ёi͋SSKrSSKrSSKrSSKrSSKrSSKrSSKrSSKrSSK r SSK J r SSK J r J r JrJr SSKJrJr SSKJrJr SSKJrJr SSKJr SSKJrJrJr S S KJ r J!r!J"r"J#r#J$r$ \"\%\RLS S 9r'S r(Sr)Sr*Sr+Sr,Sr-SSjr.Sr/\SSj5r0SSjr1\SSj5r2\SSj5r3\S5r4\S5r5g)N)pir)ValueSetget_real_op_inputsget_real_op_outputs some_in_set)coredefault_main_program)Executor global_scope)dygraph_not_support static_only) get_logger)_pack_loaded_dict_pickle_loads_mac_unpack_saved_dict) _check_args _check_vars_get_valid_program_normalize_path_prefix_safe_load_picklez&%(asctime)s-%(levelname)s: %(message)s)fmtcZ/n/nUR5HnUR(d"UR5R5S:XaUR U5 MIUR (dM\UR5R5S:XdMUR U5 M X4$)z Get parameters and optimizer variables from program. Args: program(Program): The program to get parameters and optimizer variables. zbuiltin.parameter pd_op.data) list_vars is_parameterget_defining_opnameappend persistable)programparamsoptsvars T/var/www/html/banglarbhumi/venv/lib/python3.13/site-packages/paddle/static/pir_io.pyget_pir_parametersr&>s F D  "   ""$))+/BB MM#  ___!4!4!6!;!;!=!M KK # <cP/n/nUR5RHnUR5S:XdUR5S:Xa"URUR 5S5 UR5S:XdMcUR UR 55 M X4$)Nr pd_op.feedr pd_op.fetch) global_blockopsrrattrsextendoperands_source)r!feed_name_list fetch_targetsops r%get_pir_feed_and_fetchr3QsNM""$(( 779 $ \(A  ! !"((*V"4 5 779 %  !3!3!5 6 )  ((r'c[5RU5R5nUR5nUR 5(a [ R R5nGOUR5(a [ R R5nGOUR5(av[ R RR5nURUR55 [ R RUR55nGO-UR!5(a[ R RR5nURUR55 [ R R#[ R$R'5R)S5SUR+55nOt[ R RR5nURUR55 [ R R-UR/55nUR1X5 g)N:r)r find_var get_tensor_place is_cpu_placepaddlebaseCPUPlaceis_cuda_pinned_placeCUDAPinnedPlace is_xpu_placerPlace set_placeXPUPlace xpu_device_idis_custom_place CustomPlacedevice get_devicesplitcustom_device_id CUDAPlace gpu_device_idset)rndarraytpplaces r%set_varrQ\s%002A  A~~ $$&   ! ! ++-    KK   " " $ AHHJ $$Q__%67     KK   " " $ AHHJ '' MM $ $ & , ,S 1! 4a6H6H6J  KK   " " $ AHHJ %%aoo&78EE'r'c[U5Hup#UR5nUR5S:wdM+UR5S:wdMA[RR S[ U5-URUR[RRR55nURU5 UR5RU5 M [U5H`up#UR5nUR5S:wdM+UR5S:wdMAUR5RU5 Mb g)z Append feed ops to the program. Args: program(Program): Specify a program you want to append fetch op. feed_vars(Value | list[Value]): Values should be feed. Returns: modify program r)r feed_name_N) enumeraterrr:_pir_opsdatastrshapedtyper;rr@replace_all_uses_with move_beforeget_parent_block remove_op)r! feed_varsir$orig_opvalues r%append_pir_feed_opsrbusI&%%' <<>\ )glln .LOO((s1v%     &&( E  % %e ,  ! ! # / / 8'I&%%' <<>\ )glln .L  $ $ & 0 0 9'r'c[U5H/unup4[RRX4U5nSUlM1 g)z Append fetch ops to the program. Args: program(Program): Specify a program you want to append fetch op. fetch_vars(Tensor | list[Tensor]): Values returned by inference. Returns: modify program TN)rTr:rUfetchr )r!fetch_name_var_mapsr_r$routs r%append_pir_fetch_opsrgs7$$78;Coo##Cq19r'c[U[RR5(d[ S[ U5S35eUR 5RnS/[U5-n[U5n[[[U555HHupg[[U5U5(a%[U5HnUR!U5 M MDSXF'MJ [[[U555HoupgXF(aM[[U5[U55(a[#SUR%5S35eUR 5R'U5 Mq g)a Prune a program according to feed_vars and target_vars. Args: program(Program): Specify a program you want to prune. feed_vars(Tensor | list[Tensor]): Values needed by inference. target_vars(Tensor | list[Tensor]): Values returned by inference. Returns modify program  gv}}44 5 5J4PW=/YZ [  $$&**I!FS^3K(L$y345 *2. = =-b1  )2(- ! $ 6$y345$''.r2HY4GHH / {;89  " , ,R 06r'cU(aURHTnURHAnURS5(aURSS5 UR 5S:XdMAMC MV g)a6 This method will create a new program and do following adjustments on it: 1. Remove all reader variables and their creator ops if exist. 