zjunlp/LookAheadTuning
[WSDM 2026] LookAhead Tuning: Safer Language Models via Partial Answer Previews
This project helps large language model (LLM) developers make their models safer and more reliable. By modifying training data to include 'partial answer previews', you can train models that are less prone to generating unsafe or incorrect outputs. You provide your existing LLM training datasets, and it outputs modified datasets ready for a more robust fine-tuning process, specifically designed for those building or enhancing LLMs.
Use this if you are a machine learning engineer or researcher responsible for fine-tuning large language models and need to improve their safety and reduce undesirable outputs.
Not ideal if you are an end-user simply looking to apply an existing language model without deep involvement in its training or fine-tuning process.
Stars
17
Forks
—
Language
Python
License
MIT
Category
Last pushed
Dec 14, 2025
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/llm-tools/zjunlp/LookAheadTuning"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
aalok-sathe/surprisal
A unified interface for computing surprisal (log probabilities) from language models! Supports...
EvolvingLMMs-Lab/lmms-engine
A simple, unified multimodal models training engine. Lean, flexible, and built for hacking at scale.
FunnySaltyFish/Better-Ruozhiba
【逐条处理完成】人为审核+修改每一条的弱智吧精选问题QA数据集
reasoning-machines/pal
PaL: Program-Aided Language Models (ICML 2023)
microsoft/monitors4codegen
Code and Data artifact for NeurIPS 2023 paper - "Monitor-Guided Decoding of Code LMs with Static...