Training in progress, step 4587, 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 sha256:9fd7f4d081b21537033b2e160ee74b52ff15f87f0e71f38ed6e387185a25b5e5
|
| 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:c737b1b79e1bfe3e9e148430849f11da864c1df799a3157441e39c1df3ea568e
|
| 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:b8946718702c3c2d38c431f0f7665cb26032dd3547d26860b2c656bc09d58caf
|
| 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 sha256:7b642c5acc0130ce5250f6eb7e0e04c4b290c2f7e9e26d0ace9cf16a6273c57c
|
| 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":
|
| 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,
|
|
@@ -2159,6 +2159,1085 @@
|
|
| 2159 |
"eval_samples_per_second": 3.816,
|
| 2160 |
"eval_steps_per_second": 3.816,
|
| 2161 |
"step": 3058
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 2162 |
}
|
| 2163 |
],
|
| 2164 |
"logging_steps": 10,
|
|
@@ -2173,7 +3252,7 @@
|
|
| 2173 |
"early_stopping_threshold": 0.0
|
| 2174 |
},
|
| 2175 |
"attributes": {
|
| 2176 |
-
"early_stopping_patience_counter":
|
| 2177 |
}
|
| 2178 |
},
|
| 2179 |
"TrainerControl": {
|
|
@@ -2182,12 +3261,12 @@
|
|
| 2182 |
"should_evaluate": false,
|
| 2183 |
"should_log": false,
|
| 2184 |
"should_save": true,
|
| 2185 |
-
"should_training_stop":
|
| 2186 |
},
|
| 2187 |
"attributes": {}
|
| 2188 |
}
|
| 2189 |
},
|
| 2190 |
-
"total_flos":
|
| 2191 |
"train_batch_size": 1,
|
| 2192 |
"trial_name": null,
|
| 2193 |
"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": 1.4903330625507718,
|
| 6 |
"eval_steps": 1529,
|
| 7 |
+
"global_step": 4587,
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
|
|
| 2159 |
"eval_samples_per_second": 3.816,
|
| 2160 |
"eval_steps_per_second": 3.816,
|
| 2161 |
"step": 3058
|
| 2162 |
+
},
|
| 2163 |
+
{
|
| 2164 |
+
"epoch": 0.9943135662063363,
|
| 2165 |
+
"grad_norm": 0.6474146246910095,
|
| 2166 |
+
"learning_rate": 2.7320081563371948e-05,
|
| 2167 |
+
"loss": 0.0085,
|
| 2168 |
+
"step": 3060
|
| 2169 |
+
},
|
| 2170 |
+
{
|
| 2171 |
+
"epoch": 0.9975629569455727,
|
| 2172 |
+
"grad_norm": 0.4375881850719452,
|
| 2173 |
+
"learning_rate": 2.718632630219907e-05,
|
| 2174 |
+
"loss": 0.0114,
|
| 2175 |
+
"step": 3070
|
| 2176 |
+
},
|
| 2177 |
+
{
|
| 2178 |
+
"epoch": 1.0006498781478472,
|
| 2179 |
+
"grad_norm": 0.327910840511322,
|
| 2180 |
+
"learning_rate": 2.7052507945443927e-05,
|
| 2181 |
+
"loss": 0.009,
|
| 2182 |
+
"step": 3080
|
| 2183 |
+
},
|
| 2184 |
+
{
|
| 2185 |
+
"epoch": 1.0038992688870836,
|
| 2186 |
+
"grad_norm": 0.4386765658855438,
|
| 2187 |
+
"learning_rate": 2.6918630354994434e-05,
|
| 2188 |
+
"loss": 0.0081,
|
| 2189 |
+
"step": 3090
|
| 2190 |
+
},
|
| 2191 |
+
{
|
| 2192 |
+
"epoch": 1.0071486596263202,
|
| 2193 |
+
"grad_norm": 0.3415273129940033,
|
| 2194 |
+
"learning_rate": 2.6784697394447942e-05,
|
| 2195 |
+
"loss": 0.007,
|
| 2196 |
+
"step": 3100
|
| 2197 |
+
},
|
| 2198 |
+
{
|
| 2199 |
+
"epoch": 1.0103980503655565,
|
| 2200 |
+
"grad_norm": 0.31640973687171936,
|
| 2201 |
+
"learning_rate": 2.6650712928999755e-05,
|
| 2202 |
+
"loss": 0.009,
|
| 2203 |
+
"step": 3110
|
| 2204 |
+
},
|
| 2205 |
+
{
|
| 2206 |
+
"epoch": 1.0136474411047929,
|
| 2207 |
+
"grad_norm": 0.4809754490852356,
|
| 2208 |
+
"learning_rate": 2.6516680825331548e-05,
|
| 2209 |
+
"loss": 0.0065,
|
| 2210 |
+
"step": 3120
|
| 2211 |
+
},
|
| 2212 |
+
{
|
| 2213 |
+
"epoch": 1.0168968318440292,
|
| 2214 |
+
"grad_norm": 0.48411983251571655,
|
| 2215 |
+
"learning_rate": 2.6382604951499802e-05,
|
| 2216 |
+
"loss": 0.0064,
|
| 2217 |
+
"step": 3130
|
| 2218 |
+
},
|
| 2219 |
+
{
|
| 2220 |
+
"epoch": 1.0201462225832656,
|
| 2221 |
+
"grad_norm": 0.5264196395874023,
|
| 2222 |
+
"learning_rate": 2.624848917682417e-05,
|
| 2223 |
+
"loss": 0.0076,
|
| 2224 |
+
"step": 3140
|
| 2225 |
+
},
|
| 2226 |
+
{
|
| 2227 |
+
"epoch": 1.023395613322502,
|
| 2228 |
+
"grad_norm": 0.5818049311637878,
|
| 2229 |
+
"learning_rate": 2.6114337371775815e-05,
|
| 2230 |
+
"loss": 0.0069,
|
| 2231 |
+
"step": 3150
|
| 2232 |
+
},
|
| 2233 |
+
{
|
| 2234 |
+
"epoch": 1.0266450040617385,
|
| 2235 |
+
"grad_norm": 0.20237238705158234,
|
| 2236 |
+
"learning_rate": 2.5980153407865694e-05,
|
| 2237 |
+
"loss": 0.0059,
|
| 2238 |
+
"step": 3160
|
| 2239 |
+
},
|
| 2240 |
+
{
|
| 2241 |
+
"epoch": 1.0298943948009749,
|
| 2242 |
+
"grad_norm": 0.4652661681175232,
|
| 2243 |
+
"learning_rate": 2.5845941157532856e-05,
|
| 2244 |
+
"loss": 0.0058,
|
| 2245 |
+
"step": 3170
|
| 2246 |
+
},
|
| 2247 |
+
{
|
| 2248 |
+
"epoch": 1.0331437855402112,
|
| 2249 |
+
"grad_norm": 0.42343178391456604,
|
| 2250 |
+
"learning_rate": 2.5711704494032662e-05,
|
| 2251 |
