mlx-examples
mlx-examples copied to clipboard
Examples in the MLX framework
Building on the amazing work by @mzbac and @nkasmanoff in https://github.com/ml-explore/mlx-examples/pull/461, I'd really love an example of how LLaVA 1.6 (aka llava next) can be fine-tuned with a LoRA. I...
see #583
Firstly I finished the LoRA fine-tuning with some training datasets, the result is good. Later I hope to add more training datasets, I have tried the following two methods, both...
Since whisper encoder significantly slower, than wav2vec2, it would be nice addition to MLX
## Request: Add grok-1 to mlx-examples and help with M1 Pro compatibility Hello, I would like to request the addition of the `grok-1` repository () to the `mlx-examples` repository. I...
this is about clip/tokenizer.py The example: `tokenizer(["a photo of a cat", "a photo of a dog"])` only works 'cos they have exact same # of tokens. this one will fail:...
I was testing the new quantize by @angeloskath with some Italian prompts that were failing with previous version and now are PERFECT! But while doing this I have seen extreme...
@awni I have PaliGemma working on MLX. In most cases works great. But there 4 issues, I don't see in the transformers implementation: - 4bit quantisation just doesn't work while...
Implements all MoEs with the new `mx.block_sparse_mm` and `mx.block_sparse_qmm` (the latter is at https://github.com/ml-explore/mlx/pull/1124 ). One notable change is in the quantization predicate which accepts modules that define `to_quantized()` and...
My os system is 13.0, and I have tried mlx v0.13 to mlx v0.7, it all reports: libmlx.dylib (built for macOS 13.5 which is newer than running OS) I wonder...