+ Pja0t$^RIHt^RIt^RIt^RIHt^RIHtHtH t H t ^RI H t H t Ht^RIt^RIHt^RIHt^RIHtHt]'d ^R IHt^R IHt^R IHt.R%Ot]!R4t ] !R4t!]"!]PFR4'g^]!R4]PFPHR&]!R4]PFPHR&]!R4]PFPHR&^RI%H&t&H't'H(t(RRlt)RRlt*!RR ]'4t+!RR 4t,] R]R]-3,3,t.R]/R&]R&RR ll4t0]R&R!R"ll4t0R&R#R$llt0R# ]d RtRtELi;i)') annotationsN)Callable)overload TYPE_CHECKING TypeAliasUnion) ParamSpecSelfTypeVar)Tensor)_check_cuda_bindings)driverruntime) _POOL_HANDLE_CUDAGraphInputLivenessTracker) _dummy_type CUDAGraphgraph_R_P_CudaStreamBase _CUDAGraph_graph_pool_handle_cuda_isCurrentStreamCapturing)rrrcV^8dQhRR/#)returnbool)formats"j/Users/ahmed/devFolder/Ultron/claude-voice/gateway/.venv/lib/python3.14/site-packages/torch/cuda/graphs.py __annotate__r#9s,,T,c\4#)zReturn True if CUDA graph capture is underway on the current CUDA stream, False otherwise. If a CUDA context does not exist on the current device, returns False without initializing the context. )rr r$r"is_current_stream_capturingr&9s * ++r$cV^8dQhRR/#rrrr )r!s"r"r#r#Bs99<9r$cP\PP\44#)zReturn an opaque token representing the id of a graph memory pool. See :ref:`Graph memory management`. .. warning:: This API is in beta and may change in future releases. )torchcudarrr r$r"graph_pool_handler,Bs :: " "#5#7 88r$ca]tRt^Nt$RtR]R&RRV3RllltRRltR RV3R llltR V3R llt R V3R llt RV3Rllt RV3Rllt RV3Rllt RV3RlltRV3RlltRV3RlltRV3RlltRRltRtV;t#)!raWrapper around a CUDA graph. Arguments: keep_graph (bool, optional): If ``keep_graph=False``, the cudaGraphExec_t will be instantiated on GPU at the end of ``capture_end`` and the underlying cudaGraph_t will be destroyed. Users who want to query or otherwise modify the underlying cudaGraph_t before instantiation can set ``keep_graph=True`` and access it via ``raw_cuda_graph`` after ``capture_end``. Note that the cudaGraphExec_t will not be instantiated at the end of ``capture_end`` in this case. Instead, it will be instantiated via an explicit called to ``instantiate`` or automatically on the first call to ``replay`` if ``instantiate`` was not already called. Calling ``instantiate`` manually before ``replay`` is recommended to prevent increased latency on the first call to ``replay``. It is allowed to modify the raw cudaGraph_t after first calling ``instantiate``, but the user must call ``instantiate`` again manually to make sure the instantiated graph has these changes. Pytorch has no means of tracking these changes. .. warning:: This API is in beta and may change in future releases. z%_CUDAGraphInputLivenessTracker | None_trackerc V^8dQhRRRR/#)r keep_graphrrr r )r!s"r"r#CUDAGraph.__annotate__ks$r$c 4<\SV`W4pRVnV#N)super__new__r.)clsr0instance __class__s&& r"r5CUDAGraph.__new__ks7?33 r$cV^8dQhRR/#rrNoner )r!s"r"r#r1psr$c ~VPRuqnVeVP4R#R# \dR#i;ir3)r.stop Exception)selftrackers& r"__del__CUDAGraph.__del__ps= %)]]D "G]" #   s '- <<c(V^8dQhRRRRRRRR/#) rpool_POOL_HANDLE | Nonecapture_error_modestrcheck_input_livenessrrr<r )r!s"r"r#r1xs2&"&"!