Training in progress, step 525, 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 1311381296
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:240b5dc839c6a052475de2c65478c6ae63f48f86c374757e8b75e4ee54277d3b
|
| 3 |
size 1311381296
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1543168721
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:84f90b4ab1f233fcbef72dbed1b18cc78e04aca81bf0cc7244a6348a41da9084
|
| 3 |
size 1543168721
|
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 sha256:3d6d8fafcd1ee268414be5acf0366296af5b03d60871978712eac1979cb42d65
|
| 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:b341c034af54d94ea6acd4c2c1460069e92f327e2bc7c02119cd9b60167f4580
|
| 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": 175,
|
| 7 |
-
"global_step":
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
@@ -2474,6 +2474,1239 @@
|
|
| 2474 |
"eval_samples_per_second": 8.576,
|
| 2475 |
"eval_steps_per_second": 2.144,
|
| 2476 |
"step": 350
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 2477 |
}
|
| 2478 |
],
|
| 2479 |
"logging_steps": 1,
|
|
@@ -2493,7 +3726,7 @@
|
|
| 2493 |
"attributes": {}
|
| 2494 |
}
|
| 2495 |
},
|
| 2496 |
-
"total_flos":
|
| 2497 |
"train_batch_size": 16,
|
| 2498 |
"trial_name": null,
|
| 2499 |
"trial_params": null
|
|
|
|
| 2 |
"best_global_step": null,
|
| 3 |
"best_metric": null,
|
| 4 |
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 0.15,
|
| 6 |
"eval_steps": 175,
|
| 7 |
+
"global_step": 525,
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
|
|
| 2474 |
"eval_samples_per_second": 8.576,
|
| 2475 |
"eval_steps_per_second": 2.144,
|
| 2476 |
"step": 350
|
| 2477 |
+
},
|
| 2478 |
+
{
|
| 2479 |
+
"epoch": 0.10028571428571428,
|
| 2480 |
+
"grad_norm": 1.3528292179107666,
|
| 2481 |
+
"learning_rate": 0.001,
|
| 2482 |
+
"loss": 6.034597396850586,
|
| 2483 |
+
"step": 351
|
| 2484 |
+
},
|
| 2485 |
+
{
|
| 2486 |
+
"epoch": 0.10057142857142858,
|
| 2487 |
+
"grad_norm": 1.392505407333374,
|
| 2488 |
+
"learning_rate": 0.001,
|
| 2489 |
+
"loss": 5.979316711425781,
|
| 2490 |
+
"step": 352
|
| 2491 |
+
},
|
| 2492 |
+
{
|
| 2493 |
+
"epoch": 0.10085714285714285,
|
| 2494 |
+
"grad_norm": 1.362580418586731,
|
| 2495 |
+
"learning_rate": 0.001,
|
| 2496 |
+
"loss": 5.992738723754883,
|
| 2497 |
+
"step": 353
|
| 2498 |
+
},
|
| 2499 |
+
{
|
| 2500 |
+
"epoch": 0.10114285714285715,
|
| 2501 |
+
"grad_norm": 1.3664690256118774,
|
| 2502 |
+
"learning_rate": 0.001,
|
| 2503 |
+
"loss": 6.036233901977539,
|
| 2504 |
+
"step": 354
|
| 2505 |
+
},
|
| 2506 |
+
{
|
| 2507 |
+
"epoch": 0.10142857142857142,
|
| 2508 |
+
"grad_norm": 1.4197261333465576,
|
| 2509 |
+
"learning_rate": 0.001,
|
| 2510 |
+
"loss": 5.977957725524902,
|
| 2511 |
+
"step": 355
|
| 2512 |
+
},
|
| 2513 |
+
{
|
| 2514 |
+
"epoch": 0.10171428571428572,
|
| 2515 |
+
"grad_norm": 1.4244575500488281,
|
| 2516 |
+
"learning_rate": 0.001,
|
| 2517 |
+
"loss": 5.941583156585693,
|
| 2518 |
+
"step": 356
|
| 2519 |
+
},
|
| 2520 |
+
{
|
| 2521 |
+
"epoch": 0.102,
|
| 2522 |
+
"grad_norm": 1.3508753776550293,
|
| 2523 |
+
"learning_rate": 0.001,
|
| 2524 |
+
"loss": 5.954012870788574,
|
| 2525 |
+
"step": 357
|
| 2526 |
+
},
|
| 2527 |
+
{
|
| 2528 |
+
"epoch": 0.10228571428571429,
|
| 2529 |
+
"grad_norm": 1.4688359498977661,
|
| 2530 |
+
"learning_rate": 0.001,
|
| 2531 |
+
"loss": 5.932170867919922,
|
| 2532 |
+
"step": 358
|
| 2533 |
+
},
|
| 2534 |
+
{
|
| 2535 |
+
"epoch": 0.10257142857142858,
|
| 2536 |
+
"grad_norm": 1.3655815124511719,
|
| 2537 |
+
"learning_rate": 0.001,
|
| 2538 |
+
"loss": 5.964273929595947,
|
| 2539 |
+
"step": 359
|
| 2540 |
+
},
|
| 2541 |
+
{
|
| 2542 |
+
"epoch": 0.10285714285714286,
|
| 2543 |
+
"grad_norm": 1.3773704767227173,
|
| 2544 |
+
"learning_rate": 0.001,
|
| 2545 |
+
"loss": 5.929989814758301,
|
| 2546 |
+
"step": 360
|
| 2547 |
+
},
|
| 2548 |
+
{
|
| 2549 |
+
"epoch": 0.10314285714285715,
|
| 2550 |
+
"grad_norm": 1.414300560951233,
|
| 2551 |
+
"learning_rate": 0.001,
|
| 2552 |
+
"loss": 5.952170372009277,
|
| 2553 |
+
"step": 361
|
| 2554 |
+
},
|
| 2555 |
+
{
|
| 2556 |
+
"epoch": 0.10342857142857143,
|
| 2557 |
+
"grad_norm": 1.4066667556762695,
|
| 2558 |
+
"learning_rate": 0.001,
|
| 2559 |
+
"loss": 5.928333759307861,
|
| 2560 |
+
"step": 362
|
| 2561 |
+
},
|
| 2562 |
+
{
|
| 2563 |
+
"epoch": 0.10371428571428572,
|
| 2564 |
+
"grad_norm": 1.350995659828186,
|
| 2565 |
+
"learning_rate": 0.001,
|
| 2566 |
+
"loss": 5.880458831787109,
|
| 2567 |
+
"step": 363
|
| 2568 |
+
},
|
| 2569 |
+
{
|
| 2570 |
+
"epoch": 0.104,
|
| 2571 |
+
"grad_norm": 1.2922935485839844,
|
| 2572 |
+
"learning_rate": 0.001,
|
| 2573 |
+
"loss": 5.87301778793335,
|
| 2574 |
+
"step": 364
|
| 2575 |
+
},
|
| 2576 |
+
{
|
| 2577 |
+
"epoch": 0.10428571428571429,
|
| 2578 |
+
"grad_norm": 1.3252061605453491,
|
| 2579 |
+
"learning_rate": 0.001,
|
| 2580 |
+
"loss": 5.878377437591553,
|
| 2581 |
+
"step": 365
|
| 2582 |
+
},
|
| 2583 |
+
{
|
| 2584 |
+
"epoch": 0.10457142857142857,
|
| 2585 |
+
"grad_norm": 1.408143401145935,
|
