A PyTorch-native inference engine with cache, parallelism, quantization and cpu offload for 100+ DiTs.
-
Updated
Nov 13, 2026 - Python
#
svdquant
Here are 7 public repositories matching this topic...
Hybrid Automatic Video Colorizer (HAVC) server that exposes a GPU-accelerated colorization pipeline for B&W images and video frames based on Diffusion Transformer (DiT) models: Qwen-Image-Edit-2511, Qwen-Image 2.1 and LongCat ImageEdit Turbo. Colorized reference frames are propagated using CMNET2.
-
Updated
Oct 25, 2026 - Python
W4A4 and INT8 KV-cache quantization for Infinity VAR models. Optimized for high-fidelity generative AI deployment on edge GPUs (e.g. NVIDIA Jetson).
computer-vision pytorch gpu-acceleration quantization model-compression nvidia-jetson inference-optimization edge-ai on-device-ml weight-quantization post-training-quantization autoregressive-models generative-ai kv-cache-quantization activation-quantization visual-autoregressive-model svdquant infinity-var
-
Updated
Jun 11, 2026 - Python
4-bit SVDQuant (Nunchaku) diffusion inference on AMD GPUs (ROCm): Triton backend, native RDNA3 int4 kernels, FLUX.1, FLUX.2 [klein], Qwen-Image, Z-Image-Turbo, SDXL, LoRA, ComfyUI nodes
flux amd pytorch triton lora quantization text-to-image rocm radeon 4-bit amd-gpu diffusion-models int4 generative-ai comfyui sdxl nunchaku qwen-image z-image svdquant
-
Updated
Nov 11, 2026 - Python
Krea 2 Turbo at 4-bit weights and activations on Nunchaku fused kernels. Runtime port + deepcompressor conversion.
-
Updated
Jul 21, 2026 - Python
ComfyUI nodes for Nunchaku 4-bit models on AMD GPUs (ROCm): FLUX.2 [klein], FLUX.1, Qwen-Image, Z-Image-Turbo, SDXL
-
Updated
Nov 12, 2026 - Python
Add this topic to your repo
To associate your repository with the svdquant topic, visit your repo's landing page and select "manage topics."