+ jL^RIt^RIt^RIt^RIt^RIHt^RIHt^RIH t ^RI H t H t ^RI Ht^RIt^RIHt^RIHt^RIHt^R IHt^R IHt^R IHt^R IHt^R IH t ^RI!H"t"H#t#^RI$H%t%H&t&H't'^RI(H)t)^RI*H+t+H,t,^RI-H.t.^RI/H0t0H1t1H2t2] 'd^RI3H4t4.R0Ot5!RR] 4t6]!R]7R7R1RRll4t8]!RR7]0RRR RR!RR"RR#RR$R/R%R&ll44t9]!R'R44t:]!RR7RRR RR!RR"RR(]6PvR)RR#RR$R/R*R+ll4t<]!RR7R,R-l4t=R2R.R/llt>R#)3NFuture) dataclass)Enum)cast TYPE_CHECKING) deprecatedSTATE_DICT_TYPE)$_ProcessBasedAsyncCheckpointExecutor)#_ThreadBasedAsyncCheckpointExecutor)_storage_setup)DefaultSavePlanner)_dcp_method_logger)Metadata)SavePlan SavePlanner) AsyncStager DefaultStagerStagingOptions)Stateful) StorageWriter WriteResult)_get_default_group) _api_bc_check _DistWrapper_profile)_AsyncCheckpointExecutorAsyncCheckpointerTypeAsyncSaveResponsec"]tRt^4tRtRtRtRtR#)rz!Enum for async checkpointer type.threadprocessN)__name__ __module__ __qualname____firstlineno____doc__THREADPROCESS__static_attributes__r$n/home/luca/screen-checking/.venv/lib/python3.14/site-packages/torch/distributed/checkpoint/state_dict_saver.pyrr4s+ FGr-za`save_state_dict` is deprecated and will be removed in future versions.Please use `save` instead.)categoryFcV^8dQhR\R\R\PR,R\R\ R\ R,R\/#)  state_dictstorage_writer process_groupNcoordinator_rankno_distplannerreturnr rdist ProcessGroupintboolrr)formats"r. __annotate__r?@s^   ! $$t+     4    r-c VP4\4;_uu_4\VVVVVV4uuRRR4# +'giR#;i)z3This method is deprecated. Please switch to 'save'.N)resetr_save_state_dict)r2r3r4r5r6r7s&&&&&&r.save_state_dictrC;s?          s > A Tlog_exceptions checkpoint_idr3r7r4r6use_collectivescV^8dQhR\R\\P,R,R\R,R\ R,R\ PR,R\R\R \/#) r1r2rFNr3r7r4r6rGr8) r strosPathLikerrr:r;r=r)r>s"r.r?r?Ysr r r $t+r "D( r 4  r $$t+ r r r r r-c \PPR4T;'g=\P!4'*;'g\P !4'*pV'd\ P!R^R7\4;_uu_4\\\W!RR74p\\V4VVVVVR7uuRRR4# +'giR#;i)a, Save a distributed model in SPMD style. This function is different from ``torch.save()`` as it handles ``ShardedTensor`` , and ``DTensor`` by having each rank only save their local shards. For each ``Stateful`` object (having both a ``state_dict`` and a ``load_state_dict``), save will call ``state_dict`` before serialization. .. warning:: There are no guarantees of Backwards Compatibility across PyTorch versions for saved state_dicts. .. warning:: If using the `process_group` argument, make sure that only its ranks call `save_state_dict` and that all data in state_dict belong to it. .. note:: When saving checkpoint for FSDP's `ShardingStrategy.HYBRID_SHARD`, only one of the shard_group should be calling `save_state_dict` and the corresponding process group needs to be passed in. .. note:: If no process group is available, this function assumes the intention is to save the state_dict in the local process. .. note: Rank 0 is assumed to be the coordinator rank. Args: state_dict (Dict[str, Any]): The state_dict to save. 