Training in progress, step 786, 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 50365768
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:918a19dd5be289b1d5252ee6055bfa93acf056d7fe55513047fc352c573af832
|
| 3 |
size 50365768
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 25859412
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b2703b9b4c18158b143d2eb763d8d45a59efddef2997f61805afe60f82977066
|
| 3 |
size 25859412
|
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:da8aafa153797c09b7590357c4762191d38eb5a4cabcce380889e0c381a3b264
|
| 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:d21999b2c3064c25cbac458cfab467b7d49efded67b922250d5be6cf196ac859
|
| 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": 393,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -2774,6 +2774,2765 @@
|
|
| 2774 |
"eval_samples_per_second": 18.04,
|
| 2775 |
"eval_steps_per_second": 9.03,
|
| 2776 |
"step": 393
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 2777 |
}
|
| 2778 |
],
|
| 2779 |
"logging_steps": 1,
|
|
@@ -2793,7 +5552,7 @@
|
|
| 2793 |
"attributes": {}
|
| 2794 |
}
|
| 2795 |
},
|
| 2796 |
-
"total_flos":
|
| 2797 |
"train_batch_size": 2,
|
| 2798 |
"trial_name": null,
|
| 2799 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.3557365919891378,
|
| 5 |
"eval_steps": 393,
|
| 6 |
+
"global_step": 786,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 2774 |
"eval_samples_per_second": 18.04,
|
| 2775 |
"eval_steps_per_second": 9.03,
|
| 2776 |
"step": 393
|
| 2777 |
+
},
|
| 2778 |
+
{
|
| 2779 |
+
"epoch": 0.17832088707852456,
|
| 2780 |
+
"grad_norm": 0.4007515609264374,
|
| 2781 |
+
"learning_rate": 0.00017155989607441213,
|
| 2782 |
+
"loss": 0.6961,
|
| 2783 |
+
"step": 394
|
| 2784 |
+
},
|
| 2785 |
+
{
|
| 2786 |
+
"epoch": 0.1787734781624802,
|
| 2787 |
+
"grad_norm": 0.4367770850658417,
|
| 2788 |
+
"learning_rate": 0.00017141908194822446,
|
| 2789 |
+
"loss": 0.7962,
|
| 2790 |
+
"step": 395
|
| 2791 |
+
},
|
| 2792 |
+
{
|
| 2793 |
+
"epoch": 0.17922606924643583,
|
| 2794 |
+
"grad_norm": 0.5173846483230591,
|
| 2795 |
+
"learning_rate": 0.000171277978177951,
|
| 2796 |
+
"loss": 0.837,
|
| 2797 |
+
"step": 396
|
| 2798 |
+
},
|
| 2799 |
+
{
|
| 2800 |
+
"epoch": 0.17967866033039148,
|
| 2801 |
+
"grad_norm": 0.5975651741027832,
|
| 2802 |
+
"learning_rate": 0.00017113658533584594,
|
| 2803 |
+
"loss": 0.9096,
|
| 2804 |
+
"step": 397
|
| 2805 |
+
},
|
| 2806 |
+
{
|
| 2807 |
+
"epoch": 0.18013125141434713,
|
| 2808 |
+
"grad_norm": 0.512630045413971,
|
| 2809 |
+
"learning_rate": 0.00017099490399533583,
|
| 2810 |
+
"loss": 0.8301,
|
| 2811 |
+
"step": 398
|
| 2812 |
+
},
|
| 2813 |
+
{
|
| 2814 |
+
"epoch": 0.18058384249830278,
|
| 2815 |
+
"grad_norm": 0.5926880836486816,
|
| 2816 |
+
"learning_rate": 0.00017085293473101734,
|
| 2817 |
+
"loss": 0.8294,
|
| 2818 |
+
"step": 399
|
| 2819 |
+
},
|
| 2820 |
+
{
|
| 2821 |
+
"epoch": 0.18103643358225843,
|
| 2822 |
+
"grad_norm": 0.4605947732925415,
|
| 2823 |
+
"learning_rate": 0.00017071067811865476,
|
| 2824 |
+
"loss": 0.819,
|
| 2825 |
+
"step": 400
|
| 2826 |
+
},
|
| 2827 |
+
{
|
| 2828 |
+
"epoch": 0.18148902466621408,
|
| 2829 |
+
"grad_norm": 0.5397257208824158,
|
| 2830 |
+
"learning_rate": 0.00017056813473517775,
|
| 2831 |
+
"loss": 0.958,
|
| 2832 |
+
"step": 401
|
| 2833 |
+
},
|
| 2834 |
+
{
|
| 2835 |
+
"epoch": 0.18194161575016973,
|
| 2836 |
+
"grad_norm": 0.5134234428405762,
|
| 2837 |
+
"learning_rate": 0.00017042530515867896,
|
| 2838 |
+
"loss": 0.8052,
|
| 2839 |
+
"step": 402
|
| 2840 |
+
},
|
| 2841 |
+
{
|
| 2842 |
+
"epoch": 0.18239420683412536,
|
| 2843 |
+
"grad_norm": 0.48100152611732483,
|
| 2844 |
+
"learning_rate": 0.00017028218996841172,
|
| 2845 |
+
"loss": 0.8442,
|
| 2846 |
+
"step": 403
|
| 2847 |
+
},
|
| 2848 |
+
{
|
| 2849 |
+
"epoch": 0.182846797918081,
|
| 2850 |
+
"grad_norm": 0.48078837990760803,
|
| 2851 |
+
"learning_rate": 0.00017013878974478776,
|
| 2852 |
+
"loss": 0.715,
|
| 2853 |
+
"step": 404
|
| 2854 |
+
},
|
| 2855 |
+
{
|
| 2856 |
+
"epoch": 0.18329938900203666,
|
| 2857 |
+
"grad_norm": 0.4258125424385071,
|
| 2858 |
+
"learning_rate": 0.00016999510506937466,
|
| 2859 |
+
"loss": 0.7623,
|
| 2860 |
+
"step": 405
|
| 2861 |
+
},
|
| 2862 |
+
{
|
| 2863 |
+
"epoch": 0.1837519800859923,
|
| 2864 |
+
"grad_norm": 0.5635977983474731,
|
| 2865 |
+
"learning_rate": 0.00016985113652489374,
|
| 2866 |
+
"loss": 0.8729,
|
| 2867 |
+
"step": 406
|
| 2868 |
+
},
|
| 2869 |
+
{
|
| 2870 |
+
"epoch": 0.18420457116994796,
|
| 2871 |
+
"grad_norm": 0.4123055338859558,
|
| 2872 |
+
"learning_rate": 0.00016970688469521736,
|
| 2873 |
+
"loss": 0.6492,
|
| 2874 |
+
"step": 407
|
| 2875 |
+
},
|
| 2876 |
+
{
|
| 2877 |
+
"epoch": 0.1846571622539036,
|
| 2878 |
+
"grad_norm": 0.5301768183708191,
|
| 2879 |
+
"learning_rate": 0.000169562350165367,
|
| 2880 |
+
"loss": 0.974,
|
| 2881 |
+
"step": 408
|
| 2882 |
+
},
|
| 2883 |
+
{
|
| 2884 |
+
"epoch": 0.18510975333785926,
|
| 2885 |
+
"grad_norm": 0.5563939213752747,
|
| 2886 |
+
"learning_rate": 0.00016941753352151055,
|
| 2887 |
+
"loss": 1.0331,
|
| 2888 |
+
"step": 409
|
| 2889 |
+
},
|
| 2890 |
+
{
|
| 2891 |
+
"epoch": 0.18556234442181488,
|
| 2892 |
+
"grad_norm": 0.4835430681705475,
|
| 2893 |
+
"learning_rate": 0.00016927243535095997,
|
| 2894 |
+
"loss": 0.5977,
|
| 2895 |
+
"step": 410
|
| 2896 |
+
},
|
| 2897 |
+
{
|
| 2898 |
+
"epoch": 0.18601493550577053,
|
| 2899 |
+
"grad_norm": 0.4503084719181061,
|
| 2900 |
+
"learning_rate": 0.000169127056242169,
|
| 2901 |
+
"loss": 0.7688,
|
| 2902 |
+
"step": 411
|
| 2903 |
+
},
|
| 2904 |
+
{
|
| 2905 |
+
"epoch": 0.18646752658972618,
|
| 2906 |
+
"grad_norm": 0.461060106754303,
|
| 2907 |
+
"learning_rate": 0.00016898139678473076,
|
| 2908 |
+
"loss": 0.7793,
|
| 2909 |
+
"step": 412
|
| 2910 |
+
},
|
| 2911 |
+
{
|
| 2912 |
+
"epoch": 0.18692011767368183,
|
| 2913 |
+
"grad_norm": 0.49685123562812805,
|
| 2914 |
+
"learning_rate": 0.0001688354575693754,
|
| 2915 |
+
"loss": 0.9484,
|
| 2916 |
+
"step": 413
|
| 2917 |
+
},
|
| 2918 |
+
{
|
| 2919 |
+
"epoch": 0.18737270875763748,
|
| 2920 |
+
"grad_norm": 0.48521700501441956,
|
| 2921 |
+
"learning_rate": 0.00016868923918796753,
|
| 2922 |
+
"loss": 0.7861,
|
| 2923 |
+
"step": 414
|
| 2924 |
+
},
|
| 2925 |
+
{
|
| 2926 |
+
"epoch": 0.18782529984159313,
|
| 2927 |
+
"grad_norm": 0.4642525017261505,
|
| 2928 |
+
"learning_rate": 0.00016854274223350397,
|
| 2929 |
+
"loss": 0.8415,
|
| 2930 |
+
"step": 415
|
| 2931 |
+
},
|
| 2932 |
+
{
|
| 2933 |
+
"epoch": 0.18827789092554878,
|
| 2934 |
+
"grad_norm": 0.5125812292098999,
|
| 2935 |
+
"learning_rate": 0.00016839596730011138,
|
| 2936 |
+
"loss": 0.8479,
|
| 2937 |
+
"step": 416
|
| 2938 |
+
},
|
| 2939 |
+
{
|
| 2940 |
+
"epoch": 0.1887304820095044,
|
| 2941 |
+
"grad_norm": 0.6952138543128967,
|
| 2942 |
+
"learning_rate": 0.00016824891498304364,
|
| 2943 |
+
"loss": 1.1171,
|
| 2944 |
+
"step": 417
|
| 2945 |
+
},
|
| 2946 |
+
{
|
| 2947 |
+
"epoch": 0.18918307309346005,
|
| 2948 |
+
"grad_norm": 0.5572001934051514,
|
| 2949 |
+
"learning_rate": 0.00016810158587867973,
|
| 2950 |
+
"loss": 0.9817,
|
| 2951 |
+
"step": 418
|
| 2952 |
+
},
|
| 2953 |
+
{
|
| 2954 |
+
"epoch": 0.1896356641774157,
|
| 2955 |
+
"grad_norm": 0.47780361771583557,
|
| 2956 |
+
"learning_rate": 0.00016795398058452104,
|
| 2957 |
+
"loss": 0.9416,
|
| 2958 |
+
"step": 419
|
| 2959 |
+
},
|
| 2960 |
+
{
|
| 2961 |
+
"epoch": 0.19008825526137135,
|
| 2962 |
+
"grad_norm": 0.4838089048862457,
|
| 2963 |
+
"learning_rate": 0.0001678060996991891,
|
| 2964 |
+
"loss": 0.8315,
|
| 2965 |
+
"step": 420
|
| 2966 |
+
},
|
| 2967 |
+
{
|
| 2968 |
+
"epoch": 0.190540846345327,
|
| 2969 |
+
"grad_norm": 0.4793652892112732,
|
| 2970 |
+
"learning_rate": 0.00016765794382242314,
|
| 2971 |
+
"loss": 0.908,
|
| 2972 |
+
"step": 421
|
| 2973 |
+
},
|
| 2974 |
+
{
|
| 2975 |
+
"epoch": 0.19099343742928265,
|
| 2976 |
+
"grad_norm": 0.4631733000278473,
|
| 2977 |
+
"learning_rate": 0.00016750951355507763,
|
| 2978 |
+
"loss": 0.689,
|
| 2979 |
+
"step": 422
|
| 2980 |
+
},
|
| 2981 |
+
{
|
| 2982 |
+
"epoch": 0.19144602851323828,
|
| 2983 |
+
"grad_norm": 0.5332536101341248,
|
| 2984 |
+
"learning_rate": 0.00016736080949911978,
|
| 2985 |
+
"loss": 0.9532,
|
| 2986 |
+
"step": 423
|
| 2987 |
+
},
|
| 2988 |
+
{
|
| 2989 |
+
"epoch": 0.19189861959719393,
|
| 2990 |
+
"grad_norm": 0.5441685914993286,
|
| 2991 |
+
"learning_rate": 0.00016721183225762727,
|
| 2992 |
+
"loss": 0.8802,
|
| 2993 |
+
"step": 424
|
| 2994 |
+
},
|
| 2995 |
+
{
|
| 2996 |
+
"epoch": 0.19235121068114958,
|
| 2997 |
+
"grad_norm": 0.687839686870575,
|
| 2998 |
+
"learning_rate": 0.00016706258243478562,
|
| 2999 |
+
"loss": 0.771,
|
| 3000 |
+
"step": 425
|
| 3001 |
+
},
|
| 3002 |
+
{
|
| 3003 |
+
"epoch": 0.19280380176510523,
|
| 3004 |
+
"grad_norm": 0.5836572647094727,
|
| 3005 |
+
"learning_rate": 0.00016691306063588583,
|
| 3006 |
+
"loss": 1.0452,
|
| 3007 |
+
"step": 426
|
| 3008 |
+
},
|
| 3009 |
+
{
|
| 3010 |
+
"epoch": 0.19325639284906088,
|
| 3011 |
+
"grad_norm": 0.5064947605133057,
|
| 3012 |
+
"learning_rate": 0.00016676326746732195,
|
| 3013 |
+
"loss": 0.93,
|
| 3014 |
+
"step": 427
|
| 3015 |
+
},
|
| 3016 |
+
{
|
| 3017 |
+
"epoch": 0.19370898393301653,
|
| 3018 |
+
"grad_norm": 0.4416975975036621,
|
| 3019 |
+
"learning_rate": 0.00016661320353658857,
|
| 3020 |
+
"loss": 0.7413,
|
| 3021 |
+
"step": 428
|
| 3022 |
+
},
|
| 3023 |
+
{
|
| 3024 |
+
"epoch": 0.19416157501697218,
|
| 3025 |
+
"grad_norm": 0.6051701903343201,
|
| 3026 |
+
"learning_rate": 0.0001664628694522783,
|
| 3027 |
+
"loss": 0.7676,
|
| 3028 |
+
"step": 429
|
| 3029 |
+
},
|
| 3030 |
+
{
|
| 3031 |
+
"epoch": 0.1946141661009278,
|
| 3032 |
+
"grad_norm": 0.6898373961448669,
|
| 3033 |
+
"learning_rate": 0.00016631226582407952,
|
| 3034 |
+
"loss": 1.0243,
|
| 3035 |
+
"step": 430
|
| 3036 |
+
},
|
| 3037 |
+
{
|
| 3038 |
+
"epoch": 0.19506675718488345,
|
| 3039 |
+
"grad_norm": 0.4763202369213104,
|
| 3040 |
+
"learning_rate": 0.00016616139326277365,
|
| 3041 |
+
"loss": 0.6903,
|
| 3042 |
+
"step": 431
|
| 3043 |
+
},
|
| 3044 |
+
{
|
| 3045 |
+
"epoch": 0.1955193482688391,
|
| 3046 |
+
"grad_norm": 0.5373172163963318,
|
| 3047 |
+
"learning_rate": 0.0001660102523802328,
|
| 3048 |
+
"loss": 0.9375,
|
| 3049 |
+
"step": 432
|
| 3050 |
+
},
|
| 3051 |
+
{
|
| 3052 |
+
"epoch": 0.19597193935279475,
|
| 3053 |
+
"grad_norm": 0.4901677072048187,
|
| 3054 |
+
"learning_rate": 0.00016585884378941725,
|
| 3055 |
+
"loss": 0.8726,
|
| 3056 |
+
"step": 433
|
| 3057 |
+
},
|
| 3058 |
+
{
|
| 3059 |
+
"epoch": 0.1964245304367504,
|
| 3060 |
+
"grad_norm": 0.440965473651886,
|
| 3061 |
+
"learning_rate": 0.0001657071681043731,
|
| 3062 |
+
"loss": 0.7162,
|
| 3063 |
+
"step": 434
|
| 3064 |
+
},
|
| 3065 |
+
{
|
| 3066 |
+
"epoch": 0.19687712152070605,
|
| 3067 |
+
"grad_norm": 0.5141696333885193,
|
| 3068 |
+
"learning_rate": 0.0001655552259402295,
|
| 3069 |
+
"loss": 0.7813,
|
| 3070 |
+
"step": 435
|
| 3071 |
+
},
|
| 3072 |
+
{
|
| 3073 |
+
"epoch": 0.1973297126046617,
|
| 3074 |
+
"grad_norm": 0.44578734040260315,
|
| 3075 |
+
"learning_rate": 0.00016540301791319645,
|
| 3076 |
+
"loss": 0.7619,
|
| 3077 |
+
"step": 436
|
| 3078 |
+
},
|
| 3079 |
+
{
|
| 3080 |
+
"epoch": 0.19778230368861732,
|
| 3081 |
+
"grad_norm": 0.4049105644226074,
|
| 3082 |
+
