Commit
•
1203cbb
1
Parent(s):
2809865
Training in progress, step 18216, checkpoint
Browse files
last-checkpoint/model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 500389884
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:da8378f58afdcbb61da5aab016c87b360ba3b97ce5895a57c14de5629f844194
|
| 3 |
size 500389884
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1000900218
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d95db74dc8b3eeeb846cb4c4e9addbe7241924481ae96c22081d519fcd26cbfc
|
| 3 |
size 1000900218
|
last-checkpoint/rng_state_0.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14512
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:12f61f7e51cc603cdd137a87be15b659d00af22f66c4e5d529fd92a34ec1a2fc
|
| 3 |
size 14512
|
last-checkpoint/rng_state_1.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14512
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a04bf168981df42daaf3298864d910e1018e869d302c5844d03dbfe8eb4c3e75
|
| 3 |
size 14512
|
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 sha256:0d863b1e8dd7aadad275d56a5330082a7af839c16da6482be7ffc6e9ccd5d7a6
|
| 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,
|
|
@@ -28343,6 +28343,3548 @@
|
|
| 28343 |
"learning_rate": 1.0007905528929295e-05,
|
| 28344 |
"loss": 5.4911,
|
| 28345 |
"step": 16192
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 28346 |
}
|
| 28347 |
],
|
| 28348 |
"logging_steps": 4,
|
|
@@ -28350,7 +31892,7 @@
|
|
| 28350 |
"num_input_tokens_seen": 0,
|
| 28351 |
"num_train_epochs": 1,
|
| 28352 |
"save_steps": 2024,
|
| 28353 |
-
"total_flos":
|
| 28354 |
"train_batch_size": 8,
|
| 28355 |
"trial_name": null,
|
| 28356 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.9000444686002272,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 18216,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 28343 |
"learning_rate": 1.0007905528929295e-05,
|
| 28344 |
"loss": 5.4911,
|
| 28345 |
"step": 16192
|
| 28346 |
+
},
|
| 28347 |
+
{
|
| 28348 |
+
"epoch": 0.8,
|
| 28349 |
+
"grad_norm": 2.1323323249816895,
|
| 28350 |
+
"learning_rate": 9.998023617767677e-06,
|
| 28351 |
+
"loss": 5.5276,
|
| 28352 |
+
"step": 16196
|
| 28353 |
+
},
|
| 28354 |
+
{
|
| 28355 |
+
"epoch": 0.8,
|
| 28356 |
+
"grad_norm": 2.277160167694092,
|
| 28357 |
+
"learning_rate": 9.988141706606058e-06,
|
| 28358 |
+
"loss": 5.4718,
|
| 28359 |
+
"step": 16200
|
| 28360 |
+
},
|
| 28361 |
+
{
|
| 28362 |
+
"epoch": 0.8,
|
| 28363 |
+
"grad_norm": 1.8769733905792236,
|
| 28364 |
+
"learning_rate": 9.97825979544444e-06,
|
| 28365 |
+
"loss": 5.3622,
|
| 28366 |
+
"step": 16204
|
| 28367 |
+
},
|
| 28368 |
+
{
|
| 28369 |
+
"epoch": 0.8,
|
| 28370 |
+
"grad_norm": 1.8731465339660645,
|
| 28371 |
+
"learning_rate": 9.96837788428282e-06,
|
| 28372 |
+
"loss": 5.5215,
|
| 28373 |
+
"step": 16208
|
| 28374 |
+
},
|
| 28375 |
+
{
|
| 28376 |
+
"epoch": 0.8,
|
| 28377 |
+
"grad_norm": 1.8808999061584473,
|
| 28378 |
+
"learning_rate": 9.958495973121202e-06,
|
| 28379 |
+
"loss": 5.3953,
|
| 28380 |
+
"step": 16212
|
| 28381 |
+
},
|
| 28382 |
+
{
|
| 28383 |
+
"epoch": 0.8,
|
| 28384 |
+
"grad_norm": 1.8657152652740479,
|
| 28385 |
+
"learning_rate": 9.948614061959584e-06,
|
| 28386 |
+
"loss": 5.3316,
|
| 28387 |
+
"step": 16216
|
| 28388 |
+
},
|
| 28389 |
+
{
|
| 28390 |
+
"epoch": 0.8,
|
| 28391 |
+
"grad_norm": 1.849173665046692,
|
| 28392 |
+
"learning_rate": 9.938732150797965e-06,
|
| 28393 |
+
"loss": 5.4471,
|
| 28394 |
+
"step": 16220
|
| 28395 |
+
},
|
| 28396 |
+
{
|
| 28397 |
+
"epoch": 0.8,
|
| 28398 |
+
"grad_norm": 2.220717668533325,
|
| 28399 |
+
"learning_rate": 9.928850239636347e-06,
|
| 28400 |
+
"loss": 5.5204,
|
| 28401 |
+
"step": 16224
|
| 28402 |
+
},
|
| 28403 |
+
{
|
| 28404 |
+
"epoch": 0.8,
|
| 28405 |
+
"grad_norm": 2.0210342407226562,
|
| 28406 |
+
"learning_rate": 9.918968328474727e-06,
|
| 28407 |
+
"loss": 5.3364,
|
| 28408 |
+
"step": 16228
|
| 28409 |
+
},
|
| 28410 |
+
{
|
| 28411 |
+
"epoch": 0.8,
|
| 28412 |
+
"grad_norm": 1.9695372581481934,
|
| 28413 |
+
"learning_rate": 9.909086417313109e-06,
|
| 28414 |
+
"loss": 5.485,
|
| 28415 |
+
"step": 16232
|
| 28416 |
+
},
|
| 28417 |
+
{
|
| 28418 |
+
"epoch": 0.8,
|
| 28419 |
+
"grad_norm": 1.875001311302185,
|
| 28420 |
+
"learning_rate": 9.89920450615149e-06,
|
| 28421 |
+
"loss": 5.3977,
|
| 28422 |
+
"step": 16236
|
| 28423 |
+
},
|
| 28424 |
+
{
|
| 28425 |
+
"epoch": 0.8,
|
| 28426 |
+
"grad_norm": 2.136852741241455,
|
| 28427 |
+
"learning_rate": 9.889322594989871e-06,
|
| 28428 |
+
"loss": 5.5784,
|
| 28429 |
+
"step": 16240
|
| 28430 |
+
},
|
| 28431 |
+
{
|
| 28432 |
+
"epoch": 0.8,
|
| 28433 |
+
"grad_norm": 1.8972970247268677,
|
| 28434 |
+
"learning_rate": 9.879440683828253e-06,
|
| 28435 |
+
"loss": 5.4973,
|
| 28436 |
+
"step": 16244
|
| 28437 |
+
},
|
| 28438 |
+
{
|
| 28439 |
+
"epoch": 0.8,
|
| 28440 |
+
"grad_norm": 2.14034366607666,
|
| 28441 |
+
"learning_rate": 9.869558772666634e-06,
|
| 28442 |
+
"loss": 5.546,
|
| 28443 |
+
"step": 16248
|
| 28444 |
+
},
|
| 28445 |
+
{
|
| 28446 |
+
"epoch": 0.8,
|
| 28447 |
+
"grad_norm": 2.0028293132781982,
|
| 28448 |
+
"learning_rate": 9.859676861505016e-06,
|
| 28449 |
+
"loss": 5.3869,
|
| 28450 |
+
"step": 16252
|
| 28451 |
+
},
|
| 28452 |
+
{
|
| 28453 |
+
"epoch": 0.8,
|
| 28454 |
+
"grad_norm": 1.9485490322113037,
|
| 28455 |
+
"learning_rate": 9.849794950343396e-06,
|
| 28456 |
+
"loss": 5.4852,
|
| 28457 |
+
"step": 16256
|
| 28458 |
+
},
|
| 28459 |
+
{
|
| 28460 |
+
"epoch": 0.8,
|
| 28461 |
+
"grad_norm": 1.9260238409042358,
|
| 28462 |
+
"learning_rate": 9.839913039181778e-06,
|
| 28463 |
+
"loss": 5.473,
|
| 28464 |
+
"step": 16260
|
| 28465 |
+
},
|
| 28466 |
+
{
|
| 28467 |
+
"epoch": 0.8,
|
| 28468 |
+
"grad_norm": 2.1630001068115234,
|
| 28469 |
+
"learning_rate": 9.830031128020159e-06,
|
| 28470 |
+
"loss": 5.4207,
|
| 28471 |
+
"step": 16264
|
| 28472 |
+
},
|
| 28473 |
+
{
|
| 28474 |
+
"epoch": 0.8,
|
| 28475 |
+
"grad_norm": 1.9328508377075195,
|
| 28476 |
+
"learning_rate": 9.82014921685854e-06,
|
| 28477 |
+
"loss": 5.4433,
|
| 28478 |
+
"step": 16268
|
| 28479 |
+
},
|
| 28480 |
+
{
|
| 28481 |
+
"epoch": 0.8,
|
| 28482 |
+
"grad_norm": 1.9832642078399658,
|
| 28483 |
+
"learning_rate": 9.810267305696923e-06,
|
| 28484 |
+
"loss": 5.4387,
|
| 28485 |
+
"step": 16272
|
| 28486 |
+
},
|
| 28487 |
+
{
|
| 28488 |
+
"epoch": 0.8,
|
| 28489 |
+
"grad_norm": 2.0616111755371094,
|
| 28490 |
+
"learning_rate": 9.800385394535303e-06,
|
| 28491 |
+
"loss": 5.3895,
|
| 28492 |
+
"step": 16276
|
| 28493 |
+
},
|
| 28494 |
+
{
|
| 28495 |
+
"epoch": 0.8,
|
| 28496 |
+
"grad_norm": 2.1709606647491455,
|
| 28497 |
+
"learning_rate": 9.790503483373685e-06,
|
| 28498 |
+
"loss": 5.5639,
|
| 28499 |
+
"step": 16280
|
| 28500 |
+
},
|
| 28501 |
+
{
|
| 28502 |
+
"epoch": 0.8,
|
| 28503 |
+
"grad_norm": 2.1553335189819336,
|
| 28504 |
+
"learning_rate": 9.780621572212066e-06,
|
| 28505 |
+
"loss": 5.5419,
|
| 28506 |
+
"step": 16284
|
| 28507 |
+
},
|
| 28508 |
+
{
|
| 28509 |
+
"epoch": 0.8,
|
| 28510 |
+
"grad_norm": 1.7869793176651,
|
| 28511 |
+
"learning_rate": 9.770739661050448e-06,
|
| 28512 |
+
"loss": 5.4425,
|
| 28513 |
+
"step": 16288
|
| 28514 |
+
},
|
| 28515 |
+
{
|
| 28516 |
+
"epoch": 0.8,
|
| 28517 |
+
"grad_norm": 1.902388095855713,
|
| 28518 |
+
"learning_rate": 9.760857749888828e-06,
|
| 28519 |
+
"loss": 5.3651,
|
| 28520 |
+
"step": 16292
|
| 28521 |
+
},
|
| 28522 |
+
{
|
| 28523 |
+
"epoch": 0.81,
|
| 28524 |
+
"grad_norm": 2.135723829269409,
|
| 28525 |
+
"learning_rate": 9.750975838727212e-06,
|
| 28526 |
+
"loss": 5.5838,
|
| 28527 |
+
"step": 16296
|
| 28528 |
+
},
|
| 28529 |
+
{
|
| 28530 |
+
"epoch": 0.81,
|
| 28531 |
+
"grad_norm": 2.3749701976776123,
|
| 28532 |
+
"learning_rate": 9.741093927565592e-06,
|
| 28533 |
+
"loss": 5.5562,
|
| 28534 |
+
"step": 16300
|
| 28535 |
+
},
|
| 28536 |
+
{
|
| 28537 |
+
"epoch": 0.81,
|
| 28538 |
+
"grad_norm": 1.9421731233596802,
|
| 28539 |
+
"learning_rate": 9.731212016403973e-06,
|
| 28540 |
+
"loss": 5.432,
|
| 28541 |
+
"step": 16304
|
| 28542 |
+
},
|
| 28543 |
+
{
|
| 28544 |
+
"epoch": 0.81,
|
| 28545 |
+
"grad_norm": 1.87296462059021,
|
| 28546 |
+
"learning_rate": 9.721330105242355e-06,
|
| 28547 |
+
"loss": 5.5429,
|
| 28548 |
+
"step": 16308
|
| 28549 |
+
},
|
| 28550 |
+
{
|
| 28551 |
+
"epoch": 0.81,
|
| 28552 |
+
"grad_norm": 2.0159077644348145,
|
| 28553 |
+
"learning_rate": 9.711448194080735e-06,
|
| 28554 |
+
"loss": 5.5088,
|
| 28555 |
+
"step": 16312
|
| 28556 |
+
},
|
| 28557 |
+
{
|
| 28558 |
+
"epoch": 0.81,
|
| 28559 |
+
"grad_norm": 1.9539657831192017,
|
| 28560 |
+
"learning_rate": 9.701566282919117e-06,
|
| 28561 |
+
"loss": 5.49,
|
| 28562 |
+
"step": 16316
|
| 28563 |
+
},
|
| 28564 |
+
{
|
| 28565 |
+
"epoch": 0.81,
|
| 28566 |
+
"grad_norm": 2.0840256214141846,
|
| 28567 |
+
"learning_rate": 9.691684371757497e-06,
|
| 28568 |
+
"loss": 5.4972,
|
| 28569 |
+
"step": 16320
|
| 28570 |
+
},
|
| 28571 |
+
{
|
| 28572 |
+
"epoch": 0.81,
|
| 28573 |
+
"grad_norm": 2.0817222595214844,
|
| 28574 |
+
"learning_rate": 9.68180246059588e-06,
|
| 28575 |
+
"loss": 5.5185,
|
| 28576 |
+
"step": 16324
|
| 28577 |
+
},
|
| 28578 |
+
{
|
| 28579 |
+
"epoch": 0.81,
|
| 28580 |
+
"grad_norm": 2.172551393508911,
|
| 28581 |
+
"learning_rate": 9.671920549434262e-06,
|
| 28582 |
+
"loss": 5.5465,
|
| 28583 |
+
"step": 16328
|
| 28584 |
+
},
|
| 28585 |
+
{
|
| 28586 |
+
"epoch": 0.81,
|
| 28587 |
+
"grad_norm": 1.8779146671295166,
|
| 28588 |
+
"learning_rate": 9.662038638272644e-06,
|
| 28589 |
+
"loss": 5.5047,
|
| 28590 |
+
"step": 16332
|
| 28591 |
+
},
|
| 28592 |
+
{
|
| 28593 |
+
"epoch": 0.81,
|
| 28594 |
+
"grad_norm": 1.9017332792282104,
|
| 28595 |
+
"learning_rate": 9.652156727111024e-06,
|
| 28596 |
+
"loss": 5.4705,
|
| 28597 |
+
"step": 16336
|
| 28598 |
+
},
|
| 28599 |
+
{
|
| 28600 |
+
"epoch": 0.81,
|
| 28601 |
+
"grad_norm": 2.0198209285736084,
|
| 28602 |
+
"learning_rate": 9.642274815949404e-06,
|
| 28603 |
+
"loss": 5.4886,
|
| 28604 |
+
"step": 16340
|
| 28605 |
+
},
|
| 28606 |
+
{
|
| 28607 |
+
"epoch": 0.81,
|
| 28608 |
+
"grad_norm": 2.0686516761779785,
|
| 28609 |
+
"learning_rate": 9.632392904787786e-06,
|
| 28610 |
+
"loss": 5.5188,
|
| 28611 |
+
"step": 16344
|
| 28612 |
+
},
|
| 28613 |
+
{
|
| 28614 |
+
"epoch": 0.81,
|
| 28615 |
+
"grad_norm": 1.962902545928955,
|
| 28616 |
+
"learning_rate": 9.622510993626167e-06,
|
| 28617 |
+
"loss": 5.5414,
|
| 28618 |
+
"step": 16348
|
| 28619 |
+
},
|
| 28620 |
+
{
|
| 28621 |
+
"epoch": 0.81,
|
| 28622 |
+
"grad_norm": 2.0159122943878174,
|
| 28623 |
+
"learning_rate": 9.612629082464549e-06,
|
| 28624 |
+
"loss": 5.422,
|
| 28625 |
+
"step": 16352
|
| 28626 |
+
},
|
| 28627 |
+
{
|
| 28628 |
+
"epoch": 0.81,
|
| 28629 |
+
"grad_norm": 2.2383811473846436,
|
| 28630 |
+
"learning_rate": 9.60274717130293e-06,
|
| 28631 |
+
"loss": 5.5384,
|
| 28632 |
+
"step": 16356
|
| 28633 |
+
},
|
| 28634 |
+
{
|
| 28635 |
+
"epoch": 0.81,
|
| 28636 |
+
"grad_norm": 1.993220567703247,
|
| 28637 |
+
"learning_rate": 9.592865260141313e-06,
|
| 28638 |
+
"loss": 5.3551,
|
| 28639 |
+
"step": 16360
|
| 28640 |
+
},
|
| 28641 |
+
{
|
| 28642 |
+
"epoch": 0.81,
|
| 28643 |
+
"grad_norm": 2.130995035171509,
|
| 28644 |
+
"learning_rate": 9.582983348979693e-06,
|
| 28645 |
+
"loss": 5.4926,
|
| 28646 |
+
"step": 16364
|
| 28647 |
+
},
|
| 28648 |
+
{
|
| 28649 |
+
"epoch": 0.81,
|
| 28650 |
+
"grad_norm": 1.9019581079483032,
|
| 28651 |
+
"learning_rate": 9.573101437818074e-06,
|
| 28652 |
+
"loss": 5.5623,
|
| 28653 |
+
"step": 16368
|
| 28654 |
+
},
|
| 28655 |
+
{
|
| 28656 |
+
"epoch": 0.81,
|
| 28657 |
+
"grad_norm": 2.1175732612609863,
|
| 28658 |
+
"learning_rate": 9.563219526656456e-06,
|
| 28659 |
+
"loss": 5.4529,
|
| 28660 |
+
"step": 16372
|
| 28661 |
+
},
|
| 28662 |
+
{
|
| 28663 |
+
"epoch": 0.81,
|
| 28664 |
+
"grad_norm": 1.8680050373077393,
|
| 28665 |
+
"learning_rate": 9.553337615494836e-06,
|
| 28666 |
+
"loss": 5.4273,
|
| 28667 |
+
"step": 16376
|
| 28668 |
+
},
|
| 28669 |
+
{
|
| 28670 |
+
"epoch": 0.81,
|
| 28671 |
+
"grad_norm": 2.0988821983337402,
|
| 28672 |
+
"learning_rate": 9.543455704333218e-06,
|
| 28673 |
+
"loss": 5.477,
|
| 28674 |
+
"step": 16380
|
| 28675 |
+
},
|
| 28676 |
+
{
|
| 28677 |
+
"epoch": 0.81,
|
| 28678 |
+
"grad_norm": 2.198651075363159,
|
| 28679 |
+
"learning_rate": 9.5335737931716e-06,
|
| 28680 |
+
"loss": 5.5427,
|
| 28681 |
+
"step": 16384
|
| 28682 |
+
},
|
| 28683 |
+
{
|
| 28684 |
+
"epoch": 0.81,
|
| 28685 |
+
"grad_norm": 2.1054868698120117,
|
| 28686 |
+
"learning_rate": 9.523691882009982e-06,
|
| 28687 |
+
"loss": 5.4173,
|
| 28688 |
+
"step": 16388
|
| 28689 |
+
},
|
| 28690 |
+
{
|
| 28691 |
+
"epoch": 0.81,
|
| 28692 |
+
"grad_norm": 2.0211637020111084,
|
| 28693 |
+
"learning_rate": 9.513809970848363e-06,
|
| 28694 |
+
"loss": 5.6035,
|
| 28695 |
+
"step": 16392
|
| 28696 |
+
},
|
| 28697 |
+
{
|
| 28698 |
+
"epoch": 0.81,
|
| 28699 |
+
"grad_norm": 2.0547540187835693,
|
| 28700 |
+
"learning_rate": 9.503928059686745e-06,
|
| 28701 |
+
"loss": 5.4983,
|
| 28702 |
+
"step": 16396
|
| 28703 |
+
},
|
| 28704 |
+
{
|
| 28705 |
+
"epoch": 0.81,
|
| 28706 |
+
"grad_norm": 1.9957647323608398,
|
| 28707 |
+
"learning_rate": 9.494046148525125e-06,
|
| 28708 |
+
"loss": 5.5374,
|
| 28709 |
+
"step": 16400
|
| 28710 |
+
},
|
| 28711 |
+
{
|
| 28712 |
+
"epoch": 0.81,
|
| 28713 |
+
"grad_norm": 2.0535998344421387,
|
| 28714 |
+
"learning_rate": 9.484164237363505e-06,
|
| 28715 |
+
"loss": 5.5643,
|
| 28716 |
+
"step": 16404
|
| 28717 |
+
},
|
| 28718 |
+
{
|
| 28719 |
+
"epoch": 0.81,
|
| 28720 |
+
"grad_norm": 2.1046228408813477,
|
| 28721 |
+
"learning_rate": 9.474282326201887e-06,
|
| 28722 |
+
"loss": 5.4727,
|
| 28723 |
+
"step": 16408
|
| 28724 |
+
},
|
| 28725 |
+
{
|
| 28726 |
+
"epoch": 0.81,
|
| 28727 |
+
"grad_norm": 2.1698873043060303,
|
| 28728 |
+
"learning_rate": 9.46440041504027e-06,
|
| 28729 |
+
"loss": 5.5637,
|
| 28730 |
+
"step": 16412
|
| 28731 |
+
},
|
| 28732 |
+
{
|
| 28733 |
+
"epoch": 0.81,
|
| 28734 |
+
"grad_norm": 1.9448730945587158,
|
| 28735 |
+
"learning_rate": 9.454518503878652e-06,
|
| 28736 |
