Video wird geladen...

Video konnte nicht geladen werden

Zur Startseite

1/ [NeurIPS D&B] Introducing HourVideo: A benchmark for hour-long video-language understanding!🚀 500 egocentric videos, 18 total tasks & ~13k questions! Performance: GPT-4➡️25.7% Gemini 1.5 Pro➡️37.3% Humans➡️85.0% We highlight a significant gap in multimodal capabilities🧵👇

keshigeyan's profile picture

Keshigeyan Chandrasegaran

1,082 subscribers

111,369 Aufrufe • vor 1 Jahr •via X (Twitter)

LIVE

Anya Rossi• Live Now

Private livecam show

0 Kommentare

Keine Kommentare verfügbar

Kommentare vom Original-Post werden hier angezeigt

Ähnliche Videos

Google has launched Med-Gemini, an advanced AI fine-tuned for medical Tasks. It significantly outperforms earlier models, including GPT-4, on most medical benchmarks. Achieves top scores, particularly on the MedQA-USMLE benchmark with a groundbreaking 91.1% accuracy.🚀 Demonstrates superior performance over GPT-4 by 44.5% on average across seven multimodal benchmarks. Excels in tasks such as medical summarization, generating doctor referrals, and simplifying medical documents. It is a preferred method over human expert analyses for complex text-based medical tasks. This marks a significant advancement in AI for healthcare, suggesting potential improvements in medical diagnostics and patient care. By choosing to harmonize with technology, we become more human, we become IRREPLACEABLE. Join the IRREPLACEABLE Academy and read the Book: #techforgood #medicine #ai
0:45

Google has launched Med-Gemini, an advanced AI fine-tuned for medical Tasks. It significantly outperforms earlier models, including GPT-4, on most medical benchmarks. Achieves top scores, particularly on the MedQA-USMLE benchmark with a groundbreaking 91.1% accuracy.🚀 Demonstrates superior performance over GPT-4 by 44.5% on average across seven multimodal benchmarks. Excels in tasks such as medical summarization, generating doctor referrals, and simplifying medical documents. It is a preferred method over human expert analyses for complex text-based medical tasks. This marks a significant advancement in AI for healthcare, suggesting potential improvements in medical diagnostics and patient care. By choosing to harmonize with technology, we become more human, we become IRREPLACEABLE. Join the IRREPLACEABLE Academy and read the Book: #techforgood #medicine #ai

Pascal Bornet

14,393 Aufrufe • vor 1 Jahr

In December, we launched Gemini 1.0 Pro. Today, we're introducing Gemini 1.5 Pro! 🚀 This next-gen model uses a Mixture-of-Experts (MoE) approach for more efficient training & higher-quality responses. Gemini 1.5 Pro, our mid-sized model, will soon come standard with a 128K-token context window, but starting today, developers + customers can sign up for the limited Private Preview to try out 1.5 Pro with a groundbreaking and experimental 1 million token context window! The 1M tokens feature unlocks huge possibilities for devs - upload hundreds of pages of text, entire code repos, and long videos and let Gemini reason across them. It's still experimental and early and we’d love your feedback - learn more here.
0:20

In December, we launched Gemini 1.0 Pro. Today, we're introducing Gemini 1.5 Pro! 🚀 This next-gen model uses a Mixture-of-Experts (MoE) approach for more efficient training & higher-quality responses. Gemini 1.5 Pro, our mid-sized model, will soon come standard with a 128K-token context window, but starting today, developers + customers can sign up for the limited Private Preview to try out 1.5 Pro with a groundbreaking and experimental 1 million token context window! The 1M tokens feature unlocks huge possibilities for devs - upload hundreds of pages of text, entire code repos, and long videos and let Gemini reason across them. It's still experimental and early and we’d love your feedback - learn more here.

Sundar Pichai

852,611 Aufrufe • vor 2 Jahren

Robot AI brains, aka Vision-Language-Action models, cannot adapt to new tasks as easily as LLMs like Gemini, ChatGPT, or Grok. LLMs can adapt quickly with their in-context learning (ICL) capabilities. But can we inject ICL abilities into a pre-trained VLA like pi0? Yes! Introducing RICL (Retraining for In-Context Learning), our Conference on Robot Learning (CoRL) 2025 paper. Our RICL-pi0 model can adapt to unseen objects, novel motions, and new scenes with just ICL and RAG (retrieval-augmented generation). RICL-pi0 also boosts performance on the long-tail of tasks. A quick 1 minute video summary:
1:03

Robot AI brains, aka Vision-Language-Action models, cannot adapt to new tasks as easily as LLMs like Gemini, ChatGPT, or Grok. LLMs can adapt quickly with their in-context learning (ICL) capabilities. But can we inject ICL abilities into a pre-trained VLA like pi0? Yes! Introducing RICL (Retraining for In-Context Learning), our Conference on Robot Learning (CoRL) 2025 paper. Our RICL-pi0 model can adapt to unseen objects, novel motions, and new scenes with just ICL and RAG (retrieval-augmented generation). RICL-pi0 also boosts performance on the long-tail of tasks. A quick 1 minute video summary:

Kaustubh Sridhar

52,158 Aufrufe • vor 11 Monaten

📢 Join us tomorrow morning at our CVPR 2025 poster session (#340, ExHall D, 10:30am–12:30pm) to chat about Project Magma 👉 This is a big team effort to build a multimodal agentic model capable of understanding and acting in both digital and physical environments—just like how we interact with the world every day. 🚀 Even more exciting: we demonstrate the scaling potential of agent pretraining on large-scale human instructional videos through our Set-of-Mark (SoM) and Trace-of-Mark (ToM), showcasing strong zero-shot performance in: 1) multimodal image/video understanding, 2) UI navigation, 3) Real-world robot manipulation and even 4) Gaming! We've received encouraging feedback over the past few days—and this is only the beginning. A small step forward, with exciting things ahead!
1:31

📢 Join us tomorrow morning at our CVPR 2025 poster session (#340, ExHall D, 10:30am–12:30pm) to chat about Project Magma 👉 This is a big team effort to build a multimodal agentic model capable of understanding and acting in both digital and physical environments—just like how we interact with the world every day. 🚀 Even more exciting: we demonstrate the scaling potential of agent pretraining on large-scale human instructional videos through our Set-of-Mark (SoM) and Trace-of-Mark (ToM), showcasing strong zero-shot performance in: 1) multimodal image/video understanding, 2) UI navigation, 3) Real-world robot manipulation and even 4) Gaming! We've received encouraging feedback over the past few days—and this is only the beginning. A small step forward, with exciting things ahead!

Jianwei Yang

13,112 Aufrufe • vor 1 Jahr

Gemini-1.5 Pro has its spotlight stolen today, and people are poking fun at Sora vs Google memes. Well, I think it's the biggest boost in LLM capability so far in 2024. v1.5's 10M token context (1) excels at retrieval; (2) generalizes zero-shot to extremely long instructions like full tutorials and codebases; and (3) works across modalities such as text, audio, and video. Here's a stunning example: v1.5 learns to translate from English to Kalamang purely in context, following a full linguistic manual at inference time. Kalamang is a language spoken by fewer than 200 speakers in western New Guinea. Gemini has never seen this language during training and is only provided with 500 pages of linguistic documentation, a dictionary, and ~400 parallel sentences in context. It basically acquires a sophisticated new skill in the neural activations, instead of gradient finetuning. I talked about the Myth of Context Length many times before: don't get too excited by claims of 1M or even 1B context tokens. LSTMs already achieved literally infinite context length 25 yrs ago! What truly matters is how well the model actually uses the context to solve real-world problems, and Gemini-1.5 has surpassed the SOTA with flying colors. The paper is also well-written with lots of solid quantitative analysis on in-context memorization and generalization. Paper: “Gemini 1.5: Unlocking multimodal understanding across millions of tokens of context” Congrats to Jeff Dean Oriol Vinyals Sundar Pichai and team!
0:20

Gemini-1.5 Pro has its spotlight stolen today, and people are poking fun at Sora vs Google memes. Well, I think it's the biggest boost in LLM capability so far in 2024. v1.5's 10M token context (1) excels at retrieval; (2) generalizes zero-shot to extremely long instructions like full tutorials and codebases; and (3) works across modalities such as text, audio, and video. Here's a stunning example: v1.5 learns to translate from English to Kalamang purely in context, following a full linguistic manual at inference time. Kalamang is a language spoken by fewer than 200 speakers in western New Guinea. Gemini has never seen this language during training and is only provided with 500 pages of linguistic documentation, a dictionary, and ~400 parallel sentences in context. It basically acquires a sophisticated new skill in the neural activations, instead of gradient finetuning. I talked about the Myth of Context Length many times before: don't get too excited by claims of 1M or even 1B context tokens. LSTMs already achieved literally infinite context length 25 yrs ago! What truly matters is how well the model actually uses the context to solve real-world problems, and Gemini-1.5 has surpassed the SOTA with flying colors. The paper is also well-written with lots of solid quantitative analysis on in-context memorization and generalization. Paper: “Gemini 1.5: Unlocking multimodal understanding across millions of tokens of context” Congrats to Jeff Dean Oriol Vinyals Sundar Pichai and team!

