Training in progress, step 16192, 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:c2f0dbf644a9677e79e68a1d141370f9fc19d80aafd4fa3703ba7daaf4679cdf
|
| 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 sha256:1f9ef24481a5454737bb36638668624276fc1de872bfb96458291412f8b836da
|
| 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 sha256:0ea443f9780df3e4f55b7a8fe78f5f54f5f5faaadff43eb645514989afd8f776
|
| 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 sha256:6bde8530504f368a95ffd37dfe12eaf937597762fada8d945ae1e071babd54fa
|
| 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 sha256:5539ffd5320a73f47cc2092b2ac4690b8e116f5ffa1613c0c9ef2d49e29add72
|
| 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,
|
|
@@ -24801,6 +24801,3548 @@
|
|
| 24801 |
"learning_rate": 1.500568209891793e-05,
|
| 24802 |
"loss": 5.5133,
|
| 24803 |
"step": 14168
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 24804 |
}
|
| 24805 |
],
|
| 24806 |
"logging_steps": 4,
|
|
@@ -24808,7 +28350,7 @@
|
|
| 24808 |
"num_input_tokens_seen": 0,
|
| 24809 |
"num_train_epochs": 1,
|
| 24810 |
"save_steps": 2024,
|
| 24811 |
-
"total_flos":
|
| 24812 |
"train_batch_size": 8,
|
| 24813 |
"trial_name": null,
|
| 24814 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.8000395276446465,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 16192,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 24801 |
"learning_rate": 1.500568209891793e-05,
|
| 24802 |
"loss": 5.5133,
|
| 24803 |
"step": 14168
|
| 24804 |
+
},
|
| 24805 |
+
{
|
| 24806 |
+
"epoch": 0.7,
|
| 24807 |
+
"grad_norm": 2.001925468444824,
|
| 24808 |
+
"learning_rate": 1.4995800187756313e-05,
|
| 24809 |
+
"loss": 5.4878,
|
| 24810 |
+
"step": 14172
|
| 24811 |
+
},
|
| 24812 |
+
{
|
| 24813 |
+
"epoch": 0.7,
|
| 24814 |
+
"grad_norm": 2.0786337852478027,
|
| 24815 |
+
"learning_rate": 1.4985918276594693e-05,
|
| 24816 |
+
"loss": 5.5475,
|
| 24817 |
+
"step": 14176
|
| 24818 |
+
},
|
| 24819 |
+
{
|
| 24820 |
+
"epoch": 0.7,
|
| 24821 |
+
"grad_norm": 2.0157644748687744,
|
| 24822 |
+
"learning_rate": 1.4976036365433073e-05,
|
| 24823 |
+
"loss": 5.4686,
|
| 24824 |
+
"step": 14180
|
| 24825 |
+
},
|
| 24826 |
+
{
|
| 24827 |
+
"epoch": 0.7,
|
| 24828 |
+
"grad_norm": 2.1091387271881104,
|
| 24829 |
+
"learning_rate": 1.4966154454271456e-05,
|
| 24830 |
+
"loss": 5.5175,
|
| 24831 |
+
"step": 14184
|
| 24832 |
+
},
|
| 24833 |
+
{
|
| 24834 |
+
"epoch": 0.7,
|
| 24835 |
+
"grad_norm": 1.8892532587051392,
|
| 24836 |
+
"learning_rate": 1.495627254310984e-05,
|
| 24837 |
+
"loss": 5.57,
|
| 24838 |
+
"step": 14188
|
| 24839 |
+
},
|
| 24840 |
+
{
|
| 24841 |
+
"epoch": 0.7,
|
| 24842 |
+
"grad_norm": 2.0681941509246826,
|
| 24843 |
+
"learning_rate": 1.4946390631948221e-05,
|
| 24844 |
+
"loss": 5.4503,
|
| 24845 |
+
"step": 14192
|
| 24846 |
+
},
|
| 24847 |
+
{
|
| 24848 |
+
"epoch": 0.7,
|
| 24849 |
+
"grad_norm": 2.028324842453003,
|
| 24850 |
+
"learning_rate": 1.4936508720786602e-05,
|
| 24851 |
+
"loss": 5.5231,
|
| 24852 |
+
"step": 14196
|
| 24853 |
+
},
|
| 24854 |
+
{
|
| 24855 |
+
"epoch": 0.7,
|
| 24856 |
+
"grad_norm": 1.8590697050094604,
|
| 24857 |
+
"learning_rate": 1.4926626809624982e-05,
|
| 24858 |
+
"loss": 5.4046,
|
| 24859 |
+
"step": 14200
|
| 24860 |
+
},
|
| 24861 |
+
{
|
| 24862 |
+
"epoch": 0.7,
|
| 24863 |
+
"grad_norm": 2.1684091091156006,
|
| 24864 |
+
"learning_rate": 1.4916744898463364e-05,
|
| 24865 |
+
"loss": 5.5428,
|
| 24866 |
+
"step": 14204
|
| 24867 |
+
},
|
| 24868 |
+
{
|
| 24869 |
+
"epoch": 0.7,
|
| 24870 |
+
"grad_norm": 2.4600932598114014,
|
| 24871 |
+
"learning_rate": 1.4906862987301745e-05,
|
| 24872 |
+
"loss": 5.4973,
|
| 24873 |
+
"step": 14208
|
| 24874 |
+
},
|
| 24875 |
+
{
|
| 24876 |
+
"epoch": 0.7,
|
| 24877 |
+
"grad_norm": 2.182546615600586,
|
| 24878 |
+
"learning_rate": 1.4896981076140127e-05,
|
| 24879 |
+
"loss": 5.5168,
|
| 24880 |
+
"step": 14212
|
| 24881 |
+
},
|
| 24882 |
+
{
|
| 24883 |
+
"epoch": 0.7,
|
| 24884 |
+
"grad_norm": 2.0207765102386475,
|
| 24885 |
+
"learning_rate": 1.4887099164978507e-05,
|
| 24886 |
+
"loss": 5.4892,
|
| 24887 |
+
"step": 14216
|
| 24888 |
+
},
|
| 24889 |
+
{
|
| 24890 |
+
"epoch": 0.7,
|
| 24891 |
+
"grad_norm": 2.07978892326355,
|
| 24892 |
+
"learning_rate": 1.4877217253816889e-05,
|
| 24893 |
+
"loss": 5.5074,
|
| 24894 |
+
"step": 14220
|
| 24895 |
+
},
|
| 24896 |
+
{
|
| 24897 |
+
"epoch": 0.7,
|
| 24898 |
+
"grad_norm": 2.0122103691101074,
|
| 24899 |
+
"learning_rate": 1.486733534265527e-05,
|
| 24900 |
+
"loss": 5.5007,
|
| 24901 |
+
"step": 14224
|
| 24902 |
+
},
|
| 24903 |
+
{
|
| 24904 |
+
"epoch": 0.7,
|
| 24905 |
+
"grad_norm": 1.951704502105713,
|
| 24906 |
+
"learning_rate": 1.4857453431493651e-05,
|
| 24907 |
+
"loss": 5.5281,
|
| 24908 |
+
"step": 14228
|
| 24909 |
+
},
|
| 24910 |
+
{
|
| 24911 |
+
"epoch": 0.7,
|
| 24912 |
+
"grad_norm": 1.903686761856079,
|
| 24913 |
+
"learning_rate": 1.4847571520332032e-05,
|
| 24914 |
+
"loss": 5.483,
|
| 24915 |
+
"step": 14232
|
| 24916 |
+
},
|
| 24917 |
+
{
|
| 24918 |
+
"epoch": 0.7,
|
| 24919 |
+
"grad_norm": 1.7512837648391724,
|
| 24920 |
+
"learning_rate": 1.4837689609170414e-05,
|
| 24921 |
+
"loss": 5.5026,
|
| 24922 |
+
"step": 14236
|
| 24923 |
+
},
|
| 24924 |
+
{
|
| 24925 |
+
"epoch": 0.7,
|
| 24926 |
+
"grad_norm": 1.9716506004333496,
|
| 24927 |
+
"learning_rate": 1.4827807698008794e-05,
|
| 24928 |
+
"loss": 5.5547,
|
| 24929 |
+
"step": 14240
|
| 24930 |
+
},
|
| 24931 |
+
{
|
| 24932 |
+
"epoch": 0.7,
|
| 24933 |
+
"grad_norm": 2.1064813137054443,
|
| 24934 |
+
"learning_rate": 1.4817925786847178e-05,
|
| 24935 |
+
"loss": 5.6868,
|
| 24936 |
+
"step": 14244
|
| 24937 |
+
},
|
| 24938 |
+
{
|
| 24939 |
+
"epoch": 0.7,
|
| 24940 |
+
"grad_norm": 1.920245885848999,
|
| 24941 |
+
"learning_rate": 1.480804387568556e-05,
|
| 24942 |
+
"loss": 5.5195,
|
| 24943 |
+
"step": 14248
|
| 24944 |
+
},
|
| 24945 |
+
{
|
| 24946 |
+
"epoch": 0.7,
|
| 24947 |
+
"grad_norm": 1.8807624578475952,
|
| 24948 |
+
"learning_rate": 1.479816196452394e-05,
|
| 24949 |
+
"loss": 5.5605,
|
| 24950 |
+
"step": 14252
|
| 24951 |
+
},
|
| 24952 |
+
{
|
| 24953 |
+
"epoch": 0.7,
|
| 24954 |
+
"grad_norm": 2.1558022499084473,
|
| 24955 |
+
"learning_rate": 1.4788280053362322e-05,
|
| 24956 |
+
"loss": 5.54,
|
| 24957 |
+
"step": 14256
|
| 24958 |
+
},
|
| 24959 |
+
{
|
| 24960 |
+
"epoch": 0.7,
|
| 24961 |
+
"grad_norm": 1.9825403690338135,
|
| 24962 |
+
"learning_rate": 1.4778398142200703e-05,
|
| 24963 |
+
"loss": 5.5888,
|
| 24964 |
+
"step": 14260
|
| 24965 |
+
},
|
| 24966 |
+
{
|
| 24967 |
+
"epoch": 0.7,
|
| 24968 |
+
"grad_norm": 1.974697232246399,
|
| 24969 |
+
"learning_rate": 1.4768516231039083e-05,
|
| 24970 |
+
"loss": 5.5791,
|
| 24971 |
+
"step": 14264
|
| 24972 |
+
},
|
| 24973 |
+
{
|
| 24974 |
+
"epoch": 0.7,
|
| 24975 |
+
"grad_norm": 2.0667243003845215,
|
| 24976 |
+
"learning_rate": 1.4758634319877465e-05,
|
| 24977 |
+
"loss": 5.4354,
|
| 24978 |
+
"step": 14268
|
| 24979 |
+
},
|
| 24980 |
+
{
|
| 24981 |
+
"epoch": 0.71,
|
| 24982 |
+
"grad_norm": 1.8902255296707153,
|
| 24983 |
+
"learning_rate": 1.4748752408715846e-05,
|
| 24984 |
+
"loss": 5.5167,
|
| 24985 |
+
"step": 14272
|
| 24986 |
+
},
|
| 24987 |
+
{
|
| 24988 |
+
"epoch": 0.71,
|
| 24989 |
+
"grad_norm": 1.882251501083374,
|
| 24990 |
+
"learning_rate": 1.4738870497554228e-05,
|
| 24991 |
+
"loss": 5.4551,
|
| 24992 |
+
"step": 14276
|
| 24993 |
+
},
|
| 24994 |
+
{
|
| 24995 |
+
"epoch": 0.71,
|
| 24996 |
+
"grad_norm": 2.0265276432037354,
|
| 24997 |
+
"learning_rate": 1.4728988586392608e-05,
|
| 24998 |
+
"loss": 5.3989,
|
| 24999 |
+
"step": 14280
|
| 25000 |
+
},
|
| 25001 |
+
{
|
| 25002 |
+
"epoch": 0.71,
|
| 25003 |
+
"grad_norm": 2.1499834060668945,
|
| 25004 |
+
"learning_rate": 1.471910667523099e-05,
|
| 25005 |
+
"loss": 5.3913,
|
| 25006 |
+
"step": 14284
|
| 25007 |
+
},
|
| 25008 |
+
{
|
| 25009 |
+
"epoch": 0.71,
|
| 25010 |
+
"grad_norm": 1.9374068975448608,
|
| 25011 |
+
"learning_rate": 1.470922476406937e-05,
|
| 25012 |
+
"loss": 5.4951,
|
| 25013 |
+
"step": 14288
|
| 25014 |
+
},
|
| 25015 |
+
{
|
| 25016 |
+
"epoch": 0.71,
|
| 25017 |
+
"grad_norm": 2.04496693611145,
|
| 25018 |
+
"learning_rate": 1.4699342852907753e-05,
|
| 25019 |
+
"loss": 5.5568,
|
| 25020 |
+
"step": 14292
|
| 25021 |
+
},
|
| 25022 |
+
{
|
| 25023 |
+
"epoch": 0.71,
|
| 25024 |
+
"grad_norm": 2.1718828678131104,
|
| 25025 |
+
"learning_rate": 1.4689460941746133e-05,
|
| 25026 |
+
"loss": 5.4003,
|
| 25027 |
+
"step": 14296
|
| 25028 |
+
},
|
| 25029 |
+
{
|
| 25030 |
+
"epoch": 0.71,
|
| 25031 |
+
"grad_norm": 2.170172929763794,
|
| 25032 |
+
"learning_rate": 1.4679579030584517e-05,
|
| 25033 |
+
"loss": 5.4535,
|
| 25034 |
+
"step": 14300
|
| 25035 |
+
},
|
| 25036 |
+
{
|
| 25037 |
+
"epoch": 0.71,
|
| 25038 |
+
"grad_norm": 1.9548550844192505,
|
| 25039 |
+
"learning_rate": 1.4669697119422899e-05,
|
| 25040 |
+
"loss": 5.5077,
|
| 25041 |
+
"step": 14304
|
| 25042 |
+
},
|
| 25043 |
+
{
|
| 25044 |
+
"epoch": 0.71,
|
| 25045 |
+
"grad_norm": 1.7335338592529297,
|
| 25046 |
+
"learning_rate": 1.4659815208261279e-05,
|
| 25047 |
+
"loss": 5.4265,
|
| 25048 |
+
"step": 14308
|
| 25049 |
+
},
|
| 25050 |
+
{
|
| 25051 |
+
"epoch": 0.71,
|
| 25052 |
+
"grad_norm": 2.0178778171539307,
|
| 25053 |
+
"learning_rate": 1.4649933297099661e-05,
|
| 25054 |
+
"loss": 5.3215,
|
| 25055 |
+
"step": 14312
|
| 25056 |
+
},
|
| 25057 |
+
{
|
| 25058 |
+
"epoch": 0.71,
|
| 25059 |
+
"grad_norm": 2.198841094970703,
|
| 25060 |
+
"learning_rate": 1.4640051385938042e-05,
|
| 25061 |
+
"loss": 5.5392,
|
| 25062 |
+
"step": 14316
|
| 25063 |
+
},
|
| 25064 |
+
{
|
| 25065 |
+
"epoch": 0.71,
|
| 25066 |
+
"grad_norm": 2.093581438064575,
|
| 25067 |
+
"learning_rate": 1.4630169474776424e-05,
|
| 25068 |
+
"loss": 5.4572,
|
| 25069 |
+
"step": 14320
|
| 25070 |
+
},
|
| 25071 |
+
{
|
| 25072 |
+
"epoch": 0.71,
|
| 25073 |
+
"grad_norm": 1.7876406908035278,
|
| 25074 |
+
"learning_rate": 1.4620287563614804e-05,
|
| 25075 |
+
"loss": 5.3976,
|
| 25076 |
+
"step": 14324
|
| 25077 |
+
},
|
| 25078 |
+
{
|
| 25079 |
+
"epoch": 0.71,
|
| 25080 |
+
"grad_norm": 2.3379602432250977,
|
| 25081 |
+
"learning_rate": 1.4610405652453184e-05,
|
| 25082 |
+
"loss": 5.5972,
|
| 25083 |
+
"step": 14328
|
| 25084 |
+
},
|
| 25085 |
+
{
|
| 25086 |
+
"epoch": 0.71,
|
| 25087 |
+
"grad_norm": 2.0745928287506104,
|
| 25088 |
+
"learning_rate": 1.4600523741291566e-05,
|
| 25089 |
+
"loss": 5.4489,
|
| 25090 |
+
"step": 14332
|
| 25091 |
+
},
|
| 25092 |
+
{
|
| 25093 |
+
"epoch": 0.71,
|
| 25094 |
+
"grad_norm": 1.9868686199188232,
|
| 25095 |
+
"learning_rate": 1.4590641830129947e-05,
|
| 25096 |
+
"loss": 5.5921,
|
| 25097 |
+
"step": 14336
|
| 25098 |
+
},
|
| 25099 |
+
{
|
| 25100 |
+
"epoch": 0.71,
|
| 25101 |
+
"grad_norm": 2.052441358566284,
|
| 25102 |
+
"learning_rate": 1.4580759918968329e-05,
|
| 25103 |
+
"loss": 5.5621,
|
| 25104 |
+
"step": 14340
|
| 25105 |
+
},
|
| 25106 |
+
{
|
| 25107 |
+
"epoch": 0.71,
|
| 25108 |
+
"grad_norm": 2.250218629837036,
|
| 25109 |
+
"learning_rate": 1.457087800780671e-05,
|
| 25110 |
+
"loss": 5.3768,
|
| 25111 |
+
"step": 14344
|
| 25112 |
+
},
|
| 25113 |
+
{
|
| 25114 |
+
"epoch": 0.71,
|
| 25115 |
+
"grad_norm": 2.0471935272216797,
|
| 25116 |
+
"learning_rate": 1.4560996096645091e-05,
|
| 25117 |
+
"loss": 5.5863,
|
| 25118 |
+
"step": 14348
|
| 25119 |
+
},
|
| 25120 |
+
{
|
| 25121 |
+
"epoch": 0.71,
|
| 25122 |
+
"grad_norm": 2.0607481002807617,
|
| 25123 |
+
"learning_rate": 1.4551114185483472e-05,
|
| 25124 |
+
"loss": 5.5404,
|
| 25125 |
+
"step": 14352
|
| 25126 |
+
},
|
| 25127 |
+
{
|
| 25128 |
+
"epoch": 0.71,
|
| 25129 |
+
"grad_norm": 1.7724336385726929,
|
| 25130 |
+
"learning_rate": 1.4541232274321854e-05,
|
| 25131 |
+
"loss": 5.4719,
|
| 25132 |
+
"step": 14356
|
| 25133 |
+
},
|
| 25134 |
+
{
|
| 25135 |
+
"epoch": 0.71,
|
| 25136 |
+
"grad_norm": 2.1439015865325928,
|
| 25137 |
+
"learning_rate": 1.4531350363160237e-05,
|
| 25138 |
+
"loss": 5.5045,
|
| 25139 |
+
"step": 14360
|
| 25140 |
+
},
|
| 25141 |
+
{
|
| 25142 |
+
"epoch": 0.71,
|
| 25143 |
+
"grad_norm": 2.067152500152588,
|
| 25144 |
+
"learning_rate": 1.4521468451998618e-05,
|
| 25145 |
+
"loss": 5.5518,
|
| 25146 |
+
"step": 14364
|
| 25147 |
+
},
|
| 25148 |
+
{
|
| 25149 |
+
"epoch": 0.71,
|
| 25150 |
+
"grad_norm": 1.8666131496429443,
|
| 25151 |
+
"learning_rate": 1.4511586540837e-05,
|
| 25152 |
+
"loss": 5.4791,
|
| 25153 |
+
"step": 14368
|
| 25154 |
+
},
|
| 25155 |
+
{
|
| 25156 |
+
"epoch": 0.71,
|
| 25157 |
+
"grad_norm": 1.9369677305221558,
|
| 25158 |
+
"learning_rate": 1.450170462967538e-05,
|
| 25159 |
+
"loss": 5.4379,
|
| 25160 |
+
"step": 14372
|
| 25161 |
+
},
|
| 25162 |
+
{
|
| 25163 |
+
"epoch": 0.71,
|
| 25164 |
+
"grad_norm": 2.0378055572509766,
|
| 25165 |
+
"learning_rate": 1.4491822718513762e-05,
|
| 25166 |
+
"loss": 5.4874,
|
| 25167 |
+
"step": 14376
|
| 25168 |
+
},
|
| 25169 |
+
{
|
| 25170 |
+
"epoch": 0.71,
|
| 25171 |
+
"grad_norm": 1.8923200368881226,
|
| 25172 |
+
"learning_rate": 1.4481940807352143e-05,
|
| 25173 |
+
"loss": 5.5813,
|
| 25174 |
+
"step": 14380
|
| 25175 |
+
},
|
| 25176 |
+
{
|
| 25177 |
+
"epoch": 0.71,
|
| 25178 |
+
"grad_norm": 1.9572350978851318,
|
| 25179 |
+
"learning_rate": 1.4472058896190525e-05,
|
| 25180 |
+
"loss": 5.5161,
|
| 25181 |
+
"step": 14384
|
| 25182 |
+
},
|
| 25183 |
+
{
|
| 25184 |
+
"epoch": 0.71,
|
| 25185 |
+
"grad_norm": 1.9131194353103638,
|
| 25186 |
+
"learning_rate": 1.4462176985028905e-05,
|
| 25187 |
+
"loss": 5.6163,
|
| 25188 |
+
"step": 14388
|
| 25189 |
+
},
|
| 25190 |
+
{
|
| 25191 |
+
"epoch": 0.71,
|
| 25192 |
+
"grad_norm": 2.5276191234588623,
|
| 25193 |
+
"learning_rate": 1.4452295073867287e-05,
|
| 25194 |
+
"loss": 5.5442,
|
| 25195 |
+
"step": 14392
|
| 25196 |
+
},
|
| 25197 |
+
{
|
| 25198 |
+
"epoch": 0.71,
|
| 25199 |
+
"grad_norm": 2.0798611640930176,
|
| 25200 |
+
"learning_rate": 1.4442413162705667e-05,
|
| 25201 |
+
"loss": 5.6086,
|
| 25202 |
+
"step": 14396
|
| 25203 |
+
},
|
| 25204 |
+
{
|
| 25205 |
+
"epoch": 0.71,
|
| 25206 |
+
"grad_norm": 2.1294023990631104,
|
| 25207 |
