Загрузка видео...

Не удалось загрузить видео

На главную

Microsoft's new Florence 2 is big for Computer Vision. It's a merge between Text and Vision. With a single prompt you can instruct the model to do CV tasks like captioning, object detection, grounding, and segmentation. The best part, it only uses a single backbone to handle everything. ▸...

LiorOnAI's profile picture

Lior Alexander

116,681 subscribers

186,579 просмотров • 2 лет назад •via X (Twitter)

Наука и технологии
LIVE

Anya Rossi• Live Now

Private livecam show

Комментарии: 8

Фото профиля AlphaSignal AI
AlphaSignal AI2 лет назад

@AlphaSignalAI One step closer to AGI..

Фото профиля Dash
Dash2 лет назад

@AlphaSignalAI Holy shit

Фото профиля Lior⚡
Lior⚡2 лет назад

@Mrosenmer Can't wait for the repo 👀

Фото профиля Ariyan
Ariyan2 лет назад

@AlphaSignalAI @jxnlco @skalskip92 you've seen this?

Фото профиля zachary austin
zachary austin2 лет назад

@AlphaSignalAI Look away NSA

Фото профиля ThisAndThat
ThisAndThat2 лет назад

@AlphaSignalAI Well at least the demo is not much different than YOLOv8 or similar. We have been combining a few models to achieve what you have described. If this model can do all that together and with even better performance then great. But I don't trust Microsoft. They suck.

Фото профиля Waseem
Waseem2 лет назад

@AlphaSignalAI I've attempted to do something like this with images and GPT-4V. Results have been pretty good but working on improving it. Plan to put something like this on a robot with a raspberry pi.

Фото профиля alejandro cartagena
alejandro cartagena2 лет назад

@AlphaSignalAI Look @elmanmansimov

Похожие видео

Florence-2, the new vision foundation model by Microsoft, can now run 100% locally in your browser on WebGPU, thanks to Transformers.js! 🤗🤯 It supports tasks like image captioning, optical character recognition, object detection, and many more! 😍 WOW! Demo (+ source code) 👇
1:05

Florence-2, the new vision foundation model by Microsoft, can now run 100% locally in your browser on WebGPU, thanks to Transformers.js! 🤗🤯 It supports tasks like image captioning, optical character recognition, object detection, and many more! 😍 WOW! Demo (+ source code) 👇

Xenova

88,762 просмотров • 2 лет назад

Big News! Meta just released Segment Anything, a new AI model that can "cut out" any object, in any image/video, with a single click. The model is designed and trained to be promptable, so it can transfer zero-shot to new image distributions and tasks.
0:22

Big News! Meta just released Segment Anything, a new AI model that can "cut out" any object, in any image/video, with a single click. The model is designed and trained to be promptable, so it can transfer zero-shot to new image distributions and tasks.

Lior Alexander

290,330 просмотров • 3 лет назад

LLMs are great for human in the loop applications, but fail at deterministic developer tasks. Interfaze (YC P26) is a new AI model that outperforms general LLMs on high accuracy tasks like: OCR, Object Detection, Web scraping, Speech-to-text, Classification and more. Congrats on the launch, Yoeven and Harsha!
1:21

LLMs are great for human in the loop applications, but fail at deterministic developer tasks. Interfaze (YC P26) is a new AI model that outperforms general LLMs on high accuracy tasks like: OCR, Object Detection, Web scraping, Speech-to-text, Classification and more. Congrats on the launch, Yoeven and Harsha!