&" &"# &"  &"r$c <VPe"VPP4RVn\SV` WR7V'd/^RIHpV!4VnVPP 4R#R#)aeBegin capturing CUDA work on the current stream. Typically, you shouldn't call ``capture_begin`` yourself. Use :class:`~torch.cuda.graph` or :func:`~torch.cuda.make_graphed_callables`, which call ``capture_begin`` internally. Arguments: pool (optional): Token (returned by :func:`~torch.cuda.graph_pool_handle` or :meth:`other_Graph_instance.pool()`) that hints this graph may share memory with the indicated pool. See :ref:`Graph memory management`. capture_error_mode (str, optional): specifies the cudaStreamCaptureMode for the graph capture stream. Can be "global", "thread_local" or "relaxed". During cuda graph capture, some actions, such as cudaMalloc, may be unsafe. "global" will error on actions in other threads, "thread_local" will only error for actions in the current thread, and "relaxed" will not error on these actions. Do NOT change this setting unless you're familiar with `cudaStreamCaptureMode `_ check_input_liveness (bool, optional): If ``True``, tracks external tensor inputs during graph capture and raises an error if any are deallocated before replay. This helps debug "use after free" errors where input tensors are garbage collected between capture and replay. Default: ``False``. .. note:: Custom CUDA kernels added outside PyTorch (e.g., via cuLaunchKernel or DLPack) are not tracked by this mechanism. N)rErGr)r.r>r4 capture_begintorch.utils._cuda_debugrstart)r@rErGrIrr8s&&&& r"rKCUDAGraph.capture_beginxs[< == $ MM    DM 4O  N:r@r8s&r"rQCUDAGraph.capture_ends/  == $ MM    %r$cV^8dQhRR/#r;r )r!s"r"r#r1sTr$c $<\SV`4R#)zInstantiate the CUDA graph. Will be called by ``capture_end`` if ``keep_graph=False``, or by ``replay`` if ``keep_graph=True`` and ``instantiate`` has not already been explicitly called. Does not destroy the cudaGraph_t returned by ``raw_cuda_graph``. N)r4 instantiaterRs&r"rVCUDAGraph.instantiates r$cV^8dQhRR/#r;r )r!s"r"r#r1sr$c <VPe*VPPVP44\SV`4R#)z,Replay the CUDA work captured by this graph.N)r. check_aliverEr4replayrRs&r"r[CUDAGraph.replays/ == $ MM % %diik 2 r$cV^8dQhRR/#r;r )r!s"r"r#r1str$c <VPe"VPP4RVn\SV` 4R#)z1Delete the graph currently held by this instance.N)r.r>r4resetrRs&r"r_CUDAGraph.resets- == $ MM    DM  r$cV^8dQhRR/#r(r )r!s"r"r#r1slr$c <\SV`4#)zReturn an opaque token representing the id of this graph's memory pool. This id can optionally be passed to another graph's ``capture_begin``, which hints the other graph may share the same memory pool. )r4rErRs&r"rECUDAGraph.pools w|~r$cV^8dQhRR/#r;r )r!s"r"r#r1s++4+r$c <\SV`4#)z/Enable debugging mode for CUDAGraph.debug_dump.)r4enable_debug_moderRs&r"rfCUDAGraph.enable_debug_modesw(**r$c V^8dQhRRRR/#)r debug_pathrHrr<r )r!s"r"r#r1s..S.T.r$c "<\SV`V4#)z Arguments: debug_path (required): Path to dump the graph to. Calls a debugging function to dump the graph if the debugging is enabled via CUDAGraph.enable_debug_mode() )r4 debug_dump)r@rir8s&&r"rkCUDAGraph.debug_dumpsw!