#post-training3 articles

3 articles

AIBullisharXiv – CS AI · 6h ago1

🧠

Improving Text-to-Image Generation with Intrinsic Self-Confidence Rewards

Researchers introduced ARC (Adaptive Rewarding by self-Confidence), a new framework for improving text-to-image generation models through self-confidence signals rather than external rewards. The method uses internal self-denoising probes to evaluate model accuracy and converts this into scalar rewards for unsupervised optimization, showing improvements in compositional generation and text-image alignment.

AINeutralarXiv – CS AI · 6h ago1

🧠

Theoretical Perspectives on Data Quality and Synergistic Effects in Pre- and Post-Training Reasoning Models

New theoretical research analyzes how Large Language Models learn during pretraining versus post-training phases, revealing that balanced pretraining data creates latent capabilities activated later, while supervised fine-tuning works best on small, challenging datasets and reinforcement learning requires large-scale data that isn't overly difficult.

AIBullisharXiv – CS AI · 6h ago3

🧠

Surgical Post-Training: Cutting Errors, Keeping Knowledge

Researchers introduce Surgical Post-Training (SPoT), a new method to improve Large Language Model reasoning while preventing catastrophic forgetting. SPoT achieved 6.2% accuracy improvement on Qwen3-8B using only 4k data pairs and 28 minutes of training, offering a more efficient alternative to traditional post-training approaches.