AIdevsmartdata/chimere

Rust-native MoE inference runtime with custom CUDA kernels for Blackwell GPUs. Includes DFlash speculative decoding, multi-tier Engram memory, and entropy-adaptive routing. Targets Qwen3.5-35B-A3B on a single RTX 5060 Ti 16GB.

语言：Rust · GitHub 星标：9 · Fork：1

话题：blackwell · cuda · deltanet · engram · ffi · hybrid-ssm · inference · llm

在 GitHub 上查看原仓库