Training in progress, step 1179, 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 50365768
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:776e3a60fc3ddacab8a74c719807c6ac17fc0ea6578b478112a3887015d61df1
|
| 3 |
size 50365768
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 25859412
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f0ed3cfac196364e7f703086c34678624ce9cdab1779b3183c7652ea39248a09
|
| 3 |
size 25859412
|
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:e51717d4b453ecdefb3e1b21a1d457b9afb8eea5b3ba753b7f2ece6e2f24fa15
|
| 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:d6f940fecfc43794835ec25ad54cdeb0d35006f38f4199f153111d4cfd7b09dd
|
| 3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch": 0.
|
| 5 |
"eval_steps": 393,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -5533,6 +5533,2765 @@
|
|
| 5533 |
"eval_samples_per_second": 18.003,
|
| 5534 |
"eval_steps_per_second": 9.011,
|
| 5535 |
"step": 786
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 5536 |
}
|
| 5537 |
],
|
| 5538 |
"logging_steps": 1,
|
|
@@ -5552,7 +8311,7 @@
|
|
| 5552 |
"attributes": {}
|
| 5553 |
}
|
| 5554 |
},
|
| 5555 |
-
"total_flos":
|
| 5556 |
"train_batch_size": 2,
|
| 5557 |
"trial_name": null,
|
| 5558 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.5336048879837068,
|
| 5 |
"eval_steps": 393,
|
| 6 |
+
"global_step": 1179,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 5533 |
"eval_samples_per_second": 18.003,
|
| 5534 |
"eval_steps_per_second": 9.011,
|
| 5535 |
"step": 786
|
| 5536 |
+
},
|
| 5537 |
+
{
|
| 5538 |
+
"epoch": 0.35618918307309344,
|
| 5539 |
+
"grad_norm": 0.46580770611763,
|
| 5540 |
+
"learning_rate": 0.00010060414875812709,
|
| 5541 |
+
"loss": 0.7772,
|
| 5542 |
+
"step": 787
|
| 5543 |
+
},
|
| 5544 |
+
{
|
| 5545 |
+
"epoch": 0.3566417741570491,
|
| 5546 |
+
"grad_norm": 0.5495899319648743,
|
| 5547 |
+
"learning_rate": 0.00010040276719995605,
|
| 5548 |
+
"loss": 0.846,
|
| 5549 |
+
"step": 788
|
| 5550 |
+
},
|
| 5551 |
+
{
|
| 5552 |
+
"epoch": 0.35709436524100474,
|
| 5553 |
+
"grad_norm": 0.555011510848999,
|
| 5554 |
+
"learning_rate": 0.00010020138400834011,
|
| 5555 |
+
"loss": 0.7422,
|
| 5556 |
+
"step": 789
|
| 5557 |
+
},
|
| 5558 |
+
{
|
| 5559 |
+
"epoch": 0.3575469563249604,
|
| 5560 |
+
"grad_norm": 0.5138972997665405,
|
| 5561 |
+
"learning_rate": 0.0001,
|
| 5562 |
+
"loss": 0.923,
|
| 5563 |
+
"step": 790
|
| 5564 |
+
},
|
| 5565 |
+
{
|
| 5566 |
+
"epoch": 0.35799954740891604,
|
| 5567 |
+
"grad_norm": 0.48092055320739746,
|
| 5568 |
+
"learning_rate": 9.979861599165991e-05,
|
| 5569 |
+
"loss": 0.933,
|
| 5570 |
+
"step": 791
|
| 5571 |
+
},
|
| 5572 |
+
{
|
| 5573 |
+
"epoch": 0.35845213849287166,
|
| 5574 |
+
"grad_norm": 0.44474145770072937,
|
| 5575 |
+
"learning_rate": 9.959723280004396e-05,
|
| 5576 |
+
"loss": 0.6501,
|
| 5577 |
+
"step": 792
|
| 5578 |
+
},
|
| 5579 |
+
{
|
| 5580 |
+
"epoch": 0.35890472957682734,
|
| 5581 |
+
"grad_norm": 0.49980682134628296,
|
| 5582 |
+
"learning_rate": 9.939585124187292e-05,
|
| 5583 |
+
"loss": 0.7661,
|
| 5584 |
+
"step": 793
|
| 5585 |
+
},
|
| 5586 |
+
{
|
| 5587 |
+
"epoch": 0.35935732066078296,
|
| 5588 |
+
"grad_norm": 0.7595360279083252,
|
| 5589 |
+
"learning_rate": 9.919447213386103e-05,
|
| 5590 |
+
"loss": 0.8846,
|
| 5591 |
+
"step": 794
|
| 5592 |
+
},
|
| 5593 |
+
{
|
| 5594 |
+
"epoch": 0.35980991174473864,
|
| 5595 |
+
"grad_norm": 0.492086797952652,
|
| 5596 |
+
"learning_rate": 9.899309629271246e-05,
|
| 5597 |
+
"loss": 0.8221,
|
| 5598 |
+
"step": 795
|
| 5599 |
+
},
|
| 5600 |
+
{
|
| 5601 |
+
"epoch": 0.36026250282869426,
|
| 5602 |
+
"grad_norm": 0.5320361852645874,
|
| 5603 |
+
"learning_rate": 9.879172453511827e-05,
|
| 5604 |
+
"loss": 0.8432,
|
| 5605 |
+
"step": 796
|
| 5606 |
+
},
|
| 5607 |
+
{
|
| 5608 |
+
"epoch": 0.36071509391264994,
|
| 5609 |
+
"grad_norm": 0.4166679084300995,
|
| 5610 |
+
"learning_rate": 9.85903576777529e-05,
|
| 5611 |
+
"loss": 0.7284,
|
| 5612 |
+
"step": 797
|
| 5613 |
+
},
|
| 5614 |
+
{
|
| 5615 |
+
"epoch": 0.36116768499660556,
|
| 5616 |
+
"grad_norm": 0.5354281067848206,
|
| 5617 |
+
"learning_rate": 9.838899653727088e-05,
|
| 5618 |
+
"loss": 0.9363,
|
| 5619 |
+
"step": 798
|
| 5620 |
+
},
|
| 5621 |
+
{
|
| 5622 |
+
"epoch": 0.3616202760805612,
|
| 5623 |
+
"grad_norm": 0.4781079888343811,
|
| 5624 |
+
"learning_rate": 9.818764193030363e-05,
|
| 5625 |
+
"loss": 0.7322,
|
| 5626 |
+
"step": 799
|
| 5627 |
+
},
|
| 5628 |
+
{
|
| 5629 |
+
"epoch": 0.36207286716451687,
|
| 5630 |
+
"grad_norm": 0.4732087552547455,
|
| 5631 |
+
"learning_rate": 9.798629467345599e-05,
|
| 5632 |
+
"loss": 0.7354,
|
| 5633 |
+
"step": 800
|
| 5634 |
+
},
|
| 5635 |
+
{
|
| 5636 |
+
"epoch": 0.3625254582484725,
|
| 5637 |
+
"grad_norm": 0.4598158597946167,
|
| 5638 |
+
"learning_rate": 9.778495558330305e-05,
|
| 5639 |
+
"loss": 0.6928,
|
| 5640 |
+
"step": 801
|
| 5641 |
+
},
|
| 5642 |
+
{
|
| 5643 |
+
"epoch": 0.36297804933242817,
|
| 5644 |
+
"grad_norm": 0.451435387134552,
|
| 5645 |
+
"learning_rate": 9.75836254763868e-05,
|
| 5646 |
+
"loss": 0.7661,
|
| 5647 |
+
"step": 802
|
| 5648 |
+
},
|
| 5649 |
+
{
|
| 5650 |
+
"epoch": 0.3634306404163838,
|
| 5651 |
+
"grad_norm": 0.5164604187011719,
|
| 5652 |
+
"learning_rate": 9.73823051692127e-05,
|
| 5653 |
+
"loss": 0.8549,
|
| 5654 |
+
"step": 803
|
| 5655 |
+
},
|
| 5656 |
+
{
|
| 5657 |
+
"epoch": 0.36388323150033947,
|
| 5658 |
+
"grad_norm": 0.44017529487609863,
|
| 5659 |
+
"learning_rate": 9.718099547824658e-05,
|
| 5660 |
+
"loss": 0.7798,
|
| 5661 |
+
"step": 804
|
| 5662 |
+
},
|
| 5663 |
+
{
|
| 5664 |
+
"epoch": 0.3643358225842951,
|
| 5665 |
+
"grad_norm": 0.45070794224739075,
|
| 5666 |
+
"learning_rate": 9.697969721991114e-05,
|
| 5667 |
+
"loss": 0.7063,
|
| 5668 |
+
"step": 805
|
| 5669 |
+
},
|
| 5670 |
+
{
|
| 5671 |
+
"epoch": 0.3647884136682507,
|
| 5672 |
+
"grad_norm": 0.5448845624923706,
|
| 5673 |
+
"learning_rate": 9.677841121058273e-05,
|
| 5674 |
+
"loss": 1.1638,
|
| 5675 |
+
"step": 806
|
| 5676 |
+
},
|
| 5677 |
+
{
|
| 5678 |
+
"epoch": 0.3652410047522064,
|
| 5679 |
+
"grad_norm": 0.5377791523933411,
|
| 5680 |
+
"learning_rate": 9.657713826658806e-05,
|
| 5681 |
+
"loss": 0.9093,
|
| 5682 |
+
"step": 807
|
| 5683 |
+
},
|
| 5684 |
+
{
|
| 5685 |
+
"epoch": 0.365693595836162,
|
| 5686 |
+
"grad_norm": 0.5310801863670349,
|
| 5687 |
+
"learning_rate": 9.63758792042008e-05,
|
| 5688 |
+
"loss": 0.9624,
|
| 5689 |
+
"step": 808
|
| 5690 |
+
},
|
| 5691 |
+
{
|
| 5692 |
+
"epoch": 0.3661461869201177,
|
| 5693 |
+
"grad_norm": 0.5090137124061584,
|
| 5694 |
+
"learning_rate": 9.617463483963834e-05,
|
| 5695 |
+
"loss": 0.8736,
|
| 5696 |
+
"step": 809
|
| 5697 |
+
},
|
| 5698 |
+
{
|
| 5699 |
+
"epoch": 0.3665987780040733,
|
| 5700 |
+
"grad_norm": 0.47381317615509033,
|
| 5701 |
+
"learning_rate": 9.597340598905852e-05,
|
| 5702 |
+
"loss": 0.8111,
|
| 5703 |
+
"step": 810
|
| 5704 |
+
},
|
| 5705 |
+
{
|
| 5706 |
+
"epoch": 0.367051369088029,
|
| 5707 |
+
"grad_norm": 0.5369988679885864,
|
| 5708 |
+
"learning_rate": 9.577219346855613e-05,
|
| 5709 |
+
"loss": 0.733,
|
| 5710 |
+
"step": 811
|
| 5711 |
+
},
|
| 5712 |
+
{
|
| 5713 |
+
"epoch": 0.3675039601719846,
|
| 5714 |
+
"grad_norm": 0.6559973359107971,
|
| 5715 |
+
"learning_rate": 9.55709980941599e-05,
|
| 5716 |
+
"loss": 0.7097,
|
| 5717 |
+
"step": 812
|
| 5718 |
+
},
|
| 5719 |
+
{
|
| 5720 |
+
"epoch": 0.36795655125594023,
|
| 5721 |
+
"grad_norm": 0.5182902216911316,
|
| 5722 |
+
"learning_rate": 9.536982068182887e-05,
|
| 5723 |
+
"loss": 0.8151,
|
| 5724 |
+
"step": 813
|
| 5725 |
+
},
|
| 5726 |
+
{
|
| 5727 |
+
"epoch": 0.3684091423398959,
|
| 5728 |
+
"grad_norm": 0.49896934628486633,
|
| 5729 |
+
"learning_rate": 9.516866204744931e-05,
|
| 5730 |
+
"loss": 1.012,
|
| 5731 |
+
"step": 814
|
| 5732 |
+
},
|
| 5733 |
+
{
|
| 5734 |
+
"epoch": 0.36886173342385153,
|
| 5735 |
+
"grad_norm": 0.5400479435920715,
|
| 5736 |
+
"learning_rate": 9.496752300683133e-05,
|
| 5737 |
+
"loss": 0.9837,
|
| 5738 |
+
"step": 815
|
| 5739 |
+
},
|
| 5740 |
+
{
|
| 5741 |
+
"epoch": 0.3693143245078072,
|
| 5742 |
+
"grad_norm": 0.4777016341686249,
|
| 5743 |
+
"learning_rate": 9.476640437570562e-05,
|
| 5744 |
+
"loss": 0.7982,
|
| 5745 |
+
"step": 816
|
| 5746 |
+
},
|
| 5747 |
+
{
|
| 5748 |
+
"epoch": 0.36976691559176283,
|
| 5749 |
+
"grad_norm": 0.4785037040710449,
|
| 5750 |
+
"learning_rate": 9.456530696971999e-05,
|
| 5751 |
+
"loss": 0.701,
|
| 5752 |
+
"step": 817
|
| 5753 |
+
},
|
| 5754 |
+
{
|
| 5755 |
+
"epoch": 0.3702195066757185,
|
| 5756 |
+
"grad_norm": 0.4874170422554016,
|
| 5757 |
+
"learning_rate": 9.436423160443625e-05,
|
| 5758 |
+
"loss": 0.8509,
|
| 5759 |
+
"step": 818
|
| 5760 |
+
},
|
| 5761 |
+
{
|
| 5762 |
+
"epoch": 0.37067209775967414,
|
| 5763 |
+
"grad_norm": 0.5621728897094727,
|
| 5764 |
+
"learning_rate": 9.416317909532683e-05,
|
| 5765 |
+
"loss": 1.0424,
|
| 5766 |
+
"step": 819
|
| 5767 |
+
},
|
| 5768 |
+
{
|
| 5769 |
+
"epoch": 0.37112468884362976,
|
| 5770 |
+
"grad_norm": 0.5205615162849426,
|
| 5771 |
+
"learning_rate": 9.396215025777139e-05,
|
| 5772 |
+
"loss": 0.8152,
|
| 5773 |
+
"step": 820
|
| 5774 |
+
},
|
| 5775 |
+
{
|
| 5776 |
+
"epoch": 0.37157727992758544,
|
| 5777 |
+
"grad_norm": 0.47887948155403137,
|
| 5778 |
+
"learning_rate": 9.376114590705368e-05,
|
| 5779 |
+
"loss": 0.7467,
|
| 5780 |
+
"step": 821
|
| 5781 |
+
},
|
| 5782 |
+
{
|
| 5783 |
+
"epoch": 0.37202987101154106,
|
| 5784 |
+
"grad_norm": 0.511816680431366,
|
| 5785 |
+
"learning_rate": 9.356016685835806e-05,
|
| 5786 |
+
"loss": 0.8711,
|
| 5787 |
+
"step": 822
|
| 5788 |
+
},
|
| 5789 |
+
{
|
| 5790 |
+
"epoch": 0.37248246209549674,
|
| 5791 |
+
"grad_norm": 0.4856937527656555,
|
| 5792 |
+
"learning_rate": 9.335921392676631e-05,
|
| 5793 |
+
"loss": 0.8454,
|
| 5794 |
+
"step": 823
|
| 5795 |
+
},
|
| 5796 |
+
{
|
| 5797 |
+
"epoch": 0.37293505317945236,
|
| 5798 |
+
"grad_norm": 0.563502311706543,
|
| 5799 |
+
"learning_rate": 9.315828792725438e-05,
|
| 5800 |
+
"loss": 0.9715,
|
| 5801 |
+
"step": 824
|
| 5802 |
+
},
|
| 5803 |
+
{
|
| 5804 |
+
"epoch": 0.37338764426340804,
|
| 5805 |
+
"grad_norm": 0.5120623111724854,
|
| 5806 |
+
"learning_rate": 9.295738967468881e-05,
|
| 5807 |
+
"loss": 0.8401,
|
| 5808 |
+
"step": 825
|
| 5809 |
+
},
|
| 5810 |
+
{
|
| 5811 |
+
"epoch": 0.37384023534736366,
|
| 5812 |
+
"grad_norm": 0.43110066652297974,
|
| 5813 |
+
"learning_rate": 9.275651998382377e-05,
|
| 5814 |
+
"loss": 0.7352,
|
| 5815 |
+
"step": 826
|
| 5816 |
+
},
|
| 5817 |
+
{
|
| 5818 |
+
"epoch": 0.3742928264313193,
|
| 5819 |
+
"grad_norm": 0.5547953248023987,
|
| 5820 |
+
"learning_rate": 9.255567966929751e-05,
|
| 5821 |
+
"loss": 1.0781,
|
| 5822 |
+
"step": 827
|
| 5823 |
+
},
|
| 5824 |
+
{
|
| 5825 |
+
"epoch": 0.37474541751527496,
|
| 5826 |
+
"grad_norm": 0.5874584913253784,
|
| 5827 |
+
"learning_rate": 9.235486954562919e-05,
|
| 5828 |
+
"loss": 0.7856,
|
| 5829 |
+
"step": 828
|
| 5830 |
+
},
|
| 5831 |
+
{
|
| 5832 |
+
"epoch": 0.3751980085992306,
|
| 5833 |
+
"grad_norm": 0.4563722610473633,
|
| 5834 |
+
"learning_rate": 9.215409042721552e-05,
|
| 5835 |
+
"loss": 0.737,
|
| 5836 |
+
"step": 829
|
| 5837 |
+
},
|
| 5838 |
+
{
|
| 5839 |
+
"epoch": 0.37565059968318626,
|
| 5840 |
+
"grad_norm": 0.5010003447532654,
|
| 5841 |
+
