Training in progress, step 1392, 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 17640136
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e9c50ec66b0a5205ea686ab7f963037939e2290e7c424e63d9b5f687a94adb1b
|
| 3 |
size 17640136
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 9569204
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e5da7829272c9a2d0de8fc92720b5720028da7b325c5337059a306a8e92ab65e
|
| 3 |
size 9569204
|
last-checkpoint/rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14244
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0664753154575bfb6a2520e17a9f7f5686f941595d2615eeacbc06d1a09a97fb
|
| 3 |
size 14244
|
last-checkpoint/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1064
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2e5e580be9b534c9686b8622eb7165521326ac290f4c34eef1bfcf943b6e153a
|
| 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": 348,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -7347,6 +7347,2450 @@
|
|
| 7347 |
"eval_samples_per_second": 45.478,
|
| 7348 |
"eval_steps_per_second": 22.739,
|
| 7349 |
"step": 1044
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 7350 |
}
|
| 7351 |
],
|
| 7352 |
"logging_steps": 1,
|
|
@@ -7361,12 +9805,12 @@
|
|
| 7361 |
"should_evaluate": false,
|
| 7362 |
"should_log": false,
|
| 7363 |
"should_save": true,
|
| 7364 |
-
"should_training_stop":
|
| 7365 |
},
|
| 7366 |
"attributes": {}
|
| 7367 |
}
|
| 7368 |
},
|
| 7369 |
-
"total_flos":
|
| 7370 |
"train_batch_size": 2,
|
| 7371 |
"trial_name": null,
|
| 7372 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.7920341394025604,
|
| 5 |
"eval_steps": 348,
|
| 6 |
+
"global_step": 1392,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 7347 |
"eval_samples_per_second": 45.478,
|
| 7348 |
"eval_steps_per_second": 22.739,
|
| 7349 |
"step": 1044
|
| 7350 |
+
},
|
| 7351 |
+
{
|
| 7352 |
+
"epoch": 0.5945945945945946,
|
| 7353 |
+
"grad_norm": 0.5834308862686157,
|
| 7354 |
+
"learning_rate": 2.9530844080984565e-05,
|
| 7355 |
+
"loss": 1.4174,
|
| 7356 |
+
"step": 1045
|
| 7357 |
+
},
|
| 7358 |
+
{
|
| 7359 |
+
"epoch": 0.5951635846372688,
|
| 7360 |
+
"grad_norm": 0.5811535120010376,
|
| 7361 |
+
"learning_rate": 2.9369738069767107e-05,
|
| 7362 |
+
"loss": 1.2859,
|
| 7363 |
+
"step": 1046
|
| 7364 |
+
},
|
| 7365 |
+
{
|
| 7366 |
+
"epoch": 0.5957325746799431,
|
| 7367 |
+
"grad_norm": 0.6040303707122803,
|
| 7368 |
+
"learning_rate": 2.920899704293849e-05,
|
| 7369 |
+
"loss": 1.7526,
|
| 7370 |
+
"step": 1047
|
| 7371 |
+
},
|
| 7372 |
+
{
|
| 7373 |
+
"epoch": 0.5963015647226173,
|
| 7374 |
+
"grad_norm": 0.5936810970306396,
|
| 7375 |
+
"learning_rate": 2.9048621831133616e-05,
|
| 7376 |
+
"loss": 1.3031,
|
| 7377 |
+
"step": 1048
|
| 7378 |
+
},
|
| 7379 |
+
{
|
| 7380 |
+
"epoch": 0.5968705547652916,
|
| 7381 |
+
"grad_norm": 0.5825332999229431,
|
| 7382 |
+
"learning_rate": 2.8888613263097153e-05,
|
| 7383 |
+
"loss": 1.3483,
|
| 7384 |
+
"step": 1049
|
| 7385 |
+
},
|
| 7386 |
+
{
|
| 7387 |
+
"epoch": 0.5974395448079659,
|
| 7388 |
+
"grad_norm": 0.6082255244255066,
|
| 7389 |
+
"learning_rate": 2.8728972165679067e-05,
|
| 7390 |
+
"loss": 1.528,
|
| 7391 |
+
"step": 1050
|
| 7392 |
+
},
|
| 7393 |
+
{
|
| 7394 |
+
"epoch": 0.5980085348506401,
|
| 7395 |
+
"grad_norm": 0.594572126865387,
|
| 7396 |
+
"learning_rate": 2.8569699363830316e-05,
|
| 7397 |
+
"loss": 1.5789,
|
| 7398 |
+
"step": 1051
|
| 7399 |
+
},
|
| 7400 |
+
{
|
| 7401 |
+
"epoch": 0.5985775248933144,
|
| 7402 |
+
"grad_norm": 0.6006420850753784,
|
| 7403 |
+
"learning_rate": 2.8410795680598846e-05,
|
| 7404 |
+
"loss": 1.5638,
|
| 7405 |
+
"step": 1052
|
| 7406 |
+
},
|
| 7407 |
+
{
|
| 7408 |
+
"epoch": 0.5991465149359886,
|
| 7409 |
+
"grad_norm": 0.5715523958206177,
|
| 7410 |
+
"learning_rate": 2.825226193712507e-05,
|
| 7411 |
+
"loss": 1.5222,
|
| 7412 |
+
"step": 1053
|
| 7413 |
+
},
|
| 7414 |
+
{
|
| 7415 |
+
"epoch": 0.5997155049786629,
|
| 7416 |
+
"grad_norm": 0.5750184059143066,
|
| 7417 |
+
"learning_rate": 2.8094098952637692e-05,
|
| 7418 |
+
"loss": 1.5154,
|
| 7419 |
+
"step": 1054
|
| 7420 |
+
},
|
| 7421 |
+
{
|
| 7422 |
+
"epoch": 0.6002844950213371,
|
| 7423 |
+
"grad_norm": 0.5836694240570068,
|
| 7424 |
+
"learning_rate": 2.793630754444967e-05,
|
| 7425 |
+
"loss": 1.4624,
|
| 7426 |
+
"step": 1055
|
| 7427 |
+
},
|
| 7428 |
+
{
|
| 7429 |
+
"epoch": 0.6008534850640114,
|
| 7430 |
+
"grad_norm": 0.5644353628158569,
|
| 7431 |
+
"learning_rate": 2.7778888527953572e-05,
|
| 7432 |
+
"loss": 1.564,
|
| 7433 |
+
"step": 1056
|
| 7434 |
+
},
|
| 7435 |
+
{
|
| 7436 |
+
"epoch": 0.6014224751066857,
|
| 7437 |
+
"grad_norm": 0.6327478885650635,
|
| 7438 |
+
"learning_rate": 2.762184271661785e-05,
|
| 7439 |
+
"loss": 1.4707,
|
| 7440 |
+
"step": 1057
|
| 7441 |
+
},
|
| 7442 |
+
{
|
| 7443 |
+
"epoch": 0.6019914651493599,
|
| 7444 |
+
"grad_norm": 0.5783342719078064,
|
| 7445 |
+
"learning_rate": 2.746517092198231e-05,
|
| 7446 |
+
"loss": 1.4888,
|
| 7447 |
+
"step": 1058
|
| 7448 |
+
},
|
| 7449 |
+
{
|
| 7450 |
+
"epoch": 0.6025604551920342,
|
| 7451 |
+
"grad_norm": 0.5796740651130676,
|
| 7452 |
+
"learning_rate": 2.730887395365397e-05,
|
| 7453 |
+
"loss": 1.5201,
|
| 7454 |
+
"step": 1059
|
| 7455 |
+
},
|
| 7456 |
+
{
|
| 7457 |
+
"epoch": 0.6031294452347084,
|
| 7458 |
+
"grad_norm": 0.5543321967124939,
|
| 7459 |
+
"learning_rate": 2.715295261930306e-05,
|
| 7460 |
+
"loss": 1.4378,
|
| 7461 |
+
"step": 1060
|
| 7462 |
+
},
|
| 7463 |
+
{
|
| 7464 |
+
"epoch": 0.6036984352773827,
|
| 7465 |
+
"grad_norm": 0.6244597434997559,
|
| 7466 |
+
"learning_rate": 2.699740772465851e-05,
|
| 7467 |
+
"loss": 1.4242,
|
| 7468 |
+
"step": 1061
|
| 7469 |
+
},
|
| 7470 |
+
{
|
| 7471 |
+
"epoch": 0.604267425320057,
|
| 7472 |
+
"grad_norm": 0.5890554785728455,
|
| 7473 |
+
"learning_rate": 2.6842240073504165e-05,
|
| 7474 |
+
"loss": 1.4732,
|
| 7475 |
+
"step": 1062
|
| 7476 |
+
},
|
| 7477 |
+
{
|
| 7478 |
+
"epoch": 0.6048364153627311,
|
| 7479 |
+
"grad_norm": 0.5934953689575195,
|
| 7480 |
+
"learning_rate": 2.668745046767436e-05,
|
| 7481 |
+
"loss": 1.517,
|
| 7482 |
+
"step": 1063
|
| 7483 |
+
},
|
| 7484 |
+
{
|
| 7485 |
+
"epoch": 0.6054054054054054,
|
| 7486 |
+
"grad_norm": 0.5716105103492737,
|
| 7487 |
+
"learning_rate": 2.6533039707049834e-05,
|
| 7488 |
+
"loss": 1.2859,
|
| 7489 |
+
"step": 1064
|
| 7490 |
+
},
|
| 7491 |
+
{
|
| 7492 |
+
"epoch": 0.6059743954480796,
|
| 7493 |
+
"grad_norm": 0.5798661708831787,
|
| 7494 |
+
"learning_rate": 2.63790085895538e-05,
|
| 7495 |
+
"loss": 1.5015,
|
| 7496 |
+
"step": 1065
|
| 7497 |
+
},
|
| 7498 |
+
{
|
| 7499 |
+
"epoch": 0.6065433854907539,
|
| 7500 |
+
"grad_norm": 0.600385844707489,
|
| 7501 |
+
"learning_rate": 2.6225357911147385e-05,
|
| 7502 |
+
"loss": 1.4027,
|
| 7503 |
+
"step": 1066
|
| 7504 |
+
},
|
| 7505 |
+
{
|
| 7506 |
+
"epoch": 0.6071123755334281,
|
| 7507 |
+
"grad_norm": 0.5749977231025696,
|
| 7508 |
+
"learning_rate": 2.6072088465826038e-05,
|
| 7509 |
+
"loss": 1.5876,
|
| 7510 |
+
"step": 1067
|
| 7511 |
+
},
|
| 7512 |
+
{
|
| 7513 |
+
"epoch": 0.6076813655761024,
|
| 7514 |
+
"grad_norm": 0.5585724711418152,
|
| 7515 |
+
"learning_rate": 2.591920104561503e-05,
|
| 7516 |
+
"loss": 1.4756,
|
| 7517 |
+
"step": 1068
|
| 7518 |
+
},
|
| 7519 |
+
{
|
| 7520 |
+
"epoch": 0.6082503556187767,
|
| 7521 |
+
"grad_norm": 0.5597994327545166,
|
| 7522 |
+
"learning_rate": 2.5766696440565496e-05,
|
| 7523 |
+
"loss": 1.4621,
|
| 7524 |
+
"step": 1069
|
| 7525 |
+
},
|
| 7526 |
+
{
|
| 7527 |
+
"epoch": 0.6088193456614509,
|
| 7528 |
+
"grad_norm": 0.5746705532073975,
|
| 7529 |
+
"learning_rate": 2.5614575438750522e-05,
|
| 7530 |
+
"loss": 1.1686,
|
| 7531 |
+
"step": 1070
|
| 7532 |
+
},
|
| 7533 |
+
{
|
| 7534 |
+
"epoch": 0.6093883357041252,
|
| 7535 |
+
"grad_norm": 0.6218065023422241,
|
| 7536 |
+
"learning_rate": 2.546283882626065e-05,
|
| 7537 |
+
"loss": 1.4789,
|
| 7538 |
+
"step": 1071
|
| 7539 |
+
},
|
| 7540 |
+
{
|
| 7541 |
+
"epoch": 0.6099573257467994,
|
| 7542 |
+
"grad_norm": 0.6003706455230713,
|
| 7543 |
+
"learning_rate": 2.5311487387200306e-05,
|
| 7544 |
+
"loss": 1.3938,
|
| 7545 |
+
"step": 1072
|
| 7546 |
+
},
|
| 7547 |
+
{
|
| 7548 |
+
"epoch": 0.6105263157894737,
|
| 7549 |
+
"grad_norm": 0.6234976649284363,
|
| 7550 |
+
"learning_rate": 2.516052190368341e-05,
|
| 7551 |
+
"loss": 1.4399,
|
| 7552 |
+
"step": 1073
|
| 7553 |
+
},
|
| 7554 |
+
{
|
| 7555 |
+
"epoch": 0.6110953058321479,
|
| 7556 |
+
"grad_norm": 0.5926810503005981,
|
| 7557 |
+
"learning_rate": 2.500994315582943e-05,
|
| 7558 |
+
"loss": 1.3032,
|
| 7559 |
+
"step": 1074
|
| 7560 |
+
},
|
| 7561 |
+
{
|
| 7562 |
+
"epoch": 0.6116642958748222,
|
| 7563 |
+
"grad_norm": 0.5906057953834534,
|
| 7564 |
+
"learning_rate": 2.485975192175949e-05,
|
| 7565 |
+
"loss": 1.2748,
|
| 7566 |
+
"step": 1075
|
| 7567 |
+
},
|
| 7568 |
+
{
|
| 7569 |
+
"epoch": 0.6122332859174965,
|
| 7570 |
+
"grad_norm": 0.5761781334877014,
|
| 7571 |
+
"learning_rate": 2.4709948977592034e-05,
|
| 7572 |
+
"loss": 1.4486,
|
| 7573 |
+
"step": 1076
|
| 7574 |
+
},
|
| 7575 |
+
{
|
| 7576 |
+
"epoch": 0.6128022759601707,
|
| 7577 |
+
"grad_norm": 0.6170504093170166,
|
| 7578 |
+
"learning_rate": 2.4560535097439108e-05,
|
| 7579 |
+
"loss": 1.5943,
|
| 7580 |
+
"step": 1077
|
| 7581 |
+
},
|
| 7582 |
+
{
|
| 7583 |
+
"epoch": 0.613371266002845,
|
| 7584 |
+
"grad_norm": 0.6018140912055969,
|
| 7585 |
+
"learning_rate": 2.4411511053402302e-05,
|
| 7586 |
+
"loss": 1.5996,
|
| 7587 |
+
"step": 1078
|
| 7588 |
+
},
|
| 7589 |
+
{
|
| 7590 |
+
"epoch": 0.6139402560455192,
|
| 7591 |
+
"grad_norm": 0.5538153052330017,
|
| 7592 |
+
"learning_rate": 2.4262877615568626e-05,
|
| 7593 |
+
"loss": 1.4874,
|
| 7594 |
+
"step": 1079
|
| 7595 |
+
},
|
| 7596 |
+
{
|
| 7597 |
+
"epoch": 0.6145092460881935,
|
| 7598 |
+
"grad_norm": 0.5843609571456909,
|
| 7599 |
+
"learning_rate": 2.411463555200667e-05,
|
| 7600 |
+
"loss": 1.269,
|
| 7601 |
+
"step": 1080
|
| 7602 |
+
},
|
| 7603 |
+
{
|
| 7604 |
+
"epoch": 0.6150782361308678,
|
| 7605 |
+
"grad_norm": 0.5559793710708618,
|
| 7606 |
+
"learning_rate": 2.3966785628762546e-05,
