Training in progress, step 1555, 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 51431872
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4ab74bb3f72388a5b307e11494e47e1b17188fbf5f88ff661a7b68ef7abc9306
|
| 3 |
size 51431872
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 26550644
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f865f38fc2299a9d4f4ed4259ad07d67093713fc963165e19bec963563430689
|
| 3 |
size 26550644
|
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 sha256:4004e1f96447bd8e09b1cfa50921da8e9f031903dc7b269172c5df49e147a2b3
|
| 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:be15ac73f67d359fd96b19cffe1b76da47c48d24be9f8cf124ada645a823a514
|
| 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": 389,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -8208,6 +8208,2722 @@
|
|
| 8208 |
"eval_samples_per_second": 41.406,
|
| 8209 |
"eval_steps_per_second": 20.703,
|
| 8210 |
"step": 1167
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 8211 |
}
|
| 8212 |
],
|
| 8213 |
"logging_steps": 1,
|
|
@@ -8222,12 +10938,12 @@
|
|
| 8222 |
"should_evaluate": false,
|
| 8223 |
"should_log": false,
|
| 8224 |
"should_save": true,
|
| 8225 |
-
"should_training_stop":
|
| 8226 |
},
|
| 8227 |
"attributes": {}
|
| 8228 |
}
|
| 8229 |
},
|
| 8230 |
-
"total_flos":
|
| 8231 |
"train_batch_size": 2,
|
| 8232 |
"trial_name": null,
|
| 8233 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.005167816962902219,
|
| 5 |
"eval_steps": 389,
|
| 6 |
+
"global_step": 1555,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 8208 |
"eval_samples_per_second": 41.406,
|
| 8209 |
"eval_steps_per_second": 20.703,
|
| 8210 |
"step": 1167
|
| 8211 |
+
},
|
| 8212 |
+
{
|
| 8213 |
+
"epoch": 0.0038816785933567795,
|
| 8214 |
+
"grad_norm": 3.088469982147217,
|
| 8215 |
+
"learning_rate": 2.9397241032157764e-05,
|
| 8216 |
+
"loss": 0.6386,
|
| 8217 |
+
"step": 1168
|
| 8218 |
+
},
|
| 8219 |
+
{
|
| 8220 |
+
"epoch": 0.0038850019483168455,
|
| 8221 |
+
"grad_norm": 4.252261638641357,
|
| 8222 |
+
"learning_rate": 2.9253385364510477e-05,
|
| 8223 |
+
"loss": 0.7193,
|
| 8224 |
+
"step": 1169
|
| 8225 |
+
},
|
| 8226 |
+
{
|
| 8227 |
+
"epoch": 0.003888325303276911,
|
| 8228 |
+
"grad_norm": 1.8697190284729004,
|
| 8229 |
+
"learning_rate": 2.9109822211964043e-05,
|
| 8230 |
+
"loss": 1.1484,
|
| 8231 |
+
"step": 1170
|
| 8232 |
+
},
|
| 8233 |
+
{
|
| 8234 |
+
"epoch": 0.003891648658236977,
|
| 8235 |
+
"grad_norm": 1.8480234146118164,
|
| 8236 |
+
"learning_rate": 2.896655216810713e-05,
|
| 8237 |
+
"loss": 0.8281,
|
| 8238 |
+
"step": 1171
|
| 8239 |
+
},
|
| 8240 |
+
{
|
| 8241 |
+
"epoch": 0.0038949720131970424,
|
| 8242 |
+
"grad_norm": 2.3371095657348633,
|
| 8243 |
+
"learning_rate": 2.8823575825316518e-05,
|
| 8244 |
+
"loss": 1.0603,
|
| 8245 |
+
"step": 1172
|
| 8246 |
+
},
|
| 8247 |
+
{
|
| 8248 |
+
"epoch": 0.003898295368157108,
|
| 8249 |
+
"grad_norm": 2.347442150115967,
|
| 8250 |
+
"learning_rate": 2.8680893774754612e-05,
|
| 8251 |
+
"loss": 0.9536,
|
| 8252 |
+
"step": 1173
|
| 8253 |
+
},
|
| 8254 |
+
{
|
| 8255 |
+
"epoch": 0.003901618723117174,
|
| 8256 |
+
"grad_norm": 1.7480069398880005,
|
| 8257 |
+
"learning_rate": 2.8538506606367032e-05,
|
| 8258 |
+
"loss": 0.9155,
|
| 8259 |
+
"step": 1174
|
| 8260 |
+
},
|
| 8261 |
+
{
|
| 8262 |
+
"epoch": 0.0039049420780772398,
|
| 8263 |
+
"grad_norm": 2.047581195831299,
|
| 8264 |
+
"learning_rate": 2.8396414908880098e-05,
|
| 8265 |
+
"loss": 1.1513,
|
| 8266 |
+
"step": 1175
|
| 8267 |
+
},
|
| 8268 |
+
{
|
| 8269 |
+
"epoch": 0.003908265433037306,
|
| 8270 |
+
"grad_norm": 1.8007320165634155,
|
| 8271 |
+
"learning_rate": 2.8254619269798555e-05,
|
| 8272 |
+
"loss": 0.7208,
|
| 8273 |
+
"step": 1176
|
| 8274 |
+
},
|
| 8275 |
+
{
|
| 8276 |
+
"epoch": 0.003911588787997371,
|
| 8277 |
+
"grad_norm": 1.9428542852401733,
|
| 8278 |
+
"learning_rate": 2.811312027540294e-05,
|
| 8279 |
+
"loss": 0.9256,
|
| 8280 |
+
"step": 1177
|
| 8281 |
+
},
|
| 8282 |
+
{
|
| 8283 |
+
"epoch": 0.003914912142957437,
|
| 8284 |
+
"grad_norm": 2.215702533721924,
|
| 8285 |
+
"learning_rate": 2.7971918510747286e-05,
|
| 8286 |
+
"loss": 0.7165,
|
| 8287 |
+
"step": 1178
|
| 8288 |
+
},
|
| 8289 |
+
{
|
| 8290 |
+
"epoch": 0.003918235497917502,
|
| 8291 |
+
"grad_norm": 2.1010520458221436,
|
| 8292 |
+
"learning_rate": 2.7831014559656698e-05,
|
| 8293 |
+
"loss": 0.919,
|
| 8294 |
+
"step": 1179
|
| 8295 |
+
},
|
| 8296 |
+
{
|
| 8297 |
+
"epoch": 0.003921558852877568,
|
| 8298 |
+
"grad_norm": 1.8713068962097168,
|
| 8299 |
+
"learning_rate": 2.769040900472488e-05,
|
| 8300 |
+
"loss": 0.9832,
|
| 8301 |
+
"step": 1180
|
| 8302 |
+
},
|
| 8303 |
+
{
|
| 8304 |
+
"epoch": 0.0039248822078376345,
|
| 8305 |
+
"grad_norm": 2.1189072132110596,
|
| 8306 |
+
"learning_rate": 2.7550102427311786e-05,
|
| 8307 |
+
"loss": 1.2858,
|
| 8308 |
+
"step": 1181
|
| 8309 |
+
},
|
| 8310 |
+
{
|
| 8311 |
+
"epoch": 0.0039282055627977,
|
| 8312 |
+
"grad_norm": 2.4399776458740234,
|
| 8313 |
+
"learning_rate": 2.741009540754118e-05,
|
| 8314 |
+
"loss": 0.8947,
|
| 8315 |
+
"step": 1182
|
| 8316 |
+
},
|
| 8317 |
+
{
|
| 8318 |
+
"epoch": 0.003931528917757766,
|
| 8319 |
+
"grad_norm": 2.8238000869750977,
|
| 8320 |
+
"learning_rate": 2.7270388524298262e-05,
|
| 8321 |
+
"loss": 0.8025,
|
| 8322 |
+
"step": 1183
|
| 8323 |
+
},
|
| 8324 |
+
{
|
| 8325 |
+
"epoch": 0.003934852272717832,
|
| 8326 |
+
"grad_norm": 2.2132227420806885,
|
| 8327 |
+
"learning_rate": 2.713098235522724e-05,
|
| 8328 |
+
"loss": 0.9481,
|
| 8329 |
+
"step": 1184
|
| 8330 |
+
},
|
| 8331 |
+
{
|
| 8332 |
+
"epoch": 0.003938175627677897,
|
| 8333 |
+
"grad_norm": 2.1194279193878174,
|
| 8334 |
+
"learning_rate": 2.699187747672899e-05,
|
| 8335 |
+
"loss": 0.8331,
|
| 8336 |
+
"step": 1185
|
| 8337 |
+
},
|
| 8338 |
+
{
|
| 8339 |
+
"epoch": 0.003941498982637963,
|
| 8340 |
+
"grad_norm": 3.092690944671631,
|
| 8341 |
+
"learning_rate": 2.6853074463958615e-05,
|
| 8342 |
+
"loss": 0.8494,
|
| 8343 |
+
"step": 1186
|
| 8344 |
+
},
|
| 8345 |
+
{
|
| 8346 |
+
"epoch": 0.003944822337598028,
|
| 8347 |
+
"grad_norm": 1.9737895727157593,
|
| 8348 |
+
"learning_rate": 2.671457389082309e-05,
|
| 8349 |
+
"loss": 0.6099,
|
| 8350 |
+
"step": 1187
|
| 8351 |
+
},
|
| 8352 |
+
{
|
| 8353 |
+
"epoch": 0.003948145692558094,
|
| 8354 |
+
"grad_norm": 2.717967987060547,
|
| 8355 |
+
"learning_rate": 2.657637632997898e-05,
|
| 8356 |
+
"loss": 0.8857,
|
| 8357 |
+
"step": 1188
|
| 8358 |
+
},
|
| 8359 |
+
{
|
| 8360 |
+
"epoch": 0.00395146904751816,
|
| 8361 |
+
"grad_norm": 2.020359516143799,
|
| 8362 |
+
"learning_rate": 2.6438482352829896e-05,
|
| 8363 |
+
"loss": 1.0062,
|
| 8364 |
+
"step": 1189
|
| 8365 |
+
},
|
| 8366 |
+
{
|
| 8367 |
+
"epoch": 0.003954792402478226,
|
| 8368 |
+
"grad_norm": 2.472184658050537,
|
| 8369 |
+
"learning_rate": 2.630089252952427e-05,
|
| 8370 |
+
"loss": 0.5646,
|
| 8371 |
+
"step": 1190
|
| 8372 |
+
},
|
| 8373 |
+
{
|
| 8374 |
+
"epoch": 0.003958115757438292,
|
| 8375 |
+
"grad_norm": 3.495150566101074,
|
| 8376 |
+
"learning_rate": 2.616360742895293e-05,
|
| 8377 |
+
"loss": 1.0481,
|
| 8378 |
+
"step": 1191
|
| 8379 |
+
},
|
| 8380 |
+
{
|
| 8381 |
+
"epoch": 0.003961439112398357,
|
| 8382 |
+
"grad_norm": 1.8180619478225708,
|
| 8383 |
+
"learning_rate": 2.602662761874679e-05,
|
| 8384 |
+
"loss": 0.9938,
|
| 8385 |
+
"step": 1192
|
| 8386 |
+
},
|
| 8387 |
+
{
|
| 8388 |
+
"epoch": 0.003964762467358423,
|
| 8389 |
+
"grad_norm": 2.7407257556915283,
|
| 8390 |
+
"learning_rate": 2.588995366527448e-05,
|
| 8391 |
+
"loss": 1.0012,
|
| 8392 |
+
"step": 1193
|
| 8393 |
+
},
|
| 8394 |
+
{
|
| 8395 |
+
"epoch": 0.003968085822318489,
|
| 8396 |
+
"grad_norm": 2.4475646018981934,
|
| 8397 |
+
"learning_rate": 2.5753586133640005e-05,
|
| 8398 |
+
"loss": 1.2745,
|
| 8399 |
+
"step": 1194
|
| 8400 |
+
},
|
| 8401 |
+
{
|
| 8402 |
+
"epoch": 0.003971409177278554,
|
| 8403 |
+
"grad_norm": 2.805459976196289,
|
| 8404 |
+
"learning_rate": 2.5617525587680402e-05,
|
| 8405 |
+
"loss": 0.9023,
|
| 8406 |
+
"step": 1195
|
| 8407 |
+
},
|
| 8408 |
+
{
|
| 8409 |
+
"epoch": 0.00397473253223862,
|
| 8410 |
+
"grad_norm": 2.728306293487549,
|
| 8411 |
+
"learning_rate": 2.5481772589963426e-05,
|
| 8412 |
+
"loss": 0.9511,
|
| 8413 |
+
"step": 1196
|
| 8414 |
+
},
|
| 8415 |
+
{
|
| 8416 |
+
"epoch": 0.003978055887198686,
|
| 8417 |
+
"grad_norm": 2.340111017227173,
|
| 8418 |
+
"learning_rate": 2.5346327701785234e-05,
|
| 8419 |
+
"loss": 1.0199,
|
| 8420 |
+
"step": 1197
|
| 8421 |
+
},
|
| 8422 |
+
{
|
| 8423 |
+
"epoch": 0.0039813792421587516,
|
| 8424 |
+
"grad_norm": 4.539111614227295,
|
| 8425 |
+
"learning_rate": 2.521119148316803e-05,
|
| 8426 |
+
"loss": 0.9674,
|
| 8427 |
+
"step": 1198
|
| 8428 |
+
},
|
| 8429 |
+
{
|
| 8430 |
+
"epoch": 0.003984702597118818,
|
| 8431 |
+
"grad_norm": 2.8320844173431396,
|
| 8432 |
+
"learning_rate": 2.5076364492857753e-05,
|
| 8433 |
+
"loss": 0.7264,
|
| 8434 |
+
"step": 1199
|
| 8435 |
+
},
|
| 8436 |
+
{
|
| 8437 |
+
"epoch": 0.003988025952078883,
|
| 8438 |
+
"grad_norm": 2.41717267036438,
|
| 8439 |
+
"learning_rate": 2.4941847288321797e-05,
|
| 8440 |
+
"loss": 0.77,
|
| 8441 |
+
"step": 1200
|
| 8442 |
+
},
|
| 8443 |
+
{
|
| 8444 |
+
"epoch": 0.003991349307038949,
|
| 8445 |
+
"grad_norm": 1.9176758527755737,
|
| 8446 |
+
"learning_rate": 2.4807640425746693e-05,
|
| 8447 |
+
"loss": 0.7773,
|
| 8448 |
+
"step": 1201
|
| 8449 |
+
},
|
| 8450 |
+
{
|
| 8451 |
+
"epoch": 0.003994672661999015,
|
| 8452 |
+
"grad_norm": 3.3391666412353516,
|
| 8453 |
+
"learning_rate": 2.4673744460035765e-05,
|
| 8454 |
+
"loss": 0.8565,
|
| 8455 |
+
"step": 1202
|
| 8456 |
+
},
|
| 8457 |
+
{
|
| 8458 |
+
"epoch": 0.00399799601695908,
|
| 8459 |
+
"grad_norm": 1.898715615272522,
|
| 8460 |
+
"learning_rate": 2.4540159944806974e-05,
|
| 8461 |
+
"loss": 0.8248,
|
| 8462 |
+
"step": 1203
|
| 8463 |
+
},
|
| 8464 |
+
{
|
| 8465 |
+
"epoch": 0.004001319371919146,
|
| 8466 |
+
"grad_norm": 2.5850183963775635,
|
| 8467 |
+
"learning_rate": 2.4406887432390424e-05,
|
| 8468 |
+
"loss": 0.9095,
|
| 8469 |
+
"step": 1204
|
| 8470 |
+
},
|
| 8471 |
+
{
|
| 8472 |
+
"epoch": 0.004004642726879211,
|
| 8473 |
+
"grad_norm": 2.522801637649536,
|
| 8474 |
+
"learning_rate": 2.427392747382623e-05,
|
| 8475 |
+
"loss": 0.6626,
|
| 8476 |
+
"step": 1205
|
| 8477 |
+
},
|
| 8478 |
+
{
|
| 8479 |
+
"epoch": 0.0040079660818392775,
|
| 8480 |
+
"grad_norm": 1.7238844633102417,
|
| 8481 |
+
"learning_rate": 2.4141280618862194e-05,
|
| 8482 |
+
"loss": 0.9149,
|
| 8483 |
+
"step": 1206
|
| 8484 |
+
},
|
| 8485 |
+
{
|
| 8486 |
+
"epoch": 0.004011289436799344,
|
| 8487 |
+
"grad_norm": 2.0944669246673584,
|
| 8488 |
+
"learning_rate": 2.4008947415951522e-05,
|
| 8489 |
+
"loss": 1.0417,
|
| 8490 |
+
"step": 1207
|
| 8491 |
+
},
|
| 8492 |
+
{
|
| 8493 |
+
"epoch": 0.004014612791759409,
|
| 8494 |
+
"grad_norm": 2.466402053833008,
|
| 8495 |
+
"learning_rate": 2.3876928412250588e-05,
|
| 8496 |
+
"loss": 0.9066,
|
| 8497 |
+
"step": 1208
|
| 8498 |
+
},
|
| 8499 |
+
{
|
| 8500 |
+
"epoch": 0.004017936146719475,
|
| 8501 |
+
"grad_norm": 3.359685182571411,
|
| 8502 |
+
"learning_rate": 2.374522415361663e-05,
|
| 8503 |
+
"loss": 0.7668,
|
| 8504 |
+
"step": 1209
|
| 8505 |
+
},
|
| 8506 |
+
{
|
| 8507 |
+
"epoch": 0.004021259501679541,
|
| 8508 |
+
"grad_norm": 2.3471083641052246,
|
| 8509 |
+
"learning_rate": 2.3613835184605525e-05,
