Training in progress, step 1200, checkpoint
Browse files
last-checkpoint/adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 194563400
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f4b5f1eef8e7013f79a444f9ea028ef8ae496fd8b4c9d5bcc1c2b7042e6dfaa4
|
| 3 |
size 194563400
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 100255938
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d975d28d45a1322a651452aaaf97a8106605aeb8742d88cabc76f193e21cc481
|
| 3 |
size 100255938
|
last-checkpoint/rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14244
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7532d75c0ba8850f0b92d41ab0ce24556d470c4da7cda005b659f7bcc8bf4b4d
|
| 3 |
size 14244
|
last-checkpoint/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1064
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b20da93e50834667a3a10e5c0a114d4f1a917750b3a4009ff729ec7f84e3ff65
|
| 3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": 0.7061967849731445,
|
| 3 |
"best_model_checkpoint": "miner_id_24/checkpoint-200",
|
| 4 |
-
"epoch": 0.
|
| 5 |
"eval_steps": 200,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -7055,6 +7055,1414 @@
|
|
| 7055 |
"eval_samples_per_second": 10.131,
|
| 7056 |
"eval_steps_per_second": 5.072,
|
| 7057 |
"step": 1000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 7058 |
}
|
| 7059 |
],
|
| 7060 |
"logging_steps": 1,
|
|
@@ -7069,7 +8477,7 @@
|
|
| 7069 |
"early_stopping_threshold": 0.0
|
| 7070 |
},
|
| 7071 |
"attributes": {
|
| 7072 |
-
"early_stopping_patience_counter":
|
| 7073 |
}
|
| 7074 |
},
|
| 7075 |
"TrainerControl": {
|
|
@@ -7078,12 +8486,12 @@
|
|
| 7078 |
"should_evaluate": false,
|
| 7079 |
"should_log": false,
|
| 7080 |
"should_save": true,
|
| 7081 |
-
"should_training_stop":
|
| 7082 |
},
|
| 7083 |
"attributes": {}
|
| 7084 |
}
|
| 7085 |
},
|
| 7086 |
-
"total_flos": 1.
|
| 7087 |
"train_batch_size": 2,
|
| 7088 |
"trial_name": null,
|
| 7089 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": 0.7061967849731445,
|
| 3 |
"best_model_checkpoint": "miner_id_24/checkpoint-200",
|
| 4 |
+
"epoch": 0.13107054599074314,
|
| 5 |
"eval_steps": 200,
|
| 6 |
+
"global_step": 1200,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 7055 |
"eval_samples_per_second": 10.131,
|
| 7056 |
"eval_steps_per_second": 5.072,
|
| 7057 |
"step": 1000
|
| 7058 |
+
},
|
| 7059 |
+
{
|
| 7060 |
+
"epoch": 0.10933468044727823,
|
| 7061 |
+
"grad_norm": 1.2205915451049805,
|
| 7062 |
+
"learning_rate": 0.0001993824803270106,
|
| 7063 |
+
"loss": 0.7065,
|
| 7064 |
+
"step": 1001
|
| 7065 |
+
},
|
| 7066 |
+
{
|
| 7067 |
+
"epoch": 0.10944390590227053,
|
| 7068 |
+
"grad_norm": 1.3085845708847046,
|
| 7069 |
+
"learning_rate": 0.00019938120906093872,
|
| 7070 |
+
"loss": 0.9705,
|
| 7071 |
+
"step": 1002
|
| 7072 |
+
},
|
| 7073 |
+
{
|
| 7074 |
+
"epoch": 0.10955313135726281,
|
| 7075 |
+
"grad_norm": 1.2457107305526733,
|
| 7076 |
+
"learning_rate": 0.0001993799364917178,
|
| 7077 |
+
"loss": 0.6375,
|
| 7078 |
+
"step": 1003
|
| 7079 |
+
},
|
| 7080 |
+
{
|
| 7081 |
+
"epoch": 0.1096623568122551,
|
| 7082 |
+
"grad_norm": 1.124176025390625,
|
| 7083 |
+
"learning_rate": 0.00019937866261936452,
|
| 7084 |
+
"loss": 1.0772,
|
| 7085 |
+
"step": 1004
|
| 7086 |
+
},
|
| 7087 |
+
{
|
| 7088 |
+
"epoch": 0.10977158226724738,
|
| 7089 |
+
"grad_norm": 0.9982814788818359,
|
| 7090 |
+
"learning_rate": 0.00019937738744389558,
|
| 7091 |
+
"loss": 0.695,
|
| 7092 |
+
"step": 1005
|
| 7093 |
+
},
|
| 7094 |
+
{
|
| 7095 |
+
"epoch": 0.10988080772223967,
|
| 7096 |
+
"grad_norm": 1.127756118774414,
|
| 7097 |
+
"learning_rate": 0.00019937611096532772,
|
| 7098 |
+
"loss": 0.925,
|
| 7099 |
+
"step": 1006
|
| 7100 |
+
},
|
| 7101 |
+
{
|
| 7102 |
+
"epoch": 0.10999003317723195,
|
| 7103 |
+
"grad_norm": 1.3581675291061401,
|
| 7104 |
+
"learning_rate": 0.00019937483318367762,
|
| 7105 |
+
"loss": 0.9221,
|
| 7106 |
+
"step": 1007
|
| 7107 |
+
},
|
| 7108 |
+
{
|
| 7109 |
+
"epoch": 0.11009925863222424,
|
| 7110 |
+
"grad_norm": 1.3347697257995605,
|
| 7111 |
+
"learning_rate": 0.0001993735540989621,
|
| 7112 |
+
"loss": 0.9107,
|
| 7113 |
+
"step": 1008
|
| 7114 |
+
},
|
| 7115 |
+
{
|
| 7116 |
+
"epoch": 0.11020848408721652,
|
| 7117 |
+
"grad_norm": 0.9323390126228333,
|
| 7118 |
+
"learning_rate": 0.0001993722737111979,
|
| 7119 |
+
"loss": 0.948,
|
| 7120 |
+
"step": 1009
|
| 7121 |
+
},
|
| 7122 |
+
{
|
| 7123 |
+
"epoch": 0.1103177095422088,
|
| 7124 |
+
"grad_norm": 1.101404070854187,
|
| 7125 |
+
"learning_rate": 0.00019937099202040184,
|
| 7126 |
+
"loss": 0.8408,
|
| 7127 |
+
"step": 1010
|
| 7128 |
+
},
|
| 7129 |
+
{
|
| 7130 |
+
"epoch": 0.1104269349972011,
|
| 7131 |
+
"grad_norm": 1.327126383781433,
|
| 7132 |
+
"learning_rate": 0.00019936970902659064,
|
| 7133 |
+
"loss": 0.7098,
|
| 7134 |
+
"step": 1011
|
| 7135 |
+
},
|
| 7136 |
+
{
|
| 7137 |
+
"epoch": 0.11053616045219339,
|
| 7138 |
+
"grad_norm": 1.6174031496047974,
|
| 7139 |
+
"learning_rate": 0.00019936842472978125,
|
| 7140 |
+
"loss": 1.0018,
|
| 7141 |
+
"step": 1012
|
| 7142 |
+
},
|
| 7143 |
+
{
|
| 7144 |
+
"epoch": 0.11064538590718567,
|
| 7145 |
+
"grad_norm": 1.1017050743103027,
|
| 7146 |
+
"learning_rate": 0.0001993671391299904,
|
| 7147 |
+
"loss": 0.7131,
|
| 7148 |
+
"step": 1013
|
| 7149 |
+
},
|
| 7150 |
+
{
|
| 7151 |
+
"epoch": 0.11075461136217796,
|
| 7152 |
+
"grad_norm": 1.4763091802597046,
|
| 7153 |
+
"learning_rate": 0.00019936585222723503,
|
| 7154 |
+
"loss": 0.5779,
|
| 7155 |
+
"step": 1014
|
| 7156 |
+
},
|
| 7157 |
+
{
|
| 7158 |
+
"epoch": 0.11086383681717024,
|
| 7159 |
+
"grad_norm": 1.215286135673523,
|
| 7160 |
+
"learning_rate": 0.00019936456402153195,
|
| 7161 |
+
"loss": 1.131,
|
| 7162 |
+
"step": 1015
|
| 7163 |
+
},
|
| 7164 |
+
{
|
| 7165 |
+
"epoch": 0.11097306227216253,
|
| 7166 |
+
"grad_norm": 1.2176015377044678,
|
| 7167 |
+
"learning_rate": 0.00019936327451289813,
|
| 7168 |
+
"loss": 0.8717,
|
| 7169 |
+
"step": 1016
|
| 7170 |
+
},
|
| 7171 |
+
{
|
| 7172 |
+
"epoch": 0.11108228772715481,
|
| 7173 |
+
"grad_norm": 1.297967553138733,
