Training in progress, step 1598, 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 80013120
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6407ab1a98f357bb2f56078b41d18c55d4e32be183deab593de741a109d194f3
|
| 3 |
size 80013120
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 41120084
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a892b456fb6306016c02bc81c8cf10c08d42ffbb557543c7c944ad1f57e0f4c2
|
| 3 |
size 41120084
|
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 sha256:46266452dabee304c03ff26394dee6b60227d26e47caca8f6aff1080b24d1c26
|
| 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:c325a1545638f21bf398668656d35f0eaa4863b2daa719b4b45455e3cbd688e2
|
| 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": 500,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -8407,6 +8407,2792 @@
|
|
| 8407 |
"learning_rate": 1.4627233965158744e-05,
|
| 8408 |
"loss": 2.2994,
|
| 8409 |
"step": 1200
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 8410 |
}
|
| 8411 |
],
|
| 8412 |
"logging_steps": 1,
|
|
@@ -8421,12 +11207,12 @@
|
|
| 8421 |
"should_evaluate": false,
|
| 8422 |
"should_log": false,
|
| 8423 |
"should_save": true,
|
| 8424 |
-
"should_training_stop":
|
| 8425 |
},
|
| 8426 |
"attributes": {}
|
| 8427 |
}
|
| 8428 |
},
|
| 8429 |
-
"total_flos":
|
| 8430 |
"train_batch_size": 4,
|
| 8431 |
"trial_name": null,
|
| 8432 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.6656946469485524,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 1598,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 8407 |
"learning_rate": 1.4627233965158744e-05,
|
| 8408 |
"loss": 2.2994,
|
| 8409 |
"step": 1200
|
| 8410 |
+
},
|
| 8411 |
+
{
|
| 8412 |
+
"epoch": 0.5003124349093939,
|
| 8413 |
+
"grad_norm": 0.35549497604370117,
|
| 8414 |
+
"learning_rate": 1.4557612089349854e-05,
|
| 8415 |
+
"loss": 2.1406,
|
| 8416 |
+
"step": 1201
|
| 8417 |
+
},
|
| 8418 |
+
{
|
| 8419 |
+
"epoch": 0.5007290147885857,
|
| 8420 |
+
"grad_norm": 0.3743707239627838,
|
| 8421 |
+
"learning_rate": 1.4488128058544098e-05,
|
| 8422 |
+
"loss": 2.2135,
|
| 8423 |
+
"step": 1202
|
| 8424 |
+
},
|
| 8425 |
+
{
|
| 8426 |
+
"epoch": 0.5011455946677775,
|
| 8427 |
+
"grad_norm": 0.37623918056488037,
|
| 8428 |
+
"learning_rate": 1.4418782142983589e-05,
|
| 8429 |
+
"loss": 2.123,
|
| 8430 |
+
"step": 1203
|
| 8431 |
+
},
|
| 8432 |
+
{
|
| 8433 |
+
"epoch": 0.5015621745469694,
|
| 8434 |
+
"grad_norm": 0.38898077607154846,
|
| 8435 |
+
"learning_rate": 1.4349574612373312e-05,
|
| 8436 |
+
"loss": 2.2361,
|
| 8437 |
+
"step": 1204
|
| 8438 |
+
},
|
| 8439 |
+
{
|
| 8440 |
+
"epoch": 0.5019787544261612,
|
| 8441 |
+
"grad_norm": 0.36593276262283325,
|
| 8442 |
+
"learning_rate": 1.428050573587999e-05,
|
| 8443 |
+
"loss": 2.1007,
|
| 8444 |
+
"step": 1205
|
| 8445 |
+
},
|
| 8446 |
+
{
|
| 8447 |
+
"epoch": 0.5023953343053531,
|
| 8448 |
+
"grad_norm": 0.40702638030052185,
|
| 8449 |
+
"learning_rate": 1.4211575782131131e-05,
|
| 8450 |
+
"loss": 2.1984,
|
| 8451 |
+
"step": 1206
|
| 8452 |
+
},
|
| 8453 |
+
{
|
| 8454 |
+
"epoch": 0.5028119141845449,
|
| 8455 |
+
"grad_norm": 0.41245660185813904,
|
| 8456 |
+
"learning_rate": 1.4142785019213906e-05,
|
| 8457 |
+
"loss": 2.1878,
|
| 8458 |
+
"step": 1207
|
| 8459 |
+
},
|
| 8460 |
+
{
|
| 8461 |
+
"epoch": 0.5032284940637367,
|
| 8462 |
+
"grad_norm": 0.40169423818588257,
|
| 8463 |
+
"learning_rate": 1.4074133714674115e-05,
|
| 8464 |
+
"loss": 2.1697,
|
| 8465 |
+
"step": 1208
|
| 8466 |
+
},
|
| 8467 |
+
{
|
| 8468 |
+
"epoch": 0.5036450739429286,
|
| 8469 |
+
"grad_norm": 0.43156102299690247,
|
| 8470 |
+
"learning_rate": 1.4005622135515223e-05,
|
| 8471 |
+
"loss": 2.151,
|
| 8472 |
+
"step": 1209
|
| 8473 |
+
},
|
| 8474 |
+
{
|
| 8475 |
+
"epoch": 0.5040616538221204,
|
| 8476 |
+
"grad_norm": 0.41381898522377014,
|
| 8477 |
+
"learning_rate": 1.3937250548197212e-05,
|
| 8478 |
+
"loss": 2.1351,
|
| 8479 |
+
"step": 1210
|
| 8480 |
+
},
|
| 8481 |
+
{
|
| 8482 |
+
"epoch": 0.5044782337013122,
|
| 8483 |
+
"grad_norm": 0.4139593541622162,
|
| 8484 |
+
"learning_rate": 1.3869019218635642e-05,
|
| 8485 |
+
"loss": 2.23,
|
| 8486 |
+
"step": 1211
|
| 8487 |
+
},
|
| 8488 |
+
{
|
| 8489 |
+
"epoch": 0.504894813580504,
|
| 8490 |
+
"grad_norm": 0.408181756734848,
|
| 8491 |
+
"learning_rate": 1.3800928412200547e-05,
|
| 8492 |
+
"loss": 2.0367,
|
| 8493 |
+
"step": 1212
|
| 8494 |
+
},
|
| 8495 |
+
{
|
| 8496 |
+
"epoch": 0.5053113934596959,
|
| 8497 |
+
"grad_norm": 0.4455416798591614,
|
| 8498 |
+
"learning_rate": 1.3732978393715407e-05,
|
| 8499 |
+
"loss": 2.1086,
|
| 8500 |
+
"step": 1213
|
| 8501 |
+
},
|
| 8502 |
+
{
|
| 8503 |
+
"epoch": 0.5057279733388877,
|
| 8504 |
+
"grad_norm": 0.41871967911720276,
|
| 8505 |
+
"learning_rate": 1.3665169427456203e-05,
|
| 8506 |
+
"loss": 2.1236,
|
| 8507 |
+
"step": 1214
|
| 8508 |
+
},
|
| 8509 |
+
{
|
| 8510 |
+
"epoch": 0.5061445532180796,
|
| 8511 |
+
"grad_norm": 0.4171868562698364,
|
| 8512 |
+
"learning_rate": 1.3597501777150252e-05,
|
| 8513 |
+
"loss": 2.3884,
|
| 8514 |
+
"step": 1215
|
| 8515 |
+
},
|
| 8516 |
+
{
|
| 8517 |
+
"epoch": 0.5065611330972714,
|
| 8518 |
+
"grad_norm": 0.4312697947025299,
|
| 8519 |
+
"learning_rate": 1.3529975705975328e-05,
|
| 8520 |
+
"loss": 2.1986,
|
| 8521 |
+
"step": 1216
|
| 8522 |
+
},
|
| 8523 |
+
{
|
| 8524 |
+
"epoch": 0.5069777129764632,
|
| 8525 |
+
"grad_norm": 0.4134281575679779,
|
| 8526 |
+
"learning_rate": 1.3462591476558511e-05,
|
| 8527 |
+
"loss": 2.0586,
|
| 8528 |
+
"step": 1217
|
| 8529 |
+
},
|
| 8530 |
+
{
|
| 8531 |
+
"epoch": 0.5073942928556551,
|
| 8532 |
+
"grad_norm": 0.39930659532546997,
|
| 8533 |
+
"learning_rate": 1.339534935097524e-05,
|
| 8534 |
+
"loss": 2.2032,
|
| 8535 |
+
"step": 1218
|
| 8536 |
+
},
|
| 8537 |
+
{
|
| 8538 |
+
"epoch": 0.5078108727348469,
|
| 8539 |
+
"grad_norm": 0.4186604917049408,
|
| 8540 |
+
"learning_rate": 1.3328249590748255e-05,
|
| 8541 |
+
"loss": 2.1334,
|
| 8542 |
+
"step": 1219
|
| 8543 |
+
},
|
| 8544 |
+
{
|
| 8545 |
+
"epoch": 0.5082274526140388,
|
| 8546 |
+
"grad_norm": 0.4806564152240753,
|
| 8547 |
+
"learning_rate": 1.3261292456846647e-05,
|
| 8548 |
+
"loss": 2.1565,
|
| 8549 |
+
"step": 1220
|
| 8550 |
+
},
|
| 8551 |
+
{
|
| 8552 |
+
"epoch": 0.5086440324932305,
|
| 8553 |
+
"grad_norm": 0.4354822039604187,
|
| 8554 |
+
"learning_rate": 1.3194478209684746e-05,
|
| 8555 |
+
"loss": 2.1295,
|
| 8556 |
+
"step": 1221
|
| 8557 |
+
},
|
| 8558 |
+
{
|
| 8559 |
+
"epoch": 0.5090606123724224,
|
| 8560 |
+
"grad_norm": 0.4317784011363983,
|
| 8561 |
+
"learning_rate": 1.3127807109121165e-05,
|
| 8562 |
+
"loss": 2.2652,
|
| 8563 |
+
"step": 1222
|
| 8564 |
+
},
|
| 8565 |
+
{
|
| 8566 |
+
"epoch": 0.5094771922516143,
|
| 8567 |
+
"grad_norm": 0.46810969710350037,
|
| 8568 |
+
"learning_rate": 1.3061279414457823e-05,
|
| 8569 |
+
"loss": 2.2531,
|
| 8570 |
+
"step": 1223
|
| 8571 |
+
},
|
| 8572 |
+
{
|
| 8573 |
+
"epoch": 0.5098937721308061,
|
| 8574 |
+
"grad_norm": 0.44056519865989685,
|
| 8575 |
+
"learning_rate": 1.2994895384438827e-05,
|
| 8576 |
+
"loss": 2.1596,
|
| 8577 |
+
"step": 1224
|
| 8578 |
+
},
|
| 8579 |
+
{
|
| 8580 |
+
"epoch": 0.510310352009998,
|
| 8581 |
+
"grad_norm": 0.43828773498535156,
|
| 8582 |
+
"learning_rate": 1.2928655277249597e-05,
|
| 8583 |
+
"loss": 2.2656,
|
| 8584 |
+
"step": 1225
|
| 8585 |
+
},
|
| 8586 |
+
{
|
| 8587 |
+
"epoch": 0.5107269318891897,
|
| 8588 |
+
"grad_norm": 0.42964041233062744,
|
| 8589 |
+
"learning_rate": 1.2862559350515746e-05,
|
| 8590 |
+
"loss": 2.0963,
|
| 8591 |
+
"step": 1226
|
| 8592 |
+
},
|
| 8593 |
+
{
|
| 8594 |
+
"epoch": 0.5111435117683816,
|
| 8595 |
+
"grad_norm": 0.43426331877708435,
|
| 8596 |
+
"learning_rate": 1.2796607861302184e-05,
|
| 8597 |
+
"loss": 2.1752,
|
| 8598 |
+
"step": 1227
|
| 8599 |
+
},
|
| 8600 |
+
{
|
| 8601 |
+
"epoch": 0.5115600916475734,
|
| 8602 |
+
"grad_norm": 0.4176805019378662,
|
| 8603 |
+
"learning_rate": 1.2730801066112024e-05,
|
| 8604 |
+
"loss": 2.1394,
|
| 8605 |
+
"step": 1228
|
| 8606 |
+
},
|
| 8607 |
+
{
|
| 8608 |
+
"epoch": 0.5119766715267653,
|
| 8609 |
+
"grad_norm": 0.4687243700027466,
|
| 8610 |
+
"learning_rate": 1.2665139220885613e-05,
|
| 8611 |
+
"loss": 2.0872,
|
| 8612 |
+
"step": 1229
|
| 8613 |
+
},
|
| 8614 |
+
{
|
| 8615 |
+
"epoch": 0.512393251405957,
|
| 8616 |
+
"grad_norm": 0.45004138350486755,
|
| 8617 |
+
"learning_rate": 1.2599622580999605e-05,
|
| 8618 |
+
"loss": 2.2393,
|
| 8619 |
+
"step": 1230
|
| 8620 |
+
},
|
| 8621 |
+
{
|
| 8622 |
+
"epoch": 0.5128098312851489,
|
| 8623 |
+
"grad_norm": 0.4343072772026062,
|
| 8624 |
+
"learning_rate": 1.253425140126585e-05,
|
| 8625 |
+
"loss": 2.2669,
|
| 8626 |
+
"step": 1231
|
| 8627 |
+
},
|
| 8628 |
+
{
|
| 8629 |
+
"epoch": 0.5132264111643408,
|
| 8630 |
+
"grad_norm": 0.4603501856327057,
|
| 8631 |
+
"learning_rate": 1.2469025935930523e-05,
|
| 8632 |
+
"loss": 2.1012,
|
| 8633 |
+
"step": 1232
|
| 8634 |
+
},
|
| 8635 |
+
{
|
| 8636 |
+
"epoch": 0.5136429910435326,
|
| 8637 |
+
"grad_norm": 0.46814849972724915,
|
| 8638 |
+
"learning_rate": 1.2403946438673031e-05,
|
| 8639 |
+
"loss": 2.1392,
|
| 8640 |
+
"step": 1233
|
| 8641 |
+
},
|
| 8642 |
+
{
|
| 8643 |
+
"epoch": 0.5140595709227245,
|
| 8644 |
+
"grad_norm": 0.44013842940330505,
|
| 8645 |
+
"learning_rate": 1.2339013162605078e-05,
|
| 8646 |
+
"loss": 2.0884,
|
| 8647 |
+
"step": 1234
|
| 8648 |
+
},
|
| 8649 |
+
{
|
| 8650 |
+
"epoch": 0.5144761508019162,
|
| 8651 |
+
"grad_norm": 0.5072870850563049,
|
| 8652 |
+
"learning_rate": 1.2274226360269687e-05,
|
| 8653 |
+
"loss": 2.0433,
|
| 8654 |
+
"step": 1235
|
| 8655 |
+
},
|
| 8656 |
+
{
|
| 8657 |
+
"epoch": 0.5148927306811081,
|
| 8658 |
+
"grad_norm": 0.44100427627563477,
|
| 8659 |
+
"learning_rate": 1.2209586283640206e-05,
|
| 8660 |
+
"loss": 2.2232,
|
| 8661 |
+
"step": 1236
|
| 8662 |
+
},
|
| 8663 |
+
{
|
| 8664 |
+
"epoch": 0.5153093105602999,
|
| 8665 |
+
"grad_norm": 0.4575675129890442,
|
| 8666 |
+
"learning_rate": 1.2145093184119327e-05,
|
| 8667 |
+
"loss": 2.1936,
|
| 8668 |
+
"step": 1237
|
| 8669 |
+
},
|
| 8670 |
+
{
|
| 8671 |
+
"epoch": 0.5157258904394918,
|
| 8672 |
+
"grad_norm": 0.4817796051502228,
|
| 8673 |
+
"learning_rate": 1.2080747312538083e-05,
|
| 8674 |
+
"loss": 2.2495,
|
| 8675 |
+
"step": 1238
|
| 8676 |
+
},
|
| 8677 |
+
{
|
| 8678 |
+
"epoch": 0.5161424703186837,
|
| 8679 |
+
"grad_norm": 0.5097609162330627,
|
| 8680 |
+
"learning_rate": 1.2016548919154947e-05,
|
| 8681 |
+
"loss": 2.0267,
|
| 8682 |
+
"step": 1239
|
| 8683 |
+
},
|
| 8684 |
+
{
|
| 8685 |
+
"epoch": 0.5165590501978754,
|
| 8686 |
+
"grad_norm": 0.5130468606948853,
|
| 8687 |
+
"learning_rate": 1.1952498253654754e-05,
|
| 8688 |
+
"loss": 2.1539,
|
| 8689 |
+
"step": 1240
|
| 8690 |
+
},
|
| 8691 |
+
{
|
| 8692 |
+
"epoch": 0.5169756300770673,
|
| 8693 |
+
"grad_norm": 0.47017228603363037,
|
| 8694 |
+
"learning_rate": 1.1888595565147836e-05,
|
| 8695 |
+
"loss": 2.231,
|
| 8696 |
+
"step": 1241
|
| 8697 |
+
},
|
| 8698 |
+
{
|
| 8699 |
+
"epoch": 0.5173922099562591,
|
| 8700 |
+
"grad_norm": 0.49151870608329773,
|
| 8701 |
+
"learning_rate": 1.1824841102168965e-05,
|
| 8702 |
+
"loss": 2.202,
|
| 8703 |
+
"step": 1242
|
| 8704 |
+
},
|
| 8705 |
+
{
|
| 8706 |
+
"epoch": 0.517808789835451,
|
| 8707 |
+
"grad_norm": 0.5390906929969788,
|
| 8708 |
+
"learning_rate": 1.1761235112676433e-05,
|
| 8709 |
+
"loss": 2.1915,
|
| 8710 |
+
"step": 1243
|
| 8711 |
+
},
|
| 8712 |
+
{
|
| 8713 |
+
"epoch": 0.5182253697146428,
|
| 8714 |
+
"grad_norm": 0.5682397484779358,
|
| 8715 |
+
"learning_rate": 1.1697777844051105e-05,
|
| 8716 |
+
"loss": 2.3685,
|
| 8717 |
+
"step": 1244
|
| 8718 |
+
},
|
