metadata
base_model:
- meta-llama/Llama-3.2-3B-Instruct
base_model_relation: quantized
This is a version of the Llama-3.2-3B-Instruct model quantized to 4-bit via Half-Quadratic Quantization (HQQ): https://mobiusml.github.io/hqq_blog/
To run the model, install the HQQ library from https://github.com/mobiusml/hqq and use it as follows:
model_id = 'brunopio/Llama-3.2-3B-Instruct-nbits4-GS64-Axis1-HQQ'
from hqq.engine.hf import HQQModelForCausalLM, AutoTokenizer
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = HQQModelForCausalLM.from_quantized(model_id)
Limitations:
-Only supports single GPU runtime.
-Not compatible with HuggingFace's PEFT.