| 2586 |
+
"learning_rate": 0.001,
|
| 2587 |
+
"loss": 5.818056583404541,
|
| 2588 |
+
"step": 366
|
| 2589 |
+
},
|
| 2590 |
+
{
|
| 2591 |
+
"epoch": 0.10485714285714286,
|
| 2592 |
+
"grad_norm": 1.360413670539856,
|
| 2593 |
+
"learning_rate": 0.001,
|
| 2594 |
+
"loss": 5.873591423034668,
|
| 2595 |
+
"step": 367
|
| 2596 |
+
},
|
| 2597 |
+
{
|
| 2598 |
+
"epoch": 0.10514285714285715,
|
| 2599 |
+
"grad_norm": 1.344091773033142,
|
| 2600 |
+
"learning_rate": 0.001,
|
| 2601 |
+
"loss": 5.825854778289795,
|
| 2602 |
+
"step": 368
|
| 2603 |
+
},
|
| 2604 |
+
{
|
| 2605 |
+
"epoch": 0.10542857142857143,
|
| 2606 |
+
"grad_norm": 1.3215280771255493,
|
| 2607 |
+
"learning_rate": 0.001,
|
| 2608 |
+
"loss": 5.826833248138428,
|
| 2609 |
+
"step": 369
|
| 2610 |
+
},
|
| 2611 |
+
{
|
| 2612 |
+
"epoch": 0.10571428571428572,
|
| 2613 |
+
"grad_norm": 1.3262429237365723,
|
| 2614 |
+
"learning_rate": 0.001,
|
| 2615 |
+
"loss": 5.827884197235107,
|
| 2616 |
+
"step": 370
|
| 2617 |
+
},
|
| 2618 |
+
{
|
| 2619 |
+
"epoch": 0.106,
|
| 2620 |
+
"grad_norm": 1.2807199954986572,
|
| 2621 |
+
"learning_rate": 0.001,
|
| 2622 |
+
"loss": 5.807199954986572,
|
| 2623 |
+
"step": 371
|
| 2624 |
+
},
|
| 2625 |
+
{
|
| 2626 |
+
"epoch": 0.10628571428571429,
|
| 2627 |
+
"grad_norm": 1.302396535873413,
|
| 2628 |
+
"learning_rate": 0.001,
|
| 2629 |
+
"loss": 5.773519515991211,
|
| 2630 |
+
"step": 372
|
| 2631 |
+
},
|
| 2632 |
+
{
|
| 2633 |
+
"epoch": 0.10657142857142857,
|
| 2634 |
+
"grad_norm": 1.2010446786880493,
|
| 2635 |
+
"learning_rate": 0.001,
|
| 2636 |
+
"loss": 5.818293571472168,
|
| 2637 |
+
"step": 373
|
| 2638 |
+
},
|
| 2639 |
+
{
|
| 2640 |
+
"epoch": 0.10685714285714286,
|
| 2641 |
+
"grad_norm": 1.297667384147644,
|
| 2642 |
+
"learning_rate": 0.001,
|
| 2643 |
+
"loss": 5.738192558288574,
|
| 2644 |
+
"step": 374
|
| 2645 |
+
},
|
| 2646 |
+
{
|
| 2647 |
+
"epoch": 0.10714285714285714,
|
| 2648 |
+
"grad_norm": 1.260051965713501,
|
| 2649 |
+
"learning_rate": 0.001,
|
| 2650 |
+
"loss": 5.853949069976807,
|
| 2651 |
+
"step": 375
|
| 2652 |
+
},
|
| 2653 |
+
{
|
| 2654 |
+
"epoch": 0.10742857142857143,
|
| 2655 |
+
"grad_norm": 1.266003131866455,
|
| 2656 |
+
"learning_rate": 0.001,
|
| 2657 |
+
"loss": 5.801722526550293,
|
| 2658 |
+
"step": 376
|
| 2659 |
+
},
|
| 2660 |
+
{
|
| 2661 |
+
"epoch": 0.10771428571428572,
|
| 2662 |
+
"grad_norm": 1.2530462741851807,
|
| 2663 |
+
"learning_rate": 0.001,
|
| 2664 |
+
"loss": 5.817451000213623,
|
| 2665 |
+
"step": 377
|
| 2666 |
+
},
|
| 2667 |
+
{
|
| 2668 |
+
"epoch": 0.108,
|
| 2669 |
+
"grad_norm": 1.253485083580017,
|
| 2670 |
+
"learning_rate": 0.001,
|
| 2671 |
+
"loss": 5.770527362823486,
|
| 2672 |
+
"step": 378
|
| 2673 |
+
},
|
| 2674 |
+
{
|
| 2675 |
+
"epoch": 0.10828571428571429,
|
| 2676 |
+
"grad_norm": 1.301617980003357,
|
| 2677 |
+
"learning_rate": 0.001,
|
| 2678 |
+
"loss": 5.79968786239624,
|
| 2679 |
+
"step": 379
|
| 2680 |
+
},
|
| 2681 |
+
{
|
| 2682 |
+
"epoch": 0.10857142857142857,
|
| 2683 |
+
"grad_norm": 1.264508843421936,
|
| 2684 |
+
"learning_rate": 0.001,
|
| 2685 |
+
"loss": 5.702681541442871,
|
| 2686 |
+
"step": 380
|
| 2687 |
+
},
|
| 2688 |
+
{
|
| 2689 |
+
"epoch": 0.10885714285714286,
|
| 2690 |
+
"grad_norm": 1.1902637481689453,
|
| 2691 |
+
"learning_rate": 0.001,
|
| 2692 |
+
"loss": 5.78672456741333,
|
| 2693 |
+
"step": 381
|
| 2694 |
+
},
|
| 2695 |
+
{
|
| 2696 |
+
"epoch": 0.10914285714285714,
|
| 2697 |
+
"grad_norm": 1.2222683429718018,
|
| 2698 |
+
"learning_rate": 0.001,
|
| 2699 |
+
"loss": 5.722955226898193,
|
| 2700 |
+
"step": 382
|
| 2701 |
+
},
|
| 2702 |
+
{
|
| 2703 |
+
"epoch": 0.10942857142857143,
|
| 2704 |
+
"grad_norm": 1.2311407327651978,
|
| 2705 |
+
"learning_rate": 0.001,
|
| 2706 |
+
"loss": 5.774104595184326,
|
| 2707 |
+
"step": 383
|
| 2708 |
+
},
|
| 2709 |
+
{
|
| 2710 |
+
"epoch": 0.10971428571428571,
|
| 2711 |
+
"grad_norm": 1.3318625688552856,
|
| 2712 |
+
"learning_rate": 0.001,
|
| 2713 |
+
"loss": 5.682190418243408,
|
| 2714 |
+
"step": 384
|
| 2715 |
+
},
|
| 2716 |
+
{
|
| 2717 |
+
"epoch": 0.11,
|
| 2718 |
+
"grad_norm": 1.240397572517395,
|
| 2719 |
+
"learning_rate": 0.001,
|
| 2720 |
+
"loss": 5.739994049072266,
|
| 2721 |
+
"step": 385
|
| 2722 |
+
},
|
| 2723 |
+
{
|
| 2724 |
+
"epoch": 0.11028571428571429,
|
| 2725 |
+
"grad_norm": 1.198891520500183,
|
| 2726 |
+
"learning_rate": 0.001,
|
| 2727 |
+
"loss": 5.662315845489502,
|
| 2728 |
+
"step": 386
|
| 2729 |
+
},
|
| 2730 |
+
{
|
| 2731 |
+
"epoch": 0.11057142857142857,
|
| 2732 |
+
"grad_norm": 1.1602271795272827,
|
| 2733 |
+
"learning_rate": 0.001,
|
| 2734 |
+
"loss": 5.753437519073486,
|
| 2735 |
+
"step": 387
|
| 2736 |
+
},
|
| 2737 |
+
{
|
| 2738 |
+
"epoch": 0.11085714285714286,
|
| 2739 |
+
"grad_norm": 1.2685130834579468,
|
| 2740 |
+
"learning_rate": 0.001,
|
| 2741 |
+
"loss": 5.619208335876465,
|
| 2742 |
+
"step": 388
|
| 2743 |
+
},
|
| 2744 |
+
{
|
| 2745 |
+
"epoch": 0.11114285714285714,
|
| 2746 |
+
"grad_norm": 1.3128584623336792,
|
| 2747 |
+
"learning_rate": 0.001,
