |
+
"learning_rate": 3.034e-05,
|
| 21302 |
+
"loss": 13.7578,
|
| 21303 |
+
"step": 3035
|
| 21304 |
+
},
|
| 21305 |
+
{
|
| 21306 |
+
"epoch": 0.0003036,
|
| 21307 |
+
"grad_norm": 4.1412811279296875,
|
| 21308 |
+
"learning_rate": 3.035e-05,
|
| 21309 |
+
"loss": 12.9766,
|
| 21310 |
+
"step": 3036
|
| 21311 |
+
},
|
| 21312 |
+
{
|
| 21313 |
+
"epoch": 0.0003037,
|
| 21314 |
+
"grad_norm": 4.127572536468506,
|
| 21315 |
+
"learning_rate": 3.036e-05,
|
| 21316 |
+
"loss": 12.9609,
|
| 21317 |
+
"step": 3037
|
| 21318 |
+
},
|
| 21319 |
+
{
|
| 21320 |
+
"epoch": 0.0003038,
|
| 21321 |
+
"grad_norm": 4.17681884765625,
|
| 21322 |
+
"learning_rate": 3.037e-05,
|
| 21323 |
+
"loss": 13.8047,
|
| 21324 |
+
"step": 3038
|
| 21325 |
+
},
|
| 21326 |
+
{
|
| 21327 |
+
"epoch": 0.0003039,
|
| 21328 |
+
"grad_norm": 4.279952049255371,
|
| 21329 |
+
"learning_rate": 3.0380000000000004e-05,
|
| 21330 |
+
"loss": 13.7656,
|
| 21331 |
+
"step": 3039
|
| 21332 |
+
},
|
| 21333 |
+
{
|
| 21334 |
+
"epoch": 0.000304,
|
| 21335 |
+
"grad_norm": 4.562777996063232,
|
| 21336 |
+
"learning_rate": 3.0390000000000002e-05,
|
| 21337 |
+
"loss": 13.2734,
|
| 21338 |
+
"step": 3040
|
| 21339 |
+
},
|
| 21340 |
+
{
|
| 21341 |
+
"epoch": 0.0003041,
|
| 21342 |
+
"grad_norm": 4.5680999755859375,
|
| 21343 |
+
"learning_rate": 3.04e-05,
|
| 21344 |
+
"loss": 13.8828,
|
| 21345 |
+
"step": 3041
|
| 21346 |
+
},
|
| 21347 |
+
{
|
| 21348 |
+
"epoch": 0.0003042,
|
| 21349 |
+
"grad_norm": 4.2030439376831055,
|
| 21350 |
+
"learning_rate": 3.0410000000000002e-05,
|
| 21351 |
+
"loss": 13.8203,
|
| 21352 |
+
"step": 3042
|
| 21353 |
+
},
|
| 21354 |
+
{
|
| 21355 |
+
"epoch": 0.0003043,
|
| 21356 |
+
"grad_norm": 4.62878942489624,
|
| 21357 |
+
"learning_rate": 3.042e-05,
|
| 21358 |
+
"loss": 15.5625,
|
| 21359 |
+
"step": 3043
|
| 21360 |
+
},
|
| 21361 |
+
{
|
| 21362 |
+
"epoch": 0.0003044,
|
| 21363 |
+
"grad_norm": 4.339844226837158,
|
| 21364 |
+
"learning_rate": 3.0430000000000002e-05,
|
| 21365 |
+
"loss": 13.7109,
|
| 21366 |
+
"step": 3044
|
| 21367 |
+
},
|
| 21368 |
+
{
|
| 21369 |
+
"epoch": 0.0003045,
|
| 21370 |
+
"grad_norm": 4.207889080047607,
|
| 21371 |
+
"learning_rate": 3.044e-05,
|
| 21372 |
+
"loss": 14.1406,
|
| 21373 |
+
"step": 3045
|
| 21374 |
+
},
|
| 21375 |
+
{
|
| 21376 |
+
"epoch": 0.0003046,
|
| 21377 |
+
"grad_norm": 4.4827961921691895,
|
| 21378 |
+
"learning_rate": 3.0449999999999998e-05,
|
| 21379 |
+
"loss": 13.8359,
|
| 21380 |
+
"step": 3046
|
| 21381 |
+
},
|
| 21382 |
+
{
|
| 21383 |
+
"epoch": 0.0003047,
|
| 21384 |
+
"grad_norm": 4.142913341522217,
|
| 21385 |
+
"learning_rate": 3.0460000000000003e-05,
|
| 21386 |
+
"loss": 13.0703,
|
| 21387 |
+
"step": 3047
|
| 21388 |
+
},
|
| 21389 |
+
{
|
| 21390 |
+
"epoch": 0.0003048,
|
| 21391 |
+
"grad_norm": 3.8258612155914307,
|
| 21392 |
+
"learning_rate": 3.0469999999999998e-05,
|
| 21393 |
+
"loss": 12.3672,
|
| 21394 |
+
"step": 3048
|
| 21395 |
+
},
|
| 21396 |
+
{
|
| 21397 |
+
"epoch": 0.0003049,
|
| 21398 |
+
"grad_norm": 4.040428161621094,
|
| 21399 |
+
"learning_rate": 3.0480000000000003e-05,
|
| 21400 |
+
"loss": 12.3984,
|
| 21401 |
+
"step": 3049
|
| 21402 |
+
},
|
| 21403 |
+
{
|
| 21404 |
+
"epoch": 0.000305,
|
| 21405 |
+
"grad_norm": 4.279740810394287,
|
| 21406 |
+
"learning_rate": 3.049e-05,
|
| 21407 |
+
"loss": 13.8125,
|
| 21408 |
+
"step": 3050
|
| 21409 |
+
},
|
| 21410 |
+
{
|
| 21411 |
+
"epoch": 0.0003051,
|
| 21412 |
+
"grad_norm": 4.1018853187561035,
|
| 21413 |
+
"learning_rate": 3.05e-05,
|
| 21414 |
+
"loss": 12.8125,
|
| 21415 |
+
"step": 3051
|
| 21416 |
+
},
|
| 21417 |
+
{
|
| 21418 |
+
"epoch": 0.0003052,
|
| 21419 |
+
"grad_norm": 4.0386834144592285,
|
| 21420 |
+
"learning_rate": 3.051e-05,
|
| 21421 |
+
"loss": 12.5703,
|
| 21422 |
+
"step": 3052
|
| 21423 |
+
},
|
| 21424 |
+
{
|
| 21425 |
+
"epoch": 0.0003053,
|
| 21426 |
+
"grad_norm": 4.915343284606934,
|
| 21427 |
+
"learning_rate": 3.052e-05,
|
| 21428 |
+
"loss": 13.9766,
|
| 21429 |
+
"step": 3053
|
| 21430 |
+
},
|
| 21431 |
+
{
|
| 21432 |
+
"epoch": 0.0003054,
|
| 21433 |
+
"grad_norm": 4.776984214782715,
|
| 21434 |
+
"learning_rate": 3.053e-05,
|
| 21435 |
+
"loss": 14.0078,
|
| 21436 |
+
"step": 3054
|
| 21437 |
+
},
|
| 21438 |
+
{
|
| 21439 |
+
"epoch": 0.0003055,
|
| 21440 |
+
"grad_norm": 4.403343200683594,
|
| 21441 |
+
"learning_rate": 3.054e-05,
|
| 21442 |
+
"loss": 13.8594,
|
| 21443 |
+
"step": 3055
|
| 21444 |
+
},
|
| 21445 |
+
{
|
| 21446 |
+
"epoch": 0.0003056,
|
| 21447 |
+
"grad_norm": 4.060154914855957,
|
| 21448 |
+
"learning_rate": 3.0550000000000004e-05,
|
| 21449 |
+
"loss": 13.0781,
|
| 21450 |
+
"step": 3056
|
| 21451 |
+
},
|
| 21452 |
+
{
|
| 21453 |
+
"epoch": 0.0003057,
|
| 21454 |
+
"grad_norm": 4.458163261413574,
|
| 21455 |
+
"learning_rate": 3.056e-05,
|
| 21456 |
+
"loss": 13.9844,
|
| 21457 |
+
"step": 3057
|
| 21458 |
+
},
|
| 21459 |
+
{
|
| 21460 |
+
"epoch": 0.0003058,
|
| 21461 |
+
"grad_norm": 4.36746072769165,
|
| 21462 |
+
"learning_rate": 3.057e-05,
|
| 21463 |
+
"loss": 14.2422,
|
| 21464 |
+
"step": 3058
|
| 21465 |
+
},
|
| 21466 |
+
{
|
| 21467 |
+
"epoch": 0.0003059,
|
| 21468 |
+
"grad_norm": 4.294203758239746,
|
| 21469 |
+
"learning_rate": 3.058e-05,
|
| 21470 |
+
"loss": 13.6406,
|
| 21471 |
+
"step": 3059
|
| 21472 |
+
},
|
| 21473 |
+
{
|
| 21474 |
+
"epoch": 0.000306,
|
| 21475 |
+
"grad_norm": 4.018357276916504,
|
| 21476 |
+
"learning_rate": 3.059e-05,
|
| 21477 |
+
"loss": 13.2578,
|
| 21478 |
+
"step": 3060
|
| 21479 |
+
},
|
| 21480 |
+
{
|
| 21481 |
+
"epoch": 0.0003061,
|
| 21482 |
+
"grad_norm": 4.065389156341553,
|
| 21483 |
+
"learning_rate": 3.0600000000000005e-05,
|
| 21484 |
+
"loss": 13.3594,
|
| 21485 |
+
"step": 3061
|
| 21486 |
+
},
|
| 21487 |
+
{
|
| 21488 |
+
"epoch": 0.0003062,
|
| 21489 |
+
"grad_norm": 4.399044513702393,
|
| 21490 |
+
"learning_rate": 3.061e-05,
|
| 21491 |
+
"loss": 13.2109,
|
| 21492 |
+
"step": 3062
|
| 21493 |
+
},
|
| 21494 |
+
{
|
| 21495 |
+
"epoch": 0.0003063,
|
| 21496 |
+
"grad_norm": 4.423494815826416,
|
| 21497 |
+
"learning_rate": 3.062e-05,
|
| 21498 |
+
"loss": 13.0156,
|
| 21499 |
+
"step": 3063
|
| 21500 |
+
},
|
| 21501 |
+
{
|
| 21502 |
+
"epoch": 0.0003064,
|
| 21503 |
+
"grad_norm": 4.337902069091797,
|
| 21504 |
+
"learning_rate": 3.063e-05,
|
| 21505 |
+
"loss": 13.8281,
|
| 21506 |
+
"step": 3064
|
| 21507 |
+
},
|
| 21508 |
+
{
|
| 21509 |
+
"epoch": 0.0003065,
|
| 21510 |
+
"grad_norm": 4.550607681274414,
|
| 21511 |
+
"learning_rate": 3.064e-05,
|
| 21512 |
+
"loss": 13.8672,
|
| 21513 |
+
"step": 3065
|
| 21514 |
+
},
|
| 21515 |
+
{
|
| 21516 |
+
"epoch": 0.0003066,
|
| 21517 |
+
"grad_norm": 4.459248065948486,
|
| 21518 |
+
"learning_rate": 3.065e-05,
|
| 21519 |
+
"loss": 12.2578,
|
| 21520 |
+
"step": 3066
|
| 21521 |
+
},
|
| 21522 |
+
{
|
| 21523 |
+
"epoch": 0.0003067,
|
| 21524 |
+
"grad_norm": 4.459362983703613,
|
| 21525 |
+
"learning_rate": 3.066e-05,
|
| 21526 |
+
"loss": 13.7266,
|
| 21527 |
+
"step": 3067
|
| 21528 |
+
},
|
| 21529 |
+
{
|
| 21530 |
+
"epoch": 0.0003068,
|
| 21531 |
+
"grad_norm": 4.087619781494141,
|
| 21532 |
+
"learning_rate": 3.0669999999999996e-05,
|
| 21533 |
+
"loss": 13.1875,
|
| 21534 |
+
"step": 3068
|
| 21535 |
+
},
|
| 21536 |
+
{
|
| 21537 |
+
"epoch": 0.0003069,
|
| 21538 |
+
"grad_norm": 3.956347942352295,
|
| 21539 |
+
"learning_rate": 3.0680000000000004e-05,
|
| 21540 |
+
"loss": 12.6875,
|
| 21541 |
+
"step": 3069
|
| 21542 |
+
},
|
| 21543 |
+
{
|
| 21544 |
+
"epoch": 0.000307,
|
| 21545 |
+
"grad_norm": 3.932720184326172,
|
| 21546 |
+
"learning_rate": 3.069e-05,
|
| 21547 |
+
"loss": 12.0156,
|
| 21548 |
+
"step": 3070
|
| 21549 |
+
},
|
| 21550 |
+
{
|
| 21551 |
+
"epoch": 0.0003071,
|
| 21552 |
+
"grad_norm": 5.017730712890625,
|
| 21553 |
+
"learning_rate": 3.07e-05,
|
| 21554 |
+
"loss": 13.1641,
|
| 21555 |
+
"step": 3071
|
| 21556 |
+
},
|
| 21557 |
+
{
|
| 21558 |
+
"epoch": 0.0003072,
|
| 21559 |
+
"grad_norm": 4.067151069641113,
|
| 21560 |
+
"learning_rate": 3.071e-05,
|
| 21561 |
+
"loss": 12.5234,
|
| 21562 |
+
"step": 3072
|
| 21563 |
+
},
|
| 21564 |
+
{
|
| 21565 |
+
"epoch": 0.0003073,
|
| 21566 |
+
"grad_norm": 4.07366418838501,
|
| 21567 |
+
"learning_rate": 3.0720000000000004e-05,
|
| 21568 |
+
"loss": 13.0547,
|
| 21569 |
+
"step": 3073
|
| 21570 |
+
},
|
| 21571 |
+
{
|
| 21572 |
+
"epoch": 0.0003074,
|
| 21573 |
+
"grad_norm": 4.133744239807129,
|
| 21574 |
+
"learning_rate": 3.073e-05,
|
| 21575 |
+
"loss": 12.9453,
|
| 21576 |
+
"step": 3074
|
| 21577 |
+
},
|
| 21578 |
+
{
|
| 21579 |
+
"epoch": 0.0003075,
|
| 21580 |
+
"grad_norm": 4.064107894897461,
|
| 21581 |
+
"learning_rate": 3.074e-05,
|
| 21582 |
+
"loss": 12.8281,
|
| 21583 |
+
"step": 3075
|
| 21584 |
+
},
|
| 21585 |
+
{
|
| 21586 |
+
"epoch": 0.0003076,
|
| 21587 |
+
"grad_norm": 4.300269603729248,
|
| 21588 |
+
"learning_rate": 3.075e-05,
|
| 21589 |
+
"loss": 13.8203,
|
| 21590 |
+
"step": 3076
|
| 21591 |
+
},
|
| 21592 |
+
{
|
| 21593 |
+
"epoch": 0.0003077,
|
| 21594 |
+
"grad_norm": 4.334654331207275,
|
| 21595 |
+
"learning_rate": 3.076e-05,
|
| 21596 |
+
"loss": 13.5625,
|
| 21597 |
+
"step": 3077
|
| 21598 |
+
},
|
| 21599 |
+
{
|
| 21600 |
+
"epoch": 0.0003078,
|
| 21601 |
+
"grad_norm": 4.894828796386719,
|
| 21602 |
+
"learning_rate": 3.0770000000000005e-05,
|
| 21603 |
+
"loss": 14.5469,
|
| 21604 |
+
"step": 3078
|
| 21605 |
+
},
|
| 21606 |
+
{
|
| 21607 |
+
"epoch": 0.0003079,
|
| 21608 |
+
"grad_norm": 3.7898101806640625,
|
| 21609 |
+
"learning_rate": 3.078e-05,
|
| 21610 |
+
"loss": 12.1797,
|
| 21611 |
+
"step": 3079
|
| 21612 |
+
},
|
| 21613 |
+
{
|
| 21614 |
+
"epoch": 0.000308,
|
| 21615 |
+
"grad_norm": 3.9666337966918945,
|
| 21616 |
+
"learning_rate": 3.079e-05,
|
| 21617 |
+
"loss": 12.125,
|
| 21618 |
+
"step": 3080
|
| 21619 |
+
},
|
| 21620 |
+
{
|
| 21621 |
+
"epoch": 0.0003081,
|
| 21622 |
+
"grad_norm": 4.044882297515869,
|
| 21623 |
+
"learning_rate": 3.08e-05,
|
| 21624 |
+
"loss": 12.6406,
|
| 21625 |
+
"step": 3081
|
| 21626 |
+
},
|
| 21627 |
+
{
|
| 21628 |
+
"epoch": 0.0003082,
|
| 21629 |
+
"grad_norm": 3.975879192352295,
|
| 21630 |
+
"learning_rate": 3.081e-05,
|
| 21631 |
+
"loss": 12.4531,
|
| 21632 |
+
"step": 3082
|
| 21633 |
+
},
|
| 21634 |
+
{
|
| 21635 |
+
"epoch": 0.0003083,
|
| 21636 |
+
"grad_norm": 4.260823726654053,
|
| 21637 |
+
"learning_rate": 3.082e-05,
|
| 21638 |
+
"loss": 13.3672,
|
| 21639 |
+
"step": 3083
|
| 21640 |
+
},
|
| 21641 |
+
{
|
| 21642 |
+
"epoch": 0.0003084,
|
| 21643 |
+
"grad_norm": 4.446697235107422,
|
| 21644 |
+
"learning_rate": 3.083e-05,
|
| 21645 |
+
"loss": 12.7422,
|
| 21646 |
+
"step": 3084
|
| 21647 |
+
},
|
| 21648 |
+
{
|
| 21649 |
+
"epoch": 0.0003085,
|
| 21650 |
+
"grad_norm": 4.369131565093994,
|
| 21651 |
+
"learning_rate": 3.0839999999999996e-05,
|
| 21652 |
+
"loss": 13.3281,
|
| 21653 |
+
"step": 3085
|
| 21654 |
+
},
|
| 21655 |
+
{
|
| 21656 |
+
"epoch": 0.0003086,
|
| 21657 |
+
"grad_norm": 4.1950364112854,
|
| 21658 |
+
"learning_rate": 3.0850000000000004e-05,
|
| 21659 |
+
"loss": 13.2266,
|
| 21660 |
+
"step": 3086
|
| 21661 |
+
},
|
| 21662 |
+
{
|
| 21663 |
+
"epoch": 0.0003087,
|
| 21664 |
+
"grad_norm": 4.131595611572266,
|
| 21665 |
+
"learning_rate": 3.086e-05,
|
| 21666 |
+
"loss": 13.5,
|
| 21667 |
+
"step": 3087
|
| 21668 |
+
},
|
| 21669 |
+
{
|
| 21670 |
+
"epoch": 0.0003088,
|
| 21671 |
+
"grad_norm": 4.693575859069824,
|
| 21672 |
+
"learning_rate": 3.087e-05,
|
| 21673 |
+
"loss": 13.6953,
|
| 21674 |
+
"step": 3088
|
| 21675 |
+
},
|
| 21676 |
+
{
|
| 21677 |
+
"epoch": 0.0003089,
|
| 21678 |
+
"grad_norm": 4.081747531890869,
|
| 21679 |
+
"learning_rate": 3.088e-05,
|
| 21680 |
+
"loss": 12.4375,
|
| 21681 |
+
"step": 3089
|
| 21682 |
+
},
|
| 21683 |
+
{
|
| 21684 |
+
"epoch": 0.000309,
|
| 21685 |
+
"grad_norm": 4.189774513244629,
|
| 21686 |
+
"learning_rate": 3.0890000000000004e-05,
|
| 21687 |
+
"loss": 12.4453,
|
| 21688 |
+
"step": 3090
|
| 21689 |
+
},
|
| 21690 |
+
{
|
| 21691 |
+
"epoch": 0.0003091,
|
| 21692 |
+
"grad_norm": 4.040273666381836,
|
| 21693 |
+
"learning_rate": 3.09e-05,
|
| 21694 |
+
"loss": 11.8047,
|
| 21695 |
+
"step": 3091
|
| 21696 |
+
},
|
| 21697 |
+
{
|
| 21698 |
+
"epoch": 0.0003092,
|
| 21699 |
+
"grad_norm": 4.0818047523498535,
|
| 21700 |
+
"learning_rate": 3.091e-05,
|
| 21701 |
+
"loss": 12.4609,
|
| 21702 |
+
"step": 3092
|
| 21703 |
+
},
|
| 21704 |
+
{
|
| 21705 |
+
"epoch": 0.0003093,
|
| 21706 |
+
"grad_norm": 4.071326732635498,
|
| 21707 |
+
"learning_rate": 3.092e-05,
