Training in progress, step 1330, 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 48679352
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:c1dc2cbbe57c97401a90d24ef883fe3c5e0b1f443286befb1eb3e141c611a7fd
|
3 |
size 48679352
|
last-checkpoint/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 25152884
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:2baa0a73c29389e323435f73ecbbaa841f07ea042c06114f9fa5f92ecdc9f230
|
3 |
size 25152884
|
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:c41965c9ba80b80aa013ab39ca7605c7004ac8b7b3982e9a1f27800d44b27d05
|
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 sha256:4dd83d88b5559b4d0fca903fbe38b787a7fe386065782a09f912476ae97c46fd
|
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": 500,
|
6 |
-
"global_step":
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
@@ -7000,6 +7000,2323 @@
|
|
7000 |
"learning_rate": 1.4623710158072663e-05,
|
7001 |
"loss": 0.772,
|
7002 |
"step": 999
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
7003 |
}
|
7004 |
],
|
7005 |
"logging_steps": 1,
|
@@ -7014,12 +9331,12 @@
|
|
7014 |
"should_evaluate": false,
|
7015 |
"should_log": false,
|
7016 |
"should_save": true,
|
7017 |
-
"should_training_stop":
|
7018 |
},
|
7019 |
"attributes": {}
|
7020 |
}
|
7021 |
},
|
7022 |
-
"total_flos":
|
7023 |
"train_batch_size": 4,
|
7024 |
"trial_name": null,
|
7025 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 0.008840116849063313,
|
5 |
"eval_steps": 500,
|
6 |
+
"global_step": 1330,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
7000 |
"learning_rate": 1.4623710158072663e-05,
|
7001 |
"loss": 0.772,
|
7002 |
"step": 999
|
7003 |
+
},
|
7004 |
+
{
|
7005 |
+
"epoch": 0.006646704397791965,
|
7006 |
+
"grad_norm": 2.8134305477142334,
|
7007 |
+
"learning_rate": 1.4540031398113335e-05,
|
7008 |
+
"loss": 0.7747,
|
7009 |
+
"step": 1000
|
7010 |
+
},
|
7011 |
+
{
|
7012 |
+
"epoch": 0.006653351102189757,
|
7013 |
+
"grad_norm": 0.31304219365119934,
|
7014 |
+
"learning_rate": 1.4456551983514333e-05,
|
7015 |
+
"loss": 0.9119,
|
7016 |
+
"step": 1001
|
7017 |
+
},
|
7018 |
+
{
|
7019 |
+
"epoch": 0.0066599978065875485,
|
7020 |
+
"grad_norm": 0.6188201308250427,
|
7021 |
+
"learning_rate": 1.4373272383572028e-05,
|
7022 |
+
"loss": 0.7513,
|
7023 |
+
"step": 1002
|
7024 |
+
},
|
7025 |
+
{
|
7026 |
+
"epoch": 0.006666644510985341,
|
7027 |
+
"grad_norm": 0.6836460828781128,
|
7028 |
+
"learning_rate": 1.4290193066459457e-05,
|
7029 |
+
"loss": 0.944,
|
7030 |
+
"step": 1003
|
7031 |
+
},
|
7032 |
+
{
|
7033 |
+
"epoch": 0.006673291215383133,
|
7034 |
+
"grad_norm": 0.6879714131355286,
|
7035 |
+
"learning_rate": 1.4207314499223745e-05,
|
7036 |
+
"loss": 0.7964,
|
7037 |
+
"step": 1004
|
7038 |
+
},
|
7039 |
+
{
|
7040 |
+
"epoch": 0.0066799379197809245,
|
7041 |
+
"grad_norm": 0.5281482934951782,
|
7042 |
+
"learning_rate": 1.4124637147783432e-05,
|
7043 |
+
"loss": 0.8331,
|
7044 |
+
"step": 1005
|
7045 |
+
},
|
7046 |
+
{
|
7047 |
+
"epoch": 0.006686584624178716,
|
7048 |
+
"grad_norm": 0.5473004579544067,
|
7049 |
+
"learning_rate": 1.404216147692598e-05,
|
7050 |
+
"loss": 0.9916,
|
7051 |
+
"step": 1006
|
7052 |
+
},
|
7053 |
+
{
|
7054 |
+
"epoch": 0.006693231328576509,
|
7055 |
+
"grad_norm": 0.6183127760887146,
|
7056 |
+
"learning_rate": 1.395988795030495e-05,
|
7057 |
+
"loss": 0.6854,
|
7058 |
+
"step": 1007
|
7059 |
+
},
|
7060 |
+
{
|
7061 |
+
"epoch": 0.006699878032974301,
|
7062 |
+
"grad_norm": 0.5756447911262512,
|
7063 |
+
"learning_rate": 1.3877817030437568e-05,
|
7064 |
+
"loss": 0.7835,
|
7065 |
+
"step": 1008
|
7066 |
+
},
|
7067 |
+
{
|
7068 |
+
"epoch": 0.006706524737372092,
|
7069 |
+
"grad_norm": 0.5753706693649292,
|
7070 |
+
"learning_rate": 1.3795949178702033e-05,
|
7071 |
+
"loss": 1.0322,
|
7072 |
+
"step": 1009
|
7073 |
+
},
|
7074 |
+
{
|
7075 |
+
"epoch": 0.006713171441769884,
|
7076 |
+
"grad_norm": 0.6262528300285339,
|
7077 |
+
"learning_rate": 1.371428485533498e-05,
|
7078 |
+
"loss": 0.7982,
|
7079 |
+
"step": 1010
|
7080 |
+
},
|
7081 |
+
{
|
7082 |
+
"epoch": 0.006719818146167677,
|
7083 |
+
"grad_norm": 0.6056203246116638,
|
7084 |
+
"learning_rate": 1.3632824519428889e-05,
|
7085 |
+
"loss": 0.9525,
|
7086 |
+
"step": 1011
|
7087 |
+
},
|
7088 |
+
{
|
7089 |
+
"epoch": 0.0067264648505654685,
|
7090 |
+
"grad_norm": 0.6044523119926453,
|
7091 |
+
"learning_rate": 1.3551568628929434e-05,
|
7092 |
+
"loss": 0.8489,
|
7093 |
+
"step": 1012
|
7094 |
+
},
|
7095 |
+
{
|
7096 |
+
"epoch": 0.00673311155496326,
|
7097 |
+
"grad_norm": 1.2201472520828247,
|
7098 |
+
"learning_rate": 1.3470517640632969e-05,
|
7099 |
+
"loss": 1.0784,
|
7100 |
+
"step": 1013
|
7101 |
+
},
|
7102 |
+
{
|
7103 |
+
"epoch": 0.006739758259361052,
|
7104 |
+
"grad_norm": 0.5861047506332397,
|
7105 |
+
"learning_rate": 1.3389672010183984e-05,
|
7106 |
+
"loss": 0.8827,
|
7107 |
+
"step": 1014
|
7108 |
+
},
|
7109 |
+
{
|
7110 |
+
"epoch": 0.0067464049637588445,
|
7111 |
+
"grad_norm": 0.8151621222496033,
|
7112 |
+
"learning_rate": 1.3309032192072463e-05,
|
7113 |
+
"loss": 0.9368,
|
7114 |
+
"step": 1015
|
7115 |
+
},
|
7116 |
+
{
|
7117 |
+
"epoch": 0.006753051668156636,
|
7118 |
+
"grad_norm": 0.6700717806816101,
|
7119 |
+
"learning_rate": 1.3228598639631418e-05,
|
7120 |
+
"loss": 0.8357,
|
7121 |
+
"step": 1016
|
7122 |
+
},
|
7123 |
+
{
|
7124 |
+
"epoch": 0.006759698372554428,
|
7125 |
+
"grad_norm": 0.9705390334129333,
|
7126 |
+
"learning_rate": 1.3148371805034253e-05,
|
7127 |
+
"loss": 0.8453,
|
7128 |
+
"step": 1017
|
7129 |
+
},
|
7130 |
+
{
|
7131 |
+
"epoch": 0.00676634507695222,
|
7132 |
+
"grad_norm": 0.7800853252410889,
|
7133 |
+
"learning_rate": 1.3068352139292312e-05,
|
7134 |
+
"loss": 0.728,
|
7135 |
+
"step": 1018
|
7136 |
+
},
|
7137 |
+
{
|
7138 |
+
"epoch": 0.006772991781350012,
|
7139 |
+
"grad_norm": 0.7995185852050781,
|
7140 |
+
"learning_rate": 1.2988540092252266e-05,
|
7141 |
+
"loss": 0.9883,
|
7142 |
+
"step": 1019
|
7143 |
+
},
|
7144 |
+
{
|
7145 |
+
"epoch": 0.006779638485747804,
|
7146 |
+
"grad_norm": 0.7061491012573242,
|
7147 |
+
"learning_rate": 1.29089361125936e-05,
|
7148 |
+
"loss": 0.7797,
|
7149 |
+
"step": 1020
|
7150 |
+
},
|
7151 |
+
{
|
7152 |
+
"epoch": 0.006786285190145596,
|
7153 |
+
"grad_norm": 0.926814079284668,
|
7154 |
+
"learning_rate": 1.2829540647826149e-05,
|
7155 |
+
"loss": 0.7926,
|
7156 |
+
"step": 1021
|
7157 |
+
},
|
7158 |
+
{
|
7159 |
+
"epoch": 0.006792931894543388,
|
7160 |
+
"grad_norm": 0.8915033340454102,
|
7161 |
+
"learning_rate": 1.275035414428753e-05,
|
7162 |
+
"loss": 0.9411,
|
7163 |
+
"step": 1022
|
7164 |
+
},
|
7165 |
+
{
|
7166 |
+
"epoch": 0.00679957859894118,
|
7167 |
+
"grad_norm": 1.2160683870315552,
|
7168 |
+
"learning_rate": 1.2671377047140625e-05,
|
7169 |
+
"loss": 0.8545,
|
7170 |
+
"step": 1023
|
7171 |
+
},
|
7172 |
+
{
|
7173 |
+
"epoch": 0.006806225303338972,
|
7174 |
+
"grad_norm": 0.8952360153198242,
|
7175 |
+
"learning_rate": 1.2592609800371086e-05,
|
7176 |
+
"loss": 0.9481,
|
7177 |
+
"step": 1024
|
7178 |
+
},
|
7179 |
+
{
|
7180 |
+
"epoch": 0.006812872007736764,
|
7181 |
+
"grad_norm": 0.7336769700050354,
|
7182 |
+
"learning_rate": 1.251405284678488e-05,
|
7183 |
+
"loss": 0.7532,
|
7184 |
+
"step": 1025
|
7185 |
+
},
|
7186 |
+
{
|
7187 |
+
"epoch": 0.006819518712134556,
|
7188 |
+
"grad_norm": 0.9183105230331421,
|
7189 |
+
"learning_rate": 1.2435706628005767e-05,
|
7190 |
+
"loss": 0.8165,
|
7191 |
+
"step": 1026
|
7192 |
+
},
|
7193 |
+
{
|
7194 |
+
"epoch": 0.006826165416532348,
|
7195 |
+
"grad_norm": 0.7961068153381348,
|
7196 |
+
"learning_rate": 1.2357571584472794e-05,
|
7197 |
+
"loss": 0.8613,
|
7198 |
+
"step": 1027
|
7199 |
+
},
|
7200 |
+
{
|
7201 |
+
"epoch": 0.00683281212093014,
|
7202 |
+
"grad_norm": 0.920566737651825,
|
7203 |
+
"learning_rate": 1.227964815543784e-05,
|
7204 |
+
"loss": 0.9217,
|
7205 |
+
"step": 1028
|
7206 |
+
},
|
7207 |
+
{
|
7208 |
+
"epoch": 0.0068394588253279315,
|
7209 |
+
"grad_norm": 0.9314897656440735,
|
7210 |
+
"learning_rate": 1.2201936778963192e-05,
|
7211 |
+
"loss": 0.7957,
|
7212 |
+
"step": 1029
|
7213 |
+
},
|
7214 |
+
{
|
7215 |
+
"epoch": 0.006846105529725724,
|
7216 |
+
"grad_norm": 1.002401351928711,
|
7217 |
+
"learning_rate": 1.2124437891918993e-05,
|
7218 |
+
"loss": 0.8704,
|
7219 |
+
"step": 1030
|
7220 |
+
},
|
7221 |
+
{
|
7222 |
+
"epoch": 0.006852752234123516,
|
7223 |
+
"grad_norm": 1.2262808084487915,
|
7224 |
+
"learning_rate": 1.2047151929980888e-05,
|
7225 |
+
"loss": 0.8852,
|
7226 |
+
"step": 1031
|
7227 |
+
},
|
7228 |
+
{
|
7229 |
+
"epoch": 0.006859398938521308,
|
7230 |
+
"grad_norm": 1.0216968059539795,
|
7231 |
+
"learning_rate": 1.1970079327627453e-05,
|
7232 |
+
"loss": 0.7399,
|
7233 |
+
"step": 1032
|
7234 |
+
},
|
7235 |
+
{
|
7236 |
+
"epoch": 0.006866045642919099,
|
7237 |
+
"grad_norm": 1.1232874393463135,
|
7238 |
+
"learning_rate": 1.1893220518137882e-05,
|
7239 |
+
"loss": 0.7159,
|
7240 |
+
"step": 1033
|
7241 |
+
},
|
7242 |
+
{
|
7243 |
+
"epoch": 0.006872692347316892,
|
7244 |
+
"grad_norm": 1.0752450227737427,
|
7245 |
+
"learning_rate": 1.1816575933589458e-05,
