[CVPR'24] Validation-free few-shot adaptation of CLIP, using a well-initialized Linear Probe (ZSLP) and class-adaptive constraints (CLAP).

Code for the paper "Generalizing Importance Weighting to A Universal Solver for Distribution Shift Problems".

Code for Continuously Changing Corruptions (CCC) benchmark + evaluation

A framework for merging models solving different tasks with different initializations into one multi-task model without any additional training

🔥 🔥 [ECCV 2024 Oral ] Official code for "Weighted Ensemble Models Are Strong Continual Learners"

Source code for the NeurIPS 2023 paper: "CSOT: Curriculum and Structure-Aware Optimal Transport for Learning with Noisy Labels"

Personal implementation of ASIF by Antonio Norelli

A prompt injection game to collect data for robust ML research

Paper list for the survey "Combating Misinformation in the Age of LLMs: Opportunities and Challenges" and the initiative "LLMs Meet Misinformation", accepted by AI Magazine 2024

LLM Unlearning

Test-time Prompt Tuning (TPT) for zero-shot generalization in vision-language models (NeurIPS 2022))

The lastest paper about detection of LLM-generated text and code

CIFAR-10-Warehouse: Towards Broad and More Realistic Testbeds in Model Generalization Analysis

Official implementation of the ICCV2023 paper: Enhancing Generalization of Universal Adversarial Perturbation through Gradient Aggregation

A curation of awesome tools, documents and projects about LLM Security.

APBench: A Unified Availability Poisoning Attack and Defenses Benchmark (TMLR 08/2024)

This may be the simplest implement of DDPM. You can directly run Main.py to train the UNet on CIFAR-10 dataset and see the amazing process of denoising.

Starting kit for the NeurIPS 2023 unlearning challenge

A list of papers that studies Novel Class Discovery

Supplementary code for the paper "UnSplit: Data-Oblivious Model Inversion, Model Stealing, and Label Inference Attacks Against Split Learning".

BMAD hold a Creative Commons Attribution-NonCommercial-ShareAlike (CC BY-NC-SA) license

[ICML23] On Pitfalls of Test-Time Adaptation

Code and results accompanying our paper titled RLSbench: Domain Adaptation under Relaxed Label Shift

[ACL2023] We introduce LLM-Blender, an innovative ensembling framework to attain consistently superior performance by leveraging the diverse strengths of multiple open-source LLMs. LLM-Blender cut …

The is the official implementation of ICML 2023 paper "Revisiting Weighted Aggregation in Federated Learning with Neural Networks".