| 8719 |
+
{
|
| 8720 |
+
"epoch": 0.5186419495938346,
|
| 8721 |
+
"grad_norm": 0.5339843034744263,
|
| 8722 |
+
"learning_rate": 1.1634469543095383e-05,
|
| 8723 |
+
"loss": 2.22,
|
| 8724 |
+
"step": 1245
|
| 8725 |
+
},
|
| 8726 |
+
{
|
| 8727 |
+
"epoch": 0.5190585294730264,
|
| 8728 |
+
"grad_norm": 0.5852044820785522,
|
| 8729 |
+
"learning_rate": 1.1571310456032342e-05,
|
| 8730 |
+
"loss": 2.1708,
|
| 8731 |
+
"step": 1246
|
| 8732 |
+
},
|
| 8733 |
+
{
|
| 8734 |
+
"epoch": 0.5194751093522183,
|
| 8735 |
+
"grad_norm": 0.5521949529647827,
|
| 8736 |
+
"learning_rate": 1.150830082850468e-05,
|
| 8737 |
+
"loss": 2.1607,
|
| 8738 |
+
"step": 1247
|
| 8739 |
+
},
|
| 8740 |
+
{
|
| 8741 |
+
"epoch": 0.5198916892314102,
|
| 8742 |
+
"grad_norm": 0.5825779438018799,
|
| 8743 |
+
"learning_rate": 1.1445440905573856e-05,
|
| 8744 |
+
"loss": 2.2721,
|
| 8745 |
+
"step": 1248
|
| 8746 |
+
},
|
| 8747 |
+
{
|
| 8748 |
+
"epoch": 0.5203082691106019,
|
| 8749 |
+
"grad_norm": 0.6723480820655823,
|
| 8750 |
+
"learning_rate": 1.1382730931719043e-05,
|
| 8751 |
+
"loss": 2.3048,
|
| 8752 |
+
"step": 1249
|
| 8753 |
+
},
|
| 8754 |
+
{
|
| 8755 |
+
"epoch": 0.5207248489897938,
|
| 8756 |
+
"grad_norm": 0.7712786793708801,
|
| 8757 |
+
"learning_rate": 1.1320171150836228e-05,
|
| 8758 |
+
"loss": 2.2836,
|
| 8759 |
+
"step": 1250
|
| 8760 |
+
},
|
| 8761 |
+
{
|
| 8762 |
+
"epoch": 0.5211414288689856,
|
| 8763 |
+
"grad_norm": 0.362064003944397,
|
| 8764 |
+
"learning_rate": 1.1257761806237299e-05,
|
| 8765 |
+
"loss": 2.1024,
|
| 8766 |
+
"step": 1251
|
| 8767 |
+
},
|
| 8768 |
+
{
|
| 8769 |
+
"epoch": 0.5215580087481775,
|
| 8770 |
+
"grad_norm": 0.3768676817417145,
|
| 8771 |
+
"learning_rate": 1.1195503140648984e-05,
|
| 8772 |
+
"loss": 2.0597,
|
| 8773 |
+
"step": 1252
|
| 8774 |
+
},
|
| 8775 |
+
{
|
| 8776 |
+
"epoch": 0.5219745886273693,
|
| 8777 |
+
"grad_norm": 0.39301928877830505,
|
| 8778 |
+
"learning_rate": 1.1133395396212049e-05,
|
| 8779 |
+
"loss": 2.0886,
|
| 8780 |
+
"step": 1253
|
| 8781 |
+
},
|
| 8782 |
+
{
|
| 8783 |
+
"epoch": 0.5223911685065611,
|
| 8784 |
+
"grad_norm": 0.37058496475219727,
|
| 8785 |
+
"learning_rate": 1.1071438814480239e-05,
|
| 8786 |
+
"loss": 2.167,
|
| 8787 |
+
"step": 1254
|
| 8788 |
+
},
|
| 8789 |
+
{
|
| 8790 |
+
"epoch": 0.522807748385753,
|
| 8791 |
+
"grad_norm": 0.3886200189590454,
|
| 8792 |
+
"learning_rate": 1.100963363641943e-05,
|
| 8793 |
+
"loss": 2.1336,
|
| 8794 |
+
"step": 1255
|
| 8795 |
+
},
|
| 8796 |
+
{
|
| 8797 |
+
"epoch": 0.5232243282649448,
|
| 8798 |
+
"grad_norm": 0.4241747260093689,
|
| 8799 |
+
"learning_rate": 1.0947980102406596e-05,
|
| 8800 |
+
"loss": 1.9278,
|
| 8801 |
+
"step": 1256
|
| 8802 |
+
},
|
| 8803 |
+
{
|
| 8804 |
+
"epoch": 0.5236409081441367,
|
| 8805 |
+
"grad_norm": 0.41185516119003296,
|
| 8806 |
+
"learning_rate": 1.0886478452228988e-05,
|
| 8807 |
+
"loss": 2.1514,
|
| 8808 |
+
"step": 1257
|
| 8809 |
+
},
|
| 8810 |
+
{
|
| 8811 |
+
"epoch": 0.5240574880233285,
|
| 8812 |
+
"grad_norm": 0.40949538350105286,
|
| 8813 |
+
"learning_rate": 1.0825128925083089e-05,
|
| 8814 |
+
"loss": 2.2136,
|
| 8815 |
+
"step": 1258
|
| 8816 |
+
},
|
| 8817 |
+
{
|
| 8818 |
+
"epoch": 0.5244740679025203,
|
| 8819 |
+
"grad_norm": 0.42916616797447205,
|
| 8820 |
+
"learning_rate": 1.0763931759573736e-05,
|
| 8821 |
+
"loss": 2.2336,
|
| 8822 |
+
"step": 1259
|
| 8823 |
+
},
|
| 8824 |
+
{
|
| 8825 |
+
"epoch": 0.5248906477817121,
|
| 8826 |
+
"grad_norm": 0.39636459946632385,
|
| 8827 |
+
"learning_rate": 1.070288719371324e-05,
|
| 8828 |
+
"loss": 2.1677,
|
| 8829 |
+
"step": 1260
|
| 8830 |
+
},
|
| 8831 |
+
{
|
| 8832 |
+
"epoch": 0.525307227660904,
|
| 8833 |
+
"grad_norm": 0.445295125246048,
|
| 8834 |
+
"learning_rate": 1.0641995464920346e-05,
|
| 8835 |
+
"loss": 2.1093,
|
| 8836 |
+
"step": 1261
|
| 8837 |
+
},
|
| 8838 |
+
{
|
| 8839 |
+
"epoch": 0.5257238075400958,
|
| 8840 |
+
"grad_norm": 0.43316540122032166,
|
| 8841 |
+
"learning_rate": 1.0581256810019435e-05,
|
| 8842 |
+
"loss": 2.2007,
|
| 8843 |
+
"step": 1262
|
| 8844 |
+
},
|
| 8845 |
+
{
|
| 8846 |
+
"epoch": 0.5261403874192877,
|
| 8847 |
+
"grad_norm": 0.4185689091682434,
|
| 8848 |
+
"learning_rate": 1.0520671465239495e-05,
|
| 8849 |
+
"loss": 2.0491,
|
| 8850 |
+
"step": 1263
|
| 8851 |
+
},
|
| 8852 |
+
{
|
| 8853 |
+
"epoch": 0.5265569672984795,
|
| 8854 |
+
"grad_norm": 0.4575265049934387,
|
| 8855 |
+
"learning_rate": 1.0460239666213262e-05,
|
| 8856 |
+
"loss": 2.2749,
|
| 8857 |
+
"step": 1264
|
| 8858 |
+
},
|
| 8859 |
+
{
|
| 8860 |
+
"epoch": 0.5269735471776713,
|
| 8861 |
+
"grad_norm": 0.44808733463287354,
|
| 8862 |
+
"learning_rate": 1.0399961647976314e-05,
|
| 8863 |
+
"loss": 2.124,
|
| 8864 |
+
"step": 1265
|
| 8865 |
+
},
|
| 8866 |
+
{
|
| 8867 |
+
"epoch": 0.5273901270568632,
|
| 8868 |
+
"grad_norm": 0.43795716762542725,
|
| 8869 |
+
"learning_rate": 1.0339837644966099e-05,
|
| 8870 |
+
"loss": 2.1715,
|
| 8871 |
+
"step": 1266
|
| 8872 |
+
},
|
| 8873 |
+
{
|
| 8874 |
+
"epoch": 0.527806706936055,
|
| 8875 |
+
"grad_norm": 0.4253341257572174,
|
| 8876 |
+
"learning_rate": 1.02798678910211e-05,
|
| 8877 |
+
"loss": 2.2266,
|
| 8878 |
+
"step": 1267
|
| 8879 |
+
},
|
| 8880 |
+
{
|
| 8881 |
+
"epoch": 0.5282232868152468,
|
| 8882 |
+
"grad_norm": 0.44820019602775574,
|
| 8883 |
+
"learning_rate": 1.0220052619379843e-05,
|
| 8884 |
+
"loss": 2.2208,
|
| 8885 |
+
"step": 1268
|
| 8886 |
+
},
|
| 8887 |
+
{
|
| 8888 |
+
"epoch": 0.5286398666944386,
|
| 8889 |
+
"grad_norm": 0.4370150566101074,
|
| 8890 |
+
"learning_rate": 1.0160392062680057e-05,
|
| 8891 |
+
"loss": 2.2378,
|
| 8892 |
+
"step": 1269
|
| 8893 |
+
},
|
| 8894 |
+
{
|
| 8895 |
+
"epoch": 0.5290564465736305,
|
| 8896 |
+
"grad_norm": 0.60475754737854,
|
| 8897 |
+
"learning_rate": 1.0100886452957714e-05,
|
| 8898 |
+
"loss": 2.1901,
|
| 8899 |
+
"step": 1270
|
| 8900 |
+
},
|
| 8901 |
+
{
|
| 8902 |
+
"epoch": 0.5294730264528223,
|
| 8903 |
+
"grad_norm": 0.4462258517742157,
|
| 8904 |
+
"learning_rate": 1.0041536021646204e-05,
|
| 8905 |
+
"loss": 2.2266,
|
| 8906 |
+
"step": 1271
|
| 8907 |
+
},
|
| 8908 |
+
{
|
| 8909 |
+
"epoch": 0.5298896063320142,
|
| 8910 |
+
"grad_norm": 0.452957421541214,
|
| 8911 |
+
"learning_rate": 9.982340999575335e-06,
|
| 8912 |
+
"loss": 2.1806,
|
| 8913 |
+
"step": 1272
|
| 8914 |
+
},
|
| 8915 |
+
{
|
| 8916 |
+
"epoch": 0.530306186211206,
|
| 8917 |
+
"grad_norm": 0.434084951877594,
|
| 8918 |
+
"learning_rate": 9.92330161697051e-06,
|
| 8919 |
+
"loss": 2.1724,
|
| 8920 |
+
"step": 1273
|
| 8921 |
+
},
|
| 8922 |
+
{
|
| 8923 |
+
"epoch": 0.5307227660903978,
|
| 8924 |
+
"grad_norm": 0.4277683198451996,
|
| 8925 |
+
"learning_rate": 9.864418103451828e-06,
|
| 8926 |
+
"loss": 2.2438,
|
| 8927 |
+
"step": 1274
|
| 8928 |
+
},
|
| 8929 |
+
{
|
| 8930 |
+
"epoch": 0.5311393459695897,
|
| 8931 |
+
"grad_norm": 0.435316801071167,
|
| 8932 |
+
"learning_rate": 9.805690688033114e-06,
|
| 8933 |
+
"loss": 2.2084,
|
| 8934 |
+
"step": 1275
|
| 8935 |
+
},
|
| 8936 |
+
{
|
| 8937 |
+
"epoch": 0.5315559258487815,
|
| 8938 |
+
"grad_norm": 0.45000767707824707,
|
| 8939 |
+
"learning_rate": 9.747119599121163e-06,
|
| 8940 |
+
"loss": 2.1589,
|
| 8941 |
+
"step": 1276
|
| 8942 |
+
},
|
| 8943 |
+
{
|
| 8944 |
+
"epoch": 0.5319725057279734,
|
| 8945 |
+
"grad_norm": 0.45371347665786743,
|
| 8946 |
+
"learning_rate": 9.688705064514703e-06,
|
| 8947 |
+
"loss": 2.2112,
|
| 8948 |
+
"step": 1277
|
| 8949 |
+
},
|
| 8950 |
+
{
|
| 8951 |
+
"epoch": 0.5323890856071651,
|
| 8952 |
+
"grad_norm": 0.44338762760162354,
|
| 8953 |
+
"learning_rate": 9.630447311403623e-06,
|
| 8954 |
+
"loss": 2.277,
|
| 8955 |
+
"step": 1278
|
| 8956 |
+
},
|
| 8957 |
+
{
|
| 8958 |
+
"epoch": 0.532805665486357,
|
| 8959 |
+
"grad_norm": 0.4100877344608307,
|
| 8960 |
+
"learning_rate": 9.57234656636804e-06,
|
| 8961 |
+
"loss": 2.1358,
|
| 8962 |
+
"step": 1279
|
| 8963 |
+
},
|
| 8964 |
+
{
|
| 8965 |
+
"epoch": 0.5332222453655489,
|
| 8966 |
+
"grad_norm": 0.4822656214237213,
|
| 8967 |
+
"learning_rate": 9.514403055377386e-06,
|
| 8968 |
+
"loss": 2.2666,
|
| 8969 |
+
"step": 1280
|
| 8970 |
+
},
|
| 8971 |
+
{
|
| 8972 |
+
"epoch": 0.5336388252447407,
|
| 8973 |
+
"grad_norm": 0.4593331217765808,
|
| 8974 |
+
"learning_rate": 9.456617003789642e-06,
|
| 8975 |
+
"loss": 2.1719,
|
| 8976 |
+
"step": 1281
|
| 8977 |
+
},
|
| 8978 |
+
{
|
| 8979 |
+
"epoch": 0.5340554051239326,
|
| 8980 |
+
"grad_norm": 0.5039366483688354,
|
| 8981 |
+
"learning_rate": 9.398988636350303e-06,
|
| 8982 |
+
"loss": 2.1991,
|
| 8983 |
+
"step": 1282
|
| 8984 |
+
},
|
| 8985 |
+
{
|
| 8986 |
+
"epoch": 0.5344719850031243,
|
| 8987 |
+
"grad_norm": 0.4868943989276886,
|
| 8988 |
+
"learning_rate": 9.34151817719166e-06,
|
| 8989 |
+
"loss": 2.2524,
|
| 8990 |
+
"step": 1283
|
| 8991 |
+
},
|
| 8992 |
+
{
|
| 8993 |
+
"epoch": 0.5348885648823162,
|
| 8994 |
+
"grad_norm": 0.48140257596969604,
|
| 8995 |
+
"learning_rate": 9.284205849831817e-06,
|
| 8996 |
+
"loss": 2.2549,
|
| 8997 |
+
"step": 1284
|
| 8998 |
+
},
|
| 8999 |
+
{
|
| 9000 |
+
"epoch": 0.535305144761508,
|
| 9001 |
+
"grad_norm": 0.4989936649799347,
|
| 9002 |
+
"learning_rate": 9.227051877173859e-06,
|
| 9003 |
+
"loss": 2.2195,
|
| 9004 |
+
"step": 1285
|
| 9005 |
+
},
|
| 9006 |
+
{
|
| 9007 |
+
"epoch": 0.5357217246406999,
|
| 9008 |
+
"grad_norm": 0.4500278830528259,
|
| 9009 |
+
"learning_rate": 9.170056481504996e-06,
|
| 9010 |
+
"loss": 2.1515,
|
| 9011 |
+
"step": 1286
|
| 9012 |
+
},
|
| 9013 |
+
{
|
| 9014 |
+
"epoch": 0.5361383045198916,
|
| 9015 |
+
"grad_norm": 0.4891306459903717,
|
| 9016 |
+
"learning_rate": 9.1132198844957e-06,
|
| 9017 |
+
"loss": 2.1373,
|
| 9018 |
+
"step": 1287
|
| 9019 |
+
},
|
| 9020 |
+
{
|
| 9021 |
+
"epoch": 0.5365548843990835,
|
| 9022 |
+
"grad_norm": 0.4580099284648895,
|
| 9023 |
+
"learning_rate": 9.056542307198823e-06,
|
| 9024 |
+
"loss": 2.2233,
|
| 9025 |
+
"step": 1288
|
| 9026 |
+
},
|
| 9027 |
+
{
|
| 9028 |
+
"epoch": 0.5369714642782754,
|
| 9029 |
+
"grad_norm": 0.4709134101867676,
|
| 9030 |
+
"learning_rate": 9.000023970048726e-06,
|
| 9031 |
+
"loss": 2.1904,
|
| 9032 |
+
"step": 1289
|
| 9033 |
+
},
|
| 9034 |
+
{
|
| 9035 |
+
"epoch": 0.5373880441574672,
|
| 9036 |
+
"grad_norm": 0.492691308259964,
|
| 9037 |
+
"learning_rate": 8.943665092860488e-06,
|
| 9038 |
+
"loss": 2.2083,
|
| 9039 |
+
"step": 1290
|
| 9040 |
+
},
|
| 9041 |
+
{
|
| 9042 |
+
"epoch": 0.5378046240366591,
|
| 9043 |
+
"grad_norm": 0.5778969526290894,
|
| 9044 |
+
"learning_rate": 8.88746589482895e-06,
|
| 9045 |
+
"loss": 2.2656,
|
| 9046 |
+
"step": 1291
|
| 9047 |
+
},
|
| 9048 |
+
{
|
| 9049 |
+
"epoch": 0.5382212039158508,
|
| 9050 |
+
"grad_norm": 0.5076452493667603,
|
| 9051 |
+
"learning_rate": 8.831426594527975e-06,
|
| 9052 |
+
"loss": 2.2225,
|
| 9053 |
+
"step": 1292
|
| 9054 |
+
},
|
| 9055 |
+
{
|
| 9056 |
+
"epoch": 0.5386377837950427,
|
| 9057 |
+
"grad_norm": 0.5400701761245728,
|
| 9058 |
+
"learning_rate": 8.775547409909502e-06,
|
| 9059 |
+
"loss": 2.1506,
|
| 9060 |
+
"step": 1293
|
| 9061 |
+
},
|
| 9062 |
+
{
|
| 9063 |
+
"epoch": 0.5390543636742345,
|
| 9064 |
+
"grad_norm": 0.4912501573562622,
|
| 9065 |
+
"learning_rate": 8.719828558302733e-06,
|
| 9066 |
+
"loss": 2.2068,
|
| 9067 |
+
"step": 1294
|
| 9068 |
+
},
|
| 9069 |
+
{
|
| 9070 |
+
"epoch": 0.5394709435534264,
|
| 9071 |
+
"grad_norm": 0.5658887028694153,
|
| 9072 |
+
"learning_rate": 8.664270256413332e-06,
|
| 9073 |
+
"loss": 2.1972,
|
| 9074 |
+
"step": 1295
|
| 9075 |
+
},
|
| 9076 |
+
{
|
| 9077 |
+
"epoch": 0.5398875234326183,
|
| 9078 |
+
"grad_norm": 0.5720379948616028,
|
| 9079 |
+
