Training in progress, step 14168, checkpoint
Browse files
last-checkpoint/model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 500389884
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8de22d450769d69a96c63c76531a3fb7e01f6423c491441830387e9cb7ed916e
|
| 3 |
size 500389884
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1000900218
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f43ce093c117315e7acfb01d8346e842f77cf100d65506fdd1c745797e57ff5c
|
| 3 |
size 1000900218
|
last-checkpoint/rng_state_0.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14512
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e9faa9a63eb45764eaaacb970ffffa47beb58be3388211b399565fed7ba6f8ef
|
| 3 |
size 14512
|
last-checkpoint/rng_state_1.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14512
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b1c08e7ac0c6c146323ce2c3786f8affd631063335ef26e898e86831cecaafa1
|
| 3 |
size 14512
|
last-checkpoint/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1064
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a25ecdb576c7c60f2386731ab1defd8898d170e70fb4eb9008331e7fc07ff9c3
|
| 3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch": 0.
|
| 5 |
"eval_steps": 500,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -21259,6 +21259,3548 @@
|
|
| 21259 |
"learning_rate": 2.0003458668906565e-05,
|
| 21260 |
"loss": 5.5015,
|
| 21261 |
"step": 12144
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 21262 |
}
|
| 21263 |
],
|
| 21264 |
"logging_steps": 4,
|
|
@@ -21266,7 +24808,7 @@
|
|
| 21266 |
"num_input_tokens_seen": 0,
|
| 21267 |
"num_train_epochs": 1,
|
| 21268 |
"save_steps": 2024,
|
| 21269 |
-
"total_flos": 5.
|
| 21270 |
"train_batch_size": 8,
|
| 21271 |
"trial_name": null,
|
| 21272 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.7000345866890657,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 14168,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 21259 |
"learning_rate": 2.0003458668906565e-05,
|
| 21260 |
"loss": 5.5015,
|
| 21261 |
"step": 12144
|
| 21262 |
+
},
|
| 21263 |
+
{
|
| 21264 |
+
"epoch": 0.6,
|
| 21265 |
+
"grad_norm": 1.796938180923462,
|
| 21266 |
+
"learning_rate": 1.999357675774495e-05,
|
| 21267 |
+
"loss": 5.4519,
|
| 21268 |
+
"step": 12148
|
| 21269 |
+
},
|
| 21270 |
+
{
|
| 21271 |
+
"epoch": 0.6,
|
| 21272 |
+
"grad_norm": 2.2093305587768555,
|
| 21273 |
+
"learning_rate": 1.9983694846583332e-05,
|
| 21274 |
+
"loss": 5.6164,
|
| 21275 |
+
"step": 12152
|
| 21276 |
+
},
|
| 21277 |
+
{
|
| 21278 |
+
"epoch": 0.6,
|
| 21279 |
+
"grad_norm": 1.8350774049758911,
|
| 21280 |
+
"learning_rate": 1.997381293542171e-05,
|
| 21281 |
+
"loss": 5.583,
|
| 21282 |
+
"step": 12156
|
| 21283 |
+
},
|
| 21284 |
+
{
|
| 21285 |
+
"epoch": 0.6,
|
| 21286 |
+
"grad_norm": 1.8653216361999512,
|
| 21287 |
+
"learning_rate": 1.9963931024260093e-05,
|
| 21288 |
+
"loss": 5.656,
|
| 21289 |
+
"step": 12160
|
| 21290 |
+
},
|
| 21291 |
+
{
|
| 21292 |
+
"epoch": 0.6,
|
| 21293 |
+
"grad_norm": 1.7206474542617798,
|
| 21294 |
+
"learning_rate": 1.9954049113098475e-05,
|
| 21295 |
+
"loss": 5.5226,
|
| 21296 |
+
"step": 12164
|
| 21297 |
+
},
|
| 21298 |
+
{
|
| 21299 |
+
"epoch": 0.6,
|
| 21300 |
+
"grad_norm": 1.8878382444381714,
|
| 21301 |
+
"learning_rate": 1.9944167201936857e-05,
|
| 21302 |
+
"loss": 5.6363,
|
| 21303 |
+
"step": 12168
|
| 21304 |
+
},
|
| 21305 |
+
{
|
| 21306 |
+
"epoch": 0.6,
|
| 21307 |
+
"grad_norm": 2.0668506622314453,
|
| 21308 |
+
"learning_rate": 1.9934285290775236e-05,
|
| 21309 |
+
"loss": 5.5106,
|
| 21310 |
+
"step": 12172
|
| 21311 |
+
},
|
| 21312 |
+
{
|
| 21313 |
+
"epoch": 0.6,
|
| 21314 |
+
"grad_norm": 2.0027477741241455,
|
| 21315 |
+
"learning_rate": 1.9924403379613618e-05,
|
| 21316 |
+
"loss": 5.5217,
|
| 21317 |
+
"step": 12176
|
| 21318 |
+
},
|
| 21319 |
+
{
|
| 21320 |
+
"epoch": 0.6,
|
| 21321 |
+
"grad_norm": 2.1123950481414795,
|
| 21322 |
+
"learning_rate": 1.9914521468452e-05,
|
| 21323 |
+
"loss": 5.6275,
|
| 21324 |
+
"step": 12180
|
| 21325 |
+
},
|
| 21326 |
+
{
|
| 21327 |
+
"epoch": 0.6,
|
| 21328 |
+
"grad_norm": 2.1055514812469482,
|
| 21329 |
+
"learning_rate": 1.9904639557290382e-05,
|
| 21330 |
+
"loss": 5.6186,
|
| 21331 |
+
"step": 12184
|
| 21332 |
+
},
|
| 21333 |
+
{
|
| 21334 |
+
"epoch": 0.6,
|
| 21335 |
+
"grad_norm": 2.0140175819396973,
|
| 21336 |
+
"learning_rate": 1.989475764612876e-05,
|
| 21337 |
+
"loss": 5.5965,
|
| 21338 |
+
"step": 12188
|
| 21339 |
+
},
|
| 21340 |
+
{
|
| 21341 |
+
"epoch": 0.6,
|
| 21342 |
+
"grad_norm": 1.902254343032837,
|
| 21343 |
+
"learning_rate": 1.9884875734967142e-05,
|
| 21344 |
+
"loss": 5.5557,
|
| 21345 |
+
"step": 12192
|
| 21346 |
+
},
|
| 21347 |
+
{
|
| 21348 |
+
"epoch": 0.6,
|
| 21349 |
+
"grad_norm": 1.9510375261306763,
|
| 21350 |
+
"learning_rate": 1.9874993823805525e-05,
|
| 21351 |
+
"loss": 5.4804,
|
| 21352 |
+
"step": 12196
|
| 21353 |
+
},
|
| 21354 |
+
{
|
| 21355 |
+
"epoch": 0.6,
|
| 21356 |
+
"grad_norm": 1.950716495513916,
|
| 21357 |
+
"learning_rate": 1.9865111912643907e-05,
|
| 21358 |
+
"loss": 5.5887,
|
| 21359 |
+
"step": 12200
|
| 21360 |
+
},
|
| 21361 |
+
{
|
| 21362 |
+
"epoch": 0.6,
|
| 21363 |
+
"grad_norm": 2.0936696529388428,
|
| 21364 |
+
"learning_rate": 1.985523000148229e-05,
|
| 21365 |
+
"loss": 5.5463,
|
| 21366 |
+
"step": 12204
|
| 21367 |
+
},
|
| 21368 |
+
{
|
| 21369 |
+
"epoch": 0.6,
|
| 21370 |
+
"grad_norm": 1.6361807584762573,
|
| 21371 |
+
"learning_rate": 1.984534809032067e-05,
|
| 21372 |
+
"loss": 5.5516,
|
| 21373 |
+
"step": 12208
|
| 21374 |
+
},
|
| 21375 |
+
{
|
| 21376 |
+
"epoch": 0.6,
|
| 21377 |
+
"grad_norm": 2.1499600410461426,
|
| 21378 |
+
"learning_rate": 1.9835466179159053e-05,
|
| 21379 |
+
"loss": 5.5467,
|
| 21380 |
+
"step": 12212
|
| 21381 |
+
},
|
| 21382 |
+
{
|
| 21383 |
+
"epoch": 0.6,
|
| 21384 |
+
"grad_norm": 1.9551507234573364,
|
| 21385 |
+
"learning_rate": 1.982558426799743e-05,
|
| 21386 |
+
"loss": 5.549,
|
| 21387 |
+
"step": 12216
|
| 21388 |
+
},
|
| 21389 |
+
{
|
| 21390 |
+
"epoch": 0.6,
|
| 21391 |
+
"grad_norm": 1.9426565170288086,
|
| 21392 |
+
"learning_rate": 1.9815702356835814e-05,
|
| 21393 |
+
"loss": 5.5726,
|
| 21394 |
+
"step": 12220
|
| 21395 |
+
},
|
| 21396 |
+
{
|
| 21397 |
+
"epoch": 0.6,
|
| 21398 |
+
"grad_norm": 1.9303926229476929,
|
| 21399 |
+
"learning_rate": 1.9805820445674196e-05,
|
| 21400 |
+
"loss": 5.5571,
|
| 21401 |
+
"step": 12224
|
| 21402 |
+
},
|
| 21403 |
+
{
|
| 21404 |
+
"epoch": 0.6,
|
| 21405 |
+
"grad_norm": 1.859390139579773,
|
| 21406 |
+
"learning_rate": 1.9795938534512574e-05,
|
| 21407 |
+
"loss": 5.499,
|
| 21408 |
+
"step": 12228
|
| 21409 |
+
},
|
| 21410 |
+
{
|
| 21411 |
+
"epoch": 0.6,
|
| 21412 |
+
"grad_norm": 1.8791084289550781,
|
| 21413 |
+
"learning_rate": 1.9786056623350956e-05,
|
| 21414 |
+
"loss": 5.5617,
|
| 21415 |
+
"step": 12232
|
| 21416 |
+
},
|
| 21417 |
+
{
|
| 21418 |
+
"epoch": 0.6,
|
| 21419 |
+
"grad_norm": 2.1466195583343506,
|
| 21420 |
+
"learning_rate": 1.977617471218934e-05,
|
| 21421 |
+
"loss": 5.3845,
|
| 21422 |
+
"step": 12236
|
| 21423 |
+
},
|
| 21424 |
+
{
|
| 21425 |
+
"epoch": 0.6,
|
| 21426 |
+
"grad_norm": 1.7159942388534546,
|
| 21427 |
+
"learning_rate": 1.976629280102772e-05,
|
| 21428 |
+
"loss": 5.551,
|
| 21429 |
+
"step": 12240
|
| 21430 |
+
},
|
| 21431 |
+
{
|
| 21432 |
+
"epoch": 0.6,
|
| 21433 |
+
"grad_norm": 1.8822176456451416,
|
| 21434 |
+
"learning_rate": 1.97564108898661e-05,
|
| 21435 |
+
"loss": 5.4277,
|
| 21436 |
+
"step": 12244
|
| 21437 |
+
},
|
| 21438 |
+
{
|
| 21439 |
+
"epoch": 0.61,
|
| 21440 |
+
"grad_norm": 1.798642635345459,
|
| 21441 |
+
"learning_rate": 1.974652897870448e-05,
|
| 21442 |
+
"loss": 5.6463,
|
| 21443 |
+
"step": 12248
|
| 21444 |
+
},
|
| 21445 |
+
{
|
| 21446 |
+
"epoch": 0.61,
|
| 21447 |
+
"grad_norm": 1.9936386346817017,
|
| 21448 |
+
"learning_rate": 1.9736647067542863e-05,
|
| 21449 |
+
"loss": 5.6342,
|
| 21450 |
+
"step": 12252
|
| 21451 |
+
},
|
| 21452 |
+
{
|
| 21453 |
+
"epoch": 0.61,
|
| 21454 |
+
"grad_norm": 2.1285390853881836,
|
| 21455 |
+
"learning_rate": 1.9726765156381245e-05,
|
| 21456 |
+
"loss": 5.5005,
|
| 21457 |
+
"step": 12256
|
| 21458 |
+
},
|
| 21459 |
+
{
|
| 21460 |
+
"epoch": 0.61,
|
| 21461 |
+
"grad_norm": 1.88554048538208,
|
| 21462 |
+
"learning_rate": 1.9716883245219624e-05,
|
| 21463 |
+
"loss": 5.5671,
|
| 21464 |
+
"step": 12260
|
| 21465 |
+
},
|
| 21466 |
+
{
|
| 21467 |
+
"epoch": 0.61,
|
| 21468 |
+
"grad_norm": 2.1367435455322266,
|
| 21469 |
+
"learning_rate": 1.970700133405801e-05,
|
| 21470 |
+
"loss": 5.4724,
|
| 21471 |
+
"step": 12264
|
| 21472 |
+
},
|
| 21473 |
+
{
|
| 21474 |
+
"epoch": 0.61,
|
| 21475 |
+
"grad_norm": 1.9615224599838257,
|
| 21476 |
+
"learning_rate": 1.969711942289639e-05,
|
| 21477 |
+
"loss": 5.4797,
|
| 21478 |
+
"step": 12268
|
| 21479 |
+
},
|
| 21480 |
+
{
|
| 21481 |
+
"epoch": 0.61,
|
| 21482 |
+
"grad_norm": 2.0672008991241455,
|
| 21483 |
+
"learning_rate": 1.968723751173477e-05,
|
| 21484 |
+
"loss": 5.4903,
|
| 21485 |
+
"step": 12272
|
| 21486 |
+
},
|
| 21487 |
+
{
|
| 21488 |
+
"epoch": 0.61,
|
| 21489 |
+
"grad_norm": 1.8822599649429321,
|
| 21490 |
+
"learning_rate": 1.9677355600573152e-05,
|
| 21491 |
+
"loss": 5.4196,
|
| 21492 |
+
"step": 12276
|
| 21493 |
+
},
|
| 21494 |
+
{
|
| 21495 |
+
"epoch": 0.61,
|
| 21496 |
+
"grad_norm": 1.9987417459487915,
|
| 21497 |
+
"learning_rate": 1.9667473689411534e-05,
|
| 21498 |
+
"loss": 5.5575,
|
| 21499 |
+
"step": 12280
|
| 21500 |
+
},
|
| 21501 |
+
{
|
| 21502 |
+
"epoch": 0.61,
|
| 21503 |
+
"grad_norm": 1.8995615243911743,
|
| 21504 |
+
"learning_rate": 1.9657591778249916e-05,
|
| 21505 |
+
"loss": 5.5343,
|
| 21506 |
+
"step": 12284
|
| 21507 |
+
},
|
| 21508 |
+
{
|
| 21509 |
+
"epoch": 0.61,
|
| 21510 |
+
"grad_norm": 1.9003994464874268,
|
| 21511 |
+
"learning_rate": 1.9647709867088295e-05,
|
| 21512 |
+
"loss": 5.5341,
|
| 21513 |
+
"step": 12288
|
| 21514 |
+
},
|
| 21515 |
+
{
|
| 21516 |
+
"epoch": 0.61,
|
| 21517 |
+
"grad_norm": 1.9870491027832031,
|
| 21518 |
+
"learning_rate": 1.9637827955926677e-05,
|
| 21519 |
+
"loss": 5.5928,
|
| 21520 |
+
"step": 12292
|
| 21521 |
+
},
|
| 21522 |
+
{
|
| 21523 |
+
"epoch": 0.61,
|
| 21524 |
+
"grad_norm": 2.072319269180298,
|
| 21525 |
+
"learning_rate": 1.962794604476506e-05,
|
| 21526 |
+
"loss": 5.4821,
|
| 21527 |
+
"step": 12296
|
| 21528 |
+
},
|
| 21529 |
+
{
|
| 21530 |
+
"epoch": 0.61,
|
| 21531 |
+
"grad_norm": 1.9005805253982544,
|
| 21532 |
+
"learning_rate": 1.9618064133603438e-05,
|
| 21533 |
+
"loss": 5.5296,
|
| 21534 |
+
"step": 12300
|
| 21535 |
+
},
|
| 21536 |
+
{
|
| 21537 |
+
"epoch": 0.61,
|
| 21538 |
+
"grad_norm": 1.8590092658996582,
|
| 21539 |
+
"learning_rate": 1.960818222244182e-05,
|
| 21540 |
+
"loss": 5.568,
|
| 21541 |
+
"step": 12304
|
| 21542 |
+
},
|
| 21543 |
+
{
|
| 21544 |
+
"epoch": 0.61,
|
| 21545 |
+
"grad_norm": 2.030334949493408,
|
| 21546 |
+
"learning_rate": 1.9598300311280202e-05,
|
| 21547 |
+
"loss": 5.4239,
|
| 21548 |
+
"step": 12308
|
| 21549 |
+
},
|
| 21550 |
+
{
|
| 21551 |
+
"epoch": 0.61,
|
| 21552 |
+
"grad_norm": 1.9241563081741333,
|
| 21553 |
+
"learning_rate": 1.9588418400118584e-05,
|
| 21554 |
+
"loss": 5.5618,
|
| 21555 |
+
"step": 12312
|
| 21556 |
+
},
|
| 21557 |
+
{
|
| 21558 |
+
"epoch": 0.61,
|
| 21559 |
+
"grad_norm": 1.9295763969421387,
|
| 21560 |
+
"learning_rate": 1.9578536488956963e-05,
|
| 21561 |
+
"loss": 5.6011,
|
| 21562 |
+
"step": 12316
|
| 21563 |
+
},
|
| 21564 |
+
{
|
| 21565 |
+
"epoch": 0.61,
|
| 21566 |
+
"grad_norm": 1.9351352453231812,
|
| 21567 |
+
"learning_rate": 1.9568654577795348e-05,
|
| 21568 |
+
"loss": 5.5549,
|
| 21569 |
+
"step": 12320
|
| 21570 |
+
},
|
| 21571 |
+
{
|
| 21572 |
+
"epoch": 0.61,
|
| 21573 |
+
"grad_norm": 2.0557003021240234,
|
| 21574 |
+
"learning_rate": 1.955877266663373e-05,
|
| 21575 |
+
"loss": 5.531,
|
| 21576 |
+
"step": 12324
|
| 21577 |
+
},
|
| 21578 |
+
{
|
| 21579 |
+
"epoch": 0.61,
|
| 21580 |
+
"grad_norm": 2.0774848461151123,
|
| 21581 |
+
"learning_rate": 1.954889075547211e-05,
|
| 21582 |
+
"loss": 5.5031,
|
| 21583 |
+
"step": 12328
|
| 21584 |
+
},
|
| 21585 |
+
{
|
| 21586 |
+
"epoch": 0.61,
|
| 21587 |
+
"grad_norm": 1.9636493921279907,
|
| 21588 |
+
"learning_rate": 1.953900884431049e-05,
|
| 21589 |
+
"loss": 5.5733,
|
| 21590 |
+
"step": 12332
|
| 21591 |
+
},
|
| 21592 |
+
{
|
| 21593 |
+
"epoch": 0.61,
|
| 21594 |
+
"grad_norm": 2.006387710571289,
|
| 21595 |
+
"learning_rate": 1.9529126933148873e-05,
|
| 21596 |
+
"loss": 5.5184,
|
| 21597 |
+
"step": 12336
|
| 21598 |
+
},
|
| 21599 |
+
{
|
| 21600 |
+
"epoch": 0.61,
|
| 21601 |
+
"grad_norm": 1.8912975788116455,
|
| 21602 |
+
"learning_rate": 1.9519245021987255e-05,
|
| 21603 |
+
"loss": 5.4529,
|
| 21604 |
+
"step": 12340
|
| 21605 |
+
},
|
| 21606 |
+
{
|
| 21607 |
+
"epoch": 0.61,
|
| 21608 |
+
"grad_norm": 2.028090715408325,
|
| 21609 |
+
"learning_rate": 1.9509363110825634e-05,
|
| 21610 |
+
"loss": 5.4242,
|
| 21611 |
+
"step": 12344
|
| 21612 |
+
},
|
| 21613 |
+
{
|
| 21614 |
+
"epoch": 0.61,
|
| 21615 |
+
"grad_norm": 2.042482376098633,
|
| 21616 |
+
"learning_rate": 1.9499481199664016e-05,
|
| 21617 |
+
"loss": 5.5455,
|
| 21618 |
+
"step": 12348
|
| 21619 |
+
},
|
| 21620 |
+
{
|
| 21621 |
+
"epoch": 0.61,
|
| 21622 |
+
"grad_norm": 2.1823337078094482,
|
| 21623 |
+
"learning_rate": 1.9489599288502398e-05,
|
| 21624 |
+
"loss": 5.4165,
|
| 21625 |
+
"step": 12352
|
| 21626 |
+
},
|
| 21627 |
+
{
|
| 21628 |
+
"epoch": 0.61,
|
| 21629 |
+
"grad_norm": 1.8578866720199585,
|
| 21630 |
+
"learning_rate": 1.947971737734078e-05,
|
| 21631 |
+
"loss": 5.4873,
|
| 21632 |
+
"step": 12356
|
| 21633 |
+
},
|
| 21634 |
+
{
|
| 21635 |
+
"epoch": 0.61,
|
| 21636 |
+
"grad_norm": 1.964311957359314,
|
| 21637 |
+
"learning_rate": 1.946983546617916e-05,
|
| 21638 |
+
"loss": 5.5457,
|
| 21639 |
+
"step": 12360
|
| 21640 |
+
},
|
| 21641 |
+
{
|
| 21642 |
+
"epoch": 0.61,
|
| 21643 |
+
"grad_norm": 2.030364513397217,
|
| 21644 |
+
"learning_rate": 1.945995355501754e-05,
|
| 21645 |
+
"loss": 5.474,
|
| 21646 |
+
"step": 12364
|
| 21647 |
+
},
|
| 21648 |
+
{
|
| 21649 |
+
"epoch": 0.61,
|
| 21650 |
+