*--r$cV^8dQhRR/#rrintr )r!s"r"r#r1s(((r$c <\SV`4#)alReturns the underlying cudaGraph_t. ``keep_graph`` must be True. See the following for APIs for how to manipulate this object: `Graph Management `_ and `cuda-python Graph Management bindings `_ )r4raw_cuda_graphrRs&r"rqCUDAGraph.raw_cuda_graphs w%''r$cV^8dQhRR/#rnr )r!s"r"r#r1s--S-r$c <\SV`4#)aReturns the underlying cudaGraphExec_t. ``instantiate`` must have been called if ``keep_graph`` is True, or ``capture_end`` must have been called if ``keep_graph`` is False. If you call ``instantiate()`` after ``raw_cuda_graph_exec()``, the previously returned cudaGraphExec_t will be destroyed. It is your responsibility not to use this object after destruction. See the following for APIs for how to manipulate this object: `Graph Execution `_ and `cuda-python Graph Execution bindings `_ )r4raw_cuda_graph_execrRs&r"ruCUDAGraph.raw_cuda_graph_execs w*,,r$cV^8dQhRR/#)rrdictr )r!s"r"r#r1sC C C r$c ^RIHp\e \f \ R4hV!4'd \ R4h\P P R\P PR\P PR\P PR\P PR \P PR \P PR \P PR \P PR \P PR/ pVP!4p\#\P$!V^R74wrE\#\P$!W5R74wre/p.p\'V4EFp Wi,p W\)V 4&\#\P*!V 44p \#\P,!V 44p V ^ , p V R,pRpV \P P 8Xd\P.!\)V 4R7p\P0!V4wppV\P2P48Xd\)VP64'd\P8!\)VP64R7p\P:!V4wppV\P2P48Xd)\=V\>4'dVPA4MTpVPCRV RVPEV \GV 44RV RV RVRVR.R./4EK \#\PH!V^R74wppV^8d\#\PH!VVR74wpppp\'V4Fp VPE\)VV ,44pVPE\)VV ,44pVfKKVfKQVV,R,PCV4VV,R,PCV4K \PJ!VPM4R7p\#\PN!V44pVF"pV^ ,VR,,VR&VVR&K$ RVRV/#)aReturn a dictionary describing the graph's topology and node metadata. ``keep_graph`` must be True. The graph must have been instantiated (via :meth:`instantiate`) before calling this method. Requires the ``cuda.bindings`` package. Returns a dictionary with structure:: { "exec_graph_id": int, "nodes": [ { "index": int, "node_type": str, "tools_id": int, "graph_id": int, "node_id": int, "kernel_name": str or None, "dependencies": [int, ...], "dependents": [int, ...], }, ..., ], } Each node's ``graph_id`` is remapped to the exec graph id so that ``tools_id`` values match those reported by CUPTI-based profilers. ``dependencies`` and ``dependents`` are lists of node indices within the ``nodes`` list. This structure is useful for inspecting a profiler trace and establishing whether a particular dependency observed in the profile is a true dependency (encoded in the graph) or a fake dependency caused by mapping of independent streams to the same hardware channel. )_is_tools_id_unavailableNz1get_graph_data requires the cuda.bindings packagezget_graph_data requires cudaGraphNodeGetToolsId which needs cuda.bindings >= 13.1 and CUDA driver >= 13.1 (or cuda-compat >= 13.1 in LD_LIBRARY_PATH)kernelmemcpymemsethost child_graphempty wait_event event_record mem_allocmem_free)numNodesl) init_valueindex