Upload folder using huggingface_hub
Browse files- adapter_config.json +4 -4
- adapter_model.safetensors +1 -1
- optimizer.pt +2 -2
- rng_state.pth +1 -1
- scaler.pt +1 -1
- scheduler.pt +1 -1
- trainer_state.json +929 -5
- training_args.bin +1 -1
adapter_config.json
CHANGED
|
@@ -29,13 +29,13 @@
|
|
| 29 |
"rank_pattern": {},
|
| 30 |
"revision": null,
|
| 31 |
"target_modules": [
|
| 32 |
-
"down_proj",
|
| 33 |
-
"up_proj",
|
| 34 |
"gate_proj",
|
| 35 |
-
"v_proj",
|
| 36 |
"o_proj",
|
|
|
|
|
|
|
| 37 |
"q_proj",
|
| 38 |
-
"k_proj"
|
|
|
|
| 39 |
],
|
| 40 |
"task_type": "CAUSAL_LM",
|
| 41 |
"trainable_token_indices": null,
|
|
|
|
| 29 |
"rank_pattern": {},
|
| 30 |
"revision": null,
|
| 31 |
"target_modules": [
|
|
|
|
|
|
|
| 32 |
"gate_proj",
|
|
|
|
| 33 |
"o_proj",
|
| 34 |
+
"down_proj",
|
| 35 |
+
"up_proj",
|
| 36 |
"q_proj",
|
| 37 |
+
"k_proj",
|
| 38 |
+
"v_proj"
|
| 39 |
],
|
| 40 |
"task_type": "CAUSAL_LM",
|
| 41 |
"trainable_token_indices": null,
|
adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 264308896
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:43fd707af932d69932231314b4398195f40859ba5286a2a0d59c081c87ccb77b
|
| 3 |
size 264308896
|
optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:843040dcad4f852768a06d9e594c999813eac472c3f65f7443969358b67099aa
|
| 3 |
+
size 136089907
|
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:c12bd7452a7c59fb850dbff4b78e1bad4cd5c1923c2b1c3ec3478be5c61e7f77
|
| 3 |
size 14645
|
scaler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1383
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:fc1d69bd3f86e65d28af2330480e9d8542f90d84e33fa16c1fd5116fa1a5b336
|
| 3 |
size 1383
|
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 sha256:6025d642d364613419905274d3700452dbaf1fd69dfd7c5f8f57368bc61dfbbe
|
| 3 |
size 1465
|
trainer_state.json
CHANGED
|
@@ -2,9 +2,9 @@
|
|
| 2 |
"best_global_step": null,
|
| 3 |
"best_metric": null,
|
| 4 |
"best_model_checkpoint": null,
|
| 5 |
-
"epoch": 0.
|
| 6 |
"eval_steps": 500,
|
| 7 |
-
"global_step":
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
@@ -64,11 +64,935 @@
|
|
| 64 |
"rewards/match_format_exactly/mean": 1.0,
|
| 65 |
"rewards/match_format_exactly/std": 0.0,
|
| 66 |
"step": 20
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 67 |
}
|
| 68 |
],
|
| 69 |
"logging_steps": 10,
|
| 70 |
-
"max_steps":
|
| 71 |
-
"num_input_tokens_seen":
|
| 72 |
"num_train_epochs": 1,
|
| 73 |
"save_steps": 10,
|
| 74 |
"stateful_callbacks": {
|
|
@@ -78,7 +1002,7 @@
|
|
| 78 |
"should_evaluate": false,
|
| 79 |
"should_log": false,
|
| 80 |
"should_save": true,
|
| 81 |
-
"should_training_stop":
|
| 82 |
},
|
| 83 |
"attributes": {}
|
| 84 |
}
|
|
|
|
| 2 |
"best_global_step": null,
|
| 3 |
"best_metric": null,
|
| 4 |
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 0.02406821620134782,
|
| 6 |
"eval_steps": 500,
|
| 7 |
+
"global_step": 350,
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
|
|
| 64 |
"rewards/match_format_exactly/mean": 1.0,
|
| 65 |
"rewards/match_format_exactly/std": 0.0,
|
| 66 |
"step": 20
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"completion_length": 299.05,
|
| 70 |
+
"completions/clipped_ratio": 0.025,
|
| 71 |
+
"completions/max_length": 684.5,
|
| 72 |
+
"completions/max_terminated_length": 582.1,
|
| 73 |
+
"completions/mean_length": 299.05,
|
| 74 |
+
"completions/mean_terminated_length": 268.775,
|
| 75 |
+
"completions/min_length": 12.0,
|
| 76 |
+
"completions/min_terminated_length": 12.0,
|
| 77 |
+
"epoch": 0.0020629899601155273,
|
| 78 |
+
"frac_reward_zero_std": 0.0,
|
| 79 |
+
"grad_norm": 1.6133402585983276,
|
| 80 |
+
"kl": 1.2915641874074937,
|
| 81 |
+
"learning_rate": 4.1428571428571435e-06,
|
| 82 |
+
"loss": 0.0013,
|
| 83 |
+
"num_tokens": 75404.0,
|
| 84 |
+
"reward": 4.5625,
|
| 85 |
+
"reward_std": 0.6107304871082306,
|
| 86 |
+
"rewards/check_coherence/mean": 0.6125,
|
| 87 |
+
"rewards/check_coherence/std": 0.403445702791214,
|
| 88 |
+
"rewards/check_response_quality/mean": 2.2125,
|
| 89 |
+
"rewards/check_response_quality/std": 0.2752987265586853,
|
| 90 |
+
"rewards/match_format_approximately/mean": 0.7375,
|
| 91 |
+
"rewards/match_format_approximately/std": 0.2404700517654419,
|
| 92 |
+
"rewards/match_format_exactly/mean": 1.0,
|
| 93 |
+
"rewards/match_format_exactly/std": 0.0,
|
| 94 |
+
"step": 30
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"completion_length": 347.025,
|
| 98 |
+
"completions/clipped_ratio": 0.075,
|
| 99 |
+
"completions/max_length": 862.6,
|
| 100 |
+
"completions/max_terminated_length": 509.7,
|
| 101 |
+
"completions/mean_length": 347.025,
|
| 102 |
+
"completions/mean_terminated_length": 242.34166870117187,
|
| 103 |
+
"completions/min_length": 58.0,
|
| 104 |
+
"completions/min_terminated_length": 58.0,
|
| 105 |
+
"epoch": 0.0027506532801540367,
|
| 106 |
+
"frac_reward_zero_std": 0.1,
|
| 107 |
+
"grad_norm": 2.190549850463867,
|
| 108 |
+
"kl": 0.8691788390278816,
|
| 109 |
+
"learning_rate": 4.936507936507937e-06,
|
| 110 |
+
"loss": 0.0009,
|
| 111 |
+
"num_tokens": 102089.0,
|
| 112 |
+
"reward": 4.775,
|
| 113 |
+
"reward_std": 0.5918105363845825,
|
| 114 |
+
"rewards/check_coherence/mean": 0.85,
|
| 115 |
+
"rewards/check_coherence/std": 0.345650315284729,
|
| 116 |
+
"rewards/check_response_quality/mean": 2.2,
|
| 117 |
+
"rewards/check_response_quality/std": 0.20347774028778076,
|
| 118 |
+
"rewards/match_format_approximately/mean": 0.725,
|
| 119 |
+
"rewards/match_format_approximately/std": 0.16547005176544188,
|
| 120 |
+
"rewards/match_format_exactly/mean": 1.0,
|
| 121 |
+
"rewards/match_format_exactly/std": 0.0,
|
| 122 |
+
"step": 40