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``) storage_writer (Optional[StorageWriter]): Instance of StorageWriter used to perform writes. If this is not specified, DCP will automatically infer the writer based on the checkpoint_id. If checkpoint_id is also None, an exception will be raised. (Default: ``None``) planner (Optional[SavePlanner]): Instance of SavePlanner. If this is not specified, the default planner will be used. (Default: ``None``) process_group (Optional[ProcessGroup]): ProcessGroup to be used for cross-rank synchronization. (Default: ``None``) no_dist (bool): If ``True``, this function will assume the intent is to load a checkpoint on a single rank/process. (Default: ``False``) use_collectives (bool): If ``False``, this function will assume the intent is to save a checkpoint without using cross-rank synchronization. (Default: ``True``) This configuration is experimental and should be used with caution. It will change the format of the saved checkpoint and may not be backward compatible. Returns: Metadata: Metadata object for the saved checkpoint. Example: >>> # xdoctest: +SKIP >>> my_model = MyModule() >>> state_dict = {"model": my_model} >>> fs_storage_writer = torch.distributed.checkpoint.FileSystemWriter( ... "/checkpoint/1" ... ) >>> torch.distributed.checkpoint.save( >>> state_dict=state_dict, >>> storage_writer=fs_storage_writer, >>> ) .. note:: save_state_dict uses collectives to coordinate writes across ranks. For NCCL-based process groups, internal tensor representations of objects must be moved to the GPU device before communication takes place. In this case, the device used is given by ``torch.cuda.current_device()`` and it is the user's responsibility to ensure that this is set so that each rank has an individual GPU, via ``torch.cuda.set_device()``. z!torch.distributed.checkpoint.savezptorch.distributed is disabled, unavailable or uninitialized, assuming the intent is to save in a single process. stacklevelF)reader)r2r3r4r6r7rGN)torch_C_log_api_usage_oncer: is_availableis_initializedwarningswarnrrrrrB_stateful_to_state_dict)r2rFr3r7r4r6rGs&$$$$$$r.saverXWs~ HH  !DEQQd//11QQ4;N;N;P7PG ~  >.PUV  .z:)'+  s 5C C) c0a]tRt^toRtV3RltRtVtR#)r a This class contains futures for staging and upload completion. It is returned by async_save(). staging_completion is a future that indicates when local copy of state_dict is complete. upload_completion is a future that indicates when a checkpoint completed saving. cN<V^8dQh/S[R,;R&S[R,;R&#)r1Nstaging_completionupload_completionr)r> __classdict__s"r.r?AsyncSaveResponse.__annotate__s't $d|#r-r$N)r%r&r'r(r)__annotate_func__r,__classdictcell__)r]s@r.r r sr-async_checkpointer_type async_stagerc*V^8dQhR\R\\P,R,R\R,R\ R,R\ PR,R\R\R,R \R \R \\,/ #) r1r2rFNr3r7r4rarbr6rGr8) r rIrJrKrrr:r;rrr=rr )r>s"r.r?r?snnn$t+n"D( n 4  n $$t+ n3n$nnn nr-c aaaaa\PPR4\P!4'd`\P !4'dET;'g \ 4p \P!R4V P9d \R4hRoSf7Ve\V\4'dVoM\\RRRR44oSo\S4oRoRVV3Rllo\RR 7R VV3R ll4p \ P"!4;_uu_4p SeV P%S4V !4p V\&P(8Xd \+4M \-4p V P/V VVVVVVR 7pSe&VP1V3R l4V P34RRR4\X \44'd\T p\54pV3RRllpVP74'gVP1V4MVP9R4\;VXR7#\RR 7V3Rl4pV!4X# +'giL;i)a, Asynchronous version of ``save``. This code first de-stages the state_dict on to the staging storage (defaults to CPU memory), and then calls the `save` in a separate thread. .. warning:: This feature is experimental and subject to change. If you provide an ``async_stager``, call ``close()`` after the last checkpoint is saved. Internally created default stagers are closed automatically. Args: state_dict (Dict[str, Any]): The state_dict to save. 