Training in progress, step 1540, 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 104902272
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f14ef6fec965648f63b4dd06aded7a022596349469ccda0f6b809177f64c1366
|
| 3 |
size 104902272
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 53623316
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:fab1fe8e9127e017ba37f7631c72f3013b210b71d943e4d88caf38ae3a1af234
|
| 3 |
size 53623316
|
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 sha256:1c9a72ec94b181ba8d96906a3923edb1ec85538edfacf304bfb540b17ada23ed
|
| 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:f799a99d50707205d4f44f1c07481ab2c434510a2f5e2b9ca9c7eb27035c6d6c
|
| 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": 385,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -8124,6 +8124,2709 @@
|
|
| 8124 |
"eval_samples_per_second": 18.265,
|
| 8125 |
"eval_steps_per_second": 9.134,
|
| 8126 |
"step": 1155
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 8127 |
}
|
| 8128 |
],
|
| 8129 |
"logging_steps": 1,
|
|
@@ -8138,12 +10841,12 @@
|
|
| 8138 |
"should_evaluate": false,
|
| 8139 |
"should_log": false,
|
| 8140 |
"should_save": true,
|
| 8141 |
-
"should_training_stop":
|
| 8142 |
},
|
| 8143 |
"attributes": {}
|
| 8144 |
}
|
| 8145 |
},
|
| 8146 |
-
"total_flos":
|
| 8147 |
"train_batch_size": 2,
|
| 8148 |
"trial_name": null,
|
| 8149 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.1231753649270146,
|
| 5 |
"eval_steps": 385,
|
| 6 |
+
"global_step": 1540,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 8124 |
"eval_samples_per_second": 18.265,
|
| 8125 |
"eval_steps_per_second": 9.134,
|
| 8126 |
"step": 1155
|
| 8127 |
+
},
|
| 8128 |
+
{
|
| 8129 |
+
"epoch": 0.0924615076984603,
|
| 8130 |
+
"grad_norm": 4.510709285736084,
|
| 8131 |
+
"learning_rate": 2.9507445309385294e-05,
|
| 8132 |
+
"loss": 4.4372,
|
| 8133 |
+
"step": 1156
|
| 8134 |
+
},
|
| 8135 |
+
{
|
| 8136 |
+
"epoch": 0.09254149170165966,
|
| 8137 |
+
"grad_norm": 3.9571750164031982,
|
| 8138 |
+
"learning_rate": 2.9361955260953722e-05,
|
| 8139 |
+
"loss": 3.642,
|
| 8140 |
+
"step": 1157
|
| 8141 |
+
},
|
| 8142 |
+
{
|
| 8143 |
+
"epoch": 0.09262147570485903,
|
| 8144 |
+
"grad_norm": 4.990582466125488,
|
| 8145 |
+
"learning_rate": 2.9216763033579097e-05,
|
| 8146 |
+
"loss": 3.3757,
|
| 8147 |
+
"step": 1158
|
| 8148 |
+
},
|
| 8149 |
+
{
|
| 8150 |
+
"epoch": 0.09270145970805839,
|
| 8151 |
+
"grad_norm": 3.8319714069366455,
|
| 8152 |
+
"learning_rate": 2.9071869239414663e-05,
|
| 8153 |
+
"loss": 3.0402,
|
| 8154 |
+
"step": 1159
|
| 8155 |
+
},
|
| 8156 |
+
{
|
| 8157 |
+
"epoch": 0.09278144371125775,
|
| 8158 |
+
"grad_norm": 5.654752731323242,
|
| 8159 |
+
"learning_rate": 2.8927274489355293e-05,
|
| 8160 |
+
"loss": 4.2507,
|
| 8161 |
+
"step": 1160
|
| 8162 |
+
},
|
| 8163 |
+
{
|
| 8164 |
+
"epoch": 0.09286142771445711,
|
| 8165 |
+
"grad_norm": 4.614495754241943,
|
| 8166 |
+
"learning_rate": 2.878297939303507e-05,
|
| 8167 |
+
"loss": 4.291,
|
| 8168 |
+
"step": 1161
|
| 8169 |
+
},
|
| 8170 |
+
{
|
| 8171 |
+
"epoch": 0.09294141171765646,
|
| 8172 |
+
"grad_norm": 4.101933479309082,
|
| 8173 |
+
"learning_rate": 2.8638984558824777e-05,
|
| 8174 |
+
"loss": 3.3441,
|
| 8175 |
+
"step": 1162
|
| 8176 |
+
},
|
| 8177 |
+
{
|
| 8178 |
+
"epoch": 0.09302139572085583,
|
| 8179 |
+
"grad_norm": 3.650947332382202,
|
| 8180 |
+
"learning_rate": 2.849529059382915e-05,
|
| 8181 |
+
"loss": 2.8172,
|
| 8182 |
+
"step": 1163
|
| 8183 |
+
},
|
| 8184 |
+
{
|
| 8185 |
+
"epoch": 0.09310137972405519,
|
| 8186 |
+
"grad_norm": 4.197967052459717,
|
| 8187 |
+
"learning_rate": 2.835189810388441e-05,
|
| 8188 |
+
"loss": 3.5285,
|
| 8189 |
+
"step": 1164
|
| 8190 |
+
},
|
| 8191 |
+
{
|
| 8192 |
+
"epoch": 0.09318136372725455,
|
| 8193 |
+
"grad_norm": 4.3074727058410645,
|
| 8194 |
+
"learning_rate": 2.8208807693555818e-05,
|
| 8195 |
+
"loss": 3.8336,
|
| 8196 |
+
"step": 1165
|
| 8197 |
+
},
|
| 8198 |
+
{
|
| 8199 |
+
"epoch": 0.09326134773045391,
|
| 8200 |
+
"grad_norm": 4.938948631286621,
|
| 8201 |
+
"learning_rate": 2.8066019966134904e-05,
|
| 8202 |
+
"loss": 3.812,
|
| 8203 |
+
"step": 1166
|
| 8204 |
+
},
|
| 8205 |
+
{
|
| 8206 |
+
"epoch": 0.09334133173365326,
|
| 8207 |
+
"grad_norm": 4.886793613433838,
|
| 8208 |
+
"learning_rate": 2.792353552363707e-05,
|
| 8209 |
+
"loss": 3.9102,
|
| 8210 |
+
"step": 1167
|
| 8211 |
+
},
|
| 8212 |
+
{
|
| 8213 |
+
"epoch": 0.09342131573685263,
|
| 8214 |
+
"grad_norm": 4.254992961883545,
|
| 8215 |
+
"learning_rate": 2.7781354966799078e-05,
|
| 8216 |
+
"loss": 3.3508,
|
| 8217 |
+
"step": 1168
|
| 8218 |
+
},
|
| 8219 |
+
{
|
| 8220 |
+
"epoch": 0.09350129974005199,
|
| 8221 |
+
"grad_norm": 3.545367956161499,
|
| 8222 |
+
"learning_rate": 2.7639478895076397e-05,
|
| 8223 |
+
"loss": 3.0721,
|
| 8224 |
+
"step": 1169
|
| 8225 |
+
},
|
| 8226 |
+
{
|
| 8227 |
+
"epoch": 0.09358128374325135,
|
| 8228 |
+
"grad_norm": 2.9493346214294434,
|
| 8229 |
+
"learning_rate": 2.7497907906640742e-05,
|
| 8230 |
+
"loss": 2.0804,
|
| 8231 |
+
"step": 1170
|
| 8232 |
+
},
|
| 8233 |
+
{
|
| 8234 |
+
"epoch": 0.09366126774645071,
|
| 8235 |
+
"grad_norm": 3.6416776180267334,
|
| 8236 |
+
"learning_rate": 2.7356642598377603e-05,
|
| 8237 |
+
"loss": 3.1277,
|
| 8238 |
+
"step": 1171
|
| 8239 |
+
},
|
| 8240 |
+
{
|
| 8241 |
+
"epoch": 0.09374125174965008,
|
| 8242 |
+
"grad_norm": 5.29701566696167,
|
| 8243 |
+
"learning_rate": 2.721568356588362e-05,
|
| 8244 |
+
"loss": 3.3519,
|
| 8245 |
+
"step": 1172
|
| 8246 |
+
},
|
| 8247 |
+
{
|
| 8248 |
+
"epoch": 0.09382123575284942,
|
| 8249 |
+
"grad_norm": 4.363441467285156,
|
| 8250 |
+
"learning_rate": 2.7075031403464123e-05,
|
| 8251 |
+
"loss": 3.1857,
|
| 8252 |
+
"step": 1173
|
| 8253 |
+
},
|
| 8254 |
+
{
|
| 8255 |
+
"epoch": 0.09390121975604879,
|
| 8256 |
+
"grad_norm": 3.976638078689575,
|
| 8257 |
+
"learning_rate": 2.6934686704130696e-05,
|
| 8258 |
+
"loss": 4.1215,
|
| 8259 |
+
"step": 1174
|
| 8260 |
+
},
|
| 8261 |
+
{
|
| 8262 |
+
"epoch": 0.09398120375924815,
|
| 8263 |
+
"grad_norm": 3.825456142425537,
|
| 8264 |
+
"learning_rate": 2.679465005959856e-05,
|
| 8265 |
+
"loss": 3.581,
|
| 8266 |
+
"step": 1175
|
| 8267 |
+
},
|
| 8268 |
+
{
|
| 8269 |
+
"epoch": 0.09406118776244751,
|
| 8270 |
+
"grad_norm": 3.7466976642608643,
|
| 8271 |
+
"learning_rate": 2.665492206028407e-05,
|
| 8272 |
+
"loss": 3.265,
|
| 8273 |
+
"step": 1176
|
| 8274 |
+
},
|
| 8275 |
+
{
|
| 8276 |
+
"epoch": 0.09414117176564687,
|
| 8277 |
+
"grad_norm": 4.0751872062683105,
|
| 8278 |
+
"learning_rate": 2.6515503295302447e-05,
|
| 8279 |
+
"loss": 4.1583,
|
| 8280 |
+
"step": 1177
|
| 8281 |
+
},
|
| 8282 |
+
{
|
| 8283 |
+
"epoch": 0.09422115576884624,
|
| 8284 |
+
"grad_norm": 3.220518112182617,
|
| 8285 |
+
"learning_rate": 2.6376394352464972e-05,
|
| 8286 |
+
"loss": 2.3114,
|
| 8287 |
+
"step": 1178
|
| 8288 |
+
},
|
| 8289 |
+
{
|
| 8290 |
+
"epoch": 0.09430113977204559,
|
| 8291 |
+
"grad_norm": 4.122797012329102,
|
| 8292 |
+
"learning_rate": 2.6237595818276728e-05,
|
| 8293 |
+
"loss": 2.8948,
|
| 8294 |
+
"step": 1179
|
| 8295 |
+
},
|
| 8296 |
+
{
|
| 8297 |
+
"epoch": 0.09438112377524495,
|
| 8298 |
+
"grad_norm": 3.5874483585357666,
|
| 8299 |
+
"learning_rate": 2.6099108277934103e-05,
|
| 8300 |
+
"loss": 3.2958,
|
| 8301 |
+
"step": 1180
|
| 8302 |
+
},
|
| 8303 |
+
{
|
| 8304 |
+
"epoch": 0.09446110777844431,
|
| 8305 |
+
"grad_norm": 4.130005359649658,
|
| 8306 |
+
"learning_rate": 2.596093231532224e-05,
|
| 8307 |
+
"loss": 3.3648,
|
| 8308 |
+
"step": 1181
|
| 8309 |
+
},
|
| 8310 |
+
{
|
| 8311 |
+
"epoch": 0.09454109178164367,
|
| 8312 |
+
"grad_norm": 4.924187183380127,
|
| 8313 |
+
"learning_rate": 2.5823068513012595e-05,
|
| 8314 |
+
"loss": 2.7937,
|
| 8315 |
+
"step": 1182
|
| 8316 |
+
},
|
| 8317 |
+
{
|
| 8318 |
+
"epoch": 0.09462107578484304,
|
| 8319 |
+
"grad_norm": 3.9850120544433594,
|
| 8320 |
+
"learning_rate": 2.5685517452260567e-05,
|
| 8321 |
+
"loss": 3.303,
|
| 8322 |
+
"step": 1183
|
| 8323 |
+
},
|
| 8324 |
+
{
|
| 8325 |
+
"epoch": 0.09470105978804239,
|
| 8326 |
+
"grad_norm": 4.6082353591918945,
|
| 8327 |
+
"learning_rate": 2.5548279713002997e-05,
|
| 8328 |
+
"loss": 3.3408,
|
| 8329 |
+
"step": 1184
|
| 8330 |
+
},
|
| 8331 |
+
{
|
| 8332 |
+
"epoch": 0.09478104379124175,
|
| 8333 |
+
"grad_norm": 3.862121820449829,
|
| 8334 |
+
"learning_rate": 2.541135587385568e-05,
|
| 8335 |
+
"loss": 3.7493,
|
| 8336 |
+
"step": 1185
|
| 8337 |
+
},
|
| 8338 |
+
{
|
| 8339 |
+
"epoch": 0.09486102779444111,
|
| 8340 |
+
"grad_norm": 4.668119430541992,
|
| 8341 |
+
"learning_rate": 2.527474651211089e-05,
|
| 8342 |
+
"loss": 4.6341,
|
| 8343 |
+
"step": 1186
|
| 8344 |
+
},
|
| 8345 |
+
{
|
| 8346 |
+
"epoch": 0.09494101179764047,
|
| 8347 |
+
"grad_norm": 4.146975517272949,
|
| 8348 |
+
"learning_rate": 2.5138452203735173e-05,
|
| 8349 |
+
"loss": 3.2399,
|
| 8350 |
+
"step": 1187
|
| 8351 |
+
},
|
| 8352 |
+
{
|
| 8353 |
+
"epoch": 0.09502099580083984,
|
| 8354 |
+
"grad_norm": 4.673194885253906,
|
| 8355 |
+
"learning_rate": 2.500247352336664e-05,
|
| 8356 |
+
"loss": 4.7629,
|
| 8357 |
+
"step": 1188
|
| 8358 |
+
},
|
| 8359 |
+
{
|
| 8360 |
+
"epoch": 0.0951009798040392,
|
| 8361 |
+
"grad_norm": 4.4623122215271,
|
| 8362 |
+
"learning_rate": 2.4866811044312665e-05,
|
| 8363 |
+
"loss": 3.9425,
|
| 8364 |
+
"step": 1189
|
| 8365 |
+
},
|
| 8366 |
+
{
|
| 8367 |
+
"epoch": 0.09518096380723855,
|
| 8368 |
+
"grad_norm": 4.428549766540527,
|
| 8369 |
+
"learning_rate": 2.4731465338547556e-05,
|
| 8370 |
+
"loss": 3.7749,
|
| 8371 |
+
"step": 1190
|
| 8372 |
+
},
|
| 8373 |
+
{
|
| 8374 |
+
"epoch": 0.09526094781043791,
|
| 8375 |
+
"grad_norm": 4.835275650024414,
|
| 8376 |
+
"learning_rate": 2.459643697670998e-05,
|
| 8377 |
+
"loss": 3.5015,
|
| 8378 |
+
"step": 1191
|
| 8379 |
+
},
|
| 8380 |
+
{
|
| 8381 |
+
"epoch": 0.09534093181363727,
|
| 8382 |
+
"grad_norm": 3.94173526763916,
|
| 8383 |
+
"learning_rate": 2.4461726528100615e-05,
|
| 8384 |
+
"loss": 3.3298,
|
| 8385 |
+
"step": 1192
|
| 8386 |
+
},
|
| 8387 |
+
{
|
| 8388 |
+
"epoch": 0.09542091581683664,
|
| 8389 |
+
"grad_norm": 3.751077651977539,
|
| 8390 |
+
"learning_rate": 2.432733456067987e-05,
|
| 8391 |
+
"loss": 3.2805,
|
| 8392 |
+
"step": 1193
|
| 8393 |
+
},
|
| 8394 |
+
{
|
| 8395 |
+
"epoch": 0.095500899820036,
|
| 8396 |
+
"grad_norm": 4.054288864135742,
|
| 8397 |
+
"learning_rate": 2.41932616410653e-05,
|
| 8398 |
+
"loss": 3.4902,
|
| 8399 |
+
"step": 1194
|
| 8400 |
+
},
|
| 8401 |
+
{
|
| 8402 |
+
"epoch": 0.09558088382323535,
|
| 8403 |
+
"grad_norm": 4.829845905303955,
|
| 8404 |
+
"learning_rate": 2.405950833452928e-05,
|
| 8405 |
+
"loss": 4.494,
|
| 8406 |
+
"step": 1195
|
| 8407 |
+
},
|
| 8408 |
+
{
|
| 8409 |
+
"epoch": 0.09566086782643471,
|
| 8410 |
+
"grad_norm": 4.179841995239258,
|
| 8411 |
+
"learning_rate": 2.392607520499677e-05,
|
| 8412 |
+
"loss": 3.782,
|
| 8413 |
+
"step": 1196
|
| 8414 |
+
},
|
| 8415 |
+
{
|
| 8416 |
+
"epoch": 0.09574085182963407,
|
| 8417 |
+
"grad_norm": 4.335528373718262,
|
| 8418 |
+
"learning_rate": 2.3792962815042685e-05,
|
| 8419 |
+
"loss": 3.9263,
|
| 8420 |
+
"step": 1197
|
| 8421 |
+
},
|
| 8422 |
+
{
|
| 8423 |
+
"epoch": 0.09582083583283343,
|
| 8424 |
+
"grad_norm": 3.8097429275512695,
