--- license: apache-2.0 datasets: - NeelNanda/pile-10k --- ## Model Details This model is an int4 model with group_size 128 of [Qwen/Qwen2-57B-A14B-Instruct](https://huggingface.co/Qwen/Qwen2-57B-A14B-Instruct) generated by [intel/auto-round](https://github.com/intel/auto-round), auto-round is needed to run this model ## How To Use ### INT4 CPU/CUDA Inference ```python ##git clone https://github.com/intel/auto-round.git ##cd auto-round && pip install -vvv --no-build-isolation -e . from auto_round import AutoHfQuantizer ##must import import torch from transformers import AutoModelForCausalLM,AutoTokenizer quantized_model_dir = "Intel/Qwen2-57B-A14B-Instruct-int4-inc" tokenizer = AutoTokenizer.from_pretrained(quantized_model_dir) model = AutoModelForCausalLM.from_pretrained( quantized_model_dir, torch_dtype=torch.float16, device_map="auto", ) prompt = "There is a girl who likes adventure," messages = [ {"role": "system", "content": "You are a helpful assistant."}, {"role": "user", "content": prompt} ] tokenizer = AutoTokenizer.from_pretrained(quantized_model_dir) text = tokenizer.apply_chat_template( messages, tokenize=False, add_generation_prompt=True ) model_inputs = tokenizer([text], return_tensors="pt").to(model.device) generated_ids = model.generate( model_inputs.input_ids, max_new_tokens=50, ##change this to align with the official usage do_sample=False ##change this to align with the official usage ) generated_ids = [ output_ids[len(input_ids):] for input_ids, output_ids in zip(model_inputs.input_ids, generated_ids) ] response = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0] print(response) ##prompt = "请介绍一下阿里巴巴公司" ##阿里巴巴集团是一家中国跨国科技公司，成立于1999年，总部位于杭州。阿里巴巴的业务涵盖了电子商务、零售、金融、物流、云计算等多个领域，是全球最大的电子商务公司之一。\n 阿里巴巴旗下拥有淘宝网、天猫、 ##prompt = "9.8大还是9.11大" ##9.8和9.11都是小数，但是9.8比9.11大。在数学中，小数的大小取决于它们的数值，数值越大则越“大”。在这个情况下，9.8的 ##prompt = "Once upon a time," ##there was a kingdom far, far away. In this kingdom, there lived a beautiful princess who had hair as golden as the sun and eyes as blue as the sea. The princess was kind and gentle, and everyone in the kingdom loved her dearly. ##prompt = "There is a girl who likes adventure," ##That's great to hear! Adventure can be a wonderful way to explore new places, learn new things, and challenge yourself in exciting ways. If you're looking for ideas on how to embark on an adventure, here are a few suggestions: 1. ``` ### Evaluate the model pip3 install lm-eval==0.4.2 ```bash git clone https://github.com/intel/auto-round cd auto-round/examples/language-modeling python3 eval_042/evluation.py --model_name "Intel/Qwen2-57B-A14B-Instruct-int4-inc" --eval_bs 16 --tasks lambada_openai,hellaswag,piqa,winogrande,truthfulqa_mc1,openbookqa,boolq,arc_easy,arc_challenge,mmlu,gsm8k,cmmlu,ceval-valid --trust_remote_code ``` | Metric | BF16 | INT4-AutoRound | [official GPTQ](https://huggingface.co/Qwen/Qwen2-57B-A14B-Instruct-GPTQ-Int4) | | :---------------------- | :------ | :-------------- | :------------------------------------------------------------ | | Avg | 0.7040 | 0.7043 | 0.6990 | | mmlu | 0.7438 | 0.7408 | 0.7409 | | cmmlu | 0.8505 | 0.8448 | 0.8475 | | ceval-valid | 0.8767 | 0.8611 | 0.8507 | | gsm8k 5 shots (strict) | 0.7627 | 0.7657 | 0.7597 | | lambada_openai | 0.7452 | 0.7444 | 0.7524 | | hellaswag | 0.6517 | 0.6475 | 0.6471 | | winogrande | 0.7245 | 0.7285 | 0.7198 | | piqa | 0.8058 | 0.8058 | 0.8041 | | truthfulqa_mc1 | 0.4345 | 0.4321 | 0.4272 | | openbookqa | 0.3400 | 0.3560 | 0.3300 | | boolq | 0.8835 | 0.8844 | 0.8810 | | arc_easy | 0.8035 | 0.8051 | 0.8001 | | arc_challenge | 0.5299 | 0.5392 | 0.5265 | ## Reproduce Here is the sample command to reproduce the model. ```bash git clone https://github.com/intel/auto-round cd auto-round/examples/language-modeling pip install -r requirements.txt python3 main.py \ --model_name Qwen/Qwen2-57B-A14B-Instruct \ --device 0 \ --group_size 128 \ --nsamples 512 \ --bits 4 \ --iter 1000 \ --disable_eval \ --fp_layers "shared_expert_gate,gate" \ --deployment_device 'auto_round' \ --output_dir "./tmp_autoround" ``` we found the output of model.layers.3.mlp.shared_expert.down_proj could be up to ~50k if adding chat template and will cause some backend like exllamav2 oeverflow. so after quantizing the model, please manually add this to config.json ~~~bash "extra_config": { "model.layers.3.mlp.shared_expert.down_proj": { "clip": true }, } ~~~ ## 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](https://github.com/intel/neural-compressor) * Intel Extension for Transformers [link](https://github.com/intel/intel-extension-for-transformers) ## 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](https://arxiv.org/abs/2309.05516) [github](https://github.com/intel/auto-round)