id
int64 0
458k
| file_name
stringlengths 4
119
| file_path
stringlengths 14
227
| content
stringlengths 24
9.96M
| size
int64 24
9.96M
| language
stringclasses 1
value | extension
stringclasses 14
values | total_lines
int64 1
219k
| avg_line_length
float64 2.52
4.63M
| max_line_length
int64 5
9.91M
| alphanum_fraction
float64 0
1
| repo_name
stringlengths 7
101
| repo_stars
int64 100
139k
| repo_forks
int64 0
26.4k
| repo_open_issues
int64 0
2.27k
| repo_license
stringclasses 12
values | repo_extraction_date
stringclasses 433
values |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
2,286,800 | serverperavg.cpython-39.pyc | hkgdifyu_pFedPT/system/flcore/servers/__pycache__/serverperavg.cpython-39.pyc | a
f¾`c‚ ã @ sH d dl Z d dlZd dlmZ d dlmZ d dlmZ G dd„ deƒZdS )é N)Ú clientPerAvg)ÚServer)ÚThreadc s, e Zd Z‡ fdd„Zdd„ Zdd„ Z‡ ZS )ÚPerAvgc sF t ƒ ||¡ | ¡ | |t¡ td| j› d| j› �ƒ tdƒ d S )Nz
Join ratio / total clients: z / z%Finished creating server and clients.)ÚsuperÚ__init__Úset_slow_clientsÚ set_clientsr ÚprintÚ
join_ratioÚ num_clients)ÚselfÚargsÚtimes©Ú __class__© úu/media/sim812/391e55df-b6f2-4fe9-a920-53434a8506fa/lgh/pdept/PFL-Non-IID-master/system/flcore/servers/serverperavg.pyr s
zPerAvg.__init__c C sª t | jd ƒD ]p}| ¡ | _| ¡ || j dkrRtd|› d�ƒ tdƒ | ¡ | jD ]}| ¡ | ¡ qX| ¡ |
¡ qtdƒ tt | j ƒƒ |
¡ | ¡ d S )Né r z
-------------Round number: z
-------------z+
Evaluate global model with one step updatez
Best global accuracy.)ÚrangeÚ
global_roundsÚselect_clientsÚselected_clientsÚ send_modelsÚeval_gapr
Úevaluate_one_stepÚtrainÚreceive_modelsÚaggregate_parametersÚmaxÚ rs_test_accÚ save_resultsÚsave_global_model)r
ÚiÚclientr r r r s
z PerAvg.trainc C s’ g }| j D ] }| t |j¡¡ | ¡ q
| ¡ }t| j ƒD ]\}}| || |j¡ q<t |d ƒd t |d ƒ }| j
|¡ t d |¡ƒ d S )Né g ğ?r z Average Test Accurancy: {:.4f})
ÚclientsÚappendÚcopyÚdeepcopyÚmodelZtrain_one_stepÚ test_metricsÚ enumerateÚ clone_modelÚsumr r
Úformat)r
Z models_tempÚcÚstatsr# Útest_accr r r r 4 s
zPerAvg.evaluate_one_step)Ú__name__Ú
__module__Ú __qualname__r r r Ú
__classcell__r r r r r s
!r ) r( ÚtorchZflcore.clients.clientperavgr Úflcore.servers.serverbaser Ú threadingr r r r r r Ú<module> s
| 2,060 | Python | .py | 22 | 92.363636 | 347 | 0.491417 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,801 | serverreppt.cpython-37.pyc | hkgdifyu_pFedPT/system/flcore/servers/__pycache__/serverreppt.cpython-37.pyc | B
Ðecý ã @ sh d dl mZ d dlmZ d dlmZ d dlZd dlZd dlZd dl Z d dl
Z
d dlZG dd„ deƒZ dS )é )Ú clientREPPT)ÚServer)ÚThreadNc s< e Zd Z‡ fdd„Zdd„ Zdd„ Zdd„ Zd d
„ Z‡ ZS ) Ú PFedRepPTc sn t ƒ ||¡ || _| ¡ | |t¡ t |jj ¡| _
g | _ g | _ t
d| j› d| j› �ƒ t
dƒ g | _d S )Nz
Join ratio / total clients: z / z%Finished creating server and clients.)ÚsuperÚ__init__ÚargsÚset_slow_clientsÚ set_clientsr ÚcopyÚdeepcopyÚmodelÚbaseÚ global_modelÚdiff_proÚclients_divergeÚprintÚ
join_ratioÚ num_clientsÚBudget)Úselfr Útimes)Ú __class__© úH/root/autodl-tmp/PFL-Non-IID-master/system/flcore/servers/serverreppt.pyr s zPFedRepPT.__init__c
C sÌ �xft | jd ƒD �]R}t ¡ }| ¡ | _| ¡ || j dkr`td|› d�ƒ tdƒ | ¡ d}x"| jD ]}| ¡ }||
¡ }qlW d}xdt | j d j
j ¡ | j d j
j ¡ ƒD ]:\}}|| } t | dk| t | ¡| ¡} |t | ¡ }q´W td |
¡ ¡ƒ | j |
¡ ¡ td |¡ƒ | j |¡ | ¡ | ¡ | j t ¡ | ¡ tdd d| jd
ƒ qW td ƒ tt| j ƒƒ td ƒ tt| jdd … ƒt | jdd … ƒ ƒ | ¡ | ¡ | ¡ d S )
Né r z
-------------Round number: z
-------------z
Evaluate global modelz"0 and 1 clients difference: {:.4f}z"Averaged prompr difference: {:.4f}z-------------------------z time costéÿÿÿÿz
Best global accuracy.z
Average time cost per round.)!ÚrangeÚ
global_roundsÚtimeÚselect_clientsÚselected_clientsÚ send_modelsÚeval_gapr ÚevaluateÚtrainÚitemÚzipÚclientsr
Ú generatorÚ
parametersÚtorchÚwhereÚ
zeros_likeÚsumÚformatr Úappendr Úreceive_modelsÚaggregate_parametersr ÚmaxÚ rs_test_accÚlenÚ save_resultsÚsave_global_modelÚsave_client_model)
r ÚiÚs_tÚ
temp_diff_proÚclientÚtemp_diff_pro_clientÚdiverge_clentsÚ new_paramÚ old_paramr r r r r% s@
0 (zPFedRepPT.trainc C sŒ t | jƒdkst‚d}x| jD ]}||j7 }q W g | _g | _g | _xD| jD ]:}| j |j| ¡ | j |j¡ | j t
|j j ¡¡ qJW d S )Nr )
r5 r! ÚAssertionErrorÚ
train_samplesÚuploaded_weightsÚ uploaded_idsÚuploaded_modelsr0 Úidr r r
r )r Úactive_train_samplesr< r r r r1 M s zPFedRepPT.receive_modelsc C s t j d| jd¡}t j |¡s(t |¡ x”t| jƒD ]†\}}t j || jd t |ƒ d t | j
j ƒ d t | j
j ƒ d t | j
j
ƒ d t | j
jƒ d t | j
jƒ d ¡}t |j|¡ q4W d S )NÚmodelsr< Ú_clientÚ_z.pt)ÚosÚpathÚjoinÚdatasetÚexistsÚmakedirsÚ enumerater( Ú algorithmÚstrr Ú
num_promptr r Ú plocal_stepsr r+ Úsaver
)r Ú
model_pathÚc_idxÚcÚmodel_path_saver r r r8 [ s
pzPFedRepPT.save_client_modelc C s8 | j d | j }d}tj |¡s*t |¡ t| jƒ�r4|d | j d t | j
ƒ d t | j j ƒ d t | j j
ƒ d t | j jƒ d t | j jƒ d t | j jƒ }|d |¡ }td| ƒ t |d¡�f}|jd| jd� |jd| jd� |jd | jd� |jd
| jd� |jd | jd� |jd | jd� W d Q R X d S )
NrJ z ../results/z{}.h5z File path: Úwr4 )ÚdataÚrs_test_acc_stdÚ rs_test_aucÚ
rs_train_lossr r )rN rR rK rL rO rP r5 r4 ÚgoalrS r r rT r r rU r r/ r Úh5pyÚFileÚcreate_datasetr] r^ r_ r r )r ÚalgoÚ result_pathÚ file_pathÚhfr r r r6 c s
l zPFedRepPT.save_results) Ú__name__Ú
__module__Ú __qualname__r r% r1 r8 r6 Ú
__classcell__r r )r r r s
0r ) Z!system.flcore.clients.clientrepptr Ú system.flcore.servers.serverbaser Ú threadingr r r r+ rK ra r r r r r Ú<module> s | 4,071 | Python | .py | 35 | 115.057143 | 397 | 0.461977 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,802 | servermtl.cpython-37.pyc | hkgdifyu_pFedPT/system/flcore/servers/__pycache__/servermtl.cpython-37.pyc | B
·:cc ã @ s@ d dl Z d dlmZ d dlmZ d dlmZ G dd„ deƒZdS )é N)Ú clientMTL)ÚServer)ÚThreadc s4 e Zd Z‡ fdd„Zdd„ Zdd„ Zdd„ Z‡ ZS ) ÚFedMTLc sÈ t ƒ ||¡ t| | j¡ƒ| _tj| j| jf|j d�| _
|j | _ t | j| jf¡}t | jdf¡}|d| j | |j
¡ d }| |j ¡| _| ¡ | |t¡ td| j› d| j› �ƒ tdƒ d S )N)Údeviceé é z
Join clients / total clients: z / z%Finished creating server and clients.)ÚsuperÚ__init__ÚlenÚflattenÚ global_modelÚdimÚtorchÚzerosÚ join_clientsr ÚW_globÚonesÚmmÚTÚtoÚomegaÚset_slow_clientsÚ set_clientsr ÚprintÚ num_clients)ÚselfÚargsÚtimesÚIÚir )Ú __class__© úF/root/autodl-tmp/PFL-Non-IID-master/system/flcore/servers/servermtl.pyr
s zFedMTL.__init__c C s´ xˆt | jd ƒD ]v}| ¡ | _| ¡ || j dkrTtd|› d�ƒ tdƒ | ¡ x0t| jƒD ]"\}}| | j
| j |¡ | ¡ q`W qW tdƒ tt
| jƒƒ | ¡ | ¡ d S )Nr r z
-------------Round number: z
-------------z
Evaluate global modelz
Best global accuracy.)ÚrangeÚ
global_roundsÚselect_clientsÚselected_clientsÚaggregate_parametersÚeval_gapr ÚevaluateÚ enumerateZreceive_valuesr r ÚtrainÚmaxÚ rs_test_accÚ save_resultsÚsave_global_model)r r ÚidxÚclientr" r" r# r, s
z FedMTL.trainc s, | ¡ ‰ ˆ ¡ }‡ fdd„|D ƒ}t |¡S )Nc s g | ]}ˆ | ¡ ‘qS r" )r )Ú.0Úkey)Ú
state_dictr" r# ú
<listcomp>< s z"FedMTL.flatten.<locals>.<listcomp>)r5 Úkeysr Úcat)r Úmodelr7 ÚWr" )r5 r# r 9 s zFedMTL.flattenc C sP t j| j| jf| jd�| _x0t| jƒD ]"\}}| |j ¡| jd d …|f< q&W d S )N)r )
r r r r r r r+ r' r r9 )r r1 r2 r" r" r# r( ? s zFedMTL.aggregate_parameters)Ú__name__Ú
__module__Ú __qualname__r
r, r r( Ú
__classcell__r" r" )r! r# r s r )r Zflcore.clients.clientmtlr Úflcore.servers.serverbaser Ú threadingr r r" r" r" r# Ú<module> s | 2,354 | Python | .py | 15 | 155.733333 | 367 | 0.440598 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,803 | serverbase.cpython-39.pyc | hkgdifyu_pFedPT/system/flcore/servers/__pycache__/serverbase.cpython-39.pyc | a
ubc‹# ã @ sX d dl Z d dlZd dlZd dlZd dlZd dlZd dlZd dlm Z G dd„ de
ƒZ dS )é N)Úread_client_datac @ s° e Zd Zdd„ Zdd„ Zdd„ Zdd„ Zd d
„ Zd d „ Zd
d„ Z dd„ Z
dd„ Z dd„ Z dd„ Z
dd„ Zdd„ Zdd „ Zd d „ Zd d „ Zd!d"„ Zd*d$d%„Zd&d'„ Zd+d(d)„Zd#S ),ÚServerc C sô |j | _ |j| _|j| _|j| _|j| _|j| _t |j ¡| _
|j | _ |j | _ t
| j | j ƒ| _|j| _|j| _|j| _|j| _|j| _d| _g | _g | _g | _g | _g | _g | _g | _g | _ g | _ g | _ || _ |j | _ |j!| _!|j"| _"|j#| _#d S )Néd )$ÚdeviceÚdatasetÚ
global_roundsÚ local_stepsÚ
batch_sizeÚlocal_learning_rateÚ
learning_rateÚcopyÚdeepcopyÚmodelÚ global_modelÚ num_clientsÚ
join_ratioÚintÚ join_clientsÚ algorithmÚ time_selectÚgoalÚtime_thretholdÚsave_folder_nameÚtop_cntÚclientsÚselected_clientsÚtrain_slow_clientsÚsend_slow_clientsÚuploaded_weightsÚ uploaded_idsÚuploaded_modelsÚ rs_test_accÚ rs_test_aucÚ
rs_train_lossÚtimesÚeval_gapÚclient_drop_rateÚtrain_slow_rateÚsend_slow_rate)ÚselfÚargsr$ © r+ ús/media/sim812/391e55df-b6f2-4fe9-a920-53434a8506fa/lgh/pdept/PFL-Non-IID-master/system/flcore/servers/serverbase.pyÚ__init__
s> zServer.__init__c C sn t t| jƒ| j| jƒD ]R\}}}t| j|dd�}t| j|dd�}|||t|ƒt|ƒ||d�}| j |¡ qd S )NT)Úis_trainF)ÚidÚ
train_samplesÚ test_samplesÚ
train_slowÚ send_slow)
ÚzipÚranger r r r r Úlenr Úappend) r) r* Z clientObjÚir2 r3 Ú
train_dataÚ test_dataÚclientr+ r+ r, Ú set_clients4 s ûzServer.set_clientsc C sV dd„ t | jƒD ƒ}dd„ t | jƒD ƒ}tj |t|| j ƒ¡}|D ] }d||< qD|S )Nc S s g | ]}d ‘qS )Fr+ ©Ú.0r8 r+ r+ r, Ú
<listcomp>E ó z.Server.select_slow_clients.<locals>.<listcomp>c S s g | ]}|‘qS r+ r+ r= r+ r+ r, r? F r@ T)r5 r ÚnpÚrandomÚchoicer )r) Z slow_rateZ slow_clientsÚidxÚidx_r8 r+ r+ r, Úselect_slow_clientsD s
zServer.select_slow_clientsc C s | | j¡| _| | j¡| _d S ©N)rF r' r r( r )r) r+ r+ r, Úset_slow_clientsM s ÿÿzServer.set_slow_clientsc C s t tjj| j| jdd�ƒ}|S )NF)Úreplace)ÚlistrA rB rC r r )r) r r+ r+ r, Úselect_clientsS s zServer.select_clientsc C s. t | jƒdksJ ‚| jD ]}| | j¡ qd S ©Nr )r6 r Úset_parametersr )r) r; r+ r+ r, Ú send_modelsX s
zServer.send_modelsc C s� t | jƒdksJ ‚g | _d}g | _g | _| jD ]8}| j |j¡ ||j7 }| j |j¡ | j |j¡ q.t | jƒD ]\}}|| | j|< qrd S rL )
r6 r r r r r7 r0 r/ r Ú enumerate)r) Z tot_samplesr; r8 Úwr+ r+ r, Úreceive_models` s
zServer.receive_modelsc C sf t | jƒdksJ ‚t | jd ¡| _| j ¡ D ]}|j ¡ q.t| j | jƒD ]\}}|
||¡ qLd S rL ) r6 r r r
r Ú
parametersÚdataÚzero_r4 r Úadd_parameters)r) ÚparamrP Ú client_modelr+ r+ r, Úaggregate_parameterso s zServer.aggregate_parametersc C s: t | j ¡ | ¡ ƒD ] \}}| j|j ¡ | 7 _qd S rG )r4 r rR rS Úclone)r) rP rW Z server_paramZ client_paramr+ r+ r, rU y s zServer.add_parametersc C sP t j d| j¡}t j |¡s&t |¡ t j || jd d ¡}t | j |¡ d S ©NÚmodelsÚ_serverú.pt)
ÚosÚpathÚjoinr ÚexistsÚmakedirsr ÚtorchÚsaver ©r) Z
model_pathr+ r+ r, Úsave_global_model} s
zServer.save_global_modelc C sH t j d| j¡}t j || jd d ¡}t j |¡s8J ‚t |¡| _d S rZ ) r^ r_ r` r r ra rc Úloadr re r+ r+ r, Ú
load_model„ s zServer.load_modelc C s4 t j d| j¡}t j || jd d ¡}t j |¡S rZ )r^ r_ r` r r ra re r+ r+ r, Ú model_existsŠ s zServer.model_existsc C sÊ | j d | j }d}tj |¡s*t |¡ t| jƒrÆ|d | j d t | j
ƒ }|d |¡ }t d| ƒ t
|d¡�@}|jd| jd� |jd| jd� |jd | jd� W d ƒ n1 s¼0 Y d S )
NÚ_z ../results/z{}.h5z File path: rP r! )rS r" r# )r r r^ r_ ra rb r6 r! r Ústrr$ ÚformatÚprintÚh5pyÚFileZcreate_datasetr" r# )r) ÚalgoÚ result_pathÚ file_pathZhfr+ r+ r, Ú save_results� s
zServer.save_resultsc C s> t j | j¡st | j¡ t |t j | jd| d ¡¡ d S ©NZserver_r] )r^ r_ ra r rb rc rd r` )r) ÚitemÚ item_namer+ r+ r, Ú save_itemŸ s zServer.save_itemc C s t tj | jd| d ¡¡S rt )rc rg r^ r_ r` r )r) rv r+ r+ r, Ú load_item¤ s zServer.load_itemc C sh g }g }g }| j D ]8}| ¡ \}}}| |d ¡ | || ¡ | |¡ qdd„ | j D ƒ}||||fS )Nç ğ?c S s g | ]
}|j ‘qS r+ ©r/ ©r> Úcr+ r+ r, r? ´ r@ z'Server.test_metrics.<locals>.<listcomp>)r Ú test_metricsr7 ) r) Ú num_samplesZ tot_correctZtot_aucr| ÚctÚnsÚaucÚidsr+ r+ r, r} § s
zServer.test_metricsc C sR g }g }| j D ](}| ¡ \}}| |¡ | |d ¡ qdd„ | j D ƒ}|||fS )Nry c S s g | ]
}|j ‘qS r+ rz r{ r+ r+ r, r? À r@ z(Server.train_metrics.<locals>.<listcomp>)r Ú
train_metricsr7 )r) r~ Úlossesr| Úclr€ r‚ r+ r+ r, rƒ ¸ s
zServer.train_metricsNc
C s> | ¡ }| ¡ }t|d ƒd t|d ƒ }t|d ƒd t|d ƒ }t|d ƒd t|d ƒ }dd„ t|d |d ƒD ƒ}dd„ t|d |d ƒD ƒ} |d kr²| j |¡ n
| |¡ | j |¡ |d krŞ| j |¡ n
| |¡ td |¡ƒ td |¡ƒ td
|¡ƒ td t
|¡¡ƒ td t
| ¡¡ƒ d S )
Né ry é é c S s g | ]\}}|| ‘qS r+ r+ ©r> ÚaÚnr+ r+ r, r? Í r@ z#Server.evaluate.<locals>.<listcomp>c S s g | ]\}}|| ‘qS r+ r+ r‰ r+ r+ r, r? Î r@ zAveraged Train Loss: {:.4f}z Averaged Test Accurancy: {:.4f}zAveraged Test AUC: {:.4f}zStd Test Accurancy: {:.4f}zStd Test AUC: {:.4f}) r} rƒ Úsumr4 r! r7 r" r# rm rl rA Ústd)
r) ÚaccÚlossÚstatsZ stats_trainÚtest_accÚtest_aucÚ
train_lossZaccsZaucsr+ r+ r, ÚevaluateÅ s&
zServer.evaluatec C s. t d |¡ƒ t d |¡ƒ t d |¡ƒ d S )Nz Average Test Accurancy: {:.4f}zAverage Test AUC: {:.4f}zAverage Train Loss: {:.4f})rm rl )r) r‘ r’ r“ r+ r+ r, Úprint_á s z
Server.print_c C sğ |D ]æ}|d krr|d krrt |ƒt t |¡d¡jd |k}t |ƒdko^t || d … ¡|k }|rj|rjqê dS q|d kr¬t |ƒt t |¡d¡jd |k}|r¤qê dS q|d kræt |ƒdkoÖt || d … ¡|k }|rŞqê dS qt‚qdS )Nr‡ r FT)r6 rc ÚtopkÚtensorÚindicesrA r� ÚNotImplementedError)r) Zacc_lssr Ú div_valueZacc_lsZfind_topZfind_divr+ r+ r, Ú
check_doneæ s& $$$$zServer.check_done)NN)NN)Ú__name__Ú
__module__Ú __qualname__r- r< rF rH rK rN rQ rX rU rf rh ri rs rw rx r} rƒ r” r• r› r+ r+ r+ r, r s( '
r ) rc r^ ÚnumpyrA rn r ÚtimerB Úutils.data_utilsr Úobjectr r+ r+ r+ r, Ú<module> s | 8,366 | Python | .py | 47 | 176.787234 | 760 | 0.4 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,804 | serverbn.cpython-39.pyc | hkgdifyu_pFedPT/system/flcore/servers/__pycache__/serverbn.cpython-39.pyc | a
f¾`cã ã @ sL d dl mZ d dlmZ d dlmZ d dlmZ d dlZG dd„ deƒZ dS )é )ÚclientBN)ÚServer)Úread_client_data)ÚThreadNc s$ e Zd Z‡ fdd„Zdd„ Z‡ ZS )ÚFedBNc sF t ƒ ||¡ | ¡ | |t¡ td| j› d| j› �ƒ tdƒ d S )Nz
Join ratio / total clients: z / z%Finished creating server and clients.)ÚsuperÚ__init__Úset_slow_clientsÚ set_clientsr ÚprintÚ
join_ratioÚ num_clients)ÚselfÚargsÚtimes©Ú __class__© úq/media/sim812/391e55df-b6f2-4fe9-a920-53434a8506fa/lgh/pdept/PFL-Non-IID-master/system/flcore/servers/serverbn.pyr s
zFedBN.__init__c C s¢ t | jd ƒD ]h}| ¡ | _| ¡ || j dkrRtd|› d�ƒ tdƒ | ¡ | jD ] }| ¡ qX| ¡ |
¡ qtdƒ tt | j ƒƒ |
¡ | ¡ d S )Né r z
-------------Round number: z
-------------z
Evaluate global modelz
Best global accuracy.)ÚrangeÚ
global_roundsÚselect_clientsÚselected_clientsÚ send_modelsÚeval_gapr ÚevaluateÚtrainÚreceive_modelsÚaggregate_parametersÚmaxÚ rs_test_accÚ save_resultsÚsave_global_model)r ÚiÚclientr r r r s
z FedBN.train)Ú__name__Ú
__module__Ú __qualname__r r Ú
__classcell__r r r r r s
r )
Zflcore.clients.clientbnr Úflcore.servers.serverbaser Úutils.data_utilsr Ú threadingr Útimer r r r r Ú<module> s
| 1,562 | Python | .py | 16 | 96.375 | 358 | 0.505495 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,805 | serverpfedpt.cpython-38.pyc | hkgdifyu_pFedPT/system/flcore/servers/__pycache__/serverpfedpt.cpython-38.pyc | U
Õ şcš ã @ sh d dl mZ d dlmZ d dlmZ d dlZd dlZd dlZd dl Z d dl
Z
d dl Z G dd„ deƒZ
dS )é )ÚclientPT)ÚServer)ÚThreadNc sN e Zd Z‡ fdd„Zdd„ Zdd„ Zdd„ Zd d
„ Zdd d
„Zdd„ Z ‡ Z
S )ÚPFedPTc sn t ƒ ||¡ | ¡ || _| |t¡ t |jj ¡| _
g | _ g | _ t
d| j› d| j› �ƒ t
dƒ g | _d S )Nz
Join ratio / total clients: z / z%Finished creating server and clients.)ÚsuperÚ__init__Úset_slow_clientsÚargsÚ set_clientsr ÚcopyÚdeepcopyÚmodelÚbaseÚ global_modelÚdiff_proÚclients_divergeÚprintÚ
join_ratioÚ num_clientsÚBudget)Úselfr Útimes©Ú __class__© õd D:\京东\promot\第二次投稿\å®�验\æœ�务器\native - pro\system\flcore\servers\serverpfedpt.pyr s zPFedPT.__init__c C sú g }t | jd ƒD �]n}t ¡ }| ¡ | _| ¡ || j dkr`td|› d�ƒ tdƒ | ¡ d}| jD ]}| ¡ }||
¡ }qjtd |¡ƒ | j
|¡ d}t| jd jj ¡ | jd jj ¡ ƒD ]:\}} || }
t |
dk|
t |
¡|
¡}
|t |
¡ }qÈtd |
¡ ¡ƒ | j
|
¡ ¡ || j dk�rJtdƒ | j|d � | ¡ | ¡ | j
t ¡ | ¡ td
d d
| jd ƒ qtd
ƒ tt| j ƒƒ tdƒ tt| jdd … ƒt | jdd … ƒ ƒ tdƒ tt|ƒƒ | ¡ | ¡ | ¡ d S )Né r z
-------------Round number: z
-------------z
Evaluate global modelz"Averaged prompr difference: {:.4f}z"0 and 1 clients difference: {:.4f}z
Evaluate local model)Úaccz-------------------------z time costéÿÿÿÿz
Best global accuracy.z
Average time cost per round.z
Best local accuracy.)!ÚrangeÚ
global_roundsÚtimeÚselect_clientsÚselected_clientsÚ send_modelsÚeval_gapr ÚevaluateÚtrainÚitemÚformatr ÚappendÚzipÚclientsr
Ú generatorÚ
parametersÚtorchÚwhereÚ
zeros_likeÚsumr Úreceive_modelsÚaggregate_parametersr ÚmaxÚ rs_test_accÚlenÚ save_resultsÚsave_global_modelÚsave_client_model) r Ú local_accÚiÚs_tÚ
temp_diff_proÚclientÚtemp_diff_pro_clientÚdiverge_clentsÚ new_paramÚ old_paramr r r r r' sL
. ( z PFedPT.trainc C s„ t | jƒdkst‚d}| jD ]}||j7 }q g | _g | _g | _| jD ]:}| j |j| ¡ | j |j¡ | j t
|j j ¡¡ qDd S )Nr )
r7 r# ÚAssertionErrorÚ
train_samplesÚuploaded_weightsÚ uploaded_idsÚuploaded_modelsr* Úidr r r
r )r Úactive_train_samplesr? r r r r3 S s
zPFedPT.receive_modelsc C s t j d| jd| jjd | jj d | jj d | jj d | jj d | jj
¡}t j |¡sjt |¡ t
| jƒD ]†\}}t j || jd t|ƒ d t| jjƒ d t| jjƒ d t| jjƒ d t| jjƒ d t| jjƒ d ¡}t |j|¡ qtd S )NÚmodelsr? Ú*Ú_clientÚ_z.pt)ÚosÚpathÚjoinÚdatasetr Úarv1Úarv2Úarv3Úarv4Úarv5Úarv6ÚexistsÚmakedirsÚ enumerater, Ú algorithmÚstrÚ
num_promptr r Ú plocal_stepsr r/ Úsaver
)r Ú
model_pathÚc_idxÚcÚmodel_path_saver r r r: a s T
pzPFedPT.save_client_modelc C s€ | j d | j }d| jj d | jj d | jj d | jj d | jj d | jj d }t j
|¡srt |¡ t
| jƒ�r||d | j d t| jƒ d t| jjƒ d t| jjƒ d t| jjƒ d t| jjƒ d t| jjƒ }|d |¡ }td| ƒ t |d¡�f}|jd| jd � |jd
| j d � |jd | j d � |jd | j d � |jd
| j d � |jd| j d � W 5 Q R X d S )NrN z ../results/rL ú/z{}.h5z File path: Úwr6 )ÚdataÚrs_test_acc_stdÚ rs_test_aucÚ
rs_train_lossr r )!rR r\ r rS rT rU rV rW rX rO rP rY rZ r7 r6 Úgoalr] r r^ r r r_ r r) r Úh5pyÚFileÚcreate_datasetrh ri rj r r )r ÚalgoÚ result_pathÚ file_pathÚhfr r r r8 i s L
l zPFedPT.save_resultsNc C s| | ¡ }| ¡ }t|d ƒd t|d ƒ }t|d ƒd t|d ƒ }t|d ƒd t|d ƒ }t|d ƒd t|d ƒ }dd„ t|d |d ƒD ƒ} dd„ t|d |d ƒD ƒ}
|d krÎ| j |¡ n
| |¡ | j |¡ | j t | ¡¡ |d k�r| j
|¡ n
| |¡ t d |¡ƒ t d
|¡ƒ t d |¡ƒ t d |¡ƒ t d
t | ¡¡ƒ t d t |
¡¡ƒ d S )Né ç ğ?r é é c S s g | ]\}}|| ‘qS r r ©Ú.0ÚaÚnr r r Ú
<listcomp>… s z#PFedPT.evaluate.<locals>.<listcomp>c S s g | ]\}}|| ‘qS r r rw r r r r{ † s zAveraged Train Loss: {:.4f}z Averaged Test Accurancy: {:.4f}z$Averaged Test oral Accurancy: {:.4f}zAveraged Test AUC: {:.4f}zStd Test Accurancy: {:.4f}zStd Test AUC: {:.4f})
Ú test_metricsÚ
train_metricsr2 r+ r6 r* ri rh ÚnpÚstdrj r r) ) r r ÚlossÚstatsÚ stats_trainÚtest_accÚ test_acc2Útest_aucÚ
train_lossÚaccsÚaucsr r r r& | s,
zPFedPT.evaluatec C s~ g }g }g }g }| j D ]H}| ¡ \}}}} | |d ¡ | |d ¡ | | | ¡ | |¡ qdd„ | j D ƒ}
|
||||fS )Nrt c S s g | ]
}|j ‘qS r )rI )rx rc r r r r{ ¨ s z'PFedPT.test_metrics.<locals>.<listcomp>)r, r| r* ) r Ú num_samplesÚ tot_correctÚ tot_correct2Útot_aucrc ÚctÚct2ÚnsÚaucÚidsr r r r| › s
zPFedPT.test_metrics)NN) Ú__name__Ú
__module__Ú __qualname__r r' r3 r: r8 r& r| Ú
__classcell__r r r r r s 4
r )Zflcore.clients.clientptr Úflcore.servers.serverbaser Ú threadingr r! r/ rO rl r Únumpyr~ r r r r r Ú<module> s | 6,091 | Python | .py | 58 | 103.758621 | 494 | 0.426086 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,806 | serverbabu.cpython-38.pyc | hkgdifyu_pFedPT/system/flcore/servers/__pycache__/serverbabu.cpython-38.pyc | U
[Ğıcr ã @ s8 d dl mZ d dlmZ d dlmZ G dd„ deƒZdS )é )Ú
clientBABU)ÚServer)ÚThreadc s, e Zd Z‡ fdd„Zdd„ Zdd„ Z‡ ZS )ÚFedBABUc sF t ƒ ||¡ | ¡ | |t¡ td| j› d| j› �ƒ tdƒ d S )Nz
Join ratio / total clients: z / z%Finished creating server and clients.)ÚsuperÚ__init__Úset_slow_clientsÚ set_clientsr ÚprintÚ
join_ratioÚ num_clients)ÚselfÚargsÚtimes©Ú __class__© õX D:\京东\promot\第二次投稿\å®�验\native - pro\system\flcore\servers\serverbabu.pyr s
zFedBABU.__init__c C s g }t | jd ƒD ]Š}| ¡ | _| ¡ || j dkrVtd|› d�ƒ tdƒ | ¡ | jD ] }| ¡ q\|| j dkrŒtdƒ | j|d� | ¡ |
¡ qtdƒ tt | j ƒƒ td ƒ tt |ƒƒ | j
D ] }| ¡ qÎtd
ƒ | ¡ | ¡ | ¡ d S ) Né r z
-------------Round number: z
-------------z
Evaluate global modelz
Evaluate local model)Úaccz
Best global accuracy.z
Best local accuracy.z4
-------------Evaluate fine-tuned model-------------)ÚrangeÚ
global_roundsÚselect_clientsÚselected_clientsÚ send_modelsÚeval_gapr
ÚevaluateÚtrainÚreceive_modelsÚaggregate_parametersÚmaxÚ rs_test_accÚclientsZ fine_tuneÚ save_resultsÚsave_global_model)r
Ú local_accÚiÚclientr r r r s2
z
FedBABU.trainc C s� t | jƒdkst‚g | _d}g | _g | _| jD ]:}| j |j¡ ||j7 }| j |j¡ | j |j j
¡ q.t | jƒD ]\}}|| | j|< qtd S )Nr ) Úlenr ÚAssertionErrorÚuploaded_weightsÚ uploaded_idsÚuploaded_modelsÚappendÚ
train_samplesÚidÚmodelÚbaseÚ enumerate)r
Ú tot_samplesr' r& Úwr r r r ; s
zFedBABU.receive_models)Ú__name__Ú
__module__Ú __qualname__r r r Ú
__classcell__r r r r r s
'r N)Zflcore.clients.clientbabur Úflcore.servers.serverbaser Ú threadingr r r r r r Ú<module> s | 2,164 | Python | .py | 25 | 85.4 | 254 | 0.477103 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,807 | serverrep.cpython-38.pyc | hkgdifyu_pFedPT/system/flcore/servers/__pycache__/serverrep.cpython-38.pyc | U
ºĞıcí ã @ sH d dl mZ d dlmZ d dlmZ d dlZd dlZG dd„ deƒZdS )é )Ú clientRep)ÚServer)ÚThreadNc s, e Zd Z‡ fdd„Zdd„ Zdd„ Z‡ ZS )ÚFedRepc sL t ƒ ||¡ | ¡ | |t¡ td| j› d| j› �ƒ tdƒ g | _d S )Nz
Join ratio / total clients: z / z%Finished creating server and clients.) ÚsuperÚ__init__Úset_slow_clientsÚ set_clientsr ÚprintÚ
join_ratioÚ num_clientsÚBudget)ÚselfÚargsÚtimes©Ú __class__© õW D:\京东\promot\第二次投稿\å®�验\native - pro\system\flcore\servers\serverrep.pyr s zFedRep.__init__c C s< g }t | jd ƒD ]º}t ¡ }| ¡ | _| ¡ || j dkr^td|› d�ƒ tdƒ | ¡ | jD ] }| ¡ qd|| j dkr”tdƒ | j|d� |
¡ | ¡ | j
t ¡ | ¡ tdd d| j d
ƒ qtd ƒ tt| jƒƒ td ƒ tt| j dd … ƒt| j dd … ƒ ƒ td
ƒ tt|ƒƒ | ¡ | ¡ d S )Né r z
-------------Round number: z
-------------z
Evaluate global modelz
Evaluate local model)Úaccz-------------------------z time costéÿÿÿÿz
Best global accuracy.z
Average time cost per round.z
Best local accuracy.)ÚrangeÚ
global_roundsÚtimeÚselect_clientsÚselected_clientsÚ send_modelsÚeval_gapr
ÚevaluateÚtrainÚreceive_modelsÚaggregate_parametersr
ÚappendÚmaxÚ rs_test_accÚsumÚlenÚ save_resultsÚsave_global_model)r Ú local_accÚiÚs_tÚclientr r r r s4
( z FedRep.trainc C s„ t | jƒdkst‚d}| jD ]}||j7 }q g | _g | _g | _| jD ]:}| j |j| ¡ | j |j¡ | j t
|j j ¡¡ qDd S )Nr )
r' r ÚAssertionErrorÚ
train_samplesÚuploaded_weightsÚ uploaded_idsÚuploaded_modelsr# ÚidÚcopyÚdeepcopyÚmodelÚbase)r Úactive_train_samplesr- r r r r! > s
zFedRep.receive_models)Ú__name__Ú
__module__Ú __qualname__r r r! Ú
__classcell__r r r r r s 'r ) Zflcore.clients.clientrepr Úflcore.servers.serverbaser Ú threadingr r r4 r r r r r Ú<module> s
| 2,288 | Python | .py | 21 | 107.619048 | 383 | 0.472222 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,808 | serverrod.cpython-39.pyc | hkgdifyu_pFedPT/system/flcore/servers/__pycache__/serverrod.cpython-39.pyc | a
f¾`cç ã @ sL d dl mZ d dlmZ d dlmZ d dlmZ d dlZG dd„ deƒZ dS )é )Ú clientROD)ÚServer)Úread_client_data)ÚThreadNc s$ e Zd Z‡ fdd„Zdd„ Z‡ ZS )ÚFedRODc sF t ƒ ||¡ | ¡ | |t¡ td| j› d| j› �ƒ tdƒ d S )Nz
Join ratio / total clients: z / z%Finished creating server and clients.)ÚsuperÚ__init__Úset_slow_clientsÚ set_clientsr ÚprintÚ
join_ratioÚ num_clients)ÚselfÚargsÚtimes©Ú __class__© úr/media/sim812/391e55df-b6f2-4fe9-a920-53434a8506fa/lgh/pdept/PFL-Non-IID-master/system/flcore/servers/serverrod.pyr s
zFedROD.__init__c C s¢ t | jd ƒD ]h}| ¡ | _| ¡ || j dkrRtd|› d�ƒ tdƒ | ¡ | jD ] }| ¡ qX| ¡ |
¡ qtdƒ tt | j ƒƒ |
¡ | ¡ d S )Né r z
-------------Round number: z
-------------z
Evaluate global modelz
Best global accuracy.)ÚrangeÚ
global_roundsÚselect_clientsÚselected_clientsÚ send_modelsÚeval_gapr ÚevaluateÚtrainÚreceive_modelsÚaggregate_parametersÚmaxÚ rs_test_accÚ save_resultsÚsave_global_model)r ÚiÚclientr r r r s
z FedROD.train)Ú__name__Ú
__module__Ú __qualname__r r Ú
__classcell__r r r r r s
r )
Zflcore.clients.clientrodr Úflcore.servers.serverbaser Úutils.data_utilsr Ú threadingr Útimer r r r r Ú<module> s
| 1,568 | Python | .py | 16 | 96.75 | 360 | 0.507405 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,809 | serveramp.cpython-38.pyc | hkgdifyu_pFedPT/system/flcore/servers/__pycache__/serveramp.cpython-38.pyc | U
[Ğıc¢ ã @ sd d dl Z d dlZd dlZd dlZd dlZd dlmZmZ d dl m
Z
d dl m Z G dd„ de
ƒZ
dS )é N)Ú clientAMPÚweight_flatten)ÚServer)ÚThreadc s4 e Zd Z‡ fdd„Zdd„ Zdd„ Zdd„ Z‡ ZS ) ÚFedAMPc sV t ƒ ||¡ | ¡ | |t¡ |j| _|j| _td| j› d| j › �ƒ tdƒ d S )Nz
Join ratio / total clients: z / z%Finished creating server and clients.)
