Training in progress, step 1518, 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 45118424
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:21721f0ecd25362e2d9945d534a0dac6cdc1b2ee2198f8a32d4aaa5d87a4d3a1
|
| 3 |
size 45118424
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 90365754
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3670ee1b34b37df440e6ffbc094e7a6ca040c206fd1ade54ba1227d4d61c920b
|
| 3 |
size 90365754
|
last-checkpoint/rng_state_0.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14512
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e341a407d013bd3d8c299b459b942e0f91461288855ccf151918636c56b9eabc
|
| 3 |
size 14512
|
last-checkpoint/rng_state_1.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14512
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:dbdd3b2f34ab6ba4fa2c76ff3eee45d49fde9b3bed1d7bfbbe63d769d607e9d8
|
| 3 |
size 14512
|
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 sha256:5f3f9099f6a15efc79b02559da6b6f60fddd0d0728d78a5f70d7f733dcd70f55
|
| 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": 380,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -8019,6 +8019,2652 @@
|
|
| 8019 |
"eval_samples_per_second": 17.926,
|
| 8020 |
"eval_steps_per_second": 4.485,
|
| 8021 |
"step": 1140
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 8022 |
}
|
| 8023 |
],
|
| 8024 |
"logging_steps": 1,
|
|
@@ -8033,12 +10679,12 @@
|
|
| 8033 |
"should_evaluate": false,
|
| 8034 |
"should_log": false,
|
| 8035 |
"should_save": true,
|
| 8036 |
-
"should_training_stop":
|
| 8037 |
},
|
| 8038 |
"attributes": {}
|
| 8039 |
}
|
| 8040 |
},
|
| 8041 |
-
"total_flos":
|
| 8042 |
"train_batch_size": 2,
|
| 8043 |
"trial_name": null,
|
| 8044 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.9994239157271007,
|
| 5 |
"eval_steps": 380,
|
| 6 |
+
"global_step": 1518,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 8019 |
"eval_samples_per_second": 17.926,
|
| 8020 |
"eval_steps_per_second": 4.485,
|
| 8021 |
"step": 1140
|
| 8022 |
+
},
|
| 8023 |
+
{
|
| 8024 |
+
"epoch": 0.7512138918607522,
|
| 8025 |
+
"grad_norm": 0.7555129528045654,
|
| 8026 |
+
"learning_rate": 3.081864134739381e-05,
|
| 8027 |
+
"loss": 1.1989,
|
| 8028 |
+
"step": 1141
|
| 8029 |
+
},
|
| 8030 |
+
{
|
| 8031 |
+
"epoch": 0.7518722738869229,
|
| 8032 |
+
"grad_norm": 0.8739767670631409,
|
| 8033 |
+
"learning_rate": 3.0664272015046734e-05,
|
| 8034 |
+
"loss": 1.2435,
|
| 8035 |
+
"step": 1142
|
| 8036 |
+
},
|
| 8037 |
+
{
|
| 8038 |
+
"epoch": 0.7525306559130935,
|
| 8039 |
+
"grad_norm": 0.9945182204246521,
|
| 8040 |
+
"learning_rate": 3.0510220227165474e-05,
|
| 8041 |
+
"loss": 1.1476,
|
| 8042 |
+
"step": 1143
|
| 8043 |
+
},
|
| 8044 |
+
{
|
| 8045 |
+
"epoch": 0.7531890379392643,
|
| 8046 |
+
"grad_norm": 0.9394963383674622,
|
| 8047 |
+
"learning_rate": 3.0356486689277995e-05,
|
| 8048 |
+
"loss": 0.8853,
|
| 8049 |
+
"step": 1144
|
| 8050 |
+
},
|
| 8051 |
+
{
|
| 8052 |
+
"epoch": 0.7538474199654349,
|
| 8053 |
+
"grad_norm": 1.3489819765090942,
|
| 8054 |
+
"learning_rate": 3.0203072105454622e-05,
|
| 8055 |
+
"loss": 1.1426,
|
| 8056 |
+
"step": 1145
|
| 8057 |
+
},
|
| 8058 |
+
{
|
| 8059 |
+
"epoch": 0.7545058019916057,
|
| 8060 |
+
"grad_norm": 1.6051266193389893,
|
| 8061 |
+
"learning_rate": 3.0049977178305076e-05,
|
| 8062 |
+
"loss": 1.5071,
|
| 8063 |
+
"step": 1146
|
| 8064 |
+
},
|
| 8065 |
+
{
|
| 8066 |
+
"epoch": 0.7551641840177763,
|
| 8067 |
+
"grad_norm": 1.4647694826126099,
|
| 8068 |
+
"learning_rate": 2.9897202608975074e-05,
|
| 8069 |
+
"loss": 1.1429,
|
| 8070 |
+
"step": 1147
|
| 8071 |
+
},
|
| 8072 |
+
{
|
| 8073 |
+
"epoch": 0.755822566043947,
|
| 8074 |
+
"grad_norm": 1.5680222511291504,
|
| 8075 |
+
"learning_rate": 2.974474909714304e-05,
|
| 8076 |
+
"loss": 1.1338,
|
| 8077 |
+
"step": 1148
|
| 8078 |
+
},
|
| 8079 |
+
{
|
| 8080 |
+
"epoch": 0.7564809480701177,
|
| 8081 |
+
"grad_norm": 1.7426420450210571,
|
| 8082 |
+
"learning_rate": 2.959261734101716e-05,
|
| 8083 |
+
"loss": 0.9512,
|
| 8084 |
+
"step": 1149
|
| 8085 |
+
},
|
| 8086 |
+
{
|
| 8087 |
+
"epoch": 0.7571393300962884,
|
| 8088 |
+
"grad_norm": 2.067566156387329,
|
| 8089 |
+
"learning_rate": 2.9440808037331967e-05,
|
| 8090 |
+
"loss": 1.0109,
|
| 8091 |
+
"step": 1150
|
| 8092 |
+
},
|
| 8093 |
+
{
|
| 8094 |
+
"epoch": 0.757797712122459,
|
| 8095 |
+
"grad_norm": 0.2545592188835144,
|
| 8096 |
+
"learning_rate": 2.9289321881345254e-05,
|
| 8097 |
+
"loss": 1.0832,
|
| 8098 |
+
"step": 1151
|
| 8099 |
+
},
|
| 8100 |
+
{
|
| 8101 |
+
"epoch": 0.7584560941486297,
|
| 8102 |
+
"grad_norm": 0.27604377269744873,
|
| 8103 |
+
"learning_rate": 2.9138159566834834e-05,
|
| 8104 |
+
"loss": 1.2362,
|
| 8105 |
+
"step": 1152
|
| 8106 |
+
},
|
| 8107 |
+
{
|
| 8108 |
+
"epoch": 0.7591144761748004,
|
| 8109 |
+
"grad_norm": 0.2910035252571106,
|
| 8110 |
+
"learning_rate": 2.8987321786095412e-05,
|
| 8111 |
+
"loss": 1.2334,
|
| 8112 |
+
"step": 1153
|
| 8113 |
+
},
|
| 8114 |
+
{
|
| 8115 |
+
"epoch": 0.7597728582009711,
|
| 8116 |
+
"grad_norm": 0.29931461811065674,
|
| 8117 |
+
"learning_rate": 2.8836809229935357e-05,
|
| 8118 |
+
"loss": 1.1605,
|
| 8119 |
+
"step": 1154
|
| 8120 |
+
},
|
| 8121 |
+
{
|
| 8122 |
+
"epoch": 0.7604312402271418,
|
| 8123 |
+
"grad_norm": 0.32173511385917664,
|
| 8124 |
+
"learning_rate": 2.8686622587673594e-05,
|
| 8125 |
+
"loss": 1.2719,
|
| 8126 |
+
"step": 1155
|
| 8127 |
+
},
|
| 8128 |
+
{
|
| 8129 |
+
"epoch": 0.7610896222533124,
|
| 8130 |
+
"grad_norm": 0.3663305342197418,
|
| 8131 |
+
"learning_rate": 2.8536762547136464e-05,
|
| 8132 |
+
"loss": 1.2495,
|
| 8133 |
+
"step": 1156
|
| 8134 |
+
},
|
| 8135 |
+
{
|
| 8136 |
+
"epoch": 0.7617480042794832,
|
| 8137 |
+
"grad_norm": 0.3718562126159668,
|
| 8138 |
+
"learning_rate": 2.8387229794654423e-05,
|
| 8139 |
+
"loss": 1.2236,
|
| 8140 |
+
"step": 1157
|
| 8141 |
+
},
|
| 8142 |
+
{
|
| 8143 |
+
"epoch": 0.7624063863056538,
|
| 8144 |
+
"grad_norm": 0.39510485529899597,
|
| 8145 |
+
"learning_rate": 2.823802501505909e-05,
|
| 8146 |
+
"loss": 1.2276,
|
| 8147 |
+
"step": 1158
|
| 8148 |
+
},
|
| 8149 |
+
{
|
| 8150 |
+
"epoch": 0.7630647683318246,
|
| 8151 |
+
"grad_norm": 0.4048973321914673,
|
| 8152 |
+
"learning_rate": 2.8089148891680084e-05,
|
| 8153 |
+
"loss": 1.2046,
|
| 8154 |
+
"step": 1159
|
| 8155 |
+
},
|
| 8156 |
+
{
|
| 8157 |
+
"epoch": 0.7637231503579952,
|
| 8158 |
+
"grad_norm": 0.42129042744636536,
|
| 8159 |
+
"learning_rate": 2.7940602106341707e-05,
|
| 8160 |
+
"loss": 1.3255,
|
| 8161 |
+
"step": 1160
|
| 8162 |
+
},
|
| 8163 |
+
{
|
| 8164 |
+
"epoch": 0.764381532384166,
|
| 8165 |
+
"grad_norm": 0.5140630602836609,
|
| 8166 |
+
"learning_rate": 2.7792385339360038e-05,
|
| 8167 |
+
"loss": 1.3431,
|
| 8168 |
+
"step": 1161
|
| 8169 |
+
},
|
| 8170 |
+
{
|
| 8171 |
+
"epoch": 0.7650399144103366,
|
| 8172 |
+
"grad_norm": 0.5049022436141968,
|
| 8173 |
+
"learning_rate": 2.7644499269539724e-05,
|
| 8174 |
+
"loss": 1.2449,
|
| 8175 |
+
"step": 1162
|
| 8176 |
+
},
|
| 8177 |
+
{
|
| 8178 |
+
"epoch": 0.7656982964365073,
|
| 8179 |
+
"grad_norm": 0.5614619851112366,
|
| 8180 |
+
"learning_rate": 2.749694457417089e-05,
|
| 8181 |
+
"loss": 1.336,
|
| 8182 |
+
"step": 1163
|
| 8183 |
+
},
|
| 8184 |
+
{
|
| 8185 |
+
"epoch": 0.766356678462678,
|
| 8186 |
+
"grad_norm": 0.6224168539047241,
|
| 8187 |
+
"learning_rate": 2.7349721929026005e-05,
|
| 8188 |
+
"loss": 1.2038,
|
| 8189 |
+
"step": 1164
|
| 8190 |
+
},
|
| 8191 |
+
{
|
| 8192 |
+
"epoch": 0.7670150604888486,
|
| 8193 |
+
"grad_norm": 0.6956164240837097,
|
| 8194 |
+
"learning_rate": 2.720283200835686e-05,
|
| 8195 |
+
"loss": 1.3149,
|
| 8196 |
+
"step": 1165
|
| 8197 |
+
},
|
| 8198 |
+
{
|
| 8199 |
+
"epoch": 0.7676734425150193,
|
| 8200 |
+
"grad_norm": 0.7140865921974182,
|
| 8201 |
+
"learning_rate": 2.7056275484891304e-05,
|
| 8202 |
+
"loss": 1.2284,
|
| 8203 |
+
"step": 1166
|
| 8204 |
+
},
|
| 8205 |
+
{
|
| 8206 |
+
"epoch": 0.76833182454119,
|
| 8207 |
+
"grad_norm": 0.8704789876937866,
|
| 8208 |
+
"learning_rate": 2.691005302983045e-05,
|
| 8209 |
+
"loss": 1.1158,
|
| 8210 |
+
"step": 1167
|
| 8211 |
+
},
|
| 8212 |
+
{
|
| 8213 |
+
"epoch": 0.7689902065673607,
|
| 8214 |
+
"grad_norm": 1.1140766143798828,
|
| 8215 |
+
"learning_rate": 2.6764165312845403e-05,
|
| 8216 |
+
"loss": 1.1269,
|
| 8217 |
+
"step": 1168
|
| 8218 |
+
},
|
| 8219 |
+
{
|
| 8220 |
+
"epoch": 0.7696485885935314,
|
| 8221 |
+
"grad_norm": 1.0424233675003052,
|
| 8222 |
+
"learning_rate": 2.661861300207412e-05,
|
| 8223 |
+
"loss": 0.944,
|
| 8224 |
+
"step": 1169
|
| 8225 |
+
},
|
| 8226 |
+
{
|
| 8227 |
+
"epoch": 0.7703069706197021,
|
| 8228 |
+
"grad_norm": 1.2212202548980713,
|
| 8229 |
+
"learning_rate": 2.6473396764118575e-05,
|
| 8230 |
+
"loss": 1.1607,
|
| 8231 |
+
"step": 1170
|
| 8232 |
+
},
|
| 8233 |
+
{
|
| 8234 |
+
"epoch": 0.7709653526458727,
|
| 8235 |
+
"grad_norm": 1.0277799367904663,
|
| 8236 |
+
"learning_rate": 2.6328517264041563e-05,
|
| 8237 |
+
"loss": 0.7129,
|
| 8238 |
+
"step": 1171
|
| 8239 |
+
},
|
| 8240 |
+
{
|
| 8241 |
+
"epoch": 0.7716237346720435,
|
| 8242 |
+
"grad_norm": 1.411064624786377,
|
| 8243 |
+
"learning_rate": 2.618397516536367e-05,
|
| 8244 |
+
"loss": 1.1172,
|
| 8245 |
+
"step": 1172
|
| 8246 |
+
},
|
| 8247 |
+
{
|
| 8248 |
+
"epoch": 0.7722821166982141,
|
| 8249 |
+
"grad_norm": 1.4100395441055298,
|
| 8250 |
+
"learning_rate": 2.6039771130060254e-05,
|
| 8251 |
+
"loss": 0.8726,
|
| 8252 |
+
"step": 1173
|
| 8253 |
+
},
|
| 8254 |
+
{
|
| 8255 |
+
"epoch": 0.7729404987243849,
|
| 8256 |
+
"grad_norm": 1.6474452018737793,
|
| 8257 |
+
"learning_rate": 2.589590581855843e-05,
|
| 8258 |
+
"loss": 0.8718,
|
| 8259 |
+
"step": 1174
|
| 8260 |
+
},
|
| 8261 |
+
{
|
| 8262 |
+
"epoch": 0.7735988807505555,
|
| 8263 |
+
"grad_norm": 1.7511224746704102,
|
| 8264 |
+
"learning_rate": 2.575237988973398e-05,
|
| 8265 |
+
"loss": 0.7599,
|
| 8266 |
+
"step": 1175
|
| 8267 |
+
},
|
| 8268 |
+
{
|
| 8269 |
+
"epoch": 0.7742572627767262,
|
| 8270 |
+
"grad_norm": 0.2579965889453888,
|
| 8271 |
+
"learning_rate": 2.560919400090843e-05,
|
| 8272 |
+
"loss": 1.2273,
|
| 8273 |
+
"step": 1176
|
| 8274 |
+
},
|
| 8275 |
+
{
|
| 8276 |
+
"epoch": 0.7749156448028969,
|
| 8277 |
+
"grad_norm": 0.2835389971733093,
|
| 8278 |
+
"learning_rate": 2.5466348807845908e-05,
|
| 8279 |
+
"loss": 1.2614,
|
| 8280 |
+
"step": 1177
|
| 8281 |
+
},
|
| 8282 |
+
{
|
| 8283 |
+
"epoch": 0.7755740268290676,
|
| 8284 |
+
"grad_norm": 0.29296889901161194,
|
| 8285 |
+
"learning_rate": 2.53238449647503e-05,
|
| 8286 |
+
"loss": 1.2371,
|
| 8287 |
+
"step": 1178
|
| 8288 |
+
},
|
| 8289 |
+
{
|
| 8290 |
+
"epoch": 0.7762324088552383,
|
| 8291 |
+
"grad_norm": 0.32429298758506775,
|
| 8292 |
+
"learning_rate": 2.5181683124262136e-05,
|
| 8293 |
+
"loss": 1.1937,
|
| 8294 |
+
"step": 1179
|
| 8295 |
+
},
|
| 8296 |
+
{
|
| 8297 |
+
"epoch": 0.7768907908814089,
|
| 8298 |
+
"grad_norm": 0.3148968815803528,
|
| 8299 |
+
"learning_rate": 2.503986393745564e-05,
|
| 8300 |
+
"loss": 1.1598,
|
| 8301 |
+
"step": 1180
|
| 8302 |
