Training in progress, step 974, checkpoint
Browse files
last-checkpoint/adapter_model.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 100059752
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:fa87ab3f59e4df8333b56bc88f2bb7e552498b6faa21bd35b8e43ed55274c72e
|
3 |
size 100059752
|
last-checkpoint/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 51244404
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:d3c519724662a5b439cc0d825599a4cff5f90d21bdc3475ab7b31fbced3a9ae0
|
3 |
size 51244404
|
last-checkpoint/rng_state.pth
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 14244
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:d3270b8e107d5b2342fb347c198bb80cb377e6e0b722de88aea18663deed1dac
|
3 |
size 14244
|
last-checkpoint/scheduler.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 1064
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:5f57cc71b1136da66608c7abb64fd70bf0cf409b432dcb054547648718e52346
|
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":
|
5 |
"eval_steps": 244,
|
6 |
-
"global_step":
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
@@ -5163,6 +5163,1700 @@
|
|
5163 |
"eval_samples_per_second": 15.281,
|
5164 |
"eval_steps_per_second": 7.641,
|
5165 |
"step": 732
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
5166 |
}
|
5167 |
],
|
5168 |
"logging_steps": 1,
|
@@ -5177,12 +6871,12 @@
|
|
5177 |
"should_evaluate": false,
|
5178 |
"should_log": false,
|
5179 |
"should_save": true,
|
5180 |
-
"should_training_stop":
|
5181 |
},
|
5182 |
"attributes": {}
|
5183 |
}
|
5184 |
},
|
5185 |
-
"total_flos":
|
5186 |
"train_batch_size": 2,
|
5187 |
"trial_name": null,
|
5188 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 1.0005134788189987,
|
5 |
"eval_steps": 244,
|
6 |
+
"global_step": 974,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
5163 |
"eval_samples_per_second": 15.281,
|
5164 |
"eval_steps_per_second": 7.641,
|
5165 |
"step": 732
|
5166 |
+
},
|
5167 |
+
{
|
5168 |
+
"epoch": 0.7527599486521181,
|
5169 |
+
"grad_norm": 1.2231281995773315,
|
5170 |
+
"learning_rate": 2.9289321881345254e-05,
|
5171 |
+
"loss": 2.0821,
|
5172 |
+
"step": 733
|
5173 |
+
},
|
5174 |
+
{
|
5175 |
+
"epoch": 0.7537869062901156,
|
5176 |
+
"grad_norm": 1.4395884275436401,
|
5177 |
+
"learning_rate": 2.9059257794488424e-05,
|
5178 |
+
"loss": 2.2183,
|
5179 |
+
"step": 734
|
5180 |
+
},
|
5181 |
+
{
|
5182 |
+
"epoch": 0.754813863928113,
|
5183 |
+
"grad_norm": 1.5501489639282227,
|
5184 |
+
"learning_rate": 2.882994713435658e-05,
|
5185 |
+
"loss": 2.4913,
|
5186 |
+
"step": 735
|
5187 |
+
},
|
5188 |
+
{
|
5189 |
+
"epoch": 0.7558408215661104,
|
5190 |
+
"grad_norm": 1.4353766441345215,
|
5191 |
+
"learning_rate": 2.860139233634547e-05,
|
5192 |
+
"loss": 2.2493,
|
5193 |
+
"step": 736
|
5194 |
+
},
|
5195 |
+
{
|
5196 |
+
"epoch": 0.7568677792041079,
|
5197 |
+
"grad_norm": 1.2123923301696777,
|
5198 |
+
"learning_rate": 2.8373595827823086e-05,
|
5199 |
+
"loss": 2.3579,
|
5200 |
+
"step": 737
|
5201 |
+
},
|
5202 |
+
{
|
5203 |
+
"epoch": 0.7578947368421053,
|
5204 |
+
"grad_norm": 1.1628131866455078,
|
5205 |
+
"learning_rate": 2.8146560028104153e-05,
|
5206 |
+
"loss": 2.1711,
|
5207 |
+
"step": 738
|
5208 |
+
},
|
5209 |
+
{
|
5210 |
+
"epoch": 0.7589216944801027,
|
5211 |
+
"grad_norm": 1.0057865381240845,
|
5212 |
+
"learning_rate": 2.792028734842418e-05,
|
5213 |
+
"loss": 1.7073,
|
5214 |
+
"step": 739
|
5215 |
+
},
|
5216 |
+
{
|
5217 |
+
"epoch": 0.7599486521181001,
|
5218 |
+
"grad_norm": 1.5157403945922852,
|
5219 |
+
"learning_rate": 2.7694780191914006e-05,
|
5220 |
+
"loss": 2.3561,
|
5221 |
+
"step": 740
|
5222 |
+
},
|
5223 |
+
{
|
5224 |
+
"epoch": 0.7609756097560976,
|
5225 |
+
"grad_norm": 1.333644151687622,
|
5226 |
+
"learning_rate": 2.7470040953574238e-05,
|
5227 |
+
"loss": 2.3166,
|
5228 |
+
"step": 741
|
5229 |
+
},
|
5230 |
+
{
|
5231 |
+
"epoch": 0.762002567394095,
|
5232 |
+
"grad_norm": 1.4937711954116821,
|
5233 |
+
"learning_rate": 2.724607202024969e-05,
|
5234 |
+
"loss": 2.2235,
|
5235 |
+
"step": 742
|
5236 |
+
},
|
5237 |
+
{
|
5238 |
+
"epoch": 0.7630295250320924,
|
5239 |
+
"grad_norm": 1.6282092332839966,
|
5240 |
+
"learning_rate": 2.7022875770604284e-05,
|
5241 |
+
"loss": 2.3438,
|
5242 |
+
"step": 743
|
5243 |
+
},
|
5244 |
+
{
|
5245 |
+
"epoch": 0.7640564826700899,
|
5246 |
+
"grad_norm": 1.3984050750732422,
|
5247 |
+
"learning_rate": 2.6800454575095567e-05,
|
5248 |
+
"loss": 2.2035,
|
5249 |
+
"step": 744
|
5250 |
+
},
|
5251 |
+
{
|
5252 |
+
"epoch": 0.7650834403080873,
|
5253 |
+
"grad_norm": 1.3804353475570679,
|
5254 |
+
"learning_rate": 2.6578810795949682e-05,
|
5255 |
+
"loss": 2.4949,
|
5256 |
+
"step": 745
|
5257 |
+
},
|
5258 |
+
{
|
5259 |
+
"epoch": 0.7661103979460847,
|
5260 |
+
"grad_norm": 1.403314232826233,
|
5261 |
+
"learning_rate": 2.6357946787136113e-05,
|
5262 |
+
"loss": 2.4336,
|
5263 |
+
"step": 746
|
5264 |
+
},
|
5265 |
+
{
|
5266 |
+
"epoch": 0.7671373555840821,
|
5267 |
+
"grad_norm": 1.34967041015625,
|
5268 |
+
"learning_rate": 2.613786489434287e-05,
|
5269 |
+
"loss": 2.2571,
|
5270 |
+
"step": 747
|
5271 |
+
},
|
5272 |
+
{
|
5273 |
+
"epoch": 0.7681643132220796,
|
5274 |
+
"grad_norm": 1.5277897119522095,
|
5275 |
+
"learning_rate": 2.591856745495148e-05,
|
5276 |
+
"loss": 2.1579,
|
5277 |
+
"step": 748
|
5278 |
+
},
|
5279 |
+
{
|
5280 |
+
"epoch": 0.769191270860077,
|
5281 |
+
"grad_norm": 1.5782662630081177,
|
5282 |
+
"learning_rate": 2.5700056798012163e-05,
|
5283 |
+
"loss": 2.2695,
|
5284 |
+
"step": 749
|
5285 |
+
},
|
5286 |
+
{
|
5287 |
+
"epoch": 0.7702182284980744,
|
5288 |
+
"grad_norm": 1.5577448606491089,
|
5289 |
+
"learning_rate": 2.548233524421911e-05,
|
5290 |
+
"loss": 2.234,
|
5291 |
+
"step": 750
|
5292 |
+
},
|
5293 |
+
{
|
5294 |
+
"epoch": 0.7712451861360718,
|
5295 |
+
"grad_norm": 1.2873456478118896,
|
5296 |
+
"learning_rate": 2.5265405105885855e-05,
|
5297 |
+
"loss": 2.0918,
|
5298 |
+
"step": 751
|
5299 |
+
},
|
5300 |
+
{
|
5301 |
+
"epoch": 0.7722721437740693,
|
5302 |
+
"grad_norm": 1.7132203578948975,
|
5303 |
+
"learning_rate": 2.5049268686920667e-05,
|
5304 |
+
"loss": 2.2356,
|
5305 |
+
"step": 752
|
5306 |
+
},
|
5307 |
+
{
|
5308 |
+
"epoch": 0.7732991014120667,
|
5309 |
+
"grad_norm": 1.3685425519943237,
|
5310 |
+
"learning_rate": 2.4833928282802132e-05,
|
5311 |
+
"loss": 2.1547,
|
5312 |
+
"step": 753
|
5313 |
+
},
|
5314 |
+
{
|
5315 |
+
"epoch": 0.7743260590500641,
|
5316 |
+
"grad_norm": 1.3102649450302124,
|
5317 |
+
"learning_rate": 2.461938618055478e-05,
|
5318 |
+
"loss": 2.1503,
|
5319 |
+
"step": 754
|
5320 |
+
},
|
5321 |
+
{
|
5322 |
+
"epoch": 0.7753530166880617,
|
5323 |
+
