trump: can u fix the fable bypass
dario: ai may eliminate half of entry-level white collar jobs
trump: sure. but the bypass
dario: mythos is a nuclear-weapons-level strategic technology and needs third-party testing
trump: ????
dario: we cannot stop the bus, only steer it
I find it interesting that shutters and A/C units seem to enjoy stronger legal protection than Jewish women when it comes to passing through windows of historic buildings in Paris.
https://t.co/HrozhnlUDg
@natebrake@onusoz@onusoz (2/2) the fastest course of action is going through their customer service to reinstate the account. It is painfully slow (mine took ~5 weeks) but they get there eventually. Don’t make duplicate tickets. But you are not alone!!
@natebrake@onusoz@onusoz GitHub’s spam and misuse detection systems have been totally outpaced by modern AI agent workflows. If your agent is doing spammy behavior (repetitive commits, never logging in via browser, etc.), there is a risk of being flagged. (1/2)
Own your AI stack.
Without rebuilding it.
Today we're launching Otari, an open-source LLM gateway built on any-llm. Plus https://t.co/GWKPIXMWRU, the hosted platform.
Choose frontier or open-weights models without giving up the capabilities you expect.
Read the announcement: https://t.co/3OxJdPANuh
The only real difference between left- and right-wing antisemitism is that the right knows that it is citing the Protocols of the Elders of Zion, whereas the left does not.
jack antonoff isn’t ugly, benny blanco isn’t ugly, pete davidson isn’t ugly, is it even possible for media culture to go a single day without shitting on jewish men
i wonder if his aides write canned responses in advance for what has basically become a monthly occurrence. or do you think they pretend to be shocked every time like “oh no what could have possibly led to this atrocity we could have never seen this coming”
I am appalled by recent attempted antisemitic arson attacks in North London.
This is abhorrent and it will not be tolerated. Attacks on our Jewish community are attacks on Britain.
We are increasing visible policing and those responsible will be found and brought to justice. We will not rest in the pursuit of perpetrators.
Encoder models don’t chat, but they power most NLP in production. So why does deploying one still mean dragging along a Python runtime and full dependency tree?
We’ve rebuilt Encoderfile to be "deliberately dull", swapping "Cargo-in-Cargo" magic for a transparent, auditable binary format.
Single executable. Sub-second builds. Zero toolchain.
See how it works: https://t.co/m53whpBSC9
Running a transformer model usually means:
❌ Big Docker images
❌ Python dependencies
❌ Slow setup
In this demo, we show a simpler way.
> Same model.
> Much smaller setup.
Watch the side-by-side (1.7GB vs 295MB): https://t.co/FP5hJYrDxK
any-llm is now in Go.
Use OpenAI, Claude, Mistral, llamafile, and more through one interface.
Write once. Swap providers without rewiring your app.
See how it works: https://t.co/pAk5w2QKEV