|
| 21708 |
+
"loss": 12.0547,
|
| 21709 |
+
"step": 3093
|
| 21710 |
+
},
|
| 21711 |
+
{
|
| 21712 |
+
"epoch": 0.0003094,
|
| 21713 |
+
"grad_norm": 3.987217426300049,
|
| 21714 |
+
"learning_rate": 3.093e-05,
|
| 21715 |
+
"loss": 11.9219,
|
| 21716 |
+
"step": 3094
|
| 21717 |
+
},
|
| 21718 |
+
{
|
| 21719 |
+
"epoch": 0.0003095,
|
| 21720 |
+
"grad_norm": 4.264516830444336,
|
| 21721 |
+
"learning_rate": 3.0940000000000005e-05,
|
| 21722 |
+
"loss": 14.3672,
|
| 21723 |
+
"step": 3095
|
| 21724 |
+
},
|
| 21725 |
+
{
|
| 21726 |
+
"epoch": 0.0003096,
|
| 21727 |
+
"grad_norm": 4.392807960510254,
|
| 21728 |
+
"learning_rate": 3.095e-05,
|
| 21729 |
+
"loss": 13.4609,
|
| 21730 |
+
"step": 3096
|
| 21731 |
+
},
|
| 21732 |
+
{
|
| 21733 |
+
"epoch": 0.0003097,
|
| 21734 |
+
"grad_norm": 4.537779808044434,
|
| 21735 |
+
"learning_rate": 3.096e-05,
|
| 21736 |
+
"loss": 13.3359,
|
| 21737 |
+
"step": 3097
|
| 21738 |
+
},
|
| 21739 |
+
{
|
| 21740 |
+
"epoch": 0.0003098,
|
| 21741 |
+
"grad_norm": 4.28918981552124,
|
| 21742 |
+
"learning_rate": 3.097e-05,
|
| 21743 |
+
"loss": 13.6328,
|
| 21744 |
+
"step": 3098
|
| 21745 |
+
},
|
| 21746 |
+
{
|
| 21747 |
+
"epoch": 0.0003099,
|
| 21748 |
+
"grad_norm": 4.60708475112915,
|
| 21749 |
+
"learning_rate": 3.098e-05,
|
| 21750 |
+
"loss": 13.2578,
|
| 21751 |
+
"step": 3099
|
| 21752 |
+
},
|
| 21753 |
+
{
|
| 21754 |
+
"epoch": 0.00031,
|
| 21755 |
+
"grad_norm": 4.293633937835693,
|
| 21756 |
+
"learning_rate": 3.099e-05,
|
| 21757 |
+
"loss": 12.5312,
|
| 21758 |
+
"step": 3100
|
| 21759 |
+
},
|
| 21760 |
+
{
|
| 21761 |
+
"epoch": 0.0003101,
|
| 21762 |
+
"grad_norm": 4.165204048156738,
|
| 21763 |
+
"learning_rate": 3.1e-05,
|
| 21764 |
+
"loss": 13.0,
|
| 21765 |
+
"step": 3101
|
| 21766 |
+
},
|
| 21767 |
+
{
|
| 21768 |
+
"epoch": 0.0003102,
|
| 21769 |
+
"grad_norm": 4.12003231048584,
|
| 21770 |
+
"learning_rate": 3.1009999999999996e-05,
|
| 21771 |
+
"loss": 12.5469,
|
| 21772 |
+
"step": 3102
|
| 21773 |
+
},
|
| 21774 |
+
{
|
| 21775 |
+
"epoch": 0.0003103,
|
| 21776 |
+
"grad_norm": 4.180938243865967,
|
| 21777 |
+
"learning_rate": 3.1020000000000005e-05,
|
| 21778 |
+
"loss": 12.8828,
|
| 21779 |
+
"step": 3103
|
| 21780 |
+
},
|
| 21781 |
+
{
|
| 21782 |
+
"epoch": 0.0003104,
|
| 21783 |
+
"grad_norm": 4.337403297424316,
|
| 21784 |
+
"learning_rate": 3.103e-05,
|
| 21785 |
+
"loss": 12.875,
|
| 21786 |
+
"step": 3104
|
| 21787 |
+
},
|
| 21788 |
+
{
|
| 21789 |
+
"epoch": 0.0003105,
|
| 21790 |
+
"grad_norm": 4.510208606719971,
|
| 21791 |
+
"learning_rate": 3.104e-05,
|
| 21792 |
+
"loss": 13.625,
|
| 21793 |
+
"step": 3105
|
| 21794 |
+
},
|
| 21795 |
+
{
|
| 21796 |
+
"epoch": 0.0003106,
|
| 21797 |
+
"grad_norm": 3.9945015907287598,
|
| 21798 |
+
"learning_rate": 3.105e-05,
|
| 21799 |
+
"loss": 12.2031,
|
| 21800 |
+
"step": 3106
|
| 21801 |
+
},
|
| 21802 |
+
{
|
| 21803 |
+
"epoch": 0.0003107,
|
| 21804 |
+
"grad_norm": 4.681159496307373,
|
| 21805 |
+
"learning_rate": 3.1060000000000004e-05,
|
| 21806 |
+
"loss": 13.5078,
|
| 21807 |
+
"step": 3107
|
| 21808 |
+
},
|
| 21809 |
+
{
|
| 21810 |
+
"epoch": 0.0003108,
|
| 21811 |
+
"grad_norm": 4.287213325500488,
|
| 21812 |
+
"learning_rate": 3.107e-05,
|
| 21813 |
+
"loss": 12.5391,
|
| 21814 |
+
"step": 3108
|
| 21815 |
+
},
|
| 21816 |
+
{
|
| 21817 |
+
"epoch": 0.0003109,
|
| 21818 |
+
"grad_norm": 4.070905685424805,
|
| 21819 |
+
"learning_rate": 3.108e-05,
|
| 21820 |
+
"loss": 12.6094,
|
| 21821 |
+
"step": 3109
|
| 21822 |
+
},
|
| 21823 |
+
{
|
| 21824 |
+
"epoch": 0.000311,
|
| 21825 |
+
"grad_norm": 4.0991926193237305,
|
| 21826 |
+
"learning_rate": 3.109e-05,
|
| 21827 |
+
"loss": 11.7734,
|
| 21828 |
+
"step": 3110
|
| 21829 |
+
},
|
| 21830 |
+
{
|
| 21831 |
+
"epoch": 0.0003111,
|
| 21832 |
+
"grad_norm": 3.793558120727539,
|
| 21833 |
+
"learning_rate": 3.11e-05,
|
| 21834 |
+
"loss": 11.7422,
|
| 21835 |
+
"step": 3111
|
| 21836 |
+
},
|
| 21837 |
+
{
|
| 21838 |
+
"epoch": 0.0003112,
|
| 21839 |
+
"grad_norm": 4.245021820068359,
|
| 21840 |
+
"learning_rate": 3.1110000000000005e-05,
|
| 21841 |
+
"loss": 13.0625,
|
| 21842 |
+
"step": 3112
|
| 21843 |
+
},
|
| 21844 |
+
{
|
| 21845 |
+
"epoch": 0.0003113,
|
| 21846 |
+
"grad_norm": 5.156881809234619,
|
| 21847 |
+
"learning_rate": 3.112e-05,
|
| 21848 |
+
"loss": 12.9297,
|
| 21849 |
+
"step": 3113
|
| 21850 |
+
},
|
| 21851 |
+
{
|
| 21852 |
+
"epoch": 0.0003114,
|
| 21853 |
+
"grad_norm": 4.007889747619629,
|
| 21854 |
+
"learning_rate": 3.113e-05,
|
| 21855 |
+
"loss": 11.8281,
|
| 21856 |
+
"step": 3114
|
| 21857 |
+
},
|
| 21858 |
+
{
|
| 21859 |
+
"epoch": 0.0003115,
|
| 21860 |
+
"grad_norm": 4.115943908691406,
|
| 21861 |
+
"learning_rate": 3.1140000000000003e-05,
|
| 21862 |
+
"loss": 12.2891,
|
| 21863 |
+
"step": 3115
|
| 21864 |
+
},
|
| 21865 |
+
{
|
| 21866 |
+
"epoch": 0.0003116,
|
| 21867 |
+
"grad_norm": 4.325364589691162,
|
| 21868 |
+
"learning_rate": 3.115e-05,
|
| 21869 |
+
"loss": 13.0156,
|
| 21870 |
+
"step": 3116
|
| 21871 |
+
},
|
| 21872 |
+
{
|
| 21873 |
+
"epoch": 0.0003117,
|
| 21874 |
+
"grad_norm": 4.555175304412842,
|
| 21875 |
+
"learning_rate": 3.116e-05,
|
| 21876 |
+
"loss": 13.9141,
|
| 21877 |
+
"step": 3117
|
| 21878 |
+
},
|
| 21879 |
+
{
|
| 21880 |
+
"epoch": 0.0003118,
|
| 21881 |
+
"grad_norm": 4.521339416503906,
|
| 21882 |
+
"learning_rate": 3.117e-05,
|
| 21883 |
+
"loss": 12.9531,
|
| 21884 |
+
"step": 3118
|
| 21885 |
+
},
|
| 21886 |
+
{
|
| 21887 |
+
"epoch": 0.0003119,
|
| 21888 |
+
"grad_norm": 4.389863967895508,
|
| 21889 |
+
"learning_rate": 3.1179999999999996e-05,
|
| 21890 |
+
"loss": 13.6094,
|
| 21891 |
+
"step": 3119
|
| 21892 |
+
},
|
| 21893 |
+
{
|
| 21894 |
+
"epoch": 0.000312,
|
| 21895 |
+
"grad_norm": 4.105045795440674,
|
| 21896 |
+
"learning_rate": 3.1190000000000005e-05,
|
| 21897 |
+
"loss": 13.3906,
|
| 21898 |
+
"step": 3120
|
| 21899 |
+
},
|
| 21900 |
+
{
|
| 21901 |
+
"epoch": 0.0003121,
|
| 21902 |
+
"grad_norm": 4.395541667938232,
|
| 21903 |
+
"learning_rate": 3.12e-05,
|
| 21904 |
+
"loss": 12.8203,
|
| 21905 |
+
"step": 3121
|
| 21906 |
+
},
|
| 21907 |
+
{
|
| 21908 |
+
"epoch": 0.0003122,
|
| 21909 |
+
"grad_norm": 4.227506637573242,
|
| 21910 |
+
"learning_rate": 3.121e-05,
|
| 21911 |
+
"loss": 13.3828,
|
| 21912 |
+
"step": 3122
|
| 21913 |
+
},
|
| 21914 |
+
{
|
| 21915 |
+
"epoch": 0.0003123,
|
| 21916 |
+
"grad_norm": 4.824154853820801,
|
| 21917 |
+
"learning_rate": 3.122e-05,
|
| 21918 |
+
"loss": 14.5469,
|
| 21919 |
+
"step": 3123
|
| 21920 |
+
},
|
| 21921 |
+
{
|
| 21922 |
+
"epoch": 0.0003124,
|
| 21923 |
+
"grad_norm": 4.500116348266602,
|
| 21924 |
+
"learning_rate": 3.1230000000000004e-05,
|
| 21925 |
+
"loss": 13.6484,
|
| 21926 |
+
"step": 3124
|
| 21927 |
+
},
|
| 21928 |
+
{
|
| 21929 |
+
"epoch": 0.0003125,
|
| 21930 |
+
"grad_norm": 4.317654609680176,
|
| 21931 |
+
"learning_rate": 3.124e-05,
|
| 21932 |
+
"loss": 13.0781,
|
| 21933 |
+
"step": 3125
|
| 21934 |
+
},
|
| 21935 |
+
{
|
| 21936 |
+
"epoch": 0.0003126,
|
| 21937 |
+
"grad_norm": 3.8973302841186523,
|
| 21938 |
+
"learning_rate": 3.125e-05,
|
| 21939 |
+
"loss": 12.1172,
|
| 21940 |
+
"step": 3126
|
| 21941 |
+
},
|
| 21942 |
+
{
|
| 21943 |
+
"epoch": 0.0003127,
|
| 21944 |
+
"grad_norm": 4.212470531463623,
|
| 21945 |
+
"learning_rate": 3.126e-05,
|
| 21946 |
+
"loss": 12.4375,
|
| 21947 |
+
"step": 3127
|
| 21948 |
+
},
|
| 21949 |
+
{
|
| 21950 |
+
"epoch": 0.0003128,
|
| 21951 |
+
"grad_norm": 4.022023677825928,
|
| 21952 |
+
"learning_rate": 3.127e-05,
|
| 21953 |
+
"loss": 12.2344,
|
| 21954 |
+
"step": 3128
|
| 21955 |
+
},
|
| 21956 |
+
{
|
| 21957 |
+
"epoch": 0.0003129,
|
| 21958 |
+
"grad_norm": 4.173778057098389,
|
| 21959 |
+
"learning_rate": 3.1280000000000005e-05,
|
| 21960 |
+
"loss": 12.6328,
|
| 21961 |
+
"step": 3129
|
| 21962 |
+
},
|
| 21963 |
+
{
|
| 21964 |
+
"epoch": 0.000313,
|
| 21965 |
+
"grad_norm": 3.9178037643432617,
|
| 21966 |
+
"learning_rate": 3.129e-05,
|
| 21967 |
+
"loss": 11.5859,
|
| 21968 |
+
"step": 3130
|
| 21969 |
+
},
|
| 21970 |
+
{
|
| 21971 |
+
"epoch": 0.0003131,
|
| 21972 |
+
"grad_norm": 4.367602825164795,
|
| 21973 |
+
"learning_rate": 3.13e-05,
|
| 21974 |
+
"loss": 13.0469,
|
| 21975 |
+
"step": 3131
|
| 21976 |
+
},
|
| 21977 |
+
{
|
| 21978 |
+
"epoch": 0.0003132,
|
| 21979 |
+
"grad_norm": 4.052750587463379,
|
| 21980 |
+
"learning_rate": 3.1310000000000003e-05,
|
| 21981 |
+
"loss": 11.6016,
|
| 21982 |
+
"step": 3132
|
| 21983 |
+
},
|
| 21984 |
+
{
|
| 21985 |
+
"epoch": 0.0003133,
|
| 21986 |
+
"grad_norm": 3.8811826705932617,
|
| 21987 |
+
"learning_rate": 3.132e-05,
|
| 21988 |
+
"loss": 11.6562,
|
| 21989 |
+
"step": 3133
|
| 21990 |
+
},
|
| 21991 |
+
{
|
| 21992 |
+
"epoch": 0.0003134,
|
| 21993 |
+
"grad_norm": 4.070820331573486,
|
| 21994 |
+
"learning_rate": 3.133e-05,
|
| 21995 |
+
"loss": 11.5,
|
| 21996 |
+
"step": 3134
|
| 21997 |
+
},
|
| 21998 |
+
{
|
| 21999 |
+
"epoch": 0.0003135,
|
| 22000 |
+
"grad_norm": 4.166890621185303,
|
| 22001 |
+
"learning_rate": 3.134e-05,
|
| 22002 |
+
"loss": 11.9297,
|
| 22003 |
+
"step": 3135
|
| 22004 |
+
},
|
| 22005 |
+
{
|
| 22006 |
+
"epoch": 0.0003136,
|
| 22007 |
+
"grad_norm": 5.06160306930542,
|
| 22008 |
+
"learning_rate": 3.1349999999999996e-05,
|
| 22009 |
+
"loss": 13.4609,
|
| 22010 |
+
"step": 3136
|
| 22011 |
+
},
|
| 22012 |
+
{
|
| 22013 |
+
"epoch": 0.0003137,
|
| 22014 |
+
"grad_norm": 4.107486724853516,
|
| 22015 |
+
"learning_rate": 3.1360000000000005e-05,
|
| 22016 |
+
"loss": 12.1719,
|
| 22017 |
+
"step": 3137
|
| 22018 |
+
},
|
| 22019 |
+
{
|
| 22020 |
+
"epoch": 0.0003138,
|
| 22021 |
+
"grad_norm": 4.159612655639648,
|
| 22022 |
+
"learning_rate": 3.137e-05,
|
| 22023 |
+
"loss": 12.9219,
|
| 22024 |
+
"step": 3138
|
| 22025 |
+
},
|
| 22026 |
+
{
|
| 22027 |
+
"epoch": 0.0003139,
|
| 22028 |
+
"grad_norm": 4.197668552398682,
|
| 22029 |
+
"learning_rate": 3.138e-05,
|
| 22030 |
+
"loss": 12.625,
|
| 22031 |
+
"step": 3139
|
| 22032 |
+
},
|
| 22033 |
+
{
|
| 22034 |
+
"epoch": 0.000314,
|
| 22035 |
+
"grad_norm": 4.047123908996582,
|
| 22036 |
+
"learning_rate": 3.139e-05,
|
| 22037 |
+
"loss": 12.1406,
|
| 22038 |
+
"step": 3140
|
| 22039 |
+
},
|
| 22040 |
+
{
|
| 22041 |
+
"epoch": 0.0003141,
|
| 22042 |
+
"grad_norm": 4.3158860206604,
|
| 22043 |
+
"learning_rate": 3.1400000000000004e-05,
|
| 22044 |
+
"loss": 12.1016,
|
| 22045 |
+
"step": 3141
|
| 22046 |
+
},
|
| 22047 |
+
{
|
| 22048 |
+
"epoch": 0.0003142,
|
| 22049 |
+
"grad_norm": 3.9308483600616455,
|
| 22050 |
+
"learning_rate": 3.141e-05,
|
| 22051 |
+
"loss": 12.2188,
|
| 22052 |
+
"step": 3142
|
| 22053 |
+
},
|
| 22054 |
+
{
|
| 22055 |
+
"epoch": 0.0003143,
|
| 22056 |
+
"grad_norm": 4.07874059677124,
|
| 22057 |
+
"learning_rate": 3.142e-05,
|
| 22058 |
+
"loss": 12.0703,
|
| 22059 |
+
"step": 3143
|
| 22060 |
+
},
|
| 22061 |
+
{
|
| 22062 |
+
"epoch": 0.0003144,
|
| 22063 |
+
"grad_norm": 4.216789722442627,
|
| 22064 |
+
"learning_rate": 3.143e-05,
|
| 22065 |
+
"loss": 12.9844,
|
| 22066 |
+
"step": 3144
|
| 22067 |
+
},
|
| 22068 |
+
{
|
| 22069 |
+
"epoch": 0.0003145,
|
| 22070 |
+
"grad_norm": 4.4545512199401855,
|
| 22071 |
+
"learning_rate": 3.144e-05,
|
| 22072 |
+
"loss": 12.8359,
|
| 22073 |
+
"step": 3145
|
| 22074 |
+
},
|
| 22075 |
+
{
|
| 22076 |
+
"epoch": 0.0003146,
|
| 22077 |
+
"grad_norm": 3.9575791358947754,
|
| 22078 |
+
"learning_rate": 3.1450000000000006e-05,
|
| 22079 |
+
"loss": 11.7266,
|
| 22080 |
+
"step": 3146
|
| 22081 |
+
},
|
| 22082 |
+
{
|
| 22083 |
+
"epoch": 0.0003147,
|
| 22084 |
+
"grad_norm": 4.6207380294799805,
|
| 22085 |
+
"learning_rate": 3.146e-05,
|
| 22086 |
+
"loss": 13.1094,
|
| 22087 |
+
"step": 3147
|
| 22088 |
+
},
|
| 22089 |
+
{
|
| 22090 |
+
"epoch": 0.0003148,
|
| 22091 |
+
"grad_norm": 4.516714096069336,
|
| 22092 |
+
"learning_rate": 3.147e-05,
|
| 22093 |
+
"loss": 14.0156,
|
| 22094 |
+
"step": 3148
|
| 22095 |
+
},
|
| 22096 |
+
{
|
| 22097 |
+
"epoch": 0.0003149,
|
| 22098 |
+
"grad_norm": 4.093890190124512,
|
| 22099 |
+
"learning_rate": 3.1480000000000004e-05,
|
| 22100 |
+
"loss": 11.6406,
|
| 22101 |
+
"step": 3149
|
| 22102 |
+
},
|
| 22103 |
+
{
|
| 22104 |
+
"epoch": 0.000315,
|
| 22105 |
+
"grad_norm": 4.16150426864624,
|
| 22106 |
+
"learning_rate": 3.149e-05,
|
| 22107 |
+
"loss": 11.8594,
|
| 22108 |
+
"step": 3150
|
| 22109 |
+
},
|
| 22110 |
+
{
|
| 22111 |
+
"epoch": 0.0003151,
|
| 22112 |
+
"grad_norm": 4.188538074493408,
|
| 22113 |
+
"learning_rate": 3.15e-05,
|
| 22114 |
+
"loss": 12.3906,