ÚsuperÚ__init__Úset_slow_clientsÚ set_clientsr ÚalphaKÚsigmaÚprintÚ
join_ratioÚ num_clients)ÚselfÚargsÚtimes©Ú __class__© õW D:\京东\promot\第二次投稿\å®�验\native - pro\system\flcore\servers\serveramp.pyr s zFedAMP.__init__c C sÔ g }t | jd ƒD ]‚}| ¡ | _| ¡ || j dkrVtd|› d�ƒ tdƒ | ¡ | jD ] }| ¡ q\|| j dkrŒtdƒ | j|d� | ¡ qtdƒ tt
| j ƒƒ td ƒ tt
|ƒƒ | ¡ |
¡ d S )
Né r z
-------------Round number: z
-------------z
Evaluate global modelz
Evaluate local model)Úaccz
Best global accuracy.z
Best local accuracy.)ÚrangeÚ
global_roundsÚselect_clientsÚselected_clientsÚ send_modelsÚeval_gapr
ÚevaluateÚtrainÚreceive_modelsÚmaxÚ rs_test_accÚ save_resultsÚsave_global_model)r Ú local_accÚiÚclientr r r r s(
z FedAMP.trainc
C sŒ t | jƒdkst‚t | jƒdk�rˆ| jD �]\}t | j¡}| ¡ D ]}|j
¡ qBt | j
¡}t| jƒD ]b\}}|j| j| krÂt|jƒ}t|ƒ}|| d¡} t | | ¡} | j| | ¡ ||< qhd||< qhdt |¡ }
t| jƒD ]:\}}t| ¡ | ¡ ƒD ] \}} | j || | 7 _ qşqät ¡ } |j�rJt dt t j ¡ ¡ ¡ | ||
¡ |j!d d7 < |j!d dt ¡ | 7 < q(d S )Nr éÿÿÿÿr gš™™™™™¹?Ú
num_roundsÚ
total_costé )"Úlenr ÚAssertionErrorÚuploaded_modelsÚclientsÚcopyÚdeepcopyÚ global_modelÚ
parametersÚdataÚzero_ÚtorchÚzerosÚ join_clientsÚ enumerateÚidÚ uploaded_idsr ÚmodelÚviewÚdotr ÚeÚsumÚzipÚtimeÚ send_slowÚsleepÚnpÚabsÚrandomÚrandÚset_parametersÚsend_time_cost)
r ÚcÚmuÚparamÚcoefÚjZmwZ weights_iZ weights_jÚsubZ coef_selfZparam_jÚ
start_timer r r r ; s2
zFedAMP.send_modelsc C s t | | j ¡| j S )N)ÚmathÚexpr )r Úxr r r r@ _ s zFedAMP.e)Ú__name__Ú
__module__Ú __qualname__r r r r@ Ú
__classcell__r r r r r s
"$r )r7 r1 rC ÚnumpyrF rS Zflcore.clients.clientampr r Úflcore.servers.serverbaser Ú threadingr r r r r r Ú<module> s | 2,831 | Python | .py | 30 | 93.2 | 387 | 0.467523 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,810 | serverrodpt.cpython-38.pyc | hkgdifyu_pFedPT/system/flcore/servers/__pycache__/serverrodpt.cpython-38.pyc | U
Õ şcÒ ã @ st d dl mZ d dlmZ d dlmZ d dlmZ d dlZd dl Z d dl
Z
d dl Z d dl Z d dl
ZG dd„ deƒZdS )é )Ú clientRODPT)ÚServer)Úread_client_data)ÚThreadNc sN e Zd Z‡ fdd„Zdd„ Zdd„ Zdd„ Zd d
„ Zdd d
„Zdd„ Z ‡ Z
S )ÚFedRODPTc sf t ƒ ||¡ || _| ¡ | |t¡ g | _t |j ¡| _
g | _ t d| j
› d| j› �ƒ t dƒ d S )Nz
Join ratio / total clients: z / z%Finished creating server and clients.)ÚsuperÚ__init__ÚargsÚset_slow_clientsÚ set_clientsr Úclients_divergeÚcopyÚdeepcopyÚmodelÚ global_modelÚdiff_proÚprintÚ
join_ratioÚ num_clients)Úselfr Útimes©Ú __class__© õc D:\京东\promot\第二次投稿\å®�验\æœ�务器\native - pro\system\flcore\servers\serverrodpt.pyr
s zFedRODPT.__init__c
C sš g }t | jd ƒD �]>}| ¡ | _| ¡ || j dkrXtd|› d�ƒ tdƒ | ¡ d}| jD ]}| ¡ }|| ¡ }qbtd
|¡ƒ | j |¡ d}t
| jd jj ¡ | jd jj ¡ ƒD ]:\}}|| } t | dk| t | ¡| ¡} |t | ¡ }qÀtd
| ¡ ¡ƒ | j | ¡ ¡ || j dk�rBtdƒ | j|d � | ¡ | ¡ qtd
ƒ tt| jƒƒ td ƒ tt|ƒƒ | ¡ | ¡ | ¡ d S ) Né r z
-------------Round number: z
-------------z
Evaluate global modelz"Averaged prompr difference: {:.4f}z"0 and 1 clients difference: {:.4f}z
Evaluate local model)Úaccz
Best global accuracy.z
Best local accuracy.) ÚrangeÚ
global_roundsÚselect_clientsÚselected_clientsÚ send_modelsÚeval_gapr ÚevaluateÚtrainÚitemÚformatr ÚappendÚzipÚclientsr Ú generatorÚ
parametersÚtorchÚwhereÚ
zeros_likeÚsumr Úreceive_modelsÚaggregate_parametersÚmaxÚ rs_test_accÚ save_resultsÚsave_global_modelÚsave_client_model)
r Ú local_accÚiÚ
temp_diff_proÚclientÚtemp_diff_pro_clientÚdiverge_clentsÚ new_paramÚ old_paramr r r r r$ sF
ÿ
zFedRODPT.trainc C s‚ t | jƒdkst‚d}| jD ]}||j7 }q g | _g | _g | _| jD ]8}| j |j| ¡ | j |j¡ | j t
|j ¡¡ qDd S )Nr ) Úlenr ÚAssertionErrorÚ
train_samplesÚuploaded_weightsÚ uploaded_idsÚuploaded_modelsr' Úidr
r r )r Úactive_train_samplesr: r r r r0 I s
zFedRODPT.receive_modelsc C s t j d| jd| jjd | jj d | jj d | jj d | jj d | jj
¡}t j |¡sjt |¡ t
| jƒD ]†\}}t j || jd t|ƒ d t| jjƒ d t| jjƒ d t| jjƒ d t| jjƒ d t| jjƒ d ¡}t |j|¡ qtd S )NÚmodelsr: Ú*Ú_clientÚ_z.pt)ÚosÚpathÚjoinÚdatasetr Úarv1Úarv2Úarv3Úarv4Úarv5Úarv6ÚexistsÚmakedirsÚ enumerater) Ú algorithmÚstrÚ
num_promptr r Ú plocal_stepsr r, Úsaver )r Ú
model_pathÚc_idxÚcÚmodel_path_saver r r r6 W s T
pzFedRODPT.save_client_modelc C s€ | j d | j }d| jj d | jj d | jj d | jj d | jj d | jj d }t j
|¡srt |¡ t
| jƒ�r||d | j d t| jƒ d t| jjƒ d t| jjƒ d t| jjƒ d t| jjƒ d t| jjƒ }|d |¡ }td| ƒ t |d¡�f}|jd| jd � |jd
| j d � |jd | j d � |jd | j d � |jd
| j d � |jd| j d � W 5 Q R X d S )NrJ z ../results/rH ú/z{}.h5z File path: Úwr3 )ÚdataÚrs_test_acc_stdÚ rs_test_aucÚ
rs_train_lossr r )!rN rX r rO rP rQ rR rS rT rK rL rU rV r? r3 ÚgoalrY r rZ r r r[ r r& r Úh5pyÚFileÚcreate_datasetrd re rf r r )r ÚalgoÚ result_pathÚ file_pathÚhfr r r r4 _ s L
l zFedRODPT.save_resultsNc C s| | ¡ }| ¡ }t|d ƒd t|d ƒ }t|d ƒd t|d ƒ }t|d ƒd t|d ƒ }t|d ƒd t|d ƒ }dd„ t|d |d ƒD ƒ} dd„ t|d |d ƒD ƒ}
|d krÎ| j |¡ n
| |¡ | j |¡ | j t | ¡¡ |d k�r| j
|¡ n
| |¡ t d |¡ƒ t d
|¡ƒ t d |¡ƒ t d |¡ƒ t d
t | ¡¡ƒ t d t |
¡¡ƒ d S )Né ç ğ?r é é c S s g | ]\}}|| ‘qS r r ©Ú.0ÚaÚnr r r Ú
<listcomp>{ s z%FedRODPT.evaluate.<locals>.<listcomp>c S s g | ]\}}|| ‘qS r r rs r r r rw | s zAveraged Train Loss: {:.4f}z Averaged Test Accurancy: {:.4f}z$Averaged Test oral Accurancy: {:.4f}zAveraged Test AUC: {:.4f}zStd Test Accurancy: {:.4f}zStd Test AUC: {:.4f})
Ú test_metricsÚ
train_metricsr/ r( r3 r' re rd ÚnpÚstdrf r r& ) r r ÚlossÚstatsÚ stats_trainÚtest_accÚ test_acc2Útest_aucÚ
train_lossÚaccsÚaucsr r r r# r s,
zFedRODPT.evaluatec C s~ g }g }g }g }| j D ]H}| ¡ \}}}} | |d ¡ | |d ¡ | | | ¡ | |¡ qdd„ | j D ƒ}
|
||||fS )Nrp c S s g | ]
}|j ‘qS r )rE )rt r_ r r r rw � s z)FedRODPT.test_metrics.<locals>.<listcomp>)r) rx r' ) r Ú num_samplesÚ tot_correctÚ tot_correct2Útot_aucr_ ÚctÚct2ÚnsÚaucÚidsr r r rx ‘ s
zFedRODPT.test_metrics)NN) Ú__name__Ú
__module__Ú __qualname__r r$ r0 r6 r4 r# rx Ú
__classcell__r r r r r s -
r )Zflcore.clients.clientrodptr Úflcore.servers.serverbaser Úutils.data_utilsr Ú threadingr Útimerh r
rK r, Únumpyrz r r r r r Ú<module> s | 5,946 | Python | .py | 51 | 115.333333 | 496 | 0.433005 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,811 | serverpfedpt.cpython-39.pyc | hkgdifyu_pFedPT/system/flcore/servers/__pycache__/serverpfedpt.cpython-39.pyc | a
@=bc_ ã @ sX d dl mZ d dlmZ d dlmZ d dlZd dlZd dlZd dl Z G dd„ deƒZ
dS )é )ÚclientPT)ÚServer)ÚThreadNc s4 e Zd Z‡ fdd„Zdd„ Zdd„ Zdd„ Z‡ ZS ) ÚPFedPTc s\ t ƒ ||¡ | ¡ | |t¡ t |jj¡| _ t
d| j › d| j › �ƒ t
dƒ g | _
d S )Nz
Join ratio / total clients: z / z%Finished creating server and clients.)ÚsuperÚ__init__Úset_slow_clientsÚ set_clientsr ÚcopyÚdeepcopyÚmodelÚbaseÚ global_modelÚprintÚ
join_ratioÚ num_clientsÚBudget)ÚselfÚargsÚtimes©Ú __class__© úu/media/sim812/391e55df-b6f2-4fe9-a920-53434a8506fa/lgh/pdept/PFL-Non-IID-master/system/flcore/servers/serverpfedpt.pyr s zPFedPT.__init__c C s
t | jd ƒD ]˜}t ¡ }| ¡ | _| ¡ || j dkrZtd|› d�ƒ tdƒ | ¡ | jD ] }| ¡ q`|
¡ | ¡ | j
t ¡ | ¡ tddd| j d ƒ qtd ƒ tt| jƒƒ td
ƒ tt| j dd … ƒt| j dd … ƒ ƒ | ¡ | ¡ | ¡ d S ) Né r z
-------------Round number: z
-------------z
Evaluate global modelz-------------------------z time costéÿÿÿÿz
Best global accuracy.z
Average time cost per round.)ÚrangeÚ
global_roundsÚtimeÚselect_clientsÚselected_clientsÚ send_modelsÚeval_gapr ÚevaluateÚtrainÚreceive_modelsÚaggregate_parametersr ÚappendÚmaxÚ rs_test_accÚsumÚlenÚ save_resultsÚsave_global_modelÚsave_client_model)r ÚiÚs_tÚclientr r r r$ s*
(z PFedPT.trainc C s„ t | jƒdksJ ‚d}| jD ]}||j7 }q g | _g | _g | _| jD ]:}| j |j| ¡ | j |j¡ | j t |j
j ¡¡ qDd S )Nr ) r+ r Ú
train_samplesÚuploaded_weightsÚ uploaded_idsÚuploaded_modelsr' Úidr
r r r
)r Úactive_train_samplesr1 r r r r% @ s
zPFedPT.receive_modelsc C sn t j d| jd¡}t j |¡s(t |¡ t| jƒD ]6\}}t j || jd t |ƒ d ¡}t
|j |¡ q2d S )NÚmodelsr1 Ú_serverz.pt)
ÚosÚpathÚjoinÚdatasetÚexistsÚmakedirsÚ enumerateÚclientsÚ algorithmÚstrÚtorchÚsaver )r Ú
model_pathÚc_idxÚcZmodel_path_saver r r r. N s
zPFedPT.save_client_model)Ú__name__Ú
__module__Ú __qualname__r r$ r% r. Ú
__classcell__r r r r r
s &r ) Z system.flcore.clients.clientptr Ú system.flcore.servers.serverbaser Ú threadingr r r
rD r: r r r r r Ú<module> s | 2,685 | Python | .py | 27 | 98.185185 | 326 | 0.478751 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,812 | servermoon.cpython-37.pyc | hkgdifyu_pFedPT/system/flcore/servers/__pycache__/servermoon.cpython-37.pyc | B
¿:cc{ ã @ sL d dl mZ d dlmZ d dlmZ d dlmZ d dlZG dd„ deƒZ dS )é )Ú
clientMOON)ÚServer)Úread_client_data)ÚThreadNc s$ e Zd Z‡ fdd„Zdd„ Z‡ ZS )ÚMOONc sL t ƒ ||¡ | ¡ | |t¡ td| j› d| j› �ƒ tdƒ g | _d S )Nz
Join ratio / total clients: z / z%Finished creating server and clients.) ÚsuperÚ__init__Úset_slow_clientsÚ set_clientsr ÚprintÚ
join_ratioÚ num_clientsÚBudget)ÚselfÚargsÚtimes)Ú __class__© úG/root/autodl-tmp/PFL-Non-IID-master/system/flcore/servers/servermoon.pyr s z
MOON.__init__c C s\ g }d| _ d}xŞ| j sìt ¡ }| ¡ | _| ¡ || j dkr^td|› d�ƒ tdƒ | ¡ x| jD ] }| ¡ qfW || j dkr˜tdƒ | j|d� | ¡ |
¡ | j t ¡ | ¡ td| j d ƒ | j
| jg| jd
�| _ |d 7 }qW td ƒ tt| jƒƒ td
ƒ tt|ƒƒ tdƒ tt| j d d … ƒt| j d d … ƒ ƒ | ¡ | ¡ d S )NFr z
-------------Round number: z
-------------z
Evaluate global modelz
Evaluate local model)Úaccz2--------------------------------------------------éÿÿÿÿ)Úacc_lssÚtop_cnté z
Best global accuracy.z
Best local accuracy.z
Averaged time per iteration.)ÚdoneÚtimeÚselect_clientsÚselected_clientsÚ send_modelsÚeval_gapr ÚevaluateÚtrainÚreceive_modelsÚaggregate_parametersr ÚappendÚ
check_doneÚ rs_test_accr ÚmaxÚsumÚlenÚ save_resultsÚsave_global_model)r Ú local_accÚiÚs_tÚclientr r r r! s<
(z
MOON.train)Ú__name__Ú
__module__Ú __qualname__r r! Ú
__classcell__r r )r r r s r )
Zflcore.clients.clientmoonr Úflcore.servers.serverbaser Úutils.data_utilsr Ú threadingr r r r r r r Ú<module> s
| 1,946 | Python | .py | 19 | 101.157895 | 370 | 0.475622 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,813 | serverapfl.cpython-37.pyc | hkgdifyu_pFedPT/system/flcore/servers/__pycache__/serverapfl.cpython-37.pyc | B
¸:cc® ã @ s8 d dl mZ d dlmZ d dlmZ G dd„ deƒZdS )é )Ú
clientAPFL)ÚServer)ÚThreadc s$ e Zd Z‡ fdd„Zdd„ Z‡ ZS )ÚAPFLc sF t ƒ ||¡ | ¡ | |t¡ td| j› d| j› �ƒ tdƒ d S )Nz
Join ratio / total clients: z / z%Finished creating server and clients.)ÚsuperÚ__init__Úset_slow_clientsÚ set_clientsr ÚprintÚ
join_ratioÚ num_clients)ÚselfÚargsÚtimes)Ú __class__© úG/root/autodl-tmp/PFL-Non-IID-master/system/flcore/servers/serverapfl.pyr s
z
APFL.__init__c C sª x~t | jd ƒD ]l}| ¡ | _| ¡ || j dkrTtd|› d�ƒ tdƒ | ¡ x| jD ] }| ¡ q\W | ¡ |
¡ qW tdƒ tt | j ƒƒ |
¡ | ¡ d S )Né r z
-------------Round number: z
-------------z
Evaluate global modelz
Best global accuracy.)ÚrangeÚ
global_roundsÚselect_clientsÚselected_clientsÚ send_modelsÚeval_gapr
ÚevaluateÚtrainÚreceive_modelsÚaggregate_parametersÚmaxÚ rs_test_accÚ save_resultsÚsave_global_model)r
ÚiÚclientr r r r s
z
APFL.train)Ú__name__Ú
__module__Ú __qualname__r r Ú
__classcell__r r )r r r s
r N)Zflcore.clients.clientapflr Úflcore.servers.serverbaser Ú threadingr r r r r r Ú<module> s | 1,442 | Python | .py | 14 | 101.714286 | 227 | 0.492652 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,814 | serverbabu.cpython-37.pyc | hkgdifyu_pFedPT/system/flcore/servers/__pycache__/serverbabu.cpython-37.pyc | B
¸:cc• ã @ s8 d dl mZ d dlmZ d dlmZ G dd„ deƒZdS )é )Ú
clientBABU)ÚServer)ÚThreadc s, e Zd Z‡ fdd„Zdd„ Zdd„ Z‡ ZS )ÚFedBABUc sF t ƒ ||¡ | ¡ | |t¡ td| j› d| j› �ƒ tdƒ d S )Nz
Join ratio / total clients: z / z%Finished creating server and clients.)ÚsuperÚ__init__Úset_slow_clientsÚ set_clientsr ÚprintÚ
join_ratioÚ num_clients)ÚselfÚargsÚtimes)Ú __class__© úG/root/autodl-tmp/PFL-Non-IID-master/system/flcore/servers/serverbabu.pyr s
zFedBABU.__init__c C sÒ x~t | jd ƒD ]l}| ¡ | _| ¡ || j dkrTtd|› d�ƒ tdƒ | ¡ x| jD ] }| ¡ q\W | ¡ |
¡ qW tdƒ tt | j ƒƒ x| j
D ] }| ¡ q�W tdƒ | ¡ | ¡ | ¡ d S )Né r z
-------------Round number: z
-------------z
Evaluate global modelz
Best global accuracy.z4
-------------Evaluate fine-tuned model-------------)ÚrangeÚ
global_roundsÚselect_clientsÚselected_clientsÚ send_modelsÚeval_gapr
ÚevaluateÚtrainÚreceive_modelsÚaggregate_parametersÚmaxÚ rs_test_accÚclientsZ fine_tuneÚ save_resultsÚsave_global_model)r
ÚiÚclientr r r r s&
z
FedBABU.trainc C s˜ t | jƒdkst‚g | _d}g | _g | _xD| jD ]:}| j |j¡ ||j7 }| j |j¡ | j |j j
¡ q0W x$t | jƒD ]\}}|| | j|< qzW d S )Nr ) Úlenr ÚAssertionErrorÚuploaded_weightsÚ uploaded_idsÚuploaded_modelsÚappendÚ
train_samplesÚidÚmodelÚbaseÚ enumerate)r
Ú tot_samplesr$ r# Úwr r r r 7 s
zFedBABU.receive_models)Ú__name__Ú
__module__Ú __qualname__r r r Ú
__classcell__r r )r r r s
#r N)Zflcore.clients.clientbabur Úflcore.servers.serverbaser Ú threadingr r r r r r Ú<module> s | 2,012 | Python | .py | 16 | 124.5 | 264 | 0.48022 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,815 | serverdyn.cpython-39.pyc | hkgdifyu_pFedPT/system/flcore/servers/__pycache__/serverdyn.cpython-39.pyc | a
»¢bc ã @ sP d dl Z d dlZd dlmZ d dlmZ d dlmZ d dlZG dd„ deƒZ dS )é N)Ú clientDyn)ÚServer)ÚThreadc s< e Zd Z‡ fdd„Zdd„ Zdd„ Zdd„ Zd d
„ Z‡ ZS ) ÚFedDync s€ t ƒ ||¡ | ¡ | |t¡ td| j› d| j› �ƒ tdƒ g | _|j | _ t
|j ¡| _
| j
¡ D ]}t |j¡|_qhd S )Nz
Join ratio / total clients: z / z%Finished creating server and clients.)ÚsuperÚ__init__Úset_slow_clientsÚ set_clientsr ÚprintÚ
join_ratioÚ num_clientsÚBudgetÚalphaÚcopyÚdeepcopyÚmodelÚ server_stateÚ
parametersÚtorchÚ
zeros_likeÚdata)ÚselfÚargsÚtimesÚparam©Ú __class__© úr/media/sim812/391e55df-b6f2-4fe9-a920-53434a8506fa/lgh/pdept/PFL-Non-IID-master/system/flcore/servers/serverdyn.pyr
s zFedDyn.__init__c C sh g }d| _ d}t| jd ƒD ]Ü}t ¡ }| ¡ | _| ¡ || j dkrhtd|› d�ƒ tdƒ | ¡ | jD ] }|
¡ qn|| j dkržtdƒ | j |d� | ¡ | ¡ |
¡ | j t ¡ | ¡ td | jd
ƒ | j| jg| jd �| _ |d7 }q td ƒ tt| jƒƒ td
ƒ tt|ƒƒ tdƒ tt| jdd … ƒt| jdd … ƒ ƒ | ¡ | ¡ d S )NFr é z
-------------Round number: z
-------------z
Evaluate global modelz
Evaluate local model)Úaccz2--------------------------------------------------éÿÿÿÿ)Úacc_lssÚtop_cntz
Best global accuracy.z
Best local accuracy.z
Averaged time per iteration.)ÚdoneÚrangeÚ
global_roundsÚtimeÚselect_clientsÚselected_clientsÚ send_modelsÚeval_gapr
ÚevaluateÚtrainÚreceive_modelsÚupdate_server_stateÚaggregate_parametersr
ÚappendÚ
check_doneÚ rs_test_accr# ÚmaxÚsumÚlenÚ save_resultsÚsave_global_model)r Z local_accÚiÚs_tÚclientr r r r- s>
(z FedDyn.trainc C s< t | j ¡ | ¡ ƒD ]"\}}| j|j ¡ | j 7 _qd S )N)ÚzipÚ global_modelr r ÚcloneÚ join_clients)r Ú client_modelÚ server_paramÚ client_paramr r r Úadd_parametersO s zFedDyn.add_parametersc C s” t | jƒdksJ ‚t | jd ¡| _| j ¡ D ]}t |j¡|_q.| jD ]}| |¡ qHt
| j ¡ | j ¡ ƒD ] \}}| jd| j | 8 _qnd S )Nr r )
r6 Úuploaded_modelsr r r= r r r r rC r< r r )r r r@ rA Ú state_paramr r r r0 S s
zFedDyn.aggregate_parametersc C s¾ t | jƒdksJ ‚t | jd ¡}| ¡ D ]}t |j¡|_q*| jD ]B}t| j ¡ | ¡ | ¡ ƒD ]"\}}}| j|| | j
7 _qbqDt| j ¡ | ¡ ƒD ] \}}| j| j | 8 _qœd S )Nr )
r6 rD r r r r r r r< r= r r r )r Z model_deltar r@ rA rB Z delta_paramrE r r r r/ ` s
$ zFedDyn.update_server_state) Ú__name__Ú
__module__Ú __qualname__r r- rC r0 r/ Ú
__classcell__r r r r r s
1
r )
r r Z system.flcore.clients.clientdynr Ú system.flcore.servers.serverbaser Ú threadingr r' r r r r r Ú<module> s | 3,312 | Python | .py | 30 | 109.166667 | 409 | 0.439842 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,816 | servermoonpt.cpython-38.pyc | hkgdifyu_pFedPT/system/flcore/servers/__pycache__/servermoonpt.cpython-38.pyc | U
şcÿ ã @ st d dl mZ d dlmZ d dlmZ d dlmZ d dlZd dl Z d dl
Z
d dl Z d dl Z d dl
ZG dd„ deƒZdS )é )Ú clientMOONPT)ÚServer)Úread_client_data)ÚThreadNc sN e Zd Z‡ fdd„Zdd„ Zdd„ Zdd„ Zd d
„ Zdd d
„Zdd„ Z ‡ Z
S )ÚMOONPTc sl t ƒ ||¡ || _| ¡ | |t¡ g | _g | _td| j › d| j
› �ƒ tdƒ t |j
¡| _g | _d S )Nz
Join ratio / total clients: z / z%Finished creating server and clients.)ÚsuperÚ__init__ÚargsÚset_slow_clientsÚ set_clientsr Úclients_divergeÚdiff_proÚprintÚ
join_ratioÚ num_clientsÚcopyÚdeepcopyÚmodelÚ global_modelÚBudget)Úselfr Útimes©Ú __class__© õd D:\京东\promot\第二次投稿\å®�验\æœ�务器\native - pro\system\flcore\servers\servermoonpt.pyr
s zMOONPT.__init__c C s g }d| _ d}t| jd ƒD �]ˆ}t ¡ }| ¡ | _| ¡ || j dkrjtd|› d�ƒ tdƒ | ¡ d}| jD ]}|
¡ }|| ¡ }qttd |¡ƒ | j
|¡ d}t| jd jj ¡ | jd jj ¡ ƒD ]:\}} || }
t |
dk|
t |
¡|
¡}
|t |
¡ }qÒtd | ¡ ¡ƒ | j | ¡ ¡ || j dk�rTtd ƒ | j |d
� | ¡ | ¡ | j t ¡ | ¡ td | jd ƒ | j | j g| j d
�| _ |d7 }q tdƒ tt | j ƒƒ tdƒ tt |ƒƒ tdƒ tt| jdd … ƒt | jdd … ƒ ƒ | !¡ | "¡ | #¡ d S )NFr é z
-------------Round number: z
-------------z
Evaluate global modelz"Averaged prompr difference: {:.4f}z"0 and 1 clients difference: {:.4f}z
Evaluate local model)Úaccz2--------------------------------------------------éÿÿÿÿ)Úacc_lssÚtop_cntz
Best global accuracy.z
Best local accuracy.z
Averaged time per iteration.)$ÚdoneÚrangeÚ
global_roundsÚtimeÚselect_clientsÚselected_clientsÚ send_modelsÚeval_gapr ÚevaluateÚtrainÚitemÚformatr
ÚappendÚzipÚclientsr Ú generatorÚ
parametersÚtorchÚwhereÚ
zeros_likeÚsumr Úreceive_modelsÚaggregate_parametersr Ú
check_doneÚ rs_test_accr ÚmaxÚlenÚ save_resultsÚsave_global_modelÚsave_client_model) r Ú local_accÚiÚs_tÚ
temp_diff_proÚclientÚtemp_diff_pro_clientÚdiverge_clentsÚ new_paramÚ old_paramr
r r r r* sX
ÿ
(z MOONPT.trainc C s‚ t | jƒdkst‚d}| jD ]}||j7 }q g | _g | _g | _| jD ]8}| j |j| ¡ | j |j¡ | j t
|j ¡¡ qDd S )Nr ) r; r& ÚAssertionErrorÚ
train_samplesÚuploaded_weightsÚ uploaded_idsÚuploaded_modelsr- Úidr r r )r Úactive_train_samplesrC r r r r6 Y s
zMOONPT.receive_modelsc C s t j d| jd| jjd | jj d | jj d | jj d | jj d | jj
¡}t j |¡sjt |¡ t
| jƒD ]†\}}t j || jd t|ƒ d t| jjƒ d t| jjƒ d t| jjƒ d t| jjƒ d t| jjƒ d ¡}t |j|¡ qtd S )NÚmodelsrC Ú*Ú_clientÚ_z.pt)ÚosÚpathÚjoinÚdatasetr Úarv1Úarv2Úarv3Úarv4Úarv5Úarv6ÚexistsÚmakedirsÚ enumerater/ Ú algorithmÚstrÚ
num_promptr r Ú plocal_stepsr# r2 Úsaver )r Ú
model_pathÚc_idxÚcÚmodel_path_saver r r r> h s< T
ÿÿ
ÿÿÿÿş
şşÿşızMOONPT.save_client_modelc C s€ | j d | j }d| jj d | jj d | jj d | jj d | jj d | jj d }t j
|¡srt |¡ t
| jƒ�r||d | j d t| jƒ d t| jjƒ d t| jjƒ d t| jjƒ d t| jjƒ d t| jjƒ }|d |¡ }td| ƒ t |d¡�f}|jd| jd � |jd
| j d � |jd | j d � |jd | j d � |jd
| j d � |jd| j d � W 5 Q R X d S )NrR z ../results/rP ú/z{}.h5z File path: Úwr9 )ÚdataÚrs_test_acc_stdÚ rs_test_aucÚ
rs_train_lossr
r )!rV r` r rW rX rY rZ r[ r\ rS rT r] r^ r; r9 Úgoalra r rb r r rc r# r, r Úh5pyÚFileÚcreate_datasetrl rm rn r
r )r ÚalgoÚ result_pathÚ file_pathÚhfr r r r< t s> L
0ÿÿ
ÿÿÿÿş
ş zMOONPT.save_resultsNc C s| | ¡ }| ¡ }t|d ƒd t|d ƒ }t|d ƒd t|d ƒ }t|d ƒd t|d ƒ }t|d ƒd t|d ƒ }dd„ t|d |d ƒD ƒ} dd„ t|d |d ƒD ƒ}
|d krÎ| j |¡ n
| |¡ | j |¡ | j t | ¡¡ |d k�r| j
|¡ n
| |¡ t d |¡ƒ t d
|¡ƒ t d |¡ƒ t d |¡ƒ t d
t | ¡¡ƒ t d t |
¡¡ƒ d S )Né ç ğ?r é é c S s g | ]\}}|| ‘qS r r ©Ú.0ÚaÚnr r r Ú
<listcomp>’ s z#MOONPT.evaluate.<locals>.<listcomp>c S s g | ]\}}|| ‘qS r r r{ r r r r “ s zAveraged Train Loss: {:.4f}z Averaged Test Accurancy: {:.4f}z$Averaged Test oral Accurancy: {:.4f}zAveraged Test AUC: {:.4f}zStd Test Accurancy: {:.4f}zStd Test AUC: {:.4f})
Ú test_metricsÚ
train_metricsr5 r. r9 r- rm rl ÚnpÚstdrn r r, ) r r ÚlossÚstatsÚ stats_trainÚtest_accÚ test_acc2Útest_aucÚ
train_lossÚaccsÚaucsr r r r) ‰ s,
zMOONPT.evaluatec C s~ g }g }g }g }| j D ]H}| ¡ \}}}} | |d ¡ | |d ¡ | | | ¡ | |¡ qdd„ | j D ƒ}
|
||||fS )Nrx c S s g | ]
}|j ‘qS r )rM )r| rg r r r r µ s z'MOONPT.test_metrics.<locals>.<listcomp>)r/ r€ r- ) r Ú num_samplesÚ tot_correctÚ tot_correct2Útot_aucrg ÚctÚct2ÚnsÚaucÚidsr r r r€ ¨ s
zMOONPT.test_metrics)NN) Ú__name__Ú
__module__Ú __qualname__r r* r6 r> r< r) r€ Ú
__classcell__r r r r r s =
r )Zflcore.clients.clientmoonptr Úflcore.servers.serverbaser Úutils.data_utilsr Ú threadingr r$ rp r rS r2 Únumpyr‚ r r r r r Ú<module> s | 6,327 | Python | .py | 62 | 100.822581 | 494 | 0.423715 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,817 | serverper.cpython-39.pyc | hkgdifyu_pFedPT/system/flcore/servers/__pycache__/serverper.cpython-39.pyc | a
f¾`cî ã @ s8 d dl mZ d dlmZ d dlmZ G dd„ deƒZdS )é )Ú clientPer)ÚServer)ÚThreadc s, e Zd Z‡ fdd„Zdd„ Zdd„ Z‡ ZS )ÚFedPerc sF t ƒ ||¡ | ¡ | |t¡ td| j› d| j› �ƒ tdƒ d S )Nz
Join ratio / total clients: z / z%Finished creating server and clients.)ÚsuperÚ__init__Úset_slow_clientsÚ set_clientsr ÚprintÚ
join_ratioÚ num_clients)ÚselfÚargsÚtimes©Ú __class__© úr/media/sim812/391e55df-b6f2-4fe9-a920-53434a8506fa/lgh/pdept/PFL-Non-IID-master/system/flcore/servers/serverper.pyr s
zFedPer.__init__c C s¢ t | jd ƒD ]h}| ¡ | _| ¡ || j dkrRtd|› d�ƒ tdƒ | ¡ | jD ] }| ¡ qX| ¡ |
¡ qtdƒ tt | j ƒƒ |
¡ | ¡ d S )Né r z
-------------Round number: z
-------------z
Evaluate global modelz
Best global accuracy.)ÚrangeÚ
global_roundsÚselect_clientsÚselected_clientsÚ send_modelsÚeval_gapr
ÚevaluateÚtrainÚreceive_modelsÚaggregate_parametersÚmaxÚ rs_test_accÚ save_resultsÚsave_global_model)r
ÚiÚclientr r r r s
z FedPer.trainc C s� t | jƒdksJ ‚g | _d}g | _g | _| jD ]:}| j |j¡ ||j7 }| j |j¡ | j |jj ¡ q.t
| jƒD ]\}}|| | j|< qtd S )Nr ) Úlenr Úuploaded_weightsÚ uploaded_idsÚuploaded_modelsÚappendÚ
train_samplesÚidÚmodelÚbaseÚ enumerate)r
Ú tot_samplesr$ r# Úwr r r r 2 s
zFedPer.receive_models)Ú__name__Ú
__module__Ú __qualname__r r r Ú
__classcell__r r r r r s
r N)Zflcore.clients.clientperr Úflcore.servers.serverbaser Ú threadingr r r r r r Ú<module> s | 1,913 | Python | .py | 18 | 105.055556 | 327 | 0.484177 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,818 | serverapfl.cpython-38.pyc | hkgdifyu_pFedPT/system/flcore/servers/__pycache__/serverapfl.cpython-38.pyc | U
”jfc® ã @ s8 d dl mZ d dlmZ d dlmZ G dd„ deƒZdS )é )Ú
clientAPFL)ÚServer)ÚThreadc s$ e Zd Z‡ fdd„Zdd„ Z‡ ZS )ÚAPFLc sF t ƒ ||¡ | ¡ | |t¡ td| j› d| j› �ƒ tdƒ d S )Nz
Join ratio / total clients: z / z%Finished creating server and clients.)ÚsuperÚ__init__Úset_slow_clientsÚ set_clientsr ÚprintÚ
join_ratioÚ num_clients)ÚselfÚargsÚtimes©Ú __class__© õL D:\京东\promot\cifar\cifar\Cifar10_iid\system\flcore\servers\serverapfl.pyr s
z
APFL.__init__c C s¢ t | jd ƒD ]h}| ¡ | _| ¡ || j dkrRtd|› d�ƒ tdƒ | ¡ | jD ] }| ¡ qX| ¡ |
¡ qtdƒ tt | j ƒƒ |
¡ | ¡ d S )Né r z
-------------Round number: z
-------------z
Evaluate global modelz
Best global accuracy.)ÚrangeÚ
global_roundsÚselect_clientsÚselected_clientsÚ send_modelsÚeval_gapr
ÚevaluateÚtrainÚreceive_modelsÚaggregate_parametersÚmaxÚ rs_test_accÚ save_resultsÚsave_global_model)r
ÚiÚclientr r r r s
z
APFL.train)Ú__name__Ú
__module__Ú __qualname__r r Ú
__classcell__r r r r r s
r N)Zflcore.clients.clientapflr Úflcore.servers.serverbaser Ú threadingr r r r r r Ú<module> s | 1,456 | Python | .py | 17 | 84.411765 | 225 | 0.488889 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,819 | serveravg.cpython-38.pyc | hkgdifyu_pFedPT/system/flcore/servers/__pycache__/serveravg.cpython-38.pyc | U
[Ğıc· ã @ sH d dl Z d dlZd dlmZ d dlmZ d dlmZ G dd„ deƒZdS )é N)Ú clientAVG)ÚServer)ÚThreadc s$ e Zd Z‡ fdd„Zdd„ Z‡ ZS )ÚFedAvgc sL t ƒ ||¡ | ¡ | |t¡ td| j› d| j› �ƒ tdƒ g | _d S )Nz
Join ratio / total clients: z / z%Finished creating server and clients.) ÚsuperÚ__init__Zset_slow_clientsZ set_clientsr ÚprintÚ
join_ratioÚ num_clientsÚBudget)ÚselfÚargsÚtimes©Ú __class__© õW D:\京东\promot\第二次投稿\å®�验\native - pro\system\flcore\servers\serveravg.pyr s zFedAvg.__init__c C s< g }t | jd ƒD ]º}t ¡ }| ¡ | _| ¡ || j dkr^td|› d�ƒ tdƒ | ¡ | jD ] }| ¡ qd|| j dkr”tdƒ | j|d� |
¡ | ¡ | j
t ¡ | ¡ tdd d| j d
ƒ qtd ƒ tt| jƒƒ td ƒ tt|ƒƒ td
ƒ tt| j dd … ƒt| j dd … ƒ ƒ | ¡ | ¡ d S )Né r z
-------------Round number: z
-------------z
Evaluate global modelz
Evaluate local model)Úaccz-------------------------z time costéÿÿÿÿz
Best global accuracy.z
Best local accuracy.z
Average time cost per round.)ÚrangeÚ
global_roundsÚtimeZselect_clientsZselected_clientsZ send_modelsZeval_gapr ZevaluateÚtrainZreceive_modelsZaggregate_parametersr ÚappendÚmaxZ rs_test_accÚsumÚlenZ save_resultsZsave_global_model)r Z local_accÚiZs_tÚclientr r r r s4
(z FedAvg.train)Ú__name__Ú
__module__Ú __qualname__r r Ú
__classcell__r r r r r s r ) r ÚtorchZflcore.clients.clientavgr Zflcore.servers.serverbaser Ú threadingr r r r r r Ú<module> s
| 1,867 | Python | .py | 17 | 108.529412 | 383 | 0.47866 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,820 | serverditto.cpython-37.pyc | hkgdifyu_pFedPT/system/flcore/servers/__pycache__/serverditto.cpython-37.pyc | B
¾:ccü ã @ sH d dl Z d dlmZ d dlmZ d dlmZ d dlZG dd„ deƒZdS )é N)Ú clientDitto)ÚServer)ÚThreadc s$ e Zd Z‡ fdd„Zdd„ Z‡ ZS )ÚDittoc sL t ƒ ||¡ | ¡ | |t¡ td| j› d| j› �ƒ tdƒ g | _d S )Nz
Join ratio / total clients: z / z%Finished creating server and clients.) ÚsuperÚ__init__Úset_slow_clientsÚ set_clientsr ÚprintÚ
join_ratioÚ num_clientsÚBudget)ÚselfÚargsÚtimes)Ú __class__© úH/root/autodl-tmp/PFL-Non-IID-master/system/flcore/servers/serverditto.pyr s zDitto.__init__c C s x¶t | jd ƒD ]¤}t ¡ }| ¡ | _| ¡ || j dkr\td|› d�ƒ tdƒ | ¡ x | jD ]}| ¡ |
¡ qdW | ¡ | ¡ | j
t ¡ | ¡ tddd| j
d ƒ qW td ƒ tt| jƒƒ td
ƒ tt| j
dd … ƒt| j
dd … ƒ ƒ | ¡ | ¡ d S ) Né r z
-------------Round number: z
-------------z
Evaluate global modelz-------------------------z time costéÿÿÿÿz
Best global accuracy.z
Average time cost per round.)ÚrangeÚ
global_roundsÚtimeÚselect_clientsÚselected_clientsÚ send_modelsÚeval_gapr
ÚevaluateZptrainÚtrainÚreceive_modelsÚaggregate_parametersr
ÚappendÚmaxÚ rs_test_accÚsumÚlenÚ save_resultsÚsave_global_model)r ÚiÚs_tÚclientr r r r s*
(z Ditto.train)Ú__name__Ú
__module__Ú __qualname__r r Ú
__classcell__r r )r r r s r ) ÚcopyZflcore.clients.clientdittor Úflcore.servers.serverbaser Ú threadingr r r r r r r Ú<module> s
| 1,727 | Python | .py | 14 | 122 | 331 | 0.482497 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,821 | serverpFedMe.cpython-37.pyc | hkgdifyu_pFedPT/system/flcore/servers/__pycache__/serverpFedMe.cpython-37.pyc | B
¸:ccE ã @ sP d dl Z d dlZd dlZd dlmZ d dlmZ d dlmZ G dd„ deƒZ dS )é N)Ú clientpFedMe)ÚServer)ÚThreadc sL e Zd Z‡ fdd„Zdd„ Zdd„ Zdd„ Zd d
„ Zd d „ Zd
d„ Z ‡ Z
S )ÚpFedMec s` t ƒ ||¡ | ¡ | |t¡ |j| _g | _g | _g | _t d| j
› d| j › �ƒ t dƒ d S )Nz
Join ratio / total clients: z / z%Finished creating server and clients.) ÚsuperÚ__init__Úset_slow_clientsÚ set_clientsr ÚbetaÚrs_train_acc_perÚrs_train_loss_perÚrs_test_acc_perÚprintÚ
join_ratioÚ num_clients)ÚselfÚargsÚtimes)Ú __class__© úI/root/autodl-tmp/PFL-Non-IID-master/system/flcore/servers/serverpFedMe.pyr
s zpFedMe.__init__c C sÈ xœt | jd ƒD ]Š}| ¡ | _| ¡ x| jD ] }| ¡ q.W || j dkrltd|› d�ƒ tdƒ | ¡ t
t | j
¡ ƒ¡| _| ¡ | ¡ | ¡ qW tdƒ tt| jƒƒ | ¡ | ¡ d S )Né r z
-------------Round number: z
-------------z
Evaluate personalized modelz
Best personalized results.)ÚrangeÚ
global_roundsÚselect_clientsÚselected_clientsÚ send_modelsÚtrainÚeval_gapr Úevaluate_personalized_modelÚcopyÚdeepcopyÚlistÚ global_modelÚ
parametersÚprevious_global_modelÚreceive_modelsÚaggregate_parametersÚbeta_aggregate_parametersÚmaxr
Ú save_resultsÚsave_global_model)r ÚiÚclientr r r r s"
z pFedMe.trainc C sB x<t | j| j ¡ ƒD ]&\}}d| j |j | j|j |_qW d S )Nr )Úzipr% r# r$ r
Údata)r Z pre_paramÚparamr r r r( D s z pFedMe.beta_aggregate_parametersc C sV g }g }x2| j D ](}| ¡ \}}| |d ¡ | |¡ qW dd„ | j D ƒ}|||fS )Ng ğ?c S s g | ]
}|j ‘qS r )Úid)Ú.0Úcr r r ú
<listcomp>P s z4pFedMe.test_metrics_personalized.<locals>.<listcomp>)ÚclientsÚtest_metrics_personalizedÚappend)r Ú num_samplesÚ tot_correctr3 ÚctÚnsÚidsr r r r6 I s z pFedMe.test_metrics_personalizedc C sl g }g }g }xB| j D ]8}| ¡ \}}}| |d ¡ | |¡ | |d ¡ qW dd„ | j D ƒ}||||fS )Ng ğ?c S s g | ]
}|j ‘qS r )r1 )r2 r3 r r r r4 ^ s z5pFedMe.train_metrics_personalized.<locals>.<listcomp>)r5 Útrain_metrics_personalizedr7 ) r r8 r9 Úlossesr3 r: Úclr; r< r r r r= T s
z!pFedMe.train_metrics_personalizedc C sB | ¡ }t|d ƒd t|d ƒ }| j |¡ td |¡ƒ d S )Né g ğ?r z+Average Personalized Test Accurancy: {:.4f})r6 Úsumr
r7 r Úformat)r ÚstatsÚtest_accr r r r b s z"pFedMe.evaluate_personalized_modelc C s¦ | j d | j }d}tj |¡s*t |¡ t| jƒr¢|d | j d t | j
ƒ }t |d
|¡ d¡�6}|jd| jd� |jd| jd� |jd| jd� W d Q R X d S ) NÚ_z ../results/z{}.h5ÚwÚ rs_test_acc)r/ Z rs_train_accÚ
rs_train_loss)ÚdatasetÚ algorithmÚosÚpathÚexistsÚmakedirsÚlenr
ÚgoalÚstrr Úh5pyÚFilerB Úcreate_datasetr r )r ÚalgoÚ result_pathZalgo2Úhfr r r r* q s
zpFedMe.save_results) Ú__name__Ú
__module__Ú __qualname__r r r( r6 r= r r* Ú
__classcell__r r )r r r s + r )
rK r rR Zflcore.clients.clientpFedMer Zflcore.servers.serverbaser Ú threadingr r r r r r Ú<module> s | 3,763 | Python | .py | 23 | 162.26087 | 430 | 0.466167 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,822 | serveravg.cpython-39.pyc | hkgdifyu_pFedPT/system/flcore/servers/__pycache__/serveravg.cpython-39.pyc | a
Ú'bcç ã @ sH d dl Z d dlZd dlmZ d dlmZ d dlmZ G dd„ deƒZdS )é N)Ú clientAVG)ÚServer)ÚThreadc s$ e Zd Z‡ fdd„Zdd„ Z‡ ZS )ÚFedAvgc sL t ƒ ||¡ | ¡ | |t¡ td| j› d| j› �ƒ tdƒ g | _d S )Nz
Join ratio / total clients: z / z%Finished creating server and clients.) ÚsuperÚ__init__Zset_slow_clientsZ set_clientsr ÚprintÚ
join_ratioÚ num_clientsÚBudget)ÚselfÚargsÚtimes©Ú __class__© úr/media/sim812/391e55df-b6f2-4fe9-a920-53434a8506fa/lgh/pdept/PFL-Non-IID-master/system/flcore/servers/serveravg.pyr s zFedAvg.__init__c C s t | jd ƒD ]˜}t ¡ }| ¡ | _| ¡ || j dkrZtd|› d�ƒ tdƒ | ¡ | jD ] }| ¡ q`|
¡ | ¡ | j
t ¡ | ¡ tddd| j d ƒ qtd ƒ tt| jƒƒ td
ƒ tt| j dd … ƒt| j dd … ƒ ƒ | ¡ | ¡ d S ) Né r z
-------------Round number: z
-------------z
Evaluate global modelz-------------------------z time costéÿÿÿÿz
Best global accuracy.z
Average time cost per round.)ÚrangeÚ
global_roundsÚtimeZselect_clientsZselected_clientsZ send_modelsÚeval_gapr ZevaluateÚtrainZreceive_modelsZaggregate_parametersr ÚappendÚmaxZ rs_test_accÚsumÚlenZ save_resultsZsave_global_model)r ÚiZs_tÚclientr r r r s(
(z FedAvg.train)Ú__name__Ú
__module__Ú __qualname__r r Ú
__classcell__r r r r r s r ) r ÚtorchZ system.flcore.clients.clientavgr Z system.flcore.servers.serverbaser Ú threadingr r r r r r Ú<module> s
| 1,771 | Python | .py | 15 | 116.733333 | 369 | 0.491747 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,823 | serverphp.cpython-38.pyc | hkgdifyu_pFedPT/system/flcore/servers/__pycache__/serverphp.cpython-38.pyc | U
•ic‚ ã @ sH d dl mZ d dlmZ d dlmZ d dlZd dlZG dd„ deƒZdS )é )Ú clientPHP)ÚServer)ÚThreadNc s, e Zd Z‡ fdd„Zdd„ Zdd„ Z‡ ZS )ÚFedPHPc sL t ƒ ||¡ | ¡ | |t¡ td| j› d| j› �ƒ tdƒ g | _d S )Nz
Join ratio / total clients: z / z%Finished creating server and clients.) ÚsuperÚ__init__Úset_slow_clientsÚ set_clientsr ÚprintÚ
join_ratioÚ num_clientsÚBudget)ÚselfÚargsÚtimes©Ú __class__© õD D:\京东\promot\cifar\cifar\tiny\system\flcore\servers\serverphp.pyr s zFedPHP.__init__c C s¤ t | jd ƒD ]j}| ¡ | _| |¡ || j dkrTtd|› d�ƒ tdƒ | ¡ | jD ] }| ¡ qZ| ¡ |
¡ qtdƒ tt | j ƒƒ |
¡ | ¡ d S )Né r z
-------------Round number: z
-------------z
Evaluate global modelz
Best global accuracy.)ÚrangeÚ
global_roundsÚselect_clientsÚselected_clientsÚ send_modelsÚeval_gapr
ÚevaluateÚtrainÚreceive_modelsÚaggregate_parametersÚmaxÚ rs_test_accÚ save_resultsÚsave_global_model)r ÚiÚclientr r r r s
z FedPHP.trainc C s0 t | jƒdkst‚| jD ]}| | j|¡ qd S )Nr )ÚlenÚclientsÚAssertionErrorÚset_parametersÚ global_model)r ÚRr% r r r r 5 s
zFedPHP.send_models)Ú__name__Ú
__module__Ú __qualname__r r r Ú
__classcell__r r r r r s r ) Zflcore.clients.clientphpr Úflcore.servers.serverbaser Ú threadingr ÚtimeÚcopyr r r r r Ú<module> s
| 1,735 | Python | .py | 17 | 100.823529 | 350 | 0.482257 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,824 | serverbn.cpython-37.pyc | hkgdifyu_pFedPT/system/flcore/servers/__pycache__/serverbn.cpython-37.pyc | B
¿:ccã ã @ sL d dl mZ d dlmZ d dlmZ d dlmZ d dlZG dd„ deƒZ dS )é )ÚclientBN)ÚServer)Úread_client_data)ÚThreadNc s$ e Zd Z‡ fdd„Zdd„ Z‡ ZS )ÚFedBNc sF t ƒ ||¡ | ¡ | |t¡ td| j› d| j› �ƒ tdƒ d S )Nz
Join ratio / total clients: z / z%Finished creating server and clients.)ÚsuperÚ__init__Úset_slow_clientsÚ set_clientsr ÚprintÚ
join_ratioÚ num_clients)ÚselfÚargsÚtimes)Ú __class__© úE/root/autodl-tmp/PFL-Non-IID-master/system/flcore/servers/serverbn.pyr s
zFedBN.__init__c C sª x~t | jd ƒD ]l}| ¡ | _| ¡ || j dkrTtd|› d�ƒ tdƒ | ¡ x| jD ] }| ¡ q\W | ¡ |
¡ qW tdƒ tt | j ƒƒ |
¡ | ¡ d S )Né r z
-------------Round number: z
-------------z
Evaluate global modelz
Best global accuracy.)ÚrangeÚ
global_roundsÚselect_clientsÚselected_clientsÚ send_modelsÚeval_gapr ÚevaluateÚtrainÚreceive_modelsÚaggregate_parametersÚmaxÚ rs_test_accÚ save_resultsÚsave_global_model)r ÚiÚclientr r r r s
z FedBN.train)Ú__name__Ú
__module__Ú __qualname__r r Ú
__classcell__r r )r r r s
r )
Zflcore.clients.clientbnr Úflcore.servers.serverbaser Úutils.data_utilsr Ú threadingr Útimer r r r r Ú<module> s
| 1,512 | Python | .py | 13 | 115 | 346 | 0.498 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,825 | serverrod.cpython-37.pyc | hkgdifyu_pFedPT/system/flcore/servers/__pycache__/serverrod.cpython-37.pyc | B
¸:ccç ã @ sL d dl mZ d dlmZ d dlmZ d dlmZ d dlZG dd„ deƒZ dS )é )Ú clientROD)ÚServer)Úread_client_data)ÚThreadNc s$ e Zd Z‡ fdd„Zdd„ Z‡ ZS )ÚFedRODc sF t ƒ ||¡ | ¡ | |t¡ td| j› d| j› �ƒ tdƒ d S )Nz
Join ratio / total clients: z / z%Finished creating server and clients.)ÚsuperÚ__init__Úset_slow_clientsÚ set_clientsr ÚprintÚ
join_ratioÚ num_clients)ÚselfÚargsÚtimes)Ú __class__© úF/root/autodl-tmp/PFL-Non-IID-master/system/flcore/servers/serverrod.pyr s
zFedROD.__init__c C sª x~t | jd ƒD ]l}| ¡ | _| ¡ || j dkrTtd|› d�ƒ tdƒ | ¡ x| jD ] }| ¡ q\W | ¡ |
¡ qW tdƒ tt | j ƒƒ |
¡ | ¡ d S )Né r z
-------------Round number: z
-------------z
Evaluate global modelz
Best global accuracy.)ÚrangeÚ
global_roundsÚselect_clientsÚselected_clientsÚ send_modelsÚeval_gapr ÚevaluateÚtrainÚreceive_modelsÚaggregate_parametersÚmaxÚ rs_test_accÚ save_resultsÚsave_global_model)r ÚiÚclientr r r r s
z FedROD.train)Ú__name__Ú
__module__Ú __qualname__r r Ú
__classcell__r r )r r r s
r )
Zflcore.clients.clientrodr Úflcore.servers.serverbaser Úutils.data_utilsr Ú threadingr Útimer r r r r Ú<module> s
| 1,518 | Python | .py | 13 | 115.461538 | 348 | 0.5 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,826 | serveramp.cpython-37.pyc | hkgdifyu_pFedPT/system/flcore/servers/__pycache__/serveramp.cpython-37.pyc | B
·:ccÍ ã @ sd d dl Z d dlZd dlZd dlZd dlZd dlmZmZ d dl m
Z
d dl m Z G dd„ de
ƒZ
dS )é N)Ú clientAMPÚweight_flatten)ÚServer)ÚThreadc s4 e Zd Z‡ fdd„Zdd„ Zdd„ Zdd„ Z‡ ZS ) ÚFedAMPc sV t ƒ ||¡ | ¡ | |t¡ |j| _|j| _td| j› d| j › �ƒ tdƒ d S )Nz
Join ratio / total clients: z / z%Finished creating server and clients.)