|
| 2748 |
+
"loss": 5.722102642059326,
|
| 2749 |
+
"step": 389
|
| 2750 |
+
},
|
| 2751 |
+
{
|
| 2752 |
+
"epoch": 0.11142857142857143,
|
| 2753 |
+
"grad_norm": 1.204474925994873,
|
| 2754 |
+
"learning_rate": 0.001,
|
| 2755 |
+
"loss": 5.635885238647461,
|
| 2756 |
+
"step": 390
|
| 2757 |
+
},
|
| 2758 |
+
{
|
| 2759 |
+
"epoch": 0.11171428571428571,
|
| 2760 |
+
"grad_norm": 1.1874816417694092,
|
| 2761 |
+
"learning_rate": 0.001,
|
| 2762 |
+
"loss": 5.661076068878174,
|
| 2763 |
+
"step": 391
|
| 2764 |
+
},
|
| 2765 |
+
{
|
| 2766 |
+
"epoch": 0.112,
|
| 2767 |
+
"grad_norm": 1.1582341194152832,
|
| 2768 |
+
"learning_rate": 0.001,
|
| 2769 |
+
"loss": 5.608723163604736,
|
| 2770 |
+
"step": 392
|
| 2771 |
+
},
|
| 2772 |
+
{
|
| 2773 |
+
"epoch": 0.11228571428571428,
|
| 2774 |
+
"grad_norm": 1.2210676670074463,
|
| 2775 |
+
"learning_rate": 0.001,
|
| 2776 |
+
"loss": 5.618552207946777,
|
| 2777 |
+
"step": 393
|
| 2778 |
+
},
|
| 2779 |
+
{
|
| 2780 |
+
"epoch": 0.11257142857142857,
|
| 2781 |
+
"grad_norm": 1.192152976989746,
|
| 2782 |
+
"learning_rate": 0.001,
|
| 2783 |
+
"loss": 5.620550632476807,
|
| 2784 |
+
"step": 394
|
| 2785 |
+
},
|
| 2786 |
+
{
|
| 2787 |
+
"epoch": 0.11285714285714285,
|
| 2788 |
+
"grad_norm": 1.1243138313293457,
|
| 2789 |
+
"learning_rate": 0.001,
|
| 2790 |
+
"loss": 5.688298225402832,
|
| 2791 |
+
"step": 395
|
| 2792 |
+
},
|
| 2793 |
+
{
|
| 2794 |
+
"epoch": 0.11314285714285714,
|
| 2795 |
+
"grad_norm": 1.1195775270462036,
|
| 2796 |
+
"learning_rate": 0.001,
|
| 2797 |
+
"loss": 5.562441825866699,
|
| 2798 |
+
"step": 396
|
| 2799 |
+
},
|
| 2800 |
+
{
|
| 2801 |
+
"epoch": 0.11342857142857143,
|
| 2802 |
+
"grad_norm": 1.2325485944747925,
|
| 2803 |
+
"learning_rate": 0.001,
|
| 2804 |
+
"loss": 5.615917682647705,
|
| 2805 |
+
"step": 397
|
| 2806 |
+
},
|
| 2807 |
+
{
|
| 2808 |
+
"epoch": 0.11371428571428571,
|
| 2809 |
+
"grad_norm": 1.094494342803955,
|
| 2810 |
+
"learning_rate": 0.001,
|
| 2811 |
+
"loss": 5.563663959503174,
|
| 2812 |
+
"step": 398
|
| 2813 |
+
},
|
| 2814 |
+
{
|
| 2815 |
+
"epoch": 0.114,
|
| 2816 |
+
"grad_norm": 1.1367664337158203,
|
| 2817 |
+
"learning_rate": 0.001,
|
| 2818 |
+
"loss": 5.659440040588379,
|
| 2819 |
+
"step": 399
|
| 2820 |
+
},
|
| 2821 |
+
{
|
| 2822 |
+
"epoch": 0.11428571428571428,
|
| 2823 |
+
"grad_norm": 1.2361277341842651,
|
| 2824 |
+
"learning_rate": 0.001,
|
| 2825 |
+
"loss": 5.571027755737305,
|
| 2826 |
+
"step": 400
|
| 2827 |
+
},
|
| 2828 |
+
{
|
| 2829 |
+
"epoch": 0.11457142857142857,
|
| 2830 |
+
"grad_norm": 1.089381456375122,
|
| 2831 |
+
"learning_rate": 0.001,
|
| 2832 |
+
"loss": 5.635144233703613,
|
| 2833 |
+
"step": 401
|
| 2834 |
+
},
|
| 2835 |
+
{
|
| 2836 |
+
"epoch": 0.11485714285714285,
|
| 2837 |
+
"grad_norm": 1.3162202835083008,
|
| 2838 |
+
"learning_rate": 0.001,
|
| 2839 |
+
"loss": 5.620790958404541,
|
| 2840 |
+
"step": 402
|
| 2841 |
+
},
|
| 2842 |
+
{
|
| 2843 |
+
"epoch": 0.11514285714285714,
|
| 2844 |
+
"grad_norm": 1.1435638666152954,
|
| 2845 |
+
"learning_rate": 0.001,
|
| 2846 |
+
"loss": 5.574390411376953,
|
| 2847 |
+
"step": 403
|
| 2848 |
+
},
|
| 2849 |
+
{
|
| 2850 |
+
"epoch": 0.11542857142857142,
|
| 2851 |
+
"grad_norm": 1.3203638792037964,
|
| 2852 |
+
"learning_rate": 0.001,
|
| 2853 |
+
"loss": 5.517210006713867,
|
| 2854 |
+
"step": 404
|
| 2855 |
+
},
|
| 2856 |
+
{
|
| 2857 |
+
"epoch": 0.11571428571428571,
|
| 2858 |
+
"grad_norm": 1.0997134447097778,
|
| 2859 |
+
"learning_rate": 0.001,
|
| 2860 |
+
"loss": 5.55522346496582,
|
| 2861 |
+
"step": 405
|
| 2862 |
+
},
|
| 2863 |
+
{
|
| 2864 |
+
"epoch": 0.116,
|
| 2865 |
+
"grad_norm": 1.1040183305740356,
|
| 2866 |
+
"learning_rate": 0.001,
|
| 2867 |
+
"loss": 5.548467636108398,
|
| 2868 |
+
"step": 406
|
| 2869 |
+
},
|
| 2870 |
+
{
|
| 2871 |
+
"epoch": 0.11628571428571428,
|
| 2872 |
+
"grad_norm": 1.1761415004730225,
|
| 2873 |
+
"learning_rate": 0.001,
|
| 2874 |
+
"loss": 5.560356616973877,
|
| 2875 |
+
"step": 407
|
| 2876 |
+
},
|
| 2877 |
+
{
|
| 2878 |
+
"epoch": 0.11657142857142858,
|
| 2879 |
+
"grad_norm": 1.050987958908081,
|
| 2880 |
+
"learning_rate": 0.001,
|
| 2881 |
+
"loss": 5.533413887023926,
|
| 2882 |
+
"step": 408
|
| 2883 |
+
},
|
| 2884 |
+
{
|
| 2885 |
+
"epoch": 0.11685714285714285,
|
| 2886 |
+
"grad_norm": 1.2728478908538818,
|
| 2887 |
+
"learning_rate": 0.001,
|
| 2888 |
+
"loss": 5.5438971519470215,
|
| 2889 |
+
"step": 409
|
| 2890 |
+
},
|
| 2891 |
+
{
|
| 2892 |
+
"epoch": 0.11714285714285715,
|
| 2893 |
+
"grad_norm": 1.0529109239578247,
|
| 2894 |
+
"learning_rate": 0.001,
|
| 2895 |
+
"loss": 5.479302406311035,
|
| 2896 |
+
"step": 410
|
| 2897 |
+
},
|
| 2898 |
+
{
|
| 2899 |
+
"epoch": 0.11742857142857142,
|
| 2900 |
+
"grad_norm": 1.0749348402023315,
|
| 2901 |
+
"learning_rate": 0.001,
|
| 2902 |
+
"loss": 5.514036655426025,
|
| 2903 |
+
"step": 411
|
| 2904 |
+
},
|
| 2905 |
+
{
|
| 2906 |
+
"epoch": 0.11771428571428572,
|
| 2907 |
+
"grad_norm": 1.1333913803100586,
|
| 2908 |
+
"learning_rate": 0.001,
|
| 2909 |