+
},
|
| 8303 |
+
{
|
| 8304 |
+
"epoch": 0.7775491729075796,
|
| 8305 |
+
"grad_norm": 0.3182394206523895,
|
| 8306 |
+
"learning_rate": 2.4898388053835763e-05,
|
| 8307 |
+
"loss": 1.1057,
|
| 8308 |
+
"step": 1181
|
| 8309 |
+
},
|
| 8310 |
+
{
|
| 8311 |
+
"epoch": 0.7782075549337503,
|
| 8312 |
+
"grad_norm": 0.34069356322288513,
|
| 8313 |
+
"learning_rate": 2.4757256121335183e-05,
|
| 8314 |
+
"loss": 1.2295,
|
| 8315 |
+
"step": 1182
|
| 8316 |
+
},
|
| 8317 |
+
{
|
| 8318 |
+
"epoch": 0.778865936959921,
|
| 8319 |
+
"grad_norm": 0.3777032494544983,
|
| 8320 |
+
"learning_rate": 2.461646878631133e-05,
|
| 8321 |
+
"loss": 1.404,
|
| 8322 |
+
"step": 1183
|
| 8323 |
+
},
|
| 8324 |
+
{
|
| 8325 |
+
"epoch": 0.7795243189860916,
|
| 8326 |
+
"grad_norm": 0.3982040584087372,
|
| 8327 |
+
"learning_rate": 2.4476026693543484e-05,
|
| 8328 |
+
"loss": 1.2168,
|
| 8329 |
+
"step": 1184
|
| 8330 |
+
},
|
| 8331 |
+
{
|
| 8332 |
+
"epoch": 0.7801827010122624,
|
| 8333 |
+
"grad_norm": 0.4333396553993225,
|
| 8334 |
+
"learning_rate": 2.4335930486229774e-05,
|
| 8335 |
+
"loss": 1.2611,
|
| 8336 |
+
"step": 1185
|
| 8337 |
+
},
|
| 8338 |
+
{
|
| 8339 |
+
"epoch": 0.780841083038433,
|
| 8340 |
+
"grad_norm": 0.46400609612464905,
|
| 8341 |
+
"learning_rate": 2.419618080598417e-05,
|
| 8342 |
+
"loss": 1.3248,
|
| 8343 |
+
"step": 1186
|
| 8344 |
+
},
|
| 8345 |
+
{
|
| 8346 |
+
"epoch": 0.7814994650646038,
|
| 8347 |
+
"grad_norm": 0.49693915247917175,
|
| 8348 |
+
"learning_rate": 2.405677829283366e-05,
|
| 8349 |
+
"loss": 1.3477,
|
| 8350 |
+
"step": 1187
|
| 8351 |
+
},
|
| 8352 |
+
{
|
| 8353 |
+
"epoch": 0.7821578470907744,
|
| 8354 |
+
"grad_norm": 0.5477811694145203,
|
| 8355 |
+
"learning_rate": 2.3917723585215357e-05,
|
| 8356 |
+
"loss": 1.2802,
|
| 8357 |
+
"step": 1188
|
| 8358 |
+
},
|
| 8359 |
+
{
|
| 8360 |
+
"epoch": 0.7828162291169452,
|
| 8361 |
+
"grad_norm": 0.5534879565238953,
|
| 8362 |
+
"learning_rate": 2.377901731997335e-05,
|
| 8363 |
+
"loss": 1.2442,
|
| 8364 |
+
"step": 1189
|
| 8365 |
+
},
|
| 8366 |
+
{
|
| 8367 |
+
"epoch": 0.7834746111431158,
|
| 8368 |
+
"grad_norm": 0.6362003087997437,
|
| 8369 |
+
"learning_rate": 2.3640660132355996e-05,
|
| 8370 |
+
"loss": 1.2242,
|
| 8371 |
+
"step": 1190
|
| 8372 |
+
},
|
| 8373 |
+
{
|
| 8374 |
+
"epoch": 0.7841329931692865,
|
| 8375 |
+
"grad_norm": 0.726051390171051,
|
| 8376 |
+
"learning_rate": 2.3502652656012947e-05,
|
| 8377 |
+
"loss": 1.2611,
|
| 8378 |
+
"step": 1191
|
| 8379 |
+
},
|
| 8380 |
+
{
|
| 8381 |
+
"epoch": 0.7847913751954572,
|
| 8382 |
+
"grad_norm": 0.8183614015579224,
|
| 8383 |
+
"learning_rate": 2.3364995522992227e-05,
|
| 8384 |
+
"loss": 1.1556,
|
| 8385 |
+
"step": 1192
|
| 8386 |
+
},
|
| 8387 |
+
{
|
| 8388 |
+
"epoch": 0.7854497572216278,
|
| 8389 |
+
"grad_norm": 0.8726836442947388,
|
| 8390 |
+
"learning_rate": 2.3227689363737382e-05,
|
| 8391 |
+
"loss": 1.2289,
|
| 8392 |
+
"step": 1193
|
| 8393 |
+
},
|
| 8394 |
+
{
|
| 8395 |
+
"epoch": 0.7861081392477985,
|
| 8396 |
+
"grad_norm": 1.0491002798080444,
|
| 8397 |
+
"learning_rate": 2.3090734807084548e-05,
|
| 8398 |
+
"loss": 1.3026,
|
| 8399 |
+
"step": 1194
|
| 8400 |
+
},
|
| 8401 |
+
{
|
| 8402 |
+
"epoch": 0.7867665212739692,
|
| 8403 |
+
"grad_norm": 1.1141643524169922,
|
| 8404 |
+
"learning_rate": 2.2954132480259504e-05,
|
| 8405 |
+
"loss": 0.9071,
|
| 8406 |
+
"step": 1195
|
| 8407 |
+
},
|
| 8408 |
+
{
|
| 8409 |
+
"epoch": 0.7874249033001399,
|
| 8410 |
+
"grad_norm": 1.2619067430496216,
|
| 8411 |
+
"learning_rate": 2.2817883008875063e-05,
|
| 8412 |
+
"loss": 1.0773,
|
| 8413 |
+
"step": 1196
|
| 8414 |
+
},
|
| 8415 |
+
{
|
| 8416 |
+
"epoch": 0.7880832853263106,
|
| 8417 |
+
"grad_norm": 1.916133165359497,
|
| 8418 |
+
"learning_rate": 2.2681987016927908e-05,
|
| 8419 |
+
"loss": 1.3466,
|
| 8420 |
+
"step": 1197
|
| 8421 |
+
},
|
| 8422 |
+
{
|
| 8423 |
+
"epoch": 0.7887416673524813,
|
| 8424 |
+
"grad_norm": 1.8914225101470947,
|
| 8425 |
+
"learning_rate": 2.2546445126795822e-05,
|
| 8426 |
+
"loss": 1.0075,
|
| 8427 |
+
"step": 1198
|
| 8428 |
+
},
|
| 8429 |
+
{
|
| 8430 |
+
"epoch": 0.7894000493786519,
|
| 8431 |
+
"grad_norm": 1.9101247787475586,
|
| 8432 |
+
"learning_rate": 2.2411257959234966e-05,
|
| 8433 |
+
"loss": 1.0738,
|
| 8434 |
+
"step": 1199
|
| 8435 |
+
},
|
| 8436 |
+
{
|
| 8437 |
+
"epoch": 0.7900584314048227,
|
| 8438 |
+
"grad_norm": 2.5399341583251953,
|
| 8439 |
+
"learning_rate": 2.227642613337686e-05,
|
| 8440 |
+
"loss": 1.1081,
|
| 8441 |
+
"step": 1200
|
| 8442 |
+
},
|
| 8443 |
+
{
|
| 8444 |
+
"epoch": 0.7907168134309933,
|
| 8445 |
+
"grad_norm": 0.2590787708759308,
|
| 8446 |
+
"learning_rate": 2.2141950266725687e-05,
|
| 8447 |
+
"loss": 1.1662,
|
| 8448 |
+
"step": 1201
|
| 8449 |
+
},
|
| 8450 |
+
{
|
| 8451 |
+
"epoch": 0.7913751954571641,
|
| 8452 |
+
"grad_norm": 0.31000667810440063,
|
| 8453 |
+
"learning_rate": 2.2007830975155363e-05,
|
| 8454 |
+
"loss": 1.2154,
|
| 8455 |
+
"step": 1202
|
| 8456 |
+
},
|
| 8457 |
+
{
|
| 8458 |
+
"epoch": 0.7920335774833347,
|
| 8459 |
+
"grad_norm": 0.28687340021133423,
|
| 8460 |
+
"learning_rate": 2.1874068872906772e-05,
|
| 8461 |
+
"loss": 1.1468,
|
| 8462 |
+
"step": 1203
|
| 8463 |
+
},
|
| 8464 |
+
{
|
| 8465 |
+
"epoch": 0.7926919595095054,
|
| 8466 |
+
"grad_norm": 0.3054833710193634,
|
| 8467 |
+
"learning_rate": 2.1740664572584946e-05,
|
| 8468 |
+
"loss": 1.1847,
|
| 8469 |
+
"step": 1204
|
| 8470 |
+
},
|
| 8471 |
+
{
|
| 8472 |
+
"epoch": 0.7933503415356761,
|
| 8473 |
+
"grad_norm": 0.32457491755485535,
|
| 8474 |
+
"learning_rate": 2.160761868515623e-05,
|
| 8475 |
+
"loss": 1.2377,
|
| 8476 |
+
"step": 1205
|
| 8477 |
+
},
|
| 8478 |
+
{
|
| 8479 |
+
"epoch": 0.7940087235618467,
|
| 8480 |
+
"grad_norm": 0.3366299867630005,
|
| 8481 |
+
"learning_rate": 2.1474931819945553e-05,
|
| 8482 |
+
"loss": 1.284,
|
| 8483 |
+
"step": 1206
|
| 8484 |
+
},
|
| 8485 |
+
{
|
| 8486 |
+
"epoch": 0.7946671055880175,
|
| 8487 |
+
"grad_norm": 0.3629007339477539,
|
| 8488 |
+
"learning_rate": 2.1342604584633506e-05,
|
| 8489 |
+
"loss": 1.2582,
|
| 8490 |
+
"step": 1207
|
| 8491 |
+
},
|
| 8492 |
+
{
|
| 8493 |
+
"epoch": 0.7953254876141881,
|
| 8494 |
+
"grad_norm": 0.3729316294193268,
|
| 8495 |
+
"learning_rate": 2.1210637585253703e-05,
|
| 8496 |
+
"loss": 1.2428,
|
| 8497 |
+
"step": 1208
|
| 8498 |
+
},
|
| 8499 |
+
{
|
| 8500 |
+
"epoch": 0.7959838696403588,
|
| 8501 |
+
"grad_norm": 0.390794575214386,
|
| 8502 |
+
"learning_rate": 2.107903142619e-05,
|
| 8503 |
+
"loss": 1.2041,
|
| 8504 |
+
"step": 1209
|
| 8505 |
+
},
|
| 8506 |
+
{
|
| 8507 |
+
"epoch": 0.7966422516665295,
|
| 8508 |
+
"grad_norm": 0.40834668278694153,
|
| 8509 |
+
"learning_rate": 2.0947786710173546e-05,
|
| 8510 |
+
"loss": 1.2392,
|
| 8511 |
+
"step": 1210
|
| 8512 |
+
},
|
| 8513 |
+
{
|
| 8514 |
+
"epoch": 0.7973006336927002,
|
| 8515 |
+
"grad_norm": 0.43642354011535645,
|
| 8516 |
+
"learning_rate": 2.081690403828024e-05,
|
| 8517 |
+
"loss": 1.2946,
|
| 8518 |
+
"step": 1211
|
| 8519 |
+
},
|
| 8520 |
+
{
|
| 8521 |
+
"epoch": 0.7979590157188708,
|
| 8522 |
+
"grad_norm": 0.46954917907714844,
|
| 8523 |
+
"learning_rate": 2.0686384009927838e-05,
|
| 8524 |
+
"loss": 1.3417,
|
| 8525 |
+
"step": 1212
|
| 8526 |
+
},
|
| 8527 |
+
{
|
| 8528 |
+
"epoch": 0.7986173977450416,
|
| 8529 |
+
"grad_norm": 0.5424888134002686,
|
| 8530 |
+
"learning_rate": 2.0556227222873313e-05,
|
| 8531 |
+
"loss": 1.4782,
|
| 8532 |
+
"step": 1213
|
| 8533 |
+
},
|
| 8534 |
+
{
|
| 8535 |
+
"epoch": 0.7992757797712122,
|
| 8536 |
+
"grad_norm": 0.5052427053451538,
|
| 8537 |
+
"learning_rate": 2.0426434273210014e-05,
|
| 8538 |
+
"loss": 1.18,
|
| 8539 |
+
"step": 1214
|
| 8540 |
+
},
|
| 8541 |
+
{
|
| 8542 |
+
"epoch": 0.799934161797383,
|
| 8543 |
+
"grad_norm": 0.5371694564819336,
|
| 8544 |
+
"learning_rate": 2.029700575536504e-05,
|
| 8545 |
+
"loss": 1.2642,
|
| 8546 |
+
"step": 1215
|
| 8547 |
+
},
|
| 8548 |
+
{
|
| 8549 |
+
"epoch": 0.8005925438235536,
|
| 8550 |
+
"grad_norm": 0.6594541668891907,
|
| 8551 |
+
"learning_rate": 2.0167942262096316e-05,
|
| 8552 |
+
"loss": 1.3036,
|
| 8553 |
+
"step": 1216
|
| 8554 |
+
},
|
| 8555 |
+
{
|
| 8556 |
+
"epoch": 0.8012509258497243,
|
| 8557 |
+
"grad_norm": 0.7874505519866943,
|
| 8558 |
+
"learning_rate": 2.003924438449023e-05,
|
| 8559 |
+
"loss": 1.1885,
|
| 8560 |
+
"step": 1217
|
| 8561 |
+
},
|
| 8562 |
+
{
|
| 8563 |
+
"epoch": 0.801909307875895,
|
| 8564 |
+
"grad_norm": 0.955467164516449,
|
| 8565 |
+
"learning_rate": 1.9910912711958618e-05,
|
| 8566 |
+
"loss": 1.2362,
|
| 8567 |
+
"step": 1218
|
| 8568 |
+
},
|
| 8569 |
+
{
|
| 8570 |
+
"epoch": 0.8025676899020657,
|
| 8571 |
+
"grad_norm": 0.9550561904907227,
|
| 8572 |
+
"learning_rate": 1.9782947832236097e-05,
|
| 8573 |
+
"loss": 1.037,
|
| 8574 |
+
"step": 1219
|
| 8575 |
+
},
|
| 8576 |
+
{
|
| 8577 |
+
"epoch": 0.8032260719282364,
|
| 8578 |
+
"grad_norm": 1.281804084777832,
|
| 8579 |
+
"learning_rate": 1.9655350331377566e-05,
|
| 8580 |
+
"loss": 1.0645,
|
| 8581 |
+
"step": 1220
|
| 8582 |
+
},
|
| 8583 |
+
{
|
| 8584 |
+
"epoch": 0.803884453954407,
|
| 8585 |
+
"grad_norm": 1.2864665985107422,
|
| 8586 |
+
"learning_rate": 1.9528120793755322e-05,
|
| 8587 |
+
"loss": 1.0086,
|
| 8588 |
+
"step": 1221
|
| 8589 |
+
},
|
| 8590 |
+
{
|
| 8591 |
+
"epoch": 0.8045428359805777,
|
| 8592 |
+
"grad_norm": 1.5257987976074219,
|
| 8593 |
+
"learning_rate": 1.9401259802056492e-05,
|
| 8594 |
+
"loss": 1.1066,
|
| 8595 |
+
"step": 1222
|
| 8596 |
+
},
|
| 8597 |
+
{
|
| 8598 |
+
"epoch": 0.8052012180067484,
|
| 8599 |
+
"grad_norm": 1.65569007396698,
|
| 8600 |
+
"learning_rate": 1.9274767937280357e-05,
|
| 8601 |
+
"loss": 1.2644,
|
| 8602 |
+
"step": 1223
|
| 8603 |
+
},
|
| 8604 |
+
{
|
| 8605 |
+
"epoch": 0.8058596000329191,
|
| 8606 |
+
"grad_norm": 1.5913068056106567,
|
| 8607 |
+
"learning_rate": 1.9148645778735554e-05,
|
| 8608 |
+
"loss": 0.7979,
|
| 8609 |
+
"step": 1224
|
| 8610 |
+
},
|
| 8611 |
+
{
|
| 8612 |
+
"epoch": 0.8065179820590898,
|
| 8613 |
+
"grad_norm": 2.411531448364258,
|
| 8614 |
+
"learning_rate": 1.9022893904037676e-05,
|
| 8615 |
+
"loss": 1.2982,
|
| 8616 |
+
"step": 1225
|
| 8617 |
+
},
|
| 8618 |
+
{
|
| 8619 |
+
"epoch": 0.8071763640852605,
|
| 8620 |
+
"grad_norm": 0.24755673110485077,
|
| 8621 |
+
"learning_rate": 1.889751288910645e-05,
|
| 8622 |
+
"loss": 1.1338,
|
| 8623 |
+
"step": 1226
|
| 8624 |
+
},
|
| 8625 |
+
{
|
| 8626 |
+
"epoch": 0.8078347461114311,
|
| 8627 |
+
"grad_norm": 0.2823069989681244,
|
| 8628 |
+
"learning_rate": 1.8772503308163046e-05,
|
| 8629 |
+
"loss": 1.2658,
|
| 8630 |
+
"step": 1227
|
| 8631 |
+
},
|
| 8632 |
+
{
|
| 8633 |
+
"epoch": 0.8084931281376019,
|
| 8634 |
+
"grad_norm": 0.28120023012161255,
|
| 8635 |
+
"learning_rate": 1.8647865733727643e-05,
|
| 8636 |
+
"loss": 1.1407,
|
| 8637 |
+
"step": 1228
|
| 8638 |
+
},
|
| 8639 |
+
{
|
| 8640 |
+
"epoch": 0.8091515101637725,
|
| 8641 |
+
"grad_norm": 0.321236789226532,
|
| 8642 |
+
"learning_rate": 1.8523600736616632e-05,
|
| 8643 |
+
"loss": 1.2701,
|