|
7246 |
+
"loss": 0.6177,
|
7247 |
+
"step": 1034
|
7248 |
+
},
|
7249 |
+
{
|
7250 |
+
"epoch": 0.006879339051714684,
|
7251 |
+
"grad_norm": 0.911821722984314,
|
7252 |
+
"learning_rate": 1.174014600485514e-05,
|
7253 |
+
"loss": 0.7266,
|
7254 |
+
"step": 1035
|
7255 |
+
},
|
7256 |
+
{
|
7257 |
+
"epoch": 0.006885985756112475,
|
7258 |
+
"grad_norm": 1.2482151985168457,
|
7259 |
+
"learning_rate": 1.1663931161601188e-05,
|
7260 |
+
"loss": 0.8125,
|
7261 |
+
"step": 1036
|
7262 |
+
},
|
7263 |
+
{
|
7264 |
+
"epoch": 0.006892632460510267,
|
7265 |
+
"grad_norm": 1.0357885360717773,
|
7266 |
+
"learning_rate": 1.1587931832284726e-05,
|
7267 |
+
"loss": 0.7536,
|
7268 |
+
"step": 1037
|
7269 |
+
},
|
7270 |
+
{
|
7271 |
+
"epoch": 0.00689927916490806,
|
7272 |
+
"grad_norm": 1.0621405839920044,
|
7273 |
+
"learning_rate": 1.1512148444151283e-05,
|
7274 |
+
"loss": 0.6836,
|
7275 |
+
"step": 1038
|
7276 |
+
},
|
7277 |
+
{
|
7278 |
+
"epoch": 0.0069059258693058515,
|
7279 |
+
"grad_norm": 1.1723105907440186,
|
7280 |
+
"learning_rate": 1.1436581423232434e-05,
|
7281 |
+
"loss": 0.6806,
|
7282 |
+
"step": 1039
|
7283 |
+
},
|
7284 |
+
{
|
7285 |
+
"epoch": 0.006912572573703643,
|
7286 |
+
"grad_norm": 1.1845663785934448,
|
7287 |
+
"learning_rate": 1.1361231194343436e-05,
|
7288 |
+
"loss": 0.815,
|
7289 |
+
"step": 1040
|
7290 |
+
},
|
7291 |
+
{
|
7292 |
+
"epoch": 0.006919219278101435,
|
7293 |
+
"grad_norm": 1.6141482591629028,
|
7294 |
+
"learning_rate": 1.1286098181080795e-05,
|
7295 |
+
"loss": 0.9749,
|
7296 |
+
"step": 1041
|
7297 |
+
},
|
7298 |
+
{
|
7299 |
+
"epoch": 0.006925865982499228,
|
7300 |
+
"grad_norm": 1.34913969039917,
|
7301 |
+
"learning_rate": 1.1211182805819881e-05,
|
7302 |
+
"loss": 0.6698,
|
7303 |
+
"step": 1042
|
7304 |
+
},
|
7305 |
+
{
|
7306 |
+
"epoch": 0.006932512686897019,
|
7307 |
+
"grad_norm": 1.377968668937683,
|
7308 |
+
"learning_rate": 1.1136485489712556e-05,
|
7309 |
+
"loss": 0.5934,
|
7310 |
+
"step": 1043
|
7311 |
+
},
|
7312 |
+
{
|
7313 |
+
"epoch": 0.006939159391294811,
|
7314 |
+
"grad_norm": 1.7394827604293823,
|
7315 |
+
"learning_rate": 1.1062006652684864e-05,
|
7316 |
+
"loss": 0.7956,
|
7317 |
+
"step": 1044
|
7318 |
+
},
|
7319 |
+
{
|
7320 |
+
"epoch": 0.006945806095692603,
|
7321 |
+
"grad_norm": 1.4188005924224854,
|
7322 |
+
"learning_rate": 1.0987746713434576e-05,
|
7323 |
+
"loss": 0.7873,
|
7324 |
+
"step": 1045
|
7325 |
+
},
|
7326 |
+
{
|
7327 |
+
"epoch": 0.0069524528000903954,
|
7328 |
+
"grad_norm": 1.7469269037246704,
|
7329 |
+
"learning_rate": 1.0913706089428932e-05,
|
7330 |
+
"loss": 1.0878,
|
7331 |
+
"step": 1046
|
7332 |
+
},
|
7333 |
+
{
|
7334 |
+
"epoch": 0.006959099504488187,
|
7335 |
+
"grad_norm": 2.151268482208252,
|
7336 |
+
"learning_rate": 1.0839885196902194e-05,
|
7337 |
+
"loss": 1.1574,
|
7338 |
+
"step": 1047
|
7339 |
+
},
|
7340 |
+
{
|
7341 |
+
"epoch": 0.006965746208885979,
|
7342 |
+
"grad_norm": 2.0133984088897705,
|
7343 |
+
"learning_rate": 1.0766284450853415e-05,
|
7344 |
+
"loss": 0.716,
|
7345 |
+
"step": 1048
|
7346 |
+
},
|
7347 |
+
{
|
7348 |
+
"epoch": 0.006972392913283771,
|
7349 |
+
"grad_norm": 3.2061398029327393,
|
7350 |
+
"learning_rate": 1.0692904265044012e-05,
|
7351 |
+
"loss": 0.915,
|
7352 |
+
"step": 1049
|
7353 |
+
},
|
7354 |
+
{
|
7355 |
+
"epoch": 0.006979039617681563,
|
7356 |
+
"grad_norm": 2.7918288707733154,
|
7357 |
+
"learning_rate": 1.0619745051995472e-05,
|
7358 |
+
"loss": 1.219,
|
7359 |
+
"step": 1050
|
7360 |
+
},
|
7361 |
+
{
|
7362 |
+
"epoch": 0.006985686322079355,
|
7363 |
+
"grad_norm": 0.3655790686607361,
|
7364 |
+
"learning_rate": 1.0546807222987071e-05,
|
7365 |
+
"loss": 1.0736,
|
7366 |
+
"step": 1051
|
7367 |
+
},
|
7368 |
+
{
|
7369 |
+
"epoch": 0.006992333026477147,
|
7370 |
+
"grad_norm": 0.4837401509284973,
|
7371 |
+
"learning_rate": 1.047409118805353e-05,
|
7372 |
+
"loss": 0.9356,
|
7373 |
+
"step": 1052
|
7374 |
+
},
|
7375 |
+
{
|
7376 |
+
"epoch": 0.006998979730874939,
|
7377 |
+
"grad_norm": 0.45560112595558167,
|
7378 |
+
"learning_rate": 1.0401597355982678e-05,
|
7379 |
+
"loss": 0.8369,
|
7380 |
+
"step": 1053
|
7381 |
+
},
|
7382 |
+
{
|
7383 |
+
"epoch": 0.007005626435272731,
|
7384 |
+
"grad_norm": 0.5554627180099487,
|
7385 |
+
"learning_rate": 1.03293261343132e-05,
|
7386 |
+
"loss": 0.8466,
|
7387 |
+
"step": 1054
|
7388 |
+
},
|
7389 |
+
{
|
7390 |
+
"epoch": 0.007012273139670523,
|
7391 |
+
"grad_norm": 0.7247781753540039,
|
7392 |
+
"learning_rate": 1.0257277929332332e-05,
|
7393 |
+
"loss": 1.0188,
|
7394 |
+
"step": 1055
|
7395 |
+
},
|
7396 |
+
{
|
7397 |
+
"epoch": 0.007018919844068315,
|
7398 |
+
"grad_norm": 0.5876094102859497,
|
7399 |
+
"learning_rate": 1.0185453146073604e-05,
|
7400 |
+
"loss": 0.8758,
|
7401 |
+
"step": 1056
|
7402 |
+
},
|
7403 |
+
{
|
7404 |
+
"epoch": 0.007025566548466107,
|
7405 |
+
"grad_norm": 0.5293012857437134,
|
7406 |
+
"learning_rate": 1.0113852188314493e-05,
|
7407 |
+
"loss": 0.8288,
|
7408 |
+
"step": 1057
|
7409 |
+
},
|
7410 |
+
{
|
7411 |
+
"epoch": 0.007032213252863899,
|
7412 |
+
"grad_norm": 0.5751774311065674,
|
7413 |
+
"learning_rate": 1.0042475458574202e-05,
|
7414 |
+
"loss": 0.8582,
|
7415 |
+
"step": 1058
|
7416 |
+
},
|
7417 |
+
{
|
7418 |
+
"epoch": 0.007038859957261691,
|
7419 |
+
"grad_norm": 0.7522995471954346,
|
7420 |
+
"learning_rate": 9.971323358111434e-06,
|
7421 |
+
"loss": 0.7952,
|
7422 |
+
"step": 1059
|
7423 |
+
},
|
7424 |
+
{
|
7425 |
+
"epoch": 0.007045506661659482,
|
7426 |
+
"grad_norm": 0.6282753348350525,
|
7427 |
+
"learning_rate": 9.900396286922026e-06,
|
7428 |
+
"loss": 0.7768,
|
7429 |
+
"step": 1060
|
7430 |
+
},
|
7431 |
+
{
|
7432 |
+
"epoch": 0.007052153366057275,
|
7433 |
+
"grad_norm": 0.6144838333129883,
|
7434 |
+
"learning_rate": 9.829694643736836e-06,
|
7435 |
+
"loss": 0.7815,
|
7436 |
+
"step": 1061
|
7437 |
+
},
|
7438 |
+
{
|
7439 |
+
"epoch": 0.007058800070455067,
|
7440 |
+
"grad_norm": 0.6692332029342651,
|
7441 |
+
"learning_rate": 9.759218826019378e-06,
|
7442 |
+
"loss": 0.7859,
|
7443 |
+
"step": 1062
|
7444 |
+
},
|
7445 |
+
{
|
7446 |
+
"epoch": 0.0070654467748528585,
|
7447 |
+
"grad_norm": 0.8164696097373962,
|
7448 |
+
"learning_rate": 9.688969229963685e-06,
|
7449 |
+
"loss": 0.8053,
|
7450 |
+
"step": 1063
|
7451 |
+
},
|
7452 |
+
{
|
7453 |
+
"epoch": 0.00707209347925065,
|
7454 |
+
"grad_norm": 0.7802616953849792,
|
7455 |
+
"learning_rate": 9.618946250492011e-06,
|
7456 |
+
"loss": 0.93,
|
7457 |
+
"step": 1064
|
7458 |
+
},
|
7459 |
+
{
|
7460 |
+
"epoch": 0.007078740183648443,
|
7461 |
+
"grad_norm": 0.7973951697349548,
|
7462 |
+
"learning_rate": 9.549150281252633e-06,
|
7463 |
+
"loss": 0.931,
|
7464 |
+
"step": 1065
|
7465 |
+
},
|
7466 |
+
{
|
7467 |
+
"epoch": 0.007085386888046235,
|
7468 |
+
"grad_norm": 0.6746591925621033,
|
7469 |
+
"learning_rate": 9.479581714617668e-06,
|
7470 |
+
"loss": 0.9751,
|
7471 |
+
"step": 1066
|
7472 |
+
},
|
7473 |
+
{
|
7474 |
+
"epoch": 0.007092033592444026,
|
7475 |
+
"grad_norm": 0.8076299428939819,
|
7476 |
+
"learning_rate": 9.410240941680859e-06,
|
7477 |
+
"loss": 0.7754,
|
7478 |
+
"step": 1067
|
7479 |
+
},
|
7480 |
+
{
|
7481 |
+
"epoch": 0.007098680296841818,
|
7482 |
+
"grad_norm": 2.197793483734131,
|
7483 |
+
"learning_rate": 9.341128352255313e-06,
|
7484 |
+
"loss": 0.8414,
|
7485 |
+
"step": 1068
|
7486 |
+
},
|
7487 |
+
{
|
7488 |
+
"epoch": 0.007105327001239611,
|
7489 |
+
"grad_norm": 0.7308064699172974,
|
7490 |
+
"learning_rate": 9.272244334871377e-06,
|
7491 |
+
"loss": 0.7137,
|
7492 |
+
"step": 1069
|
7493 |
+
},
|
7494 |
+
{
|
7495 |
+
"epoch": 0.007111973705637402,
|
7496 |
+
"grad_norm": 0.8364386558532715,
|
7497 |
+
"learning_rate": 9.203589276774439e-06,
|
7498 |
+
"loss": 0.858,
|
7499 |
+
"step": 1070
|
7500 |
+
},
|
7501 |
+
{
|
7502 |
+
"epoch": 0.007118620410035194,
|
7503 |
+
"grad_norm": 0.8338163495063782,
|
7504 |
+
"learning_rate": 9.135163563922766e-06,
|
7505 |
+
"loss": 0.709,
|
7506 |
+
"step": 1071
|
7507 |
+
},
|
7508 |
+
{
|
7509 |
+
"epoch": 0.007125267114432986,
|
7510 |
+
"grad_norm": 0.7769277691841125,
|
7511 |
+
"learning_rate": 9.06696758098528e-06,
|
7512 |
+
"loss": 0.7744,
|
7513 |
+
"step": 1072
|
7514 |
+
},
|
7515 |
+
{
|
7516 |
+
"epoch": 0.0071319138188307785,
|
7517 |
+
"grad_norm": 0.8195592164993286,
|
7518 |
+
"learning_rate": 8.999001711339434e-06,
|
7519 |
+
"loss": 0.8514,
|
7520 |
+
"step": 1073
|
7521 |
+
},
|
7522 |
+
{
|
7523 |
+
"epoch": 0.00713856052322857,
|
7524 |
+
"grad_norm": 0.7981391549110413,
|
7525 |
+
"learning_rate": 8.931266337069083e-06,
|
7526 |
+
"loss": 0.7724,
|
7527 |
+
"step": 1074
|
7528 |
+
},
|
7529 |
+
{
|
7530 |
+
"epoch": 0.007145207227626362,
|
7531 |
+
"grad_norm": 1.2387161254882812,
|
7532 |
+
"learning_rate": 8.86376183896226e-06,
|
7533 |
+
"loss": 0.6626,
|
7534 |
+
"step": 1075
|
7535 |
+
},
|
7536 |
+
{
|
7537 |
+
"epoch": 0.007151853932024154,
|
7538 |
+
"grad_norm": 1.1157495975494385,
|
7539 |
+
"learning_rate": 8.796488596509133e-06,
|
7540 |
+
"loss": 0.643,
|
7541 |
+
"step": 1076
|
7542 |
+
},
|
7543 |
+
{
|
7544 |
+
"epoch": 0.007158500636421946,
