Training in progress, step 1422, checkpoint
Browse files
last-checkpoint/adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 195125192
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2ede535db8a810ce9fb2aa5a7eb1b672ece64d74f6757610be67ecc0c91415aa
|
| 3 |
size 195125192
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 99831781
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:570fd83bc0bae2a7b50e9b70be935dcee8b3db3cf98f901c96efc7ab465d04e2
|
| 3 |
size 99831781
|
last-checkpoint/rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14645
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3d6d8fafcd1ee268414be5acf0366296af5b03d60871978712eac1979cb42d65
|
| 3 |
size 14645
|
last-checkpoint/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1465
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6f21374c25f9f9af3295300d053f8f4a6e967c8bc8059c5c4e45da73a44f75b2
|
| 3 |
size 1465
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -2,9 +2,9 @@
|
|
| 2 |
"best_global_step": null,
|
| 3 |
"best_metric": null,
|
| 4 |
"best_model_checkpoint": null,
|
| 5 |
-
"epoch": 0.
|
| 6 |
"eval_steps": 474,
|
| 7 |
-
"global_step":
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
@@ -6660,6 +6660,3332 @@
|
|
| 6660 |
"eval_samples_per_second": 22.111,
|
| 6661 |
"eval_steps_per_second": 5.532,
|
| 6662 |
"step": 948
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 6663 |
}
|
| 6664 |
],
|
| 6665 |
"logging_steps": 1,
|
|
@@ -6679,7 +10005,7 @@
|
|
| 6679 |
"attributes": {}
|
| 6680 |
}
|
| 6681 |
},
|
| 6682 |
-
"total_flos":
|
| 6683 |
"train_batch_size": 4,
|
| 6684 |
"trial_name": null,
|
| 6685 |
"trial_params": null
|
|
|
|
| 2 |
"best_global_step": null,
|
| 3 |
"best_metric": null,
|
| 4 |
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 0.6914240564030876,
|
| 6 |
"eval_steps": 474,
|
| 7 |
+
"global_step": 1422,
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
|
|
| 6660 |
"eval_samples_per_second": 22.111,
|
| 6661 |
"eval_steps_per_second": 5.532,
|
| 6662 |
"step": 948
|
| 6663 |
+
},
|
| 6664 |
+
{
|
| 6665 |
+
"epoch": 0.46143560444903664,
|
| 6666 |
+
"grad_norm": 0.01333213597536087,
|
| 6667 |
+
"learning_rate": 0.00030705933304460805,
|
| 6668 |
+
"loss": 0.037060752511024475,
|
| 6669 |
+
"step": 949
|
| 6670 |
+
},
|
| 6671 |
+
{
|
| 6672 |
+
"epoch": 0.4619218379626816,
|
| 6673 |
+
"grad_norm": 0.014666610397398472,
|
| 6674 |
+
"learning_rate": 0.0003068427890861845,
|
| 6675 |
+
"loss": 0.039563264697790146,
|
| 6676 |
+
"step": 950
|
| 6677 |
+
},
|
| 6678 |
+
{
|
| 6679 |
+
"epoch": 0.4624080714763265,
|
| 6680 |
+
"grad_norm": 0.014413428492844105,
|
| 6681 |
+
"learning_rate": 0.0003066262451277609,
|
| 6682 |
+
"loss": 0.03490311652421951,
|
| 6683 |
+
"step": 951
|
| 6684 |
+
},
|
| 6685 |
+
{
|
| 6686 |
+
"epoch": 0.4628943049899714,
|
| 6687 |
+
"grad_norm": 0.014327336102724075,
|
| 6688 |
+
"learning_rate": 0.0003064097011693374,
|
| 6689 |
+
"loss": 0.03744923323392868,
|
| 6690 |
+
"step": 952
|
| 6691 |
+
},
|
| 6692 |
+
{
|
| 6693 |
+
"epoch": 0.4633805385036164,
|
| 6694 |
+
"grad_norm": 0.013454711064696312,
|
| 6695 |
+
"learning_rate": 0.00030619315721091384,
|
| 6696 |
+
"loss": 0.03523600846529007,
|
| 6697 |
+
"step": 953
|
| 6698 |
+
},
|
| 6699 |
+
{
|
| 6700 |
+
"epoch": 0.4638667720172613,
|
| 6701 |
+
"grad_norm": 0.013872881419956684,
|
| 6702 |
+
"learning_rate": 0.00030597661325249025,
|
| 6703 |
+
"loss": 0.036383256316185,
|
| 6704 |
+
"step": 954
|
| 6705 |
+
},
|
| 6706 |
+
{
|
| 6707 |
+
"epoch": 0.4643530055309062,
|
| 6708 |
+
"grad_norm": 0.013156497851014137,
|
| 6709 |
+
"learning_rate": 0.0003057600692940667,
|
| 6710 |
+
"loss": 0.03556698188185692,
|
| 6711 |
+
"step": 955
|
| 6712 |
+
},
|
| 6713 |
+
{
|
| 6714 |
+
"epoch": 0.46483923904455116,
|
| 6715 |
+
"grad_norm": 0.01402540784329176,
|
| 6716 |
+
"learning_rate": 0.0003055435253356432,
|
| 6717 |
+
"loss": 0.0337364487349987,
|
| 6718 |
+
"step": 956
|
| 6719 |
+
},
|
| 6720 |
+
{
|
| 6721 |
+
"epoch": 0.46532547255819606,
|
| 6722 |
+
"grad_norm": 0.014185097999870777,
|
| 6723 |
+
"learning_rate": 0.0003053269813772196,
|
| 6724 |
+
"loss": 0.034214798361063004,
|
| 6725 |
+
"step": 957
|
| 6726 |
+
},
|
| 6727 |
+
{
|
| 6728 |
+
"epoch": 0.465811706071841,
|
| 6729 |
+
"grad_norm": 0.01380988396704197,
|
| 6730 |
+
"learning_rate": 0.000305110437418796,
|
| 6731 |
+
"loss": 0.03797744959592819,
|
| 6732 |
+
"step": 958
|
| 6733 |
+
},
|
| 6734 |
+
{
|
| 6735 |
+
"epoch": 0.46629793958548593,
|
| 6736 |
+
"grad_norm": 0.015561358071863651,
|
| 6737 |
+
"learning_rate": 0.00030489389346037246,
|
| 6738 |
+
"loss": 0.03637092933058739,
|
| 6739 |
+
"step": 959
|
| 6740 |
+
},
|
| 6741 |
+
{
|
| 6742 |
+
"epoch": 0.46678417309913084,
|
| 6743 |
+
"grad_norm": 0.01449553668498993,
|
| 6744 |
+
"learning_rate": 0.0003046773495019489,
|
| 6745 |
+
"loss": 0.03891671076416969,
|
| 6746 |
+
"step": 960
|
| 6747 |
+
},
|
| 6748 |
+
{
|
| 6749 |
+
"epoch": 0.4672704066127758,
|
| 6750 |
+
"grad_norm": 0.01277115661650896,
|
| 6751 |
+
"learning_rate": 0.0003044608055435253,
|
| 6752 |
+
"loss": 0.03399049863219261,
|
| 6753 |
+
"step": 961
|
| 6754 |
+
},
|
| 6755 |
+
{
|
| 6756 |
+
"epoch": 0.4677566401264207,
|
| 6757 |
+
"grad_norm": 0.013732589781284332,
|
| 6758 |
+
"learning_rate": 0.0003042442615851018,
|
| 6759 |
+
"loss": 0.03488261625170708,
|
| 6760 |
+
"step": 962
|
| 6761 |
+
},
|
| 6762 |
+
{
|
| 6763 |
+
"epoch": 0.4682428736400656,
|
| 6764 |
+
"grad_norm": 0.013535597361624241,
|
| 6765 |
+
"learning_rate": 0.00030402771762667825,
|
| 6766 |
+
"loss": 0.03977413475513458,
|
| 6767 |
+
"step": 963
|
| 6768 |
+
},
|
| 6769 |
+
{
|
| 6770 |
+
"epoch": 0.4687291071537106,
|
| 6771 |
+
"grad_norm": 0.014227787964046001,
|
| 6772 |
+
"learning_rate": 0.00030381117366825466,
|
| 6773 |
+
"loss": 0.039457306265830994,
|
| 6774 |
+
"step": 964
|
| 6775 |
+
},
|
| 6776 |
+
{
|
| 6777 |
+
"epoch": 0.4692153406673555,
|
| 6778 |
+
"grad_norm": 0.013504140079021454,
|
| 6779 |
+
"learning_rate": 0.0003035946297098311,
|
| 6780 |
+
"loss": 0.03363119810819626,
|
| 6781 |
+
"step": 965
|
| 6782 |
+
},
|
| 6783 |
+
{
|
| 6784 |
+
"epoch": 0.46970157418100045,
|
| 6785 |
+
"grad_norm": 0.013607289642095566,
|
| 6786 |
+
"learning_rate": 0.00030337808575140753,
|
| 6787 |
+
"loss": 0.037966132164001465,
|
| 6788 |
+
"step": 966
|
| 6789 |
+
},
|
| 6790 |
+
{
|
| 6791 |
+
"epoch": 0.47018780769464535,
|
| 6792 |
+
"grad_norm": 0.013094695284962654,
|
| 6793 |
+
"learning_rate": 0.000303161541792984,
|
| 6794 |
+
"loss": 0.03316368907690048,
|
| 6795 |
+
"step": 967
|
| 6796 |
+
},
|
| 6797 |
+
{
|
| 6798 |
+
"epoch": 0.47067404120829026,
|
| 6799 |
+
"grad_norm": 0.013641368597745895,
|
| 6800 |
+
"learning_rate": 0.0003029449978345604,
|
| 6801 |
+
"loss": 0.03873325139284134,
|
| 6802 |
+
"step": 968
|
| 6803 |
+
},
|
| 6804 |
+
{
|
| 6805 |
+
"epoch": 0.4711602747219352,
|
| 6806 |
+
"grad_norm": 0.01359693892300129,
|
| 6807 |
+
"learning_rate": 0.00030272845387613686,
|
| 6808 |
+
"loss": 0.039252910763025284,
|
| 6809 |
+
"step": 969
|
| 6810 |
+
},
|
| 6811 |
+
{
|
| 6812 |
+
"epoch": 0.47164650823558013,
|
| 6813 |
+
"grad_norm": 0.0134733272716403,
|
| 6814 |
+
"learning_rate": 0.0003025119099177133,
|
| 6815 |
+
"loss": 0.03847449645400047,
|
| 6816 |
+
"step": 970
|
| 6817 |
+
},
|
| 6818 |
+
{
|
| 6819 |
+
"epoch": 0.47213274174922504,
|
| 6820 |
+
"grad_norm": 0.013345838524401188,
|
| 6821 |
+
"learning_rate": 0.00030229536595928973,
|
| 6822 |
+
"loss": 0.039488643407821655,
|
| 6823 |
+
"step": 971
|
| 6824 |
+
},
|
| 6825 |
+
{
|
| 6826 |
+
"epoch": 0.47261897526287,
|
| 6827 |
+
"grad_norm": 0.01357495877891779,
|
| 6828 |
+
"learning_rate": 0.0003020788220008662,
|
| 6829 |
+
"loss": 0.037975043058395386,
|
| 6830 |
+
"step": 972
|
| 6831 |
+
},
|
| 6832 |
+
{
|
| 6833 |
+
"epoch": 0.4731052087765149,
|
| 6834 |
+
"grad_norm": 0.01503952220082283,
|
| 6835 |
+
"learning_rate": 0.0003018622780424426,
|
| 6836 |
+
"loss": 0.03439665213227272,
|
| 6837 |
+
"step": 973
|
| 6838 |
+
},
|
| 6839 |
+
{
|
| 6840 |
+
"epoch": 0.47359144229015987,
|
| 6841 |
+
"grad_norm": 0.014459135942161083,
|
| 6842 |
+
"learning_rate": 0.00030164573408401907,
|
| 6843 |
+
"loss": 0.036652304232120514,
|
| 6844 |
+
"step": 974
|
| 6845 |
+
},
|
| 6846 |
+
{
|
| 6847 |
+
"epoch": 0.4740776758038048,
|
| 6848 |
+
"grad_norm": 0.014494093134999275,
|
| 6849 |
+
"learning_rate": 0.0003014291901255955,
|
| 6850 |
+
"loss": 0.03853532671928406,
|
| 6851 |
+
"step": 975
|
| 6852 |
+
},
|
| 6853 |
+
{
|
| 6854 |
+
"epoch": 0.4745639093174497,
|
| 6855 |
+
"grad_norm": 0.014746518805623055,
|
| 6856 |
+
"learning_rate": 0.00030121264616717194,
|
| 6857 |
+
"loss": 0.034450870007276535,
|
| 6858 |
+
"step": 976
|
| 6859 |
+
},
|
| 6860 |
+
{
|
| 6861 |
+
"epoch": 0.47505014283109465,
|
| 6862 |
+
"grad_norm": 0.014133110642433167,
|
| 6863 |
+
"learning_rate": 0.0003009961022087484,
|
| 6864 |
+
"loss": 0.04049959033727646,
|
| 6865 |
+
"step": 977
|
| 6866 |
+
},
|
| 6867 |
+
{
|
| 6868 |
+
"epoch": 0.47553637634473955,
|
| 6869 |
+
"grad_norm": 0.014087921939790249,
|
| 6870 |
+
"learning_rate": 0.0003007795582503248,
|
| 6871 |
+
"loss": 0.0359988659620285,
|
| 6872 |
+
"step": 978
|
| 6873 |
+
},
|
| 6874 |
+
{
|
| 6875 |
+
"epoch": 0.4760226098583845,
|
| 6876 |
+
"grad_norm": 0.013674001209437847,
|
| 6877 |
+
"learning_rate": 0.00030056301429190127,
|
| 6878 |
+
"loss": 0.0319087952375412,
|
| 6879 |
+
"step": 979
|
| 6880 |
+
},
|
| 6881 |
+
{
|
| 6882 |
+
"epoch": 0.4765088433720294,
|
| 6883 |
+
"grad_norm": 0.013482606038451195,
|
| 6884 |
+
"learning_rate": 0.00030034647033347773,
|
| 6885 |
+
"loss": 0.03861702233552933,
|
| 6886 |
+
"step": 980
|
| 6887 |
+
},
|
| 6888 |
+
{
|
| 6889 |
+
"epoch": 0.47699507688567433,
|
| 6890 |
+
"grad_norm": 0.013383050449192524,
|
| 6891 |
+
"learning_rate": 0.00030012992637505414,
|
| 6892 |
+
"loss": 0.03758491203188896,
|
| 6893 |
+
"step": 981
|
| 6894 |
+
},
|
| 6895 |
+
{
|
| 6896 |
+
"epoch": 0.4774813103993193,
|
| 6897 |
+
"grad_norm": 0.014612327329814434,
|
| 6898 |
+
"learning_rate": 0.0002999133824166306,
|
| 6899 |
+
"loss": 0.03936506062746048,
|
| 6900 |
+
"step": 982
|
| 6901 |
+
},
|
| 6902 |
+
{
|
| 6903 |
+
"epoch": 0.4779675439129642,
|
| 6904 |
+
"grad_norm": 0.013832930475473404,
|
| 6905 |
+
"learning_rate": 0.000299696838458207,
|
| 6906 |
+
"loss": 0.03548000752925873,
|
| 6907 |
+
"step": 983
|
| 6908 |
+
},
|
| 6909 |
+
{
|
| 6910 |
+
"epoch": 0.4784537774266091,
|
| 6911 |
+
"grad_norm": 0.013834511861205101,
|
| 6912 |
+
"learning_rate": 0.00029948029449978347,
|
| 6913 |
+
"loss": 0.033962223678827286,
|
| 6914 |
+
"step": 984
|
| 6915 |
+
},
|
| 6916 |
+
{
|
| 6917 |
+
"epoch": 0.47894001094025407,
|
| 6918 |
+
"grad_norm": 0.012873651459813118,
|
| 6919 |
+
"learning_rate": 0.0002992637505413599,
|
| 6920 |
+
"loss": 0.03505862504243851,
|
| 6921 |
+
"step": 985
|
| 6922 |
+
},
|
| 6923 |
+
{
|
| 6924 |
+
"epoch": 0.479426244453899,
|
| 6925 |
+
"grad_norm": 0.013968316838145256,
|
| 6926 |
+
"learning_rate": 0.00029904720658293634,
|
| 6927 |
+
"loss": 0.03907528147101402,
|
| 6928 |
+
"step": 986
|
| 6929 |
+
},
|
| 6930 |
+
{
|
| 6931 |
+
"epoch": 0.47991247796754394,
|
| 6932 |
+
"grad_norm": 0.013454949483275414,
|
| 6933 |
+
"learning_rate": 0.0002988306626245128,
|
| 6934 |
+
"loss": 0.03765419125556946,
|
| 6935 |
+
"step": 987
|
| 6936 |
+
},
|
| 6937 |
+
{
|
| 6938 |
+
"epoch": 0.48039871148118884,
|
| 6939 |
+
"grad_norm": 0.014067943207919598,
|
| 6940 |
+
"learning_rate": 0.0002986141186660892,
|
| 6941 |
+
"loss": 0.03700315207242966,
|
| 6942 |
+
"step": 988
|
| 6943 |
+
},
|
| 6944 |
+
{
|
| 6945 |
+
"epoch": 0.48088494499483375,
|
| 6946 |
+
"grad_norm": 0.013319096527993679,
|
| 6947 |
+
"learning_rate": 0.0002983975747076657,
|
| 6948 |
+
"loss": 0.03290332108736038,
|
| 6949 |
+
"step": 989
|
| 6950 |
+
},
|
| 6951 |
+
{
|
| 6952 |
+
"epoch": 0.4813711785084787,
|
| 6953 |
+
"grad_norm": 0.01260996051132679,
|
| 6954 |
+
"learning_rate": 0.0002981810307492421,
|
| 6955 |
+
"loss": 0.03568773716688156,
|
| 6956 |
+
"step": 990
|
| 6957 |
+
},
|
| 6958 |
+
{
|
| 6959 |
+
"epoch": 0.4818574120221236,
|
| 6960 |
+
"grad_norm": 0.013793015852570534,
|
| 6961 |
+
"learning_rate": 0.00029796448679081855,
|
| 6962 |
+
"loss": 0.035640228539705276,
|
| 6963 |
+
"step": 991
|
| 6964 |
+
},
|
| 6965 |
+
{
|
| 6966 |
+
"epoch": 0.48234364553576853,
|
| 6967 |
+
"grad_norm": 0.01395094208419323,
|
| 6968 |
+
"learning_rate": 0.00029774794283239495,
|
| 6969 |
+
"loss": 0.037107307463884354,
|
| 6970 |
+
"step": 992
|
| 6971 |
+
},
|
| 6972 |
+
{
|
| 6973 |
+
"epoch": 0.4828298790494135,
|
| 6974 |
+
"grad_norm": 0.014228583313524723,
|
| 6975 |
+
"learning_rate": 0.0002975313988739714,
|
| 6976 |
+
"loss": 0.03581860661506653,
|
| 6977 |
+
"step": 993
|
| 6978 |
+
},
|
| 6979 |
+
{
|
| 6980 |
+
"epoch": 0.4833161125630584,
|
| 6981 |
+
"grad_norm": 0.013966365717351437,
|
| 6982 |
+
"learning_rate": 0.0002973148549155479,
|
| 6983 |
+
"loss": 0.0373714454472065,
|
| 6984 |
+
"step": 994
|
| 6985 |
+
},
|
| 6986 |
+
{
|
| 6987 |
+
"epoch": 0.48380234607670336,
|
| 6988 |
+
"grad_norm": 0.014881118200719357,
|
| 6989 |
+
"learning_rate": 0.0002970983109571243,
|
| 6990 |
+
"loss": 0.039440978318452835,
|
| 6991 |
+
"step": 995
|
| 6992 |
+
},
|
| 6993 |
+
{
|
| 6994 |
+
"epoch": 0.48428857959034827,
|
| 6995 |
+
"grad_norm": 0.01362524926662445,
|
| 6996 |
+
"learning_rate": 0.00029688176699870075,
|
| 6997 |
+
"loss": 0.03773489594459534,
|
| 6998 |
+
"step": 996
|
| 6999 |
+
},
|
| 7000 |
+
{
|
| 7001 |
+
"epoch": 0.4847748131039932,
|
| 7002 |
+
"grad_norm": 0.014135688543319702,
|
| 7003 |
+
"learning_rate": 0.0002966652230402772,
|
| 7004 |
+
"loss": 0.03612879291176796,
|
| 7005 |
+
"step": 997
|
| 7006 |
+
},
|
| 7007 |
+
{
|
| 7008 |
+
"epoch": 0.48526104661763814,
|
| 7009 |
+
"grad_norm": 0.01322716660797596,
|
| 7010 |
+
"learning_rate": 0.0002964486790818536,
|
| 7011 |
+
"loss": 0.038729891180992126,
|
| 7012 |
+
"step": 998
|
| 7013 |
+
},
|
| 7014 |
+
{
|
| 7015 |
+
"epoch": 0.48574728013128304,
|
| 7016 |
+
"grad_norm": 0.014626462943851948,
|
| 7017 |
+
"learning_rate": 0.0002962321351234301,
|
| 7018 |
+
"loss": 0.03782185539603233,
|
| 7019 |
+
"step": 999
|
| 7020 |
+
},
|
| 7021 |
+
{
|
| 7022 |
+
"epoch": 0.48623351364492795,
|
| 7023 |
+
"grad_norm": 0.014553732238709927,
|
| 7024 |
+
"learning_rate": 0.0002960155911650065,
|
| 7025 |
+
"loss": 0.03561050817370415,
|
| 7026 |
+
"step": 1000
|
| 7027 |
+
},
|
| 7028 |
+
{
|
| 7029 |
+
"epoch": 0.4867197471585729,
|
| 7030 |