"learning_rate": 0.00016525054464056212,
|
| 3083 |
+
"loss": 0.6142,
|
| 3084 |
+
"step": 437
|
| 3085 |
+
},
|
| 3086 |
+
{
|
| 3087 |
+
"epoch": 0.19823489477257297,
|
| 3088 |
+
"grad_norm": 0.46392935514450073,
|
| 3089 |
+
"learning_rate": 0.0001650978067406904,
|
| 3090 |
+
"loss": 0.8162,
|
| 3091 |
+
"step": 438
|
| 3092 |
+
},
|
| 3093 |
+
{
|
| 3094 |
+
"epoch": 0.19868748585652862,
|
| 3095 |
+
"grad_norm": 0.4067855775356293,
|
| 3096 |
+
"learning_rate": 0.00016494480483301836,
|
| 3097 |
+
"loss": 0.7385,
|
| 3098 |
+
"step": 439
|
| 3099 |
+
},
|
| 3100 |
+
{
|
| 3101 |
+
"epoch": 0.19914007694048427,
|
| 3102 |
+
"grad_norm": 0.47872698307037354,
|
| 3103 |
+
"learning_rate": 0.0001647915395380539,
|
| 3104 |
+
"loss": 0.7628,
|
| 3105 |
+
"step": 440
|
| 3106 |
+
},
|
| 3107 |
+
{
|
| 3108 |
+
"epoch": 0.19959266802443992,
|
| 3109 |
+
"grad_norm": 0.4723743200302124,
|
| 3110 |
+
"learning_rate": 0.0001646380114773729,
|
| 3111 |
+
"loss": 0.8442,
|
| 3112 |
+
"step": 441
|
| 3113 |
+
},
|
| 3114 |
+
{
|
| 3115 |
+
"epoch": 0.20004525910839557,
|
| 3116 |
+
"grad_norm": 0.4618031084537506,
|
| 3117 |
+
"learning_rate": 0.00016448422127361706,
|
| 3118 |
+
"loss": 0.9227,
|
| 3119 |
+
"step": 442
|
| 3120 |
+
},
|
| 3121 |
+
{
|
| 3122 |
+
"epoch": 0.20049785019235122,
|
| 3123 |
+
"grad_norm": 0.4881635010242462,
|
| 3124 |
+
"learning_rate": 0.00016433016955049115,
|
| 3125 |
+
"loss": 0.8212,
|
| 3126 |
+
"step": 443
|
| 3127 |
+
},
|
| 3128 |
+
{
|
| 3129 |
+
"epoch": 0.20095044127630685,
|
| 3130 |
+
"grad_norm": 0.49687886238098145,
|
| 3131 |
+
"learning_rate": 0.00016417585693276058,
|
| 3132 |
+
"loss": 0.9704,
|
| 3133 |
+
"step": 444
|
| 3134 |
+
},
|
| 3135 |
+
{
|
| 3136 |
+
"epoch": 0.2014030323602625,
|
| 3137 |
+
"grad_norm": 0.5278611183166504,
|
| 3138 |
+
"learning_rate": 0.00016402128404624882,
|
| 3139 |
+
"loss": 0.9722,
|
| 3140 |
+
"step": 445
|
| 3141 |
+
},
|
| 3142 |
+
{
|
| 3143 |
+
"epoch": 0.20185562344421815,
|
| 3144 |
+
"grad_norm": 0.5232154726982117,
|
| 3145 |
+
"learning_rate": 0.0001638664515178348,
|
| 3146 |
+
"loss": 1.0113,
|
| 3147 |
+
"step": 446
|
| 3148 |
+
},
|
| 3149 |
+
{
|
| 3150 |
+
"epoch": 0.2023082145281738,
|
| 3151 |
+
"grad_norm": 0.4725476801395416,
|
| 3152 |
+
"learning_rate": 0.00016371135997545058,
|
| 3153 |
+
"loss": 0.7659,
|
| 3154 |
+
"step": 447
|
| 3155 |
+
},
|
| 3156 |
+
{
|
| 3157 |
+
"epoch": 0.20276080561212945,
|
| 3158 |
+
"grad_norm": 0.458344966173172,
|
| 3159 |
+
"learning_rate": 0.00016355601004807856,
|
| 3160 |
+
"loss": 0.7465,
|
| 3161 |
+
"step": 448
|
| 3162 |
+
},
|
| 3163 |
+
{
|
| 3164 |
+
"epoch": 0.2032133966960851,
|
| 3165 |
+
"grad_norm": 1.0862559080123901,
|
| 3166 |
+
"learning_rate": 0.00016340040236574902,
|
| 3167 |
+
"loss": 1.1804,
|
| 3168 |
+
"step": 449
|
| 3169 |
+
},
|
| 3170 |
+
{
|
| 3171 |
+
"epoch": 0.20366598778004075,
|
| 3172 |
+
"grad_norm": 0.496658593416214,
|
| 3173 |
+
"learning_rate": 0.00016324453755953773,
|
| 3174 |
+
"loss": 0.766,
|
| 3175 |
+
"step": 450
|
| 3176 |
+
},
|
| 3177 |
+
{
|
| 3178 |
+
"epoch": 0.20411857886399637,
|
| 3179 |
+
"grad_norm": 0.6810061931610107,
|
| 3180 |
+
"learning_rate": 0.00016308841626156307,
|
| 3181 |
+
"loss": 1.0167,
|
| 3182 |
+
"step": 451
|
| 3183 |
+
},
|
| 3184 |
+
{
|
| 3185 |
+
"epoch": 0.20457116994795202,
|
| 3186 |
+
"grad_norm": 0.5212746858596802,
|
| 3187 |
+
"learning_rate": 0.00016293203910498376,
|
| 3188 |
+
"loss": 0.7169,
|
| 3189 |
+
"step": 452
|
| 3190 |
+
},
|
| 3191 |
+
{
|
| 3192 |
+
"epoch": 0.20502376103190767,
|
| 3193 |
+
"grad_norm": 0.46362701058387756,
|
| 3194 |
+
"learning_rate": 0.00016277540672399608,
|
| 3195 |
+
"loss": 0.8142,
|
| 3196 |
+
"step": 453
|
| 3197 |
+
},
|
| 3198 |
+
{
|
| 3199 |
+
"epoch": 0.20547635211586332,
|
| 3200 |
+
"grad_norm": 0.5136375427246094,
|
| 3201 |
+
"learning_rate": 0.00016261851975383137,
|
| 3202 |
+
"loss": 0.844,
|
| 3203 |
+
"step": 454
|
| 3204 |
+
},
|
| 3205 |
+
{
|
| 3206 |
+
"epoch": 0.20592894319981897,
|
| 3207 |
+
"grad_norm": 0.47330260276794434,
|
| 3208 |
+
"learning_rate": 0.00016246137883075363,
|
| 3209 |
+
"loss": 0.7709,
|
| 3210 |
+
"step": 455
|
| 3211 |
+
},
|
| 3212 |
+
{
|
| 3213 |
+
"epoch": 0.20638153428377462,
|
| 3214 |
+
"grad_norm": 0.6377627849578857,
|
| 3215 |
+
"learning_rate": 0.0001623039845920566,
|
| 3216 |
+
"loss": 0.7982,
|
| 3217 |
+
"step": 456
|
| 3218 |
+
},
|
| 3219 |
+
{
|
| 3220 |
+
"epoch": 0.20683412536773024,
|
| 3221 |
+
"grad_norm": 0.6286625266075134,
|
| 3222 |
+
"learning_rate": 0.00016214633767606143,
|
| 3223 |
+
"loss": 0.8048,
|
| 3224 |
+
"step": 457
|
| 3225 |
+
},
|
| 3226 |
+
{
|
| 3227 |
+
"epoch": 0.2072867164516859,
|
| 3228 |
+
"grad_norm": 0.4780018925666809,
|
| 3229 |
+
"learning_rate": 0.00016198843872211404,
|
| 3230 |
+
"loss": 0.8316,
|
| 3231 |
+
"step": 458
|
| 3232 |
+
},
|
| 3233 |
+
{
|
| 3234 |
+
"epoch": 0.20773930753564154,
|
| 3235 |
+
"grad_norm": 0.5281711220741272,
|
| 3236 |
+
"learning_rate": 0.00016183028837058247,
|
| 3237 |
+
"loss": 0.9401,
|
| 3238 |
+
"step": 459
|
| 3239 |
+
},
|
| 3240 |
+
{
|
| 3241 |
+
"epoch": 0.2081918986195972,
|
| 3242 |
+
"grad_norm": 0.46986955404281616,
|
| 3243 |
+
"learning_rate": 0.00016167188726285434,
|
| 3244 |
+
"loss": 0.728,
|
| 3245 |
+
"step": 460
|
| 3246 |
+
},
|
| 3247 |
+
{
|
| 3248 |
+
"epoch": 0.20864448970355284,
|
| 3249 |
+
"grad_norm": 0.4838784635066986,
|
| 3250 |
+
"learning_rate": 0.00016151323604133414,
|
| 3251 |
+
"loss": 0.7384,
|
| 3252 |
+
"step": 461
|
| 3253 |
+
},
|
| 3254 |
+
{
|
| 3255 |
+
"epoch": 0.2090970807875085,
|
| 3256 |
+
"grad_norm": 0.441522479057312,
|
| 3257 |
+
"learning_rate": 0.0001613543353494409,
|
| 3258 |
+
"loss": 0.738,
|
| 3259 |
+
"step": 462
|
| 3260 |
+
},
|
| 3261 |
+
{
|
| 3262 |
+
"epoch": 0.20954967187146414,
|
| 3263 |
+
"grad_norm": 0.5253183841705322,
|
| 3264 |
+
"learning_rate": 0.0001611951858316052,
|
| 3265 |
+
"loss": 0.8767,
|
| 3266 |
+
"step": 463
|
| 3267 |
+
},
|
| 3268 |
+
{
|
| 3269 |
+
"epoch": 0.21000226295541977,
|
| 3270 |
+
"grad_norm": 0.542733371257782,
|
| 3271 |
+
"learning_rate": 0.00016103578813326684,
|
| 3272 |
+
"loss": 0.9233,
|
| 3273 |
+
"step": 464
|
| 3274 |
+
},
|
| 3275 |
+
{
|
| 3276 |
+
"epoch": 0.21045485403937542,
|
| 3277 |
+
"grad_norm": 0.548744261264801,
|
| 3278 |
+
"learning_rate": 0.00016087614290087208,
|
| 3279 |
+
"loss": 0.9597,
|
| 3280 |
+
"step": 465
|
| 3281 |
+
},
|
| 3282 |
+
{
|
| 3283 |
+
"epoch": 0.21090744512333107,
|
| 3284 |
+
"grad_norm": 0.4585428833961487,
|
| 3285 |
+
"learning_rate": 0.00016071625078187114,
|
| 3286 |
+
"loss": 0.7097,
|
| 3287 |
+
"step": 466
|
| 3288 |
+
},
|
| 3289 |
+
{
|
| 3290 |
+
"epoch": 0.21136003620728672,
|
| 3291 |
+
"grad_norm": 0.4707442820072174,
|
| 3292 |
+
"learning_rate": 0.0001605561124247154,
|
| 3293 |
+
"loss": 0.6851,
|
| 3294 |
+
"step": 467
|
| 3295 |
+
},
|
| 3296 |
+
{
|
| 3297 |
+
"epoch": 0.21181262729124237,
|
| 3298 |
+
"grad_norm": 0.44789552688598633,
|
| 3299 |
+
"learning_rate": 0.00016039572847885504,
|
| 3300 |
+
"loss": 0.65,
|
| 3301 |
+
"step": 468
|
| 3302 |
+
},
|
| 3303 |
+
{
|
| 3304 |
+
"epoch": 0.21226521837519802,
|
| 3305 |
+
"grad_norm": 0.4923850893974304,
|
| 3306 |
+
"learning_rate": 0.00016023509959473605,
|
| 3307 |
+
"loss": 0.9482,
|
| 3308 |
+
"step": 469
|
| 3309 |
+
},
|
| 3310 |
+
{
|
| 3311 |
+
"epoch": 0.21271780945915367,
|
| 3312 |
+
"grad_norm": 0.5251060724258423,
|
| 3313 |
+
"learning_rate": 0.0001600742264237979,
|
| 3314 |
+
"loss": 0.865,
|
| 3315 |
+
"step": 470
|
| 3316 |
+
},
|
| 3317 |
+
{
|
| 3318 |
+
"epoch": 0.2131704005431093,
|
| 3319 |
+
"grad_norm": 0.43079182505607605,
|
| 3320 |
+
"learning_rate": 0.0001599131096184707,
|
| 3321 |
+
"loss": 0.6448,
|
| 3322 |
+
"step": 471
|
| 3323 |
+
},
|
| 3324 |
+
{
|
| 3325 |
+
"epoch": 0.21362299162706494,
|
| 3326 |
+
"grad_norm": 0.4741304814815521,
|
| 3327 |
+
"learning_rate": 0.00015975174983217275,
|
| 3328 |
+
"loss": 0.7992,
|
| 3329 |
+
"step": 472
|
| 3330 |
+
},
|
| 3331 |
+
{
|
| 3332 |
+
"epoch": 0.2140755827110206,
|
| 3333 |
+
"grad_norm": 0.5322284698486328,
|
| 3334 |
+
"learning_rate": 0.0001595901477193076,
|
| 3335 |
+
"loss": 0.7397,
|
| 3336 |
+
"step": 473
|
| 3337 |
+
},
|
| 3338 |
+
{
|
| 3339 |
+
"epoch": 0.21452817379497624,
|
| 3340 |
+
"grad_norm": 0.490506112575531,
|
| 3341 |
+
"learning_rate": 0.00015942830393526176,
|
| 3342 |
+
"loss": 0.8015,
|
| 3343 |
+
"step": 474
|
| 3344 |
+
},
|
| 3345 |
+
{
|
| 3346 |
+
"epoch": 0.2149807648789319,
|
| 3347 |
+
"grad_norm": 0.5263440012931824,
|
| 3348 |
+
"learning_rate": 0.0001592662191364017,
|
| 3349 |
+
"loss": 0.9122,
|
| 3350 |
+
"step": 475
|
| 3351 |
+
},
|
| 3352 |
+
{
|
| 3353 |
+
"epoch": 0.21543335596288754,
|
| 3354 |
+
"grad_norm": 0.5945788621902466,
|
| 3355 |
+
"learning_rate": 0.0001591038939800714,
|
| 3356 |
+
"loss": 1.0784,
|
| 3357 |
+
"step": 476
|
| 3358 |
+
},
|
| 3359 |
+
{
|
| 3360 |
+
"epoch": 0.2158859470468432,
|
| 3361 |
+
"grad_norm": 0.510407030582428,
|
| 3362 |
+
"learning_rate": 0.00015894132912458968,
|
| 3363 |
+
"loss": 1.0042,
|
| 3364 |
+
"step": 477
|
| 3365 |
+
},
|
| 3366 |
+
{
|
| 3367 |
+
"epoch": 0.2163385381307988,
|
| 3368 |
+
"grad_norm": 0.4436274766921997,
|
| 3369 |
+
"learning_rate": 0.00015877852522924732,
|
| 3370 |
+
"loss": 0.7253,
|
| 3371 |
+
"step": 478
|
| 3372 |
+
},
|
| 3373 |
+
{
|
| 3374 |
+
"epoch": 0.21679112921475446,
|
| 3375 |
+
"grad_norm": 0.45981746912002563,
|
| 3376 |
+
"learning_rate": 0.0001586154829543047,
|
| 3377 |
+
"loss": 0.8309,
|
| 3378 |
+
"step": 479
|
| 3379 |
+
},
|
| 3380 |
+
{
|
| 3381 |
+
"epoch": 0.2172437202987101,
|
| 3382 |
+
"grad_norm": 1.0382171869277954,
|
| 3383 |
+
"learning_rate": 0.0001584522029609889,
|
| 3384 |
+
"loss": 0.9175,
|
| 3385 |
+
"step": 480
|
| 3386 |
+
},
|
| 3387 |
+
{
|
| 3388 |
+
"epoch": 0.21769631138266576,
|
| 3389 |
+
"grad_norm": 0.4632301330566406,
|
| 3390 |
+
"learning_rate": 0.00015828868591149104,
|
| 3391 |
+
"loss": 0.771,
|
| 3392 |
+
"step": 481
|
| 3393 |
+
},
|
| 3394 |
+
{
|
| 3395 |
+
"epoch": 0.2181489024666214,
|
| 3396 |
+
"grad_norm": 0.49221765995025635,
|
| 3397 |
+
"learning_rate": 0.00015812493246896366,
|
| 3398 |
+
"loss": 0.7167,
|
| 3399 |
+
"step": 482
|
| 3400 |
+
},
|
| 3401 |
+
{
|
| 3402 |
+
"epoch": 0.21860149355057706,
|
| 3403 |
+
"grad_norm": 0.5978690385818481,
|
| 3404 |
+
"learning_rate": 0.00015796094329751807,
|
| 3405 |
+
"loss": 0.9947,
|
| 3406 |
+
"step": 483
|
| 3407 |
+
},
|
| 3408 |
+
{
|
| 3409 |
+
"epoch": 0.2190540846345327,
|
| 3410 |
+
"grad_norm": 0.48670709133148193,
|
| 3411 |
+
"learning_rate": 0.0001577967190622215,
|
| 3412 |
+
"loss": 0.935,
|
| 3413 |
+
"step": 484
|
| 3414 |
+
},
|
| 3415 |
+
{
|
| 3416 |
+
"epoch": 0.21950667571848834,
|
| 3417 |
+
"grad_norm": 0.48809000849723816,
|
| 3418 |
+
"learning_rate": 0.00015763226042909455,
|
| 3419 |
+
"loss": 0.9351,
|
| 3420 |
+
"step": 485
|
| 3421 |
+
},
|
| 3422 |
+
{
|
| 3423 |
+
"epoch": 0.219959266802444,
|
| 3424 |
+
"grad_norm": 0.6401401162147522,
|
| 3425 |
+
"learning_rate": 0.00015746756806510838,
|
| 3426 |
+
"loss": 1.2269,
|
| 3427 |
+
"step": 486
|
| 3428 |
+
},
|
| 3429 |
+
{
|
| 3430 |
+
"epoch": 0.22041185788639964,
|
| 3431 |
+
"grad_norm": 0.6913933753967285,
|
| 3432 |
+
"learning_rate": 0.00015730264263818212,
|
| 3433 |
+
