ml-explore / ml-explore/mlx-examples

Running larger MobileCLIP and AIMv2 on MLX

Open
#1,227 0 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
Python
Stars
9k
Forks
1.2k
PR merge metrics
No merged PRs in 30d

Description

It would be great to have an addition to the examples on how to run Apple's MobileCLIP and AIMv2 models on MLX.
Currently we only have the two smaller MobileCLIP versions in MLX format available, and there are no examples on how to encode text using the Zero-shot Adapted AIMv2. I'm curious to benchmark similar tasks against PyTorch with MPS or CUDA backends.

Maybe there is some conversion script I could run myself for MobileCLIP-B-LT?

AIMv2 already runs in MLX, just missing some MLX examples on how to encode text and run it in a different way than the other variants.

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start by reviewing the existing MLX MobileCLIP examples and the current AIMv2 MLX support. Compare the Zero-shot Adapted AIMv2 usage described in the linked ml-aim documentation with the available examples, and investigate whether MobileCLIP-B-LT has a conversion path. Done means documented examples for the requested models and text encoding workflow, with benchmarking against PyTorch MPS or CUDA where practical.

Written by the indexing model from the issue text.

Assessment

Tech stack
python, pytorch
Domain
machine-learning
Issue type
Feature
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Mostly clear
Newbie friendliness
35/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.