Hugging Apps

community

AI & ML interests

None defined yet.

Recent Activity

multimodalart published a Space about 3 hours ago

hugging-apps/jeffan-31b-1710

multimodalart updated a Space about 4 hours ago

hugging-apps/jeffan-31b-1710

multimodalart published a Space about 7 hours ago

hugging-apps/trackeverything

View all activity

hugging-apps 's Spaces 688

Jeffan-31b-1710

Decision probabilities from a Gemma-4-31B finetune

TrackEverything

Long-horizon dense 3D point tracking for any video

elephant-1

Typed decision model, a probability for every option

Kimodo Motion

Text & constraints to 3D human motion (Kimodo SOMA)

Granite Speech 5.0 2B

Multilingual speech recognition & translation demo

WinARDet

Dense-scene grounding via compressed window auto-regression

Underdog Bark 3.0 Pro

State-of-the-art open English speech-to-text (0.6B)

ExMR - Extremity Motion Refinement

3D body, hand and foot motion from monocular video

SolarRenderer

Few-step structure-to-video rendering on MiniMax-H3

ActionRoPE

Action-controlled game world model video generation

OvisOCR3 Unified Document Understanding

OCR, parsing, KIE & cross-page tables with OvisOCR3

Policy Structure Extractor (Qwen3.5-9B LoRA)

Turn a policy page image into a nested JSON tree

FastH3 Trim

Prompt to video with synced audio in 8 fast forwards

OXR-1.0 Document Parsing

Document images/PDFs to Markdown & JSON with layout boxes

Ropedia W0

Action-conditioned egocentric world model rollouts

FastH3-Omni

FastVideo 8-step video+audio from image/video/audio refs

Flora-TempSeg

Video temporal segmentation VLM (Qwen3.5-9B LoRA)

FastH3-QAD-v2

MiniMax-H3 in 8 steps — NVFP4, sparse video + audio

MiniMax-H3 Ref2VA · PTD per-hop LoRA

References + prompt to video with audio, in 8 or 4 steps

FLUX 3 Image Klein Consistency

Instruction edits that keep everything else the same

PPMem: Persistent Planar Memory for Video World Models

Camera-posed video world model with planar memory

FineSplat

Feed-forward 3D Gaussian splats from unposed images

SMD MiniMax-H3 Ref2VA (4 & 8 step)

Few-step reference-to-video+audio with MiniMax-H3

LILIE Text-to-Image

JiT-T2I trained with a language-informed loss (LILIE)