+
"loss": 0.0072,
|
| 2252 |
+
"step": 3180
|
| 2253 |
+
},
|
| 2254 |
+
{
|
| 2255 |
+
"epoch": 1.0363931762794476,
|
| 2256 |
+
"grad_norm": 0.31664910912513733,
|
| 2257 |
+
"learning_rate": 2.557744729132503e-05,
|
| 2258 |
+
"loss": 0.0062,
|
| 2259 |
+
"step": 3190
|
| 2260 |
+
},
|
| 2261 |
+
{
|
| 2262 |
+
"epoch": 1.039642567018684,
|
| 2263 |
+
"grad_norm": 0.4600023627281189,
|
| 2264 |
+
"learning_rate": 2.5443173423962606e-05,
|
| 2265 |
+
"loss": 0.0065,
|
| 2266 |
+
"step": 3200
|
| 2267 |
+
},
|
| 2268 |
+
{
|
| 2269 |
+
"epoch": 1.0428919577579203,
|
| 2270 |
+
"grad_norm": 0.21867072582244873,
|
| 2271 |
+
"learning_rate": 2.5308886766978985e-05,
|
| 2272 |
+
"loss": 0.0075,
|
| 2273 |
+
"step": 3210
|
| 2274 |
+
},
|
| 2275 |
+
{
|
| 2276 |
+
"epoch": 1.0461413484971567,
|
| 2277 |
+
"grad_norm": 0.2808971405029297,
|
| 2278 |
+
"learning_rate": 2.517459119577685e-05,
|
| 2279 |
+
"loss": 0.0064,
|
| 2280 |
+
"step": 3220
|
| 2281 |
+
},
|
| 2282 |
+
{
|
| 2283 |
+
"epoch": 1.0493907392363933,
|
| 2284 |
+
"grad_norm": 0.3623920679092407,
|
| 2285 |
+
"learning_rate": 2.504029058601612e-05,
|
| 2286 |
+
"loss": 0.0057,
|
| 2287 |
+
"step": 3230
|
| 2288 |
+
},
|
| 2289 |
+
{
|
| 2290 |
+
"epoch": 1.0526401299756296,
|
| 2291 |
+
"grad_norm": 0.7856932282447815,
|
| 2292 |
+
"learning_rate": 2.490598881350215e-05,
|
| 2293 |
+
"loss": 0.0069,
|
| 2294 |
+
"step": 3240
|
| 2295 |
+
},
|
| 2296 |
+
{
|
| 2297 |
+
"epoch": 1.055889520714866,
|
| 2298 |
+
"grad_norm": 0.3538922965526581,
|
| 2299 |
+
"learning_rate": 2.4771689754073858e-05,
|
| 2300 |
+
"loss": 0.0067,
|
| 2301 |
+
"step": 3250
|
| 2302 |
+
},
|
| 2303 |
+
{
|
| 2304 |
+
"epoch": 1.0591389114541023,
|
| 2305 |
+
"grad_norm": 0.40585437417030334,
|
| 2306 |
+
"learning_rate": 2.4637397283491828e-05,
|
| 2307 |
+
"loss": 0.0068,
|
| 2308 |
+
"step": 3260
|
| 2309 |
+
},
|
| 2310 |
+
{
|
| 2311 |
+
"epoch": 1.0623883021933387,
|
| 2312 |
+
"grad_norm": 0.3392549753189087,
|
| 2313 |
+
"learning_rate": 2.450311527732653e-05,
|
| 2314 |
+
"loss": 0.0075,
|
| 2315 |
+
"step": 3270
|
| 2316 |
+
},
|
| 2317 |
+
{
|
| 2318 |
+
"epoch": 1.065637692932575,
|
| 2319 |
+
"grad_norm": 0.4685852527618408,
|
| 2320 |
+
"learning_rate": 2.436884761084642e-05,
|
| 2321 |
+
"loss": 0.0076,
|
| 2322 |
+
"step": 3280
|
| 2323 |
+
},
|
| 2324 |
+
{
|
| 2325 |
+
"epoch": 1.0688870836718116,
|
| 2326 |
+
"grad_norm": 0.2836906611919403,
|
| 2327 |
+
"learning_rate": 2.423459815890614e-05,
|
| 2328 |
+
"loss": 0.0063,
|
| 2329 |
+
"step": 3290
|
| 2330 |
+
},
|
| 2331 |
+
{
|
| 2332 |
+
"epoch": 1.072136474411048,
|
| 2333 |
+
"grad_norm": 0.3446894586086273,
|
| 2334 |
+
"learning_rate": 2.4100370795834652e-05,
|
| 2335 |
+
"loss": 0.0056,
|
| 2336 |
+
"step": 3300
|
| 2337 |
+
},
|
| 2338 |
+
{
|
| 2339 |
+
"epoch": 1.0753858651502843,
|
| 2340 |
+
"grad_norm": 0.6391892433166504,
|
| 2341 |
+
"learning_rate": 2.3966169395323466e-05,
|
| 2342 |
+
"loss": 0.007,
|
| 2343 |
+
"step": 3310
|
| 2344 |
+
},
|
| 2345 |
+
{
|
| 2346 |
+
"epoch": 1.0786352558895207,
|
| 2347 |
+
"grad_norm": 0.4981069564819336,
|
| 2348 |
+
"learning_rate": 2.383199783031484e-05,
|
| 2349 |
+
"loss": 0.0069,
|
| 2350 |
+
"step": 3320
|
| 2351 |
+
},
|
| 2352 |
+
{
|
| 2353 |
+
"epoch": 1.081884646628757,
|
| 2354 |
+
"grad_norm": 0.3438924252986908,
|
| 2355 |
+
"learning_rate": 2.369785997288998e-05,
|
| 2356 |
+
"loss": 0.0062,
|
| 2357 |
+
"step": 3330
|
| 2358 |
+
},
|
| 2359 |
+
{
|
| 2360 |
+
"epoch": 1.0851340373679934,
|
| 2361 |
+
"grad_norm": 0.6114248633384705,
|
| 2362 |
+
"learning_rate": 2.356375969415735e-05,
|
| 2363 |
+
"loss": 0.006,
|
| 2364 |
+
"step": 3340
|
| 2365 |
+
},
|
| 2366 |
+
{
|
| 2367 |
+
"epoch": 1.08838342810723,
|
| 2368 |
+
"grad_norm": 0.658473551273346,
|
| 2369 |
+
"learning_rate": 2.3429700864140892e-05,
|
| 2370 |
+
"loss": 0.0057,
|
| 2371 |
+
"step": 3350
|
| 2372 |
+
},
|
| 2373 |
+
{
|
| 2374 |
+
"epoch": 1.0916328188464663,
|
| 2375 |
+
"grad_norm": 0.2530268430709839,
|
| 2376 |
+
"learning_rate": 2.3295687351668407e-05,
|
| 2377 |
+
"loss": 0.0057,
|
| 2378 |
+
"step": 3360
|
| 2379 |
+
},
|
| 2380 |
+
{
|
| 2381 |
+
"epoch": 1.0948822095857027,
|
| 2382 |
+
"grad_norm": 0.33400505781173706,
|
| 2383 |
+
"learning_rate": 2.3161723024259832e-05,
|
| 2384 |
+
"loss": 0.0058,
|
| 2385 |
+
"step": 3370
|
| 2386 |
+
},
|
| 2387 |
+
{
|
| 2388 |
+
"epoch": 1.098131600324939,
|
| 2389 |
+
"grad_norm": 0.47318515181541443,
|
| 2390 |
+
"learning_rate": 2.302781174801569e-05,
|
| 2391 |
+
"loss": 0.0058,
|
| 2392 |
+
"step": 3380
|
| 2393 |
+
},
|
| 2394 |
+
{
|
| 2395 |
+
"epoch": 1.1013809910641754,
|
| 2396 |
+