Y Combinator

71,358 просмотров • 3 месяцев назад

Our vision is for AI that uses world models to adapt in new and dynamic environments and efficiently learn new skills. We’re sharing V-JEPA 2, a new world model with state-of-the-art performance in visual understanding and prediction. V-JEPA 2 is a 1.2 billion-parameter model, trained on video, that can enable zero-shot planning in robots—allowing them to plan and execute tasks in unfamiliar environments. Learn more about V-JEPA 2 ➡️ As we continue working toward our goal of achieving advanced machine intelligence (AMI), we’re also releasing three new benchmarks for evaluating how well existing models can reason about the physical world from video. Learn more and download the new benchmarks ➡️
2:03

Our vision is for AI that uses world models to adapt in new and dynamic environments and efficiently learn new skills. We’re sharing V-JEPA 2, a new world model with state-of-the-art performance in visual understanding and prediction. V-JEPA 2 is a 1.2 billion-parameter model, trained on video, that can enable zero-shot planning in robots—allowing them to plan and execute tasks in unfamiliar environments. Learn more about V-JEPA 2 ➡️ As we continue working toward our goal of achieving advanced machine intelligence (AMI), we’re also releasing three new benchmarks for evaluating how well existing models can reason about the physical world from video. Learn more and download the new benchmarks ➡️

AI at Meta

310,340 просмотров • 1 год назад

SAM 2 from Meta FAIR is the first unified model for real-time, promptable object segmentation in images & videos. Using the model in our web-based demo you can segment, track and apply effects to objects in video in just a few clicks. Try SAM 2 ➡️
0:40

SAM 2 from Meta FAIR is the first unified model for real-time, promptable object segmentation in images & videos. Using the model in our web-based demo you can segment, track and apply effects to objects in video in just a few clicks. Try SAM 2 ➡️

AI at Meta

88,918 просмотров • 2 лет назад

The teams shipping AI agents right now are bleeding money on the dumbest possible expense: teaching a 400B-parameter model to read a file name. Every time an AI agent needs to "see" something today, it routes an image through a frontier model. OCR, object detection, checking if a button exists on screen. You're paying GPT-4o or Claude pricing for tasks that require perception, not reasoning. One agent workflow processing a few thousand screenshots per day can burn through more on vision calls than on the actual thinking. Perceptron's Isaac is 2B parameters. Built by the team that created Meta's Chameleon multimodal models. On perceptive benchmarks, it matches or beats models 50x its size. The VQA, OCR, and object detection scores are competitive with models running on infrastructure that costs orders of magnitude more. The MCP wrapper is the distribution play. One install command and every Claude Code agent can offload vision tasks to a model that runs on a single consumer GPU. The agent keeps its reasoning in the frontier model and routes perception to a specialist. That split is how you get vision-heavy agent workflows from "technically possible but expensive" to "cheap enough to run on everything." This is the same pattern that won in every other compute-intensive stack. General-purpose handles orchestration. Specialists handle the heavy lifting. Graphics went through it. Audio went through it. Video encoding went through it. Vision in AI agents is next. The teams building agents that see 10,000 images a day will care about this before anyone else does.
0:38

The teams shipping AI agents right now are bleeding money on the dumbest possible expense: teaching a 400B-parameter model to read a file name. Every time an AI agent needs to "see" something today, it routes an image through a frontier model. OCR, object detection, checking if a button exists on screen. You're paying GPT-4o or Claude pricing for tasks that require perception, not reasoning. One agent workflow processing a few thousand screenshots per day can burn through more on vision calls than on the actual thinking. Perceptron's Isaac is 2B parameters. Built by the team that created Meta's Chameleon multimodal models. On perceptive benchmarks, it matches or beats models 50x its size. The VQA, OCR, and object detection scores are competitive with models running on infrastructure that costs orders of magnitude more. The MCP wrapper is the distribution play. One install command and every Claude Code agent can offload vision tasks to a model that runs on a single consumer GPU. The agent keeps its reasoning in the frontier model and routes perception to a specialist. That split is how you get vision-heavy agent workflows from "technically possible but expensive" to "cheap enough to run on everything." This is the same pattern that won in every other compute-intensive stack. General-purpose handles orchestration. Specialists handle the heavy lifting. Graphics went through it. Audio went through it. Video encoding went through it. Vision in AI agents is next. The teams building agents that see 10,000 images a day will care about this before anyone else does.

Aakash Gupta

55,978 просмотров • 4 месяцев назад

Sam Altman says the goal is to run a GPT-5 quality model locally on a single device We don't know how to do it yet, but the recent efficiency gains were also once thought impossible The vision is a new computer form factor for the AI era: "it should run locally... if you want it to"
0:47

Sam Altman says the goal is to run a GPT-5 quality model locally on a single device We don't know how to do it yet, but the recent efficiency gains were also once thought impossible The vision is a new computer form factor for the AI era: "it should run locally... if you want it to"

Haider.

