---
license: llama3.1
train: false
inference: false
pipeline_tag: text-generation
---
This is an HQQ all 4-bit (group-size=64) quantized Llama3.1-8B-Instruct model.
## Model Size
| Models | fp16| HQQ 4-bit/gs-64 | AWQ 4-bit |
|:-------------------:|:--------:|:----------------:|:----------------:|
| Bitrate (Linear layers) | 16 | 4.5 | 4.25 |
| VRAM | 15.7 (GB) | 6.1 (GB) | 6.3 (GB) |
## Model Decoding Speed
| Models | fp16| HQQ 4-bit/gs-64| AWQ 4-bit |
|:-------------------:|:--------:|:----------------:|:----------------:|
| Decoding* - short seq (tokens/sec)| 53 | 125 | 67 |
| Decoding* - long seq (tokens/sec)| 50 | 97 | 65 |
*: RTX 3090
## Performance
| Models | fp16 | HQQ 4-bit/gs-64 | AWQ 4-bit |
|:-------------------:|:--------:|:----------------:|:----------------:|
| ARC (25-shot) | 60.49 | 60.32 | 57.85 |
| HellaSwag (10-shot)| 80.16 | 79.21 | 79.28 |
| MMLU (5-shot) | 68.98 | | 67.14 |
| TruthfulQA-MC2 | 54.03 | 53.89 | 51.87 |
| Winogrande (5-shot)| 77.98 | 76.24 | 76.4 |
| GSM8K (5-shot) | 75.44 | | 73.47 |
| Average | 69.51 | | 67.67 |
## Usage
First, install the dependecies:
```
pip install git+https://github.com/mobiusml/hqq.git #master branch fix
pip install bitblas
```
Also, make sure you use at least torch `2.4.0` or the nightly build.
Then you can use the sample code below:
``` Python
import torch
from transformers import AutoTokenizer
from hqq.models.hf.base import AutoHQQHFModel
from hqq.utils.patching import *
from hqq.core.quantize import *
from hqq.utils.generation_hf import HFGenerator
#Load the model
###################################################
model_id = 'mobiuslabsgmbh/Llama-3.1-8b-instruct_4bitgs64_hqq'
compute_dtype = torch.float16 #bfloat16 for torchao, float16 for bitblas
cache_dir = '.'
model = AutoHQQHFModel.from_quantized(model_id, cache_dir=cache_dir, compute_dtype=compute_dtype)
tokenizer = AutoTokenizer.from_pretrained(model_id, cache_dir=cache_dir)
quant_config = BaseQuantizeConfig(nbits=4, group_size=64, quant_scale=False, quant_zero=False, axis=1)
patch_linearlayers(model, patch_add_quant_config, quant_config)
#Use optimized inference kernels
###################################################
HQQLinear.set_backend(HQQBackend.PYTORCH)
#prepare_for_inference(model) #default backend
#prepare_for_inference(model, backend="torchao_int4")
prepare_for_inference(model, backend="bitblas") #takes a while to init...
#Generate
###################################################
gen = HFGenerator(model, tokenizer, max_new_tokens=1000, do_sample=True, compile="partial").warmup() #Warm-up takes a while
gen.generate("Write an essay about large language models", print_tokens=True)
gen.generate("Tell me a funny joke!", print_tokens=True)
gen.generate("How to make a yummy chocolate cake?", print_tokens=True)
```