"grad_norm": 2.0638349056243896,
|
| 21651 |
+
"learning_rate": 1.9450071643855923e-05,
|
| 21652 |
+
"loss": 5.5748,
|
| 21653 |
+
"step": 12368
|
| 21654 |
+
},
|
| 21655 |
+
{
|
| 21656 |
+
"epoch": 0.61,
|
| 21657 |
+
"grad_norm": 2.002610445022583,
|
| 21658 |
+
"learning_rate": 1.94401897326943e-05,
|
| 21659 |
+
"loss": 5.546,
|
| 21660 |
+
"step": 12372
|
| 21661 |
+
},
|
| 21662 |
+
{
|
| 21663 |
+
"epoch": 0.61,
|
| 21664 |
+
"grad_norm": 1.8753662109375,
|
| 21665 |
+
"learning_rate": 1.9430307821532687e-05,
|
| 21666 |
+
"loss": 5.5053,
|
| 21667 |
+
"step": 12376
|
| 21668 |
+
},
|
| 21669 |
+
{
|
| 21670 |
+
"epoch": 0.61,
|
| 21671 |
+
"grad_norm": 2.1950864791870117,
|
| 21672 |
+
"learning_rate": 1.942042591037107e-05,
|
| 21673 |
+
"loss": 5.4716,
|
| 21674 |
+
"step": 12380
|
| 21675 |
+
},
|
| 21676 |
+
{
|
| 21677 |
+
"epoch": 0.61,
|
| 21678 |
+
"grad_norm": 2.067065715789795,
|
| 21679 |
+
"learning_rate": 1.9410543999209447e-05,
|
| 21680 |
+
"loss": 5.5028,
|
| 21681 |
+
"step": 12384
|
| 21682 |
+
},
|
| 21683 |
+
{
|
| 21684 |
+
"epoch": 0.61,
|
| 21685 |
+
"grad_norm": 2.0129029750823975,
|
| 21686 |
+
"learning_rate": 1.940066208804783e-05,
|
| 21687 |
+
"loss": 5.522,
|
| 21688 |
+
"step": 12388
|
| 21689 |
+
},
|
| 21690 |
+
{
|
| 21691 |
+
"epoch": 0.61,
|
| 21692 |
+
"grad_norm": 1.5361533164978027,
|
| 21693 |
+
"learning_rate": 1.939078017688621e-05,
|
| 21694 |
+
"loss": 5.4896,
|
| 21695 |
+
"step": 12392
|
| 21696 |
+
},
|
| 21697 |
+
{
|
| 21698 |
+
"epoch": 0.61,
|
| 21699 |
+
"grad_norm": 1.945295810699463,
|
| 21700 |
+
"learning_rate": 1.9380898265724594e-05,
|
| 21701 |
+
"loss": 5.5306,
|
| 21702 |
+
"step": 12396
|
| 21703 |
+
},
|
| 21704 |
+
{
|
| 21705 |
+
"epoch": 0.61,
|
| 21706 |
+
"grad_norm": 1.7914199829101562,
|
| 21707 |
+
"learning_rate": 1.9371016354562972e-05,
|
| 21708 |
+
"loss": 5.5514,
|
| 21709 |
+
"step": 12400
|
| 21710 |
+
},
|
| 21711 |
+
{
|
| 21712 |
+
"epoch": 0.61,
|
| 21713 |
+
"grad_norm": 2.061509609222412,
|
| 21714 |
+
"learning_rate": 1.9361134443401354e-05,
|
| 21715 |
+
"loss": 5.5803,
|
| 21716 |
+
"step": 12404
|
| 21717 |
+
},
|
| 21718 |
+
{
|
| 21719 |
+
"epoch": 0.61,
|
| 21720 |
+
"grad_norm": 1.9697644710540771,
|
| 21721 |
+
"learning_rate": 1.9351252532239736e-05,
|
| 21722 |
+
"loss": 5.4989,
|
| 21723 |
+
"step": 12408
|
| 21724 |
+
},
|
| 21725 |
+
{
|
| 21726 |
+
"epoch": 0.61,
|
| 21727 |
+
"grad_norm": 1.801199197769165,
|
| 21728 |
+
"learning_rate": 1.934137062107812e-05,
|
| 21729 |
+
"loss": 5.6458,
|
| 21730 |
+
"step": 12412
|
| 21731 |
+
},
|
| 21732 |
+
{
|
| 21733 |
+
"epoch": 0.61,
|
| 21734 |
+
"grad_norm": 1.985129475593567,
|
| 21735 |
+
"learning_rate": 1.9331488709916497e-05,
|
| 21736 |
+
"loss": 5.4291,
|
| 21737 |
+
"step": 12416
|
| 21738 |
+
},
|
| 21739 |
+
{
|
| 21740 |
+
"epoch": 0.61,
|
| 21741 |
+
"grad_norm": 1.8330814838409424,
|
| 21742 |
+
"learning_rate": 1.932160679875488e-05,
|
| 21743 |
+
"loss": 5.3954,
|
| 21744 |
+
"step": 12420
|
| 21745 |
+
},
|
| 21746 |
+
{
|
| 21747 |
+
"epoch": 0.61,
|
| 21748 |
+
"grad_norm": 2.2082693576812744,
|
| 21749 |
+
"learning_rate": 1.931172488759326e-05,
|
| 21750 |
+
"loss": 5.5882,
|
| 21751 |
+
"step": 12424
|
| 21752 |
+
},
|
| 21753 |
+
{
|
| 21754 |
+
"epoch": 0.61,
|
| 21755 |
+
"grad_norm": 1.6590445041656494,
|
| 21756 |
+
"learning_rate": 1.930184297643164e-05,
|
| 21757 |
+
"loss": 5.4728,
|
| 21758 |
+
"step": 12428
|
| 21759 |
+
},
|
| 21760 |
+
{
|
| 21761 |
+
"epoch": 0.61,
|
| 21762 |
+
"grad_norm": 2.0365848541259766,
|
| 21763 |
+
"learning_rate": 1.9291961065270022e-05,
|
| 21764 |
+
"loss": 5.5032,
|
| 21765 |
+
"step": 12432
|
| 21766 |
+
},
|
| 21767 |
+
{
|
| 21768 |
+
"epoch": 0.61,
|
| 21769 |
+
"grad_norm": 1.9523799419403076,
|
| 21770 |
+
"learning_rate": 1.9282079154108407e-05,
|
| 21771 |
+
"loss": 5.4974,
|
| 21772 |
+
"step": 12436
|
| 21773 |
+
},
|
| 21774 |
+
{
|
| 21775 |
+
"epoch": 0.61,
|
| 21776 |
+
"grad_norm": 1.9144923686981201,
|
| 21777 |
+
"learning_rate": 1.9272197242946786e-05,
|
| 21778 |
+
"loss": 5.5348,
|
| 21779 |
+
"step": 12440
|
| 21780 |
+
},
|
| 21781 |
+
{
|
| 21782 |
+
"epoch": 0.61,
|
| 21783 |
+
"grad_norm": 1.7671104669570923,
|
| 21784 |
+
"learning_rate": 1.9262315331785168e-05,
|
| 21785 |
+
"loss": 5.5292,
|
| 21786 |
+
"step": 12444
|
| 21787 |
+
},
|
| 21788 |
+
{
|
| 21789 |
+
"epoch": 0.62,
|
| 21790 |
+
"grad_norm": 2.0515549182891846,
|
| 21791 |
+
"learning_rate": 1.925243342062355e-05,
|
| 21792 |
+
"loss": 5.4845,
|
| 21793 |
+
"step": 12448
|
| 21794 |
+
},
|
| 21795 |
+
{
|
| 21796 |
+
"epoch": 0.62,
|
| 21797 |
+
"grad_norm": 1.9381266832351685,
|
| 21798 |
+
"learning_rate": 1.9242551509461932e-05,
|
| 21799 |
+
"loss": 5.5455,
|
| 21800 |
+
"step": 12452
|
| 21801 |
+
},
|
| 21802 |
+
{
|
| 21803 |
+
"epoch": 0.62,
|
| 21804 |
+
"grad_norm": 2.017817497253418,
|
| 21805 |
+
"learning_rate": 1.923266959830031e-05,
|
| 21806 |
+
"loss": 5.3844,
|
| 21807 |
+
"step": 12456
|
| 21808 |
+
},
|
| 21809 |
+
{
|
| 21810 |
+
"epoch": 0.62,
|
| 21811 |
+
"grad_norm": 1.8400537967681885,
|
| 21812 |
+
"learning_rate": 1.9222787687138693e-05,
|
| 21813 |
+
"loss": 5.4878,
|
| 21814 |
+
"step": 12460
|
| 21815 |
+
},
|
| 21816 |
+
{
|
| 21817 |
+
"epoch": 0.62,
|
| 21818 |
+
"grad_norm": 2.263641119003296,
|
| 21819 |
+
"learning_rate": 1.9212905775977075e-05,
|
| 21820 |
+
"loss": 5.5841,
|
| 21821 |
+
"step": 12464
|
| 21822 |
+
},
|
| 21823 |
+
{
|
| 21824 |
+
"epoch": 0.62,
|
| 21825 |
+
"grad_norm": 1.8722437620162964,
|
| 21826 |
+
"learning_rate": 1.9203023864815457e-05,
|
| 21827 |
+
"loss": 5.4782,
|
| 21828 |
+
"step": 12468
|
| 21829 |
+
},
|
| 21830 |
+
{
|
| 21831 |
+
"epoch": 0.62,
|
| 21832 |
+
"grad_norm": 2.020585060119629,
|
| 21833 |
+
"learning_rate": 1.9193141953653836e-05,
|
| 21834 |
+
"loss": 5.4668,
|
| 21835 |
+
"step": 12472
|
| 21836 |
+
},
|
| 21837 |
+
{
|
| 21838 |
+
"epoch": 0.62,
|
| 21839 |
+
"grad_norm": 1.7479088306427002,
|
| 21840 |
+
"learning_rate": 1.9183260042492218e-05,
|
| 21841 |
+
"loss": 5.5514,
|
| 21842 |
+
"step": 12476
|
| 21843 |
+
},
|
| 21844 |
+
{
|
| 21845 |
+
"epoch": 0.62,
|
| 21846 |
+
"grad_norm": 1.9156551361083984,
|
| 21847 |
+
"learning_rate": 1.91733781313306e-05,
|
| 21848 |
+
"loss": 5.4994,
|
| 21849 |
+
"step": 12480
|
| 21850 |
+
},
|
| 21851 |
+
{
|
| 21852 |
+
"epoch": 0.62,
|
| 21853 |
+
"grad_norm": 1.882408618927002,
|
| 21854 |
+
"learning_rate": 1.9163496220168982e-05,
|
| 21855 |
+
"loss": 5.4834,
|
| 21856 |
+
"step": 12484
|
| 21857 |
+
},
|
| 21858 |
+
{
|
| 21859 |
+
"epoch": 0.62,
|
| 21860 |
+
"grad_norm": 2.083282232284546,
|
| 21861 |
+
"learning_rate": 1.915361430900736e-05,
|
| 21862 |
+
"loss": 5.5116,
|
| 21863 |
+
"step": 12488
|
| 21864 |
+
},
|
| 21865 |
+
{
|
| 21866 |
+
"epoch": 0.62,
|
| 21867 |
+
"grad_norm": 1.9320555925369263,
|
| 21868 |
+
"learning_rate": 1.9143732397845746e-05,
|
| 21869 |
+
"loss": 5.5378,
|
| 21870 |
+
"step": 12492
|
| 21871 |
+
},
|
| 21872 |
+
{
|
| 21873 |
+
"epoch": 0.62,
|
| 21874 |
+
"grad_norm": 2.215940475463867,
|
| 21875 |
+
"learning_rate": 1.9133850486684128e-05,
|
| 21876 |
+
"loss": 5.4937,
|
| 21877 |
+
"step": 12496
|
| 21878 |
+
},
|
| 21879 |
+
{
|
| 21880 |
+
"epoch": 0.62,
|
| 21881 |
+
"grad_norm": 2.308119297027588,
|
| 21882 |
+
"learning_rate": 1.9123968575522507e-05,
|
| 21883 |
+
"loss": 5.555,
|
| 21884 |
+
"step": 12500
|
| 21885 |
+
},
|
| 21886 |
+
{
|
| 21887 |
+
"epoch": 0.62,
|
| 21888 |
+
"grad_norm": 2.178675413131714,
|
| 21889 |
+
"learning_rate": 1.911408666436089e-05,
|
| 21890 |
+
"loss": 5.487,
|
| 21891 |
+
"step": 12504
|
| 21892 |
+
},
|
| 21893 |
+
{
|
| 21894 |
+
"epoch": 0.62,
|
| 21895 |
+
"grad_norm": 2.004458427429199,
|
| 21896 |
+
"learning_rate": 1.910420475319927e-05,
|
| 21897 |
+
"loss": 5.5416,
|
| 21898 |
+
"step": 12508
|
| 21899 |
+
},
|
| 21900 |
+
{
|
| 21901 |
+
"epoch": 0.62,
|
| 21902 |
+
"grad_norm": 2.0435168743133545,
|
| 21903 |
+
"learning_rate": 1.909432284203765e-05,
|
| 21904 |
+
"loss": 5.4122,
|
| 21905 |
+
"step": 12512
|
| 21906 |
+
},
|
| 21907 |
+
{
|
| 21908 |
+
"epoch": 0.62,
|
| 21909 |
+
"grad_norm": 2.0281968116760254,
|
| 21910 |
+
"learning_rate": 1.9084440930876032e-05,
|
| 21911 |
+
"loss": 5.5032,
|
| 21912 |
+
"step": 12516
|
| 21913 |
+
},
|
| 21914 |
+
{
|
| 21915 |
+
"epoch": 0.62,
|
| 21916 |
+
"grad_norm": 1.9901241064071655,
|
| 21917 |
+
"learning_rate": 1.9074559019714414e-05,
|
| 21918 |
+
"loss": 5.5346,
|
| 21919 |
+
"step": 12520
|
| 21920 |
+
},
|
| 21921 |
+
{
|
| 21922 |
+
"epoch": 0.62,
|
| 21923 |
+
"grad_norm": 2.0608649253845215,
|
| 21924 |
+
"learning_rate": 1.9064677108552796e-05,
|
| 21925 |
+
"loss": 5.5318,
|
| 21926 |
+
"step": 12524
|
| 21927 |
+
},
|
| 21928 |
+
{
|
| 21929 |
+
"epoch": 0.62,
|
| 21930 |
+
"grad_norm": 2.1132655143737793,
|
| 21931 |
+
"learning_rate": 1.9054795197391175e-05,
|
| 21932 |
+
"loss": 5.5227,
|
| 21933 |
+
"step": 12528
|
| 21934 |
+
},
|
| 21935 |
+
{
|
| 21936 |
+
"epoch": 0.62,
|
| 21937 |
+
"grad_norm": 2.1006295680999756,
|
| 21938 |
+
"learning_rate": 1.9044913286229557e-05,
|
| 21939 |
+
"loss": 5.4704,
|
| 21940 |
+
"step": 12532
|
| 21941 |
+
},
|
| 21942 |
+
{
|
| 21943 |
+
"epoch": 0.62,
|
| 21944 |
+
"grad_norm": 1.8386894464492798,
|
| 21945 |
+
"learning_rate": 1.903503137506794e-05,
|
| 21946 |
+
"loss": 5.4159,
|
| 21947 |
+
"step": 12536
|
| 21948 |
+
},
|
| 21949 |
+
{
|
| 21950 |
+
"epoch": 0.62,
|
| 21951 |
+
"grad_norm": 1.9647696018218994,
|
| 21952 |
+
"learning_rate": 1.902514946390632e-05,
|
| 21953 |
+
"loss": 5.6805,
|
| 21954 |
+
"step": 12540
|
| 21955 |
+
},
|
| 21956 |
+
{
|
| 21957 |
+
"epoch": 0.62,
|
| 21958 |
+
"grad_norm": 2.1188244819641113,
|
| 21959 |
+
"learning_rate": 1.90152675527447e-05,
|
| 21960 |
+
"loss": 5.4004,
|
| 21961 |
+
"step": 12544
|
| 21962 |
+
},
|
| 21963 |
+
{
|
| 21964 |
+
"epoch": 0.62,
|
| 21965 |
+
"grad_norm": 2.0802998542785645,
|
| 21966 |
+
"learning_rate": 1.9005385641583085e-05,
|
| 21967 |
+
"loss": 5.4416,
|
| 21968 |
+
"step": 12548
|
| 21969 |
+
},
|
| 21970 |
+
{
|
| 21971 |
+
"epoch": 0.62,
|
| 21972 |
+
"grad_norm": 1.834084153175354,
|
| 21973 |
+
"learning_rate": 1.8995503730421467e-05,
|
| 21974 |
+
"loss": 5.6306,
|
| 21975 |
+
"step": 12552
|
| 21976 |
+
},
|
| 21977 |
+
{
|
| 21978 |
+
"epoch": 0.62,
|
| 21979 |
+
"grad_norm": 1.8112331628799438,
|
| 21980 |
+
"learning_rate": 1.8985621819259846e-05,
|
| 21981 |
+
"loss": 5.5333,
|
| 21982 |
+
"step": 12556
|
| 21983 |
+
},
|
| 21984 |
+
{
|
| 21985 |
+
"epoch": 0.62,
|
| 21986 |
+
"grad_norm": 1.899707317352295,
|
| 21987 |
+
"learning_rate": 1.8975739908098228e-05,
|
| 21988 |
+
"loss": 5.4976,
|
| 21989 |
+
"step": 12560
|
| 21990 |
+
},
|
| 21991 |
+
{
|
| 21992 |
+
"epoch": 0.62,
|
| 21993 |
+
"grad_norm": 2.298161506652832,
|
| 21994 |
+
"learning_rate": 1.896585799693661e-05,
|
| 21995 |
+
"loss": 5.594,
|
| 21996 |
+
"step": 12564
|
| 21997 |
+
},
|
| 21998 |
+
{
|
| 21999 |
+
"epoch": 0.62,
|
| 22000 |
+
"grad_norm": 2.173597574234009,
|
| 22001 |
+
"learning_rate": 1.8955976085774992e-05,
|
| 22002 |
+
"loss": 5.572,
|
| 22003 |
+
"step": 12568
|
| 22004 |
+
},
|
| 22005 |
+
{
|
| 22006 |
+
"epoch": 0.62,
|
| 22007 |
+
"grad_norm": 1.8348718881607056,
|
| 22008 |
+
"learning_rate": 1.894609417461337e-05,
|
| 22009 |
+
"loss": 5.4743,
|
| 22010 |
+
"step": 12572
|
| 22011 |
+
},
|
| 22012 |
+
{
|
| 22013 |
+
"epoch": 0.62,
|
| 22014 |
+
"grad_norm": 1.7967536449432373,
|
| 22015 |
+
"learning_rate": 1.8936212263451753e-05,
|
| 22016 |
+
"loss": 5.5196,
|
| 22017 |
+
"step": 12576
|
| 22018 |
+
},
|
| 22019 |
+
{
|
| 22020 |
+
"epoch": 0.62,
|
| 22021 |
+
"grad_norm": 1.9703553915023804,
|
| 22022 |
+
"learning_rate": 1.8926330352290135e-05,
|
| 22023 |
+
"loss": 5.532,
|
| 22024 |
+
"step": 12580
|
| 22025 |
+
},
|
| 22026 |
+
{
|
| 22027 |
+
"epoch": 0.62,
|
| 22028 |
+
"grad_norm": 1.7581743001937866,
|
| 22029 |
+
"learning_rate": 1.8916448441128513e-05,
|
| 22030 |
+
"loss": 5.5114,
|
| 22031 |
+
"step": 12584
|
| 22032 |
+
},
|
| 22033 |
+
{
|
| 22034 |
+
"epoch": 0.62,
|
| 22035 |
+
"grad_norm": 2.0711758136749268,
|
| 22036 |
+
"learning_rate": 1.8906566529966895e-05,
|
| 22037 |
+
"loss": 5.4094,
|
| 22038 |
+
"step": 12588
|
| 22039 |
+
},
|
| 22040 |
+
{
|
| 22041 |
+
"epoch": 0.62,
|
| 22042 |
+
"grad_norm": 1.7996923923492432,
|
| 22043 |
+
"learning_rate": 1.8896684618805277e-05,
|
| 22044 |
+
"loss": 5.5836,
|
| 22045 |
+
"step": 12592
|
| 22046 |
+
},
|
| 22047 |
+
{
|
| 22048 |
+
"epoch": 0.62,
|
| 22049 |
+
"grad_norm": 1.97800612449646,
|
| 22050 |
+
"learning_rate": 1.888680270764366e-05,
|
| 22051 |
+
"loss": 5.4898,
|
| 22052 |
+
"step": 12596
|
| 22053 |
+
},
|
| 22054 |
+
{
|
| 22055 |
+
"epoch": 0.62,
|
| 22056 |
+
"grad_norm": 1.7940218448638916,
|
| 22057 |
+
"learning_rate": 1.8876920796482038e-05,
|
| 22058 |
+
"loss": 5.6005,
|
| 22059 |
+
"step": 12600
|
| 22060 |
+
},
|
| 22061 |
+
{
|
| 22062 |
+
"epoch": 0.62,
|
| 22063 |
+
"grad_norm": 1.8298521041870117,
|
| 22064 |
+
"learning_rate": 1.886703888532042e-05,
|
| 22065 |
+
"loss": 5.55,
|
| 22066 |
+
"step": 12604
|
| 22067 |
+
},
|
| 22068 |
+
{
|
| 22069 |
+
"epoch": 0.62,
|
| 22070 |
+
"grad_norm": 2.040109872817993,
|
| 22071 |
+
"learning_rate": 1.8857156974158806e-05,
|
| 22072 |
+
"loss": 5.5685,
|
| 22073 |
+
"step": 12608
|
| 22074 |
+
},
|
| 22075 |
+
{
|
| 22076 |
+
"epoch": 0.62,
|
| 22077 |
+
"grad_norm": 1.8531662225723267,
|
| 22078 |
+
"learning_rate": 1.8847275062997184e-05,
|
| 22079 |
+
"loss": 5.5343,
|
| 22080 |
+
"step": 12612
|
| 22081 |
+
},
|
| 22082 |
+
{
|
| 22083 |
+
"epoch": 0.62,
|
| 22084 |
+
"grad_norm": 2.1842970848083496,
|
| 22085 |
+
"learning_rate": 1.8837393151835566e-05,
|
| 22086 |
+
"loss": 5.4704,
|
| 22087 |
+
"step": 12616
|
| 22088 |
+
},
|
| 22089 |
+
{
|
| 22090 |
+
"epoch": 0.62,
|
| 22091 |
+
"grad_norm": 1.876779556274414,
|
| 22092 |
+
"learning_rate": 1.882751124067395e-05,
|
| 22093 |
+
"loss": 5.5272,
|
| 22094 |
+
"step": 12620
|
| 22095 |
+
},
|
| 22096 |
+
{
|
| 22097 |
+
"epoch": 0.62,
|
| 22098 |
+
"grad_norm": 1.9100033044815063,
|
| 22099 |
+
"learning_rate": 1.881762932951233e-05,
|
| 22100 |
+
"loss": 5.4669,
|
| 22101 |
+