"learning_rate": 9.195334312832742e-05,
|
| 5842 |
+
"loss": 0.8458,
|
| 5843 |
+
"step": 830
|
| 5844 |
+
},
|
| 5845 |
+
{
|
| 5846 |
+
"epoch": 0.3761031907671419,
|
| 5847 |
+
"grad_norm": 0.44281336665153503,
|
| 5848 |
+
"learning_rate": 9.175262846310682e-05,
|
| 5849 |
+
"loss": 0.686,
|
| 5850 |
+
"step": 831
|
| 5851 |
+
},
|
| 5852 |
+
{
|
| 5853 |
+
"epoch": 0.37655578185109756,
|
| 5854 |
+
"grad_norm": 0.5412707328796387,
|
| 5855 |
+
"learning_rate": 9.155194724556331e-05,
|
| 5856 |
+
"loss": 1.0169,
|
| 5857 |
+
"step": 832
|
| 5858 |
+
},
|
| 5859 |
+
{
|
| 5860 |
+
"epoch": 0.3770083729350532,
|
| 5861 |
+
"grad_norm": 0.4889048933982849,
|
| 5862 |
+
"learning_rate": 9.135130028957076e-05,
|
| 5863 |
+
"loss": 0.7573,
|
| 5864 |
+
"step": 833
|
| 5865 |
+
},
|
| 5866 |
+
{
|
| 5867 |
+
"epoch": 0.3774609640190088,
|
| 5868 |
+
"grad_norm": 0.5044034123420715,
|
| 5869 |
+
"learning_rate": 9.115068840886417e-05,
|
| 5870 |
+
"loss": 0.7249,
|
| 5871 |
+
"step": 834
|
| 5872 |
+
},
|
| 5873 |
+
{
|
| 5874 |
+
"epoch": 0.3779135551029645,
|
| 5875 |
+
"grad_norm": 0.45018038153648376,
|
| 5876 |
+
"learning_rate": 9.095011241703623e-05,
|
| 5877 |
+
"loss": 0.6717,
|
| 5878 |
+
"step": 835
|
| 5879 |
+
},
|
| 5880 |
+
{
|
| 5881 |
+
"epoch": 0.3783661461869201,
|
| 5882 |
+
"grad_norm": 0.5517880320549011,
|
| 5883 |
+
"learning_rate": 9.07495731275341e-05,
|
| 5884 |
+
"loss": 1.0841,
|
| 5885 |
+
"step": 836
|
| 5886 |
+
},
|
| 5887 |
+
{
|
| 5888 |
+
"epoch": 0.3788187372708758,
|
| 5889 |
+
"grad_norm": 0.41919058561325073,
|
| 5890 |
+
"learning_rate": 9.054907135365615e-05,
|
| 5891 |
+
"loss": 0.6097,
|
| 5892 |
+
"step": 837
|
| 5893 |
+
},
|
| 5894 |
+
{
|
| 5895 |
+
"epoch": 0.3792713283548314,
|
| 5896 |
+
"grad_norm": 0.4952158033847809,
|
| 5897 |
+
"learning_rate": 9.034860790854849e-05,
|
| 5898 |
+
"loss": 0.7001,
|
| 5899 |
+
"step": 838
|
| 5900 |
+
},
|
| 5901 |
+
{
|
| 5902 |
+
"epoch": 0.3797239194387871,
|
| 5903 |
+
"grad_norm": 0.5002701878547668,
|
| 5904 |
+
"learning_rate": 9.01481836052019e-05,
|
| 5905 |
+
"loss": 0.7987,
|
| 5906 |
+
"step": 839
|
| 5907 |
+
},
|
| 5908 |
+
{
|
| 5909 |
+
"epoch": 0.3801765105227427,
|
| 5910 |
+
"grad_norm": 0.42661139369010925,
|
| 5911 |
+
"learning_rate": 8.994779925644831e-05,
|
| 5912 |
+
"loss": 0.6824,
|
| 5913 |
+
"step": 840
|
| 5914 |
+
},
|
| 5915 |
+
{
|
| 5916 |
+
"epoch": 0.38062910160669833,
|
| 5917 |
+
"grad_norm": 0.5633373856544495,
|
| 5918 |
+
"learning_rate": 8.974745567495768e-05,
|
| 5919 |
+
"loss": 0.896,
|
| 5920 |
+
"step": 841
|
| 5921 |
+
},
|
| 5922 |
+
{
|
| 5923 |
+
"epoch": 0.381081692690654,
|
| 5924 |
+
"grad_norm": 0.5182005167007446,
|
| 5925 |
+
"learning_rate": 8.954715367323468e-05,
|
| 5926 |
+
"loss": 0.8394,
|
| 5927 |
+
"step": 842
|
| 5928 |
+
},
|
| 5929 |
+
{
|
| 5930 |
+
"epoch": 0.38153428377460963,
|
| 5931 |
+
"grad_norm": 0.4467911720275879,
|
| 5932 |
+
"learning_rate": 8.934689406361521e-05,
|
| 5933 |
+
"loss": 0.8438,
|
| 5934 |
+
"step": 843
|
| 5935 |
+
},
|
| 5936 |
+
{
|
| 5937 |
+
"epoch": 0.3819868748585653,
|
| 5938 |
+
"grad_norm": 0.3690691590309143,
|
| 5939 |
+
"learning_rate": 8.914667765826338e-05,
|
| 5940 |
+
"loss": 0.5102,
|
| 5941 |
+
"step": 844
|
| 5942 |
+
},
|
| 5943 |
+
{
|
| 5944 |
+
"epoch": 0.38243946594252093,
|
| 5945 |
+
"grad_norm": 0.5445937514305115,
|
| 5946 |
+
"learning_rate": 8.894650526916803e-05,
|
| 5947 |
+
"loss": 0.8618,
|
| 5948 |
+
"step": 845
|
| 5949 |
+
},
|
| 5950 |
+
{
|
| 5951 |
+
"epoch": 0.38289205702647655,
|
| 5952 |
+
"grad_norm": 0.5911957621574402,
|
| 5953 |
+
"learning_rate": 8.874637770813946e-05,
|
| 5954 |
+
"loss": 1.0281,
|
| 5955 |
+
"step": 846
|
| 5956 |
+
},
|
| 5957 |
+
{
|
| 5958 |
+
"epoch": 0.38334464811043223,
|
| 5959 |
+
"grad_norm": 0.4994654953479767,
|
| 5960 |
+
"learning_rate": 8.854629578680624e-05,
|
| 5961 |
+
"loss": 1.0337,
|
| 5962 |
+
"step": 847
|
| 5963 |
+
},
|
| 5964 |
+
{
|
| 5965 |
+
"epoch": 0.38379723919438785,
|
| 5966 |
+
"grad_norm": 0.48666390776634216,
|
| 5967 |
+
"learning_rate": 8.834626031661178e-05,
|
| 5968 |
+
"loss": 0.79,
|
| 5969 |
+
"step": 848
|
| 5970 |
+
},
|
| 5971 |
+
{
|
| 5972 |
+
"epoch": 0.38424983027834353,
|
| 5973 |
+
"grad_norm": 0.4750109314918518,
|
| 5974 |
+
"learning_rate": 8.814627210881117e-05,
|
| 5975 |
+
"loss": 0.8021,
|
| 5976 |
+
"step": 849
|
| 5977 |
+
},
|
| 5978 |
+
{
|
| 5979 |
+
"epoch": 0.38470242136229915,
|
| 5980 |
+
"grad_norm": 0.46276208758354187,
|
| 5981 |
+
"learning_rate": 8.79463319744677e-05,
|
| 5982 |
+
"loss": 0.7748,
|
| 5983 |
+
"step": 850
|
| 5984 |
+
},
|
| 5985 |
+
{
|
| 5986 |
+
"epoch": 0.38515501244625483,
|
| 5987 |
+
"grad_norm": 0.5422043204307556,
|
| 5988 |
+
"learning_rate": 8.774644072444986e-05,
|
| 5989 |
+
"loss": 0.87,
|
| 5990 |
+
"step": 851
|
| 5991 |
+
},
|
| 5992 |
+
{
|
| 5993 |
+
"epoch": 0.38560760353021045,
|
| 5994 |
+
"grad_norm": 0.5052550435066223,
|
| 5995 |
+
"learning_rate": 8.75465991694278e-05,
|
| 5996 |
+
"loss": 0.8123,
|
| 5997 |
+
"step": 852
|
| 5998 |
+
},
|
| 5999 |
+
{
|
| 6000 |
+
"epoch": 0.3860601946141661,
|
| 6001 |
+
"grad_norm": 0.4825555384159088,
|
| 6002 |
+
"learning_rate": 8.73468081198701e-05,
|
| 6003 |
+
"loss": 0.7966,
|
| 6004 |
+
"step": 853
|
| 6005 |
+
},
|
| 6006 |
+
{
|
| 6007 |
+
"epoch": 0.38651278569812175,
|
| 6008 |
+
"grad_norm": 0.5406911373138428,
|
| 6009 |
+
"learning_rate": 8.714706838604055e-05,
|
| 6010 |
+
"loss": 1.0109,
|
| 6011 |
+
"step": 854
|
| 6012 |
+
},
|
| 6013 |
+
{
|
| 6014 |
+
"epoch": 0.3869653767820774,
|
| 6015 |
+
"grad_norm": 0.5366398096084595,
|
| 6016 |
+
"learning_rate": 8.694738077799488e-05,
|
| 6017 |
+
"loss": 1.0179,
|
| 6018 |
+
"step": 855
|
| 6019 |
+
},
|
| 6020 |
+
{
|
| 6021 |
+
"epoch": 0.38741796786603305,
|
| 6022 |
+
"grad_norm": 0.5232101678848267,
|
| 6023 |
+
"learning_rate": 8.674774610557728e-05,
|
| 6024 |
+
"loss": 0.7802,
|
| 6025 |
+
"step": 856
|
| 6026 |
+
},
|
| 6027 |
+
{
|
| 6028 |
+
"epoch": 0.3878705589499887,
|
| 6029 |
+
"grad_norm": 0.47566697001457214,
|
| 6030 |
+
"learning_rate": 8.654816517841741e-05,
|
| 6031 |
+
"loss": 0.8271,
|
| 6032 |
+
"step": 857
|
| 6033 |
+
},
|
| 6034 |
+
{
|
| 6035 |
+
"epoch": 0.38832315003394435,
|
| 6036 |
+
"grad_norm": 0.5479469895362854,
|
| 6037 |
+
"learning_rate": 8.634863880592686e-05,
|
| 6038 |
+
"loss": 1.0688,
|
| 6039 |
+
"step": 858
|
| 6040 |
+
},
|
| 6041 |
+
{
|
| 6042 |
+
"epoch": 0.3887757411179,
|
| 6043 |
+
"grad_norm": 0.47462597489356995,
|
| 6044 |
+
"learning_rate": 8.614916779729603e-05,
|
| 6045 |
+
"loss": 0.8378,
|
| 6046 |
+
"step": 859
|
| 6047 |
+
},
|
| 6048 |
+
{
|
| 6049 |
+
"epoch": 0.3892283322018556,
|
| 6050 |
+
"grad_norm": 0.5492944121360779,
|
| 6051 |
+
"learning_rate": 8.594975296149076e-05,
|
| 6052 |
+
"loss": 0.9606,
|
| 6053 |
+
"step": 860
|
| 6054 |
+
},
|
| 6055 |
+
{
|
| 6056 |
+
"epoch": 0.3896809232858113,
|
| 6057 |
+
"grad_norm": 0.40749794244766235,
|
| 6058 |
+
"learning_rate": 8.575039510724905e-05,
|
| 6059 |
+
"loss": 0.6763,
|
| 6060 |
+
"step": 861
|
| 6061 |
+
},
|
| 6062 |
+
{
|
| 6063 |
+
"epoch": 0.3901335143697669,
|
| 6064 |
+
"grad_norm": 0.5637528300285339,
|
| 6065 |
+
"learning_rate": 8.55510950430779e-05,
|
| 6066 |
+
"loss": 0.9517,
|
| 6067 |
+
"step": 862
|
| 6068 |
+
},
|
| 6069 |
+
{
|
| 6070 |
+
"epoch": 0.3905861054537226,
|
| 6071 |
+
"grad_norm": 0.48673295974731445,
|
| 6072 |
+
"learning_rate": 8.535185357724982e-05,
|
| 6073 |
+
"loss": 0.7629,
|
| 6074 |
+
"step": 863
|
| 6075 |
+
},
|
| 6076 |
+
{
|
| 6077 |
+
"epoch": 0.3910386965376782,
|
| 6078 |
+
"grad_norm": 0.49013152718544006,
|
| 6079 |
+
"learning_rate": 8.515267151779974e-05,
|
| 6080 |
+
"loss": 0.7738,
|
| 6081 |
+
"step": 864
|
| 6082 |
+
},
|
| 6083 |
+
{
|
| 6084 |
+
"epoch": 0.3914912876216339,
|
| 6085 |
+
"grad_norm": 0.5754150748252869,
|
| 6086 |
+
"learning_rate": 8.495354967252169e-05,
|
| 6087 |
+
"loss": 1.0042,
|
| 6088 |
+
"step": 865
|
| 6089 |
+
},
|
| 6090 |
+
{
|
| 6091 |
+
"epoch": 0.3919438787055895,
|
| 6092 |
+
"grad_norm": 0.5181888937950134,
|
| 6093 |
+
"learning_rate": 8.475448884896547e-05,
|
| 6094 |
+
"loss": 1.0482,
|
| 6095 |
+
"step": 866
|
| 6096 |
+
},
|
| 6097 |
+
{
|
| 6098 |
+
"epoch": 0.3923964697895451,
|
| 6099 |
+
"grad_norm": 0.6057239770889282,
|
| 6100 |
+
"learning_rate": 8.455548985443334e-05,
|
| 6101 |
+
"loss": 1.2377,
|
| 6102 |
+
"step": 867
|
| 6103 |
+
},
|
| 6104 |
+
{
|
| 6105 |
+
"epoch": 0.3928490608735008,
|
| 6106 |
+
"grad_norm": 0.5204604864120483,
|
| 6107 |
+
"learning_rate": 8.435655349597689e-05,
|
| 6108 |
+
"loss": 0.8495,
|
| 6109 |
+
"step": 868
|
| 6110 |
+
},
|
| 6111 |
+
{
|
| 6112 |
+
"epoch": 0.3933016519574564,
|
| 6113 |
+
"grad_norm": 0.48216450214385986,
|
| 6114 |
+
"learning_rate": 8.41576805803937e-05,
|
| 6115 |
+
"loss": 0.6864,
|
| 6116 |
+
"step": 869
|
| 6117 |
+
},
|
| 6118 |
+
{
|
| 6119 |
+
"epoch": 0.3937542430414121,
|
| 6120 |
+
"grad_norm": 0.40034219622612,
|
| 6121 |
+
"learning_rate": 8.395887191422397e-05,
|
| 6122 |
+
"loss": 0.5482,
|
| 6123 |
+
"step": 870
|
| 6124 |
+
},
|
| 6125 |
+
{
|
| 6126 |
+
"epoch": 0.3942068341253677,
|
| 6127 |
+
"grad_norm": 0.49549373984336853,
|
| 6128 |
+
"learning_rate": 8.37601283037474e-05,
|
| 6129 |
+
"loss": 0.8129,
|
| 6130 |
+
"step": 871
|
| 6131 |
+
},
|
| 6132 |
+
{
|
| 6133 |
+
"epoch": 0.3946594252093234,
|
| 6134 |
+
"grad_norm": 0.4993366599082947,
|
| 6135 |
+
"learning_rate": 8.356145055497981e-05,
|
| 6136 |
+
"loss": 0.7104,
|
| 6137 |
+
"step": 872
|
| 6138 |
+
},
|
| 6139 |
+
{
|
| 6140 |
+
"epoch": 0.395112016293279,
|
| 6141 |
+
"grad_norm": 0.5507094264030457,
|
| 6142 |
+
"learning_rate": 8.336283947366992e-05,
|
| 6143 |
+
"loss": 0.9944,
|
| 6144 |
+
"step": 873
|
| 6145 |
+
},
|
| 6146 |
+
{
|
| 6147 |
+
"epoch": 0.39556460737723464,
|
| 6148 |
+
"grad_norm": 0.5083698034286499,
|
| 6149 |
+
"learning_rate": 8.316429586529615e-05,
|
| 6150 |
+
"loss": 0.9294,
|
| 6151 |
+
"step": 874
|
| 6152 |
+
},
|
| 6153 |
+
{
|
| 6154 |
+
"epoch": 0.3960171984611903,
|
| 6155 |
+
"grad_norm": 0.5367676615715027,
|
| 6156 |
+
"learning_rate": 8.296582053506317e-05,
|
| 6157 |
+
"loss": 0.996,
|
| 6158 |
+
"step": 875
|
| 6159 |
+
},
|
| 6160 |
+
{
|
| 6161 |
+
"epoch": 0.39646978954514595,
|
| 6162 |
+
"grad_norm": 0.4346739649772644,
|
| 6163 |
+
"learning_rate": 8.276741428789883e-05,
|
| 6164 |
+
"loss": 0.633,
|
| 6165 |
+
"step": 876
|
| 6166 |
+
},
|
| 6167 |
+
{
|
| 6168 |
+
"epoch": 0.3969223806291016,
|
| 6169 |
+
"grad_norm": 0.5170437097549438,
|
| 6170 |
+
"learning_rate": 8.256907792845072e-05,
|
| 6171 |
+
"loss": 1.0517,
|
| 6172 |
+
"step": 877
|
| 6173 |
+
},
|
| 6174 |
+
{
|
| 6175 |
+
"epoch": 0.39737497171305725,
|
| 6176 |
+
"grad_norm": 0.45783472061157227,
|
| 6177 |
+
"learning_rate": 8.237081226108311e-05,
|
| 6178 |
+
"loss": 0.7076,
|
| 6179 |
+
"step": 878
|
| 6180 |
+
},
|
| 6181 |
+
{
|
| 6182 |
+
"epoch": 0.3978275627970129,
|
| 6183 |
+
"grad_norm": 0.7020736336708069,
|
| 6184 |
+
"learning_rate": 8.217261808987351e-05,
|
| 6185 |
+
"loss": 0.9009,
|
| 6186 |
+
"step": 879
|
| 6187 |
+
},
|
| 6188 |
+
{
|
| 6189 |
+
"epoch": 0.39828015388096855,
|
| 6190 |
+
"grad_norm": 0.5104503631591797,
|
| 6191 |
+
"learning_rate": 8.197449621860943e-05,
|
| 6192 |
+
"loss": 0.9576,
|
| 6193 |
+
"step": 880
|