node_typetools_idgraph_idnode_id kernel_name dependencies dependents)numEdges exec_graph_idnodes)(torch.cuda._graph_annotationsrz _cuda_runtime _cuda_driver RuntimeErrorcudaGraphNodeTypecudaGraphNodeTypeKernelcudaGraphNodeTypeMemcpycudaGraphNodeTypeMemsetcudaGraphNodeTypeHostcudaGraphNodeTypeGraphcudaGraphNodeTypeEmptycudaGraphNodeTypeWaitEventcudaGraphNodeTypeEventRecordcudaGraphNodeTypeMemAlloccudaGraphNodeTypeMemFreerqr cudaGraphGetNodesrangerocudaGraphNodeGetTypecudaGraphNodeGetToolsId CUgraphNodecuGraphKernelNodeGetParamsCUresult CUDA_SUCCESSfunc CUfunction cuFuncGetName isinstancebytesdecodeappendgetrHcudaGraphGetEdgescudaGraphExec_trucudaGraphExecGetId)r@rznode_type_namesraw_numr handle_to_idx node_infosinodentyperrrrcu_nodeerrparamscu_funcname num_edges from_nodesto_nodes _edge_datasrcdst exec_handlerinfos& r"get_graph_dataCUDAGraph.get_graph_datasJ K  L$8RS S # % %>   + + C CX  + + C CX  + + C CX  + + A A6  + + B BM  + + B BG  + + F F  + + H H.  + + E E{  + + D Dj  !!#%m&E&EcTU&VW)  + +C > )+ !# sA8D'(#d) $()K)KD)QRE+M,Q,QRV,WXH2~H+GK 77OOO&22c$iH*EEgN V,//<<<V[[AQAQ*55V[[AQRG , : :7 CICl33@@@7A$7N7NdkkmTX   Q!4!4UCJ!Gw!;"B "  '@2  + +C! < 1a q=:N//iH; 7J*i9%#''JqM(:;#''HQK(89?ssOL188=sON3::3? &$33//1 -  , ,[ 9 D - 3tIFD ,D  ] Z  r$)r.)F)NglobalF)__name__ __module__ __qualname____firstlineno____doc____annotations__r5rBrKrQrVr[r_rErfrkrqrur__static_attributes__ __classcell__)r8s@r"rrNs443 &"&"P ! ! ++..((--C C r$cR]tRtRt$RtRtR]R&R RRlltRR ltR R lt R t R#)rima Context-manager that captures CUDA work into a :class:`torch.cuda.CUDAGraph` object for later replay. See :ref:`CUDA Graphs ` for a general introduction, detailed use, and constraints. Arguments: cuda_graph (torch.cuda.CUDAGraph): Graph object used for capture. pool (optional): Opaque token (returned by a call to :func:`~torch.cuda.graph_pool_handle()` or :meth:`other_Graph_instance.pool()`) hinting this graph's capture may share memory from the specified pool. See :ref:`Graph memory management`. stream (torch.cuda.Stream, optional): If supplied, will be set as the current stream in the context. If not supplied, ``graph`` sets its own internal side stream as the current stream in the context. capture_error_mode (str, optional): specifies the cudaStreamCaptureMode for the graph capture stream. Can be "global", "thread_local" or "relaxed". During cuda graph capture, some actions, such as cudaMalloc, may be unsafe. "global" will error on actions in other threads, "thread_local" will only error for actions in the current thread, and "relaxed" will not error on actions. Do NOT change this setting unless you're familiar with `cudaStreamCaptureMode `_ enable_annotations (bool, optional): If ``True``, enables kernel annotation recording on entry and automatically calls :func:`~torch.cuda._graph_annotations.resolve_pending_annotations` before