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。

True computer use is fully general. FDM-1 uses arrow keys on a computer to steer a car in San Francisco with less than 1 hour of fine-tuning data. The action policy is critical: tuning FDM-1 to drive gets much higher accuracy than tuning just the video encoder on the... show more

Standard Intelligence

10,412 subscribers

72,372 次观看 • 6 个月前 •via X (Twitter)

Anya Rossi• Live Now

Private livecam show

14 条评论

Standard Intelligence 的头像

Standard Intelligence6 个月前

FDM-1 completes complex tasks and navigates interfaces well enough to use CAD applications. Forking VMs allow us to snapshot when a successful operation completes (extrusion, selection, etc.), letting us apply test-time compute to computer use.

Standard Intelligence 的头像

Standard Intelligence6 个月前

Inspired by VPT (Baker et al), we train an inverse dynamics model (IDM) to predict frame-by-frame computer actions. The IDM leverages 40k hours of contractor data to label 11 million hours of screen recordings—550,000x larger than the largest open-source computer use dataset.

Standard Intelligence 的头像

Standard Intelligence6 个月前

We’ve made two main advances: the ability to train on our 11M+ hour computer action dataset and understand long-context video. Our video encoder can fit nearly two hours of 30FPS, high-resolution video into a 1M token context window, ~50x more efficient than existing SOTA.

Standard Intelligence 的头像

Standard Intelligence6 个月前

Computer use models shouldn't learn from screenshots. We built a new foundation model that learns from video like humans do. FDM-1 can construct a gear in Blender, find software bugs, and even drive a real car through San Francisco using arrow keys.

Standard Intelligence 的头像

Standard Intelligence6 个月前

Here’s the blog post, where you can learn more about how we trained this model:

agniv 的头像

agniv6 个月前

HOLY DUCK

_____ 的头像

_____6 个月前

amazing work! any idea why the baseline for the self driving task starts out worse than random guessing though?

Yuze 的头像

Yuze6 个月前

fascinating work on the context compression part! Wondering if the model would be able to know what is a good quality action at system 2 level. eg. when to raise/fall on playing card games 😀

Yuval Grossman 的头像

Yuval Grossman6 个月前

Can you explain what do you mean here by “tuning the video encoder”?

Samu 的头像

Samu6 个月前

nice

madhavan 的头像

madhavan6 个月前

holy fire

Gary Basin 的头像

Gary Basin6 个月前

> less than 1 hour of fine-tuning data

FallenOne 的头像

FallenOne6 个月前

that's so cool

Haris Poljo 的头像

Haris Poljo6 个月前

@comma_ai

相关视频

The countdown is on—less than 90 days until championship week at Oakmont! From fine-tuning agronomic practices to dialing in playing conditions, the team is laser-focused on delivering a world-class test of golf.

The countdown is on—less than 90 days until championship week at Oakmont! From fine-tuning agronomic practices to dialing in playing conditions, the team is laser-focused on delivering a world-class test of golf.

Tyler Bloom

44,468 次观看 • 1 年前

OpenAI engineer explained every form of fine-tuning in 1 hour 46 minutes - better than $3000 fine-tuning bootcamps. pick your task -> try prompting first -> drop to SFT for style -> run DPO for preference -> run RFT for verifiable rewards -> ship a model 10x cheaper than GPT-5. That loop is why teams that fine-tune beat teams that only prompt on cost and speed. Prompting + SFT + DPO + RFT + OpenAI fine-tuning API - that's the stack. Watch and save it, then pick the right fine-tune for your task this week.