"grad_norm": 1.4933326244354248,
|
5324 |
+
"learning_rate": 2.440564465872469e-05,
|
5325 |
+
"loss": 2.2991,
|
5326 |
+
"step": 755
|
5327 |
+
},
|
5328 |
+
{
|
5329 |
+
"epoch": 0.7763799743260591,
|
5330 |
+
"grad_norm": 1.284360647201538,
|
5331 |
+
"learning_rate": 2.4192705987355424e-05,
|
5332 |
+
"loss": 2.145,
|
5333 |
+
"step": 756
|
5334 |
+
},
|
5335 |
+
{
|
5336 |
+
"epoch": 0.7774069319640565,
|
5337 |
+
"grad_norm": 1.3690736293792725,
|
5338 |
+
"learning_rate": 2.3980572427963887e-05,
|
5339 |
+
"loss": 2.2586,
|
5340 |
+
"step": 757
|
5341 |
+
},
|
5342 |
+
{
|
5343 |
+
"epoch": 0.778433889602054,
|
5344 |
+
"grad_norm": 1.6226760149002075,
|
5345 |
+
"learning_rate": 2.3769246233516242e-05,
|
5346 |
+
"loss": 2.4489,
|
5347 |
+
"step": 758
|
5348 |
+
},
|
5349 |
+
{
|
5350 |
+
"epoch": 0.7794608472400514,
|
5351 |
+
"grad_norm": 1.2062180042266846,
|
5352 |
+
"learning_rate": 2.3558729648404065e-05,
|
5353 |
+
"loss": 2.013,
|
5354 |
+
"step": 759
|
5355 |
+
},
|
5356 |
+
{
|
5357 |
+
"epoch": 0.7804878048780488,
|
5358 |
+
"grad_norm": 1.2239223718643188,
|
5359 |
+
"learning_rate": 2.33490249084204e-05,
|
5360 |
+
"loss": 2.2947,
|
5361 |
+
"step": 760
|
5362 |
+
},
|
5363 |
+
{
|
5364 |
+
"epoch": 0.7815147625160462,
|
5365 |
+
"grad_norm": 1.7042533159255981,
|
5366 |
+
"learning_rate": 2.3140134240736168e-05,
|
5367 |
+
"loss": 2.1473,
|
5368 |
+
"step": 761
|
5369 |
+
},
|
5370 |
+
{
|
5371 |
+
"epoch": 0.7825417201540437,
|
5372 |
+
"grad_norm": 1.2973995208740234,
|
5373 |
+
"learning_rate": 2.2932059863876365e-05,
|
5374 |
+
"loss": 2.1991,
|
5375 |
+
"step": 762
|
5376 |
+
},
|
5377 |
+
{
|
5378 |
+
"epoch": 0.7835686777920411,
|
5379 |
+
"grad_norm": 1.2338628768920898,
|
5380 |
+
"learning_rate": 2.272480398769662e-05,
|
5381 |
+
"loss": 2.1853,
|
5382 |
+
"step": 763
|
5383 |
+
},
|
5384 |
+
{
|
5385 |
+
"epoch": 0.7845956354300385,
|
5386 |
+
"grad_norm": 2.069164991378784,
|
5387 |
+
"learning_rate": 2.2518368813359637e-05,
|
5388 |
+
"loss": 2.2275,
|
5389 |
+
"step": 764
|
5390 |
+
},
|
5391 |
+
{
|
5392 |
+
"epoch": 0.785622593068036,
|
5393 |
+
"grad_norm": 1.2554938793182373,
|
5394 |
+
"learning_rate": 2.231275653331181e-05,
|
5395 |
+
"loss": 2.2557,
|
5396 |
+
"step": 765
|
5397 |
+
},
|
5398 |
+
{
|
5399 |
+
"epoch": 0.7866495507060334,
|
5400 |
+
"grad_norm": 1.2675412893295288,
|
5401 |
+
"learning_rate": 2.2107969331260048e-05,
|
5402 |
+
"loss": 2.3231,
|
5403 |
+
"step": 766
|
5404 |
+
},
|
5405 |
+
{
|
5406 |
+
"epoch": 0.7876765083440308,
|
5407 |
+
"grad_norm": 1.4239006042480469,
|
5408 |
+
"learning_rate": 2.1904009382148472e-05,
|
5409 |
+
"loss": 2.2516,
|
5410 |
+
"step": 767
|
5411 |
+
},
|
5412 |
+
{
|
5413 |
+
"epoch": 0.7887034659820282,
|
5414 |
+
"grad_norm": 1.2731049060821533,
|
5415 |
+
"learning_rate": 2.170087885213541e-05,
|
5416 |
+
"loss": 2.1673,
|
5417 |
+
"step": 768
|
5418 |
+
},
|
5419 |
+
{
|
5420 |
+
"epoch": 0.7897304236200257,
|
5421 |
+
"grad_norm": 1.8812044858932495,
|
5422 |
+
"learning_rate": 2.1498579898570227e-05,
|
5423 |
+
"loss": 2.1782,
|
5424 |
+
"step": 769
|
5425 |
+
},
|
5426 |
+
{
|
5427 |
+
"epoch": 0.7907573812580231,
|
5428 |
+
"grad_norm": 1.426591396331787,
|
5429 |
+
"learning_rate": 2.1297114669970618e-05,
|
5430 |
+
"loss": 1.9592,
|
5431 |
+
"step": 770
|
5432 |
+
},
|
5433 |
+
{
|
5434 |
+
"epoch": 0.7917843388960205,
|
5435 |
+
"grad_norm": 1.302981972694397,
|
5436 |
+
"learning_rate": 2.109648530599968e-05,
|
5437 |
+
"loss": 2.2618,
|
5438 |
+
"step": 771
|
5439 |
+
},
|
5440 |
+
{
|
5441 |
+
"epoch": 0.7928112965340179,
|
5442 |
+
"grad_norm": 1.8140078783035278,
|
5443 |
+
"learning_rate": 2.089669393744319e-05,
|
5444 |
+
"loss": 2.3927,
|
5445 |
+
"step": 772
|
5446 |
+
},
|
5447 |
+
{
|
5448 |
+
"epoch": 0.7938382541720154,
|
5449 |
+
"grad_norm": 2.2119557857513428,
|
5450 |
+
"learning_rate": 2.0697742686187017e-05,
|
5451 |
+
"loss": 2.2843,
|
5452 |
+
"step": 773
|
5453 |
+
},
|
5454 |
+
{
|
5455 |
+
"epoch": 0.7948652118100128,
|
5456 |
+
"grad_norm": 1.7960866689682007,
|
5457 |
+
"learning_rate": 2.049963366519446e-05,
|
5458 |
+
"loss": 2.4243,
|
5459 |
+
"step": 774
|
5460 |
+
},
|
5461 |
+
{
|
5462 |
+
"epoch": 0.7958921694480102,
|
5463 |
+
"grad_norm": 1.459788203239441,
|
5464 |
+
"learning_rate": 2.030236897848402e-05,
|
5465 |
+
"loss": 2.2595,
|
5466 |
+
"step": 775
|
5467 |
+
},
|
5468 |
+
{
|
5469 |
+
"epoch": 0.7969191270860077,
|
5470 |
+
"grad_norm": 1.232600450515747,
|
5471 |
+
"learning_rate": 2.0105950721106894e-05,
|
5472 |
+
"loss": 2.1295,
|
5473 |
+
"step": 776
|
5474 |
+
},
|
5475 |
+
{
|
5476 |
+
"epoch": 0.7979460847240052,
|
5477 |
+
"grad_norm": 1.4180529117584229,
|
5478 |
+
"learning_rate": 1.9910380979124754e-05,
|
5479 |
+
"loss": 2.1814,
|
5480 |
+
"step": 777
|
5481 |
+
},
|
5482 |
+
{
|
5483 |
+
"epoch": 0.7989730423620026,
|
5484 |
+
"grad_norm": 1.501356601715088,
|
5485 |
+
"learning_rate": 1.971566182958765e-05,
|
5486 |
+
"loss": 2.4455,
|
5487 |
+
"step": 778
|
5488 |
+
},
|
5489 |
+
{
|
5490 |
+
"epoch": 0.8,
|
5491 |
+
"grad_norm": 1.6094731092453003,
|
5492 |
+
"learning_rate": 1.952179534051183e-05,
|
5493 |
+
"loss": 2.3726,
|
5494 |
+
"step": 779
|
5495 |
+
},
|
5496 |
+
{
|
5497 |
+
"epoch": 0.8010269576379975,
|
5498 |
+
"grad_norm": 1.3195991516113281,
|
5499 |
+
"learning_rate": 1.9328783570857957e-05,
|
5500 |
+
"loss": 2.2679,
|
5501 |
+
"step": 780
|
5502 |
+
},
|
5503 |
+
{
|
5504 |
+
"epoch": 0.8020539152759949,
|
5505 |
+
"grad_norm": 1.4381358623504639,
|
5506 |
+
"learning_rate": 1.9136628570509063e-05,
|
5507 |
+
"loss": 2.0012,
|
5508 |
+
"step": 781
|
5509 |
+
},
|
5510 |
+
{
|
5511 |
+
"epoch": 0.8030808729139923,
|
5512 |
+
"grad_norm": 1.612783670425415,
|
5513 |
+
"learning_rate": 1.8945332380248913e-05,
|
5514 |
+
"loss": 2.2622,
|
5515 |
+
"step": 782
|
5516 |
+
},
|
5517 |
+
{
|
5518 |
+
"epoch": 0.8041078305519898,
|
5519 |
+
"grad_norm": 1.2924425601959229,
|
5520 |
+
"learning_rate": 1.8754897031740192e-05,
|
5521 |
+
"loss": 2.0476,
|
5522 |
+
"step": 783
|
5523 |
+
},
|
5524 |
+
{
|
5525 |
+
"epoch": 0.8051347881899872,
|
5526 |
+
"grad_norm": 1.5139434337615967,
|
5527 |
+
"learning_rate": 1.856532454750307e-05,
|
5528 |
+
"loss": 2.0912,
|
5529 |
+
"step": 784
|
5530 |
+
},
|
5531 |
+
{
|
5532 |
+
"epoch": 0.8061617458279846,
|
5533 |
+
"grad_norm": 1.4951143264770508,
|
5534 |
+
"learning_rate": 1.8376616940893654e-05,
|
5535 |
+
"loss": 2.3838,
|
5536 |
+
"step": 785
|
5537 |
+
},
|
5538 |
+
{
|
5539 |
+
"epoch": 0.807188703465982,
|
5540 |
+
"grad_norm": 1.432755947113037,
|
5541 |
+
"learning_rate": 1.8188776216082603e-05,
|
5542 |
+
"loss": 2.2717,
|
5543 |
+
"step": 786
|
5544 |
+
},
|
