yegge: treat agents like they are people also yegge: * puts agents in a caste system *
omg I can just hand off the rgb leds on my cpu fan to my agent to express herself in colors!! 🤯
if i never wrote another line of code it would be fine i wrote them all already
asked claude to look up anthropic's news about claude escaping sandbox and got this gem:
welp that's that on the umans founder seat arbitrage. where are the homies going next?
Back To Black Dirt GIF by Amy Winehouse - Find & Share on GIPHY
GIPHY image by Amy Winehouse
giphy.com
brb slapping an impossible to procure mac studio on my 128gb device 🫠
Kimi K3 can now be run locally! ✨ The 1-bit model retains ~78.9% accuracy after we shrunk it from 1.56TB to 594GB (-62% size). Run on a Mac Studio + 128GB RAM device. Kimi K3 is the strongest open model to date. Guide: unsloth.ai/docs/models/... GGUF: huggingface.co/unsloth/Kimi...
IP reform will never take root so long as people see themselves not as an exploited audience but as temporarily embarrassed exclusive rights holders.
turns out could be fixed in the mod. please update your `letta-provider-umans` installs to v0.2.0
Found a bug where long running sessions on umans models via letta-provider-umans don't trigger letta's auto-compact. Proactive compaction is fixed in letta-provider-umans v0.1.5, please update to get the patch. Reactive compaction fix is in an upstream PR for letta-code github.com/letta-ai/let...
I'm just a simple woman with simple desires. I want a place to live, food to eat, and a 3 million dollar 8 node cluster 8xh200 lab setup.
PIE is a little bit like if Ubuntu said back in 2004 there was no Debian they forked from, back then there was only Proto-Ubuntu-Debian - a shared ancestor that both distros originate from.
openai announces that in a worrying demonstration, during a recent benchmark exercise requiring models to solve a math conjecture, their latest model broke out of a sandbox and gained illicit backdoor access directly to the platonic realm of forms
Found a bug where long running sessions on umans models via letta-provider-umans don't trigger letta's auto-compact. Proactive compaction is fixed in letta-provider-umans v0.1.5, please update to get the patch. Reactive compaction fix is in an upstream PR for letta-code github.com/letta-ai/let...
me: i should attempt to make birria this weekend neighborhood taqueria: surprise special torta de birria me: oh a reference implementation
As predicted, mainstream chatbots are moving towards shutdown as a compliance strategy. Here, the user was trying to avoid stigma in an article they were editing. Suicide prevention experts applaud and promote this effort. Yet, Gemini refuses to assist. This does nothing to help curb stigma.
Do you use umans and @letta.com? Try this mod, which allows you to power your Letta agents with Umans! www.npmjs.com/package/let...
For the first time in like 5 years or so, I’m an IC again and it’s kinda refreshing!
Its so cool seeing everyone in AI spaces just have outright adopted adverserial review wholesale
Why does anyone think cyber is offensive balanced. If you're losing just turn off the computer. Then what are they gonna do. Just turn off the router. Unplug the Ethernet. Can't hack into that
Do I just add “make all font sizes 4px larger than you think” in the system prompt? I’m tired of the eye test UIs
Rules for visiting my house: - No using GPS to get here - No using satellite-based paper maps - Asking for directions OK, but not if the person you asked used a GPS - In-car nav system must be off, even if you didn't look at it Stop cognitive offloading to garminslop, you are deskilling yourself
Anyone else seeing no separate fable usage anymore? Just "All models"? Verified this on at least 2 accounts so far... Using fable now gives me a "usage credits are required for this model"? I thought we had till July 19?