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"completion_length": 278.275,
|
| 126 |
+
"completions/clipped_ratio": 0.025,
|
| 127 |
+
"completions/max_length": 695.3,
|
| 128 |
+
"completions/max_terminated_length": 556.3,
|
| 129 |
+
"completions/mean_length": 278.275,
|
| 130 |
+
"completions/mean_terminated_length": 240.375,
|
| 131 |
+
"completions/min_length": 19.8,
|
| 132 |
+
"completions/min_terminated_length": 19.8,
|
| 133 |
+
"epoch": 0.0034383166001925457,
|
| 134 |
+
"frac_reward_zero_std": 0.0,
|
| 135 |
+
"grad_norm": 1.1459013223648071,
|
| 136 |
+
"kl": 0.7370456486940384,
|
| 137 |
+
"learning_rate": 4.793650793650794e-06,
|
| 138 |
+
"loss": 0.0007,
|
| 139 |
+
"num_tokens": 125192.0,
|
| 140 |
+
"reward": 4.7375,
|
| 141 |
+
"reward_std": 0.9061064124107361,
|
| 142 |
+
"rewards/check_coherence/mean": 0.8625,
|
| 143 |
+
"rewards/check_coherence/std": 0.4861203670501709,
|
| 144 |
+
"rewards/check_response_quality/mean": 2.175,
|
| 145 |
+
"rewards/check_response_quality/std": 0.34289742112159727,
|
| 146 |
+
"rewards/match_format_approximately/mean": 0.725,
|
| 147 |
+
"rewards/match_format_approximately/std": 0.2654700517654419,
|
| 148 |
+
"rewards/match_format_exactly/mean": 0.975,
|
| 149 |
+
"rewards/match_format_exactly/std": 0.05,
|
| 150 |
+
"step": 50
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"completion_length": 127.625,
|
| 154 |
+
"completions/clipped_ratio": 0.0,
|
| 155 |
+
"completions/max_length": 249.7,
|
| 156 |
+
"completions/max_terminated_length": 249.7,
|
| 157 |
+
"completions/mean_length": 127.625,
|
| 158 |
+
"completions/mean_terminated_length": 127.625,
|
| 159 |
+
"completions/min_length": 11.6,
|
| 160 |
+
"completions/min_terminated_length": 11.6,
|
| 161 |
+
"epoch": 0.004125979920231055,
|
| 162 |
+
"frac_reward_zero_std": 0.1,
|
| 163 |
+
"grad_norm": 1.772578477859497,
|
| 164 |
+
"kl": 1.2247508466243744,
|
| 165 |
+
"learning_rate": 4.634920634920635e-06,
|
| 166 |
+
"loss": 0.0012,
|
| 167 |
+
"num_tokens": 140377.0,
|
| 168 |
+
"reward": 5.075,
|
| 169 |
+
"reward_std": 0.583526349067688,
|
| 170 |
+
"rewards/check_coherence/mean": 0.925,
|
| 171 |
+
"rewards/check_coherence/std": 0.4689477920532227,
|
| 172 |
+
"rewards/check_response_quality/mean": 2.325,
|
| 173 |
+
"rewards/check_response_quality/std": 0.17886751294136047,
|
| 174 |
+
"rewards/match_format_approximately/mean": 0.825,
|
| 175 |
+
"rewards/match_format_approximately/std": 0.17886751294136047,
|
| 176 |
+
"rewards/match_format_exactly/mean": 1.0,
|
| 177 |
+
"rewards/match_format_exactly/std": 0.0,
|
| 178 |
+
"step": 60
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"completion_length": 161.4,
|
| 182 |
+
"completions/clipped_ratio": 0.0,
|
| 183 |
+
"completions/max_length": 423.5,
|
| 184 |
+
"completions/max_terminated_length": 423.5,
|
| 185 |
+
"completions/mean_length": 161.4,
|
| 186 |
+
"completions/mean_terminated_length": 161.4,
|
| 187 |
+
"completions/min_length": 17.8,
|
| 188 |
+
"completions/min_terminated_length": 17.8,
|
| 189 |
+
"epoch": 0.004813643240269564,
|
| 190 |
+
"frac_reward_zero_std": 0.0,
|
| 191 |
+
"grad_norm": 3.128584861755371,
|
| 192 |
+
"kl": 0.8261611372232437,
|
| 193 |
+
"learning_rate": 4.476190476190477e-06,
|
| 194 |
+
"loss": 0.0008,
|
| 195 |
+
"num_tokens": 156317.0,
|
| 196 |
+
"reward": 5.1625,
|
| 197 |
+
"reward_std": 0.6853798747062683,
|
| 198 |
+
"rewards/check_coherence/mean": 1.025,
|
| 199 |
+
"rewards/check_coherence/std": 0.4077350258827209,
|
| 200 |
+
"rewards/check_response_quality/mean": 2.3125,
|
| 201 |
+
"rewards/check_response_quality/std": 0.21582483053207396,
|
| 202 |
+
"rewards/match_format_approximately/mean": 0.825,
|
| 203 |
+
"rewards/match_format_approximately/std": 0.2,
|
| 204 |
+
"rewards/match_format_exactly/mean": 1.0,
|
| 205 |
+
"rewards/match_format_exactly/std": 0.0,
|
| 206 |
+
"step": 70
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"completion_length": 120.125,
|
| 210 |
+
"completions/clipped_ratio": 0.0,
|
| 211 |
+
"completions/max_length": 248.3,
|
| 212 |
+
"completions/max_terminated_length": 248.3,
|
| 213 |
+
"completions/mean_length": 120.125,
|
| 214 |
+
"completions/mean_terminated_length": 120.125,
|
| 215 |
+
"completions/min_length": 16.5,
|
| 216 |
+
"completions/min_terminated_length": 16.5,
|
| 217 |
+
"epoch": 0.0055013065603080735,
|
| 218 |
+
"frac_reward_zero_std": 0.4,
|
| 219 |
+
"grad_norm": 2.667182683944702,
|
| 220 |
+
"kl": 0.8017235696315765,
|
| 221 |
+
"learning_rate": 4.317460317460318e-06,
|
| 222 |
+
"loss": 0.0008,
|
| 223 |
+
"num_tokens": 171822.0,
|
| 224 |
+
"reward": 5.4,
|
| 225 |
+
"reward_std": 0.41746232509613035,
|
| 226 |
+
"rewards/check_coherence/mean": 1.2625,
|
| 227 |
+
"rewards/check_coherence/std": 0.2978713572025299,
|
| 228 |
+
"rewards/check_response_quality/mean": 2.3125,
|
| 229 |
+
"rewards/check_response_quality/std": 0.11582483053207397,
|
| 230 |
+
"rewards/match_format_approximately/mean": 0.825,
|
| 231 |
+
"rewards/match_format_approximately/std": 0.1,
|
| 232 |
+
"rewards/match_format_exactly/mean": 1.0,
|
| 233 |
+
"rewards/match_format_exactly/std": 0.0,
|
| 234 |
+
"step": 80
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"completion_length": 164.875,
|
| 238 |
+
"completions/clipped_ratio": 0.0,
|
| 239 |
+
"completions/max_length": 490.9,
|
| 240 |
+
"completions/max_terminated_length": 490.9,
|
| 241 |
+
"completions/mean_length": 164.875,
|
| 242 |
+
"completions/mean_terminated_length": 164.875,
|
| 243 |
+
"completions/min_length": 13.5,
|
| 244 |
+
"completions/min_terminated_length": 13.5,
|
| 245 |
+
"epoch": 0.006188969880346582,
|
| 246 |
+
"frac_reward_zero_std": 0.0,
|
| 247 |
+
"grad_norm": 1.8317680358886719,