+
"loss": 5.504560947418213,
|
| 2910 |
+
"step": 412
|
| 2911 |
+
},
|
| 2912 |
+
{
|
| 2913 |
+
"epoch": 0.118,
|
| 2914 |
+
"grad_norm": 1.136358618736267,
|
| 2915 |
+
"learning_rate": 0.001,
|
| 2916 |
+
"loss": 5.4899373054504395,
|
| 2917 |
+
"step": 413
|
| 2918 |
+
},
|
| 2919 |
+
{
|
| 2920 |
+
"epoch": 0.11828571428571429,
|
| 2921 |
+
"grad_norm": 1.0938942432403564,
|
| 2922 |
+
"learning_rate": 0.001,
|
| 2923 |
+
"loss": 5.436349868774414,
|
| 2924 |
+
"step": 414
|
| 2925 |
+
},
|
| 2926 |
+
{
|
| 2927 |
+
"epoch": 0.11857142857142858,
|
| 2928 |
+
"grad_norm": 1.1970667839050293,
|
| 2929 |
+
"learning_rate": 0.001,
|
| 2930 |
+
"loss": 5.5028557777404785,
|
| 2931 |
+
"step": 415
|
| 2932 |
+
},
|
| 2933 |
+
{
|
| 2934 |
+
"epoch": 0.11885714285714286,
|
| 2935 |
+
"grad_norm": 1.1001304388046265,
|
| 2936 |
+
"learning_rate": 0.001,
|
| 2937 |
+
"loss": 5.438455581665039,
|
| 2938 |
+
"step": 416
|
| 2939 |
+
},
|
| 2940 |
+
{
|
| 2941 |
+
"epoch": 0.11914285714285715,
|
| 2942 |
+
"grad_norm": 1.0444203615188599,
|
| 2943 |
+
"learning_rate": 0.001,
|
| 2944 |
+
"loss": 5.500203609466553,
|
| 2945 |
+
"step": 417
|
| 2946 |
+
},
|
| 2947 |
+
{
|
| 2948 |
+
"epoch": 0.11942857142857143,
|
| 2949 |
+
"grad_norm": 1.2863494157791138,
|
| 2950 |
+
"learning_rate": 0.001,
|
| 2951 |
+
"loss": 5.4192891120910645,
|
| 2952 |
+
"step": 418
|
| 2953 |
+
},
|
| 2954 |
+
{
|
| 2955 |
+
"epoch": 0.11971428571428572,
|
| 2956 |
+
"grad_norm": 1.1580034494400024,
|
| 2957 |
+
"learning_rate": 0.001,
|
| 2958 |
+
"loss": 5.442489147186279,
|
| 2959 |
+
"step": 419
|
| 2960 |
+
},
|
| 2961 |
+
{
|
| 2962 |
+
"epoch": 0.12,
|
| 2963 |
+
"grad_norm": 1.0923646688461304,
|
| 2964 |
+
"learning_rate": 0.001,
|
| 2965 |
+
"loss": 5.414194583892822,
|
| 2966 |
+
"step": 420
|
| 2967 |
+
},
|
| 2968 |
+
{
|
| 2969 |
+
"epoch": 0.12028571428571429,
|
| 2970 |
+
"grad_norm": 1.1441943645477295,
|
| 2971 |
+
"learning_rate": 0.001,
|
| 2972 |
+
"loss": 5.431639671325684,
|
| 2973 |
+
"step": 421
|
| 2974 |
+
},
|
| 2975 |
+
{
|
| 2976 |
+
"epoch": 0.12057142857142857,
|
| 2977 |
+
"grad_norm": 1.0411007404327393,
|
| 2978 |
+
"learning_rate": 0.001,
|
| 2979 |
+
"loss": 5.472476005554199,
|
| 2980 |
+
"step": 422
|
| 2981 |
+
},
|
| 2982 |
+
{
|
| 2983 |
+
"epoch": 0.12085714285714286,
|
| 2984 |
+
"grad_norm": 1.1856043338775635,
|
| 2985 |
+
"learning_rate": 0.001,
|
| 2986 |
+
"loss": 5.41392183303833,
|
| 2987 |
+
"step": 423
|
| 2988 |
+
},
|
| 2989 |
+
{
|
| 2990 |
+
"epoch": 0.12114285714285715,
|
| 2991 |
+
"grad_norm": 1.1088652610778809,
|
| 2992 |
+
"learning_rate": 0.001,
|
| 2993 |
+
"loss": 5.409942626953125,
|
| 2994 |
+
"step": 424
|
| 2995 |
+
},
|
| 2996 |
+
{
|
| 2997 |
+
"epoch": 0.12142857142857143,
|
| 2998 |
+
"grad_norm": 1.1382297277450562,
|
| 2999 |
+
"learning_rate": 0.001,
|
| 3000 |
+
"loss": 5.413122177124023,
|
| 3001 |
+
"step": 425
|
| 3002 |
+
},
|
| 3003 |
+
{
|
| 3004 |
+
"epoch": 0.12171428571428572,
|
| 3005 |
+
"grad_norm": 1.060284972190857,
|
| 3006 |
+
"learning_rate": 0.001,
|
| 3007 |
+
"loss": 5.390390872955322,
|
| 3008 |
+
"step": 426
|
| 3009 |
+
},
|
| 3010 |
+
{
|
| 3011 |
+
"epoch": 0.122,
|
| 3012 |
+
"grad_norm": 1.0070810317993164,
|
| 3013 |
+
"learning_rate": 0.001,
|
| 3014 |
+
"loss": 5.417789936065674,
|
| 3015 |
+
"step": 427
|
| 3016 |
+
},
|
| 3017 |
+
{
|
| 3018 |
+
"epoch": 0.12228571428571429,
|
| 3019 |
+
"grad_norm": 1.0594487190246582,
|
| 3020 |
+
"learning_rate": 0.001,
|
| 3021 |
+
"loss": 5.350903511047363,
|
| 3022 |
+
"step": 428
|
| 3023 |
+
},
|
| 3024 |
+
{
|
| 3025 |
+
"epoch": 0.12257142857142857,
|
| 3026 |
+
"grad_norm": 1.1363307237625122,
|
| 3027 |
+
"learning_rate": 0.001,
|
| 3028 |
+
"loss": 5.364880084991455,
|
| 3029 |
+
"step": 429
|
| 3030 |
+
},
|
| 3031 |
+
{
|
| 3032 |
+
"epoch": 0.12285714285714286,
|
| 3033 |
+
"grad_norm": 1.1017429828643799,
|
| 3034 |
+
"learning_rate": 0.001,
|
| 3035 |
+
"loss": 5.394749164581299,
|
| 3036 |
+
"step": 430
|
| 3037 |
+
},
|
| 3038 |
+
{
|
| 3039 |
+
"epoch": 0.12314285714285714,
|
| 3040 |
+
"grad_norm": 1.138629674911499,
|
| 3041 |
+
"learning_rate": 0.001,
|
| 3042 |
+
"loss": 5.366091251373291,
|
| 3043 |
+
"step": 431
|
| 3044 |
+
},
|
| 3045 |
+
{
|
| 3046 |
+
"epoch": 0.12342857142857143,
|
| 3047 |
+
"grad_norm": 1.1799721717834473,
|
| 3048 |
+
"learning_rate": 0.001,
|
| 3049 |
+
"loss": 5.366257667541504,
|
| 3050 |
+
"step": 432
|
| 3051 |
+
},
|
| 3052 |
+
{
|
| 3053 |
+
"epoch": 0.12371428571428572,
|
| 3054 |
+
"grad_norm": 1.0385410785675049,
|
| 3055 |
+
"learning_rate": 0.001,
|
| 3056 |
+
"loss": 5.342431545257568,
|
| 3057 |
+
"step": 433
|
| 3058 |
+
},
|
| 3059 |
+
{
|
| 3060 |
+
"epoch": 0.124,
|
| 3061 |
+
"grad_norm": 1.051072359085083,
|
| 3062 |
+
"learning_rate": 0.001,
|
| 3063 |
+
"loss": 5.35132360458374,
|
| 3064 |
+
"step": 434
|
| 3065 |
+
},
|
| 3066 |
+
{
|
| 3067 |
+
"epoch": 0.12428571428571429,
|
| 3068 |
+
"grad_norm": 1.0489829778671265,
|
| 3069 |
+
"learning_rate": 0.001,
|
| 3070 |
+
"loss": 5.30415678024292,
|
