Training in progress, step 831, 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 25271744
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c1dbd19107319c058bd00055bb44a8c7f5eb208b888d6b52575bd7776291612d
|
| 3 |
size 25271744
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 13685836
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:70faa927d2d54c8ea7e7448c78084d6e6bba62b12460f6e68ddb25a4eda68fbd
|
| 3 |
size 13685836
|
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 sha256:2738afd85fa000763436a2d636f7e58177e0ad5cf69c1697cda8d874f4897ae9
|
| 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 sha256:896d4cf2739e2a4f3e21a7eda08b38acf92ea653e52c8f1d68438ebfb7a26d99
|
| 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": 277,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -3909,6 +3909,1953 @@
|
|
| 3909 |
"eval_samples_per_second": 55.531,
|
| 3910 |
"eval_steps_per_second": 27.825,
|
| 3911 |
"step": 554
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 3912 |
}
|
| 3913 |
],
|
| 3914 |
"logging_steps": 1,
|
|
@@ -3928,7 +5875,7 @@
|
|
| 3928 |
"attributes": {}
|
| 3929 |
}
|
| 3930 |
},
|
| 3931 |
-
"total_flos":
|
| 3932 |
"train_batch_size": 2,
|
| 3933 |
"trial_name": null,
|
| 3934 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.7503386004514673,
|
| 5 |
"eval_steps": 277,
|
| 6 |
+
"global_step": 831,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 3909 |
"eval_samples_per_second": 55.531,
|
| 3910 |
"eval_steps_per_second": 27.825,
|
| 3911 |
"step": 554
|
| 3912 |
+
},
|
| 3913 |
+
{
|
| 3914 |
+
"epoch": 0.5011286681715575,
|
| 3915 |
+
"grad_norm": 1.9571224451065063,
|
| 3916 |
+
"learning_rate": 0.00010114445321365482,
|
| 3917 |
+
"loss": 1.4151,
|
| 3918 |
+
"step": 555
|
| 3919 |
+
},
|
| 3920 |
+
{
|
| 3921 |
+
"epoch": 0.5020316027088036,
|
| 3922 |
+
"grad_norm": 1.3597534894943237,
|
| 3923 |
+
"learning_rate": 0.00010085834810820871,
|
| 3924 |
+
"loss": 1.1126,
|
| 3925 |
+
"step": 556
|
| 3926 |
+
},
|
| 3927 |
+
{
|
| 3928 |
+
"epoch": 0.5029345372460496,
|
| 3929 |
+
"grad_norm": 1.4733638763427734,
|
| 3930 |
+
"learning_rate": 0.00010057223597595243,
|
| 3931 |
+
"loss": 1.2403,
|
| 3932 |
+
"step": 557
|
| 3933 |
+
},
|
| 3934 |
+
{
|
| 3935 |
+
"epoch": 0.5038374717832957,
|
| 3936 |
+
"grad_norm": 1.2308521270751953,
|
| 3937 |
+
"learning_rate": 0.00010028611915912405,
|
| 3938 |
+
"loss": 0.7117,
|
| 3939 |
+
"step": 558
|
| 3940 |
+
},
|
| 3941 |
+
{
|
| 3942 |
+
"epoch": 0.5047404063205417,
|
| 3943 |
+
"grad_norm": 0.9335219860076904,
|
| 3944 |
+
"learning_rate": 0.0001,
|
| 3945 |
+
"loss": 0.6408,
|
| 3946 |
+
"step": 559
|
| 3947 |
+
},
|
| 3948 |
+
{
|
| 3949 |
+
"epoch": 0.5056433408577878,
|
| 3950 |
+
"grad_norm": 1.7657417058944702,
|
| 3951 |
+
"learning_rate": 9.9713880840876e-05,
|
| 3952 |
+
"loss": 1.53,
|
| 3953 |
+
"step": 560
|
| 3954 |
+
},
|
| 3955 |
+
{
|
| 3956 |
+
"epoch": 0.5065462753950338,
|
| 3957 |
+
"grad_norm": 1.1229721307754517,
|
| 3958 |
+
"learning_rate": 9.94277640240476e-05,
|
| 3959 |
+
"loss": 0.6362,
|
| 3960 |
+
"step": 561
|
| 3961 |
+
},
|
| 3962 |
+
{
|
| 3963 |
+
"epoch": 0.5074492099322799,
|
| 3964 |
+
"grad_norm": 1.421522855758667,
|
| 3965 |
+
"learning_rate": 9.914165189179131e-05,
|
| 3966 |
+
"loss": 1.0754,
|
| 3967 |
+
"step": 562
|
| 3968 |
+
},
|
| 3969 |
+
{
|
| 3970 |
+
"epoch": 0.5083521444695259,
|
| 3971 |
+
"grad_norm": 1.4801396131515503,
|
| 3972 |
+
"learning_rate": 9.885554678634523e-05,
|
| 3973 |
+
"loss": 1.2265,
|
| 3974 |
+
"step": 563
|
| 3975 |
+
},
|
| 3976 |
+
{
|
| 3977 |
+
"epoch": 0.509255079006772,
|
| 3978 |
+
"grad_norm": 1.18939208984375,
|
| 3979 |
+
"learning_rate": 9.856945104988989e-05,
|
| 3980 |
+
"loss": 1.0753,
|
| 3981 |
+
"step": 564
|
| 3982 |
+
},
|
| 3983 |
+
{
|
| 3984 |
+
"epoch": 0.510158013544018,
|
| 3985 |
+
"grad_norm": 1.6806690692901611,
|
| 3986 |
+
"learning_rate": 9.828336702452927e-05,
|
| 3987 |
+
"loss": 1.4325,
|
| 3988 |
+
"step": 565
|
| 3989 |
+
},
|
| 3990 |
+
{
|
| 3991 |
+
"epoch": 0.5110609480812641,
|
| 3992 |
+
"grad_norm": 1.2649235725402832,
|
| 3993 |
+
"learning_rate": 9.799729705227133e-05,
|
| 3994 |
+
"loss": 0.9571,
|
| 3995 |
+
"step": 566
|
| 3996 |
+
},
|
| 3997 |
+
{
|
| 3998 |
+
"epoch": 0.5119638826185101,
|
| 3999 |
+
"grad_norm": 1.5114171504974365,
|
| 4000 |
+
"learning_rate": 9.771124347500903e-05,
|
| 4001 |
+
"loss": 1.3416,
|
| 4002 |
+
"step": 567
|
| 4003 |
+
},
|
| 4004 |
+
{
|
| 4005 |
+
"epoch": 0.5128668171557562,
|
| 4006 |
+
"grad_norm": 1.5993719100952148,
|
| 4007 |
+
"learning_rate": 9.742520863450115e-05,
|
| 4008 |
+
"loss": 1.4492,
|
| 4009 |
+
"step": 568
|
| 4010 |
+
},
|
| 4011 |
+
{
|
| 4012 |
+
"epoch": 0.5137697516930022,
|
| 4013 |
+
"grad_norm": 1.540785789489746,
|
| 4014 |
+
"learning_rate": 9.713919487235306e-05,
|
| 4015 |
+
"loss": 1.4828,
|
| 4016 |
+
"step": 569
|
| 4017 |
+
},
|
| 4018 |
+
{
|
| 4019 |
+
"epoch": 0.5146726862302483,
|
| 4020 |
+
"grad_norm": 1.5753827095031738,
|
| 4021 |
+
"learning_rate": 9.685320452999751e-05,
|
| 4022 |
+
"loss": 1.2541,
|
| 4023 |
+
"step": 570
|
| 4024 |
+
},
|
| 4025 |
+
{
|
| 4026 |
+
"epoch": 0.5155756207674943,
|
| 4027 |
+
"grad_norm": 1.5678349733352661,
|
| 4028 |
+
"learning_rate": 9.656723994867566e-05,
|
| 4029 |
+
"loss": 1.6312,
|
| 4030 |
+
"step": 571
|
| 4031 |
+
},
|
| 4032 |
+
{
|
| 4033 |
+
"epoch": 0.5164785553047404,
|
| 4034 |
+
"grad_norm": 1.3026212453842163,
|
| 4035 |
+
"learning_rate": 9.628130346941767e-05,
|
| 4036 |
+
"loss": 1.431,
|
| 4037 |
+
"step": 572
|
| 4038 |
+
},
|
| 4039 |
+
{
|
| 4040 |
+
"epoch": 0.5173814898419865,
|
| 4041 |
+
"grad_norm": 1.6213833093643188,
|
| 4042 |
+
"learning_rate": 9.599539743302364e-05,
|
| 4043 |
+
"loss": 1.4334,
|
| 4044 |
+
"step": 573
|
| 4045 |
+
},
|
| 4046 |
+
{
|
| 4047 |
+
"epoch": 0.5182844243792325,
|
| 4048 |
+
"grad_norm": 1.3841042518615723,
|
| 4049 |
+
"learning_rate": 9.570952418004455e-05,
|
| 4050 |
+
"loss": 1.2536,
|
| 4051 |
+
"step": 574
|
| 4052 |
+
},
|
| 4053 |
+
{
|
| 4054 |
+
"epoch": 0.5191873589164786,
|
| 4055 |
+
"grad_norm": 1.253650426864624,
|
| 4056 |
+
"learning_rate": 9.542368605076296e-05,
|
| 4057 |
+
"loss": 1.1555,
|
| 4058 |
+
"step": 575
|
| 4059 |
+
},
|
| 4060 |
+
{
|
| 4061 |
+
"epoch": 0.5200902934537246,
|
| 4062 |
+
"grad_norm": 1.06034255027771,
|
| 4063 |
+
"learning_rate": 9.513788538517375e-05,
|
| 4064 |
+
"loss": 0.8516,
|
| 4065 |
+
"step": 576
|
| 4066 |
+
},
|
| 4067 |
+
{
|
| 4068 |
+
"epoch": 0.5209932279909707,
|
| 4069 |
+
"grad_norm": 1.22532320022583,
|
| 4070 |
+
"learning_rate": 9.485212452296535e-05,
|
| 4071 |
+
"loss": 0.7823,
|
| 4072 |
+
"step": 577
|
| 4073 |
+
},
|
| 4074 |
+
{
|
| 4075 |
+
"epoch": 0.5218961625282167,
|
| 4076 |
+
"grad_norm": 1.5111454725265503,
|
| 4077 |
+
"learning_rate": 9.456640580350018e-05,
|
| 4078 |
+
"loss": 1.3708,
|
| 4079 |
+
"step": 578
|
| 4080 |
+
},
|
| 4081 |
+
{
|
| 4082 |
+
"epoch": 0.5227990970654628,
|
| 4083 |
+
"grad_norm": 1.2933988571166992,
|
| 4084 |
+
"learning_rate": 9.428073156579569e-05,
|
| 4085 |
+
"loss": 1.0624,
|
| 4086 |
+
"step": 579
|
| 4087 |
+
},
|
| 4088 |
+
{
|
| 4089 |
+
"epoch": 0.5237020316027088,
|
| 4090 |
+
"grad_norm": 1.409776210784912,
|
| 4091 |
+
"learning_rate": 9.399510414850518e-05,
|
| 4092 |
+
"loss": 1.2465,
|
| 4093 |
+
"step": 580
|
| 4094 |
+
},
|
| 4095 |
+
{
|
| 4096 |
+
"epoch": 0.5246049661399549,
|
| 4097 |
+
"grad_norm": 1.200989842414856,
|
| 4098 |
+
"learning_rate": 9.370952588989872e-05,
|
| 4099 |
+
"loss": 0.9357,
|
| 4100 |
+
"step": 581
|
| 4101 |
+
},
|
| 4102 |
+
{
|
| 4103 |
+
"epoch": 0.5255079006772009,
|
| 4104 |
+
"grad_norm": 1.2518566846847534,
|
| 4105 |
+
"learning_rate": 9.342399912784386e-05,
|
| 4106 |
+
"loss": 1.2201,
|
| 4107 |
+
"step": 582
|
| 4108 |
+
},
|
| 4109 |
+
{
|
| 4110 |
+
"epoch": 0.526410835214447,
|
| 4111 |
+
"grad_norm": 1.6580489873886108,
|
| 4112 |
+
"learning_rate": 9.313852619978659e-05,
|
| 4113 |
+
"loss": 1.7361,
|
| 4114 |
+
"step": 583
|
| 4115 |