|
7545 |
+
"grad_norm": 0.8519596457481384,
|
7546 |
+
"learning_rate": 8.72944698789977e-06,
|
7547 |
+
"loss": 0.6917,
|
7548 |
+
"step": 1077
|
7549 |
+
},
|
7550 |
+
{
|
7551 |
+
"epoch": 0.007165147340819738,
|
7552 |
+
"grad_norm": 1.1191961765289307,
|
7553 |
+
"learning_rate": 8.66263739002211e-06,
|
7554 |
+
"loss": 0.659,
|
7555 |
+
"step": 1078
|
7556 |
+
},
|
7557 |
+
{
|
7558 |
+
"epoch": 0.00717179404521753,
|
7559 |
+
"grad_norm": 0.961754322052002,
|
7560 |
+
"learning_rate": 8.596060178459758e-06,
|
7561 |
+
"loss": 0.9327,
|
7562 |
+
"step": 1079
|
7563 |
+
},
|
7564 |
+
{
|
7565 |
+
"epoch": 0.007178440749615322,
|
7566 |
+
"grad_norm": 1.1048482656478882,
|
7567 |
+
"learning_rate": 8.529715727489912e-06,
|
7568 |
+
"loss": 0.8056,
|
7569 |
+
"step": 1080
|
7570 |
+
},
|
7571 |
+
{
|
7572 |
+
"epoch": 0.007185087454013114,
|
7573 |
+
"grad_norm": 0.9219740033149719,
|
7574 |
+
"learning_rate": 8.463604410081293e-06,
|
7575 |
+
"loss": 0.6779,
|
7576 |
+
"step": 1081
|
7577 |
+
},
|
7578 |
+
{
|
7579 |
+
"epoch": 0.007191734158410906,
|
7580 |
+
"grad_norm": 1.114214539527893,
|
7581 |
+
"learning_rate": 8.397726597892008e-06,
|
7582 |
+
"loss": 0.9183,
|
7583 |
+
"step": 1082
|
7584 |
+
},
|
7585 |
+
{
|
7586 |
+
"epoch": 0.007198380862808698,
|
7587 |
+
"grad_norm": 0.9252001643180847,
|
7588 |
+
"learning_rate": 8.332082661267443e-06,
|
7589 |
+
"loss": 0.5787,
|
7590 |
+
"step": 1083
|
7591 |
+
},
|
7592 |
+
{
|
7593 |
+
"epoch": 0.00720502756720649,
|
7594 |
+
"grad_norm": 1.4290266036987305,
|
7595 |
+
"learning_rate": 8.266672969238216e-06,
|
7596 |
+
"loss": 0.7914,
|
7597 |
+
"step": 1084
|
7598 |
+
},
|
7599 |
+
{
|
7600 |
+
"epoch": 0.007211674271604282,
|
7601 |
+
"grad_norm": 1.0707231760025024,
|
7602 |
+
"learning_rate": 8.201497889518073e-06,
|
7603 |
+
"loss": 0.6836,
|
7604 |
+
"step": 1085
|
7605 |
+
},
|
7606 |
+
{
|
7607 |
+
"epoch": 0.007218320976002074,
|
7608 |
+
"grad_norm": 1.3955053091049194,
|
7609 |
+
"learning_rate": 8.136557788501903e-06,
|
7610 |
+
"loss": 0.5338,
|
7611 |
+
"step": 1086
|
7612 |
+
},
|
7613 |
+
{
|
7614 |
+
"epoch": 0.0072249676803998655,
|
7615 |
+
"grad_norm": 1.097495675086975,
|
7616 |
+
"learning_rate": 8.071853031263554e-06,
|
7617 |
+
"loss": 0.7974,
|
7618 |
+
"step": 1087
|
7619 |
+
},
|
7620 |
+
{
|
7621 |
+
"epoch": 0.007231614384797658,
|
7622 |
+
"grad_norm": 1.1199660301208496,
|
7623 |
+
"learning_rate": 8.007383981553857e-06,
|
7624 |
+
"loss": 0.8519,
|
7625 |
+
"step": 1088
|
7626 |
+
},
|
7627 |
+
{
|
7628 |
+
"epoch": 0.00723826108919545,
|
7629 |
+
"grad_norm": 0.9631024599075317,
|
7630 |
+
"learning_rate": 7.943151001798554e-06,
|
7631 |
+
"loss": 0.5342,
|
7632 |
+
"step": 1089
|
7633 |
+
},
|
7634 |
+
{
|
7635 |
+
"epoch": 0.0072449077935932416,
|
7636 |
+
"grad_norm": 1.3137375116348267,
|
7637 |
+
"learning_rate": 7.879154453096304e-06,
|
7638 |
+
"loss": 0.9322,
|
7639 |
+
"step": 1090
|
7640 |
+
},
|
7641 |
+
{
|
7642 |
+
"epoch": 0.007251554497991033,
|
7643 |
+
"grad_norm": 1.220885992050171,
|
7644 |
+
"learning_rate": 7.81539469521661e-06,
|
7645 |
+
"loss": 0.8128,
|
7646 |
+
"step": 1091
|
7647 |
+
},
|
7648 |
+
{
|
7649 |
+
"epoch": 0.007258201202388826,
|
7650 |
+
"grad_norm": 1.5835233926773071,
|
7651 |
+
"learning_rate": 7.751872086597783e-06,
|
7652 |
+
"loss": 0.7819,
|
7653 |
+
"step": 1092
|
7654 |
+
},
|
7655 |
+
{
|
7656 |
+
"epoch": 0.007264847906786618,
|
7657 |
+
"grad_norm": 1.3007620573043823,
|
7658 |
+
"learning_rate": 7.688586984344992e-06,
|
7659 |
+
"loss": 0.8538,
|
7660 |
+
"step": 1093
|
7661 |
+
},
|
7662 |
+
{
|
7663 |
+
"epoch": 0.007271494611184409,
|
7664 |
+
"grad_norm": 1.304919719696045,
|
7665 |
+
"learning_rate": 7.6255397442281825e-06,
|
7666 |
+
"loss": 0.5072,
|
7667 |
+
"step": 1094
|
7668 |
+
},
|
7669 |
+
{
|
7670 |
+
"epoch": 0.007278141315582201,
|
7671 |
+
"grad_norm": 1.584822654724121,
|
7672 |
+
"learning_rate": 7.562730720680112e-06,
|
7673 |
+
"loss": 0.983,
|
7674 |
+
"step": 1095
|
7675 |
+
},
|
7676 |
+
{
|
7677 |
+
"epoch": 0.007284788019979994,
|
7678 |
+
"grad_norm": 1.5479097366333008,
|
7679 |
+
"learning_rate": 7.500160266794371e-06,
|
7680 |
+
"loss": 0.9376,
|
7681 |
+
"step": 1096
|
7682 |
+
},
|
7683 |
+
{
|
7684 |
+
"epoch": 0.0072914347243777855,
|
7685 |
+
"grad_norm": 1.8360904455184937,
|
7686 |
+
"learning_rate": 7.437828734323393e-06,
|
7687 |
+
"loss": 1.0823,
|
7688 |
+
"step": 1097
|
7689 |
+
},
|
7690 |
+
{
|
7691 |
+
"epoch": 0.007298081428775577,
|
7692 |
+
"grad_norm": 1.8177344799041748,
|
7693 |
+
"learning_rate": 7.375736473676442e-06,
|
7694 |
+
"loss": 0.704,
|
7695 |
+
"step": 1098
|
7696 |
+
},
|
7697 |
+
{
|
7698 |
+
"epoch": 0.007304728133173369,
|
7699 |
+
"grad_norm": 2.0094518661499023,
|
7700 |
+
"learning_rate": 7.3138838339176675e-06,
|
7701 |
+
"loss": 0.8205,
|
7702 |
+
"step": 1099
|
7703 |
+
},
|
7704 |
+
{
|
7705 |
+
"epoch": 0.0073113748375711616,
|
7706 |
+
"grad_norm": 3.6242754459381104,
|
7707 |
+
"learning_rate": 7.252271162764129e-06,
|
7708 |
+
"loss": 1.2131,
|
7709 |
+
"step": 1100
|
7710 |
+
},
|
7711 |
+
{
|
7712 |
+
"epoch": 0.007318021541968953,
|
7713 |
+
"grad_norm": 0.379312127828598,
|
7714 |
+
"learning_rate": 7.190898806583929e-06,
|
7715 |
+
"loss": 1.0036,
|
7716 |
+
"step": 1101
|
7717 |
+
},
|
7718 |
+
{
|
7719 |
+
"epoch": 0.007324668246366745,
|
7720 |
+
"grad_norm": 0.6471967697143555,
|
7721 |
+
"learning_rate": 7.1297671103941035e-06,
|
7722 |
+
"loss": 1.1373,
|
7723 |
+
"step": 1102
|
7724 |
+
},
|
7725 |
+
{
|
7726 |
+
"epoch": 0.007331314950764537,
|
7727 |
+
"grad_norm": 0.5435198545455933,
|
7728 |
+
"learning_rate": 7.068876417858811e-06,
|
7729 |
+
"loss": 0.7672,
|
7730 |
+
"step": 1103
|
7731 |
+
},
|
7732 |
+
{
|
7733 |
+
"epoch": 0.007337961655162329,
|
7734 |
+
"grad_norm": 0.6249924898147583,
|
7735 |
+
"learning_rate": 7.008227071287338e-06,
|
7736 |
+
"loss": 0.9681,
|
7737 |
+
"step": 1104
|
7738 |
+
},
|
7739 |
+
{
|
7740 |
+
"epoch": 0.007344608359560121,
|
7741 |
+
"grad_norm": 0.8528724312782288,
|
7742 |
+
"learning_rate": 6.947819411632223e-06,
|
7743 |
+
"loss": 0.848,
|
7744 |
+
"step": 1105
|
7745 |
+
},
|
7746 |
+
{
|
7747 |
+
"epoch": 0.007351255063957913,
|
7748 |
+
"grad_norm": 0.6179990768432617,
|
7749 |
+
"learning_rate": 6.887653778487307e-06,
|
7750 |
+
"loss": 0.901,
|
7751 |
+
"step": 1106
|
7752 |
+
},
|
7753 |
+
{
|
7754 |
+
"epoch": 0.007357901768355705,
|
7755 |
+
"grad_norm": 0.5665982365608215,
|
7756 |
+
"learning_rate": 6.827730510085817e-06,
|
7757 |
+
"loss": 0.8572,
|
7758 |
+
"step": 1107
|
7759 |
+
},
|
7760 |
+
{
|
7761 |
+
"epoch": 0.007364548472753497,
|
7762 |
+
"grad_norm": 0.5486406087875366,
|
7763 |
+
"learning_rate": 6.7680499432984654e-06,
|
7764 |
+
"loss": 0.5723,
|
7765 |
+
"step": 1108
|
7766 |
+
},
|
7767 |
+
{
|
7768 |
+
"epoch": 0.007371195177151289,
|
7769 |
+
"grad_norm": 0.5960477590560913,
|
7770 |
+
"learning_rate": 6.708612413631615e-06,
|
7771 |
+
"loss": 0.8061,
|
7772 |
+
"step": 1109
|
7773 |
+
},
|
7774 |
+
{
|
7775 |
+
"epoch": 0.007377841881549081,
|
7776 |
+
"grad_norm": 0.7446190714836121,
|
7777 |
+
"learning_rate": 6.649418255225298e-06,
|
7778 |
+
"loss": 0.8753,
|
7779 |
+
"step": 1110
|
7780 |
+
},
|
7781 |
+
{
|
7782 |
+
"epoch": 0.007384488585946873,
|
7783 |
+
"grad_norm": 0.6754137873649597,
|
7784 |
+
"learning_rate": 6.590467800851419e-06,
|
7785 |
+
"loss": 0.7474,
|
7786 |
+
"step": 1111
|
7787 |
+
},
|
7788 |
+
{
|
7789 |
+
"epoch": 0.007391135290344665,
|
7790 |
+
"grad_norm": 0.7214033603668213,
|
7791 |
+
"learning_rate": 6.531761381911827e-06,
|
7792 |
+
"loss": 0.8932,
|
7793 |
+
"step": 1112
|
7794 |
+
},
|
7795 |
+
{
|
7796 |
+
"epoch": 0.007397781994742457,
|
7797 |
+
"grad_norm": 0.6419174075126648,
|
7798 |
+
"learning_rate": 6.473299328436499e-06,
|
7799 |
+
"loss": 0.7526,
|
7800 |
+
"step": 1113
|
7801 |
+
},
|
7802 |
+
{
|
7803 |
+
"epoch": 0.0074044286991402485,
|
7804 |
+
"grad_norm": 0.6379873156547546,
|
7805 |
+
"learning_rate": 6.415081969081649e-06,
|
7806 |
+
"loss": 0.7635,
|
7807 |
+
"step": 1114
|
7808 |
+
},
|
7809 |
+
{
|
7810 |
+
"epoch": 0.007411075403538041,
|
7811 |
+
"grad_norm": 0.7169990539550781,
|
7812 |
+
"learning_rate": 6.357109631127889e-06,
|
7813 |
+
"loss": 0.7471,
|
7814 |
+
"step": 1115
|
7815 |
+
},
|
7816 |
+
{
|
7817 |
+
"epoch": 0.007417722107935833,
|
7818 |
+
"grad_norm": 0.64534592628479,
|
7819 |
+
"learning_rate": 6.2993826404783965e-06,
|
7820 |
+
"loss": 1.0755,
|
7821 |
+
"step": 1116
|
7822 |
+
},
|
7823 |
+
{
|
7824 |
+
"epoch": 0.007424368812333625,
|
7825 |
+
"grad_norm": 0.6701549291610718,
|
7826 |
+
"learning_rate": 6.241901321657112e-06,
|
7827 |
+
"loss": 0.9706,
|
7828 |
+
"step": 1117
|
7829 |
+
},
|
7830 |
+
{
|
7831 |
+
"epoch": 0.007431015516731416,
|
7832 |
+
"grad_norm": 0.7817533612251282,
|
7833 |
+
"learning_rate": 6.184665997806832e-06,
|
7834 |
+
"loss": 1.0468,
|
7835 |
+
"step": 1118
|
7836 |
+
},
|
7837 |
+
{
|
7838 |
+
"epoch": 0.007437662221129209,
|
7839 |
+
"grad_norm": 0.6312013864517212,
|
7840 |
+
"learning_rate": 6.127676990687453e-06,
|
7841 |
+
"loss": 0.9234,
|
