Large Language Models

Chain of Thought prompting in LLMs offers limited generalizability, providing performance gains only when prompts are highly specific to problem types; highlighting a critical trade-off between perfor…

Chain of Preference Optimization (CPO) dramatically improves LLM reasoning by leveraging ToT’s search tree for efficient fine-tuning, achieving similar or better performance with significantly reduced…

Chain-of-Agents (CoA) framework uses multi-agent collaboration to efficiently process long contexts for LLMs, significantly improving performance on various tasks.

LLMs surprisingly master complex logic puzzles like Sudoku and Zebra puzzles after training on strategically ordered solution steps, revealing hidden reasoning abilities.

Cascade Speculative Drafting (CS Drafting) dramatically speeds up large language model inference by using a multi-stage drafting process, optimizing both time allocation and autoregressive generation.

Smaller language models can learn skill composition from limited examples, substantially improving their ability to combine skills in novel ways through fine-tuning.

LLMs can improve reasoning by teaching weaker models, a process called Learning by Teaching (LbT), as shown in this preliminary study. LbT enhances not just student models, but also the teacher model…

LLMs struggle to reliably generate valid data science code due to a lack of true understanding of numerical constraints in APIs, despite seemingly mastering common patterns through extensive training.

LLMs struggle with in-context exploration, needing substantial prompt engineering or training interventions to effectively explore multi-armed bandit environments.

LLM agents surprisingly exhibit human-like trust behavior, especially GPT-4, paving the way for simulating complex human interactions in various applications.

LLMs struggle with noisy rationales in chain-of-thought prompting. This paper introduces the NoRa dataset, showing that existing methods struggle. A new method, CD-CoT, significantly improves accura…

Language models learn to skip steps in reasoning, improving efficiency and generalization, showcasing emergent human-like cognitive abilities.

GNNs enhance LLM-based task planning by improving the ability to process task graphs, surpassing existing solutions even without training.

LLMs’ reasoning can be improved by using internal consistency to calibrate their outputs.

Cal-DPO calibrates implicit rewards in contrastive preference learning, dramatically improving large language model alignment with human preferences.

Training large language models efficiently is key; this paper shows how using structured feedforward layers and a novel training regime significantly reduces computational costs and improves training …

Buffer of Thoughts (BoT) boosts Large Language Model reasoning by storing and reusing high-level ’thought-templates’, achieving significant accuracy and efficiency gains across diverse tasks.

Householder Reflection Adaptation (HRA) bridges low-rank and orthogonal LLM adaptation, achieving superior performance with fewer parameters than existing methods. By using a chain of Householder refl…

Bridge-IF, a novel generative diffusion model, excels at inverse protein folding by learning probabilistic dependencies between protein structures and sequences, significantly outperforming existing m…

Boosting LLMs with an intelligent information assistant, ASSISTRAG, significantly improves accuracy and reasoning, especially for less advanced models.