|
| 7607 |
+
"loss": 1.5796,
|
| 7608 |
+
"step": 1081
|
| 7609 |
+
},
|
| 7610 |
+
{
|
| 7611 |
+
"epoch": 0.615647226173542,
|
| 7612 |
+
"grad_norm": 0.5636264085769653,
|
| 7613 |
+
"learning_rate": 2.381932860985596e-05,
|
| 7614 |
+
"loss": 1.2805,
|
| 7615 |
+
"step": 1082
|
| 7616 |
+
},
|
| 7617 |
+
{
|
| 7618 |
+
"epoch": 0.6162162162162163,
|
| 7619 |
+
"grad_norm": 0.6079363226890564,
|
| 7620 |
+
"learning_rate": 2.3672265257276383e-05,
|
| 7621 |
+
"loss": 1.5295,
|
| 7622 |
+
"step": 1083
|
| 7623 |
+
},
|
| 7624 |
+
{
|
| 7625 |
+
"epoch": 0.6167852062588904,
|
| 7626 |
+
"grad_norm": 0.6165335178375244,
|
| 7627 |
+
"learning_rate": 2.352559633097885e-05,
|
| 7628 |
+
"loss": 1.5551,
|
| 7629 |
+
"step": 1084
|
| 7630 |
+
},
|
| 7631 |
+
{
|
| 7632 |
+
"epoch": 0.6173541963015647,
|
| 7633 |
+
"grad_norm": 0.6137623190879822,
|
| 7634 |
+
"learning_rate": 2.337932258888028e-05,
|
| 7635 |
+
"loss": 1.4585,
|
| 7636 |
+
"step": 1085
|
| 7637 |
+
},
|
| 7638 |
+
{
|
| 7639 |
+
"epoch": 0.6179231863442389,
|
| 7640 |
+
"grad_norm": 0.5806836485862732,
|
| 7641 |
+
"learning_rate": 2.3233444786855407e-05,
|
| 7642 |
+
"loss": 1.5539,
|
| 7643 |
+
"step": 1086
|
| 7644 |
+
},
|
| 7645 |
+
{
|
| 7646 |
+
"epoch": 0.6184921763869132,
|
| 7647 |
+
"grad_norm": 0.5768011212348938,
|
| 7648 |
+
"learning_rate": 2.308796367873296e-05,
|
| 7649 |
+
"loss": 1.4415,
|
| 7650 |
+
"step": 1087
|
| 7651 |
+
},
|
| 7652 |
+
{
|
| 7653 |
+
"epoch": 0.6190611664295875,
|
| 7654 |
+
"grad_norm": 0.5644312500953674,
|
| 7655 |
+
"learning_rate": 2.294288001629177e-05,
|
| 7656 |
+
"loss": 1.4668,
|
| 7657 |
+
"step": 1088
|
| 7658 |
+
},
|
| 7659 |
+
{
|
| 7660 |
+
"epoch": 0.6196301564722617,
|
| 7661 |
+
"grad_norm": 0.5748885869979858,
|
| 7662 |
+
"learning_rate": 2.2798194549256792e-05,
|
| 7663 |
+
"loss": 1.3066,
|
| 7664 |
+
"step": 1089
|
| 7665 |
+
},
|
| 7666 |
+
{
|
| 7667 |
+
"epoch": 0.620199146514936,
|
| 7668 |
+
"grad_norm": 0.5609626770019531,
|
| 7669 |
+
"learning_rate": 2.2653908025295323e-05,
|
| 7670 |
+
"loss": 1.3779,
|
| 7671 |
+
"step": 1090
|
| 7672 |
+
},
|
| 7673 |
+
{
|
| 7674 |
+
"epoch": 0.6207681365576102,
|
| 7675 |
+
"grad_norm": 0.5691306591033936,
|
| 7676 |
+
"learning_rate": 2.251002119001312e-05,
|
| 7677 |
+
"loss": 1.442,
|
| 7678 |
+
"step": 1091
|
| 7679 |
+
},
|
| 7680 |
+
{
|
| 7681 |
+
"epoch": 0.6213371266002845,
|
| 7682 |
+
"grad_norm": 0.589314877986908,
|
| 7683 |
+
"learning_rate": 2.2366534786950467e-05,
|
| 7684 |
+
"loss": 1.4482,
|
| 7685 |
+
"step": 1092
|
| 7686 |
+
},
|
| 7687 |
+
{
|
| 7688 |
+
"epoch": 0.6219061166429587,
|
| 7689 |
+
"grad_norm": 0.5820268392562866,
|
| 7690 |
+
"learning_rate": 2.222344955757851e-05,
|
| 7691 |
+
"loss": 1.4195,
|
| 7692 |
+
"step": 1093
|
| 7693 |
+
},
|
| 7694 |
+
{
|
| 7695 |
+
"epoch": 0.622475106685633,
|
| 7696 |
+
"grad_norm": 0.6211294531822205,
|
| 7697 |
+
"learning_rate": 2.2080766241295235e-05,
|
| 7698 |
+
"loss": 1.549,
|
| 7699 |
+
"step": 1094
|
| 7700 |
+
},
|
| 7701 |
+
{
|
| 7702 |
+
"epoch": 0.6230440967283073,
|
| 7703 |
+
"grad_norm": 0.6313804984092712,
|
| 7704 |
+
"learning_rate": 2.1938485575421752e-05,
|
| 7705 |
+
"loss": 1.6662,
|
| 7706 |
+
"step": 1095
|
| 7707 |
+
},
|
| 7708 |
+
{
|
| 7709 |
+
"epoch": 0.6236130867709815,
|
| 7710 |
+
"grad_norm": 0.5776501297950745,
|
| 7711 |
+
"learning_rate": 2.1796608295198462e-05,
|
| 7712 |
+
"loss": 1.3551,
|
| 7713 |
+
"step": 1096
|
| 7714 |
+
},
|
| 7715 |
+
{
|
| 7716 |
+
"epoch": 0.6241820768136558,
|
| 7717 |
+
"grad_norm": 0.5959988236427307,
|
| 7718 |
+
"learning_rate": 2.165513513378121e-05,
|
| 7719 |
+
"loss": 1.4321,
|
| 7720 |
+
"step": 1097
|
| 7721 |
+
},
|
| 7722 |
+
{
|
| 7723 |
+
"epoch": 0.62475106685633,
|
| 7724 |
+
"grad_norm": 0.5878854393959045,
|
| 7725 |
+
"learning_rate": 2.1514066822237665e-05,
|
| 7726 |
+
"loss": 1.428,
|
| 7727 |
+
"step": 1098
|
| 7728 |
+
},
|
| 7729 |
+
{
|
| 7730 |
+
"epoch": 0.6253200568990043,
|
| 7731 |
+
"grad_norm": 0.5653113722801208,
|
| 7732 |
+
"learning_rate": 2.137340408954329e-05,
|
| 7733 |
+
"loss": 1.3464,
|
| 7734 |
+
"step": 1099
|
| 7735 |
+
},
|
| 7736 |
+
{
|
| 7737 |
+
"epoch": 0.6258890469416786,
|
| 7738 |
+
"grad_norm": 0.5969840884208679,
|
| 7739 |
+
"learning_rate": 2.1233147662577767e-05,
|
| 7740 |
+
"loss": 1.4497,
|
| 7741 |
+
"step": 1100
|
| 7742 |
+
},
|
| 7743 |
+
{
|
| 7744 |
+
"epoch": 0.6264580369843528,
|
| 7745 |
+
"grad_norm": 0.5675022602081299,
|
| 7746 |
+
"learning_rate": 2.1093298266121165e-05,
|
| 7747 |
+
"loss": 1.4289,
|
| 7748 |
+
"step": 1101
|
| 7749 |
+
},
|
| 7750 |
+
{
|
| 7751 |
+
"epoch": 0.6270270270270271,
|
| 7752 |
+
"grad_norm": 0.6396809816360474,
|
| 7753 |
+
"learning_rate": 2.0953856622850176e-05,
|
| 7754 |
+
"loss": 1.4908,
|
| 7755 |
+
"step": 1102
|
| 7756 |
+
},
|
| 7757 |
+
{
|
| 7758 |
+
"epoch": 0.6275960170697013,
|
| 7759 |
+
"grad_norm": 0.5843429565429688,
|
| 7760 |
+
"learning_rate": 2.081482345333452e-05,
|
| 7761 |
+
"loss": 1.6213,
|
| 7762 |
+
"step": 1103
|
| 7763 |
+
},
|
| 7764 |
+
{
|
| 7765 |
+
"epoch": 0.6281650071123756,
|
| 7766 |
+
"grad_norm": 0.5792785882949829,
|
| 7767 |
+
"learning_rate": 2.0676199476033e-05,
|
| 7768 |
+
"loss": 1.57,
|
| 7769 |
+
"step": 1104
|
| 7770 |
+
},
|
| 7771 |
+
{
|
| 7772 |
+
"epoch": 0.6287339971550497,
|
| 7773 |
+
"grad_norm": 0.6015857458114624,
|
| 7774 |
+
"learning_rate": 2.053798540728995e-05,
|
| 7775 |
+
"loss": 1.5818,
|
| 7776 |
+
"step": 1105
|
| 7777 |
+
},
|
| 7778 |
+
{
|
| 7779 |
+
"epoch": 0.629302987197724,
|
| 7780 |
+
"grad_norm": 0.5723267197608948,
|
| 7781 |
+
"learning_rate": 2.0400181961331478e-05,
|
| 7782 |
+
"loss": 1.3799,
|
| 7783 |
+
"step": 1106
|
| 7784 |
+
},
|
| 7785 |
+
{
|
| 7786 |
+
"epoch": 0.6298719772403983,
|
| 7787 |
+
"grad_norm": 0.6322827339172363,
|
| 7788 |
+
"learning_rate": 2.0262789850261798e-05,
|
| 7789 |
+
"loss": 1.4456,
|
| 7790 |
+
"step": 1107
|
| 7791 |
+
},
|
| 7792 |
+
{
|
| 7793 |
+
"epoch": 0.6304409672830725,
|
| 7794 |
+
"grad_norm": 0.6475574970245361,
|
| 7795 |
+
"learning_rate": 2.012580978405949e-05,
|
| 7796 |
+
"loss": 1.6081,
|
| 7797 |
+
"step": 1108
|
| 7798 |
+
},
|
| 7799 |
+
{
|
| 7800 |
+
"epoch": 0.6310099573257468,
|
| 7801 |
+
"grad_norm": 0.5577263832092285,
|
| 7802 |
+
"learning_rate": 1.9989242470573975e-05,
|
| 7803 |
+
"loss": 1.319,
|
| 7804 |
+
"step": 1109
|
| 7805 |
+
},
|
| 7806 |
+
{
|
| 7807 |
+
"epoch": 0.631578947368421,
|
| 7808 |
+
"grad_norm": 0.5767825245857239,
|
| 7809 |
+
"learning_rate": 1.9853088615521663e-05,
|
| 7810 |
+
"loss": 1.1708,
|
| 7811 |
+
"step": 1110
|
| 7812 |
+
},
|
| 7813 |
+
{
|
| 7814 |
+
"epoch": 0.6321479374110953,
|
| 7815 |
+
"grad_norm": 0.5954525470733643,
|
| 7816 |
+
"learning_rate": 1.9717348922482458e-05,
|
| 7817 |
+
"loss": 1.3891,
|
| 7818 |
+
"step": 1111
|
| 7819 |
+
},
|
| 7820 |
+
{
|
| 7821 |
+
"epoch": 0.6327169274537695,
|
| 7822 |
+
"grad_norm": 0.6102060079574585,
|
| 7823 |
+
"learning_rate": 1.9582024092896033e-05,
|
| 7824 |
+
"loss": 1.3531,
|
| 7825 |
+
"step": 1112
|
| 7826 |
+
},
|
| 7827 |
+
{
|
| 7828 |
+
"epoch": 0.6332859174964438,
|
| 7829 |
+
"grad_norm": 0.5929975509643555,
|
| 7830 |
+
"learning_rate": 1.9447114826058233e-05,
|
| 7831 |
+
"loss": 1.5927,
|
| 7832 |
+
"step": 1113
|
| 7833 |
+
},
|
| 7834 |
+
{
|
| 7835 |
+
"epoch": 0.6338549075391181,
|
| 7836 |
+
"grad_norm": 0.5874947905540466,
|
| 7837 |
+
"learning_rate": 1.931262181911754e-05,
|
| 7838 |
+
"loss": 1.4828,
|
| 7839 |
+
"step": 1114
|
| 7840 |
+
},
|
| 7841 |
+
{
|
| 7842 |
+
"epoch": 0.6344238975817923,
|
| 7843 |
+
"grad_norm": 0.5605891346931458,
|
| 7844 |
+
"learning_rate": 1.9178545767071322e-05,
|
| 7845 |
+
"loss": 1.655,
|
| 7846 |
+
"step": 1115
|
| 7847 |
+
},
|
| 7848 |
+
{
|
| 7849 |
+
"epoch": 0.6349928876244666,
|
| 7850 |
+
"grad_norm": 0.5770267248153687,
|
| 7851 |
+
"learning_rate": 1.9044887362762343e-05,
|
| 7852 |
+
"loss": 1.3424,
|
| 7853 |
+
"step": 1116
|
| 7854 |
+
},
|
| 7855 |
+
{
|
| 7856 |
+
"epoch": 0.6355618776671408,
|
| 7857 |
+
"grad_norm": 0.639620840549469,
|
| 7858 |
+
"learning_rate": 1.8911647296875147e-05,
|
| 7859 |
+
"loss": 1.3701,
|
| 7860 |
+
"step": 1117
|
| 7861 |
+
},
|
| 7862 |
+
{
|
| 7863 |
+
"epoch": 0.6361308677098151,
|
| 7864 |
+
"grad_norm": 0.6051335334777832,
|
| 7865 |
+
"learning_rate": 1.87788262579325e-05,
|
| 7866 |
+
"loss": 1.3458,
|
| 7867 |
+
"step": 1118
|
| 7868 |
+
},
|
| 7869 |
+
{
|
| 7870 |
+
"epoch": 0.6366998577524894,
|
| 7871 |
+
"grad_norm": 0.5964054465293884,
|
| 7872 |
+
"learning_rate": 1.8646424932291896e-05,
|
| 7873 |
+
"loss": 1.5532,
|
| 7874 |
+
"step": 1119
|
| 7875 |
+
},
|
| 7876 |
+
{
|
| 7877 |
+
"epoch": 0.6372688477951636,
|
| 7878 |
+
"grad_norm": 0.6007367968559265,
|
| 7879 |
+
"learning_rate": 1.851444400414185e-05,
|
| 7880 |
+
"loss": 1.5373,
|
| 7881 |
+
"step": 1120
|
| 7882 |
+
},
|
| 7883 |
+
{
|
| 7884 |
+
"epoch": 0.6378378378378379,
|
| 7885 |
+
"grad_norm": 0.5970923900604248,
|
| 7886 |
+
"learning_rate": 1.8382884155498514e-05,
|
| 7887 |
+
"loss": 1.5256,
|
| 7888 |
+
"step": 1121
|
| 7889 |
+
},
|
| 7890 |
+
{
|
| 7891 |
+
"epoch": 0.6384068278805121,
|
| 7892 |
+
"grad_norm": 0.6180837750434875,
|
| 7893 |
+
"learning_rate": 1.8251746066202058e-05,
|
| 7894 |
+
"loss": 1.4781,
|
| 7895 |
+
"step": 1122
|
| 7896 |
+
},
|
| 7897 |
+
{
|
| 7898 |
+
"epoch": 0.6389758179231864,
|
| 7899 |
+
"grad_norm": 0.6046369671821594,
|
| 7900 |
+
"learning_rate": 1.812103041391322e-05,
|
| 7901 |
+
"loss": 1.4899,
|
| 7902 |
+
"step": 1123
|
| 7903 |
+
},
|
| 7904 |
+
{
|
| 7905 |
+
"epoch": 0.6395448079658606,
|
| 7906 |
+
"grad_norm": 0.5703504085540771,
|
| 7907 |
+
"learning_rate": 1.799073787410982e-05,
|
| 7908 |
+
"loss": 1.5633,
|
| 7909 |
+
"step": 1124
|
| 7910 |
+
},
|
| 7911 |
+
{
|
| 7912 |
+
"epoch": 0.6401137980085349,
|
| 7913 |
+
"grad_norm": 0.6019449830055237,
|
| 7914 |
+
"learning_rate": 1.786086912008316e-05,
|
| 7915 |
+
"loss": 1.3685,
|
| 7916 |
+
"step": 1125
|
| 7917 |
+
},
|
| 7918 |
+
{
|
| 7919 |
+
"epoch": 0.6406827880512092,
|
| 7920 |
+
"grad_norm": 0.5852835774421692,
|
| 7921 |
+
