pylibwholegraph.torch.embedding.WholeMemoryEmbedding#
- class pylibwholegraph.torch.embedding.WholeMemoryEmbedding(wmb_embedding: PyWholeMemoryEmbedding, wmb_cache_policy: Optional[WholeMemoryCachePolicy])#
WholeMemory Embedding
- Attributes:
- shape
Methods
add_gradients
apply_gradients
dim
drop_all_cache
gather
get_embedding_tensor
get_optimizer_state
get_optimizer_state_names
load
need_grad
save
set_adjust_cache
writeback_all_cache
- __init__(wmb_embedding: PyWholeMemoryEmbedding, wmb_cache_policy: Optional[WholeMemoryCachePolicy])#
Methods
__init__
(wmb_embedding, wmb_cache_policy)add_gradients
(indice, grad_outputs)apply_gradients
(lr)dim
()drop_all_cache
()gather
(indice, *[, is_training, force_dtype])get_embedding_tensor
()get_optimizer_state
(state_name)get_optimizer_state_names
()load
(file_prefix, *[, ignore_embedding, ...])need_grad
()save
(file_prefix)set_adjust_cache
(adjust_cache)writeback_all_cache
()Attributes
shape