| 6194 |
+
},
|
| 6195 |
+
{
|
| 6196 |
+
"epoch": 0.39873274496492417,
|
| 6197 |
+
"grad_norm": 0.4474295675754547,
|
| 6198 |
+
"learning_rate": 8.177644745078526e-05,
|
| 6199 |
+
"loss": 0.6848,
|
| 6200 |
+
"step": 881
|
| 6201 |
+
},
|
| 6202 |
+
{
|
| 6203 |
+
"epoch": 0.39918533604887985,
|
| 6204 |
+
"grad_norm": 0.5501548051834106,
|
| 6205 |
+
"learning_rate": 8.157847258959885e-05,
|
| 6206 |
+
"loss": 1.0901,
|
| 6207 |
+
"step": 882
|
| 6208 |
+
},
|
| 6209 |
+
{
|
| 6210 |
+
"epoch": 0.39963792713283547,
|
| 6211 |
+
"grad_norm": 0.4706266522407532,
|
| 6212 |
+
"learning_rate": 8.138057243794833e-05,
|
| 6213 |
+
"loss": 0.7566,
|
| 6214 |
+
"step": 883
|
| 6215 |
+
},
|
| 6216 |
+
{
|
| 6217 |
+
"epoch": 0.40009051821679115,
|
| 6218 |
+
"grad_norm": 0.46165910363197327,
|
| 6219 |
+
"learning_rate": 8.118274779842888e-05,
|
| 6220 |
+
"loss": 0.7071,
|
| 6221 |
+
"step": 884
|
| 6222 |
+
},
|
| 6223 |
+
{
|
| 6224 |
+
"epoch": 0.40054310930074677,
|
| 6225 |
+
"grad_norm": 0.5035961270332336,
|
| 6226 |
+
"learning_rate": 8.098499947332934e-05,
|
| 6227 |
+
"loss": 0.8654,
|
| 6228 |
+
"step": 885
|
| 6229 |
+
},
|
| 6230 |
+
{
|
| 6231 |
+
"epoch": 0.40099570038470245,
|
| 6232 |
+
"grad_norm": 0.464796781539917,
|
| 6233 |
+
"learning_rate": 8.078732826462915e-05,
|
| 6234 |
+
"loss": 0.6857,
|
| 6235 |
+
"step": 886
|
| 6236 |
+
},
|
| 6237 |
+
{
|
| 6238 |
+
"epoch": 0.40144829146865807,
|
| 6239 |
+
"grad_norm": 0.5464978814125061,
|
| 6240 |
+
"learning_rate": 8.0589734973995e-05,
|
| 6241 |
+
"loss": 0.9781,
|
| 6242 |
+
"step": 887
|
| 6243 |
+
},
|
| 6244 |
+
{
|
| 6245 |
+
"epoch": 0.4019008825526137,
|
| 6246 |
+
"grad_norm": 0.42914673686027527,
|
| 6247 |
+
"learning_rate": 8.03922204027775e-05,
|
| 6248 |
+
"loss": 0.7287,
|
| 6249 |
+
"step": 888
|
| 6250 |
+
},
|
| 6251 |
+
{
|
| 6252 |
+
"epoch": 0.40235347363656937,
|
| 6253 |
+
"grad_norm": 0.5215734839439392,
|
| 6254 |
+
"learning_rate": 8.019478535200806e-05,
|
| 6255 |
+
"loss": 0.9839,
|
| 6256 |
+
"step": 889
|
| 6257 |
+
},
|
| 6258 |
+
{
|
| 6259 |
+
"epoch": 0.402806064720525,
|
| 6260 |
+
"grad_norm": 0.5467308163642883,
|
| 6261 |
+
"learning_rate": 7.999743062239557e-05,
|
| 6262 |
+
"loss": 0.8938,
|
| 6263 |
+
"step": 890
|
| 6264 |
+
},
|
| 6265 |
+
{
|
| 6266 |
+
"epoch": 0.40325865580448067,
|
| 6267 |
+
"grad_norm": 0.5061361789703369,
|
| 6268 |
+
"learning_rate": 7.98001570143232e-05,
|
| 6269 |
+
"loss": 0.7561,
|
| 6270 |
+
"step": 891
|
| 6271 |
+
},
|
| 6272 |
+
{
|
| 6273 |
+
"epoch": 0.4037112468884363,
|
| 6274 |
+
"grad_norm": 0.4637202322483063,
|
| 6275 |
+
"learning_rate": 7.960296532784515e-05,
|
| 6276 |
+
"loss": 0.8056,
|
| 6277 |
+
"step": 892
|
| 6278 |
+
},
|
| 6279 |
+
{
|
| 6280 |
+
"epoch": 0.40416383797239197,
|
| 6281 |
+
"grad_norm": 0.47489434480667114,
|
| 6282 |
+
"learning_rate": 7.940585636268327e-05,
|
| 6283 |
+
"loss": 0.833,
|
| 6284 |
+
"step": 893
|
| 6285 |
+
},
|
| 6286 |
+
{
|
| 6287 |
+
"epoch": 0.4046164290563476,
|
| 6288 |
+
"grad_norm": 0.555900514125824,
|
| 6289 |
+
"learning_rate": 7.920883091822408e-05,
|
| 6290 |
+
"loss": 0.8873,
|
| 6291 |
+
"step": 894
|
| 6292 |
+
},
|
| 6293 |
+
{
|
| 6294 |
+
"epoch": 0.4050690201403032,
|
| 6295 |
+
"grad_norm": 0.5026564002037048,
|
| 6296 |
+
"learning_rate": 7.901188979351526e-05,
|
| 6297 |
+
"loss": 1.0811,
|
| 6298 |
+
"step": 895
|
| 6299 |
+
},
|
| 6300 |
+
{
|
| 6301 |
+
"epoch": 0.4055216112242589,
|
| 6302 |
+
"grad_norm": 0.5691704154014587,
|
| 6303 |
+
"learning_rate": 7.881503378726258e-05,
|
| 6304 |
+
"loss": 0.7082,
|
| 6305 |
+
"step": 896
|
| 6306 |
+
},
|
| 6307 |
+
{
|
| 6308 |
+
"epoch": 0.4059742023082145,
|
| 6309 |
+
"grad_norm": 0.42396095395088196,
|
| 6310 |
+
"learning_rate": 7.86182636978266e-05,
|
| 6311 |
+
"loss": 0.6481,
|
| 6312 |
+
"step": 897
|
| 6313 |
+
},
|
| 6314 |
+
{
|
| 6315 |
+
"epoch": 0.4064267933921702,
|
| 6316 |
+
"grad_norm": 0.4842861294746399,
|
| 6317 |
+
"learning_rate": 7.84215803232194e-05,
|
| 6318 |
+
"loss": 0.6791,
|
| 6319 |
+
"step": 898
|
| 6320 |
+
},
|
| 6321 |
+
{
|
| 6322 |
+
"epoch": 0.4068793844761258,
|
| 6323 |
+
"grad_norm": 0.4645484685897827,
|
| 6324 |
+
"learning_rate": 7.822498446110145e-05,
|
| 6325 |
+
"loss": 0.6694,
|
| 6326 |
+
"step": 899
|
| 6327 |
+
},
|
| 6328 |
+
{
|
| 6329 |
+
"epoch": 0.4073319755600815,
|
| 6330 |
+
"grad_norm": 0.5308927297592163,
|
| 6331 |
+
"learning_rate": 7.802847690877832e-05,
|
| 6332 |
+
"loss": 0.9689,
|
| 6333 |
+
"step": 900
|
| 6334 |
+
},
|
| 6335 |
+
{
|
| 6336 |
+
"epoch": 0.4077845666440371,
|
| 6337 |
+
"grad_norm": 0.4862045347690582,
|
| 6338 |
+
"learning_rate": 7.78320584631973e-05,
|
| 6339 |
+
"loss": 0.8439,
|
| 6340 |
+
"step": 901
|
| 6341 |
+
},
|
| 6342 |
+
{
|
| 6343 |
+
"epoch": 0.40823715772799274,
|
| 6344 |
+
"grad_norm": 0.5172306299209595,
|
| 6345 |
+
"learning_rate": 7.763572992094447e-05,
|
| 6346 |
+
"loss": 0.9196,
|
| 6347 |
+
"step": 902
|
| 6348 |
+
},
|
| 6349 |
+
{
|
| 6350 |
+
"epoch": 0.4086897488119484,
|
| 6351 |
+
"grad_norm": 0.4229103624820709,
|
| 6352 |
+
"learning_rate": 7.743949207824119e-05,
|
| 6353 |
+
"loss": 0.5522,
|
| 6354 |
+
"step": 903
|
| 6355 |
+
},
|
| 6356 |
+
{
|
| 6357 |
+
"epoch": 0.40914233989590404,
|
| 6358 |
+
"grad_norm": 0.45848751068115234,
|
| 6359 |
+
"learning_rate": 7.7243345730941e-05,
|
| 6360 |
+
"loss": 0.8773,
|
| 6361 |
+
"step": 904
|
| 6362 |
+
},
|
| 6363 |
+
{
|
| 6364 |
+
"epoch": 0.4095949309798597,
|
| 6365 |
+
"grad_norm": 0.4387413263320923,
|
| 6366 |
+
"learning_rate": 7.704729167452647e-05,
|
| 6367 |
+
"loss": 0.8242,
|
| 6368 |
+
"step": 905
|
| 6369 |
+
},
|
| 6370 |
+
{
|
| 6371 |
+
"epoch": 0.41004752206381534,
|
| 6372 |
+
"grad_norm": 0.49762117862701416,
|
| 6373 |
+
"learning_rate": 7.685133070410571e-05,
|
| 6374 |
+
"loss": 0.8721,
|
| 6375 |
+
"step": 906
|
| 6376 |
+
},
|
| 6377 |
+
{
|
| 6378 |
+
"epoch": 0.41050011314777096,
|
| 6379 |
+
"grad_norm": 0.47309303283691406,
|
| 6380 |
+
"learning_rate": 7.66554636144095e-05,
|
| 6381 |
+
"loss": 0.7279,
|
| 6382 |
+
"step": 907
|
| 6383 |
+
},
|
| 6384 |
+
{
|
| 6385 |
+
"epoch": 0.41095270423172664,
|
| 6386 |
+
"grad_norm": 0.6089125275611877,
|
| 6387 |
+
"learning_rate": 7.64596911997877e-05,
|
| 6388 |
+
"loss": 0.9264,
|
| 6389 |
+
"step": 908
|
| 6390 |
+
},
|
| 6391 |
+
{
|
| 6392 |
+
"epoch": 0.41140529531568226,
|
| 6393 |
+
"grad_norm": 0.46339258551597595,
|
| 6394 |
+
"learning_rate": 7.626401425420634e-05,
|
| 6395 |
+
"loss": 0.6112,
|
| 6396 |
+
"step": 909
|
| 6397 |
+
},
|
| 6398 |
+
{
|
| 6399 |
+
"epoch": 0.41185788639963794,
|
| 6400 |
+
"grad_norm": 0.5053806900978088,
|
| 6401 |
+
"learning_rate": 7.606843357124426e-05,
|
| 6402 |
+
"loss": 0.8201,
|
| 6403 |
+
"step": 910
|
| 6404 |
+
},
|
| 6405 |
+
{
|
| 6406 |
+
"epoch": 0.41231047748359356,
|
| 6407 |
+
"grad_norm": 0.5062787532806396,
|
| 6408 |
+
"learning_rate": 7.587294994408981e-05,
|
| 6409 |
+
"loss": 0.8513,
|
| 6410 |
+
"step": 911
|
| 6411 |
+
},
|
| 6412 |
+
{
|
| 6413 |
+
"epoch": 0.41276306856754924,
|
| 6414 |
+
"grad_norm": 0.5039675235748291,
|
| 6415 |
+
"learning_rate": 7.56775641655378e-05,
|
| 6416 |
+
"loss": 0.8486,
|
| 6417 |
+
"step": 912
|
| 6418 |
+
},
|
| 6419 |
+
{
|
| 6420 |
+
"epoch": 0.41321565965150486,
|
| 6421 |
+
"grad_norm": 0.528287410736084,
|
| 6422 |
+
"learning_rate": 7.548227702798624e-05,
|
| 6423 |
+
"loss": 0.9241,
|
| 6424 |
+
"step": 913
|
| 6425 |
+
},
|
| 6426 |
+
{
|
| 6427 |
+
"epoch": 0.4136682507354605,
|
| 6428 |
+
"grad_norm": 0.5342898368835449,
|
| 6429 |
+
"learning_rate": 7.528708932343304e-05,
|
| 6430 |
+
"loss": 0.911,
|
| 6431 |
+
"step": 914
|
| 6432 |
+
},
|
| 6433 |
+
{
|
| 6434 |
+
"epoch": 0.41412084181941616,
|
| 6435 |
+
"grad_norm": 0.5621039271354675,
|
| 6436 |
+
"learning_rate": 7.509200184347282e-05,
|
| 6437 |
+
"loss": 1.0581,
|
| 6438 |
+
"step": 915
|
| 6439 |
+
},
|
| 6440 |
+
{
|
| 6441 |
+
"epoch": 0.4145734329033718,
|
| 6442 |
+
"grad_norm": 0.5111497044563293,
|
| 6443 |
+
"learning_rate": 7.489701537929384e-05,
|
| 6444 |
+
"loss": 0.8369,
|
| 6445 |
+
"step": 916
|
| 6446 |
+
},
|
| 6447 |
+
{
|
| 6448 |
+
"epoch": 0.41502602398732746,
|
| 6449 |
+
"grad_norm": 0.5669052600860596,
|
| 6450 |
+
"learning_rate": 7.470213072167456e-05,
|
| 6451 |
+
"loss": 0.9215,
|
| 6452 |
+
"step": 917
|
| 6453 |
+
},
|
| 6454 |
+
{
|
| 6455 |
+
"epoch": 0.4154786150712831,
|
| 6456 |
+
"grad_norm": 0.4463648200035095,
|
| 6457 |
+
"learning_rate": 7.450734866098066e-05,
|
| 6458 |
+
"loss": 0.7924,
|
| 6459 |
+
"step": 918
|
| 6460 |
+
},
|
| 6461 |
+
{
|
| 6462 |
+
"epoch": 0.41593120615523876,
|
| 6463 |
+
"grad_norm": 0.5927613973617554,
|
| 6464 |
+
"learning_rate": 7.431266998716171e-05,
|
| 6465 |
+
"loss": 1.084,
|
| 6466 |
+
"step": 919
|
| 6467 |
+
},
|
| 6468 |
+
{
|
| 6469 |
+
"epoch": 0.4163837972391944,
|
| 6470 |
+
"grad_norm": 0.47022631764411926,
|
| 6471 |
+
"learning_rate": 7.411809548974792e-05,
|
| 6472 |
+
"loss": 0.6784,
|
| 6473 |
+
"step": 920
|
| 6474 |
+
},
|
| 6475 |
+
{
|
| 6476 |
+
"epoch": 0.41683638832315,
|
| 6477 |
+
"grad_norm": 0.46409207582473755,
|
| 6478 |
+
"learning_rate": 7.392362595784711e-05,
|
| 6479 |
+
"loss": 0.6136,
|
| 6480 |
+
"step": 921
|
| 6481 |
+
},
|
| 6482 |
+
{
|
| 6483 |
+
"epoch": 0.4172889794071057,
|
| 6484 |
+
"grad_norm": 0.5186243653297424,
|
| 6485 |
+
"learning_rate": 7.372926218014131e-05,
|
| 6486 |
+
"loss": 0.8043,
|
| 6487 |
+
"step": 922
|
| 6488 |
+
},
|
| 6489 |
+
{
|
| 6490 |
+
"epoch": 0.4177415704910613,
|
| 6491 |
+
"grad_norm": 0.5423625707626343,
|
| 6492 |
+
"learning_rate": 7.353500494488373e-05,
|
| 6493 |
+
"loss": 0.931,
|
| 6494 |
+
"step": 923
|
| 6495 |
+
},
|
| 6496 |
+
{
|
| 6497 |
+
"epoch": 0.418194161575017,
|
| 6498 |
+
"grad_norm": 0.5815045833587646,
|
| 6499 |
+
"learning_rate": 7.334085503989547e-05,
|
| 6500 |
+
"loss": 0.9268,
|
| 6501 |
+
"step": 924
|
| 6502 |
+
},
|
| 6503 |
+
{
|
| 6504 |
+
"epoch": 0.4186467526589726,
|
| 6505 |
+
"grad_norm": 0.5250252485275269,
|
| 6506 |
+
"learning_rate": 7.314681325256232e-05,
|
| 6507 |
+
"loss": 0.962,
|
| 6508 |
+
"step": 925
|
| 6509 |
+
},
|
| 6510 |
+
{
|
| 6511 |
+
"epoch": 0.4190993437429283,
|
| 6512 |
+
"grad_norm": 0.48460298776626587,
|
| 6513 |
+
"learning_rate": 7.295288036983163e-05,
|
| 6514 |
+
"loss": 0.6581,
|
| 6515 |
+
"step": 926
|
| 6516 |
+
},
|
| 6517 |
+
{
|
| 6518 |
+
"epoch": 0.4195519348268839,
|
| 6519 |
+
"grad_norm": 0.5446643829345703,
|
| 6520 |
+
"learning_rate": 7.275905717820908e-05,
|
| 6521 |
+
"loss": 1.0548,
|
| 6522 |
+
"step": 927
|
| 6523 |
+
},
|
| 6524 |
+
{
|
| 6525 |
+
"epoch": 0.42000452591083953,
|
| 6526 |
+
"grad_norm": 0.4789769649505615,
|
| 6527 |
+
"learning_rate": 7.256534446375542e-05,
|
| 6528 |
+
"loss": 0.8688,
|
| 6529 |
+
"step": 928
|
| 6530 |
+
},
|
| 6531 |
+
{
|
| 6532 |
+
"epoch": 0.4204571169947952,
|
| 6533 |
+
"grad_norm": 0.4612770080566406,
|
| 6534 |
+
"learning_rate": 7.237174301208349e-05,
|
| 6535 |
+
"loss": 0.748,
|
| 6536 |
+
"step": 929
|
| 6537 |
+
},
|
| 6538 |
+
{
|
| 6539 |
+
"epoch": 0.42090970807875083,
|
| 6540 |
+
"grad_norm": 0.4356364607810974,
|
| 6541 |
+
"learning_rate": 7.217825360835473e-05,
|
| 6542 |
+
"loss": 0.6854,
|
| 6543 |
+
"step": 930
|
| 6544 |
+
},
|
| 6545 |
+
{
|
| 6546 |
+
"epoch": 0.4213622991627065,
|
| 6547 |
+