the capture ends. Annotations are **not** cleared on exit so that multiple graphs in the same workload can accumulate annotations. Requires ``cuda.bindings`` package and cuda-compat >= 13.1 or CUDA driver >= 13.1. check_input_liveness (bool, optional): If ``True``, tracks external tensor inputs during graph capture and raises an error if any are deallocated before replay. This helps debug "use after free" errors where input tensors are garbage collected between capture and replay. Default: ``False``. .. note:: Custom CUDA kernels added outside PyTorch (e.g., via cuLaunchKernel or DLPack) are not tracked by this mechanism. .. note:: For effective memory sharing, if you pass a ``pool`` used by a previous capture and the previous capture used an explicit ``stream`` argument, you should pass the same ``stream`` argument to this capture. .. warning:: This API is in beta and may change in future releases. .. _cudaStreamCaptureMode: https://docs.nvidia.com/cuda/cuda-runtime-api/group__CUDART__STREAM.html#group__CUDART__STREAM_1g9d0535d93a214cbf126835257b16ba85 Ntorch.cuda.Stream | Nonedefault_capture_streamc 0V^8dQhRRRRRRRRR R R R /#) r cuda_graphrrErFstreamrrGrHenable_annotationsrrIr )r!s"r"r#graph.__annotate__sF999"9) 9  9 ! 9#9r$c VfFVPPf.\PP 4VPnVfRMV3VnVeTMVPPVnVP f \R4h\PPVP 4Vn Wn W@n WPn W`n R#)Nzcapture_stream must not be Noner )r8rr*r+StreamrEcapture_streamAssertionErrorr stream_ctxrrG_enable_annotationsrI)r@rrErrGrrIs&&&&&&&r"__init__graph.__init__s >dnnCCK49JJ4E4E4GDNN 1;?> 3 3 3 JJL   !!#  # # # W M !!# %% YY  $66  "&!:!:  r$c V^8dQhRRRR/#)rargsobjectrr<r )r!s"r"r#rs 1 1f 1 1r$c VP'd^RIHpV!4VPP 4VP P !V!VP'd^RIHpV!VP4R#R#)r)resolve_pending_annotations)remap_to_exec_graphN)rrrrrQr__exit__r)r@rrrs&* r"rgraph.__exit__sY  # # # Q ' ) ##%   $'  # # # I  0 $r$)rrGrrIrrEr)NNrFF) rrrrrrrrrrrr r$r"rrms+(T8<4;96  D 1 1r$torch.nn.Module.r_ModuleOrCallablec 0V^8dQhRRRRRRRRR R R R/#) r callablesr sample_argstuple[Tensor, ...]num_warmup_itersroallow_unused_inputrrErFrr )r!s"r"r#r#sD #     r$cR#r3r rrrrrEs&&&&&r"make_graphed_callablesrsr$c 0V^8dQhRRRRRRRRR R R R/#) rrztuple[_ModuleOrCallable, ...]rztuple[tuple[Tensor, ...], ...]rrorrrErFrr )r!s"r"r#r#sD((,(/(( (  ( # (r$cR#r3r rs&&&&&r"rrs%(r$c 0V^8dQhRRRRRRRRR R R R/#) rrz1_ModuleOrCallable | tuple[_ModuleOrCallable, ...]rz3tuple[Tensor, ...] | tuple[tuple[Tensor, ...], ...]rrorrrErFrr )r!s"r"r#r#sLJJ@JDJJ J  J 7 Jr$c \P!4'd'\P!4'd \R4hRp\ V\ 4'g0RpV3p\ P!\ \R3,V43pM5\ P!