7842 |
+
"step": 1119
|
7843 |
+
},
|
7844 |
+
{
|
7845 |
+
"epoch": 0.007444308925527001,
|
7846 |
+
"grad_norm": 0.7704412937164307,
|
7847 |
+
"learning_rate": 6.07093462067419e-06,
|
7848 |
+
"loss": 0.8273,
|
7849 |
+
"step": 1120
|
7850 |
+
},
|
7851 |
+
{
|
7852 |
+
"epoch": 0.0074509556299247924,
|
7853 |
+
"grad_norm": 0.9202284216880798,
|
7854 |
+
"learning_rate": 6.014439206755706e-06,
|
7855 |
+
"loss": 0.6653,
|
7856 |
+
"step": 1121
|
7857 |
+
},
|
7858 |
+
{
|
7859 |
+
"epoch": 0.007457602334322584,
|
7860 |
+
"grad_norm": 0.922299861907959,
|
7861 |
+
"learning_rate": 5.958191066532354e-06,
|
7862 |
+
"loss": 0.9426,
|
7863 |
+
"step": 1122
|
7864 |
+
},
|
7865 |
+
{
|
7866 |
+
"epoch": 0.007464249038720377,
|
7867 |
+
"grad_norm": 0.8126717209815979,
|
7868 |
+
"learning_rate": 5.902190516214384e-06,
|
7869 |
+
"loss": 1.0114,
|
7870 |
+
"step": 1123
|
7871 |
+
},
|
7872 |
+
{
|
7873 |
+
"epoch": 0.0074708957431181685,
|
7874 |
+
"grad_norm": 1.1028474569320679,
|
7875 |
+
"learning_rate": 5.846437870620192e-06,
|
7876 |
+
"loss": 0.891,
|
7877 |
+
"step": 1124
|
7878 |
+
},
|
7879 |
+
{
|
7880 |
+
"epoch": 0.00747754244751596,
|
7881 |
+
"grad_norm": 0.8131560683250427,
|
7882 |
+
"learning_rate": 5.79093344317449e-06,
|
7883 |
+
"loss": 0.7927,
|
7884 |
+
"step": 1125
|
7885 |
+
},
|
7886 |
+
{
|
7887 |
+
"epoch": 0.007484189151913752,
|
7888 |
+
"grad_norm": 0.8165668845176697,
|
7889 |
+
"learning_rate": 5.735677545906626e-06,
|
7890 |
+
"loss": 0.7592,
|
7891 |
+
"step": 1126
|
7892 |
+
},
|
7893 |
+
{
|
7894 |
+
"epoch": 0.007490835856311545,
|
7895 |
+
"grad_norm": 0.9261493682861328,
|
7896 |
+
"learning_rate": 5.680670489448742e-06,
|
7897 |
+
"loss": 0.7585,
|
7898 |
+
"step": 1127
|
7899 |
+
},
|
7900 |
+
{
|
7901 |
+
"epoch": 0.007497482560709336,
|
7902 |
+
"grad_norm": 0.8316916823387146,
|
7903 |
+
"learning_rate": 5.625912583034115e-06,
|
7904 |
+
"loss": 0.7128,
|
7905 |
+
"step": 1128
|
7906 |
+
},
|
7907 |
+
{
|
7908 |
+
"epoch": 0.007504129265107128,
|
7909 |
+
"grad_norm": 0.9080417156219482,
|
7910 |
+
"learning_rate": 5.5714041344953445e-06,
|
7911 |
+
"loss": 0.663,
|
7912 |
+
"step": 1129
|
7913 |
+
},
|
7914 |
+
{
|
7915 |
+
"epoch": 0.00751077596950492,
|
7916 |
+
"grad_norm": 0.9159788489341736,
|
7917 |
+
"learning_rate": 5.51714545026264e-06,
|
7918 |
+
"loss": 0.8107,
|
7919 |
+
"step": 1130
|
7920 |
+
},
|
7921 |
+
{
|
7922 |
+
"epoch": 0.0075174226739027124,
|
7923 |
+
"grad_norm": 1.1240578889846802,
|
7924 |
+
"learning_rate": 5.463136835362148e-06,
|
7925 |
+
"loss": 0.6353,
|
7926 |
+
"step": 1131
|
7927 |
+
},
|
7928 |
+
{
|
7929 |
+
"epoch": 0.007524069378300504,
|
7930 |
+
"grad_norm": 1.2321254014968872,
|
7931 |
+
"learning_rate": 5.409378593414194e-06,
|
7932 |
+
"loss": 0.995,
|
7933 |
+
"step": 1132
|
7934 |
+
},
|
7935 |
+
{
|
7936 |
+
"epoch": 0.007530716082698296,
|
7937 |
+
"grad_norm": 0.9713672399520874,
|
7938 |
+
"learning_rate": 5.355871026631554e-06,
|
7939 |
+
"loss": 0.7501,
|
7940 |
+
"step": 1133
|
7941 |
+
},
|
7942 |
+
{
|
7943 |
+
"epoch": 0.007537362787096088,
|
7944 |
+
"grad_norm": 1.0002257823944092,
|
7945 |
+
"learning_rate": 5.302614435817793e-06,
|
7946 |
+
"loss": 0.8235,
|
7947 |
+
"step": 1134
|
7948 |
+
},
|
7949 |
+
{
|
7950 |
+
"epoch": 0.00754400949149388,
|
7951 |
+
"grad_norm": 1.2293847799301147,
|
7952 |
+
"learning_rate": 5.249609120365578e-06,
|
7953 |
+
"loss": 0.731,
|
7954 |
+
"step": 1135
|
7955 |
+
},
|
7956 |
+
{
|
7957 |
+
"epoch": 0.007550656195891672,
|
7958 |
+
"grad_norm": 0.8931113481521606,
|
7959 |
+
"learning_rate": 5.196855378254989e-06,
|
7960 |
+
"loss": 0.7542,
|
7961 |
+
"step": 1136
|
7962 |
+
},
|
7963 |
+
{
|
7964 |
+
"epoch": 0.007557302900289464,
|
7965 |
+
"grad_norm": 1.4215387105941772,
|
7966 |
+
"learning_rate": 5.144353506051797e-06,
|
7967 |
+
"loss": 0.751,
|
7968 |
+
"step": 1137
|
7969 |
+
},
|
7970 |
+
{
|
7971 |
+
"epoch": 0.007563949604687256,
|
7972 |
+
"grad_norm": 1.4619780778884888,
|
7973 |
+
"learning_rate": 5.0921037989058614e-06,
|
7974 |
+
"loss": 0.8514,
|
7975 |
+
"step": 1138
|
7976 |
+
},
|
7977 |
+
{
|
7978 |
+
"epoch": 0.007570596309085048,
|
7979 |
+
"grad_norm": 2.04518723487854,
|
7980 |
+
"learning_rate": 5.0401065505494445e-06,
|
7981 |
+
"loss": 1.0564,
|
7982 |
+
"step": 1139
|
7983 |
+
},
|
7984 |
+
{
|
7985 |
+
"epoch": 0.00757724301348284,
|
7986 |
+
"grad_norm": 1.1913663148880005,
|
7987 |
+
"learning_rate": 4.988362053295564e-06,
|
7988 |
+
"loss": 0.8242,
|
7989 |
+
"step": 1140
|
7990 |
+
},
|
7991 |
+
{
|
7992 |
+
"epoch": 0.007583889717880632,
|
7993 |
+
"grad_norm": 1.0685111284255981,
|
7994 |
+
"learning_rate": 4.9368705980363415e-06,
|
7995 |
+
"loss": 0.5443,
|
7996 |
+
"step": 1141
|
7997 |
+
},
|
7998 |
+
{
|
7999 |
+
"epoch": 0.007590536422278424,
|
8000 |
+
"grad_norm": 1.0553430318832397,
|
8001 |
+
"learning_rate": 4.885632474241347e-06,
|
8002 |
+
"loss": 0.624,
|
8003 |
+
"step": 1142
|
8004 |
+
},
|
8005 |
+
{
|
8006 |
+
"epoch": 0.007597183126676216,
|
8007 |
+
"grad_norm": 1.1717983484268188,
|
8008 |
+
"learning_rate": 4.834647969956052e-06,
|
8009 |
+
"loss": 0.7475,
|
8010 |
+
"step": 1143
|
8011 |
+
},
|
8012 |
+
{
|
8013 |
+
"epoch": 0.007603829831074008,
|
8014 |
+
"grad_norm": 1.592492938041687,
|
8015 |
+
"learning_rate": 4.783917371800101e-06,
|
8016 |
+
"loss": 0.7852,
|
8017 |
+
"step": 1144
|
8018 |
+
},
|
8019 |
+
{
|
8020 |
+
"epoch": 0.007610476535471799,
|
8021 |
+
"grad_norm": 1.7482496500015259,
|
8022 |
+
"learning_rate": 4.733440964965791e-06,
|
8023 |
+
"loss": 1.0631,
|
8024 |
+
"step": 1145
|
8025 |
+
},
|
8026 |
+
{
|
8027 |
+
"epoch": 0.007617123239869592,
|
8028 |
+
"grad_norm": 1.701235055923462,
|
8029 |
+
"learning_rate": 4.683219033216402e-06,
|
8030 |
+
"loss": 0.9337,
|
8031 |
+
"step": 1146
|
8032 |
+
},
|
8033 |
+
{
|
8034 |
+
"epoch": 0.007623769944267384,
|
8035 |
+
"grad_norm": 2.070103168487549,
|
8036 |
+
"learning_rate": 4.633251858884657e-06,
|
8037 |
+
"loss": 0.814,
|
8038 |
+
"step": 1147
|
8039 |
+
},
|
8040 |
+
{
|
8041 |
+
"epoch": 0.0076304166486651755,
|
8042 |
+
"grad_norm": 2.205002784729004,
|
8043 |
+
"learning_rate": 4.583539722871094e-06,
|
8044 |
+
"loss": 0.7706,
|
8045 |
+
"step": 1148
|
8046 |
+
},
|
8047 |
+
{
|
8048 |
+
"epoch": 0.007637063353062967,
|
8049 |
+
"grad_norm": 2.1740975379943848,
|
8050 |
+
"learning_rate": 4.534082904642495e-06,
|
8051 |
+
"loss": 0.8495,
|
8052 |
+
"step": 1149
|
8053 |
+
},
|
8054 |
+
{
|
8055 |
+
"epoch": 0.00764371005746076,
|
8056 |
+
"grad_norm": 3.0469541549682617,
|
8057 |
+
"learning_rate": 4.484881682230341e-06,
|
8058 |
+
"loss": 0.7312,
|
8059 |
+
"step": 1150
|
8060 |
+
},
|
8061 |
+
{
|
8062 |
+
"epoch": 0.007650356761858552,
|
8063 |
+
"grad_norm": 0.3347352147102356,
|
8064 |
+
"learning_rate": 4.435936332229229e-06,
|
8065 |
+
"loss": 0.8578,
|
8066 |
+
"step": 1151
|
8067 |
+
},
|
8068 |
+
{
|
8069 |
+
"epoch": 0.007657003466256343,
|
8070 |
+
"grad_norm": 0.734880805015564,
|
8071 |
+
"learning_rate": 4.3872471297952965e-06,
|
8072 |
+
"loss": 0.9773,
|
8073 |
+
"step": 1152
|
8074 |
+
},
|
8075 |
+
{
|
8076 |
+
"epoch": 0.007663650170654135,
|
8077 |
+
"grad_norm": 0.7428154349327087,
|
8078 |
+
"learning_rate": 4.3388143486447045e-06,
|
8079 |
+
"loss": 0.8733,
|
8080 |
+
"step": 1153
|
8081 |
+
},
|
8082 |
+
{
|
8083 |
+
"epoch": 0.007670296875051928,
|
8084 |
+
"grad_norm": 0.5158607363700867,
|
8085 |
+
"learning_rate": 4.290638261052099e-06,
|
8086 |
+
"loss": 0.9492,
|
8087 |
+
"step": 1154
|
8088 |
+
},
|
8089 |
+
{
|
8090 |
+
"epoch": 0.007676943579449719,
|
8091 |
+
"grad_norm": 0.5684289932250977,
|
8092 |
+
"learning_rate": 4.242719137849077e-06,
|
8093 |
+
"loss": 0.8136,
|
8094 |
+
"step": 1155
|
8095 |
+
},
|
8096 |
+
{
|
8097 |
+
"epoch": 0.007683590283847511,
|
8098 |
+
"grad_norm": 0.7612115144729614,
|
8099 |
+
"learning_rate": 4.1950572484226345e-06,
|
8100 |
+
"loss": 0.914,
|
8101 |
+
"step": 1156
|
8102 |
+
},
|
8103 |
+
{
|
8104 |
+
"epoch": 0.007690236988245303,
|
8105 |
+
"grad_norm": 0.6157774925231934,
|
8106 |
+
"learning_rate": 4.147652860713685e-06,
|
8107 |
+
"loss": 0.947,
|
8108 |
+
"step": 1157
|
8109 |
+
},
|
8110 |
+
{
|
8111 |
+
"epoch": 0.0076968836926430955,
|
8112 |
+
"grad_norm": 0.7647441029548645,
|
8113 |
+
"learning_rate": 4.100506241215562e-06,
|
8114 |
+
"loss": 0.8584,
|
8115 |
+
"step": 1158
|
8116 |
+
},
|
8117 |
+
{
|
8118 |
+
"epoch": 0.007703530397040887,
|
8119 |
+
"grad_norm": 0.7151400446891785,
|
8120 |
+
"learning_rate": 4.0536176549724806e-06,
|
8121 |
+
"loss": 0.6786,
|
8122 |
+
"step": 1159
|
8123 |
+
},
|
8124 |
+
{
|
8125 |
+
"epoch": 0.007710177101438679,
|
8126 |
+
"grad_norm": 0.6203306913375854,
|
8127 |
+
"learning_rate": 4.00698736557808e-06,
|
8128 |
+
"loss": 0.9418,
|
8129 |
+
"step": 1160
|
8130 |
+
},
|
8131 |
+
{
|
8132 |
+
"epoch": 0.007716823805836471,
|
8133 |
+
"grad_norm": 0.7037209868431091,
|
8134 |
+
"learning_rate": 3.960615635173925e-06,
|
8135 |
+
"loss": 0.8577,
|
8136 |
+
"step": 1161
|
8137 |
+
},
|
8138 |
+
{
|
8139 |
+
"epoch": 0.007723470510234263,
|
8140 |
+
"grad_norm": 0.6116876006126404,
|