"loss": 0.972,
|
| 3434 |
+
"step": 487
|
| 3435 |
+
},
|
| 3436 |
+
{
|
| 3437 |
+
"epoch": 0.2208644489703553,
|
| 3438 |
+
"grad_norm": 0.44599664211273193,
|
| 3439 |
+
"learning_rate": 0.0001571374848171801,
|
| 3440 |
+
"loss": 0.8677,
|
| 3441 |
+
"step": 488
|
| 3442 |
+
},
|
| 3443 |
+
{
|
| 3444 |
+
"epoch": 0.22131704005431094,
|
| 3445 |
+
"grad_norm": 1.1593374013900757,
|
| 3446 |
+
"learning_rate": 0.00015697209527190906,
|
| 3447 |
+
"loss": 0.7672,
|
| 3448 |
+
"step": 489
|
| 3449 |
+
},
|
| 3450 |
+
{
|
| 3451 |
+
"epoch": 0.2217696311382666,
|
| 3452 |
+
"grad_norm": 0.47981521487236023,
|
| 3453 |
+
"learning_rate": 0.00015680647467311557,
|
| 3454 |
+
"loss": 0.8651,
|
| 3455 |
+
"step": 490
|
| 3456 |
+
},
|
| 3457 |
+
{
|
| 3458 |
+
"epoch": 0.2222222222222222,
|
| 3459 |
+
"grad_norm": 0.44539451599121094,
|
| 3460 |
+
"learning_rate": 0.00015664062369248328,
|
| 3461 |
+
"loss": 0.7102,
|
| 3462 |
+
"step": 491
|
| 3463 |
+
},
|
| 3464 |
+
{
|
| 3465 |
+
"epoch": 0.22267481330617786,
|
| 3466 |
+
"grad_norm": 0.6086403131484985,
|
| 3467 |
+
"learning_rate": 0.00015647454300263012,
|
| 3468 |
+
"loss": 0.9978,
|
| 3469 |
+
"step": 492
|
| 3470 |
+
},
|
| 3471 |
+
{
|
| 3472 |
+
"epoch": 0.2231274043901335,
|
| 3473 |
+
"grad_norm": 0.512725293636322,
|
| 3474 |
+
"learning_rate": 0.00015630823327710558,
|
| 3475 |
+
"loss": 0.846,
|
| 3476 |
+
"step": 493
|
| 3477 |
+
},
|
| 3478 |
+
{
|
| 3479 |
+
"epoch": 0.22357999547408916,
|
| 3480 |
+
"grad_norm": 0.5239939093589783,
|
| 3481 |
+
"learning_rate": 0.0001561416951903881,
|
| 3482 |
+
"loss": 1.01,
|
| 3483 |
+
"step": 494
|
| 3484 |
+
},
|
| 3485 |
+
{
|
| 3486 |
+
"epoch": 0.2240325865580448,
|
| 3487 |
+
"grad_norm": 0.4727892279624939,
|
| 3488 |
+
"learning_rate": 0.00015597492941788222,
|
| 3489 |
+
"loss": 0.7412,
|
| 3490 |
+
"step": 495
|
| 3491 |
+
},
|
| 3492 |
+
{
|
| 3493 |
+
"epoch": 0.22448517764200046,
|
| 3494 |
+
"grad_norm": 0.576046347618103,
|
| 3495 |
+
"learning_rate": 0.00015580793663591585,
|
| 3496 |
+
"loss": 0.9968,
|
| 3497 |
+
"step": 496
|
| 3498 |
+
},
|
| 3499 |
+
{
|
| 3500 |
+
"epoch": 0.2249377687259561,
|
| 3501 |
+
"grad_norm": 0.4874511957168579,
|
| 3502 |
+
"learning_rate": 0.00015564071752173758,
|
| 3503 |
+
"loss": 0.7964,
|
| 3504 |
+
"step": 497
|
| 3505 |
+
},
|
| 3506 |
+
{
|
| 3507 |
+
"epoch": 0.22539035980991173,
|
| 3508 |
+
"grad_norm": 0.3970421552658081,
|
| 3509 |
+
"learning_rate": 0.0001554732727535139,
|
| 3510 |
+
"loss": 0.5187,
|
| 3511 |
+
"step": 498
|
| 3512 |
+
},
|
| 3513 |
+
{
|
| 3514 |
+
"epoch": 0.22584295089386738,
|
| 3515 |
+
"grad_norm": 0.430368036031723,
|
| 3516 |
+
"learning_rate": 0.0001553056030103264,
|
| 3517 |
+
"loss": 0.6658,
|
| 3518 |
+
"step": 499
|
| 3519 |
+
},
|
| 3520 |
+
{
|
| 3521 |
+
"epoch": 0.22629554197782303,
|
| 3522 |
+
"grad_norm": 0.5428101420402527,
|
| 3523 |
+
"learning_rate": 0.00015513770897216918,
|
| 3524 |
+
"loss": 1.1194,
|
| 3525 |
+
"step": 500
|
| 3526 |
+
},
|
| 3527 |
+
{
|
| 3528 |
+
"epoch": 0.22674813306177868,
|
| 3529 |
+
"grad_norm": 0.6447327136993408,
|
| 3530 |
+
"learning_rate": 0.00015496959131994586,
|
| 3531 |
+
"loss": 1.0849,
|
| 3532 |
+
"step": 501
|
| 3533 |
+
},
|
| 3534 |
+
{
|
| 3535 |
+
"epoch": 0.22720072414573433,
|
| 3536 |
+
"grad_norm": 0.4934568405151367,
|
| 3537 |
+
"learning_rate": 0.00015480125073546704,
|
| 3538 |
+
"loss": 0.8384,
|
| 3539 |
+
"step": 502
|
| 3540 |
+
},
|
| 3541 |
+
{
|
| 3542 |
+
"epoch": 0.22765331522968998,
|
| 3543 |
+
"grad_norm": 0.5266446471214294,
|
| 3544 |
+
"learning_rate": 0.0001546326879014473,
|
| 3545 |
+
"loss": 1.0149,
|
| 3546 |
+
"step": 503
|
| 3547 |
+
},
|
| 3548 |
+
{
|
| 3549 |
+
"epoch": 0.22810590631364563,
|
| 3550 |
+
"grad_norm": 0.5469086766242981,
|
| 3551 |
+
"learning_rate": 0.00015446390350150273,
|
| 3552 |
+
"loss": 0.9152,
|
| 3553 |
+
"step": 504
|
| 3554 |
+
},
|
| 3555 |
+
{
|
| 3556 |
+
"epoch": 0.22855849739760126,
|
| 3557 |
+
"grad_norm": 0.4938475787639618,
|
| 3558 |
+
"learning_rate": 0.0001542948982201479,
|
| 3559 |
+
"loss": 0.8845,
|
| 3560 |
+
"step": 505
|
| 3561 |
+
},
|
| 3562 |
+
{
|
| 3563 |
+
"epoch": 0.2290110884815569,
|
| 3564 |
+
"grad_norm": 0.40693241357803345,
|
| 3565 |
+
"learning_rate": 0.00015412567274279316,
|
| 3566 |
+
"loss": 0.6758,
|
| 3567 |
+
"step": 506
|
| 3568 |
+
},
|
| 3569 |
+
{
|
| 3570 |
+
"epoch": 0.22946367956551256,
|
| 3571 |
+
"grad_norm": 0.46620872616767883,
|
| 3572 |
+
"learning_rate": 0.00015395622775574193,
|
| 3573 |
+
"loss": 0.7524,
|
| 3574 |
+
"step": 507
|
| 3575 |
+
},
|
| 3576 |
+
{
|
| 3577 |
+
"epoch": 0.2299162706494682,
|
| 3578 |
+
"grad_norm": 0.4872758984565735,
|
| 3579 |
+
"learning_rate": 0.00015378656394618787,
|
| 3580 |
+
"loss": 0.8768,
|
| 3581 |
+
"step": 508
|
| 3582 |
+
},
|
| 3583 |
+
{
|
| 3584 |
+
"epoch": 0.23036886173342386,
|
| 3585 |
+
"grad_norm": 0.5335472226142883,
|
| 3586 |
+
"learning_rate": 0.00015361668200221203,
|
| 3587 |
+
"loss": 0.9473,
|
| 3588 |
+
"step": 509
|
| 3589 |
+
},
|
| 3590 |
+
{
|
| 3591 |
+
"epoch": 0.2308214528173795,
|
| 3592 |
+
"grad_norm": 0.5027713179588318,
|
| 3593 |
+
"learning_rate": 0.0001534465826127801,
|
| 3594 |
+
"loss": 0.9785,
|
| 3595 |
+
"step": 510
|
| 3596 |
+
},
|
| 3597 |
+
{
|
| 3598 |
+
"epoch": 0.23127404390133513,
|
| 3599 |
+
"grad_norm": 0.4740453064441681,
|
| 3600 |
+
"learning_rate": 0.00015327626646773976,
|
| 3601 |
+
"loss": 0.8903,
|
| 3602 |
+
"step": 511
|
| 3603 |
+
},
|
| 3604 |
+
{
|
| 3605 |
+
"epoch": 0.23172663498529078,
|
| 3606 |
+
"grad_norm": 0.5740635395050049,
|
| 3607 |
+
"learning_rate": 0.00015310573425781767,
|
| 3608 |
+
"loss": 0.9505,
|
| 3609 |
+
"step": 512
|
| 3610 |
+
},
|
| 3611 |
+
{
|
| 3612 |
+
"epoch": 0.23217922606924643,
|
| 3613 |
+
"grad_norm": 0.5359417796134949,
|
| 3614 |
+
"learning_rate": 0.0001529349866746167,
|
| 3615 |
+
"loss": 0.7476,
|
| 3616 |
+
"step": 513
|
| 3617 |
+
},
|
| 3618 |
+
{
|
| 3619 |
+
"epoch": 0.23263181715320208,
|
| 3620 |
+
"grad_norm": 0.37038734555244446,
|
| 3621 |
+
"learning_rate": 0.0001527640244106133,
|
| 3622 |
+
"loss": 0.5349,
|
| 3623 |
+
"step": 514
|
| 3624 |
+
},
|
| 3625 |
+
{
|
| 3626 |
+
"epoch": 0.23308440823715773,
|
| 3627 |
+
"grad_norm": 0.43701088428497314,
|
| 3628 |
+
"learning_rate": 0.0001525928481591544,
|
| 3629 |
+
"loss": 0.4619,
|
| 3630 |
+
"step": 515
|
| 3631 |
+
},
|
| 3632 |
+
{
|
| 3633 |
+
"epoch": 0.23353699932111338,
|
| 3634 |
+
"grad_norm": 0.5714133977890015,
|
| 3635 |
+
"learning_rate": 0.00015242145861445498,
|
| 3636 |
+
"loss": 1.015,
|
| 3637 |
+
"step": 516
|
| 3638 |
+
},
|
| 3639 |
+
{
|
| 3640 |
+
"epoch": 0.23398959040506903,
|
| 3641 |
+
"grad_norm": 0.4387613832950592,
|
| 3642 |
+
"learning_rate": 0.0001522498564715949,
|
| 3643 |
+
"loss": 0.6386,
|
| 3644 |
+
"step": 517
|
| 3645 |
+
},
|
| 3646 |
+
{
|
| 3647 |
+
"epoch": 0.23444218148902465,
|
| 3648 |
+
"grad_norm": 0.5844204425811768,
|
| 3649 |
+
"learning_rate": 0.00015207804242651626,
|
| 3650 |
+
"loss": 0.8942,
|
| 3651 |
+
"step": 518
|
| 3652 |
+
},
|
| 3653 |
+
{
|
| 3654 |
+
"epoch": 0.2348947725729803,
|
| 3655 |
+
"grad_norm": 0.5000221133232117,
|
| 3656 |
+
"learning_rate": 0.0001519060171760205,
|
| 3657 |
+
"loss": 0.8325,
|
| 3658 |
+
"step": 519
|
| 3659 |
+
},
|
| 3660 |
+
{
|
| 3661 |
+
"epoch": 0.23534736365693595,
|
| 3662 |
+
"grad_norm": 0.5425203442573547,
|
| 3663 |
+
"learning_rate": 0.00015173378141776568,
|
| 3664 |
+
"loss": 1.0931,
|
| 3665 |
+
"step": 520
|
| 3666 |
+
},
|
| 3667 |
+
{
|
| 3668 |
+
"epoch": 0.2357999547408916,
|
| 3669 |
+
"grad_norm": 0.5176980495452881,
|
| 3670 |
+
"learning_rate": 0.00015156133585026357,
|
| 3671 |
+
"loss": 0.9856,
|
| 3672 |
+
"step": 521
|
| 3673 |
+
},
|
| 3674 |
+
{
|
| 3675 |
+
"epoch": 0.23625254582484725,
|
| 3676 |
+
"grad_norm": 0.42920199036598206,
|
| 3677 |
+
"learning_rate": 0.0001513886811728769,
|
| 3678 |
+
"loss": 0.6077,
|
| 3679 |
+
"step": 522
|
| 3680 |
+
},
|
| 3681 |
+
{
|
| 3682 |
+
"epoch": 0.2367051369088029,
|
| 3683 |
+
"grad_norm": 0.4761313498020172,
|
| 3684 |
+
"learning_rate": 0.00015121581808581622,
|
| 3685 |
+
"loss": 0.7548,
|
| 3686 |
+
"step": 523
|
| 3687 |
+
},
|
| 3688 |
+
{
|
| 3689 |
+
"epoch": 0.23715772799275855,
|
| 3690 |
+
"grad_norm": 0.5511419773101807,
|
| 3691 |
+
"learning_rate": 0.0001510427472901376,
|
| 3692 |
+
"loss": 0.9048,
|
| 3693 |
+
"step": 524
|
| 3694 |
+
},
|
| 3695 |
+
{
|
| 3696 |
+
"epoch": 0.23761031907671418,
|
| 3697 |
+
"grad_norm": 0.5572141408920288,
|
| 3698 |
+
"learning_rate": 0.00015086946948773922,
|
| 3699 |
+
"loss": 0.9789,
|
| 3700 |
+
"step": 525
|
| 3701 |
+
},
|
| 3702 |
+
{
|
| 3703 |
+
"epoch": 0.23806291016066983,
|
| 3704 |
+
"grad_norm": 0.4818233847618103,
|
| 3705 |
+
"learning_rate": 0.00015069598538135906,
|
| 3706 |
+
"loss": 0.9283,
|
| 3707 |
+
"step": 526
|
| 3708 |
+
},
|
| 3709 |
+
{
|
| 3710 |
+
"epoch": 0.23851550124462548,
|
| 3711 |
+
"grad_norm": 0.5013587474822998,
|
| 3712 |
+
"learning_rate": 0.0001505222956745715,
|
| 3713 |
+
"loss": 0.9243,
|
| 3714 |
+
"step": 527
|
| 3715 |
+
},
|
| 3716 |
+
{
|
| 3717 |
+
"epoch": 0.23896809232858113,
|
| 3718 |
+
"grad_norm": 0.4967738389968872,
|
| 3719 |
+
"learning_rate": 0.0001503484010717849,
|
| 3720 |
+
"loss": 0.8075,
|
| 3721 |
+
"step": 528
|
| 3722 |
+
},
|
| 3723 |
+
{
|
| 3724 |
+
"epoch": 0.23942068341253678,
|
| 3725 |
+
"grad_norm": 0.48400700092315674,
|
| 3726 |
+
"learning_rate": 0.00015017430227823864,
|
| 3727 |
+
"loss": 0.784,
|
| 3728 |
+
"step": 529
|
| 3729 |
+
},
|
| 3730 |
+
{
|
| 3731 |
+
"epoch": 0.23987327449649243,
|
| 3732 |
+
"grad_norm": 0.4794735908508301,
|
| 3733 |
+
"learning_rate": 0.00015000000000000001,
|
| 3734 |
+
"loss": 0.7469,
|
| 3735 |
+
"step": 530
|
| 3736 |
+
},
|
| 3737 |
+
{
|
| 3738 |
+
"epoch": 0.24032586558044808,
|
| 3739 |
+
"grad_norm": 0.5944988131523132,
|
| 3740 |
+
"learning_rate": 0.0001498254949439617,
|
| 3741 |
+
"loss": 1.2721,
|
| 3742 |
+
"step": 531
|
| 3743 |
+
},
|
| 3744 |
+
{
|
| 3745 |
+
"epoch": 0.2407784566644037,
|
| 3746 |
+
"grad_norm": 0.5433883666992188,
|
| 3747 |
+
"learning_rate": 0.0001496507878178388,
|
| 3748 |
+
"loss": 0.9277,
|
| 3749 |
+
"step": 532
|
| 3750 |
+
},
|
| 3751 |
+
{
|
| 3752 |
+
"epoch": 0.24123104774835935,
|
| 3753 |
+
"grad_norm": 0.5507298707962036,
|
| 3754 |
+
"learning_rate": 0.00014947587933016577,
|
| 3755 |
+
"loss": 0.94,
|
| 3756 |
+
"step": 533
|
| 3757 |
+
},
|
| 3758 |
+
{
|
| 3759 |
+
"epoch": 0.241683638832315,
|
| 3760 |
+
"grad_norm": 0.6457071900367737,
|
| 3761 |
+
"learning_rate": 0.00014930077019029375,
|
| 3762 |
+
"loss": 1.1681,
|
| 3763 |
+
"step": 534
|
| 3764 |
+
},
|
| 3765 |
+
{
|
| 3766 |
+
"epoch": 0.24213622991627065,
|
| 3767 |
+
"grad_norm": 0.5488935112953186,
|
| 3768 |
+
"learning_rate": 0.00014912546110838775,
|
| 3769 |
+
"loss": 0.9509,
|
| 3770 |
+
"step": 535
|
| 3771 |
+
},
|
| 3772 |
+
{
|
| 3773 |
+
"epoch": 0.2425888210002263,
|
| 3774 |
+
"grad_norm": 0.5333803296089172,
|
| 3775 |
+
"learning_rate": 0.0001489499527954235,
|
| 3776 |
+
"loss": 0.9315,
|
| 3777 |
+
"step": 536
|
| 3778 |
+
},
|
| 3779 |
+
{
|
| 3780 |
+
"epoch": 0.24304141208418195,
|
| 3781 |
+
"grad_norm": 0.4848494231700897,
|
| 3782 |
+
"learning_rate": 0.00014877424596318483,
|
| 3783 |
+
"loss": 0.7865,
|
| 3784 |
+
"step": 537
|
| 3785 |
+
},
|