|
| 7174 |
+
"learning_rate": 0.00019936198370135036,
|
| 7175 |
+
"loss": 1.0072,
|
| 7176 |
+
"step": 1017
|
| 7177 |
+
},
|
| 7178 |
+
{
|
| 7179 |
+
"epoch": 0.1111915131821471,
|
| 7180 |
+
"grad_norm": 1.1566557884216309,
|
| 7181 |
+
"learning_rate": 0.00019936069158690568,
|
| 7182 |
+
"loss": 0.8495,
|
| 7183 |
+
"step": 1018
|
| 7184 |
+
},
|
| 7185 |
+
{
|
| 7186 |
+
"epoch": 0.11130073863713938,
|
| 7187 |
+
"grad_norm": 1.104823350906372,
|
| 7188 |
+
"learning_rate": 0.00019935939816958097,
|
| 7189 |
+
"loss": 0.9517,
|
| 7190 |
+
"step": 1019
|
| 7191 |
+
},
|
| 7192 |
+
{
|
| 7193 |
+
"epoch": 0.11140996409213168,
|
| 7194 |
+
"grad_norm": 1.3326419591903687,
|
| 7195 |
+
"learning_rate": 0.00019935810344939321,
|
| 7196 |
+
"loss": 1.0257,
|
| 7197 |
+
"step": 1020
|
| 7198 |
+
},
|
| 7199 |
+
{
|
| 7200 |
+
"epoch": 0.11151918954712396,
|
| 7201 |
+
"grad_norm": 1.1882355213165283,
|
| 7202 |
+
"learning_rate": 0.0001993568074263594,
|
| 7203 |
+
"loss": 0.9735,
|
| 7204 |
+
"step": 1021
|
| 7205 |
+
},
|
| 7206 |
+
{
|
| 7207 |
+
"epoch": 0.11162841500211625,
|
| 7208 |
+
"grad_norm": 1.0751303434371948,
|
| 7209 |
+
"learning_rate": 0.00019935551010049647,
|
| 7210 |
+
"loss": 0.6276,
|
| 7211 |
+
"step": 1022
|
| 7212 |
+
},
|
| 7213 |
+
{
|
| 7214 |
+
"epoch": 0.11173764045710853,
|
| 7215 |
+
"grad_norm": 1.2646431922912598,
|
| 7216 |
+
"learning_rate": 0.00019935421147182147,
|
| 7217 |
+
"loss": 0.7494,
|
| 7218 |
+
"step": 1023
|
| 7219 |
+
},
|
| 7220 |
+
{
|
| 7221 |
+
"epoch": 0.11184686591210082,
|
| 7222 |
+
"grad_norm": 1.2806205749511719,
|
| 7223 |
+
"learning_rate": 0.00019935291154035145,
|
| 7224 |
+
"loss": 1.0176,
|
| 7225 |
+
"step": 1024
|
| 7226 |
+
},
|
| 7227 |
+
{
|
| 7228 |
+
"epoch": 0.1119560913670931,
|
| 7229 |
+
"grad_norm": 1.3170644044876099,
|
| 7230 |
+
"learning_rate": 0.0001993516103061034,
|
| 7231 |
+
"loss": 1.1641,
|
| 7232 |
+
"step": 1025
|
| 7233 |
+
},
|
| 7234 |
+
{
|
| 7235 |
+
"epoch": 0.11206531682208538,
|
| 7236 |
+
"grad_norm": 1.271410346031189,
|
| 7237 |
+
"learning_rate": 0.00019935030776909446,
|
| 7238 |
+
"loss": 1.1157,
|
| 7239 |
+
"step": 1026
|
| 7240 |
+
},
|
| 7241 |
+
{
|
| 7242 |
+
"epoch": 0.11217454227707767,
|
| 7243 |
+
"grad_norm": 1.4487496614456177,
|
| 7244 |
+
"learning_rate": 0.00019934900392934167,
|
| 7245 |
+
"loss": 0.832,
|
| 7246 |
+
"step": 1027
|
| 7247 |
+
},
|
| 7248 |
+
{
|
| 7249 |
+
"epoch": 0.11228376773206995,
|
| 7250 |
+
"grad_norm": 1.3541709184646606,
|
| 7251 |
+
"learning_rate": 0.00019934769878686212,
|
| 7252 |
+
"loss": 0.722,
|
| 7253 |
+
"step": 1028
|
| 7254 |
+
},
|
| 7255 |
+
{
|
| 7256 |
+
"epoch": 0.11239299318706225,
|
| 7257 |
+
"grad_norm": 1.5500739812850952,
|
| 7258 |
+
"learning_rate": 0.0001993463923416729,
|
| 7259 |
+
"loss": 0.896,
|
| 7260 |
+
"step": 1029
|
| 7261 |
+
},
|
| 7262 |
+
{
|
| 7263 |
+
"epoch": 0.11250221864205454,
|
| 7264 |
+
"grad_norm": 1.046573519706726,
|
| 7265 |
+
"learning_rate": 0.00019934508459379117,
|
| 7266 |
+
"loss": 0.6733,
|
| 7267 |
+
"step": 1030
|
| 7268 |
+
},
|
| 7269 |
+
{
|
| 7270 |
+
"epoch": 0.11261144409704682,
|
| 7271 |
+
"grad_norm": 1.2942837476730347,
|
| 7272 |
+
"learning_rate": 0.0001993437755432341,
|
| 7273 |
+
"loss": 0.8235,
|
| 7274 |
+
"step": 1031
|
| 7275 |
+
},
|
| 7276 |
+
{
|
| 7277 |
+
"epoch": 0.1127206695520391,
|
| 7278 |
+
"grad_norm": 2.2381908893585205,
|
| 7279 |
+
"learning_rate": 0.00019934246519001883,
|
| 7280 |
+
"loss": 0.9397,
|
| 7281 |
+
"step": 1032
|
| 7282 |
+
},
|
| 7283 |
+
{
|
| 7284 |
+
"epoch": 0.11282989500703139,
|
| 7285 |
+
"grad_norm": 1.2537754774093628,
|
| 7286 |
+
"learning_rate": 0.0001993411535341625,
|
| 7287 |
+
"loss": 0.9351,
|
| 7288 |
+
"step": 1033
|
| 7289 |
+
},
|
| 7290 |
+
{
|
| 7291 |
+
"epoch": 0.11293912046202367,
|
| 7292 |
+
"grad_norm": 1.1798999309539795,
|
| 7293 |
+
"learning_rate": 0.00019933984057568242,
|
| 7294 |
+
"loss": 0.9141,
|
| 7295 |
+
"step": 1034
|
| 7296 |
+
},
|
| 7297 |
+
{
|
| 7298 |
+
"epoch": 0.11304834591701596,
|
| 7299 |
+
"grad_norm": 1.300783395767212,
|
| 7300 |
+
"learning_rate": 0.00019933852631459571,
|
| 7301 |
+
"loss": 0.7434,
|
| 7302 |
+
"step": 1035
|
| 7303 |
+
},
|
| 7304 |
+
{
|
| 7305 |
+
"epoch": 0.11315757137200824,
|
| 7306 |
+
"grad_norm": 1.6943469047546387,
|
| 7307 |
+
"learning_rate": 0.00019933721075091962,
|
| 7308 |
+
"loss": 1.4925,
|
| 7309 |
+
"step": 1036
|
| 7310 |
+
},
|
| 7311 |
+
{
|
| 7312 |
+
"epoch": 0.11326679682700053,
|
| 7313 |
+
"grad_norm": 1.3298465013504028,
|
| 7314 |
+
"learning_rate": 0.00019933589388467142,
|
| 7315 |
+
"loss": 0.7144,
|
| 7316 |
+
"step": 1037
|
| 7317 |
+
},
|
| 7318 |
+
{
|
| 7319 |
+
"epoch": 0.11337602228199282,
|
| 7320 |
+
"grad_norm": 1.3957929611206055,
|
| 7321 |
+
"learning_rate": 0.00019933457571586838,
|
| 7322 |
+
"loss": 1.335,
|
| 7323 |
+
"step": 1038
|
| 7324 |
+
},
|
| 7325 |
+
{
|
| 7326 |
+
"epoch": 0.11348524773698511,
|
| 7327 |
+
"grad_norm": 1.2654701471328735,
|
| 7328 |
+
"learning_rate": 0.00019933325624452776,
|
| 7329 |
+
"loss": 0.8992,
|
| 7330 |
+
"step": 1039
|
| 7331 |
+
},
|
| 7332 |
+
{
|
| 7333 |
+
"epoch": 0.1135944731919774,
|
| 7334 |
+
"grad_norm": 1.1495540142059326,
|
| 7335 |
+
"learning_rate": 0.00019933193547066688,
|
| 7336 |
+
"loss": 0.8485,
|
| 7337 |
+
"step": 1040
|
| 7338 |
+
},
|
| 7339 |
+
{
|
| 7340 |
+
"epoch": 0.11370369864696968,
|
| 7341 |
+
"grad_norm": 1.567428708076477,
|
| 7342 |
+
"learning_rate": 0.00019933061339430305,
|
| 7343 |
+
"loss": 0.8859,
|
| 7344 |
+
"step": 1041
|
| 7345 |
+
},
|
| 7346 |
+
{
|
| 7347 |
+
"epoch": 0.11381292410196196,
|
| 7348 |
+
"grad_norm": 1.3406460285186768,
|
| 7349 |
+
"learning_rate": 0.00019932929001545367,
|
| 7350 |
+
"loss": 0.9023,
|
| 7351 |
+
"step": 1042
|
| 7352 |
+
},
|
| 7353 |
+
{
|
| 7354 |
+
"epoch": 0.11392214955695425,
|
| 7355 |
+
"grad_norm": 1.2080565690994263,
|
| 7356 |
+
"learning_rate": 0.00019932796533413596,
|
| 7357 |
+
"loss": 0.5623,
|
| 7358 |
+
"step": 1043
|
| 7359 |
+
},
|
| 7360 |
+
{
|
| 7361 |
+
"epoch": 0.11403137501194653,
