Ë ” ij&ãóô—ddlZddlZddlmZddlmZmZmZddlm Z m Z m Z ddl m Z mZmZmZddlmZgd¢Zed¬ «Gd „d ««ZGd „d ej*«ZGd„dej*«Zy)éN)Ú dataclass)ÚAnyÚOptionalÚUnion)ÚMetadataÚ MetadataIndexÚ StorageMeta)ÚLoadPlanÚ LoadPlannerÚSavePlanÚ SavePlanner)ÚFuture)Ú WriteResultÚ StorageWriterÚ StorageReaderT)Úfrozencó,—eZdZUeed<eed<eed<y)rÚindexÚ size_in_bytesÚ storage_dataN)Ú__name__Ú __module__Ú __qualname__rÚ__annotations__Úintr©óú^/var/www/html/engine/venv/lib/python3.12/site-packages/torch/distributed/checkpoint/storage.pyrrs…à ÓàÓØÔrrc óâ—eZdZdZej ddeeejdfddfd„«Z ej de de de ddfd „«Z ej d edefd „«Zej d eedeefd „«Zej d ededeeefd„«Zej dedeeeddfd„«Zeej deeejfde fd„««Zdeefd„Zy)ra8 Interface used by ``save_state_dict`` to write to storage. One StorageWriter instance acts as both the coordinator and the follower in a distributed checkpoint. As part of initialization, each instance is told its role. A subclass should expect the following sequence of calls. 0) (all ranks) set checkpoint_id if users pass a valid checkpoint_id. 1) (all ranks) set_up_storage_writer() 2) (all ranks) prepare_local_plan() 3) (coordinator) prepare_global_plan() 4) (all ranks) write_data() 5) (coordinator) finish() NÚ checkpoint_idÚreturncó—y)a© Calls to indicates a brand new checkpoint write is going to happen. A checkpoint_id may be present if users set the checkpoint_id for this checkpoint write. The meaning of the checkpiont_id is storage-dependent. It can be a path to a folder/file or a key for a key-value storage. Args: checkpoint_id (Union[str, os.PathLike, None]): The ID of this checkpoint instance. The meaning of the checkpoint_id depends on the storage. It can be a path to a folder or to a file. It can also be a key if the storage is a key-value store. (Default: ``None``) Nr©Úselfr s rÚresetzStorageWriter.reset-ó€ð rÚis_coordinatorÚargsÚkwargscó—y)z± Initialize this instance. Args: is_coordinator (bool): Whether this instance is responsible for coordinating the checkpoint. Nr)r$r'r(r)s rÚset_up_storage_writerz#StorageWriter.set_up_storage_writer?órÚplancó—y)a‰ Perform storage-specific local planning. While this method can produce a completely different plan, the recommended way is to store storage specific data in SavePlan::storage_data. Args: plan (SavePlan): The local plan from the ``SavePlanner`` in use. Returns: A transformed ``SavePlan`` after storage local planning Nr©r$r-s rÚprepare_local_planz StorageWriter.prepare_local_planKr,rÚplanscó—y)aÌ Perform centralized planning of storage. This method is only called on the coordinator instance. While this method can produce a completely different plan, the preferred way is to store storage specific data in SavePlan::storage_data. Args: plans: A list of ``SavePlan`` instances, one for each rank. Returns: A list of transformed ``SavePlan`` after storage global planning Nr©r$r1s rÚprepare_global_planz!StorageWriter.prepare_global_planZr,rÚplannercó—y)a Write all items from ``plan`` using ``planner`` to resolve the data. A subclass should call ``SavePlanner::resolve_data`` on each item from the plan to get access to the underlying object to write. Subclasses should lazily call `resolve_data` as it can allocate memory. In case of tensors, make following assumptions: - They might be on any device, including not matching the one on ``WriteItem::tensor_data`` - They might be views or not contiguous. Only the projection needs to be saved. Args: plan (SavePlan): The save plan to execute. planner (SavePlanner): Planner object to be used to resolve items to data. Returns: A future that completes to a list of WriteResult Nr©r$r-r5s rÚ write_datazStorageWriter.write_datakr,rÚmetadataÚresultscó—y)a½ Write the metadata and marks the current checkpoint as successful. The actual format/schema used for serializing `metadata` is an implementation detail. The only requirement is that it's recoverable in to the same object graph. Args: metadata (Metadata): metadata for the new checkpoint results: A list of WriteResults from all ranks. Returns: None Nr)r$r9r:s rÚfinishzStorageWriter.finishƒr,rcó—y©zŒ Check if the given checkpoint_id is supported by the storage. This allow us to enable automatic storage selection. Nr©Úclsr s rÚvalidate_checkpoint_idz$StorageWriter.validate_checkpoint_id”ó€ð rcó—y)a= Return the storage-specific metadata. This is used to store additional information in a checkpoint that can be useful for providing request-level observability. StorageMeta is passed to the ``SavePlanner`` during save calls. Returns None by default. TODO: provide an example Nr)r$s rÚ storage_metazStorageWriter.storage_metas€ðr©N)rrrÚ__doc__ÚabcÚabstractmethodrÚstrÚosÚPathLiker%Úboolrr+r r0Úlistr4r rrr8rr<Ú classmethodrArr rDrrrrrs•„ñð" ×Ññ  5¨¨b¯k©k¸4Ð)?