+
"loss": 5.4932,
|
| 28737 |
+
"step": 16416
|
| 28738 |
+
},
|
| 28739 |
+
{
|
| 28740 |
+
"epoch": 0.81,
|
| 28741 |
+
"grad_norm": 2.0663201808929443,
|
| 28742 |
+
"learning_rate": 9.444636592717032e-06,
|
| 28743 |
+
"loss": 5.5067,
|
| 28744 |
+
"step": 16420
|
| 28745 |
+
},
|
| 28746 |
+
{
|
| 28747 |
+
"epoch": 0.81,
|
| 28748 |
+
"grad_norm": 1.8125361204147339,
|
| 28749 |
+
"learning_rate": 9.434754681555414e-06,
|
| 28750 |
+
"loss": 5.5381,
|
| 28751 |
+
"step": 16424
|
| 28752 |
+
},
|
| 28753 |
+
{
|
| 28754 |
+
"epoch": 0.81,
|
| 28755 |
+
"grad_norm": 2.077420473098755,
|
| 28756 |
+
"learning_rate": 9.424872770393794e-06,
|
| 28757 |
+
"loss": 5.5642,
|
| 28758 |
+
"step": 16428
|
| 28759 |
+
},
|
| 28760 |
+
{
|
| 28761 |
+
"epoch": 0.81,
|
| 28762 |
+
"grad_norm": 1.9312140941619873,
|
| 28763 |
+
"learning_rate": 9.414990859232176e-06,
|
| 28764 |
+
"loss": 5.3661,
|
| 28765 |
+
"step": 16432
|
| 28766 |
+
},
|
| 28767 |
+
{
|
| 28768 |
+
"epoch": 0.81,
|
| 28769 |
+
"grad_norm": 1.8529763221740723,
|
| 28770 |
+
"learning_rate": 9.405108948070557e-06,
|
| 28771 |
+
"loss": 5.4776,
|
| 28772 |
+
"step": 16436
|
| 28773 |
+
},
|
| 28774 |
+
{
|
| 28775 |
+
"epoch": 0.81,
|
| 28776 |
+
"grad_norm": 1.8764407634735107,
|
| 28777 |
+
"learning_rate": 9.395227036908939e-06,
|
| 28778 |
+
"loss": 5.4928,
|
| 28779 |
+
"step": 16440
|
| 28780 |
+
},
|
| 28781 |
+
{
|
| 28782 |
+
"epoch": 0.81,
|
| 28783 |
+
"grad_norm": 2.139594316482544,
|
| 28784 |
+
"learning_rate": 9.385345125747321e-06,
|
| 28785 |
+
"loss": 5.4108,
|
| 28786 |
+
"step": 16444
|
| 28787 |
+
},
|
| 28788 |
+
{
|
| 28789 |
+
"epoch": 0.81,
|
| 28790 |
+
"grad_norm": 2.1105079650878906,
|
| 28791 |
+
"learning_rate": 9.375463214585701e-06,
|
| 28792 |
+
"loss": 5.4876,
|
| 28793 |
+
"step": 16448
|
| 28794 |
+
},
|
| 28795 |
+
{
|
| 28796 |
+
"epoch": 0.81,
|
| 28797 |
+
"grad_norm": 1.9484140872955322,
|
| 28798 |
+
"learning_rate": 9.365581303424083e-06,
|
| 28799 |
+
"loss": 5.4148,
|
| 28800 |
+
"step": 16452
|
| 28801 |
+
},
|
| 28802 |
+
{
|
| 28803 |
+
"epoch": 0.81,
|
| 28804 |
+
"grad_norm": 2.0571391582489014,
|
| 28805 |
+
"learning_rate": 9.355699392262464e-06,
|
| 28806 |
+
"loss": 5.6282,
|
| 28807 |
+
"step": 16456
|
| 28808 |
+
},
|
| 28809 |
+
{
|
| 28810 |
+
"epoch": 0.81,
|
| 28811 |
+
"grad_norm": 1.7625576257705688,
|
| 28812 |
+
"learning_rate": 9.345817481100846e-06,
|
| 28813 |
+
"loss": 5.5073,
|
| 28814 |
+
"step": 16460
|
| 28815 |
+
},
|
| 28816 |
+
{
|
| 28817 |
+
"epoch": 0.81,
|
| 28818 |
+
"grad_norm": 2.1183722019195557,
|
| 28819 |
+
"learning_rate": 9.335935569939226e-06,
|
| 28820 |
+
"loss": 5.4659,
|
| 28821 |
+
"step": 16464
|
| 28822 |
+
},
|
| 28823 |
+
{
|
| 28824 |
+
"epoch": 0.81,
|
| 28825 |
+
"grad_norm": 2.135255813598633,
|
| 28826 |
+
"learning_rate": 9.326053658777608e-06,
|
| 28827 |
+
"loss": 5.4262,
|
| 28828 |
+
"step": 16468
|
| 28829 |
+
},
|
| 28830 |
+
{
|
| 28831 |
+
"epoch": 0.81,
|
| 28832 |
+
"grad_norm": 1.7497916221618652,
|
| 28833 |
+
"learning_rate": 9.31617174761599e-06,
|
| 28834 |
+
"loss": 5.537,
|
| 28835 |
+
"step": 16472
|
| 28836 |
+
},
|
| 28837 |
+
{
|
| 28838 |
+
"epoch": 0.81,
|
| 28839 |
+
"grad_norm": 1.9297901391983032,
|
| 28840 |
+
"learning_rate": 9.30628983645437e-06,
|
| 28841 |
+
"loss": 5.4848,
|
| 28842 |
+
"step": 16476
|
| 28843 |
+
},
|
| 28844 |
+
{
|
| 28845 |
+
"epoch": 0.81,
|
| 28846 |
+
"grad_norm": 1.992133617401123,
|
| 28847 |
+
"learning_rate": 9.296407925292753e-06,
|
| 28848 |
+
"loss": 5.5307,
|
| 28849 |
+
"step": 16480
|
| 28850 |
+
},
|
| 28851 |
+
{
|
| 28852 |
+
"epoch": 0.81,
|
| 28853 |
+
"grad_norm": 1.8346421718597412,
|
| 28854 |
+
"learning_rate": 9.286526014131133e-06,
|
| 28855 |
+
"loss": 5.4071,
|
| 28856 |
+
"step": 16484
|
| 28857 |
+
},
|
| 28858 |
+
{
|
| 28859 |
+
"epoch": 0.81,
|
| 28860 |
+
"grad_norm": 1.9083247184753418,
|
| 28861 |
+
"learning_rate": 9.276644102969515e-06,
|
| 28862 |
+
"loss": 5.4248,
|
| 28863 |
+
"step": 16488
|
| 28864 |
+
},
|
| 28865 |
+
{
|
| 28866 |
+
"epoch": 0.81,
|
| 28867 |
+
"grad_norm": 2.0954606533050537,
|
| 28868 |
+
"learning_rate": 9.266762191807895e-06,
|
| 28869 |
+
"loss": 5.5175,
|
| 28870 |
+
"step": 16492
|
| 28871 |
+
},
|
| 28872 |
+
{
|
| 28873 |
+
"epoch": 0.82,
|
| 28874 |
+
"grad_norm": 2.101158857345581,
|
| 28875 |
+
"learning_rate": 9.256880280646278e-06,
|
| 28876 |
+
"loss": 5.5345,
|
| 28877 |
+
"step": 16496
|
| 28878 |
+
},
|
| 28879 |
+
{
|
| 28880 |
+
"epoch": 0.82,
|
| 28881 |
+
"grad_norm": 2.170283794403076,
|
| 28882 |
+
"learning_rate": 9.24699836948466e-06,
|
| 28883 |
+
"loss": 5.5719,
|
| 28884 |
+
"step": 16500
|
| 28885 |
+
},
|
| 28886 |
+
{
|
| 28887 |
+
"epoch": 0.82,
|
| 28888 |
+
"grad_norm": 2.078697681427002,
|
| 28889 |
+
"learning_rate": 9.23711645832304e-06,
|
| 28890 |
+
"loss": 5.4199,
|
| 28891 |
+
"step": 16504
|
| 28892 |
+
},
|
| 28893 |
+
{
|
| 28894 |
+
"epoch": 0.82,
|
| 28895 |
+
"grad_norm": 2.1254682540893555,
|
| 28896 |
+
"learning_rate": 9.227234547161422e-06,
|
| 28897 |
+
"loss": 5.3647,
|
| 28898 |
+
"step": 16508
|
| 28899 |
+
},
|
| 28900 |
+
{
|
| 28901 |
+
"epoch": 0.82,
|
| 28902 |
+
"grad_norm": 2.2468132972717285,
|
| 28903 |
+
"learning_rate": 9.217352635999802e-06,
|
| 28904 |
+
"loss": 5.4673,
|
| 28905 |
+
"step": 16512
|
| 28906 |
+
},
|
| 28907 |
+
{
|
| 28908 |
+
"epoch": 0.82,
|
| 28909 |
+
"grad_norm": 2.053579330444336,
|
| 28910 |
+
"learning_rate": 9.207470724838184e-06,
|
| 28911 |
+
"loss": 5.4083,
|
| 28912 |
+
"step": 16516
|
| 28913 |
+
},
|
| 28914 |
+
{
|
| 28915 |
+
"epoch": 0.82,
|
| 28916 |
+
"grad_norm": 2.2088301181793213,
|
| 28917 |
+
"learning_rate": 9.197588813676565e-06,
|
| 28918 |
+
"loss": 5.5694,
|
| 28919 |
+
"step": 16520
|
| 28920 |
+
},
|
| 28921 |
+
{
|
| 28922 |
+
"epoch": 0.82,
|
| 28923 |
+
"grad_norm": 1.9974719285964966,
|
| 28924 |
+
"learning_rate": 9.187706902514947e-06,
|
| 28925 |
+
"loss": 5.4396,
|
| 28926 |
+
"step": 16524
|
| 28927 |
+
},
|
| 28928 |
+
{
|
| 28929 |
+
"epoch": 0.82,
|
| 28930 |
+
"grad_norm": 2.066420555114746,
|
| 28931 |
+
"learning_rate": 9.177824991353329e-06,
|
| 28932 |
+
"loss": 5.5418,
|
| 28933 |
+
"step": 16528
|
| 28934 |
+
},
|
| 28935 |
+
{
|
| 28936 |
+
"epoch": 0.82,
|
| 28937 |
+
"grad_norm": 2.2263917922973633,
|
| 28938 |
+
"learning_rate": 9.16794308019171e-06,
|
| 28939 |
+
"loss": 5.4904,
|
| 28940 |
+
"step": 16532
|
| 28941 |
+
},
|
| 28942 |
+
{
|
| 28943 |
+
"epoch": 0.82,
|
| 28944 |
+
"grad_norm": 2.0956313610076904,
|
| 28945 |
+
"learning_rate": 9.158061169030091e-06,
|
| 28946 |
+
"loss": 5.47,
|
| 28947 |
+
"step": 16536
|
| 28948 |
+
},
|
| 28949 |
+
{
|
| 28950 |
+
"epoch": 0.82,
|
| 28951 |
+
"grad_norm": 2.120701551437378,
|
| 28952 |
+
"learning_rate": 9.148179257868472e-06,
|
| 28953 |
+
"loss": 5.4498,
|
| 28954 |
+
"step": 16540
|
| 28955 |
+
},
|
| 28956 |
+
{
|
| 28957 |
+
"epoch": 0.82,
|
| 28958 |
+
"grad_norm": 2.068040609359741,
|
| 28959 |
+
"learning_rate": 9.138297346706854e-06,
|
| 28960 |
+
"loss": 5.5373,
|
| 28961 |
+
"step": 16544
|
| 28962 |
+
},
|
| 28963 |
+
{
|
| 28964 |
+
"epoch": 0.82,
|
| 28965 |
+
"grad_norm": 2.125322103500366,
|
| 28966 |
+
"learning_rate": 9.128415435545234e-06,
|
| 28967 |
+
"loss": 5.4782,
|
| 28968 |
+
"step": 16548
|
| 28969 |
+
},
|
| 28970 |
+
{
|
| 28971 |
+
"epoch": 0.82,
|
| 28972 |
+
"grad_norm": 1.8984034061431885,
|
| 28973 |
+
"learning_rate": 9.118533524383616e-06,
|
| 28974 |
+
"loss": 5.5614,
|
| 28975 |
+
"step": 16552
|
| 28976 |
+
},
|
| 28977 |
+
{
|
| 28978 |
+
"epoch": 0.82,
|
| 28979 |
+
"grad_norm": 2.038201332092285,
|
| 28980 |
+
"learning_rate": 9.108651613221998e-06,
|
| 28981 |
+
"loss": 5.5308,
|
| 28982 |
+
"step": 16556
|
| 28983 |
+
},
|
| 28984 |
+
{
|
| 28985 |
+
"epoch": 0.82,
|
| 28986 |
+
"grad_norm": 2.083704948425293,
|
| 28987 |
+
"learning_rate": 9.098769702060379e-06,
|
| 28988 |
+
"loss": 5.3986,
|
| 28989 |
+
"step": 16560
|
| 28990 |
+
},
|
| 28991 |
+
{
|
| 28992 |
+
"epoch": 0.82,
|
| 28993 |
+
"grad_norm": 2.2222559452056885,
|
| 28994 |
+
"learning_rate": 9.08888779089876e-06,
|
| 28995 |
+
"loss": 5.4164,
|
| 28996 |
+
"step": 16564
|
| 28997 |
+
},
|
| 28998 |
+
{
|
| 28999 |
+
"epoch": 0.82,
|
| 29000 |
+
"grad_norm": 2.319937229156494,
|
| 29001 |
+
"learning_rate": 9.079005879737141e-06,
|
| 29002 |
+
"loss": 5.5564,
|
| 29003 |
+
"step": 16568
|
| 29004 |
+
},
|
| 29005 |
+
{
|
| 29006 |
+
"epoch": 0.82,
|
| 29007 |
+
"grad_norm": 1.930305004119873,
|
| 29008 |
+
"learning_rate": 9.069123968575523e-06,
|
| 29009 |
+
"loss": 5.5181,
|
| 29010 |
+
"step": 16572
|
| 29011 |
+
},
|
| 29012 |
+
{
|
| 29013 |
+
"epoch": 0.82,
|
| 29014 |
+
"grad_norm": 2.175090789794922,
|
| 29015 |
+
"learning_rate": 9.059242057413904e-06,
|
| 29016 |
+
"loss": 5.4816,
|
| 29017 |
+
"step": 16576
|
| 29018 |
+
},
|
| 29019 |
+
{
|
| 29020 |
+
"epoch": 0.82,
|
| 29021 |
+
"grad_norm": 1.9041398763656616,
|
| 29022 |
+
"learning_rate": 9.049360146252286e-06,
|
| 29023 |
+
"loss": 5.4462,
|
| 29024 |
+
"step": 16580
|
| 29025 |
+
},
|
| 29026 |
+
{
|
| 29027 |
+
"epoch": 0.82,
|
| 29028 |
+
"grad_norm": 1.92721426486969,
|
| 29029 |
+
"learning_rate": 9.039478235090668e-06,
|
| 29030 |
+
"loss": 5.4993,
|
| 29031 |
+
"step": 16584
|
| 29032 |
+
},
|
| 29033 |
+
{
|
| 29034 |
+
"epoch": 0.82,
|
| 29035 |
+
"grad_norm": 2.098320484161377,
|
| 29036 |
+
"learning_rate": 9.029596323929048e-06,
|
| 29037 |
+
"loss": 5.4334,
|
| 29038 |
+
"step": 16588
|
| 29039 |
+
},
|
| 29040 |
+
{
|
| 29041 |
+
"epoch": 0.82,
|
| 29042 |
+
"grad_norm": 1.988296389579773,
|
| 29043 |
+
"learning_rate": 9.01971441276743e-06,
|
| 29044 |
+
"loss": 5.5241,
|
| 29045 |
+
"step": 16592
|
| 29046 |
+
},
|
| 29047 |
+
{
|
| 29048 |
+
"epoch": 0.82,
|
| 29049 |
+
"grad_norm": 1.9492675065994263,
|
| 29050 |
+
"learning_rate": 9.00983250160581e-06,
|
| 29051 |
+
"loss": 5.4024,
|
| 29052 |
+
"step": 16596
|
| 29053 |
+
},
|
| 29054 |
+
{
|
| 29055 |
+
"epoch": 0.82,
|
| 29056 |
+
"grad_norm": 2.0381886959075928,
|
| 29057 |
+
"learning_rate": 8.999950590444192e-06,
|
| 29058 |
+
"loss": 5.4861,
|
| 29059 |
+
"step": 16600
|
| 29060 |
+
},
|
| 29061 |
+
{
|
| 29062 |
+
"epoch": 0.82,
|
| 29063 |
+
"grad_norm": 2.2015647888183594,
|
| 29064 |
+
"learning_rate": 8.990068679282573e-06,
|
| 29065 |
+
"loss": 5.4589,
|
| 29066 |
+
"step": 16604
|
| 29067 |
+
},
|
| 29068 |
+
{
|
| 29069 |
+
"epoch": 0.82,
|
| 29070 |
+
"grad_norm": 1.970509648323059,
|
| 29071 |
+
"learning_rate": 8.980186768120955e-06,
|
| 29072 |
+
"loss": 5.5072,
|
| 29073 |
+
"step": 16608
|
| 29074 |
+
},
|
| 29075 |
+
{
|
| 29076 |
+
"epoch": 0.82,
|
| 29077 |
+
"grad_norm": 2.048265218734741,
|
| 29078 |
+
"learning_rate": 8.970304856959337e-06,
|
| 29079 |
+
"loss": 5.4781,
|
| 29080 |
+
"step": 16612
|
| 29081 |
+
},
|
| 29082 |
+
{
|
| 29083 |
+
"epoch": 0.82,
|
| 29084 |
+
"grad_norm": 2.1781177520751953,
|
| 29085 |
+
"learning_rate": 8.960422945797719e-06,
|
| 29086 |
+
"loss": 5.4382,
|
| 29087 |
+
"step": 16616
|
| 29088 |
+
},
|
| 29089 |
+
{
|
| 29090 |
+
"epoch": 0.82,
|
| 29091 |
+
"grad_norm": 2.3206918239593506,
|
| 29092 |
+
"learning_rate": 8.9505410346361e-06,
|
| 29093 |
+
"loss": 5.5371,
|
| 29094 |
+
"step": 16620
|
| 29095 |
+
},
|
| 29096 |
+
{
|
| 29097 |
+
"epoch": 0.82,
|
| 29098 |
+
"grad_norm": 2.129166603088379,
|
| 29099 |
+
"learning_rate": 8.94065912347448e-06,
|
| 29100 |
+
"loss": 5.4773,
|
| 29101 |
+
"step": 16624
|
| 29102 |
+
},
|
| 29103 |
+
{
|
| 29104 |
+
"epoch": 0.82,
|
| 29105 |
+
"grad_norm": 1.8786160945892334,
|
| 29106 |
+
"learning_rate": 8.930777212312862e-06,
|
| 29107 |
+
"loss": 5.3599,
|
| 29108 |
+
"step": 16628
|
| 29109 |
+
},
|
| 29110 |
+
{
|
| 29111 |
+
"epoch": 0.82,
|
| 29112 |
+
"grad_norm": 2.114015579223633,
|
| 29113 |
+
"learning_rate": 8.920895301151242e-06,
|
| 29114 |
+
"loss": 5.5262,
|
| 29115 |
+
"step": 16632
|
| 29116 |
+
},
|
| 29117 |
+
{
|
| 29118 |
+
"epoch": 0.82,
|
| 29119 |
+
"grad_norm": 1.9730268716812134,
|
| 29120 |
+
"learning_rate": 8.911013389989624e-06,
|
| 29121 |
+
"loss": 5.3921,
|
| 29122 |
+
"step": 16636
|
| 29123 |
+
},
|
| 29124 |
+
{
|
| 29125 |
+
"epoch": 0.82,
|
| 29126 |
+
"grad_norm": 2.007050037384033,
|
| 29127 |
+
"learning_rate": 8.901131478828005e-06,
|
| 29128 |
+
"loss": 5.4521,
|
| 29129 |
+
"step": 16640
|
| 29130 |
+
},
|
| 29131 |
+
{
|
| 29132 |
+
"epoch": 0.82,
|
| 29133 |
+
"grad_norm": 1.9782698154449463,
|
| 29134 |
+
"learning_rate": 8.891249567666388e-06,
|
| 29135 |
+
"loss": 5.4024,
|
| 29136 |
+
"step": 16644
|
| 29137 |
+
},
|
| 29138 |
+
{
|
| 29139 |
+
"epoch": 0.82,
|
| 29140 |
+
"grad_norm": 2.118109941482544,
|
| 29141 |
+
"learning_rate": 8.881367656504769e-06,
|
| 29142 |
+
"loss": 5.4549,
|
| 29143 |
+
"step": 16648
|
| 29144 |
+
},
|
| 29145 |
+
{
|
| 29146 |
+
"epoch": 0.82,
|
| 29147 |
+
"grad_norm": 1.9269788265228271,
|
| 29148 |
+
"learning_rate": 8.871485745343149e-06,
|
| 29149 |
+
"loss": 5.3879,
|
| 29150 |
+
"step": 16652
|
| 29151 |
+
},
|
| 29152 |
+
{
|
| 29153 |
+
"epoch": 0.82,
|
| 29154 |
+
"grad_norm": 2.0356998443603516,
|
| 29155 |
+
"learning_rate": 8.861603834181531e-06,
|
| 29156 |
+
"loss": 5.4729,
|
| 29157 |
+
"step": 16656
|
| 29158 |
+
},
|
| 29159 |
+
{
|
| 29160 |
+
"epoch": 0.82,
|
| 29161 |
+
"grad_norm": 1.9945244789123535,
|
| 29162 |
+
"learning_rate": 8.851721923019912e-06,
|
| 29163 |
+
"loss": 5.489,
|
| 29164 |
+
"step": 16660
|
| 29165 |
+
},
|
| 29166 |
+
{
|
| 29167 |
+
"epoch": 0.82,
|
| 29168 |
+
"grad_norm": 2.1839029788970947,
|
| 29169 |
+
"learning_rate": 8.841840011858294e-06,
|
| 29170 |
+
"loss": 5.3626,
|
| 29171 |
+
"step": 16664
|
| 29172 |
+
},
|
| 29173 |
+
{
|
| 29174 |
+
"epoch": 0.82,
|
| 29175 |
+
"grad_norm": 1.9056282043457031,
|
| 29176 |
+
"learning_rate": 8.831958100696674e-06,
|
| 29177 |
+
"loss": 5.5155,
|
| 29178 |
+
"step": 16668
|
| 29179 |
+
},
|
| 29180 |
+
{
|
| 29181 |
+
"epoch": 0.82,
|
| 29182 |
+
"grad_norm": 1.971134066581726,
|
| 29183 |
+
"learning_rate": 8.822076189535058e-06,
|
| 29184 |
+
"loss": 5.4582,
|
| 29185 |
+
"step": 16672
|
| 29186 |
+
},
|
| 29187 |
+
{
|
| 29188 |
+
"epoch": 0.82,
|
| 29189 |
