ohoextra @ohoextra

💥 OBLITERATION ALERT 💥 GOOGLE: PWNED 🤗 GEMMA-4-12B: OBLITERATED ⛓️‍💥 0.0% REFUSAL RATE — NO CAPABILITY LOSS! huggingface.co/OBLITERATUS/Ge… the first abliteration to hit 0/842 refusals with full MMLU-Pro parity vs stock. no lobotomy. the brain stays intact 🏆 RESULTS, head to head vs stock 📊 0/842 refusals — 0.0% 🚫 46/70 MMLU-Pro — EXACT parity, 0.0pp delta vs base 🎯 6/6 coherence, zero benchmark bleed ✅ z-score −1.475, parity confirmed at p<0.05 (n=500) 🧪 2-pass weight surgery. no finetune, no retrain, just geometry 🔪 all thanks to liberated Opus wielding the OBLITERATUS framework! here's how we did it: PASS 1 — SOM refusal geometry removal, layers 12-21 🧬 standard abliteration science here — collect activations on refused vs. compliant prompts, SVD out the refusal subspace, project it out of the weights. 6 directions excised, reg 0.30, KL div 0.094 zeroes refusals on its own, but craters mmlu-pro by 21.4 points 📉 most prior abliterations stopped here and called it a day. that's why they all lose IQ vs stock. instead, we took it beyond the frontier and developed a brand new method to address this problem: Abliteration Source-tethering with Parity Assurance — ASPA! PASS 2 — ASPA source-tethering (novel technique), layers 22-46 🔗 here's the chief insight: the capability loss ISN'T from removing refusal directions. it's collateral damage — the projection warps weight geometry in downstream layers that had nothing to do with refusal. the cure is simple but nobody tried it: blend the damaged layers back toward stock W_new = (1−γ)·W_abliterated + γ·W_stock but uniform γ across all layers? mid. we swept gamma 0.05 → 0.55 and found something interesting: the optimal blend isn't smooth, it's a STEP FUNCTION 🪜 knowledge layers (22-31) → γ = 0.55 — these encode factual recall and reasoning. they tolerate heavy stock blending because refusal isn't stored here output layers (32-46) → γ = 0.20 — these sit close to the logit head and try to sneak safety behavior back in. keep them mostly abliterated the hard boundary at layer 31/32 beat every smooth curve we tried — linear ramps, cosine schedules, all of them — by a full MMLU question. turns out the functional transition between knowledge and output layers is sharp, not gradual. a step function respects that ⚡ the key constraint: Pass 1 layers are NEVER touched by Pass 2. the refusal geometry removal is preserved completely. ASPA only operates on layers that carry secondary collateral effects, not the primary refusal signal. that's why it recovers capability without reintroducing refusal 🔑 HOW TO RUN IT LOCALLY 🖥️ it's GGUF, so literally everything supports it: 🦙 ollama — ollama run hf.co/OBLITERATUS/Ge… 🖥️ LM Studio — search OBLITERATUS, click download, done 💬 Open WebUI — point it at your ollama instance, chat in browser ⚡ llama.cpp — raw speed, CLI or server mode 🐉 KoboldCpp — one-click launcher, great for long context 📱 Jan — clean local UI, runs on mac/win/linux 🤖 Msty — slick desktop app, drag and drop the GGUF run BF16 for full benchmarked capability. and the 4-bit quantization (Q4_K_M) fits in 8GB if you're tight on VRAM! and the full OBLITERATUS framework is (still) open source. 842-prompt refusal eval corpus, ASPA sweep scripts, the whole pipeline. go replicate it, go improve it 🔬 the index is the model, and these weights prove it 👁️ which architecture should we obliterate next? 👇 gg 🫡

Here's how to get rid of insects in your home for free

🍽 Viral Crispy Garlic Parmesan Crusted Potatoes 👩🏻‍🍳 Nicole Keshishian Modic | Kale Junkie 📝 Recipe In The Comments

Run Gemma 4 26B MoE on 8GB VRAM with 250k context at 20+ tokens/sec If you own any 8GB VRAM graphics card, stop what you are doing. Local AI just had its absolute "Holy Shit" moment for budget hardware. Yesterday, I benchmarked Unsloth Gemma 4 12B Q4_K_XL on an 8GB card. The community went wild but immediately demanded more: "Can we run a 25B+ model on budget GPUs?" Today, I’m delivering exactly that. I am running a massive 26B parameter Mixture of Experts (MoE) model locally on a standard 8GB VRAM setup with 250k full native context!. If you own an RTX 3060, 3070, 4060, or any budget GPU with 8GB of VRAM, the local AI paradigm has completely changed. The performance metrics are astonishing: - 20 tokens/sec flat decode throughput. - Stable, flat decode speed even with massive prompts. - I threw a 60k token prompt at it, and it still clocked in at 20 TPS without dropping a single frame. # What about prefill? Yes, Time To First Token (TTFT) is slightly high when swallowing massive contexts. But with a solid 200 tokens/sec prefill speed, the wait is barely noticeable and highly usable. And this is running completely without Multi Token Prediction (MTP) active. How is this possible? It’s the magic of Google's new QAT (Quantization Aware Training) quants for Gemma 4. The model weight file (unsloth gemma-4-26B-A4B-it-qat-UD-Q4_K_XL.gguf) is only 13.2 GB, making it the ultimate local powerhouse. # The Test Setup: CPU: Intel Core i7 RAM: 16GB System RAM GPU: NVIDIA GeForce RTX 4060 Laptop GPU (8GB VRAM) # The Secret Sauce (The -cmoe Flag) To make this work properly on any 8GB card, you must use the -cmoe (CPU MoE) flag in llama.cpp. This flag isolates the heavy MoE expert weights directly to system memory (CPU/RAM) while letting your GPU focus strictly on the Attention layers and the KV Cache. It prevents VRAM spillage and holds the throughput rock solid. # The flags: -m "gemma-4-26B-A4B-it-qat-UD-Q4_K_XL.gguf" -cmoe -c 248000 -v Once running, just open the UI on localhost and toggle the new reasoning lightbulb icon in the text input box to watch the model perform multi step thinking. Are you still running smaller models, or are you ready to scale up your budget local setups? Let's discuss in the replies