|
| 8425 |
+
"learning_rate": 2.36601717258897e-05,
|
| 8426 |
+
"loss": 3.6231,
|
| 8427 |
+
"step": 1198
|
| 8428 |
+
},
|
| 8429 |
+
{
|
| 8430 |
+
"epoch": 0.0959008198360328,
|
| 8431 |
+
"grad_norm": 3.452497959136963,
|
| 8432 |
+
"learning_rate": 2.352770249740588e-05,
|
| 8433 |
+
"loss": 2.9945,
|
| 8434 |
+
"step": 1199
|
| 8435 |
+
},
|
| 8436 |
+
{
|
| 8437 |
+
"epoch": 0.09598080383923216,
|
| 8438 |
+
"grad_norm": 4.345489025115967,
|
| 8439 |
+
"learning_rate": 2.339555568810221e-05,
|
| 8440 |
+
"loss": 3.0404,
|
| 8441 |
+
"step": 1200
|
| 8442 |
+
},
|
| 8443 |
+
{
|
| 8444 |
+
"epoch": 0.09606078784243151,
|
| 8445 |
+
"grad_norm": 4.521251201629639,
|
| 8446 |
+
"learning_rate": 2.3263731855130298e-05,
|
| 8447 |
+
"loss": 3.9112,
|
| 8448 |
+
"step": 1201
|
| 8449 |
+
},
|
| 8450 |
+
{
|
| 8451 |
+
"epoch": 0.09614077184563087,
|
| 8452 |
+
"grad_norm": 3.881998300552368,
|
| 8453 |
+
"learning_rate": 2.3132231554280136e-05,
|
| 8454 |
+
"loss": 3.3303,
|
| 8455 |
+
"step": 1202
|
| 8456 |
+
},
|
| 8457 |
+
{
|
| 8458 |
+
"epoch": 0.09622075584883023,
|
| 8459 |
+
"grad_norm": 5.4891676902771,
|
| 8460 |
+
"learning_rate": 2.300105533997753e-05,
|
| 8461 |
+
"loss": 3.3695,
|
| 8462 |
+
"step": 1203
|
| 8463 |
+
},
|
| 8464 |
+
{
|
| 8465 |
+
"epoch": 0.0963007398520296,
|
| 8466 |
+
"grad_norm": 3.869205951690674,
|
| 8467 |
+
"learning_rate": 2.2870203765281926e-05,
|
| 8468 |
+
"loss": 2.9992,
|
| 8469 |
+
"step": 1204
|
| 8470 |
+
},
|
| 8471 |
+
{
|
| 8472 |
+
"epoch": 0.09638072385522896,
|
| 8473 |
+
"grad_norm": 6.634833335876465,
|
| 8474 |
+
"learning_rate": 2.2739677381884115e-05,
|
| 8475 |
+
"loss": 4.5587,
|
| 8476 |
+
"step": 1205
|
| 8477 |
+
},
|
| 8478 |
+
{
|
| 8479 |
+
"epoch": 0.09646070785842832,
|
| 8480 |
+
"grad_norm": 4.484224796295166,
|
| 8481 |
+
"learning_rate": 2.260947674010372e-05,
|
| 8482 |
+
"loss": 3.9581,
|
| 8483 |
+
"step": 1206
|
| 8484 |
+
},
|
| 8485 |
+
{
|
| 8486 |
+
"epoch": 0.09654069186162767,
|
| 8487 |
+
"grad_norm": 4.545680999755859,
|
| 8488 |
+
"learning_rate": 2.2479602388887012e-05,
|
| 8489 |
+
"loss": 4.4443,
|
| 8490 |
+
"step": 1207
|
| 8491 |
+
},
|
| 8492 |
+
{
|
| 8493 |
+
"epoch": 0.09662067586482703,
|
| 8494 |
+
"grad_norm": 5.094995021820068,
|
| 8495 |
+
"learning_rate": 2.235005487580466e-05,
|
| 8496 |
+
"loss": 3.5916,
|
| 8497 |
+
"step": 1208
|
| 8498 |
+
},
|
| 8499 |
+
{
|
| 8500 |
+
"epoch": 0.0967006598680264,
|
| 8501 |
+
"grad_norm": 4.446115493774414,
|
| 8502 |
+
"learning_rate": 2.2220834747049214e-05,
|
| 8503 |
+
"loss": 3.7843,
|
| 8504 |
+
"step": 1209
|
| 8505 |
+
},
|
| 8506 |
+
{
|
| 8507 |
+
"epoch": 0.09678064387122576,
|
| 8508 |
+
"grad_norm": 4.460416316986084,
|
| 8509 |
+
"learning_rate": 2.2091942547432955e-05,
|
| 8510 |
+
"loss": 3.7152,
|
| 8511 |
+
"step": 1210
|
| 8512 |
+
},
|
| 8513 |
+
{
|
| 8514 |
+
"epoch": 0.09686062787442512,
|
| 8515 |
+
"grad_norm": 3.8882133960723877,
|
| 8516 |
+
"learning_rate": 2.1963378820385626e-05,
|
| 8517 |
+
"loss": 2.9874,
|
| 8518 |
+
"step": 1211
|
| 8519 |
+
},
|
| 8520 |
+
{
|
| 8521 |
+
"epoch": 0.09694061187762447,
|
| 8522 |
+
"grad_norm": 4.029484748840332,
|
| 8523 |
+
"learning_rate": 2.1835144107952022e-05,
|
| 8524 |
+
"loss": 3.9031,
|
| 8525 |
+
"step": 1212
|
| 8526 |
+
},
|
| 8527 |
+
{
|
| 8528 |
+
"epoch": 0.09702059588082383,
|
| 8529 |
+
"grad_norm": 4.002587795257568,
|
| 8530 |
+
"learning_rate": 2.170723895078972e-05,
|
| 8531 |
+
"loss": 3.0681,
|
| 8532 |
+
"step": 1213
|
| 8533 |
+
},
|
| 8534 |
+
{
|
| 8535 |
+
"epoch": 0.0971005798840232,
|
| 8536 |
+
"grad_norm": 6.7918620109558105,
|
| 8537 |
+
"learning_rate": 2.1579663888166956e-05,
|
| 8538 |
+
"loss": 4.8938,
|
| 8539 |
+
"step": 1214
|
| 8540 |
+
},
|
| 8541 |
+
{
|
| 8542 |
+
"epoch": 0.09718056388722256,
|
| 8543 |
+
"grad_norm": 3.8774309158325195,
|
| 8544 |
+
"learning_rate": 2.1452419457960137e-05,
|
| 8545 |
+
"loss": 3.4964,
|
| 8546 |
+
"step": 1215
|
| 8547 |
+
},
|
| 8548 |
+
{
|
| 8549 |
+
"epoch": 0.09726054789042192,
|
| 8550 |
+
"grad_norm": 3.694997549057007,
|
| 8551 |
+
"learning_rate": 2.132550619665168e-05,
|
| 8552 |
+
"loss": 3.0601,
|
| 8553 |
+
"step": 1216
|
| 8554 |
+
},
|
| 8555 |
+
{
|
| 8556 |
+
"epoch": 0.09734053189362128,
|
| 8557 |
+
"grad_norm": 4.705177307128906,
|
| 8558 |
+
"learning_rate": 2.119892463932781e-05,
|
| 8559 |
+
"loss": 3.7202,
|
| 8560 |
+
"step": 1217
|
| 8561 |
+
},
|
| 8562 |
+
{
|
| 8563 |
+
"epoch": 0.09742051589682063,
|
| 8564 |
+
"grad_norm": 4.710257053375244,
|
| 8565 |
+
"learning_rate": 2.107267531967618e-05,
|
| 8566 |
+
"loss": 4.0198,
|
| 8567 |
+
"step": 1218
|
| 8568 |
+
},
|
| 8569 |
+
{
|
| 8570 |
+
"epoch": 0.09750049990002,
|
| 8571 |
+
"grad_norm": 4.042464733123779,
|
| 8572 |
+
"learning_rate": 2.0946758769983665e-05,
|
| 8573 |
+
"loss": 3.5091,
|
| 8574 |
+
"step": 1219
|
| 8575 |
+
},
|
| 8576 |
+
{
|
| 8577 |
+
"epoch": 0.09758048390321936,
|
| 8578 |
+
"grad_norm": 4.7746734619140625,
|
| 8579 |
+
"learning_rate": 2.0821175521134207e-05,
|
| 8580 |
+
"loss": 3.528,
|
| 8581 |
+
"step": 1220
|
| 8582 |
+
},
|
| 8583 |
+
{
|
| 8584 |
+
"epoch": 0.09766046790641872,
|
| 8585 |
+
"grad_norm": 4.828027725219727,
|
| 8586 |
+
"learning_rate": 2.06959261026064e-05,
|
| 8587 |
+
"loss": 4.1874,
|
| 8588 |
+
"step": 1221
|
| 8589 |
+
},
|
| 8590 |
+
{
|
| 8591 |
+
"epoch": 0.09774045190961808,
|
| 8592 |
+
"grad_norm": 4.168130397796631,
|
| 8593 |
+
"learning_rate": 2.05710110424714e-05,
|
| 8594 |
+
"loss": 3.4702,
|
| 8595 |
+
"step": 1222
|
| 8596 |
+
},
|
| 8597 |
+
{
|
| 8598 |
+
"epoch": 0.09782043591281743,
|
| 8599 |
+
"grad_norm": 6.053627967834473,
|
| 8600 |
+
"learning_rate": 2.0446430867390696e-05,
|
| 8601 |
+
"loss": 4.2077,
|
| 8602 |
+
"step": 1223
|
| 8603 |
+
},
|
| 8604 |
+
{
|
| 8605 |
+
"epoch": 0.09790041991601679,
|
| 8606 |
+
"grad_norm": 4.410123825073242,
|
| 8607 |
+
"learning_rate": 2.0322186102613795e-05,
|
| 8608 |
+
"loss": 3.7704,
|
| 8609 |
+
"step": 1224
|
| 8610 |
+
},
|
| 8611 |
+
{
|
| 8612 |
+
"epoch": 0.09798040391921616,
|
| 8613 |
+
"grad_norm": 2.8216817378997803,
|
| 8614 |
+
"learning_rate": 2.0198277271976052e-05,
|
| 8615 |
+
"loss": 2.1769,
|
| 8616 |
+
"step": 1225
|
| 8617 |
+
},
|
| 8618 |
+
{
|
| 8619 |
+
"epoch": 0.09806038792241552,
|
| 8620 |
+
"grad_norm": 3.798919200897217,
|
| 8621 |
+
"learning_rate": 2.0074704897896558e-05,
|
| 8622 |
+
"loss": 2.9967,
|
| 8623 |
+
"step": 1226
|
| 8624 |
+
},
|
| 8625 |
+
{
|
| 8626 |
+
"epoch": 0.09814037192561488,
|
| 8627 |
+
"grad_norm": 3.748995542526245,
|
| 8628 |
+
"learning_rate": 1.995146950137575e-05,
|
| 8629 |
+
"loss": 2.989,
|
| 8630 |
+
"step": 1227
|
| 8631 |
+
},
|
| 8632 |
+
{
|
| 8633 |
+
"epoch": 0.09822035592881424,
|
| 8634 |
+
"grad_norm": 4.26412296295166,
|
| 8635 |
+
"learning_rate": 1.982857160199334e-05,
|
| 8636 |
+
"loss": 3.5748,
|
| 8637 |
+
"step": 1228
|
| 8638 |
+
},
|
| 8639 |
+
{
|
| 8640 |
+
"epoch": 0.09830033993201359,
|
| 8641 |
+
"grad_norm": 4.9580078125,
|
| 8642 |
+
"learning_rate": 1.970601171790616e-05,
|
| 8643 |
+
"loss": 3.6186,
|
| 8644 |
+
"step": 1229
|
| 8645 |
+
},
|
| 8646 |
+
{
|
| 8647 |
+
"epoch": 0.09838032393521295,
|
| 8648 |
+
"grad_norm": 4.893904685974121,
|
| 8649 |
+
"learning_rate": 1.9583790365845822e-05,
|
| 8650 |
+
"loss": 3.1621,
|
| 8651 |
+
"step": 1230
|
| 8652 |
+
},
|
| 8653 |
+
{
|
| 8654 |
+
"epoch": 0.09846030793841232,
|
| 8655 |
+
"grad_norm": 3.505159616470337,
|
| 8656 |
+
"learning_rate": 1.946190806111674e-05,
|
| 8657 |
+
"loss": 3.0279,
|
| 8658 |
+
"step": 1231
|
| 8659 |
+
},
|
| 8660 |
+
{
|
| 8661 |
+
"epoch": 0.09854029194161168,
|
| 8662 |
+
"grad_norm": 4.74012565612793,
|
| 8663 |
+
"learning_rate": 1.9340365317593746e-05,
|
| 8664 |
+
"loss": 4.0141,
|
| 8665 |
+
"step": 1232
|
| 8666 |
+
},
|
| 8667 |
+
{
|
| 8668 |
+
"epoch": 0.09862027594481104,
|
| 8669 |
+
"grad_norm": 4.515071868896484,
|
| 8670 |
+
"learning_rate": 1.921916264772011e-05,
|
| 8671 |
+
"loss": 3.3258,
|
| 8672 |
+
"step": 1233
|
| 8673 |
+
},
|
| 8674 |
+
{
|
| 8675 |
+
"epoch": 0.0987002599480104,
|
| 8676 |
+
"grad_norm": 5.331474304199219,
|
| 8677 |
+
"learning_rate": 1.9098300562505266e-05,
|
| 8678 |
+
"loss": 3.2355,
|
| 8679 |
+
"step": 1234
|
| 8680 |
+
},
|
| 8681 |
+
{
|
| 8682 |
+
"epoch": 0.09878024395120975,
|
| 8683 |
+
"grad_norm": 4.327870845794678,
|
| 8684 |
+
"learning_rate": 1.8977779571522646e-05,
|
| 8685 |
+
"loss": 3.7563,
|
| 8686 |
+
"step": 1235
|
| 8687 |
+
},
|
| 8688 |
+
{
|
| 8689 |
+
"epoch": 0.09886022795440912,
|
| 8690 |
+
"grad_norm": 4.85066032409668,
|
| 8691 |
+
"learning_rate": 1.8857600182907675e-05,
|
| 8692 |
+
"loss": 3.9254,
|
| 8693 |
+
"step": 1236
|
| 8694 |
+
},
|
| 8695 |
+
{
|
| 8696 |
+
"epoch": 0.09894021195760848,
|
| 8697 |
+
"grad_norm": 5.018810272216797,
|
| 8698 |
+
"learning_rate": 1.873776290335545e-05,
|
| 8699 |
+
"loss": 3.7063,
|
| 8700 |
+
"step": 1237
|
| 8701 |
+
},
|
| 8702 |
+
{
|
| 8703 |
+
"epoch": 0.09902019596080784,
|
| 8704 |
+
"grad_norm": 4.827825546264648,
|
| 8705 |
+
"learning_rate": 1.8618268238118675e-05,
|
| 8706 |
+
"loss": 3.0606,
|
| 8707 |
+
"step": 1238
|
| 8708 |
+
},
|
| 8709 |
+
{
|
| 8710 |
+
"epoch": 0.0991001799640072,
|
| 8711 |
+
"grad_norm": 3.860900640487671,
|
| 8712 |
+
"learning_rate": 1.8499116691005613e-05,
|
| 8713 |
+
"loss": 2.6762,
|
| 8714 |
+
"step": 1239
|
| 8715 |
+
},
|
| 8716 |
+
{
|
| 8717 |
+
"epoch": 0.09918016396720655,
|
| 8718 |
+
"grad_norm": 4.694439888000488,
|
| 8719 |
+
"learning_rate": 1.8380308764377842e-05,
|
| 8720 |
+
"loss": 3.3674,
|
| 8721 |
+
"step": 1240
|
| 8722 |
+
},
|
| 8723 |
+
{
|
| 8724 |
+
"epoch": 0.09926014797040592,
|
| 8725 |
+
"grad_norm": 3.836242198944092,
|
| 8726 |
+
"learning_rate": 1.8261844959148135e-05,
|
| 8727 |
+
"loss": 3.128,
|
| 8728 |
+
"step": 1241
|
| 8729 |
+
},
|
| 8730 |
+
{
|
| 8731 |
+
"epoch": 0.09934013197360528,
|
| 8732 |
+
"grad_norm": 4.42763090133667,
|
| 8733 |
+
"learning_rate": 1.8143725774778508e-05,
|
| 8734 |
+
"loss": 3.1887,
|
| 8735 |
+
"step": 1242
|
| 8736 |
+
},
|
| 8737 |
+
{
|
| 8738 |
+
"epoch": 0.09942011597680464,
|
| 8739 |
+
"grad_norm": 4.029833793640137,
|
| 8740 |
+
"learning_rate": 1.8025951709277898e-05,
|
| 8741 |
+
"loss": 3.4147,
|
| 8742 |
+
"step": 1243
|
| 8743 |
+
},
|
| 8744 |
+
{
|
| 8745 |
+
"epoch": 0.099500099980004,
|
| 8746 |
+
"grad_norm": 3.6723241806030273,
|
| 8747 |
+
"learning_rate": 1.7908523259200192e-05,
|
| 8748 |
+
"loss": 2.8744,
|
| 8749 |
+
"step": 1244
|
| 8750 |
+
},
|
| 8751 |
+
{
|
| 8752 |
+
"epoch": 0.09958008398320337,
|
| 8753 |
+
"grad_norm": 3.8148231506347656,
|
| 8754 |
+
"learning_rate": 1.7791440919642177e-05,
|
| 8755 |
+
"loss": 2.7896,
|
| 8756 |
+
"step": 1245
|
| 8757 |
+
},
|
| 8758 |
+
{
|
| 8759 |
+
"epoch": 0.09966006798640271,
|
| 8760 |
+
"grad_norm": 4.132092475891113,
|
| 8761 |
+
"learning_rate": 1.767470518424129e-05,
|
| 8762 |
+
"loss": 2.6917,
|
| 8763 |
+
"step": 1246
|
| 8764 |
+
},
|
| 8765 |
+
{
|
| 8766 |
+
"epoch": 0.09974005198960208,
|
| 8767 |
+
"grad_norm": 3.8130404949188232,
|
| 8768 |
+
"learning_rate": 1.755831654517365e-05,
|
| 8769 |
+
"loss": 3.3571,
|
| 8770 |
+
"step": 1247
|
| 8771 |
+
},
|
| 8772 |
+
{
|
| 8773 |
+
"epoch": 0.09982003599280144,
|
| 8774 |