+
},
|
| 4116 |
+
{
|
| 4117 |
+
"epoch": 0.527313769751693,
|
| 4118 |
+
"grad_norm": 1.4227832555770874,
|
| 4119 |
+
"learning_rate": 9.285310944273232e-05,
|
| 4120 |
+
"loss": 1.2134,
|
| 4121 |
+
"step": 584
|
| 4122 |
+
},
|
| 4123 |
+
{
|
| 4124 |
+
"epoch": 0.5282167042889391,
|
| 4125 |
+
"grad_norm": 1.3826041221618652,
|
| 4126 |
+
"learning_rate": 9.256775119322642e-05,
|
| 4127 |
+
"loss": 1.0884,
|
| 4128 |
+
"step": 585
|
| 4129 |
+
},
|
| 4130 |
+
{
|
| 4131 |
+
"epoch": 0.5291196388261851,
|
| 4132 |
+
"grad_norm": 1.5409297943115234,
|
| 4133 |
+
"learning_rate": 9.228245378733537e-05,
|
| 4134 |
+
"loss": 1.0612,
|
| 4135 |
+
"step": 586
|
| 4136 |
+
},
|
| 4137 |
+
{
|
| 4138 |
+
"epoch": 0.5300225733634312,
|
| 4139 |
+
"grad_norm": 1.4004051685333252,
|
| 4140 |
+
"learning_rate": 9.199721956062766e-05,
|
| 4141 |
+
"loss": 1.0878,
|
| 4142 |
+
"step": 587
|
| 4143 |
+
},
|
| 4144 |
+
{
|
| 4145 |
+
"epoch": 0.5309255079006772,
|
| 4146 |
+
"grad_norm": 1.718218445777893,
|
| 4147 |
+
"learning_rate": 9.171205084815444e-05,
|
| 4148 |
+
"loss": 1.6136,
|
| 4149 |
+
"step": 588
|
| 4150 |
+
},
|
| 4151 |
+
{
|
| 4152 |
+
"epoch": 0.5318284424379233,
|
| 4153 |
+
"grad_norm": 1.3253271579742432,
|
| 4154 |
+
"learning_rate": 9.142694998443056e-05,
|
| 4155 |
+
"loss": 1.0533,
|
| 4156 |
+
"step": 589
|
| 4157 |
+
},
|
| 4158 |
+
{
|
| 4159 |
+
"epoch": 0.5327313769751693,
|
| 4160 |
+
"grad_norm": 1.449233055114746,
|
| 4161 |
+
"learning_rate": 9.11419193034155e-05,
|
| 4162 |
+
"loss": 1.4656,
|
| 4163 |
+
"step": 590
|
| 4164 |
+
},
|
| 4165 |
+
{
|
| 4166 |
+
"epoch": 0.5336343115124154,
|
| 4167 |
+
"grad_norm": 1.109061360359192,
|
| 4168 |
+
"learning_rate": 9.08569611384941e-05,
|
| 4169 |
+
"loss": 0.8056,
|
| 4170 |
+
"step": 591
|
| 4171 |
+
},
|
| 4172 |
+
{
|
| 4173 |
+
"epoch": 0.5345372460496614,
|
| 4174 |
+
"grad_norm": 1.4240176677703857,
|
| 4175 |
+
"learning_rate": 9.057207782245757e-05,
|
| 4176 |
+
"loss": 0.8339,
|
| 4177 |
+
"step": 592
|
| 4178 |
+
},
|
| 4179 |
+
{
|
| 4180 |
+
"epoch": 0.5354401805869075,
|
| 4181 |
+
"grad_norm": 1.1838033199310303,
|
| 4182 |
+
"learning_rate": 9.028727168748444e-05,
|
| 4183 |
+
"loss": 0.6906,
|
| 4184 |
+
"step": 593
|
| 4185 |
+
},
|
| 4186 |
+
{
|
| 4187 |
+
"epoch": 0.5363431151241534,
|
| 4188 |
+
"grad_norm": 1.3202743530273438,
|
| 4189 |
+
"learning_rate": 9.000254506512133e-05,
|
| 4190 |
+
"loss": 1.0769,
|
| 4191 |
+
"step": 594
|
| 4192 |
+
},
|
| 4193 |
+
{
|
| 4194 |
+
"epoch": 0.5372460496613995,
|
| 4195 |
+
"grad_norm": 1.3041194677352905,
|
| 4196 |
+
"learning_rate": 8.971790028626395e-05,
|
| 4197 |
+
"loss": 1.1342,
|
| 4198 |
+
"step": 595
|
| 4199 |
+
},
|
| 4200 |
+
{
|
| 4201 |
+
"epoch": 0.5381489841986457,
|
| 4202 |
+
"grad_norm": 1.382552981376648,
|
| 4203 |
+
"learning_rate": 8.943333968113808e-05,
|
| 4204 |
+
"loss": 1.0642,
|
| 4205 |
+
"step": 596
|
| 4206 |
+
},
|
| 4207 |
+
{
|
| 4208 |
+
"epoch": 0.5390519187358916,
|
| 4209 |
+
"grad_norm": 1.3533730506896973,
|
| 4210 |
+
"learning_rate": 8.914886557928031e-05,
|
| 4211 |
+
"loss": 1.3536,
|
| 4212 |
+
"step": 597
|
| 4213 |
+
},
|
| 4214 |
+
{
|
| 4215 |
+
"epoch": 0.5399548532731377,
|
| 4216 |
+
"grad_norm": 1.3678534030914307,
|
| 4217 |
+
"learning_rate": 8.886448030951912e-05,
|
| 4218 |
+
"loss": 0.9511,
|
| 4219 |
+
"step": 598
|
| 4220 |
+
},
|
| 4221 |
+
{
|
| 4222 |
+
"epoch": 0.5408577878103837,
|
| 4223 |
+
"grad_norm": 1.6321760416030884,
|
| 4224 |
+
"learning_rate": 8.85801861999558e-05,
|
| 4225 |
+
"loss": 1.0163,
|
| 4226 |
+
"step": 599
|
| 4227 |
+
},
|
| 4228 |
+
{
|
| 4229 |
+
"epoch": 0.5417607223476298,
|
| 4230 |
+
"grad_norm": 1.5338650941848755,
|
| 4231 |
+
"learning_rate": 8.82959855779453e-05,
|
| 4232 |
+
"loss": 1.3266,
|
| 4233 |
+
"step": 600
|
| 4234 |
+
},
|
| 4235 |
+
{
|
| 4236 |
+
"epoch": 0.5426636568848758,
|
| 4237 |
+
"grad_norm": 1.57759690284729,
|
| 4238 |
+
"learning_rate": 8.801188077007728e-05,
|
| 4239 |
+
"loss": 0.9428,
|
| 4240 |
+
"step": 601
|
| 4241 |
+
},
|
| 4242 |
+
{
|
| 4243 |
+
"epoch": 0.5435665914221219,
|
| 4244 |
+
"grad_norm": 1.4672127962112427,
|
| 4245 |
+
"learning_rate": 8.772787410215707e-05,
|
| 4246 |
+
"loss": 1.3282,
|
| 4247 |
+
"step": 602
|
| 4248 |
+
},
|
| 4249 |
+
{
|
| 4250 |
+
"epoch": 0.5444695259593679,
|
| 4251 |
+
"grad_norm": 1.4892001152038574,
|
| 4252 |
+
"learning_rate": 8.744396789918647e-05,
|
| 4253 |
+
"loss": 1.0562,
|
| 4254 |
+
"step": 603
|
| 4255 |
+
},
|
| 4256 |
+
{
|
| 4257 |
+
"epoch": 0.545372460496614,
|
| 4258 |
+
"grad_norm": 1.5642857551574707,
|
| 4259 |
+
"learning_rate": 8.71601644853449e-05,
|
| 4260 |
+
"loss": 1.0113,
|
| 4261 |
+
"step": 604
|
| 4262 |
+
},
|
| 4263 |
+
{
|
| 4264 |
+
"epoch": 0.54627539503386,
|
| 4265 |
+
"grad_norm": 1.4271188974380493,
|
| 4266 |
+
"learning_rate": 8.687646618397035e-05,
|
| 4267 |
+
"loss": 0.8131,
|
| 4268 |
+
"step": 605
|
| 4269 |
+
},
|
| 4270 |
+
{
|
| 4271 |
+
"epoch": 0.5471783295711061,
|
| 4272 |
+
"grad_norm": 1.384011149406433,
|
| 4273 |
+
"learning_rate": 8.659287531754024e-05,
|
| 4274 |
+
"loss": 0.9166,
|
| 4275 |
+
"step": 606
|
| 4276 |
+
},
|
| 4277 |
+
{
|
| 4278 |
+
"epoch": 0.5480812641083521,
|
| 4279 |
+
"grad_norm": 1.609860897064209,
|
| 4280 |
+
"learning_rate": 8.630939420765247e-05,
|
| 4281 |
+
"loss": 1.2092,
|
| 4282 |
+
"step": 607
|
| 4283 |
+
},
|
| 4284 |
+
{
|
| 4285 |
+
"epoch": 0.5489841986455982,
|
| 4286 |
+
"grad_norm": 1.5530509948730469,
|
| 4287 |
+
"learning_rate": 8.602602517500651e-05,
|
| 4288 |
+
"loss": 1.0349,
|
| 4289 |
+
"step": 608
|
| 4290 |
+
},
|
| 4291 |
+
{
|
| 4292 |
+
"epoch": 0.5498871331828442,
|
| 4293 |
+
"grad_norm": 1.5369367599487305,
|
| 4294 |
+
"learning_rate": 8.574277053938423e-05,
|
| 4295 |
+
"loss": 1.2942,
|
| 4296 |
+
"step": 609
|
| 4297 |
+
},
|
| 4298 |
+
{
|
| 4299 |
+
"epoch": 0.5507900677200903,
|
| 4300 |
+
"grad_norm": 1.3397136926651,
|
| 4301 |
+
"learning_rate": 8.545963261963102e-05,
|
| 4302 |
+
"loss": 0.9546,
|
| 4303 |
+
"step": 610
|
| 4304 |
+
},
|
| 4305 |
+
{
|
| 4306 |
+
"epoch": 0.5516930022573363,
|
| 4307 |
+
"grad_norm": 1.3543927669525146,
|
| 4308 |
+
"learning_rate": 8.517661373363684e-05,
|
| 4309 |
+
"loss": 1.0404,
|
| 4310 |
+
"step": 611
|
| 4311 |
+
},
|
| 4312 |
+
{
|
| 4313 |
+
"epoch": 0.5525959367945824,
|
| 4314 |
+
"grad_norm": 1.440779685974121,
|
| 4315 |
+
"learning_rate": 8.48937161983171e-05,
|
| 4316 |
+
"loss": 1.0322,
|
| 4317 |
+
"step": 612
|
| 4318 |
+
},
|
| 4319 |
+
{
|
| 4320 |
+
"epoch": 0.5534988713318284,
|
| 4321 |
+
"grad_norm": 1.3192452192306519,
|
| 4322 |
+
"learning_rate": 8.461094232959381e-05,
|
| 4323 |
+
"loss": 0.8567,
|
| 4324 |
+
"step": 613
|
| 4325 |
+
},
|
| 4326 |
+
{
|
| 4327 |
+
"epoch": 0.5544018058690745,
|
| 4328 |
+
"grad_norm": 1.2406502962112427,
|
| 4329 |
+
"learning_rate": 8.432829444237666e-05,
|
| 4330 |
+
"loss": 1.0435,
|
| 4331 |
+
"step": 614
|
| 4332 |
+
},
|
| 4333 |
+
{
|
| 4334 |
+
"epoch": 0.5553047404063205,
|
| 4335 |
+
"grad_norm": 1.4534803628921509,
|
| 4336 |
+
"learning_rate": 8.404577485054394e-05,
|
| 4337 |
+
"loss": 1.0001,
|
| 4338 |
+
"step": 615
|
| 4339 |
+
},
|
| 4340 |
+
{
|
| 4341 |
+
"epoch": 0.5562076749435666,
|
| 4342 |
+
"grad_norm": 1.1070985794067383,
|
| 4343 |
+
"learning_rate": 8.376338586692366e-05,
|
| 4344 |
+
"loss": 0.5828,
|
| 4345 |
+
"step": 616
|
| 4346 |
+
},
|
| 4347 |
+
{
|
| 4348 |
+
"epoch": 0.5571106094808126,
|
| 4349 |
+
"grad_norm": 1.6241154670715332,
|
| 4350 |
+
"learning_rate": 8.348112980327454e-05,
|
| 4351 |
+
"loss": 1.3879,
|
| 4352 |
+
"step": 617
|
| 4353 |
+
},
|
| 4354 |
+
{
|
| 4355 |
+
"epoch": 0.5580135440180587,
|
| 4356 |
+
"grad_norm": 1.4151923656463623,
|
| 4357 |
+
"learning_rate": 8.319900897026733e-05,
|
| 4358 |
+
"loss": 1.2522,
|
| 4359 |
+
"step": 618
|
| 4360 |
+
},
|
| 4361 |
+
{
|
| 4362 |
+
"epoch": 0.5589164785553047,
|
| 4363 |
+
"grad_norm": 1.5467820167541504,
|
| 4364 |
+
"learning_rate": 8.29170256774656e-05,
|
| 4365 |
+