+
"grad_norm": 0.014309097081422806,
|
| 7031 |
+
"learning_rate": 0.00029579904720658295,
|
| 7032 |
+
"loss": 0.03861964866518974,
|
| 7033 |
+
"step": 1001
|
| 7034 |
+
},
|
| 7035 |
+
{
|
| 7036 |
+
"epoch": 0.4872059806722178,
|
| 7037 |
+
"grad_norm": 0.012629122473299503,
|
| 7038 |
+
"learning_rate": 0.00029558250324815936,
|
| 7039 |
+
"loss": 0.03389890491962433,
|
| 7040 |
+
"step": 1002
|
| 7041 |
+
},
|
| 7042 |
+
{
|
| 7043 |
+
"epoch": 0.4876922141858628,
|
| 7044 |
+
"grad_norm": 0.014117905870079994,
|
| 7045 |
+
"learning_rate": 0.0002953659592897358,
|
| 7046 |
+
"loss": 0.03747197985649109,
|
| 7047 |
+
"step": 1003
|
| 7048 |
+
},
|
| 7049 |
+
{
|
| 7050 |
+
"epoch": 0.4881784476995077,
|
| 7051 |
+
"grad_norm": 0.013692461885511875,
|
| 7052 |
+
"learning_rate": 0.0002951494153313123,
|
| 7053 |
+
"loss": 0.03489881008863449,
|
| 7054 |
+
"step": 1004
|
| 7055 |
+
},
|
| 7056 |
+
{
|
| 7057 |
+
"epoch": 0.4886646812131526,
|
| 7058 |
+
"grad_norm": 0.01396162249147892,
|
| 7059 |
+
"learning_rate": 0.0002949328713728887,
|
| 7060 |
+
"loss": 0.035344578325748444,
|
| 7061 |
+
"step": 1005
|
| 7062 |
+
},
|
| 7063 |
+
{
|
| 7064 |
+
"epoch": 0.48915091472679756,
|
| 7065 |
+
"grad_norm": 0.013573478907346725,
|
| 7066 |
+
"learning_rate": 0.00029471632741446516,
|
| 7067 |
+
"loss": 0.03927745297551155,
|
| 7068 |
+
"step": 1006
|
| 7069 |
+
},
|
| 7070 |
+
{
|
| 7071 |
+
"epoch": 0.48963714824044247,
|
| 7072 |
+
"grad_norm": 0.013765616342425346,
|
| 7073 |
+
"learning_rate": 0.00029449978345604157,
|
| 7074 |
+
"loss": 0.033977653831243515,
|
| 7075 |
+
"step": 1007
|
| 7076 |
+
},
|
| 7077 |
+
{
|
| 7078 |
+
"epoch": 0.4901233817540874,
|
| 7079 |
+
"grad_norm": 0.013294212520122528,
|
| 7080 |
+
"learning_rate": 0.00029428323949761803,
|
| 7081 |
+
"loss": 0.034197743982076645,
|
| 7082 |
+
"step": 1008
|
| 7083 |
+
},
|
| 7084 |
+
{
|
| 7085 |
+
"epoch": 0.49060961526773234,
|
| 7086 |
+
"grad_norm": 0.013792756013572216,
|
| 7087 |
+
"learning_rate": 0.00029406669553919444,
|
| 7088 |
+
"loss": 0.0338798388838768,
|
| 7089 |
+
"step": 1009
|
| 7090 |
+
},
|
| 7091 |
+
{
|
| 7092 |
+
"epoch": 0.49109584878137724,
|
| 7093 |
+
"grad_norm": 0.014132252894341946,
|
| 7094 |
+
"learning_rate": 0.0002938501515807709,
|
| 7095 |
+
"loss": 0.03649807721376419,
|
| 7096 |
+
"step": 1010
|
| 7097 |
+
},
|
| 7098 |
+
{
|
| 7099 |
+
"epoch": 0.4915820822950222,
|
| 7100 |
+
"grad_norm": 0.013721713796257973,
|
| 7101 |
+
"learning_rate": 0.00029363360762234736,
|
| 7102 |
+
"loss": 0.0340237021446228,
|
| 7103 |
+
"step": 1011
|
| 7104 |
+
},
|
| 7105 |
+
{
|
| 7106 |
+
"epoch": 0.4920683158086671,
|
| 7107 |
+
"grad_norm": 0.015485446900129318,
|
| 7108 |
+
"learning_rate": 0.00029341706366392377,
|
| 7109 |
+
"loss": 0.03557199239730835,
|
| 7110 |
+
"step": 1012
|
| 7111 |
+
},
|
| 7112 |
+
{
|
| 7113 |
+
"epoch": 0.492554549322312,
|
| 7114 |
+
"grad_norm": 0.013667070306837559,
|
| 7115 |
+
"learning_rate": 0.00029320051970550023,
|
| 7116 |
+
"loss": 0.03545474633574486,
|
| 7117 |
+
"step": 1013
|
| 7118 |
+
},
|
| 7119 |
+
{
|
| 7120 |
+
"epoch": 0.493040782835957,
|
| 7121 |
+
"grad_norm": 0.015222440473735332,
|
| 7122 |
+
"learning_rate": 0.00029298397574707664,
|
| 7123 |
+
"loss": 0.03371709585189819,
|
| 7124 |
+
"step": 1014
|
| 7125 |
+
},
|
| 7126 |
+
{
|
| 7127 |
+
"epoch": 0.4935270163496019,
|
| 7128 |
+
"grad_norm": 0.012989448383450508,
|
| 7129 |
+
"learning_rate": 0.0002927674317886531,
|
| 7130 |
+
"loss": 0.03332573175430298,
|
| 7131 |
+
"step": 1015
|
| 7132 |
+
},
|
| 7133 |
+
{
|
| 7134 |
+
"epoch": 0.49401324986324685,
|
| 7135 |
+
"grad_norm": 0.014377813786268234,
|
| 7136 |
+
"learning_rate": 0.00029255088783022956,
|
| 7137 |
+
"loss": 0.03789186850190163,
|
| 7138 |
+
"step": 1016
|
| 7139 |
+
},
|
| 7140 |
+
{
|
| 7141 |
+
"epoch": 0.49449948337689176,
|
| 7142 |
+
"grad_norm": 0.01331800315529108,
|
| 7143 |
+
"learning_rate": 0.00029233434387180597,
|
| 7144 |
+
"loss": 0.035854436457157135,
|
| 7145 |
+
"step": 1017
|
| 7146 |
+
},
|
| 7147 |
+
{
|
| 7148 |
+
"epoch": 0.49498571689053666,
|
| 7149 |
+
"grad_norm": 0.01412629522383213,
|
| 7150 |
+
"learning_rate": 0.00029211779991338243,
|
| 7151 |
+
"loss": 0.0383642315864563,
|
| 7152 |
+
"step": 1018
|
| 7153 |
+
},
|
| 7154 |
+
{
|
| 7155 |
+
"epoch": 0.4954719504041816,
|
| 7156 |
+
"grad_norm": 0.013764998875558376,
|
| 7157 |
+
"learning_rate": 0.00029190125595495884,
|
| 7158 |
+
"loss": 0.039679210633039474,
|
| 7159 |
+
"step": 1019
|
| 7160 |
+
},
|
| 7161 |
+
{
|
| 7162 |
+
"epoch": 0.49595818391782653,
|
| 7163 |
+
"grad_norm": 0.014185120351612568,
|
| 7164 |
+
"learning_rate": 0.0002916847119965353,
|
| 7165 |
+
"loss": 0.036174945533275604,
|
| 7166 |
+
"step": 1020
|
| 7167 |
+
},
|
| 7168 |
+
{
|
| 7169 |
+
"epoch": 0.49644441743147144,
|
| 7170 |
+
"grad_norm": 0.014051356352865696,
|
| 7171 |
+
"learning_rate": 0.00029146816803811177,
|
| 7172 |
+
"loss": 0.036626048386096954,
|
| 7173 |
+
"step": 1021
|
| 7174 |
+
},
|
| 7175 |
+
{
|
| 7176 |
+
"epoch": 0.4969306509451164,
|
| 7177 |
+
"grad_norm": 0.013894816860556602,
|
| 7178 |
+
"learning_rate": 0.0002912516240796882,
|
| 7179 |
+
"loss": 0.03644995018839836,
|
| 7180 |
+
"step": 1022
|
| 7181 |
+
},
|
| 7182 |
+
{
|
| 7183 |
+
"epoch": 0.4974168844587613,
|
| 7184 |
+
"grad_norm": 0.013736812397837639,
|
| 7185 |
+
"learning_rate": 0.00029103508012126464,
|
| 7186 |
+
"loss": 0.03344986215233803,
|
| 7187 |
+
"step": 1023
|
| 7188 |
+
},
|
| 7189 |
+
{
|
| 7190 |
+
"epoch": 0.4979031179724063,
|
| 7191 |
+
"grad_norm": 0.013827759772539139,
|
| 7192 |
+
"learning_rate": 0.00029081853616284105,
|
| 7193 |
+
"loss": 0.03501196205615997,
|
| 7194 |
+
"step": 1024
|
| 7195 |
+
},
|
| 7196 |
+
{
|
| 7197 |
+
"epoch": 0.4983893514860512,
|
| 7198 |
+
"grad_norm": 0.01374312024563551,
|
| 7199 |
+
"learning_rate": 0.0002906019922044175,
|
| 7200 |
+
"loss": 0.03503570705652237,
|
| 7201 |
+
"step": 1025
|
| 7202 |
+
},
|
| 7203 |
+
{
|
| 7204 |
+
"epoch": 0.4988755849996961,
|
| 7205 |
+
"grad_norm": 0.013449519872665405,
|
| 7206 |
+
"learning_rate": 0.0002903854482459939,
|
| 7207 |
+
"loss": 0.033825989812612534,
|
| 7208 |
+
"step": 1026
|
| 7209 |
+
},
|
| 7210 |
+
{
|
| 7211 |
+
"epoch": 0.49936181851334105,
|
| 7212 |
+
"grad_norm": 0.013138077221810818,
|
| 7213 |
+
"learning_rate": 0.0002901689042875704,
|
| 7214 |
+
"loss": 0.0354132317006588,
|
| 7215 |
+
"step": 1027
|
| 7216 |
+
},
|
| 7217 |
+
{
|
| 7218 |
+
"epoch": 0.49984805202698596,
|
| 7219 |
+
"grad_norm": 0.013364890590310097,
|
| 7220 |
+
"learning_rate": 0.00028995236032914684,
|
| 7221 |
+
"loss": 0.03656283766031265,
|
| 7222 |
+
"step": 1028
|
| 7223 |
+
},
|
| 7224 |
+
{
|
| 7225 |
+
"epoch": 0.5003342855406309,
|
| 7226 |
+
"grad_norm": 0.013499929569661617,
|
| 7227 |
+
"learning_rate": 0.00028973581637072325,
|
| 7228 |
+
"loss": 0.038244593888521194,
|
| 7229 |
+
"step": 1029
|
| 7230 |
+
},
|
| 7231 |
+
{
|
| 7232 |
+
"epoch": 0.5008205190542758,
|
| 7233 |
+
"grad_norm": 0.013432961888611317,
|
| 7234 |
+
"learning_rate": 0.0002895192724122997,
|
| 7235 |
+
"loss": 0.0363604873418808,
|
| 7236 |
+
"step": 1030
|
| 7237 |
+
},
|
| 7238 |
+
{
|
| 7239 |
+
"epoch": 0.5013067525679208,
|
| 7240 |
+
"grad_norm": 0.014093254692852497,
|
| 7241 |
+
"learning_rate": 0.0002893027284538761,
|
| 7242 |
+
"loss": 0.0386497899889946,
|
| 7243 |
+
"step": 1031
|
| 7244 |
+
},
|
| 7245 |
+
{
|
| 7246 |
+
"epoch": 0.5017929860815656,
|
| 7247 |
+
"grad_norm": 0.013268879614770412,
|
| 7248 |
+
"learning_rate": 0.0002890861844954526,
|
| 7249 |
+
"loss": 0.033929694443941116,
|
| 7250 |
+
"step": 1032
|
| 7251 |
+
},
|
| 7252 |
+
{
|
| 7253 |
+
"epoch": 0.5022792195952106,
|
| 7254 |
+
"grad_norm": 0.01348187681287527,
|
| 7255 |
+
"learning_rate": 0.00028886964053702905,
|
| 7256 |
+
"loss": 0.03745168447494507,
|
| 7257 |
+
"step": 1033
|
| 7258 |
+
},
|
| 7259 |
+
{
|
| 7260 |
+
"epoch": 0.5027654531088556,
|
| 7261 |
+
"grad_norm": 0.013327261433005333,
|
| 7262 |
+
"learning_rate": 0.00028865309657860545,
|
| 7263 |
+
"loss": 0.03401700779795647,
|
| 7264 |
+
"step": 1034
|
| 7265 |
+
},
|
| 7266 |
+
{
|
| 7267 |
+
"epoch": 0.5032516866225004,
|
| 7268 |
+
"grad_norm": 0.012692646123468876,
|
| 7269 |
+
"learning_rate": 0.0002884365526201819,
|
| 7270 |
+
"loss": 0.03298821672797203,
|
| 7271 |
+
"step": 1035
|
| 7272 |
+
},
|
| 7273 |
+
{
|
| 7274 |
+
"epoch": 0.5037379201361454,
|
| 7275 |
+
"grad_norm": 0.0130177466198802,
|
| 7276 |
+
"learning_rate": 0.0002882200086617583,
|
| 7277 |
+
"loss": 0.03459744155406952,
|
| 7278 |
+
"step": 1036
|
| 7279 |
+
},
|
| 7280 |
+
{
|
| 7281 |
+
"epoch": 0.5042241536497903,
|
| 7282 |
+
"grad_norm": 0.01413513533771038,
|
| 7283 |
+
"learning_rate": 0.0002880034647033348,
|
| 7284 |
+
"loss": 0.03635701164603233,
|
| 7285 |
+
"step": 1037
|
| 7286 |
+
},
|
| 7287 |
+
{
|
| 7288 |
+
"epoch": 0.5047103871634352,
|
| 7289 |
+
"grad_norm": 0.01338459923863411,
|
| 7290 |
+
"learning_rate": 0.0002877869207449112,
|
| 7291 |
+
"loss": 0.03807484358549118,
|
| 7292 |
+
"step": 1038
|
| 7293 |
+
},
|
| 7294 |
+
{
|
| 7295 |
+
"epoch": 0.5051966206770802,
|
| 7296 |
+
"grad_norm": 0.013206271454691887,
|
| 7297 |
+
"learning_rate": 0.00028757037678648766,
|
| 7298 |
+
"loss": 0.03469417616724968,
|
| 7299 |
+
"step": 1039
|
| 7300 |
+
},
|
| 7301 |
+
{
|
| 7302 |
+
"epoch": 0.5056828541907251,
|
| 7303 |
+
"grad_norm": 0.013839577324688435,
|
| 7304 |
+
"learning_rate": 0.0002873538328280641,
|
| 7305 |
+
"loss": 0.03586322069168091,
|
| 7306 |
+
"step": 1040
|
| 7307 |
+
},
|
| 7308 |
+
{
|
| 7309 |
+
"epoch": 0.50616908770437,
|
| 7310 |
+
"grad_norm": 0.014376025646924973,
|
| 7311 |
+
"learning_rate": 0.00028713728886964053,
|
| 7312 |
+
"loss": 0.03774120286107063,
|
| 7313 |
+
"step": 1041
|
| 7314 |
+
},
|
| 7315 |
+
{
|
| 7316 |
+
"epoch": 0.5066553212180149,
|
| 7317 |
+
"grad_norm": 0.013656928203999996,
|
| 7318 |
+
"learning_rate": 0.000286920744911217,
|
| 7319 |
+
"loss": 0.03591923788189888,
|
| 7320 |
+
"step": 1042
|
| 7321 |
+
},
|
| 7322 |
+
{
|
| 7323 |
+
"epoch": 0.5071415547316599,
|
| 7324 |
+
"grad_norm": 0.013080078177154064,
|
| 7325 |
+
"learning_rate": 0.00028670420095279345,
|
| 7326 |
+
"loss": 0.03551328927278519,
|
| 7327 |
+
"step": 1043
|
| 7328 |
+
},
|
| 7329 |
+
{
|
| 7330 |
+
"epoch": 0.5076277882453049,
|
| 7331 |
+
"grad_norm": 0.0133149903267622,
|
| 7332 |
+
"learning_rate": 0.00028648765699436986,
|
| 7333 |
+
"loss": 0.03520733863115311,
|
| 7334 |
+
"step": 1044
|
| 7335 |
+
},
|
| 7336 |
+
{
|
| 7337 |
+
"epoch": 0.5081140217589497,
|
| 7338 |
+
"grad_norm": 0.013456271030008793,
|
| 7339 |
+
"learning_rate": 0.0002862711130359463,
|
| 7340 |
+
"loss": 0.034662459045648575,
|
| 7341 |
+
"step": 1045
|
| 7342 |
+
},
|
| 7343 |
+
{
|
| 7344 |
+
"epoch": 0.5086002552725947,
|
| 7345 |
+
"grad_norm": 0.0138524966314435,
|
| 7346 |
+
"learning_rate": 0.00028605456907752273,
|
| 7347 |
+
"loss": 0.03426341712474823,
|
| 7348 |
+
"step": 1046
|
| 7349 |
+
},
|
| 7350 |
+
{
|
| 7351 |
+
"epoch": 0.5090864887862396,
|
| 7352 |
+
"grad_norm": 0.013234579935669899,
|
| 7353 |
+
"learning_rate": 0.0002858380251190992,
|
| 7354 |
+
"loss": 0.036201220005750656,
|
| 7355 |
+
"step": 1047
|
| 7356 |
+
},
|
| 7357 |
+
{
|
| 7358 |
+
"epoch": 0.5095727222998845,
|
| 7359 |
+
"grad_norm": 0.01478514727205038,
|
| 7360 |
+
"learning_rate": 0.0002856214811606756,
|
| 7361 |
+
"loss": 0.03706587478518486,
|
| 7362 |
+
"step": 1048
|
| 7363 |
+
},
|
| 7364 |
+
{
|
| 7365 |
+
"epoch": 0.5100589558135294,
|
| 7366 |
+
"grad_norm": 0.014010661281645298,
|
| 7367 |
+
"learning_rate": 0.00028540493720225206,
|
| 7368 |
+
"loss": 0.033495090901851654,
|
| 7369 |
+
"step": 1049
|
| 7370 |
+
},
|
| 7371 |
+
{
|
| 7372 |
+
"epoch": 0.5105451893271744,
|
| 7373 |
+
"grad_norm": 0.014482982456684113,
|
| 7374 |
+
"learning_rate": 0.0002851883932438285,
|
| 7375 |
+
"loss": 0.03460167720913887,
|
| 7376 |
+
"step": 1050
|
| 7377 |
+
},
|
| 7378 |
+
{
|
| 7379 |
+
"epoch": 0.5110314228408193,
|
| 7380 |
+
"grad_norm": 0.013520138338208199,
|
| 7381 |
+
"learning_rate": 0.00028497184928540493,
|
| 7382 |
+
"loss": 0.034577250480651855,
|
| 7383 |
+
"step": 1051
|
| 7384 |
+
},
|
| 7385 |
+
{
|
| 7386 |
+
"epoch": 0.5115176563544642,
|
| 7387 |
+
"grad_norm": 0.013442403636872768,
|
| 7388 |
+
"learning_rate": 0.0002847553053269814,
|
| 7389 |
+
"loss": 0.03726118803024292,
|
| 7390 |
+
"step": 1052
|
| 7391 |
+
},
|
| 7392 |
+
{
|
| 7393 |
+
"epoch": 0.5120038898681092,
|
| 7394 |
+
"grad_norm": 0.013599686324596405,
|
| 7395 |
+
"learning_rate": 0.0002845387613685578,
|
| 7396 |
+
"loss": 0.034997228533029556,
|
| 7397 |
+
"step": 1053
|
| 7398 |
+
},
|
| 7399 |
+
{
|
| 7400 |
+
"epoch": 0.512490123381754,
|
| 7401 |
+
"grad_norm": 0.014052558690309525,
|
| 7402 |
+
"learning_rate": 0.00028432221741013427,
|
| 7403 |
+
"loss": 0.033227983862161636,
|
| 7404 |
+
"step": 1054
|
| 7405 |
+
},
|
| 7406 |
+
{
|
| 7407 |
+
"epoch": 0.512976356895399,
|
| 7408 |
+
"grad_norm": 0.014964016154408455,
|
| 7409 |
+
"learning_rate": 0.0002841056734517107,
|
| 7410 |
+
"loss": 0.03668046370148659,
|
| 7411 |
+
"step": 1055
|
| 7412 |
+
},
|
| 7413 |
+
{
|
| 7414 |
+
"epoch": 0.513462590409044,
|
| 7415 |
+
"grad_norm": 0.013480198569595814,
|
| 7416 |
+
"learning_rate": 0.00028388912949328714,
|
| 7417 |
+
"loss": 0.03563865274190903,
|
| 7418 |
+
"step": 1056
|
| 7419 |
+
},
|
| 7420 |
+
{
|
| 7421 |
+
"epoch": 0.5139488239226889,
|
| 7422 |
+
"grad_norm": 0.014785866253077984,
|
| 7423 |
+
"learning_rate": 0.0002836725855348636,
|
| 7424 |
+
"loss": 0.03554994612932205,
|
| 7425 |
+
"step": 1057
|
| 7426 |
+
},
|
| 7427 |
+
{
|
| 7428 |
+
"epoch": 0.5144350574363338,
|
| 7429 |
+
"grad_norm": 0.014100163243710995,
|
| 7430 |
+
"learning_rate": 0.00028345604157644,
|
| 7431 |
+
"loss": 0.036969512701034546,
|
| 7432 |
+
"step": 1058
|
| 7433 |
+
},
|
| 7434 |
+
{
|
| 7435 |
+
"epoch": 0.5149212909499787,
|
| 7436 |
+
"grad_norm": 0.013631733134388924,
|
| 7437 |
+
"learning_rate": 0.00028323949761801647,
|
| 7438 |
+
"loss": 0.03874921426177025,
|
| 7439 |
+
"step": 1059
|
| 7440 |
+
},
|
| 7441 |
+
{
|
| 7442 |
+
"epoch": 0.5154075244636237,
|
| 7443 |
+
"grad_norm": 0.01346477773040533,
|
| 7444 |
+
"learning_rate": 0.00028302295365959293,
|
| 7445 |
+
"loss": 0.03435073420405388,
|
| 7446 |
+
"step": 1060
|
| 7447 |
+
},
|
| 7448 |
+
{
|
| 7449 |
+
"epoch": 0.5158937579772686,
|
| 7450 |
+
"grad_norm": 0.013825018890202045,
|
| 7451 |
+
"learning_rate": 0.00028280640970116934,
|
| 7452 |
+
"loss": 0.03786073252558708,
