osiriszjq/impulse_init
Convolutional Initialization for Data-Efficient Vision Transformers
This project helps machine learning researchers and engineers improve the efficiency of training Vision Transformers (ViT) for image classification. By using a specialized initialization method, it allows these models to achieve strong performance even when less training data is available. You input image datasets like CIFAR-10 or Tiny ImageNet, and it outputs a more robustly initialized ViT model ready for training.
Use this if you are developing computer vision models and need to train Vision Transformers effectively with limited image data.
Not ideal if you are working with non-image data or if your existing Vision Transformer models already perform well with abundant data.
Stars
16
Forks
—
Language
Jupyter Notebook
License
MIT
Category
Last pushed
Dec 09, 2025
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/transformers/osiriszjq/impulse_init"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
lucidrains/x-transformers
A concise but complete full-attention transformer with a set of promising experimental features...
kanishkamisra/minicons
Utility for behavioral and representational analyses of Language Models
lucidrains/simple-hierarchical-transformer
Experiments around a simple idea for inducing multiple hierarchical predictive model within a GPT
lucidrains/dreamer4
Implementation of Danijar's latest iteration for his Dreamer line of work
Nicolepcx/Transformers-in-Action
This is the corresponding code for the book Transformers in Action