197,384 просмотров • 8 месяцев назад

MolmoAct2 runs zero-shot on an SO-ARM101, no training required 🦾 Ai2 open Action Reasoning Model is now in LeRobot with the full lifecycle covered: fine-tuning (full or LoRA), evaluation, and real-robot deployment. It pairs a Molmo2-ER vision-language backbone with a flow-matching action expert to turn images, language, and proprioceptive state into action chunks. The best part: ready-made checkpoints ship with calibration correction baked in, so you can point it at your SO-100/101 and just watch it work. Inference fits in ~12GB at bf16, and LoRA fine-tuning runs on a single 24GB GPU. Big thanks to the Ai2 team for building this in the open.
0:30

MolmoAct2 runs zero-shot on an SO-ARM101, no training required 🦾 Ai2 open Action Reasoning Model is now in LeRobot with the full lifecycle covered: fine-tuning (full or LoRA), evaluation, and real-robot deployment. It pairs a Molmo2-ER vision-language backbone with a flow-matching action expert to turn images, language, and proprioceptive state into action chunks. The best part: ready-made checkpoints ship with calibration correction baked in, so you can point it at your SO-100/101 and just watch it work. Inference fits in ~12GB at bf16, and LoRA fine-tuning runs on a single 24GB GPU. Big thanks to the Ai2 team for building this in the open.

LeRobot

22,200 просмотров • 1 месяц назад

1/N Most Vision-Language-Action models need tons of data for finetuning, and still fail for new objects and instructions. Introducing OTTER, a lightweight, easy-to-train model that uses text-aware visual features to nail unseen tasks out of the box! Here's how it works 👇
0:48

1/N Most Vision-Language-Action models need tons of data for finetuning, and still fail for new objects and instructions. Introducing OTTER, a lightweight, easy-to-train model that uses text-aware visual features to nail unseen tasks out of the box! Here's how it works 👇

Fangchen Liu

68,366 просмотров • 1 год назад

Today we're excited to introduce Vy, our AI that sees and acts on your computer. At Vercept, our mission is to reinvent how humans use computers–enabling you to accomplish orders of magnitude more than what you can do today. Vy is a first glimpse at AI that sees and uses your computer just like you do. It runs natively on your computer, with access to all your applications and accounts that you’re already signed into. It is available for download starting today! As part of the release, we’ve built VyUI, an AI model bridging the gap between language and your screen. It sets a new state-of-the-art on UI grounding benchmarks like ScreenSpot v1, ScreenSpot v2, and GroundUI Web–beating leading models from OpenAI, Google, and Anthropic. Check out what Vy can do in the thread below: 🧵👇
1:41

Today we're excited to introduce Vy, our AI that sees and acts on your computer. At Vercept, our mission is to reinvent how humans use computers–enabling you to accomplish orders of magnitude more than what you can do today. Vy is a first glimpse at AI that sees and uses your computer just like you do. It runs natively on your computer, with access to all your applications and accounts that you’re already signed into. It is available for download starting today! As part of the release, we’ve built VyUI, an AI model bridging the gap between language and your screen. It sets a new state-of-the-art on UI grounding benchmarks like ScreenSpot v1, ScreenSpot v2, and GroundUI Web–beating leading models from OpenAI, Google, and Anthropic. Check out what Vy can do in the thread below: 🧵👇

Vercept

123,748 просмотров • 1 год назад

Woow Nvidia has just released a 2.6B open-source world model 🔥 You can turn a single image, text prompt and trajectory into controllable worlds... And on a single GPU! - Code available on GitHub - Paper as well on arxiv You can use it for many things like embodied AI and robotics research, simulations, etc. Because it can run on a single GPU (like an RTX 5090 or H100) it makes world models accessible to basically everyone!
1:14

Woow Nvidia has just released a 2.6B open-source world model 🔥 You can turn a single image, text prompt and trajectory into controllable worlds... And on a single GPU! - Code available on GitHub - Paper as well on arxiv You can use it for many things like embodied AI and robotics research, simulations, etc. Because it can run on a single GPU (like an RTX 5090 or H100) it makes world models accessible to basically everyone!

