Training in progress, step 3058, 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 167832240
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:316b5b66c011ab655d2de0d0f9b4e3fb9737869533eff5fc1cb619905299a286
|
| 3 |
size 167832240
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 335929123
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:51fea0149fe130b7385c5c2c92755ef0f42f6621fac6fad2ea562cd36563f8ab
|
| 3 |
size 335929123
|
last-checkpoint/rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14645
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:aac6559199f848eb2e15456c1e5cee636970b1ba1477e47f5e59da91c5e0c169
|
| 3 |
size 14645
|
last-checkpoint/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1465
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3571d042b38772fa078cbf10ae0e9c85dcf4749dee484ca4e2ce1b0fc9a1dab4
|
| 3 |
size 1465
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -2,9 +2,9 @@
|
|
| 2 |
"best_global_step": 1529,
|
| 3 |
"best_metric": 0.15474164485931396,
|
| 4 |
"best_model_checkpoint": "./results_ner_lora_entity_aware/checkpoint-1529",
|
| 5 |
-
"epoch": 0.
|
| 6 |
"eval_steps": 1529,
|
| 7 |
-
"global_step":
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
@@ -1080,6 +1080,1085 @@
|
|
| 1080 |
"eval_samples_per_second": 3.817,
|
| 1081 |
"eval_steps_per_second": 3.817,
|
| 1082 |
"step": 1529
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1083 |
}
|
| 1084 |
],
|
| 1085 |
"logging_steps": 10,
|
|
@@ -1094,7 +2173,7 @@
|
|
| 1094 |
"early_stopping_threshold": 0.0
|
| 1095 |
},
|
| 1096 |
"attributes": {
|
| 1097 |
-
"early_stopping_patience_counter":
|
| 1098 |
}
|
| 1099 |
},
|
| 1100 |
"TrainerControl": {
|
|
@@ -1108,7 +2187,7 @@
|
|
| 1108 |
"attributes": {}
|
| 1109 |
}
|
| 1110 |
},
|
| 1111 |
-
"total_flos":
|
| 1112 |
"train_batch_size": 1,
|
| 1113 |
"trial_name": null,
|
| 1114 |
"trial_params": null
|
|
|
|
| 2 |
"best_global_step": 1529,
|
| 3 |
"best_metric": 0.15474164485931396,
|
| 4 |
"best_model_checkpoint": "./results_ner_lora_entity_aware/checkpoint-1529",
|
| 5 |
+
"epoch": 0.993663688058489,
|
| 6 |
"eval_steps": 1529,
|
| 7 |
+
"global_step": 3058,
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
|
|
| 1080 |
"eval_samples_per_second": 3.817,
|
| 1081 |
"eval_steps_per_second": 3.817,
|
| 1082 |
"step": 1529
|
| 1083 |
+
},
|
| 1084 |
+
{
|
| 1085 |
+
"epoch": 0.49715678310316813,
|
| 1086 |
+
"grad_norm": 0.4904119372367859,
|
| 1087 |
+
"learning_rate": 4.481201106059251e-05,
|
| 1088 |
+
"loss": 0.0326,
|
| 1089 |
+
"step": 1530
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"epoch": 0.5004061738424046,
|
| 1093 |
+
"grad_norm": 0.41244634985923767,
|
| 1094 |
+
"learning_rate": 4.472981513766538e-05,
|
| 1095 |
+
"loss": 0.0296,
|
| 1096 |
+
"step": 1540
|
| 1097 |
+
},
|
| 1098 |
+
{
|
| 1099 |
+
"epoch": 0.503655564581641,
|
| 1100 |
+
"grad_norm": 0.4068795442581177,
|
| 1101 |
+
"learning_rate": 4.464704982847008e-05,
|
| 1102 |
+
"loss": 0.0345,
|
| 1103 |
+
"step": 1550
|
| 1104 |
+
},
|
| 1105 |
+
{
|
| 1106 |
+
"epoch": 0.5069049553208773,
|
| 1107 |
+
"grad_norm": 0.3623734712600708,
|
| 1108 |
+
"learning_rate": 4.456371752154549e-05,
|
| 1109 |
+
"loss": 0.0309,
|
| 1110 |
+
"step": 1560
|
| 1111 |
+
},
|
| 1112 |
+
{
|
| 1113 |
+
"epoch": 0.5101543460601138,
|
| 1114 |
+
"grad_norm": 0.3473750054836273,
|
| 1115 |
+
"learning_rate": 4.447982062179358e-05,
|
| 1116 |
+
"loss": 0.0313,
|
| 1117 |
+
"step": 1570
|
| 1118 |
+
},
|
| 1119 |
+
{
|
| 1120 |
+
"epoch": 0.5134037367993501,
|
| 1121 |
+
"grad_norm": 0.4023807644844055,
|
| 1122 |
+
"learning_rate": 4.439536155041e-05,
|
| 1123 |
+
"loss": 0.0328,
|
| 1124 |
+
"step": 1580
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 0.5166531275385865,
|
| 1128 |
+
"grad_norm": 0.3513451814651489,
|
| 1129 |
+
"learning_rate": 4.4310342744814214e-05,
|
| 1130 |
+
"loss": 0.0343,
|
| 1131 |
+
"step": 1590
|
| 1132 |
+
},
|
| 1133 |
+
{
|
| 1134 |
+
"epoch": 0.5199025182778229,
|
| 1135 |
+
"grad_norm": 0.37276962399482727,
|
| 1136 |
+
"learning_rate": 4.4224766658579166e-05,
|
| 1137 |
+
"loss": 0.031,
|
| 1138 |
+
"step": 1600
|
| 1139 |
+
},
|
| 1140 |
+
{
|
| 1141 |
+
"epoch": 0.5231519090170593,
|
| 1142 |
+
"grad_norm": 0.456547349691391,
|
| 1143 |
+
"learning_rate": 4.413863576136044e-05,
|
| 1144 |
+
"loss": 0.0317,
|
| 1145 |
+
"step": 1610
|
| 1146 |
+
},
|
| 1147 |
+
{
|
| 1148 |
+
"epoch": 0.5264012997562957,
|
| 1149 |
+
"grad_norm": 0.3823520541191101,
|
| 1150 |
+
"learning_rate": 4.4051952538825034e-05,
|
| 1151 |
+
"loss": 0.0302,
|
| 1152 |
+
"step": 1620
|
| 1153 |
+
},
|
| 1154 |
+
{
|
| 1155 |
+
"epoch": 0.529650690495532,
|
| 1156 |
+
"grad_norm": 0.42634570598602295,