|
| 248 |
+
"kl": 0.8454075694084168,
|
| 249 |
+
"learning_rate": 4.158730158730159e-06,
|
| 250 |
+
"loss": 0.0008,
|
| 251 |
+
"num_tokens": 189509.0,
|
| 252 |
+
"reward": 5.325,
|
| 253 |
+
"reward_std": 0.7779198408126831,
|
| 254 |
+
"rewards/check_coherence/mean": 1.15,
|
| 255 |
+
"rewards/check_coherence/std": 0.47320507764816283,
|
| 256 |
+
"rewards/check_response_quality/mean": 2.3375,
|
| 257 |
+
"rewards/check_response_quality/std": 0.21933756470680238,
|
| 258 |
+
"rewards/match_format_approximately/mean": 0.8375,
|
| 259 |
+
"rewards/match_format_approximately/std": 0.21933756470680238,
|
| 260 |
+
"rewards/match_format_exactly/mean": 1.0,
|
| 261 |
+
"rewards/match_format_exactly/std": 0.0,
|
| 262 |
+
"step": 90
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"completion_length": 128.325,
|
| 266 |
+
"completions/clipped_ratio": 0.0,
|
| 267 |
+
"completions/max_length": 322.5,
|
| 268 |
+
"completions/max_terminated_length": 322.5,
|
| 269 |
+
"completions/mean_length": 128.325,
|
| 270 |
+
"completions/mean_terminated_length": 128.325,
|
| 271 |
+
"completions/min_length": 46.3,
|
| 272 |
+
"completions/min_terminated_length": 46.3,
|
| 273 |
+
"epoch": 0.006876633200385091,
|
| 274 |
+
"frac_reward_zero_std": 0.3,
|
| 275 |
+
"grad_norm": 0.003549874061718583,
|
| 276 |
+
"kl": 0.820201675593853,
|
| 277 |
+
"learning_rate": 4.000000000000001e-06,
|
| 278 |
+
"loss": 0.0008,
|
| 279 |
+
"num_tokens": 205998.0,
|
| 280 |
+
"reward": 5.45,
|
| 281 |
+
"reward_std": 0.5343478560447693,
|
| 282 |
+
"rewards/check_coherence/mean": 1.225,
|
| 283 |
+
"rewards/check_coherence/std": 0.3809401035308838,
|
| 284 |
+
"rewards/check_response_quality/mean": 2.3625,
|
| 285 |
+
"rewards/check_response_quality/std": 0.10386751294136047,
|
| 286 |
+
"rewards/match_format_approximately/mean": 0.8625,
|
| 287 |
+
"rewards/match_format_approximately/std": 0.10386751294136047,
|
| 288 |
+
"rewards/match_format_exactly/mean": 1.0,
|
| 289 |
+
"rewards/match_format_exactly/std": 0.0,
|
| 290 |
+
"step": 100
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"completion_length": 147.575,
|
| 294 |
+
"completions/clipped_ratio": 0.0,
|
| 295 |
+
"completions/max_length": 452.6,
|
| 296 |
+
"completions/max_terminated_length": 452.6,
|
| 297 |
+
"completions/mean_length": 147.575,
|
| 298 |
+
"completions/mean_terminated_length": 147.575,
|
| 299 |
+
"completions/min_length": 8.8,
|
| 300 |
+
"completions/min_terminated_length": 8.8,
|
| 301 |
+
"epoch": 0.007564296520423601,
|
| 302 |
+
"frac_reward_zero_std": 0.1,
|
| 303 |
+
"grad_norm": 2.094895124435425,
|
| 304 |
+
"kl": 1.165158998966217,
|
| 305 |
+
"learning_rate": 3.857142857142858e-06,
|
| 306 |
+
"loss": 0.0012,
|
| 307 |
+
"num_tokens": 224325.0,
|
| 308 |
+
"reward": 5.3,
|
| 309 |
+
"reward_std": 0.8748959302902222,
|
| 310 |
+
"rewards/check_coherence/mean": 1.2125,
|
| 311 |
+
"rewards/check_coherence/std": 0.38273502588272096,
|
| 312 |
+
"rewards/check_response_quality/mean": 2.2875,
|
| 313 |
+
"rewards/check_response_quality/std": 0.28898603916168214,
|
| 314 |
+
"rewards/match_format_approximately/mean": 0.825,
|
| 315 |
+
"rewards/match_format_approximately/std": 0.22320507764816283,
|
| 316 |
+
"rewards/match_format_exactly/mean": 0.975,
|
| 317 |
+
"rewards/match_format_exactly/std": 0.05,
|
| 318 |
+
"step": 110
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"completion_length": 93.475,
|
| 322 |
+
"completions/clipped_ratio": 0.0,
|
| 323 |
+
"completions/max_length": 232.9,
|
| 324 |
+
"completions/max_terminated_length": 232.9,
|
| 325 |
+
"completions/mean_length": 93.475,
|
| 326 |
+
"completions/mean_terminated_length": 93.475,
|
| 327 |
+
"completions/min_length": 8.8,
|
| 328 |
+
"completions/min_terminated_length": 8.8,
|
| 329 |
+
"epoch": 0.00825195984046211,
|
| 330 |
+
"frac_reward_zero_std": 0.3,
|
| 331 |
+
"grad_norm": 1.8320742845535278,
|
| 332 |
+
"kl": 0.8784003466367721,
|
| 333 |
+
"learning_rate": 3.6984126984126987e-06,
|
| 334 |
+
"loss": 0.0009,
|
| 335 |
+
"num_tokens": 238872.0,
|
| 336 |
+
"reward": 5.6125,
|
| 337 |
+
"reward_std": 0.5023834943771363,
|
| 338 |
+
"rewards/check_coherence/mean": 1.3875,
|
| 339 |
+
"rewards/check_coherence/std": 0.1978713572025299,
|
| 340 |
+
"rewards/check_response_quality/mean": 2.3625,
|
| 341 |
+
"rewards/check_response_quality/std": 0.18273502588272095,
|
| 342 |
+
"rewards/match_format_approximately/mean": 0.8625,
|
| 343 |
+
"rewards/match_format_approximately/std": 0.18273502588272095,
|
| 344 |
+
"rewards/match_format_exactly/mean": 1.0,
|
| 345 |
+
"rewards/match_format_exactly/std": 0.0,
|
| 346 |
+
"step": 120
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"completion_length": 63.425,
|
| 350 |
+
"completions/clipped_ratio": 0.0,
|
| 351 |
+
"completions/max_length": 180.7,
|
| 352 |
+
"completions/max_terminated_length": 180.7,
|
| 353 |
+
"completions/mean_length": 63.425,
|
| 354 |
+
"completions/mean_terminated_length": 63.425,
|
| 355 |
+
"completions/min_length": 9.8,
|
| 356 |
+
"completions/min_terminated_length": 9.8,
|
| 357 |
+
"epoch": 0.008939623160500619,
|
| 358 |
+
"frac_reward_zero_std": 0.5,
|
| 359 |
+
"grad_norm": 1.388489007949829,
|
| 360 |
+
"kl": 0.855793622136116,
|
| 361 |
+
"learning_rate": 3.53968253968254e-06,
|
| 362 |
+
"loss": 0.0009,
|
| 363 |
+
"num_tokens": 252433.0,
|
| 364 |
+
"reward": 5.75,
|
| 365 |
+
"reward_std": 0.4457427144050598,
|
| 366 |
+
"rewards/check_coherence/mean": 1.4,
|
| 367 |
+
"rewards/check_coherence/std": 0.2,
|
| 368 |
+
"rewards/check_response_quality/mean": 2.425,
|
| 369 |
+
"rewards/check_response_quality/std": 0.12886751294136048,
|
| 370 |
+
"rewards/match_format_approximately/mean": 0.925,