|
| 22115 |
+
"step": 3151
|
| 22116 |
+
},
|
| 22117 |
+
{
|
| 22118 |
+
"epoch": 0.0003152,
|
| 22119 |
+
"grad_norm": 4.296939849853516,
|
| 22120 |
+
"learning_rate": 3.151e-05,
|
| 22121 |
+
"loss": 13.1172,
|
| 22122 |
+
"step": 3152
|
| 22123 |
+
},
|
| 22124 |
+
{
|
| 22125 |
+
"epoch": 0.0003153,
|
| 22126 |
+
"grad_norm": 4.2411417961120605,
|
| 22127 |
+
"learning_rate": 3.1519999999999996e-05,
|
| 22128 |
+
"loss": 12.0859,
|
| 22129 |
+
"step": 3153
|
| 22130 |
+
},
|
| 22131 |
+
{
|
| 22132 |
+
"epoch": 0.0003154,
|
| 22133 |
+
"grad_norm": 4.440285682678223,
|
| 22134 |
+
"learning_rate": 3.1530000000000005e-05,
|
| 22135 |
+
"loss": 13.3047,
|
| 22136 |
+
"step": 3154
|
| 22137 |
+
},
|
| 22138 |
+
{
|
| 22139 |
+
"epoch": 0.0003155,
|
| 22140 |
+
"grad_norm": 4.174127578735352,
|
| 22141 |
+
"learning_rate": 3.154e-05,
|
| 22142 |
+
"loss": 12.1875,
|
| 22143 |
+
"step": 3155
|
| 22144 |
+
},
|
| 22145 |
+
{
|
| 22146 |
+
"epoch": 0.0003156,
|
| 22147 |
+
"grad_norm": 4.034418106079102,
|
| 22148 |
+
"learning_rate": 3.155e-05,
|
| 22149 |
+
"loss": 12.3594,
|
| 22150 |
+
"step": 3156
|
| 22151 |
+
},
|
| 22152 |
+
{
|
| 22153 |
+
"epoch": 0.0003157,
|
| 22154 |
+
"grad_norm": 4.488844871520996,
|
| 22155 |
+
"learning_rate": 3.156e-05,
|
| 22156 |
+
"loss": 12.8984,
|
| 22157 |
+
"step": 3157
|
| 22158 |
+
},
|
| 22159 |
+
{
|
| 22160 |
+
"epoch": 0.0003158,
|
| 22161 |
+
"grad_norm": 4.429039478302002,
|
| 22162 |
+
"learning_rate": 3.1570000000000004e-05,
|
| 22163 |
+
"loss": 12.5859,
|
| 22164 |
+
"step": 3158
|
| 22165 |
+
},
|
| 22166 |
+
{
|
| 22167 |
+
"epoch": 0.0003159,
|
| 22168 |
+
"grad_norm": 4.253639221191406,
|
| 22169 |
+
"learning_rate": 3.158e-05,
|
| 22170 |
+
"loss": 12.8984,
|
| 22171 |
+
"step": 3159
|
| 22172 |
+
},
|
| 22173 |
+
{
|
| 22174 |
+
"epoch": 0.000316,
|
| 22175 |
+
"grad_norm": 4.056889533996582,
|
| 22176 |
+
"learning_rate": 3.159e-05,
|
| 22177 |
+
"loss": 11.7266,
|
| 22178 |
+
"step": 3160
|
| 22179 |
+
},
|
| 22180 |
+
{
|
| 22181 |
+
"epoch": 0.0003161,
|
| 22182 |
+
"grad_norm": 4.251730918884277,
|
| 22183 |
+
"learning_rate": 3.16e-05,
|
| 22184 |
+
"loss": 12.3672,
|
| 22185 |
+
"step": 3161
|
| 22186 |
+
},
|
| 22187 |
+
{
|
| 22188 |
+
"epoch": 0.0003162,
|
| 22189 |
+
"grad_norm": 4.355313777923584,
|
| 22190 |
+
"learning_rate": 3.161e-05,
|
| 22191 |
+
"loss": 12.4531,
|
| 22192 |
+
"step": 3162
|
| 22193 |
+
},
|
| 22194 |
+
{
|
| 22195 |
+
"epoch": 0.0003163,
|
| 22196 |
+
"grad_norm": 4.073619842529297,
|
| 22197 |
+
"learning_rate": 3.1620000000000006e-05,
|
| 22198 |
+
"loss": 12.2344,
|
| 22199 |
+
"step": 3163
|
| 22200 |
+
},
|
| 22201 |
+
{
|
| 22202 |
+
"epoch": 0.0003164,
|
| 22203 |
+
"grad_norm": 3.9847664833068848,
|
| 22204 |
+
"learning_rate": 3.163e-05,
|
| 22205 |
+
"loss": 12.0469,
|
| 22206 |
+
"step": 3164
|
| 22207 |
+
},
|
| 22208 |
+
{
|
| 22209 |
+
"epoch": 0.0003165,
|
| 22210 |
+
"grad_norm": 4.491983413696289,
|
| 22211 |
+
"learning_rate": 3.164e-05,
|
| 22212 |
+
"loss": 12.6328,
|
| 22213 |
+
"step": 3165
|
| 22214 |
+
},
|
| 22215 |
+
{
|
| 22216 |
+
"epoch": 0.0003166,
|
| 22217 |
+
"grad_norm": 4.011117458343506,
|
| 22218 |
+
"learning_rate": 3.1650000000000004e-05,
|
| 22219 |
+
"loss": 11.8047,
|
| 22220 |
+
"step": 3166
|
| 22221 |
+
},
|
| 22222 |
+
{
|
| 22223 |
+
"epoch": 0.0003167,
|
| 22224 |
+
"grad_norm": 4.118056297302246,
|
| 22225 |
+
"learning_rate": 3.166e-05,
|
| 22226 |
+
"loss": 12.7812,
|
| 22227 |
+
"step": 3167
|
| 22228 |
+
},
|
| 22229 |
+
{
|
| 22230 |
+
"epoch": 0.0003168,
|
| 22231 |
+
"grad_norm": 3.8982393741607666,
|
| 22232 |
+
"learning_rate": 3.167e-05,
|
| 22233 |
+
"loss": 12.0,
|
| 22234 |
+
"step": 3168
|
| 22235 |
+
},
|
| 22236 |
+
{
|
| 22237 |
+
"epoch": 0.0003169,
|
| 22238 |
+
"grad_norm": 4.089506149291992,
|
| 22239 |
+
"learning_rate": 3.168e-05,
|
| 22240 |
+
"loss": 11.6484,
|
| 22241 |
+
"step": 3169
|
| 22242 |
+
},
|
| 22243 |
+
{
|
| 22244 |
+
"epoch": 0.000317,
|
| 22245 |
+
"grad_norm": 3.823589563369751,
|
| 22246 |
+
"learning_rate": 3.1689999999999996e-05,
|
| 22247 |
+
"loss": 11.6172,
|
| 22248 |
+
"step": 3170
|
| 22249 |
+
},
|
| 22250 |
+
{
|
| 22251 |
+
"epoch": 0.0003171,
|
| 22252 |
+
"grad_norm": 4.54476261138916,
|
| 22253 |
+
"learning_rate": 3.17e-05,
|
| 22254 |
+
"loss": 12.0312,
|
| 22255 |
+
"step": 3171
|
| 22256 |
+
},
|
| 22257 |
+
{
|
| 22258 |
+
"epoch": 0.0003172,
|
| 22259 |
+
"grad_norm": 3.9650237560272217,
|
| 22260 |
+
"learning_rate": 3.171e-05,
|
| 22261 |
+
"loss": 11.4609,
|
| 22262 |
+
"step": 3172
|
| 22263 |
+
},
|
| 22264 |
+
{
|
| 22265 |
+
"epoch": 0.0003173,
|
| 22266 |
+
"grad_norm": 3.9006104469299316,
|
| 22267 |
+
"learning_rate": 3.172e-05,
|
| 22268 |
+
"loss": 11.1328,
|
| 22269 |
+
"step": 3173
|
| 22270 |
+
},
|
| 22271 |
+
{
|
| 22272 |
+
"epoch": 0.0003174,
|
| 22273 |
+
"grad_norm": 4.44875431060791,
|
| 22274 |
+
"learning_rate": 3.173e-05,
|
| 22275 |
+
"loss": 12.3359,
|
| 22276 |
+
"step": 3174
|
| 22277 |
+
},
|
| 22278 |
+
{
|
| 22279 |
+
"epoch": 0.0003175,
|
| 22280 |
+
"grad_norm": 4.258243083953857,
|
| 22281 |
+
"learning_rate": 3.1740000000000004e-05,
|
| 22282 |
+
"loss": 11.4609,
|
| 22283 |
+
"step": 3175
|
| 22284 |
+
},
|
| 22285 |
+
{
|
| 22286 |
+
"epoch": 0.0003176,
|
| 22287 |
+
"grad_norm": 5.055328369140625,
|
| 22288 |
+
"learning_rate": 3.175e-05,
|
| 22289 |
+
"loss": 13.3125,
|
| 22290 |
+
"step": 3176
|
| 22291 |
+
},
|
| 22292 |
+
{
|
| 22293 |
+
"epoch": 0.0003177,
|
| 22294 |
+
"grad_norm": 4.481571197509766,
|
| 22295 |
+
"learning_rate": 3.176e-05,
|
| 22296 |
+
"loss": 12.8672,
|
| 22297 |
+
"step": 3177
|
| 22298 |
+
},
|
| 22299 |
+
{
|
| 22300 |
+
"epoch": 0.0003178,
|
| 22301 |
+
"grad_norm": 4.156774044036865,
|
| 22302 |
+
"learning_rate": 3.177e-05,
|
| 22303 |
+
"loss": 11.6328,
|
| 22304 |
+
"step": 3178
|
| 22305 |
+
},
|
| 22306 |
+
{
|
| 22307 |
+
"epoch": 0.0003179,
|
| 22308 |
+
"grad_norm": 3.8571271896362305,
|
| 22309 |
+
"learning_rate": 3.178e-05,
|
| 22310 |
+
"loss": 11.8516,
|
| 22311 |
+
"step": 3179
|
| 22312 |
+
},
|
| 22313 |
+
{
|
| 22314 |
+
"epoch": 0.000318,
|
| 22315 |
+
"grad_norm": 3.8231124877929688,
|
| 22316 |
+
"learning_rate": 3.1790000000000006e-05,
|
| 22317 |
+
"loss": 11.3594,
|
| 22318 |
+
"step": 3180
|
| 22319 |
+
},
|
| 22320 |
+
{
|
| 22321 |
+
"epoch": 0.0003181,
|
| 22322 |
+
"grad_norm": 3.8678410053253174,
|
| 22323 |
+
"learning_rate": 3.18e-05,
|
| 22324 |
+
"loss": 11.3672,
|
| 22325 |
+
"step": 3181
|
| 22326 |
+
},
|
| 22327 |
+
{
|
| 22328 |
+
"epoch": 0.0003182,
|
| 22329 |
+
"grad_norm": 3.991071939468384,
|
| 22330 |
+
"learning_rate": 3.1809999999999995e-05,
|
| 22331 |
+
"loss": 11.7266,
|
| 22332 |
+
"step": 3182
|
| 22333 |
+
},
|
| 22334 |
+
{
|
| 22335 |
+
"epoch": 0.0003183,
|
| 22336 |
+
"grad_norm": 4.208522796630859,
|
| 22337 |
+
"learning_rate": 3.1820000000000004e-05,
|
| 22338 |
+
"loss": 12.8438,
|
| 22339 |
+
"step": 3183
|
| 22340 |
+
},
|
| 22341 |
+
{
|
| 22342 |
+
"epoch": 0.0003184,
|
| 22343 |
+
"grad_norm": 4.157339572906494,
|
| 22344 |
+
"learning_rate": 3.183e-05,
|
| 22345 |
+
"loss": 11.5469,
|
| 22346 |
+
"step": 3184
|
| 22347 |
+
},
|
| 22348 |
+
{
|
| 22349 |
+
"epoch": 0.0003185,
|
| 22350 |
+
"grad_norm": 4.158442974090576,
|
| 22351 |
+
"learning_rate": 3.184e-05,
|
| 22352 |
+
"loss": 11.9219,
|
| 22353 |
+
"step": 3185
|
| 22354 |
+
},
|
| 22355 |
+
{
|
| 22356 |
+
"epoch": 0.0003186,
|
| 22357 |
+
"grad_norm": 3.9469897747039795,
|
| 22358 |
+
"learning_rate": 3.185e-05,
|
| 22359 |
+
"loss": 11.3906,
|
| 22360 |
+
"step": 3186
|
| 22361 |
+
},
|
| 22362 |
+
{
|
| 22363 |
+
"epoch": 0.0003187,
|
| 22364 |
+
"grad_norm": 4.372818946838379,
|
| 22365 |
+
"learning_rate": 3.1859999999999997e-05,
|
| 22366 |
+
"loss": 12.2578,
|
| 22367 |
+
"step": 3187
|
| 22368 |
+
},
|
| 22369 |
+
{
|
| 22370 |
+
"epoch": 0.0003188,
|
| 22371 |
+
"grad_norm": 4.051171779632568,
|
| 22372 |
+
"learning_rate": 3.187e-05,
|
| 22373 |
+
"loss": 11.3359,
|
| 22374 |
+
"step": 3188
|
| 22375 |
+
},
|
| 22376 |
+
{
|
| 22377 |
+
"epoch": 0.0003189,
|
| 22378 |
+
"grad_norm": 3.806670665740967,
|
| 22379 |
+
"learning_rate": 3.188e-05,
|
| 22380 |
+
"loss": 10.9844,
|
| 22381 |
+
"step": 3189
|
| 22382 |
+
},
|
| 22383 |
+
{
|
| 22384 |
+
"epoch": 0.000319,
|
| 22385 |
+
"grad_norm": 4.212409019470215,
|
| 22386 |
+
"learning_rate": 3.189e-05,
|
| 22387 |
+
"loss": 12.5156,
|
| 22388 |
+
"step": 3190
|
| 22389 |
+
},
|
| 22390 |
+
{
|
| 22391 |
+
"epoch": 0.0003191,
|
| 22392 |
+
"grad_norm": 4.613623142242432,
|
| 22393 |
+
"learning_rate": 3.19e-05,
|
| 22394 |
+
"loss": 12.9609,
|
| 22395 |
+
"step": 3191
|
| 22396 |
+
},
|
| 22397 |
+
{
|
| 22398 |
+
"epoch": 0.0003192,
|
| 22399 |
+
"grad_norm": 4.048587322235107,
|
| 22400 |
+
"learning_rate": 3.1910000000000004e-05,
|
| 22401 |
+
"loss": 12.1328,
|
| 22402 |
+
"step": 3192
|
| 22403 |
+
},
|
| 22404 |
+
{
|
| 22405 |
+
"epoch": 0.0003193,
|
| 22406 |
+
"grad_norm": 4.698572635650635,
|
| 22407 |
+
"learning_rate": 3.192e-05,
|
| 22408 |
+
"loss": 12.1562,
|
| 22409 |
+
"step": 3193
|
| 22410 |
+
},
|
| 22411 |
+
{
|
| 22412 |
+
"epoch": 0.0003194,
|
| 22413 |
+
"grad_norm": 3.892582893371582,
|
| 22414 |
+
"learning_rate": 3.193e-05,
|
| 22415 |
+
"loss": 11.4062,
|
| 22416 |
+
"step": 3194
|
| 22417 |
+
},
|
| 22418 |
+
{
|
| 22419 |
+
"epoch": 0.0003195,
|
| 22420 |
+
"grad_norm": 4.513314247131348,
|
| 22421 |
+
"learning_rate": 3.194e-05,
|
| 22422 |
+
"loss": 12.7891,
|
| 22423 |
+
"step": 3195
|
| 22424 |
+
},
|
| 22425 |
+
{
|
| 22426 |
+
"epoch": 0.0003196,
|
| 22427 |
+
"grad_norm": 4.335578918457031,
|
| 22428 |
+
"learning_rate": 3.195e-05,
|
| 22429 |
+
"loss": 11.75,
|
| 22430 |
+
"step": 3196
|
| 22431 |
+
},
|
| 22432 |
+
{
|
| 22433 |
+
"epoch": 0.0003197,
|
| 22434 |
+
"grad_norm": 4.395205497741699,
|
| 22435 |
+
"learning_rate": 3.1960000000000006e-05,
|
| 22436 |
+
"loss": 12.2109,
|
| 22437 |
+
"step": 3197
|
| 22438 |
+
},
|
| 22439 |
+
{
|
| 22440 |
+
"epoch": 0.0003198,
|
| 22441 |
+
"grad_norm": 4.159797668457031,
|
| 22442 |
+
"learning_rate": 3.197e-05,
|
| 22443 |
+
"loss": 12.0,
|
| 22444 |
+
"step": 3198
|
| 22445 |
+
},
|
| 22446 |
+
{
|
| 22447 |
+
"epoch": 0.0003199,
|
| 22448 |
+
"grad_norm": 4.623922824859619,
|
| 22449 |
+
"learning_rate": 3.1979999999999995e-05,
|
| 22450 |
+
"loss": 13.2109,
|
| 22451 |
+
"step": 3199
|
| 22452 |
+
},
|
| 22453 |
+
{
|
| 22454 |
+
"epoch": 0.00032,
|
| 22455 |
+
"grad_norm": 4.834364414215088,
|
| 22456 |
+
"learning_rate": 3.1990000000000004e-05,
|
| 22457 |
+
"loss": 12.6797,
|
| 22458 |
+
"step": 3200
|
| 22459 |
+
},
|
| 22460 |
+
{
|
| 22461 |
+
"epoch": 0.0003201,
|
| 22462 |
+
"grad_norm": 4.239382266998291,
|
| 22463 |
+
"learning_rate": 3.2e-05,
|
| 22464 |
+
"loss": 11.1797,
|
| 22465 |
+
"step": 3201
|
| 22466 |
+
},
|
| 22467 |
+
{
|
| 22468 |
+
"epoch": 0.0003202,
|
| 22469 |
+
"grad_norm": 4.145647048950195,
|
| 22470 |
+
"learning_rate": 3.201e-05,
|
| 22471 |
+
"loss": 11.9219,
|
| 22472 |
+
"step": 3202
|
| 22473 |
+
},
|
| 22474 |
+
{
|
| 22475 |
+
"epoch": 0.0003203,
|
| 22476 |
+
"grad_norm": 3.926037311553955,
|
| 22477 |
+
"learning_rate": 3.202e-05,
|
| 22478 |
+
"loss": 10.5156,
|
| 22479 |
+
"step": 3203
|
| 22480 |
+
},
|
| 22481 |
+
{
|
| 22482 |
+
"epoch": 0.0003204,
|
| 22483 |
+
"grad_norm": 4.056522846221924,
|
| 22484 |
+
"learning_rate": 3.2029999999999997e-05,
|
| 22485 |
+
"loss": 11.8438,
|
| 22486 |
+
"step": 3204
|
| 22487 |
+
},
|
| 22488 |
+
{
|
| 22489 |
+
"epoch": 0.0003205,
|
| 22490 |
+
"grad_norm": 4.285017013549805,
|
| 22491 |
+
"learning_rate": 3.204e-05,
|
| 22492 |
+
"loss": 12.3672,
|
| 22493 |
+
"step": 3205
|
| 22494 |
+
},
|
| 22495 |
+
{
|
| 22496 |
+
"epoch": 0.0003206,
|
| 22497 |
+
"grad_norm": 4.016689300537109,
|
| 22498 |
+
"learning_rate": 3.205e-05,
|
| 22499 |
+
"loss": 11.2109,
|
| 22500 |
+
"step": 3206
|
| 22501 |
+
},
|
| 22502 |
+
{
|
| 22503 |
+
"epoch": 0.0003207,
|
| 22504 |
+
"grad_norm": 3.714569330215454,
|
| 22505 |
+
"learning_rate": 3.206e-05,
|
| 22506 |
+
"loss": 10.5781,
|
| 22507 |
+
"step": 3207
|
| 22508 |
+
},
|
| 22509 |
+
{
|
| 22510 |
+
"epoch": 0.0003208,
|
| 22511 |
+
"grad_norm": 4.070149898529053,
|
| 22512 |
+
"learning_rate": 3.207e-05,
|
| 22513 |
+
"loss": 12.9141,
|
| 22514 |
+
"step": 3208
|
| 22515 |
+
},
|
| 22516 |
+
{
|
| 22517 |
+
"epoch": 0.0003209,
|
| 22518 |
+
"grad_norm": 4.141865253448486,
|
| 22519 |
+
"learning_rate": 3.208e-05,
|
| 22520 |
+
"loss": 11.5703,
|
| 22521 |
+
"step": 3209
|