|
| 7362 |
+
"grad_norm": 1.5534919500350952,
|
| 7363 |
+
"learning_rate": 0.0001993266393503674,
|
| 7364 |
+
"loss": 0.7808,
|
| 7365 |
+
"step": 1044
|
| 7366 |
+
},
|
| 7367 |
+
{
|
| 7368 |
+
"epoch": 0.11414060046693882,
|
| 7369 |
+
"grad_norm": 1.3839794397354126,
|
| 7370 |
+
"learning_rate": 0.00019932531206416534,
|
| 7371 |
+
"loss": 0.8238,
|
| 7372 |
+
"step": 1045
|
| 7373 |
+
},
|
| 7374 |
+
{
|
| 7375 |
+
"epoch": 0.1142498259219311,
|
| 7376 |
+
"grad_norm": 1.5013834238052368,
|
| 7377 |
+
"learning_rate": 0.0001993239834755472,
|
| 7378 |
+
"loss": 0.7241,
|
| 7379 |
+
"step": 1046
|
| 7380 |
+
},
|
| 7381 |
+
{
|
| 7382 |
+
"epoch": 0.1143590513769234,
|
| 7383 |
+
"grad_norm": 1.3011348247528076,
|
| 7384 |
+
"learning_rate": 0.00019932265358453034,
|
| 7385 |
+
"loss": 0.6477,
|
| 7386 |
+
"step": 1047
|
| 7387 |
+
},
|
| 7388 |
+
{
|
| 7389 |
+
"epoch": 0.11446827683191568,
|
| 7390 |
+
"grad_norm": 1.2873181104660034,
|
| 7391 |
+
"learning_rate": 0.0001993213223911323,
|
| 7392 |
+
"loss": 0.8507,
|
| 7393 |
+
"step": 1048
|
| 7394 |
+
},
|
| 7395 |
+
{
|
| 7396 |
+
"epoch": 0.11457750228690797,
|
| 7397 |
+
"grad_norm": 1.3292242288589478,
|
| 7398 |
+
"learning_rate": 0.00019931998989537045,
|
| 7399 |
+
"loss": 0.8086,
|
| 7400 |
+
"step": 1049
|
| 7401 |
+
},
|
| 7402 |
+
{
|
| 7403 |
+
"epoch": 0.11468672774190025,
|
| 7404 |
+
"grad_norm": 1.728370189666748,
|
| 7405 |
+
"learning_rate": 0.0001993186560972623,
|
| 7406 |
+
"loss": 0.6108,
|
| 7407 |
+
"step": 1050
|
| 7408 |
+
},
|
| 7409 |
+
{
|
| 7410 |
+
"epoch": 0.11479595319689254,
|
| 7411 |
+
"grad_norm": 1.2135319709777832,
|
| 7412 |
+
"learning_rate": 0.00019931732099682533,
|
| 7413 |
+
"loss": 0.9925,
|
| 7414 |
+
"step": 1051
|
| 7415 |
+
},
|
| 7416 |
+
{
|
| 7417 |
+
"epoch": 0.11490517865188482,
|
| 7418 |
+
"grad_norm": 1.5249863862991333,
|
| 7419 |
+
"learning_rate": 0.00019931598459407704,
|
| 7420 |
+
"loss": 0.4253,
|
| 7421 |
+
"step": 1052
|
| 7422 |
+
},
|
| 7423 |
+
{
|
| 7424 |
+
"epoch": 0.1150144041068771,
|
| 7425 |
+
"grad_norm": 1.195080041885376,
|
| 7426 |
+
"learning_rate": 0.000199314646889035,
|
| 7427 |
+
"loss": 0.7426,
|
| 7428 |
+
"step": 1053
|
| 7429 |
+
},
|
| 7430 |
+
{
|
| 7431 |
+
"epoch": 0.11512362956186939,
|
| 7432 |
+
"grad_norm": 1.8309367895126343,
|
| 7433 |
+
"learning_rate": 0.00019931330788171668,
|
| 7434 |
+
"loss": 0.8132,
|
| 7435 |
+
"step": 1054
|
| 7436 |
+
},
|
| 7437 |
+
{
|
| 7438 |
+
"epoch": 0.11523285501686167,
|
| 7439 |
+
"grad_norm": 1.301666259765625,
|
| 7440 |
+
"learning_rate": 0.0001993119675721397,
|
| 7441 |
+
"loss": 0.7908,
|
| 7442 |
+
"step": 1055
|
| 7443 |
+
},
|
| 7444 |
+
{
|
| 7445 |
+
"epoch": 0.11534208047185397,
|
| 7446 |
+
"grad_norm": 1.4406108856201172,
|
| 7447 |
+
"learning_rate": 0.0001993106259603216,
|
| 7448 |
+
"loss": 0.7856,
|
| 7449 |
+
"step": 1056
|
| 7450 |
+
},
|
| 7451 |
+
{
|
| 7452 |
+
"epoch": 0.11545130592684626,
|
| 7453 |
+
"grad_norm": 1.672108769416809,
|
| 7454 |
+
"learning_rate": 0.00019930928304628,
|
| 7455 |
+
"loss": 0.8696,
|
| 7456 |
+
"step": 1057
|
| 7457 |
+
},
|
| 7458 |
+
{
|
| 7459 |
+
"epoch": 0.11556053138183854,
|
| 7460 |
+
"grad_norm": 1.634285807609558,
|
| 7461 |
+
"learning_rate": 0.00019930793883003243,
|
| 7462 |
+
"loss": 0.6701,
|
| 7463 |
+
"step": 1058
|
| 7464 |
+
},
|
| 7465 |
+
{
|
| 7466 |
+
"epoch": 0.11566975683683083,
|
| 7467 |
+
"grad_norm": 1.3943419456481934,
|
| 7468 |
+
"learning_rate": 0.0001993065933115966,
|
| 7469 |
+
"loss": 1.0329,
|
| 7470 |
+
"step": 1059
|
| 7471 |
+
},
|
| 7472 |
+
{
|
| 7473 |
+
"epoch": 0.11577898229182311,
|
| 7474 |
+
"grad_norm": 1.7570948600769043,
|
| 7475 |
+
"learning_rate": 0.00019930524649099013,
|
| 7476 |
+
"loss": 0.8156,
|
| 7477 |
+
"step": 1060
|
| 7478 |
+
},
|
| 7479 |
+
{
|
| 7480 |
+
"epoch": 0.1158882077468154,
|
| 7481 |
+
"grad_norm": 1.2345890998840332,
|
| 7482 |
+
"learning_rate": 0.0001993038983682307,
|
| 7483 |
+
"loss": 0.7043,
|
| 7484 |
+
"step": 1061
|
| 7485 |
+
},
|
| 7486 |
+
{
|
| 7487 |
+
"epoch": 0.11599743320180768,
|
| 7488 |
+
"grad_norm": 1.6712225675582886,
|
| 7489 |
+
"learning_rate": 0.00019930254894333596,
|
| 7490 |
+
"loss": 0.7725,
|
| 7491 |
+
"step": 1062
|
| 7492 |
+
},
|
| 7493 |
+
{
|
| 7494 |
+
"epoch": 0.11610665865679996,
|
| 7495 |
+
"grad_norm": 1.473175048828125,
|
| 7496 |
+
"learning_rate": 0.0001993011982163236,
|
| 7497 |
+
"loss": 0.7347,
|
| 7498 |
+
"step": 1063
|
| 7499 |
+
},
|
| 7500 |
+
{
|
| 7501 |
+
"epoch": 0.11621588411179225,
|
| 7502 |
+
"grad_norm": 1.3786654472351074,
|
| 7503 |
+
"learning_rate": 0.00019929984618721137,
|
| 7504 |
+
"loss": 0.8501,
|
| 7505 |
+
"step": 1064
|
| 7506 |
+
},
|
| 7507 |
+
{
|
| 7508 |
+
"epoch": 0.11632510956678455,
|
| 7509 |
+
"grad_norm": 1.1373165845870972,
|
| 7510 |
+
"learning_rate": 0.00019929849285601692,
|
| 7511 |
+
"loss": 0.8806,
|
| 7512 |
+
"step": 1065
|
| 7513 |
+
},
|
| 7514 |
+
{
|
| 7515 |
+
"epoch": 0.11643433502177683,
|
| 7516 |
+
"grad_norm": 1.8122903108596802,
|
| 7517 |
+
"learning_rate": 0.00019929713822275805,
|
| 7518 |
+
"loss": 0.9973,
|
| 7519 |
+
"step": 1066
|
| 7520 |
+
},
|
| 7521 |
+
{
|
| 7522 |
+
"epoch": 0.11654356047676911,
|
| 7523 |
+
"grad_norm": 0.9967430233955383,
|
| 7524 |
+
"learning_rate": 0.00019929578228745256,
|
| 7525 |
+
"loss": 0.783,
|
| 7526 |
+
"step": 1067
|
| 7527 |
+
},
|
| 7528 |
+
{
|
| 7529 |
+
"epoch": 0.1166527859317614,
|
| 7530 |
+
"grad_norm": 1.4985566139221191,
|
| 7531 |
+
"learning_rate": 0.00019929442505011818,
|
| 7532 |
+
"loss": 0.8065,
|
| 7533 |
+
"step": 1068
|
| 7534 |
+
},
|
| 7535 |
+
{
|
| 7536 |
+
"epoch": 0.11676201138675368,
|
| 7537 |
+
"grad_norm": 1.4771041870117188,
|
| 7538 |
+
"learning_rate": 0.0001992930665107727,
|
| 7539 |
+
"loss": 0.7213,
|
| 7540 |
+
"step": 1069
|
| 7541 |
+
},
|
| 7542 |
+
{
|
| 7543 |
+
"epoch": 0.11687123684174597,
|
| 7544 |
+
"grad_norm": 1.803648829460144,
|
| 7545 |
+
"learning_rate": 0.00019929170666943397,
|
| 7546 |
+
"loss": 0.7813,
|
| 7547 |
+
"step": 1070
|
| 7548 |
+