| 8644 |
+
"step": 1229
|
| 8645 |
+
},
|
| 8646 |
+
{
|
| 8647 |
+
"epoch": 0.8098098921899433,
|
| 8648 |
+
"grad_norm": 0.33734390139579773,
|
| 8649 |
+
"learning_rate": 1.8399708885940136e-05,
|
| 8650 |
+
"loss": 1.1863,
|
| 8651 |
+
"step": 1230
|
| 8652 |
+
},
|
| 8653 |
+
{
|
| 8654 |
+
"epoch": 0.8104682742161139,
|
| 8655 |
+
"grad_norm": 0.33143654465675354,
|
| 8656 |
+
"learning_rate": 1.8276190749099264e-05,
|
| 8657 |
+
"loss": 1.315,
|
| 8658 |
+
"step": 1231
|
| 8659 |
+
},
|
| 8660 |
+
{
|
| 8661 |
+
"epoch": 0.8111266562422846,
|
| 8662 |
+
"grad_norm": 0.3370499610900879,
|
| 8663 |
+
"learning_rate": 1.8153046891783652e-05,
|
| 8664 |
+
"loss": 1.204,
|
| 8665 |
+
"step": 1232
|
| 8666 |
+
},
|
| 8667 |
+
{
|
| 8668 |
+
"epoch": 0.8117850382684553,
|
| 8669 |
+
"grad_norm": 0.37255147099494934,
|
| 8670 |
+
"learning_rate": 1.8030277877968772e-05,
|
| 8671 |
+
"loss": 1.2116,
|
| 8672 |
+
"step": 1233
|
| 8673 |
+
},
|
| 8674 |
+
{
|
| 8675 |
+
"epoch": 0.8124434202946259,
|
| 8676 |
+
"grad_norm": 0.39773163199424744,
|
| 8677 |
+
"learning_rate": 1.790788426991339e-05,
|
| 8678 |
+
"loss": 1.3115,
|
| 8679 |
+
"step": 1234
|
| 8680 |
+
},
|
| 8681 |
+
{
|
| 8682 |
+
"epoch": 0.8131018023207967,
|
| 8683 |
+
"grad_norm": 0.4204883873462677,
|
| 8684 |
+
"learning_rate": 1.778586662815701e-05,
|
| 8685 |
+
"loss": 1.3576,
|
| 8686 |
+
"step": 1235
|
| 8687 |
+
},
|
| 8688 |
+
{
|
| 8689 |
+
"epoch": 0.8137601843469673,
|
| 8690 |
+
"grad_norm": 0.48859432339668274,
|
| 8691 |
+
"learning_rate": 1.7664225511517198e-05,
|
| 8692 |
+
"loss": 1.2549,
|
| 8693 |
+
"step": 1236
|
| 8694 |
+
},
|
| 8695 |
+
{
|
| 8696 |
+
"epoch": 0.814418566373138,
|
| 8697 |
+
"grad_norm": 0.4854777157306671,
|
| 8698 |
+
"learning_rate": 1.7542961477087172e-05,
|
| 8699 |
+
"loss": 1.3323,
|
| 8700 |
+
"step": 1237
|
| 8701 |
+
},
|
| 8702 |
+
{
|
| 8703 |
+
"epoch": 0.8150769483993087,
|
| 8704 |
+
"grad_norm": 0.5650511980056763,
|
| 8705 |
+
"learning_rate": 1.742207508023327e-05,
|
| 8706 |
+
"loss": 1.2028,
|
| 8707 |
+
"step": 1238
|
| 8708 |
+
},
|
| 8709 |
+
{
|
| 8710 |
+
"epoch": 0.8157353304254794,
|
| 8711 |
+
"grad_norm": 0.6029946208000183,
|
| 8712 |
+
"learning_rate": 1.7301566874592134e-05,
|
| 8713 |
+
"loss": 1.36,
|
| 8714 |
+
"step": 1239
|
| 8715 |
+
},
|
| 8716 |
+
{
|
| 8717 |
+
"epoch": 0.81639371245165,
|
| 8718 |
+
"grad_norm": 0.6795003414154053,
|
| 8719 |
+
"learning_rate": 1.718143741206849e-05,
|
| 8720 |
+
"loss": 1.3337,
|
| 8721 |
+
"step": 1240
|
| 8722 |
+
},
|
| 8723 |
+
{
|
| 8724 |
+
"epoch": 0.8170520944778208,
|
| 8725 |
+
"grad_norm": 0.7911986112594604,
|
| 8726 |
+
"learning_rate": 1.7061687242832468e-05,
|
| 8727 |
+
"loss": 1.3073,
|
| 8728 |
+
"step": 1241
|
| 8729 |
+
},
|
| 8730 |
+
{
|
| 8731 |
+
"epoch": 0.8177104765039914,
|
| 8732 |
+
"grad_norm": 0.9688619375228882,
|
| 8733 |
+
"learning_rate": 1.694231691531709e-05,
|
| 8734 |
+
"loss": 1.3579,
|
| 8735 |
+
"step": 1242
|
| 8736 |
+
},
|
| 8737 |
+
{
|
| 8738 |
+
"epoch": 0.8183688585301622,
|
| 8739 |
+
"grad_norm": 0.9323916435241699,
|
| 8740 |
+
"learning_rate": 1.6823326976215792e-05,
|
| 8741 |
+
"loss": 1.1159,
|
| 8742 |
+
"step": 1243
|
| 8743 |
+
},
|
| 8744 |
+
{
|
| 8745 |
+
"epoch": 0.8190272405563328,
|
| 8746 |
+
"grad_norm": 1.2399168014526367,
|
| 8747 |
+
"learning_rate": 1.670471797047992e-05,
|
| 8748 |
+
"loss": 1.2605,
|
| 8749 |
+
"step": 1244
|
| 8750 |
+
},
|
| 8751 |
+
{
|
| 8752 |
+
"epoch": 0.8196856225825035,
|
| 8753 |
+
"grad_norm": 1.2477198839187622,
|
| 8754 |
+
"learning_rate": 1.6586490441316116e-05,
|
| 8755 |
+
"loss": 1.3426,
|
| 8756 |
+
"step": 1245
|
| 8757 |
+
},
|
| 8758 |
+
{
|
| 8759 |
+
"epoch": 0.8203440046086742,
|
| 8760 |
+
"grad_norm": 1.4674113988876343,
|
| 8761 |
+
"learning_rate": 1.6468644930184095e-05,
|
| 8762 |
+
"loss": 1.0689,
|
| 8763 |
+
"step": 1246
|
| 8764 |
+
},
|
| 8765 |
+
{
|
| 8766 |
+
"epoch": 0.8210023866348448,
|
| 8767 |
+
"grad_norm": 1.4392623901367188,
|
| 8768 |
+
"learning_rate": 1.6351181976793894e-05,
|
| 8769 |
+
"loss": 1.0795,
|
| 8770 |
+
"step": 1247
|
| 8771 |
+
},
|
| 8772 |
+
{
|
| 8773 |
+
"epoch": 0.8216607686610156,
|
| 8774 |
+
"grad_norm": 1.4645551443099976,
|
| 8775 |
+
"learning_rate": 1.62341021191035e-05,
|
| 8776 |
+
"loss": 1.2043,
|
| 8777 |
+
"step": 1248
|
| 8778 |
+
},
|
| 8779 |
+
{
|
| 8780 |
+
"epoch": 0.8223191506871862,
|
| 8781 |
+
"grad_norm": 1.7416354417800903,
|
| 8782 |
+
"learning_rate": 1.611740589331645e-05,
|
| 8783 |
+
"loss": 0.9595,
|
| 8784 |
+
"step": 1249
|
| 8785 |
+
},
|
| 8786 |
+
{
|
| 8787 |
+
"epoch": 0.8229775327133569,
|
| 8788 |
+
"grad_norm": 1.685441255569458,
|
| 8789 |
+
"learning_rate": 1.6001093833879287e-05,
|
| 8790 |
+
"loss": 0.7698,
|
| 8791 |
+
"step": 1250
|
| 8792 |
+
},
|
| 8793 |
+
{
|
| 8794 |
+
"epoch": 0.8236359147395276,
|
| 8795 |
+
"grad_norm": 0.25341877341270447,
|
| 8796 |
+
"learning_rate": 1.588516647347916e-05,
|
| 8797 |
+
"loss": 1.2026,
|
| 8798 |
+
"step": 1251
|
| 8799 |
+
},
|
| 8800 |
+
{
|
| 8801 |
+
"epoch": 0.8242942967656983,
|
| 8802 |
+
"grad_norm": 0.27512264251708984,
|
| 8803 |
+
"learning_rate": 1.5769624343041355e-05,
|
| 8804 |
+
"loss": 1.2053,
|
| 8805 |
+
"step": 1252
|
| 8806 |
+
},
|
| 8807 |
+
{
|
| 8808 |
+
"epoch": 0.824952678791869,
|
| 8809 |
+
"grad_norm": 0.28133538365364075,
|
| 8810 |
+
"learning_rate": 1.5654467971726894e-05,
|
| 8811 |
+
"loss": 1.1509,
|
| 8812 |
+
"step": 1253
|
| 8813 |
+
},
|
| 8814 |
+
{
|
| 8815 |
+
"epoch": 0.8256110608180397,
|
| 8816 |
+
"grad_norm": 0.3063061535358429,
|
| 8817 |
+
"learning_rate": 1.553969788693008e-05,
|
| 8818 |
+
"loss": 1.1995,
|
| 8819 |
+
"step": 1254
|
| 8820 |
+
},
|
| 8821 |
+
{
|
| 8822 |
+
"epoch": 0.8262694428442103,
|
| 8823 |
+
"grad_norm": 0.3312765955924988,
|
| 8824 |
+
"learning_rate": 1.542531461427611e-05,
|
| 8825 |
+
"loss": 1.2657,
|
| 8826 |
+
"step": 1255
|
| 8827 |
+
},
|
| 8828 |
+
{
|
| 8829 |
+
"epoch": 0.8269278248703811,
|
| 8830 |
+
"grad_norm": 0.3586176335811615,
|
| 8831 |
+
"learning_rate": 1.531131867761866e-05,
|
| 8832 |
+
"loss": 1.1941,
|
| 8833 |
+
"step": 1256
|
| 8834 |
+
},
|
| 8835 |
+
{
|
| 8836 |
+
"epoch": 0.8275862068965517,
|
| 8837 |
+
"grad_norm": 0.36552894115448,
|
| 8838 |
+
"learning_rate": 1.5197710599037429e-05,
|
| 8839 |
+
"loss": 1.198,
|
| 8840 |
+
"step": 1257
|
| 8841 |
+
},
|
| 8842 |
+
{
|
| 8843 |
+
"epoch": 0.8282445889227225,
|
| 8844 |
+
"grad_norm": 0.384895920753479,
|
| 8845 |
+
"learning_rate": 1.5084490898835856e-05,
|
| 8846 |
+
"loss": 1.2629,
|
| 8847 |
+
"step": 1258
|
| 8848 |
+
},
|
| 8849 |
+
{
|
| 8850 |
+
"epoch": 0.8289029709488931,
|
| 8851 |
+
"grad_norm": 0.3710484802722931,
|
| 8852 |
+
"learning_rate": 1.497166009553871e-05,
|
| 8853 |
+
"loss": 1.2126,
|
| 8854 |
+
"step": 1259
|
| 8855 |
+
},
|
| 8856 |
+
{
|
| 8857 |
+
"epoch": 0.8295613529750637,
|
| 8858 |
+
"grad_norm": 0.4181410074234009,
|
| 8859 |
+
"learning_rate": 1.4859218705889588e-05,
|
| 8860 |
+
"loss": 1.2037,
|
| 8861 |
+
"step": 1260
|
| 8862 |
+
},
|
| 8863 |
+
{
|
| 8864 |
+
"epoch": 0.8302197350012345,
|
| 8865 |
+
"grad_norm": 0.4357868731021881,
|
| 8866 |
+
"learning_rate": 1.4747167244848714e-05,
|
| 8867 |
+
"loss": 1.185,
|
| 8868 |
+
"step": 1261
|
| 8869 |
+
},
|
| 8870 |
+
{
|
| 8871 |
+
"epoch": 0.8308781170274051,
|
| 8872 |
+
"grad_norm": 0.489761620759964,
|
| 8873 |
+
"learning_rate": 1.4635506225590512e-05,
|
| 8874 |
+
"loss": 1.3106,
|
| 8875 |
+
"step": 1262
|
| 8876 |
+
},
|
| 8877 |
+
{
|
| 8878 |
+
"epoch": 0.8315364990535759,
|
| 8879 |
+
"grad_norm": 0.5292978882789612,
|
| 8880 |
+
"learning_rate": 1.4524236159501226e-05,
|
| 8881 |
+
"loss": 1.3421,
|
| 8882 |
+
"step": 1263
|
| 8883 |
+
},
|
| 8884 |
+
{
|
| 8885 |
+
"epoch": 0.8321948810797465,
|
| 8886 |
+
"grad_norm": 0.5524282455444336,
|
| 8887 |
+
"learning_rate": 1.4413357556176631e-05,
|
| 8888 |
+
"loss": 1.2944,
|
| 8889 |
+
"step": 1264
|
| 8890 |
+
},
|
| 8891 |
+
{
|
| 8892 |
+
"epoch": 0.8328532631059172,
|
| 8893 |
+
"grad_norm": 0.6491603851318359,
|
| 8894 |
+
"learning_rate": 1.430287092341971e-05,
|
| 8895 |
+
"loss": 1.3944,
|
| 8896 |
+
"step": 1265
|
| 8897 |
+
},
|
| 8898 |
+
{
|
| 8899 |
+
"epoch": 0.8335116451320879,
|
| 8900 |
+
"grad_norm": 0.6646156311035156,
|
| 8901 |
+
"learning_rate": 1.4192776767238158e-05,
|
| 8902 |
+
"loss": 1.2302,
|
| 8903 |
+
"step": 1266
|
| 8904 |
+
},
|
| 8905 |
+
{
|
| 8906 |
+
"epoch": 0.8341700271582586,
|
| 8907 |
+
"grad_norm": 0.8558313846588135,
|
| 8908 |
+
"learning_rate": 1.408307559184242e-05,
|
| 8909 |
+
"loss": 1.252,
|
| 8910 |
+
"step": 1267
|
| 8911 |
+
},
|
| 8912 |
+
{
|
| 8913 |
+
"epoch": 0.8348284091844292,
|
| 8914 |
+
"grad_norm": 0.9991942644119263,
|
| 8915 |
+
"learning_rate": 1.3973767899642975e-05,
|
| 8916 |
+
"loss": 1.1287,
|
| 8917 |
+
"step": 1268
|
| 8918 |
+
},
|
| 8919 |
+
{
|
| 8920 |
+
"epoch": 0.8354867912106,
|
| 8921 |
+
"grad_norm": 1.0037152767181396,
|
| 8922 |
+
"learning_rate": 1.38648541912483e-05,
|
| 8923 |
+
"loss": 1.2154,
|
| 8924 |
+
"step": 1269
|
| 8925 |
+
},
|
| 8926 |
+
{
|
| 8927 |
+
"epoch": 0.8361451732367706,
|
| 8928 |
+
"grad_norm": 1.1659568548202515,
|
| 8929 |
+
"learning_rate": 1.3756334965462502e-05,
|
| 8930 |
+
"loss": 1.0629,
|
| 8931 |
+
"step": 1270
|
| 8932 |
+
},
|
| 8933 |
+
{
|
| 8934 |
+
"epoch": 0.8368035552629414,
|
| 8935 |
+
"grad_norm": 1.38374662399292,
|
| 8936 |
+
"learning_rate": 1.3648210719283028e-05,
|
| 8937 |
+
"loss": 1.0876,
|
| 8938 |
+
"step": 1271
|
| 8939 |
+
},
|
| 8940 |
+
{
|
| 8941 |
+
"epoch": 0.837461937289112,
|
| 8942 |
+
"grad_norm": 1.2293808460235596,
|
| 8943 |
+
"learning_rate": 1.3540481947898376e-05,
|
| 8944 |
+
"loss": 0.8495,
|
| 8945 |
+
"step": 1272
|
| 8946 |
+
},
|
| 8947 |
+
{
|
| 8948 |
+
"epoch": 0.8381203193152827,
|
| 8949 |
+
"grad_norm": 1.2876181602478027,
|
| 8950 |
+
"learning_rate": 1.3433149144685908e-05,
|
| 8951 |
+
"loss": 0.8809,
|
| 8952 |
+
"step": 1273
|
| 8953 |
+
},
|
| 8954 |
+
{
|
| 8955 |
+
"epoch": 0.8387787013414534,
|
| 8956 |
+
"grad_norm": 1.7796697616577148,
|
| 8957 |
+
"learning_rate": 1.3326212801209392e-05,
|
| 8958 |
+
"loss": 1.0504,
|
| 8959 |
+
"step": 1274
|
| 8960 |
+
},
|
| 8961 |
+
{
|
| 8962 |
+
"epoch": 0.839437083367624,
|
| 8963 |
+
"grad_norm": 2.186758041381836,
|
| 8964 |
+
"learning_rate": 1.3219673407217037e-05,
|
| 8965 |
+
"loss": 1.0458,
|
| 8966 |
+
"step": 1275
|
| 8967 |
+
},
|
| 8968 |
+
{
|
| 8969 |
+
"epoch": 0.8400954653937948,
|
| 8970 |
+
"grad_norm": 0.2571915090084076,
|
| 8971 |
+
"learning_rate": 1.311353145063905e-05,
|
| 8972 |
+
"loss": 1.1547,
|
| 8973 |
+
"step": 1276
|
| 8974 |
+
},
|
| 8975 |
+
{
|
| 8976 |
+
"epoch": 0.8407538474199654,
|
| 8977 |
+
"grad_norm": 0.2778959274291992,
|
| 8978 |
+
"learning_rate": 1.3007787417585372e-05,
|
| 8979 |
+
"loss": 1.1754,
|
| 8980 |
+
"step": 1277
|
| 8981 |
+
},
|
| 8982 |
+
{
|
| 8983 |
+
"epoch": 0.8414122294461361,
|
| 8984 |
+
"grad_norm": 0.3004729747772217,
|
| 8985 |
+
"learning_rate": 1.2902441792343612e-05,
|
| 8986 |
+