+
"learning_rate": 1.4432531251544048e-05,
|
| 25208 |
+
"loss": 5.3953,
|
| 25209 |
+
"step": 14400
|
| 25210 |
+
},
|
| 25211 |
+
{
|
| 25212 |
+
"epoch": 0.71,
|
| 25213 |
+
"grad_norm": 2.01753568649292,
|
| 25214 |
+
"learning_rate": 1.442264934038243e-05,
|
| 25215 |
+
"loss": 5.537,
|
| 25216 |
+
"step": 14404
|
| 25217 |
+
},
|
| 25218 |
+
{
|
| 25219 |
+
"epoch": 0.71,
|
| 25220 |
+
"grad_norm": 2.0679104328155518,
|
| 25221 |
+
"learning_rate": 1.441276742922081e-05,
|
| 25222 |
+
"loss": 5.4632,
|
| 25223 |
+
"step": 14408
|
| 25224 |
+
},
|
| 25225 |
+
{
|
| 25226 |
+
"epoch": 0.71,
|
| 25227 |
+
"grad_norm": 1.921613097190857,
|
| 25228 |
+
"learning_rate": 1.4402885518059192e-05,
|
| 25229 |
+
"loss": 5.2852,
|
| 25230 |
+
"step": 14412
|
| 25231 |
+
},
|
| 25232 |
+
{
|
| 25233 |
+
"epoch": 0.71,
|
| 25234 |
+
"grad_norm": 1.9771466255187988,
|
| 25235 |
+
"learning_rate": 1.4393003606897576e-05,
|
| 25236 |
+
"loss": 5.5808,
|
| 25237 |
+
"step": 14416
|
| 25238 |
+
},
|
| 25239 |
+
{
|
| 25240 |
+
"epoch": 0.71,
|
| 25241 |
+
"grad_norm": 1.8879215717315674,
|
| 25242 |
+
"learning_rate": 1.4383121695735956e-05,
|
| 25243 |
+
"loss": 5.4957,
|
| 25244 |
+
"step": 14420
|
| 25245 |
+
},
|
| 25246 |
+
{
|
| 25247 |
+
"epoch": 0.71,
|
| 25248 |
+
"grad_norm": 2.0186474323272705,
|
| 25249 |
+
"learning_rate": 1.4373239784574339e-05,
|
| 25250 |
+
"loss": 5.4864,
|
| 25251 |
+
"step": 14424
|
| 25252 |
+
},
|
| 25253 |
+
{
|
| 25254 |
+
"epoch": 0.71,
|
| 25255 |
+
"grad_norm": 2.147420883178711,
|
| 25256 |
+
"learning_rate": 1.4363357873412719e-05,
|
| 25257 |
+
"loss": 5.5098,
|
| 25258 |
+
"step": 14428
|
| 25259 |
+
},
|
| 25260 |
+
{
|
| 25261 |
+
"epoch": 0.71,
|
| 25262 |
+
"grad_norm": 1.9963911771774292,
|
| 25263 |
+
"learning_rate": 1.4353475962251101e-05,
|
| 25264 |
+
"loss": 5.3886,
|
| 25265 |
+
"step": 14432
|
| 25266 |
+
},
|
| 25267 |
+
{
|
| 25268 |
+
"epoch": 0.71,
|
| 25269 |
+
"grad_norm": 2.03368878364563,
|
| 25270 |
+
"learning_rate": 1.4343594051089481e-05,
|
| 25271 |
+
"loss": 5.4533,
|
| 25272 |
+
"step": 14436
|
| 25273 |
+
},
|
| 25274 |
+
{
|
| 25275 |
+
"epoch": 0.71,
|
| 25276 |
+
"grad_norm": 2.274022340774536,
|
| 25277 |
+
"learning_rate": 1.4333712139927863e-05,
|
| 25278 |
+
"loss": 5.5222,
|
| 25279 |
+
"step": 14440
|
| 25280 |
+
},
|
| 25281 |
+
{
|
| 25282 |
+
"epoch": 0.71,
|
| 25283 |
+
"grad_norm": 2.0925748348236084,
|
| 25284 |
+
"learning_rate": 1.4323830228766244e-05,
|
| 25285 |
+
"loss": 5.3865,
|
| 25286 |
+
"step": 14444
|
| 25287 |
+
},
|
| 25288 |
+
{
|
| 25289 |
+
"epoch": 0.71,
|
| 25290 |
+
"grad_norm": 2.25508451461792,
|
| 25291 |
+
"learning_rate": 1.4313948317604626e-05,
|
| 25292 |
+
"loss": 5.388,
|
| 25293 |
+
"step": 14448
|
| 25294 |
+
},
|
| 25295 |
+
{
|
| 25296 |
+
"epoch": 0.71,
|
| 25297 |
+
"grad_norm": 1.9655667543411255,
|
| 25298 |
+
"learning_rate": 1.4304066406443006e-05,
|
| 25299 |
+
"loss": 5.5274,
|
| 25300 |
+
"step": 14452
|
| 25301 |
+
},
|
| 25302 |
+
{
|
| 25303 |
+
"epoch": 0.71,
|
| 25304 |
+
"grad_norm": 1.7187186479568481,
|
| 25305 |
+
"learning_rate": 1.4294184495281388e-05,
|
| 25306 |
+
"loss": 5.5024,
|
| 25307 |
+
"step": 14456
|
| 25308 |
+
},
|
| 25309 |
+
{
|
| 25310 |
+
"epoch": 0.71,
|
| 25311 |
+
"grad_norm": 1.9573662281036377,
|
| 25312 |
+
"learning_rate": 1.4284302584119769e-05,
|
| 25313 |
+
"loss": 5.5709,
|
| 25314 |
+
"step": 14460
|
| 25315 |
+
},
|
| 25316 |
+
{
|
| 25317 |
+
"epoch": 0.71,
|
| 25318 |
+
"grad_norm": 1.9070782661437988,
|
| 25319 |
+
"learning_rate": 1.4274420672958149e-05,
|
| 25320 |
+
"loss": 5.5638,
|
| 25321 |
+
"step": 14464
|
| 25322 |
+
},
|
| 25323 |
+
{
|
| 25324 |
+
"epoch": 0.71,
|
| 25325 |
+
"grad_norm": 1.99477219581604,
|
| 25326 |
+
"learning_rate": 1.4264538761796531e-05,
|
| 25327 |
+
"loss": 5.554,
|
| 25328 |
+
"step": 14468
|
| 25329 |
+
},
|
| 25330 |
+
{
|
| 25331 |
+
"epoch": 0.72,
|
| 25332 |
+
"grad_norm": 2.1351230144500732,
|
| 25333 |
+
"learning_rate": 1.4254656850634915e-05,
|
| 25334 |
+
"loss": 5.4401,
|
| 25335 |
+
"step": 14472
|
| 25336 |
+
},
|
| 25337 |
+
{
|
| 25338 |
+
"epoch": 0.72,
|
| 25339 |
+
"grad_norm": 2.12768292427063,
|
| 25340 |
+
"learning_rate": 1.4244774939473297e-05,
|
| 25341 |
+
"loss": 5.5931,
|
| 25342 |
+
"step": 14476
|
| 25343 |
+
},
|
| 25344 |
+
{
|
| 25345 |
+
"epoch": 0.72,
|
| 25346 |
+
"grad_norm": 1.834502100944519,
|
| 25347 |
+
"learning_rate": 1.4234893028311677e-05,
|
| 25348 |
+
"loss": 5.3383,
|
| 25349 |
+
"step": 14480
|
| 25350 |
+
},
|
| 25351 |
+
{
|
| 25352 |
+
"epoch": 0.72,
|
| 25353 |
+
"grad_norm": 1.9401111602783203,
|
| 25354 |
+
"learning_rate": 1.4225011117150058e-05,
|
| 25355 |
+
"loss": 5.5206,
|
| 25356 |
+
"step": 14484
|
| 25357 |
+
},
|
| 25358 |
+
{
|
| 25359 |
+
"epoch": 0.72,
|
| 25360 |
+
"grad_norm": 2.1304268836975098,
|
| 25361 |
+
"learning_rate": 1.421512920598844e-05,
|
| 25362 |
+
"loss": 5.4791,
|
| 25363 |
+
"step": 14488
|
| 25364 |
+
},
|
| 25365 |
+
{
|
| 25366 |
+
"epoch": 0.72,
|
| 25367 |
+
"grad_norm": 1.8855656385421753,
|
| 25368 |
+
"learning_rate": 1.420524729482682e-05,
|
| 25369 |
+
"loss": 5.4404,
|
| 25370 |
+
"step": 14492
|
| 25371 |
+
},
|
| 25372 |
+
{
|
| 25373 |
+
"epoch": 0.72,
|
| 25374 |
+
"grad_norm": 2.1601414680480957,
|
| 25375 |
+
"learning_rate": 1.4195365383665202e-05,
|
| 25376 |
+
"loss": 5.509,
|
| 25377 |
+
"step": 14496
|
| 25378 |
+
},
|
| 25379 |
+
{
|
| 25380 |
+
"epoch": 0.72,
|
| 25381 |
+
"grad_norm": 2.054011821746826,
|
| 25382 |
+
"learning_rate": 1.4185483472503582e-05,
|
| 25383 |
+
"loss": 5.5097,
|
| 25384 |
+
"step": 14500
|
| 25385 |
+
},
|
| 25386 |
+
{
|
| 25387 |
+
"epoch": 0.72,
|
| 25388 |
+
"grad_norm": 2.2468581199645996,
|
| 25389 |
+
"learning_rate": 1.4175601561341964e-05,
|
| 25390 |
+
"loss": 5.5658,
|
| 25391 |
+
"step": 14504
|
| 25392 |
+
},
|
| 25393 |
+
{
|
| 25394 |
+
"epoch": 0.72,
|
| 25395 |
+
"grad_norm": 1.8028780221939087,
|
| 25396 |
+
"learning_rate": 1.4165719650180345e-05,
|
| 25397 |
+
"loss": 5.4528,
|
| 25398 |
+
"step": 14508
|
| 25399 |
+
},
|
| 25400 |
+
{
|
| 25401 |
+
"epoch": 0.72,
|
| 25402 |
+
"grad_norm": 1.7557624578475952,
|
| 25403 |
+
"learning_rate": 1.4155837739018727e-05,
|
| 25404 |
+
"loss": 5.4267,
|
| 25405 |
+
"step": 14512
|
| 25406 |
+
},
|
| 25407 |
+
{
|
| 25408 |
+
"epoch": 0.72,
|
| 25409 |
+
"grad_norm": 2.2908835411071777,
|
| 25410 |
+
"learning_rate": 1.4145955827857107e-05,
|
| 25411 |
+
"loss": 5.5676,
|
| 25412 |
+
"step": 14516
|
| 25413 |
+
},
|
| 25414 |
+
{
|
| 25415 |
+
"epoch": 0.72,
|
| 25416 |
+
"grad_norm": 1.8863370418548584,
|
| 25417 |
+
"learning_rate": 1.413607391669549e-05,
|
| 25418 |
+
"loss": 5.5873,
|
| 25419 |
+
"step": 14520
|
| 25420 |
+
},
|
| 25421 |
+
{
|
| 25422 |
+
"epoch": 0.72,
|
| 25423 |
+
"grad_norm": 1.9381186962127686,
|
| 25424 |
+
"learning_rate": 1.412619200553387e-05,
|
| 25425 |
+
"loss": 5.5573,
|
| 25426 |
+
"step": 14524
|
| 25427 |
+
},
|
| 25428 |
+
{
|
| 25429 |
+
"epoch": 0.72,
|
| 25430 |
+
"grad_norm": 1.9330207109451294,
|
| 25431 |
+
"learning_rate": 1.4116310094372252e-05,
|
| 25432 |
+
"loss": 5.3673,
|
| 25433 |
+
"step": 14528
|
| 25434 |
+
},
|
| 25435 |
+
{
|
| 25436 |
+
"epoch": 0.72,
|
| 25437 |
+
"grad_norm": 2.0471458435058594,
|
| 25438 |
+
"learning_rate": 1.4106428183210636e-05,
|
| 25439 |
+
"loss": 5.4435,
|
| 25440 |
+
"step": 14532
|
| 25441 |
+
},
|
| 25442 |
+
{
|
| 25443 |
+
"epoch": 0.72,
|
| 25444 |
+
"grad_norm": 1.922804594039917,
|
| 25445 |
+
"learning_rate": 1.4096546272049016e-05,
|
| 25446 |
+
"loss": 5.4511,
|
| 25447 |
+
"step": 14536
|
| 25448 |
+
},
|
| 25449 |
+
{
|
| 25450 |
+
"epoch": 0.72,
|
| 25451 |
+
"grad_norm": 1.946153998374939,
|
| 25452 |
+
"learning_rate": 1.4086664360887398e-05,
|
| 25453 |
+
"loss": 5.4641,
|
| 25454 |
+
"step": 14540
|
| 25455 |
+
},
|
| 25456 |
+
{
|
| 25457 |
+
"epoch": 0.72,
|
| 25458 |
+
"grad_norm": 2.0340843200683594,
|
| 25459 |
+
"learning_rate": 1.4076782449725778e-05,
|
| 25460 |
+
"loss": 5.5205,
|
| 25461 |
+
"step": 14544
|
| 25462 |
+
},
|
| 25463 |
+
{
|
| 25464 |
+
"epoch": 0.72,
|
| 25465 |
+
"grad_norm": 2.218997001647949,
|
| 25466 |
+
"learning_rate": 1.4066900538564159e-05,
|
| 25467 |
+
"loss": 5.4572,
|
| 25468 |
+
"step": 14548
|
| 25469 |
+
},
|
| 25470 |
+
{
|
| 25471 |
+
"epoch": 0.72,
|
| 25472 |
+
"grad_norm": 1.8929450511932373,
|
| 25473 |
+
"learning_rate": 1.405701862740254e-05,
|
| 25474 |
+
"loss": 5.4428,
|
| 25475 |
+
"step": 14552
|
| 25476 |
+
},
|
| 25477 |
+
{
|
| 25478 |
+
"epoch": 0.72,
|
| 25479 |
+
"grad_norm": 1.828822374343872,
|
| 25480 |
+
"learning_rate": 1.4047136716240921e-05,
|
| 25481 |
+
"loss": 5.4737,
|
| 25482 |
+
"step": 14556
|
| 25483 |
+
},
|
| 25484 |
+
{
|
| 25485 |
+
"epoch": 0.72,
|
| 25486 |
+
"grad_norm": 2.1129636764526367,
|
| 25487 |
+
"learning_rate": 1.4037254805079303e-05,
|
| 25488 |
+
"loss": 5.4952,
|
| 25489 |
+
"step": 14560
|
| 25490 |
+
},
|
| 25491 |
+
{
|
| 25492 |
+
"epoch": 0.72,
|
| 25493 |
+
"grad_norm": 2.0485777854919434,
|
| 25494 |
+
"learning_rate": 1.4027372893917684e-05,
|
| 25495 |
+
"loss": 5.5007,
|
| 25496 |
+
"step": 14564
|
| 25497 |
+
},
|
| 25498 |
+
{
|
| 25499 |
+
"epoch": 0.72,
|
| 25500 |
+
"grad_norm": 2.2085585594177246,
|
| 25501 |
+
"learning_rate": 1.4017490982756066e-05,
|
| 25502 |
+
"loss": 5.5036,
|
| 25503 |
+
"step": 14568
|
| 25504 |
+
},
|
| 25505 |
+
{
|
| 25506 |
+
"epoch": 0.72,
|
| 25507 |
+
"grad_norm": 2.0381059646606445,
|
| 25508 |
+
"learning_rate": 1.4007609071594446e-05,
|
| 25509 |
+
"loss": 5.5335,
|
| 25510 |
+
"step": 14572
|
| 25511 |
+
},
|
| 25512 |
+
{
|
| 25513 |
+
"epoch": 0.72,
|
| 25514 |
+
"grad_norm": 1.994460940361023,
|
| 25515 |
+
"learning_rate": 1.3997727160432828e-05,
|
| 25516 |
+
"loss": 5.5262,
|
| 25517 |
+
"step": 14576
|
| 25518 |
+
},
|
| 25519 |
+
{
|
| 25520 |
+
"epoch": 0.72,
|
| 25521 |
+
"grad_norm": 1.9527006149291992,
|
| 25522 |
+
"learning_rate": 1.3987845249271208e-05,
|
| 25523 |
+
"loss": 5.4994,
|
| 25524 |
+
"step": 14580
|
| 25525 |
+
},
|
| 25526 |
+
{
|
| 25527 |
+
"epoch": 0.72,
|
| 25528 |
+
"grad_norm": 2.07161808013916,
|
| 25529 |
+
"learning_rate": 1.397796333810959e-05,
|
| 25530 |
+
"loss": 5.5022,
|
| 25531 |
+
"step": 14584
|
| 25532 |
+
},
|
| 25533 |
+
{
|
| 25534 |
+
"epoch": 0.72,
|
| 25535 |
+
"grad_norm": 2.083160638809204,
|
| 25536 |
+
"learning_rate": 1.3968081426947974e-05,
|
| 25537 |
+
"loss": 5.46,
|
| 25538 |
+
"step": 14588
|
| 25539 |
+
},
|
| 25540 |
+
{
|
| 25541 |
+
"epoch": 0.72,
|
| 25542 |
+
"grad_norm": 1.9139795303344727,
|
| 25543 |
+
"learning_rate": 1.3958199515786355e-05,
|
| 25544 |
+
"loss": 5.3918,
|
| 25545 |
+
"step": 14592
|
| 25546 |
+
},
|
| 25547 |
+
{
|
| 25548 |
+
"epoch": 0.72,
|
| 25549 |
+
"grad_norm": 2.016361951828003,
|
| 25550 |
+
"learning_rate": 1.3948317604624737e-05,
|
| 25551 |
+
"loss": 5.3577,
|
| 25552 |
+
"step": 14596
|
| 25553 |
+
},
|
| 25554 |
+
{
|
| 25555 |
+
"epoch": 0.72,
|
| 25556 |
+
"grad_norm": 1.9085208177566528,
|
| 25557 |
+
"learning_rate": 1.3938435693463117e-05,
|
| 25558 |
+
"loss": 5.5672,
|
| 25559 |
+
"step": 14600
|
| 25560 |
+
},
|
| 25561 |
+
{
|
| 25562 |
+
"epoch": 0.72,
|
| 25563 |
+
"grad_norm": 2.156883716583252,
|
| 25564 |
+
"learning_rate": 1.3928553782301499e-05,
|
| 25565 |
+
"loss": 5.4402,
|
| 25566 |
+
"step": 14604
|
| 25567 |
+
},
|
| 25568 |
+
{
|
| 25569 |
+
"epoch": 0.72,
|
| 25570 |
+
"grad_norm": 2.1130199432373047,
|
| 25571 |
+
"learning_rate": 1.391867187113988e-05,
|
| 25572 |
+
"loss": 5.5375,
|
| 25573 |
+
"step": 14608
|
| 25574 |
+
},
|
| 25575 |
+
{
|
| 25576 |
+
"epoch": 0.72,
|
| 25577 |
+
"grad_norm": 2.0779714584350586,
|
| 25578 |
+
"learning_rate": 1.3908789959978261e-05,
|
| 25579 |
+
"loss": 5.4568,
|
| 25580 |
+
"step": 14612
|
| 25581 |
+
},
|
| 25582 |
+
{
|
| 25583 |
+
"epoch": 0.72,
|
| 25584 |
+
"grad_norm": 2.0961239337921143,
|
| 25585 |
+
"learning_rate": 1.3898908048816642e-05,
|
| 25586 |
+
"loss": 5.5331,
|
| 25587 |
+
"step": 14616
|
| 25588 |
+
},
|
| 25589 |
+
{
|
| 25590 |
+
"epoch": 0.72,
|
| 25591 |
+
"grad_norm": 1.9321538209915161,
|
| 25592 |
+
"learning_rate": 1.3889026137655022e-05,
|
| 25593 |
+
"loss": 5.4607,
|
| 25594 |
+
"step": 14620
|
| 25595 |
+
},
|
| 25596 |
+
{
|
| 25597 |
+
"epoch": 0.72,
|
| 25598 |
+
"grad_norm": 2.171811103820801,
|
| 25599 |
+
"learning_rate": 1.3879144226493404e-05,
|
| 25600 |
+
"loss": 5.5705,
|
| 25601 |
+
"step": 14624
|
| 25602 |
+
},
|
| 25603 |
+
{
|
| 25604 |
+
"epoch": 0.72,
|
| 25605 |
+
"grad_norm": 2.0111582279205322,
|
| 25606 |
+
"learning_rate": 1.3869262315331785e-05,
|
| 25607 |
+
"loss": 5.5365,
|
| 25608 |
+
"step": 14628
|
| 25609 |
+
},
|
| 25610 |
+
{
|
| 25611 |
+
"epoch": 0.72,
|
| 25612 |
+
"grad_norm": 2.2230725288391113,
|
| 25613 |
+
"learning_rate": 1.3859380404170167e-05,
|
| 25614 |
+
"loss": 5.4544,
|
| 25615 |
+
"step": 14632
|
| 25616 |
+
},
|
| 25617 |
+
{
|
| 25618 |
+
"epoch": 0.72,
|
| 25619 |
+
"grad_norm": 2.1781952381134033,
|
| 25620 |
+
"learning_rate": 1.3849498493008547e-05,
|
| 25621 |
+
"loss": 5.5207,
|
| 25622 |
+
"step": 14636
|
| 25623 |
+
},
|
| 25624 |
+
{
|
| 25625 |
+
"epoch": 0.72,
|
| 25626 |
+
"grad_norm": 2.1648108959198,
|
| 25627 |
+
"learning_rate": 1.3839616581846929e-05,
|
| 25628 |
+
"loss": 5.542,
|
| 25629 |
+
"step": 14640
|
| 25630 |
+
},
|
| 25631 |
+
{
|
| 25632 |
+
"epoch": 0.72,
|
| 25633 |
+
"grad_norm": 2.008898973464966,
|
| 25634 |
+
"learning_rate": 1.3829734670685313e-05,
|
| 25635 |
+
"loss": 5.4819,
|
| 25636 |
+
"step": 14644
|
| 25637 |
+
},
|
| 25638 |
+
{
|
| 25639 |
+
"epoch": 0.72,
|
| 25640 |
+
"grad_norm": 1.8579167127609253,
|
| 25641 |
+
"learning_rate": 1.3819852759523693e-05,
|
| 25642 |
+
"loss": 5.5307,
|
| 25643 |
+
"step": 14648
|
| 25644 |
+
},
|
| 25645 |
+
{
|
| 25646 |
+
"epoch": 0.72,
|
| 25647 |
+
"grad_norm": 1.951011300086975,
|
| 25648 |
+
"learning_rate": 1.3809970848362075e-05,
|
| 25649 |
+
"loss": 5.5083,
|
| 25650 |
+
"step": 14652
|
| 25651 |
+
},
|
| 25652 |
+
{
|
| 25653 |
+
"epoch": 0.72,
|
| 25654 |
+
"grad_norm": 2.622732639312744,
|
| 25655 |
+
"learning_rate": 1.3800088937200456e-05,
|
| 25656 |
+