"grad_norm": 0.6402705907821655,
|
| 6548 |
+
"learning_rate": 7.198487703727632e-05,
|
| 6549 |
+
"loss": 0.9116,
|
| 6550 |
+
"step": 931
|
| 6551 |
+
},
|
| 6552 |
+
{
|
| 6553 |
+
"epoch": 0.42181489024666213,
|
| 6554 |
+
"grad_norm": 0.4785003066062927,
|
| 6555 |
+
"learning_rate": 7.179161408309776e-05,
|
| 6556 |
+
"loss": 0.9363,
|
| 6557 |
+
"step": 932
|
| 6558 |
+
},
|
| 6559 |
+
{
|
| 6560 |
+
"epoch": 0.4222674813306178,
|
| 6561 |
+
"grad_norm": 0.5057145357131958,
|
| 6562 |
+
"learning_rate": 7.159846552960774e-05,
|
| 6563 |
+
"loss": 0.7782,
|
| 6564 |
+
"step": 933
|
| 6565 |
+
},
|
| 6566 |
+
{
|
| 6567 |
+
"epoch": 0.42272007241457343,
|
| 6568 |
+
"grad_norm": 0.5009840726852417,
|
| 6569 |
+
"learning_rate": 7.14054321601311e-05,
|
| 6570 |
+
"loss": 0.8482,
|
| 6571 |
+
"step": 934
|
| 6572 |
+
},
|
| 6573 |
+
{
|
| 6574 |
+
"epoch": 0.42317266349852906,
|
| 6575 |
+
"grad_norm": 0.5882793664932251,
|
| 6576 |
+
"learning_rate": 7.121251475752539e-05,
|
| 6577 |
+
"loss": 1.0156,
|
| 6578 |
+
"step": 935
|
| 6579 |
+
},
|
| 6580 |
+
{
|
| 6581 |
+
"epoch": 0.42362525458248473,
|
| 6582 |
+
"grad_norm": 0.5955524444580078,
|
| 6583 |
+
"learning_rate": 7.101971410417803e-05,
|
| 6584 |
+
"loss": 0.9331,
|
| 6585 |
+
"step": 936
|
| 6586 |
+
},
|
| 6587 |
+
{
|
| 6588 |
+
"epoch": 0.42407784566644036,
|
| 6589 |
+
"grad_norm": 0.44543153047561646,
|
| 6590 |
+
"learning_rate": 7.082703098200282e-05,
|
| 6591 |
+
"loss": 0.7012,
|
| 6592 |
+
"step": 937
|
| 6593 |
+
},
|
| 6594 |
+
{
|
| 6595 |
+
"epoch": 0.42453043675039603,
|
| 6596 |
+
"grad_norm": 0.5537339448928833,
|
| 6597 |
+
"learning_rate": 7.063446617243694e-05,
|
| 6598 |
+
"loss": 0.71,
|
| 6599 |
+
"step": 938
|
| 6600 |
+
},
|
| 6601 |
+
{
|
| 6602 |
+
"epoch": 0.42498302783435166,
|
| 6603 |
+
"grad_norm": 0.6707674264907837,
|
| 6604 |
+
"learning_rate": 7.04420204564378e-05,
|
| 6605 |
+
"loss": 0.9452,
|
| 6606 |
+
"step": 939
|
| 6607 |
+
},
|
| 6608 |
+
{
|
| 6609 |
+
"epoch": 0.42543561891830733,
|
| 6610 |
+
"grad_norm": 0.4922984540462494,
|
| 6611 |
+
"learning_rate": 7.024969461447972e-05,
|
| 6612 |
+
"loss": 0.7476,
|
| 6613 |
+
"step": 940
|
| 6614 |
+
},
|
| 6615 |
+
{
|
| 6616 |
+
"epoch": 0.42588821000226296,
|
| 6617 |
+
"grad_norm": 0.5744497179985046,
|
| 6618 |
+
"learning_rate": 7.005748942655095e-05,
|
| 6619 |
+
"loss": 0.8709,
|
| 6620 |
+
"step": 941
|
| 6621 |
+
},
|
| 6622 |
+
{
|
| 6623 |
+
"epoch": 0.4263408010862186,
|
| 6624 |
+
"grad_norm": 0.49316924810409546,
|
| 6625 |
+
"learning_rate": 6.986540567215044e-05,
|
| 6626 |
+
"loss": 0.6702,
|
| 6627 |
+
"step": 942
|
| 6628 |
+
},
|
| 6629 |
+
{
|
| 6630 |
+
"epoch": 0.42679339217017426,
|
| 6631 |
+
"grad_norm": 0.5075472593307495,
|
| 6632 |
+
"learning_rate": 6.967344413028452e-05,
|
| 6633 |
+
"loss": 0.8238,
|
| 6634 |
+
"step": 943
|
| 6635 |
+
},
|
| 6636 |
+
{
|
| 6637 |
+
"epoch": 0.4272459832541299,
|
| 6638 |
+
"grad_norm": 0.5124140381813049,
|
| 6639 |
+
"learning_rate": 6.948160557946404e-05,
|
| 6640 |
+
"loss": 0.8917,
|
| 6641 |
+
"step": 944
|
| 6642 |
+
},
|
| 6643 |
+
{
|
| 6644 |
+
"epoch": 0.42769857433808556,
|
| 6645 |
+
"grad_norm": 0.49430105090141296,
|
| 6646 |
+
"learning_rate": 6.928989079770094e-05,
|
| 6647 |
+
"loss": 0.9076,
|
| 6648 |
+
"step": 945
|
| 6649 |
+
},
|
| 6650 |
+
{
|
| 6651 |
+
"epoch": 0.4281511654220412,
|
| 6652 |
+
"grad_norm": 0.536520779132843,
|
| 6653 |
+
"learning_rate": 6.909830056250527e-05,
|
| 6654 |
+
"loss": 0.9427,
|
| 6655 |
+
"step": 946
|
| 6656 |
+
},
|
| 6657 |
+
{
|
| 6658 |
+
"epoch": 0.42860375650599686,
|
| 6659 |
+
"grad_norm": 0.514545202255249,
|
| 6660 |
+
"learning_rate": 6.890683565088198e-05,
|
| 6661 |
+
"loss": 0.8755,
|
| 6662 |
+
"step": 947
|
| 6663 |
+
},
|
| 6664 |
+
{
|
| 6665 |
+
"epoch": 0.4290563475899525,
|
| 6666 |
+
"grad_norm": 0.43919333815574646,
|
| 6667 |
+
"learning_rate": 6.87154968393277e-05,
|
| 6668 |
+
"loss": 0.6828,
|
| 6669 |
+
"step": 948
|
| 6670 |
+
},
|
| 6671 |
+
{
|
| 6672 |
+
"epoch": 0.4295089386739081,
|
| 6673 |
+
"grad_norm": 0.6306238770484924,
|
| 6674 |
+
"learning_rate": 6.852428490382773e-05,
|
| 6675 |
+
"loss": 1.099,
|
| 6676 |
+
"step": 949
|
| 6677 |
+
},
|
| 6678 |
+
{
|
| 6679 |
+
"epoch": 0.4299615297578638,
|
| 6680 |
+
"grad_norm": 0.7785803079605103,
|
| 6681 |
+
"learning_rate": 6.833320061985277e-05,
|
| 6682 |
+
"loss": 0.7355,
|
| 6683 |
+
"step": 950
|
| 6684 |
+
},
|
| 6685 |
+
{
|
| 6686 |
+
"epoch": 0.4304141208418194,
|
| 6687 |
+
"grad_norm": 0.4950277805328369,
|
| 6688 |
+
"learning_rate": 6.814224476235581e-05,
|
| 6689 |
+
"loss": 0.6854,
|
| 6690 |
+
"step": 951
|
| 6691 |
+
},
|
| 6692 |
+
{
|
| 6693 |
+
"epoch": 0.4308667119257751,
|
| 6694 |
+
"grad_norm": 0.6018353700637817,
|
| 6695 |
+
"learning_rate": 6.795141810576906e-05,
|
| 6696 |
+
"loss": 1.2355,
|
| 6697 |
+
"step": 952
|
| 6698 |
+
},
|
| 6699 |
+
{
|
| 6700 |
+
"epoch": 0.4313193030097307,
|
| 6701 |
+
"grad_norm": 0.4305374026298523,
|
| 6702 |
+
"learning_rate": 6.776072142400067e-05,
|
| 6703 |
+
"loss": 0.6234,
|
| 6704 |
+
"step": 953
|
| 6705 |
+
},
|
| 6706 |
+
{
|
| 6707 |
+
"epoch": 0.4317718940936864,
|
| 6708 |
+
"grad_norm": 0.39642637968063354,
|
| 6709 |
+
"learning_rate": 6.757015549043175e-05,
|
| 6710 |
+
"loss": 0.4528,
|
| 6711 |
+
"step": 954
|
| 6712 |
+
},
|
| 6713 |
+
{
|
| 6714 |
+
"epoch": 0.432224485177642,
|
| 6715 |
+
"grad_norm": 0.44513657689094543,
|
| 6716 |
+
"learning_rate": 6.73797210779131e-05,
|
| 6717 |
+
"loss": 0.6681,
|
| 6718 |
+
"step": 955
|
| 6719 |
+
},
|
| 6720 |
+
{
|
| 6721 |
+
"epoch": 0.4326770762615976,
|
| 6722 |
+
"grad_norm": 0.47976213693618774,
|
| 6723 |
+
"learning_rate": 6.718941895876212e-05,
|
| 6724 |
+
"loss": 0.6925,
|
| 6725 |
+
"step": 956
|
| 6726 |
+
},
|
| 6727 |
+
{
|
| 6728 |
+
"epoch": 0.4331296673455533,
|
| 6729 |
+
"grad_norm": 0.4550023078918457,
|
| 6730 |
+
"learning_rate": 6.699924990475974e-05,
|
| 6731 |
+
"loss": 0.6719,
|
| 6732 |
+
"step": 957
|
| 6733 |
+
},
|
| 6734 |
+
{
|
| 6735 |
+
"epoch": 0.4335822584295089,
|
| 6736 |
+
"grad_norm": 0.4960489273071289,
|
| 6737 |
+
"learning_rate": 6.680921468714719e-05,
|
| 6738 |
+
"loss": 0.9812,
|
| 6739 |
+
"step": 958
|
| 6740 |
+
},
|
| 6741 |
+
{
|
| 6742 |
+
"epoch": 0.4340348495134646,
|
| 6743 |
+
"grad_norm": 0.4192532002925873,
|
| 6744 |
+
"learning_rate": 6.661931407662292e-05,
|
| 6745 |
+
"loss": 0.5652,
|
| 6746 |
+
"step": 959
|
| 6747 |
+
},
|
| 6748 |
+
{
|
| 6749 |
+
"epoch": 0.4344874405974202,
|
| 6750 |
+
"grad_norm": 0.7804063558578491,
|
| 6751 |
+
"learning_rate": 6.642954884333955e-05,
|
| 6752 |
+
"loss": 0.9131,
|
| 6753 |
+
"step": 960
|
| 6754 |
+
},
|
| 6755 |
+
{
|
| 6756 |
+
"epoch": 0.43494003168137585,
|
| 6757 |
+
"grad_norm": 0.5083318948745728,
|
| 6758 |
+
"learning_rate": 6.623991975690051e-05,
|
| 6759 |
+
"loss": 0.7885,
|
| 6760 |
+
"step": 961
|
| 6761 |
+
},
|
| 6762 |
+
{
|
| 6763 |
+
"epoch": 0.4353926227653315,
|
| 6764 |
+
"grad_norm": 0.4479636549949646,
|
| 6765 |
+
"learning_rate": 6.605042758635729e-05,
|
| 6766 |
+
"loss": 0.7053,
|
| 6767 |
+
"step": 962
|
| 6768 |
+
},
|
| 6769 |
+
{
|
| 6770 |
+
"epoch": 0.43584521384928715,
|
| 6771 |
+
"grad_norm": 0.5218529105186462,
|
| 6772 |
+
"learning_rate": 6.586107310020591e-05,
|
| 6773 |
+
"loss": 0.8669,
|
| 6774 |
+
"step": 963
|
| 6775 |
+
},
|
| 6776 |
+
{
|
| 6777 |
+
"epoch": 0.4362978049332428,
|
| 6778 |
+
"grad_norm": 0.42930254340171814,
|
| 6779 |
+
"learning_rate": 6.567185706638417e-05,
|
| 6780 |
+
"loss": 0.6542,
|
| 6781 |
+
"step": 964
|
| 6782 |
+
},
|
| 6783 |
+
{
|
| 6784 |
+
"epoch": 0.43675039601719845,
|
| 6785 |
+
"grad_norm": 0.4791978895664215,
|
| 6786 |
+
"learning_rate": 6.548278025226821e-05,
|
| 6787 |
+
"loss": 0.8341,
|
| 6788 |
+
"step": 965
|
| 6789 |
+
},
|
| 6790 |
+
{
|
| 6791 |
+
"epoch": 0.43720298710115413,
|
| 6792 |
+
"grad_norm": 0.5348832011222839,
|
| 6793 |
+
"learning_rate": 6.52938434246697e-05,
|
| 6794 |
+
"loss": 0.8451,
|
| 6795 |
+
"step": 966
|
| 6796 |
+
},
|
| 6797 |
+
{
|
| 6798 |
+
"epoch": 0.43765557818510975,
|
| 6799 |
+
"grad_norm": 0.662418007850647,
|
| 6800 |
+
"learning_rate": 6.510504734983249e-05,
|
| 6801 |
+
"loss": 0.8109,
|
| 6802 |
+
"step": 967
|
| 6803 |
+
},
|
| 6804 |
+
{
|
| 6805 |
+
"epoch": 0.4381081692690654,
|
| 6806 |
+
"grad_norm": 0.48103776574134827,
|
| 6807 |
+
"learning_rate": 6.491639279342963e-05,
|
| 6808 |
+
"loss": 0.8619,
|
| 6809 |
+
"step": 968
|
| 6810 |
+
},
|
| 6811 |
+
{
|
| 6812 |
+
"epoch": 0.43856076035302105,
|
| 6813 |
+
"grad_norm": 0.5524132251739502,
|
| 6814 |
+
"learning_rate": 6.472788052056027e-05,
|
| 6815 |
+
"loss": 0.8855,
|
| 6816 |
+
"step": 969
|
| 6817 |
+
},
|
| 6818 |
+
{
|
| 6819 |
+
"epoch": 0.4390133514369767,
|
| 6820 |
+
"grad_norm": 0.5164328217506409,
|
| 6821 |
+
"learning_rate": 6.453951129574644e-05,
|
| 6822 |
+
"loss": 0.7278,
|
| 6823 |
+
"step": 970
|
| 6824 |
+
},
|
| 6825 |
+
{
|
| 6826 |
+
"epoch": 0.43946594252093235,
|
| 6827 |
+
"grad_norm": 0.5808908343315125,
|
| 6828 |
+
"learning_rate": 6.435128588293012e-05,
|
| 6829 |
+
"loss": 0.8667,
|
| 6830 |
+
"step": 971
|
| 6831 |
+
},
|
| 6832 |
+
{
|
| 6833 |
+
"epoch": 0.439918533604888,
|
| 6834 |
+
"grad_norm": 0.5309447646141052,
|
| 6835 |
+
"learning_rate": 6.416320504546997e-05,
|
| 6836 |
+
"loss": 0.8923,
|
| 6837 |
+
"step": 972
|
| 6838 |
+
},
|
| 6839 |
+
{
|
| 6840 |
+
"epoch": 0.44037112468884365,
|
| 6841 |
+
"grad_norm": 0.4193963408470154,
|
| 6842 |
+
"learning_rate": 6.397526954613839e-05,
|
| 6843 |
+
"loss": 0.7471,
|
| 6844 |
+
"step": 973
|
| 6845 |
+
},
|
| 6846 |
+
{
|
| 6847 |
+
"epoch": 0.4408237157727993,
|
| 6848 |
+
"grad_norm": 0.4619600772857666,
|
| 6849 |
+
"learning_rate": 6.378748014711834e-05,
|
| 6850 |
+
"loss": 0.7625,
|
| 6851 |
+
"step": 974
|
| 6852 |
+
},
|
| 6853 |
+
{
|
| 6854 |
+
"epoch": 0.4412763068567549,
|
| 6855 |
+
"grad_norm": 0.4345484673976898,
|
| 6856 |
+
"learning_rate": 6.359983761000018e-05,
|
| 6857 |
+
"loss": 0.6369,
|
| 6858 |
+
"step": 975
|
| 6859 |
+
},
|
| 6860 |
+
{
|
| 6861 |
+
"epoch": 0.4417288979407106,
|
| 6862 |
+
"grad_norm": 0.4431121051311493,
|
| 6863 |
+
"learning_rate": 6.341234269577879e-05,
|
| 6864 |
+
"loss": 0.804,
|
| 6865 |
+
"step": 976
|
| 6866 |
+
},
|
| 6867 |
+
{
|
| 6868 |
+
"epoch": 0.4421814890246662,
|
| 6869 |
+
"grad_norm": 0.49312248826026917,
|
| 6870 |
+
"learning_rate": 6.322499616485025e-05,
|
| 6871 |
+
"loss": 0.8464,
|
| 6872 |
+
"step": 977
|
| 6873 |
+
},
|
| 6874 |
+
{
|
| 6875 |
+
"epoch": 0.4426340801086219,
|
| 6876 |
+
"grad_norm": 0.4761488139629364,
|
| 6877 |
+
"learning_rate": 6.30377987770089e-05,
|
| 6878 |
+
"loss": 0.7871,
|
| 6879 |
+
"step": 978
|
| 6880 |
+
},
|
| 6881 |
+
{
|
| 6882 |
+
"epoch": 0.4430866711925775,
|
| 6883 |
+
"grad_norm": 0.48850172758102417,
|
| 6884 |
+
"learning_rate": 6.285075129144428e-05,
|
| 6885 |
+
"loss": 0.8349,
|
| 6886 |
+
"step": 979
|
| 6887 |
+
},
|
| 6888 |
+
{
|
| 6889 |
+
"epoch": 0.4435392622765332,
|
| 6890 |
+
"grad_norm": 0.4868282973766327,
|
| 6891 |
+
"learning_rate": 6.26638544667379e-05,
|
| 6892 |
+
"loss": 0.777,
|
| 6893 |
+
"step": 980
|
| 6894 |
+
},
|
| 6895 |
+
{
|
| 6896 |
+
"epoch": 0.4439918533604888,
|
| 6897 |
+
"grad_norm": 0.5236530303955078,
|
| 6898 |
+
"learning_rate": 6.247710906086031e-05,
|
| 6899 |