"learning_rate": 1.773142482293464e-05,
|
| 7922 |
+
"loss": 1.5065,
|
| 7923 |
+
"step": 1126
|
| 7924 |
+
},
|
| 7925 |
+
{
|
| 7926 |
+
"epoch": 0.6412517780938833,
|
| 7927 |
+
"grad_norm": 0.5664365887641907,
|
| 7928 |
+
"learning_rate": 1.7602405651572275e-05,
|
| 7929 |
+
"loss": 1.5823,
|
| 7930 |
+
"step": 1127
|
| 7931 |
+
},
|
| 7932 |
+
{
|
| 7933 |
+
"epoch": 0.6418207681365576,
|
| 7934 |
+
"grad_norm": 0.5778409242630005,
|
| 7935 |
+
"learning_rate": 1.747381227270718e-05,
|
| 7936 |
+
"loss": 1.4294,
|
| 7937 |
+
"step": 1128
|
| 7938 |
+
},
|
| 7939 |
+
{
|
| 7940 |
+
"epoch": 0.6423897581792318,
|
| 7941 |
+
"grad_norm": 0.5901049375534058,
|
| 7942 |
+
"learning_rate": 1.734564535085028e-05,
|
| 7943 |
+
"loss": 1.3996,
|
| 7944 |
+
"step": 1129
|
| 7945 |
+
},
|
| 7946 |
+
{
|
| 7947 |
+
"epoch": 0.6429587482219061,
|
| 7948 |
+
"grad_norm": 0.6099653244018555,
|
| 7949 |
+
"learning_rate": 1.721790554830869e-05,
|
| 7950 |
+
"loss": 1.5873,
|
| 7951 |
+
"step": 1130
|
| 7952 |
+
},
|
| 7953 |
+
{
|
| 7954 |
+
"epoch": 0.6435277382645803,
|
| 7955 |
+
"grad_norm": 0.5981472730636597,
|
| 7956 |
+
"learning_rate": 1.7090593525182287e-05,
|
| 7957 |
+
"loss": 1.5958,
|
| 7958 |
+
"step": 1131
|
| 7959 |
+
},
|
| 7960 |
+
{
|
| 7961 |
+
"epoch": 0.6440967283072546,
|
| 7962 |
+
"grad_norm": 0.6043581366539001,
|
| 7963 |
+
"learning_rate": 1.6963709939360585e-05,
|
| 7964 |
+
"loss": 1.561,
|
| 7965 |
+
"step": 1132
|
| 7966 |
+
},
|
| 7967 |
+
{
|
| 7968 |
+
"epoch": 0.6446657183499289,
|
| 7969 |
+
"grad_norm": 0.6230269074440002,
|
| 7970 |
+
"learning_rate": 1.6837255446518964e-05,
|
| 7971 |
+
"loss": 1.4484,
|
| 7972 |
+
"step": 1133
|
| 7973 |
+
},
|
| 7974 |
+
{
|
| 7975 |
+
"epoch": 0.6452347083926031,
|
| 7976 |
+
"grad_norm": 0.579458475112915,
|
| 7977 |
+
"learning_rate": 1.671123070011551e-05,
|
| 7978 |
+
"loss": 1.597,
|
| 7979 |
+
"step": 1134
|
| 7980 |
+
},
|
| 7981 |
+
{
|
| 7982 |
+
"epoch": 0.6458036984352774,
|
| 7983 |
+
"grad_norm": 0.5982540845870972,
|
| 7984 |
+
"learning_rate": 1.6585636351387635e-05,
|
| 7985 |
+
"loss": 1.5299,
|
| 7986 |
+
"step": 1135
|
| 7987 |
+
},
|
| 7988 |
+
{
|
| 7989 |
+
"epoch": 0.6463726884779516,
|
| 7990 |
+
"grad_norm": 0.6335290670394897,
|
| 7991 |
+
"learning_rate": 1.646047304934851e-05,
|
| 7992 |
+
"loss": 1.6529,
|
| 7993 |
+
"step": 1136
|
| 7994 |
+
},
|
| 7995 |
+
{
|
| 7996 |
+
"epoch": 0.6469416785206259,
|
| 7997 |
+
"grad_norm": 0.580467164516449,
|
| 7998 |
+
"learning_rate": 1.6335741440784035e-05,
|
| 7999 |
+
"loss": 1.5459,
|
| 8000 |
+
"step": 1137
|
| 8001 |
+
},
|
| 8002 |
+
{
|
| 8003 |
+
"epoch": 0.6475106685633002,
|
| 8004 |
+
"grad_norm": 0.5840801000595093,
|
| 8005 |
+
"learning_rate": 1.621144217024918e-05,
|
| 8006 |
+
"loss": 1.3808,
|
| 8007 |
+
"step": 1138
|
| 8008 |
+
},
|
| 8009 |
+
{
|
| 8010 |
+
"epoch": 0.6480796586059744,
|
| 8011 |
+
"grad_norm": 0.592555582523346,
|
| 8012 |
+
"learning_rate": 1.608757588006483e-05,
|
| 8013 |
+
"loss": 1.5013,
|
| 8014 |
+
"step": 1139
|
| 8015 |
+
},
|
| 8016 |
+
{
|
| 8017 |
+
"epoch": 0.6486486486486487,
|
| 8018 |
+
"grad_norm": 0.5938240885734558,
|
| 8019 |
+
"learning_rate": 1.596414321031452e-05,
|
| 8020 |
+
"loss": 1.3971,
|
| 8021 |
+
"step": 1140
|
| 8022 |
+
},
|
| 8023 |
+
{
|
| 8024 |
+
"epoch": 0.6492176386913229,
|
| 8025 |
+
"grad_norm": 0.5719125270843506,
|
| 8026 |
+
"learning_rate": 1.5841144798840855e-05,
|
| 8027 |
+
"loss": 1.4372,
|
| 8028 |
+
"step": 1141
|
| 8029 |
+
},
|
| 8030 |
+
{
|
| 8031 |
+
"epoch": 0.6497866287339972,
|
| 8032 |
+
"grad_norm": 0.6199617981910706,
|
| 8033 |
+
"learning_rate": 1.5718581281242572e-05,
|
| 8034 |
+
"loss": 1.6019,
|
| 8035 |
+
"step": 1142
|
| 8036 |
+
},
|
| 8037 |
+
{
|
| 8038 |
+
"epoch": 0.6503556187766714,
|
| 8039 |
+
"grad_norm": 0.594205379486084,
|
| 8040 |
+
"learning_rate": 1.5596453290870982e-05,
|
| 8041 |
+
"loss": 1.5322,
|
| 8042 |
+
"step": 1143
|
| 8043 |
+
},
|
| 8044 |
+
{
|
| 8045 |
+
"epoch": 0.6509246088193457,
|
| 8046 |
+
"grad_norm": 0.6198005676269531,
|
| 8047 |
+
"learning_rate": 1.5474761458826793e-05,
|
| 8048 |
+
"loss": 1.4777,
|
| 8049 |
+
"step": 1144
|
| 8050 |
+
},
|
| 8051 |
+
{
|
| 8052 |
+
"epoch": 0.65149359886202,
|
| 8053 |
+
"grad_norm": 0.567058265209198,
|
| 8054 |
+
"learning_rate": 1.5353506413956932e-05,
|
| 8055 |
+
"loss": 1.5108,
|
| 8056 |
+
"step": 1145
|
| 8057 |
+
},
|
| 8058 |
+
{
|
| 8059 |
+
"epoch": 0.6520625889046942,
|
| 8060 |
+
"grad_norm": 0.5950532555580139,
|
| 8061 |
+
"learning_rate": 1.5232688782851068e-05,
|
| 8062 |
+
"loss": 1.5038,
|
| 8063 |
+
"step": 1146
|
| 8064 |
+
},
|
| 8065 |
+
{
|
| 8066 |
+
"epoch": 0.6526315789473685,
|
| 8067 |
+
"grad_norm": 0.6178238987922668,
|
| 8068 |
+
"learning_rate": 1.511230918983867e-05,
|
| 8069 |
+
"loss": 1.5458,
|
| 8070 |
+
"step": 1147
|
| 8071 |
+
},
|
| 8072 |
+
{
|
| 8073 |
+
"epoch": 0.6532005689900426,
|
| 8074 |
+
"grad_norm": 0.5962685346603394,
|
| 8075 |
+
"learning_rate": 1.4992368256985546e-05,
|
| 8076 |
+
"loss": 1.432,
|
| 8077 |
+
"step": 1148
|
| 8078 |
+
},
|
| 8079 |
+
{
|
| 8080 |
+
"epoch": 0.6537695590327169,
|
| 8081 |
+
"grad_norm": 0.5979620218276978,
|
| 8082 |
+
"learning_rate": 1.4872866604090696e-05,
|
| 8083 |
+
"loss": 1.5035,
|
| 8084 |
+
"step": 1149
|
| 8085 |
+
},
|
| 8086 |
+
{
|
| 8087 |
+
"epoch": 0.6543385490753911,
|
| 8088 |
+
"grad_norm": 0.5819264650344849,
|
| 8089 |
+
"learning_rate": 1.475380484868325e-05,
|
| 8090 |
+
"loss": 1.4169,
|
| 8091 |
+
"step": 1150
|
| 8092 |
+
},
|
| 8093 |
+
{
|
| 8094 |
+
"epoch": 0.6549075391180654,
|
| 8095 |
+
"grad_norm": 0.6469606757164001,
|
| 8096 |
+
"learning_rate": 1.4635183606018943e-05,
|
| 8097 |
+
"loss": 1.4442,
|
| 8098 |
+
"step": 1151
|
| 8099 |
+
},
|
| 8100 |
+
{
|
| 8101 |
+
"epoch": 0.6554765291607397,
|
| 8102 |
+
"grad_norm": 0.5789610147476196,
|
| 8103 |
+
"learning_rate": 1.451700348907734e-05,
|
| 8104 |
+
"loss": 1.122,
|
| 8105 |
+
"step": 1152
|
| 8106 |
+
},
|
| 8107 |
+
{
|
| 8108 |
+
"epoch": 0.6560455192034139,
|
| 8109 |
+
"grad_norm": 0.5797150135040283,
|
| 8110 |
+
"learning_rate": 1.4399265108558379e-05,
|
| 8111 |
+
"loss": 1.5795,
|
| 8112 |
+
"step": 1153
|
| 8113 |
+
},
|
| 8114 |
+
{
|
| 8115 |
+
"epoch": 0.6566145092460882,
|
| 8116 |
+
"grad_norm": 0.6014512777328491,
|
| 8117 |
+
"learning_rate": 1.4281969072879298e-05,
|
| 8118 |
+
"loss": 1.3673,
|
| 8119 |
+
"step": 1154
|
| 8120 |
+
},
|
| 8121 |
+
{
|
| 8122 |
+
"epoch": 0.6571834992887624,
|
| 8123 |
+
"grad_norm": 0.566061794757843,
|
| 8124 |
+
"learning_rate": 1.4165115988171596e-05,
|
| 8125 |
+
"loss": 1.4255,
|
| 8126 |
+
"step": 1155
|
| 8127 |
+
},
|
| 8128 |
+
{
|
| 8129 |
+
"epoch": 0.6577524893314367,
|
| 8130 |
+
"grad_norm": 0.599322497844696,
|
| 8131 |
+
"learning_rate": 1.4048706458277672e-05,
|
| 8132 |
+
"loss": 1.4538,
|
| 8133 |
+
"step": 1156
|
| 8134 |
+
},
|
| 8135 |
+
{
|
| 8136 |
+
"epoch": 0.658321479374111,
|
| 8137 |
+
"grad_norm": 0.6258883476257324,
|
| 8138 |
+
"learning_rate": 1.3932741084747913e-05,
|
| 8139 |
+
"loss": 1.5197,
|
| 8140 |
+
"step": 1157
|
| 8141 |
+
},
|
| 8142 |
+
{
|
| 8143 |
+
"epoch": 0.6588904694167852,
|
| 8144 |
+
"grad_norm": 0.603754460811615,
|
| 8145 |
+
"learning_rate": 1.3817220466837566e-05,
|
| 8146 |
+
"loss": 1.5596,
|
| 8147 |
+
"step": 1158
|
| 8148 |
+
},
|
| 8149 |
+
{
|
| 8150 |
+
"epoch": 0.6594594594594595,
|
| 8151 |
+
"grad_norm": 0.5680553317070007,
|
| 8152 |
+
"learning_rate": 1.3702145201503458e-05,
|
| 8153 |
+
"loss": 1.3882,
|
| 8154 |
+
"step": 1159
|
| 8155 |
+
},
|
| 8156 |
+
{
|
| 8157 |
+
"epoch": 0.6600284495021337,
|
| 8158 |
+
"grad_norm": 0.6317921280860901,
|
| 8159 |
+
"learning_rate": 1.3587515883401202e-05,
|
| 8160 |
+
"loss": 1.4051,
|
| 8161 |
+
"step": 1160
|
| 8162 |
+
},
|
| 8163 |
+
{
|
| 8164 |
+
"epoch": 0.660597439544808,
|
| 8165 |
+
"grad_norm": 0.5998348593711853,
|
| 8166 |
+
"learning_rate": 1.3473333104881792e-05,
|
| 8167 |
+
"loss": 1.5309,
|
| 8168 |
+
"step": 1161
|
| 8169 |
+
},
|
| 8170 |
+
{
|
| 8171 |
+
"epoch": 0.6611664295874822,
|
| 8172 |
+
"grad_norm": 0.5694242715835571,
|
| 8173 |
+
"learning_rate": 1.3359597455988803e-05,
|
| 8174 |
+
"loss": 1.4933,
|
| 8175 |
+
"step": 1162
|
| 8176 |
+
},
|
| 8177 |
+
{
|
| 8178 |
+
"epoch": 0.6617354196301565,
|
| 8179 |
+
"grad_norm": 0.6193349361419678,
|
| 8180 |
+
"learning_rate": 1.3246309524455291e-05,
|
| 8181 |
+
"loss": 1.5781,
|
| 8182 |
+
"step": 1163
|
| 8183 |
+
},
|
| 8184 |
+
{
|
| 8185 |
+
"epoch": 0.6623044096728308,
|
| 8186 |
+
"grad_norm": 0.5579991340637207,
|
| 8187 |
+
"learning_rate": 1.3133469895700634e-05,
|
| 8188 |
+
"loss": 1.3616,
|
| 8189 |
+
"step": 1164
|
| 8190 |
+
},
|
| 8191 |
+
{
|
| 8192 |
+
"epoch": 0.662873399715505,
|
| 8193 |
+
"grad_norm": 0.5790702104568481,
|
| 8194 |
+
"learning_rate": 1.3021079152827631e-05,
|
| 8195 |
+
"loss": 1.3994,
|
| 8196 |
+
"step": 1165
|
| 8197 |
+
},
|
| 8198 |
+
{
|
| 8199 |
+
"epoch": 0.6634423897581793,
|
| 8200 |
+
"grad_norm": 0.5730209946632385,
|
| 8201 |
+
"learning_rate": 1.2909137876619448e-05,
|
| 8202 |
+
"loss": 1.3269,
|
| 8203 |
+
"step": 1166
|
| 8204 |
+
},
|
| 8205 |
+
{
|
| 8206 |
+
"epoch": 0.6640113798008535,
|
| 8207 |
+
"grad_norm": 0.6066268086433411,
|
| 8208 |
+
"learning_rate": 1.2797646645536566e-05,
|
| 8209 |
+
"loss": 1.6239,
|
| 8210 |
+
"step": 1167
|
| 8211 |
+
},
|
| 8212 |
+
{
|
| 8213 |
+
"epoch": 0.6645803698435278,
|
| 8214 |
+
"grad_norm": 0.649182140827179,
|
| 8215 |
+
"learning_rate": 1.2686606035713944e-05,
|
| 8216 |
+
"loss": 1.7304,
|
| 8217 |
+
"step": 1168
|
| 8218 |
+
},
|
| 8219 |
+
{
|
| 8220 |
+
"epoch": 0.6651493598862019,
|
| 8221 |
+
"grad_norm": 0.6383649110794067,
|
| 8222 |
+
"learning_rate": 1.2576016620957853e-05,
|
| 8223 |
+
"loss": 1.4477,
|
| 8224 |
+
"step": 1169
|
| 8225 |
+
},
|
| 8226 |
+
{
|
| 8227 |
+
"epoch": 0.6657183499288762,
|
| 8228 |
+
"grad_norm": 0.5763673782348633,
|
| 8229 |
+
"learning_rate": 1.2465878972743028e-05,
|
| 8230 |
+
"loss": 1.4846,
|
| 8231 |
+
"step": 1170
|
| 8232 |
+
},
|
| 8233 |
+
{
|
| 8234 |
+
"epoch": 0.6662873399715505,
|
| 8235 |
+
"grad_norm": 0.5865679383277893,