"learning_rate": 8.60887272032248e-06,
|
| 9080 |
+
"loss": 2.113,
|
| 9081 |
+
"step": 1296
|
| 9082 |
+
},
|
| 9083 |
+
{
|
| 9084 |
+
"epoch": 0.54030410331181,
|
| 9085 |
+
"grad_norm": 0.5988136529922485,
|
| 9086 |
+
"learning_rate": 8.553636165486173e-06,
|
| 9087 |
+
"loss": 2.2126,
|
| 9088 |
+
"step": 1297
|
| 9089 |
+
},
|
| 9090 |
+
{
|
| 9091 |
+
"epoch": 0.5407206831910019,
|
| 9092 |
+
"grad_norm": 0.5860940217971802,
|
| 9093 |
+
"learning_rate": 8.498560806734229e-06,
|
| 9094 |
+
"loss": 2.244,
|
| 9095 |
+
"step": 1298
|
| 9096 |
+
},
|
| 9097 |
+
{
|
| 9098 |
+
"epoch": 0.5411372630701937,
|
| 9099 |
+
"grad_norm": 0.6165693402290344,
|
| 9100 |
+
"learning_rate": 8.443646858269604e-06,
|
| 9101 |
+
"loss": 2.3292,
|
| 9102 |
+
"step": 1299
|
| 9103 |
+
},
|
| 9104 |
+
{
|
| 9105 |
+
"epoch": 0.5415538429493856,
|
| 9106 |
+
"grad_norm": 0.7384406924247742,
|
| 9107 |
+
"learning_rate": 8.38889453366743e-06,
|
| 9108 |
+
"loss": 2.3799,
|
| 9109 |
+
"step": 1300
|
| 9110 |
+
},
|
| 9111 |
+
{
|
| 9112 |
+
"epoch": 0.5419704228285773,
|
| 9113 |
+
"grad_norm": 0.3716765344142914,
|
| 9114 |
+
"learning_rate": 8.334304045874247e-06,
|
| 9115 |
+
"loss": 2.0695,
|
| 9116 |
+
"step": 1301
|
| 9117 |
+
},
|
| 9118 |
+
{
|
| 9119 |
+
"epoch": 0.5423870027077692,
|
| 9120 |
+
"grad_norm": 0.386123925447464,
|
| 9121 |
+
"learning_rate": 8.279875607207227e-06,
|
| 9122 |
+
"loss": 2.0412,
|
| 9123 |
+
"step": 1302
|
| 9124 |
+
},
|
| 9125 |
+
{
|
| 9126 |
+
"epoch": 0.542803582586961,
|
| 9127 |
+
"grad_norm": 0.37405717372894287,
|
| 9128 |
+
"learning_rate": 8.225609429353187e-06,
|
| 9129 |
+
"loss": 2.147,
|
| 9130 |
+
"step": 1303
|
| 9131 |
+
},
|
| 9132 |
+
{
|
| 9133 |
+
"epoch": 0.5432201624661529,
|
| 9134 |
+
"grad_norm": 0.40637415647506714,
|
| 9135 |
+
"learning_rate": 8.171505723367961e-06,
|
| 9136 |
+
"loss": 2.1846,
|
| 9137 |
+
"step": 1304
|
| 9138 |
+
},
|
| 9139 |
+
{
|
| 9140 |
+
"epoch": 0.5436367423453448,
|
| 9141 |
+
"grad_norm": 0.4203828275203705,
|
| 9142 |
+
"learning_rate": 8.117564699675433e-06,
|
| 9143 |
+
"loss": 2.0723,
|
| 9144 |
+
"step": 1305
|
| 9145 |
+
},
|
| 9146 |
+
{
|
| 9147 |
+
"epoch": 0.5440533222245365,
|
| 9148 |
+
"grad_norm": 0.4045223891735077,
|
| 9149 |
+
"learning_rate": 8.063786568066806e-06,
|
| 9150 |
+
"loss": 2.2134,
|
| 9151 |
+
"step": 1306
|
| 9152 |
+
},
|
| 9153 |
+
{
|
| 9154 |
+
"epoch": 0.5444699021037284,
|
| 9155 |
+
"grad_norm": 0.42567864060401917,
|
| 9156 |
+
"learning_rate": 8.010171537699712e-06,
|
| 9157 |
+
"loss": 2.178,
|
| 9158 |
+
"step": 1307
|
| 9159 |
+
},
|
| 9160 |
+
{
|
| 9161 |
+
"epoch": 0.5448864819829202,
|
| 9162 |
+
"grad_norm": 0.39315065741539,
|
| 9163 |
+
"learning_rate": 7.956719817097492e-06,
|
| 9164 |
+
"loss": 2.1352,
|
| 9165 |
+
"step": 1308
|
| 9166 |
+
},
|
| 9167 |
+
{
|
| 9168 |
+
"epoch": 0.5453030618621121,
|
| 9169 |
+
"grad_norm": 0.4117122292518616,
|
| 9170 |
+
"learning_rate": 7.903431614148277e-06,
|
| 9171 |
+
"loss": 2.0137,
|
| 9172 |
+
"step": 1309
|
| 9173 |
+
},
|
| 9174 |
+
{
|
| 9175 |
+
"epoch": 0.5457196417413039,
|
| 9176 |
+
"grad_norm": 0.41751304268836975,
|
| 9177 |
+
"learning_rate": 7.850307136104247e-06,
|
| 9178 |
+
"loss": 2.0473,
|
| 9179 |
+
"step": 1310
|
| 9180 |
+
},
|
| 9181 |
+
{
|
| 9182 |
+
"epoch": 0.5461362216204957,
|
| 9183 |
+
"grad_norm": 0.42515861988067627,
|
| 9184 |
+
"learning_rate": 7.79734658958085e-06,
|
| 9185 |
+
"loss": 2.1861,
|
| 9186 |
+
"step": 1311
|
| 9187 |
+
},
|
| 9188 |
+
{
|
| 9189 |
+
"epoch": 0.5465528014996875,
|
| 9190 |
+
"grad_norm": 0.42132726311683655,
|
| 9191 |
+
"learning_rate": 7.744550180555905e-06,
|
| 9192 |
+
"loss": 2.1065,
|
| 9193 |
+
"step": 1312
|
| 9194 |
+
},
|
| 9195 |
+
{
|
| 9196 |
+
"epoch": 0.5469693813788794,
|
| 9197 |
+
"grad_norm": 0.43172234296798706,
|
| 9198 |
+
"learning_rate": 7.691918114368907e-06,
|
| 9199 |
+
"loss": 2.1498,
|
| 9200 |
+
"step": 1313
|
| 9201 |
+
},
|
| 9202 |
+
{
|
| 9203 |
+
"epoch": 0.5473859612580713,
|
| 9204 |
+
"grad_norm": 0.41604796051979065,
|
| 9205 |
+
"learning_rate": 7.639450595720137e-06,
|
| 9206 |
+
"loss": 2.1492,
|
| 9207 |
+
"step": 1314
|
| 9208 |
+
},
|
| 9209 |
+
{
|
| 9210 |
+
"epoch": 0.5478025411372631,
|
| 9211 |
+
"grad_norm": 0.42018142342567444,
|
| 9212 |
+
"learning_rate": 7.587147828669916e-06,
|
| 9213 |
+
"loss": 2.2407,
|
| 9214 |
+
"step": 1315
|
| 9215 |
+
},
|
| 9216 |
+
{
|
| 9217 |
+
"epoch": 0.5482191210164549,
|
| 9218 |
+
"grad_norm": 0.4484560489654541,
|
| 9219 |
+
"learning_rate": 7.535010016637828e-06,
|
| 9220 |
+
"loss": 2.1838,
|
| 9221 |
+
"step": 1316
|
| 9222 |
+
},
|
| 9223 |
+
{
|
| 9224 |
+
"epoch": 0.5486357008956467,
|
| 9225 |
+
"grad_norm": 0.4491097629070282,
|
| 9226 |
+
"learning_rate": 7.483037362401857e-06,
|
| 9227 |
+
"loss": 2.1785,
|
| 9228 |
+
"step": 1317
|
| 9229 |
+
},
|
| 9230 |
+
{
|
| 9231 |
+
"epoch": 0.5490522807748386,
|
| 9232 |
+
"grad_norm": 0.4612688422203064,
|
| 9233 |
+
"learning_rate": 7.4312300680977e-06,
|
| 9234 |
+
"loss": 2.1611,
|
| 9235 |
+
"step": 1318
|
| 9236 |
+
},
|
| 9237 |
+
{
|
| 9238 |
+
"epoch": 0.5494688606540304,
|
| 9239 |
+
"grad_norm": 0.44412896037101746,
|
| 9240 |
+
"learning_rate": 7.379588335217874e-06,
|
| 9241 |
+
"loss": 2.194,
|
| 9242 |
+
"step": 1319
|
| 9243 |
+
},
|
| 9244 |
+
{
|
| 9245 |
+
"epoch": 0.5498854405332222,
|
| 9246 |
+
"grad_norm": 0.4443187415599823,
|
| 9247 |
+
"learning_rate": 7.328112364611012e-06,
|
| 9248 |
+
"loss": 2.1786,
|
| 9249 |
+
"step": 1320
|
| 9250 |
+
},
|
| 9251 |
+
{
|
| 9252 |
+
"epoch": 0.550302020412414,
|
| 9253 |
+
"grad_norm": 0.4697517454624176,
|
| 9254 |
+
"learning_rate": 7.2768023564810165e-06,
|
| 9255 |
+
"loss": 2.153,
|
| 9256 |
+
"step": 1321
|
| 9257 |
+
},
|
| 9258 |
+
{
|
| 9259 |
+
"epoch": 0.5507186002916059,
|
| 9260 |
+
"grad_norm": 0.4328586161136627,
|
| 9261 |
+
"learning_rate": 7.225658510386385e-06,
|
| 9262 |
+
"loss": 2.1508,
|
| 9263 |
+
"step": 1322
|
| 9264 |
+
},
|
| 9265 |
+
{
|
| 9266 |
+
"epoch": 0.5511351801707978,
|
| 9267 |
+
"grad_norm": 0.4476320147514343,
|
| 9268 |
+
"learning_rate": 7.1746810252393e-06,
|
| 9269 |
+
"loss": 2.3088,
|
| 9270 |
+
"step": 1323
|
| 9271 |
+
},
|
| 9272 |
+
{
|
| 9273 |
+
"epoch": 0.5515517600499896,
|
| 9274 |
+
"grad_norm": 0.42715781927108765,
|
| 9275 |
+
"learning_rate": 7.123870099304941e-06,
|
| 9276 |
+
"loss": 2.2181,
|
| 9277 |
+
"step": 1324
|
| 9278 |
+
},
|
| 9279 |
+
{
|
| 9280 |
+
"epoch": 0.5519683399291814,
|
| 9281 |
+
"grad_norm": 0.4416360855102539,
|
| 9282 |
+
"learning_rate": 7.073225930200722e-06,
|
| 9283 |
+
"loss": 2.2013,
|
| 9284 |
+
"step": 1325
|
| 9285 |
+
},
|
| 9286 |
+
{
|
| 9287 |
+
"epoch": 0.5523849198083732,
|
| 9288 |
+
"grad_norm": 0.4616955518722534,
|
| 9289 |
+
"learning_rate": 7.022748714895444e-06,
|
| 9290 |
+
"loss": 2.0927,
|
| 9291 |
+
"step": 1326
|
| 9292 |
+
},
|
| 9293 |
+
{
|
| 9294 |
+
"epoch": 0.5528014996875651,
|
| 9295 |
+
"grad_norm": 0.47019249200820923,
|
| 9296 |
+
"learning_rate": 6.97243864970864e-06,
|
| 9297 |
+
"loss": 2.1574,
|
| 9298 |
+
"step": 1327
|
| 9299 |
+
},
|
| 9300 |
+
{
|
| 9301 |
+
"epoch": 0.553218079566757,
|
| 9302 |
+
"grad_norm": 0.46532902121543884,
|
| 9303 |
+
"learning_rate": 6.922295930309691e-06,
|
| 9304 |
+
"loss": 2.1184,
|
| 9305 |
+
"step": 1328
|
| 9306 |
+
},
|
| 9307 |
+
{
|
| 9308 |
+
"epoch": 0.5536346594459488,
|
| 9309 |
+
"grad_norm": 0.4631338119506836,
|
| 9310 |
+
"learning_rate": 6.872320751717176e-06,
|
| 9311 |
+
"loss": 2.1675,
|
| 9312 |
+
"step": 1329
|
| 9313 |
+
},
|
| 9314 |
+
{
|
| 9315 |
+
"epoch": 0.5540512393251406,
|
| 9316 |
+
"grad_norm": 0.45464131236076355,
|
| 9317 |
+
"learning_rate": 6.8225133082980284e-06,
|
| 9318 |
+
"loss": 2.1569,
|
| 9319 |
+
"step": 1330
|
| 9320 |
+
},
|
| 9321 |
+
{
|
| 9322 |
+
"epoch": 0.5544678192043324,
|
| 9323 |
+
"grad_norm": 0.48318782448768616,
|
| 9324 |
+
"learning_rate": 6.772873793766815e-06,
|
| 9325 |
+
"loss": 2.2317,
|
| 9326 |
+
"step": 1331
|
| 9327 |
+
},
|
| 9328 |
+
{
|
| 9329 |
+
"epoch": 0.5548843990835243,
|
| 9330 |
+
"grad_norm": 0.46803736686706543,
|
| 9331 |
+
"learning_rate": 6.723402401185014e-06,
|
| 9332 |
+
"loss": 2.1849,
|
| 9333 |
+
"step": 1332
|
| 9334 |
+
},
|
| 9335 |
+
{
|
| 9336 |
+
"epoch": 0.5553009789627161,
|
| 9337 |
+
"grad_norm": 0.5041208863258362,
|
| 9338 |
+
"learning_rate": 6.6740993229601864e-06,
|
| 9339 |
+
"loss": 2.1887,
|
| 9340 |
+
"step": 1333
|
| 9341 |
+
},
|
| 9342 |
+
{
|
| 9343 |
+
"epoch": 0.555717558841908,
|
| 9344 |
+
"grad_norm": 0.4831056594848633,
|
| 9345 |
+
"learning_rate": 6.624964750845325e-06,
|
| 9346 |
+
"loss": 2.3047,
|
| 9347 |
+
"step": 1334
|
| 9348 |
+
},
|
| 9349 |
+
{
|
| 9350 |
+
"epoch": 0.5561341387210997,
|
| 9351 |
+
"grad_norm": 0.475440114736557,
|
| 9352 |
+
"learning_rate": 6.575998875938017e-06,
|
| 9353 |
+
"loss": 2.2391,
|
| 9354 |
+
"step": 1335
|
| 9355 |
+
},
|
| 9356 |
+
{
|
| 9357 |
+
"epoch": 0.5565507186002916,
|
| 9358 |
+
"grad_norm": 0.5054152607917786,
|
| 9359 |
+
"learning_rate": 6.527201888679746e-06,
|
| 9360 |
+
"loss": 2.2116,
|
| 9361 |
+
"step": 1336
|
| 9362 |
+
},
|
| 9363 |
+
{
|
| 9364 |
+
"epoch": 0.5569672984794835,
|
| 9365 |
+
"grad_norm": 0.5019363760948181,
|
| 9366 |
+
"learning_rate": 6.478573978855146e-06,
|
| 9367 |
+
"loss": 2.232,
|
| 9368 |
+
"step": 1337
|
| 9369 |
+
},
|
| 9370 |
+
{
|
| 9371 |
+
"epoch": 0.5573838783586753,
|
| 9372 |
+
"grad_norm": 0.5251423716545105,
|
| 9373 |
+
"learning_rate": 6.4301153355912916e-06,
|
| 9374 |
+
"loss": 2.2459,
|
| 9375 |
+
"step": 1338
|
| 9376 |
+
},
|
| 9377 |
+
{
|
| 9378 |
+
"epoch": 0.5578004582378671,
|
| 9379 |
+
"grad_norm": 0.4702129662036896,
|
| 9380 |
+
"learning_rate": 6.381826147356906e-06,
|
| 9381 |
+
"loss": 2.3169,
|
| 9382 |
+
"step": 1339
|
| 9383 |
+
},
|
| 9384 |
+
{
|
| 9385 |
+
"epoch": 0.5582170381170589,
|
| 9386 |
+
"grad_norm": 0.5291473865509033,
|
| 9387 |
+
"learning_rate": 6.333706601961642e-06,
|
| 9388 |
+
"loss": 2.1435,
|
| 9389 |
+
"step": 1340
|
| 9390 |
+
},
|
| 9391 |
+
{
|
| 9392 |
+
"epoch": 0.5586336179962508,
|
| 9393 |
+
"grad_norm": 0.5018637776374817,
|
| 9394 |
+
"learning_rate": 6.285756886555422e-06,
|
| 9395 |
+
"loss": 2.1405,
|
| 9396 |
+
"step": 1341
|
| 9397 |
+
},
|
| 9398 |
+
{
|
| 9399 |
+
"epoch": 0.5590501978754426,
|
| 9400 |
+
"grad_norm": 0.4844542443752289,
|
| 9401 |
+
"learning_rate": 6.237977187627586e-06,
|
| 9402 |
+
"loss": 2.1674,
|
| 9403 |
+
"step": 1342
|
| 9404 |
+
},
|
| 9405 |
+
{
|
| 9406 |
+
"epoch": 0.5594667777546345,
|
| 9407 |
+
"grad_norm": 0.5339523553848267,
|
| 9408 |
+
"learning_rate": 6.190367691006288e-06,
|
| 9409 |
+
"loss": 2.236,
|
| 9410 |
+
"step": 1343
|
| 9411 |
+
},
|
| 9412 |
+
{
|
| 9413 |
+
"epoch": 0.5598833576338262,
|
| 9414 |
+
"grad_norm": 0.5443533658981323,
|
| 9415 |
+
"learning_rate": 6.142928581857693e-06,
|
| 9416 |
+
"loss": 2.3391,
|
| 9417 |
+
"step": 1344
|
| 9418 |
+
},
|
| 9419 |
+
{
|
| 9420 |
+
"epoch": 0.5602999375130181,
|
| 9421 |
+
"grad_norm": 0.534054160118103,
|
| 9422 |
+
"learning_rate": 6.095660044685275e-06,
|
| 9423 |
+
"loss": 2.2016,
|
| 9424 |
+
"step": 1345
|
| 9425 |
+
},
|
| 9426 |
+
{
|
| 9427 |
+
"epoch": 0.56071651739221,
|
| 9428 |
+
"grad_norm": 0.5672020316123962,
|
| 9429 |
+
"learning_rate": 6.048562263329138e-06,
|
| 9430 |
+
"loss": 2.2499,
|
| 9431 |
+
"step": 1346
|
| 9432 |
+
},
|
| 9433 |
+
{
|
| 9434 |
+
"epoch": 0.5611330972714018,
|
| 9435 |
+
"grad_norm": 0.5929094552993774,
|
| 9436 |
+
"learning_rate": 6.001635420965235e-06,
|
| 9437 |
+
"loss": 2.2559,
|
| 9438 |
+
"step": 1347
|
| 9439 |
+
},
|
| 9440 |
+
{
|
| 9441 |
+
"epoch": 0.5615496771505937,
|
