pylibwholegraph.torch.embedding.create_builtin_cache_policy

pylibwholegraph.torch.embedding.create_builtin_cache_policy(builtin_cache_type: str, embedding_memory_type: str, embedding_memory_location: str, access_type: str, cache_ratio: float, *, cache_memory_type: str = '', cache_memory_location: str = '')

Create builtin cache policy

Parameters:

• builtin_cache_type – supported types are none, local_device, local_node and all_devices

• embedding_memory_type – WholeMemory type of raw embedding

• embedding_memory_location – WholeMemory location of raw embedding

• access_type – Access type needed

• cache_ratio – Ratio of cache

• cache_memory_type – WholeMemory type of cache

• cache_memory_location – WholeMemory location of cache

Returns:

WholeMemoryCachePolicy or None