8141 |
+
"learning_rate": 3.914502724448061e-06,
|
8142 |
+
"loss": 0.9221,
|
8143 |
+
"step": 1162
|
8144 |
+
},
|
8145 |
+
{
|
8146 |
+
"epoch": 0.007730117214632055,
|
8147 |
+
"grad_norm": 0.8219679594039917,
|
8148 |
+
"learning_rate": 3.868648892633497e-06,
|
8149 |
+
"loss": 0.8342,
|
8150 |
+
"step": 1163
|
8151 |
+
},
|
8152 |
+
{
|
8153 |
+
"epoch": 0.007736763919029847,
|
8154 |
+
"grad_norm": 0.6384825110435486,
|
8155 |
+
"learning_rate": 3.823054397506781e-06,
|
8156 |
+
"loss": 0.8674,
|
8157 |
+
"step": 1164
|
8158 |
+
},
|
8159 |
+
{
|
8160 |
+
"epoch": 0.007743410623427639,
|
8161 |
+
"grad_norm": 0.6831299662590027,
|
8162 |
+
"learning_rate": 3.7777194953865667e-06,
|
8163 |
+
"loss": 0.7916,
|
8164 |
+
"step": 1165
|
8165 |
+
},
|
8166 |
+
{
|
8167 |
+
"epoch": 0.007750057327825431,
|
8168 |
+
"grad_norm": 0.7946675419807434,
|
8169 |
+
"learning_rate": 3.7326444411321547e-06,
|
8170 |
+
"loss": 0.8928,
|
8171 |
+
"step": 1166
|
8172 |
+
},
|
8173 |
+
{
|
8174 |
+
"epoch": 0.007756704032223223,
|
8175 |
+
"grad_norm": 0.6539303064346313,
|
8176 |
+
"learning_rate": 3.6878294881420363e-06,
|
8177 |
+
"loss": 0.9344,
|
8178 |
+
"step": 1167
|
8179 |
+
},
|
8180 |
+
{
|
8181 |
+
"epoch": 0.007763350736621015,
|
8182 |
+
"grad_norm": 1.1968379020690918,
|
8183 |
+
"learning_rate": 3.6432748883524935e-06,
|
8184 |
+
"loss": 0.9727,
|
8185 |
+
"step": 1168
|
8186 |
+
},
|
8187 |
+
{
|
8188 |
+
"epoch": 0.007769997441018807,
|
8189 |
+
"grad_norm": 0.6433477401733398,
|
8190 |
+
"learning_rate": 3.598980892236198e-06,
|
8191 |
+
"loss": 0.6713,
|
8192 |
+
"step": 1169
|
8193 |
+
},
|
8194 |
+
{
|
8195 |
+
"epoch": 0.007776644145416599,
|
8196 |
+
"grad_norm": 0.9618440866470337,
|
8197 |
+
"learning_rate": 3.5549477488007854e-06,
|
8198 |
+
"loss": 1.0181,
|
8199 |
+
"step": 1170
|
8200 |
+
},
|
8201 |
+
{
|
8202 |
+
"epoch": 0.007783290849814391,
|
8203 |
+
"grad_norm": 0.8468819856643677,
|
8204 |
+
"learning_rate": 3.511175705587433e-06,
|
8205 |
+
"loss": 0.8858,
|
8206 |
+
"step": 1171
|
8207 |
+
},
|
8208 |
+
{
|
8209 |
+
"epoch": 0.0077899375542121825,
|
8210 |
+
"grad_norm": 0.7299633026123047,
|
8211 |
+
"learning_rate": 3.4676650086695016e-06,
|
8212 |
+
"loss": 0.7895,
|
8213 |
+
"step": 1172
|
8214 |
+
},
|
8215 |
+
{
|
8216 |
+
"epoch": 0.007796584258609975,
|
8217 |
+
"grad_norm": 0.9545450210571289,
|
8218 |
+
"learning_rate": 3.4244159026511566e-06,
|
8219 |
+
"loss": 0.8403,
|
8220 |
+
"step": 1173
|
8221 |
+
},
|
8222 |
+
{
|
8223 |
+
"epoch": 0.007803230963007767,
|
8224 |
+
"grad_norm": 0.9891371726989746,
|
8225 |
+
"learning_rate": 3.3814286306659502e-06,
|
8226 |
+
"loss": 0.9864,
|
8227 |
+
"step": 1174
|
8228 |
+
},
|
8229 |
+
{
|
8230 |
+
"epoch": 0.0078098776674055586,
|
8231 |
+
"grad_norm": 0.7887678146362305,
|
8232 |
+
"learning_rate": 3.3387034343755065e-06,
|
8233 |
+
"loss": 0.9753,
|
8234 |
+
"step": 1175
|
8235 |
+
},
|
8236 |
+
{
|
8237 |
+
"epoch": 0.00781652437180335,
|
8238 |
+
"grad_norm": 0.9488683938980103,
|
8239 |
+
"learning_rate": 3.2962405539681217e-06,
|
8240 |
+
"loss": 0.6611,
|
8241 |
+
"step": 1176
|
8242 |
+
},
|
8243 |
+
{
|
8244 |
+
"epoch": 0.007823171076201143,
|
8245 |
+
"grad_norm": 1.0425969362258911,
|
8246 |
+
"learning_rate": 3.25404022815744e-06,
|
8247 |
+
"loss": 0.7946,
|
8248 |
+
"step": 1177
|
8249 |
+
},
|
8250 |
+
{
|
8251 |
+
"epoch": 0.007829817780598934,
|
8252 |
+
"grad_norm": 1.1102572679519653,
|
8253 |
+
"learning_rate": 3.2121026941811015e-06,
|
8254 |
+
"loss": 0.4979,
|
8255 |
+
"step": 1178
|
8256 |
+
},
|
8257 |
+
{
|
8258 |
+
"epoch": 0.007836464484996726,
|
8259 |
+
"grad_norm": 0.8967680931091309,
|
8260 |
+
"learning_rate": 3.1704281877993903e-06,
|
8261 |
+
"loss": 0.8126,
|
8262 |
+
"step": 1179
|
8263 |
+
},
|
8264 |
+
{
|
8265 |
+
"epoch": 0.007843111189394519,
|
8266 |
+
"grad_norm": 0.7562099099159241,
|
8267 |
+
"learning_rate": 3.1290169432939553e-06,
|
8268 |
+
"loss": 0.7026,
|
8269 |
+
"step": 1180
|
8270 |
+
},
|
8271 |
+
{
|
8272 |
+
"epoch": 0.00784975789379231,
|
8273 |
+
"grad_norm": 0.9412071108818054,
|
8274 |
+
"learning_rate": 3.087869193466458e-06,
|
8275 |
+
"loss": 0.8846,
|
8276 |
+
"step": 1181
|
8277 |
+
},
|
8278 |
+
{
|
8279 |
+
"epoch": 0.007856404598190102,
|
8280 |
+
"grad_norm": 1.510528564453125,
|
8281 |
+
"learning_rate": 3.0469851696372564e-06,
|
8282 |
+
"loss": 1.1582,
|
8283 |
+
"step": 1182
|
8284 |
+
},
|
8285 |
+
{
|
8286 |
+
"epoch": 0.007863051302587895,
|
8287 |
+
"grad_norm": 1.3257495164871216,
|
8288 |
+
"learning_rate": 3.0063651016441428e-06,
|
8289 |
+
"loss": 0.7229,
|
8290 |
+
"step": 1183
|
8291 |
+
},
|
8292 |
+
{
|
8293 |
+
"epoch": 0.007869698006985686,
|
8294 |
+
"grad_norm": 1.0000684261322021,
|
8295 |
+
"learning_rate": 2.9660092178409927e-06,
|
8296 |
+
"loss": 0.8639,
|
8297 |
+
"step": 1184
|
8298 |
+
},
|
8299 |
+
{
|
8300 |
+
"epoch": 0.007876344711383479,
|
8301 |
+
"grad_norm": 1.1794826984405518,
|
8302 |
+
"learning_rate": 2.9259177450965682e-06,
|
8303 |
+
"loss": 0.8521,
|
8304 |
+
"step": 1185
|
8305 |
+
},
|
8306 |
+
{
|
8307 |
+
"epoch": 0.00788299141578127,
|
8308 |
+
"grad_norm": 1.2365461587905884,
|
8309 |
+
"learning_rate": 2.8860909087931543e-06,
|
8310 |
+
"loss": 0.7315,
|
8311 |
+
"step": 1186
|
8312 |
+
},
|
8313 |
+
{
|
8314 |
+
"epoch": 0.007889638120179062,
|
8315 |
+
"grad_norm": 1.2363101243972778,
|
8316 |
+
"learning_rate": 2.8465289328253376e-06,
|
8317 |
+
"loss": 0.873,
|
8318 |
+
"step": 1187
|
8319 |
+
},
|
8320 |
+
{
|
8321 |
+
"epoch": 0.007896284824576855,
|
8322 |
+
"grad_norm": 1.118465781211853,
|
8323 |
+
"learning_rate": 2.8072320395987285e-06,
|
8324 |
+
"loss": 0.8294,
|
8325 |
+
"step": 1188
|
8326 |
+
},
|
8327 |
+
{
|
8328 |
+
"epoch": 0.007902931528974646,
|
8329 |
+
"grad_norm": 1.2116254568099976,
|
8330 |
+
"learning_rate": 2.7682004500287464e-06,
|
8331 |
+
"loss": 0.7468,
|
8332 |
+
"step": 1189
|
8333 |
+
},
|
8334 |
+
{
|
8335 |
+
"epoch": 0.007909578233372438,
|
8336 |
+
"grad_norm": 1.4756786823272705,
|
8337 |
+
"learning_rate": 2.7294343835393368e-06,
|
8338 |
+
"loss": 0.7626,
|
8339 |
+
"step": 1190
|
8340 |
+
},
|
8341 |
+
{
|
8342 |
+
"epoch": 0.00791622493777023,
|
8343 |
+
"grad_norm": 1.321200966835022,
|
8344 |
+
"learning_rate": 2.690934058061756e-06,
|
8345 |
+
"loss": 0.9202,
|
8346 |
+
"step": 1191
|
8347 |
+
},
|
8348 |
+
{
|
8349 |
+
"epoch": 0.007922871642168022,
|
8350 |
+
"grad_norm": 1.2931435108184814,
|
8351 |
+
"learning_rate": 2.6526996900333277e-06,
|
8352 |
+
"loss": 0.7292,
|
8353 |
+
"step": 1192
|
8354 |
+
},
|
8355 |
+
{
|
8356 |
+
"epoch": 0.007929518346565814,
|
8357 |
+
"grad_norm": 1.1711453199386597,
|
8358 |
+
"learning_rate": 2.614731494396283e-06,
|
8359 |
+
"loss": 0.7659,
|
8360 |
+
"step": 1193
|
8361 |
+
},
|
8362 |
+
{
|
8363 |
+
"epoch": 0.007936165050963607,
|
8364 |
+
"grad_norm": 1.4512163400650024,
|
8365 |
+
"learning_rate": 2.577029684596466e-06,
|
8366 |
+
"loss": 0.7325,
|
8367 |
+
"step": 1194
|
8368 |
+
},
|
8369 |
+
{
|
8370 |
+
"epoch": 0.007942811755361398,
|
8371 |
+
"grad_norm": 1.4844986200332642,
|
8372 |
+
"learning_rate": 2.539594472582213e-06,
|
8373 |
+
"loss": 0.7598,
|
8374 |
+
"step": 1195
|
8375 |
+
},
|
8376 |
+
{
|
8377 |
+
"epoch": 0.00794945845975919,
|
8378 |
+
"grad_norm": 1.782923936843872,
|
8379 |
+
"learning_rate": 2.5024260688030987e-06,
|
8380 |
+
"loss": 0.659,
|
8381 |
+
"step": 1196
|
8382 |
+
},
|
8383 |
+
{
|
8384 |
+
"epoch": 0.007956105164156981,
|
8385 |
+
"grad_norm": 2.0441620349884033,
|
8386 |
+
"learning_rate": 2.465524682208814e-06,
|
8387 |
+
"loss": 0.9054,
|
8388 |
+
"step": 1197
|
8389 |
+
},
|
8390 |
+
{
|
8391 |
+
"epoch": 0.007962751868554774,
|
8392 |
+
"grad_norm": 2.2920122146606445,
|
8393 |
+
"learning_rate": 2.4288905202479283e-06,
|
8394 |
+
"loss": 0.9065,
|
8395 |
+
"step": 1198
|
8396 |
+
},
|
8397 |
+
{
|
8398 |
+
"epoch": 0.007969398572952566,
|
8399 |
+
"grad_norm": 2.007251501083374,
|
8400 |
+
"learning_rate": 2.3925237888667572e-06,
|
8401 |
+
"loss": 0.8426,
|
8402 |
+
"step": 1199
|
8403 |
+
},
|
8404 |
+
{
|
8405 |
+
"epoch": 0.007976045277350357,
|
8406 |
+
"grad_norm": 2.4775331020355225,
|
8407 |
+
"learning_rate": 2.3564246925082357e-06,
|
8408 |
+
"loss": 0.6222,
|
8409 |
+
"step": 1200
|
8410 |
+
},
|
8411 |
+
{
|
8412 |
+
"epoch": 0.00798269198174815,
|
8413 |
+
"grad_norm": 0.31705576181411743,
|
8414 |
+
"learning_rate": 2.320593434110696e-06,
|
8415 |
+
"loss": 1.0729,
|
8416 |
+
"step": 1201
|
8417 |
+
},
|
8418 |
+
{
|
8419 |
+
"epoch": 0.007989338686145942,
|
8420 |
+
"grad_norm": 0.6680214405059814,
|
8421 |
+
"learning_rate": 2.2850302151067814e-06,
|
8422 |
+
"loss": 0.884,
|
8423 |
+
"step": 1202
|
8424 |
+
},
|
8425 |
+
{
|
8426 |
+
"epoch": 0.007995985390543733,
|
8427 |
+
"grad_norm": 0.5763593316078186,
|
8428 |
+
"learning_rate": 2.2497352354222902e-06,
|
8429 |
+
"loss": 0.7567,
|
8430 |
+
"step": 1203
|
8431 |
+
},
|
8432 |
+
{
|
8433 |
+
"epoch": 0.008002632094941526,
|
8434 |
+
"grad_norm": 0.5514530539512634,
|
8435 |
+
"learning_rate": 2.214708693475065e-06,
|
8436 |
+
"loss": 0.768,
|
8437 |
+
"step": 1204
|
8438 |