"grad_norm": 0.18566595017910004,
|
| 2397 |
+
"learning_rate": 2.2893957387505488e-05,
|
| 2398 |
+
"loss": 0.0048,
|
| 2399 |
+
"step": 3390
|
| 2400 |
+
},
|
| 2401 |
+
{
|
| 2402 |
+
"epoch": 1.1046303818034118,
|
| 2403 |
+
"grad_norm": 0.43583136796951294,
|
| 2404 |
+
"learning_rate": 2.2760163805656172e-05,
|
| 2405 |
+
"loss": 0.0062,
|
| 2406 |
+
"step": 3400
|
| 2407 |
+
},
|
| 2408 |
+
{
|
| 2409 |
+
"epoch": 1.1078797725426484,
|
| 2410 |
+
"grad_norm": 0.09334202855825424,
|
| 2411 |
+
"learning_rate": 2.262643486364069e-05,
|
| 2412 |
+
"loss": 0.0054,
|
| 2413 |
+
"step": 3410
|
| 2414 |
+
},
|
| 2415 |
+
{
|
| 2416 |
+
"epoch": 1.1111291632818847,
|
| 2417 |
+
"grad_norm": 0.4679158627986908,
|
| 2418 |
+
"learning_rate": 2.2492774420766518e-05,
|
| 2419 |
+
"loss": 0.0058,
|
| 2420 |
+
"step": 3420
|
| 2421 |
+
},
|
| 2422 |
+
{
|
| 2423 |
+
"epoch": 1.114378554021121,
|
| 2424 |
+
"grad_norm": 0.2904549539089203,
|
| 2425 |
+
"learning_rate": 2.2359186334364314e-05,
|
| 2426 |
+
"loss": 0.0062,
|
| 2427 |
+
"step": 3430
|
| 2428 |
+
},
|
| 2429 |
+
{
|
| 2430 |
+
"epoch": 1.1176279447603574,
|
| 2431 |
+
"grad_norm": 0.2544921338558197,
|
| 2432 |
+
"learning_rate": 2.22256744596766e-05,
|
| 2433 |
+
"loss": 0.0054,
|
| 2434 |
+
"step": 3440
|
| 2435 |
+
},
|
| 2436 |
+
{
|
| 2437 |
+
"epoch": 1.1208773354995938,
|
| 2438 |
+
"grad_norm": 0.3968259394168854,
|
| 2439 |
+
"learning_rate": 2.2092242649746468e-05,
|
| 2440 |
+
"loss": 0.007,
|
| 2441 |
+
"step": 3450
|
| 2442 |
+
},
|
| 2443 |
+
{
|
| 2444 |
+
"epoch": 1.1241267262388301,
|
| 2445 |
+
"grad_norm": 0.5143113732337952,
|
| 2446 |
+
"learning_rate": 2.195889475530641e-05,
|
| 2447 |
+
"loss": 0.0063,
|
| 2448 |
+
"step": 3460
|
| 2449 |
+
},
|
| 2450 |
+
{
|
| 2451 |
+
"epoch": 1.1273761169780667,
|
| 2452 |
+
"grad_norm": 0.39842620491981506,
|
| 2453 |
+
"learning_rate": 2.1825634624667188e-05,
|
| 2454 |
+
"loss": 0.0043,
|
| 2455 |
+
"step": 3470
|
| 2456 |
+
},
|
| 2457 |
+
{
|
| 2458 |
+
"epoch": 1.130625507717303,
|
| 2459 |
+
"grad_norm": 0.7474611401557922,
|
| 2460 |
+
"learning_rate": 2.169246610360679e-05,
|
| 2461 |
+
"loss": 0.0044,
|
| 2462 |
+
"step": 3480
|
| 2463 |
+
},
|
| 2464 |
+
{
|
| 2465 |
+
"epoch": 1.1338748984565394,
|
| 2466 |
+
"grad_norm": 0.43382173776626587,
|
| 2467 |
+
"learning_rate": 2.15593930352594e-05,
|
| 2468 |
+
"loss": 0.0068,
|
| 2469 |
+
"step": 3490
|
| 2470 |
+
},
|
| 2471 |
+
{
|
| 2472 |
+
"epoch": 1.1371242891957758,
|
| 2473 |
+
"grad_norm": 0.43323034048080444,
|
| 2474 |
+
"learning_rate": 2.1426419260004533e-05,
|
| 2475 |
+
"loss": 0.0066,
|
| 2476 |
+
"step": 3500
|
| 2477 |
+
},
|
| 2478 |
+
{
|
| 2479 |
+
"epoch": 1.1403736799350122,
|
| 2480 |
+
"grad_norm": 0.5351451635360718,
|
| 2481 |
+
"learning_rate": 2.1293548615356175e-05,
|
| 2482 |
+
"loss": 0.0059,
|
| 2483 |
+
"step": 3510
|
| 2484 |
+
},
|
| 2485 |
+
{
|
| 2486 |
+
"epoch": 1.1436230706742485,
|
| 2487 |
+
"grad_norm": 0.3306446373462677,
|
| 2488 |
+
"learning_rate": 2.1160784935852065e-05,
|
| 2489 |
+
"loss": 0.0063,
|
| 2490 |
+
"step": 3520
|
| 2491 |
+
},
|
| 2492 |
+
{
|
| 2493 |
+
"epoch": 1.1468724614134849,
|
| 2494 |
+
"grad_norm": 0.8898324370384216,
|
| 2495 |
+
"learning_rate": 2.1028132052942995e-05,
|
| 2496 |
+
"loss": 0.0059,
|
| 2497 |
+
"step": 3530
|
| 2498 |
+
},
|
| 2499 |
+
{
|
| 2500 |
+
"epoch": 1.1501218521527214,
|
| 2501 |
+
"grad_norm": 0.3271152973175049,
|
| 2502 |
+
"learning_rate": 2.0895593794882268e-05,
|
| 2503 |
+
"loss": 0.0055,
|
| 2504 |
+
"step": 3540
|
| 2505 |
+
},
|
| 2506 |
+
{
|
| 2507 |
+
"epoch": 1.1533712428919578,
|
| 2508 |
+
"grad_norm": 0.22426745295524597,
|
| 2509 |
+
"learning_rate": 2.0763173986615216e-05,
|
| 2510 |
+
"loss": 0.005,
|
| 2511 |
+
"step": 3550
|
| 2512 |
+
},
|
| 2513 |
+
{
|
| 2514 |
+
"epoch": 1.1566206336311942,
|
| 2515 |
+
"grad_norm": 0.42375096678733826,
|
| 2516 |
+
"learning_rate": 2.063087644966879e-05,
|
| 2517 |
+
"loss": 0.0049,
|
| 2518 |
+
"step": 3560
|
| 2519 |
+
},
|
| 2520 |
+
{
|
| 2521 |
+
"epoch": 1.1598700243704305,
|
| 2522 |
+
"grad_norm": 0.4661770761013031,
|
| 2523 |
+
"learning_rate": 2.04987050020413e-05,
|
| 2524 |
+
"loss": 0.0048,
|
| 2525 |
+
"step": 3570
|
| 2526 |
+
},
|
| 2527 |
+
{
|
| 2528 |
+
"epoch": 1.1631194151096669,
|
| 2529 |
+
"grad_norm": 0.4002689719200134,
|
| 2530 |
+
"learning_rate": 2.0366663458092224e-05,
|
| 2531 |
+
"loss": 0.0045,
|
| 2532 |
+
"step": 3580
|
| 2533 |
+
},
|
| 2534 |
+
{
|
| 2535 |
+
"epoch": 1.1663688058489032,
|
| 2536 |
+
"grad_norm": 0.5181555151939392,
|
| 2537 |
+
"learning_rate": 2.0234755628432133e-05,
|
| 2538 |
+
"loss": 0.0043,
|
| 2539 |
+
"step": 3590
|
