正在加载视频...

视频加载失败

返回首页

Can reinforcement learning from AI feedback unlock new capabilities in AI agents? Introducing Motif, an LLM-powered method for intrinsic motivation from AI feedback. Motif extracts reward functions from Llama 2's preferences and uses them to train agents with reinforcement learning. On the complex NetHack game, Motif solves previously unsolved...

proceduralia's profile picture

Pierluca D'Oro

2,535 subscribers

311,883 次观看 • 2 年前 •via X (Twitter)

游戏科学技术教育
LIVE

Anya Rossi• Live Now

Private livecam show

10 条评论

Pierluca D'Oro 的头像
Pierluca D'Oro2 年前

To benchmark the capabilities of Motif, we apply it to NetHack, a challenging rogue-like videogame, in which a player has to go through different levels of a dungeon, killing monsters, gathering objects and overcoming significant difficulties. Yet common sense can take you very far in such an environment! We use the messages from the game (i.e., even captions shown in 20% of interactions) to ask Llama 2 about its preferences about game situations. In this image, for instance, the event caption is "You kill the yellow mold!", which is understood by the Llama 2 model due to its knowledge of NetHack. / 2

Pierluca D'Oro 的头像
Pierluca D'Oro2 年前

Motif leverages recent ideas from RLAIF, asking an LLM to rank event captions and then distilling those preferences into a reward function. Motif has three phases: • Dataset annotation: given a dataset of observations with event captions, Motif uses Llama 2 to give preferences on sampled pairs according to its perception of how good and promising they are in the environment • Reward training: the resulting dataset of annotated pairs is used to learn a reward model from preferences • RL training: the reward function is given to an agent interacting with the environment and used to train it with RL, possibly alongside an external reward / 3

Pierluca D'Oro 的头像
Pierluca D'Oro2 年前

Our prompting strategy is very simple: we ask the model (with a form of chain of thought) to reason about its knowledge of NetHack and to give preferences about two messages coming from NetHack. We add the possibility to have a modifier in the prompt, to further influence and steer the behavior of an agent. / 4

Pierluca D'Oro 的头像
Pierluca D'Oro2 年前

We measure the performance of Motif in the task of NetHack score maximization, finding, surprisingly, that the reward extracted from Llama 2 yields an agent obtaining better score than an agent directly optimizing the score itself. This is the first time an intrinsically motivated agent outperforms the task-driven baseline on such a complex environment. When combining the score reward to the one coming from the LLM, the performance improves even further, both in terms of sample efficiency and final score. / 5

Pierluca D'Oro 的头像
Pierluca D'Oro2 年前

Motif also solves extremely sparse reward tasks, when other methods never find the solution. Motif does not need any demonstration or expert data, and simply bootstraps its knowledge from the LLM's common sense, to effectively explore the environment. Motif distills the capability of finding new information into AI agents. For instance, they will already know that opening a door is good for exploration even before having ever interacted with one. / 6

Pierluca D'Oro 的头像
Pierluca D'Oro2 年前

To understand where the performance of Motif is coming from, we analyzed the behavior of agents trained with Motif’s and the extrinsic reward function. We discovered that, when compared to an agent just trained using the score from the game, the behavior of agents trained with Motif is more aligned with human intuitions on playing NetHack: instead of going to the levels as fast as possible, encountering monsters too strong for its experience level, Motif’s reward encourages survival. An agent trained with both the intrinsic and extrinsic rewards is thus able to leverage this survival attitude to get significantly more score than an agent trained to maximize the score. / 7

Pierluca D'Oro 的头像
Pierluca D'Oro2 年前

We show that by just modifying the prompt given to the model with a single sentence, you can easily change the resulting behavior of the AI agent. Do you want an agent that maximizes gold, goes down the dungeon or kills more monsters? No problem, just tell the LLM to prefer that particular type of agent. / END

Pierluca D'Oro 的头像
Pierluca D'Oro2 年前

@AlphaSignalAI @DrJimFan @TalkRLPodcast @MLStreetTalk @TheTuringPost @_akhaliq @arankomatsuzaki @natolambert @Mila_Quebec @AIatMeta @ylecun @egrefen @_rockt

TalkRL Podcast 的头像
TalkRL Podcast2 年前

🔥 Congrats! Would one of the co-authors want to tell us about this work on @TalkRLPodcast ?