"loss": 5.4687,
|
| 25657 |
+
"step": 14656
|
| 25658 |
+
},
|
| 25659 |
+
{
|
| 25660 |
+
"epoch": 0.72,
|
| 25661 |
+
"grad_norm": 1.8241809606552124,
|
| 25662 |
+
"learning_rate": 1.3790207026038838e-05,
|
| 25663 |
+
"loss": 5.5702,
|
| 25664 |
+
"step": 14660
|
| 25665 |
+
},
|
| 25666 |
+
{
|
| 25667 |
+
"epoch": 0.72,
|
| 25668 |
+
"grad_norm": 2.2202184200286865,
|
| 25669 |
+
"learning_rate": 1.3780325114877218e-05,
|
| 25670 |
+
"loss": 5.5717,
|
| 25671 |
+
"step": 14664
|
| 25672 |
+
},
|
| 25673 |
+
{
|
| 25674 |
+
"epoch": 0.72,
|
| 25675 |
+
"grad_norm": 2.125215530395508,
|
| 25676 |
+
"learning_rate": 1.37704432037156e-05,
|
| 25677 |
+
"loss": 5.4724,
|
| 25678 |
+
"step": 14668
|
| 25679 |
+
},
|
| 25680 |
+
{
|
| 25681 |
+
"epoch": 0.72,
|
| 25682 |
+
"grad_norm": 2.0341951847076416,
|
| 25683 |
+
"learning_rate": 1.376056129255398e-05,
|
| 25684 |
+
"loss": 5.4859,
|
| 25685 |
+
"step": 14672
|
| 25686 |
+
},
|
| 25687 |
+
{
|
| 25688 |
+
"epoch": 0.73,
|
| 25689 |
+
"grad_norm": 1.9950518608093262,
|
| 25690 |
+
"learning_rate": 1.3750679381392363e-05,
|
| 25691 |
+
"loss": 5.4983,
|
| 25692 |
+
"step": 14676
|
| 25693 |
+
},
|
| 25694 |
+
{
|
| 25695 |
+
"epoch": 0.73,
|
| 25696 |
+
"grad_norm": 1.9778791666030884,
|
| 25697 |
+
"learning_rate": 1.3740797470230743e-05,
|
| 25698 |
+
"loss": 5.3464,
|
| 25699 |
+
"step": 14680
|
| 25700 |
+
},
|
| 25701 |
+
{
|
| 25702 |
+
"epoch": 0.73,
|
| 25703 |
+
"grad_norm": 2.052849054336548,
|
| 25704 |
+
"learning_rate": 1.3730915559069123e-05,
|
| 25705 |
+
"loss": 5.4766,
|
| 25706 |
+
"step": 14684
|
| 25707 |
+
},
|
| 25708 |
+
{
|
| 25709 |
+
"epoch": 0.73,
|
| 25710 |
+
"grad_norm": 2.0208585262298584,
|
| 25711 |
+
"learning_rate": 1.3721033647907505e-05,
|
| 25712 |
+
"loss": 5.4662,
|
| 25713 |
+
"step": 14688
|
| 25714 |
+
},
|
| 25715 |
+
{
|
| 25716 |
+
"epoch": 0.73,
|
| 25717 |
+
"grad_norm": 2.0381686687469482,
|
| 25718 |
+
"learning_rate": 1.3711151736745886e-05,
|
| 25719 |
+
"loss": 5.4048,
|
| 25720 |
+
"step": 14692
|
| 25721 |
+
},
|
| 25722 |
+
{
|
| 25723 |
+
"epoch": 0.73,
|
| 25724 |
+
"grad_norm": 2.0504868030548096,
|
| 25725 |
+
"learning_rate": 1.3701269825584268e-05,
|
| 25726 |
+
"loss": 5.5422,
|
| 25727 |
+
"step": 14696
|
| 25728 |
+
},
|
| 25729 |
+
{
|
| 25730 |
+
"epoch": 0.73,
|
| 25731 |
+
"grad_norm": 1.9056216478347778,
|
| 25732 |
+
"learning_rate": 1.3691387914422648e-05,
|
| 25733 |
+
"loss": 5.5829,
|
| 25734 |
+
"step": 14700
|
| 25735 |
+
},
|
| 25736 |
+
{
|
| 25737 |
+
"epoch": 0.73,
|
| 25738 |
+
"grad_norm": 2.1584672927856445,
|
| 25739 |
+
"learning_rate": 1.3681506003261032e-05,
|
| 25740 |
+
"loss": 5.4341,
|
| 25741 |
+
"step": 14704
|
| 25742 |
+
},
|
| 25743 |
+
{
|
| 25744 |
+
"epoch": 0.73,
|
| 25745 |
+
"grad_norm": 1.8544187545776367,
|
| 25746 |
+
"learning_rate": 1.3671624092099414e-05,
|
| 25747 |
+
"loss": 5.5552,
|
| 25748 |
+
"step": 14708
|
| 25749 |
+
},
|
| 25750 |
+
{
|
| 25751 |
+
"epoch": 0.73,
|
| 25752 |
+
"grad_norm": 1.9360532760620117,
|
| 25753 |
+
"learning_rate": 1.3661742180937794e-05,
|
| 25754 |
+
"loss": 5.413,
|
| 25755 |
+
"step": 14712
|
| 25756 |
+
},
|
| 25757 |
+
{
|
| 25758 |
+
"epoch": 0.73,
|
| 25759 |
+
"grad_norm": 1.7825835943222046,
|
| 25760 |
+
"learning_rate": 1.3651860269776176e-05,
|
| 25761 |
+
"loss": 5.4916,
|
| 25762 |
+
"step": 14716
|
| 25763 |
+
},
|
| 25764 |
+
{
|
| 25765 |
+
"epoch": 0.73,
|
| 25766 |
+
"grad_norm": 2.0297420024871826,
|
| 25767 |
+
"learning_rate": 1.3641978358614557e-05,
|
| 25768 |
+
"loss": 5.4825,
|
| 25769 |
+
"step": 14720
|
| 25770 |
+
},
|
| 25771 |
+
{
|
| 25772 |
+
"epoch": 0.73,
|
| 25773 |
+
"grad_norm": 2.0158278942108154,
|
| 25774 |
+
"learning_rate": 1.3632096447452939e-05,
|
| 25775 |
+
"loss": 5.4502,
|
| 25776 |
+
"step": 14724
|
| 25777 |
+
},
|
| 25778 |
+
{
|
| 25779 |
+
"epoch": 0.73,
|
| 25780 |
+
"grad_norm": 2.265287160873413,
|
| 25781 |
+
"learning_rate": 1.362221453629132e-05,
|
| 25782 |
+
"loss": 5.5644,
|
| 25783 |
+
"step": 14728
|
| 25784 |
+
},
|
| 25785 |
+
{
|
| 25786 |
+
"epoch": 0.73,
|
| 25787 |
+
"grad_norm": 2.0918755531311035,
|
| 25788 |
+
"learning_rate": 1.3612332625129701e-05,
|
| 25789 |
+
"loss": 5.6084,
|
| 25790 |
+
"step": 14732
|
| 25791 |
+
},
|
| 25792 |
+
{
|
| 25793 |
+
"epoch": 0.73,
|
| 25794 |
+
"grad_norm": 2.141103744506836,
|
| 25795 |
+
"learning_rate": 1.3602450713968082e-05,
|
| 25796 |
+
"loss": 5.4327,
|
| 25797 |
+
"step": 14736
|
| 25798 |
+
},
|
| 25799 |
+
{
|
| 25800 |
+
"epoch": 0.73,
|
| 25801 |
+
"grad_norm": 2.070760488510132,
|
| 25802 |
+
"learning_rate": 1.3592568802806464e-05,
|
| 25803 |
+
"loss": 5.5944,
|
| 25804 |
+
"step": 14740
|
| 25805 |
+
},
|
| 25806 |
+
{
|
| 25807 |
+
"epoch": 0.73,
|
| 25808 |
+
"grad_norm": 1.9426864385604858,
|
| 25809 |
+
"learning_rate": 1.3582686891644844e-05,
|
| 25810 |
+
"loss": 5.4326,
|
| 25811 |
+
"step": 14744
|
| 25812 |
+
},
|
| 25813 |
+
{
|
| 25814 |
+
"epoch": 0.73,
|
| 25815 |
+
"grad_norm": 1.8541244268417358,
|
| 25816 |
+
"learning_rate": 1.3572804980483224e-05,
|
| 25817 |
+
"loss": 5.5438,
|
| 25818 |
+
"step": 14748
|
| 25819 |
+
},
|
| 25820 |
+
{
|
| 25821 |
+
"epoch": 0.73,
|
| 25822 |
+
"grad_norm": 2.081083297729492,
|
| 25823 |
+
"learning_rate": 1.3562923069321606e-05,
|
| 25824 |
+
"loss": 5.5656,
|
| 25825 |
+
"step": 14752
|
| 25826 |
+
},
|
| 25827 |
+
{
|
| 25828 |
+
"epoch": 0.73,
|
| 25829 |
+
"grad_norm": 1.923309326171875,
|
| 25830 |
+
"learning_rate": 1.3553041158159987e-05,
|
| 25831 |
+
"loss": 5.5546,
|
| 25832 |
+
"step": 14756
|
| 25833 |
+
},
|
| 25834 |
+
{
|
| 25835 |
+
"epoch": 0.73,
|
| 25836 |
+
"grad_norm": 1.746881365776062,
|
| 25837 |
+
"learning_rate": 1.3543159246998372e-05,
|
| 25838 |
+
"loss": 5.4799,
|
| 25839 |
+
"step": 14760
|
| 25840 |
+
},
|
| 25841 |
+
{
|
| 25842 |
+
"epoch": 0.73,
|
| 25843 |
+
"grad_norm": 1.91403067111969,
|
| 25844 |
+
"learning_rate": 1.3533277335836753e-05,
|
| 25845 |
+
"loss": 5.4892,
|
| 25846 |
+
"step": 14764
|
| 25847 |
+
},
|
| 25848 |
+
{
|
| 25849 |
+
"epoch": 0.73,
|
| 25850 |
+
"grad_norm": 2.0929675102233887,
|
| 25851 |
+
"learning_rate": 1.3523395424675133e-05,
|
| 25852 |
+
"loss": 5.5852,
|
| 25853 |
+
"step": 14768
|
| 25854 |
+
},
|
| 25855 |
+
{
|
| 25856 |
+
"epoch": 0.73,
|
| 25857 |
+
"grad_norm": 1.9422674179077148,
|
| 25858 |
+
"learning_rate": 1.3513513513513515e-05,
|
| 25859 |
+
"loss": 5.5581,
|
| 25860 |
+
"step": 14772
|
| 25861 |
+
},
|
| 25862 |
+
{
|
| 25863 |
+
"epoch": 0.73,
|
| 25864 |
+
"grad_norm": 1.684131383895874,
|
| 25865 |
+
"learning_rate": 1.3503631602351895e-05,
|
| 25866 |
+
"loss": 5.553,
|
| 25867 |
+
"step": 14776
|
| 25868 |
+
},
|
| 25869 |
+
{
|
| 25870 |
+
"epoch": 0.73,
|
| 25871 |
+
"grad_norm": 1.8600718975067139,
|
| 25872 |
+
"learning_rate": 1.3493749691190278e-05,
|
| 25873 |
+
"loss": 5.5774,
|
| 25874 |
+
"step": 14780
|
| 25875 |
+
},
|
| 25876 |
+
{
|
| 25877 |
+
"epoch": 0.73,
|
| 25878 |
+
"grad_norm": 1.7786033153533936,
|
| 25879 |
+
"learning_rate": 1.3483867780028658e-05,
|
| 25880 |
+
"loss": 5.4997,
|
| 25881 |
+
"step": 14784
|
| 25882 |
+
},
|
| 25883 |
+
{
|
| 25884 |
+
"epoch": 0.73,
|
| 25885 |
+
"grad_norm": 2.239494562149048,
|
| 25886 |
+
"learning_rate": 1.347398586886704e-05,
|
| 25887 |
+
"loss": 5.6186,
|
| 25888 |
+
"step": 14788
|
| 25889 |
+
},
|
| 25890 |
+
{
|
| 25891 |
+
"epoch": 0.73,
|
| 25892 |
+
"grad_norm": 2.078833818435669,
|
| 25893 |
+
"learning_rate": 1.346410395770542e-05,
|
| 25894 |
+
"loss": 5.5609,
|
| 25895 |
+
"step": 14792
|
| 25896 |
+
},
|
| 25897 |
+
{
|
| 25898 |
+
"epoch": 0.73,
|
| 25899 |
+
"grad_norm": 1.867530107498169,
|
| 25900 |
+
"learning_rate": 1.3454222046543802e-05,
|
| 25901 |
+
"loss": 5.5624,
|
| 25902 |
+
"step": 14796
|
| 25903 |
+
},
|
| 25904 |
+
{
|
| 25905 |
+
"epoch": 0.73,
|
| 25906 |
+
"grad_norm": 1.8725926876068115,
|
| 25907 |
+
"learning_rate": 1.3444340135382183e-05,
|
| 25908 |
+
"loss": 5.5675,
|
| 25909 |
+
"step": 14800
|
| 25910 |
+
},
|
| 25911 |
+
{
|
| 25912 |
+
"epoch": 0.73,
|
| 25913 |
+
"grad_norm": 1.9592565298080444,
|
| 25914 |
+
"learning_rate": 1.3434458224220565e-05,
|
| 25915 |
+
"loss": 5.4278,
|
| 25916 |
+
"step": 14804
|
| 25917 |
+
},
|
| 25918 |
+
{
|
| 25919 |
+
"epoch": 0.73,
|
| 25920 |
+
"grad_norm": 2.0653085708618164,
|
| 25921 |
+
"learning_rate": 1.3424576313058945e-05,
|
| 25922 |
+
"loss": 5.4732,
|
| 25923 |
+
"step": 14808
|
| 25924 |
+
},
|
| 25925 |
+
{
|
| 25926 |
+
"epoch": 0.73,
|
| 25927 |
+
"grad_norm": 2.0164620876312256,
|
| 25928 |
+
"learning_rate": 1.3414694401897327e-05,
|
| 25929 |
+
"loss": 5.4636,
|
| 25930 |
+
"step": 14812
|
| 25931 |
+
},
|
| 25932 |
+
{
|
| 25933 |
+
"epoch": 0.73,
|
| 25934 |
+
"grad_norm": 1.9894605875015259,
|
| 25935 |
+
"learning_rate": 1.3404812490735708e-05,
|
| 25936 |
+
"loss": 5.5102,
|
| 25937 |
+
"step": 14816
|
| 25938 |
+
},
|
| 25939 |
+
{
|
| 25940 |
+
"epoch": 0.73,
|
| 25941 |
+
"grad_norm": 2.0743892192840576,
|
| 25942 |
+
"learning_rate": 1.3394930579574091e-05,
|
| 25943 |
+
"loss": 5.4852,
|
| 25944 |
+
"step": 14820
|
| 25945 |
+
},
|
| 25946 |
+
{
|
| 25947 |
+
"epoch": 0.73,
|
| 25948 |
+
"grad_norm": 2.000945806503296,
|
| 25949 |
+
"learning_rate": 1.3385048668412473e-05,
|
| 25950 |
+
"loss": 5.4698,
|
| 25951 |
+
"step": 14824
|
| 25952 |
+
},
|
| 25953 |
+
{
|
| 25954 |
+
"epoch": 0.73,
|
| 25955 |
+
"grad_norm": 2.219153881072998,
|
| 25956 |
+
"learning_rate": 1.3375166757250854e-05,
|
| 25957 |
+
"loss": 5.5966,
|
| 25958 |
+
"step": 14828
|
| 25959 |
+
},
|
| 25960 |
+
{
|
| 25961 |
+
"epoch": 0.73,
|
| 25962 |
+
"grad_norm": 2.0266921520233154,
|
| 25963 |
+
"learning_rate": 1.3365284846089234e-05,
|
| 25964 |
+
"loss": 5.5036,
|
| 25965 |
+
"step": 14832
|
| 25966 |
+
},
|
| 25967 |
+
{
|
| 25968 |
+
"epoch": 0.73,
|
| 25969 |
+
"grad_norm": 2.1432504653930664,
|
| 25970 |
+
"learning_rate": 1.3355402934927616e-05,
|
| 25971 |
+
"loss": 5.5122,
|
| 25972 |
+
"step": 14836
|
| 25973 |
+
},
|
| 25974 |
+
{
|
| 25975 |
+
"epoch": 0.73,
|
| 25976 |
+
"grad_norm": 1.975110411643982,
|
| 25977 |
+
"learning_rate": 1.3345521023765997e-05,
|
| 25978 |
+
"loss": 5.4677,
|
| 25979 |
+
"step": 14840
|
| 25980 |
+
},
|
| 25981 |
+
{
|
| 25982 |
+
"epoch": 0.73,
|
| 25983 |
+
"grad_norm": 2.1612823009490967,
|
| 25984 |
+
"learning_rate": 1.3335639112604379e-05,
|
| 25985 |
+
"loss": 5.544,
|
| 25986 |
+
"step": 14844
|
| 25987 |
+
},
|
| 25988 |
+
{
|
| 25989 |
+
"epoch": 0.73,
|
| 25990 |
+
"grad_norm": 2.0813982486724854,
|
| 25991 |
+
"learning_rate": 1.3325757201442759e-05,
|
| 25992 |
+
"loss": 5.3651,
|
| 25993 |
+
"step": 14848
|
| 25994 |
+
},
|
| 25995 |
+
{
|
| 25996 |
+
"epoch": 0.73,
|
| 25997 |
+
"grad_norm": 1.8772978782653809,
|
| 25998 |
+
"learning_rate": 1.3315875290281141e-05,
|
| 25999 |
+
"loss": 5.4319,
|
| 26000 |
+
"step": 14852
|
| 26001 |
+
},
|
| 26002 |
+
{
|
| 26003 |
+
"epoch": 0.73,
|
| 26004 |
+
"grad_norm": 1.8763823509216309,
|
| 26005 |
+
"learning_rate": 1.3305993379119521e-05,
|
| 26006 |
+
"loss": 5.6508,
|
| 26007 |
+
"step": 14856
|
| 26008 |
+
},
|
| 26009 |
+
{
|
| 26010 |
+
"epoch": 0.73,
|
| 26011 |
+
"grad_norm": 2.0054590702056885,
|
| 26012 |
+
"learning_rate": 1.3296111467957903e-05,
|
| 26013 |
+
"loss": 5.4899,
|
| 26014 |
+
"step": 14860
|
| 26015 |
+
},
|
| 26016 |
+
{
|
| 26017 |
+
"epoch": 0.73,
|
| 26018 |
+
"grad_norm": 2.0155956745147705,
|
| 26019 |
+
"learning_rate": 1.3286229556796284e-05,
|
| 26020 |
+
"loss": 5.4597,
|
| 26021 |
+
"step": 14864
|
| 26022 |
+
},
|
| 26023 |
+
{
|
| 26024 |
+
"epoch": 0.73,
|
| 26025 |
+
"grad_norm": 1.9215456247329712,
|
| 26026 |
+
"learning_rate": 1.3276347645634666e-05,
|
| 26027 |
+
"loss": 5.5587,
|
| 26028 |
+
"step": 14868
|
| 26029 |
+
},
|
| 26030 |
+
{
|
| 26031 |
+
"epoch": 0.73,
|
| 26032 |
+
"grad_norm": 2.2192320823669434,
|
| 26033 |
+
"learning_rate": 1.3266465734473046e-05,
|
| 26034 |
+
"loss": 5.5397,
|
| 26035 |
+
"step": 14872
|
| 26036 |
+
},
|
| 26037 |
+
{
|
| 26038 |
+
"epoch": 0.74,
|
| 26039 |
+
"grad_norm": 1.8037538528442383,
|
| 26040 |
+
"learning_rate": 1.325658382331143e-05,
|
| 26041 |
+
"loss": 5.503,
|
| 26042 |
+
"step": 14876
|
| 26043 |
+
},
|
| 26044 |
+
{
|
| 26045 |
+
"epoch": 0.74,
|
| 26046 |
+
"grad_norm": 2.096147298812866,
|
| 26047 |
+
"learning_rate": 1.3246701912149812e-05,
|
| 26048 |
+
"loss": 5.4808,
|
| 26049 |
+
"step": 14880
|
| 26050 |
+
},
|
| 26051 |
+
{
|
| 26052 |
+
"epoch": 0.74,
|
| 26053 |
+
"grad_norm": 2.0961217880249023,
|
| 26054 |
+
"learning_rate": 1.3236820000988192e-05,
|
| 26055 |
+
"loss": 5.5546,
|
| 26056 |
+
"step": 14884
|
| 26057 |
+
},
|
| 26058 |
+
{
|
| 26059 |
+
"epoch": 0.74,
|
| 26060 |
+
"grad_norm": 2.1718828678131104,
|
| 26061 |
+
"learning_rate": 1.3226938089826574e-05,
|
| 26062 |
+
"loss": 5.4201,
|
| 26063 |
+
"step": 14888
|
| 26064 |
+
},
|
| 26065 |
+
{
|
| 26066 |
+
"epoch": 0.74,
|
| 26067 |
+
"grad_norm": 2.062138795852661,
|
| 26068 |
+
"learning_rate": 1.3217056178664955e-05,
|
| 26069 |
+
"loss": 5.46,
|
| 26070 |
+
"step": 14892
|
| 26071 |
+
},
|
| 26072 |
+
{
|
| 26073 |
+
"epoch": 0.74,
|
| 26074 |
+
"grad_norm": 2.0585410594940186,
|
| 26075 |
+
"learning_rate": 1.3207174267503337e-05,
|
| 26076 |
+
"loss": 5.5545,
|
| 26077 |
+
"step": 14896
|
| 26078 |
+
},
|
| 26079 |
+
{
|
| 26080 |
+
"epoch": 0.74,
|
| 26081 |
+
"grad_norm": 1.8028877973556519,
|
| 26082 |
+
"learning_rate": 1.3197292356341717e-05,
|
| 26083 |
+
"loss": 5.4255,
|
| 26084 |
+
"step": 14900
|
| 26085 |
+
},
|
| 26086 |
+
{
|
| 26087 |
+
"epoch": 0.74,
|
| 26088 |
+
"grad_norm": 2.002005100250244,
|
| 26089 |
+
"learning_rate": 1.3187410445180098e-05,
|
| 26090 |
+
"loss": 5.5847,
|
| 26091 |
+
"step": 14904
|
| 26092 |
+
},
|
| 26093 |
+
{
|
| 26094 |
+
"epoch": 0.74,
|
| 26095 |
+
"grad_norm": 2.1855287551879883,
|
| 26096 |
+
"learning_rate": 1.317752853401848e-05,
|
| 26097 |
+
"loss": 5.536,
|
| 26098 |
+
"step": 14908
|
| 26099 |
+
},
|
| 26100 |
+
{
|
| 26101 |
+
"epoch": 0.74,
|
| 26102 |
+
"grad_norm": 2.193171977996826,
|
| 26103 |
+
"learning_rate": 1.316764662285686e-05,
|
| 26104 |
+
"loss": 5.4059,
|
| 26105 |
+
"step": 14912
|
| 26106 |