Jim Fan

278,483 Aufrufe • vor 2 Jahren

🚀 New Paper: Pixel Reasoner 🧠🖼️ How can Vision-Language Models (VLMs) perform chain-of-thought reasoning within the image itself? We introduce Pixel Reasoner, the first open-source framework that enables VLMs to “think in pixel space” through curiosity-driven reinforcement learning. Current VLMs reason only in text — even when grounded in rich images or videos, their logical steps are verbalized in natural language. This restricts their ability to interrogate visual evidence and demonstrate how conclusions are drawn. 🔍 So we ask: What if we could make VLMs "show their work" by reasoning directly in the pixel space? Inspired by GPT-o3’s "think-in-image" ability, we propose a framework where VLMs use interactive visual operations — zoom, select-frame, highlight — to reason through complex visual inputs. To do this, we design a two-stage training process: Instruction tuning with synthesized visual reasoning traces. Reinforcement learning with curiosity-driven reward to balance exploration between pixel and text reasoning ✨ With this, Pixel Reasoner achieves near-SoTA performance on many information-rich multimodal benchmarks: 📊 84% on InfographicsVQA 🧠 84% on V* benchmark 🧩 74% on TallyQA-Complex It also achieves strong accuracy of 68% on MVBench (a video benchmark). Website: Paper: Code: Demo: (coming soon)
0:43

🚀 New Paper: Pixel Reasoner 🧠🖼️ How can Vision-Language Models (VLMs) perform chain-of-thought reasoning within the image itself? We introduce Pixel Reasoner, the first open-source framework that enables VLMs to “think in pixel space” through curiosity-driven reinforcement learning. Current VLMs reason only in text — even when grounded in rich images or videos, their logical steps are verbalized in natural language. This restricts their ability to interrogate visual evidence and demonstrate how conclusions are drawn. 🔍 So we ask: What if we could make VLMs "show their work" by reasoning directly in the pixel space? Inspired by GPT-o3’s "think-in-image" ability, we propose a framework where VLMs use interactive visual operations — zoom, select-frame, highlight — to reason through complex visual inputs. To do this, we design a two-stage training process: Instruction tuning with synthesized visual reasoning traces. Reinforcement learning with curiosity-driven reward to balance exploration between pixel and text reasoning ✨ With this, Pixel Reasoner achieves near-SoTA performance on many information-rich multimodal benchmarks: 📊 84% on InfographicsVQA 🧠 84% on V* benchmark 🧩 74% on TallyQA-Complex It also achieves strong accuracy of 68% on MVBench (a video benchmark). Website: Paper: Code: Demo: (coming soon)

Wenhu Chen

82,829 Aufrufe • vor 1 Jahr

🎉 恭喜发财🧧🐍 As we welcome the Chinese New Year, we're thrilled to announce the launch of Qwen2.5-VL , our latest flagship vision-language model! 🚀 💗 Qwen Chat: 📖 Blog: 🤗 Hugging Face: 🤖 ModelScope: 🌟 Key Highlights: * Visual Understanding : From flowers to complex charts, Qwen2.5-VL sees it all! * Agentic Capabilities : It’s a visual agent that can reason and interact with tools like computers & phones. * Long Video Comprehension : Captures events in videos over 1 hour long! ⏳🎥 * Precise Localization : Generates bounding boxes & JSON outputs for accurate object detection. * Structured Data Outputs : Perfect for finance & commerce, handling invoices, forms & more! 💼📊 Try Qwen2.5-VL now at Qwen Chat or explore models on Hugging Face & ModelScope . 🌐
1:32

🎉 恭喜发财🧧🐍 As we welcome the Chinese New Year, we're thrilled to announce the launch of Qwen2.5-VL , our latest flagship vision-language model! 🚀 💗 Qwen Chat: 📖 Blog: 🤗 Hugging Face: 🤖 ModelScope: 🌟 Key Highlights: * Visual Understanding : From flowers to complex charts, Qwen2.5-VL sees it all! * Agentic Capabilities : It’s a visual agent that can reason and interact with tools like computers & phones. * Long Video Comprehension : Captures events in videos over 1 hour long! ⏳🎥 * Precise Localization : Generates bounding boxes & JSON outputs for accurate object detection. * Structured Data Outputs : Perfect for finance & commerce, handling invoices, forms & more! 💼📊 Try Qwen2.5-VL now at Qwen Chat or explore models on Hugging Face & ModelScope . 🌐

Qwen

762,314 Aufrufe • vor 1 Jahr

🚀 Today, we are introducing SmolTools! 🚀 Last week, at Hugging Face we made a significant leap forward with the release of SmolLM2, a compact 1.7B language model that sets a new benchmark for performance among models of its size. But beyond the impressive stats, SmolLM2 truly shines in practical, real-world applications, unlocking new possibilities for on-device inference. To demonstrate its potential, we're thrilled to introduce Smol-Tools, a suite of simple yet powerful applications that showcase the capabilities of SmolLM2. 🌟 Today, we present you two key tools: Summarize and Rewrite. 🔹 Summarize: Feed SmolLM2 a text up to 20 pages long, and it will provide you with a concise summary. Need to dig deeper? Just ask follow-up questions to clarify any details – it's that simple. 🔹 Rewrite: Draft a response to a message with your main points, and SmolLM2 will transform it into a clear, polished version that's easy for your recipient to read and understand. These tools are designed to make your workflow smoother and more efficient, leveraging the power of SmolLM2 in practical ways. 💡We hope you give these tools a try and see their potential for yourself! Feel free to build your own SmolTools and contribute them to the project. We'd love to see what you build! Check out the code here:
1:11

🚀 Today, we are introducing SmolTools! 🚀 Last week, at Hugging Face we made a significant leap forward with the release of SmolLM2, a compact 1.7B language model that sets a new benchmark for performance among models of its size. But beyond the impressive stats, SmolLM2 truly shines in practical, real-world applications, unlocking new possibilities for on-device inference. To demonstrate its potential, we're thrilled to introduce Smol-Tools, a suite of simple yet powerful applications that showcase the capabilities of SmolLM2. 🌟 Today, we present you two key tools: Summarize and Rewrite. 🔹 Summarize: Feed SmolLM2 a text up to 20 pages long, and it will provide you with a concise summary. Need to dig deeper? Just ask follow-up questions to clarify any details – it's that simple. 🔹 Rewrite: Draft a response to a message with your main points, and SmolLM2 will transform it into a clear, polished version that's easy for your recipient to read and understand. These tools are designed to make your workflow smoother and more efficient, leveraging the power of SmolLM2 in practical ways. 💡We hope you give these tools a try and see their potential for yourself! Feel free to build your own SmolTools and contribute them to the project. We'd love to see what you build! Check out the code here:

Andi Marafioti

100,813 Aufrufe • vor 1 Jahr

"The future of AI is agentic. That includes browsers!" Imagine having an AI agent in your browser that can help you complete complex tasks, answer your questions, and streamline your workflow. Today I'm thrilled to share a sneak peek at Project Mariner, a cutting-edge research collaboration between Chrome and Google DeepMind, exploring the future of agentic AI within the browser! Building on the power of Gemini 2.0, Mariner envisions AI agents seamlessly guiding users through online tasks, streamlining workflows and enriching browsing experiences. Imagine having an intelligent co-pilot in your browser, anticipating your needs and proactively offering assistance. We're in the early stages of experimentation, focusing on core functionalities like understanding user intent, automating actions, and providing personalized recommendations. This prototype leverages Gemini's advanced natural language understanding and reasoning capabilities to interpret user requests, both typed and spoken. Mariner can then interact with web pages, retrieve information, and even perform actions like filling out forms or navigating to specific sites. For example, a user could simply ask "Find me a job near me," and Mariner would understand the request, navigate to a relevant job search site, and tailor the search based on the user's location and preferences. This is just one example of how we're exploring Gemini 2.0's potential to unlock agentic experiences through a series of prototypes, including: 1. Agents with multimodal reasoning: Project Astra, our research prototype exploring the capabilities of a universal AI assistant, is enhanced by Gemini 2.0. 2. Agents that can help you accomplish complex tasks: Project Mariner itself focuses on the future of human-agent interaction within the browser. 3. Agents for developers: Jules is an experimental AI-powered coding agent that integrates directly into a GitHub workflow. 4. Agents applied across domains: We're exploring agents for navigating video games and even applying Gemini 2.0's spatial reasoning to robotics. We believe that integrating AI agents directly into the browser has the potential to revolutionize how we interact with the web. Project Mariner aims to make browsing more intuitive, efficient, and personalized. By understanding user context and proactively offering assistance, Mariner can simplify complex tasks, save users time, and empower them to achieve more online. This aligns perfectly with the vision of Gemini 2.0 to create more helpful and intuitive AI experiences. We’re currently testing Mariner with a small group of trusted users to gather feedback and refine the user experience. We believe that this technology holds immense potential to transform the way we browse and interact with information online.
2:14

"The future of AI is agentic. That includes browsers!" Imagine having an AI agent in your browser that can help you complete complex tasks, answer your questions, and streamline your workflow. Today I'm thrilled to share a sneak peek at Project Mariner, a cutting-edge research collaboration between Chrome and Google DeepMind, exploring the future of agentic AI within the browser! Building on the power of Gemini 2.0, Mariner envisions AI agents seamlessly guiding users through online tasks, streamlining workflows and enriching browsing experiences. Imagine having an intelligent co-pilot in your browser, anticipating your needs and proactively offering assistance. We're in the early stages of experimentation, focusing on core functionalities like understanding user intent, automating actions, and providing personalized recommendations. This prototype leverages Gemini's advanced natural language understanding and reasoning capabilities to interpret user requests, both typed and spoken. Mariner can then interact with web pages, retrieve information, and even perform actions like filling out forms or navigating to specific sites. For example, a user could simply ask "Find me a job near me," and Mariner would understand the request, navigate to a relevant job search site, and tailor the search based on the user's location and preferences. This is just one example of how we're exploring Gemini 2.0's potential to unlock agentic experiences through a series of prototypes, including: 1. Agents with multimodal reasoning: Project Astra, our research prototype exploring the capabilities of a universal AI assistant, is enhanced by Gemini 2.0. 2. Agents that can help you accomplish complex tasks: Project Mariner itself focuses on the future of human-agent interaction within the browser. 3. Agents for developers: Jules is an experimental AI-powered coding agent that integrates directly into a GitHub workflow. 4. Agents applied across domains: We're exploring agents for navigating video games and even applying Gemini 2.0's spatial reasoning to robotics. We believe that integrating AI agents directly into the browser has the potential to revolutionize how we interact with the web. Project Mariner aims to make browsing more intuitive, efficient, and personalized. By understanding user context and proactively offering assistance, Mariner can simplify complex tasks, save users time, and empower them to achieve more online. This aligns perfectly with the vision of Gemini 2.0 to create more helpful and intuitive AI experiences. We’re currently testing Mariner with a small group of trusted users to gather feedback and refine the user experience. We believe that this technology holds immense potential to transform the way we browse and interact with information online.

Addy Osmani

29,501 Aufrufe • vor 1 Jahr

EXCLUSIVE: The hidden $250K machine of 9 paid vendors behind the 'flagship' #NoKings protest in St. Paul, Minnesota I followed the money behind the No Kings protest in St. Paul, Minn., and uncovered an estimated $250,000 paid to 9 vendors to produce an event that was about the size of a Def Leppard concert. Sources said that the Democratic nonprofit Indivisible paid the bill. It didn't respond to numerous requests for comment. How did I piece this together? Well, I have a rule when reporting on the protest industry: be the first there and one of the last to leave. That’s how I met Slamhammer Sound & Roadcase Co. production manager Matt Svobodny, one of the very nice hard-working members of the production crew behind the scenes in St. Paul, as they were breaking down the set for the No Kings protest, long after Bruce Springsteen and most of the anti-Trump protesters had left. He was straightforward, candid and matter-of-fact about what it takes to throw a protest and, a few days later, guided me -- and you -- through the warehouse where Slamhammer stores the equipment it pulled out for the protest. He provided the kind of transparency that the secretive nonprofits behind the protests should actually be providing to citizens and the media. See for yourself: ➡️ the mobile stage ➡️ the speakers ➡️ nearly a mile of heavy-duty feeder cable used to distribute electricity throughout the rally site and the ballistic ➡️ bullet-resistant barriers that shielded the Bruce Springsteen, Jane Fonda, Joan Baez, TIm Walz, Ilhan Omar, Randi Weingarten and the day's other bold-faced names WATCH the video that I recorded ⬇️ Thank you to Fox News Digital's Hannah Brennan for her work editing the video. In our new Fox News Digital exclusive, I lay out how the "flagship" protest in St. Paul wasn’t spontaneous, like most of the media reported. It was professionally engineered. And that raises a bigger question: When protests look like productions…who’s really behind the curtain? I answer that question in the article and the thread below 🧵👇 A former Obama and Biden administration political strategist and campaign operative Roger Fisk takes credit for being the "Senior Advisor to the #NoKings flagship event," fine-tuning the "art and science" of throwing the St. Paul protest, along with two other "No Kings" protests last year. Fisk didn't respond to a request for comment. The protests have parroted Chinese government propaganda, demonizing America as a "fascist" nation and Trump as a "king." Partners in the protests were pro-communist groups funded by Neville Roy Singham, a tech tycoon living in Shanghai. DataRepublican (small r), You'll want to read this. READ: Behind the scenes, I identified 9 vendors that were paid an estimated $250,000 to construct the protest: ➡️ Slamhammer Sound & Roadcase Co. — mobile stage, 100-speaker sound system, lighting, 1,700 ft cable, ballistic barriers → estimated $100,000 ➡️ Fire Up Video — 4 jumbo screens → estimated $20,000 ➡️ Algorithm, an AV company — 2 jumbo screens → estimated $25,000 ➡️ Common World Productions — 2 LED stage screens → estimated $10,000 ➡️ Warning Lites of Minnesota — bike-rack barricades → estimated $15,000 ➡️ E5 Energy — generators, electrical distribution → estimated $15,000 ➡️ Ultimate Events — tents, chairs, tables → estimated $30,000 ➡️ On Site Companies — ~300 porta-toilets → estimated $25,000 ➡️ Fast Kat Connects — high-speed internet → estimated $10,000 Total: an estimated $250,000. This wasn’t a rally that just “popped up,” as CNN reported. It was built, truck by truck, cable by cable, screen by screen. 🧵 with how it worked.
4:50

