codefuse-ai/CodeFuse-CGE-Large

CodeFuse-CGE-Large

Homepage: 🏡 https://github.com/codefuse-ai/CodeFuse-CGE (Please give us your support with a Star🌟 + Fork🚀 + Watch👀)

Model Description

CodeFuse-CGE-Large is the Large version of the CodeFuse-CGE family which is fine-tuned based on CodeQwen1.5-7B. CodeFuse-CGE-Large is distinguish on text2code task for it's powerful ability of capturing the semantic relationship between code and text.

This model has the following notable features:
● Instruction-tuning is enabled for both query and code snippet sides.
● The model obtains sentence-level and code-level representations through a layer of cross-attention computation module.
● The model has a smaller dimensional size without significant degradation in performance.

Model Configuration
Model Size: 7B
Embedding Dimension: 1024
Hidden Layers: 32
Max Input Tokens: 1024

Requirements

flash_attn==2.4.2
torch==2.1.0
accelerate==0.28.0
transformers==4.39.2 
vllm=0.5.3

How to Use

transformers

from transformers import AutoTokenizer, AutoModel
import torch

model_name_or_path = "codefuse-ai/CodeFuse-CGE-Large"
model = AutoModel.from_pretrained(model_name_or_path, trust_remote_code=True)
tokenizer = AutoTokenizer.from_pretrained(model_name_or_path, trust_remote_code=True, truncation_side='right', padding_side='right')

if torch.cuda.is_available():
    device = 'cuda'
else:
    device = 'cpu'
model.to(device)

prefix_dict =  {'python':{'query':'Retrieve the Python code that solves the following query:', 'passage':'Python code:'},
                'java':{'query':'Retrieve the Java code that solves the following query:', 'passage':'Java code:'},
                'go':{'query':'Retrieve the Go code that solves the following query:', 'passage':'Go code:'},
                'c++':{'query':'Retrieve the C++ code that solves the following query:', 'passage':'C++ code:'},
                'javascript':{'query':'Retrieve the Javascript code that solves the following query:', 'passage':'Javascript code:'},
                'php':{'query':'Retrieve the PHP code that solves the following query:', 'passage':'PHP code:'},
                'ruby':{'query':'Retrieve the Ruby code that solves the following query:', 'passage':'Ruby code:'},
                'default':{'query':'Retrieve the code that solves the following query:', 'passage':'Code:'}
                }

text = ["Writes a Boolean to the stream.",
        "def writeBoolean(self, n): t = TYPE_BOOL_TRUE if n is False: t = TYPE_BOOL_FALSE self.stream.write(t)"]
text[0] += prefix_dict['python']['query']
text[1] += prefix_dict['python']['passage']
embed = model.encode(tokenizer, text)
score = embed[0] @ embed[1].T
print("score", score)

Benchmark the Performance

We use MRR metric to evaluate the ability on text2code retrieval tasks: AdvTest, CosQA, CSN

Acknowledgement

Thanks to the authors of open-sourced datasets, including CSN, Adv, CoSQA.

License

Since CodeFuse-CGE-Large is fine-tuned based on CodeQwen1.5-7B model, our usage license follows the same terms as that of CodeQwen1.5-7B model.