3.1 KiB
GPT-Code-Clippy (GPT-CC)
code_clippy_logo.jpg
Open Source GitHub Copilot for auto generating code
I would like to train an open source version of the new awesome GitHub Copilot AI tool, which is based on GPT3. Similar to the awesome people behind GPT-Neo, having such an open source model would greatly help researchers understand what this type of biases and limitations this kind of code autocompletion model might have such as generating insecure code (i do research in this area and i know my team would love an open sourced version to run experiments on, i.e. try and break it 🤓)
Getting the data
Downloading the data
Further processing the data
Finetuning the model
Evaluating the model
Using the model
Possible links to publicly available datasets include:
Some additional datasets may need creating that are not just method level.
5. Training scripts
I believe the standard CLM language model script would do for this.
We can make use of https://www.github.com/huggingface/transformers/tree/master/examples%2Fflax%2Flanguage-modeling%2Frun_clm_flax.py
for training the scripts you can run:
python run_clm_streaming_flax_v2.py
6. Usage
code for running the code generation is done by using:
bash run_cln_straming.sh
run_cln_straming.sh contains all the hyperparameters and will be used to generate code.
we have also generated the code for the following languages: python javascript c++ c java
We have used GPT-Neo using 13B and 27B parameter settings.
to run for following files:
13B:
bash run_cln_gpt_neo_13b.sh
27B:
bash run_cln_gpt_neo_27b.sh
7. (Optional) Challenges
The data additional data may be a challenge. From what I can see in copilot, it looks to be training on entire files, not code snippets. There are file level datasets that exist but they are a few years old and i don't think they cover many programming languages. The ones I listed above have multiple languages but are only methods.
However, githubs API is pretty easy to use and so it would be pretty easy to create one from scratch, especially if we get some insights into how the copilot dataset was generated 🤓
8. (Optional) Desired project outcome
I'd love to have this open source model setup in a similar Visual Studio Code extension to the GitHub Copilot one. I've actually made a tutorial on doing this using the GPT-Neo model, so we could easily clean it up and release it free of charge forever because from what I've seen on Twitter the GitHub Copilot might eventually be put behind a paywall 😢.
9. (Optional) Reads
The following links can be useful to better understand the project and what has previously been done.
- https://github.blog/2021-06-29-introducing-github-copilot-ai-pair-programmer/
- https://youtu.be/nC3NrhoNeP4 (tutorial on how we could setup the demo of the model once it's done cooking)