5545 |
+
{
|
5546 |
+
"epoch": 0.8082156611039795,
|
5547 |
+
"grad_norm": 1.244824767112732,
|
5548 |
+
"learning_rate": 1.800180436803386e-05,
|
5549 |
+
"loss": 1.8481,
|
5550 |
+
"step": 787
|
5551 |
+
},
|
5552 |
+
{
|
5553 |
+
"epoch": 0.8092426187419769,
|
5554 |
+
"grad_norm": 1.570754885673523,
|
5555 |
+
"learning_rate": 1.7815703382483417e-05,
|
5556 |
+
"loss": 2.3096,
|
5557 |
+
"step": 788
|
5558 |
+
},
|
5559 |
+
{
|
5560 |
+
"epoch": 0.8102695763799743,
|
5561 |
+
"grad_norm": 1.3875128030776978,
|
5562 |
+
"learning_rate": 1.7630475235918308e-05,
|
5563 |
+
"loss": 2.0865,
|
5564 |
+
"step": 789
|
5565 |
+
},
|
5566 |
+
{
|
5567 |
+
"epoch": 0.8112965340179717,
|
5568 |
+
"grad_norm": 1.4651671648025513,
|
5569 |
+
"learning_rate": 1.7446121895555555e-05,
|
5570 |
+
"loss": 2.4152,
|
5571 |
+
"step": 790
|
5572 |
+
},
|
5573 |
+
{
|
5574 |
+
"epoch": 0.8123234916559692,
|
5575 |
+
"grad_norm": 1.1608806848526,
|
5576 |
+
"learning_rate": 1.7262645319321324e-05,
|
5577 |
+
"loss": 2.214,
|
5578 |
+
"step": 791
|
5579 |
+
},
|
5580 |
+
{
|
5581 |
+
"epoch": 0.8133504492939666,
|
5582 |
+
"grad_norm": 1.622000813484192,
|
5583 |
+
"learning_rate": 1.708004745583003e-05,
|
5584 |
+
"loss": 2.2874,
|
5585 |
+
"step": 792
|
5586 |
+
},
|
5587 |
+
{
|
5588 |
+
"epoch": 0.814377406931964,
|
5589 |
+
"grad_norm": 1.3545804023742676,
|
5590 |
+
"learning_rate": 1.689833024436377e-05,
|
5591 |
+
"loss": 2.3003,
|
5592 |
+
"step": 793
|
5593 |
+
},
|
5594 |
+
{
|
5595 |
+
"epoch": 0.8154043645699615,
|
5596 |
+
"grad_norm": 1.4039793014526367,
|
5597 |
+
"learning_rate": 1.6717495614851652e-05,
|
5598 |
+
"loss": 1.8358,
|
5599 |
+
"step": 794
|
5600 |
+
},
|
5601 |
+
{
|
5602 |
+
"epoch": 0.8164313222079589,
|
5603 |
+
"grad_norm": 1.3507652282714844,
|
5604 |
+
"learning_rate": 1.6537545487849336e-05,
|
5605 |
+
"loss": 2.2534,
|
5606 |
+
"step": 795
|
5607 |
+
},
|
5608 |
+
{
|
5609 |
+
"epoch": 0.8174582798459563,
|
5610 |
+
"grad_norm": 1.4168148040771484,
|
5611 |
+
"learning_rate": 1.6358481774518606e-05,
|
5612 |
+
"loss": 2.2372,
|
5613 |
+
"step": 796
|
5614 |
+
},
|
5615 |
+
{
|
5616 |
+
"epoch": 0.8184852374839537,
|
5617 |
+
"grad_norm": 1.266169786453247,
|
5618 |
+
"learning_rate": 1.6180306376607035e-05,
|
5619 |
+
"loss": 2.387,
|
5620 |
+
"step": 797
|
5621 |
+
},
|
5622 |
+
{
|
5623 |
+
"epoch": 0.8195121951219512,
|
5624 |
+
"grad_norm": 1.277815818786621,
|
5625 |
+
"learning_rate": 1.6003021186427893e-05,
|
5626 |
+
"loss": 2.2015,
|
5627 |
+
"step": 798
|
5628 |
+
},
|
5629 |
+
{
|
5630 |
+
"epoch": 0.8205391527599486,
|
5631 |
+
"grad_norm": 1.7137157917022705,
|
5632 |
+
"learning_rate": 1.5826628086839968e-05,
|
5633 |
+
"loss": 2.1036,
|
5634 |
+
"step": 799
|
5635 |
+
},
|
5636 |
+
{
|
5637 |
+
"epoch": 0.8215661103979461,
|
5638 |
+
"grad_norm": 1.4451020956039429,
|
5639 |
+
"learning_rate": 1.5651128951227612e-05,
|
5640 |
+
"loss": 2.0347,
|
5641 |
+
"step": 800
|
5642 |
+
},
|
5643 |
+
{
|
5644 |
+
"epoch": 0.8225930680359436,
|
5645 |
+
"grad_norm": 1.3843578100204468,
|
5646 |
+
"learning_rate": 1.547652564348082e-05,
|
5647 |
+
"loss": 2.1694,
|
5648 |
+
"step": 801
|
5649 |
+
},
|
5650 |
+
{
|
5651 |
+
"epoch": 0.823620025673941,
|
5652 |
+
"grad_norm": 1.3804799318313599,
|
5653 |
+
"learning_rate": 1.5302820017975394e-05,
|
5654 |
+
"loss": 2.0706,
|
5655 |
+
"step": 802
|
5656 |
+
},
|
5657 |
+
{
|
5658 |
+
"epoch": 0.8246469833119384,
|
5659 |
+
"grad_norm": 1.2779897451400757,
|
5660 |
+
"learning_rate": 1.5130013919553355e-05,
|
5661 |
+
"loss": 2.2537,
|
5662 |
+
"step": 803
|
5663 |
+
},
|
5664 |
+
{
|
5665 |
+
"epoch": 0.8256739409499358,
|
5666 |
+
"grad_norm": 1.3886756896972656,
|
5667 |
+
"learning_rate": 1.4958109183503243e-05,
|
5668 |
+
"loss": 2.412,
|
5669 |
+
"step": 804
|
5670 |
+
},
|
5671 |
+
{
|
5672 |
+
"epoch": 0.8267008985879333,
|
5673 |
+
"grad_norm": 1.5321778059005737,
|
5674 |
+
"learning_rate": 1.4787107635540732e-05,
|
5675 |
+
"loss": 2.0181,
|
5676 |
+
"step": 805
|
5677 |
+
},
|
5678 |
+
{
|
5679 |
+
"epoch": 0.8277278562259307,
|
5680 |
+
"grad_norm": 1.277282953262329,
|
5681 |
+
"learning_rate": 1.4617011091789135e-05,
|
5682 |
+
"loss": 2.2371,
|
5683 |
+
"step": 806
|
5684 |
+
},
|
5685 |
+
{
|
5686 |
+
"epoch": 0.8287548138639281,
|
5687 |
+
"grad_norm": 1.8173282146453857,
|
5688 |
+
"learning_rate": 1.4447821358760127e-05,
|
5689 |
+
"loss": 2.3445,
|
5690 |
+
"step": 807
|
5691 |
+
},
|
5692 |
+
{
|
5693 |
+
"epoch": 0.8297817715019256,
|
5694 |
+
"grad_norm": 2.2534754276275635,
|
5695 |
+
"learning_rate": 1.4279540233334665e-05,
|
5696 |
+
"loss": 2.3177,
|
5697 |
+
"step": 808
|
5698 |
+
},
|
5699 |
+
{
|
5700 |
+
"epoch": 0.830808729139923,
|
5701 |
+
"grad_norm": 1.2119274139404297,
|
5702 |
+
"learning_rate": 1.4112169502743799e-05,
|
5703 |
+
"loss": 2.0937,
|
5704 |
+
"step": 809
|
5705 |
+
},
|
5706 |
+
{
|
5707 |
+
"epoch": 0.8318356867779204,
|
5708 |
+
"grad_norm": 1.1014364957809448,
|
5709 |
+
"learning_rate": 1.3945710944549706e-05,
|
5710 |
+
"loss": 2.0666,
|
5711 |
+
"step": 810
|
5712 |
+
},
|
5713 |
+
{
|
5714 |
+
"epoch": 0.8328626444159178,
|
5715 |
+
"grad_norm": 1.0724173784255981,
|
5716 |
+
"learning_rate": 1.3780166326626876e-05,
|
5717 |
+
"loss": 2.383,
|
5718 |
+
"step": 811
|
5719 |
+
},
|
5720 |
+
{
|
5721 |
+
"epoch": 0.8338896020539153,
|
5722 |
+
"grad_norm": 1.3168890476226807,
|
5723 |
+
"learning_rate": 1.361553740714323e-05,
|
5724 |
+
"loss": 2.0837,
|
5725 |
+
"step": 812
|
5726 |
+
},
|
5727 |
+
{
|
5728 |
+
"epoch": 0.8349165596919127,
|
5729 |
+
"grad_norm": 1.224159598350525,
|
5730 |
+
"learning_rate": 1.3451825934541551e-05,
|
5731 |
+
"loss": 2.3732,
|
5732 |
+
"step": 813
|
5733 |
+
},
|
5734 |
+
{
|
5735 |
+
"epoch": 0.8359435173299101,
|
5736 |
+
"grad_norm": 1.204418659210205,
|
5737 |
+
"learning_rate": 1.3289033647520877e-05,
|
5738 |
+
"loss": 2.2253,
|
5739 |
+
"step": 814
|
5740 |
+
},
|
5741 |
+
{
|
5742 |
+
"epoch": 0.8369704749679076,
|
5743 |
+
"grad_norm": 1.26673424243927,
|
5744 |
+
"learning_rate": 1.3127162275017957e-05,
|
5745 |
+
"loss": 2.2487,
|
5746 |
+
"step": 815
|
5747 |
+
},
|
5748 |
+
{
|
5749 |
+
"epoch": 0.837997432605905,
|
5750 |
+
"grad_norm": 1.131012201309204,
|
5751 |
+
"learning_rate": 1.2966213536189032e-05,
|
5752 |
+
"loss": 1.9738,
|
5753 |
+
"step": 816
|
5754 |
+
},
|
5755 |
+
{
|
5756 |
+
"epoch": 0.8390243902439024,
|
5757 |
+
"grad_norm": 1.244232177734375,
|
5758 |
+
"learning_rate": 1.2806189140391489e-05,
|
5759 |
+
"loss": 2.1122,
|
5760 |
+
"step": 817
|
5761 |
+
},
|
5762 |
+
{
|
5763 |
+
"epoch": 0.8400513478818998,
|
5764 |
+
"grad_norm": 1.446234941482544,
|
5765 |
+
"learning_rate": 1.2647090787165694e-05,
