Pumpkin's starred repositories
SMDL-Attribution
[ICLR 2024 Oral] Less is More: Fewer Interpretable Region via Submodular Subset Selection
easy-to-hard
Easy-to-Hard Generalization: Scalable Alignment Beyond Human Supervision
rlhf-poisoning
Code for paper "Universal Jailbreak Backdoors from Poisoned Human Feedback"
EasyContext
Memory optimization and training recipes to extrapolate language models' context length to 1 million tokens, with minimal hardware.
Expert_Sparsity
[ACL 2024] Not All Experts are Equal: Efficient Expert Pruning and Skipping for Mixture-of-Experts Large Language Models
awesome-mixture-of-experts
A collection of AWESOME things about mixture-of-experts
Awesome-Model-Merging
:couple: A curated list of Model Merging methods.
one-person-businesses-methodology-v2.0
《一人企业方法论》第二版,也适合做其他副业(比如自媒体、电商、数字商品)的非技术人群。