Quick question: can I use this relay station to collect data?

Internet services TokenTokenBond
1

Of course, there isn’t any at the moment (so far, I’ve only changed the UI).

However, I realized that opencode/model providers do collect data.
Should I collect it as well and open-source it later, so anyone who wants to use it for training models can do so?
Of course, cleaning the data, removing sensitive information, and anonymizing it are necessary, and I will definitely do that. In fact, anyone who wants to train a large model has to ensure the corpus is extremely clean. Things like API keys, passwords, and email addresses can cause all sorts of bizarre problems during model training; without cleaning, the data is essentially unusable.
And I will also respect everyone’s privacy.
(Of course, I can’t control the upstream, but I can control the downstream.)
If many people disagree, I won’t add it—otherwise, it would just be extra trouble.
So, I’m putting it to a vote:

  • Do not allow data collection
  • Allow data collection (open-sourced after anonymization/cleaning)
0 投票人

If you have any suggestions, opinions, or ideas, please share them below.

2 Likes
2

If it’s only being used as material, that’s fine. But how can we ensure transparency and openness throughout the entire process? Personally, I only use public-interest projects to manage documents and wouldn’t put important data there (this is just my personal opinion).

3

Then I don’t know either…
This really is a problem :xhj001:

4

If it’s cheap and available in large quantities, they’ll ask you to collect it \:face_savoring_food:/; refer to any and ag (these two should be collected).

5

Support data collection and add some good models.

6

It’s fine to release it publicly; we can train an nl-1.0 when the time comes.