| 9442 |
+
"grad_norm": 0.5988344550132751,
|
| 9443 |
+
"learning_rate": 5.954879700104732e-06,
|
| 9444 |
+
"loss": 2.0733,
|
| 9445 |
+
"step": 1348
|
| 9446 |
+
},
|
| 9447 |
+
{
|
| 9448 |
+
"epoch": 0.5619662570297854,
|
| 9449 |
+
"grad_norm": 0.6177953481674194,
|
| 9450 |
+
"learning_rate": 5.908295282593207e-06,
|
| 9451 |
+
"loss": 2.2322,
|
| 9452 |
+
"step": 1349
|
| 9453 |
+
},
|
| 9454 |
+
{
|
| 9455 |
+
"epoch": 0.5623828369089773,
|
| 9456 |
+
"grad_norm": 0.7406448721885681,
|
| 9457 |
+
"learning_rate": 5.861882349610042e-06,
|
| 9458 |
+
"loss": 2.327,
|
| 9459 |
+
"step": 1350
|
| 9460 |
+
},
|
| 9461 |
+
{
|
| 9462 |
+
"epoch": 0.5627994167881691,
|
| 9463 |
+
"grad_norm": 0.3616834878921509,
|
| 9464 |
+
"learning_rate": 5.81564108166765e-06,
|
| 9465 |
+
"loss": 2.1104,
|
| 9466 |
+
"step": 1351
|
| 9467 |
+
},
|
| 9468 |
+
{
|
| 9469 |
+
"epoch": 0.563215996667361,
|
| 9470 |
+
"grad_norm": 0.41988110542297363,
|
| 9471 |
+
"learning_rate": 5.769571658610762e-06,
|
| 9472 |
+
"loss": 2.1414,
|
| 9473 |
+
"step": 1352
|
| 9474 |
+
},
|
| 9475 |
+
{
|
| 9476 |
+
"epoch": 0.5636325765465529,
|
| 9477 |
+
"grad_norm": 0.39019477367401123,
|
| 9478 |
+
"learning_rate": 5.72367425961583e-06,
|
| 9479 |
+
"loss": 2.2014,
|
| 9480 |
+
"step": 1353
|
| 9481 |
+
},
|
| 9482 |
+
{
|
| 9483 |
+
"epoch": 0.5640491564257446,
|
| 9484 |
+
"grad_norm": 0.3775304853916168,
|
| 9485 |
+
"learning_rate": 5.677949063190169e-06,
|
| 9486 |
+
"loss": 2.1051,
|
| 9487 |
+
"step": 1354
|
| 9488 |
+
},
|
| 9489 |
+
{
|
| 9490 |
+
"epoch": 0.5644657363049365,
|
| 9491 |
+
"grad_norm": 0.40709546208381653,
|
| 9492 |
+
"learning_rate": 5.6323962471714286e-06,
|
| 9493 |
+
"loss": 2.2534,
|
| 9494 |
+
"step": 1355
|
| 9495 |
+
},
|
| 9496 |
+
{
|
| 9497 |
+
"epoch": 0.5648823161841283,
|
| 9498 |
+
"grad_norm": 0.4010683298110962,
|
| 9499 |
+
"learning_rate": 5.587015988726774e-06,
|
| 9500 |
+
"loss": 2.1041,
|
| 9501 |
+
"step": 1356
|
| 9502 |
+
},
|
| 9503 |
+
{
|
| 9504 |
+
"epoch": 0.5652988960633202,
|
| 9505 |
+
"grad_norm": 0.4118007719516754,
|
| 9506 |
+
"learning_rate": 5.541808464352277e-06,
|
| 9507 |
+
"loss": 2.0971,
|
| 9508 |
+
"step": 1357
|
| 9509 |
+
},
|
| 9510 |
+
{
|
| 9511 |
+
"epoch": 0.5657154759425119,
|
| 9512 |
+
"grad_norm": 0.40107977390289307,
|
| 9513 |
+
"learning_rate": 5.496773849872183e-06,
|
| 9514 |
+
"loss": 2.2204,
|
| 9515 |
+
"step": 1358
|
| 9516 |
+
},
|
| 9517 |
+
{
|
| 9518 |
+
"epoch": 0.5661320558217038,
|
| 9519 |
+
"grad_norm": 0.39281153678894043,
|
| 9520 |
+
"learning_rate": 5.4519123204382625e-06,
|
| 9521 |
+
"loss": 2.1446,
|
| 9522 |
+
"step": 1359
|
| 9523 |
+
},
|
| 9524 |
+
{
|
| 9525 |
+
"epoch": 0.5665486357008956,
|
| 9526 |
+
"grad_norm": 0.4124067723751068,
|
| 9527 |
+
"learning_rate": 5.407224050529097e-06,
|
| 9528 |
+
"loss": 2.2212,
|
| 9529 |
+
"step": 1360
|
| 9530 |
+
},
|
| 9531 |
+
{
|
| 9532 |
+
"epoch": 0.5669652155800875,
|
| 9533 |
+
"grad_norm": 0.390995591878891,
|
| 9534 |
+
"learning_rate": 5.362709213949396e-06,
|
| 9535 |
+
"loss": 2.138,
|
| 9536 |
+
"step": 1361
|
| 9537 |
+
},
|
| 9538 |
+
{
|
| 9539 |
+
"epoch": 0.5673817954592794,
|
| 9540 |
+
"grad_norm": 0.4064914286136627,
|
| 9541 |
+
"learning_rate": 5.318367983829392e-06,
|
| 9542 |
+
"loss": 2.1965,
|
| 9543 |
+
"step": 1362
|
| 9544 |
+
},
|
| 9545 |
+
{
|
| 9546 |
+
"epoch": 0.5677983753384711,
|
| 9547 |
+
"grad_norm": 0.4465165436267853,
|
| 9548 |
+
"learning_rate": 5.2742005326240575e-06,
|
| 9549 |
+
"loss": 2.1492,
|
| 9550 |
+
"step": 1363
|
| 9551 |
+
},
|
| 9552 |
+
{
|
| 9553 |
+
"epoch": 0.568214955217663,
|
| 9554 |
+
"grad_norm": 0.422794371843338,
|
| 9555 |
+
"learning_rate": 5.23020703211255e-06,
|
| 9556 |
+
"loss": 2.3525,
|
| 9557 |
+
"step": 1364
|
| 9558 |
+
},
|
| 9559 |
+
{
|
| 9560 |
+
"epoch": 0.5686315350968548,
|
| 9561 |
+
"grad_norm": 0.4358288645744324,
|
| 9562 |
+
"learning_rate": 5.186387653397435e-06,
|
| 9563 |
+
"loss": 2.2029,
|
| 9564 |
+
"step": 1365
|
| 9565 |
+
},
|
| 9566 |
+
{
|
| 9567 |
+
"epoch": 0.5690481149760467,
|
| 9568 |
+
"grad_norm": 0.46458449959754944,
|
| 9569 |
+
"learning_rate": 5.142742566904079e-06,
|
| 9570 |
+
"loss": 2.1966,
|
| 9571 |
+
"step": 1366
|
| 9572 |
+
},
|
| 9573 |
+
{
|
| 9574 |
+
"epoch": 0.5694646948552385,
|
| 9575 |
+
"grad_norm": 0.4231399595737457,
|
| 9576 |
+
"learning_rate": 5.0992719423800236e-06,
|
| 9577 |
+
"loss": 2.2435,
|
| 9578 |
+
"step": 1367
|
| 9579 |
+
},
|
| 9580 |
+
{
|
| 9581 |
+
"epoch": 0.5698812747344303,
|
| 9582 |
+
"grad_norm": 0.44632214307785034,
|
| 9583 |
+
"learning_rate": 5.055975948894226e-06,
|
| 9584 |
+
"loss": 2.261,
|
| 9585 |
+
"step": 1368
|
| 9586 |
+
},
|
| 9587 |
+
{
|
| 9588 |
+
"epoch": 0.5702978546136221,
|
| 9589 |
+
"grad_norm": 0.41880112886428833,
|
| 9590 |
+
"learning_rate": 5.012854754836488e-06,
|
| 9591 |
+
"loss": 2.0232,
|
| 9592 |
+
"step": 1369
|
| 9593 |
+
},
|
| 9594 |
+
{
|
| 9595 |
+
"epoch": 0.570714434492814,
|
| 9596 |
+
"grad_norm": 0.43773967027664185,
|
| 9597 |
+
"learning_rate": 4.969908527916766e-06,
|
| 9598 |
+
"loss": 2.1273,
|
| 9599 |
+
"step": 1370
|
| 9600 |
+
},
|
| 9601 |
+
{
|
| 9602 |
+
"epoch": 0.5711310143720059,
|
| 9603 |
+
"grad_norm": 0.4529377520084381,
|
| 9604 |
+
"learning_rate": 4.9271374351645024e-06,
|
| 9605 |
+
"loss": 2.1708,
|
| 9606 |
+
"step": 1371
|
| 9607 |
+
},
|
| 9608 |
+
{
|
| 9609 |
+
"epoch": 0.5715475942511977,
|
| 9610 |
+
"grad_norm": 0.4359259307384491,
|
| 9611 |
+
"learning_rate": 4.884541642928009e-06,
|
| 9612 |
+
"loss": 1.9911,
|
| 9613 |
+
"step": 1372
|
| 9614 |
+
},
|
| 9615 |
+
{
|
| 9616 |
+
"epoch": 0.5719641741303895,
|
| 9617 |
+
"grad_norm": 0.4691868722438812,
|
| 9618 |
+
"learning_rate": 4.842121316873821e-06,
|
| 9619 |
+
"loss": 2.1426,
|
| 9620 |
+
"step": 1373
|
| 9621 |
+
},
|
| 9622 |
+
{
|
| 9623 |
+
"epoch": 0.5723807540095813,
|
| 9624 |
+
"grad_norm": 0.4630512297153473,
|
| 9625 |
+
"learning_rate": 4.799876621986021e-06,
|
| 9626 |
+
"loss": 2.2009,
|
| 9627 |
+
"step": 1374
|
| 9628 |
+
},
|
| 9629 |
+
{
|
| 9630 |
+
"epoch": 0.5727973338887732,
|
| 9631 |
+
"grad_norm": 0.456673264503479,
|
| 9632 |
+
"learning_rate": 4.757807722565605e-06,
|
| 9633 |
+
"loss": 2.124,
|
| 9634 |
+
"step": 1375
|
| 9635 |
+
},
|
| 9636 |
+
{
|
| 9637 |
+
"epoch": 0.573213913767965,
|
| 9638 |
+
"grad_norm": 0.44640928506851196,
|
| 9639 |
+
"learning_rate": 4.71591478222988e-06,
|
| 9640 |
+
"loss": 2.1812,
|
| 9641 |
+
"step": 1376
|
| 9642 |
+
},
|
| 9643 |
+
{
|
| 9644 |
+
"epoch": 0.5736304936471568,
|
| 9645 |
+
"grad_norm": 0.4493945837020874,
|
| 9646 |
+
"learning_rate": 4.674197963911775e-06,
|
| 9647 |
+
"loss": 2.0659,
|
| 9648 |
+
"step": 1377
|
| 9649 |
+
},
|
| 9650 |
+
{
|
| 9651 |
+
"epoch": 0.5740470735263487,
|
| 9652 |
+
"grad_norm": 0.4972690939903259,
|
| 9653 |
+
"learning_rate": 4.632657429859266e-06,
|
| 9654 |
+
"loss": 2.1043,
|
| 9655 |
+
"step": 1378
|
| 9656 |
+
},
|
| 9657 |
+
{
|
| 9658 |
+
"epoch": 0.5744636534055405,
|
| 9659 |
+
"grad_norm": 0.48750877380371094,
|
| 9660 |
+
"learning_rate": 4.591293341634661e-06,
|
| 9661 |
+
"loss": 2.2463,
|
| 9662 |
+
"step": 1379
|
| 9663 |
+
},
|
| 9664 |
+
{
|
| 9665 |
+
"epoch": 0.5748802332847324,
|
| 9666 |
+
"grad_norm": 0.47315627336502075,
|
| 9667 |
+
"learning_rate": 4.550105860114085e-06,
|
| 9668 |
+
"loss": 2.1493,
|
| 9669 |
+
"step": 1380
|
| 9670 |
+
},
|
| 9671 |
+
{
|
| 9672 |
+
"epoch": 0.5752968131639242,
|
| 9673 |
+
"grad_norm": 0.5010528564453125,
|
| 9674 |
+
"learning_rate": 4.509095145486736e-06,
|
| 9675 |
+
"loss": 2.1929,
|
| 9676 |
+
"step": 1381
|
| 9677 |
+
},
|
| 9678 |
+
{
|
| 9679 |
+
"epoch": 0.575713393043116,
|
| 9680 |
+
"grad_norm": 0.4589388966560364,
|
| 9681 |
+
"learning_rate": 4.468261357254339e-06,
|
| 9682 |
+
"loss": 2.1532,
|
| 9683 |
+
"step": 1382
|
| 9684 |
+
},
|
| 9685 |
+
{
|
| 9686 |
+
"epoch": 0.5761299729223078,
|
| 9687 |
+
"grad_norm": 0.47072404623031616,
|
| 9688 |
+
"learning_rate": 4.4276046542305276e-06,
|
| 9689 |
+
"loss": 2.096,
|
| 9690 |
+
"step": 1383
|
| 9691 |
+
},
|
| 9692 |
+
{
|
| 9693 |
+
"epoch": 0.5765465528014997,
|
| 9694 |
+
"grad_norm": 0.49293652176856995,
|
| 9695 |
+
"learning_rate": 4.387125194540159e-06,
|
| 9696 |
+
"loss": 2.2295,
|
| 9697 |
+
"step": 1384
|
| 9698 |
+
},
|
| 9699 |
+
{
|
| 9700 |
+
"epoch": 0.5769631326806915,
|
| 9701 |
+
"grad_norm": 0.4684501588344574,
|
| 9702 |
+
"learning_rate": 4.346823135618788e-06,
|
| 9703 |
+
"loss": 2.2086,
|
| 9704 |
+
"step": 1385
|
| 9705 |
+
},
|
| 9706 |
+
{
|
| 9707 |
+
"epoch": 0.5773797125598834,
|
| 9708 |
+
"grad_norm": 0.48839181661605835,
|
| 9709 |
+
"learning_rate": 4.306698634211976e-06,
|
| 9710 |
+
"loss": 2.1639,
|
| 9711 |
+
"step": 1386
|
| 9712 |
+
},
|
| 9713 |
+
{
|
| 9714 |
+
"epoch": 0.5777962924390752,
|
| 9715 |
+
"grad_norm": 0.501790463924408,
|
| 9716 |
+
"learning_rate": 4.266751846374734e-06,
|
| 9717 |
+
"loss": 2.2215,
|
| 9718 |
+
"step": 1387
|
| 9719 |
+
},
|
| 9720 |
+
{
|
| 9721 |
+
"epoch": 0.578212872318267,
|
| 9722 |
+
"grad_norm": 0.5091081857681274,
|
| 9723 |
+
"learning_rate": 4.226982927470874e-06,
|
| 9724 |
+
"loss": 2.2341,
|
| 9725 |
+
"step": 1388
|
| 9726 |
+
},
|
| 9727 |
+
{
|
| 9728 |
+
"epoch": 0.5786294521974589,
|
| 9729 |
+
"grad_norm": 0.5062838792800903,
|
| 9730 |
+
"learning_rate": 4.187392032172471e-06,
|
| 9731 |
+
"loss": 2.1799,
|
| 9732 |
+
"step": 1389
|
| 9733 |
+
},
|
| 9734 |
+
{
|
| 9735 |
+
"epoch": 0.5790460320766507,
|
| 9736 |
+
"grad_norm": 0.5266156792640686,
|
| 9737 |
+
"learning_rate": 4.1479793144591775e-06,
|
| 9738 |
+
"loss": 2.2925,
|
| 9739 |
+
"step": 1390
|
| 9740 |
+
},
|
| 9741 |
+
{
|
| 9742 |
+
"epoch": 0.5794626119558426,
|
| 9743 |
+
"grad_norm": 0.5020082592964172,
|
| 9744 |
+
"learning_rate": 4.108744927617669e-06,
|
| 9745 |
+
"loss": 2.2214,
|
| 9746 |
+
"step": 1391
|
| 9747 |
+
},
|
| 9748 |
+
{
|
| 9749 |
+
"epoch": 0.5798791918350343,
|
| 9750 |
+
"grad_norm": 0.5312255620956421,
|
| 9751 |
+
"learning_rate": 4.06968902424108e-06,
|
| 9752 |
+
"loss": 2.3272,
|
| 9753 |
+
"step": 1392
|
| 9754 |
+
},
|
| 9755 |
+
{
|
| 9756 |
+
"epoch": 0.5802957717142262,
|
| 9757 |
+
"grad_norm": 0.5147542953491211,
|
| 9758 |
+
"learning_rate": 4.030811756228331e-06,
|
| 9759 |
+
"loss": 2.3233,
|
| 9760 |
+
"step": 1393
|
| 9761 |
+
},
|
| 9762 |
+
{
|
| 9763 |
+
"epoch": 0.580712351593418,
|
| 9764 |
+
"grad_norm": 0.5263500213623047,
|
| 9765 |
+
"learning_rate": 3.99211327478361e-06,
|
| 9766 |
+
"loss": 2.2691,
|
| 9767 |
+
"step": 1394
|
| 9768 |
+
},
|
| 9769 |
+
{
|
| 9770 |
+
"epoch": 0.5811289314726099,
|
| 9771 |
+
"grad_norm": 0.5341267585754395,
|
| 9772 |
+
"learning_rate": 3.953593730415745e-06,
|
| 9773 |
+
"loss": 2.144,
|
| 9774 |
+
"step": 1395
|
| 9775 |
+
},
|
| 9776 |
+
{
|
| 9777 |
+
"epoch": 0.5815455113518017,
|
| 9778 |
+
"grad_norm": 0.5725787281990051,
|
| 9779 |
+
"learning_rate": 3.915253272937613e-06,
|
| 9780 |
+
"loss": 2.1749,
|
| 9781 |
+
"step": 1396
|
| 9782 |
+
},
|
| 9783 |
+
{
|
| 9784 |
+
"epoch": 0.5819620912309935,
|
| 9785 |
+
"grad_norm": 0.5373120307922363,
|
| 9786 |
+
"learning_rate": 3.877092051465614e-06,
|
| 9787 |
+
"loss": 2.1293,
|
| 9788 |
+
"step": 1397
|
| 9789 |
+
},
|
| 9790 |
+
{
|
| 9791 |
+
"epoch": 0.5823786711101854,
|
| 9792 |
+
"grad_norm": 0.5862066149711609,
|
| 9793 |
+
"learning_rate": 3.839110214419017e-06,
|
| 9794 |
+
"loss": 2.2286,
|
| 9795 |
+
"step": 1398
|
| 9796 |
+
},
|
| 9797 |
+
{
|
| 9798 |
+
"epoch": 0.5827952509893772,
|
| 9799 |
+
"grad_norm": 0.6132252216339111,
|
| 9800 |
+
"learning_rate": 3.8013079095194304e-06,
|
| 9801 |
+
"loss": 2.2992,
|
| 9802 |
+
"step": 1399