+
},
|
| 26107 |
+
{
|
| 26108 |
+
"epoch": 0.74,
|
| 26109 |
+
"grad_norm": 2.0406908988952637,
|
| 26110 |
+
"learning_rate": 1.3157764711695242e-05,
|
| 26111 |
+
"loss": 5.5851,
|
| 26112 |
+
"step": 14916
|
| 26113 |
+
},
|
| 26114 |
+
{
|
| 26115 |
+
"epoch": 0.74,
|
| 26116 |
+
"grad_norm": 2.342682123184204,
|
| 26117 |
+
"learning_rate": 1.3147882800533623e-05,
|
| 26118 |
+
"loss": 5.5068,
|
| 26119 |
+
"step": 14920
|
| 26120 |
+
},
|
| 26121 |
+
{
|
| 26122 |
+
"epoch": 0.74,
|
| 26123 |
+
"grad_norm": 2.0584092140197754,
|
| 26124 |
+
"learning_rate": 1.3138000889372005e-05,
|
| 26125 |
+
"loss": 5.3973,
|
| 26126 |
+
"step": 14924
|
| 26127 |
+
},
|
| 26128 |
+
{
|
| 26129 |
+
"epoch": 0.74,
|
| 26130 |
+
"grad_norm": 1.7771698236465454,
|
| 26131 |
+
"learning_rate": 1.3128118978210385e-05,
|
| 26132 |
+
"loss": 5.4812,
|
| 26133 |
+
"step": 14928
|
| 26134 |
+
},
|
| 26135 |
+
{
|
| 26136 |
+
"epoch": 0.74,
|
| 26137 |
+
"grad_norm": 1.923229694366455,
|
| 26138 |
+
"learning_rate": 1.3118237067048769e-05,
|
| 26139 |
+
"loss": 5.4176,
|
| 26140 |
+
"step": 14932
|
| 26141 |
+
},
|
| 26142 |
+
{
|
| 26143 |
+
"epoch": 0.74,
|
| 26144 |
+
"grad_norm": 1.9571110010147095,
|
| 26145 |
+
"learning_rate": 1.310835515588715e-05,
|
| 26146 |
+
"loss": 5.5085,
|
| 26147 |
+
"step": 14936
|
| 26148 |
+
},
|
| 26149 |
+
{
|
| 26150 |
+
"epoch": 0.74,
|
| 26151 |
+
"grad_norm": 1.9827457666397095,
|
| 26152 |
+
"learning_rate": 1.3098473244725531e-05,
|
| 26153 |
+
"loss": 5.4454,
|
| 26154 |
+
"step": 14940
|
| 26155 |
+
},
|
| 26156 |
+
{
|
| 26157 |
+
"epoch": 0.74,
|
| 26158 |
+
"grad_norm": 2.2986981868743896,
|
| 26159 |
+
"learning_rate": 1.3088591333563913e-05,
|
| 26160 |
+
"loss": 5.562,
|
| 26161 |
+
"step": 14944
|
| 26162 |
+
},
|
| 26163 |
+
{
|
| 26164 |
+
"epoch": 0.74,
|
| 26165 |
+
"grad_norm": 2.087019920349121,
|
| 26166 |
+
"learning_rate": 1.3078709422402294e-05,
|
| 26167 |
+
"loss": 5.5164,
|
| 26168 |
+
"step": 14948
|
| 26169 |
+
},
|
| 26170 |
+
{
|
| 26171 |
+
"epoch": 0.74,
|
| 26172 |
+
"grad_norm": 1.9258410930633545,
|
| 26173 |
+
"learning_rate": 1.3068827511240676e-05,
|
| 26174 |
+
"loss": 5.5252,
|
| 26175 |
+
"step": 14952
|
| 26176 |
+
},
|
| 26177 |
+
{
|
| 26178 |
+
"epoch": 0.74,
|
| 26179 |
+
"grad_norm": 1.833736538887024,
|
| 26180 |
+
"learning_rate": 1.3058945600079056e-05,
|
| 26181 |
+
"loss": 5.4423,
|
| 26182 |
+
"step": 14956
|
| 26183 |
+
},
|
| 26184 |
+
{
|
| 26185 |
+
"epoch": 0.74,
|
| 26186 |
+
"grad_norm": 2.1392202377319336,
|
| 26187 |
+
"learning_rate": 1.3049063688917438e-05,
|
| 26188 |
+
"loss": 5.5637,
|
| 26189 |
+
"step": 14960
|
| 26190 |
+
},
|
| 26191 |
+
{
|
| 26192 |
+
"epoch": 0.74,
|
| 26193 |
+
"grad_norm": 1.9656065702438354,
|
| 26194 |
+
"learning_rate": 1.3039181777755818e-05,
|
| 26195 |
+
"loss": 5.474,
|
| 26196 |
+
"step": 14964
|
| 26197 |
+
},
|
| 26198 |
+
{
|
| 26199 |
+
"epoch": 0.74,
|
| 26200 |
+
"grad_norm": 1.8016414642333984,
|
| 26201 |
+
"learning_rate": 1.3029299866594199e-05,
|
| 26202 |
+
"loss": 5.4101,
|
| 26203 |
+
"step": 14968
|
| 26204 |
+
},
|
| 26205 |
+
{
|
| 26206 |
+
"epoch": 0.74,
|
| 26207 |
+
"grad_norm": 2.141220808029175,
|
| 26208 |
+
"learning_rate": 1.301941795543258e-05,
|
| 26209 |
+
"loss": 5.4747,
|
| 26210 |
+
"step": 14972
|
| 26211 |
+
},
|
| 26212 |
+
{
|
| 26213 |
+
"epoch": 0.74,
|
| 26214 |
+
"grad_norm": 2.1080868244171143,
|
| 26215 |
+
"learning_rate": 1.3009536044270961e-05,
|
| 26216 |
+
"loss": 5.3842,
|
| 26217 |
+
"step": 14976
|
| 26218 |
+
},
|
| 26219 |
+
{
|
| 26220 |
+
"epoch": 0.74,
|
| 26221 |
+
"grad_norm": 1.9357255697250366,
|
| 26222 |
+
"learning_rate": 1.2999654133109343e-05,
|
| 26223 |
+
"loss": 5.5716,
|
| 26224 |
+
"step": 14980
|
| 26225 |
+
},
|
| 26226 |
+
{
|
| 26227 |
+
"epoch": 0.74,
|
| 26228 |
+
"grad_norm": 2.3347742557525635,
|
| 26229 |
+
"learning_rate": 1.2989772221947724e-05,
|
| 26230 |
+
"loss": 5.5935,
|
| 26231 |
+
"step": 14984
|
| 26232 |
+
},
|
| 26233 |
+
{
|
| 26234 |
+
"epoch": 0.74,
|
| 26235 |
+
"grad_norm": 1.999286413192749,
|
| 26236 |
+
"learning_rate": 1.2979890310786106e-05,
|
| 26237 |
+
"loss": 5.4976,
|
| 26238 |
+
"step": 14988
|
| 26239 |
+
},
|
| 26240 |
+
{
|
| 26241 |
+
"epoch": 0.74,
|
| 26242 |
+
"grad_norm": 2.040515184402466,
|
| 26243 |
+
"learning_rate": 1.297000839962449e-05,
|
| 26244 |
+
"loss": 5.6029,
|
| 26245 |
+
"step": 14992
|
| 26246 |
+
},
|
| 26247 |
+
{
|
| 26248 |
+
"epoch": 0.74,
|
| 26249 |
+
"grad_norm": 1.9321497678756714,
|
| 26250 |
+
"learning_rate": 1.296012648846287e-05,
|
| 26251 |
+
"loss": 5.503,
|
| 26252 |
+
"step": 14996
|
| 26253 |
+
},
|
| 26254 |
+
{
|
| 26255 |
+
"epoch": 0.74,
|
| 26256 |
+
"grad_norm": 1.834435224533081,
|
| 26257 |
+
"learning_rate": 1.2950244577301252e-05,
|
| 26258 |
+
"loss": 5.5076,
|
| 26259 |
+
"step": 15000
|
| 26260 |
+
},
|
| 26261 |
+
{
|
| 26262 |
+
"epoch": 0.74,
|
| 26263 |
+
"grad_norm": 2.410569906234741,
|
| 26264 |
+
"learning_rate": 1.2940362666139632e-05,
|
| 26265 |
+
"loss": 5.6112,
|
| 26266 |
+
"step": 15004
|
| 26267 |
+
},
|
| 26268 |
+
{
|
| 26269 |
+
"epoch": 0.74,
|
| 26270 |
+
"grad_norm": 2.110943555831909,
|
| 26271 |
+
"learning_rate": 1.2930480754978014e-05,
|
| 26272 |
+
"loss": 5.4589,
|
| 26273 |
+
"step": 15008
|
| 26274 |
+
},
|
| 26275 |
+
{
|
| 26276 |
+
"epoch": 0.74,
|
| 26277 |
+
"grad_norm": 2.2104272842407227,
|
| 26278 |
+
"learning_rate": 1.2920598843816395e-05,
|
| 26279 |
+
"loss": 5.4494,
|
| 26280 |
+
"step": 15012
|
| 26281 |
+
},
|
| 26282 |
+
{
|
| 26283 |
+
"epoch": 0.74,
|
| 26284 |
+
"grad_norm": 2.0038228034973145,
|
| 26285 |
+
"learning_rate": 1.2910716932654777e-05,
|
| 26286 |
+
"loss": 5.4229,
|
| 26287 |
+
"step": 15016
|
| 26288 |
+
},
|
| 26289 |
+
{
|
| 26290 |
+
"epoch": 0.74,
|
| 26291 |
+
"grad_norm": 2.0080738067626953,
|
| 26292 |
+
"learning_rate": 1.2900835021493157e-05,
|
| 26293 |
+
"loss": 5.5645,
|
| 26294 |
+
"step": 15020
|
| 26295 |
+
},
|
| 26296 |
+
{
|
| 26297 |
+
"epoch": 0.74,
|
| 26298 |
+
"grad_norm": 2.075934886932373,
|
| 26299 |
+
"learning_rate": 1.2890953110331539e-05,
|
| 26300 |
+
"loss": 5.6104,
|
| 26301 |
+
"step": 15024
|
| 26302 |
+
},
|
| 26303 |
+
{
|
| 26304 |
+
"epoch": 0.74,
|
| 26305 |
+
"grad_norm": 1.9630203247070312,
|
| 26306 |
+
"learning_rate": 1.288107119916992e-05,
|
| 26307 |
+
"loss": 5.4888,
|
| 26308 |
+
"step": 15028
|
| 26309 |
+
},
|
| 26310 |
+
{
|
| 26311 |
+
"epoch": 0.74,
|
| 26312 |
+
"grad_norm": 2.1975886821746826,
|
| 26313 |
+
"learning_rate": 1.28711892880083e-05,
|
| 26314 |
+
"loss": 5.5283,
|
| 26315 |
+
"step": 15032
|
| 26316 |
+
},
|
| 26317 |
+
{
|
| 26318 |
+
"epoch": 0.74,
|
| 26319 |
+
"grad_norm": 1.9507403373718262,
|
| 26320 |
+
"learning_rate": 1.2861307376846682e-05,
|
| 26321 |
+
"loss": 5.6203,
|
| 26322 |
+
"step": 15036
|
| 26323 |
+
},
|
| 26324 |
+
{
|
| 26325 |
+
"epoch": 0.74,
|
| 26326 |
+
"grad_norm": 2.0065085887908936,
|
| 26327 |
+
"learning_rate": 1.2851425465685062e-05,
|
| 26328 |
+
"loss": 5.4493,
|
| 26329 |
+
"step": 15040
|
| 26330 |
+
},
|
| 26331 |
+
{
|
| 26332 |
+
"epoch": 0.74,
|
| 26333 |
+
"grad_norm": 2.170250177383423,
|
| 26334 |
+
"learning_rate": 1.2841543554523444e-05,
|
| 26335 |
+
"loss": 5.5062,
|
| 26336 |
+
"step": 15044
|
| 26337 |
+
},
|
| 26338 |
+
{
|
| 26339 |
+
"epoch": 0.74,
|
| 26340 |
+
"grad_norm": 1.9755939245224,
|
| 26341 |
+
"learning_rate": 1.2831661643361828e-05,
|
| 26342 |
+
"loss": 5.4901,
|
| 26343 |
+
"step": 15048
|
| 26344 |
+
},
|
| 26345 |
+
{
|
| 26346 |
+
"epoch": 0.74,
|
| 26347 |
+
"grad_norm": 2.0391461849212646,
|
| 26348 |
+
"learning_rate": 1.2821779732200208e-05,
|
| 26349 |
+
"loss": 5.4896,
|
| 26350 |
+
"step": 15052
|
| 26351 |
+
},
|
| 26352 |
+
{
|
| 26353 |
+
"epoch": 0.74,
|
| 26354 |
+
"grad_norm": 1.8779929876327515,
|
| 26355 |
+
"learning_rate": 1.281189782103859e-05,
|
| 26356 |
+
"loss": 5.4311,
|
| 26357 |
+
"step": 15056
|
| 26358 |
+
},
|
| 26359 |
+
{
|
| 26360 |
+
"epoch": 0.74,
|
| 26361 |
+
"grad_norm": 2.123504400253296,
|
| 26362 |
+
"learning_rate": 1.2802015909876971e-05,
|
| 26363 |
+
"loss": 5.5117,
|
| 26364 |
+
"step": 15060
|
| 26365 |
+
},
|
| 26366 |
+
{
|
| 26367 |
+
"epoch": 0.74,
|
| 26368 |
+
"grad_norm": 1.885321021080017,
|
| 26369 |
+
"learning_rate": 1.2792133998715353e-05,
|
| 26370 |
+
"loss": 5.515,
|
| 26371 |
+
"step": 15064
|
| 26372 |
+
},
|
| 26373 |
+
{
|
| 26374 |
+
"epoch": 0.74,
|
| 26375 |
+
"grad_norm": 2.281585454940796,
|
| 26376 |
+
"learning_rate": 1.2782252087553733e-05,
|
| 26377 |
+
"loss": 5.4979,
|
| 26378 |
+
"step": 15068
|
| 26379 |
+
},
|
| 26380 |
+
{
|
| 26381 |
+
"epoch": 0.74,
|
| 26382 |
+
"grad_norm": 2.1277706623077393,
|
| 26383 |
+
"learning_rate": 1.2772370176392115e-05,
|
| 26384 |
+
"loss": 5.4285,
|
| 26385 |
+
"step": 15072
|
| 26386 |
+
},
|
| 26387 |
+
{
|
| 26388 |
+
"epoch": 0.74,
|
| 26389 |
+
"grad_norm": 2.0085339546203613,
|
| 26390 |
+
"learning_rate": 1.2762488265230496e-05,
|
| 26391 |
+
"loss": 5.4741,
|
| 26392 |
+
"step": 15076
|
| 26393 |
+
},
|
| 26394 |
+
{
|
| 26395 |
+
"epoch": 0.75,
|
| 26396 |
+
"grad_norm": 2.0528817176818848,
|
| 26397 |
+
"learning_rate": 1.2752606354068878e-05,
|
| 26398 |
+
"loss": 5.5861,
|
| 26399 |
+
"step": 15080
|
| 26400 |
+
},
|
| 26401 |
+
{
|
| 26402 |
+
"epoch": 0.75,
|
| 26403 |
+
"grad_norm": 2.0713131427764893,
|
| 26404 |
+
"learning_rate": 1.2742724442907258e-05,
|
| 26405 |
+
"loss": 5.5236,
|
| 26406 |
+
"step": 15084
|
| 26407 |
+
},
|
| 26408 |
+
{
|
| 26409 |
+
"epoch": 0.75,
|
| 26410 |
+
"grad_norm": 1.9570958614349365,
|
| 26411 |
+
"learning_rate": 1.273284253174564e-05,
|
| 26412 |
+
"loss": 5.4983,
|
| 26413 |
+
"step": 15088
|
| 26414 |
+
},
|
| 26415 |
+
{
|
| 26416 |
+
"epoch": 0.75,
|
| 26417 |
+
"grad_norm": 1.8373630046844482,
|
| 26418 |
+
"learning_rate": 1.272296062058402e-05,
|
| 26419 |
+
"loss": 5.4913,
|
| 26420 |
+
"step": 15092
|
| 26421 |
+
},
|
| 26422 |
+
{
|
| 26423 |
+
"epoch": 0.75,
|
| 26424 |
+
"grad_norm": 2.0442957878112793,
|
| 26425 |
+
"learning_rate": 1.2713078709422403e-05,
|
| 26426 |
+
"loss": 5.4757,
|
| 26427 |
+
"step": 15096
|
| 26428 |
+
},
|
| 26429 |
+
{
|
| 26430 |
+
"epoch": 0.75,
|
| 26431 |
+
"grad_norm": 2.090407609939575,
|
| 26432 |
+
"learning_rate": 1.2703196798260783e-05,
|
| 26433 |
+
"loss": 5.4915,
|
| 26434 |
+
"step": 15100
|
| 26435 |
+
},
|
| 26436 |
+
{
|
| 26437 |
+
"epoch": 0.75,
|
| 26438 |
+
"grad_norm": 2.1005630493164062,
|
| 26439 |
+
"learning_rate": 1.2693314887099167e-05,
|
| 26440 |
+
"loss": 5.5734,
|
| 26441 |
+
"step": 15104
|
| 26442 |
+
},
|
| 26443 |
+
{
|
| 26444 |
+
"epoch": 0.75,
|
| 26445 |
+
"grad_norm": 2.062366008758545,
|
| 26446 |
+
"learning_rate": 1.2683432975937549e-05,
|
| 26447 |
+
"loss": 5.5959,
|
| 26448 |
+
"step": 15108
|
| 26449 |
+
},
|
| 26450 |
+
{
|
| 26451 |
+
"epoch": 0.75,
|
| 26452 |
+
"grad_norm": 1.9203088283538818,
|
| 26453 |
+
"learning_rate": 1.267355106477593e-05,
|
| 26454 |
+
"loss": 5.4388,
|
| 26455 |
+
"step": 15112
|
| 26456 |
+
},
|
| 26457 |
+
{
|
| 26458 |
+
"epoch": 0.75,
|
| 26459 |
+
"grad_norm": 2.0604312419891357,
|
| 26460 |
+
"learning_rate": 1.266366915361431e-05,
|
| 26461 |
+
"loss": 5.5115,
|
| 26462 |
+
"step": 15116
|
| 26463 |
+
},
|
| 26464 |
+
{
|
| 26465 |
+
"epoch": 0.75,
|
| 26466 |
+
"grad_norm": 2.230130672454834,
|
| 26467 |
+
"learning_rate": 1.2653787242452692e-05,
|
| 26468 |
+
"loss": 5.596,
|
| 26469 |
+
"step": 15120
|
| 26470 |
+
},
|
| 26471 |
+
{
|
| 26472 |
+
"epoch": 0.75,
|
| 26473 |
+
"grad_norm": 2.0392181873321533,
|
| 26474 |
+
"learning_rate": 1.2643905331291072e-05,
|
| 26475 |
+
"loss": 5.57,
|
| 26476 |
+
"step": 15124
|
| 26477 |
+
},
|
| 26478 |
+
{
|
| 26479 |
+
"epoch": 0.75,
|
| 26480 |
+
"grad_norm": 2.253793954849243,
|
| 26481 |
+
"learning_rate": 1.2634023420129454e-05,
|
| 26482 |
+
"loss": 5.5402,
|
| 26483 |
+
"step": 15128
|
| 26484 |
+
},
|
| 26485 |
+
{
|
| 26486 |
+
"epoch": 0.75,
|
| 26487 |
+
"grad_norm": 1.9613460302352905,
|
| 26488 |
+
"learning_rate": 1.2624141508967834e-05,
|
| 26489 |
+
"loss": 5.3467,
|
| 26490 |
+
"step": 15132
|
| 26491 |
+
},
|
| 26492 |
+
{
|
| 26493 |
+
"epoch": 0.75,
|
| 26494 |
+
"grad_norm": 1.8354169130325317,
|
| 26495 |
+
"learning_rate": 1.2614259597806216e-05,
|
| 26496 |
+
"loss": 5.5377,
|
| 26497 |
+
"step": 15136
|
| 26498 |
+
},
|
| 26499 |
+
{
|
| 26500 |
+
"epoch": 0.75,
|
| 26501 |
+
"grad_norm": 1.8670376539230347,
|
| 26502 |
+
"learning_rate": 1.2604377686644597e-05,
|
| 26503 |
+
"loss": 5.4526,
|
| 26504 |
+
"step": 15140
|
| 26505 |
+
},
|
| 26506 |
+
{
|
| 26507 |
+
"epoch": 0.75,
|
| 26508 |
+
"grad_norm": 2.1439993381500244,
|
| 26509 |
+
"learning_rate": 1.2594495775482979e-05,
|
| 26510 |
+
"loss": 5.4915,
|
| 26511 |
+
"step": 15144
|
| 26512 |
+
},
|
| 26513 |
+
{
|
| 26514 |
+
"epoch": 0.75,
|
| 26515 |
+
"grad_norm": 1.9621843099594116,
|
| 26516 |
+
"learning_rate": 1.258461386432136e-05,
|
| 26517 |
+
"loss": 5.415,
|
| 26518 |
+
"step": 15148
|
| 26519 |
+
},
|
| 26520 |
+
{
|
| 26521 |
+
"epoch": 0.75,
|
| 26522 |
+
"grad_norm": 2.0130698680877686,
|
| 26523 |
+
"learning_rate": 1.2574731953159741e-05,
|
| 26524 |
+
"loss": 5.5646,
|
| 26525 |
+
"step": 15152
|
| 26526 |
+
},
|
| 26527 |
+
{
|
| 26528 |
+
"epoch": 0.75,
|
| 26529 |
+
"grad_norm": 2.1963608264923096,
|
| 26530 |
+
"learning_rate": 1.2564850041998122e-05,
|
| 26531 |
+
"loss": 5.5678,
|
| 26532 |
+
"step": 15156
|
| 26533 |
+
},
|
| 26534 |
+
{
|
| 26535 |
+
"epoch": 0.75,
|
| 26536 |
+
"grad_norm": 2.007394790649414,
|
| 26537 |
+
"learning_rate": 1.2554968130836504e-05,
|
| 26538 |
+
"loss": 5.4874,
|
| 26539 |
+
"step": 15160
|
| 26540 |
+
},
|
| 26541 |
+
{
|
| 26542 |
+
"epoch": 0.75,
|
| 26543 |
+
"grad_norm": 2.03348708152771,
|
| 26544 |
+
"learning_rate": 1.2545086219674888e-05,
|
| 26545 |
+
"loss": 5.3595,
|
| 26546 |
+
"step": 15164
|
| 26547 |
+
},
|
| 26548 |
+
{
|
| 26549 |
+
"epoch": 0.75,
|
| 26550 |
+
"grad_norm": 2.1427929401397705,
|
| 26551 |
+
"learning_rate": 1.2535204308513268e-05,
|
| 26552 |
+
"loss": 5.5416,
|
| 26553 |
+
"step": 15168
|
| 26554 |
+
},
|
| 26555 |
+
{
|
| 26556 |
+
"epoch": 0.75,
|
| 26557 |