Andrzej Białecki 的头像
Andrzej Białecki2 年前

A new era of MMO bots are upon us! Great work.

相关视频

Can AI agents adapt zero-shot, to complex multi-step language instructions in open-ended environments? We present MaestroMotif, a method for AI-assisted skill design that produces highly capable and steerable hierarchical agents. To the best of our knowledge, it is the first method that, without expert labeled datasets, solves compositional tasks requiring hundreds of steps for completion. All the modules within MaestroMotif are learned from interaction: from the highest level of planning to the lowest-level of sensorimotor control. On the open-ended domain of NetHack, it surpasses existing approaches, including those that are fine-tuned specifically for each task. At the heart of MaestroMotif is the idea that decomposing a task into subtasks significantly helps decision making. MaestroMotif leverages an agent designer's intuition about a domain to identify important skills and describe them in natural language. These short descriptions then get converted into adaptable hierarchical agents through AI feedback and in-context learning. Our paper was recently published at ICLR 2025 and we open-source the whole project including the code, prompts and pre-trained models. Paper: Code: NotebookLM Podcast: This work was done with the amazing Mikael Henaff, Roberta Raileanu, Shagun Sodhani, Pascal Vincent, Amy Zhang, Pierre-Luc Bacon, Doina Precup, with equal supervision by Marlos C. Machado and Pierluca D'Oro. Take a look at the following thread:
1:12

Can AI agents adapt zero-shot, to complex multi-step language instructions in open-ended environments? We present MaestroMotif, a method for AI-assisted skill design that produces highly capable and steerable hierarchical agents. To the best of our knowledge, it is the first method that, without expert labeled datasets, solves compositional tasks requiring hundreds of steps for completion. All the modules within MaestroMotif are learned from interaction: from the highest level of planning to the lowest-level of sensorimotor control. On the open-ended domain of NetHack, it surpasses existing approaches, including those that are fine-tuned specifically for each task. At the heart of MaestroMotif is the idea that decomposing a task into subtasks significantly helps decision making. MaestroMotif leverages an agent designer's intuition about a domain to identify important skills and describe them in natural language. These short descriptions then get converted into adaptable hierarchical agents through AI feedback and in-context learning. Our paper was recently published at ICLR 2025 and we open-source the whole project including the code, prompts and pre-trained models. Paper: Code: NotebookLM Podcast: This work was done with the amazing Mikael Henaff, Roberta Raileanu, Shagun Sodhani, Pascal Vincent, Amy Zhang, Pierre-Luc Bacon, Doina Precup, with equal supervision by Marlos C. Machado and Pierluca D'Oro. Take a look at the following thread:

Martin Klissarov

80,217 次观看 • 1 年前

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:
2:41

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

86,457 次观看 • 1 年前

New short course on Reinforcement Learning from Human Feedback! RLHF is one of the key techniques that led to the rise of modern LLMs. It is used to align LLMs with human preferences, to make them more honest, helpful and harmless, by (i) learning a reward function that mimics human preferences, as expressed in human-provided labels, then, (ii) tuning an LLM to generate outputs that receive a high reward. In this course, taught by Nikita Namjoshi, Developer Advocate for GenAI at Google Cloud, you'll learn the details of how RLHF works, including how to apply it to tune an LLM for your own applications. You'll also use an open source library to tune a base LLM to align with human preferences expressed in a training set, and evaluate the tuned model by comparing its responses before and after RLHF-tuning. Please sign up here!
3:26

New short course on Reinforcement Learning from Human Feedback! RLHF is one of the key techniques that led to the rise of modern LLMs. It is used to align LLMs with human preferences, to make them more honest, helpful and harmless, by (i) learning a reward function that mimics human preferences, as expressed in human-provided labels, then, (ii) tuning an LLM to generate outputs that receive a high reward. In this course, taught by Nikita Namjoshi, Developer Advocate for GenAI at Google Cloud, you'll learn the details of how RLHF works, including how to apply it to tune an LLM for your own applications. You'll also use an open source library to tune a base LLM to align with human preferences expressed in a training set, and evaluate the tuned model by comparing its responses before and after RLHF-tuning. Please sign up here!