|
| 7453 |
+
"step": 1061
|
| 7454 |
+
},
|
| 7455 |
+
{
|
| 7456 |
+
"epoch": 0.5163799914909135,
|
| 7457 |
+
"grad_norm": 0.013739226385951042,
|
| 7458 |
+
"learning_rate": 0.0002825898657427458,
|
| 7459 |
+
"loss": 0.03493267297744751,
|
| 7460 |
+
"step": 1062
|
| 7461 |
+
},
|
| 7462 |
+
{
|
| 7463 |
+
"epoch": 0.5168662250045585,
|
| 7464 |
+
"grad_norm": 0.014403737150132656,
|
| 7465 |
+
"learning_rate": 0.0002823733217843222,
|
| 7466 |
+
"loss": 0.03776548057794571,
|
| 7467 |
+
"step": 1063
|
| 7468 |
+
},
|
| 7469 |
+
{
|
| 7470 |
+
"epoch": 0.5173524585182033,
|
| 7471 |
+
"grad_norm": 0.013879086822271347,
|
| 7472 |
+
"learning_rate": 0.0002821567778258987,
|
| 7473 |
+
"loss": 0.03671635687351227,
|
| 7474 |
+
"step": 1064
|
| 7475 |
+
},
|
| 7476 |
+
{
|
| 7477 |
+
"epoch": 0.5178386920318483,
|
| 7478 |
+
"grad_norm": 0.014006221666932106,
|
| 7479 |
+
"learning_rate": 0.0002819402338674751,
|
| 7480 |
+
"loss": 0.03714973106980324,
|
| 7481 |
+
"step": 1065
|
| 7482 |
+
},
|
| 7483 |
+
{
|
| 7484 |
+
"epoch": 0.5183249255454933,
|
| 7485 |
+
"grad_norm": 0.013886251486837864,
|
| 7486 |
+
"learning_rate": 0.00028172368990905155,
|
| 7487 |
+
"loss": 0.0382484570145607,
|
| 7488 |
+
"step": 1066
|
| 7489 |
+
},
|
| 7490 |
+
{
|
| 7491 |
+
"epoch": 0.5188111590591381,
|
| 7492 |
+
"grad_norm": 0.021207468584179878,
|
| 7493 |
+
"learning_rate": 0.000281507145950628,
|
| 7494 |
+
"loss": 0.035265255719423294,
|
| 7495 |
+
"step": 1067
|
| 7496 |
+
},
|
| 7497 |
+
{
|
| 7498 |
+
"epoch": 0.5192973925727831,
|
| 7499 |
+
"grad_norm": 0.01562880165874958,
|
| 7500 |
+
"learning_rate": 0.0002812906019922044,
|
| 7501 |
+
"loss": 0.03547072783112526,
|
| 7502 |
+
"step": 1068
|
| 7503 |
+
},
|
| 7504 |
+
{
|
| 7505 |
+
"epoch": 0.519783626086428,
|
| 7506 |
+
"grad_norm": 0.014076780527830124,
|
| 7507 |
+
"learning_rate": 0.0002810740580337809,
|
| 7508 |
+
"loss": 0.03398030251264572,
|
| 7509 |
+
"step": 1069
|
| 7510 |
+
},
|
| 7511 |
+
{
|
| 7512 |
+
"epoch": 0.5202698596000729,
|
| 7513 |
+
"grad_norm": 0.01447069551795721,
|
| 7514 |
+
"learning_rate": 0.0002808575140753573,
|
| 7515 |
+
"loss": 0.03421202301979065,
|
| 7516 |
+
"step": 1070
|
| 7517 |
+
},
|
| 7518 |
+
{
|
| 7519 |
+
"epoch": 0.5207560931137178,
|
| 7520 |
+
"grad_norm": 0.01564360037446022,
|
| 7521 |
+
"learning_rate": 0.00028064097011693375,
|
| 7522 |
+
"loss": 0.03529740869998932,
|
| 7523 |
+
"step": 1071
|
| 7524 |
+
},
|
| 7525 |
+
{
|
| 7526 |
+
"epoch": 0.5212423266273628,
|
| 7527 |
+
"grad_norm": 0.013676514849066734,
|
| 7528 |
+
"learning_rate": 0.00028042442615851016,
|
| 7529 |
+
"loss": 0.03715941309928894,
|
| 7530 |
+
"step": 1072
|
| 7531 |
+
},
|
| 7532 |
+
{
|
| 7533 |
+
"epoch": 0.5217285601410078,
|
| 7534 |
+
"grad_norm": 0.015466811135411263,
|
| 7535 |
+
"learning_rate": 0.0002802078822000866,
|
| 7536 |
+
"loss": 0.037360869348049164,
|
| 7537 |
+
"step": 1073
|
| 7538 |
+
},
|
| 7539 |
+
{
|
| 7540 |
+
"epoch": 0.5222147936546526,
|
| 7541 |
+
"grad_norm": 0.013988148421049118,
|
| 7542 |
+
"learning_rate": 0.0002799913382416631,
|
| 7543 |
+
"loss": 0.0363808311522007,
|
| 7544 |
+
"step": 1074
|
| 7545 |
+
},
|
| 7546 |
+
{
|
| 7547 |
+
"epoch": 0.5227010271682976,
|
| 7548 |
+
"grad_norm": 0.01467287726700306,
|
| 7549 |
+
"learning_rate": 0.0002797747942832395,
|
| 7550 |
+
"loss": 0.03347277641296387,
|
| 7551 |
+
"step": 1075
|
| 7552 |
+
},
|
| 7553 |
+
{
|
| 7554 |
+
"epoch": 0.5231872606819425,
|
| 7555 |
+
"grad_norm": 0.014155564829707146,
|
| 7556 |
+
"learning_rate": 0.00027955825032481595,
|
| 7557 |
+
"loss": 0.033481333404779434,
|
| 7558 |
+
"step": 1076
|
| 7559 |
+
},
|
| 7560 |
+
{
|
| 7561 |
+
"epoch": 0.5236734941955874,
|
| 7562 |
+
"grad_norm": 0.015523151494562626,
|
| 7563 |
+
"learning_rate": 0.0002793417063663924,
|
| 7564 |
+
"loss": 0.038126830011606216,
|
| 7565 |
+
"step": 1077
|
| 7566 |
+
},
|
| 7567 |
+
{
|
| 7568 |
+
"epoch": 0.5241597277092324,
|
| 7569 |
+
"grad_norm": 0.014107078313827515,
|
| 7570 |
+
"learning_rate": 0.0002791251624079688,
|
| 7571 |
+
"loss": 0.0383860319852829,
|
| 7572 |
+
"step": 1078
|
| 7573 |
+
},
|
| 7574 |
+
{
|
| 7575 |
+
"epoch": 0.5246459612228773,
|
| 7576 |
+
"grad_norm": 0.013211668469011784,
|
| 7577 |
+
"learning_rate": 0.00027890861844954523,
|
| 7578 |
+
"loss": 0.037652213126420975,
|
| 7579 |
+
"step": 1079
|
| 7580 |
+
},
|
| 7581 |
+
{
|
| 7582 |
+
"epoch": 0.5251321947365222,
|
| 7583 |
+
"grad_norm": 0.013383075594902039,
|
| 7584 |
+
"learning_rate": 0.0002786920744911217,
|
| 7585 |
+
"loss": 0.03486897051334381,
|
| 7586 |
+
"step": 1080
|
| 7587 |
+
},
|
| 7588 |
+
{
|
| 7589 |
+
"epoch": 0.5256184282501671,
|
| 7590 |
+
"grad_norm": 0.01312405988574028,
|
| 7591 |
+
"learning_rate": 0.00027847553053269816,
|
| 7592 |
+
"loss": 0.03551631048321724,
|
| 7593 |
+
"step": 1081
|
| 7594 |
+
},
|
| 7595 |
+
{
|
| 7596 |
+
"epoch": 0.5261046617638121,
|
| 7597 |
+
"grad_norm": 0.014280078001320362,
|
| 7598 |
+
"learning_rate": 0.00027825898657427456,
|
| 7599 |
+
"loss": 0.0327209047973156,
|
| 7600 |
+
"step": 1082
|
| 7601 |
+
},
|
| 7602 |
+
{
|
| 7603 |
+
"epoch": 0.526590895277457,
|
| 7604 |
+
"grad_norm": 0.014230197295546532,
|
| 7605 |
+
"learning_rate": 0.000278042442615851,
|
| 7606 |
+
"loss": 0.03499278426170349,
|
| 7607 |
+
"step": 1083
|
| 7608 |
+
},
|
| 7609 |
+
{
|
| 7610 |
+
"epoch": 0.5270771287911019,
|
| 7611 |
+
"grad_norm": 0.0142115643247962,
|
| 7612 |
+
"learning_rate": 0.0002778258986574275,
|
| 7613 |
+
"loss": 0.03594350814819336,
|
| 7614 |
+
"step": 1084
|
| 7615 |
+
},
|
| 7616 |
+
{
|
| 7617 |
+
"epoch": 0.5275633623047469,
|
| 7618 |
+
"grad_norm": 0.013110818341374397,
|
| 7619 |
+
"learning_rate": 0.0002776093546990039,
|
| 7620 |
+
"loss": 0.03481375426054001,
|
| 7621 |
+
"step": 1085
|
| 7622 |
+
},
|
| 7623 |
+
{
|
| 7624 |
+
"epoch": 0.5280495958183918,
|
| 7625 |
+
"grad_norm": 0.01580216735601425,
|
| 7626 |
+
"learning_rate": 0.00027739281074058036,
|
| 7627 |
+
"loss": 0.03467171639204025,
|
| 7628 |
+
"step": 1086
|
| 7629 |
+
},
|
| 7630 |
+
{
|
| 7631 |
+
"epoch": 0.5285358293320367,
|
| 7632 |
+
"grad_norm": 0.014446753077208996,
|
| 7633 |
+
"learning_rate": 0.00027717626678215677,
|
| 7634 |
+
"loss": 0.034862712025642395,
|
| 7635 |
+
"step": 1087
|
| 7636 |
+
},
|
| 7637 |
+
{
|
| 7638 |
+
"epoch": 0.5290220628456817,
|
| 7639 |
+
"grad_norm": 0.013800714164972305,
|
| 7640 |
+
"learning_rate": 0.00027695972282373323,
|
| 7641 |
+
"loss": 0.03640086203813553,
|
| 7642 |
+
"step": 1088
|
| 7643 |
+
},
|
| 7644 |
+
{
|
| 7645 |
+
"epoch": 0.5295082963593266,
|
| 7646 |
+
"grad_norm": 0.013880511745810509,
|
| 7647 |
+
"learning_rate": 0.00027674317886530964,
|
| 7648 |
+
"loss": 0.037139032036066055,
|
| 7649 |
+
"step": 1089
|
| 7650 |
+
},
|
| 7651 |
+
{
|
| 7652 |
+
"epoch": 0.5299945298729715,
|
| 7653 |
+
"grad_norm": 0.03105057217180729,
|
| 7654 |
+
"learning_rate": 0.0002765266349068861,
|
| 7655 |
+
"loss": 0.036529600620269775,
|
| 7656 |
+
"step": 1090
|
| 7657 |
+
},
|
| 7658 |
+
{
|
| 7659 |
+
"epoch": 0.5304807633866164,
|
| 7660 |
+
"grad_norm": 0.014402316883206367,
|
| 7661 |
+
"learning_rate": 0.00027631009094846256,
|
| 7662 |
+
"loss": 0.03081696107983589,
|
| 7663 |
+
"step": 1091
|
| 7664 |
+
},
|
| 7665 |
+
{
|
| 7666 |
+
"epoch": 0.5309669969002614,
|
| 7667 |
+
"grad_norm": 0.014533761888742447,
|
| 7668 |
+
"learning_rate": 0.00027609354699003897,
|
| 7669 |
+
"loss": 0.034763358533382416,
|
| 7670 |
+
"step": 1092
|
| 7671 |
+
},
|
| 7672 |
+
{
|
| 7673 |
+
"epoch": 0.5314532304139062,
|
| 7674 |
+
"grad_norm": 0.013627896085381508,
|
| 7675 |
+
"learning_rate": 0.00027587700303161543,
|
| 7676 |
+
"loss": 0.03388526290655136,
|
| 7677 |
+
"step": 1093
|
| 7678 |
+
},
|
| 7679 |
+
{
|
| 7680 |
+
"epoch": 0.5319394639275512,
|
| 7681 |
+
"grad_norm": 0.013394586741924286,
|
| 7682 |
+
"learning_rate": 0.0002756604590731919,
|
| 7683 |
+
"loss": 0.03367890045046806,
|
| 7684 |
+
"step": 1094
|
| 7685 |
+
},
|
| 7686 |
+
{
|
| 7687 |
+
"epoch": 0.5324256974411962,
|
| 7688 |
+
"grad_norm": 0.014889057725667953,
|
| 7689 |
+
"learning_rate": 0.0002754439151147683,
|
| 7690 |
+
"loss": 0.033912476152181625,
|
| 7691 |
+
"step": 1095
|
| 7692 |
+
},
|
| 7693 |
+
{
|
| 7694 |
+
"epoch": 0.532911930954841,
|
| 7695 |
+
"grad_norm": 0.014293133281171322,
|
| 7696 |
+
"learning_rate": 0.0002752273711563447,
|
| 7697 |
+
"loss": 0.03397766500711441,
|
| 7698 |
+
"step": 1096
|
| 7699 |
+
},
|
| 7700 |
+
{
|
| 7701 |
+
"epoch": 0.533398164468486,
|
| 7702 |
+
"grad_norm": 0.013831854797899723,
|
| 7703 |
+
"learning_rate": 0.0002750108271979212,
|
| 7704 |
+
"loss": 0.03504156321287155,
|
| 7705 |
+
"step": 1097
|
| 7706 |
+
},
|
| 7707 |
+
{
|
| 7708 |
+
"epoch": 0.5338843979821309,
|
| 7709 |
+
"grad_norm": 0.014174986630678177,
|
| 7710 |
+
"learning_rate": 0.00027479428323949764,
|
| 7711 |
+
"loss": 0.03770895674824715,
|
| 7712 |
+
"step": 1098
|
| 7713 |
+
},
|
| 7714 |
+
{
|
| 7715 |
+
"epoch": 0.5343706314957758,
|
| 7716 |
+
"grad_norm": 0.01447014044970274,
|
| 7717 |
+
"learning_rate": 0.00027457773928107405,
|
| 7718 |
+
"loss": 0.036136239767074585,
|
| 7719 |
+
"step": 1099
|
| 7720 |
+
},
|
| 7721 |
+
{
|
| 7722 |
+
"epoch": 0.5348568650094208,
|
| 7723 |
+
"grad_norm": 0.013747083023190498,
|
| 7724 |
+
"learning_rate": 0.0002743611953226505,
|
| 7725 |
+
"loss": 0.034260399639606476,
|
| 7726 |
+
"step": 1100
|
| 7727 |
+
},
|
| 7728 |
+
{
|
| 7729 |
+
"epoch": 0.5353430985230657,
|
| 7730 |
+
"grad_norm": 0.013516520150005817,
|
| 7731 |
+
"learning_rate": 0.00027414465136422697,
|
| 7732 |
+
"loss": 0.034295883029699326,
|
| 7733 |
+
"step": 1101
|
| 7734 |
+
},
|
| 7735 |
+
{
|
| 7736 |
+
"epoch": 0.5358293320367107,
|
| 7737 |
+
"grad_norm": 0.013682578690350056,
|
| 7738 |
+
"learning_rate": 0.0002739281074058034,
|
| 7739 |
+
"loss": 0.03732244670391083,
|
| 7740 |
+
"step": 1102
|
| 7741 |
+
},
|
| 7742 |
+
{
|
| 7743 |
+
"epoch": 0.5363155655503555,
|
| 7744 |
+
"grad_norm": 0.014351158402860165,
|
| 7745 |
+
"learning_rate": 0.0002737115634473798,
|
| 7746 |
+
"loss": 0.034668367356061935,
|
| 7747 |
+
"step": 1103
|
| 7748 |
+
},
|
| 7749 |
+
{
|
| 7750 |
+
"epoch": 0.5368017990640005,
|
| 7751 |
+
"grad_norm": 0.013111933134496212,
|
| 7752 |
+
"learning_rate": 0.0002734950194889563,
|
| 7753 |
+
"loss": 0.035882238298654556,
|
| 7754 |
+
"step": 1104
|
| 7755 |
+
},
|
| 7756 |
+
{
|
| 7757 |
+
"epoch": 0.5372880325776455,
|
| 7758 |
+
"grad_norm": 0.013734894804656506,
|
| 7759 |
+
"learning_rate": 0.0002732784755305327,
|
| 7760 |
+
"loss": 0.03722135350108147,
|
| 7761 |
+
"step": 1105
|
| 7762 |
+
},
|
| 7763 |
+
{
|
| 7764 |
+
"epoch": 0.5377742660912903,
|
| 7765 |
+
"grad_norm": 0.013723678886890411,
|
| 7766 |
+
"learning_rate": 0.0002730619315721091,
|
| 7767 |
+
"loss": 0.03608832508325577,
|
| 7768 |
+
"step": 1106
|
| 7769 |
+
},
|
| 7770 |
+
{
|
| 7771 |
+
"epoch": 0.5382604996049353,
|
| 7772 |
+
"grad_norm": 0.014407295733690262,
|
| 7773 |
+
"learning_rate": 0.0002728453876136856,
|
| 7774 |
+
"loss": 0.036822110414505005,
|
| 7775 |
+
"step": 1107
|
| 7776 |
+
},
|
| 7777 |
+
{
|
| 7778 |
+
"epoch": 0.5387467331185802,
|
| 7779 |
+
"grad_norm": 0.014280757866799831,
|
| 7780 |
+
"learning_rate": 0.00027262884365526204,
|
| 7781 |
+
"loss": 0.03326963633298874,
|
| 7782 |
+
"step": 1108
|
| 7783 |
+
},
|
| 7784 |
+
{
|
| 7785 |
+
"epoch": 0.5392329666322251,
|
| 7786 |
+
"grad_norm": 0.01311193872243166,
|
| 7787 |
+
"learning_rate": 0.00027241229969683845,
|
| 7788 |
+
"loss": 0.031371016055345535,
|
| 7789 |
+
"step": 1109
|
| 7790 |
+
},
|
| 7791 |
+
{
|
| 7792 |
+
"epoch": 0.53971920014587,
|
| 7793 |
+
"grad_norm": 0.013727921061217785,
|
| 7794 |
+
"learning_rate": 0.0002721957557384149,
|
| 7795 |
+
"loss": 0.03663400560617447,
|
| 7796 |
+
"step": 1110
|
| 7797 |
+
},
|
| 7798 |
+
{
|
| 7799 |
+
"epoch": 0.540205433659515,
|
| 7800 |
+
"grad_norm": 0.01490043569356203,
|
| 7801 |
+
"learning_rate": 0.0002719792117799914,
|
| 7802 |
+
"loss": 0.03613126277923584,
|
| 7803 |
+
"step": 1111
|
| 7804 |
+
},
|
| 7805 |
+
{
|
| 7806 |
+
"epoch": 0.5406916671731599,
|
| 7807 |
+
"grad_norm": 0.014557641930878162,
|
| 7808 |
+
"learning_rate": 0.0002717626678215678,
|
| 7809 |
+
"loss": 0.03574630990624428,
|
| 7810 |
+
"step": 1112
|
| 7811 |
+
},
|
| 7812 |
+
{
|
| 7813 |
+
"epoch": 0.5411779006868048,
|
| 7814 |
+
"grad_norm": 0.01450525876134634,
|
| 7815 |
+
"learning_rate": 0.0002715461238631442,
|
| 7816 |
+
"loss": 0.036301519721746445,
|
| 7817 |
+
"step": 1113
|
| 7818 |
+
},
|
| 7819 |
+
{
|
| 7820 |
+
"epoch": 0.5416641342004498,
|
| 7821 |
+
"grad_norm": 0.01451620738953352,
|
| 7822 |
+
"learning_rate": 0.00027132957990472066,
|
| 7823 |
+
"loss": 0.03540552407503128,
|
| 7824 |
+
"step": 1114
|
| 7825 |
+
},
|
| 7826 |
+
{
|
| 7827 |
+
"epoch": 0.5421503677140946,
|
| 7828 |
+
"grad_norm": 0.013807685114443302,
|
| 7829 |
+
"learning_rate": 0.0002711130359462971,
|
| 7830 |
+
"loss": 0.039537493139505386,
|
| 7831 |
+
"step": 1115
|
| 7832 |
+
},
|
| 7833 |
+
{
|
| 7834 |
+
"epoch": 0.5426366012277396,
|
| 7835 |
+
"grad_norm": 0.01461933646351099,
|
| 7836 |
+
"learning_rate": 0.0002708964919878735,
|
| 7837 |
+
"loss": 0.034039728343486786,
|
| 7838 |
+
"step": 1116
|
| 7839 |
+
},
|
| 7840 |
+
{
|
| 7841 |
+
"epoch": 0.5431228347413846,
|
| 7842 |
+
"grad_norm": 0.013832468539476395,
|
| 7843 |
+
"learning_rate": 0.00027067994802945,
|
| 7844 |
+
"loss": 0.03415941447019577,
|
| 7845 |
+
"step": 1117
|
| 7846 |
+
},
|
| 7847 |
+
{
|
| 7848 |
+
"epoch": 0.5436090682550295,
|
| 7849 |
+
"grad_norm": 0.013654129579663277,
|
| 7850 |
+
"learning_rate": 0.00027046340407102645,
|
| 7851 |
+
"loss": 0.0373041145503521,
|
| 7852 |
+
"step": 1118
|
| 7853 |
+
},
|
| 7854 |
+
{
|
| 7855 |
+
"epoch": 0.5440953017686744,
|
| 7856 |
+
"grad_norm": 0.01301703229546547,
|
| 7857 |
+
"learning_rate": 0.00027024686011260286,
|
| 7858 |
+
"loss": 0.036253511905670166,
|
| 7859 |
+
"step": 1119
|
| 7860 |
+
},
|
| 7861 |
+
{
|
| 7862 |
+
"epoch": 0.5445815352823193,
|
| 7863 |
+
"grad_norm": 0.013760874979197979,
|
| 7864 |
+
"learning_rate": 0.00027003031615417927,
|
| 7865 |
+
"loss": 0.032219234853982925,
|
| 7866 |
+
"step": 1120
|
| 7867 |
+
},
|
| 7868 |
+
{
|
| 7869 |
+
"epoch": 0.5450677687959643,
|
| 7870 |
+
"grad_norm": 0.015250683762133121,
|
| 7871 |
+
"learning_rate": 0.0002698137721957558,
|
| 7872 |
+
"loss": 0.0377640575170517,
|
| 7873 |
+
"step": 1121
|
| 7874 |
+
},
|
| 7875 |
+
{
|
| 7876 |
+
"epoch": 0.5455540023096092,
|
| 7877 |
+
"grad_norm": 0.015328595414757729,
|
| 7878 |
