mit-han-lab/offsite-tuning
Offsite-Tuning: Transfer Learning without Full Model
This helps organizations adapt powerful, pre-trained AI models to their specific data and tasks without sharing their sensitive information or needing massive computing resources. A model owner sends a small adapter and a compressed emulator to a data owner, who then uses their specific data to improve the model's performance. The fine-tuned adapter is returned to the model owner, resulting in an AI model that works better for the data owner's unique needs while keeping all data private. This is for AI practitioners, data scientists, or business units who want to customize large foundation models.
387 stars. No commits in the last 6 months.
Use this if you need to customize a large, proprietary AI model with your specific data, but you cannot or do not want to share your data with the model owner, and you have limited computational resources for fine-tuning.
Not ideal if you have full access to the AI model's internal workings and weights, or if you prefer to perform traditional fine-tuning directly on your own infrastructure without privacy constraints.
Stars
387
Forks
40
Language
Python
License
MIT
Category
Last pushed
Nov 27, 2023
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/ml-frameworks/mit-han-lab/offsite-tuning"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
Jittor/jittor
Jittor is a high-performance deep learning framework based on JIT compiling and meta-operators.
zhanghang1989/ResNeSt
ResNeSt: Split-Attention Networks
berniwal/swin-transformer-pytorch
Implementation of the Swin Transformer in PyTorch.
NVlabs/FasterViT
[ICLR 2024] Official PyTorch implementation of FasterViT: Fast Vision Transformers with...
ViTAE-Transformer/ViTPose
The official repo for [NeurIPS'22] "ViTPose: Simple Vision Transformer Baselines for Human Pose...