| 2540 |
+
},
|
| 2541 |
+
{
|
| 2542 |
+
"epoch": 1.1696181965881398,
|
| 2543 |
+
"grad_norm": 0.4838791787624359,
|
| 2544 |
+
"learning_rate": 2.0102985319812688e-05,
|
| 2545 |
+
"loss": 0.0059,
|
| 2546 |
+
"step": 3600
|
| 2547 |
+
},
|
| 2548 |
+
{
|
| 2549 |
+
"epoch": 1.1728675873273762,
|
| 2550 |
+
"grad_norm": 0.5321421027183533,
|
| 2551 |
+
"learning_rate": 1.9971356335016834e-05,
|
| 2552 |
+
"loss": 0.0062,
|
| 2553 |
+
"step": 3610
|
| 2554 |
+
},
|
| 2555 |
+
{
|
| 2556 |
+
"epoch": 1.1761169780666125,
|
| 2557 |
+
"grad_norm": 0.17948143184185028,
|
| 2558 |
+
"learning_rate": 1.9839872472749013e-05,
|
| 2559 |
+
"loss": 0.0046,
|
| 2560 |
+
"step": 3620
|
| 2561 |
+
},
|
| 2562 |
+
{
|
| 2563 |
+
"epoch": 1.1793663688058489,
|
| 2564 |
+
"grad_norm": 0.5810254216194153,
|
| 2565 |
+
"learning_rate": 1.9708537527525544e-05,
|
| 2566 |
+
"loss": 0.0051,
|
| 2567 |
+
"step": 3630
|
| 2568 |
+
},
|
| 2569 |
+
{
|
| 2570 |
+
"epoch": 1.1826157595450852,
|
| 2571 |
+
"grad_norm": 0.32746565341949463,
|
| 2572 |
+
"learning_rate": 1.957735528956514e-05,
|
| 2573 |
+
"loss": 0.0061,
|
| 2574 |
+
"step": 3640
|
| 2575 |
+
},
|
| 2576 |
+
{
|
| 2577 |
+
"epoch": 1.1858651502843216,
|
| 2578 |
+
"grad_norm": 0.359332412481308,
|
| 2579 |
+
"learning_rate": 1.9446329544679488e-05,
|
| 2580 |
+
"loss": 0.0058,
|
| 2581 |
+
"step": 3650
|
| 2582 |
+
},
|
| 2583 |
+
{
|
| 2584 |
+
"epoch": 1.189114541023558,
|
| 2585 |
+
"grad_norm": 0.4285711944103241,
|
| 2586 |
+
"learning_rate": 1.9315464074164036e-05,
|
| 2587 |
+
"loss": 0.004,
|
| 2588 |
+
"step": 3660
|
| 2589 |
+
},
|
| 2590 |
+
{
|
| 2591 |
+
"epoch": 1.1923639317627945,
|
| 2592 |
+
"grad_norm": 0.42309272289276123,
|
| 2593 |
+
"learning_rate": 1.918476265468882e-05,
|
| 2594 |
+
"loss": 0.0049,
|
| 2595 |
+
"step": 3670
|
| 2596 |
+
},
|
| 2597 |
+
{
|
| 2598 |
+
"epoch": 1.195613322502031,
|
| 2599 |
+
"grad_norm": 0.495802640914917,
|
| 2600 |
+
"learning_rate": 1.9054229058189514e-05,
|
| 2601 |
+
"loss": 0.0046,
|
| 2602 |
+
"step": 3680
|
| 2603 |
+
},
|
| 2604 |
+
{
|
| 2605 |
+
"epoch": 1.1988627132412673,
|
| 2606 |
+
"grad_norm": 0.39304494857788086,
|
| 2607 |
+
"learning_rate": 1.892386705175856e-05,
|
| 2608 |
+
"loss": 0.0043,
|
| 2609 |
+
"step": 3690
|
| 2610 |
+
},
|
| 2611 |
+
{
|
| 2612 |
+
"epoch": 1.2021121039805036,
|
| 2613 |
+
"grad_norm": 0.30474936962127686,
|
| 2614 |
+
"learning_rate": 1.879368039753644e-05,
|
| 2615 |
+
"loss": 0.0042,
|
| 2616 |
+
"step": 3700
|
| 2617 |
+
},
|
| 2618 |
+
{
|
| 2619 |
+
"epoch": 1.20536149471974,
|
| 2620 |
+
"grad_norm": 0.19054022431373596,
|
| 2621 |
+
"learning_rate": 1.866367285260312e-05,
|
| 2622 |
+
"loss": 0.0045,
|
| 2623 |
+
"step": 3710
|
| 2624 |
+
},
|
| 2625 |
+
{
|
| 2626 |
+
"epoch": 1.2086108854589765,
|
| 2627 |
+
"grad_norm": 0.38206177949905396,
|
| 2628 |
+
"learning_rate": 1.853384816886962e-05,
|
| 2629 |
+
"loss": 0.0056,
|
| 2630 |
+
"step": 3720
|
| 2631 |
+
},
|
| 2632 |
+
{
|
| 2633 |
+
"epoch": 1.211860276198213,
|
| 2634 |
+
"grad_norm": 0.34758618474006653,
|
| 2635 |
+
"learning_rate": 1.840421009296975e-05,
|
| 2636 |
+
"loss": 0.0034,
|
| 2637 |
+
"step": 3730
|
| 2638 |
+
},
|
| 2639 |
+
{
|
| 2640 |
+
"epoch": 1.2151096669374493,
|
| 2641 |
+
"grad_norm": 0.3336513638496399,
|
| 2642 |
+
"learning_rate": 1.827476236615194e-05,
|
| 2643 |
+
"loss": 0.0051,
|
| 2644 |
+
"step": 3740
|
| 2645 |
+
},
|
| 2646 |
+
{
|
| 2647 |
+
"epoch": 1.2183590576766856,
|
| 2648 |
+
"grad_norm": 0.6265475153923035,
|
| 2649 |
+
"learning_rate": 1.8145508724171316e-05,
|
| 2650 |
+
"loss": 0.0032,
|
| 2651 |
+
"step": 3750
|
| 2652 |
+
},
|
| 2653 |
+
{
|
| 2654 |
+
"epoch": 1.221608448415922,
|
| 2655 |
+
"grad_norm": 0.4233214259147644,
|
| 2656 |
+
"learning_rate": 1.80164528971819e-05,
|
| 2657 |
+
"loss": 0.0063,
|
| 2658 |
+
"step": 3760
|
| 2659 |
+
},
|
| 2660 |
+
{
|
| 2661 |
+
"epoch": 1.2248578391551583,
|
| 2662 |
+
"grad_norm": 0.2057565301656723,
|
| 2663 |
+
"learning_rate": 1.7887598609628897e-05,
|
| 2664 |
+
"loss": 0.0037,
|
| 2665 |
+
"step": 3770
|
| 2666 |
+
},
|
| 2667 |
+
{
|
| 2668 |
+
"epoch": 1.2281072298943947,
|
| 2669 |
+
"grad_norm": 0.2737014889717102,
|
| 2670 |
+
"learning_rate": 1.7758949580141276e-05,
|
| 2671 |
+
"loss": 0.0047,
|
| 2672 |
+
"step": 3780
|
| 2673 |
+
},
|
| 2674 |
+
{
|
| 2675 |
+
"epoch": 1.2313566206336313,
|
| 2676 |
+
"grad_norm": 0.38340964913368225,
|
| 2677 |
+
"learning_rate": 1.7630509521424407e-05,
|
| 2678 |
+
"loss": 0.0049,
|
| 2679 |
+
"step": 3790
|
| 2680 |
+
},
|
| 2681 |
+
{
|
| 2682 |
+
"epoch": 1.2346060113728676,
|
| 2683 |
+
"grad_norm": 0.3404456377029419,
|
| 2684 |
+
