Results 6 issues of Abhinav Goel

Opening a draft pull request while we debug some XLA pattern matcher issues.

@rwitten this is a draft. This type of change would be specific to a few transformer models (e.g., Gemma, LLama, GPT, etc.). It wouldn't work with MoE, or some new...

I am new to programming, and I don't completely understand how you converted float32 to fixed. I am trying to implement a Neural Net with Keras, and Im not sure...

@zhangqiaorjc please review this change to support per core batch size < 1 with the synthetic dataset.

We notice a buggy behavior with bitcasts and dynamic update slices. When we turn on activation checkpointing (e.g., saving outputs of projection layers using the `SAVE_OUT_PROJ` flag in PAXML) we...