Research map / Probabilistic and scientific modeling / Variational and generative sampling
Variational Open-Domain Question Answering
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Adding distillation of the MCQA retriever into a query-only student retriever (ablation) | VOD retriever for FindZebra semantic search · FindZebra retrieval | MRR: 27.8 -> 31.7 | Helped |
| Combining the distilled VOD retriever with BM25 score at evaluation (ablation) | VOD distilled retriever · FindZebra retrieval | MRR: 31.7 -> 38.9 | Helped |
| Pretraining on MedMCQA before USMLE finetuning (vs USMLE-only training) (ablation) | VOD 2xBioLinkBERT · USMLE | test accuracy: 44.7 -> 55.0 | Helped |
| Replacing static BM25 retriever with VOD-learned retriever (2x BioLinkBERT) on MedMCQA (ablation) | BioLinkBERT reader-retriever · MedMCQA | test accuracy: 55.3 -> 62.9 | Helped |
| Annealing the RVB parameter alpha from 1 (ELBO) to 0 versus fixed ELBO or IWB bounds (ablation) | VOD training · ablation on variational bounds (P=100) | answering accuracy and retriever entropy | Helped |
| Increasing truncated retriever support P from 8/32 to 100 (ablation) | VOD truncated retriever · MCQA ablation on MedMCQA | answering accuracy | Helped |
| Using VOD (2x BioLinkBERT, 220M) instead of Med-PaLM (540B) | MedMCQA test · MedMCQA | test accuracy: 62.9 vs 57.6 | Helped |
| Using VOD on USMLE with MedMCQA pretraining, compared to Med-PaLM | USMLE test · USMLE | test accuracy: 55.0 vs 67.6 | Hurt |
Closest papers by meaning
- Making Retrieval-Augmented Language Models Robust to Irrelevant Context (ICLR 2024 · similarity 0.87)
- A Statistical Framework for Data-dependent Retrieval-Augmented Models (ICML 2024 · similarity 0.87)
- BLADE: Enhancing Black-Box Large Language Models with Small Domain-Specific Models (AAAI 2025 · similarity 0.86)
- SuRe: Summarizing Retrievals using Answer Candidates for Open-domain QA of LLMs (ICLR 2024 · similarity 0.86)
- FreeRet: MLLMs as Training-Free Retrievers (ICML 2026 · similarity 0.86)
- TAIA: Large Language Models are Out-of-Distribution Data Learners (NeurIPS 2024 · similarity 0.85)
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.