DriCo-adapters / train_args.json
anonymous-24421's picture
Upload DriCo LoRA adapters
8a0e578 verified
Raw
History Blame Contribute Delete
1.66 kB
{
"data_dir": "dataset/planner/",
"base_model": "Qwen/Qwen2.5-7B-Instruct",
"output_dir": "out/",
"resume_lora_path": "",
"prompt_module": "prompts.prompt",
"recipe_dir": "prompts/recipe",
"reference_dir": "prompts/reference",
"only_actor": "all",
"system_filter": "planner",
"drop_wait": false,
"train_on_prompt": false,
"epochs": 3,
"batch_size": 8,
"grad_accum_steps": 1,
"lr": 1e-05,
"weight_decay": 0.0,
"warmup_ratio": 0.01,
"max_grad_norm": 1.0,
"max_length": 2048,
"valid_ratio": 0.0,
"log_every": 10,
"seed": 42,
"lora_r": 64,
"lora_alpha": 32,
"lora_dropout": 0.1,
"target_modules": [
"q_proj",
"k_proj",
"v_proj",
"o_proj",
"gate_proj",
"up_proj",
"down_proj"
],
"fp16": false,
"bf16": true,
"eval_seen_recipes": [
"1_baked_eggplant",
"4_sliced_beef_and_lentil_stew",
"6_zucchini_lentil_and_mushroom_patty"
],
"eval_unseen_recipes": [
"1_baked_bell_pepper",
"4_sliced_pumpkin_and_chickpea_stew",
"6_zucchini_green_pea_and_onion_patty"
],
"holdout_recipes": [
"1_baked_bell_pepper",
"4_sliced_pumpkin_and_chickpea_stew",
"6_zucchini_green_pea_and_onion_patty"
],
"exclude_holdout_from_train": true,
"eval_every": 1,
"eval_every_steps": 0,
"eval_episodes": 1,
"eval_horizon": 120,
"eval_layout": "new_env",
"eval_chunk_size": 3,
"eval_chunk_max_steps": 50,
"eval_match_threshold": 0.3,
"eval_stall_limit": 10,
"eval_max_new_tokens": 100,
"eval_verbose": true,
"eval_slot_threshold": 0.66,
"wandb_project": "collab-overcooked",
"wandb_run_name": "planner_only",
"use_flash_attn": false
}