},
|
| 7549 |
+
{
|
| 7550 |
+
"epoch": 0.11698046229673825,
|
| 7551 |
+
"grad_norm": 1.6353493928909302,
|
| 7552 |
+
"learning_rate": 0.00019929034552611979,
|
| 7553 |
+
"loss": 0.7494,
|
| 7554 |
+
"step": 1071
|
| 7555 |
+
},
|
| 7556 |
+
{
|
| 7557 |
+
"epoch": 0.11708968775173054,
|
| 7558 |
+
"grad_norm": 1.5146071910858154,
|
| 7559 |
+
"learning_rate": 0.00019928898308084803,
|
| 7560 |
+
"loss": 0.8089,
|
| 7561 |
+
"step": 1072
|
| 7562 |
+
},
|
| 7563 |
+
{
|
| 7564 |
+
"epoch": 0.11719891320672282,
|
| 7565 |
+
"grad_norm": 1.3005802631378174,
|
| 7566 |
+
"learning_rate": 0.00019928761933363652,
|
| 7567 |
+
"loss": 0.7843,
|
| 7568 |
+
"step": 1073
|
| 7569 |
+
},
|
| 7570 |
+
{
|
| 7571 |
+
"epoch": 0.11730813866171512,
|
| 7572 |
+
"grad_norm": 1.4477092027664185,
|
| 7573 |
+
"learning_rate": 0.0001992862542845032,
|
| 7574 |
+
"loss": 0.8091,
|
| 7575 |
+
"step": 1074
|
| 7576 |
+
},
|
| 7577 |
+
{
|
| 7578 |
+
"epoch": 0.1174173641167074,
|
| 7579 |
+
"grad_norm": 1.0689550638198853,
|
| 7580 |
+
"learning_rate": 0.0001992848879334659,
|
| 7581 |
+
"loss": 0.9451,
|
| 7582 |
+
"step": 1075
|
| 7583 |
+
},
|
| 7584 |
+
{
|
| 7585 |
+
"epoch": 0.11752658957169969,
|
| 7586 |
+
"grad_norm": 1.581805944442749,
|
| 7587 |
+
"learning_rate": 0.00019928352028054258,
|
| 7588 |
+
"loss": 0.8532,
|
| 7589 |
+
"step": 1076
|
| 7590 |
+
},
|
| 7591 |
+
{
|
| 7592 |
+
"epoch": 0.11763581502669197,
|
| 7593 |
+
"grad_norm": 1.5796600580215454,
|
| 7594 |
+
"learning_rate": 0.0001992821513257512,
|
| 7595 |
+
"loss": 0.6943,
|
| 7596 |
+
"step": 1077
|
| 7597 |
+
},
|
| 7598 |
+
{
|
| 7599 |
+
"epoch": 0.11774504048168426,
|
| 7600 |
+
"grad_norm": 1.5918270349502563,
|
| 7601 |
+
"learning_rate": 0.00019928078106910965,
|
| 7602 |
+
"loss": 0.9145,
|
| 7603 |
+
"step": 1078
|
| 7604 |
+
},
|
| 7605 |
+
{
|
| 7606 |
+
"epoch": 0.11785426593667654,
|
| 7607 |
+
"grad_norm": 1.4922096729278564,
|
| 7608 |
+
"learning_rate": 0.00019927940951063592,
|
| 7609 |
+
"loss": 0.6328,
|
| 7610 |
+
"step": 1079
|
| 7611 |
+
},
|
| 7612 |
+
{
|
| 7613 |
+
"epoch": 0.11796349139166883,
|
| 7614 |
+
"grad_norm": 0.9377450346946716,
|
| 7615 |
+
"learning_rate": 0.000199278036650348,
|
| 7616 |
+
"loss": 1.0577,
|
| 7617 |
+
"step": 1080
|
| 7618 |
+
},
|
| 7619 |
+
{
|
| 7620 |
+
"epoch": 0.11807271684666111,
|
| 7621 |
+
"grad_norm": 1.3035157918930054,
|
| 7622 |
+
"learning_rate": 0.00019927666248826392,
|
| 7623 |
+
"loss": 0.7525,
|
| 7624 |
+
"step": 1081
|
| 7625 |
+
},
|
| 7626 |
+
{
|
| 7627 |
+
"epoch": 0.1181819423016534,
|
| 7628 |
+
"grad_norm": 1.5068557262420654,
|
| 7629 |
+
"learning_rate": 0.00019927528702440165,
|
| 7630 |
+
"loss": 0.7421,
|
| 7631 |
+
"step": 1082
|
| 7632 |
+
},
|
| 7633 |
+
{
|
| 7634 |
+
"epoch": 0.1182911677566457,
|
| 7635 |
+
"grad_norm": 1.2622627019882202,
|
| 7636 |
+
"learning_rate": 0.00019927391025877924,
|
| 7637 |
+
"loss": 0.6874,
|
| 7638 |
+
"step": 1083
|
| 7639 |
+
},
|
| 7640 |
+
{
|
| 7641 |
+
"epoch": 0.11840039321163798,
|
| 7642 |
+
"grad_norm": 1.7203670740127563,
|
| 7643 |
+
"learning_rate": 0.00019927253219141476,
|
| 7644 |
+
"loss": 0.6967,
|
| 7645 |
+
"step": 1084
|
| 7646 |
+
},
|
| 7647 |
+
{
|
| 7648 |
+
"epoch": 0.11850961866663026,
|
| 7649 |
+
"grad_norm": 1.436240792274475,
|
| 7650 |
+
"learning_rate": 0.00019927115282232624,
|
| 7651 |
+
"loss": 0.6359,
|
| 7652 |
+
"step": 1085
|
| 7653 |
+
},
|
| 7654 |
+
{
|
| 7655 |
+
"epoch": 0.11861884412162255,
|
| 7656 |
+
"grad_norm": 1.0763415098190308,
|
| 7657 |
+
"learning_rate": 0.00019926977215153183,
|
| 7658 |
+
"loss": 0.8655,
|
| 7659 |
+
"step": 1086
|
| 7660 |
+
},
|
| 7661 |
+
{
|
| 7662 |
+
"epoch": 0.11872806957661483,
|
| 7663 |
+
"grad_norm": 1.2961541414260864,
|
| 7664 |
+
"learning_rate": 0.0001992683901790496,
|
| 7665 |
+
"loss": 0.494,
|
| 7666 |
+
"step": 1087
|
| 7667 |
+
},
|
| 7668 |
+
{
|
| 7669 |
+
"epoch": 0.11883729503160712,
|
| 7670 |
+
"grad_norm": 1.5096957683563232,
|
| 7671 |
+
"learning_rate": 0.00019926700690489765,
|
| 7672 |
+
"loss": 0.6852,
|
| 7673 |
+
"step": 1088
|
| 7674 |
+
},
|
| 7675 |
+
{
|
| 7676 |
+
"epoch": 0.1189465204865994,
|
| 7677 |
+
"grad_norm": 1.5299978256225586,
|
| 7678 |
+
"learning_rate": 0.00019926562232909417,
|
| 7679 |
+
"loss": 1.0133,
|
| 7680 |
+
"step": 1089
|
| 7681 |
+
},
|
| 7682 |
+
{
|
| 7683 |
+
"epoch": 0.11905574594159168,
|
| 7684 |
+
"grad_norm": 1.0444854497909546,
|
| 7685 |
+
"learning_rate": 0.00019926423645165727,
|
| 7686 |
+
"loss": 0.9608,
|
| 7687 |
+
"step": 1090
|
| 7688 |
+
},
|
| 7689 |
+
{
|
| 7690 |
+
"epoch": 0.11916497139658397,
|
| 7691 |
+
"grad_norm": 1.2627832889556885,
|
| 7692 |
+
"learning_rate": 0.00019926284927260514,
|
| 7693 |
+
"loss": 1.0715,
|
| 7694 |
+
"step": 1091
|
| 7695 |
+
},
|
| 7696 |
+
{
|
| 7697 |
+
"epoch": 0.11927419685157625,
|
| 7698 |
+
"grad_norm": 1.2448171377182007,
|
| 7699 |
+
"learning_rate": 0.00019926146079195594,
|
| 7700 |
+
"loss": 0.9528,
|
| 7701 |
+
"step": 1092
|
| 7702 |
+
},
|
| 7703 |
+
{
|
| 7704 |
+
"epoch": 0.11938342230656855,
|
| 7705 |
+
"grad_norm": 1.0306788682937622,
|
| 7706 |
+
"learning_rate": 0.00019926007100972798,
|
| 7707 |
+
"loss": 0.908,
|
| 7708 |
+
"step": 1093
|
| 7709 |
+
},
|
| 7710 |
+
{
|
| 7711 |
+
"epoch": 0.11949264776156084,
|
| 7712 |
+
"grad_norm": 1.3586934804916382,
|
| 7713 |
+
"learning_rate": 0.00019925867992593932,
|
| 7714 |
+
"loss": 0.8233,
|
| 7715 |
+
"step": 1094
|
| 7716 |
+
},
|
| 7717 |
+
{
|
| 7718 |
+
"epoch": 0.11960187321655312,
|
| 7719 |
+
"grad_norm": 1.3546695709228516,
|
| 7720 |
+
"learning_rate": 0.00019925728754060834,
|
| 7721 |
+
"loss": 0.6921,
|
| 7722 |
+
"step": 1095
|
| 7723 |
+
},
|
| 7724 |
+
{
|
| 7725 |
+
"epoch": 0.1197110986715454,
|
| 7726 |
+
"grad_norm": 1.4747517108917236,
|
| 7727 |
+
"learning_rate": 0.00019925589385375324,
|
| 7728 |
+
"loss": 0.3998,
|
| 7729 |
+
"step": 1096
|
| 7730 |
+
},
|
| 7731 |
+
{
|
| 7732 |
+
"epoch": 0.11982032412653769,
|
| 7733 |
+
"grad_norm": 1.6195718050003052,
|
| 7734 |
+
"learning_rate": 0.0001992544988653923,
|
| 7735 |
