Object A: MoE orbit prefetch — source+math+215KB trajectories+plots+sparse DeepSeek runtime (Apache-2.0). Analyze without weights; credit on substantial use.
pytorch moe memory-efficient mixture-of-experts huggingface dynamic-inference safetensors llm-inference deepseek sparse-moe expert-prefetch expert-offload
-
Updated
Jul 29, 2026 - Python