BF16 models losslessly compressed to 12-bit: 99.7% of weights share exponent bits
AlpinDale · x · 2026-11-07
AlpinDale demonstrated lossless compression of BF16 models down to 12 bits, based on NyoomFloat12 from Together Compute. Over 99.7% of weights share the same 0111 upper exponent bits, so those bits can be omitted for most values and restored at decode time like classic file compression. Since no hardware natively supports this dtype, significant work went into making inference efficient.
Related event: Developer open-sources nf12: lossless 12-bit compression for BF16 models(10 posts)→
More from Infra
- US AI scale now hinges on coordinating power, components, agentic workflows and humanoid pilots — ingliguori · 2026-11-07
- Fudan's D3D diffusion drafter enables async speculative decoding, up to 7.48x speedup on Qwen3 — pmttyji · 2026-11-07
- Study: data centers have only marginal impact on residential electricity prices — TinfoilTricorn · 2026-11-07
- YouJiacheng: post-training needs high tps for long-tail rollouts — YouJiacheng · 2026-11-07
- vLLM contributor: lossless KV cache quantization viable but not worth it — AlpinDale · 2026-11-07
- NVIDIA's first co-packaged optics Ethernet switch, Spectrum-X Photonics, enters full production — TinfoilTricorn · 2026-11-07