Model Details
This model is an int4 auto-round model with group_size 128 of Qwen/Qwen2-7B generated by intel/auto-round. If you need AutoGPTQ format, please load the model with revision 07a117c
How To Use
INT4 Inference
from auto_round import AutoRoundConfig
from transformers import AutoModelForCausalLM,AutoTokenizer
quantized_model_dir = "Intel/Qwen2-7B-int4-inc"
tokenizer = AutoTokenizer.from_pretrained(quantized_model_dir)
model = AutoModelForCausalLM.from_pretrained(quantized_model_dir,
device_map="auto"
)
text = "下面我来介绍一下阿里巴巴公司,"
text = "9.8和9.11哪个数字大?答案是"
text = "Once upon a time,"
text = "There is a girl who likes adventure,"
inputs = tokenizer(text, return_tensors="pt").to(model.device)
print(tokenizer.decode(model.generate(**inputs, max_new_tokens=50, do_sample=False)[0]))
Intel Gaudi-2 INT4 Inference
docker image with Gaudi Software Stack is recommended. More details can be found in Gaudi Guide.
import habana_frameworks.torch.core as htcore
import habana_frameworks.torch.hpu as hthpu
from auto_round import AutoRoundConfig
from transformers import AutoModelForCausalLM,AutoTokenizer
quantized_model_dir = "Intel/Qwen2-7B-int4-inc"
tokenizer = AutoTokenizer.from_pretrained(quantized_model_dir)
model = AutoModelForCausalLM.from_pretrained(quantized_model_dir).to('hpu').to(bfloat16)
text = "下面我来介绍一下阿里巴巴公司,"
inputs = tokenizer(text, return_tensors="pt").to(model.device)
print(tokenizer.decode(model.generate(**inputs, max_new_tokens=50, do_sample=False)[0]))
Evaluate the model
pip3 install lm-eval==0.4.4,auto-round
auto-round --model "Intel/Qwen2-7B-int4-inc" --eval --eval_bs 16 --tasks lambada_openai,hellaswag,piqa,winogrande,truthfulqa_mc1,openbookqa,boolq,arc_easy,arc_challenge,mmlu,gsm8k,cmmlu,ceval-valid
Metric |
BF16 |
INT4 |
Avg |
0.6659 |
0.6604 |
mmlu |
0.6697 |
0.6646 |
cmmlu |
0.8254 |
0.8118 |
ceval-valid |
0.8339 |
0.8053 |
lambada_openai |
0.7182 |
0.7136 |
hellaswag |
0.5823 |
0.5752 |
winogrande |
0.7222 |
0.7277 |
piqa |
0.7911 |
0.7933 |
truthfulqa_mc1 |
0.3647 |
0.3476 |
openbookqa |
0.3520 |
0.3440 |
boolq |
0.8183 |
0.8223 |
arc_easy |
0.7660 |
0.7635 |
arc_challenge |
0.4505 |
0.4633 |
gsm8k 5 shots(strict match) |
0.7619 |
0.7528 |
Generate the model
Here is the sample command to reproduce the model. We observed a larger accuracy drop in Chinese tasks and recommend using a high-quality Chinese dataset for calibration. However, we did not achieve better accuracy with some public datasets.
auto-round
--model_name Qwen/Qwen2-7B \
--device 0 \
--group_size 128 \
--nsamples 512 \
--bits 4 \
--iter 1000 \
--disable_eval \
--model_dtype "float16" \
--format 'auto_round' \
--output_dir "./tmp_autoround"
Ethical Considerations and Limitations
The model can produce factually incorrect output, and should not be relied on to produce factually accurate information. Because of the limitations of the pretrained model and the finetuning datasets, it is possible that this model could generate lewd, biased or otherwise offensive outputs.
Therefore, before deploying any applications of the model, developers should perform safety testing.
Caveats and Recommendations
Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model.
Here are a couple of useful links to learn more about Intel's AI software:
- Intel Neural Compressor link
- Intel Extension for Transformers link
Disclaimer
The license on this model does not constitute legal advice. We are not responsible for the actions of third parties who use this model. Please consult an attorney before using this model for commercial purposes.
Cite
@article{cheng2023optimize, title={Optimize weight rounding via signed gradient descent for the quantization of llms}, author={Cheng, Wenhua and Zhang, Weiwei and Shen, Haihao and Cai, Yiyang and He, Xin and Lv, Kaokao and Liu, Yi}, journal={arXiv preprint arXiv:2309.05516}, year={2023} }
arxiv github