EXCLUSIVE: The hidden $250K machine of 9 paid vendors behind the 'flagship' #NoKings protest in St. Paul, Minnesota I followed the money behind the No Kings protest in St. Paul, Minn., and uncovered an estimated $250,000 paid to 9 vendors to produce an event that was about the size of a Def Leppard concert. Sources said that the Democratic nonprofit Indivisible paid the bill. It didn't respond to numerous requests for comment. How did I piece this together? Well, I have a rule when reporting on the protest industry: be the first there and one of the last to leave. That’s how I met Slamhammer Sound & Roadcase Co. production manager Matt Svobodny, one of the very nice hard-working members of the production crew behind the scenes in St. Paul, as they were breaking down the set for the No Kings protest, long after Bruce Springsteen and most of the anti-Trump protesters had left. He was straightforward, candid and matter-of-fact about what it takes to throw a protest and, a few days later, guided me -- and you -- through the warehouse where Slamhammer stores the equipment it pulled out for the protest. He provided the kind of transparency that the secretive nonprofits behind the protests should actually be providing to citizens and the media. See for yourself: ➡️ the mobile stage ➡️ the speakers ➡️ nearly a mile of heavy-duty feeder cable used to distribute electricity throughout the rally site and the ballistic ➡️ bullet-resistant barriers that shielded the Bruce Springsteen, Jane Fonda, Joan Baez, TIm Walz, Ilhan Omar, Randi Weingarten and the day's other bold-faced names WATCH the video that I recorded ⬇️ Thank you to Fox News Digital's Hannah Brennan for her work editing the video. In our new Fox News Digital exclusive, I lay out how the "flagship" protest in St. Paul wasn’t spontaneous, like most of the media reported. It was professionally engineered. And that raises a bigger question: When protests look like productions…who’s really behind the curtain? I answer that question in the article and the thread below 🧵👇 A former Obama and Biden administration political strategist and campaign operative Roger Fisk takes credit for being the "Senior Advisor to the #NoKings flagship event," fine-tuning the "art and science" of throwing the St. Paul protest, along with two other "No Kings" protests last year. Fisk didn't respond to a request for comment. The protests have parroted Chinese government propaganda, demonizing America as a "fascist" nation and Trump as a "king." Partners in the protests were pro-communist groups funded by Neville Roy Singham, a tech tycoon living in Shanghai. DataRepublican (small r), You'll want to read this. READ: Behind the scenes, I identified 9 vendors that were paid an estimated $250,000 to construct the protest: ➡️ Slamhammer Sound & Roadcase Co. — mobile stage, 100-speaker sound system, lighting, 1,700 ft cable, ballistic barriers → estimated $100,000 ➡️ Fire Up Video — 4 jumbo screens → estimated $20,000 ➡️ Algorithm, an AV company — 2 jumbo screens → estimated $25,000 ➡️ Common World Productions — 2 LED stage screens → estimated $10,000 ➡️ Warning Lites of Minnesota — bike-rack barricades → estimated $15,000 ➡️ E5 Energy — generators, electrical distribution → estimated $15,000 ➡️ Ultimate Events — tents, chairs, tables → estimated $30,000 ➡️ On Site Companies — ~300 porta-toilets → estimated $25,000 ➡️ Fast Kat Connects — high-speed internet → estimated $10,000 Total: an estimated $250,000. This wasn’t a rally that just “popped up,” as CNN reported. It was built, truck by truck, cable by cable, screen by screen. 🧵 with how it worked.

Asra Nomani

881,300 Aufrufe • vor 3 Monaten

Can GPT-4 teach a robot hand to do pen spinning tricks better than you do? I'm excited to announce Eureka, an open-ended agent that designs reward functions for robot dexterity at super-human level. It’s like Voyager in the space of a physics simulator API! Eureka bridges the gap between high-level reasoning (coding) and low-level motor control. It is a “hybrid-gradient architecture”: a black box, inference-only LLM instructs a white box, learnable neural network. The outer loop runs GPT-4 to refine the reward function (gradient-free), while the inner loop runs reinforcement learning to train a robot controller (gradient-based). We are able to scale up Eureka thanks to IsaacGym, a GPU-accelerated physics simulator that speeds up reality by 1000x. On a benchmark suite of 29 tasks across 10 robots, Eureka rewards outperform expert human-written ones on 83% of the tasks by 52% improvement margin on average. We are surprised that Eureka is able to learn pen spinning tricks, which are very difficult even for CGI artists to animate frame by frame! Eureka also enables a new form of in-context RLHF, which is able to incorporate a human operator’s feedback in natural language to steer and align the reward functions. It can serve as a powerful co-pilot for robot engineers to design sophisticated motor behaviors. As usual, we open-source everything! Welcome you all to check out our video gallery and try the codebase today: Paper: Code: Deep dive with me: 🧵
0:24

Can GPT-4 teach a robot hand to do pen spinning tricks better than you do? I'm excited to announce Eureka, an open-ended agent that designs reward functions for robot dexterity at super-human level. It’s like Voyager in the space of a physics simulator API! Eureka bridges the gap between high-level reasoning (coding) and low-level motor control. It is a “hybrid-gradient architecture”: a black box, inference-only LLM instructs a white box, learnable neural network. The outer loop runs GPT-4 to refine the reward function (gradient-free), while the inner loop runs reinforcement learning to train a robot controller (gradient-based). We are able to scale up Eureka thanks to IsaacGym, a GPU-accelerated physics simulator that speeds up reality by 1000x. On a benchmark suite of 29 tasks across 10 robots, Eureka rewards outperform expert human-written ones on 83% of the tasks by 52% improvement margin on average. We are surprised that Eureka is able to learn pen spinning tricks, which are very difficult even for CGI artists to animate frame by frame! Eureka also enables a new form of in-context RLHF, which is able to incorporate a human operator’s feedback in natural language to steer and align the reward functions. It can serve as a powerful co-pilot for robot engineers to design sophisticated motor behaviors. As usual, we open-source everything! Welcome you all to check out our video gallery and try the codebase today: Paper: Code: Deep dive with me: 🧵

Jim Fan

2,674,056 Aufrufe • vor 2 Jahren

My Premium League SBC Grind👇 ➡️ An Infinite coins generator🤑 ⬅️ 🔮 FUT Fantasy Edition 🔮 💚+🔁 appreciated ● Liga Portugal🇵🇹/Pro 1🇧🇪 In average I need to buy 8 cards per run. Sometimes less (after doing the Daily Upgrades and Winter Provision), sometimes less, sometimes all 11! ➡️ cost at the moment: 300 coins ➡️ 8×300=2.400 (Max. 3.300) ● Ligue 1🇫🇷/Eredivisie🇳🇱 I buy lots of Ligue 1 or Eredivisie silvers (rating 67+) for 150 on bid or 200 on the market, as I use 6-8 silvers per run. I fill them up with golds I receive back during the runs, dailies and Winter Provision Upgrades! ➡️ cost: ~1.4k per run ● Bundesliga🇩🇪/Serie A🇮🇹 I buy some silvers (70-74 rating) for 150 on bid or 200 on the market, as I use 3-5 per run (4 in average). I fill them up with golds I receive back during the runs, dailies and Winter Provision Upgrades! ➡️ cost: ~800 ● Prem🏴󠁧󠁢󠁥󠁮󠁧󠁿/La Liga🇪🇸 I dont buy anything here. Ive doing them everytime with players of my club. You can add 1-2 silvers (72-74) for 200 coins each, if you want. But no need tbf! ➡️ cost: 0 Max. cost in total: 5.500 coins Average cost per run: ~3-4k coins Average return: ~6-8k coins ➡️ based on my experience ● How I deal with tradeable packs • I sell: ☆ 86+ ☆ TOTWs ☆ Chem styles with a value ➡️ Hunter, Shadow, Anchor, Engine,... ☆ Managers above discard ☆ Players above discard • I keep in the club: ☆ League SBC league cards ☆ 83-85s ☆ managers with discard value • I discard: ☆ off league cards with discard value (You can also keep them) ☆ consumables with discard value ☆ discard Promo cards ☆ ALL dupes with discard value ●How to deal with untrade cards: • Premium leagues (<82) into Premium • Off league golds (<82) into: ☆ 80+ Pick (4/10) ☆ 81+ Pick (1/3) • 83s-84s + TOTW into 83+×14 • 82-83s into 85-87 Upgrade (if you want) • 85s into Winter Provision🌟 • 85+ into Player/Icon SBC ➡️ tradeable packs back ● Conclusion You will end up with profit all the time, if you do it the way I explained above🤑 As you pack a discard Promo card, a TOTW, or a chemstyle with value, sometimes even a gold Walkout here and there, your coins will rise quite fastly💥 You also get tradeable silver packs back from the 80+ (4/10) Picks and other tradeable packs from the Player/Icon SBCs🔥 I would also recommend doing Marquee Matchups + UCL Marquee Matchups within for even more tradeable pack return✌🏻 The video is just for the visualization! Make sure to Follow🤝🏻
0:23

