Upload ao_config.json with huggingface_hub
Browse files- ao_config.json +64 -0
ao_config.json
ADDED
|
@@ -0,0 +1,64 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"layer_combinations": [
|
| 3 |
+
[
|
| 4 |
+
25,
|
| 5 |
+
50,
|
| 6 |
+
75
|
| 7 |
+
]
|
| 8 |
+
],
|
| 9 |
+
"act_layer_combinations": [
|
| 10 |
+
[
|
| 11 |
+
9,
|
| 12 |
+
18,
|
| 13 |
+
27
|
| 14 |
+
]
|
| 15 |
+
],
|
| 16 |
+
"schema_version": 1,
|
| 17 |
+
"special_token": " ?",
|
| 18 |
+
"prefix_template": "Layer: {layer}\\n{special_token} * {num_positions} \\n",
|
| 19 |
+
"model_name": "Qwen/Qwen3-8B",
|
| 20 |
+
"hook_onto_layer": 1,
|
| 21 |
+
"use_decoder_vectors": true,
|
| 22 |
+
"generation_kwargs": {
|
| 23 |
+
"do_sample": false,
|
| 24 |
+
"max_new_tokens": 20
|
| 25 |
+
},
|
| 26 |
+
"steering_coefficient": 1.0,
|
| 27 |
+
"use_lora": true,
|
| 28 |
+
"lora_r": 64,
|
| 29 |
+
"lora_alpha": 128,
|
| 30 |
+
"lora_dropout": 0.05,
|
| 31 |
+
"lora_target_modules": "all-linear",
|
| 32 |
+
"dataset_configs": [],
|
| 33 |
+
"dataset_loader_names": [],
|
| 34 |
+
"dataset_folder": "",
|
| 35 |
+
"train_batch_size": 16,
|
| 36 |
+
"eval_batch_size": 128,
|
| 37 |
+
"train_batches_per_materialization_block": 16,
|
| 38 |
+
"num_epochs": 1,
|
| 39 |
+
"lr": 1e-05,
|
| 40 |
+
"gradient_accumulation_steps": 1,
|
| 41 |
+
"max_grad_norm": 1.0,
|
| 42 |
+
"eval_steps": 999999,
|
| 43 |
+
"eval_on_start": false,
|
| 44 |
+
"gradient_checkpointing": true,
|
| 45 |
+
"window_mult": 20,
|
| 46 |
+
"save_steps": 5000,
|
| 47 |
+
"save_dir": "checkpoints",
|
| 48 |
+
"max_train_examples": null,
|
| 49 |
+
"seed": 42,
|
| 50 |
+
"eval_logs_path": "",
|
| 51 |
+
"load_lora_path": null,
|
| 52 |
+
"created_at_utc": "",
|
| 53 |
+
"git_commit": "",
|
| 54 |
+
"wandb_project": "cot-oracle",
|
| 55 |
+
"wandb_run_name": "",
|
| 56 |
+
"wandb_suffix": "",
|
| 57 |
+
"hf_push_to_hub": false,
|
| 58 |
+
"hf_private_repo": false,
|
| 59 |
+
"hf_repo_name": "",
|
| 60 |
+
"hf_repo_id": "",
|
| 61 |
+
"load_in_8bit": false,
|
| 62 |
+
"open_ended_eval_include": null,
|
| 63 |
+
"positive_negative_examples": false
|
| 64 |
+
}
|