正在加载视频...

视频加载失败

加载此视频时出现问题。这可能是由于临时网络问题，或视频可能不可用。

Meta's Joe Spisak explains how AI models can train themselves by generating images, asking itself questions about them, and choosing the best answers, in order to move beyond human data and human fine-tuning, and teach itself from synthetic data

Tsarathustra

83,522 subscribers

42,390 次观看 • 1 年前 •via X (Twitter)

Anya Rossi• Live Now

Private livecam show

10 条评论

Tsarathustra 的头像

Tsarathustra1 年前

Source (thanks to @curiousgangsta):

Bunagaya 的头像

Bunagaya1 年前

doesn't feel like that much time left before recursive self improvement

Rob Leclerc 的头像

Rob Leclerc1 年前

We do this as well, but we don’t get far without testing our hypotheses against reality. AI can’t side step this epistemological constraint no more than we can.

Narasimha R N 🤗 AI - Disciple @69 的头像

Narasimha R N 🤗 AI - Disciple @691 年前

AI self-training is revolutionary. Synthetic data bypasses human limits, speeding progress. However, guardrails are crucial to ensure ethical outcomes and prevent unintended consequences.

Chandan Ganwani 的头像

Chandan Ganwani1 年前

Hallucinations squared and compounded exponentially. This is brilliant!

Rob Parker 的头像

Rob Parker1 年前

Human limitations

𝕏one - exo/acc 🧙🏽‍♂️🎨 的头像

𝕏one - exo/acc 🧙🏽‍♂️🎨1 年前

Self play is the correct way forward.

Shawn 的头像

Shawn1 年前

Sounds like Meta is closing up OAIs moat

Tsarathustra 的头像

Tsarathustra1 年前

have you tried Meta AI's voice model? It's pretty good. I think it's only a 1B/3B model based on Llama 3.2

Timi Okoya 的头像

Timi Okoya1 年前

This allows AI to explore beyond the limitations of human-labeled data, increasing its ability to learn, adapt, & improve autonomously. It opens up possibilities for faster, more scalable training, enabling AI systems to tackle complex tasks with minimal human intervention.

相关视频

Revolutionizing Move Programming with OpenLedger In this demo, we showcase how Move datasets contributed by data providers to OpenLedger’s datanets are used to fine-tune specialized models with LoRA fine-tuning. As seen in the video, we showcase an example on how builders can deploy a Move-specialized model that powers Co-pilot agents using our no-code model fine-tuning platform. This is the future of AI and Web3 innovation. Watch this space to see more specialised models and data feeds being built for next generation agents on top of OpenLedger #Move

Revolutionizing Move Programming with OpenLedger In this demo, we showcase how Move datasets contributed by data providers to OpenLedger’s datanets are used to fine-tune specialized models with LoRA fine-tuning. As seen in the video, we showcase an example on how builders can deploy a Move-specialized model that powers Co-pilot agents using our no-code model fine-tuning platform. This is the future of AI and Web3 innovation. Watch this space to see more specialised models and data feeds being built for next generation agents on top of OpenLedger #Move

OpenLedger

61,662 次观看 • 1 年前

Intelligence Factory is building human intelligence for robots. They train general-purpose manipulation models on human demonstration data (vision, action, and touch), then deploy them in warehouses, grocery stores, and data centers. Congrats on the launch, Yash and Jalaj!

Intelligence Factory is building human intelligence for robots. They train general-purpose manipulation models on human demonstration data (vision, action, and touch), then deploy them in warehouses, grocery stores, and data centers. Congrats on the launch, Yash and Jalaj!

Y Combinator

43,947 次观看 • 1 个月前

As AI becomes more capable and agentic, models themselves become more of a commodity, and all value gets created by how you steer, ground, and finetune these models with your business data and workflow — and how they compose with the UI layer of human to AI to human interaction. That’s why today we are announcing wave two of Microsoft 365 Copilot.

As AI becomes more capable and agentic, models themselves become more of a commodity, and all value gets created by how you steer, ground, and finetune these models with your business data and workflow — and how they compose with the UI layer of human to AI to human interaction. That’s why today we are announcing wave two of Microsoft 365 Copilot.