+
"loss": 0.793,
|
| 6900 |
+
"step": 981
|
| 6901 |
+
},
|
| 6902 |
+
{
|
| 6903 |
+
"epoch": 0.4444444444444444,
|
| 6904 |
+
"grad_norm": 0.6053972840309143,
|
| 6905 |
+
"learning_rate": 6.229051583116796e-05,
|
| 6906 |
+
"loss": 1.0982,
|
| 6907 |
+
"step": 982
|
| 6908 |
+
},
|
| 6909 |
+
{
|
| 6910 |
+
"epoch": 0.4448970355284001,
|
| 6911 |
+
"grad_norm": 0.4751902222633362,
|
| 6912 |
+
"learning_rate": 6.210407553440015e-05,
|
| 6913 |
+
"loss": 0.7435,
|
| 6914 |
+
"step": 983
|
| 6915 |
+
},
|
| 6916 |
+
{
|
| 6917 |
+
"epoch": 0.4453496266123557,
|
| 6918 |
+
"grad_norm": 0.610129177570343,
|
| 6919 |
+
"learning_rate": 6.191778892667591e-05,
|
| 6920 |
+
"loss": 1.1366,
|
| 6921 |
+
"step": 984
|
| 6922 |
+
},
|
| 6923 |
+
{
|
| 6924 |
+
"epoch": 0.4458022176963114,
|
| 6925 |
+
"grad_norm": 0.5146445631980896,
|
| 6926 |
+
"learning_rate": 6.173165676349103e-05,
|
| 6927 |
+
"loss": 0.8541,
|
| 6928 |
+
"step": 985
|
| 6929 |
+
},
|
| 6930 |
+
{
|
| 6931 |
+
"epoch": 0.446254808780267,
|
| 6932 |
+
"grad_norm": 0.42905983328819275,
|
| 6933 |
+
"learning_rate": 6.154567979971493e-05,
|
| 6934 |
+
"loss": 0.6065,
|
| 6935 |
+
"step": 986
|
| 6936 |
+
},
|
| 6937 |
+
{
|
| 6938 |
+
"epoch": 0.4467073998642227,
|
| 6939 |
+
"grad_norm": 0.4927644431591034,
|
| 6940 |
+
"learning_rate": 6.135985878958759e-05,
|
| 6941 |
+
"loss": 0.7915,
|
| 6942 |
+
"step": 987
|
| 6943 |
+
},
|
| 6944 |
+
{
|
| 6945 |
+
"epoch": 0.4471599909481783,
|
| 6946 |
+
"grad_norm": 0.5202013850212097,
|
| 6947 |
+
"learning_rate": 6.117419448671651e-05,
|
| 6948 |
+
"loss": 0.7805,
|
| 6949 |
+
"step": 988
|
| 6950 |
+
},
|
| 6951 |
+
{
|
| 6952 |
+
"epoch": 0.44761258203213394,
|
| 6953 |
+
"grad_norm": 0.45878079533576965,
|
| 6954 |
+
"learning_rate": 6.098868764407371e-05,
|
| 6955 |
+
"loss": 0.8074,
|
| 6956 |
+
"step": 989
|
| 6957 |
+
},
|
| 6958 |
+
{
|
| 6959 |
+
"epoch": 0.4480651731160896,
|
| 6960 |
+
"grad_norm": 0.5383991599082947,
|
| 6961 |
+
"learning_rate": 6.080333901399251e-05,
|
| 6962 |
+
"loss": 0.7669,
|
| 6963 |
+
"step": 990
|
| 6964 |
+
},
|
| 6965 |
+
{
|
| 6966 |
+
"epoch": 0.44851776420004524,
|
| 6967 |
+
"grad_norm": 0.47985830903053284,
|
| 6968 |
+
"learning_rate": 6.0618149348164696e-05,
|
| 6969 |
+
"loss": 0.7841,
|
| 6970 |
+
"step": 991
|
| 6971 |
+
},
|
| 6972 |
+
{
|
| 6973 |
+
"epoch": 0.4489703552840009,
|
| 6974 |
+
"grad_norm": 0.49758774042129517,
|
| 6975 |
+
"learning_rate": 6.043311939763734e-05,
|
| 6976 |
+
"loss": 0.7702,
|
| 6977 |
+
"step": 992
|
| 6978 |
+
},
|
| 6979 |
+
{
|
| 6980 |
+
"epoch": 0.44942294636795654,
|
| 6981 |
+
"grad_norm": 0.582724392414093,
|
| 6982 |
+
"learning_rate": 6.024824991280974e-05,
|
| 6983 |
+
"loss": 0.9283,
|
| 6984 |
+
"step": 993
|
| 6985 |
+
},
|
| 6986 |
+
{
|
| 6987 |
+
"epoch": 0.4498755374519122,
|
| 6988 |
+
"grad_norm": 0.5484330058097839,
|
| 6989 |
+
"learning_rate": 6.006354164343046e-05,
|
| 6990 |
+
"loss": 0.9831,
|
| 6991 |
+
"step": 994
|
| 6992 |
+
},
|
| 6993 |
+
{
|
| 6994 |
+
"epoch": 0.45032812853586784,
|
| 6995 |
+
"grad_norm": 0.5154725909233093,
|
| 6996 |
+
"learning_rate": 5.9878995338594224e-05,
|
| 6997 |
+
"loss": 0.8368,
|
| 6998 |
+
"step": 995
|
| 6999 |
+
},
|
| 7000 |
+
{
|
| 7001 |
+
"epoch": 0.45078071961982347,
|
| 7002 |
+
"grad_norm": 0.5858908891677856,
|
| 7003 |
+
"learning_rate": 5.969461174673889e-05,
|
| 7004 |
+
"loss": 1.2287,
|
| 7005 |
+
"step": 996
|
| 7006 |
+
},
|
| 7007 |
+
{
|
| 7008 |
+
"epoch": 0.45123331070377914,
|
| 7009 |
+
"grad_norm": 0.46006911993026733,
|
| 7010 |
+
"learning_rate": 5.9510391615642466e-05,
|
| 7011 |
+
"loss": 0.8285,
|
| 7012 |
+
"step": 997
|
| 7013 |
+
},
|
| 7014 |
+
{
|
| 7015 |
+
"epoch": 0.45168590178773477,
|
| 7016 |
+
"grad_norm": 0.5323997139930725,
|
| 7017 |
+
"learning_rate": 5.9326335692419995e-05,
|
| 7018 |
+
"loss": 1.0234,
|
| 7019 |
+
"step": 998
|
| 7020 |
+
},
|
| 7021 |
+
{
|
| 7022 |
+
"epoch": 0.45213849287169044,
|
| 7023 |
+
"grad_norm": 0.5194817781448364,
|
| 7024 |
+
"learning_rate": 5.914244472352057e-05,
|
| 7025 |
+
"loss": 0.8194,
|
| 7026 |
+
"step": 999
|
| 7027 |
+
},
|
| 7028 |
+
{
|
| 7029 |
+
"epoch": 0.45259108395564607,
|
| 7030 |
+
"grad_norm": 0.5339561700820923,
|
| 7031 |
+
"learning_rate": 5.8958719454724346e-05,
|
| 7032 |
+
"loss": 0.9298,
|
| 7033 |
+
"step": 1000
|
| 7034 |
+
},
|
| 7035 |
+
{
|
| 7036 |
+
"epoch": 0.45304367503960175,
|
| 7037 |
+
"grad_norm": 0.5592015981674194,
|
| 7038 |
+
"learning_rate": 5.877516063113939e-05,
|
| 7039 |
+
"loss": 1.0499,
|
| 7040 |
+
"step": 1001
|
| 7041 |
+
},
|
| 7042 |
+
{
|
| 7043 |
+
"epoch": 0.45349626612355737,
|
| 7044 |
+
"grad_norm": 0.476650208234787,
|
| 7045 |
+
"learning_rate": 5.859176899719883e-05,
|
| 7046 |
+
"loss": 0.7622,
|
| 7047 |
+
"step": 1002
|
| 7048 |
+
},
|
| 7049 |
+
{
|
| 7050 |
+
"epoch": 0.453948857207513,
|
| 7051 |
+
"grad_norm": 0.5177839398384094,
|
| 7052 |
+
"learning_rate": 5.840854529665767e-05,
|
| 7053 |
+
"loss": 0.7409,
|
| 7054 |
+
"step": 1003
|
| 7055 |
+
},
|
| 7056 |
+
{
|
| 7057 |
+
"epoch": 0.45440144829146867,
|
| 7058 |
+
"grad_norm": 0.48446404933929443,
|
| 7059 |
+
"learning_rate": 5.8225490272589933e-05,
|
| 7060 |
+
"loss": 0.7528,
|
| 7061 |
+
"step": 1004
|
| 7062 |
+
},
|
| 7063 |
+
{
|
| 7064 |
+
"epoch": 0.4548540393754243,
|
| 7065 |
+
"grad_norm": 0.5130894780158997,
|
| 7066 |
+
"learning_rate": 5.804260466738547e-05,
|
| 7067 |
+
"loss": 0.6405,
|
| 7068 |
+
"step": 1005
|
| 7069 |
+
},
|
| 7070 |
+
{
|
| 7071 |
+
"epoch": 0.45530663045937997,
|
| 7072 |
+
"grad_norm": 0.38971513509750366,
|
| 7073 |
+
"learning_rate": 5.785988922274711e-05,
|
| 7074 |
+
"loss": 0.4938,
|
| 7075 |
+
"step": 1006
|
| 7076 |
+
},
|
| 7077 |
+
{
|
| 7078 |
+
"epoch": 0.4557592215433356,
|
| 7079 |
+
"grad_norm": 0.5229670405387878,
|
| 7080 |
+
"learning_rate": 5.767734467968757e-05,
|
| 7081 |
+
"loss": 0.8747,
|
| 7082 |
+
"step": 1007
|
| 7083 |
+
},
|
| 7084 |
+
{
|
| 7085 |
+
"epoch": 0.45621181262729127,
|
| 7086 |
+
"grad_norm": 0.4669683873653412,
|
| 7087 |
+
"learning_rate": 5.7494971778526486e-05,
|
| 7088 |
+
"loss": 0.7369,
|
| 7089 |
+
"step": 1008
|
| 7090 |
+
},
|
| 7091 |
+
{
|
| 7092 |
+
"epoch": 0.4566644037112469,
|
| 7093 |
+
"grad_norm": 0.5144645571708679,
|
| 7094 |
+
"learning_rate": 5.7312771258887386e-05,
|
| 7095 |
+
"loss": 0.8298,
|
| 7096 |
+
"step": 1009
|
| 7097 |
+
},
|
| 7098 |
+
{
|
| 7099 |
+
"epoch": 0.4571169947952025,
|
| 7100 |
+
"grad_norm": 0.47048383951187134,
|
| 7101 |
+
"learning_rate": 5.713074385969457e-05,
|
| 7102 |
+
"loss": 0.6594,
|
| 7103 |
+
"step": 1010
|
| 7104 |
+
},
|
| 7105 |
+
{
|
| 7106 |
+
"epoch": 0.4575695858791582,
|
| 7107 |
+
"grad_norm": 0.5354034304618835,
|
| 7108 |
+
"learning_rate": 5.694889031917047e-05,
|
| 7109 |
+
"loss": 0.8264,
|
| 7110 |
+
"step": 1011
|
| 7111 |
+
},
|
| 7112 |
+
{
|
| 7113 |
+
"epoch": 0.4580221769631138,
|
| 7114 |
+
"grad_norm": 0.5005225539207458,
|
| 7115 |
+
"learning_rate": 5.676721137483225e-05,
|
| 7116 |
+
"loss": 0.7441,
|
| 7117 |
+
"step": 1012
|
| 7118 |
+
},
|
| 7119 |
+
{
|
| 7120 |
+
"epoch": 0.4584747680470695,
|
| 7121 |
+
"grad_norm": 0.5038537979125977,
|
| 7122 |
+
"learning_rate": 5.658570776348911e-05,
|
| 7123 |
+
"loss": 0.8668,
|
| 7124 |
+
"step": 1013
|
| 7125 |
+
},
|
| 7126 |
+
{
|
| 7127 |
+
"epoch": 0.4589273591310251,
|
| 7128 |
+
"grad_norm": 0.4663868546485901,
|
| 7129 |
+
"learning_rate": 5.6404380221238985e-05,
|
| 7130 |
+
"loss": 0.7556,
|
| 7131 |
+
"step": 1014
|
| 7132 |
+
},
|
| 7133 |
+
{
|
| 7134 |
+
"epoch": 0.4593799502149808,
|
| 7135 |
+
"grad_norm": 0.5414367318153381,
|
| 7136 |
+
"learning_rate": 5.622322948346594e-05,
|
| 7137 |
+
"loss": 0.9308,
|
| 7138 |
+
"step": 1015
|
| 7139 |
+
},
|
| 7140 |
+
{
|
| 7141 |
+
"epoch": 0.4598325412989364,
|
| 7142 |
+
"grad_norm": 0.49297547340393066,
|
| 7143 |
+
"learning_rate": 5.604225628483695e-05,
|
| 7144 |
+
"loss": 0.9007,
|
| 7145 |
+
"step": 1016
|
| 7146 |
+
},
|
| 7147 |
+
{
|
| 7148 |
+
"epoch": 0.46028513238289204,
|
| 7149 |
+
"grad_norm": 0.4712825417518616,
|
| 7150 |
+
"learning_rate": 5.586146135929893e-05,
|
| 7151 |
+
"loss": 0.7742,
|
| 7152 |
+
"step": 1017
|
| 7153 |
+
},
|
| 7154 |
+
{
|
| 7155 |
+
"epoch": 0.4607377234668477,
|
| 7156 |
+
"grad_norm": 0.46004435420036316,
|
| 7157 |
+
"learning_rate": 5.568084544007588e-05,
|
| 7158 |
+
"loss": 0.5906,
|
| 7159 |
+
"step": 1018
|
| 7160 |
+
},
|
| 7161 |
+
{
|
| 7162 |
+
"epoch": 0.46119031455080334,
|
| 7163 |
+
"grad_norm": 0.48832663893699646,
|
| 7164 |
+
"learning_rate": 5.550040925966569e-05,
|
| 7165 |
+
"loss": 0.8536,
|
| 7166 |
+
"step": 1019
|
| 7167 |
+
},
|
| 7168 |
+
{
|
| 7169 |
+
"epoch": 0.461642905634759,
|
| 7170 |
+
"grad_norm": 0.5183213949203491,
|
| 7171 |
+
"learning_rate": 5.5320153549837415e-05,
|
| 7172 |
+
"loss": 0.8216,
|
| 7173 |
+
"step": 1020
|
| 7174 |
+
},
|
| 7175 |
+
{
|
| 7176 |
+
"epoch": 0.46209549671871464,
|
| 7177 |
+
"grad_norm": 0.5739022493362427,
|
| 7178 |
+
"learning_rate": 5.5140079041628214e-05,
|
| 7179 |
+
"loss": 0.9626,
|
| 7180 |
+
"step": 1021
|
| 7181 |
+
},
|
| 7182 |
+
{
|
| 7183 |
+
"epoch": 0.46254808780267026,
|
| 7184 |
+
"grad_norm": 0.5354647636413574,
|
| 7185 |
+
"learning_rate": 5.4960186465340316e-05,
|
| 7186 |
+
"loss": 0.952,
|
| 7187 |
+
"step": 1022
|
| 7188 |
+
},
|
| 7189 |
+
{
|
| 7190 |
+
"epoch": 0.46300067888662594,
|
| 7191 |
+
"grad_norm": 0.5027756094932556,
|
| 7192 |
+
"learning_rate": 5.4780476550538174e-05,
|
| 7193 |
+
"loss": 0.6244,
|
| 7194 |
+
"step": 1023
|
| 7195 |
+
},
|
| 7196 |
+
{
|
| 7197 |
+
"epoch": 0.46345326997058156,
|
| 7198 |
+
"grad_norm": 0.4999634623527527,
|
| 7199 |
+
"learning_rate": 5.4600950026045326e-05,
|
| 7200 |
+
"loss": 0.6141,
|
| 7201 |
+
"step": 1024
|
| 7202 |
+
},
|
| 7203 |
+
{
|
| 7204 |
+
"epoch": 0.46390586105453724,
|
| 7205 |
+
"grad_norm": 0.5566645860671997,
|
| 7206 |
+
"learning_rate": 5.4421607619941706e-05,
|
| 7207 |
+
"loss": 1.0411,
|
| 7208 |
+
"step": 1025
|
| 7209 |
+
},
|
| 7210 |
+
{
|
| 7211 |
+
"epoch": 0.46435845213849286,
|
| 7212 |
+
"grad_norm": 0.5371447801589966,
|
| 7213 |
+
"learning_rate": 5.424245005956048e-05,
|
| 7214 |
+
"loss": 0.932,
|
| 7215 |
+
"step": 1026
|
| 7216 |
+
},
|
| 7217 |
+
{
|
| 7218 |
+
"epoch": 0.46481104322244854,
|
| 7219 |
+
"grad_norm": 0.546616792678833,
|
| 7220 |
+
"learning_rate": 5.406347807148515e-05,
|
| 7221 |
+
"loss": 0.8634,
|
| 7222 |
+
"step": 1027
|
| 7223 |
+
},
|
| 7224 |
+
{
|
| 7225 |
+
"epoch": 0.46526363430640416,
|
| 7226 |
+
"grad_norm": 0.5544886589050293,
|
| 7227 |
+
"learning_rate": 5.3884692381546676e-05,
|
| 7228 |
+
"loss": 0.9725,
|
| 7229 |
+
"step": 1028
|
| 7230 |
+
},
|
| 7231 |
+
{
|
| 7232 |
+
"epoch": 0.4657162253903598,
|
| 7233 |
+
"grad_norm": 0.4882948398590088,
|
| 7234 |
+
"learning_rate": 5.370609371482036e-05,
|
| 7235 |
+
"loss": 0.8355,
|
| 7236 |
+
"step": 1029
|
| 7237 |
+
},
|
| 7238 |
+
{
|
| 7239 |
+
"epoch": 0.46616881647431546,
|
| 7240 |
+
"grad_norm": 0.4552709758281708,
|
| 7241 |
+
"learning_rate": 5.3527682795623146e-05,
|
| 7242 |
+
"loss": 0.6487,
|
| 7243 |
+
"step": 1030
|
| 7244 |
+
},
|
| 7245 |
+
{
|
| 7246 |
+
"epoch": 0.4666214075582711,
|
| 7247 |
+
"grad_norm": 0.46513643860816956,
|
| 7248 |
+
"learning_rate": 5.334946034751049e-05,
|
| 7249 |
+
"loss": 0.7491,
|
| 7250 |