|
| 371 |
+
"rewards/match_format_approximately/std": 0.12886751294136048,
|
| 372 |
+
"rewards/match_format_exactly/mean": 1.0,
|
| 373 |
+
"rewards/match_format_exactly/std": 0.0,
|
| 374 |
+
"step": 130
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"completion_length": 41.35,
|
| 378 |
+
"completions/clipped_ratio": 0.0,
|
| 379 |
+
"completions/max_length": 127.7,
|
| 380 |
+
"completions/max_terminated_length": 127.7,
|
| 381 |
+
"completions/mean_length": 41.35,
|
| 382 |
+
"completions/mean_terminated_length": 41.35,
|
| 383 |
+
"completions/min_length": 7.7,
|
| 384 |
+
"completions/min_terminated_length": 7.7,
|
| 385 |
+
"epoch": 0.009627286480539128,
|
| 386 |
+
"frac_reward_zero_std": 0.0,
|
| 387 |
+
"grad_norm": 2.7645645141601562,
|
| 388 |
+
"kl": 1.5349723994731903,
|
| 389 |
+
"learning_rate": 3.3809523809523814e-06,
|
| 390 |
+
"loss": 0.0015,
|
| 391 |
+
"num_tokens": 264579.0,
|
| 392 |
+
"reward": 5.0875,
|
| 393 |
+
"reward_std": 1.1462337374687195,
|
| 394 |
+
"rewards/check_coherence/mean": 1.15,
|
| 395 |
+
"rewards/check_coherence/std": 0.48867512941360475,
|
| 396 |
+
"rewards/check_response_quality/mean": 2.1875,
|
| 397 |
+
"rewards/check_response_quality/std": 0.3914190471172333,
|
| 398 |
+
"rewards/match_format_approximately/mean": 0.85,
|
| 399 |
+
"rewards/match_format_approximately/std": 0.20773502588272094,
|
| 400 |
+
"rewards/match_format_exactly/mean": 0.9,
|
| 401 |
+
"rewards/match_format_exactly/std": 0.1154700517654419,
|
| 402 |
+
"step": 140
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"completion_length": 46.625,
|
| 406 |
+
"completions/clipped_ratio": 0.0,
|
| 407 |
+
"completions/max_length": 135.3,
|
| 408 |
+
"completions/max_terminated_length": 135.3,
|
| 409 |
+
"completions/mean_length": 46.625,
|
| 410 |
+
"completions/mean_terminated_length": 46.625,
|
| 411 |
+
"completions/min_length": 10.0,
|
| 412 |
+
"completions/min_terminated_length": 10.0,
|
| 413 |
+
"epoch": 0.010314949800577638,
|
| 414 |
+
"frac_reward_zero_std": 0.4,
|
| 415 |
+
"grad_norm": 0.36925405263900757,
|
| 416 |
+
"kl": 1.2759959518909454,
|
| 417 |
+
"learning_rate": 3.2222222222222227e-06,
|
| 418 |
+
"loss": 0.0013,
|
| 419 |
+
"num_tokens": 278280.0,
|
| 420 |
+
"reward": 5.7,
|
| 421 |
+
"reward_std": 0.4686140716075897,
|
| 422 |
+
"rewards/check_coherence/mean": 1.3875,
|
| 423 |
+
"rewards/check_coherence/std": 0.225,
|
| 424 |
+
"rewards/check_response_quality/mean": 2.4,
|
| 425 |
+
"rewards/check_response_quality/std": 0.15173887014389037,
|
| 426 |
+
"rewards/match_format_approximately/mean": 0.9125,
|
| 427 |
+
"rewards/match_format_approximately/std": 0.13273502588272096,
|
| 428 |
+
"rewards/match_format_exactly/mean": 1.0,
|
| 429 |
+
"rewards/match_format_exactly/std": 0.0,
|
| 430 |
+
"step": 150
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"completion_length": 37.85,
|
| 434 |
+
"completions/clipped_ratio": 0.0,
|
| 435 |
+
"completions/max_length": 108.9,
|
| 436 |
+
"completions/max_terminated_length": 108.9,
|
| 437 |
+
"completions/mean_length": 37.85,
|
| 438 |
+
"completions/mean_terminated_length": 37.85,
|
| 439 |
+
"completions/min_length": 10.3,
|
| 440 |
+
"completions/min_terminated_length": 10.3,
|
| 441 |
+
"epoch": 0.011002613120616147,
|
| 442 |
+
"frac_reward_zero_std": 0.5,
|
| 443 |
+
"grad_norm": 0.003718956606462598,
|
| 444 |
+
"kl": 1.1356925666332245,
|
| 445 |
+
"learning_rate": 3.063492063492064e-06,
|
| 446 |
+
"loss": 0.0011,
|
| 447 |
+
"num_tokens": 289698.0,
|
| 448 |
+
"reward": 5.8625,
|
| 449 |
+
"reward_std": 0.275,
|
| 450 |
+
"rewards/check_coherence/mean": 1.4,
|
| 451 |
+
"rewards/check_coherence/std": 0.2,
|
| 452 |
+
"rewards/check_response_quality/mean": 2.475,
|
| 453 |
+
"rewards/check_response_quality/std": 0.05,
|
| 454 |
+
"rewards/match_format_approximately/mean": 0.9875,
|
| 455 |
+
"rewards/match_format_approximately/std": 0.025,
|
| 456 |
+
"rewards/match_format_exactly/mean": 1.0,
|
| 457 |
+
"rewards/match_format_exactly/std": 0.0,
|
| 458 |
+
"step": 160
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"completion_length": 24.475,
|
| 462 |
+
"completions/clipped_ratio": 0.0,
|
| 463 |
+
"completions/max_length": 58.6,
|
| 464 |
+
"completions/max_terminated_length": 58.6,
|
| 465 |
+
"completions/mean_length": 24.475,
|
| 466 |
+
"completions/mean_terminated_length": 24.475,
|
| 467 |
+
"completions/min_length": 10.1,
|
| 468 |
+
"completions/min_terminated_length": 10.1,
|
| 469 |
+
"epoch": 0.011690276440654656,
|
| 470 |
+
"frac_reward_zero_std": 0.8,
|
| 471 |
+
"grad_norm": 0.0026451745070517063,
|
| 472 |
+
"kl": 1.1320037961006164,
|
| 473 |
+
"learning_rate": 2.9047619047619053e-06,
|
| 474 |
+
"loss": 0.0011,
|
| 475 |
+
"num_tokens": 302601.0,
|
| 476 |
+
"reward": 5.9,
|
| 477 |
+
"reward_std": 0.14574271440505981,
|
| 478 |
+
"rewards/check_coherence/mean": 1.425,
|
| 479 |
+
"rewards/check_coherence/std": 0.10773502588272095,
|
| 480 |
+
"rewards/check_response_quality/mean": 2.4875,
|
| 481 |
+
"rewards/check_response_quality/std": 0.025,
|
| 482 |
+
"rewards/match_format_approximately/mean": 0.9875,
|
| 483 |
+
"rewards/match_format_approximately/std": 0.025,
|
| 484 |
+
"rewards/match_format_exactly/mean": 1.0,
|
| 485 |
+
"rewards/match_format_exactly/std": 0.0,
|
| 486 |
+
"step": 170
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"completion_length": 65.375,
|
| 490 |
+
"completions/clipped_ratio": 0.025,
|
| 491 |
+
"completions/max_length": 216.0,
|
| 492 |
+
"completions/max_terminated_length": 67.0,
|
| 493 |
+
"completions/mean_length": 65.375,
|
| 494 |
+
"completions/mean_terminated_length": 26.866666793823242,
|
| 495 |
+