Satya Nadella

624,278 次观看 • 1 年前

Cohere CEO Aidan Gomez says when human data is too expensive, AI learns to scale itself. Why hire 100,000 doctors when 100 will do? Train on trusted signals -- then generate 1,000× synthetic lookalikes. The result? Scalable expertise, born from mimicry.

Cohere CEO Aidan Gomez says when human data is too expensive, AI learns to scale itself. Why hire 100,000 doctors when 100 will do? Train on trusted signals -- then generate 1,000× synthetic lookalikes. The result? Scalable expertise, born from mimicry.

vitrupo

40,704 次观看 • 1 年前

Amjad Masad says true AGI means an AI that can enter new situations and quickly learn how to achieve goals Even the latest models still struggle with tasks outside their training data. The real breakthrough will come when AI can train itself -- like AlphaGo playing billions of games against itself

Amjad Masad says true AGI means an AI that can enter new situations and quickly learn how to achieve goals Even the latest models still struggle with tasks outside their training data. The real breakthrough will come when AI can train itself -- like AlphaGo playing billions of games against itself

Haider.

163,410 次观看 • 1 年前

Satya Nadella predicts, - if you have AI + Quantum Computing, you may use quantum to generate synthetic data - AI can then use that data to train better models for complex fields like chemistry and physics.

Satya Nadella predicts, - if you have AI + Quantum Computing, you may use quantum to generate synthetic data - AI can then use that data to train better models for complex fields like chemistry and physics.

Haider.

78,677 次观看 • 1 年前

Dario Amodei says static data is no longer the most central thing in AI development Models are now relying on 'dynamic data', learning through trial and error by generating their own synthetic data in reinforcement environments "as AI creates its own experience, the open web matters less"

Dario Amodei says static data is no longer the most central thing in AI development Models are now relying on 'dynamic data', learning through trial and error by generating their own synthetic data in reinforcement environments "as AI creates its own experience, the open web matters less"

Haider.

16,326 次观看 • 4 个月前

New short course on Fine-tuning LLMs! Many developers are moving beyond only prompting, to also fine-tuning LLMs - that is, taking a pre-trained model and training it further on your own data, which can deliver superior results inexpensively. In this course, Sharon Zhou, CEO of Lamini (disclosure: I’m a minor shareholder) shows you how to recognize when fine-tuning can be help, and how to train an open-source LLM on your own data. I hope you enjoy the course!

New short course on Fine-tuning LLMs! Many developers are moving beyond only prompting, to also fine-tuning LLMs - that is, taking a pre-trained model and training it further on your own data, which can deliver superior results inexpensively. In this course, Sharon Zhou, CEO of Lamini (disclosure: I’m a minor shareholder) shows you how to recognize when fine-tuning can be help, and how to train an open-source LLM on your own data. I hope you enjoy the course!

Andrew Ng

502,781 次观看 • 2 年前

Have a sneak preview of our revolutionary $AIT Data Annotation Platform designed to train #AI models 🤖 Join us and see how you can answer questions, validate data, and earn rewards on our platform 👀 👉

Have a sneak preview of our revolutionary $AIT Data Annotation Platform designed to train #AI models 🤖 Join us and see how you can answer questions, validate data, and earn rewards on our platform 👀 👉

AIT Protocol

715,615 次观看 • 2 年前

Excited to introduce DeepSeek R1 fine-tuning with Firecrawl 🔥 Learn how to fine-tune DeepSeek's R1 model on custom instruction datasets generated by Firecrawl's web data extraction. Train your own AI expert to answer domain-specific questions about your website!

Excited to introduce DeepSeek R1 fine-tuning with Firecrawl 🔥 Learn how to fine-tune DeepSeek's R1 model on custom instruction datasets generated by Firecrawl's web data extraction. Train your own AI expert to answer domain-specific questions about your website!