+
"learning_rate": 0.0002695972282373322,
|
| 7879 |
+
"loss": 0.0361119844019413,
|
| 7880 |
+
"step": 1122
|
| 7881 |
+
},
|
| 7882 |
+
{
|
| 7883 |
+
"epoch": 0.5460402358232541,
|
| 7884 |
+
"grad_norm": 0.03255344182252884,
|
| 7885 |
+
"learning_rate": 0.0002693806842789086,
|
| 7886 |
+
"loss": 0.03720467910170555,
|
| 7887 |
+
"step": 1123
|
| 7888 |
+
},
|
| 7889 |
+
{
|
| 7890 |
+
"epoch": 0.5465264693368991,
|
| 7891 |
+
"grad_norm": 0.013812856748700142,
|
| 7892 |
+
"learning_rate": 0.00026916414032048506,
|
| 7893 |
+
"loss": 0.03531163930892944,
|
| 7894 |
+
"step": 1124
|
| 7895 |
+
},
|
| 7896 |
+
{
|
| 7897 |
+
"epoch": 0.5470127028505439,
|
| 7898 |
+
"grad_norm": 0.013879380188882351,
|
| 7899 |
+
"learning_rate": 0.0002689475963620615,
|
| 7900 |
+
"loss": 0.03696723282337189,
|
| 7901 |
+
"step": 1125
|
| 7902 |
+
},
|
| 7903 |
+
{
|
| 7904 |
+
"epoch": 0.5474989363641889,
|
| 7905 |
+
"grad_norm": 0.02719070576131344,
|
| 7906 |
+
"learning_rate": 0.00026873105240363793,
|
| 7907 |
+
"loss": 0.03985331580042839,
|
| 7908 |
+
"step": 1126
|
| 7909 |
+
},
|
| 7910 |
+
{
|
| 7911 |
+
"epoch": 0.5479851698778339,
|
| 7912 |
+
"grad_norm": 0.014068419113755226,
|
| 7913 |
+
"learning_rate": 0.0002685145084452144,
|
| 7914 |
+
"loss": 0.03203361853957176,
|
| 7915 |
+
"step": 1127
|
| 7916 |
+
},
|
| 7917 |
+
{
|
| 7918 |
+
"epoch": 0.5484714033914787,
|
| 7919 |
+
"grad_norm": 0.014416170306503773,
|
| 7920 |
+
"learning_rate": 0.00026829796448679086,
|
| 7921 |
+
"loss": 0.035285789519548416,
|
| 7922 |
+
"step": 1128
|
| 7923 |
+
},
|
| 7924 |
+
{
|
| 7925 |
+
"epoch": 0.5489576369051237,
|
| 7926 |
+
"grad_norm": 0.014336498454213142,
|
| 7927 |
+
"learning_rate": 0.00026808142052836727,
|
| 7928 |
+
"loss": 0.03428525850176811,
|
| 7929 |
+
"step": 1129
|
| 7930 |
+
},
|
| 7931 |
+
{
|
| 7932 |
+
"epoch": 0.5494438704187686,
|
| 7933 |
+
"grad_norm": 0.05851925164461136,
|
| 7934 |
+
"learning_rate": 0.0002678648765699437,
|
| 7935 |
+
"loss": 0.03579060360789299,
|
| 7936 |
+
"step": 1130
|
| 7937 |
+
},
|
| 7938 |
+
{
|
| 7939 |
+
"epoch": 0.5499301039324136,
|
| 7940 |
+
"grad_norm": 0.07212135195732117,
|
| 7941 |
+
"learning_rate": 0.00026764833261152014,
|
| 7942 |
+
"loss": 0.037139203399419785,
|
| 7943 |
+
"step": 1131
|
| 7944 |
+
},
|
| 7945 |
+
{
|
| 7946 |
+
"epoch": 0.5504163374460584,
|
| 7947 |
+
"grad_norm": 0.025306610390543938,
|
| 7948 |
+
"learning_rate": 0.0002674317886530966,
|
| 7949 |
+
"loss": 0.03558531403541565,
|
| 7950 |
+
"step": 1132
|
| 7951 |
+
},
|
| 7952 |
+
{
|
| 7953 |
+
"epoch": 0.5509025709597034,
|
| 7954 |
+
"grad_norm": 0.016330786049365997,
|
| 7955 |
+
"learning_rate": 0.000267215244694673,
|
| 7956 |
+
"loss": 0.03798871859908104,
|
| 7957 |
+
"step": 1133
|
| 7958 |
+
},
|
| 7959 |
+
{
|
| 7960 |
+
"epoch": 0.5513888044733484,
|
| 7961 |
+
"grad_norm": 0.02563410997390747,
|
| 7962 |
+
"learning_rate": 0.00026699870073624947,
|
| 7963 |
+
"loss": 0.035177167505025864,
|
| 7964 |
+
"step": 1134
|
| 7965 |
+
},
|
| 7966 |
+
{
|
| 7967 |
+
"epoch": 0.5518750379869932,
|
| 7968 |
+
"grad_norm": 0.034003619104623795,
|
| 7969 |
+
"learning_rate": 0.00026678215677782593,
|
| 7970 |
+
"loss": 0.03756334260106087,
|
| 7971 |
+
"step": 1135
|
| 7972 |
+
},
|
| 7973 |
+
{
|
| 7974 |
+
"epoch": 0.5523612715006382,
|
| 7975 |
+
"grad_norm": 0.10675495862960815,
|
| 7976 |
+
"learning_rate": 0.00026656561281940234,
|
| 7977 |
+
"loss": 0.037493519484996796,
|
| 7978 |
+
"step": 1136
|
| 7979 |
+
},
|
| 7980 |
+
{
|
| 7981 |
+
"epoch": 0.5528475050142831,
|
| 7982 |
+
"grad_norm": 0.015743186697363853,
|
| 7983 |
+
"learning_rate": 0.00026634906886097875,
|
| 7984 |
+
"loss": 0.03329642489552498,
|
| 7985 |
+
"step": 1137
|
| 7986 |
+
},
|
| 7987 |
+
{
|
| 7988 |
+
"epoch": 0.553333738527928,
|
| 7989 |
+
"grad_norm": 0.025118406862020493,
|
| 7990 |
+
"learning_rate": 0.00026613252490255526,
|
| 7991 |
+
"loss": 0.03610823675990105,
|
| 7992 |
+
"step": 1138
|
| 7993 |
+
},
|
| 7994 |
+
{
|
| 7995 |
+
"epoch": 0.553819972041573,
|
| 7996 |
+
"grad_norm": 0.03117513842880726,
|
| 7997 |
+
"learning_rate": 0.0002659159809441317,
|
| 7998 |
+
"loss": 0.03637993335723877,
|
| 7999 |
+
"step": 1139
|
| 8000 |
+
},
|
| 8001 |
+
{
|
| 8002 |
+
"epoch": 0.5543062055552179,
|
| 8003 |
+
"grad_norm": 0.02941029518842697,
|
| 8004 |
+
"learning_rate": 0.0002656994369857081,
|
| 8005 |
+
"loss": 0.035395268350839615,
|
| 8006 |
+
"step": 1140
|
| 8007 |
+
},
|
| 8008 |
+
{
|
| 8009 |
+
"epoch": 0.5547924390688628,
|
| 8010 |
+
"grad_norm": 0.01681404374539852,
|
| 8011 |
+
"learning_rate": 0.00026548289302728454,
|
| 8012 |
+
"loss": 0.03673194721341133,
|
| 8013 |
+
"step": 1141
|
| 8014 |
+
},
|
| 8015 |
+
{
|
| 8016 |
+
"epoch": 0.5552786725825077,
|
| 8017 |
+
"grad_norm": 0.01851377636194229,
|
| 8018 |
+
"learning_rate": 0.000265266349068861,
|
| 8019 |
+
"loss": 0.03434037044644356,
|
| 8020 |
+
"step": 1142
|
| 8021 |
+
},
|
| 8022 |
+
{
|
| 8023 |
+
"epoch": 0.5557649060961527,
|
| 8024 |
+
"grad_norm": 0.016019780188798904,
|
| 8025 |
+
"learning_rate": 0.0002650498051104374,
|
| 8026 |
+
"loss": 0.03344109654426575,
|
| 8027 |
+
"step": 1143
|
| 8028 |
+
},
|
| 8029 |
+
{
|
| 8030 |
+
"epoch": 0.5562511396097976,
|
| 8031 |
+
"grad_norm": 0.015543187968432903,
|
| 8032 |
+
"learning_rate": 0.0002648332611520138,
|
| 8033 |
+
"loss": 0.03443222865462303,
|
| 8034 |
+
"step": 1144
|
| 8035 |
+
},
|
| 8036 |
+
{
|
| 8037 |
+
"epoch": 0.5567373731234425,
|
| 8038 |
+
"grad_norm": 0.016184071078896523,
|
| 8039 |
+
"learning_rate": 0.00026461671719359034,
|
| 8040 |
+
"loss": 0.03171855956315994,
|
| 8041 |
+
"step": 1145
|
| 8042 |
+
},
|
| 8043 |
+
{
|
| 8044 |
+
"epoch": 0.5572236066370875,
|
| 8045 |
+
"grad_norm": 0.015442226082086563,
|
| 8046 |
+
"learning_rate": 0.00026440017323516675,
|
| 8047 |
+
"loss": 0.03837261721491814,
|
| 8048 |
+
"step": 1146
|
| 8049 |
+
},
|
| 8050 |
+
{
|
| 8051 |
+
"epoch": 0.5577098401507324,
|
| 8052 |
+
"grad_norm": 0.014598467387259007,
|
| 8053 |
+
"learning_rate": 0.00026418362927674316,
|
| 8054 |
+
"loss": 0.03551751747727394,
|
| 8055 |
+
"step": 1147
|
| 8056 |
+
},
|
| 8057 |
+
{
|
| 8058 |
+
"epoch": 0.5581960736643773,
|
| 8059 |
+
"grad_norm": 0.01574649289250374,
|
| 8060 |
+
"learning_rate": 0.0002639670853183196,
|
| 8061 |
+
"loss": 0.03703949600458145,
|
| 8062 |
+
"step": 1148
|
| 8063 |
+
},
|
| 8064 |
+
{
|
| 8065 |
+
"epoch": 0.5586823071780223,
|
| 8066 |
+
"grad_norm": 0.01476812269538641,
|
| 8067 |
+
"learning_rate": 0.0002637505413598961,
|
| 8068 |
+
"loss": 0.036938779056072235,
|
| 8069 |
+
"step": 1149
|
| 8070 |
+
},
|
| 8071 |
+
{
|
| 8072 |
+
"epoch": 0.5591685406916672,
|
| 8073 |
+
"grad_norm": 0.015625180676579475,
|
| 8074 |
+
"learning_rate": 0.0002635339974014725,
|
| 8075 |
+
"loss": 0.03427194431424141,
|
| 8076 |
+
"step": 1150
|
| 8077 |
+
},
|
| 8078 |
+
{
|
| 8079 |
+
"epoch": 0.5596547742053121,
|
| 8080 |
+
"grad_norm": 0.015583393163979053,
|
| 8081 |
+
"learning_rate": 0.00026331745344304895,
|
| 8082 |
+
"loss": 0.03499074652791023,
|
| 8083 |
+
"step": 1151
|
| 8084 |
+
},
|
| 8085 |
+
{
|
| 8086 |
+
"epoch": 0.560141007718957,
|
| 8087 |
+
"grad_norm": 0.015638794749975204,
|
| 8088 |
+
"learning_rate": 0.0002631009094846254,
|
| 8089 |
+
"loss": 0.03275633603334427,
|
| 8090 |
+
"step": 1152
|
| 8091 |
+
},
|
| 8092 |
+
{
|
| 8093 |
+
"epoch": 0.560627241232602,
|
| 8094 |
+
"grad_norm": 0.013819956220686436,
|
| 8095 |
+
"learning_rate": 0.0002628843655262018,
|
| 8096 |
+
"loss": 0.03441233187913895,
|
| 8097 |
+
"step": 1153
|
| 8098 |
+
},
|
| 8099 |
+
{
|
| 8100 |
+
"epoch": 0.5611134747462468,
|
| 8101 |
+
"grad_norm": 0.015144738368690014,
|
| 8102 |
+
"learning_rate": 0.00026266782156777823,
|
| 8103 |
+
"loss": 0.03544919192790985,
|
| 8104 |
+
"step": 1154
|
| 8105 |
+
},
|
| 8106 |
+
{
|
| 8107 |
+
"epoch": 0.5615997082598918,
|
| 8108 |
+
"grad_norm": 0.014519906602799892,
|
| 8109 |
+
"learning_rate": 0.00026245127760935475,
|
| 8110 |
+
"loss": 0.03368879482150078,
|
| 8111 |
+
"step": 1155
|
| 8112 |
+
},
|
| 8113 |
+
{
|
| 8114 |
+
"epoch": 0.5620859417735368,
|
| 8115 |
+
"grad_norm": 0.015124383382499218,
|
| 8116 |
+
"learning_rate": 0.00026223473365093115,
|
| 8117 |
+
"loss": 0.03450941666960716,
|
| 8118 |
+
"step": 1156
|
| 8119 |
+
},
|
| 8120 |
+
{
|
| 8121 |
+
"epoch": 0.5625721752871816,
|
| 8122 |
+
"grad_norm": 0.014417370781302452,
|
| 8123 |
+
"learning_rate": 0.00026201818969250756,
|
| 8124 |
+
"loss": 0.03814588487148285,
|
| 8125 |
+
"step": 1157
|
| 8126 |
+
},
|
| 8127 |
+
{
|
| 8128 |
+
"epoch": 0.5630584088008266,
|
| 8129 |
+
"grad_norm": 0.016265632584691048,
|
| 8130 |
+
"learning_rate": 0.000261801645734084,
|
| 8131 |
+
"loss": 0.03617558628320694,
|
| 8132 |
+
"step": 1158
|
| 8133 |
+
},
|
| 8134 |
+
{
|
| 8135 |
+
"epoch": 0.5635446423144715,
|
| 8136 |
+
"grad_norm": 0.014344267547130585,
|
| 8137 |
+
"learning_rate": 0.0002615851017756605,
|
| 8138 |
+
"loss": 0.03676125034689903,
|
| 8139 |
+
"step": 1159
|
| 8140 |
+
},
|
| 8141 |
+
{
|
| 8142 |
+
"epoch": 0.5640308758281165,
|
| 8143 |
+
"grad_norm": 0.016097700223326683,
|
| 8144 |
+
"learning_rate": 0.0002613685578172369,
|
| 8145 |
+
"loss": 0.03329729288816452,
|
| 8146 |
+
"step": 1160
|
| 8147 |
+
},
|
| 8148 |
+
{
|
| 8149 |
+
"epoch": 0.5645171093417614,
|
| 8150 |
+
"grad_norm": 0.015015493147075176,
|
| 8151 |
+
"learning_rate": 0.0002611520138588133,
|
| 8152 |
+
"loss": 0.036959245800971985,
|
| 8153 |
+
"step": 1161
|
| 8154 |
+
},
|
| 8155 |
+
{
|
| 8156 |
+
"epoch": 0.5650033428554063,
|
| 8157 |
+
"grad_norm": 0.015185183845460415,
|
| 8158 |
+
"learning_rate": 0.0002609354699003898,
|
| 8159 |
+
"loss": 0.03682759404182434,
|
| 8160 |
+
"step": 1162
|
| 8161 |
+
},
|
| 8162 |
+
{
|
| 8163 |
+
"epoch": 0.5654895763690513,
|
| 8164 |
+
"grad_norm": 0.014621437527239323,
|
| 8165 |
+
"learning_rate": 0.00026071892594196623,
|
| 8166 |
+
"loss": 0.035395123064517975,
|
| 8167 |
+
"step": 1163
|
| 8168 |
+
},
|
| 8169 |
+
{
|
| 8170 |
+
"epoch": 0.5659758098826961,
|
| 8171 |
+
"grad_norm": 0.015326340682804585,
|
| 8172 |
+
"learning_rate": 0.00026050238198354264,
|
| 8173 |
+
"loss": 0.03279733657836914,
|
| 8174 |
+
"step": 1164
|
| 8175 |
+
},
|
| 8176 |
+
{
|
| 8177 |
+
"epoch": 0.5664620433963411,
|
| 8178 |
+
"grad_norm": 0.014678249135613441,
|
| 8179 |
+
"learning_rate": 0.0002602858380251191,
|
| 8180 |
+
"loss": 0.03424302488565445,
|
| 8181 |
+
"step": 1165
|
| 8182 |
+
},
|
| 8183 |
+
{
|
| 8184 |
+
"epoch": 0.5669482769099861,
|
| 8185 |
+
"grad_norm": 0.014298009686172009,
|
| 8186 |
+
"learning_rate": 0.00026006929406669556,
|
| 8187 |
+
"loss": 0.035760778933763504,
|
| 8188 |
+
"step": 1166
|
| 8189 |
+
},
|
| 8190 |
+
{
|
| 8191 |
+
"epoch": 0.5674345104236309,
|
| 8192 |
+
"grad_norm": 0.014259879477322102,
|
| 8193 |
+
"learning_rate": 0.00025985275010827197,
|
| 8194 |
+
"loss": 0.03151150047779083,
|
| 8195 |
+
"step": 1167
|
| 8196 |
+
},
|
| 8197 |
+
{
|
| 8198 |
+
"epoch": 0.5679207439372759,
|
| 8199 |
+
"grad_norm": 0.014827264472842216,
|
| 8200 |
+
"learning_rate": 0.0002596362061498484,
|
| 8201 |
+
"loss": 0.03375959396362305,
|
| 8202 |
+
"step": 1168
|
| 8203 |
+
},
|
| 8204 |
+
{
|
| 8205 |
+
"epoch": 0.5684069774509208,
|
| 8206 |
+
"grad_norm": 0.013862174935638905,
|
| 8207 |
+
"learning_rate": 0.0002594196621914249,
|
| 8208 |
+
"loss": 0.034797750413417816,
|
| 8209 |
+
"step": 1169
|
| 8210 |
+
},
|
| 8211 |
+
{
|
| 8212 |
+
"epoch": 0.5688932109645657,
|
| 8213 |
+
"grad_norm": 0.013984517194330692,
|
| 8214 |
+
"learning_rate": 0.0002592031182330013,
|
| 8215 |
+
"loss": 0.03391830623149872,
|
| 8216 |
+
"step": 1170
|
| 8217 |
+
},
|
| 8218 |
+
{
|
| 8219 |
+
"epoch": 0.5693794444782107,
|
| 8220 |
+
"grad_norm": 0.015051658265292645,
|
| 8221 |
+
"learning_rate": 0.0002589865742745777,
|
| 8222 |
+
"loss": 0.03376283496618271,
|
| 8223 |
+
"step": 1171
|
| 8224 |
+
},
|
| 8225 |
+
{
|
| 8226 |
+
"epoch": 0.5698656779918556,
|
| 8227 |
+
"grad_norm": 0.015616999007761478,
|
| 8228 |
+
"learning_rate": 0.00025877003031615423,
|
| 8229 |
+
"loss": 0.03637155145406723,
|
| 8230 |
+
"step": 1172
|
| 8231 |
+
},
|
| 8232 |
+
{
|
| 8233 |
+
"epoch": 0.5703519115055005,
|
| 8234 |
+
"grad_norm": 0.015244786627590656,
|
| 8235 |
+
"learning_rate": 0.00025855348635773064,
|
| 8236 |
+
"loss": 0.03279532864689827,
|
| 8237 |
+
"step": 1173
|
| 8238 |
+
},
|
| 8239 |
+
{
|
| 8240 |
+
"epoch": 0.5708381450191454,
|
| 8241 |
+
"grad_norm": 0.015110467560589314,
|
| 8242 |
+
"learning_rate": 0.00025833694239930704,
|
| 8243 |
+
"loss": 0.028855204582214355,
|
| 8244 |
+
"step": 1174
|
| 8245 |
+
},
|
| 8246 |
+
{
|
| 8247 |
+
"epoch": 0.5713243785327904,
|
| 8248 |
+
"grad_norm": 0.014544975012540817,
|
| 8249 |
+
"learning_rate": 0.0002581203984408835,
|
| 8250 |
+
"loss": 0.0331791490316391,
|
| 8251 |
+
"step": 1175
|
| 8252 |
+
},
|
| 8253 |
+
{
|
| 8254 |
+
"epoch": 0.5718106120464354,
|
| 8255 |
+
"grad_norm": 0.016088686883449554,
|
| 8256 |
+
"learning_rate": 0.00025790385448245997,
|
| 8257 |
+
"loss": 0.035294558852910995,
|
| 8258 |
+
"step": 1176
|
| 8259 |
+
},
|
| 8260 |
+
{
|
| 8261 |
+
"epoch": 0.5722968455600802,
|
| 8262 |
+
"grad_norm": 0.015287506394088268,
|
| 8263 |
+
"learning_rate": 0.0002576873105240364,
|
| 8264 |
+
"loss": 0.0316205769777298,
|
| 8265 |
+
"step": 1177
|
| 8266 |
+
},
|
| 8267 |
+
{
|
| 8268 |
+
"epoch": 0.5727830790737252,
|
| 8269 |
+
"grad_norm": 0.014967241324484348,
|
| 8270 |
+
"learning_rate": 0.0002574707665656128,
|
| 8271 |
+
"loss": 0.035786937922239304,
|
| 8272 |
+
"step": 1178
|
| 8273 |
+
},
|
| 8274 |
+
{
|
| 8275 |
+
"epoch": 0.5732693125873701,
|
| 8276 |
+
"grad_norm": 0.013769371435046196,
|
| 8277 |
+
"learning_rate": 0.0002572542226071893,
|
| 8278 |
+
"loss": 0.034682467579841614,
|
| 8279 |
+
"step": 1179
|
| 8280 |
+
},
|
| 8281 |
+
{
|
| 8282 |
+
"epoch": 0.573755546101015,
|
| 8283 |
+
"grad_norm": 0.013882031664252281,
|
| 8284 |
+
"learning_rate": 0.0002570376786487657,
|
| 8285 |
+
"loss": 0.031522758305072784,
|
| 8286 |
+
"step": 1180
|
| 8287 |
+
},
|
| 8288 |
+
{
|
| 8289 |
+
"epoch": 0.5742417796146599,
|
| 8290 |
+
"grad_norm": 0.014171222224831581,
|
| 8291 |
+
"learning_rate": 0.0002568211346903421,
|
| 8292 |
+
"loss": 0.0348481647670269,
|
| 8293 |
+
"step": 1181
|
| 8294 |
+
},
|
| 8295 |
+
{
|
| 8296 |
+
"epoch": 0.5747280131283049,
|
| 8297 |
+
"grad_norm": 0.014809303916990757,
|
| 8298 |
+
"learning_rate": 0.00025660459073191863,
|
| 8299 |
+
"loss": 0.0344531424343586,
|
| 8300 |
+
"step": 1182
|
| 8301 |
+
},
|
| 8302 |
+
{
|