|
| 8236 |
+
"learning_rate": 1.2356193660209681e-05,
|
| 8237 |
+
"loss": 1.5687,
|
| 8238 |
+
"step": 1171
|
| 8239 |
+
},
|
| 8240 |
+
{
|
| 8241 |
+
"epoch": 0.6668563300142247,
|
| 8242 |
+
"grad_norm": 0.5898412466049194,
|
| 8243 |
+
"learning_rate": 1.2246961250160527e-05,
|
| 8244 |
+
"loss": 1.5227,
|
| 8245 |
+
"step": 1172
|
| 8246 |
+
},
|
| 8247 |
+
{
|
| 8248 |
+
"epoch": 0.667425320056899,
|
| 8249 |
+
"grad_norm": 0.6910015344619751,
|
| 8250 |
+
"learning_rate": 1.2138182307057987e-05,
|
| 8251 |
+
"loss": 1.245,
|
| 8252 |
+
"step": 1173
|
| 8253 |
+
},
|
| 8254 |
+
{
|
| 8255 |
+
"epoch": 0.6679943100995732,
|
| 8256 |
+
"grad_norm": 0.5660498142242432,
|
| 8257 |
+
"learning_rate": 1.2029857393021094e-05,
|
| 8258 |
+
"loss": 1.2887,
|
| 8259 |
+
"step": 1174
|
| 8260 |
+
},
|
| 8261 |
+
{
|
| 8262 |
+
"epoch": 0.6685633001422475,
|
| 8263 |
+
"grad_norm": 0.5966072082519531,
|
| 8264 |
+
"learning_rate": 1.1921987067822672e-05,
|
| 8265 |
+
"loss": 1.3417,
|
| 8266 |
+
"step": 1175
|
| 8267 |
+
},
|
| 8268 |
+
{
|
| 8269 |
+
"epoch": 0.6691322901849218,
|
| 8270 |
+
"grad_norm": 0.5854772329330444,
|
| 8271 |
+
"learning_rate": 1.1814571888886483e-05,
|
| 8272 |
+
"loss": 1.474,
|
| 8273 |
+
"step": 1176
|
| 8274 |
+
},
|
| 8275 |
+
{
|
| 8276 |
+
"epoch": 0.669701280227596,
|
| 8277 |
+
"grad_norm": 0.6000230312347412,
|
| 8278 |
+
"learning_rate": 1.1707612411284253e-05,
|
| 8279 |
+
"loss": 1.4276,
|
| 8280 |
+
"step": 1177
|
| 8281 |
+
},
|
| 8282 |
+
{
|
| 8283 |
+
"epoch": 0.6702702702702703,
|
| 8284 |
+
"grad_norm": 0.5757988691329956,
|
| 8285 |
+
"learning_rate": 1.1601109187732928e-05,
|
| 8286 |
+
"loss": 1.5459,
|
| 8287 |
+
"step": 1178
|
| 8288 |
+
},
|
| 8289 |
+
{
|
| 8290 |
+
"epoch": 0.6708392603129445,
|
| 8291 |
+
"grad_norm": 0.5930068492889404,
|
| 8292 |
+
"learning_rate": 1.149506276859167e-05,
|
| 8293 |
+
"loss": 1.4149,
|
| 8294 |
+
"step": 1179
|
| 8295 |
+
},
|
| 8296 |
+
{
|
| 8297 |
+
"epoch": 0.6714082503556188,
|
| 8298 |
+
"grad_norm": 0.5741011500358582,
|
| 8299 |
+
"learning_rate": 1.1389473701859121e-05,
|
| 8300 |
+
"loss": 1.2504,
|
| 8301 |
+
"step": 1180
|
| 8302 |
+
},
|
| 8303 |
+
{
|
| 8304 |
+
"epoch": 0.671977240398293,
|
| 8305 |
+
"grad_norm": 0.588571310043335,
|
| 8306 |
+
"learning_rate": 1.1284342533170545e-05,
|
| 8307 |
+
"loss": 1.6301,
|
| 8308 |
+
"step": 1181
|
| 8309 |
+
},
|
| 8310 |
+
{
|
| 8311 |
+
"epoch": 0.6725462304409673,
|
| 8312 |
+
"grad_norm": 0.5500454306602478,
|
| 8313 |
+
"learning_rate": 1.1179669805794968e-05,
|
| 8314 |
+
"loss": 1.4952,
|
| 8315 |
+
"step": 1182
|
| 8316 |
+
},
|
| 8317 |
+
{
|
| 8318 |
+
"epoch": 0.6731152204836416,
|
| 8319 |
+
"grad_norm": 0.5811514854431152,
|
| 8320 |
+
"learning_rate": 1.1075456060632472e-05,
|
| 8321 |
+
"loss": 1.447,
|
| 8322 |
+
"step": 1183
|
| 8323 |
+
},
|
| 8324 |
+
{
|
| 8325 |
+
"epoch": 0.6736842105263158,
|
| 8326 |
+
"grad_norm": 0.6315092444419861,
|
| 8327 |
+
"learning_rate": 1.0971701836211268e-05,
|
| 8328 |
+
"loss": 1.5707,
|
| 8329 |
+
"step": 1184
|
| 8330 |
+
},
|
| 8331 |
+
{
|
| 8332 |
+
"epoch": 0.6742532005689901,
|
| 8333 |
+
"grad_norm": 0.6309195756912231,
|
| 8334 |
+
"learning_rate": 1.0868407668684998e-05,
|
| 8335 |
+
"loss": 1.2443,
|
| 8336 |
+
"step": 1185
|
| 8337 |
+
},
|
| 8338 |
+
{
|
| 8339 |
+
"epoch": 0.6748221906116643,
|
| 8340 |
+
"grad_norm": 0.5840954780578613,
|
| 8341 |
+
"learning_rate": 1.0765574091829933e-05,
|
| 8342 |
+
"loss": 1.4682,
|
| 8343 |
+
"step": 1186
|
| 8344 |
+
},
|
| 8345 |
+
{
|
| 8346 |
+
"epoch": 0.6753911806543386,
|
| 8347 |
+
"grad_norm": 0.6113706827163696,
|
| 8348 |
+
"learning_rate": 1.0663201637042252e-05,
|
| 8349 |
+
"loss": 1.4292,
|
| 8350 |
+
"step": 1187
|
| 8351 |
+
},
|
| 8352 |
+
{
|
| 8353 |
+
"epoch": 0.6759601706970128,
|
| 8354 |
+
"grad_norm": 0.6047906279563904,
|
| 8355 |
+
"learning_rate": 1.0561290833335224e-05,
|
| 8356 |
+
"loss": 1.627,
|
| 8357 |
+
"step": 1188
|
| 8358 |
+
},
|
| 8359 |
+
{
|
| 8360 |
+
"epoch": 0.676529160739687,
|
| 8361 |
+
"grad_norm": 0.5755859613418579,
|
| 8362 |
+
"learning_rate": 1.04598422073366e-05,
|
| 8363 |
+
"loss": 1.3449,
|
| 8364 |
+
"step": 1189
|
| 8365 |
+
},
|
| 8366 |
+
{
|
| 8367 |
+
"epoch": 0.6770981507823614,
|
| 8368 |
+
"grad_norm": 0.5938130021095276,
|
| 8369 |
+
"learning_rate": 1.0358856283285722e-05,
|
| 8370 |
+
"loss": 1.389,
|
| 8371 |
+
"step": 1190
|
| 8372 |
+
},
|
| 8373 |
+
{
|
| 8374 |
+
"epoch": 0.6776671408250355,
|
| 8375 |
+
"grad_norm": 0.5975162386894226,
|
| 8376 |
+
"learning_rate": 1.0258333583030955e-05,
|
| 8377 |
+
"loss": 1.4868,
|
| 8378 |
+
"step": 1191
|
| 8379 |
+
},
|
| 8380 |
+
{
|
| 8381 |
+
"epoch": 0.6782361308677098,
|
| 8382 |
+
"grad_norm": 0.6362975239753723,
|
| 8383 |
+
"learning_rate": 1.0158274626026931e-05,
|
| 8384 |
+
"loss": 1.6409,
|
| 8385 |
+
"step": 1192
|
| 8386 |
+
},
|
| 8387 |
+
{
|
| 8388 |
+
"epoch": 0.678805120910384,
|
| 8389 |
+
"grad_norm": 0.6175844669342041,
|
| 8390 |
+
"learning_rate": 1.0058679929331827e-05,
|
| 8391 |
+
"loss": 1.4914,
|
| 8392 |
+
"step": 1193
|
| 8393 |
+
},
|
| 8394 |
+
{
|
| 8395 |
+
"epoch": 0.6793741109530583,
|
| 8396 |
+
"grad_norm": 0.5870533585548401,
|
| 8397 |
+
"learning_rate": 9.959550007604835e-06,
|
| 8398 |
+
"loss": 1.3655,
|
| 8399 |
+
"step": 1194
|
| 8400 |
+
},
|
| 8401 |
+
{
|
| 8402 |
+
"epoch": 0.6799431009957326,
|
| 8403 |
+
"grad_norm": 0.5993149280548096,
|
| 8404 |
+
"learning_rate": 9.860885373103324e-06,
|
| 8405 |
+
"loss": 1.4203,
|
| 8406 |
+
"step": 1195
|
| 8407 |
+
},
|
| 8408 |
+
{
|
| 8409 |
+
"epoch": 0.6805120910384068,
|
| 8410 |
+
"grad_norm": 0.5798912048339844,
|
| 8411 |
+
"learning_rate": 9.7626865356803e-06,
|
| 8412 |
+
"loss": 1.4378,
|
| 8413 |
+
"step": 1196
|
| 8414 |
+
},
|
| 8415 |
+
{
|
| 8416 |
+
"epoch": 0.6810810810810811,
|
| 8417 |
+
"grad_norm": 0.5729113221168518,
|
| 8418 |
+
"learning_rate": 9.664954002781745e-06,
|
| 8419 |
+
"loss": 1.4054,
|
| 8420 |
+
"step": 1197
|
| 8421 |
+
},
|
| 8422 |
+
{
|
| 8423 |
+
"epoch": 0.6816500711237553,
|
| 8424 |
+
"grad_norm": 0.6329131126403809,
|
| 8425 |
+
"learning_rate": 9.567688279443964e-06,
|
| 8426 |
+
"loss": 1.4381,
|
| 8427 |
+
"step": 1198
|
| 8428 |
+
},
|
| 8429 |
+
{
|
| 8430 |
+
"epoch": 0.6822190611664296,
|
| 8431 |
+
"grad_norm": 0.6088592410087585,
|
| 8432 |
+
"learning_rate": 9.4708898682911e-06,
|
| 8433 |
+
"loss": 1.4094,
|
| 8434 |
+
"step": 1199
|
| 8435 |
+
},
|
| 8436 |
+
{
|
| 8437 |
+
"epoch": 0.6827880512091038,
|
| 8438 |
+
"grad_norm": 0.5889382362365723,
|
| 8439 |
+
"learning_rate": 9.374559269532346e-06,
|
| 8440 |
+
"loss": 1.5365,
|
| 8441 |
+
"step": 1200
|
| 8442 |
+
},
|
| 8443 |
+
{
|
| 8444 |
+
"epoch": 0.6833570412517781,
|
| 8445 |
+
"grad_norm": 0.6487043499946594,
|
| 8446 |
+
"learning_rate": 9.27869698095951e-06,
|
| 8447 |
+
"loss": 1.4747,
|
| 8448 |
+
"step": 1201
|
| 8449 |
+
},
|
| 8450 |
+
{
|
| 8451 |
+
"epoch": 0.6839260312944524,
|
| 8452 |
+
"grad_norm": 0.6006666421890259,
|
| 8453 |
+
"learning_rate": 9.183303497944361e-06,
|
| 8454 |
+
"loss": 1.3953,
|
| 8455 |
+
"step": 1202
|
| 8456 |
+
},
|
| 8457 |
+
{
|
| 8458 |
+
"epoch": 0.6844950213371266,
|
| 8459 |
+
"grad_norm": 0.5925318002700806,
|
| 8460 |
+
"learning_rate": 9.088379313436113e-06,
|
| 8461 |
+
"loss": 1.5679,
|
| 8462 |
+
"step": 1203
|
| 8463 |
+
},
|
| 8464 |
+
{
|
| 8465 |
+
"epoch": 0.6850640113798009,
|
| 8466 |
+
"grad_norm": 0.5964149832725525,
|
| 8467 |
+
"learning_rate": 8.993924917958874e-06,
|
| 8468 |
+
"loss": 1.4872,
|
| 8469 |
+
"step": 1204
|
| 8470 |
+
},
|
| 8471 |
+
{
|
| 8472 |
+
"epoch": 0.6856330014224751,
|
| 8473 |
+
"grad_norm": 0.5567532777786255,
|
| 8474 |
+
"learning_rate": 8.899940799609096e-06,
|
| 8475 |
+
"loss": 1.3922,
|
| 8476 |
+
"step": 1205
|
| 8477 |
+
},
|
| 8478 |
+
{
|
| 8479 |
+
"epoch": 0.6862019914651494,
|
| 8480 |
+
"grad_norm": 0.5803432464599609,
|
| 8481 |
+
"learning_rate": 8.806427444053033e-06,
|
| 8482 |
+
"loss": 1.319,
|
| 8483 |
+
"step": 1206
|
| 8484 |
+
},
|
| 8485 |
+
{
|
| 8486 |
+
"epoch": 0.6867709815078236,
|
| 8487 |
+
"grad_norm": 0.583640456199646,
|
| 8488 |
+
"learning_rate": 8.713385334524283e-06,
|
| 8489 |
+
"loss": 1.4564,
|
| 8490 |
+
"step": 1207
|
| 8491 |
+
},
|
| 8492 |
+
{
|
| 8493 |
+
"epoch": 0.6873399715504979,
|
| 8494 |
+
"grad_norm": 0.6316723227500916,
|
| 8495 |
+
"learning_rate": 8.620814951821232e-06,
|
| 8496 |
+
"loss": 1.4586,
|
| 8497 |
+
"step": 1208
|
| 8498 |
+
},
|
| 8499 |
+
{
|
| 8500 |
+
"epoch": 0.6879089615931722,
|
| 8501 |
+
"grad_norm": 0.5926545262336731,
|
| 8502 |
+
"learning_rate": 8.528716774304658e-06,
|
| 8503 |
+
"loss": 1.5008,
|
| 8504 |
+
"step": 1209
|
| 8505 |
+
},
|
| 8506 |
+
{
|
| 8507 |
+
"epoch": 0.6884779516358464,
|
| 8508 |
+
"grad_norm": 0.5738364458084106,
|
| 8509 |
+
"learning_rate": 8.43709127789517e-06,
|
| 8510 |
+
"loss": 1.3766,
|
| 8511 |
+
"step": 1210
|
| 8512 |
+
},
|
| 8513 |
+
{
|
| 8514 |
+
"epoch": 0.6890469416785207,
|
| 8515 |
+
"grad_norm": 0.5985202193260193,
|
| 8516 |
+
"learning_rate": 8.345938936070718e-06,
|
| 8517 |
+
"loss": 1.5175,
|
| 8518 |
+
"step": 1211
|
| 8519 |
+
},
|
| 8520 |
+
{
|
| 8521 |
+
"epoch": 0.6896159317211948,
|
| 8522 |
+
"grad_norm": 0.6196452379226685,
|
| 8523 |
+
"learning_rate": 8.255260219864324e-06,
|
| 8524 |
+
"loss": 1.6161,
|
| 8525 |
+
"step": 1212
|
| 8526 |
+
},
|
| 8527 |
+
{
|
| 8528 |
+
"epoch": 0.6901849217638691,
|
| 8529 |
+
"grad_norm": 0.6303586959838867,
|
| 8530 |
+
"learning_rate": 8.16505559786146e-06,
|
| 8531 |
+
"loss": 1.6054,
|
| 8532 |
+
"step": 1213
|
| 8533 |
+
},
|
| 8534 |
+
{
|
| 8535 |
+
"epoch": 0.6907539118065433,
|
| 8536 |
+
"grad_norm": 0.5856702327728271,
|
| 8537 |
+
"learning_rate": 8.07532553619772e-06,
|
| 8538 |
+
"loss": 1.5131,
|
| 8539 |
+
"step": 1214
|
| 8540 |
+
},
|
| 8541 |
+
{
|
| 8542 |
+
"epoch": 0.6913229018492176,
|
| 8543 |
+
"grad_norm": 0.5996472239494324,
|
| 8544 |
+
"learning_rate": 7.986070498556397e-06,
|
| 8545 |
+
"loss": 1.3462,
|
| 8546 |
+
"step": 1215
|
| 8547 |
+
},
|
| 8548 |
+
{
|
| 8549 |
+
"epoch": 0.6918918918918919,
|
| 8550 |