+
"step": 1031
|
| 7251 |
+
},
|
| 7252 |
+
{
|
| 7253 |
+
"epoch": 0.46707399864222676,
|
| 7254 |
+
"grad_norm": 0.5479960441589355,
|
| 7255 |
+
"learning_rate": 5.317142709327354e-05,
|
| 7256 |
+
"loss": 1.0504,
|
| 7257 |
+
"step": 1032
|
| 7258 |
+
},
|
| 7259 |
+
{
|
| 7260 |
+
"epoch": 0.4675265897261824,
|
| 7261 |
+
"grad_norm": 0.5057997703552246,
|
| 7262 |
+
"learning_rate": 5.2993583754936126e-05,
|
| 7263 |
+
"loss": 0.9572,
|
| 7264 |
+
"step": 1033
|
| 7265 |
+
},
|
| 7266 |
+
{
|
| 7267 |
+
"epoch": 0.46797918081013806,
|
| 7268 |
+
"grad_norm": 0.4689159691333771,
|
| 7269 |
+
"learning_rate": 5.28159310537518e-05,
|
| 7270 |
+
"loss": 0.6862,
|
| 7271 |
+
"step": 1034
|
| 7272 |
+
},
|
| 7273 |
+
{
|
| 7274 |
+
"epoch": 0.4684317718940937,
|
| 7275 |
+
"grad_norm": 0.4798116385936737,
|
| 7276 |
+
"learning_rate": 5.263846971020108e-05,
|
| 7277 |
+
"loss": 0.698,
|
| 7278 |
+
"step": 1035
|
| 7279 |
+
},
|
| 7280 |
+
{
|
| 7281 |
+
"epoch": 0.4688843629780493,
|
| 7282 |
+
"grad_norm": 0.5007181167602539,
|
| 7283 |
+
"learning_rate": 5.246120044398839e-05,
|
| 7284 |
+
"loss": 0.811,
|
| 7285 |
+
"step": 1036
|
| 7286 |
+
},
|
| 7287 |
+
{
|
| 7288 |
+
"epoch": 0.469336954062005,
|
| 7289 |
+
"grad_norm": 0.46686723828315735,
|
| 7290 |
+
"learning_rate": 5.2284123974039154e-05,
|
| 7291 |
+
"loss": 0.7168,
|
| 7292 |
+
"step": 1037
|
| 7293 |
+
},
|
| 7294 |
+
{
|
| 7295 |
+
"epoch": 0.4697895451459606,
|
| 7296 |
+
"grad_norm": 0.4993205666542053,
|
| 7297 |
+
"learning_rate": 5.210724101849696e-05,
|
| 7298 |
+
"loss": 0.8641,
|
| 7299 |
+
"step": 1038
|
| 7300 |
+
},
|
| 7301 |
+
{
|
| 7302 |
+
"epoch": 0.4702421362299163,
|
| 7303 |
+
"grad_norm": 0.45969900488853455,
|
| 7304 |
+
"learning_rate": 5.193055229472045e-05,
|
| 7305 |
+
"loss": 0.6912,
|
| 7306 |
+
"step": 1039
|
| 7307 |
+
},
|
| 7308 |
+
{
|
| 7309 |
+
"epoch": 0.4706947273138719,
|
| 7310 |
+
"grad_norm": 0.570006251335144,
|
| 7311 |
+
"learning_rate": 5.175405851928068e-05,
|
| 7312 |
+
"loss": 1.0181,
|
| 7313 |
+
"step": 1040
|
| 7314 |
+
},
|
| 7315 |
+
{
|
| 7316 |
+
"epoch": 0.4711473183978276,
|
| 7317 |
+
"grad_norm": 0.5396348237991333,
|
| 7318 |
+
"learning_rate": 5.157776040795804e-05,
|
| 7319 |
+
"loss": 0.8275,
|
| 7320 |
+
"step": 1041
|
| 7321 |
+
},
|
| 7322 |
+
{
|
| 7323 |
+
"epoch": 0.4715999094817832,
|
| 7324 |
+
"grad_norm": 0.5312735438346863,
|
| 7325 |
+
"learning_rate": 5.14016586757394e-05,
|
| 7326 |
+
"loss": 0.8963,
|
| 7327 |
+
"step": 1042
|
| 7328 |
+
},
|
| 7329 |
+
{
|
| 7330 |
+
"epoch": 0.47205250056573883,
|
| 7331 |
+
"grad_norm": 0.5774843096733093,
|
| 7332 |
+
"learning_rate": 5.122575403681521e-05,
|
| 7333 |
+
"loss": 0.9631,
|
| 7334 |
+
"step": 1043
|
| 7335 |
+
},
|
| 7336 |
+
{
|
| 7337 |
+
"epoch": 0.4725050916496945,
|
| 7338 |
+
"grad_norm": 0.5320578813552856,
|
| 7339 |
+
"learning_rate": 5.105004720457653e-05,
|
| 7340 |
+
"loss": 0.9338,
|
| 7341 |
+
"step": 1044
|
| 7342 |
+
},
|
| 7343 |
+
{
|
| 7344 |
+
"epoch": 0.47295768273365013,
|
| 7345 |
+
"grad_norm": 0.45046183466911316,
|
| 7346 |
+
"learning_rate": 5.087453889161229e-05,
|
| 7347 |
+
"loss": 0.7609,
|
| 7348 |
+
"step": 1045
|
| 7349 |
+
},
|
| 7350 |
+
{
|
| 7351 |
+
"epoch": 0.4734102738176058,
|
| 7352 |
+
"grad_norm": 0.5660046935081482,
|
| 7353 |
+
"learning_rate": 5.069922980970626e-05,
|
| 7354 |
+
"loss": 0.9304,
|
| 7355 |
+
"step": 1046
|
| 7356 |
+
},
|
| 7357 |
+
{
|
| 7358 |
+
"epoch": 0.47386286490156143,
|
| 7359 |
+
"grad_norm": 0.45797666907310486,
|
| 7360 |
+
"learning_rate": 5.052412066983427e-05,
|
| 7361 |
+
"loss": 0.7153,
|
| 7362 |
+
"step": 1047
|
| 7363 |
+
},
|
| 7364 |
+
{
|
| 7365 |
+
"epoch": 0.4743154559855171,
|
| 7366 |
+
"grad_norm": 0.49813908338546753,
|
| 7367 |
+
"learning_rate": 5.0349212182161254e-05,
|
| 7368 |
+
"loss": 0.9467,
|
| 7369 |
+
"step": 1048
|
| 7370 |
+
},
|
| 7371 |
+
{
|
| 7372 |
+
"epoch": 0.47476804706947273,
|
| 7373 |
+
"grad_norm": 0.42716631293296814,
|
| 7374 |
+
"learning_rate": 5.017450505603831e-05,
|
| 7375 |
+
"loss": 0.6052,
|
| 7376 |
+
"step": 1049
|
| 7377 |
+
},
|
| 7378 |
+
{
|
| 7379 |
+
"epoch": 0.47522063815342835,
|
| 7380 |
+
"grad_norm": 0.48885422945022583,
|
| 7381 |
+
"learning_rate": 5.000000000000002e-05,
|
| 7382 |
+
"loss": 0.8695,
|
| 7383 |
+
"step": 1050
|
| 7384 |
+
},
|
| 7385 |
+
{
|
| 7386 |
+
"epoch": 0.47567322923738403,
|
| 7387 |
+
"grad_norm": 0.46352043747901917,
|
| 7388 |
+
"learning_rate": 4.98256977217614e-05,
|
| 7389 |
+
"loss": 0.7086,
|
| 7390 |
+
"step": 1051
|
| 7391 |
+
},
|
| 7392 |
+
{
|
| 7393 |
+
"epoch": 0.47612582032133965,
|
| 7394 |
+
"grad_norm": 0.5039620995521545,
|
| 7395 |
+
"learning_rate": 4.965159892821509e-05,
|
| 7396 |
+
"loss": 0.6385,
|
| 7397 |
+
"step": 1052
|
| 7398 |
+
},
|
| 7399 |
+
{
|
| 7400 |
+
"epoch": 0.47657841140529533,
|
| 7401 |
+
"grad_norm": 0.5147011876106262,
|
| 7402 |
+
"learning_rate": 4.9477704325428554e-05,
|
| 7403 |
+
"loss": 0.8561,
|
| 7404 |
+
"step": 1053
|
| 7405 |
+
},
|
| 7406 |
+
{
|
| 7407 |
+
"epoch": 0.47703100248925095,
|
| 7408 |
+
"grad_norm": 0.46321332454681396,
|
| 7409 |
+
"learning_rate": 4.9304014618640995e-05,
|
| 7410 |
+
"loss": 0.8561,
|
| 7411 |
+
"step": 1054
|
| 7412 |
+
},
|
| 7413 |
+
{
|
| 7414 |
+
"epoch": 0.47748359357320663,
|
| 7415 |
+
"grad_norm": 0.5085314512252808,
|
| 7416 |
+
"learning_rate": 4.913053051226079e-05,
|
| 7417 |
+
"loss": 0.7139,
|
| 7418 |
+
"step": 1055
|
| 7419 |
+
},
|
| 7420 |
+
{
|
| 7421 |
+
"epoch": 0.47793618465716226,
|
| 7422 |
+
"grad_norm": 0.5991241335868835,
|
| 7423 |
+
"learning_rate": 4.895725270986244e-05,
|
| 7424 |
+
"loss": 1.1444,
|
| 7425 |
+
"step": 1056
|
| 7426 |
+
},
|
| 7427 |
+
{
|
| 7428 |
+
"epoch": 0.4783887757411179,
|
| 7429 |
+
"grad_norm": 0.4854184687137604,
|
| 7430 |
+
"learning_rate": 4.87841819141838e-05,
|
| 7431 |
+
"loss": 0.7903,
|
| 7432 |
+
"step": 1057
|
| 7433 |
+
},
|
| 7434 |
+
{
|
| 7435 |
+
"epoch": 0.47884136682507356,
|
| 7436 |
+
"grad_norm": 0.5258088111877441,
|
| 7437 |
+
"learning_rate": 4.861131882712314e-05,
|
| 7438 |
+
"loss": 0.9007,
|
| 7439 |
+
"step": 1058
|
| 7440 |
+
},
|
| 7441 |
+
{
|
| 7442 |
+
"epoch": 0.4792939579090292,
|
| 7443 |
+
"grad_norm": 0.4994451105594635,
|
| 7444 |
+
"learning_rate": 4.843866414973641e-05,
|
| 7445 |
+
"loss": 0.7204,
|
| 7446 |
+
"step": 1059
|
| 7447 |
+
},
|
| 7448 |
+
{
|
| 7449 |
+
"epoch": 0.47974654899298486,
|
| 7450 |
+
"grad_norm": 0.483104944229126,
|
| 7451 |
+
"learning_rate": 4.826621858223431e-05,
|
| 7452 |
+
"loss": 0.8054,
|
| 7453 |
+
"step": 1060
|
| 7454 |
+
},
|
| 7455 |
+
{
|
| 7456 |
+
"epoch": 0.4801991400769405,
|
| 7457 |
+
"grad_norm": 0.5926142930984497,
|
| 7458 |
+
"learning_rate": 4.809398282397951e-05,
|
| 7459 |
+
"loss": 0.9783,
|
| 7460 |
+
"step": 1061
|
| 7461 |
+
},
|
| 7462 |
+
{
|
| 7463 |
+
"epoch": 0.48065173116089616,
|
| 7464 |
+
"grad_norm": 0.5814982056617737,
|
| 7465 |
+
"learning_rate": 4.7921957573483754e-05,
|
| 7466 |
+
"loss": 1.1615,
|
| 7467 |
+
"step": 1062
|
| 7468 |
+
},
|
| 7469 |
+
{
|
| 7470 |
+
"epoch": 0.4811043222448518,
|
| 7471 |
+
"grad_norm": 0.5472806692123413,
|
| 7472 |
+
"learning_rate": 4.7750143528405126e-05,
|
| 7473 |
+
"loss": 0.8907,
|
| 7474 |
+
"step": 1063
|
| 7475 |
+
},
|
| 7476 |
+
{
|
| 7477 |
+
"epoch": 0.4815569133288074,
|
| 7478 |
+
"grad_norm": 0.5430026054382324,
|
| 7479 |
+
"learning_rate": 4.7578541385545014e-05,
|
| 7480 |
+
"loss": 1.1033,
|
| 7481 |
+
"step": 1064
|
| 7482 |
+
},
|
| 7483 |
+
{
|
| 7484 |
+
"epoch": 0.4820095044127631,
|
| 7485 |
+
"grad_norm": 0.4955959916114807,
|
| 7486 |
+
"learning_rate": 4.740715184084559e-05,
|
| 7487 |
+
"loss": 0.8295,
|
| 7488 |
+
"step": 1065
|
| 7489 |
+
},
|
| 7490 |
+
{
|
| 7491 |
+
"epoch": 0.4824620954967187,
|
| 7492 |
+
"grad_norm": 0.44894957542419434,
|
| 7493 |
+
"learning_rate": 4.723597558938672e-05,
|
| 7494 |
+
"loss": 0.6721,
|
| 7495 |
+
"step": 1066
|
| 7496 |
+
},
|
| 7497 |
+
{
|
| 7498 |
+
"epoch": 0.4829146865806744,
|
| 7499 |
+
"grad_norm": 0.4289432764053345,
|
| 7500 |
+
"learning_rate": 4.7065013325383275e-05,
|
| 7501 |
+
"loss": 0.6993,
|
| 7502 |
+
"step": 1067
|
| 7503 |
+
},
|
| 7504 |
+
{
|
| 7505 |
+
"epoch": 0.48336727766463,
|
| 7506 |
+
"grad_norm": 0.5182718634605408,
|
| 7507 |
+
"learning_rate": 4.6894265742182344e-05,
|
| 7508 |
+
"loss": 0.6838,
|
| 7509 |
+
"step": 1068
|
| 7510 |
+
},
|
| 7511 |
+
{
|
| 7512 |
+
"epoch": 0.4838198687485857,
|
| 7513 |
+
"grad_norm": 0.6270569562911987,
|
| 7514 |
+
"learning_rate": 4.672373353226023e-05,
|
| 7515 |
+
"loss": 1.0023,
|
| 7516 |
+
"step": 1069
|
| 7517 |
+
},
|
| 7518 |
+
{
|
| 7519 |
+
"epoch": 0.4842724598325413,
|
| 7520 |
+
"grad_norm": 0.4682745039463043,
|
| 7521 |
+
"learning_rate": 4.6553417387219886e-05,
|
| 7522 |
+
"loss": 0.7835,
|
| 7523 |
+
"step": 1070
|
| 7524 |
+
},
|
| 7525 |
+
{
|
| 7526 |
+
"epoch": 0.4847250509164969,
|
| 7527 |
+
"grad_norm": 0.5611541271209717,
|
| 7528 |
+
"learning_rate": 4.6383317997787986e-05,
|
| 7529 |
+
"loss": 0.8775,
|
| 7530 |
+
"step": 1071
|
| 7531 |
+
},
|
| 7532 |
+
{
|
| 7533 |
+
"epoch": 0.4851776420004526,
|
| 7534 |
+
"grad_norm": 0.5803195238113403,
|
| 7535 |
+
"learning_rate": 4.6213436053812144e-05,
|
| 7536 |
+
"loss": 1.2508,
|
| 7537 |
+
"step": 1072
|
| 7538 |
+
},
|
| 7539 |
+
{
|
| 7540 |
+
"epoch": 0.4856302330844082,
|
| 7541 |
+
"grad_norm": 0.4491642117500305,
|
| 7542 |
+
"learning_rate": 4.6043772244258096e-05,
|
| 7543 |
+
"loss": 0.7083,
|
| 7544 |
+
"step": 1073
|
| 7545 |
+
},
|
| 7546 |
+
{
|
| 7547 |
+
"epoch": 0.4860828241683639,
|
| 7548 |
+
"grad_norm": 0.5523327589035034,
|
| 7549 |
+
"learning_rate": 4.587432725720687e-05,
|
| 7550 |
+
"loss": 0.9192,
|
| 7551 |
+
"step": 1074
|
| 7552 |
+
},
|
| 7553 |
+
{
|
| 7554 |
+
"epoch": 0.4865354152523195,
|
| 7555 |
+
"grad_norm": 0.5153692960739136,
|
| 7556 |
+
"learning_rate": 4.5705101779852135e-05,
|
| 7557 |
+
"loss": 0.8575,
|
| 7558 |
+
"step": 1075
|
| 7559 |
+
},
|
| 7560 |
+
{
|
| 7561 |
+
"epoch": 0.48698800633627515,
|
| 7562 |
+
"grad_norm": 0.5204266309738159,
|
| 7563 |
+
"learning_rate": 4.5536096498497295e-05,
|
| 7564 |
+
"loss": 0.7705,
|
| 7565 |
+
"step": 1076
|
| 7566 |
+
},
|
| 7567 |
+
{
|
| 7568 |
+
"epoch": 0.4874405974202308,
|
| 7569 |
+
"grad_norm": 0.5776071548461914,
|
| 7570 |
+
"learning_rate": 4.5367312098552705e-05,
|
| 7571 |
+
"loss": 1.0659,
|
| 7572 |
+
"step": 1077
|
| 7573 |
+
},
|
| 7574 |
+
{
|
| 7575 |
+
"epoch": 0.48789318850418645,
|
| 7576 |
+
"grad_norm": 0.4952472448348999,
|
| 7577 |
+
"learning_rate": 4.519874926453302e-05,
|
| 7578 |
+
"loss": 0.8679,
|
| 7579 |
+
"step": 1078
|
| 7580 |
+
},
|
| 7581 |
+
{
|
| 7582 |
+
"epoch": 0.4883457795881421,
|
| 7583 |
+
"grad_norm": 0.4542064964771271,
|
| 7584 |
+
"learning_rate": 4.503040868005416e-05,
|
| 7585 |
+
"loss": 0.796,
|
| 7586 |
+
"step": 1079
|
| 7587 |
+
},
|
| 7588 |
+
{
|
| 7589 |
+
"epoch": 0.48879837067209775,
|
| 7590 |
+
"grad_norm": 0.5350983142852783,
|
| 7591 |
+
"learning_rate": 4.486229102783084e-05,
|
| 7592 |
+
"loss": 0.9615,
|
| 7593 |
+
"step": 1080
|
| 7594 |
+
},
|
| 7595 |
+
{
|
| 7596 |
+
"epoch": 0.4892509617560534,
|
| 7597 |
+
"grad_norm": 0.6435551047325134,
|
| 7598 |
+
"learning_rate": 4.469439698967359e-05,
|
| 7599 |
+
"loss": 1.4427,
|
| 7600 |
+
"step": 1081
|