Paul Couvert

174,008 просмотров • 3 месяцев назад

LP-MusicCaps: LLM-Based Pseudo Music Captioning paper page: Automatic music captioning, which generates natural language descriptions for given music tracks, holds significant potential for enhancing the understanding and organization of large volumes of musical data. Despite its importance, researchers face challenges due to the costly and time-consuming collection process of existing music-language datasets, which are limited in size. To address this data scarcity issue, we propose the use of large language models (LLMs) to artificially generate the description sentences from large-scale tag datasets. This results in approximately 2.2M captions paired with 0.5M audio clips. We term it Large Language Model based Pseudo music caption dataset, shortly, LP-MusicCaps. We conduct a systemic evaluation of the large-scale music captioning dataset with various quantitative evaluation metrics used in the field of natural language processing as well as human evaluation. In addition, we trained a transformer-based music captioning model with the dataset and evaluated it under zero-shot and transfer-learning settings. The results demonstrate that our proposed approach outperforms the supervised baseline model.
1:11

LP-MusicCaps: LLM-Based Pseudo Music Captioning paper page: Automatic music captioning, which generates natural language descriptions for given music tracks, holds significant potential for enhancing the understanding and organization of large volumes of musical data. Despite its importance, researchers face challenges due to the costly and time-consuming collection process of existing music-language datasets, which are limited in size. To address this data scarcity issue, we propose the use of large language models (LLMs) to artificially generate the description sentences from large-scale tag datasets. This results in approximately 2.2M captions paired with 0.5M audio clips. We term it Large Language Model based Pseudo music caption dataset, shortly, LP-MusicCaps. We conduct a systemic evaluation of the large-scale music captioning dataset with various quantitative evaluation metrics used in the field of natural language processing as well as human evaluation. In addition, we trained a transformer-based music captioning model with the dataset and evaluated it under zero-shot and transfer-learning settings. The results demonstrate that our proposed approach outperforms the supervised baseline model.

AK

78,794 просмотров • 3 лет назад

QVAC SDK 0.13.0 is live, and this version brings a lot of exciting updates! Local AI now plugs into your coding agent, ships as a desktop app in one command, and runs even more models. Highlights: NEW INTEGRATIONS - OpenCode and coding agents: the new @qvac/ai-sdk-provider makes QVAC a local provider. Less setup, same-model requests queue cleanly, and managed mode starts and supervises qvac serve for you. - Broader OpenAI-compatible API, validated across supported flows so covered capabilities stay consistent and testable. - Turn your QVAC project into a real desktop app for Mac, Windows, or Linux with a single command. The new Electron plugin handles the packaging and keeps the app small by including only what it needs. NEW MODELS - New pi0.5 model support - run a vision-language "robot brain" on a single ordinary graphics card, at full accuracy. - Image-to-video, fully local, via the Wan2.1 model in the Diffusion addon. - New BCI add-on: brain-computer interface transcription, fully local. Decode recorded neural signals into text on-device via the Whisper.cpp-based BCI model. IMPROVEMENTS - Whisper GPU transcription on Android, auto-picking the best backend (OpenCL on Adreno 700+, Vulkan elsewhere), unified on one ggml engine. - Parakeet steadier on mobile, with real end-of-utterance detection for streaming. - Supertonic TTS now runs full GPU across Metal, Vulkan, and OpenCL, with native streaming.
1:39

