dynamic-fine-tuning

Tag

Cards List
#dynamic-fine-tuning

@maximelabonne: This is so neat! Dynamic Fine-Tuning (DFT) reweights the SFT loss by the model's own token probability, which creates a…

X AI KOLs Following · 2026-05-20 Cached

Dynamic Fine-Tuning (DFT) is introduced as a method that reweights the SFT loss using the model's own token probability, creating a feedback loop, and adds forward KL to penalize tokens the base model finds likely but the policy has pushed toward zero probability. The tweet expresses skepticism about SFT papers in practice but praises the attempt.

0 favorites 0 likes
#dynamic-fine-tuning

On the Generalization of SFT: A Reinforcement Learning Perspective with Reward Rectification

Papers with Code Trending · 2025-08-07 Cached

This paper analyzes limitations in standard supervised fine-tuning (SFT) from a reinforcement learning perspective and proposes Dynamic Fine-Tuning (DFT), a simple gradient-rescaling method that improves LLM generalization and matches offline RL performance.

0 favorites 0 likes
← Back to home

Submit Feedback