ctxc

I didn't think about this parallel, but that's just so sad.

Thanks - I didn't verify this personally, but still stand corrected.

If training wasn't considered outside the law, this goes on to make the point about double standards for US vs Chinese model training methods.

ctxc

hierdiscuss

Anthropic was asked to pay book authors because it trained on pirated downloaded books.

What about books and art where the author/artist does not authorise AI to train on it? They do happily train on it, ignoring their "ToS".

This is just double standards, a slap on the wrist to not worsen the situation with authors imo.

You could also say the Chinese companies are doing the same - they _do_ pay for their Anthropic subscriptions after all.

ctxc

il y a 13 joursdiscuss

Nice work!

ctxc

il y a 15 joursdiscuss

It was always this way!

ctxc

il y a 15 joursdiscuss

Yeah weird, bc I just checked and it isn't oss. Only the SDKs are OSS, which...doesn't help much

ctxc

il y a 24 joursdiscuss

/s not needed, it's the easiest path

ctxc

il y a 2 moisdiscuss

(you mean it's _not_ AI generated)

ctxc

il y a 2 moisdiscuss

I feel like with time, 50 is the new 70.

ctxc

il y a 2 moisdiscuss

Those numbers make sense to me...120 million input tokens is like 120 sessions of hitting the full context limit, which seems like a lot to me though

ctxc

il y a 2 moisdiscuss

Haha! Made me chuckle

ctxc

il y a 2 moisdiscuss

Reads like a LOT of tokens to me. What does your usage /workflow look like? I'm v curious because although I do use Claude code, my token counts aren't nearly as much

I want to know if I'm missing something cool!

ctxc

il y a 2 moisdiscuss

When you have a general idea of what smells bad vs what's okay...why?

I'd rather get it from the LLM and review

ctxc

il y a 2 moisdiscuss

[dead]

ctxc

il y a 3 moisdiscuss

They report an hour less than average sleep time the first 3 months?!

How did get so lucky?

ctxc

il y a 3 moisdiscuss

WOW YES OMG

do you remember what the wrestling kinda game with ed ed and eddy was?

ctxc

il y a 3 moisdiscuss

Out of curiosity, what kind of custom harness are you thinking?

ctxc

il y a 3 moisdiscuss

I do something tangential. If you can pick out identifiers for these pages with deterministic features (financial pages have the most numbers on the page, or have the word "director", "general manager" and "managing director" on the same page, etc)

Pick those pages out and pipe only those to the llm

Luckily I work with excel w lots of sheets and don't need to do pdf to text conversion etc

ctxc

il y a 4 moisdiscuss

Scale, my friend, scale...

ctxc

il y a 4 moisdiscuss

Interesting, I'd love to know more. Are parts of it public?

Submissions

Human Intelligence Portal (demo, awareness)

Agentic browsers: a note for my friends

[untitled]

Ask HN: Have you directly authored a high-profile massive bug? How does it feel?

Buzzwords for the Busy: LLMs

The kind of company I want to be a part of

Making the Most of a HN Launch

The Perfect Hacker News Launch (MagicPatterns HN Login Button)

Security 101 for devs: a presentation (2014)

Poor Man's Datadog: Discord

Ask HN: Good AI tool for UI mockups without code?

Using LLMs to steganographically encode secret messages

A journey of building progressively cooler things

Ask HN: Tips to get started on my own server

What's Hashing? (2022)

Ask HN: What are some quotes that stuck with you and change the way you think?

Ask HN: Most interesting technical articles you've bookmarked

India's largest stock broker has 30 devs (2020)

QR code creation: Dynamic, step by step guide

Indian govt requires VPNs to store PII for 5 years

comments