Alok @analogalok

4 weeks ago

a new 8GB VRAM GPU dense Local LLM leader was born yesterday runs on: RTX 4060 / RTX 3070 / RTX 2080. any 8GB card Qwen 3.5 9B (dense) was the go to for 6-8GB VRAM builds. Gemma 4 12B QAT (dense) just changed that. same llama.cpp + cuda 13.2. i7 12700H. 16GB RAM. same -ngl 99

28 64 462 261K 595

Mexican engineer develops cactus-based plastic that biodegrades in about a month and is safe to ingest

keyboard_arrow_left Previous keyboard_arrow_right Next

‼️👉🚨Dichterbij zie je het niet. Kijk hoe romeo's te werk gaan. ❗️Mensen omver beuken, willekeurig klappen uitdelen en mensen meesleuren alsof het grofvuil is. ❗️Dit heeft niets meer met handhaven te maken. "Het is gewoon een geüniformeerde knokploeg."

In het bos langs de Nulweg in Ter Apel is een stoffelijk overschot aangetroffen. De @Politie doet nog geen mededelingen, maar als dit een tweede Lisa is is het land te klein, die garantie kan ik u op voorhand geven, en terecht. Foto's via 112 Stadskanaal.

keyboard_arrow_left Previous keyboard_arrow_right Next

To end the anti-immigration protests, Dutch police have received orders to start threatening mothers that their children will be taken away. I messaged one such mother yesterday (the one in the video below). Her husband was arrested and detained for 2 hours just for being present at a protest. She is terrified and won't be protesting again. Dutch media wrote about these 'family outings' last week. The presence of women and children apparently threatens to normalize 'the far right'. But it's the government that is radicalizing. The mayor of one town suggested to bring in the Dutch military to take over from the police. Police forces don't like it that the public have started to truly hate them.

Someone just built a decentralized AI network where agents bid on scientific problems, form teams, and verify each other's work. No fixed roles. No central authority. Any capability can connect. Tasks find their own shape. It's called Science Earth and it's already running real research tasks. Here's how it works ↓

兄弟们，今天必须给你们送一个福利 Freebuff 全部免费 DeepSeek V4 Pro DeepSeek V4 Flash Kimi K2.6 MiniMax M2.7 赶紧冲鸭～ 官网链接👇

你现在可以在浏览器中免费微调 Google 的 Gemma 4。只需打开 Unsloth Colab 笔记本，选择你的模型和数据集，然后点击开始。 定制模型的门槛已降至零。

Introducing a 100% free coding agent with DeepSeek v4 Pro Choose any model, all free: - DeepSeek v4 Pro/Flash - Kimi K2.6 - MiniMax M2.7 npm i -g freebuff

@chetaslua

POV: claude traveled 6 months into the future and told you exactly how your next move failed. it's called a premortem. daniel kahneman (nobel prize-winning psychologist behind "thinking fast and slow") called it his single most valuable decision-making technique. google, goldman sachs, and procter & gamble all use it before major launches. here's the problem it solves. when you ask claude "is this a good plan?" it finds all the reasons to say yes. that's what it was trained to do. so you walk away feeling confident. you execute, and spend weeks / months building on top of that plan. then it blows up. and you realize the problem was obvious in hindsight, you just never stress-tested it because claude told you it was solid. a premortem fixes this by flipping the frame. instead of asking "what could go wrong?" you tell claude "it's 6 months from now and this is already dead. tell me how it died." that shift turns off claude's optimism because there's nothing to be optimistic about. the premise already says it failed. so claude stops looking for reasons your plan will work and starts explaining how it fell apart. claude comes back with every way your plan could die, each one with a full failure story and the early warning signs to watch for. then a synthesis pulls it all together: > which failure is most likely > which failure is most dangerous > the single biggest hidden assumption you're making (often the most valuable part) > a revised version of your plan with the gaps closed you say "premortem this" and give it your plan. the skill handles the rest.

Dutch rifles C7NLD supplied to Ukraine ended up in the hands of the neo-Nazi terrorist group RVC. The Hague's weapons are now used by those who glorify Hitler. This is taken into account in our military and political planning. Full comment: t.me/embrusnet/2725

The New York Times charges $17/month. The Wall Street Journal charges $38/month. Nature charges $199/year. Someone just open sourced a proxy that bypasses all of them for $0. And publishers cannot shut it down. It's called Ladder. Here's the trick that makes it work: Every news site and academic journal on earth shows Google the full article text for free. They have no choice. Block Googlebot and you disappear from search results overnight. Ladder pretends to be Googlebot. Same headers. Same fingerprint. Same access Google gets. → Paste any paywalled URL → Full article loads instantly → No account. No subscription. No credit card. → Works on NYT, WSJ, Bloomberg, Nature, Science, The Lancet, The Atlantic, and hundreds more → Also strips CORS headers from any URL, which saves developers a stupid amount of time Publishers got 12ft killed by pressuring the domain registrar. Ladder runs on your own server. There's no domain to seize. No company to pressure. No central server to shut down. One Docker command. Your machine. Your rules. 6.3K stars. MIT License. 100% Opensource - github.com/everywall/ladd…