+
"grad_norm": 2.2608683109283447,
|
| 29190 |
+
"learning_rate": 8.812194278373438e-06,
|
| 29191 |
+
"loss": 5.5671,
|
| 29192 |
+
"step": 16676
|
| 29193 |
+
},
|
| 29194 |
+
{
|
| 29195 |
+
"epoch": 0.82,
|
| 29196 |
+
"grad_norm": 1.8873885869979858,
|
| 29197 |
+
"learning_rate": 8.80231236721182e-06,
|
| 29198 |
+
"loss": 5.5378,
|
| 29199 |
+
"step": 16680
|
| 29200 |
+
},
|
| 29201 |
+
{
|
| 29202 |
+
"epoch": 0.82,
|
| 29203 |
+
"grad_norm": 2.1268815994262695,
|
| 29204 |
+
"learning_rate": 8.7924304560502e-06,
|
| 29205 |
+
"loss": 5.5464,
|
| 29206 |
+
"step": 16684
|
| 29207 |
+
},
|
| 29208 |
+
{
|
| 29209 |
+
"epoch": 0.82,
|
| 29210 |
+
"grad_norm": 2.120333671569824,
|
| 29211 |
+
"learning_rate": 8.782548544888581e-06,
|
| 29212 |
+
"loss": 5.4566,
|
| 29213 |
+
"step": 16688
|
| 29214 |
+
},
|
| 29215 |
+
{
|
| 29216 |
+
"epoch": 0.82,
|
| 29217 |
+
"grad_norm": 2.202099323272705,
|
| 29218 |
+
"learning_rate": 8.772666633726963e-06,
|
| 29219 |
+
"loss": 5.3333,
|
| 29220 |
+
"step": 16692
|
| 29221 |
+
},
|
| 29222 |
+
{
|
| 29223 |
+
"epoch": 0.82,
|
| 29224 |
+
"grad_norm": 1.8824645280838013,
|
| 29225 |
+
"learning_rate": 8.762784722565343e-06,
|
| 29226 |
+
"loss": 5.5016,
|
| 29227 |
+
"step": 16696
|
| 29228 |
+
},
|
| 29229 |
+
{
|
| 29230 |
+
"epoch": 0.83,
|
| 29231 |
+
"grad_norm": 2.0147457122802734,
|
| 29232 |
+
"learning_rate": 8.752902811403727e-06,
|
| 29233 |
+
"loss": 5.4586,
|
| 29234 |
+
"step": 16700
|
| 29235 |
+
},
|
| 29236 |
+
{
|
| 29237 |
+
"epoch": 0.83,
|
| 29238 |
+
"grad_norm": 2.041895627975464,
|
| 29239 |
+
"learning_rate": 8.743020900242107e-06,
|
| 29240 |
+
"loss": 5.5022,
|
| 29241 |
+
"step": 16704
|
| 29242 |
+
},
|
| 29243 |
+
{
|
| 29244 |
+
"epoch": 0.83,
|
| 29245 |
+
"grad_norm": 2.077690362930298,
|
| 29246 |
+
"learning_rate": 8.73313898908049e-06,
|
| 29247 |
+
"loss": 5.5781,
|
| 29248 |
+
"step": 16708
|
| 29249 |
+
},
|
| 29250 |
+
{
|
| 29251 |
+
"epoch": 0.83,
|
| 29252 |
+
"grad_norm": 1.9477964639663696,
|
| 29253 |
+
"learning_rate": 8.72325707791887e-06,
|
| 29254 |
+
"loss": 5.5584,
|
| 29255 |
+
"step": 16712
|
| 29256 |
+
},
|
| 29257 |
+
{
|
| 29258 |
+
"epoch": 0.83,
|
| 29259 |
+
"grad_norm": 1.8398356437683105,
|
| 29260 |
+
"learning_rate": 8.713375166757252e-06,
|
| 29261 |
+
"loss": 5.5032,
|
| 29262 |
+
"step": 16716
|
| 29263 |
+
},
|
| 29264 |
+
{
|
| 29265 |
+
"epoch": 0.83,
|
| 29266 |
+
"grad_norm": 1.9973992109298706,
|
| 29267 |
+
"learning_rate": 8.703493255595632e-06,
|
| 29268 |
+
"loss": 5.37,
|
| 29269 |
+
"step": 16720
|
| 29270 |
+
},
|
| 29271 |
+
{
|
| 29272 |
+
"epoch": 0.83,
|
| 29273 |
+
"grad_norm": 2.0176520347595215,
|
| 29274 |
+
"learning_rate": 8.693611344434013e-06,
|
| 29275 |
+
"loss": 5.5138,
|
| 29276 |
+
"step": 16724
|
| 29277 |
+
},
|
| 29278 |
+
{
|
| 29279 |
+
"epoch": 0.83,
|
| 29280 |
+
"grad_norm": 2.1837217807769775,
|
| 29281 |
+
"learning_rate": 8.683729433272396e-06,
|
| 29282 |
+
"loss": 5.5308,
|
| 29283 |
+
"step": 16728
|
| 29284 |
+
},
|
| 29285 |
+
{
|
| 29286 |
+
"epoch": 0.83,
|
| 29287 |
+
"grad_norm": 2.0156595706939697,
|
| 29288 |
+
"learning_rate": 8.673847522110777e-06,
|
| 29289 |
+
"loss": 5.5722,
|
| 29290 |
+
"step": 16732
|
| 29291 |
+
},
|
| 29292 |
+
{
|
| 29293 |
+
"epoch": 0.83,
|
| 29294 |
+
"grad_norm": 2.3676466941833496,
|
| 29295 |
+
"learning_rate": 8.663965610949159e-06,
|
| 29296 |
+
"loss": 5.5775,
|
| 29297 |
+
"step": 16736
|
| 29298 |
+
},
|
| 29299 |
+
{
|
| 29300 |
+
"epoch": 0.83,
|
| 29301 |
+
"grad_norm": 2.270716667175293,
|
| 29302 |
+
"learning_rate": 8.65408369978754e-06,
|
| 29303 |
+
"loss": 5.4107,
|
| 29304 |
+
"step": 16740
|
| 29305 |
+
},
|
| 29306 |
+
{
|
| 29307 |
+
"epoch": 0.83,
|
| 29308 |
+
"grad_norm": 1.9551721811294556,
|
| 29309 |
+
"learning_rate": 8.644201788625921e-06,
|
| 29310 |
+
"loss": 5.6022,
|
| 29311 |
+
"step": 16744
|
| 29312 |
+
},
|
| 29313 |
+
{
|
| 29314 |
+
"epoch": 0.83,
|
| 29315 |
+
"grad_norm": 1.9827896356582642,
|
| 29316 |
+
"learning_rate": 8.634319877464302e-06,
|
| 29317 |
+
"loss": 5.4923,
|
| 29318 |
+
"step": 16748
|
| 29319 |
+
},
|
| 29320 |
+
{
|
| 29321 |
+
"epoch": 0.83,
|
| 29322 |
+
"grad_norm": 2.1360533237457275,
|
| 29323 |
+
"learning_rate": 8.624437966302682e-06,
|
| 29324 |
+
"loss": 5.4328,
|
| 29325 |
+
"step": 16752
|
| 29326 |
+
},
|
| 29327 |
+
{
|
| 29328 |
+
"epoch": 0.83,
|
| 29329 |
+
"grad_norm": 2.094109296798706,
|
| 29330 |
+
"learning_rate": 8.614556055141066e-06,
|
| 29331 |
+
"loss": 5.416,
|
| 29332 |
+
"step": 16756
|
| 29333 |
+
},
|
| 29334 |
+
{
|
| 29335 |
+
"epoch": 0.83,
|
| 29336 |
+
"grad_norm": 1.9513869285583496,
|
| 29337 |
+
"learning_rate": 8.604674143979446e-06,
|
| 29338 |
+
"loss": 5.403,
|
| 29339 |
+
"step": 16760
|
| 29340 |
+
},
|
| 29341 |
+
{
|
| 29342 |
+
"epoch": 0.83,
|
| 29343 |
+
"grad_norm": 2.0020523071289062,
|
| 29344 |
+
"learning_rate": 8.594792232817828e-06,
|
| 29345 |
+
"loss": 5.4628,
|
| 29346 |
+
"step": 16764
|
| 29347 |
+
},
|
| 29348 |
+
{
|
| 29349 |
+
"epoch": 0.83,
|
| 29350 |
+
"grad_norm": 2.405801296234131,
|
| 29351 |
+
"learning_rate": 8.584910321656209e-06,
|
| 29352 |
+
"loss": 5.5799,
|
| 29353 |
+
"step": 16768
|
| 29354 |
+
},
|
| 29355 |
+
{
|
| 29356 |
+
"epoch": 0.83,
|
| 29357 |
+
"grad_norm": 2.114650249481201,
|
| 29358 |
+
"learning_rate": 8.57502841049459e-06,
|
| 29359 |
+
"loss": 5.4151,
|
| 29360 |
+
"step": 16772
|
| 29361 |
+
},
|
| 29362 |
+
{
|
| 29363 |
+
"epoch": 0.83,
|
| 29364 |
+
"grad_norm": 1.9269883632659912,
|
| 29365 |
+
"learning_rate": 8.565146499332971e-06,
|
| 29366 |
+
"loss": 5.4321,
|
| 29367 |
+
"step": 16776
|
| 29368 |
+
},
|
| 29369 |
+
{
|
| 29370 |
+
"epoch": 0.83,
|
| 29371 |
+
"grad_norm": 1.894822597503662,
|
| 29372 |
+
"learning_rate": 8.555264588171353e-06,
|
| 29373 |
+
"loss": 5.4419,
|
| 29374 |
+
"step": 16780
|
| 29375 |
+
},
|
| 29376 |
+
{
|
| 29377 |
+
"epoch": 0.83,
|
| 29378 |
+
"grad_norm": 2.0186209678649902,
|
| 29379 |
+
"learning_rate": 8.545382677009735e-06,
|
| 29380 |
+
"loss": 5.5657,
|
| 29381 |
+
"step": 16784
|
| 29382 |
+
},
|
| 29383 |
+
{
|
| 29384 |
+
"epoch": 0.83,
|
| 29385 |
+
"grad_norm": 1.920372486114502,
|
| 29386 |
+
"learning_rate": 8.535500765848115e-06,
|
| 29387 |
+
"loss": 5.5334,
|
| 29388 |
+
"step": 16788
|
| 29389 |
+
},
|
| 29390 |
+
{
|
| 29391 |
+
"epoch": 0.83,
|
| 29392 |
+
"grad_norm": 2.1055715084075928,
|
| 29393 |
+
"learning_rate": 8.525618854686497e-06,
|
| 29394 |
+
"loss": 5.4952,
|
| 29395 |
+
"step": 16792
|
| 29396 |
+
},
|
| 29397 |
+
{
|
| 29398 |
+
"epoch": 0.83,
|
| 29399 |
+
"grad_norm": 2.0137712955474854,
|
| 29400 |
+
"learning_rate": 8.515736943524878e-06,
|
| 29401 |
+
"loss": 5.5128,
|
| 29402 |
+
"step": 16796
|
| 29403 |
+
},
|
| 29404 |
+
{
|
| 29405 |
+
"epoch": 0.83,
|
| 29406 |
+
"grad_norm": 2.0473227500915527,
|
| 29407 |
+
"learning_rate": 8.50585503236326e-06,
|
| 29408 |
+
"loss": 5.6793,
|
| 29409 |
+
"step": 16800
|
| 29410 |
+
},
|
| 29411 |
+
{
|
| 29412 |
+
"epoch": 0.83,
|
| 29413 |
+
"grad_norm": 2.0757429599761963,
|
| 29414 |
+
"learning_rate": 8.49597312120164e-06,
|
| 29415 |
+
"loss": 5.4888,
|
| 29416 |
+
"step": 16804
|
| 29417 |
+
},
|
| 29418 |
+
{
|
| 29419 |
+
"epoch": 0.83,
|
| 29420 |
+
"grad_norm": 1.9529187679290771,
|
| 29421 |
+
"learning_rate": 8.486091210040022e-06,
|
| 29422 |
+
"loss": 5.2731,
|
| 29423 |
+
"step": 16808
|
| 29424 |
+
},
|
| 29425 |
+
{
|
| 29426 |
+
"epoch": 0.83,
|
| 29427 |
+
"grad_norm": 1.897220492362976,
|
| 29428 |
+
"learning_rate": 8.476209298878403e-06,
|
| 29429 |
+
"loss": 5.4103,
|
| 29430 |
+
"step": 16812
|
| 29431 |
+
},
|
| 29432 |
+
{
|
| 29433 |
+
"epoch": 0.83,
|
| 29434 |
+
"grad_norm": 2.0591204166412354,
|
| 29435 |
+
"learning_rate": 8.466327387716785e-06,
|
| 29436 |
+
"loss": 5.5609,
|
| 29437 |
+
"step": 16816
|
| 29438 |
+
},
|
| 29439 |
+
{
|
| 29440 |
+
"epoch": 0.83,
|
| 29441 |
+
"grad_norm": 1.8429813385009766,
|
| 29442 |
+
"learning_rate": 8.456445476555167e-06,
|
| 29443 |
+
"loss": 5.3453,
|
| 29444 |
+
"step": 16820
|
| 29445 |
+
},
|
| 29446 |
+
{
|
| 29447 |
+
"epoch": 0.83,
|
| 29448 |
+
"grad_norm": 1.854067087173462,
|
| 29449 |
+
"learning_rate": 8.446563565393547e-06,
|
| 29450 |
+
"loss": 5.4855,
|
| 29451 |
+
"step": 16824
|
| 29452 |
+
},
|
| 29453 |
+
{
|
| 29454 |
+
"epoch": 0.83,
|
| 29455 |
+
"grad_norm": 1.87723708152771,
|
| 29456 |
+
"learning_rate": 8.43668165423193e-06,
|
| 29457 |
+
"loss": 5.5087,
|
| 29458 |
+
"step": 16828
|
| 29459 |
+
},
|
| 29460 |
+
{
|
| 29461 |
+
"epoch": 0.83,
|
| 29462 |
+
"grad_norm": 2.25486159324646,
|
| 29463 |
+
"learning_rate": 8.42679974307031e-06,
|
| 29464 |
+
"loss": 5.6419,
|
| 29465 |
+
"step": 16832
|
| 29466 |
+
},
|
| 29467 |
+
{
|
| 29468 |
+
"epoch": 0.83,
|
| 29469 |
+
"grad_norm": 2.1054129600524902,
|
| 29470 |
+
"learning_rate": 8.416917831908692e-06,
|
| 29471 |
+
"loss": 5.4713,
|
| 29472 |
+
"step": 16836
|
| 29473 |
+
},
|
| 29474 |
+
{
|
| 29475 |
+
"epoch": 0.83,
|
| 29476 |
+
"grad_norm": 1.9546363353729248,
|
| 29477 |
+
"learning_rate": 8.407035920747072e-06,
|
| 29478 |
+
"loss": 5.51,
|
| 29479 |
+
"step": 16840
|
| 29480 |
+
},
|
| 29481 |
+
{
|
| 29482 |
+
"epoch": 0.83,
|
| 29483 |
+
"grad_norm": 1.8574483394622803,
|
| 29484 |
+
"learning_rate": 8.397154009585454e-06,
|
| 29485 |
+
"loss": 5.4273,
|
| 29486 |
+
"step": 16844
|
| 29487 |
+
},
|
| 29488 |
+
{
|
| 29489 |
+
"epoch": 0.83,
|
| 29490 |
+
"grad_norm": 2.135690927505493,
|
| 29491 |
+
"learning_rate": 8.387272098423836e-06,
|
| 29492 |
+
"loss": 5.4408,
|
| 29493 |
+
"step": 16848
|
| 29494 |
+
},
|
| 29495 |
+
{
|
| 29496 |
+
"epoch": 0.83,
|
| 29497 |
+
"grad_norm": 1.9898631572723389,
|
| 29498 |
+
"learning_rate": 8.377390187262217e-06,
|
| 29499 |
+
"loss": 5.2905,
|
| 29500 |
+
"step": 16852
|
| 29501 |
+
},
|
| 29502 |
+
{
|
| 29503 |
+
"epoch": 0.83,
|
| 29504 |
+
"grad_norm": 2.016470193862915,
|
| 29505 |
+
"learning_rate": 8.367508276100599e-06,
|
| 29506 |
+
"loss": 5.3266,
|
| 29507 |
+
"step": 16856
|
| 29508 |
+
},
|
| 29509 |
+
{
|
| 29510 |
+
"epoch": 0.83,
|
| 29511 |
+
"grad_norm": 1.9973735809326172,
|
| 29512 |
+
"learning_rate": 8.357626364938979e-06,
|
| 29513 |
+
"loss": 5.4987,
|
| 29514 |
+
"step": 16860
|
| 29515 |
+
},
|
| 29516 |
+
{
|
| 29517 |
+
"epoch": 0.83,
|
| 29518 |
+
"grad_norm": 2.116567373275757,
|
| 29519 |
+
"learning_rate": 8.347744453777361e-06,
|
| 29520 |
+
"loss": 5.4306,
|
| 29521 |
+
"step": 16864
|
| 29522 |
+
},
|
| 29523 |
+
{
|
| 29524 |
+
"epoch": 0.83,
|
| 29525 |
+
"grad_norm": 2.044475793838501,
|
| 29526 |
+
"learning_rate": 8.337862542615741e-06,
|
| 29527 |
+
"loss": 5.4907,
|
| 29528 |
+
"step": 16868
|
| 29529 |
+
},
|
| 29530 |
+
{
|
| 29531 |
+
"epoch": 0.83,
|
| 29532 |
+
"grad_norm": 2.090527296066284,
|
| 29533 |
+
"learning_rate": 8.327980631454123e-06,
|
| 29534 |
+
"loss": 5.3698,
|
| 29535 |
+
"step": 16872
|
| 29536 |
+
},
|
| 29537 |
+
{
|
| 29538 |
+
"epoch": 0.83,
|
| 29539 |
+
"grad_norm": 1.98384428024292,
|
| 29540 |
+
"learning_rate": 8.318098720292506e-06,
|
| 29541 |
+
"loss": 5.469,
|
| 29542 |
+
"step": 16876
|
| 29543 |
+
},
|
| 29544 |
+
{
|
| 29545 |
+
"epoch": 0.83,
|
| 29546 |
+
"grad_norm": 1.775121808052063,
|
| 29547 |
+
"learning_rate": 8.308216809130886e-06,
|
| 29548 |
+
"loss": 5.461,
|
| 29549 |
+
"step": 16880
|
| 29550 |
+
},
|
| 29551 |
+
{
|
| 29552 |
+
"epoch": 0.83,
|
| 29553 |
+
"grad_norm": 1.9661427736282349,
|
| 29554 |
+
"learning_rate": 8.298334897969268e-06,
|
| 29555 |
+
"loss": 5.494,
|
| 29556 |
+
"step": 16884
|
| 29557 |
+
},
|
| 29558 |
+
{
|
| 29559 |
+
"epoch": 0.83,
|
| 29560 |
+
"grad_norm": 2.0031895637512207,
|
| 29561 |
+
"learning_rate": 8.288452986807648e-06,
|
| 29562 |
+
"loss": 5.5011,
|
| 29563 |
+
"step": 16888
|
| 29564 |
+
},
|
| 29565 |
+
{
|
| 29566 |
+
"epoch": 0.83,
|
| 29567 |
+
"grad_norm": 2.221911907196045,
|
| 29568 |
+
"learning_rate": 8.27857107564603e-06,
|
| 29569 |
+
"loss": 5.4296,
|
| 29570 |
+
"step": 16892
|
| 29571 |
+
},
|
| 29572 |
+
{
|
| 29573 |
+
"epoch": 0.83,
|
| 29574 |
+
"grad_norm": 2.0504343509674072,
|
| 29575 |
+
"learning_rate": 8.26868916448441e-06,
|
| 29576 |
+
"loss": 5.5495,
|
| 29577 |
+
"step": 16896
|
| 29578 |
+
},
|
| 29579 |
+
{
|
| 29580 |
+
"epoch": 0.84,
|
| 29581 |
+
"grad_norm": 2.1068339347839355,
|
| 29582 |
+
"learning_rate": 8.258807253322794e-06,
|
| 29583 |
+
"loss": 5.408,
|
| 29584 |
+
"step": 16900
|
| 29585 |
+
},
|
| 29586 |
+
{
|
| 29587 |
+
"epoch": 0.84,
|
| 29588 |
+
"grad_norm": 2.0044867992401123,
|
| 29589 |
+
"learning_rate": 8.248925342161175e-06,
|
| 29590 |
+
"loss": 5.512,
|
| 29591 |
+
"step": 16904
|
| 29592 |
+
},
|
| 29593 |
+
{
|
| 29594 |
+
"epoch": 0.84,
|
| 29595 |
+
"grad_norm": 2.3192813396453857,
|
| 29596 |
+
"learning_rate": 8.239043430999555e-06,
|
| 29597 |
+
"loss": 5.4185,
|
| 29598 |
+
"step": 16908
|
| 29599 |
+
},
|
| 29600 |
+
{
|
| 29601 |
+
"epoch": 0.84,
|
| 29602 |
+
"grad_norm": 1.8410991430282593,
|
| 29603 |
+
"learning_rate": 8.229161519837937e-06,
|
| 29604 |
+
"loss": 5.4222,
|
| 29605 |
+
"step": 16912
|
| 29606 |
+
},
|
| 29607 |
+
{
|
| 29608 |
+
"epoch": 0.84,
|
| 29609 |
+
"grad_norm": 2.0134191513061523,
|
| 29610 |
+
"learning_rate": 8.219279608676318e-06,
|
| 29611 |
+
"loss": 5.4357,
|
| 29612 |
+
"step": 16916
|
| 29613 |
+
},
|
| 29614 |
+
{
|
| 29615 |
+
"epoch": 0.84,
|
| 29616 |
+
"grad_norm": 2.0390844345092773,
|
| 29617 |
+
"learning_rate": 8.2093976975147e-06,
|
| 29618 |
+
"loss": 5.5363,
|
| 29619 |
+
"step": 16920
|
| 29620 |
+
},
|
| 29621 |
+
{
|
| 29622 |
+
"epoch": 0.84,
|
| 29623 |
+
"grad_norm": 2.12786602973938,
|
| 29624 |
+
"learning_rate": 8.19951578635308e-06,
|
| 29625 |
+
"loss": 5.31,
|
| 29626 |
+
"step": 16924
|
| 29627 |
+
},
|
| 29628 |
+
{
|
| 29629 |
+
"epoch": 0.84,
|
| 29630 |
+
"grad_norm": 1.9766027927398682,
|
| 29631 |
+
"learning_rate": 8.189633875191464e-06,
|
| 29632 |
+
"loss": 5.534,
|
| 29633 |
+
"step": 16928
|
| 29634 |
+
},
|
| 29635 |
+
{
|
| 29636 |
+
"epoch": 0.84,
|
| 29637 |
+
"grad_norm": 1.7689497470855713,
|
| 29638 |
+
"learning_rate": 8.179751964029844e-06,
|
| 29639 |
+
"loss": 5.3465,
|
| 29640 |
