#gradient-based2 articles

2 articles

AIBullisharXiv – CS AI · 10h ago7/10

🧠

Inference-time Alignment in Continuous Space

Researchers propose Simple Energy Adaptation (SEA), a new algorithm for aligning large language models with human feedback at inference time. SEA uses gradient-based sampling in continuous latent space rather than searching discrete response spaces, achieving up to 77.51% improvement on AdvBench and 16.36% on MATH benchmarks.

AINeutralarXiv – CS AI · 10h ago5/10

🧠

Jacobian Scopes: token-level causal attributions in LLMs

Researchers introduce Jacobian Scopes, a new gradient-based method for interpreting how individual tokens influence Large Language Model predictions. The technique uses perturbation theory and information geometry to reveal model biases, translation strategies, and learning mechanisms, with open-source implementations and an interactive demo available.

🏢 Hugging Face