\ \ \R3,R3,V4p.p\W4EFqwr\ V\PP4'd\VP4^8Xd5\VP4^8Xd\VP4^8Xg \!R4h\";QJd*RVP%44F 'dK RM RM!RVP%444'g \!R4h\P&P(P*!V !p VP-\ V 44\";QJdRV 4F 'dK RM RM !RV 44'dEKi\!R4h VU u.uFp \V 4NK p p VUu.uFHp\ V\PP4'd\ VP/44MRNKJ p p\1\V44U u.uFp W},W,,NK pp \1\V44Uu.uF!p\P2P54NK# pp\1\V44Uu.uF!p\P2P54NK# ppVf \74MTp\P2P94\P2P;\P2P=44;_uu_4\WV4EFwpp pRRRppp\1V4Fp\P&P(P?V!V !4p\ ;QJd.RV4FNK 5M !RV44p\V4^8gKn\P@PCT\ ;QJd.R V4FNK 5M !R V44\ ;QJd.R V4FNK 5M !R V44RVR 7pK VVV3Fp?K EK RRR4\P2P94.p.p\WV4Fwpp p\P2PEVVR 7;_uu_4V!V !pRRR4\P&P(PGX4wppVP-\ V44VP-V4K .p.p \\IV4\IV4\IV44EFwpp!p"\ ;QJd.R V!4FNK 5M !R V!44p#\ ;QJd.RV!4FNK 5M !RV!44pRp\V4^8d\P2PEV"VR 7;_uu_4\P@PCT\ ;QJd.RV4FNK 5M !RV44\ ;QJd.RV#4FNK 5M !RV#44RVR 7pRRR4.p$^p%VFMp&V&PJ'd(Ve$V$P-VV%,4V%^, p%K<V$P-R4KO \ V$4p$VP-V#4V P-V$4EK VPM4V PM4RRlp'.p(\OV4Fwp pV'!VV ,VV ,W,W,VV ,W,VV ,VV ,V V ,4 p)\ V\PP4'd>RRlp*V*!VVPPV)VPR4Vn)V(P-V4KV(P-V)4K V'd V(^,#\ V(4#uup iuupiuup iuupiuupi +'giEL;i +'giEL;i +'giEL;i)a Accept callables (functions or :class:`nn.Module`\ s) and returns graphed versions. Each graphed callable's forward pass runs its source callable's forward CUDA work as a CUDA graph inside a single autograd node. The graphed callable's forward pass also appends a backward node to the autograd graph. During backward, this node runs the callable's backward work as a CUDA graph. Therefore, each graphed callable should be a drop-in replacement for its source callable in an autograd-enabled training loop. See :ref:`Partial-network capture` for detailed use and constraints. If you pass a tuple of several callables, their captures will use the same memory pool. See :ref:`Graph memory management` for when this is appropriate. Arguments: callables (torch.nn.Module or Python function, or tuple of these): Callable or callables to graph. See :ref:`Graph memory management` for when passing a tuple of callables is appropriate. If you pass a tuple of callables, their order in the tuple must be the same order they'll run in the live workload. sample_args (tuple of Tensors, or tuple of tuples of Tensors): Samples args for each callable. If a single callable was passed, ``sample_args`` must be a single tuple of argument Tensors. If a tuple of callables was passed, ``sample_args`` must be tuple of tuples of argument Tensors. num_warmup_iters (int): The number of warmup iterations. Currently, ``DataDistributedParallel`` needs 11 iterations for warm up. Default: ``3``. allow_unused_input (bool): If False, specifying inputs that were not used when computing outputs (and therefore their grad is always zero) is an error. Defaults to False. pool (optional): Token (returned by :func:`~torch.cuda.graph_pool_handle` or :meth:`other_Graph_instance.pool()`) that hints this graph may share memory with the indicated pool. See :ref:`Graph memory management`. .. note:: The ``requires_grad`` state of each Tensor in ``sample_args`` must match the state that's expected for the corresponding real input in the training loop. .. warning:: This API is in beta and may change in future releases. .. warning:: ``sample_args`` for each callable must contain only Tensors. Other types are not allowed. .. warning:: Returned callables do not support higher order differentiation (e.g., double