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``) storage_writer (Optional[StorageWriter]): Instance of StorageWriter used to perform 'stage' and 'save'. If this is not specified, DCP will automatically infer the writer based on the checkpoint_id. If checkpoint_id is also None, an exception will be raised. (Default: ``None``) planner (Optional[SavePlanner]): Instance of SavePlanner. If this is not specified, the default planner will be used. (Default: ``None``) process_group (Optional[ProcessGroup]): ProcessGroup to be used for cross-rank synchronization. (Default: ``None``) async_checkpointer_type (AsyncCheckpointerType): whether to do checkpoint in separate thread or process (Default: ``AsyncCheckpointerType.THREAD``) async_stager (AsyncStager): provides staging implementation. If ``storage_writer`` implements ``AsyncStager`` and ``async_stager`` is not provided, ``storage_writer`` will be used for staging. User-provided stagers remain owned by the caller and should be closed after the last checkpoint is saved. no_dist (bool): If ``True``, this function will assume the intent is to save a checkpoint on a single rank/process. (Default: ``False``) use_collectives: If False, Save the checkpoint without rank coordination. (Default: ``True``) This configuration is experimental and should be used with caution. It will change the format of the saved checkpoint and may not be backward compatible. Returns: Future: A future holding the resultant Metadata object from `save`. Example: >>> # xdoctest: +SKIP >>> my_model = MyModule() >>> state_dict = {"model": my_model} >>> fs_storage_writer = torch.distributed.checkpoint.FileSystemWriter( ... "/checkpoint/1" ... ) >>> checkpoint_future = torch.distributed.checkpoint.async_save( >>> state_dict=state_dict, >>> storage_writer=fs_storage_writer, >>> ) >>> >>> # ... do some work ... >>> >>> checkpoint_future.result() z'torch.distributed.checkpoint.async_savecpuzfA CPU backend must be enabled for async save; try initializing process group with 'cpu:gloo,cuda:nccl'NFcV^8dQhRR/#)r1r8Nr$)r>s"r.r? async_save..__annotate__Gs''D'r-cL<SeS'gRoSP4R#R#R#)NT)close)owned_async_stagerowned_async_stager_closedsr.maybe_close_owned_async_stager2async_save..maybe_close_owned_async_stagerGs'  )2K(, %  $ $ &3L )r-TrDcPV^8dQhR\\,\,/#)r1r8rr )r>s"r.r?rgNs..f_5G.r-c&<SPS4#N)stage)rbr2sr.stage_state_dict$async_save..stage_state_dictMs!!*--r-)rFr3r7r4r6rGc<S!4#rqr$)_rls&r.async_save..hs 6T6Vr-cTV^8dQhR\\,R\R,/#)r1original_staging_futurereturn_staging_futureNro)r>s"r.r?rgps% 7 7%+O%< 7#)$< 7r-cVP4VPR4R# \dpTPT4Rp?R#Rp?ii;irq)result set_result Exception set_exception)rzr{es&& r.callbackasync_save..callbackpsB 7'..0%006 7%33A66 7s!% A AA )r[r\cP<SP'dSP4R#R#rq) should_synchronize_after_executesynchronize_staging)rbsr.maybe_synchronize_staging-async_save..maybe_synchronize_stagings <<<002=r-)rPrQrRr:rSrTrdevice _device_typesAssertionError isinstancerrrrWr contextlib ExitStackrrr+r r execute_saveadd_done_callbackpop_allrdoner~r )r2rFr3r7r4rarbr6rGpgrsstackstaging_future_or_state_dictupload_executor upload_futurestaging_futurer{rrrlrjrksf$$$$$d$$ @@@r. async_savers/\ HH  !JK t2244  2 202 << b&6&6 6 x .2  %*^[*Q*Q)L( L". (4J %'' t,..