"loss": 1.3313,
|
| 4366 |
+
"step": 619
|
| 4367 |
+
},
|
| 4368 |
+
{
|
| 4369 |
+
"epoch": 0.5598194130925508,
|
| 4370 |
+
"grad_norm": 1.7408809661865234,
|
| 4371 |
+
"learning_rate": 8.263518223330697e-05,
|
| 4372 |
+
"loss": 1.3344,
|
| 4373 |
+
"step": 620
|
| 4374 |
+
},
|
| 4375 |
+
{
|
| 4376 |
+
"epoch": 0.5607223476297969,
|
| 4377 |
+
"grad_norm": 1.752617597579956,
|
| 4378 |
+
"learning_rate": 8.235348094508426e-05,
|
| 4379 |
+
"loss": 1.1067,
|
| 4380 |
+
"step": 621
|
| 4381 |
+
},
|
| 4382 |
+
{
|
| 4383 |
+
"epoch": 0.5616252821670429,
|
| 4384 |
+
"grad_norm": 1.5512912273406982,
|
| 4385 |
+
"learning_rate": 8.207192411892646e-05,
|
| 4386 |
+
"loss": 1.2465,
|
| 4387 |
+
"step": 622
|
| 4388 |
+
},
|
| 4389 |
+
{
|
| 4390 |
+
"epoch": 0.562528216704289,
|
| 4391 |
+
"grad_norm": 1.5489177703857422,
|
| 4392 |
+
"learning_rate": 8.179051405977993e-05,
|
| 4393 |
+
"loss": 1.1418,
|
| 4394 |
+
"step": 623
|
| 4395 |
+
},
|
| 4396 |
+
{
|
| 4397 |
+
"epoch": 0.563431151241535,
|
| 4398 |
+
"grad_norm": 1.3127493858337402,
|
| 4399 |
+
"learning_rate": 8.150925307138969e-05,
|
| 4400 |
+
"loss": 0.7634,
|
| 4401 |
+
"step": 624
|
| 4402 |
+
},
|
| 4403 |
+
{
|
| 4404 |
+
"epoch": 0.5643340857787811,
|
| 4405 |
+
"grad_norm": 1.4094018936157227,
|
| 4406 |
+
"learning_rate": 8.122814345628016e-05,
|
| 4407 |
+
"loss": 0.8071,
|
| 4408 |
+
"step": 625
|
| 4409 |
+
},
|
| 4410 |
+
{
|
| 4411 |
+
"epoch": 0.5652370203160271,
|
| 4412 |
+
"grad_norm": 1.5054543018341064,
|
| 4413 |
+
"learning_rate": 8.094718751573667e-05,
|
| 4414 |
+
"loss": 1.0688,
|
| 4415 |
+
"step": 626
|
| 4416 |
+
},
|
| 4417 |
+
{
|
| 4418 |
+
"epoch": 0.5661399548532732,
|
| 4419 |
+
"grad_norm": 1.7666995525360107,
|
| 4420 |
+
"learning_rate": 8.06663875497866e-05,
|
| 4421 |
+
"loss": 1.1685,
|
| 4422 |
+
"step": 627
|
| 4423 |
+
},
|
| 4424 |
+
{
|
| 4425 |
+
"epoch": 0.5670428893905192,
|
| 4426 |
+
"grad_norm": 1.767767071723938,
|
| 4427 |
+
"learning_rate": 8.038574585718032e-05,
|
| 4428 |
+
"loss": 1.6177,
|
| 4429 |
+
"step": 628
|
| 4430 |
+
},
|
| 4431 |
+
{
|
| 4432 |
+
"epoch": 0.5679458239277653,
|
| 4433 |
+
"grad_norm": 1.3216131925582886,
|
| 4434 |
+
"learning_rate": 8.010526473537249e-05,
|
| 4435 |
+
"loss": 0.673,
|
| 4436 |
+
"step": 629
|
| 4437 |
+
},
|
| 4438 |
+
{
|
| 4439 |
+
"epoch": 0.5688487584650113,
|
| 4440 |
+
"grad_norm": 1.6139782667160034,
|
| 4441 |
+
"learning_rate": 7.98249464805034e-05,
|
| 4442 |
+
"loss": 1.2622,
|
| 4443 |
+
"step": 630
|
| 4444 |
+
},
|
| 4445 |
+
{
|
| 4446 |
+
"epoch": 0.5697516930022574,
|
| 4447 |
+
"grad_norm": 1.6862844228744507,
|
| 4448 |
+
"learning_rate": 7.954479338737995e-05,
|
| 4449 |
+
"loss": 1.4653,
|
| 4450 |
+
"step": 631
|
| 4451 |
+
},
|
| 4452 |
+
{
|
| 4453 |
+
"epoch": 0.5706546275395034,
|
| 4454 |
+
"grad_norm": 1.3713010549545288,
|
| 4455 |
+
"learning_rate": 7.926480774945689e-05,
|
| 4456 |
+
"loss": 0.9346,
|
| 4457 |
+
"step": 632
|
| 4458 |
+
},
|
| 4459 |
+
{
|
| 4460 |
+
"epoch": 0.5715575620767495,
|
| 4461 |
+
"grad_norm": 1.6012693643569946,
|
| 4462 |
+
"learning_rate": 7.89849918588183e-05,
|
| 4463 |
+
"loss": 0.862,
|
| 4464 |
+
"step": 633
|
| 4465 |
+
},
|
| 4466 |
+
{
|
| 4467 |
+
"epoch": 0.5724604966139955,
|
| 4468 |
+
"grad_norm": 1.5387890338897705,
|
| 4469 |
+
"learning_rate": 7.870534800615845e-05,
|
| 4470 |
+
"loss": 1.0758,
|
| 4471 |
+
"step": 634
|
| 4472 |
+
},
|
| 4473 |
+
{
|
| 4474 |
+
"epoch": 0.5733634311512416,
|
| 4475 |
+
"grad_norm": 1.8564848899841309,
|
| 4476 |
+
"learning_rate": 7.84258784807633e-05,
|
| 4477 |
+
"loss": 1.4123,
|
| 4478 |
+
"step": 635
|
| 4479 |
+
},
|
| 4480 |
+
{
|
| 4481 |
+
"epoch": 0.5742663656884875,
|
| 4482 |
+
"grad_norm": 1.7306365966796875,
|
| 4483 |
+
"learning_rate": 7.814658557049175e-05,
|
| 4484 |
+
"loss": 0.9185,
|
| 4485 |
+
"step": 636
|
| 4486 |
+
},
|
| 4487 |
+
{
|
| 4488 |
+
"epoch": 0.5751693002257336,
|
| 4489 |
+
"grad_norm": 1.6491857767105103,
|
| 4490 |
+
"learning_rate": 7.786747156175676e-05,
|
| 4491 |
+
"loss": 1.067,
|
| 4492 |
+
"step": 637
|
| 4493 |
+
},
|
| 4494 |
+
{
|
| 4495 |
+
"epoch": 0.5760722347629796,
|
| 4496 |
+
"grad_norm": 1.5658364295959473,
|
| 4497 |
+
"learning_rate": 7.758853873950676e-05,
|
| 4498 |
+
"loss": 1.2779,
|
| 4499 |
+
"step": 638
|
| 4500 |
+
},
|
| 4501 |
+
{
|
| 4502 |
+
"epoch": 0.5769751693002257,
|
| 4503 |
+
"grad_norm": 1.6873422861099243,
|
| 4504 |
+
"learning_rate": 7.730978938720693e-05,
|
| 4505 |
+
"loss": 1.2657,
|
| 4506 |
+
"step": 639
|
| 4507 |
+
},
|
| 4508 |
+
{
|
| 4509 |
+
"epoch": 0.5778781038374717,
|
| 4510 |
+
"grad_norm": 1.5114984512329102,
|
| 4511 |
+
"learning_rate": 7.703122578682046e-05,
|
| 4512 |
+
"loss": 1.1484,
|
| 4513 |
+
"step": 640
|
| 4514 |
+
},
|
| 4515 |
+
{
|
| 4516 |
+
"epoch": 0.5787810383747178,
|
| 4517 |
+
"grad_norm": 1.3105080127716064,
|
| 4518 |
+
"learning_rate": 7.675285021878989e-05,
|
| 4519 |
+
"loss": 0.9168,
|
| 4520 |
+
"step": 641
|
| 4521 |
+
},
|
| 4522 |
+
{
|
| 4523 |
+
"epoch": 0.5796839729119638,
|
| 4524 |
+
"grad_norm": 1.3600938320159912,
|
| 4525 |
+
"learning_rate": 7.647466496201847e-05,
|
| 4526 |
+
"loss": 0.9933,
|
| 4527 |
+
"step": 642
|
| 4528 |
+
},
|
| 4529 |
+
{
|
| 4530 |
+
"epoch": 0.5805869074492099,
|
| 4531 |
+
"grad_norm": 1.7162373065948486,
|
| 4532 |
+
"learning_rate": 7.619667229385146e-05,
|
| 4533 |
+
"loss": 1.456,
|
| 4534 |
+
"step": 643
|
| 4535 |
+
},
|
| 4536 |
+
{
|
| 4537 |
+
"epoch": 0.581489841986456,
|
| 4538 |
+
"grad_norm": 1.4823421239852905,
|
| 4539 |
+
"learning_rate": 7.591887449005748e-05,
|
| 4540 |
+
"loss": 0.9799,
|
| 4541 |
+
"step": 644
|
| 4542 |
+
},
|
| 4543 |
+
{
|
| 4544 |
+
"epoch": 0.582392776523702,
|
| 4545 |
+
"grad_norm": 1.467466115951538,
|
| 4546 |
+
"learning_rate": 7.564127382481e-05,
|
| 4547 |
+
"loss": 1.2861,
|
| 4548 |
+
"step": 645
|
| 4549 |
+
},
|
| 4550 |
+
{
|
| 4551 |
+
"epoch": 0.5832957110609481,
|
| 4552 |
+
"grad_norm": 1.736372470855713,
|
| 4553 |
+
"learning_rate": 7.536387257066854e-05,
|
| 4554 |
+
"loss": 1.4079,
|
| 4555 |
+
"step": 646
|
| 4556 |
+
},
|
| 4557 |
+
{
|
| 4558 |
+
"epoch": 0.5841986455981941,
|
| 4559 |
+
"grad_norm": 1.3974703550338745,
|
| 4560 |
+
"learning_rate": 7.508667299856014e-05,
|
| 4561 |
+
"loss": 0.6523,
|
| 4562 |
+
"step": 647
|
| 4563 |
+
},
|
| 4564 |
+
{
|
| 4565 |
+
"epoch": 0.5851015801354402,
|
| 4566 |
+
"grad_norm": 1.7177207469940186,
|
| 4567 |
+
"learning_rate": 7.480967737776089e-05,
|
| 4568 |
+
"loss": 1.2047,
|
| 4569 |
+
"step": 648
|
| 4570 |
+
},
|
| 4571 |
+
{
|
| 4572 |
+
"epoch": 0.5860045146726862,
|
| 4573 |
+
"grad_norm": 1.6860179901123047,
|
| 4574 |
+
"learning_rate": 7.453288797587714e-05,
|
| 4575 |
+
"loss": 1.6312,
|
| 4576 |
+
"step": 649
|
| 4577 |
+
},
|
| 4578 |
+
{
|
| 4579 |
+
"epoch": 0.5869074492099323,
|
| 4580 |
+
"grad_norm": 1.3072704076766968,
|
| 4581 |
+
"learning_rate": 7.425630705882707e-05,
|
| 4582 |
+
"loss": 0.8851,
|
| 4583 |
+
"step": 650
|
| 4584 |
+
},
|
| 4585 |
+
{
|
| 4586 |
+
"epoch": 0.5878103837471783,
|
| 4587 |
+
"grad_norm": 1.2708418369293213,
|
| 4588 |
+
"learning_rate": 7.397993689082216e-05,
|
| 4589 |
+
"loss": 0.826,
|
| 4590 |
+
"step": 651
|
| 4591 |
+
},
|
| 4592 |
+
{
|
| 4593 |
+
"epoch": 0.5887133182844244,
|
| 4594 |
+
"grad_norm": 1.4069938659667969,
|
| 4595 |
+
"learning_rate": 7.370377973434855e-05,
|
| 4596 |
+
"loss": 1.0625,
|
| 4597 |
+
"step": 652
|
| 4598 |
+
},
|
| 4599 |
+
{
|
| 4600 |
+
"epoch": 0.5896162528216704,
|
| 4601 |
+
"grad_norm": 1.5313855409622192,
|
| 4602 |
+
"learning_rate": 7.34278378501486e-05,
|
| 4603 |
+
"loss": 0.9191,
|
| 4604 |
+
"step": 653
|
| 4605 |
+
},
|
| 4606 |
+
{
|
| 4607 |
+
"epoch": 0.5905191873589165,
|
| 4608 |
+
"grad_norm": 1.6954371929168701,
|
| 4609 |
+
"learning_rate": 7.31521134972023e-05,
|
| 4610 |
+
"loss": 1.564,
|
| 4611 |
+
"step": 654
|
| 4612 |
+
},
|
| 4613 |
+
{
|
| 4614 |
+
"epoch": 0.5914221218961625,
|
| 4615 |
+
"grad_norm": 1.5678001642227173,
|
| 4616 |
+