+
"grad_norm": 0.6016293168067932,
|
| 8551 |
+
"learning_rate": 7.897290946166037e-06,
|
| 8552 |
+
"loss": 1.3177,
|
| 8553 |
+
"step": 1216
|
| 8554 |
+
},
|
| 8555 |
+
{
|
| 8556 |
+
"epoch": 0.6924608819345661,
|
| 8557 |
+
"grad_norm": 0.5805103182792664,
|
| 8558 |
+
"learning_rate": 7.808987337798158e-06,
|
| 8559 |
+
"loss": 1.1701,
|
| 8560 |
+
"step": 1217
|
| 8561 |
+
},
|
| 8562 |
+
{
|
| 8563 |
+
"epoch": 0.6930298719772404,
|
| 8564 |
+
"grad_norm": 0.5823555588722229,
|
| 8565 |
+
"learning_rate": 7.721160129764792e-06,
|
| 8566 |
+
"loss": 1.275,
|
| 8567 |
+
"step": 1218
|
| 8568 |
+
},
|
| 8569 |
+
{
|
| 8570 |
+
"epoch": 0.6935988620199146,
|
| 8571 |
+
"grad_norm": 0.5881485939025879,
|
| 8572 |
+
"learning_rate": 7.633809775916135e-06,
|
| 8573 |
+
"loss": 1.3304,
|
| 8574 |
+
"step": 1219
|
| 8575 |
+
},
|
| 8576 |
+
{
|
| 8577 |
+
"epoch": 0.6941678520625889,
|
| 8578 |
+
"grad_norm": 0.610157310962677,
|
| 8579 |
+
"learning_rate": 7.546936727638298e-06,
|
| 8580 |
+
"loss": 1.325,
|
| 8581 |
+
"step": 1220
|
| 8582 |
+
},
|
| 8583 |
+
{
|
| 8584 |
+
"epoch": 0.6947368421052632,
|
| 8585 |
+
"grad_norm": 0.6015896797180176,
|
| 8586 |
+
"learning_rate": 7.460541433850788e-06,
|
| 8587 |
+
"loss": 1.4739,
|
| 8588 |
+
"step": 1221
|
| 8589 |
+
},
|
| 8590 |
+
{
|
| 8591 |
+
"epoch": 0.6953058321479374,
|
| 8592 |
+
"grad_norm": 0.6073941588401794,
|
| 8593 |
+
"learning_rate": 7.374624341004388e-06,
|
| 8594 |
+
"loss": 1.6308,
|
| 8595 |
+
"step": 1222
|
| 8596 |
+
},
|
| 8597 |
+
{
|
| 8598 |
+
"epoch": 0.6958748221906117,
|
| 8599 |
+
"grad_norm": 0.5897748470306396,
|
| 8600 |
+
"learning_rate": 7.289185893078721e-06,
|
| 8601 |
+
"loss": 1.4808,
|
| 8602 |
+
"step": 1223
|
| 8603 |
+
},
|
| 8604 |
+
{
|
| 8605 |
+
"epoch": 0.6964438122332859,
|
| 8606 |
+
"grad_norm": 0.6318244338035583,
|
| 8607 |
+
"learning_rate": 7.204226531579994e-06,
|
| 8608 |
+
"loss": 1.5134,
|
| 8609 |
+
"step": 1224
|
| 8610 |
+
},
|
| 8611 |
+
{
|
| 8612 |
+
"epoch": 0.6970128022759602,
|
| 8613 |
+
"grad_norm": 0.5809091329574585,
|
| 8614 |
+
"learning_rate": 7.119746695538765e-06,
|
| 8615 |
+
"loss": 1.4117,
|
| 8616 |
+
"step": 1225
|
| 8617 |
+
},
|
| 8618 |
+
{
|
| 8619 |
+
"epoch": 0.6975817923186344,
|
| 8620 |
+
"grad_norm": 0.6108141541481018,
|
| 8621 |
+
"learning_rate": 7.0357468215075275e-06,
|
| 8622 |
+
"loss": 1.3201,
|
| 8623 |
+
"step": 1226
|
| 8624 |
+
},
|
| 8625 |
+
{
|
| 8626 |
+
"epoch": 0.6981507823613087,
|
| 8627 |
+
"grad_norm": 0.566813051700592,
|
| 8628 |
+
"learning_rate": 6.952227343558671e-06,
|
| 8629 |
+
"loss": 1.502,
|
| 8630 |
+
"step": 1227
|
| 8631 |
+
},
|
| 8632 |
+
{
|
| 8633 |
+
"epoch": 0.698719772403983,
|
| 8634 |
+
"grad_norm": 0.5999999046325684,
|
| 8635 |
+
"learning_rate": 6.869188693282036e-06,
|
| 8636 |
+
"loss": 1.3958,
|
| 8637 |
+
"step": 1228
|
| 8638 |
+
},
|
| 8639 |
+
{
|
| 8640 |
+
"epoch": 0.6992887624466572,
|
| 8641 |
+
"grad_norm": 0.6325690150260925,
|
| 8642 |
+
"learning_rate": 6.786631299782797e-06,
|
| 8643 |
+
"loss": 1.4682,
|
| 8644 |
+
"step": 1229
|
| 8645 |
+
},
|
| 8646 |
+
{
|
| 8647 |
+
"epoch": 0.6998577524893315,
|
| 8648 |
+
"grad_norm": 0.5865020155906677,
|
| 8649 |
+
"learning_rate": 6.704555589679262e-06,
|
| 8650 |
+
"loss": 1.4662,
|
| 8651 |
+
"step": 1230
|
| 8652 |
+
},
|
| 8653 |
+
{
|
| 8654 |
+
"epoch": 0.7004267425320057,
|
| 8655 |
+
"grad_norm": 0.5978051424026489,
|
| 8656 |
+
"learning_rate": 6.622961987100518e-06,
|
| 8657 |
+
"loss": 1.4549,
|
| 8658 |
+
"step": 1231
|
| 8659 |
+
},
|
| 8660 |
+
{
|
| 8661 |
+
"epoch": 0.70099573257468,
|
| 8662 |
+
"grad_norm": 0.6172093152999878,
|
| 8663 |
+
"learning_rate": 6.541850913684444e-06,
|
| 8664 |
+
"loss": 1.52,
|
| 8665 |
+
"step": 1232
|
| 8666 |
+
},
|
| 8667 |
+
{
|
| 8668 |
+
"epoch": 0.7015647226173541,
|
| 8669 |
+
"grad_norm": 0.6080772280693054,
|
| 8670 |
+
"learning_rate": 6.461222788575394e-06,
|
| 8671 |
+
"loss": 1.5765,
|
| 8672 |
+
"step": 1233
|
| 8673 |
+
},
|
| 8674 |
+
{
|
| 8675 |
+
"epoch": 0.7021337126600284,
|
| 8676 |
+
"grad_norm": 0.6048703193664551,
|
| 8677 |
+
"learning_rate": 6.3810780284220495e-06,
|
| 8678 |
+
"loss": 1.6723,
|
| 8679 |
+
"step": 1234
|
| 8680 |
+
},
|
| 8681 |
+
{
|
| 8682 |
+
"epoch": 0.7027027027027027,
|
| 8683 |
+
"grad_norm": 0.5950552225112915,
|
| 8684 |
+
"learning_rate": 6.301417047375347e-06,
|
| 8685 |
+
"loss": 1.4492,
|
| 8686 |
+
"step": 1235
|
| 8687 |
+
},
|
| 8688 |
+
{
|
| 8689 |
+
"epoch": 0.7032716927453769,
|
| 8690 |
+
"grad_norm": 0.5936811566352844,
|
| 8691 |
+
"learning_rate": 6.222240257086176e-06,
|
| 8692 |
+
"loss": 1.4721,
|
| 8693 |
+
"step": 1236
|
| 8694 |
+
},
|
| 8695 |
+
{
|
| 8696 |
+
"epoch": 0.7038406827880512,
|
| 8697 |
+
"grad_norm": 0.5990265011787415,
|
| 8698 |
+
"learning_rate": 6.143548066703475e-06,
|
| 8699 |
+
"loss": 1.3644,
|
| 8700 |
+
"step": 1237
|
| 8701 |
+
},
|
| 8702 |
+
{
|
| 8703 |
+
"epoch": 0.7044096728307254,
|
| 8704 |
+
"grad_norm": 0.5738005638122559,
|
| 8705 |
+
"learning_rate": 6.065340882871906e-06,
|
| 8706 |
+
"loss": 1.4847,
|
| 8707 |
+
"step": 1238
|
| 8708 |
+
},
|
| 8709 |
+
{
|
| 8710 |
+
"epoch": 0.7049786628733997,
|
| 8711 |
+
"grad_norm": 0.5998217463493347,
|
| 8712 |
+
"learning_rate": 5.9876191097298475e-06,
|
| 8713 |
+
"loss": 1.4917,
|
| 8714 |
+
"step": 1239
|
| 8715 |
+
},
|
| 8716 |
+
{
|
| 8717 |
+
"epoch": 0.705547652916074,
|
| 8718 |
+
"grad_norm": 0.5693299174308777,
|
| 8719 |
+
"learning_rate": 5.910383148907395e-06,
|
| 8720 |
+
"loss": 1.3934,
|
| 8721 |
+
"step": 1240
|
| 8722 |
+
},
|
| 8723 |
+
{
|
| 8724 |
+
"epoch": 0.7061166429587482,
|
| 8725 |
+
"grad_norm": 0.5998255014419556,
|
| 8726 |
+
"learning_rate": 5.8336333995240526e-06,
|
| 8727 |
+
"loss": 1.6348,
|
| 8728 |
+
"step": 1241
|
| 8729 |
+
},
|
| 8730 |
+
{
|
| 8731 |
+
"epoch": 0.7066856330014225,
|
| 8732 |
+
"grad_norm": 0.6056730151176453,
|
| 8733 |
+
"learning_rate": 5.757370258186889e-06,
|
| 8734 |
+
"loss": 1.4748,
|
| 8735 |
+
"step": 1242
|
| 8736 |
+
},
|
| 8737 |
+
{
|
| 8738 |
+
"epoch": 0.7072546230440967,
|
| 8739 |
+
"grad_norm": 0.6141417622566223,
|
| 8740 |
+
"learning_rate": 5.6815941189884315e-06,
|
| 8741 |
+
"loss": 1.4371,
|
| 8742 |
+
"step": 1243
|
| 8743 |
+
},
|
| 8744 |
+
{
|
| 8745 |
+
"epoch": 0.707823613086771,
|
| 8746 |
+
"grad_norm": 0.5924522280693054,
|
| 8747 |
+
"learning_rate": 5.606305373504528e-06,
|
| 8748 |
+
"loss": 1.4896,
|
| 8749 |
+
"step": 1244
|
| 8750 |
+
},
|
| 8751 |
+
{
|
| 8752 |
+
"epoch": 0.7083926031294452,
|
| 8753 |
+
"grad_norm": 0.5907067656517029,
|
| 8754 |
+
"learning_rate": 5.5315044107925094e-06,
|
| 8755 |
+
"loss": 1.5258,
|
| 8756 |
+
"step": 1245
|
| 8757 |
+
},
|
| 8758 |
+
{
|
| 8759 |
+
"epoch": 0.7089615931721195,
|
| 8760 |
+
"grad_norm": 0.5856897234916687,
|
| 8761 |
+
"learning_rate": 5.457191617388957e-06,
|
| 8762 |
+
"loss": 1.3751,
|
| 8763 |
+
"step": 1246
|
| 8764 |
+
},
|
| 8765 |
+
{
|
| 8766 |
+
"epoch": 0.7095305832147938,
|
| 8767 |
+
"grad_norm": 0.5863030552864075,
|
| 8768 |
+
"learning_rate": 5.383367377307857e-06,
|
| 8769 |
+
"loss": 1.2607,
|
| 8770 |
+
"step": 1247
|
| 8771 |
+
},
|
| 8772 |
+
{
|
| 8773 |
+
"epoch": 0.710099573257468,
|
| 8774 |
+
"grad_norm": 0.5891332626342773,
|
| 8775 |
+
"learning_rate": 5.310032072038651e-06,
|
| 8776 |
+
"loss": 1.3852,
|
| 8777 |
+
"step": 1248
|
| 8778 |
+
},
|
| 8779 |
+
{
|
| 8780 |
+
"epoch": 0.7106685633001423,
|
| 8781 |
+
"grad_norm": 0.5775113701820374,
|
| 8782 |
+
"learning_rate": 5.237186080544098e-06,
|
| 8783 |
+
"loss": 1.5867,
|
| 8784 |
+
"step": 1249
|
| 8785 |
+
},
|
| 8786 |
+
{
|
| 8787 |
+
"epoch": 0.7112375533428165,
|
| 8788 |
+
"grad_norm": 0.5843526721000671,
|
| 8789 |
+
"learning_rate": 5.164829779258451e-06,
|
| 8790 |
+
"loss": 1.5694,
|
| 8791 |
+
"step": 1250
|
| 8792 |
+
},
|
| 8793 |
+
{
|
| 8794 |
+
"epoch": 0.7118065433854908,
|
| 8795 |
+
"grad_norm": 0.7019409537315369,
|
| 8796 |
+
"learning_rate": 5.092963542085483e-06,
|
| 8797 |
+
"loss": 1.4444,
|
| 8798 |
+
"step": 1251
|
| 8799 |
+
},
|
| 8800 |
+
{
|
| 8801 |
+
"epoch": 0.712375533428165,
|
| 8802 |
+
"grad_norm": 0.6279569864273071,
|
| 8803 |
+
"learning_rate": 5.021587740396505e-06,
|
| 8804 |
+
"loss": 1.5798,
|
| 8805 |
+
"step": 1252
|
| 8806 |
+
},
|
| 8807 |
+
{
|
| 8808 |
+
"epoch": 0.7129445234708393,
|
| 8809 |
+
"grad_norm": 0.6179226040840149,
|
| 8810 |
+
"learning_rate": 4.950702743028535e-06,
|
| 8811 |
+
"loss": 1.3976,
|
| 8812 |
+
"step": 1253
|
| 8813 |
+
},
|
| 8814 |
+
{
|
| 8815 |
+
"epoch": 0.7135135135135136,
|
| 8816 |
+
"grad_norm": 0.5874016284942627,
|
| 8817 |
+
"learning_rate": 4.880308916282305e-06,
|
| 8818 |
+
"loss": 1.5384,
|
| 8819 |
+
"step": 1254
|
| 8820 |
+
},
|
| 8821 |
+
{
|
| 8822 |
+
"epoch": 0.7140825035561877,
|
| 8823 |
+
"grad_norm": 0.5691651701927185,
|
| 8824 |
+
"learning_rate": 4.810406623920427e-06,
|
| 8825 |
+
"loss": 1.3594,
|
| 8826 |
+
"step": 1255
|
| 8827 |
+
},
|
| 8828 |
+
{
|
| 8829 |
+
"epoch": 0.714651493598862,
|
| 8830 |
+
"grad_norm": 0.5660080909729004,
|
| 8831 |
+
"learning_rate": 4.740996227165462e-06,
|
| 8832 |
+
"loss": 1.5635,
|
| 8833 |
+
"step": 1256
|
| 8834 |
+
},
|
| 8835 |
+
{
|
| 8836 |
+
"epoch": 0.7152204836415362,
|
| 8837 |
+
"grad_norm": 0.6053207516670227,
|
| 8838 |
+
"learning_rate": 4.672078084698095e-06,
|
| 8839 |
+
"loss": 1.5981,
|
| 8840 |
+
"step": 1257
|
| 8841 |
+
},
|
| 8842 |
+
{
|
| 8843 |
+
"epoch": 0.7157894736842105,
|
| 8844 |
+
"grad_norm": 0.5976085066795349,
|
| 8845 |
+
"learning_rate": 4.603652552655302e-06,
|
| 8846 |
+
"loss": 1.5909,
|
| 8847 |
+
"step": 1258
|
| 8848 |
+
},
|
| 8849 |
+
{
|
| 8850 |
+
"epoch": 0.7163584637268848,
|
| 8851 |
+
"grad_norm": 0.547666609287262,
|
| 8852 |
+
"learning_rate": 4.53571998462845e-06,
|
| 8853 |
+
"loss": 1.3622,
|
| 8854 |
+
"step": 1259
|
| 8855 |
+
},
|
| 8856 |
+
{
|
| 8857 |
+
"epoch": 0.716927453769559,
|
| 8858 |
+
"grad_norm": 0.6441154479980469,
|
| 8859 |
+
"learning_rate": 4.468280731661489e-06,
|
| 8860 |
+
"loss": 1.4626,
|
| 8861 |
+
"step": 1260
|
| 8862 |
+
},
|
| 8863 |
+
{
|
| 8864 |
+
"epoch": 0.7174964438122333,
