[EMNLP 25] Personalized Language Models via Privacy-Preserving Evolutionary Model Merging

Official implementation of Wan et al's paper "Everyone's Voice Matters: Quantifying Annotation Disagreement Using Demographic Information" (AAAI 2023)

Official implementation of Tabular Transfer Learning via Prompting LLMs (COLM 2024).

Official implementation of Optimized Feature Generation for Tabular Data via LLMs with Decision Tree Reasoning (NeurIPS 2024).

Steer LLM outputs towards a certain topic/subject and enhance response capabilities using activation engineering by adding steering vectors

An Easy-to-use, Scalable and High-performance RLHF Framework based on Ray (PPO & GRPO & REINFORCE++ & vLLM & Ray & Dynamic Sampling & Async Agentic RL)

[ICLR'25 Oral] Representation Alignment for Generation: Training Diffusion Transformers Is Easier Than You Think

Lightweight Adapting for Black-Box Large Language Models

Official implementation of Hierarchical Context Merging: Better Long Context Understanding for Pre-trained LLMs (ICLR 2024).

Contriever: Unsupervised Dense Information Retrieval with Contrastive Learning

This includes the original implementation of SELF-RAG: Learning to Retrieve, Generate and Critique through self-reflection by Akari Asai, Zeqiu Wu, Yizhong Wang, Avirup Sil, and Hannaneh Hajishirzi.

Online Adaptation of Language Models with a Memory of Amortized Contexts (NeurIPS 2024)

800,000 step-level correctness labels on LLM solutions to MATH problems

Benchmarking large language models' complex reasoning ability with chain-of-thought prompting

Official Code for the paper "SuRe: Summarizing Retrievals using Answer Candidates for Open-domain QA of LLMs" (ICLR 2024)

Parkar and Kim et al.'s paper on :SelectLLM: Can LLMs Select Important Instructions to Annotate?"

An automatic evaluator for instruction-following language models. Human-validated, high-quality, cheap, and fast.

Official repository of NEFTune: Noisy Embeddings Improves Instruction Finetuning

Codes for papers on Large Language Models Personalization (LaMP)

Code for the paper "RoAST: Robustifying Language Models via Adversarial Perturbation with Selective Training" (EMNLP 2023)

Reference implementation for DPO (Direct Preference Optimization)