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。

In-context learning as the mysterious ability in LMs. We propose ✨Deep-thinking✨ to boost ICL by iterative forward tuning. It is possible to tune LMs without backpropagation! 🤯 Paper: Gradio Demo:

Binyuan Hui

43,886 subscribers

53,380 次观看 • 3 年前 •via X (Twitter)

Anya Rossi• Live Now

Private livecam show

0 条评论

暂无评论

原始帖子的评论将显示在这里

相关视频

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:

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 次观看 • 11 个月前

“We noticed that along with extended thinking, with GPT-5.5, we were able to do something that we didn't think was possible,” Shopify’s goal is to have more entrepreneurs in the world and the way they do that is by removing as much friction as possible from the builder journey. There was life before Shopify with GPT-5.5, and after. We prefer after. ✨

“We noticed that along with extended thinking, with GPT-5.5, we were able to do something that we didn't think was possible,” Shopify’s goal is to have more entrepreneurs in the world and the way they do that is by removing as much friction as possible from the builder journey. There was life before Shopify with GPT-5.5, and after. We prefer after. ✨

Farhan Thawar

13,857 次观看 • 1 个月前

Sending children to public school in the age of AI is a massive mistake. The system was already a joke, and it lacks the slightest ability to adjust to a rapidly changing world. The process of learning, adaptability, critical thinking, & problem-solving is the way forward.

Sending children to public school in the age of AI is a massive mistake. The system was already a joke, and it lacks the slightest ability to adjust to a rapidly changing world. The process of learning, adaptability, critical thinking, & problem-solving is the way forward.

Brett Pike

82,964 次观看 • 4 个月前

Storing too many tools in your context window increases latency and can lead to wrong tool selection. In this demo, we used LFM2.5-ColBERT-350M as a filter to only select the five most relevant tools among 151 options. It's fast and reliable, even without any specific fine-tuning. Try the demo on Hugging Face! And learn more on our blog:

Storing too many tools in your context window increases latency and can lead to wrong tool selection. In this demo, we used LFM2.5-ColBERT-350M as a filter to only select the five most relevant tools among 151 options. It's fast and reliable, even without any specific fine-tuning. Try the demo on Hugging Face! And learn more on our blog:

Liquid AI

33,392 次观看 • 1 个月前

We’ve heard your complaints You think the lyrics added in the newest update is as you fans call it “complete and utter dogshit” So we went back into the labs and decided to show a sneak peak of the NEW and IMPROVED version of the LMS!

We’ve heard your complaints You think the lyrics added in the newest update is as you fans call it “complete and utter dogshit” So we went back into the labs and decided to show a sneak peak of the NEW and IMPROVED version of the LMS!

Legit Forsaken Leaks

44,987 次观看 • 2 个月前

Rewrote tiny-diffusion to be 3x smaller! Went from 951 lines to just 364, all contained in one file. As simple as possible, but not simpler. I also added a tiny GPT implementation as a comparison (312 lines, inspired by Andrej Karpathy). The two implementations are ~80% identical. The model architecture, training loop, tokenization, etc, only differ in 19 lines of code. The main differences are contained within two functions (generate and get_batch). The reason to include the GPT implementation was to show how similar autoregressive LMs are to diffusion LMs on an architectural level. Only *1* line of code in the architecture needs to be modified to support masked language diffusion instead of next-token prediction (by disabling causal masking). Link to the repo is in the comments

Rewrote tiny-diffusion to be 3x smaller! Went from 951 lines to just 364, all contained in one file. As simple as possible, but not simpler. I also added a tiny GPT implementation as a comparison (312 lines, inspired by Andrej Karpathy). The two implementations are ~80% identical. The model architecture, training loop, tokenization, etc, only differ in 19 lines of code. The main differences are contained within two functions (generate and get_batch). The reason to include the GPT implementation was to show how similar autoregressive LMs are to diffusion LMs on an architectural level. Only 1 line of code in the architecture needs to be modified to support masked language diffusion instead of next-token prediction (by disabling causal masking). Link to the repo is in the comments

Nathan Barry

161,578 次观看 • 6 个月前

In our #ICML2023 paper (led by Ghada Sokar and with Rishabh Agarwal and utku ) we highlight the prevalence of dormant neurons in deep RL, and propose a simple method to mitigate it. Quick video summary here, and more details in thread!