"learning_rate": 7.2876608932709e-05,
|
| 4617 |
+
"loss": 1.1355,
|
| 4618 |
+
"step": 655
|
| 4619 |
+
},
|
| 4620 |
+
{
|
| 4621 |
+
"epoch": 0.5923250564334086,
|
| 4622 |
+
"grad_norm": 1.3186235427856445,
|
| 4623 |
+
"learning_rate": 7.26013264120686e-05,
|
| 4624 |
+
"loss": 0.842,
|
| 4625 |
+
"step": 656
|
| 4626 |
+
},
|
| 4627 |
+
{
|
| 4628 |
+
"epoch": 0.5932279909706546,
|
| 4629 |
+
"grad_norm": 1.612156629562378,
|
| 4630 |
+
"learning_rate": 7.232626818886326e-05,
|
| 4631 |
+
"loss": 0.8356,
|
| 4632 |
+
"step": 657
|
| 4633 |
+
},
|
| 4634 |
+
{
|
| 4635 |
+
"epoch": 0.5941309255079007,
|
| 4636 |
+
"grad_norm": 1.3746870756149292,
|
| 4637 |
+
"learning_rate": 7.205143651483906e-05,
|
| 4638 |
+
"loss": 0.7663,
|
| 4639 |
+
"step": 658
|
| 4640 |
+
},
|
| 4641 |
+
{
|
| 4642 |
+
"epoch": 0.5950338600451467,
|
| 4643 |
+
"grad_norm": 1.4078235626220703,
|
| 4644 |
+
"learning_rate": 7.177683363988736e-05,
|
| 4645 |
+
"loss": 0.9621,
|
| 4646 |
+
"step": 659
|
| 4647 |
+
},
|
| 4648 |
+
{
|
| 4649 |
+
"epoch": 0.5959367945823928,
|
| 4650 |
+
"grad_norm": 1.5795187950134277,
|
| 4651 |
+
"learning_rate": 7.150246181202648e-05,
|
| 4652 |
+
"loss": 0.8875,
|
| 4653 |
+
"step": 660
|
| 4654 |
+
},
|
| 4655 |
+
{
|
| 4656 |
+
"epoch": 0.5968397291196388,
|
| 4657 |
+
"grad_norm": 1.5938857793807983,
|
| 4658 |
+
"learning_rate": 7.122832327738331e-05,
|
| 4659 |
+
"loss": 0.9685,
|
| 4660 |
+
"step": 661
|
| 4661 |
+
},
|
| 4662 |
+
{
|
| 4663 |
+
"epoch": 0.5977426636568849,
|
| 4664 |
+
"grad_norm": 1.5371569395065308,
|
| 4665 |
+
"learning_rate": 7.095442028017492e-05,
|
| 4666 |
+
"loss": 1.0103,
|
| 4667 |
+
"step": 662
|
| 4668 |
+
},
|
| 4669 |
+
{
|
| 4670 |
+
"epoch": 0.5986455981941309,
|
| 4671 |
+
"grad_norm": 1.641914963722229,
|
| 4672 |
+
"learning_rate": 7.068075506269005e-05,
|
| 4673 |
+
"loss": 1.1853,
|
| 4674 |
+
"step": 663
|
| 4675 |
+
},
|
| 4676 |
+
{
|
| 4677 |
+
"epoch": 0.599548532731377,
|
| 4678 |
+
"grad_norm": 1.546448826789856,
|
| 4679 |
+
"learning_rate": 7.040732986527108e-05,
|
| 4680 |
+
"loss": 0.9724,
|
| 4681 |
+
"step": 664
|
| 4682 |
+
},
|
| 4683 |
+
{
|
| 4684 |
+
"epoch": 0.600451467268623,
|
| 4685 |
+
"grad_norm": 1.414829134941101,
|
| 4686 |
+
"learning_rate": 7.01341469262953e-05,
|
| 4687 |
+
"loss": 1.0056,
|
| 4688 |
+
"step": 665
|
| 4689 |
+
},
|
| 4690 |
+
{
|
| 4691 |
+
"epoch": 0.6013544018058691,
|
| 4692 |
+
"grad_norm": 2.0793347358703613,
|
| 4693 |
+
"learning_rate": 6.986120848215678e-05,
|
| 4694 |
+
"loss": 1.5461,
|
| 4695 |
+
"step": 666
|
| 4696 |
+
},
|
| 4697 |
+
{
|
| 4698 |
+
"epoch": 0.6022573363431152,
|
| 4699 |
+
"grad_norm": 1.2860586643218994,
|
| 4700 |
+
"learning_rate": 6.958851676724823e-05,
|
| 4701 |
+
"loss": 0.7181,
|
| 4702 |
+
"step": 667
|
| 4703 |
+
},
|
| 4704 |
+
{
|
| 4705 |
+
"epoch": 0.6031602708803612,
|
| 4706 |
+
"grad_norm": 1.9850184917449951,
|
| 4707 |
+
"learning_rate": 6.931607401394229e-05,
|
| 4708 |
+
"loss": 1.6305,
|
| 4709 |
+
"step": 668
|
| 4710 |
+
},
|
| 4711 |
+
{
|
| 4712 |
+
"epoch": 0.6040632054176073,
|
| 4713 |
+
"grad_norm": 1.7549471855163574,
|
| 4714 |
+
"learning_rate": 6.904388245257363e-05,
|
| 4715 |
+
"loss": 1.4215,
|
| 4716 |
+
"step": 669
|
| 4717 |
+
},
|
| 4718 |
+
{
|
| 4719 |
+
"epoch": 0.6049661399548533,
|
| 4720 |
+
"grad_norm": 1.667541742324829,
|
| 4721 |
+
"learning_rate": 6.877194431142055e-05,
|
| 4722 |
+
"loss": 1.323,
|
| 4723 |
+
"step": 670
|
| 4724 |
+
},
|
| 4725 |
+
{
|
| 4726 |
+
"epoch": 0.6058690744920994,
|
| 4727 |
+
"grad_norm": 1.388028860092163,
|
| 4728 |
+
"learning_rate": 6.850026181668668e-05,
|
| 4729 |
+
"loss": 0.9346,
|
| 4730 |
+
"step": 671
|
| 4731 |
+
},
|
| 4732 |
+
{
|
| 4733 |
+
"epoch": 0.6067720090293454,
|
| 4734 |
+
"grad_norm": 1.8620973825454712,
|
| 4735 |
+
"learning_rate": 6.822883719248283e-05,
|
| 4736 |
+
"loss": 1.5254,
|
| 4737 |
+
"step": 672
|
| 4738 |
+
},
|
| 4739 |
+
{
|
| 4740 |
+
"epoch": 0.6076749435665915,
|
| 4741 |
+
"grad_norm": 1.5007909536361694,
|
| 4742 |
+
"learning_rate": 6.79576726608089e-05,
|
| 4743 |
+
"loss": 1.1383,
|
| 4744 |
+
"step": 673
|
| 4745 |
+
},
|
| 4746 |
+
{
|
| 4747 |
+
"epoch": 0.6085778781038375,
|
| 4748 |
+
"grad_norm": 1.584130048751831,
|
| 4749 |
+
"learning_rate": 6.768677044153535e-05,
|
| 4750 |
+
"loss": 1.0465,
|
| 4751 |
+
"step": 674
|
| 4752 |
+
},
|
| 4753 |
+
{
|
| 4754 |
+
"epoch": 0.6094808126410836,
|
| 4755 |
+
"grad_norm": 1.9776028394699097,
|
| 4756 |
+
"learning_rate": 6.741613275238534e-05,
|
| 4757 |
+
"loss": 1.3916,
|
| 4758 |
+
"step": 675
|
| 4759 |
+
},
|
| 4760 |
+
{
|
| 4761 |
+
"epoch": 0.6103837471783295,
|
| 4762 |
+
"grad_norm": 2.143301248550415,
|
| 4763 |
+
"learning_rate": 6.714576180891654e-05,
|
| 4764 |
+
"loss": 1.4605,
|
| 4765 |
+
"step": 676
|
| 4766 |
+
},
|
| 4767 |
+
{
|
| 4768 |
+
"epoch": 0.6112866817155757,
|
| 4769 |
+
"grad_norm": 1.706349492073059,
|
| 4770 |
+
"learning_rate": 6.68756598245028e-05,
|
| 4771 |
+
"loss": 1.248,
|
| 4772 |
+
"step": 677
|
| 4773 |
+
},
|
| 4774 |
+
{
|
| 4775 |
+
"epoch": 0.6121896162528216,
|
| 4776 |
+
"grad_norm": 1.7645604610443115,
|
| 4777 |
+
"learning_rate": 6.660582901031621e-05,
|
| 4778 |
+
"loss": 1.3162,
|
| 4779 |
+
"step": 678
|
| 4780 |
+
},
|
| 4781 |
+
{
|
| 4782 |
+
"epoch": 0.6130925507900677,
|
| 4783 |
+
"grad_norm": 0.9834024906158447,
|
| 4784 |
+
"learning_rate": 6.633627157530899e-05,
|
| 4785 |
+
"loss": 0.4117,
|
| 4786 |
+
"step": 679
|
| 4787 |
+
},
|
| 4788 |
+
{
|
| 4789 |
+
"epoch": 0.6139954853273137,
|
| 4790 |
+
"grad_norm": 1.4688466787338257,
|
| 4791 |
+
"learning_rate": 6.606698972619526e-05,
|
| 4792 |
+
"loss": 0.9471,
|
| 4793 |
+
"step": 680
|
| 4794 |
+
},
|
| 4795 |
+
{
|
| 4796 |
+
"epoch": 0.6148984198645598,
|
| 4797 |
+
"grad_norm": 1.369699239730835,
|
| 4798 |
+
"learning_rate": 6.579798566743314e-05,
|
| 4799 |
+
"loss": 0.8842,
|
| 4800 |
+
"step": 681
|
| 4801 |
+
},
|
| 4802 |
+
{
|
| 4803 |
+
"epoch": 0.6158013544018058,
|
| 4804 |
+
"grad_norm": 1.4613699913024902,
|
| 4805 |
+
"learning_rate": 6.552926160120663e-05,
|
| 4806 |
+
"loss": 1.0846,
|
| 4807 |
+
"step": 682
|
| 4808 |
+
},
|
| 4809 |
+
{
|
| 4810 |
+
"epoch": 0.6167042889390519,
|
| 4811 |
+
"grad_norm": 1.527610182762146,
|
| 4812 |
+
"learning_rate": 6.526081972740758e-05,
|
| 4813 |
+
"loss": 1.0844,
|
| 4814 |
+
"step": 683
|
| 4815 |
+
},
|
| 4816 |
+
{
|
| 4817 |
+
"epoch": 0.6176072234762979,
|
| 4818 |
+
"grad_norm": 1.742908239364624,
|
| 4819 |
+
"learning_rate": 6.499266224361767e-05,
|
| 4820 |
+
"loss": 1.3409,
|
| 4821 |
+
"step": 684
|
| 4822 |
+
},
|
| 4823 |
+
{
|
| 4824 |
+
"epoch": 0.618510158013544,
|
| 4825 |
+
"grad_norm": 1.8235154151916504,
|
| 4826 |
+
"learning_rate": 6.472479134509052e-05,
|
| 4827 |
+
"loss": 1.308,
|
| 4828 |
+
"step": 685
|
| 4829 |
+
},
|
| 4830 |
+
{
|
| 4831 |
+
"epoch": 0.61941309255079,
|
| 4832 |
+
"grad_norm": 1.399803876876831,
|
| 4833 |
+
"learning_rate": 6.445720922473355e-05,
|
| 4834 |
+
"loss": 0.9267,
|
| 4835 |
+
"step": 686
|
| 4836 |
+
},
|
| 4837 |
+
{
|
| 4838 |
+
"epoch": 0.6203160270880361,
|
| 4839 |
+
"grad_norm": 1.8101701736450195,
|
| 4840 |
+
"learning_rate": 6.418991807309012e-05,
|
| 4841 |
+
"loss": 0.9509,
|
| 4842 |
+
"step": 687
|
| 4843 |
+
},
|
| 4844 |
+
{
|
| 4845 |
+
"epoch": 0.6212189616252821,
|
| 4846 |
+
"grad_norm": 1.5924134254455566,
|
| 4847 |
+
"learning_rate": 6.392292007832168e-05,
|
| 4848 |
+
"loss": 1.2864,
|
| 4849 |
+
"step": 688
|
| 4850 |
+
},
|
| 4851 |
+
{
|
| 4852 |
+
"epoch": 0.6221218961625282,
|
| 4853 |
+
"grad_norm": 1.2724056243896484,
|
| 4854 |
+
"learning_rate": 6.365621742618966e-05,
|
| 4855 |
+
"loss": 0.7186,
|
| 4856 |
+
"step": 689
|
| 4857 |
+
},
|
| 4858 |
+
{
|
| 4859 |
+
"epoch": 0.6230248306997742,
|
| 4860 |
+
"grad_norm": 1.3141125440597534,
|
| 4861 |
+
"learning_rate": 6.338981230003778e-05,
|
| 4862 |
+
"loss": 0.9604,
|
| 4863 |
+
"step": 690
|
| 4864 |
+
},
|
| 4865 |
+
{
|
| 4866 |
+
"epoch": 0.6239277652370203,
|
| 4867 |
+
"grad_norm": 1.656320571899414,