QVAC SDK 0.13.0 is live, and this version brings a lot of exciting updates! Local AI now plugs into your coding agent, ships as a desktop app in one command, and runs even more models. Highlights: NEW INTEGRATIONS - OpenCode and coding agents: the new @qvac/ai-sdk-provider makes QVAC a local provider. Less setup, same-model requests queue cleanly, and managed mode starts and supervises qvac serve for you. - Broader OpenAI-compatible API, validated across supported flows so covered capabilities stay consistent and testable. - Turn your QVAC project into a real desktop app for Mac, Windows, or Linux with a single command. The new Electron plugin handles the packaging and keeps the app small by including only what it needs. NEW MODELS - New pi0.5 model support - run a vision-language "robot brain" on a single ordinary graphics card, at full accuracy. - Image-to-video, fully local, via the Wan2.1 model in the Diffusion addon. - New BCI add-on: brain-computer interface transcription, fully local. Decode recorded neural signals into text on-device via the Whisper.cpp-based BCI model. IMPROVEMENTS - Whisper GPU transcription on Android, auto-picking the best backend (OpenCL on Adreno 700+, Vulkan elsewhere), unified on one ggml engine. - Parakeet steadier on mobile, with real end-of-utterance detection for streaming. - Supertonic TTS now runs full GPU across Metal, Vulkan, and OpenCL, with native streaming.

QVAC

20,931,912 просмотров • 2 месяцев назад

🤯 Building computer-vision applications is never going to be the same again. As Andrej Karpathy says, "English is going to be the new programming-language", and we just made that happen for programming visual AI. 👨‍💻 This past month, the VLM Run team cooked up something magical that I'm especially proud of - a natural-language interface to build and run computer vision workflows (object-detection, OCR, parsing, and so much more) – all via our new VLM Run MCP server. 🔌 It works with any MCP-compatible AI agent (Claude, OpenAI, Cursor, etc.) and abstracts complex visual workflows into simple English commands. No need to write Python or fiddle with bespoke computer-vision code – just say what you want your agent to do, and it does it. Here’s a glimpse into what you can build today ( ☑️ Image AI agents – run detection, tagging, blurring, cropping, and more traditional CV tasks with English ☑️ Document AI agents – extract totals from invoices, search visual content, or act on what’s found ☑️ Video AI agents – understand scene content, trim clips, add subtitles, and create ready-to-share edits ☑️ And best-of-all, it's fully programmable with English. 🧑‍🍳 MCP Showcase: 📚 MCP Cookbooks: 📚 MCP Docs: #computervision #ai #agents #llm #vlm #mcp #vibecoding #genai
1:49

🤯 Building computer-vision applications is never going to be the same again. As Andrej Karpathy says, "English is going to be the new programming-language", and we just made that happen for programming visual AI. 👨‍💻 This past month, the VLM Run team cooked up something magical that I'm especially proud of - a natural-language interface to build and run computer vision workflows (object-detection, OCR, parsing, and so much more) – all via our new VLM Run MCP server. 🔌 It works with any MCP-compatible AI agent (Claude, OpenAI, Cursor, etc.) and abstracts complex visual workflows into simple English commands. No need to write Python or fiddle with bespoke computer-vision code – just say what you want your agent to do, and it does it. Here’s a glimpse into what you can build today ( ☑️ Image AI agents – run detection, tagging, blurring, cropping, and more traditional CV tasks with English ☑️ Document AI agents – extract totals from invoices, search visual content, or act on what’s found ☑️ Video AI agents – understand scene content, trim clips, add subtitles, and create ready-to-share edits ☑️ And best-of-all, it's fully programmable with English. 🧑‍🍳 MCP Showcase: 📚 MCP Cookbooks: 📚 MCP Docs: #computervision #ai #agents #llm #vlm #mcp #vibecoding #genai

Sudeep Pillai

17,789 просмотров • 1 год назад

Google presents AudioPaLM: A Large Language Model That Can Speak and Listen paper page: introduce AudioPaLM, a large language model for speech understanding and generation. AudioPaLM fuses text-based and speech-based language models, PaLM-2 [Anil et al., 2023] and AudioLM [Borsos et al., 2022], into a unified multimodal architecture that can process and generate text and speech with applications including speech recognition and speech-to-speech translation. AudioPaLM inherits the capability to preserve paralinguistic information such as speaker identity and intonation from AudioLM and the linguistic knowledge present only in text large language models such as PaLM-2. We demonstrate that initializing AudioPaLM with the weights of a text-only large language model improves speech processing, successfully leveraging the larger quantity of text training data used in pretraining to assist with the speech tasks. The resulting model significantly outperforms existing systems for speech translation tasks and has the ability to perform zero-shot speech-to-text translation for many languages for which input/target language combinations were not seen in training. AudioPaLM also demonstrates features of audio language models, such as transferring a voice across languages based on a short spoken prompt.
4:10