Andrew Ng

205,527 次观看 • 2 年前

Turn any open-source LLM into reasoning powerhouse! Using reinforcement finetuning you can add reasoning abilities to any LLM, even without a labelled dataset. Step-by-step explanation with code:
10:41

Turn any open-source LLM into reasoning powerhouse! Using reinforcement finetuning you can add reasoning abilities to any LLM, even without a labelled dataset. Step-by-step explanation with code:

Akshay 🚀

50,423 次观看 • 1 年前

Karpathy's prediction about RL is coming true now! He called reward functions unreliable and argued that a single reward number is too low-dimensional to teach an agent what "good" means for complex tasks. To solve this, Agents need a knowledge-guided review as a higher-dimensional feedback channel. Every major AI lab trains models with RL today (OpenAI, Anthropic, DeepSeek). And their key bottleneck has always been the reward functions. GRPO by DeepSeek worked well for math and code because the environment gave a binary signal. But for real agent tasks, someone still has to hand-code the scoring function. That takes days and breaks every time the pipeline changes. RULER (implemented in OpenPipe ART, 10k stars) addresses the exact problem Karpathy identified. The reward criteria are defined in plain English, and an LLM evaluates each trajectory against that description to provide feedback for training. I trained a Qwen3 1.4B agent that plays 2048 using GRPO with this exact workflow. In this case, the agent saw the board, picked a direction, and RULER evaluated the outcome, all from this natural language definition. You can see the full implementation on GitHub and try it yourself. Here's the ART Repo: (don't forget to star it ⭐ ) Just like RLHF replaced manual rankings and GRPO replaced the critic model, natural language rewards are replacing hand-coded scoring functions. RL reward engineering is now prompt engineering. I wrote a full walkthrough covering RL for LLM agents, from RLHF to GRPO to RULER, in the article below.
0:25

Karpathy's prediction about RL is coming true now! He called reward functions unreliable and argued that a single reward number is too low-dimensional to teach an agent what "good" means for complex tasks. To solve this, Agents need a knowledge-guided review as a higher-dimensional feedback channel. Every major AI lab trains models with RL today (OpenAI, Anthropic, DeepSeek). And their key bottleneck has always been the reward functions. GRPO by DeepSeek worked well for math and code because the environment gave a binary signal. But for real agent tasks, someone still has to hand-code the scoring function. That takes days and breaks every time the pipeline changes. RULER (implemented in OpenPipe ART, 10k stars) addresses the exact problem Karpathy identified. The reward criteria are defined in plain English, and an LLM evaluates each trajectory against that description to provide feedback for training. I trained a Qwen3 1.4B agent that plays 2048 using GRPO with this exact workflow. In this case, the agent saw the board, picked a direction, and RULER evaluated the outcome, all from this natural language definition. You can see the full implementation on GitHub and try it yourself. Here's the ART Repo: (don't forget to star it ⭐ ) Just like RLHF replaced manual rankings and GRPO replaced the critic model, natural language rewards are replacing hand-coded scoring functions. RL reward engineering is now prompt engineering. I wrote a full walkthrough covering RL for LLM agents, from RLHF to GRPO to RULER, in the article below.

Avi Chawla

349,743 次观看 • 1 个月前

Reinforcement learning is mind-blowing! Here's a reinforcement learning agent that mastered Breakout and is playing at a superhuman level (score 350+). The crazy part: We didn't write the code or the pipeline. NEO, the first autonomous AI Engineer, did it. From a prompt! It trained over 24 million steps. Evaluated itself. Iterated. Everything under 50 minutes. NEO designed the environment, the reward function, and the entire pipeline. Agentic AI is here!
0:24

Reinforcement learning is mind-blowing! Here's a reinforcement learning agent that mastered Breakout and is playing at a superhuman level (score 350+). The crazy part: We didn't write the code or the pipeline. NEO, the first autonomous AI Engineer, did it. From a prompt! It trained over 24 million steps. Evaluated itself. Iterated. Everything under 50 minutes. NEO designed the environment, the reward function, and the entire pipeline. Agentic AI is here!

