skainet-io-gguf/sk.ainet.io.gguf.llama/MmapLlamaLoader/loadToMap

loadToMap

fun <T : DType, V> loadToMap(ctx: ExecutionContext, dtype: KClass<T>): LlamaWeights<T, V>(source)

Load model weights as mmap-backed tensors.

Return

LlamaWeights containing mmap-backed tensors

Parameters

T

the data type (must be FP32 for mmap loading)

V

the value type (Float for FP32)

ctx

execution context (used to wrap TensorData with ops)

inline fun <T : DType, V> loadToMap(ctx: ExecutionContext): LlamaWeights<T, V>(source)