2. Remove the :code:`read_op` if exists. 3. change the :code:`is_test` attribute of operators to :code:`True`. All the :code:`Parameter` information will be lost. Args: prune_read_op(bool): remove the read ops that are added by py_reader for cpp inference library Notes: This API is a very low level API. Use :code:`Program.clone(for_test=True)` instead. Returns: Program: The new program. is_testTzpd_op.batch_normN)blocksr,has_attr set_bool_attrr)r! prune_read_opblockr2s r%_inference_optimizersY. ))B{{9%%  D1wwy..  r'c [U[RR5(d[ S[ U5S35e[U[ 5(dU/n[SU55(d [ S5e[U[ 5(dU/n[SU55(d [ S5e[UR5R5S:Xa [S5eUR5RH/nUR5S :XdM[R"S 5 O [RR!5nUR#U5nUR5nUVs/sHoR%U5PM n nUVs/sHoR%U5PM n nURH*nUR5S :XdMUR'U5 M, UR)S S 5n U (d [+XiU 5 [-USS9 /n [/U 5HupUR15nUnUR5S:Xa]UR3S5R15nUR5S5(a(UR75SS:XaUR3S5nUR8(aU R;UUR45 MU R;US[=U 5-45 M [RR?U5 [AXi5 [CXl5 SSS5 U$s snfs snf!,(df  U$=f)a' Normalize/Optimize a program according to feed_vars and fetch_vars. Args: program(Program): Specify a program you want to optimize. feed_vars(Tensor | list[Tensor]): Values needed by inference. fetch_vars(Tensor | list[Tensor]): Values returned by inference. kwargs: Supported keys including ``skip_prune_program``. - skip_prune_program(bool): whether to skip pruning program. Defaults to False. Returns: Program: Normalized/Optimized program. Examples: .. code-block:: python >>> import paddle >>> paddle.enable_static() >>> path_prefix = "./infer_model" # User defined network, here a softmax regression example >>> image = paddle.static.data(name='img', shape=[None, 28, 28], dtype='float32') >>> label = paddle.static.data(name='label', shape=[None, 1], dtype='int64') >>> predict = paddle.static.nn.fc(image, 10, activation='softmax') >>> loss = paddle.nn.functional.cross_entropy(predict, label) >>> exe = paddle.static.Executor(paddle.CPUPlace()) >>> exe.run(paddle.static.default_startup_program()) # normalize main program. >>> program = paddle.static.default_main_program() >>> normalized_program = paddle.static.normalize_program(program, [image], [predict]) rirjc3V# UHn[U[R5v M! g7fNrkrValue.0vs r% (normalize_pir_program..s;Az!SYY''')z2feed_vars type must be a Value or a list of Value.c3V# UHn[U[R5v M! g7frrrs r%rrs<Az!SYY''rz3fetch_vars type must be a Value or a list of Value.rz=program must not be empty. at least one operator is required!z pd_op.auczHBe sure that you have set auc states to 0 before saving inference model.r*skip_prune_programFT)rz pd_op.scalerrag? fetch_name_N)"rkr:rlrmrnrorrallrpr+r,rtrwarningswarnr IrMappingclonelook_upr]getrzrrTroperand_sourcer~r-has_namerrW program_guardrbrg)r!r^ fetch_varskwargsr2 value_map copy_programr+rclone_feed_varsclone_fetch_varsrfetch_vars_tupler_r$scale_oporig_varfull_ops r%normalize_pir_programrsN gv}}44 5 5J4PW=/YZ [   i & &K ;; ; ;LMM j$ ' ' \ << < <MNN 7   ! % %&!+ K  ""$(( 779 # MMZ   ) $$&I==+L,,.L5>?Y((+YO?6@Aj))!,jA 779 %  " "2 &  $8%@ \  # #S-#a&*@$A B.  $ $\ 2L:\< 3 =@A2 3 2 s8MM $M McxSnUcUcSn[U5nUc@[U5upVXV-n[UUUVs/sHoR(dMUPM snUS9$Sn [ [ U55S:Xa[ R"S5 g0n UHn [5RU R5nUcUSLa{[RR[RRU5U R5n [R "UR#5U RU SS5 MUR#5XR'M UcU(a/n /n[%U R'55H'nU R)X5 UR)U5 M) SnUSLa=[RR[RRU5U5n[R*"U UUSSU5 U(a'[5RU 5R-5$gs snf) a Save specific variables in the `Program` to files. There are two ways to specify the variables to be saved: set variables in a list and assign it to the `vars`, or use the `predicate` function to select variables that make `predicate(variable) == True`. The first way has a higher priority. The `dirname` is used to specify the folder where to save variables. If you prefer to save variables in separate files in the `dirname` folder, do not set `filename`. If you prefer to save all variables in a single file, use `filename` to specify it. Args: dirname(str, optional): The folder to save variables. When you need to save the parameter to the memory, set it to None. main_program(Program, optional): The program whose variables will be saved. If it is None, the default main program will be used automatically. Default: None vars(list[Variable], optional): The list contains all variables to be saved. Default: None filename(str, optional): If you prefer to save all variables in a single file, use `filename` to specify it. Otherwise, let `filename` be None. Default: None Returns: str: When saving parameters to a file, returns None. When saving parameters to memory, returns a binary string containing parameters. FNT) main_programdirnamevarsfilename saved_paramsrzVno variable in your model, please ensure there are any variables in your model to save)rr& save_vars_pirr rprrrrr r6rospathjoinnormpathr save_funcr7sortedkeysrsave_combine_func get_bytes)rrrrsave_to_memoryparamopt vars_listr$params_var_name save_var_maprsave_file_path save_var_listsave_var_namesr save_paths r%rrMsJN8+%l3L |' 5 K %!*>#oo#>   ) tDz?a  MMh  A.))!&&1CNe$;!#bgg.>.>w.G!PNN$affndE(+~~'7 VV$  >MN|0023$$\%78%%d+4I&GGLL)9)9')BHM  " "   >**?;EEG G W?s H7 H7 c Ub[U[5(deSnUb[RR U5nUS:XaSnUcUc [ 5n[U[ RR5(d[S[U535e[U5upgXg-n[ RRRRU[!5UR"5 [%UUUUVs/sHoR&(dMUPM snUS9 gUc [ 5n[U[ RR5(d[S[U535e0n UHn [!5R)U R*5n[U[ RRR,5(deUczUc [/S5e[RR1XR*5n [2R4"U S/SUR75UR"R955 MXU R*'M Ub/n /n [;U R=55H5nU R?XR755 U R?U5 M7 USLa[RR1X5n[2R@"UU U SUR"R955 [CX5H%up[EU[FRH"U55 M' ggs snf)a7 :api_attr: PIR Static Graph This API loads variables from files by C++ function. There are two ways to specify the variables to be loaded: the first way, set variables in a list and assign it to the `vars`; the second way, use the `predicate` function to select variables that make `predicate(variable) == True`. The first way has a higher priority. The `dirname` is used to specify the folder where to load variables. If variables were saved in separate files in the folder `dirname`, set `filename` None. If all variables were saved in a single file, use `filename` to specify it. Args: executor(Executor): The executor to create variables in scope. dirname(str): The folder where to load the variables. main_program(Program, optional): The program whose variables will be loaded. If it is None, the default main program will be used automatically. Default: None vars(list[Variable], optional): The list that contains all variables to be loaded. Default: None filename(str, optional): The file which saved all required variables. If variables were saved in separate files, set it to be None. Default: None Returns: None NFrz`The type of input main_program is invalid, expected type is paddle.static.Program, but received )rrrrz>The directory path and params cannot be None at the same time.)