🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

I stumbled upon a highly underrated paper It studies the dimensionality self-attention geometry via algebraic methods, focusing on a simplified version without softmax but also a short section with it at the end, they have some conjectures if anyone will can prove it

you can just do things

Distributed intelligence > super intelligence!

Reject MCP, embrace SLOP. Simple Language Open Protocol

float i,e,R,s;vec3 q,p,d=vec3(FC.xy/r-vec2(.6,.5),.7);for(q.zx--;i++<99.;){o.rgb+=hsv(.1,.2,min(e*s,.7-e)/35.);s=1.;p=q+=d*e*R*.1;p=vec3(log2(R=length(p))-t,exp(1.-p.z/R),atan(p.y,p.x)+cos(t)*.2);for(e=--p.y;s<3e2;s+=s)e+=sin(dot(sin(p.zxy*s)-.5,1.-cos(p.yxz*s)))/s;}#つぶやきGLSL

RWKV7-G1 "GooseOne" first release: reasoning @ 0.1b params, pure RNN (attention-free), fully multilingual. Demo & weights on RWKV.com 🪿 Larger G1 training in progress.

Qwen2.5-Math-7B-Instruct can scale to o1 level accuracy in only 32 rollouts. This paper's methods has a 4–16x better scaling rate over our deterministic search counterparts. Current inference-time scaling often relies on imperfect reward models that cause “reward hacking.”

Also, does this make Gemma3 27B to be the best non-reasoning LLM????

At Ledger, you might know that we have the Ledger Donjon, our dedicated team constantly conducting open security research. We recently worked with Trezor, revealing that their Trezor Safe 3 was susceptible to physical supply chain attacks. Here's a thread on our findings:🧵

Wow ⚡️ Google back at it

The open source Conversational Speech Model of Sesame is out. Already trendy as #2 on Hugging Face within less than 24 hours huggingface.co/models

🚀We just dropped SmolDocling: a 256M open-source vision LM for complete document OCR!📄✨ It's lightning fast, process a page in 0.35 sec on consumer GPU using < 500MB VRAM⚡ SOTA in document conversion, beating every competing model we tested up to 27x larger🤯 But how? 🧶⬇️

Another week, another cool release 🥰

Nvidia dropping a 2B open foundation model for humanoid robots on the hub!🔥

⚡ AutoQuant I updated AutoQuant to make the GGUF versions of Gemma 3 abliterated. It implements imatrix and can split the model into multiple files. The GGUF code is based on gguf-my-repo, maintained by Xuan-Son Nguyen and Vaibhav (VB) Srivastav It also supports GPTQ, ExLlamaV2, AWQ, and HQQ!

Quantum Gravity remains unsolved because only recently physics account for perception ⚡️

float i,e,R,s;vec3 q,p,d=vec3(FC.xy/r,1);for(q.zy--;i++<229.;){e+=i/6e4;o.rgb+=hsv(q.z-.75,R,e*R*i/30.);s=2.;p=q+=d*e*R*R*.15;p=vec3(log(R=length(p-.2))-t*.25,exp(mod(-p.z,s)/R),p+sin(t)*.07);for(e=--p.y;s<4e3;s+=s)e+=-abs(dot(sin(p*s),max(p,exp(atan(.4,e))))/s*.1);}#つぶやきGLSL

Light in the newly discovered “solid” form.

Llama 4 Omni is releasing very soon 👀 my system has detected a new official page for the upcoming models