ajyl Goto Github PK

adversarialreprogramming

Code for my blog post on Adversarial Reprogramming

ajyl.github.io

calm-pytorch

Implementation of CALM from the paper "LLM Augmented LLMs: Expanding Capabilities through Composition", out of Google Deepmind

clip_text_span

official implementation of "Interpreting CLIP's Image Representation via Text-Based Decomposition"

concept_graphs

direct-preference-optimization

Reference implementation for DPO (Direct Preference Optimization)

dpo_toxic

A Mechanistic Understanding of Alignment Algorithms: A Case Study on DPO and Toxicity.

emotions-micromodels

Micromodels -- A framework for accurate, explainable, data efficient, and reusable NLP models.

epsilon-transformers

epsilon machines and transformers!

f-divergence-dpo

Direct preference optimization with f-divergences.

llm_categorical_hierarchical_representations

mech_int_othellogpt

olmo

Modeling, training, eval, and inference code for OLMo

open_clip

An open source implementation of CLIP.

othello_world

Emergent world representations: Exploring a sequence model trained on a synthetic task

pplm

Plug and Play Language Model implementation. Allows to steer topic and attributes of GPT-2 models.

prompt-to-prompt

pyreft

ReFT: Representation Finetuning for Language Models

q_probe

Q-Probe: A Lightweight Approach to Reward Maximization for Language Models

slurm-cheatsheet

steering-vectors

Steering vectors for transformer language models in Pytorch / Huggingface

syntax-guided-generation

transformer-debugger

transformerlens

vit-prisma

ViT Prisma is a mechanistic interpretability library for Vision Transformers (ViTs).

youtube-comment-downloader

Simple script for downloading Youtube comments without using the Youtube API