Failed to load the model (The requested number of bits 3 is not supported.)

[certik]

I downloaded 30GB of mlx-community/Llama-3.3-70B-Instruct-3bit only to find out:

🥲 Failed to load the model

Failed to load model

Error when loading model: ValueError: [quantize] The requested number of bits 3 is not supported. The supported bits are 2, 4 and 8.

LMStudio should know ahead of time that it does not support 3 bit and prevent me to download it, since it will not work.

[yagil]

Can you please try with the latest beta? https://lmstudio.ai/beta-releases

[ehartford]

Same here

🥲 Failed to load the model

Failed to load model

Error when loading model: ValueError: [quantize] The requested number of bits 3 is not supported. The supported bits are 2, 4 and 8.

[yagil]

@ehartford should work in the latest beta (build 9). Is it not working there? Get from https://lmstudio.ai/beta-releases

[certik]

I can confirm that this is fixed in the latest beta. Thanks @yagil !