%rkr rrrr r:rlrmrnror&r; libpaddlercreate_loaded_parameterr _default_executor load_vars_pirr r6rVariablertrr load_funcr7 get_placerrrload_combine_funcziprQnparray)executorrrrrvars_from_memoryrrr$ load_var_mapr file_path load_var_listload_var_namesrs r%rrsL  z(H=== =''""7+2~ |  /1L, (=(=>>rswyEtFsGH ( 5 { !!99 ,.("<"<   %!%9##9    /1L, (=(=>>rswyEtFsGH   A.))!&&1Cc6;;#8#8#A#ABB BB?$XGGLL&&9 NN$..88: (+QVV$%(  MN|0023$$\%7%B%B%DE%%d+4 5(77<<:  " "**446  !? bhhsm,@# G:s ;MMc ^ ^[RRU5nUS:wdS5eSU;aUSn[R"S5 [ U[ 5(d[S[U535eUS:dUS:a[SU35e[RRU5nU(a:[RRU5(d[R"U5 S n[U5upxUV s0sH(oR(dMU RU"U 5_M* n n UV s0sH)n U R(dMU RU"U 5_M+ n n [X5n [ R"S :Xa}[ R$R&S :Xa_[(R*"XS 9m[-US -S5n Sm U R/U U4Sj[1S[3T5T 555 SSS5 O-[-US -S5n [(R4"XUS 9 SSS5 [-US-S5n [(R4"XUS 9 SSS5 [6R8R;XS-5 gs sn fs sn f!,(df  Nh=f!,(df  Ny=f!,(df  N]=f)a  This function saves parameters, optimizer information and network description to model_path. The parameters contain all the trainable Tensor, and save to a file with suffix ".pdparams". The optimizer information contains all the Tensor used by optimizer. For Adam optimizer, contains beta1, beta2, momentum etc. All the information will be saved to a file with suffix ".pdopt". (If the optimizer has no Tensor to save (like SGD), the file will not be generated). The network description is the description of the program. It's only used for deployment. The description will be saved to a file with a suffix ".pdmodel". Args: program(Program) : The program to be saved. model_path(str): The file prefix to save the program. The format is "dirname/file_prefix". If file_prefix is an empty str, an exception will be raised. protocol(int, optional): The protocol version of pickle module must be greater than 1 and less than 5. Default: 4 configs(dict, optional) : Optional keyword arguments. Returns: None rzThe input model_path MUST be format of dirname/filename [dirname\filename in Windows system], but received model_path is empty string.pickle_protocolzJ'pickle_protocol' is a deprecated argument. Please use 'protocol' instead.z+The 'protocol' MUST be `int`, but received z/Expected 1<'protocol'<5, but received protocol=c[5RUR5R5n[R "U5$r)r r6rr7rr)r$rNs r%r7save_pir..get_tensorRs/ N # #CHH - 8 8 :xx{r'darwin)protocol .pdparamswbi@c32># UH nTXT-v M g7fr)rr_ max_bytes pickle_bytess r%rsave_pir..is!?AQY/?srN.pdopt.json)rrbasenamerrrkintrtrorexistsmakedirsr&r rrsysplatform version_infomajorpickledumpsopen writelinesrangerpdumpr:rserialize_pir_program)r! model_pathrconfigs base_namedir_namer7parameter_listoptimizer_param_listr$ param_dictopt_dictfrrs @@r%save_pirr&sb(  ,I ? R?G#,- X  h $ $9$x.9I J  !|x!|=hZ H  wwz*Hx00 H ,>g+F(N-;-;c!*S/!^ ('C ?? "*S/!' $J9J ||xC$4$4$:$:a$?