+
"grad_norm": 4.179472923278809,
|
| 8775 |
+
"learning_rate": 1.7442275493152037e-05,
|
| 8776 |
+
"loss": 3.2753,
|
| 8777 |
+
"step": 1248
|
| 8778 |
+
},
|
| 8779 |
+
{
|
| 8780 |
+
"epoch": 0.0999000199960008,
|
| 8781 |
+
"grad_norm": 4.515778064727783,
|
| 8782 |
+
"learning_rate": 1.732658251742366e-05,
|
| 8783 |
+
"loss": 3.341,
|
| 8784 |
+
"step": 1249
|
| 8785 |
+
},
|
| 8786 |
+
{
|
| 8787 |
+
"epoch": 0.09998000399920016,
|
| 8788 |
+
"grad_norm": 3.3752856254577637,
|
| 8789 |
+
"learning_rate": 1.7211238105768214e-05,
|
| 8790 |
+
"loss": 2.6897,
|
| 8791 |
+
"step": 1250
|
| 8792 |
+
},
|
| 8793 |
+
{
|
| 8794 |
+
"epoch": 0.10005998800239951,
|
| 8795 |
+
"grad_norm": 3.830716609954834,
|
| 8796 |
+
"learning_rate": 1.7096242744495837e-05,
|
| 8797 |
+
"loss": 3.3853,
|
| 8798 |
+
"step": 1251
|
| 8799 |
+
},
|
| 8800 |
+
{
|
| 8801 |
+
"epoch": 0.10013997200559888,
|
| 8802 |
+
"grad_norm": 13.131118774414062,
|
| 8803 |
+
"learning_rate": 1.6981596918444953e-05,
|
| 8804 |
+
"loss": 3.8576,
|
| 8805 |
+
"step": 1252
|
| 8806 |
+
},
|
| 8807 |
+
{
|
| 8808 |
+
"epoch": 0.10021995600879824,
|
| 8809 |
+
"grad_norm": 4.193271636962891,
|
| 8810 |
+
"learning_rate": 1.6867301110980283e-05,
|
| 8811 |
+
"loss": 3.5221,
|
| 8812 |
+
"step": 1253
|
| 8813 |
+
},
|
| 8814 |
+
{
|
| 8815 |
+
"epoch": 0.1002999400119976,
|
| 8816 |
+
"grad_norm": 4.261105537414551,
|
| 8817 |
+
"learning_rate": 1.6753355803990912e-05,
|
| 8818 |
+
"loss": 4.0092,
|
| 8819 |
+
"step": 1254
|
| 8820 |
+
},
|
| 8821 |
+
{
|
| 8822 |
+
"epoch": 0.10037992401519696,
|
| 8823 |
+
"grad_norm": 4.289058208465576,
|
| 8824 |
+
"learning_rate": 1.663976147788806e-05,
|
| 8825 |
+
"loss": 3.6114,
|
| 8826 |
+
"step": 1255
|
| 8827 |
+
},
|
| 8828 |
+
{
|
| 8829 |
+
"epoch": 0.10045990801839633,
|
| 8830 |
+
"grad_norm": 4.8529181480407715,
|
| 8831 |
+
"learning_rate": 1.652651861160318e-05,
|
| 8832 |
+
"loss": 3.8212,
|
| 8833 |
+
"step": 1256
|
| 8834 |
+
},
|
| 8835 |
+
{
|
| 8836 |
+
"epoch": 0.10053989202159568,
|
| 8837 |
+
"grad_norm": 5.128431797027588,
|
| 8838 |
+
"learning_rate": 1.6413627682585965e-05,
|
| 8839 |
+
"loss": 4.3689,
|
| 8840 |
+
"step": 1257
|
| 8841 |
+
},
|
| 8842 |
+
{
|
| 8843 |
+
"epoch": 0.10061987602479504,
|
| 8844 |
+
"grad_norm": 4.320888519287109,
|
| 8845 |
+
"learning_rate": 1.630108916680223e-05,
|
| 8846 |
+
"loss": 3.6578,
|
| 8847 |
+
"step": 1258
|
| 8848 |
+
},
|
| 8849 |
+
{
|
| 8850 |
+
"epoch": 0.1006998600279944,
|
| 8851 |
+
"grad_norm": 5.23969030380249,
|
| 8852 |
+
"learning_rate": 1.6188903538731947e-05,
|
| 8853 |
+
"loss": 4.8745,
|
| 8854 |
+
"step": 1259
|
| 8855 |
+
},
|
| 8856 |
+
{
|
| 8857 |
+
"epoch": 0.10077984403119376,
|
| 8858 |
+
"grad_norm": 4.180932998657227,
|
| 8859 |
+
"learning_rate": 1.607707127136734e-05,
|
| 8860 |
+
"loss": 3.701,
|
| 8861 |
+
"step": 1260
|
| 8862 |
+
},
|
| 8863 |
+
{
|
| 8864 |
+
"epoch": 0.10085982803439313,
|
| 8865 |
+
"grad_norm": 4.653079032897949,
|
| 8866 |
+
"learning_rate": 1.5965592836210743e-05,
|
| 8867 |
+
"loss": 3.2031,
|
| 8868 |
+
"step": 1261
|
| 8869 |
+
},
|
| 8870 |
+
{
|
| 8871 |
+
"epoch": 0.10093981203759249,
|
| 8872 |
+
"grad_norm": 3.652902603149414,
|
| 8873 |
+
"learning_rate": 1.5854468703272663e-05,
|
| 8874 |
+
"loss": 2.9304,
|
| 8875 |
+
"step": 1262
|
| 8876 |
+
},
|
| 8877 |
+
{
|
| 8878 |
+
"epoch": 0.10101979604079184,
|
| 8879 |
+
"grad_norm": 3.9545648097991943,
|
| 8880 |
+
"learning_rate": 1.5743699341069895e-05,
|
| 8881 |
+
"loss": 4.0015,
|
| 8882 |
+
"step": 1263
|
| 8883 |
+
},
|
| 8884 |
+
{
|
| 8885 |
+
"epoch": 0.1010997800439912,
|
| 8886 |
+
"grad_norm": 4.3861212730407715,
|
| 8887 |
+
"learning_rate": 1.5633285216623385e-05,
|
| 8888 |
+
"loss": 3.6873,
|
| 8889 |
+
"step": 1264
|
| 8890 |
+
},
|
| 8891 |
+
{
|
| 8892 |
+
"epoch": 0.10117976404719056,
|
| 8893 |
+
"grad_norm": 3.4465882778167725,
|
| 8894 |
+
"learning_rate": 1.5523226795456347e-05,
|
| 8895 |
+
"loss": 3.1494,
|
| 8896 |
+
"step": 1265
|
| 8897 |
+
},
|
| 8898 |
+
{
|
| 8899 |
+
"epoch": 0.10125974805038992,
|
| 8900 |
+
"grad_norm": 4.774640083312988,
|
| 8901 |
+
"learning_rate": 1.541352454159237e-05,
|
| 8902 |
+
"loss": 5.2294,
|
| 8903 |
+
"step": 1266
|
| 8904 |
+
},
|
| 8905 |
+
{
|
| 8906 |
+
"epoch": 0.10133973205358929,
|
| 8907 |
+
"grad_norm": 4.005011081695557,
|
| 8908 |
+
"learning_rate": 1.5304178917553304e-05,
|
| 8909 |
+
"loss": 3.2125,
|
| 8910 |
+
"step": 1267
|
| 8911 |
+
},
|
| 8912 |
+
{
|
| 8913 |
+
"epoch": 0.10141971605678864,
|
| 8914 |
+
"grad_norm": 3.908224582672119,
|
| 8915 |
+
"learning_rate": 1.5195190384357404e-05,
|
| 8916 |
+
"loss": 2.9691,
|
| 8917 |
+
"step": 1268
|
| 8918 |
+
},
|
| 8919 |
+
{
|
| 8920 |
+
"epoch": 0.101499700059988,
|
| 8921 |
+
"grad_norm": 5.306667804718018,
|
| 8922 |
+
"learning_rate": 1.508655940151743e-05,
|
| 8923 |
+
"loss": 3.5111,
|
| 8924 |
+
"step": 1269
|
| 8925 |
+
},
|
| 8926 |
+
{
|
| 8927 |
+
"epoch": 0.10157968406318736,
|
| 8928 |
+
"grad_norm": 3.8672940731048584,
|
| 8929 |
+
"learning_rate": 1.4978286427038601e-05,
|
| 8930 |
+
"loss": 3.1632,
|
| 8931 |
+
"step": 1270
|
| 8932 |
+
},
|
| 8933 |
+
{
|
| 8934 |
+
"epoch": 0.10165966806638672,
|
| 8935 |
+
"grad_norm": 4.651573657989502,
|
| 8936 |
+
"learning_rate": 1.4870371917416703e-05,
|
| 8937 |
+
"loss": 3.8594,
|
| 8938 |
+
"step": 1271
|
| 8939 |
+
},
|
| 8940 |
+
{
|
| 8941 |
+
"epoch": 0.10173965206958609,
|
| 8942 |
+
"grad_norm": 5.070247650146484,
|
| 8943 |
+
"learning_rate": 1.4762816327636241e-05,
|
| 8944 |
+
"loss": 3.9816,
|
| 8945 |
+
"step": 1272
|
| 8946 |
+
},
|
| 8947 |
+
{
|
| 8948 |
+
"epoch": 0.10181963607278545,
|
| 8949 |
+
"grad_norm": 3.7501707077026367,
|
| 8950 |
+
"learning_rate": 1.4655620111168423e-05,
|
| 8951 |
+
"loss": 2.7913,
|
| 8952 |
+
"step": 1273
|
| 8953 |
+
},
|
| 8954 |
+
{
|
| 8955 |
+
"epoch": 0.1018996200759848,
|
| 8956 |
+
"grad_norm": 5.562809467315674,
|
| 8957 |
+
"learning_rate": 1.4548783719969239e-05,
|
| 8958 |
+
"loss": 4.0055,
|
| 8959 |
+
"step": 1274
|
| 8960 |
+
},
|
| 8961 |
+
{
|
| 8962 |
+
"epoch": 0.10197960407918416,
|
| 8963 |
+
"grad_norm": 4.067861557006836,
|
| 8964 |
+
"learning_rate": 1.444230760447769e-05,
|
| 8965 |
+
"loss": 3.2748,
|
| 8966 |
+
"step": 1275
|
| 8967 |
+
},
|
| 8968 |
+
{
|
| 8969 |
+
"epoch": 0.10205958808238352,
|
| 8970 |
+
"grad_norm": 3.589777708053589,
|
| 8971 |
+
"learning_rate": 1.4336192213613742e-05,
|
| 8972 |
+
"loss": 2.5628,
|
| 8973 |
+
"step": 1276
|
| 8974 |
+
},
|
| 8975 |
+
{
|
| 8976 |
+
"epoch": 0.10213957208558289,
|
| 8977 |
+
"grad_norm": 3.3339505195617676,
|
| 8978 |
+
"learning_rate": 1.4230437994776457e-05,
|
| 8979 |
+
"loss": 2.7425,
|
| 8980 |
+
"step": 1277
|
| 8981 |
+
},
|
| 8982 |
+
{
|
| 8983 |
+
"epoch": 0.10221955608878225,
|
| 8984 |
+
"grad_norm": 4.104794979095459,
|
| 8985 |
+
"learning_rate": 1.4125045393842219e-05,
|
| 8986 |
+
"loss": 3.3508,
|
| 8987 |
+
"step": 1278
|
| 8988 |
+
},
|
| 8989 |
+
{
|
| 8990 |
+
"epoch": 0.1022995400919816,
|
| 8991 |
+
"grad_norm": 4.165839195251465,
|
| 8992 |
+
"learning_rate": 1.4020014855162755e-05,
|
| 8993 |
+
"loss": 3.6149,
|
| 8994 |
+
"step": 1279
|
| 8995 |
+
},
|
| 8996 |
+
{
|
| 8997 |
+
"epoch": 0.10237952409518096,
|
| 8998 |
+
"grad_norm": 4.0906291007995605,
|
| 8999 |
+
"learning_rate": 1.3915346821563235e-05,
|
| 9000 |
+
"loss": 3.7574,
|
| 9001 |
+
"step": 1280
|
| 9002 |
+
},
|
| 9003 |
+
{
|
| 9004 |
+
"epoch": 0.10245950809838032,
|
| 9005 |
+
"grad_norm": 3.9316253662109375,
|
| 9006 |
+
"learning_rate": 1.3811041734340446e-05,
|
| 9007 |
+
"loss": 4.107,
|
| 9008 |
+
"step": 1281
|
| 9009 |
+
},
|
| 9010 |
+
{
|
| 9011 |
+
"epoch": 0.10253949210157969,
|
| 9012 |
+
"grad_norm": 3.3988406658172607,
|
| 9013 |
+
"learning_rate": 1.3707100033261034e-05,
|
| 9014 |
+
"loss": 2.7007,
|
| 9015 |
+
"step": 1282
|
| 9016 |
+
},
|
| 9017 |
+
{
|
| 9018 |
+
"epoch": 0.10261947610477905,
|
| 9019 |
+
"grad_norm": 3.6011903285980225,
|
| 9020 |
+
"learning_rate": 1.3603522156559446e-05,
|
| 9021 |
+
"loss": 2.6981,
|
| 9022 |
+
"step": 1283
|
| 9023 |
+
},
|
| 9024 |
+
{
|
| 9025 |
+
"epoch": 0.10269946010797841,
|
| 9026 |
+
"grad_norm": 4.022884845733643,
|
| 9027 |
+
"learning_rate": 1.3500308540936201e-05,
|
| 9028 |
+
"loss": 3.754,
|
| 9029 |
+
"step": 1284
|
| 9030 |
+
},
|
| 9031 |
+
{
|
| 9032 |
+
"epoch": 0.10277944411117776,
|
| 9033 |
+
"grad_norm": 4.574496746063232,
|
| 9034 |
+
"learning_rate": 1.339745962155613e-05,
|
| 9035 |
+
"loss": 3.4626,
|
| 9036 |
+
"step": 1285
|
| 9037 |
+
},
|
| 9038 |
+
{
|
| 9039 |
+
"epoch": 0.10285942811437712,
|
| 9040 |
+
"grad_norm": 3.933546543121338,
|
| 9041 |
+
"learning_rate": 1.3294975832046353e-05,
|
| 9042 |
+
"loss": 3.0226,
|
| 9043 |
+
"step": 1286
|
| 9044 |
+
},
|
| 9045 |
+
{
|
| 9046 |
+
"epoch": 0.10293941211757648,
|
| 9047 |
+
"grad_norm": 3.2701051235198975,
|
| 9048 |
+
"learning_rate": 1.3192857604494535e-05,
|
| 9049 |
+
"loss": 2.3685,
|
| 9050 |
+
"step": 1287
|
| 9051 |
+
},
|
| 9052 |
+
{
|
| 9053 |
+
"epoch": 0.10301939612077585,
|
| 9054 |
+
"grad_norm": 3.9511404037475586,
|
| 9055 |
+
"learning_rate": 1.3091105369447165e-05,
|
| 9056 |
+
"loss": 3.695,
|
| 9057 |
+
"step": 1288
|
| 9058 |
+
},
|
| 9059 |
+
{
|
| 9060 |
+
"epoch": 0.10309938012397521,
|
| 9061 |
+
"grad_norm": 3.584268569946289,
|
| 9062 |
+
"learning_rate": 1.2989719555907587e-05,
|
| 9063 |
+
"loss": 2.7863,
|
| 9064 |
+
"step": 1289
|
| 9065 |
+
},
|
| 9066 |
+
{
|
| 9067 |
+
"epoch": 0.10317936412717456,
|
| 9068 |
+
"grad_norm": 5.1404571533203125,
|
| 9069 |
+
"learning_rate": 1.2888700591334223e-05,
|
| 9070 |
+
"loss": 4.2853,
|
| 9071 |
+
"step": 1290
|
| 9072 |
+
},
|
| 9073 |
+
{
|
| 9074 |
+
"epoch": 0.10325934813037392,
|
| 9075 |
+
"grad_norm": 4.278750896453857,
|
| 9076 |
+
"learning_rate": 1.2788048901638917e-05,
|
| 9077 |
+
"loss": 3.5956,
|
| 9078 |
+
"step": 1291
|
| 9079 |
+
},
|
| 9080 |
+
{
|
| 9081 |
+
"epoch": 0.10333933213357328,
|
| 9082 |
+
"grad_norm": 4.515839099884033,
|
| 9083 |
+
"learning_rate": 1.2687764911184907e-05,
|
| 9084 |
+
"loss": 4.0274,
|
| 9085 |
+
"step": 1292
|
| 9086 |
+
},
|
| 9087 |
+
{
|
| 9088 |
+
"epoch": 0.10341931613677265,
|
| 9089 |
+
"grad_norm": 4.482051849365234,
|
| 9090 |
+
"learning_rate": 1.2587849042785183e-05,
|
| 9091 |
+
"loss": 3.4618,
|
| 9092 |
+
"step": 1293
|
| 9093 |
+
},
|
| 9094 |
+
{
|
| 9095 |
+
"epoch": 0.10349930013997201,
|
| 9096 |
+
"grad_norm": 4.138375759124756,
|
| 9097 |
+
"learning_rate": 1.2488301717700735e-05,
|
| 9098 |
+
"loss": 3.0179,
|
| 9099 |
+
"step": 1294
|
| 9100 |
+
},
|
| 9101 |
+
{
|
| 9102 |
+
"epoch": 0.10357928414317137,
|
| 9103 |
+
"grad_norm": 3.6072168350219727,
|
| 9104 |
+
"learning_rate": 1.2389123355638654e-05,
|
| 9105 |
+
"loss": 3.0613,
|
| 9106 |
+
"step": 1295
|
| 9107 |
+
},
|
| 9108 |
+
{
|
| 9109 |
+
"epoch": 0.10365926814637072,
|
| 9110 |
+
"grad_norm": 4.473779678344727,
|
| 9111 |
+
"learning_rate": 1.2290314374750422e-05,
|
| 9112 |
+
"loss": 3.642,
|
| 9113 |
+
"step": 1296
|
| 9114 |
+
},
|
| 9115 |
+
{
|
| 9116 |
+
"epoch": 0.10373925214957008,
|
| 9117 |
+
"grad_norm": 3.3354222774505615,
|
| 9118 |
+
"learning_rate": 1.2191875191630209e-05,
|
| 9119 |
+
"loss": 2.9358,
|
| 9120 |
+
"step": 1297
|
| 9121 |
+
},
|
| 9122 |
+
{
|