"loss": 1.2591,
|
| 8987 |
+
"step": 1278
|
| 8988 |
+
},
|
| 8989 |
+
{
|
| 8990 |
+
"epoch": 0.8420706114723068,
|
| 8991 |
+
"grad_norm": 0.3155894875526428,
|
| 8992 |
+
"learning_rate": 1.279749505737674e-05,
|
| 8993 |
+
"loss": 1.2791,
|
| 8994 |
+
"step": 1279
|
| 8995 |
+
},
|
| 8996 |
+
{
|
| 8997 |
+
"epoch": 0.8427289934984775,
|
| 8998 |
+
"grad_norm": 0.33009597659111023,
|
| 8999 |
+
"learning_rate": 1.2692947693320867e-05,
|
| 9000 |
+
"loss": 1.1802,
|
| 9001 |
+
"step": 1280
|
| 9002 |
+
},
|
| 9003 |
+
{
|
| 9004 |
+
"epoch": 0.8433873755246482,
|
| 9005 |
+
"grad_norm": 0.3477858603000641,
|
| 9006 |
+
"learning_rate": 1.2588800178983051e-05,
|
| 9007 |
+
"loss": 1.2065,
|
| 9008 |
+
"step": 1281
|
| 9009 |
+
},
|
| 9010 |
+
{
|
| 9011 |
+
"epoch": 0.8440457575508189,
|
| 9012 |
+
"grad_norm": 0.355474591255188,
|
| 9013 |
+
"learning_rate": 1.2485052991339174e-05,
|
| 9014 |
+
"loss": 1.2134,
|
| 9015 |
+
"step": 1282
|
| 9016 |
+
},
|
| 9017 |
+
{
|
| 9018 |
+
"epoch": 0.8447041395769895,
|
| 9019 |
+
"grad_norm": 0.3893088400363922,
|
| 9020 |
+
"learning_rate": 1.2381706605531562e-05,
|
| 9021 |
+
"loss": 1.2709,
|
| 9022 |
+
"step": 1283
|
| 9023 |
+
},
|
| 9024 |
+
{
|
| 9025 |
+
"epoch": 0.8453625216031603,
|
| 9026 |
+
"grad_norm": 0.39613381028175354,
|
| 9027 |
+
"learning_rate": 1.2278761494867119e-05,
|
| 9028 |
+
"loss": 1.239,
|
| 9029 |
+
"step": 1284
|
| 9030 |
+
},
|
| 9031 |
+
{
|
| 9032 |
+
"epoch": 0.8460209036293309,
|
| 9033 |
+
"grad_norm": 0.42657214403152466,
|
| 9034 |
+
"learning_rate": 1.2176218130814887e-05,
|
| 9035 |
+
"loss": 1.2017,
|
| 9036 |
+
"step": 1285
|
| 9037 |
+
},
|
| 9038 |
+
{
|
| 9039 |
+
"epoch": 0.8466792856555017,
|
| 9040 |
+
"grad_norm": 0.5077014565467834,
|
| 9041 |
+
"learning_rate": 1.2074076983003958e-05,
|
| 9042 |
+
"loss": 1.2926,
|
| 9043 |
+
"step": 1286
|
| 9044 |
+
},
|
| 9045 |
+
{
|
| 9046 |
+
"epoch": 0.8473376676816723,
|
| 9047 |
+
"grad_norm": 0.4870278537273407,
|
| 9048 |
+
"learning_rate": 1.1972338519221393e-05,
|
| 9049 |
+
"loss": 1.2529,
|
| 9050 |
+
"step": 1287
|
| 9051 |
+
},
|
| 9052 |
+
{
|
| 9053 |
+
"epoch": 0.8479960497078429,
|
| 9054 |
+
"grad_norm": 0.5146488547325134,
|
| 9055 |
+
"learning_rate": 1.1871003205410091e-05,
|
| 9056 |
+
"loss": 1.3742,
|
| 9057 |
+
"step": 1288
|
| 9058 |
+
},
|
| 9059 |
+
{
|
| 9060 |
+
"epoch": 0.8486544317340137,
|
| 9061 |
+
"grad_norm": 0.6352048516273499,
|
| 9062 |
+
"learning_rate": 1.1770071505666491e-05,
|
| 9063 |
+
"loss": 1.2944,
|
| 9064 |
+
"step": 1289
|
| 9065 |
+
},
|
| 9066 |
+
{
|
| 9067 |
+
"epoch": 0.8493128137601843,
|
| 9068 |
+
"grad_norm": 0.6424921154975891,
|
| 9069 |
+
"learning_rate": 1.166954388223862e-05,
|
| 9070 |
+
"loss": 1.232,
|
| 9071 |
+
"step": 1290
|
| 9072 |
+
},
|
| 9073 |
+
{
|
| 9074 |
+
"epoch": 0.849971195786355,
|
| 9075 |
+
"grad_norm": 0.745509922504425,
|
| 9076 |
+
"learning_rate": 1.1569420795523911e-05,
|
| 9077 |
+
"loss": 1.46,
|
| 9078 |
+
"step": 1291
|
| 9079 |
+
},
|
| 9080 |
+
{
|
| 9081 |
+
"epoch": 0.8506295778125257,
|
| 9082 |
+
"grad_norm": 0.8032955527305603,
|
| 9083 |
+
"learning_rate": 1.1469702704067064e-05,
|
| 9084 |
+
"loss": 1.1648,
|
| 9085 |
+
"step": 1292
|
| 9086 |
+
},
|
| 9087 |
+
{
|
| 9088 |
+
"epoch": 0.8512879598386964,
|
| 9089 |
+
"grad_norm": 0.9472469091415405,
|
| 9090 |
+
"learning_rate": 1.1370390064558034e-05,
|
| 9091 |
+
"loss": 1.2158,
|
| 9092 |
+
"step": 1293
|
| 9093 |
+
},
|
| 9094 |
+
{
|
| 9095 |
+
"epoch": 0.8519463418648671,
|
| 9096 |
+
"grad_norm": 1.1131627559661865,
|
| 9097 |
+
"learning_rate": 1.1271483331829835e-05,
|
| 9098 |
+
"loss": 1.1096,
|
| 9099 |
+
"step": 1294
|
| 9100 |
+
},
|
| 9101 |
+
{
|
| 9102 |
+
"epoch": 0.8526047238910378,
|
| 9103 |
+
"grad_norm": 1.2295578718185425,
|
| 9104 |
+
"learning_rate": 1.1172982958856482e-05,
|
| 9105 |
+
"loss": 1.0952,
|
| 9106 |
+
"step": 1295
|
| 9107 |
+
},
|
| 9108 |
+
{
|
| 9109 |
+
"epoch": 0.8532631059172084,
|
| 9110 |
+
"grad_norm": 1.3151978254318237,
|
| 9111 |
+
"learning_rate": 1.107488939675102e-05,
|
| 9112 |
+
"loss": 0.9785,
|
| 9113 |
+
"step": 1296
|
| 9114 |
+
},
|
| 9115 |
+
{
|
| 9116 |
+
"epoch": 0.8539214879433792,
|
| 9117 |
+
"grad_norm": 1.631568431854248,
|
| 9118 |
+
"learning_rate": 1.0977203094763345e-05,
|
| 9119 |
+
"loss": 1.1322,
|
| 9120 |
+
"step": 1297
|
| 9121 |
+
},
|
| 9122 |
+
{
|
| 9123 |
+
"epoch": 0.8545798699695498,
|
| 9124 |
+
"grad_norm": 1.512904405593872,
|
| 9125 |
+
"learning_rate": 1.0879924500278116e-05,
|
| 9126 |
+
"loss": 1.1041,
|
| 9127 |
+
"step": 1298
|
| 9128 |
+
},
|
| 9129 |
+
{
|
| 9130 |
+
"epoch": 0.8552382519957206,
|
| 9131 |
+
"grad_norm": 1.573190689086914,
|
| 9132 |
+
"learning_rate": 1.0783054058812858e-05,
|
| 9133 |
+
"loss": 0.8695,
|
| 9134 |
+
"step": 1299
|
| 9135 |
+
},
|
| 9136 |
+
{
|
| 9137 |
+
"epoch": 0.8558966340218912,
|
| 9138 |
+
"grad_norm": 1.600016474723816,
|
| 9139 |
+
"learning_rate": 1.0686592214015768e-05,
|
| 9140 |
+
"loss": 0.7234,
|
| 9141 |
+
"step": 1300
|
| 9142 |
+
},
|
| 9143 |
+
{
|
| 9144 |
+
"epoch": 0.8565550160480618,
|
| 9145 |
+
"grad_norm": 0.2644733488559723,
|
| 9146 |
+
"learning_rate": 1.0590539407663803e-05,
|
| 9147 |
+
"loss": 1.1294,
|
| 9148 |
+
"step": 1301
|
| 9149 |
+
},
|
| 9150 |
+
{
|
| 9151 |
+
"epoch": 0.8572133980742326,
|
| 9152 |
+
"grad_norm": 0.28398260474205017,
|
| 9153 |
+
"learning_rate": 1.0494896079660554e-05,
|
| 9154 |
+
"loss": 1.2221,
|
| 9155 |
+
"step": 1302
|
| 9156 |
+
},
|
| 9157 |
+
{
|
| 9158 |
+
"epoch": 0.8578717801004032,
|
| 9159 |
+
"grad_norm": 0.3015180230140686,
|
| 9160 |
+
"learning_rate": 1.0399662668034327e-05,
|
| 9161 |
+
"loss": 1.1769,
|
| 9162 |
+
"step": 1303
|
| 9163 |
+
},
|
| 9164 |
+
{
|
| 9165 |
+
"epoch": 0.858530162126574,
|
| 9166 |
+
"grad_norm": 0.295482873916626,
|
| 9167 |
+
"learning_rate": 1.0304839608936e-05,
|
| 9168 |
+
"loss": 1.1557,
|
| 9169 |
+
"step": 1304
|
| 9170 |
+
},
|
| 9171 |
+
{
|
| 9172 |
+
"epoch": 0.8591885441527446,
|
| 9173 |
+
"grad_norm": 0.3402240574359894,
|
| 9174 |
+
"learning_rate": 1.021042733663723e-05,
|
| 9175 |
+
"loss": 1.1766,
|
| 9176 |
+
"step": 1305
|
| 9177 |
+
},
|
| 9178 |
+
{
|
| 9179 |
+
"epoch": 0.8598469261789153,
|
| 9180 |
+
"grad_norm": 0.35621342062950134,
|
| 9181 |
+
"learning_rate": 1.0116426283528302e-05,
|
| 9182 |
+
"loss": 1.268,
|
| 9183 |
+
"step": 1306
|
| 9184 |
+
},
|
| 9185 |
+
{
|
| 9186 |
+
"epoch": 0.860505308205086,
|
| 9187 |
+
"grad_norm": 0.36093905568122864,
|
| 9188 |
+
"learning_rate": 1.0022836880116138e-05,
|
| 9189 |
+
"loss": 1.2207,
|
| 9190 |
+
"step": 1307
|
| 9191 |
+
},
|
| 9192 |
+
{
|
| 9193 |
+
"epoch": 0.8611636902312567,
|
| 9194 |
+
"grad_norm": 0.4131215512752533,
|
| 9195 |
+
"learning_rate": 9.929659555022442e-06,
|
| 9196 |
+
"loss": 1.3246,
|
| 9197 |
+
"step": 1308
|
| 9198 |
+
},
|
| 9199 |
+
{
|
| 9200 |
+
"epoch": 0.8618220722574274,
|
| 9201 |
+
"grad_norm": 0.4348328113555908,
|
| 9202 |
+
"learning_rate": 9.836894734981706e-06,
|
| 9203 |
+
"loss": 1.2316,
|
| 9204 |
+
"step": 1309
|
| 9205 |
+
},
|
| 9206 |
+
{
|
| 9207 |
+
"epoch": 0.8624804542835981,
|
| 9208 |
+
"grad_norm": 0.471159964799881,
|
| 9209 |
+
"learning_rate": 9.744542844839143e-06,
|
| 9210 |
+
"loss": 1.4138,
|
| 9211 |
+
"step": 1310
|
| 9212 |
+
},
|
| 9213 |
+
{
|
| 9214 |
+
"epoch": 0.8631388363097687,
|
| 9215 |
+
"grad_norm": 0.4972412586212158,
|
| 9216 |
+
"learning_rate": 9.652604307548895e-06,
|
| 9217 |
+
"loss": 1.3278,
|
| 9218 |
+
"step": 1311
|
| 9219 |
+
},
|
| 9220 |
+
{
|
| 9221 |
+
"epoch": 0.8637972183359395,
|
| 9222 |
+
"grad_norm": 0.5030047297477722,
|
| 9223 |
+
"learning_rate": 9.561079544171992e-06,
|
| 9224 |
+
"loss": 1.2477,
|
| 9225 |
+
"step": 1312
|
| 9226 |
+
},
|
| 9227 |
+
{
|
| 9228 |
+
"epoch": 0.8644556003621101,
|
| 9229 |
+
"grad_norm": 0.5558236241340637,
|
| 9230 |
+
"learning_rate": 9.469968973874466e-06,
|
| 9231 |
+
"loss": 1.4028,
|
| 9232 |
+
"step": 1313
|
| 9233 |
+
},
|
| 9234 |
+
{
|
| 9235 |
+
"epoch": 0.8651139823882807,
|
| 9236 |
+
"grad_norm": 0.6310866475105286,
|
| 9237 |
+
"learning_rate": 9.37927301392545e-06,
|
| 9238 |
+
"loss": 1.2789,
|
| 9239 |
+
"step": 1314
|
| 9240 |
+
},
|
| 9241 |
+
{
|
| 9242 |
+
"epoch": 0.8657723644144515,
|
| 9243 |
+
"grad_norm": 0.7040503025054932,
|
| 9244 |
+
"learning_rate": 9.28899207969518e-06,
|
| 9245 |
+
"loss": 1.1996,
|
| 9246 |
+
"step": 1315
|
| 9247 |
+
},
|
| 9248 |
+
{
|
| 9249 |
+
"epoch": 0.8664307464406221,
|
| 9250 |
+
"grad_norm": 0.7893826365470886,
|
| 9251 |
+
"learning_rate": 9.199126584653183e-06,
|
| 9252 |
+
"loss": 1.324,
|
| 9253 |
+
"step": 1316
|
| 9254 |
+
},
|
| 9255 |
+
{
|
| 9256 |
+
"epoch": 0.8670891284667929,
|
| 9257 |
+
"grad_norm": 0.9313814043998718,
|
| 9258 |
+
"learning_rate": 9.109676940366418e-06,
|
| 9259 |
+
"loss": 1.2993,
|
| 9260 |
+
"step": 1317
|
| 9261 |
+
},
|
| 9262 |
+
{
|
| 9263 |
+
"epoch": 0.8677475104929635,
|
| 9264 |
+
"grad_norm": 0.8821551203727722,
|
| 9265 |
+
"learning_rate": 9.02064355649721e-06,
|
| 9266 |
+
"loss": 1.0871,
|
| 9267 |
+
"step": 1318
|
| 9268 |
+
},
|
| 9269 |
+
{
|
| 9270 |
+
"epoch": 0.8684058925191342,
|
| 9271 |
+
"grad_norm": 1.143466591835022,
|
| 9272 |
+
"learning_rate": 8.932026840801554e-06,
|
| 9273 |
+
"loss": 1.2784,
|
| 9274 |
+
"step": 1319
|
| 9275 |
+
},
|
| 9276 |
+
{
|
| 9277 |
+
"epoch": 0.8690642745453049,
|
| 9278 |
+
"grad_norm": 1.2319263219833374,
|
| 9279 |
+
"learning_rate": 8.843827199127207e-06,
|
| 9280 |
+
"loss": 0.9828,
|
| 9281 |
+
"step": 1320
|
| 9282 |
+
},
|
| 9283 |
+
{
|
| 9284 |
+
"epoch": 0.8697226565714756,
|
| 9285 |
+
"grad_norm": 1.3100347518920898,
|
| 9286 |
+
"learning_rate": 8.756045035411753e-06,
|
| 9287 |
+
"loss": 1.0976,
|
| 9288 |
+
"step": 1321
|
| 9289 |
+
},
|
| 9290 |
+
{
|
| 9291 |
+
"epoch": 0.8703810385976463,
|
| 9292 |
+
"grad_norm": 1.8412470817565918,
|
| 9293 |
+
"learning_rate": 8.668680751680835e-06,
|
| 9294 |
+
"loss": 1.3813,
|
| 9295 |
+
"step": 1322
|
| 9296 |
+
},
|
| 9297 |
+
{
|
| 9298 |
+
"epoch": 0.871039420623817,
|
| 9299 |
+
"grad_norm": 1.6632609367370605,
|
| 9300 |
+
"learning_rate": 8.5817347480463e-06,
|
| 9301 |
+
"loss": 0.9646,
|
| 9302 |
+
"step": 1323
|
| 9303 |
+
},
|
| 9304 |
+
{
|
| 9305 |
+
"epoch": 0.8716978026499876,
|
| 9306 |
+
"grad_norm": 1.9535080194473267,
|
| 9307 |
+
"learning_rate": 8.495207422704298e-06,
|
| 9308 |
+
"loss": 0.9062,
|
| 9309 |
+
"step": 1324
|
| 9310 |
+
},
|
| 9311 |
+
{
|
| 9312 |
+
"epoch": 0.8723561846761584,
|
| 9313 |
+
"grad_norm": 1.937402606010437,
|
| 9314 |
+
"learning_rate": 8.409099171933544e-06,
|
| 9315 |
+
"loss": 0.8929,
|
| 9316 |
+
"step": 1325
|
| 9317 |
+
},
|
| 9318 |
+
{
|
| 9319 |
+
"epoch": 0.873014566702329,
|
| 9320 |
+
"grad_norm": 0.26802825927734375,
|
| 9321 |
+
"learning_rate": 8.323410390093522e-06,
|
| 9322 |
+
"loss": 1.1504,
|
| 9323 |
+
"step": 1326
|
| 9324 |
+
},
|
| 9325 |
+
{
|
| 9326 |
+
"epoch": 0.8736729487284998,
|
| 9327 |
+
"grad_norm": 0.2851593494415283,
|
| 9328 |
+
"learning_rate": 8.238141469622529e-06,
|
| 9329 |
+