| 3071 |
+
"step": 435
|
| 3072 |
+
},
|
| 3073 |
+
{
|
| 3074 |
+
"epoch": 0.12457142857142857,
|
| 3075 |
+
"grad_norm": 1.0064418315887451,
|
| 3076 |
+
"learning_rate": 0.001,
|
| 3077 |
+
"loss": 5.359162330627441,
|
| 3078 |
+
"step": 436
|
| 3079 |
+
},
|
| 3080 |
+
{
|
| 3081 |
+
"epoch": 0.12485714285714286,
|
| 3082 |
+
"grad_norm": 1.045552134513855,
|
| 3083 |
+
"learning_rate": 0.001,
|
| 3084 |
+
"loss": 5.32468318939209,
|
| 3085 |
+
"step": 437
|
| 3086 |
+
},
|
| 3087 |
+
{
|
| 3088 |
+
"epoch": 0.12514285714285714,
|
| 3089 |
+
"grad_norm": 1.1107268333435059,
|
| 3090 |
+
"learning_rate": 0.001,
|
| 3091 |
+
"loss": 5.395519256591797,
|
| 3092 |
+
"step": 438
|
| 3093 |
+
},
|
| 3094 |
+
{
|
| 3095 |
+
"epoch": 0.12542857142857142,
|
| 3096 |
+
"grad_norm": 1.0607658624649048,
|
| 3097 |
+
"learning_rate": 0.001,
|
| 3098 |
+
"loss": 5.265985488891602,
|
| 3099 |
+
"step": 439
|
| 3100 |
+
},
|
| 3101 |
+
{
|
| 3102 |
+
"epoch": 0.12571428571428572,
|
| 3103 |
+
"grad_norm": 1.0155754089355469,
|
| 3104 |
+
"learning_rate": 0.001,
|
| 3105 |
+
"loss": 5.302718162536621,
|
| 3106 |
+
"step": 440
|
| 3107 |
+
},
|
| 3108 |
+
{
|
| 3109 |
+
"epoch": 0.126,
|
| 3110 |
+
"grad_norm": 1.0499167442321777,
|
| 3111 |
+
"learning_rate": 0.001,
|
| 3112 |
+
"loss": 5.3036274909973145,
|
| 3113 |
+
"step": 441
|
| 3114 |
+
},
|
| 3115 |
+
{
|
| 3116 |
+
"epoch": 0.12628571428571428,
|
| 3117 |
+
"grad_norm": 0.9737458229064941,
|
| 3118 |
+
"learning_rate": 0.001,
|
| 3119 |
+
"loss": 5.297948360443115,
|
| 3120 |
+
"step": 442
|
| 3121 |
+
},
|
| 3122 |
+
{
|
| 3123 |
+
"epoch": 0.12657142857142858,
|
| 3124 |
+
"grad_norm": 1.0175093412399292,
|
| 3125 |
+
"learning_rate": 0.001,
|
| 3126 |
+
"loss": 5.300426483154297,
|
| 3127 |
+
"step": 443
|
| 3128 |
+
},
|
| 3129 |
+
{
|
| 3130 |
+
"epoch": 0.12685714285714286,
|
| 3131 |
+
"grad_norm": 1.1759352684020996,
|
| 3132 |
+
"learning_rate": 0.001,
|
| 3133 |
+
"loss": 5.339329242706299,
|
| 3134 |
+
"step": 444
|
| 3135 |
+
},
|
| 3136 |
+
{
|
| 3137 |
+
"epoch": 0.12714285714285714,
|
| 3138 |
+
"grad_norm": 1.0861457586288452,
|
| 3139 |
+
"learning_rate": 0.001,
|
| 3140 |
+
"loss": 5.220453262329102,
|
| 3141 |
+
"step": 445
|
| 3142 |
+
},
|
| 3143 |
+
{
|
| 3144 |
+
"epoch": 0.12742857142857142,
|
| 3145 |
+
"grad_norm": 0.9744501709938049,
|
| 3146 |
+
"learning_rate": 0.001,
|
| 3147 |
+
"loss": 5.312012195587158,
|
| 3148 |
+
"step": 446
|
| 3149 |
+
},
|
| 3150 |
+
{
|
| 3151 |
+
"epoch": 0.12771428571428572,
|
| 3152 |
+
"grad_norm": 0.9547734260559082,
|
| 3153 |
+
"learning_rate": 0.001,
|
| 3154 |
+
"loss": 5.241524696350098,
|
| 3155 |
+
"step": 447
|
| 3156 |
+
},
|
| 3157 |
+
{
|
| 3158 |
+
"epoch": 0.128,
|
| 3159 |
+
"grad_norm": 1.0768176317214966,
|
| 3160 |
+
"learning_rate": 0.001,
|
| 3161 |
+
"loss": 5.31060791015625,
|
| 3162 |
+
"step": 448
|
| 3163 |
+
},
|
| 3164 |
+
{
|
| 3165 |
+
"epoch": 0.12828571428571428,
|
| 3166 |
+
"grad_norm": 1.01326584815979,
|
| 3167 |
+
"learning_rate": 0.001,
|
| 3168 |
+
"loss": 5.250446796417236,
|
| 3169 |
+
"step": 449
|
| 3170 |
+
},
|
| 3171 |
+
{
|
| 3172 |
+
"epoch": 0.12857142857142856,
|
| 3173 |
+
"grad_norm": 1.0076360702514648,
|
| 3174 |
+
"learning_rate": 0.001,
|
| 3175 |
+
"loss": 5.276961803436279,
|
| 3176 |
+
"step": 450
|
| 3177 |
+
},
|
| 3178 |
+
{
|
| 3179 |
+
"epoch": 0.12885714285714286,
|
| 3180 |
+
"grad_norm": 0.969070315361023,
|
| 3181 |
+
"learning_rate": 0.001,
|
| 3182 |
+
"loss": 5.206024169921875,
|
| 3183 |
+
"step": 451
|
| 3184 |
+
},
|
| 3185 |
+
{
|
| 3186 |
+
"epoch": 0.12914285714285714,
|
| 3187 |
+
"grad_norm": 1.0394223928451538,
|
| 3188 |
+
"learning_rate": 0.001,
|
| 3189 |
+
"loss": 5.295047760009766,
|
| 3190 |
+
"step": 452
|
| 3191 |
+
},
|
| 3192 |
+
{
|
| 3193 |
+
"epoch": 0.12942857142857142,
|
| 3194 |
+
"grad_norm": 0.9847224354743958,
|
| 3195 |
+
"learning_rate": 0.001,
|
| 3196 |
+
"loss": 5.222926139831543,
|
| 3197 |
+
"step": 453
|
| 3198 |
+
},
|
| 3199 |
+
{
|
| 3200 |
+
"epoch": 0.12971428571428573,
|
| 3201 |
+
"grad_norm": 0.9708535671234131,
|
| 3202 |
+
"learning_rate": 0.001,
|
| 3203 |
+
"loss": 5.248390197753906,
|
| 3204 |
+
"step": 454
|
| 3205 |
+
},
|
| 3206 |
+
{
|
| 3207 |
+
"epoch": 0.13,
|
| 3208 |
+
"grad_norm": 0.9936147928237915,
|
| 3209 |
+
"learning_rate": 0.001,
|
| 3210 |
+
"loss": 5.206098556518555,
|
| 3211 |
+
"step": 455
|
| 3212 |
+
},
|
| 3213 |
+
{
|
| 3214 |
+
"epoch": 0.13028571428571428,
|
| 3215 |
+
"grad_norm": 0.9675630927085876,
|
| 3216 |
+
"learning_rate": 0.001,
|
| 3217 |
+
"loss": 5.243976593017578,
|
| 3218 |
+
"step": 456
|
| 3219 |
+
},
|
| 3220 |
+
{
|
| 3221 |
+
"epoch": 0.13057142857142856,
|
| 3222 |
+
"grad_norm": 1.0233958959579468,
|
| 3223 |
+
"learning_rate": 0.001,
|
| 3224 |
+
"loss": 5.20659065246582,
|
| 3225 |
+
"step": 457
|
| 3226 |
+
},
|
| 3227 |
+
{
|
| 3228 |
+
"epoch": 0.13085714285714287,
|
| 3229 |
+
"grad_norm": 0.9825986623764038,
|
| 3230 |
+
"learning_rate": 0.001,
|
| 3231 |
+
"loss": 5.270219802856445,
|
| 3232 |
+
"step": 458
