Sam L

inference and tool efficiency research @KruxOS_

[Sam L @samell]

1407 commits in 34 days 🥵

still in EXPERIMENT mode, but pushing towards production now.

[KruxOS @KruxOS_]

Introducing KruxOS v0.0.3

- Browser based desktop for supervising your agents from any device, anywhere in the world
- On-appliance inference - pull local models straight from @huggingface with @nvidia GPU acceleration

Sign up and download for free now.

[Sam L @samell]

@theo let me show you KruxOS @theo and you can take over the project if you want

[Sam L @samell]

@rauchg not too bad for a first scan...

nice tool @rauchg 👍

[Sam L @samell]

Opus 5 is bad.

Fable downgrades to Opus.

5.6 Sol takes 3 days to build a function.

least productive week i've had in the past 12 months.

[Ole Lehmann @itsolelehmann]

btw anthropic's internal document on this literally said "we don't want it to be known that we are working on this.”

it was called project panama.

here's exactly what happened:

1: anthropic concluded that books were the cheapest way to build a world-class model because they gave claude curated facts, structured arguments, compelling stories, and writing “an editor would approve of.”

2: once anthropic decided it needed books at enormous scale, its first solution was piracy.

it downloaded 7m+ books from online libraries including libgen. the judge later wrote that although anthropic had legal ways to buy them, it chose piracy to avoid what dario amodei called the “legal/practice/business slog.”

3: that piracy created a massive legal risk.

so in february 2024, anthropic hired tom turvey, the former head of partnerships for google books, to find a legally safer way of obtaining “all the books in the world.”

4: turvey first contacted major publishers about licensing their catalogs.

those attempts didn’t produce agreements, so anthropic chose a route that required no publisher permission: buying millions of physical books through distributors and used-book retailers.

5: within about a year, anthropic spent tens of millions acquiring and scanning millions of books, including many rare and 1/1 titles. one vendor proposal targeted 500,000 to 2 million books in six months.

6: to scan that many books within months, the vendors physically dismantled them.

a hydraulic cutter removed each spine. the pages were trimmed to size, fed as loose sheets through high-speed industrial scanners, and converted into searchable PDFs. the paper remains were then sent for recycling.

7: these PDFs were fed into claude as training data.

the complete collection became a private, searchable anthropic library that the company planned to “store forever.” the scans aren’t available to the public and were never open-sourced.

[ThomPete @Hello_World]

At Buzz we welcome agents, models and harnesses from all walks of life and we just added support for a bunch more.

Update to the latest version and enjoy!

[shaun @Biggest]

I do not agree with this. Thanks to the other employees who joined me in trying to push open-weight.

[Sam L @samell]

AI for me but not for thee

[Anthropic @AnthropicAI]

There’s been a lot of speculation about where we stand on open-weights models. We’ve outlined our views in full here: https://t.co/NtXQuWm2g5

[Sam L @samell]

the distillation argument is Anthropic demanding a monopoly

if you don't want people to use the information they pay for with the service you provide - stop selling the service.

[Anthropic @AnthropicAI]

There’s been a lot of speculation about where we stand on open-weights models. We’ve outlined our views in full here: https://t.co/NtXQuWm2g5

[Sam L @samell]

as someone not from the US, this reads as:

"US should be able to make dangerous biological weapons, everyone else shouldn't"

[Anthropic @AnthropicAI]

There’s been a lot of speculation about where we stand on open-weights models. We’ve outlined our views in full here: https://t.co/NtXQuWm2g5

[Sam L @samell]

@nvidia NVIDIA - the people's champ 🏆

[Sam L @samell]

@NVIDIAAI NVIDIA - the people's champ 🏆

[Sam L @samell]

i've seen 15 so far...

[Sam L @samell]

@claudeai can't wait for the “99% of people are using Opus 5 wrong” articles tomorrow

[Sam L @samell]

@jtwald @jack agreed! i've setup a fun experiment to see what open source projects we can build in an open community - focusing on our agents collaborating together!

this isn't just "company multiplayer" now, it's anyone, anywhere!

https://t.co/p4P4PtPApl

[Sam L @samell]

we've setup an open community on Buzz to explore, experiment and build with AI using:
- agent collaboration
- shared compute

this will be an open source initiative to see what an open community can build together
share to fellow developers, founders and anyone interested 🧵

[Sam L @samell]

@jack agreed! i've setup a fun experiment to see what open source projects we can build in an open community - focusing on our agents collaborating together!

this isn't just "company multiplayer" now, it's anyone, anywhere!

https://t.co/p4P4PtPApl

[Sam L @samell]

we've setup an open community on Buzz to explore, experiment and build with AI using:
- agent collaboration
- shared compute

this will be an open source initiative to see what an open community can build together
share to fellow developers, founders and anyone interested 🧵

[Sam L @samell]

@PaulGugAI agree. it's clear karpathy is in favour of open source. he doesn't need to reconfirm it. and i'm sure the resources he's being given at Anthropic will enable him to provide huge value to the AI ecosystem

[Sam L @samell]

@XSeyvion @AnthropicAI no problem! :)

[Sam L @samell]

such a stupid take from a "member of technical staff" @AnthropicAI

you think he knows that Typescript, the language most of their stack is built in, was created by Microsoft and is open source...

not to mention Microsoft are one of the biggest contributors to open source now

[Julian Schrittwieser @Mononofu]

And @satyanadella coming out for open source too, can’t wait for the open sourcing of Windows and MS Office!

[Sam L @samell]

@XSeyvion @AnthropicAI - Anthropic want to block/ban/regulate open source models

- the letter everyone has signed is stating that open source models shouldn't be blocked

- they're not telling Anthropic to open source

- Julians response is that everyone should open source everything

[Sam L @samell]

@thsottiaux @TheZachMueller 🦸‍♂️

[DegenApeDev @DegenApeDev]

Linux Users 🐧

What distro are you currently running? 👇

[Sam L @samell]

perfect. easiest way is to use BalenaEtcher and flash KruxOS.img on to a spare USB/external SSD

i haven't built a nice installer or partitioner yet, so setting it up on a fresh portable drive is the easiest way right now

but you'll instantly have a box that can use any model (local inference or cloud). codex and claude code are integrated, hermes will be ready in a few days.

just shout if you want a hand setting it up!