| 7601 |
+
},
|
| 7602 |
+
{
|
| 7603 |
+
"epoch": 0.48970355284000905,
|
| 7604 |
+
"grad_norm": 0.5093064308166504,
|
| 7605 |
+
"learning_rate": 4.452672724648611e-05,
|
| 7606 |
+
"loss": 0.9063,
|
| 7607 |
+
"step": 1082
|
| 7608 |
+
},
|
| 7609 |
+
{
|
| 7610 |
+
"epoch": 0.49015614392396467,
|
| 7611 |
+
"grad_norm": 0.5286267399787903,
|
| 7612 |
+
"learning_rate": 4.4359282478262454e-05,
|
| 7613 |
+
"loss": 0.7893,
|
| 7614 |
+
"step": 1083
|
| 7615 |
+
},
|
| 7616 |
+
{
|
| 7617 |
+
"epoch": 0.49060873500792035,
|
| 7618 |
+
"grad_norm": 0.46403610706329346,
|
| 7619 |
+
"learning_rate": 4.419206336408418e-05,
|
| 7620 |
+
"loss": 0.6207,
|
| 7621 |
+
"step": 1084
|
| 7622 |
+
},
|
| 7623 |
+
{
|
| 7624 |
+
"epoch": 0.49106132609187597,
|
| 7625 |
+
"grad_norm": 0.5149431824684143,
|
| 7626 |
+
"learning_rate": 4.40250705821178e-05,
|
| 7627 |
+
"loss": 0.8015,
|
| 7628 |
+
"step": 1085
|
| 7629 |
+
},
|
| 7630 |
+
{
|
| 7631 |
+
"epoch": 0.49151391717583165,
|
| 7632 |
+
"grad_norm": 0.543387234210968,
|
| 7633 |
+
"learning_rate": 4.385830480961192e-05,
|
| 7634 |
+
"loss": 0.9494,
|
| 7635 |
+
"step": 1086
|
| 7636 |
+
},
|
| 7637 |
+
{
|
| 7638 |
+
"epoch": 0.49196650825978727,
|
| 7639 |
+
"grad_norm": 0.5150254964828491,
|
| 7640 |
+
"learning_rate": 4.3691766722894435e-05,
|
| 7641 |
+
"loss": 0.8872,
|
| 7642 |
+
"step": 1087
|
| 7643 |
+
},
|
| 7644 |
+
{
|
| 7645 |
+
"epoch": 0.49241909934374295,
|
| 7646 |
+
"grad_norm": 0.48482635617256165,
|
| 7647 |
+
"learning_rate": 4.3525456997369926e-05,
|
| 7648 |
+
"loss": 0.5804,
|
| 7649 |
+
"step": 1088
|
| 7650 |
+
},
|
| 7651 |
+
{
|
| 7652 |
+
"epoch": 0.49287169042769857,
|
| 7653 |
+
"grad_norm": 0.5497539639472961,
|
| 7654 |
+
"learning_rate": 4.335937630751674e-05,
|
| 7655 |
+
"loss": 1.0462,
|
| 7656 |
+
"step": 1089
|
| 7657 |
+
},
|
| 7658 |
+
{
|
| 7659 |
+
"epoch": 0.4933242815116542,
|
| 7660 |
+
"grad_norm": 0.4731244742870331,
|
| 7661 |
+
"learning_rate": 4.3193525326884435e-05,
|
| 7662 |
+
"loss": 0.7626,
|
| 7663 |
+
"step": 1090
|
| 7664 |
+
},
|
| 7665 |
+
{
|
| 7666 |
+
"epoch": 0.4937768725956099,
|
| 7667 |
+
"grad_norm": 0.5599950551986694,
|
| 7668 |
+
"learning_rate": 4.3027904728090954e-05,
|
| 7669 |
+
"loss": 0.893,
|
| 7670 |
+
"step": 1091
|
| 7671 |
+
},
|
| 7672 |
+
{
|
| 7673 |
+
"epoch": 0.4942294636795655,
|
| 7674 |
+
"grad_norm": 0.5236995220184326,
|
| 7675 |
+
"learning_rate": 4.2862515182819904e-05,
|
| 7676 |
+
"loss": 0.9606,
|
| 7677 |
+
"step": 1092
|
| 7678 |
+
},
|
| 7679 |
+
{
|
| 7680 |
+
"epoch": 0.4946820547635212,
|
| 7681 |
+
"grad_norm": 0.6051554679870605,
|
| 7682 |
+
"learning_rate": 4.26973573618179e-05,
|
| 7683 |
+
"loss": 0.9776,
|
| 7684 |
+
"step": 1093
|
| 7685 |
+
},
|
| 7686 |
+
{
|
| 7687 |
+
"epoch": 0.4951346458474768,
|
| 7688 |
+
"grad_norm": 0.550380527973175,
|
| 7689 |
+
"learning_rate": 4.253243193489165e-05,
|
| 7690 |
+
"loss": 0.8356,
|
| 7691 |
+
"step": 1094
|
| 7692 |
+
},
|
| 7693 |
+
{
|
| 7694 |
+
"epoch": 0.4955872369314325,
|
| 7695 |
+
"grad_norm": 0.4601718783378601,
|
| 7696 |
+
"learning_rate": 4.236773957090548e-05,
|
| 7697 |
+
"loss": 0.75,
|
| 7698 |
+
"step": 1095
|
| 7699 |
+
},
|
| 7700 |
+
{
|
| 7701 |
+
"epoch": 0.4960398280153881,
|
| 7702 |
+
"grad_norm": 0.5237348675727844,
|
| 7703 |
+
"learning_rate": 4.220328093777851e-05,
|
| 7704 |
+
"loss": 0.8505,
|
| 7705 |
+
"step": 1096
|
| 7706 |
+
},
|
| 7707 |
+
{
|
| 7708 |
+
"epoch": 0.4964924190993437,
|
| 7709 |
+
"grad_norm": 0.5931401252746582,
|
| 7710 |
+
"learning_rate": 4.203905670248194e-05,
|
| 7711 |
+
"loss": 0.8769,
|
| 7712 |
+
"step": 1097
|
| 7713 |
+
},
|
| 7714 |
+
{
|
| 7715 |
+
"epoch": 0.4969450101832994,
|
| 7716 |
+
"grad_norm": 0.5065420269966125,
|
| 7717 |
+
"learning_rate": 4.1875067531036374e-05,
|
| 7718 |
+
"loss": 0.9189,
|
| 7719 |
+
"step": 1098
|
| 7720 |
+
},
|
| 7721 |
+
{
|
| 7722 |
+
"epoch": 0.497397601267255,
|
| 7723 |
+
"grad_norm": 0.49355390667915344,
|
| 7724 |
+
"learning_rate": 4.1711314088509e-05,
|
| 7725 |
+
"loss": 0.7092,
|
| 7726 |
+
"step": 1099
|
| 7727 |
+
},
|
| 7728 |
+
{
|
| 7729 |
+
"epoch": 0.4978501923512107,
|
| 7730 |
+
"grad_norm": 0.5571166276931763,
|
| 7731 |
+
"learning_rate": 4.154779703901114e-05,
|
| 7732 |
+
"loss": 0.9334,
|
| 7733 |
+
"step": 1100
|
| 7734 |
+
},
|
| 7735 |
+
{
|
| 7736 |
+
"epoch": 0.4983027834351663,
|
| 7737 |
+
"grad_norm": 0.4783550500869751,
|
| 7738 |
+
"learning_rate": 4.1384517045695316e-05,
|
| 7739 |
+
"loss": 0.8249,
|
| 7740 |
+
"step": 1101
|
| 7741 |
+
},
|
| 7742 |
+
{
|
| 7743 |
+
"epoch": 0.498755374519122,
|
| 7744 |
+
"grad_norm": 0.42690062522888184,
|
| 7745 |
+
"learning_rate": 4.12214747707527e-05,
|
| 7746 |
+
"loss": 0.5945,
|
| 7747 |
+
"step": 1102
|
| 7748 |
+
},
|
| 7749 |
+
{
|
| 7750 |
+
"epoch": 0.4992079656030776,
|
| 7751 |
+
"grad_norm": 0.5278857946395874,
|
| 7752 |
+
"learning_rate": 4.1058670875410386e-05,
|
| 7753 |
+
"loss": 0.9126,
|
| 7754 |
+
"step": 1103
|
| 7755 |
+
},
|
| 7756 |
+
{
|
| 7757 |
+
"epoch": 0.49966055668703324,
|
| 7758 |
+
"grad_norm": 0.5173430442810059,
|
| 7759 |
+
"learning_rate": 4.089610601992864e-05,
|
| 7760 |
+
"loss": 1.0533,
|
| 7761 |
+
"step": 1104
|
| 7762 |
+
},
|
| 7763 |
+
{
|
| 7764 |
+
"epoch": 0.5001131477709889,
|
| 7765 |
+
"grad_norm": 0.6061631441116333,
|
| 7766 |
+
"learning_rate": 4.0733780863598335e-05,
|
| 7767 |
+
"loss": 1.1095,
|
| 7768 |
+
"step": 1105
|
| 7769 |
+
},
|
| 7770 |
+
{
|
| 7771 |
+
"epoch": 0.5005657388549446,
|
| 7772 |
+
"grad_norm": 0.5131955742835999,
|
| 7773 |
+
"learning_rate": 4.057169606473827e-05,
|
| 7774 |
+
"loss": 0.765,
|
| 7775 |
+
"step": 1106
|
| 7776 |
+
},
|
| 7777 |
+
{
|
| 7778 |
+
"epoch": 0.5010183299389002,
|
| 7779 |
+
"grad_norm": 0.6769198775291443,
|
| 7780 |
+
"learning_rate": 4.04098522806924e-05,
|
| 7781 |
+
"loss": 0.5315,
|
| 7782 |
+
"step": 1107
|
| 7783 |
+
},
|
| 7784 |
+
{
|
| 7785 |
+
"epoch": 0.5014709210228558,
|
| 7786 |
+
"grad_norm": 0.5840115547180176,
|
| 7787 |
+
"learning_rate": 4.0248250167827275e-05,
|
| 7788 |
+
"loss": 0.8269,
|
| 7789 |
+
"step": 1108
|
| 7790 |
+
},
|
| 7791 |
+
{
|
| 7792 |
+
"epoch": 0.5019235121068115,
|
| 7793 |
+
"grad_norm": 0.5163032412528992,
|
| 7794 |
+
"learning_rate": 4.00868903815293e-05,
|
| 7795 |
+
"loss": 0.9032,
|
| 7796 |
+
"step": 1109
|
| 7797 |
+
},
|
| 7798 |
+
{
|
| 7799 |
+
"epoch": 0.5023761031907671,
|
| 7800 |
+
"grad_norm": 0.5210195779800415,
|
| 7801 |
+
"learning_rate": 3.99257735762021e-05,
|
| 7802 |
+
"loss": 0.6677,
|
| 7803 |
+
"step": 1110
|
| 7804 |
+
},
|
| 7805 |
+
{
|
| 7806 |
+
"epoch": 0.5028286942747228,
|
| 7807 |
+
"grad_norm": 0.4478912651538849,
|
| 7808 |
+
"learning_rate": 3.976490040526394e-05,
|
| 7809 |
+
"loss": 0.7087,
|
| 7810 |
+
"step": 1111
|
| 7811 |
+
},
|
| 7812 |
+
{
|
| 7813 |
+
"epoch": 0.5032812853586784,
|
| 7814 |
+
"grad_norm": 0.6642077565193176,
|
| 7815 |
+
"learning_rate": 3.960427152114494e-05,
|
| 7816 |
+
"loss": 1.0827,
|
| 7817 |
+
"step": 1112
|
| 7818 |
+
},
|
| 7819 |
+
{
|
| 7820 |
+
"epoch": 0.5037338764426341,
|
| 7821 |
+
"grad_norm": 0.5077605247497559,
|
| 7822 |
+
"learning_rate": 3.9443887575284586e-05,
|
| 7823 |
+
"loss": 0.8169,
|
| 7824 |
+
"step": 1113
|
| 7825 |
+
},
|
| 7826 |
+
{
|
| 7827 |
+
"epoch": 0.5041864675265897,
|
| 7828 |
+
"grad_norm": 0.5410700440406799,
|
| 7829 |
+
"learning_rate": 3.9283749218128885e-05,
|
| 7830 |
+
"loss": 0.929,
|
| 7831 |
+
"step": 1114
|
| 7832 |
+
},
|
| 7833 |
+
{
|
| 7834 |
+
"epoch": 0.5046390586105454,
|
| 7835 |
+
"grad_norm": 0.5579005479812622,
|
| 7836 |
+
"learning_rate": 3.9123857099127936e-05,
|
| 7837 |
+
"loss": 0.8948,
|
| 7838 |
+
"step": 1115
|
| 7839 |
+
},
|
| 7840 |
+
{
|
| 7841 |
+
"epoch": 0.505091649694501,
|
| 7842 |
+
"grad_norm": 0.44208788871765137,
|
| 7843 |
+
"learning_rate": 3.896421186673318e-05,
|
| 7844 |
+
"loss": 0.7895,
|
| 7845 |
+
"step": 1116
|
| 7846 |
+
},
|
| 7847 |
+
{
|
| 7848 |
+
"epoch": 0.5055442407784566,
|
| 7849 |
+
"grad_norm": 0.6037198305130005,
|
| 7850 |
+
"learning_rate": 3.88048141683948e-05,
|
| 7851 |
+
"loss": 0.9484,
|
| 7852 |
+
"step": 1117
|
| 7853 |
+
},
|
| 7854 |
+
{
|
| 7855 |
+
"epoch": 0.5059968318624123,
|
| 7856 |
+
"grad_norm": 0.5426963567733765,
|
| 7857 |
+
"learning_rate": 3.864566465055912e-05,
|
| 7858 |
+
"loss": 0.8515,
|
| 7859 |
+
"step": 1118
|
| 7860 |
+
},
|
| 7861 |
+
{
|
| 7862 |
+
"epoch": 0.506449422946368,
|
| 7863 |
+
"grad_norm": 0.5471535921096802,
|
| 7864 |
+
"learning_rate": 3.848676395866586e-05,
|
| 7865 |
+
"loss": 0.952,
|
| 7866 |
+
"step": 1119
|
| 7867 |
+
},
|
| 7868 |
+
{
|
| 7869 |
+
"epoch": 0.5069020140303236,
|
| 7870 |
+
"grad_norm": 0.5362775921821594,
|
| 7871 |
+
"learning_rate": 3.832811273714569e-05,
|
| 7872 |
+
"loss": 0.8603,
|
| 7873 |
+
"step": 1120
|
| 7874 |
+
},
|
| 7875 |
+
{
|
| 7876 |
+
"epoch": 0.5073546051142792,
|
| 7877 |
+
"grad_norm": 0.47162777185440063,
|
| 7878 |
+
"learning_rate": 3.816971162941755e-05,
|
| 7879 |
+
"loss": 0.737,
|
| 7880 |
+
"step": 1121
|
| 7881 |
+
},
|
| 7882 |
+
{
|
| 7883 |
+
"epoch": 0.5078071961982349,
|
| 7884 |
+
"grad_norm": 0.6005630493164062,
|
| 7885 |
+
"learning_rate": 3.8011561277885964e-05,
|
| 7886 |
+
"loss": 0.5545,
|
| 7887 |
+
"step": 1122
|
| 7888 |
+
},
|
| 7889 |
+
{
|
| 7890 |
+
"epoch": 0.5082597872821906,
|
| 7891 |
+
"grad_norm": 0.585284948348999,
|
| 7892 |
+
"learning_rate": 3.785366232393861e-05,
|
| 7893 |
+
"loss": 0.8955,
|
| 7894 |
+
"step": 1123
|
| 7895 |
+
},
|
| 7896 |
+
{
|
| 7897 |
+
"epoch": 0.5087123783661461,
|
| 7898 |
+
"grad_norm": 0.47206246852874756,
|
| 7899 |
+
"learning_rate": 3.769601540794344e-05,
|
| 7900 |
+
"loss": 0.4943,
|
| 7901 |
+
"step": 1124
|
| 7902 |
+
},
|
| 7903 |
+
{
|
| 7904 |
+
"epoch": 0.5091649694501018,
|
| 7905 |
+
"grad_norm": 0.5819693803787231,
|
| 7906 |
+
"learning_rate": 3.75386211692464e-05,
|
| 7907 |
+
"loss": 0.8945,
|
| 7908 |
+
"step": 1125
|
| 7909 |
+
},
|
| 7910 |
+
{
|
| 7911 |
+
"epoch": 0.5096175605340575,
|
| 7912 |
+
"grad_norm": 0.5495473146438599,
|
| 7913 |
+
"learning_rate": 3.738148024616863e-05,
|
| 7914 |
+
"loss": 1.0552,
|
| 7915 |
+
"step": 1126
|
| 7916 |
+
},
|
| 7917 |
+
{
|
| 7918 |
+
"epoch": 0.5100701516180132,
|
| 7919 |
+
"grad_norm": 0.5882697105407715,
|
| 7920 |
+
"learning_rate": 3.722459327600395e-05,
|
| 7921 |
+
"loss": 0.9964,
|
| 7922 |
+
"step": 1127
|
| 7923 |
+
},
|
| 7924 |
+
{
|
| 7925 |
+
"epoch": 0.5105227427019687,
|
| 7926 |
+
"grad_norm": 0.5525474548339844,
|
| 7927 |
+
"learning_rate": 3.7067960895016275e-05,
|
| 7928 |
+
"loss": 0.9653,
|
| 7929 |
+
"step": 1128
|
| 7930 |
+
},
|
| 7931 |
+
{
|
| 7932 |
+
"epoch": 0.5109753337859244,
|
| 7933 |
+
"grad_norm": 0.4547380208969116,
|
| 7934 |
+
"learning_rate": 3.691158373843694e-05,
|
| 7935 |
+
"loss": 0.6892,
|
| 7936 |
+
"step": 1129
|
| 7937 |
+
},
|
| 7938 |
+
{
|
| 7939 |
+
"epoch": 0.5114279248698801,
|
| 7940 |
+
"grad_norm": 0.5613864064216614,
|
| 7941 |
+
"learning_rate": 3.675546244046228e-05,
|
| 7942 |
+
"loss": 0.8805,
|
| 7943 |
+
"step": 1130
|
| 7944 |
+
},
|
| 7945 |
+
{
|
| 7946 |
+
"epoch": 0.5118805159538357,
|
| 7947 |
+
"grad_norm": 0.5161027908325195,
|
| 7948 |
+
"learning_rate": 3.659959763425098e-05,
|
| 7949 |
+
"loss": 0.8082,
|
| 7950 |
+
"step": 1131
|
| 7951 |
+
},
|
| 7952 |
+
{
|
| 7953 |
+