+
},
|
8439 |
+
{
|
8440 |
+
"epoch": 0.008009278799339317,
|
8441 |
+
"grad_norm": 0.6078202128410339,
|
8442 |
+
"learning_rate": 2.179950786173879e-06,
|
8443 |
+
"loss": 0.839,
|
8444 |
+
"step": 1205
|
8445 |
+
},
|
8446 |
+
{
|
8447 |
+
"epoch": 0.00801592550373711,
|
8448 |
+
"grad_norm": 0.630172610282898,
|
8449 |
+
"learning_rate": 2.145461708917312e-06,
|
8450 |
+
"loss": 0.8652,
|
8451 |
+
"step": 1206
|
8452 |
+
},
|
8453 |
+
{
|
8454 |
+
"epoch": 0.008022572208134902,
|
8455 |
+
"grad_norm": 0.7208753824234009,
|
8456 |
+
"learning_rate": 2.1112416555926497e-06,
|
8457 |
+
"loss": 0.9369,
|
8458 |
+
"step": 1207
|
8459 |
+
},
|
8460 |
+
{
|
8461 |
+
"epoch": 0.008029218912532693,
|
8462 |
+
"grad_norm": 0.8084291815757751,
|
8463 |
+
"learning_rate": 2.077290818574834e-06,
|
8464 |
+
"loss": 0.7762,
|
8465 |
+
"step": 1208
|
8466 |
+
},
|
8467 |
+
{
|
8468 |
+
"epoch": 0.008035865616930486,
|
8469 |
+
"grad_norm": 0.5719032287597656,
|
8470 |
+
"learning_rate": 2.043609388725326e-06,
|
8471 |
+
"loss": 0.9558,
|
8472 |
+
"step": 1209
|
8473 |
+
},
|
8474 |
+
{
|
8475 |
+
"epoch": 0.008042512321328278,
|
8476 |
+
"grad_norm": 0.6796174645423889,
|
8477 |
+
"learning_rate": 2.01019755539108e-06,
|
8478 |
+
"loss": 1.0185,
|
8479 |
+
"step": 1210
|
8480 |
+
},
|
8481 |
+
{
|
8482 |
+
"epoch": 0.008049159025726069,
|
8483 |
+
"grad_norm": 0.6527438759803772,
|
8484 |
+
"learning_rate": 1.9770555064034467e-06,
|
8485 |
+
"loss": 0.7037,
|
8486 |
+
"step": 1211
|
8487 |
+
},
|
8488 |
+
{
|
8489 |
+
"epoch": 0.008055805730123862,
|
8490 |
+
"grad_norm": 0.776106595993042,
|
8491 |
+
"learning_rate": 1.944183428077145e-06,
|
8492 |
+
"loss": 0.8553,
|
8493 |
+
"step": 1212
|
8494 |
+
},
|
8495 |
+
{
|
8496 |
+
"epoch": 0.008062452434521653,
|
8497 |
+
"grad_norm": 0.5876369476318359,
|
8498 |
+
"learning_rate": 1.911581505209176e-06,
|
8499 |
+
"loss": 0.9049,
|
8500 |
+
"step": 1213
|
8501 |
+
},
|
8502 |
+
{
|
8503 |
+
"epoch": 0.008069099138919445,
|
8504 |
+
"grad_norm": 0.7080634832382202,
|
8505 |
+
"learning_rate": 1.8792499210778191e-06,
|
8506 |
+
"loss": 0.8793,
|
8507 |
+
"step": 1214
|
8508 |
+
},
|
8509 |
+
{
|
8510 |
+
"epoch": 0.008075745843317238,
|
8511 |
+
"grad_norm": 0.9152222275733948,
|
8512 |
+
"learning_rate": 1.8471888574415951e-06,
|
8513 |
+
"loss": 0.6502,
|
8514 |
+
"step": 1215
|
8515 |
+
},
|
8516 |
+
{
|
8517 |
+
"epoch": 0.008082392547715029,
|
8518 |
+
"grad_norm": 0.7787750959396362,
|
8519 |
+
"learning_rate": 1.8153984945382452e-06,
|
8520 |
+
"loss": 0.8039,
|
8521 |
+
"step": 1216
|
8522 |
+
},
|
8523 |
+
{
|
8524 |
+
"epoch": 0.008089039252112821,
|
8525 |
+
"grad_norm": 0.8649187088012695,
|
8526 |
+
"learning_rate": 1.783879011083689e-06,
|
8527 |
+
"loss": 0.8385,
|
8528 |
+
"step": 1217
|
8529 |
+
},
|
8530 |
+
{
|
8531 |
+
"epoch": 0.008095685956510614,
|
8532 |
+
"grad_norm": 0.6336601376533508,
|
8533 |
+
"learning_rate": 1.7526305842710532e-06,
|
8534 |
+
"loss": 0.8588,
|
8535 |
+
"step": 1218
|
8536 |
+
},
|
8537 |
+
{
|
8538 |
+
"epoch": 0.008102332660908405,
|
8539 |
+
"grad_norm": 0.6925593614578247,
|
8540 |
+
"learning_rate": 1.7216533897696675e-06,
|
8541 |
+
"loss": 0.9039,
|
8542 |
+
"step": 1219
|
8543 |
+
},
|
8544 |
+
{
|
8545 |
+
"epoch": 0.008108979365306197,
|
8546 |
+
"grad_norm": 0.7388715147972107,
|
8547 |
+
"learning_rate": 1.6909476017240912e-06,
|
8548 |
+
"loss": 0.8403,
|
8549 |
+
"step": 1220
|
8550 |
+
},
|
8551 |
+
{
|
8552 |
+
"epoch": 0.008115626069703988,
|
8553 |
+
"grad_norm": 0.7469501495361328,
|
8554 |
+
"learning_rate": 1.6605133927530825e-06,
|
8555 |
+
"loss": 0.7552,
|
8556 |
+
"step": 1221
|
8557 |
+
},
|
8558 |
+
{
|
8559 |
+
"epoch": 0.00812227277410178,
|
8560 |
+
"grad_norm": 0.8448594212532043,
|
8561 |
+
"learning_rate": 1.6303509339486823e-06,
|
8562 |
+
"loss": 0.8702,
|
8563 |
+
"step": 1222
|
8564 |
+
},
|
8565 |
+
{
|
8566 |
+
"epoch": 0.008128919478499573,
|
8567 |
+
"grad_norm": 0.8358093500137329,
|
8568 |
+
"learning_rate": 1.6004603948752473e-06,
|
8569 |
+
"loss": 0.8922,
|
8570 |
+
"step": 1223
|
8571 |
+
},
|
8572 |
+
{
|
8573 |
+
"epoch": 0.008135566182897364,
|
8574 |
+
"grad_norm": 1.0381255149841309,
|
8575 |
+
"learning_rate": 1.5708419435684462e-06,
|
8576 |
+
"loss": 0.7692,
|
8577 |
+
"step": 1224
|
8578 |
+
},
|
8579 |
+
{
|
8580 |
+
"epoch": 0.008142212887295157,
|
8581 |
+
"grad_norm": 0.8204221129417419,
|
8582 |
+
"learning_rate": 1.5414957465343882e-06,
|
8583 |
+
"loss": 0.8635,
|
8584 |
+
"step": 1225
|
8585 |
+
},
|
8586 |
+
{
|
8587 |
+
"epoch": 0.00814885959169295,
|
8588 |
+
"grad_norm": 1.3811793327331543,
|
8589 |
+
"learning_rate": 1.512421968748623e-06,
|
8590 |
+
"loss": 1.0606,
|
8591 |
+
"step": 1226
|
8592 |
+
},
|
8593 |
+
{
|
8594 |
+
"epoch": 0.00815550629609074,
|
8595 |
+
"grad_norm": 0.7987362742424011,
|
8596 |
+
"learning_rate": 1.4836207736552642e-06,
|
8597 |
+
"loss": 0.8876,
|
8598 |
+
"step": 1227
|
8599 |
+
},
|
8600 |
+
{
|
8601 |
+
"epoch": 0.008162153000488533,
|
8602 |
+
"grad_norm": 1.040132761001587,
|
8603 |
+
"learning_rate": 1.455092323166024e-06,
|
8604 |
+
"loss": 0.8985,
|
8605 |
+
"step": 1228
|
8606 |
+
},
|
8607 |
+
{
|
8608 |
+
"epoch": 0.008168799704886326,
|
8609 |
+
"grad_norm": 0.9249858260154724,
|
8610 |
+
"learning_rate": 1.4268367776593405e-06,
|
8611 |
+
"loss": 0.8616,
|
8612 |
+
"step": 1229
|
8613 |
+
},
|
8614 |
+
{
|
8615 |
+
"epoch": 0.008175446409284116,
|
8616 |
+
"grad_norm": 1.1802650690078735,
|
8617 |
+
"learning_rate": 1.3988542959794627e-06,
|
8618 |
+
"loss": 0.6785,
|
8619 |
+
"step": 1230
|
8620 |
+
},
|
8621 |
+
{
|
8622 |
+
"epoch": 0.008182093113681909,
|
8623 |
+
"grad_norm": 1.0302826166152954,
|
8624 |
+
"learning_rate": 1.3711450354355449e-06,
|
8625 |
+
"loss": 0.6362,
|
8626 |
+
"step": 1231
|
8627 |
+
},
|
8628 |
+
{
|
8629 |
+
"epoch": 0.0081887398180797,
|
8630 |
+
"grad_norm": 1.121963381767273,
|
8631 |
+
"learning_rate": 1.3437091518007816e-06,
|
8632 |
+
"loss": 1.023,
|
8633 |
+
"step": 1232
|
8634 |
+
},
|
8635 |
+
{
|
8636 |
+
"epoch": 0.008195386522477493,
|
8637 |
+
"grad_norm": 1.0118569135665894,
|
8638 |
+
"learning_rate": 1.3165467993115244e-06,
|
8639 |
+
"loss": 0.7098,
|
8640 |
+
"step": 1233
|
8641 |
+
},
|
8642 |
+
{
|
8643 |
+
"epoch": 0.008202033226875285,
|
8644 |
+
"grad_norm": 1.052306890487671,
|
8645 |
+
"learning_rate": 1.2896581306664047e-06,
|
8646 |
+
"loss": 0.7376,
|
8647 |
+
"step": 1234
|
8648 |
+
},
|
8649 |
+
{
|
8650 |
+
"epoch": 0.008208679931273076,
|
8651 |
+
"grad_norm": 1.0704679489135742,
|
8652 |
+
"learning_rate": 1.2630432970255013e-06,
|
8653 |
+
"loss": 1.0065,
|
8654 |
+
"step": 1235
|
8655 |
+
},
|
8656 |
+
{
|
8657 |
+
"epoch": 0.008215326635670869,
|
8658 |
+
"grad_norm": 1.216379165649414,
|
8659 |
+
"learning_rate": 1.2367024480094691e-06,
|
8660 |
+
"loss": 0.515,
|
8661 |
+
"step": 1236
|
8662 |
+
},
|
8663 |
+
{
|
8664 |
+
"epoch": 0.008221973340068661,
|
8665 |
+
"grad_norm": 1.3317914009094238,
|
8666 |
+
"learning_rate": 1.2106357316986838e-06,
|
8667 |
+
"loss": 0.7376,
|
8668 |
+
"step": 1237
|
8669 |
+
},
|
8670 |
+
{
|
8671 |
+
"epoch": 0.008228620044466452,
|
8672 |
+
"grad_norm": 1.7680779695510864,
|
8673 |
+
"learning_rate": 1.1848432946324594e-06,
|
8674 |
+
"loss": 0.9229,
|
8675 |
+
"step": 1238
|
8676 |
+
},
|
8677 |
+
{
|
8678 |
+
"epoch": 0.008235266748864245,
|
8679 |
+
"grad_norm": 1.3019485473632812,
|
8680 |
+
"learning_rate": 1.1593252818081658e-06,
|
8681 |
+
"loss": 1.0404,
|
8682 |
+
"step": 1239
|
8683 |
+
},
|
8684 |
+
{
|
8685 |
+
"epoch": 0.008241913453262036,
|
8686 |
+
"grad_norm": 1.3333247900009155,
|
8687 |
+
"learning_rate": 1.1340818366804729e-06,
|
8688 |
+
"loss": 0.6831,
|
8689 |
+
"step": 1240
|
8690 |
+
},
|
8691 |
+
{
|
8692 |
+
"epoch": 0.008248560157659828,
|
8693 |
+
"grad_norm": 1.3237649202346802,
|
8694 |
+
"learning_rate": 1.1091131011604804e-06,
|
8695 |
+
"loss": 0.7026,
|
8696 |
+
"step": 1241
|
8697 |
+
},
|
8698 |
+
{
|
8699 |
+
"epoch": 0.00825520686205762,
|
8700 |
+
"grad_norm": 1.5985360145568848,
|
8701 |
+
"learning_rate": 1.084419215614979e-06,
|
8702 |
+
"loss": 0.854,
|
8703 |
+
"step": 1242
|
8704 |
+
},
|
8705 |
+
{
|
8706 |
+
"epoch": 0.008261853566455412,
|
8707 |
+
"grad_norm": 1.1438730955123901,
|
8708 |
+
"learning_rate": 1.0600003188656117e-06,
|
8709 |
+
"loss": 0.5545,
|
8710 |
+
"step": 1243
|
8711 |
+
},
|
8712 |
+
{
|
8713 |
+
"epoch": 0.008268500270853204,
|
8714 |
+
"grad_norm": 1.2783243656158447,
|
8715 |
+
"learning_rate": 1.0358565481881356e-06,
|
8716 |
+
"loss": 0.4873,
|
8717 |
+
"step": 1244
|
8718 |
+
},
|
8719 |
+
{
|
8720 |
+
"epoch": 0.008275146975250997,
|
8721 |
+
"grad_norm": 1.4403537511825562,
|
8722 |
+
"learning_rate": 1.0119880393116176e-06,
|
8723 |
+
"loss": 0.7985,
|
8724 |
+
"step": 1245
|
8725 |
+
},
|
8726 |
+
{
|
8727 |
+
"epoch": 0.008281793679648788,
|
8728 |
+
"grad_norm": 1.6538810729980469,
|
8729 |
+
"learning_rate": 9.883949264176962e-07,
|
8730 |
+
"loss": 0.8637,
|
8731 |
+
"step": 1246
|
8732 |
+
},
|
8733 |
+
{
|
8734 |
+
"epoch": 0.00828844038404658,
|
8735 |
+
"grad_norm": 1.6638885736465454,
|