"step": 12624
|
| 22102 |
+
},
|
| 22103 |
+
{
|
| 22104 |
+
"epoch": 0.62,
|
| 22105 |
+
"grad_norm": 2.233772039413452,
|
| 22106 |
+
"learning_rate": 1.880774741835071e-05,
|
| 22107 |
+
"loss": 5.5786,
|
| 22108 |
+
"step": 12628
|
| 22109 |
+
},
|
| 22110 |
+
{
|
| 22111 |
+
"epoch": 0.62,
|
| 22112 |
+
"grad_norm": 2.021141767501831,
|
| 22113 |
+
"learning_rate": 1.879786550718909e-05,
|
| 22114 |
+
"loss": 5.5469,
|
| 22115 |
+
"step": 12632
|
| 22116 |
+
},
|
| 22117 |
+
{
|
| 22118 |
+
"epoch": 0.62,
|
| 22119 |
+
"grad_norm": 1.8748712539672852,
|
| 22120 |
+
"learning_rate": 1.8787983596027473e-05,
|
| 22121 |
+
"loss": 5.4186,
|
| 22122 |
+
"step": 12636
|
| 22123 |
+
},
|
| 22124 |
+
{
|
| 22125 |
+
"epoch": 0.62,
|
| 22126 |
+
"grad_norm": 2.0556745529174805,
|
| 22127 |
+
"learning_rate": 1.8778101684865855e-05,
|
| 22128 |
+
"loss": 5.6404,
|
| 22129 |
+
"step": 12640
|
| 22130 |
+
},
|
| 22131 |
+
{
|
| 22132 |
+
"epoch": 0.62,
|
| 22133 |
+
"grad_norm": 2.089085102081299,
|
| 22134 |
+
"learning_rate": 1.8768219773704234e-05,
|
| 22135 |
+
"loss": 5.6056,
|
| 22136 |
+
"step": 12644
|
| 22137 |
+
},
|
| 22138 |
+
{
|
| 22139 |
+
"epoch": 0.62,
|
| 22140 |
+
"grad_norm": 1.8434518575668335,
|
| 22141 |
+
"learning_rate": 1.8758337862542616e-05,
|
| 22142 |
+
"loss": 5.6488,
|
| 22143 |
+
"step": 12648
|
| 22144 |
+
},
|
| 22145 |
+
{
|
| 22146 |
+
"epoch": 0.63,
|
| 22147 |
+
"grad_norm": 2.003434658050537,
|
| 22148 |
+
"learning_rate": 1.8748455951380998e-05,
|
| 22149 |
+
"loss": 5.5057,
|
| 22150 |
+
"step": 12652
|
| 22151 |
+
},
|
| 22152 |
+
{
|
| 22153 |
+
"epoch": 0.63,
|
| 22154 |
+
"grad_norm": 2.292663335800171,
|
| 22155 |
+
"learning_rate": 1.8738574040219377e-05,
|
| 22156 |
+
"loss": 5.5022,
|
| 22157 |
+
"step": 12656
|
| 22158 |
+
},
|
| 22159 |
+
{
|
| 22160 |
+
"epoch": 0.63,
|
| 22161 |
+
"grad_norm": 1.9476063251495361,
|
| 22162 |
+
"learning_rate": 1.872869212905776e-05,
|
| 22163 |
+
"loss": 5.5844,
|
| 22164 |
+
"step": 12660
|
| 22165 |
+
},
|
| 22166 |
+
{
|
| 22167 |
+
"epoch": 0.63,
|
| 22168 |
+
"grad_norm": 2.138032913208008,
|
| 22169 |
+
"learning_rate": 1.8718810217896144e-05,
|
| 22170 |
+
"loss": 5.4383,
|
| 22171 |
+
"step": 12664
|
| 22172 |
+
},
|
| 22173 |
+
{
|
| 22174 |
+
"epoch": 0.63,
|
| 22175 |
+
"grad_norm": 1.8477308750152588,
|
| 22176 |
+
"learning_rate": 1.8708928306734523e-05,
|
| 22177 |
+
"loss": 5.5665,
|
| 22178 |
+
"step": 12668
|
| 22179 |
+
},
|
| 22180 |
+
{
|
| 22181 |
+
"epoch": 0.63,
|
| 22182 |
+
"grad_norm": 1.9181241989135742,
|
| 22183 |
+
"learning_rate": 1.8699046395572905e-05,
|
| 22184 |
+
"loss": 5.5661,
|
| 22185 |
+
"step": 12672
|
| 22186 |
+
},
|
| 22187 |
+
{
|
| 22188 |
+
"epoch": 0.63,
|
| 22189 |
+
"grad_norm": 1.928312063217163,
|
| 22190 |
+
"learning_rate": 1.8689164484411287e-05,
|
| 22191 |
+
"loss": 5.5366,
|
| 22192 |
+
"step": 12676
|
| 22193 |
+
},
|
| 22194 |
+
{
|
| 22195 |
+
"epoch": 0.63,
|
| 22196 |
+
"grad_norm": 2.210855484008789,
|
| 22197 |
+
"learning_rate": 1.867928257324967e-05,
|
| 22198 |
+
"loss": 5.5136,
|
| 22199 |
+
"step": 12680
|
| 22200 |
+
},
|
| 22201 |
+
{
|
| 22202 |
+
"epoch": 0.63,
|
| 22203 |
+
"grad_norm": 2.030755043029785,
|
| 22204 |
+
"learning_rate": 1.8669400662088048e-05,
|
| 22205 |
+
"loss": 5.4813,
|
| 22206 |
+
"step": 12684
|
| 22207 |
+
},
|
| 22208 |
+
{
|
| 22209 |
+
"epoch": 0.63,
|
| 22210 |
+
"grad_norm": 2.1717166900634766,
|
| 22211 |
+
"learning_rate": 1.865951875092643e-05,
|
| 22212 |
+
"loss": 5.579,
|
| 22213 |
+
"step": 12688
|
| 22214 |
+
},
|
| 22215 |
+
{
|
| 22216 |
+
"epoch": 0.63,
|
| 22217 |
+
"grad_norm": 2.068718671798706,
|
| 22218 |
+
"learning_rate": 1.8649636839764812e-05,
|
| 22219 |
+
"loss": 5.4381,
|
| 22220 |
+
"step": 12692
|
| 22221 |
+
},
|
| 22222 |
+
{
|
| 22223 |
+
"epoch": 0.63,
|
| 22224 |
+
"grad_norm": 2.0134997367858887,
|
| 22225 |
+
"learning_rate": 1.8639754928603194e-05,
|
| 22226 |
+
"loss": 5.6513,
|
| 22227 |
+
"step": 12696
|
| 22228 |
+
},
|
| 22229 |
+
{
|
| 22230 |
+
"epoch": 0.63,
|
| 22231 |
+
"grad_norm": 2.061288356781006,
|
| 22232 |
+
"learning_rate": 1.8629873017441573e-05,
|
| 22233 |
+
"loss": 5.4894,
|
| 22234 |
+
"step": 12700
|
| 22235 |
+
},
|
| 22236 |
+
{
|
| 22237 |
+
"epoch": 0.63,
|
| 22238 |
+
"grad_norm": 2.0297648906707764,
|
| 22239 |
+
"learning_rate": 1.8619991106279955e-05,
|
| 22240 |
+
"loss": 5.5485,
|
| 22241 |
+
"step": 12704
|
| 22242 |
+
},
|
| 22243 |
+
{
|
| 22244 |
+
"epoch": 0.63,
|
| 22245 |
+
"grad_norm": 2.0792784690856934,
|
| 22246 |
+
"learning_rate": 1.8610109195118337e-05,
|
| 22247 |
+
"loss": 5.5622,
|
| 22248 |
+
"step": 12708
|
| 22249 |
+
},
|
| 22250 |
+
{
|
| 22251 |
+
"epoch": 0.63,
|
| 22252 |
+
"grad_norm": 2.003371238708496,
|
| 22253 |
+
"learning_rate": 1.8600227283956715e-05,
|
| 22254 |
+
"loss": 5.5616,
|
| 22255 |
+
"step": 12712
|
| 22256 |
+
},
|
| 22257 |
+
{
|
| 22258 |
+
"epoch": 0.63,
|
| 22259 |
+
"grad_norm": 2.0913472175598145,
|
| 22260 |
+
"learning_rate": 1.8590345372795098e-05,
|
| 22261 |
+
"loss": 5.6751,
|
| 22262 |
+
"step": 12716
|
| 22263 |
+
},
|
| 22264 |
+
{
|
| 22265 |
+
"epoch": 0.63,
|
| 22266 |
+
"grad_norm": 1.9094600677490234,
|
| 22267 |
+
"learning_rate": 1.8580463461633483e-05,
|
| 22268 |
+
"loss": 5.5047,
|
| 22269 |
+
"step": 12720
|
| 22270 |
+
},
|
| 22271 |
+
{
|
| 22272 |
+
"epoch": 0.63,
|
| 22273 |
+
"grad_norm": 2.1928985118865967,
|
| 22274 |
+
"learning_rate": 1.857058155047186e-05,
|
| 22275 |
+
"loss": 5.6044,
|
| 22276 |
+
"step": 12724
|
| 22277 |
+
},
|
| 22278 |
+
{
|
| 22279 |
+
"epoch": 0.63,
|
| 22280 |
+
"grad_norm": 1.9102288484573364,
|
| 22281 |
+
"learning_rate": 1.8560699639310244e-05,
|
| 22282 |
+
"loss": 5.4843,
|
| 22283 |
+
"step": 12728
|
| 22284 |
+
},
|
| 22285 |
+
{
|
| 22286 |
+
"epoch": 0.63,
|
| 22287 |
+
"grad_norm": 1.908156156539917,
|
| 22288 |
+
"learning_rate": 1.8550817728148626e-05,
|
| 22289 |
+
"loss": 5.5217,
|
| 22290 |
+
"step": 12732
|
| 22291 |
+
},
|
| 22292 |
+
{
|
| 22293 |
+
"epoch": 0.63,
|
| 22294 |
+
"grad_norm": 1.8940017223358154,
|
| 22295 |
+
"learning_rate": 1.8540935816987008e-05,
|
| 22296 |
+
"loss": 5.5211,
|
| 22297 |
+
"step": 12736
|
| 22298 |
+
},
|
| 22299 |
+
{
|
| 22300 |
+
"epoch": 0.63,
|
| 22301 |
+
"grad_norm": 1.6733046770095825,
|
| 22302 |
+
"learning_rate": 1.8531053905825386e-05,
|
| 22303 |
+
"loss": 5.6507,
|
| 22304 |
+
"step": 12740
|
| 22305 |
+
},
|
| 22306 |
+
{
|
| 22307 |
+
"epoch": 0.63,
|
| 22308 |
+
"grad_norm": 1.9494881629943848,
|
| 22309 |
+
"learning_rate": 1.852117199466377e-05,
|
| 22310 |
+
"loss": 5.6028,
|
| 22311 |
+
"step": 12744
|
| 22312 |
+
},
|
| 22313 |
+
{
|
| 22314 |
+
"epoch": 0.63,
|
| 22315 |
+
"grad_norm": 2.361642360687256,
|
| 22316 |
+
"learning_rate": 1.851129008350215e-05,
|
| 22317 |
+
"loss": 5.5678,
|
| 22318 |
+
"step": 12748
|
| 22319 |
+
},
|
| 22320 |
+
{
|
| 22321 |
+
"epoch": 0.63,
|
| 22322 |
+
"grad_norm": 1.9810757637023926,
|
| 22323 |
+
"learning_rate": 1.8501408172340533e-05,
|
| 22324 |
+
"loss": 5.5055,
|
| 22325 |
+
"step": 12752
|
| 22326 |
+
},
|
| 22327 |
+
{
|
| 22328 |
+
"epoch": 0.63,
|
| 22329 |
+
"grad_norm": 2.196544885635376,
|
| 22330 |
+
"learning_rate": 1.849152626117891e-05,
|
| 22331 |
+
"loss": 5.4622,
|
| 22332 |
+
"step": 12756
|
| 22333 |
+
},
|
| 22334 |
+
{
|
| 22335 |
+
"epoch": 0.63,
|
| 22336 |
+
"grad_norm": 1.841874599456787,
|
| 22337 |
+
"learning_rate": 1.8481644350017293e-05,
|
| 22338 |
+
"loss": 5.525,
|
| 22339 |
+
"step": 12760
|
| 22340 |
+
},
|
| 22341 |
+
{
|
| 22342 |
+
"epoch": 0.63,
|
| 22343 |
+
"grad_norm": 1.982703685760498,
|
| 22344 |
+
"learning_rate": 1.8471762438855675e-05,
|
| 22345 |
+
"loss": 5.5797,
|
| 22346 |
+
"step": 12764
|
| 22347 |
+
},
|
| 22348 |
+
{
|
| 22349 |
+
"epoch": 0.63,
|
| 22350 |
+
"grad_norm": 2.193528413772583,
|
| 22351 |
+
"learning_rate": 1.8461880527694058e-05,
|
| 22352 |
+
"loss": 5.4744,
|
| 22353 |
+
"step": 12768
|
| 22354 |
+
},
|
| 22355 |
+
{
|
| 22356 |
+
"epoch": 0.63,
|
| 22357 |
+
"grad_norm": 1.6755714416503906,
|
| 22358 |
+
"learning_rate": 1.8451998616532436e-05,
|
| 22359 |
+
"loss": 5.4944,
|
| 22360 |
+
"step": 12772
|
| 22361 |
+
},
|
| 22362 |
+
{
|
| 22363 |
+
"epoch": 0.63,
|
| 22364 |
+
"grad_norm": 1.9214802980422974,
|
| 22365 |
+
"learning_rate": 1.8442116705370818e-05,
|
| 22366 |
+
"loss": 5.4525,
|
| 22367 |
+
"step": 12776
|
| 22368 |
+
},
|
| 22369 |
+
{
|
| 22370 |
+
"epoch": 0.63,
|
| 22371 |
+
"grad_norm": 2.1332216262817383,
|
| 22372 |
+
"learning_rate": 1.8432234794209204e-05,
|
| 22373 |
+
"loss": 5.4974,
|
| 22374 |
+
"step": 12780
|
| 22375 |
+
},
|
| 22376 |
+
{
|
| 22377 |
+
"epoch": 0.63,
|
| 22378 |
+
"grad_norm": 1.9983022212982178,
|
| 22379 |
+
"learning_rate": 1.8422352883047582e-05,
|
| 22380 |
+
"loss": 5.4737,
|
| 22381 |
+
"step": 12784
|
| 22382 |
+
},
|
| 22383 |
+
{
|
| 22384 |
+
"epoch": 0.63,
|
| 22385 |
+
"grad_norm": 2.090367555618286,
|
| 22386 |
+
"learning_rate": 1.8412470971885964e-05,
|
| 22387 |
+
"loss": 5.4413,
|
| 22388 |
+
"step": 12788
|
| 22389 |
+
},
|
| 22390 |
+
{
|
| 22391 |
+
"epoch": 0.63,
|
| 22392 |
+
"grad_norm": 1.903393030166626,
|
| 22393 |
+
"learning_rate": 1.8402589060724346e-05,
|
| 22394 |
+
"loss": 5.5338,
|
| 22395 |
+
"step": 12792
|
| 22396 |
+
},
|
| 22397 |
+
{
|
| 22398 |
+
"epoch": 0.63,
|
| 22399 |
+
"grad_norm": 1.8425928354263306,
|
| 22400 |
+
"learning_rate": 1.8392707149562725e-05,
|
| 22401 |
+
"loss": 5.5461,
|
| 22402 |
+
"step": 12796
|
| 22403 |
+
},
|
| 22404 |
+
{
|
| 22405 |
+
"epoch": 0.63,
|
| 22406 |
+
"grad_norm": 1.7911487817764282,
|
| 22407 |
+
"learning_rate": 1.8382825238401107e-05,
|
| 22408 |
+
"loss": 5.5144,
|
| 22409 |
+
"step": 12800
|
| 22410 |
+
},
|
| 22411 |
+
{
|
| 22412 |
+
"epoch": 0.63,
|
| 22413 |
+
"grad_norm": 1.9388378858566284,
|
| 22414 |
+
"learning_rate": 1.837294332723949e-05,
|
| 22415 |
+
"loss": 5.4777,
|
| 22416 |
+
"step": 12804
|
| 22417 |
+
},
|
| 22418 |
+
{
|
| 22419 |
+
"epoch": 0.63,
|
| 22420 |
+
"grad_norm": 1.9651098251342773,
|
| 22421 |
+
"learning_rate": 1.836306141607787e-05,
|
| 22422 |
+
"loss": 5.583,
|
| 22423 |
+
"step": 12808
|
| 22424 |
+
},
|
| 22425 |
+
{
|
| 22426 |
+
"epoch": 0.63,
|
| 22427 |
+
"grad_norm": 2.097846269607544,
|
| 22428 |
+
"learning_rate": 1.835317950491625e-05,
|
| 22429 |
+
"loss": 5.4835,
|
| 22430 |
+
"step": 12812
|
| 22431 |
+
},
|
| 22432 |
+
{
|
| 22433 |
+
"epoch": 0.63,
|
| 22434 |
+
"grad_norm": 1.8683522939682007,
|
| 22435 |
+
"learning_rate": 1.8343297593754632e-05,
|
| 22436 |
+
"loss": 5.542,
|
| 22437 |
+
"step": 12816
|
| 22438 |
+
},
|
| 22439 |
+
{
|
| 22440 |
+
"epoch": 0.63,
|
| 22441 |
+
"grad_norm": 1.9476234912872314,
|
| 22442 |
+
"learning_rate": 1.8333415682593014e-05,
|
| 22443 |
+
"loss": 5.5376,
|
| 22444 |
+
"step": 12820
|
| 22445 |
+
},
|
| 22446 |
+
{
|
| 22447 |
+
"epoch": 0.63,
|
| 22448 |
+
"grad_norm": 2.049328088760376,
|
| 22449 |
+
"learning_rate": 1.8323533771431396e-05,
|
| 22450 |
+
"loss": 5.4911,
|
| 22451 |
+
"step": 12824
|
| 22452 |
+
},
|
| 22453 |
+
{
|
| 22454 |
+
"epoch": 0.63,
|
| 22455 |
+
"grad_norm": 1.8876453638076782,
|
| 22456 |
+
"learning_rate": 1.8313651860269775e-05,
|
| 22457 |
+
"loss": 5.5226,
|
| 22458 |
+
"step": 12828
|
| 22459 |
+
},
|
| 22460 |
+
{
|
| 22461 |
+
"epoch": 0.63,
|
| 22462 |
+
"grad_norm": 1.9597487449645996,
|
| 22463 |
+
"learning_rate": 1.8303769949108157e-05,
|
| 22464 |
+
"loss": 5.4785,
|
| 22465 |
+
"step": 12832
|
| 22466 |
+
},
|
| 22467 |
+
{
|
| 22468 |
+
"epoch": 0.63,
|
| 22469 |
+
"grad_norm": 2.0029759407043457,
|
| 22470 |
+
"learning_rate": 1.8293888037946542e-05,
|
| 22471 |
+
"loss": 5.5109,
|
| 22472 |
+
"step": 12836
|
| 22473 |
+
},
|
| 22474 |
+
{
|
| 22475 |
+
"epoch": 0.63,
|
| 22476 |
+
"grad_norm": 1.9219965934753418,
|
| 22477 |
+
"learning_rate": 1.828400612678492e-05,
|
| 22478 |
+
"loss": 5.508,
|
| 22479 |
+
"step": 12840
|
| 22480 |
+
},
|
| 22481 |
+
{
|
| 22482 |
+
"epoch": 0.63,
|
| 22483 |
+
"grad_norm": 1.9963774681091309,
|
| 22484 |
+
"learning_rate": 1.8274124215623303e-05,
|
| 22485 |
+
"loss": 5.664,
|
| 22486 |
+
"step": 12844
|
| 22487 |
+
},
|
| 22488 |
+
{
|
| 22489 |
+
"epoch": 0.63,
|
| 22490 |
+
"grad_norm": 2.181628942489624,
|
| 22491 |
+
"learning_rate": 1.8264242304461685e-05,
|
| 22492 |
+
"loss": 5.4549,
|
| 22493 |
+
"step": 12848
|
| 22494 |
+
},
|
| 22495 |
+
{
|
| 22496 |
+
"epoch": 0.64,
|
| 22497 |
+
"grad_norm": 1.9105952978134155,
|
| 22498 |
+
"learning_rate": 1.8254360393300067e-05,
|
| 22499 |
+
"loss": 5.6288,
|
| 22500 |
+
"step": 12852
|
| 22501 |
+
},
|
| 22502 |
+
{
|
| 22503 |
+
"epoch": 0.64,
|
| 22504 |
+
"grad_norm": 1.8461229801177979,
|
| 22505 |
+
"learning_rate": 1.8244478482138446e-05,
|
| 22506 |
+
"loss": 5.5195,
|
| 22507 |
+
"step": 12856
|
| 22508 |
+
},
|
| 22509 |
+
{
|
| 22510 |
+
"epoch": 0.64,
|
| 22511 |
+
"grad_norm": 1.9446773529052734,
|
| 22512 |
+
"learning_rate": 1.8234596570976828e-05,
|
| 22513 |
+
"loss": 5.4313,
|
| 22514 |
+
"step": 12860
|
| 22515 |
+
},
|
| 22516 |
+
{
|
| 22517 |
+
"epoch": 0.64,
|
| 22518 |
+
"grad_norm": 2.007297992706299,
|
| 22519 |
+
"learning_rate": 1.822471465981521e-05,
|
| 22520 |
+
"loss": 5.6664,
|
| 22521 |
+
"step": 12864
|
| 22522 |
+
},
|
| 22523 |
+
{
|
| 22524 |
+
"epoch": 0.64,
|
| 22525 |
+
"grad_norm": 2.0537166595458984,
|
| 22526 |
+
"learning_rate": 1.821483274865359e-05,
|
| 22527 |
+
"loss": 5.5447,
|
| 22528 |
+
"step": 12868
|
| 22529 |
+
},
|
| 22530 |
+
{
|
| 22531 |
+
"epoch": 0.64,
|
| 22532 |
+
"grad_norm": 2.0639407634735107,
|
| 22533 |
+
"learning_rate": 1.820495083749197e-05,
|
| 22534 |
+
"loss": 5.4532,
|
| 22535 |
+
"step": 12872
|
| 22536 |
+
},
|
| 22537 |
+
{
|
| 22538 |
+
"epoch": 0.64,
|
| 22539 |
+
"grad_norm": 1.9469093084335327,
|
| 22540 |
+
"learning_rate": 1.8195068926330353e-05,
|
| 22541 |
+
"loss": 5.5461,
|
| 22542 |
+
"step": 12876
|
| 22543 |
+
},
|
| 22544 |
+
{
|
| 22545 |
+
"epoch": 0.64,
|
| 22546 |
+
"grad_norm": 1.766298770904541,
|
| 22547 |
+
"learning_rate": 1.8185187015168735e-05,
|
| 22548 |
+
"loss": 5.4775,
|
| 22549 |
+
"step": 12880
|
| 22550 |
+
},
|
| 22551 |
+
{
|
| 22552 |
+
"epoch": 0.64,
|
| 22553 |
+
"grad_norm": 1.7954472303390503,