|
| 1157 |
+
"learning_rate": 4.3964719492579584e-05,
|
| 1158 |
+
"loss": 0.0288,
|
| 1159 |
+
"step": 1630
|
| 1160 |
+
},
|
| 1161 |
+
{
|
| 1162 |
+
"epoch": 0.5329000812347685,
|
| 1163 |
+
"grad_norm": 0.33916255831718445,
|
| 1164 |
+
"learning_rate": 4.387693914009819e-05,
|
| 1165 |
+
"loss": 0.0298,
|
| 1166 |
+
"step": 1640
|
| 1167 |
+
},
|
| 1168 |
+
{
|
| 1169 |
+
"epoch": 0.5361494719740049,
|
| 1170 |
+
"grad_norm": 0.441773921251297,
|
| 1171 |
+
"learning_rate": 4.3788614014649775e-05,
|
| 1172 |
+
"loss": 0.033,
|
| 1173 |
+
"step": 1650
|
| 1174 |
+
},
|
| 1175 |
+
{
|
| 1176 |
+
"epoch": 0.5393988627132412,
|
| 1177 |
+
"grad_norm": 0.35982316732406616,
|
| 1178 |
+
"learning_rate": 4.3699746665224945e-05,
|
| 1179 |
+
"loss": 0.0267,
|
| 1180 |
+
"step": 1660
|
| 1181 |
+
},
|
| 1182 |
+
{
|
| 1183 |
+
"epoch": 0.5426482534524777,
|
| 1184 |
+
"grad_norm": 0.3963621258735657,
|
| 1185 |
+
"learning_rate": 4.3610339656462445e-05,
|
| 1186 |
+
"loss": 0.0308,
|
| 1187 |
+
"step": 1670
|
| 1188 |
+
},
|
| 1189 |
+
{
|
| 1190 |
+
"epoch": 0.545897644191714,
|
| 1191 |
+
"grad_norm": 0.4419509172439575,
|
| 1192 |
+
"learning_rate": 4.352039556857516e-05,
|
| 1193 |
+
"loss": 0.0325,
|
| 1194 |
+
"step": 1680
|
| 1195 |
+
},
|
| 1196 |
+
{
|
| 1197 |
+
"epoch": 0.5491470349309504,
|
| 1198 |
+
"grad_norm": 0.4380287826061249,
|
| 1199 |
+
"learning_rate": 4.3429916997275626e-05,
|
| 1200 |
+
"loss": 0.0297,
|
| 1201 |
+
"step": 1690
|
| 1202 |
+
},
|
| 1203 |
+
{
|
| 1204 |
+
"epoch": 0.5523964256701869,
|
| 1205 |
+
"grad_norm": 0.27853381633758545,
|
| 1206 |
+
"learning_rate": 4.333890655370113e-05,
|
| 1207 |
+
"loss": 0.0314,
|
| 1208 |
+
"step": 1700
|
| 1209 |
+
},
|
| 1210 |
+
{
|
| 1211 |
+
"epoch": 0.5556458164094232,
|
| 1212 |
+
"grad_norm": 0.3319164514541626,
|
| 1213 |
+
"learning_rate": 4.324736686433837e-05,
|
| 1214 |
+
"loss": 0.0294,
|
| 1215 |
+
"step": 1710
|
| 1216 |
+
},
|
| 1217 |
+
{
|
| 1218 |
+
"epoch": 0.5588952071486596,
|
| 1219 |
+
"grad_norm": 0.5052310824394226,
|
| 1220 |
+
"learning_rate": 4.315530057094762e-05,
|
| 1221 |
+
"loss": 0.0314,
|
| 1222 |
+
"step": 1720
|
| 1223 |
+
},
|
| 1224 |
+
{
|
| 1225 |
+
"epoch": 0.5621445978878961,
|
| 1226 |
+
"grad_norm": 0.2669266164302826,
|
| 1227 |
+
"learning_rate": 4.306271033048655e-05,
|
| 1228 |
+
"loss": 0.0305,
|
| 1229 |
+
"step": 1730
|
| 1230 |
+
},
|
| 1231 |
+
{
|
| 1232 |
+
"epoch": 0.5653939886271324,
|
| 1233 |
+
"grad_norm": 0.3192387819290161,
|
| 1234 |
+
"learning_rate": 4.2969598815033476e-05,
|
| 1235 |
+
"loss": 0.0274,
|
| 1236 |
+
"step": 1740
|
| 1237 |
+
},
|
| 1238 |
+
{
|
| 1239 |
+
"epoch": 0.5686433793663688,
|
| 1240 |
+
"grad_norm": 0.5369754433631897,
|
| 1241 |
+
"learning_rate": 4.2875968711710286e-05,
|
| 1242 |
+
"loss": 0.032,
|
| 1243 |
+
"step": 1750
|
| 1244 |
+
},
|
| 1245 |
+
{
|
| 1246 |
+
"epoch": 0.5718927701056052,
|
| 1247 |
+
"grad_norm": 0.2641260623931885,
|
| 1248 |
+
"learning_rate": 4.2781822722604916e-05,
|
| 1249 |
+
"loss": 0.0272,
|
| 1250 |
+
"step": 1760
|
| 1251 |
+
},
|
| 1252 |
+
{
|
| 1253 |
+
"epoch": 0.5751421608448416,
|
| 1254 |
+
"grad_norm": 0.4835808575153351,
|
| 1255 |
+
"learning_rate": 4.268716356469331e-05,
|
| 1256 |
+
"loss": 0.0286,
|
| 1257 |
+
"step": 1770
|
| 1258 |
+
},
|
| 1259 |
+
{
|
| 1260 |
+
"epoch": 0.578391551584078,
|
| 1261 |
+
"grad_norm": 0.27514582872390747,
|
| 1262 |
+
"learning_rate": 4.259199396976107e-05,
|
| 1263 |
+
"loss": 0.0269,
|
| 1264 |
+
"step": 1780
|
| 1265 |
+
},
|
| 1266 |
+
{
|
| 1267 |
+
"epoch": 0.5816409423233144,
|
| 1268 |
+
"grad_norm": 0.3719632625579834,
|
| 1269 |
+
"learning_rate": 4.2496316684324585e-05,
|
| 1270 |
+
"loss": 0.029,
|
| 1271 |
+
"step": 1790
|
| 1272 |
+
},
|
| 1273 |
+
{
|
| 1274 |
+
"epoch": 0.5848903330625508,
|
| 1275 |
+
"grad_norm": 0.36428073048591614,
|
| 1276 |
+
"learning_rate": 4.2400134469551746e-05,
|
| 1277 |
+
"loss": 0.0304,
|
| 1278 |
+
"step": 1800
|
| 1279 |
+
},
|
| 1280 |
+
{
|
| 1281 |
+
"epoch": 0.5881397238017871,
|
| 1282 |
+
"grad_norm": 0.6899104714393616,
|
| 1283 |
+
"learning_rate": 4.230345010118233e-05,
|
| 1284 |
+
"loss": 0.0296,
|
| 1285 |
+
"step": 1810
|
| 1286 |
+
},
|
| 1287 |
+
{
|
| 1288 |
+
"epoch": 0.5913891145410236,
|
| 1289 |
+
"grad_norm": 0.3395729660987854,
|
| 1290 |
+
"learning_rate": 4.220626636944783e-05,
|
| 1291 |
+
"loss": 0.025,
|
| 1292 |
+
"step": 1820
|
| 1293 |
+
},
|
| 1294 |
+
{
|
| 1295 |
+
"epoch": 0.59463850528026,
|
| 1296 |
+
"grad_norm": 0.38969168066978455,
|
| 1297 |
+
"learning_rate": 4.2108586078990966e-05,
|
| 1298 |
+
"loss": 0.0248,
|
| 1299 |
+
"step": 1830
|
| 1300 |
+
},
|
| 1301 |
+
{
|
| 1302 |
+
"epoch": 0.5978878960194963,
|