+
"loss": 1.0382,
|
| 7736 |
+
"step": 1097
|
| 7737 |
+
},
|
| 7738 |
+
{
|
| 7739 |
+
"epoch": 0.11992954958152997,
|
| 7740 |
+
"grad_norm": 1.6750741004943848,
|
| 7741 |
+
"learning_rate": 0.0001992531025755438,
|
| 7742 |
+
"loss": 0.5369,
|
| 7743 |
+
"step": 1098
|
| 7744 |
+
},
|
| 7745 |
+
{
|
| 7746 |
+
"epoch": 0.12003877503652226,
|
| 7747 |
+
"grad_norm": 1.1470670700073242,
|
| 7748 |
+
"learning_rate": 0.00019925170498422606,
|
| 7749 |
+
"loss": 0.8429,
|
| 7750 |
+
"step": 1099
|
| 7751 |
+
},
|
| 7752 |
+
{
|
| 7753 |
+
"epoch": 0.12014800049151454,
|
| 7754 |
+
"grad_norm": 1.1155179738998413,
|
| 7755 |
+
"learning_rate": 0.00019925030609145743,
|
| 7756 |
+
"loss": 0.6548,
|
| 7757 |
+
"step": 1100
|
| 7758 |
+
},
|
| 7759 |
+
{
|
| 7760 |
+
"epoch": 0.12025722594650683,
|
| 7761 |
+
"grad_norm": 1.6656816005706787,
|
| 7762 |
+
"learning_rate": 0.00019924890589725623,
|
| 7763 |
+
"loss": 0.5107,
|
| 7764 |
+
"step": 1101
|
| 7765 |
+
},
|
| 7766 |
+
{
|
| 7767 |
+
"epoch": 0.12036645140149912,
|
| 7768 |
+
"grad_norm": 1.3539464473724365,
|
| 7769 |
+
"learning_rate": 0.00019924750440164081,
|
| 7770 |
+
"loss": 0.7774,
|
| 7771 |
+
"step": 1102
|
| 7772 |
+
},
|
| 7773 |
+
{
|
| 7774 |
+
"epoch": 0.12047567685649141,
|
| 7775 |
+
"grad_norm": 1.571033239364624,
|
| 7776 |
+
"learning_rate": 0.00019924610160462956,
|
| 7777 |
+
"loss": 0.6545,
|
| 7778 |
+
"step": 1103
|
| 7779 |
+
},
|
| 7780 |
+
{
|
| 7781 |
+
"epoch": 0.1205849023114837,
|
| 7782 |
+
"grad_norm": 1.665709376335144,
|
| 7783 |
+
"learning_rate": 0.0001992446975062409,
|
| 7784 |
+
"loss": 0.8734,
|
| 7785 |
+
"step": 1104
|
| 7786 |
+
},
|
| 7787 |
+
{
|
| 7788 |
+
"epoch": 0.12069412776647598,
|
| 7789 |
+
"grad_norm": 1.2241404056549072,
|
| 7790 |
+
"learning_rate": 0.00019924329210649317,
|
| 7791 |
+
"loss": 0.8143,
|
| 7792 |
+
"step": 1105
|
| 7793 |
+
},
|
| 7794 |
+
{
|
| 7795 |
+
"epoch": 0.12080335322146826,
|
| 7796 |
+
"grad_norm": 0.9677228331565857,
|
| 7797 |
+
"learning_rate": 0.00019924188540540488,
|
| 7798 |
+
"loss": 0.7972,
|
| 7799 |
+
"step": 1106
|
| 7800 |
+
},
|
| 7801 |
+
{
|
| 7802 |
+
"epoch": 0.12091257867646055,
|
| 7803 |
+
"grad_norm": 1.384819507598877,
|
| 7804 |
+
"learning_rate": 0.00019924047740299444,
|
| 7805 |
+
"loss": 0.5386,
|
| 7806 |
+
"step": 1107
|
| 7807 |
+
},
|
| 7808 |
+
{
|
| 7809 |
+
"epoch": 0.12102180413145283,
|
| 7810 |
+
"grad_norm": 1.4157319068908691,
|
| 7811 |
+
"learning_rate": 0.0001992390680992803,
|
| 7812 |
+
"loss": 0.7179,
|
| 7813 |
+
"step": 1108
|
| 7814 |
+
},
|
| 7815 |
+
{
|
| 7816 |
+
"epoch": 0.12113102958644512,
|
| 7817 |
+
"grad_norm": 1.3099032640457153,
|
| 7818 |
+
"learning_rate": 0.00019923765749428092,
|
| 7819 |
+
"loss": 0.851,
|
| 7820 |
+
"step": 1109
|
| 7821 |
+
},
|
| 7822 |
+
{
|
| 7823 |
+
"epoch": 0.1212402550414374,
|
| 7824 |
+
"grad_norm": 1.1710662841796875,
|
| 7825 |
+
"learning_rate": 0.0001992362455880149,
|
| 7826 |
+
"loss": 0.7325,
|
| 7827 |
+
"step": 1110
|
| 7828 |
+
},
|
| 7829 |
+
{
|
| 7830 |
+
"epoch": 0.1213494804964297,
|
| 7831 |
+
"grad_norm": 1.4912019968032837,
|
| 7832 |
+
"learning_rate": 0.0001992348323805006,
|
| 7833 |
+
"loss": 0.7942,
|
| 7834 |
+
"step": 1111
|
| 7835 |
+
},
|
| 7836 |
+
{
|
| 7837 |
+
"epoch": 0.12145870595142198,
|
| 7838 |
+
"grad_norm": 1.0365915298461914,
|
| 7839 |
+
"learning_rate": 0.00019923341787175667,
|
| 7840 |
+
"loss": 1.0589,
|
| 7841 |
+
"step": 1112
|
| 7842 |
+
},
|
| 7843 |
+
{
|
| 7844 |
+
"epoch": 0.12156793140641427,
|
| 7845 |
+
"grad_norm": 1.487493872642517,
|
| 7846 |
+
"learning_rate": 0.00019923200206180164,
|
| 7847 |
+
"loss": 0.9355,
|
| 7848 |
+
"step": 1113
|
| 7849 |
+
},
|
| 7850 |
+
{
|
| 7851 |
+
"epoch": 0.12167715686140655,
|
| 7852 |
+
"grad_norm": 1.150929570198059,
|
| 7853 |
+
"learning_rate": 0.00019923058495065402,
|
| 7854 |
+
"loss": 0.813,
|
| 7855 |
+
"step": 1114
|
| 7856 |
+
},
|
| 7857 |
+
{
|
| 7858 |
+
"epoch": 0.12178638231639884,
|
| 7859 |
+
"grad_norm": 1.6531025171279907,
|
| 7860 |
+
"learning_rate": 0.00019922916653833248,
|
| 7861 |
+
"loss": 0.8806,
|
| 7862 |
+
"step": 1115
|
| 7863 |
+
},
|
| 7864 |
+
{
|
| 7865 |
+
"epoch": 0.12189560777139112,
|
| 7866 |
+
"grad_norm": 1.3603259325027466,
|
| 7867 |
+
"learning_rate": 0.00019922774682485552,
|
| 7868 |
+
"loss": 1.1474,
|
| 7869 |
+
"step": 1116
|
| 7870 |
+
},
|
| 7871 |
+
{
|
| 7872 |
+
"epoch": 0.1220048332263834,
|
| 7873 |
+
"grad_norm": 1.6009835004806519,
|
| 7874 |
+
"learning_rate": 0.00019922632581024182,
|
| 7875 |
+
"loss": 0.7006,
|
| 7876 |
+
"step": 1117
|
| 7877 |
+
},
|
| 7878 |
+
{
|
| 7879 |
+
"epoch": 0.12211405868137569,
|
| 7880 |
+
"grad_norm": 1.4102089405059814,
|
| 7881 |
+
"learning_rate": 0.00019922490349451,
|
| 7882 |
+
"loss": 0.4627,
|
| 7883 |
+
"step": 1118
|
| 7884 |
+
},
|
| 7885 |
+
{
|
| 7886 |
+
"epoch": 0.12222328413636797,
|
| 7887 |
+
"grad_norm": 1.0573030710220337,
|
| 7888 |
+
"learning_rate": 0.0001992234798776787,
|
| 7889 |
+
"loss": 0.9204,
|
| 7890 |
+
"step": 1119
|
| 7891 |
+
},
|
| 7892 |
+
{
|
| 7893 |
+
"epoch": 0.12233250959136027,
|
| 7894 |
+
"grad_norm": 1.8316164016723633,
|
| 7895 |
+
"learning_rate": 0.00019922205495976657,
|
| 7896 |
+
"loss": 0.7667,
|
| 7897 |
+
"step": 1120
|
| 7898 |
+
},
|
| 7899 |
+
{
|
| 7900 |
+
"epoch": 0.12244173504635256,
|
| 7901 |
+
"grad_norm": 1.232649803161621,
|
| 7902 |
+
"learning_rate": 0.00019922062874079237,
|
| 7903 |
+
"loss": 0.8237,
|
| 7904 |
+
"step": 1121
|
| 7905 |
+
},
|
| 7906 |
+
{
|
| 7907 |
+
"epoch": 0.12255096050134484,
|
| 7908 |
+
"grad_norm": 1.6198135614395142,
|
| 7909 |
+
"learning_rate": 0.0001992192012207747,
|
| 7910 |
+
"loss": 0.9329,
|
| 7911 |
+
"step": 1122
|
| 7912 |
+
},
|
| 7913 |
+
{
|
| 7914 |
+
"epoch": 0.12266018595633713,
|
| 7915 |
+
"grad_norm": 1.4416582584381104,
|
| 7916 |
+
"learning_rate": 0.00019921777239973238,
|
| 7917 |
+
"loss": 0.7153,
|
| 7918 |
+
"step": 1123
|
| 7919 |
+
},
|
| 7920 |
+
{
|
| 7921 |
+
"epoch": 0.12276941141132941,
|
| 7922 |
+
"grad_norm": 1.1062138080596924,
|
| 7923 |