+
"grad_norm": 2.0188114643096924,
|
| 26558 |
+
"learning_rate": 1.252532239735165e-05,
|
| 26559 |
+
"loss": 5.4923,
|
| 26560 |
+
"step": 15172
|
| 26561 |
+
},
|
| 26562 |
+
{
|
| 26563 |
+
"epoch": 0.75,
|
| 26564 |
+
"grad_norm": 1.9621504545211792,
|
| 26565 |
+
"learning_rate": 1.251544048619003e-05,
|
| 26566 |
+
"loss": 5.5167,
|
| 26567 |
+
"step": 15176
|
| 26568 |
+
},
|
| 26569 |
+
{
|
| 26570 |
+
"epoch": 0.75,
|
| 26571 |
+
"grad_norm": 2.1401383876800537,
|
| 26572 |
+
"learning_rate": 1.2505558575028412e-05,
|
| 26573 |
+
"loss": 5.3883,
|
| 26574 |
+
"step": 15180
|
| 26575 |
+
},
|
| 26576 |
+
{
|
| 26577 |
+
"epoch": 0.75,
|
| 26578 |
+
"grad_norm": 2.0091779232025146,
|
| 26579 |
+
"learning_rate": 1.2495676663866793e-05,
|
| 26580 |
+
"loss": 5.5855,
|
| 26581 |
+
"step": 15184
|
| 26582 |
+
},
|
| 26583 |
+
{
|
| 26584 |
+
"epoch": 0.75,
|
| 26585 |
+
"grad_norm": 2.0588550567626953,
|
| 26586 |
+
"learning_rate": 1.2485794752705173e-05,
|
| 26587 |
+
"loss": 5.4961,
|
| 26588 |
+
"step": 15188
|
| 26589 |
+
},
|
| 26590 |
+
{
|
| 26591 |
+
"epoch": 0.75,
|
| 26592 |
+
"grad_norm": 2.0023372173309326,
|
| 26593 |
+
"learning_rate": 1.2475912841543555e-05,
|
| 26594 |
+
"loss": 5.3909,
|
| 26595 |
+
"step": 15192
|
| 26596 |
+
},
|
| 26597 |
+
{
|
| 26598 |
+
"epoch": 0.75,
|
| 26599 |
+
"grad_norm": 2.395747184753418,
|
| 26600 |
+
"learning_rate": 1.2466030930381936e-05,
|
| 26601 |
+
"loss": 5.4807,
|
| 26602 |
+
"step": 15196
|
| 26603 |
+
},
|
| 26604 |
+
{
|
| 26605 |
+
"epoch": 0.75,
|
| 26606 |
+
"grad_norm": 1.8098511695861816,
|
| 26607 |
+
"learning_rate": 1.2456149019220318e-05,
|
| 26608 |
+
"loss": 5.4387,
|
| 26609 |
+
"step": 15200
|
| 26610 |
+
},
|
| 26611 |
+
{
|
| 26612 |
+
"epoch": 0.75,
|
| 26613 |
+
"grad_norm": 1.9918867349624634,
|
| 26614 |
+
"learning_rate": 1.24462671080587e-05,
|
| 26615 |
+
"loss": 5.396,
|
| 26616 |
+
"step": 15204
|
| 26617 |
+
},
|
| 26618 |
+
{
|
| 26619 |
+
"epoch": 0.75,
|
| 26620 |
+
"grad_norm": 2.0140202045440674,
|
| 26621 |
+
"learning_rate": 1.2436385196897082e-05,
|
| 26622 |
+
"loss": 5.4098,
|
| 26623 |
+
"step": 15208
|
| 26624 |
+
},
|
| 26625 |
+
{
|
| 26626 |
+
"epoch": 0.75,
|
| 26627 |
+
"grad_norm": 1.9749282598495483,
|
| 26628 |
+
"learning_rate": 1.2426503285735462e-05,
|
| 26629 |
+
"loss": 5.4362,
|
| 26630 |
+
"step": 15212
|
| 26631 |
+
},
|
| 26632 |
+
{
|
| 26633 |
+
"epoch": 0.75,
|
| 26634 |
+
"grad_norm": 2.1992135047912598,
|
| 26635 |
+
"learning_rate": 1.2416621374573842e-05,
|
| 26636 |
+
"loss": 5.5986,
|
| 26637 |
+
"step": 15216
|
| 26638 |
+
},
|
| 26639 |
+
{
|
| 26640 |
+
"epoch": 0.75,
|
| 26641 |
+
"grad_norm": 2.3077468872070312,
|
| 26642 |
+
"learning_rate": 1.2406739463412225e-05,
|
| 26643 |
+
"loss": 5.5058,
|
| 26644 |
+
"step": 15220
|
| 26645 |
+
},
|
| 26646 |
+
{
|
| 26647 |
+
"epoch": 0.75,
|
| 26648 |
+
"grad_norm": 1.964931845664978,
|
| 26649 |
+
"learning_rate": 1.2396857552250605e-05,
|
| 26650 |
+
"loss": 5.3997,
|
| 26651 |
+
"step": 15224
|
| 26652 |
+
},
|
| 26653 |
+
{
|
| 26654 |
+
"epoch": 0.75,
|
| 26655 |
+
"grad_norm": 1.9411903619766235,
|
| 26656 |
+
"learning_rate": 1.2386975641088987e-05,
|
| 26657 |
+
"loss": 5.4008,
|
| 26658 |
+
"step": 15228
|
| 26659 |
+
},
|
| 26660 |
+
{
|
| 26661 |
+
"epoch": 0.75,
|
| 26662 |
+
"grad_norm": 2.01727557182312,
|
| 26663 |
+
"learning_rate": 1.2377093729927367e-05,
|
| 26664 |
+
"loss": 5.6187,
|
| 26665 |
+
"step": 15232
|
| 26666 |
+
},
|
| 26667 |
+
{
|
| 26668 |
+
"epoch": 0.75,
|
| 26669 |
+
"grad_norm": 1.9451463222503662,
|
| 26670 |
+
"learning_rate": 1.2367211818765751e-05,
|
| 26671 |
+
"loss": 5.5272,
|
| 26672 |
+
"step": 15236
|
| 26673 |
+
},
|
| 26674 |
+
{
|
| 26675 |
+
"epoch": 0.75,
|
| 26676 |
+
"grad_norm": 2.1012978553771973,
|
| 26677 |
+
"learning_rate": 1.2357329907604131e-05,
|
| 26678 |
+
"loss": 5.3657,
|
| 26679 |
+
"step": 15240
|
| 26680 |
+
},
|
| 26681 |
+
{
|
| 26682 |
+
"epoch": 0.75,
|
| 26683 |
+
"grad_norm": 2.219510078430176,
|
| 26684 |
+
"learning_rate": 1.2347447996442513e-05,
|
| 26685 |
+
"loss": 5.526,
|
| 26686 |
+
"step": 15244
|
| 26687 |
+
},
|
| 26688 |
+
{
|
| 26689 |
+
"epoch": 0.75,
|
| 26690 |
+
"grad_norm": 2.013444185256958,
|
| 26691 |
+
"learning_rate": 1.2337566085280894e-05,
|
| 26692 |
+
"loss": 5.4749,
|
| 26693 |
+
"step": 15248
|
| 26694 |
+
},
|
| 26695 |
+
{
|
| 26696 |
+
"epoch": 0.75,
|
| 26697 |
+
"grad_norm": 1.7549456357955933,
|
| 26698 |
+
"learning_rate": 1.2327684174119274e-05,
|
| 26699 |
+
"loss": 5.3478,
|
| 26700 |
+
"step": 15252
|
| 26701 |
+
},
|
| 26702 |
+
{
|
| 26703 |
+
"epoch": 0.75,
|
| 26704 |
+
"grad_norm": 1.8471717834472656,
|
| 26705 |
+
"learning_rate": 1.2317802262957656e-05,
|
| 26706 |
+
"loss": 5.5201,
|
| 26707 |
+
"step": 15256
|
| 26708 |
+
},
|
| 26709 |
+
{
|
| 26710 |
+
"epoch": 0.75,
|
| 26711 |
+
"grad_norm": 2.324028491973877,
|
| 26712 |
+
"learning_rate": 1.2307920351796037e-05,
|
| 26713 |
+
"loss": 5.5589,
|
| 26714 |
+
"step": 15260
|
| 26715 |
+
},
|
| 26716 |
+
{
|
| 26717 |
+
"epoch": 0.75,
|
| 26718 |
+
"grad_norm": 2.0255486965179443,
|
| 26719 |
+
"learning_rate": 1.229803844063442e-05,
|
| 26720 |
+
"loss": 5.5575,
|
| 26721 |
+
"step": 15264
|
| 26722 |
+
},
|
| 26723 |
+
{
|
| 26724 |
+
"epoch": 0.75,
|
| 26725 |
+
"grad_norm": 2.1873011589050293,
|
| 26726 |
+
"learning_rate": 1.22881565294728e-05,
|
| 26727 |
+
"loss": 5.5165,
|
| 26728 |
+
"step": 15268
|
| 26729 |
+
},
|
| 26730 |
+
{
|
| 26731 |
+
"epoch": 0.75,
|
| 26732 |
+
"grad_norm": 2.1346065998077393,
|
| 26733 |
+
"learning_rate": 1.2278274618311183e-05,
|
| 26734 |
+
"loss": 5.5053,
|
| 26735 |
+
"step": 15272
|
| 26736 |
+
},
|
| 26737 |
+
{
|
| 26738 |
+
"epoch": 0.75,
|
| 26739 |
+
"grad_norm": 2.1076815128326416,
|
| 26740 |
+
"learning_rate": 1.2268392707149563e-05,
|
| 26741 |
+
"loss": 5.4105,
|
| 26742 |
+
"step": 15276
|
| 26743 |
+
},
|
| 26744 |
+
{
|
| 26745 |
+
"epoch": 0.75,
|
| 26746 |
+
"grad_norm": 1.9122978448867798,
|
| 26747 |
+
"learning_rate": 1.2258510795987945e-05,
|
| 26748 |
+
"loss": 5.4235,
|
| 26749 |
+
"step": 15280
|
| 26750 |
+
},
|
| 26751 |
+
{
|
| 26752 |
+
"epoch": 0.76,
|
| 26753 |
+
"grad_norm": 2.054979085922241,
|
| 26754 |
+
"learning_rate": 1.2248628884826326e-05,
|
| 26755 |
+
"loss": 5.5838,
|
| 26756 |
+
"step": 15284
|
| 26757 |
+
},
|
| 26758 |
+
{
|
| 26759 |
+
"epoch": 0.76,
|
| 26760 |
+
"grad_norm": 1.9995005130767822,
|
| 26761 |
+
"learning_rate": 1.2238746973664706e-05,
|
| 26762 |
+
"loss": 5.4584,
|
| 26763 |
+
"step": 15288
|
| 26764 |
+
},
|
| 26765 |
+
{
|
| 26766 |
+
"epoch": 0.76,
|
| 26767 |
+
"grad_norm": 2.0757248401641846,
|
| 26768 |
+
"learning_rate": 1.222886506250309e-05,
|
| 26769 |
+
"loss": 5.5064,
|
| 26770 |
+
"step": 15292
|
| 26771 |
+
},
|
| 26772 |
+
{
|
| 26773 |
+
"epoch": 0.76,
|
| 26774 |
+
"grad_norm": 1.831465721130371,
|
| 26775 |
+
"learning_rate": 1.221898315134147e-05,
|
| 26776 |
+
"loss": 5.5416,
|
| 26777 |
+
"step": 15296
|
| 26778 |
+
},
|
| 26779 |
+
{
|
| 26780 |
+
"epoch": 0.76,
|
| 26781 |
+
"grad_norm": 2.3364105224609375,
|
| 26782 |
+
"learning_rate": 1.2209101240179852e-05,
|
| 26783 |
+
"loss": 5.5332,
|
| 26784 |
+
"step": 15300
|
| 26785 |
+
},
|
| 26786 |
+
{
|
| 26787 |
+
"epoch": 0.76,
|
| 26788 |
+
"grad_norm": 1.9546422958374023,
|
| 26789 |
+
"learning_rate": 1.2199219329018233e-05,
|
| 26790 |
+
"loss": 5.4726,
|
| 26791 |
+
"step": 15304
|
| 26792 |
+
},
|
| 26793 |
+
{
|
| 26794 |
+
"epoch": 0.76,
|
| 26795 |
+
"grad_norm": 1.8395370244979858,
|
| 26796 |
+
"learning_rate": 1.2189337417856615e-05,
|
| 26797 |
+
"loss": 5.4443,
|
| 26798 |
+
"step": 15308
|
| 26799 |
+
},
|
| 26800 |
+
{
|
| 26801 |
+
"epoch": 0.76,
|
| 26802 |
+
"grad_norm": 2.080458402633667,
|
| 26803 |
+
"learning_rate": 1.2179455506694995e-05,
|
| 26804 |
+
"loss": 5.5154,
|
| 26805 |
+
"step": 15312
|
| 26806 |
+
},
|
| 26807 |
+
{
|
| 26808 |
+
"epoch": 0.76,
|
| 26809 |
+
"grad_norm": 2.017116069793701,
|
| 26810 |
+
"learning_rate": 1.2169573595533375e-05,
|
| 26811 |
+
"loss": 5.5048,
|
| 26812 |
+
"step": 15316
|
| 26813 |
+
},
|
| 26814 |
+
{
|
| 26815 |
+
"epoch": 0.76,
|
| 26816 |
+
"grad_norm": 2.0325398445129395,
|
| 26817 |
+
"learning_rate": 1.2159691684371759e-05,
|
| 26818 |
+
"loss": 5.3668,
|
| 26819 |
+
"step": 15320
|
| 26820 |
+
},
|
| 26821 |
+
{
|
| 26822 |
+
"epoch": 0.76,
|
| 26823 |
+
"grad_norm": 2.246100664138794,
|
| 26824 |
+
"learning_rate": 1.214980977321014e-05,
|
| 26825 |
+
"loss": 5.5176,
|
| 26826 |
+
"step": 15324
|
| 26827 |
+
},
|
| 26828 |
+
{
|
| 26829 |
+
"epoch": 0.76,
|
| 26830 |
+
"grad_norm": 2.0287086963653564,
|
| 26831 |
+
"learning_rate": 1.2139927862048522e-05,
|
| 26832 |
+
"loss": 5.5352,
|
| 26833 |
+
"step": 15328
|
| 26834 |
+
},
|
| 26835 |
+
{
|
| 26836 |
+
"epoch": 0.76,
|
| 26837 |
+
"grad_norm": 2.278211832046509,
|
| 26838 |
+
"learning_rate": 1.2130045950886902e-05,
|
| 26839 |
+
"loss": 5.441,
|
| 26840 |
+
"step": 15332
|
| 26841 |
+
},
|
| 26842 |
+
{
|
| 26843 |
+
"epoch": 0.76,
|
| 26844 |
+
"grad_norm": 2.143902063369751,
|
| 26845 |
+
"learning_rate": 1.2120164039725284e-05,
|
| 26846 |
+
"loss": 5.5263,
|
| 26847 |
+
"step": 15336
|
| 26848 |
+
},
|
| 26849 |
+
{
|
| 26850 |
+
"epoch": 0.76,
|
| 26851 |
+
"grad_norm": 2.0122079849243164,
|
| 26852 |
+
"learning_rate": 1.2110282128563664e-05,
|
| 26853 |
+
"loss": 5.3925,
|
| 26854 |
+
"step": 15340
|
| 26855 |
+
},
|
| 26856 |
+
{
|
| 26857 |
+
"epoch": 0.76,
|
| 26858 |
+
"grad_norm": 2.094726324081421,
|
| 26859 |
+
"learning_rate": 1.2100400217402046e-05,
|
| 26860 |
+
"loss": 5.4837,
|
| 26861 |
+
"step": 15344
|
| 26862 |
+
},
|
| 26863 |
+
{
|
| 26864 |
+
"epoch": 0.76,
|
| 26865 |
+
"grad_norm": 2.0904650688171387,
|
| 26866 |
+
"learning_rate": 1.2090518306240428e-05,
|
| 26867 |
+
"loss": 5.4732,
|
| 26868 |
+
"step": 15348
|
| 26869 |
+
},
|
| 26870 |
+
{
|
| 26871 |
+
"epoch": 0.76,
|
| 26872 |
+
"grad_norm": 1.869597315788269,
|
| 26873 |
+
"learning_rate": 1.2080636395078809e-05,
|
| 26874 |
+
"loss": 5.4292,
|
| 26875 |
+
"step": 15352
|
| 26876 |
+
},
|
| 26877 |
+
{
|
| 26878 |
+
"epoch": 0.76,
|
| 26879 |
+
"grad_norm": 2.0959110260009766,
|
| 26880 |
+
"learning_rate": 1.2070754483917191e-05,
|
| 26881 |
+
"loss": 5.5033,
|
| 26882 |
+
"step": 15356
|
| 26883 |
+
},
|
| 26884 |
+
{
|
| 26885 |
+
"epoch": 0.76,
|
| 26886 |
+
"grad_norm": 2.153909921646118,
|
| 26887 |
+
"learning_rate": 1.2060872572755571e-05,
|
| 26888 |
+
"loss": 5.4692,
|
| 26889 |
+
"step": 15360
|
| 26890 |
+
},
|
| 26891 |
+
{
|
| 26892 |
+
"epoch": 0.76,
|
| 26893 |
+
"grad_norm": 2.0445823669433594,
|
| 26894 |
+
"learning_rate": 1.2050990661593953e-05,
|
| 26895 |
+
"loss": 5.3999,
|
| 26896 |
+
"step": 15364
|
| 26897 |
+
},
|
| 26898 |
+
{
|
| 26899 |
+
"epoch": 0.76,
|
| 26900 |
+
"grad_norm": 2.0181336402893066,
|
| 26901 |
+
"learning_rate": 1.2041108750432334e-05,
|
| 26902 |
+
"loss": 5.4045,
|
| 26903 |
+
"step": 15368
|
| 26904 |
+
},
|
| 26905 |
+
{
|
| 26906 |
+
"epoch": 0.76,
|
| 26907 |
+
"grad_norm": 2.008654832839966,
|
| 26908 |
+
"learning_rate": 1.2031226839270716e-05,
|
| 26909 |
+
"loss": 5.6017,
|
| 26910 |
+
"step": 15372
|
| 26911 |
+
},
|
| 26912 |
+
{
|
| 26913 |
+
"epoch": 0.76,
|
| 26914 |
+
"grad_norm": 2.126469850540161,
|
| 26915 |
+
"learning_rate": 1.2021344928109096e-05,
|
| 26916 |
+
"loss": 5.4578,
|
| 26917 |
+
"step": 15376
|
| 26918 |
+
},
|
| 26919 |
+
{
|
| 26920 |
+
"epoch": 0.76,
|
| 26921 |
+
"grad_norm": 2.031398057937622,
|
| 26922 |
+
"learning_rate": 1.2011463016947478e-05,
|
| 26923 |
+
"loss": 5.5091,
|
| 26924 |
+
"step": 15380
|
| 26925 |
+
},
|
| 26926 |
+
{
|
| 26927 |
+
"epoch": 0.76,
|
| 26928 |
+
"grad_norm": 2.102151870727539,
|
| 26929 |
+
"learning_rate": 1.200158110578586e-05,
|
| 26930 |
+
"loss": 5.4637,
|
| 26931 |
+
"step": 15384
|
| 26932 |
+
},
|
| 26933 |
+
{
|
| 26934 |
+
"epoch": 0.76,
|
| 26935 |
+
"grad_norm": 1.9239962100982666,
|
| 26936 |
+
"learning_rate": 1.199169919462424e-05,
|
| 26937 |
+
"loss": 5.5014,
|
| 26938 |
+
"step": 15388
|
| 26939 |
+
},
|
| 26940 |
+
{
|
| 26941 |
+
"epoch": 0.76,
|
| 26942 |
+
"grad_norm": 1.9613525867462158,
|
| 26943 |
+
"learning_rate": 1.1981817283462623e-05,
|
| 26944 |
+
"loss": 5.5283,
|
| 26945 |
+
"step": 15392
|
| 26946 |
+
},
|
| 26947 |
+
{
|
| 26948 |
+
"epoch": 0.76,
|
| 26949 |
+
"grad_norm": 1.973003625869751,
|
| 26950 |
+
"learning_rate": 1.1971935372301003e-05,
|
| 26951 |
+
"loss": 5.4865,
|
| 26952 |
+
"step": 15396
|
| 26953 |
+
},
|
| 26954 |
+
{
|
| 26955 |
+
"epoch": 0.76,
|
| 26956 |
+
"grad_norm": 2.3486452102661133,
|
| 26957 |
+
"learning_rate": 1.1962053461139385e-05,
|
| 26958 |
+
"loss": 5.5125,
|
| 26959 |
+
"step": 15400
|
| 26960 |
+
},
|
| 26961 |
+
{
|
| 26962 |
+
"epoch": 0.76,
|
| 26963 |
+
"grad_norm": 2.088740825653076,
|
| 26964 |
+
"learning_rate": 1.1952171549977765e-05,
|
| 26965 |
+
"loss": 5.4416,
|
| 26966 |
+
"step": 15404
|
| 26967 |
+
},
|
| 26968 |
+
{
|
| 26969 |
+
"epoch": 0.76,
|
| 26970 |
+
"grad_norm": 1.9701464176177979,
|
| 26971 |
+
"learning_rate": 1.1942289638816147e-05,
|
| 26972 |
+
"loss": 5.2871,
|
| 26973 |
+
"step": 15408
|
| 26974 |
+
},
|
| 26975 |
+
{
|
| 26976 |
+
"epoch": 0.76,
|
| 26977 |
+
"grad_norm": 2.2388153076171875,
|
| 26978 |
+
"learning_rate": 1.193240772765453e-05,
|
| 26979 |
+
"loss": 5.551,
|
| 26980 |
+
"step": 15412
|
| 26981 |
+
},
|
| 26982 |
+
{
|
| 26983 |
+
"epoch": 0.76,
|
| 26984 |
+
"grad_norm": 2.2453413009643555,
|
| 26985 |
+
"learning_rate": 1.192252581649291e-05,
|
| 26986 |
+
"loss": 5.4372,
|
| 26987 |
+
"step": 15416
|
| 26988 |
+
},
|
| 26989 |
+
{
|
| 26990 |
+
"epoch": 0.76,
|
| 26991 |
+
"grad_norm": 2.1105456352233887,
|
| 26992 |
+
"learning_rate": 1.1912643905331292e-05,
|
| 26993 |
+
"loss": 5.5284,
|
| 26994 |
+
"step": 15420
|
| 26995 |
+
},
|
| 26996 |
+
{
|
| 26997 |
+
"epoch": 0.76,
|
| 26998 |
+
"grad_norm": 2.197547197341919,
|
| 26999 |
+
"learning_rate": 1.1902761994169672e-05,
|
| 27000 |
+
"loss": 5.5067,
|
| 27001 |
+
"step": 15424
|
| 27002 |
+
},
|
| 27003 |
+
{
|
| 27004 |
+
"epoch": 0.76,
|
| 27005 |
+
"grad_norm": 1.9585908651351929,
|
| 27006 |
+