In our #ICML2023 paper (led by Ghada Sokar and with Rishabh Agarwal and utku ) we highlight the prevalence of dormant neurons in deep RL, and propose a simple method to mitigate it. Quick video summary here, and more details in thread!

Pablo Samuel Castro

40,603 次观看 • 3 年前

Googles Project Suncatcher aims to build data centers (TPUs) in space by 2027! Project Suncatcher makes it possible. It's absolutely incredible how daring and forward-thinking Google is!

Googles Project Suncatcher aims to build data centers (TPUs) in space by 2027! Project Suncatcher makes it possible. It's absolutely incredible how daring and forward-thinking Google is!

Chubby♨️

27,219 次观看 • 8 个月前

🚀 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)

🚀 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 次观看 • 1 年前

🚨 New Paper Training an LLM to speak low-resource language (EACL workshop, 2026) Tulu is spoken by 2M+ people in coastal Karnataka and LLMs basically can't speak it. We got to 85% grammar accuracy without fine-tuning anything or collecting a single new training example.

🚨 New Paper Training an LLM to speak low-resource language (EACL workshop, 2026) Tulu is spoken by 2M+ people in coastal Karnataka and LLMs basically can't speak it. We got to 85% grammar accuracy without fine-tuning anything or collecting a single new training example.

Lossfunk

120,884 次观看 • 4 个月前

In modern language, we think the word “perfect” means “to perform without errors.” Thankfully, this is not what it means in the scriptural context. Thanks to God’s grace, we can be perfected in Him.

In modern language, we think the word “perfect” means “to perform without errors.” Thankfully, this is not what it means in the scriptural context. Thanks to God’s grace, we can be perfected in Him.

David A. Bednar

54,426 次观看 • 1 年前

This lady said in her old liberal mindset, she would have reacted angrily to Charlie Kirk's comment on black pilots, assuming racism without context. His comment on black pilots questions DEI policies that lower qualifications to boost representation from just 2%. Without context, it sounds inflammatory, but he's critiquing the system, not the people. We must stop being defensive, embrace critical thinking, self-reflection, and listen to hard truths from outsiders to grow, without instantly crying racism.

This lady said in her old liberal mindset, she would have reacted angrily to Charlie Kirk's comment on black pilots, assuming racism without context. His comment on black pilots questions DEI policies that lower qualifications to boost representation from just 2%. Without context, it sounds inflammatory, but he's critiquing the system, not the people. We must stop being defensive, embrace critical thinking, self-reflection, and listen to hard truths from outsiders to grow, without instantly crying racism.

ꪻꫝể ꪻꫝể

361,206 次观看 • 10 个月前

Our last Stanford guest lecture - Ekdeep Singh Lubana on what counts as an explanation & a neuro-inspired "model systems approach" to interp Plus, how in-context learning and many-shot jailbreaking are explained by LLM representations changing in-context (as a case study for that approach) 00:33 - What counts as an explanation? 04:47 - Levels of analysis & standard interpretability approaches 18:19 - The "model systems" approach to interp [Case study on in-context learning] 23:36 - How LLM representations change in-context 44:10 - Modeling ICL with rational analysis 1:10:54 - Conclusion & questions Thanks again to Surya Ganguli for having us in his class!

Our last Stanford guest lecture - Ekdeep Singh Lubana on what counts as an explanation & a neuro-inspired "model systems approach" to interp Plus, how in-context learning and many-shot jailbreaking are explained by LLM representations changing in-context (as a case study for that approach) 00:33 - What counts as an explanation? 04:47 - Levels of analysis & standard interpretability approaches 18:19 - The "model systems" approach to interp [Case study on in-context learning] 23:36 - How LLM representations change in-context 44:10 - Modeling ICL with rational analysis 1:10:54 - Conclusion & questions Thanks again to Surya Ganguli for having us in his class!