Google presents AudioPaLM: A Large Language Model That Can Speak and Listen paper page: introduce AudioPaLM, a large language model for speech understanding and generation. AudioPaLM fuses text-based and speech-based language models, PaLM-2 [Anil et al., 2023] and AudioLM [Borsos et al., 2022], into a unified multimodal architecture that can process and generate text and speech with applications including speech recognition and speech-to-speech translation. AudioPaLM inherits the capability to preserve paralinguistic information such as speaker identity and intonation from AudioLM and the linguistic knowledge present only in text large language models such as PaLM-2. We demonstrate that initializing AudioPaLM with the weights of a text-only large language model improves speech processing, successfully leveraging the larger quantity of text training data used in pretraining to assist with the speech tasks. The resulting model significantly outperforms existing systems for speech translation tasks and has the ability to perform zero-shot speech-to-text translation for many languages for which input/target language combinations were not seen in training. AudioPaLM also demonstrates features of audio language models, such as transferring a voice across languages based on a short spoken prompt.

AK

290,517 просмотров • 3 лет назад

MiniMax H3 is native in ComfyUI, day zero with the weights. Model highlights: → Text-to-video, prompt only → Image-to-video → First-and-last-frame - control the opening frame, the closing frame, or both → Reference-to-video - carry a subject, a motion, or a voice through the clip → In-place editing - modify a shot you already have instead of regenerating it The capability MiniMax (official) leads with is multimodal context understanding, and it's what collapses those five tasks into one model. H3 takes images, audio, and video together and resolves them against a prompt that explains how they relate. Download the workflows & learn more in our blog 👇
1:06

MiniMax H3 is native in ComfyUI, day zero with the weights. Model highlights: → Text-to-video, prompt only → Image-to-video → First-and-last-frame - control the opening frame, the closing frame, or both → Reference-to-video - carry a subject, a motion, or a voice through the clip → In-place editing - modify a shot you already have instead of regenerating it The capability MiniMax (official) leads with is multimodal context understanding, and it's what collapses those five tasks into one model. H3 takes images, audio, and video together and resolves them against a prompt that explains how they relate. Download the workflows & learn more in our blog 👇

ComfyUI

230,129 просмотров • 14 дней назад

MotionGPT: Human Motion as a Foreign Language paper page: Though the advancement of pre-trained large language models unfolds, the exploration of building a unified model for language and other multi-modal data, such as motion, remains challenging and untouched so far. Fortunately, human motion displays a semantic coupling akin to human language, often perceived as a form of body language. By fusing language data with large-scale motion models, motion-language pre-training that can enhance the performance of motion-related tasks becomes feasible. Driven by this insight, we propose MotionGPT, a unified, versatile, and user-friendly motion-language model to handle multiple motion-relevant tasks. Specifically, we employ the discrete vector quantization for human motion and transfer 3D motion into motion tokens, similar to the generation process of word tokens. Building upon this "motion vocabulary", we perform language modeling on both motion and text in a unified manner, treating human motion as a specific language. Moreover, inspired by prompt learning, we pre-train MotionGPT with a mixture of motion-language data and fine-tune it on prompt-based question-and-answer tasks. Extensive experiments demonstrate that MotionGPT achieves state-of-the-art performances on multiple motion tasks including text-driven motion generation, motion captioning, motion prediction, and motion in-between.
0:48