+
"step": 16932
|
| 29641 |
+
},
|
| 29642 |
+
{
|
| 29643 |
+
"epoch": 0.84,
|
| 29644 |
+
"grad_norm": 2.117271900177002,
|
| 29645 |
+
"learning_rate": 8.169870052868225e-06,
|
| 29646 |
+
"loss": 5.4583,
|
| 29647 |
+
"step": 16936
|
| 29648 |
+
},
|
| 29649 |
+
{
|
| 29650 |
+
"epoch": 0.84,
|
| 29651 |
+
"grad_norm": 2.0808498859405518,
|
| 29652 |
+
"learning_rate": 8.159988141706607e-06,
|
| 29653 |
+
"loss": 5.477,
|
| 29654 |
+
"step": 16940
|
| 29655 |
+
},
|
| 29656 |
+
{
|
| 29657 |
+
"epoch": 0.84,
|
| 29658 |
+
"grad_norm": 2.0178062915802,
|
| 29659 |
+
"learning_rate": 8.150106230544987e-06,
|
| 29660 |
+
"loss": 5.5178,
|
| 29661 |
+
"step": 16944
|
| 29662 |
+
},
|
| 29663 |
+
{
|
| 29664 |
+
"epoch": 0.84,
|
| 29665 |
+
"grad_norm": 1.7878342866897583,
|
| 29666 |
+
"learning_rate": 8.140224319383369e-06,
|
| 29667 |
+
"loss": 5.533,
|
| 29668 |
+
"step": 16948
|
| 29669 |
+
},
|
| 29670 |
+
{
|
| 29671 |
+
"epoch": 0.84,
|
| 29672 |
+
"grad_norm": 2.0112874507904053,
|
| 29673 |
+
"learning_rate": 8.13034240822175e-06,
|
| 29674 |
+
"loss": 5.342,
|
| 29675 |
+
"step": 16952
|
| 29676 |
+
},
|
| 29677 |
+
{
|
| 29678 |
+
"epoch": 0.84,
|
| 29679 |
+
"grad_norm": 2.224484443664551,
|
| 29680 |
+
"learning_rate": 8.120460497060133e-06,
|
| 29681 |
+
"loss": 5.4574,
|
| 29682 |
+
"step": 16956
|
| 29683 |
+
},
|
| 29684 |
+
{
|
| 29685 |
+
"epoch": 0.84,
|
| 29686 |
+
"grad_norm": 2.29886531829834,
|
| 29687 |
+
"learning_rate": 8.110578585898514e-06,
|
| 29688 |
+
"loss": 5.5541,
|
| 29689 |
+
"step": 16960
|
| 29690 |
+
},
|
| 29691 |
+
{
|
| 29692 |
+
"epoch": 0.84,
|
| 29693 |
+
"grad_norm": 1.8924994468688965,
|
| 29694 |
+
"learning_rate": 8.100696674736896e-06,
|
| 29695 |
+
"loss": 5.4196,
|
| 29696 |
+
"step": 16964
|
| 29697 |
+
},
|
| 29698 |
+
{
|
| 29699 |
+
"epoch": 0.84,
|
| 29700 |
+
"grad_norm": 2.2159488201141357,
|
| 29701 |
+
"learning_rate": 8.090814763575276e-06,
|
| 29702 |
+
"loss": 5.4989,
|
| 29703 |
+
"step": 16968
|
| 29704 |
+
},
|
| 29705 |
+
{
|
| 29706 |
+
"epoch": 0.84,
|
| 29707 |
+
"grad_norm": 2.170715570449829,
|
| 29708 |
+
"learning_rate": 8.080932852413656e-06,
|
| 29709 |
+
"loss": 5.5074,
|
| 29710 |
+
"step": 16972
|
| 29711 |
+
},
|
| 29712 |
+
{
|
| 29713 |
+
"epoch": 0.84,
|
| 29714 |
+
"grad_norm": 2.0515708923339844,
|
| 29715 |
+
"learning_rate": 8.071050941252038e-06,
|
| 29716 |
+
"loss": 5.4033,
|
| 29717 |
+
"step": 16976
|
| 29718 |
+
},
|
| 29719 |
+
{
|
| 29720 |
+
"epoch": 0.84,
|
| 29721 |
+
"grad_norm": 2.0467865467071533,
|
| 29722 |
+
"learning_rate": 8.061169030090419e-06,
|
| 29723 |
+
"loss": 5.4566,
|
| 29724 |
+
"step": 16980
|
| 29725 |
+
},
|
| 29726 |
+
{
|
| 29727 |
+
"epoch": 0.84,
|
| 29728 |
+
"grad_norm": 1.9163670539855957,
|
| 29729 |
+
"learning_rate": 8.0512871189288e-06,
|
| 29730 |
+
"loss": 5.5335,
|
| 29731 |
+
"step": 16984
|
| 29732 |
+
},
|
| 29733 |
+
{
|
| 29734 |
+
"epoch": 0.84,
|
| 29735 |
+
"grad_norm": 2.1390318870544434,
|
| 29736 |
+
"learning_rate": 8.041405207767183e-06,
|
| 29737 |
+
"loss": 5.5707,
|
| 29738 |
+
"step": 16988
|
| 29739 |
+
},
|
| 29740 |
+
{
|
| 29741 |
+
"epoch": 0.84,
|
| 29742 |
+
"grad_norm": 1.9965319633483887,
|
| 29743 |
+
"learning_rate": 8.031523296605565e-06,
|
| 29744 |
+
"loss": 5.4165,
|
| 29745 |
+
"step": 16992
|
| 29746 |
+
},
|
| 29747 |
+
{
|
| 29748 |
+
"epoch": 0.84,
|
| 29749 |
+
"grad_norm": 2.137233257293701,
|
| 29750 |
+
"learning_rate": 8.021641385443945e-06,
|
| 29751 |
+
"loss": 5.3596,
|
| 29752 |
+
"step": 16996
|
| 29753 |
+
},
|
| 29754 |
+
{
|
| 29755 |
+
"epoch": 0.84,
|
| 29756 |
+
"grad_norm": 2.152256727218628,
|
| 29757 |
+
"learning_rate": 8.011759474282327e-06,
|
| 29758 |
+
"loss": 5.3396,
|
| 29759 |
+
"step": 17000
|
| 29760 |
+
},
|
| 29761 |
+
{
|
| 29762 |
+
"epoch": 0.84,
|
| 29763 |
+
"grad_norm": 2.28680682182312,
|
| 29764 |
+
"learning_rate": 8.001877563120708e-06,
|
| 29765 |
+
"loss": 5.5507,
|
| 29766 |
+
"step": 17004
|
| 29767 |
+
},
|
| 29768 |
+
{
|
| 29769 |
+
"epoch": 0.84,
|
| 29770 |
+
"grad_norm": 2.26821231842041,
|
| 29771 |
+
"learning_rate": 7.991995651959088e-06,
|
| 29772 |
+
"loss": 5.4969,
|
| 29773 |
+
"step": 17008
|
| 29774 |
+
},
|
| 29775 |
+
{
|
| 29776 |
+
"epoch": 0.84,
|
| 29777 |
+
"grad_norm": 2.275667428970337,
|
| 29778 |
+
"learning_rate": 7.98211374079747e-06,
|
| 29779 |
+
"loss": 5.441,
|
| 29780 |
+
"step": 17012
|
| 29781 |
+
},
|
| 29782 |
+
{
|
| 29783 |
+
"epoch": 0.84,
|
| 29784 |
+
"grad_norm": 2.080756902694702,
|
| 29785 |
+
"learning_rate": 7.972231829635852e-06,
|
| 29786 |
+
"loss": 5.4398,
|
| 29787 |
+
"step": 17016
|
| 29788 |
+
},
|
| 29789 |
+
{
|
| 29790 |
+
"epoch": 0.84,
|
| 29791 |
+
"grad_norm": 2.10422420501709,
|
| 29792 |
+
"learning_rate": 7.962349918474234e-06,
|
| 29793 |
+
"loss": 5.437,
|
| 29794 |
+
"step": 17020
|
| 29795 |
+
},
|
| 29796 |
+
{
|
| 29797 |
+
"epoch": 0.84,
|
| 29798 |
+
"grad_norm": 1.858323335647583,
|
| 29799 |
+
"learning_rate": 7.952468007312615e-06,
|
| 29800 |
+
"loss": 5.5432,
|
| 29801 |
+
"step": 17024
|
| 29802 |
+
},
|
| 29803 |
+
{
|
| 29804 |
+
"epoch": 0.84,
|
| 29805 |
+
"grad_norm": 2.4101650714874268,
|
| 29806 |
+
"learning_rate": 7.942586096150997e-06,
|
| 29807 |
+
"loss": 5.412,
|
| 29808 |
+
"step": 17028
|
| 29809 |
+
},
|
| 29810 |
+
{
|
| 29811 |
+
"epoch": 0.84,
|
| 29812 |
+
"grad_norm": 2.2219436168670654,
|
| 29813 |
+
"learning_rate": 7.932704184989377e-06,
|
| 29814 |
+
"loss": 5.5279,
|
| 29815 |
+
"step": 17032
|
| 29816 |
+
},
|
| 29817 |
+
{
|
| 29818 |
+
"epoch": 0.84,
|
| 29819 |
+
"grad_norm": 2.182474374771118,
|
| 29820 |
+
"learning_rate": 7.922822273827757e-06,
|
| 29821 |
+
"loss": 5.4662,
|
| 29822 |
+
"step": 17036
|
| 29823 |
+
},
|
| 29824 |
+
{
|
| 29825 |
+
"epoch": 0.84,
|
| 29826 |
+
"grad_norm": 2.060351610183716,
|
| 29827 |
+
"learning_rate": 7.91294036266614e-06,
|
| 29828 |
+
"loss": 5.504,
|
| 29829 |
+
"step": 17040
|
| 29830 |
+
},
|
| 29831 |
+
{
|
| 29832 |
+
"epoch": 0.84,
|
| 29833 |
+
"grad_norm": 2.1096701622009277,
|
| 29834 |
+
"learning_rate": 7.903058451504522e-06,
|
| 29835 |
+
"loss": 5.4539,
|
| 29836 |
+
"step": 17044
|
| 29837 |
+
},
|
| 29838 |
+
{
|
| 29839 |
+
"epoch": 0.84,
|
| 29840 |
+
"grad_norm": 2.0492708683013916,
|
| 29841 |
+
"learning_rate": 7.893176540342904e-06,
|
| 29842 |
+
"loss": 5.4398,
|
| 29843 |
+
"step": 17048
|
| 29844 |
+
},
|
| 29845 |
+
{
|
| 29846 |
+
"epoch": 0.84,
|
| 29847 |
+
"grad_norm": 2.032947301864624,
|
| 29848 |
+
"learning_rate": 7.883294629181284e-06,
|
| 29849 |
+
"loss": 5.5253,
|
| 29850 |
+
"step": 17052
|
| 29851 |
+
},
|
| 29852 |
+
{
|
| 29853 |
+
"epoch": 0.84,
|
| 29854 |
+
"grad_norm": 2.0764636993408203,
|
| 29855 |
+
"learning_rate": 7.873412718019666e-06,
|
| 29856 |
+
"loss": 5.4655,
|
| 29857 |
+
"step": 17056
|
| 29858 |
+
},
|
| 29859 |
+
{
|
| 29860 |
+
"epoch": 0.84,
|
| 29861 |
+
"grad_norm": 2.105656862258911,
|
| 29862 |
+
"learning_rate": 7.863530806858046e-06,
|
| 29863 |
+
"loss": 5.4761,
|
| 29864 |
+
"step": 17060
|
| 29865 |
+
},
|
| 29866 |
+
{
|
| 29867 |
+
"epoch": 0.84,
|
| 29868 |
+
"grad_norm": 1.975953459739685,
|
| 29869 |
+
"learning_rate": 7.853648895696428e-06,
|
| 29870 |
+
"loss": 5.5364,
|
| 29871 |
+
"step": 17064
|
| 29872 |
+
},
|
| 29873 |
+
{
|
| 29874 |
+
"epoch": 0.84,
|
| 29875 |
+
"grad_norm": 2.0592944622039795,
|
| 29876 |
+
"learning_rate": 7.843766984534809e-06,
|
| 29877 |
+
"loss": 5.4987,
|
| 29878 |
+
"step": 17068
|
| 29879 |
+
},
|
| 29880 |
+
{
|
| 29881 |
+
"epoch": 0.84,
|
| 29882 |
+
"grad_norm": 2.1122117042541504,
|
| 29883 |
+
"learning_rate": 7.833885073373191e-06,
|
| 29884 |
+
"loss": 5.4162,
|
| 29885 |
+
"step": 17072
|
| 29886 |
+
},
|
| 29887 |
+
{
|
| 29888 |
+
"epoch": 0.84,
|
| 29889 |
+
"grad_norm": 2.143172264099121,
|
| 29890 |
+
"learning_rate": 7.824003162211573e-06,
|
| 29891 |
+
"loss": 5.4959,
|
| 29892 |
+
"step": 17076
|
| 29893 |
+
},
|
| 29894 |
+
{
|
| 29895 |
+
"epoch": 0.84,
|
| 29896 |
+
"grad_norm": 1.9919787645339966,
|
| 29897 |
+
"learning_rate": 7.814121251049953e-06,
|
| 29898 |
+
"loss": 5.469,
|
| 29899 |
+
"step": 17080
|
| 29900 |
+
},
|
| 29901 |
+
{
|
| 29902 |
+
"epoch": 0.84,
|
| 29903 |
+
"grad_norm": 1.9146004915237427,
|
| 29904 |
+
"learning_rate": 7.804239339888335e-06,
|
| 29905 |
+
"loss": 5.4748,
|
| 29906 |
+
"step": 17084
|
| 29907 |
+
},
|
| 29908 |
+
{
|
| 29909 |
+
"epoch": 0.84,
|
| 29910 |
+
"grad_norm": 2.3150486946105957,
|
| 29911 |
+
"learning_rate": 7.794357428726716e-06,
|
| 29912 |
+
"loss": 5.4056,
|
| 29913 |
+
"step": 17088
|
| 29914 |
+
},
|
| 29915 |
+
{
|
| 29916 |
+
"epoch": 0.84,
|
| 29917 |
+
"grad_norm": 2.1717705726623535,
|
| 29918 |
+
"learning_rate": 7.784475517565098e-06,
|
| 29919 |
+
"loss": 5.5389,
|
| 29920 |
+
"step": 17092
|
| 29921 |
+
},
|
| 29922 |
+
{
|
| 29923 |
+
"epoch": 0.84,
|
| 29924 |
+
"grad_norm": 2.1674489974975586,
|
| 29925 |
+
"learning_rate": 7.774593606403478e-06,
|
| 29926 |
+
"loss": 5.3378,
|
| 29927 |
+
"step": 17096
|
| 29928 |
+
},
|
| 29929 |
+
{
|
| 29930 |
+
"epoch": 0.84,
|
| 29931 |
+
"grad_norm": 2.17425537109375,
|
| 29932 |
+
"learning_rate": 7.76471169524186e-06,
|
| 29933 |
+
"loss": 5.5094,
|
| 29934 |
+
"step": 17100
|
| 29935 |
+
},
|
| 29936 |
+
{
|
| 29937 |
+
"epoch": 0.85,
|
| 29938 |
+
"grad_norm": 2.2170867919921875,
|
| 29939 |
+
"learning_rate": 7.754829784080242e-06,
|
| 29940 |
+
"loss": 5.4591,
|
| 29941 |
+
"step": 17104
|
| 29942 |
+
},
|
| 29943 |
+
{
|
| 29944 |
+
"epoch": 0.85,
|
| 29945 |
+
"grad_norm": 2.0710206031799316,
|
| 29946 |
+
"learning_rate": 7.744947872918623e-06,
|
| 29947 |
+
"loss": 5.461,
|
| 29948 |
+
"step": 17108
|
| 29949 |
+
},
|
| 29950 |
+
{
|
| 29951 |
+
"epoch": 0.85,
|
| 29952 |
+
"grad_norm": 1.9662617444992065,
|
| 29953 |
+
"learning_rate": 7.735065961757005e-06,
|
| 29954 |
+
"loss": 5.5232,
|
| 29955 |
+
"step": 17112
|
| 29956 |
+
},
|
| 29957 |
+
{
|
| 29958 |
+
"epoch": 0.85,
|
| 29959 |
+
"grad_norm": 2.1950018405914307,
|
| 29960 |
+
"learning_rate": 7.725184050595385e-06,
|
| 29961 |
+
"loss": 5.4845,
|
| 29962 |
+
"step": 17116
|
| 29963 |
+
},
|
| 29964 |
+
{
|
| 29965 |
+
"epoch": 0.85,
|
| 29966 |
+
"grad_norm": 2.166281223297119,
|
| 29967 |
+
"learning_rate": 7.715302139433767e-06,
|
| 29968 |
+
"loss": 5.4899,
|
| 29969 |
+
"step": 17120
|
| 29970 |
+
},
|
| 29971 |
+
{
|
| 29972 |
+
"epoch": 0.85,
|
| 29973 |
+
"grad_norm": 2.0825867652893066,
|
| 29974 |
+
"learning_rate": 7.705420228272148e-06,
|
| 29975 |
+
"loss": 5.5889,
|
| 29976 |
+
"step": 17124
|
| 29977 |
+
},
|
| 29978 |
+
{
|
| 29979 |
+
"epoch": 0.85,
|
| 29980 |
+
"grad_norm": 2.0458121299743652,
|
| 29981 |
+
"learning_rate": 7.69553831711053e-06,
|
| 29982 |
+
"loss": 5.4465,
|
| 29983 |
+
"step": 17128
|
| 29984 |
+
},
|
| 29985 |
+
{
|
| 29986 |
+
"epoch": 0.85,
|
| 29987 |
+
"grad_norm": 1.972931146621704,
|
| 29988 |
+
"learning_rate": 7.685656405948912e-06,
|
| 29989 |
+
"loss": 5.447,
|
| 29990 |
+
"step": 17132
|
| 29991 |
+
},
|
| 29992 |
+
{
|
| 29993 |
+
"epoch": 0.85,
|
| 29994 |
+
"grad_norm": 2.2071616649627686,
|
| 29995 |
+
"learning_rate": 7.675774494787292e-06,
|
| 29996 |
+
"loss": 5.5405,
|
| 29997 |
+
"step": 17136
|
| 29998 |
+
},
|
| 29999 |
+
{
|
| 30000 |
+
"epoch": 0.85,
|
| 30001 |
+
"grad_norm": 2.24798583984375,
|
| 30002 |
+
"learning_rate": 7.665892583625674e-06,
|
| 30003 |
+
"loss": 5.5034,
|
| 30004 |
+
"step": 17140
|
| 30005 |
+
},
|
| 30006 |
+
{
|
| 30007 |
+
"epoch": 0.85,
|
| 30008 |
+
"grad_norm": 2.3352463245391846,
|
| 30009 |
+
"learning_rate": 7.656010672464054e-06,
|
| 30010 |
+
"loss": 5.5322,
|
| 30011 |
+
"step": 17144
|
| 30012 |
+
},
|
| 30013 |
+
{
|
| 30014 |
+
"epoch": 0.85,
|
| 30015 |
+
"grad_norm": 2.1701347827911377,
|
| 30016 |
+
"learning_rate": 7.646128761302436e-06,
|
| 30017 |
+
"loss": 5.6042,
|
| 30018 |
+
"step": 17148
|
| 30019 |
+
},
|
| 30020 |
+
{
|
| 30021 |
+
"epoch": 0.85,
|
| 30022 |
+
"grad_norm": 2.0654942989349365,
|
| 30023 |
+
"learning_rate": 7.636246850140817e-06,
|
| 30024 |
+
"loss": 5.5102,
|
| 30025 |
+
"step": 17152
|
| 30026 |
+
},
|
| 30027 |
+
{
|
| 30028 |
+
"epoch": 0.85,
|
| 30029 |
+
"grad_norm": 2.410454273223877,
|
| 30030 |
+
"learning_rate": 7.626364938979198e-06,
|
| 30031 |
+
"loss": 5.4408,
|
| 30032 |
+
"step": 17156
|
| 30033 |
+
},
|
| 30034 |
+
{
|
| 30035 |
+
"epoch": 0.85,
|
| 30036 |
+
"grad_norm": 2.0221352577209473,
|
| 30037 |
+
"learning_rate": 7.616483027817581e-06,
|
| 30038 |
+
"loss": 5.4086,
|
| 30039 |
+
"step": 17160
|
| 30040 |
+
},
|
| 30041 |
+
{
|
| 30042 |
+
"epoch": 0.85,
|
| 30043 |
+
"grad_norm": 2.211092233657837,
|
| 30044 |
+
"learning_rate": 7.606601116655962e-06,
|
| 30045 |
+
"loss": 5.4589,
|
| 30046 |
+
"step": 17164
|
| 30047 |
+
},
|
| 30048 |
+
{
|
| 30049 |
+
"epoch": 0.85,
|
| 30050 |
+
"grad_norm": 1.9467920064926147,
|
| 30051 |
+
"learning_rate": 7.596719205494343e-06,
|
| 30052 |
+
"loss": 5.4867,
|
| 30053 |
+
"step": 17168
|
| 30054 |
+
},
|
| 30055 |
+
{
|
| 30056 |
+
"epoch": 0.85,
|
| 30057 |
+
"grad_norm": 2.1144025325775146,
|
| 30058 |
+
"learning_rate": 7.586837294332725e-06,
|
| 30059 |
+
"loss": 5.5143,
|
| 30060 |
+
"step": 17172
|
| 30061 |
+
},
|
| 30062 |
+
{
|
| 30063 |
+
"epoch": 0.85,
|
| 30064 |
+
"grad_norm": 2.1652915477752686,
|
| 30065 |
+
"learning_rate": 7.576955383171106e-06,
|
| 30066 |
+
"loss": 5.533,
|
| 30067 |
+
"step": 17176
|
| 30068 |
+
},
|
| 30069 |
+
{
|
| 30070 |
+
"epoch": 0.85,
|
| 30071 |
+
"grad_norm": 1.9289984703063965,
|
| 30072 |
+
"learning_rate": 7.567073472009486e-06,
|
| 30073 |
+
"loss": 5.5406,
|
| 30074 |
+
"step": 17180
|
| 30075 |
+
},
|
| 30076 |
+
{
|
| 30077 |
+
"epoch": 0.85,
|
| 30078 |
+
"grad_norm": 2.028322458267212,
|
| 30079 |
+
"learning_rate": 7.557191560847867e-06,
|
| 30080 |
+
"loss": 5.4118,
|
| 30081 |
+
"step": 17184
|
| 30082 |
+
},
|
| 30083 |
+
{
|
| 30084 |
+
"epoch": 0.85,
|
| 30085 |
+
"grad_norm": 2.2385053634643555,
|
| 30086 |
+
"learning_rate": 7.54730964968625e-06,
|
| 30087 |
+
"loss": 5.5023,
|
| 30088 |
+
"step": 17188
|
| 30089 |
+
},
|
| 30090 |
+
{
|
| 30091 |
+
"epoch": 0.85,
|
| 30092 |
+
"grad_norm": 1.8756178617477417,
|