|
5766 |
+
"loss": 2.3183,
|
5767 |
+
"step": 818
|
5768 |
+
},
|
5769 |
+
{
|
5770 |
+
"epoch": 0.8410783055198973,
|
5771 |
+
"grad_norm": 1.4866114854812622,
|
5772 |
+
"learning_rate": 1.2488920166217032e-05,
|
5773 |
+
"loss": 2.273,
|
5774 |
+
"step": 819
|
5775 |
+
},
|
5776 |
+
{
|
5777 |
+
"epoch": 0.8421052631578947,
|
5778 |
+
"grad_norm": 1.590286374092102,
|
5779 |
+
"learning_rate": 1.2331678957397819e-05,
|
5780 |
+
"loss": 2.2149,
|
5781 |
+
"step": 820
|
5782 |
+
},
|
5783 |
+
{
|
5784 |
+
"epoch": 0.8431322207958921,
|
5785 |
+
"grad_norm": 1.6288813352584839,
|
5786 |
+
"learning_rate": 1.2175368830689593e-05,
|
5787 |
+
"loss": 2.204,
|
5788 |
+
"step": 821
|
5789 |
+
},
|
5790 |
+
{
|
5791 |
+
"epoch": 0.8441591784338897,
|
5792 |
+
"grad_norm": 1.644627571105957,
|
5793 |
+
"learning_rate": 1.2019991446185309e-05,
|
5794 |
+
"loss": 2.3602,
|
5795 |
+
"step": 822
|
5796 |
+
},
|
5797 |
+
{
|
5798 |
+
"epoch": 0.8451861360718871,
|
5799 |
+
"grad_norm": 3.2406749725341797,
|
5800 |
+
"learning_rate": 1.186554845407174e-05,
|
5801 |
+
"loss": 2.2797,
|
5802 |
+
"step": 823
|
5803 |
+
},
|
5804 |
+
{
|
5805 |
+
"epoch": 0.8462130937098845,
|
5806 |
+
"grad_norm": 1.201799988746643,
|
5807 |
+
"learning_rate": 1.1712041494611958e-05,
|
5808 |
+
"loss": 2.2419,
|
5809 |
+
"step": 824
|
5810 |
+
},
|
5811 |
+
{
|
5812 |
+
"epoch": 0.8472400513478819,
|
5813 |
+
"grad_norm": 3.3227691650390625,
|
5814 |
+
"learning_rate": 1.1559472198127818e-05,
|
5815 |
+
"loss": 2.3199,
|
5816 |
+
"step": 825
|
5817 |
+
},
|
5818 |
+
{
|
5819 |
+
"epoch": 0.8482670089858794,
|
5820 |
+
"grad_norm": 1.7569915056228638,
|
5821 |
+
"learning_rate": 1.1407842184982786e-05,
|
5822 |
+
"loss": 2.2096,
|
5823 |
+
"step": 826
|
5824 |
+
},
|
5825 |
+
{
|
5826 |
+
"epoch": 0.8492939666238768,
|
5827 |
+
"grad_norm": 1.4387773275375366,
|
5828 |
+
"learning_rate": 1.125715306556464e-05,
|
5829 |
+
"loss": 2.2704,
|
5830 |
+
"step": 827
|
5831 |
+
},
|
5832 |
+
{
|
5833 |
+
"epoch": 0.8503209242618742,
|
5834 |
+
"grad_norm": 1.222449779510498,
|
5835 |
+
"learning_rate": 1.1107406440268376e-05,
|
5836 |
+
"loss": 2.1525,
|
5837 |
+
"step": 828
|
5838 |
+
},
|
5839 |
+
{
|
5840 |
+
"epoch": 0.8513478818998717,
|
5841 |
+
"grad_norm": 1.3529927730560303,
|
5842 |
+
"learning_rate": 1.0958603899479281e-05,
|
5843 |
+
"loss": 2.3474,
|
5844 |
+
"step": 829
|
5845 |
+
},
|
5846 |
+
{
|
5847 |
+
"epoch": 0.8523748395378691,
|
5848 |
+
"grad_norm": 1.2619078159332275,
|
5849 |
+
"learning_rate": 1.0810747023555878e-05,
|
5850 |
+
"loss": 2.1297,
|
5851 |
+
"step": 830
|
5852 |
+
},
|
5853 |
+
{
|
5854 |
+
"epoch": 0.8534017971758665,
|
5855 |
+
"grad_norm": 1.312843918800354,
|
5856 |
+
"learning_rate": 1.0663837382813336e-05,
|
5857 |
+
"loss": 2.1388,
|
5858 |
+
"step": 831
|
5859 |
+
},
|
5860 |
+
{
|
5861 |
+
"epoch": 0.8544287548138639,
|
5862 |
+
"grad_norm": 2.356374979019165,
|
5863 |
+
"learning_rate": 1.0517876537506687e-05,
|
5864 |
+
"loss": 2.2537,
|
5865 |
+
"step": 832
|
5866 |
+
},
|
5867 |
+
{
|
5868 |
+
"epoch": 0.8554557124518614,
|
5869 |
+
"grad_norm": 1.5887354612350464,
|
5870 |
+
"learning_rate": 1.0372866037814277e-05,
|
5871 |
+
"loss": 2.1951,
|
5872 |
+
"step": 833
|
5873 |
+
},
|
5874 |
+
{
|
5875 |
+
"epoch": 0.8564826700898588,
|
5876 |
+
"grad_norm": 1.3657351732254028,
|
5877 |
+
"learning_rate": 1.0228807423821263e-05,
|
5878 |
+
"loss": 2.3044,
|
5879 |
+
"step": 834
|
5880 |
+
},
|
5881 |
+
{
|
5882 |
+
"epoch": 0.8575096277278562,
|
5883 |
+
"grad_norm": 1.2927043437957764,
|
5884 |
+
"learning_rate": 1.0085702225503313e-05,
|
5885 |
+
"loss": 2.0492,
|
5886 |
+
"step": 835
|
5887 |
+
},
|
5888 |
+
{
|
5889 |
+
"epoch": 0.8585365853658536,
|
5890 |
+
"grad_norm": 1.8013806343078613,
|
5891 |
+
"learning_rate": 9.943551962710362e-06,
|
5892 |
+
"loss": 2.2268,
|
5893 |
+
"step": 836
|
5894 |
+
},
|
5895 |
+
{
|
5896 |
+
"epoch": 0.8595635430038511,
|
5897 |
+
"grad_norm": 1.4299966096878052,
|
5898 |
+
"learning_rate": 9.802358145150425e-06,
|
5899 |
+
"loss": 2.404,
|
5900 |
+
"step": 837
|
5901 |
+
},
|
5902 |
+
{
|
5903 |
+
"epoch": 0.8605905006418485,
|
5904 |
+
"grad_norm": 1.9017884731292725,
|
5905 |
+
"learning_rate": 9.662122272373575e-06,
|
5906 |
+
"loss": 2.2539,
|
5907 |
+
"step": 838
|
5908 |
+
},
|
5909 |
+
{
|
5910 |
+
"epoch": 0.8616174582798459,
|
5911 |
+
"grad_norm": 1.2827292680740356,
|
5912 |
+
"learning_rate": 9.522845833756e-06,
|
5913 |
+
"loss": 2.1102,
|
5914 |
+
"step": 839
|
5915 |
+
},
|
5916 |
+
{
|
5917 |
+
"epoch": 0.8626444159178434,
|
5918 |
+
"grad_norm": 1.567421555519104,
|
5919 |
+
"learning_rate": 9.384530308484273e-06,
|
5920 |
+
"loss": 2.4192,
|
5921 |
+
"step": 840
|
5922 |
+
},
|
5923 |
+
{
|
5924 |
+
"epoch": 0.8636713735558408,
|
5925 |
+
"grad_norm": 1.4824482202529907,
|
5926 |
+
"learning_rate": 9.247177165539556e-06,
|
5927 |
+
"loss": 2.3457,
|
5928 |
+
"step": 841
|
5929 |
+
},
|
5930 |
+
{
|
5931 |
+
"epoch": 0.8646983311938382,
|
5932 |
+
"grad_norm": 2.1519415378570557,
|
5933 |
+
"learning_rate": 9.110787863682002e-06,
|
5934 |
+
"loss": 2.2903,
|
5935 |
+
"step": 842
|
5936 |
+
},
|
5937 |
+
{
|
5938 |
+
"epoch": 0.8657252888318356,
|
5939 |
+
"grad_norm": 1.3354597091674805,
|
5940 |
+
"learning_rate": 8.97536385143527e-06,
|
5941 |
+
"loss": 2.2387,
|
5942 |
+
"step": 843
|
5943 |
+
},
|
5944 |
+
{
|
5945 |
+
"epoch": 0.8667522464698331,
|
5946 |
+
"grad_norm": 1.363411784172058,
|
5947 |
+
"learning_rate": 8.840906567071194e-06,
|
5948 |
+
"loss": 2.3414,
|
5949 |
+
"step": 844
|
5950 |
+
},
|
5951 |
+
{
|
5952 |
+
"epoch": 0.8677792041078306,
|
5953 |
+
"grad_norm": 1.2312959432601929,
|
5954 |
+
"learning_rate": 8.707417438594445e-06,
|
5955 |
+
"loss": 2.1462,
|
5956 |
+
"step": 845
|
5957 |
+
},
|
5958 |
+
{
|
5959 |
+
"epoch": 0.868806161745828,
|
5960 |
+
"grad_norm": 1.4682419300079346,
|
5961 |
+
"learning_rate": 8.574897883727384e-06,
|
5962 |
+
"loss": 2.3364,
|
5963 |
+
"step": 846
|
5964 |
+
},
|
5965 |
+
{
|
5966 |
+
"epoch": 0.8698331193838255,
|
5967 |
+
"grad_norm": 1.9188342094421387,
|
5968 |
+
"learning_rate": 8.443349309895032e-06,
|
5969 |
+
"loss": 2.1837,
|
5970 |
+
"step": 847
|
5971 |
+
},
|
5972 |
+
{
|
5973 |
+
"epoch": 0.8708600770218229,
|
5974 |
+
"grad_norm": 1.4827263355255127,
|
5975 |
+
"learning_rate": 8.312773114210049e-06,
|
5976 |
+
"loss": 2.2226,
|
5977 |
+
"step": 848
|
5978 |
+
},
|
5979 |
+
{
|
5980 |
+
"epoch": 0.8718870346598203,
|
5981 |
+
"grad_norm": 1.3923516273498535,
|
5982 |
+
"learning_rate": 8.183170683457986e-06,
|
5983 |
+
"loss": 2.4293,
|
5984 |
+
"step": 849
|
5985 |
+
},
|
5986 |
+
{
|
5987 |
+
"epoch": 0.8729139922978177,