OpenAI engineer explained every form of fine-tuning in 1 hour 46 minutes - better than $3000 fine-tuning bootcamps. pick your task -> try prompting first -> drop to SFT for style -> run DPO for preference -> run RFT for verifiable rewards -> ship a model 10x cheaper than GPT-5. That loop is why teams that fine-tune beat teams that only prompt on cost and speed. Prompting + SFT + DPO + RFT + OpenAI fine-tuning API - that's the stack. Watch and save it, then pick the right fine-tune for your task this week.

h100envy

41,245 次观看 • 2 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

502,941 次观看 • 3 年前

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

61,693 次观看 • 1 年前

The era of Stable Intelligence is here 🤖 Tether’s QVAC Fabric just released the world’s first cross-platform 1-bit LLM LoRA fine-tuning framework. QVAC Fabric extends Microsoft's ultra-efficient BitNet architecture, allowing fine-tuning and inference of LLMs directly on your smartphone—no NVIDIA GPU/CUDA lock-in or expensive server required. The Breakthrough: - Total Sovereignty: LoRA fine-tune ultra-efficient models locally on any smartphone, including iPhones, Pixel phones, Samsung Galaxy phones and any desktop/laptop operating systems using Vulkan and Metal backends. - Extreme Efficiency: 1-bit architecture uses up to 90% less memory and runs up to 11x faster than traditional models. - Universal Access: What used to require a data center now runs on the chip in your pocket. Own your intelligence. The era of stable, local AI is here. 📱🧠 Read the full details on Hugging Face and grab the binaries to build on your own hardware.

The era of Stable Intelligence is here 🤖 Tether’s QVAC Fabric just released the world’s first cross-platform 1-bit LLM LoRA fine-tuning framework. QVAC Fabric extends Microsoft's ultra-efficient BitNet architecture, allowing fine-tuning and inference of LLMs directly on your smartphone—no NVIDIA GPU/CUDA lock-in or expensive server required. The Breakthrough: - Total Sovereignty: LoRA fine-tune ultra-efficient models locally on any smartphone, including iPhones, Pixel phones, Samsung Galaxy phones and any desktop/laptop operating systems using Vulkan and Metal backends. - Extreme Efficiency: 1-bit architecture uses up to 90% less memory and runs up to 11x faster than traditional models. - Universal Access: What used to require a data center now runs on the chip in your pocket. Own your intelligence. The era of stable, local AI is here. 📱🧠 Read the full details on Hugging Face and grab the binaries to build on your own hardware.

QVAC

108,875 次观看 • 6 个月前

In 1983, singer Barry White urged people to learn how to code. At the time, less than 1% of the population owned a computer.

In 1983, singer Barry White urged people to learn how to code. At the time, less than 1% of the population owned a computer.

Jon Erlichman

344,967 次观看 • 1 年前

Today we're thrilled to announce the first end-to-end platform for Reinforcement Fine-Tuning. With just a dozen labeled data points, you can outperform #OpenAI o1 and #DeepSeekR1 on complex tasks. Built on the #GRPO methodology that DeepSeek-R1 popularized, our platform delivers exceptional results. In our real-world PyTorch to Triton transpilation case study, we achieved 3x higher accuracy than OpenAI o1 and DeepSeek-R1 when writing GPU code. Check out the thread below to learn how you can adapt an #opensource #LLM to your use cases with unmatched efficiency. #rft

Today we're thrilled to announce the first end-to-end platform for Reinforcement Fine-Tuning. With just a dozen labeled data points, you can outperform #OpenAI o1 and #DeepSeekR1 on complex tasks. Built on the #GRPO methodology that DeepSeek-R1 popularized, our platform delivers exceptional results. In our real-world PyTorch to Triton transpilation case study, we achieved 3x higher accuracy than OpenAI o1 and DeepSeek-R1 when writing GPU code. Check out the thread below to learn how you can adapt an #opensource #LLM to your use cases with unmatched efficiency. #rft

Predibase by Rubrik

1,022,546 次观看 • 1 年前

I got to ask Brett Adcock a couple of questions on Royden's livestream yesterday: - Data scarcity and compute constraint outlook - Pre-training vs. fine-tuning balance Brett believes generalization is the key, and Figure will be the first to solve it.

I got to ask Brett Adcock a couple of questions on Royden's livestream yesterday: - Data scarcity and compute constraint outlook - Pre-training vs. fine-tuning balance Brett believes generalization is the key, and Figure will be the first to solve it.

