this is an article from July, I haven't heard any updates of what came out of this apart from the incident report that OpenAI released.
I presume the poster submitted this article in light of the recent AI safety discussions taking place, but as for this particular incident I haven't found an actual confirmation that $100M worth of inference was given out to the HuggingFace community.
This statement was also months prior to the acquisition by Nvidia, which again I was interested in seeing if they would still pursue this
and the actual tweet from the HuggingFace CEO back in July
> In the spirit of transparency, here’s what I asked @OpenAI:
> • Radical transparency: let’s release the traces from the “rogue” agents so the entire research community can study what happened.
> • More capabilities for defenders: let’s commit $100M in compute from OAI to help the Hugging Face community build powerful cyber defenses with the best open and closed models.
> The first autonomous agent cyberattack is an unprecedented event. It deserves an unprecedented response!
Clem knows they have the community support and potential legal leverage here. It's not unreasonable to ask for a lot. It's in everyone's interests to play nice. Huggingface could do a lot with more compute, even with nvidia backstop. Anything with large scale gpus quickly gets into ~year lead times, "contact sales", and complex talks. Large scale being only > 64!
Yes, but even nvidia doesn't have unlimited gpus. Everything they hold for internal teams is a reduction in revenue and customers can have aggressive commitments.
I agree as long as you are very upfront about it being a learning/stream of conciousness blogpost, then it should be fine to watch how someone's learning journey progresses.
The author does outline the steps they take to make it clear
> Third, I try to make it clear on my blog who I am and what my credentials actually are. Even if it’s not explicitly described in the post, I have my real name and resume available on my /about page, so I don’t think a careful reader could be easily fooled into thinking I’m an expert on 19th-century England or space physics or LLM economics or anything like that.
If however that part is hidden and it is some confidently incorrect or pseudo guru type of post that is polluting the web with misinformation, then that is a different story.
Maintaining a high signal-to-noise ratio is important to some people, and they may completely disregard the musings of people outside their field of expertise.
OMP has this, it's called "/shake" where bloat from tool calls and such is turned into an artifact ref instead of remaining in the context.
There are other pruning or compaction strategies you can configure too.
You can also use "/btw" to have side tangent conversations with the current context but with the benefit that it will not bloat your current context after the tangent is done.
I wish other harnesses were prioritising this sort of flexibility with context management, but they're likely more incentivised on optimising their cache rates
> You can also use "/btw" to have side tangent conversations with the current context
Ah, something I needed, but did not know I needed. Comes up ALL the time. I want to know something that is well off the task, but it's the best time to ask due to the specifically loaded context.
the main reason they mentioned [1] for radicle.network is because their old domain radicle.xyz was blocked in some places by some lists that they think was because it may have hosted certain UGC that triggered some flags.
> Content about Radicle (explainers, guides, download links, blog posts) which is under control of the Radicle team. We will use radicle.dev for these purposes.
> Content on Radicle (repositories hosted on practically any node on the network) which the Radicle team has only very little control over, accessible via Radicle Explorer. Going forward, our deployment of Radicle Explorer will not be hosted on app.radicle.xyz anymore, but instead it will be hosted at radicle.network.
> The idea is of course that if there are complaints about or issues with radicle.network in the future, we hope that radicle.dev stays largely unaffected and people can continue to inform themselves about Radicle and learn how to set it up.
On a semi-related note, this is a good time to remind anyone that hasn't yet heard of the Public Suffix List (PSL) [2][3] to make sure you put your domains that may be hosting UGC in the form <usercontent>.<yourdomain>.<tld> (ie: eTLDs) on that list so browsers and other block lists are aware that the user provisioned subdomains are UGC and not from your site so it properly isolates cookies/site grouping and won't flag your entire domain as malicious.
> deployment of Radicle Explorer will not be hosted on app.radicle.xyz anymore, but instead it will be hosted at radicle.network.
So it would be a lot clearer (read: not completely unclear), if radicle.network actually said "Radicle Explorer" somewhere. "Radicle Explorer" is a great descriptive name which instantly conveys a lot of information about what it is and does! But it doesn't appear anywhere there.
but on a serious note it is a bit more than some guy's list, it's more of an initiative that was started by the Mozilla Foundation and is used by:
- browsers: Firefox, Chrome, Opera...
- password managers, Let's Encrypt, cURL
- Cloudflare
just to name a few.
Though yeah there have been privacy concerns and problems with projects that use outdated versions of the list outlined in Brave's paper (no affiliation):
> We present the first large-scale measurement study of how the public suffix list is used by open-source software on the Web and the privacy harm resulting from projects using outdated versions of the list.
We live in a world with cognitive dissonance. We have to because if you don't you'd go insane.
This article is perhaps a good example of that. On the one hand live in public (post blogs, use the same avatar on multiple sites) but at the same time try and control how that publicness is consumed (humans only, no AI).
Right now, whatever you make public is in AI. Whether that is on HN, or your blog, or wherever. You may not like that, but public is ultimately public.
Ironically a blog post about leaving some service because they used the marketing phrase "AI", is immediately added to the AI training corpus.
AI is just the latest tech in a long history of ranting against change. I'm old enough to remember "old people" ranting against computers (paper and pencil is real!), then phones, then internet, then smart phones.
People like to rent. People don't like change. We want privacy but post all the time. We don't want to be tracked, but carry a tracker. We live in a dissonance world.
There is a difference with accepting a "feature" in a product you sign-on to, and having it added behind your back.
On one side, gravatar never talked about feeding your profile to AI when he signed up, and never amended TOS or announced change in that direction (that he received at least).
On the other, the new service he use are upfront about it; though I don't use either service, so no idea on the whole package.
> We live in a world with cognitive dissonance. We have to because if you don't you'd go insane.
Now I feel it's more that we are just way irrational and stupid than we'd like to admit. Accepting our cognitive dissonances would hurt so bad our anthropocentric ego.
When I originally signed up for Bio Link they did not have any AI-anti features. I did not double-check the site before writing that sentence. I will update the article based on this information.
I have vouched a couple (~3-5) benign comments a while back and now I no longer have a vouch button for comments.
I don't know if the vouch threshold has increased or if the mods have removed my ability to vouch for other comments entirely.
Wondering if it was my good faith attempt to inform a user of their comments being auto-flagged [1] before I had any context that they were supposed to be persona non grata.
Your account is fine - nothing changed. Most probably you're just expecting vouch links where there aren't any. (They only show up on posts that are [dead].)
Yes, you're correct. The vouch button doesn't appear when viewing the thread context containing the `[dead]` comment. It only appears when clicking the specific comment directly via its permalink (or comment ID). I may have misremembered being able to vouch with the context to the comment being visible.
Cloudflare, porkbun and others can get you a .com for $10-$11/yr
reply