WHY JAX? Why the Hell a 3rd ML framework in 2023?
HTML-код
- Опубликовано: 26 авг 2024
- 100x faster on GPU. Why do I need a 3rd Machine Learning Framework, next to TensorFlow2 and PyTorch? Automatically scalable Cloud infrastructure (TPU Pods v4) on Google Cloud and Nvidia H100 SuperPods (... AWS) with XLA compiler are fast! Really fast!
Based on my last video ( ) we look at a new powerful ML framework from the cloud infrastructure side. Primary factors are SPEED, more SPEED, SCALE, Memory and COSTS ($$) for ML training and inference at cloud supercomputer.
#ai
#datascience
#pytorch
#tensorflow2
#jax
#machinelearning
I might have gone through every JAX related video on RUclips, imho this one tops it thanks to the way you put it all together. I'd love to see more on advanced JAX from your perspective, cheers!
Thank you for your comment.
I wonder if its posible have a whole framework in Rust like polars library vs pandas