Abstract: DALL·E mini is an attempt at reproducing OpenAI's DALL·E with a smaller architecture. DALL·E can generate new images from any text prompt. The model is 27 times smaller than the original version and was trained on a single TPU v3-8 for only 3 days.
It was made possible by simplifying the model architecture and memory requirements as well as leveraging open-source code and pre-trained models available.
Speaker Bio: Boris Dayma is a Machine Learning Engineer.
He works on making AI more accessible through articles and user-friendly projects. He is a contributor to Weights & Biases, Transformers, fastai, Pytorch Lightning and many other AI frameworks. He is the lead author of the DALL·E mini project, which also involves Suraj Patil, Pedro Cuenca, Khalid Saifullah, Tanishq Abraham, Phúc Lê Khắc, Luke Melas and Ritobrata Ghosh.
---
Please subscribe and like the video to help me keep motivated to make awesome videos like this one. :)
To buy my book, Approaching (Almost) Any Machine Learning problem, please visit: https://bit.ly/buyaaml
Follow me on:
Twitter: https://twitter.com/abhi1thakur
LinkedIn: https://www.linkedin.com/in/abhi1thakur/
Kaggle: https://kaggle.com/abhishek
Instagram: https://instagram.com/abhi1thakur
0 Comments