| 3786 |
+
{
|
| 3787 |
+
"epoch": 0.24349400316813757,
|
| 3788 |
+
"grad_norm": 0.5007054209709167,
|
| 3789 |
+
"learning_rate": 0.0001485983413242606,
|
| 3790 |
+
"loss": 0.9231,
|
| 3791 |
+
"step": 538
|
| 3792 |
+
},
|
| 3793 |
+
{
|
| 3794 |
+
"epoch": 0.24394659425209322,
|
| 3795 |
+
"grad_norm": 0.5695723295211792,
|
| 3796 |
+
"learning_rate": 0.00014842223959204198,
|
| 3797 |
+
"loss": 0.9934,
|
| 3798 |
+
"step": 539
|
| 3799 |
+
},
|
| 3800 |
+
{
|
| 3801 |
+
"epoch": 0.24439918533604887,
|
| 3802 |
+
"grad_norm": 0.4827041029930115,
|
| 3803 |
+
"learning_rate": 0.00014824594148071934,
|
| 3804 |
+
"loss": 0.8522,
|
| 3805 |
+
"step": 540
|
| 3806 |
+
},
|
| 3807 |
+
{
|
| 3808 |
+
"epoch": 0.24485177642000452,
|
| 3809 |
+
"grad_norm": 0.5187906622886658,
|
| 3810 |
+
"learning_rate": 0.00014806944770527958,
|
| 3811 |
+
"loss": 0.9672,
|
| 3812 |
+
"step": 541
|
| 3813 |
+
},
|
| 3814 |
+
{
|
| 3815 |
+
"epoch": 0.24530436750396017,
|
| 3816 |
+
"grad_norm": 0.712968111038208,
|
| 3817 |
+
"learning_rate": 0.00014789275898150308,
|
| 3818 |
+
"loss": 0.9316,
|
| 3819 |
+
"step": 542
|
| 3820 |
+
},
|
| 3821 |
+
{
|
| 3822 |
+
"epoch": 0.24575695858791582,
|
| 3823 |
+
"grad_norm": 0.6967286467552185,
|
| 3824 |
+
"learning_rate": 0.00014771587602596084,
|
| 3825 |
+
"loss": 0.9736,
|
| 3826 |
+
"step": 543
|
| 3827 |
+
},
|
| 3828 |
+
{
|
| 3829 |
+
"epoch": 0.24620954967187147,
|
| 3830 |
+
"grad_norm": 0.3656039535999298,
|
| 3831 |
+
"learning_rate": 0.00014753879955601163,
|
| 3832 |
+
"loss": 0.4932,
|
| 3833 |
+
"step": 544
|
| 3834 |
+
},
|
| 3835 |
+
{
|
| 3836 |
+
"epoch": 0.2466621407558271,
|
| 3837 |
+
"grad_norm": 0.864976704120636,
|
| 3838 |
+
"learning_rate": 0.00014736153028979893,
|
| 3839 |
+
"loss": 1.0662,
|
| 3840 |
+
"step": 545
|
| 3841 |
+
},
|
| 3842 |
+
{
|
| 3843 |
+
"epoch": 0.24711473183978275,
|
| 3844 |
+
"grad_norm": 0.4753805994987488,
|
| 3845 |
+
"learning_rate": 0.0001471840689462482,
|
| 3846 |
+
"loss": 0.8889,
|
| 3847 |
+
"step": 546
|
| 3848 |
+
},
|
| 3849 |
+
{
|
| 3850 |
+
"epoch": 0.2475673229237384,
|
| 3851 |
+
"grad_norm": 0.4921295642852783,
|
| 3852 |
+
"learning_rate": 0.00014700641624506392,
|
| 3853 |
+
"loss": 0.8106,
|
| 3854 |
+
"step": 547
|
| 3855 |
+
},
|
| 3856 |
+
{
|
| 3857 |
+
"epoch": 0.24801991400769405,
|
| 3858 |
+
"grad_norm": 0.7320112586021423,
|
| 3859 |
+
"learning_rate": 0.00014682857290672648,
|
| 3860 |
+
"loss": 0.845,
|
| 3861 |
+
"step": 548
|
| 3862 |
+
},
|
| 3863 |
+
{
|
| 3864 |
+
"epoch": 0.2484725050916497,
|
| 3865 |
+
"grad_norm": 0.4472196698188782,
|
| 3866 |
+
"learning_rate": 0.0001466505396524895,
|
| 3867 |
+
"loss": 0.8076,
|
| 3868 |
+
"step": 549
|
| 3869 |
+
},
|
| 3870 |
+
{
|
| 3871 |
+
"epoch": 0.24892509617560535,
|
| 3872 |
+
"grad_norm": 0.43544888496398926,
|
| 3873 |
+
"learning_rate": 0.00014647231720437686,
|
| 3874 |
+
"loss": 0.8158,
|
| 3875 |
+
"step": 550
|
| 3876 |
+
},
|
| 3877 |
+
{
|
| 3878 |
+
"epoch": 0.249377687259561,
|
| 3879 |
+
"grad_norm": 0.39511433243751526,
|
| 3880 |
+
"learning_rate": 0.00014629390628517964,
|
| 3881 |
+
"loss": 0.6934,
|
| 3882 |
+
"step": 551
|
| 3883 |
+
},
|
| 3884 |
+
{
|
| 3885 |
+
"epoch": 0.24983027834351662,
|
| 3886 |
+
"grad_norm": 0.44961658120155334,
|
| 3887 |
+
"learning_rate": 0.00014611530761845335,
|
| 3888 |
+
"loss": 0.7865,
|
| 3889 |
+
"step": 552
|
| 3890 |
+
},
|
| 3891 |
+
{
|
| 3892 |
+
"epoch": 0.2502828694274723,
|
| 3893 |
+
"grad_norm": 0.4339733421802521,
|
| 3894 |
+
"learning_rate": 0.00014593652192851486,
|
| 3895 |
+
"loss": 0.5448,
|
| 3896 |
+
"step": 553
|
| 3897 |
+
},
|
| 3898 |
+
{
|
| 3899 |
+
"epoch": 0.2507354605114279,
|
| 3900 |
+
"grad_norm": 0.4500328600406647,
|
| 3901 |
+
"learning_rate": 0.00014575754994043956,
|
| 3902 |
+
"loss": 0.8066,
|
| 3903 |
+
"step": 554
|
| 3904 |
+
},
|
| 3905 |
+
{
|
| 3906 |
+
"epoch": 0.25118805159538354,
|
| 3907 |
+
"grad_norm": 0.5119399428367615,
|
| 3908 |
+
"learning_rate": 0.00014557839238005832,
|
| 3909 |
+
"loss": 1.0859,
|
| 3910 |
+
"step": 555
|
| 3911 |
+
},
|
| 3912 |
+
{
|
| 3913 |
+
"epoch": 0.2516406426793392,
|
| 3914 |
+
"grad_norm": 0.5100839138031006,
|
| 3915 |
+
"learning_rate": 0.00014539904997395468,
|
| 3916 |
+
"loss": 0.9779,
|
| 3917 |
+
"step": 556
|
| 3918 |
+
},
|
| 3919 |
+
{
|
| 3920 |
+
"epoch": 0.25209323376329484,
|
| 3921 |
+
"grad_norm": 0.5187854766845703,
|
| 3922 |
+
"learning_rate": 0.00014521952344946187,
|
| 3923 |
+
"loss": 1.0203,
|
| 3924 |
+
"step": 557
|
| 3925 |
+
},
|
| 3926 |
+
{
|
| 3927 |
+
"epoch": 0.2525458248472505,
|
| 3928 |
+
"grad_norm": 0.45579731464385986,
|
| 3929 |
+
"learning_rate": 0.0001450398135346597,
|
| 3930 |
+
"loss": 0.7401,
|
| 3931 |
+
"step": 558
|
| 3932 |
+
},
|
| 3933 |
+
{
|
| 3934 |
+
"epoch": 0.25299841593120614,
|
| 3935 |
+
"grad_norm": 0.47840386629104614,
|
| 3936 |
+
"learning_rate": 0.00014485992095837177,
|
| 3937 |
+
"loss": 0.78,
|
| 3938 |
+
"step": 559
|
| 3939 |
+
},
|
| 3940 |
+
{
|
| 3941 |
+
"epoch": 0.2534510070151618,
|
| 3942 |
+
"grad_norm": 0.49256467819213867,
|
| 3943 |
+
"learning_rate": 0.00014467984645016258,
|
| 3944 |
+
"loss": 0.9524,
|
| 3945 |
+
"step": 560
|
| 3946 |
+
},
|
| 3947 |
+
{
|
| 3948 |
+
"epoch": 0.25390359809911744,
|
| 3949 |
+
"grad_norm": 0.3887495696544647,
|
| 3950 |
+
"learning_rate": 0.00014449959074033434,
|
| 3951 |
+
"loss": 0.6893,
|
| 3952 |
+
"step": 561
|
| 3953 |
+
},
|
| 3954 |
+
{
|
| 3955 |
+
"epoch": 0.25435618918307307,
|
| 3956 |
+
"grad_norm": 0.46293124556541443,
|
| 3957 |
+
"learning_rate": 0.00014431915455992414,
|
| 3958 |
+
"loss": 0.7996,
|
| 3959 |
+
"step": 562
|
| 3960 |
+
},
|
| 3961 |
+
{
|
| 3962 |
+
"epoch": 0.25480878026702874,
|
| 3963 |
+
"grad_norm": 0.5254295468330383,
|
| 3964 |
+
"learning_rate": 0.00014413853864070103,
|
| 3965 |
+
"loss": 0.9435,
|
| 3966 |
+
"step": 563
|
| 3967 |
+
},
|
| 3968 |
+
{
|
| 3969 |
+
"epoch": 0.25526137135098437,
|
| 3970 |
+
"grad_norm": 0.4905354380607605,
|
| 3971 |
+
"learning_rate": 0.00014395774371516305,
|
| 3972 |
+
"loss": 0.7951,
|
| 3973 |
+
"step": 564
|
| 3974 |
+
},
|
| 3975 |
+
{
|
| 3976 |
+
"epoch": 0.25571396243494005,
|
| 3977 |
+
"grad_norm": 0.47427988052368164,
|
| 3978 |
+
"learning_rate": 0.00014377677051653404,
|
| 3979 |
+
"loss": 0.7587,
|
| 3980 |
+
"step": 565
|
| 3981 |
+
},
|
| 3982 |
+
{
|
| 3983 |
+
"epoch": 0.25616655351889567,
|
| 3984 |
+
"grad_norm": 0.5358110070228577,
|
| 3985 |
+
"learning_rate": 0.00014359561977876102,
|
| 3986 |
+
"loss": 0.9288,
|
| 3987 |
+
"step": 566
|
| 3988 |
+
},
|
| 3989 |
+
{
|
| 3990 |
+
"epoch": 0.25661914460285135,
|
| 3991 |
+
"grad_norm": 0.4997360110282898,
|
| 3992 |
+
"learning_rate": 0.00014341429223651092,
|
| 3993 |
+
"loss": 0.9648,
|
| 3994 |
+
"step": 567
|
| 3995 |
+
},
|
| 3996 |
+
{
|
| 3997 |
+
"epoch": 0.25707173568680697,
|
| 3998 |
+
"grad_norm": 0.4709185063838959,
|
| 3999 |
+
"learning_rate": 0.00014323278862516775,
|
| 4000 |
+
"loss": 0.8978,
|
| 4001 |
+
"step": 568
|
| 4002 |
+
},
|
| 4003 |
+
{
|
| 4004 |
+
"epoch": 0.2575243267707626,
|
| 4005 |
+
"grad_norm": 0.6130642890930176,
|
| 4006 |
+
"learning_rate": 0.00014305110968082952,
|
| 4007 |
+
"loss": 1.2076,
|
| 4008 |
+
"step": 569
|
| 4009 |
+
},
|
| 4010 |
+
{
|
| 4011 |
+
"epoch": 0.25797691785471827,
|
| 4012 |
+
"grad_norm": 0.5134180188179016,
|
| 4013 |
+
"learning_rate": 0.00014286925614030542,
|
| 4014 |
+
"loss": 0.8747,
|
| 4015 |
+
"step": 570
|
| 4016 |
+
},
|
| 4017 |
+
{
|
| 4018 |
+
"epoch": 0.2584295089386739,
|
| 4019 |
+
"grad_norm": 0.5592136383056641,
|
| 4020 |
+
"learning_rate": 0.00014268722874111265,
|
| 4021 |
+
"loss": 1.0155,
|
| 4022 |
+
"step": 571
|
| 4023 |
+
},
|
| 4024 |
+
{
|
| 4025 |
+
"epoch": 0.25888210002262957,
|
| 4026 |
+
"grad_norm": 0.41666361689567566,
|
| 4027 |
+
"learning_rate": 0.00014250502822147354,
|
| 4028 |
+
"loss": 0.7715,
|
| 4029 |
+
"step": 572
|
| 4030 |
+
},
|
| 4031 |
+
{
|
| 4032 |
+
"epoch": 0.2593346911065852,
|
| 4033 |
+
"grad_norm": 0.45632311701774597,
|
| 4034 |
+
"learning_rate": 0.00014232265532031243,
|
| 4035 |
+
"loss": 0.7487,
|
| 4036 |
+
"step": 573
|
| 4037 |
+
},
|
| 4038 |
+
{
|
| 4039 |
+
"epoch": 0.25978728219054087,
|
| 4040 |
+
"grad_norm": 0.5512887239456177,
|
| 4041 |
+
"learning_rate": 0.00014214011077725292,
|
| 4042 |
+
"loss": 0.9464,
|
| 4043 |
+
"step": 574
|
| 4044 |
+
},
|
| 4045 |
+
{
|
| 4046 |
+
"epoch": 0.2602398732744965,
|
| 4047 |
+
"grad_norm": 0.47320300340652466,
|
| 4048 |
+
"learning_rate": 0.00014195739533261458,
|
| 4049 |
+
"loss": 0.9381,
|
| 4050 |
+
"step": 575
|
| 4051 |
+
},
|
| 4052 |
+
{
|
| 4053 |
+
"epoch": 0.2606924643584521,
|
| 4054 |
+
"grad_norm": 0.5143305659294128,
|
| 4055 |
+
"learning_rate": 0.0001417745097274101,
|
| 4056 |
+
"loss": 0.7783,
|
| 4057 |
+
"step": 576
|
| 4058 |
+
},
|
| 4059 |
+
{
|
| 4060 |
+
"epoch": 0.2611450554424078,
|
| 4061 |
+
"grad_norm": 0.4237568974494934,
|
| 4062 |
+
"learning_rate": 0.00014159145470334235,
|
| 4063 |
+
"loss": 0.5311,
|
| 4064 |
+
"step": 577
|
| 4065 |
+
},
|
| 4066 |
+
{
|
| 4067 |
+
"epoch": 0.2615976465263634,
|
| 4068 |
+
"grad_norm": 0.5218071937561035,
|
| 4069 |
+
"learning_rate": 0.0001414082310028012,
|
| 4070 |
+
"loss": 0.9782,
|
| 4071 |
+
"step": 578
|
| 4072 |
+
},
|
| 4073 |
+
{
|
| 4074 |
+
"epoch": 0.2620502376103191,
|
| 4075 |
+
"grad_norm": 0.411541223526001,
|
| 4076 |
+
"learning_rate": 0.00014122483936886067,
|
| 4077 |
+
"loss": 0.6654,
|
| 4078 |
+
"step": 579
|
| 4079 |
+
},
|
| 4080 |
+
{
|
| 4081 |
+
"epoch": 0.2625028286942747,
|
| 4082 |
+
"grad_norm": 0.508083701133728,
|
| 4083 |
+
"learning_rate": 0.0001410412805452757,
|
| 4084 |
+
"loss": 1.0775,
|
| 4085 |
+
"step": 580
|
| 4086 |
+
},
|
| 4087 |
+
{
|
| 4088 |
+
"epoch": 0.2629554197782304,
|
| 4089 |
+
"grad_norm": 0.43752411007881165,
|
| 4090 |
+
"learning_rate": 0.00014085755527647946,
|
| 4091 |
+
"loss": 0.6292,
|
| 4092 |
+
"step": 581
|
| 4093 |
+
},
|
| 4094 |
+
{
|
| 4095 |
+
"epoch": 0.263408010862186,
|
| 4096 |
+
"grad_norm": 0.591759443283081,
|
| 4097 |
+
"learning_rate": 0.00014067366430758004,
|
| 4098 |
+
"loss": 1.0097,
|
| 4099 |
+
"step": 582
|
| 4100 |
+
},
|
| 4101 |
+
{
|
| 4102 |
+
"epoch": 0.26386060194614164,
|
| 4103 |
+
"grad_norm": 0.411795973777771,
|
| 4104 |
+
"learning_rate": 0.00014048960838435753,
|
| 4105 |
+
"loss": 0.5143,
|
| 4106 |
+
"step": 583
|
| 4107 |
+
},
|
| 4108 |
+
{
|
| 4109 |
+
"epoch": 0.2643131930300973,
|
| 4110 |
+
"grad_norm": 0.4778439402580261,
|
| 4111 |
+
"learning_rate": 0.00014030538825326113,
|
| 4112 |
+
"loss": 0.6781,
|
| 4113 |
+
"step": 584
|
| 4114 |
+
},
|
| 4115 |
+
{
|
| 4116 |
+
"epoch": 0.26476578411405294,
|
| 4117 |
+
"grad_norm": 0.5669144988059998,
|
| 4118 |
+
"learning_rate": 0.00014012100466140578,
|
| 4119 |
+
"loss": 0.9513,
|
| 4120 |
+
"step": 585
|
| 4121 |
+
},
|
| 4122 |
+
{
|
| 4123 |
+
"epoch": 0.2652183751980086,
|
| 4124 |
+
"grad_norm": 0.5837613344192505,
|
| 4125 |
+
"learning_rate": 0.00013993645835656953,
|
| 4126 |
+
"loss": 1.0309,
|
| 4127 |
+
"step": 586
|
| 4128 |
+
},
|
| 4129 |
+
{
|
| 4130 |
+
"epoch": 0.26567096628196424,
|
| 4131 |
+
"grad_norm": 0.42293286323547363,
|
| 4132 |
+
"learning_rate": 0.00013975175008719027,
|
| 4133 |
+
"loss": 0.6944,
|
| 4134 |
+
"step": 587
|
| 4135 |
+
},
|
| 4136 |
+
{
|
| 4137 |
+
"epoch": 0.2661235573659199,