|
| 22554 |
+
"learning_rate": 1.8175305104007114e-05,
|
| 22555 |
+
"loss": 5.5326,
|
| 22556 |
+
"step": 12884
|
| 22557 |
+
},
|
| 22558 |
+
{
|
| 22559 |
+
"epoch": 0.64,
|
| 22560 |
+
"grad_norm": 1.8361804485321045,
|
| 22561 |
+
"learning_rate": 1.8165423192845496e-05,
|
| 22562 |
+
"loss": 5.5217,
|
| 22563 |
+
"step": 12888
|
| 22564 |
+
},
|
| 22565 |
+
{
|
| 22566 |
+
"epoch": 0.64,
|
| 22567 |
+
"grad_norm": 2.2192041873931885,
|
| 22568 |
+
"learning_rate": 1.815554128168388e-05,
|
| 22569 |
+
"loss": 5.4166,
|
| 22570 |
+
"step": 12892
|
| 22571 |
+
},
|
| 22572 |
+
{
|
| 22573 |
+
"epoch": 0.64,
|
| 22574 |
+
"grad_norm": 2.092569351196289,
|
| 22575 |
+
"learning_rate": 1.814565937052226e-05,
|
| 22576 |
+
"loss": 5.5973,
|
| 22577 |
+
"step": 12896
|
| 22578 |
+
},
|
| 22579 |
+
{
|
| 22580 |
+
"epoch": 0.64,
|
| 22581 |
+
"grad_norm": 1.8529601097106934,
|
| 22582 |
+
"learning_rate": 1.8135777459360642e-05,
|
| 22583 |
+
"loss": 5.4088,
|
| 22584 |
+
"step": 12900
|
| 22585 |
+
},
|
| 22586 |
+
{
|
| 22587 |
+
"epoch": 0.64,
|
| 22588 |
+
"grad_norm": 1.7156609296798706,
|
| 22589 |
+
"learning_rate": 1.8125895548199024e-05,
|
| 22590 |
+
"loss": 5.5506,
|
| 22591 |
+
"step": 12904
|
| 22592 |
+
},
|
| 22593 |
+
{
|
| 22594 |
+
"epoch": 0.64,
|
| 22595 |
+
"grad_norm": 2.3277995586395264,
|
| 22596 |
+
"learning_rate": 1.8116013637037406e-05,
|
| 22597 |
+
"loss": 5.4852,
|
| 22598 |
+
"step": 12908
|
| 22599 |
+
},
|
| 22600 |
+
{
|
| 22601 |
+
"epoch": 0.64,
|
| 22602 |
+
"grad_norm": 2.039177894592285,
|
| 22603 |
+
"learning_rate": 1.8106131725875785e-05,
|
| 22604 |
+
"loss": 5.509,
|
| 22605 |
+
"step": 12912
|
| 22606 |
+
},
|
| 22607 |
+
{
|
| 22608 |
+
"epoch": 0.64,
|
| 22609 |
+
"grad_norm": 2.2541489601135254,
|
| 22610 |
+
"learning_rate": 1.8096249814714167e-05,
|
| 22611 |
+
"loss": 5.5551,
|
| 22612 |
+
"step": 12916
|
| 22613 |
+
},
|
| 22614 |
+
{
|
| 22615 |
+
"epoch": 0.64,
|
| 22616 |
+
"grad_norm": 1.9068487882614136,
|
| 22617 |
+
"learning_rate": 1.808636790355255e-05,
|
| 22618 |
+
"loss": 5.5223,
|
| 22619 |
+
"step": 12920
|
| 22620 |
+
},
|
| 22621 |
+
{
|
| 22622 |
+
"epoch": 0.64,
|
| 22623 |
+
"grad_norm": 1.8464723825454712,
|
| 22624 |
+
"learning_rate": 1.807648599239093e-05,
|
| 22625 |
+
"loss": 5.4688,
|
| 22626 |
+
"step": 12924
|
| 22627 |
+
},
|
| 22628 |
+
{
|
| 22629 |
+
"epoch": 0.64,
|
| 22630 |
+
"grad_norm": 1.8630855083465576,
|
| 22631 |
+
"learning_rate": 1.806660408122931e-05,
|
| 22632 |
+
"loss": 5.521,
|
| 22633 |
+
"step": 12928
|
| 22634 |
+
},
|
| 22635 |
+
{
|
| 22636 |
+
"epoch": 0.64,
|
| 22637 |
+
"grad_norm": 1.9354201555252075,
|
| 22638 |
+
"learning_rate": 1.805672217006769e-05,
|
| 22639 |
+
"loss": 5.5234,
|
| 22640 |
+
"step": 12932
|
| 22641 |
+
},
|
| 22642 |
+
{
|
| 22643 |
+
"epoch": 0.64,
|
| 22644 |
+
"grad_norm": 2.2278544902801514,
|
| 22645 |
+
"learning_rate": 1.8046840258906074e-05,
|
| 22646 |
+
"loss": 5.563,
|
| 22647 |
+
"step": 12936
|
| 22648 |
+
},
|
| 22649 |
+
{
|
| 22650 |
+
"epoch": 0.64,
|
| 22651 |
+
"grad_norm": 2.0102896690368652,
|
| 22652 |
+
"learning_rate": 1.8036958347744452e-05,
|
| 22653 |
+
"loss": 5.6071,
|
| 22654 |
+
"step": 12940
|
| 22655 |
+
},
|
| 22656 |
+
{
|
| 22657 |
+
"epoch": 0.64,
|
| 22658 |
+
"grad_norm": 2.12506103515625,
|
| 22659 |
+
"learning_rate": 1.8027076436582834e-05,
|
| 22660 |
+
"loss": 5.5518,
|
| 22661 |
+
"step": 12944
|
| 22662 |
+
},
|
| 22663 |
+
{
|
| 22664 |
+
"epoch": 0.64,
|
| 22665 |
+
"grad_norm": 2.134568929672241,
|
| 22666 |
+
"learning_rate": 1.8017194525421216e-05,
|
| 22667 |
+
"loss": 5.4289,
|
| 22668 |
+
"step": 12948
|
| 22669 |
+
},
|
| 22670 |
+
{
|
| 22671 |
+
"epoch": 0.64,
|
| 22672 |
+
"grad_norm": 1.866938829421997,
|
| 22673 |
+
"learning_rate": 1.80073126142596e-05,
|
| 22674 |
+
"loss": 5.4047,
|
| 22675 |
+
"step": 12952
|
| 22676 |
+
},
|
| 22677 |
+
{
|
| 22678 |
+
"epoch": 0.64,
|
| 22679 |
+
"grad_norm": 2.0489606857299805,
|
| 22680 |
+
"learning_rate": 1.799743070309798e-05,
|
| 22681 |
+
"loss": 5.4668,
|
| 22682 |
+
"step": 12956
|
| 22683 |
+
},
|
| 22684 |
+
{
|
| 22685 |
+
"epoch": 0.64,
|
| 22686 |
+
"grad_norm": 2.130350351333618,
|
| 22687 |
+
"learning_rate": 1.7987548791936363e-05,
|
| 22688 |
+
"loss": 5.599,
|
| 22689 |
+
"step": 12960
|
| 22690 |
+
},
|
| 22691 |
+
{
|
| 22692 |
+
"epoch": 0.64,
|
| 22693 |
+
"grad_norm": 2.1276466846466064,
|
| 22694 |
+
"learning_rate": 1.7977666880774745e-05,
|
| 22695 |
+
"loss": 5.3925,
|
| 22696 |
+
"step": 12964
|
| 22697 |
+
},
|
| 22698 |
+
{
|
| 22699 |
+
"epoch": 0.64,
|
| 22700 |
+
"grad_norm": 1.8620883226394653,
|
| 22701 |
+
"learning_rate": 1.7967784969613123e-05,
|
| 22702 |
+
"loss": 5.4918,
|
| 22703 |
+
"step": 12968
|
| 22704 |
+
},
|
| 22705 |
+
{
|
| 22706 |
+
"epoch": 0.64,
|
| 22707 |
+
"grad_norm": 1.7407227754592896,
|
| 22708 |
+
"learning_rate": 1.7957903058451505e-05,
|
| 22709 |
+
"loss": 5.5406,
|
| 22710 |
+
"step": 12972
|
| 22711 |
+
},
|
| 22712 |
+
{
|
| 22713 |
+
"epoch": 0.64,
|
| 22714 |
+
"grad_norm": 1.9221688508987427,
|
| 22715 |
+
"learning_rate": 1.7948021147289887e-05,
|
| 22716 |
+
"loss": 5.5111,
|
| 22717 |
+
"step": 12976
|
| 22718 |
+
},
|
| 22719 |
+
{
|
| 22720 |
+
"epoch": 0.64,
|
| 22721 |
+
"grad_norm": 2.0765016078948975,
|
| 22722 |
+
"learning_rate": 1.793813923612827e-05,
|
| 22723 |
+
"loss": 5.6063,
|
| 22724 |
+
"step": 12980
|
| 22725 |
+
},
|
| 22726 |
+
{
|
| 22727 |
+
"epoch": 0.64,
|
| 22728 |
+
"grad_norm": 2.040132999420166,
|
| 22729 |
+
"learning_rate": 1.7928257324966648e-05,
|
| 22730 |
+
"loss": 5.5904,
|
| 22731 |
+
"step": 12984
|
| 22732 |
+
},
|
| 22733 |
+
{
|
| 22734 |
+
"epoch": 0.64,
|
| 22735 |
+
"grad_norm": 1.8433319330215454,
|
| 22736 |
+
"learning_rate": 1.791837541380503e-05,
|
| 22737 |
+
"loss": 5.4806,
|
| 22738 |
+
"step": 12988
|
| 22739 |
+
},
|
| 22740 |
+
{
|
| 22741 |
+
"epoch": 0.64,
|
| 22742 |
+
"grad_norm": 2.065800428390503,
|
| 22743 |
+
"learning_rate": 1.7908493502643412e-05,
|
| 22744 |
+
"loss": 5.563,
|
| 22745 |
+
"step": 12992
|
| 22746 |
+
},
|
| 22747 |
+
{
|
| 22748 |
+
"epoch": 0.64,
|
| 22749 |
+
"grad_norm": 2.199831485748291,
|
| 22750 |
+
"learning_rate": 1.789861159148179e-05,
|
| 22751 |
+
"loss": 5.5233,
|
| 22752 |
+
"step": 12996
|
| 22753 |
+
},
|
| 22754 |
+
{
|
| 22755 |
+
"epoch": 0.64,
|
| 22756 |
+
"grad_norm": 1.8309836387634277,
|
| 22757 |
+
"learning_rate": 1.7888729680320173e-05,
|
| 22758 |
+
"loss": 5.4911,
|
| 22759 |
+
"step": 13000
|
| 22760 |
+
},
|
| 22761 |
+
{
|
| 22762 |
+
"epoch": 0.64,
|
| 22763 |
+
"grad_norm": 1.900347113609314,
|
| 22764 |
+
"learning_rate": 1.7878847769158555e-05,
|
| 22765 |
+
"loss": 5.5083,
|
| 22766 |
+
"step": 13004
|
| 22767 |
+
},
|
| 22768 |
+
{
|
| 22769 |
+
"epoch": 0.64,
|
| 22770 |
+
"grad_norm": 1.8298702239990234,
|
| 22771 |
+
"learning_rate": 1.786896585799694e-05,
|
| 22772 |
+
"loss": 5.5197,
|
| 22773 |
+
"step": 13008
|
| 22774 |
+
},
|
| 22775 |
+
{
|
| 22776 |
+
"epoch": 0.64,
|
| 22777 |
+
"grad_norm": 1.8966771364212036,
|
| 22778 |
+
"learning_rate": 1.785908394683532e-05,
|
| 22779 |
+
"loss": 5.4034,
|
| 22780 |
+
"step": 13012
|
| 22781 |
+
},
|
| 22782 |
+
{
|
| 22783 |
+
"epoch": 0.64,
|
| 22784 |
+
"grad_norm": 1.9861708879470825,
|
| 22785 |
+
"learning_rate": 1.78492020356737e-05,
|
| 22786 |
+
"loss": 5.3773,
|
| 22787 |
+
"step": 13016
|
| 22788 |
+
},
|
| 22789 |
+
{
|
| 22790 |
+
"epoch": 0.64,
|
| 22791 |
+
"grad_norm": 2.1633896827697754,
|
| 22792 |
+
"learning_rate": 1.7839320124512083e-05,
|
| 22793 |
+
"loss": 5.5735,
|
| 22794 |
+
"step": 13020
|
| 22795 |
+
},
|
| 22796 |
+
{
|
| 22797 |
+
"epoch": 0.64,
|
| 22798 |
+
"grad_norm": 1.8944169282913208,
|
| 22799 |
+
"learning_rate": 1.7829438213350462e-05,
|
| 22800 |
+
"loss": 5.4982,
|
| 22801 |
+
"step": 13024
|
| 22802 |
+
},
|
| 22803 |
+
{
|
| 22804 |
+
"epoch": 0.64,
|
| 22805 |
+
"grad_norm": 2.358996868133545,
|
| 22806 |
+
"learning_rate": 1.7819556302188844e-05,
|
| 22807 |
+
"loss": 5.5851,
|
| 22808 |
+
"step": 13028
|
| 22809 |
+
},
|
| 22810 |
+
{
|
| 22811 |
+
"epoch": 0.64,
|
| 22812 |
+
"grad_norm": 1.8002705574035645,
|
| 22813 |
+
"learning_rate": 1.7809674391027226e-05,
|
| 22814 |
+
"loss": 5.5903,
|
| 22815 |
+
"step": 13032
|
| 22816 |
+
},
|
| 22817 |
+
{
|
| 22818 |
+
"epoch": 0.64,
|
| 22819 |
+
"grad_norm": 2.174081325531006,
|
| 22820 |
+
"learning_rate": 1.7799792479865608e-05,
|
| 22821 |
+
"loss": 5.5327,
|
| 22822 |
+
"step": 13036
|
| 22823 |
+
},
|
| 22824 |
+
{
|
| 22825 |
+
"epoch": 0.64,
|
| 22826 |
+
"grad_norm": 2.115267515182495,
|
| 22827 |
+
"learning_rate": 1.7789910568703987e-05,
|
| 22828 |
+
"loss": 5.6085,
|
| 22829 |
+
"step": 13040
|
| 22830 |
+
},
|
| 22831 |
+
{
|
| 22832 |
+
"epoch": 0.64,
|
| 22833 |
+
"grad_norm": 2.197908878326416,
|
| 22834 |
+
"learning_rate": 1.778002865754237e-05,
|
| 22835 |
+
"loss": 5.5499,
|
| 22836 |
+
"step": 13044
|
| 22837 |
+
},
|
| 22838 |
+
{
|
| 22839 |
+
"epoch": 0.64,
|
| 22840 |
+
"grad_norm": 2.2714781761169434,
|
| 22841 |
+
"learning_rate": 1.777014674638075e-05,
|
| 22842 |
+
"loss": 5.6221,
|
| 22843 |
+
"step": 13048
|
| 22844 |
+
},
|
| 22845 |
+
{
|
| 22846 |
+
"epoch": 0.64,
|
| 22847 |
+
"grad_norm": 2.1368441581726074,
|
| 22848 |
+
"learning_rate": 1.7760264835219133e-05,
|
| 22849 |
+
"loss": 5.5613,
|
| 22850 |
+
"step": 13052
|
| 22851 |
+
},
|
| 22852 |
+
{
|
| 22853 |
+
"epoch": 0.65,
|
| 22854 |
+
"grad_norm": 2.0283219814300537,
|
| 22855 |
+
"learning_rate": 1.775038292405751e-05,
|
| 22856 |
+
"loss": 5.4734,
|
| 22857 |
+
"step": 13056
|
| 22858 |
+
},
|
| 22859 |
+
{
|
| 22860 |
+
"epoch": 0.65,
|
| 22861 |
+
"grad_norm": 1.7801271677017212,
|
| 22862 |
+
"learning_rate": 1.7740501012895894e-05,
|
| 22863 |
+
"loss": 5.4928,
|
| 22864 |
+
"step": 13060
|
| 22865 |
+
},
|
| 22866 |
+
{
|
| 22867 |
+
"epoch": 0.65,
|
| 22868 |
+
"grad_norm": 1.8141615390777588,
|
| 22869 |
+
"learning_rate": 1.7730619101734276e-05,
|
| 22870 |
+
"loss": 5.4943,
|
| 22871 |
+
"step": 13064
|
| 22872 |
+
},
|
| 22873 |
+
{
|
| 22874 |
+
"epoch": 0.65,
|
| 22875 |
+
"grad_norm": 1.9424062967300415,
|
| 22876 |
+
"learning_rate": 1.7720737190572658e-05,
|
| 22877 |
+
"loss": 5.4888,
|
| 22878 |
+
"step": 13068
|
| 22879 |
+
},
|
| 22880 |
+
{
|
| 22881 |
+
"epoch": 0.65,
|
| 22882 |
+
"grad_norm": 1.7956730127334595,
|
| 22883 |
+
"learning_rate": 1.771085527941104e-05,
|
| 22884 |
+
"loss": 5.3367,
|
| 22885 |
+
"step": 13072
|
| 22886 |
+
},
|
| 22887 |
+
{
|
| 22888 |
+
"epoch": 0.65,
|
| 22889 |
+
"grad_norm": 1.8769028186798096,
|
| 22890 |
+
"learning_rate": 1.7700973368249422e-05,
|
| 22891 |
+
"loss": 5.5365,
|
| 22892 |
+
"step": 13076
|
| 22893 |
+
},
|
| 22894 |
+
{
|
| 22895 |
+
"epoch": 0.65,
|
| 22896 |
+
"grad_norm": 1.8470765352249146,
|
| 22897 |
+
"learning_rate": 1.76910914570878e-05,
|
| 22898 |
+
"loss": 5.6281,
|
| 22899 |
+
"step": 13080
|
| 22900 |
+
},
|
| 22901 |
+
{
|
| 22902 |
+
"epoch": 0.65,
|
| 22903 |
+
"grad_norm": 1.9143494367599487,
|
| 22904 |
+
"learning_rate": 1.7681209545926183e-05,
|
| 22905 |
+
"loss": 5.403,
|
| 22906 |
+
"step": 13084
|
| 22907 |
+
},
|
| 22908 |
+
{
|
| 22909 |
+
"epoch": 0.65,
|
| 22910 |
+
"grad_norm": 1.8906290531158447,
|
| 22911 |
+
"learning_rate": 1.7671327634764565e-05,
|
| 22912 |
+
"loss": 5.4201,
|
| 22913 |
+
"step": 13088
|
| 22914 |
+
},
|
| 22915 |
+
{
|
| 22916 |
+
"epoch": 0.65,
|
| 22917 |
+
"grad_norm": 1.9759694337844849,
|
| 22918 |
+
"learning_rate": 1.7661445723602947e-05,
|
| 22919 |
+
"loss": 5.5,
|
| 22920 |
+
"step": 13092
|
| 22921 |
+
},
|
| 22922 |
+
{
|
| 22923 |
+
"epoch": 0.65,
|
| 22924 |
+
"grad_norm": 2.16597580909729,
|
| 22925 |
+
"learning_rate": 1.7651563812441325e-05,
|
| 22926 |
+
"loss": 5.4831,
|
| 22927 |
+
"step": 13096
|
| 22928 |
+
},
|
| 22929 |
+
{
|
| 22930 |
+
"epoch": 0.65,
|
| 22931 |
+
"grad_norm": 2.142273187637329,
|
| 22932 |
+
"learning_rate": 1.7641681901279708e-05,
|
| 22933 |
+
"loss": 5.5608,
|
| 22934 |
+
"step": 13100
|
| 22935 |
+
},
|
| 22936 |
+
{
|
| 22937 |
+
"epoch": 0.65,
|
| 22938 |
+
"grad_norm": 2.0143542289733887,
|
| 22939 |
+
"learning_rate": 1.763179999011809e-05,
|
| 22940 |
+
"loss": 5.4914,
|
| 22941 |
+
"step": 13104
|
| 22942 |
+
},
|
| 22943 |
+
{
|
| 22944 |
+
"epoch": 0.65,
|
| 22945 |
+
"grad_norm": 2.0241010189056396,
|
| 22946 |
+
"learning_rate": 1.762191807895647e-05,
|
| 22947 |
+
"loss": 5.5402,
|
| 22948 |
+
"step": 13108
|
| 22949 |
+
},
|
| 22950 |
+
{
|
| 22951 |
+
"epoch": 0.65,
|
| 22952 |
+
"grad_norm": 2.111691951751709,
|
| 22953 |
+
"learning_rate": 1.761203616779485e-05,
|
| 22954 |
+
"loss": 5.5417,
|
| 22955 |
+
"step": 13112
|
| 22956 |
+
},
|
| 22957 |
+
{
|
| 22958 |
+
"epoch": 0.65,
|
| 22959 |
+
"grad_norm": 2.0966546535491943,
|
| 22960 |
+
"learning_rate": 1.7602154256633232e-05,
|
| 22961 |
+
"loss": 5.4544,
|
| 22962 |
+
"step": 13116
|
| 22963 |
+
},
|
| 22964 |
+
{
|
| 22965 |
+
"epoch": 0.65,
|
| 22966 |
+
"grad_norm": 2.2063889503479004,
|
| 22967 |
+
"learning_rate": 1.7592272345471614e-05,
|
| 22968 |
+
"loss": 5.5447,
|
| 22969 |
+
"step": 13120
|
| 22970 |
+
},
|
| 22971 |
+
{
|
| 22972 |
+
"epoch": 0.65,
|
| 22973 |
+
"grad_norm": 2.0416488647460938,
|
| 22974 |
+
"learning_rate": 1.7582390434309997e-05,
|
| 22975 |
+
"loss": 5.56,
|
| 22976 |
+
"step": 13124
|
| 22977 |
+
},
|
| 22978 |
+
{
|
| 22979 |
+
"epoch": 0.65,
|
| 22980 |
+
"grad_norm": 2.1015825271606445,
|
| 22981 |
+
"learning_rate": 1.757250852314838e-05,
|
| 22982 |
+
"loss": 5.5632,
|
| 22983 |
+
"step": 13128
|
| 22984 |
+
},
|
| 22985 |
+
{
|
| 22986 |
+
"epoch": 0.65,
|
| 22987 |
+
"grad_norm": 2.154283046722412,
|
| 22988 |
+
"learning_rate": 1.756262661198676e-05,
|
| 22989 |
+
"loss": 5.4369,
|
| 22990 |
+
"step": 13132
|
| 22991 |
+
},
|
| 22992 |
+
{
|
| 22993 |
+
"epoch": 0.65,
|
| 22994 |
+
"grad_norm": 1.9613300561904907,
|
| 22995 |
+
"learning_rate": 1.7552744700825143e-05,
|
| 22996 |
+
"loss": 5.4493,
|
| 22997 |
+
"step": 13136
|
| 22998 |
+
},
|
| 22999 |
+
{
|
| 23000 |
+
"epoch": 0.65,
|
| 23001 |
+
"grad_norm": 1.936897873878479,
|
| 23002 |
+
"learning_rate": 1.754286278966352e-05,
|
| 23003 |
+
"loss": 5.6429,
|
| 23004 |
+
"step": 13140
|
| 23005 |
+