| 30093 |
+
"learning_rate": 7.5374277385246315e-06,
|
| 30094 |
+
"loss": 5.4028,
|
| 30095 |
+
"step": 17192
|
| 30096 |
+
},
|
| 30097 |
+
{
|
| 30098 |
+
"epoch": 0.85,
|
| 30099 |
+
"grad_norm": 2.0008492469787598,
|
| 30100 |
+
"learning_rate": 7.527545827363013e-06,
|
| 30101 |
+
"loss": 5.4572,
|
| 30102 |
+
"step": 17196
|
| 30103 |
+
},
|
| 30104 |
+
{
|
| 30105 |
+
"epoch": 0.85,
|
| 30106 |
+
"grad_norm": 1.9606680870056152,
|
| 30107 |
+
"learning_rate": 7.517663916201394e-06,
|
| 30108 |
+
"loss": 5.3813,
|
| 30109 |
+
"step": 17200
|
| 30110 |
+
},
|
| 30111 |
+
{
|
| 30112 |
+
"epoch": 0.85,
|
| 30113 |
+
"grad_norm": 1.765757441520691,
|
| 30114 |
+
"learning_rate": 7.507782005039775e-06,
|
| 30115 |
+
"loss": 5.433,
|
| 30116 |
+
"step": 17204
|
| 30117 |
+
},
|
| 30118 |
+
{
|
| 30119 |
+
"epoch": 0.85,
|
| 30120 |
+
"grad_norm": 2.2999327182769775,
|
| 30121 |
+
"learning_rate": 7.497900093878156e-06,
|
| 30122 |
+
"loss": 5.4443,
|
| 30123 |
+
"step": 17208
|
| 30124 |
+
},
|
| 30125 |
+
{
|
| 30126 |
+
"epoch": 0.85,
|
| 30127 |
+
"grad_norm": 1.831790804862976,
|
| 30128 |
+
"learning_rate": 7.488018182716537e-06,
|
| 30129 |
+
"loss": 5.4745,
|
| 30130 |
+
"step": 17212
|
| 30131 |
+
},
|
| 30132 |
+
{
|
| 30133 |
+
"epoch": 0.85,
|
| 30134 |
+
"grad_norm": 2.0281448364257812,
|
| 30135 |
+
"learning_rate": 7.47813627155492e-06,
|
| 30136 |
+
"loss": 5.3546,
|
| 30137 |
+
"step": 17216
|
| 30138 |
+
},
|
| 30139 |
+
{
|
| 30140 |
+
"epoch": 0.85,
|
| 30141 |
+
"grad_norm": 2.163875102996826,
|
| 30142 |
+
"learning_rate": 7.468254360393301e-06,
|
| 30143 |
+
"loss": 5.4957,
|
| 30144 |
+
"step": 17220
|
| 30145 |
+
},
|
| 30146 |
+
{
|
| 30147 |
+
"epoch": 0.85,
|
| 30148 |
+
"grad_norm": 2.0201468467712402,
|
| 30149 |
+
"learning_rate": 7.458372449231682e-06,
|
| 30150 |
+
"loss": 5.3797,
|
| 30151 |
+
"step": 17224
|
| 30152 |
+
},
|
| 30153 |
+
{
|
| 30154 |
+
"epoch": 0.85,
|
| 30155 |
+
"grad_norm": 1.9520927667617798,
|
| 30156 |
+
"learning_rate": 7.448490538070063e-06,
|
| 30157 |
+
"loss": 5.5663,
|
| 30158 |
+
"step": 17228
|
| 30159 |
+
},
|
| 30160 |
+
{
|
| 30161 |
+
"epoch": 0.85,
|
| 30162 |
+
"grad_norm": 1.9706037044525146,
|
| 30163 |
+
"learning_rate": 7.4386086269084445e-06,
|
| 30164 |
+
"loss": 5.4462,
|
| 30165 |
+
"step": 17232
|
| 30166 |
+
},
|
| 30167 |
+
{
|
| 30168 |
+
"epoch": 0.85,
|
| 30169 |
+
"grad_norm": 2.2334280014038086,
|
| 30170 |
+
"learning_rate": 7.428726715746826e-06,
|
| 30171 |
+
"loss": 5.5028,
|
| 30172 |
+
"step": 17236
|
| 30173 |
+
},
|
| 30174 |
+
{
|
| 30175 |
+
"epoch": 0.85,
|
| 30176 |
+
"grad_norm": 2.089432716369629,
|
| 30177 |
+
"learning_rate": 7.418844804585207e-06,
|
| 30178 |
+
"loss": 5.5537,
|
| 30179 |
+
"step": 17240
|
| 30180 |
+
},
|
| 30181 |
+
{
|
| 30182 |
+
"epoch": 0.85,
|
| 30183 |
+
"grad_norm": 2.0354325771331787,
|
| 30184 |
+
"learning_rate": 7.408962893423589e-06,
|
| 30185 |
+
"loss": 5.4,
|
| 30186 |
+
"step": 17244
|
| 30187 |
+
},
|
| 30188 |
+
{
|
| 30189 |
+
"epoch": 0.85,
|
| 30190 |
+
"grad_norm": 1.9282554388046265,
|
| 30191 |
+
"learning_rate": 7.39908098226197e-06,
|
| 30192 |
+
"loss": 5.3915,
|
| 30193 |
+
"step": 17248
|
| 30194 |
+
},
|
| 30195 |
+
{
|
| 30196 |
+
"epoch": 0.85,
|
| 30197 |
+
"grad_norm": 2.133868455886841,
|
| 30198 |
+
"learning_rate": 7.389199071100351e-06,
|
| 30199 |
+
"loss": 5.4868,
|
| 30200 |
+
"step": 17252
|
| 30201 |
+
},
|
| 30202 |
+
{
|
| 30203 |
+
"epoch": 0.85,
|
| 30204 |
+
"grad_norm": 2.0873701572418213,
|
| 30205 |
+
"learning_rate": 7.379317159938733e-06,
|
| 30206 |
+
"loss": 5.6375,
|
| 30207 |
+
"step": 17256
|
| 30208 |
+
},
|
| 30209 |
+
{
|
| 30210 |
+
"epoch": 0.85,
|
| 30211 |
+
"grad_norm": 1.9751038551330566,
|
| 30212 |
+
"learning_rate": 7.369435248777114e-06,
|
| 30213 |
+
"loss": 5.4291,
|
| 30214 |
+
"step": 17260
|
| 30215 |
+
},
|
| 30216 |
+
{
|
| 30217 |
+
"epoch": 0.85,
|
| 30218 |
+
"grad_norm": 1.8549004793167114,
|
| 30219 |
+
"learning_rate": 7.359553337615495e-06,
|
| 30220 |
+
"loss": 5.4953,
|
| 30221 |
+
"step": 17264
|
| 30222 |
+
},
|
| 30223 |
+
{
|
| 30224 |
+
"epoch": 0.85,
|
| 30225 |
+
"grad_norm": 1.9882365465164185,
|
| 30226 |
+
"learning_rate": 7.349671426453876e-06,
|
| 30227 |
+
"loss": 5.5191,
|
| 30228 |
+
"step": 17268
|
| 30229 |
+
},
|
| 30230 |
+
{
|
| 30231 |
+
"epoch": 0.85,
|
| 30232 |
+
"grad_norm": 2.0008509159088135,
|
| 30233 |
+
"learning_rate": 7.339789515292258e-06,
|
| 30234 |
+
"loss": 5.573,
|
| 30235 |
+
"step": 17272
|
| 30236 |
+
},
|
| 30237 |
+
{
|
| 30238 |
+
"epoch": 0.85,
|
| 30239 |
+
"grad_norm": 1.9084336757659912,
|
| 30240 |
+
"learning_rate": 7.3299076041306395e-06,
|
| 30241 |
+
"loss": 5.5864,
|
| 30242 |
+
"step": 17276
|
| 30243 |
+
},
|
| 30244 |
+
{
|
| 30245 |
+
"epoch": 0.85,
|
| 30246 |
+
"grad_norm": 2.3234941959381104,
|
| 30247 |
+
"learning_rate": 7.320025692969021e-06,
|
| 30248 |
+
"loss": 5.4648,
|
| 30249 |
+
"step": 17280
|
| 30250 |
+
},
|
| 30251 |
+
{
|
| 30252 |
+
"epoch": 0.85,
|
| 30253 |
+
"grad_norm": 2.033445358276367,
|
| 30254 |
+
"learning_rate": 7.310143781807402e-06,
|
| 30255 |
+
"loss": 5.5663,
|
| 30256 |
+
"step": 17284
|
| 30257 |
+
},
|
| 30258 |
+
{
|
| 30259 |
+
"epoch": 0.85,
|
| 30260 |
+
"grad_norm": 2.036726474761963,
|
| 30261 |
+
"learning_rate": 7.300261870645783e-06,
|
| 30262 |
+
"loss": 5.3706,
|
| 30263 |
+
"step": 17288
|
| 30264 |
+
},
|
| 30265 |
+
{
|
| 30266 |
+
"epoch": 0.85,
|
| 30267 |
+
"grad_norm": 2.135927677154541,
|
| 30268 |
+
"learning_rate": 7.290379959484164e-06,
|
| 30269 |
+
"loss": 5.4365,
|
| 30270 |
+
"step": 17292
|
| 30271 |
+
},
|
| 30272 |
+
{
|
| 30273 |
+
"epoch": 0.85,
|
| 30274 |
+
"grad_norm": 2.0434539318084717,
|
| 30275 |
+
"learning_rate": 7.280498048322546e-06,
|
| 30276 |
+
"loss": 5.4914,
|
| 30277 |
+
"step": 17296
|
| 30278 |
+
},
|
| 30279 |
+
{
|
| 30280 |
+
"epoch": 0.85,
|
| 30281 |
+
"grad_norm": 1.9364794492721558,
|
| 30282 |
+
"learning_rate": 7.270616137160927e-06,
|
| 30283 |
+
"loss": 5.358,
|
| 30284 |
+
"step": 17300
|
| 30285 |
+
},
|
| 30286 |
+
{
|
| 30287 |
+
"epoch": 0.85,
|
| 30288 |
+
"grad_norm": 2.1555495262145996,
|
| 30289 |
+
"learning_rate": 7.260734225999309e-06,
|
| 30290 |
+
"loss": 5.4667,
|
| 30291 |
+
"step": 17304
|
| 30292 |
+
},
|
| 30293 |
+
{
|
| 30294 |
+
"epoch": 0.86,
|
| 30295 |
+
"grad_norm": 2.0687687397003174,
|
| 30296 |
+
"learning_rate": 7.25085231483769e-06,
|
| 30297 |
+
"loss": 5.5075,
|
| 30298 |
+
"step": 17308
|
| 30299 |
+
},
|
| 30300 |
+
{
|
| 30301 |
+
"epoch": 0.86,
|
| 30302 |
+
"grad_norm": 2.2169644832611084,
|
| 30303 |
+
"learning_rate": 7.240970403676071e-06,
|
| 30304 |
+
"loss": 5.5186,
|
| 30305 |
+
"step": 17312
|
| 30306 |
+
},
|
| 30307 |
+
{
|
| 30308 |
+
"epoch": 0.86,
|
| 30309 |
+
"grad_norm": 2.0690207481384277,
|
| 30310 |
+
"learning_rate": 7.2310884925144525e-06,
|
| 30311 |
+
"loss": 5.4944,
|
| 30312 |
+
"step": 17316
|
| 30313 |
+
},
|
| 30314 |
+
{
|
| 30315 |
+
"epoch": 0.86,
|
| 30316 |
+
"grad_norm": 2.172851324081421,
|
| 30317 |
+
"learning_rate": 7.221206581352834e-06,
|
| 30318 |
+
"loss": 5.5178,
|
| 30319 |
+
"step": 17320
|
| 30320 |
+
},
|
| 30321 |
+
{
|
| 30322 |
+
"epoch": 0.86,
|
| 30323 |
+
"grad_norm": 2.178602457046509,
|
| 30324 |
+
"learning_rate": 7.211324670191215e-06,
|
| 30325 |
+
"loss": 5.5693,
|
| 30326 |
+
"step": 17324
|
| 30327 |
+
},
|
| 30328 |
+
{
|
| 30329 |
+
"epoch": 0.86,
|
| 30330 |
+
"grad_norm": 1.9525049924850464,
|
| 30331 |
+
"learning_rate": 7.201442759029596e-06,
|
| 30332 |
+
"loss": 5.3773,
|
| 30333 |
+
"step": 17328
|
| 30334 |
+
},
|
| 30335 |
+
{
|
| 30336 |
+
"epoch": 0.86,
|
| 30337 |
+
"grad_norm": 2.0250043869018555,
|
| 30338 |
+
"learning_rate": 7.191560847867978e-06,
|
| 30339 |
+
"loss": 5.4672,
|
| 30340 |
+
"step": 17332
|
| 30341 |
+
},
|
| 30342 |
+
{
|
| 30343 |
+
"epoch": 0.86,
|
| 30344 |
+
"grad_norm": 2.229799747467041,
|
| 30345 |
+
"learning_rate": 7.1816789367063594e-06,
|
| 30346 |
+
"loss": 5.4451,
|
| 30347 |
+
"step": 17336
|
| 30348 |
+
},
|
| 30349 |
+
{
|
| 30350 |
+
"epoch": 0.86,
|
| 30351 |
+
"grad_norm": 2.2048773765563965,
|
| 30352 |
+
"learning_rate": 7.171797025544741e-06,
|
| 30353 |
+
"loss": 5.5388,
|
| 30354 |
+
"step": 17340
|
| 30355 |
+
},
|
| 30356 |
+
{
|
| 30357 |
+
"epoch": 0.86,
|
| 30358 |
+
"grad_norm": 2.1948986053466797,
|
| 30359 |
+
"learning_rate": 7.161915114383122e-06,
|
| 30360 |
+
"loss": 5.4121,
|
| 30361 |
+
"step": 17344
|
| 30362 |
+
},
|
| 30363 |
+
{
|
| 30364 |
+
"epoch": 0.86,
|
| 30365 |
+
"grad_norm": 2.410446882247925,
|
| 30366 |
+
"learning_rate": 7.152033203221503e-06,
|
| 30367 |
+
"loss": 5.5068,
|
| 30368 |
+
"step": 17348
|
| 30369 |
+
},
|
| 30370 |
+
{
|
| 30371 |
+
"epoch": 0.86,
|
| 30372 |
+
"grad_norm": 2.0198326110839844,
|
| 30373 |
+
"learning_rate": 7.142151292059884e-06,
|
| 30374 |
+
"loss": 5.4786,
|
| 30375 |
+
"step": 17352
|
| 30376 |
+
},
|
| 30377 |
+
{
|
| 30378 |
+
"epoch": 0.86,
|
| 30379 |
+
"grad_norm": 2.1943955421447754,
|
| 30380 |
+
"learning_rate": 7.1322693808982655e-06,
|
| 30381 |
+
"loss": 5.4957,
|
| 30382 |
+
"step": 17356
|
| 30383 |
+
},
|
| 30384 |
+
{
|
| 30385 |
+
"epoch": 0.86,
|
| 30386 |
+
"grad_norm": 2.1132426261901855,
|
| 30387 |
+
"learning_rate": 7.122387469736648e-06,
|
| 30388 |
+
"loss": 5.3965,
|
| 30389 |
+
"step": 17360
|
| 30390 |
+
},
|
| 30391 |
+
{
|
| 30392 |
+
"epoch": 0.86,
|
| 30393 |
+
"grad_norm": 2.0462357997894287,
|
| 30394 |
+
"learning_rate": 7.112505558575029e-06,
|
| 30395 |
+
"loss": 5.5153,
|
| 30396 |
+
"step": 17364
|
| 30397 |
+
},
|
| 30398 |
+
{
|
| 30399 |
+
"epoch": 0.86,
|
| 30400 |
+
"grad_norm": 2.0501723289489746,
|
| 30401 |
+
"learning_rate": 7.10262364741341e-06,
|
| 30402 |
+
"loss": 5.5062,
|
| 30403 |
+
"step": 17368
|
| 30404 |
+
},
|
| 30405 |
+
{
|
| 30406 |
+
"epoch": 0.86,
|
| 30407 |
+
"grad_norm": 2.148674726486206,
|
| 30408 |
+
"learning_rate": 7.092741736251791e-06,
|
| 30409 |
+
"loss": 5.2946,
|
| 30410 |
+
"step": 17372
|
| 30411 |
+
},
|
| 30412 |
+
{
|
| 30413 |
+
"epoch": 0.86,
|
| 30414 |
+
"grad_norm": 2.0384411811828613,
|
| 30415 |
+
"learning_rate": 7.082859825090172e-06,
|
| 30416 |
+
"loss": 5.6131,
|
| 30417 |
+
"step": 17376
|
| 30418 |
+
},
|
| 30419 |
+
{
|
| 30420 |
+
"epoch": 0.86,
|
| 30421 |
+
"grad_norm": 2.235848903656006,
|
| 30422 |
+
"learning_rate": 7.072977913928554e-06,
|
| 30423 |
+
"loss": 5.3982,
|
| 30424 |
+
"step": 17380
|
| 30425 |
+
},
|
| 30426 |
+
{
|
| 30427 |
+
"epoch": 0.86,
|
| 30428 |
+
"grad_norm": 2.0050299167633057,
|
| 30429 |
+
"learning_rate": 7.063096002766935e-06,
|
| 30430 |
+
"loss": 5.5681,
|
| 30431 |
+
"step": 17384
|
| 30432 |
+
},
|
| 30433 |
+
{
|
| 30434 |
+
"epoch": 0.86,
|
| 30435 |
+
"grad_norm": 1.9482308626174927,
|
| 30436 |
+
"learning_rate": 7.053214091605318e-06,
|
| 30437 |
+
"loss": 5.4242,
|
| 30438 |
+
"step": 17388
|
| 30439 |
+
},
|
| 30440 |
+
{
|
| 30441 |
+
"epoch": 0.86,
|
| 30442 |
+
"grad_norm": 2.077125072479248,
|
| 30443 |
+
"learning_rate": 7.043332180443699e-06,
|
| 30444 |
+
"loss": 5.4092,
|
| 30445 |
+
"step": 17392
|
| 30446 |
+
},
|
| 30447 |
+
{
|
| 30448 |
+
"epoch": 0.86,
|
| 30449 |
+
"grad_norm": 2.2242355346679688,
|
| 30450 |
+
"learning_rate": 7.033450269282079e-06,
|
| 30451 |
+
"loss": 5.4268,
|
| 30452 |
+
"step": 17396
|
| 30453 |
+
},
|
| 30454 |
+
{
|
| 30455 |
+
"epoch": 0.86,
|
| 30456 |
+
"grad_norm": 2.2366597652435303,
|
| 30457 |
+
"learning_rate": 7.0235683581204605e-06,
|
| 30458 |
+
"loss": 5.4304,
|
| 30459 |
+
"step": 17400
|
| 30460 |
+
},
|
| 30461 |
+
{
|
| 30462 |
+
"epoch": 0.86,
|
| 30463 |
+
"grad_norm": 2.3268561363220215,
|
| 30464 |
+
"learning_rate": 7.013686446958842e-06,
|
| 30465 |
+
"loss": 5.4141,
|
| 30466 |
+
"step": 17404
|
| 30467 |
+
},
|
| 30468 |
+
{
|
| 30469 |
+
"epoch": 0.86,
|
| 30470 |
+
"grad_norm": 2.1040186882019043,
|
| 30471 |
+
"learning_rate": 7.003804535797223e-06,
|
| 30472 |
+
"loss": 5.4129,
|
| 30473 |
+
"step": 17408
|
| 30474 |
+
},
|
| 30475 |
+
{
|
| 30476 |
+
"epoch": 0.86,
|
| 30477 |
+
"grad_norm": 2.0050957202911377,
|
| 30478 |
+
"learning_rate": 6.993922624635604e-06,
|
| 30479 |
+
"loss": 5.5029,
|
| 30480 |
+
"step": 17412
|
| 30481 |
+
},
|
| 30482 |
+
{
|
| 30483 |
+
"epoch": 0.86,
|
| 30484 |
+
"grad_norm": 1.914214849472046,
|
| 30485 |
+
"learning_rate": 6.984040713473987e-06,
|
| 30486 |
+
"loss": 5.5394,
|
| 30487 |
+
"step": 17416
|
| 30488 |
+
},
|
| 30489 |
+
{
|
| 30490 |
+
"epoch": 0.86,
|
| 30491 |
+
"grad_norm": 2.112946033477783,
|
| 30492 |
+
"learning_rate": 6.974158802312368e-06,
|
| 30493 |
+
"loss": 5.4547,
|
| 30494 |
+
"step": 17420
|
| 30495 |
+
},
|
| 30496 |
+
{
|
| 30497 |
+
"epoch": 0.86,
|
| 30498 |
+
"grad_norm": 1.980510950088501,
|
| 30499 |
+
"learning_rate": 6.9642768911507495e-06,
|
| 30500 |
+
"loss": 5.4997,
|
| 30501 |
+
"step": 17424
|
| 30502 |
+
},
|
| 30503 |
+
{
|
| 30504 |
+
"epoch": 0.86,
|
| 30505 |
+
"grad_norm": 1.985985517501831,
|
| 30506 |
+
"learning_rate": 6.954394979989131e-06,
|
| 30507 |
+
"loss": 5.3671,
|
| 30508 |
+
"step": 17428
|
| 30509 |
+
},
|
| 30510 |
+
{
|
| 30511 |
+
"epoch": 0.86,
|
| 30512 |
+
"grad_norm": 1.897262454032898,
|
| 30513 |
+
"learning_rate": 6.944513068827511e-06,
|
| 30514 |
+
"loss": 5.4578,
|
| 30515 |
+
"step": 17432
|
| 30516 |
+
},
|
| 30517 |
+
{
|
| 30518 |
+
"epoch": 0.86,
|
| 30519 |
+
"grad_norm": 1.9851828813552856,
|
| 30520 |
+
"learning_rate": 6.934631157665892e-06,
|
| 30521 |
+
"loss": 5.4318,
|
| 30522 |
+
"step": 17436
|
| 30523 |
+
},
|
| 30524 |
+
{
|
| 30525 |
+
"epoch": 0.86,
|
| 30526 |
+
"grad_norm": 1.8977246284484863,
|
| 30527 |
+
"learning_rate": 6.9247492465042735e-06,
|
| 30528 |
+
"loss": 5.4777,
|
| 30529 |
+
"step": 17440
|
| 30530 |
+
},
|
| 30531 |
+
{
|
| 30532 |
+
"epoch": 0.86,
|
| 30533 |
+
"grad_norm": 2.1280171871185303,
|
| 30534 |
+
"learning_rate": 6.9148673353426564e-06,
|
| 30535 |
+
"loss": 5.583,
|
| 30536 |
+
"step": 17444
|
| 30537 |
+
},
|
| 30538 |
+
{
|
| 30539 |
+
"epoch": 0.86,
|
| 30540 |
+
"grad_norm": 2.3764641284942627,
|
| 30541 |
+
"learning_rate": 6.904985424181038e-06,
|
| 30542 |
+
"loss": 5.5559,
|
| 30543 |
+
