Formulas and Code for Adapting Inference to GDN/KDA Linear Modules
Click the blue text to follow us When adapting hybrid models on the inference framework (vLLM/SGLang), the key is to construct the corresponding linear modules. The linear module used by the QwenNext model is GDN (Gated Delta Networks), while KimiLinear uses KDA (Kimi Delta Attention). The basic structure of both modules is similar, both being … Read more