| 1303 |
+
"grad_norm": 0.3905599117279053,
|
| 1304 |
+
"learning_rate": 4.2010412048784733e-05,
|
| 1305 |
+
"loss": 0.0257,
|
| 1306 |
+
"step": 1840
|
| 1307 |
+
},
|
| 1308 |
+
{
|
| 1309 |
+
"epoch": 0.6011372867587328,
|
| 1310 |
+
"grad_norm": 0.3760197162628174,
|
| 1311 |
+
"learning_rate": 4.191174711205105e-05,
|
| 1312 |
+
"loss": 0.0278,
|
| 1313 |
+
"step": 1850
|
| 1314 |
+
},
|
| 1315 |
+
{
|
| 1316 |
+
"epoch": 0.6043866774979691,
|
| 1317 |
+
"grad_norm": 0.4263547956943512,
|
| 1318 |
+
"learning_rate": 4.181259411617898e-05,
|
| 1319 |
+
"loss": 0.0248,
|
| 1320 |
+
"step": 1860
|
| 1321 |
+
},
|
| 1322 |
+
{
|
| 1323 |
+
"epoch": 0.6076360682372055,
|
| 1324 |
+
"grad_norm": 0.46628740429878235,
|
| 1325 |
+
"learning_rate": 4.1712955922642614e-05,
|
| 1326 |
+
"loss": 0.0274,
|
| 1327 |
+
"step": 1870
|
| 1328 |
+
},
|
| 1329 |
+
{
|
| 1330 |
+
"epoch": 0.6108854589764419,
|
| 1331 |
+
"grad_norm": 0.46346044540405273,
|
| 1332 |
+
"learning_rate": 4.161283540691841e-05,
|
| 1333 |
+
"loss": 0.0248,
|
| 1334 |
+
"step": 1880
|
| 1335 |
+
},
|
| 1336 |
+
{
|
| 1337 |
+
"epoch": 0.6141348497156783,
|
| 1338 |
+
"grad_norm": 0.37473928928375244,
|
| 1339 |
+
"learning_rate": 4.151223545840225e-05,
|
| 1340 |
+
"loss": 0.0272,
|
| 1341 |
+
"step": 1890
|
| 1342 |
+
},
|
| 1343 |
+
{
|
| 1344 |
+
"epoch": 0.6173842404549147,
|
| 1345 |
+
"grad_norm": 0.4056950807571411,
|
| 1346 |
+
"learning_rate": 4.141115898032607e-05,
|
| 1347 |
+
"loss": 0.024,
|
| 1348 |
+
"step": 1900
|
| 1349 |
+
},
|
| 1350 |
+
{
|
| 1351 |
+
"epoch": 0.620633631194151,
|
| 1352 |
+
"grad_norm": 0.4770098626613617,
|
| 1353 |
+
"learning_rate": 4.130960888967405e-05,
|
| 1354 |
+
"loss": 0.0237,
|
| 1355 |
+
"step": 1910
|
| 1356 |
+
},
|
| 1357 |
+
{
|
| 1358 |
+
"epoch": 0.6238830219333875,
|
| 1359 |
+
"grad_norm": 0.41782036423683167,
|
| 1360 |
+
"learning_rate": 4.1207588117098445e-05,
|
| 1361 |
+
"loss": 0.0272,
|
| 1362 |
+
"step": 1920
|
| 1363 |
+
},
|
| 1364 |
+
{
|
| 1365 |
+
"epoch": 0.6271324126726239,
|
| 1366 |
+
"grad_norm": 0.4040960669517517,
|
| 1367 |
+
"learning_rate": 4.1105099606835e-05,
|
| 1368 |
+
"loss": 0.0256,
|
| 1369 |
+
"step": 1930
|
| 1370 |
+
},
|
| 1371 |
+
{
|
| 1372 |
+
"epoch": 0.6303818034118602,
|
| 1373 |
+
"grad_norm": 0.3390992283821106,
|
| 1374 |
+
"learning_rate": 4.1002146316617986e-05,
|
| 1375 |
+
"loss": 0.0276,
|
| 1376 |
+
"step": 1940
|
| 1377 |
+
},
|
| 1378 |
+
{
|
| 1379 |
+
"epoch": 0.6336311941510967,
|
| 1380 |
+
"grad_norm": 0.4464505910873413,
|
| 1381 |
+
"learning_rate": 4.0898731217594836e-05,
|
| 1382 |
+
"loss": 0.0265,
|
| 1383 |
+
"step": 1950
|
| 1384 |
+
},
|
| 1385 |
+
{
|
| 1386 |
+
"epoch": 0.6368805848903331,
|
| 1387 |
+
"grad_norm": 0.46650540828704834,
|
| 1388 |
+
"learning_rate": 4.0794857294240415e-05,
|
| 1389 |
+
"loss": 0.0229,
|
| 1390 |
+
"step": 1960
|
| 1391 |
+
},
|
| 1392 |
+
{
|
| 1393 |
+
"epoch": 0.6401299756295694,
|
| 1394 |
+
"grad_norm": 0.39290061593055725,
|
| 1395 |
+
"learning_rate": 4.0690527544270886e-05,
|
| 1396 |
+
"loss": 0.0207,
|
| 1397 |
+
"step": 1970
|
| 1398 |
+
},
|
| 1399 |
+
{
|
| 1400 |
+
"epoch": 0.6433793663688059,
|
| 1401 |
+
"grad_norm": 0.38553354144096375,
|
| 1402 |
+
"learning_rate": 4.0585744978557174e-05,
|
| 1403 |
+
"loss": 0.0267,
|
| 1404 |
+
"step": 1980
|
| 1405 |
+
},
|
| 1406 |
+
{
|
| 1407 |
+
"epoch": 0.6466287571080422,
|
| 1408 |
+
"grad_norm": 0.456086128950119,
|
| 1409 |
+
"learning_rate": 4.048051262103811e-05,
|
| 1410 |
+
"loss": 0.0264,
|
| 1411 |
+
"step": 1990
|
| 1412 |
+
},
|
| 1413 |
+
{
|
| 1414 |
+
"epoch": 0.6498781478472786,
|
| 1415 |
+
"grad_norm": 0.5561078786849976,
|
| 1416 |
+
"learning_rate": 4.0374833508633156e-05,
|
| 1417 |
+
"loss": 0.0218,
|
| 1418 |
+
"step": 2000
|
| 1419 |
+
},
|
| 1420 |
+
{
|
| 1421 |
+
"epoch": 0.6531275385865151,
|
| 1422 |
+
"grad_norm": 0.3495825231075287,
|
| 1423 |
+
"learning_rate": 4.0268710691154724e-05,
|
| 1424 |
+
"loss": 0.0226,
|
| 1425 |
+
"step": 2010
|
| 1426 |
+
},
|
| 1427 |
+
{
|
| 1428 |
+
"epoch": 0.6563769293257514,
|
| 1429 |
+
"grad_norm": 0.4553760588169098,
|
| 1430 |
+
"learning_rate": 4.0162147231220216e-05,
|
| 1431 |
+
"loss": 0.024,
|
| 1432 |
+
"step": 2020
|
| 1433 |
+
},
|
| 1434 |
+
{
|
| 1435 |
+
"epoch": 0.6596263200649878,
|
| 1436 |
+
"grad_norm": 0.40624141693115234,
|
| 1437 |
+
"learning_rate": 4.0055146204163605e-05,
|
| 1438 |
+
"loss": 0.022,
|
| 1439 |
+
"step": 2030
|
| 1440 |
+
},
|
| 1441 |
+
{
|
| 1442 |
+
"epoch": 0.6628757108042242,
|
| 1443 |
+
"grad_norm": 0.37093663215637207,
|
| 1444 |
+
"learning_rate": 3.994771069794668e-05,
|
| 1445 |
+
"loss": 0.0241,
|
| 1446 |
+
"step": 2040
|
| 1447 |
+
},
|
| 1448 |
+
{