|
| 9803 |
+
},
|
| 9804 |
+
{
|
| 9805 |
+
"epoch": 0.5832118308685691,
|
| 9806 |
+
"grad_norm": 1.0293792486190796,
|
| 9807 |
+
"learning_rate": 3.763685283790208e-06,
|
| 9808 |
+
"loss": 2.2717,
|
| 9809 |
+
"step": 1400
|
| 9810 |
+
},
|
| 9811 |
+
{
|
| 9812 |
+
"epoch": 0.5836284107477608,
|
| 9813 |
+
"grad_norm": 0.3628023564815521,
|
| 9814 |
+
"learning_rate": 3.7262424835558917e-06,
|
| 9815 |
+
"loss": 2.1814,
|
| 9816 |
+
"step": 1401
|
| 9817 |
+
},
|
| 9818 |
+
{
|
| 9819 |
+
"epoch": 0.5840449906269527,
|
| 9820 |
+
"grad_norm": 0.38038867712020874,
|
| 9821 |
+
"learning_rate": 3.6889796544416267e-06,
|
| 9822 |
+
"loss": 2.1976,
|
| 9823 |
+
"step": 1402
|
| 9824 |
+
},
|
| 9825 |
+
{
|
| 9826 |
+
"epoch": 0.5844615705061446,
|
| 9827 |
+
"grad_norm": 0.395766943693161,
|
| 9828 |
+
"learning_rate": 3.6518969413725903e-06,
|
| 9829 |
+
"loss": 2.2299,
|
| 9830 |
+
"step": 1403
|
| 9831 |
+
},
|
| 9832 |
+
{
|
| 9833 |
+
"epoch": 0.5848781503853364,
|
| 9834 |
+
"grad_norm": 0.44534817337989807,
|
| 9835 |
+
"learning_rate": 3.6149944885734756e-06,
|
| 9836 |
+
"loss": 2.1294,
|
| 9837 |
+
"step": 1404
|
| 9838 |
+
},
|
| 9839 |
+
{
|
| 9840 |
+
"epoch": 0.5852947302645283,
|
| 9841 |
+
"grad_norm": 0.4051525890827179,
|
| 9842 |
+
"learning_rate": 3.578272439567848e-06,
|
| 9843 |
+
"loss": 2.2731,
|
| 9844 |
+
"step": 1405
|
| 9845 |
+
},
|
| 9846 |
+
{
|
| 9847 |
+
"epoch": 0.58571131014372,
|
| 9848 |
+
"grad_norm": 0.37931153178215027,
|
| 9849 |
+
"learning_rate": 3.5417309371776675e-06,
|
| 9850 |
+
"loss": 2.1129,
|
| 9851 |
+
"step": 1406
|
| 9852 |
+
},
|
| 9853 |
+
{
|
| 9854 |
+
"epoch": 0.5861278900229119,
|
| 9855 |
+
"grad_norm": 0.44215741753578186,
|
| 9856 |
+
"learning_rate": 3.505370123522689e-06,
|
| 9857 |
+
"loss": 2.0779,
|
| 9858 |
+
"step": 1407
|
| 9859 |
+
},
|
| 9860 |
+
{
|
| 9861 |
+
"epoch": 0.5865444699021037,
|
| 9862 |
+
"grad_norm": 0.4249580502510071,
|
| 9863 |
+
"learning_rate": 3.4691901400199235e-06,
|
| 9864 |
+
"loss": 2.2216,
|
| 9865 |
+
"step": 1408
|
| 9866 |
+
},
|
| 9867 |
+
{
|
| 9868 |
+
"epoch": 0.5869610497812956,
|
| 9869 |
+
"grad_norm": 0.4146334230899811,
|
| 9870 |
+
"learning_rate": 3.4331911273830784e-06,
|
| 9871 |
+
"loss": 2.1824,
|
| 9872 |
+
"step": 1409
|
| 9873 |
+
},
|
| 9874 |
+
{
|
| 9875 |
+
"epoch": 0.5873776296604873,
|
| 9876 |
+
"grad_norm": 0.3934686481952667,
|
| 9877 |
+
"learning_rate": 3.3973732256220246e-06,
|
| 9878 |
+
"loss": 2.1035,
|
| 9879 |
+
"step": 1410
|
| 9880 |
+
},
|
| 9881 |
+
{
|
| 9882 |
+
"epoch": 0.5877942095396792,
|
| 9883 |
+
"grad_norm": 0.4455997943878174,
|
| 9884 |
+
"learning_rate": 3.3617365740422456e-06,
|
| 9885 |
+
"loss": 2.1365,
|
| 9886 |
+
"step": 1411
|
| 9887 |
+
},
|
| 9888 |
+
{
|
| 9889 |
+
"epoch": 0.5882107894188711,
|
| 9890 |
+
"grad_norm": 0.4235437214374542,
|
| 9891 |
+
"learning_rate": 3.326281311244267e-06,
|
| 9892 |
+
"loss": 2.2011,
|
| 9893 |
+
"step": 1412
|
| 9894 |
+
},
|
| 9895 |
+
{
|
| 9896 |
+
"epoch": 0.5886273692980629,
|
| 9897 |
+
"grad_norm": 0.4142230153083801,
|
| 9898 |
+
"learning_rate": 3.2910075751231895e-06,
|
| 9899 |
+
"loss": 2.1224,
|
| 9900 |
+
"step": 1413
|
| 9901 |
+
},
|
| 9902 |
+
{
|
| 9903 |
+
"epoch": 0.5890439491772548,
|
| 9904 |
+
"grad_norm": 0.42072489857673645,
|
| 9905 |
+
"learning_rate": 3.255915502868062e-06,
|
| 9906 |
+
"loss": 2.1745,
|
| 9907 |
+
"step": 1414
|
| 9908 |
+
},
|
| 9909 |
+
{
|
| 9910 |
+
"epoch": 0.5894605290564465,
|
| 9911 |
+
"grad_norm": 0.43629273772239685,
|
| 9912 |
+
"learning_rate": 3.2210052309614437e-06,
|
| 9913 |
+
"loss": 2.1543,
|
| 9914 |
+
"step": 1415
|
| 9915 |
+
},
|
| 9916 |
+
{
|
| 9917 |
+
"epoch": 0.5898771089356384,
|
| 9918 |
+
"grad_norm": 0.42446547746658325,
|
| 9919 |
+
"learning_rate": 3.186276895178775e-06,
|
| 9920 |
+
"loss": 2.117,
|
| 9921 |
+
"step": 1416
|
| 9922 |
+
},
|
| 9923 |
+
{
|
| 9924 |
+
"epoch": 0.5902936888148302,
|
| 9925 |
+
"grad_norm": 0.43842992186546326,
|
| 9926 |
+
"learning_rate": 3.151730630587912e-06,
|
| 9927 |
+
"loss": 2.0253,
|
| 9928 |
+
"step": 1417
|
| 9929 |
+
},
|
| 9930 |
+
{
|
| 9931 |
+
"epoch": 0.5907102686940221,
|
| 9932 |
+
"grad_norm": 0.44594523310661316,
|
| 9933 |
+
"learning_rate": 3.117366571548608e-06,
|
| 9934 |
+
"loss": 2.3175,
|
| 9935 |
+
"step": 1418
|
| 9936 |
+
},
|
| 9937 |
+
{
|
| 9938 |
+
"epoch": 0.591126848573214,
|
| 9939 |
+
"grad_norm": 0.43836236000061035,
|
| 9940 |
+
"learning_rate": 3.0831848517119454e-06,
|
| 9941 |
+
"loss": 2.1398,
|
| 9942 |
+
"step": 1419
|
| 9943 |
+
},
|
| 9944 |
+
{
|
| 9945 |
+
"epoch": 0.5915434284524057,
|
| 9946 |
+
"grad_norm": 0.42096608877182007,
|
| 9947 |
+
"learning_rate": 3.0491856040198606e-06,
|
| 9948 |
+
"loss": 2.2345,
|
| 9949 |
+
"step": 1420
|
| 9950 |
+
},
|
| 9951 |
+
{
|
| 9952 |
+
"epoch": 0.5919600083315976,
|
| 9953 |
+
"grad_norm": 0.4303555190563202,
|
| 9954 |
+
"learning_rate": 3.0153689607045845e-06,
|
| 9955 |
+
"loss": 2.0476,
|
| 9956 |
+
"step": 1421
|
| 9957 |
+
},
|
| 9958 |
+
{
|
| 9959 |
+
"epoch": 0.5923765882107894,
|
| 9960 |
+
"grad_norm": 0.49172818660736084,
|
| 9961 |
+
"learning_rate": 2.981735053288165e-06,
|
| 9962 |
+
"loss": 2.2606,
|
| 9963 |
+
"step": 1422
|
| 9964 |
+
},
|
| 9965 |
+
{
|
| 9966 |
+
"epoch": 0.5927931680899813,
|
| 9967 |
+
"grad_norm": 0.4514530599117279,
|
| 9968 |
+
"learning_rate": 2.9482840125819254e-06,
|
| 9969 |
+
"loss": 2.2732,
|
| 9970 |
+
"step": 1423
|
| 9971 |
+
},
|
| 9972 |
+
{
|
| 9973 |
+
"epoch": 0.5932097479691731,
|
| 9974 |
+
"grad_norm": 0.46950235962867737,
|
| 9975 |
+
"learning_rate": 2.915015968686002e-06,
|
| 9976 |
+
"loss": 2.2154,
|
| 9977 |
+
"step": 1424
|
| 9978 |
+
},
|
| 9979 |
+
{
|
| 9980 |
+
"epoch": 0.5936263278483649,
|
| 9981 |
+
"grad_norm": 0.6510283946990967,
|
| 9982 |
+
"learning_rate": 2.8819310509887844e-06,
|
| 9983 |
+
"loss": 2.2183,
|
| 9984 |
+
"step": 1425
|
| 9985 |
+
},
|
| 9986 |
+
{
|
| 9987 |
+
"epoch": 0.5940429077275567,
|
| 9988 |
+
"grad_norm": 0.4346058964729309,
|
| 9989 |
+
"learning_rate": 2.8490293881664187e-06,
|
| 9990 |
+
"loss": 2.2303,
|
| 9991 |
+
"step": 1426
|
| 9992 |
+
},
|
| 9993 |
+
{
|
| 9994 |
+
"epoch": 0.5944594876067486,
|
| 9995 |
+
"grad_norm": 0.46655917167663574,
|
| 9996 |
+
"learning_rate": 2.816311108182368e-06,
|
| 9997 |
+
"loss": 2.2119,
|
| 9998 |
+
"step": 1427
|
| 9999 |
+
},
|
| 10000 |
+
{
|
| 10001 |
+
"epoch": 0.5948760674859405,
|
| 10002 |
+
"grad_norm": 0.4653722941875458,
|
| 10003 |
+
"learning_rate": 2.7837763382868255e-06,
|
| 10004 |
+
"loss": 2.2797,
|
| 10005 |
+
"step": 1428
|
| 10006 |
+
},
|
| 10007 |
+
{
|
| 10008 |
+
"epoch": 0.5952926473651322,
|
| 10009 |
+
"grad_norm": 0.436707466840744,
|
| 10010 |
+
"learning_rate": 2.7514252050163e-06,
|
| 10011 |
+
"loss": 2.1124,
|
| 10012 |
+
"step": 1429
|
| 10013 |
+
},
|
| 10014 |
+
{
|
| 10015 |
+
"epoch": 0.5957092272443241,
|
| 10016 |
+
"grad_norm": 0.48595985770225525,
|
| 10017 |
+
"learning_rate": 2.7192578341930618e-06,
|
| 10018 |
+
"loss": 2.1141,
|
| 10019 |
+
"step": 1430
|
| 10020 |
+
},
|
| 10021 |
+
{
|
| 10022 |
+
"epoch": 0.5961258071235159,
|
| 10023 |
+
"grad_norm": 0.45693573355674744,
|
| 10024 |
+
"learning_rate": 2.6872743509246957e-06,
|
| 10025 |
+
"loss": 2.1458,
|
| 10026 |
+
"step": 1431
|
| 10027 |
+
},
|
| 10028 |
+
{
|
| 10029 |
+
"epoch": 0.5965423870027078,
|
| 10030 |
+
"grad_norm": 0.505688488483429,
|
| 10031 |
+
"learning_rate": 2.6554748796035866e-06,
|
| 10032 |
+
"loss": 2.2593,
|
| 10033 |
+
"step": 1432
|
| 10034 |
+
},
|
| 10035 |
+
{
|
| 10036 |
+
"epoch": 0.5969589668818996,
|
| 10037 |
+
"grad_norm": 0.48864802718162537,
|
| 10038 |
+
"learning_rate": 2.6238595439064484e-06,
|
| 10039 |
+
"loss": 2.2106,
|
| 10040 |
+
"step": 1433
|
| 10041 |
+
},
|
| 10042 |
+
{
|
| 10043 |
+
"epoch": 0.5973755467610914,
|
| 10044 |
+
"grad_norm": 0.49326953291893005,
|
| 10045 |
+
"learning_rate": 2.5924284667938614e-06,
|
| 10046 |
+
"loss": 2.2302,
|
| 10047 |
+
"step": 1434
|
| 10048 |
+
},
|
| 10049 |
+
{
|
| 10050 |
+
"epoch": 0.5977921266402833,
|
| 10051 |
+
"grad_norm": 0.46278807520866394,
|
| 10052 |
+
"learning_rate": 2.561181770509741e-06,
|
| 10053 |
+
"loss": 2.0505,
|
| 10054 |
+
"step": 1435
|
| 10055 |
+
},
|
| 10056 |
+
{
|
| 10057 |
+
"epoch": 0.5982087065194751,
|
| 10058 |
+
"grad_norm": 0.440238356590271,
|
| 10059 |
+
"learning_rate": 2.530119576580936e-06,
|
| 10060 |
+
"loss": 2.2414,
|
| 10061 |
+
"step": 1436
|
| 10062 |
+
},
|
| 10063 |
+
{
|
| 10064 |
+
"epoch": 0.598625286398667,
|
| 10065 |
+
"grad_norm": 0.4914673864841461,
|
| 10066 |
+
"learning_rate": 2.4992420058166765e-06,
|
| 10067 |
+
"loss": 2.1614,
|
| 10068 |
+
"step": 1437
|
| 10069 |
+
},
|
| 10070 |
+
{
|
| 10071 |
+
"epoch": 0.5990418662778588,
|
| 10072 |
+
"grad_norm": 0.4985724985599518,
|
| 10073 |
+
"learning_rate": 2.4685491783081714e-06,
|
| 10074 |
+
"loss": 2.2197,
|
| 10075 |
+
"step": 1438
|
| 10076 |
+
},
|
| 10077 |
+
{
|
| 10078 |
+
"epoch": 0.5994584461570506,
|
| 10079 |
+
"grad_norm": 0.5183117389678955,
|
| 10080 |
+
"learning_rate": 2.4380412134280884e-06,
|
| 10081 |
+
"loss": 2.2139,
|
| 10082 |
+
"step": 1439
|
| 10083 |
+
},
|
| 10084 |
+
{
|
| 10085 |
+
"epoch": 0.5998750260362424,
|
| 10086 |
+
"grad_norm": 0.5157926082611084,
|
| 10087 |
+
"learning_rate": 2.4077182298301416e-06,
|
| 10088 |
+
"loss": 2.3266,
|
| 10089 |
+
"step": 1440
|
| 10090 |
+
},
|
| 10091 |
+
{
|
| 10092 |
+
"epoch": 0.6002916059154343,
|
| 10093 |
+
"grad_norm": 0.5392321348190308,
|
| 10094 |
+
"learning_rate": 2.3775803454485934e-06,
|
| 10095 |
+
"loss": 2.2959,
|
| 10096 |
+
"step": 1441
|
| 10097 |
+
},
|
| 10098 |
+
{
|
| 10099 |
+
"epoch": 0.6007081857946261,
|
| 10100 |
+
"grad_norm": 0.5534506440162659,
|
| 10101 |
+
"learning_rate": 2.347627677497788e-06,
|
| 10102 |
+
"loss": 2.27,
|
| 10103 |
+
"step": 1442
|
| 10104 |
+
},
|
| 10105 |
+
{
|
| 10106 |
+
"epoch": 0.601124765673818,
|
| 10107 |
+
"grad_norm": 0.5261732339859009,
|
| 10108 |
+
"learning_rate": 2.3178603424717395e-06,
|
| 10109 |
+
"loss": 2.2333,
|
| 10110 |
+
"step": 1443
|
| 10111 |
+
},
|
| 10112 |
+
{
|
| 10113 |
+
"epoch": 0.6015413455530098,
|
| 10114 |
+
"grad_norm": 0.5616102814674377,
|
| 10115 |
+
"learning_rate": 2.2882784561436275e-06,
|
| 10116 |
+
"loss": 2.282,
|
| 10117 |
+
"step": 1444
|
| 10118 |
+
},
|
| 10119 |
+
{
|
| 10120 |
+
"epoch": 0.6019579254322016,
|
| 10121 |
+
"grad_norm": 0.5356156229972839,
|
| 10122 |
+
"learning_rate": 2.258882133565404e-06,
|
| 10123 |
+
"loss": 2.0994,
|
| 10124 |
+
"step": 1445
|
| 10125 |
+
},
|
| 10126 |
+
{
|
| 10127 |
+
"epoch": 0.6023745053113935,
|
| 10128 |
+
"grad_norm": 0.5574761629104614,
|
| 10129 |
+
"learning_rate": 2.229671489067281e-06,
|
| 10130 |
+
"loss": 2.2651,
|
| 10131 |
+
"step": 1446
|
| 10132 |
+
},
|
| 10133 |
+
{
|
| 10134 |
+
"epoch": 0.6027910851905853,
|
| 10135 |
+
"grad_norm": 0.561643123626709,
|
| 10136 |
+
"learning_rate": 2.2006466362573306e-06,
|
| 10137 |
+
"loss": 2.2072,
|
| 10138 |
+
"step": 1447
|
| 10139 |
+
},
|
| 10140 |
+
{
|
| 10141 |
+
"epoch": 0.6032076650697771,
|
| 10142 |
+
"grad_norm": 0.5972204208374023,
|
| 10143 |
+
"learning_rate": 2.1718076880210327e-06,
|
| 10144 |
+
"loss": 2.3453,
|
| 10145 |
+
"step": 1448
|
| 10146 |
+
},
|
| 10147 |
+
{
|
| 10148 |
+
"epoch": 0.6036242449489689,
|
| 10149 |
+
"grad_norm": 0.5803166627883911,
|
| 10150 |
+
"learning_rate": 2.143154756520832e-06,
|
| 10151 |
+
"loss": 2.2736,
|
| 10152 |
+
"step": 1449
|
| 10153 |
+
},
|
| 10154 |
+
{
|
| 10155 |
+
"epoch": 0.6040408248281608,
|
| 10156 |
+
"grad_norm": 0.6789401769638062,
|
| 10157 |
+
"learning_rate": 2.1146879531957155e-06,