"learning_rate": 1.750228214015295e-05,
|
| 2685 |
+
"loss": 0.0034,
|
| 2686 |
+
"step": 3800
|
| 2687 |
+
},
|
| 2688 |
+
{
|
| 2689 |
+
"epoch": 1.237855402112104,
|
| 2690 |
+
"grad_norm": 0.40690556168556213,
|
| 2691 |
+
"learning_rate": 1.7374271136863863e-05,
|
| 2692 |
+
"loss": 0.0042,
|
| 2693 |
+
"step": 3810
|
| 2694 |
+
},
|
| 2695 |
+
{
|
| 2696 |
+
"epoch": 1.2411047928513403,
|
| 2697 |
+
"grad_norm": 0.22202616930007935,
|
| 2698 |
+
"learning_rate": 1.7246480205849613e-05,
|
| 2699 |
+
"loss": 0.0046,
|
| 2700 |
+
"step": 3820
|
| 2701 |
+
},
|
| 2702 |
+
{
|
| 2703 |
+
"epoch": 1.2443541835905767,
|
| 2704 |
+
"grad_norm": 0.37885475158691406,
|
| 2705 |
+
"learning_rate": 1.7118913035051564e-05,
|
| 2706 |
+
"loss": 0.004,
|
| 2707 |
+
"step": 3830
|
| 2708 |
+
},
|
| 2709 |
+
{
|
| 2710 |
+
"epoch": 1.2476035743298133,
|
| 2711 |
+
"grad_norm": 0.16762322187423706,
|
| 2712 |
+
"learning_rate": 1.6991573305953533e-05,
|
| 2713 |
+
"loss": 0.0034,
|
| 2714 |
+
"step": 3840
|
| 2715 |
+
},
|
| 2716 |
+
{
|
| 2717 |
+
"epoch": 1.2508529650690496,
|
| 2718 |
+
"grad_norm": 0.5190407633781433,
|
| 2719 |
+
"learning_rate": 1.686446469347558e-05,
|
| 2720 |
+
"loss": 0.0042,
|
| 2721 |
+
"step": 3850
|
| 2722 |
+
},
|
| 2723 |
+
{
|
| 2724 |
+
"epoch": 1.254102355808286,
|
| 2725 |
+
"grad_norm": 0.290955126285553,
|
| 2726 |
+
"learning_rate": 1.6737590865867907e-05,
|
| 2727 |
+
"loss": 0.0056,
|
| 2728 |
+
"step": 3860
|
| 2729 |
+
},
|
| 2730 |
+
{
|
| 2731 |
+
"epoch": 1.2573517465475224,
|
| 2732 |
+
"grad_norm": 0.240234836935997,
|
| 2733 |
+
"learning_rate": 1.6610955484605023e-05,
|
| 2734 |
+
"loss": 0.0034,
|
| 2735 |
+
"step": 3870
|
| 2736 |
+
},
|
| 2737 |
+
{
|
| 2738 |
+
"epoch": 1.2606011372867587,
|
| 2739 |
+
"grad_norm": 0.29844263195991516,
|
| 2740 |
+
"learning_rate": 1.6484562204280075e-05,
|
| 2741 |
+
"loss": 0.0038,
|
| 2742 |
+
"step": 3880
|
| 2743 |
+
},
|
| 2744 |
+
{
|
| 2745 |
+
"epoch": 1.263850528025995,
|
| 2746 |
+
"grad_norm": 0.3929229974746704,
|
| 2747 |
+
"learning_rate": 1.6358414672499377e-05,
|
| 2748 |
+
"loss": 0.0051,
|
| 2749 |
+
"step": 3890
|
| 2750 |
+
},
|
| 2751 |
+
{
|
| 2752 |
+
"epoch": 1.2670999187652314,
|
| 2753 |
+
"grad_norm": 0.3525027632713318,
|
| 2754 |
+
"learning_rate": 1.623251652977713e-05,
|
| 2755 |
+
"loss": 0.0037,
|
| 2756 |
+
"step": 3900
|
| 2757 |
+
},
|
| 2758 |
+
{
|
| 2759 |
+
"epoch": 1.2703493095044678,
|
| 2760 |
+
"grad_norm": 0.20320917665958405,
|
| 2761 |
+
"learning_rate": 1.6106871409430387e-05,
|
| 2762 |
+
"loss": 0.0044,
|
| 2763 |
+
"step": 3910
|
| 2764 |
+
},
|
| 2765 |
+
{
|
| 2766 |
+
"epoch": 1.2735987002437044,
|
| 2767 |
+
"grad_norm": 0.1730404794216156,
|
| 2768 |
+
"learning_rate": 1.5981482937474172e-05,
|
| 2769 |
+
"loss": 0.0039,
|
| 2770 |
+
"step": 3920
|
| 2771 |
+
},
|
| 2772 |
+
{
|
| 2773 |
+
"epoch": 1.2768480909829407,
|
| 2774 |
+
"grad_norm": 0.3508148789405823,
|
| 2775 |
+
"learning_rate": 1.5856354732516865e-05,
|
| 2776 |
+
"loss": 0.0041,
|
| 2777 |
+
"step": 3930
|
| 2778 |
+
},
|
| 2779 |
+
{
|
| 2780 |
+
"epoch": 1.280097481722177,
|
| 2781 |
+
"grad_norm": 0.35546788573265076,
|
| 2782 |
+
"learning_rate": 1.573149040565572e-05,
|
| 2783 |
+
"loss": 0.0043,
|
| 2784 |
+
"step": 3940
|
| 2785 |
+
},
|
| 2786 |
+
{
|
| 2787 |
+
"epoch": 1.2833468724614134,
|
| 2788 |
+
"grad_norm": 0.2085341513156891,
|
| 2789 |
+
"learning_rate": 1.5606893560372714e-05,
|
| 2790 |
+
"loss": 0.0043,
|
| 2791 |
+
"step": 3950
|
| 2792 |
+
},
|
| 2793 |
+
{
|
| 2794 |
+
"epoch": 1.28659626320065,
|
| 2795 |
+
"grad_norm": 0.18922321498394012,
|
| 2796 |
+
"learning_rate": 1.548256779243052e-05,
|
| 2797 |
+
"loss": 0.0037,
|
| 2798 |
+
"step": 3960
|
| 2799 |
+
},
|
| 2800 |
+
{
|
| 2801 |
+
"epoch": 1.2898456539398864,
|
| 2802 |
+
"grad_norm": 0.3243728280067444,
|
| 2803 |
+
"learning_rate": 1.5358516689768734e-05,
|
| 2804 |
+
"loss": 0.0044,
|
| 2805 |
+
"step": 3970
|
| 2806 |
+
},
|
| 2807 |
+
{
|
| 2808 |
+
"epoch": 1.2930950446791227,
|
| 2809 |
+
"grad_norm": 0.37964773178100586,
|
| 2810 |
+
"learning_rate": 1.5234743832400344e-05,
|
| 2811 |
+
"loss": 0.0031,
|
| 2812 |
+
"step": 3980
|
| 2813 |
+
},
|
| 2814 |
+
{
|
| 2815 |
+
"epoch": 1.296344435418359,
|
| 2816 |
+
"grad_norm": 0.3203120529651642,
|
| 2817 |
+
"learning_rate": 1.5111252792308406e-05,
|
| 2818 |
+
"loss": 0.0031,
|
| 2819 |
+
"step": 3990
|
| 2820 |
+
},
|
| 2821 |
+
{
|
| 2822 |
+
"epoch": 1.2995938261575954,
|
| 2823 |
+
"grad_norm": 0.1457476019859314,
|
| 2824 |
+
"learning_rate": 1.4988047133342964e-05,
|
| 2825 |
+
"loss": 0.0042,
|
| 2826 |
+
"step": 4000
|
| 2827 |
+
},
|
| 2828 |
+
{
|
| 2829 |
