⚞ Black ⚟

its better than getting votekicked in cs i guess

@Invisible_TSW you can just call it estradiol enanthate you know

CAS: 4956-37-0 For molecular biological use only, not for resale. I call it an ammo in TERF wars.

ahahahbahbaHahAhahAHAHahahaAHAHAhahah mmmmmmmmm its IBM all over again

@AutismEgregore @TeksEdge ye its not a criticism by any means, its awesome. id like to see the tradeoff between intelligence/speed for the dumber small models. would being able to process at 15kt/s outweigh the intelligence of running kimik2.5 at 15t/s?

@lostboys @TeksEdge Don't care. If they do qwen3.5-27b-abliterated, I would pay an arm and a leg for one of those cards.

🎗️ "Medium-Sized" LLM Burners Coming Soon! 🔥 This Could Make Local HyperToken Generation a Reality. ⚡️ NVIDIA’s worst nightmare? 😱 ⚙️ Application-Specific Hardware Taalas new PCIe ASIC board would burn the entire medium-sized Qwen 3.5-27B LLM straight into silicon 🤯 (already doing it with small models) Taalos said medium models on ASIC would be available in their lab by Spring '26. 💭Imagine: 🚫 No more loading weights 🚀 ~10,000 Tokens Per Second locally (Llama 3.1 8B already @ 17,000 tps) 💻 Standard PC slot, ultra-low power (10x less) 🔋 🌍 100% offline with no cloud, no GPU farm 💰 Reddit unit cost rumor $300 to $400 🖥️ Imagine HyperToken generation on your desktop. 🤖 AI agents that think at light speed. ⚡️ Are you ready? 👀

@tempstat @TeksEdge or you could just take a half measure and stick the autoregressive transformer in the silicon without the weights. that way you still have silly fast inference speed and a dynamic model. im pretty sure its what groq and cerebras do

@lostboys @TeksEdge What about separate cards and interface to switch between or run in parallel? I mean you could have 4 or 5 of this cards each asic, a different model.

using AG/AS as generic carriers has sort of "been a thing" since at least 2018. PMID: 26598698 for population reduction with a 99.5% rate. PMID: 32839345 for population modification with a 82.3% rate. im honestly kinda suprised nobody has used this for spreading a custom disease yet, using something like OX513A/OX5034 to make sure nobody yoinks and steals it as the males are infertile

Holy shit it's real The combat bug

you could probably keep the fixed weights in silicon and glue the MoE layers to it, and have a little arm processor or something work the routing logic. depends how complex that routing logic is, and it wouldnt work for distrubuted systems where 1 gpu runs many chats. itd innately tie 1 pcie : 1 chat session in its first iteration which isnt very marketable to inference providers so they'd never do it. i also wonder where the kv cache actually sits in this.

@lostboys Yeah works best w/dense models but I bet they could get MoE working w/some Verilog magic

@VodafoneUK why does your THG3000 router come with a preinstalled cocktail recipe?

@IsraelMFA maybe its because the united states, japan, india and egypt arent anywhere near europe?

Israel, as a sovereign state, legislates its own laws. Unbelievably, the European Union is intervening in the midst of a sovereign state’s legislative process. Israel is facing murderous and brutal terrorism on a scale unmatched anywhere in the world. Palestinian terrorism is fueled by payments from the Palestinian Authority, as well as by abductions that lead to shortened prison terms for terrorists. Israeli law already permits capital punishment, and under the proposed new law, discretion over its application will remain with the courts. EU double standards: as always, the EU obsessively singles out Israel - we haven’t seen a tweet like this when it comes to capital punishment in the United States, Japan, India, Egypt, or other countries.

i too have made a recursive self improving agentic ai harness. ok maybe im being a little silly, but instead of citing a summary DAG with a map-reduce tool stuck to a coding agent as evidence to credibility, why dont you cite something actually interesting thats related to context management? SSM type models like mamba3 suffer much less from the rolling context window reductionism & detail loss from the not-really-adequate compression in transformers, have linearly growing self-attention compared to a quadratic self-attention in SSM, etc? those seem like much more technically credible things to be honest, other than attribing a solution to a problem everyone kinda knows is terrible. another thing, the PR that adds custom context management into openclaw was authored by a member of marian.engineering and solo-merged with no review. Not to be prying, but considering that ME and Voltropy seem to be innately linked, doesnt this seem like a conflict of interest? github.com/openclaw/openc… the paper itself on voltropy.com has `arXiv:submit/7269166` in its watermark. this... isnt announced. its the papers internal tracking number i guess? considering that preliminary review has a 14-day inactivity limit for authors to respond to edit requests and its been about 40 days now i believe this has either expired, or been denied. I dont know why voltropy stamped their own paper on their own website with the internal arxiv tracker but im not a researcher so maybe its just "a thing" you do ¯\_(ツ)_/¯, but its not a requirement for submission as per info.arxiv.org/help/faq/whyno… This doesnt come off as organic at all and seems exceptionally veiled on purpose. I dont buy it.

Public Service Announcement:

@Clav0Updates isnt 500mg of trenbolone similar to whats given to breeding bulls that weigh about 500kg??!?! ??!? !? ? WTF???

Clavicular has officially BEATEN the “decaying” allegations and returned to his PRIME physique not even 2 weeks into his cycle of 500 mg of Tren 😳👀

@emilymontes100 @vuppygirl ewwwwwwww someone doesnt micronise their raws

@vuppygirl this is my OGL... the raw is literally separating...

@vuppygirl also, instead of using serum testosterone to garner if something actually has an estradiol ester in it, wouldnt a better idea be to... actually look at the serum estradiol levels? Unless this is just an ST test.

@vuppygirl isnt that kinda expected if you have overactive adrenals? GNRH doesnt affect them IIRC which is why monotherapy doesnt result in 0.

@Reelix @vxunderground yes, i know. it just spawned an interesting observation that privacy-oriented people never really seem to research what is actually compromised or not and instead seem to just prefer going off of the early life section

@lostboys @vxunderground This was posted to a public discord server.

Chat, this does NOT look good on paper

@sheltr64 @KovaaKs Yeah thats the only problem, itd just be funny to see people abstractly have it.

@lostboys @KovaaKs That'd be kinda lame. I do like 100%ing my games so being forced to use SAM to build my completion checklist would really suck

We're working on new Steam achievements! Reply with your idea.👇 If it ships, we'll credit you.

OpenAI puts erotic chatbot plans on hold ‘indefinitely’ ft.trib.al/4Q2hLpT

@tobyliiiiiiiiii not quite returning, just flying overhead. still cool though

Humans are returning to the Moon for the first time in 54 years exactly one week from today. This is not being talked about enough.

arc-3-agi benchmark brings forth a super cool question of when people wont be able to pass these newer and more complex benchmarks :D