| 22522 |
+
},
|
| 22523 |
+
{
|
| 22524 |
+
"epoch": 0.000321,
|
| 22525 |
+
"grad_norm": 3.913559913635254,
|
| 22526 |
+
"learning_rate": 3.209e-05,
|
| 22527 |
+
"loss": 11.0781,
|
| 22528 |
+
"step": 3210
|
| 22529 |
+
},
|
| 22530 |
+
{
|
| 22531 |
+
"epoch": 0.0003211,
|
| 22532 |
+
"grad_norm": 4.005146503448486,
|
| 22533 |
+
"learning_rate": 3.21e-05,
|
| 22534 |
+
"loss": 11.5703,
|
| 22535 |
+
"step": 3211
|
| 22536 |
+
},
|
| 22537 |
+
{
|
| 22538 |
+
"epoch": 0.0003212,
|
| 22539 |
+
"grad_norm": 4.054205417633057,
|
| 22540 |
+
"learning_rate": 3.211e-05,
|
| 22541 |
+
"loss": 12.1016,
|
| 22542 |
+
"step": 3212
|
| 22543 |
+
},
|
| 22544 |
+
{
|
| 22545 |
+
"epoch": 0.0003213,
|
| 22546 |
+
"grad_norm": 4.258857250213623,
|
| 22547 |
+
"learning_rate": 3.212e-05,
|
| 22548 |
+
"loss": 11.5312,
|
| 22549 |
+
"step": 3213
|
| 22550 |
+
},
|
| 22551 |
+
{
|
| 22552 |
+
"epoch": 0.0003214,
|
| 22553 |
+
"grad_norm": 4.0924835205078125,
|
| 22554 |
+
"learning_rate": 3.2130000000000006e-05,
|
| 22555 |
+
"loss": 11.8984,
|
| 22556 |
+
"step": 3214
|
| 22557 |
+
},
|
| 22558 |
+
{
|
| 22559 |
+
"epoch": 0.0003215,
|
| 22560 |
+
"grad_norm": 4.040266990661621,
|
| 22561 |
+
"learning_rate": 3.214e-05,
|
| 22562 |
+
"loss": 11.5391,
|
| 22563 |
+
"step": 3215
|
| 22564 |
+
},
|
| 22565 |
+
{
|
| 22566 |
+
"epoch": 0.0003216,
|
| 22567 |
+
"grad_norm": 4.984847068786621,
|
| 22568 |
+
"learning_rate": 3.2149999999999995e-05,
|
| 22569 |
+
"loss": 13.5234,
|
| 22570 |
+
"step": 3216
|
| 22571 |
+
},
|
| 22572 |
+
{
|
| 22573 |
+
"epoch": 0.0003217,
|
| 22574 |
+
"grad_norm": 3.912262201309204,
|
| 22575 |
+
"learning_rate": 3.2160000000000004e-05,
|
| 22576 |
+
"loss": 11.625,
|
| 22577 |
+
"step": 3217
|
| 22578 |
+
},
|
| 22579 |
+
{
|
| 22580 |
+
"epoch": 0.0003218,
|
| 22581 |
+
"grad_norm": 3.9006030559539795,
|
| 22582 |
+
"learning_rate": 3.217e-05,
|
| 22583 |
+
"loss": 11.1641,
|
| 22584 |
+
"step": 3218
|
| 22585 |
+
},
|
| 22586 |
+
{
|
| 22587 |
+
"epoch": 0.0003219,
|
| 22588 |
+
"grad_norm": 4.044284343719482,
|
| 22589 |
+
"learning_rate": 3.218e-05,
|
| 22590 |
+
"loss": 11.5312,
|
| 22591 |
+
"step": 3219
|
| 22592 |
+
},
|
| 22593 |
+
{
|
| 22594 |
+
"epoch": 0.000322,
|
| 22595 |
+
"grad_norm": 4.083791732788086,
|
| 22596 |
+
"learning_rate": 3.219e-05,
|
| 22597 |
+
"loss": 12.5625,
|
| 22598 |
+
"step": 3220
|
| 22599 |
+
},
|
| 22600 |
+
{
|
| 22601 |
+
"epoch": 0.0003221,
|
| 22602 |
+
"grad_norm": 4.190662384033203,
|
| 22603 |
+
"learning_rate": 3.22e-05,
|
| 22604 |
+
"loss": 11.5469,
|
| 22605 |
+
"step": 3221
|
| 22606 |
+
},
|
| 22607 |
+
{
|
| 22608 |
+
"epoch": 0.0003222,
|
| 22609 |
+
"grad_norm": 4.039274215698242,
|
| 22610 |
+
"learning_rate": 3.221e-05,
|
| 22611 |
+
"loss": 11.7031,
|
| 22612 |
+
"step": 3222
|
| 22613 |
+
},
|
| 22614 |
+
{
|
| 22615 |
+
"epoch": 0.0003223,
|
| 22616 |
+
"grad_norm": 3.9752583503723145,
|
| 22617 |
+
"learning_rate": 3.222e-05,
|
| 22618 |
+
"loss": 10.8125,
|
| 22619 |
+
"step": 3223
|
| 22620 |
+
},
|
| 22621 |
+
{
|
| 22622 |
+
"epoch": 0.0003224,
|
| 22623 |
+
"grad_norm": 4.353899955749512,
|
| 22624 |
+
"learning_rate": 3.223e-05,
|
| 22625 |
+
"loss": 11.1406,
|
| 22626 |
+
"step": 3224
|
| 22627 |
+
},
|
| 22628 |
+
{
|
| 22629 |
+
"epoch": 0.0003225,
|
| 22630 |
+
"grad_norm": 4.121684551239014,
|
| 22631 |
+
"learning_rate": 3.224e-05,
|
| 22632 |
+
"loss": 11.9766,
|
| 22633 |
+
"step": 3225
|
| 22634 |
+
},
|
| 22635 |
+
{
|
| 22636 |
+
"epoch": 0.0003226,
|
| 22637 |
+
"grad_norm": 4.066223621368408,
|
| 22638 |
+
"learning_rate": 3.225e-05,
|
| 22639 |
+
"loss": 11.3672,
|
| 22640 |
+
"step": 3226
|
| 22641 |
+
},
|
| 22642 |
+
{
|
| 22643 |
+
"epoch": 0.0003227,
|
| 22644 |
+
"grad_norm": 4.243929386138916,
|
| 22645 |
+
"learning_rate": 3.226e-05,
|
| 22646 |
+
"loss": 12.2109,
|
| 22647 |
+
"step": 3227
|
| 22648 |
+
},
|
| 22649 |
+
{
|
| 22650 |
+
"epoch": 0.0003228,
|
| 22651 |
+
"grad_norm": 3.994262218475342,
|
| 22652 |
+
"learning_rate": 3.227e-05,
|
| 22653 |
+
"loss": 11.4688,
|
| 22654 |
+
"step": 3228
|
| 22655 |
+
},
|
| 22656 |
+
{
|
| 22657 |
+
"epoch": 0.0003229,
|
| 22658 |
+
"grad_norm": 3.9189815521240234,
|
| 22659 |
+
"learning_rate": 3.228e-05,
|
| 22660 |
+
"loss": 11.125,
|
| 22661 |
+
"step": 3229
|
| 22662 |
+
},
|
| 22663 |
+
{
|
| 22664 |
+
"epoch": 0.000323,
|
| 22665 |
+
"grad_norm": 4.217655181884766,
|
| 22666 |
+
"learning_rate": 3.229e-05,
|
| 22667 |
+
"loss": 11.7969,
|
| 22668 |
+
"step": 3230
|
| 22669 |
+
},
|
| 22670 |
+
{
|
| 22671 |
+
"epoch": 0.0003231,
|
| 22672 |
+
"grad_norm": 4.277668476104736,
|
| 22673 |
+
"learning_rate": 3.2300000000000006e-05,
|
| 22674 |
+
"loss": 12.2734,
|
| 22675 |
+
"step": 3231
|
| 22676 |
+
},
|
| 22677 |
+
{
|
| 22678 |
+
"epoch": 0.0003232,
|
| 22679 |
+
"grad_norm": 4.184340476989746,
|
| 22680 |
+
"learning_rate": 3.231e-05,
|
| 22681 |
+
"loss": 11.8281,
|
| 22682 |
+
"step": 3232
|
| 22683 |
+
},
|
| 22684 |
+
{
|
| 22685 |
+
"epoch": 0.0003233,
|
| 22686 |
+
"grad_norm": 3.999751567840576,
|
| 22687 |
+
"learning_rate": 3.2319999999999995e-05,
|
| 22688 |
+
"loss": 11.4766,
|
| 22689 |
+
"step": 3233
|
| 22690 |
+
},
|
| 22691 |
+
{
|
| 22692 |
+
"epoch": 0.0003234,
|
| 22693 |
+
"grad_norm": 4.286815643310547,
|
| 22694 |
+
"learning_rate": 3.2330000000000004e-05,
|
| 22695 |
+
"loss": 11.4453,
|
| 22696 |
+
"step": 3234
|
| 22697 |
+
},
|
| 22698 |
+
{
|
| 22699 |
+
"epoch": 0.0003235,
|
| 22700 |
+
"grad_norm": 4.097026824951172,
|
| 22701 |
+
"learning_rate": 3.234e-05,
|
| 22702 |
+
"loss": 11.2734,
|
| 22703 |
+
"step": 3235
|
| 22704 |
+
},
|
| 22705 |
+
{
|
| 22706 |
+
"epoch": 0.0003236,
|
| 22707 |
+
"grad_norm": 4.420822620391846,
|
| 22708 |
+
"learning_rate": 3.235e-05,
|
| 22709 |
+
"loss": 11.9766,
|
| 22710 |
+
"step": 3236
|
| 22711 |
+
},
|
| 22712 |
+
{
|
| 22713 |
+
"epoch": 0.0003237,
|
| 22714 |
+
"grad_norm": 4.046294689178467,
|
| 22715 |
+
"learning_rate": 3.236e-05,
|
| 22716 |
+
"loss": 10.9688,
|
| 22717 |
+
"step": 3237
|
| 22718 |
+
},
|
| 22719 |
+
{
|
| 22720 |
+
"epoch": 0.0003238,
|
| 22721 |
+
"grad_norm": 4.0677809715271,
|
| 22722 |
+
"learning_rate": 3.237e-05,
|
| 22723 |
+
"loss": 11.6641,
|
| 22724 |
+
"step": 3238
|
| 22725 |
+
},
|
| 22726 |
+
{
|
| 22727 |
+
"epoch": 0.0003239,
|
| 22728 |
+
"grad_norm": 4.110946178436279,
|
| 22729 |
+
"learning_rate": 3.238e-05,
|
| 22730 |
+
"loss": 11.1562,
|
| 22731 |
+
"step": 3239
|
| 22732 |
+
},
|
| 22733 |
+
{
|
| 22734 |
+
"epoch": 0.000324,
|
| 22735 |
+
"grad_norm": 4.498655319213867,
|
| 22736 |
+
"learning_rate": 3.239e-05,
|
| 22737 |
+
"loss": 12.2891,
|
| 22738 |
+
"step": 3240
|
| 22739 |
+
},
|
| 22740 |
+
{
|
| 22741 |
+
"epoch": 0.0003241,
|
| 22742 |
+
"grad_norm": 4.171398162841797,
|
| 22743 |
+
"learning_rate": 3.24e-05,
|
| 22744 |
+
"loss": 11.6875,
|
| 22745 |
+
"step": 3241
|
| 22746 |
+
},
|
| 22747 |
+
{
|
| 22748 |
+
"epoch": 0.0003242,
|
| 22749 |
+
"grad_norm": 4.259690284729004,
|
| 22750 |
+
"learning_rate": 3.241e-05,
|
| 22751 |
+
"loss": 11.9062,
|
| 22752 |
+
"step": 3242
|
| 22753 |
+
},
|
| 22754 |
+
{
|
| 22755 |
+
"epoch": 0.0003243,
|
| 22756 |
+
"grad_norm": 4.196627140045166,
|
| 22757 |
+
"learning_rate": 3.242e-05,
|
| 22758 |
+
"loss": 11.4844,
|
| 22759 |
+
"step": 3243
|
| 22760 |
+
},
|
| 22761 |
+
{
|
| 22762 |
+
"epoch": 0.0003244,
|
| 22763 |
+
"grad_norm": 4.299365043640137,
|
| 22764 |
+
"learning_rate": 3.243e-05,
|
| 22765 |
+
"loss": 11.9297,
|
| 22766 |
+
"step": 3244
|
| 22767 |
+
},
|
| 22768 |
+
{
|
| 22769 |
+
"epoch": 0.0003245,
|
| 22770 |
+
"grad_norm": 4.672417163848877,
|
| 22771 |
+
"learning_rate": 3.244e-05,
|
| 22772 |
+
"loss": 12.8359,
|
| 22773 |
+
"step": 3245
|
| 22774 |
+
},
|
| 22775 |
+
{
|
| 22776 |
+
"epoch": 0.0003246,
|
| 22777 |
+
"grad_norm": 4.864752769470215,
|
| 22778 |
+
"learning_rate": 3.245e-05,
|
| 22779 |
+
"loss": 12.2266,
|
| 22780 |
+
"step": 3246
|
| 22781 |
+
},
|
| 22782 |
+
{
|
| 22783 |
+
"epoch": 0.0003247,
|
| 22784 |
+
"grad_norm": 3.9772098064422607,
|
| 22785 |
+
"learning_rate": 3.246e-05,
|
| 22786 |
+
"loss": 11.5391,
|
| 22787 |
+
"step": 3247
|
| 22788 |
+
},
|
| 22789 |
+
{
|
| 22790 |
+
"epoch": 0.0003248,
|
| 22791 |
+
"grad_norm": 3.984213352203369,
|
| 22792 |
+
"learning_rate": 3.2470000000000006e-05,
|
| 22793 |
+
"loss": 11.2188,
|
| 22794 |
+
"step": 3248
|
| 22795 |
+
},
|
| 22796 |
+
{
|
| 22797 |
+
"epoch": 0.0003249,
|
| 22798 |
+
"grad_norm": 4.254142761230469,
|
| 22799 |
+
"learning_rate": 3.248e-05,
|
| 22800 |
+
"loss": 11.0781,
|
| 22801 |
+
"step": 3249
|
| 22802 |
+
},
|
| 22803 |
+
{
|
| 22804 |
+
"epoch": 0.000325,
|
| 22805 |
+
"grad_norm": 4.016782283782959,
|
| 22806 |
+
"learning_rate": 3.2489999999999996e-05,
|
| 22807 |
+
"loss": 10.7266,
|
| 22808 |
+
"step": 3250
|
| 22809 |
+
},
|
| 22810 |
+
{
|
| 22811 |
+
"epoch": 0.0003251,
|
| 22812 |
+
"grad_norm": 4.111670970916748,
|
| 22813 |
+
"learning_rate": 3.2500000000000004e-05,
|
| 22814 |
+
"loss": 10.8047,
|
| 22815 |
+
"step": 3251
|
| 22816 |
+
},
|
| 22817 |
+
{
|
| 22818 |
+
"epoch": 0.0003252,
|
| 22819 |
+
"grad_norm": 3.729321002960205,
|
| 22820 |
+
"learning_rate": 3.251e-05,
|
| 22821 |
+
"loss": 10.0312,
|
| 22822 |
+
"step": 3252
|
| 22823 |
+
},
|
| 22824 |
+
{
|
| 22825 |
+
"epoch": 0.0003253,
|
| 22826 |
+
"grad_norm": 4.472454071044922,
|
| 22827 |
+
"learning_rate": 3.252e-05,
|
| 22828 |
+
"loss": 11.3984,
|
| 22829 |
+
"step": 3253
|
| 22830 |
+
},
|
| 22831 |
+
{
|
| 22832 |
+
"epoch": 0.0003254,
|
| 22833 |
+
"grad_norm": 5.1932196617126465,
|
| 22834 |
+
"learning_rate": 3.253e-05,
|
| 22835 |
+
"loss": 12.0156,
|
| 22836 |
+
"step": 3254
|
| 22837 |
+
},
|
| 22838 |
+
{
|
| 22839 |
+
"epoch": 0.0003255,
|
| 22840 |
+
"grad_norm": 4.058501243591309,
|
| 22841 |
+
"learning_rate": 3.254e-05,
|
| 22842 |
+
"loss": 10.6094,
|
| 22843 |
+
"step": 3255
|
| 22844 |
+
},
|
| 22845 |
+
{
|
| 22846 |
+
"epoch": 0.0003256,
|
| 22847 |
+
"grad_norm": 4.450124263763428,
|
| 22848 |
+
"learning_rate": 3.255e-05,
|
| 22849 |
+
"loss": 11.2656,
|
| 22850 |
+
"step": 3256
|
| 22851 |
+
},
|
| 22852 |
+
{
|
| 22853 |
+
"epoch": 0.0003257,
|
| 22854 |
+
"grad_norm": 4.315848350524902,
|
| 22855 |
+
"learning_rate": 3.256e-05,
|
| 22856 |
+
"loss": 11.5625,
|
| 22857 |
+
"step": 3257
|
| 22858 |
+
},
|
| 22859 |
+
{
|
| 22860 |
+
"epoch": 0.0003258,
|
| 22861 |
+
"grad_norm": 4.343723773956299,
|
| 22862 |
+
"learning_rate": 3.257e-05,
|
| 22863 |
+
"loss": 12.2266,
|
| 22864 |
+
"step": 3258
|
| 22865 |
+
},
|
| 22866 |
+
{
|
| 22867 |
+
"epoch": 0.0003259,
|
| 22868 |
+
"grad_norm": 4.783463478088379,
|
| 22869 |
+
"learning_rate": 3.258e-05,
|
| 22870 |
+
"loss": 12.2188,
|
| 22871 |
+
"step": 3259
|
| 22872 |
+
},
|
| 22873 |
+
{
|
| 22874 |
+
"epoch": 0.000326,
|
| 22875 |
+
"grad_norm": 4.320399761199951,
|
| 22876 |
+
"learning_rate": 3.259e-05,
|
| 22877 |
+
"loss": 11.8281,
|
| 22878 |
+
"step": 3260
|
| 22879 |
+
},
|
| 22880 |
+
{
|
| 22881 |
+
"epoch": 0.0003261,
|
| 22882 |
+
"grad_norm": 4.1972551345825195,
|
| 22883 |
+
"learning_rate": 3.26e-05,
|
| 22884 |
+
"loss": 11.5156,
|
| 22885 |
+
"step": 3261
|
| 22886 |
+
},
|
| 22887 |
+
{
|
| 22888 |
+
"epoch": 0.0003262,
|
| 22889 |
+
"grad_norm": 3.796144485473633,
|
| 22890 |
+
"learning_rate": 3.261e-05,
|
| 22891 |
+
"loss": 10.1094,
|
| 22892 |
+
"step": 3262
|
| 22893 |
+
},
|
| 22894 |
+
{
|
| 22895 |
+
"epoch": 0.0003263,
|
| 22896 |
+
"grad_norm": 3.7178866863250732,
|
| 22897 |
+
"learning_rate": 3.262e-05,
|
| 22898 |
+
"loss": 10.7422,
|
| 22899 |
+
"step": 3263
|
| 22900 |
+
},
|
| 22901 |
+
{
|
| 22902 |
+
"epoch": 0.0003264,
|
| 22903 |
+
"grad_norm": 4.453627586364746,
|
| 22904 |
+
"learning_rate": 3.263e-05,
|
| 22905 |
+
"loss": 12.5859,
|
| 22906 |
+
"step": 3264
|
| 22907 |
+
},
|
| 22908 |
+
{
|
| 22909 |
+
"epoch": 0.0003265,
|
| 22910 |
+
"grad_norm": 4.911516189575195,
|
| 22911 |
+
"learning_rate": 3.2640000000000006e-05,
|
| 22912 |
+
"loss": 12.3359,
|
| 22913 |
+
"step": 3265
|
| 22914 |
+
},
|
| 22915 |
+
{
|
| 22916 |
+
"epoch": 0.0003266,
|
| 22917 |
+
"grad_norm": 4.017552375793457,
|
| 22918 |
+
"learning_rate": 3.265e-05,
|
| 22919 |
+
"loss": 10.2344,
|
| 22920 |
+
"step": 3266
|
| 22921 |
+
},
|
| 22922 |
+
{
|
| 22923 |
+
"epoch": 0.0003267,
|
| 22924 |
+
"grad_norm": 4.349109649658203,
|
| 22925 |
+
"learning_rate": 3.2659999999999996e-05,
|
| 22926 |
+
"loss": 11.1641,
|
| 22927 |
+
"step": 3267
|
| 22928 |
+
},
|
| 22929 |
+
{
|
| 22930 |
+
"epoch": 0.0003268,
|
| 22931 |
+
"grad_norm": 4.3839430809021,