+
"learning_rate": 0.00019921634227768403,
|
| 7924 |
+
"loss": 1.0572,
|
| 7925 |
+
"step": 1124
|
| 7926 |
+
},
|
| 7927 |
+
{
|
| 7928 |
+
"epoch": 0.1228786368663217,
|
| 7929 |
+
"grad_norm": 1.6167114973068237,
|
| 7930 |
+
"learning_rate": 0.0001992149108546485,
|
| 7931 |
+
"loss": 0.6347,
|
| 7932 |
+
"step": 1125
|
| 7933 |
+
},
|
| 7934 |
+
{
|
| 7935 |
+
"epoch": 0.12298786232131398,
|
| 7936 |
+
"grad_norm": 1.4493283033370972,
|
| 7937 |
+
"learning_rate": 0.00019921347813064454,
|
| 7938 |
+
"loss": 0.8981,
|
| 7939 |
+
"step": 1126
|
| 7940 |
+
},
|
| 7941 |
+
{
|
| 7942 |
+
"epoch": 0.12309708777630626,
|
| 7943 |
+
"grad_norm": 1.386709213256836,
|
| 7944 |
+
"learning_rate": 0.00019921204410569093,
|
| 7945 |
+
"loss": 0.8438,
|
| 7946 |
+
"step": 1127
|
| 7947 |
+
},
|
| 7948 |
+
{
|
| 7949 |
+
"epoch": 0.12320631323129855,
|
| 7950 |
+
"grad_norm": 1.011716604232788,
|
| 7951 |
+
"learning_rate": 0.00019921060877980647,
|
| 7952 |
+
"loss": 0.8014,
|
| 7953 |
+
"step": 1128
|
| 7954 |
+
},
|
| 7955 |
+
{
|
| 7956 |
+
"epoch": 0.12331553868629085,
|
| 7957 |
+
"grad_norm": 1.101968765258789,
|
| 7958 |
+
"learning_rate": 0.00019920917215300997,
|
| 7959 |
+
"loss": 0.9873,
|
| 7960 |
+
"step": 1129
|
| 7961 |
+
},
|
| 7962 |
+
{
|
| 7963 |
+
"epoch": 0.12342476414128313,
|
| 7964 |
+
"grad_norm": 1.388913631439209,
|
| 7965 |
+
"learning_rate": 0.00019920773422532024,
|
| 7966 |
+
"loss": 0.7366,
|
| 7967 |
+
"step": 1130
|
| 7968 |
+
},
|
| 7969 |
+
{
|
| 7970 |
+
"epoch": 0.12353398959627541,
|
| 7971 |
+
"grad_norm": 1.0777970552444458,
|
| 7972 |
+
"learning_rate": 0.0001992062949967562,
|
| 7973 |
+
"loss": 0.9902,
|
| 7974 |
+
"step": 1131
|
| 7975 |
+
},
|
| 7976 |
+
{
|
| 7977 |
+
"epoch": 0.1236432150512677,
|
| 7978 |
+
"grad_norm": 1.364811658859253,
|
| 7979 |
+
"learning_rate": 0.0001992048544673367,
|
| 7980 |
+
"loss": 0.9963,
|
| 7981 |
+
"step": 1132
|
| 7982 |
+
},
|
| 7983 |
+
{
|
| 7984 |
+
"epoch": 0.12375244050625998,
|
| 7985 |
+
"grad_norm": 1.068984866142273,
|
| 7986 |
+
"learning_rate": 0.0001992034126370806,
|
| 7987 |
+
"loss": 0.7391,
|
| 7988 |
+
"step": 1133
|
| 7989 |
+
},
|
| 7990 |
+
{
|
| 7991 |
+
"epoch": 0.12386166596125227,
|
| 7992 |
+
"grad_norm": 1.4525978565216064,
|
| 7993 |
+
"learning_rate": 0.0001992019695060068,
|
| 7994 |
+
"loss": 0.743,
|
| 7995 |
+
"step": 1134
|
| 7996 |
+
},
|
| 7997 |
+
{
|
| 7998 |
+
"epoch": 0.12397089141624455,
|
| 7999 |
+
"grad_norm": 1.5010894536972046,
|
| 8000 |
+
"learning_rate": 0.0001992005250741343,
|
| 8001 |
+
"loss": 0.5288,
|
| 8002 |
+
"step": 1135
|
| 8003 |
+
},
|
| 8004 |
+
{
|
| 8005 |
+
"epoch": 0.12408011687123684,
|
| 8006 |
+
"grad_norm": 1.6246229410171509,
|
| 8007 |
+
"learning_rate": 0.000199199079341482,
|
| 8008 |
+
"loss": 0.564,
|
| 8009 |
+
"step": 1136
|
| 8010 |
+
},
|
| 8011 |
+
{
|
| 8012 |
+
"epoch": 0.12418934232622912,
|
| 8013 |
+
"grad_norm": 1.2848314046859741,
|
| 8014 |
+
"learning_rate": 0.00019919763230806886,
|
| 8015 |
+
"loss": 1.01,
|
| 8016 |
+
"step": 1137
|
| 8017 |
+
},
|
| 8018 |
+
{
|
| 8019 |
+
"epoch": 0.12429856778122142,
|
| 8020 |
+
"grad_norm": 1.1841753721237183,
|
| 8021 |
+
"learning_rate": 0.00019919618397391383,
|
| 8022 |
+
"loss": 0.8131,
|
| 8023 |
+
"step": 1138
|
| 8024 |
+
},
|
| 8025 |
+
{
|
| 8026 |
+
"epoch": 0.1244077932362137,
|
| 8027 |
+
"grad_norm": 1.2205997705459595,
|
| 8028 |
+
"learning_rate": 0.00019919473433903587,
|
| 8029 |
+
"loss": 0.6815,
|
| 8030 |
+
"step": 1139
|
| 8031 |
+
},
|
| 8032 |
+
{
|
| 8033 |
+
"epoch": 0.12451701869120599,
|
| 8034 |
+
"grad_norm": 1.003227710723877,
|
| 8035 |
+
"learning_rate": 0.00019919328340345407,
|
| 8036 |
+
"loss": 1.0282,
|
| 8037 |
+
"step": 1140
|
| 8038 |
+
},
|
| 8039 |
+
{
|
| 8040 |
+
"epoch": 0.12462624414619827,
|
| 8041 |
+
"grad_norm": 1.6726300716400146,
|
| 8042 |
+
"learning_rate": 0.00019919183116718741,
|
| 8043 |
+
"loss": 0.5959,
|
| 8044 |
+
"step": 1141
|
| 8045 |
+
},
|
| 8046 |
+
{
|
| 8047 |
+
"epoch": 0.12473546960119056,
|
| 8048 |
+
"grad_norm": 1.4816073179244995,
|
| 8049 |
+
"learning_rate": 0.00019919037763025497,
|
| 8050 |
+
"loss": 0.8751,
|
| 8051 |
+
"step": 1142
|
| 8052 |
+
},
|
| 8053 |
+
{
|
| 8054 |
+
"epoch": 0.12484469505618284,
|
| 8055 |
+
"grad_norm": 1.4230906963348389,
|
| 8056 |
+
"learning_rate": 0.00019918892279267573,
|
| 8057 |
+
"loss": 0.7128,
|
| 8058 |
+
"step": 1143
|
| 8059 |
+
},
|
| 8060 |
+
{
|
| 8061 |
+
"epoch": 0.12495392051117513,
|
| 8062 |
+
"grad_norm": 1.585444450378418,
|
| 8063 |
+
"learning_rate": 0.00019918746665446887,
|
| 8064 |
+
"loss": 0.987,
|
| 8065 |
+
"step": 1144
|
| 8066 |
+
},
|
| 8067 |
+
{
|
| 8068 |
+
"epoch": 0.1250631459661674,
|
| 8069 |
+
"grad_norm": 1.2062503099441528,
|
| 8070 |
+
"learning_rate": 0.0001991860092156534,
|
| 8071 |
+
"loss": 0.8526,
|
| 8072 |
+
"step": 1145
|
| 8073 |
+
},
|
| 8074 |
+
{
|
| 8075 |
+
"epoch": 0.1251723714211597,
|
| 8076 |
+
"grad_norm": 1.2922080755233765,
|
| 8077 |
+
"learning_rate": 0.00019918455047624847,
|
| 8078 |
+
"loss": 0.8382,
|
| 8079 |
+
"step": 1146
|
| 8080 |
+
},
|
| 8081 |
+
{
|
| 8082 |
+
"epoch": 0.12528159687615198,
|
| 8083 |
+
"grad_norm": 1.0248481035232544,
|
| 8084 |
+
"learning_rate": 0.0001991830904362732,
|
| 8085 |
+
"loss": 0.8086,
|
| 8086 |
+
"step": 1147
|
| 8087 |
+
},
|
| 8088 |
+
{
|
| 8089 |
+
"epoch": 0.12539082233114426,
|
| 8090 |
+
"grad_norm": 1.1466559171676636,
|
| 8091 |
+
"learning_rate": 0.00019918162909574676,
|
| 8092 |
+
"loss": 1.0704,
|
| 8093 |
+
"step": 1148
|
| 8094 |
+
},
|
| 8095 |
+
{
|
| 8096 |
+
"epoch": 0.12550004778613655,
|
| 8097 |
+
"grad_norm": 1.1013227701187134,
|
| 8098 |
+
"learning_rate": 0.00019918016645468828,
|
| 8099 |
+
"loss": 0.7709,
|
| 8100 |
+
"step": 1149
|
| 8101 |
+
},
|
| 8102 |
+
{
|
| 8103 |
+
"epoch": 0.12560927324112883,
|
| 8104 |
+
"grad_norm": 1.3349677324295044,
|
| 8105 |
+
"learning_rate": 0.0001991787025131169,
|
| 8106 |
+
"loss": 0.8724,
|
| 8107 |
+
"step": 1150
|
| 8108 |
+
},
|
| 8109 |
+
{
|
| 8110 |
+
"epoch": 0.12571849869612114,
|