|
| 4138 |
+
"grad_norm": 0.5234153866767883,
|
| 4139 |
+
"learning_rate": 0.00013956688060236266,
|
| 4140 |
+
"loss": 1.1786,
|
| 4141 |
+
"step": 588
|
| 4142 |
+
},
|
| 4143 |
+
{
|
| 4144 |
+
"epoch": 0.26657614844987554,
|
| 4145 |
+
"grad_norm": 0.4235227108001709,
|
| 4146 |
+
"learning_rate": 0.00013938185065183532,
|
| 4147 |
+
"loss": 0.7063,
|
| 4148 |
+
"step": 589
|
| 4149 |
+
},
|
| 4150 |
+
{
|
| 4151 |
+
"epoch": 0.26702873953383116,
|
| 4152 |
+
"grad_norm": 0.5143097043037415,
|
| 4153 |
+
"learning_rate": 0.00013919666098600753,
|
| 4154 |
+
"loss": 0.8122,
|
| 4155 |
+
"step": 590
|
| 4156 |
+
},
|
| 4157 |
+
{
|
| 4158 |
+
"epoch": 0.26748133061778684,
|
| 4159 |
+
"grad_norm": 0.5661123991012573,
|
| 4160 |
+
"learning_rate": 0.00013901131235592634,
|
| 4161 |
+
"loss": 0.9807,
|
| 4162 |
+
"step": 591
|
| 4163 |
+
},
|
| 4164 |
+
{
|
| 4165 |
+
"epoch": 0.26793392170174246,
|
| 4166 |
+
"grad_norm": 0.47588860988616943,
|
| 4167 |
+
"learning_rate": 0.0001388258055132835,
|
| 4168 |
+
"loss": 0.7905,
|
| 4169 |
+
"step": 592
|
| 4170 |
+
},
|
| 4171 |
+
{
|
| 4172 |
+
"epoch": 0.26838651278569814,
|
| 4173 |
+
"grad_norm": 0.43809354305267334,
|
| 4174 |
+
"learning_rate": 0.0001386401412104124,
|
| 4175 |
+
"loss": 0.6649,
|
| 4176 |
+
"step": 593
|
| 4177 |
+
},
|
| 4178 |
+
{
|
| 4179 |
+
"epoch": 0.26883910386965376,
|
| 4180 |
+
"grad_norm": 0.4953427314758301,
|
| 4181 |
+
"learning_rate": 0.0001384543202002851,
|
| 4182 |
+
"loss": 0.9997,
|
| 4183 |
+
"step": 594
|
| 4184 |
+
},
|
| 4185 |
+
{
|
| 4186 |
+
"epoch": 0.26929169495360944,
|
| 4187 |
+
"grad_norm": 0.5393937826156616,
|
| 4188 |
+
"learning_rate": 0.000138268343236509,
|
| 4189 |
+
"loss": 1.0959,
|
| 4190 |
+
"step": 595
|
| 4191 |
+
},
|
| 4192 |
+
{
|
| 4193 |
+
"epoch": 0.26974428603756506,
|
| 4194 |
+
"grad_norm": 0.4869440495967865,
|
| 4195 |
+
"learning_rate": 0.0001380822110733241,
|
| 4196 |
+
"loss": 0.7566,
|
| 4197 |
+
"step": 596
|
| 4198 |
+
},
|
| 4199 |
+
{
|
| 4200 |
+
"epoch": 0.2701968771215207,
|
| 4201 |
+
"grad_norm": 0.5863755345344543,
|
| 4202 |
+
"learning_rate": 0.00013789592446559988,
|
| 4203 |
+
"loss": 1.073,
|
| 4204 |
+
"step": 597
|
| 4205 |
+
},
|
| 4206 |
+
{
|
| 4207 |
+
"epoch": 0.27064946820547636,
|
| 4208 |
+
"grad_norm": 0.46900475025177,
|
| 4209 |
+
"learning_rate": 0.00013770948416883205,
|
| 4210 |
+
"loss": 0.8824,
|
| 4211 |
+
"step": 598
|
| 4212 |
+
},
|
| 4213 |
+
{
|
| 4214 |
+
"epoch": 0.271102059289432,
|
| 4215 |
+
"grad_norm": 0.5337939262390137,
|
| 4216 |
+
"learning_rate": 0.0001375228909391397,
|
| 4217 |
+
"loss": 1.1455,
|
| 4218 |
+
"step": 599
|
| 4219 |
+
},
|
| 4220 |
+
{
|
| 4221 |
+
"epoch": 0.27155465037338766,
|
| 4222 |
+
"grad_norm": 0.6933276653289795,
|
| 4223 |
+
"learning_rate": 0.00013733614553326212,
|
| 4224 |
+
"loss": 0.9099,
|
| 4225 |
+
"step": 600
|
| 4226 |
+
},
|
| 4227 |
+
{
|
| 4228 |
+
"epoch": 0.2720072414573433,
|
| 4229 |
+
"grad_norm": 0.41781508922576904,
|
| 4230 |
+
"learning_rate": 0.00013714924870855571,
|
| 4231 |
+
"loss": 0.569,
|
| 4232 |
+
"step": 601
|
| 4233 |
+
},
|
| 4234 |
+
{
|
| 4235 |
+
"epoch": 0.27245983254129896,
|
| 4236 |
+
"grad_norm": 0.47960004210472107,
|
| 4237 |
+
"learning_rate": 0.00013696220122299112,
|
| 4238 |
+
"loss": 0.9258,
|
| 4239 |
+
"step": 602
|
| 4240 |
+
},
|
| 4241 |
+
{
|
| 4242 |
+
"epoch": 0.2729124236252546,
|
| 4243 |
+
"grad_norm": 0.4569396674633026,
|
| 4244 |
+
"learning_rate": 0.00013677500383514976,
|
| 4245 |
+
"loss": 0.7346,
|
| 4246 |
+
"step": 603
|
| 4247 |
+
},
|
| 4248 |
+
{
|
| 4249 |
+
"epoch": 0.2733650147092102,
|
| 4250 |
+
"grad_norm": 0.48138466477394104,
|
| 4251 |
+
"learning_rate": 0.00013658765730422125,
|
| 4252 |
+
"loss": 0.865,
|
| 4253 |
+
"step": 604
|
| 4254 |
+
},
|
| 4255 |
+
{
|
| 4256 |
+
"epoch": 0.2738176057931659,
|
| 4257 |
+
"grad_norm": 0.5245850682258606,
|
| 4258 |
+
"learning_rate": 0.00013640016238999984,
|
| 4259 |
+
"loss": 0.9676,
|
| 4260 |
+
"step": 605
|
| 4261 |
+
},
|
| 4262 |
+
{
|
| 4263 |
+
"epoch": 0.2742701968771215,
|
| 4264 |
+
"grad_norm": 0.6183745861053467,
|
| 4265 |
+
"learning_rate": 0.0001362125198528817,
|
| 4266 |
+
"loss": 1.0605,
|
| 4267 |
+
"step": 606
|
| 4268 |
+
},
|
| 4269 |
+
{
|
| 4270 |
+
"epoch": 0.2747227879610772,
|
| 4271 |
+
"grad_norm": 0.4640631079673767,
|
| 4272 |
+
"learning_rate": 0.00013602473045386165,
|
| 4273 |
+
"loss": 0.9091,
|
| 4274 |
+
"step": 607
|
| 4275 |
+
},
|
| 4276 |
+
{
|
| 4277 |
+
"epoch": 0.2751753790450328,
|
| 4278 |
+
"grad_norm": 0.4996570646762848,
|
| 4279 |
+
"learning_rate": 0.00013583679495453,
|
| 4280 |
+
"loss": 0.8421,
|
| 4281 |
+
"step": 608
|
| 4282 |
+
},
|
| 4283 |
+
{
|
| 4284 |
+
"epoch": 0.2756279701289885,
|
| 4285 |
+
"grad_norm": 0.5220595002174377,
|
| 4286 |
+
"learning_rate": 0.0001356487141170699,
|
| 4287 |
+
"loss": 0.5935,
|
| 4288 |
+
"step": 609
|
| 4289 |
+
},
|
| 4290 |
+
{
|
| 4291 |
+
"epoch": 0.2760805612129441,
|
| 4292 |
+
"grad_norm": 0.46406427025794983,
|
| 4293 |
+
"learning_rate": 0.00013546048870425356,
|
| 4294 |
+
"loss": 0.9066,
|
| 4295 |
+
"step": 610
|
| 4296 |
+
},
|
| 4297 |
+
{
|
| 4298 |
+
"epoch": 0.27653315229689973,
|
| 4299 |
+
"grad_norm": 0.4911267161369324,
|
| 4300 |
+
"learning_rate": 0.00013527211947943974,
|
| 4301 |
+
"loss": 0.8333,
|
| 4302 |
+
"step": 611
|
| 4303 |
+
},
|
| 4304 |
+
{
|
| 4305 |
+
"epoch": 0.2769857433808554,
|
| 4306 |
+
"grad_norm": 0.516521692276001,
|
| 4307 |
+
"learning_rate": 0.00013508360720657038,
|
| 4308 |
+
"loss": 0.808,
|
| 4309 |
+
"step": 612
|
| 4310 |
+
},
|
| 4311 |
+
{
|
| 4312 |
+
"epoch": 0.27743833446481103,
|
| 4313 |
+
"grad_norm": 0.5458274483680725,
|
| 4314 |
+
"learning_rate": 0.0001348949526501675,
|
| 4315 |
+
"loss": 1.0222,
|
| 4316 |
+
"step": 613
|
| 4317 |
+
},
|
| 4318 |
+
{
|
| 4319 |
+
"epoch": 0.2778909255487667,
|
| 4320 |
+
"grad_norm": 0.54526686668396,
|
| 4321 |
+
"learning_rate": 0.0001347061565753303,
|
| 4322 |
+
"loss": 0.7527,
|
| 4323 |
+
"step": 614
|
| 4324 |
+
},
|
| 4325 |
+
{
|
| 4326 |
+
"epoch": 0.27834351663272233,
|
| 4327 |
+
"grad_norm": 0.4712691307067871,
|
| 4328 |
+
"learning_rate": 0.0001345172197477318,
|
| 4329 |
+
"loss": 0.7782,
|
| 4330 |
+
"step": 615
|
| 4331 |
+
},
|
| 4332 |
+
{
|
| 4333 |
+
"epoch": 0.27879610771667795,
|
| 4334 |
+
"grad_norm": 0.44933852553367615,
|
| 4335 |
+
"learning_rate": 0.00013432814293361584,
|
| 4336 |
+
"loss": 0.7569,
|
| 4337 |
+
"step": 616
|
| 4338 |
+
},
|
| 4339 |
+
{
|
| 4340 |
+
"epoch": 0.27924869880063363,
|
| 4341 |
+
"grad_norm": 0.5856960415840149,
|
| 4342 |
+
"learning_rate": 0.00013413892689979407,
|
| 4343 |
+
"loss": 1.2136,
|
| 4344 |
+
"step": 617
|
| 4345 |
+
},
|
| 4346 |
+
{
|
| 4347 |
+
"epoch": 0.27970128988458925,
|
| 4348 |
+
"grad_norm": 0.5791141390800476,
|
| 4349 |
+
"learning_rate": 0.00013394957241364273,
|
| 4350 |
+
"loss": 1.1147,
|
| 4351 |
+
"step": 618
|
| 4352 |
+
},
|
| 4353 |
+
{
|
| 4354 |
+
"epoch": 0.28015388096854493,
|
| 4355 |
+
"grad_norm": 0.5399816036224365,
|
| 4356 |
+
"learning_rate": 0.00013376008024309948,
|
| 4357 |
+
"loss": 0.9661,
|
| 4358 |
+
"step": 619
|
| 4359 |
+
},
|
| 4360 |
+
{
|
| 4361 |
+
"epoch": 0.28060647205250056,
|
| 4362 |
+
"grad_norm": 0.4976983666419983,
|
| 4363 |
+
"learning_rate": 0.0001335704511566605,
|
| 4364 |
+
"loss": 0.8953,
|
| 4365 |
+
"step": 620
|
| 4366 |
+
},
|
| 4367 |
+
{
|
| 4368 |
+
"epoch": 0.28105906313645623,
|
| 4369 |
+
"grad_norm": 0.5449846386909485,
|
| 4370 |
+
"learning_rate": 0.0001333806859233771,
|
| 4371 |
+
"loss": 0.9998,
|
| 4372 |
+
"step": 621
|
| 4373 |
+
},
|
| 4374 |
+
{
|
| 4375 |
+
"epoch": 0.28151165422041186,
|
| 4376 |
+
"grad_norm": 0.4902433156967163,
|
| 4377 |
+
"learning_rate": 0.00013319078531285285,
|
| 4378 |
+
"loss": 0.8005,
|
| 4379 |
+
"step": 622
|
| 4380 |
+
},
|
| 4381 |
+
{
|
| 4382 |
+
"epoch": 0.2819642453043675,
|
| 4383 |
+
"grad_norm": 0.5565148591995239,
|
| 4384 |
+
"learning_rate": 0.0001330007500952403,
|
| 4385 |
+
"loss": 0.8355,
|
| 4386 |
+
"step": 623
|
| 4387 |
+
},
|
| 4388 |
+
{
|
| 4389 |
+
"epoch": 0.28241683638832316,
|
| 4390 |
+
"grad_norm": 0.5662787556648254,
|
| 4391 |
+
"learning_rate": 0.00013281058104123793,
|
| 4392 |
+
"loss": 0.9444,
|
| 4393 |
+
"step": 624
|
| 4394 |
+
},
|
| 4395 |
+
{
|
| 4396 |
+
"epoch": 0.2828694274722788,
|
| 4397 |
+
"grad_norm": 0.42661720514297485,
|
| 4398 |
+
"learning_rate": 0.00013262027892208694,
|
| 4399 |
+
"loss": 0.6726,
|
| 4400 |
+
"step": 625
|
| 4401 |
+
},
|
| 4402 |
+
{
|
| 4403 |
+
"epoch": 0.28332201855623446,
|
| 4404 |
+
"grad_norm": 1.0109385251998901,
|
| 4405 |
+
"learning_rate": 0.00013242984450956828,
|
| 4406 |
+
"loss": 0.882,
|
| 4407 |
+
"step": 626
|
| 4408 |
+
},
|
| 4409 |
+
{
|
| 4410 |
+
"epoch": 0.2837746096401901,
|
| 4411 |
+
"grad_norm": 0.4351411759853363,
|
| 4412 |
+
"learning_rate": 0.00013223927857599935,
|
| 4413 |
+
"loss": 0.7719,
|
| 4414 |
+
"step": 627
|
| 4415 |
+
},
|
| 4416 |
+
{
|
| 4417 |
+
"epoch": 0.28422720072414576,
|
| 4418 |
+
"grad_norm": 0.5186506509780884,
|
| 4419 |
+
"learning_rate": 0.00013204858189423097,
|
| 4420 |
+
"loss": 0.7782,
|
| 4421 |
+
"step": 628
|
| 4422 |
+
},
|
| 4423 |
+
{
|
| 4424 |
+
"epoch": 0.2846797918081014,
|
| 4425 |
+
"grad_norm": 0.7090187668800354,
|
| 4426 |
+
"learning_rate": 0.00013185775523764424,
|
| 4427 |
+
"loss": 0.9125,
|
| 4428 |
+
"step": 629
|
| 4429 |
+
},
|
| 4430 |
+
{
|
| 4431 |
+
"epoch": 0.285132382892057,
|
| 4432 |
+
"grad_norm": 0.5328108072280884,
|
| 4433 |
+
"learning_rate": 0.00013166679938014726,
|
| 4434 |
+
"loss": 0.9438,
|
| 4435 |
+
"step": 630
|
| 4436 |
+
},
|
| 4437 |
+
{
|
| 4438 |
+
"epoch": 0.2855849739760127,
|
| 4439 |
+
"grad_norm": 0.4744925796985626,
|
| 4440 |
+
"learning_rate": 0.00013147571509617228,
|
| 4441 |
+
"loss": 0.8791,
|
| 4442 |
+
"step": 631
|
| 4443 |
+
},
|
| 4444 |
+
{
|
| 4445 |
+
"epoch": 0.2860375650599683,
|
| 4446 |
+
"grad_norm": 0.5229085683822632,
|
| 4447 |
+
"learning_rate": 0.00013128450316067232,
|
| 4448 |
+
"loss": 0.881,
|
| 4449 |
+
"step": 632
|
| 4450 |
+
},
|
| 4451 |
+
{
|
| 4452 |
+
"epoch": 0.286490156143924,
|
| 4453 |
+
"grad_norm": 0.5308219790458679,
|
| 4454 |
+
"learning_rate": 0.00013109316434911804,
|
| 4455 |
+
"loss": 0.9303,
|
| 4456 |
+
"step": 633
|
| 4457 |
+
},
|
| 4458 |
+
{
|
| 4459 |
+
"epoch": 0.2869427472278796,
|
| 4460 |
+
"grad_norm": 0.5029545426368713,
|
| 4461 |
+
"learning_rate": 0.00013090169943749476,
|
| 4462 |
+
"loss": 0.9326,
|
| 4463 |
+
"step": 634
|
| 4464 |
+
},
|
| 4465 |
+
{
|
| 4466 |
+
"epoch": 0.2873953383118353,
|
| 4467 |
+
"grad_norm": 0.5018476843833923,
|
| 4468 |
+
"learning_rate": 0.00013071010920229909,
|
| 4469 |
+
"loss": 0.8697,
|
| 4470 |
+
"step": 635
|
| 4471 |
+
},
|
| 4472 |
+
{
|
| 4473 |
+
"epoch": 0.2878479293957909,
|
| 4474 |
+
"grad_norm": 0.5224341154098511,
|
| 4475 |
+
"learning_rate": 0.00013051839442053599,
|
| 4476 |
+
"loss": 1.0156,
|
| 4477 |
+
"step": 636
|
| 4478 |
+
},
|
| 4479 |
+
{
|
| 4480 |
+
"epoch": 0.2883005204797465,
|
| 4481 |
+
"grad_norm": 0.3911929726600647,
|
| 4482 |
+
"learning_rate": 0.00013032655586971552,
|
| 4483 |
+
"loss": 0.6028,
|
| 4484 |
+
"step": 637
|
| 4485 |
+
},
|
| 4486 |
+
{
|
| 4487 |
+
"epoch": 0.2887531115637022,
|
| 4488 |
+
"grad_norm": 0.6230449676513672,
|
| 4489 |
+