+
"epoch": 1.3028432168968318,
|
| 2830 |
+
"grad_norm": 0.14640583097934723,
|
| 2831 |
+
"learning_rate": 1.486513041111819e-05,
|
| 2832 |
+
"loss": 0.003,
|
| 2833 |
+
"step": 4010
|
| 2834 |
+
},
|
| 2835 |
+
{
|
| 2836 |
+
"epoch": 1.3060926076360682,
|
| 2837 |
+
"grad_norm": 0.5449049472808838,
|
| 2838 |
+
"learning_rate": 1.4742506172909775e-05,
|
| 2839 |
+
"loss": 0.0033,
|
| 2840 |
+
"step": 4020
|
| 2841 |
+
},
|
| 2842 |
+
{
|
| 2843 |
+
"epoch": 1.3093419983753045,
|
| 2844 |
+
"grad_norm": 0.40858981013298035,
|
| 2845 |
+
"learning_rate": 1.4620177957552578e-05,
|
| 2846 |
+
"loss": 0.0038,
|
| 2847 |
+
"step": 4030
|
| 2848 |
+
},
|
| 2849 |
+
{
|
| 2850 |
+
"epoch": 1.312591389114541,
|
| 2851 |
+
"grad_norm": 0.5728728175163269,
|
| 2852 |
+
"learning_rate": 1.4498149295338464e-05,
|
| 2853 |
+
"loss": 0.0031,
|
| 2854 |
+
"step": 4040
|
| 2855 |
+
},
|
| 2856 |
+
{
|
| 2857 |
+
"epoch": 1.3158407798537775,
|
| 2858 |
+
"grad_norm": 0.24417610466480255,
|
| 2859 |
+
"learning_rate": 1.437642370791446e-05,
|
| 2860 |
+
"loss": 0.0034,
|
| 2861 |
+
"step": 4050
|
| 2862 |
+
},
|
| 2863 |
+
{
|
| 2864 |
+
"epoch": 1.3190901705930138,
|
| 2865 |
+
"grad_norm": 0.317682683467865,
|
| 2866 |
+
"learning_rate": 1.4255004708181075e-05,
|
| 2867 |
+
"loss": 0.0028,
|
| 2868 |
+
"step": 4060
|
| 2869 |
+
},
|
| 2870 |
+
{
|
| 2871 |
+
"epoch": 1.3223395613322502,
|
| 2872 |
+
"grad_norm": 0.6588785648345947,
|
| 2873 |
+
"learning_rate": 1.4133895800190983e-05,
|
| 2874 |
+
"loss": 0.004,
|
| 2875 |
+
"step": 4070
|
| 2876 |
+
},
|
| 2877 |
+
{
|
| 2878 |
+
"epoch": 1.3255889520714865,
|
| 2879 |
+
"grad_norm": 0.44252049922943115,
|
| 2880 |
+
"learning_rate": 1.4013100479047825e-05,
|
| 2881 |
+
"loss": 0.0034,
|
| 2882 |
+
"step": 4080
|
| 2883 |
+
},
|
| 2884 |
+
{
|
| 2885 |
+
"epoch": 1.328838342810723,
|
| 2886 |
+
"grad_norm": 0.4248214066028595,
|
| 2887 |
+
"learning_rate": 1.3892622230805436e-05,
|
| 2888 |
+
"loss": 0.005,
|
| 2889 |
+
"step": 4090
|
| 2890 |
+
},
|
| 2891 |
+
{
|
| 2892 |
+
"epoch": 1.3320877335499595,
|
| 2893 |
+
"grad_norm": 0.31828951835632324,
|
| 2894 |
+
"learning_rate": 1.3772464532367125e-05,
|
| 2895 |
+
"loss": 0.0026,
|
| 2896 |
+
"step": 4100
|
| 2897 |
+
},
|
| 2898 |
+
{
|
| 2899 |
+
"epoch": 1.3353371242891958,
|
| 2900 |
+
"grad_norm": 0.25999122858047485,
|
| 2901 |
+
"learning_rate": 1.3652630851385454e-05,
|
| 2902 |
+
"loss": 0.0035,
|
| 2903 |
+
"step": 4110
|
| 2904 |
+
},
|
| 2905 |
+
{
|
| 2906 |
+
"epoch": 1.3385865150284322,
|
| 2907 |
+
"grad_norm": 0.15871325135231018,
|
| 2908 |
+
"learning_rate": 1.353312464616207e-05,
|
| 2909 |
+
"loss": 0.0035,
|
| 2910 |
+
"step": 4120
|
| 2911 |
+
},
|
| 2912 |
+
{
|
| 2913 |
+
"epoch": 1.3418359057676685,
|
| 2914 |
+
"grad_norm": 0.44427499175071716,
|
| 2915 |
+
"learning_rate": 1.341394936554794e-05,
|
| 2916 |
+
"loss": 0.0028,
|
| 2917 |
+
"step": 4130
|
| 2918 |
+
},
|
| 2919 |
+
{
|
| 2920 |
+
"epoch": 1.345085296506905,
|
| 2921 |
+
"grad_norm": 0.306194543838501,
|
| 2922 |
+
"learning_rate": 1.329510844884385e-05,
|
| 2923 |
+
"loss": 0.0026,
|
| 2924 |
+
"step": 4140
|
| 2925 |
+
},
|
| 2926 |
+
{
|
| 2927 |
+
"epoch": 1.3483346872461413,
|
| 2928 |
+
"grad_norm": 0.19543297588825226,
|
| 2929 |
+
"learning_rate": 1.3176605325701086e-05,
|
| 2930 |
+
"loss": 0.0028,
|
| 2931 |
+
"step": 4150
|
| 2932 |
+
},
|
| 2933 |
+
{
|
| 2934 |
+
"epoch": 1.3515840779853776,
|
| 2935 |
+
"grad_norm": 0.21941304206848145,
|
| 2936 |
+
"learning_rate": 1.305844341602249e-05,
|
| 2937 |
+
"loss": 0.0018,
|
| 2938 |
+
"step": 4160
|
| 2939 |
+
},
|
| 2940 |
+
{
|
| 2941 |
+
"epoch": 1.3548334687246142,
|
| 2942 |
+
"grad_norm": 0.36652112007141113,
|
| 2943 |
+
"learning_rate": 1.2940626129863792e-05,
|
| 2944 |
+
"loss": 0.003,
|
| 2945 |
+
"step": 4170
|
| 2946 |
+
},
|
| 2947 |
+
{
|
| 2948 |
+
"epoch": 1.3580828594638505,
|
| 2949 |
+
"grad_norm": 0.47468826174736023,
|
| 2950 |
+
"learning_rate": 1.282315686733514e-05,
|
| 2951 |
+
"loss": 0.0025,
|
| 2952 |
+
"step": 4180
|
| 2953 |
+
},
|
| 2954 |
+
{
|
| 2955 |
+
"epoch": 1.361332250203087,
|
| 2956 |
+
"grad_norm": 0.3629470765590668,
|
| 2957 |
+
"learning_rate": 1.2706039018503013e-05,
|
| 2958 |
+
"loss": 0.003,
|
| 2959 |
+
"step": 4190
|
| 2960 |
+
},
|
| 2961 |
+
{
|
| 2962 |
+
"epoch": 1.3645816409423233,
|
| 2963 |
+
"grad_norm": 0.3716108500957489,
|
| 2964 |
+
"learning_rate": 1.2589275963292397e-05,
|
| 2965 |
+
"loss": 0.0027,
|
| 2966 |
+
"step": 4200
|
| 2967 |
+
},
|
| 2968 |
+
{
|
| 2969 |
+
"epoch": 1.3678310316815598,
|
| 2970 |
+
"grad_norm": 0.8420029878616333,
|
| 2971 |
+
"learning_rate": 1.2472871071389205e-05,