8736 |
+
"learning_rate": 9.65077342139814e-07,
|
8737 |
+
"loss": 0.7477,
|
8738 |
+
"step": 1247
|
8739 |
+
},
|
8740 |
+
{
|
8741 |
+
"epoch": 0.008295087088444371,
|
8742 |
+
"grad_norm": 1.9725651741027832,
|
8743 |
+
"learning_rate": 9.420354175624591e-07,
|
8744 |
+
"loss": 0.9379,
|
8745 |
+
"step": 1248
|
8746 |
+
},
|
8747 |
+
{
|
8748 |
+
"epoch": 0.008301733792842164,
|
8749 |
+
"grad_norm": 2.747512102127075,
|
8750 |
+
"learning_rate": 9.19269282220464e-07,
|
8751 |
+
"loss": 0.8132,
|
8752 |
+
"step": 1249
|
8753 |
+
},
|
8754 |
+
{
|
8755 |
+
"epoch": 0.008308380497239956,
|
8756 |
+
"grad_norm": 3.339989423751831,
|
8757 |
+
"learning_rate": 8.967790640982465e-07,
|
8758 |
+
"loss": 0.9292,
|
8759 |
+
"step": 1250
|
8760 |
+
},
|
8761 |
+
{
|
8762 |
+
"epoch": 0.008315027201637747,
|
8763 |
+
"grad_norm": 0.36648422479629517,
|
8764 |
+
"learning_rate": 8.745648896290981e-07,
|
8765 |
+
"loss": 0.9416,
|
8766 |
+
"step": 1251
|
8767 |
+
},
|
8768 |
+
{
|
8769 |
+
"epoch": 0.00832167390603554,
|
8770 |
+
"grad_norm": 0.429190456867218,
|
8771 |
+
"learning_rate": 8.52626883694474e-07,
|
8772 |
+
"loss": 0.9262,
|
8773 |
+
"step": 1252
|
8774 |
+
},
|
8775 |
+
{
|
8776 |
+
"epoch": 0.008328320610433333,
|
8777 |
+
"grad_norm": 0.6208512187004089,
|
8778 |
+
"learning_rate": 8.309651696233045e-07,
|
8779 |
+
"loss": 0.9098,
|
8780 |
+
"step": 1253
|
8781 |
+
},
|
8782 |
+
{
|
8783 |
+
"epoch": 0.008334967314831123,
|
8784 |
+
"grad_norm": 0.5772883892059326,
|
8785 |
+
"learning_rate": 8.095798691912737e-07,
|
8786 |
+
"loss": 0.8938,
|
8787 |
+
"step": 1254
|
8788 |
+
},
|
8789 |
+
{
|
8790 |
+
"epoch": 0.008341614019228916,
|
8791 |
+
"grad_norm": 0.6024505496025085,
|
8792 |
+
"learning_rate": 7.884711026201585e-07,
|
8793 |
+
"loss": 0.6757,
|
8794 |
+
"step": 1255
|
8795 |
+
},
|
8796 |
+
{
|
8797 |
+
"epoch": 0.008348260723626709,
|
8798 |
+
"grad_norm": 0.47945520281791687,
|
8799 |
+
"learning_rate": 7.676389885771518e-07,
|
8800 |
+
"loss": 0.769,
|
8801 |
+
"step": 1256
|
8802 |
+
},
|
8803 |
+
{
|
8804 |
+
"epoch": 0.0083549074280245,
|
8805 |
+
"grad_norm": 0.5948473215103149,
|
8806 |
+
"learning_rate": 7.470836441741736e-07,
|
8807 |
+
"loss": 0.8862,
|
8808 |
+
"step": 1257
|
8809 |
+
},
|
8810 |
+
{
|
8811 |
+
"epoch": 0.008361554132422292,
|
8812 |
+
"grad_norm": 0.5873749852180481,
|
8813 |
+
"learning_rate": 7.268051849672441e-07,
|
8814 |
+
"loss": 0.8728,
|
8815 |
+
"step": 1258
|
8816 |
+
},
|
8817 |
+
{
|
8818 |
+
"epoch": 0.008368200836820083,
|
8819 |
+
"grad_norm": 0.5595722794532776,
|
8820 |
+
"learning_rate": 7.068037249557957e-07,
|
8821 |
+
"loss": 0.8866,
|
8822 |
+
"step": 1259
|
8823 |
+
},
|
8824 |
+
{
|
8825 |
+
"epoch": 0.008374847541217876,
|
8826 |
+
"grad_norm": 0.6838152408599854,
|
8827 |
+
"learning_rate": 6.870793765820782e-07,
|
8828 |
+
"loss": 0.7628,
|
8829 |
+
"step": 1260
|
8830 |
+
},
|
8831 |
+
{
|
8832 |
+
"epoch": 0.008381494245615668,
|
8833 |
+
"grad_norm": 0.6558036804199219,
|
8834 |
+
"learning_rate": 6.676322507304877e-07,
|
8835 |
+
"loss": 0.9166,
|
8836 |
+
"step": 1261
|
8837 |
+
},
|
8838 |
+
{
|
8839 |
+
"epoch": 0.008388140950013459,
|
8840 |
+
"grad_norm": 0.6153919100761414,
|
8841 |
+
"learning_rate": 6.484624567269615e-07,
|
8842 |
+
"loss": 0.7852,
|
8843 |
+
"step": 1262
|
8844 |
+
},
|
8845 |
+
{
|
8846 |
+
"epoch": 0.008394787654411252,
|
8847 |
+
"grad_norm": 0.7325903177261353,
|
8848 |
+
"learning_rate": 6.29570102338356e-07,
|
8849 |
+
"loss": 0.9253,
|
8850 |
+
"step": 1263
|
8851 |
+
},
|
8852 |
+
{
|
8853 |
+
"epoch": 0.008401434358809044,
|
8854 |
+
"grad_norm": 0.7558137774467468,
|
8855 |
+
"learning_rate": 6.109552937718588e-07,
|
8856 |
+
"loss": 0.6489,
|
8857 |
+
"step": 1264
|
8858 |
+
},
|
8859 |
+
{
|
8860 |
+
"epoch": 0.008408081063206835,
|
8861 |
+
"grad_norm": 0.6297517418861389,
|
8862 |
+
"learning_rate": 5.92618135674361e-07,
|
8863 |
+
"loss": 0.9842,
|
8864 |
+
"step": 1265
|
8865 |
+
},
|
8866 |
+
{
|
8867 |
+
"epoch": 0.008414727767604628,
|
8868 |
+
"grad_norm": 0.7743880748748779,
|
8869 |
+
"learning_rate": 5.745587311318968e-07,
|
8870 |
+
"loss": 0.8271,
|
8871 |
+
"step": 1266
|
8872 |
+
},
|
8873 |
+
{
|
8874 |
+
"epoch": 0.008421374472002419,
|
8875 |
+
"grad_norm": 0.7147794365882874,
|
8876 |
+
"learning_rate": 5.567771816690381e-07,
|
8877 |
+
"loss": 0.9824,
|
8878 |
+
"step": 1267
|
8879 |
+
},
|
8880 |
+
{
|
8881 |
+
"epoch": 0.008428021176400211,
|
8882 |
+
"grad_norm": 0.831110954284668,
|
8883 |
+
"learning_rate": 5.392735872483623e-07,
|
8884 |
+
"loss": 0.8822,
|
8885 |
+
"step": 1268
|
8886 |
+
},
|
8887 |
+
{
|
8888 |
+
"epoch": 0.008434667880798004,
|
8889 |
+
"grad_norm": 0.7674936652183533,
|
8890 |
+
"learning_rate": 5.220480462698462e-07,
|
8891 |
+
"loss": 0.7066,
|
8892 |
+
"step": 1269
|
8893 |
+
},
|
8894 |
+
{
|
8895 |
+
"epoch": 0.008441314585195795,
|
8896 |
+
"grad_norm": 0.8405891060829163,
|
8897 |
+
"learning_rate": 5.051006555703453e-07,
|
8898 |
+
"loss": 0.7432,
|
8899 |
+
"step": 1270
|
8900 |
+
},
|
8901 |
+
{
|
8902 |
+
"epoch": 0.008447961289593587,
|
8903 |
+
"grad_norm": 0.8006777167320251,
|
8904 |
+
"learning_rate": 4.884315104230264e-07,
|
8905 |
+
"loss": 0.8399,
|
8906 |
+
"step": 1271
|
8907 |
+
},
|
8908 |
+
{
|
8909 |
+
"epoch": 0.00845460799399138,
|
8910 |
+
"grad_norm": 0.8176466226577759,
|
8911 |
+
"learning_rate": 4.7204070453685244e-07,
|
8912 |
+
"loss": 0.6448,
|
8913 |
+
"step": 1272
|
8914 |
+
},
|
8915 |
+
{
|
8916 |
+
"epoch": 0.00846125469838917,
|
8917 |
+
"grad_norm": 0.74027019739151,
|
8918 |
+
"learning_rate": 4.5592833005603796e-07,
|
8919 |
+
"loss": 0.8987,
|
8920 |
+
"step": 1273
|
8921 |
+
},
|
8922 |
+
{
|
8923 |
+
"epoch": 0.008467901402786963,
|
8924 |
+
"grad_norm": 0.9606915712356567,
|
8925 |
+
"learning_rate": 4.4009447755954944e-07,
|
8926 |
+
"loss": 1.0994,
|
8927 |
+
"step": 1274
|
8928 |
+
},
|
8929 |
+
{
|
8930 |
+
"epoch": 0.008474548107184754,
|
8931 |
+
"grad_norm": 1.3926945924758911,
|
8932 |
+
"learning_rate": 4.2453923606057265e-07,
|
8933 |
+
"loss": 0.7428,
|
8934 |
+
"step": 1275
|
8935 |
+
},
|
8936 |
+
{
|
8937 |
+
"epoch": 0.008481194811582547,
|
8938 |
+
"grad_norm": 0.9520140290260315,
|
8939 |
+
"learning_rate": 4.0926269300603503e-07,
|
8940 |
+
"loss": 0.9258,
|
8941 |
+
"step": 1276
|
8942 |
+
},
|
8943 |
+
{
|
8944 |
+
"epoch": 0.00848784151598034,
|
8945 |
+
"grad_norm": 0.8461837768554688,
|
8946 |
+
"learning_rate": 3.9426493427611177e-07,
|
8947 |
+
"loss": 0.7099,
|
8948 |
+
"step": 1277
|
8949 |
+
},
|
8950 |
+
{
|
8951 |
+
"epoch": 0.00849448822037813,
|
8952 |
+
"grad_norm": 0.9905888438224792,
|
8953 |
+
"learning_rate": 3.795460441837095e-07,
|
8954 |
+
"loss": 0.8662,
|
8955 |
+
"step": 1278
|
8956 |
+
},
|
8957 |
+
{
|
8958 |
+
"epoch": 0.008501134924775923,
|
8959 |
+
"grad_norm": 0.9792956113815308,
|
8960 |
+
"learning_rate": 3.651061054740501e-07,
|
8961 |
+
"loss": 0.8404,
|
8962 |
+
"step": 1279
|
8963 |
+
},
|
8964 |
+
{
|
8965 |
+
"epoch": 0.008507781629173716,
|
8966 |
+
"grad_norm": 1.0338367223739624,
|
8967 |
+
"learning_rate": 3.5094519932415417e-07,
|
8968 |
+
"loss": 0.7575,
|
8969 |
+
"step": 1280
|
8970 |
+
},
|
8971 |
+
{
|
8972 |
+
"epoch": 0.008514428333571506,
|
8973 |
+
"grad_norm": 1.1163334846496582,
|
8974 |
+
"learning_rate": 3.370634053424082e-07,
|
8975 |
+
"loss": 0.7396,
|
8976 |
+
"step": 1281
|
8977 |
+
},
|
8978 |
+
{
|
8979 |
+
"epoch": 0.008521075037969299,
|
8980 |
+
"grad_norm": 1.0261398553848267,
|
8981 |
+
"learning_rate": 3.234608015681151e-07,
|
8982 |
+
"loss": 0.7362,
|
8983 |
+
"step": 1282
|
8984 |
+
},
|
8985 |
+
{
|
8986 |
+
"epoch": 0.008527721742367092,
|
8987 |
+
"grad_norm": 0.927574872970581,
|
8988 |
+
"learning_rate": 3.1013746447104975e-07,
|
8989 |
+
"loss": 0.8644,
|
8990 |
+
"step": 1283
|
8991 |
+
},
|
8992 |
+
{
|
8993 |
+
"epoch": 0.008534368446764883,
|
8994 |
+
"grad_norm": 1.1198316812515259,
|
8995 |
+
"learning_rate": 2.970934689510485e-07,
|
8996 |
+
"loss": 0.7248,
|
8997 |
+
"step": 1284
|
8998 |
+
},
|
8999 |
+
{
|
9000 |
+
"epoch": 0.008541015151162675,
|
9001 |
+
"grad_norm": 1.0062837600708008,
|
9002 |
+
"learning_rate": 2.843288883375539e-07,
|
9003 |
+
"loss": 0.7071,
|
9004 |
+
"step": 1285
|
9005 |
+
},
|
9006 |
+
{
|
9007 |
+
"epoch": 0.008547661855560466,
|
9008 |
+
"grad_norm": 1.5024248361587524,
|
9009 |
+
"learning_rate": 2.71843794389226e-07,
|
9010 |
+
"loss": 0.7523,
|
9011 |
+
"step": 1286
|
9012 |
+
},
|
9013 |
+
{
|
9014 |
+
"epoch": 0.008554308559958259,
|
9015 |
+
"grad_norm": 1.3791800737380981,
|
9016 |
+
"learning_rate": 2.596382572935374e-07,
|
9017 |
+
"loss": 0.7903,
|
9018 |
+
"step": 1287
|
9019 |
+
},
|
9020 |
+
{
|
9021 |
+
"epoch": 0.008560955264356051,
|
9022 |
+
"grad_norm": 1.1591501235961914,
|
9023 |
+
"learning_rate": 2.477123456663788e-07,
|
9024 |
+
"loss": 0.7717,
|
9025 |
+
"step": 1288
|
9026 |
+
},
|
9027 |
+
{
|
9028 |
+
"epoch": 0.008567601968753842,
|
9029 |
+
"grad_norm": 1.3647840023040771,
|
9030 |
+
"learning_rate": 2.3606612655166504e-07,
|
9031 |
+
"loss": 0.7711,
|
9032 |
+
"step": 1289
|
9033 |
+
},
|
9034 |
+
{
|
9035 |