|
| 22932 |
+
"learning_rate": 3.2670000000000004e-05,
|
| 22933 |
+
"loss": 11.7188,
|
| 22934 |
+
"step": 3268
|
| 22935 |
+
},
|
| 22936 |
+
{
|
| 22937 |
+
"epoch": 0.0003269,
|
| 22938 |
+
"grad_norm": 4.069872856140137,
|
| 22939 |
+
"learning_rate": 3.268e-05,
|
| 22940 |
+
"loss": 11.5312,
|
| 22941 |
+
"step": 3269
|
| 22942 |
+
},
|
| 22943 |
+
{
|
| 22944 |
+
"epoch": 0.000327,
|
| 22945 |
+
"grad_norm": 4.946285724639893,
|
| 22946 |
+
"learning_rate": 3.269e-05,
|
| 22947 |
+
"loss": 12.8594,
|
| 22948 |
+
"step": 3270
|
| 22949 |
+
},
|
| 22950 |
+
{
|
| 22951 |
+
"epoch": 0.0003271,
|
| 22952 |
+
"grad_norm": 4.320945739746094,
|
| 22953 |
+
"learning_rate": 3.27e-05,
|
| 22954 |
+
"loss": 11.7188,
|
| 22955 |
+
"step": 3271
|
| 22956 |
+
},
|
| 22957 |
+
{
|
| 22958 |
+
"epoch": 0.0003272,
|
| 22959 |
+
"grad_norm": 4.11987829208374,
|
| 22960 |
+
"learning_rate": 3.271e-05,
|
| 22961 |
+
"loss": 11.2891,
|
| 22962 |
+
"step": 3272
|
| 22963 |
+
},
|
| 22964 |
+
{
|
| 22965 |
+
"epoch": 0.0003273,
|
| 22966 |
+
"grad_norm": 4.189293384552002,
|
| 22967 |
+
"learning_rate": 3.272e-05,
|
| 22968 |
+
"loss": 11.2812,
|
| 22969 |
+
"step": 3273
|
| 22970 |
+
},
|
| 22971 |
+
{
|
| 22972 |
+
"epoch": 0.0003274,
|
| 22973 |
+
"grad_norm": 4.540557384490967,
|
| 22974 |
+
"learning_rate": 3.273e-05,
|
| 22975 |
+
"loss": 11.7266,
|
| 22976 |
+
"step": 3274
|
| 22977 |
+
},
|
| 22978 |
+
{
|
| 22979 |
+
"epoch": 0.0003275,
|
| 22980 |
+
"grad_norm": 4.37672233581543,
|
| 22981 |
+
"learning_rate": 3.274e-05,
|
| 22982 |
+
"loss": 11.25,
|
| 22983 |
+
"step": 3275
|
| 22984 |
+
},
|
| 22985 |
+
{
|
| 22986 |
+
"epoch": 0.0003276,
|
| 22987 |
+
"grad_norm": 4.313999176025391,
|
| 22988 |
+
"learning_rate": 3.275e-05,
|
| 22989 |
+
"loss": 11.3672,
|
| 22990 |
+
"step": 3276
|
| 22991 |
+
},
|
| 22992 |
+
{
|
| 22993 |
+
"epoch": 0.0003277,
|
| 22994 |
+
"grad_norm": 4.196670055389404,
|
| 22995 |
+
"learning_rate": 3.276e-05,
|
| 22996 |
+
"loss": 12.3125,
|
| 22997 |
+
"step": 3277
|
| 22998 |
+
},
|
| 22999 |
+
{
|
| 23000 |
+
"epoch": 0.0003278,
|
| 23001 |
+
"grad_norm": 3.7270560264587402,
|
| 23002 |
+
"learning_rate": 3.277e-05,
|
| 23003 |
+
"loss": 10.4453,
|
| 23004 |
+
"step": 3278
|
| 23005 |
+
},
|
| 23006 |
+
{
|
| 23007 |
+
"epoch": 0.0003279,
|
| 23008 |
+
"grad_norm": 3.9468445777893066,
|
| 23009 |
+
"learning_rate": 3.278e-05,
|
| 23010 |
+
"loss": 10.5625,
|
| 23011 |
+
"step": 3279
|
| 23012 |
+
},
|
| 23013 |
+
{
|
| 23014 |
+
"epoch": 0.000328,
|
| 23015 |
+
"grad_norm": 4.433849334716797,
|
| 23016 |
+
"learning_rate": 3.279e-05,
|
| 23017 |
+
"loss": 12.7969,
|
| 23018 |
+
"step": 3280
|
| 23019 |
+
},
|
| 23020 |
+
{
|
| 23021 |
+
"epoch": 0.0003281,
|
| 23022 |
+
"grad_norm": 4.000179290771484,
|
| 23023 |
+
"learning_rate": 3.28e-05,
|
| 23024 |
+
"loss": 10.4688,
|
| 23025 |
+
"step": 3281
|
| 23026 |
+
},
|
| 23027 |
+
{
|
| 23028 |
+
"epoch": 0.0003282,
|
| 23029 |
+
"grad_norm": 4.158862113952637,
|
| 23030 |
+
"learning_rate": 3.2810000000000006e-05,
|
| 23031 |
+
"loss": 10.7656,
|
| 23032 |
+
"step": 3282
|
| 23033 |
+
},
|
| 23034 |
+
{
|
| 23035 |
+
"epoch": 0.0003283,
|
| 23036 |
+
"grad_norm": 4.051787853240967,
|
| 23037 |
+
"learning_rate": 3.282e-05,
|
| 23038 |
+
"loss": 10.8594,
|
| 23039 |
+
"step": 3283
|
| 23040 |
+
},
|
| 23041 |
+
{
|
| 23042 |
+
"epoch": 0.0003284,
|
| 23043 |
+
"grad_norm": 4.009098529815674,
|
| 23044 |
+
"learning_rate": 3.2829999999999996e-05,
|
| 23045 |
+
"loss": 11.3594,
|
| 23046 |
+
"step": 3284
|
| 23047 |
+
},
|
| 23048 |
+
{
|
| 23049 |
+
"epoch": 0.0003285,
|
| 23050 |
+
"grad_norm": 4.038074493408203,
|
| 23051 |
+
"learning_rate": 3.2840000000000004e-05,
|
| 23052 |
+
"loss": 11.25,
|
| 23053 |
+
"step": 3285
|
| 23054 |
+
},
|
| 23055 |
+
{
|
| 23056 |
+
"epoch": 0.0003286,
|
| 23057 |
+
"grad_norm": 4.692169666290283,
|
| 23058 |
+
"learning_rate": 3.285e-05,
|
| 23059 |
+
"loss": 11.8828,
|
| 23060 |
+
"step": 3286
|
| 23061 |
+
},
|
| 23062 |
+
{
|
| 23063 |
+
"epoch": 0.0003287,
|
| 23064 |
+
"grad_norm": 4.001228332519531,
|
| 23065 |
+
"learning_rate": 3.286e-05,
|
| 23066 |
+
"loss": 10.6562,
|
| 23067 |
+
"step": 3287
|
| 23068 |
+
},
|
| 23069 |
+
{
|
| 23070 |
+
"epoch": 0.0003288,
|
| 23071 |
+
"grad_norm": 4.636309623718262,
|
| 23072 |
+
"learning_rate": 3.287e-05,
|
| 23073 |
+
"loss": 11.6328,
|
| 23074 |
+
"step": 3288
|
| 23075 |
+
},
|
| 23076 |
+
{
|
| 23077 |
+
"epoch": 0.0003289,
|
| 23078 |
+
"grad_norm": 4.129543781280518,
|
| 23079 |
+
"learning_rate": 3.288e-05,
|
| 23080 |
+
"loss": 11.5547,
|
| 23081 |
+
"step": 3289
|
| 23082 |
+
},
|
| 23083 |
+
{
|
| 23084 |
+
"epoch": 0.000329,
|
| 23085 |
+
"grad_norm": 4.099421977996826,
|
| 23086 |
+
"learning_rate": 3.289e-05,
|
| 23087 |
+
"loss": 11.0469,
|
| 23088 |
+
"step": 3290
|
| 23089 |
+
},
|
| 23090 |
+
{
|
| 23091 |
+
"epoch": 0.0003291,
|
| 23092 |
+
"grad_norm": 4.2602643966674805,
|
| 23093 |
+
"learning_rate": 3.29e-05,
|
| 23094 |
+
"loss": 11.9375,
|
| 23095 |
+
"step": 3291
|
| 23096 |
+
},
|
| 23097 |
+
{
|
| 23098 |
+
"epoch": 0.0003292,
|
| 23099 |
+
"grad_norm": 4.596462726593018,
|
| 23100 |
+
"learning_rate": 3.291e-05,
|
| 23101 |
+
"loss": 12.2734,
|
| 23102 |
+
"step": 3292
|
| 23103 |
+
},
|
| 23104 |
+
{
|
| 23105 |
+
"epoch": 0.0003293,
|
| 23106 |
+
"grad_norm": 4.166189670562744,
|
| 23107 |
+
"learning_rate": 3.292e-05,
|
| 23108 |
+
"loss": 11.4375,
|
| 23109 |
+
"step": 3293
|
| 23110 |
+
},
|
| 23111 |
+
{
|
| 23112 |
+
"epoch": 0.0003294,
|
| 23113 |
+
"grad_norm": 5.362889766693115,
|
| 23114 |
+
"learning_rate": 3.293e-05,
|
| 23115 |
+
"loss": 12.2266,
|
| 23116 |
+
"step": 3294
|
| 23117 |
+
},
|
| 23118 |
+
{
|
| 23119 |
+
"epoch": 0.0003295,
|
| 23120 |
+
"grad_norm": 4.621495246887207,
|
| 23121 |
+
"learning_rate": 3.294e-05,
|
| 23122 |
+
"loss": 12.7969,
|
| 23123 |
+
"step": 3295
|
| 23124 |
+
},
|
| 23125 |
+
{
|
| 23126 |
+
"epoch": 0.0003296,
|
| 23127 |
+
"grad_norm": 4.488149642944336,
|
| 23128 |
+
"learning_rate": 3.295e-05,
|
| 23129 |
+
"loss": 11.625,
|
| 23130 |
+
"step": 3296
|
| 23131 |
+
},
|
| 23132 |
+
{
|
| 23133 |
+
"epoch": 0.0003297,
|
| 23134 |
+
"grad_norm": 3.998244524002075,
|
| 23135 |
+
"learning_rate": 3.296e-05,
|
| 23136 |
+
"loss": 10.5703,
|
| 23137 |
+
"step": 3297
|
| 23138 |
+
},
|
| 23139 |
+
{
|
| 23140 |
+
"epoch": 0.0003298,
|
| 23141 |
+
"grad_norm": 4.318847179412842,
|
| 23142 |
+
"learning_rate": 3.297e-05,
|
| 23143 |
+
"loss": 11.5781,
|
| 23144 |
+
"step": 3298
|
| 23145 |
+
},
|
| 23146 |
+
{
|
| 23147 |
+
"epoch": 0.0003299,
|
| 23148 |
+
"grad_norm": 4.119663715362549,
|
| 23149 |
+
"learning_rate": 3.2980000000000006e-05,
|
| 23150 |
+
"loss": 10.5,
|
| 23151 |
+
"step": 3299
|
| 23152 |
+
},
|
| 23153 |
+
{
|
| 23154 |
+
"epoch": 0.00033,
|
| 23155 |
+
"grad_norm": 3.9446024894714355,
|
| 23156 |
+
"learning_rate": 3.299e-05,
|
| 23157 |
+
"loss": 10.4844,
|
| 23158 |
+
"step": 3300
|
| 23159 |
+
},
|
| 23160 |
+
{
|
| 23161 |
+
"epoch": 0.0003301,
|
| 23162 |
+
"grad_norm": 4.198339939117432,
|
| 23163 |
+
"learning_rate": 3.2999999999999996e-05,
|
| 23164 |
+
"loss": 11.1875,
|
| 23165 |
+
"step": 3301
|
| 23166 |
+
},
|
| 23167 |
+
{
|
| 23168 |
+
"epoch": 0.0003302,
|
| 23169 |
+
"grad_norm": 4.08008337020874,
|
| 23170 |
+
"learning_rate": 3.3010000000000004e-05,
|
| 23171 |
+
"loss": 10.8828,
|
| 23172 |
+
"step": 3302
|
| 23173 |
+
},
|
| 23174 |
+
{
|
| 23175 |
+
"epoch": 0.0003303,
|
| 23176 |
+
"grad_norm": 4.160445213317871,
|
| 23177 |
+
"learning_rate": 3.302e-05,
|
| 23178 |
+
"loss": 10.6875,
|
| 23179 |
+
"step": 3303
|
| 23180 |
+
},
|
| 23181 |
+
{
|
| 23182 |
+
"epoch": 0.0003304,
|
| 23183 |
+
"grad_norm": 4.125695705413818,
|
| 23184 |
+
"learning_rate": 3.303e-05,
|
| 23185 |
+
"loss": 11.8438,
|
| 23186 |
+
"step": 3304
|
| 23187 |
+
},
|
| 23188 |
+
{
|
| 23189 |
+
"epoch": 0.0003305,
|
| 23190 |
+
"grad_norm": 4.040487766265869,
|
| 23191 |
+
"learning_rate": 3.304e-05,
|
| 23192 |
+
"loss": 11.0938,
|
| 23193 |
+
"step": 3305
|
| 23194 |
+
},
|
| 23195 |
+
{
|
| 23196 |
+
"epoch": 0.0003306,
|
| 23197 |
+
"grad_norm": 4.203553676605225,
|
| 23198 |
+
"learning_rate": 3.305e-05,
|
| 23199 |
+
"loss": 11.1328,
|
| 23200 |
+
"step": 3306
|
| 23201 |
+
},
|
| 23202 |
+
{
|
| 23203 |
+
"epoch": 0.0003307,
|
| 23204 |
+
"grad_norm": 4.719578742980957,
|
| 23205 |
+
"learning_rate": 3.306e-05,
|
| 23206 |
+
"loss": 11.2891,
|
| 23207 |
+
"step": 3307
|
| 23208 |
+
},
|
| 23209 |
+
{
|
| 23210 |
+
"epoch": 0.0003308,
|
| 23211 |
+
"grad_norm": 3.790029525756836,
|
| 23212 |
+
"learning_rate": 3.307e-05,
|
| 23213 |
+
"loss": 9.7578,
|
| 23214 |
+
"step": 3308
|
| 23215 |
+
},
|
| 23216 |
+
{
|
| 23217 |
+
"epoch": 0.0003309,
|
| 23218 |
+
"grad_norm": 4.178408622741699,
|
| 23219 |
+
"learning_rate": 3.308e-05,
|
| 23220 |
+
"loss": 11.0547,
|
| 23221 |
+
"step": 3309
|
| 23222 |
+
},
|
| 23223 |
+
{
|
| 23224 |
+
"epoch": 0.000331,
|
| 23225 |
+
"grad_norm": 3.88385009765625,
|
| 23226 |
+
"learning_rate": 3.309e-05,
|
| 23227 |
+
"loss": 10.5547,
|
| 23228 |
+
"step": 3310
|
| 23229 |
+
},
|
| 23230 |
+
{
|
| 23231 |
+
"epoch": 0.0003311,
|
| 23232 |
+
"grad_norm": 4.04526424407959,
|
| 23233 |
+
"learning_rate": 3.31e-05,
|
| 23234 |
+
"loss": 11.1953,
|
| 23235 |
+
"step": 3311
|
| 23236 |
+
},
|
| 23237 |
+
{
|
| 23238 |
+
"epoch": 0.0003312,
|
| 23239 |
+
"grad_norm": 4.0054121017456055,
|
| 23240 |
+
"learning_rate": 3.311e-05,
|
| 23241 |
+
"loss": 10.7812,
|
| 23242 |
+
"step": 3312
|
| 23243 |
+
},
|
| 23244 |
+
{
|
| 23245 |
+
"epoch": 0.0003313,
|
| 23246 |
+
"grad_norm": 4.404824733734131,
|
| 23247 |
+
"learning_rate": 3.312e-05,
|
| 23248 |
+
"loss": 11.3281,
|
| 23249 |
+
"step": 3313
|
| 23250 |
+
},
|
| 23251 |
+
{
|
| 23252 |
+
"epoch": 0.0003314,
|
| 23253 |
+
"grad_norm": 4.11653470993042,
|
| 23254 |
+
"learning_rate": 3.313e-05,
|
| 23255 |
+
"loss": 10.4531,
|
| 23256 |
+
"step": 3314
|
| 23257 |
+
},
|
| 23258 |
+
{
|
| 23259 |
+
"epoch": 0.0003315,
|
| 23260 |
+
"grad_norm": 4.128344535827637,
|
| 23261 |
+
"learning_rate": 3.314e-05,
|
| 23262 |
+
"loss": 10.1562,
|
| 23263 |
+
"step": 3315
|
| 23264 |
+
},
|
| 23265 |
+
{
|
| 23266 |
+
"epoch": 0.0003316,
|
| 23267 |
+
"grad_norm": 4.2458600997924805,
|
| 23268 |
+
"learning_rate": 3.3150000000000006e-05,
|
| 23269 |
+
"loss": 11.1094,
|
| 23270 |
+
"step": 3316
|
| 23271 |
+
},
|
| 23272 |
+
{
|
| 23273 |
+
"epoch": 0.0003317,
|
| 23274 |
+
"grad_norm": 3.804975986480713,
|
| 23275 |
+
"learning_rate": 3.316e-05,
|
| 23276 |
+
"loss": 10.3438,
|
| 23277 |
+
"step": 3317
|
| 23278 |
+
},
|
| 23279 |
+
{
|
| 23280 |
+
"epoch": 0.0003318,
|
| 23281 |
+
"grad_norm": 4.2271809577941895,
|
| 23282 |
+
"learning_rate": 3.3169999999999996e-05,
|
| 23283 |
+
"loss": 11.125,
|
| 23284 |
+
"step": 3318
|
| 23285 |
+
},
|
| 23286 |
+
{
|
| 23287 |
+
"epoch": 0.0003319,
|
| 23288 |
+
"grad_norm": 3.787332057952881,
|
| 23289 |
+
"learning_rate": 3.3180000000000004e-05,
|
| 23290 |
+
"loss": 10.2812,
|
| 23291 |
+
"step": 3319
|
| 23292 |
+
},
|
| 23293 |
+
{
|
| 23294 |
+
"epoch": 0.000332,
|
| 23295 |
+
"grad_norm": 4.239166259765625,
|
| 23296 |
+
"learning_rate": 3.319e-05,
|
| 23297 |
+
"loss": 10.9141,
|
| 23298 |
+
"step": 3320
|
| 23299 |
+
},
|
| 23300 |
+
{
|
| 23301 |
+
"epoch": 0.0003321,
|
| 23302 |
+
"grad_norm": 4.154354572296143,
|
| 23303 |
+
"learning_rate": 3.32e-05,
|
| 23304 |
+
"loss": 10.8984,
|
| 23305 |
+
"step": 3321
|
| 23306 |
+
},
|
| 23307 |
+
{
|
| 23308 |
+
"epoch": 0.0003322,
|
| 23309 |
+
"grad_norm": 3.8980112075805664,
|
| 23310 |
+
"learning_rate": 3.321e-05,
|
| 23311 |
+
"loss": 10.875,
|
| 23312 |
+
"step": 3322
|
| 23313 |
+
},
|
| 23314 |
+
{
|
| 23315 |
+
"epoch": 0.0003323,
|
| 23316 |
+
"grad_norm": 3.9956798553466797,
|
| 23317 |
+
"learning_rate": 3.322e-05,
|
| 23318 |
+
"loss": 11.0078,
|
| 23319 |
+
"step": 3323
|
| 23320 |
+
},
|
| 23321 |
+
{
|
| 23322 |
+
"epoch": 0.0003324,
|
| 23323 |
+
"grad_norm": 4.027379512786865,
|
| 23324 |
+
"learning_rate": 3.323e-05,
|
| 23325 |
+
"loss": 10.4844,
|
| 23326 |
+
"step": 3324
|
| 23327 |
+
},
|
| 23328 |
+
{
|
| 23329 |
+
"epoch": 0.0003325,
|
| 23330 |
+
"grad_norm": 4.109697341918945,
|
| 23331 |
+
"learning_rate": 3.324e-05,
|
| 23332 |
+
"loss": 10.7266,
|
| 23333 |
+
"step": 3325
|
| 23334 |
+
},
|
| 23335 |
+
{
|
| 23336 |
+
"epoch": 0.0003326,
|
| 23337 |
+
"grad_norm": 4.49432897567749,
|
| 23338 |
+
"learning_rate": 3.325e-05,
|
| 23339 |
+
"loss": 11.5391,
|
| 23340 |
+
"step": 3326
|
| 23341 |