The Humanoid Hub

28,915 次观看 • 4 个月前

Announcing RoboFuME🤖💨, a system for autonomous & efficient real-world robot learning that 1. pre-trains a VLM reward model and a multi-task RL policy from diverse off-the-shelf demo data; 2. runs RL fine-tuning online with the VLM reward model. 🔗 🧵↓

Announcing RoboFuME🤖💨, a system for autonomous & efficient real-world robot learning that 1. pre-trains a VLM reward model and a multi-task RL policy from diverse off-the-shelf demo data; 2. runs RL fine-tuning online with the VLM reward model. 🔗 🧵↓

Jingyun Yang

88,294 次观看 • 2 年前

Vision-language models perform diverse tasks via in-context learning. Time for robots to do the same! Introducing In-Context Robot Transformer (ICRT): a robot policy that learns new tasks by prompting with robot trajectories, without any fine-tuning. [1/N]

Vision-language models perform diverse tasks via in-context learning. Time for robots to do the same! Introducing In-Context Robot Transformer (ICRT): a robot policy that learns new tasks by prompting with robot trajectories, without any fine-tuning. [1/N]

Max Fu

40,466 次观看 • 2 年前

New short course on LLMOps! LLMOps (large language model operations) is a rapidly developing field that takes ideas from MLOps (machine learning operations) and specializes them to building and deploying LLM-based applications. In this course, taught by Google Cloud's Erwin Huizenga, you'll learn to use automation to make building, tuning and deploying an LLM-based application less manual and more efficient. You'll learn how to: - Apply supervised fine-tuning to tune an LLM to a specific task - Automate and orchestrate LLM-tuning and deployment by customizing a pre-built tuning pipeline - Apply best practices for preparing training data for supervised fine-tuning of an LLM - Create an LLMOps workflow you can adapt to other LLM-tuning jobs This course doesn't assume any prior MLOps or LLMOps experience. Sign up here to learn about this emerging field!

New short course on LLMOps! LLMOps (large language model operations) is a rapidly developing field that takes ideas from MLOps (machine learning operations) and specializes them to building and deploying LLM-based applications. In this course, taught by Google Cloud's Erwin Huizenga, you'll learn to use automation to make building, tuning and deploying an LLM-based application less manual and more efficient. You'll learn how to: - Apply supervised fine-tuning to tune an LLM to a specific task - Automate and orchestrate LLM-tuning and deployment by customizing a pre-built tuning pipeline - Apply best practices for preparing training data for supervised fine-tuning of an LLM - Create an LLMOps workflow you can adapt to other LLM-tuning jobs This course doesn't assume any prior MLOps or LLMOps experience. Sign up here to learn about this emerging field!

Andrew Ng

221,803 次观看 • 2 年前

This is me talking to my computer without making a sound. After just a month of collecting data, our model is already approaching dictation in accuracy. We were surprised to see that it generalizes to unseen participants as well! (1/n)

This is me talking to my computer without making a sound. After just a month of collecting data, our model is already approaching dictation in accuracy. We were surprised to see that it generalizes to unseen participants as well! (1/n)

Vadims

581,578 次观看 • 2 个月前

New Course: Reinforcement Fine-Tuning LLMs with GRPO! Learn to use reinforcement learning to improve your LLM performance in this short course, built in collaboration with Predibase by Rubrik, and taught by Travis Addair, its Co-Founder and CTO, and Arnav Garg, its Senior Engineer and Machine Learning Lead. Reasoning models have been one of the most important developments in LLMs. Reinforcement Fine-Tuning (RFT) uses rewards to encourage LLMs to find solutions to multi-step reasoning tasks such as solving math problems and debugging code - without needing pre-existing training examples like in traditional supervised fine-tuning. Group Relative Policy Optimization (GRPO) is a reinforcement fine-tuning algorithm gaining rapid adoption. Developed by the DeepSeek team and used to train the R1 reasoning model, GRPO uses reward functions that you can write in Python to assign rewards to model responses. It’s beneficial for tasks with verifiable outcomes and can work well even with fewer than 100 training examples. It can also significantly improve the reasoning ability of smaller LLMs, making applications faster and more cost effective. In this course, you’ll take a technical deep dive into RFT with GRPO. You’ll learn to build reward functions that you can use in the GRPO training process to guide an LLM toward better performance on multi-step reasoning tasks. In detail, you’ll: - Learn when reinforcement fine-tuning is a better fit than supervised fine-tuning, especially for tasks involving multi-step reasoning or limited labeled data. - Understand how GRPO uses programmable reward functions as a more scalable alternative to the human feedback required for other reinforcement learning algorithms, such as RLHF and DPO. - Frame the Wordle game as a reinforcement fine-tuning problem and see how an LLM can learn to plan, analyze feedback, and improve its strategy over time. - Design reward functions that power the reinforcement fine-tuning process. - Learn techniques for evaluating more subjective tasks, such as rating the quality of a text summary, using an LLM as a judge. - Understand why reward hacking happens and how to avoid it by adding penalty functions to discourage undesirable behaviors. - Learn the four key components of the loss calculation in the GRPO algorithm: token probability distribution ratios, advantages, clipping, and KL-divergence. - Launch reinforcement fine-tuning jobs using Predibase’s hosted training services. By the end of this course, you’ll be able to build and fine-tune LLMs using reinforcement learning to improve reasoning without relying on large labeled datasets or subjective human feedback. Please sign up here:

New Course: Reinforcement Fine-Tuning LLMs with GRPO! Learn to use reinforcement learning to improve your LLM performance in this short course, built in collaboration with Predibase by Rubrik, and taught by Travis Addair, its Co-Founder and CTO, and Arnav Garg, its Senior Engineer and Machine Learning Lead. Reasoning models have been one of the most important developments in LLMs. Reinforcement Fine-Tuning (RFT) uses rewards to encourage LLMs to find solutions to multi-step reasoning tasks such as solving math problems and debugging code - without needing pre-existing training examples like in traditional supervised fine-tuning. Group Relative Policy Optimization (GRPO) is a reinforcement fine-tuning algorithm gaining rapid adoption. Developed by the DeepSeek team and used to train the R1 reasoning model, GRPO uses reward functions that you can write in Python to assign rewards to model responses. It’s beneficial for tasks with verifiable outcomes and can work well even with fewer than 100 training examples. It can also significantly improve the reasoning ability of smaller LLMs, making applications faster and more cost effective. In this course, you’ll take a technical deep dive into RFT with GRPO. You’ll learn to build reward functions that you can use in the GRPO training process to guide an LLM toward better performance on multi-step reasoning tasks. In detail, you’ll: - Learn when reinforcement fine-tuning is a better fit than supervised fine-tuning, especially for tasks involving multi-step reasoning or limited labeled data. - Understand how GRPO uses programmable reward functions as a more scalable alternative to the human feedback required for other reinforcement learning algorithms, such as RLHF and DPO. - Frame the Wordle game as a reinforcement fine-tuning problem and see how an LLM can learn to plan, analyze feedback, and improve its strategy over time. - Design reward functions that power the reinforcement fine-tuning process. - Learn techniques for evaluating more subjective tasks, such as rating the quality of a text summary, using an LLM as a judge. - Understand why reward hacking happens and how to avoid it by adding penalty functions to discourage undesirable behaviors. - Learn the four key components of the loss calculation in the GRPO algorithm: token probability distribution ratios, advantages, clipping, and KL-divergence. - Launch reinforcement fine-tuning jobs using Predibase’s hosted training services. By the end of this course, you’ll be able to build and fine-tune LLMs using reinforcement learning to improve reasoning without relying on large labeled datasets or subjective human feedback. Please sign up here:

Andrew Ng

86,697 次观看 • 1 年前

We've been getting Fable 5.1 in the background this whole time, and if you're running Fable 5 in a Claude Code it's probably you too. Day 5 of building CALI open world with a loop of ai agents. Today's progress: - map is fine tuning - buildings going up across it - car hijacking going in now We can't even drive it yet. Too much on screen and the fps is at 1. However I just spawn codex to fix the FPS issue. Can't wait to get in it.

We've been getting Fable 5.1 in the background this whole time, and if you're running Fable 5 in a Claude Code it's probably you too. Day 5 of building CALI open world with a loop of ai agents. Today's progress: - map is fine tuning - buildings going up across it - car hijacking going in now We can't even drive it yet. Too much on screen and the fps is at 1. However I just spawn codex to fix the FPS issue. Can't wait to get in it.

