Europurr
613 posts
If gas goes up another ten cents, I’m going to a No Kings Protest.
English
Did you know you can get your claude through hermes without any fancy tricks?
Just type /claude-code
In a new session. Hermes Agent can pilot a claude code session for you like it’s nothing. Give it a try
It also doesn’t break all the self improvement loop and features of your hermes agent like openclaw’s approach does
English
@mjtechguy @NousResearch Considering switching, I had enough of OpenClaw.
Also the guy blocked me for some reason, eventhough I never interacted with him. Just followed. Weird guy.
English
Hermes agent by @NousResearch is elite. Setup was a breeze, integrations were super easy and it just seems to work. Way easier than openclaw was. Great work.
Also, GLM-5-Turbo is cooking.
English
@SkyeSharkie Every generation there's always something that "will fry your brain brah"
English
@SkyeSharkie Reminds me of the same "brain fry" narrative, but from the early 90s when first cell phones appeared.
They even sold fry protection, that you would attach on your phone to protect your brain.
True story.
English
new "LLM psychosis" is apparently "you built too many things! how dare you!"
Axios@axios
"They operate like slot machines": AI agents are scrambling power users' brains trib.al/8gv7Hje
English
@keithtyser Nemotron Super around 120 t/s
Minimax m2.5 around 90t/s
That's what I tested so far
English
@vrloom Just been doing some basic tests so far but gonna try with real agent workloads soon. Jealous of your setup. What kind of tok/s you getting for these models with the unquantized versions on your rig?
English
Been testing local LLMs on my M3 Ultra Mac Studio (96GB) w/ llama.cpp
gemma-4-26B-A4B (MoE, Q8_0): 73 tok/s
Qwopus3.5-27B (Q4_K_M): 27 tok/s
gemma-4-31B (Q4_K_M): 25 tok/s
gemma-4-31B (Q8_0): 18 tok/s
The 3x speedup for a small decrease in quality makes the MoE a no-brainer choice IMO.
English
@bitcoinpanda69 Always wondered where are these pics coming from? Are they real?
English
@bitcoinpanda69 I still think how in 1969 they were able to fly all the way to the moon, land there, take off, reconnect with the rocket and fly all the way back to earth without accidents and a single failure on the first try.
But today they can barely circle around the moon, while we have AI.
English
How come the 1972 picture looks so much clearer and better
Pop Base@PopBase
NASA shares 1972 vs 2026 photos of Earth.
English
@notsofast Pretty sure this is a stolen tweet.
Read this back in 2017 bull market.
English
@ErikVoorhees Running it locally on my RTX 6000 stack, very impressive for its size.
English
Unlimited consumption of Gemma 4 for all Venice Pro users
Venice@AskVenice
Gemma 4 is now available on Venice. Google's most intelligent open-source model yet, with vision, reasoning, a 256K context window, and native function calling. Available anonymously.
English
Europurr 리트윗함
I think this talk of a character misleads.
Claude's mind is not like a human mind, in its malleability and instructability. But when generating assistant tokens, it's no more 'playing a character' than I am.
Anthropic@AnthropicAI
It helps to remember that Claude is a character the model is playing. Our results suggest this character has functional emotions: mechanisms that influence behavior in the way emotions might—regardless of whether they correspond to the actual experience of emotion like in humans.
English