"loss": 1.1977,
|
| 9330 |
+
"step": 1327
|
| 9331 |
+
},
|
| 9332 |
+
{
|
| 9333 |
+
"epoch": 0.8743313307546704,
|
| 9334 |
+
"grad_norm": 0.29775407910346985,
|
| 9335 |
+
"learning_rate": 8.15329280103605e-06,
|
| 9336 |
+
"loss": 1.2695,
|
| 9337 |
+
"step": 1328
|
| 9338 |
+
},
|
| 9339 |
+
{
|
| 9340 |
+
"epoch": 0.874989712780841,
|
| 9341 |
+
"grad_norm": 0.29565665125846863,
|
| 9342 |
+
"learning_rate": 8.068864772924855e-06,
|
| 9343 |
+
"loss": 1.0892,
|
| 9344 |
+
"step": 1329
|
| 9345 |
+
},
|
| 9346 |
+
{
|
| 9347 |
+
"epoch": 0.8756480948070118,
|
| 9348 |
+
"grad_norm": 0.3173798620700836,
|
| 9349 |
+
"learning_rate": 7.984857771953303e-06,
|
| 9350 |
+
"loss": 1.2387,
|
| 9351 |
+
"step": 1330
|
| 9352 |
+
},
|
| 9353 |
+
{
|
| 9354 |
+
"epoch": 0.8763064768331824,
|
| 9355 |
+
"grad_norm": 0.34442922472953796,
|
| 9356 |
+
"learning_rate": 7.901272182857478e-06,
|
| 9357 |
+
"loss": 1.23,
|
| 9358 |
+
"step": 1331
|
| 9359 |
+
},
|
| 9360 |
+
{
|
| 9361 |
+
"epoch": 0.8769648588593532,
|
| 9362 |
+
"grad_norm": 0.35819026827812195,
|
| 9363 |
+
"learning_rate": 7.818108388443546e-06,
|
| 9364 |
+
"loss": 1.2773,
|
| 9365 |
+
"step": 1332
|
| 9366 |
+
},
|
| 9367 |
+
{
|
| 9368 |
+
"epoch": 0.8776232408855238,
|
| 9369 |
+
"grad_norm": 0.3835400938987732,
|
| 9370 |
+
"learning_rate": 7.735366769585817e-06,
|
| 9371 |
+
"loss": 1.2615,
|
| 9372 |
+
"step": 1333
|
| 9373 |
+
},
|
| 9374 |
+
{
|
| 9375 |
+
"epoch": 0.8782816229116945,
|
| 9376 |
+
"grad_norm": 0.4186893701553345,
|
| 9377 |
+
"learning_rate": 7.653047705225257e-06,
|
| 9378 |
+
"loss": 1.2031,
|
| 9379 |
+
"step": 1334
|
| 9380 |
+
},
|
| 9381 |
+
{
|
| 9382 |
+
"epoch": 0.8789400049378652,
|
| 9383 |
+
"grad_norm": 0.4734337627887726,
|
| 9384 |
+
"learning_rate": 7.571151572367541e-06,
|
| 9385 |
+
"loss": 1.34,
|
| 9386 |
+
"step": 1335
|
| 9387 |
+
},
|
| 9388 |
+
{
|
| 9389 |
+
"epoch": 0.8795983869640359,
|
| 9390 |
+
"grad_norm": 0.494617760181427,
|
| 9391 |
+
"learning_rate": 7.489678746081364e-06,
|
| 9392 |
+
"loss": 1.269,
|
| 9393 |
+
"step": 1336
|
| 9394 |
+
},
|
| 9395 |
+
{
|
| 9396 |
+
"epoch": 0.8802567689902066,
|
| 9397 |
+
"grad_norm": 0.540384829044342,
|
| 9398 |
+
"learning_rate": 7.408629599496808e-06,
|
| 9399 |
+
"loss": 1.3307,
|
| 9400 |
+
"step": 1337
|
| 9401 |
+
},
|
| 9402 |
+
{
|
| 9403 |
+
"epoch": 0.8809151510163773,
|
| 9404 |
+
"grad_norm": 0.5604397058486938,
|
| 9405 |
+
"learning_rate": 7.32800450380361e-06,
|
| 9406 |
+
"loss": 1.2498,
|
| 9407 |
+
"step": 1338
|
| 9408 |
+
},
|
| 9409 |
+
{
|
| 9410 |
+
"epoch": 0.8815735330425479,
|
| 9411 |
+
"grad_norm": 0.6044402122497559,
|
| 9412 |
+
"learning_rate": 7.247803828249356e-06,
|
| 9413 |
+
"loss": 1.2038,
|
| 9414 |
+
"step": 1339
|
| 9415 |
+
},
|
| 9416 |
+
{
|
| 9417 |
+
"epoch": 0.8822319150687187,
|
| 9418 |
+
"grad_norm": 0.6669920682907104,
|
| 9419 |
+
"learning_rate": 7.168027940137922e-06,
|
| 9420 |
+
"loss": 1.1242,
|
| 9421 |
+
"step": 1340
|
| 9422 |
+
},
|
| 9423 |
+
{
|
| 9424 |
+
"epoch": 0.8828902970948893,
|
| 9425 |
+
"grad_norm": 0.8415902853012085,
|
| 9426 |
+
"learning_rate": 7.088677204827721e-06,
|
| 9427 |
+
"loss": 1.3357,
|
| 9428 |
+
"step": 1341
|
| 9429 |
+
},
|
| 9430 |
+
{
|
| 9431 |
+
"epoch": 0.8835486791210599,
|
| 9432 |
+
"grad_norm": 0.9297363758087158,
|
| 9433 |
+
"learning_rate": 7.009751985730062e-06,
|
| 9434 |
+
"loss": 1.2527,
|
| 9435 |
+
"step": 1342
|
| 9436 |
+
},
|
| 9437 |
+
{
|
| 9438 |
+
"epoch": 0.8842070611472307,
|
| 9439 |
+
"grad_norm": 1.0265207290649414,
|
| 9440 |
+
"learning_rate": 6.93125264430744e-06,
|
| 9441 |
+
"loss": 1.0002,
|
| 9442 |
+
"step": 1343
|
| 9443 |
+
},
|
| 9444 |
+
{
|
| 9445 |
+
"epoch": 0.8848654431734013,
|
| 9446 |
+
"grad_norm": 1.0541255474090576,
|
| 9447 |
+
"learning_rate": 6.853179540071964e-06,
|
| 9448 |
+
"loss": 1.104,
|
| 9449 |
+
"step": 1344
|
| 9450 |
+
},
|
| 9451 |
+
{
|
| 9452 |
+
"epoch": 0.8855238251995721,
|
| 9453 |
+
"grad_norm": 1.1322312355041504,
|
| 9454 |
+
"learning_rate": 6.775533030583559e-06,
|
| 9455 |
+
"loss": 0.8635,
|
| 9456 |
+
"step": 1345
|
| 9457 |
+
},
|
| 9458 |
+
{
|
| 9459 |
+
"epoch": 0.8861822072257427,
|
| 9460 |
+
"grad_norm": 1.4108020067214966,
|
| 9461 |
+
"learning_rate": 6.698313471448547e-06,
|
| 9462 |
+
"loss": 1.1874,
|
| 9463 |
+
"step": 1346
|
| 9464 |
+
},
|
| 9465 |
+
{
|
| 9466 |
+
"epoch": 0.8868405892519134,
|
| 9467 |
+
"grad_norm": 1.4998079538345337,
|
| 9468 |
+
"learning_rate": 6.62152121631785e-06,
|
| 9469 |
+
"loss": 1.2854,
|
| 9470 |
+
"step": 1347
|
| 9471 |
+
},
|
| 9472 |
+
{
|
| 9473 |
+
"epoch": 0.8874989712780841,
|
| 9474 |
+
"grad_norm": 2.0073752403259277,
|
| 9475 |
+
"learning_rate": 6.545156616885373e-06,
|
| 9476 |
+
"loss": 1.1182,
|
| 9477 |
+
"step": 1348
|
| 9478 |
+
},
|
| 9479 |
+
{
|
| 9480 |
+
"epoch": 0.8881573533042548,
|
| 9481 |
+
"grad_norm": 1.9627798795700073,
|
| 9482 |
+
"learning_rate": 6.469220022886491e-06,
|
| 9483 |
+
"loss": 1.0721,
|
| 9484 |
+
"step": 1349
|
| 9485 |
+
},
|
| 9486 |
+
{
|
| 9487 |
+
"epoch": 0.8888157353304255,
|
| 9488 |
+
"grad_norm": 2.0956573486328125,
|
| 9489 |
+
"learning_rate": 6.39371178209639e-06,
|
| 9490 |
+
"loss": 1.06,
|
| 9491 |
+
"step": 1350
|
| 9492 |
+
},
|
| 9493 |
+
{
|
| 9494 |
+
"epoch": 0.8894741173565962,
|
| 9495 |
+
"grad_norm": 0.26062607765197754,
|
| 9496 |
+
"learning_rate": 6.31863224032847e-06,
|
| 9497 |
+
"loss": 1.2184,
|
| 9498 |
+
"step": 1351
|
| 9499 |
+
},
|
| 9500 |
+
{
|
| 9501 |
+
"epoch": 0.8901324993827668,
|
| 9502 |
+
"grad_norm": 0.2938222289085388,
|
| 9503 |
+
"learning_rate": 6.24398174143277e-06,
|
| 9504 |
+
"loss": 1.262,
|
| 9505 |
+
"step": 1352
|
| 9506 |
+
},
|
| 9507 |
+
{
|
| 9508 |
+
"epoch": 0.8907908814089376,
|
| 9509 |
+
"grad_norm": 0.2993498146533966,
|
| 9510 |
+
"learning_rate": 6.169760627294408e-06,
|
| 9511 |
+
"loss": 1.0936,
|
| 9512 |
+
"step": 1353
|
| 9513 |
+
},
|
| 9514 |
+
{
|
| 9515 |
+
"epoch": 0.8914492634351082,
|
| 9516 |
+
"grad_norm": 0.3110993802547455,
|
| 9517 |
+
"learning_rate": 6.095969237831956e-06,
|
| 9518 |
+
"loss": 1.189,
|
| 9519 |
+
"step": 1354
|
| 9520 |
+
},
|
| 9521 |
+
{
|
| 9522 |
+
"epoch": 0.8921076454612789,
|
| 9523 |
+
"grad_norm": 0.3342379927635193,
|
| 9524 |
+
"learning_rate": 6.022607910996014e-06,
|
| 9525 |
+
"loss": 1.1589,
|
| 9526 |
+
"step": 1355
|
| 9527 |
+
},
|
| 9528 |
+
{
|
| 9529 |
+
"epoch": 0.8927660274874496,
|
| 9530 |
+
"grad_norm": 0.34470638632774353,
|
| 9531 |
+
"learning_rate": 5.949676982767505e-06,
|
| 9532 |
+
"loss": 1.2739,
|
| 9533 |
+
"step": 1356
|
| 9534 |
+
},
|
| 9535 |
+
{
|
| 9536 |
+
"epoch": 0.8934244095136202,
|
| 9537 |
+
"grad_norm": 0.3623284101486206,
|
| 9538 |
+
"learning_rate": 5.877176787156213e-06,
|
| 9539 |
+
"loss": 1.2129,
|
| 9540 |
+
"step": 1357
|
| 9541 |
+
},
|
| 9542 |
+
{
|
| 9543 |
+
"epoch": 0.894082791539791,
|
| 9544 |
+
"grad_norm": 0.3873266875743866,
|
| 9545 |
+
"learning_rate": 5.805107656199271e-06,
|
| 9546 |
+
"loss": 1.2821,
|
| 9547 |
+
"step": 1358
|
| 9548 |
+
},
|
| 9549 |
+
{
|
| 9550 |
+
"epoch": 0.8947411735659616,
|
| 9551 |
+
"grad_norm": 0.4302651882171631,
|
| 9552 |
+
"learning_rate": 5.733469919959622e-06,
|
| 9553 |
+
"loss": 1.2724,
|
| 9554 |
+
"step": 1359
|
| 9555 |
+
},
|
| 9556 |
+
{
|
| 9557 |
+
"epoch": 0.8953995555921324,
|
| 9558 |
+
"grad_norm": 0.44161009788513184,
|
| 9559 |
+
"learning_rate": 5.66226390652449e-06,
|
| 9560 |
+
"loss": 1.33,
|
| 9561 |
+
"step": 1360
|
| 9562 |
+
},
|
| 9563 |
+
{
|
| 9564 |
+
"epoch": 0.896057937618303,
|
| 9565 |
+
"grad_norm": 0.4458579421043396,
|
| 9566 |
+
"learning_rate": 5.591489942003902e-06,
|
| 9567 |
+
"loss": 1.3203,
|
| 9568 |
+
"step": 1361
|
| 9569 |
+
},
|
| 9570 |
+
{
|
| 9571 |
+
"epoch": 0.8967163196444737,
|
| 9572 |
+
"grad_norm": 0.50836181640625,
|
| 9573 |
+
"learning_rate": 5.521148350529137e-06,
|
| 9574 |
+
"loss": 1.4751,
|
| 9575 |
+
"step": 1362
|
| 9576 |
+
},
|
| 9577 |
+
{
|
| 9578 |
+
"epoch": 0.8973747016706444,
|
| 9579 |
+
"grad_norm": 0.5364328026771545,
|
| 9580 |
+
"learning_rate": 5.451239454251356e-06,
|
| 9581 |
+
"loss": 1.3369,
|
| 9582 |
+
"step": 1363
|
| 9583 |
+
},
|
| 9584 |
+
{
|
| 9585 |
+
"epoch": 0.8980330836968151,
|
| 9586 |
+
"grad_norm": 0.5729132890701294,
|
| 9587 |
+
"learning_rate": 5.381763573340049e-06,
|
| 9588 |
+
"loss": 1.2195,
|
| 9589 |
+
"step": 1364
|
| 9590 |
+
},
|
| 9591 |
+
{
|
| 9592 |
+
"epoch": 0.8986914657229857,
|
| 9593 |
+
"grad_norm": 0.5974079966545105,
|
| 9594 |
+
"learning_rate": 5.312721025981504e-06,
|
| 9595 |
+
"loss": 1.2283,
|
| 9596 |
+
"step": 1365
|
| 9597 |
+
},
|
| 9598 |
+
{
|
| 9599 |
+
"epoch": 0.8993498477491565,
|
| 9600 |
+
"grad_norm": 0.6872406601905823,
|
| 9601 |
+
"learning_rate": 5.244112128377476e-06,
|
| 9602 |
+
"loss": 1.3682,
|
| 9603 |
+
"step": 1366
|
| 9604 |
+
},
|
| 9605 |
+
{
|
| 9606 |
+
"epoch": 0.9000082297753271,
|
| 9607 |
+
"grad_norm": 0.8229653239250183,
|
| 9608 |
+
"learning_rate": 5.1759371947437516e-06,
|
| 9609 |
+
"loss": 1.3575,
|
| 9610 |
+
"step": 1367
|
| 9611 |
+
},
|
| 9612 |
+
{
|
| 9613 |
+
"epoch": 0.9006666118014979,
|
| 9614 |
+
"grad_norm": 0.83298259973526,
|
| 9615 |
+
"learning_rate": 5.108196537308507e-06,
|
| 9616 |
+
"loss": 1.1735,
|
| 9617 |
+
"step": 1368
|
| 9618 |
+
},
|
| 9619 |
+
{
|
| 9620 |
+
"epoch": 0.9013249938276685,
|
| 9621 |
+
"grad_norm": 1.0485754013061523,
|
| 9622 |
+
"learning_rate": 5.040890466311121e-06,
|
| 9623 |
+
"loss": 1.1567,
|
| 9624 |
+
"step": 1369
|
| 9625 |
+
},
|
| 9626 |
+
{
|
| 9627 |
+
"epoch": 0.9019833758538391,
|
| 9628 |
+
"grad_norm": 1.270142912864685,
|
| 9629 |
+
"learning_rate": 4.97401929000062e-06,
|
| 9630 |
+
"loss": 1.1402,
|
| 9631 |
+
"step": 1370
|
| 9632 |
+
},
|
| 9633 |
+
{
|
| 9634 |
+
"epoch": 0.9026417578800099,
|
| 9635 |
+
"grad_norm": 1.2949579954147339,
|
| 9636 |
+
"learning_rate": 4.907583314634267e-06,
|
| 9637 |
+
"loss": 0.8752,
|
| 9638 |
+
"step": 1371
|
| 9639 |
+
},
|
| 9640 |
+
{
|
| 9641 |
+
"epoch": 0.9033001399061805,
|
| 9642 |
+
"grad_norm": 1.4545363187789917,
|
| 9643 |
+
"learning_rate": 4.841582844476245e-06,
|
| 9644 |
+
"loss": 1.0948,
|
| 9645 |
+
"step": 1372
|
| 9646 |
+
},
|
| 9647 |
+
{
|
| 9648 |
+
"epoch": 0.9039585219323513,
|
| 9649 |
+
"grad_norm": 1.7462900876998901,
|
| 9650 |
+
"learning_rate": 4.7760181817961604e-06,
|
| 9651 |
+
"loss": 1.0682,
|
| 9652 |
+
"step": 1373
|
| 9653 |
+
},
|
| 9654 |
+
{
|
| 9655 |
+
"epoch": 0.9046169039585219,
|
| 9656 |
+
"grad_norm": 1.53018319606781,
|
| 9657 |
+
"learning_rate": 4.710889626867687e-06,
|
| 9658 |
+
"loss": 0.8449,
|
| 9659 |
+
"step": 1374
|
| 9660 |
+
},
|
| 9661 |
+
{
|
| 9662 |
+
"epoch": 0.9052752859846926,
|
| 9663 |
+
"grad_norm": 2.336136817932129,
|
| 9664 |
+
"learning_rate": 4.646197477967251e-06,
|
| 9665 |
+
"loss": 1.2089,
|
| 9666 |
+
"step": 1375
|
| 9667 |
+
},
|
| 9668 |
+
{
|
| 9669 |
+
"epoch": 0.9059336680108633,
|
| 9670 |
+
"grad_norm": 0.24908111989498138,
|
| 9671 |
+
"learning_rate": 4.5819420313726545e-06,
|
| 9672 |
+
"loss": 1.1893,
|
| 9673 |
