Hacker News: ai_what

New comment by ai_what in "EU Council to Vote on Chat Scanning Proposal on Thursday"

ai_what — Wed, 19 Jun 2024 08:54:36 +0000

So you also believe the "socialist parties" aren't actually socialist?

New comment by ai_what in "Sharing new research, models, and datasets from Meta FAIR"

ai_what — Tue, 18 Jun 2024 20:48:04 +0000

I can't speculate that for you. Try reaching out to them if you must know.

New comment by ai_what in "Sharing new research, models, and datasets from Meta FAIR"

ai_what — Tue, 18 Jun 2024 19:42:26 +0000

Try a different email address, that's what fixed it for me.

New comment by ai_what in "Sharing new research, models, and datasets from Meta FAIR"

ai_what — Tue, 18 Jun 2024 19:38:00 +0000

Yes, Yann LeCun posted about it on twitter ~2 weeks ago.

Edit, it was a month ago: https://twitter.com/ylecun/status/1793181068943639014

New comment by ai_what in "Deepseek Coder V2 Beats GPT-4o, Opus"

ai_what — Mon, 17 Jun 2024 19:57:08 +0000

Was the lite model also tested?

New comment by ai_what in "Show HN: Paste2Download – No Login, No Ads, Download Videos Watermark Free"

ai_what — Sun, 16 Jun 2024 08:25:40 +0000

Very nice, I was looking for something like this.

The only feedback I have is that on the frontpage it says "Download any tiktok video" but you support more websites than just tiktok. So maybe add that piece of information :)

New comment by ai_what in "Tell HN: YC Hired Wire Fraud Soliciting. Ask: Best Way to Report to Authorities?"

ai_what — Sat, 15 Jun 2024 23:09:29 +0000

Isn't that just what the average spam email these days does? Scrape emails and try to engage in fraud? I'm not saying it's ok but....

It seems kind of a waste of time to hunt one particular instance of this down. I can give you the emails in my spam folder, it's full of these kinds of people.

New comment by ai_what in "Show HN: Yesterday's Domains levels up expired domain finding platform"

ai_what — Sat, 15 Jun 2024 08:37:23 +0000

My 2 cents: you should offer a limited free plan so people can see if it's worth it.

New comment by ai_what in "Vercel ends open-source sponsorship program giving projects 24hr notice"

ai_what — Fri, 14 Jun 2024 20:00:19 +0000

Yeah this is strange. If they are using, lets say $1000 in credits per day, then they will be charged $700 starting tomorrow. Like you said, they might not even see the email yet.

New comment by ai_what in "Mozilla defies Kremlin, restores banned Firefox add-ons in Russia"

ai_what — Fri, 14 Jun 2024 17:22:36 +0000

Russia often bans entire websites and IP ranges, making it harder for non-tech people to get certain software.

Mozilla was likely hoping that nobody from the West would notice that they complied and removed the addons.

Just goes to show you how all those "values" they stand for only matter when it doesn't make them lose any users.

New comment by ai_what in "Meta Open-Sources Megalodon LLM for Efficient Long Sequence Modeling"

ai_what — Tue, 11 Jun 2024 19:02:16 +0000

Didn't this happen in April?

Edit: Yes it did, https://github.com/XuezheMax/megalodon

New comment by ai_what in "Surprisingly high levels of toxic gas found in Louisiana"

ai_what — Tue, 11 Jun 2024 13:48:03 +0000

From the article:

> The EPA just announced stricter rules on ethylene oxide that would require commercial sterilizing outfits to dramatically reduce emissions.

New comment by ai_what in "Google has been blocking Invidious with error "This helps protect our community""

ai_what — Mon, 10 Jun 2024 19:04:38 +0000

yt-dlp also: https://github.com/yt-dlp/yt-dlp/issues/10128

New comment by ai_what in "Qwen2 LLM Released"

ai_what — Thu, 06 Jun 2024 17:22:41 +0000

It won't. Amazon kind of when that angle with MistralLite[1] (a 7B finetune), and it was barely passing in terms of being an effective summarizer. 0.5B are pretty much useless.

https://huggingface.co/amazon/MistralLite

New comment by ai_what in "Qwen2 LLM Released"

ai_what — Thu, 06 Jun 2024 17:16:56 +0000

I agree. I personally don't have high hopes for the 0.5B model.

Phi-2 was 2.7B and it was already regularly outputting complete nonsense.

I ran the 0.5B model of the previous Qwen version (1.5) and it reminded me of one of those lorum ipsum word generators.

The other new Qwen models (7B and up) look good though.

New comment by ai_what in "Ask HN: Is Booking.com Under Attack?"

ai_what — Thu, 06 Jun 2024 13:43:41 +0000

https://news.ycombinator.com/item?id=38481258

New comment by ai_what in "GameStop stock influencer Roaring Kitty may lose access to E-Trade, report says"

ai_what — Thu, 06 Jun 2024 03:56:15 +0000

Care to explain why not? I can't think of anything he did besides post his position.

New comment by ai_what in "Spotify CEO: "cost of creating content is close to zero""

ai_what — Mon, 03 Jun 2024 21:51:23 +0000

He likely knows better, so that makes me wonder why he would say this.

New comment by ai_what in "Homegrown Chinese CPUs are catching up to AMD and Intel"

ai_what — Mon, 03 Jun 2024 21:49:22 +0000

It looks like the gap is closer faster than what was expected.

New comment by ai_what in "Ask HN: Where do you keep your messy stack of Markdown notes?"

ai_what — Mon, 03 Jun 2024 07:05:39 +0000

I wanted to like Joplin but it's not very fast. It takes like 5 seconds just to start.

So now I use: https://github.com/nuttyartist/notes