|
| 3233 |
+
},
|
| 3234 |
+
{
|
| 3235 |
+
"epoch": 0.13114285714285714,
|
| 3236 |
+
"grad_norm": 0.9673742055892944,
|
| 3237 |
+
"learning_rate": 0.001,
|
| 3238 |
+
"loss": 5.157027721405029,
|
| 3239 |
+
"step": 459
|
| 3240 |
+
},
|
| 3241 |
+
{
|
| 3242 |
+
"epoch": 0.13142857142857142,
|
| 3243 |
+
"grad_norm": 1.0010325908660889,
|
| 3244 |
+
"learning_rate": 0.001,
|
| 3245 |
+
"loss": 5.232192516326904,
|
| 3246 |
+
"step": 460
|
| 3247 |
+
},
|
| 3248 |
+
{
|
| 3249 |
+
"epoch": 0.1317142857142857,
|
| 3250 |
+
"grad_norm": 1.0134440660476685,
|
| 3251 |
+
"learning_rate": 0.001,
|
| 3252 |
+
"loss": 5.191030979156494,
|
| 3253 |
+
"step": 461
|
| 3254 |
+
},
|
| 3255 |
+
{
|
| 3256 |
+
"epoch": 0.132,
|
| 3257 |
+
"grad_norm": 0.959584653377533,
|
| 3258 |
+
"learning_rate": 0.001,
|
| 3259 |
+
"loss": 5.21228551864624,
|
| 3260 |
+
"step": 462
|
| 3261 |
+
},
|
| 3262 |
+
{
|
| 3263 |
+
"epoch": 0.13228571428571428,
|
| 3264 |
+
"grad_norm": 1.0412331819534302,
|
| 3265 |
+
"learning_rate": 0.001,
|
| 3266 |
+
"loss": 5.119292259216309,
|
| 3267 |
+
"step": 463
|
| 3268 |
+
},
|
| 3269 |
+
{
|
| 3270 |
+
"epoch": 0.13257142857142856,
|
| 3271 |
+
"grad_norm": 0.9348019957542419,
|
| 3272 |
+
"learning_rate": 0.001,
|
| 3273 |
+
"loss": 5.231322765350342,
|
| 3274 |
+
"step": 464
|
| 3275 |
+
},
|
| 3276 |
+
{
|
| 3277 |
+
"epoch": 0.13285714285714287,
|
| 3278 |
+
"grad_norm": 0.968283474445343,
|
| 3279 |
+
"learning_rate": 0.001,
|
| 3280 |
+
"loss": 5.146182060241699,
|
| 3281 |
+
"step": 465
|
| 3282 |
+
},
|
| 3283 |
+
{
|
| 3284 |
+
"epoch": 0.13314285714285715,
|
| 3285 |
+
"grad_norm": 0.9431299567222595,
|
| 3286 |
+
"learning_rate": 0.001,
|
| 3287 |
+
"loss": 5.187849998474121,
|
| 3288 |
+
"step": 466
|
| 3289 |
+
},
|
| 3290 |
+
{
|
| 3291 |
+
"epoch": 0.13342857142857142,
|
| 3292 |
+
"grad_norm": 0.9716904759407043,
|
| 3293 |
+
"learning_rate": 0.001,
|
| 3294 |
+
"loss": 5.079883575439453,
|
| 3295 |
+
"step": 467
|
| 3296 |
+
},
|
| 3297 |
+
{
|
| 3298 |
+
"epoch": 0.1337142857142857,
|
| 3299 |
+
"grad_norm": 0.9341455101966858,
|
| 3300 |
+
"learning_rate": 0.001,
|
| 3301 |
+
"loss": 5.129624366760254,
|
| 3302 |
+
"step": 468
|
| 3303 |
+
},
|
| 3304 |
+
{
|
| 3305 |
+
"epoch": 0.134,
|
| 3306 |
+
"grad_norm": 0.9602625966072083,
|
| 3307 |
+
"learning_rate": 0.001,
|
| 3308 |
+
"loss": 5.135655403137207,
|
| 3309 |
+
"step": 469
|
| 3310 |
+
},
|
| 3311 |
+
{
|
| 3312 |
+
"epoch": 0.13428571428571429,
|
| 3313 |
+
"grad_norm": 1.1513607501983643,
|
| 3314 |
+
"learning_rate": 0.001,
|
| 3315 |
+
"loss": 5.141994476318359,
|
| 3316 |
+
"step": 470
|
| 3317 |
+
},
|
| 3318 |
+
{
|
| 3319 |
+
"epoch": 0.13457142857142856,
|
| 3320 |
+
"grad_norm": 0.935822069644928,
|
| 3321 |
+
"learning_rate": 0.001,
|
| 3322 |
+
"loss": 5.1337385177612305,
|
| 3323 |
+
"step": 471
|
| 3324 |
+
},
|
| 3325 |
+
{
|
| 3326 |
+
"epoch": 0.13485714285714287,
|
| 3327 |
+
"grad_norm": 0.9868633151054382,
|
| 3328 |
+
"learning_rate": 0.001,
|
| 3329 |
+
"loss": 5.136781692504883,
|
| 3330 |
+
"step": 472
|
| 3331 |
+
},
|
| 3332 |
+
{
|
| 3333 |
+
"epoch": 0.13514285714285715,
|
| 3334 |
+
"grad_norm": 0.9418619275093079,
|
| 3335 |
+
"learning_rate": 0.001,
|
| 3336 |
+
"loss": 5.1065449714660645,
|
| 3337 |
+
"step": 473
|
| 3338 |
+
},
|
| 3339 |
+
{
|
| 3340 |
+
"epoch": 0.13542857142857143,
|
| 3341 |
+
"grad_norm": 0.947436511516571,
|
| 3342 |
+
"learning_rate": 0.001,
|
| 3343 |
+
"loss": 5.105165004730225,
|
| 3344 |
+
"step": 474
|
| 3345 |
+
},
|
| 3346 |
+
{
|
| 3347 |
+
"epoch": 0.1357142857142857,
|
| 3348 |
+
"grad_norm": 1.02706778049469,
|
| 3349 |
+
"learning_rate": 0.001,
|
| 3350 |
+
"loss": 5.121386528015137,
|
| 3351 |
+
"step": 475
|
| 3352 |
+
},
|
| 3353 |
+
{
|
| 3354 |
+
"epoch": 0.136,
|
| 3355 |
+
"grad_norm": 0.9679945707321167,
|
| 3356 |
+
"learning_rate": 0.001,
|
| 3357 |
+
"loss": 5.128341197967529,
|
| 3358 |
+
"step": 476
|
| 3359 |
+
},
|
| 3360 |
+
{
|
| 3361 |
+
"epoch": 0.1362857142857143,
|
| 3362 |
+
"grad_norm": 0.9815688729286194,
|
| 3363 |
+
"learning_rate": 0.001,
|
| 3364 |
+
"loss": 5.099264144897461,
|
| 3365 |
+
"step": 477
|
| 3366 |
+
},
|
| 3367 |
+
{
|
| 3368 |
+
"epoch": 0.13657142857142857,
|
| 3369 |
+
"grad_norm": 0.9458716511726379,
|
| 3370 |
+
"learning_rate": 0.001,
|
| 3371 |
+
"loss": 5.144362926483154,
|
| 3372 |
+
"step": 478
|
| 3373 |
+
},
|
| 3374 |
+
{
|
| 3375 |
+
"epoch": 0.13685714285714284,
|
| 3376 |
+
"grad_norm": 0.9795572757720947,
|
| 3377 |
+
"learning_rate": 0.001,
|
| 3378 |
+
"loss": 5.091696739196777,
|
| 3379 |
+
"step": 479
|
| 3380 |
+
},
|
| 3381 |
+
{
|
| 3382 |
+
"epoch": 0.13714285714285715,
|
| 3383 |
+
"grad_norm": 0.9244929552078247,
|
| 3384 |
+
"learning_rate": 0.001,
|
| 3385 |
+
"loss": 5.091777801513672,
|
| 3386 |
+
"step": 480
|
| 3387 |
+
},
|
| 3388 |
+
{
|
| 3389 |
+
"epoch": 0.13742857142857143,
|
| 3390 |
+
"grad_norm": 0.9584999680519104,
|
| 3391 |
+
"learning_rate": 0.001,
|
| 3392 |
+
"loss": 5.077735900878906,
|
| 3393 |
+
"step": 481
|
| 3394 |
+
},
|