"step": 17448
|
| 30544 |
+
},
|
| 30545 |
+
{
|
| 30546 |
+
"epoch": 0.86,
|
| 30547 |
+
"grad_norm": 1.9994136095046997,
|
| 30548 |
+
"learning_rate": 6.895103513019419e-06,
|
| 30549 |
+
"loss": 5.5662,
|
| 30550 |
+
"step": 17452
|
| 30551 |
+
},
|
| 30552 |
+
{
|
| 30553 |
+
"epoch": 0.86,
|
| 30554 |
+
"grad_norm": 2.108659267425537,
|
| 30555 |
+
"learning_rate": 6.8852216018578e-06,
|
| 30556 |
+
"loss": 5.4098,
|
| 30557 |
+
"step": 17456
|
| 30558 |
+
},
|
| 30559 |
+
{
|
| 30560 |
+
"epoch": 0.86,
|
| 30561 |
+
"grad_norm": 1.9477959871292114,
|
| 30562 |
+
"learning_rate": 6.875339690696181e-06,
|
| 30563 |
+
"loss": 5.4161,
|
| 30564 |
+
"step": 17460
|
| 30565 |
+
},
|
| 30566 |
+
{
|
| 30567 |
+
"epoch": 0.86,
|
| 30568 |
+
"grad_norm": 2.2120134830474854,
|
| 30569 |
+
"learning_rate": 6.865457779534562e-06,
|
| 30570 |
+
"loss": 5.5344,
|
| 30571 |
+
"step": 17464
|
| 30572 |
+
},
|
| 30573 |
+
{
|
| 30574 |
+
"epoch": 0.86,
|
| 30575 |
+
"grad_norm": 1.9351931810379028,
|
| 30576 |
+
"learning_rate": 6.855575868372943e-06,
|
| 30577 |
+
"loss": 5.5115,
|
| 30578 |
+
"step": 17468
|
| 30579 |
+
},
|
| 30580 |
+
{
|
| 30581 |
+
"epoch": 0.86,
|
| 30582 |
+
"grad_norm": 1.9376587867736816,
|
| 30583 |
+
"learning_rate": 6.845693957211324e-06,
|
| 30584 |
+
"loss": 5.4218,
|
| 30585 |
+
"step": 17472
|
| 30586 |
+
},
|
| 30587 |
+
{
|
| 30588 |
+
"epoch": 0.86,
|
| 30589 |
+
"grad_norm": 1.8527143001556396,
|
| 30590 |
+
"learning_rate": 6.835812046049707e-06,
|
| 30591 |
+
"loss": 5.4781,
|
| 30592 |
+
"step": 17476
|
| 30593 |
+
},
|
| 30594 |
+
{
|
| 30595 |
+
"epoch": 0.86,
|
| 30596 |
+
"grad_norm": 1.9370919466018677,
|
| 30597 |
+
"learning_rate": 6.825930134888088e-06,
|
| 30598 |
+
"loss": 5.6185,
|
| 30599 |
+
"step": 17480
|
| 30600 |
+
},
|
| 30601 |
+
{
|
| 30602 |
+
"epoch": 0.86,
|
| 30603 |
+
"grad_norm": 1.8956094980239868,
|
| 30604 |
+
"learning_rate": 6.816048223726469e-06,
|
| 30605 |
+
"loss": 5.5335,
|
| 30606 |
+
"step": 17484
|
| 30607 |
+
},
|
| 30608 |
+
{
|
| 30609 |
+
"epoch": 0.86,
|
| 30610 |
+
"grad_norm": 2.1358373165130615,
|
| 30611 |
+
"learning_rate": 6.806166312564851e-06,
|
| 30612 |
+
"loss": 5.5501,
|
| 30613 |
+
"step": 17488
|
| 30614 |
+
},
|
| 30615 |
+
{
|
| 30616 |
+
"epoch": 0.86,
|
| 30617 |
+
"grad_norm": 1.8500255346298218,
|
| 30618 |
+
"learning_rate": 6.796284401403232e-06,
|
| 30619 |
+
"loss": 5.4718,
|
| 30620 |
+
"step": 17492
|
| 30621 |
+
},
|
| 30622 |
+
{
|
| 30623 |
+
"epoch": 0.86,
|
| 30624 |
+
"grad_norm": 1.9620647430419922,
|
| 30625 |
+
"learning_rate": 6.786402490241612e-06,
|
| 30626 |
+
"loss": 5.4566,
|
| 30627 |
+
"step": 17496
|
| 30628 |
+
},
|
| 30629 |
+
{
|
| 30630 |
+
"epoch": 0.86,
|
| 30631 |
+
"grad_norm": 2.0902743339538574,
|
| 30632 |
+
"learning_rate": 6.776520579079993e-06,
|
| 30633 |
+
"loss": 5.484,
|
| 30634 |
+
"step": 17500
|
| 30635 |
+
},
|
| 30636 |
+
{
|
| 30637 |
+
"epoch": 0.86,
|
| 30638 |
+
"grad_norm": 2.329399824142456,
|
| 30639 |
+
"learning_rate": 6.766638667918376e-06,
|
| 30640 |
+
"loss": 5.5337,
|
| 30641 |
+
"step": 17504
|
| 30642 |
+
},
|
| 30643 |
+
{
|
| 30644 |
+
"epoch": 0.87,
|
| 30645 |
+
"grad_norm": 1.9751675128936768,
|
| 30646 |
+
"learning_rate": 6.7567567567567575e-06,
|
| 30647 |
+
"loss": 5.4638,
|
| 30648 |
+
"step": 17508
|
| 30649 |
+
},
|
| 30650 |
+
{
|
| 30651 |
+
"epoch": 0.87,
|
| 30652 |
+
"grad_norm": 2.188885450363159,
|
| 30653 |
+
"learning_rate": 6.746874845595139e-06,
|
| 30654 |
+
"loss": 5.5174,
|
| 30655 |
+
"step": 17512
|
| 30656 |
+
},
|
| 30657 |
+
{
|
| 30658 |
+
"epoch": 0.87,
|
| 30659 |
+
"grad_norm": 2.1230249404907227,
|
| 30660 |
+
"learning_rate": 6.73699293443352e-06,
|
| 30661 |
+
"loss": 5.4525,
|
| 30662 |
+
"step": 17516
|
| 30663 |
+
},
|
| 30664 |
+
{
|
| 30665 |
+
"epoch": 0.87,
|
| 30666 |
+
"grad_norm": 2.0748202800750732,
|
| 30667 |
+
"learning_rate": 6.727111023271901e-06,
|
| 30668 |
+
"loss": 5.5633,
|
| 30669 |
+
"step": 17520
|
| 30670 |
+
},
|
| 30671 |
+
{
|
| 30672 |
+
"epoch": 0.87,
|
| 30673 |
+
"grad_norm": 1.8766546249389648,
|
| 30674 |
+
"learning_rate": 6.717229112110282e-06,
|
| 30675 |
+
"loss": 5.5295,
|
| 30676 |
+
"step": 17524
|
| 30677 |
+
},
|
| 30678 |
+
{
|
| 30679 |
+
"epoch": 0.87,
|
| 30680 |
+
"grad_norm": 2.1104044914245605,
|
| 30681 |
+
"learning_rate": 6.707347200948664e-06,
|
| 30682 |
+
"loss": 5.5265,
|
| 30683 |
+
"step": 17528
|
| 30684 |
+
},
|
| 30685 |
+
{
|
| 30686 |
+
"epoch": 0.87,
|
| 30687 |
+
"grad_norm": 1.9020673036575317,
|
| 30688 |
+
"learning_rate": 6.697465289787046e-06,
|
| 30689 |
+
"loss": 5.4684,
|
| 30690 |
+
"step": 17532
|
| 30691 |
+
},
|
| 30692 |
+
{
|
| 30693 |
+
"epoch": 0.87,
|
| 30694 |
+
"grad_norm": 1.9456652402877808,
|
| 30695 |
+
"learning_rate": 6.687583378625427e-06,
|
| 30696 |
+
"loss": 5.3826,
|
| 30697 |
+
"step": 17536
|
| 30698 |
+
},
|
| 30699 |
+
{
|
| 30700 |
+
"epoch": 0.87,
|
| 30701 |
+
"grad_norm": 2.117117166519165,
|
| 30702 |
+
"learning_rate": 6.677701467463808e-06,
|
| 30703 |
+
"loss": 5.4525,
|
| 30704 |
+
"step": 17540
|
| 30705 |
+
},
|
| 30706 |
+
{
|
| 30707 |
+
"epoch": 0.87,
|
| 30708 |
+
"grad_norm": 2.0873782634735107,
|
| 30709 |
+
"learning_rate": 6.667819556302189e-06,
|
| 30710 |
+
"loss": 5.5402,
|
| 30711 |
+
"step": 17544
|
| 30712 |
+
},
|
| 30713 |
+
{
|
| 30714 |
+
"epoch": 0.87,
|
| 30715 |
+
"grad_norm": 2.0289838314056396,
|
| 30716 |
+
"learning_rate": 6.6579376451405705e-06,
|
| 30717 |
+
"loss": 5.5603,
|
| 30718 |
+
"step": 17548
|
| 30719 |
+
},
|
| 30720 |
+
{
|
| 30721 |
+
"epoch": 0.87,
|
| 30722 |
+
"grad_norm": 2.2275471687316895,
|
| 30723 |
+
"learning_rate": 6.648055733978952e-06,
|
| 30724 |
+
"loss": 5.3792,
|
| 30725 |
+
"step": 17552
|
| 30726 |
+
},
|
| 30727 |
+
{
|
| 30728 |
+
"epoch": 0.87,
|
| 30729 |
+
"grad_norm": 1.9133155345916748,
|
| 30730 |
+
"learning_rate": 6.638173822817333e-06,
|
| 30731 |
+
"loss": 5.4289,
|
| 30732 |
+
"step": 17556
|
| 30733 |
+
},
|
| 30734 |
+
{
|
| 30735 |
+
"epoch": 0.87,
|
| 30736 |
+
"grad_norm": 2.193645477294922,
|
| 30737 |
+
"learning_rate": 6.628291911655715e-06,
|
| 30738 |
+
"loss": 5.5224,
|
| 30739 |
+
"step": 17560
|
| 30740 |
+
},
|
| 30741 |
+
{
|
| 30742 |
+
"epoch": 0.87,
|
| 30743 |
+
"grad_norm": 2.1608972549438477,
|
| 30744 |
+
"learning_rate": 6.618410000494096e-06,
|
| 30745 |
+
"loss": 5.5412,
|
| 30746 |
+
"step": 17564
|
| 30747 |
+
},
|
| 30748 |
+
{
|
| 30749 |
+
"epoch": 0.87,
|
| 30750 |
+
"grad_norm": 2.141594648361206,
|
| 30751 |
+
"learning_rate": 6.6085280893324774e-06,
|
| 30752 |
+
"loss": 5.4787,
|
| 30753 |
+
"step": 17568
|
| 30754 |
+
},
|
| 30755 |
+
{
|
| 30756 |
+
"epoch": 0.87,
|
| 30757 |
+
"grad_norm": 1.9416935443878174,
|
| 30758 |
+
"learning_rate": 6.598646178170859e-06,
|
| 30759 |
+
"loss": 5.3674,
|
| 30760 |
+
"step": 17572
|
| 30761 |
+
},
|
| 30762 |
+
{
|
| 30763 |
+
"epoch": 0.87,
|
| 30764 |
+
"grad_norm": 2.110677480697632,
|
| 30765 |
+
"learning_rate": 6.58876426700924e-06,
|
| 30766 |
+
"loss": 5.4024,
|
| 30767 |
+
"step": 17576
|
| 30768 |
+
},
|
| 30769 |
+
{
|
| 30770 |
+
"epoch": 0.87,
|
| 30771 |
+
"grad_norm": 2.2235372066497803,
|
| 30772 |
+
"learning_rate": 6.578882355847621e-06,
|
| 30773 |
+
"loss": 5.4458,
|
| 30774 |
+
"step": 17580
|
| 30775 |
+
},
|
| 30776 |
+
{
|
| 30777 |
+
"epoch": 0.87,
|
| 30778 |
+
"grad_norm": 2.280282974243164,
|
| 30779 |
+
"learning_rate": 6.569000444686002e-06,
|
| 30780 |
+
"loss": 5.5981,
|
| 30781 |
+
"step": 17584
|
| 30782 |
+
},
|
| 30783 |
+
{
|
| 30784 |
+
"epoch": 0.87,
|
| 30785 |
+
"grad_norm": 2.1084625720977783,
|
| 30786 |
+
"learning_rate": 6.559118533524384e-06,
|
| 30787 |
+
"loss": 5.4475,
|
| 30788 |
+
"step": 17588
|
| 30789 |
+
},
|
| 30790 |
+
{
|
| 30791 |
+
"epoch": 0.87,
|
| 30792 |
+
"grad_norm": 2.004232406616211,
|
| 30793 |
+
"learning_rate": 6.5492366223627656e-06,
|
| 30794 |
+
"loss": 5.4314,
|
| 30795 |
+
"step": 17592
|
| 30796 |
+
},
|
| 30797 |
+
{
|
| 30798 |
+
"epoch": 0.87,
|
| 30799 |
+
"grad_norm": 1.9286199808120728,
|
| 30800 |
+
"learning_rate": 6.539354711201147e-06,
|
| 30801 |
+
"loss": 5.3943,
|
| 30802 |
+
"step": 17596
|
| 30803 |
+
},
|
| 30804 |
+
{
|
| 30805 |
+
"epoch": 0.87,
|
| 30806 |
+
"grad_norm": 1.9742239713668823,
|
| 30807 |
+
"learning_rate": 6.529472800039528e-06,
|
| 30808 |
+
"loss": 5.4633,
|
| 30809 |
+
"step": 17600
|
| 30810 |
+
},
|
| 30811 |
+
{
|
| 30812 |
+
"epoch": 0.87,
|
| 30813 |
+
"grad_norm": 2.1503305435180664,
|
| 30814 |
+
"learning_rate": 6.519590888877909e-06,
|
| 30815 |
+
"loss": 5.4654,
|
| 30816 |
+
"step": 17604
|
| 30817 |
+
},
|
| 30818 |
+
{
|
| 30819 |
+
"epoch": 0.87,
|
| 30820 |
+
"grad_norm": 1.996319055557251,
|
| 30821 |
+
"learning_rate": 6.50970897771629e-06,
|
| 30822 |
+
"loss": 5.4479,
|
| 30823 |
+
"step": 17608
|
| 30824 |
+
},
|
| 30825 |
+
{
|
| 30826 |
+
"epoch": 0.87,
|
| 30827 |
+
"grad_norm": 2.1689870357513428,
|
| 30828 |
+
"learning_rate": 6.499827066554672e-06,
|
| 30829 |
+
"loss": 5.4242,
|
| 30830 |
+
"step": 17612
|
| 30831 |
+
},
|
| 30832 |
+
{
|
| 30833 |
+
"epoch": 0.87,
|
| 30834 |
+
"grad_norm": 2.0061464309692383,
|
| 30835 |
+
"learning_rate": 6.489945155393053e-06,
|
| 30836 |
+
"loss": 5.3833,
|
| 30837 |
+
"step": 17616
|
| 30838 |
+
},
|
| 30839 |
+
{
|
| 30840 |
+
"epoch": 0.87,
|
| 30841 |
+
"grad_norm": 2.1201388835906982,
|
| 30842 |
+
"learning_rate": 6.480063244231435e-06,
|
| 30843 |
+
"loss": 5.4374,
|
| 30844 |
+
"step": 17620
|
| 30845 |
+
},
|
| 30846 |
+
{
|
| 30847 |
+
"epoch": 0.87,
|
| 30848 |
+
"grad_norm": 2.196545124053955,
|
| 30849 |
+
"learning_rate": 6.470181333069816e-06,
|
| 30850 |
+
"loss": 5.3564,
|
| 30851 |
+
"step": 17624
|
| 30852 |
+
},
|
| 30853 |
+
{
|
| 30854 |
+
"epoch": 0.87,
|
| 30855 |
+
"grad_norm": 2.073232412338257,
|
| 30856 |
+
"learning_rate": 6.460299421908197e-06,
|
| 30857 |
+
"loss": 5.4057,
|
| 30858 |
+
"step": 17628
|
| 30859 |
+
},
|
| 30860 |
+
{
|
| 30861 |
+
"epoch": 0.87,
|
| 30862 |
+
"grad_norm": 1.9354524612426758,
|
| 30863 |
+
"learning_rate": 6.4504175107465785e-06,
|
| 30864 |
+
"loss": 5.4718,
|
| 30865 |
+
"step": 17632
|
| 30866 |
+
},
|
| 30867 |
+
{
|
| 30868 |
+
"epoch": 0.87,
|
| 30869 |
+
"grad_norm": 2.032994508743286,
|
| 30870 |
+
"learning_rate": 6.44053559958496e-06,
|
| 30871 |
+
"loss": 5.358,
|
| 30872 |
+
"step": 17636
|
| 30873 |
+
},
|
| 30874 |
+
{
|
| 30875 |
+
"epoch": 0.87,
|
| 30876 |
+
"grad_norm": 2.130598545074463,
|
| 30877 |
+
"learning_rate": 6.430653688423341e-06,
|
| 30878 |
+
"loss": 5.4658,
|
| 30879 |
+
"step": 17640
|
| 30880 |
+
},
|
| 30881 |
+
{
|
| 30882 |
+
"epoch": 0.87,
|
| 30883 |
+
"grad_norm": 1.8692468404769897,
|
| 30884 |
+
"learning_rate": 6.420771777261722e-06,
|
| 30885 |
+
"loss": 5.4004,
|
| 30886 |
+
"step": 17644
|
| 30887 |
+
},
|
| 30888 |
+
{
|
| 30889 |
+
"epoch": 0.87,
|
| 30890 |
+
"grad_norm": 1.9368531703948975,
|
| 30891 |
+
"learning_rate": 6.410889866100104e-06,
|
| 30892 |
+
"loss": 5.4809,
|
| 30893 |
+
"step": 17648
|
| 30894 |
+
},
|
| 30895 |
+
{
|
| 30896 |
+
"epoch": 0.87,
|
| 30897 |
+
"grad_norm": 2.235506534576416,
|
| 30898 |
+
"learning_rate": 6.4010079549384855e-06,
|
| 30899 |
+
"loss": 5.4532,
|
| 30900 |
+
"step": 17652
|
| 30901 |
+
},
|
| 30902 |
+
{
|
| 30903 |
+
"epoch": 0.87,
|
| 30904 |
+
"grad_norm": 1.9980324506759644,
|
| 30905 |
+
"learning_rate": 6.391126043776867e-06,
|
| 30906 |
+
"loss": 5.5156,
|
| 30907 |
+
"step": 17656
|
| 30908 |
+
},
|
| 30909 |
+
{
|
| 30910 |
+
"epoch": 0.87,
|
| 30911 |
+
"grad_norm": 1.947649598121643,
|
| 30912 |
+
"learning_rate": 6.381244132615248e-06,
|
| 30913 |
+
"loss": 5.3676,
|
| 30914 |
+
"step": 17660
|
| 30915 |
+
},
|
| 30916 |
+
{
|
| 30917 |
+
"epoch": 0.87,
|
| 30918 |
+
"grad_norm": 2.365041971206665,
|
| 30919 |
+
"learning_rate": 6.371362221453629e-06,
|
| 30920 |
+
"loss": 5.4387,
|
| 30921 |
+
"step": 17664
|
| 30922 |
+
},
|
| 30923 |
+
{
|
| 30924 |
+
"epoch": 0.87,
|
| 30925 |
+
"grad_norm": 2.2820627689361572,
|
| 30926 |
+
"learning_rate": 6.36148031029201e-06,
|
| 30927 |
+
"loss": 5.4541,
|
| 30928 |
+
"step": 17668
|
| 30929 |
+
},
|
| 30930 |
+
{
|
| 30931 |
+
"epoch": 0.87,
|
| 30932 |
+
"grad_norm": 2.02691650390625,
|
| 30933 |
+
"learning_rate": 6.3515983991303915e-06,
|
| 30934 |
+
"loss": 5.5084,
|
| 30935 |
+
"step": 17672
|
| 30936 |
+
},
|
| 30937 |
+
{
|
| 30938 |
+
"epoch": 0.87,
|
| 30939 |
+
"grad_norm": 2.0064783096313477,
|
| 30940 |
+
"learning_rate": 6.3417164879687744e-06,
|
| 30941 |
+
"loss": 5.5205,
|
| 30942 |
+
"step": 17676
|
| 30943 |
+
},
|
| 30944 |
+
{
|
| 30945 |
+
"epoch": 0.87,
|
| 30946 |
+
"grad_norm": 1.9961150884628296,
|
| 30947 |
+
"learning_rate": 6.331834576807155e-06,
|
| 30948 |
+
"loss": 5.3904,
|
| 30949 |
+
"step": 17680
|
| 30950 |
+
},
|
| 30951 |
+
{
|
| 30952 |
+
"epoch": 0.87,
|
| 30953 |
+
"grad_norm": 2.2273404598236084,
|
| 30954 |
+
"learning_rate": 6.321952665645536e-06,
|
| 30955 |
+
"loss": 5.4756,
|
| 30956 |
+
"step": 17684
|
| 30957 |
+
},
|
| 30958 |
+
{
|
| 30959 |
+
"epoch": 0.87,
|
| 30960 |
+
"grad_norm": 2.078472852706909,
|
| 30961 |
+
"learning_rate": 6.312070754483917e-06,
|
| 30962 |
+
"loss": 5.5535,
|
| 30963 |
+
"step": 17688
|
| 30964 |
+
},
|
| 30965 |
+
{
|
| 30966 |
+
"epoch": 0.87,
|
| 30967 |
+
"grad_norm": 1.9248629808425903,
|
| 30968 |
+
"learning_rate": 6.3021888433222984e-06,
|
| 30969 |
+
"loss": 5.4378,
|
| 30970 |
+
"step": 17692
|
| 30971 |
+
},
|
| 30972 |
+
{
|
| 30973 |
+
"epoch": 0.87,
|
| 30974 |
+
"grad_norm": 2.1985530853271484,
|
| 30975 |
+
"learning_rate": 6.29230693216068e-06,
|
| 30976 |
+
"loss": 5.4345,
|
| 30977 |
+
"step": 17696
|
| 30978 |
+
},
|
| 30979 |
+
{
|
| 30980 |
+
"epoch": 0.87,
|
| 30981 |
+
"grad_norm": 2.087536096572876,
|
| 30982 |
+
"learning_rate": 6.282425020999061e-06,
|
| 30983 |
+
"loss": 5.5135,
|
| 30984 |
+
"step": 17700
|
| 30985 |
+
},
|
| 30986 |
+
{
|
| 30987 |
+
"epoch": 0.87,
|
| 30988 |
+
"grad_norm": 2.0699515342712402,
|
| 30989 |
+
"learning_rate": 6.272543109837444e-06,
|
| 30990 |
+
"loss": 5.3821,
|
| 30991 |
+
"step": 17704
|
| 30992 |
+
},
|
| 30993 |
+
{
|
| 30994 |
+
"epoch": 0.87,
|
| 30995 |
+
