Hugging Face's logo
Dan Santos's picture

Dan Santos

DanVance
5 4 18
Ā·

AI & ML interests

None yet

Recent Activity

liked a model 6 days ago
microsoft/Mage-Flow-Turbo
liked a model 6 days ago
microsoft/Mage-Flow
repliedto SeaWolf-AI's post 14 days ago
šŸ”µ VKUE ā€” No GPU? Runs anyway. "Frontier models need a datacenter GPU" rests on a hidden assumption: that the model reads ALL its parameters every token. Decode is memory-bandwidth bound ā€” sweep 34B params/token and an 8 GB card dies at 1ā€“2 tok/s. So we ran ONE 34.7B reasoning model ā€” Ourbox-35B-JGOS, a sparse Mixture-of-Experts ā€” as the identical weights across the whole hardware spectrum. All measured: ā€¢ B200: 18,057 tok/s (aggregate) ā€¢ 1Ɨ A10G: 126 tok/s ā€¢ 8 GB laptop (RTX 5060): 20 tok/s ā€¢ GPU-less CPU: 17 tok/s Why it works: Ourbox holds 34.7B params but only ~3B are active per token (256 experts, top-8). Since decode is bandwidth-bound, a dense 34B moves ~16.7 GB/token while Ourbox moves ~1.45 GB ā€” ~11Ɨ less traffic. Put the experts in system RAM, keep attention/router/shared on the GPU, and a 34.7B reasoner runs on an 8 GB laptop ā€” or no GPU at all. Sparsity alone, proven (same laptop, same quant, ~same footprint): Ourbox-35B (A3B) 20.01 tok/s vs Qwen2.5-32B (dense) 5.36 ā†’ 3.7Ɨ from sparsity alone, ~2Ɨ the best dense-32B on any 8 GB machine. Not a toy: GPQA Diamond 86.4% (maj@8). Try it live (same prompt, GPU vs GPU-less CPU, live tok/s). Honest scope: one machine's measurements; the CPU path proves it RUNS without a GPU, not that it beats one. šŸ“ Article: https://huggingface.co/blog/FINAL-Bench/vkue šŸ”µ GPU vs CPU demo: https://final-bench-ourbox-35b-vkue-demo.hf.space/ šŸ”µ CPU-only demo: https://final-bench-ourbox-35b-vkue-cpu.hf.space šŸ“Š VKUE leaderboard: https://huggingface.co/spaces/FINAL-Bench/VKUE šŸ¤— Model: https://huggingface.co/FINAL-Bench/Ourbox-35B-JGOS-GGUF āš” VKAE (speed): https://huggingface.co/spaces/VIDraft/vkae VKUE is the "runs anywhere" side of our serving line; VKAE the "fast on datacenter GPUs" side. VKAE is fast; VKUE is everywhere.
View all activity

Organizations

None yet