|
| 8510 |
+
"loss": 1.0475,
|
| 8511 |
+
"step": 1210
|
| 8512 |
+
},
|
| 8513 |
+
{
|
| 8514 |
+
"epoch": 0.004024582856639606,
|
| 8515 |
+
"grad_norm": 2.58257794380188,
|
| 8516 |
+
"learning_rate": 2.348276204846951e-05,
|
| 8517 |
+
"loss": 1.0036,
|
| 8518 |
+
"step": 1211
|
| 8519 |
+
},
|
| 8520 |
+
{
|
| 8521 |
+
"epoch": 0.004027906211599672,
|
| 8522 |
+
"grad_norm": 1.622631549835205,
|
| 8523 |
+
"learning_rate": 2.3352005287154977e-05,
|
| 8524 |
+
"loss": 0.7247,
|
| 8525 |
+
"step": 1212
|
| 8526 |
+
},
|
| 8527 |
+
{
|
| 8528 |
+
"epoch": 0.004031229566559737,
|
| 8529 |
+
"grad_norm": 2.188223123550415,
|
| 8530 |
+
"learning_rate": 2.3221565441300197e-05,
|
| 8531 |
+
"loss": 0.9965,
|
| 8532 |
+
"step": 1213
|
| 8533 |
+
},
|
| 8534 |
+
{
|
| 8535 |
+
"epoch": 0.0040345529215198035,
|
| 8536 |
+
"grad_norm": 2.9655275344848633,
|
| 8537 |
+
"learning_rate": 2.3091443050233074e-05,
|
| 8538 |
+
"loss": 1.0074,
|
| 8539 |
+
"step": 1214
|
| 8540 |
+
},
|
| 8541 |
+
{
|
| 8542 |
+
"epoch": 0.0040378762764798695,
|
| 8543 |
+
"grad_norm": 1.9351389408111572,
|
| 8544 |
+
"learning_rate": 2.2961638651968975e-05,
|
| 8545 |
+
"loss": 0.7356,
|
| 8546 |
+
"step": 1215
|
| 8547 |
+
},
|
| 8548 |
+
{
|
| 8549 |
+
"epoch": 0.004041199631439935,
|
| 8550 |
+
"grad_norm": 1.64227294921875,
|
| 8551 |
+
"learning_rate": 2.2832152783208395e-05,
|
| 8552 |
+
"loss": 0.6714,
|
| 8553 |
+
"step": 1216
|
| 8554 |
+
},
|
| 8555 |
+
{
|
| 8556 |
+
"epoch": 0.004044522986400001,
|
| 8557 |
+
"grad_norm": 2.866924285888672,
|
| 8558 |
+
"learning_rate": 2.2702985979334935e-05,
|
| 8559 |
+
"loss": 1.3895,
|
| 8560 |
+
"step": 1217
|
| 8561 |
+
},
|
| 8562 |
+
{
|
| 8563 |
+
"epoch": 0.004047846341360066,
|
| 8564 |
+
"grad_norm": 1.7436803579330444,
|
| 8565 |
+
"learning_rate": 2.2574138774412857e-05,
|
| 8566 |
+
"loss": 0.8873,
|
| 8567 |
+
"step": 1218
|
| 8568 |
+
},
|
| 8569 |
+
{
|
| 8570 |
+
"epoch": 0.004051169696320132,
|
| 8571 |
+
"grad_norm": 2.3957839012145996,
|
| 8572 |
+
"learning_rate": 2.2445611701184998e-05,
|
| 8573 |
+
"loss": 0.921,
|
| 8574 |
+
"step": 1219
|
| 8575 |
+
},
|
| 8576 |
+
{
|
| 8577 |
+
"epoch": 0.004054493051280198,
|
| 8578 |
+
"grad_norm": 2.0950446128845215,
|
| 8579 |
+
"learning_rate": 2.231740529107057e-05,
|
| 8580 |
+
"loss": 1.1943,
|
| 8581 |
+
"step": 1220
|
| 8582 |
+
},
|
| 8583 |
+
{
|
| 8584 |
+
"epoch": 0.004057816406240263,
|
| 8585 |
+
"grad_norm": 3.499908924102783,
|
| 8586 |
+
"learning_rate": 2.218952007416294e-05,
|
| 8587 |
+
"loss": 0.5183,
|
| 8588 |
+
"step": 1221
|
| 8589 |
+
},
|
| 8590 |
+
{
|
| 8591 |
+
"epoch": 0.004061139761200329,
|
| 8592 |
+
"grad_norm": 4.9227294921875,
|
| 8593 |
+
"learning_rate": 2.206195657922745e-05,
|
| 8594 |
+
"loss": 0.843,
|
| 8595 |
+
"step": 1222
|
| 8596 |
+
},
|
| 8597 |
+
{
|
| 8598 |
+
"epoch": 0.0040644631161603955,
|
| 8599 |
+
"grad_norm": 2.2580618858337402,
|
| 8600 |
+
"learning_rate": 2.1934715333699187e-05,
|
| 8601 |
+
"loss": 0.8151,
|
| 8602 |
+
"step": 1223
|
| 8603 |
+
},
|
| 8604 |
+
{
|
| 8605 |
+
"epoch": 0.004067786471120461,
|
| 8606 |
+
"grad_norm": 1.6244633197784424,
|
| 8607 |
+
"learning_rate": 2.1807796863680886e-05,
|
| 8608 |
+
"loss": 1.0736,
|
| 8609 |
+
"step": 1224
|
| 8610 |
+
},
|
| 8611 |
+
{
|
| 8612 |
+
"epoch": 0.004071109826080527,
|
| 8613 |
+
"grad_norm": 2.2781097888946533,
|
| 8614 |
+
"learning_rate": 2.1681201693940668e-05,
|
| 8615 |
+
"loss": 0.7883,
|
| 8616 |
+
"step": 1225
|
| 8617 |
+
},
|
| 8618 |
+
{
|
| 8619 |
+
"epoch": 0.004074433181040592,
|
| 8620 |
+
"grad_norm": 1.6174863576889038,
|
| 8621 |
+
"learning_rate": 2.1554930347909918e-05,
|
| 8622 |
+
"loss": 1.0568,
|
| 8623 |
+
"step": 1226
|
| 8624 |
+
},
|
| 8625 |
+
{
|
| 8626 |
+
"epoch": 0.004077756536000658,
|
| 8627 |
+
"grad_norm": 2.1196584701538086,
|
| 8628 |
+
"learning_rate": 2.142898334768114e-05,
|
| 8629 |
+
"loss": 0.7304,
|
| 8630 |
+
"step": 1227
|
| 8631 |
+
},
|
| 8632 |
+
{
|
| 8633 |
+
"epoch": 0.004081079890960724,
|
| 8634 |
+
"grad_norm": 3.7660393714904785,
|
| 8635 |
+
"learning_rate": 2.1303361214005724e-05,
|
| 8636 |
+
"loss": 0.7623,
|
| 8637 |
+
"step": 1228
|
| 8638 |
+
},
|
| 8639 |
+
{
|
| 8640 |
+
"epoch": 0.004084403245920789,
|
| 8641 |
+
"grad_norm": 2.0749154090881348,
|
| 8642 |
+
"learning_rate": 2.117806446629188e-05,
|
| 8643 |
+
"loss": 0.2008,
|
| 8644 |
+
"step": 1229
|
| 8645 |
+
},
|
| 8646 |
+
{
|
| 8647 |
+
"epoch": 0.004087726600880855,
|
| 8648 |
+
"grad_norm": 2.255093812942505,
|
| 8649 |
+
"learning_rate": 2.1053093622602404e-05,
|
| 8650 |
+
"loss": 0.7191,
|
| 8651 |
+
"step": 1230
|
| 8652 |
+
},
|
| 8653 |
+
{
|
| 8654 |
+
"epoch": 0.0040910499558409206,
|
| 8655 |
+
"grad_norm": 3.07987380027771,
|
| 8656 |
+
"learning_rate": 2.09284491996526e-05,
|
| 8657 |
+
"loss": 0.7715,
|
| 8658 |
+
"step": 1231
|
| 8659 |
+
},
|
| 8660 |
+
{
|
| 8661 |
+
"epoch": 0.004094373310800987,
|
| 8662 |
+
"grad_norm": 2.6831705570220947,
|
| 8663 |
+
"learning_rate": 2.0804131712808174e-05,
|
| 8664 |
+
"loss": 0.8619,
|
| 8665 |
+
"step": 1232
|
| 8666 |
+
},
|
| 8667 |
+
{
|
| 8668 |
+
"epoch": 0.004097696665761053,
|
| 8669 |
+
"grad_norm": 2.8853580951690674,
|
| 8670 |
+
"learning_rate": 2.0680141676082997e-05,
|
| 8671 |
+
"loss": 0.8456,
|
| 8672 |
+
"step": 1233
|
| 8673 |
+
},
|
| 8674 |
+
{
|
| 8675 |
+
"epoch": 0.004101020020721118,
|
| 8676 |
+
"grad_norm": 2.1765217781066895,
|
| 8677 |
+
"learning_rate": 2.0556479602137035e-05,
|
| 8678 |
+
"loss": 0.9191,
|
| 8679 |
+
"step": 1234
|
| 8680 |
+
},
|
| 8681 |
+
{
|
| 8682 |
+
"epoch": 0.004104343375681184,
|
| 8683 |
+
"grad_norm": 1.470539927482605,
|
| 8684 |
+
"learning_rate": 2.043314600227425e-05,
|
| 8685 |
+
"loss": 0.7136,
|
| 8686 |
+
"step": 1235
|
| 8687 |
+
},
|
| 8688 |
+
{
|
| 8689 |
+
"epoch": 0.00410766673064125,
|
| 8690 |
+
"grad_norm": 2.399249792098999,
|
| 8691 |
+
"learning_rate": 2.0310141386440462e-05,
|
| 8692 |
+
"loss": 1.102,
|
| 8693 |
+
"step": 1236
|
| 8694 |
+
},
|
| 8695 |
+
{
|
| 8696 |
+
"epoch": 0.004110990085601315,
|
| 8697 |
+
"grad_norm": 2.0476646423339844,
|
| 8698 |
+
"learning_rate": 2.0187466263221245e-05,
|
| 8699 |
+
"loss": 0.9299,
|
| 8700 |
+
"step": 1237
|
| 8701 |
+
},
|
| 8702 |
+
{
|
| 8703 |
+
"epoch": 0.004114313440561381,
|
| 8704 |
+
"grad_norm": 2.2156193256378174,
|
| 8705 |
+
"learning_rate": 2.0065121139839826e-05,
|
| 8706 |
+
"loss": 0.8462,
|
| 8707 |
+
"step": 1238
|
| 8708 |
+
},
|
| 8709 |
+
{
|
| 8710 |
+
"epoch": 0.0041176367955214465,
|
| 8711 |
+
"grad_norm": 2.8007566928863525,
|
| 8712 |
+
"learning_rate": 1.9943106522154987e-05,
|
| 8713 |
+
"loss": 0.7217,
|
| 8714 |
+
"step": 1239
|
| 8715 |
+
},
|
| 8716 |
+
{
|
| 8717 |
+
"epoch": 0.004120960150481513,
|
| 8718 |
+
"grad_norm": 2.013310194015503,
|
| 8719 |
+
"learning_rate": 1.982142291465896e-05,
|
| 8720 |
+
"loss": 0.4966,
|
| 8721 |
+
"step": 1240
|
| 8722 |
+
},
|
| 8723 |
+
{
|
| 8724 |
+
"epoch": 0.004124283505441579,
|
| 8725 |
+
"grad_norm": 1.9652122259140015,
|
| 8726 |
+
"learning_rate": 1.970007082047537e-05,
|
| 8727 |
+
"loss": 0.7634,
|
| 8728 |
+
"step": 1241
|
| 8729 |
+
},
|
| 8730 |
+
{
|
| 8731 |
+
"epoch": 0.004127606860401644,
|
| 8732 |
+
"grad_norm": 3.3954484462738037,
|
| 8733 |
+
"learning_rate": 1.957905074135712e-05,
|
| 8734 |
+
"loss": 0.564,
|
| 8735 |
+
"step": 1242
|
| 8736 |
+
},
|
| 8737 |
+
{
|
| 8738 |
+
"epoch": 0.00413093021536171,
|
| 8739 |
+
"grad_norm": 1.620543122291565,
|
| 8740 |
+
"learning_rate": 1.945836317768437e-05,
|
| 8741 |
+
"loss": 0.9842,
|
| 8742 |
+
"step": 1243
|
| 8743 |
+
},
|
| 8744 |
+
{
|
| 8745 |
+
"epoch": 0.004134253570321775,
|
| 8746 |
+
"grad_norm": 1.858948826789856,
|
| 8747 |
+
"learning_rate": 1.933800862846239e-05,
|
| 8748 |
+
"loss": 0.8381,
|
| 8749 |
+
"step": 1244
|
| 8750 |
+
},
|
| 8751 |
+
{
|
| 8752 |
+
"epoch": 0.004137576925281841,
|
| 8753 |
+
"grad_norm": 1.7152379751205444,
|
| 8754 |
+
"learning_rate": 1.921798759131953e-05,
|
| 8755 |
+
"loss": 1.0301,
|
| 8756 |
+
"step": 1245
|
| 8757 |
+
},
|
| 8758 |
+
{
|
| 8759 |
+
"epoch": 0.004140900280241907,
|
| 8760 |
+
"grad_norm": 2.675569772720337,
|
| 8761 |
+
"learning_rate": 1.9098300562505266e-05,
|
| 8762 |
+
"loss": 0.664,
|
| 8763 |
+
"step": 1246
|
| 8764 |
+
},
|
| 8765 |
+
{
|
| 8766 |
+
"epoch": 0.0041442236352019724,
|
| 8767 |
+
"grad_norm": 1.6856192350387573,
|
| 8768 |
+
"learning_rate": 1.8978948036887946e-05,
|
| 8769 |
+
"loss": 0.8006,
|
| 8770 |
+
"step": 1247
|
| 8771 |
+
},
|
| 8772 |
+
{
|
| 8773 |
+
"epoch": 0.0041475469901620385,
|
| 8774 |
+
"grad_norm": 2.1971328258514404,
|
| 8775 |
+
"learning_rate": 1.8859930507952917e-05,
|
| 8776 |
+
"loss": 0.6714,
|
| 8777 |
+
"step": 1248
|
| 8778 |
+
},
|
| 8779 |
+
{
|
| 8780 |
+
"epoch": 0.004150870345122105,
|
| 8781 |
+
"grad_norm": 2.3259854316711426,
|
| 8782 |
+
"learning_rate": 1.8741248467800366e-05,
|
| 8783 |
+
"loss": 0.9986,
|
| 8784 |
+
"step": 1249
|
| 8785 |
+
},
|
| 8786 |
+
{
|
| 8787 |
+
"epoch": 0.00415419370008217,
|
| 8788 |
+
"grad_norm": 2.4596996307373047,
|
| 8789 |
+
"learning_rate": 1.8622902407143394e-05,
|
| 8790 |
+
"loss": 0.864,
|
| 8791 |
+
"step": 1250
|
| 8792 |
+
},
|
| 8793 |
+
{
|
| 8794 |
+
"epoch": 0.004157517055042236,
|
| 8795 |
+
"grad_norm": 2.8073737621307373,
|
| 8796 |
+
"learning_rate": 1.8504892815305906e-05,
|
| 8797 |
+
"loss": 1.0837,
|
| 8798 |
+
"step": 1251
|
| 8799 |
+
},
|
| 8800 |
+
{
|
| 8801 |
+
"epoch": 0.004160840410002301,
|
| 8802 |
+
"grad_norm": 2.7789711952209473,
|
| 8803 |
+
"learning_rate": 1.838722018022061e-05,
|
| 8804 |
+
"loss": 1.0432,
|
| 8805 |
+
"step": 1252
|
| 8806 |
+
},
|
| 8807 |
+
{
|
| 8808 |
+
"epoch": 0.004164163764962367,
|
| 8809 |
+
"grad_norm": 2.148526668548584,
|
| 8810 |
+
"learning_rate": 1.8269884988427e-05,
|
| 8811 |
+
"loss": 1.0993,
|
| 8812 |
+
"step": 1253
|
| 8813 |
+
},
|
| 8814 |
+
{
|
| 8815 |
+
"epoch": 0.004167487119922433,
|
| 8816 |
+
"grad_norm": 2.7591512203216553,
|
| 8817 |
+
"learning_rate": 1.8152887725069345e-05,
|
| 8818 |
+
"loss": 0.9319,
|
| 8819 |
+
"step": 1254
|
| 8820 |
+
},
|
| 8821 |
+
{
|
| 8822 |
+
"epoch": 0.004170810474882498,
|
| 8823 |
+
"grad_norm": 2.959975004196167,
|
| 8824 |
+
"learning_rate": 1.8036228873894746e-05,
|
| 8825 |
+
"loss": 0.8672,
|
| 8826 |
+
"step": 1255
|
| 8827 |
+
},
|
| 8828 |
+
{
|
| 8829 |
+
"epoch": 0.0041741338298425645,
|
| 8830 |
+
"grad_norm": 2.444984197616577,
|
| 8831 |
+
"learning_rate": 1.7919908917250992e-05,
|
| 8832 |
+
"loss": 0.9582,
|
| 8833 |
+
"step": 1256
|
| 8834 |
+
},
|
| 8835 |
+
{
|
| 8836 |
+
"epoch": 0.0041774571848026305,
|
| 8837 |
+
"grad_norm": 2.0599467754364014,
|
| 8838 |
+
"learning_rate": 1.7803928336084697e-05,
|
| 8839 |
+
"loss": 0.7514,
|
| 8840 |
+
"step": 1257
|
| 8841 |
+
},
|
| 8842 |
+
{
|
| 8843 |
+
"epoch": 0.004180780539762696,
|
| 8844 |
+
"grad_norm": 1.6798958778381348,
|
| 8845 |
+
"learning_rate": 1.7688287609939246e-05,
|
| 8846 |
+
"loss": 0.6838,
|
| 8847 |
+
"step": 1258
|
| 8848 |
+
},
|
| 8849 |
+
{
|
| 8850 |
+
"epoch": 0.004184103894722762,
|
| 8851 |
+
"grad_norm": 1.6114916801452637,
|
| 8852 |
+
"learning_rate": 1.757298721695283e-05,
|
| 8853 |
+
"loss": 1.0065,
|
| 8854 |
+
"step": 1259
|
| 8855 |
+
},
|
| 8856 |
+
{
|
| 8857 |
+
"epoch": 0.004187427249682827,
|
| 8858 |
+
"grad_norm": 3.5097157955169678,
|
| 8859 |
+
"learning_rate": 1.7458027633856478e-05,
|
| 8860 |
+
"loss": 0.7438,