Ziwen

28,510 次观看 • 22 天前

Fine tuning your game just got even easier. The @bushnellgolf Launch Pro is back with improved technology to help you dial in your data this off season. @rdsbarath breaks down the new models and how it can help your game.

Fine tuning your game just got even easier. The @bushnellgolf Launch Pro is back with improved technology to help you dial in your data this off season. @rdsbarath breaks down the new models and how it can help your game.

Skratch

157,045 次观看 • 10 个月前

Excited to share our latest progress on DYNA-1 pre-training! 🤖 The base model now can perform diverse, dexterous tasks (laundry folding, package sorting, …) without any post-training, even in unseen environments. This powerful base also allows extremely efficient fine-tuning to ~100% success on challenging new tasks with as little as 1 hour of data! 🤯 Watch it master two of them: cup stacking & celery chopping on repeat, no failures. 👇

Excited to share our latest progress on DYNA-1 pre-training! 🤖 The base model now can perform diverse, dexterous tasks (laundry folding, package sorting, …) without any post-training, even in unseen environments. This powerful base also allows extremely efficient fine-tuning to ~100% success on challenging new tasks with as little as 1 hour of data! 🤯 Watch it master two of them: cup stacking & celery chopping on repeat, no failures. 👇

Dyna Robotics

68,112 次观看 • 10 个月前

Many people are flat wrong about DeepSeek. You might think everyone freaked out about DeepSeek because the model is really good—which it is—or because it was from China, which is also true. But there's a more subtle reason: DeepSeek showed they could achieve those results using reinforcement learning instead of supervised fine-tuning. In other words, they didn't need an expensive phase where a bunch of people tuned the model by answering questions. They were able to do all of this faster and cheaper using reinforcement fine-tuning. This is a huge deal! I recorded the attached video with the following goals in mind: 1. Explain how reinforcement fine-tuning works 2. Show you how you can fine-tune your own models 3. Walk you through a complete code example

Many people are flat wrong about DeepSeek. You might think everyone freaked out about DeepSeek because the model is really good—which it is—or because it was from China, which is also true. But there's a more subtle reason: DeepSeek showed they could achieve those results using reinforcement learning instead of supervised fine-tuning. In other words, they didn't need an expensive phase where a bunch of people tuned the model by answering questions. They were able to do all of this faster and cheaper using reinforcement fine-tuning. This is a huge deal! I recorded the attached video with the following goals in mind: 1. Explain how reinforcement fine-tuning works 2. Show you how you can fine-tune your own models 3. Walk you through a complete code example

Santiago

105,589 次观看 • 1 年前

This is the Siri update Apple never shipped 🤯 Lemon just launched a "Voice-to-Action" engine that instead of just transcribing your voice, it actually uses your computer to complete tasks. → 5x faster than typing → Works on every Mac app → Free to use option

This is the Siri update Apple never shipped 🤯 Lemon just launched a "Voice-to-Action" engine that instead of just transcribing your voice, it actually uses your computer to complete tasks. → 5x faster than typing → Works on every Mac app → Free to use option

Alvaro Cintas

29,620 次观看 • 7 个月前

Internet Computer Just Activated a New Level of Security... DFINITY Foundation officially launches SEV Subnets on the Internet Computer to eliminate plaintext data exposure at the hardware level. This integration uses Secure Encrypted Virtualization (SEV) to ensure sensitive data remains encrypted even when viewed by entities with direct physical access to the server hardware. Engineering Lead Björn Tackmann confirms that this upgrade resolves a fundamental vulnerability in traditional cloud computing, in which data is processed in plaintext within the memory layer.

Internet Computer Just Activated a New Level of Security... DFINITY Foundation officially launches SEV Subnets on the Internet Computer to eliminate plaintext data exposure at the hardware level. This integration uses Secure Encrypted Virtualization (SEV) to ensure sensitive data remains encrypted even when viewed by entities with direct physical access to the server hardware. Engineering Lead Björn Tackmann confirms that this upgrade resolves a fundamental vulnerability in traditional cloud computing, in which data is processed in plaintext within the memory layer.

BSCN

32,480 次观看 • 2 个月前