|
| 1449 |
+
"epoch": 0.6661251015434606,
|
| 1450 |
+
"grad_norm": 0.5362465977668762,
|
| 1451 |
+
"learning_rate": 3.9839843813069984e-05,
|
| 1452 |
+
"loss": 0.0246,
|
| 1453 |
+
"step": 2050
|
| 1454 |
+
},
|
| 1455 |
+
{
|
| 1456 |
+
"epoch": 0.669374492282697,
|
| 1457 |
+
"grad_norm": 0.4092622995376587,
|
| 1458 |
+
"learning_rate": 3.9731548662483234e-05,
|
| 1459 |
+
"loss": 0.0241,
|
| 1460 |
+
"step": 2060
|
| 1461 |
+
},
|
| 1462 |
+
{
|
| 1463 |
+
"epoch": 0.6726238830219334,
|
| 1464 |
+
"grad_norm": 0.41743412613868713,
|
| 1465 |
+
"learning_rate": 3.962282837149558e-05,
|
| 1466 |
+
"loss": 0.0232,
|
| 1467 |
+
"step": 2070
|
| 1468 |
+
},
|
| 1469 |
+
{
|
| 1470 |
+
"epoch": 0.6758732737611698,
|
| 1471 |
+
"grad_norm": 0.5507500171661377,
|
| 1472 |
+
"learning_rate": 3.951368607768537e-05,
|
| 1473 |
+
"loss": 0.0223,
|
| 1474 |
+
"step": 2080
|
| 1475 |
+
},
|
| 1476 |
+
{
|
| 1477 |
+
"epoch": 0.6791226645004061,
|
| 1478 |
+
"grad_norm": 0.45379722118377686,
|
| 1479 |
+
"learning_rate": 3.9404124930809625e-05,
|
| 1480 |
+
"loss": 0.0242,
|
| 1481 |
+
"step": 2090
|
| 1482 |
+
},
|
| 1483 |
+
{
|
| 1484 |
+
"epoch": 0.6823720552396426,
|
| 1485 |
+
"grad_norm": 0.36823663115501404,
|
| 1486 |
+
"learning_rate": 3.929414809271308e-05,
|
| 1487 |
+
"loss": 0.0265,
|
| 1488 |
+
"step": 2100
|
| 1489 |
+
},
|
| 1490 |
+
{
|
| 1491 |
+
"epoch": 0.685621445978879,
|
| 1492 |
+
"grad_norm": 0.3751804530620575,
|
| 1493 |
+
"learning_rate": 3.918375873723701e-05,
|
| 1494 |
+
"loss": 0.0245,
|
| 1495 |
+
"step": 2110
|
| 1496 |
+
},
|
| 1497 |
+
{
|
| 1498 |
+
"epoch": 0.6888708367181153,
|
| 1499 |
+
"grad_norm": 0.4553548991680145,
|
| 1500 |
+
"learning_rate": 3.907296005012758e-05,
|
| 1501 |
+
"loss": 0.0257,
|
| 1502 |
+
"step": 2120
|
| 1503 |
+
},
|
| 1504 |
+
{
|
| 1505 |
+
"epoch": 0.6921202274573518,
|
| 1506 |
+
"grad_norm": 0.3335023820400238,
|
| 1507 |
+
"learning_rate": 3.896175522894395e-05,
|
| 1508 |
+
"loss": 0.0226,
|
| 1509 |
+
"step": 2130
|
| 1510 |
+
},
|
| 1511 |
+
{
|
| 1512 |
+
"epoch": 0.6953696181965882,
|
| 1513 |
+
"grad_norm": 0.4702470600605011,
|
| 1514 |
+
"learning_rate": 3.8850147482965973e-05,
|
| 1515 |
+
"loss": 0.0218,
|
| 1516 |
+
"step": 2140
|
| 1517 |
+
},
|
| 1518 |
+
{
|
| 1519 |
+
"epoch": 0.6986190089358245,
|
| 1520 |
+
"grad_norm": 0.39953991770744324,
|
| 1521 |
+
"learning_rate": 3.873814003310158e-05,
|
| 1522 |
+
"loss": 0.0194,
|
| 1523 |
+
"step": 2150
|
| 1524 |
+
},
|
| 1525 |
+
{
|
| 1526 |
+
"epoch": 0.7018683996750609,
|
| 1527 |
+
"grad_norm": 0.5067981481552124,
|
| 1528 |
+
"learning_rate": 3.862573611179381e-05,
|
| 1529 |
+
"loss": 0.022,
|
| 1530 |
+
"step": 2160
|
| 1531 |
+
},
|
| 1532 |
+
{
|
| 1533 |
+
"epoch": 0.7051177904142973,
|
| 1534 |
+
"grad_norm": 0.30548062920570374,
|
| 1535 |
+
"learning_rate": 3.851293896292756e-05,
|
| 1536 |
+
"loss": 0.0171,
|
| 1537 |
+
"step": 2170
|
| 1538 |
+
},
|
| 1539 |
+
{
|
| 1540 |
+
"epoch": 0.7083671811535337,
|
| 1541 |
+
"grad_norm": 0.39521142840385437,
|
| 1542 |
+
"learning_rate": 3.839975184173596e-05,
|
| 1543 |
+
"loss": 0.0192,
|
| 1544 |
+
"step": 2180
|
| 1545 |
+
},
|
| 1546 |
+
{
|
| 1547 |
+
"epoch": 0.7116165718927701,
|
| 1548 |
+
"grad_norm": 0.36969834566116333,
|
| 1549 |
+
"learning_rate": 3.8286178014706395e-05,
|
| 1550 |
+
"loss": 0.0245,
|
| 1551 |
+
"step": 2190
|
| 1552 |
+
},
|
| 1553 |
+
{
|
| 1554 |
+
"epoch": 0.7148659626320065,
|
| 1555 |
+
"grad_norm": 0.4855635166168213,
|
| 1556 |
+
"learning_rate": 3.8172220759486287e-05,
|
| 1557 |
+
"loss": 0.0229,
|
| 1558 |
+
"step": 2200
|
| 1559 |
+
},
|
| 1560 |
+
{
|
| 1561 |
+
"epoch": 0.7181153533712429,
|
| 1562 |
+
"grad_norm": 0.45929041504859924,
|
| 1563 |
+
"learning_rate": 3.8057883364788475e-05,
|
| 1564 |
+
"loss": 0.0186,
|
| 1565 |
+
"step": 2210
|
| 1566 |
+
},
|
| 1567 |
+
{
|
| 1568 |
+
"epoch": 0.7213647441104792,
|
| 1569 |
+
"grad_norm": 0.2941083610057831,
|
| 1570 |
+
"learning_rate": 3.7943169130296295e-05,
|
| 1571 |
+
"loss": 0.0188,
|
| 1572 |
+
"step": 2220
|
| 1573 |
+
},
|
| 1574 |
+
{
|
| 1575 |
+
"epoch": 0.7246141348497157,
|
| 1576 |
+
"grad_norm": 0.3690025806427002,
|
| 1577 |
+
"learning_rate": 3.782808136656839e-05,
|
| 1578 |
+
"loss": 0.0188,
|
| 1579 |
+
"step": 2230
|
| 1580 |
+
},
|
| 1581 |
+
{
|
| 1582 |
+
"epoch": 0.7278635255889521,
|
| 1583 |
+
"grad_norm": 0.38714373111724854,
|
| 1584 |
+
"learning_rate": 3.771262339494314e-05,
|
| 1585 |
+
"loss": 0.0191,
|
| 1586 |
+
"step": 2240
|
| 1587 |
+
},
|
| 1588 |
+
{
|
| 1589 |
+
"epoch": 0.7311129163281884,
|
| 1590 |
+
"grad_norm": 0.40680810809135437,
|
| 1591 |
+
"learning_rate": 3.759679854744282e-05,
|
| 1592 |
+
"loss": 0.0197,
|
| 1593 |
