An official quantization of meta-llama/Llama-2-7b using PV-Tuning on top of AQLM. For this quantization, we used 1 codebook of 16 bits for groups of 16 weights, totalling about 1.58 bits per weight.

The 1x16g16 models require aqlm inference library v1.1.6 or newer:

pip install aqlm[gpu,cpu]>=1.1.6

Model	AQLM scheme	WikiText 2 PPL	Model size, Gb	Hub link
Llama-2-7b	1x16	5.68	2.4	Link
Llama-2-7b	2x8	5.90	2.2	Link
Llama-2-7b (this)	1x16g16	9.21	1.7	Link
Llama-2-13b	1x16	5.05	4.1	Link
Llama-2-70b	1x16	3.78	18.8	Link

To learn more about the inference, as well as the information on how to quantize models yourself, please refer to the official GitHub repo. The original code for PV-Tuning can be found in the AQLM@pv-tuning branch.

ISTA-DASLab/Llama-2-7b-AQLM-PV-1Bit-1x16-hf

ISTA-DASLab/
Llama-2-7b-AQLM-PV-1Bit-1x16-hf