My Premium League SBC Grind👇 ➡️ An Infinite coins generator🤑 ⬅️ 🔮 FUT Fantasy Edition 🔮 💚+🔁 appreciated ● Liga Portugal🇵🇹/Pro 1🇧🇪 In average I need to buy 8 cards per run. Sometimes less (after doing the Daily Upgrades and Winter Provision), sometimes less, sometimes all 11! ➡️ cost at the moment: 300 coins ➡️ 8×300=2.400 (Max. 3.300) ● Ligue 1🇫🇷/Eredivisie🇳🇱 I buy lots of Ligue 1 or Eredivisie silvers (rating 67+) for 150 on bid or 200 on the market, as I use 6-8 silvers per run. I fill them up with golds I receive back during the runs, dailies and Winter Provision Upgrades! ➡️ cost: ~1.4k per run ● Bundesliga🇩🇪/Serie A🇮🇹 I buy some silvers (70-74 rating) for 150 on bid or 200 on the market, as I use 3-5 per run (4 in average). I fill them up with golds I receive back during the runs, dailies and Winter Provision Upgrades! ➡️ cost: ~800 ● Prem🏴󠁧󠁢󠁥󠁮󠁧󠁿/La Liga🇪🇸 I dont buy anything here. Ive doing them everytime with players of my club. You can add 1-2 silvers (72-74) for 200 coins each, if you want. But no need tbf! ➡️ cost: 0 Max. cost in total: 5.500 coins Average cost per run: ~3-4k coins Average return: ~6-8k coins ➡️ based on my experience ● How I deal with tradeable packs • I sell: ☆ 86+ ☆ TOTWs ☆ Chem styles with a value ➡️ Hunter, Shadow, Anchor, Engine,... ☆ Managers above discard ☆ Players above discard • I keep in the club: ☆ League SBC league cards ☆ 83-85s ☆ managers with discard value • I discard: ☆ off league cards with discard value (You can also keep them) ☆ consumables with discard value ☆ discard Promo cards ☆ ALL dupes with discard value ●How to deal with untrade cards: • Premium leagues (<82) into Premium • Off league golds (<82) into: ☆ 80+ Pick (4/10) ☆ 81+ Pick (1/3) • 83s-84s + TOTW into 83+×14 • 82-83s into 85-87 Upgrade (if you want) • 85s into Winter Provision🌟 • 85+ into Player/Icon SBC ➡️ tradeable packs back ● Conclusion You will end up with profit all the time, if you do it the way I explained above🤑 As you pack a discard Promo card, a TOTW, or a chemstyle with value, sometimes even a gold Walkout here and there, your coins will rise quite fastly💥 You also get tradeable silver packs back from the 80+ (4/10) Picks and other tradeable packs from the Player/Icon SBCs🔥 I would also recommend doing Marquee Matchups + UCL Marquee Matchups within for even more tradeable pack return✌🏻 The video is just for the visualization! Make sure to Follow🤝🏻

Chem Expert 🐦 EA FC

43,802 Aufrufe • vor 4 Monaten

My Premium League SBC Grind👇 ➡️ An Infinite coins generator🤑 ⬅️ 🌟 Future Stars Grind Edition🌟 💚+🔁 appreciated ● Liga Portugal🇵🇹/Pro 1🇧🇪 In average I need to buy 8 cards per run. Sometimes less (after doing the Daily Upgrades and Winter Provision), sometimes all 11, if I grind a lot after the helpful Daily Upgrades ➡️ cost at the moment: 400 coins ➡️ 8×400=3.200 (Max. 4.400) ● Ligue 1🇫🇷/Eredivisie🇳🇱 I buy lots of Ligue 1 silvers (rating 69+) for 150 on bid or 200 on the market, as I use 7 silvers per run. I fill them up with golds I receive back during the runs, dailies and Winter Provision Upgrades ➡️ cost: max. 1,4k ● Bundesliga🇩🇪/Serie A🇮🇹 I buy some silvers (70-74 rating) for 150 on bid or 200 on the market, as I use 3-5 per run (4 in average). I fill them up with golds I receive back during the runs, dailies and Winter Provision Upgrades. ➡️ cost: max. 800 ● Prem🏴󠁧󠁢󠁥󠁮󠁧󠁿/La Liga🇪🇸 I dont buy anything here. Ive doing them everytime with players of my club. You can add 1-2 silvers (72-74) for 200 coins each, if you want. But no need tbf! ➡️ cost: 0 Max. cost in total: 6200 coins Average cost per run: ~4-5k coins Average return: ~6-8k coins ➡️ based on my experience ● How I deal with tradeable packs • I sell: ☆ 85+ ☆ TOTWs ☆ Chem styles with a value ➡️ Hunter, Shadow, Anchor, Engine,... ☆ Managers above discard ☆ Players above discard • I keep in the club: ☆ League SBC league cards ☆ 83s and 84s ☆ managers with discard value • I discard: ☆ off league cards with discard value (You can also keep them) ☆ consumables with discard value ☆ discard Promo cards ☆ ALL dupes with discard value ●How to deal with untrade cards: • Premium leagues (<82) into Premium • Off league golds (<82) into: ☆ 80+ Pick (4/10) ☆ Crafting Upgrades or ☆ 80+ Pick (1/3) • 82s-84s into 85+×3 • 83s-84s + TOTW into 83+×10 • 84-86 into 88-90 Upgrade • 85s into Winter Provision • 85+ into Player/Icon SBC ➡️ tradeable packs back ● Conclusion You will end up with profit all the time, if you do it on that way, even if you pack nothing with a value. You don't even need the Bundesliga/Serie A silvers mostly, I just buy and use them, to hoard more golds from these leagues for other Upgrades. As you pack a discard Promo card, a TOTW, or a chemstyle with value, sometimes even a gold Walkout here and there, your coins will rise quite fastly💥 You also get tradeable silver packs back from the 80+ (4/10) Picks and other tradeable packs from the 85+×3 and Player/Icon SBCs🔥 The video is just for the visualization✌🏻 Make sure to Follow🤝🏻
0:31

