Structs§
- Batch
Inference - Batch inference helper. Currently delegates to CPU-based predict_batch. Future: keep intermediate data on GPU between batches.
- GpuDevice
- A handle to a GPU device and its command queue.
- GpuOps
- GPU operations dispatcher with cached pipelines.
- GpuTensor
- A tensor stored on the GPU as an f32 buffer.
Enums§
- DType
- Data type for GPU tensors.