Views
No views yet
pip install protobuf transformers==4.30.2 cpm_kernels torch>=2.0 gradio mdtex2html sentencepiece accelerate1from transformers import AutoTokenizer, AutoModel
2tokenizer = AutoTokenizer.from_pretrained("THUDM/codegeex2-6b", trust_remote_code=True)
3model = AutoModel.from_pretrained("THUDM/codegeex2-6b", trust_remote_code=True, device='cuda')
4model = model.eval()
5
6# remember adding a language tag for better performance
7prompt = "# language: Python\n# write a bubble sort function\n"
8inputs = tokenizer.encode(prompt, return_tensors="pt").to(model.device)
9outputs = model.generate(inputs, max_length=256, top_k=1)
10response = tokenizer.decode(outputs[0])
11
12>>> print(response)
13# language: Python
14# write a bubble sort function
15
16
17def bubble_sort(list):
18 for i in range(len(list) - 1):
19 for j in range(len(list) - 1):
20 if list[j] > list[j + 1]:
21 list[j], list[j + 1] = list[j + 1], list[j]
22 return list
23
24
25print(bubble_sort([5, 2, 1, 8, 4]))@inproceedings{zheng2023codegeex,
title={CodeGeeX: A Pre-Trained Model for Code Generation with Multilingual Evaluations on HumanEval-X},
author={Qinkai Zheng and Xiao Xia and Xu Zou and Yuxiao Dong and Shan Wang and Yufei Xue and Zihan Wang and Lei Shen and Andi Wang and Yang Li and Teng Su and Zhilin Yang and Jie Tang},
booktitle={KDD},
year={2023}
}