"epoch": 0.5123331070377913,
|
| 7954 |
+
"grad_norm": 0.550102710723877,
|
| 7955 |
+
"learning_rate": 3.644398995192147e-05,
|
| 7956 |
+
"loss": 0.865,
|
| 7957 |
+
"step": 1132
|
| 7958 |
+
},
|
| 7959 |
+
{
|
| 7960 |
+
"epoch": 0.512785698121747,
|
| 7961 |
+
"grad_norm": 0.562684178352356,
|
| 7962 |
+
"learning_rate": 3.628864002454947e-05,
|
| 7963 |
+
"loss": 0.9495,
|
| 7964 |
+
"step": 1133
|
| 7965 |
+
},
|
| 7966 |
+
{
|
| 7967 |
+
"epoch": 0.5132382892057027,
|
| 7968 |
+
"grad_norm": 0.5930166840553284,
|
| 7969 |
+
"learning_rate": 3.6133548482165225e-05,
|
| 7970 |
+
"loss": 1.1959,
|
| 7971 |
+
"step": 1134
|
| 7972 |
+
},
|
| 7973 |
+
{
|
| 7974 |
+
"epoch": 0.5136908802896583,
|
| 7975 |
+
"grad_norm": 0.5563511848449707,
|
| 7976 |
+
"learning_rate": 3.597871595375121e-05,
|
| 7977 |
+
"loss": 1.0088,
|
| 7978 |
+
"step": 1135
|
| 7979 |
+
},
|
| 7980 |
+
{
|
| 7981 |
+
"epoch": 0.5141434713736139,
|
| 7982 |
+
"grad_norm": 0.4959903061389923,
|
| 7983 |
+
"learning_rate": 3.582414306723941e-05,
|
| 7984 |
+
"loss": 0.8666,
|
| 7985 |
+
"step": 1136
|
| 7986 |
+
},
|
| 7987 |
+
{
|
| 7988 |
+
"epoch": 0.5145960624575696,
|
| 7989 |
+
"grad_norm": 0.5866261720657349,
|
| 7990 |
+
"learning_rate": 3.5669830449508836e-05,
|
| 7991 |
+
"loss": 0.9295,
|
| 7992 |
+
"step": 1137
|
| 7993 |
+
},
|
| 7994 |
+
{
|
| 7995 |
+
"epoch": 0.5150486535415252,
|
| 7996 |
+
"grad_norm": 0.46393883228302,
|
| 7997 |
+
"learning_rate": 3.5515778726382966e-05,
|
| 7998 |
+
"loss": 0.8025,
|
| 7999 |
+
"step": 1138
|
| 8000 |
+
},
|
| 8001 |
+
{
|
| 8002 |
+
"epoch": 0.5155012446254809,
|
| 8003 |
+
"grad_norm": 0.49082475900650024,
|
| 8004 |
+
"learning_rate": 3.536198852262713e-05,
|
| 8005 |
+
"loss": 0.9733,
|
| 8006 |
+
"step": 1139
|
| 8007 |
+
},
|
| 8008 |
+
{
|
| 8009 |
+
"epoch": 0.5159538357094365,
|
| 8010 |
+
"grad_norm": 0.5178489089012146,
|
| 8011 |
+
"learning_rate": 3.520846046194614e-05,
|
| 8012 |
+
"loss": 0.8229,
|
| 8013 |
+
"step": 1140
|
| 8014 |
+
},
|
| 8015 |
+
{
|
| 8016 |
+
"epoch": 0.5164064267933922,
|
| 8017 |
+
"grad_norm": 0.42519283294677734,
|
| 8018 |
+
"learning_rate": 3.5055195166981645e-05,
|
| 8019 |
+
"loss": 0.6869,
|
| 8020 |
+
"step": 1141
|
| 8021 |
+
},
|
| 8022 |
+
{
|
| 8023 |
+
"epoch": 0.5168590178773478,
|
| 8024 |
+
"grad_norm": 0.5158776044845581,
|
| 8025 |
+
"learning_rate": 3.490219325930962e-05,
|
| 8026 |
+
"loss": 0.8341,
|
| 8027 |
+
"step": 1142
|
| 8028 |
+
},
|
| 8029 |
+
{
|
| 8030 |
+
"epoch": 0.5173116089613035,
|
| 8031 |
+
"grad_norm": 0.6154707670211792,
|
| 8032 |
+
"learning_rate": 3.474945535943793e-05,
|
| 8033 |
+
"loss": 1.0029,
|
| 8034 |
+
"step": 1143
|
| 8035 |
+
},
|
| 8036 |
+
{
|
| 8037 |
+
"epoch": 0.5177642000452591,
|
| 8038 |
+
"grad_norm": 0.45592236518859863,
|
| 8039 |
+
"learning_rate": 3.459698208680359e-05,
|
| 8040 |
+
"loss": 0.7145,
|
| 8041 |
+
"step": 1144
|
| 8042 |
+
},
|
| 8043 |
+
{
|
| 8044 |
+
"epoch": 0.5182167911292147,
|
| 8045 |
+
"grad_norm": 0.546812117099762,
|
| 8046 |
+
"learning_rate": 3.4444774059770536e-05,
|
| 8047 |
+
"loss": 0.7955,
|
| 8048 |
+
"step": 1145
|
| 8049 |
+
},
|
| 8050 |
+
{
|
| 8051 |
+
"epoch": 0.5186693822131704,
|
| 8052 |
+
"grad_norm": 0.5020256638526917,
|
| 8053 |
+
"learning_rate": 3.429283189562694e-05,
|
| 8054 |
+
"loss": 0.8758,
|
| 8055 |
+
"step": 1146
|
| 8056 |
+
},
|
| 8057 |
+
{
|
| 8058 |
+
"epoch": 0.5191219732971261,
|
| 8059 |
+
"grad_norm": 0.45688191056251526,
|
| 8060 |
+
"learning_rate": 3.4141156210582756e-05,
|
| 8061 |
+
"loss": 0.7281,
|
| 8062 |
+
"step": 1147
|
| 8063 |
+
},
|
| 8064 |
+
{
|
| 8065 |
+
"epoch": 0.5195745643810817,
|
| 8066 |
+
"grad_norm": 0.5124973058700562,
|
| 8067 |
+
"learning_rate": 3.398974761976725e-05,
|
| 8068 |
+
"loss": 0.9177,
|
| 8069 |
+
"step": 1148
|
| 8070 |
+
},
|
| 8071 |
+
{
|
| 8072 |
+
"epoch": 0.5200271554650373,
|
| 8073 |
+
"grad_norm": 0.5134281516075134,
|
| 8074 |
+
"learning_rate": 3.383860673722639e-05,
|
| 8075 |
+
"loss": 0.6541,
|
| 8076 |
+
"step": 1149
|
| 8077 |
+
},
|
| 8078 |
+
{
|
| 8079 |
+
"epoch": 0.520479746548993,
|
| 8080 |
+
"grad_norm": 0.47039201855659485,
|
| 8081 |
+
"learning_rate": 3.36877341759205e-05,
|
| 8082 |
+
"loss": 0.6795,
|
| 8083 |
+
"step": 1150
|
| 8084 |
+
},
|
| 8085 |
+
{
|
| 8086 |
+
"epoch": 0.5209323376329487,
|
| 8087 |
+
"grad_norm": 0.41255295276641846,
|
| 8088 |
+
"learning_rate": 3.353713054772171e-05,
|
| 8089 |
+
"loss": 0.4722,
|
| 8090 |
+
"step": 1151
|
| 8091 |
+
},
|
| 8092 |
+
{
|
| 8093 |
+
"epoch": 0.5213849287169042,
|
| 8094 |
+
"grad_norm": 0.5607662796974182,
|
| 8095 |
+
"learning_rate": 3.338679646341146e-05,
|
| 8096 |
+
"loss": 0.7539,
|
| 8097 |
+
"step": 1152
|
| 8098 |
+
},
|
| 8099 |
+
{
|
| 8100 |
+
"epoch": 0.5218375198008599,
|
| 8101 |
+
"grad_norm": 0.475154310464859,
|
| 8102 |
+
"learning_rate": 3.3236732532678096e-05,
|
| 8103 |
+
"loss": 0.753,
|
| 8104 |
+
"step": 1153
|
| 8105 |
+
},
|
| 8106 |
+
{
|
| 8107 |
+
"epoch": 0.5222901108848156,
|
| 8108 |
+
"grad_norm": 0.5787057280540466,
|
| 8109 |
+
"learning_rate": 3.308693936411421e-05,
|
| 8110 |
+
"loss": 1.0778,
|
| 8111 |
+
"step": 1154
|
| 8112 |
+
},
|
| 8113 |
+
{
|
| 8114 |
+
"epoch": 0.5227427019687713,
|
| 8115 |
+
"grad_norm": 0.49256566166877747,
|
| 8116 |
+
"learning_rate": 3.293741756521442e-05,
|
| 8117 |
+
"loss": 0.7686,
|
| 8118 |
+
"step": 1155
|
| 8119 |
+
},
|
| 8120 |
+
{
|
| 8121 |
+
"epoch": 0.5231952930527268,
|
| 8122 |
+
"grad_norm": 0.5235592722892761,
|
| 8123 |
+
"learning_rate": 3.2788167742372725e-05,
|
| 8124 |
+
"loss": 1.0292,
|
| 8125 |
+
"step": 1156
|
| 8126 |
+
},
|
| 8127 |
+
{
|
| 8128 |
+
"epoch": 0.5236478841366825,
|
| 8129 |
+
"grad_norm": 0.4235650300979614,
|
| 8130 |
+
"learning_rate": 3.263919050088023e-05,
|
| 8131 |
+
"loss": 0.6254,
|
| 8132 |
+
"step": 1157
|
| 8133 |
+
},
|
| 8134 |
+
{
|
| 8135 |
+
"epoch": 0.5241004752206382,
|
| 8136 |
+
"grad_norm": 0.4604433476924896,
|
| 8137 |
+
"learning_rate": 3.24904864449224e-05,
|
| 8138 |
+
"loss": 0.7407,
|
| 8139 |
+
"step": 1158
|
| 8140 |
+
},
|
| 8141 |
+
{
|
| 8142 |
+
"epoch": 0.5245530663045938,
|
| 8143 |
+
"grad_norm": 0.41169530153274536,
|
| 8144 |
+
"learning_rate": 3.234205617757686e-05,
|
| 8145 |
+
"loss": 0.6558,
|
| 8146 |
+
"step": 1159
|
| 8147 |
+
},
|
| 8148 |
+
{
|
| 8149 |
+
"epoch": 0.5250056573885494,
|
| 8150 |
+
"grad_norm": 0.4996432065963745,
|
| 8151 |
+
"learning_rate": 3.219390030081091e-05,
|
| 8152 |
+
"loss": 0.7899,
|
| 8153 |
+
"step": 1160
|
| 8154 |
+
},
|
| 8155 |
+
{
|
| 8156 |
+
"epoch": 0.5254582484725051,
|
| 8157 |
+
"grad_norm": 0.4621078073978424,
|
| 8158 |
+
"learning_rate": 3.204601941547897e-05,
|
| 8159 |
+
"loss": 0.8085,
|
| 8160 |
+
"step": 1161
|
| 8161 |
+
},
|
| 8162 |
+
{
|
| 8163 |
+
"epoch": 0.5259108395564608,
|
| 8164 |
+
"grad_norm": 0.46135109663009644,
|
| 8165 |
+
"learning_rate": 3.1898414121320276e-05,
|
| 8166 |
+
"loss": 0.7147,
|
| 8167 |
+
"step": 1162
|
| 8168 |
+
},
|
| 8169 |
+
{
|
| 8170 |
+
"epoch": 0.5263634306404164,
|
| 8171 |
+
"grad_norm": 0.47625815868377686,
|
| 8172 |
+
"learning_rate": 3.1751085016956374e-05,
|
| 8173 |
+
"loss": 0.8159,
|
| 8174 |
+
"step": 1163
|
| 8175 |
+
},
|
| 8176 |
+
{
|
| 8177 |
+
"epoch": 0.526816021724372,
|
| 8178 |
+
"grad_norm": 0.5127959251403809,
|
| 8179 |
+
"learning_rate": 3.160403269988864e-05,
|
| 8180 |
+
"loss": 0.8973,
|
| 8181 |
+
"step": 1164
|
| 8182 |
+
},
|
| 8183 |
+
{
|
| 8184 |
+
"epoch": 0.5272686128083277,
|
| 8185 |
+
"grad_norm": 0.43038976192474365,
|
| 8186 |
+
"learning_rate": 3.1457257766496015e-05,
|
| 8187 |
+
"loss": 0.4513,
|
| 8188 |
+
"step": 1165
|
| 8189 |
+
},
|
| 8190 |
+
{
|
| 8191 |
+
"epoch": 0.5277212038922833,
|
| 8192 |
+
"grad_norm": 0.38239526748657227,
|
| 8193 |
+
"learning_rate": 3.131076081203247e-05,
|
| 8194 |
+
"loss": 0.589,
|
| 8195 |
+
"step": 1166
|
| 8196 |
+
},
|
| 8197 |
+
{
|
| 8198 |
+
"epoch": 0.528173794976239,
|
| 8199 |
+
"grad_norm": 0.40409696102142334,
|
| 8200 |
+
"learning_rate": 3.116454243062459e-05,
|
| 8201 |
+
"loss": 0.5848,
|
| 8202 |
+
"step": 1167
|
| 8203 |
+
},
|
| 8204 |
+
{
|
| 8205 |
+
"epoch": 0.5286263860601946,
|
| 8206 |
+
"grad_norm": 0.5500158667564392,
|
| 8207 |
+
"learning_rate": 3.101860321526924e-05,
|
| 8208 |
+
"loss": 0.9011,
|
| 8209 |
+
"step": 1168
|
| 8210 |
+
},
|
| 8211 |
+
{
|
| 8212 |
+
"epoch": 0.5290789771441503,
|
| 8213 |
+
"grad_norm": 0.5197808742523193,
|
| 8214 |
+
"learning_rate": 3.087294375783103e-05,
|
| 8215 |
+
"loss": 0.7307,
|
| 8216 |
+
"step": 1169
|
| 8217 |
+
},
|
| 8218 |
+
{
|
| 8219 |
+
"epoch": 0.5295315682281059,
|
| 8220 |
+
"grad_norm": 0.4519001841545105,
|
| 8221 |
+
"learning_rate": 3.072756464904006e-05,
|
| 8222 |
+
"loss": 0.6155,
|
| 8223 |
+
"step": 1170
|
| 8224 |
+
},
|
| 8225 |
+
{
|
| 8226 |
+
"epoch": 0.5299841593120616,
|
| 8227 |
+
"grad_norm": 0.5058773756027222,
|
| 8228 |
+
"learning_rate": 3.0582466478489455e-05,
|
| 8229 |
+
"loss": 0.8084,
|
| 8230 |
+
"step": 1171
|
| 8231 |
+
},
|
| 8232 |
+
{
|
| 8233 |
+
"epoch": 0.5304367503960172,
|
| 8234 |
+
"grad_norm": 0.5140249133110046,
|
| 8235 |
+
"learning_rate": 3.0437649834632977e-05,
|
| 8236 |
+
"loss": 0.8015,
|
| 8237 |
+
"step": 1172
|
| 8238 |
+
},
|
| 8239 |
+
{
|
| 8240 |
+
"epoch": 0.5308893414799728,
|
| 8241 |
+
"grad_norm": 0.4667006731033325,
|
| 8242 |
+
"learning_rate": 3.029311530478266e-05,
|
| 8243 |
+
"loss": 0.7651,
|
| 8244 |
+
"step": 1173
|
| 8245 |
+
},
|
| 8246 |
+
{
|
| 8247 |
+
"epoch": 0.5313419325639285,
|
| 8248 |
+
"grad_norm": 0.5502526164054871,
|
| 8249 |
+
"learning_rate": 3.0148863475106314e-05,
|
| 8250 |
+
"loss": 0.9117,
|
| 8251 |
+
"step": 1174
|
| 8252 |
+
},
|
| 8253 |
+
{
|
| 8254 |
+
"epoch": 0.5317945236478842,
|
| 8255 |
+
"grad_norm": 0.45357632637023926,
|
| 8256 |
+
"learning_rate": 3.000489493062535e-05,
|
| 8257 |
+
"loss": 0.6998,
|
| 8258 |
+
"step": 1175
|
| 8259 |
+
},
|
| 8260 |
+
{
|
| 8261 |
+
"epoch": 0.5322471147318398,
|
| 8262 |
+
"grad_norm": 0.5894111394882202,
|
| 8263 |
+
"learning_rate": 2.9861210255212245e-05,
|
| 8264 |
+
"loss": 0.88,
|
| 8265 |
+
"step": 1176
|
| 8266 |
+
},
|
| 8267 |
+
{
|
| 8268 |
+
"epoch": 0.5326997058157954,
|
| 8269 |
+
"grad_norm": 0.4162091314792633,
|
| 8270 |
+
"learning_rate": 2.9717810031588277e-05,
|
| 8271 |
+
"loss": 0.5883,
|
| 8272 |
+
"step": 1177
|
| 8273 |
+
},
|
| 8274 |
+
{
|
| 8275 |
+
"epoch": 0.5331522968997511,
|
| 8276 |
+
"grad_norm": 0.5021141171455383,
|
| 8277 |
+
"learning_rate": 2.9574694841321082e-05,
|
| 8278 |
+
"loss": 0.7265,
|
| 8279 |
+
"step": 1178
|
| 8280 |
+
},
|
| 8281 |
+
{
|
| 8282 |
+
"epoch": 0.5336048879837068,
|
| 8283 |
+
"grad_norm": 0.5346493721008301,
|
| 8284 |
+
"learning_rate": 2.943186526482229e-05,
|
| 8285 |
+
"loss": 0.84,
|
| 8286 |
+
"step": 1179
|
| 8287 |
+
},
|
| 8288 |
+
{
|
| 8289 |
+
"epoch": 0.5336048879837068,
|
| 8290 |
+
"eval_loss": 0.20086538791656494,
|
| 8291 |
+
"eval_runtime": 51.6164,
|
| 8292 |
+
"eval_samples_per_second": 18.037,
|
| 8293 |
+
"eval_steps_per_second": 9.028,
|
| 8294 |
+
"step": 1179
|
| 8295 |
}
|
| 8296 |
],
|
| 8297 |
"logging_steps": 1,
|
|
|
|
| 8311 |
"attributes": {}
|
| 8312 |
}
|
| 8313 |
},
|
| 8314 |
+
"total_flos": 2.150076298989404e+17,
|
| 8315 |
"train_batch_size": 2,
|
| 8316 |
"trial_name": null,
|
| 8317 |
"trial_params": null
|