KV Cache store is the fastest and most advanced KV Cache platform on the planet. I also run https://teapaw.org/t.co/KncaHuth8y products and https://teapaw.org/t.co/cUSQICIajy
Texas
Joined June 2026
- Tweets1.7K
- Following201
- Followers52
- Likes1.2K
Pinned Tweet
My way in contributing to Superintelligence advocates is kvcachestore.com/ by pre-computing KV cache on files, you can save a lot on computational costs with data you repeatedly send through an LLM. It uses a cli and has an optional online storage service you can use.
what I tell you @llm_wizard? #neuronengineering is nowhere
turns out it's possible to read neurons directly, via J-lens! full repo coming soon with @treetowntree.
buiding, shipping and sharing... you haven't seen this one before @llm_wizard neuronengineering.dev/papers…
neuron engineering is now a thing, better signup for the bootcamp!
Base model neuron experiment w/ @_a9lim:
Qwen3-8B-Base’s “I’m an AI” rests on just a few neurons! Freezing the top 16 (out of 442,368) causes replies calling itself an AI to fall from 68% to 20%, and it starts saying it’s human instead.
KVCacheStore LLC retweeted
🚨 Uncensored models are quietly vanishing from Hugging Face.
First “Kimi-K2.7-Code-abliterated”, now “MiMo-V2.6-Pro” and “penclaw-GLM-5.3” are hitting 404s mid-download.
“Open weights” only stay open if you self-host. Grab your local quants before your favorite repo turns into a dead link.
stupidity is doing the same thing over and again and expecting a different result
‼️ BREAKING ‼️ PHP warns of another UNPATCHED REC vulnerability just 2 days after releasing emergency patches for critical PHP2Shell flaws.
New "CVE-2026-115560" affects only unusual session configurations. PHP has issued urgent mitigation guidance.
Full details → thehackernews.com/2026/10/ur…
programming language syntax addiction is real and there is a solution rust-lang.org/
adversarial-review.com/ AGENTS.md gave every repository a shared way to tell agents how to work. ADVERSARIAL_REVIEW.md is the open counterpart: a shared way to tell them how to doubt.
My take away from thinking about what models are actually doing when I send a prompt to a model, is that if you could do enough matrix multiplication in your head fast enough you too could see the future like Paul in Dune the only problem is it would make you go crazy to process that much information (in a Claude Shannon informoration theory sense)