||JB *{*D 1QI LLq#l"3Y? 2 1*{*D 1Q KK  92 j8#T *a H(3 + KK%%gG/CD=2 12 1 + *s< J "J ?JJ3JJ% J6 J"% J36 KcnUb[U[5(deUS-n[U5upV[US5n[R S:Xa*[R RS:Xa [XG5nO [USS9n[U5nSSS5 UH`n U R(dMU RW;dSU RS US 35e[U RXR5 Mb [U5S :GaUS -n [R R#U 5(d S U S35eU(aG[$R&R(R*R-U[/5UR05 [U S5n[USS9nSSS5 UH`n U R(dMU RW;dSU RS U S 35e[U RXR5 Mb gg!,(df  GN=f!,(df  N=f)a :api_attr: PIR Static Graph This function gets parameters and optimizer information from program, and then gets corresponding value from file. An exception will be thrown if shape or dtype of the parameters does not match. This function can also load model file saved with [ save_params, save_persistables, save_vars ]. var_list can not be None when loading a single model file ( filename is not None when save_params, save_persistables or save_vars is called ). Args: program(Program): The program to be loaded model_prefix(str): The file prefix to store the program executor(Executor, optional): The executor used for initializing the parameter when startup program is not run. var_list(list|tuple, optional): The Tensor list/tuple to load a single model file saved with [ save_params, save_persistables, save_vars ]. Default: None Returns: None Nrrbrrlatin1)encodingzCan not find [z] in model file []rrzOptimizer file [z ] not exits)rkr r&rrrrrrrrr rrQrprrrr:r;rrrr r) r! model_prefixrvar_listparameter_file_namerrr load_dictr$ opt_file_names r%load_pirr ys2  z(H=== =&4,>g+F(N !4 (A <<8 #(8(8(>(>!(C)*=AI)!h?I%i0 ) ???88y(   *;>> import paddle >>> import numpy as np >>> paddle.enable_static() # Build the model >>> startup_prog = paddle.static.default_startup_program() >>> main_prog = paddle.static.default_main_program() >>> with paddle.static.program_guard(main_prog, startup_prog): ... image = paddle.static.data(name="img", shape=[64, 784]) ... w = paddle.create_parameter(shape=[784, 200], dtype='float32') ... b = paddle.create_parameter(shape=[200], dtype='float32') ... hidden_w = paddle.matmul(x=image, y=w) ... hidden_b = paddle.add(hidden_w, b) >>> exe = paddle.static.Executor(paddle.CPUPlace()) >>> exe.run(startup_prog) # Save the inference model >>> path_prefix = "./infer_model" >>> paddle.static.save_inference_model(path_prefix, [image], [hidden_b], exe) >>> [inference_program, feed_target_names, fetch_targets] = ( ... paddle.static.load_inference_model(path_prefix, exe)) >>> tensor_img = np.array(np.random.random((64, 784)), dtype=np.float32) >>> results = exe.run(inference_program, ... feed={feed_target_names[0]: tensor_img}, ... fetch_list=fetch_targets) # In this example, the inference program was saved in file # "./infer_model.pdmodel" and parameters were saved in file # " ./infer_model.pdiparams". # By the inference program, feed_target_names and # fetch_targets, we can use an executor to run the inference # program to get the inference result. )model_filenamer)pserver_endpointsNzKLoad inference model from memory is deprecated. Please specify path_prefix.rrz8params_filename cannot be None when path_prefix is None.rrzThere is no directory named rr  __model__rz[The old way to load inference model is deprecated. Please specify path_prefix. model path: z, params path: )inspect currentframef_codeco_namer_loggerwarningrrtr:rlrmr;rdeserialize_pir_programr&r rprrrrrrrrrr3)rrrsupported_argsdeprecated_argscallerrrr!r"r#rr$dir_pathrr load_dirname feed_namesr1s r%load_inference_model_pirr,sF;N,O  ! ! # * * 2 2F@ Y  $4d; **%6=  "J  --'') 00I)'2 }#7tt7 t9q= $(  -[9 77??;/ww}}X&&;H:FG G$w.J% 4K$ZZ(8$?N$jj):DAO%WW\\+{C WW\\+7OP ww~~j11!#k!JJ& ggll;;  ggllr8s.    8.   gll H  &)2:6 "1J!HcL  \H\HD  v-r OE OEd @7 @7F S Sl \0 \0r'