-.     5  ) NN9 :'7'9$'*?*G*GG 1 246  !0 < < (')'+!=!   )  + +,V W MMO3 6.775.4h3H 7""$$  , ,X 6 ! , ,T 2!4   4 0 3 1 3 "#u s (BI I c0V^8dQhR\R\/#)r1r2r8r )r>s"r.r?r?s   O r-c/pVP4F,wr#RpRV 2Vn\RR7!V4!V4W&K. V#)z]Creates a shallow copy of `state_dict` where `state_dict` is called for each Stateful object.cR\V\4'dVP4#T#rq)rrr2)elems&r._elem_to_state_dict4_stateful_to_state_dict.._elem_to_state_dicts (24(B(B4??$ L Lr-z_stateful_to_state_dict.TrD)itemsr%r)r2stateful_state_dictkeyrrs& r.rWrWs^%%'  M*B#'G$#5T#J $ $ ( r-cV^8dQhR\R\R\PR,R\R\ R\ R,R\ R \/#) r1r2r3r4Nr5r6r7rGr8r9)r>s"r.r?r?srhhh!h$$t+h h  h 4  hhhr-caaaaaaa\PPR4\W$'*V4oSf \ 4oSf \ R4hRo/p\ SRR4;peWR&SPVR&\R /VBVVVVV3Rl4p \R /VBVVV3Rl4p RoS'dSPRW4oMV !4p V !V .4p V ^,o\R /VBVVV3Rl4p \R /VBVV3R l4pS'dSPR W4pV#V !4pV!V.4pSP4V#) z,torch.distributed.checkpoint.save_state_dictNplanner is NonerFr4cr<Sf \R4hSP4pR\P!SP4P 9d6\ P!R^R7SP SSP4MSP SVSPR7R\P!SP4P 9d*SPSPSPSR7MSPSP4SP4pSPV4pV#)Nr storage_metazThe function definition for SavePlanner.set_up_planner has been updated to include the storage_meta argument. Please update your implementation to include this parameter.rM)r2ris_coordinatorkwargs)rankrG) rrinspect signatureset_up_planner parametersrUrVrset_up_storage_writerrcreate_local_planprepare_local_plan)r local_plandistWr7r2r3rGs r. local_step$_save_state_dict..local_steps ? !23 3%224 !2!273I3I!J!U!U U MM.    " ":u/C/C D  " "%)$33 #    !E!EFQQ R  0 0$$ZZ / 1   0 01E1E F..0 #66zB r-cp<Sf \R4hSPV4wpoSPV4pV#)Nr)rcreate_global_planprepare_global_plan)all_local_plansglobal_metadatar7r3s&r. global_step%_save_state_dict..global_stepsA ? !23 3+2+E+Eo+V((<<_Mr-planc<Sf \R4hSf \R4hSPS4pSPVS4pVP4VP 4#)Nrzcentral_plan is None)r finish_plan write_datawaitvalue)final_local_plan all_writes central_planr7r3s r.r$_save_state_dict..write_datasc ? !23 3   !78 8"..|<#../?I !!r-cL<Sf \R4hSPSVR7S#)Nzglobal_metadata is None)metadataresults)rfinish) all_resultsrr3s&r.finish_checkpoint+_save_state_dict..finish_checkpoints.  " !:; ; Lr-writer$) rPrQrRrrrgetattrgrouprreduce_scatter all_reducebarrier)r2r3r4r5r6r7rG ckpt_kwargsckpt_idrrr global_planrrr write_resultsrrrsff&&&ff @@@r.rBrBsd HH  !OP  5E FE$&.//OK>?DAAN'.O$',{{ O$&+&!!'!F&+&'%)L++FJL )| &1:,&? "1~ &+& "' "&+&' ##GZK O ,6< $m_5  Or-)rCrXrrr )NrFN)NrFNT)?rrrJrUconcurrent.futuresr dataclassesrenumrtypingrrtyping_extensionsr rPtorch.distributed distributedr:#torch.distributed._state_dict_utilsr 4torch.distributed.checkpoint._async_process_executorr 3torch.distributed.checkpoint._async_thread_executorr +torch.distributed.checkpoint._storage_utilsr,torch.distributed.checkpoint.default_plannerr#torch.distributed.checkpoint.loggerr%torch.distributed.checkpoint.metadatar$torch.distributed.checkpoint.plannerrr$torch.distributed.checkpoint.stagingrrr%torch.distributed.checkpoint.statefulr$torch.distributed.checkpoint.storagerr"torch.distributed.distributed_c10drutilsrrr,torch.distributed.checkpoint._async_executorr__all__r FutureWarningrCrXr r*rrWrBr$r-r.rs %!&( ?GKB:F ;KA88U D !     .4(r /3r ,0 r #' r /3 r r !r )r j  $ $  $4(n/3n,0 n #' n /3 n6K5Q5Qn(,nn!n)nb4( )  hhr-