"completions/min_length": 9.4,
|
| 496 |
+
"completions/min_terminated_length": 9.4,
|
| 497 |
+
"epoch": 0.012377939760693164,
|
| 498 |
+
"frac_reward_zero_std": 0.3,
|
| 499 |
+
"grad_norm": 2.4312984943389893,
|
| 500 |
+
"kl": 1.0950138330459596,
|
| 501 |
+
"learning_rate": 2.7460317460317466e-06,
|
| 502 |
+
"loss": 0.0011,
|
| 503 |
+
"num_tokens": 317912.0,
|
| 504 |
+
"reward": 5.775,
|
| 505 |
+
"reward_std": 0.3957427144050598,
|
| 506 |
+
"rewards/check_coherence/mean": 1.4,
|
| 507 |
+
"rewards/check_coherence/std": 0.2,
|
| 508 |
+
"rewards/check_response_quality/mean": 2.4375,
|
| 509 |
+
"rewards/check_response_quality/std": 0.10386751294136047,
|
| 510 |
+
"rewards/match_format_approximately/mean": 0.9375,
|
| 511 |
+
"rewards/match_format_approximately/std": 0.10386751294136047,
|
| 512 |
+
"rewards/match_format_exactly/mean": 1.0,
|
| 513 |
+
"rewards/match_format_exactly/std": 0.0,
|
| 514 |
+
"step": 180
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"completion_length": 16.275,
|
| 518 |
+
"completions/clipped_ratio": 0.0,
|
| 519 |
+
"completions/max_length": 25.2,
|
| 520 |
+
"completions/max_terminated_length": 25.2,
|
| 521 |
+
"completions/mean_length": 16.275,
|
| 522 |
+
"completions/mean_terminated_length": 16.275,
|
| 523 |
+
"completions/min_length": 9.8,
|
| 524 |
+
"completions/min_terminated_length": 9.8,
|
| 525 |
+
"epoch": 0.013065603080731673,
|
| 526 |
+
"frac_reward_zero_std": 0.6,
|
| 527 |
+
"grad_norm": 0.004265956114977598,
|
| 528 |
+
"kl": 1.1890327751636505,
|
| 529 |
+
"learning_rate": 2.587301587301588e-06,
|
| 530 |
+
"loss": 0.0012,
|
| 531 |
+
"num_tokens": 332055.0,
|
| 532 |
+
"reward": 5.8875,
|
| 533 |
+
"reward_std": 0.18273502588272095,
|
| 534 |
+
"rewards/check_coherence/mean": 1.3875,
|
| 535 |
+
"rewards/check_coherence/std": 0.18273502588272095,
|
| 536 |
+
"rewards/check_response_quality/mean": 2.5,
|
| 537 |
+
"rewards/check_response_quality/std": 0.0,
|
| 538 |
+
"rewards/match_format_approximately/mean": 1.0,
|
| 539 |
+
"rewards/match_format_approximately/std": 0.0,
|
| 540 |
+
"rewards/match_format_exactly/mean": 1.0,
|
| 541 |
+
"rewards/match_format_exactly/std": 0.0,
|
| 542 |
+
"step": 190
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"completion_length": 37.9,
|
| 546 |
+
"completions/clipped_ratio": 0.0,
|
| 547 |
+
"completions/max_length": 112.8,
|
| 548 |
+
"completions/max_terminated_length": 112.8,
|
| 549 |
+
"completions/mean_length": 37.9,
|
| 550 |
+
"completions/mean_terminated_length": 37.9,
|
| 551 |
+
"completions/min_length": 8.1,
|
| 552 |
+
"completions/min_terminated_length": 8.1,
|
| 553 |
+
"epoch": 0.013753266400770183,
|
| 554 |
+
"frac_reward_zero_std": 0.4,
|
| 555 |
+
"grad_norm": 16.73357582092285,
|
| 556 |
+
"kl": 1.1858276724815369,
|
| 557 |
+
"learning_rate": 2.428571428571429e-06,
|
| 558 |
+
"loss": 0.0012,
|
| 559 |
+
"num_tokens": 344875.0,
|
| 560 |
+
"reward": 5.75,
|
| 561 |
+
"reward_std": 0.4457427144050598,
|
| 562 |
+
"rewards/check_coherence/mean": 1.35,
|
| 563 |
+
"rewards/check_coherence/std": 0.25773502588272096,
|
| 564 |
+
"rewards/check_response_quality/mean": 2.45,
|
| 565 |
+
"rewards/check_response_quality/std": 0.1,
|
| 566 |
+
"rewards/match_format_approximately/mean": 0.95,
|
| 567 |
+
"rewards/match_format_approximately/std": 0.1,
|
| 568 |
+
"rewards/match_format_exactly/mean": 1.0,
|
| 569 |
+
"rewards/match_format_exactly/std": 0.0,
|
| 570 |
+
"step": 200
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"completion_length": 57.225,
|
| 574 |
+
"completions/clipped_ratio": 0.0,
|
| 575 |
+
"completions/max_length": 142.9,
|
| 576 |
+
"completions/max_terminated_length": 142.9,
|
| 577 |
+
"completions/mean_length": 57.225,
|
| 578 |
+
"completions/mean_terminated_length": 57.225,
|
| 579 |
+
"completions/min_length": 9.4,
|
| 580 |
+
"completions/min_terminated_length": 9.4,
|
| 581 |
+
"epoch": 0.014440929720808692,
|
| 582 |
+
"frac_reward_zero_std": 0.6,
|
| 583 |
+
"grad_norm": 0.006797971669584513,
|
| 584 |
+
"kl": 1.0238984107971192,
|
| 585 |
+
"learning_rate": 2.26984126984127e-06,
|
| 586 |
+
"loss": 0.001,
|
| 587 |
+
"num_tokens": 359308.0,
|
| 588 |
+
"reward": 5.7625,
|
| 589 |
+
"reward_std": 0.2946484684944153,
|
| 590 |
+
"rewards/check_coherence/mean": 1.375,
|
| 591 |
+
"rewards/check_coherence/std": 0.20773502588272094,
|
| 592 |
+
"rewards/check_response_quality/mean": 2.4375,
|
| 593 |
+
"rewards/check_response_quality/std": 0.06582483053207397,
|
| 594 |
+
"rewards/match_format_approximately/mean": 0.95,
|
| 595 |
+
"rewards/match_format_approximately/std": 0.05,
|
| 596 |
+
"rewards/match_format_exactly/mean": 1.0,
|
| 597 |
+
"rewards/match_format_exactly/std": 0.0,
|
| 598 |
+
"step": 210
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"completion_length": 60.225,
|
| 602 |
+
"completions/clipped_ratio": 0.0,
|
| 603 |
+
"completions/max_length": 198.7,
|
| 604 |
+
"completions/max_terminated_length": 198.7,
|
| 605 |
+
"completions/mean_length": 60.225,
|
| 606 |
+
"completions/mean_terminated_length": 60.225,
|
| 607 |
+
"completions/min_length": 10.7,
|
| 608 |
+
"completions/min_terminated_length": 10.7,
|
| 609 |
+
"epoch": 0.015128593040847202,
|
| 610 |
+
"frac_reward_zero_std": 0.4,
|
| 611 |
+
"grad_norm": 1.2490910291671753,
|
| 612 |
+
"kl": 1.139722502231598,
|
| 613 |
+
"learning_rate": 2.1111111111111114e-06,
|
| 614 |
+
"loss": 0.0011,
|
| 615 |
+
"num_tokens": 372089.0,
|
| 616 |
+
"reward": 5.8,
|
| 617 |
+
"reward_std": 0.4,
|
| 618 |
+
"rewards/check_coherence/mean": 1.375,
|
| 619 |
+
"rewards/check_coherence/std": 0.25,
|
| 620 |
+
"rewards/check_response_quality/mean": 2.4625,
|
| 621 |
+
"rewards/check_response_quality/std": 0.075,
