The following models have been tested and are known to work with llm
.
We are collecting models in the rustformers organization, but this work is ongoing.
- https://huggingface.co/lxe/Cerebras-GPT-2.7B-Alpaca-SP-ggml: note that this is
f16
-only and we recommend you quantize it usingllm
for best performance.
At the time of writing, there are no publicly-released models with the GPT-J architecture and the currently-supported GGML quantization version.
You will need to obtain a F16 model and quantize it using llm
.
We have chosen not to include any models here until we have a better understanding of the licensing situation.