|
| 8865 |
+
"grad_norm": 0.6044400930404663,
|
| 8866 |
+
"learning_rate": 4.4013351422491635e-06,
|
| 8867 |
+
"loss": 1.5432,
|
| 8868 |
+
"step": 1261
|
| 8869 |
+
},
|
| 8870 |
+
{
|
| 8871 |
+
"epoch": 0.7180654338549075,
|
| 8872 |
+
"grad_norm": 0.5658133029937744,
|
| 8873 |
+
"learning_rate": 4.334883562335157e-06,
|
| 8874 |
+
"loss": 1.4528,
|
| 8875 |
+
"step": 1262
|
| 8876 |
+
},
|
| 8877 |
+
{
|
| 8878 |
+
"epoch": 0.7186344238975818,
|
| 8879 |
+
"grad_norm": 0.6291137933731079,
|
| 8880 |
+
"learning_rate": 4.268926335310408e-06,
|
| 8881 |
+
"loss": 1.2975,
|
| 8882 |
+
"step": 1263
|
| 8883 |
+
},
|
| 8884 |
+
{
|
| 8885 |
+
"epoch": 0.719203413940256,
|
| 8886 |
+
"grad_norm": 0.5724123120307922,
|
| 8887 |
+
"learning_rate": 4.20346380201122e-06,
|
| 8888 |
+
"loss": 1.223,
|
| 8889 |
+
"step": 1264
|
| 8890 |
+
},
|
| 8891 |
+
{
|
| 8892 |
+
"epoch": 0.7197724039829303,
|
| 8893 |
+
"grad_norm": 0.6324455142021179,
|
| 8894 |
+
"learning_rate": 4.138496300717565e-06,
|
| 8895 |
+
"loss": 1.3516,
|
| 8896 |
+
"step": 1265
|
| 8897 |
+
},
|
| 8898 |
+
{
|
| 8899 |
+
"epoch": 0.7203413940256046,
|
| 8900 |
+
"grad_norm": 0.5916242599487305,
|
| 8901 |
+
"learning_rate": 4.0740241671513025e-06,
|
| 8902 |
+
"loss": 1.6546,
|
| 8903 |
+
"step": 1266
|
| 8904 |
+
},
|
| 8905 |
+
{
|
| 8906 |
+
"epoch": 0.7209103840682788,
|
| 8907 |
+
"grad_norm": 0.5780736804008484,
|
| 8908 |
+
"learning_rate": 4.010047734474454e-06,
|
| 8909 |
+
"loss": 1.4467,
|
| 8910 |
+
"step": 1267
|
| 8911 |
+
},
|
| 8912 |
+
{
|
| 8913 |
+
"epoch": 0.7214793741109531,
|
| 8914 |
+
"grad_norm": 0.580437958240509,
|
| 8915 |
+
"learning_rate": 3.946567333287566e-06,
|
| 8916 |
+
"loss": 1.2151,
|
| 8917 |
+
"step": 1268
|
| 8918 |
+
},
|
| 8919 |
+
{
|
| 8920 |
+
"epoch": 0.7220483641536273,
|
| 8921 |
+
"grad_norm": 0.631999135017395,
|
| 8922 |
+
"learning_rate": 3.883583291627823e-06,
|
| 8923 |
+
"loss": 1.6731,
|
| 8924 |
+
"step": 1269
|
| 8925 |
+
},
|
| 8926 |
+
{
|
| 8927 |
+
"epoch": 0.7226173541963016,
|
| 8928 |
+
"grad_norm": 0.5912725329399109,
|
| 8929 |
+
"learning_rate": 3.821095934967511e-06,
|
| 8930 |
+
"loss": 1.5419,
|
| 8931 |
+
"step": 1270
|
| 8932 |
+
},
|
| 8933 |
+
{
|
| 8934 |
+
"epoch": 0.7231863442389758,
|
| 8935 |
+
"grad_norm": 0.5841814279556274,
|
| 8936 |
+
"learning_rate": 3.759105586212275e-06,
|
| 8937 |
+
"loss": 1.36,
|
| 8938 |
+
"step": 1271
|
| 8939 |
+
},
|
| 8940 |
+
{
|
| 8941 |
+
"epoch": 0.7237553342816501,
|
| 8942 |
+
"grad_norm": 0.620486319065094,
|
| 8943 |
+
"learning_rate": 3.6976125656994376e-06,
|
| 8944 |
+
"loss": 1.4474,
|
| 8945 |
+
"step": 1272
|
| 8946 |
+
},
|
| 8947 |
+
{
|
| 8948 |
+
"epoch": 0.7243243243243244,
|
| 8949 |
+
"grad_norm": 0.5620819330215454,
|
| 8950 |
+
"learning_rate": 3.6366171911963455e-06,
|
| 8951 |
+
"loss": 1.3565,
|
| 8952 |
+
"step": 1273
|
| 8953 |
+
},
|
| 8954 |
+
{
|
| 8955 |
+
"epoch": 0.7248933143669986,
|
| 8956 |
+
"grad_norm": 0.6318161487579346,
|
| 8957 |
+
"learning_rate": 3.576119777898812e-06,
|
| 8958 |
+
"loss": 1.5721,
|
| 8959 |
+
"step": 1274
|
| 8960 |
+
},
|
| 8961 |
+
{
|
| 8962 |
+
"epoch": 0.7254623044096729,
|
| 8963 |
+
"grad_norm": 0.5643869638442993,
|
| 8964 |
+
"learning_rate": 3.516120638429332e-06,
|
| 8965 |
+
"loss": 1.3681,
|
| 8966 |
+
"step": 1275
|
| 8967 |
+
},
|
| 8968 |
+
{
|
| 8969 |
+
"epoch": 0.726031294452347,
|
| 8970 |
+
"grad_norm": 0.5829715132713318,
|
| 8971 |
+
"learning_rate": 3.4566200828356157e-06,
|
| 8972 |
+
"loss": 1.3699,
|
| 8973 |
+
"step": 1276
|
| 8974 |
+
},
|
| 8975 |
+
{
|
| 8976 |
+
"epoch": 0.7266002844950213,
|
| 8977 |
+
"grad_norm": 0.5623791813850403,
|
| 8978 |
+
"learning_rate": 3.397618418588877e-06,
|
| 8979 |
+
"loss": 1.3686,
|
| 8980 |
+
"step": 1277
|
| 8981 |
+
},
|
| 8982 |
+
{
|
| 8983 |
+
"epoch": 0.7271692745376956,
|
| 8984 |
+
"grad_norm": 0.5907699465751648,
|
| 8985 |
+
"learning_rate": 3.3391159505823165e-06,
|
| 8986 |
+
"loss": 1.5019,
|
| 8987 |
+
"step": 1278
|
| 8988 |
+
},
|
| 8989 |
+
{
|
| 8990 |
+
"epoch": 0.7277382645803698,
|
| 8991 |
+
"grad_norm": 0.5887323617935181,
|
| 8992 |
+
"learning_rate": 3.2811129811295416e-06,
|
| 8993 |
+
"loss": 1.5161,
|
| 8994 |
+
"step": 1279
|
| 8995 |
+
},
|
| 8996 |
+
{
|
| 8997 |
+
"epoch": 0.7283072546230441,
|
| 8998 |
+
"grad_norm": 0.6375420093536377,
|
| 8999 |
+
"learning_rate": 3.2236098099629353e-06,
|
| 9000 |
+
"loss": 1.53,
|
| 9001 |
+
"step": 1280
|
| 9002 |
+
},
|
| 9003 |
+
{
|
| 9004 |
+
"epoch": 0.7288762446657183,
|
| 9005 |
+
"grad_norm": 0.569848358631134,
|
| 9006 |
+
"learning_rate": 3.16660673423218e-06,
|
| 9007 |
+
"loss": 1.5462,
|
| 9008 |
+
"step": 1281
|
| 9009 |
+
},
|
| 9010 |
+
{
|
| 9011 |
+
"epoch": 0.7294452347083926,
|
| 9012 |
+
"grad_norm": 0.5773903727531433,
|
| 9013 |
+
"learning_rate": 3.1101040485027043e-06,
|
| 9014 |
+
"loss": 1.4332,
|
| 9015 |
+
"step": 1282
|
| 9016 |
+
},
|
| 9017 |
+
{
|
| 9018 |
+
"epoch": 0.7300142247510668,
|
| 9019 |
+
"grad_norm": 0.5759513974189758,
|
| 9020 |
+
"learning_rate": 3.0541020447541256e-06,
|
| 9021 |
+
"loss": 1.4906,
|
| 9022 |
+
"step": 1283
|
| 9023 |
+
},
|
| 9024 |
+
{
|
| 9025 |
+
"epoch": 0.7305832147937411,
|
| 9026 |
+
"grad_norm": 0.5894652009010315,
|
| 9027 |
+
"learning_rate": 2.99860101237881e-06,
|
| 9028 |
+
"loss": 1.3007,
|
| 9029 |
+
"step": 1284
|
| 9030 |
+
},
|
| 9031 |
+
{
|
| 9032 |
+
"epoch": 0.7311522048364154,
|
| 9033 |
+
"grad_norm": 0.5720746517181396,
|
| 9034 |
+
"learning_rate": 2.9436012381803156e-06,
|
| 9035 |
+
"loss": 1.5254,
|
| 9036 |
+
"step": 1285
|
| 9037 |
+
},
|
| 9038 |
+
{
|
| 9039 |
+
"epoch": 0.7317211948790896,
|
| 9040 |
+
"grad_norm": 0.6133726239204407,
|
| 9041 |
+
"learning_rate": 2.8891030063719183e-06,
|
| 9042 |
+
"loss": 1.6029,
|
| 9043 |
+
"step": 1286
|
| 9044 |
+
},
|
| 9045 |
+
{
|
| 9046 |
+
"epoch": 0.7322901849217639,
|
| 9047 |
+
"grad_norm": 0.6293920874595642,
|
| 9048 |
+
"learning_rate": 2.8351065985751766e-06,
|
| 9049 |
+
"loss": 1.5918,
|
| 9050 |
+
"step": 1287
|
| 9051 |
+
},
|
| 9052 |
+
{
|
| 9053 |
+
"epoch": 0.7328591749644381,
|
| 9054 |
+
"grad_norm": 0.5941974520683289,
|
| 9055 |
+
"learning_rate": 2.7816122938184255e-06,
|
| 9056 |
+
"loss": 1.43,
|
| 9057 |
+
"step": 1288
|
| 9058 |
+
},
|
| 9059 |
+
{
|
| 9060 |
+
"epoch": 0.7334281650071124,
|
| 9061 |
+
"grad_norm": 0.5790094137191772,
|
| 9062 |
+
"learning_rate": 2.7286203685354063e-06,
|
| 9063 |
+
"loss": 1.4635,
|
| 9064 |
+
"step": 1289
|
| 9065 |
+
},
|
| 9066 |
+
{
|
| 9067 |
+
"epoch": 0.7339971550497866,
|
| 9068 |
+
"grad_norm": 0.593591570854187,
|
| 9069 |
+
"learning_rate": 2.6761310965637833e-06,
|
| 9070 |
+
"loss": 1.554,
|
| 9071 |
+
"step": 1290
|
| 9072 |
+
},
|
| 9073 |
+
{
|
| 9074 |
+
"epoch": 0.7345661450924609,
|
| 9075 |
+
"grad_norm": 0.6287367939949036,
|
| 9076 |
+
"learning_rate": 2.62414474914372e-06,
|
| 9077 |
+
"loss": 1.2736,
|
| 9078 |
+
"step": 1291
|
| 9079 |
+
},
|
| 9080 |
+
{
|
| 9081 |
+
"epoch": 0.7351351351351352,
|
| 9082 |
+
"grad_norm": 0.586243748664856,
|
| 9083 |
+
"learning_rate": 2.5726615949165254e-06,
|
| 9084 |
+
"loss": 1.6281,
|
| 9085 |
+
"step": 1292
|
| 9086 |
+
},
|
| 9087 |
+
{
|
| 9088 |
+
"epoch": 0.7357041251778094,
|
| 9089 |
+
"grad_norm": 0.6094790697097778,
|
| 9090 |
+
"learning_rate": 2.5216818999232117e-06,
|
| 9091 |
+
"loss": 1.4495,
|
| 9092 |
+
"step": 1293
|
| 9093 |
+
},
|
| 9094 |
+
{
|
| 9095 |
+
"epoch": 0.7362731152204837,
|
| 9096 |
+
"grad_norm": 0.5789740681648254,
|
| 9097 |
+
"learning_rate": 2.4712059276031816e-06,
|
| 9098 |
+
"loss": 1.6063,
|
| 9099 |
+
"step": 1294
|
| 9100 |
+
},
|
| 9101 |
+
{
|
| 9102 |
+
"epoch": 0.7368421052631579,
|
| 9103 |
+
"grad_norm": 0.5999897122383118,
|
| 9104 |
+
"learning_rate": 2.421233938792811e-06,
|
| 9105 |
+
"loss": 1.3805,
|
| 9106 |
+
"step": 1295
|
| 9107 |
+
},
|
| 9108 |
+
{
|
| 9109 |
+
"epoch": 0.7374110953058322,
|
| 9110 |
+
"grad_norm": 0.5815314054489136,
|
| 9111 |
+
"learning_rate": 2.3717661917241117e-06,
|
| 9112 |
+
"loss": 1.4289,
|
| 9113 |
+
"step": 1296
|
| 9114 |
+
},
|
| 9115 |
+
{
|
| 9116 |
+
"epoch": 0.7379800853485065,
|
| 9117 |
+
"grad_norm": 0.5862295031547546,
|
| 9118 |
+
"learning_rate": 2.322802942023461e-06,
|
| 9119 |
+
"loss": 1.4672,
|
| 9120 |
+
"step": 1297
|
| 9121 |
+
},
|
| 9122 |
+
{
|
| 9123 |
+
"epoch": 0.7385490753911806,
|
| 9124 |
+
"grad_norm": 0.5870852470397949,
|
| 9125 |
+
"learning_rate": 2.2743444427101525e-06,
|
| 9126 |
+
"loss": 1.5368,
|
| 9127 |
+
"step": 1298
|
| 9128 |
+
},
|
| 9129 |
+
{
|
| 9130 |
+
"epoch": 0.7391180654338549,
|
| 9131 |
+
"grad_norm": 0.5981742143630981,
|
| 9132 |
+
"learning_rate": 2.2263909441952226e-06,
|
| 9133 |
+
"loss": 1.4996,
|
| 9134 |
+
"step": 1299
|
| 9135 |
+
},
|
| 9136 |
+
{
|
| 9137 |
+
"epoch": 0.7396870554765291,
|
| 9138 |
+
"grad_norm": 0.6381643414497375,
|
| 9139 |
+
"learning_rate": 2.178942694280095e-06,
|
| 9140 |
+
"loss": 1.4773,
|
| 9141 |
+
"step": 1300
|
| 9142 |
+
},
|
| 9143 |
+
{
|
| 9144 |
+
"epoch": 0.7402560455192034,
|
| 9145 |
+
"grad_norm": 0.5861015915870667,
|
| 9146 |
+
"learning_rate": 2.1319999381552604e-06,
|
| 9147 |
+
"loss": 1.3885,
|
| 9148 |
+
"step": 1301
|
| 9149 |
+
},
|
| 9150 |
+
{
|
| 9151 |
+
"epoch": 0.7408250355618776,
|
| 9152 |
+
"grad_norm": 0.5836819410324097,
|
| 9153 |
+
"learning_rate": 2.0855629183990867e-06,
|
| 9154 |
+
"loss": 1.4594,
|
| 9155 |
+
"step": 1302
|
| 9156 |
+
},
|
| 9157 |
+
{
|
| 9158 |
+
"epoch": 0.7413940256045519,
|
| 9159 |
+
"grad_norm": 0.57367342710495,
|
| 9160 |
+
"learning_rate": 2.039631874976533e-06,
|
| 9161 |
+
"loss": 1.5536,
|
| 9162 |
+
"step": 1303
|
| 9163 |
+
},
|
| 9164 |
+
{
|
| 9165 |
+
"epoch": 0.7419630156472262,
|
| 9166 |
+
"grad_norm": 0.6635316014289856,
|
| 9167 |
+
"learning_rate": 1.9942070452378836e-06,
|
| 9168 |
+
"loss": 1.3837,
|
| 9169 |
+
"step": 1304
|
| 9170 |
+
},
|
| 9171 |
+
{
|
| 9172 |
+
"epoch": 0.7425320056899004,
|
| 9173 |
+
"grad_norm": 0.6087902784347534,
|
| 9174 |
+
"learning_rate": 1.9492886639175922e-06,
|
| 9175 |
+
"loss": 1.5232,
|
| 9176 |
+
"step": 1305
|
| 9177 |
+
},
|
| 9178 |
+
{
|
| 9179 |
+