backward). .. warning:: In any :class:`~torch.nn.Module` passed to :func:`~make_graphed_callables`, only parameters may be trainable. Buffers must have ``requires_grad=False``. .. warning:: After you pass a :class:`torch.nn.Module` through :func:`~make_graphed_callables`, you may not add or remove any of that Module's parameters or buffers. .. warning:: :class:`torch.nn.Module`\s passed to :func:`~torch.cuda.make_graphed_callables` must not have module hooks registered on them at the time they are passed. However, registering hooks on modules *after* passing them through :func:`~torch.cuda.make_graphed_callables` is allowed. .. warning:: When running a graphed callable, you must pass its arguments in the same order and format they appeared in that callable's ``sample_args``. .. warning:: The automatic mixed precision is supported in :func:`~torch.cuda.make_graphed_callables` only with disabled caching. The context manager `torch.cuda.amp.autocast()` must have `cache_enabled=False`. z_make_graphed_callables does not support the autocast caching. Please set `cache_enabled=False`.FT.c3<"TFqPRJxK R#5i)FN requires_grad.0bs& r" )make_graphed_callables..dsEA%/sc3V"TFp\V\P4xK! R#5ir3)rr*r )rargs& r"rrlsHKS:c5<<00Ks')Nc3L"TFqP'gKVxK R#5ir3r ros& r"rrs$K1??QQ$ $c3L"TFqP'gKVxK R#5ir3r rrs& r"rrs%';!AA';rc3t"TF.qP'gK\P!V4xK0 R#5ir3r r* empty_likers& r"rrs(+9@AOO/E,,Q//s88)outputsinputs grad_outputs only_inputs allow_unused)rEc3t"TF.qP'd\P!V4MRxK0 R#5ir3rrs& r"rrs)$ FT???E  Q  .__annotate__sl====6= = < = 1 =+=7=/= =r$c aaaaaaaaaa !VVVVVVV3RlR\PP4o RV VV3Rllp V #)c<]tRtRt]RVVVV3Rll4t]]PPPRVVV3Rll44t Rt R#)Omake_graphed_callables..make_graphed_autograd_function..Graphedic$V^8dQhRRRRRR/#)rctxrrr rrr )r!s"r"r#\make_graphed_callables..make_graphed_autograd_function..Graphed.__annotate__s' A AV Af A9K Ar$c <\S4FSpSV,P4W,P48wgK5SV,PW,4KU SP4\ S\ 4'g\ R\S4 24h\ ;QJd.RS4FNK 5#!RS44#)z"static_outputs must be tuple, got c3@"TFqP4xK R#5ir3detachrs& r"rjmake_graphed_callables..make_graphed_autograd_function..Graphed.forward..s@AXXZZs)rdata_ptrcopy_r[rtuplertype)r5rrr'r*r,r-s&* r"forwardWmake_graphed_callables..make_graphed_autograd_function..Graphed.forwards}-A+A.779VY=O=O=QQ,Q/55fi@.  "!.%88(.make_graphed_autograd_function..Graphed.backward.. s(0#$-AHHJQ6/s#%) lenrzipr<r=r[rr>r?)r5rCggradr(r/r.s&* r"backwardXmake_graphed_callables..make_graphed_autograd_function..Graphed.backwardsu:%8!99(%c%j\1NsSfOgNhi ##6>GA}::<4==?:GGDM ?   """4e<<(@FXAY@Z[u0uu0r$r N) rrrr staticmethodr@r*autogradfunctiononce_differentiablerKr)r(r'r*r/r.r,r-sr"Graphedr3sN  A A A ^^ $ $ 8 8  9 r$rQc V^8dQhRRRR/#)r user_argsrrr )r!s"r"r#Tmake_graphed_callables..make_graphed_autograd_function..__annotate__s R Rv R& Rr$c<\PPP!V!pSP!\ V4S,!p\PPP VS4#r3)r*utils_pytreearg_tree_leavesapplyr>tree_unflatten)rSflatten_user_argsoutrQr)r+s* r"functionalizedVmake_graphed_callables..make_graphed_autograd_function..functionalizedsW!