type
status
date
slug
summary
tags
category
icon
password
📢【新聞標題】
Hugging Face says its new robotics model is so efficient it can run on a MacBook
Hugging Face 表示其新型機器人模型非常高效，可以在 MacBook 上運行
📰【摘要】
Hugging Face released an open AI model for robotics called SmolVLA. Trained on community-shared datasets, SmolVLA outperforms larger models in virtual and real-world environments. SmolVLA aims to democratize access to vision-language-action models and accelerate research toward generalist robotic agents. The model is small enough to run on a single consumer GPU or even a MacBook.
Hugging Face 發布了一個名為 SmolVLA 的開放 AI 機器人模型。SmolVLA 在社群共享數據集上進行訓練，在虛擬和現實世界環境中均優於更大的模型。SmolVLA 旨在普及對視覺-語言-動作模型的訪問，並加速對通用機器人代理的研究。該模型足夠小，可以在單個消費級 GPU 甚至 MacBook 上運行。
🗝️【關鍵詞彙表】
📝 democratize (v.)
- 使民主化、普及
- 例句: SmolVLA aims to democratize access to vision-language-action models.
- 翻譯: SmolVLA 旨在普及對視覺-語言-動作模型的訪問。
📝 generalist (adj.)
- 通才的、多面手的
- 例句: accelerate research toward generalist robotic agents.
- 翻譯: 加速對通用機器人代理的研究。
📝 parameters (n.)
- 參數
- 例句: SmolVLA, which is 450 million parameters in size.
- 翻譯: SmolVLA 的大小為 4.5 億個參數。
📝 asynchronous (adj.)
- 異步的
- 例句: SmolVLA also supports an asynchronous inference stack.
- 翻譯: SmolVLA 還支持異步推理堆疊。
📝 inference (n.)
- 推理
- 例句: SmolVLA also supports an asynchronous inference stack.
- 翻譯: SmolVLA 還支持異步推理堆疊。
📝 nascent (adj.)
- 新興的、剛開始的
- 例句: Hugging Face is far from the only player in the nascent open robotics race.
- 翻譯: Hugging Face 遠非新興的開放機器人競賽中唯一的參與者。
📝 outperform (v.)
- 勝過、優於
- 例句: SmolVLA outperforms much larger models for robotics.
- 翻譯: SmolVLA 在機器人技術上勝過更大的模型。
✍️【文法與句型】
📝 It’s becoming a little easier to...
- 說明: Used to express that something is gradually becoming less difficult.
- 翻譯: 用於表達某事正逐漸變得不那麼困難。
- 例句: It’s becoming a little easier to build sophisticated robotics projects at home.
- 翻譯: 在家構建複雜的機器人項目正變得越來越容易。
📝 is far from the only player in...
- 說明: Used to indicate that someone or something is not the only one involved in a particular activity or situation.
- 翻譯: 用於表明某人或某事不是參與特定活動或情況的唯一一方。
- 例句: Hugging Face is far from the only player in the nascent open robotics race.
- 翻譯: Hugging Face 遠非新興的開放機器人競賽中唯一的參與者。
📝 Trained on ..., SmolVLA outperforms...
- 說明: A participial phrase indicating the basis or method by which SmolVLA achieves better performance.
- 翻譯: 一個分詞短語，表示 SmolVLA 實現更好性能的基礎或方法。
- 例句: Trained on “compatibly licensed,” community-shared datasets, SmolVLA outperforms much larger models for robotics.
- 翻譯: 在「兼容許可」的社群共享數據集上進行訓練，SmolVLA 在機器人技術上勝過更大的模型。
📖【全文與翻譯】
It’s becoming a little easier to build sophisticated robotics projects at home.
在家構建複雜的機器人項目正變得越來越容易。
Earlier this week, AI dev platform Hugging Face released an open AI model for robotics called SmolVLA [https://huggingface.co/lerobot/smolvla_base].