| 9123 |
+
"epoch": 0.10381923615276945,
|
| 9124 |
+
"grad_norm": 58.084869384765625,
|
| 9125 |
+
"learning_rate": 1.2093806221313008e-05,
|
| 9126 |
+
"loss": 3.8033,
|
| 9127 |
+
"step": 1298
|
| 9128 |
+
},
|
| 9129 |
+
{
|
| 9130 |
+
"epoch": 0.10389922015596881,
|
| 9131 |
+
"grad_norm": 4.5894646644592285,
|
| 9132 |
+
"learning_rate": 1.1996107877272944e-05,
|
| 9133 |
+
"loss": 4.5699,
|
| 9134 |
+
"step": 1299
|
| 9135 |
+
},
|
| 9136 |
+
{
|
| 9137 |
+
"epoch": 0.10397920415916817,
|
| 9138 |
+
"grad_norm": 3.8106391429901123,
|
| 9139 |
+
"learning_rate": 1.1898780571421552e-05,
|
| 9140 |
+
"loss": 3.0554,
|
| 9141 |
+
"step": 1300
|
| 9142 |
+
},
|
| 9143 |
+
{
|
| 9144 |
+
"epoch": 0.10405918816236753,
|
| 9145 |
+
"grad_norm": 3.7163541316986084,
|
| 9146 |
+
"learning_rate": 1.1801824714105992e-05,
|
| 9147 |
+
"loss": 2.7762,
|
| 9148 |
+
"step": 1301
|
| 9149 |
+
},
|
| 9150 |
+
{
|
| 9151 |
+
"epoch": 0.10413917216556688,
|
| 9152 |
+
"grad_norm": 4.408644676208496,
|
| 9153 |
+
"learning_rate": 1.1705240714107302e-05,
|
| 9154 |
+
"loss": 2.8706,
|
| 9155 |
+
"step": 1302
|
| 9156 |
+
},
|
| 9157 |
+
{
|
| 9158 |
+
"epoch": 0.10421915616876624,
|
| 9159 |
+
"grad_norm": 5.071597576141357,
|
| 9160 |
+
"learning_rate": 1.1609028978638803e-05,
|
| 9161 |
+
"loss": 3.8516,
|
| 9162 |
+
"step": 1303
|
| 9163 |
+
},
|
| 9164 |
+
{
|
| 9165 |
+
"epoch": 0.1042991401719656,
|
| 9166 |
+
"grad_norm": 4.749671936035156,
|
| 9167 |
+
"learning_rate": 1.1513189913344214e-05,
|
| 9168 |
+
"loss": 3.2928,
|
| 9169 |
+
"step": 1304
|
| 9170 |
+
},
|
| 9171 |
+
{
|
| 9172 |
+
"epoch": 0.10437912417516497,
|
| 9173 |
+
"grad_norm": 4.901279449462891,
|
| 9174 |
+
"learning_rate": 1.141772392229601e-05,
|
| 9175 |
+
"loss": 3.632,
|
| 9176 |
+
"step": 1305
|
| 9177 |
+
},
|
| 9178 |
+
{
|
| 9179 |
+
"epoch": 0.10445910817836433,
|
| 9180 |
+
"grad_norm": 4.000324726104736,
|
| 9181 |
+
"learning_rate": 1.1322631407993811e-05,
|
| 9182 |
+
"loss": 3.5759,
|
| 9183 |
+
"step": 1306
|
| 9184 |
+
},
|
| 9185 |
+
{
|
| 9186 |
+
"epoch": 0.10453909218156368,
|
| 9187 |
+
"grad_norm": 3.7913856506347656,
|
| 9188 |
+
"learning_rate": 1.1227912771362526e-05,
|
| 9189 |
+
"loss": 3.0771,
|
| 9190 |
+
"step": 1307
|
| 9191 |
+
},
|
| 9192 |
+
{
|
| 9193 |
+
"epoch": 0.10461907618476304,
|
| 9194 |
+
"grad_norm": 3.757390260696411,
|
| 9195 |
+
"learning_rate": 1.1133568411750727e-05,
|
| 9196 |
+
"loss": 3.5889,
|
| 9197 |
+
"step": 1308
|
| 9198 |
+
},
|
| 9199 |
+
{
|
| 9200 |
+
"epoch": 0.1046990601879624,
|
| 9201 |
+
"grad_norm": 3.9638967514038086,
|
| 9202 |
+
"learning_rate": 1.1039598726929045e-05,
|
| 9203 |
+
"loss": 3.413,
|
| 9204 |
+
"step": 1309
|
| 9205 |
+
},
|
| 9206 |
+
{
|
| 9207 |
+
"epoch": 0.10477904419116177,
|
| 9208 |
+
"grad_norm": 4.465587139129639,
|
| 9209 |
+
"learning_rate": 1.0946004113088381e-05,
|
| 9210 |
+
"loss": 3.9682,
|
| 9211 |
+
"step": 1310
|
| 9212 |
+
},
|
| 9213 |
+
{
|
| 9214 |
+
"epoch": 0.10485902819436113,
|
| 9215 |
+
"grad_norm": 5.90559720993042,
|
| 9216 |
+
"learning_rate": 1.0852784964838247e-05,
|
| 9217 |
+
"loss": 4.3722,
|
| 9218 |
+
"step": 1311
|
| 9219 |
+
},
|
| 9220 |
+
{
|
| 9221 |
+
"epoch": 0.1049390121975605,
|
| 9222 |
+
"grad_norm": 3.717155933380127,
|
| 9223 |
+
"learning_rate": 1.0759941675205221e-05,
|
| 9224 |
+
"loss": 2.7307,
|
| 9225 |
+
"step": 1312
|
| 9226 |
+
},
|
| 9227 |
+
{
|
| 9228 |
+
"epoch": 0.10501899620075984,
|
| 9229 |
+
"grad_norm": 5.315234184265137,
|
| 9230 |
+
"learning_rate": 1.0667474635631125e-05,
|
| 9231 |
+
"loss": 4.1049,
|
| 9232 |
+
"step": 1313
|
| 9233 |
+
},
|
| 9234 |
+
{
|
| 9235 |
+
"epoch": 0.1050989802039592,
|
| 9236 |
+
"grad_norm": 3.982090473175049,
|
| 9237 |
+
"learning_rate": 1.0575384235971465e-05,
|
| 9238 |
+
"loss": 3.6139,
|
| 9239 |
+
"step": 1314
|
| 9240 |
+
},
|
| 9241 |
+
{
|
| 9242 |
+
"epoch": 0.10517896420715857,
|
| 9243 |
+
"grad_norm": 3.9381678104400635,
|
| 9244 |
+
"learning_rate": 1.0483670864493778e-05,
|
| 9245 |
+
"loss": 2.9638,
|
| 9246 |
+
"step": 1315
|
| 9247 |
+
},
|
| 9248 |
+
{
|
| 9249 |
+
"epoch": 0.10525894821035793,
|
| 9250 |
+
"grad_norm": 4.617581367492676,
|
| 9251 |
+
"learning_rate": 1.0392334907876022e-05,
|
| 9252 |
+
"loss": 3.9213,
|
| 9253 |
+
"step": 1316
|
| 9254 |
+
},
|
| 9255 |
+
{
|
| 9256 |
+
"epoch": 0.1053389322135573,
|
| 9257 |
+
"grad_norm": 4.367071151733398,
|
| 9258 |
+
"learning_rate": 1.0301376751204895e-05,
|
| 9259 |
+
"loss": 4.1807,
|
| 9260 |
+
"step": 1317
|
| 9261 |
+
},
|
| 9262 |
+
{
|
| 9263 |
+
"epoch": 0.10541891621675664,
|
| 9264 |
+
"grad_norm": 3.6065549850463867,
|
| 9265 |
+
"learning_rate": 1.0210796777974197e-05,
|
| 9266 |
+
"loss": 2.5606,
|
| 9267 |
+
"step": 1318
|
| 9268 |
+
},
|
| 9269 |
+
{
|
| 9270 |
+
"epoch": 0.105498900219956,
|
| 9271 |
+
"grad_norm": 6.54305362701416,
|
| 9272 |
+
"learning_rate": 1.0120595370083318e-05,
|
| 9273 |
+
"loss": 4.5298,
|
| 9274 |
+
"step": 1319
|
| 9275 |
+
},
|
| 9276 |
+
{
|
| 9277 |
+
"epoch": 0.10557888422315537,
|
| 9278 |
+
"grad_norm": 5.466845512390137,
|
| 9279 |
+
"learning_rate": 1.0030772907835483e-05,
|
| 9280 |
+
"loss": 3.7131,
|
| 9281 |
+
"step": 1320
|
| 9282 |
+
},
|
| 9283 |
+
{
|
| 9284 |
+
"epoch": 0.10565886822635473,
|
| 9285 |
+
"grad_norm": 4.155703544616699,
|
| 9286 |
+
"learning_rate": 9.94132976993627e-06,
|
| 9287 |
+
"loss": 3.0387,
|
| 9288 |
+
"step": 1321
|
| 9289 |
+
},
|
| 9290 |
+
{
|
| 9291 |
+
"epoch": 0.10573885222955409,
|
| 9292 |
+
"grad_norm": 4.413890838623047,
|
| 9293 |
+
"learning_rate": 9.852266333491954e-06,
|
| 9294 |
+
"loss": 4.255,
|
| 9295 |
+
"step": 1322
|
| 9296 |
+
},
|
| 9297 |
+
{
|
| 9298 |
+
"epoch": 0.10581883623275345,
|
| 9299 |
+
"grad_norm": 3.982755661010742,
|
| 9300 |
+
"learning_rate": 9.763582974007934e-06,
|
| 9301 |
+
"loss": 3.3927,
|
| 9302 |
+
"step": 1323
|
| 9303 |
+
},
|
| 9304 |
+
{
|
| 9305 |
+
"epoch": 0.1058988202359528,
|
| 9306 |
+
"grad_norm": 3.7549726963043213,
|
| 9307 |
+
"learning_rate": 9.675280065387116e-06,
|
| 9308 |
+
"loss": 3.5822,
|
| 9309 |
+
"step": 1324
|
| 9310 |
+
},
|
| 9311 |
+
{
|
| 9312 |
+
"epoch": 0.10597880423915217,
|
| 9313 |
+
"grad_norm": 3.8914010524749756,
|
| 9314 |
+
"learning_rate": 9.587357979928413e-06,
|
| 9315 |
+
"loss": 3.4797,
|
| 9316 |
+
"step": 1325
|
| 9317 |
+
},
|
| 9318 |
+
{
|
| 9319 |
+
"epoch": 0.10605878824235153,
|
| 9320 |
+
"grad_norm": 3.8833811283111572,
|
| 9321 |
+
"learning_rate": 9.499817088325102e-06,
|
| 9322 |
+
"loss": 3.7417,
|
| 9323 |
+
"step": 1326
|
| 9324 |
+
},
|
| 9325 |
+
{
|
| 9326 |
+
"epoch": 0.10613877224555089,
|
| 9327 |
+
"grad_norm": 3.68992018699646,
|
| 9328 |
+
"learning_rate": 9.412657759663279e-06,
|
| 9329 |
+
"loss": 3.0596,
|
| 9330 |
+
"step": 1327
|
| 9331 |
+
},
|
| 9332 |
+
{
|
| 9333 |
+
"epoch": 0.10621875624875025,
|
| 9334 |
+
"grad_norm": 3.560129404067993,
|
| 9335 |
+
"learning_rate": 9.325880361420336e-06,
|
| 9336 |
+
"loss": 3.1886,
|
| 9337 |
+
"step": 1328
|
| 9338 |
+
},
|
| 9339 |
+
{
|
| 9340 |
+
"epoch": 0.10629874025194962,
|
| 9341 |
+
"grad_norm": 4.636751651763916,
|
| 9342 |
+
"learning_rate": 9.239485259463365e-06,
|
| 9343 |
+
"loss": 4.7497,
|
| 9344 |
+
"step": 1329
|
| 9345 |
+
},
|
| 9346 |
+
{
|
| 9347 |
+
"epoch": 0.10637872425514897,
|
| 9348 |
+
"grad_norm": 4.269623756408691,
|
| 9349 |
+
"learning_rate": 9.153472818047625e-06,
|
| 9350 |
+
"loss": 3.162,
|
| 9351 |
+
"step": 1330
|
| 9352 |
+
},
|
| 9353 |
+
{
|
| 9354 |
+
"epoch": 0.10645870825834833,
|
| 9355 |
+
"grad_norm": 4.405839920043945,
|
| 9356 |
+
"learning_rate": 9.067843399815068e-06,
|
| 9357 |
+
"loss": 3.7649,
|
| 9358 |
+
"step": 1331
|
| 9359 |
+
},
|
| 9360 |
+
{
|
| 9361 |
+
"epoch": 0.10653869226154769,
|
| 9362 |
+
"grad_norm": 4.116036891937256,
|
| 9363 |
+
"learning_rate": 8.982597365792711e-06,
|
| 9364 |
+
"loss": 3.2195,
|
| 9365 |
+
"step": 1332
|
| 9366 |
+
},
|
| 9367 |
+
{
|
| 9368 |
+
"epoch": 0.10661867626474705,
|
| 9369 |
+
"grad_norm": 3.428410530090332,
|
| 9370 |
+
"learning_rate": 8.897735075391155e-06,
|
| 9371 |
+
"loss": 2.9777,
|
| 9372 |
+
"step": 1333
|
| 9373 |
+
},
|
| 9374 |
+
{
|
| 9375 |
+
"epoch": 0.10669866026794642,
|
| 9376 |
+
"grad_norm": 4.532532215118408,
|
| 9377 |
+
"learning_rate": 8.813256886403164e-06,
|
| 9378 |
+
"loss": 3.6915,
|
| 9379 |
+
"step": 1334
|
| 9380 |
+
},
|
| 9381 |
+
{
|
| 9382 |
+
"epoch": 0.10677864427114576,
|
| 9383 |
+
"grad_norm": 4.025850772857666,
|
| 9384 |
+
"learning_rate": 8.729163155001974e-06,
|
| 9385 |
+
"loss": 3.0048,
|
| 9386 |
+
"step": 1335
|
| 9387 |
+
},
|
| 9388 |
+
{
|
| 9389 |
+
"epoch": 0.10685862827434513,
|
| 9390 |
+
"grad_norm": 4.763420104980469,
|
| 9391 |
+
"learning_rate": 8.645454235739903e-06,
|
| 9392 |
+
"loss": 4.045,
|
| 9393 |
+
"step": 1336
|
| 9394 |
+
},
|
| 9395 |
+
{
|
| 9396 |
+
"epoch": 0.10693861227754449,
|
| 9397 |
+
"grad_norm": 6.281853199005127,
|
| 9398 |
+
"learning_rate": 8.562130481546903e-06,
|
| 9399 |
+
"loss": 2.8622,
|
| 9400 |
+
"step": 1337
|
| 9401 |
+
},
|
| 9402 |
+
{
|
| 9403 |
+
"epoch": 0.10701859628074385,
|
| 9404 |
+
"grad_norm": 3.804391384124756,
|
| 9405 |
+
"learning_rate": 8.479192243728962e-06,
|
| 9406 |
+
"loss": 2.9874,
|
| 9407 |
+
"step": 1338
|
| 9408 |
+
},
|
| 9409 |
+
{
|
| 9410 |
+
"epoch": 0.10709858028394321,
|
| 9411 |
+
"grad_norm": 4.4125447273254395,
|
| 9412 |
+
"learning_rate": 8.39663987196665e-06,
|
| 9413 |
+
"loss": 3.7106,
|
| 9414 |
+
"step": 1339
|
| 9415 |
+
},
|
| 9416 |
+
{
|
| 9417 |
+
"epoch": 0.10717856428714258,
|
| 9418 |
+
"grad_norm": 3.222792863845825,
|
| 9419 |
+
"learning_rate": 8.314473714313719e-06,
|
| 9420 |
+
"loss": 2.6159,
|
| 9421 |
+
"step": 1340
|
| 9422 |
+
},
|
| 9423 |
+
{
|
| 9424 |
+
"epoch": 0.10725854829034193,
|
| 9425 |
+
"grad_norm": 4.708176612854004,
|
| 9426 |
+
"learning_rate": 8.232694117195538e-06,
|
| 9427 |
+
"loss": 3.6817,
|
| 9428 |
+
"step": 1341
|
| 9429 |
+
},
|
| 9430 |
+
{
|
| 9431 |
+
"epoch": 0.10733853229354129,
|
| 9432 |
+
"grad_norm": 5.3699631690979,
|
| 9433 |
+
"learning_rate": 8.151301425407699e-06,
|
| 9434 |
+
"loss": 3.5441,
|
| 9435 |
+
"step": 1342
|
| 9436 |
+
},
|
| 9437 |
+
{
|
| 9438 |
+
"epoch": 0.10741851629674065,
|
| 9439 |
+
"grad_norm": 3.633037805557251,
|
| 9440 |
+
"learning_rate": 8.070295982114539e-06,
|
| 9441 |
+
"loss": 3.1749,
|
| 9442 |
+
"step": 1343
|
| 9443 |
+
},
|
| 9444 |
+
{
|
| 9445 |
+
"epoch": 0.10749850029994001,
|
| 9446 |
+
"grad_norm": 3.714905261993408,
|
| 9447 |
+
"learning_rate": 7.9896781288477e-06,
|
| 9448 |
+
"loss": 2.5333,
|
| 9449 |
+
"step": 1344
|
| 9450 |
+
},
|
| 9451 |
+
{
|
| 9452 |
+
"epoch": 0.10757848430313938,
|
| 9453 |
+
"grad_norm": 4.083629131317139,
|
| 9454 |
+
"learning_rate": 7.909448205504632e-06,
|
| 9455 |
+
"loss": 3.3521,
|
| 9456 |
+
"step": 1345
|
| 9457 |
+
},
|
| 9458 |
+
{
|
| 9459 |
+
"epoch": 0.10765846830633873,
|
| 9460 |
+
"grad_norm": 4.123989105224609,
|
| 9461 |
+
"learning_rate": 7.829606550347313e-06,
|
| 9462 |
+
"loss": 3.6748,
|
| 9463 |
+
"step": 1346
|
| 9464 |
+
},
|
| 9465 |
+
{
|
| 9466 |
+
"epoch": 0.10773845230953809,
|
| 9467 |
+
"grad_norm": 3.817969560623169,
|
| 9468 |
+
"learning_rate": 7.750153500000624e-06,
|
| 9469 |
+
"loss": 2.841,
|
| 9470 |
+
"step": 1347
|
| 9471 |
+
},
