A native-Rust deep learning framework with explicit memory safety and hardware acceleration
3D parallelism for RUMUS: Tensor Parallelism, Pipeline Parallelism, async collectives
Sparse graph engine for RUMUS — WebGPU-native fused SpMM for GNNs
Proc macros for the RUMUS deep learning framework
High-throughput inference server for RUMUS with continuous batching
Spatial CNN engine for RUMUS — direct-convolution and pooling via WebGPU