本週早些時候，AI 開發平台 Hugging Face 發布了一個名為 SmolVLA 的開放 AI 機器人模型。
Trained on “compatibly licensed,” community-shared datasets, SmolVLA outperforms much larger models for robotics in both virtual and real-world environments, Hugging Face claims.
Hugging Face 聲稱，SmolVLA 在「兼容許可」的社群共享數據集上進行訓練，在虛擬和現實世界環境中均優於更大的機器人模型。
“SmolVLA aims to democratize access to vision-language-action [VLA] models and accelerate research toward generalist robotic agents,” writes Hugging Face in a blog post [https://huggingface.co/blog/smolvla].
Hugging Face 在一篇博文中寫道：「SmolVLA 旨在普及對視覺-語言-動作 [VLA] 模型的訪問，並加速對通用機器人代理的研究。」
“SmolVLA is not only a lightweight yet capable model, but also a method for training and evaluating generalist robotics [technologies].”
「SmolVLA 不僅是一個輕量級但功能強大的模型，而且是一種用於訓練和評估通用機器人 [技術] 的方法。」
SmolVLA is a part of Hugging Face’s rapidly expanding effort to establish an ecosystem of low-cost robotics hardware and software.
SmolVLA 是 Hugging Face 迅速擴張的努力的一部分，旨在建立一個低成本機器人硬體和軟體的生態系統。
Last year, the company launched LeRobot, a collection of robotics-focused models, datasets, and tools.
去年，該公司推出了 LeRobot，這是一系列以機器人為中心的模型、數據集和工具。
More recently, Hugging Face acquired Pollen Robotics[https://techcrunch.com/2025/04/14/hugging-face-buys-a-humanoid-robotics-startup/], a robotics startup based in France, and unveiled several inexpensive [https://techcrunch.com/2025/04/28/hugging-face-releases-a-3d-printed-robotic-arm-starting-at-100/] robotics systems, including humanoids [https://techcrunch.com/2025/05/29/hugging-face-unveils-two-new-humanoid-robots/], for purchase.
最近，Hugging Face 收購了位於法國的機器人新創公司 Pollen Robotics，並推出了幾款廉價的機器人系統，包括可供購買的類人機器人。
SmolVLA, which is 450 million parameters in size, was trained on data from LeRobot Community Datasets, specially marked robotics datasets shared on Hugging Face’s AI development platform.
SmolVLA 的大小為 4.5 億個參數，它是在來自 LeRobot 社群數據集的數據上訓練的，這些數據集是在 Hugging Face 的 AI 開發平台上共享的特別標記的機器人數據集。
Parameters, sometimes referred to as “weights,” are the internal components of a model that guide its behavior.
參數，有時稱為「權重」，是指導模型行為的內部組件。
Hugging Face claims that SmolVLA is small enough to run on a single consumer GPU— or even a MacBook — and can be tested and deployed on “affordable” hardware, including the company’s own robotics systems.
Hugging Face 聲稱，SmolVLA 足夠小，可以在單個消費級 GPU 甚至 MacBook 上運行，並且可以在「價格合理」的硬體上進行測試和部署，包括該公司自己的機器人系統。
In an interesting twist, SmolVLA also supports an “asynchronous inference stack,” which Hugging Face says allows the model to separate the processing of a robot’s actions from the processing of what it sees and hears.
有趣的是，SmolVLA 還支持「異步推理堆疊」，Hugging Face 表示，這允許模型將機器人動作的處理與其所見所聞的處理分開。
As the company explains in its blog post, “[b]ecause of this separation, robots can respond more quickly in fast-changing environments.”
