Alex Cheema

Results 418 comments of Alex Cheema

Here's mine. Example for Llama-3.1-8B. ``` (.venv) alex@Alexs-MBP exo % ls -R ~/.cache/huggingface/hub/models--mlx-community--Meta-Llama-3.1-8B-Instruct-4bit blobs cachedreqs refs snapshots /Users/alex/.cache/huggingface/hub/models--mlx-community--Meta-Llama-3.1-8B-Instruct-4bit/blobs: 02ee80b6196926a5ad790a004d9efd6ab1ba6542 421cda369d1e01e742b01d82e3a39c7cc82a8586 412806596a1283ba83602123e49ec97da18e7f16 6d37edeb7623a2fd95ffe7f719a6475dc9a2b1ea /Users/alex/.cache/huggingface/hub/models--mlx-community--Meta-Llama-3.1-8B-Instruct-4bit/cachedreqs: efc01dc1fd006f88344400c099cda5b3e8e524ef /Users/alex/.cache/huggingface/hub/models--mlx-community--Meta-Llama-3.1-8B-Instruct-4bit/cachedreqs/efc01dc1fd006f88344400c099cda5b3e8e524ef: fetch_file_list.json /Users/alex/.cache/huggingface/hub/models--mlx-community--Meta-Llama-3.1-8B-Instruct-4bit/refs: main /Users/alex/.cache/huggingface/hub/models--mlx-community--Meta-Llama-3.1-8B-Instruct-4bit/snapshots: efc01dc1fd006f88344400c099cda5b3e8e524ef...

What is your use case exactly? You can't connect to huggingface? You can try running with the flag `--download-quick-check` which will skip all huggingface calls and look for the files...

Request: Raspberry Pi. We could bundle it with Coral USB TPU (https://coral.ai/products/) which could be super cost effective home ai inference.

List of supported hardware:

2x4090 on exo vs bunch of 4090s in one pc ![IMG_0076](https://github.com/user-attachments/assets/e2b223fd-fe23-46e8-87cf-d1517f462303)

The people want to know how fast it is. ![IMG_7523](https://github.com/user-attachments/assets/e8f18dd6-1130-4ac1-8bf9-4532d032afa5)

![IMG_0093](https://github.com/user-attachments/assets/cfff1e8c-c872-4795-acff-f52c05ff4a8d)

> it is able to run on my 32gb mac :) can i take this up @AlexCheema ? wanted to explore sd, flux for the longest, this will be a...

> Want to participate too! Sure, if you and @varshith15 agree to work together I will award the full $500 to both of you.