|
| 622 |
+
"rewards/match_format_approximately/mean": 0.9625,
|
| 623 |
+
"rewards/match_format_approximately/std": 0.075,
|
| 624 |
+
"rewards/match_format_exactly/mean": 1.0,
|
| 625 |
+
"rewards/match_format_exactly/std": 0.0,
|
| 626 |
+
"step": 220
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"completion_length": 30.075,
|
| 630 |
+
"completions/clipped_ratio": 0.0,
|
| 631 |
+
"completions/max_length": 84.5,
|
| 632 |
+
"completions/max_terminated_length": 84.5,
|
| 633 |
+
"completions/mean_length": 30.075,
|
| 634 |
+
"completions/mean_terminated_length": 30.075,
|
| 635 |
+
"completions/min_length": 9.8,
|
| 636 |
+
"completions/min_terminated_length": 9.8,
|
| 637 |
+
"epoch": 0.01581625636088571,
|
| 638 |
+
"frac_reward_zero_std": 0.7,
|
| 639 |
+
"grad_norm": 0.004948398098349571,
|
| 640 |
+
"kl": 1.1260765612125396,
|
| 641 |
+
"learning_rate": 1.9523809523809527e-06,
|
| 642 |
+
"loss": 0.0011,
|
| 643 |
+
"num_tokens": 384276.0,
|
| 644 |
+
"reward": 5.875,
|
| 645 |
+
"reward_std": 0.1728713572025299,
|
| 646 |
+
"rewards/check_coherence/mean": 1.4375,
|
| 647 |
+
"rewards/check_coherence/std": 0.0978713572025299,
|
| 648 |
+
"rewards/check_response_quality/mean": 2.4625,
|
| 649 |
+
"rewards/check_response_quality/std": 0.075,
|
| 650 |
+
"rewards/match_format_approximately/mean": 0.975,
|
| 651 |
+
"rewards/match_format_approximately/std": 0.05,
|
| 652 |
+
"rewards/match_format_exactly/mean": 1.0,
|
| 653 |
+
"rewards/match_format_exactly/std": 0.0,
|
| 654 |
+
"step": 230
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"completion_length": 65.25,
|
| 658 |
+
"completions/clipped_ratio": 0.025,
|
| 659 |
+
"completions/max_length": 223.1,
|
| 660 |
+
"completions/max_terminated_length": 65.9,
|
| 661 |
+
"completions/mean_length": 65.25,
|
| 662 |
+
"completions/mean_terminated_length": 25.908333396911623,
|
| 663 |
+
"completions/min_length": 9.1,
|
| 664 |
+
"completions/min_terminated_length": 9.1,
|
| 665 |
+
"epoch": 0.01650391968092422,
|
| 666 |
+
"frac_reward_zero_std": 0.6,
|
| 667 |
+
"grad_norm": 0.0049158609472215176,
|
| 668 |
+
"kl": 1.044507622718811,
|
| 669 |
+
"learning_rate": 1.7936507936507938e-06,
|
| 670 |
+
"loss": 0.001,
|
| 671 |
+
"num_tokens": 399686.0,
|
| 672 |
+
"reward": 5.8375,
|
| 673 |
+
"reward_std": 0.325,
|
| 674 |
+
"rewards/check_coherence/mean": 1.45,
|
| 675 |
+
"rewards/check_coherence/std": 0.1,
|
| 676 |
+
"rewards/check_response_quality/mean": 2.4375,
|
| 677 |
+
"rewards/check_response_quality/std": 0.125,
|
| 678 |
+
"rewards/match_format_approximately/mean": 0.95,
|
| 679 |
+
"rewards/match_format_approximately/std": 0.1,
|
| 680 |
+
"rewards/match_format_exactly/mean": 1.0,
|
| 681 |
+
"rewards/match_format_exactly/std": 0.0,
|
| 682 |
+
"step": 240
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"completion_length": 34.25,
|
| 686 |
+
"completions/clipped_ratio": 0.0,
|
| 687 |
+
"completions/max_length": 97.6,
|
| 688 |
+
"completions/max_terminated_length": 97.6,
|
| 689 |
+
"completions/mean_length": 34.25,
|
| 690 |
+
"completions/mean_terminated_length": 34.25,
|
| 691 |
+
"completions/min_length": 9.6,
|
| 692 |
+
"completions/min_terminated_length": 9.6,
|
| 693 |
+
"epoch": 0.01719158300096273,
|
| 694 |
+
"frac_reward_zero_std": 0.6,
|
| 695 |
+
"grad_norm": 0.0035998751409351826,
|
| 696 |
+
"kl": 1.131579464673996,
|
| 697 |
+
"learning_rate": 1.6349206349206351e-06,
|
| 698 |
+
"loss": 0.0011,
|
| 699 |
+
"num_tokens": 412936.0,
|
| 700 |
+
"reward": 5.875,
|
| 701 |
+
"reward_std": 0.25,
|
| 702 |
+
"rewards/check_coherence/mean": 1.45,
|
| 703 |
+
"rewards/check_coherence/std": 0.1,
|
| 704 |
+
"rewards/check_response_quality/mean": 2.4625,
|
| 705 |
+
"rewards/check_response_quality/std": 0.075,
|
| 706 |
+
"rewards/match_format_approximately/mean": 0.9625,
|
| 707 |
+
"rewards/match_format_approximately/std": 0.075,
|
| 708 |
+
"rewards/match_format_exactly/mean": 1.0,
|
| 709 |
+
"rewards/match_format_exactly/std": 0.0,
|
| 710 |
+
"step": 250
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"completion_length": 44.525,
|
| 714 |
+
"completions/clipped_ratio": 0.0,
|
| 715 |
+
"completions/max_length": 138.9,
|
| 716 |
+
"completions/max_terminated_length": 138.9,
|
| 717 |
+
"completions/mean_length": 44.525,
|
| 718 |
+
"completions/mean_terminated_length": 44.525,
|
| 719 |
+
"completions/min_length": 9.5,
|
| 720 |
+
"completions/min_terminated_length": 9.5,
|
| 721 |
+
"epoch": 0.017879246321001237,
|
| 722 |
+
"frac_reward_zero_std": 0.8,
|
| 723 |
+
"grad_norm": 0.003912392072379589,
|
| 724 |
+
"kl": 1.0597735822200776,
|
| 725 |
+
"learning_rate": 1.4761904761904762e-06,
|
| 726 |
+
"loss": 0.0011,
|
| 727 |
+
"num_tokens": 427709.0,
|
| 728 |
+
"reward": 5.9375,
|
| 729 |
+
"reward_std": 0.125,
|
| 730 |
+
"rewards/check_coherence/mean": 1.475,
|
| 731 |
+
"rewards/check_coherence/std": 0.05,
|
| 732 |
+
"rewards/check_response_quality/mean": 2.475,
|
| 733 |
+
"rewards/check_response_quality/std": 0.05,
|
| 734 |
+
"rewards/match_format_approximately/mean": 0.9875,
|
| 735 |
+
"rewards/match_format_approximately/std": 0.025,
|
| 736 |
+
"rewards/match_format_exactly/mean": 1.0,
|
| 737 |
+
"rewards/match_format_exactly/std": 0.0,
|
| 738 |
+
"step": 260
|
| 739 |
+
},
|
| 740 |
+
{
|
| 741 |
+
"completion_length": 13.175,
|
| 742 |
+
"completions/clipped_ratio": 0.0,
|
| 743 |
+
"completions/max_length": 22.9,
|
| 744 |
+
"completions/max_terminated_length": 22.9,
|
| 745 |
+
"completions/mean_length": 13.175,
|
| 746 |
+
"completions/mean_terminated_length": 13.175,
|
| 747 |
+
"completions/min_length": 7.3,
|
| 748 |
+
"completions/min_terminated_length": 7.3,
|
| 749 |
+
