shadow-cann/hispark-modelzoo-act

ACT

ACT（Action Chunking with Transformers）是面向机器人学习场景的高性能端到端动作控制模型。相比传统模块化机器人控制模型，ACT采用轻量化Transformer架构作为核心骨干进行动作表征学习，结合多模态感知融合模块和时序动作优化网络，在控制精度和实时响应速度上均有显著提升。

Mirror Metadata

• Hugging Face repo: shadow-cann/hispark-modelzoo-act
• Portal model id: ivcifqkd0400
• Created at: 2026-03-03 10:30:33
• Updated at: 2026-03-04 16:06:22
• Category: 多模态

Framework

• PyTorch

Supported OS

• OpenEuler

Computing Power

• Hi3403V100 SVP_NNN

Tags

• 具身智能

Detail Parameters

• 输入: 1 x 6；1 x 3 x 240 x 320；1 x 3 x 240 x 320
• 参数量: 87 M
• 计算量: 8.02 GFLOPs

Files In This Repo

• ACT.zip (源模型 / 源模型下载; 源模型 / 源模型元数据)
• act_distill_fp32_for_mindcmd_simp_release.om (编译模型 / OM 元数据 / a16w8)
• SVP_NNN_PC_V1.0.6.0.tgz (附加资源 / 附加资源)

Upstream Links

• Portal card: https://gitbubble.github.io/hisilicon-developer-portal-mirror/model-detail.html?id=ivcifqkd0400
• Upstream repository: https://gitee.com/HiSpark/modelzoo/blob/master/samples/contribute/ACT/README.md
• License reference: https://github.com/tonyzhaozh/act/blob/main/LICENSE

Notes

• This repository was mirrored from the HiSilicon Developer Portal model card and local downloads captured on 2026-03-27.
• File ownership follows the portal card mapping, not just filename similarity.
• Cover image: 1731868158459906_____.png