Video wird geladen...

Video konnte nicht geladen werden

Zur Startseite

Randomly came across a tool called ResearchStudio that automatically creates videos for papers. Found it interesting so gave it a try for my recent preprint with Google on how SFT reasoning diversity impacts post-RL gains. Google

dylan_works_'s profile picture

Dylan Zhang

1,680 subscribers

15,055 Aufrufe • vor 26 Tagen •via X (Twitter)

Wissenschaft & TechnologieBildung
LIVE

Anya Rossi• Live Now

Private livecam show

0 Kommentare

Keine Kommentare verfügbar

Kommentare vom Original-Post werden hier angezeigt

Ähnliche Videos

Lecture 10 of my course! Nominally on regularization in RL, so I discuss the evolving role of the KL penalty in RL, but also a set of nice RL papers that explain what RL helps models generalize better than SFT -- with theory supporting it. When going through these, it's so interesting how seasonal problems in ML are. Lots of problems from controlling reward models overopt will rhyme as we try to control rubrics for agents. 00:00 Intro & the role of regularization 02:50 The KL penalty in RL 10:53 RL as a reverse KL loss 21:09 Why RL generalizes better than SFT 25:15 Other regularization tools Just a few videos left as I get to the end of the course. Thanks all, and keep sending questions. Spread the word if you have a second.
31:33

Lecture 10 of my course! Nominally on regularization in RL, so I discuss the evolving role of the KL penalty in RL, but also a set of nice RL papers that explain what RL helps models generalize better than SFT -- with theory supporting it. When going through these, it's so interesting how seasonal problems in ML are. Lots of problems from controlling reward models overopt will rhyme as we try to control rubrics for agents. 00:00 Intro & the role of regularization 02:50 The KL penalty in RL 10:53 RL as a reverse KL loss 21:09 Why RL generalizes better than SFT 25:15 Other regularization tools Just a few videos left as I get to the end of the course. Thanks all, and keep sending questions. Spread the word if you have a second.

Nathan Lambert

89,819 Aufrufe • vor 3 Monaten

Came across this early iteration of Thief while editing. Found it interesting so I'm posting here for posterity. Source is a German gaming publication called PC Games magazine reporting on E3 1997.
1:02

Came across this early iteration of Thief while editing. Found it interesting so I'm posting here for posterity. Source is a German gaming publication called PC Games magazine reporting on E3 1997.

Majuular

15,606 Aufrufe • vor 3 Monaten

Here’s how to go Live with Search. Try it out today in the Google app for Android and iOS.
0:24

Here’s how to go Live with Search. Try it out today in the Google app for Android and iOS.

Google

135,548 Aufrufe • vor 1 Jahr

GOOGLE JUST LEAKED A NEW TOOL CALLED GEMINI SPARK AND IT IS NOT A CHATBOT. IT RUNS IN THE BACKGROUND, READS YOUR GMAIL AND CLICKS BUTTONS FOR YOU.
7:18

GOOGLE JUST LEAKED A NEW TOOL CALLED GEMINI SPARK AND IT IS NOT A CHATBOT. IT RUNS IN THE BACKGROUND, READS YOUR GMAIL AND CLICKS BUTTONS FOR YOU.

0xMarioNawfal

52,151 Aufrufe • vor 5 Monaten

Video understanding isn't just recognizing —it demands reasoning across thousands of frames. Meet Long-RL🚀 Highlights: 🧠 Dataset: LongVideo-Reason — 52K QAs with reasoning. ⚡ System: MR-SP - 2.1× faster RL for long videos. 📈 Scalability: Hour-long videos (3,600 frames) RL on a single node (8×A100s). 🖼️📝🎵 RL training for video, text, audio — works with VILA, Qwen series, and image/video generation models 🎨🎬 📄 Paper: 🎥 Demo: 💻 Code:
1:19

Video understanding isn't just recognizing —it demands reasoning across thousands of frames. Meet Long-RL🚀 Highlights: 🧠 Dataset: LongVideo-Reason — 52K QAs with reasoning. ⚡ System: MR-SP - 2.1× faster RL for long videos. 📈 Scalability: Hour-long videos (3,600 frames) RL on a single node (8×A100s). 🖼️📝🎵 RL training for video, text, audio — works with VILA, Qwen series, and image/video generation models 🎨🎬 📄 Paper: 🎥 Demo: 💻 Code:

Yukang Chen

31,876 Aufrufe • vor 1 Jahr

Tomorrow, we’re releasing a small but mighty feature that moves us closer to... A ✨Post-Google✨ Internet Now Arc will automatically suggest @ChatGPTapp — instead of Google — for certain search queries. Why is this such a big deal, and what does it tease about our future?
0:26

Tomorrow, we’re releasing a small but mighty feature that moves us closer to... A ✨Post-Google✨ Internet Now Arc will automatically suggest @ChatGPTapp — instead of Google — for certain search queries. Why is this such a big deal, and what does it tease about our future?

Josh Miller

287,498 Aufrufe • vor 2 Jahren

How can we get VLMs to move their eyes—and reason step-by-step in visually grounded ways? 👀 We introduce ViGoRL, a RL method that anchors reasoning to image regions. 🎯 It outperforms vanilla GRPO and SFT across grounding, spatial tasks, and visual search (86.4% on V*). 👇🧵
0:39

How can we get VLMs to move their eyes—and reason step-by-step in visually grounded ways? 👀 We introduce ViGoRL, a RL method that anchors reasoning to image regions. 🎯 It outperforms vanilla GRPO and SFT across grounding, spatial tasks, and visual search (86.4% on V*). 👇🧵

Gabriel Sarch

76,824 Aufrufe • vor 1 Jahr

With Google Flow Tools, you can now use natural language to create bespoke tools and workflows. In this video, you'll see how you can: ➡️ Explore and try the various Tools within Google Flow ➡️ Remix a tool to make it even more relevant for your unique workflow ➡️Create your own tool from scratch with a natural language prompt ➡️ Share your tools with a friend for them to try out Whether you’re looking for an image editor, video resizer or custom shaders, you can now create them in Google Flow, no coding experience required.
2:30

With Google Flow Tools, you can now use natural language to create bespoke tools and workflows. In this video, you'll see how you can: ➡️ Explore and try the various Tools within Google Flow ➡️ Remix a tool to make it even more relevant for your unique workflow ➡️Create your own tool from scratch with a natural language prompt ➡️ Share your tools with a friend for them to try out Whether you’re looking for an image editor, video resizer or custom shaders, you can now create them in Google Flow, no coding experience required.

Google Flow

20,037 Aufrufe • vor 4 Monaten

BREAKING: You can now build your own video game with Google. Google's new AI tool, Playground, creates playable games without writing a single line of code. You can even change the game by simply typing what you want. Here's how to try it:
1:26

BREAKING: You can now build your own video game with Google. Google's new AI tool, Playground, creates playable games without writing a single line of code. You can even change the game by simply typing what you want. Here's how to try it:

The AI Colony

10,723 Aufrufe • vor 24 Tagen

We integrated Gemini Omni 1.1 Flash and a new suite of creative controls into so anyone with a Google account can now create in Google Vids for free. Here's how you can use it to create videos — without needing professional editing skills.🧵↓
1:06

We integrated Gemini Omni 1.1 Flash and a new suite of creative controls into so anyone with a Google account can now create in Google Vids for free. Here's how you can use it to create videos — without needing professional editing skills.🧵↓

News from Google

117,317 Aufrufe • vor 1 Monat

Kirsten Dunst and Gabrielle Union reunite once again for a new campaign promoting Google Shopping’s Try It On feature.
1:19

Kirsten Dunst and Gabrielle Union reunite once again for a new campaign promoting Google Shopping’s Try It On feature.

Kirsten Dunst Updates

1,183,260 Aufrufe • vor 10 Monaten

We are back with a new episode of What's New in Google Accessibility. This update covers recent features for Android , Google Pixel , YouTube , Chrome, and Google Gemini that help make our products more accessible. Watch the video for the details. #GoogleAccessibility #Accessibility #Google #a11y Google
7:42

We are back with a new episode of What's New in Google Accessibility. This update covers recent features for Android , Google Pixel , YouTube , Chrome, and Google Gemini that help make our products more accessible. Watch the video for the details. #GoogleAccessibility #Accessibility #Google #a11y Google

Google Accessibility

34,178 Aufrufe • vor 6 Monaten

Interviewed at Google for 6.5 months. 6 rounds. Every round was a diagnostic, not a gate. Google didn’t give me the job. It gave me the standard.
0:44

Interviewed at Google for 6.5 months. 6 rounds. Every round was a diagnostic, not a gate. Google didn’t give me the job. It gave me the standard.

Agentic Amit

67,124 Aufrufe • vor 3 Monaten