Baseten 工程博客：pyannote 说话人分离模型提速 9.6 倍
Blog
Baseten 让 pyannote 的说话人分离（diarization）模型推理速度提升 9.6 倍。该博客还介绍了面向托管 RL rollout 的 delta 权重同步优化、Baseten 在 Coval 语音 AI 基准测试中领先，以及供编码智能体使用 Baseten 的新 MCP 与技能。
正文
Blog Inference engineering Model performance How Baseten makes pyannote’s diarization models 9.6x faster Matte Lim 1 other Model performance Optimizing delta weight syncs for managed RL rollouts Paras Stefanopoulos Model performance Baseten leads Coval’s voice AI benchmark Ansel Erol 1 other AI engineering New MCP and skill for coding agents to use Baseten Marius Killinger 1 other AI models Best open-source models for post-training Chloe Florit Model performance The efficient frontier of LLM inference Philip Kiely Model performance Agentic kernels in production Brian Li 2 others AI models GLM-5.3: Scaling with post-training, intuitively explained Chloe Florit 1 other AI engineering How to run any open model inside DeepSeek Harness Alex Ker 1 2 3 ... 27 requestAnimationFrame(function(){$RT=performance.now()}); Popular models