& 3 3 C CY O --%(9":]"JLC;;&&55c;PQ Qr$)r*rNFunction) r'r(r)r*r+r,r-r.r/r]rQs fffffffff @r"make_graphed_autograd_function>make_graphed_callables..make_graphed_autograd_functions4( ( enn--( T R Rr$c ,V^8dQhRRRRRRRRRR/#) rrrgraph_training_statergraphedzCallable[_P, _R]orig_fwdrr )r!s"r"r#r0.s:  % &* * +   "  r$c&aaaaRVVVV3RllpV#)c$V^8dQhRRRRRR/#)rrSz_P.args user_kwargsz _P.kwargsrrr )r!s"r"r#Jmake_graphed_callables..make_graphed_forward..__annotate__4s&CCC CbCr$cF<SPS8Xd S!V/VB#S!V/VB#r3)training)rSrhrrcrdres*,r"new_fwdEmake_graphed_callables..make_graphed_forward..new_fwd4s2}}(<<& A[AA'BkBBr$r )rrcrdrerlsffff r"make_graphed_forward4make_graphed_callables..make_graphed_forward.s CCr$zModules must not have hooks registered at the time they are passed. However, registering hooks on modules after passing them through make_graphed_callables is allowed.zIn any :class:`~torch.nn.Module` passed to :func:`~make_graphed_callables`, only parameters may be trainable. All buffers must have ``requires_grad=False``.zfIn the beta API, sample_args for each callable must contain only Tensors. Other types are not allowed.r )*r*is_autocast_enabledis_autocast_cache_enabledrrr>typingcastr rHnnModulerG_backward_hooks_forward_hooks_forward_pre_hooksrallbuffersrVrWrXr parametersrr+rr,rrr tree_leavesrNrJr tree_flattenreversedr reverse enumeraterkr@)+rrrrrEjust_one_callable _sample_argsflatten_sample_argscr flatten_argper_callable_len_user_argsper_callable_module_paramsr"per_callable_static_input_surfacesr fwd_graphs bwd_graphsmempoolrr, grad_inputsr outputs_gradvper_callable_static_outputs"per_callable_output_unflatten_specr' func_outputsflatten_outputsspec per_callable_static_grad_outputsper_callable_static_grad_inputsr-r(r.r/grad_idxrr`retrdrns+&&&&& r"rrsrT   ""u'F'F'H'H m   i ' ' L  E&#+$6 DF {{5vs{);S)@#A;O y/ a ) )A%%&!+(()Q.,,-2$a3EE333EEEE$1 kk))994@ ""5#56sHKHsssHKHHH ^ )069L!L8K#d)8K!L"A",Auxx!?!?allnRG" s9~&*&A !;!>>>&'* 38I2GH2GQ%**&&(2GJH27I2GH2GQ%**&&(2GJH%)\!tG  JJ   5::,,. / /03 %G1 ,D$,26tT,K+,++--99$+F$u$K$Kuu$K$KK |$q("'.."5"5 ,$u%';%uu%';% &+U+9@+UU+9@+&%)%7#6 #K -|[9:'1  0. JJ#%)+&!$Yj!IdI ZZ  ig  6 6;L7!& 3 3 @ @ N#**5+AB*11$7 "J(*$&(#;>34,-<7ni$e$ FT$ ee$ FT$  uJJuuJJJ  | q !!)'!::#nn11( 5 T,@ T55 T,@ TT!&&W2E&W&W2E&W!W $!3 2 ; 'C   [%<"))+h*?@A "))$/ ( ##56(//0CD'../ABK<P%,,.#++-=@$&CY'40 qM qM & ) & ) .q 1 . 1 ' * ,Q / +A .   dEHHOO , ,  0dmmWdllDL JJt  JJw I(L1v :i"M"* IH 0 / /B7 6 60;::s]b4Ab9b>;'c:'cBc ($c  *c 8>c c! Ac5(c5 c ! c2 5 d )r&r,rrr)FN)1__conditional_annotations__ __future__rrrrcollections.abcrrrrrtyping_extensionsr r r r*r torch.cuda._utilsr cuda.bindingsrrrr ImportError torch.cudarrLr_utilsr__all__rrhasattrr__dict__torch._Crrrr&r,rrrrrr)rs@r"rsb"" $<<66 2'F   T]t_uxx*++&1,&?EHHl#.9:N.OEHH*+:E(;EHH67TS,9\ \ ~v1v1t %%6f8M%MN9N   ( (JJULMsE E#"E#