"epoch": 0.01856690964103975,
|
| 750 |
+
"frac_reward_zero_std": 0.6,
|
| 751 |
+
"grad_norm": 4.749101638793945,
|
| 752 |
+
"kl": 1.352827501296997,
|
| 753 |
+
"learning_rate": 1.3174603174603175e-06,
|
| 754 |
+
"loss": 0.0014,
|
| 755 |
+
"num_tokens": 437888.0,
|
| 756 |
+
"reward": 5.7625,
|
| 757 |
+
"reward_std": 0.41304759979248046,
|
| 758 |
+
"rewards/check_coherence/mean": 1.4125,
|
| 759 |
+
"rewards/check_coherence/std": 0.175,
|
| 760 |
+
"rewards/check_response_quality/mean": 2.4125,
|
| 761 |
+
"rewards/check_response_quality/std": 0.14464847445487977,
|
| 762 |
+
"rewards/match_format_approximately/mean": 0.9625,
|
| 763 |
+
"rewards/match_format_approximately/std": 0.053867512941360475,
|
| 764 |
+
"rewards/match_format_exactly/mean": 0.975,
|
| 765 |
+
"rewards/match_format_exactly/std": 0.05,
|
| 766 |
+
"step": 270
|
| 767 |
+
},
|
| 768 |
+
{
|
| 769 |
+
"completion_length": 29.05,
|
| 770 |
+
"completions/clipped_ratio": 0.0,
|
| 771 |
+
"completions/max_length": 75.4,
|
| 772 |
+
"completions/max_terminated_length": 75.4,
|
| 773 |
+
"completions/mean_length": 29.05,
|
| 774 |
+
"completions/mean_terminated_length": 29.05,
|
| 775 |
+
"completions/min_length": 9.3,
|
| 776 |
+
"completions/min_terminated_length": 9.3,
|
| 777 |
+
"epoch": 0.019254572961078256,
|
| 778 |
+
"frac_reward_zero_std": 0.7,
|
| 779 |
+
"grad_norm": 0.0032444808166474104,
|
| 780 |
+
"kl": 1.0743911743164063,
|
| 781 |
+
"learning_rate": 1.1587301587301589e-06,
|
| 782 |
+
"loss": 0.0011,
|
| 783 |
+
"num_tokens": 450566.0,
|
| 784 |
+
"reward": 5.8875,
|
| 785 |
+
"reward_std": 0.225,
|
| 786 |
+
"rewards/check_coherence/mean": 1.4375,
|
| 787 |
+
"rewards/check_coherence/std": 0.125,
|
| 788 |
+
"rewards/check_response_quality/mean": 2.475,
|
| 789 |
+
"rewards/check_response_quality/std": 0.05,
|
| 790 |
+
"rewards/match_format_approximately/mean": 0.975,
|
| 791 |
+
"rewards/match_format_approximately/std": 0.05,
|
| 792 |
+
"rewards/match_format_exactly/mean": 1.0,
|
| 793 |
+
"rewards/match_format_exactly/std": 0.0,
|
| 794 |
+
"step": 280
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"completion_length": 35.0,
|
| 798 |
+
"completions/clipped_ratio": 0.0,
|
| 799 |
+
"completions/max_length": 100.7,
|
| 800 |
+
"completions/max_terminated_length": 100.7,
|
| 801 |
+
"completions/mean_length": 35.0,
|
| 802 |
+
"completions/mean_terminated_length": 35.0,
|
| 803 |
+
"completions/min_length": 9.7,
|
| 804 |
+
"completions/min_terminated_length": 9.7,
|
| 805 |
+
"epoch": 0.019942236281116764,
|
| 806 |
+
"frac_reward_zero_std": 0.7,
|
| 807 |
+
"grad_norm": 4.597572326660156,
|
| 808 |
+
"kl": 9.488562166690826,
|
| 809 |
+
"learning_rate": 1.0000000000000002e-06,
|
| 810 |
+
"loss": 0.0095,
|
| 811 |
+
"num_tokens": 462366.0,
|
| 812 |
+
"reward": 5.875,
|
| 813 |
+
"reward_std": 0.20773502588272094,
|
| 814 |
+
"rewards/check_coherence/mean": 1.4,
|
| 815 |
+
"rewards/check_coherence/std": 0.15773502588272095,
|
| 816 |
+
"rewards/check_response_quality/mean": 2.4875,
|
| 817 |
+
"rewards/check_response_quality/std": 0.025,
|
| 818 |
+
"rewards/match_format_approximately/mean": 0.9875,
|
| 819 |
+
"rewards/match_format_approximately/std": 0.025,
|
| 820 |
+
"rewards/match_format_exactly/mean": 1.0,
|
| 821 |
+
"rewards/match_format_exactly/std": 0.0,
|
| 822 |
+
"step": 290
|
| 823 |
+
},
|
| 824 |
+
{
|
| 825 |
+
"completion_length": 16.875,
|
| 826 |
+
"completions/clipped_ratio": 0.0,
|
| 827 |
+
"completions/max_length": 31.8,
|
| 828 |
+
"completions/max_terminated_length": 31.8,
|
| 829 |
+
"completions/mean_length": 16.875,
|
| 830 |
+
"completions/mean_terminated_length": 16.875,
|
| 831 |
+
"completions/min_length": 8.6,
|
| 832 |
+
"completions/min_terminated_length": 8.6,
|
| 833 |
+
"epoch": 0.020629899601155275,
|
| 834 |
+
"frac_reward_zero_std": 0.8,
|
| 835 |
+
"grad_norm": 0.0025285985320806503,
|
| 836 |
+
"kl": 1.0927811563014984,
|
| 837 |
+
"learning_rate": 8.412698412698414e-07,
|
| 838 |
+
"loss": 0.0011,
|
| 839 |
+
"num_tokens": 473501.0,
|
| 840 |
+
"reward": 5.9375,
|
| 841 |
+
"reward_std": 0.125,
|
| 842 |
+
"rewards/check_coherence/mean": 1.5,
|
| 843 |
+
"rewards/check_coherence/std": 0.0,
|
| 844 |
+
"rewards/check_response_quality/mean": 2.4625,
|
| 845 |
+
"rewards/check_response_quality/std": 0.075,
|
| 846 |
+
"rewards/match_format_approximately/mean": 0.975,
|
| 847 |
+
"rewards/match_format_approximately/std": 0.05,
|
| 848 |
+
"rewards/match_format_exactly/mean": 1.0,
|
| 849 |
+
"rewards/match_format_exactly/std": 0.0,
|
| 850 |
+
"step": 300
|
| 851 |
+
},
|
| 852 |
+
{
|
| 853 |
+
"completion_length": 34.825,
|
| 854 |
+
"completions/clipped_ratio": 0.0,
|
| 855 |
+
"completions/max_length": 104.8,
|
| 856 |
+
"completions/max_terminated_length": 104.8,
|
| 857 |
+
"completions/mean_length": 34.825,
|
| 858 |
+
"completions/mean_terminated_length": 34.825,
|
| 859 |
+
"completions/min_length": 9.6,
|
| 860 |
+
"completions/min_terminated_length": 9.6,
|
| 861 |
+
"epoch": 0.021317562921193783,
|
| 862 |
+
"frac_reward_zero_std": 0.9,
|
| 863 |
+
"grad_norm": 0.004729569889605045,
|
| 864 |
+
"kl": 1.0428565800189973,
|
| 865 |
+
"learning_rate": 6.825396825396826e-07,
|
| 866 |
+
"loss": 0.001,
|
| 867 |
+
"num_tokens": 487282.0,
|
| 868 |
+
"reward": 5.975,
|
| 869 |
+
"reward_std": 0.05,
|
| 870 |
+
"rewards/check_coherence/mean": 1.5,
|
| 871 |
+
"rewards/check_coherence/std": 0.0,
|
| 872 |
+
"rewards/check_response_quality/mean": 2.4875,
|
| 873 |
+
"rewards/check_response_quality/std": 0.025,
|
| 874 |
+
"rewards/match_format_approximately/mean": 0.9875,
|
| 875 |
+
"rewards/match_format_approximately/std": 0.025,
|
| 876 |
+
"rewards/match_format_exactly/mean": 1.0,
