Find a story

Search Spins

Search titles, summaries, and missing voices across published articles — press releases, announcements, and media coverage.

3 results for “supervised fine-tuning”

SPIN Processed News Frame: The Hype

Capacity-Dependent Effects of Data Selection for Reasoning

A new arXiv preprint challenges the assumption that high-likelihood responses are universally optimal for reasoning-focused fine-tuning, demonstrating instead that data selection effectiveness depends critically on model size and training duration.

Spin 40% Claim Present in Source AI Risk Moderate
arXiv Machine Learning

Aug 17, 2026

SPIN Processed News Frame: The Hype

Weightless Fine-Tuning: Personalizing LLMs via Logit-Space Transport

A new method called Weightless Fine-Tuning (WFT) enables personalization of large language models at decoding time without updating model weights, reducing computational cost while approximating the distributional effect of supervised fine-tuning.

Spin 65% Claim Present in Source AI Risk High
arXiv Machine Learning

Aug 13, 2026

SPIN Processed News Frame: The Hype

Enough is as good as a feast: A Comprehensive Analysis of How Reinforcement Learning Mitigates Task Conflicts in LLMs

A new arXiv preprint claims reinforcement learning (RL) training reduces task conflicts during model merging in LLMs compared to supervised fine-tuning, citing three empirical and theoretical mechanisms.

Spin 45% Claim Present in Source AI Risk Moderate
arXiv Computation and Language

Jul 27, 2026