|
| 8861 |
+
"step": 1260
|
| 8862 |
+
},
|
| 8863 |
+
{
|
| 8864 |
+
"epoch": 0.004190750604642893,
|
| 8865 |
+
"grad_norm": 2.6307899951934814,
|
| 8866 |
+
"learning_rate": 1.7343409335972073e-05,
|
| 8867 |
+
"loss": 0.975,
|
| 8868 |
+
"step": 1261
|
| 8869 |
+
},
|
| 8870 |
+
{
|
| 8871 |
+
"epoch": 0.004194073959602959,
|
| 8872 |
+
"grad_norm": 2.486320734024048,
|
| 8873 |
+
"learning_rate": 1.7229132797210356e-05,
|
| 8874 |
+
"loss": 0.8089,
|
| 8875 |
+
"step": 1262
|
| 8876 |
+
},
|
| 8877 |
+
{
|
| 8878 |
+
"epoch": 0.004197397314563024,
|
| 8879 |
+
"grad_norm": 2.3524551391601562,
|
| 8880 |
+
"learning_rate": 1.7115198490069095e-05,
|
| 8881 |
+
"loss": 0.6894,
|
| 8882 |
+
"step": 1263
|
| 8883 |
+
},
|
| 8884 |
+
{
|
| 8885 |
+
"epoch": 0.00420072066952309,
|
| 8886 |
+
"grad_norm": 2.064492702484131,
|
| 8887 |
+
"learning_rate": 1.700160688563095e-05,
|
| 8888 |
+
"loss": 0.8705,
|
| 8889 |
+
"step": 1264
|
| 8890 |
+
},
|
| 8891 |
+
{
|
| 8892 |
+
"epoch": 0.004204044024483156,
|
| 8893 |
+
"grad_norm": 1.568864107131958,
|
| 8894 |
+
"learning_rate": 1.6888358453561648e-05,
|
| 8895 |
+
"loss": 0.9711,
|
| 8896 |
+
"step": 1265
|
| 8897 |
+
},
|
| 8898 |
+
{
|
| 8899 |
+
"epoch": 0.004207367379443222,
|
| 8900 |
+
"grad_norm": 2.4715378284454346,
|
| 8901 |
+
"learning_rate": 1.6775453662107976e-05,
|
| 8902 |
+
"loss": 0.9384,
|
| 8903 |
+
"step": 1266
|
| 8904 |
+
},
|
| 8905 |
+
{
|
| 8906 |
+
"epoch": 0.004210690734403288,
|
| 8907 |
+
"grad_norm": 1.937577486038208,
|
| 8908 |
+
"learning_rate": 1.666289297809591e-05,
|
| 8909 |
+
"loss": 0.7837,
|
| 8910 |
+
"step": 1267
|
| 8911 |
+
},
|
| 8912 |
+
{
|
| 8913 |
+
"epoch": 0.004214014089363353,
|
| 8914 |
+
"grad_norm": 2.4397103786468506,
|
| 8915 |
+
"learning_rate": 1.655067686692865e-05,
|
| 8916 |
+
"loss": 0.5114,
|
| 8917 |
+
"step": 1268
|
| 8918 |
+
},
|
| 8919 |
+
{
|
| 8920 |
+
"epoch": 0.004217337444323419,
|
| 8921 |
+
"grad_norm": 2.0045480728149414,
|
| 8922 |
+
"learning_rate": 1.6438805792584656e-05,
|
| 8923 |
+
"loss": 1.1449,
|
| 8924 |
+
"step": 1269
|
| 8925 |
+
},
|
| 8926 |
+
{
|
| 8927 |
+
"epoch": 0.004220660799283485,
|
| 8928 |
+
"grad_norm": 2.310739040374756,
|
| 8929 |
+
"learning_rate": 1.6327280217615792e-05,
|
| 8930 |
+
"loss": 0.9197,
|
| 8931 |
+
"step": 1270
|
| 8932 |
+
},
|
| 8933 |
+
{
|
| 8934 |
+
"epoch": 0.00422398415424355,
|
| 8935 |
+
"grad_norm": 1.959408164024353,
|
| 8936 |
+
"learning_rate": 1.6216100603145425e-05,
|
| 8937 |
+
"loss": 1.0219,
|
| 8938 |
+
"step": 1271
|
| 8939 |
+
},
|
| 8940 |
+
{
|
| 8941 |
+
"epoch": 0.004227307509203616,
|
| 8942 |
+
"grad_norm": 2.1857080459594727,
|
| 8943 |
+
"learning_rate": 1.610526740886642e-05,
|
| 8944 |
+
"loss": 1.0474,
|
| 8945 |
+
"step": 1272
|
| 8946 |
+
},
|
| 8947 |
+
{
|
| 8948 |
+
"epoch": 0.0042306308641636816,
|
| 8949 |
+
"grad_norm": 2.136479616165161,
|
| 8950 |
+
"learning_rate": 1.5994781093039335e-05,
|
| 8951 |
+
"loss": 1.0238,
|
| 8952 |
+
"step": 1273
|
| 8953 |
+
},
|
| 8954 |
+
{
|
| 8955 |
+
"epoch": 0.004233954219123748,
|
| 8956 |
+
"grad_norm": 2.034966230392456,
|
| 8957 |
+
"learning_rate": 1.588464211249049e-05,
|
| 8958 |
+
"loss": 0.8942,
|
| 8959 |
+
"step": 1274
|
| 8960 |
+
},
|
| 8961 |
+
{
|
| 8962 |
+
"epoch": 0.004237277574083814,
|
| 8963 |
+
"grad_norm": 2.3870041370391846,
|
| 8964 |
+
"learning_rate": 1.577485092261012e-05,
|
| 8965 |
+
"loss": 0.7661,
|
| 8966 |
+
"step": 1275
|
| 8967 |
+
},
|
| 8968 |
+
{
|
| 8969 |
+
"epoch": 0.004240600929043879,
|
| 8970 |
+
"grad_norm": 2.3408772945404053,
|
| 8971 |
+
"learning_rate": 1.566540797735039e-05,
|
| 8972 |
+
"loss": 1.0518,
|
| 8973 |
+
"step": 1276
|
| 8974 |
+
},
|
| 8975 |
+
{
|
| 8976 |
+
"epoch": 0.004243924284003945,
|
| 8977 |
+
"grad_norm": 1.9745877981185913,
|
| 8978 |
+
"learning_rate": 1.5556313729223615e-05,
|
| 8979 |
+
"loss": 0.9101,
|
| 8980 |
+
"step": 1277
|
| 8981 |
+
},
|
| 8982 |
+
{
|
| 8983 |
+
"epoch": 0.00424724763896401,
|
| 8984 |
+
"grad_norm": 2.037195920944214,
|
| 8985 |
+
"learning_rate": 1.5447568629300357e-05,
|
| 8986 |
+
"loss": 0.8533,
|
| 8987 |
+
"step": 1278
|
| 8988 |
+
},
|
| 8989 |
+
{
|
| 8990 |
+
"epoch": 0.004250570993924076,
|
| 8991 |
+
"grad_norm": 2.7731986045837402,
|
| 8992 |
+
"learning_rate": 1.5339173127207564e-05,
|
| 8993 |
+
"loss": 0.87,
|
| 8994 |
+
"step": 1279
|
| 8995 |
+
},
|
| 8996 |
+
{
|
| 8997 |
+
"epoch": 0.004253894348884142,
|
| 8998 |
+
"grad_norm": 1.7042630910873413,
|
| 8999 |
+
"learning_rate": 1.5231127671126677e-05,
|
| 9000 |
+
"loss": 1.0944,
|
| 9001 |
+
"step": 1280
|
| 9002 |
+
},
|
| 9003 |
+
{
|
| 9004 |
+
"epoch": 0.0042572177038442075,
|
| 9005 |
+
"grad_norm": 2.3835747241973877,
|
| 9006 |
+
"learning_rate": 1.5123432707791851e-05,
|
| 9007 |
+
"loss": 0.9447,
|
| 9008 |
+
"step": 1281
|
| 9009 |
+
},
|
| 9010 |
+
{
|
| 9011 |
+
"epoch": 0.004260541058804274,
|
| 9012 |
+
"grad_norm": 1.7946707010269165,
|
| 9013 |
+
"learning_rate": 1.5016088682488028e-05,
|
| 9014 |
+
"loss": 0.6378,
|
| 9015 |
+
"step": 1282
|
| 9016 |
+
},
|
| 9017 |
+
{
|
| 9018 |
+
"epoch": 0.00426386441376434,
|
| 9019 |
+
"grad_norm": 2.4652082920074463,
|
| 9020 |
+
"learning_rate": 1.4909096039049152e-05,
|
| 9021 |
+
"loss": 0.8965,
|
| 9022 |
+
"step": 1283
|
| 9023 |
+
},
|
| 9024 |
+
{
|
| 9025 |
+
"epoch": 0.004267187768724405,
|
| 9026 |
+
"grad_norm": 1.9793140888214111,
|
| 9027 |
+
"learning_rate": 1.4802455219856293e-05,
|
| 9028 |
+
"loss": 0.9664,
|
| 9029 |
+
"step": 1284
|
| 9030 |
+
},
|
| 9031 |
+
{
|
| 9032 |
+
"epoch": 0.004270511123684471,
|
| 9033 |
+
"grad_norm": 2.0774343013763428,
|
| 9034 |
+
"learning_rate": 1.4696166665835853e-05,
|
| 9035 |
+
"loss": 0.9969,
|
| 9036 |
+
"step": 1285
|
| 9037 |
+
},
|
| 9038 |
+
{
|
| 9039 |
+
"epoch": 0.004273834478644536,
|
| 9040 |
+
"grad_norm": 2.2134058475494385,
|
| 9041 |
+
"learning_rate": 1.4590230816457728e-05,
|
| 9042 |
+
"loss": 0.9955,
|
| 9043 |
+
"step": 1286
|
| 9044 |
+
},
|
| 9045 |
+
{
|
| 9046 |
+
"epoch": 0.004277157833604602,
|
| 9047 |
+
"grad_norm": 2.212571859359741,
|
| 9048 |
+
"learning_rate": 1.4484648109733489e-05,
|
| 9049 |
+
"loss": 0.8405,
|
| 9050 |
+
"step": 1287
|
| 9051 |
+
},
|
| 9052 |
+
{
|
| 9053 |
+
"epoch": 0.004280481188564668,
|
| 9054 |
+
"grad_norm": 3.2596254348754883,
|
| 9055 |
+
"learning_rate": 1.4379418982214543e-05,
|
| 9056 |
+
"loss": 0.7455,
|
| 9057 |
+
"step": 1288
|
| 9058 |
+
},
|
| 9059 |
+
{
|
| 9060 |
+
"epoch": 0.0042838045435247335,
|
| 9061 |
+
"grad_norm": 3.176002264022827,
|
| 9062 |
+
"learning_rate": 1.427454386899042e-05,
|
| 9063 |
+
"loss": 1.0073,
|
| 9064 |
+
"step": 1289
|
| 9065 |
+
},
|
| 9066 |
+
{
|
| 9067 |
+
"epoch": 0.0042871278984847995,
|
| 9068 |
+
"grad_norm": 2.9529361724853516,
|
| 9069 |
+
"learning_rate": 1.4170023203686878e-05,
|
| 9070 |
+
"loss": 0.9066,
|
| 9071 |
+
"step": 1290
|
| 9072 |
+
},
|
| 9073 |
+
{
|
| 9074 |
+
"epoch": 0.004290451253444865,
|
| 9075 |
+
"grad_norm": 3.7993085384368896,
|
| 9076 |
+
"learning_rate": 1.4065857418464123e-05,
|
| 9077 |
+
"loss": 0.9157,
|
| 9078 |
+
"step": 1291
|
| 9079 |
+
},
|
| 9080 |
+
{
|
| 9081 |
+
"epoch": 0.004293774608404931,
|
| 9082 |
+
"grad_norm": 2.430190324783325,
|
| 9083 |
+
"learning_rate": 1.3962046944015072e-05,
|
| 9084 |
+
"loss": 1.2987,
|
| 9085 |
+
"step": 1292
|
| 9086 |
+
},
|
| 9087 |
+
{
|
| 9088 |
+
"epoch": 0.004297097963364997,
|
| 9089 |
+
"grad_norm": 3.1696853637695312,
|
| 9090 |
+
"learning_rate": 1.3858592209563525e-05,
|
| 9091 |
+
"loss": 0.9778,
|
| 9092 |
+
"step": 1293
|
| 9093 |
+
},
|
| 9094 |
+
{
|
| 9095 |
+
"epoch": 0.004300421318325062,
|
| 9096 |
+
"grad_norm": 4.211546897888184,
|
| 9097 |
+
"learning_rate": 1.3755493642862438e-05,
|
| 9098 |
+
"loss": 0.7872,
|
| 9099 |
+
"step": 1294
|
| 9100 |
+
},
|
| 9101 |
+
{
|
| 9102 |
+
"epoch": 0.004303744673285128,
|
| 9103 |
+
"grad_norm": 3.354203462600708,
|
| 9104 |
+
"learning_rate": 1.3652751670192077e-05,
|
| 9105 |
+
"loss": 0.8334,
|
| 9106 |
+
"step": 1295
|
| 9107 |
+
},
|
| 9108 |
+
{
|
| 9109 |
+
"epoch": 0.004307068028245194,
|
| 9110 |
+
"grad_norm": 1.8831151723861694,
|
| 9111 |
+
"learning_rate": 1.3550366716358354e-05,
|
| 9112 |
+
"loss": 1.1835,
|
| 9113 |
+
"step": 1296
|
| 9114 |
+
},
|
| 9115 |
+
{
|
| 9116 |
+
"epoch": 0.004310391383205259,
|
| 9117 |
+
"grad_norm": 3.2071986198425293,
|
| 9118 |
+
"learning_rate": 1.3448339204690975e-05,
|
| 9119 |
+
"loss": 0.5194,
|
| 9120 |
+
"step": 1297
|
| 9121 |
+
},
|
| 9122 |
+
{
|
| 9123 |
+
"epoch": 0.0043137147381653255,
|
| 9124 |
+
"grad_norm": 2.459270477294922,
|
| 9125 |
+
"learning_rate": 1.3346669557041779e-05,
|
| 9126 |
+
"loss": 1.0003,
|
| 9127 |
+
"step": 1298
|
| 9128 |
+
},
|
| 9129 |
+
{
|
| 9130 |
+
"epoch": 0.004317038093125391,
|
| 9131 |
+
"grad_norm": 3.5556893348693848,
|
| 9132 |
+
"learning_rate": 1.3245358193782909e-05,
|
| 9133 |
+
"loss": 1.3607,
|
| 9134 |
+
"step": 1299
|
| 9135 |
+
},
|
| 9136 |
+
{
|
| 9137 |
+
"epoch": 0.004320361448085457,
|
| 9138 |
+
"grad_norm": 5.798487186431885,
|
| 9139 |
+
"learning_rate": 1.3144405533805138e-05,
|
| 9140 |
+
"loss": 0.8204,
|
| 9141 |
+
"step": 1300
|
| 9142 |
+
},
|
| 9143 |
+
{
|
| 9144 |
+
"epoch": 0.004323684803045523,
|
| 9145 |
+
"grad_norm": 2.429063320159912,
|
| 9146 |
+
"learning_rate": 1.3043811994516097e-05,
|
| 9147 |
+
"loss": 0.6399,
|
| 9148 |
+
"step": 1301
|
| 9149 |
+
},
|
| 9150 |
+
{
|
| 9151 |
+
"epoch": 0.004327008158005588,
|
| 9152 |
+
"grad_norm": 1.6566165685653687,
|
| 9153 |
+
"learning_rate": 1.2943577991838596e-05,
|
| 9154 |
+
"loss": 0.6076,
|
| 9155 |
+
"step": 1302
|
| 9156 |
+
},
|
| 9157 |
+
{
|
| 9158 |
+
"epoch": 0.004330331512965654,
|
| 9159 |
+
"grad_norm": 2.3414995670318604,
|
| 9160 |
+
"learning_rate": 1.2843703940208818e-05,
|
| 9161 |
+
"loss": 1.1178,
|
| 9162 |
+
"step": 1303
|
| 9163 |
+
},
|
| 9164 |
+
{
|
| 9165 |
+
"epoch": 0.004333654867925719,
|
| 9166 |
+
"grad_norm": 3.708766222000122,
|
| 9167 |
+
"learning_rate": 1.2744190252574739e-05,
|
| 9168 |
+
"loss": 0.387,
|
| 9169 |
+
"step": 1304
|
| 9170 |
+
},
|
| 9171 |
+
{
|
| 9172 |
+
"epoch": 0.004336978222885785,
|
| 9173 |
+
"grad_norm": 1.8660900592803955,
|
| 9174 |
+
"learning_rate": 1.2645037340394284e-05,
|
| 9175 |
+
"loss": 0.5099,
|
| 9176 |
+
"step": 1305
|
| 9177 |
+
},
|
| 9178 |
+
{
|
| 9179 |
+
"epoch": 0.004340301577845851,
|
| 9180 |
+
"grad_norm": 1.8044732809066772,
|
| 9181 |
+
"learning_rate": 1.254624561363369e-05,
|
| 9182 |
+
"loss": 1.0002,
|
| 9183 |
+
"step": 1306
|
| 9184 |
+
},
|
| 9185 |
+
{
|
| 9186 |
+
"epoch": 0.004343624932805917,
|
| 9187 |
+
"grad_norm": 2.6514997482299805,
|
| 9188 |
+
"learning_rate": 1.2447815480765834e-05,
|
| 9189 |
+
"loss": 0.9949,
|
| 9190 |
+
"step": 1307
|
| 9191 |
+
},
|
| 9192 |
+
{
|
| 9193 |
+
"epoch": 0.004346948287765983,
|
| 9194 |
+
"grad_norm": 2.5847949981689453,
|
| 9195 |
+
"learning_rate": 1.234974734876847e-05,
|
| 9196 |
+
"loss": 1.0075,
|
| 9197 |
+
"step": 1308
|
| 9198 |
+
},
|
| 9199 |
+
{
|
| 9200 |
+
"epoch": 0.004350271642726049,
|
| 9201 |
+
"grad_norm": 1.7802176475524902,
|
| 9202 |
+
"learning_rate": 1.2252041623122645e-05,
|
| 9203 |
+
"loss": 1.0541,
|
| 9204 |
+
"step": 1309
|
| 9205 |
+
},
|
| 9206 |
+
{
|
| 9207 |
+
"epoch": 0.004353594997686114,
|
| 9208 |
+
"grad_norm": 2.3329453468322754,
|
| 9209 |
+
"learning_rate": 1.215469870781093e-05,
|
| 9210 |
+
"loss": 0.9778,
|
| 9211 |