| 3395 |
+
{
|
| 3396 |
+
"epoch": 0.1377142857142857,
|
| 3397 |
+
"grad_norm": 0.9418224692344666,
|
| 3398 |
+
"learning_rate": 0.001,
|
| 3399 |
+
"loss": 5.0613484382629395,
|
| 3400 |
+
"step": 482
|
| 3401 |
+
},
|
| 3402 |
+
{
|
| 3403 |
+
"epoch": 0.138,
|
| 3404 |
+
"grad_norm": 0.965401291847229,
|
| 3405 |
+
"learning_rate": 0.001,
|
| 3406 |
+
"loss": 5.018246650695801,
|
| 3407 |
+
"step": 483
|
| 3408 |
+
},
|
| 3409 |
+
{
|
| 3410 |
+
"epoch": 0.1382857142857143,
|
| 3411 |
+
"grad_norm": 1.0144182443618774,
|
| 3412 |
+
"learning_rate": 0.001,
|
| 3413 |
+
"loss": 5.066267013549805,
|
| 3414 |
+
"step": 484
|
| 3415 |
+
},
|
| 3416 |
+
{
|
| 3417 |
+
"epoch": 0.13857142857142857,
|
| 3418 |
+
"grad_norm": 0.890421986579895,
|
| 3419 |
+
"learning_rate": 0.001,
|
| 3420 |
+
"loss": 5.023804664611816,
|
| 3421 |
+
"step": 485
|
| 3422 |
+
},
|
| 3423 |
+
{
|
| 3424 |
+
"epoch": 0.13885714285714285,
|
| 3425 |
+
"grad_norm": 0.945013165473938,
|
| 3426 |
+
"learning_rate": 0.001,
|
| 3427 |
+
"loss": 5.0201311111450195,
|
| 3428 |
+
"step": 486
|
| 3429 |
+
},
|
| 3430 |
+
{
|
| 3431 |
+
"epoch": 0.13914285714285715,
|
| 3432 |
+
"grad_norm": 1.0092036724090576,
|
| 3433 |
+
"learning_rate": 0.001,
|
| 3434 |
+
"loss": 5.059299945831299,
|
| 3435 |
+
"step": 487
|
| 3436 |
+
},
|
| 3437 |
+
{
|
| 3438 |
+
"epoch": 0.13942857142857143,
|
| 3439 |
+
"grad_norm": 0.9381226301193237,
|
| 3440 |
+
"learning_rate": 0.001,
|
| 3441 |
+
"loss": 5.054329872131348,
|
| 3442 |
+
"step": 488
|
| 3443 |
+
},
|
| 3444 |
+
{
|
| 3445 |
+
"epoch": 0.1397142857142857,
|
| 3446 |
+
"grad_norm": 0.9205765724182129,
|
| 3447 |
+
"learning_rate": 0.001,
|
| 3448 |
+
"loss": 5.034371376037598,
|
| 3449 |
+
"step": 489
|
| 3450 |
+
},
|
| 3451 |
+
{
|
| 3452 |
+
"epoch": 0.14,
|
| 3453 |
+
"grad_norm": 0.9698334336280823,
|
| 3454 |
+
"learning_rate": 0.001,
|
| 3455 |
+
"loss": 5.058368682861328,
|
| 3456 |
+
"step": 490
|
| 3457 |
+
},
|
| 3458 |
+
{
|
| 3459 |
+
"epoch": 0.1402857142857143,
|
| 3460 |
+
"grad_norm": 1.0283845663070679,
|
| 3461 |
+
"learning_rate": 0.001,
|
| 3462 |
+
"loss": 5.019568920135498,
|
| 3463 |
+
"step": 491
|
| 3464 |
+
},
|
| 3465 |
+
{
|
| 3466 |
+
"epoch": 0.14057142857142857,
|
| 3467 |
+
"grad_norm": 0.9763932824134827,
|
| 3468 |
+
"learning_rate": 0.001,
|
| 3469 |
+
"loss": 5.045557022094727,
|
| 3470 |
+
"step": 492
|
| 3471 |
+
},
|
| 3472 |
+
{
|
| 3473 |
+
"epoch": 0.14085714285714285,
|
| 3474 |
+
"grad_norm": 1.035258412361145,
|
| 3475 |
+
"learning_rate": 0.001,
|
| 3476 |
+
"loss": 5.058183193206787,
|
| 3477 |
+
"step": 493
|
| 3478 |
+
},
|
| 3479 |
+
{
|
| 3480 |
+
"epoch": 0.14114285714285715,
|
| 3481 |
+
"grad_norm": 1.00905179977417,
|
| 3482 |
+
"learning_rate": 0.001,
|
| 3483 |
+
"loss": 4.991766929626465,
|
| 3484 |
+
"step": 494
|
| 3485 |
+
},
|
| 3486 |
+
{
|
| 3487 |
+
"epoch": 0.14142857142857143,
|
| 3488 |
+
"grad_norm": 0.9244449138641357,
|
| 3489 |
+
"learning_rate": 0.001,
|
| 3490 |
+
"loss": 4.998722553253174,
|
| 3491 |
+
"step": 495
|
| 3492 |
+
},
|
| 3493 |
+
{
|
| 3494 |
+
"epoch": 0.1417142857142857,
|
| 3495 |
+
"grad_norm": 0.9540549516677856,
|
| 3496 |
+
"learning_rate": 0.001,
|
| 3497 |
+
"loss": 4.964876651763916,
|
| 3498 |
+
"step": 496
|
| 3499 |
+
},
|
| 3500 |
+
{
|
| 3501 |
+
"epoch": 0.142,
|
| 3502 |
+
"grad_norm": 0.8989408612251282,
|
| 3503 |
+
"learning_rate": 0.001,
|
| 3504 |
+
"loss": 4.988190650939941,
|
| 3505 |
+
"step": 497
|
| 3506 |
+
},
|
| 3507 |
+
{
|
| 3508 |
+
"epoch": 0.1422857142857143,
|
| 3509 |
+
"grad_norm": 0.964568555355072,
|
| 3510 |
+
"learning_rate": 0.001,
|
| 3511 |
+
"loss": 5.018099784851074,
|
| 3512 |
+
"step": 498
|
| 3513 |
+
},
|
| 3514 |
+
{
|
| 3515 |
+
"epoch": 0.14257142857142857,
|
| 3516 |
+
"grad_norm": 0.9217190742492676,
|
| 3517 |
+
"learning_rate": 0.001,
|
| 3518 |
+
"loss": 5.0018815994262695,
|
| 3519 |
+
"step": 499
|
| 3520 |
+
},
|
| 3521 |
+
{
|
| 3522 |
+
"epoch": 0.14285714285714285,
|
| 3523 |
+
"grad_norm": 0.9222206473350525,
|
| 3524 |
+
"learning_rate": 0.001,
|
| 3525 |
+
"loss": 4.977924346923828,
|
| 3526 |
+
"step": 500
|
| 3527 |
+
},
|
| 3528 |
+
{
|
| 3529 |
+
"epoch": 0.14314285714285716,
|
| 3530 |
+
"grad_norm": 0.9363248944282532,
|
| 3531 |
+
"learning_rate": 0.001,
|
| 3532 |
+
"loss": 4.953252792358398,
|
| 3533 |
+
"step": 501
|
| 3534 |
+
},
|
| 3535 |
+
{
|
| 3536 |
+
"epoch": 0.14342857142857143,
|
| 3537 |
+
"grad_norm": 1.0123873949050903,
|
| 3538 |
+
"learning_rate": 0.001,
|
| 3539 |
+
"loss": 4.943054676055908,
|
| 3540 |
+
"step": 502
|
| 3541 |
+
},
|
| 3542 |
+
{
|
| 3543 |
+
"epoch": 0.1437142857142857,
|
| 3544 |
+
"grad_norm": 0.9083608388900757,
|
| 3545 |
+
"learning_rate": 0.001,
|
| 3546 |
+
"loss": 5.012940883636475,
|
| 3547 |
+
"step": 503
|
| 3548 |
+
},
|
| 3549 |
+
{
|
| 3550 |
+
"epoch": 0.144,
|
| 3551 |
+
"grad_norm": 0.882636547088623,
|
| 3552 |
+
"learning_rate": 0.001,
|
| 3553 |
+
"loss": 4.9350666999816895,
|
| 3554 |
+
"step": 504
|
| 3555 |
+
},
|
| 3556 |
+
{
|
| 3557 |
+
"epoch": 0.1442857142857143,