|
| 2972 |
+
"loss": 0.0027,
|
| 2973 |
+
"step": 4210
|
| 2974 |
+
},
|
| 2975 |
+
{
|
| 2976 |
+
"epoch": 1.3710804224207962,
|
| 2977 |
+
"grad_norm": 0.43932396173477173,
|
| 2978 |
+
"learning_rate": 1.2356827702143048e-05,
|
| 2979 |
+
"loss": 0.003,
|
| 2980 |
+
"step": 4220
|
| 2981 |
+
},
|
| 2982 |
+
{
|
| 2983 |
+
"epoch": 1.3743298131600326,
|
| 2984 |
+
"grad_norm": 0.28611013293266296,
|
| 2985 |
+
"learning_rate": 1.2241149204470314e-05,
|
| 2986 |
+
"loss": 0.0034,
|
| 2987 |
+
"step": 4230
|
| 2988 |
+
},
|
| 2989 |
+
{
|
| 2990 |
+
"epoch": 1.377579203899269,
|
| 2991 |
+
"grad_norm": 0.30783408880233765,
|
| 2992 |
+
"learning_rate": 1.2125838916757471e-05,
|
| 2993 |
+
"loss": 0.0034,
|
| 2994 |
+
"step": 4240
|
| 2995 |
+
},
|
| 2996 |
+
{
|
| 2997 |
+
"epoch": 1.3808285946385053,
|
| 2998 |
+
"grad_norm": 0.16370636224746704,
|
| 2999 |
+
"learning_rate": 1.2010900166764774e-05,
|
| 3000 |
+
"loss": 0.0028,
|
| 3001 |
+
"step": 4250
|
| 3002 |
+
},
|
| 3003 |
+
{
|
| 3004 |
+
"epoch": 1.3840779853777416,
|
| 3005 |
+
"grad_norm": 0.055241186171770096,
|
| 3006 |
+
"learning_rate": 1.1896336271530187e-05,
|
| 3007 |
+
"loss": 0.0035,
|
| 3008 |
+
"step": 4260
|
| 3009 |
+
},
|
| 3010 |
+
{
|
| 3011 |
+
"epoch": 1.387327376116978,
|
| 3012 |
+
"grad_norm": 0.42322880029678345,
|
| 3013 |
+
"learning_rate": 1.1782150537273665e-05,
|
| 3014 |
+
"loss": 0.0029,
|
| 3015 |
+
"step": 4270
|
| 3016 |
+
},
|
| 3017 |
+
{
|
| 3018 |
+
"epoch": 1.3905767668562143,
|
| 3019 |
+
"grad_norm": 0.12644466757774353,
|
| 3020 |
+
"learning_rate": 1.166834625930178e-05,
|
| 3021 |
+
"loss": 0.0022,
|
| 3022 |
+
"step": 4280
|
| 3023 |
+
},
|
| 3024 |
+
{
|
| 3025 |
+
"epoch": 1.393826157595451,
|
| 3026 |
+
"grad_norm": 0.24443909525871277,
|
| 3027 |
+
"learning_rate": 1.1554926721912562e-05,
|
| 3028 |
+
"loss": 0.0022,
|
| 3029 |
+
"step": 4290
|
| 3030 |
+
},
|
| 3031 |
+
{
|
| 3032 |
+
"epoch": 1.3970755483346873,
|
| 3033 |
+
"grad_norm": 0.12778101861476898,
|
| 3034 |
+
"learning_rate": 1.144189519830074e-05,
|
| 3035 |
+
"loss": 0.003,
|
| 3036 |
+
"step": 4300
|
| 3037 |
+
},
|
| 3038 |
+
{
|
| 3039 |
+
"epoch": 1.4003249390739236,
|
| 3040 |
+
"grad_norm": 0.15853983163833618,
|
| 3041 |
+
"learning_rate": 1.1329254950463315e-05,
|
| 3042 |
+
"loss": 0.0023,
|
| 3043 |
+
"step": 4310
|
| 3044 |
+
},
|
| 3045 |
+
{
|
| 3046 |
+
"epoch": 1.40357432981316,
|
| 3047 |
+
"grad_norm": 0.11951529234647751,
|
| 3048 |
+
"learning_rate": 1.1217009229105357e-05,
|
| 3049 |
+
"loss": 0.0031,
|
| 3050 |
+
"step": 4320
|
| 3051 |
+
},
|
| 3052 |
+
{
|
| 3053 |
+
"epoch": 1.4068237205523964,
|
| 3054 |
+
"grad_norm": 0.21437332034111023,
|
| 3055 |
+
"learning_rate": 1.1105161273546236e-05,
|
| 3056 |
+
"loss": 0.0019,
|
| 3057 |
+
"step": 4330
|
| 3058 |
+
},
|
| 3059 |
+
{
|
| 3060 |
+
"epoch": 1.410073111291633,
|
| 3061 |
+
"grad_norm": 0.8929743766784668,
|
| 3062 |
+
"learning_rate": 1.0993714311626146e-05,
|
| 3063 |
+
"loss": 0.0024,
|
| 3064 |
+
"step": 4340
|
| 3065 |
+
},
|
| 3066 |
+
{
|
| 3067 |
+
"epoch": 1.4133225020308693,
|
| 3068 |
+
"grad_norm": 0.2045769989490509,
|
| 3069 |
+
"learning_rate": 1.0882671559612909e-05,
|
| 3070 |
+
"loss": 0.0029,
|
| 3071 |
+
"step": 4350
|
| 3072 |
+
},
|
| 3073 |
+
{
|
| 3074 |
+
"epoch": 1.4165718927701056,
|
| 3075 |
+
"grad_norm": 0.08286549896001816,
|
| 3076 |
+
"learning_rate": 1.0772036222109182e-05,
|
| 3077 |
+
"loss": 0.0029,
|
| 3078 |
+
"step": 4360
|
| 3079 |
+
},
|
| 3080 |
+
{
|
| 3081 |
+
"epoch": 1.419821283509342,
|
| 3082 |
+
"grad_norm": 0.09282595664262772,
|
| 3083 |
+
"learning_rate": 1.066181149196e-05,
|
| 3084 |
+
"loss": 0.0018,
|
| 3085 |
+
"step": 4370
|
| 3086 |
+
},
|
| 3087 |
+
{
|
| 3088 |
+
"epoch": 1.4230706742485784,
|
| 3089 |
+
"grad_norm": 0.2858879566192627,
|
| 3090 |
+
"learning_rate": 1.055200055016057e-05,
|
| 3091 |
+
"loss": 0.0029,
|
| 3092 |
+
"step": 4380
|
| 3093 |
+
},
|
| 3094 |
+
{
|
| 3095 |
+
"epoch": 1.4263200649878147,
|
| 3096 |
+
"grad_norm": 0.2344265580177307,
|
| 3097 |
+
"learning_rate": 1.0442606565764534e-05,
|
| 3098 |
+
"loss": 0.0019,
|
| 3099 |
+
"step": 4390
|
| 3100 |
+
},
|
| 3101 |
+
{
|
| 3102 |
+
"epoch": 1.429569455727051,
|
| 3103 |
+
"grad_norm": 0.346629798412323,
|
| 3104 |
+
"learning_rate": 1.0333632695792492e-05,
|
| 3105 |
+
"loss": 0.0025,
|
| 3106 |
+
"step": 4400
|
| 3107 |
+
},
|
| 3108 |
+
{
|
| 3109 |
+
"epoch": 1.4328188464662874,
|
| 3110 |
+
"grad_norm": 0.29290953278541565,
|
| 3111 |
+
"learning_rate": 1.0225082085140856e-05,
|
| 3112 |
+
"loss": 0.0022,
|
| 3113 |
+
"step": 4410
|
| 3114 |
+
},
|
| 3115 |
+
{
|
| 3116 |
+
"epoch": 1.436068237205524,
|