+
"step": 1310
|
| 9212 |
+
},
|
| 9213 |
+
{
|
| 9214 |
+
"epoch": 0.00435691835264618,
|
| 9215 |
+
"grad_norm": 2.180158853530884,
|
| 9216 |
+
"learning_rate": 1.2057719005315793e-05,
|
| 9217 |
+
"loss": 0.8749,
|
| 9218 |
+
"step": 1311
|
| 9219 |
+
},
|
| 9220 |
+
{
|
| 9221 |
+
"epoch": 0.004360241707606245,
|
| 9222 |
+
"grad_norm": 2.902301788330078,
|
| 9223 |
+
"learning_rate": 1.1961102916617961e-05,
|
| 9224 |
+
"loss": 0.7362,
|
| 9225 |
+
"step": 1312
|
| 9226 |
+
},
|
| 9227 |
+
{
|
| 9228 |
+
"epoch": 0.004363565062566311,
|
| 9229 |
+
"grad_norm": 2.000943660736084,
|
| 9230 |
+
"learning_rate": 1.1864850841194697e-05,
|
| 9231 |
+
"loss": 0.8745,
|
| 9232 |
+
"step": 1313
|
| 9233 |
+
},
|
| 9234 |
+
{
|
| 9235 |
+
"epoch": 0.004366888417526377,
|
| 9236 |
+
"grad_norm": 3.558805227279663,
|
| 9237 |
+
"learning_rate": 1.17689631770182e-05,
|
| 9238 |
+
"loss": 0.794,
|
| 9239 |
+
"step": 1314
|
| 9240 |
+
},
|
| 9241 |
+
{
|
| 9242 |
+
"epoch": 0.004370211772486443,
|
| 9243 |
+
"grad_norm": 2.757683038711548,
|
| 9244 |
+
"learning_rate": 1.167344032055394e-05,
|
| 9245 |
+
"loss": 1.2417,
|
| 9246 |
+
"step": 1315
|
| 9247 |
+
},
|
| 9248 |
+
{
|
| 9249 |
+
"epoch": 0.004373535127446509,
|
| 9250 |
+
"grad_norm": 2.3887696266174316,
|
| 9251 |
+
"learning_rate": 1.1578282666759021e-05,
|
| 9252 |
+
"loss": 1.017,
|
| 9253 |
+
"step": 1316
|
| 9254 |
+
},
|
| 9255 |
+
{
|
| 9256 |
+
"epoch": 0.004376858482406574,
|
| 9257 |
+
"grad_norm": 2.794624090194702,
|
| 9258 |
+
"learning_rate": 1.1483490609080517e-05,
|
| 9259 |
+
"loss": 0.7826,
|
| 9260 |
+
"step": 1317
|
| 9261 |
+
},
|
| 9262 |
+
{
|
| 9263 |
+
"epoch": 0.00438018183736664,
|
| 9264 |
+
"grad_norm": 2.7033398151397705,
|
| 9265 |
+
"learning_rate": 1.1389064539453953e-05,
|
| 9266 |
+
"loss": 0.746,
|
| 9267 |
+
"step": 1318
|
| 9268 |
+
},
|
| 9269 |
+
{
|
| 9270 |
+
"epoch": 0.004383505192326706,
|
| 9271 |
+
"grad_norm": 1.7172794342041016,
|
| 9272 |
+
"learning_rate": 1.129500484830156e-05,
|
| 9273 |
+
"loss": 1.0663,
|
| 9274 |
+
"step": 1319
|
| 9275 |
+
},
|
| 9276 |
+
{
|
| 9277 |
+
"epoch": 0.004386828547286771,
|
| 9278 |
+
"grad_norm": 2.080857515335083,
|
| 9279 |
+
"learning_rate": 1.120131192453069e-05,
|
| 9280 |
+
"loss": 0.7509,
|
| 9281 |
+
"step": 1320
|
| 9282 |
+
},
|
| 9283 |
+
{
|
| 9284 |
+
"epoch": 0.004390151902246837,
|
| 9285 |
+
"grad_norm": 2.202035427093506,
|
| 9286 |
+
"learning_rate": 1.1107986155532247e-05,
|
| 9287 |
+
"loss": 1.1129,
|
| 9288 |
+
"step": 1321
|
| 9289 |
+
},
|
| 9290 |
+
{
|
| 9291 |
+
"epoch": 0.004393475257206903,
|
| 9292 |
+
"grad_norm": 2.652700424194336,
|
| 9293 |
+
"learning_rate": 1.1015027927179079e-05,
|
| 9294 |
+
"loss": 0.8909,
|
| 9295 |
+
"step": 1322
|
| 9296 |
+
},
|
| 9297 |
+
{
|
| 9298 |
+
"epoch": 0.0043967986121669685,
|
| 9299 |
+
"grad_norm": 2.584562063217163,
|
| 9300 |
+
"learning_rate": 1.0922437623824333e-05,
|
| 9301 |
+
"loss": 0.9037,
|
| 9302 |
+
"step": 1323
|
| 9303 |
+
},
|
| 9304 |
+
{
|
| 9305 |
+
"epoch": 0.004400121967127035,
|
| 9306 |
+
"grad_norm": 2.0051112174987793,
|
| 9307 |
+
"learning_rate": 1.0830215628299956e-05,
|
| 9308 |
+
"loss": 1.0131,
|
| 9309 |
+
"step": 1324
|
| 9310 |
+
},
|
| 9311 |
+
{
|
| 9312 |
+
"epoch": 0.0044034453220871,
|
| 9313 |
+
"grad_norm": 2.6688451766967773,
|
| 9314 |
+
"learning_rate": 1.0738362321914997e-05,
|
| 9315 |
+
"loss": 0.6386,
|
| 9316 |
+
"step": 1325
|
| 9317 |
+
},
|
| 9318 |
+
{
|
| 9319 |
+
"epoch": 0.004406768677047166,
|
| 9320 |
+
"grad_norm": 3.4277102947235107,
|
| 9321 |
+
"learning_rate": 1.0646878084454138e-05,
|
| 9322 |
+
"loss": 1.2131,
|
| 9323 |
+
"step": 1326
|
| 9324 |
+
},
|
| 9325 |
+
{
|
| 9326 |
+
"epoch": 0.004410092032007232,
|
| 9327 |
+
"grad_norm": 2.226804256439209,
|
| 9328 |
+
"learning_rate": 1.0555763294176047e-05,
|
| 9329 |
+
"loss": 0.773,
|
| 9330 |
+
"step": 1327
|
| 9331 |
+
},
|
| 9332 |
+
{
|
| 9333 |
+
"epoch": 0.004413415386967297,
|
| 9334 |
+
"grad_norm": 2.501744508743286,
|
| 9335 |
+
"learning_rate": 1.0465018327811849e-05,
|
| 9336 |
+
"loss": 0.7347,
|
| 9337 |
+
"step": 1328
|
| 9338 |
+
},
|
| 9339 |
+
{
|
| 9340 |
+
"epoch": 0.004416738741927363,
|
| 9341 |
+
"grad_norm": 1.8200547695159912,
|
| 9342 |
+
"learning_rate": 1.0374643560563579e-05,
|
| 9343 |
+
"loss": 0.7342,
|
| 9344 |
+
"step": 1329
|
| 9345 |
+
},
|
| 9346 |
+
{
|
| 9347 |
+
"epoch": 0.004420062096887428,
|
| 9348 |
+
"grad_norm": 2.6898293495178223,
|
| 9349 |
+
"learning_rate": 1.02846393661026e-05,
|
| 9350 |
+
"loss": 0.5875,
|
| 9351 |
+
"step": 1330
|
| 9352 |
+
},
|
| 9353 |
+
{
|
| 9354 |
+
"epoch": 0.0044233854518474945,
|
| 9355 |
+
"grad_norm": 2.5434093475341797,
|
| 9356 |
+
"learning_rate": 1.0195006116568074e-05,
|
| 9357 |
+
"loss": 0.5544,
|
| 9358 |
+
"step": 1331
|
| 9359 |
+
},
|
| 9360 |
+
{
|
| 9361 |
+
"epoch": 0.0044267088068075605,
|
| 9362 |
+
"grad_norm": 2.0328285694122314,
|
| 9363 |
+
"learning_rate": 1.0105744182565401e-05,
|
| 9364 |
+
"loss": 0.8736,
|
| 9365 |
+
"step": 1332
|
| 9366 |
+
},
|
| 9367 |
+
{
|
| 9368 |
+
"epoch": 0.004430032161767626,
|
| 9369 |
+
"grad_norm": 1.593604326248169,
|
| 9370 |
+
"learning_rate": 1.0016853933164772e-05,
|
| 9371 |
+
"loss": 0.6265,
|
| 9372 |
+
"step": 1333
|
| 9373 |
+
},
|
| 9374 |
+
{
|
| 9375 |
+
"epoch": 0.004433355516727692,
|
| 9376 |
+
"grad_norm": 2.4271087646484375,
|
| 9377 |
+
"learning_rate": 9.92833573589953e-06,
|
| 9378 |
+
"loss": 0.8883,
|
| 9379 |
+
"step": 1334
|
| 9380 |
+
},
|
| 9381 |
+
{
|
| 9382 |
+
"epoch": 0.004436678871687758,
|
| 9383 |
+
"grad_norm": 2.8914899826049805,
|
| 9384 |
+
"learning_rate": 9.840189956764677e-06,
|
| 9385 |
+
"loss": 0.5926,
|
| 9386 |
+
"step": 1335
|
| 9387 |
+
},
|
| 9388 |
+
{
|
| 9389 |
+
"epoch": 0.004440002226647823,
|
| 9390 |
+
"grad_norm": 2.0368692874908447,
|
| 9391 |
+
"learning_rate": 9.752416960215438e-06,
|
| 9392 |
+
"loss": 0.8281,
|
| 9393 |
+
"step": 1336
|
| 9394 |
+
},
|
| 9395 |
+
{
|
| 9396 |
+
"epoch": 0.004443325581607889,
|
| 9397 |
+
"grad_norm": 3.7236480712890625,
|
| 9398 |
+
"learning_rate": 9.665017109165663e-06,
|
| 9399 |
+
"loss": 0.829,
|
| 9400 |
+
"step": 1337
|
| 9401 |
+
},
|
| 9402 |
+
{
|
| 9403 |
+
"epoch": 0.004446648936567954,
|
| 9404 |
+
"grad_norm": 3.2966623306274414,
|
| 9405 |
+
"learning_rate": 9.577990764986366e-06,
|
| 9406 |
+
"loss": 1.2795,
|
| 9407 |
+
"step": 1338
|
| 9408 |
+
},
|
| 9409 |
+
{
|
| 9410 |
+
"epoch": 0.00444997229152802,
|
| 9411 |
+
"grad_norm": 2.6626784801483154,
|
| 9412 |
+
"learning_rate": 9.491338287504249e-06,
|
| 9413 |
+
"loss": 0.4713,
|
| 9414 |
+
"step": 1339
|
| 9415 |
+
},
|
| 9416 |
+
{
|
| 9417 |
+
"epoch": 0.0044532956464880865,
|
| 9418 |
+
"grad_norm": 4.007009506225586,
|
| 9419 |
+
"learning_rate": 9.405060035000135e-06,
|
| 9420 |
+
"loss": 0.5719,
|
| 9421 |
+
"step": 1340
|
| 9422 |
+
},
|
| 9423 |
+
{
|
| 9424 |
+
"epoch": 0.004456619001448152,
|
| 9425 |
+
"grad_norm": 2.7674596309661865,
|
| 9426 |
+
"learning_rate": 9.319156364207637e-06,
|
| 9427 |
+
"loss": 1.0912,
|
| 9428 |
+
"step": 1341
|
| 9429 |
+
},
|
| 9430 |
+
{
|
| 9431 |
+
"epoch": 0.004459942356408218,
|
| 9432 |
+
"grad_norm": 2.6112825870513916,
|
| 9433 |
+
"learning_rate": 9.233627630311504e-06,
|
| 9434 |
+
"loss": 0.9995,
|
| 9435 |
+
"step": 1342
|
| 9436 |
+
},
|
| 9437 |
+
{
|
| 9438 |
+
"epoch": 0.004463265711368283,
|
| 9439 |
+
"grad_norm": 2.9278571605682373,
|
| 9440 |
+
"learning_rate": 9.148474186946288e-06,
|
| 9441 |
+
"loss": 0.8194,
|
| 9442 |
+
"step": 1343
|
| 9443 |
+
},
|
| 9444 |
+
{
|
| 9445 |
+
"epoch": 0.004466589066328349,
|
| 9446 |
+
"grad_norm": 4.50659704208374,
|
| 9447 |
+
"learning_rate": 9.06369638619481e-06,
|
| 9448 |
+
"loss": 0.8752,
|
| 9449 |
+
"step": 1344
|
| 9450 |
+
},
|
| 9451 |
+
{
|
| 9452 |
+
"epoch": 0.004469912421288415,
|
| 9453 |
+
"grad_norm": 2.8715479373931885,
|
| 9454 |
+
"learning_rate": 8.979294578586738e-06,
|
| 9455 |
+
"loss": 0.9225,
|
| 9456 |
+
"step": 1345
|
| 9457 |
+
},
|
| 9458 |
+
{
|
| 9459 |
+
"epoch": 0.00447323577624848,
|
| 9460 |
+
"grad_norm": 2.1400842666625977,
|
| 9461 |
+
"learning_rate": 8.895269113097104e-06,
|
| 9462 |
+
"loss": 0.731,
|
| 9463 |
+
"step": 1346
|
| 9464 |
+
},
|
| 9465 |
+
{
|
| 9466 |
+
"epoch": 0.004476559131208546,
|
| 9467 |
+
"grad_norm": 1.7783732414245605,
|
| 9468 |
+
"learning_rate": 8.81162033714491e-06,
|
| 9469 |
+
"loss": 0.8749,
|
| 9470 |
+
"step": 1347
|
| 9471 |
+
},
|
| 9472 |
+
{
|
| 9473 |
+
"epoch": 0.0044798824861686124,
|
| 9474 |
+
"grad_norm": 2.487292528152466,
|
| 9475 |
+
"learning_rate": 8.72834859659164e-06,
|
| 9476 |
+
"loss": 1.1521,
|
| 9477 |
+
"step": 1348
|
| 9478 |
+
},
|
| 9479 |
+
{
|
| 9480 |
+
"epoch": 0.004483205841128678,
|
| 9481 |
+
"grad_norm": 3.0707662105560303,
|
| 9482 |
+
"learning_rate": 8.645454235739903e-06,
|
| 9483 |
+
"loss": 0.575,
|
| 9484 |
+
"step": 1349
|
| 9485 |
+
},
|
| 9486 |
+
{
|
| 9487 |
+
"epoch": 0.004486529196088744,
|
| 9488 |
+
"grad_norm": 2.8332557678222656,
|
| 9489 |
+
"learning_rate": 8.562937597331899e-06,
|
| 9490 |
+
"loss": 0.7608,
|
| 9491 |
+
"step": 1350
|
| 9492 |
+
},
|
| 9493 |
+
{
|
| 9494 |
+
"epoch": 0.004489852551048809,
|
| 9495 |
+
"grad_norm": 1.9897592067718506,
|
| 9496 |
+
"learning_rate": 8.480799022548114e-06,
|
| 9497 |
+
"loss": 0.9413,
|
| 9498 |
+
"step": 1351
|
| 9499 |
+
},
|
| 9500 |
+
{
|
| 9501 |
+
"epoch": 0.004493175906008875,
|
| 9502 |
+
"grad_norm": 3.2400360107421875,
|
| 9503 |
+
"learning_rate": 8.399038851005802e-06,
|
| 9504 |
+
"loss": 0.7033,
|
| 9505 |
+
"step": 1352
|
| 9506 |
+
},
|
| 9507 |
+
{
|
| 9508 |
+
"epoch": 0.004496499260968941,
|
| 9509 |
+
"grad_norm": 2.4915127754211426,
|
| 9510 |
+
"learning_rate": 8.317657420757707e-06,
|
| 9511 |
+
"loss": 0.5627,
|
| 9512 |
+
"step": 1353
|
| 9513 |
+
},
|
| 9514 |
+
{
|
| 9515 |
+
"epoch": 0.004499822615929006,
|
| 9516 |
+
"grad_norm": 2.3583595752716064,
|
| 9517 |
+
"learning_rate": 8.236655068290555e-06,
|
| 9518 |
+
"loss": 0.8612,
|
| 9519 |
+
"step": 1354
|
| 9520 |
+
},
|
| 9521 |
+
{
|
| 9522 |
+
"epoch": 0.004503145970889072,
|
| 9523 |
+
"grad_norm": 4.30778694152832,
|
| 9524 |
+
"learning_rate": 8.156032128523694e-06,
|
| 9525 |
+
"loss": 0.8659,
|
| 9526 |
+
"step": 1355
|
| 9527 |
+
},
|
| 9528 |
+
{
|
| 9529 |
+
"epoch": 0.004506469325849138,
|
| 9530 |
+
"grad_norm": 1.7405524253845215,
|
| 9531 |
+
"learning_rate": 8.07578893480776e-06,
|
| 9532 |
+
"loss": 0.7578,
|
| 9533 |
+
"step": 1356
|
| 9534 |
+
},
|
| 9535 |
+
{
|
| 9536 |
+
"epoch": 0.004509792680809204,
|
| 9537 |
+
"grad_norm": 2.5440306663513184,
|
| 9538 |
+
"learning_rate": 7.995925818923223e-06,
|
| 9539 |
+
"loss": 1.1294,
|
| 9540 |
+
"step": 1357
|
| 9541 |
+
},
|
| 9542 |
+
{
|
| 9543 |
+
"epoch": 0.00451311603576927,
|
| 9544 |
+
"grad_norm": 1.922419786453247,
|
| 9545 |
+
"learning_rate": 7.916443111079052e-06,
|
| 9546 |
+
"loss": 0.8335,
|
| 9547 |
+
"step": 1358
|
| 9548 |
+
},
|
| 9549 |
+
{
|
| 9550 |
+
"epoch": 0.004516439390729335,
|
| 9551 |
+
"grad_norm": 2.027200698852539,
|
| 9552 |
+
"learning_rate": 7.83734113991138e-06,
|
| 9553 |
+
"loss": 0.9653,
|
| 9554 |
+
"step": 1359
|
| 9555 |
+
},
|
| 9556 |
+
{
|
| 9557 |
+
"epoch": 0.004519762745689401,
|
| 9558 |
+
"grad_norm": 2.6092543601989746,
|
| 9559 |
+
"learning_rate": 7.758620232482084e-06,
|
| 9560 |
+
"loss": 0.6077,
|
| 9561 |
+
"step": 1360
|
| 9562 |
+
},
|
| 9563 |
+
{
|