| 8111 |
+
"grad_norm": 1.3502088785171509,
|
| 8112 |
+
"learning_rate": 0.0001991772372710519,
|
| 8113 |
+
"loss": 0.7469,
|
| 8114 |
+
"step": 1151
|
| 8115 |
+
},
|
| 8116 |
+
{
|
| 8117 |
+
"epoch": 0.12582772415111343,
|
| 8118 |
+
"grad_norm": 1.935926079750061,
|
| 8119 |
+
"learning_rate": 0.00019917577072851245,
|
| 8120 |
+
"loss": 0.6738,
|
| 8121 |
+
"step": 1152
|
| 8122 |
+
},
|
| 8123 |
+
{
|
| 8124 |
+
"epoch": 0.1259369496061057,
|
| 8125 |
+
"grad_norm": 1.0898360013961792,
|
| 8126 |
+
"learning_rate": 0.0001991743028855178,
|
| 8127 |
+
"loss": 0.9042,
|
| 8128 |
+
"step": 1153
|
| 8129 |
+
},
|
| 8130 |
+
{
|
| 8131 |
+
"epoch": 0.126046175061098,
|
| 8132 |
+
"grad_norm": 1.331155776977539,
|
| 8133 |
+
"learning_rate": 0.00019917283374208718,
|
| 8134 |
+
"loss": 0.7953,
|
| 8135 |
+
"step": 1154
|
| 8136 |
+
},
|
| 8137 |
+
{
|
| 8138 |
+
"epoch": 0.12615540051609028,
|
| 8139 |
+
"grad_norm": 1.3799457550048828,
|
| 8140 |
+
"learning_rate": 0.00019917136329823985,
|
| 8141 |
+
"loss": 0.8777,
|
| 8142 |
+
"step": 1155
|
| 8143 |
+
},
|
| 8144 |
+
{
|
| 8145 |
+
"epoch": 0.12626462597108257,
|
| 8146 |
+
"grad_norm": 1.4449254274368286,
|
| 8147 |
+
"learning_rate": 0.0001991698915539951,
|
| 8148 |
+
"loss": 0.762,
|
| 8149 |
+
"step": 1156
|
| 8150 |
+
},
|
| 8151 |
+
{
|
| 8152 |
+
"epoch": 0.12637385142607485,
|
| 8153 |
+
"grad_norm": 2.1223926544189453,
|
| 8154 |
+
"learning_rate": 0.00019916841850937223,
|
| 8155 |
+
"loss": 0.89,
|
| 8156 |
+
"step": 1157
|
| 8157 |
+
},
|
| 8158 |
+
{
|
| 8159 |
+
"epoch": 0.12648307688106714,
|
| 8160 |
+
"grad_norm": 2.0117380619049072,
|
| 8161 |
+
"learning_rate": 0.00019916694416439055,
|
| 8162 |
+
"loss": 0.9165,
|
| 8163 |
+
"step": 1158
|
| 8164 |
+
},
|
| 8165 |
+
{
|
| 8166 |
+
"epoch": 0.12659230233605942,
|
| 8167 |
+
"grad_norm": 1.5511128902435303,
|
| 8168 |
+
"learning_rate": 0.00019916546851906942,
|
| 8169 |
+
"loss": 0.7501,
|
| 8170 |
+
"step": 1159
|
| 8171 |
+
},
|
| 8172 |
+
{
|
| 8173 |
+
"epoch": 0.1267015277910517,
|
| 8174 |
+
"grad_norm": 1.7130296230316162,
|
| 8175 |
+
"learning_rate": 0.0001991639915734281,
|
| 8176 |
+
"loss": 0.7706,
|
| 8177 |
+
"step": 1160
|
| 8178 |
+
},
|
| 8179 |
+
{
|
| 8180 |
+
"epoch": 0.126810753246044,
|
| 8181 |
+
"grad_norm": 1.9127705097198486,
|
| 8182 |
+
"learning_rate": 0.0001991625133274861,
|
| 8183 |
+
"loss": 0.6702,
|
| 8184 |
+
"step": 1161
|
| 8185 |
+
},
|
| 8186 |
+
{
|
| 8187 |
+
"epoch": 0.12691997870103627,
|
| 8188 |
+
"grad_norm": 0.9722485542297363,
|
| 8189 |
+
"learning_rate": 0.0001991610337812627,
|
| 8190 |
+
"loss": 0.8523,
|
| 8191 |
+
"step": 1162
|
| 8192 |
+
},
|
| 8193 |
+
{
|
| 8194 |
+
"epoch": 0.12702920415602856,
|
| 8195 |
+
"grad_norm": 1.625596046447754,
|
| 8196 |
+
"learning_rate": 0.0001991595529347773,
|
| 8197 |
+
"loss": 0.8436,
|
| 8198 |
+
"step": 1163
|
| 8199 |
+
},
|
| 8200 |
+
{
|
| 8201 |
+
"epoch": 0.12713842961102084,
|
| 8202 |
+
"grad_norm": 0.9808928370475769,
|
| 8203 |
+
"learning_rate": 0.00019915807078804937,
|
| 8204 |
+
"loss": 1.0688,
|
| 8205 |
+
"step": 1164
|
| 8206 |
+
},
|
| 8207 |
+
{
|
| 8208 |
+
"epoch": 0.12724765506601313,
|
| 8209 |
+
"grad_norm": 1.1023699045181274,
|
| 8210 |
+
"learning_rate": 0.0001991565873410983,
|
| 8211 |
+
"loss": 0.5671,
|
| 8212 |
+
"step": 1165
|
| 8213 |
+
},
|
| 8214 |
+
{
|
| 8215 |
+
"epoch": 0.1273568805210054,
|
| 8216 |
+
"grad_norm": 1.5466971397399902,
|
| 8217 |
+
"learning_rate": 0.00019915510259394358,
|
| 8218 |
+
"loss": 0.873,
|
| 8219 |
+
"step": 1166
|
| 8220 |
+
},
|
| 8221 |
+
{
|
| 8222 |
+
"epoch": 0.1274661059759977,
|
| 8223 |
+
"grad_norm": 1.235567331314087,
|
| 8224 |
+
"learning_rate": 0.00019915361654660466,
|
| 8225 |
+
"loss": 0.4493,
|
| 8226 |
+
"step": 1167
|
| 8227 |
+
},
|
| 8228 |
+
{
|
| 8229 |
+
"epoch": 0.12757533143098998,
|
| 8230 |
+
"grad_norm": 1.0934616327285767,
|
| 8231 |
+
"learning_rate": 0.000199152129199101,
|
| 8232 |
+
"loss": 0.8694,
|
| 8233 |
+
"step": 1168
|
| 8234 |
+
},
|
| 8235 |
+
{
|
| 8236 |
+
"epoch": 0.12768455688598226,
|
| 8237 |
+
"grad_norm": 1.2065255641937256,
|
| 8238 |
+
"learning_rate": 0.00019915064055145217,
|
| 8239 |
+
"loss": 0.7965,
|
| 8240 |
+
"step": 1169
|
| 8241 |
+
},
|
| 8242 |
+
{
|
| 8243 |
+
"epoch": 0.12779378234097458,
|
| 8244 |
+
"grad_norm": 1.1741219758987427,
|
| 8245 |
+
"learning_rate": 0.00019914915060367764,
|
| 8246 |
+
"loss": 1.0232,
|
| 8247 |
+
"step": 1170
|
| 8248 |
+
},
|
| 8249 |
+
{
|
| 8250 |
+
"epoch": 0.12790300779596686,
|
| 8251 |
+
"grad_norm": 1.2858492136001587,
|
| 8252 |
+
"learning_rate": 0.00019914765935579695,
|
| 8253 |
+
"loss": 0.5655,
|
| 8254 |
+
"step": 1171
|
| 8255 |
+
},
|
| 8256 |
+
{
|
| 8257 |
+
"epoch": 0.12801223325095915,
|
| 8258 |
+
"grad_norm": 1.2976844310760498,
|
| 8259 |
+
"learning_rate": 0.00019914616680782965,
|
| 8260 |
+
"loss": 1.1432,
|
| 8261 |
+
"step": 1172
|
| 8262 |
+
},
|
| 8263 |
+
{
|
| 8264 |
+
"epoch": 0.12812145870595143,
|
| 8265 |
+
"grad_norm": 1.2354950904846191,
|
| 8266 |
+
"learning_rate": 0.00019914467295979534,
|
| 8267 |
+
"loss": 0.7539,
|
| 8268 |
+
"step": 1173
|
| 8269 |
+
},
|
| 8270 |
+
{
|
| 8271 |
+
"epoch": 0.12823068416094371,
|
| 8272 |
+
"grad_norm": 1.6397385597229004,
|
| 8273 |
+
"learning_rate": 0.00019914317781171356,
|
| 8274 |
+
"loss": 0.7583,
|
| 8275 |
+
"step": 1174
|
| 8276 |
+
},
|
| 8277 |
+
{
|
| 8278 |
+
"epoch": 0.128339909615936,
|
| 8279 |
+
"grad_norm": 1.0528595447540283,
|
| 8280 |
+
"learning_rate": 0.00019914168136360395,
|
| 8281 |
+
"loss": 0.7553,
|
| 8282 |
+
"step": 1175
|
| 8283 |
+
},
|
| 8284 |
+
{
|
| 8285 |
+
"epoch": 0.12844913507092828,
|
| 8286 |
+
"grad_norm": 1.6798441410064697,
|
| 8287 |
+
"learning_rate": 0.00019914018361548617,
|
| 8288 |
+
"loss": 0.782,
|
| 8289 |
+
"step": 1176
|
| 8290 |
+
},
|
| 8291 |
+
{
|
| 8292 |
+
"epoch": 0.12855836052592057,
|
| 8293 |
+
"grad_norm": 1.258589744567871,
|
| 8294 |
+
"learning_rate": 0.0001991386845673798,
|
| 8295 |
+
"loss": 0.6446,
|
| 8296 |
+
"step": 1177
|
| 8297 |
+
},
|
| 8298 |
+
{
|
