public DeviceQuantizedMatrix(reader as SafetensorsReader, info as TensorInfo, device as Device, buffer as Byte[])
Loads a published F32/F16/BF16 Safetensors matrix in its original encoding, without silently quantizing it. Shape and byte ranges come from the validated reader. The caller owns the reader; upload memory stays bounded.
| Parameter | Description |
|---|---|
reader | Validated shared reader. @param info Exact tensor descriptor. |
device | Execution selector. @param buffer Reusable upload storage. |
matrix = DeviceQuantizedMatrix(reader, info, Device.cpu, buffer)