| 8303 |
+
"epoch": 0.5752142466419498,
|
| 8304 |
+
"grad_norm": 0.014414481818675995,
|
| 8305 |
+
"learning_rate": 0.00025638804677349504,
|
| 8306 |
+
"loss": 0.037036582827568054,
|
| 8307 |
+
"step": 1183
|
| 8308 |
+
},
|
| 8309 |
+
{
|
| 8310 |
+
"epoch": 0.5757004801555947,
|
| 8311 |
+
"grad_norm": 0.016130995005369186,
|
| 8312 |
+
"learning_rate": 0.00025617150281507145,
|
| 8313 |
+
"loss": 0.03491906449198723,
|
| 8314 |
+
"step": 1184
|
| 8315 |
+
},
|
| 8316 |
+
{
|
| 8317 |
+
"epoch": 0.5761867136692397,
|
| 8318 |
+
"grad_norm": 0.014523262158036232,
|
| 8319 |
+
"learning_rate": 0.00025595495885664786,
|
| 8320 |
+
"loss": 0.033943112939596176,
|
| 8321 |
+
"step": 1185
|
| 8322 |
+
},
|
| 8323 |
+
{
|
| 8324 |
+
"epoch": 0.5766729471828845,
|
| 8325 |
+
"grad_norm": 0.014005020260810852,
|
| 8326 |
+
"learning_rate": 0.0002557384148982244,
|
| 8327 |
+
"loss": 0.03315407782793045,
|
| 8328 |
+
"step": 1186
|
| 8329 |
+
},
|
| 8330 |
+
{
|
| 8331 |
+
"epoch": 0.5771591806965295,
|
| 8332 |
+
"grad_norm": 0.014882663264870644,
|
| 8333 |
+
"learning_rate": 0.0002555218709398008,
|
| 8334 |
+
"loss": 0.03499560058116913,
|
| 8335 |
+
"step": 1187
|
| 8336 |
+
},
|
| 8337 |
+
{
|
| 8338 |
+
"epoch": 0.5776454142101745,
|
| 8339 |
+
"grad_norm": 0.014436531811952591,
|
| 8340 |
+
"learning_rate": 0.0002553053269813772,
|
| 8341 |
+
"loss": 0.03628615289926529,
|
| 8342 |
+
"step": 1188
|
| 8343 |
+
},
|
| 8344 |
+
{
|
| 8345 |
+
"epoch": 0.5781316477238194,
|
| 8346 |
+
"grad_norm": 0.014366181567311287,
|
| 8347 |
+
"learning_rate": 0.0002550887830229537,
|
| 8348 |
+
"loss": 0.0350048765540123,
|
| 8349 |
+
"step": 1189
|
| 8350 |
+
},
|
| 8351 |
+
{
|
| 8352 |
+
"epoch": 0.5786178812374643,
|
| 8353 |
+
"grad_norm": 0.01463479083031416,
|
| 8354 |
+
"learning_rate": 0.0002548722390645301,
|
| 8355 |
+
"loss": 0.037951964884996414,
|
| 8356 |
+
"step": 1190
|
| 8357 |
+
},
|
| 8358 |
+
{
|
| 8359 |
+
"epoch": 0.5791041147511092,
|
| 8360 |
+
"grad_norm": 0.014090332202613354,
|
| 8361 |
+
"learning_rate": 0.0002546556951061065,
|
| 8362 |
+
"loss": 0.035452209413051605,
|
| 8363 |
+
"step": 1191
|
| 8364 |
+
},
|
| 8365 |
+
{
|
| 8366 |
+
"epoch": 0.5795903482647542,
|
| 8367 |
+
"grad_norm": 0.014285637997090816,
|
| 8368 |
+
"learning_rate": 0.000254439151147683,
|
| 8369 |
+
"loss": 0.03433792665600777,
|
| 8370 |
+
"step": 1192
|
| 8371 |
+
},
|
| 8372 |
+
{
|
| 8373 |
+
"epoch": 0.580076581778399,
|
| 8374 |
+
"grad_norm": 0.015341908670961857,
|
| 8375 |
+
"learning_rate": 0.00025422260718925945,
|
| 8376 |
+
"loss": 0.03899161145091057,
|
| 8377 |
+
"step": 1193
|
| 8378 |
+
},
|
| 8379 |
+
{
|
| 8380 |
+
"epoch": 0.580562815292044,
|
| 8381 |
+
"grad_norm": 0.014338607899844646,
|
| 8382 |
+
"learning_rate": 0.00025400606323083586,
|
| 8383 |
+
"loss": 0.028708798810839653,
|
| 8384 |
+
"step": 1194
|
| 8385 |
+
},
|
| 8386 |
+
{
|
| 8387 |
+
"epoch": 0.581049048805689,
|
| 8388 |
+
"grad_norm": 0.014068455435335636,
|
| 8389 |
+
"learning_rate": 0.00025378951927241227,
|
| 8390 |
+
"loss": 0.03241080045700073,
|
| 8391 |
+
"step": 1195
|
| 8392 |
+
},
|
| 8393 |
+
{
|
| 8394 |
+
"epoch": 0.5815352823193338,
|
| 8395 |
+
"grad_norm": 0.014995142817497253,
|
| 8396 |
+
"learning_rate": 0.0002535729753139888,
|
| 8397 |
+
"loss": 0.03661258518695831,
|
| 8398 |
+
"step": 1196
|
| 8399 |
+
},
|
| 8400 |
+
{
|
| 8401 |
+
"epoch": 0.5820215158329788,
|
| 8402 |
+
"grad_norm": 0.014894435182213783,
|
| 8403 |
+
"learning_rate": 0.0002533564313555652,
|
| 8404 |
+
"loss": 0.03713268041610718,
|
| 8405 |
+
"step": 1197
|
| 8406 |
+
},
|
| 8407 |
+
{
|
| 8408 |
+
"epoch": 0.5825077493466237,
|
| 8409 |
+
"grad_norm": 0.014744212850928307,
|
| 8410 |
+
"learning_rate": 0.0002531398873971416,
|
| 8411 |
+
"loss": 0.031973473727703094,
|
| 8412 |
+
"step": 1198
|
| 8413 |
+
},
|
| 8414 |
+
{
|
| 8415 |
+
"epoch": 0.5829939828602686,
|
| 8416 |
+
"grad_norm": 0.014299332164227962,
|
| 8417 |
+
"learning_rate": 0.0002529233434387181,
|
| 8418 |
+
"loss": 0.03431545943021774,
|
| 8419 |
+
"step": 1199
|
| 8420 |
+
},
|
| 8421 |
+
{
|
| 8422 |
+
"epoch": 0.5834802163739136,
|
| 8423 |
+
"grad_norm": 0.014882242307066917,
|
| 8424 |
+
"learning_rate": 0.0002527067994802945,
|
| 8425 |
+
"loss": 0.03584323078393936,
|
| 8426 |
+
"step": 1200
|
| 8427 |
+
},
|
| 8428 |
+
{
|
| 8429 |
+
"epoch": 0.5839664498875585,
|
| 8430 |
+
"grad_norm": 0.014170056208968163,
|
| 8431 |
+
"learning_rate": 0.00025249025552187093,
|
| 8432 |
+
"loss": 0.034378085285425186,
|
| 8433 |
+
"step": 1201
|
| 8434 |
+
},
|
| 8435 |
+
{
|
| 8436 |
+
"epoch": 0.5844526834012034,
|
| 8437 |
+
"grad_norm": 0.014155167154967785,
|
| 8438 |
+
"learning_rate": 0.00025227371156344734,
|
| 8439 |
+
"loss": 0.034098751842975616,
|
| 8440 |
+
"step": 1202
|
| 8441 |
+
},
|
| 8442 |
+
{
|
| 8443 |
+
"epoch": 0.5849389169148483,
|
| 8444 |
+
"grad_norm": 0.015772607177495956,
|
| 8445 |
+
"learning_rate": 0.00025205716760502386,
|
| 8446 |
+
"loss": 0.031616367399692535,
|
| 8447 |
+
"step": 1203
|
| 8448 |
+
},
|
| 8449 |
+
{
|
| 8450 |
+
"epoch": 0.5854251504284933,
|
| 8451 |
+
"grad_norm": 0.013857257552444935,
|
| 8452 |
+
"learning_rate": 0.00025184062364660026,
|
| 8453 |
+
"loss": 0.034753598272800446,
|
| 8454 |
+
"step": 1204
|
| 8455 |
+
},
|
| 8456 |
+
{
|
| 8457 |
+
"epoch": 0.5859113839421383,
|
| 8458 |
+
"grad_norm": 0.01471875049173832,
|
| 8459 |
+
"learning_rate": 0.00025162407968817667,
|
| 8460 |
+
"loss": 0.03113594464957714,
|
| 8461 |
+
"step": 1205
|
| 8462 |
+
},
|
| 8463 |
+
{
|
| 8464 |
+
"epoch": 0.5863976174557831,
|
| 8465 |
+
"grad_norm": 0.014296879060566425,
|
| 8466 |
+
"learning_rate": 0.0002514075357297532,
|
| 8467 |
+
"loss": 0.03316786140203476,
|
| 8468 |
+
"step": 1206
|
| 8469 |
+
},
|
| 8470 |
+
{
|
| 8471 |
+
"epoch": 0.5868838509694281,
|
| 8472 |
+
"grad_norm": 0.013809135183691978,
|
| 8473 |
+
"learning_rate": 0.0002511909917713296,
|
| 8474 |
+
"loss": 0.03172443062067032,
|
| 8475 |
+
"step": 1207
|
| 8476 |
+
},
|
| 8477 |
+
{
|
| 8478 |
+
"epoch": 0.587370084483073,
|
| 8479 |
+
"grad_norm": 0.014519846998155117,
|
| 8480 |
+
"learning_rate": 0.000250974447812906,
|
| 8481 |
+
"loss": 0.03601034730672836,
|
| 8482 |
+
"step": 1208
|
| 8483 |
+
},
|
| 8484 |
+
{
|
| 8485 |
+
"epoch": 0.5878563179967179,
|
| 8486 |
+
"grad_norm": 0.01552652195096016,
|
| 8487 |
+
"learning_rate": 0.0002507579038544824,
|
| 8488 |
+
"loss": 0.03719080239534378,
|
| 8489 |
+
"step": 1209
|
| 8490 |
+
},
|
| 8491 |
+
{
|
| 8492 |
+
"epoch": 0.5883425515103629,
|
| 8493 |
+
"grad_norm": 0.015220316126942635,
|
| 8494 |
+
"learning_rate": 0.00025054135989605893,
|
| 8495 |
+
"loss": 0.03144580125808716,
|
| 8496 |
+
"step": 1210
|
| 8497 |
+
},
|
| 8498 |
+
{
|
| 8499 |
+
"epoch": 0.5888287850240078,
|
| 8500 |
+
"grad_norm": 0.01446231734007597,
|
| 8501 |
+
"learning_rate": 0.00025032481593763534,
|
| 8502 |
+
"loss": 0.03810115531086922,
|
| 8503 |
+
"step": 1211
|
| 8504 |
+
},
|
| 8505 |
+
{
|
| 8506 |
+
"epoch": 0.5893150185376527,
|
| 8507 |
+
"grad_norm": 0.014120371080935001,
|
| 8508 |
+
"learning_rate": 0.00025010827197921175,
|
| 8509 |
+
"loss": 0.03405807167291641,
|
| 8510 |
+
"step": 1212
|
| 8511 |
+
},
|
| 8512 |
+
{
|
| 8513 |
+
"epoch": 0.5898012520512976,
|
| 8514 |
+
"grad_norm": 0.014574095606803894,
|
| 8515 |
+
"learning_rate": 0.0002498917280207882,
|
| 8516 |
+
"loss": 0.032808732241392136,
|
| 8517 |
+
"step": 1213
|
| 8518 |
+
},
|
| 8519 |
+
{
|
| 8520 |
+
"epoch": 0.5902874855649426,
|
| 8521 |
+
"grad_norm": 0.01470918022096157,
|
| 8522 |
+
"learning_rate": 0.00024967518406236467,
|
| 8523 |
+
"loss": 0.035265203565359116,
|
| 8524 |
+
"step": 1214
|
| 8525 |
+
},
|
| 8526 |
+
{
|
| 8527 |
+
"epoch": 0.5907737190785874,
|
| 8528 |
+
"grad_norm": 0.014463575556874275,
|
| 8529 |
+
"learning_rate": 0.0002494586401039411,
|
| 8530 |
+
"loss": 0.03249891847372055,
|
| 8531 |
+
"step": 1215
|
| 8532 |
+
},
|
| 8533 |
+
{
|
| 8534 |
+
"epoch": 0.5912599525922324,
|
| 8535 |
+
"grad_norm": 0.01371840015053749,
|
| 8536 |
+
"learning_rate": 0.00024924209614551754,
|
| 8537 |
+
"loss": 0.03212244436144829,
|
| 8538 |
+
"step": 1216
|
| 8539 |
+
},
|
| 8540 |
+
{
|
| 8541 |
+
"epoch": 0.5917461861058774,
|
| 8542 |
+
"grad_norm": 0.013899889774620533,
|
| 8543 |
+
"learning_rate": 0.000249025552187094,
|
| 8544 |
+
"loss": 0.033638618886470795,
|
| 8545 |
+
"step": 1217
|
| 8546 |
+
},
|
| 8547 |
+
{
|
| 8548 |
+
"epoch": 0.5922324196195222,
|
| 8549 |
+
"grad_norm": 0.014456473290920258,
|
| 8550 |
+
"learning_rate": 0.0002488090082286704,
|
| 8551 |
+
"loss": 0.03235378488898277,
|
| 8552 |
+
"step": 1218
|
| 8553 |
+
},
|
| 8554 |
+
{
|
| 8555 |
+
"epoch": 0.5927186531331672,
|
| 8556 |
+
"grad_norm": 0.014223506674170494,
|
| 8557 |
+
"learning_rate": 0.0002485924642702469,
|
| 8558 |
+
"loss": 0.03393813222646713,
|
| 8559 |
+
"step": 1219
|
| 8560 |
+
},
|
| 8561 |
+
{
|
| 8562 |
+
"epoch": 0.5932048866468121,
|
| 8563 |
+
"grad_norm": 0.0147509491071105,
|
| 8564 |
+
"learning_rate": 0.0002483759203118233,
|
| 8565 |
+
"loss": 0.03550096973776817,
|
| 8566 |
+
"step": 1220
|
| 8567 |
+
},
|
| 8568 |
+
{
|
| 8569 |
+
"epoch": 0.5936911201604571,
|
| 8570 |
+
"grad_norm": 0.013944664038717747,
|
| 8571 |
+
"learning_rate": 0.00024815937635339975,
|
| 8572 |
+
"loss": 0.0346209742128849,
|
| 8573 |
+
"step": 1221
|
| 8574 |
+
},
|
| 8575 |
+
{
|
| 8576 |
+
"epoch": 0.594177353674102,
|
| 8577 |
+
"grad_norm": 0.017330098897218704,
|
| 8578 |
+
"learning_rate": 0.0002479428323949762,
|
| 8579 |
+
"loss": 0.03434748202562332,
|
| 8580 |
+
"step": 1222
|
| 8581 |
+
},
|
| 8582 |
+
{
|
| 8583 |
+
"epoch": 0.5946635871877469,
|
| 8584 |
+
"grad_norm": 0.013888601213693619,
|
| 8585 |
+
"learning_rate": 0.0002477262884365526,
|
| 8586 |
+
"loss": 0.03292950987815857,
|
| 8587 |
+
"step": 1223
|
| 8588 |
+
},
|
| 8589 |
+
{
|
| 8590 |
+
"epoch": 0.5951498207013919,
|
| 8591 |
+
"grad_norm": 0.013507389463484287,
|
| 8592 |
+
"learning_rate": 0.0002475097444781291,
|
| 8593 |
+
"loss": 0.030300376936793327,
|
| 8594 |
+
"step": 1224
|
| 8595 |
+
},
|
| 8596 |
+
{
|
| 8597 |
+
"epoch": 0.5956360542150367,
|
| 8598 |
+
"grad_norm": 0.014849923551082611,
|
| 8599 |
+
"learning_rate": 0.0002472932005197055,
|
| 8600 |
+
"loss": 0.03718724846839905,
|
| 8601 |
+
"step": 1225
|
| 8602 |
+
},
|
| 8603 |
+
{
|
| 8604 |
+
"epoch": 0.5961222877286817,
|
| 8605 |
+
"grad_norm": 0.013127702288329601,
|
| 8606 |
+
"learning_rate": 0.00024707665656128195,
|
| 8607 |
+
"loss": 0.03642948344349861,
|
| 8608 |
+
"step": 1226
|
| 8609 |
+
},
|
| 8610 |
+
{
|
| 8611 |
+
"epoch": 0.5966085212423267,
|
| 8612 |
+
"grad_norm": 0.014370061457157135,
|
| 8613 |
+
"learning_rate": 0.00024686011260285836,
|
| 8614 |
+
"loss": 0.036296501755714417,
|
| 8615 |
+
"step": 1227
|
| 8616 |
+
},
|
| 8617 |
+
{
|
| 8618 |
+
"epoch": 0.5970947547559715,
|
| 8619 |
+
"grad_norm": 0.014229833148419857,
|
| 8620 |
+
"learning_rate": 0.0002466435686444348,
|
| 8621 |
+
"loss": 0.03772621974349022,
|
| 8622 |
+
"step": 1228
|
| 8623 |
+
},
|
| 8624 |
+
{
|
| 8625 |
+
"epoch": 0.5975809882696165,
|
| 8626 |
+
"grad_norm": 0.014395033940672874,
|
| 8627 |
+
"learning_rate": 0.0002464270246860113,
|
| 8628 |
+
"loss": 0.03394228592514992,
|
| 8629 |
+
"step": 1229
|
| 8630 |
+
},
|
| 8631 |
+
{
|
| 8632 |
+
"epoch": 0.5980672217832614,
|
| 8633 |
+
"grad_norm": 0.014203759841620922,
|
| 8634 |
+
"learning_rate": 0.0002462104807275877,
|
| 8635 |
+
"loss": 0.03398077189922333,
|
| 8636 |
+
"step": 1230
|
| 8637 |
+
},
|
| 8638 |
+
{
|
| 8639 |
+
"epoch": 0.5985534552969063,
|
| 8640 |
+
"grad_norm": 0.014437762089073658,
|
| 8641 |
+
"learning_rate": 0.00024599393676916415,
|
| 8642 |
+
"loss": 0.028665797784924507,
|
| 8643 |
+
"step": 1231
|
| 8644 |
+
},
|
| 8645 |
+
{
|
| 8646 |
+
"epoch": 0.5990396888105513,
|
| 8647 |
+
"grad_norm": 0.014124961569905281,
|
| 8648 |
+
"learning_rate": 0.00024577739281074056,
|
| 8649 |
+
"loss": 0.035838231444358826,
|
| 8650 |
+
"step": 1232
|
| 8651 |
+
},
|
| 8652 |
+
{
|
| 8653 |
+
"epoch": 0.5995259223241962,
|
| 8654 |
+
"grad_norm": 0.013696972280740738,
|
| 8655 |
+
"learning_rate": 0.000245560848852317,
|
| 8656 |
+
"loss": 0.034617070108652115,
|
| 8657 |
+
"step": 1233
|
| 8658 |
+
},
|
| 8659 |
+
{
|
| 8660 |
+
"epoch": 0.6000121558378412,
|
| 8661 |
+
"grad_norm": 0.013875085860490799,
|
| 8662 |
+
"learning_rate": 0.0002453443048938935,
|
| 8663 |
+
"loss": 0.03617449849843979,
|
| 8664 |
+
"step": 1234
|
| 8665 |
+
},
|
| 8666 |
+
{
|
| 8667 |
+
"epoch": 0.600498389351486,
|
| 8668 |
+
"grad_norm": 0.01313305739313364,
|
| 8669 |
+
"learning_rate": 0.0002451277609354699,
|
| 8670 |
+
"loss": 0.03542593866586685,
|
| 8671 |
+
"step": 1235
|
| 8672 |
+
},
|
| 8673 |
+
{
|
| 8674 |
+
"epoch": 0.600984622865131,
|
| 8675 |
+
"grad_norm": 0.014472343027591705,
|
| 8676 |
+
"learning_rate": 0.00024491121697704636,
|
| 8677 |
+
"loss": 0.032386451959609985,
|
| 8678 |
+
"step": 1236
|
| 8679 |
+
},
|
| 8680 |
+
{
|
| 8681 |
+
"epoch": 0.601470856378776,
|
| 8682 |
+
"grad_norm": 0.014684909023344517,
|
| 8683 |
+
"learning_rate": 0.00024469467301862276,
|
| 8684 |
+
"loss": 0.0343945138156414,
|
| 8685 |
+
"step": 1237
|
| 8686 |
+
},
|
| 8687 |
+
{
|
| 8688 |
+
"epoch": 0.6019570898924208,
|
| 8689 |
+
"grad_norm": 0.014352599158883095,
|
| 8690 |
+
"learning_rate": 0.0002444781290601992,
|
| 8691 |
+
"loss": 0.03243408724665642,
|
| 8692 |
+
"step": 1238
|
| 8693 |
+
},
|
| 8694 |
+
{
|
| 8695 |
+
"epoch": 0.6024433234060658,
|
| 8696 |
+
"grad_norm": 0.014815051108598709,
|
| 8697 |
+
"learning_rate": 0.00024426158510177564,
|
| 8698 |
+
"loss": 0.036052584648132324,
|
| 8699 |
+
"step": 1239
|
| 8700 |
+
},
|
| 8701 |
+
{
|
| 8702 |
+
"epoch": 0.6029295569197107,
|
| 8703 |
+
"grad_norm": 0.015110349282622337,
|
| 8704 |
+
"learning_rate": 0.0002440450411433521,
|
| 8705 |
+
"loss": 0.035057857632637024,
|
| 8706 |
+
"step": 1240
|
| 8707 |
+
},
|
| 8708 |
+
{
|
| 8709 |
+
"epoch": 0.6034157904333556,
|
| 8710 |
+
"grad_norm": 0.01499603409320116,
|
| 8711 |
+
"learning_rate": 0.00024382849718492856,
|
| 8712 |
+
"loss": 0.03447789326310158,
|
| 8713 |
+
"step": 1241
|
| 8714 |
+
},
|
| 8715 |
+
{
|
| 8716 |
+
"epoch": 0.6039020239470005,
|
| 8717 |
+
"grad_norm": 0.013547827489674091,
|
| 8718 |
+
"learning_rate": 0.00024361195322650497,
|
| 8719 |
+
"loss": 0.036869995296001434,
|
| 8720 |
+
"step": 1242
|
| 8721 |
+
},
|
| 8722 |
+
{
|
| 8723 |
+
"epoch": 0.6043882574606455,
|
| 8724 |
+
"grad_norm": 0.014528974890708923,
|
| 8725 |
+
"learning_rate": 0.00024339540926808143,
|