Santiago

21,701 次观看 • 7 个月前

New short course: Building Code Agents with Hugging Face smolagents! Learn how to build code agents in this course, created in collaboration with Hugging Face, and taught by Thomas Wolf, its co-founder and CSO, and m_ric, Hugging Face’s Project Lead on Agents. Tool-calling agents use LLMs to generate multiple function calls sequentially to complete a complex sequence of tasks. They generate one function call, execute it, observe, reason, and decide what to do next. Code agents take a different approach. They consolidate all these calls into a single block of code, letting the LLM lay out an entire action plan at once, which can be executed efficiently to provide more reliable results. You’ll learn how to code agents using smolagents, a lightweight agentic framework from Hugging Face. Along the way, you’ll learn how to run LLM-generated code safely and develop an evaluation system to optimize your code agent for production. In detail, you’ll learn: - How agentic systems have evolved, gaining greater levels of agency over time—and why code agents are a next step. - How code agents write their actions in code. - When code agents outperform function-calling agents. - How to run code agents safely in your system using a constrained Python interpreter and sandboxing using E2B. - To trace, debug, and assess the code agent to optimize its behaviours for complex requests. - How to build a research multi-agent system that can find information online and organize it into an interactive report. By the end of this course, you’ll know how to build and run code agents using smolagents, and deploy them safely with a structured evaluation system in your projects. Please sign up here!
2:17

New short course: Building Code Agents with Hugging Face smolagents! Learn how to build code agents in this course, created in collaboration with Hugging Face, and taught by Thomas Wolf, its co-founder and CSO, and m_ric, Hugging Face’s Project Lead on Agents. Tool-calling agents use LLMs to generate multiple function calls sequentially to complete a complex sequence of tasks. They generate one function call, execute it, observe, reason, and decide what to do next. Code agents take a different approach. They consolidate all these calls into a single block of code, letting the LLM lay out an entire action plan at once, which can be executed efficiently to provide more reliable results. You’ll learn how to code agents using smolagents, a lightweight agentic framework from Hugging Face. Along the way, you’ll learn how to run LLM-generated code safely and develop an evaluation system to optimize your code agent for production. In detail, you’ll learn: - How agentic systems have evolved, gaining greater levels of agency over time—and why code agents are a next step. - How code agents write their actions in code. - When code agents outperform function-calling agents. - How to run code agents safely in your system using a constrained Python interpreter and sandboxing using E2B. - To trace, debug, and assess the code agent to optimize its behaviours for complex requests. - How to build a research multi-agent system that can find information online and organize it into an interactive report. By the end of this course, you’ll know how to build and run code agents using smolagents, and deploy them safely with a structured evaluation system in your projects. Please sign up here!

Andrew Ng

124,382 次观看 • 1 年前

Nothing beats open-source! We now have a fully open-source coding agent that can compete with any proprietary solution out there: • 500+ available models • It doesn't train on your code • It doesn't charge a penny for tokens • It's open. You can change if you want. Kilo Code is the #1 app on OpenRouter right now, and the community is releasing new features on GitHub at rocket speed (about 20 releases in the last 30 days). I installed it in Visual Studio Code, but you can install it in pretty much any IDE. I need to spend more time with it, but here is a very interesting design choice: Most coding agents support two modes: an "agentic" mode for writing code and an "ask" mode for talking to the model without making any changes. Kilo Code supports 5 different modes: 1. Architect mode - Research, design, and planning 2. Debug mode - Problem solving and troubleshooting 3. Code mode - To write code. 4. Ask mode - To talk to the model. 5. Orchestrator mode - Combine every agent to solve problems. On paper, I really like this. I'll need more time to try out the specialized agents and see whether niching becomes an advantage. Here is the link to the website: Thanks to the Kilo for partnering with me on this post.
6:17

Nothing beats open-source! We now have a fully open-source coding agent that can compete with any proprietary solution out there: • 500+ available models • It doesn't train on your code • It doesn't charge a penny for tokens • It's open. You can change if you want. Kilo Code is the #1 app on OpenRouter right now, and the community is releasing new features on GitHub at rocket speed (about 20 releases in the last 30 days). I installed it in Visual Studio Code, but you can install it in pretty much any IDE. I need to spend more time with it, but here is a very interesting design choice: Most coding agents support two modes: an "agentic" mode for writing code and an "ask" mode for talking to the model without making any changes. Kilo Code supports 5 different modes: 1. Architect mode - Research, design, and planning 2. Debug mode - Problem solving and troubleshooting 3. Code mode - To write code. 4. Ask mode - To talk to the model. 5. Orchestrator mode - Combine every agent to solve problems. On paper, I really like this. I'll need more time to try out the specialized agents and see whether niching becomes an advantage. Here is the link to the website: Thanks to the Kilo for partnering with me on this post.