|
| 9472 |
+
{
|
| 9473 |
+
"epoch": 0.10781843631273745,
|
| 9474 |
+
"grad_norm": 3.5868868827819824,
|
| 9475 |
+
"learning_rate": 7.671089389451058e-06,
|
| 9476 |
+
"loss": 3.1438,
|
| 9477 |
+
"step": 1348
|
| 9478 |
+
},
|
| 9479 |
+
{
|
| 9480 |
+
"epoch": 0.10789842031593681,
|
| 9481 |
+
"grad_norm": 4.903133392333984,
|
| 9482 |
+
"learning_rate": 7.592414552045324e-06,
|
| 9483 |
+
"loss": 3.7662,
|
| 9484 |
+
"step": 1349
|
| 9485 |
+
},
|
| 9486 |
+
{
|
| 9487 |
+
"epoch": 0.10797840431913618,
|
| 9488 |
+
"grad_norm": 3.7668721675872803,
|
| 9489 |
+
"learning_rate": 7.514129319488839e-06,
|
| 9490 |
+
"loss": 3.4625,
|
| 9491 |
+
"step": 1350
|
| 9492 |
+
},
|
| 9493 |
+
{
|
| 9494 |
+
"epoch": 0.10805838832233554,
|
| 9495 |
+
"grad_norm": 4.727697372436523,
|
| 9496 |
+
"learning_rate": 7.43623402184438e-06,
|
| 9497 |
+
"loss": 3.9206,
|
| 9498 |
+
"step": 1351
|
| 9499 |
+
},
|
| 9500 |
+
{
|
| 9501 |
+
"epoch": 0.10813837232553489,
|
| 9502 |
+
"grad_norm": 4.111035346984863,
|
| 9503 |
+
"learning_rate": 7.358728987530728e-06,
|
| 9504 |
+
"loss": 3.4021,
|
| 9505 |
+
"step": 1352
|
| 9506 |
+
},
|
| 9507 |
+
{
|
| 9508 |
+
"epoch": 0.10821835632873425,
|
| 9509 |
+
"grad_norm": 5.181998252868652,
|
| 9510 |
+
"learning_rate": 7.281614543321269e-06,
|
| 9511 |
+
"loss": 4.448,
|
| 9512 |
+
"step": 1353
|
| 9513 |
+
},
|
| 9514 |
+
{
|
| 9515 |
+
"epoch": 0.10829834033193361,
|
| 9516 |
+
"grad_norm": 4.558880805969238,
|
| 9517 |
+
"learning_rate": 7.204891014342552e-06,
|
| 9518 |
+
"loss": 4.3893,
|
| 9519 |
+
"step": 1354
|
| 9520 |
+
},
|
| 9521 |
+
{
|
| 9522 |
+
"epoch": 0.10837832433513298,
|
| 9523 |
+
"grad_norm": 4.556720733642578,
|
| 9524 |
+
"learning_rate": 7.128558724072976e-06,
|
| 9525 |
+
"loss": 2.9553,
|
| 9526 |
+
"step": 1355
|
| 9527 |
+
},
|
| 9528 |
+
{
|
| 9529 |
+
"epoch": 0.10845830833833234,
|
| 9530 |
+
"grad_norm": 3.7258121967315674,
|
| 9531 |
+
"learning_rate": 7.052617994341448e-06,
|
| 9532 |
+
"loss": 2.8388,
|
| 9533 |
+
"step": 1356
|
| 9534 |
+
},
|
| 9535 |
+
{
|
| 9536 |
+
"epoch": 0.1085382923415317,
|
| 9537 |
+
"grad_norm": 3.7325799465179443,
|
| 9538 |
+
"learning_rate": 6.977069145325987e-06,
|
| 9539 |
+
"loss": 2.4219,
|
| 9540 |
+
"step": 1357
|
| 9541 |
+
},
|
| 9542 |
+
{
|
| 9543 |
+
"epoch": 0.10861827634473105,
|
| 9544 |
+
"grad_norm": 4.148155212402344,
|
| 9545 |
+
"learning_rate": 6.901912495552332e-06,
|
| 9546 |
+
"loss": 3.422,
|
| 9547 |
+
"step": 1358
|
| 9548 |
+
},
|
| 9549 |
+
{
|
| 9550 |
+
"epoch": 0.10869826034793041,
|
| 9551 |
+
"grad_norm": 3.9268195629119873,
|
| 9552 |
+
"learning_rate": 6.827148361892721e-06,
|
| 9553 |
+
"loss": 2.9873,
|
| 9554 |
+
"step": 1359
|
| 9555 |
+
},
|
| 9556 |
+
{
|
| 9557 |
+
"epoch": 0.10877824435112977,
|
| 9558 |
+
"grad_norm": 3.8969569206237793,
|
| 9559 |
+
"learning_rate": 6.75277705956443e-06,
|
| 9560 |
+
"loss": 3.2832,
|
| 9561 |
+
"step": 1360
|
| 9562 |
+
},
|
| 9563 |
+
{
|
| 9564 |
+
"epoch": 0.10885822835432914,
|
| 9565 |
+
"grad_norm": 3.398939847946167,
|
| 9566 |
+
"learning_rate": 6.6787989021285135e-06,
|
| 9567 |
+
"loss": 2.8888,
|
| 9568 |
+
"step": 1361
|
| 9569 |
+
},
|
| 9570 |
+
{
|
| 9571 |
+
"epoch": 0.1089382123575285,
|
| 9572 |
+
"grad_norm": 3.8946545124053955,
|
| 9573 |
+
"learning_rate": 6.605214201488486e-06,
|
| 9574 |
+
"loss": 3.8039,
|
| 9575 |
+
"step": 1362
|
| 9576 |
+
},
|
| 9577 |
+
{
|
| 9578 |
+
"epoch": 0.10901819636072785,
|
| 9579 |
+
"grad_norm": 4.92830228805542,
|
| 9580 |
+
"learning_rate": 6.5320232678889536e-06,
|
| 9581 |
+
"loss": 3.6133,
|
| 9582 |
+
"step": 1363
|
| 9583 |
+
},
|
| 9584 |
+
{
|
| 9585 |
+
"epoch": 0.10909818036392721,
|
| 9586 |
+
"grad_norm": 3.7663941383361816,
|
| 9587 |
+
"learning_rate": 6.459226409914332e-06,
|
| 9588 |
+
"loss": 3.2744,
|
| 9589 |
+
"step": 1364
|
| 9590 |
+
},
|
| 9591 |
+
{
|
| 9592 |
+
"epoch": 0.10917816436712657,
|
| 9593 |
+
"grad_norm": 4.027180194854736,
|
| 9594 |
+
"learning_rate": 6.386823934487618e-06,
|
| 9595 |
+
"loss": 3.7264,
|
| 9596 |
+
"step": 1365
|
| 9597 |
+
},
|
| 9598 |
+
{
|
| 9599 |
+
"epoch": 0.10925814837032594,
|
| 9600 |
+
"grad_norm": 4.098419189453125,
|
| 9601 |
+
"learning_rate": 6.314816146868952e-06,
|
| 9602 |
+
"loss": 3.0471,
|
| 9603 |
+
"step": 1366
|
| 9604 |
+
},
|
| 9605 |
+
{
|
| 9606 |
+
"epoch": 0.1093381323735253,
|
| 9607 |
+
"grad_norm": 5.7975263595581055,
|
| 9608 |
+
"learning_rate": 6.243203350654437e-06,
|
| 9609 |
+
"loss": 3.7878,
|
| 9610 |
+
"step": 1367
|
| 9611 |
+
},
|
| 9612 |
+
{
|
| 9613 |
+
"epoch": 0.10941811637672466,
|
| 9614 |
+
"grad_norm": 4.690611362457275,
|
| 9615 |
+
"learning_rate": 6.171985847774864e-06,
|
| 9616 |
+
"loss": 3.6505,
|
| 9617 |
+
"step": 1368
|
| 9618 |
+
},
|
| 9619 |
+
{
|
| 9620 |
+
"epoch": 0.10949810037992401,
|
| 9621 |
+
"grad_norm": 4.184288501739502,
|
| 9622 |
+
"learning_rate": 6.1011639384943585e-06,
|
| 9623 |
+
"loss": 3.7925,
|
| 9624 |
+
"step": 1369
|
| 9625 |
+
},
|
| 9626 |
+
{
|
| 9627 |
+
"epoch": 0.10957808438312337,
|
| 9628 |
+
"grad_norm": 5.111532211303711,
|
| 9629 |
+
"learning_rate": 6.030737921409169e-06,
|
| 9630 |
+
"loss": 2.8176,
|
| 9631 |
+
"step": 1370
|
| 9632 |
+
},
|
| 9633 |
+
{
|
| 9634 |
+
"epoch": 0.10965806838632274,
|
| 9635 |
+
"grad_norm": 4.371936798095703,
|
| 9636 |
+
"learning_rate": 5.960708093446432e-06,
|
| 9637 |
+
"loss": 4.6534,
|
| 9638 |
+
"step": 1371
|
| 9639 |
+
},
|
| 9640 |
+
{
|
| 9641 |
+
"epoch": 0.1097380523895221,
|
| 9642 |
+
"grad_norm": 3.959766149520874,
|
| 9643 |
+
"learning_rate": 5.891074749862857e-06,
|
| 9644 |
+
"loss": 3.7773,
|
| 9645 |
+
"step": 1372
|
| 9646 |
+
},
|
| 9647 |
+
{
|
| 9648 |
+
"epoch": 0.10981803639272146,
|
| 9649 |
+
"grad_norm": 4.102706432342529,
|
| 9650 |
+
"learning_rate": 5.8218381842435e-06,
|
| 9651 |
+
"loss": 3.2274,
|
| 9652 |
+
"step": 1373
|
| 9653 |
+
},
|
| 9654 |
+
{
|
| 9655 |
+
"epoch": 0.10989802039592081,
|
| 9656 |
+
"grad_norm": 4.045333385467529,
|
| 9657 |
+
"learning_rate": 5.75299868850061e-06,
|
| 9658 |
+
"loss": 3.2656,
|
| 9659 |
+
"step": 1374
|
| 9660 |
+
},
|
| 9661 |
+
{
|
| 9662 |
+
"epoch": 0.10997800439912017,
|
| 9663 |
+
"grad_norm": 4.71077823638916,
|
| 9664 |
+
"learning_rate": 5.684556552872256e-06,
|
| 9665 |
+
"loss": 4.4475,
|
| 9666 |
+
"step": 1375
|
| 9667 |
+
},
|
| 9668 |
+
{
|
| 9669 |
+
"epoch": 0.11005798840231953,
|
| 9670 |
+
"grad_norm": 3.2355942726135254,
|
| 9671 |
+
"learning_rate": 5.616512065921187e-06,
|
| 9672 |
+
"loss": 3.0672,
|
| 9673 |
+
"step": 1376
|
| 9674 |
+
},
|
| 9675 |
+
{
|
| 9676 |
+
"epoch": 0.1101379724055189,
|
| 9677 |
+
"grad_norm": 4.2143731117248535,
|
| 9678 |
+
"learning_rate": 5.548865514533641e-06,
|
| 9679 |
+
"loss": 3.3803,
|
| 9680 |
+
"step": 1377
|
| 9681 |
+
},
|
| 9682 |
+
{
|
| 9683 |
+
"epoch": 0.11021795640871826,
|
| 9684 |
+
"grad_norm": 4.2557172775268555,
|
| 9685 |
+
"learning_rate": 5.481617183918053e-06,
|
| 9686 |
+
"loss": 3.6468,
|
| 9687 |
+
"step": 1378
|
| 9688 |
+
},
|
| 9689 |
+
{
|
| 9690 |
+
"epoch": 0.11029794041191762,
|
| 9691 |
+
"grad_norm": 3.4788947105407715,
|
| 9692 |
+
"learning_rate": 5.4147673576039e-06,
|
| 9693 |
+
"loss": 2.6879,
|
| 9694 |
+
"step": 1379
|
| 9695 |
+
},
|
| 9696 |
+
{
|
| 9697 |
+
"epoch": 0.11037792441511697,
|
| 9698 |
+
"grad_norm": 3.5003573894500732,
|
| 9699 |
+
"learning_rate": 5.348316317440549e-06,
|
| 9700 |
+
"loss": 3.0834,
|
| 9701 |
+
"step": 1380
|
| 9702 |
+
},
|
| 9703 |
+
{
|
| 9704 |
+
"epoch": 0.11045790841831633,
|
| 9705 |
+
"grad_norm": 4.086651802062988,
|
| 9706 |
+
"learning_rate": 5.282264343595978e-06,
|
| 9707 |
+
"loss": 3.3918,
|
| 9708 |
+
"step": 1381
|
| 9709 |
+
},
|
| 9710 |
+
{
|
| 9711 |
+
"epoch": 0.1105378924215157,
|
| 9712 |
+
"grad_norm": 3.577976942062378,
|
| 9713 |
+
"learning_rate": 5.21661171455563e-06,
|
| 9714 |
+
"loss": 2.9443,
|
| 9715 |
+
"step": 1382
|
| 9716 |
+
},
|
| 9717 |
+
{
|
| 9718 |
+
"epoch": 0.11061787642471506,
|
| 9719 |
+
"grad_norm": 3.582505464553833,
|
| 9720 |
+
"learning_rate": 5.151358707121301e-06,
|
| 9721 |
+
"loss": 2.7432,
|
| 9722 |
+
"step": 1383
|
| 9723 |
+
},
|
| 9724 |
+
{
|
| 9725 |
+
"epoch": 0.11069786042791442,
|
| 9726 |
+
"grad_norm": 5.2246198654174805,
|
| 9727 |
+
"learning_rate": 5.086505596409885e-06,
|
| 9728 |
+
"loss": 3.2871,
|
| 9729 |
+
"step": 1384
|
| 9730 |
+
},
|
| 9731 |
+
{
|
| 9732 |
+
"epoch": 0.11077784443111378,
|
| 9733 |
+
"grad_norm": 4.0024614334106445,
|
| 9734 |
+
"learning_rate": 5.0220526558522274e-06,
|
| 9735 |
+
"loss": 3.3569,
|
| 9736 |
+
"step": 1385
|
| 9737 |
+
},
|
| 9738 |
+
{
|
| 9739 |
+
"epoch": 0.11085782843431313,
|
| 9740 |
+
"grad_norm": 5.208253383636475,
|
| 9741 |
+
"learning_rate": 4.958000157192022e-06,
|
| 9742 |
+
"loss": 4.7191,
|
| 9743 |
+
"step": 1386
|
| 9744 |
+
},
|
| 9745 |
+
{
|
| 9746 |
+
"epoch": 0.1109378124375125,
|
| 9747 |
+
"grad_norm": 4.149219512939453,
|
| 9748 |
+
"learning_rate": 4.8943483704846475e-06,
|
| 9749 |
+
"loss": 3.8205,
|
| 9750 |
+
"step": 1387
|
| 9751 |
+
},
|
| 9752 |
+
{
|
| 9753 |
+
"epoch": 0.11101779644071186,
|
| 9754 |
+
"grad_norm": 4.099031448364258,
|
| 9755 |
+
"learning_rate": 4.831097564095999e-06,
|
| 9756 |
+
"loss": 3.699,
|
| 9757 |
+
"step": 1388
|
| 9758 |
+
},
|
| 9759 |
+
{
|
| 9760 |
+
"epoch": 0.11109778044391122,
|
| 9761 |
+
"grad_norm": 5.425239562988281,
|
| 9762 |
+
"learning_rate": 4.768248004701359e-06,
|
| 9763 |
+
"loss": 4.9931,
|
| 9764 |
+
"step": 1389
|
| 9765 |
+
},
|
| 9766 |
+
{
|
| 9767 |
+
"epoch": 0.11117776444711058,
|
| 9768 |
+
"grad_norm": 4.155279159545898,
|
| 9769 |
+
"learning_rate": 4.705799957284351e-06,
|
| 9770 |
+
"loss": 2.499,
|
| 9771 |
+
"step": 1390
|
| 9772 |
+
},
|
| 9773 |
+
{
|
| 9774 |
+
"epoch": 0.11125774845030993,
|
| 9775 |
+
"grad_norm": 4.97507381439209,
|
| 9776 |
+
"learning_rate": 4.643753685135688e-06,
|
| 9777 |
+
"loss": 3.4487,
|
| 9778 |
+
"step": 1391
|
| 9779 |
+
},
|
| 9780 |
+
{
|
| 9781 |
+
"epoch": 0.1113377324535093,
|
| 9782 |
+
"grad_norm": 3.454404592514038,
|
| 9783 |
+
"learning_rate": 4.582109449852168e-06,
|
| 9784 |
+
"loss": 3.2001,
|
| 9785 |
+
"step": 1392
|
| 9786 |
+
},
|
| 9787 |
+
{
|
| 9788 |
+
"epoch": 0.11141771645670866,
|
| 9789 |
+
"grad_norm": 4.447368621826172,
|
| 9790 |
+
"learning_rate": 4.520867511335569e-06,
|
| 9791 |
+
"loss": 4.0092,
|
| 9792 |
+
"step": 1393
|
| 9793 |
+
},
|
| 9794 |
+
{
|
| 9795 |
+
"epoch": 0.11149770045990802,
|
| 9796 |
+
"grad_norm": 3.826348066329956,
|
| 9797 |
+
"learning_rate": 4.4600281277914715e-06,
|
| 9798 |
+
"loss": 3.0956,
|
| 9799 |
+
"step": 1394
|
| 9800 |
+
},
|
| 9801 |
+
{
|
| 9802 |
+
"epoch": 0.11157768446310738,
|
| 9803 |
+
"grad_norm": 3.4276773929595947,
|
| 9804 |
+
"learning_rate": 4.399591555728233e-06,
|
| 9805 |
+
"loss": 2.9562,
|
| 9806 |
+
"step": 1395
|
| 9807 |
+
},
|
| 9808 |
+
{
|
| 9809 |
+
"epoch": 0.11165766846630674,
|
| 9810 |
+
"grad_norm": 3.9827113151550293,
|
| 9811 |
+
"learning_rate": 4.339558049955927e-06,
|
| 9812 |
+
"loss": 2.9519,
|
| 9813 |
+
"step": 1396
|
| 9814 |
+
},
|
| 9815 |
+
{
|
| 9816 |
+
"epoch": 0.1117376524695061,
|
| 9817 |
+
"grad_norm": 3.802614212036133,
|
| 9818 |
+
"learning_rate": 4.279927863585198e-06,
|
| 9819 |
+
"loss": 2.7602,
|
| 9820 |
+
"step": 1397
|
| 9821 |
