CHATS-lab/verbalized-sampling

Controlled text generation with programmable constraints

An autoregressive character-level language model for making more things

A collection of LogitsProcessors to customize and enhance LLM behavior for specific tasks.

Probabilistic programming with large language models

Antidoom generates and trains targeted preference data for reducing model repetition loops (doom loops).

DataDreamer: Prompt. Generate Synthetic Data. Train & Align Models.   🤖💤

Interpretable Causal Diffusion Language Models

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

An Automatic Prompt Optimization Framework for Large Language Models