|
| 10158 |
+
"loss": 2.3797,
|
| 10159 |
+
"step": 1450
|
| 10160 |
+
},
|
| 10161 |
+
{
|
| 10162 |
+
"epoch": 0.6044574047073527,
|
| 10163 |
+
"grad_norm": 0.37630099058151245,
|
| 10164 |
+
"learning_rate": 2.0864073887607416e-06,
|
| 10165 |
+
"loss": 2.0858,
|
| 10166 |
+
"step": 1451
|
| 10167 |
+
},
|
| 10168 |
+
{
|
| 10169 |
+
"epoch": 0.6048739845865445,
|
| 10170 |
+
"grad_norm": 0.3744637370109558,
|
| 10171 |
+
"learning_rate": 2.058313173206661e-06,
|
| 10172 |
+
"loss": 2.1679,
|
| 10173 |
+
"step": 1452
|
| 10174 |
+
},
|
| 10175 |
+
{
|
| 10176 |
+
"epoch": 0.6052905644657363,
|
| 10177 |
+
"grad_norm": 0.3858438730239868,
|
| 10178 |
+
"learning_rate": 2.0304054157994525e-06,
|
| 10179 |
+
"loss": 2.1499,
|
| 10180 |
+
"step": 1453
|
| 10181 |
+
},
|
| 10182 |
+
{
|
| 10183 |
+
"epoch": 0.6057071443449281,
|
| 10184 |
+
"grad_norm": 0.4133443832397461,
|
| 10185 |
+
"learning_rate": 2.0026842250799038e-06,
|
| 10186 |
+
"loss": 2.1605,
|
| 10187 |
+
"step": 1454
|
| 10188 |
+
},
|
| 10189 |
+
{
|
| 10190 |
+
"epoch": 0.60612372422412,
|
| 10191 |
+
"grad_norm": 0.40560242533683777,
|
| 10192 |
+
"learning_rate": 1.975149708863205e-06,
|
| 10193 |
+
"loss": 2.0838,
|
| 10194 |
+
"step": 1455
|
| 10195 |
+
},
|
| 10196 |
+
{
|
| 10197 |
+
"epoch": 0.6065403041033118,
|
| 10198 |
+
"grad_norm": 0.391827791929245,
|
| 10199 |
+
"learning_rate": 1.9478019742385113e-06,
|
| 10200 |
+
"loss": 2.1384,
|
| 10201 |
+
"step": 1456
|
| 10202 |
+
},
|
| 10203 |
+
{
|
| 10204 |
+
"epoch": 0.6069568839825037,
|
| 10205 |
+
"grad_norm": 0.4195277988910675,
|
| 10206 |
+
"learning_rate": 1.92064112756854e-06,
|
| 10207 |
+
"loss": 2.0542,
|
| 10208 |
+
"step": 1457
|
| 10209 |
+
},
|
| 10210 |
+
{
|
| 10211 |
+
"epoch": 0.6073734638616954,
|
| 10212 |
+
"grad_norm": 0.40138891339302063,
|
| 10213 |
+
"learning_rate": 1.893667274489136e-06,
|
| 10214 |
+
"loss": 2.1876,
|
| 10215 |
+
"step": 1458
|
| 10216 |
+
},
|
| 10217 |
+
{
|
| 10218 |
+
"epoch": 0.6077900437408873,
|
| 10219 |
+
"grad_norm": 0.42072543501853943,
|
| 10220 |
+
"learning_rate": 1.866880519908909e-06,
|
| 10221 |
+
"loss": 2.1337,
|
| 10222 |
+
"step": 1459
|
| 10223 |
+
},
|
| 10224 |
+
{
|
| 10225 |
+
"epoch": 0.6082066236200792,
|
| 10226 |
+
"grad_norm": 0.4126233458518982,
|
| 10227 |
+
"learning_rate": 1.8402809680087452e-06,
|
| 10228 |
+
"loss": 2.1252,
|
| 10229 |
+
"step": 1460
|
| 10230 |
+
},
|
| 10231 |
+
{
|
| 10232 |
+
"epoch": 0.608623203499271,
|
| 10233 |
+
"grad_norm": 0.4317125380039215,
|
| 10234 |
+
"learning_rate": 1.8138687222414962e-06,
|
| 10235 |
+
"loss": 2.2389,
|
| 10236 |
+
"step": 1461
|
| 10237 |
+
},
|
| 10238 |
+
{
|
| 10239 |
+
"epoch": 0.6090397833784629,
|
| 10240 |
+
"grad_norm": 0.4322243928909302,
|
| 10241 |
+
"learning_rate": 1.787643885331497e-06,
|
| 10242 |
+
"loss": 2.1097,
|
| 10243 |
+
"step": 1462
|
| 10244 |
+
},
|
| 10245 |
+
{
|
| 10246 |
+
"epoch": 0.6094563632576546,
|
| 10247 |
+
"grad_norm": 0.3994523882865906,
|
| 10248 |
+
"learning_rate": 1.7616065592742038e-06,
|
| 10249 |
+
"loss": 2.1334,
|
| 10250 |
+
"step": 1463
|
| 10251 |
+
},
|
| 10252 |
+
{
|
| 10253 |
+
"epoch": 0.6098729431368465,
|
| 10254 |
+
"grad_norm": 0.439184308052063,
|
| 10255 |
+
"learning_rate": 1.735756845335812e-06,
|
| 10256 |
+
"loss": 2.2396,
|
| 10257 |
+
"step": 1464
|
| 10258 |
+
},
|
| 10259 |
+
{
|
| 10260 |
+
"epoch": 0.6102895230160383,
|
| 10261 |
+
"grad_norm": 0.42550763487815857,
|
| 10262 |
+
"learning_rate": 1.710094844052812e-06,
|
| 10263 |
+
"loss": 2.2612,
|
| 10264 |
+
"step": 1465
|
| 10265 |
+
},
|
| 10266 |
+
{
|
| 10267 |
+
"epoch": 0.6107061028952302,
|
| 10268 |
+
"grad_norm": 0.4296604096889496,
|
| 10269 |
+
"learning_rate": 1.6846206552316612e-06,
|
| 10270 |
+
"loss": 2.1427,
|
| 10271 |
+
"step": 1466
|
| 10272 |
+
},
|
| 10273 |
+
{
|
| 10274 |
+
"epoch": 0.6111226827744219,
|
| 10275 |
+
"grad_norm": 0.47978174686431885,
|
| 10276 |
+
"learning_rate": 1.6593343779483517e-06,
|
| 10277 |
+
"loss": 2.1704,
|
| 10278 |
+
"step": 1467
|
| 10279 |
+
},
|
| 10280 |
+
{
|
| 10281 |
+
"epoch": 0.6115392626536138,
|
| 10282 |
+
"grad_norm": 0.43665987253189087,
|
| 10283 |
+
"learning_rate": 1.6342361105480096e-06,
|
| 10284 |
+
"loss": 2.1245,
|
| 10285 |
+
"step": 1468
|
| 10286 |
+
},
|
| 10287 |
+
{
|
| 10288 |
+
"epoch": 0.6119558425328057,
|
| 10289 |
+
"grad_norm": 0.467197060585022,
|
| 10290 |
+
"learning_rate": 1.6093259506445913e-06,
|
| 10291 |
+
"loss": 2.2824,
|
| 10292 |
+
"step": 1469
|
| 10293 |
+
},
|
| 10294 |
+
{
|
| 10295 |
+
"epoch": 0.6123724224119975,
|
| 10296 |
+
"grad_norm": 0.4877012073993683,
|
| 10297 |
+
"learning_rate": 1.5846039951204096e-06,
|
| 10298 |
+
"loss": 2.1825,
|
| 10299 |
+
"step": 1470
|
| 10300 |
+
},
|
| 10301 |
+
{
|
| 10302 |
+
"epoch": 0.6127890022911894,
|
| 10303 |
+
"grad_norm": 0.42274004220962524,
|
| 10304 |
+
"learning_rate": 1.5600703401258298e-06,
|
| 10305 |
+
"loss": 2.2681,
|
| 10306 |
+
"step": 1471
|
| 10307 |
+
},
|
| 10308 |
+
{
|
| 10309 |
+
"epoch": 0.6132055821703811,
|
| 10310 |
+
"grad_norm": 0.44951850175857544,
|
| 10311 |
+
"learning_rate": 1.5357250810788314e-06,
|
| 10312 |
+
"loss": 2.1808,
|
| 10313 |
+
"step": 1472
|
| 10314 |
+
},
|
| 10315 |
+
{
|
| 10316 |
+
"epoch": 0.613622162049573,
|
| 10317 |
+
"grad_norm": 0.46898871660232544,
|
| 10318 |
+
"learning_rate": 1.5115683126647073e-06,
|
| 10319 |
+
"loss": 2.3322,
|
| 10320 |
+
"step": 1473
|
| 10321 |
+
},
|
| 10322 |
+
{
|
| 10323 |
+
"epoch": 0.6140387419287648,
|
| 10324 |
+
"grad_norm": 0.47944626212120056,
|
| 10325 |
+
"learning_rate": 1.4876001288356312e-06,
|
| 10326 |
+
"loss": 2.2514,
|
| 10327 |
+
"step": 1474
|
| 10328 |
+
},
|
| 10329 |
+
{
|
| 10330 |
+
"epoch": 0.6144553218079567,
|
| 10331 |
+
"grad_norm": 0.472414493560791,
|
| 10332 |
+
"learning_rate": 1.4638206228103413e-06,
|
| 10333 |
+
"loss": 2.1953,
|
| 10334 |
+
"step": 1475
|
| 10335 |
+
},
|
| 10336 |
+
{
|
| 10337 |
+
"epoch": 0.6148719016871486,
|
| 10338 |
+
"grad_norm": 0.4496370255947113,
|
| 10339 |
+
"learning_rate": 1.4402298870737519e-06,
|
| 10340 |
+
"loss": 2.1316,
|
| 10341 |
+
"step": 1476
|
| 10342 |
+
},
|
| 10343 |
+
{
|
| 10344 |
+
"epoch": 0.6152884815663403,
|
| 10345 |
+
"grad_norm": 0.47646018862724304,
|
| 10346 |
+
"learning_rate": 1.416828013376581e-06,
|
| 10347 |
+
"loss": 2.2501,
|
| 10348 |
+
"step": 1477
|
| 10349 |
+
},
|
| 10350 |
+
{
|
| 10351 |
+
"epoch": 0.6157050614455322,
|
| 10352 |
+
"grad_norm": 0.46072840690612793,
|
| 10353 |
+
"learning_rate": 1.39361509273504e-06,
|
| 10354 |
+
"loss": 2.2411,
|
| 10355 |
+
"step": 1478
|
| 10356 |
+
},
|
| 10357 |
+
{
|
| 10358 |
+
"epoch": 0.616121641324724,
|
| 10359 |
+
"grad_norm": 0.4801056385040283,
|
| 10360 |
+
"learning_rate": 1.3705912154304224e-06,
|
| 10361 |
+
"loss": 2.1061,
|
| 10362 |
+
"step": 1479
|
| 10363 |
+
},
|
| 10364 |
+
{
|
| 10365 |
+
"epoch": 0.6165382212039159,
|
| 10366 |
+
"grad_norm": 0.4991438388824463,
|
| 10367 |
+
"learning_rate": 1.3477564710088098e-06,
|
| 10368 |
+
"loss": 2.1523,
|
| 10369 |
+
"step": 1480
|
| 10370 |
+
},
|
| 10371 |
+
{
|
| 10372 |
+
"epoch": 0.6169548010831077,
|
| 10373 |
+
"grad_norm": 0.46082231402397156,
|
| 10374 |
+
"learning_rate": 1.3251109482806666e-06,
|
| 10375 |
+
"loss": 2.0674,
|
| 10376 |
+
"step": 1481
|
| 10377 |
+
},
|
| 10378 |
+
{
|
| 10379 |
+
"epoch": 0.6173713809622995,
|
| 10380 |
+
"grad_norm": 0.5152682065963745,
|
| 10381 |
+
"learning_rate": 1.302654735320552e-06,
|
| 10382 |
+
"loss": 2.1793,
|
| 10383 |
+
"step": 1482
|
| 10384 |
+
},
|
| 10385 |
+
{
|
| 10386 |
+
"epoch": 0.6177879608414913,
|
| 10387 |
+
"grad_norm": 0.47470808029174805,
|
| 10388 |
+
"learning_rate": 1.2803879194667246e-06,
|
| 10389 |
+
"loss": 2.1302,
|
| 10390 |
+
"step": 1483
|
| 10391 |
+
},
|
| 10392 |
+
{
|
| 10393 |
+
"epoch": 0.6182045407206832,
|
| 10394 |
+
"grad_norm": 0.4618239402770996,
|
| 10395 |
+
"learning_rate": 1.2583105873208267e-06,
|
| 10396 |
+
"loss": 2.0966,
|
| 10397 |
+
"step": 1484
|
| 10398 |
+
},
|
| 10399 |
+
{
|
| 10400 |
+
"epoch": 0.6186211205998751,
|
| 10401 |
+
"grad_norm": 0.483643501996994,
|
| 10402 |
+
"learning_rate": 1.2364228247475683e-06,
|
| 10403 |
+
"loss": 2.2631,
|
| 10404 |
+
"step": 1485
|
| 10405 |
+
},
|
| 10406 |
+
{
|
| 10407 |
+
"epoch": 0.6190377004790668,
|
| 10408 |
+
"grad_norm": 0.4786933660507202,
|
| 10409 |
+
"learning_rate": 1.2147247168743547e-06,
|
| 10410 |
+
"loss": 2.0714,
|
| 10411 |
+
"step": 1486
|
| 10412 |
+
},
|
| 10413 |
+
{
|
| 10414 |
+
"epoch": 0.6194542803582587,
|
| 10415 |
+
"grad_norm": 0.5250565409660339,
|
| 10416 |
+
"learning_rate": 1.1932163480909865e-06,
|
| 10417 |
+
"loss": 2.265,
|
| 10418 |
+
"step": 1487
|
| 10419 |
+
},
|
| 10420 |
+
{
|
| 10421 |
+
"epoch": 0.6198708602374505,
|
| 10422 |
+
"grad_norm": 0.5278326272964478,
|
| 10423 |
+
"learning_rate": 1.1718978020492988e-06,
|
| 10424 |
+
"loss": 2.3252,
|
| 10425 |
+
"step": 1488
|
| 10426 |
+
},
|
| 10427 |
+
{
|
| 10428 |
+
"epoch": 0.6202874401166424,
|
| 10429 |
+
"grad_norm": 0.4978008568286896,
|
| 10430 |
+
"learning_rate": 1.150769161662868e-06,
|
| 10431 |
+
"loss": 2.1351,
|
| 10432 |
+
"step": 1489
|
| 10433 |
+
},
|
| 10434 |
+
{
|
| 10435 |
+
"epoch": 0.6207040199958342,
|
| 10436 |
+
"grad_norm": 0.503657341003418,
|
| 10437 |
+
"learning_rate": 1.1298305091066664e-06,
|
| 10438 |
+
"loss": 2.1816,
|
| 10439 |
+
"step": 1490
|
| 10440 |
+
},
|
| 10441 |
+
{
|
| 10442 |
+
"epoch": 0.621120599875026,
|
| 10443 |
+
"grad_norm": 0.4968448281288147,
|
| 10444 |
+
"learning_rate": 1.1090819258167795e-06,
|
| 10445 |
+
"loss": 2.1719,
|
| 10446 |
+
"step": 1491
|
| 10447 |
+
},
|
| 10448 |
+
{
|
| 10449 |
+
"epoch": 0.6215371797542179,
|
| 10450 |
+
"grad_norm": 0.47265180945396423,
|
| 10451 |
+
"learning_rate": 1.0885234924900233e-06,
|
| 10452 |
+
"loss": 2.1128,
|
| 10453 |
+
"step": 1492
|
| 10454 |
+
},
|
| 10455 |
+
{
|
| 10456 |
+
"epoch": 0.6219537596334097,
|
| 10457 |
+
"grad_norm": 0.5009278059005737,
|
| 10458 |
+
"learning_rate": 1.0681552890836944e-06,
|
| 10459 |
+
"loss": 2.1928,
|
| 10460 |
+
"step": 1493
|
| 10461 |
+
},
|
| 10462 |
+
{
|
| 10463 |
+
"epoch": 0.6223703395126016,
|
| 10464 |
+
"grad_norm": 0.5196132063865662,
|
| 10465 |
+
"learning_rate": 1.0479773948152306e-06,
|
| 10466 |
+
"loss": 2.1684,
|
| 10467 |
+
"step": 1494
|
| 10468 |
+
},
|
| 10469 |
+
{
|
| 10470 |
+
"epoch": 0.6227869193917934,
|
| 10471 |
+
"grad_norm": 0.5334997177124023,
|
| 10472 |
+
"learning_rate": 1.027989888161901e-06,
|
| 10473 |
+
"loss": 2.2935,
|
| 10474 |
+
"step": 1495
|
| 10475 |
+
},
|
| 10476 |
+
{
|
| 10477 |
+
"epoch": 0.6232034992709852,
|
| 10478 |
+
"grad_norm": 0.5558013319969177,
|
| 10479 |
+
"learning_rate": 1.0081928468605118e-06,
|
| 10480 |
+
"loss": 2.2524,
|
| 10481 |
+
"step": 1496
|
| 10482 |
+
},
|
| 10483 |
+
{
|
| 10484 |
+
"epoch": 0.623620079150177,
|
| 10485 |
+
"grad_norm": 0.5635661482810974,
|
| 10486 |
+
"learning_rate": 9.88586347907089e-07,
|
| 10487 |
+
"loss": 2.0852,
|
| 10488 |
+
"step": 1497
|
| 10489 |
+
},
|
| 10490 |
+
{
|
| 10491 |
+
"epoch": 0.6240366590293689,
|
| 10492 |
+
"grad_norm": 0.5811219811439514,
|
| 10493 |
+
"learning_rate": 9.691704675565905e-07,
|
| 10494 |
+
"loss": 2.2478,
|
| 10495 |
+
"step": 1498
|
| 10496 |
+
},
|
| 10497 |
+
{
|
| 10498 |
+
"epoch": 0.6244532389085607,
|
| 10499 |
+
"grad_norm": 0.6354938745498657,
|
| 10500 |
+
"learning_rate": 9.499452813226284e-07,
|
| 10501 |
+
"loss": 2.2547,
|
| 10502 |
+
"step": 1499
|
| 10503 |
+
},
|
| 10504 |
+
{
|
| 10505 |
+
"epoch": 0.6248698187877526,
|
| 10506 |
+
"grad_norm": 0.7180959582328796,
|
| 10507 |
+
"learning_rate": 9.309108639771136e-07,
