I recently made my first contribution to a very good chat completion application built using TypeScript, in collaboration with the owner of the project, Majd Al Mounayer. Majd has a great understanding of development, ranging from ESLint to CI actions. As I reviewed his project, I noticed that a solid foundational structure makes an application scalable and easier for new contributors to add features.
Based on the Groq instances, it would be a good idea to provide users with an option --token_usage or -t to monitor how many tokens are being returned or used for optimization purposes. This is important because some blocks of code might exceed the token limit applied to each model.
From dicussing several comment and understand what to do to add this feature in this project.
Adding handleTokenFlag to check argument passing that have --token-usage or -tu if yes we stderr for the token usage.
[x] Build does not fail.
[x] Tested locally.
[x] Linting errors if any are resolved.
Adding handleTokenFlag to check argument passing that have --token-usage or -tu if yes we stderr for the token usage.
[x] Build does not fail.
[x] Tested locally.
[x] Linting errors if any are resolved.
The suggested from him pointing out that on the argument checking into CLI we don't need to wrapped it with try-catch block so I fixed and make a change.
Adding handleTokenFlag to check argument passing that have --token-usage or -tu if yes we stderr for the token usage.
[x] Build does not fail.
[x] Tested locally.
[x] Linting errors if any are resolved.
When optimizeit processes several files at once the output can be very big. This would then force the user to scroll all the way up to see the tokens which is not user friendly. I think that the token info needs to be displayed at the bottom most of the program output below all processed file outputs.
From the quote I made a change by calling stderr at the end of application to display the token at the bottom most of the application.
Adding handleTokenFlag to check argument passing that have --token-usage or -tu if yes we stderr for the token usage.
[x] Build does not fail.
[x] Tested locally.
[x] Linting errors if any are resolved.
This bug because of wrong if condition inside completion processing the if condition check the flag of --token-usage being pass to the argument or not but then I'm using it with else condition which throw the error if not found response.token so instead of
if (tokenUsageInformation && chatCompletion?.usage) {
this.saveTokenUsageInfo(chatCompletion?.usage);
} else {
throw new Error(`
Token Usage Information is not available for file: ${fileName}
`);
change to
if (tokenUsageInformation) {
if (!chatCompletion.usage) {
throw new Error('Token usage information is not available');
}
this.accumulateToken(chatCompletion?.usage);
This ensures that the token usage is properly handled, and if the information is unavailable, an appropriate error is raised without breaking the application flow.
Overall, contributing to this project allowed me to learn more about different coding styles and adapt to them. Majd’s consistency in using ESLint made the development pattern very clear, helping ensure a smooth contribution process.
The above is the detailed content of First Contribute. For more information, please follow other related articles on the PHP Chinese website!