+
"step": 2250
|
| 1594 |
+
},
|
| 1595 |
+
{
|
| 1596 |
+
"epoch": 0.7343623070674249,
|
| 1597 |
+
"grad_norm": 0.41902831196784973,
|
| 1598 |
+
"learning_rate": 3.748061016667745e-05,
|
| 1599 |
+
"loss": 0.0205,
|
| 1600 |
+
"step": 2260
|
| 1601 |
+
},
|
| 1602 |
+
{
|
| 1603 |
+
"epoch": 0.7376116978066612,
|
| 1604 |
+
"grad_norm": 0.4369294047355652,
|
| 1605 |
+
"learning_rate": 3.736406160574833e-05,
|
| 1606 |
+
"loss": 0.019,
|
| 1607 |
+
"step": 2270
|
| 1608 |
+
},
|
| 1609 |
+
{
|
| 1610 |
+
"epoch": 0.7408610885458976,
|
| 1611 |
+
"grad_norm": 0.3856930732727051,
|
| 1612 |
+
"learning_rate": 3.724715622815122e-05,
|
| 1613 |
+
"loss": 0.022,
|
| 1614 |
+
"step": 2280
|
| 1615 |
+
},
|
| 1616 |
+
{
|
| 1617 |
+
"epoch": 0.7441104792851341,
|
| 1618 |
+
"grad_norm": 0.34679755568504333,
|
| 1619 |
+
"learning_rate": 3.712989740767938e-05,
|
| 1620 |
+
"loss": 0.0164,
|
| 1621 |
+
"step": 2290
|
| 1622 |
+
},
|
| 1623 |
+
{
|
| 1624 |
+
"epoch": 0.7473598700243704,
|
| 1625 |
+
"grad_norm": 0.3927323818206787,
|
| 1626 |
+
"learning_rate": 3.7012288528326086e-05,
|
| 1627 |
+
"loss": 0.0181,
|
| 1628 |
+
"step": 2300
|
| 1629 |
+
},
|
| 1630 |
+
{
|
| 1631 |
+
"epoch": 0.7506092607636068,
|
| 1632 |
+
"grad_norm": 0.4021192491054535,
|
| 1633 |
+
"learning_rate": 3.689433298418706e-05,
|
| 1634 |
+
"loss": 0.0159,
|
| 1635 |
+
"step": 2310
|
| 1636 |
+
},
|
| 1637 |
+
{
|
| 1638 |
+
"epoch": 0.7538586515028433,
|
| 1639 |
+
"grad_norm": 0.48003751039505005,
|
| 1640 |
+
"learning_rate": 3.6776034179362474e-05,
|
| 1641 |
+
"loss": 0.0177,
|
| 1642 |
+
"step": 2320
|
| 1643 |
+
},
|
| 1644 |
+
{
|
| 1645 |
+
"epoch": 0.7571080422420796,
|
| 1646 |
+
"grad_norm": 0.3487580716609955,
|
| 1647 |
+
"learning_rate": 3.66573955278587e-05,
|
| 1648 |
+
"loss": 0.0162,
|
| 1649 |
+
"step": 2330
|
| 1650 |
+
},
|
| 1651 |
+
{
|
| 1652 |
+
"epoch": 0.760357432981316,
|
| 1653 |
+
"grad_norm": 0.47422289848327637,
|
| 1654 |
+
"learning_rate": 3.653842045348985e-05,
|
| 1655 |
+
"loss": 0.018,
|
| 1656 |
+
"step": 2340
|
| 1657 |
+
},
|
| 1658 |
+
{
|
| 1659 |
+
"epoch": 0.7636068237205524,
|
| 1660 |
+
"grad_norm": 0.38853368163108826,
|
| 1661 |
+
"learning_rate": 3.64191123897789e-05,
|
| 1662 |
+
"loss": 0.0229,
|
| 1663 |
+
"step": 2350
|
| 1664 |
+
},
|
| 1665 |
+
{
|
| 1666 |
+
"epoch": 0.7668562144597888,
|
| 1667 |
+
"grad_norm": 0.36860230565071106,
|
| 1668 |
+
"learning_rate": 3.62994747798586e-05,
|
| 1669 |
+
"loss": 0.018,
|
| 1670 |
+
"step": 2360
|
| 1671 |
+
},
|
| 1672 |
+
{
|
| 1673 |
+
"epoch": 0.7701056051990252,
|
| 1674 |
+
"grad_norm": 0.4562481939792633,
|
| 1675 |
+
"learning_rate": 3.617951107637219e-05,
|
| 1676 |
+
"loss": 0.0191,
|
| 1677 |
+
"step": 2370
|
| 1678 |
+
},
|
| 1679 |
+
{
|
| 1680 |
+
"epoch": 0.7733549959382616,
|
| 1681 |
+
"grad_norm": 0.708402156829834,
|
| 1682 |
+
"learning_rate": 3.605922474137366e-05,
|
| 1683 |
+
"loss": 0.019,
|
| 1684 |
+
"step": 2380
|
| 1685 |
+
},
|
| 1686 |
+
{
|
| 1687 |
+
"epoch": 0.776604386677498,
|
| 1688 |
+
"grad_norm": 0.48525258898735046,
|
| 1689 |
+
"learning_rate": 3.5938619246227884e-05,
|
| 1690 |
+
"loss": 0.0217,
|
| 1691 |
+
"step": 2390
|
| 1692 |
+
},
|
| 1693 |
+
{
|
| 1694 |
+
"epoch": 0.7798537774167343,
|
| 1695 |
+
"grad_norm": 0.3320712447166443,
|
| 1696 |
+
"learning_rate": 3.581769807151044e-05,
|
| 1697 |
+
"loss": 0.0195,
|
| 1698 |
+
"step": 2400
|
| 1699 |
+
},
|
| 1700 |
+
{
|
| 1701 |
+
"epoch": 0.7831031681559708,
|
| 1702 |
+
"grad_norm": 0.36696651577949524,
|
| 1703 |
+
"learning_rate": 3.56964647069072e-05,
|
| 1704 |
+
"loss": 0.0192,
|
| 1705 |
+
"step": 2410
|
| 1706 |
+
},
|
| 1707 |
+
{
|
| 1708 |
+
"epoch": 0.7863525588952072,
|
| 1709 |
+
"grad_norm": 1.0005451440811157,
|
| 1710 |
+
"learning_rate": 3.55749226511135e-05,
|
| 1711 |
+
"loss": 0.0196,
|
| 1712 |
+
"step": 2420
|
| 1713 |
+
},
|
| 1714 |
+
{
|
| 1715 |
+
"epoch": 0.7896019496344435,
|
| 1716 |
+
"grad_norm": 0.45593878626823425,
|
| 1717 |
+
"learning_rate": 3.54530754117333e-05,
|
| 1718 |
+
"loss": 0.0225,
|
| 1719 |
+
"step": 2430
|
| 1720 |
+
},
|
| 1721 |
+
{
|
| 1722 |
+
"epoch": 0.7928513403736799,
|
| 1723 |
+
"grad_norm": 0.33774876594543457,
|
| 1724 |
+
"learning_rate": 3.533092650517793e-05,
|
| 1725 |
+
"loss": 0.0199,
|
| 1726 |
+
"step": 2440
|
| 1727 |
+
},
|
| 1728 |
+
{
|
| 1729 |
+
"epoch": 0.7961007311129163,
|
| 1730 |
+
"grad_norm": 0.483853816986084,
|
| 1731 |
+
"learning_rate": 3.5208479456564524e-05,
|
| 1732 |
+
"loss": 0.0206,
|
| 1733 |
+
"step": 2450
|
| 1734 |
+
},
|
| 1735 |
+
{
|
| 1736 |
+
"epoch": 0.7993501218521527,
|
| 1737 |
+
"grad_norm": 0.36340585350990295,
|
| 1738 |
+
"learning_rate": 3.508573779961441e-05,
|
| 1739 |
