DeepSpec
★ 7.2KDeepSpec is a full-stack Python codebase for training and evaluating speculative-decoding draft models (DSpark, DFlash, Eagle3) against targets like Qwen3 and Gemma. It covers data prep, 8-GPU training, and benchmark evaluation, plus released checkpoints.
AI Frameworks | Python · speculative-decoding · llm-inference
View Project →