vllm.models.mistral.mistral_large4.amd
¶
Modules:
-
dspark–Mistral Large 4 DSpark draft model (dense MLA backbone + Markov head).
-
model–Mistral Large 4: MLA + DSA sparse attention, MoE, NoPE layers, KV sharing.
-
model_state–Mistral Large 4 model-runner state: builds the specialized decode after load.
-
specialized_decode–