Загрузка видео...

Не удалось загрузить видео

На главную

Flow-based generative models trained with flow matching tend to learn curved trajectories, which are challenging to approximate in a few steps. Rectified flows aim to learn straight trajectories, which are easier to simulate with less computation.

alec_helbling's profile picture

Alec Helbling

10,182 subscribers

34,778 просмотров • 5 месяцев назад •via X (Twitter)

LIVE

Anya Rossi• Live Now

Private livecam show

Комментарии: 0

Нет доступных комментариев

Здесь появятся комментарии из оригинального поста

Похожие видео

In flow matching, a coupling determines how noise and data samples are paired during training. The choice of coupling is important because it influences the geometry of trajectories at inference time. The simplest choice is the independent coupling, where noise and data points are paired arbitrarily. This can lead to curved trajectories as the model averages over many conflicting pairings. However, if we use optimal transport on batches of pairs, this leads to fewer ambiguous intersections that the model must resolve, leading to straighter trajectories at inference time.
0:18

In flow matching, a coupling determines how noise and data samples are paired during training. The choice of coupling is important because it influences the geometry of trajectories at inference time. The simplest choice is the independent coupling, where noise and data points are paired arbitrarily. This can lead to curved trajectories as the model averages over many conflicting pairings. However, if we use optimal transport on batches of pairs, this leads to fewer ambiguous intersections that the model must resolve, leading to straighter trajectories at inference time.

Alec Helbling

65,060 просмотров • 1 месяц назад

Agent workflow, tinkering with agent workflow visualisation, designed in Figma, animated with Motion It was also a good opportunity to learn xyflow · React Flow · Svelte Flow
0:13

Agent workflow, tinkering with agent workflow visualisation, designed in Figma, animated with Motion It was also a good opportunity to learn xyflow · React Flow · Svelte Flow

Jeet

14,182 просмотров • 1 месяц назад

Flow matching produces smooth, deterministic trajectories. In contrast, the sampling process of a diffusion model is chaotic, resembling the random motion of gas particles.
0:13

Flow matching produces smooth, deterministic trajectories. In contrast, the sampling process of a diffusion model is chaotic, resembling the random motion of gas particles.

Alec Helbling

428,834 просмотров • 1 год назад

We are also releasing self-contained lecture notes that explain flow matching and diffusion models from scratch. This goes from "zero" to the state-of-the-art in modern Generative AI. 📖 Read the notes here: Joint work with Ezra Erives.
0:15

We are also releasing self-contained lecture notes that explain flow matching and diffusion models from scratch. This goes from "zero" to the state-of-the-art in modern Generative AI. 📖 Read the notes here: Joint work with Ezra Erives.

Peter Holderrieth

474,533 просмотров • 3 месяцев назад

Flow Matching aims to learn a "flow" that transforms a simple source distribution (e.g. Gaussian) to an arbitrarily complex target distribution. This video shows the evolution of the marginal probability path as a source distribution is transformed to a target distribution.
0:13

Flow Matching aims to learn a "flow" that transforms a simple source distribution (e.g. Gaussian) to an arbitrarily complex target distribution. This video shows the evolution of the marginal probability path as a source distribution is transformed to a target distribution.

Alec Helbling

150,399 просмотров • 1 год назад

The #RTXRemix REST API integration with ComfyUI will enable modders to remaster the classics with 1,000s of generative AI models. Learn More →
0:10

The #RTXRemix REST API integration with ComfyUI will enable modders to remaster the classics with 1,000s of generative AI models. Learn More →

NVIDIA GeForce

76,660 просмотров • 2 лет назад

3D-printing enables the fabrication of miniaturized, ultrafast delta robots that can follow trajectories with extreme precision and generate enough power to launch projectiles into the air. Learn more in Science Robotics:
0:14

3D-printing enables the fabrication of miniaturized, ultrafast delta robots that can follow trajectories with extreme precision and generate enough power to launch projectiles into the air. Learn more in Science Robotics:

Science Magazine

15,809 просмотров • 6 месяцев назад

New research from BlackForestLabsAI - Unofficial 🥳 Meet Self-Flow: our self-supervised framework for image, audio, video & world models 🤖 Do generative models really need DINO to learn strong representations? We propose teaching them directly via a joint framework instead 🧵
2:39

New research from BlackForestLabsAI - Unofficial 🥳 Meet Self-Flow: our self-supervised framework for image, audio, video & world models 🤖 Do generative models really need DINO to learn strong representations? We propose teaching them directly via a joint framework instead 🧵

Hila Chefer

67,523 просмотров • 3 месяцев назад

“When you go to learn from a magician, you tend to learn magic” 😂🤣
0:17

“When you go to learn from a magician, you tend to learn magic” 😂🤣

Urban Cowboy

50,299 просмотров • 1 год назад

🇮🇷| Armed rioters were observed firing with Kalashnikov in Ilam city, few hours ago Ilam is 40km from the border with Iraq, and is easier for rioters to gain weapons from weapon smugglers. These are trained individuals with the sole aim of creating chaos.
0:12

🇮🇷| Armed rioters were observed firing with Kalashnikov in Ilam city, few hours ago Ilam is 40km from the border with Iraq, and is easier for rioters to gain weapons from weapon smugglers. These are trained individuals with the sole aim of creating chaos.

Arya - آریا

50,024 просмотров • 5 месяцев назад

Sometimes Flow is refered to as a silent film, which isn't true. Dialogue free would be a better description. Without dialogue we can use the other tools of cinema in a much more expressive way. The music, the sound, the camera, lighting and animation don't have to compete with the dialogue which can tend to attract all the attention.
0:15

Sometimes Flow is refered to as a silent film, which isn't true. Dialogue free would be a better description. Without dialogue we can use the other tools of cinema in a much more expressive way. The music, the sound, the camera, lighting and animation don't have to compete with the dialogue which can tend to attract all the attention.

Gints Zilbalodis

163,338 просмотров • 1 год назад

Within Quai Network, cross-chain transactions travel through coincident blocks, which are blocks that are simultaneously valid in multiple chains. Coincident blocks are produced by miners, and enable liquidity to flow seamlessly between Quai chains with no centralized bridges.
0:11

Within Quai Network, cross-chain transactions travel through coincident blocks, which are blocks that are simultaneously valid in multiple chains. Coincident blocks are produced by miners, and enable liquidity to flow seamlessly between Quai chains with no centralized bridges.

Quai Network ⚡️💵

16,795 просмотров • 2 лет назад

hi all, excited to join! i'm building an expressive mini "shoggoth" robot which will eventually be hooked up to gpt4o realtime voice. i'm currently working on the low-level policies, which are trained in a mujoco simulation with RL. to delay working on raw-pixels for now, i trained a pose-estimation model using deeplabcut and triangulate the position in 3d space using the stereo cameras. eventually, i'll use gpt4o's tool calling capabilities to activate several of these policies (closed and open loop) based on the dialog flow! captions: manual actuation of the tentacle / 3d pose estimation / target design
0:29

hi all, excited to join! i'm building an expressive mini "shoggoth" robot which will eventually be hooked up to gpt4o realtime voice. i'm currently working on the low-level policies, which are trained in a mujoco simulation with RL. to delay working on raw-pixels for now, i trained a pose-estimation model using deeplabcut and triangulate the position in 3d space using the stereo cameras. eventually, i'll use gpt4o's tool calling capabilities to activate several of these policies (closed and open loop) based on the dialog flow! captions: manual actuation of the tentacle / 3d pose estimation / target design

Matthieu LC

45,506 просмотров • 1 год назад

Instead of going to filmschool, I made my first feature film Away which I started at 20 and finished at 24. I think the best way to learn is by making films so I did everything on my own. It went on to win the Contrechamp award in Annecy which gave me the opportunity to make Flow
0:18

Instead of going to filmschool, I made my first feature film Away which I started at 20 and finished at 24. I think the best way to learn is by making films so I did everything on my own. It went on to win the Contrechamp award in Annecy which gave me the opportunity to make Flow

Gints Zilbalodis

1,345,810 просмотров • 1 год назад

From the beginning, Flow has been a tool built by artists, for artists. In that spirit, we have some news: 1) Welcome Henry Daubrez 🌸💀 as our first filmmaker in residence who will work directly with our team to shape the future of Flow. 2) We're launching Flow Sessions, a pilot program that will provide mentorship and education for filmmakers on how they can best use Flow to bring their stories to life. Learn more:
0:08

From the beginning, Flow has been a tool built by artists, for artists. In that spirit, we have some news: 1) Welcome Henry Daubrez 🌸💀 as our first filmmaker in residence who will work directly with our team to shape the future of Flow. 2) We're launching Flow Sessions, a pilot program that will provide mentorship and education for filmmakers on how they can best use Flow to bring their stories to life. Learn more:

Flow by Google

137,072 просмотров • 9 месяцев назад

A quick CFD test visualizes the flow geometry which might have contributed to these features.
0:19

A quick CFD test visualizes the flow geometry which might have contributed to these features.

Craig Stone

27,651 просмотров • 3 месяцев назад

Having seen a few comments, I’d like to clarify something. We haven’t used generative AI in the design team. We have the human capabilities to create what we want, and when our capabilities are limited, we learn and strive to achieve our goals
0:14

Having seen a few comments, I’d like to clarify something. We haven’t used generative AI in the design team. We have the human capabilities to create what we want, and when our capabilities are limited, we learn and strive to achieve our goals

Adrián San Gil | G2 Enelthion

221,977 просмотров • 3 месяцев назад

Longer tracks are here with Lyria 3 Pro in Gemini! From experimenting with different styles to generating tracks with complex transitions, Lyria 3 Pro makes it easier to bring your full vision to life. Rolling out today to Google AI Plus, Pro, and Ultra users. Learn more 🧵
0:14

Longer tracks are here with Lyria 3 Pro in Gemini! From experimenting with different styles to generating tracks with complex transitions, Lyria 3 Pro makes it easier to bring your full vision to life. Rolling out today to Google AI Plus, Pro, and Ultra users. Learn more 🧵

Google Gemini

6,725,137 просмотров • 2 месяцев назад