|
| 4868 |
+
"learning_rate": 6.312370688077399e-05,
|
| 4869 |
+
"loss": 1.2993,
|
| 4870 |
+
"step": 691
|
| 4871 |
+
},
|
| 4872 |
+
{
|
| 4873 |
+
"epoch": 0.6248306997742664,
|
| 4874 |
+
"grad_norm": 1.3529797792434692,
|
| 4875 |
+
"learning_rate": 6.285790334685282e-05,
|
| 4876 |
+
"loss": 0.4246,
|
| 4877 |
+
"step": 692
|
| 4878 |
+
},
|
| 4879 |
+
{
|
| 4880 |
+
"epoch": 0.6257336343115124,
|
| 4881 |
+
"grad_norm": 1.2153328657150269,
|
| 4882 |
+
"learning_rate": 6.259240387425736e-05,
|
| 4883 |
+
"loss": 0.9525,
|
| 4884 |
+
"step": 693
|
| 4885 |
+
},
|
| 4886 |
+
{
|
| 4887 |
+
"epoch": 0.6266365688487585,
|
| 4888 |
+
"grad_norm": 1.6419172286987305,
|
| 4889 |
+
"learning_rate": 6.232721063648148e-05,
|
| 4890 |
+
"loss": 0.9395,
|
| 4891 |
+
"step": 694
|
| 4892 |
+
},
|
| 4893 |
+
{
|
| 4894 |
+
"epoch": 0.6275395033860045,
|
| 4895 |
+
"grad_norm": 1.4371968507766724,
|
| 4896 |
+
"learning_rate": 6.206232580451225e-05,
|
| 4897 |
+
"loss": 1.4503,
|
| 4898 |
+
"step": 695
|
| 4899 |
+
},
|
| 4900 |
+
{
|
| 4901 |
+
"epoch": 0.6284424379232506,
|
| 4902 |
+
"grad_norm": 2.1209614276885986,
|
| 4903 |
+
"learning_rate": 6.179775154681184e-05,
|
| 4904 |
+
"loss": 1.5918,
|
| 4905 |
+
"step": 696
|
| 4906 |
+
},
|
| 4907 |
+
{
|
| 4908 |
+
"epoch": 0.6293453724604966,
|
| 4909 |
+
"grad_norm": 1.8328834772109985,
|
| 4910 |
+
"learning_rate": 6.153349002929987e-05,
|
| 4911 |
+
"loss": 1.1838,
|
| 4912 |
+
"step": 697
|
| 4913 |
+
},
|
| 4914 |
+
{
|
| 4915 |
+
"epoch": 0.6302483069977427,
|
| 4916 |
+
"grad_norm": 1.7110613584518433,
|
| 4917 |
+
"learning_rate": 6.126954341533599e-05,
|
| 4918 |
+
"loss": 1.4122,
|
| 4919 |
+
"step": 698
|
| 4920 |
+
},
|
| 4921 |
+
{
|
| 4922 |
+
"epoch": 0.6311512415349887,
|
| 4923 |
+
"grad_norm": 1.826737642288208,
|
| 4924 |
+
"learning_rate": 6.100591386570167e-05,
|
| 4925 |
+
"loss": 1.5058,
|
| 4926 |
+
"step": 699
|
| 4927 |
+
},
|
| 4928 |
+
{
|
| 4929 |
+
"epoch": 0.6320541760722348,
|
| 4930 |
+
"grad_norm": 1.8801661729812622,
|
| 4931 |
+
"learning_rate": 6.0742603538582835e-05,
|
| 4932 |
+
"loss": 1.5475,
|
| 4933 |
+
"step": 700
|
| 4934 |
+
},
|
| 4935 |
+
{
|
| 4936 |
+
"epoch": 0.6329571106094808,
|
| 4937 |
+
"grad_norm": 1.763385534286499,
|
| 4938 |
+
"learning_rate": 6.047961458955214e-05,
|
| 4939 |
+
"loss": 1.3673,
|
| 4940 |
+
"step": 701
|
| 4941 |
+
},
|
| 4942 |
+
{
|
| 4943 |
+
"epoch": 0.6338600451467269,
|
| 4944 |
+
"grad_norm": 1.508298397064209,
|
| 4945 |
+
"learning_rate": 6.021694917155129e-05,
|
| 4946 |
+
"loss": 1.1175,
|
| 4947 |
+
"step": 702
|
| 4948 |
+
},
|
| 4949 |
+
{
|
| 4950 |
+
"epoch": 0.6347629796839729,
|
| 4951 |
+
"grad_norm": 1.5054030418395996,
|
| 4952 |
+
"learning_rate": 5.9954609434873344e-05,
|
| 4953 |
+
"loss": 1.32,
|
| 4954 |
+
"step": 703
|
| 4955 |
+
},
|
| 4956 |
+
{
|
| 4957 |
+
"epoch": 0.635665914221219,
|
| 4958 |
+
"grad_norm": 1.6304901838302612,
|
| 4959 |
+
"learning_rate": 5.9692597527145354e-05,
|
| 4960 |
+
"loss": 0.9387,
|
| 4961 |
+
"step": 704
|
| 4962 |
+
},
|
| 4963 |
+
{
|
| 4964 |
+
"epoch": 0.636568848758465,
|
| 4965 |
+
"grad_norm": 1.3840839862823486,
|
| 4966 |
+
"learning_rate": 5.943091559331053e-05,
|
| 4967 |
+
"loss": 0.627,
|
| 4968 |
+
"step": 705
|
| 4969 |
+
},
|
| 4970 |
+
{
|
| 4971 |
+
"epoch": 0.6374717832957111,
|
| 4972 |
+
"grad_norm": 1.8724223375320435,
|
| 4973 |
+
"learning_rate": 5.9169565775610656e-05,
|
| 4974 |
+
"loss": 1.6199,
|
| 4975 |
+
"step": 706
|
| 4976 |
+
},
|
| 4977 |
+
{
|
| 4978 |
+
"epoch": 0.6383747178329571,
|
| 4979 |
+
"grad_norm": 1.563077449798584,
|
| 4980 |
+
"learning_rate": 5.890855021356891e-05,
|
| 4981 |
+
"loss": 1.1355,
|
| 4982 |
+
"step": 707
|
| 4983 |
+
},
|
| 4984 |
+
{
|
| 4985 |
+
"epoch": 0.6392776523702032,
|
| 4986 |
+
"grad_norm": 1.436528205871582,
|
| 4987 |
+
"learning_rate": 5.864787104397194e-05,
|
| 4988 |
+
"loss": 1.3004,
|
| 4989 |
+
"step": 708
|
| 4990 |
+
},
|
| 4991 |
+
{
|
| 4992 |
+
"epoch": 0.6401805869074492,
|
| 4993 |
+
"grad_norm": 1.5161725282669067,
|
| 4994 |
+
"learning_rate": 5.838753040085256e-05,
|
| 4995 |
+
"loss": 0.8622,
|
| 4996 |
+
"step": 709
|
| 4997 |
+
},
|
| 4998 |
+
{
|
| 4999 |
+
"epoch": 0.6410835214446953,
|
| 5000 |
+
"grad_norm": 1.5266764163970947,
|
| 5001 |
+
"learning_rate": 5.81275304154723e-05,
|
| 5002 |
+
"loss": 0.9748,
|
| 5003 |
+
"step": 710
|
| 5004 |
+
},
|
| 5005 |
+
{
|
| 5006 |
+
"epoch": 0.6419864559819413,
|
| 5007 |
+
"grad_norm": 1.5332485437393188,
|
| 5008 |
+
"learning_rate": 5.786787321630394e-05,
|
| 5009 |
+
"loss": 0.8654,
|
| 5010 |
+
"step": 711
|
| 5011 |
+
},
|
| 5012 |
+
{
|
| 5013 |
+
"epoch": 0.6428893905191874,
|
| 5014 |
+
"grad_norm": 1.5081924200057983,
|
| 5015 |
+
"learning_rate": 5.7608560929013946e-05,
|
| 5016 |
+
"loss": 1.2023,
|
| 5017 |
+
"step": 712
|
| 5018 |
+
},
|
| 5019 |
+
{
|
| 5020 |
+
"epoch": 0.6437923250564334,
|
| 5021 |
+
"grad_norm": 1.880600094795227,
|
| 5022 |
+
"learning_rate": 5.7349595676445286e-05,
|
| 5023 |
+
"loss": 1.3637,
|
| 5024 |
+
"step": 713
|
| 5025 |
+
},
|
| 5026 |
+
{
|
| 5027 |
+
"epoch": 0.6446952595936795,
|
| 5028 |
+
"grad_norm": 1.742201328277588,
|
| 5029 |
+
"learning_rate": 5.709097957860001e-05,
|
| 5030 |
+
"loss": 1.4513,
|
| 5031 |
+
"step": 714
|
| 5032 |
+
},
|
| 5033 |
+
{
|
| 5034 |
+
"epoch": 0.6455981941309256,
|
| 5035 |
+
"grad_norm": 1.4408034086227417,
|
| 5036 |
+
"learning_rate": 5.683271475262164e-05,
|
| 5037 |
+
"loss": 1.0124,
|
| 5038 |
+
"step": 715
|
| 5039 |
+
},
|
| 5040 |
+
{
|
| 5041 |
+
"epoch": 0.6465011286681716,
|
| 5042 |
+
"grad_norm": 1.2453595399856567,
|
| 5043 |
+
"learning_rate": 5.657480331277819e-05,
|
| 5044 |
+
"loss": 0.908,
|
| 5045 |
+
"step": 716
|
| 5046 |
+
},
|
| 5047 |
+
{
|
| 5048 |
+
"epoch": 0.6474040632054177,
|
| 5049 |
+
"grad_norm": 1.4883902072906494,
|
| 5050 |
+
"learning_rate": 5.6317247370444715e-05,
|
| 5051 |
+
"loss": 1.15,
|
| 5052 |
+
"step": 717
|
| 5053 |
+
},
|
| 5054 |
+
{
|
| 5055 |
+
"epoch": 0.6483069977426636,
|
| 5056 |
+
"grad_norm": 1.3922057151794434,
|
| 5057 |
+
"learning_rate": 5.6060049034085815e-05,
|
| 5058 |
+
"loss": 1.2913,
|
| 5059 |
+
"step": 718
|
| 5060 |
+
},
|
| 5061 |
+
{
|
| 5062 |
+
"epoch": 0.6492099322799098,
|
| 5063 |
+
"grad_norm": 1.6199557781219482,
|
| 5064 |
+
"learning_rate": 5.58032104092389e-05,
|
| 5065 |
+
"loss": 1.0053,
|
| 5066 |
+
"step": 719
|
| 5067 |
+
},
|
| 5068 |
+
{
|
| 5069 |
+
"epoch": 0.6501128668171557,
|
| 5070 |
+
"grad_norm": 1.8710253238677979,
|
| 5071 |
+
"learning_rate": 5.5546733598496314e-05,
|
| 5072 |
+
"loss": 1.4156,
|
| 5073 |
+
"step": 720
|
| 5074 |
+
},
|
| 5075 |
+
{
|
| 5076 |
+
"epoch": 0.6510158013544018,
|
| 5077 |
+
"grad_norm": 1.3759031295776367,
|
| 5078 |
+
"learning_rate": 5.5290620701488594e-05,
|
| 5079 |
+
"loss": 0.9242,
|
| 5080 |
+
"step": 721
|
| 5081 |
+
},
|
| 5082 |
+
{
|
| 5083 |
+
"epoch": 0.6519187358916478,
|
| 5084 |
+
"grad_norm": 1.869899034500122,
|
| 5085 |
+
"learning_rate": 5.5034873814867125e-05,
|
| 5086 |
+
"loss": 1.4667,
|
| 5087 |
+
"step": 722
|
| 5088 |
+
},
|
| 5089 |
+
{
|
| 5090 |
+
"epoch": 0.6528216704288939,
|
| 5091 |
+
"grad_norm": 1.673433780670166,
|
| 5092 |
+
"learning_rate": 5.477949503228686e-05,
|
| 5093 |
+
"loss": 1.4162,
|
| 5094 |
+
"step": 723
|
| 5095 |
+
},
|
| 5096 |
+
{
|
| 5097 |
+
"epoch": 0.6537246049661399,
|
| 5098 |
+
"grad_norm": 1.4881922006607056,
|
| 5099 |
+
"learning_rate": 5.452448644438946e-05,
|
| 5100 |
+
"loss": 1.4349,
|
| 5101 |
+
"step": 724
|
| 5102 |
+
},
|
| 5103 |
+
{
|
| 5104 |
+
"epoch": 0.654627539503386,
|
| 5105 |
+
"grad_norm": 1.577567219734192,
|
| 5106 |
+
"learning_rate": 5.42698501387858e-05,
|
| 5107 |
+
"loss": 1.3597,
|
| 5108 |
+
"step": 725
|
| 5109 |
+
},
|
| 5110 |
+
{
|
| 5111 |
+
"epoch": 0.655530474040632,
|
| 5112 |
+
"grad_norm": 1.6177291870117188,
|
| 5113 |
+
"learning_rate": 5.401558820003929e-05,
|
| 5114 |
+
"loss": 1.5005,
|
| 5115 |
+
"step": 726
|
| 5116 |
+
},
|
| 5117 |
+
{
|
| 5118 |
+