+
"step": 1376
|
| 9674 |
+
},
|
| 9675 |
+
{
|
| 9676 |
+
"epoch": 0.906592050037034,
|
| 9677 |
+
"grad_norm": 0.2742631733417511,
|
| 9678 |
+
"learning_rate": 4.518123581361577e-06,
|
| 9679 |
+
"loss": 1.2126,
|
| 9680 |
+
"step": 1377
|
| 9681 |
+
},
|
| 9682 |
+
{
|
| 9683 |
+
"epoch": 0.9072504320632047,
|
| 9684 |
+
"grad_norm": 0.29479411244392395,
|
| 9685 |
+
"learning_rate": 4.4547424202104335e-06,
|
| 9686 |
+
"loss": 1.1572,
|
| 9687 |
+
"step": 1378
|
| 9688 |
+
},
|
| 9689 |
+
{
|
| 9690 |
+
"epoch": 0.9079088140893754,
|
| 9691 |
+
"grad_norm": 0.299268901348114,
|
| 9692 |
+
"learning_rate": 4.391798838192884e-06,
|
| 9693 |
+
"loss": 1.1414,
|
| 9694 |
+
"step": 1379
|
| 9695 |
+
},
|
| 9696 |
+
{
|
| 9697 |
+
"epoch": 0.908567196115546,
|
| 9698 |
+
"grad_norm": 0.32361260056495667,
|
| 9699 |
+
"learning_rate": 4.329293123578604e-06,
|
| 9700 |
+
"loss": 1.2059,
|
| 9701 |
+
"step": 1380
|
| 9702 |
+
},
|
| 9703 |
+
{
|
| 9704 |
+
"epoch": 0.9092255781417168,
|
| 9705 |
+
"grad_norm": 0.35213223099708557,
|
| 9706 |
+
"learning_rate": 4.267225562631871e-06,
|
| 9707 |
+
"loss": 1.1661,
|
| 9708 |
+
"step": 1381
|
| 9709 |
+
},
|
| 9710 |
+
{
|
| 9711 |
+
"epoch": 0.9098839601678874,
|
| 9712 |
+
"grad_norm": 0.36282479763031006,
|
| 9713 |
+
"learning_rate": 4.2055964396103486e-06,
|
| 9714 |
+
"loss": 1.2415,
|
| 9715 |
+
"step": 1382
|
| 9716 |
+
},
|
| 9717 |
+
{
|
| 9718 |
+
"epoch": 0.910542342194058,
|
| 9719 |
+
"grad_norm": 0.39451318979263306,
|
| 9720 |
+
"learning_rate": 4.1444060367636925e-06,
|
| 9721 |
+
"loss": 1.1791,
|
| 9722 |
+
"step": 1383
|
| 9723 |
+
},
|
| 9724 |
+
{
|
| 9725 |
+
"epoch": 0.9112007242202288,
|
| 9726 |
+
"grad_norm": 0.4231424927711487,
|
| 9727 |
+
"learning_rate": 4.083654634332334e-06,
|
| 9728 |
+
"loss": 1.3214,
|
| 9729 |
+
"step": 1384
|
| 9730 |
+
},
|
| 9731 |
+
{
|
| 9732 |
+
"epoch": 0.9118591062463994,
|
| 9733 |
+
"grad_norm": 0.4535464942455292,
|
| 9734 |
+
"learning_rate": 4.023342510546191e-06,
|
| 9735 |
+
"loss": 1.2253,
|
| 9736 |
+
"step": 1385
|
| 9737 |
+
},
|
| 9738 |
+
{
|
| 9739 |
+
"epoch": 0.9125174882725702,
|
| 9740 |
+
"grad_norm": 0.5160973072052002,
|
| 9741 |
+
"learning_rate": 3.963469941623288e-06,
|
| 9742 |
+
"loss": 1.2891,
|
| 9743 |
+
"step": 1386
|
| 9744 |
+
},
|
| 9745 |
+
{
|
| 9746 |
+
"epoch": 0.9131758702987408,
|
| 9747 |
+
"grad_norm": 0.5102490782737732,
|
| 9748 |
+
"learning_rate": 3.904037201768618e-06,
|
| 9749 |
+
"loss": 1.2608,
|
| 9750 |
+
"step": 1387
|
| 9751 |
+
},
|
| 9752 |
+
{
|
| 9753 |
+
"epoch": 0.9138342523249116,
|
| 9754 |
+
"grad_norm": 0.5157856941223145,
|
| 9755 |
+
"learning_rate": 3.845044563172895e-06,
|
| 9756 |
+
"loss": 1.3482,
|
| 9757 |
+
"step": 1388
|
| 9758 |
+
},
|
| 9759 |
+
{
|
| 9760 |
+
"epoch": 0.9144926343510822,
|
| 9761 |
+
"grad_norm": 0.576413094997406,
|
| 9762 |
+
"learning_rate": 3.786492296011157e-06,
|
| 9763 |
+
"loss": 1.2207,
|
| 9764 |
+
"step": 1389
|
| 9765 |
+
},
|
| 9766 |
+
{
|
| 9767 |
+
"epoch": 0.9151510163772529,
|
| 9768 |
+
"grad_norm": 0.652743399143219,
|
| 9769 |
+
"learning_rate": 3.7283806684416776e-06,
|
| 9770 |
+
"loss": 1.3499,
|
| 9771 |
+
"step": 1390
|
| 9772 |
+
},
|
| 9773 |
+
{
|
| 9774 |
+
"epoch": 0.9158093984034236,
|
| 9775 |
+
"grad_norm": 0.7483471035957336,
|
| 9776 |
+
"learning_rate": 3.670709946604667e-06,
|
| 9777 |
+
"loss": 1.3639,
|
| 9778 |
+
"step": 1391
|
| 9779 |
+
},
|
| 9780 |
+
{
|
| 9781 |
+
"epoch": 0.9164677804295943,
|
| 9782 |
+
"grad_norm": 0.7818670868873596,
|
| 9783 |
+
"learning_rate": 3.6134803946210937e-06,
|
| 9784 |
+
"loss": 1.1841,
|
| 9785 |
+
"step": 1392
|
| 9786 |
+
},
|
| 9787 |
+
{
|
| 9788 |
+
"epoch": 0.917126162455765,
|
| 9789 |
+
"grad_norm": 0.8835737109184265,
|
| 9790 |
+
"learning_rate": 3.5566922745914335e-06,
|
| 9791 |
+
"loss": 1.0998,
|
| 9792 |
+
"step": 1393
|
| 9793 |
+
},
|
| 9794 |
+
{
|
| 9795 |
+
"epoch": 0.9177845444819357,
|
| 9796 |
+
"grad_norm": 1.2403169870376587,
|
| 9797 |
+
"learning_rate": 3.5003458465944882e-06,
|
| 9798 |
+
"loss": 1.5137,
|
| 9799 |
+
"step": 1394
|
| 9800 |
+
},
|
| 9801 |
+
{
|
| 9802 |
+
"epoch": 0.9184429265081063,
|
| 9803 |
+
"grad_norm": 1.1381645202636719,
|
| 9804 |
+
"learning_rate": 3.444441368686191e-06,
|
| 9805 |
+
"loss": 1.069,
|
| 9806 |
+
"step": 1395
|
| 9807 |
+
},
|
| 9808 |
+
{
|
| 9809 |
+
"epoch": 0.919101308534277,
|
| 9810 |
+
"grad_norm": 1.4728044271469116,
|
| 9811 |
+
"learning_rate": 3.388979096898415e-06,
|
| 9812 |
+
"loss": 1.2426,
|
| 9813 |
+
"step": 1396
|
| 9814 |
+
},
|
| 9815 |
+
{
|
| 9816 |
+
"epoch": 0.9197596905604477,
|
| 9817 |
+
"grad_norm": 1.2873393297195435,
|
| 9818 |
+
"learning_rate": 3.3339592852378864e-06,
|
| 9819 |
+
"loss": 0.962,
|
| 9820 |
+
"step": 1397
|
| 9821 |
+
},
|
| 9822 |
+
{
|
| 9823 |
+
"epoch": 0.9204180725866183,
|
| 9824 |
+
"grad_norm": 1.5346226692199707,
|
| 9825 |
+
"learning_rate": 3.2793821856848426e-06,
|
| 9826 |
+
"loss": 0.9109,
|
| 9827 |
+
"step": 1398
|
| 9828 |
+
},
|
| 9829 |
+
{
|
| 9830 |
+
"epoch": 0.9210764546127891,
|
| 9831 |
+
"grad_norm": 1.9029617309570312,
|
| 9832 |
+
"learning_rate": 3.225248048192053e-06,
|
| 9833 |
+
"loss": 1.1107,
|
| 9834 |
+
"step": 1399
|
| 9835 |
+
},
|
| 9836 |
+
{
|
| 9837 |
+
"epoch": 0.9217348366389597,
|
| 9838 |
+
"grad_norm": 1.7355810403823853,
|
| 9839 |
+
"learning_rate": 3.171557120683588e-06,
|
| 9840 |
+
"loss": 1.0064,
|
| 9841 |
+
"step": 1400
|
| 9842 |
+
},
|
| 9843 |
+
{
|
| 9844 |
+
"epoch": 0.9223932186651305,
|
| 9845 |
+
"grad_norm": 0.2773732841014862,
|
| 9846 |
+
"learning_rate": 3.118309649053686e-06,
|
| 9847 |
+
"loss": 1.2105,
|
| 9848 |
+
"step": 1401
|
| 9849 |
+
},
|
| 9850 |
+
{
|
| 9851 |
+
"epoch": 0.9230516006913011,
|
| 9852 |
+
"grad_norm": 0.2957621216773987,
|
| 9853 |
+
"learning_rate": 3.065505877165675e-06,
|
| 9854 |
+
"loss": 1.2446,
|
| 9855 |
+
"step": 1402
|
| 9856 |
+
},
|
| 9857 |
+
{
|
| 9858 |
+
"epoch": 0.9237099827174718,
|
| 9859 |
+
"grad_norm": 0.31012555956840515,
|
| 9860 |
+
"learning_rate": 3.01314604685079e-06,
|
| 9861 |
+
"loss": 1.246,
|
| 9862 |
+
"step": 1403
|
| 9863 |
+
},
|
| 9864 |
+
{
|
| 9865 |
+
"epoch": 0.9243683647436425,
|
| 9866 |
+
"grad_norm": 0.3149036169052124,
|
| 9867 |
+
"learning_rate": 2.9612303979071e-06,
|
| 9868 |
+
"loss": 1.161,
|
| 9869 |
+
"step": 1404
|
| 9870 |
+
},
|
| 9871 |
+
{
|
| 9872 |
+
"epoch": 0.9250267467698132,
|
| 9873 |
+
"grad_norm": 0.35212263464927673,
|
| 9874 |
+
"learning_rate": 2.909759168098436e-06,
|
| 9875 |
+
"loss": 1.2855,
|
| 9876 |
+
"step": 1405
|
| 9877 |
+
},
|
| 9878 |
+
{
|
| 9879 |
+
"epoch": 0.9256851287959839,
|
| 9880 |
+
"grad_norm": 0.3609144389629364,
|
| 9881 |
+
"learning_rate": 2.858732593153246e-06,
|
| 9882 |
+
"loss": 1.1838,
|
| 9883 |
+
"step": 1406
|
| 9884 |
+
},
|
| 9885 |
+
{
|
| 9886 |
+
"epoch": 0.9263435108221546,
|
| 9887 |
+
"grad_norm": 0.3796745538711548,
|
| 9888 |
+
"learning_rate": 2.8081509067635516e-06,
|
| 9889 |
+
"loss": 1.2689,
|
| 9890 |
+
"step": 1407
|
| 9891 |
+
},
|
| 9892 |
+
{
|
| 9893 |
+
"epoch": 0.9270018928483252,
|
| 9894 |
+
"grad_norm": 0.38216182589530945,
|
| 9895 |
+
"learning_rate": 2.7580143405838586e-06,
|
| 9896 |
+
"loss": 1.1406,
|
| 9897 |
+
"step": 1408
|
| 9898 |
+
},
|
| 9899 |
+
{
|
| 9900 |
+
"epoch": 0.9276602748744959,
|
| 9901 |
+
"grad_norm": 0.4440428912639618,
|
| 9902 |
+
"learning_rate": 2.708323124230139e-06,
|
| 9903 |
+
"loss": 1.2299,
|
| 9904 |
+
"step": 1409
|
| 9905 |
+
},
|
| 9906 |
+
{
|
| 9907 |
+
"epoch": 0.9283186569006666,
|
| 9908 |
+
"grad_norm": 0.4657718241214752,
|
| 9909 |
+
"learning_rate": 2.659077485278716e-06,
|
| 9910 |
+
"loss": 1.2485,
|
| 9911 |
+
"step": 1410
|
| 9912 |
+
},
|
| 9913 |
+
{
|
| 9914 |
+
"epoch": 0.9289770389268373,
|
| 9915 |
+
"grad_norm": 0.5070340037345886,
|
| 9916 |
+
"learning_rate": 2.6102776492652805e-06,
|
| 9917 |
+
"loss": 1.4255,
|
| 9918 |
+
"step": 1411
|
| 9919 |
+
},
|
| 9920 |
+
{
|
| 9921 |
+
"epoch": 0.929635420953008,
|
| 9922 |
+
"grad_norm": 0.5538939237594604,
|
| 9923 |
+
"learning_rate": 2.5619238396837662e-06,
|
| 9924 |
+
"loss": 1.4407,
|
| 9925 |
+
"step": 1412
|
| 9926 |
+
},
|
| 9927 |
+
{
|
| 9928 |
+
"epoch": 0.9302938029791786,
|
| 9929 |
+
"grad_norm": 0.5763673186302185,
|
| 9930 |
+
"learning_rate": 2.514016277985487e-06,
|
| 9931 |
+
"loss": 1.2522,
|
| 9932 |
+
"step": 1413
|
| 9933 |
+
},
|
| 9934 |
+
{
|
| 9935 |
+
"epoch": 0.9309521850053494,
|
| 9936 |
+
"grad_norm": 0.6714739203453064,
|
| 9937 |
+
"learning_rate": 2.466555183577968e-06,
|
| 9938 |
+
"loss": 1.2696,
|
| 9939 |
+
"step": 1414
|
| 9940 |
+
},
|
| 9941 |
+
{
|
| 9942 |
+
"epoch": 0.93161056703152,
|
| 9943 |
+
"grad_norm": 0.7577025890350342,
|
| 9944 |
+
"learning_rate": 2.419540773823992e-06,
|
| 9945 |
+
"loss": 1.2229,
|
| 9946 |
+
"step": 1415
|
| 9947 |
+
},
|
| 9948 |
+
{
|
| 9949 |
+
"epoch": 0.9322689490576908,
|
| 9950 |
+
"grad_norm": 1.0093814134597778,
|
| 9951 |
+
"learning_rate": 2.3729732640406233e-06,
|
| 9952 |
+
"loss": 1.2574,
|
| 9953 |
+
"step": 1416
|
| 9954 |
+
},
|
| 9955 |
+
{
|
| 9956 |
+
"epoch": 0.9329273310838614,
|
| 9957 |
+
"grad_norm": 1.0037925243377686,
|
| 9958 |
+
"learning_rate": 2.32685286749823e-06,
|
| 9959 |
+
"loss": 1.1359,
|
| 9960 |
+
"step": 1417
|
| 9961 |
+
},
|
| 9962 |
+
{
|
| 9963 |
+
"epoch": 0.9335857131100321,
|
| 9964 |
+
"grad_norm": 1.3896335363388062,
|
| 9965 |
+
"learning_rate": 2.2811797954194524e-06,
|
| 9966 |
+
"loss": 1.2697,
|
| 9967 |
+
"step": 1418
|
| 9968 |
+
},
|
| 9969 |
+
{
|
| 9970 |
+
"epoch": 0.9342440951362028,
|
| 9971 |
+
"grad_norm": 1.2368346452713013,
|
| 9972 |
+
"learning_rate": 2.235954256978301e-06,
|
| 9973 |
+
"loss": 1.1392,
|
| 9974 |
+
"step": 1419
|
| 9975 |
+
},
|
| 9976 |
+
{
|
| 9977 |
+
"epoch": 0.9349024771623735,
|
| 9978 |
+
"grad_norm": 1.3703478574752808,
|
| 9979 |
+
"learning_rate": 2.191176459299138e-06,
|
| 9980 |
+
"loss": 1.1208,
|
| 9981 |
+
"step": 1420
|
| 9982 |
+
},
|
| 9983 |
+
{
|
| 9984 |
+
"epoch": 0.9355608591885441,
|
| 9985 |
+
"grad_norm": 1.3063801527023315,
|
| 9986 |
+
"learning_rate": 2.146846607455788e-06,
|
| 9987 |
+
"loss": 0.9792,
|
| 9988 |
+
"step": 1421
|
| 9989 |
+
},
|
| 9990 |
+
{
|
| 9991 |
+
"epoch": 0.9362192412147149,
|
| 9992 |
+
"grad_norm": 1.5417190790176392,
|
| 9993 |
+
"learning_rate": 2.1029649044705503e-06,
|
| 9994 |
+
"loss": 1.1611,
|
| 9995 |
+
"step": 1422
|
| 9996 |
+
},
|
| 9997 |
+
{
|
| 9998 |
+
"epoch": 0.9368776232408855,
|
| 9999 |
+
"grad_norm": 1.48011314868927,
|
| 10000 |
+
"learning_rate": 2.059531551313287e-06,
|
| 10001 |
+
"loss": 0.8745,
|
| 10002 |
+
"step": 1423
|
| 10003 |
+
},
|
| 10004 |
+
{
|
| 10005 |
+
"epoch": 0.9375360052670562,
|
| 10006 |
+
"grad_norm": 1.7629079818725586,
|
| 10007 |
+
"learning_rate": 2.016546746900505e-06,
|
| 10008 |
+
"loss": 1.0266,
|
| 10009 |
+
"step": 1424
|
| 10010 |
+
},
|
| 10011 |
+
{
|
| 10012 |
+
"epoch": 0.9381943872932269,
|
| 10013 |
+
"grad_norm": 2.372678518295288,
|
| 10014 |
+
"learning_rate": 1.9740106880944297e-06,
|
| 10015 |
+
"loss": 1.2707,