+
},
|
| 9822 |
+
{
|
| 9823 |
+
"epoch": 0.11181763647270546,
|
| 9824 |
+
"grad_norm": 3.994422435760498,
|
| 9825 |
+
"learning_rate": 4.220701248026248e-06,
|
| 9826 |
+
"loss": 4.1273,
|
| 9827 |
+
"step": 1398
|
| 9828 |
+
},
|
| 9829 |
+
{
|
| 9830 |
+
"epoch": 0.11189762047590482,
|
| 9831 |
+
"grad_norm": 5.223484992980957,
|
| 9832 |
+
"learning_rate": 4.161878452987778e-06,
|
| 9833 |
+
"loss": 4.2757,
|
| 9834 |
+
"step": 1399
|
| 9835 |
+
},
|
| 9836 |
+
{
|
| 9837 |
+
"epoch": 0.11197760447910418,
|
| 9838 |
+
"grad_norm": 3.9088292121887207,
|
| 9839 |
+
"learning_rate": 4.103459726475889e-06,
|
| 9840 |
+
"loss": 3.6747,
|
| 9841 |
+
"step": 1400
|
| 9842 |
+
},
|
| 9843 |
+
{
|
| 9844 |
+
"epoch": 0.11205758848230354,
|
| 9845 |
+
"grad_norm": 7.126443862915039,
|
| 9846 |
+
"learning_rate": 4.045445314793073e-06,
|
| 9847 |
+
"loss": 3.1352,
|
| 9848 |
+
"step": 1401
|
| 9849 |
+
},
|
| 9850 |
+
{
|
| 9851 |
+
"epoch": 0.11213757248550289,
|
| 9852 |
+
"grad_norm": 4.137068271636963,
|
| 9853 |
+
"learning_rate": 3.987835462537193e-06,
|
| 9854 |
+
"loss": 3.4167,
|
| 9855 |
+
"step": 1402
|
| 9856 |
+
},
|
| 9857 |
+
{
|
| 9858 |
+
"epoch": 0.11221755648870226,
|
| 9859 |
+
"grad_norm": 3.08945631980896,
|
| 9860 |
+
"learning_rate": 3.9306304126004e-06,
|
| 9861 |
+
"loss": 2.4006,
|
| 9862 |
+
"step": 1403
|
| 9863 |
+
},
|
| 9864 |
+
{
|
| 9865 |
+
"epoch": 0.11229754049190162,
|
| 9866 |
+
"grad_norm": 4.775023937225342,
|
| 9867 |
+
"learning_rate": 3.873830406168111e-06,
|
| 9868 |
+
"loss": 4.2971,
|
| 9869 |
+
"step": 1404
|
| 9870 |
+
},
|
| 9871 |
+
{
|
| 9872 |
+
"epoch": 0.11237752449510098,
|
| 9873 |
+
"grad_norm": 5.087342739105225,
|
| 9874 |
+
"learning_rate": 3.817435682718096e-06,
|
| 9875 |
+
"loss": 4.5239,
|
| 9876 |
+
"step": 1405
|
| 9877 |
+
},
|
| 9878 |
+
{
|
| 9879 |
+
"epoch": 0.11245750849830034,
|
| 9880 |
+
"grad_norm": 18.801504135131836,
|
| 9881 |
+
"learning_rate": 3.761446480019315e-06,
|
| 9882 |
+
"loss": 4.4807,
|
| 9883 |
+
"step": 1406
|
| 9884 |
+
},
|
| 9885 |
+
{
|
| 9886 |
+
"epoch": 0.1125374925014997,
|
| 9887 |
+
"grad_norm": 4.829584121704102,
|
| 9888 |
+
"learning_rate": 3.7058630341310075e-06,
|
| 9889 |
+
"loss": 3.4067,
|
| 9890 |
+
"step": 1407
|
| 9891 |
+
},
|
| 9892 |
+
{
|
| 9893 |
+
"epoch": 0.11261747650469905,
|
| 9894 |
+
"grad_norm": 4.73903226852417,
|
| 9895 |
+
"learning_rate": 3.6506855794016913e-06,
|
| 9896 |
+
"loss": 4.2005,
|
| 9897 |
+
"step": 1408
|
| 9898 |
+
},
|
| 9899 |
+
{
|
| 9900 |
+
"epoch": 0.11269746050789842,
|
| 9901 |
+
"grad_norm": 3.212651014328003,
|
| 9902 |
+
"learning_rate": 3.595914348468166e-06,
|
| 9903 |
+
"loss": 2.418,
|
| 9904 |
+
"step": 1409
|
| 9905 |
+
},
|
| 9906 |
+
{
|
| 9907 |
+
"epoch": 0.11277744451109778,
|
| 9908 |
+
"grad_norm": 4.14359188079834,
|
| 9909 |
+
"learning_rate": 3.541549572254488e-06,
|
| 9910 |
+
"loss": 3.7558,
|
| 9911 |
+
"step": 1410
|
| 9912 |
+
},
|
| 9913 |
+
{
|
| 9914 |
+
"epoch": 0.11285742851429714,
|
| 9915 |
+
"grad_norm": 4.382094383239746,
|
| 9916 |
+
"learning_rate": 3.4875914799710974e-06,
|
| 9917 |
+
"loss": 3.9129,
|
| 9918 |
+
"step": 1411
|
| 9919 |
+
},
|
| 9920 |
+
{
|
| 9921 |
+
"epoch": 0.1129374125174965,
|
| 9922 |
+
"grad_norm": 3.705204725265503,
|
| 9923 |
+
"learning_rate": 3.43404029911375e-06,
|
| 9924 |
+
"loss": 3.7869,
|
| 9925 |
+
"step": 1412
|
| 9926 |
+
},
|
| 9927 |
+
{
|
| 9928 |
+
"epoch": 0.11301739652069587,
|
| 9929 |
+
"grad_norm": 4.701824188232422,
|
| 9930 |
+
"learning_rate": 3.3808962554625958e-06,
|
| 9931 |
+
"loss": 4.0207,
|
| 9932 |
+
"step": 1413
|
| 9933 |
+
},
|
| 9934 |
+
{
|
| 9935 |
+
"epoch": 0.11309738052389522,
|
| 9936 |
+
"grad_norm": 5.10697603225708,
|
| 9937 |
+
"learning_rate": 3.3281595730812575e-06,
|
| 9938 |
+
"loss": 4.2201,
|
| 9939 |
+
"step": 1414
|
| 9940 |
+
},
|
| 9941 |
+
{
|
| 9942 |
+
"epoch": 0.11317736452709458,
|
| 9943 |
+
"grad_norm": 3.4379613399505615,
|
| 9944 |
+
"learning_rate": 3.275830474315855e-06,
|
| 9945 |
+
"loss": 2.4332,
|
| 9946 |
+
"step": 1415
|
| 9947 |
+
},
|
| 9948 |
+
{
|
| 9949 |
+
"epoch": 0.11325734853029394,
|
| 9950 |
+
"grad_norm": 4.493072509765625,
|
| 9951 |
+
"learning_rate": 3.223909179794027e-06,
|
| 9952 |
+
"loss": 3.4741,
|
| 9953 |
+
"step": 1416
|
| 9954 |
+
},
|
| 9955 |
+
{
|
| 9956 |
+
"epoch": 0.1133373325334933,
|
| 9957 |
+
"grad_norm": 3.4767467975616455,
|
| 9958 |
+
"learning_rate": 3.1723959084241194e-06,
|
| 9959 |
+
"loss": 2.7417,
|
| 9960 |
+
"step": 1417
|
| 9961 |
+
},
|
| 9962 |
+
{
|
| 9963 |
+
"epoch": 0.11341731653669267,
|
| 9964 |
+
"grad_norm": 4.231232166290283,
|
| 9965 |
+
"learning_rate": 3.121290877394134e-06,
|
| 9966 |
+
"loss": 3.027,
|
| 9967 |
+
"step": 1418
|
| 9968 |
+
},
|
| 9969 |
+
{
|
| 9970 |
+
"epoch": 0.11349730053989202,
|
| 9971 |
+
"grad_norm": 4.284804344177246,
|
| 9972 |
+
"learning_rate": 3.0705943021708704e-06,
|
| 9973 |
+
"loss": 3.9684,
|
| 9974 |
+
"step": 1419
|
| 9975 |
+
},
|
| 9976 |
+
{
|
| 9977 |
+
"epoch": 0.11357728454309138,
|
| 9978 |
+
"grad_norm": 4.5746283531188965,
|
| 9979 |
+
"learning_rate": 3.0203063964990617e-06,
|
| 9980 |
+
"loss": 4.3571,
|
| 9981 |
+
"step": 1420
|
| 9982 |
+
},
|
| 9983 |
+
{
|
| 9984 |
+
"epoch": 0.11365726854629074,
|
| 9985 |
+
"grad_norm": 4.382323741912842,
|
| 9986 |
+
"learning_rate": 2.970427372400353e-06,
|
| 9987 |
+
"loss": 3.6244,
|
| 9988 |
+
"step": 1421
|
| 9989 |
+
},
|
| 9990 |
+
{
|
| 9991 |
+
"epoch": 0.1137372525494901,
|
| 9992 |
+
"grad_norm": 3.819322109222412,
|
| 9993 |
+
"learning_rate": 2.9209574401725557e-06,
|
| 9994 |
+
"loss": 3.1486,
|
| 9995 |
+
"step": 1422
|
| 9996 |
+
},
|
| 9997 |
+
{
|
| 9998 |
+
"epoch": 0.11381723655268947,
|
| 9999 |
+
"grad_norm": 4.852266788482666,
|
| 10000 |
+
"learning_rate": 2.8718968083886075e-06,
|
| 10001 |
+
"loss": 3.8628,
|
| 10002 |
+
"step": 1423
|
| 10003 |
+
},
|
| 10004 |
+
{
|
| 10005 |
+
"epoch": 0.11389722055588883,
|
| 10006 |
+
"grad_norm": 4.21525764465332,
|
| 10007 |
+
"learning_rate": 2.82324568389587e-06,
|
| 10008 |
+
"loss": 2.9995,
|
| 10009 |
+
"step": 1424
|
| 10010 |
+
},
|
| 10011 |
+
{
|
| 10012 |
+
"epoch": 0.11397720455908818,
|
| 10013 |
+
"grad_norm": 5.286691665649414,
|
| 10014 |
+
"learning_rate": 2.7750042718150516e-06,
|
| 10015 |
+
"loss": 4.7328,
|
| 10016 |
+
"step": 1425
|
| 10017 |
+
},
|
| 10018 |
+
{
|
| 10019 |
+
"epoch": 0.11405718856228754,
|
| 10020 |
+
"grad_norm": 4.92333459854126,
|
| 10021 |
+
"learning_rate": 2.7271727755395214e-06,
|
| 10022 |
+
"loss": 4.0131,
|
| 10023 |
+
"step": 1426
|
| 10024 |
+
},
|
| 10025 |
+
{
|
| 10026 |
+
"epoch": 0.1141371725654869,
|
| 10027 |
+
"grad_norm": 4.312798976898193,
|
| 10028 |
+
"learning_rate": 2.6797513967343645e-06,
|
| 10029 |
+
"loss": 3.5426,
|
| 10030 |
+
"step": 1427
|
| 10031 |
+
},
|
| 10032 |
+
{
|
| 10033 |
+
"epoch": 0.11421715656868626,
|
| 10034 |
+
"grad_norm": 3.709238052368164,
|
| 10035 |
+
"learning_rate": 2.6327403353355264e-06,
|
| 10036 |
+
"loss": 3.4297,
|
| 10037 |
+
"step": 1428
|
| 10038 |
+
},
|
| 10039 |
+
{
|
| 10040 |
+
"epoch": 0.11429714057188563,
|
| 10041 |
+
"grad_norm": 3.7097280025482178,
|
| 10042 |
+
"learning_rate": 2.586139789548991e-06,
|
| 10043 |
+
"loss": 3.3895,
|
| 10044 |
+
"step": 1429
|
| 10045 |
+
},
|
| 10046 |
+
{
|
| 10047 |
+
"epoch": 0.11437712457508498,
|
| 10048 |
+
"grad_norm": 3.967355251312256,
|
| 10049 |
+
"learning_rate": 2.539949955849985e-06,
|
| 10050 |
+
"loss": 3.6372,
|
| 10051 |
+
"step": 1430
|
| 10052 |
+
},
|
| 10053 |
+
{
|
| 10054 |
+
"epoch": 0.11445710857828434,
|
| 10055 |
+
"grad_norm": 4.879059791564941,
|
| 10056 |
+
"learning_rate": 2.494171028982062e-06,
|
| 10057 |
+
"loss": 2.6665,
|
| 10058 |
+
"step": 1431
|
| 10059 |
+
},
|
| 10060 |
+
{
|
| 10061 |
+
"epoch": 0.1145370925814837,
|
| 10062 |
+
"grad_norm": 4.5773515701293945,
|
| 10063 |
+
"learning_rate": 2.4488032019563402e-06,
|
| 10064 |
+
"loss": 3.8109,
|
| 10065 |
+
"step": 1432
|
| 10066 |
+
},
|
| 10067 |
+
{
|
| 10068 |
+
"epoch": 0.11461707658468306,
|
| 10069 |
+
"grad_norm": 6.647262096405029,
|
| 10070 |
+
"learning_rate": 2.403846666050691e-06,
|
| 10071 |
+
"loss": 2.8738,
|
| 10072 |
+
"step": 1433
|
| 10073 |
+
},
|
| 10074 |
+
{
|
| 10075 |
+
"epoch": 0.11469706058788243,
|
| 10076 |
+
"grad_norm": 4.358283996582031,
|
| 10077 |
+
"learning_rate": 2.359301610808917e-06,
|
| 10078 |
+
"loss": 3.696,
|
| 10079 |
+
"step": 1434
|
| 10080 |
+
},
|
| 10081 |
+
{
|
| 10082 |
+
"epoch": 0.11477704459108179,
|
| 10083 |
+
"grad_norm": 5.082062244415283,
|
| 10084 |
+
"learning_rate": 2.315168224039932e-06,
|
| 10085 |
+
"loss": 4.6404,
|
| 10086 |
+
"step": 1435
|
| 10087 |
+
},
|
| 10088 |
+
{
|
| 10089 |
+
"epoch": 0.11485702859428114,
|
| 10090 |
+
"grad_norm": 4.232576847076416,
|
| 10091 |
+
"learning_rate": 2.271446691817014e-06,
|
| 10092 |
+
"loss": 3.0782,
|
| 10093 |
+
"step": 1436
|
| 10094 |
+
},
|
| 10095 |
+
{
|
| 10096 |
+
"epoch": 0.1149370125974805,
|
| 10097 |
+
"grad_norm": 4.3032755851745605,
|
| 10098 |
+
"learning_rate": 2.228137198476976e-06,
|
| 10099 |
+
"loss": 3.2745,
|
| 10100 |
+
"step": 1437
|
| 10101 |
+
},
|
| 10102 |
+
{
|
| 10103 |
+
"epoch": 0.11501699660067986,
|
| 10104 |
+
"grad_norm": 4.733532428741455,
|
| 10105 |
+
"learning_rate": 2.1852399266194314e-06,
|
| 10106 |
+
"loss": 5.058,
|
| 10107 |
+
"step": 1438
|
| 10108 |
+
},
|
| 10109 |
+
{
|
| 10110 |
+
"epoch": 0.11509698060387923,
|
| 10111 |
+
"grad_norm": 4.020907878875732,
|
| 10112 |
+
"learning_rate": 2.1427550571060053e-06,
|
| 10113 |
+
"loss": 2.7802,
|
| 10114 |
+
"step": 1439
|
| 10115 |
+
},
|
| 10116 |
+
{
|
| 10117 |
+
"epoch": 0.11517696460707859,
|
| 10118 |
+
"grad_norm": 5.103706359863281,
|
| 10119 |
+
"learning_rate": 2.100682769059548e-06,
|
| 10120 |
+
"loss": 4.2234,
|
| 10121 |
+
"step": 1440
|
| 10122 |
+
},
|
| 10123 |
+
{
|
| 10124 |
+
"epoch": 0.11525694861027795,
|
| 10125 |
+
"grad_norm": 3.8497917652130127,
|
| 10126 |
+
"learning_rate": 2.0590232398634114e-06,
|
| 10127 |
+
"loss": 3.1467,
|
| 10128 |
+
"step": 1441
|
| 10129 |
+
},
|
| 10130 |
+
{
|
| 10131 |
+
"epoch": 0.1153369326134773,
|
| 10132 |
+
"grad_norm": 3.565351724624634,
|
| 10133 |
+
"learning_rate": 2.017776645160707e-06,
|
| 10134 |
+
"loss": 3.2625,
|
| 10135 |
+
"step": 1442
|
| 10136 |
+
},
|
| 10137 |
+
{
|
| 10138 |
+
"epoch": 0.11541691661667666,
|
| 10139 |
+
"grad_norm": 4.117542743682861,
|
| 10140 |
+
"learning_rate": 1.976943158853517e-06,
|
| 10141 |
+
"loss": 3.9284,
|
| 10142 |
+
"step": 1443
|
| 10143 |
+
},
|
| 10144 |
+
{
|
| 10145 |
+
"epoch": 0.11549690061987603,
|
| 10146 |
+
"grad_norm": 3.3290624618530273,
|
| 10147 |
+
"learning_rate": 1.9365229531022264e-06,
|
| 10148 |
+
"loss": 3.1838,
|
| 10149 |
+
"step": 1444
|
| 10150 |
+
},
|
| 10151 |
+
{
|
| 10152 |
+
"epoch": 0.11557688462307539,
|
| 10153 |
+
"grad_norm": 4.1593523025512695,
|
| 10154 |
+
"learning_rate": 1.8965161983247493e-06,
|
| 10155 |
+
"loss": 3.0462,
|
| 10156 |
+
"step": 1445
|
| 10157 |
+
},
|
| 10158 |
+
{
|
| 10159 |
+
"epoch": 0.11565686862627475,
|
| 10160 |
+
"grad_norm": 4.247846603393555,
|
| 10161 |
+
"learning_rate": 1.8569230631958256e-06,
|
| 10162 |
+
"loss": 2.9795,
|
| 10163 |
+
"step": 1446
|
| 10164 |
+
},
|
| 10165 |
+
{
|
| 10166 |
+
