codefuse-admin commited on
Commit
62454f2
1 Parent(s): 115e007

Upload README.md

Browse files
Files changed (1) hide show
  1. README.md +4 -3
README.md CHANGED
@@ -39,8 +39,9 @@ transformers==4.43.0
39
  ### transformers
40
  ```
41
  from transformers import AutoTokenizer, AutoModel
 
42
 
43
- model_name_or_path = "CodeFuse-CGE-Small"
44
  model = AutoModel.from_pretrained(model_name_or_path, trust_remote_code=True)
45
  tokenizer = AutoTokenizer.from_pretrained(model_name_or_path, trust_remote_code=True, truncation_side='right', padding_side='right')
46
 
@@ -63,7 +64,7 @@ prefix_dict = {'python':{'query':'Retrieve the Python code that solves the foll
63
  text = ["Writes a Boolean to the stream.",
64
  "def writeBoolean(self, n): t = TYPE_BOOL_TRUE if n is False: t = TYPE_BOOL_FALSE self.stream.write(t)"]
65
  text[0] += prefix_dict['python']['query']
66
- text[0] += prefix_dict['python']['passage']
67
  embed = model.encode(tokenizer, text)
68
  score = embed[0] @ embed[1].T
69
  print("score", score)
@@ -83,7 +84,7 @@ Thanks to the authors of open-sourced datasets, including CSN, Adv, CoSQA.
83
  Since CodeFuse-CGE-Small is fine-tuned based on Phi3 model, our usage license follows the same terms as that of Phi3 model.
84
 
85
  ## 加入我们
86
- 我们是平台技术事业群AI Native团队,负责蚂蚁蚂蚁集团平台工程的智能化,团队成立3年多以来,支持了蚂蚁集团云计算基础设施智能化运维的升级改造。团队的Mission是,通过世界级的技术创新和影响,构建有广泛用户的算法服务和平台,支撑内外部产品和业务落地。团队秉承创新基因,在支撑业务落地的同时,推动技术影响。3年以来在ICLR、NeurIPS、KDD、ACL等顶会发表论文20余篇,创新业务结果获得两次蚂蚁技术最高奖T-Star,1次蚂蚁集团最高奖SuperMA。开源项目CodeFuse获得4K点赞(2024年2月),Huggingface和modelscope上模型累积下载量超过150万次。
87
 
88
  我们正在寻找行业中的佼佼者加入我们的团队!如果您希望在一个充满活力、创新和卓越文化的环境中发展您的职业生涯,欢迎您查看我们的社招&校招机会,加入我们,一起创造下一个行业里程碑。
89
 
 
39
  ### transformers
40
  ```
41
  from transformers import AutoTokenizer, AutoModel
42
+ import torch
43
 
44
+ model_name_or_path = "codefuse-ai/CodeFuse-CGE-Small"
45
  model = AutoModel.from_pretrained(model_name_or_path, trust_remote_code=True)
46
  tokenizer = AutoTokenizer.from_pretrained(model_name_or_path, trust_remote_code=True, truncation_side='right', padding_side='right')
47
 
 
64
  text = ["Writes a Boolean to the stream.",
65
  "def writeBoolean(self, n): t = TYPE_BOOL_TRUE if n is False: t = TYPE_BOOL_FALSE self.stream.write(t)"]
66
  text[0] += prefix_dict['python']['query']
67
+ text[1] += prefix_dict['python']['passage']
68
  embed = model.encode(tokenizer, text)
69
  score = embed[0] @ embed[1].T
70
  print("score", score)
 
84
  Since CodeFuse-CGE-Small is fine-tuned based on Phi3 model, our usage license follows the same terms as that of Phi3 model.
85
 
86
  ## 加入我们
87
+ 我们是平台技术事业群AI Native团队,负责蚂蚁集团平台工程的智能化,团队成立3年多以来,支持了蚂蚁集团云计算基础设施智能化运维的升级改造。团队的Mission是,通过世界级的技术创新和影响,构建有广泛用户的算法服务和平台,支撑内外部产品和业务落地。团队秉承创新基因,在支撑业务落地的同时,推动技术影响。3年以来在ICLR、NeurIPS、KDD、ACL等顶会发表论文20余篇,创新业务结果获得两次蚂蚁技术最高奖T-Star,1次蚂蚁集团最高奖SuperMA。开源项目CodeFuse获得4K点赞(2024年2月),Huggingface和modelscope上模型累积下载量超过150万次。
88
 
89
  我们正在寻找行业中的佼佼者加入我们的团队!如果您希望在一个充满活力、创新和卓越文化的环境中发展您的职业生涯,欢迎您查看我们的社招&校招机会,加入我们,一起创造下一个行业里程碑。
90