Models

Perceptron AI 发布 36B 参数机器人模型预训练检查点

Why we picked it

Perceptron AI 把 36B 的机器人模型做成了 NLP 圈那种预训练检查点玩法,机械臂用户下载对应权重就能用文本指令操控,还能继续微调。有实体机械臂的开发者可以直接去 Hugging Face 试试,省掉从零训练的成本。

Perceptron AI 在 Hugging Face 上发布了 36B 参数的机器人模型,提供多个可直接运行的检查点版本。

Body

每个检查点已针对特定的机械臂做过微调,用户下载与自己设备匹配的版本后,机械臂即可执行纯文本指令。用户还可以在这些检查点基础上继续微调,适配自己的任务场景。

这种做法类似于语言模型领域的预训练检查点生态:下游用户不用从头训练,而是基于现成权重做少量适配。发布渠道为 Hugging Face,覆盖已支持的多款机械臂。

Original · rohanpaul_ai

Robotics is getting the pretrained-checkpoint economy that made language models move so fast.

Perceptron AI just released ready-to-run versions of its 36B-parameter robot model on Hugging Face, each already fine-tuned for a specific robot arm.

Owners of a supported arm download the matching checkpoint, and the arm acts on plain-text instructions, with the option to fine-tune it further for their own tasks.

Industry Coordinates & Source Verification

This report is categorized under AITOP's "Models" desk, focusing on US frontier AI labs and Silicon Valley technology leaders. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows and US-China foundation model benchmarks and commercialization comparison. Key entities and technical topics under continuous tracking include Perceptron AI, Hugging Face, robots, open-source models.

Source attribution: curated from rohanpaul_ai (@rohanpaul_ai).