| 8299 |
+
"epoch": 0.12866758598091285,
|
| 8300 |
+
"grad_norm": 1.3399349451065063,
|
| 8301 |
+
"learning_rate": 0.00019913718421930447,
|
| 8302 |
+
"loss": 0.6295,
|
| 8303 |
+
"step": 1178
|
| 8304 |
+
},
|
| 8305 |
+
{
|
| 8306 |
+
"epoch": 0.12877681143590514,
|
| 8307 |
+
"grad_norm": 1.269821286201477,
|
| 8308 |
+
"learning_rate": 0.00019913568257127996,
|
| 8309 |
+
"loss": 0.9165,
|
| 8310 |
+
"step": 1179
|
| 8311 |
+
},
|
| 8312 |
+
{
|
| 8313 |
+
"epoch": 0.12888603689089742,
|
| 8314 |
+
"grad_norm": 1.0398304462432861,
|
| 8315 |
+
"learning_rate": 0.00019913417962332585,
|
| 8316 |
+
"loss": 0.7205,
|
| 8317 |
+
"step": 1180
|
| 8318 |
+
},
|
| 8319 |
+
{
|
| 8320 |
+
"epoch": 0.1289952623458897,
|
| 8321 |
+
"grad_norm": 1.5040866136550903,
|
| 8322 |
+
"learning_rate": 0.0001991326753754619,
|
| 8323 |
+
"loss": 0.7214,
|
| 8324 |
+
"step": 1181
|
| 8325 |
+
},
|
| 8326 |
+
{
|
| 8327 |
+
"epoch": 0.129104487800882,
|
| 8328 |
+
"grad_norm": 1.4522327184677124,
|
| 8329 |
+
"learning_rate": 0.00019913116982770785,
|
| 8330 |
+
"loss": 0.7847,
|
| 8331 |
+
"step": 1182
|
| 8332 |
+
},
|
| 8333 |
+
{
|
| 8334 |
+
"epoch": 0.12921371325587427,
|
| 8335 |
+
"grad_norm": 1.994592308998108,
|
| 8336 |
+
"learning_rate": 0.00019912966298008344,
|
| 8337 |
+
"loss": 0.8897,
|
| 8338 |
+
"step": 1183
|
| 8339 |
+
},
|
| 8340 |
+
{
|
| 8341 |
+
"epoch": 0.12932293871086656,
|
| 8342 |
+
"grad_norm": 1.1971591711044312,
|
| 8343 |
+
"learning_rate": 0.00019912815483260837,
|
| 8344 |
+
"loss": 0.758,
|
| 8345 |
+
"step": 1184
|
| 8346 |
+
},
|
| 8347 |
+
{
|
| 8348 |
+
"epoch": 0.12943216416585884,
|
| 8349 |
+
"grad_norm": 1.4321714639663696,
|
| 8350 |
+
"learning_rate": 0.00019912664538530248,
|
| 8351 |
+
"loss": 0.8213,
|
| 8352 |
+
"step": 1185
|
| 8353 |
+
},
|
| 8354 |
+
{
|
| 8355 |
+
"epoch": 0.12954138962085113,
|
| 8356 |
+
"grad_norm": 1.096415400505066,
|
| 8357 |
+
"learning_rate": 0.00019912513463818553,
|
| 8358 |
+
"loss": 0.917,
|
| 8359 |
+
"step": 1186
|
| 8360 |
+
},
|
| 8361 |
+
{
|
| 8362 |
+
"epoch": 0.1296506150758434,
|
| 8363 |
+
"grad_norm": 1.1453442573547363,
|
| 8364 |
+
"learning_rate": 0.00019912362259127733,
|
| 8365 |
+
"loss": 1.0078,
|
| 8366 |
+
"step": 1187
|
| 8367 |
+
},
|
| 8368 |
+
{
|
| 8369 |
+
"epoch": 0.12975984053083572,
|
| 8370 |
+
"grad_norm": 1.2599525451660156,
|
| 8371 |
+
"learning_rate": 0.00019912210924459772,
|
| 8372 |
+
"loss": 1.135,
|
| 8373 |
+
"step": 1188
|
| 8374 |
+
},
|
| 8375 |
+
{
|
| 8376 |
+
"epoch": 0.129869065985828,
|
| 8377 |
+
"grad_norm": 1.5952235460281372,
|
| 8378 |
+
"learning_rate": 0.00019912059459816655,
|
| 8379 |
+
"loss": 1.0051,
|
| 8380 |
+
"step": 1189
|
| 8381 |
+
},
|
| 8382 |
+
{
|
| 8383 |
+
"epoch": 0.1299782914408203,
|
| 8384 |
+
"grad_norm": 1.5863276720046997,
|
| 8385 |
+
"learning_rate": 0.00019911907865200368,
|
| 8386 |
+
"loss": 0.9173,
|
| 8387 |
+
"step": 1190
|
| 8388 |
+
},
|
| 8389 |
+
{
|
| 8390 |
+
"epoch": 0.13008751689581258,
|
| 8391 |
+
"grad_norm": 1.192514419555664,
|
| 8392 |
+
"learning_rate": 0.00019911756140612895,
|
| 8393 |
+
"loss": 1.1793,
|
| 8394 |
+
"step": 1191
|
| 8395 |
+
},
|
| 8396 |
+
{
|
| 8397 |
+
"epoch": 0.13019674235080486,
|
| 8398 |
+
"grad_norm": 1.3725534677505493,
|
| 8399 |
+
"learning_rate": 0.0001991160428605623,
|
| 8400 |
+
"loss": 0.6035,
|
| 8401 |
+
"step": 1192
|
| 8402 |
+
},
|
| 8403 |
+
{
|
| 8404 |
+
"epoch": 0.13030596780579715,
|
| 8405 |
+
"grad_norm": 1.4018157720565796,
|
| 8406 |
+
"learning_rate": 0.00019911452301532364,
|
| 8407 |
+
"loss": 0.5395,
|
| 8408 |
+
"step": 1193
|
| 8409 |
+
},
|
| 8410 |
+
{
|
| 8411 |
+
"epoch": 0.13041519326078943,
|
| 8412 |
+
"grad_norm": 1.6074763536453247,
|
| 8413 |
+
"learning_rate": 0.00019911300187043285,
|
| 8414 |
+
"loss": 0.7839,
|
| 8415 |
+
"step": 1194
|
| 8416 |
+
},
|
| 8417 |
+
{
|
| 8418 |
+
"epoch": 0.13052441871578171,
|
| 8419 |
+
"grad_norm": 1.3953901529312134,
|
| 8420 |
+
"learning_rate": 0.00019911147942590994,
|
| 8421 |
+
"loss": 0.5896,
|
| 8422 |
+
"step": 1195
|
| 8423 |
+
},
|
| 8424 |
+
{
|
| 8425 |
+
"epoch": 0.130633644170774,
|
| 8426 |
+
"grad_norm": 1.767909049987793,
|
| 8427 |
+
"learning_rate": 0.00019910995568177484,
|
| 8428 |
+
"loss": 0.7127,
|
| 8429 |
+
"step": 1196
|
| 8430 |
+
},
|
| 8431 |
+
{
|
| 8432 |
+
"epoch": 0.13074286962576628,
|
| 8433 |
+
"grad_norm": 2.1226041316986084,
|
| 8434 |
+
"learning_rate": 0.00019910843063804753,
|
| 8435 |
+
"loss": 0.7028,
|
| 8436 |
+
"step": 1197
|
| 8437 |
+
},
|
| 8438 |
+
{
|
| 8439 |
+
"epoch": 0.13085209508075857,
|
| 8440 |
+
"grad_norm": 1.4294666051864624,
|
| 8441 |
+
"learning_rate": 0.000199106904294748,
|
| 8442 |
+
"loss": 0.8648,
|
| 8443 |
+
"step": 1198
|
| 8444 |
+
},
|
| 8445 |
+
{
|
| 8446 |
+
"epoch": 0.13096132053575085,
|
| 8447 |
+
"grad_norm": 1.0620681047439575,
|
| 8448 |
+
"learning_rate": 0.0001991053766518963,
|
| 8449 |
+
"loss": 0.7604,
|
| 8450 |
+
"step": 1199
|
| 8451 |
+
},
|
| 8452 |
+
{
|
| 8453 |
+
"epoch": 0.13107054599074314,
|
| 8454 |
+
"grad_norm": 1.41582453250885,
|
| 8455 |
+
"learning_rate": 0.00019910384770951243,
|
| 8456 |
+
"loss": 0.9233,
|
| 8457 |
+
"step": 1200
|
| 8458 |
+
},
|
| 8459 |
+
{
|
| 8460 |
+
"epoch": 0.13107054599074314,
|
| 8461 |
+
"eval_loss": 0.7380471229553223,
|
| 8462 |
+
"eval_runtime": 72.7237,
|
| 8463 |
+
"eval_samples_per_second": 10.134,
|
| 8464 |
+
"eval_steps_per_second": 5.074,
|
| 8465 |
+
"step": 1200
|
| 8466 |
}
|
| 8467 |
],
|
| 8468 |
"logging_steps": 1,
|
|
|
|
| 8477 |
"early_stopping_threshold": 0.0
|
| 8478 |
},
|
| 8479 |
"attributes": {
|
| 8480 |
+
"early_stopping_patience_counter": 3
|
| 8481 |
}
|
| 8482 |
},
|
| 8483 |
"TrainerControl": {
|
|
|
|
| 8486 |
"should_evaluate": false,
|
| 8487 |
"should_log": false,
|
| 8488 |
"should_save": true,
|
| 8489 |
+
"should_training_stop": true
|
| 8490 |
},
|
| 8491 |
"attributes": {}
|
| 8492 |
}
|
| 8493 |
},
|
| 8494 |
+
"total_flos": 1.716967696194601e+17,
|
| 8495 |
"train_batch_size": 2,
|
| 8496 |
"trial_name": null,
|
| 8497 |
"trial_params": null
|