"learning_rate": 0.00013013459432784961,
|
| 4490 |
+
"loss": 1.3624,
|
| 4491 |
+
"step": 638
|
| 4492 |
+
},
|
| 4493 |
+
{
|
| 4494 |
+
"epoch": 0.2892057026476578,
|
| 4495 |
+
"grad_norm": 0.482842355966568,
|
| 4496 |
+
"learning_rate": 0.00012994251057344905,
|
| 4497 |
+
"loss": 0.8734,
|
| 4498 |
+
"step": 639
|
| 4499 |
+
},
|
| 4500 |
+
{
|
| 4501 |
+
"epoch": 0.2896582937316135,
|
| 4502 |
+
"grad_norm": 0.48907703161239624,
|
| 4503 |
+
"learning_rate": 0.00012975030538552032,
|
| 4504 |
+
"loss": 0.8193,
|
| 4505 |
+
"step": 640
|
| 4506 |
+
},
|
| 4507 |
+
{
|
| 4508 |
+
"epoch": 0.2901108848155691,
|
| 4509 |
+
"grad_norm": 0.5242980718612671,
|
| 4510 |
+
"learning_rate": 0.00012955797954356224,
|
| 4511 |
+
"loss": 0.8471,
|
| 4512 |
+
"step": 641
|
| 4513 |
+
},
|
| 4514 |
+
{
|
| 4515 |
+
"epoch": 0.2905634758995248,
|
| 4516 |
+
"grad_norm": 0.4441068172454834,
|
| 4517 |
+
"learning_rate": 0.0001293655338275631,
|
| 4518 |
+
"loss": 0.693,
|
| 4519 |
+
"step": 642
|
| 4520 |
+
},
|
| 4521 |
+
{
|
| 4522 |
+
"epoch": 0.2910160669834804,
|
| 4523 |
+
"grad_norm": 0.4508345127105713,
|
| 4524 |
+
"learning_rate": 0.0001291729690179972,
|
| 4525 |
+
"loss": 0.6663,
|
| 4526 |
+
"step": 643
|
| 4527 |
+
},
|
| 4528 |
+
{
|
| 4529 |
+
"epoch": 0.29146865806743605,
|
| 4530 |
+
"grad_norm": 0.4511308968067169,
|
| 4531 |
+
"learning_rate": 0.00012898028589582202,
|
| 4532 |
+
"loss": 0.8183,
|
| 4533 |
+
"step": 644
|
| 4534 |
+
},
|
| 4535 |
+
{
|
| 4536 |
+
"epoch": 0.2919212491513917,
|
| 4537 |
+
"grad_norm": 0.5768519639968872,
|
| 4538 |
+
"learning_rate": 0.00012878748524247462,
|
| 4539 |
+
"loss": 0.7828,
|
| 4540 |
+
"step": 645
|
| 4541 |
+
},
|
| 4542 |
+
{
|
| 4543 |
+
"epoch": 0.29237384023534735,
|
| 4544 |
+
"grad_norm": 0.4653044044971466,
|
| 4545 |
+
"learning_rate": 0.00012859456783986893,
|
| 4546 |
+
"loss": 0.6586,
|
| 4547 |
+
"step": 646
|
| 4548 |
+
},
|
| 4549 |
+
{
|
| 4550 |
+
"epoch": 0.292826431319303,
|
| 4551 |
+
"grad_norm": 0.5396013855934143,
|
| 4552 |
+
"learning_rate": 0.00012840153447039228,
|
| 4553 |
+
"loss": 0.8975,
|
| 4554 |
+
"step": 647
|
| 4555 |
+
},
|
| 4556 |
+
{
|
| 4557 |
+
"epoch": 0.29327902240325865,
|
| 4558 |
+
"grad_norm": 0.4867834746837616,
|
| 4559 |
+
"learning_rate": 0.00012820838591690227,
|
| 4560 |
+
"loss": 0.8606,
|
| 4561 |
+
"step": 648
|
| 4562 |
+
},
|
| 4563 |
+
{
|
| 4564 |
+
"epoch": 0.2937316134872143,
|
| 4565 |
+
"grad_norm": 0.4207919239997864,
|
| 4566 |
+
"learning_rate": 0.00012801512296272368,
|
| 4567 |
+
"loss": 0.6362,
|
| 4568 |
+
"step": 649
|
| 4569 |
+
},
|
| 4570 |
+
{
|
| 4571 |
+
"epoch": 0.29418420457116995,
|
| 4572 |
+
"grad_norm": 0.6996460556983948,
|
| 4573 |
+
"learning_rate": 0.0001278217463916453,
|
| 4574 |
+
"loss": 0.7404,
|
| 4575 |
+
"step": 650
|
| 4576 |
+
},
|
| 4577 |
+
{
|
| 4578 |
+
"epoch": 0.29463679565512557,
|
| 4579 |
+
"grad_norm": 0.5210298299789429,
|
| 4580 |
+
"learning_rate": 0.00012762825698791653,
|
| 4581 |
+
"loss": 0.7207,
|
| 4582 |
+
"step": 651
|
| 4583 |
+
},
|
| 4584 |
+
{
|
| 4585 |
+
"epoch": 0.29508938673908125,
|
| 4586 |
+
"grad_norm": 0.440009742975235,
|
| 4587 |
+
"learning_rate": 0.0001274346555362446,
|
| 4588 |
+
"loss": 0.6941,
|
| 4589 |
+
"step": 652
|
| 4590 |
+
},
|
| 4591 |
+
{
|
| 4592 |
+
"epoch": 0.29554197782303687,
|
| 4593 |
+
"grad_norm": 0.5195272564888,
|
| 4594 |
+
"learning_rate": 0.00012724094282179094,
|
| 4595 |
+
"loss": 0.7714,
|
| 4596 |
+
"step": 653
|
| 4597 |
+
},
|
| 4598 |
+
{
|
| 4599 |
+
"epoch": 0.29599456890699255,
|
| 4600 |
+
"grad_norm": 0.5177645087242126,
|
| 4601 |
+
"learning_rate": 0.0001270471196301684,
|
| 4602 |
+
"loss": 0.8669,
|
| 4603 |
+
"step": 654
|
| 4604 |
+
},
|
| 4605 |
+
{
|
| 4606 |
+
"epoch": 0.2964471599909482,
|
| 4607 |
+
"grad_norm": 0.46505850553512573,
|
| 4608 |
+
"learning_rate": 0.0001268531867474377,
|
| 4609 |
+
"loss": 0.914,
|
| 4610 |
+
"step": 655
|
| 4611 |
+
},
|
| 4612 |
+
{
|
| 4613 |
+
"epoch": 0.29689975107490385,
|
| 4614 |
+
"grad_norm": 0.578664243221283,
|
| 4615 |
+
"learning_rate": 0.00012665914496010454,
|
| 4616 |
+
"loss": 1.2505,
|
| 4617 |
+
"step": 656
|
| 4618 |
+
},
|
| 4619 |
+
{
|
| 4620 |
+
"epoch": 0.2973523421588595,
|
| 4621 |
+
"grad_norm": 0.4832736551761627,
|
| 4622 |
+
"learning_rate": 0.00012646499505511625,
|
| 4623 |
+
"loss": 0.8235,
|
| 4624 |
+
"step": 657
|
| 4625 |
+
},
|
| 4626 |
+
{
|
| 4627 |
+
"epoch": 0.2978049332428151,
|
| 4628 |
+
"grad_norm": 0.5135257244110107,
|
| 4629 |
+
"learning_rate": 0.0001262707378198587,
|
| 4630 |
+
"loss": 0.8036,
|
| 4631 |
+
"step": 658
|
| 4632 |
+
},
|
| 4633 |
+
{
|
| 4634 |
+
"epoch": 0.2982575243267708,
|
| 4635 |
+
"grad_norm": 0.4872123599052429,
|
| 4636 |
+
"learning_rate": 0.0001260763740421529,
|
| 4637 |
+
"loss": 0.8361,
|
| 4638 |
+
"step": 659
|
| 4639 |
+
},
|
| 4640 |
+
{
|
| 4641 |
+
"epoch": 0.2987101154107264,
|
| 4642 |
+
"grad_norm": 0.5487096905708313,
|
| 4643 |
+
"learning_rate": 0.00012588190451025207,
|
| 4644 |
+
"loss": 1.1053,
|
| 4645 |
+
"step": 660
|
| 4646 |
+
},
|
| 4647 |
+
{
|
| 4648 |
+
"epoch": 0.2991627064946821,
|
| 4649 |
+
"grad_norm": 0.5271202921867371,
|
| 4650 |
+
"learning_rate": 0.00012568733001283827,
|
| 4651 |
+
"loss": 0.775,
|
| 4652 |
+
"step": 661
|
| 4653 |
+
},
|
| 4654 |
+
{
|
| 4655 |
+
"epoch": 0.2996152975786377,
|
| 4656 |
+
"grad_norm": 0.49594709277153015,
|
| 4657 |
+
"learning_rate": 0.00012549265133901934,
|
| 4658 |
+
"loss": 0.747,
|
| 4659 |
+
"step": 662
|
| 4660 |
+
},
|
| 4661 |
+
{
|
| 4662 |
+
"epoch": 0.3000678886625934,
|
| 4663 |
+
"grad_norm": 0.5582691431045532,
|
| 4664 |
+
"learning_rate": 0.00012529786927832542,
|
| 4665 |
+
"loss": 0.8814,
|
| 4666 |
+
"step": 663
|
| 4667 |
+
},
|
| 4668 |
+
{
|
| 4669 |
+
"epoch": 0.300520479746549,
|
| 4670 |
+
"grad_norm": 0.5153692960739136,
|
| 4671 |
+
"learning_rate": 0.00012510298462070619,
|
| 4672 |
+
"loss": 0.9651,
|
| 4673 |
+
"step": 664
|
| 4674 |
+
},
|
| 4675 |
+
{
|
| 4676 |
+
"epoch": 0.3009730708305046,
|
| 4677 |
+
"grad_norm": 0.5268899202346802,
|
| 4678 |
+
"learning_rate": 0.0001249079981565272,
|
| 4679 |
+
"loss": 0.6635,
|
| 4680 |
+
"step": 665
|
| 4681 |
+
},
|
| 4682 |
+
{
|
| 4683 |
+
"epoch": 0.3014256619144603,
|
| 4684 |
+
"grad_norm": 0.5442325472831726,
|
| 4685 |
+
"learning_rate": 0.00012471291067656697,
|
| 4686 |
+
"loss": 0.9908,
|
| 4687 |
+
"step": 666
|
| 4688 |
+
},
|
| 4689 |
+
{
|
| 4690 |
+
"epoch": 0.3018782529984159,
|
| 4691 |
+
"grad_norm": 0.4943901300430298,
|
| 4692 |
+
"learning_rate": 0.00012451772297201376,
|
| 4693 |
+
"loss": 0.7379,
|
| 4694 |
+
"step": 667
|
| 4695 |
+
},
|
| 4696 |
+
{
|
| 4697 |
+
"epoch": 0.3023308440823716,
|
| 4698 |
+
"grad_norm": 0.49773284792900085,
|
| 4699 |
+
"learning_rate": 0.0001243224358344622,
|
| 4700 |
+
"loss": 0.968,
|
| 4701 |
+
"step": 668
|
| 4702 |
+
},
|
| 4703 |
+
{
|
| 4704 |
+
"epoch": 0.3027834351663272,
|
| 4705 |
+
"grad_norm": 0.4361281991004944,
|
| 4706 |
+
"learning_rate": 0.00012412705005591024,
|
| 4707 |
+
"loss": 0.8096,
|
| 4708 |
+
"step": 669
|
| 4709 |
+
},
|
| 4710 |
+
{
|
| 4711 |
+
"epoch": 0.30323602625028284,
|
| 4712 |
+
"grad_norm": 0.47975584864616394,
|
| 4713 |
+
"learning_rate": 0.0001239315664287558,
|
| 4714 |
+
"loss": 0.7846,
|
| 4715 |
+
"step": 670
|
| 4716 |
+
},
|
| 4717 |
+
{
|
| 4718 |
+
"epoch": 0.3036886173342385,
|
| 4719 |
+
"grad_norm": 0.5244975090026855,
|
| 4720 |
+
"learning_rate": 0.00012373598574579367,
|
| 4721 |
+
"loss": 0.9548,
|
| 4722 |
+
"step": 671
|
| 4723 |
+
},
|
| 4724 |
+
{
|
| 4725 |
+
"epoch": 0.30414120841819414,
|
| 4726 |
+
"grad_norm": 0.5144295692443848,
|
| 4727 |
+
"learning_rate": 0.00012354030880021234,
|
| 4728 |
+
"loss": 0.7269,
|
| 4729 |
+
"step": 672
|
| 4730 |
+
},
|
| 4731 |
+
{
|
| 4732 |
+
"epoch": 0.3045937995021498,
|
| 4733 |
+
"grad_norm": 0.5258602499961853,
|
| 4734 |
+
"learning_rate": 0.00012334453638559057,
|
| 4735 |
+
"loss": 0.9435,
|
| 4736 |
+
"step": 673
|
| 4737 |
+
},
|
| 4738 |
+
{
|
| 4739 |
+
"epoch": 0.30504639058610544,
|
| 4740 |
+
"grad_norm": 0.5024265050888062,
|
| 4741 |
+
"learning_rate": 0.00012314866929589432,
|
| 4742 |
+
"loss": 0.8495,
|
| 4743 |
+
"step": 674
|
| 4744 |
+
},
|
| 4745 |
+
{
|
| 4746 |
+
"epoch": 0.3054989816700611,
|
| 4747 |
+
"grad_norm": 0.4053838849067688,
|
| 4748 |
+
"learning_rate": 0.00012295270832547356,
|
| 4749 |
+
"loss": 0.4879,
|
| 4750 |
+
"step": 675
|
| 4751 |
+
},
|
| 4752 |
+
{
|
| 4753 |
+
"epoch": 0.30595157275401674,
|
| 4754 |
+
"grad_norm": 0.5322363972663879,
|
| 4755 |
+
"learning_rate": 0.000122756654269059,
|
| 4756 |
+
"loss": 0.7895,
|
| 4757 |
+
"step": 676
|
| 4758 |
+
},
|
| 4759 |
+
{
|
| 4760 |
+
"epoch": 0.30640416383797237,
|
| 4761 |
+
"grad_norm": 0.5888680815696716,
|
| 4762 |
+
"learning_rate": 0.00012256050792175887,
|
| 4763 |
+
"loss": 1.1324,
|
| 4764 |
+
"step": 677
|
| 4765 |
+
},
|
| 4766 |
+
{
|
| 4767 |
+
"epoch": 0.30685675492192804,
|
| 4768 |
+
"grad_norm": 0.5343974232673645,
|
| 4769 |
+
"learning_rate": 0.00012236427007905558,
|
| 4770 |
+
"loss": 0.9867,
|
| 4771 |
+
"step": 678
|
| 4772 |
+
},
|
| 4773 |
+
{
|
| 4774 |
+
"epoch": 0.30730934600588367,
|
| 4775 |
+
"grad_norm": 0.4725770950317383,
|
| 4776 |
+
"learning_rate": 0.00012216794153680274,
|
| 4777 |
+
"loss": 0.8123,
|
| 4778 |
+
"step": 679
|
| 4779 |
+
},
|
| 4780 |
+
{
|
| 4781 |
+
"epoch": 0.30776193708983934,
|
| 4782 |
+
"grad_norm": 0.4887194335460663,
|
| 4783 |
+
"learning_rate": 0.00012197152309122173,
|
| 4784 |
+
"loss": 0.9044,
|
| 4785 |
+
"step": 680
|
| 4786 |
+
},
|
| 4787 |
+
{
|
| 4788 |
+
"epoch": 0.30821452817379497,
|
| 4789 |
+
"grad_norm": 0.4865034818649292,
|
| 4790 |
+
"learning_rate": 0.00012177501553889856,
|
| 4791 |
+
"loss": 0.7338,
|
| 4792 |
+
"step": 681
|
| 4793 |
+
},
|
| 4794 |
+
{
|
| 4795 |
+
"epoch": 0.30866711925775064,
|
| 4796 |
+
"grad_norm": 0.472418874502182,
|
| 4797 |
+
"learning_rate": 0.00012157841967678063,
|
| 4798 |
+
"loss": 0.7923,
|
| 4799 |
+
"step": 682
|
| 4800 |
+
},
|
| 4801 |
+
{
|
| 4802 |
+
"epoch": 0.30911971034170627,
|
| 4803 |
+
"grad_norm": 0.48899978399276733,
|
| 4804 |
+
"learning_rate": 0.00012138173630217342,
|
| 4805 |
+
"loss": 0.8421,
|
| 4806 |
+
"step": 683
|
| 4807 |
+
},
|
| 4808 |
+
{
|
| 4809 |
+
"epoch": 0.3095723014256619,
|
| 4810 |
+
"grad_norm": 0.4492488503456116,
|
| 4811 |
+
"learning_rate": 0.00012118496621273745,
|
| 4812 |
+
"loss": 0.6645,
|
| 4813 |
+
"step": 684
|
| 4814 |
+
},
|
| 4815 |
+
{
|
| 4816 |
+
"epoch": 0.31002489250961757,
|
| 4817 |
+
"grad_norm": 0.40700769424438477,
|
| 4818 |
+
"learning_rate": 0.00012098811020648475,
|
| 4819 |
+
"loss": 0.5403,
|
| 4820 |
+
"step": 685
|
| 4821 |
+
},
|
| 4822 |
+
{
|
| 4823 |
+
"epoch": 0.3104774835935732,
|
| 4824 |
+
"grad_norm": 0.543043315410614,
|
| 4825 |
+
"learning_rate": 0.00012079116908177593,
|
| 4826 |
+
"loss": 0.9634,
|
| 4827 |
+
"step": 686
|
| 4828 |
+
},
|
| 4829 |
+
{
|
| 4830 |
+
"epoch": 0.31093007467752887,
|
| 4831 |
+
"grad_norm": 0.5078130960464478,
|
| 4832 |
+
"learning_rate": 0.00012059414363731674,
|
| 4833 |
+
"loss": 0.9197,
|
| 4834 |
+
"step": 687
|
| 4835 |
+
},
|
| 4836 |
+
{
|
| 4837 |
+
"epoch": 0.3113826657614845,
|
| 4838 |
+
"grad_norm": 0.47711634635925293,
|
| 4839 |
+
"learning_rate": 0.00012039703467215488,
|
| 4840 |
+
"loss": 0.8304,
|
| 4841 |