},
|
| 23006 |
+
{
|
| 23007 |
+
"epoch": 0.65,
|
| 23008 |
+
"grad_norm": 2.1174933910369873,
|
| 23009 |
+
"learning_rate": 1.7532980878501903e-05,
|
| 23010 |
+
"loss": 5.5222,
|
| 23011 |
+
"step": 13144
|
| 23012 |
+
},
|
| 23013 |
+
{
|
| 23014 |
+
"epoch": 0.65,
|
| 23015 |
+
"grad_norm": 1.8853952884674072,
|
| 23016 |
+
"learning_rate": 1.7523098967340285e-05,
|
| 23017 |
+
"loss": 5.5823,
|
| 23018 |
+
"step": 13148
|
| 23019 |
+
},
|
| 23020 |
+
{
|
| 23021 |
+
"epoch": 0.65,
|
| 23022 |
+
"grad_norm": 1.7702313661575317,
|
| 23023 |
+
"learning_rate": 1.7513217056178664e-05,
|
| 23024 |
+
"loss": 5.5205,
|
| 23025 |
+
"step": 13152
|
| 23026 |
+
},
|
| 23027 |
+
{
|
| 23028 |
+
"epoch": 0.65,
|
| 23029 |
+
"grad_norm": 1.8901150226593018,
|
| 23030 |
+
"learning_rate": 1.7503335145017046e-05,
|
| 23031 |
+
"loss": 5.5227,
|
| 23032 |
+
"step": 13156
|
| 23033 |
+
},
|
| 23034 |
+
{
|
| 23035 |
+
"epoch": 0.65,
|
| 23036 |
+
"grad_norm": 1.9898940324783325,
|
| 23037 |
+
"learning_rate": 1.7493453233855428e-05,
|
| 23038 |
+
"loss": 5.5064,
|
| 23039 |
+
"step": 13160
|
| 23040 |
+
},
|
| 23041 |
+
{
|
| 23042 |
+
"epoch": 0.65,
|
| 23043 |
+
"grad_norm": 1.8458938598632812,
|
| 23044 |
+
"learning_rate": 1.748357132269381e-05,
|
| 23045 |
+
"loss": 5.4032,
|
| 23046 |
+
"step": 13164
|
| 23047 |
+
},
|
| 23048 |
+
{
|
| 23049 |
+
"epoch": 0.65,
|
| 23050 |
+
"grad_norm": 2.1321139335632324,
|
| 23051 |
+
"learning_rate": 1.747368941153219e-05,
|
| 23052 |
+
"loss": 5.5491,
|
| 23053 |
+
"step": 13168
|
| 23054 |
+
},
|
| 23055 |
+
{
|
| 23056 |
+
"epoch": 0.65,
|
| 23057 |
+
"grad_norm": 2.055555820465088,
|
| 23058 |
+
"learning_rate": 1.746380750037057e-05,
|
| 23059 |
+
"loss": 5.4821,
|
| 23060 |
+
"step": 13172
|
| 23061 |
+
},
|
| 23062 |
+
{
|
| 23063 |
+
"epoch": 0.65,
|
| 23064 |
+
"grad_norm": 2.0164570808410645,
|
| 23065 |
+
"learning_rate": 1.7453925589208953e-05,
|
| 23066 |
+
"loss": 5.4772,
|
| 23067 |
+
"step": 13176
|
| 23068 |
+
},
|
| 23069 |
+
{
|
| 23070 |
+
"epoch": 0.65,
|
| 23071 |
+
"grad_norm": 1.8798964023590088,
|
| 23072 |
+
"learning_rate": 1.7444043678047335e-05,
|
| 23073 |
+
"loss": 5.6292,
|
| 23074 |
+
"step": 13180
|
| 23075 |
+
},
|
| 23076 |
+
{
|
| 23077 |
+
"epoch": 0.65,
|
| 23078 |
+
"grad_norm": 1.8820830583572388,
|
| 23079 |
+
"learning_rate": 1.7434161766885717e-05,
|
| 23080 |
+
"loss": 5.5503,
|
| 23081 |
+
"step": 13184
|
| 23082 |
+
},
|
| 23083 |
+
{
|
| 23084 |
+
"epoch": 0.65,
|
| 23085 |
+
"grad_norm": 1.768707513809204,
|
| 23086 |
+
"learning_rate": 1.74242798557241e-05,
|
| 23087 |
+
"loss": 5.5907,
|
| 23088 |
+
"step": 13188
|
| 23089 |
+
},
|
| 23090 |
+
{
|
| 23091 |
+
"epoch": 0.65,
|
| 23092 |
+
"grad_norm": 1.7819797992706299,
|
| 23093 |
+
"learning_rate": 1.741439794456248e-05,
|
| 23094 |
+
"loss": 5.3631,
|
| 23095 |
+
"step": 13192
|
| 23096 |
+
},
|
| 23097 |
+
{
|
| 23098 |
+
"epoch": 0.65,
|
| 23099 |
+
"grad_norm": 1.9888780117034912,
|
| 23100 |
+
"learning_rate": 1.740451603340086e-05,
|
| 23101 |
+
"loss": 5.5318,
|
| 23102 |
+
"step": 13196
|
| 23103 |
+
},
|
| 23104 |
+
{
|
| 23105 |
+
"epoch": 0.65,
|
| 23106 |
+
"grad_norm": 1.7733293771743774,
|
| 23107 |
+
"learning_rate": 1.7394634122239242e-05,
|
| 23108 |
+
"loss": 5.4794,
|
| 23109 |
+
"step": 13200
|
| 23110 |
+
},
|
| 23111 |
+
{
|
| 23112 |
+
"epoch": 0.65,
|
| 23113 |
+
"grad_norm": 2.08803653717041,
|
| 23114 |
+
"learning_rate": 1.7384752211077624e-05,
|
| 23115 |
+
"loss": 5.5392,
|
| 23116 |
+
"step": 13204
|
| 23117 |
+
},
|
| 23118 |
+
{
|
| 23119 |
+
"epoch": 0.65,
|
| 23120 |
+
"grad_norm": 1.7837083339691162,
|
| 23121 |
+
"learning_rate": 1.7374870299916006e-05,
|
| 23122 |
+
"loss": 5.5612,
|
| 23123 |
+
"step": 13208
|
| 23124 |
+
},
|
| 23125 |
+
{
|
| 23126 |
+
"epoch": 0.65,
|
| 23127 |
+
"grad_norm": 2.200601100921631,
|
| 23128 |
+
"learning_rate": 1.7364988388754385e-05,
|
| 23129 |
+
"loss": 5.4618,
|
| 23130 |
+
"step": 13212
|
| 23131 |
+
},
|
| 23132 |
+
{
|
| 23133 |
+
"epoch": 0.65,
|
| 23134 |
+
"grad_norm": 2.062946319580078,
|
| 23135 |
+
"learning_rate": 1.7355106477592767e-05,
|
| 23136 |
+
"loss": 5.4031,
|
| 23137 |
+
"step": 13216
|
| 23138 |
+
},
|
| 23139 |
+
{
|
| 23140 |
+
"epoch": 0.65,
|
| 23141 |
+
"grad_norm": 1.8929904699325562,
|
| 23142 |
+
"learning_rate": 1.734522456643115e-05,
|
| 23143 |
+
"loss": 5.6011,
|
| 23144 |
+
"step": 13220
|
| 23145 |
+
},
|
| 23146 |
+
{
|
| 23147 |
+
"epoch": 0.65,
|
| 23148 |
+
"grad_norm": 2.175036907196045,
|
| 23149 |
+
"learning_rate": 1.7335342655269528e-05,
|
| 23150 |
+
"loss": 5.5038,
|
| 23151 |
+
"step": 13224
|
| 23152 |
+
},
|
| 23153 |
+
{
|
| 23154 |
+
"epoch": 0.65,
|
| 23155 |
+
"grad_norm": 2.1107425689697266,
|
| 23156 |
+
"learning_rate": 1.732546074410791e-05,
|
| 23157 |
+
"loss": 5.4624,
|
| 23158 |
+
"step": 13228
|
| 23159 |
+
},
|
| 23160 |
+
{
|
| 23161 |
+
"epoch": 0.65,
|
| 23162 |
+
"grad_norm": 1.8881957530975342,
|
| 23163 |
+
"learning_rate": 1.7315578832946292e-05,
|
| 23164 |
+
"loss": 5.4579,
|
| 23165 |
+
"step": 13232
|
| 23166 |
+
},
|
| 23167 |
+
{
|
| 23168 |
+
"epoch": 0.65,
|
| 23169 |
+
"grad_norm": 1.9419997930526733,
|
| 23170 |
+
"learning_rate": 1.7305696921784674e-05,
|
| 23171 |
+
"loss": 5.5691,
|
| 23172 |
+
"step": 13236
|
| 23173 |
+
},
|
| 23174 |
+
{
|
| 23175 |
+
"epoch": 0.65,
|
| 23176 |
+
"grad_norm": 2.006504535675049,
|
| 23177 |
+
"learning_rate": 1.7295815010623056e-05,
|
| 23178 |
+
"loss": 5.604,
|
| 23179 |
+
"step": 13240
|
| 23180 |
+
},
|
| 23181 |
+
{
|
| 23182 |
+
"epoch": 0.65,
|
| 23183 |
+
"grad_norm": 1.9548697471618652,
|
| 23184 |
+
"learning_rate": 1.7285933099461438e-05,
|
| 23185 |
+
"loss": 5.5202,
|
| 23186 |
+
"step": 13244
|
| 23187 |
+
},
|
| 23188 |
+
{
|
| 23189 |
+
"epoch": 0.65,
|
| 23190 |
+
"grad_norm": 1.903361201286316,
|
| 23191 |
+
"learning_rate": 1.727605118829982e-05,
|
| 23192 |
+
"loss": 5.5459,
|
| 23193 |
+
"step": 13248
|
| 23194 |
+
},
|
| 23195 |
+
{
|
| 23196 |
+
"epoch": 0.65,
|
| 23197 |
+
"grad_norm": 1.9035142660140991,
|
| 23198 |
+
"learning_rate": 1.72661692771382e-05,
|
| 23199 |
+
"loss": 5.5067,
|
| 23200 |
+
"step": 13252
|
| 23201 |
+
},
|
| 23202 |
+
{
|
| 23203 |
+
"epoch": 0.65,
|
| 23204 |
+
"grad_norm": 1.9801918268203735,
|
| 23205 |
+
"learning_rate": 1.725628736597658e-05,
|
| 23206 |
+
"loss": 5.502,
|
| 23207 |
+
"step": 13256
|
| 23208 |
+
},
|
| 23209 |
+
{
|
| 23210 |
+
"epoch": 0.66,
|
| 23211 |
+
"grad_norm": 1.8996813297271729,
|
| 23212 |
+
"learning_rate": 1.7246405454814963e-05,
|
| 23213 |
+
"loss": 5.4709,
|
| 23214 |
+
"step": 13260
|
| 23215 |
+
},
|
| 23216 |
+
{
|
| 23217 |
+
"epoch": 0.66,
|
| 23218 |
+
"grad_norm": 1.903206467628479,
|
| 23219 |
+
"learning_rate": 1.7236523543653345e-05,
|
| 23220 |
+
"loss": 5.4252,
|
| 23221 |
+
"step": 13264
|
| 23222 |
+
},
|
| 23223 |
+
{
|
| 23224 |
+
"epoch": 0.66,
|
| 23225 |
+
"grad_norm": 2.280048370361328,
|
| 23226 |
+
"learning_rate": 1.7226641632491724e-05,
|
| 23227 |
+
"loss": 5.6236,
|
| 23228 |
+
"step": 13268
|
| 23229 |
+
},
|
| 23230 |
+
{
|
| 23231 |
+
"epoch": 0.66,
|
| 23232 |
+
"grad_norm": 1.916698932647705,
|
| 23233 |
+
"learning_rate": 1.7216759721330106e-05,
|
| 23234 |
+
"loss": 5.5417,
|
| 23235 |
+
"step": 13272
|
| 23236 |
+
},
|
| 23237 |
+
{
|
| 23238 |
+
"epoch": 0.66,
|
| 23239 |
+
"grad_norm": 1.8944514989852905,
|
| 23240 |
+
"learning_rate": 1.7206877810168488e-05,
|
| 23241 |
+
"loss": 5.5273,
|
| 23242 |
+
"step": 13276
|
| 23243 |
+
},
|
| 23244 |
+
{
|
| 23245 |
+
"epoch": 0.66,
|
| 23246 |
+
"grad_norm": 2.160426139831543,
|
| 23247 |
+
"learning_rate": 1.7196995899006866e-05,
|
| 23248 |
+
"loss": 5.4925,
|
| 23249 |
+
"step": 13280
|
| 23250 |
+
},
|
| 23251 |
+
{
|
| 23252 |
+
"epoch": 0.66,
|
| 23253 |
+
"grad_norm": 1.9661693572998047,
|
| 23254 |
+
"learning_rate": 1.718711398784525e-05,
|
| 23255 |
+
"loss": 5.5566,
|
| 23256 |
+
"step": 13284
|
| 23257 |
+
},
|
| 23258 |
+
{
|
| 23259 |
+
"epoch": 0.66,
|
| 23260 |
+
"grad_norm": 2.1579537391662598,
|
| 23261 |
+
"learning_rate": 1.717723207668363e-05,
|
| 23262 |
+
"loss": 5.5,
|
| 23263 |
+
"step": 13288
|
| 23264 |
+
},
|
| 23265 |
+
{
|
| 23266 |
+
"epoch": 0.66,
|
| 23267 |
+
"grad_norm": 1.8311907052993774,
|
| 23268 |
+
"learning_rate": 1.7167350165522013e-05,
|
| 23269 |
+
"loss": 5.4402,
|
| 23270 |
+
"step": 13292
|
| 23271 |
+
},
|
| 23272 |
+
{
|
| 23273 |
+
"epoch": 0.66,
|
| 23274 |
+
"grad_norm": 2.067732810974121,
|
| 23275 |
+
"learning_rate": 1.7157468254360395e-05,
|
| 23276 |
+
"loss": 5.5876,
|
| 23277 |
+
"step": 13296
|
| 23278 |
+
},
|
| 23279 |
+
{
|
| 23280 |
+
"epoch": 0.66,
|
| 23281 |
+
"grad_norm": 1.9030920267105103,
|
| 23282 |
+
"learning_rate": 1.7147586343198777e-05,
|
| 23283 |
+
"loss": 5.5544,
|
| 23284 |
+
"step": 13300
|
| 23285 |
+
},
|
| 23286 |
+
{
|
| 23287 |
+
"epoch": 0.66,
|
| 23288 |
+
"grad_norm": 1.9689793586730957,
|
| 23289 |
+
"learning_rate": 1.713770443203716e-05,
|
| 23290 |
+
"loss": 5.5876,
|
| 23291 |
+
"step": 13304
|
| 23292 |
+
},
|
| 23293 |
+
{
|
| 23294 |
+
"epoch": 0.66,
|
| 23295 |
+
"grad_norm": 2.319972038269043,
|
| 23296 |
+
"learning_rate": 1.7127822520875537e-05,
|
| 23297 |
+
"loss": 5.6179,
|
| 23298 |
+
"step": 13308
|
| 23299 |
+
},
|
| 23300 |
+
{
|
| 23301 |
+
"epoch": 0.66,
|
| 23302 |
+
"grad_norm": 2.05124568939209,
|
| 23303 |
+
"learning_rate": 1.711794060971392e-05,
|
| 23304 |
+
"loss": 5.4808,
|
| 23305 |
+
"step": 13312
|
| 23306 |
+
},
|
| 23307 |
+
{
|
| 23308 |
+
"epoch": 0.66,
|
| 23309 |
+
"grad_norm": 2.054259777069092,
|
| 23310 |
+
"learning_rate": 1.71080586985523e-05,
|
| 23311 |
+
"loss": 5.571,
|
| 23312 |
+
"step": 13316
|
| 23313 |
+
},
|
| 23314 |
+
{
|
| 23315 |
+
"epoch": 0.66,
|
| 23316 |
+
"grad_norm": 2.1697633266448975,
|
| 23317 |
+
"learning_rate": 1.7098176787390684e-05,
|
| 23318 |
+
"loss": 5.5687,
|
| 23319 |
+
"step": 13320
|
| 23320 |
+
},
|
| 23321 |
+
{
|
| 23322 |
+
"epoch": 0.66,
|
| 23323 |
+
"grad_norm": 2.158599853515625,
|
| 23324 |
+
"learning_rate": 1.7088294876229062e-05,
|
| 23325 |
+
"loss": 5.4535,
|
| 23326 |
+
"step": 13324
|
| 23327 |
+
},
|
| 23328 |
+
{
|
| 23329 |
+
"epoch": 0.66,
|
| 23330 |
+
"grad_norm": 2.263106107711792,
|
| 23331 |
+
"learning_rate": 1.7078412965067444e-05,
|
| 23332 |
+
"loss": 5.5662,
|
| 23333 |
+
"step": 13328
|
| 23334 |
+
},
|
| 23335 |
+
{
|
| 23336 |
+
"epoch": 0.66,
|
| 23337 |
+
"grad_norm": 1.9761734008789062,
|
| 23338 |
+
"learning_rate": 1.7068531053905826e-05,
|
| 23339 |
+
"loss": 5.4655,
|
| 23340 |
+
"step": 13332
|
| 23341 |
+
},
|
| 23342 |
+
{
|
| 23343 |
+
"epoch": 0.66,
|
| 23344 |
+
"grad_norm": 2.1491572856903076,
|
| 23345 |
+
"learning_rate": 1.705864914274421e-05,
|
| 23346 |
+
"loss": 5.5403,
|
| 23347 |
+
"step": 13336
|
| 23348 |
+
},
|
| 23349 |
+
{
|
| 23350 |
+
"epoch": 0.66,
|
| 23351 |
+
"grad_norm": 1.9614084959030151,
|
| 23352 |
+
"learning_rate": 1.7048767231582587e-05,
|
| 23353 |
+
"loss": 5.4883,
|
| 23354 |
+
"step": 13340
|
| 23355 |
+
},
|
| 23356 |
+
{
|
| 23357 |
+
"epoch": 0.66,
|
| 23358 |
+
"grad_norm": 2.1084208488464355,
|
| 23359 |
+
"learning_rate": 1.703888532042097e-05,
|
| 23360 |
+
"loss": 5.4915,
|
| 23361 |
+
"step": 13344
|
| 23362 |
+
},
|
| 23363 |
+
{
|
| 23364 |
+
"epoch": 0.66,
|
| 23365 |
+
"grad_norm": 1.9315608739852905,
|
| 23366 |
+
"learning_rate": 1.702900340925935e-05,
|
| 23367 |
+
"loss": 5.531,
|
| 23368 |
+
"step": 13348
|
| 23369 |
+
},
|
| 23370 |
+
{
|
| 23371 |
+
"epoch": 0.66,
|
| 23372 |
+
"grad_norm": 1.9300886392593384,
|
| 23373 |
+
"learning_rate": 1.7019121498097733e-05,
|
| 23374 |
+
"loss": 5.5294,
|
| 23375 |
+
"step": 13352
|
| 23376 |
+
},
|
| 23377 |
+
{
|
| 23378 |
+
"epoch": 0.66,
|
| 23379 |
+
"grad_norm": 1.9439579248428345,
|
| 23380 |
+
"learning_rate": 1.7009239586936115e-05,
|
| 23381 |
+
"loss": 5.5372,
|
| 23382 |
+
"step": 13356
|
| 23383 |
+
},
|
| 23384 |
+
{
|
| 23385 |
+
"epoch": 0.66,
|
| 23386 |
+
"grad_norm": 1.8347561359405518,
|
| 23387 |
+
"learning_rate": 1.6999357675774497e-05,
|
| 23388 |
+
"loss": 5.5304,
|
| 23389 |
+
"step": 13360
|
| 23390 |
+
},
|
| 23391 |
+
{
|
| 23392 |
+
"epoch": 0.66,
|
| 23393 |
+
"grad_norm": 1.8642383813858032,
|
| 23394 |
+
"learning_rate": 1.6989475764612876e-05,
|
| 23395 |
+
"loss": 5.3163,
|
| 23396 |
+
"step": 13364
|
| 23397 |
+
},
|
| 23398 |
+
{
|
| 23399 |
+
"epoch": 0.66,
|
| 23400 |
+
"grad_norm": 1.9651672840118408,
|
| 23401 |
+
"learning_rate": 1.6979593853451258e-05,
|
| 23402 |
+
"loss": 5.4271,
|
| 23403 |
+
"step": 13368
|
| 23404 |
+
},
|
| 23405 |
+
{
|
| 23406 |
+
"epoch": 0.66,
|
| 23407 |
+
"grad_norm": 2.146197557449341,
|
| 23408 |
+
"learning_rate": 1.696971194228964e-05,
|
| 23409 |
+
"loss": 5.6012,
|
| 23410 |
+
"step": 13372
|
| 23411 |
+
},
|
| 23412 |
+
{
|
| 23413 |
+
"epoch": 0.66,
|
| 23414 |
+
"grad_norm": 1.9717754125595093,
|
| 23415 |
+
"learning_rate": 1.6959830031128022e-05,
|
| 23416 |
+
"loss": 5.4703,
|
| 23417 |
+
"step": 13376
|
| 23418 |
+
},
|
| 23419 |
+
{
|
| 23420 |
+
"epoch": 0.66,
|
| 23421 |
+
"grad_norm": 1.8049124479293823,
|
| 23422 |
+
"learning_rate": 1.69499481199664e-05,
|
| 23423 |
+
"loss": 5.4398,
|
| 23424 |
+
"step": 13380
|
| 23425 |
+
},
|
| 23426 |
+
{
|
| 23427 |
+
"epoch": 0.66,
|
| 23428 |
+
"grad_norm": 1.8642454147338867,
|
| 23429 |
+
"learning_rate": 1.6940066208804783e-05,
|
| 23430 |
+
"loss": 5.5652,
|
| 23431 |
+
"step": 13384
|
| 23432 |
+
},
|
| 23433 |
+
{
|
| 23434 |
+
"epoch": 0.66,
|
| 23435 |
+
"grad_norm": 1.8307348489761353,
|
| 23436 |
+
"learning_rate": 1.6930184297643165e-05,
|
| 23437 |
+
"loss": 5.6079,
|
| 23438 |
+
"step": 13388
|
| 23439 |
+
},
|
| 23440 |
+
{
|
| 23441 |
+
"epoch": 0.66,
|
| 23442 |
+
"grad_norm": 1.8231866359710693,
|
| 23443 |
+
"learning_rate": 1.6920302386481547e-05,
|
| 23444 |
+
"loss": 5.4477,
|
| 23445 |
+
"step": 13392
|
| 23446 |
+
},
|
| 23447 |
+
{
|
| 23448 |
+
"epoch": 0.66,
|
| 23449 |
+
"grad_norm": 1.7777388095855713,
|
| 23450 |
+
"learning_rate": 1.6910420475319926e-05,
|
| 23451 |
+
"loss": 5.5014,
|
| 23452 |
+
"step": 13396
|
| 23453 |
+
},
|
| 23454 |
+
{
|
| 23455 |
+
"epoch": 0.66,
|
| 23456 |
+
"grad_norm": 2.1190943717956543,
|
| 23457 |
+