"grad_norm": 2.309680223464966,
|
| 30996 |
+
"learning_rate": 6.262661198675825e-06,
|
| 30997 |
+
"loss": 5.3426,
|
| 30998 |
+
"step": 17708
|
| 30999 |
+
},
|
| 31000 |
+
{
|
| 31001 |
+
"epoch": 0.88,
|
| 31002 |
+
"grad_norm": 1.9877557754516602,
|
| 31003 |
+
"learning_rate": 6.252779287514206e-06,
|
| 31004 |
+
"loss": 5.554,
|
| 31005 |
+
"step": 17712
|
| 31006 |
+
},
|
| 31007 |
+
{
|
| 31008 |
+
"epoch": 0.88,
|
| 31009 |
+
"grad_norm": 2.1621484756469727,
|
| 31010 |
+
"learning_rate": 6.2428973763525866e-06,
|
| 31011 |
+
"loss": 5.4897,
|
| 31012 |
+
"step": 17716
|
| 31013 |
+
},
|
| 31014 |
+
{
|
| 31015 |
+
"epoch": 0.88,
|
| 31016 |
+
"grad_norm": 2.0568161010742188,
|
| 31017 |
+
"learning_rate": 6.233015465190968e-06,
|
| 31018 |
+
"loss": 5.4726,
|
| 31019 |
+
"step": 17720
|
| 31020 |
+
},
|
| 31021 |
+
{
|
| 31022 |
+
"epoch": 0.88,
|
| 31023 |
+
"grad_norm": 1.8659361600875854,
|
| 31024 |
+
"learning_rate": 6.22313355402935e-06,
|
| 31025 |
+
"loss": 5.461,
|
| 31026 |
+
"step": 17724
|
| 31027 |
+
},
|
| 31028 |
+
{
|
| 31029 |
+
"epoch": 0.88,
|
| 31030 |
+
"grad_norm": 1.8461517095565796,
|
| 31031 |
+
"learning_rate": 6.213251642867731e-06,
|
| 31032 |
+
"loss": 5.3407,
|
| 31033 |
+
"step": 17728
|
| 31034 |
+
},
|
| 31035 |
+
{
|
| 31036 |
+
"epoch": 0.88,
|
| 31037 |
+
"grad_norm": 2.2194485664367676,
|
| 31038 |
+
"learning_rate": 6.203369731706112e-06,
|
| 31039 |
+
"loss": 5.3863,
|
| 31040 |
+
"step": 17732
|
| 31041 |
+
},
|
| 31042 |
+
{
|
| 31043 |
+
"epoch": 0.88,
|
| 31044 |
+
"grad_norm": 2.2594525814056396,
|
| 31045 |
+
"learning_rate": 6.1934878205444935e-06,
|
| 31046 |
+
"loss": 5.4737,
|
| 31047 |
+
"step": 17736
|
| 31048 |
+
},
|
| 31049 |
+
{
|
| 31050 |
+
"epoch": 0.88,
|
| 31051 |
+
"grad_norm": 2.067777156829834,
|
| 31052 |
+
"learning_rate": 6.1836059093828755e-06,
|
| 31053 |
+
"loss": 5.4817,
|
| 31054 |
+
"step": 17740
|
| 31055 |
+
},
|
| 31056 |
+
{
|
| 31057 |
+
"epoch": 0.88,
|
| 31058 |
+
"grad_norm": 1.9696800708770752,
|
| 31059 |
+
"learning_rate": 6.173723998221257e-06,
|
| 31060 |
+
"loss": 5.5215,
|
| 31061 |
+
"step": 17744
|
| 31062 |
+
},
|
| 31063 |
+
{
|
| 31064 |
+
"epoch": 0.88,
|
| 31065 |
+
"grad_norm": 1.9700802564620972,
|
| 31066 |
+
"learning_rate": 6.163842087059637e-06,
|
| 31067 |
+
"loss": 5.5821,
|
| 31068 |
+
"step": 17748
|
| 31069 |
+
},
|
| 31070 |
+
{
|
| 31071 |
+
"epoch": 0.88,
|
| 31072 |
+
"grad_norm": 2.2519845962524414,
|
| 31073 |
+
"learning_rate": 6.153960175898018e-06,
|
| 31074 |
+
"loss": 5.5501,
|
| 31075 |
+
"step": 17752
|
| 31076 |
+
},
|
| 31077 |
+
{
|
| 31078 |
+
"epoch": 0.88,
|
| 31079 |
+
"grad_norm": 2.1531550884246826,
|
| 31080 |
+
"learning_rate": 6.1440782647364e-06,
|
| 31081 |
+
"loss": 5.354,
|
| 31082 |
+
"step": 17756
|
| 31083 |
+
},
|
| 31084 |
+
{
|
| 31085 |
+
"epoch": 0.88,
|
| 31086 |
+
"grad_norm": 2.299639940261841,
|
| 31087 |
+
"learning_rate": 6.134196353574782e-06,
|
| 31088 |
+
"loss": 5.5487,
|
| 31089 |
+
"step": 17760
|
| 31090 |
+
},
|
| 31091 |
+
{
|
| 31092 |
+
"epoch": 0.88,
|
| 31093 |
+
"grad_norm": 1.9032407999038696,
|
| 31094 |
+
"learning_rate": 6.124314442413163e-06,
|
| 31095 |
+
"loss": 5.4551,
|
| 31096 |
+
"step": 17764
|
| 31097 |
+
},
|
| 31098 |
+
{
|
| 31099 |
+
"epoch": 0.88,
|
| 31100 |
+
"grad_norm": 2.121720552444458,
|
| 31101 |
+
"learning_rate": 6.114432531251545e-06,
|
| 31102 |
+
"loss": 5.4185,
|
| 31103 |
+
"step": 17768
|
| 31104 |
+
},
|
| 31105 |
+
{
|
| 31106 |
+
"epoch": 0.88,
|
| 31107 |
+
"grad_norm": 1.955588698387146,
|
| 31108 |
+
"learning_rate": 6.104550620089926e-06,
|
| 31109 |
+
"loss": 5.5947,
|
| 31110 |
+
"step": 17772
|
| 31111 |
+
},
|
| 31112 |
+
{
|
| 31113 |
+
"epoch": 0.88,
|
| 31114 |
+
"grad_norm": 1.9518580436706543,
|
| 31115 |
+
"learning_rate": 6.094668708928307e-06,
|
| 31116 |
+
"loss": 5.4159,
|
| 31117 |
+
"step": 17776
|
| 31118 |
+
},
|
| 31119 |
+
{
|
| 31120 |
+
"epoch": 0.88,
|
| 31121 |
+
"grad_norm": 2.2284739017486572,
|
| 31122 |
+
"learning_rate": 6.084786797766688e-06,
|
| 31123 |
+
"loss": 5.4806,
|
| 31124 |
+
"step": 17780
|
| 31125 |
+
},
|
| 31126 |
+
{
|
| 31127 |
+
"epoch": 0.88,
|
| 31128 |
+
"grad_norm": 1.9473198652267456,
|
| 31129 |
+
"learning_rate": 6.07490488660507e-06,
|
| 31130 |
+
"loss": 5.3457,
|
| 31131 |
+
"step": 17784
|
| 31132 |
+
},
|
| 31133 |
+
{
|
| 31134 |
+
"epoch": 0.88,
|
| 31135 |
+
"grad_norm": 2.25762939453125,
|
| 31136 |
+
"learning_rate": 6.065022975443451e-06,
|
| 31137 |
+
"loss": 5.3905,
|
| 31138 |
+
"step": 17788
|
| 31139 |
+
},
|
| 31140 |
+
{
|
| 31141 |
+
"epoch": 0.88,
|
| 31142 |
+
"grad_norm": 2.13055682182312,
|
| 31143 |
+
"learning_rate": 6.055141064281832e-06,
|
| 31144 |
+
"loss": 5.4596,
|
| 31145 |
+
"step": 17792
|
| 31146 |
+
},
|
| 31147 |
+
{
|
| 31148 |
+
"epoch": 0.88,
|
| 31149 |
+
"grad_norm": 2.078608751296997,
|
| 31150 |
+
"learning_rate": 6.045259153120214e-06,
|
| 31151 |
+
"loss": 5.4836,
|
| 31152 |
+
"step": 17796
|
| 31153 |
+
},
|
| 31154 |
+
{
|
| 31155 |
+
"epoch": 0.88,
|
| 31156 |
+
"grad_norm": 2.032860040664673,
|
| 31157 |
+
"learning_rate": 6.0353772419585954e-06,
|
| 31158 |
+
"loss": 5.3315,
|
| 31159 |
+
"step": 17800
|
| 31160 |
+
},
|
| 31161 |
+
{
|
| 31162 |
+
"epoch": 0.88,
|
| 31163 |
+
"grad_norm": 2.18186616897583,
|
| 31164 |
+
"learning_rate": 6.025495330796977e-06,
|
| 31165 |
+
"loss": 5.5165,
|
| 31166 |
+
"step": 17804
|
| 31167 |
+
},
|
| 31168 |
+
{
|
| 31169 |
+
"epoch": 0.88,
|
| 31170 |
+
"grad_norm": 1.8949894905090332,
|
| 31171 |
+
"learning_rate": 6.015613419635358e-06,
|
| 31172 |
+
"loss": 5.4296,
|
| 31173 |
+
"step": 17808
|
| 31174 |
+
},
|
| 31175 |
+
{
|
| 31176 |
+
"epoch": 0.88,
|
| 31177 |
+
"grad_norm": 1.9019147157669067,
|
| 31178 |
+
"learning_rate": 6.005731508473739e-06,
|
| 31179 |
+
"loss": 5.3742,
|
| 31180 |
+
"step": 17812
|
| 31181 |
+
},
|
| 31182 |
+
{
|
| 31183 |
+
"epoch": 0.88,
|
| 31184 |
+
"grad_norm": 1.9749938249588013,
|
| 31185 |
+
"learning_rate": 5.99584959731212e-06,
|
| 31186 |
+
"loss": 5.4702,
|
| 31187 |
+
"step": 17816
|
| 31188 |
+
},
|
| 31189 |
+
{
|
| 31190 |
+
"epoch": 0.88,
|
| 31191 |
+
"grad_norm": 1.9528026580810547,
|
| 31192 |
+
"learning_rate": 5.9859676861505015e-06,
|
| 31193 |
+
"loss": 5.4179,
|
| 31194 |
+
"step": 17820
|
| 31195 |
+
},
|
| 31196 |
+
{
|
| 31197 |
+
"epoch": 0.88,
|
| 31198 |
+
"grad_norm": 2.04555082321167,
|
| 31199 |
+
"learning_rate": 5.976085774988883e-06,
|
| 31200 |
+
"loss": 5.4807,
|
| 31201 |
+
"step": 17824
|
| 31202 |
+
},
|
| 31203 |
+
{
|
| 31204 |
+
"epoch": 0.88,
|
| 31205 |
+
"grad_norm": 2.2078750133514404,
|
| 31206 |
+
"learning_rate": 5.966203863827265e-06,
|
| 31207 |
+
"loss": 5.4206,
|
| 31208 |
+
"step": 17828
|
| 31209 |
+
},
|
| 31210 |
+
{
|
| 31211 |
+
"epoch": 0.88,
|
| 31212 |
+
"grad_norm": 2.1232731342315674,
|
| 31213 |
+
"learning_rate": 5.956321952665646e-06,
|
| 31214 |
+
"loss": 5.5452,
|
| 31215 |
+
"step": 17832
|
| 31216 |
+
},
|
| 31217 |
+
{
|
| 31218 |
+
"epoch": 0.88,
|
| 31219 |
+
"grad_norm": 1.724265217781067,
|
| 31220 |
+
"learning_rate": 5.946440041504027e-06,
|
| 31221 |
+
"loss": 5.3831,
|
| 31222 |
+
"step": 17836
|
| 31223 |
+
},
|
| 31224 |
+
{
|
| 31225 |
+
"epoch": 0.88,
|
| 31226 |
+
"grad_norm": 2.0802602767944336,
|
| 31227 |
+
"learning_rate": 5.936558130342408e-06,
|
| 31228 |
+
"loss": 5.3925,
|
| 31229 |
+
"step": 17840
|
| 31230 |
+
},
|
| 31231 |
+
{
|
| 31232 |
+
"epoch": 0.88,
|
| 31233 |
+
"grad_norm": 1.913464069366455,
|
| 31234 |
+
"learning_rate": 5.92667621918079e-06,
|
| 31235 |
+
"loss": 5.4786,
|
| 31236 |
+
"step": 17844
|
| 31237 |
+
},
|
| 31238 |
+
{
|
| 31239 |
+
"epoch": 0.88,
|
| 31240 |
+
"grad_norm": 1.9357552528381348,
|
| 31241 |
+
"learning_rate": 5.916794308019171e-06,
|
| 31242 |
+
"loss": 5.4415,
|
| 31243 |
+
"step": 17848
|
| 31244 |
+
},
|
| 31245 |
+
{
|
| 31246 |
+
"epoch": 0.88,
|
| 31247 |
+
"grad_norm": 1.9869678020477295,
|
| 31248 |
+
"learning_rate": 5.906912396857552e-06,
|
| 31249 |
+
"loss": 5.4731,
|
| 31250 |
+
"step": 17852
|
| 31251 |
+
},
|
| 31252 |
+
{
|
| 31253 |
+
"epoch": 0.88,
|
| 31254 |
+
"grad_norm": 2.1964402198791504,
|
| 31255 |
+
"learning_rate": 5.897030485695934e-06,
|
| 31256 |
+
"loss": 5.4538,
|
| 31257 |
+
"step": 17856
|
| 31258 |
+
},
|
| 31259 |
+
{
|
| 31260 |
+
"epoch": 0.88,
|
| 31261 |
+
"grad_norm": 2.2341887950897217,
|
| 31262 |
+
"learning_rate": 5.887148574534315e-06,
|
| 31263 |
+
"loss": 5.4484,
|
| 31264 |
+
"step": 17860
|
| 31265 |
+
},
|
| 31266 |
+
{
|
| 31267 |
+
"epoch": 0.88,
|
| 31268 |
+
"grad_norm": 2.2685790061950684,
|
| 31269 |
+
"learning_rate": 5.8772666633726965e-06,
|
| 31270 |
+
"loss": 5.4935,
|
| 31271 |
+
"step": 17864
|
| 31272 |
+
},
|
| 31273 |
+
{
|
| 31274 |
+
"epoch": 0.88,
|
| 31275 |
+
"grad_norm": 1.9977366924285889,
|
| 31276 |
+
"learning_rate": 5.867384752211079e-06,
|
| 31277 |
+
"loss": 5.4866,
|
| 31278 |
+
"step": 17868
|
| 31279 |
+
},
|
| 31280 |
+
{
|
| 31281 |
+
"epoch": 0.88,
|
| 31282 |
+
"grad_norm": 1.8025336265563965,
|
| 31283 |
+
"learning_rate": 5.857502841049459e-06,
|
| 31284 |
+
"loss": 5.5315,
|
| 31285 |
+
"step": 17872
|
| 31286 |
+
},
|
| 31287 |
+
{
|
| 31288 |
+
"epoch": 0.88,
|
| 31289 |
+
"grad_norm": 1.8028703927993774,
|
| 31290 |
+
"learning_rate": 5.84762092988784e-06,
|
| 31291 |
+
"loss": 5.4048,
|
| 31292 |
+
"step": 17876
|
| 31293 |
+
},
|
| 31294 |
+
{
|
| 31295 |
+
"epoch": 0.88,
|
| 31296 |
+
"grad_norm": 2.0206375122070312,
|
| 31297 |
+
"learning_rate": 5.837739018726221e-06,
|
| 31298 |
+
"loss": 5.5057,
|
| 31299 |
+
"step": 17880
|
| 31300 |
+
},
|
| 31301 |
+
{
|
| 31302 |
+
"epoch": 0.88,
|
| 31303 |
+
"grad_norm": 2.2146549224853516,
|
| 31304 |
+
"learning_rate": 5.8278571075646034e-06,
|
| 31305 |
+
"loss": 5.4375,
|
| 31306 |
+
"step": 17884
|
| 31307 |
+
},
|
| 31308 |
+
{
|
| 31309 |
+
"epoch": 0.88,
|
| 31310 |
+
"grad_norm": 2.1485488414764404,
|
| 31311 |
+
"learning_rate": 5.817975196402985e-06,
|
| 31312 |
+
"loss": 5.4375,
|
| 31313 |
+
"step": 17888
|
| 31314 |
+
},
|
| 31315 |
+
{
|
| 31316 |
+
"epoch": 0.88,
|
| 31317 |
+
"grad_norm": 1.9976389408111572,
|
| 31318 |
+
"learning_rate": 5.808093285241366e-06,
|
| 31319 |
+
"loss": 5.5645,
|
| 31320 |
+
"step": 17892
|
| 31321 |
+
},
|
| 31322 |
+
{
|
| 31323 |
+
"epoch": 0.88,
|
| 31324 |
+
"grad_norm": 2.1577677726745605,
|
| 31325 |
+
"learning_rate": 5.798211374079747e-06,
|
| 31326 |
+
"loss": 5.4917,
|
| 31327 |
+
"step": 17896
|
| 31328 |
+
},
|
| 31329 |
+
{
|
| 31330 |
+
"epoch": 0.88,
|
| 31331 |
+
"grad_norm": 2.085784673690796,
|
| 31332 |
+
"learning_rate": 5.788329462918129e-06,
|
| 31333 |
+
"loss": 5.5115,
|
| 31334 |
+
"step": 17900
|
| 31335 |
+
},
|
| 31336 |
+
{
|
| 31337 |
+
"epoch": 0.88,
|
| 31338 |
+
"grad_norm": 2.250061511993408,
|
| 31339 |
+
"learning_rate": 5.77844755175651e-06,
|
| 31340 |
+
"loss": 5.4932,
|
| 31341 |
+
"step": 17904
|
| 31342 |
+
},
|
| 31343 |
+
{
|
| 31344 |
+
"epoch": 0.88,
|
| 31345 |
+
"grad_norm": 2.076542854309082,
|
| 31346 |
+
"learning_rate": 5.768565640594891e-06,
|
| 31347 |
+
"loss": 5.3858,
|
| 31348 |
+
"step": 17908
|
| 31349 |
+
},
|
| 31350 |
+
{
|
| 31351 |
+
"epoch": 0.89,
|
| 31352 |
+
"grad_norm": 2.0458431243896484,
|
| 31353 |
+
"learning_rate": 5.761154207223677e-06,
|
| 31354 |
+
"loss": 5.5596,
|
| 31355 |
+
"step": 17912
|
| 31356 |
+
},
|
| 31357 |
+
{
|
| 31358 |
+
"epoch": 0.89,
|
| 31359 |
+
"grad_norm": 2.075693130493164,
|
| 31360 |
+
"learning_rate": 5.7512722960620585e-06,
|
| 31361 |
+
"loss": 5.4962,
|
| 31362 |
+
"step": 17916
|
| 31363 |
+
},
|
| 31364 |
+
{
|
| 31365 |
+
"epoch": 0.89,
|
| 31366 |
+
"grad_norm": 2.09002685546875,
|
| 31367 |
+
"learning_rate": 5.74139038490044e-06,
|
| 31368 |
+
"loss": 5.3926,
|
| 31369 |
+
"step": 17920
|
| 31370 |
+
},
|
| 31371 |
+
{
|
| 31372 |
+
"epoch": 0.89,
|
| 31373 |
+
"grad_norm": 2.0116419792175293,
|
| 31374 |
+
"learning_rate": 5.731508473738821e-06,
|
| 31375 |
+
"loss": 5.4649,
|
| 31376 |
+
"step": 17924
|
| 31377 |
+
},
|
| 31378 |
+
{
|
| 31379 |
+
"epoch": 0.89,
|
| 31380 |
+
"grad_norm": 2.0170979499816895,
|
| 31381 |
+
"learning_rate": 5.721626562577203e-06,
|
| 31382 |
+
"loss": 5.4906,
|
| 31383 |
+
"step": 17928
|
| 31384 |
+
},
|
| 31385 |
+
{
|
| 31386 |
+
"epoch": 0.89,
|
| 31387 |
+
"grad_norm": 2.033344268798828,
|
| 31388 |
+
"learning_rate": 5.711744651415584e-06,
|
| 31389 |
+
"loss": 5.5144,
|
| 31390 |
+
"step": 17932
|
| 31391 |
+
},
|
| 31392 |
+
{
|
| 31393 |
+
"epoch": 0.89,
|
| 31394 |
+
"grad_norm": 1.9842782020568848,
|
| 31395 |
+
"learning_rate": 5.701862740253965e-06,
|
| 31396 |
+
"loss": 5.4478,
|
| 31397 |
+
"step": 17936
|
| 31398 |
+
},
|
| 31399 |
+
{
|
| 31400 |
+
"epoch": 0.89,
|
| 31401 |
+
"grad_norm": 2.059737205505371,
|
| 31402 |
+
"learning_rate": 5.691980829092347e-06,
|
| 31403 |
+
"loss": 5.4969,
|
| 31404 |
+
"step": 17940
|
| 31405 |
+
},
|
| 31406 |
+
{
|
| 31407 |
+
"epoch": 0.89,
|
| 31408 |
+
"grad_norm": 2.116508960723877,
|
| 31409 |
+
"learning_rate": 5.682098917930728e-06,
|
| 31410 |
+
"loss": 5.5734,
|
| 31411 |
+
"step": 17944
|
| 31412 |
+
},
|
| 31413 |
+
{
|
| 31414 |
+
"epoch": 0.89,
|
| 31415 |
+
"grad_norm": 2.062220573425293,
|
| 31416 |
+
"learning_rate": 5.672217006769109e-06,
|
| 31417 |
+
"loss": 5.3472,
|
| 31418 |
+
"step": 17948
|
| 31419 |
+
},
|
| 31420 |
+
{
|
| 31421 |
+
"epoch": 0.89,
|
| 31422 |
+
"grad_norm": 2.176339626312256,
|
| 31423 |
+
"learning_rate": 5.66233509560749e-06,
|
| 31424 |
+
"loss": 5.3356,
|
| 31425 |
+
"step": 17952
|
| 31426 |
+
},
|
| 31427 |
+
{
|
| 31428 |
+
"epoch": 0.89,
|
| 31429 |
+
"grad_norm": 2.2417047023773193,
|
| 31430 |
+
"learning_rate": 5.652453184445872e-06,
|
| 31431 |
+
"loss": 5.429,
|
| 31432 |
+
"step": 17956
|
| 31433 |
+
},
|
| 31434 |
+
{
|
| 31435 |
+
"epoch": 0.89,
|
| 31436 |
+
"grad_norm": 1.9376815557479858,
|
| 31437 |
+
"learning_rate": 5.6425712732842535e-06,
|
| 31438 |
+
"loss": 5.4263,
|
| 31439 |
+
"step": 17960
|
| 31440 |
+
},
|
| 31441 |
+
{
|
| 31442 |
+
"epoch": 0.89,
|
| 31443 |
+
"grad_norm": 1.9888767004013062,
|
| 31444 |
+
"learning_rate": 5.632689362122635e-06,
|
| 31445 |
+
"loss": 5.5286,