| 9564 |
+
"epoch": 0.004523086100649467,
|
| 9565 |
+
"grad_norm": 2.5783867835998535,
|
| 9566 |
+
"learning_rate": 7.680280714277443e-06,
|
| 9567 |
+
"loss": 1.2442,
|
| 9568 |
+
"step": 1361
|
| 9569 |
+
},
|
| 9570 |
+
{
|
| 9571 |
+
"epoch": 0.004526409455609532,
|
| 9572 |
+
"grad_norm": 2.3049886226654053,
|
| 9573 |
+
"learning_rate": 7.602322909206872e-06,
|
| 9574 |
+
"loss": 0.7618,
|
| 9575 |
+
"step": 1362
|
| 9576 |
+
},
|
| 9577 |
+
{
|
| 9578 |
+
"epoch": 0.004529732810569598,
|
| 9579 |
+
"grad_norm": 2.7443058490753174,
|
| 9580 |
+
"learning_rate": 7.524747139601473e-06,
|
| 9581 |
+
"loss": 0.6442,
|
| 9582 |
+
"step": 1363
|
| 9583 |
+
},
|
| 9584 |
+
{
|
| 9585 |
+
"epoch": 0.0045330561655296635,
|
| 9586 |
+
"grad_norm": 1.7677465677261353,
|
| 9587 |
+
"learning_rate": 7.447553726212774e-06,
|
| 9588 |
+
"loss": 0.9181,
|
| 9589 |
+
"step": 1364
|
| 9590 |
+
},
|
| 9591 |
+
{
|
| 9592 |
+
"epoch": 0.0045363795204897295,
|
| 9593 |
+
"grad_norm": 2.5418739318847656,
|
| 9594 |
+
"learning_rate": 7.370742988211365e-06,
|
| 9595 |
+
"loss": 1.3973,
|
| 9596 |
+
"step": 1365
|
| 9597 |
+
},
|
| 9598 |
+
{
|
| 9599 |
+
"epoch": 0.004539702875449796,
|
| 9600 |
+
"grad_norm": 2.5869128704071045,
|
| 9601 |
+
"learning_rate": 7.294315243185579e-06,
|
| 9602 |
+
"loss": 1.0766,
|
| 9603 |
+
"step": 1366
|
| 9604 |
+
},
|
| 9605 |
+
{
|
| 9606 |
+
"epoch": 0.004543026230409861,
|
| 9607 |
+
"grad_norm": 2.520656108856201,
|
| 9608 |
+
"learning_rate": 7.218270807140237e-06,
|
| 9609 |
+
"loss": 0.9667,
|
| 9610 |
+
"step": 1367
|
| 9611 |
+
},
|
| 9612 |
+
{
|
| 9613 |
+
"epoch": 0.004546349585369927,
|
| 9614 |
+
"grad_norm": 2.4313716888427734,
|
| 9615 |
+
"learning_rate": 7.142609994495264e-06,
|
| 9616 |
+
"loss": 0.862,
|
| 9617 |
+
"step": 1368
|
| 9618 |
+
},
|
| 9619 |
+
{
|
| 9620 |
+
"epoch": 0.004549672940329993,
|
| 9621 |
+
"grad_norm": 2.530888795852661,
|
| 9622 |
+
"learning_rate": 7.067333118084429e-06,
|
| 9623 |
+
"loss": 1.1411,
|
| 9624 |
+
"step": 1369
|
| 9625 |
+
},
|
| 9626 |
+
{
|
| 9627 |
+
"epoch": 0.004552996295290058,
|
| 9628 |
+
"grad_norm": 3.0306622982025146,
|
| 9629 |
+
"learning_rate": 6.992440489154051e-06,
|
| 9630 |
+
"loss": 0.7518,
|
| 9631 |
+
"step": 1370
|
| 9632 |
+
},
|
| 9633 |
+
{
|
| 9634 |
+
"epoch": 0.004556319650250124,
|
| 9635 |
+
"grad_norm": 1.9484350681304932,
|
| 9636 |
+
"learning_rate": 6.9179324173617056e-06,
|
| 9637 |
+
"loss": 0.7771,
|
| 9638 |
+
"step": 1371
|
| 9639 |
+
},
|
| 9640 |
+
{
|
| 9641 |
+
"epoch": 0.004559643005210189,
|
| 9642 |
+
"grad_norm": 2.236544132232666,
|
| 9643 |
+
"learning_rate": 6.8438092107749206e-06,
|
| 9644 |
+
"loss": 0.5691,
|
| 9645 |
+
"step": 1372
|
| 9646 |
+
},
|
| 9647 |
+
{
|
| 9648 |
+
"epoch": 0.0045629663601702555,
|
| 9649 |
+
"grad_norm": 1.5882771015167236,
|
| 9650 |
+
"learning_rate": 6.770071175869974e-06,
|
| 9651 |
+
"loss": 1.115,
|
| 9652 |
+
"step": 1373
|
| 9653 |
+
},
|
| 9654 |
+
{
|
| 9655 |
+
"epoch": 0.0045662897151303216,
|
| 9656 |
+
"grad_norm": 2.735529661178589,
|
| 9657 |
+
"learning_rate": 6.696718617530529e-06,
|
| 9658 |
+
"loss": 1.0615,
|
| 9659 |
+
"step": 1374
|
| 9660 |
+
},
|
| 9661 |
+
{
|
| 9662 |
+
"epoch": 0.004569613070090387,
|
| 9663 |
+
"grad_norm": 3.5812888145446777,
|
| 9664 |
+
"learning_rate": 6.623751839046455e-06,
|
| 9665 |
+
"loss": 1.0835,
|
| 9666 |
+
"step": 1375
|
| 9667 |
+
},
|
| 9668 |
+
{
|
| 9669 |
+
"epoch": 0.004572936425050453,
|
| 9670 |
+
"grad_norm": 3.308802843093872,
|
| 9671 |
+
"learning_rate": 6.551171142112544e-06,
|
| 9672 |
+
"loss": 0.6246,
|
| 9673 |
+
"step": 1376
|
| 9674 |
+
},
|
| 9675 |
+
{
|
| 9676 |
+
"epoch": 0.004576259780010518,
|
| 9677 |
+
"grad_norm": 2.600813627243042,
|
| 9678 |
+
"learning_rate": 6.4789768268272544e-06,
|
| 9679 |
+
"loss": 0.4047,
|
| 9680 |
+
"step": 1377
|
| 9681 |
+
},
|
| 9682 |
+
{
|
| 9683 |
+
"epoch": 0.004579583134970584,
|
| 9684 |
+
"grad_norm": 2.414781093597412,
|
| 9685 |
+
"learning_rate": 6.407169191691464e-06,
|
| 9686 |
+
"loss": 0.8842,
|
| 9687 |
+
"step": 1378
|
| 9688 |
+
},
|
| 9689 |
+
{
|
| 9690 |
+
"epoch": 0.00458290648993065,
|
| 9691 |
+
"grad_norm": 2.679520606994629,
|
| 9692 |
+
"learning_rate": 6.335748533607278e-06,
|
| 9693 |
+
"loss": 0.793,
|
| 9694 |
+
"step": 1379
|
| 9695 |
+
},
|
| 9696 |
+
{
|
| 9697 |
+
"epoch": 0.004586229844890715,
|
| 9698 |
+
"grad_norm": 1.5030497312545776,
|
| 9699 |
+
"learning_rate": 6.264715147876743e-06,
|
| 9700 |
+
"loss": 1.077,
|
| 9701 |
+
"step": 1380
|
| 9702 |
+
},
|
| 9703 |
+
{
|
| 9704 |
+
"epoch": 0.0045895531998507814,
|
| 9705 |
+
"grad_norm": 1.6694726943969727,
|
| 9706 |
+
"learning_rate": 6.194069328200669e-06,
|
| 9707 |
+
"loss": 1.0486,
|
| 9708 |
+
"step": 1381
|
| 9709 |
+
},
|
| 9710 |
+
{
|
| 9711 |
+
"epoch": 0.0045928765548108475,
|
| 9712 |
+
"grad_norm": 2.1914167404174805,
|
| 9713 |
+
"learning_rate": 6.123811366677401e-06,
|
| 9714 |
+
"loss": 0.898,
|
| 9715 |
+
"step": 1382
|
| 9716 |
+
},
|
| 9717 |
+
{
|
| 9718 |
+
"epoch": 0.004596199909770913,
|
| 9719 |
+
"grad_norm": 3.29081130027771,
|
| 9720 |
+
"learning_rate": 6.053941553801612e-06,
|
| 9721 |
+
"loss": 1.001,
|
| 9722 |
+
"step": 1383
|
| 9723 |
+
},
|
| 9724 |
+
{
|
| 9725 |
+
"epoch": 0.004599523264730979,
|
| 9726 |
+
"grad_norm": 2.1422948837280273,
|
| 9727 |
+
"learning_rate": 5.984460178463103e-06,
|
| 9728 |
+
"loss": 0.7816,
|
| 9729 |
+
"step": 1384
|
| 9730 |
+
},
|
| 9731 |
+
{
|
| 9732 |
+
"epoch": 0.004602846619691044,
|
| 9733 |
+
"grad_norm": 2.297849178314209,
|
| 9734 |
+
"learning_rate": 5.915367527945615e-06,
|
| 9735 |
+
"loss": 0.9205,
|
| 9736 |
+
"step": 1385
|
| 9737 |
+
},
|
| 9738 |
+
{
|
| 9739 |
+
"epoch": 0.00460616997465111,
|
| 9740 |
+
"grad_norm": 2.695786714553833,
|
| 9741 |
+
"learning_rate": 5.846663887925607e-06,
|
| 9742 |
+
"loss": 0.8756,
|
| 9743 |
+
"step": 1386
|
| 9744 |
+
},
|
| 9745 |
+
{
|
| 9746 |
+
"epoch": 0.004609493329611176,
|
| 9747 |
+
"grad_norm": 4.079002380371094,
|
| 9748 |
+
"learning_rate": 5.77834954247114e-06,
|
| 9749 |
+
"loss": 0.5973,
|
| 9750 |
+
"step": 1387
|
| 9751 |
+
},
|
| 9752 |
+
{
|
| 9753 |
+
"epoch": 0.004612816684571241,
|
| 9754 |
+
"grad_norm": 2.4570271968841553,
|
| 9755 |
+
"learning_rate": 5.7104247740406346e-06,
|
| 9756 |
+
"loss": 0.8365,
|
| 9757 |
+
"step": 1388
|
| 9758 |
+
},
|
| 9759 |
+
{
|
| 9760 |
+
"epoch": 0.004616140039531307,
|
| 9761 |
+
"grad_norm": 5.311358451843262,
|
| 9762 |
+
"learning_rate": 5.642889863481715e-06,
|
| 9763 |
+
"loss": 0.8901,
|
| 9764 |
+
"step": 1389
|
| 9765 |
+
},
|
| 9766 |
+
{
|
| 9767 |
+
"epoch": 0.004619463394491373,
|
| 9768 |
+
"grad_norm": 1.6130121946334839,
|
| 9769 |
+
"learning_rate": 5.575745090030138e-06,
|
| 9770 |
+
"loss": 0.6247,
|
| 9771 |
+
"step": 1390
|
| 9772 |
+
},
|
| 9773 |
+
{
|
| 9774 |
+
"epoch": 0.004622786749451439,
|
| 9775 |
+
"grad_norm": 1.8931605815887451,
|
| 9776 |
+
"learning_rate": 5.508990731308494e-06,
|
| 9777 |
+
"loss": 1.0028,
|
| 9778 |
+
"step": 1391
|
| 9779 |
+
},
|
| 9780 |
+
{
|
| 9781 |
+
"epoch": 0.004626110104411505,
|
| 9782 |
+
"grad_norm": 1.7890613079071045,
|
| 9783 |
+
"learning_rate": 5.442627063325156e-06,
|
| 9784 |
+
"loss": 0.8613,
|
| 9785 |
+
"step": 1392
|
| 9786 |
+
},
|
| 9787 |
+
{
|
| 9788 |
+
"epoch": 0.00462943345937157,
|
| 9789 |
+
"grad_norm": 2.1536760330200195,
|
| 9790 |
+
"learning_rate": 5.376654360473121e-06,
|
| 9791 |
+
"loss": 1.052,
|
| 9792 |
+
"step": 1393
|
| 9793 |
+
},
|
| 9794 |
+
{
|
| 9795 |
+
"epoch": 0.004632756814331636,
|
| 9796 |
+
"grad_norm": 2.666614532470703,
|
| 9797 |
+
"learning_rate": 5.311072895528857e-06,
|
| 9798 |
+
"loss": 1.1699,
|
| 9799 |
+
"step": 1394
|
| 9800 |
+
},
|
| 9801 |
+
{
|
| 9802 |
+
"epoch": 0.004636080169291702,
|
| 9803 |
+
"grad_norm": 2.1968772411346436,
|
| 9804 |
+
"learning_rate": 5.245882939651181e-06,
|
| 9805 |
+
"loss": 0.8406,
|
| 9806 |
+
"step": 1395
|
| 9807 |
+
},
|
| 9808 |
+
{
|
| 9809 |
+
"epoch": 0.004639403524251767,
|
| 9810 |
+
"grad_norm": 2.414034843444824,
|
| 9811 |
+
"learning_rate": 5.181084762380151e-06,
|
| 9812 |
+
"loss": 0.9435,
|
| 9813 |
+
"step": 1396
|
| 9814 |
+
},
|
| 9815 |
+
{
|
| 9816 |
+
"epoch": 0.004642726879211833,
|
| 9817 |
+
"grad_norm": 2.4396629333496094,
|
| 9818 |
+
"learning_rate": 5.116678631635951e-06,
|
| 9819 |
+
"loss": 0.8793,
|
| 9820 |
+
"step": 1397
|
| 9821 |
+
},
|
| 9822 |
+
{
|
| 9823 |
+
"epoch": 0.0046460502341718985,
|
| 9824 |
+
"grad_norm": 1.8852567672729492,
|
| 9825 |
+
"learning_rate": 5.0526648137177555e-06,
|
| 9826 |
+
"loss": 0.8386,
|
| 9827 |
+
"step": 1398
|
| 9828 |
+
},
|
| 9829 |
+
{
|
| 9830 |
+
"epoch": 0.004649373589131965,
|
| 9831 |
+
"grad_norm": 2.083007335662842,
|
| 9832 |
+
"learning_rate": 4.989043573302654e-06,
|
| 9833 |
+
"loss": 0.9061,
|
| 9834 |
+
"step": 1399
|
| 9835 |
+
},
|
| 9836 |
+
{
|
| 9837 |
+
"epoch": 0.004652696944092031,
|
| 9838 |
+
"grad_norm": 1.9218701124191284,
|
| 9839 |
+
"learning_rate": 4.92581517344457e-06,
|
| 9840 |
+
"loss": 0.8797,
|
| 9841 |
+
"step": 1400
|
| 9842 |
+
},
|
| 9843 |
+
{
|
| 9844 |
+
"epoch": 0.004656020299052096,
|
| 9845 |
+
"grad_norm": 2.787018060684204,
|
| 9846 |
+
"learning_rate": 4.862979875573148e-06,
|
| 9847 |
+
"loss": 0.8159,
|
| 9848 |
+
"step": 1401
|
| 9849 |
+
},
|
| 9850 |
+
{
|
| 9851 |
+
"epoch": 0.004659343654012162,
|
| 9852 |
+
"grad_norm": 3.2939648628234863,
|
| 9853 |
+
"learning_rate": 4.800537939492644e-06,
|
| 9854 |
+
"loss": 0.7294,
|
| 9855 |
+
"step": 1402
|
| 9856 |
+
},
|
| 9857 |
+
{
|
| 9858 |
+
"epoch": 0.004662667008972227,
|
| 9859 |
+
"grad_norm": 3.5180723667144775,
|
| 9860 |
+
"learning_rate": 4.7384896233809266e-06,
|
| 9861 |
+
"loss": 0.7644,
|
| 9862 |
+
"step": 1403
|
| 9863 |
+
},
|
| 9864 |
+
{
|
| 9865 |
+
"epoch": 0.004665990363932293,
|
| 9866 |
+
"grad_norm": 1.954309105873108,
|
| 9867 |
+
"learning_rate": 4.676835183788353e-06,
|
| 9868 |
+
"loss": 0.8668,
|
| 9869 |
+
"step": 1404
|
| 9870 |
+
},
|
| 9871 |
+
{
|
| 9872 |
+
"epoch": 0.004669313718892359,
|
| 9873 |
+
"grad_norm": 1.6614712476730347,
|
| 9874 |
+
"learning_rate": 4.61557487563673e-06,
|
| 9875 |
+
"loss": 0.8414,
|
| 9876 |
+
"step": 1405
|
| 9877 |
+
},
|
| 9878 |
+
{
|
| 9879 |
+
"epoch": 0.0046726370738524245,
|
| 9880 |
+
"grad_norm": 3.9828073978424072,
|
| 9881 |
+
"learning_rate": 4.554708952218267e-06,
|
| 9882 |
+
"loss": 0.9979,
|
| 9883 |
+
"step": 1406
|
| 9884 |
+
},
|
| 9885 |
+
{
|
| 9886 |
+
"epoch": 0.0046759604288124906,
|
| 9887 |
+
"grad_norm": 2.2041354179382324,
|
| 9888 |
+
"learning_rate": 4.494237665194468e-06,
|
| 9889 |
+
"loss": 0.8029,
|
| 9890 |
+
"step": 1407
|
| 9891 |
+
},
|
| 9892 |
+
{
|
| 9893 |
+
"epoch": 0.004679283783772557,
|
| 9894 |
+
"grad_norm": 2.6644821166992188,
|
| 9895 |
+
"learning_rate": 4.434161264595205e-06,
|
| 9896 |
+
"loss": 0.9607,
|
| 9897 |
+
"step": 1408
|
| 9898 |
+
},
|
| 9899 |
+
{
|
| 9900 |
+
"epoch": 0.004682607138732622,
|
| 9901 |
+
"grad_norm": 1.8666707277297974,
|
| 9902 |
+
"learning_rate": 4.37447999881756e-06,
|
| 9903 |
+
"loss": 0.9889,
|
| 9904 |
+
"step": 1409
|
| 9905 |
+
},
|
| 9906 |
+
{
|
| 9907 |
+
"epoch": 0.004685930493692688,
|
| 9908 |
+
"grad_norm": 2.3008713722229004,
|
| 9909 |
+
"learning_rate": 4.315194114624888e-06,
|
| 9910 |
+
"loss": 0.639,
|
| 9911 |
+
"step": 1410
|
| 9912 |
+
},
|
| 9913 |
+
{
|
| 9914 |
+
"epoch": 0.004689253848652753,
|
| 9915 |
