Research map / Large language models / Retrieval and knowledge
Contrastive Learning Reduces Hallucination in Conversations
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 |
|---|---|---|---|
| Add the mixed contrast objective (ablation) | MixCL · Wizard-of-Wikipedia test seen | Knowledge-F1: 19.1 -> 22.3 | Helped |
| Add the mixed contrast objective (ablation) | MixCL · Wizard-of-Wikipedia test unseen | Knowledge-F1: 16.9 -> 18.0 | Helped |
| Use hard negative sampling instead of random negatives (ablation) | MixCL · Wizard-of-Wikipedia test seen | Knowledge-F1: 20.8 -> 22.3 | Helped |
| Use hard negative sampling instead of random negatives (ablation) | MixCL · Wizard-of-Wikipedia test unseen | Knowledge-F1: 17.4 -> 18.0 | Helped |
| Add model-generated negatives (ablation) | MixCL · Wizard-of-Wikipedia test seen | Knowledge-F1: 21.7 -> 22.3 | Helped |
| Add model-generated negatives (ablation) | MixCL · Wizard-of-Wikipedia test unseen | Knowledge-F1: 17.5 -> 18.0 | Helped |
| Add the language model loss (ablation) | MixCL · Wizard-of-Wikipedia test unseen | Knowledge-F1: 16.7 -> 18.0 | Helped |
| Add the language model loss (ablation) | MixCL · Wizard-of-Wikipedia test seen | Knowledge-F1: 21.8 -> 22.3 | Helped |
| Use the mixed contrast objective and negative sampling rather than MLE alone (ablation) | MixCL · Wizard-of-Wikipedia test seen | Knowledge-F1: 18.9 -> 22.3 | Helped |
| Use the mixed contrast objective and negative sampling rather than MLE alone (ablation) | MixCL · Wizard-of-Wikipedia test unseen | Knowledge-F1: 16.0 -> 18.0 | Helped |
Closest papers by meaning
- Refine Knowledge of Large Language Models via Adaptive Contrastive Learning (ICLR 2025 · similarity 0.90)
- The Mirage of Performance Gains: Why Contrastive Decoding Fails to Mitigate Object Hallucinations in MLLMs? (NeurIPS 2025 · similarity 0.88)
- Hallucination Augmented Contrastive Learning for Multimodal Large Language Model (CVPR 2024 · similarity 0.87)
- Decoupling Contrastive Decoding: Robust Hallucination Mitigation in Multimodal Large Language Models (NeurIPS 2025 · similarity 0.87)
- DoLa: Decoding by Contrasting Layers Improves Factuality in Large Language Models (ICLR 2024 · similarity 0.86)
- Improving the Robustness of Knowledge-Grounded Dialogue via Contrastive Learning (AAAI 2024 · similarity 0.86)
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.