MotionGPT: Human Motion as a Foreign Language paper page: Though the advancement of pre-trained large language models unfolds, the exploration of building a unified model for language and other multi-modal data, such as motion, remains challenging and untouched so far. Fortunately, human motion displays a semantic coupling akin to human language, often perceived as a form of body language. By fusing language data with large-scale motion models, motion-language pre-training that can enhance the performance of motion-related tasks becomes feasible. Driven by this insight, we propose MotionGPT, a unified, versatile, and user-friendly motion-language model to handle multiple motion-relevant tasks. Specifically, we employ the discrete vector quantization for human motion and transfer 3D motion into motion tokens, similar to the generation process of word tokens. Building upon this "motion vocabulary", we perform language modeling on both motion and text in a unified manner, treating human motion as a specific language. Moreover, inspired by prompt learning, we pre-train MotionGPT with a mixture of motion-language data and fine-tune it on prompt-based question-and-answer tasks. Extensive experiments demonstrate that MotionGPT achieves state-of-the-art performances on multiple motion tasks including text-driven motion generation, motion captioning, motion prediction, and motion in-between.

AK

125,353 просмотров • 3 лет назад

Chinese robotics company Astribot released their latest World-Action Model (WAM), Lumo-2. Technical breakdown: - based on a frozen 🥶 Qwen-3.5 4B VLM - trained in 3 progressive stages: 1. Action is aligned with latent world dynamics (an abstract representation of action). Real-world actions are anchored to physical constraints, while the latent space is guided to focus on motion-relevant changes. This bidirectional relationship makes the model physically grounded -> critical for a world model. 2. Action is aligned with vision and language. Reusing the vision backbone and action encoder from the frozen VLM, the authors add a custom vocabulary (for new actions), a semantic module, an action decoder, and an action projector. This aligns the (new) action representations with the (existing) vision-language semantic space. Most importantly: it builds a direct mapping from natural-language instructions to motor execution. 3. End-to-end training on language, video, and robot data. Only the new modules (everything outside the frozen backbone) are trained end-to-end across temporal reasoning, physical understanding, long-horizon, and dexterous manipulation. At the end of the day, Lumo-2 is not the best on benchmarks, but that's not the point. What's genuinely new: - a way to combine latent world modeling and action generation through progressive alignment - a physically-grounded latent dynamics space - it lifts performance on unseen objects using un-annotated human egocentric video + Vision Pro captures, no special transfer algorithm needed Why it matters: - the whole model is thin trainable adapters (semantic module, action decoder/projector) on a frozen 4B backbone (cheap) - that scale is suited for real-time embedded inference (~2.71× decode speedup, no accuracy loss) - its real moat is long-horizon execution, where the added temporal memory pays off far more than on any other task As a result, this robot can now make your latte (5x sped up video):
0:39

Chinese robotics company Astribot released their latest World-Action Model (WAM), Lumo-2. Technical breakdown: - based on a frozen 🥶 Qwen-3.5 4B VLM - trained in 3 progressive stages: 1. Action is aligned with latent world dynamics (an abstract representation of action). Real-world actions are anchored to physical constraints, while the latent space is guided to focus on motion-relevant changes. This bidirectional relationship makes the model physically grounded -> critical for a world model. 2. Action is aligned with vision and language. Reusing the vision backbone and action encoder from the frozen VLM, the authors add a custom vocabulary (for new actions), a semantic module, an action decoder, and an action projector. This aligns the (new) action representations with the (existing) vision-language semantic space. Most importantly: it builds a direct mapping from natural-language instructions to motor execution. 3. End-to-end training on language, video, and robot data. Only the new modules (everything outside the frozen backbone) are trained end-to-end across temporal reasoning, physical understanding, long-horizon, and dexterous manipulation. At the end of the day, Lumo-2 is not the best on benchmarks, but that's not the point. What's genuinely new: - a way to combine latent world modeling and action generation through progressive alignment - a physically-grounded latent dynamics space - it lifts performance on unseen objects using un-annotated human egocentric video + Vision Pro captures, no special transfer algorithm needed Why it matters: - the whole model is thin trainable adapters (semantic module, action decoder/projector) on a frozen 4B backbone (cheap) - that scale is suited for real-time embedded inference (~2.71× decode speedup, no accuracy loss) - its real moat is long-horizon execution, where the added temporal memory pays off far more than on any other task As a result, this robot can now make your latte (5x sped up video):

Léo

32,173 просмотров • 24 дней назад