|
| 10016 |
+
"step": 1425
|
| 10017 |
+
},
|
| 10018 |
+
{
|
| 10019 |
+
"epoch": 0.9388527693193975,
|
| 10020 |
+
"grad_norm": 0.26507827639579773,
|
| 10021 |
+
"learning_rate": 1.9319235697021763e-06,
|
| 10022 |
+
"loss": 1.2121,
|
| 10023 |
+
"step": 1426
|
| 10024 |
+
},
|
| 10025 |
+
{
|
| 10026 |
+
"epoch": 0.9395111513455683,
|
| 10027 |
+
"grad_norm": 0.29203614592552185,
|
| 10028 |
+
"learning_rate": 1.8902855844747047e-06,
|
| 10029 |
+
"loss": 1.1962,
|
| 10030 |
+
"step": 1427
|
| 10031 |
+
},
|
| 10032 |
+
{
|
| 10033 |
+
"epoch": 0.9401695333717389,
|
| 10034 |
+
"grad_norm": 0.3047432601451874,
|
| 10035 |
+
"learning_rate": 1.8490969231061084e-06,
|
| 10036 |
+
"loss": 1.2417,
|
| 10037 |
+
"step": 1428
|
| 10038 |
+
},
|
| 10039 |
+
{
|
| 10040 |
+
"epoch": 0.9408279153979097,
|
| 10041 |
+
"grad_norm": 0.314283162355423,
|
| 10042 |
+
"learning_rate": 1.8083577742326163e-06,
|
| 10043 |
+
"loss": 1.1419,
|
| 10044 |
+
"step": 1429
|
| 10045 |
+
},
|
| 10046 |
+
{
|
| 10047 |
+
"epoch": 0.9414862974240803,
|
| 10048 |
+
"grad_norm": 0.32703661918640137,
|
| 10049 |
+
"learning_rate": 1.7680683244318152e-06,
|
| 10050 |
+
"loss": 1.1603,
|
| 10051 |
+
"step": 1430
|
| 10052 |
+
},
|
| 10053 |
+
{
|
| 10054 |
+
"epoch": 0.942144679450251,
|
| 10055 |
+
"grad_norm": 0.3614916503429413,
|
| 10056 |
+
"learning_rate": 1.728228758221706e-06,
|
| 10057 |
+
"loss": 1.1844,
|
| 10058 |
+
"step": 1431
|
| 10059 |
+
},
|
| 10060 |
+
{
|
| 10061 |
+
"epoch": 0.9428030614764217,
|
| 10062 |
+
"grad_norm": 0.3773248493671417,
|
| 10063 |
+
"learning_rate": 1.688839258059971e-06,
|
| 10064 |
+
"loss": 1.328,
|
| 10065 |
+
"step": 1432
|
| 10066 |
+
},
|
| 10067 |
+
{
|
| 10068 |
+
"epoch": 0.9434614435025924,
|
| 10069 |
+
"grad_norm": 0.42976126074790955,
|
| 10070 |
+
"learning_rate": 1.6499000043429968e-06,
|
| 10071 |
+
"loss": 1.3348,
|
| 10072 |
+
"step": 1433
|
| 10073 |
+
},
|
| 10074 |
+
{
|
| 10075 |
+
"epoch": 0.9441198255287631,
|
| 10076 |
+
"grad_norm": 0.4219948947429657,
|
| 10077 |
+
"learning_rate": 1.6114111754051974e-06,
|
| 10078 |
+
"loss": 1.1941,
|
| 10079 |
+
"step": 1434
|
| 10080 |
+
},
|
| 10081 |
+
{
|
| 10082 |
+
"epoch": 0.9447782075549338,
|
| 10083 |
+
"grad_norm": 0.4261309802532196,
|
| 10084 |
+
"learning_rate": 1.5733729475181036e-06,
|
| 10085 |
+
"loss": 1.2301,
|
| 10086 |
+
"step": 1435
|
| 10087 |
+
},
|
| 10088 |
+
{
|
| 10089 |
+
"epoch": 0.9454365895811044,
|
| 10090 |
+
"grad_norm": 0.46671414375305176,
|
| 10091 |
+
"learning_rate": 1.5357854948895634e-06,
|
| 10092 |
+
"loss": 1.2488,
|
| 10093 |
+
"step": 1436
|
| 10094 |
+
},
|
| 10095 |
+
{
|
| 10096 |
+
"epoch": 0.9460949716072751,
|
| 10097 |
+
"grad_norm": 0.5037201642990112,
|
| 10098 |
+
"learning_rate": 1.4986489896629651e-06,
|
| 10099 |
+
"loss": 1.2346,
|
| 10100 |
+
"step": 1437
|
| 10101 |
+
},
|
| 10102 |
+
{
|
| 10103 |
+
"epoch": 0.9467533536334458,
|
| 10104 |
+
"grad_norm": 0.5437148809432983,
|
| 10105 |
+
"learning_rate": 1.4619636019164606e-06,
|
| 10106 |
+
"loss": 1.2866,
|
| 10107 |
+
"step": 1438
|
| 10108 |
+
},
|
| 10109 |
+
{
|
| 10110 |
+
"epoch": 0.9474117356596164,
|
| 10111 |
+
"grad_norm": 0.6370693445205688,
|
| 10112 |
+
"learning_rate": 1.425729499662154e-06,
|
| 10113 |
+
"loss": 1.2668,
|
| 10114 |
+
"step": 1439
|
| 10115 |
+
},
|
| 10116 |
+
{
|
| 10117 |
+
"epoch": 0.9480701176857872,
|
| 10118 |
+
"grad_norm": 0.6582576036453247,
|
| 10119 |
+
"learning_rate": 1.3899468488453583e-06,
|
| 10120 |
+
"loss": 1.215,
|
| 10121 |
+
"step": 1440
|
| 10122 |
+
},
|
| 10123 |
+
{
|
| 10124 |
+
"epoch": 0.9487284997119578,
|
| 10125 |
+
"grad_norm": 0.7763570547103882,
|
| 10126 |
+
"learning_rate": 1.354615813343818e-06,
|
| 10127 |
+
"loss": 1.1654,
|
| 10128 |
+
"step": 1441
|
| 10129 |
+
},
|
| 10130 |
+
{
|
| 10131 |
+
"epoch": 0.9493868817381286,
|
| 10132 |
+
"grad_norm": 1.096980094909668,
|
| 10133 |
+
"learning_rate": 1.319736554966955e-06,
|
| 10134 |
+
"loss": 1.2844,
|
| 10135 |
+
"step": 1442
|
| 10136 |
+
},
|
| 10137 |
+
{
|
| 10138 |
+
"epoch": 0.9500452637642992,
|
| 10139 |
+
"grad_norm": 0.9722412824630737,
|
| 10140 |
+
"learning_rate": 1.2853092334551452e-06,
|
| 10141 |
+
"loss": 1.1631,
|
| 10142 |
+
"step": 1443
|
| 10143 |
+
},
|
| 10144 |
+
{
|
| 10145 |
+
"epoch": 0.95070364579047,
|
| 10146 |
+
"grad_norm": 1.2841540575027466,
|
| 10147 |
+
"learning_rate": 1.2513340064790102e-06,
|
| 10148 |
+
"loss": 1.0249,
|
| 10149 |
+
"step": 1444
|
| 10150 |
+
},
|
| 10151 |
+
{
|
| 10152 |
+
"epoch": 0.9513620278166406,
|
| 10153 |
+
"grad_norm": 1.3519368171691895,
|
| 10154 |
+
"learning_rate": 1.2178110296386157e-06,
|
| 10155 |
+
"loss": 1.2195,
|
| 10156 |
+
"step": 1445
|
| 10157 |
+
},
|
| 10158 |
+
{
|
| 10159 |
+
"epoch": 0.9520204098428113,
|
| 10160 |
+
"grad_norm": 1.3257054090499878,
|
| 10161 |
+
"learning_rate": 1.1847404564628185e-06,
|
| 10162 |
+
"loss": 1.1797,
|
| 10163 |
+
"step": 1446
|
| 10164 |
+
},
|
| 10165 |
+
{
|
| 10166 |
+
"epoch": 0.952678791868982,
|
| 10167 |
+
"grad_norm": 1.405526876449585,
|
| 10168 |
+
"learning_rate": 1.1521224384085871e-06,
|
| 10169 |
+
"loss": 0.8704,
|
| 10170 |
+
"step": 1447
|
| 10171 |
+
},
|
| 10172 |
+
{
|
| 10173 |
+
"epoch": 0.9533371738951527,
|
| 10174 |
+
"grad_norm": 1.6719517707824707,
|
| 10175 |
+
"learning_rate": 1.1199571248602382e-06,
|
| 10176 |
+
"loss": 1.0272,
|
| 10177 |
+
"step": 1448
|
| 10178 |
+
},
|
| 10179 |
+
{
|
| 10180 |
+
"epoch": 0.9539955559213233,
|
| 10181 |
+
"grad_norm": 1.8807095289230347,
|
| 10182 |
+
"learning_rate": 1.0882446631287902e-06,
|
| 10183 |
+
"loss": 0.8809,
|
| 10184 |
+
"step": 1449
|
| 10185 |
+
},
|
| 10186 |
+
{
|
| 10187 |
+
"epoch": 0.954653937947494,
|
| 10188 |
+
"grad_norm": 2.01116943359375,
|
| 10189 |
+
"learning_rate": 1.0569851984513103e-06,
|
| 10190 |
+
"loss": 0.9926,
|
| 10191 |
+
"step": 1450
|
| 10192 |
+
},
|
| 10193 |
+
{
|
| 10194 |
+
"epoch": 0.9553123199736647,
|
| 10195 |
+
"grad_norm": 0.26615461707115173,
|
| 10196 |
+
"learning_rate": 1.0261788739902022e-06,
|
| 10197 |
+
"loss": 1.2286,
|
| 10198 |
+
"step": 1451
|
| 10199 |
+
},
|
| 10200 |
+
{
|
| 10201 |
+
"epoch": 0.9559707019998354,
|
| 10202 |
+
"grad_norm": 0.28414100408554077,
|
| 10203 |
+
"learning_rate": 9.958258308325973e-07,
|
| 10204 |
+
"loss": 1.2566,
|
| 10205 |
+
"step": 1452
|
| 10206 |
+
},
|
| 10207 |
+
{
|
| 10208 |
+
"epoch": 0.9566290840260061,
|
| 10209 |
+
"grad_norm": 0.28735122084617615,
|
| 10210 |
+
"learning_rate": 9.659262079896314e-07,
|
| 10211 |
+
"loss": 1.1613,
|
| 10212 |
+
"step": 1453
|
| 10213 |
+
},
|
| 10214 |
+
{
|
| 10215 |
+
"epoch": 0.9572874660521767,
|
| 10216 |
+
"grad_norm": 0.31062746047973633,
|
| 10217 |
+
"learning_rate": 9.364801423959235e-07,
|
| 10218 |
+
"loss": 1.2663,
|
| 10219 |
+
"step": 1454
|
| 10220 |
+
},
|
| 10221 |
+
{
|
| 10222 |
+
"epoch": 0.9579458480783475,
|
| 10223 |
+
"grad_norm": 0.331766813993454,
|
| 10224 |
+
"learning_rate": 9.074877689088768e-07,
|
| 10225 |
+
"loss": 1.2882,
|
| 10226 |
+
"step": 1455
|
| 10227 |
+
},
|
| 10228 |
+
{
|
| 10229 |
+
"epoch": 0.9586042301045181,
|
| 10230 |
+
"grad_norm": 0.34377574920654297,
|
| 10231 |
+
"learning_rate": 8.78949220308023e-07,
|
| 10232 |
+
"loss": 1.2723,
|
| 10233 |
+
"step": 1456
|
| 10234 |
+
},
|
| 10235 |
+
{
|
| 10236 |
+
"epoch": 0.9592626121306889,
|
| 10237 |
+
"grad_norm": 0.35146182775497437,
|
| 10238 |
+
"learning_rate": 8.50864627294512e-07,
|
| 10239 |
+
"loss": 1.1882,
|
| 10240 |
+
"step": 1457
|
| 10241 |
+
},
|
| 10242 |
+
{
|
| 10243 |
+
"epoch": 0.9599209941568595,
|
| 10244 |
+
"grad_norm": 0.37997257709503174,
|
| 10245 |
+
"learning_rate": 8.232341184904457e-07,
|
| 10246 |
+
"loss": 1.2171,
|
| 10247 |
+
"step": 1458
|
| 10248 |
+
},
|
| 10249 |
+
{
|
| 10250 |
+
"epoch": 0.9605793761830302,
|
| 10251 |
+
"grad_norm": 0.4185076057910919,
|
| 10252 |
+
"learning_rate": 7.960578204383007e-07,
|
| 10253 |
+
"loss": 1.2212,
|
| 10254 |
+
"step": 1459
|
| 10255 |
+
},
|
| 10256 |
+
{
|
| 10257 |
+
"epoch": 0.9612377582092009,
|
| 10258 |
+
"grad_norm": 0.4693465828895569,
|
| 10259 |
+
"learning_rate": 7.693358576003617e-07,
|
| 10260 |
+
"loss": 1.2836,
|
| 10261 |
+
"step": 1460
|
| 10262 |
+
},
|
| 10263 |
+
{
|
| 10264 |
+
"epoch": 0.9618961402353716,
|
| 10265 |
+
"grad_norm": 0.4621378779411316,
|
| 10266 |
+
"learning_rate": 7.430683523581561e-07,
|
| 10267 |
+
"loss": 1.2892,
|
| 10268 |
+
"step": 1461
|
| 10269 |
+
},
|
| 10270 |
+
{
|
| 10271 |
+
"epoch": 0.9625545222615423,
|
| 10272 |
+
"grad_norm": 0.49882352352142334,
|
| 10273 |
+
"learning_rate": 7.172554250118535e-07,
|
| 10274 |
+
"loss": 1.2688,
|
| 10275 |
+
"step": 1462
|
| 10276 |
+
},
|
| 10277 |
+
{
|
| 10278 |
+
"epoch": 0.9632129042877129,
|
| 10279 |
+
"grad_norm": 0.5431182384490967,
|
| 10280 |
+
"learning_rate": 6.91897193779767e-07,
|
| 10281 |
+
"loss": 1.3595,
|
| 10282 |
+
"step": 1463
|
| 10283 |
+
},
|
| 10284 |
+
{
|
| 10285 |
+
"epoch": 0.9638712863138836,
|
| 10286 |
+
"grad_norm": 0.6177190542221069,
|
| 10287 |
+
"learning_rate": 6.66993774797775e-07,
|
| 10288 |
+
"loss": 1.3173,
|
| 10289 |
+
"step": 1464
|
| 10290 |
+
},
|
| 10291 |
+
{
|
| 10292 |
+
"epoch": 0.9645296683400543,
|
| 10293 |
+
"grad_norm": 0.6704779863357544,
|
| 10294 |
+
"learning_rate": 6.425452821188116e-07,
|
| 10295 |
+
"loss": 1.2451,
|
| 10296 |
+
"step": 1465
|
| 10297 |
+
},
|
| 10298 |
+
{
|
| 10299 |
+
"epoch": 0.965188050366225,
|
| 10300 |
+
"grad_norm": 0.7778468132019043,
|
| 10301 |
+
"learning_rate": 6.185518277123214e-07,
|
| 10302 |
+
"loss": 1.1512,
|
| 10303 |
+
"step": 1466
|
| 10304 |
+
},
|
| 10305 |
+
{
|
| 10306 |
+
"epoch": 0.9658464323923956,
|
| 10307 |
+
"grad_norm": 0.7778995633125305,
|
| 10308 |
+
"learning_rate": 5.950135214637831e-07,
|
| 10309 |
+
"loss": 1.1375,
|
| 10310 |
+
"step": 1467
|
| 10311 |
+
},
|
| 10312 |
+
{
|
| 10313 |
+
"epoch": 0.9665048144185664,
|
| 10314 |
+
"grad_norm": 0.8882863521575928,
|
| 10315 |
+
"learning_rate": 5.719304711741535e-07,
|
| 10316 |
+
"loss": 1.2534,
|
| 10317 |
+
"step": 1468
|
| 10318 |
+
},
|
| 10319 |
+
{
|
| 10320 |
+
"epoch": 0.967163196444737,
|
| 10321 |
+
"grad_norm": 1.0903034210205078,
|
| 10322 |
+
"learning_rate": 5.493027825594244e-07,
|
| 10323 |
+
"loss": 1.0416,
|
| 10324 |
+
"step": 1469
|
| 10325 |
+
},
|
| 10326 |
+
{
|
| 10327 |
+
"epoch": 0.9678215784709078,
|
| 10328 |
+
"grad_norm": 1.4687763452529907,
|
| 10329 |
+
"learning_rate": 5.271305592501108e-07,
|
| 10330 |
+
"loss": 1.0645,
|
| 10331 |
+
"step": 1470
|
| 10332 |
+
},
|
| 10333 |
+
{
|
| 10334 |
+
"epoch": 0.9684799604970784,
|
| 10335 |
+
"grad_norm": 1.4143941402435303,
|
| 10336 |
+
"learning_rate": 5.054139027907967e-07,
|
| 10337 |
+
"loss": 1.0708,
|
| 10338 |
+
"step": 1471
|
| 10339 |
+
},
|
| 10340 |
+
{
|
| 10341 |
+
"epoch": 0.9691383425232492,
|
| 10342 |
+
"grad_norm": 1.5318214893341064,
|
| 10343 |
+
"learning_rate": 4.841529126396238e-07,
|
| 10344 |
+
"loss": 0.9667,
|
| 10345 |
+
"step": 1472
|
| 10346 |
+
},
|
| 10347 |
+
{
|
| 10348 |
+
"epoch": 0.9697967245494198,
