ReDraft, Don't Just Distill: Reference-Driven Revision for Continual VLLM Post-Training
cs.AI
Submitted: 2026-09-15
Updated: 2026-09-22
Comments: 37pages, preprint
Code: https://github.com/huggingface/trl
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Terminology
Sources
- Qwen2.5-VL Technical Report
- Constitutional AI: Harmlessness from AI Feedback
- Learning to Foresee: Unveiling the Unlocking Efficiency of On-Policy Distillation
- Retaining by Doing: The Role of On-Policy Data in Mitigating Forgetting
- Revisiting On-Policy Distillation: Empirical Failure Modes and Simple Fixes
- Reinforced Self-Training (ReST) for Language Modeling
- Self-Distillation Zero: Self-Revision Turns Binary Rewards into Dense Supervision
- Distilling the Knowledge in a Neural Network
- Uni-OPD: Unifying On-Policy Distillation with a Dual-Perspective Recipe
- Does Math Reasoning Improve General LLM Capabilities? Understanding Transferability of LLM Reasoning
- Reinforcement Learning via Self-Distillation
- Bridging Reasoning Trajectories in On-Policy Distillation via Near-Future Guidance
- Overcoming catastrophic forgetting in neural networks
- Reinforcement Fine-Tuning Naturally Mitigates Forgetting in Continual Post-Training
- Tulu 3: Pushing Frontiers in Open Language Model Post-Training
- Surgical Post-Training: Proximal On-Policy Distillation for Reasoning with Knowledge Retention
- An Empirical Study of Catastrophic Forgetting in Large Language Models During Continual Fine-tuning
- Progressive Neural Networks
- Proximal Policy Optimization Algorithms
- DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models
Related papers
- MAVEN-T: Reinforced Heterogeneous Distillation for Real-Time Multi-Agent Trajectory Prediction
- Model Discovery Agent: LLM-assisted Bayesian experiment design for data-efficient discovery of mechanistic world models
- The Clinician's Veto: Navigating Trust, Liability, and Uncertainty in Autonomous AI Prescribing
- MindHelper: Closed-Loop Embodied Mental-State Reasoning for Precision Intervention
- Incumbent Advantage: Brand Bias and Cognitive Manipulation Dynamics in LLM Recommendation Systems
- VSAL: A Vision Solver with Adaptive Layouts for Graph Property Detection