"learning_rate": 1.1892880083008054e-05,
|
| 27007 |
+
"loss": 5.4292,
|
| 27008 |
+
"step": 15428
|
| 27009 |
+
},
|
| 27010 |
+
{
|
| 27011 |
+
"epoch": 0.76,
|
| 27012 |
+
"grad_norm": 2.037917137145996,
|
| 27013 |
+
"learning_rate": 1.1882998171846435e-05,
|
| 27014 |
+
"loss": 5.455,
|
| 27015 |
+
"step": 15432
|
| 27016 |
+
},
|
| 27017 |
+
{
|
| 27018 |
+
"epoch": 0.76,
|
| 27019 |
+
"grad_norm": 1.797452449798584,
|
| 27020 |
+
"learning_rate": 1.1873116260684817e-05,
|
| 27021 |
+
"loss": 5.4833,
|
| 27022 |
+
"step": 15436
|
| 27023 |
+
},
|
| 27024 |
+
{
|
| 27025 |
+
"epoch": 0.76,
|
| 27026 |
+
"grad_norm": 1.8614048957824707,
|
| 27027 |
+
"learning_rate": 1.1863234349523199e-05,
|
| 27028 |
+
"loss": 5.5127,
|
| 27029 |
+
"step": 15440
|
| 27030 |
+
},
|
| 27031 |
+
{
|
| 27032 |
+
"epoch": 0.76,
|
| 27033 |
+
"grad_norm": 1.9821441173553467,
|
| 27034 |
+
"learning_rate": 1.185335243836158e-05,
|
| 27035 |
+
"loss": 5.5038,
|
| 27036 |
+
"step": 15444
|
| 27037 |
+
},
|
| 27038 |
+
{
|
| 27039 |
+
"epoch": 0.76,
|
| 27040 |
+
"grad_norm": 2.3709757328033447,
|
| 27041 |
+
"learning_rate": 1.1843470527199961e-05,
|
| 27042 |
+
"loss": 5.4745,
|
| 27043 |
+
"step": 15448
|
| 27044 |
+
},
|
| 27045 |
+
{
|
| 27046 |
+
"epoch": 0.76,
|
| 27047 |
+
"grad_norm": 1.8125073909759521,
|
| 27048 |
+
"learning_rate": 1.1833588616038342e-05,
|
| 27049 |
+
"loss": 5.4642,
|
| 27050 |
+
"step": 15452
|
| 27051 |
+
},
|
| 27052 |
+
{
|
| 27053 |
+
"epoch": 0.76,
|
| 27054 |
+
"grad_norm": 2.0087850093841553,
|
| 27055 |
+
"learning_rate": 1.1823706704876724e-05,
|
| 27056 |
+
"loss": 5.4628,
|
| 27057 |
+
"step": 15456
|
| 27058 |
+
},
|
| 27059 |
+
{
|
| 27060 |
+
"epoch": 0.76,
|
| 27061 |
+
"grad_norm": 1.9169509410858154,
|
| 27062 |
+
"learning_rate": 1.1813824793715104e-05,
|
| 27063 |
+
"loss": 5.4821,
|
| 27064 |
+
"step": 15460
|
| 27065 |
+
},
|
| 27066 |
+
{
|
| 27067 |
+
"epoch": 0.76,
|
| 27068 |
+
"grad_norm": 2.01556396484375,
|
| 27069 |
+
"learning_rate": 1.1803942882553488e-05,
|
| 27070 |
+
"loss": 5.4142,
|
| 27071 |
+
"step": 15464
|
| 27072 |
+
},
|
| 27073 |
+
{
|
| 27074 |
+
"epoch": 0.76,
|
| 27075 |
+
"grad_norm": 1.9918155670166016,
|
| 27076 |
+
"learning_rate": 1.1794060971391868e-05,
|
| 27077 |
+
"loss": 5.6412,
|
| 27078 |
+
"step": 15468
|
| 27079 |
+
},
|
| 27080 |
+
{
|
| 27081 |
+
"epoch": 0.76,
|
| 27082 |
+
"grad_norm": 2.052454710006714,
|
| 27083 |
+
"learning_rate": 1.1784179060230249e-05,
|
| 27084 |
+
"loss": 5.4347,
|
| 27085 |
+
"step": 15472
|
| 27086 |
+
},
|
| 27087 |
+
{
|
| 27088 |
+
"epoch": 0.76,
|
| 27089 |
+
"grad_norm": 1.8537468910217285,
|
| 27090 |
+
"learning_rate": 1.177429714906863e-05,
|
| 27091 |
+
"loss": 5.5036,
|
| 27092 |
+
"step": 15476
|
| 27093 |
+
},
|
| 27094 |
+
{
|
| 27095 |
+
"epoch": 0.76,
|
| 27096 |
+
"grad_norm": 1.9864877462387085,
|
| 27097 |
+
"learning_rate": 1.1764415237907011e-05,
|
| 27098 |
+
"loss": 5.4979,
|
| 27099 |
+
"step": 15480
|
| 27100 |
+
},
|
| 27101 |
+
{
|
| 27102 |
+
"epoch": 0.77,
|
| 27103 |
+
"grad_norm": 2.200070858001709,
|
| 27104 |
+
"learning_rate": 1.1754533326745393e-05,
|
| 27105 |
+
"loss": 5.5684,
|
| 27106 |
+
"step": 15484
|
| 27107 |
+
},
|
| 27108 |
+
{
|
| 27109 |
+
"epoch": 0.77,
|
| 27110 |
+
"grad_norm": 1.7679542303085327,
|
| 27111 |
+
"learning_rate": 1.1744651415583773e-05,
|
| 27112 |
+
"loss": 5.4283,
|
| 27113 |
+
"step": 15488
|
| 27114 |
+
},
|
| 27115 |
+
{
|
| 27116 |
+
"epoch": 0.77,
|
| 27117 |
+
"grad_norm": 1.8458738327026367,
|
| 27118 |
+
"learning_rate": 1.1734769504422157e-05,
|
| 27119 |
+
"loss": 5.5566,
|
| 27120 |
+
"step": 15492
|
| 27121 |
+
},
|
| 27122 |
+
{
|
| 27123 |
+
"epoch": 0.77,
|
| 27124 |
+
"grad_norm": 2.3362741470336914,
|
| 27125 |
+
"learning_rate": 1.1724887593260538e-05,
|
| 27126 |
+
"loss": 5.4377,
|
| 27127 |
+
"step": 15496
|
| 27128 |
+
},
|
| 27129 |
+
{
|
| 27130 |
+
"epoch": 0.77,
|
| 27131 |
+
"grad_norm": 1.7595479488372803,
|
| 27132 |
+
"learning_rate": 1.1715005682098918e-05,
|
| 27133 |
+
"loss": 5.4785,
|
| 27134 |
+
"step": 15500
|
| 27135 |
+
},
|
| 27136 |
+
{
|
| 27137 |
+
"epoch": 0.77,
|
| 27138 |
+
"grad_norm": 2.1757829189300537,
|
| 27139 |
+
"learning_rate": 1.17051237709373e-05,
|
| 27140 |
+
"loss": 5.5964,
|
| 27141 |
+
"step": 15504
|
| 27142 |
+
},
|
| 27143 |
+
{
|
| 27144 |
+
"epoch": 0.77,
|
| 27145 |
+
"grad_norm": 2.0845088958740234,
|
| 27146 |
+
"learning_rate": 1.169524185977568e-05,
|
| 27147 |
+
"loss": 5.5294,
|
| 27148 |
+
"step": 15508
|
| 27149 |
+
},
|
| 27150 |
+
{
|
| 27151 |
+
"epoch": 0.77,
|
| 27152 |
+
"grad_norm": 1.937070608139038,
|
| 27153 |
+
"learning_rate": 1.1685359948614062e-05,
|
| 27154 |
+
"loss": 5.5338,
|
| 27155 |
+
"step": 15512
|
| 27156 |
+
},
|
| 27157 |
+
{
|
| 27158 |
+
"epoch": 0.77,
|
| 27159 |
+
"grad_norm": 2.137470006942749,
|
| 27160 |
+
"learning_rate": 1.1675478037452443e-05,
|
| 27161 |
+
"loss": 5.4788,
|
| 27162 |
+
"step": 15516
|
| 27163 |
+
},
|
| 27164 |
+
{
|
| 27165 |
+
"epoch": 0.77,
|
| 27166 |
+
"grad_norm": 2.1320180892944336,
|
| 27167 |
+
"learning_rate": 1.1665596126290827e-05,
|
| 27168 |
+
"loss": 5.6173,
|
| 27169 |
+
"step": 15520
|
| 27170 |
+
},
|
| 27171 |
+
{
|
| 27172 |
+
"epoch": 0.77,
|
| 27173 |
+
"grad_norm": 2.0288352966308594,
|
| 27174 |
+
"learning_rate": 1.1655714215129207e-05,
|
| 27175 |
+
"loss": 5.3012,
|
| 27176 |
+
"step": 15524
|
| 27177 |
+
},
|
| 27178 |
+
{
|
| 27179 |
+
"epoch": 0.77,
|
| 27180 |
+
"grad_norm": 1.784881830215454,
|
| 27181 |
+
"learning_rate": 1.1645832303967589e-05,
|
| 27182 |
+
"loss": 5.5805,
|
| 27183 |
+
"step": 15528
|
| 27184 |
+
},
|
| 27185 |
+
{
|
| 27186 |
+
"epoch": 0.77,
|
| 27187 |
+
"grad_norm": 2.08077073097229,
|
| 27188 |
+
"learning_rate": 1.163595039280597e-05,
|
| 27189 |
+
"loss": 5.4992,
|
| 27190 |
+
"step": 15532
|
| 27191 |
+
},
|
| 27192 |
+
{
|
| 27193 |
+
"epoch": 0.77,
|
| 27194 |
+
"grad_norm": 1.7977101802825928,
|
| 27195 |
+
"learning_rate": 1.162606848164435e-05,
|
| 27196 |
+
"loss": 5.4872,
|
| 27197 |
+
"step": 15536
|
| 27198 |
+
},
|
| 27199 |
+
{
|
| 27200 |
+
"epoch": 0.77,
|
| 27201 |
+
"grad_norm": 2.186459541320801,
|
| 27202 |
+
"learning_rate": 1.1616186570482732e-05,
|
| 27203 |
+
"loss": 5.578,
|
| 27204 |
+
"step": 15540
|
| 27205 |
+
},
|
| 27206 |
+
{
|
| 27207 |
+
"epoch": 0.77,
|
| 27208 |
+
"grad_norm": 2.183048725128174,
|
| 27209 |
+
"learning_rate": 1.1606304659321112e-05,
|
| 27210 |
+
"loss": 5.4738,
|
| 27211 |
+
"step": 15544
|
| 27212 |
+
},
|
| 27213 |
+
{
|
| 27214 |
+
"epoch": 0.77,
|
| 27215 |
+
"grad_norm": 1.8500030040740967,
|
| 27216 |
+
"learning_rate": 1.1596422748159494e-05,
|
| 27217 |
+
"loss": 5.5345,
|
| 27218 |
+
"step": 15548
|
| 27219 |
+
},
|
| 27220 |
+
{
|
| 27221 |
+
"epoch": 0.77,
|
| 27222 |
+
"grad_norm": 1.8691611289978027,
|
| 27223 |
+
"learning_rate": 1.1586540836997876e-05,
|
| 27224 |
+
"loss": 5.5283,
|
| 27225 |
+
"step": 15552
|
| 27226 |
+
},
|
| 27227 |
+
{
|
| 27228 |
+
"epoch": 0.77,
|
| 27229 |
+
"grad_norm": 2.1216142177581787,
|
| 27230 |
+
"learning_rate": 1.1576658925836258e-05,
|
| 27231 |
+
"loss": 5.581,
|
| 27232 |
+
"step": 15556
|
| 27233 |
+
},
|
| 27234 |
+
{
|
| 27235 |
+
"epoch": 0.77,
|
| 27236 |
+
"grad_norm": 2.0468342304229736,
|
| 27237 |
+
"learning_rate": 1.1566777014674639e-05,
|
| 27238 |
+
"loss": 5.5649,
|
| 27239 |
+
"step": 15560
|
| 27240 |
+
},
|
| 27241 |
+
{
|
| 27242 |
+
"epoch": 0.77,
|
| 27243 |
+
"grad_norm": 2.060667037963867,
|
| 27244 |
+
"learning_rate": 1.155689510351302e-05,
|
| 27245 |
+
"loss": 5.4901,
|
| 27246 |
+
"step": 15564
|
| 27247 |
+
},
|
| 27248 |
+
{
|
| 27249 |
+
"epoch": 0.77,
|
| 27250 |
+
"grad_norm": 1.803126573562622,
|
| 27251 |
+
"learning_rate": 1.1547013192351401e-05,
|
| 27252 |
+
"loss": 5.4285,
|
| 27253 |
+
"step": 15568
|
| 27254 |
+
},
|
| 27255 |
+
{
|
| 27256 |
+
"epoch": 0.77,
|
| 27257 |
+
"grad_norm": 2.05906081199646,
|
| 27258 |
+
"learning_rate": 1.1537131281189781e-05,
|
| 27259 |
+
"loss": 5.4268,
|
| 27260 |
+
"step": 15572
|
| 27261 |
+
},
|
| 27262 |
+
{
|
| 27263 |
+
"epoch": 0.77,
|
| 27264 |
+
"grad_norm": 2.0089757442474365,
|
| 27265 |
+
"learning_rate": 1.1527249370028164e-05,
|
| 27266 |
+
"loss": 5.5307,
|
| 27267 |
+
"step": 15576
|
| 27268 |
+
},
|
| 27269 |
+
{
|
| 27270 |
+
"epoch": 0.77,
|
| 27271 |
+
"grad_norm": 1.9152356386184692,
|
| 27272 |
+
"learning_rate": 1.1517367458866546e-05,
|
| 27273 |
+
"loss": 5.4297,
|
| 27274 |
+
"step": 15580
|
| 27275 |
+
},
|
| 27276 |
+
{
|
| 27277 |
+
"epoch": 0.77,
|
| 27278 |
+
"grad_norm": 1.9012370109558105,
|
| 27279 |
+
"learning_rate": 1.1507485547704928e-05,
|
| 27280 |
+
"loss": 5.5271,
|
| 27281 |
+
"step": 15584
|
| 27282 |
+
},
|
| 27283 |
+
{
|
| 27284 |
+
"epoch": 0.77,
|
| 27285 |
+
"grad_norm": 2.0786564350128174,
|
| 27286 |
+
"learning_rate": 1.1497603636543308e-05,
|
| 27287 |
+
"loss": 5.5957,
|
| 27288 |
+
"step": 15588
|
| 27289 |
+
},
|
| 27290 |
+
{
|
| 27291 |
+
"epoch": 0.77,
|
| 27292 |
+
"grad_norm": 2.03715443611145,
|
| 27293 |
+
"learning_rate": 1.148772172538169e-05,
|
| 27294 |
+
"loss": 5.5501,
|
| 27295 |
+
"step": 15592
|
| 27296 |
+
},
|
| 27297 |
+
{
|
| 27298 |
+
"epoch": 0.77,
|
| 27299 |
+
"grad_norm": 2.1424834728240967,
|
| 27300 |
+
"learning_rate": 1.147783981422007e-05,
|
| 27301 |
+
"loss": 5.5153,
|
| 27302 |
+
"step": 15596
|
| 27303 |
+
},
|
| 27304 |
+
{
|
| 27305 |
+
"epoch": 0.77,
|
| 27306 |
+
"grad_norm": 1.9324986934661865,
|
| 27307 |
+
"learning_rate": 1.146795790305845e-05,
|
| 27308 |
+
"loss": 5.5928,
|
| 27309 |
+
"step": 15600
|
| 27310 |
+
},
|
| 27311 |
+
{
|
| 27312 |
+
"epoch": 0.77,
|
| 27313 |
+
"grad_norm": 1.986244559288025,
|
| 27314 |
+
"learning_rate": 1.1458075991896833e-05,
|
| 27315 |
+
"loss": 5.4466,
|
| 27316 |
+
"step": 15604
|
| 27317 |
+
},
|
| 27318 |
+
{
|
| 27319 |
+
"epoch": 0.77,
|
| 27320 |
+
"grad_norm": 2.3215315341949463,
|
| 27321 |
+
"learning_rate": 1.1448194080735215e-05,
|
| 27322 |
+
"loss": 5.4948,
|
| 27323 |
+
"step": 15608
|
| 27324 |
+
},
|
| 27325 |
+
{
|
| 27326 |
+
"epoch": 0.77,
|
| 27327 |
+
"grad_norm": 2.2035694122314453,
|
| 27328 |
+
"learning_rate": 1.1438312169573597e-05,
|
| 27329 |
+
"loss": 5.6002,
|
| 27330 |
+
"step": 15612
|
| 27331 |
+
},
|
| 27332 |
+
{
|
| 27333 |
+
"epoch": 0.77,
|
| 27334 |
+
"grad_norm": 2.0092086791992188,
|
| 27335 |
+
"learning_rate": 1.1428430258411977e-05,
|
| 27336 |
+
"loss": 5.4083,
|
| 27337 |
+
"step": 15616
|
| 27338 |
+
},
|
| 27339 |
+
{
|
| 27340 |
+
"epoch": 0.77,
|
| 27341 |
+
"grad_norm": 2.008274555206299,
|
| 27342 |
+
"learning_rate": 1.141854834725036e-05,
|
| 27343 |
+
"loss": 5.3977,
|
| 27344 |
+
"step": 15620
|
| 27345 |
+
},
|
| 27346 |
+
{
|
| 27347 |
+
"epoch": 0.77,
|
| 27348 |
+
"grad_norm": 2.084843873977661,
|
| 27349 |
+
"learning_rate": 1.140866643608874e-05,
|
| 27350 |
+
"loss": 5.4896,
|
| 27351 |
+
"step": 15624
|
| 27352 |
+
},
|
| 27353 |
+
{
|
| 27354 |
+
"epoch": 0.77,
|
| 27355 |
+
"grad_norm": 2.138129949569702,
|
| 27356 |
+
"learning_rate": 1.1398784524927122e-05,
|
| 27357 |
+
"loss": 5.556,
|
| 27358 |
+
"step": 15628
|
| 27359 |
+
},
|
| 27360 |
+
{
|
| 27361 |
+
"epoch": 0.77,
|
| 27362 |
+
"grad_norm": 2.161590814590454,
|
| 27363 |
+
"learning_rate": 1.1388902613765502e-05,
|
| 27364 |
+
"loss": 5.4909,
|
| 27365 |
+
"step": 15632
|
| 27366 |
+
},
|
| 27367 |
+
{
|
| 27368 |
+
"epoch": 0.77,
|
| 27369 |
+
"grad_norm": 2.1914632320404053,
|
| 27370 |
+
"learning_rate": 1.1379020702603884e-05,
|
| 27371 |
+
"loss": 5.5578,
|
| 27372 |
+
"step": 15636
|
| 27373 |
+
},
|
| 27374 |
+
{
|
| 27375 |
+
"epoch": 0.77,
|
| 27376 |
+
"grad_norm": 2.254403829574585,
|
| 27377 |
+
"learning_rate": 1.1369138791442266e-05,
|
| 27378 |
+
"loss": 5.6521,
|
| 27379 |
+
"step": 15640
|
| 27380 |
+
},
|
| 27381 |
+
{
|
| 27382 |
+
"epoch": 0.77,
|
| 27383 |
+
"grad_norm": 1.9875483512878418,
|
| 27384 |
+
"learning_rate": 1.1359256880280647e-05,
|
| 27385 |
+
"loss": 5.4972,
|
| 27386 |
+
"step": 15644
|
| 27387 |
+
},
|
| 27388 |
+
{
|
| 27389 |
+
"epoch": 0.77,
|
| 27390 |
+
"grad_norm": 2.2201669216156006,
|
| 27391 |
+
"learning_rate": 1.1349374969119029e-05,
|
| 27392 |
+
"loss": 5.6261,
|
| 27393 |
+
"step": 15648
|
| 27394 |
+
},
|
| 27395 |
+
{
|
| 27396 |
+
"epoch": 0.77,
|
| 27397 |
+
"grad_norm": 2.054435968399048,
|
| 27398 |
+
"learning_rate": 1.1339493057957409e-05,
|
| 27399 |
+
"loss": 5.5603,
|
| 27400 |
+
"step": 15652
|
| 27401 |
+
},
|
| 27402 |
+
{
|
| 27403 |
+
"epoch": 0.77,
|
| 27404 |
+
"grad_norm": 1.902565598487854,
|
| 27405 |
+
"learning_rate": 1.1329611146795791e-05,
|
| 27406 |
+
"loss": 5.6618,
|
| 27407 |
+
"step": 15656
|
| 27408 |
+
},
|
| 27409 |
+
{
|
| 27410 |
+
"epoch": 0.77,
|
| 27411 |
+
"grad_norm": 2.2018725872039795,
|
| 27412 |
+
"learning_rate": 1.1319729235634172e-05,
|
| 27413 |
+
"loss": 5.429,
|
| 27414 |
+
"step": 15660
|
| 27415 |
+
},
|
| 27416 |
+
{
|
| 27417 |
+
"epoch": 0.77,
|
| 27418 |
+
"grad_norm": 1.9127063751220703,
|
| 27419 |
+
"learning_rate": 1.1309847324472554e-05,
|
| 27420 |
+
"loss": 5.5171,
|
| 27421 |
+
"step": 15664
|
| 27422 |
+
},
|
| 27423 |
+
{
|
| 27424 |
+
"epoch": 0.77,
|
| 27425 |
+
"grad_norm": 1.8942164182662964,
|
| 27426 |
+
"learning_rate": 1.1299965413310936e-05,
|
| 27427 |
+
"loss": 5.5357,
|
| 27428 |
+
"step": 15668
|
| 27429 |
+
},
|
| 27430 |
+
{
|
| 27431 |
+
"epoch": 0.77,
|
| 27432 |
+
"grad_norm": 2.2651684284210205,
|
| 27433 |
+
"learning_rate": 1.1290083502149316e-05,
|
| 27434 |
+
"loss": 5.4591,
|
| 27435 |
+
"step": 15672
|
| 27436 |
+
},
|
| 27437 |
+
{
|
| 27438 |
+
"epoch": 0.77,
|
| 27439 |
+
"grad_norm": 2.189774990081787,
|
| 27440 |
+
"learning_rate": 1.1280201590987698e-05,
|
| 27441 |
+
"loss": 5.4338,
|
| 27442 |
+
"step": 15676
|
| 27443 |
+
},
|
| 27444 |
+
{
|
| 27445 |
+
"epoch": 0.77,
|
| 27446 |
+
"grad_norm": 2.1134681701660156,
|
| 27447 |
+
"learning_rate": 1.1270319679826078e-05,
|
| 27448 |
+
"loss": 5.5294,
|
| 27449 |
+
"step": 15680
|
| 27450 |
+
},
|
| 27451 |
+
{
|
| 27452 |
+
"epoch": 0.77,
|
| 27453 |
+
"grad_norm": 2.2125091552734375,
|
| 27454 |
+
"learning_rate": 1.126043776866446e-05,
|
| 27455 |
+
"loss": 5.5269,
