loadStreaming
suspend fun <T : DType, V> loadStreaming(ctx: ExecutionContext, dtype: KClass<T>, onTensorLoaded: (String, Tensor<T, V>) -> Unit): Gemma3nModelMetadata(source)
Load weights using streaming API - parses metadata only, loads tensors on-demand. Requires randomAccessProvider constructor.
inline suspend fun <T : DType, V> loadStreaming(ctx: ExecutionContext, noinline onTensorLoaded: (String, Tensor<T, V>) -> Unit): Gemma3nModelMetadata(source)