lmstudio-ai
[Feature Request] Support for bitnet.cpp and BitNet models
bitnet.cpp is the official inference framework for 1-bit LLMs (e.g., BitNet b1.58). It offers a suite of optimized kernels, that support fast and lossless inference of 1.58-bit models on CPU (with NPU and GPU support coming next).
It would be awesome for LMS to support bitnet.cpp and therefore BitNet models, for on-device evaluation.
Are there any updates on this features, I saw that in llama.cpp, BitNet is supported ggml-org/llama.cpp#7931. I tried to load microsoft/bitnet-b1.58-2B-4T-gguf, but I didn't succeed.
With https://arxiv.org/abs/2504.12285 released I am curious to try that out on my home lab. LM Studio would be great for this
I am getting:
🥲 Failed to load the model
Failed to load model
error loading model: llama_model_loader: failed to load model from .lmstudio/models/microsoft/bitnet-b1.58-2B-4T-gguf/ggml-model-i2_s.gguf
I'm having same issues, any update yet?! trying to run microsoft/bitnet-b1.58-2B-4T on LM Studio v0.3.14 on Win10pro
Eu tbm estou com esse problema no LM Studio. I also have this problem in LM Studio.
Same issue for me. I wish that there were actually error messages with codes so that some sort of troubleshooting could be done.
🥲 Failed to load the model
Failed to load model
error loading model: llama_model_loader: failed to load model from .lmstudio\models\microsoft\bitnet-b1.58-2B-4T-gguf\ggml-model-i2_s.gguf
+1, somone explain to me how this would get added. as a runtime? or they're just waiting for transformer library to merge some code to support bitnet models.
bitnet.cpp is very much needed... just like CPU/CUDA llama.cpp.
B1.54 models not loading/working at all. Pls add the same to LMSTUDIO. Thanks in advance.
Friendly reminder: use the :thumbsup: reaction on the original post instead of posting bump comments, which can clutter up the discussion
After a century this issue will be open, developers don't even look at this issue. @ryan-the-crayon @rugvedS07 @yagil @will-lms @sergeichestakov @neilmehta24 @mattjcly @jlguenego @mayfer @YorkieDev @CinnamonRolls1, please react and answer if this is planned or will not be implemented.
I agree, we need confirmation if this is planned, being worked on, or not planned at all.
