A comprehensive index of artificial intelligence and machine-learning research with AI-generated summaries, citation metrics, and direct links to papers and code.
Juan Escalante, Austin Pack, Alex Barrett
Two longitudinal studies show AI-generated writing feedback matches human tutor efficacy for ENL learners, with near-even student preference split.
Ichiro Tsuda, Edgar Koerner, Hideki Shimizu
This paper presents an asynchronous neural network model with field-effect feedback that enables unsupervised learning and deterministic chaotic successive memory recall.
Ali Darvishi, Hassan Khosravi, Shazia Sadiq, et al.
A randomized controlled experiment with 1625 students shows that AI assistance in peer feedback reduces student agency, as students rely on rather than learn from AI, and hybrid human-AI approaches are no more effective than AI alone.
Omar Shaikh, Michelle Lam, Joey Hejna, et al.
DITTO aligns LLMs to user-specific behaviors using fewer than 10 demonstrations via online imitation learning and preference optimization.
Aniruddha Roy, Pretam Ray, Abhilash Nandy, et al.
REFINE-AF uses reinforcement learning from automated feedback to generate high-quality instruction datasets from a small seed set, outperforming Self-Instruct.
Tianhao Wu, Janice Lan, Weizhe Yuan, et al.
Iteratively trains LLMs to generate hidden thoughts that improve response quality via preference optimization using only response-level feedback.
Unknown
ReST^EM iteratively self-trains language models via EM-like steps, using binary feedback to filter model-generated samples and fine-tune the base model, improving problem-solving without human data.
David A Cook, Joshua Overgaard, V Shane Pankratz, et al.
LLM-powered virtual patients can generate authentic clinician-patient dialogues, represent preferences, and provide personalized feedback at low cost.
Unknown
Prometheus 2 introduces 7B and 8x7B evaluation LLMs that achieve high correlations with human and proprietary judges by merging Mistral models trained on curated feedback and preference collections.
Seungone Kim, Juyoung Suk, Shayne Longpre, et al.
Introduces Prometheus, a 13B fully open-source evaluation LLM trained on GPT-4-curated feedback data.
Unknown
Zephyr 7B uses distilled direct preference optimization (dDPO) and AI feedback data to improve intent alignment in chat-based language models.
Unknown
WizardMath enhances mathematical reasoning in Llama-2 via Reinforcement Learning from Evol-Instruct Feedback (RLEIF), outperforming many open and closed-source LLMs.