March ’23 Computer Vision Meetup

When

March 9, 2023 – 10AM PT

Where

Virtual / Zoom

Agenda

• Housekeeping & Lightning Talk – Voxel51
• Lighting Up Images in the Deep Learning Era – Soumik Rakshit, ML Engineer (Weights & Biases)
• Training and Fine Tuning Vision Transformers Efficiently with Colossal AI – Sumanth P, ML Engineer
• Closing Remarks

Lighting up Images in the Deep Learning Era

With the goal of recovering high-quality image content from its degraded version, image restoration enjoys numerous applications, such as in photography, security, medical imaging, and remote sensing. Recently, convolutional neural networks (CNNs) have achieved dramatic improvements over conventional approaches for image restoration tasks. We would discuss some of these state-of-the-art approaches to image restoration and would also discuss not only how we can improve the performance of such models, but also make them smaller and faster.

Soumik Rakshit is a Machine Learning Engineer working on end-to-end deep learning pipelines as part of the Growth team at Weights & Biases.

Training and Fine Tuning Vision Transformers Efficiently with Colossal AI

In the recent years we have seen the flourish of large AI Models but unfortunately, training and even fine-tuning those models are usually unaffordable, requiring huge computational resources. Existing deep learning frameworks like PyTorch and Tensorflow may not offer a satisfactory solution for very large AI models. In this talk we will go deep dive into Vision Transformers and see how to train and fine-tune the models more efficiently with Colossal AI that can perform complicated distributed training. The best part of this is, it is completely open source.

Sumanth P is a developer advocate, ML Engineer, ML content creator, and Founder of the Community (C4ML).

Don’t Forget

• Voxel51 will make a donation on behalf of the Meetup members to the charity that gets the most votes this month.
• Can’t make the date and time? No problem! Just make sure to register here so we can send you links to the playbacks.