This is a 2-bit quantization of @pankajmathur orca_mini_v3_70b using quip# (https://cornell-relaxml.github.io/quip-sharp/) with hessian context lenght 4k.
Inference with the model is a bit slow, but this should be the best 70b model you can use without offloading on a 3090.
Prompt Format:
### System:
You are an AI assistant that follows instruction extremely well. Help as much as you can.
### User:
Tell me about Orcas.
### Assistant:
I have included the quip library I have used in this repo. I am able to use this model in the widely known textgen-webui. For installation I suggest to follow these steps:
- Download the quip folder from this repo and place it inside the repositories folder of the textgen-webui folder.
- install the requirements of quip#
- compile and install the quiptools cuda lib:
pip install fast-hadamard-transform glog==0.3.1 primefac==2.0.12
cd repositories/quip-sharp/quiptools
python setup.py install --force
- reinstall the requirements of textgen-webui
- load the model with the quip# integration of textgen-webui
You can use the library of this repo also for scripts. Within the quip# folder, after installing the library, use this command:
python interactive_gen.py --hf_path path_to_the_2bitmodel --max_length 500 text