| 8726 |
+
"loss": 0.03351128101348877,
|
| 8727 |
+
"step": 1243
|
| 8728 |
+
},
|
| 8729 |
+
{
|
| 8730 |
+
"epoch": 0.6048744909742904,
|
| 8731 |
+
"grad_norm": 0.013727253302931786,
|
| 8732 |
+
"learning_rate": 0.00024317886530965787,
|
| 8733 |
+
"loss": 0.031914278864860535,
|
| 8734 |
+
"step": 1244
|
| 8735 |
+
},
|
| 8736 |
+
{
|
| 8737 |
+
"epoch": 0.6053607244879353,
|
| 8738 |
+
"grad_norm": 0.013836268335580826,
|
| 8739 |
+
"learning_rate": 0.0002429623213512343,
|
| 8740 |
+
"loss": 0.03593692183494568,
|
| 8741 |
+
"step": 1245
|
| 8742 |
+
},
|
| 8743 |
+
{
|
| 8744 |
+
"epoch": 0.6058469580015803,
|
| 8745 |
+
"grad_norm": 0.013676434755325317,
|
| 8746 |
+
"learning_rate": 0.00024274577739281076,
|
| 8747 |
+
"loss": 0.030865518376231194,
|
| 8748 |
+
"step": 1246
|
| 8749 |
+
},
|
| 8750 |
+
{
|
| 8751 |
+
"epoch": 0.6063331915152251,
|
| 8752 |
+
"grad_norm": 0.014790666289627552,
|
| 8753 |
+
"learning_rate": 0.00024252923343438717,
|
| 8754 |
+
"loss": 0.03735022619366646,
|
| 8755 |
+
"step": 1247
|
| 8756 |
+
},
|
| 8757 |
+
{
|
| 8758 |
+
"epoch": 0.6068194250288701,
|
| 8759 |
+
"grad_norm": 0.013988127000629902,
|
| 8760 |
+
"learning_rate": 0.00024231268947596363,
|
| 8761 |
+
"loss": 0.03215673193335533,
|
| 8762 |
+
"step": 1248
|
| 8763 |
+
},
|
| 8764 |
+
{
|
| 8765 |
+
"epoch": 0.6073056585425151,
|
| 8766 |
+
"grad_norm": 0.013941338285803795,
|
| 8767 |
+
"learning_rate": 0.00024209614551754007,
|
| 8768 |
+
"loss": 0.037025727331638336,
|
| 8769 |
+
"step": 1249
|
| 8770 |
+
},
|
| 8771 |
+
{
|
| 8772 |
+
"epoch": 0.60779189205616,
|
| 8773 |
+
"grad_norm": 0.013766956515610218,
|
| 8774 |
+
"learning_rate": 0.0002418796015591165,
|
| 8775 |
+
"loss": 0.03790061175823212,
|
| 8776 |
+
"step": 1250
|
| 8777 |
+
},
|
| 8778 |
+
{
|
| 8779 |
+
"epoch": 0.6082781255698049,
|
| 8780 |
+
"grad_norm": 0.01456359587609768,
|
| 8781 |
+
"learning_rate": 0.00024166305760069297,
|
| 8782 |
+
"loss": 0.03590507432818413,
|
| 8783 |
+
"step": 1251
|
| 8784 |
+
},
|
| 8785 |
+
{
|
| 8786 |
+
"epoch": 0.6087643590834498,
|
| 8787 |
+
"grad_norm": 0.014707593247294426,
|
| 8788 |
+
"learning_rate": 0.00024144651364226938,
|
| 8789 |
+
"loss": 0.03497236222028732,
|
| 8790 |
+
"step": 1252
|
| 8791 |
+
},
|
| 8792 |
+
{
|
| 8793 |
+
"epoch": 0.6092505925970948,
|
| 8794 |
+
"grad_norm": 0.0143510140478611,
|
| 8795 |
+
"learning_rate": 0.00024122996968384584,
|
| 8796 |
+
"loss": 0.031629275530576706,
|
| 8797 |
+
"step": 1253
|
| 8798 |
+
},
|
| 8799 |
+
{
|
| 8800 |
+
"epoch": 0.6097368261107396,
|
| 8801 |
+
"grad_norm": 0.014503378421068192,
|
| 8802 |
+
"learning_rate": 0.00024101342572542225,
|
| 8803 |
+
"loss": 0.033413030207157135,
|
| 8804 |
+
"step": 1254
|
| 8805 |
+
},
|
| 8806 |
+
{
|
| 8807 |
+
"epoch": 0.6102230596243846,
|
| 8808 |
+
"grad_norm": 0.014512976631522179,
|
| 8809 |
+
"learning_rate": 0.0002407968817669987,
|
| 8810 |
+
"loss": 0.03332623839378357,
|
| 8811 |
+
"step": 1255
|
| 8812 |
+
},
|
| 8813 |
+
{
|
| 8814 |
+
"epoch": 0.6107092931380296,
|
| 8815 |
+
"grad_norm": 0.013471649959683418,
|
| 8816 |
+
"learning_rate": 0.00024058033780857514,
|
| 8817 |
+
"loss": 0.03274286910891533,
|
| 8818 |
+
"step": 1256
|
| 8819 |
+
},
|
| 8820 |
+
{
|
| 8821 |
+
"epoch": 0.6111955266516744,
|
| 8822 |
+
"grad_norm": 0.014295383356511593,
|
| 8823 |
+
"learning_rate": 0.00024036379385015158,
|
| 8824 |
+
"loss": 0.033239275217056274,
|
| 8825 |
+
"step": 1257
|
| 8826 |
+
},
|
| 8827 |
+
{
|
| 8828 |
+
"epoch": 0.6116817601653194,
|
| 8829 |
+
"grad_norm": 0.015325219370424747,
|
| 8830 |
+
"learning_rate": 0.00024014724989172804,
|
| 8831 |
+
"loss": 0.03931151330471039,
|
| 8832 |
+
"step": 1258
|
| 8833 |
+
},
|
| 8834 |
+
{
|
| 8835 |
+
"epoch": 0.6121679936789644,
|
| 8836 |
+
"grad_norm": 0.014009041711688042,
|
| 8837 |
+
"learning_rate": 0.00023993070593330445,
|
| 8838 |
+
"loss": 0.035900142043828964,
|
| 8839 |
+
"step": 1259
|
| 8840 |
+
},
|
| 8841 |
+
{
|
| 8842 |
+
"epoch": 0.6126542271926092,
|
| 8843 |
+
"grad_norm": 0.013827085494995117,
|
| 8844 |
+
"learning_rate": 0.0002397141619748809,
|
| 8845 |
+
"loss": 0.03658423200249672,
|
| 8846 |
+
"step": 1260
|
| 8847 |
+
},
|
| 8848 |
+
{
|
| 8849 |
+
"epoch": 0.6131404607062542,
|
| 8850 |
+
"grad_norm": 0.024001143872737885,
|
| 8851 |
+
"learning_rate": 0.00023949761801645735,
|
| 8852 |
+
"loss": 0.035441312938928604,
|
| 8853 |
+
"step": 1261
|
| 8854 |
+
},
|
| 8855 |
+
{
|
| 8856 |
+
"epoch": 0.6136266942198991,
|
| 8857 |
+
"grad_norm": 0.013768965378403664,
|
| 8858 |
+
"learning_rate": 0.00023928107405803378,
|
| 8859 |
+
"loss": 0.036932703107595444,
|
| 8860 |
+
"step": 1262
|
| 8861 |
+
},
|
| 8862 |
+
{
|
| 8863 |
+
"epoch": 0.6141129277335441,
|
| 8864 |
+
"grad_norm": 0.013339557684957981,
|
| 8865 |
+
"learning_rate": 0.00023906453009961024,
|
| 8866 |
+
"loss": 0.03353942558169365,
|
| 8867 |
+
"step": 1263
|
| 8868 |
+
},
|
| 8869 |
+
{
|
| 8870 |
+
"epoch": 0.6145991612471889,
|
| 8871 |
+
"grad_norm": 0.014980307780206203,
|
| 8872 |
+
"learning_rate": 0.00023884798614118665,
|
| 8873 |
+
"loss": 0.03243682160973549,
|
| 8874 |
+
"step": 1264
|
| 8875 |
+
},
|
| 8876 |
+
{
|
| 8877 |
+
"epoch": 0.6150853947608339,
|
| 8878 |
+
"grad_norm": 0.014316984452307224,
|
| 8879 |
+
"learning_rate": 0.00023863144218276312,
|
| 8880 |
+
"loss": 0.03328140079975128,
|
| 8881 |
+
"step": 1265
|
| 8882 |
+
},
|
| 8883 |
+
{
|
| 8884 |
+
"epoch": 0.6155716282744789,
|
| 8885 |
+
"grad_norm": 0.014322753064334393,
|
| 8886 |
+
"learning_rate": 0.00023841489822433955,
|
| 8887 |
+
"loss": 0.03490499034523964,
|
| 8888 |
+
"step": 1266
|
| 8889 |
+
},
|
| 8890 |
+
{
|
| 8891 |
+
"epoch": 0.6160578617881237,
|
| 8892 |
+
"grad_norm": 0.01352720521390438,
|
| 8893 |
+
"learning_rate": 0.00023819835426591599,
|
| 8894 |
+
"loss": 0.03322020545601845,
|
| 8895 |
+
"step": 1267
|
| 8896 |
+
},
|
| 8897 |
+
{
|
| 8898 |
+
"epoch": 0.6165440953017687,
|
| 8899 |
+
"grad_norm": 0.01417227741330862,
|
| 8900 |
+
"learning_rate": 0.00023798181030749242,
|
| 8901 |
+
"loss": 0.034942928701639175,
|
| 8902 |
+
"step": 1268
|
| 8903 |
+
},
|
| 8904 |
+
{
|
| 8905 |
+
"epoch": 0.6170303288154136,
|
| 8906 |
+
"grad_norm": 0.014059393666684628,
|
| 8907 |
+
"learning_rate": 0.00023776526634906886,
|
| 8908 |
+
"loss": 0.03696153685450554,
|
| 8909 |
+
"step": 1269
|
| 8910 |
+
},
|
| 8911 |
+
{
|
| 8912 |
+
"epoch": 0.6175165623290585,
|
| 8913 |
+
"grad_norm": 0.014172044582664967,
|
| 8914 |
+
"learning_rate": 0.00023754872239064532,
|
| 8915 |
+
"loss": 0.039092909544706345,
|
| 8916 |
+
"step": 1270
|
| 8917 |
+
},
|
| 8918 |
+
{
|
| 8919 |
+
"epoch": 0.6180027958427035,
|
| 8920 |
+
"grad_norm": 0.01400654949247837,
|
| 8921 |
+
"learning_rate": 0.00023733217843222175,
|
| 8922 |
+
"loss": 0.036471493542194366,
|
| 8923 |
+
"step": 1271
|
| 8924 |
+
},
|
| 8925 |
+
{
|
| 8926 |
+
"epoch": 0.6184890293563484,
|
| 8927 |
+
"grad_norm": 0.013951717875897884,
|
| 8928 |
+
"learning_rate": 0.0002371156344737982,
|
| 8929 |
+
"loss": 0.03401320427656174,
|
| 8930 |
+
"step": 1272
|
| 8931 |
+
},
|
| 8932 |
+
{
|
| 8933 |
+
"epoch": 0.6189752628699933,
|
| 8934 |
+
"grad_norm": 0.014873258769512177,
|
| 8935 |
+
"learning_rate": 0.00023689909051537462,
|
| 8936 |
+
"loss": 0.03410210460424423,
|
| 8937 |
+
"step": 1273
|
| 8938 |
+
},
|
| 8939 |
+
{
|
| 8940 |
+
"epoch": 0.6194614963836382,
|
| 8941 |
+
"grad_norm": 0.014345663599669933,
|
| 8942 |
+
"learning_rate": 0.00023668254655695106,
|
| 8943 |
+
"loss": 0.03397146239876747,
|
| 8944 |
+
"step": 1274
|
| 8945 |
+
},
|
| 8946 |
+
{
|
| 8947 |
+
"epoch": 0.6199477298972832,
|
| 8948 |
+
"grad_norm": 0.014295048080384731,
|
| 8949 |
+
"learning_rate": 0.00023646600259852752,
|
| 8950 |
+
"loss": 0.030897531658411026,
|
| 8951 |
+
"step": 1275
|
| 8952 |
+
},
|
| 8953 |
+
{
|
| 8954 |
+
"epoch": 0.620433963410928,
|
| 8955 |
+
"grad_norm": 0.014968395233154297,
|
| 8956 |
+
"learning_rate": 0.00023624945864010393,
|
| 8957 |
+
"loss": 0.03531058877706528,
|
| 8958 |
+
"step": 1276
|
| 8959 |
+
},
|
| 8960 |
+
{
|
| 8961 |
+
"epoch": 0.620920196924573,
|
| 8962 |
+
"grad_norm": 0.013523002155125141,
|
| 8963 |
+
"learning_rate": 0.0002360329146816804,
|
| 8964 |
+
"loss": 0.036181703209877014,
|
| 8965 |
+
"step": 1277
|
| 8966 |
+
},
|
| 8967 |
+
{
|
| 8968 |
+
"epoch": 0.621406430438218,
|
| 8969 |
+
"grad_norm": 0.014598391018807888,
|
| 8970 |
+
"learning_rate": 0.00023581637072325683,
|
| 8971 |
+
"loss": 0.03319678455591202,
|
| 8972 |
+
"step": 1278
|
| 8973 |
+
},
|
| 8974 |
+
{
|
| 8975 |
+
"epoch": 0.6218926639518629,
|
| 8976 |
+
"grad_norm": 0.015035644173622131,
|
| 8977 |
+
"learning_rate": 0.00023559982676483326,
|
| 8978 |
+
"loss": 0.03545587882399559,
|
| 8979 |
+
"step": 1279
|
| 8980 |
+
},
|
| 8981 |
+
{
|
| 8982 |
+
"epoch": 0.6223788974655078,
|
| 8983 |
+
"grad_norm": 0.01414641086012125,
|
| 8984 |
+
"learning_rate": 0.0002353832828064097,
|
| 8985 |
+
"loss": 0.03653448820114136,
|
| 8986 |
+
"step": 1280
|
| 8987 |
+
},
|
| 8988 |
+
{
|
| 8989 |
+
"epoch": 0.6228651309791527,
|
| 8990 |
+
"grad_norm": 0.014127531088888645,
|
| 8991 |
+
"learning_rate": 0.00023516673884798613,
|
| 8992 |
+
"loss": 0.034944504499435425,
|
| 8993 |
+
"step": 1281
|
| 8994 |
+
},
|
| 8995 |
+
{
|
| 8996 |
+
"epoch": 0.6233513644927977,
|
| 8997 |
+
"grad_norm": 0.013686790131032467,
|
| 8998 |
+
"learning_rate": 0.0002349501948895626,
|
| 8999 |
+
"loss": 0.03326250612735748,
|
| 9000 |
+
"step": 1282
|
| 9001 |
+
},
|
| 9002 |
+
{
|
| 9003 |
+
"epoch": 0.6238375980064426,
|
| 9004 |
+
"grad_norm": 0.013996295630931854,
|
| 9005 |
+
"learning_rate": 0.00023473365093113903,
|
| 9006 |
+
"loss": 0.03417070955038071,
|
| 9007 |
+
"step": 1283
|
| 9008 |
+
},
|
| 9009 |
+
{
|
| 9010 |
+
"epoch": 0.6243238315200875,
|
| 9011 |
+
"grad_norm": 0.014012682251632214,
|
| 9012 |
+
"learning_rate": 0.00023451710697271547,
|
| 9013 |
+
"loss": 0.033247947692871094,
|
| 9014 |
+
"step": 1284
|
| 9015 |
+
},
|
| 9016 |
+
{
|
| 9017 |
+
"epoch": 0.6248100650337325,
|
| 9018 |
+
"grad_norm": 0.013572361320257187,
|
| 9019 |
+
"learning_rate": 0.0002343005630142919,
|
| 9020 |
+
"loss": 0.03197130188345909,
|
| 9021 |
+
"step": 1285
|
| 9022 |
+
},
|
| 9023 |
+
{
|
| 9024 |
+
"epoch": 0.6252962985473773,
|
| 9025 |
+
"grad_norm": 0.014377618208527565,
|
| 9026 |
+
"learning_rate": 0.00023408401905586834,
|
| 9027 |
+
"loss": 0.03632620722055435,
|
| 9028 |
+
"step": 1286
|
| 9029 |
+
},
|
| 9030 |
+
{
|
| 9031 |
+
"epoch": 0.6257825320610223,
|
| 9032 |
+
"grad_norm": 0.014611482620239258,
|
| 9033 |
+
"learning_rate": 0.0002338674750974448,
|
| 9034 |
+
"loss": 0.03583304584026337,
|
| 9035 |
+
"step": 1287
|
| 9036 |
+
},
|
| 9037 |
+
{
|
| 9038 |
+
"epoch": 0.6262687655746673,
|
| 9039 |
+
"grad_norm": 0.014583008363842964,
|
| 9040 |
+
"learning_rate": 0.00023365093113902123,
|
| 9041 |
+
"loss": 0.03570983558893204,
|
| 9042 |
+
"step": 1288
|
| 9043 |
+
},
|
| 9044 |
+
{
|
| 9045 |
+
"epoch": 0.6267549990883121,
|
| 9046 |
+
"grad_norm": 0.015231940895318985,
|
| 9047 |
+
"learning_rate": 0.00023343438718059767,
|
| 9048 |
+
"loss": 0.03442798927426338,
|
| 9049 |
+
"step": 1289
|
| 9050 |
+
},
|
| 9051 |
+
{
|
| 9052 |
+
"epoch": 0.6272412326019571,
|
| 9053 |
+
"grad_norm": 0.013659853488206863,
|
| 9054 |
+
"learning_rate": 0.0002332178432221741,
|
| 9055 |
+
"loss": 0.033729713410139084,
|
| 9056 |
+
"step": 1290
|
| 9057 |
+
},
|
| 9058 |
+
{
|
| 9059 |
+
"epoch": 0.627727466115602,
|
| 9060 |
+
"grad_norm": 0.014678172767162323,
|
| 9061 |
+
"learning_rate": 0.00023300129926375054,
|
| 9062 |
+
"loss": 0.03621881455183029,
|
| 9063 |
+
"step": 1291
|
| 9064 |
+
},
|
| 9065 |
+
{
|
| 9066 |
+
"epoch": 0.6282136996292469,
|
| 9067 |
+
"grad_norm": 0.013895882293581963,
|
| 9068 |
+
"learning_rate": 0.00023278475530532698,
|
| 9069 |
+
"loss": 0.03388817980885506,
|
| 9070 |
+
"step": 1292
|
| 9071 |
+
},
|
| 9072 |
+
{
|
| 9073 |
+
"epoch": 0.6286999331428919,
|
| 9074 |
+
"grad_norm": 0.014055908657610416,
|
| 9075 |
+
"learning_rate": 0.0002325682113469034,
|
| 9076 |
+
"loss": 0.0349176786839962,
|
| 9077 |
+
"step": 1293
|
| 9078 |
+
},
|
| 9079 |
+
{
|
| 9080 |
+
"epoch": 0.6291861666565368,
|
| 9081 |
+
"grad_norm": 0.013777400366961956,
|
| 9082 |
+
"learning_rate": 0.00023235166738847987,
|
| 9083 |
+
"loss": 0.037194207310676575,
|
| 9084 |
+
"step": 1294
|
| 9085 |
+
},
|
| 9086 |
+
{
|
| 9087 |
+
"epoch": 0.6296724001701818,
|
| 9088 |
+
"grad_norm": 0.013084378093481064,
|
| 9089 |
+
"learning_rate": 0.0002321351234300563,
|
| 9090 |
+
"loss": 0.03343290463089943,
|
| 9091 |
+
"step": 1295
|
| 9092 |
+
},
|
| 9093 |
+
{
|
| 9094 |
+
"epoch": 0.6301586336838266,
|
| 9095 |
+
"grad_norm": 0.014204435981810093,
|
| 9096 |
+
"learning_rate": 0.00023191857947163274,
|
| 9097 |
+
"loss": 0.03391396999359131,
|
| 9098 |
+
"step": 1296
|
| 9099 |
+
},
|
| 9100 |
+
{
|
| 9101 |
+
"epoch": 0.6306448671974716,
|
| 9102 |
+
"grad_norm": 0.01398822944611311,
|
| 9103 |
+
"learning_rate": 0.00023170203551320918,
|
| 9104 |
+
"loss": 0.031173840165138245,
|
| 9105 |
+
"step": 1297
|
| 9106 |
+
},
|
| 9107 |
+
{
|
| 9108 |
+
"epoch": 0.6311311007111166,
|
| 9109 |
+
"grad_norm": 0.014111969619989395,
|
| 9110 |
+
"learning_rate": 0.00023148549155478561,
|
| 9111 |
+
"loss": 0.03307380527257919,
|
| 9112 |
+
"step": 1298
|
| 9113 |
+
},
|
| 9114 |
+
{
|
| 9115 |
+
"epoch": 0.6316173342247614,
|
| 9116 |
+
"grad_norm": 0.014751107431948185,
|
| 9117 |
+
"learning_rate": 0.00023126894759636208,
|
| 9118 |
+
"loss": 0.034384749829769135,
|
| 9119 |
+
"step": 1299
|
| 9120 |
+
},
|
| 9121 |
+
{
|
| 9122 |
+
"epoch": 0.6321035677384064,
|
| 9123 |
+
"grad_norm": 0.01362798921763897,
|
| 9124 |
+
"learning_rate": 0.0002310524036379385,
|
| 9125 |
+
"loss": 0.03027649223804474,
|
| 9126 |
+
"step": 1300
|
| 9127 |
+
},
|
| 9128 |
+
{
|
| 9129 |
+
"epoch": 0.6325898012520513,
|
| 9130 |
+
"grad_norm": 0.015399159863591194,
|
| 9131 |
+
"learning_rate": 0.00023083585967951495,
|
| 9132 |
+
"loss": 0.03429962694644928,
|
| 9133 |
+
"step": 1301
|
| 9134 |
+
},
|
| 9135 |
+
{
|
| 9136 |
+
"epoch": 0.6330760347656962,
|
| 9137 |
+
"grad_norm": 0.014143045991659164,
|
| 9138 |
+
"learning_rate": 0.00023061931572109138,
|
| 9139 |
+
"loss": 0.03496836498379707,
|
| 9140 |
+
"step": 1302
|
| 9141 |
+
},
|
| 9142 |
+
{
|
| 9143 |
+
"epoch": 0.6335622682793411,
|
| 9144 |
+
"grad_norm": 0.013887948356568813,
|
| 9145 |
+
"learning_rate": 0.00023040277176266782,
|
| 9146 |
+
"loss": 0.03378382325172424,
|
| 9147 |
+
"step": 1303
|
| 9148 |
+
},
|
| 9149 |
+
{
|
| 9150 |
+