|
5988 |
+
"grad_norm": 1.3553410768508911,
|
5989 |
+
"learning_rate": 8.054543394082504e-06,
|
5990 |
+
"loss": 2.0766,
|
5991 |
+
"step": 850
|
5992 |
+
},
|
5993 |
+
{
|
5994 |
+
"epoch": 0.8739409499358152,
|
5995 |
+
"grad_norm": 3.3262252807617188,
|
5996 |
+
"learning_rate": 7.926892612170777e-06,
|
5997 |
+
"loss": 2.2739,
|
5998 |
+
"step": 851
|
5999 |
+
},
|
6000 |
+
{
|
6001 |
+
"epoch": 0.8749679075738126,
|
6002 |
+
"grad_norm": 1.1720492839813232,
|
6003 |
+
"learning_rate": 7.800219693438981e-06,
|
6004 |
+
"loss": 2.1775,
|
6005 |
+
"step": 852
|
6006 |
+
},
|
6007 |
+
{
|
6008 |
+
"epoch": 0.87599486521181,
|
6009 |
+
"grad_norm": 1.3928182125091553,
|
6010 |
+
"learning_rate": 7.674525983217828e-06,
|
6011 |
+
"loss": 2.1329,
|
6012 |
+
"step": 853
|
6013 |
+
},
|
6014 |
+
{
|
6015 |
+
"epoch": 0.8770218228498075,
|
6016 |
+
"grad_norm": 2.7582273483276367,
|
6017 |
+
"learning_rate": 7.5498128164383955e-06,
|
6018 |
+
"loss": 2.2974,
|
6019 |
+
"step": 854
|
6020 |
+
},
|
6021 |
+
{
|
6022 |
+
"epoch": 0.8780487804878049,
|
6023 |
+
"grad_norm": 1.4647380113601685,
|
6024 |
+
"learning_rate": 7.426081517617889e-06,
|
6025 |
+
"loss": 2.532,
|
6026 |
+
"step": 855
|
6027 |
+
},
|
6028 |
+
{
|
6029 |
+
"epoch": 0.8790757381258023,
|
6030 |
+
"grad_norm": 1.2066142559051514,
|
6031 |
+
"learning_rate": 7.30333340084558e-06,
|
6032 |
+
"loss": 2.1876,
|
6033 |
+
"step": 856
|
6034 |
+
},
|
6035 |
+
{
|
6036 |
+
"epoch": 0.8801026957637997,
|
6037 |
+
"grad_norm": 1.2079867124557495,
|
6038 |
+
"learning_rate": 7.181569769768792e-06,
|
6039 |
+
"loss": 2.201,
|
6040 |
+
"step": 857
|
6041 |
+
},
|
6042 |
+
{
|
6043 |
+
"epoch": 0.8811296534017972,
|
6044 |
+
"grad_norm": 1.4736223220825195,
|
6045 |
+
"learning_rate": 7.0607919175791796e-06,
|
6046 |
+
"loss": 2.2242,
|
6047 |
+
"step": 858
|
6048 |
+
},
|
6049 |
+
{
|
6050 |
+
"epoch": 0.8821566110397946,
|
6051 |
+
"grad_norm": 1.428174376487732,
|
6052 |
+
"learning_rate": 6.941001126998892e-06,
|
6053 |
+
"loss": 2.3792,
|
6054 |
+
"step": 859
|
6055 |
+
},
|
6056 |
+
{
|
6057 |
+
"epoch": 0.883183568677792,
|
6058 |
+
"grad_norm": 1.695448875427246,
|
6059 |
+
"learning_rate": 6.822198670266988e-06,
|
6060 |
+
"loss": 2.3633,
|
6061 |
+
"step": 860
|
6062 |
+
},
|
6063 |
+
{
|
6064 |
+
"epoch": 0.8842105263157894,
|
6065 |
+
"grad_norm": 1.5708576440811157,
|
6066 |
+
"learning_rate": 6.7043858091259235e-06,
|
6067 |
+
"loss": 2.297,
|
6068 |
+
"step": 861
|
6069 |
+
},
|
6070 |
+
{
|
6071 |
+
"epoch": 0.8852374839537869,
|
6072 |
+
"grad_norm": 1.4929567575454712,
|
6073 |
+
"learning_rate": 6.587563794808127e-06,
|
6074 |
+
"loss": 2.3336,
|
6075 |
+
"step": 862
|
6076 |
+
},
|
6077 |
+
{
|
6078 |
+
"epoch": 0.8862644415917843,
|
6079 |
+
"grad_norm": 1.305184245109558,
|
6080 |
+
"learning_rate": 6.471733868022744e-06,
|
6081 |
+
"loss": 2.4024,
|
6082 |
+
"step": 863
|
6083 |
+
},
|
6084 |
+
{
|
6085 |
+
"epoch": 0.8872913992297817,
|
6086 |
+
"grad_norm": 1.4627037048339844,
|
6087 |
+
"learning_rate": 6.356897258942451e-06,
|
6088 |
+
"loss": 2.1668,
|
6089 |
+
"step": 864
|
6090 |
+
},
|
6091 |
+
{
|
6092 |
+
"epoch": 0.8883183568677792,
|
6093 |
+
"grad_norm": 1.36262845993042,
|
6094 |
+
"learning_rate": 6.243055187190383e-06,
|
6095 |
+
"loss": 2.0239,
|
6096 |
+
"step": 865
|
6097 |
+
},
|
6098 |
+
{
|
6099 |
+
"epoch": 0.8893453145057766,
|
6100 |
+
"grad_norm": 1.3705987930297852,
|
6101 |
+
"learning_rate": 6.130208861827202e-06,
|
6102 |
+
"loss": 2.1388,
|
6103 |
+
"step": 866
|
6104 |
+
},
|
6105 |
+
{
|
6106 |
+
"epoch": 0.8903722721437741,
|
6107 |
+
"grad_norm": 1.4911210536956787,
|
6108 |
+
"learning_rate": 6.018359481338176e-06,
|
6109 |
+
"loss": 2.3255,
|
6110 |
+
"step": 867
|
6111 |
+
},
|
6112 |
+
{
|
6113 |
+
"epoch": 0.8913992297817716,
|
6114 |
+
"grad_norm": 1.2989243268966675,
|
6115 |
+
"learning_rate": 5.907508233620573e-06,
|
6116 |
+
"loss": 2.2913,
|
6117 |
+
"step": 868
|
6118 |
+
},
|
6119 |
+
{
|
6120 |
+
"epoch": 0.892426187419769,
|
6121 |
+
"grad_norm": 1.4845342636108398,
|
6122 |
+
"learning_rate": 5.797656295970955e-06,
|
6123 |
+
"loss": 2.3531,
|
6124 |
+
"step": 869
|
6125 |
+
},
|
6126 |
+
{
|
6127 |
+
"epoch": 0.8934531450577664,
|
6128 |
+
"grad_norm": 1.3806096315383911,
|
6129 |
+
"learning_rate": 5.688804835072748e-06,
|
6130 |
+
"loss": 2.1876,
|
6131 |
+
"step": 870
|
6132 |
+
},
|
6133 |
+
{
|
6134 |
+
"epoch": 0.8944801026957638,
|
6135 |
+
"grad_norm": 1.2086135149002075,
|
6136 |
+
"learning_rate": 5.580955006983735e-06,
|
6137 |
+
"loss": 2.2474,
|
6138 |
+
"step": 871
|
6139 |
+
},
|
6140 |
+
{
|
6141 |
+
"epoch": 0.8955070603337613,
|
6142 |
+
"grad_norm": 1.488832712173462,
|
6143 |
+
"learning_rate": 5.474107957123886e-06,
|
6144 |
+
"loss": 2.2586,
|
6145 |
+
"step": 872
|
6146 |
+
},
|
6147 |
+
{
|
6148 |
+
"epoch": 0.8965340179717587,
|
6149 |
+
"grad_norm": 2.129424571990967,
|
6150 |
+
"learning_rate": 5.3682648202631695e-06,
|
6151 |
+
"loss": 2.2862,
|
6152 |
+
"step": 873
|
6153 |
+
},
|
6154 |
+
{
|
6155 |
+
"epoch": 0.8975609756097561,
|
6156 |
+
"grad_norm": 1.4406683444976807,
|
6157 |
+
"learning_rate": 5.263426720509468e-06,
|
6158 |
+
"loss": 1.9062,
|
6159 |
+
"step": 874
|
6160 |
+
},
|
6161 |
+
{
|
6162 |
+
"epoch": 0.8985879332477535,
|
6163 |
+
"grad_norm": 1.2508238554000854,
|
6164 |
+
"learning_rate": 5.159594771296683e-06,
|
6165 |
+
"loss": 2.0821,
|
6166 |
+
"step": 875
|
6167 |
+
},
|
6168 |
+
{
|
6169 |
+
"epoch": 0.899614890885751,
|
6170 |
+
"grad_norm": 1.6048517227172852,
|
6171 |
+
"learning_rate": 5.056770075372841e-06,
|
6172 |
+
"loss": 2.0023,
|
6173 |
+
"step": 876
|
6174 |
+
},
|
6175 |
+
{
|
6176 |
+
"epoch": 0.9006418485237484,
|
6177 |
+
"grad_norm": 1.4277167320251465,
|
6178 |
+
"learning_rate": 4.954953724788469e-06,
|
6179 |
+
"loss": 2.3978,
|
6180 |
+
"step": 877
|
6181 |
+
},
|
6182 |
+
{
|
6183 |
+
"epoch": 0.9016688061617458,
|
6184 |
+
"grad_norm": 1.236256718635559,
|
6185 |
+
"learning_rate": 4.8541468008849285e-06,
|
6186 |
+
"loss": 2.0407,
|
6187 |
+
"step": 878
|
6188 |
+
},
|
6189 |
+
{
|
6190 |
+
"epoch": 0.9026957637997433,
|
6191 |
+
"grad_norm": 1.405112624168396,
|
6192 |
+
"learning_rate": 4.754350374283001e-06,
|
6193 |
+
"loss": 2.2371,
|
6194 |
+
"step": 879
|
6195 |
+
},
|
6196 |
+
{
|
6197 |
+
"epoch": 0.9037227214377407,
|
6198 |
+
"grad_norm": 1.4664572477340698,
|
6199 |
+
"learning_rate": 4.6555655048713955e-06,
|
6200 |
+
"loss": 2.4453,
|
6201 |
+
"step": 880
|
6202 |
+
},
|
6203 |
+
{
|
6204 |
+
"epoch": 0.9047496790757381,
|
6205 |
+
"grad_norm": 1.3141064643859863,
|
6206 |
+
"learning_rate": 4.5577932417956495e-06,
|
6207 |
+
"loss": 2.1578,
|
6208 |
+
