37 comments

[ 0.18 ms ] story [ 10.3 ms ] thread
(comment deleted)
There's no way this dataset is going to survive on HF, right? It will be hit with so many DMCA takedowns.
It's metadata only, not video content. I expect it will likely survive - that's a pretty common pattern for machine learning datasets.

Stable Diffusion was enabled by LAION, for example. That was metadata about images and URLs to those images, but not the actual image files.

True, but it includes the full text of the post. That's enough for some copyright strikes.
This is being posted all over the place. on multiple subreddits and stuff. Why?
Can any brave soul wade through the LLM prose to provide a human-readable summary?
I cannot verify whether it is technically correct, but it's about how to defeat Tiktok's bot filters to scrape it.
Man, I could use $699. If this gets even a single sale, then maybe I need to try having less decency...

(I guess that is sort of a roundabout summary...)

Would you say that this is immoral? Since it’s data that’s public anyways, I don’t see why putting it in a table and selling it is a bad thing.
Because a lot of things is public online and can't be copied and sold e.g. due to copyright, patents and trademark. Also if you access a website you are bound to a ToS contract and this is a breach of that contract.
"I uploaded 4.5 billion of those videos to Hugging Face" = I scraped the metadata (title, view count, etc) for 4.5 billion TikTok videos using the same API as their Android app. I posted the data to HF as Parquet. The video content itself is not included. I'll sell you my Go scraping code.
Is this even useful in any way?
Yes, train your model to give people more AI slop until they get sick of it
Make your own recommendation engine. Even if it's just a real literal text search that sorts chronologically.
You could use it as negative reinforcement to tell the next big AI model what it should not do
There is even a use case at the end which tracks which songs are currently trending. You basically get access to all of TikTok's data. If you don't see why analytics on this data are so valuable, lookup varoufaki's concept of cloud capital.
>Is it legal? It is against TikTok's terms of service. It is sold for research and educational use.

Oh, ok. Otherwise, very detailed deconstruction to scrape their API. Lots of layers of registration and creating a request that looks like it is valid client.

That data does not belong to the public. Why do you think it is OK to steal from TikTok?
> Everything described here is a private Go repository. One-time payment, permanent access, complete source. > $699 one time · lifetime access

Not open-source apparently.

And I cant find the reddit post but I think I read that videos/assets are not actually pre-downloaded, they have to be requested through Tiktok API using the provided code. So if Tiktok patches, the code will need updates too.

I think the post is essentially a decent technical-explainer (value adding and interesting) in exchange for effectively a little product placement (selling either just the code or code already running on a server at additional cost)

But I think this is the 289GB data (free): https://huggingface.co/datasets/kuben-developer/tiktok-video...

Absolutely no consideration for the people who, when posting things to Tiktok, would prefer for them not to get scraped.
> Three things to notice, because each one bites later:

Very LLMish language!

The AI keeps mentioning how a HTTP 200 can silently pollute your dataset. Why not just check contents of body? Usually APIs follow strict JSON contract for successful queries, alert or throw an error when that changes.
Textbook example of what not to post on the internet. Unreadable incoherent LLM hallucination slop.
Well, most engineers are familiar with GIGO. Garbage in... Err, thats where my comment stops. Garbage in. Yup.
Very impressive figuring out all 4 checks. I'm not even sure this was reverse engineered, rather than leaked. I wish it mentioned anything about the methodology they used