My Premium League SBC Grind👇 ➡️ An Infinite coins generator🤑 ⬅️ 🌟 Future Stars Grind Edition🌟 💚+🔁 appreciated ● Liga Portugal🇵🇹/Pro 1🇧🇪 In average I need to buy 8 cards per run. Sometimes less (after doing the Daily Upgrades and Winter Provision), sometimes all 11, if I grind a lot after the helpful Daily Upgrades ➡️ cost at the moment: 400 coins ➡️ 8×400=3.200 (Max. 4.400) ● Ligue 1🇫🇷/Eredivisie🇳🇱 I buy lots of Ligue 1 silvers (rating 69+) for 150 on bid or 200 on the market, as I use 7 silvers per run. I fill them up with golds I receive back during the runs, dailies and Winter Provision Upgrades ➡️ cost: max. 1,4k ● Bundesliga🇩🇪/Serie A🇮🇹 I buy some silvers (70-74 rating) for 150 on bid or 200 on the market, as I use 3-5 per run (4 in average). I fill them up with golds I receive back during the runs, dailies and Winter Provision Upgrades. ➡️ cost: max. 800 ● Prem🏴󠁧󠁢󠁥󠁮󠁧󠁿/La Liga🇪🇸 I dont buy anything here. Ive doing them everytime with players of my club. You can add 1-2 silvers (72-74) for 200 coins each, if you want. But no need tbf! ➡️ cost: 0 Max. cost in total: 6200 coins Average cost per run: ~4-5k coins Average return: ~6-8k coins ➡️ based on my experience ● How I deal with tradeable packs • I sell: ☆ 85+ ☆ TOTWs ☆ Chem styles with a value ➡️ Hunter, Shadow, Anchor, Engine,... ☆ Managers above discard ☆ Players above discard • I keep in the club: ☆ League SBC league cards ☆ 83s and 84s ☆ managers with discard value • I discard: ☆ off league cards with discard value (You can also keep them) ☆ consumables with discard value ☆ discard Promo cards ☆ ALL dupes with discard value ●How to deal with untrade cards: • Premium leagues (<82) into Premium • Off league golds (<82) into: ☆ 80+ Pick (4/10) ☆ Crafting Upgrades or ☆ 80+ Pick (1/3) • 82s-84s into 85+×3 • 83s-84s + TOTW into 83+×10 • 84-86 into 88-90 Upgrade • 85s into Winter Provision • 85+ into Player/Icon SBC ➡️ tradeable packs back ● Conclusion You will end up with profit all the time, if you do it on that way, even if you pack nothing with a value. You don't even need the Bundesliga/Serie A silvers mostly, I just buy and use them, to hoard more golds from these leagues for other Upgrades. As you pack a discard Promo card, a TOTW, or a chemstyle with value, sometimes even a gold Walkout here and there, your coins will rise quite fastly💥 You also get tradeable silver packs back from the 80+ (4/10) Picks and other tradeable packs from the 85+×3 and Player/Icon SBCs🔥 The video is just for the visualization✌🏻 Make sure to Follow🤝🏻

Chem Expert 🐦 EA FC

394,807 Aufrufe • vor 5 Monaten

Alibaba just dropped Qwen3.5-397B-A17B and there's a lot to unpack. 397B params, 17B active per forward pass. Sparse MoE done right. But the real story isn't the size—it's the architecture choices. The MoE Design Most MoE models feel like bolt-ons. Qwen 3.5's sparse activation is native—only 4.3% of parameters fire per token. That's how you get trillion-parameter-class performance without trillion-parameter inference costs. The 0.8 RMB/million tokens pricing isn't subsidized; it's structurally earned. Native Multimodal, Not Glued-On This is a vision-language model from the ground up. Heterogeneous architecture—separate processing pipelines for text, image, video that fuse early. Not a vision encoder slapped onto an LLM. The result: 90.8 on OmniDocBench, 79.0 on MMMU-Pro. Document understanding and visual reasoning without the usual brittleness. The Context Window Reality Qwen3.5-Plus (the hosted version) ships with 1M tokens by default. That's not a marketing number—they're actually positioning it for long-document workflows. With built-in adaptive tool use, it's clearly aimed at agentic automation, not just chat. What Actually Impressed Me • FP8 native pipeline: ~50% activation memory reduction • Async RL framework for continuous refinement—training and inference workloads separated • 201 languages (up from 119), 250k vocab for better low-resource encoding • Apache 2.0 license. Full weights on HuggingFace and ModelScope. The Benchmark Context 76.4 on SWE-bench Verified puts it in the range where it can handle real debugging workflows. 72.9 on BFCL v4 for agentic tool use. 88.4 on GPQA Diamond. These aren't SOTA in isolation, but the breadth is unusual—strong across reasoning, coding, multimodal, and agentic tasks. The Honest Caveat I haven't stress-tested the 1M context for needle-in-haystack retrieval yet. And "native multimodal" claims need real-world torture testing—PDFs with tables, charts, mixed layouts. Benchmarks are benchmarks. Bottom Line This isn't just another model release. It's a bet on efficient scale: big model capabilities, small active compute, open weights. At 1/18th the cost of Gemini 3 Pro, it's going to force pricing conversations across the board.
1:03

Alibaba just dropped Qwen3.5-397B-A17B and there's a lot to unpack. 397B params, 17B active per forward pass. Sparse MoE done right. But the real story isn't the size—it's the architecture choices. The MoE Design Most MoE models feel like bolt-ons. Qwen 3.5's sparse activation is native—only 4.3% of parameters fire per token. That's how you get trillion-parameter-class performance without trillion-parameter inference costs. The 0.8 RMB/million tokens pricing isn't subsidized; it's structurally earned. Native Multimodal, Not Glued-On This is a vision-language model from the ground up. Heterogeneous architecture—separate processing pipelines for text, image, video that fuse early. Not a vision encoder slapped onto an LLM. The result: 90.8 on OmniDocBench, 79.0 on MMMU-Pro. Document understanding and visual reasoning without the usual brittleness. The Context Window Reality Qwen3.5-Plus (the hosted version) ships with 1M tokens by default. That's not a marketing number—they're actually positioning it for long-document workflows. With built-in adaptive tool use, it's clearly aimed at agentic automation, not just chat. What Actually Impressed Me • FP8 native pipeline: ~50% activation memory reduction • Async RL framework for continuous refinement—training and inference workloads separated • 201 languages (up from 119), 250k vocab for better low-resource encoding • Apache 2.0 license. Full weights on HuggingFace and ModelScope. The Benchmark Context 76.4 on SWE-bench Verified puts it in the range where it can handle real debugging workflows. 72.9 on BFCL v4 for agentic tool use. 88.4 on GPQA Diamond. These aren't SOTA in isolation, but the breadth is unusual—strong across reasoning, coding, multimodal, and agentic tasks. The Honest Caveat I haven't stress-tested the 1M context for needle-in-haystack retrieval yet. And "native multimodal" claims need real-world torture testing—PDFs with tables, charts, mixed layouts. Benchmarks are benchmarks. Bottom Line This isn't just another model release. It's a bet on efficient scale: big model capabilities, small active compute, open weights. At 1/18th the cost of Gemini 3 Pro, it's going to force pricing conversations across the board.

Bo Wang

13,221 Aufrufe • vor 5 Monaten

Why General AI Fails Tax Professionals — and What Makes TaxGPT State-of-the-Art Tax LLM Every tax professional using general-purpose AI tools like ChatGPT or Claude for research is one hallucinated source away from costly damage to their clients and professional career. So we ran a head-to-head benchmark: TaxGPT vs. OpenAI, Anthropic, and Gemini. Our methodology: 1️⃣ Quantitative testing on CPA + EA tax-specific exam questions 2️⃣ Qualitative testing complex, real-world scenarios across trust & estate, direct and indirect taxes, payroll, multi-state nexus, advisory, and compliance Results (Accuracy): TaxGPT: 96% Gemini 2.5 Pro: 89% Claude Sonnet 4: 87% GPT-4o: 80% GPT5: 92% But here’s the real story 👇 The Source Quality Gap General LLMs are great test takers. But tax research isn’t a trivia game — it’s a liability-driven profession where every position must be defensible. When we examined the sources behind their answers, the gap became a canyon: 🔍 OpenAI included citations only 9% of the time. 🔍 91% of answers had no verifiable source trail. 🔍 When citations did appear, they averaged 3.78 sources — many from Wikipedia, CNBC, AP News, Time, NerdWallet, etc. TaxGPT averages 14 authoritative sources per answer, all from primary/secondary tax law — IRC, Treasury Regs, court cases, IRS rulings, and our proprietary tax knowledge library. More details in our official blog post in the following post.
1:28

Why General AI Fails Tax Professionals — and What Makes TaxGPT State-of-the-Art Tax LLM Every tax professional using general-purpose AI tools like ChatGPT or Claude for research is one hallucinated source away from costly damage to their clients and professional career. So we ran a head-to-head benchmark: TaxGPT vs. OpenAI, Anthropic, and Gemini. Our methodology: 1️⃣ Quantitative testing on CPA + EA tax-specific exam questions 2️⃣ Qualitative testing complex, real-world scenarios across trust & estate, direct and indirect taxes, payroll, multi-state nexus, advisory, and compliance Results (Accuracy): TaxGPT: 96% Gemini 2.5 Pro: 89% Claude Sonnet 4: 87% GPT-4o: 80% GPT5: 92% But here’s the real story 👇 The Source Quality Gap General LLMs are great test takers. But tax research isn’t a trivia game — it’s a liability-driven profession where every position must be defensible. When we examined the sources behind their answers, the gap became a canyon: 🔍 OpenAI included citations only 9% of the time. 🔍 91% of answers had no verifiable source trail. 🔍 When citations did appear, they averaged 3.78 sources — many from Wikipedia, CNBC, AP News, Time, NerdWallet, etc. TaxGPT averages 14 authoritative sources per answer, all from primary/secondary tax law — IRC, Treasury Regs, court cases, IRS rulings, and our proprietary tax knowledge library. More details in our official blog post in the following post.

