mlx-swift-examples

mlx-swift-examples copied to clipboard

Package.resolved

5

Has anyone else got this error: `Package.resolved file is corrupted or malformed; fix or delete the file to continue` Had to remove this file to get the project to build...

ajayjapan

When to stop in the LLMEval?

7

In the LLMEval project, the generation stops after reaching a limit on tokens. Is there a way to configure stopping when it finds a special token? I tried to look...

MatthewWaller

OpenELM example?

1

Thanks for the awesome work! I would request an OpenELM example. Or if we can tweak one of the existing examples for OpenELM, that would be great. Any guidance?

MatthewWaller

LLMEval not loading Qwen1.5 -0.5B model in to memory

17

When trying to load Qwen1.5, the model downloads fully but doesn't appear to load in to memory on MacOS or iOS. After typing a prompt, the error output is "Failed:...

mobile-appz

LLMEval crashing due to force unwrapping of optional in token decode for Gemma

6

[st.txt](https://github.com/ml-explore/mlx-swift-examples/files/14921908/st.txt) is the crash log. Model : Gemma 2B Quantized Prompt : "Write code to boot a raspberry pico"

ViRo3

Phi-2 on iOS, bad responses

2

I am running the phi-2 on iOS using the code from LLMEval. I have ported over an implementation of the CodeGen Tokenizer into swift as a standalone file: ```swift import...

taylorgoolsby

LLMEval: Model Configuration Registry Remains Empty, Preventing Dynamic Model Selection

1

I am currently working on implementing a model picker so selecting a new model does not require recompiling the app. I was trying to use https://github.com/ml-explore/mlx-swift-examples/blob/0af2efe8a04d11a7c01616e560a3bd67ac38f246/Libraries/LLM/Models.swift#L39 However, I am seeing...

1-ashraful-islam

Fine-tuned Qwen2 model inference error

6

I fine-tuned the model based on the [Qwen/Qwen1.5-0.5B-Chat](https://huggingface.co/Qwen/Qwen1.5-0.5B-Chat) model and then fused the models. The final output when reasoning with the mlx-lm model is as expected (a specific URL link)...

madroidmaq

llm-tool: crash during async generation

10

Steps to reproduce: 1. Move LLMTool.swift `@main` annotation from `SyncGenerator` to `AsyncGenerator`. 2. Run llm-tool target. Result: `Thread 4: EXC_BAD_ACCESS (code=2, address=0x16ff1bfe0)` Call stack: ``` #0 0x00000001050d4ac0 in mlx::core::eval(std::__1::vector const&)::$_3::operator()(mlx::core::array...

zach-brockway

llm-eval: not responding to 'what is your name?' or 'what is the difference between star wars and star trek?'

2

On my Mac, I see the error ```FAULT: determined it was necessary to configure to support remote view vibrancy CLIENT ERROR: TUINSRemoteViewController does not override -viewServiceDidTerminateWithError: and thus cannot react...

CharlieTLe