File size: 1,465 Bytes
2bcaf27 4968216 2bcaf27 f0b8e84 2bcaf27 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 |
---
library_name: transformers
tags:
- phi-3
- phi-3-mini
- phi-3-mini-4k-instruct
- conversational
- text-generation-inference
pipeline_tag: text-generation
language:
- en
---
Official quantization of [Mistral-7B-v0.1](https://huggingface.co/mistralai/Mistral-7B-v0.1) using [PV-Tuning](https://arxiv.org/abs/2405.14852) on top of [AQLM](https://arxiv.org/abs/2401.06118).
For this quantization, we used 1 codebook of 16 bits for groups of 8 weights.
Results (0-shot `acc`):
Results:
| Model | Quantization | ArcC| ArcE| Hellaswag | PiQA | Winogrande | Model size, Gb |
|------|------|-------|------|------|------|------|------|
| microsoft/Phi-3-mini-4k-instruct| None | 0.5529 | 0.8325 | 0.6055 | 0.8020 | 0.7364 | 7.6 |
| | 1x16 | 0.5051 | 0.7950 | 0.5532 | 0.7949 | 73.01 | 1.4 |
You can also find Phi-3-medium models compressed with AQLM+PV: [2-bit](https://huggingface.co/ISTA-DASLab/Phi-3-medium-4k-instruct-AQLM-PV-2Bit-1x16-hf) and [1-bit](https://huggingface.co/ISTA-DASLab/Phi-3-medium-4k-instruct-AQLM-PV-1Bit-1x16-hf)
The 1x16g16 (1-bit) models are on the way, as soon as we update the inference lib with their respective kernels.
To learn more about the inference, as well as the information on how to quantize models yourself, please refer to the [official GitHub repo](https://github.com/Vahe1994/AQLM).
The original code for PV-Tuning can be found in the [AQLM@pv-tuning](https://github.com/Vahe1994/AQLM/tree/pv-tuning) branch.
|