Santiago

73,062 次观看 • 8 个月前

New short course: Evaluating AI Agents! Evals are important for driving AI system improvements, and in this course you'll learn to systematically assess and improve an AI agent’s performance. This is built in partnership with Arize AI and taught by John Gilhuly, Head of Developer Relations, and , Director of Product. I've often found evals to be a critical tool in the agent development process - they can be the difference between picking the right thing to work on vs. wasting weeks of effort. Whether you’re building a shopping assistant, coding agent, or research assistant, having a structured evaluation process helps you refine its performance systematically, rather than relying on random trial and error. This course shows you how to structure your evals to assess the performance of each component of an agent and its end-to-end performance. For each component, you select the appropriate evaluators, test examples, and performance metrics. This helps you identify areas for improvement both during development and in production. (If you're familiar with error analysis in supervised learning, think of this as adapting those ideas to agentic workflows.) In this course, you'll build an AI agent, and add observability to visualize and debug its steps. You’ll learn about code-based evals, in which you write code explicitly to test a certain step, as well as LLM-as-a-Judge evals, in which you prompt an LLM to efficiently come up with ways to evaluate more open-ended outputs. In detail, you’ll: - Understand key differences between evaluating LLM-based systems and traditional software testing. - Add observability to an agent by collecting traces of the steps taken by the agent and visualizing them - Choose the appropriate evaluator - code-based, LLM-as-a-Judge, human-annotation based - for each component. - Compute a convergence score to evaluate if your agent can respond to a query in an efficient number of steps. - Run structured experiments to improve the agent’s performance by exploring changes to the prompt, LLM model, or the agent’s logic. - Understand how to deploy these evaluation techniques to monitor the agent’s performance in production. By the end of this course, you’ll know how to trace AI agents, systematically evaluate them, and improve their performance. Please sign up here:
2:07

New short course: Evaluating AI Agents! Evals are important for driving AI system improvements, and in this course you'll learn to systematically assess and improve an AI agent’s performance. This is built in partnership with Arize AI and taught by John Gilhuly, Head of Developer Relations, and , Director of Product. I've often found evals to be a critical tool in the agent development process - they can be the difference between picking the right thing to work on vs. wasting weeks of effort. Whether you’re building a shopping assistant, coding agent, or research assistant, having a structured evaluation process helps you refine its performance systematically, rather than relying on random trial and error. This course shows you how to structure your evals to assess the performance of each component of an agent and its end-to-end performance. For each component, you select the appropriate evaluators, test examples, and performance metrics. This helps you identify areas for improvement both during development and in production. (If you're familiar with error analysis in supervised learning, think of this as adapting those ideas to agentic workflows.) In this course, you'll build an AI agent, and add observability to visualize and debug its steps. You’ll learn about code-based evals, in which you write code explicitly to test a certain step, as well as LLM-as-a-Judge evals, in which you prompt an LLM to efficiently come up with ways to evaluate more open-ended outputs. In detail, you’ll: - Understand key differences between evaluating LLM-based systems and traditional software testing. - Add observability to an agent by collecting traces of the steps taken by the agent and visualizing them - Choose the appropriate evaluator - code-based, LLM-as-a-Judge, human-annotation based - for each component. - Compute a convergence score to evaluate if your agent can respond to a query in an efficient number of steps. - Run structured experiments to improve the agent’s performance by exploring changes to the prompt, LLM model, or the agent’s logic. - Understand how to deploy these evaluation techniques to monitor the agent’s performance in production. By the end of this course, you’ll know how to trace AI agents, systematically evaluate them, and improve their performance. Please sign up here:

Andrew Ng

126,406 次观看 • 1 年前

Introducing Candy Machine 🍭🤖 An AI-powered vault that doesn't open with keys, it opens with prompts. To open it? You’ve got to prompt your way in. Craft the right prompt, convince the AI, and the reward is yours. Built on Aptos by Cluster Protocol Think you can outsmart AI ->
0:47

Introducing Candy Machine 🍭🤖 An AI-powered vault that doesn't open with keys, it opens with prompts. To open it? You’ve got to prompt your way in. Craft the right prompt, convince the AI, and the reward is yours. Built on Aptos by Cluster Protocol Think you can outsmart AI ->

