Raphael Liu

RaphaelLiu's activity

• All
• Models
• Datasets
• Spaces
• Papers
• Collections
• Community
• Posts
• Upvotes
• Likes

New activity in ArtificialAnalysis/Video-Generation-Arena-Leaderboard 10 days ago

When include new models like SORA and HunYuan?

2

#2 opened 12 days ago by

RaphaelLiu

upvoted a paper 20 days ago

ROICtrl: Boosting Instance Control for Visual Generation

Paper • 2411.17949 • Published 26 days ago • 82

liked a model about 2 months ago

Etched/oasis-500m

Updated Nov 4 • 464 • 431

liked a dataset about 2 months ago

Koala-36M/Koala-36M-v1

Viewer • Updated Oct 12 • 36M • 511 • 22

reacted to singhsidhukuldeep's post with ❤️ about 2 months ago

view post

Post

2839

If you have ~300+ GB of V-RAM, you can run Mochi from @genmo

A SOTA model that dramatically closes the gap between closed and open video generation models.

Mochi 1 introduces revolutionary architecture featuring joint reasoning over 44,520 video tokens with full 3D attention. The model implements extended learnable rotary positional embeddings (RoPE) in three dimensions, with network-learned mixing frequencies for space and time axes.

The model incorporates cutting-edge improvements, including:
- SwiGLU feedforward layers
- Query-key normalization for enhanced stability
- Sandwich normalization for controlled internal activations

What is currently available?
The base model delivers impressive 480p video generation with exceptional motion quality and prompt adherence. Released under the Apache 2.0 license, it's freely available for both personal and commercial applications.

What's Coming?
Genmo has announced Mochi 1 HD, scheduled for release later this year, which will feature:
- Enhanced 720p resolution
- Improved motion fidelity
- Better handling of complex scene warping

• 
• 
• 2 replies

·

❤️

11

11

🔥

4

4

👀

2

2

+

Reply

liked a Space 2 months ago

Running

102

🏆

Ocrbench Leaderboard

authored a paper 2 months ago

Redefining Temporal Modeling in Video Diffusion: The Vectorized Timestep Approach

Paper • 2410.03160 • Published Oct 4 • 4

upvoted a paper 3 months ago

Redefining Temporal Modeling in Video Diffusion: The Vectorized Timestep Approach

Paper • 2410.03160 • Published Oct 4 • 4

commented a paper 3 months ago

Redefining Temporal Modeling in Video Diffusion: The Vectorized Timestep Approach

Paper • 2410.03160 • Published Oct 4 • 4 •

2

upvoted a paper 3 months ago

Open-MAGVIT2: An Open-Source Project Toward Democratizing Auto-regressive Visual Generation

Paper • 2409.04410 • Published Sep 6 • 23

liked 2 Spaces 4 months ago

Running on Zero

45

📈

K-Sort Arena

Efficient Image/Video K-Sort Arena

Running on Zero

261

📈

GenAI Arena

Realtime Image/Video Gen AI Arena

liked 6 models 5 months ago

THUDM/CogVideoX-2b

Text-to-Video • Updated 30 days ago • 32.6k • 307

datajuicer/Data-Juicer-T2V

Text-to-Video • Updated Jul 17 • 3

mistralai/Mistral-Large-Instruct-2407

Updated Oct 16 • 2.76M • 812

Kwai-Kolors/Kolors

Text-to-Image • Updated Jul 12 • 1.44k • 732

fal/AuraFlow

Text-to-Image • Updated Jul 18 • 5.1k • 634

xinsir/controlnet-union-sdxl-1.0

Text-to-Image • Updated Jul 30 • 76.1k • 1.19k

liked a dataset 6 months ago

TIGER-Lab/VideoScore-Bench

Viewer • Updated Jun 26 • 7.44k • 187 • 2

liked a Space 9 months ago

Running

147

📊

VBench Leaderboard