RobuQ delivers the first stable DiT image generation at W1.58A2 average bits via Hadamard-based robust activation quantization and layer-wise mixed-precision activations.
Pytorch: An imperative style, high-performance deep learning library
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
fields
cs.CV 2representative citing papers
OmniHands adds relation-aware tokenization and 4D attention fusion to a transformer so it can reconstruct interactive hand meshes from monocular or multi-view inputs.
citing papers explorer
-
RobuQ: Pushing DiTs to W1.58A2 via Robust Activation Quantization
RobuQ delivers the first stable DiT image generation at W1.58A2 average bits via Hadamard-based robust activation quantization and layer-wise mixed-precision activations.
-
OmniHands: Towards Robust 4D Hand Mesh Recovery via A Versatile Transformer
OmniHands adds relation-aware tokenization and 4D attention fusion to a transformer so it can reconstruct interactive hand meshes from monocular or multi-view inputs.