Cluster Protocol

100,600 次观看 • 1 年前

Can LLMs invent better ways to train LLMs? At Sakana AI, we’re pioneering AI-driven methods to automate AI research and discovery. We’re excited to release DiscoPOP: a new SOTA preference optimization algorithm that was discovered and written by an LLM! Our method leverages LLMs to propose and implement new preference optimization algorithms. We then train models with those algorithms and evaluate their performance, providing feedback to the LLM. By repeating this process for multiple generations in an evolutionary loop, the LLM discovers many highly-performant and novel preference optimization objectives! Paper: GitHub: Model: We proudly collaborated with the University of Oxford (Foerster Lab for AI Research (now part of BOLD)) and Cambridge University (Mihaela van der Schaar) on this groundbreaking project. Looking ahead, we envision a future where AI-driven research reduces the need for extensive human intervention and computational resources. This will accelerate scientific discoveries and innovation, pushing the boundaries of what AI can achieve.
0:22

Can LLMs invent better ways to train LLMs? At Sakana AI, we’re pioneering AI-driven methods to automate AI research and discovery. We’re excited to release DiscoPOP: a new SOTA preference optimization algorithm that was discovered and written by an LLM! Our method leverages LLMs to propose and implement new preference optimization algorithms. We then train models with those algorithms and evaluate their performance, providing feedback to the LLM. By repeating this process for multiple generations in an evolutionary loop, the LLM discovers many highly-performant and novel preference optimization objectives! Paper: GitHub: Model: We proudly collaborated with the University of Oxford (Foerster Lab for AI Research (now part of BOLD)) and Cambridge University (Mihaela van der Schaar) on this groundbreaking project. Looking ahead, we envision a future where AI-driven research reduces the need for extensive human intervention and computational resources. This will accelerate scientific discoveries and innovation, pushing the boundaries of what AI can achieve.

Sakana AI

555,859 次观看 • 2 年前

Introducing CADGenBench: measure how well AI systems produce engineering-grade 3D parts! While current models can generate 3D parts, they are far from precise enough to build functional parts. We built a benchmark to systematically measure their capabilities on two tasks: 1. Generation from an engineering drawing of a part 2. Editing: given an existing STEP file and a requested change The benchmark is tool-agnostic. It makes no assumptions about how you build the model. You can vary the LLM, and you can vary the environment. Use build123d, Onshape, Autodesk, or a model without an LLM entirely. We open sourced the scoring engine and a reference baseline on top of build123d. A collaboration between Hugging Face and Mecado! Submission space: Code repository:
0:43

Introducing CADGenBench: measure how well AI systems produce engineering-grade 3D parts! While current models can generate 3D parts, they are far from precise enough to build functional parts. We built a benchmark to systematically measure their capabilities on two tasks: 1. Generation from an engineering drawing of a part 2. Editing: given an existing STEP file and a requested change The benchmark is tool-agnostic. It makes no assumptions about how you build the model. You can vary the LLM, and you can vary the environment. Use build123d, Onshape, Autodesk, or a model without an LLM entirely. We open sourced the scoring engine and a reference baseline on top of build123d. A collaboration between Hugging Face and Mecado! Submission space: Code repository:

Michael Rabinovich

63,879 次观看 • 1 个月前

Open science is how we continue to push technology forward and today at Meta FAIR we’re sharing eight new AI research artifacts including new models, datasets and code to inspire innovation in the community. More in the video from Joelle Pineau. This work is another important step towards our goal of achieving Advanced Machine Intelligence (AMI). What we’re releasing: • Meta Spirit LM: An open source language model for seamless speech and text integration. • Meta Segment Anything Model 2.1: An updated checkpoint with improved results on visually similar objects, small objects and occlusion handling. Plus a new developer suite to make it easier for developers to build with SAM 2. • Layer Skip: Inference code and fine-tuned checkpoints demonstrating a new method for enhancing LLM performance. • SALSA: New code to enable researchers to benchmark AI-based attacks in support of validating security for post-quantum cryptography. • Meta Lingua: A lightweight and self-contained codebase designed to train language models at scale. • Meta Open Materials: New open source models and the largest dataset of its kind to accelerate AI-driven discovery of new inorganic materials. • MEXMA: A new research paper and code for our novel pre-trained cross-lingual sentence encoder with coverage across 80 languages. • Self-Taught Evaluator: a new method for generating synthetic preference data to train reward models without relying on human annotations. Access to state-of-the-art AI creates opportunities for everyone. We’re excited to share this work and look forward to seeing the community innovation that results from it. Details and access to everything released by FAIR today ➡️
1:36