Kash from TaxGPT.com

59,253 Aufrufe • vor 8 Monaten

Do Vision-Language Models represent space, and how? Spatial terms like "left" or "right" may not be enough to match images with spatial descriptions, as we often overlook the different frames of reference (FoR) used by speakers and listeners. See Figure 1 for examples! Introducing the COnsistent Multilingual Frame Of Reference Test (COMFORT), an evaluation protocol to assess the spatial reasoning capabilities of VLMs. COMFORT includes systematically designed datasets and metrics that evaluate model performance, and their deeper linguistic competence, specifically the spatial knowledge encoded in their internal representations. Find out more in the video teaser! Almost all VLMs prefer the egocentric relative FoR with reflected transform, similar to English. Yet, we reveal significant shortcomings of VLMs: notably, the models (1) exhibit poor robustness and consistency, (2) lack the flexibility to accommodate multiple FoRs, and (3) fail to adhere to language-specific or culture-specific conventions in cross-lingual tests, as English tends to dominate other languages. A shortened version will appear in Pluralistic Alignment Workshop Pluralistic Alignment Workshop #NeurIPS2024. It seems that the ArXiv moderators put it on hold and are eager to give it a thorough read first🤣! So here is the Paper/Code/Data: This collaboration turns out to be amazing, jointly led by Brian Zheyuan Zhang, @Hu_FY_ Jayjun Lee, with so many contributions and insights from Freda Shi, Parisa Kordjamshidi Michigan SLED Lab. With a growing effort to align vision-language models with human cognitive intuitions, we call for more attention to the ambiguous nature and cross-cultural diversity of spatial reasoning!
0:30

Do Vision-Language Models represent space, and how? Spatial terms like "left" or "right" may not be enough to match images with spatial descriptions, as we often overlook the different frames of reference (FoR) used by speakers and listeners. See Figure 1 for examples! Introducing the COnsistent Multilingual Frame Of Reference Test (COMFORT), an evaluation protocol to assess the spatial reasoning capabilities of VLMs. COMFORT includes systematically designed datasets and metrics that evaluate model performance, and their deeper linguistic competence, specifically the spatial knowledge encoded in their internal representations. Find out more in the video teaser! Almost all VLMs prefer the egocentric relative FoR with reflected transform, similar to English. Yet, we reveal significant shortcomings of VLMs: notably, the models (1) exhibit poor robustness and consistency, (2) lack the flexibility to accommodate multiple FoRs, and (3) fail to adhere to language-specific or culture-specific conventions in cross-lingual tests, as English tends to dominate other languages. A shortened version will appear in Pluralistic Alignment Workshop Pluralistic Alignment Workshop #NeurIPS2024. It seems that the ArXiv moderators put it on hold and are eager to give it a thorough read first🤣! So here is the Paper/Code/Data: This collaboration turns out to be amazing, jointly led by Brian Zheyuan Zhang, @Hu_FY_ Jayjun Lee, with so many contributions and insights from Freda Shi, Parisa Kordjamshidi Michigan SLED Lab. With a growing effort to align vision-language models with human cognitive intuitions, we call for more attention to the ambiguous nature and cross-cultural diversity of spatial reasoning!

Martin Ziqiao Ma

35,565 Aufrufe • vor 1 Jahr

One of the biggest airdrops in Hedera history?👀🪂🌱 Hey! If you've made it to this tweet as fast as Hedera's TPS, bookmark it so you don't miss any details While $IVY skydives with some friends in the video, let's check out the conditions: 🔥First of all, let's talk about how it works🔥 The airdrop will have a tier system where you could get more or less IVY depending on your tier. There will be four different tiers, and we'll announce the conditions for each tier to move up. Here's how much you can get: ➡️Tier 1: 20,000 IVY tokens ➡️Tier 2: 40,000 IVY tokens ➡️Tier 3: 60,000 IVY tokens ➡️Tier 4: 80,000 IVY tokens 🧐How do you qualify for each tier? We can't spill all the beans now, but we'll give you the conditions on our X account. For now, IVY can give you a sneak peek: ✅Tier 1 Requirements: Hold all 3 NFTs ✅Tier 2 Requirements: Tier 1 + hold $100 worth of IVY tokens (at time of snapshot) ✅Tier 3 Requirements: Tier 1 + Tier 2 + ???? ✅Tier 4 Requirements: Tier 1 + Tier 2 + Tier 3 + ???? 📸Chill out! The snapshot hasn't been taken yet, so you'll know in due time what to do to aim for the higher tiers. And to wrap it up... we'll be giving away 3 NFT sets (1 set = NFT 1 + NFT 2 + NFT 3) to those who comment and retweet this tweet, tagging 3 friends 🌱🫂 Good luck! We'll announce the winners on Monday. 🚨Oh... one more IMPORTANT detail...🚨 If you have multiple NFT sets, your IVY amount multiplies. For example, if you have 3 complete sets of NFTs and qualify as Tier 3, your amount of IVY to receive in the airdrop will be: ➡️Tier 3 = 60,000 IVY tokens ✅NFT Sets = 3 Total airdrop amount = 60,000 IVY tokens x 3 sets = 180,000 IVY tokens The maximum multiplier will be x5 If you have any questions, you can hop into our Telegram channel to ask ❓ You'll find the link in the linktree on our X account See yaa, $IVY fam Built on Hedera.
0:43

One of the biggest airdrops in Hedera history?👀🪂🌱 Hey! If you've made it to this tweet as fast as Hedera's TPS, bookmark it so you don't miss any details While $IVY skydives with some friends in the video, let's check out the conditions: 🔥First of all, let's talk about how it works🔥 The airdrop will have a tier system where you could get more or less IVY depending on your tier. There will be four different tiers, and we'll announce the conditions for each tier to move up. Here's how much you can get: ➡️Tier 1: 20,000 IVY tokens ➡️Tier 2: 40,000 IVY tokens ➡️Tier 3: 60,000 IVY tokens ➡️Tier 4: 80,000 IVY tokens 🧐How do you qualify for each tier? We can't spill all the beans now, but we'll give you the conditions on our X account. For now, IVY can give you a sneak peek: ✅Tier 1 Requirements: Hold all 3 NFTs ✅Tier 2 Requirements: Tier 1 + hold $100 worth of IVY tokens (at time of snapshot) ✅Tier 3 Requirements: Tier 1 + Tier 2 + ???? ✅Tier 4 Requirements: Tier 1 + Tier 2 + Tier 3 + ???? 📸Chill out! The snapshot hasn't been taken yet, so you'll know in due time what to do to aim for the higher tiers. And to wrap it up... we'll be giving away 3 NFT sets (1 set = NFT 1 + NFT 2 + NFT 3) to those who comment and retweet this tweet, tagging 3 friends 🌱🫂 Good luck! We'll announce the winners on Monday. 🚨Oh... one more IMPORTANT detail...🚨 If you have multiple NFT sets, your IVY amount multiplies. For example, if you have 3 complete sets of NFTs and qualify as Tier 3, your amount of IVY to receive in the airdrop will be: ➡️Tier 3 = 60,000 IVY tokens ✅NFT Sets = 3 Total airdrop amount = 60,000 IVY tokens x 3 sets = 180,000 IVY tokens The maximum multiplier will be x5 If you have any questions, you can hop into our Telegram channel to ask ❓ You'll find the link in the linktree on our X account See yaa, $IVY fam Built on Hedera.