|
| 3558 |
+
"grad_norm": 0.9688971042633057,
|
| 3559 |
+
"learning_rate": 0.001,
|
| 3560 |
+
"loss": 4.982104301452637,
|
| 3561 |
+
"step": 505
|
| 3562 |
+
},
|
| 3563 |
+
{
|
| 3564 |
+
"epoch": 0.14457142857142857,
|
| 3565 |
+
"grad_norm": 0.939300000667572,
|
| 3566 |
+
"learning_rate": 0.001,
|
| 3567 |
+
"loss": 4.938144207000732,
|
| 3568 |
+
"step": 506
|
| 3569 |
+
},
|
| 3570 |
+
{
|
| 3571 |
+
"epoch": 0.14485714285714285,
|
| 3572 |
+
"grad_norm": 1.0234988927841187,
|
| 3573 |
+
"learning_rate": 0.001,
|
| 3574 |
+
"loss": 4.9430251121521,
|
| 3575 |
+
"step": 507
|
| 3576 |
+
},
|
| 3577 |
+
{
|
| 3578 |
+
"epoch": 0.14514285714285713,
|
| 3579 |
+
"grad_norm": 0.9347419142723083,
|
| 3580 |
+
"learning_rate": 0.001,
|
| 3581 |
+
"loss": 4.955362796783447,
|
| 3582 |
+
"step": 508
|
| 3583 |
+
},
|
| 3584 |
+
{
|
| 3585 |
+
"epoch": 0.14542857142857143,
|
| 3586 |
+
"grad_norm": 0.9153620004653931,
|
| 3587 |
+
"learning_rate": 0.001,
|
| 3588 |
+
"loss": 4.918880462646484,
|
| 3589 |
+
"step": 509
|
| 3590 |
+
},
|
| 3591 |
+
{
|
| 3592 |
+
"epoch": 0.1457142857142857,
|
| 3593 |
+
"grad_norm": 0.9035690426826477,
|
| 3594 |
+
"learning_rate": 0.001,
|
| 3595 |
+
"loss": 4.959357738494873,
|
| 3596 |
+
"step": 510
|
| 3597 |
+
},
|
| 3598 |
+
{
|
| 3599 |
+
"epoch": 0.146,
|
| 3600 |
+
"grad_norm": 0.8607348203659058,
|
| 3601 |
+
"learning_rate": 0.001,
|
| 3602 |
+
"loss": 4.925543785095215,
|
| 3603 |
+
"step": 511
|
| 3604 |
+
},
|
| 3605 |
+
{
|
| 3606 |
+
"epoch": 0.1462857142857143,
|
| 3607 |
+
"grad_norm": 0.9107086658477783,
|
| 3608 |
+
"learning_rate": 0.001,
|
| 3609 |
+
"loss": 4.863113880157471,
|
| 3610 |
+
"step": 512
|
| 3611 |
+
},
|
| 3612 |
+
{
|
| 3613 |
+
"epoch": 0.14657142857142857,
|
| 3614 |
+
"grad_norm": 0.9495797157287598,
|
| 3615 |
+
"learning_rate": 0.001,
|
| 3616 |
+
"loss": 4.908524036407471,
|
| 3617 |
+
"step": 513
|
| 3618 |
+
},
|
| 3619 |
+
{
|
| 3620 |
+
"epoch": 0.14685714285714285,
|
| 3621 |
+
"grad_norm": 0.9713687300682068,
|
| 3622 |
+
"learning_rate": 0.001,
|
| 3623 |
+
"loss": 4.902448654174805,
|
| 3624 |
+
"step": 514
|
| 3625 |
+
},
|
| 3626 |
+
{
|
| 3627 |
+
"epoch": 0.14714285714285713,
|
| 3628 |
+
"grad_norm": 1.1436808109283447,
|
| 3629 |
+
"learning_rate": 0.001,
|
| 3630 |
+
"loss": 4.918895244598389,
|
| 3631 |
+
"step": 515
|
| 3632 |
+
},
|
| 3633 |
+
{
|
| 3634 |
+
"epoch": 0.14742857142857144,
|
| 3635 |
+
"grad_norm": 0.9062098264694214,
|
| 3636 |
+
"learning_rate": 0.001,
|
| 3637 |
+
"loss": 4.8549017906188965,
|
| 3638 |
+
"step": 516
|
| 3639 |
+
},
|
| 3640 |
+
{
|
| 3641 |
+
"epoch": 0.14771428571428571,
|
| 3642 |
+
"grad_norm": 0.9654607176780701,
|
| 3643 |
+
"learning_rate": 0.001,
|
| 3644 |
+
"loss": 4.919809818267822,
|
| 3645 |
+
"step": 517
|
| 3646 |
+
},
|
| 3647 |
+
{
|
| 3648 |
+
"epoch": 0.148,
|
| 3649 |
+
"grad_norm": 0.9809955954551697,
|
| 3650 |
+
"learning_rate": 0.001,
|
| 3651 |
+
"loss": 4.840274333953857,
|
| 3652 |
+
"step": 518
|
| 3653 |
+
},
|
| 3654 |
+
{
|
| 3655 |
+
"epoch": 0.1482857142857143,
|
| 3656 |
+
"grad_norm": 1.1917036771774292,
|
| 3657 |
+
"learning_rate": 0.001,
|
| 3658 |
+
"loss": 4.883162021636963,
|
| 3659 |
+
"step": 519
|
| 3660 |
+
},
|
| 3661 |
+
{
|
| 3662 |
+
"epoch": 0.14857142857142858,
|
| 3663 |
+
"grad_norm": 0.8811300992965698,
|
| 3664 |
+
"learning_rate": 0.001,
|
| 3665 |
+
"loss": 4.873466968536377,
|
| 3666 |
+
"step": 520
|
| 3667 |
+
},
|
| 3668 |
+
{
|
| 3669 |
+
"epoch": 0.14885714285714285,
|
| 3670 |
+
"grad_norm": 0.98333340883255,
|
| 3671 |
+
"learning_rate": 0.001,
|
| 3672 |
+
"loss": 4.905554294586182,
|
| 3673 |
+
"step": 521
|
| 3674 |
+
},
|
| 3675 |
+
{
|
| 3676 |
+
"epoch": 0.14914285714285713,
|
| 3677 |
+
"grad_norm": 0.8773059248924255,
|
| 3678 |
+
"learning_rate": 0.001,
|
| 3679 |
+
"loss": 4.782233238220215,
|
| 3680 |
+
"step": 522
|
| 3681 |
+
},
|
| 3682 |
+
{
|
| 3683 |
+
"epoch": 0.14942857142857144,
|
| 3684 |
+
"grad_norm": 0.8445463180541992,
|
| 3685 |
+
"learning_rate": 0.001,
|
| 3686 |
+
"loss": 4.898096084594727,
|
| 3687 |
+
"step": 523
|
| 3688 |
+
},
|
| 3689 |
+
{
|
| 3690 |
+
"epoch": 0.14971428571428572,
|
| 3691 |
+
"grad_norm": 0.912285327911377,
|
| 3692 |
+
"learning_rate": 0.001,
|
| 3693 |
+
"loss": 4.86422061920166,
|
| 3694 |
+
"step": 524
|
| 3695 |
+
},
|
| 3696 |
+
{
|
| 3697 |
+
"epoch": 0.15,
|
| 3698 |
+
"grad_norm": 1.040880560874939,
|
| 3699 |
+
"learning_rate": 0.001,
|
| 3700 |
+
"loss": 4.9074602127075195,
|
| 3701 |
+
"step": 525
|
| 3702 |
+
},
|
| 3703 |
+
{
|
| 3704 |
+
"epoch": 0.15,
|
| 3705 |
+
"eval_loss": 4.8575520515441895,
|
| 3706 |
+
"eval_runtime": 592.938,
|
| 3707 |
+
"eval_samples_per_second": 8.594,
|
| 3708 |
+
"eval_steps_per_second": 2.149,
|
| 3709 |
+
"step": 525
|
| 3710 |
}
|
| 3711 |
],
|
| 3712 |
"logging_steps": 1,
|
|
|
|
| 3726 |
"attributes": {}
|
| 3727 |
}
|
| 3728 |
},
|
| 3729 |
+
"total_flos": 7.38145133121289e+17,
|
| 3730 |
"train_batch_size": 16,
|
| 3731 |
"trial_name": null,
|
| 3732 |
"trial_params": null
|