Open tombenninger opened 3 weeks ago
I just realized this may be too early. The new llama.cpp to support these models needs to get into the llama-cpp-python library and then that needs to be incorporated into the AI Lab extension in order to run these models, but once that is done, I'd like them added to the catalog. =)
Oh it looks like llama-cpp-python just got updated, and the newest release (0.2.77) should work:
Is your enhancement related to a problem? Please describe
IBM's granite "code-instruct" models are now available in 3b, 8b, 20b, and 34b GGUF formats at https://huggingface.co/models?sort=trending&search=ibm-granite+granite+code-instruct+GGUF.
These models would probably be more useful than the granite "lab" 7b model from before the major market release.
Describe the solution you'd like
Include the 4 code instruct models in the model catalog.
Describe alternatives you've considered
The models can be downloaded and manually imported, but this just adds extra steps for the end user.
Additional context
No response