Research map / Vision-language and multimodal / Visual question answering
FiTs: Fine-Grained Two-Stage Training for Knowledge-Aware Question Answering
Qichen Ye, Bowen Cao, Nuo Chen, Weiyuan Xu, Yuexian Zou
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Add FiTs two-stage training to GreaseLM | GreaseLM · CommonsenseQA IHtest | accuracy: 73.6 -> 76.2 | Improved |
| Add FiTs two-stage training to GreaseLM | GreaseLM · OpenBookQA test | accuracy: 84.2 -> 86.0 | Improved |
| Add FiTs two-stage training to GreaseLM | GreaseLM · MedQA-USMLE test | accuracy: 38.6 -> 39.2 | Improved |
| Add MLM post-training objective (ablation) | GreaseLM · CommonsenseQA IHtest | accuracy: 73.6 -> 74.1 | Improved |
| Add knowledge adaptive (KA) post-training objective (ablation) | GreaseLM · CommonsenseQA IHtest | accuracy: 73.6 -> 74.3 | Improved |
| Add both MLM and KA post-training objectives (ablation) | GreaseLM · CommonsenseQA IHtest | accuracy: 73.6 -> 74.5 | Improved |
| Add post-training (ablation) | GreaseLM · CommonsenseQA test-reason | accuracy: 73.4 -> 73.9 | Improved |
| Add post-training (ablation) | GreaseLM · CommonsenseQA test-param | accuracy: 69.0 -> 70.9 | Improved |
| Add knowledge-aware fine-tuning (ablation) | GreaseLM · CommonsenseQA test-reason | accuracy: 73.4 -> 74.2 | Improved |
| Add knowledge-aware fine-tuning (ablation) | GreaseLM · CommonsenseQA test-param | accuracy: 69.0 -> 71.2 | Improved |
| Add KSD without KBR (ablation) | GreaseLM fine-tuning · CommonsenseQA IHtest | accuracy | Improved |
| Add KBR without KSD (ablation) | GreaseLM fine-tuning · CommonsenseQA IHtest | accuracy | Hurt |
Closest papers by meaning
- Cost-efficient Knowledge-based Question Answering with Large Language Models (NeurIPS 2024 · similarity 0.90)
- FlexKBQA: A Flexible LLM-Powered Framework for Few-Shot Knowledge Base Question Answering (AAAI 2024 · similarity 0.90)
- Code-Style In-Context Learning for Knowledge-Based Question Answering (AAAI 2024 · similarity 0.90)
- Promoting Knowledge Base Question Answering by Directing LLMs to Generate Task-relevant Logical Forms (AAAI 2025 · similarity 0.89)
- Graph Reasoning Transformers for Knowledge-Aware Question Answering (AAAI 2024 · similarity 0.88)
- KBQA-o1: Agentic Knowledge Base Question Answering with Monte Carlo Tree Search (ICML 2025 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
