Show Django: Is It Clickbait - Checking YouTube videos

Launching a new product built with Django :rocket:

It is still in beta, but it is live :tada:

Do you want to know if a #YouTube video is clickbait? And why?

Then paste a YouTube video link at IsItClickbait.com and check it for FREE.

:warning:Warning: Beta version :warning:

In terms of the tech stack, it uses Django (obviously), with yt-dlp to download videos and Deepgram for transcribing.

The AI analysis uses a custom fine-tuned MistralAI model. I fine-tuned the model with several curated examples of YouTube videos and corresponding analysis.

Im getting server error 500

1 Like

Sorry about that.

There was a problem with the proxy, which is now fixed.

Please try it again.

Still Server Error (500)

Thanks for the beta testing :grinning:

Optimized the download and transcribing, and also added an AI model with a larger context for longer transcriptions of hour-long videos.

It should be working now.

Youtube short. Working great :muscle:
3h course. Working great :muscle:

Well done!

Thanks for checking it out :grinning:

Btw. I think I might have pasted a “not youtube” link by mistake. Does it validate links?

Not yet. Coming in the next update.

@qvisty, if you are interested, I released a new and improved version with several fixes and a new interface.

Again, thanks for your feedback so far.

Great with the history :+1:
70% appears often :thinking:

Yep, I need to improve the fine-tuning of the AI model a bit better.

I started with a small dataset for proof of concept, need more data points as the model seems a bit biased.