ÚsuperÚ__init__Úset_slow_clientsÚ set_clientsr ÚalphaKÚsigmaÚprintÚ
join_ratioÚ num_clients)ÚselfÚargsÚtimes)Ú __class__© úF/root/autodl-tmp/PFL-Non-IID-master/system/flcore/servers/serveramp.pyr s zFedAMP.__init__c C s¢ xvt | jd ƒD ]d}| ¡ | _| ¡ || j dkrTtd|› d�ƒ tdƒ | ¡ x| jD ] }| ¡ q\W | ¡ qW tdƒ tt
| j ƒƒ | ¡ |
¡ d S )Né r z
-------------Round number: z
-------------z
Evaluate global modelz
Best global accuracy.)ÚrangeÚ
global_roundsÚselect_clientsÚselected_clientsÚ send_modelsÚeval_gapr
ÚevaluateÚtrainÚreceive_modelsÚmaxÚ rs_test_accÚ save_resultsÚsave_global_model)r ÚiÚclientr r r r s
z FedAMP.trainc
C s¤ t | jƒdkst‚t | jƒdk�r �xz| jD �]n}t | j¡}x| ¡ D ]}|j ¡ qHW t
| j ¡}xpt
| jƒD ]b\}}|j| j| krÌt|jƒ}t|ƒ}|| d¡} t
| | ¡} | j| | ¡ ||< qrd||< qrW dt
|¡ }
xNt
| jƒD ]@\}}x6t| ¡ | ¡ ƒD ] \}} | j|| | 7 _�qW qòW t ¡ } |j�r`t dt tj ¡ ¡ ¡ | ||
¡ |j d d7 < |j d dt ¡ | 7 < q,W d S )Nr éÿÿÿÿr gš™™™™™¹?Ú
num_roundsÚ
total_costé )!Úlenr ÚAssertionErrorÚuploaded_modelsÚcopyÚdeepcopyÚ global_modelÚ
parametersÚdataÚzero_ÚtorchÚzerosÚ join_clientsÚ enumerateÚidÚ uploaded_idsr ÚmodelÚviewÚdotr ÚeÚsumÚzipÚtimeÚ send_slowÚsleepÚnpÚabsÚrandomÚrandÚset_parametersÚsend_time_cost)
r ÚcÚmuÚparamÚcoefÚjÚmwZ weights_iZ weights_jÚsubZ coef_selfZparam_jÚ
start_timer r r r 6 s2
zFedAMP.send_modelsc C s t | | j ¡| j S )N)ÚmathÚexpr )r Úxr r r r< Z s zFedAMP.e)Ú__name__Ú
__module__Ú __qualname__r r r r< Ú
__classcell__r r )r r r s
$r )r3 r- r? ÚnumpyrB rP Zflcore.clients.clientampr r Úflcore.servers.serverbaser Ú threadingr r r r r r Ú<module> s | 2,678 | Python | .py | 24 | 110.291667 | 333 | 0.469303 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,827 | serverlocal.cpython-37.pyc | hkgdifyu_pFedPT/system/flcore/servers/__pycache__/serverlocal.cpython-37.pyc | B
¿:cc… ã @ s8 d dl mZ d dlmZ d dlmZ G dd„ deƒZdS )é )Ú clientAVG)ÚServer)ÚThreadc s$ e Zd Z‡ fdd„Zdd„ Z‡ ZS )ÚLocalc sF t ƒ ||¡ | ¡ | |t¡ td| j› d| j› �ƒ tdƒ d S )Nz
Join ratio / total clients: z / z%Finished creating server and clients.)ÚsuperÚ__init__Úset_slow_clientsÚ set_clientsr ÚprintÚ
join_ratioÚ num_clients)ÚselfÚargsÚtimes)Ú __class__© úH/root/autodl-tmp/PFL-Non-IID-master/system/flcore/servers/serverlocal.pyr s
zLocal.__init__c C sœ xpt | jd ƒD ]^}| ¡ | _|| j dkrLtd|› d�ƒ tdƒ | ¡ | ¡ | _x| jD ] }| ¡ q^W qW tdƒ tt| j ƒƒ |
¡ | ¡ d S )Né r z
-------------Round number: z
-------------z
Evaluate global modelz
Best global accuracy.) ÚrangeÚ
global_roundsÚselect_clientsÚselected_clientsÚeval_gapr
ÚevaluateÚtrainÚmaxÚ rs_test_accÚ save_resultsÚsave_global_model)r
ÚiÚclientr r r r s
z Local.train)Ú__name__Ú
__module__Ú __qualname__r r Ú
__classcell__r r )r r r s
r N)Zflcore.clients.clientavgr Úflcore.servers.serverbaser Ú threadingr r r r r r Ú<module> s | 1,375 | Python | .py | 13 | 104.461538 | 306 | 0.48496 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,828 | serverperavg.cpython-37.pyc | hkgdifyu_pFedPT/system/flcore/servers/__pycache__/serverperavg.cpython-37.pyc | B
¿:cc‚ ã @ sH d dl Z d dlZd dlmZ d dlmZ d dlmZ G dd„ deƒZdS )é N)Ú clientPerAvg)ÚServer)ÚThreadc s, e Zd Z‡ fdd„Zdd„ Zdd„ Z‡ ZS )ÚPerAvgc sF t ƒ ||¡ | ¡ | |t¡ td| j› d| j› �ƒ tdƒ d S )Nz
Join ratio / total clients: z / z%Finished creating server and clients.)ÚsuperÚ__init__Úset_slow_clientsÚ set_clientsr ÚprintÚ
join_ratioÚ num_clients)ÚselfÚargsÚtimes)Ú __class__© úI/root/autodl-tmp/PFL-Non-IID-master/system/flcore/servers/serverperavg.pyr s
zPerAvg.__init__c C s² x†t | jd ƒD ]t}| ¡ | _| ¡ || j dkrTtd|› d�ƒ tdƒ | ¡ x | jD ]}| ¡ | ¡ q\W | ¡ |
¡ qW tdƒ tt | j ƒƒ |
¡ | ¡ d S )Né r z
-------------Round number: z
-------------z+
Evaluate global model with one step updatez
Best global accuracy.)ÚrangeÚ
global_roundsÚselect_clientsÚselected_clientsÚ send_modelsÚeval_gapr
Úevaluate_one_stepÚtrainÚreceive_modelsÚaggregate_parametersÚmaxÚ rs_test_accÚ save_resultsÚsave_global_model)r
ÚiÚclientr r r r s
z PerAvg.trainc C sš g }x(| j D ] }| t |j¡¡ | ¡ q W | ¡ }x(t| j ƒD ]\}}| || |j¡ qBW t |d ƒd t |d ƒ }| j
|¡ t d |¡ƒ d S )Né g ğ?r z Average Test Accurancy: {:.4f})
ÚclientsÚappendÚcopyÚdeepcopyÚmodelZtrain_one_stepÚ test_metricsÚ enumerateÚ clone_modelÚsumr r
Úformat)r
Z models_tempÚcÚstatsr" Útest_accr r r r 4 s zPerAvg.evaluate_one_step)Ú__name__Ú
__module__Ú __qualname__r r r Ú
__classcell__r r )r r r s
!r ) r' ÚtorchZflcore.clients.clientperavgr Úflcore.servers.serverbaser Ú threadingr r r r r r Ú<module> s
| 2,014 | Python | .py | 16 | 124.5 | 335 | 0.486243 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,829 | servermoon.cpython-38.pyc | hkgdifyu_pFedPT/system/flcore/servers/__pycache__/servermoon.cpython-38.pyc | U
[Ðýc’ ã @ sL d dl mZ d dlmZ d dlmZ d dlmZ d dlZG dd„ deƒZ dS )é )Ú
clientMOON)ÚServer)Úread_client_data)ÚThreadNc s$ e Zd Z‡ fdd„Zdd„ Z‡ ZS )ÚMOONc sL t ƒ ||¡ | ¡ | |t¡ td| j› d| j› �ƒ tdƒ g | _d S )Nz
Join ratio / total clients: z / z%Finished creating server and clients.) ÚsuperÚ__init__Úset_slow_clientsÚ set_clientsr ÚprintÚ
join_ratioÚ num_clientsÚBudget)ÚselfÚargsÚtimes©Ú __class__© õX D:\京东\promot\第二次投稿\实验\native - pro\system\flcore\servers\servermoon.pyr s z
MOON.__init__c C sd g }g }d| _ d}t| jd ƒD ]Ô}t ¡ }| ¡ | _| ¡ || j dkrltd|› d�ƒ tdƒ | ¡ | jD ] }|
¡ qr|| j dkr¢tdƒ | j |d� | ¡ | ¡ | j
t ¡ | ¡ td | j
d
ƒ | j| jg| jd �| _ |d7 }q td ƒ tt| jƒƒ td
ƒ tt|ƒƒ tdƒ tt| j
dd … ƒt| j
dd … ƒ ƒ | ¡ | ¡ d S )NFr é z
-------------Round number: z
-------------z
Evaluate global modelz
Evaluate local model)Úaccz2--------------------------------------------------éÿÿÿÿ)Úacc_lssÚtop_cntz
Best global accuracy.z
Best local accuracy.z
Averaged time per iteration.)ÚdoneÚrangeÚ
global_roundsÚtimeÚselect_clientsÚselected_clientsÚ send_modelsÚeval_gapr ÚevaluateÚtrainÚreceive_modelsÚaggregate_parametersr ÚappendÚ
check_doneÚ rs_test_accr ÚmaxÚsumÚlenÚ save_resultsÚsave_global_model)r Ú local_accÚiÚs_tÚclientr r r r$ s>
(z
MOON.train)Ú__name__Ú
__module__Ú __qualname__r r$ Ú
__classcell__r r r r r s r )
Zflcore.clients.clientmoonr Úflcore.servers.serverbaser Úutils.data_utilsr Ú threadingr r r r r r r Ú<module> s
| 2,012 | Python | .py | 22 | 90.227273 | 352 | 0.473631 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,830 | serverbase.cpython-38.pyc | hkgdifyu_pFedPT/system/flcore/servers/__pycache__/serverbase.cpython-38.pyc | U
¨ÿcÍ) ã @ sX d dl Z d dlZd dlZd dlZd dlZd dlZd dlZd dlm Z G dd„ de
ƒZ dS )é N)Úread_client_datac @ s° e Zd Zdd„ Zdd„ Zdd„ Zdd„ Zd d
„ Zd d „ Zd
d„ Z dd„ Z
dd„ Z dd„ Z dd„ Z
dd„ Zdd„ Zdd „ Zd d „ Zd d „ Zd!d"„ Zd*d$d%„Zd&d'„ Zd+d(d)„Zd#S ),ÚServerc C s |j | _ |j| _|j| _|j| _|j| _|j| _t |j ¡| _
|j | _ |j | _ t
| j | j ƒ| _|j| _|j| _|j| _|j| _|j| _|j| _|| _|j| _d| _g | _g | _g | _g | _g | _ g | _ g | _ g | _ g | _ g | _!g | _"|| _#|j$| _$|j%| _%|j&| _&|j'| _'d S )Néd )(ÚdeviceÚdatasetÚ
global_roundsÚ local_stepsÚ
batch_sizeÚlocal_learning_rateÚ
learning_rateÚcopyÚdeepcopyÚmodelÚ global_modelÚ num_clientsÚ
join_ratioÚintÚ join_clientsÚ algorithmÚ time_selectÚgoalÚtime_thretholdÚsave_folder_nameÚ
num_promptÚargsÚ plocal_stepsÚtop_cntÚclientsÚselected_clientsÚtrain_slow_clientsÚsend_slow_clientsÚuploaded_weightsÚ uploaded_idsÚuploaded_modelsÚ rs_test_accÚrs_test_acc_stdÚ rs_test_aucÚ
rs_train_lossÚtimesÚeval_gapÚclient_drop_rateÚtrain_slow_rateÚsend_slow_rate)Úselfr r( © r. õb D:\京东\promot\第二次投稿\å®�验\æœ�务器\native - pro\system\flcore\servers\serverbase.pyÚ__init__
sF zServer.__init__c C sr t t| jƒ| j| jƒD ]V\}}}t| j||dd�}t| j||dd�}|||t|ƒt|ƒ||d�}| j |¡ qd S )NT)Úis_trainF)ÚidÚ
train_samplesÚ test_samplesÚ
train_slowÚ send_slow)
ÚzipÚranger r r r r Úlenr Úappend) r- r Z clientObjÚir5 r6 Ú
train_dataÚ test_dataÚclientr. r. r/ Ú set_clients8 s ûzServer.set_clientsc C sV dd„ t | jƒD ƒ}dd„ t | jƒD ƒ}tj |t|| j ƒ¡}|D ] }d||< qD|S )Nc S s g | ]}d ‘qS )Fr. ©Ú.0r; r. r. r/ Ú
<listcomp>I s z.Server.select_slow_clients.<locals>.<listcomp>c S s g | ]}|‘qS r. r. r@ r. r. r/ rB J s T)r8 r ÚnpÚrandomÚchoicer )r- Z slow_rateZ slow_clientsÚidxÚidx_r; r. r. r/ Úselect_slow_clientsH s
zServer.select_slow_clientsc C s | | j¡| _| | j¡| _d S ©N)rH r+ r r, r )r- r. r. r/ Úset_slow_clientsQ s ÿÿzServer.set_slow_clientsc C s t tjj| j| jdd�ƒ}|S )NF)Úreplace)ÚlistrC rD rE r r )r- r r. r. r/ Úselect_clientsW s zServer.select_clientsc C s. t | jƒdkst‚| jD ]}| | j¡ qd S ©Nr )r9 r ÚAssertionErrorÚset_parametersr )r- r> r. r. r/ Ú send_models\ s
zServer.send_modelsc C s� t | jƒdkst‚g | _d}g | _g | _| jD ]8}| j |j¡ ||j7 }| j |j¡ | j |j ¡ q.t
| jƒD ]\}}|| | j|< qrd S rN ) r9 r rO r! r" r# r: r3 r2 r Ú enumerate)r- Z tot_samplesr> r; Úwr. r. r/ Úreceive_modelsd s
zServer.receive_modelsc C sf t | jƒdkst‚t | jd ¡| _| j ¡ D ]}|j ¡ q.t | j
| jƒD ]\}}| ||¡ qLd S rN ) r9 r# rO r r
r Ú
parametersÚdataÚzero_r7 r! Úadd_parameters)r- ÚparamrS Ú client_modelr. r. r/ Úaggregate_parameterss s zServer.aggregate_parametersc C s: t | j ¡ | ¡ ƒD ] \}}| j|j ¡ | 7 _qd S rI )r7 r rU rV Úclone)r- rS rZ Z server_paramZ client_paramr. r. r/ rX } s zServer.add_parametersc C sØ t j d| j| jjd | jj d | jj d | jj d | jj d | jj
¡}t j |¡sht |¡ t j || j
d d t| jƒ d t| jƒ d t| jƒ d t| jƒ d t| jƒ d ¡}t | j|¡ d S ©NÚmodelsÚ*Ú_serverÚ_ú.pt)ÚosÚpathÚjoinr r Úarv1Úarv2Úarv3Úarv4Úarv5Úarv6ÚexistsÚmakedirsr Ústrr r r r r ÚtorchÚsaver ©r- Z
model_pathr. r. r/ Úsave_global_model� s
R
^zServer.save_global_modelc C sĞ t j d| j| jjd | jj d | jj d | jj d | jj d | jj
¡}t j || j d d t | j
ƒ d t | jƒ d t | jƒ d t | jƒ d t | jƒ d ¡}t j |¡sÀt‚t |¡| _d S r] )rc rd re r r rf rg rh ri rj rk r rn r r r r r rl rO ro Úloadr rq r. r. r/ Ú
load_modelˆ s R^zServer.load_modelc C s¸ t j d| j| jjd | jj d | jj d | jj d | jj d | jj
¡}t j || j d t | j
ƒ d t | jƒ d t | jƒ d t | jƒ d t | jƒ d ¡}t j |¡S )Nr^ r_ ra rb )rc rd re r r rf rg rh ri rj rk r rn r r r r r rl rq r. r. r/ Ú model_exists� s RZzServer.model_existsc C sV | j d | j }d| jj d | jj d | jj d | jj d | jj d | jj d }t j
|¡srt |¡ t
| jƒ�rR|d | j d t| jƒ d t| jƒ d t| jƒ d t| jƒ d t| jƒ d t| jƒ }|d |¡ }td| ƒ t |d¡�F}|jd| jd � |jd
| j d � |jd | j d � |jd | j d � W 5 Q R X d S )
Nra z ../results/r_ ú/z{}.h5z File path: rS r$ )rV r% r& r' ) r r r rf rg rh ri rj rk rc rd rl rm r9 r$ r rn r( r r r r r ÚformatÚprintÚh5pyZFileZcreate_datasetr% r& r' )r- ÚalgoÚ result_pathÚ file_pathZhfr. r. r/ Ú save_results“ s L
b zServer.save_resultsc C s„ t j | j¡st | j¡ t |t j | jd| d t| j ƒ d t| j
ƒ d t| j ƒ d t| j ƒ d t| j
ƒ d ¡¡ d S ©NZserver_ra rb )rc rd rl r rm ro rp re rn r r r r r )r- ÚitemÚ item_namer. r. r/ Ú save_item¤ s zServer.save_itemc C sd t tj | jd| d t| jƒ d t| jƒ d t| j ƒ d t| j
ƒ d t| j ƒ d ¡¡S r~ ) ro rs rc rd re r rn r r r r r )r- r€ r. r. r/ Ú load_item© s zServer.load_itemc C sh g }g }g }| j D ]8}| ¡ \}}}| |d ¡ | || ¡ | |¡ qdd„ | j D ƒ}||||fS )Nç ğ?c S s g | ]
}|j ‘qS r. ©r2 ©rA Úcr. r. r/ rB ¹ s z'Server.test_metrics.<locals>.<listcomp>)r Ú test_metricsr: ) r- Ú num_samplesZ tot_correctZtot_aucr† ÚctÚnsÚaucÚidsr. r. r/ r‡ ¬ s
zServer.test_metricsc C sR g }g }| j D ](}| ¡ \}}| |¡ | |d ¡ qdd„ | j D ƒ}|||fS )Nrƒ c S s g | ]
}|j ‘qS r. r„ r… r. r. r/ rB Å s z(Server.train_metrics.<locals>.<listcomp>)r Ú
train_metricsr: )r- rˆ Úlossesr† ÚclrŠ rŒ r. r. r/ r� ½ s
zServer.train_metricsNc
C sP | ¡ }| ¡ }t|d ƒd t|d ƒ }t|d ƒd t|d ƒ }t|d ƒd t|d ƒ }dd„ t|d |d ƒD ƒ}dd„ t|d |d ƒD ƒ} |d kr²| j |¡ n
| |¡ | j |¡ | j t |¡¡ |d krğ| j
|¡ n
| |¡ t d |¡ƒ t d |¡ƒ t d
|¡ƒ t d t |¡¡ƒ t d t | ¡¡ƒ d S )
Né rƒ é é c S s g | ]\}}|| ‘qS r. r. ©rA ÚaÚnr. r. r/ rB Ò s z#Server.evaluate.<locals>.<listcomp>c S s g | ]\}}|| ‘qS r. r. r“ r. r. r/ rB Ó s zAveraged Train Loss: {:.4f}z Averaged Test Accurancy: {:.4f}zAveraged Test AUC: {:.4f}zStd Test Accurancy: {:.4f}zStd Test AUC: {:.4f})
r‡ r� Úsumr7 r$ r: r& r% rC Ústdr' rx rw )
r- ÚaccÚlossÚstatsZ stats_trainÚtest_accÚtest_aucÚ
train_lossZaccsZaucsr. r. r/ ÚevaluateÊ s(
zServer.evaluatec C s. t d |¡ƒ t d |¡ƒ t d |¡ƒ d S )Nz Average Test Accurancy: {:.4f}zAverage Test AUC: {:.4f}zAverage Train Loss: {:.4f})rx rw )r- r› rœ r� r. r. r/ Úprint_ç s z
Server.print_c C sğ |D ]æ}|d krr|d krrt |ƒt t |¡d¡jd |k}t |ƒdko^t || d … ¡|k }|rj|rjqê dS q|d kr¬t |ƒt t |¡d¡jd |k}|r¤qê dS q|d kræt |ƒdkoÖt || d … ¡|k }|rŞqê dS qt‚qdS )Nr‘ r FT)r9 ro ÚtopkÚtensorÚindicesrC r— ÚNotImplementedError)r- Zacc_lssr Ú div_valueZacc_lsZfind_topZfind_divr. r. r/ Ú
check_doneì s& $$$$zServer.check_done)NN)NN)Ú__name__Ú
__module__Ú __qualname__r0 r? rH rJ rM rQ rT r[ rX rr rt ru r} r� r‚ r‡ r� r� rŸ r¥ r. r. r. r/ r s( +
r ) ro rc ÚnumpyrC ry r ÚtimerD Úutils.data_utilsr Úobjectr r. r. r. r/ Ú<module> s | 9,576 | Python | .py | 50 | 190.36 | 711 | 0.388055 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,831 | serverbase.cpython-37.pyc | hkgdifyu_pFedPT/system/flcore/servers/__pycache__/serverbase.cpython-37.pyc | B
vĞecû' ã @ sX d dl Z d dlZd dlZd dlZd dlZd dlZd dlZd dlm Z G dd„ de
ƒZ dS )é N)Úread_client_datac @ s° e Zd Zdd„ Zdd„ Zdd„ Zdd„ Zd d
„ Zd d „ Zd
d„ Z dd„ Z
dd„ Z dd„ Z dd„ Z
dd„ Zdd„ Zdd „ Zd d „ Zd d „ Zd!d"„ Zd*d$d%„Zd&d'„ Zd+d(d)„Zd#S ),ÚServerc C s
|j | _ |j| _|j| _|j| _|j| _|j| _t |j ¡| _
|j | _ |j | _ t
| j | j ƒ| _|j| _|j| _|j| _|j| _|j| _|j| _|j| _d| _g | _g | _g | _g | _g | _g | _ g | _ g | _ g | _ g | _ g | _!|| _"|j#| _#|j$| _$|j%| _%|j&| _&d S )Néd )'ÚdeviceÚdatasetÚ
global_roundsÚ local_stepsÚ
batch_sizeÚlocal_learning_rateÚ
learning_rateÚcopyÚdeepcopyÚmodelÚ global_modelÚ num_clientsÚ
join_ratioÚintÚ join_clientsÚ algorithmÚ time_selectÚgoalÚtime_thretholdÚsave_folder_nameÚ
num_promptÚ plocal_stepsÚtop_cntÚclientsÚselected_clientsÚtrain_slow_clientsÚsend_slow_clientsÚuploaded_weightsÚ uploaded_idsÚuploaded_modelsÚ rs_test_accÚrs_test_acc_stdÚ rs_test_aucÚ
rs_train_lossÚtimesÚeval_gapÚclient_drop_rateÚtrain_slow_rateÚsend_slow_rate)ÚselfÚargsr' © r. úG/root/autodl-tmp/PFL-Non-IID-master/system/flcore/servers/serverbase.pyÚ__init__
sD zServer.__init__c C sr xlt t| jƒ| j| jƒD ]R\}}}t| j|dd�}t| j|dd�}|||t|ƒt|ƒ||d�}| j |¡ qW d S )NT)Úis_trainF)ÚidÚ
train_samplesÚ test_samplesÚ
train_slowÚ send_slow)
ÚzipÚranger r r r r Úlenr Úappend) r, r- Z clientObjÚir5 r6 Ú
train_dataÚ test_dataÚclientr. r. r/ Ú set_clients7 s "zServer.set_clientsc C sZ dd„ t | jƒD ƒ}dd„ t | jƒD ƒ}tj |t|| j ƒ¡}x|D ] }d||< qFW |S )Nc S s g | ]}d ‘qS )Fr. )Ú.0r; r. r. r/ ú
<listcomp>H s z.Server.select_slow_clients.<locals>.<listcomp>c S s g | ]}|‘qS r. r. )r@ r; r. r. r/ rA I s T)r8 r ÚnpÚrandomÚchoicer )r, Z slow_rateZ slow_clientsÚidxÚidx_r; r. r. r/ Úselect_slow_clientsG s
zServer.select_slow_clientsc C s | | j¡| _| | j¡| _d S )N)rG r* r r+ r )r, r. r. r/ Úset_slow_clientsP s
zServer.set_slow_clientsc C s t tjj| j| jdd�ƒ}|S )NF)Úreplace)ÚlistrB rC rD r r )r, r r. r. r/ Úselect_clientsV s zServer.select_clientsc C s2 t | jƒdkst‚x| jD ]}| | j¡ qW d S )Nr )r9 r ÚAssertionErrorÚset_parametersr )r, r> r. r. r/ Ú send_models[ s zServer.send_modelsc C s– t | jƒdkst‚g | _d}g | _g | _xB| jD ]8}| j |j¡ ||j7 }| j |j¡ | j |j ¡ q0W x$t
| jƒD ]\}}|| | j|< qxW d S )Nr ) r9 r rL r r! r" r: r3 r2 r Ú enumerate)r, Z tot_samplesr> r; Úwr. r. r/ Úreceive_modelsc s
zServer.receive_modelsc C sn t | jƒdkst‚t | jd ¡| _x | j ¡ D ]}|j ¡ q0W x&t | j
| jƒD ]\}}| ||¡ qRW d S )Nr ) r9 r" rL r r
r Ú
parametersÚdataÚzero_r7 r Úadd_parameters)r, ÚparamrP Ú client_modelr. r. r/ Úaggregate_parametersr s zServer.aggregate_parametersc C s> x8t | j ¡ | ¡ ƒD ] \}}| j|j ¡ | 7 _qW d S )N)r7 r rR rS Úclone)r, rP rW Z server_paramZ client_paramr. r. r/ rU | s zServer.add_parametersc C s– t j d| j¡}t j |¡s&t |¡ t j || jd d t| jƒ d t| j ƒ d t| j
ƒ d t| j ƒ d t| j ƒ d ¡}t
| j|¡ d S )NÚmodelsÚ_serverÚ_z.pt)ÚosÚpathÚjoinr ÚexistsÚmakedirsr Ústrr r r r r ÚtorchÚsaver )r, Ú
model_pathr. r. r/ Úsave_global_model€ s
^zServer.save_global_modelc C s� t j d| j¡}t j || jd d t| jƒ d t| jƒ d t| jƒ d t| j ƒ d t| j
ƒ d ¡}t j |¡s~t ‚t
|¡| _d S )NrZ r[ r\ z.pt)r] r^ r_ r r rb r r r r r r` rL rc Úloadr )r, re r. r. r/ Ú
load_model‡ s ^zServer.load_modelc C sv t j d| j¡}t j || jd t| jƒ d t| jƒ d t| jƒ d t| j ƒ d t| j
ƒ d ¡}t j |¡S )NrZ r\ z.pt) r] r^ r_ r r rb r r r r r r` )r, re r. r. r/ Ú model_exists� s ZzServer.model_existsc C s | j d | j }d}tj |¡s*t |¡ t| jƒ�r
|d | j d t | j
ƒ d t | j ƒ d t | j ƒ d t | j
ƒ d t | jƒ d t | jƒ }|d |¡ }td| ƒ t |d¡�F}|jd| jd� |jd| jd� |jd | jd� |jd
| jd� W d Q R X d S ) Nr\ z ../results/z{}.h5z File path: rP r# )rS r$ r% r& )r r r] r^ r` ra r9 r# r rb r' r r r r r ÚformatÚprintÚh5pyÚFileZcreate_datasetr$ r% r& )r, ÚalgoÚ result_pathÚ file_pathZhfr. r. r/ Ú save_results’ s
b zServer.save_resultsc C s„ t j | j¡st | j¡ t |t j | jd| d t| j ƒ d t| j
ƒ d t| j ƒ d t| j ƒ d t| j
ƒ d ¡¡ d S )NÚserver_r\ z.pt)r] r^ r` r ra rc rd r_ rb r r r r r )r, ÚitemÚ item_namer. r. r/ Ú save_item£ s zServer.save_itemc C sd t tj | jd| d t| jƒ d t| jƒ d t| j ƒ d t| j
ƒ d t| j ƒ d ¡¡S )Nrr r\ z.pt) rc rg r] r^ r_ r rb r r r r r )r, rt r. r. r/ Ú load_item¨ s zServer.load_itemc C sl g }g }g }xB| j D ]8}| ¡ \}}}| |d ¡ | || ¡ | |¡ qW dd„ | j D ƒ}||||fS )Ng ğ?c S s g | ]
}|j ‘qS r. )r2 )r@ Úcr. r. r/ rA ¸ s z'Server.test_metrics.<locals>.<listcomp>)r Ú test_metricsr: ) r, Ú num_samplesZ tot_correctZtot_aucrw ÚctÚnsÚaucÚidsr. r. r/ rx « s zServer.test_metricsc C sV g }g }x2| j D ](}| ¡ \}}| |¡ | |d ¡ qW dd„ | j D ƒ}|||fS )Ng ğ?c S s g | ]
}|j ‘qS r. )r2 )r@ rw r. r. r/ rA Ä s z(Server.train_metrics.<locals>.<listcomp>)r Ú
train_metricsr: )r, ry Úlossesrw Úclr{ r} r. r. r/ r~ ¼ s
zServer.train_metricsNc
C sP | ¡ }| ¡ }t|d ƒd t|d ƒ }t|d ƒd t|d ƒ }t|d ƒd t|d ƒ }dd„ t|d |d ƒD ƒ}dd„ t|d |d ƒD ƒ} |d kr²| j |¡ n
| |¡ | j |¡ | j t |¡¡ |d krğ| j
|¡ n
| |¡ t d |¡ƒ t d |¡ƒ t d
|¡ƒ t d t |¡¡ƒ t d t | ¡¡ƒ d S )
Né g ğ?é é c S s g | ]\}}|| ‘qS r. r. )r@ ÚaÚnr. r. r/ rA Ñ s z#Server.evaluate.<locals>.<listcomp>c S s g | ]\}}|| ‘qS r. r. )r@ r„ r… r. r. r/ rA Ò s zAveraged Train Loss: {:.4f}z Averaged Test Accurancy: {:.4f}zAveraged Test AUC: {:.4f}zStd Test Accurancy: {:.4f}zStd Test AUC: {:.4f})
rx r~ Úsumr7 r# r: r% r$ rB Ústdr& rk rj )
r, ÚaccÚlossÚstatsZ stats_trainÚtest_accÚtest_aucÚ
train_lossZaccsZaucsr. r. r/ ÚevaluateÉ s(
zServer.evaluatec C s. t d |¡ƒ t d |¡ƒ t d |¡ƒ d S )Nz Average Test Accurancy: {:.4f}zAverage Test AUC: {:.4f}zAverage Train Loss: {:.4f})rk rj )r, r‹ rŒ r� r. r. r/ Úprint_æ s z
Server.print_c C sî xè|D ]à}|d krr|d krrt |ƒt t |¡d¡jd |k}t |ƒdko`t || d … ¡|k }|rl|rlqædS q|d krªt |ƒt t |¡d¡jd |k}|r¤qædS q|d krât |ƒdkoÔt || d … ¡|k }|rÜqædS qt‚qW dS )Nr‚ r FT)r9 rc ÚtopkÚtensorÚindicesrB r‡ ÚNotImplementedError)r, Zacc_lssr Ú div_valueZacc_lsZfind_topZfind_divr. r. r/ Ú
check_doneë s&
$$$$zServer.check_done)NN)NN)Ú__name__Ú
__module__Ú __qualname__r0 r? rG rH rK rN rQ rX rU rf rh ri rq ru rv rx r~ r� r� r• r. r. r. r/ r s( *
r ) rc r] ÚnumpyrB rl r ÚtimerC Úutils.data_utilsr Úobjectr r. r. r. r/ Ú<module> s | 9,090 | Python | .py | 50 | 180.68 | 695 | 0.394757 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,832 | serverfomo.cpython-38.pyc | hkgdifyu_pFedPT/system/flcore/servers/__pycache__/serverfomo.cpython-38.pyc | U
•icW ã @ s` d dl Z d dlZd dlZd dlZd dlZd dlmZ d dlm Z d dl
m Z G dd„ de ƒZ dS )é N)Ú
clientFomo)ÚServer)ÚThreadc s4 e Zd Z‡ fdd„Zdd„ Zdd„ Zdd„ Z‡ ZS ) ÚFedFomoc s€ t ƒ ||¡ | ¡ | |t¡ t tj| j| j d�¡| _
| j g| _ g | _
t|j| jƒ| _td| j› d| j› �ƒ tdƒ d S )N)Údevicez
Join ratio / total clients: z / z%Finished creating server and clients.)ÚsuperÚ__init__Úset_slow_clientsÚ set_clientsr ÚtorchÚdiagÚonesÚ num_clientsr ÚPÚ global_modelÚuploaded_modelsÚ uploaded_idsÚminÚMÚ join_clientsÚprintÚ
join_ratio)ÚselfÚargsÚtimes©Ú __class__© õE D:\京东\promot\cifar\cifar\tiny\system\flcore\servers\serverfomo.pyr s
zFedFomo.__init__c C sš t | jd ƒD ]`}| ¡ | _| ¡ || j dkrRtd|› d�ƒ tdƒ | ¡ | jD ] }| ¡ qX| ¡ qtdƒ tt
| j ƒƒ | ¡ |
¡ d S )Né r z
-------------Round number: z
-------------z
Evaluate global modelz
Best global accuracy.)ÚrangeÚ
global_roundsÚselect_clientsÚselected_clientsÚ send_modelsÚeval_gapr ÚevaluateÚtrainÚreceive_modelsÚmaxÚ rs_test_accÚ save_resultsÚsave_global_model)r ÚiÚclientr r r r' s
z
FedFomo.trainc C sø t | jƒdkst‚| jD ]Ú}t ¡ }|jrDt dt tj ¡ ¡ ¡ t | j
ƒdkrÂt | j t | j
ƒƒ}t | j|j | j
|¡j ¡ }g }g }|D ]$}| | j
| ¡ | | j
| ¡ q�| ||¡ |jd d7 < |jd dt ¡ | 7 < qd S )Nr gš™™™™™¹?Ú
num_roundsr Ú
total_costé )ÚlenÚclientsÚAssertionErrorÚtimeÚ send_slowÚsleepÚnpÚabsÚrandomÚrandr r r r r Útopkr ÚidÚindicesÚtolistÚappendr( Úsend_time_cost)r r. Ú
start_timeZM_r> r r r- r r r r$ 9 s
zFedFomo.send_modelsc C sø t | jƒdkst‚t | jtd| j | j ƒ¡}g | _g | _ d}g | _
|D ]†}|j d |j d |j d |j d }|| j
krJ| j |j¡ | j |j¡ ||j7 }| j
t |j¡¡ | j|j |j7 < qJt| j ƒD ]\}}|| | j |< qÜd S )Nr r r0 r/ )r2 r# r4 r: ÚsampleÚintÚclient_drop_rater r Úuploaded_weightsr Útrain_time_costrA Útime_thretholdr@ r= Ú
train_samplesÚcopyÚdeepcopyÚmodelr Z
weight_vectorÚ enumerate)r Zactive_clientsÚ tot_samplesr. Zclient_time_costr- Úwr r r r( P s* ÿÿ
zFedFomo.receive_models)Ú__name__Ú
__module__Ú __qualname__r r' r$ r( Ú
__classcell__r r r r r s r )
r r5 rJ r: Únumpyr8 Zflcore.clients.clientfomor Úflcore.servers.serverbaser Ú threadingr r r r r r Ú<module> s | 2,998 | Python | .py | 28 | 105.928571 | 428 | 0.453383 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,833 | serverproto.cpython-39.pyc | hkgdifyu_pFedPT/system/flcore/servers/__pycache__/serverproto.cpython-39.pyc | a
f¾`cÜ ã @ sh d dl mZ d dlmZ d dlmZ d dlmZ d dlZd dl Z
d dl m Z G dd„ deƒZ
d d
„ ZdS ) é )Ú clientProto)ÚServer)Úread_client_data)ÚThreadN)Ú defaultdictc s> e Zd Z‡ fdd„Zdd„ Zdd„ Zdd„ Zd d
d „Z‡ ZS )
ÚFedProtoc sj t ƒ ||¡ | ¡ | |t¡ td| j› d| j› �ƒ tdƒ g | _|j | _ dd„ t
|j ƒD ƒ| _ d S )Nz
Join ratio / total clients: z / z%Finished creating server and clients.c S s g | ]}d ‘qS )N© )Ú.0Ú_r r út/media/sim812/391e55df-b6f2-4fe9-a920-53434a8506fa/lgh/pdept/PFL-Non-IID-master/system/flcore/servers/serverproto.pyÚ
<listcomp> ó z%FedProto.__init__.<locals>.<listcomp>) ÚsuperÚ__init__Úset_slow_clientsÚ set_clientsr ÚprintÚ
join_ratioÚ num_clientsÚBudgetÚ num_classesÚrangeÚ
global_protos)ÚselfÚargsÚtimes©Ú __class__r r r s zFedProto.__init__c C s d| _ d}| j sÔt ¡ }| ¡ | _|| j dkrX|dkrXtd|› d�ƒ tdƒ | ¡ | jD ] }| ¡ q^| ¡ t | j
ƒ| _ | ¡ | j
t ¡ | ¡ td| j
d ƒ |dkrÊ| j| jg| jd�| _ |d 7 }q
td
ƒ tt| jƒƒ tt| j
d d … ƒt| j
d d … ƒ ƒ | ¡ d S ) NFr z
-------------Round number: z
-------------z
Evaluate global modelz2--------------------------------------------------éÿÿÿÿ)Úacc_lssÚtop_cnté z
Best global accuracy.)ÚdoneÚtimeÚselect_clientsÚselected_clientsÚeval_gapr ÚevaluateÚtrainÚreceive_protosÚproto_aggregationÚuploaded_protosr Ú send_protosr ÚappendÚ
check_doneÚ rs_test_accr ÚmaxÚsumÚlenÚ save_results)r ÚiÚs_tÚclientr r r r( s.