"epoch": 0.6340485017929861,
|
| 9151 |
+
"grad_norm": 0.014969154261052608,
|
| 9152 |
+
"learning_rate": 0.00023018622780424425,
|
| 9153 |
+
"loss": 0.03367835655808449,
|
| 9154 |
+
"step": 1304
|
| 9155 |
+
},
|
| 9156 |
+
{
|
| 9157 |
+
"epoch": 0.634534735306631,
|
| 9158 |
+
"grad_norm": 0.014439153485000134,
|
| 9159 |
+
"learning_rate": 0.00022996968384582072,
|
| 9160 |
+
"loss": 0.036455463618040085,
|
| 9161 |
+
"step": 1305
|
| 9162 |
+
},
|
| 9163 |
+
{
|
| 9164 |
+
"epoch": 0.6350209688202759,
|
| 9165 |
+
"grad_norm": 0.014308437705039978,
|
| 9166 |
+
"learning_rate": 0.00022975313988739715,
|
| 9167 |
+
"loss": 0.03467040881514549,
|
| 9168 |
+
"step": 1306
|
| 9169 |
+
},
|
| 9170 |
+
{
|
| 9171 |
+
"epoch": 0.6355072023339209,
|
| 9172 |
+
"grad_norm": 0.01413166057318449,
|
| 9173 |
+
"learning_rate": 0.0002295365959289736,
|
| 9174 |
+
"loss": 0.03561070188879967,
|
| 9175 |
+
"step": 1307
|
| 9176 |
+
},
|
| 9177 |
+
{
|
| 9178 |
+
"epoch": 0.6359934358475658,
|
| 9179 |
+
"grad_norm": 0.013942469842731953,
|
| 9180 |
+
"learning_rate": 0.00022932005197055002,
|
| 9181 |
+
"loss": 0.0382116436958313,
|
| 9182 |
+
"step": 1308
|
| 9183 |
+
},
|
| 9184 |
+
{
|
| 9185 |
+
"epoch": 0.6364796693612107,
|
| 9186 |
+
"grad_norm": 0.014691284857690334,
|
| 9187 |
+
"learning_rate": 0.00022910350801212646,
|
| 9188 |
+
"loss": 0.03597237169742584,
|
| 9189 |
+
"step": 1309
|
| 9190 |
+
},
|
| 9191 |
+
{
|
| 9192 |
+
"epoch": 0.6369659028748557,
|
| 9193 |
+
"grad_norm": 0.013743284158408642,
|
| 9194 |
+
"learning_rate": 0.00022888696405370292,
|
| 9195 |
+
"loss": 0.03499186411499977,
|
| 9196 |
+
"step": 1310
|
| 9197 |
+
},
|
| 9198 |
+
{
|
| 9199 |
+
"epoch": 0.6374521363885006,
|
| 9200 |
+
"grad_norm": 0.013684669509530067,
|
| 9201 |
+
"learning_rate": 0.00022867042009527935,
|
| 9202 |
+
"loss": 0.037170108407735825,
|
| 9203 |
+
"step": 1311
|
| 9204 |
+
},
|
| 9205 |
+
{
|
| 9206 |
+
"epoch": 0.6379383699021455,
|
| 9207 |
+
"grad_norm": 0.01366688683629036,
|
| 9208 |
+
"learning_rate": 0.0002284538761368558,
|
| 9209 |
+
"loss": 0.035437971353530884,
|
| 9210 |
+
"step": 1312
|
| 9211 |
+
},
|
| 9212 |
+
{
|
| 9213 |
+
"epoch": 0.6384246034157904,
|
| 9214 |
+
"grad_norm": 0.014581129886209965,
|
| 9215 |
+
"learning_rate": 0.00022823733217843223,
|
| 9216 |
+
"loss": 0.03337923437356949,
|
| 9217 |
+
"step": 1313
|
| 9218 |
+
},
|
| 9219 |
+
{
|
| 9220 |
+
"epoch": 0.6389108369294354,
|
| 9221 |
+
"grad_norm": 0.013359768316149712,
|
| 9222 |
+
"learning_rate": 0.00022802078822000866,
|
| 9223 |
+
"loss": 0.032357357442379,
|
| 9224 |
+
"step": 1314
|
| 9225 |
+
},
|
| 9226 |
+
{
|
| 9227 |
+
"epoch": 0.6393970704430803,
|
| 9228 |
+
"grad_norm": 0.014213671907782555,
|
| 9229 |
+
"learning_rate": 0.0002278042442615851,
|
| 9230 |
+
"loss": 0.03423149138689041,
|
| 9231 |
+
"step": 1315
|
| 9232 |
+
},
|
| 9233 |
+
{
|
| 9234 |
+
"epoch": 0.6398833039567252,
|
| 9235 |
+
"grad_norm": 0.014577810652554035,
|
| 9236 |
+
"learning_rate": 0.00022758770030316156,
|
| 9237 |
+
"loss": 0.03380856662988663,
|
| 9238 |
+
"step": 1316
|
| 9239 |
+
},
|
| 9240 |
+
{
|
| 9241 |
+
"epoch": 0.6403695374703702,
|
| 9242 |
+
"grad_norm": 0.01257399097084999,
|
| 9243 |
+
"learning_rate": 0.000227371156344738,
|
| 9244 |
+
"loss": 0.03570931404829025,
|
| 9245 |
+
"step": 1317
|
| 9246 |
+
},
|
| 9247 |
+
{
|
| 9248 |
+
"epoch": 0.640855770984015,
|
| 9249 |
+
"grad_norm": 0.014166644774377346,
|
| 9250 |
+
"learning_rate": 0.00022715461238631443,
|
| 9251 |
+
"loss": 0.03312509506940842,
|
| 9252 |
+
"step": 1318
|
| 9253 |
+
},
|
| 9254 |
+
{
|
| 9255 |
+
"epoch": 0.64134200449766,
|
| 9256 |
+
"grad_norm": 0.013939057476818562,
|
| 9257 |
+
"learning_rate": 0.00022693806842789086,
|
| 9258 |
+
"loss": 0.034664127975702286,
|
| 9259 |
+
"step": 1319
|
| 9260 |
+
},
|
| 9261 |
+
{
|
| 9262 |
+
"epoch": 0.641828238011305,
|
| 9263 |
+
"grad_norm": 0.013967418111860752,
|
| 9264 |
+
"learning_rate": 0.0002267215244694673,
|
| 9265 |
+
"loss": 0.03376214951276779,
|
| 9266 |
+
"step": 1320
|
| 9267 |
+
},
|
| 9268 |
+
{
|
| 9269 |
+
"epoch": 0.6423144715249498,
|
| 9270 |
+
"grad_norm": 0.01414591632783413,
|
| 9271 |
+
"learning_rate": 0.00022650498051104373,
|
| 9272 |
+
"loss": 0.03443095460534096,
|
| 9273 |
+
"step": 1321
|
| 9274 |
+
},
|
| 9275 |
+
{
|
| 9276 |
+
"epoch": 0.6428007050385948,
|
| 9277 |
+
"grad_norm": 0.01348118856549263,
|
| 9278 |
+
"learning_rate": 0.0002262884365526202,
|
| 9279 |
+
"loss": 0.03514254838228226,
|
| 9280 |
+
"step": 1322
|
| 9281 |
+
},
|
| 9282 |
+
{
|
| 9283 |
+
"epoch": 0.6432869385522397,
|
| 9284 |
+
"grad_norm": 0.014731802977621555,
|
| 9285 |
+
"learning_rate": 0.00022607189259419663,
|
| 9286 |
+
"loss": 0.032584674656391144,
|
| 9287 |
+
"step": 1323
|
| 9288 |
+
},
|
| 9289 |
+
{
|
| 9290 |
+
"epoch": 0.6437731720658847,
|
| 9291 |
+
"grad_norm": 0.01589735597372055,
|
| 9292 |
+
"learning_rate": 0.00022585534863577307,
|
| 9293 |
+
"loss": 0.03588491305708885,
|
| 9294 |
+
"step": 1324
|
| 9295 |
+
},
|
| 9296 |
+
{
|
| 9297 |
+
"epoch": 0.6442594055795295,
|
| 9298 |
+
"grad_norm": 0.013818907551467419,
|
| 9299 |
+
"learning_rate": 0.0002256388046773495,
|
| 9300 |
+
"loss": 0.03400437533855438,
|
| 9301 |
+
"step": 1325
|
| 9302 |
+
},
|
| 9303 |
+
{
|
| 9304 |
+
"epoch": 0.6447456390931745,
|
| 9305 |
+
"grad_norm": 0.014179420657455921,
|
| 9306 |
+
"learning_rate": 0.00022542226071892594,
|
| 9307 |
+
"loss": 0.033515866845846176,
|
| 9308 |
+
"step": 1326
|
| 9309 |
+
},
|
| 9310 |
+
{
|
| 9311 |
+
"epoch": 0.6452318726068195,
|
| 9312 |
+
"grad_norm": 0.014600992202758789,
|
| 9313 |
+
"learning_rate": 0.0002252057167605024,
|
| 9314 |
+
"loss": 0.036238547414541245,
|
| 9315 |
+
"step": 1327
|
| 9316 |
+
},
|
| 9317 |
+
{
|
| 9318 |
+
"epoch": 0.6457181061204643,
|
| 9319 |
+
"grad_norm": 0.013533315621316433,
|
| 9320 |
+
"learning_rate": 0.00022498917280207884,
|
| 9321 |
+
"loss": 0.03574304282665253,
|
| 9322 |
+
"step": 1328
|
| 9323 |
+
},
|
| 9324 |
+
{
|
| 9325 |
+
"epoch": 0.6462043396341093,
|
| 9326 |
+
"grad_norm": 0.014975950121879578,
|
| 9327 |
+
"learning_rate": 0.00022477262884365527,
|
| 9328 |
+
"loss": 0.03604700788855553,
|
| 9329 |
+
"step": 1329
|
| 9330 |
+
},
|
| 9331 |
+
{
|
| 9332 |
+
"epoch": 0.6466905731477542,
|
| 9333 |
+
"grad_norm": 0.014431072399020195,
|
| 9334 |
+
"learning_rate": 0.0002245560848852317,
|
| 9335 |
+
"loss": 0.03493282198905945,
|
| 9336 |
+
"step": 1330
|
| 9337 |
+
},
|
| 9338 |
+
{
|
| 9339 |
+
"epoch": 0.6471768066613991,
|
| 9340 |
+
"grad_norm": 0.013177242130041122,
|
| 9341 |
+
"learning_rate": 0.00022433954092680814,
|
| 9342 |
+
"loss": 0.03511972725391388,
|
| 9343 |
+
"step": 1331
|
| 9344 |
+
},
|
| 9345 |
+
{
|
| 9346 |
+
"epoch": 0.6476630401750441,
|
| 9347 |
+
"grad_norm": 0.013923943042755127,
|
| 9348 |
+
"learning_rate": 0.00022412299696838458,
|
| 9349 |
+
"loss": 0.03779950365424156,
|
| 9350 |
+
"step": 1332
|
| 9351 |
+
},
|
| 9352 |
+
{
|
| 9353 |
+
"epoch": 0.648149273688689,
|
| 9354 |
+
"grad_norm": 0.014116371050477028,
|
| 9355 |
+
"learning_rate": 0.000223906453009961,
|
| 9356 |
+
"loss": 0.0330006442964077,
|
| 9357 |
+
"step": 1333
|
| 9358 |
+
},
|
| 9359 |
+
{
|
| 9360 |
+
"epoch": 0.6486355072023339,
|
| 9361 |
+
"grad_norm": 0.014254424721002579,
|
| 9362 |
+
"learning_rate": 0.00022368990905153747,
|
| 9363 |
+
"loss": 0.033680692315101624,
|
| 9364 |
+
"step": 1334
|
| 9365 |
+
},
|
| 9366 |
+
{
|
| 9367 |
+
"epoch": 0.6491217407159788,
|
| 9368 |
+
"grad_norm": 0.013696153648197651,
|
| 9369 |
+
"learning_rate": 0.0002234733650931139,
|
| 9370 |
+
"loss": 0.028675612062215805,
|
| 9371 |
+
"step": 1335
|
| 9372 |
+
},
|
| 9373 |
+
{
|
| 9374 |
+
"epoch": 0.6496079742296238,
|
| 9375 |
+
"grad_norm": 0.013288403861224651,
|
| 9376 |
+
"learning_rate": 0.00022325682113469035,
|
| 9377 |
+
"loss": 0.032354529947042465,
|
| 9378 |
+
"step": 1336
|
| 9379 |
+
},
|
| 9380 |
+
{
|
| 9381 |
+
"epoch": 0.6500942077432688,
|
| 9382 |
+
"grad_norm": 0.013923591002821922,
|
| 9383 |
+
"learning_rate": 0.00022304027717626678,
|
| 9384 |
+
"loss": 0.03523080423474312,
|
| 9385 |
+
"step": 1337
|
| 9386 |
+
},
|
| 9387 |
+
{
|
| 9388 |
+
"epoch": 0.6505804412569136,
|
| 9389 |
+
"grad_norm": 0.014102572575211525,
|
| 9390 |
+
"learning_rate": 0.00022282373321784322,
|
| 9391 |
+
"loss": 0.033881984651088715,
|
| 9392 |
+
"step": 1338
|
| 9393 |
+
},
|
| 9394 |
+
{
|
| 9395 |
+
"epoch": 0.6510666747705586,
|
| 9396 |
+
"grad_norm": 0.01488335058093071,
|
| 9397 |
+
"learning_rate": 0.00022260718925941968,
|
| 9398 |
+
"loss": 0.03664050251245499,
|
| 9399 |
+
"step": 1339
|
| 9400 |
+
},
|
| 9401 |
+
{
|
| 9402 |
+
"epoch": 0.6515529082842035,
|
| 9403 |
+
"grad_norm": 0.013973576948046684,
|
| 9404 |
+
"learning_rate": 0.0002223906453009961,
|
| 9405 |
+
"loss": 0.03544414043426514,
|
| 9406 |
+
"step": 1340
|
| 9407 |
+
},
|
| 9408 |
+
{
|
| 9409 |
+
"epoch": 0.6520391417978484,
|
| 9410 |
+
"grad_norm": 0.013649104163050652,
|
| 9411 |
+
"learning_rate": 0.00022217410134257255,
|
| 9412 |
+
"loss": 0.03692401200532913,
|
| 9413 |
+
"step": 1341
|
| 9414 |
+
},
|
| 9415 |
+
{
|
| 9416 |
+
"epoch": 0.6525253753114933,
|
| 9417 |
+
"grad_norm": 0.013818228617310524,
|
| 9418 |
+
"learning_rate": 0.00022195755738414898,
|
| 9419 |
+
"loss": 0.03479582071304321,
|
| 9420 |
+
"step": 1342
|
| 9421 |
+
},
|
| 9422 |
+
{
|
| 9423 |
+
"epoch": 0.6530116088251383,
|
| 9424 |
+
"grad_norm": 0.013529524207115173,
|
| 9425 |
+
"learning_rate": 0.00022174101342572542,
|
| 9426 |
+
"loss": 0.028599372133612633,
|
| 9427 |
+
"step": 1343
|
| 9428 |
+
},
|
| 9429 |
+
{
|
| 9430 |
+
"epoch": 0.6534978423387832,
|
| 9431 |
+
"grad_norm": 0.013865238055586815,
|
| 9432 |
+
"learning_rate": 0.00022152446946730188,
|
| 9433 |
+
"loss": 0.03413660079240799,
|
| 9434 |
+
"step": 1344
|
| 9435 |
+
},
|
| 9436 |
+
{
|
| 9437 |
+
"epoch": 0.6539840758524281,
|
| 9438 |
+
"grad_norm": 0.014652731828391552,
|
| 9439 |
+
"learning_rate": 0.0002213079255088783,
|
| 9440 |
+
"loss": 0.032668400555849075,
|
| 9441 |
+
"step": 1345
|
| 9442 |
+
},
|
| 9443 |
+
{
|
| 9444 |
+
"epoch": 0.6544703093660731,
|
| 9445 |
+
"grad_norm": 0.013993294909596443,
|
| 9446 |
+
"learning_rate": 0.00022109138155045475,
|
| 9447 |
+
"loss": 0.03704202175140381,
|
| 9448 |
+
"step": 1346
|
| 9449 |
+
},
|
| 9450 |
+
{
|
| 9451 |
+
"epoch": 0.6549565428797179,
|
| 9452 |
+
"grad_norm": 0.014167597517371178,
|
| 9453 |
+
"learning_rate": 0.0002208748375920312,
|
| 9454 |
+
"loss": 0.03735867515206337,
|
| 9455 |
+
"step": 1347
|
| 9456 |
+
},
|
| 9457 |
+
{
|
| 9458 |
+
"epoch": 0.6554427763933629,
|
| 9459 |
+
"grad_norm": 0.01378459669649601,
|
| 9460 |
+
"learning_rate": 0.00022065829363360762,
|
| 9461 |
+
"loss": 0.031136203557252884,
|
| 9462 |
+
"step": 1348
|
| 9463 |
+
},
|
| 9464 |
+
{
|
| 9465 |
+
"epoch": 0.6559290099070079,
|
| 9466 |
+
"grad_norm": 0.014771642163395882,
|
| 9467 |
+
"learning_rate": 0.00022044174967518409,
|
| 9468 |
+
"loss": 0.03363046050071716,
|
| 9469 |
+
"step": 1349
|
| 9470 |
+
},
|
| 9471 |
+
{
|
| 9472 |
+
"epoch": 0.6564152434206527,
|
| 9473 |
+
"grad_norm": 0.013304132968187332,
|
| 9474 |
+
"learning_rate": 0.0002202252057167605,
|
| 9475 |
+
"loss": 0.03345319628715515,
|
| 9476 |
+
"step": 1350
|
| 9477 |
+
},
|
| 9478 |
+
{
|
| 9479 |
+
"epoch": 0.6569014769342977,
|
| 9480 |
+
"grad_norm": 0.01402937900274992,
|
| 9481 |
+
"learning_rate": 0.00022000866175833696,
|
| 9482 |
+
"loss": 0.03416682034730911,
|
| 9483 |
+
"step": 1351
|
| 9484 |
+
},
|
| 9485 |
+
{
|
| 9486 |
+
"epoch": 0.6573877104479426,
|
| 9487 |
+
"grad_norm": 0.01340491883456707,
|
| 9488 |
+
"learning_rate": 0.0002197921177999134,
|
| 9489 |
+
"loss": 0.03456851467490196,
|
| 9490 |
+
"step": 1352
|
| 9491 |
+
},
|
| 9492 |
+
{
|
| 9493 |
+
"epoch": 0.6578739439615876,
|
| 9494 |
+
"grad_norm": 0.014131614938378334,
|
| 9495 |
+
"learning_rate": 0.00021957557384148983,
|
| 9496 |
+
"loss": 0.035455308854579926,
|
| 9497 |
+
"step": 1353
|
| 9498 |
+
},
|
| 9499 |
+
{
|
| 9500 |
+
"epoch": 0.6583601774752325,
|
| 9501 |
+
"grad_norm": 0.01379065215587616,
|
| 9502 |
+
"learning_rate": 0.00021935902988306626,
|
| 9503 |
+
"loss": 0.03642420470714569,
|
| 9504 |
+
"step": 1354
|
| 9505 |
+
},
|
| 9506 |
+
{
|
| 9507 |
+
"epoch": 0.6588464109888774,
|
| 9508 |
+
"grad_norm": 0.013627709820866585,
|
| 9509 |
+
"learning_rate": 0.0002191424859246427,
|
| 9510 |
+
"loss": 0.03162863850593567,
|
| 9511 |
+
"step": 1355
|
| 9512 |
+
},
|
| 9513 |
+
{
|
| 9514 |
+
"epoch": 0.6593326445025224,
|
| 9515 |
+
"grad_norm": 0.013729275204241276,
|
| 9516 |
+
"learning_rate": 0.00021892594196621916,
|
| 9517 |
+
"loss": 0.03523639217019081,
|
| 9518 |
+
"step": 1356
|
| 9519 |
+
},
|
| 9520 |
+
{
|
| 9521 |
+
"epoch": 0.6598188780161672,
|
| 9522 |
+
"grad_norm": 0.013532226905226707,
|
| 9523 |
+
"learning_rate": 0.00021870939800779557,
|
| 9524 |
+
"loss": 0.035823140293359756,
|
| 9525 |
+
"step": 1357
|
| 9526 |
+
},
|
| 9527 |
+
{
|
| 9528 |
+
"epoch": 0.6603051115298122,
|
| 9529 |
+
"grad_norm": 0.014047990553081036,
|
| 9530 |
+
"learning_rate": 0.00021849285404937203,
|
| 9531 |
+
"loss": 0.03442424535751343,
|
| 9532 |
+
"step": 1358
|
| 9533 |
+
},
|
| 9534 |
+
{
|
| 9535 |
+
"epoch": 0.6607913450434572,
|
| 9536 |
+
"grad_norm": 0.01441075000911951,
|
| 9537 |
+
"learning_rate": 0.00021827631009094847,
|
| 9538 |
+
"loss": 0.03327252343297005,
|
| 9539 |
+
"step": 1359
|
| 9540 |
+
},
|
| 9541 |
+
{
|
| 9542 |
+
"epoch": 0.661277578557102,
|
| 9543 |
+
"grad_norm": 0.013950358144938946,
|
| 9544 |
+
"learning_rate": 0.0002180597661325249,
|
| 9545 |
+
"loss": 0.03601973131299019,
|
| 9546 |
+
"step": 1360
|
| 9547 |
+
},
|
| 9548 |
+
{
|
| 9549 |
+
"epoch": 0.661763812070747,
|
| 9550 |
+
"grad_norm": 0.014807157218456268,
|
| 9551 |
+
"learning_rate": 0.00021784322217410136,
|
| 9552 |
+
"loss": 0.03429421782493591,
|
| 9553 |
+
"step": 1361
|
| 9554 |
+
},
|
| 9555 |
+
{
|
| 9556 |
+
"epoch": 0.6622500455843919,
|
| 9557 |
+
"grad_norm": 0.01571790501475334,
|
| 9558 |
+
"learning_rate": 0.00021762667821567777,
|
| 9559 |
+
"loss": 0.03395929932594299,
|
| 9560 |
+
"step": 1362
|
| 9561 |
+
},
|
| 9562 |
+
{
|
| 9563 |
+
"epoch": 0.6627362790980368,
|
| 9564 |
+
"grad_norm": 0.01437438651919365,
|
| 9565 |
+
"learning_rate": 0.00021741013425725423,
|
| 9566 |
+
"loss": 0.03499199450016022,
|
| 9567 |
+
"step": 1363
|
| 9568 |
+
},
|
| 9569 |
+
{
|
| 9570 |
+
"epoch": 0.6632225126116817,
|
| 9571 |
+
"grad_norm": 0.015335733070969582,
|
| 9572 |
+
"learning_rate": 0.00021719359029883067,
|
| 9573 |