+
},
|
| 23342 |
+
{
|
| 23343 |
+
"epoch": 0.0003327,
|
| 23344 |
+
"grad_norm": 4.328161239624023,
|
| 23345 |
+
"learning_rate": 3.3260000000000003e-05,
|
| 23346 |
+
"loss": 11.3203,
|
| 23347 |
+
"step": 3327
|
| 23348 |
+
},
|
| 23349 |
+
{
|
| 23350 |
+
"epoch": 0.0003328,
|
| 23351 |
+
"grad_norm": 4.315303325653076,
|
| 23352 |
+
"learning_rate": 3.327e-05,
|
| 23353 |
+
"loss": 10.6797,
|
| 23354 |
+
"step": 3328
|
| 23355 |
+
},
|
| 23356 |
+
{
|
| 23357 |
+
"epoch": 0.0003329,
|
| 23358 |
+
"grad_norm": 4.174460411071777,
|
| 23359 |
+
"learning_rate": 3.328e-05,
|
| 23360 |
+
"loss": 11.0234,
|
| 23361 |
+
"step": 3329
|
| 23362 |
+
},
|
| 23363 |
+
{
|
| 23364 |
+
"epoch": 0.000333,
|
| 23365 |
+
"grad_norm": 3.93855357170105,
|
| 23366 |
+
"learning_rate": 3.329e-05,
|
| 23367 |
+
"loss": 10.3125,
|
| 23368 |
+
"step": 3330
|
| 23369 |
+
},
|
| 23370 |
+
{
|
| 23371 |
+
"epoch": 0.0003331,
|
| 23372 |
+
"grad_norm": 4.022221565246582,
|
| 23373 |
+
"learning_rate": 3.33e-05,
|
| 23374 |
+
"loss": 10.7188,
|
| 23375 |
+
"step": 3331
|
| 23376 |
+
},
|
| 23377 |
+
{
|
| 23378 |
+
"epoch": 0.0003332,
|
| 23379 |
+
"grad_norm": 4.30765438079834,
|
| 23380 |
+
"learning_rate": 3.331e-05,
|
| 23381 |
+
"loss": 10.0547,
|
| 23382 |
+
"step": 3332
|
| 23383 |
+
},
|
| 23384 |
+
{
|
| 23385 |
+
"epoch": 0.0003333,
|
| 23386 |
+
"grad_norm": 3.8584938049316406,
|
| 23387 |
+
"learning_rate": 3.3320000000000006e-05,
|
| 23388 |
+
"loss": 9.8906,
|
| 23389 |
+
"step": 3333
|
| 23390 |
+
},
|
| 23391 |
+
{
|
| 23392 |
+
"epoch": 0.0003334,
|
| 23393 |
+
"grad_norm": 4.225302219390869,
|
| 23394 |
+
"learning_rate": 3.333e-05,
|
| 23395 |
+
"loss": 10.1094,
|
| 23396 |
+
"step": 3334
|
| 23397 |
+
},
|
| 23398 |
+
{
|
| 23399 |
+
"epoch": 0.0003335,
|
| 23400 |
+
"grad_norm": 3.9800033569335938,
|
| 23401 |
+
"learning_rate": 3.3339999999999996e-05,
|
| 23402 |
+
"loss": 10.4609,
|
| 23403 |
+
"step": 3335
|
| 23404 |
+
},
|
| 23405 |
+
{
|
| 23406 |
+
"epoch": 0.0003336,
|
| 23407 |
+
"grad_norm": 4.5449299812316895,
|
| 23408 |
+
"learning_rate": 3.3350000000000004e-05,
|
| 23409 |
+
"loss": 12.0703,
|
| 23410 |
+
"step": 3336
|
| 23411 |
+
},
|
| 23412 |
+
{
|
| 23413 |
+
"epoch": 0.0003337,
|
| 23414 |
+
"grad_norm": 4.289771556854248,
|
| 23415 |
+
"learning_rate": 3.336e-05,
|
| 23416 |
+
"loss": 10.6875,
|
| 23417 |
+
"step": 3337
|
| 23418 |
+
},
|
| 23419 |
+
{
|
| 23420 |
+
"epoch": 0.0003338,
|
| 23421 |
+
"grad_norm": 4.595143795013428,
|
| 23422 |
+
"learning_rate": 3.337e-05,
|
| 23423 |
+
"loss": 10.6562,
|
| 23424 |
+
"step": 3338
|
| 23425 |
+
},
|
| 23426 |
+
{
|
| 23427 |
+
"epoch": 0.0003339,
|
| 23428 |
+
"grad_norm": 4.658059120178223,
|
| 23429 |
+
"learning_rate": 3.338e-05,
|
| 23430 |
+
"loss": 11.6562,
|
| 23431 |
+
"step": 3339
|
| 23432 |
+
},
|
| 23433 |
+
{
|
| 23434 |
+
"epoch": 0.000334,
|
| 23435 |
+
"grad_norm": 4.367034435272217,
|
| 23436 |
+
"learning_rate": 3.339e-05,
|
| 23437 |
+
"loss": 11.3984,
|
| 23438 |
+
"step": 3340
|
| 23439 |
+
},
|
| 23440 |
+
{
|
| 23441 |
+
"epoch": 0.0003341,
|
| 23442 |
+
"grad_norm": 3.9704926013946533,
|
| 23443 |
+
"learning_rate": 3.34e-05,
|
| 23444 |
+
"loss": 10.4219,
|
| 23445 |
+
"step": 3341
|
| 23446 |
+
},
|
| 23447 |
+
{
|
| 23448 |
+
"epoch": 0.0003342,
|
| 23449 |
+
"grad_norm": 4.128273010253906,
|
| 23450 |
+
"learning_rate": 3.341e-05,
|
| 23451 |
+
"loss": 10.6797,
|
| 23452 |
+
"step": 3342
|
| 23453 |
+
},
|
| 23454 |
+
{
|
| 23455 |
+
"epoch": 0.0003343,
|
| 23456 |
+
"grad_norm": 4.291218280792236,
|
| 23457 |
+
"learning_rate": 3.342e-05,
|
| 23458 |
+
"loss": 10.8672,
|
| 23459 |
+
"step": 3343
|
| 23460 |
+
},
|
| 23461 |
+
{
|
| 23462 |
+
"epoch": 0.0003344,
|
| 23463 |
+
"grad_norm": 4.879637241363525,
|
| 23464 |
+
"learning_rate": 3.3430000000000003e-05,
|
| 23465 |
+
"loss": 11.1172,
|
| 23466 |
+
"step": 3344
|
| 23467 |
+
},
|
| 23468 |
+
{
|
| 23469 |
+
"epoch": 0.0003345,
|
| 23470 |
+
"grad_norm": 4.181098461151123,
|
| 23471 |
+
"learning_rate": 3.344e-05,
|
| 23472 |
+
"loss": 10.0469,
|
| 23473 |
+
"step": 3345
|
| 23474 |
+
},
|
| 23475 |
+
{
|
| 23476 |
+
"epoch": 0.0003346,
|
| 23477 |
+
"grad_norm": 3.972691774368286,
|
| 23478 |
+
"learning_rate": 3.345e-05,
|
| 23479 |
+
"loss": 10.3828,
|
| 23480 |
+
"step": 3346
|
| 23481 |
+
},
|
| 23482 |
+
{
|
| 23483 |
+
"epoch": 0.0003347,
|
| 23484 |
+
"grad_norm": 4.436202526092529,
|
| 23485 |
+
"learning_rate": 3.346e-05,
|
| 23486 |
+
"loss": 10.8672,
|
| 23487 |
+
"step": 3347
|
| 23488 |
+
},
|
| 23489 |
+
{
|
| 23490 |
+
"epoch": 0.0003348,
|
| 23491 |
+
"grad_norm": 4.7815446853637695,
|
| 23492 |
+
"learning_rate": 3.347e-05,
|
| 23493 |
+
"loss": 11.0781,
|
| 23494 |
+
"step": 3348
|
| 23495 |
+
},
|
| 23496 |
+
{
|
| 23497 |
+
"epoch": 0.0003349,
|
| 23498 |
+
"grad_norm": 3.906643867492676,
|
| 23499 |
+
"learning_rate": 3.348e-05,
|
| 23500 |
+
"loss": 10.2266,
|
| 23501 |
+
"step": 3349
|
| 23502 |
+
},
|
| 23503 |
+
{
|
| 23504 |
+
"epoch": 0.000335,
|
| 23505 |
+
"grad_norm": 3.8373446464538574,
|
| 23506 |
+
"learning_rate": 3.3490000000000006e-05,
|
| 23507 |
+
"loss": 9.6016,
|
| 23508 |
+
"step": 3350
|
| 23509 |
+
},
|
| 23510 |
+
{
|
| 23511 |
+
"epoch": 0.0003351,
|
| 23512 |
+
"grad_norm": 3.990563154220581,
|
| 23513 |
+
"learning_rate": 3.35e-05,
|
| 23514 |
+
"loss": 10.5078,
|
| 23515 |
+
"step": 3351
|
| 23516 |
+
},
|
| 23517 |
+
{
|
| 23518 |
+
"epoch": 0.0003352,
|
| 23519 |
+
"grad_norm": 3.9109091758728027,
|
| 23520 |
+
"learning_rate": 3.3509999999999996e-05,
|
| 23521 |
+
"loss": 11.3516,
|
| 23522 |
+
"step": 3352
|
| 23523 |
+
},
|
| 23524 |
+
{
|
| 23525 |
+
"epoch": 0.0003353,
|
| 23526 |
+
"grad_norm": 4.099769592285156,
|
| 23527 |
+
"learning_rate": 3.3520000000000004e-05,
|
| 23528 |
+
"loss": 10.4609,
|
| 23529 |
+
"step": 3353
|
| 23530 |
+
},
|
| 23531 |
+
{
|
| 23532 |
+
"epoch": 0.0003354,
|
| 23533 |
+
"grad_norm": 3.9826912879943848,
|
| 23534 |
+
"learning_rate": 3.353e-05,
|
| 23535 |
+
"loss": 10.0938,
|
| 23536 |
+
"step": 3354
|
| 23537 |
+
},
|
| 23538 |
+
{
|
| 23539 |
+
"epoch": 0.0003355,
|
| 23540 |
+
"grad_norm": 4.484297752380371,
|
| 23541 |
+
"learning_rate": 3.354e-05,
|
| 23542 |
+
"loss": 10.7344,
|
| 23543 |
+
"step": 3355
|
| 23544 |
+
},
|
| 23545 |
+
{
|
| 23546 |
+
"epoch": 0.0003356,
|
| 23547 |
+
"grad_norm": 4.1387200355529785,
|
| 23548 |
+
"learning_rate": 3.355e-05,
|
| 23549 |
+
"loss": 11.4531,
|
| 23550 |
+
"step": 3356
|
| 23551 |
+
},
|
| 23552 |
+
{
|
| 23553 |
+
"epoch": 0.0003357,
|
| 23554 |
+
"grad_norm": 3.7992279529571533,
|
| 23555 |
+
"learning_rate": 3.356e-05,
|
| 23556 |
+
"loss": 9.5625,
|
| 23557 |
+
"step": 3357
|
| 23558 |
+
},
|
| 23559 |
+
{
|
| 23560 |
+
"epoch": 0.0003358,
|
| 23561 |
+
"grad_norm": 4.342659950256348,
|
| 23562 |
+
"learning_rate": 3.357e-05,
|
| 23563 |
+
"loss": 11.2422,
|
| 23564 |
+
"step": 3358
|
| 23565 |
+
},
|
| 23566 |
+
{
|
| 23567 |
+
"epoch": 0.0003359,
|
| 23568 |
+
"grad_norm": 7.785485744476318,
|
| 23569 |
+
"learning_rate": 3.358e-05,
|
| 23570 |
+
"loss": 11.9844,
|
| 23571 |
+
"step": 3359
|
| 23572 |
+
},
|
| 23573 |
+
{
|
| 23574 |
+
"epoch": 0.000336,
|
| 23575 |
+
"grad_norm": 3.9241621494293213,
|
| 23576 |
+
"learning_rate": 3.359e-05,
|
| 23577 |
+
"loss": 10.1875,
|
| 23578 |
+
"step": 3360
|
| 23579 |
+
},
|
| 23580 |
+
{
|
| 23581 |
+
"epoch": 0.0003361,
|
| 23582 |
+
"grad_norm": 4.481233596801758,
|
| 23583 |
+
"learning_rate": 3.3600000000000004e-05,
|
| 23584 |
+
"loss": 11.1797,
|
| 23585 |
+
"step": 3361
|
| 23586 |
+
},
|
| 23587 |
+
{
|
| 23588 |
+
"epoch": 0.0003362,
|
| 23589 |
+
"grad_norm": 4.896907329559326,
|
| 23590 |
+
"learning_rate": 3.361e-05,
|
| 23591 |
+
"loss": 11.3281,
|
| 23592 |
+
"step": 3362
|
| 23593 |
+
},
|
| 23594 |
+
{
|
| 23595 |
+
"epoch": 0.0003363,
|
| 23596 |
+
"grad_norm": 4.73055362701416,
|
| 23597 |
+
"learning_rate": 3.362e-05,
|
| 23598 |
+
"loss": 11.8047,
|
| 23599 |
+
"step": 3363
|
| 23600 |
+
},
|
| 23601 |
+
{
|
| 23602 |
+
"epoch": 0.0003364,
|
| 23603 |
+
"grad_norm": 4.517014026641846,
|
| 23604 |
+
"learning_rate": 3.363e-05,
|
| 23605 |
+
"loss": 11.5469,
|
| 23606 |
+
"step": 3364
|
| 23607 |
+
},
|
| 23608 |
+
{
|
| 23609 |
+
"epoch": 0.0003365,
|
| 23610 |
+
"grad_norm": 4.136028289794922,
|
| 23611 |
+
"learning_rate": 3.364e-05,
|
| 23612 |
+
"loss": 10.5234,
|
| 23613 |
+
"step": 3365
|
| 23614 |
+
},
|
| 23615 |
+
{
|
| 23616 |
+
"epoch": 0.0003366,
|
| 23617 |
+
"grad_norm": 3.85905385017395,
|
| 23618 |
+
"learning_rate": 3.365e-05,
|
| 23619 |
+
"loss": 9.8359,
|
| 23620 |
+
"step": 3366
|
| 23621 |
+
},
|
| 23622 |
+
{
|
| 23623 |
+
"epoch": 0.0003367,
|
| 23624 |
+
"grad_norm": 4.074779510498047,
|
| 23625 |
+
"learning_rate": 3.3660000000000006e-05,
|
| 23626 |
+
"loss": 10.7578,
|
| 23627 |
+
"step": 3367
|
| 23628 |
+
},
|
| 23629 |
+
{
|
| 23630 |
+
"epoch": 0.0003368,
|
| 23631 |
+
"grad_norm": 4.239984035491943,
|
| 23632 |
+
"learning_rate": 3.367e-05,
|
| 23633 |
+
"loss": 10.25,
|
| 23634 |
+
"step": 3368
|
| 23635 |
+
},
|
| 23636 |
+
{
|
| 23637 |
+
"epoch": 0.0003369,
|
| 23638 |
+
"grad_norm": 5.920673847198486,
|
| 23639 |
+
"learning_rate": 3.3679999999999996e-05,
|
| 23640 |
+
"loss": 11.5547,
|
| 23641 |
+
"step": 3369
|
| 23642 |
+
},
|
| 23643 |
+
{
|
| 23644 |
+
"epoch": 0.000337,
|
| 23645 |
+
"grad_norm": 3.863361120223999,
|
| 23646 |
+
"learning_rate": 3.3690000000000004e-05,
|
| 23647 |
+
"loss": 9.875,
|
| 23648 |
+
"step": 3370
|
| 23649 |
+
},
|
| 23650 |
+
{
|
| 23651 |
+
"epoch": 0.0003371,
|
| 23652 |
+
"grad_norm": 4.209554672241211,
|
| 23653 |
+
"learning_rate": 3.37e-05,
|
| 23654 |
+
"loss": 10.3438,
|
| 23655 |
+
"step": 3371
|
| 23656 |
+
},
|
| 23657 |
+
{
|
| 23658 |
+
"epoch": 0.0003372,
|
| 23659 |
+
"grad_norm": 4.092309474945068,
|
| 23660 |
+
"learning_rate": 3.371e-05,
|
| 23661 |
+
"loss": 10.0781,
|
| 23662 |
+
"step": 3372
|
| 23663 |
+
},
|
| 23664 |
+
{
|
| 23665 |
+
"epoch": 0.0003373,
|
| 23666 |
+
"grad_norm": 4.802425384521484,
|
| 23667 |
+
"learning_rate": 3.372e-05,
|
| 23668 |
+
"loss": 10.8828,
|
| 23669 |
+
"step": 3373
|
| 23670 |
+
},
|
| 23671 |
+
{
|
| 23672 |
+
"epoch": 0.0003374,
|
| 23673 |
+
"grad_norm": 4.862827777862549,
|
| 23674 |
+
"learning_rate": 3.373e-05,
|
| 23675 |
+
"loss": 12.1875,
|
| 23676 |
+
"step": 3374
|
| 23677 |
+
},
|
| 23678 |
+
{
|
| 23679 |
+
"epoch": 0.0003375,
|
| 23680 |
+
"grad_norm": 4.167935848236084,
|
| 23681 |
+
"learning_rate": 3.374e-05,
|
| 23682 |
+
"loss": 10.1484,
|
| 23683 |
+
"step": 3375
|
| 23684 |
+
},
|
| 23685 |
+
{
|
| 23686 |
+
"epoch": 0.0003376,
|
| 23687 |
+
"grad_norm": 4.243851661682129,
|
| 23688 |
+
"learning_rate": 3.375e-05,
|
| 23689 |
+
"loss": 10.5078,
|
| 23690 |
+
"step": 3376
|
| 23691 |
+
},
|
| 23692 |
+
{
|
| 23693 |
+
"epoch": 0.0003377,
|
| 23694 |
+
"grad_norm": 4.010030269622803,
|
| 23695 |
+
"learning_rate": 3.376e-05,
|
| 23696 |
+
"loss": 10.0938,
|
| 23697 |
+
"step": 3377
|
| 23698 |
+
},
|
| 23699 |
+
{
|
| 23700 |
+
"epoch": 0.0003378,
|
| 23701 |
+
"grad_norm": 3.8439948558807373,
|
| 23702 |
+
"learning_rate": 3.3770000000000004e-05,
|
| 23703 |
+
"loss": 10.1328,
|
| 23704 |
+
"step": 3378
|
| 23705 |
+
},
|
| 23706 |
+
{
|
| 23707 |
+
"epoch": 0.0003379,
|
| 23708 |
+
"grad_norm": 3.887462615966797,
|
| 23709 |
+
"learning_rate": 3.378e-05,
|
| 23710 |
+
"loss": 10.3516,
|
| 23711 |
+
"step": 3379
|
| 23712 |
+
},
|
| 23713 |
+
{
|
| 23714 |
+
"epoch": 0.000338,
|
| 23715 |
+
"grad_norm": 3.718141555786133,
|
| 23716 |
+
"learning_rate": 3.379e-05,
|
| 23717 |
+
"loss": 10.2578,
|
| 23718 |
+
"step": 3380
|
| 23719 |
+
},
|
| 23720 |
+
{
|
| 23721 |
+
"epoch": 0.0003381,
|
| 23722 |
+
"grad_norm": 4.037032127380371,
|
| 23723 |
+
"learning_rate": 3.38e-05,
|
| 23724 |
+
"loss": 10.2109,
|
| 23725 |
+
"step": 3381
|
| 23726 |
+
},
|
| 23727 |
+
{
|
| 23728 |
+
"epoch": 0.0003382,
|
| 23729 |
+
"grad_norm": 4.068556785583496,
|
| 23730 |
+
"learning_rate": 3.381e-05,
|
| 23731 |
+
"loss": 10.3984,
|
| 23732 |
+
"step": 3382
|
| 23733 |
+
},
|
| 23734 |
+
{
|
| 23735 |
+
"epoch": 0.0003383,
|
| 23736 |
+
"grad_norm": 4.041705131530762,
|
| 23737 |
+
"learning_rate": 3.382e-05,
|
| 23738 |
+
"loss": 10.1172,
|
| 23739 |
+
"step": 3383
|
| 23740 |
+
},
|
| 23741 |
+
{
|
| 23742 |
+
"epoch": 0.0003384,
|
| 23743 |
+
"grad_norm": 4.55108642578125,
|
| 23744 |
+
"learning_rate": 3.3830000000000006e-05,
|
| 23745 |
+
"loss": 11.0312,
|
| 23746 |
+
"step": 3384
|
| 23747 |
+
},
|
| 23748 |
+
{
|
| 23749 |
+
"epoch": 0.0003385,
|
