├── LICENSE ├── README.md ├── configs ├── llama_full_sft_8b.yaml ├── qwen_full_1.5b_mdpo.yaml ├── qwen_full_14b_mdpo.yaml ├── qwen_full_sft_1.5b.yaml ├── qwen_full_sft_14b.yaml ├── qwen_lora_32b_mdpo.yaml ├── qwen_lora_8b_mdpo.yaml └── qwen_lora_sft_32b.yaml ├── eval_datasets └── bamboogle │ └── test_subsampled.jsonl ├── inference ├── decompose_vllm.py ├── main_qa.py ├── main_reasoning.py └── utils.py ├── requirements.txt └── rollout ├── create_dpo_pairs.py ├── create_train_data.py ├── rs_cot.py ├── rs_mhqa.py ├── rs_pot.py └── utils.py /LICENSE: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ritaranx/AceSearcher/HEAD/LICENSE -------------------------------------------------------------------------------- /README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ritaranx/AceSearcher/HEAD/README.md -------------------------------------------------------------------------------- /configs/llama_full_sft_8b.yaml: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ritaranx/AceSearcher/HEAD/configs/llama_full_sft_8b.yaml -------------------------------------------------------------------------------- /configs/qwen_full_1.5b_mdpo.yaml: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ritaranx/AceSearcher/HEAD/configs/qwen_full_1.5b_mdpo.yaml -------------------------------------------------------------------------------- /configs/qwen_full_14b_mdpo.yaml: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ritaranx/AceSearcher/HEAD/configs/qwen_full_14b_mdpo.yaml -------------------------------------------------------------------------------- /configs/qwen_full_sft_1.5b.yaml: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ritaranx/AceSearcher/HEAD/configs/qwen_full_sft_1.5b.yaml -------------------------------------------------------------------------------- /configs/qwen_full_sft_14b.yaml: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ritaranx/AceSearcher/HEAD/configs/qwen_full_sft_14b.yaml -------------------------------------------------------------------------------- /configs/qwen_lora_32b_mdpo.yaml: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ritaranx/AceSearcher/HEAD/configs/qwen_lora_32b_mdpo.yaml -------------------------------------------------------------------------------- /configs/qwen_lora_8b_mdpo.yaml: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ritaranx/AceSearcher/HEAD/configs/qwen_lora_8b_mdpo.yaml -------------------------------------------------------------------------------- /configs/qwen_lora_sft_32b.yaml: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ritaranx/AceSearcher/HEAD/configs/qwen_lora_sft_32b.yaml -------------------------------------------------------------------------------- /eval_datasets/bamboogle/test_subsampled.jsonl: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ritaranx/AceSearcher/HEAD/eval_datasets/bamboogle/test_subsampled.jsonl -------------------------------------------------------------------------------- /inference/decompose_vllm.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ritaranx/AceSearcher/HEAD/inference/decompose_vllm.py -------------------------------------------------------------------------------- /inference/main_qa.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ritaranx/AceSearcher/HEAD/inference/main_qa.py -------------------------------------------------------------------------------- /inference/main_reasoning.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ritaranx/AceSearcher/HEAD/inference/main_reasoning.py -------------------------------------------------------------------------------- /inference/utils.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ritaranx/AceSearcher/HEAD/inference/utils.py -------------------------------------------------------------------------------- /requirements.txt: -------------------------------------------------------------------------------- 1 | numpy 2 | torch 3 | faiss-cpu==1.6.4 4 | tqdm 5 | transformers==4.51.0 6 | vllm==0.8.2 7 | -------------------------------------------------------------------------------- /rollout/create_dpo_pairs.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ritaranx/AceSearcher/HEAD/rollout/create_dpo_pairs.py -------------------------------------------------------------------------------- /rollout/create_train_data.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ritaranx/AceSearcher/HEAD/rollout/create_train_data.py -------------------------------------------------------------------------------- /rollout/rs_cot.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ritaranx/AceSearcher/HEAD/rollout/rs_cot.py -------------------------------------------------------------------------------- /rollout/rs_mhqa.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ritaranx/AceSearcher/HEAD/rollout/rs_mhqa.py -------------------------------------------------------------------------------- /rollout/rs_pot.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ritaranx/AceSearcher/HEAD/rollout/rs_pot.py -------------------------------------------------------------------------------- /rollout/utils.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ritaranx/AceSearcher/HEAD/rollout/utils.py --------------------------------------------------------------------------------