"learning_rate": 1.6900538564158308e-05,
|
| 23458 |
+
"loss": 5.512,
|
| 23459 |
+
"step": 13400
|
| 23460 |
+
},
|
| 23461 |
+
{
|
| 23462 |
+
"epoch": 0.66,
|
| 23463 |
+
"grad_norm": 1.7859139442443848,
|
| 23464 |
+
"learning_rate": 1.689065665299669e-05,
|
| 23465 |
+
"loss": 5.4176,
|
| 23466 |
+
"step": 13404
|
| 23467 |
+
},
|
| 23468 |
+
{
|
| 23469 |
+
"epoch": 0.66,
|
| 23470 |
+
"grad_norm": 1.937577486038208,
|
| 23471 |
+
"learning_rate": 1.6880774741835072e-05,
|
| 23472 |
+
"loss": 5.5242,
|
| 23473 |
+
"step": 13408
|
| 23474 |
+
},
|
| 23475 |
+
{
|
| 23476 |
+
"epoch": 0.66,
|
| 23477 |
+
"grad_norm": 1.9069881439208984,
|
| 23478 |
+
"learning_rate": 1.6870892830673454e-05,
|
| 23479 |
+
"loss": 5.5248,
|
| 23480 |
+
"step": 13412
|
| 23481 |
+
},
|
| 23482 |
+
{
|
| 23483 |
+
"epoch": 0.66,
|
| 23484 |
+
"grad_norm": 2.3434953689575195,
|
| 23485 |
+
"learning_rate": 1.6861010919511836e-05,
|
| 23486 |
+
"loss": 5.5887,
|
| 23487 |
+
"step": 13416
|
| 23488 |
+
},
|
| 23489 |
+
{
|
| 23490 |
+
"epoch": 0.66,
|
| 23491 |
+
"grad_norm": 2.0236103534698486,
|
| 23492 |
+
"learning_rate": 1.6851129008350218e-05,
|
| 23493 |
+
"loss": 5.5241,
|
| 23494 |
+
"step": 13420
|
| 23495 |
+
},
|
| 23496 |
+
{
|
| 23497 |
+
"epoch": 0.66,
|
| 23498 |
+
"grad_norm": 1.807458519935608,
|
| 23499 |
+
"learning_rate": 1.6841247097188597e-05,
|
| 23500 |
+
"loss": 5.5018,
|
| 23501 |
+
"step": 13424
|
| 23502 |
+
},
|
| 23503 |
+
{
|
| 23504 |
+
"epoch": 0.66,
|
| 23505 |
+
"grad_norm": 1.7654306888580322,
|
| 23506 |
+
"learning_rate": 1.683136518602698e-05,
|
| 23507 |
+
"loss": 5.5346,
|
| 23508 |
+
"step": 13428
|
| 23509 |
+
},
|
| 23510 |
+
{
|
| 23511 |
+
"epoch": 0.66,
|
| 23512 |
+
"grad_norm": 1.796311378479004,
|
| 23513 |
+
"learning_rate": 1.682148327486536e-05,
|
| 23514 |
+
"loss": 5.4943,
|
| 23515 |
+
"step": 13432
|
| 23516 |
+
},
|
| 23517 |
+
{
|
| 23518 |
+
"epoch": 0.66,
|
| 23519 |
+
"grad_norm": 2.1759448051452637,
|
| 23520 |
+
"learning_rate": 1.681160136370374e-05,
|
| 23521 |
+
"loss": 5.5745,
|
| 23522 |
+
"step": 13436
|
| 23523 |
+
},
|
| 23524 |
+
{
|
| 23525 |
+
"epoch": 0.66,
|
| 23526 |
+
"grad_norm": 1.9799151420593262,
|
| 23527 |
+
"learning_rate": 1.680171945254212e-05,
|
| 23528 |
+
"loss": 5.5721,
|
| 23529 |
+
"step": 13440
|
| 23530 |
+
},
|
| 23531 |
+
{
|
| 23532 |
+
"epoch": 0.66,
|
| 23533 |
+
"grad_norm": 1.8375133275985718,
|
| 23534 |
+
"learning_rate": 1.6791837541380504e-05,
|
| 23535 |
+
"loss": 5.5058,
|
| 23536 |
+
"step": 13444
|
| 23537 |
+
},
|
| 23538 |
+
{
|
| 23539 |
+
"epoch": 0.66,
|
| 23540 |
+
"grad_norm": 2.0697977542877197,
|
| 23541 |
+
"learning_rate": 1.6781955630218886e-05,
|
| 23542 |
+
"loss": 5.49,
|
| 23543 |
+
"step": 13448
|
| 23544 |
+
},
|
| 23545 |
+
{
|
| 23546 |
+
"epoch": 0.66,
|
| 23547 |
+
"grad_norm": 1.968781590461731,
|
| 23548 |
+
"learning_rate": 1.6772073719057264e-05,
|
| 23549 |
+
"loss": 5.414,
|
| 23550 |
+
"step": 13452
|
| 23551 |
+
},
|
| 23552 |
+
{
|
| 23553 |
+
"epoch": 0.66,
|
| 23554 |
+
"grad_norm": 2.182166814804077,
|
| 23555 |
+
"learning_rate": 1.6762191807895647e-05,
|
| 23556 |
+
"loss": 5.6226,
|
| 23557 |
+
"step": 13456
|
| 23558 |
+
},
|
| 23559 |
+
{
|
| 23560 |
+
"epoch": 0.67,
|
| 23561 |
+
"grad_norm": 2.110377073287964,
|
| 23562 |
+
"learning_rate": 1.675230989673403e-05,
|
| 23563 |
+
"loss": 5.5626,
|
| 23564 |
+
"step": 13460
|
| 23565 |
+
},
|
| 23566 |
+
{
|
| 23567 |
+
"epoch": 0.67,
|
| 23568 |
+
"grad_norm": 1.979601263999939,
|
| 23569 |
+
"learning_rate": 1.674242798557241e-05,
|
| 23570 |
+
"loss": 5.62,
|
| 23571 |
+
"step": 13464
|
| 23572 |
+
},
|
| 23573 |
+
{
|
| 23574 |
+
"epoch": 0.67,
|
| 23575 |
+
"grad_norm": 2.2129158973693848,
|
| 23576 |
+
"learning_rate": 1.6732546074410793e-05,
|
| 23577 |
+
"loss": 5.5565,
|
| 23578 |
+
"step": 13468
|
| 23579 |
+
},
|
| 23580 |
+
{
|
| 23581 |
+
"epoch": 0.67,
|
| 23582 |
+
"grad_norm": 2.0401570796966553,
|
| 23583 |
+
"learning_rate": 1.6722664163249175e-05,
|
| 23584 |
+
"loss": 5.4781,
|
| 23585 |
+
"step": 13472
|
| 23586 |
+
},
|
| 23587 |
+
{
|
| 23588 |
+
"epoch": 0.67,
|
| 23589 |
+
"grad_norm": 1.7625924348831177,
|
| 23590 |
+
"learning_rate": 1.6712782252087557e-05,
|
| 23591 |
+
"loss": 5.4104,
|
| 23592 |
+
"step": 13476
|
| 23593 |
+
},
|
| 23594 |
+
{
|
| 23595 |
+
"epoch": 0.67,
|
| 23596 |
+
"grad_norm": 1.8981072902679443,
|
| 23597 |
+
"learning_rate": 1.6702900340925936e-05,
|
| 23598 |
+
"loss": 5.5208,
|
| 23599 |
+
"step": 13480
|
| 23600 |
+
},
|
| 23601 |
+
{
|
| 23602 |
+
"epoch": 0.67,
|
| 23603 |
+
"grad_norm": 2.141097068786621,
|
| 23604 |
+
"learning_rate": 1.6693018429764318e-05,
|
| 23605 |
+
"loss": 5.505,
|
| 23606 |
+
"step": 13484
|
| 23607 |
+
},
|
| 23608 |
+
{
|
| 23609 |
+
"epoch": 0.67,
|
| 23610 |
+
"grad_norm": 2.0228660106658936,
|
| 23611 |
+
"learning_rate": 1.66831365186027e-05,
|
| 23612 |
+
"loss": 5.5653,
|
| 23613 |
+
"step": 13488
|
| 23614 |
+
},
|
| 23615 |
+
{
|
| 23616 |
+
"epoch": 0.67,
|
| 23617 |
+
"grad_norm": 1.9627779722213745,
|
| 23618 |
+
"learning_rate": 1.667325460744108e-05,
|
| 23619 |
+
"loss": 5.5383,
|
| 23620 |
+
"step": 13492
|
| 23621 |
+
},
|
| 23622 |
+
{
|
| 23623 |
+
"epoch": 0.67,
|
| 23624 |
+
"grad_norm": 1.80388605594635,
|
| 23625 |
+
"learning_rate": 1.666337269627946e-05,
|
| 23626 |
+
"loss": 5.4095,
|
| 23627 |
+
"step": 13496
|
| 23628 |
+
},
|
| 23629 |
+
{
|
| 23630 |
+
"epoch": 0.67,
|
| 23631 |
+
"grad_norm": 2.125562906265259,
|
| 23632 |
+
"learning_rate": 1.6653490785117842e-05,
|
| 23633 |
+
"loss": 5.5157,
|
| 23634 |
+
"step": 13500
|
| 23635 |
+
},
|
| 23636 |
+
{
|
| 23637 |
+
"epoch": 0.67,
|
| 23638 |
+
"grad_norm": 2.3643898963928223,
|
| 23639 |
+
"learning_rate": 1.6643608873956224e-05,
|
| 23640 |
+
"loss": 5.4778,
|
| 23641 |
+
"step": 13504
|
| 23642 |
+
},
|
| 23643 |
+
{
|
| 23644 |
+
"epoch": 0.67,
|
| 23645 |
+
"grad_norm": 1.9573434591293335,
|
| 23646 |
+
"learning_rate": 1.6633726962794603e-05,
|
| 23647 |
+
"loss": 5.3951,
|
| 23648 |
+
"step": 13508
|
| 23649 |
+
},
|
| 23650 |
+
{
|
| 23651 |
+
"epoch": 0.67,
|
| 23652 |
+
"grad_norm": 1.767749309539795,
|
| 23653 |
+
"learning_rate": 1.6623845051632985e-05,
|
| 23654 |
+
"loss": 5.5775,
|
| 23655 |
+
"step": 13512
|
| 23656 |
+
},
|
| 23657 |
+
{
|
| 23658 |
+
"epoch": 0.67,
|
| 23659 |
+
"grad_norm": 1.7242648601531982,
|
| 23660 |
+
"learning_rate": 1.6613963140471367e-05,
|
| 23661 |
+
"loss": 5.5681,
|
| 23662 |
+
"step": 13516
|
| 23663 |
+
},
|
| 23664 |
+
{
|
| 23665 |
+
"epoch": 0.67,
|
| 23666 |
+
"grad_norm": 2.098914384841919,
|
| 23667 |
+
"learning_rate": 1.660408122930975e-05,
|
| 23668 |
+
"loss": 5.4892,
|
| 23669 |
+
"step": 13520
|
| 23670 |
+
},
|
| 23671 |
+
{
|
| 23672 |
+
"epoch": 0.67,
|
| 23673 |
+
"grad_norm": 1.9471126794815063,
|
| 23674 |
+
"learning_rate": 1.659419931814813e-05,
|
| 23675 |
+
"loss": 5.5672,
|
| 23676 |
+
"step": 13524
|
| 23677 |
+
},
|
| 23678 |
+
{
|
| 23679 |
+
"epoch": 0.67,
|
| 23680 |
+
"grad_norm": 1.9518860578536987,
|
| 23681 |
+
"learning_rate": 1.6584317406986513e-05,
|
| 23682 |
+
"loss": 5.5318,
|
| 23683 |
+
"step": 13528
|
| 23684 |
+
},
|
| 23685 |
+
{
|
| 23686 |
+
"epoch": 0.67,
|
| 23687 |
+
"grad_norm": 1.9697927236557007,
|
| 23688 |
+
"learning_rate": 1.6574435495824896e-05,
|
| 23689 |
+
"loss": 5.5275,
|
| 23690 |
+
"step": 13532
|
| 23691 |
+
},
|
| 23692 |
+
{
|
| 23693 |
+
"epoch": 0.67,
|
| 23694 |
+
"grad_norm": 2.005502700805664,
|
| 23695 |
+
"learning_rate": 1.6564553584663274e-05,
|
| 23696 |
+
"loss": 5.6122,
|
| 23697 |
+
"step": 13536
|
| 23698 |
+
},
|
| 23699 |
+
{
|
| 23700 |
+
"epoch": 0.67,
|
| 23701 |
+
"grad_norm": 2.06354022026062,
|
| 23702 |
+
"learning_rate": 1.6554671673501656e-05,
|
| 23703 |
+
"loss": 5.3431,
|
| 23704 |
+
"step": 13540
|
| 23705 |
+
},
|
| 23706 |
+
{
|
| 23707 |
+
"epoch": 0.67,
|
| 23708 |
+
"grad_norm": 1.9697012901306152,
|
| 23709 |
+
"learning_rate": 1.6544789762340038e-05,
|
| 23710 |
+
"loss": 5.5376,
|
| 23711 |
+
"step": 13544
|
| 23712 |
+
},
|
| 23713 |
+
{
|
| 23714 |
+
"epoch": 0.67,
|
| 23715 |
+
"grad_norm": 1.9158987998962402,
|
| 23716 |
+
"learning_rate": 1.653490785117842e-05,
|
| 23717 |
+
"loss": 5.5176,
|
| 23718 |
+
"step": 13548
|
| 23719 |
+
},
|
| 23720 |
+
{
|
| 23721 |
+
"epoch": 0.67,
|
| 23722 |
+
"grad_norm": 2.0514771938323975,
|
| 23723 |
+
"learning_rate": 1.65250259400168e-05,
|
| 23724 |
+
"loss": 5.5724,
|
| 23725 |
+
"step": 13552
|
| 23726 |
+
},
|
| 23727 |
+
{
|
| 23728 |
+
"epoch": 0.67,
|
| 23729 |
+
"grad_norm": 1.8229410648345947,
|
| 23730 |
+
"learning_rate": 1.651514402885518e-05,
|
| 23731 |
+
"loss": 5.4933,
|
| 23732 |
+
"step": 13556
|
| 23733 |
+
},
|
| 23734 |
+
{
|
| 23735 |
+
"epoch": 0.67,
|
| 23736 |
+
"grad_norm": 1.6895649433135986,
|
| 23737 |
+
"learning_rate": 1.6505262117693563e-05,
|
| 23738 |
+
"loss": 5.5462,
|
| 23739 |
+
"step": 13560
|
| 23740 |
+
},
|
| 23741 |
+
{
|
| 23742 |
+
"epoch": 0.67,
|
| 23743 |
+
"grad_norm": 1.9636627435684204,
|
| 23744 |
+
"learning_rate": 1.6495380206531942e-05,
|
| 23745 |
+
"loss": 5.4918,
|
| 23746 |
+
"step": 13564
|
| 23747 |
+
},
|
| 23748 |
+
{
|
| 23749 |
+
"epoch": 0.67,
|
| 23750 |
+
"grad_norm": 1.728353500366211,
|
| 23751 |
+
"learning_rate": 1.6485498295370324e-05,
|
| 23752 |
+
"loss": 5.4126,
|
| 23753 |
+
"step": 13568
|
| 23754 |
+
},
|
| 23755 |
+
{
|
| 23756 |
+
"epoch": 0.67,
|
| 23757 |
+
"grad_norm": 1.8992080688476562,
|
| 23758 |
+
"learning_rate": 1.6475616384208706e-05,
|
| 23759 |
+
"loss": 5.4954,
|
| 23760 |
+
"step": 13572
|
| 23761 |
+
},
|
| 23762 |
+
{
|
| 23763 |
+
"epoch": 0.67,
|
| 23764 |
+
"grad_norm": 1.8971209526062012,
|
| 23765 |
+
"learning_rate": 1.6465734473047088e-05,
|
| 23766 |
+
"loss": 5.5205,
|
| 23767 |
+
"step": 13576
|
| 23768 |
+
},
|
| 23769 |
+
{
|
| 23770 |
+
"epoch": 0.67,
|
| 23771 |
+
"grad_norm": 1.9099971055984497,
|
| 23772 |
+
"learning_rate": 1.6455852561885467e-05,
|
| 23773 |
+
"loss": 5.5401,
|
| 23774 |
+
"step": 13580
|
| 23775 |
+
},
|
| 23776 |
+
{
|
| 23777 |
+
"epoch": 0.67,
|
| 23778 |
+
"grad_norm": 1.7198883295059204,
|
| 23779 |
+
"learning_rate": 1.6445970650723852e-05,
|
| 23780 |
+
"loss": 5.5187,
|
| 23781 |
+
"step": 13584
|
| 23782 |
+
},
|
| 23783 |
+
{
|
| 23784 |
+
"epoch": 0.67,
|
| 23785 |
+
"grad_norm": 1.854477882385254,
|
| 23786 |
+
"learning_rate": 1.6436088739562234e-05,
|
| 23787 |
+
"loss": 5.5558,
|
| 23788 |
+
"step": 13588
|
| 23789 |
+
},
|
| 23790 |
+
{
|
| 23791 |
+
"epoch": 0.67,
|
| 23792 |
+
"grad_norm": 1.9261908531188965,
|
| 23793 |
+
"learning_rate": 1.6426206828400613e-05,
|
| 23794 |
+
"loss": 5.5156,
|
| 23795 |
+
"step": 13592
|
| 23796 |
+
},
|
| 23797 |
+
{
|
| 23798 |
+
"epoch": 0.67,
|
| 23799 |
+
"grad_norm": 2.166208505630493,
|
| 23800 |
+
"learning_rate": 1.6416324917238995e-05,
|
| 23801 |
+
"loss": 5.5002,
|
| 23802 |
+
"step": 13596
|
| 23803 |
+
},
|
| 23804 |
+
{
|
| 23805 |
+
"epoch": 0.67,
|
| 23806 |
+
"grad_norm": 1.8934109210968018,
|
| 23807 |
+
"learning_rate": 1.6406443006077377e-05,
|
| 23808 |
+
"loss": 5.5092,
|
| 23809 |
+
"step": 13600
|
| 23810 |
+
},
|
| 23811 |
+
{
|
| 23812 |
+
"epoch": 0.67,
|
| 23813 |
+
"grad_norm": 1.830947756767273,
|
| 23814 |
+
"learning_rate": 1.639656109491576e-05,
|
| 23815 |
+
"loss": 5.5542,
|
| 23816 |
+
"step": 13604
|
| 23817 |
+
},
|
| 23818 |
+
{
|
| 23819 |
+
"epoch": 0.67,
|
| 23820 |
+
"grad_norm": 1.86097252368927,
|
| 23821 |
+
"learning_rate": 1.6386679183754138e-05,
|
| 23822 |
+
"loss": 5.502,
|
| 23823 |
+
"step": 13608
|
| 23824 |
+
},
|
| 23825 |
+
{
|
| 23826 |
+
"epoch": 0.67,
|
| 23827 |
+
"grad_norm": 1.9578551054000854,
|
| 23828 |
+
"learning_rate": 1.637679727259252e-05,
|
| 23829 |
+
"loss": 5.5351,
|
| 23830 |
+
"step": 13612
|
| 23831 |
+
},
|
| 23832 |
+
{
|
| 23833 |
+
"epoch": 0.67,
|
| 23834 |
+
"grad_norm": 1.9985218048095703,
|
| 23835 |
+
"learning_rate": 1.6366915361430902e-05,
|
| 23836 |
+
"loss": 5.403,
|
| 23837 |
+
"step": 13616
|
| 23838 |
+
},
|
| 23839 |
+
{
|
| 23840 |
+
"epoch": 0.67,
|
| 23841 |
+
"grad_norm": 2.346946954727173,
|
| 23842 |
+
"learning_rate": 1.6357033450269284e-05,
|
| 23843 |
+
"loss": 5.4314,
|
| 23844 |
+
"step": 13620
|
| 23845 |
+
},
|
| 23846 |
+
{
|
| 23847 |
+
"epoch": 0.67,
|
| 23848 |
+
"grad_norm": 1.9613852500915527,
|
| 23849 |
+
"learning_rate": 1.6347151539107663e-05,
|
| 23850 |
+
"loss": 5.4851,
|
| 23851 |
+
"step": 13624
|
| 23852 |
+
},
|
| 23853 |
+
{
|
| 23854 |
+
"epoch": 0.67,
|
| 23855 |
+
"grad_norm": 2.2844598293304443,
|
| 23856 |
+
"learning_rate": 1.6337269627946045e-05,
|
| 23857 |
+
"loss": 5.4646,
|
| 23858 |
+
"step": 13628
|
| 23859 |
+
},
|
| 23860 |
+
{
|
| 23861 |
+
"epoch": 0.67,
|
| 23862 |
+
"grad_norm": 2.0212109088897705,
|
| 23863 |
+
"learning_rate": 1.6327387716784427e-05,
|
| 23864 |
+
"loss": 5.5153,
|
| 23865 |
+
"step": 13632
|
| 23866 |
+
},
|
| 23867 |
+
{
|
| 23868 |
+
"epoch": 0.67,
|
| 23869 |
+
"grad_norm": 2.0245635509490967,
|
| 23870 |
+
"learning_rate": 1.6317505805622805e-05,
|
| 23871 |
+
"loss": 5.5516,
|
| 23872 |
+
"step": 13636
|
| 23873 |
+
},
|
| 23874 |
+
{
|
| 23875 |
+
"epoch": 0.67,
|
| 23876 |
+
"grad_norm": 2.051738977432251,
|
| 23877 |
+
"learning_rate": 1.630762389446119e-05,
|
| 23878 |
+
"loss": 5.6327,
|
| 23879 |
+
"step": 13640
|
| 23880 |
+
},
|
| 23881 |
+
{
|
| 23882 |
+
"epoch": 0.67,
|
| 23883 |
+
"grad_norm": 2.0657126903533936,
|
| 23884 |
+
"learning_rate": 1.6297741983299573e-05,
|
| 23885 |
+
"loss": 5.5882,
|
| 23886 |
+
"step": 13644
|
| 23887 |
+
},
|
| 23888 |
+
{
|
| 23889 |
+
"epoch": 0.67,
|
| 23890 |
+
"grad_norm": 1.979304313659668,
|
| 23891 |
+
"learning_rate": 1.628786007213795e-05,
|
| 23892 |
+
"loss": 5.3553,
|
| 23893 |
+
"step": 13648
|
| 23894 |
+
},
|
| 23895 |
+
{
|
| 23896 |
+
"epoch": 0.67,
|
| 23897 |
+
"grad_norm": 2.10304856300354,
|
| 23898 |
+
"learning_rate": 1.6277978160976334e-05,
|
| 23899 |
+
"loss": 5.5474,
|
| 23900 |
+
"step": 13652
|
| 23901 |
+
},
|
| 23902 |
+
{
|
| 23903 |
+
"epoch": 0.67,
|
| 23904 |
+
"grad_norm": 2.0482354164123535,
|
| 23905 |
+
"learning_rate": 1.6268096249814716e-05,
|
| 23906 |
+
"loss": 5.5004,
|
| 23907 |
+
"step": 13656
|
| 23908 |
+
},
|
| 23909 |
+
{
|