"epoch": 0.7431009957325747,
|
| 9180 |
+
"grad_norm": 0.5868977308273315,
|
| 9181 |
+
"learning_rate": 1.9048769631329399e-06,
|
| 9182 |
+
"loss": 1.5394,
|
| 9183 |
+
"step": 1306
|
| 9184 |
+
},
|
| 9185 |
+
{
|
| 9186 |
+
"epoch": 0.7436699857752489,
|
| 9187 |
+
"grad_norm": 0.5506064891815186,
|
| 9188 |
+
"learning_rate": 1.8609721723830132e-06,
|
| 9189 |
+
"loss": 1.3222,
|
| 9190 |
+
"step": 1307
|
| 9191 |
+
},
|
| 9192 |
+
{
|
| 9193 |
+
"epoch": 0.7442389758179232,
|
| 9194 |
+
"grad_norm": 0.6256322860717773,
|
| 9195 |
+
"learning_rate": 1.8175745185473714e-06,
|
| 9196 |
+
"loss": 1.6126,
|
| 9197 |
+
"step": 1308
|
| 9198 |
+
},
|
| 9199 |
+
{
|
| 9200 |
+
"epoch": 0.7448079658605974,
|
| 9201 |
+
"grad_norm": 0.6551912426948547,
|
| 9202 |
+
"learning_rate": 1.774684225884904e-06,
|
| 9203 |
+
"loss": 1.6678,
|
| 9204 |
+
"step": 1309
|
| 9205 |
+
},
|
| 9206 |
+
{
|
| 9207 |
+
"epoch": 0.7453769559032717,
|
| 9208 |
+
"grad_norm": 0.6695655584335327,
|
| 9209 |
+
"learning_rate": 1.7323015160327638e-06,
|
| 9210 |
+
"loss": 1.4653,
|
| 9211 |
+
"step": 1310
|
| 9212 |
+
},
|
| 9213 |
+
{
|
| 9214 |
+
"epoch": 0.745945945945946,
|
| 9215 |
+
"grad_norm": 0.5701092481613159,
|
| 9216 |
+
"learning_rate": 1.690426608005069e-06,
|
| 9217 |
+
"loss": 1.4619,
|
| 9218 |
+
"step": 1311
|
| 9219 |
+
},
|
| 9220 |
+
{
|
| 9221 |
+
"epoch": 0.7465149359886202,
|
| 9222 |
+
"grad_norm": 0.5779337882995605,
|
| 9223 |
+
"learning_rate": 1.6490597181919254e-06,
|
| 9224 |
+
"loss": 1.3819,
|
| 9225 |
+
"step": 1312
|
| 9226 |
+
},
|
| 9227 |
+
{
|
| 9228 |
+
"epoch": 0.7470839260312945,
|
| 9229 |
+
"grad_norm": 0.6129727959632874,
|
| 9230 |
+
"learning_rate": 1.6082010603582053e-06,
|
| 9231 |
+
"loss": 1.5916,
|
| 9232 |
+
"step": 1313
|
| 9233 |
+
},
|
| 9234 |
+
{
|
| 9235 |
+
"epoch": 0.7476529160739687,
|
| 9236 |
+
"grad_norm": 0.627040445804596,
|
| 9237 |
+
"learning_rate": 1.567850845642449e-06,
|
| 9238 |
+
"loss": 1.4437,
|
| 9239 |
+
"step": 1314
|
| 9240 |
+
},
|
| 9241 |
+
{
|
| 9242 |
+
"epoch": 0.748221906116643,
|
| 9243 |
+
"grad_norm": 0.5893445611000061,
|
| 9244 |
+
"learning_rate": 1.5280092825558645e-06,
|
| 9245 |
+
"loss": 1.4348,
|
| 9246 |
+
"step": 1315
|
| 9247 |
+
},
|
| 9248 |
+
{
|
| 9249 |
+
"epoch": 0.7487908961593173,
|
| 9250 |
+
"grad_norm": 0.5637263059616089,
|
| 9251 |
+
"learning_rate": 1.4886765769811072e-06,
|
| 9252 |
+
"loss": 1.4235,
|
| 9253 |
+
"step": 1316
|
| 9254 |
+
},
|
| 9255 |
+
{
|
| 9256 |
+
"epoch": 0.7493598862019915,
|
| 9257 |
+
"grad_norm": 0.5983802080154419,
|
| 9258 |
+
"learning_rate": 1.4498529321713584e-06,
|
| 9259 |
+
"loss": 1.5322,
|
| 9260 |
+
"step": 1317
|
| 9261 |
+
},
|
| 9262 |
+
{
|
| 9263 |
+
"epoch": 0.7499288762446658,
|
| 9264 |
+
"grad_norm": 0.6314471364021301,
|
| 9265 |
+
"learning_rate": 1.4115385487491583e-06,
|
| 9266 |
+
"loss": 1.511,
|
| 9267 |
+
"step": 1318
|
| 9268 |
+
},
|
| 9269 |
+
{
|
| 9270 |
+
"epoch": 0.7504978662873399,
|
| 9271 |
+
"grad_norm": 0.6288767457008362,
|
| 9272 |
+
"learning_rate": 1.3737336247054644e-06,
|
| 9273 |
+
"loss": 1.5245,
|
| 9274 |
+
"step": 1319
|
| 9275 |
+
},
|
| 9276 |
+
{
|
| 9277 |
+
"epoch": 0.7510668563300142,
|
| 9278 |
+
"grad_norm": 0.5633329153060913,
|
| 9279 |
+
"learning_rate": 1.3364383553985726e-06,
|
| 9280 |
+
"loss": 1.5002,
|
| 9281 |
+
"step": 1320
|
| 9282 |
+
},
|
| 9283 |
+
{
|
| 9284 |
+
"epoch": 0.7516358463726884,
|
| 9285 |
+
"grad_norm": 0.5696760416030884,
|
| 9286 |
+
"learning_rate": 1.2996529335530749e-06,
|
| 9287 |
+
"loss": 1.4833,
|
| 9288 |
+
"step": 1321
|
| 9289 |
+
},
|
| 9290 |
+
{
|
| 9291 |
+
"epoch": 0.7522048364153627,
|
| 9292 |
+
"grad_norm": 0.6070805191993713,
|
| 9293 |
+
"learning_rate": 1.2633775492589816e-06,
|
| 9294 |
+
"loss": 1.3631,
|
| 9295 |
+
"step": 1322
|
| 9296 |
+
},
|
| 9297 |
+
{
|
| 9298 |
+
"epoch": 0.752773826458037,
|
| 9299 |
+
"grad_norm": 0.6544724702835083,
|
| 9300 |
+
"learning_rate": 1.2276123899706227e-06,
|
| 9301 |
+
"loss": 1.3451,
|
| 9302 |
+
"step": 1323
|
| 9303 |
+
},
|
| 9304 |
+
{
|
| 9305 |
+
"epoch": 0.7533428165007112,
|
| 9306 |
+
"grad_norm": 0.5900003910064697,
|
| 9307 |
+
"learning_rate": 1.1923576405057258e-06,
|
| 9308 |
+
"loss": 1.4344,
|
| 9309 |
+
"step": 1324
|
| 9310 |
+
},
|
| 9311 |
+
{
|
| 9312 |
+
"epoch": 0.7539118065433855,
|
| 9313 |
+
"grad_norm": 0.5852407217025757,
|
| 9314 |
+
"learning_rate": 1.1576134830444619e-06,
|
| 9315 |
+
"loss": 1.3403,
|
| 9316 |
+
"step": 1325
|
| 9317 |
+
},
|
| 9318 |
+
{
|
| 9319 |
+
"epoch": 0.7544807965860597,
|
| 9320 |
+
"grad_norm": 0.557827353477478,
|
| 9321 |
+
"learning_rate": 1.1233800971285013e-06,
|
| 9322 |
+
"loss": 1.2692,
|
| 9323 |
+
"step": 1326
|
| 9324 |
+
},
|
| 9325 |
+
{
|
| 9326 |
+
"epoch": 0.755049786628734,
|
| 9327 |
+
"grad_norm": 0.5728473663330078,
|
| 9328 |
+
"learning_rate": 1.0896576596600705e-06,
|
| 9329 |
+
"loss": 1.2999,
|
| 9330 |
+
"step": 1327
|
| 9331 |
+
},
|
| 9332 |
+
{
|
| 9333 |
+
"epoch": 0.7556187766714082,
|
| 9334 |
+
"grad_norm": 0.6241177320480347,
|
| 9335 |
+
"learning_rate": 1.0564463449010852e-06,
|
| 9336 |
+
"loss": 1.3683,
|
| 9337 |
+
"step": 1328
|
| 9338 |
+
},
|
| 9339 |
+
{
|
| 9340 |
+
"epoch": 0.7561877667140825,
|
| 9341 |
+
"grad_norm": 0.5648365020751953,
|
| 9342 |
+
"learning_rate": 1.0237463244721747e-06,
|
| 9343 |
+
"loss": 1.4297,
|
| 9344 |
+
"step": 1329
|
| 9345 |
+
},
|
| 9346 |
+
{
|
| 9347 |
+
"epoch": 0.7567567567567568,
|
| 9348 |
+
"grad_norm": 0.590801477432251,
|
| 9349 |
+
"learning_rate": 9.915577673518695e-07,
|
| 9350 |
+
"loss": 1.5818,
|
| 9351 |
+
"step": 1330
|
| 9352 |
+
},
|
| 9353 |
+
{
|
| 9354 |
+
"epoch": 0.757325746799431,
|
| 9355 |
+
"grad_norm": 0.6130539178848267,
|
| 9356 |
+
"learning_rate": 9.59880839875682e-07,
|
| 9357 |
+
"loss": 1.2326,
|
| 9358 |
+
"step": 1331
|
| 9359 |
+
},
|
| 9360 |
+
{
|
| 9361 |
+
"epoch": 0.7578947368421053,
|
| 9362 |
+
"grad_norm": 0.5840870141983032,
|
| 9363 |
+
"learning_rate": 9.287157057352502e-07,
|
| 9364 |
+
"loss": 1.4967,
|
| 9365 |
+
"step": 1332
|
| 9366 |
+
},
|
| 9367 |
+
{
|
| 9368 |
+
"epoch": 0.7584637268847795,
|
| 9369 |
+
"grad_norm": 0.5912277102470398,
|
| 9370 |
+
"learning_rate": 8.980625259775277e-07,
|
| 9371 |
+
"loss": 1.5008,
|
| 9372 |
+
"step": 1333
|
| 9373 |
+
},
|
| 9374 |
+
{
|
| 9375 |
+
"epoch": 0.7590327169274538,
|
| 9376 |
+
"grad_norm": 0.5769196152687073,
|
| 9377 |
+
"learning_rate": 8.679214590039064e-07,
|
| 9378 |
+
"loss": 1.29,
|
| 9379 |
+
"step": 1334
|
| 9380 |
+
},
|
| 9381 |
+
{
|
| 9382 |
+
"epoch": 0.7596017069701281,
|
| 9383 |
+
"grad_norm": 0.6438156366348267,
|
| 9384 |
+
"learning_rate": 8.382926605694064e-07,
|
| 9385 |
+
"loss": 1.4775,
|
| 9386 |
+
"step": 1335
|
| 9387 |
+
},
|
| 9388 |
+
{
|
| 9389 |
+
"epoch": 0.7601706970128023,
|
| 9390 |
+
"grad_norm": 0.5996107459068298,
|
| 9391 |
+
"learning_rate": 8.091762837819094e-07,
|
| 9392 |
+
"loss": 1.5507,
|
| 9393 |
+
"step": 1336
|
| 9394 |
+
},
|
| 9395 |
+
{
|
| 9396 |
+
"epoch": 0.7607396870554766,
|
| 9397 |
+
"grad_norm": 0.6058631539344788,
|
| 9398 |
+
"learning_rate": 7.80572479101327e-07,
|
| 9399 |
+
"loss": 1.4726,
|
| 9400 |
+
"step": 1337
|
| 9401 |
+
},
|
| 9402 |
+
{
|
| 9403 |
+
"epoch": 0.7613086770981508,
|
| 9404 |
+
"grad_norm": 0.5997673273086548,
|
| 9405 |
+
"learning_rate": 7.524813943388331e-07,
|
| 9406 |
+
"loss": 1.5108,
|
| 9407 |
+
"step": 1338
|
| 9408 |
+
},
|
| 9409 |
+
{
|
| 9410 |
+
"epoch": 0.761877667140825,
|
| 9411 |
+
"grad_norm": 0.5843853950500488,
|
| 9412 |
+
"learning_rate": 7.249031746561108e-07,
|
| 9413 |
+
"loss": 1.4569,
|
| 9414 |
+
"step": 1339
|
| 9415 |
+
},
|
| 9416 |
+
{
|
| 9417 |
+
"epoch": 0.7624466571834992,
|
| 9418 |
+
"grad_norm": 0.5930002927780151,
|
| 9419 |
+
"learning_rate": 6.978379625645959e-07,
|
| 9420 |
+
"loss": 1.4858,
|
| 9421 |
+
"step": 1340
|
| 9422 |
+
},
|
| 9423 |
+
{
|
| 9424 |
+
"epoch": 0.7630156472261735,
|
| 9425 |
+
"grad_norm": 0.5973467826843262,
|
| 9426 |
+
"learning_rate": 6.712858979247116e-07,
|
| 9427 |
+
"loss": 1.5819,
|
| 9428 |
+
"step": 1341
|
| 9429 |
+
},
|
| 9430 |
+
{
|
| 9431 |
+
"epoch": 0.7635846372688478,
|
| 9432 |
+
"grad_norm": 0.601449191570282,
|
| 9433 |
+
"learning_rate": 6.452471179452135e-07,
|
| 9434 |
+
"loss": 1.6227,
|
| 9435 |
+
"step": 1342
|
| 9436 |
+
},
|
| 9437 |
+
{
|
| 9438 |
+
"epoch": 0.764153627311522,
|
| 9439 |
+
"grad_norm": 0.5814241170883179,
|
| 9440 |
+
"learning_rate": 6.197217571824232e-07,
|
| 9441 |
+
"loss": 1.3806,
|
| 9442 |
+
"step": 1343
|
| 9443 |
+
},
|
| 9444 |
+
{
|
| 9445 |
+
"epoch": 0.7647226173541963,
|
| 9446 |
+
"grad_norm": 0.5642287731170654,
|
| 9447 |
+
"learning_rate": 5.947099475395402e-07,
|
| 9448 |
+
"loss": 1.1583,
|
| 9449 |
+
"step": 1344
|
| 9450 |
+
},
|
| 9451 |
+
{
|
| 9452 |
+
"epoch": 0.7652916073968705,
|
| 9453 |
+
"grad_norm": 0.5667275190353394,
|
| 9454 |
+
"learning_rate": 5.702118182659866e-07,
|
| 9455 |
+
"loss": 1.5422,
|
| 9456 |
+
"step": 1345
|
| 9457 |
+
},
|
| 9458 |
+
{
|
| 9459 |
+
"epoch": 0.7658605974395448,
|
| 9460 |
+
"grad_norm": 0.5716063976287842,
|
| 9461 |
+
"learning_rate": 5.462274959567193e-07,
|
| 9462 |
+
"loss": 1.4454,
|
| 9463 |
+
"step": 1346
|
| 9464 |
+
},
|
| 9465 |
+
{
|
| 9466 |
+
"epoch": 0.766429587482219,
|
| 9467 |
+
"grad_norm": 0.5906545519828796,
|
| 9468 |
+
"learning_rate": 5.227571045515633e-07,
|
| 9469 |
+
"loss": 1.4336,
|
| 9470 |
+
"step": 1347
|
| 9471 |
+
},
|
| 9472 |
+
{
|
| 9473 |
+
"epoch": 0.7669985775248933,
|
| 9474 |
+
"grad_norm": 0.5766403079032898,
|
| 9475 |
+
"learning_rate": 4.998007653346126e-07,
|
| 9476 |
+
"loss": 1.3452,
|
| 9477 |
+
"step": 1348
|
| 9478 |
+
},
|
| 9479 |
+
{
|
| 9480 |
+
"epoch": 0.7675675675675676,
|
| 9481 |
+
"grad_norm": 0.5666573643684387,
|
| 9482 |
+
"learning_rate": 4.773585969335636e-07,
|
| 9483 |
+
"loss": 1.4239,
|
| 9484 |
+
"step": 1349
|
| 9485 |
+
},
|
| 9486 |
+
{
|
| 9487 |
+
"epoch": 0.7681365576102418,
|
| 9488 |
+
"grad_norm": 0.586741030216217,
|
| 9489 |
+
"learning_rate": 4.554307153191273e-07,
|
| 9490 |
+
"loss": 1.4837,
|
| 9491 |
+
"step": 1350
|
| 9492 |
+
},
|