"epoch": 0.1157368526294741,
|
| 10167 |
+
"grad_norm": 4.3389363288879395,
|
| 10168 |
+
"learning_rate": 1.8177437146463138e-06,
|
| 10169 |
+
"loss": 4.2456,
|
| 10170 |
+
"step": 1447
|
| 10171 |
+
},
|
| 10172 |
+
{
|
| 10173 |
+
"epoch": 0.11581683663267346,
|
| 10174 |
+
"grad_norm": 4.0213446617126465,
|
| 10175 |
+
"learning_rate": 1.7789783178624897e-06,
|
| 10176 |
+
"loss": 2.616,
|
| 10177 |
+
"step": 1448
|
| 10178 |
+
},
|
| 10179 |
+
{
|
| 10180 |
+
"epoch": 0.11589682063587282,
|
| 10181 |
+
"grad_norm": 4.378050327301025,
|
| 10182 |
+
"learning_rate": 1.7406270362853249e-06,
|
| 10183 |
+
"loss": 3.7144,
|
| 10184 |
+
"step": 1449
|
| 10185 |
+
},
|
| 10186 |
+
{
|
| 10187 |
+
"epoch": 0.11597680463907219,
|
| 10188 |
+
"grad_norm": 4.620596408843994,
|
| 10189 |
+
"learning_rate": 1.7026900316098215e-06,
|
| 10190 |
+
"loss": 3.9864,
|
| 10191 |
+
"step": 1450
|
| 10192 |
+
},
|
| 10193 |
+
{
|
| 10194 |
+
"epoch": 0.11605678864227155,
|
| 10195 |
+
"grad_norm": 5.792392730712891,
|
| 10196 |
+
"learning_rate": 1.665167463784356e-06,
|
| 10197 |
+
"loss": 3.4818,
|
| 10198 |
+
"step": 1451
|
| 10199 |
+
},
|
| 10200 |
+
{
|
| 10201 |
+
"epoch": 0.11613677264547091,
|
| 10202 |
+
"grad_norm": 4.163421630859375,
|
| 10203 |
+
"learning_rate": 1.6280594910099256e-06,
|
| 10204 |
+
"loss": 3.2571,
|
| 10205 |
+
"step": 1452
|
| 10206 |
+
},
|
| 10207 |
+
{
|
| 10208 |
+
"epoch": 0.11621675664867026,
|
| 10209 |
+
"grad_norm": 4.461169719696045,
|
| 10210 |
+
"learning_rate": 1.591366269739558e-06,
|
| 10211 |
+
"loss": 3.4497,
|
| 10212 |
+
"step": 1453
|
| 10213 |
+
},
|
| 10214 |
+
{
|
| 10215 |
+
"epoch": 0.11629674065186962,
|
| 10216 |
+
"grad_norm": 3.659543037414551,
|
| 10217 |
+
"learning_rate": 1.5550879546776364e-06,
|
| 10218 |
+
"loss": 3.6999,
|
| 10219 |
+
"step": 1454
|
| 10220 |
+
},
|
| 10221 |
+
{
|
| 10222 |
+
"epoch": 0.11637672465506899,
|
| 10223 |
+
"grad_norm": 3.0203680992126465,
|
| 10224 |
+
"learning_rate": 1.5192246987791981e-06,
|
| 10225 |
+
"loss": 2.2908,
|
| 10226 |
+
"step": 1455
|
| 10227 |
+
},
|
| 10228 |
+
{
|
| 10229 |
+
"epoch": 0.11645670865826835,
|
| 10230 |
+
"grad_norm": 3.7625060081481934,
|
| 10231 |
+
"learning_rate": 1.4837766532493468e-06,
|
| 10232 |
+
"loss": 3.1457,
|
| 10233 |
+
"step": 1456
|
| 10234 |
+
},
|
| 10235 |
+
{
|
| 10236 |
+
"epoch": 0.11653669266146771,
|
| 10237 |
+
"grad_norm": 5.014787197113037,
|
| 10238 |
+
"learning_rate": 1.4487439675426094e-06,
|
| 10239 |
+
"loss": 3.9683,
|
| 10240 |
+
"step": 1457
|
| 10241 |
+
},
|
| 10242 |
+
{
|
| 10243 |
+
"epoch": 0.11661667666466706,
|
| 10244 |
+
"grad_norm": 4.519006729125977,
|
| 10245 |
+
"learning_rate": 1.414126789362269e-06,
|
| 10246 |
+
"loss": 3.9337,
|
| 10247 |
+
"step": 1458
|
| 10248 |
+
},
|
| 10249 |
+
{
|
| 10250 |
+
"epoch": 0.11669666066786642,
|
| 10251 |
+
"grad_norm": 4.185690879821777,
|
| 10252 |
+
"learning_rate": 1.3799252646597426e-06,
|
| 10253 |
+
"loss": 3.4311,
|
| 10254 |
+
"step": 1459
|
| 10255 |
+
},
|
| 10256 |
+
{
|
| 10257 |
+
"epoch": 0.11677664467106579,
|
| 10258 |
+
"grad_norm": 3.645003318786621,
|
| 10259 |
+
"learning_rate": 1.3461395376340502e-06,
|
| 10260 |
+
"loss": 3.3593,
|
| 10261 |
+
"step": 1460
|
| 10262 |
+
},
|
| 10263 |
+
{
|
| 10264 |
+
"epoch": 0.11685662867426515,
|
| 10265 |
+
"grad_norm": 3.9734771251678467,
|
| 10266 |
+
"learning_rate": 1.312769750731091e-06,
|
| 10267 |
+
"loss": 2.8187,
|
| 10268 |
+
"step": 1461
|
| 10269 |
+
},
|
| 10270 |
+
{
|
| 10271 |
+
"epoch": 0.11693661267746451,
|
| 10272 |
+
"grad_norm": 3.9598028659820557,
|
| 10273 |
+
"learning_rate": 1.2798160446431006e-06,
|
| 10274 |
+
"loss": 2.8971,
|
| 10275 |
+
"step": 1462
|
| 10276 |
+
},
|
| 10277 |
+
{
|
| 10278 |
+
"epoch": 0.11701659668066387,
|
| 10279 |
+
"grad_norm": 4.44677209854126,
|
| 10280 |
+
"learning_rate": 1.2472785583080848e-06,
|
| 10281 |
+
"loss": 3.8034,
|
| 10282 |
+
"step": 1463
|
| 10283 |
+
},
|
| 10284 |
+
{
|
| 10285 |
+
"epoch": 0.11709658068386322,
|
| 10286 |
+
"grad_norm": 3.679844617843628,
|
| 10287 |
+
"learning_rate": 1.2151574289091749e-06,
|
| 10288 |
+
"loss": 3.3024,
|
| 10289 |
+
"step": 1464
|
| 10290 |
+
},
|
| 10291 |
+
{
|
| 10292 |
+
"epoch": 0.11717656468706258,
|
| 10293 |
+
"grad_norm": 4.139998912811279,
|
| 10294 |
+
"learning_rate": 1.1834527918740623e-06,
|
| 10295 |
+
"loss": 3.3435,
|
| 10296 |
+
"step": 1465
|
| 10297 |
+
},
|
| 10298 |
+
{
|
| 10299 |
+
"epoch": 0.11725654869026195,
|
| 10300 |
+
"grad_norm": 3.545083522796631,
|
| 10301 |
+
"learning_rate": 1.1521647808744873e-06,
|
| 10302 |
+
"loss": 2.7517,
|
| 10303 |
+
"step": 1466
|
| 10304 |
+
},
|
| 10305 |
+
{
|
| 10306 |
+
"epoch": 0.11733653269346131,
|
| 10307 |
+
"grad_norm": 5.666535377502441,
|
| 10308 |
+
"learning_rate": 1.1212935278255955e-06,
|
| 10309 |
+
"loss": 3.9504,
|
| 10310 |
+
"step": 1467
|
| 10311 |
+
},
|
| 10312 |
+
{
|
| 10313 |
+
"epoch": 0.11741651669666067,
|
| 10314 |
+
"grad_norm": 5.340794563293457,
|
| 10315 |
+
"learning_rate": 1.0908391628854041e-06,
|
| 10316 |
+
"loss": 4.1832,
|
| 10317 |
+
"step": 1468
|
| 10318 |
+
},
|
| 10319 |
+
{
|
| 10320 |
+
"epoch": 0.11749650069986003,
|
| 10321 |
+
"grad_norm": 3.769444227218628,
|
| 10322 |
+
"learning_rate": 1.060801814454304e-06,
|
| 10323 |
+
"loss": 3.2104,
|
| 10324 |
+
"step": 1469
|
| 10325 |
+
},
|
| 10326 |
+
{
|
| 10327 |
+
"epoch": 0.11757648470305938,
|
| 10328 |
+
"grad_norm": 3.579387903213501,
|
| 10329 |
+
"learning_rate": 1.0311816091744698e-06,
|
| 10330 |
+
"loss": 2.7134,
|
| 10331 |
+
"step": 1470
|
| 10332 |
+
},
|
| 10333 |
+
{
|
| 10334 |
+
"epoch": 0.11765646870625875,
|
| 10335 |
+
"grad_norm": 5.004464149475098,
|
| 10336 |
+
"learning_rate": 1.0019786719293045e-06,
|
| 10337 |
+
"loss": 3.5459,
|
| 10338 |
+
"step": 1471
|
| 10339 |
+
},
|
| 10340 |
+
{
|
| 10341 |
+
"epoch": 0.11773645270945811,
|
| 10342 |
+
"grad_norm": 6.693496227264404,
|
| 10343 |
+
"learning_rate": 9.731931258429638e-07,
|
| 10344 |
+
"loss": 4.5585,
|
| 10345 |
+
"step": 1472
|
| 10346 |
+
},
|
| 10347 |
+
{
|
| 10348 |
+
"epoch": 0.11781643671265747,
|
| 10349 |
+
"grad_norm": 4.905187606811523,
|
| 10350 |
+
"learning_rate": 9.448250922798329e-07,
|
| 10351 |
+
"loss": 3.5167,
|
| 10352 |
+
"step": 1473
|
| 10353 |
+
},
|
| 10354 |
+
{
|
| 10355 |
+
"epoch": 0.11789642071585683,
|
| 10356 |
+
"grad_norm": 3.418574333190918,
|
| 10357 |
+
"learning_rate": 9.168746908439718e-07,
|
| 10358 |
+
"loss": 2.669,
|
| 10359 |
+
"step": 1474
|
| 10360 |
+
},
|
| 10361 |
+
{
|
| 10362 |
+
"epoch": 0.11797640471905618,
|
| 10363 |
+
"grad_norm": 3.538538694381714,
|
| 10364 |
+
"learning_rate": 8.893420393786489e-07,
|
| 10365 |
+
"loss": 3.4193,
|
| 10366 |
+
"step": 1475
|
| 10367 |
+
},
|
| 10368 |
+
{
|
| 10369 |
+
"epoch": 0.11805638872225555,
|
| 10370 |
+
"grad_norm": 3.831590175628662,
|
| 10371 |
+
"learning_rate": 8.622272539658415e-07,
|
| 10372 |
+
"loss": 2.9996,
|
| 10373 |
+
"step": 1476
|
| 10374 |
+
},
|
| 10375 |
+
{
|
| 10376 |
+
"epoch": 0.11813637272545491,
|
| 10377 |
+
"grad_norm": 3.4640448093414307,
|
| 10378 |
+
"learning_rate": 8.355304489257254e-07,
|
| 10379 |
+
"loss": 2.9861,
|
| 10380 |
+
"step": 1477
|
| 10381 |
+
},
|
| 10382 |
+
{
|
| 10383 |
+
"epoch": 0.11821635672865427,
|
| 10384 |
+
"grad_norm": 4.218769073486328,
|
| 10385 |
+
"learning_rate": 8.092517368162078e-07,
|
| 10386 |
+
"loss": 3.3627,
|
| 10387 |
+
"step": 1478
|
| 10388 |
+
},
|
| 10389 |
+
{
|
| 10390 |
+
"epoch": 0.11829634073185363,
|
| 10391 |
+
"grad_norm": 3.9875335693359375,
|
| 10392 |
+
"learning_rate": 7.833912284324618e-07,
|
| 10393 |
+
"loss": 3.6138,
|
| 10394 |
+
"step": 1479
|
| 10395 |
+
},
|
| 10396 |
+
{
|
| 10397 |
+
"epoch": 0.118376324735053,
|
| 10398 |
+
"grad_norm": 4.092508316040039,
|
| 10399 |
+
"learning_rate": 7.579490328064265e-07,
|
| 10400 |
+
"loss": 3.8903,
|
| 10401 |
+
"step": 1480
|
| 10402 |
+
},
|
| 10403 |
+
{
|
| 10404 |
+
"epoch": 0.11845630873825234,
|
| 10405 |
+
"grad_norm": 4.583672046661377,
|
| 10406 |
+
"learning_rate": 7.32925257206396e-07,
|
| 10407 |
+
"loss": 3.7078,
|
| 10408 |
+
"step": 1481
|
| 10409 |
+
},
|
| 10410 |
+
{
|
| 10411 |
+
"epoch": 0.11853629274145171,
|
| 10412 |
+
"grad_norm": 5.149039268493652,
|
| 10413 |
+
"learning_rate": 7.083200071365203e-07,
|
| 10414 |
+
"loss": 3.7716,
|
| 10415 |
+
"step": 1482
|
| 10416 |
+
},
|
| 10417 |
+
{
|
| 10418 |
+
"epoch": 0.11861627674465107,
|
| 10419 |
+
"grad_norm": 4.750640392303467,
|
| 10420 |
+
"learning_rate": 6.841333863363831e-07,
|
| 10421 |
+
"loss": 2.8942,
|
| 10422 |
+
"step": 1483
|
| 10423 |
+
},
|
| 10424 |
+
{
|
| 10425 |
+
"epoch": 0.11869626074785043,
|
| 10426 |
+
"grad_norm": 4.358502388000488,
|
| 10427 |
+
"learning_rate": 6.603654967805683e-07,
|
| 10428 |
+
"loss": 3.5916,
|
| 10429 |
+
"step": 1484
|
| 10430 |
+
},
|
| 10431 |
+
{
|
| 10432 |
+
"epoch": 0.1187762447510498,
|
| 10433 |
+
"grad_norm": 3.6885571479797363,
|
| 10434 |
+
"learning_rate": 6.370164386782285e-07,
|
| 10435 |
+
"loss": 3.0587,
|
| 10436 |
+
"step": 1485
|
| 10437 |
+
},
|
| 10438 |
+
{
|
| 10439 |
+
"epoch": 0.11885622875424914,
|
| 10440 |
+
"grad_norm": 4.57819128036499,
|
| 10441 |
+
"learning_rate": 6.140863104726391e-07,
|
| 10442 |
+
"loss": 4.0528,
|
| 10443 |
+
"step": 1486
|
| 10444 |
+
},
|
| 10445 |
+
{
|
| 10446 |
+
"epoch": 0.1189362127574485,
|
| 10447 |
+
"grad_norm": 4.245659828186035,
|
| 10448 |
+
"learning_rate": 5.915752088408e-07,
|
| 10449 |
+
"loss": 3.3928,
|
| 10450 |
+
"step": 1487
|
| 10451 |
+
},
|
| 10452 |
+
{
|
| 10453 |
+
"epoch": 0.11901619676064787,
|
| 10454 |
+
"grad_norm": 4.543621063232422,
|
| 10455 |
+
"learning_rate": 5.694832286930685e-07,
|
| 10456 |
+
"loss": 3.1624,
|
| 10457 |
+
"step": 1488
|
| 10458 |
+
},
|
| 10459 |
+
{
|
| 10460 |
+
"epoch": 0.11909618076384723,
|
| 10461 |
+
"grad_norm": 4.2259039878845215,
|
| 10462 |
+
"learning_rate": 5.478104631726711e-07,
|
| 10463 |
+
"loss": 3.4839,
|
| 10464 |
+
"step": 1489
|
| 10465 |
+
},
|
| 10466 |
+
{
|
| 10467 |
+
"epoch": 0.1191761647670466,
|
| 10468 |
+
"grad_norm": 4.5847039222717285,
|
| 10469 |
+
"learning_rate": 5.265570036553813e-07,
|
| 10470 |
+
"loss": 3.5966,
|
| 10471 |
+
"step": 1490
|
| 10472 |
+
},
|
| 10473 |
+
{
|
| 10474 |
+
"epoch": 0.11925614877024596,
|
| 10475 |
+
"grad_norm": 3.7808141708374023,
|
| 10476 |
+
"learning_rate": 5.057229397491204e-07,
|
| 10477 |
+
"loss": 3.3972,
|
| 10478 |
+
"step": 1491
|
| 10479 |
+
},
|
| 10480 |
+
{
|
| 10481 |
+
"epoch": 0.1193361327734453,
|
| 10482 |
+
"grad_norm": 4.415256977081299,
|
| 10483 |
+
"learning_rate": 4.85308359293557e-07,
|
| 10484 |
+
"loss": 4.3388,
|
| 10485 |
+
"step": 1492
|
| 10486 |
+
},
|
| 10487 |
+
{
|
| 10488 |
+
"epoch": 0.11941611677664467,
|
| 10489 |
+
"grad_norm": 4.698878288269043,
|
| 10490 |
+
"learning_rate": 4.653133483597638e-07,
|
| 10491 |
+
"loss": 2.6318,
|
| 10492 |
+
"step": 1493
|
| 10493 |
+
},
|
| 10494 |
+
{
|
| 10495 |
+
"epoch": 0.11949610077984403,
|
| 10496 |
+
"grad_norm": 4.164829254150391,
|
| 10497 |
+
"learning_rate": 4.457379912498394e-07,
|
| 10498 |
+
"loss": 4.389,
|
| 10499 |
+
"step": 1494
|
| 10500 |
+
},
|
| 10501 |
+
{
|
| 10502 |
+
"epoch": 0.1195760847830434,
|
| 10503 |
+
"grad_norm": 4.140608310699463,
|
| 10504 |
+