正如該公司在其博文中解釋的那樣，「[由於這種分離，機器人可以在快速變化的環境中更快地做出反應。」
Techcrunch eventSAVE $200+ ON YOUR TECHCRUNCH ALL STAGE PASSBUILD SMARTER. SCALE FASTER. CONNECT DEEPER. JOIN VISIONARIES FROM PRECURSORVENTURES, NEA, INDEX VENTURES, UNDERSCORE VC, AND BEYOND FOR A DAY PACKED WITHSTRATEGIES, WORKSHOPS, AND MEANINGFUL CONNECTIONS.EXHIBIT AT TECHCRUNCH SESSIONS: AISECURE YOUR SPOT AT TC SESSIONS: AI AND SHOW 1,200+ DECISION-MAKERS WHAT YOU’VEBUILT — WITHOUT THE BIG SPEND. AVAILABLE THROUGH MAY 9 OR WHILE TABLES LAST.Boston, MA | July 15REGISTER NOW[https://techcrunch.com/events/techcrunch-all-stage-2025/?utm_source=tc&utm_medium=ad&utm_campaign=tcallstage2025&utm_content=ticketsales&promo=rb_inline&display=]
Techcrunch 活動節省 200 美元以上的 TECHCRUNCH ALL STAGE PASSBUILD SMARTER。SCALE FASTER。CONNECT DEEPER。與來自 PRECURSORVENTURES、NEA、INDEX VENTURES、UNDERSCORE VC 及其他公司的遠見者一起參加為期一天的活動，其中包含策略、研討會和有意義的聯繫。在 TECHCRUNCH SESSIONS：AISECURE 您的位置 AT TC SESSIONS：AI 並向 1,200 多名決策者展示您所建立的內容 — 無需大筆支出。5 月 9 日前提供或在表格用完之前提供。馬薩諸塞州波士頓 | 7 月 15 日立即註冊 [連結已移除]
SmolVLA is available for download from Hugging Face.
SmolVLA 可從 Hugging Face 下載。
Already, a user on X claims to have used the model to control a third-party robotic arm:
已經有 X 平台上的用戶聲稱使用該模型來控制第三方機器人手臂：
> 🚀 SmolVLA — feels like a BERT moment for robotics 🤖
I tried it on the Koch Arm: Inference on RTX 2050 (4GB), fine-tuned with just 31 demos, and matches/outperforms single-task baselines 🔥 Big thanks to @RemiCadene [https://twitter.com/RemiCadene?ref_src=twsrc%5Etfw] @danaubakirova @mustash97 @francesco__capu 🙌 pic.twitter.com/TiBkAZGwkM [https://t.co/TiBkAZGwkM] — Xingdong Zuo (@XingdongZ) June 4, 2025 [https://twitter.com/XingdongZ/status/1930168794099659147?ref_src=twsrc%5Etfw]** 🚀 SmolVLA — 感覺像是機器人技術的 BERT 時刻 🤖 我在 Koch Arm 上試了一下： 在 RTX 2050 (4GB) 上進行推理，僅用 31 個演示進行微調，並且 與單任務基準相符/優於單任務基準 🔥 非常感謝 @RemiCadene [連結已移除] @danaubakirova @mustash97 @francesco__capu 🙌 pic.twitter.com/TiBkAZGwkM [連結已移除] — Xingdong Zuo (@XingdongZ) 2025 年 6 月 4 日 [連結已移除]
It’s worth noting that Hugging Face is far from the only player in the nascent open robotics race.
值得注意的是，Hugging Face 遠非新興的開放機器人競賽中唯一的參與者。
Nvidia has a collection of tools for open robotics, and startup K-Scale Labs is building the components for what it’s calling “open-source humanoids.”
Nvidia 擁有一系列用於開放機器人的工具，新創公司 K-Scale Labs 正在構建其所謂的「開源類人機器人」的組件。
Other formidable firms in the segment include Dyna Robotics, Jeff Bezos-backed Physical Intelligence, and RLWRLD[https://techcrunch.com/2025/04/14/rlwrld-raises-14-4m-to-build-foundation-model-for-robotics/].
該領域的其他強大公司包括 Dyna Robotics、Jeff Bezos 支持的 Physical Intelligence 和 RLWRLD。
🔗【資料來源】
文章連結：https://techcrunch.com/2025/06/04/hugging-face-says-its-new-robotics-model-is-so-efficient-it-can-run-on-a-macbook/