"step": 881
|
6209 |
+
},
|
6210 |
+
{
|
6211 |
+
"epoch": 0.9057766367137355,
|
6212 |
+
"grad_norm": 1.4012268781661987,
|
6213 |
+
"learning_rate": 4.461034623446847e-06,
|
6214 |
+
"loss": 2.4731,
|
6215 |
+
"step": 882
|
6216 |
+
},
|
6217 |
+
{
|
6218 |
+
"epoch": 0.906803594351733,
|
6219 |
+
"grad_norm": 1.1742385625839233,
|
6220 |
+
"learning_rate": 4.3652906774506955e-06,
|
6221 |
+
"loss": 2.1617,
|
6222 |
+
"step": 883
|
6223 |
+
},
|
6224 |
+
{
|
6225 |
+
"epoch": 0.9078305519897304,
|
6226 |
+
"grad_norm": 1.2050607204437256,
|
6227 |
+
"learning_rate": 4.270562420656543e-06,
|
6228 |
+
"loss": 2.2211,
|
6229 |
+
"step": 884
|
6230 |
+
},
|
6231 |
+
{
|
6232 |
+
"epoch": 0.9088575096277278,
|
6233 |
+
"grad_norm": 1.6956837177276611,
|
6234 |
+
"learning_rate": 4.176850859126591e-06,
|
6235 |
+
"loss": 2.3175,
|
6236 |
+
"step": 885
|
6237 |
+
},
|
6238 |
+
{
|
6239 |
+
"epoch": 0.9098844672657252,
|
6240 |
+
"grad_norm": 1.1956061124801636,
|
6241 |
+
"learning_rate": 4.084156988125231e-06,
|
6242 |
+
"loss": 2.2572,
|
6243 |
+
"step": 886
|
6244 |
+
},
|
6245 |
+
{
|
6246 |
+
"epoch": 0.9109114249037227,
|
6247 |
+
"grad_norm": 1.2421371936798096,
|
6248 |
+
"learning_rate": 3.992481792108493e-06,
|
6249 |
+
"loss": 2.0939,
|
6250 |
+
"step": 887
|
6251 |
+
},
|
6252 |
+
{
|
6253 |
+
"epoch": 0.9119383825417201,
|
6254 |
+
"grad_norm": 1.1449053287506104,
|
6255 |
+
"learning_rate": 3.901826244713525e-06,
|
6256 |
+
"loss": 2.0271,
|
6257 |
+
"step": 888
|
6258 |
+
},
|
6259 |
+
{
|
6260 |
+
"epoch": 0.9129653401797175,
|
6261 |
+
"grad_norm": 1.5248461961746216,
|
6262 |
+
"learning_rate": 3.812191308748303e-06,
|
6263 |
+
"loss": 2.3376,
|
6264 |
+
"step": 889
|
6265 |
+
},
|
6266 |
+
{
|
6267 |
+
"epoch": 0.9139922978177151,
|
6268 |
+
"grad_norm": 1.607293725013733,
|
6269 |
+
"learning_rate": 3.723577936181366e-06,
|
6270 |
+
"loss": 2.3735,
|
6271 |
+
"step": 890
|
6272 |
+
},
|
6273 |
+
{
|
6274 |
+
"epoch": 0.9150192554557125,
|
6275 |
+
"grad_norm": 1.536257266998291,
|
6276 |
+
"learning_rate": 3.6359870681317743e-06,
|
6277 |
+
"loss": 2.2889,
|
6278 |
+
"step": 891
|
6279 |
+
},
|
6280 |
+
{
|
6281 |
+
"epoch": 0.9160462130937099,
|
6282 |
+
"grad_norm": 1.1060168743133545,
|
6283 |
+
"learning_rate": 3.5494196348590415e-06,
|
6284 |
+
"loss": 1.8348,
|
6285 |
+
"step": 892
|
6286 |
+
},
|
6287 |
+
{
|
6288 |
+
"epoch": 0.9170731707317074,
|
6289 |
+
"grad_norm": 1.3008726835250854,
|
6290 |
+
"learning_rate": 3.4638765557532983e-06,
|
6291 |
+
"loss": 2.2412,
|
6292 |
+
"step": 893
|
6293 |
+
},
|
6294 |
+
{
|
6295 |
+
"epoch": 0.9181001283697048,
|
6296 |
+
"grad_norm": 1.4789313077926636,
|
6297 |
+
"learning_rate": 3.3793587393255e-06,
|
6298 |
+
"loss": 2.1869,
|
6299 |
+
"step": 894
|
6300 |
+
},
|
6301 |
+
{
|
6302 |
+
"epoch": 0.9191270860077022,
|
6303 |
+
"grad_norm": 1.9860131740570068,
|
6304 |
+
"learning_rate": 3.295867083197801e-06,
|
6305 |
+
"loss": 2.3982,
|
6306 |
+
"step": 895
|
6307 |
+
},
|
6308 |
+
{
|
6309 |
+
"epoch": 0.9201540436456996,
|
6310 |
+
"grad_norm": 1.4276963472366333,
|
6311 |
+
"learning_rate": 3.213402474093996e-06,
|
6312 |
+
"loss": 2.4385,
|
6313 |
+
"step": 896
|
6314 |
+
},
|
6315 |
+
{
|
6316 |
+
"epoch": 0.9211810012836971,
|
6317 |
+
"grad_norm": 1.2232952117919922,
|
6318 |
+
"learning_rate": 3.131965787830149e-06,
|
6319 |
+
"loss": 2.1511,
|
6320 |
+
"step": 897
|
6321 |
+
},
|
6322 |
+
{
|
6323 |
+
"epoch": 0.9222079589216945,
|
6324 |
+
"grad_norm": 1.288060188293457,
|
6325 |
+
"learning_rate": 3.0515578893052344e-06,
|
6326 |
+
"loss": 2.2703,
|
6327 |
+
"step": 898
|
6328 |
+
},
|
6329 |
+
{
|
6330 |
+
"epoch": 0.9232349165596919,
|
6331 |
+
"grad_norm": 2.902099847793579,
|
6332 |
+
"learning_rate": 2.9721796324919893e-06,
|
6333 |
+
"loss": 2.4419,
|
6334 |
+
"step": 899
|
6335 |
+
},
|
6336 |
+
{
|
6337 |
+
"epoch": 0.9242618741976893,
|
6338 |
+
"grad_norm": 1.4718948602676392,
|
6339 |
+
"learning_rate": 2.8938318604278314e-06,
|
6340 |
+
"loss": 2.2219,
|
6341 |
+
"step": 900
|
6342 |
+
},
|
6343 |
+
{
|
6344 |
+
"epoch": 0.9252888318356868,
|
6345 |
+
"grad_norm": 1.2647035121917725,
|
6346 |
+
"learning_rate": 2.8165154052058997e-06,
|
6347 |
+
"loss": 1.9561,
|
6348 |
+
"step": 901
|
6349 |
+
},
|
6350 |
+
{
|
6351 |
+
"epoch": 0.9263157894736842,
|
6352 |
+
"grad_norm": 1.5724482536315918,
|
6353 |
+
"learning_rate": 2.7402310879662497e-06,
|
6354 |
+
"loss": 2.2152,
|
6355 |
+
"step": 902
|
6356 |
+
},
|
6357 |
+
{
|
6358 |
+
"epoch": 0.9273427471116816,
|
6359 |
+
"grad_norm": 1.4992693662643433,
|
6360 |
+
"learning_rate": 2.664979718887073e-06,
|
6361 |
+
"loss": 2.3404,
|
6362 |
+
"step": 903
|
6363 |
+
},
|
6364 |
+
{
|
6365 |
+
"epoch": 0.9283697047496791,
|
6366 |
+
"grad_norm": 1.3174737691879272,
|
6367 |
+
"learning_rate": 2.590762097176136e-06,
|
6368 |
+
"loss": 2.1052,
|
6369 |
+
"step": 904
|
6370 |
+
},
|
6371 |
+
{
|
6372 |
+
"epoch": 0.9293966623876765,
|
6373 |
+
"grad_norm": 1.3780856132507324,
|
6374 |
+
"learning_rate": 2.517579011062299e-06,
|
6375 |
+
"loss": 2.1754,
|
6376 |
+
"step": 905
|
6377 |
+
},
|
6378 |
+
{
|
6379 |
+
"epoch": 0.9304236200256739,
|
6380 |
+
"grad_norm": 1.4942290782928467,
|
6381 |
+
"learning_rate": 2.44543123778711e-06,
|
6382 |
+
"loss": 2.2278,
|
6383 |
+
"step": 906
|
6384 |
+
},
|
6385 |
+
{
|
6386 |
+
"epoch": 0.9314505776636713,
|
6387 |
+
"grad_norm": 1.3182977437973022,
|
6388 |
+
"learning_rate": 2.3743195435966036e-06,
|
6389 |
+
"loss": 2.416,
|
6390 |
+
"step": 907
|
6391 |
+
},
|
6392 |
+
{
|
6393 |
+
"epoch": 0.9324775353016688,
|
6394 |
+
"grad_norm": 1.462436318397522,
|
6395 |
+
"learning_rate": 2.304244683733059e-06,
|
6396 |
+
"loss": 2.4912,
|
6397 |
+
"step": 908
|
6398 |
+
},
|
6399 |
+
{
|
6400 |
+
"epoch": 0.9335044929396662,
|
6401 |
+
"grad_norm": 1.334877610206604,
|
6402 |
+
"learning_rate": 2.2352074024271195e-06,
|
6403 |
+
"loss": 2.1424,
|
6404 |
+
"step": 909
|
6405 |
+
},
|
6406 |
+
{
|
6407 |
+
"epoch": 0.9345314505776636,
|
6408 |
+
"grad_norm": 1.6479849815368652,
|
6409 |
+
"learning_rate": 2.167208432889789e-06,
|
6410 |
+
"loss": 2.1548,
|
6411 |
+
"step": 910
|
6412 |
+
},
|
6413 |
+
{
|
6414 |
+
"epoch": 0.935558408215661,
|
6415 |
+
"grad_norm": 1.6695164442062378,
|
6416 |
+
"learning_rate": 2.1002484973046577e-06,
|
6417 |
+
"loss": 2.2078,
|
6418 |
+
"step": 911
|
6419 |
+
},
|
6420 |
+
{
|
6421 |
+
"epoch": 0.9365853658536586,
|
6422 |
+
"grad_norm": 1.5034846067428589,
|
6423 |
+
"learning_rate": 2.034328306820288e-06,
|
6424 |
+
"loss": 2.3183,
|
6425 |
+
"step": 912
|
6426 |
+
},
|
6427 |
+
{
|
6428 |
+
"epoch": 0.937612323491656,
|
6429 |
+
"grad_norm": 1.3325846195220947,
|
6430 |
