腾讯混元开源了AngelSpec，一个能帮你的模型推理提速近2倍的投机解码框架，实测比DFlash还快10%以上，代码和权重都直接可用了。
腾讯混元开源了AngelSpec，一个端到端的投机解码框架，同时支持训练和部署。在Hy3-A21B模型上，DFly方法在并发数4到64范围内，相比自回归解码实现1.98到2.40倍的端到端加速。与DFlash相比，DFly的吞吐量高出10.5%到11.8%。项目提供了训练代码、Hy3-A21B MTP/DFly drafter权重，以及论文和文档。
🚀 We’ve open-sourced AngelSpec, an end-to-end speculative decoding framework supporting both train...
🚀 We’ve open-sourced AngelSpec, an end-to-end speculative decoding framework supporting both training and deployment. On Hy3-A21B, DFly delivers a 1.98–2.40× end-to-end speedup over autoregressive decoding across tested concurrency levels from 4 to 64, with 10.5–11.8% higher throughput than DFlash. Training code and Hy3-A21B MTP/DFly drafter weights are now available: GitHub: github.com/Tencent/AngelS… Y Paper: arxiv.org/abs/2607.25852 v Docs: angelspec.readthedocs.io R Hugging Face: huggingface.co/collections/An… C ModelScope: modelscope.cn/collections/An… Y #Hy3 3 #AngelSpec c #OpenSource e 💬 0 🔄 1 ❤️ 6 👀 288 📊 2 ⚡