+
"loss": 0.0172,
|
| 1740 |
+
"step": 2460
|
| 1741 |
+
},
|
| 1742 |
+
{
|
| 1743 |
+
"epoch": 0.8025995125913891,
|
| 1744 |
+
"grad_norm": 0.515352725982666,
|
| 1745 |
+
"learning_rate": 3.4962705076551026e-05,
|
| 1746 |
+
"loss": 0.0151,
|
| 1747 |
+
"step": 2470
|
| 1748 |
+
},
|
| 1749 |
+
{
|
| 1750 |
+
"epoch": 0.8058489033306255,
|
| 1751 |
+
"grad_norm": 0.3187580406665802,
|
| 1752 |
+
"learning_rate": 3.483938483799778e-05,
|
| 1753 |
+
"loss": 0.0148,
|
| 1754 |
+
"step": 2480
|
| 1755 |
+
},
|
| 1756 |
+
{
|
| 1757 |
+
"epoch": 0.8090982940698619,
|
| 1758 |
+
"grad_norm": 0.4531770646572113,
|
| 1759 |
+
"learning_rate": 3.47157806428755e-05,
|
| 1760 |
+
"loss": 0.0158,
|
| 1761 |
+
"step": 2490
|
| 1762 |
+
},
|
| 1763 |
+
{
|
| 1764 |
+
"epoch": 0.8123476848090982,
|
| 1765 |
+
"grad_norm": 0.6547293066978455,
|
| 1766 |
+
"learning_rate": 3.45918960582998e-05,
|
| 1767 |
+
"loss": 0.0144,
|
| 1768 |
+
"step": 2500
|
| 1769 |
+
},
|
| 1770 |
+
{
|
| 1771 |
+
"epoch": 0.8155970755483347,
|
| 1772 |
+
"grad_norm": 0.24430936574935913,
|
| 1773 |
+
"learning_rate": 3.446773465947809e-05,
|
| 1774 |
+
"loss": 0.0148,
|
| 1775 |
+
"step": 2510
|
| 1776 |
+
},
|
| 1777 |
+
{
|
| 1778 |
+
"epoch": 0.8188464662875711,
|
| 1779 |
+
"grad_norm": 0.425484299659729,
|
| 1780 |
+
"learning_rate": 3.4343300029606404e-05,
|
| 1781 |
+
"loss": 0.0192,
|
| 1782 |
+
"step": 2520
|
| 1783 |
+
},
|
| 1784 |
+
{
|
| 1785 |
+
"epoch": 0.8220958570268074,
|
| 1786 |
+
"grad_norm": 0.43435633182525635,
|
| 1787 |
+
"learning_rate": 3.4218595759766013e-05,
|
| 1788 |
+
"loss": 0.0192,
|
| 1789 |
+
"step": 2530
|
| 1790 |
+
},
|
| 1791 |
+
{
|
| 1792 |
+
"epoch": 0.8253452477660439,
|
| 1793 |
+
"grad_norm": 0.499104380607605,
|
| 1794 |
+
"learning_rate": 3.409362544881977e-05,
|
| 1795 |
+
"loss": 0.0172,
|
| 1796 |
+
"step": 2540
|
| 1797 |
+
},
|
| 1798 |
+
{
|
| 1799 |
+
"epoch": 0.8285946385052803,
|
| 1800 |
+
"grad_norm": 0.5690107941627502,
|
| 1801 |
+
"learning_rate": 3.3968392703308264e-05,
|
| 1802 |
+
"loss": 0.017,
|
| 1803 |
+
"step": 2550
|
| 1804 |
+
},
|
| 1805 |
+
{
|
| 1806 |
+
"epoch": 0.8318440292445166,
|
| 1807 |
+
"grad_norm": 0.4789363145828247,
|
| 1808 |
+
"learning_rate": 3.3842901137345725e-05,
|
| 1809 |
+
"loss": 0.0151,
|
| 1810 |
+
"step": 2560
|
| 1811 |
+
},
|
| 1812 |
+
{
|
| 1813 |
+
"epoch": 0.8350934199837531,
|
| 1814 |
+
"grad_norm": 0.34053072333335876,
|
| 1815 |
+
"learning_rate": 3.3717154372515716e-05,
|
| 1816 |
+
"loss": 0.0155,
|
| 1817 |
+
"step": 2570
|
| 1818 |
+
},
|
| 1819 |
+
{
|
| 1820 |
+
"epoch": 0.8383428107229894,
|
| 1821 |
+
"grad_norm": 0.44401687383651733,
|
| 1822 |
+
"learning_rate": 3.3591156037766655e-05,
|
| 1823 |
+
"loss": 0.0138,
|
| 1824 |
+
"step": 2580
|
| 1825 |
+
},
|
| 1826 |
+
{
|
| 1827 |
+
"epoch": 0.8415922014622258,
|
| 1828 |
+
"grad_norm": 0.41682168841362,
|
| 1829 |
+
"learning_rate": 3.346490976930704e-05,
|
| 1830 |
+
"loss": 0.014,
|
| 1831 |
+
"step": 2590
|
| 1832 |
+
},
|
| 1833 |
+
{
|
| 1834 |
+
"epoch": 0.8448415922014623,
|
| 1835 |
+
"grad_norm": 0.6371095776557922,
|
| 1836 |
+
"learning_rate": 3.333841921050053e-05,
|
| 1837 |
+
"loss": 0.0176,
|
| 1838 |
+
"step": 2600
|
| 1839 |
+
},
|
| 1840 |
+
{
|
| 1841 |
+
"epoch": 0.8480909829406986,
|
| 1842 |
+
"grad_norm": 0.22434203326702118,
|
| 1843 |
+
"learning_rate": 3.3211688011760835e-05,
|
| 1844 |
+
"loss": 0.0123,
|
| 1845 |
+
"step": 2610
|
| 1846 |
+
},
|
| 1847 |
+
{
|
| 1848 |
+
"epoch": 0.851340373679935,
|
| 1849 |
+
"grad_norm": 0.5385378003120422,
|
| 1850 |
+
"learning_rate": 3.30847198304463e-05,
|
| 1851 |
+
"loss": 0.016,
|
| 1852 |
+
"step": 2620
|
| 1853 |
+
},
|
| 1854 |
+
{
|
| 1855 |
+
"epoch": 0.8545897644191714,
|
| 1856 |
+
"grad_norm": 0.35033077001571655,
|
| 1857 |
+
"learning_rate": 3.2957518330754406e-05,
|
| 1858 |
+
"loss": 0.0149,
|
| 1859 |
+
"step": 2630
|
| 1860 |
+
},
|
| 1861 |
+
{
|
| 1862 |
+
"epoch": 0.8578391551584078,
|
| 1863 |
+
"grad_norm": 0.3793995976448059,
|
| 1864 |
+
"learning_rate": 3.2830087183616015e-05,
|
| 1865 |
+
"loss": 0.0153,
|
| 1866 |
+
"step": 2640
|
| 1867 |
+
},
|
| 1868 |
+
{
|
| 1869 |
+
"epoch": 0.8610885458976442,
|
| 1870 |
+
"grad_norm": 0.6376614570617676,
|
| 1871 |
+
"learning_rate": 3.270243006658942e-05,
|
| 1872 |
+
"loss": 0.0154,
|
| 1873 |
+
"step": 2650
|
| 1874 |
+
},
|
| 1875 |
+
{
|
| 1876 |
+
"epoch": 0.8643379366368806,
|
| 1877 |
+
"grad_norm": 0.3490144908428192,
|
| 1878 |
+
"learning_rate": 3.257455066375423e-05,
|
| 1879 |
+
"loss": 0.0154,
|
| 1880 |
+
"step": 2660
|
| 1881 |
+
},
|
| 1882 |
+
{
|
| 1883 |
+
"epoch": 0.867587327376117,
|
| 1884 |
+
"grad_norm": 0.40602752566337585,
|
| 1885 |
+