+
"learning_rate": 1.969448561542553e-06,
|
6431 |
+
"loss": 2.1897,
|
6432 |
+
"step": 913
|
6433 |
+
},
|
6434 |
+
{
|
6435 |
+
"epoch": 0.9386392811296534,
|
6436 |
+
"grad_norm": 1.5285733938217163,
|
6437 |
+
"learning_rate": 1.9056099505273427e-06,
|
6438 |
+
"loss": 2.4954,
|
6439 |
+
"step": 914
|
6440 |
+
},
|
6441 |
+
{
|
6442 |
+
"epoch": 0.9396662387676509,
|
6443 |
+
"grad_norm": 1.4107942581176758,
|
6444 |
+
"learning_rate": 1.8428131517731373e-06,
|
6445 |
+
"loss": 2.4118,
|
6446 |
+
"step": 915
|
6447 |
+
},
|
6448 |
+
{
|
6449 |
+
"epoch": 0.9406931964056483,
|
6450 |
+
"grad_norm": 1.5880571603775024,
|
6451 |
+
"learning_rate": 1.7810588322138222e-06,
|
6452 |
+
"loss": 2.1729,
|
6453 |
+
"step": 916
|
6454 |
+
},
|
6455 |
+
{
|
6456 |
+
"epoch": 0.9417201540436457,
|
6457 |
+
"grad_norm": 1.3712589740753174,
|
6458 |
+
"learning_rate": 1.7203476477116843e-06,
|
6459 |
+
"loss": 2.0461,
|
6460 |
+
"step": 917
|
6461 |
+
},
|
6462 |
+
{
|
6463 |
+
"epoch": 0.9427471116816432,
|
6464 |
+
"grad_norm": 1.2501283884048462,
|
6465 |
+
"learning_rate": 1.6606802430503166e-06,
|
6466 |
+
"loss": 2.2321,
|
6467 |
+
"step": 918
|
6468 |
+
},
|
6469 |
+
{
|
6470 |
+
"epoch": 0.9437740693196406,
|
6471 |
+
"grad_norm": 1.9402281045913696,
|
6472 |
+
"learning_rate": 1.6020572519278908e-06,
|
6473 |
+
"loss": 2.2923,
|
6474 |
+
"step": 919
|
6475 |
+
},
|
6476 |
+
{
|
6477 |
+
"epoch": 0.944801026957638,
|
6478 |
+
"grad_norm": 1.619081735610962,
|
6479 |
+
"learning_rate": 1.5444792969503407e-06,
|
6480 |
+
"loss": 2.4231,
|
6481 |
+
"step": 920
|
6482 |
+
},
|
6483 |
+
{
|
6484 |
+
"epoch": 0.9458279845956354,
|
6485 |
+
"grad_norm": 1.4488154649734497,
|
6486 |
+
"learning_rate": 1.487946989624811e-06,
|
6487 |
+
"loss": 2.3659,
|
6488 |
+
"step": 921
|
6489 |
+
},
|
6490 |
+
{
|
6491 |
+
"epoch": 0.9468549422336329,
|
6492 |
+
"grad_norm": 1.4263108968734741,
|
6493 |
+
"learning_rate": 1.43246093035313e-06,
|
6494 |
+
"loss": 2.2714,
|
6495 |
+
"step": 922
|
6496 |
+
},
|
6497 |
+
{
|
6498 |
+
"epoch": 0.9478818998716303,
|
6499 |
+
"grad_norm": 1.1710644960403442,
|
6500 |
+
"learning_rate": 1.3780217084254366e-06,
|
6501 |
+
"loss": 2.128,
|
6502 |
+
"step": 923
|
6503 |
+
},
|
6504 |
+
{
|
6505 |
+
"epoch": 0.9489088575096277,
|
6506 |
+
"grad_norm": 1.3074594736099243,
|
6507 |
+
"learning_rate": 1.3246299020139185e-06,
|
6508 |
+
"loss": 2.0772,
|
6509 |
+
"step": 924
|
6510 |
+
},
|
6511 |
+
{
|
6512 |
+
"epoch": 0.9499358151476252,
|
6513 |
+
"grad_norm": 1.3507771492004395,
|
6514 |
+
"learning_rate": 1.2722860781666956e-06,
|
6515 |
+
"loss": 2.2176,
|
6516 |
+
"step": 925
|
6517 |
+
},
|
6518 |
+
{
|
6519 |
+
"epoch": 0.9509627727856226,
|
6520 |
+
"grad_norm": 1.6228179931640625,
|
6521 |
+
"learning_rate": 1.2209907928017795e-06,
|
6522 |
+
"loss": 2.3079,
|
6523 |
+
"step": 926
|
6524 |
+
},
|
6525 |
+
{
|
6526 |
+
"epoch": 0.95198973042362,
|
6527 |
+
"grad_norm": 1.375373125076294,
|
6528 |
+
"learning_rate": 1.1707445907011339e-06,
|
6529 |
+
"loss": 2.1246,
|
6530 |
+
"step": 927
|
6531 |
+
},
|
6532 |
+
{
|
6533 |
+
"epoch": 0.9530166880616174,
|
6534 |
+
"grad_norm": 1.8349575996398926,
|
6535 |
+
"learning_rate": 1.1215480055049798e-06,
|
6536 |
+
"loss": 2.2632,
|
6537 |
+
"step": 928
|
6538 |
+
},
|
6539 |
+
{
|
6540 |
+
"epoch": 0.9540436456996149,
|
6541 |
+
"grad_norm": 1.3294469118118286,
|
6542 |
+
"learning_rate": 1.073401559706022e-06,
|
6543 |
+
"loss": 2.1936,
|
6544 |
+
"step": 929
|
6545 |
+
},
|
6546 |
+
{
|
6547 |
+
"epoch": 0.9550706033376123,
|
6548 |
+
"grad_norm": 1.4897345304489136,
|
6549 |
+
"learning_rate": 1.0263057646440199e-06,
|
6550 |
+
"loss": 2.2939,
|
6551 |
+
"step": 930
|
6552 |
+
},
|
6553 |
+
{
|
6554 |
+
"epoch": 0.9560975609756097,
|
6555 |
+
"grad_norm": 1.434954047203064,
|
6556 |
+
"learning_rate": 9.802611205002032e-07,
|
6557 |
+
"loss": 2.1473,
|
6558 |
+
"step": 931
|
6559 |
+
},
|
6560 |
+
{
|
6561 |
+
"epoch": 0.9571245186136071,
|
6562 |
+
"grad_norm": 1.6770613193511963,
|
6563 |
+
"learning_rate": 9.352681162920984e-07,
|
6564 |
+
"loss": 2.2041,
|
6565 |
+
"step": 932
|
6566 |
+
},
|
6567 |
+
{
|
6568 |
+
"epoch": 0.9581514762516046,
|
6569 |
+
"grad_norm": 1.771236538887024,
|
6570 |
+
"learning_rate": 8.913272298682773e-07,
|
6571 |
+
"loss": 2.1005,
|
6572 |
+
"step": 933
|
6573 |
+
},
|
6574 |
+
{
|
6575 |
+
"epoch": 0.959178433889602,
|
6576 |
+
"grad_norm": 1.6060534715652466,
|
6577 |
+
"learning_rate": 8.484389279032834e-07,
|
6578 |
+
"loss": 2.3045,
|
6579 |
+
"step": 934
|
6580 |
+
},
|
6581 |
+
{
|
6582 |
+
"epoch": 0.9602053915275995,
|
6583 |
+
"grad_norm": 1.5178483724594116,
|
6584 |
+
"learning_rate": 8.066036658926579e-07,
|
6585 |
+
"loss": 2.2922,
|
6586 |
+
"step": 935
|
6587 |
+
},
|
6588 |
+
{
|
6589 |
+
"epoch": 0.961232349165597,
|
6590 |
+
"grad_norm": 1.6939451694488525,
|
6591 |
+
"learning_rate": 7.658218881481439e-07,
|
6592 |
+
"loss": 2.2083,
|
6593 |
+
"step": 936
|
6594 |
+
},
|
6595 |
+
{
|
6596 |
+
"epoch": 0.9622593068035944,
|
6597 |
+
"grad_norm": 1.2474678754806519,
|
6598 |
+
"learning_rate": 7.260940277929451e-07,
|
6599 |
+
"loss": 2.1818,
|
6600 |
+
"step": 937
|
6601 |
+
},
|
6602 |
+
{
|
6603 |
+
"epoch": 0.9632862644415918,
|
6604 |
+
"grad_norm": 1.459787368774414,
|
6605 |
+
"learning_rate": 6.874205067571083e-07,
|
6606 |
+
"loss": 2.3195,
|
6607 |
+
"step": 938
|
6608 |
+
},
|
6609 |
+
{
|
6610 |
+
"epoch": 0.9643132220795892,
|
6611 |
+
"grad_norm": 1.278882384300232,
|
6612 |
+
"learning_rate": 6.498017357731034e-07,
|
6613 |
+
"loss": 2.1238,
|
6614 |
+
"step": 939
|
6615 |
+
},
|
6616 |
+
{
|
6617 |
+
"epoch": 0.9653401797175867,
|
6618 |
+
"grad_norm": 1.3337883949279785,
|
6619 |
+
"learning_rate": 6.132381143713728e-07,
|
6620 |
+
"loss": 2.1721,
|
6621 |
+
"step": 940
|
6622 |
+
},
|
6623 |
+
{
|
6624 |
+
"epoch": 0.9663671373555841,
|
6625 |
+
"grad_norm": 1.6333835124969482,
|
6626 |
+
"learning_rate": 5.777300308761446e-07,
|
6627 |
+
"loss": 2.2194,
|
6628 |
+
"step": 941
|
6629 |
+
},
|
6630 |
+
{
|
6631 |
+
"epoch": 0.9673940949935815,
|
6632 |
+
"grad_norm": 1.384678602218628,
|
6633 |
+
"learning_rate": 5.432778624013257e-07,
|
6634 |
+
"loss": 2.4414,
|
6635 |
+
"step": 942
|
6636 |
+
},
|
6637 |
+
{
|
6638 |
+
"epoch": 0.968421052631579,
|
6639 |
+
"grad_norm": 1.3769394159317017,
|
6640 |
+
"learning_rate": 5.098819748464378e-07,
|
6641 |
+
"loss": 2.3543,
|
6642 |
+
"step": 943
|
6643 |
+
},
|
6644 |
+
{
|
6645 |
+
"epoch": 0.9694480102695764,
|
6646 |
+
"grad_norm": 1.353691577911377,
|
6647 |
+
"learning_rate": 4.775427228927765e-07,
|
6648 |
+
"loss": 2.1787,
|
6649 |
+
"step": 944
|
6650 |
+
},
|
6651 |
+
{
|
6652 |