| 9493 |
+
{
|
| 9494 |
+
"epoch": 0.7687055476529161,
|
| 9495 |
+
"grad_norm": 0.5627419948577881,
|
| 9496 |
+
"learning_rate": 4.340172338043846e-07,
|
| 9497 |
+
"loss": 1.1588,
|
| 9498 |
+
"step": 1351
|
| 9499 |
+
},
|
| 9500 |
+
{
|
| 9501 |
+
"epoch": 0.7692745376955903,
|
| 9502 |
+
"grad_norm": 0.6001150608062744,
|
| 9503 |
+
"learning_rate": 4.131182630442876e-07,
|
| 9504 |
+
"loss": 1.6122,
|
| 9505 |
+
"step": 1352
|
| 9506 |
+
},
|
| 9507 |
+
{
|
| 9508 |
+
"epoch": 0.7698435277382646,
|
| 9509 |
+
"grad_norm": 0.6114206314086914,
|
| 9510 |
+
"learning_rate": 3.9273391103499257e-07,
|
| 9511 |
+
"loss": 1.493,
|
| 9512 |
+
"step": 1353
|
| 9513 |
+
},
|
| 9514 |
+
{
|
| 9515 |
+
"epoch": 0.7704125177809389,
|
| 9516 |
+
"grad_norm": 0.5789377689361572,
|
| 9517 |
+
"learning_rate": 3.728642831133833e-07,
|
| 9518 |
+
"loss": 1.4742,
|
| 9519 |
+
"step": 1354
|
| 9520 |
+
},
|
| 9521 |
+
{
|
| 9522 |
+
"epoch": 0.7709815078236131,
|
| 9523 |
+
"grad_norm": 0.6265093684196472,
|
| 9524 |
+
"learning_rate": 3.5350948195645993e-07,
|
| 9525 |
+
"loss": 1.4284,
|
| 9526 |
+
"step": 1355
|
| 9527 |
+
},
|
| 9528 |
+
{
|
| 9529 |
+
"epoch": 0.7715504978662874,
|
| 9530 |
+
"grad_norm": 0.6009383201599121,
|
| 9531 |
+
"learning_rate": 3.3466960758082867e-07,
|
| 9532 |
+
"loss": 1.3373,
|
| 9533 |
+
"step": 1356
|
| 9534 |
+
},
|
| 9535 |
+
{
|
| 9536 |
+
"epoch": 0.7721194879089616,
|
| 9537 |
+
"grad_norm": 0.5908554196357727,
|
| 9538 |
+
"learning_rate": 3.163447573422351e-07,
|
| 9539 |
+
"loss": 1.4147,
|
| 9540 |
+
"step": 1357
|
| 9541 |
+
},
|
| 9542 |
+
{
|
| 9543 |
+
"epoch": 0.7726884779516359,
|
| 9544 |
+
"grad_norm": 0.614896297454834,
|
| 9545 |
+
"learning_rate": 2.985350259349762e-07,
|
| 9546 |
+
"loss": 1.3802,
|
| 9547 |
+
"step": 1358
|
| 9548 |
+
},
|
| 9549 |
+
{
|
| 9550 |
+
"epoch": 0.77325746799431,
|
| 9551 |
+
"grad_norm": 0.6063140630722046,
|
| 9552 |
+
"learning_rate": 2.812405053914891e-07,
|
| 9553 |
+
"loss": 1.66,
|
| 9554 |
+
"step": 1359
|
| 9555 |
+
},
|
| 9556 |
+
{
|
| 9557 |
+
"epoch": 0.7738264580369844,
|
| 9558 |
+
"grad_norm": 0.5757827758789062,
|
| 9559 |
+
"learning_rate": 2.644612850818073e-07,
|
| 9560 |
+
"loss": 1.4361,
|
| 9561 |
+
"step": 1360
|
| 9562 |
+
},
|
| 9563 |
+
{
|
| 9564 |
+
"epoch": 0.7743954480796587,
|
| 9565 |
+
"grad_norm": 0.5920888185501099,
|
| 9566 |
+
"learning_rate": 2.481974517131502e-07,
|
| 9567 |
+
"loss": 1.3681,
|
| 9568 |
+
"step": 1361
|
| 9569 |
+
},
|
| 9570 |
+
{
|
| 9571 |
+
"epoch": 0.7749644381223328,
|
| 9572 |
+
"grad_norm": 0.5669330358505249,
|
| 9573 |
+
"learning_rate": 2.324490893294673e-07,
|
| 9574 |
+
"loss": 1.5391,
|
| 9575 |
+
"step": 1362
|
| 9576 |
+
},
|
| 9577 |
+
{
|
| 9578 |
+
"epoch": 0.7755334281650071,
|
| 9579 |
+
"grad_norm": 0.5745005011558533,
|
| 9580 |
+
"learning_rate": 2.172162793109611e-07,
|
| 9581 |
+
"loss": 1.3118,
|
| 9582 |
+
"step": 1363
|
| 9583 |
+
},
|
| 9584 |
+
{
|
| 9585 |
+
"epoch": 0.7761024182076813,
|
| 9586 |
+
"grad_norm": 0.5584404468536377,
|
| 9587 |
+
"learning_rate": 2.0249910037374308e-07,
|
| 9588 |
+
"loss": 1.4001,
|
| 9589 |
+
"step": 1364
|
| 9590 |
+
},
|
| 9591 |
+
{
|
| 9592 |
+
"epoch": 0.7766714082503556,
|
| 9593 |
+
"grad_norm": 0.6618481874465942,
|
| 9594 |
+
"learning_rate": 1.8829762856933387e-07,
|
| 9595 |
+
"loss": 1.4373,
|
| 9596 |
+
"step": 1365
|
| 9597 |
+
},
|
| 9598 |
+
{
|
| 9599 |
+
"epoch": 0.7772403982930298,
|
| 9600 |
+
"grad_norm": 0.5788954496383667,
|
| 9601 |
+
"learning_rate": 1.7461193728436353e-07,
|
| 9602 |
+
"loss": 1.5453,
|
| 9603 |
+
"step": 1366
|
| 9604 |
+
},
|
| 9605 |
+
{
|
| 9606 |
+
"epoch": 0.7778093883357041,
|
| 9607 |
+
"grad_norm": 0.6013731956481934,
|
| 9608 |
+
"learning_rate": 1.614420972401165e-07,
|
| 9609 |
+
"loss": 1.5133,
|
| 9610 |
+
"step": 1367
|
| 9611 |
+
},
|
| 9612 |
+
{
|
| 9613 |
+
"epoch": 0.7783783783783784,
|
| 9614 |
+
"grad_norm": 0.622704029083252,
|
| 9615 |
+
"learning_rate": 1.4878817649220944e-07,
|
| 9616 |
+
"loss": 1.4024,
|
| 9617 |
+
"step": 1368
|
| 9618 |
+
},
|
| 9619 |
+
{
|
| 9620 |
+
"epoch": 0.7789473684210526,
|
| 9621 |
+
"grad_norm": 0.5754665732383728,
|
| 9622 |
+
"learning_rate": 1.36650240430225e-07,
|
| 9623 |
+
"loss": 1.2369,
|
| 9624 |
+
"step": 1369
|
| 9625 |
+
},
|
| 9626 |
+
{
|
| 9627 |
+
"epoch": 0.7795163584637269,
|
| 9628 |
+
"grad_norm": 0.5812003016471863,
|
| 9629 |
+
"learning_rate": 1.250283517774009e-07,
|
| 9630 |
+
"loss": 1.5473,
|
| 9631 |
+
"step": 1370
|
| 9632 |
+
},
|
| 9633 |
+
{
|
| 9634 |
+
"epoch": 0.7800853485064011,
|
| 9635 |
+
"grad_norm": 0.6443690657615662,
|
| 9636 |
+
"learning_rate": 1.1392257059023026e-07,
|
| 9637 |
+
"loss": 1.5182,
|
| 9638 |
+
"step": 1371
|
| 9639 |
+
},
|
| 9640 |
+
{
|
| 9641 |
+
"epoch": 0.7806543385490754,
|
| 9642 |
+
"grad_norm": 0.5886285901069641,
|
| 9643 |
+
"learning_rate": 1.0333295425825063e-07,
|
| 9644 |
+
"loss": 1.6711,
|
| 9645 |
+
"step": 1372
|
| 9646 |
+
},
|
| 9647 |
+
{
|
| 9648 |
+
"epoch": 0.7812233285917497,
|
| 9649 |
+
"grad_norm": 0.5866546034812927,
|
| 9650 |
+
"learning_rate": 9.325955750367766e-08,
|
| 9651 |
+
"loss": 1.326,
|
| 9652 |
+
"step": 1373
|
| 9653 |
+
},
|
| 9654 |
+
{
|
| 9655 |
+
"epoch": 0.7817923186344239,
|
| 9656 |
+
"grad_norm": 0.5975569486618042,
|
| 9657 |
+
"learning_rate": 8.370243238113862e-08,
|
| 9658 |
+
"loss": 1.522,
|
| 9659 |
+
"step": 1374
|
| 9660 |
+
},
|
| 9661 |
+
{
|
| 9662 |
+
"epoch": 0.7823613086770982,
|
| 9663 |
+
"grad_norm": 0.5648385286331177,
|
| 9664 |
+
"learning_rate": 7.466162827742817e-08,
|
| 9665 |
+
"loss": 1.4345,
|
| 9666 |
+
"step": 1375
|
| 9667 |
+
},
|
| 9668 |
+
{
|
| 9669 |
+
"epoch": 0.7829302987197724,
|
| 9670 |
+
"grad_norm": 0.6075783371925354,
|
| 9671 |
+
"learning_rate": 6.61371919112197e-08,
|
| 9672 |
+
"loss": 1.4661,
|
| 9673 |
+
"step": 1376
|
| 9674 |
+
},
|
| 9675 |
+
{
|
| 9676 |
+
"epoch": 0.7834992887624467,
|
| 9677 |
+
"grad_norm": 0.6267833113670349,
|
| 9678 |
+
"learning_rate": 5.812916733284324e-08,
|
| 9679 |
+
"loss": 1.5151,
|
| 9680 |
+
"step": 1377
|
| 9681 |
+
},
|
| 9682 |
+
{
|
| 9683 |
+
"epoch": 0.7840682788051209,
|
| 9684 |
+
"grad_norm": 0.5820707082748413,
|
| 9685 |
+
"learning_rate": 5.063759592404127e-08,
|
| 9686 |
+
"loss": 1.4866,
|
| 9687 |
+
"step": 1378
|
| 9688 |
+
},
|
| 9689 |
+
{
|
| 9690 |
+
"epoch": 0.7846372688477952,
|
| 9691 |
+
"grad_norm": 0.5778554677963257,
|
| 9692 |
+
"learning_rate": 4.366251639777996e-08,
|
| 9693 |
+
"loss": 1.3435,
|
| 9694 |
+
"step": 1379
|
| 9695 |
+
},
|
| 9696 |
+
{
|
| 9697 |
+
"epoch": 0.7852062588904695,
|
| 9698 |
+
"grad_norm": 0.5690316557884216,
|
| 9699 |
+
"learning_rate": 3.720396479803823e-08,
|
| 9700 |
+
"loss": 1.4896,
|
| 9701 |
+
"step": 1380
|
| 9702 |
+
},
|
| 9703 |
+
{
|
| 9704 |
+
"epoch": 0.7857752489331437,
|
| 9705 |
+
"grad_norm": 0.5569249391555786,
|
| 9706 |
+
"learning_rate": 3.126197449959678e-08,
|
| 9707 |
+
"loss": 1.358,
|
| 9708 |
+
"step": 1381
|
| 9709 |
+
},
|
| 9710 |
+
{
|
| 9711 |
+
"epoch": 0.786344238975818,
|
| 9712 |
+
"grad_norm": 0.6170508861541748,
|
| 9713 |
+
"learning_rate": 2.5836576207916018e-08,
|
| 9714 |
+
"loss": 1.4965,
|
| 9715 |
+
"step": 1382
|
| 9716 |
+
},
|
| 9717 |
+
{
|
| 9718 |
+
"epoch": 0.7869132290184921,
|
| 9719 |
+
"grad_norm": 0.5334784388542175,
|
| 9720 |
+
"learning_rate": 2.092779795892508e-08,
|
| 9721 |
+
"loss": 1.2593,
|
| 9722 |
+
"step": 1383
|
| 9723 |
+
},
|
| 9724 |
+
{
|
| 9725 |
+
"epoch": 0.7874822190611664,
|
| 9726 |
+
"grad_norm": 0.6333361864089966,
|
| 9727 |
+
"learning_rate": 1.6535665118910802e-08,
|
| 9728 |
+
"loss": 1.3485,
|
| 9729 |
+
"step": 1384
|
| 9730 |
+
},
|
| 9731 |
+
{
|
| 9732 |
+
"epoch": 0.7880512091038406,
|
| 9733 |
+
"grad_norm": 0.6247609853744507,
|
| 9734 |
+
"learning_rate": 1.2660200384384536e-08,
|
| 9735 |
+
"loss": 1.4111,
|
| 9736 |
+
"step": 1385
|
| 9737 |
+
},
|
| 9738 |
+
{
|
| 9739 |
+
"epoch": 0.7886201991465149,
|
| 9740 |
+
"grad_norm": 0.5764187574386597,
|
| 9741 |
+
"learning_rate": 9.301423781926666e-09,
|
| 9742 |
+
"loss": 1.201,
|
| 9743 |
+
"step": 1386
|
| 9744 |
+
},
|
| 9745 |
+
{
|
| 9746 |
+
"epoch": 0.7891891891891892,
|
| 9747 |
+
"grad_norm": 0.5923343896865845,
|
| 9748 |
+
"learning_rate": 6.459352668164442e-09,
|
| 9749 |
+
"loss": 1.5227,
|
| 9750 |
+
"step": 1387
|
| 9751 |
+
},
|
| 9752 |
+
{
|
| 9753 |
+
"epoch": 0.7897581792318634,
|
| 9754 |
+
"grad_norm": 0.578344464302063,
|
| 9755 |
+
"learning_rate": 4.134001729583226e-09,
|
| 9756 |
+
"loss": 1.3055,
|
| 9757 |
+
"step": 1388
|
| 9758 |
+
},
|
| 9759 |
+
{
|
| 9760 |
+
"epoch": 0.7903271692745377,
|
| 9761 |
+
"grad_norm": 0.595220148563385,
|
| 9762 |
+
"learning_rate": 2.3253829825153894e-09,
|
| 9763 |
+
"loss": 1.5065,
|
| 9764 |
+
"step": 1389
|
| 9765 |
+
},
|
| 9766 |
+
{
|
| 9767 |
+
"epoch": 0.7908961593172119,
|
| 9768 |
+
"grad_norm": 0.5846388339996338,
|
| 9769 |
+
"learning_rate": 1.033505773062604e-09,
|
| 9770 |
+
"loss": 1.5331,
|
| 9771 |
+
"step": 1390
|
| 9772 |
+
},
|
| 9773 |
+
{
|
| 9774 |
+
"epoch": 0.7914651493598862,
|
| 9775 |
+
"grad_norm": 0.5945976376533508,
|
| 9776 |
+
"learning_rate": 2.5837677706253003e-10,
|
| 9777 |
+
"loss": 1.5147,
|
| 9778 |
+
"step": 1391
|
| 9779 |
+
},
|
| 9780 |
+
{
|
| 9781 |
+
"epoch": 0.7920341394025604,
|
| 9782 |
+
"grad_norm": 0.6109771728515625,
|
| 9783 |
+
"learning_rate": 0.0,
|
| 9784 |
+
"loss": 1.508,
|
| 9785 |
+
"step": 1392
|
| 9786 |
+
},
|
| 9787 |
+
{
|
| 9788 |
+
"epoch": 0.7920341394025604,
|
| 9789 |
+
"eval_loss": 1.461509108543396,
|
| 9790 |
+
"eval_runtime": 16.4544,
|
| 9791 |
+
"eval_samples_per_second": 44.973,
|
| 9792 |
+
"eval_steps_per_second": 22.486,
|
| 9793 |
+
"step": 1392
|
| 9794 |
}
|
| 9795 |
],
|
| 9796 |
"logging_steps": 1,
|
|
|
|
| 9805 |
"should_evaluate": false,
|
| 9806 |
"should_log": false,
|
| 9807 |
"should_save": true,
|
| 9808 |
+
"should_training_stop": true
|
| 9809 |
},
|
| 9810 |
"attributes": {}
|
| 9811 |
}
|
| 9812 |
},
|
| 9813 |
+
"total_flos": 2.474815762936627e+16,
|
| 9814 |
"train_batch_size": 2,
|
| 9815 |
"trial_name": null,
|
| 9816 |
"trial_params": null
|