|
| 10508 |
+
"loss": 2.3821,
|
| 10509 |
+
"step": 1500
|
| 10510 |
+
},
|
| 10511 |
+
{
|
| 10512 |
+
"epoch": 0.6252863986669444,
|
| 10513 |
+
"grad_norm": 0.34574657678604126,
|
| 10514 |
+
"learning_rate": 9.120672895500393e-07,
|
| 10515 |
+
"loss": 2.0427,
|
| 10516 |
+
"step": 1501
|
| 10517 |
+
},
|
| 10518 |
+
{
|
| 10519 |
+
"epoch": 0.6257029785461362,
|
| 10520 |
+
"grad_norm": 0.3725878596305847,
|
| 10521 |
+
"learning_rate": 8.934146313291425e-07,
|
| 10522 |
+
"loss": 2.0809,
|
| 10523 |
+
"step": 1502
|
| 10524 |
+
},
|
| 10525 |
+
{
|
| 10526 |
+
"epoch": 0.6261195584253281,
|
| 10527 |
+
"grad_norm": 0.3956831097602844,
|
| 10528 |
+
"learning_rate": 8.749529618596486e-07,
|
| 10529 |
+
"loss": 2.1458,
|
| 10530 |
+
"step": 1503
|
| 10531 |
+
},
|
| 10532 |
+
{
|
| 10533 |
+
"epoch": 0.6265361383045199,
|
| 10534 |
+
"grad_norm": 0.4244229793548584,
|
| 10535 |
+
"learning_rate": 8.56682352943966e-07,
|
| 10536 |
+
"loss": 2.1551,
|
| 10537 |
+
"step": 1504
|
| 10538 |
+
},
|
| 10539 |
+
{
|
| 10540 |
+
"epoch": 0.6269527181837117,
|
| 10541 |
+
"grad_norm": 0.4060150384902954,
|
| 10542 |
+
"learning_rate": 8.386028756414089e-07,
|
| 10543 |
+
"loss": 2.0259,
|
| 10544 |
+
"step": 1505
|
| 10545 |
+
},
|
| 10546 |
+
{
|
| 10547 |
+
"epoch": 0.6273692980629035,
|
| 10548 |
+
"grad_norm": 0.3999885022640228,
|
| 10549 |
+
"learning_rate": 8.207146002679523e-07,
|
| 10550 |
+
"loss": 2.1497,
|
| 10551 |
+
"step": 1506
|
| 10552 |
+
},
|
| 10553 |
+
{
|
| 10554 |
+
"epoch": 0.6277858779420954,
|
| 10555 |
+
"grad_norm": 0.4191332757472992,
|
| 10556 |
+
"learning_rate": 8.030175963959108e-07,
|
| 10557 |
+
"loss": 2.1944,
|
| 10558 |
+
"step": 1507
|
| 10559 |
+
},
|
| 10560 |
+
{
|
| 10561 |
+
"epoch": 0.6282024578212873,
|
| 10562 |
+
"grad_norm": 0.43755659461021423,
|
| 10563 |
+
"learning_rate": 7.855119328537109e-07,
|
| 10564 |
+
"loss": 2.1872,
|
| 10565 |
+
"step": 1508
|
| 10566 |
+
},
|
| 10567 |
+
{
|
| 10568 |
+
"epoch": 0.6286190377004791,
|
| 10569 |
+
"grad_norm": 0.43631330132484436,
|
| 10570 |
+
"learning_rate": 7.681976777255961e-07,
|
| 10571 |
+
"loss": 2.2692,
|
| 10572 |
+
"step": 1509
|
| 10573 |
+
},
|
| 10574 |
+
{
|
| 10575 |
+
"epoch": 0.6290356175796709,
|
| 10576 |
+
"grad_norm": 0.44572728872299194,
|
| 10577 |
+
"learning_rate": 7.510748983513783e-07,
|
| 10578 |
+
"loss": 2.2305,
|
| 10579 |
+
"step": 1510
|
| 10580 |
+
},
|
| 10581 |
+
{
|
| 10582 |
+
"epoch": 0.6294521974588627,
|
| 10583 |
+
"grad_norm": 0.41270124912261963,
|
| 10584 |
+
"learning_rate": 7.341436613261532e-07,
|
| 10585 |
+
"loss": 2.1961,
|
| 10586 |
+
"step": 1511
|
| 10587 |
+
},
|
| 10588 |
+
{
|
| 10589 |
+
"epoch": 0.6298687773380546,
|
| 10590 |
+
"grad_norm": 0.4134480655193329,
|
| 10591 |
+
"learning_rate": 7.174040325000797e-07,
|
| 10592 |
+
"loss": 2.187,
|
| 10593 |
+
"step": 1512
|
| 10594 |
+
},
|
| 10595 |
+
{
|
| 10596 |
+
"epoch": 0.6302853572172464,
|
| 10597 |
+
"grad_norm": 0.43170493841171265,
|
| 10598 |
+
"learning_rate": 7.008560769780792e-07,
|
| 10599 |
+
"loss": 2.12,
|
| 10600 |
+
"step": 1513
|
| 10601 |
+
},
|
| 10602 |
+
{
|
| 10603 |
+
"epoch": 0.6307019370964383,
|
| 10604 |
+
"grad_norm": 0.4855658710002899,
|
| 10605 |
+
"learning_rate": 6.844998591196138e-07,
|
| 10606 |
+
"loss": 2.1298,
|
| 10607 |
+
"step": 1514
|
| 10608 |
+
},
|
| 10609 |
+
{
|
| 10610 |
+
"epoch": 0.63111851697563,
|
| 10611 |
+
"grad_norm": 0.4488075077533722,
|
| 10612 |
+
"learning_rate": 6.683354425384259e-07,
|
| 10613 |
+
"loss": 2.1919,
|
| 10614 |
+
"step": 1515
|
| 10615 |
+
},
|
| 10616 |
+
{
|
| 10617 |
+
"epoch": 0.6315350968548219,
|
| 10618 |
+
"grad_norm": 0.4441848397254944,
|
| 10619 |
+
"learning_rate": 6.523628901022927e-07,
|
| 10620 |
+
"loss": 2.1476,
|
| 10621 |
+
"step": 1516
|
| 10622 |
+
},
|
| 10623 |
+
{
|
| 10624 |
+
"epoch": 0.6319516767340138,
|
| 10625 |
+
"grad_norm": 0.4448429048061371,
|
| 10626 |
+
"learning_rate": 6.365822639327723e-07,
|
| 10627 |
+
"loss": 2.065,
|
| 10628 |
+
"step": 1517
|
| 10629 |
+
},
|
| 10630 |
+
{
|
| 10631 |
+
"epoch": 0.6323682566132056,
|
| 10632 |
+
"grad_norm": 0.4354341924190521,
|
| 10633 |
+
"learning_rate": 6.209936254049809e-07,
|
| 10634 |
+
"loss": 2.1896,
|
| 10635 |
+
"step": 1518
|
| 10636 |
+
},
|
| 10637 |
+
{
|
| 10638 |
+
"epoch": 0.6327848364923974,
|
| 10639 |
+
"grad_norm": 0.44121915102005005,
|
| 10640 |
+
"learning_rate": 6.055970351473261e-07,
|
| 10641 |
+
"loss": 2.0895,
|
| 10642 |
+
"step": 1519
|
| 10643 |
+
},
|
| 10644 |
+
{
|
| 10645 |
+
"epoch": 0.6332014163715892,
|
| 10646 |
+
"grad_norm": 0.44546177983283997,
|
| 10647 |
+
"learning_rate": 5.903925530413135e-07,
|
| 10648 |
+
"loss": 2.0544,
|
| 10649 |
+
"step": 1520
|
| 10650 |
+
},
|
| 10651 |
+
{
|
| 10652 |
+
"epoch": 0.6336179962507811,
|
| 10653 |
+
"grad_norm": 0.4857303202152252,
|
| 10654 |
+
"learning_rate": 5.753802382212625e-07,
|
| 10655 |
+
"loss": 2.1827,
|
| 10656 |
+
"step": 1521
|
| 10657 |
+
},
|
| 10658 |
+
{
|
| 10659 |
+
"epoch": 0.6340345761299729,
|
| 10660 |
+
"grad_norm": 0.4315418004989624,
|
| 10661 |
+
"learning_rate": 5.605601490741241e-07,
|
| 10662 |
+
"loss": 2.1325,
|
| 10663 |
+
"step": 1522
|
| 10664 |
+
},
|
| 10665 |
+
{
|
| 10666 |
+
"epoch": 0.6344511560091648,
|
| 10667 |
+
"grad_norm": 0.46684661507606506,
|
| 10668 |
+
"learning_rate": 5.459323432392083e-07,
|
| 10669 |
+
"loss": 2.1638,
|
| 10670 |
+
"step": 1523
|
| 10671 |
+
},
|
| 10672 |
+
{
|
| 10673 |
+
"epoch": 0.6348677358883565,
|
| 10674 |
+
"grad_norm": 0.49080562591552734,
|
| 10675 |
+
"learning_rate": 5.314968776080009e-07,
|
| 10676 |
+
"loss": 2.2624,
|
| 10677 |
+
"step": 1524
|
| 10678 |
+
},
|
| 10679 |
+
{
|
| 10680 |
+
"epoch": 0.6352843157675484,
|
| 10681 |
+
"grad_norm": 0.4543081820011139,
|
| 10682 |
+
"learning_rate": 5.172538083239197e-07,
|
| 10683 |
+
"loss": 2.218,
|
| 10684 |
+
"step": 1525
|
| 10685 |
+
},
|
| 10686 |
+
{
|
| 10687 |
+
"epoch": 0.6357008956467403,
|
| 10688 |
+
"grad_norm": 0.4865744113922119,
|
| 10689 |
+
"learning_rate": 5.032031907821089e-07,
|
| 10690 |
+
"loss": 2.0499,
|
| 10691 |
+
"step": 1526
|
| 10692 |
+
},
|
| 10693 |
+
{
|
| 10694 |
+
"epoch": 0.6361174755259321,
|
| 10695 |
+
"grad_norm": 0.4952188730239868,
|
| 10696 |
+
"learning_rate": 4.893450796291999e-07,
|
| 10697 |
+
"loss": 2.2702,
|
| 10698 |
+
"step": 1527
|
| 10699 |
+
},
|
| 10700 |
+
{
|
| 10701 |
+
"epoch": 0.636534055405124,
|
| 10702 |
+
"grad_norm": 0.4662850499153137,
|
| 10703 |
+
"learning_rate": 4.7567952876312884e-07,
|
| 10704 |
+
"loss": 2.0191,
|
| 10705 |
+
"step": 1528
|
| 10706 |
+
},
|
| 10707 |
+
{
|
| 10708 |
+
"epoch": 0.6369506352843157,
|
| 10709 |
+
"grad_norm": 0.4703667163848877,
|
| 10710 |
+
"learning_rate": 4.6220659133290325e-07,
|
| 10711 |
+
"loss": 2.1969,
|
| 10712 |
+
"step": 1529
|
| 10713 |
+
},
|
| 10714 |
+
{
|
| 10715 |
+
"epoch": 0.6373672151635076,
|
| 10716 |
+
"grad_norm": 0.4602733850479126,
|
| 10717 |
+
"learning_rate": 4.48926319738413e-07,
|
| 10718 |
+
"loss": 2.1302,
|
| 10719 |
+
"step": 1530
|
| 10720 |
+
},
|
| 10721 |
+
{
|
| 10722 |
+
"epoch": 0.6377837950426994,
|
| 10723 |
+
"grad_norm": 0.49774935841560364,
|
| 10724 |
+
"learning_rate": 4.358387656302143e-07,
|
| 10725 |
+
"loss": 2.1577,
|
| 10726 |
+
"step": 1531
|
| 10727 |
+
},
|
| 10728 |
+
{
|
| 10729 |
+
"epoch": 0.6382003749218913,
|
| 10730 |
+
"grad_norm": 0.4763321876525879,
|
| 10731 |
+
"learning_rate": 4.2294397990932934e-07,
|
| 10732 |
+
"loss": 2.262,
|
| 10733 |
+
"step": 1532
|
| 10734 |
+
},
|
| 10735 |
+
{
|
| 10736 |
+
"epoch": 0.6386169548010832,
|
| 10737 |
+
"grad_norm": 0.5135960578918457,
|
| 10738 |
+
"learning_rate": 4.10242012727069e-07,
|
| 10739 |
+
"loss": 2.1379,
|
| 10740 |
+
"step": 1533
|
| 10741 |
+
},
|
| 10742 |
+
{
|
| 10743 |
+
"epoch": 0.6390335346802749,
|
| 10744 |
+
"grad_norm": 0.4936571419239044,
|
| 10745 |
+
"learning_rate": 3.9773291348479405e-07,
|
| 10746 |
+
"loss": 2.1328,
|
| 10747 |
+
"step": 1534
|
| 10748 |
+
},
|
| 10749 |
+
{
|
| 10750 |
+
"epoch": 0.6394501145594668,
|
| 10751 |
+
"grad_norm": 0.4480665624141693,
|
| 10752 |
+
"learning_rate": 3.854167308337708e-07,
|
| 10753 |
+
"loss": 2.1657,
|
| 10754 |
+
"step": 1535
|
| 10755 |
+
},
|
| 10756 |
+
{
|
| 10757 |
+
"epoch": 0.6398666944386586,
|
| 10758 |
+
"grad_norm": 0.5286336541175842,
|
| 10759 |
+
"learning_rate": 3.732935126749604e-07,
|
| 10760 |
+
"loss": 2.2341,
|
| 10761 |
+
"step": 1536
|
| 10762 |
+
},
|
| 10763 |
+
{
|
| 10764 |
+
"epoch": 0.6402832743178505,
|
| 10765 |
+
"grad_norm": 0.48503080010414124,
|
| 10766 |
+
"learning_rate": 3.613633061588184e-07,
|
| 10767 |
+
"loss": 2.1333,
|
| 10768 |
+
"step": 1537
|
| 10769 |
+
},
|
| 10770 |
+
{
|
| 10771 |
+
"epoch": 0.6406998541970422,
|
| 10772 |
+
"grad_norm": 0.47150424122810364,
|
| 10773 |
+
"learning_rate": 3.4962615768513453e-07,
|
| 10774 |
+
"loss": 2.1284,
|
| 10775 |
+
"step": 1538
|
| 10776 |
+
},
|
| 10777 |
+
{
|
| 10778 |
+
"epoch": 0.6411164340762341,
|
| 10779 |
+
"grad_norm": 0.5344295501708984,
|
| 10780 |
+
"learning_rate": 3.380821129028489e-07,
|
| 10781 |
+
"loss": 2.1414,
|
| 10782 |
+
"step": 1539
|
| 10783 |
+
},
|
| 10784 |
+
{
|
| 10785 |
+
"epoch": 0.641533013955426,
|
| 10786 |
+
"grad_norm": 0.5570894479751587,
|
| 10787 |
+
"learning_rate": 3.2673121670986907e-07,
|
| 10788 |
+
"loss": 2.1852,
|
| 10789 |
+
"step": 1540
|
| 10790 |
+
},
|
| 10791 |
+
{
|
| 10792 |
+
"epoch": 0.6419495938346178,
|
| 10793 |
+
"grad_norm": 0.5436847805976868,
|
| 10794 |
+
"learning_rate": 3.1557351325287586e-07,
|
| 10795 |
+
"loss": 2.1047,
|
| 10796 |
+
"step": 1541
|
| 10797 |
+
},
|
| 10798 |
+
{
|
| 10799 |
+
"epoch": 0.6423661737138097,
|
| 10800 |
+
"grad_norm": 0.515598475933075,
|
| 10801 |
+
"learning_rate": 3.046090459271955e-07,
|
| 10802 |
+
"loss": 2.1226,
|
| 10803 |
+
"step": 1542
|
| 10804 |
+
},
|
| 10805 |
+
{
|
| 10806 |
+
"epoch": 0.6427827535930014,
|
| 10807 |
+
"grad_norm": 0.5310966372489929,
|
| 10808 |
+
"learning_rate": 2.9383785737659985e-07,
|
| 10809 |
+
"loss": 2.1434,
|
| 10810 |
+
"step": 1543
|
| 10811 |
+
},
|
| 10812 |
+
{
|
| 10813 |
+
"epoch": 0.6431993334721933,
|
| 10814 |
+
"grad_norm": 0.52982097864151,
|
| 10815 |
+
"learning_rate": 2.8325998949314536e-07,
|
| 10816 |
+
"loss": 2.1901,
|
| 10817 |
+
"step": 1544
|
| 10818 |
+
},
|
| 10819 |
+
{
|
| 10820 |
+
"epoch": 0.6436159133513851,
|
| 10821 |
+
"grad_norm": 0.533441960811615,
|
| 10822 |
+
"learning_rate": 2.728754834170066e-07,
|
| 10823 |
+
"loss": 2.2461,
|
| 10824 |
+
"step": 1545
|
| 10825 |
+
},
|
| 10826 |
+
{
|
| 10827 |
+
"epoch": 0.644032493230577,
|
| 10828 |
+
"grad_norm": 0.5676113963127136,
|
| 10829 |
+
"learning_rate": 2.626843795363321e-07,
|
| 10830 |
+
"loss": 2.3176,
|
| 10831 |
+
"step": 1546
|
| 10832 |
+
},
|
| 10833 |
+
{
|
| 10834 |
+
"epoch": 0.6444490731097688,
|
| 10835 |
+
"grad_norm": 0.5590258240699768,
|
| 10836 |
+
"learning_rate": 2.526867174870717e-07,
|
| 10837 |
+
"loss": 2.1459,
|
| 10838 |
+
"step": 1547
|
| 10839 |
+
},
|
| 10840 |
+
{
|
| 10841 |
+
"epoch": 0.6448656529889606,
|
| 10842 |
+
"grad_norm": 0.6203082799911499,
|
| 10843 |
+
"learning_rate": 2.428825361528275e-07,
|
| 10844 |
+
"loss": 2.2837,
|
| 10845 |
+
"step": 1548
|
| 10846 |
+
},
|
| 10847 |
+
{
|
| 10848 |
+
"epoch": 0.6452822328681524,
|
| 10849 |
+
"grad_norm": 0.5960795879364014,
|
| 10850 |
+
"learning_rate": 2.3327187366470883e-07,
|
| 10851 |
+
"loss": 2.3547,
|
| 10852 |
+
"step": 1549
|
| 10853 |
+
},
|
| 10854 |
+
{
|
| 10855 |
+
"epoch": 0.6456988127473443,
|
| 10856 |
+
"grad_norm": 0.699540376663208,
|
| 10857 |
+
"learning_rate": 2.238547674011715e-07,
|
| 10858 |
