An interactive NVIDIA-GPU process viewer and beyond, the one-stop solution for GPU process management.
Open-source deep-learning framework for building, training, and fine-tuning deep learning models using state-of-the-art Physics-ML methods
Optimized Whisper models for streaming and on-device use
Qwen3.8-27B on 16-32 GB Nvidia GPUs one-click install for Windows / Linux
The open .char format for consistent characters. Face, body, wardrobe & voice in one file, consistent across every model.
High-throughput batched ASR for NVIDIA GPUs. Run Zipformer and NVIDIA NeMo Parakeet models through one Python API with TensorRT, custom CUDA plugins, GPU beam search, and word timestamps - up to 25,000 RTFx on B300.