| 23750 |
+
"grad_norm": 4.181764602661133,
|
| 23751 |
+
"learning_rate": 3.384e-05,
|
| 23752 |
+
"loss": 10.2734,
|
| 23753 |
+
"step": 3385
|
| 23754 |
+
},
|
| 23755 |
+
{
|
| 23756 |
+
"epoch": 0.0003386,
|
| 23757 |
+
"grad_norm": 3.910088539123535,
|
| 23758 |
+
"learning_rate": 3.3849999999999996e-05,
|
| 23759 |
+
"loss": 9.8125,
|
| 23760 |
+
"step": 3386
|
| 23761 |
+
},
|
| 23762 |
+
{
|
| 23763 |
+
"epoch": 0.0003387,
|
| 23764 |
+
"grad_norm": 4.057553768157959,
|
| 23765 |
+
"learning_rate": 3.3860000000000004e-05,
|
| 23766 |
+
"loss": 10.2891,
|
| 23767 |
+
"step": 3387
|
| 23768 |
+
},
|
| 23769 |
+
{
|
| 23770 |
+
"epoch": 0.0003388,
|
| 23771 |
+
"grad_norm": 4.039722919464111,
|
| 23772 |
+
"learning_rate": 3.387e-05,
|
| 23773 |
+
"loss": 10.2891,
|
| 23774 |
+
"step": 3388
|
| 23775 |
+
},
|
| 23776 |
+
{
|
| 23777 |
+
"epoch": 0.0003389,
|
| 23778 |
+
"grad_norm": 3.8159613609313965,
|
| 23779 |
+
"learning_rate": 3.388e-05,
|
| 23780 |
+
"loss": 10.1836,
|
| 23781 |
+
"step": 3389
|
| 23782 |
+
},
|
| 23783 |
+
{
|
| 23784 |
+
"epoch": 0.000339,
|
| 23785 |
+
"grad_norm": 3.9674181938171387,
|
| 23786 |
+
"learning_rate": 3.389e-05,
|
| 23787 |
+
"loss": 10.0938,
|
| 23788 |
+
"step": 3390
|
| 23789 |
+
},
|
| 23790 |
+
{
|
| 23791 |
+
"epoch": 0.0003391,
|
| 23792 |
+
"grad_norm": 3.9209303855895996,
|
| 23793 |
+
"learning_rate": 3.39e-05,
|
| 23794 |
+
"loss": 10.3672,
|
| 23795 |
+
"step": 3391
|
| 23796 |
+
},
|
| 23797 |
+
{
|
| 23798 |
+
"epoch": 0.0003392,
|
| 23799 |
+
"grad_norm": 4.239545822143555,
|
| 23800 |
+
"learning_rate": 3.391e-05,
|
| 23801 |
+
"loss": 9.8203,
|
| 23802 |
+
"step": 3392
|
| 23803 |
+
},
|
| 23804 |
+
{
|
| 23805 |
+
"epoch": 0.0003393,
|
| 23806 |
+
"grad_norm": 4.004146575927734,
|
| 23807 |
+
"learning_rate": 3.392e-05,
|
| 23808 |
+
"loss": 9.7422,
|
| 23809 |
+
"step": 3393
|
| 23810 |
+
},
|
| 23811 |
+
{
|
| 23812 |
+
"epoch": 0.0003394,
|
| 23813 |
+
"grad_norm": 4.08314847946167,
|
| 23814 |
+
"learning_rate": 3.393e-05,
|
| 23815 |
+
"loss": 10.8047,
|
| 23816 |
+
"step": 3394
|
| 23817 |
+
},
|
| 23818 |
+
{
|
| 23819 |
+
"epoch": 0.0003395,
|
| 23820 |
+
"grad_norm": 3.988220453262329,
|
| 23821 |
+
"learning_rate": 3.3940000000000004e-05,
|
| 23822 |
+
"loss": 10.3828,
|
| 23823 |
+
"step": 3395
|
| 23824 |
+
},
|
| 23825 |
+
{
|
| 23826 |
+
"epoch": 0.0003396,
|
| 23827 |
+
"grad_norm": 4.089816570281982,
|
| 23828 |
+
"learning_rate": 3.395e-05,
|
| 23829 |
+
"loss": 10.0664,
|
| 23830 |
+
"step": 3396
|
| 23831 |
+
},
|
| 23832 |
+
{
|
| 23833 |
+
"epoch": 0.0003397,
|
| 23834 |
+
"grad_norm": 4.212343215942383,
|
| 23835 |
+
"learning_rate": 3.396e-05,
|
| 23836 |
+
"loss": 10.5234,
|
| 23837 |
+
"step": 3397
|
| 23838 |
+
},
|
| 23839 |
+
{
|
| 23840 |
+
"epoch": 0.0003398,
|
| 23841 |
+
"grad_norm": 3.9881434440612793,
|
| 23842 |
+
"learning_rate": 3.397e-05,
|
| 23843 |
+
"loss": 10.4922,
|
| 23844 |
+
"step": 3398
|
| 23845 |
+
},
|
| 23846 |
+
{
|
| 23847 |
+
"epoch": 0.0003399,
|
| 23848 |
+
"grad_norm": 5.337203502655029,
|
| 23849 |
+
"learning_rate": 3.398e-05,
|
| 23850 |
+
"loss": 11.7422,
|
| 23851 |
+
"step": 3399
|
| 23852 |
+
},
|
| 23853 |
+
{
|
| 23854 |
+
"epoch": 0.00034,
|
| 23855 |
+
"grad_norm": 4.544561386108398,
|
| 23856 |
+
"learning_rate": 3.399e-05,
|
| 23857 |
+
"loss": 11.1953,
|
| 23858 |
+
"step": 3400
|
| 23859 |
+
},
|
| 23860 |
+
{
|
| 23861 |
+
"epoch": 0.0003401,
|
| 23862 |
+
"grad_norm": 4.235530376434326,
|
| 23863 |
+
"learning_rate": 3.4000000000000007e-05,
|
| 23864 |
+
"loss": 10.4375,
|
| 23865 |
+
"step": 3401
|
| 23866 |
+
},
|
| 23867 |
+
{
|
| 23868 |
+
"epoch": 0.0003402,
|
| 23869 |
+
"grad_norm": 3.9115917682647705,
|
| 23870 |
+
"learning_rate": 3.401e-05,
|
| 23871 |
+
"loss": 9.6172,
|
| 23872 |
+
"step": 3402
|
| 23873 |
+
},
|
| 23874 |
+
{
|
| 23875 |
+
"epoch": 0.0003403,
|
| 23876 |
+
"grad_norm": 3.863034963607788,
|
| 23877 |
+
"learning_rate": 3.4019999999999996e-05,
|
| 23878 |
+
"loss": 9.2422,
|
| 23879 |
+
"step": 3403
|
| 23880 |
+
},
|
| 23881 |
+
{
|
| 23882 |
+
"epoch": 0.0003404,
|
| 23883 |
+
"grad_norm": 3.962681293487549,
|
| 23884 |
+
"learning_rate": 3.4030000000000005e-05,
|
| 23885 |
+
"loss": 10.0391,
|
| 23886 |
+
"step": 3404
|
| 23887 |
+
},
|
| 23888 |
+
{
|
| 23889 |
+
"epoch": 0.0003405,
|
| 23890 |
+
"grad_norm": 3.9435994625091553,
|
| 23891 |
+
"learning_rate": 3.404e-05,
|
| 23892 |
+
"loss": 9.75,
|
| 23893 |
+
"step": 3405
|
| 23894 |
+
},
|
| 23895 |
+
{
|
| 23896 |
+
"epoch": 0.0003406,
|
| 23897 |
+
"grad_norm": 4.049876689910889,
|
| 23898 |
+
"learning_rate": 3.405e-05,
|
| 23899 |
+
"loss": 9.7734,
|
| 23900 |
+
"step": 3406
|
| 23901 |
+
},
|
| 23902 |
+
{
|
| 23903 |
+
"epoch": 0.0003407,
|
| 23904 |
+
"grad_norm": 3.991595983505249,
|
| 23905 |
+
"learning_rate": 3.406e-05,
|
| 23906 |
+
"loss": 9.8672,
|
| 23907 |
+
"step": 3407
|
| 23908 |
+
},
|
| 23909 |
+
{
|
| 23910 |
+
"epoch": 0.0003408,
|
| 23911 |
+
"grad_norm": 4.0078325271606445,
|
| 23912 |
+
"learning_rate": 3.407e-05,
|
| 23913 |
+
"loss": 9.6797,
|
| 23914 |
+
"step": 3408
|
| 23915 |
+
},
|
| 23916 |
+
{
|
| 23917 |
+
"epoch": 0.0003409,
|
| 23918 |
+
"grad_norm": 3.9059255123138428,
|
| 23919 |
+
"learning_rate": 3.408e-05,
|
| 23920 |
+
"loss": 10.4531,
|
| 23921 |
+
"step": 3409
|
| 23922 |
+
},
|
| 23923 |
+
{
|
| 23924 |
+
"epoch": 0.000341,
|
| 23925 |
+
"grad_norm": 3.985442638397217,
|
| 23926 |
+
"learning_rate": 3.409e-05,
|
| 23927 |
+
"loss": 9.875,
|
| 23928 |
+
"step": 3410
|
| 23929 |
+
},
|
| 23930 |
+
{
|
| 23931 |
+
"epoch": 0.0003411,
|
| 23932 |
+
"grad_norm": 4.073984622955322,
|
| 23933 |
+
"learning_rate": 3.41e-05,
|
| 23934 |
+
"loss": 9.6016,
|
| 23935 |
+
"step": 3411
|
| 23936 |
+
},
|
| 23937 |
+
{
|
| 23938 |
+
"epoch": 0.0003412,
|
| 23939 |
+
"grad_norm": 3.815159797668457,
|
| 23940 |
+
"learning_rate": 3.4110000000000004e-05,
|
| 23941 |
+
"loss": 9.5469,
|
| 23942 |
+
"step": 3412
|
| 23943 |
+
},
|
| 23944 |
+
{
|
| 23945 |
+
"epoch": 0.0003413,
|
| 23946 |
+
"grad_norm": 3.9025957584381104,
|
| 23947 |
+
"learning_rate": 3.412e-05,
|
| 23948 |
+
"loss": 9.8906,
|
| 23949 |
+
"step": 3413
|
| 23950 |
+
},
|
| 23951 |
+
{
|
| 23952 |
+
"epoch": 0.0003414,
|
| 23953 |
+
"grad_norm": 4.264541149139404,
|
| 23954 |
+
"learning_rate": 3.413e-05,
|
| 23955 |
+
"loss": 9.8906,
|
| 23956 |
+
"step": 3414
|
| 23957 |
+
},
|
| 23958 |
+
{
|
| 23959 |
+
"epoch": 0.0003415,
|
| 23960 |
+
"grad_norm": 4.153369903564453,
|
| 23961 |
+
"learning_rate": 3.414e-05,
|
| 23962 |
+
"loss": 9.9922,
|
| 23963 |
+
"step": 3415
|
| 23964 |
+
},
|
| 23965 |
+
{
|
| 23966 |
+
"epoch": 0.0003416,
|
| 23967 |
+
"grad_norm": 4.489535331726074,
|
| 23968 |
+
"learning_rate": 3.415e-05,
|
| 23969 |
+
"loss": 11.6719,
|
| 23970 |
+
"step": 3416
|
| 23971 |
+
},
|
| 23972 |
+
{
|
| 23973 |
+
"epoch": 0.0003417,
|
| 23974 |
+
"grad_norm": 4.108375549316406,
|
| 23975 |
+
"learning_rate": 3.416e-05,
|
| 23976 |
+
"loss": 10.2109,
|
| 23977 |
+
"step": 3417
|
| 23978 |
+
},
|
| 23979 |
+
{
|
| 23980 |
+
"epoch": 0.0003418,
|
| 23981 |
+
"grad_norm": 4.414920806884766,
|
| 23982 |
+
"learning_rate": 3.417e-05,
|
| 23983 |
+
"loss": 10.1953,
|
| 23984 |
+
"step": 3418
|
| 23985 |
+
},
|
| 23986 |
+
{
|
| 23987 |
+
"epoch": 0.0003419,
|
| 23988 |
+
"grad_norm": 4.087825775146484,
|
| 23989 |
+
"learning_rate": 3.418e-05,
|
| 23990 |
+
"loss": 9.6406,
|
| 23991 |
+
"step": 3419
|
| 23992 |
+
},
|
| 23993 |
+
{
|
| 23994 |
+
"epoch": 0.000342,
|
| 23995 |
+
"grad_norm": 4.137132167816162,
|
| 23996 |
+
"learning_rate": 3.4189999999999996e-05,
|
| 23997 |
+
"loss": 10.1094,
|
| 23998 |
+
"step": 3420
|
| 23999 |
+
},
|
| 24000 |
+
{
|
| 24001 |
+
"epoch": 0.0003421,
|
| 24002 |
+
"grad_norm": 4.568292617797852,
|
| 24003 |
+
"learning_rate": 3.4200000000000005e-05,
|
| 24004 |
+
"loss": 11.9609,
|
| 24005 |
+
"step": 3421
|
| 24006 |
+
},
|
| 24007 |
+
{
|
| 24008 |
+
"epoch": 0.0003422,
|
| 24009 |
+
"grad_norm": 3.8219153881073,
|
| 24010 |
+
"learning_rate": 3.421e-05,
|
| 24011 |
+
"loss": 8.9688,
|
| 24012 |
+
"step": 3422
|
| 24013 |
+
},
|
| 24014 |
+
{
|
| 24015 |
+
"epoch": 0.0003423,
|
| 24016 |
+
"grad_norm": 4.35713529586792,
|
| 24017 |
+
"learning_rate": 3.422e-05,
|
| 24018 |
+
"loss": 10.0078,
|
| 24019 |
+
"step": 3423
|
| 24020 |
+
},
|
| 24021 |
+
{
|
| 24022 |
+
"epoch": 0.0003424,
|
| 24023 |
+
"grad_norm": 4.1804423332214355,
|
| 24024 |
+
"learning_rate": 3.423e-05,
|
| 24025 |
+
"loss": 9.9375,
|
| 24026 |
+
"step": 3424
|
| 24027 |
+
},
|
| 24028 |
+
{
|
| 24029 |
+
"epoch": 0.0003425,
|
| 24030 |
+
"grad_norm": 4.542559623718262,
|
| 24031 |
+
"learning_rate": 3.424e-05,
|
| 24032 |
+
"loss": 10.3906,
|
| 24033 |
+
"step": 3425
|
| 24034 |
+
},
|
| 24035 |
+
{
|
| 24036 |
+
"epoch": 0.0003426,
|
| 24037 |
+
"grad_norm": 4.20182466506958,
|
| 24038 |
+
"learning_rate": 3.425e-05,
|
| 24039 |
+
"loss": 10.0703,
|
| 24040 |
+
"step": 3426
|
| 24041 |
+
},
|
| 24042 |
+
{
|
| 24043 |
+
"epoch": 0.0003427,
|
| 24044 |
+
"grad_norm": 3.8954553604125977,
|
| 24045 |
+
"learning_rate": 3.426e-05,
|
| 24046 |
+
"loss": 10.0234,
|
| 24047 |
+
"step": 3427
|
| 24048 |
+
},
|
| 24049 |
+
{
|
| 24050 |
+
"epoch": 0.0003428,
|
| 24051 |
+
"grad_norm": 3.8842320442199707,
|
| 24052 |
+
"learning_rate": 3.427e-05,
|
| 24053 |
+
"loss": 9.5703,
|
| 24054 |
+
"step": 3428
|
| 24055 |
+
},
|
| 24056 |
+
{
|
| 24057 |
+
"epoch": 0.0003429,
|
| 24058 |
+
"grad_norm": 3.9736902713775635,
|
| 24059 |
+
"learning_rate": 3.428e-05,
|
| 24060 |
+
"loss": 9.7656,
|
| 24061 |
+
"step": 3429
|
| 24062 |
+
},
|
| 24063 |
+
{
|
| 24064 |
+
"epoch": 0.000343,
|
| 24065 |
+
"grad_norm": 4.112651824951172,
|
| 24066 |
+
"learning_rate": 3.429e-05,
|
| 24067 |
+
"loss": 10.3047,
|
| 24068 |
+
"step": 3430
|
| 24069 |
+
},
|
| 24070 |
+
{
|
| 24071 |
+
"epoch": 0.0003431,
|
| 24072 |
+
"grad_norm": 4.080387115478516,
|
| 24073 |
+
"learning_rate": 3.43e-05,
|
| 24074 |
+
"loss": 10.4062,
|
| 24075 |
+
"step": 3431
|
| 24076 |
+
},
|
| 24077 |
+
{
|
| 24078 |
+
"epoch": 0.0003432,
|
| 24079 |
+
"grad_norm": 4.10958194732666,
|
| 24080 |
+
"learning_rate": 3.431e-05,
|
| 24081 |
+
"loss": 9.8359,
|
| 24082 |
+
"step": 3432
|
| 24083 |
+
},
|
| 24084 |
+
{
|
| 24085 |
+
"epoch": 0.0003433,
|
| 24086 |
+
"grad_norm": 3.89727783203125,
|
| 24087 |
+
"learning_rate": 3.4320000000000003e-05,
|
| 24088 |
+
"loss": 9.3984,
|
| 24089 |
+
"step": 3433
|
| 24090 |
+
},
|
| 24091 |
+
{
|
| 24092 |
+
"epoch": 0.0003434,
|
| 24093 |
+
"grad_norm": 3.99847149848938,
|
| 24094 |
+
"learning_rate": 3.433e-05,
|
| 24095 |
+
"loss": 9.6953,
|
| 24096 |
+
"step": 3434
|
| 24097 |
+
},
|
| 24098 |
+
{
|
| 24099 |
+
"epoch": 0.0003435,
|
| 24100 |
+
"grad_norm": 4.055912017822266,
|
| 24101 |
+
"learning_rate": 3.434e-05,
|
| 24102 |
+
"loss": 9.5859,
|
| 24103 |
+
"step": 3435
|
| 24104 |
+
},
|
| 24105 |
+
{
|
| 24106 |
+
"epoch": 0.0003436,
|
| 24107 |
+
"grad_norm": 4.387887001037598,
|
| 24108 |
+
"learning_rate": 3.435e-05,
|
| 24109 |
+
"loss": 10.9609,
|
| 24110 |
+
"step": 3436
|
| 24111 |
+
},
|
| 24112 |
+
{
|
| 24113 |
+
"epoch": 0.0003437,
|
| 24114 |
+
"grad_norm": 4.093634605407715,
|
| 24115 |
+
"learning_rate": 3.4359999999999996e-05,
|
| 24116 |
+
"loss": 9.4062,
|
| 24117 |
+
"step": 3437
|
| 24118 |
+
},
|
| 24119 |
+
{
|
| 24120 |
+
"epoch": 0.0003438,
|
| 24121 |
+
"grad_norm": 4.745052337646484,
|
| 24122 |
+
"learning_rate": 3.4370000000000005e-05,
|
| 24123 |
+
"loss": 10.0938,
|
| 24124 |
+
"step": 3438
|
| 24125 |
+
},
|
| 24126 |
+
{
|
| 24127 |
+
"epoch": 0.0003439,
|
| 24128 |
+
"grad_norm": 3.901541233062744,
|
| 24129 |
+
"learning_rate": 3.438e-05,
|
| 24130 |
+
"loss": 9.9844,
|
| 24131 |
+
"step": 3439
|
| 24132 |
+
},
|
| 24133 |
+
{
|
| 24134 |
+
"epoch": 0.000344,
|
| 24135 |
+
"grad_norm": 3.945960283279419,
|
| 24136 |
+
"learning_rate": 3.439e-05,
|
| 24137 |
+
"loss": 9.7578,
|
| 24138 |
+
"step": 3440
|
| 24139 |
+
},
|
| 24140 |
+
{
|
| 24141 |
+
"epoch": 0.0003441,
|
| 24142 |
+
"grad_norm": 3.955289840698242,
|
| 24143 |
+
"learning_rate": 3.44e-05,
|
| 24144 |
+
"loss": 9.8203,
|
| 24145 |
+
"step": 3441
|
| 24146 |
+
},
|
| 24147 |
+
{
|
| 24148 |
+
"epoch": 0.0003442,
|
| 24149 |
+
"grad_norm": 3.791595220565796,
|
| 24150 |
+
"learning_rate": 3.441e-05,
|
| 24151 |
+
"loss": 9.4688,
|
| 24152 |
+
"step": 3442
|
| 24153 |
+
},
|
| 24154 |
+
{
|
| 24155 |
+
"epoch": 0.0003443,
|
| 24156 |
+
"grad_norm": 4.786859512329102,
|
| 24157 |
+
"learning_rate": 3.442e-05,
|
| 24158 |
+
"loss": 10.7734,
