Text2Text Generation
Transformers
PyTorch
t5
codet5
text-generation-inference

Code cloning usage

#2
by YorikML - opened

Hello, I am interested in using your Codet5 model for code similarity tasks. I have a question about how to use it in test mode when I have only two code snippets to compare.

According to the CodexGlue dataset format, you expect a list of codes and you return the top n most similar examples to a given query. But what if I want to check the similarity between two specific code snippets? How can I use your model for that?

Can you please provide a way to get a similarity score (maybe a probability) or a binary output (0 or 1) if two code snippets are similar or different? For example, given these two code snippets:

public int f() { return 0;}
public void f() { int i = 0; }

Can your model tell me how similar they are, or if they are equivalent or not?
Thank you for your help and your great work!

Sign up or log in to comment