alexlioralexli/noncontrastive-ssl
Analyzing partial dimensional collapse in non-contrastive self-supervised learning. "Understanding Collapse in Non-Contrastive Siamese Representation Learning." In ECCV, 2022.
This repository helps AI researchers and machine learning engineers analyze how self-supervised learning models, specifically those using a 'siamese' architecture, might develop redundant or 'collapsed' internal representations. It takes an ImageNet dataset and a pre-trained SimSiam model as input to train and evaluate the model's representations. The output provides metrics like k-NN accuracy and collapse scores, along with plotting capabilities to visualize the results.
No commits in the last 6 months.
Use this if you are a machine learning researcher or engineer studying the behavior of self-supervised learning models and want to understand or mitigate representation collapse in non-contrastive methods.
Not ideal if you are looking for a pre-built model for immediate application in tasks like image classification, as this tool focuses on the research and analysis of model training dynamics rather than deployment.
Stars
16
Forks
1
Language
Jupyter Notebook
License
MIT
Category
Last pushed
Nov 12, 2023
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/ml-frameworks/alexlioralexli/noncontrastive-ssl"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
AdaptiveMotorControlLab/CEBRA
Learnable latent embeddings for joint behavioral and neural analysis - Official implementation of CEBRA
theolepage/sslsv
Toolkit for training and evaluating Self-Supervised Learning (SSL) frameworks for Speaker...
PaddlePaddle/PASSL
PASSL包含 SimCLR，MoCo v1/v2，BYOL，CLIP，PixPro，simsiam, SwAV, BEiT，MAE 等图像自监督算法以及 Vision...
YGZWQZD/LAMDA-SSL
30 Semi-Supervised Learning Algorithms
ModSSC/ModSSC
ModSSC: A Modular Framework for Semi Supervised Classification