"learning_rate": 4.2658237049655323e-07,
|
| 10505 |
+
"loss": 4.1022,
|
| 10506 |
+
"step": 1495
|
| 10507 |
+
},
|
| 10508 |
+
{
|
| 10509 |
+
"epoch": 0.11965606878624276,
|
| 10510 |
+
"grad_norm": 4.643738746643066,
|
| 10511 |
+
"learning_rate": 4.078465668629905e-07,
|
| 10512 |
+
"loss": 4.2987,
|
| 10513 |
+
"step": 1496
|
| 10514 |
+
},
|
| 10515 |
+
{
|
| 10516 |
+
"epoch": 0.11973605278944212,
|
| 10517 |
+
"grad_norm": 3.7802324295043945,
|
| 10518 |
+
"learning_rate": 3.8953065934225206e-07,
|
| 10519 |
+
"loss": 3.1774,
|
| 10520 |
+
"step": 1497
|
| 10521 |
+
},
|
| 10522 |
+
{
|
| 10523 |
+
"epoch": 0.11981603679264147,
|
| 10524 |
+
"grad_norm": 4.139101982116699,
|
| 10525 |
+
"learning_rate": 3.716347251570551e-07,
|
| 10526 |
+
"loss": 3.3893,
|
| 10527 |
+
"step": 1498
|
| 10528 |
+
},
|
| 10529 |
+
{
|
| 10530 |
+
"epoch": 0.11989602079584083,
|
| 10531 |
+
"grad_norm": 3.5190374851226807,
|
| 10532 |
+
"learning_rate": 3.541588397594664e-07,
|
| 10533 |
+
"loss": 3.531,
|
| 10534 |
+
"step": 1499
|
| 10535 |
+
},
|
| 10536 |
+
{
|
| 10537 |
+
"epoch": 0.11997600479904019,
|
| 10538 |
+
"grad_norm": 3.7342114448547363,
|
| 10539 |
+
"learning_rate": 3.371030768305583e-07,
|
| 10540 |
+
"loss": 3.1928,
|
| 10541 |
+
"step": 1500
|
| 10542 |
+
},
|
| 10543 |
+
{
|
| 10544 |
+
"epoch": 0.12005598880223955,
|
| 10545 |
+
"grad_norm": 3.8492627143859863,
|
| 10546 |
+
"learning_rate": 3.204675082800979e-07,
|
| 10547 |
+
"loss": 3.382,
|
| 10548 |
+
"step": 1501
|
| 10549 |
+
},
|
| 10550 |
+
{
|
| 10551 |
+
"epoch": 0.12013597280543892,
|
| 10552 |
+
"grad_norm": 4.815494537353516,
|
| 10553 |
+
"learning_rate": 3.042522042462359e-07,
|
| 10554 |
+
"loss": 4.0397,
|
| 10555 |
+
"step": 1502
|
| 10556 |
+
},
|
| 10557 |
+
{
|
| 10558 |
+
"epoch": 0.12021595680863827,
|
| 10559 |
+
"grad_norm": 3.6874735355377197,
|
| 10560 |
+
"learning_rate": 2.8845723309525176e-07,
|
| 10561 |
+
"loss": 2.5105,
|
| 10562 |
+
"step": 1503
|
| 10563 |
+
},
|
| 10564 |
+
{
|
| 10565 |
+
"epoch": 0.12029594081183763,
|
| 10566 |
+
"grad_norm": 4.021266460418701,
|
| 10567 |
+
"learning_rate": 2.7308266142119785e-07,
|
| 10568 |
+
"loss": 3.2706,
|
| 10569 |
+
"step": 1504
|
| 10570 |
+
},
|
| 10571 |
+
{
|
| 10572 |
+
"epoch": 0.12037592481503699,
|
| 10573 |
+
"grad_norm": 4.5269575119018555,
|
| 10574 |
+
"learning_rate": 2.5812855404568905e-07,
|
| 10575 |
+
"loss": 4.5545,
|
| 10576 |
+
"step": 1505
|
| 10577 |
+
},
|
| 10578 |
+
{
|
| 10579 |
+
"epoch": 0.12045590881823635,
|
| 10580 |
+
"grad_norm": 4.050426959991455,
|
| 10581 |
+
"learning_rate": 2.4359497401758024e-07,
|
| 10582 |
+
"loss": 3.9122,
|
| 10583 |
+
"step": 1506
|
| 10584 |
+
},
|
| 10585 |
+
{
|
| 10586 |
+
"epoch": 0.12053589282143572,
|
| 10587 |
+
"grad_norm": 3.5586445331573486,
|
| 10588 |
+
"learning_rate": 2.2948198261271148e-07,
|
| 10589 |
+
"loss": 2.8053,
|
| 10590 |
+
"step": 1507
|
| 10591 |
+
},
|
| 10592 |
+
{
|
| 10593 |
+
"epoch": 0.12061587682463508,
|
| 10594 |
+
"grad_norm": 3.978595018386841,
|
| 10595 |
+
"learning_rate": 2.1578963933367446e-07,
|
| 10596 |
+
"loss": 2.9083,
|
| 10597 |
+
"step": 1508
|
| 10598 |
+
},
|
| 10599 |
+
{
|
| 10600 |
+
"epoch": 0.12069586082783443,
|
| 10601 |
+
"grad_norm": 4.400628566741943,
|
| 10602 |
+
"learning_rate": 2.0251800190953518e-07,
|
| 10603 |
+
"loss": 3.4607,
|
| 10604 |
+
"step": 1509
|
| 10605 |
+
},
|
| 10606 |
+
{
|
| 10607 |
+
"epoch": 0.12077584483103379,
|
| 10608 |
+
"grad_norm": 4.357829570770264,
|
| 10609 |
+
"learning_rate": 1.8966712629558957e-07,
|
| 10610 |
+
"loss": 3.3244,
|
| 10611 |
+
"step": 1510
|
| 10612 |
+
},
|
| 10613 |
+
{
|
| 10614 |
+
"epoch": 0.12085582883423315,
|
| 10615 |
+
"grad_norm": 4.234547138214111,
|
| 10616 |
+
"learning_rate": 1.7723706667314155e-07,
|
| 10617 |
+
"loss": 3.36,
|
| 10618 |
+
"step": 1511
|
| 10619 |
+
},
|
| 10620 |
+
{
|
| 10621 |
+
"epoch": 0.12093581283743252,
|
| 10622 |
+
"grad_norm": 3.8055272102355957,
|
| 10623 |
+
"learning_rate": 1.6522787544926977e-07,
|
| 10624 |
+
"loss": 3.672,
|
| 10625 |
+
"step": 1512
|
| 10626 |
+
},
|
| 10627 |
+
{
|
| 10628 |
+
"epoch": 0.12101579684063188,
|
| 10629 |
+
"grad_norm": 3.644456386566162,
|
| 10630 |
+
"learning_rate": 1.5363960325660565e-07,
|
| 10631 |
+
"loss": 2.8197,
|
| 10632 |
+
"step": 1513
|
| 10633 |
+
},
|
| 10634 |
+
{
|
| 10635 |
+
"epoch": 0.12109578084383123,
|
| 10636 |
+
"grad_norm": 4.6977338790893555,
|
| 10637 |
+
"learning_rate": 1.424722989531113e-07,
|
| 10638 |
+
"loss": 3.9904,
|
| 10639 |
+
"step": 1514
|
| 10640 |
+
},
|
| 10641 |
+
{
|
| 10642 |
+
"epoch": 0.12117576484703059,
|
| 10643 |
+
"grad_norm": 3.618805170059204,
|
| 10644 |
+
"learning_rate": 1.3172600962190197e-07,
|
| 10645 |
+
"loss": 2.9669,
|
| 10646 |
+
"step": 1515
|
| 10647 |
+
},
|
| 10648 |
+
{
|
| 10649 |
+
"epoch": 0.12125574885022995,
|
| 10650 |
+
"grad_norm": 3.858250141143799,
|
| 10651 |
+
"learning_rate": 1.2140078057101266e-07,
|
| 10652 |
+
"loss": 2.7861,
|
| 10653 |
+
"step": 1516
|
| 10654 |
+
},
|
| 10655 |
+
{
|
| 10656 |
+
"epoch": 0.12133573285342932,
|
| 10657 |
+
"grad_norm": 3.7499475479125977,
|
| 10658 |
+
"learning_rate": 1.1149665533320974e-07,
|
| 10659 |
+
"loss": 3.2523,
|
| 10660 |
+
"step": 1517
|
| 10661 |
+
},
|
| 10662 |
+
{
|
| 10663 |
+
"epoch": 0.12141571685662868,
|
| 10664 |
+
"grad_norm": 4.985630512237549,
|
| 10665 |
+
"learning_rate": 1.020136756658574e-07,
|
| 10666 |
+
"loss": 3.4223,
|
| 10667 |
+
"step": 1518
|
| 10668 |
+
},
|
| 10669 |
+
{
|
| 10670 |
+
"epoch": 0.12149570085982804,
|
| 10671 |
+
"grad_norm": 3.8844850063323975,
|
| 10672 |
+
"learning_rate": 9.295188155068468e-08,
|
| 10673 |
+
"loss": 3.3358,
|
| 10674 |
+
"step": 1519
|
| 10675 |
+
},
|
| 10676 |
+
{
|
| 10677 |
+
"epoch": 0.12157568486302739,
|
| 10678 |
+
"grad_norm": 4.788154602050781,
|
| 10679 |
+
"learning_rate": 8.43113111936189e-08,
|
| 10680 |
+
"loss": 4.0249,
|
| 10681 |
+
"step": 1520
|
| 10682 |
+
},
|
| 10683 |
+
{
|
| 10684 |
+
"epoch": 0.12165566886622675,
|
| 10685 |
+
"grad_norm": 3.8909480571746826,
|
| 10686 |
+
"learning_rate": 7.609200102465241e-08,
|
| 10687 |
+
"loss": 3.1497,
|
| 10688 |
+
"step": 1521
|
| 10689 |
+
},
|
| 10690 |
+
{
|
| 10691 |
+
"epoch": 0.12173565286942611,
|
| 10692 |
+
"grad_norm": 4.890474796295166,
|
| 10693 |
+
"learning_rate": 6.829398569770939e-08,
|
| 10694 |
+
"loss": 3.4871,
|
| 10695 |
+
"step": 1522
|
| 10696 |
+
},
|
| 10697 |
+
{
|
| 10698 |
+
"epoch": 0.12181563687262548,
|
| 10699 |
+
"grad_norm": 3.6632330417633057,
|
| 10700 |
+
"learning_rate": 6.09172980904238e-08,
|
| 10701 |
+
"loss": 3.5284,
|
| 10702 |
+
"step": 1523
|
| 10703 |
+
},
|
| 10704 |
+
{
|
| 10705 |
+
"epoch": 0.12189562087582484,
|
| 10706 |
+
"grad_norm": 3.084301710128784,
|
| 10707 |
+
"learning_rate": 5.3961969304072715e-08,
|
| 10708 |
+
"loss": 2.9465,
|
| 10709 |
+
"step": 1524
|
| 10710 |
+
},
|
| 10711 |
+
{
|
| 10712 |
+
"epoch": 0.1219756048790242,
|
| 10713 |
+
"grad_norm": 4.575099468231201,
|
| 10714 |
+
"learning_rate": 4.74280286634099e-08,
|
| 10715 |
+
"loss": 4.1169,
|
| 10716 |
+
"step": 1525
|
| 10717 |
+
},
|
| 10718 |
+
{
|
| 10719 |
+
"epoch": 0.12205558888222355,
|
| 10720 |
+
"grad_norm": 4.711434841156006,
|
| 10721 |
+
"learning_rate": 4.131550371655468e-08,
|
| 10722 |
+
"loss": 4.1647,
|
| 10723 |
+
"step": 1526
|
| 10724 |
+
},
|
| 10725 |
+
{
|
| 10726 |
+
"epoch": 0.12213557288542291,
|
| 10727 |
+
"grad_norm": 3.6860179901123047,
|
| 10728 |
+
"learning_rate": 3.562442023485879e-08,
|
| 10729 |
+
"loss": 2.6378,
|
| 10730 |
+
"step": 1527
|
| 10731 |
+
},
|
| 10732 |
+
{
|
| 10733 |
+
"epoch": 0.12221555688862228,
|
| 10734 |
+
"grad_norm": 4.5305070877075195,
|
| 10735 |
+
"learning_rate": 3.0354802212839705e-08,
|
| 10736 |
+
"loss": 3.7609,
|
| 10737 |
+
"step": 1528
|
| 10738 |
+
},
|
| 10739 |
+
{
|
| 10740 |
+
"epoch": 0.12229554089182164,
|
| 10741 |
+
"grad_norm": 4.689446926116943,
|
| 10742 |
+
"learning_rate": 2.550667186801414e-08,
|
| 10743 |
+
"loss": 3.5085,
|
| 10744 |
+
"step": 1529
|
| 10745 |
+
},
|
| 10746 |
+
{
|
| 10747 |
+
"epoch": 0.122375524895021,
|
| 10748 |
+
"grad_norm": 4.543569564819336,
|
| 10749 |
+
"learning_rate": 2.108004964086474e-08,
|
| 10750 |
+
"loss": 3.4681,
|
| 10751 |
+
"step": 1530
|
| 10752 |
+
},
|
| 10753 |
+
{
|
| 10754 |
+
"epoch": 0.12245550889822035,
|
| 10755 |
+
"grad_norm": 3.1736643314361572,
|
| 10756 |
+
"learning_rate": 1.7074954194729044e-08,
|
| 10757 |
+
"loss": 3.0281,
|
| 10758 |
+
"step": 1531
|
| 10759 |
+
},
|
| 10760 |
+
{
|
| 10761 |
+
"epoch": 0.12253549290141971,
|
| 10762 |
+
"grad_norm": 26.721364974975586,
|
| 10763 |
+
"learning_rate": 1.3491402415710675e-08,
|
| 10764 |
+
"loss": 4.7551,
|
| 10765 |
+
"step": 1532
|
| 10766 |
+
},
|
| 10767 |
+
{
|
| 10768 |
+
"epoch": 0.12261547690461908,
|
| 10769 |
+
"grad_norm": 4.449186325073242,
|
| 10770 |
+
"learning_rate": 1.0329409412634938e-08,
|
| 10771 |
+
"loss": 3.7758,
|
| 10772 |
+
"step": 1533
|
| 10773 |
+
},
|
| 10774 |
+
{
|
| 10775 |
+
"epoch": 0.12269546090781844,
|
| 10776 |
+
"grad_norm": 3.931652307510376,
|
| 10777 |
+
"learning_rate": 7.58898851693779e-09,
|
| 10778 |
+
"loss": 2.9284,
|
| 10779 |
+
"step": 1534
|
| 10780 |
+
},
|
| 10781 |
+
{
|
| 10782 |
+
"epoch": 0.1227754449110178,
|
| 10783 |
+
"grad_norm": 4.318076133728027,
|
| 10784 |
+
"learning_rate": 5.270151282688041e-09,
|
| 10785 |
+
"loss": 4.0952,
|
| 10786 |
+
"step": 1535
|
| 10787 |
+
},
|
| 10788 |
+
{
|
| 10789 |
+
"epoch": 0.12285542891421716,
|
| 10790 |
+
"grad_norm": 3.975515127182007,
|
| 10791 |
+
"learning_rate": 3.3729074864541355e-09,
|
| 10792 |
+
"loss": 4.1076,
|
| 10793 |
+
"step": 1536
|
| 10794 |
+
},
|
| 10795 |
+
{
|
| 10796 |
+
"epoch": 0.12293541291741651,
|
| 10797 |
+
"grad_norm": 3.743957281112671,
|
| 10798 |
+
"learning_rate": 1.897265127315251e-09,
|
| 10799 |
+
"loss": 3.149,
|
| 10800 |
+
"step": 1537
|
| 10801 |
+
},
|
| 10802 |
+
{
|
| 10803 |
+
"epoch": 0.12301539692061587,
|
| 10804 |
+
"grad_norm": 3.694349765777588,
|
| 10805 |
+
"learning_rate": 8.432304268057856e-10,
|
| 10806 |
+
"loss": 3.6728,
|
| 10807 |
+
"step": 1538
|
| 10808 |
+
},
|
| 10809 |
+
{
|
| 10810 |
+
"epoch": 0.12309538092381524,
|
| 10811 |
+
"grad_norm": 4.057514190673828,
|
| 10812 |
+
"learning_rate": 2.108078289042581e-10,
|
| 10813 |
+
"loss": 3.2975,
|
| 10814 |
+
"step": 1539
|
| 10815 |
+
},
|
| 10816 |
+
{
|
| 10817 |
+
"epoch": 0.1231753649270146,
|
| 10818 |
+
"grad_norm": 4.6135783195495605,
|
| 10819 |
+
"learning_rate": 0.0,
|
| 10820 |
+
"loss": 3.6262,
|
| 10821 |
+
"step": 1540
|
| 10822 |
+
},
|
| 10823 |
+
{
|
| 10824 |
+
"epoch": 0.1231753649270146,
|
| 10825 |
+
"eval_loss": 0.8568549752235413,
|
| 10826 |
+
"eval_runtime": 288.5378,
|
| 10827 |
+
"eval_samples_per_second": 18.247,
|
| 10828 |
+
"eval_steps_per_second": 9.125,
|
| 10829 |
+
"step": 1540
|
| 10830 |
}
|
| 10831 |
],
|
| 10832 |
"logging_steps": 1,
|
|
|
|
| 10841 |
"should_evaluate": false,
|
| 10842 |
"should_log": false,
|
| 10843 |
"should_save": true,
|
| 10844 |
+
"should_training_stop": true
|
| 10845 |
},
|
| 10846 |
"attributes": {}
|
| 10847 |
}
|
| 10848 |
},
|
| 10849 |
+
"total_flos": 7.012705999886746e+17,
|
| 10850 |
"train_batch_size": 2,
|
| 10851 |
"trial_name": null,
|
| 10852 |
"trial_params": null
|