PyCoder Extension

This extension utillises the PyCoder model to generate code suggestions automatically as the user enters code into the workspace. The PyCoder model is trained via a multi-task training strategy, to learn the token prediction task and the type prediction task.

Features

The extension generates code suggestions as the user types in the workspace. Code suggestions are displayed inline at the user's current position in code. Alternatively, users can highlight a portion of code, right click and select the option Show Suggestions from PyCoder to suggest based on the highlighted code.

Requirements

• Latest python version
• PyTorch
• transformers
• fuzzywuzzy
• waitress

Run the server by navigating to /src/flask_server and run python waitress_server.py

Open the workspace in vscode, open extension.ts and press F5 to run the extension

Extension Settings

This extension contributes the following settings:

• pycoder.multiSuggestions: Specify the number of suggestions to be generated.
• pycodert.numberOfLinesToGenerate: Specify the number of lines of code to be generated.
• pycodert.inferenceMode: Select the mode of inference: local or remote.
• pycodert.remoteInferenceIP: Specify the IP of the remote inference server.
• pycodert.remoteInferencePort: Specify the port of the remove inference server.
• pycodert.codeCompletionModelDirectory: Specify the directory of a trained model.

Known Issues

None.

Release Notes

1.0.0

Initial release of Pycoder