--- license: cc-by-sa-4.0 dataset_info: features: - name: id dtype: string - name: source dtype: string - name: task dtype: string - name: type dtype: string - name: instruction dtype: string - name: question dtype: string - name: options dtype: string - name: answer dtype: string - name: context dtype: string - name: evidence dtype: string splits: - name: test num_bytes: 1966003072 num_examples: 1934 download_size: 660289122 dataset_size: 1966003072 configs: - config_name: default data_files: - split: test path: data/test-* ---

LooGLE v2

**The official repository of "LooGLE v2: Are LLMs Ready for Real World Long Dependency Challenges?"** **NeurIPS DB Track 2025**
Dataset Website Paper License
--- ## πŸ“‹ Overview LooGLE v2 is a comprehensive benchmark designed to evaluate large language models on their ability to understand and process long-context documents with complex dependencies. The benchmark covers diverse domains including **Finance**, **Law**, **Code**, and **Game**. --- ## πŸš€ Quick Start ### πŸ“¦ Installation ```bash # Create environment with Python 3.10 conda create -n loogle-v2 python=3.10 conda activate loogle-v2 # Install dependencies pip install -r requirements.txt # Install Flash Attention pip install flash-attn==2.6.3 --no-build-isolation # Or you can download flash_attn-2.6.3-cp310-cp310-linux_x86_64.whl pip install flash_attn-2.6.3-cp310-cp310-linux_x86_64.whl ``` --- ## πŸ“Š Dataset Download the LooGLE v2 dataset from Hugging Face: ```bash git clone https://huggingface.co/datasets/MuLabPKU/LooGLE-v2 ./datasets/LooGLE-v2 # Or use the Hugging Face CLI to download: hf download MuLabPKU/LooGLE-v2 --repo-type dataset --local-dir ./datasets/LooGLE-v2 ``` --- ## πŸ› οΈ Usage ### βš™οΈ Configuration **vLLM server (for `predict.py`):** ```bash python -m vllm.entrypoints.openai.api_server \ --model path/to/your/model \ --port 8000 \ --max-model-len 131072 ``` **Model entry (`config/models.jsonl`, shared by both scripts):** ```json { "name": "your-model-name", "model": "path/to/model", "max_len": 131072, "base_url": "http://localhost:8000/v1", "api_key": "your-api-key" } ``` Transformers mode (`predict_transformers.py`) does not need a server; it still reuses `name/model/max_len` from this config. Ensure `base_url` matches your vLLM port when using the server route. ### πŸ”Ž Pre-compute RAG Contexts (optional) If you plan to run `--use_rag`, first generate `context_rag` with the preprocessor: ```bash python rag_preprocess.py \ --input_path ./datasets/LooGLE-v2 \ --split test \ --output_path ./datasets/LooGLE-v2/test_rag.jsonl \ --embedding_model THUDM/LongCite-glm4-9b \ --devices 0,1 ``` For multi-turn refinement (using a generator model to iteratively improve retrieval queries): ```bash python rag_preprocess.py \ --input_path ./datasets/LooGLE-v2 \ --split test \ --output_path ./datasets/LooGLE-v2/test_rag_multi.jsonl \ --embedding_model THUDM/LongCite-glm4-9b \ --generator_model meta-llama/Llama-3.1-8B \ --multi_turn --devices 0,1 ``` ### 🎯 Running Predictions #### Option A: vLLM server (`predict.py`) ```bash python predict.py \ --model your-model-name \ --data_dir ./datasets/LooGLE-v2 \ --save_dir ./results \ --max_new_tokens 512 ``` #### Option B: Transformers local (`predict_transformers.py`) ```bash python predict_transformers.py \ --model your-model-name \ --data_dir ./datasets/LooGLE-v2 \ --save_dir ./results \ --max_new_tokens 512 ``` Optional prompting flags (both scripts): - `--use_cot` for Chain-of-Thought - `--use_rag --rag_topk --rag_context ` to inject precomputed `context_rag` (default file: `./datasets/LooGLE-v2/test_rag.jsonl`)
πŸ“ Core parameters (both options) | Flag | Purpose | |------|---------| | `--model` | Must match `config/models.jsonl` name | | `--data_dir` | Dataset path (jsonl or HF) | | `--save_dir` | Output directory | | `--with_context` | 1/0 to include original context | | `--n_proc` | Parallel processes | | `--max_new_tokens` | Generation length | | `--use_cot` | Enable Chain-of-Thought | | `--use_rag` | Use retrieved context | | `--rag_topk` | How many retrieved chunks to keep | | `--rag_context` | Path to `id + context_rag` jsonl |
πŸ–₯️ Transformers-only flags | Flag | Purpose | |------|---------| | `--device` | Target device (cuda/cpu, auto by default) | | `--load_in_8bit` | 8-bit quantization (needs bitsandbytes) | | `--load_in_4bit` | 4-bit quantization (needs bitsandbytes) | | `--torch_dtype` | Weight dtype: float16/bfloat16/float32 | > πŸ’‘ Install `bitsandbytes` to enable quantization: `pip install bitsandbytes`
### πŸ“ˆ Evaluation After prediction, evaluate the results: ```bash python evaluate.py --input_path ./results/your-model-name.jsonl ``` This outputs per-task accuracy for each domain and overall accuracy. For batch evaluation (e.g., multiple runs with CoT/RAG or no-context variants): ```bash python evaluate.py --input_path ./results --batch --output_json ./results/summary.json ``` This scans a folder for `.jsonl` files, reports each file’s accuracy, and optionally saves a summary. --- ## πŸ“ Project Structure ``` LooGLE-v2/ β”œβ”€β”€ src/ β”‚ β”œβ”€β”€ answer_extractor.py # Answer extraction logic β”‚ β”œβ”€β”€ evaluator.py # Evaluation metrics β”‚ β”œβ”€β”€ llm_client.py # LLM client implementations β”‚ β”œβ”€β”€ data_loader.py # Data loading utilities β”‚ └── utils.py # Common utilities β”œβ”€β”€ config/ β”‚ └── models.jsonl # Model configurations β”œβ”€β”€ predict.py # Prediction script (vLLM server) β”œβ”€β”€ predict_transformers.py # Prediction script (direct transformers) β”œβ”€β”€ rag_preprocess.py # RAG context preprocessing β”œβ”€β”€ evaluate.py # Evaluation script └── requirements.txt # Dependencies ``` --- ## πŸ“„ Results Format Prediction outputs are saved in JSONL format: ```json { "id": "sample_id", "source": "Finance", "task": "Metric Calculation", "type": "question_type", "correct_answer": "123.45", "pred_answer": "123.40", "response": "The correct answer is 123.40", "judge": true } ``` --- ## πŸ“– Citation If you use LooGLE v2 in your research, please cite: ```bibtex @article{he2025loogle, title={LooGLE v2: Are LLMs Ready for Real World Long Dependency Challenges?}, author={He, Ziyuan and Wang, Yuxuan and Li, Jiaqi and Liang, Kexin and Zhang, Muhan}, journal={arXiv preprint arXiv:2510.22548}, year={2025} } ``` --- ## πŸ“œ License This project is licensed under the MIT License - see the [LICENSE](LICENSE) file for details. ---