Goodfire

31,410 次观看 • 7 个月前

May 15th, 2024: The State of @Blast_L2 As we approach the highly anticipated TGE, let's study a bit so we can better understand... is it to late or are we still early? The latest you need to know from Learning in Public and Remint Reality ✨

May 15th, 2024: The State of @Blast_L2 As we approach the highly anticipated TGE, let's study a bit so we can better understand... is it to late or are we still early? The latest you need to know from Learning in Public and Remint Reality ✨

Jenn

47,034 次观看 • 2 年前

Have you bumped into Portofiro's resident phone sex addict yet? Tune in to the #ZEROPARADES Steam page to watch ASH 🫀✨ explore, or play for yourself by downloading our free #SteamNextFest demo:

Have you bumped into Portofiro's resident phone sex addict yet? Tune in to the #ZEROPARADES Steam page to watch ASH 🫀✨ explore, or play for yourself by downloading our free #SteamNextFest demo:

ZA/UM Studio

10,296 次观看 • 4 个月前

Your AI is only as smart as the context it runs on. Have you built your shared enterprise brain yet? At Databricks Data+AI Summit, I spoke with Prukalpa ✨, co-founder and co-CEO of Atlan, to go deeper on the AI Context Layer in practice. The concept of a shared enterprise brain resonated with me. It is a foundation that every AI agent draws from, learns from, and contributes back to. "The future of organizations are going to be the fastest learning organizations." - Prukalpa ✨ Sankar, co-founder and co-CEO of Atlan Watch the full interview on YouTube: Check out the live demo by Atlan on Building Context for Your AI Agents - #EnterpriseAI #ContextLayer #Data #AI #AtlanPartner Matthias Heidrich

Your AI is only as smart as the context it runs on. Have you built your shared enterprise brain yet? At Databricks Data+AI Summit, I spoke with Prukalpa ✨, co-founder and co-CEO of Atlan, to go deeper on the AI Context Layer in practice. The concept of a shared enterprise brain resonated with me. It is a foundation that every AI agent draws from, learns from, and contributes back to. "The future of organizations are going to be the fastest learning organizations." - Prukalpa ✨ Sankar, co-founder and co-CEO of Atlan Watch the full interview on YouTube: Check out the live demo by Atlan on Building Context for Your AI Agents - #EnterpriseAI #ContextLayer #Data #AI #AtlanPartner Matthias Heidrich

Helen Yu

10,953 次观看 • 27 天前

♥️✨Bonjour à tous! Today... Surprise!✨♥️ Today is the day we move forward in our adventure!♥ I'm so excited to see what happens next! Thanks again to all of you!✨ Link to the Patreon in comments! 💙🤍❤️

♥️✨Bonjour à tous! Today... Surprise!✨♥️ Today is the day we move forward in our adventure!♥ I'm so excited to see what happens next! Thanks again to all of you!✨ Link to the Patreon in comments! 💙🤍❤️

Luiroï 🍂

130,105 次观看 • 3 年前

How can you solve complex tasks using a Large Language Model? Here is a 2-minute introduction to everything you need to know to 10x the quality of your results. Let's talk about three techniques, in order of complexity, starting with the easiest one: • In-Context Learning • Indexing + In-Context Learning • Fine-tuning In-Context Learning The team that trained GPT-3 found something they couldn't explain: You can condition a model using examples of how you want it to behave. I included an example prompt in the attached video. You can "teach" the model how you want it to interpret questions, select the correct answers, and format the results by giving a few examples. You can also give specific knowledge to the model that will be helpful when formulating answers. We call this approach "grounding the model." There's another example in the video. Indexing + In-Context Learning Unfortunately, there is a limit to how much data you can include in a prompt. We call this the "context size." One version of GPT-4 supports a context of approximately 6,000 words, while the other supports 25,000 words. Although this sounds like a lot, many applications need more than that. Imagine you wrote a book and want to build an application to answer any questions about your story. What happens if your book is longer than the context? That's where Indexing comes in. Using a model, you can turn every book passage into an embedding. These are vectors, numbers that "encode" the passage's text. You can then store these embeddings in a particular database that supports fast retrieval of these vectors. You can then turn any question into an embedding and search the database for the list of passages that are similar to that query. Instead of using the entire book to ask the model, you can now use the relevant passages as in-context information, effectively working around the context size limitation. Fine-tuning Fine-tuning can give you an extra boost to get reliable outputs from your LLM. It is, however, the most complex approach on the list. There are different approaches to fine-tuning a model with your data. A popular technique is to process your data with your LLM and use the outputs to train a new classifier that solves your specific task. Notice that here you aren't modifying the LLM. Instead, you are chaining it with your trained classifier. Another approach is to modify the parameters of the LLM using your data. Think of this as "rewiring" the model in a way that solves your particular task. The results and costs will vary depending on how many layers you want to fine-tune from the original model. Many companies think that fine-tuning is the solution to their problems. In my experience, many will benefit from exploring the other two approaches. I love explaining Machine Learning and Artificial Intelligence ideas. If you enjoy in-depth content like this, follow me Santiago so you don't miss what comes next.

How can you solve complex tasks using a Large Language Model? Here is a 2-minute introduction to everything you need to know to 10x the quality of your results. Let's talk about three techniques, in order of complexity, starting with the easiest one: • In-Context Learning • Indexing + In-Context Learning • Fine-tuning In-Context Learning The team that trained GPT-3 found something they couldn't explain: You can condition a model using examples of how you want it to behave. I included an example prompt in the attached video. You can "teach" the model how you want it to interpret questions, select the correct answers, and format the results by giving a few examples. You can also give specific knowledge to the model that will be helpful when formulating answers. We call this approach "grounding the model." There's another example in the video. Indexing + In-Context Learning Unfortunately, there is a limit to how much data you can include in a prompt. We call this the "context size." One version of GPT-4 supports a context of approximately 6,000 words, while the other supports 25,000 words. Although this sounds like a lot, many applications need more than that. Imagine you wrote a book and want to build an application to answer any questions about your story. What happens if your book is longer than the context? That's where Indexing comes in. Using a model, you can turn every book passage into an embedding. These are vectors, numbers that "encode" the passage's text. You can then store these embeddings in a particular database that supports fast retrieval of these vectors. You can then turn any question into an embedding and search the database for the list of passages that are similar to that query. Instead of using the entire book to ask the model, you can now use the relevant passages as in-context information, effectively working around the context size limitation. Fine-tuning Fine-tuning can give you an extra boost to get reliable outputs from your LLM. It is, however, the most complex approach on the list. There are different approaches to fine-tuning a model with your data. A popular technique is to process your data with your LLM and use the outputs to train a new classifier that solves your specific task. Notice that here you aren't modifying the LLM. Instead, you are chaining it with your trained classifier. Another approach is to modify the parameters of the LLM using your data. Think of this as "rewiring" the model in a way that solves your particular task. The results and costs will vary depending on how many layers you want to fine-tune from the original model. Many companies think that fine-tuning is the solution to their problems. In my experience, many will benefit from exploring the other two approaches. I love explaining Machine Learning and Artificial Intelligence ideas. If you enjoy in-depth content like this, follow me Santiago so you don't miss what comes next.

Santiago

384,510 次观看 • 3 年前

Our work WorldScribe led by Ruei-Che Chang on generating context-aware live visual descriptions has received 🏆✨ Best Paper Award ✨ at #UIST2024!! Check out our new video below, and details in the thread! Here is what NotebookLM thought about it:

Our work WorldScribe led by Ruei-Che Chang on generating context-aware live visual descriptions has received 🏆✨ Best Paper Award ✨ at #UIST2024!! Check out our new video below, and details in the thread! Here is what NotebookLM thought about it:

Anhong Guo

13,266 次观看 • 1 年前

Today, we're introducing the new vibe coding experience in Google AI Studio!🔥 We want to make it as easy as possible for you to build AI apps :) Short demo of what's new:

Today, we're introducing the new vibe coding experience in Google AI Studio!🔥 We want to make it as easy as possible for you to build AI apps :) Short demo of what's new:

Patrick Loeber

77,772 次观看 • 9 个月前