"learning_rate": 3.244645266560501e-05,
|
| 1886 |
+
"loss": 0.0136,
|
| 1887 |
+
"step": 2670
|
| 1888 |
+
},
|
| 1889 |
+
{
|
| 1890 |
+
"epoch": 0.8708367181153533,
|
| 1891 |
+
"grad_norm": 0.37300461530685425,
|
| 1892 |
+
"learning_rate": 3.2318139768944856e-05,
|
| 1893 |
+
"loss": 0.0127,
|
| 1894 |
+
"step": 2680
|
| 1895 |
+
},
|
| 1896 |
+
{
|
| 1897 |
+
"epoch": 0.8740861088545898,
|
| 1898 |
+
"grad_norm": 0.3366054594516754,
|
| 1899 |
+
"learning_rate": 3.218961567677861e-05,
|
| 1900 |
+
"loss": 0.0142,
|
| 1901 |
+
"step": 2690
|
| 1902 |
+
},
|
| 1903 |
+
{
|
| 1904 |
+
"epoch": 0.8773354995938262,
|
| 1905 |
+
"grad_norm": 0.4088799059391022,
|
| 1906 |
+
"learning_rate": 3.206088409820606e-05,
|
| 1907 |
+
"loss": 0.0143,
|
| 1908 |
+
"step": 2700
|
| 1909 |
+
},
|
| 1910 |
+
{
|
| 1911 |
+
"epoch": 0.8805848903330625,
|
| 1912 |
+
"grad_norm": 0.3606589734554291,
|
| 1913 |
+
"learning_rate": 3.19319487483149e-05,
|
| 1914 |
+
"loss": 0.0125,
|
| 1915 |
+
"step": 2710
|
| 1916 |
+
},
|
| 1917 |
+
{
|
| 1918 |
+
"epoch": 0.8838342810722989,
|
| 1919 |
+
"grad_norm": 0.48760858178138733,
|
| 1920 |
+
"learning_rate": 3.180281334807348e-05,
|
| 1921 |
+
"loss": 0.0121,
|
| 1922 |
+
"step": 2720
|
| 1923 |
+
},
|
| 1924 |
+
{
|
| 1925 |
+
"epoch": 0.8870836718115354,
|
| 1926 |
+
"grad_norm": 0.4494096040725708,
|
| 1927 |
+
"learning_rate": 3.1673481624223426e-05,
|
| 1928 |
+
"loss": 0.0123,
|
| 1929 |
+
"step": 2730
|
| 1930 |
+
},
|
| 1931 |
+
{
|
| 1932 |
+
"epoch": 0.8903330625507717,
|
| 1933 |
+
"grad_norm": 0.4649289846420288,
|
| 1934 |
+
"learning_rate": 3.154395730917213e-05,
|
| 1935 |
+
"loss": 0.0135,
|
| 1936 |
+
"step": 2740
|
| 1937 |
+
},
|
| 1938 |
+
{
|
| 1939 |
+
"epoch": 0.8935824532900081,
|
| 1940 |
+
"grad_norm": 0.3580702841281891,
|
| 1941 |
+
"learning_rate": 3.141424414088499e-05,
|
| 1942 |
+
"loss": 0.014,
|
| 1943 |
+
"step": 2750
|
| 1944 |
+
},
|
| 1945 |
+
{
|
| 1946 |
+
"epoch": 0.8968318440292445,
|
| 1947 |
+
"grad_norm": 0.2921467423439026,
|
| 1948 |
+
"learning_rate": 3.128434586277757e-05,
|
| 1949 |
+
"loss": 0.0146,
|
| 1950 |
+
"step": 2760
|
| 1951 |
+
},
|
| 1952 |
+
{
|
| 1953 |
+
"epoch": 0.9000812347684809,
|
| 1954 |
+
"grad_norm": 0.5378055572509766,
|
| 1955 |
+
"learning_rate": 3.115426622360752e-05,
|
| 1956 |
+
"loss": 0.0131,
|
| 1957 |
+
"step": 2770
|
| 1958 |
+
},
|
| 1959 |
+
{
|
| 1960 |
+
"epoch": 0.9033306255077173,
|
| 1961 |
+
"grad_norm": 0.34166646003723145,
|
| 1962 |
+
"learning_rate": 3.102400897736645e-05,
|
| 1963 |
+
"loss": 0.0123,
|
| 1964 |
+
"step": 2780
|
| 1965 |
+
},
|
| 1966 |
+
{
|
| 1967 |
+
"epoch": 0.9065800162469537,
|
| 1968 |
+
"grad_norm": 0.43183135986328125,
|
| 1969 |
+
"learning_rate": 3.0893577883171556e-05,
|
| 1970 |
+
"loss": 0.0151,
|
| 1971 |
+
"step": 2790
|
| 1972 |
+
},
|
| 1973 |
+
{
|
| 1974 |
+
"epoch": 0.9098294069861901,
|
| 1975 |
+
"grad_norm": 0.6324542760848999,
|
| 1976 |
+
"learning_rate": 3.076297670515713e-05,
|
| 1977 |
+
"loss": 0.0128,
|
| 1978 |
+
"step": 2800
|
| 1979 |
+
},
|
| 1980 |
+
{
|
| 1981 |
+
"epoch": 0.9130787977254264,
|
| 1982 |
+
"grad_norm": 0.43282851576805115,
|
| 1983 |
+
"learning_rate": 3.063220921236598e-05,
|
| 1984 |
+
"loss": 0.0129,
|
| 1985 |
+
"step": 2810
|
| 1986 |
+
},
|
| 1987 |
+
{
|
| 1988 |
+
"epoch": 0.9163281884646629,
|
| 1989 |
+
"grad_norm": 0.2942393124103546,
|
| 1990 |
+
"learning_rate": 3.0501279178640575e-05,
|
| 1991 |
+
"loss": 0.0131,
|
| 1992 |
+
"step": 2820
|
| 1993 |
+
},
|
| 1994 |
+
{
|
| 1995 |
+
"epoch": 0.9195775792038993,
|
| 1996 |
+
"grad_norm": 0.32284924387931824,
|
| 1997 |
+
"learning_rate": 3.0370190382514213e-05,
|
| 1998 |
+
"loss": 0.0103,
|
| 1999 |
+
"step": 2830
|
| 2000 |
+
},
|
| 2001 |
+
{
|
| 2002 |
+
"epoch": 0.9228269699431356,
|
| 2003 |
+
"grad_norm": 0.38951289653778076,
|
| 2004 |
+
"learning_rate": 3.0238946607101936e-05,
|
| 2005 |
+
"loss": 0.0105,
|
| 2006 |
+
"step": 2840
|
| 2007 |
+
},
|
| 2008 |
+
{
|
| 2009 |
+
"epoch": 0.9260763606823721,
|
| 2010 |
+
"grad_norm": 0.407099187374115,
|
| 2011 |
+
"learning_rate": 3.0107551639991365e-05,
|
| 2012 |
+
"loss": 0.0109,
|
| 2013 |
+
"step": 2850
|
| 2014 |
+
},
|
| 2015 |
+
{
|
| 2016 |
+
"epoch": 0.9293257514216084,
|
| 2017 |
+
"grad_norm": 0.5025432705879211,
|
| 2018 |
+
"learning_rate": 2.997600927313338e-05,
|
| 2019 |
+
"loss": 0.0115,
|
| 2020 |
+
"step": 2860
|
| 2021 |
+
},
|
| 2022 |
+
{
|
| 2023 |
+
"epoch": 0.9325751421608448,
|
| 2024 |
+
"grad_norm": 0.18864522874355316,
|
| 2025 |
+
"learning_rate": 2.98443233027327e-05,
|
| 2026 |
+
"loss": 0.011,
|
| 2027 |
+
"step": 2870
|
| 2028 |
+
},
|
| 2029 |
+
{
|
| 2030 |
+
"epoch": 0.9358245329000813,
|
| 2031 |
+