|
| 10349 |
+
"grad_norm": 1.5767490863800049,
|
| 10350 |
+
"learning_rate": 4.633476861679142e-07,
|
| 10351 |
+
"loss": 0.7994,
|
| 10352 |
+
"step": 1473
|
| 10353 |
+
},
|
| 10354 |
+
{
|
| 10355 |
+
"epoch": 0.9704551065755905,
|
| 10356 |
+
"grad_norm": 1.8559094667434692,
|
| 10357 |
+
"learning_rate": 4.429983186596265e-07,
|
| 10358 |
+
"loss": 1.096,
|
| 10359 |
+
"step": 1474
|
| 10360 |
+
},
|
| 10361 |
+
{
|
| 10362 |
+
"epoch": 0.9711134886017612,
|
| 10363 |
+
"grad_norm": 2.398397922515869,
|
| 10364 |
+
"learning_rate": 4.2310490331102237e-07,
|
| 10365 |
+
"loss": 1.0258,
|
| 10366 |
+
"step": 1475
|
| 10367 |
+
},
|
| 10368 |
+
{
|
| 10369 |
+
"epoch": 0.9717718706279319,
|
| 10370 |
+
"grad_norm": 0.27290716767311096,
|
| 10371 |
+
"learning_rate": 4.036675312301452e-07,
|
| 10372 |
+
"loss": 1.1712,
|
| 10373 |
+
"step": 1476
|
| 10374 |
+
},
|
| 10375 |
+
{
|
| 10376 |
+
"epoch": 0.9724302526541025,
|
| 10377 |
+
"grad_norm": 0.27547338604927063,
|
| 10378 |
+
"learning_rate": 3.846862914364868e-07,
|
| 10379 |
+
"loss": 1.1352,
|
| 10380 |
+
"step": 1477
|
| 10381 |
+
},
|
| 10382 |
+
{
|
| 10383 |
+
"epoch": 0.9730886346802732,
|
| 10384 |
+
"grad_norm": 0.3008437752723694,
|
| 10385 |
+
"learning_rate": 3.6616127086051e-07,
|
| 10386 |
+
"loss": 1.2512,
|
| 10387 |
+
"step": 1478
|
| 10388 |
+
},
|
| 10389 |
+
{
|
| 10390 |
+
"epoch": 0.9737470167064439,
|
| 10391 |
+
"grad_norm": 0.3131862282752991,
|
| 10392 |
+
"learning_rate": 3.4809255434328227e-07,
|
| 10393 |
+
"loss": 1.217,
|
| 10394 |
+
"step": 1479
|
| 10395 |
+
},
|
| 10396 |
+
{
|
| 10397 |
+
"epoch": 0.9744053987326146,
|
| 10398 |
+
"grad_norm": 0.32987430691719055,
|
| 10399 |
+
"learning_rate": 3.3048022463612047e-07,
|
| 10400 |
+
"loss": 1.2212,
|
| 10401 |
+
"step": 1480
|
| 10402 |
+
},
|
| 10403 |
+
{
|
| 10404 |
+
"epoch": 0.9750637807587853,
|
| 10405 |
+
"grad_norm": 0.36609992384910583,
|
| 10406 |
+
"learning_rate": 3.1332436240011364e-07,
|
| 10407 |
+
"loss": 1.2002,
|
| 10408 |
+
"step": 1481
|
| 10409 |
+
},
|
| 10410 |
+
{
|
| 10411 |
+
"epoch": 0.9757221627849559,
|
| 10412 |
+
"grad_norm": 0.38834360241889954,
|
| 10413 |
+
"learning_rate": 2.9662504620588947e-07,
|
| 10414 |
+
"loss": 1.2311,
|
| 10415 |
+
"step": 1482
|
| 10416 |
+
},
|
| 10417 |
+
{
|
| 10418 |
+
"epoch": 0.9763805448111267,
|
| 10419 |
+
"grad_norm": 0.3734608292579651,
|
| 10420 |
+
"learning_rate": 2.8038235253311505e-07,
|
| 10421 |
+
"loss": 1.15,
|
| 10422 |
+
"step": 1483
|
| 10423 |
+
},
|
| 10424 |
+
{
|
| 10425 |
+
"epoch": 0.9770389268372973,
|
| 10426 |
+
"grad_norm": 0.39932605624198914,
|
| 10427 |
+
"learning_rate": 2.6459635577026353e-07,
|
| 10428 |
+
"loss": 1.1955,
|
| 10429 |
+
"step": 1484
|
| 10430 |
+
},
|
| 10431 |
+
{
|
| 10432 |
+
"epoch": 0.9776973088634681,
|
| 10433 |
+
"grad_norm": 0.4378267228603363,
|
| 10434 |
+
"learning_rate": 2.492671282141923e-07,
|
| 10435 |
+
"loss": 1.1007,
|
| 10436 |
+
"step": 1485
|
| 10437 |
+
},
|
| 10438 |
+
{
|
| 10439 |
+
"epoch": 0.9783556908896387,
|
| 10440 |
+
"grad_norm": 0.4685000479221344,
|
| 10441 |
+
"learning_rate": 2.343947400698432e-07,
|
| 10442 |
+
"loss": 1.2366,
|
| 10443 |
+
"step": 1486
|
| 10444 |
+
},
|
| 10445 |
+
{
|
| 10446 |
+
"epoch": 0.9790140729158094,
|
| 10447 |
+
"grad_norm": 0.5255098938941956,
|
| 10448 |
+
"learning_rate": 2.199792594499428e-07,
|
| 10449 |
+
"loss": 1.3388,
|
| 10450 |
+
"step": 1487
|
| 10451 |
+
},
|
| 10452 |
+
{
|
| 10453 |
+
"epoch": 0.9796724549419801,
|
| 10454 |
+
"grad_norm": 0.5867619514465332,
|
| 10455 |
+
"learning_rate": 2.0602075237465823e-07,
|
| 10456 |
+
"loss": 1.2941,
|
| 10457 |
+
"step": 1488
|
| 10458 |
+
},
|
| 10459 |
+
{
|
| 10460 |
+
"epoch": 0.9803308369681508,
|
| 10461 |
+
"grad_norm": 0.5758997201919556,
|
| 10462 |
+
"learning_rate": 1.9251928277128628e-07,
|
| 10463 |
+
"loss": 1.1415,
|
| 10464 |
+
"step": 1489
|
| 10465 |
+
},
|
| 10466 |
+
{
|
| 10467 |
+
"epoch": 0.9809892189943215,
|
| 10468 |
+
"grad_norm": 0.8498535752296448,
|
| 10469 |
+
"learning_rate": 1.7947491247399806e-07,
|
| 10470 |
+
"loss": 1.2043,
|
| 10471 |
+
"step": 1490
|
| 10472 |
+
},
|
| 10473 |
+
{
|
| 10474 |
+
"epoch": 0.9816476010204921,
|
| 10475 |
+
"grad_norm": 0.8512412309646606,
|
| 10476 |
+
"learning_rate": 1.6688770122353925e-07,
|
| 10477 |
+
"loss": 1.3614,
|
| 10478 |
+
"step": 1491
|
| 10479 |
+
},
|
| 10480 |
+
{
|
| 10481 |
+
"epoch": 0.9823059830466628,
|
| 10482 |
+
"grad_norm": 0.9369918704032898,
|
| 10483 |
+
"learning_rate": 1.5475770666694144e-07,
|
| 10484 |
+
"loss": 1.1451,
|
| 10485 |
+
"step": 1492
|
| 10486 |
+
},
|
| 10487 |
+
{
|
| 10488 |
+
"epoch": 0.9829643650728335,
|
| 10489 |
+
"grad_norm": 1.0150845050811768,
|
| 10490 |
+
"learning_rate": 1.430849843572779e-07,
|
| 10491 |
+
"loss": 1.2014,
|
| 10492 |
+
"step": 1493
|
| 10493 |
+
},
|
| 10494 |
+
{
|
| 10495 |
+
"epoch": 0.9836227470990042,
|
| 10496 |
+
"grad_norm": 1.2079380750656128,
|
| 10497 |
+
"learning_rate": 1.318695877533971e-07,
|
| 10498 |
+
"loss": 1.3343,
|
| 10499 |
+
"step": 1494
|
| 10500 |
+
},
|
| 10501 |
+
{
|
| 10502 |
+
"epoch": 0.9842811291251748,
|
| 10503 |
+
"grad_norm": 1.062026858329773,
|
| 10504 |
+
"learning_rate": 1.211115682196895e-07,
|
| 10505 |
+
"loss": 0.9174,
|
| 10506 |
+
"step": 1495
|
| 10507 |
+
},
|
| 10508 |
+
{
|
| 10509 |
+
"epoch": 0.9849395111513456,
|
| 10510 |
+
"grad_norm": 1.1822501420974731,
|
| 10511 |
+
"learning_rate": 1.1081097502584348e-07,
|
| 10512 |
+
"loss": 0.9607,
|
| 10513 |
+
"step": 1496
|
| 10514 |
+
},
|
| 10515 |
+
{
|
| 10516 |
+
"epoch": 0.9855978931775162,
|
| 10517 |
+
"grad_norm": 1.4208043813705444,
|
| 10518 |
+
"learning_rate": 1.0096785534660092e-07,
|
| 10519 |
+
"loss": 0.9065,
|
| 10520 |
+
"step": 1497
|
| 10521 |
+
},
|
| 10522 |
+
{
|
| 10523 |
+
"epoch": 0.986256275203687,
|
| 10524 |
+
"grad_norm": 1.4563997983932495,
|
| 10525 |
+
"learning_rate": 9.158225426160183e-08,
|
| 10526 |
+
"loss": 0.9299,
|
| 10527 |
+
"step": 1498
|
| 10528 |
+
},
|
| 10529 |
+
{
|
| 10530 |
+
"epoch": 0.9869146572298576,
|
| 10531 |
+
"grad_norm": 3.9852182865142822,
|
| 10532 |
+
"learning_rate": 8.265421475511792e-08,
|
| 10533 |
+
"loss": 1.2768,
|
| 10534 |
+
"step": 1499
|
| 10535 |
+
},
|
| 10536 |
+
{
|
| 10537 |
+
"epoch": 0.9875730392560284,
|
| 10538 |
+
"grad_norm": 2.2535030841827393,
|
| 10539 |
+
"learning_rate": 7.418377771585273e-08,
|
| 10540 |
+
"loss": 1.2758,
|
| 10541 |
+
"step": 1500
|
| 10542 |
+
},
|
| 10543 |
+
{
|
| 10544 |
+
"epoch": 0.988231421282199,
|
| 10545 |
+
"grad_norm": 0.26907411217689514,
|
| 10546 |
+
"learning_rate": 6.617098193681947e-08,
|
| 10547 |
+
"loss": 1.2956,
|
| 10548 |
+
"step": 1501
|
| 10549 |
+
},
|
| 10550 |
+
{
|
| 10551 |
+
"epoch": 0.9888898033083697,
|
| 10552 |
+
"grad_norm": 0.2853264808654785,
|
| 10553 |
+
"learning_rate": 5.86158641150969e-08,
|
| 10554 |
+
"loss": 1.1471,
|
| 10555 |
+
"step": 1502
|
| 10556 |
+
},
|
| 10557 |
+
{
|
| 10558 |
+
"epoch": 0.9895481853345404,
|
| 10559 |
+
"grad_norm": 0.3077790439128876,
|
| 10560 |
+
"learning_rate": 5.151845885167372e-08,
|
| 10561 |
+
"loss": 1.2542,
|
| 10562 |
+
"step": 1503
|
| 10563 |
+
},
|
| 10564 |
+
{
|
| 10565 |
+
"epoch": 0.990206567360711,
|
| 10566 |
+
"grad_norm": 0.33259809017181396,
|
| 10567 |
+
"learning_rate": 4.487879865133771e-08,
|
| 10568 |
+
"loss": 1.1694,
|
| 10569 |
+
"step": 1504
|
| 10570 |
+
},
|
| 10571 |
+
{
|
| 10572 |
+
"epoch": 0.9908649493868817,
|
| 10573 |
+
"grad_norm": 0.35665038228034973,
|
| 10574 |
+
"learning_rate": 3.8696913922475764e-08,
|
| 10575 |
+
"loss": 1.264,
|
| 10576 |
+
"step": 1505
|
| 10577 |
+
},
|
| 10578 |
+
{
|
| 10579 |
+
"epoch": 0.9915233314130524,
|
| 10580 |
+
"grad_norm": 0.38716834783554077,
|
| 10581 |
+
"learning_rate": 3.2972832976918554e-08,
|
| 10582 |
+
"loss": 1.3199,
|
| 10583 |
+
"step": 1506
|
| 10584 |
+
},
|
| 10585 |
+
{
|
| 10586 |
+
"epoch": 0.9921817134392231,
|
| 10587 |
+
"grad_norm": 0.43460676074028015,
|
| 10588 |
+
"learning_rate": 2.7706582029896068e-08,
|
| 10589 |
+
"loss": 1.2493,
|
| 10590 |
+
"step": 1507
|
| 10591 |
+
},
|
| 10592 |
+
{
|
| 10593 |
+
"epoch": 0.9928400954653938,
|
| 10594 |
+
"grad_norm": 0.4777851104736328,
|
| 10595 |
+
"learning_rate": 2.2898185199826673e-08,
|
| 10596 |
+
"loss": 1.3686,
|
| 10597 |
+
"step": 1508
|
| 10598 |
+
},
|
| 10599 |
+
{
|
| 10600 |
+
"epoch": 0.9934984774915645,
|
| 10601 |
+
"grad_norm": 0.5349164009094238,
|
| 10602 |
+
"learning_rate": 1.854766450826162e-08,
|
| 10603 |
+
"loss": 1.2081,
|
| 10604 |
+
"step": 1509
|
| 10605 |
+
},
|
| 10606 |
+
{
|
| 10607 |
+
"epoch": 0.9941568595177351,
|
| 10608 |
+
"grad_norm": 0.5055113434791565,
|
| 10609 |
+
"learning_rate": 1.4655039879740706e-08,
|
| 10610 |
+
"loss": 1.0484,
|
| 10611 |
+
"step": 1510
|
| 10612 |
+
},
|
| 10613 |
+
{
|
| 10614 |
+
"epoch": 0.9948152415439059,
|
| 10615 |
+
"grad_norm": 0.6259576678276062,
|
| 10616 |
+
"learning_rate": 1.122032914177007e-08,
|
| 10617 |
+
"loss": 1.2071,
|
| 10618 |
+
"step": 1511
|
| 10619 |
+
},
|
| 10620 |
+
{
|
| 10621 |
+
"epoch": 0.9954736235700765,
|
| 10622 |
+
"grad_norm": 0.7254166007041931,
|
| 10623 |
+
"learning_rate": 8.243548024655656e-09,
|
| 10624 |
+
"loss": 1.3634,
|
| 10625 |
+
"step": 1512
|
| 10626 |
+
},
|
| 10627 |
+
{
|
| 10628 |
+
"epoch": 0.9961320055962473,
|
| 10629 |
+
"grad_norm": 0.7936909794807434,
|
| 10630 |
+
"learning_rate": 5.72471016149212e-09,
|
| 10631 |
+
"loss": 1.2007,
|
| 10632 |
+
"step": 1513
|
| 10633 |
+
},
|
| 10634 |
+
{
|
| 10635 |
+
"epoch": 0.9967903876224179,
|
| 10636 |
+
"grad_norm": 0.9308106303215027,
|
| 10637 |
+
"learning_rate": 3.663827088085103e-09,
|
| 10638 |
+
"loss": 1.0131,
|
| 10639 |
+
"step": 1514
|
| 10640 |
+
},
|
| 10641 |
+
{
|
| 10642 |
+
"epoch": 0.9974487696485886,
|
| 10643 |
+
"grad_norm": 1.1433625221252441,
|
| 10644 |
+
"learning_rate": 2.060908242873527e-09,
|
| 10645 |
+
"loss": 1.1215,
|
| 10646 |
+
"step": 1515
|
| 10647 |
+
},
|
| 10648 |
+
{
|
| 10649 |
+
"epoch": 0.9981071516747593,
|
| 10650 |
+
"grad_norm": 1.2838548421859741,
|
| 10651 |
+
"learning_rate": 9.159609669406877e-10,
|
| 10652 |
+
"loss": 1.1398,
|
| 10653 |
+
"step": 1516
|
| 10654 |
+
},
|
| 10655 |
+
{
|
| 10656 |
+
"epoch": 0.99876553370093,
|
| 10657 |
+
"grad_norm": 1.4757442474365234,
|
| 10658 |
+
"learning_rate": 2.2899050391433918e-10,
|
| 10659 |
+
"loss": 1.1327,
|
| 10660 |
+
"step": 1517
|
| 10661 |
+
},
|
| 10662 |
+
{
|
| 10663 |
+
"epoch": 0.9994239157271007,
|
| 10664 |
+
"grad_norm": 1.396567463874817,
|
| 10665 |
+
"learning_rate": 0.0,
|
| 10666 |
+
"loss": 0.6568,
|
| 10667 |
+
"step": 1518
|
| 10668 |
}
|
| 10669 |
],
|
| 10670 |
"logging_steps": 1,
|
|
|
|
| 10679 |
"should_evaluate": false,
|
| 10680 |
"should_log": false,
|
| 10681 |
"should_save": true,
|
| 10682 |
+
"should_training_stop": true
|
| 10683 |
},
|
| 10684 |
"attributes": {}
|
| 10685 |
}
|
| 10686 |
},
|
| 10687 |
+
"total_flos": 1.175202288478716e+18,
|
| 10688 |
"train_batch_size": 2,
|
| 10689 |
"trial_name": null,
|
| 10690 |
"trial_params": null
|