(zFedProto.trainc C s. t | jƒdksJ ‚| jD ]}| | j¡ qd S ©Nr )r2 r% Z
set_protosr ©r r6 r r r r, C s
zFedProto.send_protosc C sJ t | jƒdksJ ‚g | _g | _| jD ] }| j |j¡ | j |j¡ q$d S r7 )r2 r% Ú uploaded_idsr+ r- ÚidÚprotosr8 r r r r) I s
zFedProto.receive_protosNc C sØ | ¡ }| ¡ }t|d ƒd t|d ƒ }t|d ƒd t|d ƒ }dd„ t|d |d ƒD ƒ}|d krz| j |¡ n
| |¡ |d krš| j |¡ n
| |¡ td |¡ƒ td |¡ƒ td t
|¡¡ƒ d S ) Né g ğ?r! c S s g | ]\}}|| ‘qS r r )r ÚaÚnr r r r X r
z%FedProto.evaluate.<locals>.<listcomp>zAveraged Train Loss: {:.4f}z Averaged Test Accurancy: {:.4f}zStd Test Accurancy: {:.4f}) Ú test_metricsÚ
train_metricsr1 Úzipr/ r- Ú
rs_train_lossr ÚformatÚnpÚstd)r ÚaccÚlossÚstatsÚ stats_trainÚtest_accÚ
train_lossÚaccsr r r r' R s
zFedProto.evaluate)NN) Ú__name__Ú
__module__Ú __qualname__r r( r, r) r' Ú
__classcell__r r r r r
s
( r c C s– t tƒ}| D ]$}| ¡ D ]}|| || ¡ qq | ¡ D ]V\}}t|ƒdkr‚d|d j }|D ]}||j7 }q`|t|ƒ ||< q:|d j||< q:|S )Nr! r )r ÚlistÚkeysr- Úitemsr2 Údata)Zlocal_protos_listZagg_protos_labelZ local_protosÚlabelZ
proto_listÚprotor4 r r r r* k s r* )Zflcore.clients.clientprotor Úflcore.servers.serverbaser Úutils.data_utilsr Ú threadingr r# ÚnumpyrD Ú collectionsr r r* r r r r Ú<module> s a | 3,711 | Python | .py | 33 | 111.30303 | 408 | 0.437075 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,834 | serverdynpt-checkpoint.py | hkgdifyu_pFedPT/system/flcore/servers/.ipynb_checkpoints/serverdynpt-checkpoint.py | import copy
import torch
from system.flcore.clients.clientdynpt import clientDynPT
from system.flcore.servers.serverbase import Server
from threading import Thread
import time
import h5py
import copy
import os
class FedDynPT(Server):
def __init__(self, args, times):
super().__init__(args, times)
# select slow clients
self.set_slow_clients()
self.set_clients(args, clientDynPT)
print(f"\nJoin ratio / total clients: {self.join_ratio} / {self.num_clients}")
print("Finished creating server and clients.")
# self.load_model()
self.Budget = []
self.diff_pro = []
self.alpha = args.alpha
self.clients_diverge = []
self.server_state = copy.deepcopy(args.model)
for param in self.server_state.parameters():
param.data = torch.zeros_like(param.data)
def train(self):
local_acc = []
self.done = False
i = 0
# while not self.done:
for i in range(self.global_rounds+1):
s_t = time.time()
self.selected_clients = self.select_clients()
self.send_models()
if i % self.eval_gap == 0:
print(f"\n-------------Round number: {i}-------------")
print("\nEvaluate global model")
self.evaluate()
temp_diff_pro = 0
for client in self.selected_clients:
temp_diff_pro_client = client.train()
temp_diff_pro = temp_diff_pro + temp_diff_pro_client.item()
print("Averaged prompr difference: {:.4f}".format(temp_diff_pro))
self.diff_pro.append(temp_diff_pro)
diverge_clents =0
for new_param, old_param in zip(self.clients[0].model.generator.parameters(), self.clients[1].model.generator.parameters()):
diff_pro = new_param - old_param
diff_pro = torch.where(diff_pro > 0, diff_pro, torch.zeros_like(diff_pro) - diff_pro)
diverge_clents = diverge_clents+torch.sum(diff_pro)
print("0 and 1 clients difference: {:.4f}".format(diverge_clents.item()))
self.clients_diverge.append(diverge_clents.item())
if i % self.eval_gap == 0:
print("\nEvaluate local model")
self.evaluate(acc=local_acc)
# threads = [Thread(target=client.train)
# for client in self.selected_clients]
# [t.start() for t in threads]
# [t.join() for t in threads]
self.receive_models()
self.update_server_state()
self.aggregate_parameters()
self.Budget.append(time.time() - s_t)
print('-' * 50, self.Budget[-1])
self.done = self.check_done(acc_lss=[self.rs_test_acc], top_cnt=self.top_cnt)
i += 1
print("\nBest global accuracy.")
# self.print_(max(self.rs_test_acc), max(
# self.rs_train_acc), min(self.rs_train_loss))
print(max(self.rs_test_acc))
print("\nBest local accuracy.")
print(max(local_acc))
print("\nAveraged time per iteration.")
print(sum(self.Budget[1:]) / len(self.Budget[1:]))
self.save_results()
self.save_global_model()
def add_parameters(self, client_model):
for server_param, client_param in zip(self.global_model.parameters(), client_model.parameters()):
server_param.data += client_param.data.clone() / self.join_clients
def aggregate_parameters(self):
assert (len(self.uploaded_models) > 0)
self.global_model = copy.deepcopy(self.uploaded_models[0])
for param in self.global_model.parameters():
param.data = torch.zeros_like(param.data)
for client_model in self.uploaded_models:
self.add_parameters(client_model)
for server_param, state_param in zip(self.global_model.parameters(), self.server_state.parameters()):
server_param.data -= (1 / self.alpha) * state_param
def update_server_state(self):
assert (len(self.uploaded_models) > 0)
model_delta = copy.deepcopy(self.uploaded_models[0])
for param in model_delta.parameters():
param.data = torch.zeros_like(param.data)
for client_model in self.uploaded_models:
for server_param, client_param, delta_param in zip(self.global_model.parameters(),
client_model.parameters(), model_delta.parameters()):
delta_param.data += (client_param - server_param) / self.num_clients
for state_param, delta_param in zip(self.server_state.parameters(), model_delta.parameters()):
state_param.data -= self.alpha * delta_param
def receive_models(self):
assert (len(self.selected_clients) > 0)
active_train_samples = 0
for client in self.selected_clients:
active_train_samples += client.train_samples
self.uploaded_weights = []
self.uploaded_ids = []
self.uploaded_models = []
for client in self.selected_clients:
self.uploaded_weights.append(client.train_samples / active_train_samples)
self.uploaded_ids.append(client.id)
self.uploaded_models.append(copy.deepcopy(client.model))
def save_results(self):
algo = self.dataset + "_" + self.algorithm
result_path = "../results/"
if not os.path.exists(result_path):
os.makedirs(result_path)
if (len(self.rs_test_acc)):
algo = algo + "_" + self.goal + "_" + str(self.times)
file_path = result_path + "{}.h5".format(algo)
print("File path: " + file_path)
with h5py.File(file_path, 'w') as hf:
hf.create_dataset('rs_test_acc', data=self.rs_test_acc)
hf.create_dataset('rs_test_acc_std', data=self.rs_test_acc_std)
hf.create_dataset('rs_test_auc', data=self.rs_test_auc)
hf.create_dataset('rs_train_loss', data=self.rs_train_loss)
hf.create_dataset('diff_pro', data=self.diff_pro) | 6,193 | Python | .py | 125 | 38.416 | 136 | 0.596259 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,835 | serverbabu-checkpoint.py | hkgdifyu_pFedPT/system/flcore/servers/.ipynb_checkpoints/serverbabu-checkpoint.py | from flcore.clients.clientbabu import clientBABU
from flcore.servers.serverbase import Server
from threading import Thread
class FedBABU(Server):
def __init__(self, args, times):
super().__init__(args, times)
# select slow clients
self.set_slow_clients()
self.set_clients(args, clientBABU)
print(f"\nJoin ratio / total clients: {self.join_ratio} / {self.num_clients}")
print("Finished creating server and clients.")
# self.load_model()
def train(self):
for i in range(self.global_rounds+1):
self.selected_clients = self.select_clients()
self.send_models()
if i%self.eval_gap == 0:
print(f"\n-------------Round number: {i}-------------")
print("\nEvaluate global model")
self.evaluate()
for client in self.selected_clients:
client.train()
# threads = [Thread(target=client.train)
# for client in self.selected_clients]
# [t.start() for t in threads]
# [t.join() for t in threads]
self.receive_models()
self.aggregate_parameters()
print("\nBest global accuracy.")
# self.print_(max(self.rs_test_acc), max(
# self.rs_train_acc), min(self.rs_train_loss))
print(max(self.rs_test_acc))
for client in self.clients:
client.fine_tune()
print("\n-------------Evaluate fine-tuned model-------------")
self.evaluate()
self.save_results()
self.save_global_model()
def receive_models(self):
assert (len(self.selected_clients) > 0)
self.uploaded_weights = []
tot_samples = 0
self.uploaded_ids = []
self.uploaded_models = []
for client in self.selected_clients:
self.uploaded_weights.append(client.train_samples)
tot_samples += client.train_samples
self.uploaded_ids.append(client.id)
self.uploaded_models.append(client.model.base)
for i, w in enumerate(self.uploaded_weights):
self.uploaded_weights[i] = w / tot_samples
| 2,197 | Python | .py | 51 | 32.72549 | 86 | 0.578675 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,836 | serverfedpt-checkpoint.py | hkgdifyu_pFedPT/system/flcore/servers/.ipynb_checkpoints/serverfedpt-checkpoint.py | from system.flcore.clients.clientt import clientT
from system.flcore.servers.serverbase import Server
from threading import Thread
import time
import torch
import os
import h5py
import copy
class FedPT(Server):
def __init__(self, args, times):
super().__init__(args, times)
# select slow clients
self.set_slow_clients()
self.set_clients(args, clientT)
self.global_model = copy.deepcopy(args.model.base)
self.diff_pro = []
self.clients_diverge = []
print(f"\nJoin ratio / total clients: {self.join_ratio} / {self.num_clients}")
print("Finished creating server and clients.")
# self.load_model()
self.Budget = []
def train(self):
for i in range(self.global_rounds+1):
s_t = time.time()
self.selected_clients = self.select_clients()
self.send_models()
if i%self.eval_gap == 0:
print(f"\n-------------Round number: {i}-------------")
print("\nEvaluate global model")
self.evaluate()
temp_diff_pro = 0
for client in self.selected_clients:
temp_diff_pro_client = client.train()
temp_diff_pro = temp_diff_pro +temp_diff_pro_client.item()
diverge_clents =0
for new_param, old_param in zip(self.clients[0].model.generator.parameters(), self.clients[1].model.generator.parameters()):
diff_pro = new_param - old_param
diff_pro = torch.where(diff_pro > 0, diff_pro, torch.zeros_like(diff_pro) - diff_pro)
diverge_clents = diverge_clents+torch.sum(diff_pro)
print("Averaged prompr difference: {:.4f}".format(temp_diff_pro))
print("0 and 1 clients difference: {:.4f}".format(diverge_clents.item()))
self.diff_pro.append(temp_diff_pro)
self.clients_diverge.append(diverge_clents.item())
# threads = [Thread(target=client.train)
# for client in self.selected_clients]
# [t.start() for t in threads]
# [t.join() for t in threads]
self.receive_models()
self.aggregate_parameters()
self.Budget.append(time.time() - s_t)
print('-'*25, 'time cost', '-'*25, self.Budget[-1])
print("\nBest global accuracy.")
# self.print_(max(self.rs_test_acc), max(
# self.rs_train_acc), min(self.rs_train_loss))
print(max(self.rs_test_acc))
print("\nAverage time cost per round.")
print(sum(self.Budget[1:])/len(self.Budget[1:]))
self.save_results()
self.save_global_model()
self.save_client_model()
def receive_models(self):
assert (len(self.selected_clients) > 0)
active_train_samples = 0
for client in self.selected_clients:
active_train_samples += client.train_samples
self.uploaded_weights = []
self.uploaded_ids = []
self.uploaded_models = []
for client in self.selected_clients:
self.uploaded_weights.append(client.train_samples / active_train_samples)
self.uploaded_ids.append(client.id)
self.uploaded_models.append(copy.deepcopy(client.model.base))
def save_client_model(self):
model_path = os.path.join("models", self.dataset,"client")
if not os.path.exists(model_path):
os.makedirs(model_path)
for c_idx,c in enumerate(self.clients):
model_path_save = os.path.join(model_path, self.algorithm + "_server" +str(c_idx)+ ".pt")
torch.save(c.model, model_path_save)
def save_results(self):
algo = self.dataset + "_" + self.algorithm
result_path = "../results/"
if not os.path.exists(result_path):
os.makedirs(result_path)
if (len(self.rs_test_acc)):
algo = algo + "_" + self.goal + "_" + str(self.times)
file_path = result_path + "{}.h5".format(algo)
print("File path: " + file_path)
with h5py.File(file_path, 'w') as hf:
hf.create_dataset('rs_test_acc', data=self.rs_test_acc)
hf.create_dataset('rs_test_acc_std', data=self.rs_test_acc_std)
hf.create_dataset('rs_test_auc', data=self.rs_test_auc)
hf.create_dataset('rs_train_loss', data=self.rs_train_loss)
hf.create_dataset('diff_pro', data=self.diff_pro) | 4,505 | Python | .py | 94 | 37.212766 | 136 | 0.588838 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,837 | serverbase-checkpoint.py | hkgdifyu_pFedPT/system/flcore/servers/.ipynb_checkpoints/serverbase-checkpoint.py | import torch
import os
import numpy as np
import h5py
import copy
import time
import random
from utils.data_utils import read_client_data
class Server(object):
def __init__(self, args, times):
# Set up the main attributes
self.device = args.device
self.dataset = args.dataset
self.global_rounds = args.global_rounds
self.local_steps = args.local_steps
self.batch_size = args.batch_size
self.learning_rate = args.local_learning_rate
self.global_model = copy.deepcopy(args.model)
self.num_clients = args.num_clients
self.join_ratio = args.join_ratio
self.join_clients = int(self.num_clients * self.join_ratio)
self.algorithm = args.algorithm
self.time_select = args.time_select
self.goal = args.goal
self.time_threthold = args.time_threthold
self.save_folder_name = args.save_folder_name
self.num_prompt = args.num_prompt
self.plocal_steps = args.plocal_steps
self.top_cnt = 100
self.clients = []
self.selected_clients = []
self.train_slow_clients = []
self.send_slow_clients = []
self.uploaded_weights = []
self.uploaded_ids = []
self.uploaded_models = []
self.rs_test_acc = []
self.rs_test_acc_std = []
self.rs_test_auc = []
self.rs_train_loss = []
self.times = times
self.eval_gap = args.eval_gap
self.client_drop_rate = args.client_drop_rate
self.train_slow_rate = args.train_slow_rate
self.send_slow_rate = args.send_slow_rate
def set_clients(self, args, clientObj):
for i, train_slow, send_slow in zip(range(self.num_clients), self.train_slow_clients, self.send_slow_clients):
train_data = read_client_data(self.dataset, i, is_train=True)
test_data = read_client_data(self.dataset, i, is_train=False)
client = clientObj(args,
id=i,
train_samples=len(train_data),
test_samples=len(test_data),
train_slow=train_slow,
send_slow=send_slow)
self.clients.append(client)
# random select slow clients
def select_slow_clients(self, slow_rate):
slow_clients = [False for i in range(self.num_clients)]
idx = [i for i in range(self.num_clients)]
idx_ = np.random.choice(idx, int(slow_rate * self.num_clients))
for i in idx_:
slow_clients[i] = True
return slow_clients
def set_slow_clients(self):
self.train_slow_clients = self.select_slow_clients(
self.train_slow_rate)
self.send_slow_clients = self.select_slow_clients(
self.send_slow_rate)
def select_clients(self):
selected_clients = list(np.random.choice(self.clients, self.join_clients, replace=False))
return selected_clients
def send_models(self):
assert (len(self.selected_clients) > 0)
for client in self.selected_clients:
client.set_parameters(self.global_model)
def receive_models(self):
assert (len(self.selected_clients) > 0)
self.uploaded_weights = []
tot_samples = 0
self.uploaded_ids = []
self.uploaded_models = []
for client in self.selected_clients:
self.uploaded_weights.append(client.train_samples)
tot_samples += client.train_samples
self.uploaded_ids.append(client.id)
self.uploaded_models.append(client.model)
for i, w in enumerate(self.uploaded_weights):
self.uploaded_weights[i] = w / tot_samples
def aggregate_parameters(self):
assert (len(self.uploaded_models) > 0)
self.global_model = copy.deepcopy(self.uploaded_models[0])
for param in self.global_model.parameters():
param.data.zero_()
for w, client_model in zip(self.uploaded_weights, self.uploaded_models):
self.add_parameters(w, client_model)
def add_parameters(self, w, client_model):
for server_param, client_param in zip(self.global_model.parameters(), client_model.parameters()):
server_param.data += client_param.data.clone() * w
def save_global_model(self):
model_path = os.path.join("models", self.dataset)
if not os.path.exists(model_path):
os.makedirs(model_path)
model_path = os.path.join(model_path, self.algorithm + "_server"+ "_" + str(self.num_prompt) + "_" + str(self.join_ratio) + "_" + str(self.num_clients)+ "_" + str(self.plocal_steps) + "_" + str(self.global_rounds) + ".pt")
torch.save(self.global_model, model_path)
def load_model(self):
model_path = os.path.join("models", self.dataset)
model_path = os.path.join(model_path, self.algorithm + "_server" + "_" + str(self.num_prompt) + "_" + str(self.join_ratio) + "_" + str(self.num_clients)+ "_" + str(self.plocal_steps) + "_" + str(self.global_rounds) +".pt")
assert (os.path.exists(model_path))
self.global_model = torch.load(model_path)
def model_exists(self):
model_path = os.path.join("models", self.dataset)
model_path = os.path.join(model_path, self.algorithm + "_" + str(self.num_prompt) + "_" + str(self.join_ratio) + "_" + str(self.num_clients)+ "_" + str(self.plocal_steps) + "_" + str(self.global_rounds)+".pt")
return os.path.exists(model_path)
def save_results(self):
algo = self.dataset + "_" + self.algorithm
result_path = "../results/"
if not os.path.exists(result_path):
os.makedirs(result_path)
if (len(self.rs_test_acc)):
algo = algo + "_" + self.goal + "_" + str(self.times)+ "_" + str(self.num_prompt) + "_" + str(self.join_ratio) + "_" + str(self.num_clients)+ "_" + str(self.plocal_steps) + "_" + str(self.global_rounds)
file_path = result_path + "{}.h5".format(algo)
print("File path: " + file_path)
with h5py.File(file_path, 'w') as hf:
hf.create_dataset('rs_test_acc', data=self.rs_test_acc)
hf.create_dataset('rs_test_acc_std', data=self.rs_test_acc_std)
hf.create_dataset('rs_test_auc', data=self.rs_test_auc)
hf.create_dataset('rs_train_loss', data=self.rs_train_loss)
def save_item(self, item, item_name):
if not os.path.exists(self.save_folder_name):
os.makedirs(self.save_folder_name)
torch.save(item, os.path.join(self.save_folder_name, "server_" + item_name + "_" + str(self.num_prompt) + "_" + str(self.join_ratio) + "_" + str(self.num_clients)+ "_" + str(self.plocal_steps) + "_" + str(self.global_rounds) + ".pt"))
def load_item(self, item_name):
return torch.load(os.path.join(self.save_folder_name, "server_" + item_name + "_" + str(self.num_prompt) + "_" + str(self.join_ratio) + "_" + str(self.num_clients)+ "_" + str(self.plocal_steps) + "_" + str(self.global_rounds) + ".pt"))
def test_metrics(self):
num_samples = []
tot_correct = []
tot_auc = []
for c in self.clients:
ct, ns, auc = c.test_metrics()
tot_correct.append(ct*1.0)
tot_auc.append(auc*ns)
num_samples.append(ns)
ids = [c.id for c in self.clients]
return ids, num_samples, tot_correct, tot_auc
def train_metrics(self):
num_samples = []
losses = []
for c in self.clients:
cl, ns = c.train_metrics()
num_samples.append(ns)
losses.append(cl*1.0)
ids = [c.id for c in self.clients]
return ids, num_samples, losses
# evaluate selected clients
def evaluate(self, acc=None, loss=None):
stats = self.test_metrics()
stats_train = self.train_metrics()
test_acc = sum(stats[2])*1.0 / sum(stats[1])
test_auc = sum(stats[3])*1.0 / sum(stats[1])
train_loss = sum(stats_train[2])*1.0 / sum(stats_train[1])
accs = [a / n for a, n in zip(stats[2], stats[1])]
aucs = [a / n for a, n in zip(stats[3], stats[1])]
if acc == None:
self.rs_test_acc.append(test_acc)
else:
acc.append(test_acc)
self.rs_test_auc.append(test_auc)
self.rs_test_acc_std.append(np.std(accs))
if loss == None:
self.rs_train_loss.append(train_loss)
else:
loss.append(train_loss)
print("Averaged Train Loss: {:.4f}".format(train_loss))
print("Averaged Test Accurancy: {:.4f}".format(test_acc))
print("Averaged Test AUC: {:.4f}".format(test_auc))
# self.print_(test_acc, train_acc, train_loss)
print("Std Test Accurancy: {:.4f}".format(np.std(accs)))
print("Std Test AUC: {:.4f}".format(np.std(aucs)))
def print_(self, test_acc, test_auc, train_loss):
print("Average Test Accurancy: {:.4f}".format(test_acc))
print("Average Test AUC: {:.4f}".format(test_auc))
print("Average Train Loss: {:.4f}".format(train_loss))
def check_done(self, acc_lss, top_cnt=None, div_value=None):
for acc_ls in acc_lss:
if top_cnt != None and div_value != None:
find_top = len(acc_ls) - torch.topk(torch.tensor(acc_ls), 1).indices[0] > top_cnt
find_div = len(acc_ls) > 1 and np.std(acc_ls[-top_cnt:]) < div_value
if find_top and find_div:
pass
else:
return False
elif top_cnt != None:
find_top = len(acc_ls) - torch.topk(torch.tensor(acc_ls), 1).indices[0] > top_cnt
if find_top:
pass
else:
return False
elif div_value != None:
find_div = len(acc_ls) > 1 and np.std(acc_ls[-top_cnt:]) < div_value
if find_div:
pass
else:
return False
else:
raise NotImplementedError
return True
| 10,235 | Python | .py | 207 | 38.661836 | 243 | 0.584682 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,838 | serverpfedpt-checkpoint.py | hkgdifyu_pFedPT/system/flcore/servers/.ipynb_checkpoints/serverpfedpt-checkpoint.py | from system.flcore.clients.clientpt import clientPT
from system.flcore.servers.serverbase import Server
from threading import Thread
import time
import torch
import os
import h5py
import copy
class PFedPT(Server):
def __init__(self, args, times):
super().__init__(args, times)
# select slow clients
self.set_slow_clients()
self.set_clients(args, clientPT)
self.global_model = copy.deepcopy(args.model.base)
self.diff_pro = []
self.clients_diverge = []
print(f"\nJoin ratio / total clients: {self.join_ratio} / {self.num_clients}")
print("Finished creating server and clients.")
# self.load_model()
self.Budget = []
def train(self):
for i in range(self.global_rounds+1):
s_t = time.time()
self.selected_clients = self.select_clients()
self.send_models()
if i%self.eval_gap == 0:
print(f"\n-------------Round number: {i}-------------")
print("\nEvaluate global model")
self.evaluate()
temp_diff_pro = 0
for client in self.selected_clients:
temp_diff_pro_client = client.train()
temp_diff_pro = temp_diff_pro +temp_diff_pro_client.item()
print("Averaged prompr difference: {:.4f}".format(temp_diff_pro))
self.diff_pro.append(temp_diff_pro)
diverge_clents =0
for new_param, old_param in zip(self.clients[0].model.generator.parameters(), self.clients[1].model.generator.parameters()):
diff_pro = new_param - old_param
diff_pro = torch.where(diff_pro > 0, diff_pro, torch.zeros_like(diff_pro) - diff_pro)
diverge_clents = diverge_clents+torch.sum(diff_pro)
print("0 and 1 clients difference: {:.4f}".format(diverge_clents.item()))
self.clients_diverge.append(diverge_clents.item())
# threads = [Thread(target=client.train)
# for client in self.selected_clients]
# [t.start() for t in threads]
# [t.join() for t in threads]
self.receive_models()
self.aggregate_parameters()
self.Budget.append(time.time() - s_t)
print('-'*25, 'time cost', '-'*25, self.Budget[-1])
print("\nBest global accuracy.")
# self.print_(max(self.rs_test_acc), max(
# self.rs_train_acc), min(self.rs_train_loss))
print(max(self.rs_test_acc))
print("\nAverage time cost per round.")
print(sum(self.Budget[1:])/len(self.Budget[1:]))
self.save_results()
self.save_global_model()
self.save_client_model()
def receive_models(self):
assert (len(self.selected_clients) > 0)
active_train_samples = 0
for client in self.selected_clients:
active_train_samples += client.train_samples
self.uploaded_weights = []
self.uploaded_ids = []
self.uploaded_models = []
for client in self.selected_clients:
self.uploaded_weights.append(client.train_samples / active_train_samples)
self.uploaded_ids.append(client.id)
self.uploaded_models.append(copy.deepcopy(client.model.base))
def save_client_model(self):
model_path = os.path.join("models", self.dataset,"client")
if not os.path.exists(model_path):
os.makedirs(model_path)
for c_idx,c in enumerate(self.clients):
model_path_save = os.path.join(model_path, self.algorithm + "_client" +str(c_idx)+ "_" + str(self.num_prompt) + "_" + str(self.join_ratio) + "_" + str(self.num_clients)+ "_" + str(self.plocal_steps) + "_" + str(self.global_rounds)+ ".pt")
torch.save(c.model, model_path_save)
def save_results(self):
algo = self.dataset + "_" + self.algorithm
result_path = "../results/"
if not os.path.exists(result_path):
os.makedirs(result_path)
if (len(self.rs_test_acc)):
algo = algo + "_" + self.goal + "_" + str(self.times)+ "_" + str(self.num_prompt) + "_" + str(self.join_ratio) + "_" + str(self.num_clients)+ "_" + str(self.plocal_steps) + "_" + str(self.global_rounds)
file_path = result_path + "{}.h5".format(algo)
print("File path: " + file_path)
with h5py.File(file_path, 'w') as hf:
hf.create_dataset('rs_test_acc', data=self.rs_test_acc)
hf.create_dataset('rs_test_acc_std', data=self.rs_test_acc_std)
hf.create_dataset('rs_test_auc', data=self.rs_test_auc)
hf.create_dataset('rs_train_loss', data=self.rs_train_loss)
hf.create_dataset('diff_pro', data=self.diff_pro)
hf.create_dataset('clients_diverge', data=self.clients_diverge) | 4,886 | Python | .py | 95 | 40.663158 | 250 | 0.589394 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,839 | serverreppt-checkpoint.py | hkgdifyu_pFedPT/system/flcore/servers/.ipynb_checkpoints/serverreppt-checkpoint.py | from system.flcore.clients.clientreppt import clientREPPT
from system.flcore.servers.serverbase import Server
from threading import Thread
import time
import copy
import torch
import os
import h5py
import copy
class PFedRepPT(Server):
def __init__(self, args, times):
super().__init__(args, times)
# select slow clients
self.set_slow_clients()
self.set_clients(args, clientREPPT)
self.global_model = copy.deepcopy(args.model.base)
self.diff_pro = []
print(f"\nJoin ratio / total clients: {self.join_ratio} / {self.num_clients}")
print("Finished creating server and clients.")
# self.load_model()
self.Budget = []
def train(self):
for i in range(self.global_rounds+1):
s_t = time.time()
self.selected_clients = self.select_clients()
self.send_models()
if i%self.eval_gap == 0:
print(f"\n-------------Round number: {i}-------------")
print("\nEvaluate global model")
self.evaluate()
temp_diff_pro = 0
for client in self.selected_clients:
temp_diff_pro_client = client.train()
temp_diff_pro = temp_diff_pro + temp_diff_pro_client.item()
print("Averaged prompr difference: {:.4f}".format(temp_diff_pro))
self.diff_pro.append(temp_diff_pro)
# threads = [Thread(target=client.train)
# for client in self.selected_clients]
# [t.start() for t in threads]
# [t.join() for t in threads]
self.receive_models()
self.aggregate_parameters()
self.Budget.append(time.time() - s_t)
print('-'*25, 'time cost', '-'*25, self.Budget[-1])
print("\nBest global accuracy.")
# self.print_(max(self.rs_test_acc), max(
# self.rs_train_acc), min(self.rs_train_loss))
print(max(self.rs_test_acc))
print("\nAverage time cost per round.")
print(sum(self.Budget[1:])/len(self.Budget[1:]))
self.save_results()
self.save_global_model()
self.save_client_model()
def receive_models(self):
assert (len(self.selected_clients) > 0)
active_train_samples = 0
for client in self.selected_clients:
active_train_samples += client.train_samples
self.uploaded_weights = []
self.uploaded_ids = []
self.uploaded_models = []
for client in self.selected_clients:
self.uploaded_weights.append(client.train_samples / active_train_samples)
self.uploaded_ids.append(client.id)
self.uploaded_models.append(copy.deepcopy(client.model.base))
def save_client_model(self):
model_path = os.path.join("models", self.dataset,"client")
if not os.path.exists(model_path):
os.makedirs(model_path)
for c_idx,c in enumerate(self.clients):
model_path_save = os.path.join(model_path, self.algorithm + "_server" +str(c_idx)+ ".pt")
torch.save(c.model, model_path_save)
def save_results(self):
algo = self.dataset + "_" + self.algorithm
result_path = "../results/"
if not os.path.exists(result_path):
os.makedirs(result_path)
if (len(self.rs_test_acc)):
algo = algo + "_" + self.goal + "_" + str(self.times)
file_path = result_path + "{}.h5".format(algo)
print("File path: " + file_path)
with h5py.File(file_path, 'w') as hf:
hf.create_dataset('rs_test_acc', data=self.rs_test_acc)
hf.create_dataset('rs_test_acc_std', data=self.rs_test_acc_std)
hf.create_dataset('rs_test_auc', data=self.rs_test_auc)
hf.create_dataset('rs_train_loss', data=self.rs_train_loss)
hf.create_dataset('diff_pro', data=self.diff_pro) | 3,962 | Python | .py | 87 | 35.275862 | 101 | 0.587503 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,840 | serverrep-checkpoint.py | hkgdifyu_pFedPT/system/flcore/servers/.ipynb_checkpoints/serverrep-checkpoint.py | from system.flcore.clients.clientrep import clientRep
from system.flcore.servers.serverbase import Server
from threading import Thread
import time
import copy
class FedRep(Server):
def __init__(self, args, times):
super().__init__(args, times)
# select slow clients
self.set_slow_clients()
self.set_clients(args, clientRep)
print(f"\nJoin ratio / total clients: {self.join_ratio} / {self.num_clients}")
print("Finished creating server and clients.")
# self.load_model()
self.Budget = []
def train(self):
for i in range(self.global_rounds+1):
s_t = time.time()
self.selected_clients = self.select_clients()
self.send_models()
if i%self.eval_gap == 0:
print(f"\n-------------Round number: {i}-------------")
print("\nEvaluate global model")
self.evaluate()
for client in self.selected_clients:
client.train()
# threads = [Thread(target=client.train)
# for client in self.selected_clients]
# [t.start() for t in threads]
# [t.join() for t in threads]
self.receive_models()
self.aggregate_parameters()
self.Budget.append(time.time() - s_t)
print('-'*25, 'time cost', '-'*25, self.Budget[-1])
print("\nBest global accuracy.")
# self.print_(max(self.rs_test_acc), max(
# self.rs_train_acc), min(self.rs_train_loss))
print(max(self.rs_test_acc))
print("\nAverage time cost per round.")
print(sum(self.Budget[1:])/len(self.Budget[1:]))
self.save_results()
self.save_global_model()
def receive_models(self):
assert (len(self.selected_clients) > 0)
active_train_samples = 0
for client in self.selected_clients:
active_train_samples += client.train_samples
self.uploaded_weights = []
self.uploaded_ids = []
self.uploaded_models = []
for client in self.selected_clients:
self.uploaded_weights.append(client.train_samples / active_train_samples)
self.uploaded_ids.append(client.id)
self.uploaded_models.append(copy.deepcopy(client.model.base)) | 2,334 | Python | .py | 54 | 33.185185 | 86 | 0.587456 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,841 | data_utils.py | hkgdifyu_pFedPT/system/utils/data_utils.py | import ujson
import numpy as np
import os
import torch
# IMAGE_SIZE = 28
# IMAGE_PIXELS = IMAGE_SIZE * IMAGE_SIZE
# NUM_CHANNELS = 1
# IMAGE_SIZE_CIFAR = 32
# NUM_CHANNELS_CIFAR = 3
def batch_data(data, batch_size):
'''
data is a dict := {'x': [numpy array], 'y': [numpy array]} (on one client)
returns x, y, which are both numpy array of length: batch_size
'''
data_x = data['x']
data_y = data['y']
# randomly shuffle data
ran_state = np.random.get_state()
np.random.shuffle(data_x)
np.random.set_state(ran_state)
np.random.shuffle(data_y)
# loop through mini-batches
for i in range(0, len(data_x), batch_size):
batched_x = data_x[i:i+batch_size]
batched_y = data_y[i:i+batch_size]
yield (batched_x, batched_y)
def get_random_batch_sample(data_x, data_y, batch_size):
num_parts = len(data_x)//batch_size + 1
if(len(data_x) > batch_size):
batch_idx = np.random.choice(list(range(num_parts + 1)))
sample_index = batch_idx*batch_size
if(sample_index + batch_size > len(data_x)):
return (data_x[sample_index:], data_y[sample_index:])
else:
return (data_x[sample_index: sample_index+batch_size], data_y[sample_index: sample_index+batch_size])
else:
return (data_x, data_y)
def get_batch_sample(data, batch_size):
data_x = data['x']
data_y = data['y']
# np.random.seed(100)
ran_state = np.random.get_state()
np.random.shuffle(data_x)
np.random.set_state(ran_state)
np.random.shuffle(data_y)
batched_x = data_x[0:batch_size]
batched_y = data_y[0:batch_size]
return (batched_x, batched_y)
def read_data(dataset, idx, args,is_train=True):
if is_train:
train_data_dir = os.path.join('../dataset', dataset, args.arv1+"*"+args.arv2+"*"+args.arv3+"*"+args.arv4+"*"+args.arv5+"*"+args.arv6,'train/')
train_file = train_data_dir + str(idx) + '.npz'
with open(train_file, 'rb') as f:
train_data = np.load(f, allow_pickle=True)['data'].tolist()
return train_data
else:
test_data_dir = os.path.join('../dataset', dataset, args.arv1+"*"+args.arv2+"*"+args.arv3+"*"+args.arv4+"*"+args.arv5+"*"+args.arv6,'test/')
test_file = test_data_dir + str(idx) + '.npz'
with open(test_file, 'rb') as f:
test_data = np.load(f, allow_pickle=True)['data'].tolist()
return test_data
def read_client_data(dataset, idx, args,is_train=True):
if dataset[:2] == "ag" or dataset[:2] == "SS":
return read_client_data_text(dataset, idx)
if is_train:
train_data = read_data(dataset, idx, args,is_train)
X_train = torch.Tensor(train_data['x']).type(torch.float32)
y_train = torch.Tensor(train_data['y']).type(torch.int64)
train_data = [(x, y) for x, y in zip(X_train, y_train)]
return train_data
else:
test_data = read_data(dataset, idx, args,is_train)
X_test = torch.Tensor(test_data['x']).type(torch.float32)
y_test = torch.Tensor(test_data['y']).type(torch.int64)
test_data = [(x, y) for x, y in zip(X_test, y_test)]
return test_data
def read_client_data_text(dataset, idx, is_train=True):
if is_train:
train_data = read_data(dataset, idx, is_train)
X_train, X_train_lens = list(zip(*train_data['x']))
y_train = train_data['y']
X_train = torch.Tensor(X_train).type(torch.int64)
X_train_lens = torch.Tensor(X_train_lens).type(torch.int64)
y_train = torch.Tensor(train_data['y']).type(torch.int64)
train_data = [((x, lens), y) for x, lens, y in zip(X_train, X_train_lens, y_train)]
return train_data
else:
test_data = read_data(dataset, idx, is_train)
X_test, X_test_lens = list(zip(*test_data['x']))
y_test = test_data['y']
X_test = torch.Tensor(X_test).type(torch.int64)
X_test_lens = torch.Tensor(X_test_lens).type(torch.int64)
y_test = torch.Tensor(test_data['y']).type(torch.int64)
test_data = [((x, lens), y) for x, lens, y in zip(X_test, X_test_lens, y_test)]
return test_data
| 4,194 | Python | .py | 95 | 37.326316 | 150 | 0.614742 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,842 | mem_utils.py | hkgdifyu_pFedPT/system/utils/mem_utils.py | import math
import gc
from collections import defaultdict
from typing import Optional, Tuple, List
import torch
from math import isnan
from calmsize import size as calmsize
def readable_size(num_bytes: int) -> str:
return '' if isnan(num_bytes) else '{:.2f}'.format(calmsize(num_bytes))
LEN = 79
# some pytorch low-level memory management constant
# the minimal allocate memory size (Byte)
PYTORCH_MIN_ALLOCATE = 2 ** 9
# the minimal cache memory size (Byte)
PYTORCH_MIN_CACHE = 2 ** 20
class MemReporter():
"""A memory reporter that collects tensors and memory usages
Parameters:
- model: an extra nn.Module can be passed to infer the name
of Tensors
"""
def __init__(self, model: Optional[torch.nn.Module] = None):
self.tensor_name = {}
self.device_mapping = defaultdict(list)
self.device_tensor_stat = {}
# to numbering the unknown tensors
self.name_idx = 0
tensor_names = defaultdict(list)
if model is not None:
assert isinstance(model, torch.nn.Module)
# for model with tying weight, multiple parameters may share
# the same underlying tensor
for name, param in model.named_parameters():
tensor_names[param].append(name)
for param, name in tensor_names.items():
self.tensor_name[id(param)] = '+'.join(name)
def _get_tensor_name(self, tensor: torch.Tensor) -> str:
tensor_id = id(tensor)
if tensor_id in self.tensor_name:
name = self.tensor_name[tensor_id]
# use numbering if no name can be inferred
else:
name = type(tensor).__name__ + str(self.name_idx)
self.tensor_name[tensor_id] = name
self.name_idx += 1
return name
def collect_tensor(self):
"""Collect all tensor objects tracked by python
NOTICE:
- the buffers for backward which is implemented in C++ are
not tracked by python's reference counting.
- the gradients(.grad) of Parameters is not collected, and
I don't know why.
"""
#FIXME: make the grad tensor collected by gc
objects = gc.get_objects()
tensors = [obj for obj in objects if isinstance(obj, torch.Tensor)]
for t in tensors:
self.device_mapping[t.device].append(t)
def get_stats(self):
"""Get the memory stat of tensors and then release them
As a memory profiler, we cannot hold the reference to any tensors, which
causes possibly inaccurate memory usage stats, so we delete the tensors after
getting required stats"""
visited_data = {}
self.device_tensor_stat.clear()
def get_tensor_stat(tensor: torch.Tensor) -> List[Tuple[str, int, int, int]]:
"""Get the stat of a single tensor
Returns:
- stat: a tuple containing (tensor_name, tensor_size,
tensor_numel, tensor_memory)
"""
assert isinstance(tensor, torch.Tensor)
name = self._get_tensor_name(tensor)
if tensor.is_sparse:
indices_stat = get_tensor_stat(tensor._indices())
values_stat = get_tensor_stat(tensor._values())
return indices_stat + values_stat
numel = tensor.numel()
element_size = tensor.element_size()
fact_numel = tensor.storage().size()
fact_memory_size = fact_numel * element_size
# since pytorch allocate at least 512 Bytes for any tensor, round
# up to a multiple of 512
memory_size = math.ceil(fact_memory_size / PYTORCH_MIN_ALLOCATE) \
* PYTORCH_MIN_ALLOCATE
# tensor.storage should be the actual object related to memory
# allocation
data_ptr = tensor.storage().data_ptr()
if data_ptr in visited_data:
name = '{}(->{})'.format(
name,
visited_data[data_ptr],
)
# don't count the memory for reusing same underlying storage
memory_size = 0
else:
visited_data[data_ptr] = name
size = tuple(tensor.size())
# torch scalar has empty size
if not size:
size = (1,)
return [(name, size, numel, memory_size)]
for device, tensors in self.device_mapping.items():
tensor_stats = []
for tensor in tensors:
if tensor.numel() == 0:
continue
stat = get_tensor_stat(tensor) # (name, shape, numel, memory_size)
tensor_stats += stat
if isinstance(tensor, torch.nn.Parameter):
if tensor.grad is not None:
# manually specify the name of gradient tensor
self.tensor_name[id(tensor.grad)] = '{}.grad'.format(
self._get_tensor_name(tensor)
)
stat = get_tensor_stat(tensor.grad)
tensor_stats += stat
self.device_tensor_stat[device] = tensor_stats
self.device_mapping.clear()
def print_stats(self, verbose: bool = False, target_device: Optional[torch.device] = None) -> None:
# header
# show_reuse = verbose
# template_format = '{:<40s}{:>20s}{:>10s}'
# print(template_format.format('Element type', 'Size', 'Used MEM') )
for device, tensor_stats in self.device_tensor_stat.items():
# By default, if the target_device is not specified,
# print tensors on all devices
if target_device is not None and device != target_device:
continue
# print('-' * LEN)
print('\nStorage on {}'.format(device))
total_mem = 0
total_numel = 0
for stat in tensor_stats:
name, size, numel, mem = stat
# if not show_reuse:
# name = name.split('(')[0]
# print(template_format.format(
# str(name),
# str(size),
# readable_size(mem),
# ))
total_mem += mem
total_numel += numel
print('-'*LEN)
print('Total Tensors: {} \tUsed Memory: {}'.format(
total_numel, readable_size(total_mem),
))
if device != torch.device('cpu'):
with torch.cuda.device(device):
memory_allocated = torch.cuda.memory_allocated()
print('The allocated memory on {}: {}'.format(
device, readable_size(memory_allocated),
))
if memory_allocated != total_mem:
print('Memory differs due to the matrix alignment or'
' invisible gradient buffer tensors')
print('-'*LEN)
def report(self, verbose: bool = False, device: Optional[torch.device] = None) -> None:
"""Interface for end-users to directly print the memory usage
args:
- verbose: flag to show tensor.storage reuse information
- device: `torch.device` object, specify the target device
to report detailed memory usage. It will print memory usage
on all devices if not specified. Usually we only want to
print the memory usage on CUDA devices.