Eric Ciarla (hiring)

40,189 次观看 • 1 年前

Robots are the bottleneck in scaling robotics, and learning from human video promises to solve it. But how can chaotic human data ever measure up to sanitized, lab-made teleoperation data? Introducing Do as I Do: establishing a much needed correspondence between human videos and dexterous robot data. Some fun insights below: 🧵

Robots are the bottleneck in scaling robotics, and learning from human video promises to solve it. But how can chaotic human data ever measure up to sanitized, lab-made teleoperation data? Introducing Do as I Do: establishing a much needed correspondence between human videos and dexterous robot data. Some fun insights below: 🧵

Mahi Shafiullah 🏠🤖

90,905 次观看 • 19 天前

꧁IP꧂ encompasses all data used to fuel AI models. On Story, anyone can register open source AI models, data sets, and fine tune their models. AI cannot exist without IP, and Story makes IP programmable.

꧁IP꧂ encompasses all data used to fuel AI models. On Story, anyone can register open source AI models, data sets, and fine tune their models. AI cannot exist without IP, and Story makes IP programmable.

The DATA Foundation

155,437 次观看 • 1 年前

Open models. Open data. Open tools. Jensen Huang explains how building AI in the open is key to trust and global progress. Learn more about our open models →

Open models. Open data. Open tools. Jensen Huang explains how building AI in the open is key to trust and global progress. Learn more about our open models →

NVIDIA AI

16,554 次观看 • 5 个月前

Learn to train an LLM with distributed data while ensuring privacy using federated learning in a new two-part short course, Intro to Federated Learning and Federated Fine-tuning of LLMs with Private Data, created with Flower and taught by Daniel J. Beutel and nic lane. Federated learning allows a single model to be trained across multiple devices, such as phones, or multiple organizations, such as hospitals, without the need to share data to a central server. This two-part course gives you an introduction to federated learning, and then teaches you how to fine-tune your large language model with distributed data using Flower Lab’s open source federated learning framework. You’ll learn: - How to use federated learning to train a variety of models, ranging from speech and vision models to LLMs, across distributed data while offering data privacy options to users and organizations. - Privacy Enhancing Technologies like differential privacy (DP), which obscures individual data by adding calibrated noise to query results. - Two variants of differential privacy - Central and Local - and how to choose depending on your use case. - How to measure and decrease bandwidth usage to make federated learning more practical and efficient with techniques like using pre-trained models and Parameter-Efficient Fine-Tuning - How federated LLM fine-tuning reduces the risk of leaking training data. Sign up here!

Learn to train an LLM with distributed data while ensuring privacy using federated learning in a new two-part short course, Intro to Federated Learning and Federated Fine-tuning of LLMs with Private Data, created with Flower and taught by Daniel J. Beutel and nic lane. Federated learning allows a single model to be trained across multiple devices, such as phones, or multiple organizations, such as hospitals, without the need to share data to a central server. This two-part course gives you an introduction to federated learning, and then teaches you how to fine-tune your large language model with distributed data using Flower Lab’s open source federated learning framework. You’ll learn: - How to use federated learning to train a variety of models, ranging from speech and vision models to LLMs, across distributed data while offering data privacy options to users and organizations. - Privacy Enhancing Technologies like differential privacy (DP), which obscures individual data by adding calibrated noise to query results. - Two variants of differential privacy - Central and Local - and how to choose depending on your use case. - How to measure and decrease bandwidth usage to make federated learning more practical and efficient with techniques like using pre-trained models and Parameter-Efficient Fine-Tuning - How federated LLM fine-tuning reduces the risk of leaking training data. Sign up here!