Ñ#@ð ÈDò óð ð" ×Ñð  Ø"ð  Ø+.ð  Ø:=ð  à ò  óð  ð ×Ñð   xð  °Hò  óð  ð ×Ñ𠨨h©ð ¸DÀ¹Nò óð ð  ×Ñð Øð Ø'2ð à [Ñ!Ñ "ò óð ð. ×Ñð ˜xð °$°t¸KÑ7HÑ2Ið Èdò óð ð Ø×Ñð °5¸¸b¿k¹kÐ9IÑ3Jð Ètò óóð ð˜h {Ñ3ôrrc ó—eZdZdZej ddeeejdfddfd„«Z ej de de de fd„«Z ej d e d ede de ddf d „«Zej d edefd „«Zej deedeefd„«Zej d edededfd„«Zeej deeejfdefd„««Zy)raV Interface used by ``load_state_dict`` to read from storage. One StorageReader instance acts as both the coordinator and the follower in a distributed checkpoint. As part of initialization, each instance is told its role. A subclass should expected the following sequence of calls by ``load_state_dict``: 0) (all ranks) set checkpoint_id if users pass a valid checkpoint_id. 1) (all ranks) read_metadata() 2) (all ranks) set_up_storage_reader() 3) (all ranks) prepare_local_plan() 4) (coordinator) prepare_global_plan() 5) (all ranks) read_data() Nr r!có—y)a± Calls to indicates a brand new checkpoint read is going to happen. A checkpoint_id may be present if users set the checkpoint_id for this checkpoint read. The meaning of the checkpiont_id is storage-dependent. It can be a path to a folder/file or a key for a key-value storage. Args: checkpoint_id (Union[str, os.PathLike, None]): The ID of this checkpoint instance. The meaning of the checkpoint_id depends on the storage. It can be a path to a folder or to a file. It can also be a key if the storage is more like a key-value store. (Default: ``None``) Nrr#s rr%zStorageReader.resetºr&rr(r)có—y)z Read the checkpoint metadata. Returns: The metadata object associated with the checkpoint being loaded. Nr)r$r(r)s rÚ read_metadatazStorageReader.read_metadataÌr,rr9r'có—y)zî Initialize this instance. Args: metadata (Metadata): The metadata schema to use. is_coordinator (bool): Whether this instance is responsible for coordinating the checkpoint. Nr)r$r9r'r(r)s rÚset_up_storage_readerz#StorageReader.set_up_storage_readerÖr,rr-có—y)a† Perform storage-specific local planning. While this method can produce a completely different plan, the recommended way is to store storage specific data in LoadPlan::storage_data. Args: plan (LoadPlan): The local plan from the ``LoadPlan`` in use. Returns: A transformed ``LoadPlan`` after storage local planning Nrr/s rr0z StorageReader.prepare_local_planãr,rr1có—y)aÔ Perform centralized planning of storage loading. This method is only called on the coordinator instance. While this method can produce a completely different plan, the preferred way is to store storage specific data in LoadPlan::storage_data. Args: plans: A list of ``LoadPlan`` instances, one for each rank. Returns: A list of transformed ``LoadPlan`` after storage global planning Nrr3s rr4z!StorageReader.prepare_global_planòr,rr5có—y)a« Read all items from ``plan`` using ``planner`` to resolve the data. A subclass should call ``LoadPlanner::load_bytes`` to deserialize a BytesIO object into the right place. A subclass should call ``LoadPlanner::resolve_tensor`` to get access to the tensors that in should load data into. It's the StorageLayer responsibility to properly schedule any cross device copies required. Args: plan (LoadPlan): The local plan to execute on planner (LoadPlanner): The planner object to use to resolve items. Returns: A future that completes once all reads are finished. Nrr7s rÚ read_datazStorageReader.read_datar,rcó—yr>rr?s rrAz$StorageReader.validate_checkpoint_idrBrrE)rrrrFrGrHrrIrJrKr%rrrRrLrTr r0rMr4r rrXrNrArrrrr¨sy„ñð" ×Ññ  5¨¨b¯k©k¸4Ð)?Ñ#@ð ÈDò óð ð" ×Ñð  3ð °#ð ¸(ò óð ð ×Ñð  Ø ð  Ø26ð  Ø?Bð  ØNQð  à ò  óð  ð ×Ñð   xð  °Hò  óð  ð ×Ñ𠨨h©ð ¸DÀ¹Nò óð ð  ×Ñð ˜hð °ð ÀÈÁò óð ð*Ø×Ñð °5¸¸b¿k¹kÐ9IÑ3Jð Ètò óóñ rr)rGrJÚ dataclassesrÚtypingrrrÚ%torch.distributed.checkpoint.metadatarrr Ú$torch.distributed.checkpoint.plannerr r r r Ú torch.futuresrÚ__all__rÚABCrrrrrúrassðÛ Û Ý!ß'Ñ'çVÑV÷óõ !ò <€ñ $Ô÷ðóðôJC—G‘GôJôZx C—G‘Gõx r