精选产品76

vLLM 为腾讯混元 Hy4 flash 提供 Day-0 支持

原标题:🎉 Day-0 support for @TencentHunyuan’s Hy4 flash is here in vLLM!

vLLM@vllm_project发布时间 (北京时间)

vLLM 宣布在腾讯混元 Hy4 flash 开源当天提供 Day-0 支持,可在 NVIDIA GPU 上通过 vLLM 运行该模型。Hy4 flash 为 224B 总参数、12B 激活,含 64B Engram,支持 1M token 上下文与多模态理解,采用混合注意力、Latent MoE 和 Engram 架构。

推荐理由vLLM 在模型开源当天给出 Hy4 flash 的部署镜像与配方,读者可据此了解这套混合注意力架构的落地方式。

🎉 Day-0 support for @TencentHunyuan’s Hy4 flash is here in vLLM!

A 1M-token context and multimodal understanding for research, documents, slides and computer use.

224B total parameters, including 64B Engram, with 12B active. Hybrid attention, Latent MoE and Engram put efficiency at the center of the architecture.

Run Hy4 flash on @NVIDIA GPUs with vLLM, with integration work by @inferact. Get started with the release image and deployment recipe:

Docker image: vllm/vllm-openai:hy4-flash
Recipe: https://recipes.vllm.ai/tencent/Hy4-flash

Hy4 flash 现已开放权重,包括 Hy4 flash-FP8 和 Hy4 flash-Base。 224B(12B 激活),1M 上下文,原生多模态。 快速、高效,为日常生产力而打造。 现在它属于你了。敬请期待接下来推出的 Hy4! 🤗:https://huggingface.co/tencent/Hy4-flash 🐙:https://github.com/Tencent-Hunyuan/Hy4-flash 📄:https://hy.tencent.ai/research/hy4-flash 依据 Tencent Hy License 发布。 在此查看完整条款:https://huggingface.co/tencent/Hy4-flash/blob/main/LICENSE
查看原文

Hy4 flash is now open-weight, including Hy4 flash-FP8 and Hy4 flash-Base. 224B (12B active), 1M context, natively multimodal. Fast, efficient and built for everyday productivity. It's yours now. And stay tuned for Hy4 coming up next! 🤗: https://huggingface.co/tencent/Hy4-flash 🐙: https://github.com/Tencent-Hunyuan/Hy4-flash 📄: https://hy.tencent.ai/research/hy4-flash Released under the Tencent Hy License. See full terms here: https://huggingface.co/tencent/Hy4-flash/blob/main/LICENSE

查看原帖
多模态开源生态部署工程产品更新#产品更新#部署/工程#开源生态#多模态