+
"grad_norm": 2.3219523429870605,
|
| 9916 |
+
"learning_rate": 4.25630385714576e-06,
|
| 9917 |
+
"loss": 0.8692,
|
| 9918 |
+
"step": 1411
|
| 9919 |
+
},
|
| 9920 |
+
{
|
| 9921 |
+
"epoch": 0.004692577203612819,
|
| 9922 |
+
"grad_norm": 2.4581737518310547,
|
| 9923 |
+
"learning_rate": 4.197809469872949e-06,
|
| 9924 |
+
"loss": 0.8731,
|
| 9925 |
+
"step": 1412
|
| 9926 |
+
},
|
| 9927 |
+
{
|
| 9928 |
+
"epoch": 0.004695900558572885,
|
| 9929 |
+
"grad_norm": 2.2886078357696533,
|
| 9930 |
+
"learning_rate": 4.139711194662432e-06,
|
| 9931 |
+
"loss": 0.8086,
|
| 9932 |
+
"step": 1413
|
| 9933 |
+
},
|
| 9934 |
+
{
|
| 9935 |
+
"epoch": 0.0046992239135329504,
|
| 9936 |
+
"grad_norm": 4.042829513549805,
|
| 9937 |
+
"learning_rate": 4.08200927173239e-06,
|
| 9938 |
+
"loss": 0.6003,
|
| 9939 |
+
"step": 1414
|
| 9940 |
+
},
|
| 9941 |
+
{
|
| 9942 |
+
"epoch": 0.0047025472684930165,
|
| 9943 |
+
"grad_norm": 1.9565678834915161,
|
| 9944 |
+
"learning_rate": 4.0247039396622e-06,
|
| 9945 |
+
"loss": 0.3641,
|
| 9946 |
+
"step": 1415
|
| 9947 |
+
},
|
| 9948 |
+
{
|
| 9949 |
+
"epoch": 0.004705870623453082,
|
| 9950 |
+
"grad_norm": 1.8477730751037598,
|
| 9951 |
+
"learning_rate": 3.967795435391486e-06,
|
| 9952 |
+
"loss": 0.6169,
|
| 9953 |
+
"step": 1416
|
| 9954 |
+
},
|
| 9955 |
+
{
|
| 9956 |
+
"epoch": 0.004709193978413148,
|
| 9957 |
+
"grad_norm": 2.4365956783294678,
|
| 9958 |
+
"learning_rate": 3.911283994219073e-06,
|
| 9959 |
+
"loss": 1.052,
|
| 9960 |
+
"step": 1417
|
| 9961 |
+
},
|
| 9962 |
+
{
|
| 9963 |
+
"epoch": 0.004712517333373214,
|
| 9964 |
+
"grad_norm": 2.995473623275757,
|
| 9965 |
+
"learning_rate": 3.855169849802076e-06,
|
| 9966 |
+
"loss": 0.635,
|
| 9967 |
+
"step": 1418
|
| 9968 |
+
},
|
| 9969 |
+
{
|
| 9970 |
+
"epoch": 0.004715840688333279,
|
| 9971 |
+
"grad_norm": 1.5376818180084229,
|
| 9972 |
+
"learning_rate": 3.7994532341549373e-06,
|
| 9973 |
+
"loss": 1.1172,
|
| 9974 |
+
"step": 1419
|
| 9975 |
+
},
|
| 9976 |
+
{
|
| 9977 |
+
"epoch": 0.004719164043293345,
|
| 9978 |
+
"grad_norm": 3.048156976699829,
|
| 9979 |
+
"learning_rate": 3.7441343776484117e-06,
|
| 9980 |
+
"loss": 0.6616,
|
| 9981 |
+
"step": 1420
|
| 9982 |
+
},
|
| 9983 |
+
{
|
| 9984 |
+
"epoch": 0.004722487398253411,
|
| 9985 |
+
"grad_norm": 2.806403636932373,
|
| 9986 |
+
"learning_rate": 3.6892135090086376e-06,
|
| 9987 |
+
"loss": 0.5784,
|
| 9988 |
+
"step": 1421
|
| 9989 |
+
},
|
| 9990 |
+
{
|
| 9991 |
+
"epoch": 0.004725810753213476,
|
| 9992 |
+
"grad_norm": 2.7289834022521973,
|
| 9993 |
+
"learning_rate": 3.634690855316214e-06,
|
| 9994 |
+
"loss": 0.6859,
|
| 9995 |
+
"step": 1422
|
| 9996 |
+
},
|
| 9997 |
+
{
|
| 9998 |
+
"epoch": 0.0047291341081735425,
|
| 9999 |
+
"grad_norm": 2.284665822982788,
|
| 10000 |
+
"learning_rate": 3.5805666420052452e-06,
|
| 10001 |
+
"loss": 0.6549,
|
| 10002 |
+
"step": 1423
|
| 10003 |
+
},
|
| 10004 |
+
{
|
| 10005 |
+
"epoch": 0.004732457463133608,
|
| 10006 |
+
"grad_norm": 2.156144618988037,
|
| 10007 |
+
"learning_rate": 3.5268410928623984e-06,
|
| 10008 |
+
"loss": 0.7808,
|
| 10009 |
+
"step": 1424
|
| 10010 |
+
},
|
| 10011 |
+
{
|
| 10012 |
+
"epoch": 0.004735780818093674,
|
| 10013 |
+
"grad_norm": 2.6756105422973633,
|
| 10014 |
+
"learning_rate": 3.473514430026026e-06,
|
| 10015 |
+
"loss": 0.6738,
|
| 10016 |
+
"step": 1425
|
| 10017 |
+
},
|
| 10018 |
+
{
|
| 10019 |
+
"epoch": 0.00473910417305374,
|
| 10020 |
+
"grad_norm": 2.9225358963012695,
|
| 10021 |
+
"learning_rate": 3.420586873985132e-06,
|
| 10022 |
+
"loss": 0.8345,
|
| 10023 |
+
"step": 1426
|
| 10024 |
+
},
|
| 10025 |
+
{
|
| 10026 |
+
"epoch": 0.004742427528013805,
|
| 10027 |
+
"grad_norm": 2.8826467990875244,
|
| 10028 |
+
"learning_rate": 3.368058643578642e-06,
|
| 10029 |
+
"loss": 1.0765,
|
| 10030 |
+
"step": 1427
|
| 10031 |
+
},
|
| 10032 |
+
{
|
| 10033 |
+
"epoch": 0.004745750882973871,
|
| 10034 |
+
"grad_norm": 7.946593761444092,
|
| 10035 |
+
"learning_rate": 3.3159299559943237e-06,
|
| 10036 |
+
"loss": 1.0672,
|
| 10037 |
+
"step": 1428
|
| 10038 |
+
},
|
| 10039 |
+
{
|
| 10040 |
+
"epoch": 0.004749074237933936,
|
| 10041 |
+
"grad_norm": 2.426325559616089,
|
| 10042 |
+
"learning_rate": 3.264201026767977e-06,
|
| 10043 |
+
"loss": 0.7406,
|
| 10044 |
+
"step": 1429
|
| 10045 |
+
},
|
| 10046 |
+
{
|
| 10047 |
+
"epoch": 0.004752397592894002,
|
| 10048 |
+
"grad_norm": 3.43176531791687,
|
| 10049 |
+
"learning_rate": 3.212872069782513e-06,
|
| 10050 |
+
"loss": 1.2007,
|
| 10051 |
+
"step": 1430
|
| 10052 |
+
},
|
| 10053 |
+
{
|
| 10054 |
+
"epoch": 0.004755720947854068,
|
| 10055 |
+
"grad_norm": 2.909109592437744,
|
| 10056 |
+
"learning_rate": 3.1619432972671224e-06,
|
| 10057 |
+
"loss": 0.9409,
|
| 10058 |
+
"step": 1431
|
| 10059 |
+
},
|
| 10060 |
+
{
|
| 10061 |
+
"epoch": 0.004759044302814134,
|
| 10062 |
+
"grad_norm": 1.658608078956604,
|
| 10063 |
+
"learning_rate": 3.111414919796318e-06,
|
| 10064 |
+
"loss": 0.9615,
|
| 10065 |
+
"step": 1432
|
| 10066 |
+
},
|
| 10067 |
+
{
|
| 10068 |
+
"epoch": 0.0047623676577742,
|
| 10069 |
+
"grad_norm": 2.9203875064849854,
|
| 10070 |
+
"learning_rate": 3.0612871462891266e-06,
|
| 10071 |
+
"loss": 0.9687,
|
| 10072 |
+
"step": 1433
|
| 10073 |
+
},
|
| 10074 |
+
{
|
| 10075 |
+
"epoch": 0.004765691012734266,
|
| 10076 |
+
"grad_norm": 2.789036512374878,
|
| 10077 |
+
"learning_rate": 3.0115601840081773e-06,
|
| 10078 |
+
"loss": 0.5726,
|
| 10079 |
+
"step": 1434
|
| 10080 |
+
},
|
| 10081 |
+
{
|
| 10082 |
+
"epoch": 0.004769014367694331,
|
| 10083 |
+
"grad_norm": 4.27374267578125,
|
| 10084 |
+
"learning_rate": 2.9622342385589254e-06,
|
| 10085 |
+
"loss": 0.8329,
|
| 10086 |
+
"step": 1435
|
| 10087 |
+
},
|
| 10088 |
+
{
|
| 10089 |
+
"epoch": 0.004772337722654397,
|
| 10090 |
+
"grad_norm": 1.9493528604507446,
|
| 10091 |
+
"learning_rate": 2.9133095138886958e-06,
|
| 10092 |
+
"loss": 1.003,
|
| 10093 |
+
"step": 1436
|
| 10094 |
+
},
|
| 10095 |
+
{
|
| 10096 |
+
"epoch": 0.004775661077614462,
|
| 10097 |
+
"grad_norm": 1.559511661529541,
|
| 10098 |
+
"learning_rate": 2.864786212285897e-06,
|
| 10099 |
+
"loss": 0.9523,
|
| 10100 |
+
"step": 1437
|
| 10101 |
+
},
|
| 10102 |
+
{
|
| 10103 |
+
"epoch": 0.004778984432574528,
|
| 10104 |
+
"grad_norm": 2.0355093479156494,
|
| 10105 |
+
"learning_rate": 2.81666453437921e-06,
|
| 10106 |
+
"loss": 1.2085,
|
| 10107 |
+
"step": 1438
|
| 10108 |
+
},
|
| 10109 |
+
{
|
| 10110 |
+
"epoch": 0.004782307787534594,
|
| 10111 |
+
"grad_norm": 2.2957348823547363,
|
| 10112 |
+
"learning_rate": 2.768944679136676e-06,
|
| 10113 |
+
"loss": 0.5447,
|
| 10114 |
+
"step": 1439
|
| 10115 |
+
},
|
| 10116 |
+
{
|
| 10117 |
+
"epoch": 0.0047856311424946596,
|
| 10118 |
+
"grad_norm": 2.943748712539673,
|
| 10119 |
+
"learning_rate": 2.7216268438649773e-06,
|
| 10120 |
+
"loss": 0.9175,
|
| 10121 |
+
"step": 1440
|
| 10122 |
+
},
|
| 10123 |
+
{
|
| 10124 |
+
"epoch": 0.004788954497454726,
|
| 10125 |
+
"grad_norm": 1.8914774656295776,
|
| 10126 |
+
"learning_rate": 2.674711224208548e-06,
|
| 10127 |
+
"loss": 0.6898,
|
| 10128 |
+
"step": 1441
|
| 10129 |
+
},
|
| 10130 |
+
{
|
| 10131 |
+
"epoch": 0.004792277852414791,
|
| 10132 |
+
"grad_norm": 3.4966208934783936,
|
| 10133 |
+
"learning_rate": 2.628198014148764e-06,
|
| 10134 |
+
"loss": 0.6667,
|
| 10135 |
+
"step": 1442
|
| 10136 |
+
},
|
| 10137 |
+
{
|
| 10138 |
+
"epoch": 0.004795601207374857,
|
| 10139 |
+
"grad_norm": 2.248499631881714,
|
| 10140 |
+
"learning_rate": 2.5820874060032196e-06,
|
| 10141 |
+
"loss": 0.607,
|
| 10142 |
+
"step": 1443
|
| 10143 |
+
},
|
| 10144 |
+
{
|
| 10145 |
+
"epoch": 0.004798924562334923,
|
| 10146 |
+
"grad_norm": 1.6048561334609985,
|
| 10147 |
+
"learning_rate": 2.536379590424809e-06,
|
| 10148 |
+
"loss": 0.9187,
|
| 10149 |
+
"step": 1444
|
| 10150 |
+
},
|
| 10151 |
+
{
|
| 10152 |
+
"epoch": 0.004802247917294988,
|
| 10153 |
+
"grad_norm": 3.612274169921875,
|
| 10154 |
+
"learning_rate": 2.4910747564010685e-06,
|
| 10155 |
+
"loss": 0.8708,
|
| 10156 |
+
"step": 1445
|
| 10157 |
+
},
|
| 10158 |
+
{
|
| 10159 |
+
"epoch": 0.004805571272255054,
|
| 10160 |
+
"grad_norm": 3.096086025238037,
|
| 10161 |
+
"learning_rate": 2.446173091253279e-06,
|
| 10162 |
+
"loss": 0.5083,
|
| 10163 |
+
"step": 1446
|
| 10164 |
+
},
|
| 10165 |
+
{
|
| 10166 |
+
"epoch": 0.00480889462721512,
|
| 10167 |
+
"grad_norm": 1.7813262939453125,
|
| 10168 |
+
"learning_rate": 2.4016747806357654e-06,
|
| 10169 |
+
"loss": 0.8703,
|
| 10170 |
+
"step": 1447
|
| 10171 |
+
},
|
| 10172 |
+
{
|
| 10173 |
+
"epoch": 0.0048122179821751855,
|
| 10174 |
+
"grad_norm": 4.915287494659424,
|
| 10175 |
+
"learning_rate": 2.3575800085351317e-06,
|
| 10176 |
+
"loss": 0.6168,
|
| 10177 |
+
"step": 1448
|
| 10178 |
+
},
|
| 10179 |
+
{
|
| 10180 |
+
"epoch": 0.004815541337135252,
|
| 10181 |
+
"grad_norm": 2.4087512493133545,
|
| 10182 |
+
"learning_rate": 2.3138889572694387e-06,
|
| 10183 |
+
"loss": 0.9247,
|
| 10184 |
+
"step": 1449
|
| 10185 |
+
},
|
| 10186 |
+
{
|
| 10187 |
+
"epoch": 0.004818864692095317,
|
| 10188 |
+
"grad_norm": 2.668449878692627,
|
| 10189 |
+
"learning_rate": 2.2706018074875045e-06,
|
| 10190 |
+
"loss": 0.893,
|
| 10191 |
+
"step": 1450
|
| 10192 |
+
},
|
| 10193 |
+
{
|
| 10194 |
+
"epoch": 0.004822188047055383,
|
| 10195 |
+
"grad_norm": 2.323046922683716,
|
| 10196 |
+
"learning_rate": 2.22771873816815e-06,
|
| 10197 |
+
"loss": 0.774,
|
| 10198 |
+
"step": 1451
|
| 10199 |
+
},
|
| 10200 |
+
{
|
| 10201 |
+
"epoch": 0.004825511402015449,
|
| 10202 |
+
"grad_norm": 2.927000045776367,
|
| 10203 |
+
"learning_rate": 2.1852399266194314e-06,
|
| 10204 |
+
"loss": 0.6814,
|
| 10205 |
+
"step": 1452
|
| 10206 |
+
},
|
| 10207 |
+
{
|
| 10208 |
+
"epoch": 0.004828834756975514,
|
| 10209 |
+
"grad_norm": 2.3023672103881836,
|
| 10210 |
+
"learning_rate": 2.1431655484779435e-06,
|
| 10211 |
+
"loss": 0.6913,
|
| 10212 |
+
"step": 1453
|
| 10213 |
+
},
|
| 10214 |
+
{
|
| 10215 |
+
"epoch": 0.00483215811193558,
|
| 10216 |
+
"grad_norm": 2.670738935470581,
|
| 10217 |
+
"learning_rate": 2.1014957777080513e-06,
|
| 10218 |
+
"loss": 1.0721,
|
| 10219 |
+
"step": 1454
|
| 10220 |
+
},
|
| 10221 |
+
{
|
| 10222 |
+
"epoch": 0.004835481466895645,
|
| 10223 |
+
"grad_norm": 2.3255677223205566,
|
| 10224 |
+
"learning_rate": 2.060230786601225e-06,
|
| 10225 |
+
"loss": 0.7952,
|
| 10226 |
+
"step": 1455
|
| 10227 |
+
},
|
| 10228 |
+
{
|
| 10229 |
+
"epoch": 0.0048388048218557115,
|
| 10230 |
+
"grad_norm": 2.105011224746704,
|
| 10231 |
+
"learning_rate": 2.019370745775273e-06,
|
| 10232 |
+
"loss": 0.5754,
|
| 10233 |
+
"step": 1456
|
| 10234 |
+
},
|
| 10235 |
+
{
|
| 10236 |
+
"epoch": 0.0048421281768157775,
|
| 10237 |
+
"grad_norm": 2.1294243335723877,
|
| 10238 |
+
"learning_rate": 1.978915824173666e-06,
|
| 10239 |
+
"loss": 0.8737,
|
| 10240 |
+
"step": 1457
|
| 10241 |
+
},
|
| 10242 |
+
{
|
| 10243 |
+
"epoch": 0.004845451531775843,
|
| 10244 |
+
"grad_norm": 1.7192840576171875,
|
| 10245 |
+
"learning_rate": 1.9388661890648584e-06,
|
| 10246 |
+
"loss": 0.9259,
|
| 10247 |
+
"step": 1458
|
| 10248 |
+
},
|
| 10249 |
+
{
|
| 10250 |
+
"epoch": 0.004848774886735909,
|
| 10251 |
+
"grad_norm": 2.1691536903381348,
|
| 10252 |
+
"learning_rate": 1.8992220060415344e-06,
|
| 10253 |
+
"loss": 0.7778,
|
| 10254 |
+
"step": 1459
|
| 10255 |
+
},
|
| 10256 |
+
{
|
| 10257 |
+
"epoch": 0.004852098241695975,
|
| 10258 |
+