+
"epoch": 0.008574248673151635,
|
9036 |
+
"grad_norm": 1.3846338987350464,
|
9037 |
+
"learning_rate": 2.2469966542096322e-07,
|
9038 |
+
"loss": 0.95,
|
9039 |
+
"step": 1290
|
9040 |
+
},
|
9041 |
+
{
|
9042 |
+
"epoch": 0.008580895377549427,
|
9043 |
+
"grad_norm": 1.1583385467529297,
|
9044 |
+
"learning_rate": 2.1361302617312619e-07,
|
9045 |
+
"loss": 0.5463,
|
9046 |
+
"step": 1291
|
9047 |
+
},
|
9048 |
+
{
|
9049 |
+
"epoch": 0.008587542081947218,
|
9050 |
+
"grad_norm": 1.452928900718689,
|
9051 |
+
"learning_rate": 2.028062711339318e-07,
|
9052 |
+
"loss": 0.8323,
|
9053 |
+
"step": 1292
|
9054 |
+
},
|
9055 |
+
{
|
9056 |
+
"epoch": 0.00859418878634501,
|
9057 |
+
"grad_norm": 1.6725733280181885,
|
9058 |
+
"learning_rate": 1.922794610557277e-07,
|
9059 |
+
"loss": 0.9138,
|
9060 |
+
"step": 1293
|
9061 |
+
},
|
9062 |
+
{
|
9063 |
+
"epoch": 0.008600835490742802,
|
9064 |
+
"grad_norm": 1.342020034790039,
|
9065 |
+
"learning_rate": 1.8203265511710365e-07,
|
9066 |
+
"loss": 0.7097,
|
9067 |
+
"step": 1294
|
9068 |
+
},
|
9069 |
+
{
|
9070 |
+
"epoch": 0.008607482195140594,
|
9071 |
+
"grad_norm": 1.666353464126587,
|
9072 |
+
"learning_rate": 1.7206591092253642e-07,
|
9073 |
+
"loss": 0.8405,
|
9074 |
+
"step": 1295
|
9075 |
+
},
|
9076 |
+
{
|
9077 |
+
"epoch": 0.008614128899538387,
|
9078 |
+
"grad_norm": 2.023453712463379,
|
9079 |
+
"learning_rate": 1.623792845020955e-07,
|
9080 |
+
"loss": 0.9479,
|
9081 |
+
"step": 1296
|
9082 |
+
},
|
9083 |
+
{
|
9084 |
+
"epoch": 0.008620775603936178,
|
9085 |
+
"grad_norm": 2.3681578636169434,
|
9086 |
+
"learning_rate": 1.529728303110989e-07,
|
9087 |
+
"loss": 0.8534,
|
9088 |
+
"step": 1297
|
9089 |
+
},
|
9090 |
+
{
|
9091 |
+
"epoch": 0.00862742230833397,
|
9092 |
+
"grad_norm": 1.3655821084976196,
|
9093 |
+
"learning_rate": 1.4384660122983007e-07,
|
9094 |
+
"loss": 0.6382,
|
9095 |
+
"step": 1298
|
9096 |
+
},
|
9097 |
+
{
|
9098 |
+
"epoch": 0.008634069012731763,
|
9099 |
+
"grad_norm": 2.0608959197998047,
|
9100 |
+
"learning_rate": 1.3500064856321603e-07,
|
9101 |
+
"loss": 0.7469,
|
9102 |
+
"step": 1299
|
9103 |
+
},
|
9104 |
+
{
|
9105 |
+
"epoch": 0.008640715717129554,
|
9106 |
+
"grad_norm": 3.251729726791382,
|
9107 |
+
"learning_rate": 1.264350220405719e-07,
|
9108 |
+
"loss": 1.0694,
|
9109 |
+
"step": 1300
|
9110 |
+
},
|
9111 |
+
{
|
9112 |
+
"epoch": 0.008647362421527346,
|
9113 |
+
"grad_norm": 0.3500362038612366,
|
9114 |
+
"learning_rate": 1.1814976981529002e-07,
|
9115 |
+
"loss": 1.0851,
|
9116 |
+
"step": 1301
|
9117 |
+
},
|
9118 |
+
{
|
9119 |
+
"epoch": 0.008654009125925137,
|
9120 |
+
"grad_norm": 0.7794168591499329,
|
9121 |
+
"learning_rate": 1.1014493846457919e-07,
|
9122 |
+
"loss": 1.0559,
|
9123 |
+
"step": 1302
|
9124 |
+
},
|
9125 |
+
{
|
9126 |
+
"epoch": 0.00866065583032293,
|
9127 |
+
"grad_norm": 0.49550482630729675,
|
9128 |
+
"learning_rate": 1.0242057298922581e-07,
|
9129 |
+
"loss": 0.791,
|
9130 |
+
"step": 1303
|
9131 |
+
},
|
9132 |
+
{
|
9133 |
+
"epoch": 0.008667302534720723,
|
9134 |
+
"grad_norm": 0.5143431425094604,
|
9135 |
+
"learning_rate": 9.497671681329423e-08,
|
9136 |
+
"loss": 0.7753,
|
9137 |
+
"step": 1304
|
9138 |
+
},
|
9139 |
+
{
|
9140 |
+
"epoch": 0.008673949239118513,
|
9141 |
+
"grad_norm": 0.49252620339393616,
|
9142 |
+
"learning_rate": 8.781341178393244e-08,
|
9143 |
+
"loss": 0.6776,
|
9144 |
+
"step": 1305
|
9145 |
+
},
|
9146 |
+
{
|
9147 |
+
"epoch": 0.008680595943516306,
|
9148 |
+
"grad_norm": 0.4753943681716919,
|
9149 |
+
"learning_rate": 8.093069817109445e-08,
|
9150 |
+
"loss": 0.9467,
|
9151 |
+
"step": 1306
|
9152 |
+
},
|
9153 |
+
{
|
9154 |
+
"epoch": 0.008687242647914099,
|
9155 |
+
"grad_norm": 0.5471718311309814,
|
9156 |
+
"learning_rate": 7.432861466734608e-08,
|
9157 |
+
"loss": 0.9961,
|
9158 |
+
"step": 1307
|
9159 |
+
},
|
9160 |
+
{
|
9161 |
+
"epoch": 0.00869388935231189,
|
9162 |
+
"grad_norm": 0.5803403258323669,
|
9163 |
+
"learning_rate": 6.800719838763182e-08,
|
9164 |
+
"loss": 0.7374,
|
9165 |
+
"step": 1308
|
9166 |
+
},
|
9167 |
+
{
|
9168 |
+
"epoch": 0.008700536056709682,
|
9169 |
+
"grad_norm": 0.7243188619613647,
|
9170 |
+
"learning_rate": 6.196648486906375e-08,
|
9171 |
+
"loss": 0.7798,
|
9172 |
+
"step": 1309
|
9173 |
+
},
|
9174 |
+
{
|
9175 |
+
"epoch": 0.008707182761107475,
|
9176 |
+
"grad_norm": 0.8324836492538452,
|
9177 |
+
"learning_rate": 5.620650807073857e-08,
|
9178 |
+
"loss": 1.129,
|
9179 |
+
"step": 1310
|
9180 |
+
},
|
9181 |
+
{
|
9182 |
+
"epoch": 0.008713829465505266,
|
9183 |
+
"grad_norm": 0.7195192575454712,
|
9184 |
+
"learning_rate": 5.072730037351536e-08,
|
9185 |
+
"loss": 0.8921,
|
9186 |
+
"step": 1311
|
9187 |
+
},
|
9188 |
+
{
|
9189 |
+
"epoch": 0.008720476169903058,
|
9190 |
+
"grad_norm": 0.6834086775779724,
|
9191 |
+
"learning_rate": 4.552889257987136e-08,
|
9192 |
+
"loss": 0.7839,
|
9193 |
+
"step": 1312
|
9194 |
+
},
|
9195 |
+
{
|
9196 |
+
"epoch": 0.008727122874300849,
|
9197 |
+
"grad_norm": 0.7733263373374939,
|
9198 |
+
"learning_rate": 4.06113139137021e-08,
|
9199 |
+
"loss": 0.8451,
|
9200 |
+
"step": 1313
|
9201 |
+
},
|
9202 |
+
{
|
9203 |
+
"epoch": 0.008733769578698642,
|
9204 |
+
"grad_norm": 0.7354872226715088,
|
9205 |
+
"learning_rate": 3.5974592020165954e-08,
|
9206 |
+
"loss": 0.8272,
|
9207 |
+
"step": 1314
|
9208 |
+
},
|
9209 |
+
{
|
9210 |
+
"epoch": 0.008740416283096434,
|
9211 |
+
"grad_norm": 0.646259069442749,
|
9212 |
+
"learning_rate": 3.161875296553429e-08,
|
9213 |
+
"loss": 0.7925,
|
9214 |
+
"step": 1315
|
9215 |
+
},
|
9216 |
+
{
|
9217 |
+
"epoch": 0.008747062987494225,
|
9218 |
+
"grad_norm": 0.7745428085327148,
|
9219 |
+
"learning_rate": 2.7543821237030475e-08,
|
9220 |
+
"loss": 0.9565,
|
9221 |
+
"step": 1316
|
9222 |
+
},
|
9223 |
+
{
|
9224 |
+
"epoch": 0.008753709691892018,
|
9225 |
+
"grad_norm": 0.6693745851516724,
|
9226 |
+
"learning_rate": 2.3749819742702185e-08,
|
9227 |
+
"loss": 0.7318,
|
9228 |
+
"step": 1317
|
9229 |
+
},
|
9230 |
+
{
|
9231 |
+
"epoch": 0.00876035639628981,
|
9232 |
+
"grad_norm": 0.731195330619812,
|
9233 |
+
"learning_rate": 2.0236769811299294e-08,
|
9234 |
+
"loss": 0.7627,
|
9235 |
+
"step": 1318
|
9236 |
+
},
|
9237 |
+
{
|
9238 |
+
"epoch": 0.008767003100687601,
|
9239 |
+
"grad_norm": 0.7857605814933777,
|
9240 |
+
"learning_rate": 1.7004691192135104e-08,
|
9241 |
+
"loss": 0.8653,
|
9242 |
+
"step": 1319
|
9243 |
+
},
|
9244 |
+
{
|
9245 |
+
"epoch": 0.008773649805085394,
|
9246 |
+
"grad_norm": 0.7077126502990723,
|
9247 |
+
"learning_rate": 1.4053602054991955e-08,
|
9248 |
+
"loss": 0.8137,
|
9249 |
+
"step": 1320
|
9250 |
+
},
|
9251 |
+
{
|
9252 |
+
"epoch": 0.008780296509483185,
|
9253 |
+
"grad_norm": 0.7648908495903015,
|
9254 |
+
"learning_rate": 1.1383518990015773e-08,
|
9255 |
+
"loss": 0.741,
|
9256 |
+
"step": 1321
|
9257 |
+
},
|
9258 |
+
{
|
9259 |
+
"epoch": 0.008786943213880977,
|
9260 |
+
"grad_norm": 0.9375500679016113,
|
9261 |
+
"learning_rate": 8.99445700761059e-09,
|
9262 |
+
"loss": 0.9017,
|
9263 |
+
"step": 1322
|
9264 |
+
},
|
9265 |
+
{
|
9266 |
+
"epoch": 0.00879358991827877,
|
9267 |
+
"grad_norm": 0.8353835940361023,
|
9268 |
+
"learning_rate": 6.886429538377481e-09,
|
9269 |
+
"loss": 0.7284,
|
9270 |
+
"step": 1323
|
9271 |
+
},
|
9272 |
+
{
|
9273 |
+
"epoch": 0.00880023662267656,
|
9274 |
+
"grad_norm": 1.0171085596084595,
|
9275 |
+
"learning_rate": 5.0594484330090955e-09,
|
9276 |
+
"loss": 0.7838,
|
9277 |
+
"step": 1324
|
9278 |
+
},
|
9279 |
+
{
|
9280 |
+
"epoch": 0.008806883327074353,
|
9281 |
+
"grad_norm": 0.786939263343811,
|
9282 |
+
"learning_rate": 3.513523962256349e-09,
|
9283 |
+
"loss": 0.8177,
|
9284 |
+
"step": 1325
|
9285 |
+
},
|
9286 |
+
{
|
9287 |
+
"epoch": 0.008813530031472146,
|
9288 |
+
"grad_norm": 0.8693004250526428,
|
9289 |
+
"learning_rate": 2.2486648168396075e-09,
|
9290 |
+
"loss": 0.8491,
|
9291 |
+
"step": 1326
|
9292 |
+
},
|
9293 |
+
{
|
9294 |
+
"epoch": 0.008820176735869937,
|
9295 |
+
"grad_norm": 0.9159846305847168,
|
9296 |
+
"learning_rate": 1.2648781074209304e-09,
|
9297 |
+
"loss": 0.7735,
|
9298 |
+
"step": 1327
|
9299 |
+
},
|
9300 |
+
{
|
9301 |
+
"epoch": 0.00882682344026773,
|
9302 |
+
"grad_norm": 1.1475635766983032,
|
9303 |
+
"learning_rate": 5.621693645541104e-10,
|
9304 |
+
"loss": 0.6395,
|
9305 |
+
"step": 1328
|
9306 |
+
},
|
9307 |
+
{
|
9308 |
+
"epoch": 0.00883347014466552,
|
9309 |
+
"grad_norm": 0.9265028238296509,
|
9310 |
+
"learning_rate": 1.4054253866246925e-10,
|
9311 |
+
"loss": 0.8467,
|
9312 |
+
"step": 1329
|
9313 |
+
},
|
9314 |
+
{
|
9315 |
+
"epoch": 0.008840116849063313,
|
9316 |
+
"grad_norm": 0.9815409779548645,
|
9317 |
+
"learning_rate": 0.0,
|
9318 |
+
"loss": 0.793,
|
9319 |
+
"step": 1330
|
9320 |
}
|
9321 |
],
|
9322 |
"logging_steps": 1,
|
|
|
9331 |
"should_evaluate": false,
|
9332 |
"should_log": false,
|
9333 |
"should_save": true,
|
9334 |
+
"should_training_stop": true
|
9335 |
},
|
9336 |
"attributes": {}
|
9337 |
}
|
9338 |
},
|
9339 |
+
"total_flos": 3.701942326080307e+17,
|
9340 |
"train_batch_size": 4,
|
9341 |
"trial_name": null,
|
9342 |
"trial_params": null
|