Agent-GenPath
Pathology patch synthesis with the Agent-GenPath DiT
None defined yet.
Pathology patch synthesis with the Agent-GenPath DiT
Transformer-based rendering of 3D Gaussian splats
Visual query segmentation — segment an object across video
Danish ASR with Klang Pianissimo, word timestamps
Page-level table extraction as an image-to-graph model
Typed decision questions over text, JSON, or images
One image to a 3D object in 5 guidance-free distilled steps
Typed questions in, calibrated probabilities out
GenFirst EiT MMDiT-XL/2 512px text-to-image
Indic speech to text and speech to decisions
Two-step audio-driven video dubbing from replacement speech
Qwen3.5-4B with a CHIP-8 computer compiled into its weights
Impute missing MRI/CT modalities with SphereFlow
Two-step video dubbing — re-sync lips to new speech
Draw per-object paths, animate an image (CogVideoX)
Detect AI-generated images with hierarchical MIL attention
One-step ImageNet-256 class-conditional image generation
Pixel-space text-to-image DiT trained for $21K
Translation across 53 languages with a 0.8B specialized LLM
Typed decisions: state in, calibrated probabilities out
1,257 hand-written params, 0 training steps
Few-step autoregressive long-horizon text-to-video
Stream-healed MOSS-Image-Base finetune, text-to-image
Depth, normals and albedo from a single image