AI is happening. The future is updating.Qwen3.6-35B-A3B-PrismaQuant-4.75bit-vllm
Mixed-precision quantization of `Qwen/Qwen3.6-35B-A3B` produced by **PrismaQuant** — a per-Linear sensitivity-driven allocator that chooses each Linear module's format individually under a total-bit budget. **Why "every layer refracts into a different format":** a naive uniform NVFP4 either leaves disk on the table (keeping everything BF16 "to be safe") or loses quality (quantizing sensitive layers to 4-bit). PrismaQuant measures the actual Fisher-weighted MSE for every (Linear, format) pair and runs a multi-choice knapsack under a total-bit budget, so every bit lives where it buys the most likelihood. This checkpoint uses **three precisions**, selected per-Linear by the allocator from measured sensitivity — not chosen uniformly:
Mixed-precision quantization of `Qwen/Qwen3.6-35B-A3B` produced by **PrismaQuant** — a per-Linear sensitivity-driven allocator that chooses each Linear module's format individually under a total-bit budget. **Why "every layer refracts into a different format":** a naive uniform NVFP4 either leaves disk on the table (keeping everything BF16 "to be safe") or loses quality (quantizing sensitive layers to 4-bit). PrismaQuant measures the actual Fisher-weighted MSE for every (Linear, format) pair and runs a multi-choice knapsack under a total-bit budget, so every bit lives where it buys the most likelihood. This checkpoint uses **three precisions**, selected per-Linear by the allocator from measured sensitivity — not chosen uniformly: