DataAI

Millions of high-quality video data for AI.

[DataAI @DataAI_THU]

低价出售😀
1. 千万级别 “无水印”“无字幕”“超高质量”“1080P摄影视频”
2. 千万级别 “无水印”“无字幕”“超高质量”“2k摄影图像”
3. 百万级别 S站 3D glb数据

[DataAI @DataAI_THU]

低价出售😀
1. 千万级别 “无水印”“无字幕”“超高质量”“1080P摄影视频”
2. 千万级别 “无水印”“无字幕”“超高质量”“2k摄影图像”
3. 百万级别 S站 3D glb数据

[DataAI @DataAI_THU]

Sell at a low price 😀
1. Million level "no watermark", "no subtitles", "ultra-high quality", "1080P photography video"
2. Million level "no watermark", "no subtitles", "ultra-high quality", "2k photographic images"
3. Million level S station 3D glb data

[DataAI @DataAI_THU]

低价出售😀
1. 千万级别 “无水印”“无字幕”“超高质量”“1080P摄影视频”
2. 千万级别 “无水印”“无字幕”“超高质量”“2k摄影图像”
3. 百万级别 S站 3D glb数据

[DataAI @DataAI_THU]

@yourtoolscc May you help me?

[DataAI @DataAI_THU]

@yourtoolscc 1

[DataAI @DataAI_THU]

@yourtoolscc hello, sir, I encountered an Authentication problem when logging in to my account, can you help me?

[DataAI @DataAI_THU]

如何训练Sora这样的视频生成模型？数据一定是最关键的部分！#Sora #dataset #video-generation

[DataAI @DataAI_THU]

DTrain SOTA text-to-video models with our vast collection of high-quality 720P portrait videos, sourced from top professional platforms! #PortraitVideos

[DataAI @DataAI_THU]

微信： 15217028256

[DataAI @DataAI_THU]

📢 High-quality 720P portrait videos sourced from professional platforms.

👋Hundreds of thousands.

😎We are eager to collaborate with AI companies.

👉 Feel free to reach out if you would like to discuss collaboration details!

#VideoData #PortraitVideos

[DataAI @DataAI_THU]

📢 High-quality 720P portrait videos sourced from professional platforms.
😇Hundreds of thousands.
📷 Feel free to reach out if you would like to discuss collaboration details!

[AK @_akhaliq]

DeepFaceEditing: Deep Face Generation and Editing with Disentangled Geometry and Appearance Control
project page: https://t.co/4YkKljS4KM
github: https://t.co/3V5hU78Srr

[DataAI @DataAI_THU]

📢 High-quality 720P portrait videos sourced from professional platforms.
😀Hundreds of thousands.
📷 Feel free to reach out if you would like to discuss collaboration details!

[OutofAi @OutofAi]

We created a series of simplified notebooks that cover essential aspects of Stable Diffusion, using the vanilla Stable Diffusion 2.1 base to utilise it as a face-editing model for building your own face app 🧵(1/3)

Github : https://t.co/il7DYkccYn

[DataAI @DataAI_THU]

We have a vast collection of high-quality 720P portrait videos sourced from professional platforms like iStock, totaling millions. We are eager to collaborate with researchers!

[DataAI @DataAI_THU]

We have a vast collection of high-quality 720P portrait videos sourced from professional platforms like iStock, totaling hundreds of thousands. We are eager to collaborate with AI companies.

[AK @_akhaliq]

TrackGo

A Flexible and Efficient Method for Controllable Video Generation

discuss: https://t.co/Xf460CdwRf

Recent years have seen substantial progress in diffusion-based controllable video generation. However, achieving precise control in complex scenarios, including fine-grained object parts, sophisticated motion trajectories, and coherent background movement, remains a challenge. In this paper, we introduce TrackGo, a novel approach that leverages free-form masks and arrows for conditional video generation. This method offers users with a flexible and precise mechanism for manipulating video content. We also propose the TrackAdapter for control implementation, an efficient and lightweight adapter designed to be seamlessly integrated into the temporal self-attention layers of a pretrained video generation model. This design leverages our observation that the attention map of these layers can accurately activate regions corresponding to motion in videos. Our experimental results demonstrate that our new approach, enhanced by the TrackAdapter, achieves state-of-the-art performance on key metrics such as FVD, FID, and ObjMC scores.

[AK @_akhaliq]

TrackGo

A Flexible and Efficient Method for Controllable Video Generation

discuss: https://t.co/Xf460CdwRf

Recent years have seen substantial progress in diffusion-based controllable video generation. However, achieving precise control in complex scenarios, including fine-grained object parts, sophisticated motion trajectories, and coherent background movement, remains a challenge. In this paper, we introduce TrackGo, a novel approach that leverages free-form masks and arrows for conditional video generation. This method offers users with a flexible and precise mechanism for manipulating video content. We also propose the TrackAdapter for control implementation, an efficient and lightweight adapter designed to be seamlessly integrated into the temporal self-attention layers of a pretrained video generation model. This design leverages our observation that the attention map of these layers can accurately activate regions corresponding to motion in videos. Our experimental results demonstrate that our new approach, enhanced by the TrackAdapter, achieves state-of-the-art performance on key metrics such as FVD, FID, and ObjMC scores.

[AK @_akhaliq]

FocusLLM

Scaling LLM's Context by Parallel Decoding

discuss: https://t.co/xaUZeLWoTf

Empowering LLMs with the ability to utilize useful information from a long context is crucial for many downstream applications. However, achieving long context lengths with the conventional transformer architecture requires substantial training and inference resources. In this paper, we present FocusLLM, a framework designed to extend the context length of any decoder-only LLM, enabling the model to focus on relevant information from very long sequences. FocusLLM processes long text inputs by dividing them into chunks based on the model's original context length to alleviate the issue of attention distraction. Then, it appends the local context to each chunk as a prompt to extract essential information from each chunk based on a novel parallel decoding mechanism, and ultimately integrates the extracted information into the local context. FocusLLM stands out for great training efficiency and versatility: trained with an 8K input length with much less training cost than previous methods, FocusLLM exhibits superior performance across downstream long-context tasks and maintains strong language modeling ability when handling extensive long texts, even up to 400K tokens.

[DataAI @DataAI_THU]

📢 High-quality 3D glb models.

📷150w+ glb/gltf format .

📷 Feel free to reach out if you would like to discuss collaboration details!