"epoch": 0.6564334085778781,
|
| 5119 |
+
"grad_norm": 1.8385038375854492,
|
| 5120 |
+
"learning_rate": 5.3761702709648556e-05,
|
| 5121 |
+
"loss": 1.3802,
|
| 5122 |
+
"step": 727
|
| 5123 |
+
},
|
| 5124 |
+
{
|
| 5125 |
+
"epoch": 0.6573363431151241,
|
| 5126 |
+
"grad_norm": 1.299546718597412,
|
| 5127 |
+
"learning_rate": 5.3508195746030296e-05,
|
| 5128 |
+
"loss": 1.1285,
|
| 5129 |
+
"step": 728
|
| 5130 |
+
},
|
| 5131 |
+
{
|
| 5132 |
+
"epoch": 0.6582392776523702,
|
| 5133 |
+
"grad_norm": 1.2800817489624023,
|
| 5134 |
+
"learning_rate": 5.325506938450273e-05,
|
| 5135 |
+
"loss": 0.8695,
|
| 5136 |
+
"step": 729
|
| 5137 |
+
},
|
| 5138 |
+
{
|
| 5139 |
+
"epoch": 0.6591422121896162,
|
| 5140 |
+
"grad_norm": 1.4967060089111328,
|
| 5141 |
+
"learning_rate": 5.300232569726804e-05,
|
| 5142 |
+
"loss": 1.4879,
|
| 5143 |
+
"step": 730
|
| 5144 |
+
},
|
| 5145 |
+
{
|
| 5146 |
+
"epoch": 0.6600451467268623,
|
| 5147 |
+
"grad_norm": 1.702681064605713,
|
| 5148 |
+
"learning_rate": 5.2749966753395686e-05,
|
| 5149 |
+
"loss": 1.2565,
|
| 5150 |
+
"step": 731
|
| 5151 |
+
},
|
| 5152 |
+
{
|
| 5153 |
+
"epoch": 0.6609480812641083,
|
| 5154 |
+
"grad_norm": 1.6170333623886108,
|
| 5155 |
+
"learning_rate": 5.249799461880569e-05,
|
| 5156 |
+
"loss": 1.2811,
|
| 5157 |
+
"step": 732
|
| 5158 |
+
},
|
| 5159 |
+
{
|
| 5160 |
+
"epoch": 0.6618510158013544,
|
| 5161 |
+
"grad_norm": 1.6071534156799316,
|
| 5162 |
+
"learning_rate": 5.224641135625119e-05,
|
| 5163 |
+
"loss": 1.3454,
|
| 5164 |
+
"step": 733
|
| 5165 |
+
},
|
| 5166 |
+
{
|
| 5167 |
+
"epoch": 0.6627539503386004,
|
| 5168 |
+
"grad_norm": 1.8343545198440552,
|
| 5169 |
+
"learning_rate": 5.1995219025302025e-05,
|
| 5170 |
+
"loss": 1.6234,
|
| 5171 |
+
"step": 734
|
| 5172 |
+
},
|
| 5173 |
+
{
|
| 5174 |
+
"epoch": 0.6636568848758465,
|
| 5175 |
+
"grad_norm": 1.4651812314987183,
|
| 5176 |
+
"learning_rate": 5.174441968232768e-05,
|
| 5177 |
+
"loss": 1.2592,
|
| 5178 |
+
"step": 735
|
| 5179 |
+
},
|
| 5180 |
+
{
|
| 5181 |
+
"epoch": 0.6645598194130925,
|
| 5182 |
+
"grad_norm": 1.6615885496139526,
|
| 5183 |
+
"learning_rate": 5.1494015380480396e-05,
|
| 5184 |
+
"loss": 0.9647,
|
| 5185 |
+
"step": 736
|
| 5186 |
+
},
|
| 5187 |
+
{
|
| 5188 |
+
"epoch": 0.6654627539503386,
|
| 5189 |
+
"grad_norm": 1.6668320894241333,
|
| 5190 |
+
"learning_rate": 5.1244008169678495e-05,
|
| 5191 |
+
"loss": 0.8748,
|
| 5192 |
+
"step": 737
|
| 5193 |
+
},
|
| 5194 |
+
{
|
| 5195 |
+
"epoch": 0.6663656884875846,
|
| 5196 |
+
"grad_norm": 1.682271122932434,
|
| 5197 |
+
"learning_rate": 5.099440009658957e-05,
|
| 5198 |
+
"loss": 1.2871,
|
| 5199 |
+
"step": 738
|
| 5200 |
+
},
|
| 5201 |
+
{
|
| 5202 |
+
"epoch": 0.6672686230248307,
|
| 5203 |
+
"grad_norm": 1.3818674087524414,
|
| 5204 |
+
"learning_rate": 5.074519320461357e-05,
|
| 5205 |
+
"loss": 1.0104,
|
| 5206 |
+
"step": 739
|
| 5207 |
+
},
|
| 5208 |
+
{
|
| 5209 |
+
"epoch": 0.6681715575620768,
|
| 5210 |
+
"grad_norm": 1.4028613567352295,
|
| 5211 |
+
"learning_rate": 5.049638953386635e-05,
|
| 5212 |
+
"loss": 1.2135,
|
| 5213 |
+
"step": 740
|
| 5214 |
+
},
|
| 5215 |
+
{
|
| 5216 |
+
"epoch": 0.6690744920993228,
|
| 5217 |
+
"grad_norm": 1.6344983577728271,
|
| 5218 |
+
"learning_rate": 5.024799112116274e-05,
|
| 5219 |
+
"loss": 1.1836,
|
| 5220 |
+
"step": 741
|
| 5221 |
+
},
|
| 5222 |
+
{
|
| 5223 |
+
"epoch": 0.6699774266365689,
|
| 5224 |
+
"grad_norm": 1.5904343128204346,
|
| 5225 |
+
"learning_rate": 5.000000000000002e-05,
|
| 5226 |
+
"loss": 1.2949,
|
| 5227 |
+
"step": 742
|
| 5228 |
+
},
|
| 5229 |
+
{
|
| 5230 |
+
"epoch": 0.6708803611738149,
|
| 5231 |
+
"grad_norm": 1.2303252220153809,
|
| 5232 |
+
"learning_rate": 4.975241820054107e-05,
|
| 5233 |
+
"loss": 0.8138,
|
| 5234 |
+
"step": 743
|
| 5235 |
+
},
|
| 5236 |
+
{
|
| 5237 |
+
"epoch": 0.671783295711061,
|
| 5238 |
+
"grad_norm": 1.8008294105529785,
|
| 5239 |
+
"learning_rate": 4.9505247749598006e-05,
|
| 5240 |
+
"loss": 1.9048,
|
| 5241 |
+
"step": 744
|
| 5242 |
+
},
|
| 5243 |
+
{
|
| 5244 |
+
"epoch": 0.672686230248307,
|
| 5245 |
+
"grad_norm": 1.6088261604309082,
|
| 5246 |
+
"learning_rate": 4.9258490670615475e-05,
|
| 5247 |
+
"loss": 1.6434,
|
| 5248 |
+
"step": 745
|
| 5249 |
+
},
|
| 5250 |
+
{
|
| 5251 |
+
"epoch": 0.6735891647855531,
|
| 5252 |
+
"grad_norm": 1.501248836517334,
|
| 5253 |
+
"learning_rate": 4.9012148983653964e-05,
|
| 5254 |
+
"loss": 1.176,
|
| 5255 |
+
"step": 746
|
| 5256 |
+
},
|
| 5257 |
+
{
|
| 5258 |
+
"epoch": 0.6744920993227991,
|
| 5259 |
+
"grad_norm": 1.6089529991149902,
|
| 5260 |
+
"learning_rate": 4.8766224705373476e-05,
|
| 5261 |
+
"loss": 1.0786,
|
| 5262 |
+
"step": 747
|
| 5263 |
+
},
|
| 5264 |
+
{
|
| 5265 |
+
"epoch": 0.6753950338600452,
|
| 5266 |
+
"grad_norm": 1.3817884922027588,
|
| 5267 |
+
"learning_rate": 4.852071984901696e-05,
|
| 5268 |
+
"loss": 0.8196,
|
| 5269 |
+
"step": 748
|
| 5270 |
+
},
|
| 5271 |
+
{
|
| 5272 |
+
"epoch": 0.6762979683972912,
|
| 5273 |
+
"grad_norm": 1.625156044960022,
|
| 5274 |
+
"learning_rate": 4.827563642439361e-05,
|
| 5275 |
+
"loss": 1.337,
|
| 5276 |
+
"step": 749
|
| 5277 |
+
},
|
| 5278 |
+
{
|
| 5279 |
+
"epoch": 0.6772009029345373,
|
| 5280 |
+
"grad_norm": 1.7386070489883423,
|
| 5281 |
+
"learning_rate": 4.803097643786288e-05,
|
| 5282 |
+
"loss": 1.5089,
|
| 5283 |
+
"step": 750
|
| 5284 |
+
},
|
| 5285 |
+
{
|
| 5286 |
+
"epoch": 0.6781038374717833,
|
| 5287 |
+
"grad_norm": 1.3682126998901367,
|
| 5288 |
+
"learning_rate": 4.778674189231751e-05,
|
| 5289 |
+
"loss": 0.9877,
|
| 5290 |
+
"step": 751
|
| 5291 |
+
},
|
| 5292 |
+
{
|
| 5293 |
+
"epoch": 0.6790067720090294,
|
| 5294 |
+
"grad_norm": 1.2330286502838135,
|
| 5295 |
+
"learning_rate": 4.754293478716755e-05,
|
| 5296 |
+
"loss": 0.864,
|
| 5297 |
+
"step": 752
|
| 5298 |
+
},
|
| 5299 |
+
{
|
| 5300 |
+
"epoch": 0.6799097065462754,
|
| 5301 |
+
"grad_norm": 1.4554107189178467,
|
| 5302 |
+
"learning_rate": 4.7299557118323814e-05,
|
| 5303 |
+
"loss": 1.3097,
|
| 5304 |
+
"step": 753
|
| 5305 |
+
},
|
| 5306 |
+
{
|
| 5307 |
+
"epoch": 0.6808126410835215,
|
| 5308 |
+
"grad_norm": 1.650094985961914,
|
| 5309 |
+
"learning_rate": 4.7056610878181486e-05,
|
| 5310 |
+
"loss": 1.2152,
|
| 5311 |
+
"step": 754
|
| 5312 |
+
},
|
| 5313 |
+
{
|
| 5314 |
+
"epoch": 0.6817155756207675,
|
| 5315 |
+
"grad_norm": 1.435055136680603,
|
| 5316 |
+
"learning_rate": 4.681409805560396e-05,
|
| 5317 |
+
"loss": 0.6956,
|
| 5318 |
+
"step": 755
|
| 5319 |
+
},
|
| 5320 |
+
{
|
| 5321 |
+
"epoch": 0.6826185101580136,
|
| 5322 |
+
"grad_norm": 1.376915454864502,
|
| 5323 |
+
"learning_rate": 4.657202063590653e-05,
|
| 5324 |
+
"loss": 0.8902,
|
| 5325 |
+
"step": 756
|
| 5326 |
+
},
|
| 5327 |
+
{
|
| 5328 |
+
"epoch": 0.6835214446952596,
|
| 5329 |
+
"grad_norm": 1.2419427633285522,
|
| 5330 |
+
"learning_rate": 4.633038060083996e-05,
|
| 5331 |
+
"loss": 0.6594,
|
| 5332 |
+
"step": 757
|
| 5333 |
+
},
|
| 5334 |
+
{
|
| 5335 |
+
"epoch": 0.6844243792325057,
|
| 5336 |
+
"grad_norm": 1.9876960515975952,
|
| 5337 |
+
"learning_rate": 4.6089179928574487e-05,
|
| 5338 |
+
"loss": 1.1539,
|
| 5339 |
+
"step": 758
|
| 5340 |
+
},
|
| 5341 |
+
{
|
| 5342 |
+
"epoch": 0.6853273137697516,
|
| 5343 |
+
"grad_norm": 1.5876964330673218,
|
| 5344 |
+
"learning_rate": 4.584842059368354e-05,
|
| 5345 |
+
"loss": 1.0687,
|
| 5346 |
+
"step": 759
|
| 5347 |
+
},
|
| 5348 |
+
{
|
| 5349 |
+
"epoch": 0.6862302483069977,
|
| 5350 |
+
"grad_norm": 1.317826271057129,
|
| 5351 |
+
"learning_rate": 4.560810456712754e-05,
|
| 5352 |
+
"loss": 0.9303,
|
| 5353 |
+
"step": 760
|
| 5354 |
+
},
|
| 5355 |
+
{
|
| 5356 |
+
"epoch": 0.6871331828442437,
|
| 5357 |
+
"grad_norm": 1.5974258184432983,
|
| 5358 |
+
"learning_rate": 4.536823381623779e-05,
|
| 5359 |
+
"loss": 1.2335,
|
| 5360 |
+
"step": 761
|
| 5361 |
+
},
|
| 5362 |
+
{
|
| 5363 |
+
"epoch": 0.6880361173814898,
|
| 5364 |
+
"grad_norm": 1.6230992078781128,
|
| 5365 |
+
"learning_rate": 4.512881030470031e-05,
|
| 5366 |
+
"loss": 1.3595,
|
| 5367 |