+
"loss": 0.03179340064525604,
|
| 9574 |
+
"step": 1364
|
| 9575 |
+
},
|
| 9576 |
+
{
|
| 9577 |
+
"epoch": 0.6637087461253267,
|
| 9578 |
+
"grad_norm": 0.015314069576561451,
|
| 9579 |
+
"learning_rate": 0.0002169770463404071,
|
| 9580 |
+
"loss": 0.03426148369908333,
|
| 9581 |
+
"step": 1365
|
| 9582 |
+
},
|
| 9583 |
+
{
|
| 9584 |
+
"epoch": 0.6641949796389716,
|
| 9585 |
+
"grad_norm": 0.014002913609147072,
|
| 9586 |
+
"learning_rate": 0.00021676050238198357,
|
| 9587 |
+
"loss": 0.035752829164266586,
|
| 9588 |
+
"step": 1366
|
| 9589 |
+
},
|
| 9590 |
+
{
|
| 9591 |
+
"epoch": 0.6646812131526165,
|
| 9592 |
+
"grad_norm": 0.014220299199223518,
|
| 9593 |
+
"learning_rate": 0.00021654395842355997,
|
| 9594 |
+
"loss": 0.03339208662509918,
|
| 9595 |
+
"step": 1367
|
| 9596 |
+
},
|
| 9597 |
+
{
|
| 9598 |
+
"epoch": 0.6651674466662615,
|
| 9599 |
+
"grad_norm": 0.014346009120345116,
|
| 9600 |
+
"learning_rate": 0.00021632741446513644,
|
| 9601 |
+
"loss": 0.03166678920388222,
|
| 9602 |
+
"step": 1368
|
| 9603 |
+
},
|
| 9604 |
+
{
|
| 9605 |
+
"epoch": 0.6656536801799064,
|
| 9606 |
+
"grad_norm": 0.014936978928744793,
|
| 9607 |
+
"learning_rate": 0.00021611087050671287,
|
| 9608 |
+
"loss": 0.03593449667096138,
|
| 9609 |
+
"step": 1369
|
| 9610 |
+
},
|
| 9611 |
+
{
|
| 9612 |
+
"epoch": 0.6661399136935513,
|
| 9613 |
+
"grad_norm": 0.014874998480081558,
|
| 9614 |
+
"learning_rate": 0.0002158943265482893,
|
| 9615 |
+
"loss": 0.035512425005435944,
|
| 9616 |
+
"step": 1370
|
| 9617 |
+
},
|
| 9618 |
+
{
|
| 9619 |
+
"epoch": 0.6666261472071963,
|
| 9620 |
+
"grad_norm": 0.0144179193302989,
|
| 9621 |
+
"learning_rate": 0.00021567778258986577,
|
| 9622 |
+
"loss": 0.03279472514986992,
|
| 9623 |
+
"step": 1371
|
| 9624 |
+
},
|
| 9625 |
+
{
|
| 9626 |
+
"epoch": 0.6671123807208412,
|
| 9627 |
+
"grad_norm": 0.014313534833490849,
|
| 9628 |
+
"learning_rate": 0.00021546123863144218,
|
| 9629 |
+
"loss": 0.03635512664914131,
|
| 9630 |
+
"step": 1372
|
| 9631 |
+
},
|
| 9632 |
+
{
|
| 9633 |
+
"epoch": 0.6675986142344861,
|
| 9634 |
+
"grad_norm": 0.014254770241677761,
|
| 9635 |
+
"learning_rate": 0.00021524469467301864,
|
| 9636 |
+
"loss": 0.03417997062206268,
|
| 9637 |
+
"step": 1373
|
| 9638 |
+
},
|
| 9639 |
+
{
|
| 9640 |
+
"epoch": 0.668084847748131,
|
| 9641 |
+
"grad_norm": 0.013667783699929714,
|
| 9642 |
+
"learning_rate": 0.00021502815071459505,
|
| 9643 |
+
"loss": 0.03116200864315033,
|
| 9644 |
+
"step": 1374
|
| 9645 |
+
},
|
| 9646 |
+
{
|
| 9647 |
+
"epoch": 0.668571081261776,
|
| 9648 |
+
"grad_norm": 0.014955972321331501,
|
| 9649 |
+
"learning_rate": 0.0002148116067561715,
|
| 9650 |
+
"loss": 0.03064017742872238,
|
| 9651 |
+
"step": 1375
|
| 9652 |
+
},
|
| 9653 |
+
{
|
| 9654 |
+
"epoch": 0.6690573147754209,
|
| 9655 |
+
"grad_norm": 0.014537026174366474,
|
| 9656 |
+
"learning_rate": 0.00021459506279774795,
|
| 9657 |
+
"loss": 0.03849161043763161,
|
| 9658 |
+
"step": 1376
|
| 9659 |
+
},
|
| 9660 |
+
{
|
| 9661 |
+
"epoch": 0.6695435482890658,
|
| 9662 |
+
"grad_norm": 0.013870895840227604,
|
| 9663 |
+
"learning_rate": 0.00021437851883932438,
|
| 9664 |
+
"loss": 0.030646931380033493,
|
| 9665 |
+
"step": 1377
|
| 9666 |
+
},
|
| 9667 |
+
{
|
| 9668 |
+
"epoch": 0.6700297818027108,
|
| 9669 |
+
"grad_norm": 0.013801373541355133,
|
| 9670 |
+
"learning_rate": 0.00021416197488090084,
|
| 9671 |
+
"loss": 0.03465121239423752,
|
| 9672 |
+
"step": 1378
|
| 9673 |
+
},
|
| 9674 |
+
{
|
| 9675 |
+
"epoch": 0.6705160153163556,
|
| 9676 |
+
"grad_norm": 0.013270103372633457,
|
| 9677 |
+
"learning_rate": 0.00021394543092247725,
|
| 9678 |
+
"loss": 0.033292222768068314,
|
| 9679 |
+
"step": 1379
|
| 9680 |
+
},
|
| 9681 |
+
{
|
| 9682 |
+
"epoch": 0.6710022488300006,
|
| 9683 |
+
"grad_norm": 0.01431484054774046,
|
| 9684 |
+
"learning_rate": 0.00021372888696405371,
|
| 9685 |
+
"loss": 0.03248829022049904,
|
| 9686 |
+
"step": 1380
|
| 9687 |
+
},
|
| 9688 |
+
{
|
| 9689 |
+
"epoch": 0.6714884823436456,
|
| 9690 |
+
"grad_norm": 0.013586247339844704,
|
| 9691 |
+
"learning_rate": 0.00021351234300563015,
|
| 9692 |
+
"loss": 0.036313146352767944,
|
| 9693 |
+
"step": 1381
|
| 9694 |
+
},
|
| 9695 |
+
{
|
| 9696 |
+
"epoch": 0.6719747158572905,
|
| 9697 |
+
"grad_norm": 0.013846389949321747,
|
| 9698 |
+
"learning_rate": 0.00021329579904720659,
|
| 9699 |
+
"loss": 0.03729093447327614,
|
| 9700 |
+
"step": 1382
|
| 9701 |
+
},
|
| 9702 |
+
{
|
| 9703 |
+
"epoch": 0.6724609493709354,
|
| 9704 |
+
"grad_norm": 0.014342756010591984,
|
| 9705 |
+
"learning_rate": 0.00021307925508878305,
|
| 9706 |
+
"loss": 0.03507140278816223,
|
| 9707 |
+
"step": 1383
|
| 9708 |
+
},
|
| 9709 |
+
{
|
| 9710 |
+
"epoch": 0.6729471828845803,
|
| 9711 |
+
"grad_norm": 0.015145371668040752,
|
| 9712 |
+
"learning_rate": 0.00021286271113035946,
|
| 9713 |
+
"loss": 0.0366177037358284,
|
| 9714 |
+
"step": 1384
|
| 9715 |
+
},
|
| 9716 |
+
{
|
| 9717 |
+
"epoch": 0.6734334163982253,
|
| 9718 |
+
"grad_norm": 0.013105987571179867,
|
| 9719 |
+
"learning_rate": 0.00021264616717193592,
|
| 9720 |
+
"loss": 0.03000234067440033,
|
| 9721 |
+
"step": 1385
|
| 9722 |
+
},
|
| 9723 |
+
{
|
| 9724 |
+
"epoch": 0.6739196499118701,
|
| 9725 |
+
"grad_norm": 0.014528613537549973,
|
| 9726 |
+
"learning_rate": 0.00021242962321351233,
|
| 9727 |
+
"loss": 0.03557267785072327,
|
| 9728 |
+
"step": 1386
|
| 9729 |
+
},
|
| 9730 |
+
{
|
| 9731 |
+
"epoch": 0.6744058834255151,
|
| 9732 |
+
"grad_norm": 0.014234558679163456,
|
| 9733 |
+
"learning_rate": 0.0002122130792550888,
|
| 9734 |
+
"loss": 0.035209693014621735,
|
| 9735 |
+
"step": 1387
|
| 9736 |
+
},
|
| 9737 |
+
{
|
| 9738 |
+
"epoch": 0.6748921169391601,
|
| 9739 |
+
"grad_norm": 0.015100352466106415,
|
| 9740 |
+
"learning_rate": 0.00021199653529666525,
|
| 9741 |
+
"loss": 0.03626418113708496,
|
| 9742 |
+
"step": 1388
|
| 9743 |
+
},
|
| 9744 |
+
{
|
| 9745 |
+
"epoch": 0.6753783504528049,
|
| 9746 |
+
"grad_norm": 0.014857640489935875,
|
| 9747 |
+
"learning_rate": 0.00021177999133824166,
|
| 9748 |
+
"loss": 0.03384890779852867,
|
| 9749 |
+
"step": 1389
|
| 9750 |
+
},
|
| 9751 |
+
{
|
| 9752 |
+
"epoch": 0.6758645839664499,
|
| 9753 |
+
"grad_norm": 0.014490040950477123,
|
| 9754 |
+
"learning_rate": 0.00021156344737981812,
|
| 9755 |
+
"loss": 0.031536661088466644,
|
| 9756 |
+
"step": 1390
|
| 9757 |
+
},
|
| 9758 |
+
{
|
| 9759 |
+
"epoch": 0.6763508174800948,
|
| 9760 |
+
"grad_norm": 0.014538592658936977,
|
| 9761 |
+
"learning_rate": 0.00021134690342139453,
|
| 9762 |
+
"loss": 0.03946738690137863,
|
| 9763 |
+
"step": 1391
|
| 9764 |
+
},
|
| 9765 |
+
{
|
| 9766 |
+
"epoch": 0.6768370509937397,
|
| 9767 |
+
"grad_norm": 0.013940773904323578,
|
| 9768 |
+
"learning_rate": 0.000211130359462971,
|
| 9769 |
+
"loss": 0.03681372106075287,
|
| 9770 |
+
"step": 1392
|
| 9771 |
+
},
|
| 9772 |
+
{
|
| 9773 |
+
"epoch": 0.6773232845073847,
|
| 9774 |
+
"grad_norm": 0.013929001055657864,
|
| 9775 |
+
"learning_rate": 0.00021091381550454743,
|
| 9776 |
+
"loss": 0.036155663430690765,
|
| 9777 |
+
"step": 1393
|
| 9778 |
+
},
|
| 9779 |
+
{
|
| 9780 |
+
"epoch": 0.6778095180210296,
|
| 9781 |
+
"grad_norm": 0.01500175055116415,
|
| 9782 |
+
"learning_rate": 0.00021069727154612386,
|
| 9783 |
+
"loss": 0.03327430784702301,
|
| 9784 |
+
"step": 1394
|
| 9785 |
+
},
|
| 9786 |
+
{
|
| 9787 |
+
"epoch": 0.6782957515346745,
|
| 9788 |
+
"grad_norm": 0.014348408207297325,
|
| 9789 |
+
"learning_rate": 0.00021048072758770033,
|
| 9790 |
+
"loss": 0.03527160733938217,
|
| 9791 |
+
"step": 1395
|
| 9792 |
+
},
|
| 9793 |
+
{
|
| 9794 |
+
"epoch": 0.6787819850483194,
|
| 9795 |
+
"grad_norm": 0.014366855844855309,
|
| 9796 |
+
"learning_rate": 0.00021026418362927673,
|
| 9797 |
+
"loss": 0.03398943692445755,
|
| 9798 |
+
"step": 1396
|
| 9799 |
+
},
|
| 9800 |
+
{
|
| 9801 |
+
"epoch": 0.6792682185619644,
|
| 9802 |
+
"grad_norm": 0.013865748420357704,
|
| 9803 |
+
"learning_rate": 0.0002100476396708532,
|
| 9804 |
+
"loss": 0.03580952808260918,
|
| 9805 |
+
"step": 1397
|
| 9806 |
+
},
|
| 9807 |
+
{
|
| 9808 |
+
"epoch": 0.6797544520756094,
|
| 9809 |
+
"grad_norm": 0.014454754069447517,
|
| 9810 |
+
"learning_rate": 0.0002098310957124296,
|
| 9811 |
+
"loss": 0.0344514362514019,
|
| 9812 |
+
"step": 1398
|
| 9813 |
+
},
|
| 9814 |
+
{
|
| 9815 |
+
"epoch": 0.6802406855892542,
|
| 9816 |
+
"grad_norm": 0.013905288651585579,
|
| 9817 |
+
"learning_rate": 0.00020961455175400607,
|
| 9818 |
+
"loss": 0.03444414958357811,
|
| 9819 |
+
"step": 1399
|
| 9820 |
+
},
|
| 9821 |
+
{
|
| 9822 |
+
"epoch": 0.6807269191028992,
|
| 9823 |
+
"grad_norm": 0.01423436775803566,
|
| 9824 |
+
"learning_rate": 0.00020939800779558253,
|
| 9825 |
+
"loss": 0.03654969856142998,
|
| 9826 |
+
"step": 1400
|
| 9827 |
+
},
|
| 9828 |
+
{
|
| 9829 |
+
"epoch": 0.6812131526165441,
|
| 9830 |
+
"grad_norm": 0.01422138512134552,
|
| 9831 |
+
"learning_rate": 0.00020918146383715894,
|
| 9832 |
+
"loss": 0.03397469222545624,
|
| 9833 |
+
"step": 1401
|
| 9834 |
+
},
|
| 9835 |
+
{
|
| 9836 |
+
"epoch": 0.681699386130189,
|
| 9837 |
+
"grad_norm": 0.014388046227395535,
|
| 9838 |
+
"learning_rate": 0.0002089649198787354,
|
| 9839 |
+
"loss": 0.036142874509096146,
|
| 9840 |
+
"step": 1402
|
| 9841 |
+
},
|
| 9842 |
+
{
|
| 9843 |
+
"epoch": 0.682185619643834,
|
| 9844 |
+
"grad_norm": 0.014639279805123806,
|
| 9845 |
+
"learning_rate": 0.0002087483759203118,
|
| 9846 |
+
"loss": 0.029921941459178925,
|
| 9847 |
+
"step": 1403
|
| 9848 |
+
},
|
| 9849 |
+
{
|
| 9850 |
+
"epoch": 0.6826718531574789,
|
| 9851 |
+
"grad_norm": 0.014388429932296276,
|
| 9852 |
+
"learning_rate": 0.00020853183196188827,
|
| 9853 |
+
"loss": 0.03260527178645134,
|
| 9854 |
+
"step": 1404
|
| 9855 |
+
},
|
| 9856 |
+
{
|
| 9857 |
+
"epoch": 0.6831580866711238,
|
| 9858 |
+
"grad_norm": 0.016170000657439232,
|
| 9859 |
+
"learning_rate": 0.00020831528800346473,
|
| 9860 |
+
"loss": 0.029670415446162224,
|
| 9861 |
+
"step": 1405
|
| 9862 |
+
},
|
| 9863 |
+
{
|
| 9864 |
+
"epoch": 0.6836443201847687,
|
| 9865 |
+
"grad_norm": 0.013937766663730145,
|
| 9866 |
+
"learning_rate": 0.00020809874404504114,
|
| 9867 |
+
"loss": 0.036622799932956696,
|
| 9868 |
+
"step": 1406
|
| 9869 |
+
},
|
| 9870 |
+
{
|
| 9871 |
+
"epoch": 0.6841305536984137,
|
| 9872 |
+
"grad_norm": 0.014217172749340534,
|
| 9873 |
+
"learning_rate": 0.0002078822000866176,
|
| 9874 |
+
"loss": 0.034993477165699005,
|
| 9875 |
+
"step": 1407
|
| 9876 |
+
},
|
| 9877 |
+
{
|
| 9878 |
+
"epoch": 0.6846167872120585,
|
| 9879 |
+
"grad_norm": 0.013956526294350624,
|
| 9880 |
+
"learning_rate": 0.000207665656128194,
|
| 9881 |
+
"loss": 0.03437182679772377,
|
| 9882 |
+
"step": 1408
|
| 9883 |
+
},
|
| 9884 |
+
{
|
| 9885 |
+
"epoch": 0.6851030207257035,
|
| 9886 |
+
"grad_norm": 0.013690595515072346,
|
| 9887 |
+
"learning_rate": 0.00020744911216977047,
|
| 9888 |
+
"loss": 0.03555164113640785,
|
| 9889 |
+
"step": 1409
|
| 9890 |
+
},
|
| 9891 |
+
{
|
| 9892 |
+
"epoch": 0.6855892542393485,
|
| 9893 |
+
"grad_norm": 0.014545619487762451,
|
| 9894 |
+
"learning_rate": 0.00020723256821134688,
|
| 9895 |
+
"loss": 0.030868198722600937,
|
| 9896 |
+
"step": 1410
|
| 9897 |
+
},
|
| 9898 |
+
{
|
| 9899 |
+
"epoch": 0.6860754877529934,
|
| 9900 |
+
"grad_norm": 0.014318596571683884,
|
| 9901 |
+
"learning_rate": 0.00020701602425292334,
|
| 9902 |
+
"loss": 0.032924726605415344,
|
| 9903 |
+
"step": 1411
|
| 9904 |
+
},
|
| 9905 |
+
{
|
| 9906 |
+
"epoch": 0.6865617212666383,
|
| 9907 |
+
"grad_norm": 0.013988622464239597,
|
| 9908 |
+
"learning_rate": 0.0002067994802944998,
|
| 9909 |
+
"loss": 0.0356634259223938,
|
| 9910 |
+
"step": 1412
|
| 9911 |
+
},
|
| 9912 |
+
{
|
| 9913 |
+
"epoch": 0.6870479547802832,
|
| 9914 |
+
"grad_norm": 0.014085984788835049,
|
| 9915 |
+
"learning_rate": 0.00020658293633607621,
|
| 9916 |
+
"loss": 0.03102702833712101,
|
| 9917 |
+
"step": 1413
|
| 9918 |
+
},
|
| 9919 |
+
{
|
| 9920 |
+
"epoch": 0.6875341882939282,
|
| 9921 |
+
"grad_norm": 0.014302864670753479,
|
| 9922 |
+
"learning_rate": 0.00020636639237765268,
|
| 9923 |
+
"loss": 0.03341764956712723,
|
| 9924 |
+
"step": 1414
|
| 9925 |
+
},
|
| 9926 |
+
{
|
| 9927 |
+
"epoch": 0.6880204218075731,
|
| 9928 |
+
"grad_norm": 0.013568436726927757,
|
| 9929 |
+
"learning_rate": 0.00020614984841922908,
|
| 9930 |
+
"loss": 0.03309115767478943,
|
| 9931 |
+
"step": 1415
|
| 9932 |
+
},
|
| 9933 |
+
{
|
| 9934 |
+
"epoch": 0.688506655321218,
|
| 9935 |
+
"grad_norm": 0.014550761319696903,
|
| 9936 |
+
"learning_rate": 0.00020593330446080555,
|
| 9937 |
+
"loss": 0.03668435662984848,
|
| 9938 |
+
"step": 1416
|
| 9939 |
+
},
|
| 9940 |
+
{
|
| 9941 |
+
"epoch": 0.688992888834863,
|
| 9942 |
+
"grad_norm": 0.014469129964709282,
|
| 9943 |
+
"learning_rate": 0.000205716760502382,
|
| 9944 |
+
"loss": 0.03193166106939316,
|
| 9945 |
+
"step": 1417
|
| 9946 |
+
},
|
| 9947 |
+
{
|
| 9948 |
+
"epoch": 0.6894791223485078,
|
| 9949 |
+
"grad_norm": 0.014508143998682499,
|
| 9950 |
+
"learning_rate": 0.00020550021654395842,
|
| 9951 |
+
"loss": 0.030632685869932175,
|
| 9952 |
+
"step": 1418
|
| 9953 |
+
},
|
| 9954 |
+
{
|
| 9955 |
+
"epoch": 0.6899653558621528,
|
| 9956 |
+
"grad_norm": 0.014048706740140915,
|
| 9957 |
+
"learning_rate": 0.00020528367258553488,
|
| 9958 |
+
"loss": 0.03671274334192276,
|
| 9959 |
+
"step": 1419
|
| 9960 |
+
},
|
| 9961 |
+
{
|
| 9962 |
+
"epoch": 0.6904515893757978,
|
| 9963 |
+
"grad_norm": 0.01405914407223463,
|
| 9964 |
+
"learning_rate": 0.0002050671286271113,
|
| 9965 |
+
"loss": 0.03378705307841301,
|
| 9966 |
+
"step": 1420
|
| 9967 |
+
},
|
| 9968 |
+
{
|
| 9969 |
+
"epoch": 0.6909378228894426,
|
| 9970 |
+
"grad_norm": 0.014757314696907997,
|
| 9971 |
+
"learning_rate": 0.00020485058466868775,
|
| 9972 |
+
"loss": 0.036013372242450714,
|
| 9973 |
+
"step": 1421
|
| 9974 |
+
},
|
| 9975 |
+
{
|
| 9976 |
+
"epoch": 0.6914240564030876,
|
| 9977 |
+
"grad_norm": 0.013872887939214706,
|
| 9978 |
+
"learning_rate": 0.00020463404071026419,
|
| 9979 |
+
"loss": 0.03474874421954155,
|
| 9980 |
+
"step": 1422
|
| 9981 |
+
},
|
| 9982 |
+
{
|
| 9983 |
+
"epoch": 0.6914240564030876,
|
| 9984 |
+
"eval_loss": 0.03418760001659393,
|
| 9985 |
+
"eval_runtime": 162.5503,
|
| 9986 |
+
"eval_samples_per_second": 22.104,
|
| 9987 |
+
"eval_steps_per_second": 5.531,
|
| 9988 |
+
"step": 1422
|
| 9989 |
}
|
| 9990 |
],
|
| 9991 |
"logging_steps": 1,
|
|
|
|
| 10005 |
"attributes": {}
|
| 10006 |
}
|
| 10007 |
},
|
| 10008 |
+
"total_flos": 3.971522304451891e+17,
|
| 10009 |
"train_batch_size": 4,
|
| 10010 |
"trial_name": null,
|
| 10011 |
"trial_params": null
|