"""
self.collect_tensor()
self.get_stats()
self.print_stats(verbose, target_device=device) | 7,749 | Python | .py | 171 | 33 | 103 | 0.565805 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,843 | privacy.py | hkgdifyu_pFedPT/system/utils/privacy.py | from opacus import PrivacyEngine
from opacus.dp_model_inspector import DPModelInspector
MAX_GRAD_NORM = 1.0
EPSILON = 50.0
DELTA = 1e-7
EPOCHS = 1 * 100
N_ACCUMULATION_STEPS = 4
def initialize_dp(model, optimizer, sample_rate, dp_sigma):
privacy_engine = PrivacyEngine(
model,
sample_rate = sample_rate * N_ACCUMULATION_STEPS,
# epochs = EPOCHS,
# target_epsilon = EPSILON,
target_delta = DELTA,
noise_multiplier = dp_sigma,
max_grad_norm = MAX_GRAD_NORM,
)
privacy_engine.attach(optimizer)
def get_dp_params(optimizer):
return optimizer.privacy_engine.get_privacy_spent(DELTA), DELTA
def check_dp(model):
inspector = DPModelInspector()
inspector.validate(model)
def dp_step(optimizer, i, len_train_loader):
# take a real optimizer step after N_VIRTUAL_STEP steps t
if ((i + 1) % N_ACCUMULATION_STEPS == 0) or ((i + 1) == len_train_loader):
optimizer.step()
else:
optimizer.virtual_step() # take a virtual step | 1,027 | Python | .py | 29 | 30.413793 | 78 | 0.688194 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,844 | result_utils.py | hkgdifyu_pFedPT/system/utils/result_utils.py | import h5py
import numpy as np
import os
def average_data(algorithm="", dataset="", goal="", times=10, length=800):
test_acc = get_all_results_for_one_algo(
algorithm, dataset, goal, times, int(length))
test_acc_data = np.average(test_acc, axis=0)
max_accurancy = []
for i in range(times):
max_accurancy.append(test_acc[i].max())
print("std for best accurancy:", np.std(max_accurancy))
print("mean for best accurancy:", np.mean(max_accurancy))
def get_all_results_for_one_algo(algorithm="", dataset="", goal="", times=10, length=800):
test_acc = np.zeros((times, length))
algorithms_list = [algorithm] * times
for i in range(times):
file_name = dataset + "_" + \
algorithms_list[i] + "_" + goal + "_" + str(i)
test_acc[i, :] = np.array(
read_data_then_delete(file_name, delete=False))[:length]
return test_acc
def read_data_then_delete(file_name, delete=False):
file_path = "../results/" + file_name + ".h5"
with h5py.File(file_path, 'r') as hf:
rs_test_acc = np.array(hf.get('rs_test_acc'))
if delete:
os.remove(file_path)
print("Length: ", len(rs_test_acc))
return rs_test_acc | 1,224 | Python | .py | 29 | 36.206897 | 90 | 0.62775 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,845 | data_utils.cpython-38.pyc | hkgdifyu_pFedPT/system/utils/__pycache__/data_utils.cpython-38.pyc | U
?şcb ã @ sZ d dl Z d dlZd dlZd dlZdd„ Zdd„ Zdd„ Zdd d
„Zdd d „Z dd
d„Z
dS )é Nc c s‚ | d }| d }t j ¡ }t j |¡ t j |¡ t j |¡ tdt|ƒ|ƒD ].}|||| … }|||| … }||fV qNdS )z—
data is a dict := {'x': [numpy array], 'y': [numpy array]} (on one client)
returns x, y, which are both numpy array of length: batch_size
ÚxÚyr N)ÚnpÚrandomÚ get_stateÚshuffleÚ set_stateÚrangeÚlen)ÚdataÚ
batch_sizeÚdata_xÚdata_yÚ ran_stateÚiÚ batched_xÚ batched_y© r õY D:\京东\promot\第二次投稿\å®�验\æœ�务器\native - pro\system\utils\data_utils.pyÚ
batch_data s
r c C s’ t | ƒ| d }t | ƒ|kr†tj tt|d ƒƒ¡}|| }|| t | ƒkrd| |d … ||d … fS | ||| … |||| … fS n| |fS d S )Né )r
r r ÚchoiceÚlistr )r
r r Z num_partsÚ batch_idxÚ sample_indexr r r Úget_random_batch_sample# s "r c C s^ | d }| d }t j ¡ }t j |¡ t j |¡ t j |¡ |d|… }|d|… }||fS )Nr r r )r r r r r )r r r
r r r r r r r Úget_batch_sample0 s
r Tc C s |rŠt j d| |jd |j d |j d |j d |j d |j d¡}|t |ƒ d }t
|dƒ� }t j |dd�d
¡ }W 5 Q R X |S t j d| |jd |j d |j d |j d |j d |j d ¡}|t |ƒ d } t
| dƒ� }t j |dd�d
¡ }
W 5 Q R X |
S d S )
Nz
../datasetÚ*ztrain/z.npzÚrbT)Ú allow_pickler ztest/)ÚosÚpathÚjoinZarv1Zarv2Zarv3Zarv4Zarv5Zarv6ÚstrÚopenr ÚloadÚtolist) ÚdatasetÚidxÚargsÚis_trainZtrain_data_dirZ
train_fileÚfÚ
train_dataZ
test_data_dirÚ test_fileÚ test_datar r r Ú read_data? s F F r/ c
C sÖ | d d… dks | d d… dkr*t | |ƒS |r€t| |||ƒ}t |d ¡ tj¡}t |d ¡ tj¡}dd„ t||ƒD ƒ}|S t| |||ƒ}t |d ¡ tj¡}t |d ¡ tj¡} dd„ t|| ƒD ƒ}|S d S ) Né ÚagZSSr r c S s g | ]\}}||f‘qS r r ©Ú.0r r r r r Ú
<listcomp>\ s z$read_client_data.<locals>.<listcomp>c S s g | ]\}}||f‘qS r r r2 r r r r4 b s )Úread_client_data_textr/ ÚtorchÚTensorÚtypeÚfloat32Úint64Úzip)
r' r( r) r* r, ÚX_trainÚy_trainr. ÚX_testÚy_testr r r Úread_client_dataS s
r@ c C s |r€t | ||ƒ}tt|d � ƒ\}}|d }t |¡ tj¡}t |¡ tj¡}t |d ¡ tj¡}dd„ t|||ƒD ƒ}|S t | ||ƒ}tt|d � ƒ\}} |d }
t |¡ tj¡}t | ¡ tj¡} t |d ¡ tj¡}
dd„ t|| |
ƒD ƒ}|S d S )Nr r c S s g | ]\}}}||f|f‘qS r r ©r3 r Úlensr r r r r4 p s z)read_client_data_text.<locals>.<listcomp>c S s g | ]\}}}||f|f‘qS r r rA r r r r4 { s )r/ r r; r6 r7 r8 r: ) r' r( r* r, r< Z X_train_lensr= r. r> Z X_test_lensr? r r r r5 f s" r5 )T)T)T) ZujsonÚnumpyr r r6 r r r r/ r@ r5 r r r r Ú<module> s
| 3,614 | Python | .py | 30 | 119.033333 | 601 | 0.35258 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,846 | data_utils.cpython-37.pyc | hkgdifyu_pFedPT/system/utils/__pycache__/data_utils.cpython-37.pyc | B
Ê:cc® ã @ sZ d dl Z d dlZd dlZd dlZdd„ Zdd„ Zdd„ Zdd d
„Zdd d „Z dd
d„Z
dS )é Nc c s† | d }| d }t j ¡ }t j |¡ t j |¡ t j |¡ xBtdt|ƒ|ƒD ].}|||| … }|||| … }||fV qPW dS )z—
data is a dict := {'x': [numpy array], 'y': [numpy array]} (on one client)
returns x, y, which are both numpy array of length: batch_size
ÚxÚyr N)ÚnpÚrandomÚ get_stateÚshuffleÚ set_stateÚrangeÚlen)ÚdataÚ
batch_sizeÚdata_xÚdata_yÚ ran_stateÚiÚ batched_xÚ batched_y© r ú>/root/autodl-tmp/PFL-Non-IID-master/system/utils/data_utils.pyÚ
batch_data s
r c C s’ t | ƒ| d }t | ƒ|kr†tj tt|d ƒƒ¡}|| }|| t | ƒkrd| |d … ||d … fS | ||| … |||| … fS n| |fS d S )Né )r
r r ÚchoiceÚlistr )r
r r Z num_partsÚ batch_idxÚ sample_indexr r r Úget_random_batch_sample# s "r c C s^ | d }| d }t j ¡ }t j |¡ t j |¡ t j |¡ |d|… }|d|… }||fS )Nr r r )r r r r r )r r r
r r r r r r r Úget_batch_sample0 s
r Tc
C s¨ |rTt j d| d¡}|t|ƒ d }t|dƒ� }tj|dd�d ¡ }W d Q R X |S t j d| d¡}|t|ƒ d }t|dƒ� }tj|dd�d ¡ } W d Q R X | S d S ) Nz
../datasetztrain/z.npzÚrbT)Ú allow_pickler ztest/)ÚosÚpathÚjoinÚstrÚopenr ÚloadÚtolist)
ÚdatasetÚidxÚis_trainZtrain_data_dirZ
train_fileÚfÚ
train_dataZ
test_data_dirÚ test_fileÚ test_datar r r Ú read_data? s r- c C sÒ | d d… dks | d d… dkr*t | |ƒS |r~t| ||ƒ}t |d ¡ tj¡}t |d ¡ tj¡}dd„ t||ƒD ƒ}|S t| ||ƒ}t |d ¡ tj¡}t |d ¡ tj¡}dd„ t||ƒD ƒ}|S d S ) Né ÚagZSSr r c S s g | ]\}}||f‘qS r r )Ú.0r r r r r ú
<listcomp>\ s z$read_client_data.<locals>.<listcomp>c S s g | ]\}}||f‘qS r r )r0 r r r r r r1 b s )Úread_client_data_textr- ÚtorchÚTensorÚtypeÚfloat32Úint64Úzip) r& r' r( r* ÚX_trainÚy_trainr, ÚX_testÚy_testr r r Úread_client_dataS s
r= c C s |r€t | ||ƒ}tt|d � ƒ\}}|d }t |¡ tj¡}t |¡ tj¡}t |d ¡ tj¡}dd„ t|||ƒD ƒ}|S t | ||ƒ}tt|d � ƒ\}} |d }
t |¡ tj¡}t | ¡ tj¡} t |d ¡ tj¡}
dd„ t|| |
ƒD ƒ}|S d S )Nr r c S s g | ]\}}}||f|f‘qS r r )r0 r Úlensr r r r r1 p s z)read_client_data_text.<locals>.<listcomp>c S s g | ]\}}}||f|f‘qS r r )r0 r r> r r r r r1 { s )r- r r8 r3 r4 r5 r7 ) r& r' r( r* r9 Z X_train_lensr: r, r; Z X_test_lensr< r r r r2 f s" r2 )T)T)T) ZujsonÚnumpyr r r3 r r r r- r= r2 r r r r Ú<module> s
| 3,403 | Python | .py | 25 | 134.6 | 606 | 0.356318 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,847 | mem_utils.cpython-37.pyc | hkgdifyu_pFedPT/system/utils/__pycache__/mem_utils.cpython-37.pyc | B
Ë:ccE ã @ s~ d dl Z d dlZd dlmZ d dlmZmZmZ d dlZd dl m Z d dl
m Z
e e
dœdd„Zd Zd
Zd ZG d d
„ d
ƒZdS )é N)Ú defaultdict)ÚOptionalÚTupleÚList)Úisnan)Úsize)Ú num_bytesÚreturnc C s t | ƒr dS d t| ƒ¡S )NÚ z{:.2f})r ÚformatÚcalmsize)r © r
ú=/root/autodl-tmp/PFL-Non-IID-master/system/utils/mem_utils.pyÚ
readable_size s r éO i i c @ s~ e Zd ZdZdeejj dœdd„Zej e
dœdd„Z d d
„ Z d d „ Z
deeej ddœdd„Zdeeej ddœdd„ZdS )Ú MemReporterz§A memory reporter that collects tensors and memory usages
Parameters:
- model: an extra nn.Module can be passed to infer the name
of Tensors
N)Úmodelc C s� i | _ ttƒ| _i | _d| _ttƒ}|d k rbt|tjj ƒs>t
‚x"| ¡ D ]\}}|| |¡ qHW x(|
¡ D ] \}}d |¡| j t|ƒ< qlW d S )Nr ú+)Ú tensor_namer ÚlistÚdevice_mappingÚdevice_tensor_statÚname_idxÚ
isinstanceÚtorchÚnnÚModuleÚAssertionErrorÚnamed_parametersÚappendÚitemsÚjoinÚid)Úselfr Ú tensor_namesÚnameÚparamr
r
r Ú__init__ s
zMemReporter.__init__)Útensorr c C sN t |ƒ}|| jkr | j| }n,t|ƒjt| jƒ }|| j|< | jd7 _|S )Né )r" r ÚtypeÚ__name__Ústrr )r# r( Z tensor_idr% r
r
r Ú_get_tensor_name0 s
z MemReporter._get_tensor_namec C s: t ¡ }dd„ |D ƒ}x |D ]}| j|j |¡ q W dS )a* Collect all tensor objects tracked by python
NOTICE:
- the buffers for backward which is implemented in C++ are
not tracked by python's reference counting.
- the gradients(.grad) of Parameters is not collected, and
I don't know why.
c S s g | ]}t |tjƒr|‘qS r
)r r ÚTensor)Ú.0Úobjr
r
r ú
<listcomp>F s z.MemReporter.collect_tensor.<locals>.<listcomp>N)ÚgcÚ get_objectsr Údevicer )r# ÚobjectsÚtensorsÚtr
r
r Úcollect_tensor; s
zMemReporter.collect_tensorc sÞ i ‰ˆj ¡ tjttttttf dœ‡ ‡‡fdd„ ‰ x–ˆj ¡ D ]ˆ\}}g }xp|D ]h}|
¡ dkrhqVˆ |ƒ}||7 }t |tj j
ƒrV|jdk rVd ˆ |¡¡ˆjt|jƒ< ˆ |jƒ}||7 }qVW |ˆj |< qDW ˆj ¡ dS )zûGet the memory stat of tensors and then release them
As a memory profiler, we cannot hold the reference to any tensors, which
causes possibly inaccurate memory usage stats, so we delete the tensors after
getting required stats)r( r c sÊ t | tjƒst‚ˆ | ¡}| jr@ˆ | ¡ ƒ}ˆ | ¡ ƒ}|| S | ¡ }| ¡ }|
¡ ¡ }|| }t
|t ¡t }|
¡ ¡ } | ˆkr d |ˆ| ¡}d}n|ˆ| < t| ¡ ƒ}
|
s¼d}
||
||fgS )z±Get the stat of a single tensor
Returns:
- stat: a tuple containing (tensor_name, tensor_size,
tensor_numel, tensor_memory)
z{}(->{})r )r) )r r r. r r- Ú is_sparseÚ_indicesÚ_valuesÚnumelÚ element_sizeÚstorager ÚmathÚceilÚPYTORCH_MIN_ALLOCATEÚdata_ptrr Útuple) r( r% Z indices_statZ values_statr< r= Z
fact_numelZfact_memory_sizeZ memory_sizerB r )Úget_tensor_statr# Ú visited_datar
r rD S s.
z.MemReporter.get_stats.<locals>.get_tensor_statr Nz{}.grad)r Úclearr r. r r r, Úintr r r< r r Ú ParameterÚgradr r- r r" )r# r4 r6 Ú tensor_statsr( Ústatr
)rD r# rE r Ú get_statsJ s$
*,
zMemReporter.get_statsF)ÚverboseÚ
target_devicer c
C sî xè| j ¡ D ]Ú\}}|d k r&||kr&q td |¡ƒ d}d}x(|D ] }|\}} }
} || 7 }||
7 }qBW tdt ƒ td |t|ƒ¡ƒ |t d¡krÚtj |¡� tj ¡ } W d Q R X td |t| ƒ¡ƒ | |krÚtdƒ tdt ƒ q W d S )Nz
Storage on {}r ú-z"Total Tensors: {} Used Memory: {}Úcpuz The allocated memory on {}: {}zOMemory differs due to the matrix alignment or invisible gradient buffer tensors)
r r Úprintr ÚLENr r r4 ÚcudaÚmemory_allocated)
r# rM rN r4 rJ Z total_memÚ total_numelrK r% r r< ÚmemrT r
r
r Ú print_stats” s*
zMemReporter.print_stats)rM r4 r c C s" | ¡ | ¡ | j||d� dS )a Interface for end-users to directly print the memory usage
args:
- verbose: flag to show tensor.storage reuse information
- device: `torch.device` object, specify the target device
to report detailed memory usage. It will print memory usage
on all devices if not specified. Usually we only want to
print the memory usage on CUDA devices.
)rN N)r8 rL rW )r# rM r4 r
r
r Úreport¾ s zMemReporter.report)N)FN)FN)r+ Ú
__module__Ú __qualname__Ú__doc__r r r r r' r. r, r- r8 rL Úboolr4 rW rX r
r
r
r r s J*r )r? r2 Ú collectionsr Útypingr r r r r r r rG r, r rR rA ZPYTORCH_MIN_CACHEr r
r
r
r Ú<module> s | 5,571 | Python | .py | 60 | 87.666667 | 390 | 0.482107 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,848 | data_utils.cpython-39.pyc | hkgdifyu_pFedPT/system/utils/__pycache__/data_utils.cpython-39.pyc | a
f¾`c® ã @ sZ d dl Z d dlZd dlZd dlZdd„ Zdd„ Zdd„ Zdd d
„Zdd d „Z dd
d„Z
dS )é Nc c s‚ | d }| d }t j ¡ }t j |¡ t j |¡ t j |¡ tdt|ƒ|ƒD ].}|||| … }|||| … }||fV qNdS )z—
data is a dict := {'x': [numpy array], 'y': [numpy array]} (on one client)
returns x, y, which are both numpy array of length: batch_size
ÚxÚyr N)ÚnpÚrandomÚ get_stateÚshuffleÚ set_stateÚrangeÚlen)ÚdataÚ
batch_sizeÚdata_xÚdata_yÚ ran_stateÚiÚ batched_xÚ batched_y© r új/media/sim812/391e55df-b6f2-4fe9-a920-53434a8506fa/lgh/pdept/PFL-Non-IID-master/system/utils/data_utils.pyÚ
batch_data s
r c C s’ t | ƒ| d }t | ƒ|kr†tj tt|d ƒƒ¡}|| }|| t | ƒkrd| |d … ||d … fS | ||| … |||| … fS n| |fS d S )Né )r
r r ÚchoiceÚlistr )r
r r Z num_partsÚ batch_idxÚ sample_indexr r r Úget_random_batch_sample# s "r c C s^ | d }| d }t j ¡ }t j |¡ t j |¡ t j |¡ |d|… }|d|… }||fS )Nr r r )r r r r r )r r r
r r r r r r r Úget_batch_sample0 s
r Tc
C sĞ |rht j d| d¡}|t|ƒ d }t|dƒ�&}tj|dd�d ¡ }W d ƒ n1 sZ0 Y |S t j d| d¡}|t|ƒ d }t|dƒ�&}tj|dd�d ¡ } W d ƒ n1 s¾0 Y | S d S ) Nz
../datasetztrain/z.npzÚrbT)Ú allow_pickler ztest/)ÚosÚpathÚjoinÚstrÚopenr ÚloadÚtolist)
ÚdatasetÚidxÚis_trainZtrain_data_dirZ
train_fileÚfÚ
train_dataZ
test_data_dirÚ test_fileÚ test_datar r r Ú read_data? s 4 4r- c C sÒ | d d… dks | d d… dkr*t | |ƒS |r~t| ||ƒ}t |d ¡ tj¡}t |d ¡ tj¡}dd„ t||ƒD ƒ}|S t| ||ƒ}t |d ¡ tj¡}t |d ¡ tj¡}dd„ t||ƒD ƒ}|S d S ) Né ÚagZSSr r c S s g | ]\}}||f‘qS r r ©Ú.0r r r r r Ú
<listcomp>\ ó z$read_client_data.<locals>.<listcomp>c S s g | ]\}}||f‘qS r r r0 r r r r2 b r3 )Úread_client_data_textr- ÚtorchÚTensorÚtypeÚfloat32Úint64Úzip) r& r' r( r* ÚX_trainÚy_trainr, ÚX_testÚy_testr r r Úread_client_dataS s
r? c C s |r€t | ||ƒ}tt|d � ƒ\}}|d }t |¡ tj¡}t |¡ tj¡}t |d ¡ tj¡}dd„ t|||ƒD ƒ}|S t | ||ƒ}tt|d � ƒ\}} |d }
t |¡ tj¡}t | ¡ tj¡} t |d ¡ tj¡}
dd„ t|| |
ƒD ƒ}|S d S )Nr r c S s g | ]\}}}||f|f‘qS r r ©r1 r Úlensr r r r r2 p r3 z)read_client_data_text.<locals>.<listcomp>c S s g | ]\}}}||f|f‘qS r r r@ r r r r2 { r3 )r- r r: r5 r6 r7 r9 ) r& r' r( r* r; Z X_train_lensr< r, r= Z X_test_lensr> r r r r4 f s" r4 )T)T)T) ZujsonÚnumpyr r r5 r r r r- r? r4 r r r r Ú<module> s
| 3,490 | Python | .py | 25 | 138.08 | 593 | 0.359492 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,849 | result_utils.cpython-37.pyc | hkgdifyu_pFedPT/system/utils/__pycache__/result_utils.cpython-37.pyc | B
Ë:ccÇ ã @ s: d dl Z d dlZd dlZd dd„Zd
dd„Zdd
d „ZdS )é NÚ é
é c C sn t | |||t|ƒƒ}tj|dd�}g }x"t|ƒD ]}| || ¡ ¡ q0W tdt |¡ƒ tdt |¡ƒ d S )Nr )Úaxiszstd for best accurancy:zmean for best accurancy:)
Ú get_all_results_for_one_algoÚintÚnpÚaverageÚrangeÚappendÚmaxÚprintÚstdÚmean) Ú algorithmÚdatasetÚgoalÚtimesÚlengthÚtest_accZ
test_acc_dataZ
max_accurancyÚi© r ú@/root/autodl-tmp/PFL-Non-IID-master/system/utils/result_utils.pyÚ average_data s r c C sx t ||f¡}| g| }xZt|ƒD ]N}|d || d | d t|ƒ }t t|dd�¡d |… ||d d …f< q"W |S )NÚ_F)Údelete)r Úzerosr
ÚstrÚarrayÚread_data_then_delete) r r r r r r Zalgorithms_listr Ú file_namer r r r s
$&r Fc C sT d| d }t |d¡�}t | d¡¡}W d Q R X |rBt |¡ tdt|ƒƒ |S )Nz ../results/z.h5ÚrÚ rs_test_acczLength: ) Úh5pyÚFiler r ÚgetÚosÚremover
Úlen)r r Ú file_pathÚhfr" r r r r s
r )r r r r r )r r r r r )F)r# Únumpyr r& r r r r r r r Ú<module> s
| 1,378 | Python | .py | 10 | 136.5 | 467 | 0.42575 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,850 | result_utils.cpython-38.pyc | hkgdifyu_pFedPT/system/utils/__pycache__/result_utils.cpython-38.pyc | U
çYgcÈ ã @ s: d dl Z d dlZd dlZd dd„Zd
dd„Zdd
d „ZdS )é NÚ é
é c C sj t | |||t|ƒƒ}tj|dd�}g }t|ƒD ]}| || ¡ ¡ q.tdt |¡ƒ tdt |¡ƒ d S )Nr )Úaxiszstd for best accurancy:zmean for best accurancy:)
Ú get_all_results_for_one_algoÚintÚnpÚaverageÚrangeÚappendÚmaxÚprintÚstdÚmean) Ú algorithmÚdatasetÚgoalÚtimesÚlengthÚtest_accZ
test_acc_dataZ
max_accurancyÚi© r õQ D:\京东\promot\第二次投稿\å®�验\native - pro\system\utils\result_utils.pyÚ average_data s ÿ r c C st t ||f¡}| g| }t|ƒD ]N}|d || d | d t|ƒ }t t|dd�¡d |… ||d d …f< q |S )NÚ_F)Údelete)r Úzerosr
ÚstrÚarrayÚread_data_then_delete) r r r r r r Zalgorithms_listr Ú file_namer r r r s(
ÿÿÿÿÿ
ÿÿr Fc C sT d| d }t |d¡�}t | d¡¡}W 5 Q R X |rBt |¡ tdt|ƒƒ |S )Nz ../results/z.h5ÚrÚ rs_test_acczLength: ) Úh5pyÚFiler r ÚgetÚosÚremover
Úlen)r r Ú file_pathÚhfr" r r r r ! s
r )r r r r r )r r r r r )F)r# Únumpyr r& r r r r r r r Ú<module> s
| 1,442 | Python | .py | 11 | 129.818182 | 497 | 0.412587 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,851 | privacy.cpython-38.pyc | hkgdifyu_pFedPT/system/utils/__pycache__/privacy.cpython-38.pyc | U
e¾`c ã @ sP d dl mZ d dlmZ dZdZdZdZdZdd „ Z d
d „ Z
d d
„ Z dd„ Z dS )é )Ú
PrivacyEngine)ÚDPModelInspectorg ğ?g I@gH¯¼šò×z>éd é c C s$ t | |t t|td�}| |¡ d S )N)Ú sample_rateZ target_deltaZnoise_multiplierZ
max_grad_norm)r ÚN_ACCUMULATION_STEPSÚDELTAÚ
MAX_GRAD_NORMÚattach)ÚmodelÚ optimizerr Údp_sigmaÚprivacy_engine© r õN D:\京东\promot\PFL-Non-IID-master\PFL-Non-IID-master\system\utils\privacy.pyÚ
initialize_dp
s ù r c C s | j t¡tfS ©N)r Zget_privacy_spentr )r r r r Ú
get_dp_params s r c C s t ƒ }| | ¡ d S r )r Úvalidate)r Z inspectorr r r Úcheck_dp s r c C s2 |d t dks |d |kr&| ¡ n| ¡ d S )Né r )r ÚstepZ virtual_step)r ÚiZlen_train_loaderr r r Údp_step s
r N)
Zopacusr Zopacus.dp_model_inspectorr r ZEPSILONr ZEPOCHSr r r r r r r r r Ú<module> s
| 1,200 | Python | .py | 6 | 198.833333 | 480 | 0.445188 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,852 | mem_utils.cpython-38.pyc | hkgdifyu_pFedPT/system/utils/__pycache__/mem_utils.cpython-38.pyc | U
–jfcE ã @ s~ d dl Z d dlZd dlmZ d dlmZmZmZ d dlZd dl m Z d dl
m Z
e e
dœdd„Zd Zd
Zd ZG d d
„ d
ƒZdS )é N)Ú defaultdict)ÚOptionalÚTupleÚList)Úisnan)Úsize)Ú num_bytesÚreturnc C s t | ƒr dS d t| ƒ¡S )NÚ z{:.2f})r ÚformatÚcalmsize)r © r
õB D:\京东\promot\cifar\cifar\Cifar10_iid\system\utils\mem_utils.pyÚ
readable_size s r éO i i c @ s~ e Zd ZdZdeejj dœdd„Zej e
dœdd„Z d d
„ Z d d „ Z
deeej ddœdd„Zdeeej ddœdd„ZdS )Ú MemReporterz§A memory reporter that collects tensors and memory usages
Parameters:
- model: an extra nn.Module can be passed to infer the name
of Tensors
N)Úmodelc C sˆ i | _ ttƒ| _i | _d| _ttƒ}|d k r^t|tjj ƒs>t
‚| ¡ D ]\}}|| |¡ qF|
¡ D ] \}}d |¡| j t|ƒ< qfd S )Nr ú+)Ú tensor_namer ÚlistÚdevice_mappingÚdevice_tensor_statÚname_idxÚ
isinstanceÚtorchÚnnÚModuleÚAssertionErrorÚnamed_parametersÚappendÚitemsÚjoinÚid)Úselfr Ú tensor_namesÚnameÚparamr
r
r Ú__init__ s
zMemReporter.__init__©Útensorr c C sN t |ƒ}|| jkr | j| }n,t|ƒjt| jƒ }|| j|< | jd7 _|S )Né )r" r ÚtypeÚ__name__Ústrr )r# r) Z tensor_idr% r
r
r Ú_get_tensor_name0 s
z MemReporter._get_tensor_namec C s6 t ¡ }dd„ |D ƒ}|D ]}| j|j |¡ qdS )a* Collect all tensor objects tracked by python
NOTICE:
- the buffers for backward which is implemented in C++ are
not tracked by python's reference counting.
- the gradients(.grad) of Parameters is not collected, and
I don't know why.
c S s g | ]}t |tjƒr|‘qS r
)r r ÚTensor)Ú.0Úobjr
r
r Ú
<listcomp>F s z.MemReporter.collect_tensor.<locals>.<listcomp>N)ÚgcÚ get_objectsr Údevicer )r# ÚobjectsÚtensorsÚtr
r
r Úcollect_tensor; s
zMemReporter.collect_tensorc sÖ i ‰ˆj ¡ tjttttttf dœ‡ ‡‡fdd„ ‰ ˆj ¡ D ]„\}}g }|D ]h}|
¡ dkrdqRˆ |ƒ}||7 }t |tj j
ƒrR|jdk rRd ˆ |¡¡ˆjt|jƒ< ˆ |jƒ}||7 }qR|ˆj |< qBˆj ¡ dS )zûGet the memory stat of tensors and then release them
As a memory profiler, we cannot hold the reference to any tensors, which
causes possibly inaccurate memory usage stats, so we delete the tensors after
getting required statsr( c sÊ t | tjƒst‚ˆ | ¡}| jr@ˆ | ¡ ƒ}ˆ | ¡ ƒ}|| S | ¡ }| ¡ }|
¡ ¡ }|| }t
|t ¡t }|
¡ ¡ } | ˆkr d |ˆ| ¡}d}n|ˆ| < t| ¡ ƒ}
|
s¼d}
||
||fgS )z±Get the stat of a single tensor
Returns:
- stat: a tuple containing (tensor_name, tensor_size,
tensor_numel, tensor_memory)
z{}(->{})r )r* )r r r/ r r. Ú is_sparseÚ_indicesÚ_valuesÚnumelÚ element_sizeÚstorager ÚmathÚceilÚPYTORCH_MIN_ALLOCATEÚdata_ptrr Útuple) r) r% Z indices_statZ values_statr= r> Z
fact_numelZfact_memory_sizeZ memory_sizerC r ©Úget_tensor_statr# Z visited_datar
r rF S s2
ÿ ş z.MemReporter.get_stats.<locals>.get_tensor_statr Nz{}.grad)r Úclearr r/ r r r- Úintr r r= r r Ú ParameterÚgradr r. r r" )r# r5 r7 Ú tensor_statsr) Ústatr
rE r Ú get_statsJ s&
*,
ÿ
zMemReporter.get_statsF)ÚverboseÚ
target_devicer c
C sæ | j ¡ D ]Ö\}}|d k r$||kr$q
td |¡ƒ d}d}|D ] }|\}} }
} || 7 }||
7 }q>tdt ƒ td |t|ƒ¡ƒ |t d¡krÔtj |¡� tj ¡ } W 5 Q R X td |t| ƒ¡ƒ | |krÔtdƒ tdt ƒ q
d S )Nz
Storage on {}r ú-z"Total Tensors: {} Used Memory: {}Úcpuz The allocated memory on {}: {}zOMemory differs due to the matrix alignment or invisible gradient buffer tensors)
r r Úprintr ÚLENr r r5 ÚcudaÚmemory_allocated)
r# rN rO r5 rK Z total_memÚ total_numelrL r% r r= ÚmemrU r
r
r Ú print_stats” s2
ÿ ÿzMemReporter.print_stats)rN r5 r c C s" | ¡ | ¡ | j||d� dS )a Interface for end-users to directly print the memory usage
args:
- verbose: flag to show tensor.storage reuse information
- device: `torch.device` object, specify the target device
to report detailed memory usage. It will print memory usage
on all devices if not specified. Usually we only want to
print the memory usage on CUDA devices.
)rO N)r9 rM rX )r# rN r5 r
r
r Úreport¾ s zMemReporter.report)N)FN)FN)r, Ú
__module__Ú __qualname__Ú__doc__r r r r r' r/ r- r. r9 rM Úboolr5 rX rY r
r
r
r r s J*r )r@ r3 Ú collectionsr Útypingr r r r r r r rH r- r rS rB ZPYTORCH_MIN_CACHEr r
r
r
r Ú<module> s | 5,594 | Python | .py | 60 | 88.05 | 406 | 0.478292 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,853 | generate_tiny_imagenet.py | hkgdifyu_pFedPT/dataset/generate_tiny_imagenet.py | import numpy as np
import os
import sys
import random
import torch
import torchvision
import torchvision.transforms as transforms
from utils.dataset_utils import check, separate_data, split_data, save_file
from torchvision.datasets import ImageFolder, DatasetFolder
random.seed(1)
np.random.seed(1)
num_clients = int(sys.argv[4])
num_classes = 200
dir_path = "Tiny-imagenet/"+sys.argv[1]+"*"+sys.argv[2]+"*"+sys.argv[3]+"*"+sys.argv[4]+"*"+sys.argv[5]+"*"+sys.argv[6]+"/"
dir_path2 = "Tiny-imagenet/"
# http://cs231n.stanford.edu/tiny-imagenet-200.zip
# https://github.com/QinbinLi/MOON/blob/6c7a4ed1b1a8c0724fa2976292a667a828e3ff5d/datasets.py#L148
class ImageFolder_custom(DatasetFolder):
def __init__(self, root, dataidxs=None, train=True, transform=None, target_transform=None):
self.root = root
self.dataidxs = dataidxs
self.train = train
self.transform = transform
self.target_transform = target_transform
imagefolder_obj = ImageFolder(self.root, self.transform, self.target_transform)
self.loader = imagefolder_obj.loader
if self.dataidxs is not None:
self.samples = np.array(imagefolder_obj.samples)[self.dataidxs]
else:
self.samples = np.array(imagefolder_obj.samples)
def __getitem__(self, index):
path = self.samples[index][0]
target = self.samples[index][1]
target = int(target)
sample = self.loader(path)
if self.transform is not None:
sample = self.transform(sample)
if self.target_transform is not None:
target = self.target_transform(target)
return sample, target
def __len__(self):
if self.dataidxs is None:
return len(self.samples)
else:
return len(self.dataidxs)
# Allocate data to users
def generate_dataset(dir_path, num_clients, num_classes, niid, balance, partition,class_per_client,alpha):
if not os.path.exists(dir_path):
os.makedirs(dir_path)
# Setup directory for train/test data
config_path = dir_path + "config.json"
train_path = dir_path + "train/"
test_path = dir_path + "test/"
if check(config_path, train_path, test_path, num_clients, num_classes, niid, balance, partition,alpha):
return
# Get data
transform = transforms.Compose(
[transforms.ToTensor(), transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5))])
trainset = ImageFolder_custom(root=dir_path2+'rawdata/tiny-imagenet-200/train/', transform=transform)
testset = ImageFolder_custom(root=dir_path2+'rawdata/tiny-imagenet-200/val/', transform=transform)
trainloader = torch.utils.data.DataLoader(
trainset, batch_size=len(trainset), shuffle=False)
testloader = torch.utils.data.DataLoader(
testset, batch_size=len(testset), shuffle=False)
for _, train_data in enumerate(trainloader, 0):
trainset.data, trainset.targets = train_data
for _, test_data in enumerate(testloader, 0):
testset.data, testset.targets = test_data
dataset_image = []
dataset_label = []
dataset_image.extend(trainset.data.cpu().detach().numpy())
dataset_image.extend(testset.data.cpu().detach().numpy())
dataset_label.extend(trainset.targets.cpu().detach().numpy())
dataset_label.extend(testset.targets.cpu().detach().numpy())
dataset_image = np.array(dataset_image)
dataset_label = np.array(dataset_label)
# dataset = []
# for i in range(num_classes):
# idx = dataset_label == i
# dataset.append(dataset_image[idx])
X, y, statistic = separate_data((dataset_image, dataset_label), num_clients, num_classes,
niid, balance, partition, class_per_client=class_per_client,alpha = alpha)
train_data, test_data = split_data(X, y)
save_file(config_path, train_path, test_path, train_data, test_data, num_clients, num_classes,
statistic, niid, balance, partition,alpha)
if __name__ == "__main__":
niid = True if sys.argv[1] == "noniid" else False
balance = True if sys.argv[2] == "balance" else False
partition = sys.argv[3] if sys.argv[3] != "-" else None
alpha = float(sys.argv[5])
class_per_client = int(sys.argv[6])
generate_dataset(dir_path, num_clients, num_classes, niid, balance, partition,class_per_client,alpha)
| 4,382 | Python | .py | 92 | 41.217391 | 123 | 0.680432 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,854 | generate_cifar10.py | hkgdifyu_pFedPT/dataset/generate_cifar10.py | import numpy as np
import os
import sys
import random
import torch
import torchvision
import torchvision.transforms as transforms
from utils.dataset_utils import check, separate_data, split_data, save_file
random.seed(1)
np.random.seed(1)
num_clients = int(sys.argv[4])
num_classes = 10
dir_path = "Cifar10/"+sys.argv[1]+"*"+sys.argv[2]+"*"+sys.argv[3]+"*"+sys.argv[4]+"*"+sys.argv[5]+"*"+sys.argv[6]+"/"
dir_path2 = "Cifar10/"
# Allocate data to users
def generate_cifar10(dir_path, num_clients, num_classes, niid, balance, partition,class_per_client,alpha):
if not os.path.exists(dir_path):
os.makedirs(dir_path)
# Setup directory for train/test data
config_path = dir_path + "config.json"
train_path = dir_path + "train/"
test_path = dir_path + "test/"
if check(config_path, train_path, test_path, num_clients, num_classes, niid, balance, partition,alpha):
return
# Get Cifar10 data
transform = transforms.Compose(
[transforms.ToTensor(), transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5))])
# transform = transforms.Compose(
# [transforms.ToTensor()])
trainset = torchvision.datasets.CIFAR10(
root=dir_path2+"rawdata", train=True, download=True, transform=transform)
testset = torchvision.datasets.CIFAR10(
root=dir_path2+"rawdata", train=False, download=True, transform=transform)
trainloader = torch.utils.data.DataLoader(
trainset, batch_size=len(trainset.data), shuffle=False)
testloader = torch.utils.data.DataLoader(
testset, batch_size=len(testset.data), shuffle=False)
for _, train_data in enumerate(trainloader, 0):
trainset.data, trainset.targets = train_data
for _, test_data in enumerate(testloader, 0):
testset.data, testset.targets = test_data
dataset_image = []
dataset_label = []
dataset_image.extend(trainset.data.cpu().detach().numpy())
dataset_image.extend(testset.data.cpu().detach().numpy())
dataset_label.extend(trainset.targets.cpu().detach().numpy())
dataset_label.extend(testset.targets.cpu().detach().numpy())
dataset_image = np.array(dataset_image)
dataset_label = np.array(dataset_label)
# dataset = []
# for i in range(num_classes):
# idx = dataset_label == i
# dataset.append(dataset_image[idx])
X, y, statistic = separate_data((dataset_image, dataset_label), num_clients, num_classes,
niid, balance, partition, class_per_client=class_per_client,alpha = alpha)
train_data, test_data = split_data(X, y)
save_file(config_path, train_path, test_path, train_data, test_data, num_clients, num_classes,
statistic, niid, balance, partition,alpha)
if __name__ == "__main__":
niid = True if sys.argv[1] == "noniid" else False
balance = True if sys.argv[2] == "balance" else False
partition = sys.argv[3] if sys.argv[3] != "-" else None
alpha = float(sys.argv[5])
class_per_client = int(sys.argv[6])
# niid = True
# balance = True
# partition = 'dir'
generate_cifar10(dir_path, num_clients, num_classes, niid, balance, partition,class_per_client,alpha) | 3,202 | Python | .py | 68 | 41.544118 | 117 | 0.679343 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,855 | generate_cifar100.py | hkgdifyu_pFedPT/dataset/generate_cifar100.py | import numpy as np
import os
import sys
import random
import torch
import torchvision
import torchvision.transforms as transforms
from utils.dataset_utils import check, separate_data, split_data, save_file
random.seed(1)
np.random.seed(1)
num_clients = int(sys.argv[4])
num_classes = 100
dir_path = "Cifar100/"+sys.argv[1]+"*"+sys.argv[2]+"*"+sys.argv[3]+"*"+sys.argv[4]+"*"+sys.argv[5]+"*"+sys.argv[6]+"/"
dir_path2 = "Cifar100/"
# Allocate data to users
def generate_cifar100(dir_path, num_clients, num_classes, niid, balance, partition,class_per_client,alpha):
if not os.path.exists(dir_path):
os.makedirs(dir_path)
# Setup directory for train/test data
config_path = dir_path + "config.json"
train_path = dir_path + "train/"
test_path = dir_path + "test/"
if check(config_path, train_path, test_path, num_clients, num_classes, niid, balance, partition,alpha):
return
# Get Cifar100 data
transform = transforms.Compose(
[transforms.ToTensor(), transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5))])
trainset = torchvision.datasets.CIFAR100(
root=dir_path2+"rawdata", train=True, download=True, transform=transform)
testset = torchvision.datasets.CIFAR100(
root=dir_path2+"rawdata", train=False, download=True, transform=transform)
trainloader = torch.utils.data.DataLoader(
trainset, batch_size=len(trainset.data), shuffle=False)
testloader = torch.utils.data.DataLoader(
testset, batch_size=len(testset.data), shuffle=False)
for _, train_data in enumerate(trainloader, 0):
trainset.data, trainset.targets = train_data
for _, test_data in enumerate(testloader, 0):
testset.data, testset.targets = test_data
dataset_image = []
dataset_label = []
dataset_image.extend(trainset.data.cpu().detach().numpy())
dataset_image.extend(testset.data.cpu().detach().numpy())
dataset_label.extend(trainset.targets.cpu().detach().numpy())
dataset_label.extend(testset.targets.cpu().detach().numpy())
dataset_image = np.array(dataset_image)
dataset_label = np.array(dataset_label)
# dataset = []
# for i in range(num_classes):
# idx = dataset_label == i
# dataset.append(dataset_image[idx])
X, y, statistic = separate_data((dataset_image, dataset_label), num_clients, num_classes,
niid, balance, partition, class_per_client=class_per_client,alpha = alpha)
train_data, test_data = split_data(X, y)
save_file(config_path, train_path, test_path, train_data, test_data, num_clients, num_classes,
statistic, niid, balance, partition,alpha)
if __name__ == "__main__":
niid = True if sys.argv[1] == "noniid" else False
balance = True if sys.argv[2] == "balance" else False
partition = sys.argv[3] if sys.argv[3] != "-" else None
alpha = float(sys.argv[5])
class_per_client = int(sys.argv[6])
generate_cifar100(dir_path, num_clients, num_classes, niid, balance, partition,class_per_client,alpha) | 3,074 | Python | .py | 63 | 43.206349 | 118 | 0.684899 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,856 | dataset_utils.py | hkgdifyu_pFedPT/dataset/utils/dataset_utils.py | import os
import ujson
import numpy as np
import gc
from sklearn.model_selection import train_test_split
batch_size = 2
train_size = 0.75 # merge original training set and test set, then split it manually.
least_samples = batch_size / (1-train_size) # least samples for each client
# for Dirichlet distribution
def check(config_path, train_path, test_path, num_clients, num_classes, niid=False,
balance=True, partition=None,alpha = 0.1):
# check existing dataset
if os.path.exists(config_path):
with open(config_path, 'r') as f:
config = ujson.load(f)
if config['num_clients'] == num_clients and \
config['num_classes'] == num_classes and \
config['non_iid'] == niid and \
config['balance'] == balance and \
config['partition'] == partition and \
config['alpha'] == alpha and \
config['batch_size'] == batch_size:
print("\nDataset already generated.\n")
return True
dir_path = os.path.dirname(train_path)
if not os.path.exists(dir_path):
os.makedirs(dir_path)
dir_path = os.path.dirname(test_path)
if not os.path.exists(dir_path):
os.makedirs(dir_path)
return False
def separate_data(data, num_clients, num_classes, niid=False, balance=False, partition=None, class_per_client=2,alpha = 0.1):
X = [[] for _ in range(num_clients)]
y = [[] for _ in range(num_clients)]
statistic = [[] for _ in range(num_clients)]
dataset_content, dataset_label = data
dataidx_map = {}
if not niid:
partition = 'pat'
class_per_client = num_classes
if partition == 'pat':
idxs = np.array(range(len(dataset_label)))
idx_for_each_class = []
for i in range(num_classes):
idx_for_each_class.append(idxs[dataset_label == i])
class_num_per_client = [class_per_client for _ in range(num_clients)]
for i in range(num_classes):
selected_clients = []
for client in range(num_clients):
if class_num_per_client[client] > 0:
selected_clients.append(client)
selected_clients = selected_clients[:int(num_clients/num_classes*class_per_client)]
num_all_samples = len(idx_for_each_class[i])
num_selected_clients = len(selected_clients)
num_per = num_all_samples / num_selected_clients
if balance:
num_samples = [int(num_per) for _ in range(num_selected_clients-1)]
else:
num_samples = np.random.randint(max(num_per/10, least_samples/num_classes), num_per, num_selected_clients-1).tolist()
num_samples.append(num_all_samples-sum(num_samples))
idx = 0
for client, num_sample in zip(selected_clients, num_samples):
if client not in dataidx_map.keys():
dataidx_map[client] = idx_for_each_class[i][idx:idx+num_sample]
else:
dataidx_map[client] = np.append(dataidx_map[client], idx_for_each_class[i][idx:idx+num_sample], axis=0)
idx += num_sample
class_num_per_client[client] -= 1
elif partition == "dir":
# https://github.com/IBM/probabilistic-federated-neural-matching/blob/master/experiment.py
min_size = 0
K = num_classes
N = len(dataset_label)
while min_size < least_samples:
idx_batch = [[] for _ in range(num_clients)]
for k in range(K):
idx_k = np.where(dataset_label == k)[0]
np.random.shuffle(idx_k)
proportions = np.random.dirichlet(np.repeat(alpha, num_clients))
proportions = np.array([p*(len(idx_j)<N/num_clients) for p,idx_j in zip(proportions,idx_batch)])
proportions = proportions/proportions.sum()
proportions = (np.cumsum(proportions)*len(idx_k)).astype(int)[:-1]
idx_batch = [idx_j + idx.tolist() for idx_j,idx in zip(idx_batch,np.split(idx_k,proportions))]
min_size = min([len(idx_j) for idx_j in idx_batch])
for j in range(num_clients):
dataidx_map[j] = idx_batch[j]
else:
raise NotImplementedError
# assign data
for client in range(num_clients):
idxs = dataidx_map[client]
X[client] = dataset_content[idxs]
y[client] = dataset_label[idxs]
for i in np.unique(y[client]):
statistic[client].append((int(i), int(sum(y[client]==i))))
del data
# gc.collect()
for client in range(num_clients):
print(f"Client {client}\t Size of data: {len(X[client])}\t Labels: ", np.unique(y[client]))
print(f"\t\t Samples of labels: ", [i for i in statistic[client]])
print("-" * 50)
return X, y, statistic
def split_data(X, y):
# Split dataset
train_data, test_data = [], []
num_samples = {'train':[], 'test':[]}
for i in range(len(y)):
unique, count = np.unique(y[i], return_counts=True)
if min(count) > 1:
X_train, X_test, y_train, y_test = train_test_split(
X[i], y[i], train_size=train_size, shuffle=True)
else:
X_train, X_test, y_train, y_test = train_test_split(
X[i], y[i], train_size=train_size, shuffle=True)
train_data.append({'x': X_train, 'y': y_train})
num_samples['train'].append(len(y_train))
test_data.append({'x': X_test, 'y': y_test})
num_samples['test'].append(len(y_test))
print("Total number of samples:", sum(num_samples['train'] + num_samples['test']))
print("The number of train samples:", num_samples['train'])
print("The number of test samples:", num_samples['test'])
print()
del X, y
# gc.collect()
return train_data, test_data
def save_file(config_path, train_path, test_path, train_data, test_data, num_clients,
num_classes, statistic, niid=False, balance=True, partition=None,alpha = 0.1):
config = {
'num_clients': num_clients,
'num_classes': num_classes,
'non_iid': niid,
'balance': balance,
'partition': partition,
'Size of samples for labels in clients': statistic,
'alpha': alpha,
'batch_size': batch_size,
}
# gc.collect()
print("Saving to disk.\n")
for idx, train_dict in enumerate(train_data):
with open(train_path + str(idx) + '.npz', 'wb') as f:
np.savez_compressed(f, data=train_dict)
for idx, test_dict in enumerate(test_data):
with open(test_path + str(idx) + '.npz', 'wb') as f:
np.savez_compressed(f, data=test_dict)
with open(config_path, 'w') as f:
ujson.dump(config, f)
print("Finish generating dataset.\n") | 6,882 | Python | .py | 148 | 36.925676 | 133 | 0.596533 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,857 | dataset_utils.cpython-38.pyc | hkgdifyu_pFedPT/dataset/utils/__pycache__/dataset_utils.cpython-38.pyc | U
)Qgcâ ã @ sj d dl Z d dlZd dlZd dlZd dlmZ dZdZede Z dd d
„Z
dd d „Z d
d„ Z ddd„Z
dS )é N)Útrain_test_splité g è?é FTçš™™™™™¹?c C sÔ t j | ¡rŒt| dƒ�} t | ¡}
W 5 Q R X |
d |krŒ|
d |krŒ|
d |krŒ|
d |krŒ|
d |krŒ|
d |krŒ|
d tkrŒtd ƒ d
S t j |¡} t j | ¡s®t | ¡ t j |¡} t j | ¡sĞt | ¡ d S ) NÚrÚ num_clientsÚ num_classesÚnon_iidÚbalanceÚ partitionÚalphaÚ
batch_sizez
Dataset already generated.
