bead README

Two main features:

• A code auto completion tool for openai/fauxpilot server.
• Export type-related prompt text.

The server part is not open source. You can download it from release page.

vscode market

bead-server

start command: .\bead-server.exe --config_file .\bead-server.config.json

config options

• openai: which model do you want to use? (use model name)
• models: model name and model info
  • promptMaxLength: max characters the client will send.
  • topicPrompt: useless, you can change it in vscode command: Bead: Set Topic Prompt.
• functionReferenceCount: when construct a prompt, how many methods(code) above this method will include?
• singleRefFunctionMaxCharCount: characters.
• textCompletionReqAllowTime: milliseconds. packet sent from vscode and packet handle time will less than this value.

deepseek-chat model

It can use, but very slow(response in seconds). Prefer fauxpilot server(my gpu is 4070TI, very good feeling).