| 23910 |
+
"epoch": 0.67,
|
| 23911 |
+
"grad_norm": 2.0741443634033203,
|
| 23912 |
+
"learning_rate": 1.6258214338653098e-05,
|
| 23913 |
+
"loss": 5.4887,
|
| 23914 |
+
"step": 13660
|
| 23915 |
+
},
|
| 23916 |
+
{
|
| 23917 |
+
"epoch": 0.68,
|
| 23918 |
+
"grad_norm": 1.9850010871887207,
|
| 23919 |
+
"learning_rate": 1.6248332427491476e-05,
|
| 23920 |
+
"loss": 5.4449,
|
| 23921 |
+
"step": 13664
|
| 23922 |
+
},
|
| 23923 |
+
{
|
| 23924 |
+
"epoch": 0.68,
|
| 23925 |
+
"grad_norm": 2.456059217453003,
|
| 23926 |
+
"learning_rate": 1.623845051632986e-05,
|
| 23927 |
+
"loss": 5.4471,
|
| 23928 |
+
"step": 13668
|
| 23929 |
+
},
|
| 23930 |
+
{
|
| 23931 |
+
"epoch": 0.68,
|
| 23932 |
+
"grad_norm": 1.8275492191314697,
|
| 23933 |
+
"learning_rate": 1.622856860516824e-05,
|
| 23934 |
+
"loss": 5.5139,
|
| 23935 |
+
"step": 13672
|
| 23936 |
+
},
|
| 23937 |
+
{
|
| 23938 |
+
"epoch": 0.68,
|
| 23939 |
+
"grad_norm": 2.166975736618042,
|
| 23940 |
+
"learning_rate": 1.6218686694006623e-05,
|
| 23941 |
+
"loss": 5.5475,
|
| 23942 |
+
"step": 13676
|
| 23943 |
+
},
|
| 23944 |
+
{
|
| 23945 |
+
"epoch": 0.68,
|
| 23946 |
+
"grad_norm": 2.0723702907562256,
|
| 23947 |
+
"learning_rate": 1.6208804782845e-05,
|
| 23948 |
+
"loss": 5.6062,
|
| 23949 |
+
"step": 13680
|
| 23950 |
+
},
|
| 23951 |
+
{
|
| 23952 |
+
"epoch": 0.68,
|
| 23953 |
+
"grad_norm": 2.227518320083618,
|
| 23954 |
+
"learning_rate": 1.6198922871683383e-05,
|
| 23955 |
+
"loss": 5.4955,
|
| 23956 |
+
"step": 13684
|
| 23957 |
+
},
|
| 23958 |
+
{
|
| 23959 |
+
"epoch": 0.68,
|
| 23960 |
+
"grad_norm": 1.9052515029907227,
|
| 23961 |
+
"learning_rate": 1.6189040960521765e-05,
|
| 23962 |
+
"loss": 5.5553,
|
| 23963 |
+
"step": 13688
|
| 23964 |
+
},
|
| 23965 |
+
{
|
| 23966 |
+
"epoch": 0.68,
|
| 23967 |
+
"grad_norm": 1.945573329925537,
|
| 23968 |
+
"learning_rate": 1.6179159049360147e-05,
|
| 23969 |
+
"loss": 5.3597,
|
| 23970 |
+
"step": 13692
|
| 23971 |
+
},
|
| 23972 |
+
{
|
| 23973 |
+
"epoch": 0.68,
|
| 23974 |
+
"grad_norm": 1.874603271484375,
|
| 23975 |
+
"learning_rate": 1.616927713819853e-05,
|
| 23976 |
+
"loss": 5.391,
|
| 23977 |
+
"step": 13696
|
| 23978 |
+
},
|
| 23979 |
+
{
|
| 23980 |
+
"epoch": 0.68,
|
| 23981 |
+
"grad_norm": 2.339505672454834,
|
| 23982 |
+
"learning_rate": 1.615939522703691e-05,
|
| 23983 |
+
"loss": 5.4194,
|
| 23984 |
+
"step": 13700
|
| 23985 |
+
},
|
| 23986 |
+
{
|
| 23987 |
+
"epoch": 0.68,
|
| 23988 |
+
"grad_norm": 1.7062286138534546,
|
| 23989 |
+
"learning_rate": 1.6149513315875294e-05,
|
| 23990 |
+
"loss": 5.4586,
|
| 23991 |
+
"step": 13704
|
| 23992 |
+
},
|
| 23993 |
+
{
|
| 23994 |
+
"epoch": 0.68,
|
| 23995 |
+
"grad_norm": 1.8526628017425537,
|
| 23996 |
+
"learning_rate": 1.6139631404713672e-05,
|
| 23997 |
+
"loss": 5.3614,
|
| 23998 |
+
"step": 13708
|
| 23999 |
+
},
|
| 24000 |
+
{
|
| 24001 |
+
"epoch": 0.68,
|
| 24002 |
+
"grad_norm": 1.8179652690887451,
|
| 24003 |
+
"learning_rate": 1.6129749493552054e-05,
|
| 24004 |
+
"loss": 5.4977,
|
| 24005 |
+
"step": 13712
|
| 24006 |
+
},
|
| 24007 |
+
{
|
| 24008 |
+
"epoch": 0.68,
|
| 24009 |
+
"grad_norm": 1.889673113822937,
|
| 24010 |
+
"learning_rate": 1.6119867582390436e-05,
|
| 24011 |
+
"loss": 5.5591,
|
| 24012 |
+
"step": 13716
|
| 24013 |
+
},
|
| 24014 |
+
{
|
| 24015 |
+
"epoch": 0.68,
|
| 24016 |
+
"grad_norm": 2.039731740951538,
|
| 24017 |
+
"learning_rate": 1.6109985671228815e-05,
|
| 24018 |
+
"loss": 5.4809,
|
| 24019 |
+
"step": 13720
|
| 24020 |
+
},
|
| 24021 |
+
{
|
| 24022 |
+
"epoch": 0.68,
|
| 24023 |
+
"grad_norm": 2.3024516105651855,
|
| 24024 |
+
"learning_rate": 1.6100103760067197e-05,
|
| 24025 |
+
"loss": 5.4518,
|
| 24026 |
+
"step": 13724
|
| 24027 |
+
},
|
| 24028 |
+
{
|
| 24029 |
+
"epoch": 0.68,
|
| 24030 |
+
"grad_norm": 1.9836472272872925,
|
| 24031 |
+
"learning_rate": 1.609022184890558e-05,
|
| 24032 |
+
"loss": 5.4812,
|
| 24033 |
+
"step": 13728
|
| 24034 |
+
},
|
| 24035 |
+
{
|
| 24036 |
+
"epoch": 0.68,
|
| 24037 |
+
"grad_norm": 1.8883979320526123,
|
| 24038 |
+
"learning_rate": 1.608033993774396e-05,
|
| 24039 |
+
"loss": 5.4914,
|
| 24040 |
+
"step": 13732
|
| 24041 |
+
},
|
| 24042 |
+
{
|
| 24043 |
+
"epoch": 0.68,
|
| 24044 |
+
"grad_norm": 1.9850894212722778,
|
| 24045 |
+
"learning_rate": 1.607045802658234e-05,
|
| 24046 |
+
"loss": 5.6323,
|
| 24047 |
+
"step": 13736
|
| 24048 |
+
},
|
| 24049 |
+
{
|
| 24050 |
+
"epoch": 0.68,
|
| 24051 |
+
"grad_norm": 2.124187707901001,
|
| 24052 |
+
"learning_rate": 1.6060576115420722e-05,
|
| 24053 |
+
"loss": 5.5609,
|
| 24054 |
+
"step": 13740
|
| 24055 |
+
},
|
| 24056 |
+
{
|
| 24057 |
+
"epoch": 0.68,
|
| 24058 |
+
"grad_norm": 1.9509978294372559,
|
| 24059 |
+
"learning_rate": 1.6050694204259104e-05,
|
| 24060 |
+
"loss": 5.5234,
|
| 24061 |
+
"step": 13744
|
| 24062 |
+
},
|
| 24063 |
+
{
|
| 24064 |
+
"epoch": 0.68,
|
| 24065 |
+
"grad_norm": 1.9340142011642456,
|
| 24066 |
+
"learning_rate": 1.6040812293097486e-05,
|
| 24067 |
+
"loss": 5.491,
|
| 24068 |
+
"step": 13748
|
| 24069 |
+
},
|
| 24070 |
+
{
|
| 24071 |
+
"epoch": 0.68,
|
| 24072 |
+
"grad_norm": 2.0254647731781006,
|
| 24073 |
+
"learning_rate": 1.6030930381935865e-05,
|
| 24074 |
+
"loss": 5.4735,
|
| 24075 |
+
"step": 13752
|
| 24076 |
+
},
|
| 24077 |
+
{
|
| 24078 |
+
"epoch": 0.68,
|
| 24079 |
+
"grad_norm": 2.143260955810547,
|
| 24080 |
+
"learning_rate": 1.602104847077425e-05,
|
| 24081 |
+
"loss": 5.5477,
|
| 24082 |
+
"step": 13756
|
| 24083 |
+
},
|
| 24084 |
+
{
|
| 24085 |
+
"epoch": 0.68,
|
| 24086 |
+
"grad_norm": 1.6931935548782349,
|
| 24087 |
+
"learning_rate": 1.6011166559612632e-05,
|
| 24088 |
+
"loss": 5.5821,
|
| 24089 |
+
"step": 13760
|
| 24090 |
+
},
|
| 24091 |
+
{
|
| 24092 |
+
"epoch": 0.68,
|
| 24093 |
+
"grad_norm": 1.9342801570892334,
|
| 24094 |
+
"learning_rate": 1.600128464845101e-05,
|
| 24095 |
+
"loss": 5.4517,
|
| 24096 |
+
"step": 13764
|
| 24097 |
+
},
|
| 24098 |
+
{
|
| 24099 |
+
"epoch": 0.68,
|
| 24100 |
+
"grad_norm": 1.8290477991104126,
|
| 24101 |
+
"learning_rate": 1.5991402737289393e-05,
|
| 24102 |
+
"loss": 5.4992,
|
| 24103 |
+
"step": 13768
|
| 24104 |
+
},
|
| 24105 |
+
{
|
| 24106 |
+
"epoch": 0.68,
|
| 24107 |
+
"grad_norm": 2.020256757736206,
|
| 24108 |
+
"learning_rate": 1.5981520826127775e-05,
|
| 24109 |
+
"loss": 5.5175,
|
| 24110 |
+
"step": 13772
|
| 24111 |
+
},
|
| 24112 |
+
{
|
| 24113 |
+
"epoch": 0.68,
|
| 24114 |
+
"grad_norm": 2.106531858444214,
|
| 24115 |
+
"learning_rate": 1.5971638914966157e-05,
|
| 24116 |
+
"loss": 5.4482,
|
| 24117 |
+
"step": 13776
|
| 24118 |
+
},
|
| 24119 |
+
{
|
| 24120 |
+
"epoch": 0.68,
|
| 24121 |
+
"grad_norm": 1.9939652681350708,
|
| 24122 |
+
"learning_rate": 1.5961757003804536e-05,
|
| 24123 |
+
"loss": 5.4952,
|
| 24124 |
+
"step": 13780
|
| 24125 |
+
},
|
| 24126 |
+
{
|
| 24127 |
+
"epoch": 0.68,
|
| 24128 |
+
"grad_norm": 2.0970144271850586,
|
| 24129 |
+
"learning_rate": 1.5951875092642918e-05,
|
| 24130 |
+
"loss": 5.5438,
|
| 24131 |
+
"step": 13784
|
| 24132 |
+
},
|
| 24133 |
+
{
|
| 24134 |
+
"epoch": 0.68,
|
| 24135 |
+
"grad_norm": 1.9521361589431763,
|
| 24136 |
+
"learning_rate": 1.59419931814813e-05,
|
| 24137 |
+
"loss": 5.5225,
|
| 24138 |
+
"step": 13788
|
| 24139 |
+
},
|
| 24140 |
+
{
|
| 24141 |
+
"epoch": 0.68,
|
| 24142 |
+
"grad_norm": 2.0148961544036865,
|
| 24143 |
+
"learning_rate": 1.593211127031968e-05,
|
| 24144 |
+
"loss": 5.4785,
|
| 24145 |
+
"step": 13792
|
| 24146 |
+
},
|
| 24147 |
+
{
|
| 24148 |
+
"epoch": 0.68,
|
| 24149 |
+
"grad_norm": 2.1378774642944336,
|
| 24150 |
+
"learning_rate": 1.592222935915806e-05,
|
| 24151 |
+
"loss": 5.5553,
|
| 24152 |
+
"step": 13796
|
| 24153 |
+
},
|
| 24154 |
+
{
|
| 24155 |
+
"epoch": 0.68,
|
| 24156 |
+
"grad_norm": 2.312819719314575,
|
| 24157 |
+
"learning_rate": 1.5912347447996443e-05,
|
| 24158 |
+
"loss": 5.4033,
|
| 24159 |
+
"step": 13800
|
| 24160 |
+
},
|
| 24161 |
+
{
|
| 24162 |
+
"epoch": 0.68,
|
| 24163 |
+
"grad_norm": 2.2060418128967285,
|
| 24164 |
+
"learning_rate": 1.5902465536834825e-05,
|
| 24165 |
+
"loss": 5.4713,
|
| 24166 |
+
"step": 13804
|
| 24167 |
+
},
|
| 24168 |
+
{
|
| 24169 |
+
"epoch": 0.68,
|
| 24170 |
+
"grad_norm": 1.8953157663345337,
|
| 24171 |
+
"learning_rate": 1.5892583625673203e-05,
|
| 24172 |
+
"loss": 5.5035,
|
| 24173 |
+
"step": 13808
|
| 24174 |
+
},
|
| 24175 |
+
{
|
| 24176 |
+
"epoch": 0.68,
|
| 24177 |
+
"grad_norm": 1.7217421531677246,
|
| 24178 |
+
"learning_rate": 1.588270171451159e-05,
|
| 24179 |
+
"loss": 5.6139,
|
| 24180 |
+
"step": 13812
|
| 24181 |
+
},
|
| 24182 |
+
{
|
| 24183 |
+
"epoch": 0.68,
|
| 24184 |
+
"grad_norm": 1.9969755411148071,
|
| 24185 |
+
"learning_rate": 1.587281980334997e-05,
|
| 24186 |
+
"loss": 5.44,
|
| 24187 |
+
"step": 13816
|
| 24188 |
+
},
|
| 24189 |
+
{
|
| 24190 |
+
"epoch": 0.68,
|
| 24191 |
+
"grad_norm": 2.071129560470581,
|
| 24192 |
+
"learning_rate": 1.586293789218835e-05,
|
| 24193 |
+
"loss": 5.4202,
|
| 24194 |
+
"step": 13820
|
| 24195 |
+
},
|
| 24196 |
+
{
|
| 24197 |
+
"epoch": 0.68,
|
| 24198 |
+
"grad_norm": 1.8004354238510132,
|
| 24199 |
+
"learning_rate": 1.5853055981026732e-05,
|
| 24200 |
+
"loss": 5.5286,
|
| 24201 |
+
"step": 13824
|
| 24202 |
+
},
|
| 24203 |
+
{
|
| 24204 |
+
"epoch": 0.68,
|
| 24205 |
+
"grad_norm": 2.0194168090820312,
|
| 24206 |
+
"learning_rate": 1.5843174069865114e-05,
|
| 24207 |
+
"loss": 5.426,
|
| 24208 |
+
"step": 13828
|
| 24209 |
+
},
|
| 24210 |
+
{
|
| 24211 |
+
"epoch": 0.68,
|
| 24212 |
+
"grad_norm": 1.7910535335540771,
|
| 24213 |
+
"learning_rate": 1.5833292158703496e-05,
|
| 24214 |
+
"loss": 5.529,
|
| 24215 |
+
"step": 13832
|
| 24216 |
+
},
|
| 24217 |
+
{
|
| 24218 |
+
"epoch": 0.68,
|
| 24219 |
+
"grad_norm": 2.0648398399353027,
|
| 24220 |
+
"learning_rate": 1.5823410247541874e-05,
|
| 24221 |
+
"loss": 5.5632,
|
| 24222 |
+
"step": 13836
|
| 24223 |
+
},
|
| 24224 |
+
{
|
| 24225 |
+
"epoch": 0.68,
|
| 24226 |
+
"grad_norm": 1.881971001625061,
|
| 24227 |
+
"learning_rate": 1.5813528336380257e-05,
|
| 24228 |
+
"loss": 5.6164,
|
| 24229 |
+
"step": 13840
|
| 24230 |
+
},
|
| 24231 |
+
{
|
| 24232 |
+
"epoch": 0.68,
|
| 24233 |
+
"grad_norm": 2.088383913040161,
|
| 24234 |
+
"learning_rate": 1.580364642521864e-05,
|
| 24235 |
+
"loss": 5.5221,
|
| 24236 |
+
"step": 13844
|
| 24237 |
+
},
|
| 24238 |
+
{
|
| 24239 |
+
"epoch": 0.68,
|
| 24240 |
+
"grad_norm": 1.9117428064346313,
|
| 24241 |
+
"learning_rate": 1.5793764514057017e-05,
|
| 24242 |
+
"loss": 5.6231,
|
| 24243 |
+
"step": 13848
|
| 24244 |
+
},
|
| 24245 |
+
{
|
| 24246 |
+
"epoch": 0.68,
|
| 24247 |
+
"grad_norm": 2.0459342002868652,
|
| 24248 |
+
"learning_rate": 1.57838826028954e-05,
|
| 24249 |
+
"loss": 5.3971,
|
| 24250 |
+
"step": 13852
|
| 24251 |
+
},
|
| 24252 |
+
{
|
| 24253 |
+
"epoch": 0.68,
|
| 24254 |
+
"grad_norm": 1.8177950382232666,
|
| 24255 |
+
"learning_rate": 1.577400069173378e-05,
|
| 24256 |
+
"loss": 5.4996,
|
| 24257 |
+
"step": 13856
|
| 24258 |
+
},
|
| 24259 |
+
{
|
| 24260 |
+
"epoch": 0.68,
|
| 24261 |
+
"grad_norm": 2.090265989303589,
|
| 24262 |
+
"learning_rate": 1.5764118780572163e-05,
|
| 24263 |
+
"loss": 5.5479,
|
| 24264 |
+
"step": 13860
|
| 24265 |
+
},
|
| 24266 |
+
{
|
| 24267 |
+
"epoch": 0.69,
|
| 24268 |
+
"grad_norm": 2.1461987495422363,
|
| 24269 |
+
"learning_rate": 1.5754236869410542e-05,
|
| 24270 |
+
"loss": 5.4959,
|
| 24271 |
+
"step": 13864
|
| 24272 |
+
},
|
| 24273 |
+
{
|
| 24274 |
+
"epoch": 0.69,
|
| 24275 |
+
"grad_norm": 2.008942127227783,
|
| 24276 |
+
"learning_rate": 1.5744354958248928e-05,
|
| 24277 |
+
"loss": 5.3582,
|
| 24278 |
+
"step": 13868
|
| 24279 |
+
},
|
| 24280 |
+
{
|
| 24281 |
+
"epoch": 0.69,
|
| 24282 |
+
"grad_norm": 1.7721225023269653,
|
| 24283 |
+
"learning_rate": 1.573447304708731e-05,
|
| 24284 |
+
"loss": 5.4618,
|
| 24285 |
+
"step": 13872
|
| 24286 |
+
},
|
| 24287 |
+
{
|
| 24288 |
+
"epoch": 0.69,
|
| 24289 |
+
"grad_norm": 2.161746025085449,
|
| 24290 |
+
"learning_rate": 1.572459113592569e-05,
|
| 24291 |
+
"loss": 5.5163,
|
| 24292 |
+
"step": 13876
|
| 24293 |
+
},
|
| 24294 |
+
{
|
| 24295 |
+
"epoch": 0.69,
|
| 24296 |
+
"grad_norm": 2.0746963024139404,
|
| 24297 |
+
"learning_rate": 1.571470922476407e-05,
|
| 24298 |
+
"loss": 5.4898,
|
| 24299 |
+
"step": 13880
|
| 24300 |
+
},
|
| 24301 |
+
{
|
| 24302 |
+
"epoch": 0.69,
|
| 24303 |
+
"grad_norm": Infinity,
|
| 24304 |
+
"learning_rate": 1.5707297791392857e-05,
|
| 24305 |
+
"loss": 5.5354,
|
| 24306 |
+
"step": 13884
|
| 24307 |
+
},
|
| 24308 |
+
{
|
| 24309 |
+
"epoch": 0.69,
|
| 24310 |
+
"grad_norm": 2.0238466262817383,
|
| 24311 |
+
"learning_rate": 1.5697415880231236e-05,
|
| 24312 |
+
"loss": 5.5099,
|
| 24313 |
+
"step": 13888
|
| 24314 |
+
},
|
| 24315 |
+
{
|
| 24316 |
+
"epoch": 0.69,
|
| 24317 |
+
"grad_norm": 1.9300113916397095,
|
| 24318 |
+
"learning_rate": 1.5687533969069618e-05,
|
| 24319 |
+
"loss": 5.682,
|
| 24320 |
+
"step": 13892
|
| 24321 |
+
},
|
| 24322 |
+
{
|
| 24323 |
+
"epoch": 0.69,
|
| 24324 |
+
"grad_norm": 1.9683018922805786,
|
| 24325 |
+
"learning_rate": 1.5677652057908e-05,
|
| 24326 |
+
"loss": 5.5058,
|
| 24327 |
+
"step": 13896
|
| 24328 |
+
},
|
| 24329 |
+
{
|
| 24330 |
+
"epoch": 0.69,
|
| 24331 |
+
"grad_norm": 1.975710391998291,
|
| 24332 |
+
"learning_rate": 1.5667770146746382e-05,
|
| 24333 |
+
"loss": 5.5139,
|
| 24334 |
+
"step": 13900
|
| 24335 |
+
},
|
| 24336 |
+
{
|
| 24337 |
+
"epoch": 0.69,
|
| 24338 |
+
"grad_norm": 1.9731028079986572,
|
| 24339 |
+
"learning_rate": 1.5657888235584764e-05,
|
| 24340 |
+
"loss": 5.565,
|
| 24341 |
+
"step": 13904
|
| 24342 |
+
},
|
| 24343 |
+
{
|
| 24344 |
+
"epoch": 0.69,
|
| 24345 |
+
"grad_norm": 2.0347113609313965,
|
| 24346 |
+
"learning_rate": 1.5648006324423146e-05,
|
| 24347 |
+
"loss": 5.6587,
|
| 24348 |
+
"step": 13908
|
| 24349 |
+
},
|
| 24350 |
+
{
|
| 24351 |
+
"epoch": 0.69,
|
| 24352 |
+
"grad_norm": 1.9666558504104614,
|
| 24353 |
+
"learning_rate": 1.5638124413261525e-05,
|
| 24354 |
+
"loss": 5.6072,
|
| 24355 |
+
"step": 13912
|
| 24356 |
+
},
|
| 24357 |
+
{
|
| 24358 |
+
"epoch": 0.69,
|
| 24359 |
+
"grad_norm": 1.808203935623169,
|
| 24360 |
+
"learning_rate": 1.5628242502099907e-05,
|
| 24361 |