TF)
ÚosÚpathÚexistsÚopenÚujsonÚloadr
ÚprintÚdirnameÚmakedirs) Ú config_pathÚ
train_pathÚ test_pathr r Úniidr
r r ÚfÚconfigÚdir_path© r õM D:\京东\promot\第二次投稿\å®�验\native\dataset\utils\dataset_utils.pyÚcheck s2
ÿ
ş
ı
ü
û
ú
r c sÚ dd„ t ˆƒD ƒ}dd„ t ˆƒD ƒ} dd„ t ˆƒD ƒ}
| \} } i }
|sNd}|‰|dk�rğt t t| ƒƒ¡}g }t |ƒD ]}| || |k ¡ qv‡fdd„t ˆƒD ƒ}t |ƒD �]<}g }t ˆƒD ]2}|| dkrØ| |¡ |d tˆ| ˆ ƒ… }q¾t|| ƒ}t|ƒ}|| ‰|�r0‡fdd„t |d ƒD ƒ}n&tj tˆd
t | ƒˆ|d ¡
¡ }| |t |ƒ ¡ d}t ||ƒD ]r\}}||
¡ k�r¦|| ||| … |
|< n(tj|
| || ||| … dd �|
|< ||7 }|| d 8 < �qvq¬�n|d k�rd}|}t| ƒ‰ |t k �ræd
d„ t ˆƒD ƒ}t |ƒD ]²} t | | k¡d } tj | ¡ tj t |ˆ¡¡} t ‡ ‡fdd„t | |ƒD ƒ¡} | | ¡ } t | ¡t| ƒ t¡d d… } dd„ t |t | | ¡ƒD ƒ}tdd„ |D ƒƒ}�q.�q
t ˆƒD ]} || |
| < �qînt‚t ˆƒD ]b}|
| }| | ||< | | | |< t | | ¡D ],}|
| t|ƒtt | | |kƒƒf¡ �qB�q~ t ˆƒD ]P}td|› dt|| ƒ› d�t | | ¡ƒ tddd„ |
| D ƒƒ tdƒ �q~|| |
fS )Nc S s g | ]}g ‘qS r r ©Ú.0Ú_r r r Ú
<listcomp>& s z!separate_data.<locals>.<listcomp>c S s g | ]}g ‘qS r r r! r r r r$ ' s c S s g | ]}g ‘qS r r r! r r r r$ ( s Úpatc s g | ]}ˆ ‘qS r r r! )Úclass_per_clientr r r$ 8 s r c s g | ] }t ˆ ƒ‘qS r )Úintr! )Únum_perr r r$ D s r é
)ÚaxisÚdirc S s g | ]}g ‘qS r r r! r r r r$ Y s c s$ g | ] \}}|t |ƒˆ ˆ k ‘qS r ©Úlen)r" ÚpÚidx_j)ÚNr r r r$ ^ s éÿÿÿÿc S s g | ]\}}|| ¡ ‘qS r )Útolist)r" r/ Úidxr r r r$ a s c S s g | ] }t |ƒ‘qS r r, )r" r/ r r r r$ b s zClient z Size of data: z
Labels: z Samples of labels: c S s g | ]}|‘qS r r )r" Úir r r r$ x s z2--------------------------------------------------)ÚrangeÚnpÚarrayr- Úappendr' ÚrandomÚrandintÚmaxÚ
least_samplesr2 ÚsumÚzipÚkeysÚwhereÚshuffleÚ dirichletÚrepeatÚcumsumÚastypeÚsplitÚminÚNotImplementedErrorÚuniquer ) Údatar r r r
r r& r ÚXÚyÚ statisticZdataset_contentÚ
dataset_labelZ dataidx_mapÚidxsZidx_for_each_classr4 Zclass_num_per_clientZselected_clientsÚclientZnum_all_samplesZnum_selected_clientsÚ num_samplesr3 Z
num_sampleÚmin_sizeÚKZ idx_batchÚkZidx_kZ proportionsÚjr )r0 r& r r( r Ú
separate_data% s€
&(
. * rV c C s g g }}g g dœ}t t|ƒƒD ]¬}tj|| dd�\}}t|ƒdkrht| | || tdd�\}} }
} n t| | || tdd�\}} }
} | ||
dœ¡ |d t|
ƒ¡ | | | dœ¡ |d t| ƒ¡ q td t |d |d ƒƒ td
|d ƒ td |d ƒ tƒ ~ ~||fS ) N)ÚtrainÚtestT)Ú
return_countsr )Ú
train_sizerA )ÚxrL rW rX zTotal number of samples:z The number of train samples:zThe number of test samples:)
r5 r- r6 rI rG r rZ r8 r r= ) rK rL Ú
train_dataÚ test_datarQ r4 rI ÚcountZX_trainZX_testZy_trainZy_testr r r Ú
split_data~ s6
ÿ ÿr_ c
C sĞ |||| |
|| t dœ} tdƒ t|ƒD ]8\}
}t|t|
ƒ d dƒ�}tj||d� W 5 Q R X q&t|ƒD ]8\}
}t|t|
ƒ d dƒ�}tj||d� W 5 Q R X qht| dƒ�}t | |¡ W 5 Q R X tdƒ d S )N)r r r r
r z%Size of samples for labels in clientsr r
zSaving to disk.
z.npzÚwb)rJ ÚwzFinish generating dataset.
) r
r Ú enumerater Ústrr6 Úsavez_compressedr Údump)r r r r\ r] r r rM r r
r r r r3 Z
train_dictr Z test_dictr r r Ú save_fileš s& ø rf )FTNr )FFNr r )FTNr )r r Únumpyr6 ÚgcZsklearn.model_selectionr r
rZ r< r rV r_ rf r r r r Ú<module> s& ÿ
Y ÿ | 5,383 | Python | .py | 63 | 84.365079 | 504 | 0.367224 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,858 | dataset_utils.cpython-37.pyc | hkgdifyu_pFedPT/dataset/utils/__pycache__/dataset_utils.cpython-37.pyc | B
4ØccÊ ã @ sn d dl Z d dlZd dlZd dlZd dlmZ dZdZede Z dZ
dd d
„Z dd d „Z d
d„ Z
ddd„ZdS )é N)Útrain_test_splité g è?é gš™™™™™¹?FTc C sÔ t j | ¡rŒt| dƒ�}t |¡} W d Q R X | d |krŒ| d |krŒ| d |krŒ| d |krŒ| d |krŒ| d tkrŒ| d tkrŒtd ƒ d
S t j |¡}
t j |
¡s®t
|
¡ t j |¡}
t j |
¡sĞt
|
¡ d S ) NÚrÚ num_clientsÚ num_classesÚnon_iidÚbalanceÚ partitionÚalphaÚ
batch_sizez
Dataset already generated.
TF) ÚosÚpathÚexistsÚopenÚujsonÚloadr r ÚprintÚdirnameÚmakedirs) Ú config_pathÚ
train_pathÚ test_pathr r Úniidr r
ÚfÚconfigÚdir_path© r úB/root/autodl-tmp/PFL-Non-IID-master/dataset/utils/dataset_utils.pyÚcheck s&
r c s dd„ t ˆƒD ƒ}dd„ t ˆƒD ƒ}dd„ t ˆƒD ƒ} | \}
} i } |sNd}|‰|dk�rt t t| ƒƒ¡}
g }x"t |ƒD ]}| |
| |k ¡ qxW ‡fdd„t ˆƒD ƒ}�xzt |ƒD �]D}g }x>t ˆƒD ]2}|| dkrâ| |¡ |d tˆ| ˆ ƒ… }qÈW t|| ƒ}t|ƒ}|| ‰|�r<‡fdd„t |d ƒD ƒ}n&tj tˆd
t | ƒˆ|d ¡
¡ }| |t |ƒ ¡ d}x€t ||ƒD ]r\}}||
¡ k�r´|| ||| … | |< n(tj| | || ||| … dd �| |< ||7 }|| d 8 < �q„W q´W �n$|d k�r"d}|}t| ƒ‰ xâ|t k �rşd
d„ t ˆƒD ƒ}x¾t |ƒD ]²}t | |k¡d } tj | ¡ tj t tˆ¡¡} t ‡ ‡fdd„t | |ƒD ƒ¡} | | ¡ } t | ¡t| ƒ t¡d d… } dd„ t |t | | ¡ƒD ƒ}tdd„ |D ƒƒ}�qDW �q W x$t ˆƒD ]} || | | < �q
W nt‚xrt ˆƒD ]f}| | }
|
|
||< | |
||< x>t || ¡D ],}| | t|ƒtt || |kƒƒf¡ �qdW �q0W ~ x\t ˆƒD ]P}td|› dt|| ƒ› d�t || ¡ƒ tddd„ | | D ƒƒ tdƒ �q¦W ||| fS )Nc S s g | ]}g ‘qS r r )Ú.0Ú_r r r ú
<listcomp>& s z!separate_data.<locals>.<listcomp>c S s g | ]}g ‘qS r r )r r! r r r r" ' s c S s g | ]}g ‘qS r r )r r! r r r r" ( s Úpatc s g | ]}ˆ ‘qS r r )r r! )Úclass_per_clientr r r" 8 s r c s g | ] }t ˆ ƒ‘qS r )Úint)r r! )Únum_perr r r" D s r é
)ÚaxisÚdirc S s g | ]}g ‘qS r r )r r! r r r r" Y s c s$ g | ] \}}|t |ƒˆ ˆ k ‘qS r )Úlen)r ÚpÚidx_j)ÚNr r r r" ^ s éÿÿÿÿc S s g | ]\}}|| ¡ ‘qS r )Útolist)r r, Úidxr r r r" a s c S s g | ] }t |ƒ‘qS r )r* )r r, r r r r" b s zClient z Size of data: z
Labels: z Samples of labels: c S s g | ]}|‘qS r r )r Úir r r r" x s z2--------------------------------------------------)ÚrangeÚnpÚarrayr* Úappendr% ÚrandomÚrandintÚmaxÚ
least_samplesr/ ÚsumÚzipÚkeysÚwhereÚshuffleÚ dirichletÚrepeatr ÚcumsumÚastypeÚsplitÚminÚNotImplementedErrorÚuniquer ) Údatar r r r r
r$ ÚXÚyÚ statisticZdataset_contentÚ
dataset_labelZ dataidx_mapÚidxsZidx_for_each_classr1 Zclass_num_per_clientZselected_clientsÚclientZnum_all_samplesZnum_selected_clientsÚ num_samplesr0 Z
num_sampleÚmin_sizeÚKZ idx_batchÚkZidx_kZ proportionsÚjr )r- r$ r r& r Ú
separate_data% s€
&(
2*rS c C s g g }}g g dœ}x¼t t|ƒƒD ]¬}tj|| dd�\}}t|ƒdkrjt| | || tdd�\}} }
} n t| | || tdd�\}} }
} | ||
dœ¡ |d t|
ƒ¡ | | | dœ¡ |d t| ƒ¡ q"W td t |d |d ƒƒ td
|d ƒ td |d ƒ tƒ ~ ~||fS ) N)ÚtrainÚtestT)Ú
return_countsr )Ú
train_sizer> )ÚxrI rT rU zTotal number of samples:z The number of train samples:zThe number of test samples:)
r2 r* r3 rF rD r rW r5 r r: ) rH rI Ú
train_dataÚ test_datarN r1 rF ÚcountZX_trainZX_testZy_trainZy_testr r r Ú
split_data~ s&
r\ c
C sØ |||| |
|t tdœ} tdƒ xDt|ƒD ]8\} }
t|t| ƒ d dƒ�}tj||
d� W d Q R X q(W xDt|ƒD ]8\} }t|t| ƒ d dƒ�}tj||d� W d Q R X qnW t| dƒ�}t | |¡ W d Q R X tdƒ d S )N)r r r r r
z%Size of samples for labels in clientsr r zSaving to disk.
z.npzÚwb)rG ÚwzFinish generating dataset.
)
r r r Ú enumerater Ústrr3 Úsavez_compressedr Údump)r r r rY rZ r r rJ r r r
r r0 Z
train_dictr Z test_dictr r r Ú save_fileš s$ rc )FTN)FFNr )FTN)r
r Únumpyr3 ÚgcZsklearn.model_selectionr r rW r9 r r rS r\ rc r r r r Ú<module> s
Y | 5,332 | Python | .py | 54 | 97.592593 | 489 | 0.37526 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,859 | dataset_utils-checkpoint.py | hkgdifyu_pFedPT/dataset/utils/.ipynb_checkpoints/dataset_utils-checkpoint.py | import os
import ujson
import numpy as np
import gc
from sklearn.model_selection import train_test_split
batch_size = 2
train_size = 0.75 # merge original training set and test set, then split it manually.
least_samples = batch_size / (1-train_size) # least samples for each client
alpha = 0.1 # for Dirichlet distribution
def check(config_path, train_path, test_path, num_clients, num_classes, niid=False,
balance=True, partition=None):
# check existing dataset
if os.path.exists(config_path):
with open(config_path, 'r') as f:
config = ujson.load(f)
if config['num_clients'] == num_clients and \
config['num_classes'] == num_classes and \
config['non_iid'] == niid and \
config['balance'] == balance and \
config['partition'] == partition and \
config['alpha'] == alpha and \
config['batch_size'] == batch_size:
print("\nDataset already generated.\n")
return True
dir_path = os.path.dirname(train_path)
if not os.path.exists(dir_path):
os.makedirs(dir_path)
dir_path = os.path.dirname(test_path)
if not os.path.exists(dir_path):
os.makedirs(dir_path)
return False
def separate_data(data, num_clients, num_classes, niid=False, balance=False, partition=None, class_per_client=2):
X = [[] for _ in range(num_clients)]
y = [[] for _ in range(num_clients)]
statistic = [[] for _ in range(num_clients)]
dataset_content, dataset_label = data
dataidx_map = {}
if not niid:
partition = 'pat'
class_per_client = num_classes
if partition == 'pat':
idxs = np.array(range(len(dataset_label)))
idx_for_each_class = []
for i in range(num_classes):
idx_for_each_class.append(idxs[dataset_label == i])
class_num_per_client = [class_per_client for _ in range(num_clients)]
for i in range(num_classes):
selected_clients = []
for client in range(num_clients):
if class_num_per_client[client] > 0:
selected_clients.append(client)
selected_clients = selected_clients[:int(num_clients/num_classes*class_per_client)]
num_all_samples = len(idx_for_each_class[i])
num_selected_clients = len(selected_clients)
num_per = num_all_samples / num_selected_clients
if balance:
num_samples = [int(num_per) for _ in range(num_selected_clients-1)]
else:
num_samples = np.random.randint(max(num_per/10, least_samples/num_classes), num_per, num_selected_clients-1).tolist()
num_samples.append(num_all_samples-sum(num_samples))
idx = 0
for client, num_sample in zip(selected_clients, num_samples):
if client not in dataidx_map.keys():
dataidx_map[client] = idx_for_each_class[i][idx:idx+num_sample]
else:
dataidx_map[client] = np.append(dataidx_map[client], idx_for_each_class[i][idx:idx+num_sample], axis=0)
idx += num_sample
class_num_per_client[client] -= 1
elif partition == "dir":
# https://github.com/IBM/probabilistic-federated-neural-matching/blob/master/experiment.py
min_size = 0
K = num_classes
N = len(dataset_label)
while min_size < least_samples:
idx_batch = [[] for _ in range(num_clients)]
for k in range(K):
idx_k = np.where(dataset_label == k)[0]
np.random.shuffle(idx_k)
proportions = np.random.dirichlet(np.repeat(alpha, num_clients))
proportions = np.array([p*(len(idx_j)<N/num_clients) for p,idx_j in zip(proportions,idx_batch)])
proportions = proportions/proportions.sum()
proportions = (np.cumsum(proportions)*len(idx_k)).astype(int)[:-1]
idx_batch = [idx_j + idx.tolist() for idx_j,idx in zip(idx_batch,np.split(idx_k,proportions))]
min_size = min([len(idx_j) for idx_j in idx_batch])
for j in range(num_clients):
dataidx_map[j] = idx_batch[j]
else:
raise NotImplementedError
# assign data
for client in range(num_clients):
idxs = dataidx_map[client]
X[client] = dataset_content[idxs]
y[client] = dataset_label[idxs]
for i in np.unique(y[client]):
statistic[client].append((int(i), int(sum(y[client]==i))))
del data
# gc.collect()
for client in range(num_clients):
print(f"Client {client}\t Size of data: {len(X[client])}\t Labels: ", np.unique(y[client]))
print(f"\t\t Samples of labels: ", [i for i in statistic[client]])
print("-" * 50)
return X, y, statistic
def split_data(X, y):
# Split dataset
train_data, test_data = [], []
num_samples = {'train':[], 'test':[]}
for i in range(len(y)):
unique, count = np.unique(y[i], return_counts=True)
if min(count) > 1:
X_train, X_test, y_train, y_test = train_test_split(
X[i], y[i], train_size=train_size, shuffle=True)
else:
X_train, X_test, y_train, y_test = train_test_split(
X[i], y[i], train_size=train_size, shuffle=True)
train_data.append({'x': X_train, 'y': y_train})
num_samples['train'].append(len(y_train))
test_data.append({'x': X_test, 'y': y_test})
num_samples['test'].append(len(y_test))
print("Total number of samples:", sum(num_samples['train'] + num_samples['test']))
print("The number of train samples:", num_samples['train'])
print("The number of test samples:", num_samples['test'])
print()
del X, y
# gc.collect()
return train_data, test_data
def save_file(config_path, train_path, test_path, train_data, test_data, num_clients,
num_classes, statistic, niid=False, balance=True, partition=None):
config = {
'num_clients': num_clients,
'num_classes': num_classes,
'non_iid': niid,
'balance': balance,
'partition': partition,
'Size of samples for labels in clients': statistic,
'alpha': alpha,
'batch_size': batch_size,
}
# gc.collect()
print("Saving to disk.\n")
for idx, train_dict in enumerate(train_data):
with open(train_path + str(idx) + '.npz', 'wb') as f:
np.savez_compressed(f, data=train_dict)
for idx, test_dict in enumerate(test_data):
with open(test_path + str(idx) + '.npz', 'wb') as f:
np.savez_compressed(f, data=test_dict)
with open(config_path, 'w') as f:
ujson.dump(config, f)
print("Finish generating dataset.\n") | 6,858 | Python | .py | 148 | 36.763514 | 133 | 0.596581 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,860 | generate_cifar10-checkpoint.py | hkgdifyu_pFedPT/dataset/.ipynb_checkpoints/generate_cifar10-checkpoint.py | import numpy as np
import os
import sys
import random
import torch
import torchvision
import torchvision.transforms as transforms
from utils.dataset_utils import check, separate_data, split_data, save_file
random.seed(1)
np.random.seed(1)
num_clients = 200
num_classes = 10
dir_path = "Cifar10/"
# Allocate data to users
def generate_cifar10(dir_path, num_clients, num_classes, niid, balance, partition):
if not os.path.exists(dir_path):
os.makedirs(dir_path)
# Setup directory for train/test data
config_path = dir_path + "config.json"
train_path = dir_path + "train/"
test_path = dir_path + "test/"
if check(config_path, train_path, test_path, num_clients, num_classes, niid, balance, partition):
return
# Get Cifar10 data
transform = transforms.Compose(
[transforms.ToTensor(), transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5))])
trainset = torchvision.datasets.CIFAR10(
root=dir_path+"rawdata", train=True, download=True, transform=transform)
testset = torchvision.datasets.CIFAR10(
root=dir_path+"rawdata", train=False, download=True, transform=transform)
trainloader = torch.utils.data.DataLoader(
trainset, batch_size=len(trainset.data), shuffle=False)
testloader = torch.utils.data.DataLoader(
testset, batch_size=len(testset.data), shuffle=False)
for _, train_data in enumerate(trainloader, 0):
trainset.data, trainset.targets = train_data
for _, test_data in enumerate(testloader, 0):
testset.data, testset.targets = test_data
dataset_image = []
dataset_label = []
dataset_image.extend(trainset.data.cpu().detach().numpy())
dataset_image.extend(testset.data.cpu().detach().numpy())
dataset_label.extend(trainset.targets.cpu().detach().numpy())
dataset_label.extend(testset.targets.cpu().detach().numpy())
dataset_image = np.array(dataset_image)
dataset_label = np.array(dataset_label)
# dataset = []
# for i in range(num_classes):
# idx = dataset_label == i
# dataset.append(dataset_image[idx])
X, y, statistic = separate_data((dataset_image, dataset_label), num_clients, num_classes,
niid, balance, partition)
train_data, test_data = split_data(X, y)
save_file(config_path, train_path, test_path, train_data, test_data, num_clients, num_classes,
statistic, niid, balance, partition)
if __name__ == "__main__":
# niid = True if sys.argv[1] == "noniid" else False
# balance = True if sys.argv[2] == "balance" else False
# partition = sys.argv[3] if sys.argv[3] != "-" else None
niid = True
balance = True
partition = 'dir'
generate_cifar10(dir_path, num_clients, num_classes, niid, balance, partition) | 2,816 | Python | .py | 63 | 39.063492 | 101 | 0.683805 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,861 | generate_agnews-checkpoint.py | hkgdifyu_pFedPT/dataset/.ipynb_checkpoints/generate_agnews-checkpoint.py | import numpy as np
import os
import sys
import random
import torchtext
from utils.dataset_utils import check, separate_data, split_data, save_file
from torchtext.data.utils import get_tokenizer
from torchtext.vocab import build_vocab_from_iterator
random.seed(1)
np.random.seed(1)
num_clients = 20
num_classes = 4
max_len = 200
dir_path = "agnews/"
# Allocate data to users
def generate_agnews(dir_path, num_clients, num_classes, niid, balance, partition):
if not os.path.exists(dir_path):
os.makedirs(dir_path)
# Setup directory for train/test data
config_path = dir_path + "config.json"
train_path = dir_path + "train/"
test_path = dir_path + "test/"
if check(config_path, train_path, test_path, num_clients, num_classes, niid, balance, partition):
return
# Get AG_News data
trainset, testset = torchtext.datasets.AG_NEWS(root=dir_path+"rawdata")
trainlabel, traintext = list(zip(*trainset))
testlabel, testtext = list(zip(*testset))
dataset_text = []
dataset_label = []
dataset_text.extend(traintext)
dataset_text.extend(testtext)
dataset_label.extend(trainlabel)
dataset_label.extend(testlabel)
tokenizer = get_tokenizer('basic_english')
vocab = build_vocab_from_iterator(map(tokenizer, iter(dataset_text)), specials=["<unk>"])
vocab.set_default_index(vocab["<unk>"])
text_pipeline = lambda x: vocab(tokenizer(x))
label_pipeline = lambda x: int(x) - 1
def text_transform(text, label, max_len=0):
label_list, text_list = [], []
for _text, _label in zip(text, label):
label_list.append(label_pipeline(_label))
text_ = text_pipeline(_text)
padding = [0 for i in range(max_len-len(text_))]
text_.extend(padding)
text_list.append(text_[:max_len])
return label_list, text_list
label_list, text_list = text_transform(dataset_text, dataset_label, max_len)
text_lens = [len(text) for text in text_list]
# max_len = max(text_lens)
# label_list, text_list = text_transform(dataset_text, dataset_label, max_len)
text_list = [(text, l) for text, l in zip(text_list, text_lens)]
text_list = np.array(text_list, dtype=object)
label_list = np.array(label_list)
# dataset = []
# for i in range(num_classes):
# idx = label_list == i
# dataset.append(text_list[idx])
X, y, statistic = separate_data((text_list, label_list), num_clients, num_classes, niid, balance, partition)
train_data, test_data = split_data(X, y)
save_file(config_path, train_path, test_path, train_data, test_data, num_clients, num_classes,
statistic, niid, balance, partition)
print("The size of vocabulary:", len(vocab))
if __name__ == "__main__":
niid = True if sys.argv[1] == "noniid" else False
balance = True if sys.argv[2] == "balance" else False
partition = sys.argv[3] if sys.argv[3] != "-" else None
generate_agnews(dir_path, num_clients, num_classes, niid, balance, partition) | 3,059 | Python | .py | 69 | 38.884058 | 112 | 0.675 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,862 | generate_cifar100-checkpoint.py | hkgdifyu_pFedPT/dataset/.ipynb_checkpoints/generate_cifar100-checkpoint.py | import numpy as np
import os
import sys
import random
import torch
import torchvision
import torchvision.transforms as transforms
from utils.dataset_utils import check, separate_data, split_data, save_file
random.seed(1)
np.random.seed(1)
num_clients = 20
num_classes = 100
dir_path = "Cifar100/"
# Allocate data to users
def generate_cifar100(dir_path, num_clients, num_classes, niid, balance, partition):
if not os.path.exists(dir_path):
os.makedirs(dir_path)
# Setup directory for train/test data
config_path = dir_path + "config.json"
train_path = dir_path + "train/"
test_path = dir_path + "test/"
if check(config_path, train_path, test_path, num_clients, num_classes, niid, balance, partition):
return
# Get Cifar100 data
transform = transforms.Compose(
[transforms.ToTensor(), transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5))])
trainset = torchvision.datasets.CIFAR100(
root=dir_path+"rawdata", train=True, download=True, transform=transform)
testset = torchvision.datasets.CIFAR100(
root=dir_path+"rawdata", train=False, download=True, transform=transform)
trainloader = torch.utils.data.DataLoader(
trainset, batch_size=len(trainset.data), shuffle=False)
testloader = torch.utils.data.DataLoader(
testset, batch_size=len(testset.data), shuffle=False)
for _, train_data in enumerate(trainloader, 0):
trainset.data, trainset.targets = train_data
for _, test_data in enumerate(testloader, 0):
testset.data, testset.targets = test_data
dataset_image = []
dataset_label = []
dataset_image.extend(trainset.data.cpu().detach().numpy())
dataset_image.extend(testset.data.cpu().detach().numpy())
dataset_label.extend(trainset.targets.cpu().detach().numpy())
dataset_label.extend(testset.targets.cpu().detach().numpy())
dataset_image = np.array(dataset_image)
dataset_label = np.array(dataset_label)
# dataset = []
# for i in range(num_classes):
# idx = dataset_label == i
# dataset.append(dataset_image[idx])
X, y, statistic = separate_data((dataset_image, dataset_label), num_clients, num_classes,
niid, balance, partition, class_per_client=20)
train_data, test_data = split_data(X, y)
save_file(config_path, train_path, test_path, train_data, test_data, num_clients, num_classes,
statistic, niid, balance, partition)
if __name__ == "__main__":
niid = True if sys.argv[1] == "noniid" else False
balance = True if sys.argv[2] == "balance" else False
partition = sys.argv[3] if sys.argv[3] != "-" else None
generate_cifar100(dir_path, num_clients, num_classes, niid, balance, partition) | 2,781 | Python | .py | 60 | 40.666667 | 101 | 0.688848 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,863 | fedoptimizer.cpython-39.pyc | hkgdifyu_pFedPT/system/flcore/optimizers/__pycache__/fedoptimizer.cpython-39.pyc | a
fæ`c¸ „ @ sh d dl Z d dlmZ G ddÑ deÉZG ddÑ deÉZG ddÑ deÉZG d d
Ñ d
eÉZG d d Ñ d eÉZdS )
È N)⁄ Optimizerc s& e Zd Zá fddÑZdddÑZá ZS )⁄PerAvgOptimizerc s t |dç}tt| Ɇ||° d S ©N)⁄lr)⁄dict⁄superr ⁄__init__©⁄self⁄paramsr ⁄defaults©⁄ __class__© ˙x/media/sim812/391e55df-b6f2-4fe9-a920-53434a8506fa/lgh/pdept/PFL-Non-IID-master/system/flcore/optimizers/fedoptimizer.pyr s
zPerAvgOptimizer.__init__r c C sd | j D ]X}|d D ]J}|jd u r"q|jj}|dkrF|jj|| dç q|jj||d dç qqd S )Nr r )⁄other⁄alphar ©⁄ param_groups⁄grad⁄data⁄add_)r
⁄beta⁄group⁄p⁄d_pr r r ⁄step
s
zPerAvgOptimizer.step)r ©⁄__name__⁄
__module__⁄ __qualname__r r ⁄
__classcell__r r r
r r s r c s& e Zd Zdá fddÑ ZddÑ Zá ZS ) ⁄
FEDLOptimizerÁ{ÆG·zÑ?NÁöôôôôôπ?c sD || _ || _|dk r"td†|°ÉÇt||dç}tt| Ɇ||° d S )NÁ ˙Invalid learning rate: {})r ⁄eta)⁄ server_grads⁄ pre_grads⁄
ValueError⁄formatr r r" r )r
r r r( r) r' r r
r r r s zFEDLOptimizer.__init__c C s` | j D ]T}d}|d D ]B}|j†|d |jj|d | j| | j| ° |d7 }qqd S )Nr r r r' È )r r r r r( r) )r
r ⁄ir r r r r s
ˇˇzFEDLOptimizer.step)r# NNr$ r r r r
r r" s r" c s& e Zd Zdá fddÑ ZddÑ Zá ZS ) ⁄pFedMeOptimizerr# r$ Á¸©Ò“MbP?c s: |dk rt d†|°ÉÇt|||dç}tt| Ɇ||° d S )Nr% r& )r ⁄lamda⁄mu)r* r+ r r r. r )r
r r r0 r1 r r
r r r ) s zpFedMeOptimizer.__init__c C s| d }|† ° }| jD ]`}t|d |ÉD ]L\}}|†|°}|j|d |jj|d |j|j |d |j |_q$q|d S )Nr r r0 r1 )⁄copyr ⁄zip⁄tor r )r
Z local_model⁄devicer Z
weight_updater ⁄ localweightr r r r / s
>zpFedMeOptimizer.step)r# r$ r/ r r r r
r r. ( s r. c s& e Zd Zá fddÑZdddÑZá ZS )⁄
APFLOptimizerc s t |dç}tt| Ɇ||° d S r )r r r7 r r r
r r r Z s
zAPFLOptimizer.__init__r, c C sN | j D ]B}|d D ]4}|jd u r"q|| |jj }|j†|d |° qqd S )Nr r r )r
r ⁄n_kr r r r r r r ^ s
zAPFLOptimizer.step)r, r, r r r r
r r7 Y s r7 c s. e Zd Zdá fddÑ Ze†° ddÑ ÉZá ZS )⁄PerturbedGradientDescentr# r% c s4 |dk rt d|õ ùÉÇt||dç}tÉ †||° d S )Nr% zInvalid learning rate: )r r1 )r* r r r )r
r r r1 ⁄defaultr
r r r h s z!PerturbedGradientDescent.__init__c C sd | j D ]X}t|d |ÉD ]D\}}|†|°}|jj|d |j|j }|jj||d dç qqd S )Nr r1 r )r )r r3 r4 r r r )r
Z
global_paramsr5 r r ⁄gr r r r r p s
z PerturbedGradientDescent.step)r# r% )r r r r ⁄torch⁄no_gradr r! r r r
r r9 g s r9 )r< Z torch.optimr r r" r. r7 r9 r r r r ⁄<module> s 1 | 3,891 | Python | .pyt | 27 | 143.074074 | 419 | 0.346701 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,864 | fedoptimizer.cpython-38.pyc | hkgdifyu_pFedPT/system/flcore/optimizers/__pycache__/fedoptimizer.cpython-38.pyc | U
îjfc¸ „ @ sh d dl Z d dlmZ G ddÑ deÉZG ddÑ deÉZG ddÑ deÉZG d d
Ñ d
eÉZG d d Ñ d eÉZdS )
È N)⁄ Optimizerc s& e Zd Zá fddÑZdddÑZá ZS )⁄PerAvgOptimizerc s t |dç}tt| Ɇ||° d S ©N)⁄lr)⁄dict⁄superr ⁄__init__©⁄self⁄paramsr ⁄defaults©⁄ __class__© ıQ D:\‰∫¨‰∏ú\promot\cifar\cifar\Cifar10_iid\system\flcore\optimizers\fedoptimizer.pyr s
zPerAvgOptimizer.__init__r c C sd | j D ]X}|d D ]J}|jd kr"q|jj}|dkrF|jj|| dç q|jj||d dç qqd S )Nr r )⁄other⁄alphar ©⁄ param_groups⁄grad⁄data⁄add_)r
⁄beta⁄group⁄p⁄d_pr r r ⁄step
s
zPerAvgOptimizer.step)r ©⁄__name__⁄
__module__⁄ __qualname__r r ⁄
__classcell__r r r
r r s r c s& e Zd Zdá fddÑ ZddÑ Zá ZS ) ⁄
FEDLOptimizerÁ{ÆG·zÑ?NÁöôôôôôπ?c sD || _ || _|dk r"td†|°ÉÇt||dç}tt| Ɇ||° d S )NÁ ˙Invalid learning rate: {})r ⁄eta)⁄ server_grads⁄ pre_grads⁄
ValueError⁄formatr r r" r )r
r r r( r) r' r r
r r r s zFEDLOptimizer.__init__c C s` | j D ]T}d}|d D ]B}|j†|d |jj|d | j| | j| ° |d7 }qqd S )Nr r r r' È )r r r r r( r) )r
r ⁄ir r r r r s
ˇˇzFEDLOptimizer.step)r# NNr$ r r r r
r r" s r" c s& e Zd Zdá fddÑ ZddÑ Zá ZS ) ⁄pFedMeOptimizerr# r$ Á¸©Ò“MbP?c s: |dk rt d†|°ÉÇt|||dç}tt| Ɇ||° d S )Nr% r& )r ⁄lamda⁄mu)r* r+ r r r. r )r
r r r0 r1 r r
r r r ) s zpFedMeOptimizer.__init__c C s| d }|† ° }| jD ]`}t|d |ÉD ]L\}}|†|°}|j|d |jj|d |j|j |d |j |_q$q|d S )Nr r r0 r1 )⁄copyr ⁄zip⁄tor r )r
Z local_model⁄devicer Z
weight_updater ⁄ localweightr r r r / s
>zpFedMeOptimizer.step)r# r$ r/ r r r r
r r. ( s r. c s& e Zd Zá fddÑZdddÑZá ZS )⁄
APFLOptimizerc s t |dç}tt| Ɇ||° d S r )r r r7 r r r
r r r Z s
zAPFLOptimizer.__init__r, c C sN | j D ]B}|d D ]4}|jd kr"q|| |jj }|j†|d |° qqd S )Nr r r )r
r ⁄n_kr r r r r r r ^ s
zAPFLOptimizer.step)r, r, r r r r
r r7 Y s r7 c s. e Zd Zdá fddÑ Ze†° ddÑ ÉZá ZS )⁄PerturbedGradientDescentr# r% c s4 |dk rt d|õ ùÉÇt||dç}tÉ †||° d S )Nr% zInvalid learning rate: )r r1 )r* r r r )r
r r r1 ⁄defaultr
r r r h s z!PerturbedGradientDescent.__init__c C sd | j D ]X}t|d |ÉD ]D\}}|†|°}|jj|d |j|j }|jj||d dç qqd S )Nr r1 r )r )r r3 r4 r r r )r
Z
global_paramsr5 r r ⁄gr r r r r p s
z PerturbedGradientDescent.step)r# r% )r r r r ⁄torch⁄no_gradr r! r r r
r r9 g s r9 )r< Z torch.optimr r r" r. r7 r9 r r r r ⁄<module> s 1 | 3,855 | Python | .pyt | 27 | 141.740741 | 383 | 0.340559 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,865 | fedoptimizer.cpython-37.pyc | hkgdifyu_pFedPT/system/flcore/optimizers/__pycache__/fedoptimizer.cpython-37.pyc | B
À:ccü ã @ sh d dl Z d dlmZ G dd„ deƒZG dd„ deƒZG dd„ deƒZG d d
„ d
eƒZG d d „ d eƒZdS )
é N)Ú Optimizerc s& e Zd Z‡ fdd„Zddd„Z‡ ZS )ÚPerAvgOptimizerc s t |d�}tt| ƒ ||¡ d S )N)Úlr)ÚdictÚsuperr Ú__init__)ÚselfÚparamsr Údefaults)Ú __class__© úL/root/autodl-tmp/PFL-Non-IID-master/system/flcore/optimizers/fedoptimizer.pyr s
zPerAvgOptimizer.__init__r c C sl xf| j D ]\}xV|d D ]J}|jd kr&q|jj}|dkrJ|jj|| d� q|jj||d d� qW qW d S )Nr r )ÚotherÚalphar )Ú param_groupsÚgradÚdataÚadd_)r ÚbetaÚgroupÚpÚd_pr r r
Ústep
s
zPerAvgOptimizer.step)r )Ú__name__Ú
__module__Ú __qualname__r r Ú
__classcell__r r )r r
r s r c s& e Zd Zd‡ fdd„ Zdd„ Z‡ ZS ) Ú
FEDLOptimizerç{®Gáz„?Nçš™™™™™¹?c sD || _ || _|dk r"td |¡ƒ‚t||d�}tt| ƒ ||¡ d S )Ng zInvalid learning rate: {})r Úeta)Ú server_gradsÚ pre_gradsÚ
