用ColBERT检索+GPT研究框架，5秒生成高准确答案
HLTCOE at LiveRAG: GPT-Researcher using ColBERT retrieval
- 用ColBERT双编码器检索，将文本转为密集向量加速查找
- 九段上下文输入，最终答案正确率1.07分（第5名）
- 适合想提升问答系统检索效率的研究者和开发者
HLTCOE在LiveRAG挑战中提交的系统基于GPT-Researcher框架，先通过上下文研究问题，再筛选结果并生成答案。检索模块采用ColBERT双编码器架构，将段落表示为多个密集向量；使用PLAID-X构建的本地压缩索引，基于FineWeb10-BT数据集，模型经微调以支持多语言检索。查询生成由Qwen2.5-7B-Instruct完成，过滤环节使用m2-bert-80M-8k-retrieval模型，最终利用Falcon3-10B模型，以最多九段文本作为上下文生成答案。该系统在LiveRAG自动评估中正确性得分1.07，位列第5。
原文摘要 · Abstract (English)
The HLTCOE LiveRAG submission utilized the GPT-researcher framework for researching the context of the question, filtering the returned results, and generating the final answer. The retrieval system was a ColBERT bi-encoder architecture, which represents a passage with many dense tokens. Retrieval used a local, compressed index of the FineWeb10-BT collection created with PLAID-X, using a model fine-tuned for multilingual retrieval. Query generation from context was done with Qwen2.5-7B-Instruct, while filtering was accomplished with m2-bert-80M-8k-retrieval. Up to nine passages were used as context to generate an answer using Falcon3-10B. This system placed 5th in the LiveRAG automatic evaluation for correctness with a score of 1.07.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。