"grad_norm": 3.173088550567627,
|
| 10259 |
+
"learning_rate": 1.8599834390199855e-06,
|
| 10260 |
+
"loss": 0.7728,
|
| 10261 |
+
"step": 1460
|
| 10262 |
+
},
|
| 10263 |
+
{
|
| 10264 |
+
"epoch": 0.00485542159665604,
|
| 10265 |
+
"grad_norm": 2.2414302825927734,
|
| 10266 |
+
"learning_rate": 1.8211506502394005e-06,
|
| 10267 |
+
"loss": 0.9072,
|
| 10268 |
+
"step": 1461
|
| 10269 |
+
},
|
| 10270 |
+
{
|
| 10271 |
+
"epoch": 0.004858744951616106,
|
| 10272 |
+
"grad_norm": 4.071821212768555,
|
| 10273 |
+
"learning_rate": 1.782723800261199e-06,
|
| 10274 |
+
"loss": 0.9316,
|
| 10275 |
+
"step": 1462
|
| 10276 |
+
},
|
| 10277 |
+
{
|
| 10278 |
+
"epoch": 0.004862068306576171,
|
| 10279 |
+
"grad_norm": 3.314189910888672,
|
| 10280 |
+
"learning_rate": 1.7447030479683767e-06,
|
| 10281 |
+
"loss": 0.6439,
|
| 10282 |
+
"step": 1463
|
| 10283 |
+
},
|
| 10284 |
+
{
|
| 10285 |
+
"epoch": 0.004865391661536237,
|
| 10286 |
+
"grad_norm": 1.6548985242843628,
|
| 10287 |
+
"learning_rate": 1.7070885505648394e-06,
|
| 10288 |
+
"loss": 0.9832,
|
| 10289 |
+
"step": 1464
|
| 10290 |
+
},
|
| 10291 |
+
{
|
| 10292 |
+
"epoch": 0.0048687150164963035,
|
| 10293 |
+
"grad_norm": 2.5467658042907715,
|
| 10294 |
+
"learning_rate": 1.6698804635747579e-06,
|
| 10295 |
+
"loss": 0.6691,
|
| 10296 |
+
"step": 1465
|
| 10297 |
+
},
|
| 10298 |
+
{
|
| 10299 |
+
"epoch": 0.004872038371456369,
|
| 10300 |
+
"grad_norm": 2.0734944343566895,
|
| 10301 |
+
"learning_rate": 1.6330789408419257e-06,
|
| 10302 |
+
"loss": 0.9293,
|
| 10303 |
+
"step": 1466
|
| 10304 |
+
},
|
| 10305 |
+
{
|
| 10306 |
+
"epoch": 0.004875361726416435,
|
| 10307 |
+
"grad_norm": 3.501375913619995,
|
| 10308 |
+
"learning_rate": 1.5966841345291029e-06,
|
| 10309 |
+
"loss": 0.9838,
|
| 10310 |
+
"step": 1467
|
| 10311 |
+
},
|
| 10312 |
+
{
|
| 10313 |
+
"epoch": 0.004878685081376501,
|
| 10314 |
+
"grad_norm": 2.155778169631958,
|
| 10315 |
+
"learning_rate": 1.5606961951174392e-06,
|
| 10316 |
+
"loss": 1.0267,
|
| 10317 |
+
"step": 1468
|
| 10318 |
+
},
|
| 10319 |
+
{
|
| 10320 |
+
"epoch": 0.004882008436336566,
|
| 10321 |
+
"grad_norm": 3.014702796936035,
|
| 10322 |
+
"learning_rate": 1.5251152714057748e-06,
|
| 10323 |
+
"loss": 1.1147,
|
| 10324 |
+
"step": 1469
|
| 10325 |
+
},
|
| 10326 |
+
{
|
| 10327 |
+
"epoch": 0.004885331791296632,
|
| 10328 |
+
"grad_norm": 1.773203730583191,
|
| 10329 |
+
"learning_rate": 1.4899415105101067e-06,
|
| 10330 |
+
"loss": 0.8503,
|
| 10331 |
+
"step": 1470
|
| 10332 |
+
},
|
| 10333 |
+
{
|
| 10334 |
+
"epoch": 0.004888655146256697,
|
| 10335 |
+
"grad_norm": 2.6014420986175537,
|
| 10336 |
+
"learning_rate": 1.4551750578629232e-06,
|
| 10337 |
+
"loss": 0.3187,
|
| 10338 |
+
"step": 1471
|
| 10339 |
+
},
|
| 10340 |
+
{
|
| 10341 |
+
"epoch": 0.004891978501216763,
|
| 10342 |
+
"grad_norm": 2.216736316680908,
|
| 10343 |
+
"learning_rate": 1.420816057212604e-06,
|
| 10344 |
+
"loss": 0.977,
|
| 10345 |
+
"step": 1472
|
| 10346 |
+
},
|
| 10347 |
+
{
|
| 10348 |
+
"epoch": 0.004895301856176829,
|
| 10349 |
+
"grad_norm": 3.4708402156829834,
|
| 10350 |
+
"learning_rate": 1.3868646506228656e-06,
|
| 10351 |
+
"loss": 0.5709,
|
| 10352 |
+
"step": 1473
|
| 10353 |
+
},
|
| 10354 |
+
{
|
| 10355 |
+
"epoch": 0.004898625211136895,
|
| 10356 |
+
"grad_norm": 2.2997395992279053,
|
| 10357 |
+
"learning_rate": 1.35332097847215e-06,
|
| 10358 |
+
"loss": 0.8748,
|
| 10359 |
+
"step": 1474
|
| 10360 |
+
},
|
| 10361 |
+
{
|
| 10362 |
+
"epoch": 0.004901948566096961,
|
| 10363 |
+
"grad_norm": 2.1913869380950928,
|
| 10364 |
+
"learning_rate": 1.3201851794530373e-06,
|
| 10365 |
+
"loss": 0.7726,
|
| 10366 |
+
"step": 1475
|
| 10367 |
+
},
|
| 10368 |
+
{
|
| 10369 |
+
"epoch": 0.004905271921057026,
|
| 10370 |
+
"grad_norm": 2.4012439250946045,
|
| 10371 |
+
"learning_rate": 1.287457390571667e-06,
|
| 10372 |
+
"loss": 1.2023,
|
| 10373 |
+
"step": 1476
|
| 10374 |
+
},
|
| 10375 |
+
{
|
| 10376 |
+
"epoch": 0.004908595276017092,
|
| 10377 |
+
"grad_norm": 1.876934289932251,
|
| 10378 |
+
"learning_rate": 1.2551377471472282e-06,
|
| 10379 |
+
"loss": 0.8011,
|
| 10380 |
+
"step": 1477
|
| 10381 |
+
},
|
| 10382 |
+
{
|
| 10383 |
+
"epoch": 0.004911918630977158,
|
| 10384 |
+
"grad_norm": 2.903610944747925,
|
| 10385 |
+
"learning_rate": 1.2232263828113155e-06,
|
| 10386 |
+
"loss": 0.9427,
|
| 10387 |
+
"step": 1478
|
| 10388 |
+
},
|
| 10389 |
+
{
|
| 10390 |
+
"epoch": 0.004915241985937223,
|
| 10391 |
+
"grad_norm": 2.968428134918213,
|
| 10392 |
+
"learning_rate": 1.1917234295074186e-06,
|
| 10393 |
+
"loss": 0.9079,
|
| 10394 |
+
"step": 1479
|
| 10395 |
+
},
|
| 10396 |
+
{
|
| 10397 |
+
"epoch": 0.004918565340897289,
|
| 10398 |
+
"grad_norm": 2.201148748397827,
|
| 10399 |
+
"learning_rate": 1.160629017490389e-06,
|
| 10400 |
+
"loss": 0.5731,
|
| 10401 |
+
"step": 1480
|
| 10402 |
+
},
|
| 10403 |
+
{
|
| 10404 |
+
"epoch": 0.004921888695857355,
|
| 10405 |
+
"grad_norm": 2.023606538772583,
|
| 10406 |
+
"learning_rate": 1.1299432753258954e-06,
|
| 10407 |
+
"loss": 1.0631,
|
| 10408 |
+
"step": 1481
|
| 10409 |
+
},
|
| 10410 |
+
{
|
| 10411 |
+
"epoch": 0.004925212050817421,
|
| 10412 |
+
"grad_norm": 1.9226667881011963,
|
| 10413 |
+
"learning_rate": 1.0996663298898591e-06,
|
| 10414 |
+
"loss": 0.8471,
|
| 10415 |
+
"step": 1482
|
| 10416 |
+
},
|
| 10417 |
+
{
|
| 10418 |
+
"epoch": 0.004928535405777487,
|
| 10419 |
+
"grad_norm": 3.9702680110931396,
|
| 10420 |
+
"learning_rate": 1.0697983063679752e-06,
|
| 10421 |
+
"loss": 0.7742,
|
| 10422 |
+
"step": 1483
|
| 10423 |
+
},
|
| 10424 |
+
{
|
| 10425 |
+
"epoch": 0.004931858760737552,
|
| 10426 |
+
"grad_norm": 2.1088435649871826,
|
| 10427 |
+
"learning_rate": 1.0403393282551799e-06,
|
| 10428 |
+
"loss": 0.9342,
|
| 10429 |
+
"step": 1484
|
| 10430 |
+
},
|
| 10431 |
+
{
|
| 10432 |
+
"epoch": 0.004935182115697618,
|
| 10433 |
+
"grad_norm": 2.5143256187438965,
|
| 10434 |
+
"learning_rate": 1.0112895173551185e-06,
|
| 10435 |
+
"loss": 1.0462,
|
| 10436 |
+
"step": 1485
|
| 10437 |
+
},
|
| 10438 |
+
{
|
| 10439 |
+
"epoch": 0.004938505470657684,
|
| 10440 |
+
"grad_norm": 1.9809662103652954,
|
| 10441 |
+
"learning_rate": 9.826489937796558e-07,
|
| 10442 |
+
"loss": 0.6669,
|
| 10443 |
+
"step": 1486
|
| 10444 |
+
},
|
| 10445 |
+
{
|
| 10446 |
+
"epoch": 0.004941828825617749,
|
| 10447 |
+
"grad_norm": 1.8066604137420654,
|
| 10448 |
+
"learning_rate": 9.54417875948388e-07,
|
| 10449 |
+
"loss": 1.1573,
|
| 10450 |
+
"step": 1487
|
| 10451 |
+
},
|
| 10452 |
+
{
|
| 10453 |
+
"epoch": 0.004945152180577815,
|
| 10454 |
+
"grad_norm": 2.3810055255889893,
|
| 10455 |
+
"learning_rate": 9.265962805881656e-07,
|
| 10456 |
+
"loss": 0.7143,
|
| 10457 |
+
"step": 1488
|
| 10458 |
+
},
|
| 10459 |
+
{
|
| 10460 |
+
"epoch": 0.0049484755355378805,
|
| 10461 |
+
"grad_norm": 2.1708288192749023,
|
| 10462 |
+
"learning_rate": 8.991843227325491e-07,
|
| 10463 |
+
"loss": 0.8713,
|
| 10464 |
+
"step": 1489
|
| 10465 |
+
},
|
| 10466 |
+
{
|
| 10467 |
+
"epoch": 0.0049517988904979465,
|
| 10468 |
+
"grad_norm": 3.1186017990112305,
|
| 10469 |
+
"learning_rate": 8.721821157214316e-07,
|
| 10470 |
+
"loss": 1.02,
|
| 10471 |
+
"step": 1490
|
| 10472 |
+
},
|
| 10473 |
+
{
|
| 10474 |
+
"epoch": 0.004955122245458013,
|
| 10475 |
+
"grad_norm": 3.0613787174224854,
|
| 10476 |
+
"learning_rate": 8.455897712004612e-07,
|
| 10477 |
+
"loss": 0.9772,
|
| 10478 |
+
"step": 1491
|
| 10479 |
+
},
|
| 10480 |
+
{
|
| 10481 |
+
"epoch": 0.004958445600418078,
|
| 10482 |
+
"grad_norm": 2.7749547958374023,
|
| 10483 |
+
"learning_rate": 8.194073991206641e-07,
|
| 10484 |
+
"loss": 1.0781,
|
| 10485 |
+
"step": 1492
|
| 10486 |
+
},
|
| 10487 |
+
{
|
| 10488 |
+
"epoch": 0.004961768955378144,
|
| 10489 |
+
"grad_norm": 2.8355884552001953,
|
| 10490 |
+
"learning_rate": 7.936351077379555e-07,
|
| 10491 |
+
"loss": 0.9089,
|
| 10492 |
+
"step": 1493
|
| 10493 |
+
},
|
| 10494 |
+
{
|
| 10495 |
+
"epoch": 0.00496509231033821,
|
| 10496 |
+
"grad_norm": 2.191807746887207,
|
| 10497 |
+
"learning_rate": 7.682730036126851e-07,
|
| 10498 |
+
"loss": 0.7164,
|
| 10499 |
+
"step": 1494
|
| 10500 |
+
},
|
| 10501 |
+
{
|
| 10502 |
+
"epoch": 0.004968415665298275,
|
| 10503 |
+
"grad_norm": 2.212839365005493,
|
| 10504 |
+
"learning_rate": 7.433211916092142e-07,
|
| 10505 |
+
"loss": 0.7442,
|
| 10506 |
+
"step": 1495
|
| 10507 |
+
},
|
| 10508 |
+
{
|
| 10509 |
+
"epoch": 0.004971739020258341,
|
| 10510 |
+
"grad_norm": 2.3809967041015625,
|
| 10511 |
+
"learning_rate": 7.187797748954728e-07,
|
| 10512 |
+
"loss": 0.7694,
|
| 10513 |
+
"step": 1496
|
| 10514 |
+
},
|
| 10515 |
+
{
|
| 10516 |
+
"epoch": 0.004975062375218406,
|
| 10517 |
+
"grad_norm": 1.7398457527160645,
|
| 10518 |
+
"learning_rate": 6.946488549425367e-07,
|
| 10519 |
+
"loss": 0.9457,
|
| 10520 |
+
"step": 1497
|
| 10521 |
+
},
|
| 10522 |
+
{
|
| 10523 |
+
"epoch": 0.0049783857301784725,
|
| 10524 |
+
"grad_norm": 2.083955764770508,
|
| 10525 |
+
"learning_rate": 6.709285315242064e-07,
|
| 10526 |
+
"loss": 0.3857,
|
| 10527 |
+
"step": 1498
|
| 10528 |
+
},
|
| 10529 |
+
{
|
| 10530 |
+
"epoch": 0.0049817090851385385,
|
| 10531 |
+
"grad_norm": 3.202343702316284,
|
| 10532 |
+
"learning_rate": 6.476189027165847e-07,
|
| 10533 |
+
"loss": 1.0602,
|
| 10534 |
+
"step": 1499
|
| 10535 |
+
},
|
| 10536 |
+
{
|
| 10537 |
+
"epoch": 0.004985032440098604,
|
| 10538 |
+
"grad_norm": 2.964104175567627,
|
| 10539 |
+
"learning_rate": 6.247200648976991e-07,
|
| 10540 |
+
"loss": 1.0195,
|
| 10541 |
+
"step": 1500
|
| 10542 |
+
},
|
| 10543 |
+
{
|
| 10544 |
+
"epoch": 0.00498835579505867,
|
| 10545 |
+
"grad_norm": 1.5855635404586792,
|
| 10546 |
+
"learning_rate": 6.022321127470698e-07,
|
| 10547 |
+
"loss": 0.8012,
|
| 10548 |
+
"step": 1501
|
| 10549 |
+
},
|
| 10550 |
+
{
|
| 10551 |
+
"epoch": 0.004991679150018735,
|
| 10552 |
+
"grad_norm": 2.3664095401763916,
|
| 10553 |
+
"learning_rate": 5.801551392453308e-07,
|
| 10554 |
+
"loss": 0.8443,
|
| 10555 |
+
"step": 1502
|
| 10556 |
+
},
|
| 10557 |
+
{
|
| 10558 |
+
"epoch": 0.004995002504978801,
|
| 10559 |
+
"grad_norm": 2.0609965324401855,
|
| 10560 |
+
"learning_rate": 5.584892356738647e-07,
|
| 10561 |
+
"loss": 0.8253,
|
| 10562 |
+
"step": 1503
|
| 10563 |
+
},
|
| 10564 |
+
{
|
| 10565 |
+
"epoch": 0.004998325859938867,
|
| 10566 |
+
"grad_norm": 3.14982271194458,
|
| 10567 |
+
"learning_rate": 5.372344916143912e-07,
|
| 10568 |
+
"loss": 0.8079,
|
| 10569 |
+
"step": 1504
|
| 10570 |
+
},
|
| 10571 |
+
{
|
| 10572 |
+
"epoch": 0.005001649214898932,
|
| 10573 |
+
"grad_norm": 2.4054172039031982,
|
| 10574 |
+
"learning_rate": 5.163909949486234e-07,
|
| 10575 |
+
"loss": 0.9403,
|
| 10576 |
+
"step": 1505
|
| 10577 |
+
},
|
| 10578 |
+
{
|
| 10579 |
+
"epoch": 0.005004972569858998,
|
| 10580 |
+
"grad_norm": 1.7414497137069702,
|
| 10581 |
+
"learning_rate": 4.959588318579123e-07,
|
| 10582 |
+
"loss": 0.8446,
|
| 10583 |
+
"step": 1506
|
| 10584 |
+
},
|
| 10585 |
+
{
|
| 10586 |
+
"epoch": 0.0050082959248190645,
|
| 10587 |
+
"grad_norm": 1.4878392219543457,
|
| 10588 |
+
"learning_rate": 4.759380868228247e-07,
|
| 10589 |
+
"loss": 1.1914,
|
| 10590 |
+
"step": 1507
|
| 10591 |
+
},
|
| 10592 |
+
{
|
| 10593 |
+
"epoch": 0.00501161927977913,
|
| 10594 |
+
"grad_norm": 2.7692201137542725,
|
| 10595 |
+
"learning_rate": 4.5632884262289955e-07,
|
| 10596 |
+
"loss": 0.873,
|
| 10597 |
+
"step": 1508
|
| 10598 |
+
},
|
| 10599 |
