Tarun Kumar Vangani

Hi! Thanks for the awesome work. Was wondering if I can use it for the above question? Thanks

How to freeze layers?

1

Thanks for the awesome project! I really enjoyed your talk too. Flux.jl has option for trainable as well as deleting things from params Certain tasks may have some layers frozen...

Is it possible to use term instead of vterm

2

I don't really like vterm as it hinders some use-cases of mine. 1. Unable to complete snippets as buffer is read only 2. Can't use Alt + Enter when editing...

How to set custom ratio?

3

Hi! I understand the default ratio is according to the balance of accounts. I would like to override this and set ratio myself. If this feature is not available can...

Support for AMD/ROCm with Pallas

15

I have JAX and Triton installed. On trying the code below I get the following error: I assume this is due to XLA treating my GPU to be a NVIDIA...

Reshape of a view breaks batched_mul on GPU (scalar indexing happens)

3

Useful Info: NNlib v0.8.16 I actually have use of this to implement very efficient multi head attention. I am sharing the minimal example to replicate below: using Flux # to...

This is a feature requests. I would like to be able to make something like below:  Thank you!

Tello EDU and Speed Support

2

Hi! Thanks for this awesome lib. I have a couple of questions: 1. Does this library work with the Edu version? 2. Does this library have extra supported features for...

Make requests async

4

Hi! Thanks for the project. I love using it. One little issue I have is that currently whenever I make a response in the Chat my whole emacs hangs. I...

Taking Gradients

1

Hi! This package looks awesome and exactly what I was looking for. Will there be support for taking gradients in the future? Also, I would like to contribute/lend-a-hand in anyway...