|
| 877 |
+
"rewards/match_format_exactly/std": 0.0,
|
| 878 |
+
"step": 310
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"completion_length": 28.475,
|
| 882 |
+
"completions/clipped_ratio": 0.0,
|
| 883 |
+
"completions/max_length": 81.7,
|
| 884 |
+
"completions/max_terminated_length": 81.7,
|
| 885 |
+
"completions/mean_length": 28.475,
|
| 886 |
+
"completions/mean_terminated_length": 28.475,
|
| 887 |
+
"completions/min_length": 9.0,
|
| 888 |
+
"completions/min_terminated_length": 9.0,
|
| 889 |
+
"epoch": 0.022005226241232294,
|
| 890 |
+
"frac_reward_zero_std": 0.8,
|
| 891 |
+
"grad_norm": 6.385797500610352,
|
| 892 |
+
"kl": 1.2766858220100403,
|
| 893 |
+
"learning_rate": 5.238095238095239e-07,
|
| 894 |
+
"loss": 0.0013,
|
| 895 |
+
"num_tokens": 500121.0,
|
| 896 |
+
"reward": 5.925,
|
| 897 |
+
"reward_std": 0.15,
|
| 898 |
+
"rewards/check_coherence/mean": 1.45,
|
| 899 |
+
"rewards/check_coherence/std": 0.1,
|
| 900 |
+
"rewards/check_response_quality/mean": 2.4875,
|
| 901 |
+
"rewards/check_response_quality/std": 0.025,
|
| 902 |
+
"rewards/match_format_approximately/mean": 0.9875,
|
| 903 |
+
"rewards/match_format_approximately/std": 0.025,
|
| 904 |
+
"rewards/match_format_exactly/mean": 1.0,
|
| 905 |
+
"rewards/match_format_exactly/std": 0.0,
|
| 906 |
+
"step": 320
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"completion_length": 18.75,
|
| 910 |
+
"completions/clipped_ratio": 0.0,
|
| 911 |
+
"completions/max_length": 33.8,
|
| 912 |
+
"completions/max_terminated_length": 33.8,
|
| 913 |
+
"completions/mean_length": 18.75,
|
| 914 |
+
"completions/mean_terminated_length": 18.75,
|
| 915 |
+
"completions/min_length": 8.4,
|
| 916 |
+
"completions/min_terminated_length": 8.4,
|
| 917 |
+
"epoch": 0.0226928895612708,
|
| 918 |
+
"frac_reward_zero_std": 0.7,
|
| 919 |
+
"grad_norm": 0.010602269321680069,
|
| 920 |
+
"kl": 1.1651100397109986,
|
| 921 |
+
"learning_rate": 3.6507936507936514e-07,
|
| 922 |
+
"loss": 0.0012,
|
| 923 |
+
"num_tokens": 514331.0,
|
| 924 |
+
"reward": 5.8875,
|
| 925 |
+
"reward_std": 0.18273502588272095,
|
| 926 |
+
"rewards/check_coherence/mean": 1.4625,
|
| 927 |
+
"rewards/check_coherence/std": 0.075,
|
| 928 |
+
"rewards/check_response_quality/mean": 2.4625,
|
| 929 |
+
"rewards/check_response_quality/std": 0.053867512941360475,
|
| 930 |
+
"rewards/match_format_approximately/mean": 0.9625,
|
| 931 |
+
"rewards/match_format_approximately/std": 0.053867512941360475,
|
| 932 |
+
"rewards/match_format_exactly/mean": 1.0,
|
| 933 |
+
"rewards/match_format_exactly/std": 0.0,
|
| 934 |
+
"step": 330
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"completion_length": 13.8,
|
| 938 |
+
"completions/clipped_ratio": 0.0,
|
| 939 |
+
"completions/max_length": 20.7,
|
| 940 |
+
"completions/max_terminated_length": 20.7,
|
| 941 |
+
"completions/mean_length": 13.8,
|
| 942 |
+
"completions/mean_terminated_length": 13.8,
|
| 943 |
+
"completions/min_length": 8.9,
|
| 944 |
+
"completions/min_terminated_length": 8.9,
|
| 945 |
+
"epoch": 0.023380552881309313,
|
| 946 |
+
"frac_reward_zero_std": 0.9,
|
| 947 |
+
"grad_norm": 0.007794269360601902,
|
| 948 |
+
"kl": 1.1532993257045745,
|
| 949 |
+
"learning_rate": 2.0634920634920635e-07,
|
| 950 |
+
"loss": 0.0012,
|
| 951 |
+
"num_tokens": 528547.0,
|
| 952 |
+
"reward": 5.975,
|
| 953 |
+
"reward_std": 0.05,
|
| 954 |
+
"rewards/check_coherence/mean": 1.475,
|
| 955 |
+
"rewards/check_coherence/std": 0.05,
|
| 956 |
+
"rewards/check_response_quality/mean": 2.5,
|
| 957 |
+
"rewards/check_response_quality/std": 0.0,
|
| 958 |
+
"rewards/match_format_approximately/mean": 1.0,
|
| 959 |
+
"rewards/match_format_approximately/std": 0.0,
|
| 960 |
+
"rewards/match_format_exactly/mean": 1.0,
|
| 961 |
+
"rewards/match_format_exactly/std": 0.0,
|
| 962 |
+
"step": 340
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"completion_length": 13.9,
|
| 966 |
+
"completions/clipped_ratio": 0.0,
|
| 967 |
+
"completions/max_length": 21.0,
|
| 968 |
+
"completions/max_terminated_length": 21.0,
|
| 969 |
+
"completions/mean_length": 13.9,
|
| 970 |
+
"completions/mean_terminated_length": 13.9,
|
| 971 |
+
"completions/min_length": 9.0,
|
| 972 |
+
"completions/min_terminated_length": 9.0,
|
| 973 |
+
"epoch": 0.02406821620134782,
|
| 974 |
+
"frac_reward_zero_std": 0.8,
|
| 975 |
+
"grad_norm": 3.0694658756256104,
|
| 976 |
+
"kl": 1.2572884202003478,
|
| 977 |
+
"learning_rate": 4.7619047619047627e-08,
|
| 978 |
+
"loss": 0.0013,
|
| 979 |
+
"num_tokens": 540139.0,
|
| 980 |
+
"reward": 5.95,
|
| 981 |
+
"reward_std": 0.1,
|
| 982 |
+
"rewards/check_coherence/mean": 1.45,
|
| 983 |
+
"rewards/check_coherence/std": 0.1,
|
| 984 |
+
"rewards/check_response_quality/mean": 2.5,
|
| 985 |
+
"rewards/check_response_quality/std": 0.0,
|
| 986 |
+
"rewards/match_format_approximately/mean": 1.0,
|
| 987 |
+
"rewards/match_format_approximately/std": 0.0,
|
| 988 |
+
"rewards/match_format_exactly/mean": 1.0,
|
| 989 |
+
"rewards/match_format_exactly/std": 0.0,
|
| 990 |
+
"step": 350
|
| 991 |
}
|
| 992 |
],
|
| 993 |
"logging_steps": 10,
|
| 994 |
+
"max_steps": 350,
|
| 995 |
+
"num_input_tokens_seen": 540139,
|
| 996 |
"num_train_epochs": 1,
|
| 997 |
"save_steps": 10,
|
| 998 |
"stateful_callbacks": {
|
|
|
|
| 1002 |
"should_evaluate": false,
|
| 1003 |
"should_log": false,
|
| 1004 |
"should_save": true,
|
| 1005 |
+
"should_training_stop": true
|
| 1006 |
},
|
| 1007 |
"attributes": {}
|
| 1008 |
}
|
training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 7313
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:99dd8fd386f6c88582c8ee70c2a210ac339045a5b9e54e02f9fae15a6dfb2047
|
| 3 |
size 7313
|