Open science is how we continue to push technology forward and today at Meta FAIR we’re sharing eight new AI research artifacts including new models, datasets and code to inspire innovation in the community. More in the video from Joelle Pineau. This work is another important step towards our goal of achieving Advanced Machine Intelligence (AMI). What we’re releasing: • Meta Spirit LM: An open source language model for seamless speech and text integration. • Meta Segment Anything Model 2.1: An updated checkpoint with improved results on visually similar objects, small objects and occlusion handling. Plus a new developer suite to make it easier for developers to build with SAM 2. • Layer Skip: Inference code and fine-tuned checkpoints demonstrating a new method for enhancing LLM performance. • SALSA: New code to enable researchers to benchmark AI-based attacks in support of validating security for post-quantum cryptography. • Meta Lingua: A lightweight and self-contained codebase designed to train language models at scale. • Meta Open Materials: New open source models and the largest dataset of its kind to accelerate AI-driven discovery of new inorganic materials. • MEXMA: A new research paper and code for our novel pre-trained cross-lingual sentence encoder with coverage across 80 languages. • Self-Taught Evaluator: a new method for generating synthetic preference data to train reward models without relying on human annotations. Access to state-of-the-art AI creates opportunities for everyone. We’re excited to share this work and look forward to seeing the community innovation that results from it. Details and access to everything released by FAIR today ➡️

AI at Meta

150,222 次观看 • 1 年前

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

2,673,990 次观看 • 2 年前

Super excited to share 🧠MLGym 🦾 – the first Gym environment for AI Research Agents 🤖🔬 We introduce MLGym and MLGym-Bench, a new framework and benchmark for evaluating and developing LLM agents on AI research tasks. The key contributions of our work are: 🕹️ Enables the exploration of different training algorithms for AI Research Agents such as RL 🛠️ Provides a flexible evaluation framework that can accommodate different artifacts such as models, algorithms, or predictions 🤖 Allows researchers to evaluate any model without the need to develop a custom agentic harness 🎯 Introduces 13 diverse open-ended AI Research tasks for evaluating AI Research Agents on a wide range of domains such as computer vision, natural language processing, reinforcement learning, game theory, and logical reasoning. 📈 Proposes a new evaluation metric for AI Research Agents MLGym makes it easy to: 1) Add new tasks 2) Evaluate new models 3) Integrate new agents Check out a video of the MLGym Agent to see how it performs the full pipeline of idea generation💡, implementation 👩‍💻, experimentation 👩‍🔬, and iteration 🔄 to improve on ML tasks. Huge thanks to the exceptionally talented Deepak Nathani who led this work and to all the other amazing collaborators who made this possible 🙏🫶🚀
1:32

Super excited to share 🧠MLGym 🦾 – the first Gym environment for AI Research Agents 🤖🔬 We introduce MLGym and MLGym-Bench, a new framework and benchmark for evaluating and developing LLM agents on AI research tasks. The key contributions of our work are: 🕹️ Enables the exploration of different training algorithms for AI Research Agents such as RL 🛠️ Provides a flexible evaluation framework that can accommodate different artifacts such as models, algorithms, or predictions 🤖 Allows researchers to evaluate any model without the need to develop a custom agentic harness 🎯 Introduces 13 diverse open-ended AI Research tasks for evaluating AI Research Agents on a wide range of domains such as computer vision, natural language processing, reinforcement learning, game theory, and logical reasoning. 📈 Proposes a new evaluation metric for AI Research Agents MLGym makes it easy to: 1) Add new tasks 2) Evaluate new models 3) Integrate new agents Check out a video of the MLGym Agent to see how it performs the full pipeline of idea generation💡, implementation 👩‍💻, experimentation 👩‍🔬, and iteration 🔄 to improve on ML tasks. Huge thanks to the exceptionally talented Deepak Nathani who led this work and to all the other amazing collaborators who made this possible 🙏🫶🚀