+
"step": 762
|
| 5368 |
+
},
|
| 5369 |
+
{
|
| 5370 |
+
"epoch": 0.688939051918736,
|
| 5371 |
+
"grad_norm": 1.4521753787994385,
|
| 5372 |
+
"learning_rate": 4.488983599254001e-05,
|
| 5373 |
+
"loss": 1.0362,
|
| 5374 |
+
"step": 763
|
| 5375 |
+
},
|
| 5376 |
+
{
|
| 5377 |
+
"epoch": 0.6898419864559819,
|
| 5378 |
+
"grad_norm": 1.566362977027893,
|
| 5379 |
+
"learning_rate": 4.4651312836104254e-05,
|
| 5380 |
+
"loss": 1.2261,
|
| 5381 |
+
"step": 764
|
| 5382 |
+
},
|
| 5383 |
+
{
|
| 5384 |
+
"epoch": 0.690744920993228,
|
| 5385 |
+
"grad_norm": 1.9560340642929077,
|
| 5386 |
+
"learning_rate": 4.441324278804717e-05,
|
| 5387 |
+
"loss": 1.4274,
|
| 5388 |
+
"step": 765
|
| 5389 |
+
},
|
| 5390 |
+
{
|
| 5391 |
+
"epoch": 0.691647855530474,
|
| 5392 |
+
"grad_norm": 1.6547048091888428,
|
| 5393 |
+
"learning_rate": 4.417562779731355e-05,
|
| 5394 |
+
"loss": 0.8835,
|
| 5395 |
+
"step": 766
|
| 5396 |
+
},
|
| 5397 |
+
{
|
| 5398 |
+
"epoch": 0.6925507900677201,
|
| 5399 |
+
"grad_norm": 1.3919938802719116,
|
| 5400 |
+
"learning_rate": 4.3938469809122776e-05,
|
| 5401 |
+
"loss": 0.776,
|
| 5402 |
+
"step": 767
|
| 5403 |
+
},
|
| 5404 |
+
{
|
| 5405 |
+
"epoch": 0.6934537246049661,
|
| 5406 |
+
"grad_norm": 1.4622743129730225,
|
| 5407 |
+
"learning_rate": 4.3701770764953124e-05,
|
| 5408 |
+
"loss": 0.9917,
|
| 5409 |
+
"step": 768
|
| 5410 |
+
},
|
| 5411 |
+
{
|
| 5412 |
+
"epoch": 0.6943566591422122,
|
| 5413 |
+
"grad_norm": 1.9236985445022583,
|
| 5414 |
+
"learning_rate": 4.346553260252574e-05,
|
| 5415 |
+
"loss": 1.7837,
|
| 5416 |
+
"step": 769
|
| 5417 |
+
},
|
| 5418 |
+
{
|
| 5419 |
+
"epoch": 0.6952595936794582,
|
| 5420 |
+
"grad_norm": 1.6482877731323242,
|
| 5421 |
+
"learning_rate": 4.32297572557887e-05,
|
| 5422 |
+
"loss": 1.66,
|
| 5423 |
+
"step": 770
|
| 5424 |
+
},
|
| 5425 |
+
{
|
| 5426 |
+
"epoch": 0.6961625282167043,
|
| 5427 |
+
"grad_norm": 1.2067683935165405,
|
| 5428 |
+
"learning_rate": 4.299444665490139e-05,
|
| 5429 |
+
"loss": 0.7246,
|
| 5430 |
+
"step": 771
|
| 5431 |
+
},
|
| 5432 |
+
{
|
| 5433 |
+
"epoch": 0.6970654627539503,
|
| 5434 |
+
"grad_norm": 1.3889422416687012,
|
| 5435 |
+
"learning_rate": 4.275960272621852e-05,
|
| 5436 |
+
"loss": 0.7338,
|
| 5437 |
+
"step": 772
|
| 5438 |
+
},
|
| 5439 |
+
{
|
| 5440 |
+
"epoch": 0.6979683972911964,
|
| 5441 |
+
"grad_norm": 1.8344866037368774,
|
| 5442 |
+
"learning_rate": 4.252522739227445e-05,
|
| 5443 |
+
"loss": 1.1381,
|
| 5444 |
+
"step": 773
|
| 5445 |
+
},
|
| 5446 |
+
{
|
| 5447 |
+
"epoch": 0.6988713318284424,
|
| 5448 |
+
"grad_norm": 1.595993161201477,
|
| 5449 |
+
"learning_rate": 4.2291322571767344e-05,
|
| 5450 |
+
"loss": 0.9127,
|
| 5451 |
+
"step": 774
|
| 5452 |
+
},
|
| 5453 |
+
{
|
| 5454 |
+
"epoch": 0.6997742663656885,
|
| 5455 |
+
"grad_norm": 1.143455982208252,
|
| 5456 |
+
"learning_rate": 4.205789017954364e-05,
|
| 5457 |
+
"loss": 0.4581,
|
| 5458 |
+
"step": 775
|
| 5459 |
+
},
|
| 5460 |
+
{
|
| 5461 |
+
"epoch": 0.7006772009029345,
|
| 5462 |
+
"grad_norm": 1.399437427520752,
|
| 5463 |
+
"learning_rate": 4.182493212658224e-05,
|
| 5464 |
+
"loss": 0.7188,
|
| 5465 |
+
"step": 776
|
| 5466 |
+
},
|
| 5467 |
+
{
|
| 5468 |
+
"epoch": 0.7015801354401806,
|
| 5469 |
+
"grad_norm": 1.8559757471084595,
|
| 5470 |
+
"learning_rate": 4.159245031997881e-05,
|
| 5471 |
+
"loss": 1.1238,
|
| 5472 |
+
"step": 777
|
| 5473 |
+
},
|
| 5474 |
+
{
|
| 5475 |
+
"epoch": 0.7024830699774266,
|
| 5476 |
+
"grad_norm": 1.4119492769241333,
|
| 5477 |
+
"learning_rate": 4.136044666293044e-05,
|
| 5478 |
+
"loss": 1.0352,
|
| 5479 |
+
"step": 778
|
| 5480 |
+
},
|
| 5481 |
+
{
|
| 5482 |
+
"epoch": 0.7033860045146727,
|
| 5483 |
+
"grad_norm": 1.46328604221344,
|
| 5484 |
+
"learning_rate": 4.112892305471974e-05,
|
| 5485 |
+
"loss": 1.0075,
|
| 5486 |
+
"step": 779
|
| 5487 |
+
},
|
| 5488 |
+
{
|
| 5489 |
+
"epoch": 0.7042889390519187,
|
| 5490 |
+
"grad_norm": 1.7197778224945068,
|
| 5491 |
+
"learning_rate": 4.0897881390699356e-05,
|
| 5492 |
+
"loss": 1.067,
|
| 5493 |
+
"step": 780
|
| 5494 |
+
},
|
| 5495 |
+
{
|
| 5496 |
+
"epoch": 0.7051918735891648,
|
| 5497 |
+
"grad_norm": 1.2641351222991943,
|
| 5498 |
+
"learning_rate": 4.0667323562276814e-05,
|
| 5499 |
+
"loss": 0.6254,
|
| 5500 |
+
"step": 781
|
| 5501 |
+
},
|
| 5502 |
+
{
|
| 5503 |
+
"epoch": 0.7060948081264108,
|
| 5504 |
+
"grad_norm": 1.173400640487671,
|
| 5505 |
+
"learning_rate": 4.043725145689846e-05,
|
| 5506 |
+
"loss": 0.8572,
|
| 5507 |
+
"step": 782
|
| 5508 |
+
},
|
| 5509 |
+
{
|
| 5510 |
+
"epoch": 0.7069977426636569,
|
| 5511 |
+
"grad_norm": 1.3645060062408447,
|
| 5512 |
+
"learning_rate": 4.0207666958034465e-05,
|
| 5513 |
+
"loss": 1.0241,
|
| 5514 |
+
"step": 783
|
| 5515 |
+
},
|
| 5516 |
+
{
|
| 5517 |
+
"epoch": 0.7079006772009029,
|
| 5518 |
+
"grad_norm": 1.5815194845199585,
|
| 5519 |
+
"learning_rate": 3.997857194516319e-05,
|
| 5520 |
+
"loss": 1.0335,
|
| 5521 |
+
"step": 784
|
| 5522 |
+
},
|
| 5523 |
+
{
|
| 5524 |
+
"epoch": 0.708803611738149,
|
| 5525 |
+
"grad_norm": 2.148763418197632,
|
| 5526 |
+
"learning_rate": 3.9749968293755834e-05,
|
| 5527 |
+
"loss": 1.6093,
|
| 5528 |
+
"step": 785
|
| 5529 |
+
},
|
| 5530 |
+
{
|
| 5531 |
+
"epoch": 0.7097065462753951,
|
| 5532 |
+
"grad_norm": 1.3801579475402832,
|
| 5533 |
+
"learning_rate": 3.9521857875261114e-05,
|
| 5534 |
+
"loss": 0.912,
|
| 5535 |
+
"step": 786
|
| 5536 |
+
},
|
| 5537 |
+
{
|
| 5538 |
+
"epoch": 0.7106094808126411,
|
| 5539 |
+
"grad_norm": 1.2974218130111694,
|
| 5540 |
+
"learning_rate": 3.929424255708999e-05,
|
| 5541 |
+
"loss": 0.7171,
|
| 5542 |
+
"step": 787
|
| 5543 |
+
},
|
| 5544 |
+
{
|
| 5545 |
+
"epoch": 0.7115124153498872,
|
| 5546 |
+
"grad_norm": 2.0133774280548096,
|
| 5547 |
+
"learning_rate": 3.9067124202600194e-05,
|
| 5548 |
+
"loss": 1.4236,
|
| 5549 |
+
"step": 788
|
| 5550 |
+
},
|
| 5551 |
+
{
|
| 5552 |
+
"epoch": 0.7124153498871332,
|
| 5553 |
+
"grad_norm": 1.2373929023742676,
|
| 5554 |
+
"learning_rate": 3.884050467108117e-05,
|
| 5555 |
+
"loss": 0.69,
|
| 5556 |
+
"step": 789
|
| 5557 |
+
},
|
| 5558 |
+
{
|
| 5559 |
+
"epoch": 0.7133182844243793,
|
| 5560 |
+
"grad_norm": 1.5660927295684814,
|
| 5561 |
+
"learning_rate": 3.8614385817738794e-05,
|
| 5562 |
+
"loss": 1.1102,
|
| 5563 |
+
"step": 790
|
| 5564 |
+
},
|
| 5565 |
+
{
|
| 5566 |
+
"epoch": 0.7142212189616253,
|
| 5567 |
+
"grad_norm": 1.4448158740997314,
|
| 5568 |
+
"learning_rate": 3.83887694936802e-05,
|
| 5569 |
+
"loss": 0.8871,
|
| 5570 |
+
"step": 791
|
| 5571 |
+
},
|
| 5572 |
+
{
|
| 5573 |
+
"epoch": 0.7151241534988714,
|
| 5574 |
+
"grad_norm": 1.6486666202545166,
|
| 5575 |
+
"learning_rate": 3.816365754589845e-05,
|
| 5576 |
+
"loss": 1.2077,
|
| 5577 |
+
"step": 792
|
| 5578 |
+
},
|
| 5579 |
+
{
|
| 5580 |
+
"epoch": 0.7160270880361174,
|
| 5581 |
+
"grad_norm": 2.0939183235168457,
|
| 5582 |
+
"learning_rate": 3.793905181725772e-05,
|
| 5583 |
+
"loss": 1.3465,
|
| 5584 |
+
"step": 793
|
| 5585 |
+
},
|
| 5586 |
+
{
|
| 5587 |
+
"epoch": 0.7169300225733635,
|
| 5588 |
+
"grad_norm": 1.6600865125656128,
|
| 5589 |
+
"learning_rate": 3.771495414647802e-05,
|
| 5590 |
+
"loss": 0.9234,
|
| 5591 |
+
"step": 794
|
| 5592 |
+
},
|
| 5593 |
+
{
|
| 5594 |
+
"epoch": 0.7178329571106095,
|
| 5595 |
+
"grad_norm": 1.938519835472107,
|
| 5596 |
+
"learning_rate": 3.7491366368120104e-05,
|
| 5597 |
+
"loss": 1.4282,
|
| 5598 |
+
"step": 795
|
| 5599 |
+
},
|
| 5600 |
+
{
|
| 5601 |
+
"epoch": 0.7187358916478556,
|
| 5602 |
+
"grad_norm": 1.9457025527954102,
|
| 5603 |
+
"learning_rate": 3.726829031257062e-05,
|
| 5604 |
+
"loss": 1.3572,
|
| 5605 |
+
"step": 796
|
| 5606 |
+
},
|
| 5607 |
+
{
|
| 5608 |
+
"epoch": 0.7196388261851016,
|
| 5609 |
+
"grad_norm": 1.8345212936401367,
|
| 5610 |
+
"learning_rate": 3.704572780602701e-05,
|
| 5611 |
+
"loss": 1.0558,
|
| 5612 |
+
"step": 797
|
| 5613 |
+
},
|
| 5614 |
+
{
|
| 5615 |
+
"epoch": 0.7205417607223477,
|
| 5616 |
+
"grad_norm": 1.7764463424682617,
|
| 5617 |
+