|
| 27456 |
+
"step": 15684
|
| 27457 |
+
},
|
| 27458 |
+
{
|
| 27459 |
+
"epoch": 0.78,
|
| 27460 |
+
"grad_norm": 1.9919397830963135,
|
| 27461 |
+
"learning_rate": 1.1250555857502841e-05,
|
| 27462 |
+
"loss": 5.4072,
|
| 27463 |
+
"step": 15688
|
| 27464 |
+
},
|
| 27465 |
+
{
|
| 27466 |
+
"epoch": 0.78,
|
| 27467 |
+
"grad_norm": 1.958975911140442,
|
| 27468 |
+
"learning_rate": 1.1240673946341223e-05,
|
| 27469 |
+
"loss": 5.4421,
|
| 27470 |
+
"step": 15692
|
| 27471 |
+
},
|
| 27472 |
+
{
|
| 27473 |
+
"epoch": 0.78,
|
| 27474 |
+
"grad_norm": 2.033118724822998,
|
| 27475 |
+
"learning_rate": 1.1230792035179605e-05,
|
| 27476 |
+
"loss": 5.5095,
|
| 27477 |
+
"step": 15696
|
| 27478 |
+
},
|
| 27479 |
+
{
|
| 27480 |
+
"epoch": 0.78,
|
| 27481 |
+
"grad_norm": 2.1201207637786865,
|
| 27482 |
+
"learning_rate": 1.1220910124017985e-05,
|
| 27483 |
+
"loss": 5.4767,
|
| 27484 |
+
"step": 15700
|
| 27485 |
+
},
|
| 27486 |
+
{
|
| 27487 |
+
"epoch": 0.78,
|
| 27488 |
+
"grad_norm": 1.9773985147476196,
|
| 27489 |
+
"learning_rate": 1.1211028212856367e-05,
|
| 27490 |
+
"loss": 5.424,
|
| 27491 |
+
"step": 15704
|
| 27492 |
+
},
|
| 27493 |
+
{
|
| 27494 |
+
"epoch": 0.78,
|
| 27495 |
+
"grad_norm": 1.841871976852417,
|
| 27496 |
+
"learning_rate": 1.1201146301694748e-05,
|
| 27497 |
+
"loss": 5.521,
|
| 27498 |
+
"step": 15708
|
| 27499 |
+
},
|
| 27500 |
+
{
|
| 27501 |
+
"epoch": 0.78,
|
| 27502 |
+
"grad_norm": 1.9351176023483276,
|
| 27503 |
+
"learning_rate": 1.119126439053313e-05,
|
| 27504 |
+
"loss": 5.5646,
|
| 27505 |
+
"step": 15712
|
| 27506 |
+
},
|
| 27507 |
+
{
|
| 27508 |
+
"epoch": 0.78,
|
| 27509 |
+
"grad_norm": 2.0061535835266113,
|
| 27510 |
+
"learning_rate": 1.118138247937151e-05,
|
| 27511 |
+
"loss": 5.36,
|
| 27512 |
+
"step": 15716
|
| 27513 |
+
},
|
| 27514 |
+
{
|
| 27515 |
+
"epoch": 0.78,
|
| 27516 |
+
"grad_norm": 2.4020817279815674,
|
| 27517 |
+
"learning_rate": 1.1171500568209892e-05,
|
| 27518 |
+
"loss": 5.4701,
|
| 27519 |
+
"step": 15720
|
| 27520 |
+
},
|
| 27521 |
+
{
|
| 27522 |
+
"epoch": 0.78,
|
| 27523 |
+
"grad_norm": 1.9213863611221313,
|
| 27524 |
+
"learning_rate": 1.1161618657048274e-05,
|
| 27525 |
+
"loss": 5.4668,
|
| 27526 |
+
"step": 15724
|
| 27527 |
+
},
|
| 27528 |
+
{
|
| 27529 |
+
"epoch": 0.78,
|
| 27530 |
+
"grad_norm": 2.3538432121276855,
|
| 27531 |
+
"learning_rate": 1.1151736745886655e-05,
|
| 27532 |
+
"loss": 5.5128,
|
| 27533 |
+
"step": 15728
|
| 27534 |
+
},
|
| 27535 |
+
{
|
| 27536 |
+
"epoch": 0.78,
|
| 27537 |
+
"grad_norm": 2.147163152694702,
|
| 27538 |
+
"learning_rate": 1.1141854834725037e-05,
|
| 27539 |
+
"loss": 5.4579,
|
| 27540 |
+
"step": 15732
|
| 27541 |
+
},
|
| 27542 |
+
{
|
| 27543 |
+
"epoch": 0.78,
|
| 27544 |
+
"grad_norm": 2.1335911750793457,
|
| 27545 |
+
"learning_rate": 1.1131972923563417e-05,
|
| 27546 |
+
"loss": 5.4277,
|
| 27547 |
+
"step": 15736
|
| 27548 |
+
},
|
| 27549 |
+
{
|
| 27550 |
+
"epoch": 0.78,
|
| 27551 |
+
"grad_norm": 2.1882131099700928,
|
| 27552 |
+
"learning_rate": 1.11220910124018e-05,
|
| 27553 |
+
"loss": 5.6003,
|
| 27554 |
+
"step": 15740
|
| 27555 |
+
},
|
| 27556 |
+
{
|
| 27557 |
+
"epoch": 0.78,
|
| 27558 |
+
"grad_norm": 1.9644808769226074,
|
| 27559 |
+
"learning_rate": 1.111220910124018e-05,
|
| 27560 |
+
"loss": 5.5355,
|
| 27561 |
+
"step": 15744
|
| 27562 |
+
},
|
| 27563 |
+
{
|
| 27564 |
+
"epoch": 0.78,
|
| 27565 |
+
"grad_norm": 1.7521815299987793,
|
| 27566 |
+
"learning_rate": 1.1102327190078562e-05,
|
| 27567 |
+
"loss": 5.4743,
|
| 27568 |
+
"step": 15748
|
| 27569 |
+
},
|
| 27570 |
+
{
|
| 27571 |
+
"epoch": 0.78,
|
| 27572 |
+
"grad_norm": 2.1638858318328857,
|
| 27573 |
+
"learning_rate": 1.1092445278916944e-05,
|
| 27574 |
+
"loss": 5.3701,
|
| 27575 |
+
"step": 15752
|
| 27576 |
+
},
|
| 27577 |
+
{
|
| 27578 |
+
"epoch": 0.78,
|
| 27579 |
+
"grad_norm": 1.8966166973114014,
|
| 27580 |
+
"learning_rate": 1.1082563367755324e-05,
|
| 27581 |
+
"loss": 5.5355,
|
| 27582 |
+
"step": 15756
|
| 27583 |
+
},
|
| 27584 |
+
{
|
| 27585 |
+
"epoch": 0.78,
|
| 27586 |
+
"grad_norm": 1.981217861175537,
|
| 27587 |
+
"learning_rate": 1.1072681456593706e-05,
|
| 27588 |
+
"loss": 5.4366,
|
| 27589 |
+
"step": 15760
|
| 27590 |
+
},
|
| 27591 |
+
{
|
| 27592 |
+
"epoch": 0.78,
|
| 27593 |
+
"grad_norm": 2.0059125423431396,
|
| 27594 |
+
"learning_rate": 1.1062799545432086e-05,
|
| 27595 |
+
"loss": 5.4749,
|
| 27596 |
+
"step": 15764
|
| 27597 |
+
},
|
| 27598 |
+
{
|
| 27599 |
+
"epoch": 0.78,
|
| 27600 |
+
"grad_norm": 2.042475461959839,
|
| 27601 |
+
"learning_rate": 1.1052917634270469e-05,
|
| 27602 |
+
"loss": 5.4066,
|
| 27603 |
+
"step": 15768
|
| 27604 |
+
},
|
| 27605 |
+
{
|
| 27606 |
+
"epoch": 0.78,
|
| 27607 |
+
"grad_norm": 2.1431596279144287,
|
| 27608 |
+
"learning_rate": 1.1043035723108849e-05,
|
| 27609 |
+
"loss": 5.4831,
|
| 27610 |
+
"step": 15772
|
| 27611 |
+
},
|
| 27612 |
+
{
|
| 27613 |
+
"epoch": 0.78,
|
| 27614 |
+
"grad_norm": 1.903968095779419,
|
| 27615 |
+
"learning_rate": 1.1033153811947231e-05,
|
| 27616 |
+
"loss": 5.4749,
|
| 27617 |
+
"step": 15776
|
| 27618 |
+
},
|
| 27619 |
+
{
|
| 27620 |
+
"epoch": 0.78,
|
| 27621 |
+
"grad_norm": 2.2087221145629883,
|
| 27622 |
+
"learning_rate": 1.1023271900785613e-05,
|
| 27623 |
+
"loss": 5.5542,
|
| 27624 |
+
"step": 15780
|
| 27625 |
+
},
|
| 27626 |
+
{
|
| 27627 |
+
"epoch": 0.78,
|
| 27628 |
+
"grad_norm": 1.990248680114746,
|
| 27629 |
+
"learning_rate": 1.1013389989623993e-05,
|
| 27630 |
+
"loss": 5.4592,
|
| 27631 |
+
"step": 15784
|
| 27632 |
+
},
|
| 27633 |
+
{
|
| 27634 |
+
"epoch": 0.78,
|
| 27635 |
+
"grad_norm": 1.9822028875350952,
|
| 27636 |
+
"learning_rate": 1.1003508078462375e-05,
|
| 27637 |
+
"loss": 5.5245,
|
| 27638 |
+
"step": 15788
|
| 27639 |
+
},
|
| 27640 |
+
{
|
| 27641 |
+
"epoch": 0.78,
|
| 27642 |
+
"grad_norm": 2.0607571601867676,
|
| 27643 |
+
"learning_rate": 1.0993626167300756e-05,
|
| 27644 |
+
"loss": 5.5105,
|
| 27645 |
+
"step": 15792
|
| 27646 |
+
},
|
| 27647 |
+
{
|
| 27648 |
+
"epoch": 0.78,
|
| 27649 |
+
"grad_norm": 2.2510719299316406,
|
| 27650 |
+
"learning_rate": 1.0983744256139138e-05,
|
| 27651 |
+
"loss": 5.4485,
|
| 27652 |
+
"step": 15796
|
| 27653 |
+
},
|
| 27654 |
+
{
|
| 27655 |
+
"epoch": 0.78,
|
| 27656 |
+
"grad_norm": 1.9407929182052612,
|
| 27657 |
+
"learning_rate": 1.0973862344977518e-05,
|
| 27658 |
+
"loss": 5.4835,
|
| 27659 |
+
"step": 15800
|
| 27660 |
+
},
|
| 27661 |
+
{
|
| 27662 |
+
"epoch": 0.78,
|
| 27663 |
+
"grad_norm": 2.0228731632232666,
|
| 27664 |
+
"learning_rate": 1.09639804338159e-05,
|
| 27665 |
+
"loss": 5.4121,
|
| 27666 |
+
"step": 15804
|
| 27667 |
+
},
|
| 27668 |
+
{
|
| 27669 |
+
"epoch": 0.78,
|
| 27670 |
+
"grad_norm": 2.0110514163970947,
|
| 27671 |
+
"learning_rate": 1.0954098522654282e-05,
|
| 27672 |
+
"loss": 5.4097,
|
| 27673 |
+
"step": 15808
|
| 27674 |
+
},
|
| 27675 |
+
{
|
| 27676 |
+
"epoch": 0.78,
|
| 27677 |
+
"grad_norm": 2.005176305770874,
|
| 27678 |
+
"learning_rate": 1.0944216611492664e-05,
|
| 27679 |
+
"loss": 5.5651,
|
| 27680 |
+
"step": 15812
|
| 27681 |
+
},
|
| 27682 |
+
{
|
| 27683 |
+
"epoch": 0.78,
|
| 27684 |
+
"grad_norm": 2.08233380317688,
|
| 27685 |
+
"learning_rate": 1.0934334700331045e-05,
|
| 27686 |
+
"loss": 5.5942,
|
| 27687 |
+
"step": 15816
|
| 27688 |
+
},
|
| 27689 |
+
{
|
| 27690 |
+
"epoch": 0.78,
|
| 27691 |
+
"grad_norm": 1.76272714138031,
|
| 27692 |
+
"learning_rate": 1.0924452789169425e-05,
|
| 27693 |
+
"loss": 5.3883,
|
| 27694 |
+
"step": 15820
|
| 27695 |
+
},
|
| 27696 |
+
{
|
| 27697 |
+
"epoch": 0.78,
|
| 27698 |
+
"grad_norm": 1.8395804166793823,
|
| 27699 |
+
"learning_rate": 1.0914570878007807e-05,
|
| 27700 |
+
"loss": 5.372,
|
| 27701 |
+
"step": 15824
|
| 27702 |
+
},
|
| 27703 |
+
{
|
| 27704 |
+
"epoch": 0.78,
|
| 27705 |
+
"grad_norm": 2.197016954421997,
|
| 27706 |
+
"learning_rate": 1.0904688966846188e-05,
|
| 27707 |
+
"loss": 5.4239,
|
| 27708 |
+
"step": 15828
|
| 27709 |
+
},
|
| 27710 |
+
{
|
| 27711 |
+
"epoch": 0.78,
|
| 27712 |
+
"grad_norm": 2.1127779483795166,
|
| 27713 |
+
"learning_rate": 1.089480705568457e-05,
|
| 27714 |
+
"loss": 5.358,
|
| 27715 |
+
"step": 15832
|
| 27716 |
+
},
|
| 27717 |
+
{
|
| 27718 |
+
"epoch": 0.78,
|
| 27719 |
+
"grad_norm": 2.1906561851501465,
|
| 27720 |
+
"learning_rate": 1.0884925144522952e-05,
|
| 27721 |
+
"loss": 5.528,
|
| 27722 |
+
"step": 15836
|
| 27723 |
+
},
|
| 27724 |
+
{
|
| 27725 |
+
"epoch": 0.78,
|
| 27726 |
+
"grad_norm": 2.2963180541992188,
|
| 27727 |
+
"learning_rate": 1.0875043233361334e-05,
|
| 27728 |
+
"loss": 5.6399,
|
| 27729 |
+
"step": 15840
|
| 27730 |
+
},
|
| 27731 |
+
{
|
| 27732 |
+
"epoch": 0.78,
|
| 27733 |
+
"grad_norm": 2.1306509971618652,
|
| 27734 |
+
"learning_rate": 1.0865161322199714e-05,
|
| 27735 |
+
"loss": 5.4986,
|
| 27736 |
+
"step": 15844
|
| 27737 |
+
},
|
| 27738 |
+
{
|
| 27739 |
+
"epoch": 0.78,
|
| 27740 |
+
"grad_norm": 1.8235065937042236,
|
| 27741 |
+
"learning_rate": 1.0855279411038096e-05,
|
| 27742 |
+
"loss": 5.5502,
|
| 27743 |
+
"step": 15848
|
| 27744 |
+
},
|
| 27745 |
+
{
|
| 27746 |
+
"epoch": 0.78,
|
| 27747 |
+
"grad_norm": 2.058922529220581,
|
| 27748 |
+
"learning_rate": 1.0845397499876477e-05,
|
| 27749 |
+
"loss": 5.5531,
|
| 27750 |
+
"step": 15852
|
| 27751 |
+
},
|
| 27752 |
+
{
|
| 27753 |
+
"epoch": 0.78,
|
| 27754 |
+
"grad_norm": 1.818724274635315,
|
| 27755 |
+
"learning_rate": 1.0835515588714857e-05,
|
| 27756 |
+
"loss": 5.4607,
|
| 27757 |
+
"step": 15856
|
| 27758 |
+
},
|
| 27759 |
+
{
|
| 27760 |
+
"epoch": 0.78,
|
| 27761 |
+
"grad_norm": 2.0503363609313965,
|
| 27762 |
+
"learning_rate": 1.0825633677553239e-05,
|
| 27763 |
+
"loss": 5.4722,
|
| 27764 |
+
"step": 15860
|
| 27765 |
+
},
|
| 27766 |
+
{
|
| 27767 |
+
"epoch": 0.78,
|
| 27768 |
+
"grad_norm": 2.076927423477173,
|
| 27769 |
+
"learning_rate": 1.081575176639162e-05,
|
| 27770 |
+
"loss": 5.4039,
|
| 27771 |
+
"step": 15864
|
| 27772 |
+
},
|
| 27773 |
+
{
|
| 27774 |
+
"epoch": 0.78,
|
| 27775 |
+
"grad_norm": 1.991584300994873,
|
| 27776 |
+
"learning_rate": 1.0805869855230003e-05,
|
| 27777 |
+
"loss": 5.4651,
|
| 27778 |
+
"step": 15868
|
| 27779 |
+
},
|
| 27780 |
+
{
|
| 27781 |
+
"epoch": 0.78,
|
| 27782 |
+
"grad_norm": 2.276181936264038,
|
| 27783 |
+
"learning_rate": 1.0795987944068383e-05,
|
| 27784 |
+
"loss": 5.5605,
|
| 27785 |
+
"step": 15872
|
| 27786 |
+
},
|
| 27787 |
+
{
|
| 27788 |
+
"epoch": 0.78,
|
| 27789 |
+
"grad_norm": 2.0964319705963135,
|
| 27790 |
+
"learning_rate": 1.0786106032906766e-05,
|
| 27791 |
+
"loss": 5.4315,
|
| 27792 |
+
"step": 15876
|
| 27793 |
+
},
|
| 27794 |
+
{
|
| 27795 |
+
"epoch": 0.78,
|
| 27796 |
+
"grad_norm": 2.0152671337127686,
|
| 27797 |
+
"learning_rate": 1.0776224121745146e-05,
|
| 27798 |
+
"loss": 5.5034,
|
| 27799 |
+
"step": 15880
|
| 27800 |
+
},
|
| 27801 |
+
{
|
| 27802 |
+
"epoch": 0.78,
|
| 27803 |
+
"grad_norm": 2.065906524658203,
|
| 27804 |
+
"learning_rate": 1.0766342210583528e-05,
|
| 27805 |
+
"loss": 5.5383,
|
| 27806 |
+
"step": 15884
|
| 27807 |
+
},
|
| 27808 |
+
{
|
| 27809 |
+
"epoch": 0.79,
|
| 27810 |
+
"grad_norm": 1.9010733366012573,
|
| 27811 |
+
"learning_rate": 1.0756460299421908e-05,
|
| 27812 |
+
"loss": 5.5424,
|
| 27813 |
+
"step": 15888
|
| 27814 |
+
},
|
| 27815 |
+
{
|
| 27816 |
+
"epoch": 0.79,
|
| 27817 |
+
"grad_norm": 1.9711081981658936,
|
| 27818 |
+
"learning_rate": 1.0746578388260289e-05,
|
| 27819 |
+
"loss": 5.4899,
|
| 27820 |
+
"step": 15892
|
| 27821 |
+
},
|
| 27822 |
+
{
|
| 27823 |
+
"epoch": 0.79,
|
| 27824 |
+
"grad_norm": 1.9570908546447754,
|
| 27825 |
+
"learning_rate": 1.0736696477098672e-05,
|
| 27826 |
+
"loss": 5.5072,
|
| 27827 |
+
"step": 15896
|
| 27828 |
+
},
|
| 27829 |
+
{
|
| 27830 |
+
"epoch": 0.79,
|
| 27831 |
+
"grad_norm": 2.1524243354797363,
|
| 27832 |
+
"learning_rate": 1.0726814565937053e-05,
|
| 27833 |
+
"loss": 5.4421,
|
| 27834 |
+
"step": 15900
|
| 27835 |
+
},
|
| 27836 |
+
{
|
| 27837 |
+
"epoch": 0.79,
|
| 27838 |
+
"grad_norm": 1.8811511993408203,
|
| 27839 |
+
"learning_rate": 1.0716932654775435e-05,
|
| 27840 |
+
"loss": 5.3984,
|
| 27841 |
+
"step": 15904
|
| 27842 |
+
},
|
| 27843 |
+
{
|
| 27844 |
+
"epoch": 0.79,
|
| 27845 |
+
"grad_norm": 1.914753794670105,
|
| 27846 |
+
"learning_rate": 1.0707050743613815e-05,
|
| 27847 |
+
"loss": 5.5874,
|
| 27848 |
+
"step": 15908
|
| 27849 |
+
},
|
| 27850 |
+
{
|
| 27851 |
+
"epoch": 0.79,
|
| 27852 |
+
"grad_norm": 1.9757344722747803,
|
| 27853 |
+
"learning_rate": 1.0697168832452197e-05,
|
| 27854 |
+
"loss": 5.5402,
|
| 27855 |
+
"step": 15912
|
| 27856 |
+
},
|
| 27857 |
+
{
|
| 27858 |
+
"epoch": 0.79,
|
| 27859 |
+
"grad_norm": 2.1234004497528076,
|
| 27860 |
+
"learning_rate": 1.0689757399080982e-05,
|
| 27861 |
+
"loss": 5.4948,
|
| 27862 |
+
"step": 15916
|
| 27863 |
+
},
|
| 27864 |
+
{
|
| 27865 |
+
"epoch": 0.79,
|
| 27866 |
+
"grad_norm": 1.9321492910385132,
|
| 27867 |
+
"learning_rate": 1.0679875487919364e-05,
|
| 27868 |
+
"loss": 5.3983,
|
| 27869 |
+
"step": 15920
|
| 27870 |
+
},
|
| 27871 |
+
{
|
| 27872 |
+
"epoch": 0.79,
|
| 27873 |
+
"grad_norm": 2.0924885272979736,
|
| 27874 |
+
"learning_rate": 1.0669993576757746e-05,
|
| 27875 |
+
"loss": 5.3964,
|
| 27876 |
+
"step": 15924
|
| 27877 |
+
},
|
| 27878 |
+
{
|
| 27879 |
+
"epoch": 0.79,
|
| 27880 |
+
"grad_norm": 2.1206881999969482,
|
| 27881 |
+
"learning_rate": 1.0660111665596127e-05,
|
| 27882 |
+
"loss": 5.4496,
|
| 27883 |
+
"step": 15928
|
| 27884 |
+
},
|
| 27885 |
+
{
|
| 27886 |
+
"epoch": 0.79,
|
| 27887 |
+
"grad_norm": 2.230121374130249,
|
| 27888 |
+
"learning_rate": 1.0650229754434509e-05,
|
| 27889 |
+
"loss": 5.3782,
|
| 27890 |
+
"step": 15932
|
| 27891 |
+
},
|
| 27892 |
+
{
|
| 27893 |
+
"epoch": 0.79,
|
| 27894 |
+
"grad_norm": 1.9307682514190674,
|
| 27895 |
+
"learning_rate": 1.0640347843272889e-05,
|
| 27896 |
+
"loss": 5.4723,
|
| 27897 |
+
"step": 15936
|
| 27898 |
+
},
|
| 27899 |
+
{
|
| 27900 |
+
"epoch": 0.79,
|
| 27901 |
+
"grad_norm": 2.326744794845581,
|
| 27902 |
+
"learning_rate": 1.0630465932111271e-05,
|
| 27903 |
+
"loss": 5.4944,
|
| 27904 |
+
"step": 15940
|
| 27905 |
+
},
|