+
{
|
| 10600 |
+
"epoch": 0.005014942634739196,
|
| 10601 |
+
"grad_norm": 3.191706657409668,
|
| 10602 |
+
"learning_rate": 4.3713118033620325e-07,
|
| 10603 |
+
"loss": 0.8964,
|
| 10604 |
+
"step": 1509
|
| 10605 |
+
},
|
| 10606 |
+
{
|
| 10607 |
+
"epoch": 0.005018265989699261,
|
| 10608 |
+
"grad_norm": 3.778618335723877,
|
| 10609 |
+
"learning_rate": 4.1834517933907467e-07,
|
| 10610 |
+
"loss": 0.5978,
|
| 10611 |
+
"step": 1510
|
| 10612 |
+
},
|
| 10613 |
+
{
|
| 10614 |
+
"epoch": 0.005021589344659327,
|
| 10615 |
+
"grad_norm": 2.869861364364624,
|
| 10616 |
+
"learning_rate": 3.999709173057253e-07,
|
| 10617 |
+
"loss": 0.7505,
|
| 10618 |
+
"step": 1511
|
| 10619 |
+
},
|
| 10620 |
+
{
|
| 10621 |
+
"epoch": 0.005024912699619393,
|
| 10622 |
+
"grad_norm": 1.9169481992721558,
|
| 10623 |
+
"learning_rate": 3.820084702079951e-07,
|
| 10624 |
+
"loss": 0.7455,
|
| 10625 |
+
"step": 1512
|
| 10626 |
+
},
|
| 10627 |
+
{
|
| 10628 |
+
"epoch": 0.005028236054579458,
|
| 10629 |
+
"grad_norm": 2.0523924827575684,
|
| 10630 |
+
"learning_rate": 3.6445791231497494e-07,
|
| 10631 |
+
"loss": 0.9442,
|
| 10632 |
+
"step": 1513
|
| 10633 |
+
},
|
| 10634 |
+
{
|
| 10635 |
+
"epoch": 0.005031559409539524,
|
| 10636 |
+
"grad_norm": 2.7549033164978027,
|
| 10637 |
+
"learning_rate": 3.4731931619271796e-07,
|
| 10638 |
+
"loss": 0.6908,
|
| 10639 |
+
"step": 1514
|
| 10640 |
+
},
|
| 10641 |
+
{
|
| 10642 |
+
"epoch": 0.00503488276449959,
|
| 10643 |
+
"grad_norm": 2.34470272064209,
|
| 10644 |
+
"learning_rate": 3.3059275270396207e-07,
|
| 10645 |
+
"loss": 1.1811,
|
| 10646 |
+
"step": 1515
|
| 10647 |
+
},
|
| 10648 |
+
{
|
| 10649 |
+
"epoch": 0.005038206119459656,
|
| 10650 |
+
"grad_norm": 2.1795027256011963,
|
| 10651 |
+
"learning_rate": 3.142782910077968e-07,
|
| 10652 |
+
"loss": 1.0976,
|
| 10653 |
+
"step": 1516
|
| 10654 |
+
},
|
| 10655 |
+
{
|
| 10656 |
+
"epoch": 0.005041529474419722,
|
| 10657 |
+
"grad_norm": 1.8403738737106323,
|
| 10658 |
+
"learning_rate": 2.983759985594303e-07,
|
| 10659 |
+
"loss": 0.9091,
|
| 10660 |
+
"step": 1517
|
| 10661 |
+
},
|
| 10662 |
+
{
|
| 10663 |
+
"epoch": 0.005044852829379787,
|
| 10664 |
+
"grad_norm": 2.478235960006714,
|
| 10665 |
+
"learning_rate": 2.828859411098561e-07,
|
| 10666 |
+
"loss": 0.9129,
|
| 10667 |
+
"step": 1518
|
| 10668 |
+
},
|
| 10669 |
+
{
|
| 10670 |
+
"epoch": 0.005048176184339853,
|
| 10671 |
+
"grad_norm": 2.183159351348877,
|
| 10672 |
+
"learning_rate": 2.6780818270562004e-07,
|
| 10673 |
+
"loss": 0.9001,
|
| 10674 |
+
"step": 1519
|
| 10675 |
+
},
|
| 10676 |
+
{
|
| 10677 |
+
"epoch": 0.005051499539299919,
|
| 10678 |
+
"grad_norm": 1.7946475744247437,
|
| 10679 |
+
"learning_rate": 2.5314278568850935e-07,
|
| 10680 |
+
"loss": 0.7464,
|
| 10681 |
+
"step": 1520
|
| 10682 |
+
},
|
| 10683 |
+
{
|
| 10684 |
+
"epoch": 0.005054822894259984,
|
| 10685 |
+
"grad_norm": 3.7258291244506836,
|
| 10686 |
+
"learning_rate": 2.388898106953641e-07,
|
| 10687 |
+
"loss": 0.9525,
|
| 10688 |
+
"step": 1521
|
| 10689 |
+
},
|
| 10690 |
+
{
|
| 10691 |
+
"epoch": 0.00505814624922005,
|
| 10692 |
+
"grad_norm": 1.4458050727844238,
|
| 10693 |
+
"learning_rate": 2.2504931665777717e-07,
|
| 10694 |
+
"loss": 0.7825,
|
| 10695 |
+
"step": 1522
|
| 10696 |
+
},
|
| 10697 |
+
{
|
| 10698 |
+
"epoch": 0.0050614696041801155,
|
| 10699 |
+
"grad_norm": 2.4851584434509277,
|
| 10700 |
+
"learning_rate": 2.1162136080185025e-07,
|
| 10701 |
+
"loss": 1.0679,
|
| 10702 |
+
"step": 1523
|
| 10703 |
+
},
|
| 10704 |
+
{
|
| 10705 |
+
"epoch": 0.005064792959140182,
|
| 10706 |
+
"grad_norm": 2.65813946723938,
|
| 10707 |
+
"learning_rate": 1.9860599864798268e-07,
|
| 10708 |
+
"loss": 0.6658,
|
| 10709 |
+
"step": 1524
|
| 10710 |
+
},
|
| 10711 |
+
{
|
| 10712 |
+
"epoch": 0.005068116314100248,
|
| 10713 |
+
"grad_norm": 3.4035613536834717,
|
| 10714 |
+
"learning_rate": 1.8600328401061629e-07,
|
| 10715 |
+
"loss": 1.1345,
|
| 10716 |
+
"step": 1525
|
| 10717 |
+
},
|
| 10718 |
+
{
|
| 10719 |
+
"epoch": 0.005071439669060313,
|
| 10720 |
+
"grad_norm": 1.971177577972412,
|
| 10721 |
+
"learning_rate": 1.738132689980354e-07,
|
| 10722 |
+
"loss": 0.4085,
|
| 10723 |
+
"step": 1526
|
| 10724 |
+
},
|
| 10725 |
+
{
|
| 10726 |
+
"epoch": 0.005074763024020379,
|
| 10727 |
+
"grad_norm": 2.831120252609253,
|
| 10728 |
+
"learning_rate": 1.620360040121227e-07,
|
| 10729 |
+
"loss": 0.9349,
|
| 10730 |
+
"step": 1527
|
| 10731 |
+
},
|
| 10732 |
+
{
|
| 10733 |
+
"epoch": 0.005078086378980444,
|
| 10734 |
+
"grad_norm": 2.3056507110595703,
|
| 10735 |
+
"learning_rate": 1.5067153774820374e-07,
|
| 10736 |
+
"loss": 0.8728,
|
| 10737 |
+
"step": 1528
|
| 10738 |
+
},
|
| 10739 |
+
{
|
| 10740 |
+
"epoch": 0.00508140973394051,
|
| 10741 |
+
"grad_norm": 2.3262298107147217,
|
| 10742 |
+
"learning_rate": 1.397199171947694e-07,
|
| 10743 |
+
"loss": 1.0368,
|
| 10744 |
+
"step": 1529
|
| 10745 |
+
},
|
| 10746 |
+
{
|
| 10747 |
+
"epoch": 0.005084733088900576,
|
| 10748 |
+
"grad_norm": 3.811479330062866,
|
| 10749 |
+
"learning_rate": 1.2918118763335373e-07,
|
| 10750 |
+
"loss": 0.6753,
|
| 10751 |
+
"step": 1530
|
| 10752 |
+
},
|
| 10753 |
+
{
|
| 10754 |
+
"epoch": 0.0050880564438606415,
|
| 10755 |
+
"grad_norm": 1.9728143215179443,
|
| 10756 |
+
"learning_rate": 1.190553926382898e-07,
|
| 10757 |
+
"loss": 0.7752,
|
| 10758 |
+
"step": 1531
|
| 10759 |
+
},
|
| 10760 |
+
{
|
| 10761 |
+
"epoch": 0.0050913797988207075,
|
| 10762 |
+
"grad_norm": 1.8086570501327515,
|
| 10763 |
+
"learning_rate": 1.0934257407657634e-07,
|
| 10764 |
+
"loss": 0.7167,
|
| 10765 |
+
"step": 1532
|
| 10766 |
+
},
|
| 10767 |
+
{
|
| 10768 |
+
"epoch": 0.005094703153780774,
|
| 10769 |
+
"grad_norm": 2.604548931121826,
|
| 10770 |
+
"learning_rate": 1.0004277210768908e-07,
|
| 10771 |
+
"loss": 0.7697,
|
| 10772 |
+
"step": 1533
|
| 10773 |
+
},
|
| 10774 |
+
{
|
| 10775 |
+
"epoch": 0.005098026508740839,
|
| 10776 |
+
"grad_norm": 2.7422561645507812,
|
| 10777 |
+
"learning_rate": 9.115602518338095e-08,
|
| 10778 |
+
"loss": 1.0211,
|
| 10779 |
+
"step": 1534
|
| 10780 |
+
},
|
| 10781 |
+
{
|
| 10782 |
+
"epoch": 0.005101349863700905,
|
| 10783 |
+
"grad_norm": 3.702219009399414,
|
| 10784 |
+
"learning_rate": 8.268237004757096e-08,
|
| 10785 |
+
"loss": 1.1697,
|
| 10786 |
+
"step": 1535
|
| 10787 |
+
},
|
| 10788 |
+
{
|
| 10789 |
+
"epoch": 0.00510467321866097,
|
| 10790 |
+
"grad_norm": 2.1180949211120605,
|
| 10791 |
+
"learning_rate": 7.46218417361777e-08,
|
| 10792 |
+
"loss": 1.1422,
|
| 10793 |
+
"step": 1536
|
| 10794 |
+
},
|
| 10795 |
+
{
|
| 10796 |
+
"epoch": 0.005107996573621036,
|
| 10797 |
+
"grad_norm": 2.297656536102295,
|
| 10798 |
+
"learning_rate": 6.697447357695286e-08,
|
| 10799 |
+
"loss": 1.1983,
|
| 10800 |
+
"step": 1537
|
| 10801 |
+
},
|
| 10802 |
+
{
|
| 10803 |
+
"epoch": 0.005111319928581102,
|
| 10804 |
+
"grad_norm": 2.9963088035583496,
|
| 10805 |
+
"learning_rate": 5.974029718938124e-08,
|
| 10806 |
+
"loss": 0.8987,
|
| 10807 |
+
"step": 1538
|
| 10808 |
+
},
|
| 10809 |
+
{
|
| 10810 |
+
"epoch": 0.005114643283541167,
|
| 10811 |
+
"grad_norm": 4.024659633636475,
|
| 10812 |
+
"learning_rate": 5.2919342484503146e-08,
|
| 10813 |
+
"loss": 0.6689,
|
| 10814 |
+
"step": 1539
|
| 10815 |
+
},
|
| 10816 |
+
{
|
| 10817 |
+
"epoch": 0.0051179666385012335,
|
| 10818 |
+
"grad_norm": 2.6762568950653076,
|
| 10819 |
+
"learning_rate": 4.651163766484779e-08,
|
| 10820 |
+
"loss": 1.0369,
|
| 10821 |
+
"step": 1540
|
| 10822 |
+
},
|
| 10823 |
+
{
|
| 10824 |
+
"epoch": 0.005121289993461299,
|
| 10825 |
+
"grad_norm": 2.905095338821411,
|
| 10826 |
+
"learning_rate": 4.051720922424451e-08,
|
| 10827 |
+
"loss": 0.5759,
|
| 10828 |
+
"step": 1541
|
| 10829 |
+
},
|
| 10830 |
+
{
|
| 10831 |
+
"epoch": 0.005124613348421365,
|
| 10832 |
+
"grad_norm": 2.7683260440826416,
|
| 10833 |
+
"learning_rate": 3.4936081947800624e-08,
|
| 10834 |
+
"loss": 0.7106,
|
| 10835 |
+
"step": 1542
|
| 10836 |
+
},
|
| 10837 |
+
{
|
| 10838 |
+
"epoch": 0.005127936703381431,
|
| 10839 |
+
"grad_norm": 1.6640328168869019,
|
| 10840 |
+
"learning_rate": 2.9768278911723734e-08,
|
| 10841 |
+
"loss": 0.8466,
|
| 10842 |
+
"step": 1543
|
| 10843 |
+
},
|
| 10844 |
+
{
|
| 10845 |
+
"epoch": 0.005131260058341496,
|
| 10846 |
+
"grad_norm": 2.6602017879486084,
|
| 10847 |
+
"learning_rate": 2.501382148324405e-08,
|
| 10848 |
+
"loss": 0.3438,
|
| 10849 |
+
"step": 1544
|
| 10850 |
+
},
|
| 10851 |
+
{
|
| 10852 |
+
"epoch": 0.005134583413301562,
|
| 10853 |
+
"grad_norm": 2.2989859580993652,
|
| 10854 |
+
"learning_rate": 2.0672729320581065e-08,
|
| 10855 |
+
"loss": 0.8661,
|
| 10856 |
+
"step": 1545
|
| 10857 |
+
},
|
| 10858 |
+
{
|
| 10859 |
+
"epoch": 0.005137906768261628,
|
| 10860 |
+
"grad_norm": 2.373389720916748,
|
| 10861 |
+
"learning_rate": 1.674502037277703e-08,
|
| 10862 |
+
"loss": 1.1053,
|
| 10863 |
+
"step": 1546
|
| 10864 |
+
},
|
| 10865 |
+
{
|
| 10866 |
+
"epoch": 0.005141230123221693,
|
| 10867 |
+
"grad_norm": 3.106809616088867,
|
| 10868 |
+
"learning_rate": 1.3230710879674756e-08,
|
| 10869 |
+
"loss": 0.765,
|
| 10870 |
+
"step": 1547
|
| 10871 |
+
},
|
| 10872 |
+
{
|
| 10873 |
+
"epoch": 0.0051445534781817594,
|
| 10874 |
+
"grad_norm": 2.3349437713623047,
|
| 10875 |
+
"learning_rate": 1.0129815371862084e-08,
|
| 10876 |
+
"loss": 0.645,
|
| 10877 |
+
"step": 1548
|
| 10878 |
+
},
|
| 10879 |
+
{
|
| 10880 |
+
"epoch": 0.005147876833141825,
|
| 10881 |
+
"grad_norm": 1.8190383911132812,
|
| 10882 |
+
"learning_rate": 7.4423466705497714e-09,
|
| 10883 |
+
"loss": 1.3087,
|
| 10884 |
+
"step": 1549
|
| 10885 |
+
},
|
| 10886 |
+
{
|
| 10887 |
+
"epoch": 0.005151200188101891,
|
| 10888 |
+
"grad_norm": 2.419950485229492,
|
| 10889 |
+
"learning_rate": 5.1683158875937e-09,
|
| 10890 |
+
"loss": 0.5226,
|
| 10891 |
+
"step": 1550
|
| 10892 |
+
},
|
| 10893 |
+
{
|
| 10894 |
+
"epoch": 0.005154523543061957,
|
| 10895 |
+
"grad_norm": 2.1594996452331543,
|
| 10896 |
+
"learning_rate": 3.307732425383847e-09,
|
| 10897 |
+
"loss": 1.1091,
|
| 10898 |
+
"step": 1551
|
| 10899 |
+
},
|
| 10900 |
+
{
|
| 10901 |
+
"epoch": 0.005157846898022022,
|
| 10902 |
+
"grad_norm": 1.712402582168579,
|
| 10903 |
+
"learning_rate": 1.8606039768775951e-09,
|
| 10904 |
+
"loss": 1.2234,
|
| 10905 |
+
"step": 1552
|
| 10906 |
+
},
|
| 10907 |
+
{
|
| 10908 |
+
"epoch": 0.005161170252982088,
|
| 10909 |
+
"grad_norm": 3.5046451091766357,
|
| 10910 |
+
"learning_rate": 8.26936525466504e-10,
|
| 10911 |
+
"loss": 1.0062,
|
| 10912 |
+
"step": 1553
|
| 10913 |
+
},
|
| 10914 |
+
{
|
| 10915 |
+
"epoch": 0.005164493607942153,
|
| 10916 |
+
"grad_norm": 1.645281195640564,
|
| 10917 |
+
"learning_rate": 2.0673434506512935e-10,
|
| 10918 |
+
"loss": 0.7931,
|
| 10919 |
+
"step": 1554
|
| 10920 |
+
},
|
| 10921 |
+
{
|
| 10922 |
+
"epoch": 0.005167816962902219,
|
| 10923 |
+
"grad_norm": 2.038417339324951,
|
| 10924 |
+
"learning_rate": 0.0,
|
| 10925 |
+
"loss": 0.8836,
|
| 10926 |
+
"step": 1555
|
| 10927 |
}
|
| 10928 |
],
|
| 10929 |
"logging_steps": 1,
|
|
|
|
| 10938 |
"should_evaluate": false,
|
| 10939 |
"should_log": false,
|
| 10940 |
"should_save": true,
|
| 10941 |
+
"should_training_stop": true
|
| 10942 |
},
|
| 10943 |
"attributes": {}
|
| 10944 |
}
|
| 10945 |
},
|
| 10946 |
+
"total_flos": 1.184235110006784e+17,
|
| 10947 |
"train_batch_size": 2,
|
| 10948 |
"trial_name": null,
|
| 10949 |
"trial_params": null
|