ValueErrorÚformatr r r r )r r r r! r" r r
)r r r
r s zFEDLOptimizer.__init__c C sh xb| j D ]X}d}xN|d D ]B}|j |d |jj|d | j| | j| ¡ |d7 }qW qW d S )Nr r r r é )r r r r r! r" )r r Úir r r r
r s (zFEDLOptimizer.step)r NNr )r r r r r r r r )r r
r s r c s& e Zd Zd‡ fdd„ Zdd„ Z‡ ZS ) ÚpFedMeOptimizerç{®Gáz„?çš™™™™™¹?çü©ñÒMbP?c s: |dk rt d |¡ƒ‚t|||d�}tt| ƒ ||¡ d S )Ng zInvalid learning rate: {})r ÚlamdaÚmu)r# r$ r r r' r )r r r r+ r, r
)r r r
r ) s zpFedMeOptimizer.__init__c C s„ d }| ¡ }xn| jD ]d}x^t|d |ƒD ]L\}}| |¡}|j|d |jj|d |j|j |d |j |_q(W qW |d S )Nr r r+ r, )Úcopyr ÚzipÚtor r )r Z local_modelÚdevicer Z
weight_updater Ú localweightr r r
r / s
BzpFedMeOptimizer.step)r( r) r* )r r r r r r r r )r r
r' ( s r' c s& e Zd Z‡ fdd„Zddd„Z‡ ZS )Ú
APFLOptimizerc s t |d�}tt| ƒ ||¡ d S )N)r )r r r2 r )r r r r
)r r r
r Z s
zAPFLOptimizer.__init__r% c C sV xP| j D ]F}x@|d D ]4}|jd kr&q|| |jj }|j |d |¡ qW qW d S )Nr r )r r r r )r r Ún_kr r r r r r
r ^ s
zAPFLOptimizer.step)r% r% )r r r r r r r r )r r
r2 Y s r2 c s. e Zd Zd‡ fdd„ Ze ¡ dd„ ƒZ‡ ZS )ÚPerturbedGradientDescentç{®Gáz„?ç c s4 |dk rt d|› �ƒ‚t||d�}tƒ ||¡ d S )Ng zInvalid learning rate: )r r, )r# r r r )r r r r, Údefault)r r r
r h s z!PerturbedGradientDescent.__init__c C sl xf| j D ]\}xVt|d |ƒD ]D\}}| |¡}|jj|d |j|j }|jj||d d� q W qW d S )Nr r, r )r )r r. r/ r r r )r Z
global_paramsr0 r r Úgr r r r
r p s
z PerturbedGradientDescent.step)r5 r6 )r r r r ÚtorchÚno_gradr r r r )r r
r4 g s r4 )r9 Z torch.optimr r r r' r2 r4 r r r r
Ú<module> s 1 | 3,999 | Python | .pyt | 17 | 234.176471 | 747 | 0.346975 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,866 | clientamp.cpython-38.pyc | hkgdifyu_pFedPT/system/flcore/clients/__pycache__/clientamp.cpython-38.pyc | U
”jfc ã @ sT d dl Z d dlmZ d dlmZ d dlZd dlZd dlZG dd„ deƒZ dd„ Z
dS )é N)ÚClientc s, e Zd Z‡ fdd„Zdd„ Zdd„ Z‡ ZS )Ú clientAMPc s\ t ƒ j||||f|� |j| _|j| _t | j¡| _t ¡ | _
t j j
| j ¡ | jd�| _d S )N)Úlr)ÚsuperÚ__init__ÚalphaKÚlamdaÚcopyÚdeepcopyÚmodelÚclient_uÚnnÚCrossEntropyLossÚlossÚtorchÚoptimÚSGDÚ
parametersÚ
learning_rateÚ optimizer)ÚselfÚargsÚidÚ
train_samplesÚ test_samplesÚkwargs©Ú __class__© õK D:\京东\promot\cifar\cifar\Cifar10_iid\system\flcore\clients\clientamp.pyr
s
zclientAMP.__init__c C s‚ | ¡ }t ¡ }| j | j¡ | j | j¡ | j ¡ | j}| jrTt j
d|d ¡}t |ƒD ]Ş}|D ]Ô\}}t
|ƒt
g ƒkr’|d | j¡|d< n | | j¡}| | j¡}| jrÊt dt t j
¡ ¡ ¡ | j ¡ | |¡}| ||¡}t| jƒ} t| jƒ}
| |
} || j| j d t | | ¡ 7 }| ¡ | j ¡ qdq\| j ¡ | j ¡ ~| j d d7 < | j d t ¡ | 7 < d S )Né é r gš™™™™™¹?Ú
num_roundsÚ
total_cost) Úload_train_dataÚtimer ÚtoÚdevicer ÚtrainÚ local_stepsÚ
train_slowÚnpÚrandomÚrandintÚrangeÚtypeÚsleepÚabsÚrandr Ú zero_gradr Úweight_flattenr r r ÚdotÚbackwardÚstepÚcpuÚtrain_time_cost) r Ú trainloaderÚ
start_timeÚmax_local_stepsr7 ÚxÚyÚoutputr ÚparamsZparams_Úsubr r r r( s<
zclientAMP.trainc C s8 t | ¡ | j ¡ ƒD ] \}}|j||j ¡ |_qd S )N)Úzipr r ÚdataÚclone)r r Ú coef_selfÚ new_paramÚ old_paramr r r Úset_parameters? s zclientAMP.set_parameters)Ú__name__Ú
__module__Ú __qualname__r r( rH Ú
__classcell__r r r r r s
+r c C s0 g }| ¡ D ]}| | d¡¡ q t |¡}|S )Néÿÿÿÿ)r ÚappendÚviewr Úcat)r r@ Úur r r r4 D s
r4 ) r Útorch.nnr
Úflcore.clients.clientbaser Únumpyr+ r% r r r4 r r r r Ú<module> s ; | 2,270 | Python | .pyt | 26 | 86.115385 | 269 | 0.429844 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,867 | clientdyn.cpython-38.pyc | hkgdifyu_pFedPT/system/flcore/clients/__pycache__/clientdyn.cpython-38.pyc | U
”jfc ã @ sT d dl Z d dlZd dlmZ d dlZd dlZd dlmZ G dd„ deƒZ dd„ Z
dS )é N)ÚClientc s, e Zd Z‡ fdd„Zdd„ Zdd„ Z‡ ZS )Ú clientDync sl t ƒ j||||f|� t ¡ | _tjj| j ¡ | j
d�| _ |j | _ d | _
t | j¡}t|ƒ}t |¡| _d S )N)Úlr)ÚsuperÚ__init__ÚnnÚCrossEntropyLossÚlossÚtorchÚoptimÚSGDÚmodelÚ
parametersÚ
learning_rateÚ optimizerÚalphaÚglobal_model_vectorÚcopyÚdeepcopyÚmodel_parameter_vectorÚ
zeros_likeÚold_grad)ÚselfÚargsÚidÚ
train_samplesÚ test_samplesÚkwargsr ©Ú __class__© õK D:\京东\promot\cifar\cifar\Cifar10_iid\system\flcore\clients\clientdyn.pyr s
zclientDyn.__init__c C sæ | ¡ }t ¡ }| j | j¡ | j | j¡| _| j | j¡| _| j ¡ | j}| j rft
j d|d ¡}t
|ƒD ]ò}t|ƒD ]ä\}\}}t|ƒtg ƒkr¬|d | j¡|d< n | | j¡}| | j¡}| j rät dt
t
j ¡ ¡ ¡ | j ¡ | |¡}| ||¡} | jd k�rLt| jƒ}
| | jd t |
| j d¡ 7 } | t |
| j¡8 } | ¡ | j ¡ qzqn| jd k�r”t| jƒ ¡ }
| j| j|
| j | _| j ¡ | j ¡ | _| j ¡ | _| j d d7 < | j d t ¡ | 7 < d S )Né é r gš™™™™™¹?Ú
num_roundsÚ
total_cost) Úload_train_dataÚtimer
ÚtoÚdevicer r ÚtrainÚ local_stepsÚ
train_slowÚnpÚrandomÚrandintÚrangeÚ enumerateÚtypeÚsleepÚabsÚrandr Ú zero_gradr r r r
ÚnormÚdotÚbackwardÚstepÚdetachÚcpuÚtrain_time_cost) r Ú trainloaderÚ
start_timeÚmax_local_stepsr: ÚiÚxÚyÚoutputr Úv1r r r! r* sD
zclientDyn.trainc C s@ t | ¡ | j ¡ ƒD ]\}}|j ¡ |_qt|ƒ ¡ ¡ | _d S )N)Úzipr r
ÚdataÚcloner r; r )r r
Ú new_paramÚ old_paramr r r! Úset_parametersO s zclientDyn.set_parameters)Ú__name__Ú
__module__Ú __qualname__r r* rK Ú
__classcell__r r r r! r
s 1r c C s dd„ | ¡ D ƒ}tj|dd�S )Nc S s g | ]}| d ¡‘qS )éÿÿÿÿ)Úview)Ú.0Úpr r r! Ú
<listcomp>W s z*model_parameter_vector.<locals>.<listcomp>r )Údim)r r
Úcat)r
Úparamr r r! r V s r ) r r
Útorch.nnr Únumpyr- r' Úflcore.clients.clientbaser r r r r r r! Ú<module> s L | 2,595 | Python | .pyt | 30 | 85.433333 | 285 | 0.424006 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,868 | clientmoon.cpython-39.pyc | hkgdifyu_pFedPT/system/flcore/clients/__pycache__/clientmoon.cpython-39.pyc | a
ç`cE ã @ s^ d dl Z d dlZd dlmZ d dlZd dlZd dlm mZ d dl
m Z G dd„ de ƒZ dS )é N)ÚClientc s, e Zd Z‡ fdd„Zdd„ Zdd„ Z‡ ZS )Ú
clientMOONc sf t ƒ j||||fi |¤� t ¡ | _tjj| j ¡ | j
d�| _ |j | _ |j
| _
d | _t | j¡| _d S )N)Úlr)ÚsuperÚ__init__ÚnnÚCrossEntropyLossÚlossÚtorchÚoptimÚSGDÚmodelÚ
parametersÚ
learning_rateÚ optimizerÚtauÚmuÚ global_modelÚcopyÚdeepcopyÚ old_model)ÚselfÚargsÚidÚ
train_samplesÚ test_samplesÚkwargs©Ú __class__© ús/media/sim812/391e55df-b6f2-4fe9-a920-53434a8506fa/lgh/pdept/PFL-Non-IID-master/system/flcore/clients/clientmoon.pyr s
zclientMOON.__init__c C sÜ | ¡ }t ¡ }| j ¡ | j}| jr8tj d|d ¡}t |ƒD �]Z}t
|ƒD �]J\}\}}t |ƒt g ƒkr‚|d | j
¡|d< n | | j
¡}| | j
¡}| jrºt dt tj ¡ ¡ ¡ | j ¡ | j |¡}| j |¡} | | |¡}
| j |¡ ¡ } | j |¡ ¡ } t t t || ¡| j ¡t t || ¡| j ¡t t || ¡| j ¡ ¡ }
|
| j t |
¡ 7 }
|
!¡ | j"�r�t#| j|t$|ƒƒ qN| j %¡ qNq@t& '| j¡| _| j(d d7 < | j(d t ¡ | 7 < d S )Né é r gš™™™™™¹?Ú
num_roundsÚ
total_cost))Úload_train_dataÚtimer
ÚtrainÚ local_stepsÚ
train_slowÚnpÚrandomÚrandintÚrangeÚ enumerateÚtypeÚtoÚdeviceÚsleepÚabsÚrandr Ú zero_gradÚbaseÚ predictorr r Údetachr r
ÚlogÚexpÚFÚcosine_similarityr r ÚmeanÚbackwardÚprivacyÚdp_stepÚlenÚstepr r Útrain_time_cost)r Ú trainloaderÚ
start_timeÚmax_local_stepsrB ÚiÚxÚyÚrepÚoutputr Zrep_oldZ
rep_globalZloss_conr r r r' s:
PzclientMOON.trainc C s4 t | ¡ | j ¡ ƒD ]\}}|j ¡ |_q|| _d S )N)Úzipr r
ÚdataÚcloner )r r
Ú new_paramÚ old_paramr r r Úset_parametersK s zclientMOON.set_parameters)Ú__name__Ú
__module__Ú __qualname__r r' rQ Ú
__classcell__r r r r r s .r )
r r
Útorch.nnr Únumpyr* r& Útorch.nn.functionalÚ
functionalr; Úflcore.clients.clientbaser r r r r r Ú<module> s | 2,382 | Python | .pyt | 23 | 102.478261 | 291 | 0.458475 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,869 | clientmtl.cpython-38.pyc | hkgdifyu_pFedPT/system/flcore/clients/__pycache__/clientmtl.cpython-38.pyc | U
”jfcª
ã @ s\ d dl Z d dlmZ d dlmZ d dlZd dlZd dlZd dl Z G dd„ deƒZ
dd„ Z dS )é N)ÚClientc s, e Zd Z‡ fdd„Zdd„ Zdd„ Z‡ ZS )Ú clientMTLc s` t ƒ j||||f|� d | _d | _d| _|j| _d| _t ¡ | _ t
j j | j
¡ | jdd�| _d S )Nr g-C ëâ6?g à?)ÚlrÚmomentum)ÚsuperÚ__init__ÚomegaÚW_globÚidxZitkZlambaÚnnÚCrossEntropyLossÚlossÚtorchÚoptimÚSGDÚmodelÚ
parametersÚ
learning_rateÚ optimizer)ÚselfÚargsÚidÚ
train_samplesÚ test_samplesÚkwargs©Ú __class__© õK D:\京东\promot\cifar\cifar\Cifar10_iid\system\flcore\clients\clientmtl.pyr s
zclientMTL.__init__c C s | ¡ }t ¡ }| j | j¡ | j ¡ | j}| jrFtj
d|d ¡}t |ƒD �]*}|D �] \}}t |ƒt g ƒkrˆ|d | j¡|d< n | | j¡}| | j¡}| jrÀt
dt tj ¡ ¡ ¡ | j ¡ | |¡}| ||¡}t| jƒ| jd d …| jf< d} | | j ¡ d 7 } | t t | j| j d¡d ¡7 } tt | jj d ¡d ƒd }
| d|
9 } || 7 }| ¡ | j ¡ qXqN| j ¡ d | _d | _| j!d d7 < | j!d t ¡ | 7 < d S )Né é r gš™™™™™¹?é
Ú
num_roundsÚ
total_cost)"Úload_train_dataÚtimer ÚtoÚdeviceÚtrainÚ local_stepsÚ
train_slowÚnpÚrandomÚrandintÚrangeÚtypeÚsleepÚabsÚrandr Ú zero_gradr
Úflattenr r
Únormr Úsumr ÚintÚmathÚlog10ÚshapeÚbackwardÚstepÚcpuÚtrain_time_cost) r Ú trainloaderÚ
start_timeÚmax_local_stepsr< ÚxÚyÚoutputr
Zloss_regularizerÚfr r r r( s@
"
zclientMTL.trainc C s* t |d d ¡| _t |¡| _|| _d S )Nr )r Úsqrtr ÚcopyÚdeepcopyr r
)r r r r
r r r Úreceive_valuesH s zclientMTL.receive_values)Ú__name__Ú
__module__Ú __qualname__r r( rI Ú
__classcell__r r r r r
s 1r c s, | ¡ ‰ ˆ ¡ }‡ fdd„|D ƒ}t |¡S )Nc s g | ]}ˆ | ¡ ‘qS r )r4 )Ú.0Úkey©Ú
state_dictr r Ú
<listcomp>Q s zflatten.<locals>.<listcomp>)rQ Úkeysr Úcat)r rS ÚWr rP r r4 N s r4 ) r Útorch.nnr Úflcore.clients.clientbaser Únumpyr+ r% r8 rG r r4 r r r r Ú<module> s D | 2,533 | Python | .pyt | 26 | 96.269231 | 278 | 0.419856 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,870 | clientrodpt.cpython-38.pyc | hkgdifyu_pFedPT/system/flcore/clients/__pycache__/clientrodpt.cpython-38.pyc | U
Åìıcû! ã @ s d dl Z d dlZd dlmZ d dlZd dlZd dlmZ d dl m m
Z d dl m
Z
d dlmZ d dl Z d dl m
Z
d dlmZ G dd„ deƒZd
dd „ZdS ) é N)ÚClient)Úlabel_binarize)Úmetricsc s> e Zd Z‡ fdd„Zdd„ Zd dd„Zdd „ Zd
d„ Z‡ ZS ) Ú clientRODPTc
s t ƒ j||||f|� t ¡ | _tj | jj
¡ | j dœ| jj
¡ | j dœg¡| _
|j| _tjj| jj
¡ |j|jd�| _tjjj| j| j|jd�| _t | jj ¡| _tjj| j
¡ | j d�| _t | j ¡| _ | ¡ }|D ](\}}|D ]} | j | ¡ d7 < qæqÚd S )N)ÚparamsÚlr)r Úmomentum)Ú step_sizeÚgamma)r é ) ÚsuperÚ__init__ÚnnÚCrossEntropyLossÚlossÚtorchÚoptimÚSGDÚmodelÚbaseÚ
parametersÚ
learning_rateÚ predictorÚ optimizerÚ plocal_stepsÚ generatorÚpt_learning_rater Ú
poptimizerÚ lr_schedulerÚStepLRÚlearning_decayÚ schedulerÚcopyÚdeepcopyÚpredÚopt_predÚzerosÚ num_classesÚsample_per_classÚload_train_dataÚitem)
ÚselfÚargsÚidÚ
train_samplesÚ test_samplesÚkwargsÚ trainloaderÚxÚyÚyy©Ú __class__© õY D:\京东\promot\第二次投稿\å®�验\native - pro\system\flcore\clients\clientrodpt.pyr
s(
şÿÿ zclientRODPT.__init__c C sR | ¡ }t ¡ }t | jj¡}| j | j¡ | j | j¡ | j ¡ | jj
¡ D ]
}d|_ qP| jj ¡ D ]
}d|_ qh| jj
¡ D ]
}d|_ q€t| jƒD ]¶}t|ƒD ]¨\}\}}t|ƒtg ƒkrÔ|d | j¡|d< n | | j¡}| | j¡}| j�rt dt tj ¡ ¡ ¡ | j ¡ | |¡} | | |¡}
|
¡ | j ¡ | j ¡ q¢q–| j } | j�rntj d| d ¡} | jj
¡ D ] }d|_ �qz| jj ¡ D ] }d|_ �q”| jj
¡ D ] }d|_ �q®t| ƒD ]æ}t|ƒD ]Ö\}\}}t|ƒtg ƒk�r|d | j¡|d< n | | j¡}| | j¡}| j
| j |¡¡} | j
| ¡}
t ||
| j!ƒ}| j" ¡ | ¡ | j" ¡ | | #¡ ¡}| |
#¡ | |¡}
| j$ ¡ |
¡ | j$ ¡ �qĞ�qÄ| j %¡ | j %¡ t | jj¡}d}t&| ¡ | ¡ ƒD ]<\}}|| }t' (|dk|t' )|¡| ¡}|t' *|¡ }�qä| j+d d7 < | j+d t ¡ | 7 < |S ) NFTr gš™™™™™¹?r é Ú
num_roundsÚ
total_cost),r) Útimer" r# r r ÚtoÚdevicer$ Útrainr r Ú
requires_gradr Úranger Ú enumerateÚtypeÚ
train_slowÚsleepÚnpÚabsÚrandomÚrandr Ú zero_gradr ÚbackwardÚstepr! Ú local_stepsÚrandintÚbalanced_softmax_lossr( r Údetachr% ÚcpuÚzipr ÚwhereÚ
zeros_likeÚsumÚtrain_time_cost)r+ r1 Ú
start_timeÚ
old_promptÚparamrL Úir2 r3 Úoutputr Úmax_local_stepsÚrepÚout_gÚloss_bsmÚout_pÚ
new_promptÚ
diff_provalueÚ new_paramÚ old_paramÚdiff_pror7 r7 r8 r? ( s~
zclientRODPT.trainNc C sŞ | ¡ }|d kr| j}| j | j¡ | j | j¡ | ¡ d}d}g }g }t ¡ ��: |D �],\}}t|ƒtg ƒkrŠ|d | j¡|d< n | | j¡}| | j¡}| j | j
|¡¡} | j | ¡}
| | ¡ ¡} |
¡ | } |t
tj| dd�|k¡ ¡ 7 }||jd 7 }| t | ¡ ¡ ¡ ¡ ¡ | j}
| jdk�r@|
d7 }
t| ¡ ¡ ¡ t |
¡d�}| jdk�r~|d d …d d…f }| |¡ qZW 5 Q R X | j ¡ | j ¡ tj|dd�}tj|dd�}tj ||dd�}|||fS ) Nr r ©Údimr9 ©Úclasses©ÚaxisÚmicro©Úaverage) Úload_test_datar r= r> r$ Úevalr Úno_gradrC r r r rP rU Úargmaxr* ÚshapeÚappendÚFÚsoftmaxrQ Únumpyr' r rF ÚarangeÚ concatenater Ú
roc_auc_score)r+ r Ú
testloaderÚtest_accÚtest_numÚy_probÚy_truer2 r3 r] r^ r` r[ ÚncÚlbÚaucr7 r7 r8 Ú test_metricst sH
zclientRODPT.test_metricsc C s` t |j ¡ | jj ¡ ƒD ]\}}|j ¡ |_qt |j ¡ | jj ¡ ƒD ]\}}|j ¡ |_qFd S )N)rR r r r ÚdataÚcloner )r+ r rc rd r7 r7 r8 Úset_parameters s zclientRODPT.set_parametersc C s~ | ¡ }| j | j¡ | j ¡ d}d}d}g }g }t ¡ �ú |D ]î\}}t|ƒtg ƒkrp|d | j¡|d< n | | j¡}| | j¡}| |¡} | j | j |¡¡}
|t
tj | dd�|k¡ ¡ 7 }|t
tj |
dd�|k¡ ¡ 7 }||j
d 7 }| | ¡ ¡ ¡ ¡ | t| ¡ ¡ ¡ t | j¡d�¡ qBW 5 Q R X | j ¡ tj|dd�}tj|dd�}tj||dd�} |||| fS )Nr r rf rh rj rl rm )ro r r= r> rp r rq rC r r rU rr r* rs rt rP rQ rw r rF rx r' ry r rz ) r+ Útestloaderfullr| Ú test_acc2r} r~ r r2 r3 r[ Úoutput2r‚ r7 r7 r8 rƒ ¦ s4
2
)N)Ú__name__Ú
__module__Ú __qualname__r
r? rƒ r† Ú
__classcell__r7 r7 r5 r8 r s
L
,r Úmeanc C sB | |¡}| d¡ |jd d¡}|| ¡ }tj|| |d�}|S )a} Compute the Balanced Softmax Loss between `logits` and the ground truth `labels`.
Args:
labels: A int tensor of size [batch].
logits: A float tensor of size [batch, no_of_classes].
sample_per_class: A int tensor of size [no of classes].
reduction: string. One of "none", "mean", "sum"
Returns:
loss: A float tensor. Balanced Softmax Loss.
r éÿÿÿÿ)ÚinputÚtargetÚ reduction)Útype_asÚ unsqueezeÚexpandrs Úlogru Ú
cross_entropy)ÚlabelsÚlogitsr( r’ Úspcr r7 r7 r8 rO Í s
rO )r� )r" r Útorch.nnr rw rF r< Úflcore.clients.clientbaser Útorch.nn.functionalÚ
functionalru Úsklearn.preprocessingr Úsklearnr r rO r7 r7 r7 r8 Ú<module> s @ | 5,970 | Python | .pyt | 71 | 82.211268 | 539 | 0.404645 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,871 | clientbn.cpython-39.pyc | hkgdifyu_pFedPT/system/flcore/clients/__pycache__/clientbn.cpython-39.pyc | a
Öæ`cŸ ã @ sD d dl Z d dlmZ d dlZd dlZd dlmZ G dd„ deƒZdS )é N)ÚClientc s, e Zd Z‡ fdd„Zdd„ Zdd„ Z‡ ZS )ÚclientBNc sB t ƒ j||||fi |¤� t ¡ | _tjj| j ¡ | j
d�| _ d S )N)Úlr) ÚsuperÚ__init__ÚnnÚCrossEntropyLossÚlossÚtorchÚoptimÚSGDÚmodelÚ
parametersÚ
learning_rateÚ optimizer)ÚselfÚargsÚidÚ
train_samplesÚ test_samplesÚkwargs©Ú __class__© úq/media/sim812/391e55df-b6f2-4fe9-a920-53434a8506fa/lgh/pdept/PFL-Non-IID-master/system/flcore/clients/clientbn.pyr
s
zclientBN.__init__c C sz | ¡ }t ¡ }| j ¡ | j}| jr8tj d|d ¡}t |ƒD ]Ä}t
|ƒD ]¶\}\}}t |ƒt g ƒkr~|d | j
¡|d< n | | j
¡}| | j
¡}| jr¶t dt tj ¡ ¡ ¡ | j ¡ | |¡}| ||¡} | ¡ | jrøt| j|t|ƒƒ qL| j ¡ qLq@| jd d7 < | jd t ¡ | 7 < | j�rvt| jƒ\}
} td| j › �d|
d d ›d
| › d |
d › �ƒ d S ) Né é r gš™™™™™¹?Ú
num_roundsÚ
total_costzClient u (ε = z.2fu , δ = u ) for α = ) Úload_train_dataÚtimer
ÚtrainÚ local_stepsÚ
train_slowÚnpÚrandomÚrandintÚrangeÚ enumerateÚtypeÚtoÚdeviceÚsleepÚabsÚrandr Ú zero_gradr ÚbackwardÚprivacyÚdp_stepÚlenÚstepÚtrain_time_costZ
get_dp_paramsÚprintr ) r Ú trainloaderÚ
start_timeÚmax_local_stepsr4 ÚiÚxÚyÚoutputr ÚresZDELTAr r r r! s4
zclientBN.trainc C s> t | ¡ | j ¡ ƒD ]$\\}}\}}d|vr|j ¡ |_qd S )NÚbn)ÚzipÚnamed_parametersr
ÚdataÚclone)r r
r r$ ÚonÚopr r r Úset_parameters< s $zclientBN.set_parameters)Ú__name__Ú
__module__Ú __qualname__r r! rF Ú
__classcell__r r r r r s 'r ) r
Útorch.nnr Únumpyr$ r Úflcore.clients.clientbaser r r r r r Ú<module> s
| 2,077 | Python | .pyt | 21 | 97.857143 | 297 | 0.461351 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,872 | clientditto.cpython-37.pyc | hkgdifyu_pFedPT/system/flcore/clients/__pycache__/clientditto.cpython-37.pyc | B
E&fcñ ã @ s‚ d dl Z d dlZd dlZd dlZd dlmZ d dlmZ d dl m
Z
d dl m m Z
d dlmZ d dlmZ G dd„ de
ƒZdS )é N)ÚPerturbedGradientDescent)ÚClient)Úlabel_binarize)Úmetricsc s4 e Zd Z‡ fdd„Zdd„ Zdd„ Zdd„ Z‡ ZS ) Ú clientDittoc sv t ƒ j||||f|� |j| _|j| _t | j¡| _t ¡ | _
t j j
| j ¡ | jd�| _t| j ¡ | j| jd�| _d S )N)Úlr)r Úmu)ÚsuperÚ__init__r Ú plocal_stepsÚcopyÚdeepcopyÚmodelÚpmodelÚnnÚCrossEntropyLossÚlossÚtorchÚoptimÚSGDÚ
parametersÚ
learning_rateÚ optimizerr Ú
poptimizer)ÚselfÚargsÚidÚ
train_samplesÚ test_samplesÚkwargs)Ú __class__© úH/root/autodl-tmp/PFL-Non-IID-master/system/flcore/clients/clientditto.pyr
s
zclientDitto.__init__c
C sX | ¡ }t ¡ }| j | j¡ | j ¡ | j}| jrFtj
d|d ¡}xÖt |ƒD ]Ê}xÄt |ƒD ]¸\}\}}t
|ƒt
g ƒkr�|d | j¡|d< n | | j¡}| | j¡}| jrÈt dt tj ¡ ¡ ¡ | j ¡ | |¡}| ||¡} | ¡ | j�r t| j|t|ƒƒ q^| j ¡ q^W qPW | j ¡ | jd d7 < | jd t ¡ | 7 < d S )Né é r gš™™™™™¹?Ú
num_roundsÚ
total_cost)Úload_train_dataÚtimer ÚtoÚdeviceÚtrainÚ local_stepsÚ
train_slowÚnpÚrandomÚrandintÚrangeÚ enumerateÚtypeÚsleepÚabsÚrandr Ú zero_gradr ÚbackwardÚprivacyÚdp_stepÚlenÚstepÚcpuÚtrain_time_cost)
r Ú trainloaderÚ
start_timeÚmax_local_stepsr< ÚiÚxÚyÚoutputr r! r! r" r+ ! s2
zclientDitto.trainc C s. | ¡ }t ¡ }| j | j¡ | j ¡ | j}| jrFtj
d|d ¡}x¾t |ƒD ]²}x¬|D ]¤\}}t |ƒt g ƒkrˆ|d | j¡|d< n | | j¡}| | j¡}| jrÀt
dt tj ¡ ¡ ¡ | j ¡ | |¡}| ||¡}| ¡ | j | j ¡ | j¡ qZW qPW | j ¡ | jd t ¡ | 7 < d S )Nr# r$ r gš™™™™™¹?r& )r' r( r r) r* r+ r r- r. r/ r0 r1 r3 r4 r5 r6 r r7 r r8 r< r r r= r> ) r r? r@ rA r< rC rD rE r r! r! r" ÚptrainI s,
zclientDitto.ptrainc
C sN | ¡ }| j | j¡ | j ¡ d}d}g }g }t ¡ �Ğ xÈ|D ]À\}}t|ƒtg ƒkrn|d | j¡|d< n | | j¡}| | j¡}| |¡}|t tj |dd�|k¡
¡ 7 }||j d 7 }| t
|¡ ¡ ¡ ¡ ¡ | t| ¡ ¡ ¡ t | j¡d�¡ q@W W d Q R X | j ¡ tj|dd�}tj|dd�}tj||dd�} ||| fS )Nr r# )Údim)Úclasses)ÚaxisÚmicro)Úaverage)Úload_test_datar r) r* Úevalr Úno_gradr3 ÚsumÚargmaxÚitemÚshapeÚappendÚFÚsoftmaxÚdetachr= Únumpyr r. ÚarangeÚ num_classesÚ concatenater Ú
roc_auc_score)
r ÚtestloaderfullÚtest_accÚtest_numÚy_probÚy_truerC rD rE Úaucr! r! r" Ú test_metricsh s.
4
zclientDitto.test_metrics)Ú__name__Ú
__module__Ú __qualname__r
r+ rF rb Ú
__classcell__r! r! )r r" r s ( r )r rW r. r( r Útorch.nnr Ú flcore.optimizers.fedoptimizerr Úflcore.clients.clientbaser Útorch.nn.functionalÚ
functionalrT Úsklearn.preprocessingr Úsklearnr r r! r! r! r" Ú<module> s | 3,486 | Python | .pyt | 36 | 95.666667 | 490 | 0.424515 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,873 | clientbabupt.cpython-38.pyc | hkgdifyu_pFedPT/system/flcore/clients/__pycache__/clientbabupt.cpython-38.pyc | U
3Èıck ã @ sl d dl Z d dlZd dlmZ d dlZd dlZd dlmZ d dl Z d dl m
Z
d dl m Z G dd„ deƒZ
dS )é N)ÚClient)Úlabel_binarize)Úmetricsc sD e Zd Z‡ fdd„Zdd„ Zdd„ Zddgfd d
„Zd d „ Z‡ ZS )
Ú clientBABUPTc sÀ t ƒ j||||f|� t ¡ | _tjj| jj
¡ | j d�| _ |j
| _
tjj| jj
¡ | j d�| _ |j| _tjj| jj
¡ |j|jd�| _tjjj| j| j|jd�| _| jj
¡ D ]
}d|_q°d S )N)Úlr)r Úmomentum)Ú step_sizeÚgammaF)ÚsuperÚ__init__ÚnnÚCrossEntropyLossÚlossÚtorchÚoptimÚSGDÚmodelÚbaseÚ
parametersÚ
learning_rateÚ optimizerÚfine_tuning_stepsÚ plocal_stepsÚ generatorÚpt_learning_rater Ú
poptimizerÚ lr_schedulerÚStepLRÚlearning_decayÚ schedulerÚ predictorÚ
requires_grad)ÚselfÚargsÚidÚ
train_samplesÚ test_samplesÚkwargsÚparam©Ú __class__© õZ D:\京东\promot\第二次投稿\å®�验\native - pro\system\flcore\clients\clientbabupt.pyr s
ÿÿzclientBABUPT.__init__c C s | ¡ }t ¡ }t | jj¡}| j | j¡ | j ¡ | jj
¡ D ]
}d|_ qB| jj
¡ D ]
}d|_ qZ| jj
¡ D ]
}d|_ qrt
| jƒD ]´}t|ƒD ]¦\}\}}t|ƒtg ƒkrÆ|d | j¡|d< n | | j¡}| | j¡}| jrşt dt tj ¡ ¡ ¡ | j ¡ | |¡} | | |¡}
|
¡ | j ¡ | j ¡ q”qˆ| j } | j�r^tj d| d ¡} | jj
¡ D ] }d|_ �qj| jj
¡ D ] }d|_ �q„| jj
¡ D ] }d|_ �q�t
| ƒD ]²}t|ƒD ]¢\}\}}t|ƒtg ƒk�rô|d | j¡|d< n | | j¡}| | j¡}| j�r.t dt tj ¡ ¡ ¡ | j ¡ | |¡} | | |¡}
|
¡ | j ¡ �qÀ�q´| j ¡ t | jj¡} d}
t!|
¡ |
¡ ƒD ]<\}}|| }t" #|dk|t" $|¡| ¡}|
t" %|¡ }
�q–| j&d d7 < | j&d t ¡ | 7 < |
S ) NFTr gš™™™™™¹?é é Ú
num_roundsÚ
total_cost)'Úload_train_dataÚtimeÚcopyÚdeepcopyr r ÚtoÚdeviceÚtrainr r r! r Úranger Ú enumerateÚtypeÚ
train_slowÚsleepÚnpÚabsÚrandomÚrandr Ú zero_gradr ÚbackwardÚstepr Ú local_stepsÚrandintr ÚcpuÚzipr ÚwhereÚ
zeros_likeÚsumÚtrain_time_cost)r" Ú trainloaderÚ
start_timeÚ
old_promptr( rC ÚiÚxÚyÚoutputr Úmax_local_stepsÚ
new_promptÚ
diff_provalueÚ new_paramÚ old_paramÚdiff_pror+ r+ r, r7 sr
zclientBABUPT.trainc C s2 t |j ¡ | jj ¡ ƒD ]\}}|j ¡ |_qd S )N)rG r r r ÚdataÚclone)r" r rV rW r+ r+ r, Úset_parameters_ s zclientBABUPT.set_parametersr r c
C s | ¡ }| j | j¡ | j ¡ d|kr@| jj ¡ D ]
}d|_q4d|kr`| jj ¡ D ]
}d|_qTt| j ƒD ]Š}t
|ƒD ]|\}\}}t |ƒt g ƒkr¨|d | j¡|d< n | | j¡}| | j¡}| j
¡ | |¡}| ||¡} | ¡ | j ¡ qvqj| j ¡ d S )Nr Tr Fr )r1 r r5 r6 r7 r r r! r8 r r9 r: r rA r rB rC rF )
r" Ú which_modulerL r( rC rO rP rQ rR r r+ r+ r, Ú fine_tunec s*
zclientBABUPT.fine_tunec C s~ | ¡ }| j | j¡ | j ¡ d}d}d}g }g }t ¡ �ú |D ]î\}}t|ƒtg ƒkrp|d | j¡|d< n | | j¡}| | j¡}| |¡} | j | j |¡¡}
|t
tj | dd�|k¡ ¡ 7 }|t
tj |
dd�|k¡ ¡ 7 }||j
d 7 }| | ¡ ¡ ¡ ¡ | t| ¡ ¡ ¡ t | j¡d�¡ qBW 5 Q R X | j ¡ tj|dd�}tj|dd�}tj||dd�} |||| fS )Nr r- )Údim)Úclasses)ÚaxisÚmicro)Úaverage)Úload_test_datar r5 r6 Úevalr Úno_gradr: r r rJ ÚargmaxÚitemÚshapeÚappendÚdetachrF Únumpyr r= ÚarangeÚ num_classesÚ concatenater Ú
roc_auc_score) r" ÚtestloaderfullÚtest_accÚ test_acc2Útest_numÚy_probÚy_truerP rQ rR Úoutput2Úaucr+ r+ r, Ú test_metrics~ s4
2
zclientBABUPT.test_metrics) Ú__name__Ú
__module__Ú __qualname__r r7 r[ r] rx Ú
__classcell__r+ r+ r) r, r s
Ar )r3 r Útorch.nnr rk r= r2 Úflcore.clients.clientbaser Úsklearn.preprocessingr Úsklearnr r r+ r+ r+ r, Ú<module> s | 4,511 | Python | .pyt | 63 | 70.222222 | 558 | 0.404136 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,874 | clientpt.cpython-37.pyc | hkgdifyu_pFedPT/system/flcore/clients/__pycache__/clientpt.cpython-37.pyc | B
K´cc
ã @ sL d dl Z d dlmZ d dlZd dlZd dlmZ d dlZG dd„ deƒZ dS )é N)ÚClientc s, e Zd Z‡ fdd„Zdd„ Zdd„ Z‡ ZS )ÚclientPTc sd t ƒ j||||f|� t ¡ | _tjj| jj
¡ | j d�| _ tjj| jj
¡ | j d�| _|j| _d S )N)Úlr)ÚsuperÚ__init__ÚnnÚCrossEntropyLossÚlossÚtorchÚoptimÚSGDÚmodelÚ generatorÚ
parametersÚ
learning_rateÚ
poptimizerÚbaseÚ optimizerÚ plocal_steps)ÚselfÚargsÚidÚ
train_samplesÚ test_samplesÚkwargs)Ú __class__© úE/root/autodl-tmp/PFL-Non-IID-master/system/flcore/clients/clientpt.pyr
s
zclientPT.__init__c C sì | ¡ }t ¡ }t | jj¡}| j | j¡ | j ¡ x| jj
¡ D ]
}d|_ qDW x| jj
¡ D ]
}d|_ q`W x¼t | j
ƒD ]®}x¨t|ƒD ]œ\}\}}t|ƒtg ƒkrº|d | j¡|d< n | | j¡}| | j¡}| jròt dt tj ¡ ¡ ¡ | j ¡ | |¡} | | |¡}
|
¡ | j ¡ qˆW qzW | j} | j�rLtj d| d ¡} x | jj
¡ D ] }d|_ �qZW x | jj
¡ D ] }d|_ �qxW xÂt | ƒD ]¶}x®t|ƒD ]¢\}\}}t|ƒtg ƒk�rÔ|d | j¡|d< n | | j¡}| | j¡}| j�rt dt tj ¡ ¡ ¡ | j ¡ | |¡} | | |¡}
|
¡ | j ¡ �q W �q’W | j ¡ t | jj¡} d}
xRt |
¡ |
¡ ƒD ]<\}}|| }t !|dk|t "|¡| ¡}|
t #|¡ }
�q|W | j$d d7 < | j$d t ¡ | 7 < |
S ) NFTr gš™™™™™¹?é é Ú
num_roundsÚ
total_cost)%Úload_train_dataÚtimeÚcopyÚdeepcopyr
r ÚtoÚdeviceÚtrainr r Ú
requires_gradÚranger Ú enumerateÚtypeÚ
train_slowÚsleepÚnpÚabsÚrandomÚrandr Ú zero_gradr ÚbackwardÚstepÚ local_stepsÚrandintr ÚcpuÚzipr
ÚwhereÚ
zeros_likeÚsumÚtrain_time_cost)r Ú trainloaderÚ
start_timeZ
old_promptÚparamr5 ÚiÚxÚyÚoutputr Úmax_local_stepsZ
new_promptZ
diff_provalueÚ new_paramÚ old_paramÚdiff_pror r r r( sh
zclientPT.trainc C s4 x.t | ¡ | jj ¡ ƒD ]\}}|j ¡ |_qW d S )N)r9 r r
r ÚdataÚclone)r r rF rG r r r Úset_parametersY s zclientPT.set_parameters)Ú__name__Ú
__module__Ú __qualname__r r( rK Ú
__classcell__r r )r r r s
Br )
r
Útorch.nnr Únumpyr/ r# Ú system.flcore.clients.clientbaser r$ r r r r r Ú<module> s | 2,568 | Python | .pyt | 40 | 62.775 | 285 | 0.426255 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,875 | clientmoon.cpython-38.pyc | hkgdifyu_pFedPT/system/flcore/clients/__pycache__/clientmoon.cpython-38.pyc | U
”jfcÏ ã @ s^ d dl Z d dlZd dlmZ d dlZd dlZd dlm mZ d dl
m Z G dd„ de ƒZ dS )é N)ÚClientc s, e Zd Z‡ fdd„Zdd„ Zdd„ Z‡ ZS )Ú
clientMOONc sb t ƒ j||||f|� t ¡ | _tjj| j ¡ | j
d�| _ |j | _ |j
| _
d | _t | j¡| _d S )N)Úlr)ÚsuperÚ__init__ÚnnÚCrossEntropyLossÚlossÚtorchÚoptimÚSGDÚmodelÚ
parametersÚ
learning_rateÚ optimizerÚtauÚmuÚ global_modelÚcopyÚdeepcopyÚ old_model)ÚselfÚargsÚidÚ
train_samplesÚ test_samplesÚkwargs©Ú __class__© õL D:\京东\promot\cifar\cifar\Cifar10_iid\system\flcore\clients\clientmoon.pyr s
zclientMOON.__init__c C s$ | ¡ }t ¡ }| j | j¡ | j | j¡ | j | j¡ | j ¡ | j}| j rbt
j d|d ¡}t
|ƒD �]Z}t|ƒD �]J\}\}}t|ƒtg ƒkr¬|d | j¡|d< n | | j¡}| | j¡}| j rät dt
t
j ¡ ¡ ¡ | j ¡ | j |¡}| j |¡} | | |¡}
| j |¡ ¡ } | j |¡ ¡ } t t t || ¡| j ¡t t || ¡| j ¡t t || ¡| j ¡ ¡ }
|
| j t |
¡ 7 }
|
!¡ | j"�rºt#| j|t$|ƒƒ qx| j %¡ qxqj| j &¡ | j &¡ | j &¡ t' (| j¡| _| j)d d7 < | j)d t ¡ | 7 < d S )Né é r gš™™™™™¹?Ú
num_roundsÚ
total_cost)*Úload_train_dataÚtimer
ÚtoÚdevicer r ÚtrainÚ local_stepsÚ
train_slowÚnpÚrandomÚrandintÚrangeÚ enumerateÚtypeÚsleepÚabsÚrandr Ú zero_gradÚbaseÚ predictorr Údetachr
ÚlogÚexpÚFÚcosine_similarityr r ÚmeanÚbackwardÚprivacyÚdp_stepÚlenÚstepÚcpur r Útrain_time_cost)r Ú trainloaderÚ
start_timeÚmax_local_stepsrB ÚiÚxÚyÚrepÚoutputr Zrep_oldZ
rep_globalZloss_conr r r r) sF
P
zclientMOON.trainc C s4 t | ¡ | j ¡ ƒD ]\}}|j ¡ |_q|| _d S )N)Úzipr r
ÚdataÚcloner )r r
Ú new_paramÚ old_paramr r r Úset_parametersO s zclientMOON.set_parameters)Ú__name__Ú
__module__Ú __qualname__r r) rR Ú
__classcell__r r r r r s 2r )
r r
Útorch.nnr Únumpyr, r& Útorch.nn.functionalÚ
functionalr; Úflcore.clients.clientbaser r r r r r Ú<module> s | 2,431 | Python | .pyt | 28 | 85.714286 | 263 | 0.44218 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,876 | clientrod.cpython-39.pyc | hkgdifyu_pFedPT/system/flcore/clients/__pycache__/clientrod.cpython-39.pyc | a
f¾`cı ã @ s€ d dl Z d dlZd dlmZ d dlZd dlZd dlmZ d dl m m
Z d dl m
Z
d dlmZ G dd„ deƒZd
dd „ZdS ) é N)ÚClient)Úlabel_binarize)Úmetricsc s. e Zd Z‡ fdd„Zdd„ Zddd„Z‡ ZS ) Ú clientRODc
s° t ƒ j||||fi |¤� t ¡ | _tjj| j ¡ | j
d�| _ t
| jj¡| _tjj| j ¡ | j
d�| _t | j¡| _| ¡ }|D ](\}}|D ]} | j| ¡ d7 < q�q‚d S )N)Úlré )ÚsuperÚ__init__ÚnnÚCrossEntropyLossÚlossÚtorchÚoptimÚSGDÚmodelÚ
parametersÚ
learning_rateÚ optimizerÚcopyÚdeepcopyÚ predictorÚpredÚopt_predÚzerosÚ num_classesÚsample_per_classÚload_train_dataÚitem)
ÚselfÚargsÚidÚ
train_samplesÚ test_samplesÚkwargsÚ trainloaderÚxÚyÚyy©Ú __class__© úr/media/sim812/391e55df-b6f2-4fe9-a920-53434a8506fa/lgh/pdept/PFL-Non-IID-master/system/flcore/clients/clientrod.pyr
s
zclientROD.__init__c
C sJ | ¡ }t ¡ }| j ¡ | j}| jr8tj d|d ¡}t |ƒD ]Ø}t
|ƒD ]Ê\}\}}t |ƒt g ƒkr~|d | j
¡|d< n | | j
¡}| | j
¡}| j |¡}| j |¡} t|| | jƒ}
| j ¡ |
¡ | j ¡ | | ¡ ¡} | | ¡ | |¡} | j ¡ | ¡ | j ¡ qLq@| jd d7 < | jd t ¡ | 7 < d S )Nr é r Ú
num_roundsÚ
total_cost)r Útimer ÚtrainÚ local_stepsÚ
train_slowÚnpÚrandomÚrandintÚrangeÚ enumerateÚtypeÚtoÚdeviceÚbaser Úbalanced_softmax_lossr r Ú zero_gradÚbackwardÚstepr Údetachr r Útrain_time_cost)
r r$ Ú
start_timeÚmax_local_stepsr? Úir% r& ÚrepÚout_gZloss_bsmÚout_pr r* r* r+ r0 s2
zclientROD.trainNc C sº | ¡ }|d kr| j}| ¡ d}d}g }g }t ¡ ��: |D �]"\}}t|ƒtg ƒkrn|d | j¡|d< n | | j¡}| | j¡}| j |¡} | j | ¡}
|
| ¡ ¡} |
¡ | } |t tj
| dd�|k¡ ¡ 7 }||jd 7 }| t | ¡ ¡ ¡ ¡ ¡ | j}
| jdk�r |
d7 }
t| ¡ ¡ ¡ t |
¡ƒ}| jdk�rX|d d …d d…f }| |¡ q>W d ƒ n1 �sz0 Y tj|dd�}tj|dd�}tj||dd�}|||fS )Nr r )Údimr, )ÚaxisÚmicro)Úaverage) Úload_test_datar Úevalr
Úno_gradr8 r9 r: r; r r r@ ÚsumÚargmaxr ÚshapeÚappendÚFÚsoftmaxÚcpuÚnumpyr r r3 ÚarangeÚ concatenater Ú
roc_auc_score)r r Ú
testloaderÚtest_accÚtest_numÚy_probÚy_truer% r& rE rF rG ÚoutputÚncÚlbÚaucr* r* r+ Ú test_metricsB s@ ,zclientROD.test_metrics)N)Ú__name__Ú
__module__Ú __qualname__r r0 rc Ú
__classcell__r* r* r( r+ r s %r Úmeanc C sB | |¡}| d¡ |jd d¡}|| ¡ }tj|| |d�}|S )a} Compute the Balanced Softmax Loss between `logits` and the ground truth `labels`.
