llama.cpp · b11861
minimax-m3 : forward quant scales to the graph (fixes #31063) (#31097) The generic scale pass in llama-model.cpp loads the per-expert NVFP4 scale2 tensors (ffn_{up,gate,down}_exps_s) and the per-tensor MXFP8 scales for the dense FFN and the shared expert, but minimax-m3.cpp never forwarded any of them to the graph. build_moe_ffn() defaults these to nullptr, so build_lora_mm_id() skipped its w_s branch entirely and the routed-expert weights came out global_scale times too large, swamping the residual (reported G ~= 15123 for M3.1 Flash) and producing token soup. Forward all scale tensors at the three call sites, matching the pattern used by flint and the other NVFP4-capable MoE models. Co-authored-by: origin-labs-ai **Website:** - **Attestations:** - **macOS/iOS:** - [macOS Apple Silicon (a
推荐理由：按信源设置仅收录，不参与自动精选