+
"step": 688
|
| 4842 |
+
},
|
| 4843 |
+
{
|
| 4844 |
+
"epoch": 0.31183525684544017,
|
| 4845 |
+
"grad_norm": 0.4520992338657379,
|
| 4846 |
+
"learning_rate": 0.00012019984298567682,
|
| 4847 |
+
"loss": 0.5376,
|
| 4848 |
+
"step": 689
|
| 4849 |
+
},
|
| 4850 |
+
{
|
| 4851 |
+
"epoch": 0.3122878479293958,
|
| 4852 |
+
"grad_norm": 0.49447259306907654,
|
| 4853 |
+
"learning_rate": 0.00012000256937760445,
|
| 4854 |
+
"loss": 0.8525,
|
| 4855 |
+
"step": 690
|
| 4856 |
+
},
|
| 4857 |
+
{
|
| 4858 |
+
"epoch": 0.3127404390133514,
|
| 4859 |
+
"grad_norm": 0.5532546639442444,
|
| 4860 |
+
"learning_rate": 0.00011980521464799198,
|
| 4861 |
+
"loss": 1.0082,
|
| 4862 |
+
"step": 691
|
| 4863 |
+
},
|
| 4864 |
+
{
|
| 4865 |
+
"epoch": 0.3131930300973071,
|
| 4866 |
+
"grad_norm": 0.4508833885192871,
|
| 4867 |
+
"learning_rate": 0.00011960777959722253,
|
| 4868 |
+
"loss": 0.8646,
|
| 4869 |
+
"step": 692
|
| 4870 |
+
},
|
| 4871 |
+
{
|
| 4872 |
+
"epoch": 0.3136456211812627,
|
| 4873 |
+
"grad_norm": 0.4003430902957916,
|
| 4874 |
+
"learning_rate": 0.000119410265026005,
|
| 4875 |
+
"loss": 0.6842,
|
| 4876 |
+
"step": 693
|
| 4877 |
+
},
|
| 4878 |
+
{
|
| 4879 |
+
"epoch": 0.3140982122652184,
|
| 4880 |
+
"grad_norm": 0.5402839183807373,
|
| 4881 |
+
"learning_rate": 0.00011921267173537086,
|
| 4882 |
+
"loss": 0.982,
|
| 4883 |
+
"step": 694
|
| 4884 |
+
},
|
| 4885 |
+
{
|
| 4886 |
+
"epoch": 0.314550803349174,
|
| 4887 |
+
"grad_norm": 0.44808655977249146,
|
| 4888 |
+
"learning_rate": 0.00011901500052667068,
|
| 4889 |
+
"loss": 0.8136,
|
| 4890 |
+
"step": 695
|
| 4891 |
+
},
|
| 4892 |
+
{
|
| 4893 |
+
"epoch": 0.3150033944331297,
|
| 4894 |
+
"grad_norm": 0.5470143556594849,
|
| 4895 |
+
"learning_rate": 0.00011881725220157113,
|
| 4896 |
+
"loss": 1.2509,
|
| 4897 |
+
"step": 696
|
| 4898 |
+
},
|
| 4899 |
+
{
|
| 4900 |
+
"epoch": 0.3154559855170853,
|
| 4901 |
+
"grad_norm": 0.5380640625953674,
|
| 4902 |
+
"learning_rate": 0.00011861942756205169,
|
| 4903 |
+
"loss": 1.0705,
|
| 4904 |
+
"step": 697
|
| 4905 |
+
},
|
| 4906 |
+
{
|
| 4907 |
+
"epoch": 0.31590857660104094,
|
| 4908 |
+
"grad_norm": 0.49166131019592285,
|
| 4909 |
+
"learning_rate": 0.00011842152741040116,
|
| 4910 |
+
"loss": 0.7588,
|
| 4911 |
+
"step": 698
|
| 4912 |
+
},
|
| 4913 |
+
{
|
| 4914 |
+
"epoch": 0.3163611676849966,
|
| 4915 |
+
"grad_norm": 0.5186418294906616,
|
| 4916 |
+
"learning_rate": 0.00011822355254921478,
|
| 4917 |
+
"loss": 0.9667,
|
| 4918 |
+
"step": 699
|
| 4919 |
+
},
|
| 4920 |
+
{
|
| 4921 |
+
"epoch": 0.31681375876895224,
|
| 4922 |
+
"grad_norm": 0.48516249656677246,
|
| 4923 |
+
"learning_rate": 0.0001180255037813906,
|
| 4924 |
+
"loss": 0.8118,
|
| 4925 |
+
"step": 700
|
| 4926 |
+
},
|
| 4927 |
+
{
|
| 4928 |
+
"epoch": 0.3172663498529079,
|
| 4929 |
+
"grad_norm": 0.4609658420085907,
|
| 4930 |
+
"learning_rate": 0.00011782738191012652,
|
| 4931 |
+
"loss": 0.9096,
|
| 4932 |
+
"step": 701
|
| 4933 |
+
},
|
| 4934 |
+
{
|
| 4935 |
+
"epoch": 0.31771894093686354,
|
| 4936 |
+
"grad_norm": 0.4883543848991394,
|
| 4937 |
+
"learning_rate": 0.00011762918773891691,
|
| 4938 |
+
"loss": 0.8148,
|
| 4939 |
+
"step": 702
|
| 4940 |
+
},
|
| 4941 |
+
{
|
| 4942 |
+
"epoch": 0.3181715320208192,
|
| 4943 |
+
"grad_norm": 0.5130921006202698,
|
| 4944 |
+
"learning_rate": 0.00011743092207154929,
|
| 4945 |
+
"loss": 0.8609,
|
| 4946 |
+
"step": 703
|
| 4947 |
+
},
|
| 4948 |
+
{
|
| 4949 |
+
"epoch": 0.31862412310477484,
|
| 4950 |
+
"grad_norm": 0.5097033381462097,
|
| 4951 |
+
"learning_rate": 0.00011723258571210121,
|
| 4952 |
+
"loss": 1.0203,
|
| 4953 |
+
"step": 704
|
| 4954 |
+
},
|
| 4955 |
+
{
|
| 4956 |
+
"epoch": 0.31907671418873046,
|
| 4957 |
+
"grad_norm": 0.4869794249534607,
|
| 4958 |
+
"learning_rate": 0.00011703417946493683,
|
| 4959 |
+
"loss": 1.0473,
|
| 4960 |
+
"step": 705
|
| 4961 |
+
},
|
| 4962 |
+
{
|
| 4963 |
+
"epoch": 0.31952930527268614,
|
| 4964 |
+
"grad_norm": 0.45631757378578186,
|
| 4965 |
+
"learning_rate": 0.00011683570413470383,
|
| 4966 |
+
"loss": 0.706,
|
| 4967 |
+
"step": 706
|
| 4968 |
+
},
|
| 4969 |
+
{
|
| 4970 |
+
"epoch": 0.31998189635664176,
|
| 4971 |
+
"grad_norm": 0.47609448432922363,
|
| 4972 |
+
"learning_rate": 0.00011663716052633006,
|
| 4973 |
+
"loss": 0.6871,
|
| 4974 |
+
"step": 707
|
| 4975 |
+
},
|
| 4976 |
+
{
|
| 4977 |
+
"epoch": 0.32043448744059744,
|
| 4978 |
+
"grad_norm": 0.4803597629070282,
|
| 4979 |
+
"learning_rate": 0.0001164385494450202,
|
| 4980 |
+
"loss": 0.7175,
|
| 4981 |
+
"step": 708
|
| 4982 |
+
},
|
| 4983 |
+
{
|
| 4984 |
+
"epoch": 0.32088707852455306,
|
| 4985 |
+
"grad_norm": 0.42784056067466736,
|
| 4986 |
+
"learning_rate": 0.00011623987169625261,
|
| 4987 |
+
"loss": 0.7301,
|
| 4988 |
+
"step": 709
|
| 4989 |
+
},
|
| 4990 |
+
{
|
| 4991 |
+
"epoch": 0.32133966960850874,
|
| 4992 |
+
"grad_norm": 0.5752036571502686,
|
| 4993 |
+
"learning_rate": 0.00011604112808577603,
|
| 4994 |
+
"loss": 0.8522,
|
| 4995 |
+
"step": 710
|
| 4996 |
+
},
|
| 4997 |
+
{
|
| 4998 |
+
"epoch": 0.32179226069246436,
|
| 4999 |
+
"grad_norm": 0.5516340732574463,
|
| 5000 |
+
"learning_rate": 0.00011584231941960628,
|
| 5001 |
+
"loss": 0.9913,
|
| 5002 |
+
"step": 711
|
| 5003 |
+
},
|
| 5004 |
+
{
|
| 5005 |
+
"epoch": 0.32224485177642,
|
| 5006 |
+
"grad_norm": 0.4553099572658539,
|
| 5007 |
+
"learning_rate": 0.0001156434465040231,
|
| 5008 |
+
"loss": 0.7671,
|
| 5009 |
+
"step": 712
|
| 5010 |
+
},
|
| 5011 |
+
{
|
| 5012 |
+
"epoch": 0.32269744286037566,
|
| 5013 |
+
"grad_norm": 0.5408231019973755,
|
| 5014 |
+
"learning_rate": 0.00011544451014556665,
|
| 5015 |
+
"loss": 0.9212,
|
| 5016 |
+
"step": 713
|
| 5017 |
+
},
|
| 5018 |
+
{
|
| 5019 |
+
"epoch": 0.3231500339443313,
|
| 5020 |
+
"grad_norm": 0.5186082720756531,
|
| 5021 |
+
"learning_rate": 0.00011524551115103454,
|
| 5022 |
+
"loss": 0.8433,
|
| 5023 |
+
"step": 714
|
| 5024 |
+
},
|
| 5025 |
+
{
|
| 5026 |
+
"epoch": 0.32360262502828696,
|
| 5027 |
+
"grad_norm": 0.464286744594574,
|
| 5028 |
+
"learning_rate": 0.00011504645032747832,
|
| 5029 |
+
"loss": 0.8244,
|
| 5030 |
+
"step": 715
|
| 5031 |
+
},
|
| 5032 |
+
{
|
| 5033 |
+
"epoch": 0.3240552161122426,
|
| 5034 |
+
"grad_norm": 0.5196670293807983,
|
| 5035 |
+
"learning_rate": 0.00011484732848220026,
|
| 5036 |
+
"loss": 0.9368,
|
| 5037 |
+
"step": 716
|
| 5038 |
+
},
|
| 5039 |
+
{
|
| 5040 |
+
"epoch": 0.32450780719619826,
|
| 5041 |
+
"grad_norm": 0.4835793375968933,
|
| 5042 |
+
"learning_rate": 0.00011464814642275022,
|
| 5043 |
+
"loss": 0.7648,
|
| 5044 |
+
"step": 717
|
| 5045 |
+
},
|
| 5046 |
+
{
|
| 5047 |
+
"epoch": 0.3249603982801539,
|
| 5048 |
+
"grad_norm": 0.48903408646583557,
|
| 5049 |
+
"learning_rate": 0.00011444890495692213,
|
| 5050 |
+
"loss": 0.9417,
|
| 5051 |
+
"step": 718
|
| 5052 |
+
},
|
| 5053 |
+
{
|
| 5054 |
+
"epoch": 0.3254129893641095,
|
| 5055 |
+
"grad_norm": 0.5032278895378113,
|
| 5056 |
+
"learning_rate": 0.00011424960489275098,
|
| 5057 |
+
"loss": 0.8376,
|
| 5058 |
+
"step": 719
|
| 5059 |
+
},
|
| 5060 |
+
{
|
| 5061 |
+
"epoch": 0.3258655804480652,
|
| 5062 |
+
"grad_norm": 0.44192057847976685,
|
| 5063 |
+
"learning_rate": 0.00011405024703850929,
|
| 5064 |
+
"loss": 0.6973,
|
| 5065 |
+
"step": 720
|
| 5066 |
+
},
|
| 5067 |
+
{
|
| 5068 |
+
"epoch": 0.3263181715320208,
|
| 5069 |
+
"grad_norm": 0.448816180229187,
|
| 5070 |
+
"learning_rate": 0.00011385083220270401,
|
| 5071 |
+
"loss": 0.7599,
|
| 5072 |
+
"step": 721
|
| 5073 |
+
},
|
| 5074 |
+
{
|
| 5075 |
+
"epoch": 0.3267707626159765,
|
| 5076 |
+
"grad_norm": 0.44175952672958374,
|
| 5077 |
+
"learning_rate": 0.00011365136119407319,
|
| 5078 |
+
"loss": 0.6627,
|
| 5079 |
+
"step": 722
|
| 5080 |
+
},
|
| 5081 |
+
{
|
| 5082 |
+
"epoch": 0.3272233536999321,
|
| 5083 |
+
"grad_norm": 0.5739919543266296,
|
| 5084 |
+
"learning_rate": 0.00011345183482158264,
|
| 5085 |
+
"loss": 0.9778,
|
| 5086 |
+
"step": 723
|
| 5087 |
+
},
|
| 5088 |
+
{
|
| 5089 |
+
"epoch": 0.3276759447838878,
|
| 5090 |
+
"grad_norm": 0.5183276534080505,
|
| 5091 |
+
"learning_rate": 0.00011325225389442277,
|
| 5092 |
+
"loss": 0.9271,
|
| 5093 |
+
"step": 724
|
| 5094 |
+
},
|
| 5095 |
+
{
|
| 5096 |
+
"epoch": 0.3281285358678434,
|
| 5097 |
+
"grad_norm": 0.40638917684555054,
|
| 5098 |
+
"learning_rate": 0.00011305261922200519,
|
| 5099 |
+
"loss": 0.5595,
|
| 5100 |
+
"step": 725
|
| 5101 |
+
},
|
| 5102 |
+
{
|
| 5103 |
+
"epoch": 0.32858112695179903,
|
| 5104 |
+
"grad_norm": 0.5148621797561646,
|
| 5105 |
+
"learning_rate": 0.00011285293161395946,
|
| 5106 |
+
"loss": 0.9125,
|
| 5107 |
+
"step": 726
|
| 5108 |
+
},
|
| 5109 |
+
{
|
| 5110 |
+
"epoch": 0.3290337180357547,
|
| 5111 |
+
"grad_norm": 0.49269428849220276,
|
| 5112 |
+
"learning_rate": 0.00011265319188012994,
|
| 5113 |
+
"loss": 0.8463,
|
| 5114 |
+
"step": 727
|
| 5115 |
+
},
|
| 5116 |
+
{
|
| 5117 |
+
"epoch": 0.32948630911971033,
|
| 5118 |
+
"grad_norm": 0.5185051560401917,
|
| 5119 |
+
"learning_rate": 0.00011245340083057223,
|
| 5120 |
+
"loss": 0.8708,
|
| 5121 |
+
"step": 728
|
| 5122 |
+
},
|
| 5123 |
+
{
|
| 5124 |
+
"epoch": 0.329938900203666,
|
| 5125 |
+
"grad_norm": 0.48547667264938354,
|
| 5126 |
+
"learning_rate": 0.00011225355927555014,
|
| 5127 |
+
"loss": 0.9409,
|
| 5128 |
+
"step": 729
|
| 5129 |
+
},
|
| 5130 |
+
{
|
| 5131 |
+
"epoch": 0.33039149128762163,
|
| 5132 |
+
"grad_norm": 0.581288754940033,
|
| 5133 |
+
"learning_rate": 0.0001120536680255323,
|
| 5134 |
+
"loss": 1.1318,
|
| 5135 |
+
"step": 730
|
| 5136 |
+
},
|
| 5137 |
+
{
|
| 5138 |
+
"epoch": 0.33084408237157725,
|
| 5139 |
+
"grad_norm": 0.5337197184562683,
|
| 5140 |
+
"learning_rate": 0.00011185372789118887,
|
| 5141 |
+
"loss": 0.9725,
|
| 5142 |
+
"step": 731
|
| 5143 |
+
},
|
| 5144 |
+
{
|
| 5145 |
+
"epoch": 0.33129667345553293,
|
| 5146 |
+
"grad_norm": 0.49437853693962097,
|
| 5147 |
+
"learning_rate": 0.00011165373968338824,
|
| 5148 |
+
"loss": 0.9509,
|
| 5149 |
+
"step": 732
|
| 5150 |
+
},
|
| 5151 |
+
{
|
| 5152 |
+
"epoch": 0.33174926453948855,
|
| 5153 |
+
"grad_norm": 0.577504575252533,
|
| 5154 |
+
"learning_rate": 0.00011145370421319377,
|
| 5155 |
+
"loss": 1.1211,
|
| 5156 |
+
"step": 733
|
| 5157 |
+
},
|
| 5158 |
+
{
|
| 5159 |
+
"epoch": 0.33220185562344423,
|
| 5160 |
+
"grad_norm": 0.6033246517181396,
|
| 5161 |
+
"learning_rate": 0.00011125362229186057,
|
| 5162 |
+
"loss": 1.1196,
|
| 5163 |
+
"step": 734
|
| 5164 |
+
},
|
| 5165 |
+
{
|
| 5166 |
+
"epoch": 0.33265444670739985,
|
| 5167 |
+
"grad_norm": 0.46174025535583496,
|
| 5168 |
+
"learning_rate": 0.000111053494730832,
|
| 5169 |
+
"loss": 0.762,
|
| 5170 |
+
"step": 735
|
| 5171 |
+
},
|
| 5172 |
+
{
|
| 5173 |
+
"epoch": 0.33310703779135553,
|
| 5174 |
+
"grad_norm": 0.5123987793922424,
|
| 5175 |
+
"learning_rate": 0.00011085332234173664,
|
| 5176 |
+
"loss": 0.9462,
|
| 5177 |
+
"step": 736
|
| 5178 |
+
},
|
| 5179 |
+
{
|
| 5180 |
+
"epoch": 0.33355962887531115,
|
| 5181 |
+
"grad_norm": 0.4520655572414398,
|
| 5182 |
+
"learning_rate": 0.00011065310593638483,
|
| 5183 |
+
"loss": 0.8018,
|
| 5184 |
+
"step": 737
|
| 5185 |
+
},
|
| 5186 |
+
{
|
| 5187 |
+
"epoch": 0.3340122199592668,
|
| 5188 |
+
"grad_norm": 0.5619468092918396,
|
| 5189 |
+
"learning_rate": 0.00011045284632676536,
|
| 5190 |
+
"loss": 1.0392,
|
| 5191 |
+
"step": 738
|
| 5192 |
+
},
|
| 5193 |
+
{