| 27906 |
+
{
|
| 27907 |
+
"epoch": 0.79,
|
| 27908 |
+
"grad_norm": 2.0075511932373047,
|
| 27909 |
+
"learning_rate": 1.0620584020949651e-05,
|
| 27910 |
+
"loss": 5.4213,
|
| 27911 |
+
"step": 15944
|
| 27912 |
+
},
|
| 27913 |
+
{
|
| 27914 |
+
"epoch": 0.79,
|
| 27915 |
+
"grad_norm": 2.0154924392700195,
|
| 27916 |
+
"learning_rate": 1.0610702109788034e-05,
|
| 27917 |
+
"loss": 5.5201,
|
| 27918 |
+
"step": 15948
|
| 27919 |
+
},
|
| 27920 |
+
{
|
| 27921 |
+
"epoch": 0.79,
|
| 27922 |
+
"grad_norm": 2.3839402198791504,
|
| 27923 |
+
"learning_rate": 1.0600820198626416e-05,
|
| 27924 |
+
"loss": 5.5373,
|
| 27925 |
+
"step": 15952
|
| 27926 |
+
},
|
| 27927 |
+
{
|
| 27928 |
+
"epoch": 0.79,
|
| 27929 |
+
"grad_norm": 1.8833853006362915,
|
| 27930 |
+
"learning_rate": 1.0590938287464796e-05,
|
| 27931 |
+
"loss": 5.5104,
|
| 27932 |
+
"step": 15956
|
| 27933 |
+
},
|
| 27934 |
+
{
|
| 27935 |
+
"epoch": 0.79,
|
| 27936 |
+
"grad_norm": 2.0425021648406982,
|
| 27937 |
+
"learning_rate": 1.0581056376303178e-05,
|
| 27938 |
+
"loss": 5.5934,
|
| 27939 |
+
"step": 15960
|
| 27940 |
+
},
|
| 27941 |
+
{
|
| 27942 |
+
"epoch": 0.79,
|
| 27943 |
+
"grad_norm": 1.8929284811019897,
|
| 27944 |
+
"learning_rate": 1.0571174465141558e-05,
|
| 27945 |
+
"loss": 5.5095,
|
| 27946 |
+
"step": 15964
|
| 27947 |
+
},
|
| 27948 |
+
{
|
| 27949 |
+
"epoch": 0.79,
|
| 27950 |
+
"grad_norm": 1.8911198377609253,
|
| 27951 |
+
"learning_rate": 1.056129255397994e-05,
|
| 27952 |
+
"loss": 5.4591,
|
| 27953 |
+
"step": 15968
|
| 27954 |
+
},
|
| 27955 |
+
{
|
| 27956 |
+
"epoch": 0.79,
|
| 27957 |
+
"grad_norm": 1.856986403465271,
|
| 27958 |
+
"learning_rate": 1.055141064281832e-05,
|
| 27959 |
+
"loss": 5.35,
|
| 27960 |
+
"step": 15972
|
| 27961 |
+
},
|
| 27962 |
+
{
|
| 27963 |
+
"epoch": 0.79,
|
| 27964 |
+
"grad_norm": 1.8045761585235596,
|
| 27965 |
+
"learning_rate": 1.0541528731656703e-05,
|
| 27966 |
+
"loss": 5.5255,
|
| 27967 |
+
"step": 15976
|
| 27968 |
+
},
|
| 27969 |
+
{
|
| 27970 |
+
"epoch": 0.79,
|
| 27971 |
+
"grad_norm": 1.9339076280593872,
|
| 27972 |
+
"learning_rate": 1.0531646820495085e-05,
|
| 27973 |
+
"loss": 5.4632,
|
| 27974 |
+
"step": 15980
|
| 27975 |
+
},
|
| 27976 |
+
{
|
| 27977 |
+
"epoch": 0.79,
|
| 27978 |
+
"grad_norm": 2.076307535171509,
|
| 27979 |
+
"learning_rate": 1.0521764909333465e-05,
|
| 27980 |
+
"loss": 5.453,
|
| 27981 |
+
"step": 15984
|
| 27982 |
+
},
|
| 27983 |
+
{
|
| 27984 |
+
"epoch": 0.79,
|
| 27985 |
+
"grad_norm": 2.0790350437164307,
|
| 27986 |
+
"learning_rate": 1.0511882998171847e-05,
|
| 27987 |
+
"loss": 5.4504,
|
| 27988 |
+
"step": 15988
|
| 27989 |
+
},
|
| 27990 |
+
{
|
| 27991 |
+
"epoch": 0.79,
|
| 27992 |
+
"grad_norm": 2.015014171600342,
|
| 27993 |
+
"learning_rate": 1.0502001087010228e-05,
|
| 27994 |
+
"loss": 5.4082,
|
| 27995 |
+
"step": 15992
|
| 27996 |
+
},
|
| 27997 |
+
{
|
| 27998 |
+
"epoch": 0.79,
|
| 27999 |
+
"grad_norm": 1.9248104095458984,
|
| 28000 |
+
"learning_rate": 1.049211917584861e-05,
|
| 28001 |
+
"loss": 5.5261,
|
| 28002 |
+
"step": 15996
|
| 28003 |
+
},
|
| 28004 |
+
{
|
| 28005 |
+
"epoch": 0.79,
|
| 28006 |
+
"grad_norm": 1.9845161437988281,
|
| 28007 |
+
"learning_rate": 1.048223726468699e-05,
|
| 28008 |
+
"loss": 5.495,
|
| 28009 |
+
"step": 16000
|
| 28010 |
+
},
|
| 28011 |
+
{
|
| 28012 |
+
"epoch": 0.79,
|
| 28013 |
+
"grad_norm": 2.0011472702026367,
|
| 28014 |
+
"learning_rate": 1.0472355353525372e-05,
|
| 28015 |
+
"loss": 5.3597,
|
| 28016 |
+
"step": 16004
|
| 28017 |
+
},
|
| 28018 |
+
{
|
| 28019 |
+
"epoch": 0.79,
|
| 28020 |
+
"grad_norm": 2.0203723907470703,
|
| 28021 |
+
"learning_rate": 1.0462473442363753e-05,
|
| 28022 |
+
"loss": 5.5997,
|
| 28023 |
+
"step": 16008
|
| 28024 |
+
},
|
| 28025 |
+
{
|
| 28026 |
+
"epoch": 0.79,
|
| 28027 |
+
"grad_norm": 1.9304739236831665,
|
| 28028 |
+
"learning_rate": 1.0452591531202136e-05,
|
| 28029 |
+
"loss": 5.4136,
|
| 28030 |
+
"step": 16012
|
| 28031 |
+
},
|
| 28032 |
+
{
|
| 28033 |
+
"epoch": 0.79,
|
| 28034 |
+
"grad_norm": 2.1026952266693115,
|
| 28035 |
+
"learning_rate": 1.0442709620040517e-05,
|
| 28036 |
+
"loss": 5.5538,
|
| 28037 |
+
"step": 16016
|
| 28038 |
+
},
|
| 28039 |
+
{
|
| 28040 |
+
"epoch": 0.79,
|
| 28041 |
+
"grad_norm": 1.9498411417007446,
|
| 28042 |
+
"learning_rate": 1.0432827708878897e-05,
|
| 28043 |
+
"loss": 5.5066,
|
| 28044 |
+
"step": 16020
|
| 28045 |
+
},
|
| 28046 |
+
{
|
| 28047 |
+
"epoch": 0.79,
|
| 28048 |
+
"grad_norm": 2.043534517288208,
|
| 28049 |
+
"learning_rate": 1.0422945797717279e-05,
|
| 28050 |
+
"loss": 5.4616,
|
| 28051 |
+
"step": 16024
|
| 28052 |
+
},
|
| 28053 |
+
{
|
| 28054 |
+
"epoch": 0.79,
|
| 28055 |
+
"grad_norm": 1.9264411926269531,
|
| 28056 |
+
"learning_rate": 1.041306388655566e-05,
|
| 28057 |
+
"loss": 5.5081,
|
| 28058 |
+
"step": 16028
|
| 28059 |
+
},
|
| 28060 |
+
{
|
| 28061 |
+
"epoch": 0.79,
|
| 28062 |
+
"grad_norm": 1.9634205102920532,
|
| 28063 |
+
"learning_rate": 1.0403181975394042e-05,
|
| 28064 |
+
"loss": 5.436,
|
| 28065 |
+
"step": 16032
|
| 28066 |
+
},
|
| 28067 |
+
{
|
| 28068 |
+
"epoch": 0.79,
|
| 28069 |
+
"grad_norm": 1.9793167114257812,
|
| 28070 |
+
"learning_rate": 1.0393300064232422e-05,
|
| 28071 |
+
"loss": 5.4457,
|
| 28072 |
+
"step": 16036
|
| 28073 |
+
},
|
| 28074 |
+
{
|
| 28075 |
+
"epoch": 0.79,
|
| 28076 |
+
"grad_norm": 2.0029139518737793,
|
| 28077 |
+
"learning_rate": 1.0383418153070806e-05,
|
| 28078 |
+
"loss": 5.5097,
|
| 28079 |
+
"step": 16040
|
| 28080 |
+
},
|
| 28081 |
+
{
|
| 28082 |
+
"epoch": 0.79,
|
| 28083 |
+
"grad_norm": 1.93049156665802,
|
| 28084 |
+
"learning_rate": 1.0373536241909186e-05,
|
| 28085 |
+
"loss": 5.5088,
|
| 28086 |
+
"step": 16044
|
| 28087 |
+
},
|
| 28088 |
+
{
|
| 28089 |
+
"epoch": 0.79,
|
| 28090 |
+
"grad_norm": 1.9566268920898438,
|
| 28091 |
+
"learning_rate": 1.0363654330747566e-05,
|
| 28092 |
+
"loss": 5.509,
|
| 28093 |
+
"step": 16048
|
| 28094 |
+
},
|
| 28095 |
+
{
|
| 28096 |
+
"epoch": 0.79,
|
| 28097 |
+
"grad_norm": 1.9150116443634033,
|
| 28098 |
+
"learning_rate": 1.0353772419585948e-05,
|
| 28099 |
+
"loss": 5.432,
|
| 28100 |
+
"step": 16052
|
| 28101 |
+
},
|
| 28102 |
+
{
|
| 28103 |
+
"epoch": 0.79,
|
| 28104 |
+
"grad_norm": 1.9457064867019653,
|
| 28105 |
+
"learning_rate": 1.0343890508424329e-05,
|
| 28106 |
+
"loss": 5.5555,
|
| 28107 |
+
"step": 16056
|
| 28108 |
+
},
|
| 28109 |
+
{
|
| 28110 |
+
"epoch": 0.79,
|
| 28111 |
+
"grad_norm": 2.037177562713623,
|
| 28112 |
+
"learning_rate": 1.0334008597262711e-05,
|
| 28113 |
+
"loss": 5.4792,
|
| 28114 |
+
"step": 16060
|
| 28115 |
+
},
|
| 28116 |
+
{
|
| 28117 |
+
"epoch": 0.79,
|
| 28118 |
+
"grad_norm": 2.0516912937164307,
|
| 28119 |
+
"learning_rate": 1.0324126686101091e-05,
|
| 28120 |
+
"loss": 5.5053,
|
| 28121 |
+
"step": 16064
|
| 28122 |
+
},
|
| 28123 |
+
{
|
| 28124 |
+
"epoch": 0.79,
|
| 28125 |
+
"grad_norm": 1.7669730186462402,
|
| 28126 |
+
"learning_rate": 1.0314244774939475e-05,
|
| 28127 |
+
"loss": 5.3994,
|
| 28128 |
+
"step": 16068
|
| 28129 |
+
},
|
| 28130 |
+
{
|
| 28131 |
+
"epoch": 0.79,
|
| 28132 |
+
"grad_norm": 2.1283509731292725,
|
| 28133 |
+
"learning_rate": 1.0304362863777855e-05,
|
| 28134 |
+
"loss": 5.466,
|
| 28135 |
+
"step": 16072
|
| 28136 |
+
},
|
| 28137 |
+
{
|
| 28138 |
+
"epoch": 0.79,
|
| 28139 |
+
"grad_norm": 1.7946540117263794,
|
| 28140 |
+
"learning_rate": 1.0294480952616237e-05,
|
| 28141 |
+
"loss": 5.4215,
|
| 28142 |
+
"step": 16076
|
| 28143 |
+
},
|
| 28144 |
+
{
|
| 28145 |
+
"epoch": 0.79,
|
| 28146 |
+
"grad_norm": 2.273894786834717,
|
| 28147 |
+
"learning_rate": 1.0284599041454618e-05,
|
| 28148 |
+
"loss": 5.408,
|
| 28149 |
+
"step": 16080
|
| 28150 |
+
},
|
| 28151 |
+
{
|
| 28152 |
+
"epoch": 0.79,
|
| 28153 |
+
"grad_norm": 2.055126667022705,
|
| 28154 |
+
"learning_rate": 1.0274717130292998e-05,
|
| 28155 |
+
"loss": 5.5304,
|
| 28156 |
+
"step": 16084
|
| 28157 |
+
},
|
| 28158 |
+
{
|
| 28159 |
+
"epoch": 0.79,
|
| 28160 |
+
"grad_norm": 2.0089943408966064,
|
| 28161 |
+
"learning_rate": 1.026483521913138e-05,
|
| 28162 |
+
"loss": 5.4147,
|
| 28163 |
+
"step": 16088
|
| 28164 |
+
},
|
| 28165 |
+
{
|
| 28166 |
+
"epoch": 0.8,
|
| 28167 |
+
"grad_norm": 2.053406000137329,
|
| 28168 |
+
"learning_rate": 1.025495330796976e-05,
|
| 28169 |
+
"loss": 5.5624,
|
| 28170 |
+
"step": 16092
|
| 28171 |
+
},
|
| 28172 |
+
{
|
| 28173 |
+
"epoch": 0.8,
|
| 28174 |
+
"grad_norm": 1.9701050519943237,
|
| 28175 |
+
"learning_rate": 1.0245071396808144e-05,
|
| 28176 |
+
"loss": 5.3615,
|
| 28177 |
+
"step": 16096
|
| 28178 |
+
},
|
| 28179 |
+
{
|
| 28180 |
+
"epoch": 0.8,
|
| 28181 |
+
"grad_norm": 1.9590353965759277,
|
| 28182 |
+
"learning_rate": 1.0235189485646525e-05,
|
| 28183 |
+
"loss": 5.4707,
|
| 28184 |
+
"step": 16100
|
| 28185 |
+
},
|
| 28186 |
+
{
|
| 28187 |
+
"epoch": 0.8,
|
| 28188 |
+
"grad_norm": 2.097073793411255,
|
| 28189 |
+
"learning_rate": 1.0225307574484907e-05,
|
| 28190 |
+
"loss": 5.4783,
|
| 28191 |
+
"step": 16104
|
| 28192 |
+
},
|
| 28193 |
+
{
|
| 28194 |
+
"epoch": 0.8,
|
| 28195 |
+
"grad_norm": 1.9956692457199097,
|
| 28196 |
+
"learning_rate": 1.0215425663323287e-05,
|
| 28197 |
+
"loss": 5.4288,
|
| 28198 |
+
"step": 16108
|
| 28199 |
+
},
|
| 28200 |
+
{
|
| 28201 |
+
"epoch": 0.8,
|
| 28202 |
+
"grad_norm": 1.8568942546844482,
|
| 28203 |
+
"learning_rate": 1.020554375216167e-05,
|
| 28204 |
+
"loss": 5.5691,
|
| 28205 |
+
"step": 16112
|
| 28206 |
+
},
|
| 28207 |
+
{
|
| 28208 |
+
"epoch": 0.8,
|
| 28209 |
+
"grad_norm": 1.9717293977737427,
|
| 28210 |
+
"learning_rate": 1.019566184100005e-05,
|
| 28211 |
+
"loss": 5.4763,
|
| 28212 |
+
"step": 16116
|
| 28213 |
+
},
|
| 28214 |
+
{
|
| 28215 |
+
"epoch": 0.8,
|
| 28216 |
+
"grad_norm": 2.085775375366211,
|
| 28217 |
+
"learning_rate": 1.018577992983843e-05,
|
| 28218 |
+
"loss": 5.5406,
|
| 28219 |
+
"step": 16120
|
| 28220 |
+
},
|
| 28221 |
+
{
|
| 28222 |
+
"epoch": 0.8,
|
| 28223 |
+
"grad_norm": 1.9938271045684814,
|
| 28224 |
+
"learning_rate": 1.0175898018676814e-05,
|
| 28225 |
+
"loss": 5.52,
|
| 28226 |
+
"step": 16124
|
| 28227 |
+
},
|
| 28228 |
+
{
|
| 28229 |
+
"epoch": 0.8,
|
| 28230 |
+
"grad_norm": 2.1408092975616455,
|
| 28231 |
+
"learning_rate": 1.0166016107515194e-05,
|
| 28232 |
+
"loss": 5.4718,
|
| 28233 |
+
"step": 16128
|
| 28234 |
+
},
|
| 28235 |
+
{
|
| 28236 |
+
"epoch": 0.8,
|
| 28237 |
+
"grad_norm": 2.084689140319824,
|
| 28238 |
+
"learning_rate": 1.0156134196353576e-05,
|
| 28239 |
+
"loss": 5.5263,
|
| 28240 |
+
"step": 16132
|
| 28241 |
+
},
|
| 28242 |
+
{
|
| 28243 |
+
"epoch": 0.8,
|
| 28244 |
+
"grad_norm": 1.8502254486083984,
|
| 28245 |
+
"learning_rate": 1.0146252285191956e-05,
|
| 28246 |
+
"loss": 5.4056,
|
| 28247 |
+
"step": 16136
|
| 28248 |
+
},
|
| 28249 |
+
{
|
| 28250 |
+
"epoch": 0.8,
|
| 28251 |
+
"grad_norm": 1.832261323928833,
|
| 28252 |
+
"learning_rate": 1.0136370374030339e-05,
|
| 28253 |
+
"loss": 5.4597,
|
| 28254 |
+
"step": 16140
|
| 28255 |
+
},
|
| 28256 |
+
{
|
| 28257 |
+
"epoch": 0.8,
|
| 28258 |
+
"grad_norm": 1.9201068878173828,
|
| 28259 |
+
"learning_rate": 1.0126488462868719e-05,
|
| 28260 |
+
"loss": 5.5378,
|
| 28261 |
+
"step": 16144
|
| 28262 |
+
},
|
| 28263 |
+
{
|
| 28264 |
+
"epoch": 0.8,
|
| 28265 |
+
"grad_norm": 1.8467752933502197,
|
| 28266 |
+
"learning_rate": 1.0116606551707101e-05,
|
| 28267 |
+
"loss": 5.4567,
|
| 28268 |
+
"step": 16148
|
| 28269 |
+
},
|
| 28270 |
+
{
|
| 28271 |
+
"epoch": 0.8,
|
| 28272 |
+
"grad_norm": 2.1131222248077393,
|
| 28273 |
+
"learning_rate": 1.0106724640545481e-05,
|
| 28274 |
+
"loss": 5.35,
|
| 28275 |
+
"step": 16152
|
| 28276 |
+
},
|
| 28277 |
+
{
|
| 28278 |
+
"epoch": 0.8,
|
| 28279 |
+
"grad_norm": 2.0209567546844482,
|
| 28280 |
+
"learning_rate": 1.0096842729383863e-05,
|
| 28281 |
+
"loss": 5.5339,
|
| 28282 |
+
"step": 16156
|
| 28283 |
+
},
|
| 28284 |
+
{
|
| 28285 |
+
"epoch": 0.8,
|
| 28286 |
+
"grad_norm": 2.1684625148773193,
|
| 28287 |
+
"learning_rate": 1.0086960818222245e-05,
|
| 28288 |
+
"loss": 5.4267,
|
| 28289 |
+
"step": 16160
|
| 28290 |
+
},
|
| 28291 |
+
{
|
| 28292 |
+
"epoch": 0.8,
|
| 28293 |
+
"grad_norm": 2.1010987758636475,
|
| 28294 |
+
"learning_rate": 1.0077078907060626e-05,
|
| 28295 |
+
"loss": 5.4042,
|
| 28296 |
+
"step": 16164
|
| 28297 |
+
},
|
| 28298 |
+
{
|
| 28299 |
+
"epoch": 0.8,
|
| 28300 |
+
"grad_norm": 1.9701296091079712,
|
| 28301 |
+
"learning_rate": 1.0067196995899008e-05,
|
| 28302 |
+
"loss": 5.6227,
|
| 28303 |
+
"step": 16168
|
| 28304 |
+
},
|
| 28305 |
+
{
|
| 28306 |
+
"epoch": 0.8,
|
| 28307 |
+
"grad_norm": 1.8271695375442505,
|
| 28308 |
+
"learning_rate": 1.0057315084737388e-05,
|
| 28309 |
+
"loss": 5.4151,
|
| 28310 |
+
"step": 16172
|
| 28311 |
+
},
|
| 28312 |
+
{
|
| 28313 |
+
"epoch": 0.8,
|
| 28314 |
+
"grad_norm": 2.2199959754943848,
|
| 28315 |
+
"learning_rate": 1.004743317357577e-05,
|
| 28316 |
+
"loss": 5.5352,
|
| 28317 |
+
"step": 16176
|
| 28318 |
+
},
|
| 28319 |
+
{
|
| 28320 |
+
"epoch": 0.8,
|
| 28321 |
+
"grad_norm": 2.2069809436798096,
|
| 28322 |
+
"learning_rate": 1.003755126241415e-05,
|
| 28323 |
+
"loss": 5.4987,
|
| 28324 |
+
"step": 16180
|
| 28325 |
+
},
|
| 28326 |
+
{
|
| 28327 |
+
"epoch": 0.8,
|
| 28328 |
+
"grad_norm": 2.027318239212036,
|
| 28329 |
+
"learning_rate": 1.0027669351252533e-05,
|
| 28330 |
+
"loss": 5.5284,
|
| 28331 |
+
"step": 16184
|
| 28332 |
+
},
|
| 28333 |
+
{
|
| 28334 |
+
"epoch": 0.8,
|
| 28335 |
+
"grad_norm": 1.9697614908218384,
|
| 28336 |
+
"learning_rate": 1.0017787440090915e-05,
|
| 28337 |
+
"loss": 5.5906,
|
| 28338 |
+
"step": 16188
|
| 28339 |
+
},
|
| 28340 |
+
{
|
| 28341 |
+
"epoch": 0.8,
|
| 28342 |
+
"grad_norm": 2.252358913421631,
|
| 28343 |
+
"learning_rate": 1.0007905528929295e-05,
|
| 28344 |
+
"loss": 5.4911,
|
| 28345 |
+
"step": 16192
|
| 28346 |
}
|
| 28347 |
],
|
| 28348 |
"logging_steps": 4,
|
|
|
|
| 28350 |
"num_input_tokens_seen": 0,
|
| 28351 |
"num_train_epochs": 1,
|
| 28352 |
"save_steps": 2024,
|
| 28353 |
+
"total_flos": 6.820471143346995e+16,
|
| 28354 |
"train_batch_size": 8,
|
| 28355 |
"trial_name": null,
|
| 28356 |
"trial_params": null
|