Roberta Raileanu

104,982 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
2:55

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

1,022,448 次观看 • 1 年前

We release a new urban simulator, MetaUrban, to support research on AI agents for micromobility. The work will be presented at #ICLR2025, and the demo code can run on any laptop. Webpage: Code: Paper:
0:58

We release a new urban simulator, MetaUrban, to support research on AI agents for micromobility. The work will be presented at #ICLR2025, and the demo code can run on any laptop. Webpage: Code: Paper:

Bolei Zhou

30,339 次观看 • 1 年前

The best way to learn about AI is to build an AI app yourself. With tools like Bubble and OpenAI, you can create apps like an AI chatbot without writing code 🤯 Stay ahead on AI and start learning today:
0:20

The best way to learn about AI is to build an AI app yourself. With tools like Bubble and OpenAI, you can create apps like an AI chatbot without writing code 🤯 Stay ahead on AI and start learning today:

Seth Kramer

2,614,004 次观看 • 3 年前

arXiv Papers → LLM Artifacts This is how I keep up with AI research now. It's like having access to the most personalized arXiv feed. Automations run everyday to curate papers based a set of rules and insights. Curated papers are indexed and power the artifacts. Agent convert papers to LLM wikis (based on Andrej Karpathy idea), which means insights are indexed and easily searchable and reusable. I feel like LLM Artifacts is the natural evolution to LLM Wikis. It's about making that knowledge actionable. Artifacts are customizable via agents. Artifacts can interact with agents and are dynamic in nature. Anything can be injected into the artifact as needed (insights, components, suggested experiments, action items, etc). I can take action on Artifact items with my agent orchestrator (Electron app). So I can ask questions about any paper and automate experiments in the background right from within the artifact. This is more than a visual. It's not a single prompt. It's several proactive agents coordinating to surface interesting facts, knowledge, and insights that I can act on a researcher. Agents are not just for generating useful artifacts, they are useful to keep learning and staying on the cutting edge of knowledge. Stay tuned for more.
0:20

arXiv Papers → LLM Artifacts This is how I keep up with AI research now. It's like having access to the most personalized arXiv feed. Automations run everyday to curate papers based a set of rules and insights. Curated papers are indexed and power the artifacts. Agent convert papers to LLM wikis (based on Andrej Karpathy idea), which means insights are indexed and easily searchable and reusable. I feel like LLM Artifacts is the natural evolution to LLM Wikis. It's about making that knowledge actionable. Artifacts are customizable via agents. Artifacts can interact with agents and are dynamic in nature. Anything can be injected into the artifact as needed (insights, components, suggested experiments, action items, etc). I can take action on Artifact items with my agent orchestrator (Electron app). So I can ask questions about any paper and automate experiments in the background right from within the artifact. This is more than a visual. It's not a single prompt. It's several proactive agents coordinating to surface interesting facts, knowledge, and insights that I can act on a researcher. Agents are not just for generating useful artifacts, they are useful to keep learning and staying on the cutting edge of knowledge. Stay tuned for more.

elvis

58,852 次观看 • 2 个月前

What if you could train AI agents on a laptop as easily as on a GPU cluster? Researchers from UIUC's U Lab, led by Prof. Jiaxuan You, just open-sourced OpenTinker. It's a new "Reinforcement-Learning-as-a-Service" (RLaaS) system that decouples the complex training pipeline into simple, distributed services with friendly APIs. The result? It breaks down the major engineering barriers to RL, outperforming traditional frameworks in accessibility and ease of deployment, finally making agent training viable for more developers and teams. Project: Code: U Lab: Our report: 📬 #PapersAccepted by Jiqizhixin
0:42

What if you could train AI agents on a laptop as easily as on a GPU cluster? Researchers from UIUC's U Lab, led by Prof. Jiaxuan You, just open-sourced OpenTinker. It's a new "Reinforcement-Learning-as-a-Service" (RLaaS) system that decouples the complex training pipeline into simple, distributed services with friendly APIs. The result? It breaks down the major engineering barriers to RL, outperforming traditional frameworks in accessibility and ease of deployment, finally making agent training viable for more developers and teams. Project: Code: U Lab: Our report: 📬 #PapersAccepted by Jiqizhixin

机器之心 JIQIZHIXIN

15,893 次观看 • 6 个月前