+
"loss": 5.4888,
|
| 24362 |
+
"step": 13916
|
| 24363 |
+
},
|
| 24364 |
+
{
|
| 24365 |
+
"epoch": 0.69,
|
| 24366 |
+
"grad_norm": 2.0783326625823975,
|
| 24367 |
+
"learning_rate": 1.561836059093829e-05,
|
| 24368 |
+
"loss": 5.6407,
|
| 24369 |
+
"step": 13920
|
| 24370 |
+
},
|
| 24371 |
+
{
|
| 24372 |
+
"epoch": 0.69,
|
| 24373 |
+
"grad_norm": 2.006601572036743,
|
| 24374 |
+
"learning_rate": 1.560847867977667e-05,
|
| 24375 |
+
"loss": 5.5085,
|
| 24376 |
+
"step": 13924
|
| 24377 |
+
},
|
| 24378 |
+
{
|
| 24379 |
+
"epoch": 0.69,
|
| 24380 |
+
"grad_norm": 1.8633679151535034,
|
| 24381 |
+
"learning_rate": 1.559859676861505e-05,
|
| 24382 |
+
"loss": 5.4967,
|
| 24383 |
+
"step": 13928
|
| 24384 |
+
},
|
| 24385 |
+
{
|
| 24386 |
+
"epoch": 0.69,
|
| 24387 |
+
"grad_norm": 1.9505640268325806,
|
| 24388 |
+
"learning_rate": 1.558871485745343e-05,
|
| 24389 |
+
"loss": 5.4678,
|
| 24390 |
+
"step": 13932
|
| 24391 |
+
},
|
| 24392 |
+
{
|
| 24393 |
+
"epoch": 0.69,
|
| 24394 |
+
"grad_norm": 2.0986247062683105,
|
| 24395 |
+
"learning_rate": 1.5578832946291814e-05,
|
| 24396 |
+
"loss": 5.4932,
|
| 24397 |
+
"step": 13936
|
| 24398 |
+
},
|
| 24399 |
+
{
|
| 24400 |
+
"epoch": 0.69,
|
| 24401 |
+
"grad_norm": 2.256755828857422,
|
| 24402 |
+
"learning_rate": 1.5568951035130196e-05,
|
| 24403 |
+
"loss": 5.5958,
|
| 24404 |
+
"step": 13940
|
| 24405 |
+
},
|
| 24406 |
+
{
|
| 24407 |
+
"epoch": 0.69,
|
| 24408 |
+
"grad_norm": 2.2147109508514404,
|
| 24409 |
+
"learning_rate": 1.5559069123968574e-05,
|
| 24410 |
+
"loss": 5.5194,
|
| 24411 |
+
"step": 13944
|
| 24412 |
+
},
|
| 24413 |
+
{
|
| 24414 |
+
"epoch": 0.69,
|
| 24415 |
+
"grad_norm": 1.88095223903656,
|
| 24416 |
+
"learning_rate": 1.5549187212806956e-05,
|
| 24417 |
+
"loss": 5.4609,
|
| 24418 |
+
"step": 13948
|
| 24419 |
+
},
|
| 24420 |
+
{
|
| 24421 |
+
"epoch": 0.69,
|
| 24422 |
+
"grad_norm": 1.867007851600647,
|
| 24423 |
+
"learning_rate": 1.553930530164534e-05,
|
| 24424 |
+
"loss": 5.4802,
|
| 24425 |
+
"step": 13952
|
| 24426 |
+
},
|
| 24427 |
+
{
|
| 24428 |
+
"epoch": 0.69,
|
| 24429 |
+
"grad_norm": 2.029303789138794,
|
| 24430 |
+
"learning_rate": 1.552942339048372e-05,
|
| 24431 |
+
"loss": 5.5337,
|
| 24432 |
+
"step": 13956
|
| 24433 |
+
},
|
| 24434 |
+
{
|
| 24435 |
+
"epoch": 0.69,
|
| 24436 |
+
"grad_norm": 1.9142879247665405,
|
| 24437 |
+
"learning_rate": 1.5519541479322103e-05,
|
| 24438 |
+
"loss": 5.4932,
|
| 24439 |
+
"step": 13960
|
| 24440 |
+
},
|
| 24441 |
+
{
|
| 24442 |
+
"epoch": 0.69,
|
| 24443 |
+
"grad_norm": 1.8500617742538452,
|
| 24444 |
+
"learning_rate": 1.5509659568160485e-05,
|
| 24445 |
+
"loss": 5.4916,
|
| 24446 |
+
"step": 13964
|
| 24447 |
+
},
|
| 24448 |
+
{
|
| 24449 |
+
"epoch": 0.69,
|
| 24450 |
+
"grad_norm": 2.0694692134857178,
|
| 24451 |
+
"learning_rate": 1.5499777656998867e-05,
|
| 24452 |
+
"loss": 5.4766,
|
| 24453 |
+
"step": 13968
|
| 24454 |
+
},
|
| 24455 |
+
{
|
| 24456 |
+
"epoch": 0.69,
|
| 24457 |
+
"grad_norm": 2.122901201248169,
|
| 24458 |
+
"learning_rate": 1.5489895745837245e-05,
|
| 24459 |
+
"loss": 5.6467,
|
| 24460 |
+
"step": 13972
|
| 24461 |
+
},
|
| 24462 |
+
{
|
| 24463 |
+
"epoch": 0.69,
|
| 24464 |
+
"grad_norm": 2.108297109603882,
|
| 24465 |
+
"learning_rate": 1.5480013834675627e-05,
|
| 24466 |
+
"loss": 5.5902,
|
| 24467 |
+
"step": 13976
|
| 24468 |
+
},
|
| 24469 |
+
{
|
| 24470 |
+
"epoch": 0.69,
|
| 24471 |
+
"grad_norm": 1.9898920059204102,
|
| 24472 |
+
"learning_rate": 1.547013192351401e-05,
|
| 24473 |
+
"loss": 5.5772,
|
| 24474 |
+
"step": 13980
|
| 24475 |
+
},
|
| 24476 |
+
{
|
| 24477 |
+
"epoch": 0.69,
|
| 24478 |
+
"grad_norm": 2.2552645206451416,
|
| 24479 |
+
"learning_rate": 1.5460250012352388e-05,
|
| 24480 |
+
"loss": 5.507,
|
| 24481 |
+
"step": 13984
|
| 24482 |
+
},
|
| 24483 |
+
{
|
| 24484 |
+
"epoch": 0.69,
|
| 24485 |
+
"grad_norm": 1.9198163747787476,
|
| 24486 |
+
"learning_rate": 1.545036810119077e-05,
|
| 24487 |
+
"loss": 5.4749,
|
| 24488 |
+
"step": 13988
|
| 24489 |
+
},
|
| 24490 |
+
{
|
| 24491 |
+
"epoch": 0.69,
|
| 24492 |
+
"grad_norm": 2.101717233657837,
|
| 24493 |
+
"learning_rate": 1.5440486190029152e-05,
|
| 24494 |
+
"loss": 5.444,
|
| 24495 |
+
"step": 13992
|
| 24496 |
+
},
|
| 24497 |
+
{
|
| 24498 |
+
"epoch": 0.69,
|
| 24499 |
+
"grad_norm": 2.1184744834899902,
|
| 24500 |
+
"learning_rate": 1.5430604278867534e-05,
|
| 24501 |
+
"loss": 5.6017,
|
| 24502 |
+
"step": 13996
|
| 24503 |
+
},
|
| 24504 |
+
{
|
| 24505 |
+
"epoch": 0.69,
|
| 24506 |
+
"grad_norm": 1.7170560359954834,
|
| 24507 |
+
"learning_rate": 1.5420722367705913e-05,
|
| 24508 |
+
"loss": 5.4235,
|
| 24509 |
+
"step": 14000
|
| 24510 |
+
},
|
| 24511 |
+
{
|
| 24512 |
+
"epoch": 0.69,
|
| 24513 |
+
"grad_norm": 2.1068732738494873,
|
| 24514 |
+
"learning_rate": 1.5410840456544295e-05,
|
| 24515 |
+
"loss": 5.504,
|
| 24516 |
+
"step": 14004
|
| 24517 |
+
},
|
| 24518 |
+
{
|
| 24519 |
+
"epoch": 0.69,
|
| 24520 |
+
"grad_norm": 2.275245189666748,
|
| 24521 |
+
"learning_rate": 1.5400958545382677e-05,
|
| 24522 |
+
"loss": 5.5617,
|
| 24523 |
+
"step": 14008
|
| 24524 |
+
},
|
| 24525 |
+
{
|
| 24526 |
+
"epoch": 0.69,
|
| 24527 |
+
"grad_norm": 2.2603347301483154,
|
| 24528 |
+
"learning_rate": 1.539107663422106e-05,
|
| 24529 |
+
"loss": 5.4309,
|
| 24530 |
+
"step": 14012
|
| 24531 |
+
},
|
| 24532 |
+
{
|
| 24533 |
+
"epoch": 0.69,
|
| 24534 |
+
"grad_norm": 2.082984685897827,
|
| 24535 |
+
"learning_rate": 1.538119472305944e-05,
|
| 24536 |
+
"loss": 5.408,
|
| 24537 |
+
"step": 14016
|
| 24538 |
+
},
|
| 24539 |
+
{
|
| 24540 |
+
"epoch": 0.69,
|
| 24541 |
+
"grad_norm": 1.8457398414611816,
|
| 24542 |
+
"learning_rate": 1.5371312811897823e-05,
|
| 24543 |
+
"loss": 5.4611,
|
| 24544 |
+
"step": 14020
|
| 24545 |
+
},
|
| 24546 |
+
{
|
| 24547 |
+
"epoch": 0.69,
|
| 24548 |
+
"grad_norm": 2.0834155082702637,
|
| 24549 |
+
"learning_rate": 1.5361430900736205e-05,
|
| 24550 |
+
"loss": 5.4901,
|
| 24551 |
+
"step": 14024
|
| 24552 |
+
},
|
| 24553 |
+
{
|
| 24554 |
+
"epoch": 0.69,
|
| 24555 |
+
"grad_norm": 1.9145028591156006,
|
| 24556 |
+
"learning_rate": 1.5351548989574584e-05,
|
| 24557 |
+
"loss": 5.5115,
|
| 24558 |
+
"step": 14028
|
| 24559 |
+
},
|
| 24560 |
+
{
|
| 24561 |
+
"epoch": 0.69,
|
| 24562 |
+
"grad_norm": 1.7850406169891357,
|
| 24563 |
+
"learning_rate": 1.5341667078412966e-05,
|
| 24564 |
+
"loss": 5.4384,
|
| 24565 |
+
"step": 14032
|
| 24566 |
+
},
|
| 24567 |
+
{
|
| 24568 |
+
"epoch": 0.69,
|
| 24569 |
+
"grad_norm": 1.964521884918213,
|
| 24570 |
+
"learning_rate": 1.5331785167251348e-05,
|
| 24571 |
+
"loss": 5.3728,
|
| 24572 |
+
"step": 14036
|
| 24573 |
+
},
|
| 24574 |
+
{
|
| 24575 |
+
"epoch": 0.69,
|
| 24576 |
+
"grad_norm": 2.125605821609497,
|
| 24577 |
+
"learning_rate": 1.532190325608973e-05,
|
| 24578 |
+
"loss": 5.4972,
|
| 24579 |
+
"step": 14040
|
| 24580 |
+
},
|
| 24581 |
+
{
|
| 24582 |
+
"epoch": 0.69,
|
| 24583 |
+
"grad_norm": 1.7855336666107178,
|
| 24584 |
+
"learning_rate": 1.531202134492811e-05,
|
| 24585 |
+
"loss": 5.5624,
|
| 24586 |
+
"step": 14044
|
| 24587 |
+
},
|
| 24588 |
+
{
|
| 24589 |
+
"epoch": 0.69,
|
| 24590 |
+
"grad_norm": 1.9751635789871216,
|
| 24591 |
+
"learning_rate": 1.530213943376649e-05,
|
| 24592 |
+
"loss": 5.4388,
|
| 24593 |
+
"step": 14048
|
| 24594 |
+
},
|
| 24595 |
+
{
|
| 24596 |
+
"epoch": 0.69,
|
| 24597 |
+
"grad_norm": 1.943023681640625,
|
| 24598 |
+
"learning_rate": 1.5292257522604873e-05,
|
| 24599 |
+
"loss": 5.4861,
|
| 24600 |
+
"step": 14052
|
| 24601 |
+
},
|
| 24602 |
+
{
|
| 24603 |
+
"epoch": 0.69,
|
| 24604 |
+
"grad_norm": 1.7670375108718872,
|
| 24605 |
+
"learning_rate": 1.528237561144325e-05,
|
| 24606 |
+
"loss": 5.5438,
|
| 24607 |
+
"step": 14056
|
| 24608 |
+
},
|
| 24609 |
+
{
|
| 24610 |
+
"epoch": 0.69,
|
| 24611 |
+
"grad_norm": 2.040696859359741,
|
| 24612 |
+
"learning_rate": 1.5272493700281634e-05,
|
| 24613 |
+
"loss": 5.4905,
|
| 24614 |
+
"step": 14060
|
| 24615 |
+
},
|
| 24616 |
+
{
|
| 24617 |
+
"epoch": 0.69,
|
| 24618 |
+
"grad_norm": 2.2104179859161377,
|
| 24619 |
+
"learning_rate": 1.5262611789120016e-05,
|
| 24620 |
+
"loss": 5.4197,
|
| 24621 |
+
"step": 14064
|
| 24622 |
+
},
|
| 24623 |
+
{
|
| 24624 |
+
"epoch": 0.7,
|
| 24625 |
+
"grad_norm": 2.070598840713501,
|
| 24626 |
+
"learning_rate": 1.5252729877958396e-05,
|
| 24627 |
+
"loss": 5.3342,
|
| 24628 |
+
"step": 14068
|
| 24629 |
+
},
|
| 24630 |
+
{
|
| 24631 |
+
"epoch": 0.7,
|
| 24632 |
+
"grad_norm": 2.1541106700897217,
|
| 24633 |
+
"learning_rate": 1.524284796679678e-05,
|
| 24634 |
+
"loss": 5.5594,
|
| 24635 |
+
"step": 14072
|
| 24636 |
+
},
|
| 24637 |
+
{
|
| 24638 |
+
"epoch": 0.7,
|
| 24639 |
+
"grad_norm": 1.8858451843261719,
|
| 24640 |
+
"learning_rate": 1.5232966055635162e-05,
|
| 24641 |
+
"loss": 5.5494,
|
| 24642 |
+
"step": 14076
|
| 24643 |
+
},
|
| 24644 |
+
{
|
| 24645 |
+
"epoch": 0.7,
|
| 24646 |
+
"grad_norm": 2.0127689838409424,
|
| 24647 |
+
"learning_rate": 1.5223084144473542e-05,
|
| 24648 |
+
"loss": 5.5005,
|
| 24649 |
+
"step": 14080
|
| 24650 |
+
},
|
| 24651 |
+
{
|
| 24652 |
+
"epoch": 0.7,
|
| 24653 |
+
"grad_norm": 1.70602548122406,
|
| 24654 |
+
"learning_rate": 1.5213202233311924e-05,
|
| 24655 |
+
"loss": 5.4694,
|
| 24656 |
+
"step": 14084
|
| 24657 |
+
},
|
| 24658 |
+
{
|
| 24659 |
+
"epoch": 0.7,
|
| 24660 |
+
"grad_norm": 2.1480917930603027,
|
| 24661 |
+
"learning_rate": 1.5203320322150305e-05,
|
| 24662 |
+
"loss": 5.5891,
|
| 24663 |
+
"step": 14088
|
| 24664 |
+
},
|
| 24665 |
+
{
|
| 24666 |
+
"epoch": 0.7,
|
| 24667 |
+
"grad_norm": 1.8127115964889526,
|
| 24668 |
+
"learning_rate": 1.5193438410988687e-05,
|
| 24669 |
+
"loss": 5.497,
|
| 24670 |
+
"step": 14092
|
| 24671 |
+
},
|
| 24672 |
+
{
|
| 24673 |
+
"epoch": 0.7,
|
| 24674 |
+
"grad_norm": 1.768389105796814,
|
| 24675 |
+
"learning_rate": 1.5183556499827067e-05,
|
| 24676 |
+
"loss": 5.5075,
|
| 24677 |
+
"step": 14096
|
| 24678 |
+
},
|
| 24679 |
+
{
|
| 24680 |
+
"epoch": 0.7,
|
| 24681 |
+
"grad_norm": 2.2110512256622314,
|
| 24682 |
+
"learning_rate": 1.517367458866545e-05,
|
| 24683 |
+
"loss": 5.4944,
|
| 24684 |
+
"step": 14100
|
| 24685 |
+
},
|
| 24686 |
+
{
|
| 24687 |
+
"epoch": 0.7,
|
| 24688 |
+
"grad_norm": 2.2322652339935303,
|
| 24689 |
+
"learning_rate": 1.516379267750383e-05,
|
| 24690 |
+
"loss": 5.5072,
|
| 24691 |
+
"step": 14104
|
| 24692 |
+
},
|
| 24693 |
+
{
|
| 24694 |
+
"epoch": 0.7,
|
| 24695 |
+
"grad_norm": 2.103665351867676,
|
| 24696 |
+
"learning_rate": 1.5153910766342212e-05,
|
| 24697 |
+
"loss": 5.5093,
|
| 24698 |
+
"step": 14108
|
| 24699 |
+
},
|
| 24700 |
+
{
|
| 24701 |
+
"epoch": 0.7,
|
| 24702 |
+
"grad_norm": 2.0604758262634277,
|
| 24703 |
+
"learning_rate": 1.5144028855180592e-05,
|
| 24704 |
+
"loss": 5.4539,
|
| 24705 |
+
"step": 14112
|
| 24706 |
+
},
|
| 24707 |
+
{
|
| 24708 |
+
"epoch": 0.7,
|
| 24709 |
+
"grad_norm": 1.8728346824645996,
|
| 24710 |
+
"learning_rate": 1.5134146944018972e-05,
|
| 24711 |
+
"loss": 5.401,
|
| 24712 |
+
"step": 14116
|
| 24713 |
+
},
|
| 24714 |
+
{
|
| 24715 |
+
"epoch": 0.7,
|
| 24716 |
+
"grad_norm": 2.031858444213867,
|
| 24717 |
+
"learning_rate": 1.5124265032857354e-05,
|
| 24718 |
+
"loss": 5.499,
|
| 24719 |
+
"step": 14120
|
| 24720 |
+
},
|
| 24721 |
+
{
|
| 24722 |
+
"epoch": 0.7,
|
| 24723 |
+
"grad_norm": 1.828974962234497,
|
| 24724 |
+
"learning_rate": 1.5114383121695735e-05,
|
| 24725 |
+
"loss": 5.4467,
|
| 24726 |
+
"step": 14124
|
| 24727 |
+
},
|
| 24728 |
+
{
|
| 24729 |
+
"epoch": 0.7,
|
| 24730 |
+
"grad_norm": 1.8760355710983276,
|
| 24731 |
+
"learning_rate": 1.5104501210534119e-05,
|
| 24732 |
+
"loss": 5.3958,
|
| 24733 |
+
"step": 14128
|
| 24734 |
+
},
|
| 24735 |
+
{
|
| 24736 |
+
"epoch": 0.7,
|
| 24737 |
+
"grad_norm": 1.8751137256622314,
|
| 24738 |
+
"learning_rate": 1.50946192993725e-05,
|
| 24739 |
+
"loss": 5.4515,
|
| 24740 |
+
"step": 14132
|
| 24741 |
+
},
|
| 24742 |
+
{
|
| 24743 |
+
"epoch": 0.7,
|
| 24744 |
+
"grad_norm": 1.9187356233596802,
|
| 24745 |
+
"learning_rate": 1.5084737388210881e-05,
|
| 24746 |
+
"loss": 5.5581,
|
| 24747 |
+
"step": 14136
|
| 24748 |
+
},
|
| 24749 |
+
{
|
| 24750 |
+
"epoch": 0.7,
|
| 24751 |
+
"grad_norm": 2.1077044010162354,
|
| 24752 |
+
"learning_rate": 1.5074855477049263e-05,
|
| 24753 |
+
"loss": 5.5018,
|
| 24754 |
+
"step": 14140
|
| 24755 |
+
},
|
| 24756 |
+
{
|
| 24757 |
+
"epoch": 0.7,
|
| 24758 |
+
"grad_norm": 1.9304084777832031,
|
| 24759 |
+
"learning_rate": 1.5064973565887643e-05,
|
| 24760 |
+
"loss": 5.6017,
|
| 24761 |
+
"step": 14144
|
| 24762 |
+
},
|
| 24763 |
+
{
|
| 24764 |
+
"epoch": 0.7,
|
| 24765 |
+
"grad_norm": 1.7740778923034668,
|
| 24766 |
+
"learning_rate": 1.5055091654726025e-05,
|
| 24767 |
+
"loss": 5.5123,
|
| 24768 |
+
"step": 14148
|
| 24769 |
+
},
|
| 24770 |
+
{
|
| 24771 |
+
"epoch": 0.7,
|
| 24772 |
+
"grad_norm": 1.9732328653335571,
|
| 24773 |
+
"learning_rate": 1.5045209743564406e-05,
|
| 24774 |
+
"loss": 5.4148,
|
| 24775 |
+
"step": 14152
|
| 24776 |
+
},
|
| 24777 |
+
{
|
| 24778 |
+
"epoch": 0.7,
|
| 24779 |
+
"grad_norm": 1.98763906955719,
|
| 24780 |
+
"learning_rate": 1.5035327832402788e-05,
|
| 24781 |
+
"loss": 5.5671,
|
| 24782 |
+
"step": 14156
|
| 24783 |
+
},
|
| 24784 |
+
{
|
| 24785 |
+
"epoch": 0.7,
|
| 24786 |
+
"grad_norm": 2.025677442550659,
|
| 24787 |
+
"learning_rate": 1.5025445921241168e-05,
|
| 24788 |
+
"loss": 5.5388,
|
| 24789 |
+
"step": 14160
|
| 24790 |
+
},
|
| 24791 |
+
{
|
| 24792 |
+
"epoch": 0.7,
|
| 24793 |
+
"grad_norm": 2.1304235458374023,
|
| 24794 |
+
"learning_rate": 1.501556401007955e-05,
|
| 24795 |
+
"loss": 5.549,
|
| 24796 |
+
"step": 14164
|
| 24797 |
+
},
|
| 24798 |
+
{
|
| 24799 |
+
"epoch": 0.7,
|
| 24800 |
+
"grad_norm": 1.8666431903839111,
|
| 24801 |
+
"learning_rate": 1.500568209891793e-05,
|
| 24802 |
+
"loss": 5.5133,
|
| 24803 |
+
"step": 14168
|
| 24804 |
}
|
| 24805 |
],
|
| 24806 |
"logging_steps": 4,
|
|
|
|
| 24808 |
"num_input_tokens_seen": 0,
|
| 24809 |
"num_train_epochs": 1,
|
| 24810 |
"save_steps": 2024,
|
| 24811 |
+
"total_flos": 5.967912250428621e+16,
|
| 24812 |
"train_batch_size": 8,
|
| 24813 |
"trial_name": null,
|
| 24814 |
"trial_params": null
|