+
"loss": 2.2972,
|
| 10859 |
+
"step": 1550
|
| 10860 |
+
},
|
| 10861 |
+
{
|
| 10862 |
+
"epoch": 0.6461153926265362,
|
| 10863 |
+
"grad_norm": 0.37149208784103394,
|
| 10864 |
+
"learning_rate": 2.1463125398787364e-07,
|
| 10865 |
+
"loss": 2.0706,
|
| 10866 |
+
"step": 1551
|
| 10867 |
+
},
|
| 10868 |
+
{
|
| 10869 |
+
"epoch": 0.646531972505728,
|
| 10870 |
+
"grad_norm": 0.38026919960975647,
|
| 10871 |
+
"learning_rate": 2.0560136929755892e-07,
|
| 10872 |
+
"loss": 2.0399,
|
| 10873 |
+
"step": 1552
|
| 10874 |
+
},
|
| 10875 |
+
{
|
| 10876 |
+
"epoch": 0.6469485523849198,
|
| 10877 |
+
"grad_norm": 0.7115801572799683,
|
| 10878 |
+
"learning_rate": 1.9676514844987337e-07,
|
| 10879 |
+
"loss": 2.0709,
|
| 10880 |
+
"step": 1553
|
| 10881 |
+
},
|
| 10882 |
+
{
|
| 10883 |
+
"epoch": 0.6473651322641116,
|
| 10884 |
+
"grad_norm": 0.38469597697257996,
|
| 10885 |
+
"learning_rate": 1.8812262581126562e-07,
|
| 10886 |
+
"loss": 2.0582,
|
| 10887 |
+
"step": 1554
|
| 10888 |
+
},
|
| 10889 |
+
{
|
| 10890 |
+
"epoch": 0.6477817121433035,
|
| 10891 |
+
"grad_norm": 0.38373446464538574,
|
| 10892 |
+
"learning_rate": 1.796738349948368e-07,
|
| 10893 |
+
"loss": 2.0111,
|
| 10894 |
+
"step": 1555
|
| 10895 |
+
},
|
| 10896 |
+
{
|
| 10897 |
+
"epoch": 0.6481982920224953,
|
| 10898 |
+
"grad_norm": 0.4103526771068573,
|
| 10899 |
+
"learning_rate": 1.714188088602131e-07,
|
| 10900 |
+
"loss": 2.0758,
|
| 10901 |
+
"step": 1556
|
| 10902 |
+
},
|
| 10903 |
+
{
|
| 10904 |
+
"epoch": 0.6486148719016871,
|
| 10905 |
+
"grad_norm": 0.4074265956878662,
|
| 10906 |
+
"learning_rate": 1.633575795134179e-07,
|
| 10907 |
+
"loss": 2.1847,
|
| 10908 |
+
"step": 1557
|
| 10909 |
+
},
|
| 10910 |
+
{
|
| 10911 |
+
"epoch": 0.649031451780879,
|
| 10912 |
+
"grad_norm": 0.4106631577014923,
|
| 10913 |
+
"learning_rate": 1.5549017830674418e-07,
|
| 10914 |
+
"loss": 2.0789,
|
| 10915 |
+
"step": 1558
|
| 10916 |
+
},
|
| 10917 |
+
{
|
| 10918 |
+
"epoch": 0.6494480316600708,
|
| 10919 |
+
"grad_norm": 0.4176217317581177,
|
| 10920 |
+
"learning_rate": 1.4781663583863793e-07,
|
| 10921 |
+
"loss": 2.1993,
|
| 10922 |
+
"step": 1559
|
| 10923 |
+
},
|
| 10924 |
+
{
|
| 10925 |
+
"epoch": 0.6498646115392627,
|
| 10926 |
+
"grad_norm": 0.4449746012687683,
|
| 10927 |
+
"learning_rate": 1.4033698195357048e-07,
|
| 10928 |
+
"loss": 2.1613,
|
| 10929 |
+
"step": 1560
|
| 10930 |
+
},
|
| 10931 |
+
{
|
| 10932 |
+
"epoch": 0.6502811914184545,
|
| 10933 |
+
"grad_norm": 0.40782734751701355,
|
| 10934 |
+
"learning_rate": 1.3305124574193305e-07,
|
| 10935 |
+
"loss": 2.2605,
|
| 10936 |
+
"step": 1561
|
| 10937 |
+
},
|
| 10938 |
+
{
|
| 10939 |
+
"epoch": 0.6506977712976463,
|
| 10940 |
+
"grad_norm": 0.4028443992137909,
|
| 10941 |
+
"learning_rate": 1.2595945553992573e-07,
|
| 10942 |
+
"loss": 2.0443,
|
| 10943 |
+
"step": 1562
|
| 10944 |
+
},
|
| 10945 |
+
{
|
| 10946 |
+
"epoch": 0.6511143511768381,
|
| 10947 |
+
"grad_norm": 0.4169761836528778,
|
| 10948 |
+
"learning_rate": 1.1906163892942967e-07,
|
| 10949 |
+
"loss": 2.0504,
|
| 10950 |
+
"step": 1563
|
| 10951 |
+
},
|
| 10952 |
+
{
|
| 10953 |
+
"epoch": 0.65153093105603,
|
| 10954 |
+
"grad_norm": 0.3908756673336029,
|
| 10955 |
+
"learning_rate": 1.123578227379074e-07,
|
| 10956 |
+
"loss": 2.1973,
|
| 10957 |
+
"step": 1564
|
| 10958 |
+
},
|
| 10959 |
+
{
|
| 10960 |
+
"epoch": 0.6519475109352219,
|
| 10961 |
+
"grad_norm": 0.4227580726146698,
|
| 10962 |
+
"learning_rate": 1.0584803303831381e-07,
|
| 10963 |
+
"loss": 2.1167,
|
| 10964 |
+
"step": 1565
|
| 10965 |
+
},
|
| 10966 |
+
{
|
| 10967 |
+
"epoch": 0.6523640908144137,
|
| 10968 |
+
"grad_norm": 0.4246922731399536,
|
| 10969 |
+
"learning_rate": 9.953229514897411e-08,
|
| 10970 |
+
"loss": 2.1459,
|
| 10971 |
+
"step": 1566
|
| 10972 |
+
},
|
| 10973 |
+
{
|
| 10974 |
+
"epoch": 0.6527806706936055,
|
| 10975 |
+
"grad_norm": 0.4376825988292694,
|
| 10976 |
+
"learning_rate": 9.341063363349501e-08,
|
| 10977 |
+
"loss": 2.1834,
|
| 10978 |
+
"step": 1567
|
| 10979 |
+
},
|
| 10980 |
+
{
|
| 10981 |
+
"epoch": 0.6531972505727973,
|
| 10982 |
+
"grad_norm": 0.43723276257514954,
|
| 10983 |
+
"learning_rate": 8.748307230067588e-08,
|
| 10984 |
+
"loss": 2.165,
|
| 10985 |
+
"step": 1568
|
| 10986 |
+
},
|
| 10987 |
+
{
|
| 10988 |
+
"epoch": 0.6536138304519892,
|
| 10989 |
+
"grad_norm": 0.44601333141326904,
|
| 10990 |
+
"learning_rate": 8.174963420439218e-08,
|
| 10991 |
+
"loss": 2.1801,
|
| 10992 |
+
"step": 1569
|
| 10993 |
+
},
|
| 10994 |
+
{
|
| 10995 |
+
"epoch": 0.654030410331181,
|
| 10996 |
+
"grad_norm": 0.4556887745857239,
|
| 10997 |
+
"learning_rate": 7.621034164353447e-08,
|
| 10998 |
+
"loss": 2.2955,
|
| 10999 |
+
"step": 1570
|
| 11000 |
+
},
|
| 11001 |
+
{
|
| 11002 |
+
"epoch": 0.6544469902103729,
|
| 11003 |
+
"grad_norm": 0.48959264159202576,
|
| 11004 |
+
"learning_rate": 7.086521616190279e-08,
|
| 11005 |
+
"loss": 2.2596,
|
| 11006 |
+
"step": 1571
|
| 11007 |
+
},
|
| 11008 |
+
{
|
| 11009 |
+
"epoch": 0.6548635700895646,
|
| 11010 |
+
"grad_norm": 0.47076231241226196,
|
| 11011 |
+
"learning_rate": 6.571427854813461e-08,
|
| 11012 |
+
"loss": 2.0837,
|
| 11013 |
+
"step": 1572
|
| 11014 |
+
},
|
| 11015 |
+
{
|
| 11016 |
+
"epoch": 0.6552801499687565,
|
| 11017 |
+
"grad_norm": 0.46404242515563965,
|
| 11018 |
+
"learning_rate": 6.075754883561602e-08,
|
| 11019 |
+
"loss": 2.0958,
|
| 11020 |
+
"step": 1573
|
| 11021 |
+
},
|
| 11022 |
+
{
|
| 11023 |
+
"epoch": 0.6556967298479484,
|
| 11024 |
+
"grad_norm": 0.4356656074523926,
|
| 11025 |
+
"learning_rate": 5.599504630239838e-08,
|
| 11026 |
+
"loss": 2.1906,
|
| 11027 |
+
"step": 1574
|
| 11028 |
+
},
|
| 11029 |
+
{
|
| 11030 |
+
"epoch": 0.6561133097271402,
|
| 11031 |
+
"grad_norm": 0.48961523175239563,
|
| 11032 |
+
"learning_rate": 5.1426789471142876e-08,
|
| 11033 |
+
"loss": 2.1936,
|
| 11034 |
+
"step": 1575
|
| 11035 |
+
},
|
| 11036 |
+
{
|
| 11037 |
+
"epoch": 0.656529889606332,
|
| 11038 |
+
"grad_norm": 0.4800525903701782,
|
| 11039 |
+
"learning_rate": 4.705279610903168e-08,
|
| 11040 |
+
"loss": 2.1767,
|
| 11041 |
+
"step": 1576
|
| 11042 |
+
},
|
| 11043 |
+
{
|
| 11044 |
+
"epoch": 0.6569464694855238,
|
| 11045 |
+
"grad_norm": 0.461904913187027,
|
| 11046 |
+
"learning_rate": 4.287308322770689e-08,
|
| 11047 |
+
"loss": 2.0721,
|
| 11048 |
+
"step": 1577
|
| 11049 |
+
},
|
| 11050 |
+
{
|
| 11051 |
+
"epoch": 0.6573630493647157,
|
| 11052 |
+
"grad_norm": 0.48440995812416077,
|
| 11053 |
+
"learning_rate": 3.8887667083198356e-08,
|
| 11054 |
+
"loss": 2.2816,
|
| 11055 |
+
"step": 1578
|
| 11056 |
+
},
|
| 11057 |
+
{
|
| 11058 |
+
"epoch": 0.6577796292439075,
|
| 11059 |
+
"grad_norm": 0.4660238027572632,
|
| 11060 |
+
"learning_rate": 3.509656317585708e-08,
|
| 11061 |
+
"loss": 2.1412,
|
| 11062 |
+
"step": 1579
|
| 11063 |
+
},
|
| 11064 |
+
{
|
| 11065 |
+
"epoch": 0.6581962091230994,
|
| 11066 |
+
"grad_norm": 0.4756726324558258,
|
| 11067 |
+
"learning_rate": 3.149978625032191e-08,
|
| 11068 |
+
"loss": 2.0979,
|
| 11069 |
+
"step": 1580
|
| 11070 |
+
},
|
| 11071 |
+
{
|
| 11072 |
+
"epoch": 0.6586127890022911,
|
| 11073 |
+
"grad_norm": 0.44779303669929504,
|
| 11074 |
+
"learning_rate": 2.809735029541405e-08,
|
| 11075 |
+
"loss": 2.2678,
|
| 11076 |
+
"step": 1581
|
| 11077 |
+
},
|
| 11078 |
+
{
|
| 11079 |
+
"epoch": 0.659029368881483,
|
| 11080 |
+
"grad_norm": 0.4502517879009247,
|
| 11081 |
+
"learning_rate": 2.4889268544131538e-08,
|
| 11082 |
+
"loss": 2.1124,
|
| 11083 |
+
"step": 1582
|
| 11084 |
+
},
|
| 11085 |
+
{
|
| 11086 |
+
"epoch": 0.6594459487606749,
|
| 11087 |
+
"grad_norm": 0.501974880695343,
|
| 11088 |
+
"learning_rate": 2.1875553473565956e-08,
|
| 11089 |
+
"loss": 2.1945,
|
| 11090 |
+
"step": 1583
|
| 11091 |
+
},
|
| 11092 |
+
{
|
| 11093 |
+
"epoch": 0.6598625286398667,
|
| 11094 |
+
"grad_norm": 0.48524612188339233,
|
| 11095 |
+
"learning_rate": 1.9056216804874683e-08,
|
| 11096 |
+
"loss": 2.2147,
|
| 11097 |
+
"step": 1584
|
| 11098 |
+
},
|
| 11099 |
+
{
|
| 11100 |
+
"epoch": 0.6602791085190586,
|
| 11101 |
+
"grad_norm": 0.47237899899482727,
|
| 11102 |
+
"learning_rate": 1.6431269503208724e-08,
|
| 11103 |
+
"loss": 2.1895,
|
| 11104 |
+
"step": 1585
|
| 11105 |
+
},
|
| 11106 |
+
{
|
| 11107 |
+
"epoch": 0.6606956883982503,
|
| 11108 |
+
"grad_norm": 0.514725923538208,
|
| 11109 |
+
"learning_rate": 1.400072177770162e-08,
|
| 11110 |
+
"loss": 2.161,
|
| 11111 |
+
"step": 1586
|
| 11112 |
+
},
|
| 11113 |
+
{
|
| 11114 |
+
"epoch": 0.6611122682774422,
|
| 11115 |
+
"grad_norm": 0.528115451335907,
|
| 11116 |
+
"learning_rate": 1.1764583081408375e-08,
|
| 11117 |
+
"loss": 2.2763,
|
| 11118 |
+
"step": 1587
|
| 11119 |
+
},
|
| 11120 |
+
{
|
| 11121 |
+
"epoch": 0.661528848156634,
|
| 11122 |
+
"grad_norm": 0.510019063949585,
|
| 11123 |
+
"learning_rate": 9.722862111277709e-09,
|
| 11124 |
+
"loss": 2.1913,
|
| 11125 |
+
"step": 1588
|
| 11126 |
+
},
|
| 11127 |
+
{
|
| 11128 |
+
"epoch": 0.6619454280358259,
|
| 11129 |
+
"grad_norm": 0.5159466862678528,
|
| 11130 |
+
"learning_rate": 7.875566808107637e-09,
|
| 11131 |
+
"loss": 2.2889,
|
| 11132 |
+
"step": 1589
|
| 11133 |
+
},
|
| 11134 |
+
{
|
| 11135 |
+
"epoch": 0.6623620079150178,
|
| 11136 |
+
"grad_norm": 0.5133629441261292,
|
| 11137 |
+
"learning_rate": 6.2227043565288305e-09,
|
| 11138 |
+
"loss": 2.2818,
|
| 11139 |
+
"step": 1590
|
| 11140 |
+
},
|
| 11141 |
+
{
|
| 11142 |
+
"epoch": 0.6627785877942095,
|
| 11143 |
+
"grad_norm": 0.5215743184089661,
|
| 11144 |
+
"learning_rate": 4.764281184971298e-09,
|
| 11145 |
+
"loss": 2.1326,
|
| 11146 |
+
"step": 1591
|
| 11147 |
+
},
|
| 11148 |
+
{
|
| 11149 |
+
"epoch": 0.6631951676734014,
|
| 11150 |
+
"grad_norm": 0.5067861676216125,
|
| 11151 |
+
"learning_rate": 3.500302965636637e-09,
|
| 11152 |
+
"loss": 2.2611,
|
| 11153 |
+
"step": 1592
|
| 11154 |
+
},
|
| 11155 |
+
{
|
| 11156 |
+
"epoch": 0.6636117475525932,
|
| 11157 |
+
"grad_norm": 0.5310625433921814,
|
| 11158 |
+
"learning_rate": 2.430774614470277e-09,
|
| 11159 |
+
"loss": 2.1011,
|
| 11160 |
+
"step": 1593
|
| 11161 |
+
},
|
| 11162 |
+
{
|
| 11163 |
+
"epoch": 0.6640283274317851,
|
| 11164 |
+
"grad_norm": 0.5508378148078918,
|
| 11165 |
+
"learning_rate": 1.5557002911670281e-09,
|
| 11166 |
+
"loss": 2.1273,
|
| 11167 |
+
"step": 1594
|
| 11168 |
+
},
|
| 11169 |
+
{
|
| 11170 |
+
"epoch": 0.6644449073109768,
|
| 11171 |
+
"grad_norm": 0.5138291120529175,
|
| 11172 |
+
"learning_rate": 8.750833991155727e-10,
|
| 11173 |
+
"loss": 2.2155,
|
| 11174 |
+
"step": 1595
|
| 11175 |
+
},
|
| 11176 |
+
{
|
| 11177 |
+
"epoch": 0.6648614871901687,
|
| 11178 |
+
"grad_norm": 0.5404055118560791,
|
| 11179 |
+
"learning_rate": 3.889265854317703e-10,
|
| 11180 |
+
"loss": 2.1549,
|
| 11181 |
+
"step": 1596
|
| 11182 |
+
},
|
| 11183 |
+
{
|
| 11184 |
+
"epoch": 0.6652780670693605,
|
| 11185 |
+
"grad_norm": 0.56361985206604,
|
| 11186 |
+
"learning_rate": 9.723174089759646e-11,
|
| 11187 |
+
"loss": 2.0773,
|
| 11188 |
+
"step": 1597
|
| 11189 |
+
},
|
| 11190 |
+
{
|
| 11191 |
+
"epoch": 0.6656946469485524,
|
| 11192 |
+
"grad_norm": 0.6100237965583801,
|
| 11193 |
+
"learning_rate": 0.0,
|
| 11194 |
+
"loss": 2.3351,
|
| 11195 |
+
"step": 1598
|
| 11196 |
}
|
| 11197 |
],
|
| 11198 |
"logging_steps": 1,
|
|
|
|
| 11207 |
"should_evaluate": false,
|
| 11208 |
"should_log": false,
|
| 11209 |
"should_save": true,
|
| 11210 |
+
"should_training_stop": true
|
| 11211 |
},
|
| 11212 |
"attributes": {}
|
| 11213 |
}
|
| 11214 |
},
|
| 11215 |
+
"total_flos": 1.0412593904363766e+18,
|
| 11216 |
"train_batch_size": 4,
|
| 11217 |
"trial_name": null,
|
| 11218 |
"trial_params": null
|