RAVine-dense-index

This repository contains dense index files for the search tools of the RAVine: Reality-Aligned Evaluation for Agentic Search framework. The corpus is MS MARCO V2.1, encoded using Alibaba-NLP/gte-modernbert-base.

Paper: RAVine: Reality-Aligned Evaluation for Agentic Search Code: https://github.com/SwordFaith/RAVine

Abstract

Agentic search, as a more autonomous and adaptive paradigm of retrieval augmentation, is driving the evolution of intelligent search systems. However, existing evaluation frameworks fail to align well with the goals of agentic search. First, the complex queries commonly used in current benchmarks often deviate from realistic user search scenarios. Second, prior approaches tend to introduce noise when extracting ground truth for end-to-end evaluations, leading to distorted assessments at a fine-grained level. Third, most current frameworks focus solely on the quality of final answers, neglecting the evaluation of the iterative process inherent to agentic search. To address these limitations, we propose RAVine -- a Reality-Aligned eValuation framework for agentic LLMs with search. RAVine targets multi-point queries and long-form answers that better reflect user intents, and introduces an attributable ground truth construction strategy to enhance the accuracy of fine-grained evaluation. Moreover, RAVine examines model's interaction with search tools throughout the iterative process, and accounts for factors of efficiency. We benchmark a series of models using RAVine and derive several insights, which we hope will contribute to advancing the development of agentic search systems. The code and datasets are available at this https URL .

Dataset Details

RAVine (a Reality-Aligned eValuation framework for agentic LLMs with search) is a comprehensive evaluation system for agentic search, encompassing the web environment, benchmark datasets, and a novel evaluation method. It serves as a full-process, reproducible, and goal-aligned evaluation sandbox. This RAVine-dense-index dataset specifically provides the dense index files used by the search tools within the RAVine framework.

The RAVine project also provides several other related datasets on the Hugging Face Hub:

• Queries & Nuggets: https://huggingface.co/datasets/sapphirex/RAVine-nuggets
• Raw Qrels: https://huggingface.co/datasets/sapphirex/RAVine-qrels
• URL-Docid Mapper: https://huggingface.co/datasets/sapphirex/RAVine-mapper
• Running logs (for reproduction): https://huggingface.co/datasets/sapphirex/RAVine-logs

Sample Usage

To effectively utilize this dense index dataset, it is intended to be used within the full RAVine framework. Below are the general steps to get started:

1. Clone the RAVine repository and install dependencies:

   git clone https://github.com/SwordFaith/RAVine.git
   cd RAVine
   # Install vllm environment (version 0.9.0.1 as specified in RAVine repo)
   pip install vllm==0.9.0.1
   # Install main program environment
   pip install -r src/requirements_agent.txt
   

2. Download this dataset (dense index) using Git LFS:

   git lfs install
   git clone https://huggingface.co/datasets/sapphirex/RAVine-dense-index
   

   You will need to modify the corresponding path variables in your RAVine running script (configs/ examples) to point to the downloaded index files.

3. Configure and Run the RAVine framework: First, write your configuration file (examples are available in the configs/ directory of the RAVine repository). Then, launch the vllm service and run the main evaluation program:

   bash scripts/server/vllm.sh your_config_file # run the server of agentic llm
   bash scripts/evaluation/run.sh your_config_file
   

   For more detailed and advanced usage scenarios, please refer to the official RAVine GitHub repository.