"grad_norm": 0.3602016866207123,
|
| 2032 |
+
"learning_rate": 2.971249752913834e-05,
|
| 2033 |
+
"loss": 0.012,
|
| 2034 |
+
"step": 2880
|
| 2035 |
+
},
|
| 2036 |
+
{
|
| 2037 |
+
"epoch": 0.9390739236393176,
|
| 2038 |
+
"grad_norm": 0.38901951909065247,
|
| 2039 |
+
"learning_rate": 2.958053575673389e-05,
|
| 2040 |
+
"loss": 0.0113,
|
| 2041 |
+
"step": 2890
|
| 2042 |
+
},
|
| 2043 |
+
{
|
| 2044 |
+
"epoch": 0.942323314378554,
|
| 2045 |
+
"grad_norm": 0.3340912163257599,
|
| 2046 |
+
"learning_rate": 2.944844179382778e-05,
|
| 2047 |
+
"loss": 0.0102,
|
| 2048 |
+
"step": 2900
|
| 2049 |
+
},
|
| 2050 |
+
{
|
| 2051 |
+
"epoch": 0.9455727051177905,
|
| 2052 |
+
"grad_norm": 0.32333633303642273,
|
| 2053 |
+
"learning_rate": 2.931621945254334e-05,
|
| 2054 |
+
"loss": 0.0117,
|
| 2055 |
+
"step": 2910
|
| 2056 |
+
},
|
| 2057 |
+
{
|
| 2058 |
+
"epoch": 0.9488220958570268,
|
| 2059 |
+
"grad_norm": 0.45609724521636963,
|
| 2060 |
+
"learning_rate": 2.918387254870879e-05,
|
| 2061 |
+
"loss": 0.0114,
|
| 2062 |
+
"step": 2920
|
| 2063 |
+
},
|
| 2064 |
+
{
|
| 2065 |
+
"epoch": 0.9520714865962632,
|
| 2066 |
+
"grad_norm": 0.3951948285102844,
|
| 2067 |
+
"learning_rate": 2.905140490174713e-05,
|
| 2068 |
+
"loss": 0.0099,
|
| 2069 |
+
"step": 2930
|
| 2070 |
+
},
|
| 2071 |
+
{
|
| 2072 |
+
"epoch": 0.9553208773354996,
|
| 2073 |
+
"grad_norm": 0.4054865539073944,
|
| 2074 |
+
"learning_rate": 2.8918820334565905e-05,
|
| 2075 |
+
"loss": 0.0118,
|
| 2076 |
+
"step": 2940
|
| 2077 |
+
},
|
| 2078 |
+
{
|
| 2079 |
+
"epoch": 0.958570268074736,
|
| 2080 |
+
"grad_norm": 0.40964823961257935,
|
| 2081 |
+
"learning_rate": 2.8786122673446893e-05,
|
| 2082 |
+
"loss": 0.0113,
|
| 2083 |
+
"step": 2950
|
| 2084 |
+
},
|
| 2085 |
+
{
|
| 2086 |
+
"epoch": 0.9618196588139724,
|
| 2087 |
+
"grad_norm": 0.37629735469818115,
|
| 2088 |
+
"learning_rate": 2.865331574793564e-05,
|
| 2089 |
+
"loss": 0.0112,
|
| 2090 |
+
"step": 2960
|
| 2091 |
+
},
|
| 2092 |
+
{
|
| 2093 |
+
"epoch": 0.9650690495532088,
|
| 2094 |
+
"grad_norm": 0.3705214262008667,
|
| 2095 |
+
"learning_rate": 2.8520403390731e-05,
|
| 2096 |
+
"loss": 0.0117,
|
| 2097 |
+
"step": 2970
|
| 2098 |
+
},
|
| 2099 |
+
{
|
| 2100 |
+
"epoch": 0.9683184402924452,
|
| 2101 |
+
"grad_norm": 0.3900469243526459,
|
| 2102 |
+
"learning_rate": 2.8387389437574495e-05,
|
| 2103 |
+
"loss": 0.0108,
|
| 2104 |
+
"step": 2980
|
| 2105 |
+
},
|
| 2106 |
+
{
|
| 2107 |
+
"epoch": 0.9715678310316815,
|
| 2108 |
+
"grad_norm": 0.2905224561691284,
|
| 2109 |
+
"learning_rate": 2.8254277727139616e-05,
|
| 2110 |
+
"loss": 0.0112,
|
| 2111 |
+
"step": 2990
|
| 2112 |
+
},
|
| 2113 |
+
{
|
| 2114 |
+
"epoch": 0.974817221770918,
|
| 2115 |
+
"grad_norm": 0.6466222405433655,
|
| 2116 |
+
"learning_rate": 2.812107210092105e-05,
|
| 2117 |
+
"loss": 0.0124,
|
| 2118 |
+
"step": 3000
|
| 2119 |
+
},
|
| 2120 |
+
{
|
| 2121 |
+
"epoch": 0.9780666125101544,
|
| 2122 |
+
"grad_norm": 0.5542816519737244,
|
| 2123 |
+
"learning_rate": 2.798777640312381e-05,
|
| 2124 |
+
"loss": 0.0112,
|
| 2125 |
+
"step": 3010
|
| 2126 |
+
},
|
| 2127 |
+
{
|
| 2128 |
+
"epoch": 0.9813160032493907,
|
| 2129 |
+
"grad_norm": 0.3827550411224365,
|
| 2130 |
+
"learning_rate": 2.7854394480552327e-05,
|
| 2131 |
+
"loss": 0.0112,
|
| 2132 |
+
"step": 3020
|
| 2133 |
+
},
|
| 2134 |
+
{
|
| 2135 |
+
"epoch": 0.9845653939886271,
|
| 2136 |
+
"grad_norm": 0.32786279916763306,
|
| 2137 |
+
"learning_rate": 2.7720930182499367e-05,
|
| 2138 |
+
"loss": 0.0115,
|
| 2139 |
+
"step": 3030
|
| 2140 |
+
},
|
| 2141 |
+
{
|
| 2142 |
+
"epoch": 0.9878147847278635,
|
| 2143 |
+
"grad_norm": 0.537501335144043,
|
| 2144 |
+
"learning_rate": 2.7587387360635032e-05,
|
| 2145 |
+
"loss": 0.0113,
|
| 2146 |
+
"step": 3040
|
| 2147 |
+
},
|
| 2148 |
+
{
|
| 2149 |
+
"epoch": 0.9910641754670999,
|
| 2150 |
+
"grad_norm": 0.47765976190567017,
|
| 2151 |
+
"learning_rate": 2.7453769868895518e-05,
|
| 2152 |
+
"loss": 0.0141,
|
| 2153 |
+
"step": 3050
|
| 2154 |
+
},
|
| 2155 |
+
{
|
| 2156 |
+
"epoch": 0.993663688058489,
|
| 2157 |
+
"eval_loss": 0.1697877049446106,
|
| 2158 |
+
"eval_runtime": 733.8424,
|
| 2159 |
+
"eval_samples_per_second": 3.816,
|
| 2160 |
+
"eval_steps_per_second": 3.816,
|
| 2161 |
+
"step": 3058
|
| 2162 |
}
|
| 2163 |
],
|
| 2164 |
"logging_steps": 10,
|
|
|
|
| 2173 |
"early_stopping_threshold": 0.0
|
| 2174 |
},
|
| 2175 |
"attributes": {
|
| 2176 |
+
"early_stopping_patience_counter": 1
|
| 2177 |
}
|
| 2178 |
},
|
| 2179 |
"TrainerControl": {
|
|
|
|
| 2187 |
"attributes": {}
|
| 2188 |
}
|
| 2189 |
},
|
| 2190 |
+
"total_flos": 2.385847921489871e+18,
|
| 2191 |
"train_batch_size": 1,
|
| 2192 |
"trial_name": null,
|
| 2193 |
"trial_params": null
|