|
| 24159 |
+
"step": 3443
|
| 24160 |
+
},
|
| 24161 |
+
{
|
| 24162 |
+
"epoch": 0.0003444,
|
| 24163 |
+
"grad_norm": 3.852816104888916,
|
| 24164 |
+
"learning_rate": 3.443e-05,
|
| 24165 |
+
"loss": 9.4609,
|
| 24166 |
+
"step": 3444
|
| 24167 |
+
},
|
| 24168 |
+
{
|
| 24169 |
+
"epoch": 0.0003445,
|
| 24170 |
+
"grad_norm": 3.9114298820495605,
|
| 24171 |
+
"learning_rate": 3.444e-05,
|
| 24172 |
+
"loss": 10.0781,
|
| 24173 |
+
"step": 3445
|
| 24174 |
+
},
|
| 24175 |
+
{
|
| 24176 |
+
"epoch": 0.0003446,
|
| 24177 |
+
"grad_norm": 3.7254295349121094,
|
| 24178 |
+
"learning_rate": 3.445e-05,
|
| 24179 |
+
"loss": 9.8594,
|
| 24180 |
+
"step": 3446
|
| 24181 |
+
},
|
| 24182 |
+
{
|
| 24183 |
+
"epoch": 0.0003447,
|
| 24184 |
+
"grad_norm": 4.128079891204834,
|
| 24185 |
+
"learning_rate": 3.446e-05,
|
| 24186 |
+
"loss": 9.7188,
|
| 24187 |
+
"step": 3447
|
| 24188 |
+
},
|
| 24189 |
+
{
|
| 24190 |
+
"epoch": 0.0003448,
|
| 24191 |
+
"grad_norm": 3.8521411418914795,
|
| 24192 |
+
"learning_rate": 3.447e-05,
|
| 24193 |
+
"loss": 10.0938,
|
| 24194 |
+
"step": 3448
|
| 24195 |
+
},
|
| 24196 |
+
{
|
| 24197 |
+
"epoch": 0.0003449,
|
| 24198 |
+
"grad_norm": 3.7841033935546875,
|
| 24199 |
+
"learning_rate": 3.448e-05,
|
| 24200 |
+
"loss": 9.2031,
|
| 24201 |
+
"step": 3449
|
| 24202 |
+
},
|
| 24203 |
+
{
|
| 24204 |
+
"epoch": 0.000345,
|
| 24205 |
+
"grad_norm": 3.6274397373199463,
|
| 24206 |
+
"learning_rate": 3.4490000000000004e-05,
|
| 24207 |
+
"loss": 8.9766,
|
| 24208 |
+
"step": 3450
|
| 24209 |
+
},
|
| 24210 |
+
{
|
| 24211 |
+
"epoch": 0.0003451,
|
| 24212 |
+
"grad_norm": 3.8506691455841064,
|
| 24213 |
+
"learning_rate": 3.45e-05,
|
| 24214 |
+
"loss": 9.6016,
|
| 24215 |
+
"step": 3451
|
| 24216 |
+
},
|
| 24217 |
+
{
|
| 24218 |
+
"epoch": 0.0003452,
|
| 24219 |
+
"grad_norm": 4.06758975982666,
|
| 24220 |
+
"learning_rate": 3.451e-05,
|
| 24221 |
+
"loss": 9.6484,
|
| 24222 |
+
"step": 3452
|
| 24223 |
+
},
|
| 24224 |
+
{
|
| 24225 |
+
"epoch": 0.0003453,
|
| 24226 |
+
"grad_norm": 4.5061798095703125,
|
| 24227 |
+
"learning_rate": 3.452e-05,
|
| 24228 |
+
"loss": 10.5938,
|
| 24229 |
+
"step": 3453
|
| 24230 |
+
},
|
| 24231 |
+
{
|
| 24232 |
+
"epoch": 0.0003454,
|
| 24233 |
+
"grad_norm": 4.398696422576904,
|
| 24234 |
+
"learning_rate": 3.4529999999999996e-05,
|
| 24235 |
+
"loss": 10.4062,
|
| 24236 |
+
"step": 3454
|
| 24237 |
+
},
|
| 24238 |
+
{
|
| 24239 |
+
"epoch": 0.0003455,
|
| 24240 |
+
"grad_norm": 4.197793483734131,
|
| 24241 |
+
"learning_rate": 3.4540000000000005e-05,
|
| 24242 |
+
"loss": 10.7188,
|
| 24243 |
+
"step": 3455
|
| 24244 |
+
},
|
| 24245 |
+
{
|
| 24246 |
+
"epoch": 0.0003456,
|
| 24247 |
+
"grad_norm": 4.172128677368164,
|
| 24248 |
+
"learning_rate": 3.455e-05,
|
| 24249 |
+
"loss": 9.4766,
|
| 24250 |
+
"step": 3456
|
| 24251 |
+
},
|
| 24252 |
+
{
|
| 24253 |
+
"epoch": 0.0003457,
|
| 24254 |
+
"grad_norm": 4.2549943923950195,
|
| 24255 |
+
"learning_rate": 3.456e-05,
|
| 24256 |
+
"loss": 9.6719,
|
| 24257 |
+
"step": 3457
|
| 24258 |
+
},
|
| 24259 |
+
{
|
| 24260 |
+
"epoch": 0.0003458,
|
| 24261 |
+
"grad_norm": 4.00117301940918,
|
| 24262 |
+
"learning_rate": 3.457e-05,
|
| 24263 |
+
"loss": 9.3828,
|
| 24264 |
+
"step": 3458
|
| 24265 |
+
},
|
| 24266 |
+
{
|
| 24267 |
+
"epoch": 0.0003459,
|
| 24268 |
+
"grad_norm": 4.461249828338623,
|
| 24269 |
+
"learning_rate": 3.458e-05,
|
| 24270 |
+
"loss": 9.9141,
|
| 24271 |
+
"step": 3459
|
| 24272 |
+
},
|
| 24273 |
+
{
|
| 24274 |
+
"epoch": 0.000346,
|
| 24275 |
+
"grad_norm": 4.20103120803833,
|
| 24276 |
+
"learning_rate": 3.459e-05,
|
| 24277 |
+
"loss": 9.6953,
|
| 24278 |
+
"step": 3460
|
| 24279 |
+
},
|
| 24280 |
+
{
|
| 24281 |
+
"epoch": 0.0003461,
|
| 24282 |
+
"grad_norm": 4.055276393890381,
|
| 24283 |
+
"learning_rate": 3.46e-05,
|
| 24284 |
+
"loss": 9.7344,
|
| 24285 |
+
"step": 3461
|
| 24286 |
+
},
|
| 24287 |
+
{
|
| 24288 |
+
"epoch": 0.0003462,
|
| 24289 |
+
"grad_norm": 3.7303593158721924,
|
| 24290 |
+
"learning_rate": 3.461e-05,
|
| 24291 |
+
"loss": 9.0312,
|
| 24292 |
+
"step": 3462
|
| 24293 |
+
},
|
| 24294 |
+
{
|
| 24295 |
+
"epoch": 0.0003463,
|
| 24296 |
+
"grad_norm": 4.2501420974731445,
|
| 24297 |
+
"learning_rate": 3.462e-05,
|
| 24298 |
+
"loss": 9.4922,
|
| 24299 |
+
"step": 3463
|
| 24300 |
+
},
|
| 24301 |
+
{
|
| 24302 |
+
"epoch": 0.0003464,
|
| 24303 |
+
"grad_norm": 4.156157493591309,
|
| 24304 |
+
"learning_rate": 3.463e-05,
|
| 24305 |
+
"loss": 10.5234,
|
| 24306 |
+
"step": 3464
|
| 24307 |
+
},
|
| 24308 |
+
{
|
| 24309 |
+
"epoch": 0.0003465,
|
| 24310 |
+
"grad_norm": 3.9639360904693604,
|
| 24311 |
+
"learning_rate": 3.464e-05,
|
| 24312 |
+
"loss": 9.7422,
|
| 24313 |
+
"step": 3465
|
| 24314 |
+
},
|
| 24315 |
+
{
|
| 24316 |
+
"epoch": 0.0003466,
|
| 24317 |
+
"grad_norm": 4.065398216247559,
|
| 24318 |
+
"learning_rate": 3.465e-05,
|
| 24319 |
+
"loss": 9.5781,
|
| 24320 |
+
"step": 3466
|
| 24321 |
+
},
|
| 24322 |
+
{
|
| 24323 |
+
"epoch": 0.0003467,
|
| 24324 |
+
"grad_norm": 4.593960285186768,
|
| 24325 |
+
"learning_rate": 3.4660000000000004e-05,
|
| 24326 |
+
"loss": 9.9492,
|
| 24327 |
+
"step": 3467
|
| 24328 |
+
},
|
| 24329 |
+
{
|
| 24330 |
+
"epoch": 0.0003468,
|
| 24331 |
+
"grad_norm": 3.992659091949463,
|
| 24332 |
+
"learning_rate": 3.467e-05,
|
| 24333 |
+
"loss": 9.2578,
|
| 24334 |
+
"step": 3468
|
| 24335 |
+
},
|
| 24336 |
+
{
|
| 24337 |
+
"epoch": 0.0003469,
|
| 24338 |
+
"grad_norm": 4.568031311035156,
|
| 24339 |
+
"learning_rate": 3.468e-05,
|
| 24340 |
+
"loss": 10.0391,
|
| 24341 |
+
"step": 3469
|
| 24342 |
+
},
|
| 24343 |
+
{
|
| 24344 |
+
"epoch": 0.000347,
|
| 24345 |
+
"grad_norm": 3.8288345336914062,
|
| 24346 |
+
"learning_rate": 3.469e-05,
|
| 24347 |
+
"loss": 9.3359,
|
| 24348 |
+
"step": 3470
|
| 24349 |
+
},
|
| 24350 |
+
{
|
| 24351 |
+
"epoch": 0.0003471,
|
| 24352 |
+
"grad_norm": 4.219034671783447,
|
| 24353 |
+
"learning_rate": 3.4699999999999996e-05,
|
| 24354 |
+
"loss": 9.6016,
|
| 24355 |
+
"step": 3471
|
| 24356 |
+
},
|
| 24357 |
+
{
|
| 24358 |
+
"epoch": 0.0003472,
|
| 24359 |
+
"grad_norm": 4.229843616485596,
|
| 24360 |
+
"learning_rate": 3.4710000000000005e-05,
|
| 24361 |
+
"loss": 9.5312,
|
| 24362 |
+
"step": 3472
|
| 24363 |
+
},
|
| 24364 |
+
{
|
| 24365 |
+
"epoch": 0.0003473,
|
| 24366 |
+
"grad_norm": 3.826164960861206,
|
| 24367 |
+
"learning_rate": 3.472e-05,
|
| 24368 |
+
"loss": 9.6172,
|
| 24369 |
+
"step": 3473
|
| 24370 |
+
},
|
| 24371 |
+
{
|
| 24372 |
+
"epoch": 0.0003474,
|
| 24373 |
+
"grad_norm": 3.8404030799865723,
|
| 24374 |
+
"learning_rate": 3.473e-05,
|
| 24375 |
+
"loss": 8.8984,
|
| 24376 |
+
"step": 3474
|
| 24377 |
+
},
|
| 24378 |
+
{
|
| 24379 |
+
"epoch": 0.0003475,
|
| 24380 |
+
"grad_norm": 3.9308347702026367,
|
| 24381 |
+
"learning_rate": 3.474e-05,
|
| 24382 |
+
"loss": 9.4609,
|
| 24383 |
+
"step": 3475
|
| 24384 |
+
},
|
| 24385 |
+
{
|
| 24386 |
+
"epoch": 0.0003476,
|
| 24387 |
+
"grad_norm": 3.8785057067871094,
|
| 24388 |
+
"learning_rate": 3.475e-05,
|
| 24389 |
+
"loss": 9.25,
|
| 24390 |
+
"step": 3476
|
| 24391 |
+
},
|
| 24392 |
+
{
|
| 24393 |
+
"epoch": 0.0003477,
|
| 24394 |
+
"grad_norm": 4.225211143493652,
|
| 24395 |
+
"learning_rate": 3.476e-05,
|
| 24396 |
+
"loss": 9.957,
|
| 24397 |
+
"step": 3477
|
| 24398 |
+
},
|
| 24399 |
+
{
|
| 24400 |
+
"epoch": 0.0003478,
|
| 24401 |
+
"grad_norm": 3.824558734893799,
|
| 24402 |
+
"learning_rate": 3.477e-05,
|
| 24403 |
+
"loss": 8.8125,
|
| 24404 |
+
"step": 3478
|
| 24405 |
+
},
|
| 24406 |
+
{
|
| 24407 |
+
"epoch": 0.0003479,
|
| 24408 |
+
"grad_norm": 3.9660282135009766,
|
| 24409 |
+
"learning_rate": 3.478e-05,
|
| 24410 |
+
"loss": 9.2305,
|
| 24411 |
+
"step": 3479
|
| 24412 |
+
},
|
| 24413 |
+
{
|
| 24414 |
+
"epoch": 0.000348,
|
| 24415 |
+
"grad_norm": 4.11309289932251,
|
| 24416 |
+
"learning_rate": 3.479e-05,
|
| 24417 |
+
"loss": 9.7188,
|
| 24418 |
+
"step": 3480
|
| 24419 |
+
},
|
| 24420 |
+
{
|
| 24421 |
+
"epoch": 0.0003481,
|
| 24422 |
+
"grad_norm": 3.983299732208252,
|
| 24423 |
+
"learning_rate": 3.48e-05,
|
| 24424 |
+
"loss": 9.6719,
|
| 24425 |
+
"step": 3481
|
| 24426 |
+
},
|
| 24427 |
+
{
|
| 24428 |
+
"epoch": 0.0003482,
|
| 24429 |
+
"grad_norm": 3.830436944961548,
|
| 24430 |
+
"learning_rate": 3.481e-05,
|
| 24431 |
+
"loss": 9.5273,
|
| 24432 |
+
"step": 3482
|
| 24433 |
+
},
|
| 24434 |
+
{
|
| 24435 |
+
"epoch": 0.0003483,
|
| 24436 |
+
"grad_norm": 3.9448773860931396,
|
| 24437 |
+
"learning_rate": 3.482e-05,
|
| 24438 |
+
"loss": 9.5156,
|
| 24439 |
+
"step": 3483
|
| 24440 |
+
},
|
| 24441 |
+
{
|
| 24442 |
+
"epoch": 0.0003484,
|
| 24443 |
+
"grad_norm": 4.055325031280518,
|
| 24444 |
+
"learning_rate": 3.4830000000000004e-05,
|
| 24445 |
+
"loss": 8.8555,
|
| 24446 |
+
"step": 3484
|
| 24447 |
+
},
|
| 24448 |
+
{
|
| 24449 |
+
"epoch": 0.0003485,
|
| 24450 |
+
"grad_norm": 3.7771012783050537,
|
| 24451 |
+
"learning_rate": 3.484e-05,
|
| 24452 |
+
"loss": 9.1094,
|
| 24453 |
+
"step": 3485
|
| 24454 |
+
},
|
| 24455 |
+
{
|
| 24456 |
+
"epoch": 0.0003486,
|
| 24457 |
+
"grad_norm": 4.95332145690918,
|
| 24458 |
+
"learning_rate": 3.485e-05,
|
| 24459 |
+
"loss": 9.7656,
|
| 24460 |
+
"step": 3486
|
| 24461 |
+
},
|
| 24462 |
+
{
|
| 24463 |
+
"epoch": 0.0003487,
|
| 24464 |
+
"grad_norm": 4.324108600616455,
|
| 24465 |
+
"learning_rate": 3.486e-05,
|
| 24466 |
+
"loss": 10.3516,
|
| 24467 |
+
"step": 3487
|
| 24468 |
+
},
|
| 24469 |
+
{
|
| 24470 |
+
"epoch": 0.0003488,
|
| 24471 |
+
"grad_norm": 4.3664164543151855,
|
| 24472 |
+
"learning_rate": 3.4869999999999996e-05,
|
| 24473 |
+
"loss": 9.8555,
|
| 24474 |
+
"step": 3488
|
| 24475 |
+
},
|
| 24476 |
+
{
|
| 24477 |
+
"epoch": 0.0003489,
|
| 24478 |
+
"grad_norm": 4.212305068969727,
|
| 24479 |
+
"learning_rate": 3.4880000000000005e-05,
|
| 24480 |
+
"loss": 9.3047,
|
| 24481 |
+
"step": 3489
|
| 24482 |
+
},
|
| 24483 |
+
{
|
| 24484 |
+
"epoch": 0.000349,
|
| 24485 |
+
"grad_norm": 4.726918697357178,
|
| 24486 |
+
"learning_rate": 3.489e-05,
|
| 24487 |
+
"loss": 9.9375,
|
| 24488 |
+
"step": 3490
|
| 24489 |
+
},
|
| 24490 |
+
{
|
| 24491 |
+
"epoch": 0.0003491,
|
| 24492 |
+
"grad_norm": 4.540855407714844,
|
| 24493 |
+
"learning_rate": 3.49e-05,
|
| 24494 |
+
"loss": 8.9531,
|
| 24495 |
+
"step": 3491
|
| 24496 |
+
},
|
| 24497 |
+
{
|
| 24498 |
+
"epoch": 0.0003492,
|
| 24499 |
+
"grad_norm": 3.9311349391937256,
|
| 24500 |
+
"learning_rate": 3.491e-05,
|
| 24501 |
+
"loss": 8.4844,
|
| 24502 |
+
"step": 3492
|
| 24503 |
+
},
|
| 24504 |
+
{
|
| 24505 |
+
"epoch": 0.0003493,
|
| 24506 |
+
"grad_norm": 3.7078661918640137,
|
| 24507 |
+
"learning_rate": 3.492e-05,
|
| 24508 |
+
"loss": 8.4766,
|
| 24509 |
+
"step": 3493
|
| 24510 |
+
},
|
| 24511 |
+
{
|
| 24512 |
+
"epoch": 0.0003494,
|
| 24513 |
+
"grad_norm": 4.00162935256958,
|
| 24514 |
+
"learning_rate": 3.493e-05,
|
| 24515 |
+
"loss": 10.0859,
|
| 24516 |
+
"step": 3494
|
| 24517 |
+
},
|
| 24518 |
+
{
|
| 24519 |
+
"epoch": 0.0003495,
|
| 24520 |
+
"grad_norm": 4.181024074554443,
|
| 24521 |
+
"learning_rate": 3.494e-05,
|
| 24522 |
+
"loss": 9.793,
|
| 24523 |
+
"step": 3495
|
| 24524 |
+
},
|
| 24525 |
+
{
|
| 24526 |
+
"epoch": 0.0003496,
|
| 24527 |
+
"grad_norm": 4.578210830688477,
|
| 24528 |
+
"learning_rate": 3.495e-05,
|
| 24529 |
+
"loss": 10.6016,
|
| 24530 |
+
"step": 3496
|
| 24531 |
+
},
|
| 24532 |
+
{
|
| 24533 |
+
"epoch": 0.0003497,
|
| 24534 |
+
"grad_norm": 3.7809059619903564,
|
| 24535 |
+
"learning_rate": 3.496e-05,
|
| 24536 |
+
"loss": 8.1562,
|
| 24537 |
+
"step": 3497
|
| 24538 |
+
},
|
| 24539 |
+
{
|
| 24540 |
+
"epoch": 0.0003498,
|
| 24541 |
+
"grad_norm": 6.469610691070557,
|
| 24542 |
+
"learning_rate": 3.497e-05,
|
| 24543 |
+
"loss": 10.7969,
|
| 24544 |
+
"step": 3498
|
| 24545 |
+
},
|
| 24546 |
+
{
|
| 24547 |
+
"epoch": 0.0003499,
|
| 24548 |
+
"grad_norm": 3.9823861122131348,
|
| 24549 |
+
"learning_rate": 3.498e-05,
|
| 24550 |
+
"loss": 9.3828,
|
| 24551 |
+
"step": 3499
|
| 24552 |
+
},
|
| 24553 |
+
{
|
| 24554 |
+
"epoch": 0.00035,
|
| 24555 |
+
"grad_norm": 3.8631653785705566,
|
| 24556 |
+
"learning_rate": 3.499e-05,
|
| 24557 |
+
"loss": 8.6016,
|
| 24558 |
+
"step": 3500
|
| 24559 |
+
},
|
| 24560 |
+
{
|
| 24561 |
+
"epoch": 0.00035,
|
| 24562 |
+
"eval_loss": 1.0911582708358765,
|
| 24563 |
+
"eval_runtime": 406.3204,
|
| 24564 |
+
"eval_samples_per_second": 24.611,
|
| 24565 |
+
"eval_steps_per_second": 1.538,
|
| 24566 |
+
"step": 3500
|
| 24567 |
}
|
| 24568 |
],
|
| 24569 |
"logging_steps": 1,
|