主要记录大语言大模型（LLMs） 算法（应用）工程师相关的知识及面试题

A course on aligning smol models.

PyHessian is a Pytorch library for second-order based analysis and training of Neural Networks

BackdoorLLM: A Comprehensive Benchmark for Backdoor Attacks on Large Language Models

Official Repo of ICLR 24 BadChain: Backdoor Chain-of-Thought Prompting for Large Language Models

A curated list of awesome prompt/adapter learning methods for vision-language models like CLIP.

Collection of leaked system prompts

RAG Hallucination Detecting By LRP.

Official repository of our work "Finding Lottery Tickets in Vision Models via Data-driven Spectral Foresight Pruning" accepted at CVPR 2024

[ACL 2024] ERA-CoT: Improving Chain-of-Thought through Entity Relationship Analysis.

Pruning By Explaining Revisited: Optimizing Attribution Methods to Prune CNNs and Transformers, Paper accepted at eXCV workshop of ECCV 2024

GB/T 7714-2015 BibTeX Style

Advanced AI Explainability for computer vision. Support for CNNs, Vision Transformers, Classification, Object detection, Segmentation, Image similarity and more.

Official implementation for "Automatic Chain of Thought Prompting in Large Language Models" (stay tuned & more will be updated)

Code for the 2024 arXiv publication "Fine-Tuning with Divergent Chains of Thought Boosts Reasoning Through Self-Correction in Language Models"

the instructions and demonstrations for building a formal logical reasoning capable GLM

g1: Using Llama-3.1 70b on Groq to create o1-like reasoning chains

multi1: create o1-like reasoning chains with multiple AI providers (and locally). Supports LiteLLM as backend too for 100+ providers at once.

Notebooks and Code about Generative Ai, LLMs, MLOPS, NLP , CV and Graph databases

This project is a **proof of concept** that aims to replicate the reasoning capabilities of OpenAI's newly released O1 model.

awesome papers in LLM interpretability

A lightweight library for generating synthetic instruction tuning datasets for your data without GPT.

[ACL 2024] An Easy-to-use Knowledge Editing Framework for LLMs.

Mass-editing thousands of facts into a transformer memory (ICLR 2023)

Code accompanying the ICML'24 paper "Feature Contamination: Neural Networks Learn Uncorrelated Features and Fail to Generalize"

The author's officially unofficial PyTorch BigGAN implementation.

[ICLR2024] "Backdoor Federated Learning by Poisoning Backdoor-Critical Layers"