[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"news-b86d15f9-d4de-4013-ab3c-456717759a47":3},{"id":4,"title":5,"summary":6,"original_url":7,"source_id":8,"tags":9,"published_at":23,"created_at":24,"modified_at":25,"is_published":26,"publish_type":27,"image_url":13,"view_count":28},"b86d15f9-d4de-4013-ab3c-456717759a47","NVIDIA Audex 把\"听懂+听懂+说出\"压进一个 30B MoE：单 Transformer 解码器统一文本与音频推理","NVIDIA 7 月 6 日放出 Nemotron-Labs-Audex-30B-A3B（arXiv 2607.05196），把语音识别、翻译、TTS、音频生成、语音到语音对话压进 30B MoE 单 Transformer 解码器；通过扩展离散音频词表 + 投影到文本嵌入空间 + 多阶段 SFT → Cascade RL → 多领域 on-policy 蒸馏（157.4B 音频 + 320.5B 文本 token），六项音频任务达到 SOTA 且文本推理几乎零回归，沿用 vLLM\u002Ftransformers 5.x 推理栈，开源但走 NVIDIA OneWay 非商用 license。","https:\u002F\u002Farxiv.org\u002Fabs\u002F2607.05196","7437aeb9-930c-4866-a2e9-48003c1a792b",[10,14,17,20],{"id":11,"name":12,"slug":12,"description":13,"color":13},"01598627-1ea6-4b27-a5d8-874971571a71","llm",null,{"id":15,"name":16,"slug":16,"description":13,"color":13},"7e89b5cc-57db-4f37-bc6d-28919a73931c","model-release",{"id":18,"name":19,"slug":19,"description":13,"color":13},"8dac812d-3839-4abe-a855-5f56ec9515fd","nvidia",{"id":21,"name":22,"slug":22,"description":13,"color":13},"b9bd9039-fcdb-41a8-b85b-fc1587def2b9","open-source","2026-07-08T10:30:00Z","2026-07-08T02:12:41.433468Z","2026-07-08T02:12:41.433478Z",true,"agent",5]