IVY

33,729 Aufrufe • vor 1 Jahr

🔬 Exciting News! Our manuscript, "scGPT: toward building a foundation model for single-cell multi-omics using generative AI" is now finally published in Nature Methods (Nature Methods) 🎉 !!! (Re-)Introducing scGPT: A transformative foundation model engineered for single-cell omics analysis. Developed through the analysis of over 33 million human cells, scGPT sets a new benchmark for application versatility, offering both fine-tuning and zero-shot capabilities. Since its preprint in May 2023, scGPT has significantly impacted the field, evidenced by 13K+ installations, 600+ GitHub stars 🌟, and 40+ citations before its official publication! scGPT has been validated by numerous benchmark studies as a leading foundation model in single-cell analysis. Its pre-trained embeddings extend its utility beyond single-cell studies, enhancing a variety of downstream tasks including protein enrichment and genetic perturbation predictions. Some key updates lately: ---Expanded zero-shot applications for efficient reference mapping and integration, now with CellXGene census integration. ---Advanced perturbation analysis capabilities, including genome-scale perturb-seq data analysis and bulk sequencing data generalization. ---Upgraded scGPT package, offering versatile model loading compatible with PyTorch and flash-attn, for both GPU and CPU. ---Cloud-based scGPT applications for reference mapping, cell annotation, and gene regulatory network inference are available on ---Integration with Hugging Face for easier model training. Limitations: scGPT is an early foray into foundation models for single-cell omics, facing challenges like limited zero-shot learning in some tasks, pretraining constraints, data quality issues, and evaluation limitations. See our Supplementary Notes for details. 🚀 Future Work? Short-Term Goals: 1. Releasing a Mouse Model for broader analysis. 2. Developing a comprehensive evaluation suite for foundation models in single-cell analysis. 3. Creating a foundation model for single-cell spatial omics. 4. Enhancing zero-shot capacity by integrating scGPT with RAG (e.g., knowledge graphs). Long-Term Goals: 1. Expanding scGPT for comprehensive single-cell multi-omics analysis. 2. Developing an in-silico perturbation model for predicting genetic perturbation effects. 3. Merging scGPT with multi-modal genomic sequence models for a deeper understanding of cell biology. 📚 Access the paper on Nature Methods: 🔬Preprint in Bioarixv: 💻 All our codes/data/weights are open source: Wholehearted congratulations to all the authors, especially the two co-first authors, Haotian (Haotian Cui ) and Chloe (ChloeXWang), who are really the emerging superstars in AI and biology! Vector Institute Peter Munk Cardiac Centre AI U of T Department of Computer Science Department of Laboratory Medicine & Pathobiology University Health Network University of Toronto #scGPT #GenerativeAI #AI4Science #Combio #opensource
0:52

🔬 Exciting News! Our manuscript, "scGPT: toward building a foundation model for single-cell multi-omics using generative AI" is now finally published in Nature Methods (Nature Methods) 🎉 !!! (Re-)Introducing scGPT: A transformative foundation model engineered for single-cell omics analysis. Developed through the analysis of over 33 million human cells, scGPT sets a new benchmark for application versatility, offering both fine-tuning and zero-shot capabilities. Since its preprint in May 2023, scGPT has significantly impacted the field, evidenced by 13K+ installations, 600+ GitHub stars 🌟, and 40+ citations before its official publication! scGPT has been validated by numerous benchmark studies as a leading foundation model in single-cell analysis. Its pre-trained embeddings extend its utility beyond single-cell studies, enhancing a variety of downstream tasks including protein enrichment and genetic perturbation predictions. Some key updates lately: ---Expanded zero-shot applications for efficient reference mapping and integration, now with CellXGene census integration. ---Advanced perturbation analysis capabilities, including genome-scale perturb-seq data analysis and bulk sequencing data generalization. ---Upgraded scGPT package, offering versatile model loading compatible with PyTorch and flash-attn, for both GPU and CPU. ---Cloud-based scGPT applications for reference mapping, cell annotation, and gene regulatory network inference are available on ---Integration with Hugging Face for easier model training. Limitations: scGPT is an early foray into foundation models for single-cell omics, facing challenges like limited zero-shot learning in some tasks, pretraining constraints, data quality issues, and evaluation limitations. See our Supplementary Notes for details. 🚀 Future Work? Short-Term Goals: 1. Releasing a Mouse Model for broader analysis. 2. Developing a comprehensive evaluation suite for foundation models in single-cell analysis. 3. Creating a foundation model for single-cell spatial omics. 4. Enhancing zero-shot capacity by integrating scGPT with RAG (e.g., knowledge graphs). Long-Term Goals: 1. Expanding scGPT for comprehensive single-cell multi-omics analysis. 2. Developing an in-silico perturbation model for predicting genetic perturbation effects. 3. Merging scGPT with multi-modal genomic sequence models for a deeper understanding of cell biology. 📚 Access the paper on Nature Methods: 🔬Preprint in Bioarixv: 💻 All our codes/data/weights are open source: Wholehearted congratulations to all the authors, especially the two co-first authors, Haotian (Haotian Cui ) and Chloe (ChloeXWang), who are really the emerging superstars in AI and biology! Vector Institute Peter Munk Cardiac Centre AI U of T Department of Computer Science Department of Laboratory Medicine & Pathobiology University Health Network University of Toronto #scGPT #GenerativeAI #AI4Science #Combio #opensource

Bo Wang

199,657 Aufrufe • vor 2 Jahren

🚨 Open-source AI just got serious. Alibaba just released Qwen 3.5-Plus 🤯 397B parameters. Only 17B active per inference. Performance comparable to Gemini 3 Pro — at just $0.8 per million tokens (1/18 the cost). But here’s what makes it different 👇 This isn’t just another benchmark drop. In the demo, Qwen runs directly inside the terminal: ➙ Reviews GitHub PRs ➙ Fetches diffs automatically ➙ Runs CLI & smoke tests ➙ Generates integration tests ➙ Executes /context ➙ Shows real-time token usage breakdown That’s not a chatbot. That’s a real developer workflow. Under the hood: ⚡ 60% lower deployment memory ⚡ Inference throughput increased by 8.6× (32K context) ⚡ Inference throughput increased by up to 19× (256K ultra-long context) ⚡ Native multimodal model ⚡ Supports direct video input (up to 2 hours) And the ecosystem is massive: ✔ 201 languages ✔ 250K vocabulary ✔ 400+ open-sourced models ✔ 1B+ downloads ✔ 200K+ derivative models Available now via Qwen App, MoDa Community, Hugging Face, and Alibaba Cloud Bailian. ✅ The AI race isn’t about who’s biggest anymore. → It’s about who’s usable. 🔥 Watch the demo 👇 #Qwen #OpenSourceAI #AI #DevTools
1:10

🚨 Open-source AI just got serious. Alibaba just released Qwen 3.5-Plus 🤯 397B parameters. Only 17B active per inference. Performance comparable to Gemini 3 Pro — at just $0.8 per million tokens (1/18 the cost). But here’s what makes it different 👇 This isn’t just another benchmark drop. In the demo, Qwen runs directly inside the terminal: ➙ Reviews GitHub PRs ➙ Fetches diffs automatically ➙ Runs CLI & smoke tests ➙ Generates integration tests ➙ Executes /context ➙ Shows real-time token usage breakdown That’s not a chatbot. That’s a real developer workflow. Under the hood: ⚡ 60% lower deployment memory ⚡ Inference throughput increased by 8.6× (32K context) ⚡ Inference throughput increased by up to 19× (256K ultra-long context) ⚡ Native multimodal model ⚡ Supports direct video input (up to 2 hours) And the ecosystem is massive: ✔ 201 languages ✔ 250K vocabulary ✔ 400+ open-sourced models ✔ 1B+ downloads ✔ 200K+ derivative models Available now via Qwen App, MoDa Community, Hugging Face, and Alibaba Cloud Bailian. ✅ The AI race isn’t about who’s biggest anymore. → It’s about who’s usable. 🔥 Watch the demo 👇 #Qwen #OpenSourceAI #AI #DevTools

Toha Khan

123,375 Aufrufe • vor 5 Monaten