Releases: Aincvy/vscode-fauxpilot
v10.0.2 update
v10.0.1 update
Remove option reduceLineStep.
Add new option reduceLineTryTimes: When the estimated number of tokens exceeds the limit, binary search will be used to reduce the number of lines. This parameter controls the maximum number of binary search iterations.
v1.2.2 update
Add new option resendIfEmptyResponse.
If \n is in the stopWordsArray array and an empty response is received this time, should we resend the request? The new request will remove \n from the stopWordsArray array.
I think this option is effective when using the codegen2 model.
v1.2.0 update
Add parameters to automatically adjust the number of tokens sent to the server, to avoid receiving a 400 error from the server.
Add https://www.npmjs.com/package/@dqbd/tiktoken to calc token count.
v1.1.9 update
- update to v1.9
- bug fix: document.lineCount in getPrompt function is wrong
- add try catch inside provideInlineCompletionItems function
1.1.8 change
Make suggestionDelay fields effective
Add a new class FauxpilotClient to handle log and config value.
Auto reload config.
fauxpilot-1.1.7
update openai to 3.3.0, more comfortable inline completions