|
| 5194 |
+
"epoch": 0.33446481104322245,
|
| 5195 |
+
"grad_norm": 0.49872851371765137,
|
| 5196 |
+
"learning_rate": 0.00011025254432504233,
|
| 5197 |
+
"loss": 0.7856,
|
| 5198 |
+
"step": 739
|
| 5199 |
+
},
|
| 5200 |
+
{
|
| 5201 |
+
"epoch": 0.3349174021271781,
|
| 5202 |
+
"grad_norm": 0.40664106607437134,
|
| 5203 |
+
"learning_rate": 0.00011005220074355171,
|
| 5204 |
+
"loss": 0.5486,
|
| 5205 |
+
"step": 740
|
| 5206 |
+
},
|
| 5207 |
+
{
|
| 5208 |
+
"epoch": 0.33536999321113375,
|
| 5209 |
+
"grad_norm": 0.513265073299408,
|
| 5210 |
+
"learning_rate": 0.00010985181639479813,
|
| 5211 |
+
"loss": 0.8292,
|
| 5212 |
+
"step": 741
|
| 5213 |
+
},
|
| 5214 |
+
{
|
| 5215 |
+
"epoch": 0.3358225842950894,
|
| 5216 |
+
"grad_norm": 0.3986378014087677,
|
| 5217 |
+
"learning_rate": 0.00010965139209145152,
|
| 5218 |
+
"loss": 0.7516,
|
| 5219 |
+
"step": 742
|
| 5220 |
+
},
|
| 5221 |
+
{
|
| 5222 |
+
"epoch": 0.33627517537904505,
|
| 5223 |
+
"grad_norm": 0.4895348846912384,
|
| 5224 |
+
"learning_rate": 0.00010945092864634388,
|
| 5225 |
+
"loss": 0.8381,
|
| 5226 |
+
"step": 743
|
| 5227 |
+
},
|
| 5228 |
+
{
|
| 5229 |
+
"epoch": 0.3367277664630007,
|
| 5230 |
+
"grad_norm": 0.44972074031829834,
|
| 5231 |
+
"learning_rate": 0.00010925042687246592,
|
| 5232 |
+
"loss": 0.6788,
|
| 5233 |
+
"step": 744
|
| 5234 |
+
},
|
| 5235 |
+
{
|
| 5236 |
+
"epoch": 0.3371803575469563,
|
| 5237 |
+
"grad_norm": 0.5335928201675415,
|
| 5238 |
+
"learning_rate": 0.0001090498875829638,
|
| 5239 |
+
"loss": 0.7777,
|
| 5240 |
+
"step": 745
|
| 5241 |
+
},
|
| 5242 |
+
{
|
| 5243 |
+
"epoch": 0.337632948630912,
|
| 5244 |
+
"grad_norm": 0.5425425171852112,
|
| 5245 |
+
"learning_rate": 0.00010884931159113586,
|
| 5246 |
+
"loss": 0.8378,
|
| 5247 |
+
"step": 746
|
| 5248 |
+
},
|
| 5249 |
+
{
|
| 5250 |
+
"epoch": 0.3380855397148676,
|
| 5251 |
+
"grad_norm": 0.5339006185531616,
|
| 5252 |
+
"learning_rate": 0.00010864869971042925,
|
| 5253 |
+
"loss": 0.8918,
|
| 5254 |
+
"step": 747
|
| 5255 |
+
},
|
| 5256 |
+
{
|
| 5257 |
+
"epoch": 0.3385381307988233,
|
| 5258 |
+
"grad_norm": 0.5580891966819763,
|
| 5259 |
+
"learning_rate": 0.00010844805275443673,
|
| 5260 |
+
"loss": 0.9455,
|
| 5261 |
+
"step": 748
|
| 5262 |
+
},
|
| 5263 |
+
{
|
| 5264 |
+
"epoch": 0.3389907218827789,
|
| 5265 |
+
"grad_norm": 0.4360131025314331,
|
| 5266 |
+
"learning_rate": 0.00010824737153689319,
|
| 5267 |
+
"loss": 0.6515,
|
| 5268 |
+
"step": 749
|
| 5269 |
+
},
|
| 5270 |
+
{
|
| 5271 |
+
"epoch": 0.3394433129667346,
|
| 5272 |
+
"grad_norm": 0.43018674850463867,
|
| 5273 |
+
"learning_rate": 0.00010804665687167262,
|
| 5274 |
+
"loss": 0.6645,
|
| 5275 |
+
"step": 750
|
| 5276 |
+
},
|
| 5277 |
+
{
|
| 5278 |
+
"epoch": 0.3398959040506902,
|
| 5279 |
+
"grad_norm": 0.4255496561527252,
|
| 5280 |
+
"learning_rate": 0.0001078459095727845,
|
| 5281 |
+
"loss": 0.6244,
|
| 5282 |
+
"step": 751
|
| 5283 |
+
},
|
| 5284 |
+
{
|
| 5285 |
+
"epoch": 0.3403484951346458,
|
| 5286 |
+
"grad_norm": 0.4636889398097992,
|
| 5287 |
+
"learning_rate": 0.00010764513045437083,
|
| 5288 |
+
"loss": 0.791,
|
| 5289 |
+
"step": 752
|
| 5290 |
+
},
|
| 5291 |
+
{
|
| 5292 |
+
"epoch": 0.3408010862186015,
|
| 5293 |
+
"grad_norm": 0.5495937466621399,
|
| 5294 |
+
"learning_rate": 0.00010744432033070251,
|
| 5295 |
+
"loss": 0.9623,
|
| 5296 |
+
"step": 753
|
| 5297 |
+
},
|
| 5298 |
+
{
|
| 5299 |
+
"epoch": 0.3412536773025571,
|
| 5300 |
+
"grad_norm": 0.5409247875213623,
|
| 5301 |
+
"learning_rate": 0.00010724348001617625,
|
| 5302 |
+
"loss": 0.925,
|
| 5303 |
+
"step": 754
|
| 5304 |
+
},
|
| 5305 |
+
{
|
| 5306 |
+
"epoch": 0.3417062683865128,
|
| 5307 |
+
"grad_norm": 0.4871814548969269,
|
| 5308 |
+
"learning_rate": 0.0001070426103253112,
|
| 5309 |
+
"loss": 0.7726,
|
| 5310 |
+
"step": 755
|
| 5311 |
+
},
|
| 5312 |
+
{
|
| 5313 |
+
"epoch": 0.3421588594704684,
|
| 5314 |
+
"grad_norm": 0.5325297713279724,
|
| 5315 |
+
"learning_rate": 0.00010684171207274562,
|
| 5316 |
+
"loss": 1.0311,
|
| 5317 |
+
"step": 756
|
| 5318 |
+
},
|
| 5319 |
+
{
|
| 5320 |
+
"epoch": 0.3426114505544241,
|
| 5321 |
+
"grad_norm": 0.6206918358802795,
|
| 5322 |
+
"learning_rate": 0.00010664078607323367,
|
| 5323 |
+
"loss": 1.0326,
|
| 5324 |
+
"step": 757
|
| 5325 |
+
},
|
| 5326 |
+
{
|
| 5327 |
+
"epoch": 0.3430640416383797,
|
| 5328 |
+
"grad_norm": 0.45602741837501526,
|
| 5329 |
+
"learning_rate": 0.00010643983314164194,
|
| 5330 |
+
"loss": 0.6741,
|
| 5331 |
+
"step": 758
|
| 5332 |
+
},
|
| 5333 |
+
{
|
| 5334 |
+
"epoch": 0.34351663272233535,
|
| 5335 |
+
"grad_norm": 0.4545150399208069,
|
| 5336 |
+
"learning_rate": 0.00010623885409294633,
|
| 5337 |
+
"loss": 0.6257,
|
| 5338 |
+
"step": 759
|
| 5339 |
+
},
|
| 5340 |
+
{
|
| 5341 |
+
"epoch": 0.343969223806291,
|
| 5342 |
+
"grad_norm": 0.46721890568733215,
|
| 5343 |
+
"learning_rate": 0.00010603784974222861,
|
| 5344 |
+
"loss": 0.7438,
|
| 5345 |
+
"step": 760
|
| 5346 |
+
},
|
| 5347 |
+
{
|
| 5348 |
+
"epoch": 0.34442181489024665,
|
| 5349 |
+
"grad_norm": 0.44289225339889526,
|
| 5350 |
+
"learning_rate": 0.00010583682090467317,
|
| 5351 |
+
"loss": 0.7158,
|
| 5352 |
+
"step": 761
|
| 5353 |
+
},
|
| 5354 |
+
{
|
| 5355 |
+
"epoch": 0.3448744059742023,
|
| 5356 |
+
"grad_norm": 0.6556697487831116,
|
| 5357 |
+
"learning_rate": 0.00010563576839556374,
|
| 5358 |
+
"loss": 1.1567,
|
| 5359 |
+
"step": 762
|
| 5360 |
+
},
|
| 5361 |
+
{
|
| 5362 |
+
"epoch": 0.34532699705815795,
|
| 5363 |
+
"grad_norm": 0.48243972659111023,
|
| 5364 |
+
"learning_rate": 0.00010543469303028002,
|
| 5365 |
+
"loss": 0.7717,
|
| 5366 |
+
"step": 763
|
| 5367 |
+
},
|
| 5368 |
+
{
|
| 5369 |
+
"epoch": 0.3457795881421136,
|
| 5370 |
+
"grad_norm": 0.4793996214866638,
|
| 5371 |
+
"learning_rate": 0.0001052335956242944,
|
| 5372 |
+
"loss": 0.7768,
|
| 5373 |
+
"step": 764
|
| 5374 |
+
},
|
| 5375 |
+
{
|
| 5376 |
+
"epoch": 0.34623217922606925,
|
| 5377 |
+
"grad_norm": 0.4908459484577179,
|
| 5378 |
+
"learning_rate": 0.00010503247699316867,
|
| 5379 |
+
"loss": 0.856,
|
| 5380 |
+
"step": 765
|
| 5381 |
+
},
|
| 5382 |
+
{
|
| 5383 |
+
"epoch": 0.34668477031002487,
|
| 5384 |
+
"grad_norm": 0.5109564661979675,
|
| 5385 |
+
"learning_rate": 0.00010483133795255071,
|
| 5386 |
+
"loss": 0.9241,
|
| 5387 |
+
"step": 766
|
| 5388 |
+
},
|
| 5389 |
+
{
|
| 5390 |
+
"epoch": 0.34713736139398055,
|
| 5391 |
+
"grad_norm": 0.6431382894515991,
|
| 5392 |
+
"learning_rate": 0.00010463017931817118,
|
| 5393 |
+
"loss": 1.1816,
|
| 5394 |
+
"step": 767
|
| 5395 |
+
},
|
| 5396 |
+
{
|
| 5397 |
+
"epoch": 0.34758995247793617,
|
| 5398 |
+
"grad_norm": 0.40695667266845703,
|
| 5399 |
+
"learning_rate": 0.00010442900190584015,
|
| 5400 |
+
"loss": 0.6323,
|
| 5401 |
+
"step": 768
|
| 5402 |
+
},
|
| 5403 |
+
{
|
| 5404 |
+
"epoch": 0.34804254356189185,
|
| 5405 |
+
"grad_norm": 0.5081280469894409,
|
| 5406 |
+
"learning_rate": 0.0001042278065314439,
|
| 5407 |
+
"loss": 0.9465,
|
| 5408 |
+
"step": 769
|
| 5409 |
+
},
|
| 5410 |
+
{
|
| 5411 |
+
"epoch": 0.34849513464584747,
|
| 5412 |
+
"grad_norm": 0.5007328391075134,
|
| 5413 |
+
"learning_rate": 0.00010402659401094152,
|
| 5414 |
+
"loss": 0.8279,
|
| 5415 |
+
"step": 770
|
| 5416 |
+
},
|
| 5417 |
+
{
|
| 5418 |
+
"epoch": 0.34894772572980315,
|
| 5419 |
+
"grad_norm": 0.5045701861381531,
|
| 5420 |
+
"learning_rate": 0.00010382536516036168,
|
| 5421 |
+
"loss": 0.9728,
|
| 5422 |
+
"step": 771
|
| 5423 |
+
},
|
| 5424 |
+
{
|
| 5425 |
+
"epoch": 0.34940031681375877,
|
| 5426 |
+
"grad_norm": 0.48791611194610596,
|
| 5427 |
+
"learning_rate": 0.00010362412079579924,
|
| 5428 |
+
"loss": 0.8349,
|
| 5429 |
+
"step": 772
|
| 5430 |
+
},
|
| 5431 |
+
{
|
| 5432 |
+
"epoch": 0.3498529078977144,
|
| 5433 |
+
"grad_norm": 0.42378130555152893,
|
| 5434 |
+
"learning_rate": 0.00010342286173341198,
|
| 5435 |
+
"loss": 0.5704,
|
| 5436 |
+
"step": 773
|
| 5437 |
+
},
|
| 5438 |
+
{
|
| 5439 |
+
"epoch": 0.35030549898167007,
|
| 5440 |
+
"grad_norm": 0.4793219566345215,
|
| 5441 |
+
"learning_rate": 0.00010322158878941732,
|
| 5442 |
+
"loss": 0.8859,
|
| 5443 |
+
"step": 774
|
| 5444 |
+
},
|
| 5445 |
+
{
|
| 5446 |
+
"epoch": 0.3507580900656257,
|
| 5447 |
+
"grad_norm": 0.48526403307914734,
|
| 5448 |
+
"learning_rate": 0.0001030203027800889,
|
| 5449 |
+
"loss": 0.7512,
|
| 5450 |
+
"step": 775
|
| 5451 |
+
},
|
| 5452 |
+
{
|
| 5453 |
+
"epoch": 0.35121068114958137,
|
| 5454 |
+
"grad_norm": 0.5508863925933838,
|
| 5455 |
+
"learning_rate": 0.00010281900452175346,
|
| 5456 |
+
"loss": 1.1402,
|
| 5457 |
+
"step": 776
|
| 5458 |
+
},
|
| 5459 |
+
{
|
| 5460 |
+
"epoch": 0.351663272233537,
|
| 5461 |
+
"grad_norm": 0.5259900689125061,
|
| 5462 |
+
"learning_rate": 0.00010261769483078733,
|
| 5463 |
+
"loss": 0.8289,
|
| 5464 |
+
"step": 777
|
| 5465 |
+
},
|
| 5466 |
+
{
|
| 5467 |
+
"epoch": 0.35211586331749267,
|
| 5468 |
+
"grad_norm": 0.511981189250946,
|
| 5469 |
+
"learning_rate": 0.00010241637452361323,
|
| 5470 |
+
"loss": 0.796,
|
| 5471 |
+
"step": 778
|
| 5472 |
+
},
|
| 5473 |
+
{
|
| 5474 |
+
"epoch": 0.3525684544014483,
|
| 5475 |
+
"grad_norm": 0.4819547235965729,
|
| 5476 |
+
"learning_rate": 0.00010221504441669699,
|
| 5477 |
+
"loss": 0.8646,
|
| 5478 |
+
"step": 779
|
| 5479 |
+
},
|
| 5480 |
+
{
|
| 5481 |
+
"epoch": 0.3530210454854039,
|
| 5482 |
+
"grad_norm": 0.4084272086620331,
|
| 5483 |
+
"learning_rate": 0.00010201370532654404,
|
| 5484 |
+
"loss": 0.5851,
|
| 5485 |
+
"step": 780
|
| 5486 |
+
},
|
| 5487 |
+
{
|
| 5488 |
+
"epoch": 0.3534736365693596,
|
| 5489 |
+
"grad_norm": 0.4426387548446655,
|
| 5490 |
+
"learning_rate": 0.0001018123580696964,
|
| 5491 |
+
"loss": 0.7538,
|
| 5492 |
+
"step": 781
|
| 5493 |
+
},
|
| 5494 |
+
{
|
| 5495 |
+
"epoch": 0.3539262276533152,
|
| 5496 |
+
"grad_norm": 0.5691589713096619,
|
| 5497 |
+
"learning_rate": 0.00010161100346272914,
|
| 5498 |
+
"loss": 1.1445,
|
| 5499 |
+
"step": 782
|
| 5500 |
+
},
|
| 5501 |
+
{
|
| 5502 |
+
"epoch": 0.3543788187372709,
|
| 5503 |
+
"grad_norm": 0.5064798593521118,
|
| 5504 |
+
"learning_rate": 0.00010140964232224713,
|
| 5505 |
+
"loss": 0.6435,
|
| 5506 |
+
"step": 783
|
| 5507 |
+
},
|
| 5508 |
+
{
|
| 5509 |
+
"epoch": 0.3548314098212265,
|
| 5510 |
+
"grad_norm": 0.5018054842948914,
|
| 5511 |
+
"learning_rate": 0.00010120827546488174,
|
| 5512 |
+
"loss": 0.9572,
|
| 5513 |
+
"step": 784
|
| 5514 |
+
},
|
| 5515 |
+
{
|
| 5516 |
+
"epoch": 0.35528400090518214,
|
| 5517 |
+
"grad_norm": 0.5410506129264832,
|
| 5518 |
+
"learning_rate": 0.00010100690370728755,
|
| 5519 |
+
"loss": 1.003,
|
| 5520 |
+
"step": 785
|
| 5521 |
+
},
|
| 5522 |
+
{
|
| 5523 |
+
"epoch": 0.3557365919891378,
|
| 5524 |
+
"grad_norm": 0.510235071182251,
|
| 5525 |
+
"learning_rate": 0.00010080552786613899,
|
| 5526 |
+
"loss": 0.8269,
|
| 5527 |
+
"step": 786
|
| 5528 |
+
},
|
| 5529 |
+
{
|
| 5530 |
+
"epoch": 0.3557365919891378,
|
| 5531 |
+
"eval_loss": 0.20480701327323914,
|
| 5532 |
+
"eval_runtime": 51.7143,
|
| 5533 |
+
"eval_samples_per_second": 18.003,
|
| 5534 |
+
"eval_steps_per_second": 9.011,
|
| 5535 |
+
"step": 786
|
| 5536 |
}
|
| 5537 |
],
|
| 5538 |
"logging_steps": 1,
|
|
|
|
| 5552 |
"attributes": {}
|
| 5553 |
}
|
| 5554 |
},
|
| 5555 |
+
"total_flos": 1.4333841993262694e+17,
|
| 5556 |
"train_batch_size": 2,
|
| 5557 |
"trial_name": null,
|
| 5558 |
"trial_params": null
|