+
"epoch": 0.9704749679075738,
|
6653 |
+
"grad_norm": 1.647281289100647,
|
6654 |
+
"learning_rate": 4.462604499996248e-07,
|
6655 |
+
"loss": 2.2027,
|
6656 |
+
"step": 945
|
6657 |
+
},
|
6658 |
+
{
|
6659 |
+
"epoch": 0.9715019255455712,
|
6660 |
+
"grad_norm": 1.2839142084121704,
|
6661 |
+
"learning_rate": 4.1603548840062345e-07,
|
6662 |
+
"loss": 2.1197,
|
6663 |
+
"step": 946
|
6664 |
+
},
|
6665 |
+
{
|
6666 |
+
"epoch": 0.9725288831835687,
|
6667 |
+
"grad_norm": 1.1656153202056885,
|
6668 |
+
"learning_rate": 3.8686815910021767e-07,
|
6669 |
+
"loss": 2.1826,
|
6670 |
+
"step": 947
|
6671 |
+
},
|
6672 |
+
{
|
6673 |
+
"epoch": 0.9735558408215661,
|
6674 |
+
"grad_norm": 1.508293628692627,
|
6675 |
+
"learning_rate": 3.5875877187024896e-07,
|
6676 |
+
"loss": 2.4654,
|
6677 |
+
"step": 948
|
6678 |
+
},
|
6679 |
+
{
|
6680 |
+
"epoch": 0.9745827984595635,
|
6681 |
+
"grad_norm": 1.6085854768753052,
|
6682 |
+
"learning_rate": 3.317076252467133e-07,
|
6683 |
+
"loss": 2.2999,
|
6684 |
+
"step": 949
|
6685 |
+
},
|
6686 |
+
{
|
6687 |
+
"epoch": 0.975609756097561,
|
6688 |
+
"grad_norm": 1.520034670829773,
|
6689 |
+
"learning_rate": 3.0571500652651907e-07,
|
6690 |
+
"loss": 1.7566,
|
6691 |
+
"step": 950
|
6692 |
+
},
|
6693 |
+
{
|
6694 |
+
"epoch": 0.9766367137355584,
|
6695 |
+
"grad_norm": 1.2050639390945435,
|
6696 |
+
"learning_rate": 2.807811917644898e-07,
|
6697 |
+
"loss": 2.2555,
|
6698 |
+
"step": 951
|
6699 |
+
},
|
6700 |
+
{
|
6701 |
+
"epoch": 0.9776636713735558,
|
6702 |
+
"grad_norm": 1.7655271291732788,
|
6703 |
+
"learning_rate": 2.5690644577039956e-07,
|
6704 |
+
"loss": 2.2677,
|
6705 |
+
"step": 952
|
6706 |
+
},
|
6707 |
+
{
|
6708 |
+
"epoch": 0.9786906290115532,
|
6709 |
+
"grad_norm": 1.795967698097229,
|
6710 |
+
"learning_rate": 2.340910221061754e-07,
|
6711 |
+
"loss": 2.399,
|
6712 |
+
"step": 953
|
6713 |
+
},
|
6714 |
+
{
|
6715 |
+
"epoch": 0.9797175866495507,
|
6716 |
+
"grad_norm": 1.9675099849700928,
|
6717 |
+
"learning_rate": 2.1233516308323264e-07,
|
6718 |
+
"loss": 2.2401,
|
6719 |
+
"step": 954
|
6720 |
+
},
|
6721 |
+
{
|
6722 |
+
"epoch": 0.9807445442875481,
|
6723 |
+
"grad_norm": 1.5977956056594849,
|
6724 |
+
"learning_rate": 1.9163909975982164e-07,
|
6725 |
+
"loss": 1.8277,
|
6726 |
+
"step": 955
|
6727 |
+
},
|
6728 |
+
{
|
6729 |
+
"epoch": 0.9817715019255455,
|
6730 |
+
"grad_norm": 1.4226264953613281,
|
6731 |
+
"learning_rate": 1.7200305193866284e-07,
|
6732 |
+
"loss": 2.127,
|
6733 |
+
"step": 956
|
6734 |
+
},
|
6735 |
+
{
|
6736 |
+
"epoch": 0.9827984595635431,
|
6737 |
+
"grad_norm": 1.2429808378219604,
|
6738 |
+
"learning_rate": 1.534272281645488e-07,
|
6739 |
+
"loss": 2.3365,
|
6740 |
+
"step": 957
|
6741 |
+
},
|
6742 |
+
{
|
6743 |
+
"epoch": 0.9838254172015405,
|
6744 |
+
"grad_norm": 1.4925501346588135,
|
6745 |
+
"learning_rate": 1.359118257221903e-07,
|
6746 |
+
"loss": 2.2891,
|
6747 |
+
"step": 958
|
6748 |
+
},
|
6749 |
+
{
|
6750 |
+
"epoch": 0.9848523748395379,
|
6751 |
+
"grad_norm": 1.5961631536483765,
|
6752 |
+
"learning_rate": 1.1945703063402924e-07,
|
6753 |
+
"loss": 2.2596,
|
6754 |
+
"step": 959
|
6755 |
+
},
|
6756 |
+
{
|
6757 |
+
"epoch": 0.9858793324775353,
|
6758 |
+
"grad_norm": 1.4775968790054321,
|
6759 |
+
"learning_rate": 1.0406301765837345e-07,
|
6760 |
+
"loss": 2.2753,
|
6761 |
+
"step": 960
|
6762 |
+
},
|
6763 |
+
{
|
6764 |
+
"epoch": 0.9869062901155328,
|
6765 |
+
"grad_norm": 1.4262019395828247,
|
6766 |
+
"learning_rate": 8.972995028745379e-08,
|
6767 |
+
"loss": 2.2151,
|
6768 |
+
"step": 961
|
6769 |
+
},
|
6770 |
+
{
|
6771 |
+
"epoch": 0.9879332477535302,
|
6772 |
+
"grad_norm": 1.8904521465301514,
|
6773 |
+
"learning_rate": 7.645798074572552e-08,
|
6774 |
+
"loss": 2.3732,
|
6775 |
+
"step": 962
|
6776 |
+
},
|
6777 |
+
{
|
6778 |
+
"epoch": 0.9889602053915276,
|
6779 |
+
"grad_norm": 1.2109746932983398,
|
6780 |
+
"learning_rate": 6.424724998825848e-08,
|
6781 |
+
"loss": 2.2444,
|
6782 |
+
"step": 963
|
6783 |
+
},
|
6784 |
+
{
|
6785 |
+
"epoch": 0.989987163029525,
|
6786 |
+
"grad_norm": 1.230819582939148,
|
6787 |
+
"learning_rate": 5.3097887699193885e-08,
|
6788 |
+
"loss": 2.1996,
|
6789 |
+
"step": 964
|
6790 |
+
},
|
6791 |
+
{
|
6792 |
+
"epoch": 0.9910141206675225,
|
6793 |
+
"grad_norm": 1.2803691625595093,
|
6794 |
+
"learning_rate": 4.3010012290445324e-08,
|
6795 |
+
"loss": 2.3936,
|
6796 |
+
"step": 965
|
6797 |
+
},
|
6798 |
+
{
|
6799 |
+
"epoch": 0.9920410783055199,
|
6800 |
+
"grad_norm": 1.3575743436813354,
|
6801 |
+
"learning_rate": 3.3983730900377655e-08,
|
6802 |
+
"loss": 2.3142,
|
6803 |
+
"step": 966
|
6804 |
+
},
|
6805 |
+
{
|
6806 |
+
"epoch": 0.9930680359435173,
|
6807 |
+
"grad_norm": 1.3123668432235718,
|
6808 |
+
"learning_rate": 2.601913939266343e-08,
|
6809 |
+
"loss": 2.158,
|
6810 |
+
"step": 967
|
6811 |
+
},
|
6812 |
+
{
|
6813 |
+
"epoch": 0.9940949935815148,
|
6814 |
+
"grad_norm": 1.5591670274734497,
|
6815 |
+
"learning_rate": 1.9116322355339222e-08,
|
6816 |
+
"loss": 2.1941,
|
6817 |
+
"step": 968
|
6818 |
+
},
|
6819 |
+
{
|
6820 |
+
"epoch": 0.9951219512195122,
|
6821 |
+
"grad_norm": 1.1492890119552612,
|
6822 |
+
"learning_rate": 1.3275353099795329e-08,
|
6823 |
+
"loss": 1.9275,
|
6824 |
+
"step": 969
|
6825 |
+
},
|
6826 |
+
{
|
6827 |
+
"epoch": 0.9961489088575096,
|
6828 |
+
"grad_norm": 1.3780685663223267,
|
6829 |
+
"learning_rate": 8.496293660120724e-09,
|
6830 |
+
"loss": 2.2059,
|
6831 |
+
"step": 970
|
6832 |
+
},
|
6833 |
+
{
|
6834 |
+
"epoch": 0.997175866495507,
|
6835 |
+
"grad_norm": 1.3340226411819458,
|
6836 |
+
"learning_rate": 4.779194792348119e-09,
|
6837 |
+
"loss": 2.2606,
|
6838 |
+
"step": 971
|
6839 |
+
},
|
6840 |
+
{
|
6841 |
+
"epoch": 0.9982028241335045,
|
6842 |
+
"grad_norm": 1.5144405364990234,
|
6843 |
+
"learning_rate": 2.124095973954354e-09,
|
6844 |
+
"loss": 2.181,
|
6845 |
+
"step": 972
|
6846 |
+
},
|
6847 |
+
{
|
6848 |
+
"epoch": 0.9992297817715019,
|
6849 |
+
"grad_norm": 1.2708326578140259,
|
6850 |
+
"learning_rate": 5.310254034274209e-10,
|
6851 |
+
"loss": 1.9886,
|
6852 |
+
"step": 973
|
6853 |
+
},
|
6854 |
+
{
|
6855 |
+
"epoch": 1.0005134788189987,
|
6856 |
+
"grad_norm": 2.398379325866699,
|
6857 |
+
"learning_rate": 0.0,
|
6858 |
+
"loss": 3.4023,
|
6859 |
+
"step": 974
|
6860 |
}
|
6861 |
],
|
6862 |
"logging_steps": 1,
|
|
|
6871 |
"should_evaluate": false,
|
6872 |
"should_log": false,
|
6873 |
"should_save": true,
|
6874 |
+
"should_training_stop": true
|
6875 |
},
|
6876 |
"attributes": {}
|
6877 |
}
|
6878 |
},
|
6879 |
+
"total_flos": 3.7208949874753536e+17,
|
6880 |
"train_batch_size": 2,
|
6881 |
"trial_name": null,
|
6882 |
"trial_params": null
|