| 3117 |
+
"grad_norm": 0.43641456961631775,
|
| 3118 |
+
"learning_rate": 1.0116957866491128e-05,
|
| 3119 |
+
"loss": 0.0021,
|
| 3120 |
+
"step": 4420
|
| 3121 |
+
},
|
| 3122 |
+
{
|
| 3123 |
+
"epoch": 1.4393176279447604,
|
| 3124 |
+
"grad_norm": 0.3562842309474945,
|
| 3125 |
+
"learning_rate": 1.000926316021952e-05,
|
| 3126 |
+
"loss": 0.0022,
|
| 3127 |
+
"step": 4430
|
| 3128 |
+
},
|
| 3129 |
+
{
|
| 3130 |
+
"epoch": 1.4425670186839967,
|
| 3131 |
+
"grad_norm": 0.4262392520904541,
|
| 3132 |
+
"learning_rate": 9.902001074306835e-06,
|
| 3133 |
+
"loss": 0.0028,
|
| 3134 |
+
"step": 4440
|
| 3135 |
+
},
|
| 3136 |
+
{
|
| 3137 |
+
"epoch": 1.445816409423233,
|
| 3138 |
+
"grad_norm": 0.2457405924797058,
|
| 3139 |
+
"learning_rate": 9.795174704248808e-06,
|
| 3140 |
+
"loss": 0.0025,
|
| 3141 |
+
"step": 4450
|
| 3142 |
+
},
|
| 3143 |
+
{
|
| 3144 |
+
"epoch": 1.4490658001624697,
|
| 3145 |
+
"grad_norm": 0.22532138228416443,
|
| 3146 |
+
"learning_rate": 9.6887871329668e-06,
|
| 3147 |
+
"loss": 0.0019,
|
| 3148 |
+
"step": 4460
|
| 3149 |
+
},
|
| 3150 |
+
{
|
| 3151 |
+
"epoch": 1.452315190901706,
|
| 3152 |
+
"grad_norm": 0.4915858507156372,
|
| 3153 |
+
"learning_rate": 9.582841430718767e-06,
|
| 3154 |
+
"loss": 0.0023,
|
| 3155 |
+
"step": 4470
|
| 3156 |
+
},
|
| 3157 |
+
{
|
| 3158 |
+
"epoch": 1.4555645816409424,
|
| 3159 |
+
"grad_norm": 0.18646268546581268,
|
| 3160 |
+
"learning_rate": 9.477340655010716e-06,
|
| 3161 |
+
"loss": 0.0021,
|
| 3162 |
+
"step": 4480
|
| 3163 |
+
},
|
| 3164 |
+
{
|
| 3165 |
+
"epoch": 1.4588139723801787,
|
| 3166 |
+
"grad_norm": 0.10206873714923859,
|
| 3167 |
+
"learning_rate": 9.372287850508421e-06,
|
| 3168 |
+
"loss": 0.0017,
|
| 3169 |
+
"step": 4490
|
| 3170 |
+
},
|
| 3171 |
+
{
|
| 3172 |
+
"epoch": 1.462063363119415,
|
| 3173 |
+
"grad_norm": 0.3298508822917938,
|
| 3174 |
+
"learning_rate": 9.267686048949568e-06,
|
| 3175 |
+
"loss": 0.0018,
|
| 3176 |
+
"step": 4500
|
| 3177 |
+
},
|
| 3178 |
+
{
|
| 3179 |
+
"epoch": 1.4653127538586515,
|
| 3180 |
+
"grad_norm": 0.5361196398735046,
|
| 3181 |
+
"learning_rate": 9.163538269056296e-06,
|
| 3182 |
+
"loss": 0.003,
|
| 3183 |
+
"step": 4510
|
| 3184 |
+
},
|
| 3185 |
+
{
|
| 3186 |
+
"epoch": 1.4685621445978878,
|
| 3187 |
+
"grad_norm": 0.41454723477363586,
|
| 3188 |
+
"learning_rate": 9.05984751644803e-06,
|
| 3189 |
+
"loss": 0.0019,
|
| 3190 |
+
"step": 4520
|
| 3191 |
+
},
|
| 3192 |
+
{
|
| 3193 |
+
"epoch": 1.4718115353371242,
|
| 3194 |
+
"grad_norm": 0.47003769874572754,
|
| 3195 |
+
"learning_rate": 8.956616783554759e-06,
|
| 3196 |
+
"loss": 0.0025,
|
| 3197 |
+
"step": 4530
|
| 3198 |
+
},
|
| 3199 |
+
{
|
| 3200 |
+
"epoch": 1.4750609260763607,
|
| 3201 |
+
"grad_norm": 0.2163703888654709,
|
| 3202 |
+
"learning_rate": 8.853849049530703e-06,
|
| 3203 |
+
"loss": 0.0018,
|
| 3204 |
+
"step": 4540
|
| 3205 |
+
},
|
| 3206 |
+
{
|
| 3207 |
+
"epoch": 1.478310316815597,
|
| 3208 |
+
"grad_norm": 0.08466655015945435,
|
| 3209 |
+
"learning_rate": 8.751547280168297e-06,
|
| 3210 |
+
"loss": 0.0021,
|
| 3211 |
+
"step": 4550
|
| 3212 |
+
},
|
| 3213 |
+
{
|
| 3214 |
+
"epoch": 1.4815597075548335,
|
| 3215 |
+
"grad_norm": 0.2865130305290222,
|
| 3216 |
+
"learning_rate": 8.649714427812607e-06,
|
| 3217 |
+
"loss": 0.0017,
|
| 3218 |
+
"step": 4560
|
| 3219 |
+
},
|
| 3220 |
+
{
|
| 3221 |
+
"epoch": 1.4848090982940698,
|
| 3222 |
+
"grad_norm": 0.2946512699127197,
|
| 3223 |
+
"learning_rate": 8.548353431276182e-06,
|
| 3224 |
+
"loss": 0.0019,
|
| 3225 |
+
"step": 4570
|
| 3226 |
+
},
|
| 3227 |
+
{
|
| 3228 |
+
"epoch": 1.4880584890333062,
|
| 3229 |
+
"grad_norm": 0.4205271303653717,
|
| 3230 |
+
"learning_rate": 8.447467215754157e-06,
|
| 3231 |
+
"loss": 0.0021,
|
| 3232 |
+
"step": 4580
|
| 3233 |
+
},
|
| 3234 |
+
{
|
| 3235 |
+
"epoch": 1.4903330625507718,
|
| 3236 |
+
"eval_loss": 0.20176434516906738,
|
| 3237 |
+
"eval_runtime": 733.5854,
|
| 3238 |
+
"eval_samples_per_second": 3.817,
|
| 3239 |
+
"eval_steps_per_second": 3.817,
|
| 3240 |
+
"step": 4587
|
| 3241 |
}
|
| 3242 |
],
|
| 3243 |
"logging_steps": 10,
|
|
|
|
| 3252 |
"early_stopping_threshold": 0.0
|
| 3253 |
},
|
| 3254 |
"attributes": {
|
| 3255 |
+
"early_stopping_patience_counter": 2
|
| 3256 |
}
|
| 3257 |
},
|
| 3258 |
"TrainerControl": {
|
|
|
|
| 3261 |
"should_evaluate": false,
|
| 3262 |
"should_log": false,
|
| 3263 |
"should_save": true,
|
| 3264 |
+
"should_training_stop": true
|
| 3265 |
},
|
| 3266 |
"attributes": {}
|
| 3267 |
}
|
| 3268 |
},
|
| 3269 |
+
"total_flos": 3.5805200990465556e+18,
|
| 3270 |
"train_batch_size": 1,
|
| 3271 |
"trial_name": null,
|
| 3272 |
"trial_params": null
|