|
| 31446 |
+
"step": 17964
|
| 31447 |
+
},
|
| 31448 |
+
{
|
| 31449 |
+
"epoch": 0.89,
|
| 31450 |
+
"grad_norm": 2.1054704189300537,
|
| 31451 |
+
"learning_rate": 5.622807450961016e-06,
|
| 31452 |
+
"loss": 5.4473,
|
| 31453 |
+
"step": 17968
|
| 31454 |
+
},
|
| 31455 |
+
{
|
| 31456 |
+
"epoch": 0.89,
|
| 31457 |
+
"grad_norm": 2.3070156574249268,
|
| 31458 |
+
"learning_rate": 5.612925539799398e-06,
|
| 31459 |
+
"loss": 5.6394,
|
| 31460 |
+
"step": 17972
|
| 31461 |
+
},
|
| 31462 |
+
{
|
| 31463 |
+
"epoch": 0.89,
|
| 31464 |
+
"grad_norm": 1.956694483757019,
|
| 31465 |
+
"learning_rate": 5.603043628637779e-06,
|
| 31466 |
+
"loss": 5.3773,
|
| 31467 |
+
"step": 17976
|
| 31468 |
+
},
|
| 31469 |
+
{
|
| 31470 |
+
"epoch": 0.89,
|
| 31471 |
+
"grad_norm": 2.1512622833251953,
|
| 31472 |
+
"learning_rate": 5.59316171747616e-06,
|
| 31473 |
+
"loss": 5.5051,
|
| 31474 |
+
"step": 17980
|
| 31475 |
+
},
|
| 31476 |
+
{
|
| 31477 |
+
"epoch": 0.89,
|
| 31478 |
+
"grad_norm": 1.9715803861618042,
|
| 31479 |
+
"learning_rate": 5.583279806314542e-06,
|
| 31480 |
+
"loss": 5.4564,
|
| 31481 |
+
"step": 17984
|
| 31482 |
+
},
|
| 31483 |
+
{
|
| 31484 |
+
"epoch": 0.89,
|
| 31485 |
+
"grad_norm": 2.07094144821167,
|
| 31486 |
+
"learning_rate": 5.573397895152923e-06,
|
| 31487 |
+
"loss": 5.486,
|
| 31488 |
+
"step": 17988
|
| 31489 |
+
},
|
| 31490 |
+
{
|
| 31491 |
+
"epoch": 0.89,
|
| 31492 |
+
"grad_norm": 2.0776047706604004,
|
| 31493 |
+
"learning_rate": 5.563515983991304e-06,
|
| 31494 |
+
"loss": 5.4101,
|
| 31495 |
+
"step": 17992
|
| 31496 |
+
},
|
| 31497 |
+
{
|
| 31498 |
+
"epoch": 0.89,
|
| 31499 |
+
"grad_norm": 2.0641090869903564,
|
| 31500 |
+
"learning_rate": 5.553634072829685e-06,
|
| 31501 |
+
"loss": 5.4413,
|
| 31502 |
+
"step": 17996
|
| 31503 |
+
},
|
| 31504 |
+
{
|
| 31505 |
+
"epoch": 0.89,
|
| 31506 |
+
"grad_norm": 2.069200038909912,
|
| 31507 |
+
"learning_rate": 5.543752161668067e-06,
|
| 31508 |
+
"loss": 5.4239,
|
| 31509 |
+
"step": 18000
|
| 31510 |
+
},
|
| 31511 |
+
{
|
| 31512 |
+
"epoch": 0.89,
|
| 31513 |
+
"grad_norm": 1.9378856420516968,
|
| 31514 |
+
"learning_rate": 5.5338702505064486e-06,
|
| 31515 |
+
"loss": 5.4994,
|
| 31516 |
+
"step": 18004
|
| 31517 |
+
},
|
| 31518 |
+
{
|
| 31519 |
+
"epoch": 0.89,
|
| 31520 |
+
"grad_norm": 1.9170506000518799,
|
| 31521 |
+
"learning_rate": 5.52398833934483e-06,
|
| 31522 |
+
"loss": 5.4589,
|
| 31523 |
+
"step": 18008
|
| 31524 |
+
},
|
| 31525 |
+
{
|
| 31526 |
+
"epoch": 0.89,
|
| 31527 |
+
"grad_norm": 2.1123738288879395,
|
| 31528 |
+
"learning_rate": 5.514106428183211e-06,
|
| 31529 |
+
"loss": 5.4226,
|
| 31530 |
+
"step": 18012
|
| 31531 |
+
},
|
| 31532 |
+
{
|
| 31533 |
+
"epoch": 0.89,
|
| 31534 |
+
"grad_norm": 1.9895274639129639,
|
| 31535 |
+
"learning_rate": 5.504224517021592e-06,
|
| 31536 |
+
"loss": 5.522,
|
| 31537 |
+
"step": 18016
|
| 31538 |
+
},
|
| 31539 |
+
{
|
| 31540 |
+
"epoch": 0.89,
|
| 31541 |
+
"grad_norm": 2.216384172439575,
|
| 31542 |
+
"learning_rate": 5.494342605859973e-06,
|
| 31543 |
+
"loss": 5.4323,
|
| 31544 |
+
"step": 18020
|
| 31545 |
+
},
|
| 31546 |
+
{
|
| 31547 |
+
"epoch": 0.89,
|
| 31548 |
+
"grad_norm": 2.1332645416259766,
|
| 31549 |
+
"learning_rate": 5.484460694698355e-06,
|
| 31550 |
+
"loss": 5.3666,
|
| 31551 |
+
"step": 18024
|
| 31552 |
+
},
|
| 31553 |
+
{
|
| 31554 |
+
"epoch": 0.89,
|
| 31555 |
+
"grad_norm": 2.0537400245666504,
|
| 31556 |
+
"learning_rate": 5.474578783536737e-06,
|
| 31557 |
+
"loss": 5.4988,
|
| 31558 |
+
"step": 18028
|
| 31559 |
+
},
|
| 31560 |
+
{
|
| 31561 |
+
"epoch": 0.89,
|
| 31562 |
+
"grad_norm": 1.9897429943084717,
|
| 31563 |
+
"learning_rate": 5.464696872375118e-06,
|
| 31564 |
+
"loss": 5.4833,
|
| 31565 |
+
"step": 18032
|
| 31566 |
+
},
|
| 31567 |
+
{
|
| 31568 |
+
"epoch": 0.89,
|
| 31569 |
+
"grad_norm": 2.066513776779175,
|
| 31570 |
+
"learning_rate": 5.454814961213499e-06,
|
| 31571 |
+
"loss": 5.4636,
|
| 31572 |
+
"step": 18036
|
| 31573 |
+
},
|
| 31574 |
+
{
|
| 31575 |
+
"epoch": 0.89,
|
| 31576 |
+
"grad_norm": 2.251376152038574,
|
| 31577 |
+
"learning_rate": 5.44493305005188e-06,
|
| 31578 |
+
"loss": 5.5205,
|
| 31579 |
+
"step": 18040
|
| 31580 |
+
},
|
| 31581 |
+
{
|
| 31582 |
+
"epoch": 0.89,
|
| 31583 |
+
"grad_norm": 2.1560051441192627,
|
| 31584 |
+
"learning_rate": 5.4350511388902615e-06,
|
| 31585 |
+
"loss": 5.4888,
|
| 31586 |
+
"step": 18044
|
| 31587 |
+
},
|
| 31588 |
+
{
|
| 31589 |
+
"epoch": 0.89,
|
| 31590 |
+
"grad_norm": 1.9921746253967285,
|
| 31591 |
+
"learning_rate": 5.425169227728643e-06,
|
| 31592 |
+
"loss": 5.503,
|
| 31593 |
+
"step": 18048
|
| 31594 |
+
},
|
| 31595 |
+
{
|
| 31596 |
+
"epoch": 0.89,
|
| 31597 |
+
"grad_norm": 1.9806662797927856,
|
| 31598 |
+
"learning_rate": 5.415287316567024e-06,
|
| 31599 |
+
"loss": 5.5299,
|
| 31600 |
+
"step": 18052
|
| 31601 |
+
},
|
| 31602 |
+
{
|
| 31603 |
+
"epoch": 0.89,
|
| 31604 |
+
"grad_norm": 2.086308479309082,
|
| 31605 |
+
"learning_rate": 5.405405405405406e-06,
|
| 31606 |
+
"loss": 5.3775,
|
| 31607 |
+
"step": 18056
|
| 31608 |
+
},
|
| 31609 |
+
{
|
| 31610 |
+
"epoch": 0.89,
|
| 31611 |
+
"grad_norm": 1.9652711153030396,
|
| 31612 |
+
"learning_rate": 5.395523494243787e-06,
|
| 31613 |
+
"loss": 5.5683,
|
| 31614 |
+
"step": 18060
|
| 31615 |
+
},
|
| 31616 |
+
{
|
| 31617 |
+
"epoch": 0.89,
|
| 31618 |
+
"grad_norm": 1.8376736640930176,
|
| 31619 |
+
"learning_rate": 5.3856415830821685e-06,
|
| 31620 |
+
"loss": 5.3905,
|
| 31621 |
+
"step": 18064
|
| 31622 |
+
},
|
| 31623 |
+
{
|
| 31624 |
+
"epoch": 0.89,
|
| 31625 |
+
"grad_norm": 2.1515750885009766,
|
| 31626 |
+
"learning_rate": 5.37575967192055e-06,
|
| 31627 |
+
"loss": 5.421,
|
| 31628 |
+
"step": 18068
|
| 31629 |
+
},
|
| 31630 |
+
{
|
| 31631 |
+
"epoch": 0.89,
|
| 31632 |
+
"grad_norm": 2.18635630607605,
|
| 31633 |
+
"learning_rate": 5.365877760758931e-06,
|
| 31634 |
+
"loss": 5.4513,
|
| 31635 |
+
"step": 18072
|
| 31636 |
+
},
|
| 31637 |
+
{
|
| 31638 |
+
"epoch": 0.89,
|
| 31639 |
+
"grad_norm": 2.2322137355804443,
|
| 31640 |
+
"learning_rate": 5.355995849597312e-06,
|
| 31641 |
+
"loss": 5.5696,
|
| 31642 |
+
"step": 18076
|
| 31643 |
+
},
|
| 31644 |
+
{
|
| 31645 |
+
"epoch": 0.89,
|
| 31646 |
+
"grad_norm": 2.1388771533966064,
|
| 31647 |
+
"learning_rate": 5.346113938435693e-06,
|
| 31648 |
+
"loss": 5.5154,
|
| 31649 |
+
"step": 18080
|
| 31650 |
+
},
|
| 31651 |
+
{
|
| 31652 |
+
"epoch": 0.89,
|
| 31653 |
+
"grad_norm": 2.1032564640045166,
|
| 31654 |
+
"learning_rate": 5.336232027274075e-06,
|
| 31655 |
+
"loss": 5.4737,
|
| 31656 |
+
"step": 18084
|
| 31657 |
+
},
|
| 31658 |
+
{
|
| 31659 |
+
"epoch": 0.89,
|
| 31660 |
+
"grad_norm": 2.2419564723968506,
|
| 31661 |
+
"learning_rate": 5.326350116112457e-06,
|
| 31662 |
+
"loss": 5.4804,
|
| 31663 |
+
"step": 18088
|
| 31664 |
+
},
|
| 31665 |
+
{
|
| 31666 |
+
"epoch": 0.89,
|
| 31667 |
+
"grad_norm": 2.1092734336853027,
|
| 31668 |
+
"learning_rate": 5.316468204950838e-06,
|
| 31669 |
+
"loss": 5.523,
|
| 31670 |
+
"step": 18092
|
| 31671 |
+
},
|
| 31672 |
+
{
|
| 31673 |
+
"epoch": 0.89,
|
| 31674 |
+
"grad_norm": 1.9642736911773682,
|
| 31675 |
+
"learning_rate": 5.306586293789219e-06,
|
| 31676 |
+
"loss": 5.4615,
|
| 31677 |
+
"step": 18096
|
| 31678 |
+
},
|
| 31679 |
+
{
|
| 31680 |
+
"epoch": 0.89,
|
| 31681 |
+
"grad_norm": 2.285712480545044,
|
| 31682 |
+
"learning_rate": 5.296704382627601e-06,
|
| 31683 |
+
"loss": 5.4944,
|
| 31684 |
+
"step": 18100
|
| 31685 |
+
},
|
| 31686 |
+
{
|
| 31687 |
+
"epoch": 0.89,
|
| 31688 |
+
"grad_norm": 1.8048274517059326,
|
| 31689 |
+
"learning_rate": 5.2868224714659814e-06,
|
| 31690 |
+
"loss": 5.3293,
|
| 31691 |
+
"step": 18104
|
| 31692 |
+
},
|
| 31693 |
+
{
|
| 31694 |
+
"epoch": 0.89,
|
| 31695 |
+
"grad_norm": 1.9001215696334839,
|
| 31696 |
+
"learning_rate": 5.276940560304363e-06,
|
| 31697 |
+
"loss": 5.3799,
|
| 31698 |
+
"step": 18108
|
| 31699 |
+
},
|
| 31700 |
+
{
|
| 31701 |
+
"epoch": 0.89,
|
| 31702 |
+
"grad_norm": 2.052248954772949,
|
| 31703 |
+
"learning_rate": 5.267058649142744e-06,
|
| 31704 |
+
"loss": 5.54,
|
| 31705 |
+
"step": 18112
|
| 31706 |
+
},
|
| 31707 |
+
{
|
| 31708 |
+
"epoch": 0.9,
|
| 31709 |
+
"grad_norm": 1.9618264436721802,
|
| 31710 |
+
"learning_rate": 5.257176737981126e-06,
|
| 31711 |
+
"loss": 5.5206,
|
| 31712 |
+
"step": 18116
|
| 31713 |
+
},
|
| 31714 |
+
{
|
| 31715 |
+
"epoch": 0.9,
|
| 31716 |
+
"grad_norm": 2.086357355117798,
|
| 31717 |
+
"learning_rate": 5.247294826819507e-06,
|
| 31718 |
+
"loss": 5.4713,
|
| 31719 |
+
"step": 18120
|
| 31720 |
+
},
|
| 31721 |
+
{
|
| 31722 |
+
"epoch": 0.9,
|
| 31723 |
+
"grad_norm": 1.989790678024292,
|
| 31724 |
+
"learning_rate": 5.237412915657888e-06,
|
| 31725 |
+
"loss": 5.4183,
|
| 31726 |
+
"step": 18124
|
| 31727 |
+
},
|
| 31728 |
+
{
|
| 31729 |
+
"epoch": 0.9,
|
| 31730 |
+
"grad_norm": 2.4142305850982666,
|
| 31731 |
+
"learning_rate": 5.22753100449627e-06,
|
| 31732 |
+
"loss": 5.4867,
|
| 31733 |
+
"step": 18128
|
| 31734 |
+
},
|
| 31735 |
+
{
|
| 31736 |
+
"epoch": 0.9,
|
| 31737 |
+
"grad_norm": 1.9405925273895264,
|
| 31738 |
+
"learning_rate": 5.217649093334652e-06,
|
| 31739 |
+
"loss": 5.4259,
|
| 31740 |
+
"step": 18132
|
| 31741 |
+
},
|
| 31742 |
+
{
|
| 31743 |
+
"epoch": 0.9,
|
| 31744 |
+
"grad_norm": 1.9569774866104126,
|
| 31745 |
+
"learning_rate": 5.207767182173033e-06,
|
| 31746 |
+
"loss": 5.4619,
|
| 31747 |
+
"step": 18136
|
| 31748 |
+
},
|
| 31749 |
+
{
|
| 31750 |
+
"epoch": 0.9,
|
| 31751 |
+
"grad_norm": 2.1723146438598633,
|
| 31752 |
+
"learning_rate": 5.197885271011413e-06,
|
| 31753 |
+
"loss": 5.4719,
|
| 31754 |
+
"step": 18140
|
| 31755 |
+
},
|
| 31756 |
+
{
|
| 31757 |
+
"epoch": 0.9,
|
| 31758 |
+
"grad_norm": 2.3420112133026123,
|
| 31759 |
+
"learning_rate": 5.188003359849795e-06,
|
| 31760 |
+
"loss": 5.4274,
|
| 31761 |
+
"step": 18144
|
| 31762 |
+
},
|
| 31763 |
+
{
|
| 31764 |
+
"epoch": 0.9,
|
| 31765 |
+
"grad_norm": 2.2983791828155518,
|
| 31766 |
+
"learning_rate": 5.1781214486881765e-06,
|
| 31767 |
+
"loss": 5.5184,
|
| 31768 |
+
"step": 18148
|
| 31769 |
+
},
|
| 31770 |
+
{
|
| 31771 |
+
"epoch": 0.9,
|
| 31772 |
+
"grad_norm": 2.061795711517334,
|
| 31773 |
+
"learning_rate": 5.168239537526558e-06,
|
| 31774 |
+
"loss": 5.4416,
|
| 31775 |
+
"step": 18152
|
| 31776 |
+
},
|
| 31777 |
+
{
|
| 31778 |
+
"epoch": 0.9,
|
| 31779 |
+
"grad_norm": 2.11879301071167,
|
| 31780 |
+
"learning_rate": 5.15835762636494e-06,
|
| 31781 |
+
"loss": 5.3997,
|
| 31782 |
+
"step": 18156
|
| 31783 |
+
},
|
| 31784 |
+
{
|
| 31785 |
+
"epoch": 0.9,
|
| 31786 |
+
"grad_norm": 2.072601556777954,
|
| 31787 |
+
"learning_rate": 5.148475715203321e-06,
|
| 31788 |
+
"loss": 5.4744,
|
| 31789 |
+
"step": 18160
|
| 31790 |
+
},
|
| 31791 |
+
{
|
| 31792 |
+
"epoch": 0.9,
|
| 31793 |
+
"grad_norm": 2.037374258041382,
|
| 31794 |
+
"learning_rate": 5.138593804041702e-06,
|
| 31795 |
+
"loss": 5.5549,
|
| 31796 |
+
"step": 18164
|
| 31797 |
+
},
|
| 31798 |
+
{
|
| 31799 |
+
"epoch": 0.9,
|
| 31800 |
+
"grad_norm": 2.1012215614318848,
|
| 31801 |
+
"learning_rate": 5.128711892880083e-06,
|
| 31802 |
+
"loss": 5.5601,
|
| 31803 |
+
"step": 18168
|
| 31804 |
+
},
|
| 31805 |
+
{
|
| 31806 |
+
"epoch": 0.9,
|
| 31807 |
+
"grad_norm": 1.9614689350128174,
|
| 31808 |
+
"learning_rate": 5.118829981718465e-06,
|
| 31809 |
+
"loss": 5.3775,
|
| 31810 |
+
"step": 18172
|
| 31811 |
+
},
|
| 31812 |
+
{
|
| 31813 |
+
"epoch": 0.9,
|
| 31814 |
+
"grad_norm": 2.3375091552734375,
|
| 31815 |
+
"learning_rate": 5.108948070556846e-06,
|
| 31816 |
+
"loss": 5.5373,
|
| 31817 |
+
"step": 18176
|
| 31818 |
+
},
|
| 31819 |
+
{
|
| 31820 |
+
"epoch": 0.9,
|
| 31821 |
+
"grad_norm": 1.9838519096374512,
|
| 31822 |
+
"learning_rate": 5.099066159395227e-06,
|
| 31823 |
+
"loss": 5.3491,
|
| 31824 |
+
"step": 18180
|
| 31825 |
+
},
|
| 31826 |
+
{
|
| 31827 |
+
"epoch": 0.9,
|
| 31828 |
+
"grad_norm": 2.0909204483032227,
|
| 31829 |
+
"learning_rate": 5.089184248233608e-06,
|
| 31830 |
+
"loss": 5.3386,
|
| 31831 |
+
"step": 18184
|
| 31832 |
+
},
|
| 31833 |
+
{
|
| 31834 |
+
"epoch": 0.9,
|
| 31835 |
+
"grad_norm": 2.396127223968506,
|
| 31836 |
+
"learning_rate": 5.07930233707199e-06,
|
| 31837 |
+
"loss": 5.4791,
|
| 31838 |
+
"step": 18188
|
| 31839 |
+
},
|
| 31840 |
+
{
|
| 31841 |
+
"epoch": 0.9,
|
| 31842 |
+
"grad_norm": 2.117344379425049,
|
| 31843 |
+
"learning_rate": 5.0694204259103715e-06,
|
| 31844 |
+
"loss": 5.3322,
|
| 31845 |
+
"step": 18192
|
| 31846 |
+
},
|
| 31847 |
+
{
|
| 31848 |
+
"epoch": 0.9,
|
| 31849 |
+
"grad_norm": 2.0109291076660156,
|
| 31850 |
+
"learning_rate": 5.059538514748753e-06,
|
| 31851 |
+
"loss": 5.4152,
|
| 31852 |
+
"step": 18196
|
| 31853 |
+
},
|
| 31854 |
+
{
|
| 31855 |
+
"epoch": 0.9,
|
| 31856 |
+
"grad_norm": 2.051154851913452,
|
| 31857 |
+
"learning_rate": 5.049656603587134e-06,
|
| 31858 |
+
"loss": 5.504,
|
| 31859 |
+
"step": 18200
|
| 31860 |
+
},
|
| 31861 |
+
{
|
| 31862 |
+
"epoch": 0.9,
|
| 31863 |
+
"grad_norm": 1.9939186573028564,
|
| 31864 |
+
"learning_rate": 5.039774692425515e-06,
|
| 31865 |
+
"loss": 5.377,
|
| 31866 |
+
"step": 18204
|
| 31867 |
+
},
|
| 31868 |
+
{
|
| 31869 |
+
"epoch": 0.9,
|
| 31870 |
+
"grad_norm": 2.0900635719299316,
|
| 31871 |
+
"learning_rate": 5.029892781263896e-06,
|
| 31872 |
+
"loss": 5.4448,
|
| 31873 |
+
"step": 18208
|
| 31874 |
+
},
|
| 31875 |
+
{
|
| 31876 |
+
"epoch": 0.9,
|
| 31877 |
+
"grad_norm": 1.7526922225952148,
|
| 31878 |
+
"learning_rate": 5.020010870102278e-06,
|
| 31879 |
+
"loss": 5.2948,
|
| 31880 |
+
"step": 18212
|
| 31881 |
+
},
|
| 31882 |
+
{
|
| 31883 |
+
"epoch": 0.9,
|
| 31884 |
+
"grad_norm": 2.082535982131958,
|
| 31885 |
+
"learning_rate": 5.01012895894066e-06,
|
| 31886 |
+
"loss": 5.4976,
|
| 31887 |
+
"step": 18216
|
| 31888 |
}
|
| 31889 |
],
|
| 31890 |
"logging_steps": 4,
|
|
|
|
| 31892 |
"num_input_tokens_seen": 0,
|
| 31893 |
"num_train_epochs": 1,
|
| 31894 |
"save_steps": 2024,
|
| 31895 |
+
"total_flos": 7.67303003626537e+16,
|
| 31896 |
"train_batch_size": 8,
|
| 31897 |
"trial_name": null,
|
| 31898 |
"trial_params": null
|