"learning_rate": 3.6823680670482485e-05,
|
| 5618 |
+
"loss": 1.0214,
|
| 5619 |
+
"step": 798
|
| 5620 |
+
},
|
| 5621 |
+
{
|
| 5622 |
+
"epoch": 0.7214446952595936,
|
| 5623 |
+
"grad_norm": 1.5642400979995728,
|
| 5624 |
+
"learning_rate": 3.660215072371135e-05,
|
| 5625 |
+
"loss": 1.0181,
|
| 5626 |
+
"step": 799
|
| 5627 |
+
},
|
| 5628 |
+
{
|
| 5629 |
+
"epoch": 0.7223476297968398,
|
| 5630 |
+
"grad_norm": 1.6106574535369873,
|
| 5631 |
+
"learning_rate": 3.638113977925387e-05,
|
| 5632 |
+
"loss": 0.9659,
|
| 5633 |
+
"step": 800
|
| 5634 |
+
},
|
| 5635 |
+
{
|
| 5636 |
+
"epoch": 0.7232505643340857,
|
| 5637 |
+
"grad_norm": 1.6759979724884033,
|
| 5638 |
+
"learning_rate": 3.6160649646401635e-05,
|
| 5639 |
+
"loss": 1.0394,
|
| 5640 |
+
"step": 801
|
| 5641 |
+
},
|
| 5642 |
+
{
|
| 5643 |
+
"epoch": 0.7241534988713318,
|
| 5644 |
+
"grad_norm": 1.783288598060608,
|
| 5645 |
+
"learning_rate": 3.594068213018249e-05,
|
| 5646 |
+
"loss": 1.3223,
|
| 5647 |
+
"step": 802
|
| 5648 |
+
},
|
| 5649 |
+
{
|
| 5650 |
+
"epoch": 0.7250564334085778,
|
| 5651 |
+
"grad_norm": 1.7645738124847412,
|
| 5652 |
+
"learning_rate": 3.5721239031346066e-05,
|
| 5653 |
+
"loss": 1.0735,
|
| 5654 |
+
"step": 803
|
| 5655 |
+
},
|
| 5656 |
+
{
|
| 5657 |
+
"epoch": 0.7259593679458239,
|
| 5658 |
+
"grad_norm": 1.9430018663406372,
|
| 5659 |
+
"learning_rate": 3.550232214634884e-05,
|
| 5660 |
+
"loss": 1.5089,
|
| 5661 |
+
"step": 804
|
| 5662 |
+
},
|
| 5663 |
+
{
|
| 5664 |
+
"epoch": 0.7268623024830699,
|
| 5665 |
+
"grad_norm": 1.5665555000305176,
|
| 5666 |
+
"learning_rate": 3.528393326733941e-05,
|
| 5667 |
+
"loss": 1.0646,
|
| 5668 |
+
"step": 805
|
| 5669 |
+
},
|
| 5670 |
+
{
|
| 5671 |
+
"epoch": 0.727765237020316,
|
| 5672 |
+
"grad_norm": 1.7760690450668335,
|
| 5673 |
+
"learning_rate": 3.506607418214395e-05,
|
| 5674 |
+
"loss": 1.5192,
|
| 5675 |
+
"step": 806
|
| 5676 |
+
},
|
| 5677 |
+
{
|
| 5678 |
+
"epoch": 0.728668171557562,
|
| 5679 |
+
"grad_norm": 1.5534496307373047,
|
| 5680 |
+
"learning_rate": 3.4848746674251545e-05,
|
| 5681 |
+
"loss": 0.9254,
|
| 5682 |
+
"step": 807
|
| 5683 |
+
},
|
| 5684 |
+
{
|
| 5685 |
+
"epoch": 0.7295711060948081,
|
| 5686 |
+
"grad_norm": 1.872107744216919,
|
| 5687 |
+
"learning_rate": 3.463195252279939e-05,
|
| 5688 |
+
"loss": 1.2138,
|
| 5689 |
+
"step": 808
|
| 5690 |
+
},
|
| 5691 |
+
{
|
| 5692 |
+
"epoch": 0.7304740406320541,
|
| 5693 |
+
"grad_norm": 1.5574711561203003,
|
| 5694 |
+
"learning_rate": 3.441569350255868e-05,
|
| 5695 |
+
"loss": 1.206,
|
| 5696 |
+
"step": 809
|
| 5697 |
+
},
|
| 5698 |
+
{
|
| 5699 |
+
"epoch": 0.7313769751693002,
|
| 5700 |
+
"grad_norm": 1.5422251224517822,
|
| 5701 |
+
"learning_rate": 3.419997138391954e-05,
|
| 5702 |
+
"loss": 0.9381,
|
| 5703 |
+
"step": 810
|
| 5704 |
+
},
|
| 5705 |
+
{
|
| 5706 |
+
"epoch": 0.7322799097065463,
|
| 5707 |
+
"grad_norm": 1.3428008556365967,
|
| 5708 |
+
"learning_rate": 3.3984787932876814e-05,
|
| 5709 |
+
"loss": 0.982,
|
| 5710 |
+
"step": 811
|
| 5711 |
+
},
|
| 5712 |
+
{
|
| 5713 |
+
"epoch": 0.7331828442437923,
|
| 5714 |
+
"grad_norm": 1.870396614074707,
|
| 5715 |
+
"learning_rate": 3.377014491101577e-05,
|
| 5716 |
+
"loss": 1.3272,
|
| 5717 |
+
"step": 812
|
| 5718 |
+
},
|
| 5719 |
+
{
|
| 5720 |
+
"epoch": 0.7340857787810384,
|
| 5721 |
+
"grad_norm": 2.201111078262329,
|
| 5722 |
+
"learning_rate": 3.355604407549725e-05,
|
| 5723 |
+
"loss": 1.5567,
|
| 5724 |
+
"step": 813
|
| 5725 |
+
},
|
| 5726 |
+
{
|
| 5727 |
+
"epoch": 0.7349887133182844,
|
| 5728 |
+
"grad_norm": 1.5531777143478394,
|
| 5729 |
+
"learning_rate": 3.334248717904368e-05,
|
| 5730 |
+
"loss": 1.0115,
|
| 5731 |
+
"step": 814
|
| 5732 |
+
},
|
| 5733 |
+
{
|
| 5734 |
+
"epoch": 0.7358916478555305,
|
| 5735 |
+
"grad_norm": 1.5128211975097656,
|
| 5736 |
+
"learning_rate": 3.3129475969924526e-05,
|
| 5737 |
+
"loss": 0.929,
|
| 5738 |
+
"step": 815
|
| 5739 |
+
},
|
| 5740 |
+
{
|
| 5741 |
+
"epoch": 0.7367945823927765,
|
| 5742 |
+
"grad_norm": 1.2759476900100708,
|
| 5743 |
+
"learning_rate": 3.291701219194195e-05,
|
| 5744 |
+
"loss": 0.6896,
|
| 5745 |
+
"step": 816
|
| 5746 |
+
},
|
| 5747 |
+
{
|
| 5748 |
+
"epoch": 0.7376975169300226,
|
| 5749 |
+
"grad_norm": 1.4616997241973877,
|
| 5750 |
+
"learning_rate": 3.270509758441671e-05,
|
| 5751 |
+
"loss": 1.0415,
|
| 5752 |
+
"step": 817
|
| 5753 |
+
},
|
| 5754 |
+
{
|
| 5755 |
+
"epoch": 0.7386004514672686,
|
| 5756 |
+
"grad_norm": 1.445452094078064,
|
| 5757 |
+
"learning_rate": 3.24937338821738e-05,
|
| 5758 |
+
"loss": 1.0383,
|
| 5759 |
+
"step": 818
|
| 5760 |
+
},
|
| 5761 |
+
{
|
| 5762 |
+
"epoch": 0.7395033860045147,
|
| 5763 |
+
"grad_norm": 1.2269552946090698,
|
| 5764 |
+
"learning_rate": 3.2282922815528135e-05,
|
| 5765 |
+
"loss": 0.6495,
|
| 5766 |
+
"step": 819
|
| 5767 |
+
},
|
| 5768 |
+
{
|
| 5769 |
+
"epoch": 0.7404063205417607,
|
| 5770 |
+
"grad_norm": 1.6146233081817627,
|
| 5771 |
+
"learning_rate": 3.207266611027069e-05,
|
| 5772 |
+
"loss": 0.9821,
|
| 5773 |
+
"step": 820
|
| 5774 |
+
},
|
| 5775 |
+
{
|
| 5776 |
+
"epoch": 0.7413092550790068,
|
| 5777 |
+
"grad_norm": 1.2971818447113037,
|
| 5778 |
+
"learning_rate": 3.186296548765411e-05,
|
| 5779 |
+
"loss": 0.7807,
|
| 5780 |
+
"step": 821
|
| 5781 |
+
},
|
| 5782 |
+
{
|
| 5783 |
+
"epoch": 0.7422121896162528,
|
| 5784 |
+
"grad_norm": 1.6828197240829468,
|
| 5785 |
+
"learning_rate": 3.165382266437874e-05,
|
| 5786 |
+
"loss": 1.0824,
|
| 5787 |
+
"step": 822
|
| 5788 |
+
},
|
| 5789 |
+
{
|
| 5790 |
+
"epoch": 0.7431151241534989,
|
| 5791 |
+
"grad_norm": 1.1024787425994873,
|
| 5792 |
+
"learning_rate": 3.144523935257846e-05,
|
| 5793 |
+
"loss": 0.5144,
|
| 5794 |
+
"step": 823
|
| 5795 |
+
},
|
| 5796 |
+
{
|
| 5797 |
+
"epoch": 0.7440180586907449,
|
| 5798 |
+
"grad_norm": 1.3970650434494019,
|
| 5799 |
+
"learning_rate": 3.123721725980683e-05,
|
| 5800 |
+
"loss": 0.9,
|
| 5801 |
+
"step": 824
|
| 5802 |
+
},
|
| 5803 |
+
{
|
| 5804 |
+
"epoch": 0.744920993227991,
|
| 5805 |
+
"grad_norm": 1.3488242626190186,
|
| 5806 |
+
"learning_rate": 3.102975808902303e-05,
|
| 5807 |
+
"loss": 1.0081,
|
| 5808 |
+
"step": 825
|
| 5809 |
+
},
|
| 5810 |
+
{
|
| 5811 |
+
"epoch": 0.745823927765237,
|
| 5812 |
+
"grad_norm": 1.3460159301757812,
|
| 5813 |
+
"learning_rate": 3.082286353857782e-05,
|
| 5814 |
+
"loss": 1.2013,
|
| 5815 |
+
"step": 826
|
| 5816 |
+
},
|
| 5817 |
+
{
|
| 5818 |
+
"epoch": 0.7467268623024831,
|
| 5819 |
+
"grad_norm": 1.8649550676345825,
|
| 5820 |
+
"learning_rate": 3.061653530219983e-05,
|
| 5821 |
+
"loss": 1.3269,
|
| 5822 |
+
"step": 827
|
| 5823 |
+
},
|
| 5824 |
+
{
|
| 5825 |
+
"epoch": 0.7476297968397291,
|
| 5826 |
+
"grad_norm": 1.3285624980926514,
|
| 5827 |
+
"learning_rate": 3.0410775068981613e-05,
|
| 5828 |
+
"loss": 0.8373,
|
| 5829 |
+
"step": 828
|
| 5830 |
+
},
|
| 5831 |
+
{
|
| 5832 |
+
"epoch": 0.7485327313769752,
|
| 5833 |
+
"grad_norm": 1.1073821783065796,
|
| 5834 |
+
"learning_rate": 3.0205584523365626e-05,
|
| 5835 |
+
"loss": 0.6991,
|
| 5836 |
+
"step": 829
|
| 5837 |
+
},
|
| 5838 |
+
{
|
| 5839 |
+
"epoch": 0.7494356659142212,
|
| 5840 |
+
"grad_norm": 1.5901031494140625,
|
| 5841 |
+
"learning_rate": 3.0000965345130903e-05,
|
| 5842 |
+
"loss": 1.1712,
|
| 5843 |
+
"step": 830
|
| 5844 |
+
},
|
| 5845 |
+
{
|
| 5846 |
+
"epoch": 0.7503386004514673,
|
| 5847 |
+
"grad_norm": 1.817070722579956,
|
| 5848 |
+
"learning_rate": 2.979691920937875e-05,
|
| 5849 |
+
"loss": 1.4764,
|
| 5850 |
+
"step": 831
|
| 5851 |
+
},
|
| 5852 |
+
{
|
| 5853 |
+
"epoch": 0.7503386004514673,
|
| 5854 |
+
"eval_loss": 1.0855916738510132,
|
| 5855 |
+
"eval_runtime": 8.3193,
|
| 5856 |
+
"eval_samples_per_second": 56.135,
|
| 5857 |
+
"eval_steps_per_second": 28.127,
|
| 5858 |
+
"step": 831
|
| 5859 |
}
|
| 5860 |
],
|
| 5861 |
"logging_steps": 1,
|
|
|
|
| 5875 |
"attributes": {}
|
| 5876 |
}
|
| 5877 |
},
|
| 5878 |
+
"total_flos": 2.1256313392594944e+16,
|
| 5879 |
"train_batch_size": 2,
|
| 5880 |
"trial_name": null,
|
| 5881 |
"trial_params": null
|