Andrew Ng

64,558 次观看 • 1 年前

Human generated data has fueled incredible AI progress, but what comes next? 📈 On the latest episode of our podcast, Hannah Fry and David Silver, VP of Reinforcement Learning, talk about how we could move from the era of relying on human data to one where AI could learn for itself. Watch now → 00:00 Introduction 01:50 Era of experience 03:45 AlphaZero 10:19 Move 37 15:20 Reinforcement learning and human feedback 24:30 AlphaProof 29:50 Math Olympiads 35:00 Experience based methods 42:56 Hannah's reflections 44:00 Fan Hui joins

Human generated data has fueled incredible AI progress, but what comes next? 📈 On the latest episode of our podcast, Hannah Fry and David Silver, VP of Reinforcement Learning, talk about how we could move from the era of relying on human data to one where AI could learn for itself. Watch now → 00:00 Introduction 01:50 Era of experience 03:45 AlphaZero 10:19 Move 37 15:20 Reinforcement learning and human feedback 24:30 AlphaProof 29:50 Math Olympiads 35:00 Experience based methods 42:56 Hannah's reflections 44:00 Fan Hui joins

Google DeepMind

226,305 次观看 • 1 年前

Meta presents Sapiens Foundation for Human Vision Models discuss: We present Sapiens, a family of models for four fundamental human-centric vision tasks - 2D pose estimation, body-part segmentation, depth estimation, and surface normal prediction. Our models natively support 1K high-resolution inference and are extremely easy to adapt for individual tasks by simply fine-tuning models pretrained on over 300 million in-the-wild human images. We observe that, given the same computational budget, self-supervised pretraining on a curated dataset of human images significantly boosts the performance for a diverse set of human-centric tasks. The resulting models exhibit remarkable generalization to in-the-wild data, even when labeled data is scarce or entirely synthetic. Our simple model design also brings scalability - model performance across tasks improves as we scale the number of parameters from 0.3 to 2 billion. Sapiens consistently surpasses existing baselines across various human-centric benchmarks. We achieve significant improvements over the prior state-of-the-art on Humans-5K (pose) by 7.6 mAP, Humans-2K (part-seg) by 17.1 mIoU, Hi4D (depth) by 22.4% relative RMSE, and THuman2 (normal) by 53.5% relative angular error.

Meta presents Sapiens Foundation for Human Vision Models discuss: We present Sapiens, a family of models for four fundamental human-centric vision tasks - 2D pose estimation, body-part segmentation, depth estimation, and surface normal prediction. Our models natively support 1K high-resolution inference and are extremely easy to adapt for individual tasks by simply fine-tuning models pretrained on over 300 million in-the-wild human images. We observe that, given the same computational budget, self-supervised pretraining on a curated dataset of human images significantly boosts the performance for a diverse set of human-centric tasks. The resulting models exhibit remarkable generalization to in-the-wild data, even when labeled data is scarce or entirely synthetic. Our simple model design also brings scalability - model performance across tasks improves as we scale the number of parameters from 0.3 to 2 billion. Sapiens consistently surpasses existing baselines across various human-centric benchmarks. We achieve significant improvements over the prior state-of-the-art on Humans-5K (pose) by 7.6 mAP, Humans-2K (part-seg) by 17.1 mIoU, Hi4D (depth) by 22.4% relative RMSE, and THuman2 (normal) by 53.5% relative angular error.

AK

151,511 次观看 • 1 年前

Liva AI (@thelivaAI) provides high-quality human voice and video data for companies building realistic AI models. Congrats on the launch, aoi and Ashley!

Liva AI (@thelivaAI) provides high-quality human voice and video data for companies building realistic AI models. Congrats on the launch, aoi and Ashley!

Y Combinator

43,106 次观看 • 10 个月前

From airplane wings to Dyson spheres, Paul Eremenko explains why physical engineering AI has lagged behind coding AI, and how @p_1_ai's approach to synthetic training data could change everything.

From airplane wings to Dyson spheres, Paul Eremenko explains why physical engineering AI has lagged behind coding AI, and how @p_1_ai's approach to synthetic training data could change everything.

Sonya Huang 🐥

27,686 次观看 • 1 年前

Train the AI behind real-world robots We’re hiring Data Labelers to annotate images and videos to train Optimus & autonomous systems. You’ll work with real production data & directly impact how these systems learn and operate in the real world Come join!

Train the AI behind real-world robots We’re hiring Data Labelers to annotate images and videos to train Optimus & autonomous systems. You’ll work with real production data & directly impact how these systems learn and operate in the real world Come join!

Tesla Recruiting

642,569 次观看 • 2 个月前