pub async fn value_matmul(a: &Value, b: &Value) -> BuiltinResult<Value>Expand description
GPU-aware matmul entry: if both inputs are GpuTensor handles, call provider; otherwise fall back to CPU.
pub async fn value_matmul(a: &Value, b: &Value) -> BuiltinResult<Value>GPU-aware matmul entry: if both inputs are GpuTensor handles, call provider; otherwise fall back to CPU.