Args:
labels: A int tensor of size [batch].
logits: A float tensor of size [batch, no_of_classes].
sample_per_class: A int tensor of size [no of classes].
reduction: string. One of "none", "mean", "sum"
Returns:
loss: A float tensor. Balanced Softmax Loss.
r éÿÿÿÿ)ÚinputÚtargetÚ reduction)Útype_asÚ unsqueezeÚexpandrQ ÚlogrS Ú
cross_entropy)ÚlabelsÚlogitsr rl Zspcr r* r* r+ r< n s
r< )rh )r r
Útorch.nnr
rV r3 r/ Úflcore.clients.clientbaser Útorch.nn.functionalÚ
functionalrS Úsklearn.preprocessingr Úsklearnr r r< r* r* r* r+ Ú<module> s b | 3,780 | Python | .pyt | 38 | 97.210526 | 523 | 0.456708 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,877 | clientmtlpt.cpython-38.pyc | hkgdifyu_pFedPT/system/flcore/clients/__pycache__/clientmtlpt.cpython-38.pyc | U
GçýcÓ ã @ s| d dl Z d dlmZ d dlmZ d dlZd dlZd dlZd dl Z d dl Z d dl
m Z d dl m
Z
G dd„ deƒZdd„ ZdS ) é N)ÚClient)Úlabel_binarize)Úmetricsc s< e Zd Z‡ fdd„Zdd„ Zdd„ Zdd„ Zd d
„ Z‡ ZS ) Ú clientMTLPTc s¨ t ƒ j||||f|Ž d | _d | _d| _|j| _d| _t ¡ | _ t
j | j
j ¡ | jdœg¡| _|j| _t
j j | j
j ¡ |j|jd�| _t
j jj| j| j|jd�| _d S )Nr g-C ëâ6?)ÚparamsÚlr)r Úmomentum)Ú step_sizeÚgamma)ÚsuperÚ__init__ÚomegaÚW_globÚidxÚitkÚlambaÚnnÚCrossEntropyLossÚlossÚtorchÚoptimÚSGDÚmodelÚbaseÚ
parametersÚ
learning_rateÚ optimizerÚ plocal_stepsÚ generatorÚpt_learning_rater Ú
poptimizerÚ lr_schedulerÚStepLRÚlearning_decayÚ scheduler)ÚselfÚargsÚidÚ
train_samplesÚ test_samplesÚkwargs©Ú __class__© õY D:\京东\promot\第二次投稿\实验\native - pro\system\flcore\clients\clientmtlpt.pyr
s"
ÿÿÿzclientMTLPT.__init__c C s` | ¡ }t ¡ }t | jj¡}| j | j¡ | j ¡ | jj
¡ D ]
}d|_ qB| jj
¡ D ]
}d|_ qZt | j
ƒD ]´}t|ƒD ]¦\}\}}t|ƒtg ƒkr®|d | j¡|d< n | | j¡}| | j¡}| jræt dt tj ¡ ¡ ¡ | j ¡ | |¡} | | |¡}
|
¡ | j ¡ | j ¡ q|qp| j } | j�rFtj d| d ¡} | jj
¡ D ] }d|_ �qR| jj
¡ D ] }d|_ �qlt | ƒD �]2}|D �]$\}}t|ƒtg ƒk�r¾|d | j¡|d< n | | j¡}| | j¡}| j�røt dt tj ¡ ¡ ¡ | j ¡ | |¡} | | |¡}
t | jƒ| j d d …| j!f< d} | | j "¡ d 7 } | t# $t# $| j | j% d¡d ¡7 } t&t' (| j j)d ¡d ƒd }
| d|
9 } |
| 7 }
|
¡ | j ¡ �qŒ�q‚| j *¡ d | _%d | _ t | jj¡}d}t+|
¡ |
¡ ƒD ]<\}}|| }t# ,|dk|t# -|¡| ¡}|t# $|¡ }�qò| j.d d7 < | j.d t ¡ | 7 < |S )
NFTr gš™™™™™¹?é é é
Ú
num_roundsÚ
total_cost)/Úload_train_dataÚtimeÚcopyÚdeepcopyr r ÚtoÚdeviceÚtrainr r Ú
requires_gradÚranger Ú enumerateÚtypeÚ
train_slowÚsleepÚnpÚabsÚrandomÚrandr Ú zero_gradr ÚbackwardÚstepr$ Ú local_stepsÚrandintr Úflattenr r Únormr Úsumr
ÚintÚmathÚlog10ÚshapeÚcpuÚzipÚwhereÚ
zeros_likeÚtrain_time_cost)r% Ú trainloaderÚ
start_timeÚ
old_promptÚparamrG ÚiÚxÚyÚoutputr Úmax_local_stepsÚloss_regularizerÚfÚ
new_promptÚ
diff_provalueÚ new_paramÚ old_paramÚdiff_pror- r- r. r: # s|
"
zclientMTLPT.trainc C s* t |d d ¡| _t |¡| _|| _d S )Nr )r Úsqrtr
r6 r7 r r )r% r r
r r- r- r. Úreceive_valuess s zclientMTLPT.receive_valuesc C s2 t |j ¡ | jj ¡ ƒD ]\}}|j ¡ |_qd S )N)rR r r r ÚdataÚclone)r% r rc rd r- r- r. Úset_parametersw s zclientMTLPT.set_parametersc C sv | ¡ }| j | j¡ | j ¡ d}d}d}g }g }t ¡ �ò |D ]æ\}}t|ƒtg ƒkrp|d | j¡|d< n | | j¡}| | j¡}| |¡} | j |¡}
|t tj
| dd�|k¡ ¡ 7 }|t tj
|
dd�|k¡ ¡ 7 }||j d 7 }|
| ¡ ¡ ¡ ¡ |
t| ¡ ¡ ¡ t | j¡d�¡ qBW 5 Q R X | j ¡ tj|dd�}tj|dd�}tj||dd�} |||| fS )Nr r/ )Údim)Úclasses)ÚaxisÚmicro)Úaverage)Úload_test_datar r8 r9 Úevalr Úno_gradr> r rL ÚargmaxÚitemrP ÚappendÚdetachrQ Únumpyr rA ÚarangeÚ num_classesÚ concatenater Ú
roc_auc_score) r% ÚtestloaderfullÚtest_accÚ test_acc2Útest_numÚy_probÚy_truer[ r\ r] Úoutput2Úaucr- r- r. Ú test_metrics{ s4
2
zclientMTLPT.test_metrics) Ú__name__Ú
__module__Ú __qualname__r r: rg rj r„ Ú
__classcell__r- r- r+ r. r s
Pr c s, | ¡ ‰ ˆ ¡ }‡ fdd„|D ƒ}t |¡S )Nc s g | ]}ˆ | ¡ ‘qS r- )rJ )Ú.0Úkey©Ú
state_dictr- r. Ú
<listcomp>¦ s zflatten.<locals>.<listcomp>)rŒ Úkeysr Úcat)r rŽ ÚWr- r‹ r. rJ £ s rJ )r Útorch.nnr Úflcore.clients.clientbaser rw rA r5 rN r6 Úsklearn.preprocessingr Úsklearnr r rJ r- r- r- r. Ú<module> s | 4,639 | Python | .pyt | 54 | 84.703704 | 653 | 0.41147 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,878 | clientper.cpython-38.pyc | hkgdifyu_pFedPT/system/flcore/clients/__pycache__/clientper.cpython-38.pyc | U
”jfc€ ã @ sL d dl Z d dlZd dlmZ d dlZd dlZd dlmZ G dd„ deƒZ dS )é N)ÚClientc s, e Zd Z‡ fdd„Zdd„ Zdd„ Z‡ ZS )Ú clientPerc s> t ƒ j||||f|� t ¡ | _tjj| j ¡ | j
d�| _ d S )N)Úlr) ÚsuperÚ__init__ÚnnÚCrossEntropyLossÚlossÚtorchÚoptimÚSGDÚmodelÚ
parametersÚ
learning_rateÚ optimizer)ÚselfÚargsÚidÚ
train_samplesÚ test_samplesÚkwargs©Ú __class__© õK D:\京东\promot\cifar\cifar\Cifar10_iid\system\flcore\clients\clientper.pyr
s
zclientPer.__init__c
C s4 | ¡ }t ¡ }| j | j¡ | j ¡ | j}| jrFtj
d|d ¡}t |ƒD ]ª}t |ƒD ]œ\}\}}t
|ƒt
g ƒkrŒ|d | j¡|d< n | | j¡}| | j¡}| jrÄt dt tj ¡ ¡ ¡ | j ¡ | |¡}| ||¡} | ¡ | j ¡ qZqN| j ¡ | jd d7 < | jd t ¡ | 7 < d S )Né é r gš™™™™™¹?Ú
num_roundsÚ
total_cost)Úload_train_dataÚtimer
ÚtoÚdeviceÚtrainÚ local_stepsÚ
train_slowÚnpÚrandomÚrandintÚrangeÚ enumerateÚtypeÚsleepÚabsÚrandr Ú zero_gradr ÚbackwardÚstepÚcpuÚtrain_time_cost)
r Ú trainloaderÚ
start_timeÚmax_local_stepsr1 ÚiÚxÚyÚoutputr r r r r# s.
zclientPer.trainc C s0 t | ¡ | jj ¡ ƒD ]\}}|j ¡ |_qd S )N)Úzipr r
ÚbaseÚdataÚclone)r r
Ú new_paramÚ old_paramr r r Úset_parameters0 s zclientPer.set_parameters)Ú__name__Ú
__module__Ú __qualname__r r# rA Ú
__classcell__r r r r r s r )
Úcopyr
Útorch.nnr Únumpyr& r Úflcore.clients.clientbaser r r r r r Ú<module> s | 1,849 | Python | .pyt | 20 | 91.3 | 302 | 0.457923 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,879 | clientmtl.cpython-39.pyc | hkgdifyu_pFedPT/system/flcore/clients/__pycache__/clientmtl.cpython-39.pyc | a
f¾`c®
ã @ s\ d dl Z d dlmZ d dlmZ d dlZd dlZd dlZd dl Z G dd„ deƒZ
dd„ Z dS )é N)ÚClientc s, e Zd Z‡ fdd„Zdd„ Zdd„ Z‡ ZS )Ú clientMTLc sd t ƒ j||||fi |¤� d | _d | _d| _|j| _d| _t ¡ | _ t
j j | j
¡ | jdd�| _d S )Nr g-C ëâ6?g à?)ÚlrÚmomentum)ÚsuperÚ__init__ÚomegaÚW_globÚidxZitkZlambaÚnnÚCrossEntropyLossÚlossÚtorchÚoptimÚSGDÚmodelÚ
parametersÚ
learning_rateÚ optimizer)ÚselfÚargsÚidÚ
train_samplesÚ test_samplesÚkwargs©Ú __class__© úr/media/sim812/391e55df-b6f2-4fe9-a920-53434a8506fa/lgh/pdept/PFL-Non-IID-master/system/flcore/clients/clientmtl.pyr s
zclientMTL.__init__c C sª | ¡ }t ¡ }| j ¡ | j}| jr8tj d|d ¡}t |ƒD �]*}|D �] \}}t
|ƒt
g ƒkrz|d | j ¡|d< n | | j ¡}| | j ¡}| jr²t
dt tj ¡ ¡ ¡ | j ¡ | |¡}| ||¡}t| jƒ| jd d …| jf< d} | | j ¡ d 7 } | t t | j| j d¡d ¡7 } tt | jj d ¡d ƒd }
| d|
9 } || 7 }| ¡ | j ¡ qJq@d | _d | _| j d d7 < | j d t ¡ | 7 < d S )Né é r gš™™™™™¹?é
Ú
num_roundsÚ
total_cost)!Úload_train_dataÚtimer ÚtrainÚ local_stepsÚ
train_slowÚnpÚrandomÚrandintÚrangeÚtypeÚtoÚdeviceÚsleepÚabsÚrandr Ú zero_gradr
Úflattenr r
Únormr Úsumr ÚintÚmathÚlog10ÚshapeÚbackwardÚstepÚtrain_time_cost) r Ú trainloaderÚ
start_timeÚmax_local_stepsr< ÚxÚyÚoutputr
Zloss_regularizerÚfr r r r& s<
" zclientMTL.trainc C s* t |d d ¡| _t |¡| _|| _d S )Nr )r Úsqrtr ÚcopyÚdeepcopyr r
)r r r r
r r r Úreceive_valuesH s zclientMTL.receive_values)Ú__name__Ú
__module__Ú __qualname__r r& rH Ú
__classcell__r r r r r
s 1r c s, | ¡ ‰ ˆ ¡ }‡ fdd„|D ƒ}t |¡S )Nc s g | ]}ˆ | ¡ ‘qS r )r4 )Ú.0Úkey©Ú
state_dictr r Ú
<listcomp>Q ó zflatten.<locals>.<listcomp>)rP Úkeysr Úcat)r rS ÚWr rO r r4 N s r4 ) r Útorch.nnr Úflcore.clients.clientbaser Únumpyr) r% r8 rF r r4 r r r r Ú<module> s D | 2,536 | Python | .pyt | 26 | 96.461538 | 274 | 0.430506 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,880 | clientperpt.cpython-38.pyc | hkgdifyu_pFedPT/system/flcore/clients/__pycache__/clientperpt.cpython-38.pyc | U
aÿcP ã @ sl d dl Z d dlZd dlmZ d dlZd dlZd dlmZ d dl Z d dl m
Z
d dl m Z G dd„ deƒZ
dS )é N)ÚClient)Úlabel_binarize)Úmetricsc s4 e Zd Z‡ fdd„Zdd„ Zdd„ Zdd„ Z‡ ZS ) Ú clientPerPTc sš t ƒ j||||f|� t ¡ | _tj | jj
¡ | j dœ| jj
¡ | j dœg¡| _
|j| _tjj| jj
¡ |j|jd�| _tjjj| j| j|jd�| _d S )N)ÚparamsÚlr)r Úmomentum)Ú step_sizeÚgamma)ÚsuperÚ__init__ÚnnÚCrossEntropyLossÚlossÚtorchÚoptimÚSGDÚmodelÚbaseÚ
parametersÚ
learning_rateÚ predictorÚ optimizerÚ plocal_stepsÚ generatorÚpt_learning_rater Ú
poptimizerÚ lr_schedulerÚStepLRÚlearning_decayÚ scheduler)ÚselfÚargsÚidÚ
train_samplesÚ test_samplesÚkwargs©Ú __class__© õc D:\京东\promot\第二次投稿\å®�验\æœ�务器\native - pro\system\flcore\clients\clientperpt.pyr s
ş ÿzclientPerPT.__init__c C s0 | ¡ }t ¡ }t | jj¡}| j | j¡ | j ¡ | jj
¡ D ]
}d|_ qB| jj
¡ D ]
}d|_ qZ| jj
¡ D ]
}d|_ qrt
| jƒD ]´}t|ƒD ]¦\}\}}t|ƒtg ƒkrÆ|d | j¡|d< n | | j¡}| | j¡}| jrşt dt tj ¡ ¡ ¡ | j ¡ | |¡} | | |¡}
|
¡ | j ¡ | j ¡ q”qˆ| j } | j�r^tj d| d ¡} | jj
¡ D ] }d|_ �qj| jj
¡ D ] }d|_ �q„| jj
¡ D ] }d|_ �q�t
| ƒD ]²}t|ƒD ]¢\}\}}t|ƒtg ƒk�rô|d | j¡|d< n | | j¡}| | j¡}| j�r.t dt tj ¡ ¡ ¡ | j ¡ | |¡} | | |¡}
|
¡ | j ¡ �qÀ�q´| j ¡ t | jj¡} d}
| j!d d7 < | j!d t ¡ | 7 < t"|
¡ |
¡ ƒD ]<\}}|| }t# $|dk|t# %|¡| ¡}|
t# &|¡ }
�qÂ| j!d d7 < | j!d t ¡ | 7 < |
S ) NFTr gš™™™™™¹?é é Ú
num_roundsÚ
total_cost)'Úload_train_dataÚtimeÚcopyÚdeepcopyr r ÚtoÚdeviceÚtrainr r Ú
requires_gradr Úranger Ú enumerateÚtypeÚ
train_slowÚsleepÚnpÚabsÚrandomÚrandr Ú zero_gradr ÚbackwardÚstepr Ú local_stepsÚrandintr ÚcpuÚtrain_time_costÚzipr ÚwhereÚ
zeros_likeÚsum)r! Ú trainloaderÚ
start_timeÚ
old_promptÚparamrB ÚiÚxÚyÚoutputr Úmax_local_stepsÚ
new_promptÚ
diff_provalueÚ new_paramÚ old_paramÚdiff_pror) r) r* r5 sv
zclientPerPT.trainc C s2 t |j ¡ | jj ¡ ƒD ]\}}|j ¡ |_qd S )N)rG r r r ÚdataÚclone)r! r rV rW r) r) r* Úset_parameters^ s zclientPerPT.set_parametersc C s~ | ¡ }| j | j¡ | j ¡ d}d}d}g }g }t ¡ �ú |D ]î\}}t|ƒtg ƒkrp|d | j¡|d< n | | j¡}| | j¡}| |¡} | j | j |¡¡}
|t
tj | dd�|k¡ ¡ 7 }|t
tj |
dd�|k¡ ¡ 7 }||j
d 7 }| | ¡ ¡ ¡ ¡ | t| ¡ ¡ ¡ t | j¡d�¡ qBW 5 Q R X | j ¡ tj|dd�}tj|dd�}tj||dd�} |||| fS )Nr r+ )Údim)Úclasses)ÚaxisÚmicro)Úaverage)Úload_test_datar r3 r4 Úevalr Úno_gradr9 r r rJ ÚargmaxÚitemÚshapeÚappendÚdetachrE Únumpyr r< ÚarangeÚ num_classesÚ concatenater Ú
roc_auc_score) r! ÚtestloaderfullÚtest_accÚ test_acc2Útest_numÚy_probÚy_truerP rQ rR Úoutput2Úaucr) r) r* Ú test_metricsa s4
2
zclientPerPT.test_metrics)Ú__name__Ú
__module__Ú __qualname__r r5 r[ rv Ú
__classcell__r) r) r' r* r
s Br )r1 r Útorch.nnr
ri r< r0 Úflcore.clients.clientbaser Úsklearn.preprocessingr Úsklearnr r r) r) r) r* Ú<module> s | 3,900 | Python | .pyt | 52 | 73.557692 | 558 | 0.423227 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,881 | clientproxpt.cpython-38.pyc | hkgdifyu_pFedPT/system/flcore/clients/__pycache__/clientproxpt.cpython-38.pyc | U
Ñıc ã @ sx d dl Z d dlZd dlZd dlZd dlmZ d dlmZ d dl m
Z
d dlZd dl m Z d dl
mZ G dd„ de
ƒZdS )é N)ÚPerturbedGradientDescent)ÚClient)Úlabel_binarize)Úmetricsc s4 e Zd Z‡ fdd„Zdd„ Zdd„ Zdd„ Z‡ ZS ) Ú clientProxPTc sª t ƒ j||||f|� |j| _t t| jj ¡ ƒ¡| _ t
¡ | _ t
| jj ¡ | jdœg| jd�| _|j| _tjj| jj ¡ |j|jd�| _tjjj| j| j|jd�| _d S )N)ÚparamsÚlr)Úmu)r Úmomentum)Ú step_sizeÚgamma) ÚsuperÚ__init__r ÚcopyÚdeepcopyÚlistÚmodelÚbaseÚ
parametersÚ
global_paramsÚnnÚCrossEntropyLossÚlossr Ú
learning_rateÚ optimizerÚ plocal_stepsÚtorchÚoptimÚSGDÚ generatorZpt_learning_rater
Ú
poptimizerÚ lr_schedulerÚStepLRZlearning_decayÚ scheduler)ÚselfÚargsÚidÚ
train_samplesÚ test_samplesÚkwargs©Ú __class__© õZ D:\京东\promot\第二次投稿\å®�验\native - pro\system\flcore\clients\clientproxpt.pyr
s
ÿıÿÿzclientProxPT.__init__c C s° | ¡ }t ¡ }t | jj¡}| j | j¡ | j ¡ | jj
¡ D ]
}d|_ qB| jj
¡ D ]
}d|_ qZt | j
ƒD ]´}t|ƒD ]¦\}\}}t|ƒtg ƒkr®|d | j¡|d< n | | j¡}| | j¡}| jræt dt tj ¡ ¡ ¡ | j ¡ | |¡} | | |¡}
|
¡ | j ¡ | j ¡ q|qp| j } | j�rFtj d| d ¡} | jj
¡ D ] }d|_ �qR| jj
¡ D ] }d|_ �qlt | ƒD ]�}|D ]„\}}t|ƒtg ƒk�rº|d | j¡|d< n | | j¡}| | j¡}| j ¡ | |¡} | | |¡}
|
¡ | j | j | j¡ �qŠ�q‚| j ¡ t | jj¡} d}
t!|
¡ |
¡ ƒD ]<\}}|| }t" #|dk|t" $|¡| ¡}|
t" %|¡ }
�qB| j&d d7 < | j&d t ¡ | 7 < |
S ) NFTr gš™™™™™¹?é é Ú
num_roundsÚ
total_cost)'Úload_train_dataÚtimer r r r ÚtoÚdeviceÚtrainr r Ú
requires_gradÚranger Ú enumerateÚtypeÚ
train_slowÚsleepÚnpÚabsÚrandomÚrandr Ú zero_gradr ÚbackwardÚstepr# Ú local_stepsÚrandintr r ÚcpuÚzipr ÚwhereÚ
zeros_likeÚsumÚtrain_time_cost)r$ Ú trainloaderÚ
start_timeZ
old_promptÚparamrC ÚiÚxÚyÚoutputr Úmax_local_stepsZ
new_promptZ
diff_provalueÚ new_paramÚ old_paramÚdiff_pror, r, r- r6 " sf
zclientProxPT.trainc C sD t |j ¡ | j| jj ¡ ƒD ]"\}}}|j ¡ |_|j ¡ |_q d S )N)rG r r r r ÚdataÚclone)r$ r rT Ú global_paramrN r, r, r- Úset_parametersa s & zclientProxPT.set_parametersc C sv | ¡ }| j | j¡ | j ¡ d}d}d}g }g }t ¡ �ò |D ]æ\}}t|ƒtg ƒkrp|d | j¡|d< n | | j¡}| | j¡}| |¡} | j |¡}
|t tj
| dd�|k¡ ¡ 7 }|t tj
|
dd�|k¡ ¡ 7 }||j d 7 }|
| ¡ ¡ ¡ ¡ |
t| ¡ ¡ ¡ t | j¡d�¡ qBW 5 Q R X | j ¡ tj|dd�}tj|dd�}tj||dd�} |||| fS )Nr r. )Údim)Úclasses)ÚaxisÚmicro)Úaverage)Úload_test_datar r4 r5 Úevalr Úno_gradr: r rJ ÚargmaxÚitemÚshapeÚappendÚdetachrF Únumpyr r= ÚarangeÚ num_classesÚ concatenater Ú
roc_auc_score) r$ ÚtestloaderfullÚtest_accÚ test_acc2Útest_numÚy_probÚy_truerP rQ rR Úoutput2Úaucr, r, r- Ú test_metricse s4
2
zclientProxPT.test_metrics)Ú__name__Ú
__module__Ú __qualname__r r6 rZ ru Ú
__classcell__r, r, r* r- r s ?r )r rh r= r3 r Útorch.nnr Ú flcore.optimizers.fedoptimizerr Úflcore.clients.clientbaser Úsklearn.preprocessingr Úsklearnr r r, r, r, r- Ú<module> s | 3,919 | Python | .pyt | 47 | 82.085106 | 553 | 0.437387 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,882 | clientproto.cpython-38.pyc | hkgdifyu_pFedPT/system/flcore/clients/__pycache__/clientproto.cpython-38.pyc | U
”jfc
ã @ s` d dl mZ d dlZd dlZd dlmZ d dlZd dlZd dl m
Z
G dd„ de
ƒZ dd„ Z dS )é )Ú defaultdictN)ÚClientc s> e Zd Z‡ fdd„Zdd„ Zdd„ Zdd„ Zd d
d „Z‡ ZS )
Ú clientProtoc sx t ƒ j||||f|� t ¡ | _tjj| j ¡ | j
d�| _ t | jj
¡ ƒd jd | _d | _d | _t ¡ | _|j| _d S )N)Úlrr é )ÚsuperÚ__init__ÚnnÚCrossEntropyLossÚlossÚtorchÚoptimÚSGDÚmodelÚ
parametersÚ
learning_rateÚ optimizerÚlistÚ predictorÚshapeZ feature_dimÚprotosÚ
global_protosÚMSELossÚloss_mseÚlamda)ÚselfÚargsÚidÚ
train_samplesÚ test_samplesÚkwargs©Ú __class__© õM D:\京东\promot\cifar\cifar\Cifar10_iid\system\flcore\clients\clientproto.pyr s
zclientProto.__init__c C sô | ¡ }t ¡ }| j | j¡ | j ¡ | j}| jrFtj
d|d ¡}t t ƒ}t
|ƒD �]V}t|ƒD �]F\}\}}t|ƒtg ƒkr˜|d | j¡|d< n | | j¡}| | j¡}| jrĞt dt tj ¡ ¡ ¡ | j ¡ | j |¡} | j | ¡}
| |
|¡} | jd k�r^t | ¡} t|ƒD ]*\}}
|
¡ }| j| j | |d d …f< �q | | | | ¡| j 7 } t|ƒD ]2\}}
|
¡ }|| | |d d …f ¡ j ¡ �qf| !¡ | j "¡ qdqV| j #¡ t$|ƒ| _%| j&d d7 < | j&d t ¡ | 7 < d S )Nr é r çš™™™™™¹?Ú
num_roundsÚ
total_cost)'Úload_train_dataÚtimer ÚtoÚdeviceÚtrainÚ local_stepsÚ
train_slowÚnpÚrandomÚrandintr r ÚrangeÚ enumerateÚtypeÚsleepÚabsÚrandr Ú zero_gradÚbaser r r r Ú
zeros_likeÚitemÚdatar r ÚappendÚdetachÚbackwardÚstepÚcpuÚagg_funcr Útrain_time_cost)r Ú trainloaderÚ
start_timeÚmax_local_stepsr rA ÚiÚxÚyÚrepÚoutputr Z proto_newÚyyÚy_cr# r# r$ r- sF
$
zclientProto.trainc C s t |¡| _d S )N)ÚcopyÚdeepcopyr )r r r# r# r$ Ú
set_protosN s zclientProto.set_protosc C s | ¡ }| j | j¡ | j ¡ ttƒ}t ¡ �Ê t |ƒD ]º\}\}}t
|ƒt
g ƒkrl|d | j¡|d< n | | j¡}| | j¡}| j r¤t
dt tj ¡ ¡ ¡ | j ¡ | j |¡}t |ƒD ]0\}}| ¡ }|| ||d d …f ¡ j¡ qÂq:W 5 Q R X t|ƒ| _| j ¡ d S )Nr r& ) r) r r+ r, Úevalr r r Úno_gradr4 r5 r/ r* r6 r0 r7 r1 r8 r r9 r: r< r> r? r= rC r rB ) r rE r rH rI rJ rK rM rN r# r# r$ Úcollect_protosQ s&
.
zclientProto.collect_protosNc
C s@ | ¡ }|d kr| j}| j | j¡ | ¡ d}d}t ¡ �ê |D ]Ş\}}t|ƒtg ƒkrp|d | j¡|d< n | | j¡}| | j¡}| j |¡}t dƒt
|j d | j ¡ | j¡ }t
|ƒD ]0\} }
| j ¡ D ] \} } | |
| ¡|| | f< qÒqÀ|t tj|dd�|k¡ ¡ 7 }||j d 7 }qBW 5 Q R X | j ¡ ||dfS )Nr Úinfr )Údim)Úload_test_datar r+ r, rR r rS r5 r: ÚfloatÚonesr Ú num_classesr4 r Úitemsr ÚsumÚargminr< rB )
r r Ú
testloaderÚtest_accÚtest_numrI rJ rK rL rH ÚrÚjÚpror# r# r$ Ú test_metricsj s,
$
zclientProto.test_metrics)N) Ú__name__Ú
__module__Ú __qualname__r r- rQ rT rd Ú
__classcell__r# r# r! r$ r s
3r c C sb | ¡ D ]T\}}t|ƒdkrPd|d j }|D ]}||j7 }q.|t|ƒ | |< q|d | |< q| S )z-
Returns the average of the weights.
r r )r[ Úlenr= )r ÚlabelÚ
proto_listÚprotorH r# r# r$ rC ‰ s rC )
Ú collectionsr rO r Útorch.nnr Únumpyr0 r* Úflcore.clients.clientbaser r rC r# r# r# r$ Ú<module> s ~ | 3,997 | Python | .pyt | 41 | 96.195122 | 455 | 0.392974 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,883 | clientfomo.cpython-37.pyc | hkgdifyu_pFedPT/system/flcore/clients/__pycache__/clientfomo.cpython-37.pyc | B
t fcà ã @ sd d dl Z d dlmZ d dlZd dlZd dlZd dlmZ d dl m
Z
d dl m Z G dd„ deƒZ
dS )é N)ÚClient)Ú
DataLoader)Úread_client_datac sn e Zd Z‡ fdd„Zdd„ Zddd„Zdd „ Zd
d „ Zd d
„ Zdd„ Z dd„ Z
dd„ Z dd„ Z dd„ Z
‡ ZS )Ú
clientFomoc sŒ t ƒ j||||f|� |j| _t | j¡| _g | _g | _t j
| j| j d�| _ t
¡ | _t jj| j ¡ | jd�| _d| _| jd| j | _d S )N)Údevice)Úlrgš™™™™™É?é )ÚsuperÚ__init__Ú num_clientsÚcopyÚdeepcopyÚmodelÚ old_modelÚ received_idsÚreceived_modelsÚtorchÚzerosr Ú
weight_vectorÚnnÚCrossEntropyLossÚlossÚoptimÚSGDÚ
parametersÚ
learning_rateÚ optimizerÚ val_ratioÚ
train_samples)ÚselfÚargsÚidr Ú test_samplesÚkwargs)Ú __class__© úG/root/autodl-tmp/PFL-Non-IID-master/system/flcore/clients/clientfomo.pyr
s
zclientFomo.__init__c
C sR | ¡ \}}t ¡ }| |¡ | | j| j¡ | j | j¡ | j ¡ | j }| j
rdt j
d|d ¡}x²t|ƒD ]¦}x |D ]˜\}}t|ƒtg ƒkr¦|d | j¡|d< n | | j¡}| | j¡}| j
rŞt dt t j ¡ ¡ ¡ | j ¡ | |¡}| ||¡} | ¡ | j ¡ qxW qnW | j ¡ | jd d7 < | jd t ¡ | 7 < d S )Nr é r gš™™™™™¹?Ú
num_roundsÚ
total_cost)Úload_train_dataÚtimeÚaggregate_parametersÚ clone_modelr r Útor ÚtrainÚ local_stepsÚ
train_slowÚnpÚrandomÚrandintÚrangeÚtypeÚsleepÚabsÚrandr Ú zero_gradr ÚbackwardÚstepÚcpuÚtrain_time_cost)
r Ú trainloaderÚ
val_loaderÚ
start_timeÚmax_local_stepsr< ÚxÚyÚoutputr r% r% r& r/ s2
zclientFomo.trainNc C sz |d kr| j }t| j| jdd�}t| jt|ƒ ƒ }||d … }|d |… }t|| j ddd�}t|| j | jdd�}||fS )NT)Úis_train)Ú drop_lastÚshuffle) Ú
batch_sizer Údatasetr! Úintr Úlenr Ú
has_BatchNorm)r rI Ú
train_dataÚval_idxZval_datar? r@ r% r% r& r* ? s zclientFomo.load_train_datac C sÄ | ¡ \}}| j | j¡ | j ¡ d}d}x„|D ]|\}}t|ƒtg ƒkr`|d | j¡|d< n | | j¡}| | j¡}| |¡}||jd 7 }|| ||¡ ¡ |jd 7 }q2W | j ¡ ||fS )Nr )
r* r r. r Úevalr6 Úshaper Úitemr= )r r? r@ Ú train_numr rC rD rE r% r% r& Ú
train_metricsL s
"
zclientFomo.train_metricsc C s || _ || _d S )N)r r )r ÚidsÚmodelsr% r% r& Úreceive_modelsc s zclientFomo.receive_modelsc C s� g }| | j|¡}xv| jD ]l}g }x4t| ¡ | j ¡ ƒD ] \}}| || d¡¡ q8W t |¡}| || ||¡ t |¡d ¡ qW |
|¡ t |¡S )Néÿÿÿÿgñh㈵øä>) Úrecalculate_lossr r Úzipr ÚappendÚviewr ÚcatÚnormÚweight_vector_updateÚtensor)r r@ Ú weight_listÚLÚreceived_modelZ
params_difZparam_nZparam_ir% r% r& Ú
weight_calg s
(
zclientFomo.weight_calc C sX t | j¡| _x.t|| jƒD ] \}}| j| | ¡ 7 < q W t | j¡ | j
¡| _d S )N) r2 r r r rZ r rR r r` r. r )r ra Úwr! r% r% r& r_ ‚ s z clientFomo.weight_vector_updatec C s” d}| | j¡ xn|D ]f\}}t|ƒtg ƒkrD|d | j¡|d< n | | j¡}| | j¡}||ƒ}| ||¡}|| ¡ 7 }qW | ¡ |t|ƒ S )Nr )r. r r6 r rR r= rL )r Z new_modelr@ rb rC rD rE r r% r% r& rY Œ s zclientFomo.recalculate_lossc C s> x8t | j ¡ | ¡ ƒD ] \}}| j|j ¡ | 7 _qW d S )N)rZ r r ÚdataÚclone)r re rc ÚparamZreceived_paramr% r% r& Úadd_parameters› s zclientFomo.add_parametersc C sd | | |¡¡}t|ƒdkr`x | j ¡ D ]}|j ¡ q(W x$t|| jƒD ]\}}| ||¡ qHW d S )Nr )
Ú weight_scalerd rL r r rf Úzero_rZ r ri )r r@ Úweightsrh re rc r% r% r& r, Ÿ s z clientFomo.aggregate_parametersc sN t |t d¡¡}t |¡‰ ˆ dkr@‡ fdd„|D ƒ}t |¡S t g ¡S d S )Nr c s g | ] }|ˆ ‘qS r% r% )Ú.0re )Úw_sumr% r& ú
<listcomp> s z+clientFomo.weight_scale.<locals>.<listcomp>)r Úmaximumr` Úsum)r rl r% )rn r& rj © s
zclientFomo.weight_scale)N)Ú__name__Ú
__module__Ú __qualname__r
r/ r* rT rW rd r_ rY ri r, rj Ú
__classcell__r% r% )r$ r& r s #
r )r Útorch.nnr Únumpyr2 r+ r Úflcore.clients.clientbaser Útorch.utils.datar Úutils.data_utilsr r r% r% r% r& Ú<module> s | 5,079 | Python | .pyt | 47 | 106.978723 | 876 | 0.407709 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,884 | clientdynpt.cpython-38.pyc | hkgdifyu_pFedPT/system/flcore/clients/__pycache__/clientdynpt.cpython-38.pyc | U
”jfcè ã @ sT d dl Z d dlZd dlmZ d dlZd dlZd dlmZ G dd„ deƒZ dd„ Z
dS )é N)ÚClientc s, e Zd Z‡ fdd„Zdd„ Zdd„ Z‡ ZS )Ú clientDynPTc s¬ t ƒ j||||f|� t ¡ | _tjj| j ¡ | j
d�| _ |j | _ d | _
t | j¡}t|ƒ}t |¡| _tjj| jj ¡ | j
d�| _tjj| jj ¡ | j
d�| _ |j| _d S )N)Úlr)ÚsuperÚ__init__ÚnnÚCrossEntropyLossÚlossÚtorchÚoptimÚSGDÚmodelÚ
parametersÚ
learning_rateÚ optimizerÚalphaÚglobal_model_vectorÚcopyÚdeepcopyÚmodel_parameter_vectorÚ
zeros_likeÚold_gradÚ generatorÚ
poptimizerÚbaseÚ plocal_steps)ÚselfÚargsÚidÚ
train_samplesÚ test_samplesÚkwargsr ©Ú __class__© õM D:\京东\promot\cifar\cifar\Cifar10_iid\system\flcore\clients\clientdynpt.pyr
s
zclientDynPT.__init__c C s¨ | ¡ }t ¡ }t | jj¡}| j | j¡ | j | j¡| _| j | j¡| _ | j
¡ | jj ¡ D ]
}d|_
qb| jj ¡ D ]
}d|_
qzt| jƒD ]¬}t|ƒD ]�\}\}}t|ƒtg ƒkrÎ|d | j¡|d< n | | j¡}| | j¡}| j�rt dt tj ¡ ¡ ¡ | j ¡ | |¡} | | |¡}
|
¡ | j ¡ qœq�| j } | jj ¡ D ] }d|_
�qP| jj ¡ D ] }d|_
�qj| j�r’tj d| d ¡} t| ƒD ]ú}t|ƒD ]ê\}\}}t|ƒtg ƒk�rÚ|d | j¡|d< n | | j¡}| | j¡}| j�rt dt tj ¡ ¡ ¡ | j ¡ | |¡} | | |¡}
| j d k�r|t | jƒ} |
| j!d t" #| | j d¡ 7 }
|
t" $| | j¡8 }
|
¡ | j ¡ �q¦�qš| j d k�rÈt | jƒ %¡ } | j| j!| | j | _| j &¡ | j &¡ | _| j &¡ | _ | j'd d7 < | j'd t ¡ | 7 < t | jj¡}
d}t(| ¡ |
¡ ƒD ]<\}}|| }t" )|dk|t" *|¡| ¡}|t" +|¡ }�q:| j'd d7 < | j'd t ¡ | 7 < |S ) NFTr gš™™™™™¹?é é Ú
num_roundsÚ
total_cost),Úload_train_dataÚtimer r r
r ÚtoÚdevicer r Útrainr r Ú
requires_gradÚranger Ú enumerateÚtypeÚ
train_slowÚsleepÚnpÚabsÚrandomÚrandr Ú zero_gradr ÚbackwardÚstepÚ local_stepsÚrandintr r r r
ÚnormÚdotÚdetachÚcpuÚtrain_time_costÚzipÚwherer Úsum)r Ú trainloaderÚ
start_timeÚ
old_promptÚparamr; ÚiÚxÚyÚoutputr Úmax_local_stepsÚv1Ú
new_promptÚ
diff_provalueÚ new_paramÚ old_paramÚdiff_pror$ r$ r% r. " s‚
zclientDynPT.trainc C sD t |j ¡ | jj ¡ ƒD ]\}}|j ¡ |_qt|ƒ ¡ ¡ | _d S )N) rC r r r
ÚdataÚcloner r@ r )r r
rR rS r$ r$ r% Úset_parameterss s zclientDynPT.set_parameters)Ú__name__Ú
__module__Ú __qualname__r r. rW Ú
__classcell__r$ r$ r" r% r s Qr c C s dd„ | ¡ D ƒ}tj|dd�S )Nc S s g | ]}| d ¡‘qS )éÿÿÿÿ)Úview)Ú.0Úpr$ r$ r% Ú
<listcomp>| s z*model_parameter_vector.<locals>.<listcomp>r )Údim)r r
Úcat)r
rI r$ r$ r% r { s r ) r r
Útorch.nnr Únumpyr5 r+ Úflcore.clients.clientbaser r r r$ r$ r$ r% Ú<module> s o | 3,377 | Python | .pyt | 41 | 81.268293 | 328 | 0.393168 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,885 | clientt.cpython-37.pyc | hkgdifyu_pFedPT/system/flcore/clients/__pycache__/clientt.cpython-37.pyc | B
v¹ccã ã @ sL d dl Z d dlmZ d dlZd dlZd dlmZ d dlZG dd„ deƒZ dS )é N)ÚClientc s4 e Zd Z‡ fdd„Zdd„ Zdd„ Zdd„ Z‡ ZS ) ÚclientTc sl t ƒ j||||f|� t ¡ | _tjj| jj
¡ | j d�| _ tjj| jj
¡ | j d�| _| ¡ |j| _d S )N)Úlr)ÚsuperÚ__init__ÚnnÚCrossEntropyLossÚlossÚtorchÚoptimÚSGDÚmodelÚ generatorÚ
parametersÚ
learning_rateÚ
poptimizerÚbaseÚ optimizerÚser_paraÚ plocal_steps)ÚselfÚargsÚidÚ
train_samplesÚ test_samplesÚkwargs)Ú __class__© úD/root/autodl-tmp/PFL-Non-IID-master/system/flcore/clients/clientt.pyr
s
zclientT.__init__c C sx t | j¡ t | jjjj¡| jjj_t | jjjj¡| jjj_t | jjj j¡| jjj _t | jjj
j¡| jjj
_d S )N) r
Ú manual_seedr Ú rand_liker
r Zpad_downÚdataÚpad_leftÚ pad_rightZpad_up)r r r r r s
zclientT.ser_parac C sê | ¡ }t ¡ }t | jj¡}| j | j¡ | j ¡ | j }| j
rTt j
d|d ¡}x| jj ¡ D ]
}d|_qbW x| jj ¡ D ]
}d|_q~W x¼t|ƒD ]°}xªt|ƒD ]�\}\}} t|ƒtg ƒkrÖ|d | j¡|d< n | | j¡}| | j¡} | j
�rt dt t j ¡ ¡ ¡ | j ¡ | |¡}
| |
| ¡} | ¡ | j ¡ q¤W q–W | j ¡ t | jj¡} d}
xRt | ¡ | ¡ ƒD ]<\}}|| }t |dk|t |¡| ¡}|
t !|¡ }
�qzW | j"d d7 < | j"d t ¡ | 7 < |
S ) Né é TFr gš™™™™™¹?Ú
num_roundsÚ
total_cost)#Úload_train_dataÚtimeÚcopyÚdeepcopyr
r ÚtoÚdeviceÚtrainÚ local_stepsÚ
train_slowÚnpÚrandomÚrandintr r Ú
requires_gradÚrangeÚ enumerateÚtypeÚsleepÚabsÚrandr Ú zero_gradr ÚbackwardÚstepÚcpuÚzipr
ÚwhereÚ
zeros_likeÚsumÚtrain_time_cost)r Ú trainloaderÚ
start_timeÚ
old_promptÚmax_local_stepsÚparamr= ÚiÚxÚyÚoutputr Ú
new_promptÚ
diff_provalueÚ new_paramÚ old_paramÚdiff_pror r r r. sF
z
clientT.trainc C s4 x.t | ¡ | jj ¡ ƒD ]\}}|j ¡ |_qW d S )N)r? r r
r r! Úclone)r r rO rP r r r Úset_parametersO s zclientT.set_parameters)Ú__name__Ú
__module__Ú __qualname__r r r. rS Ú
__classcell__r r )r r r s 0r )
r
Útorch.nnr Únumpyr1 r) Ú system.flcore.clients.clientbaser r* r r r r r Ú<module> s | 2,599 | Python | .pyt | 34 | 75.294118 | 292 | 0.447389 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,886 | clientproto.cpython-39.pyc | hkgdifyu_pFedPT/system/flcore/clients/__pycache__/clientproto.cpython-39.pyc | a
âæ`cš ã @ s` d dl mZ d dlZd dlZd dlmZ d dlZd dlZd dl m
Z
G dd„ de
ƒZ dd„ Z dS )é )Ú defaultdictN)ÚClientc s> e Zd Z‡ fdd„Zdd„ Zdd„ Zdd„ Zd d
d „Z‡ ZS )
Ú clientProtoc s| t ƒ j||||fi |¤� t ¡ | _tjj| j ¡ | j
d�| _ t | jj
¡ ƒd jd | _d | _d | _t ¡ | _|j| _d S )N)Úlrr é )ÚsuperÚ__init__ÚnnÚCrossEntropyLossÚlossÚtorchÚoptimÚSGDÚmodelÚ
parametersÚ
learning_rateÚ optimizerÚlistÚ predictorÚshapeZ feature_dimÚprotosÚ
global_protosÚMSELossÚloss_mseÚlamda)ÚselfÚargsÚidÚ
train_samplesÚ test_samplesÚkwargs©Ú __class__© út/media/sim812/391e55df-b6f2-4fe9-a920-53434a8506fa/lgh/pdept/PFL-Non-IID-master/system/flcore/clients/clientproto.pyr s
zclientProto.__init__c C sÜ | ¡ }t ¡ }| j ¡ | j}| jr8tj d|d ¡}t t
ƒ}t |ƒD �]V}t |ƒD �]F\}\}}t
|ƒt
g ƒkrŠ|d | j¡|d< n | | j¡}| | j¡}| jrÂt dt tj ¡ ¡ ¡ | j ¡ | j |¡} | j | ¡}
| |
|¡} | jd k�rPt | ¡} t |ƒD ]*\}}
|
¡ }| j| j | |d d …f< �q| | | | ¡| j 7 } t |ƒD ]2\}}
|
¡ }|| | |d d …f ¡ j ¡ �qX| !¡ | j "¡ qVqHt#|ƒ| _$| j%d d7 < | j%d t ¡ | 7 < d S )Nr é r çš™™™™™¹?Ú
num_roundsÚ
total_cost)&Úload_train_dataÚtimer ÚtrainÚ local_stepsÚ
train_slowÚnpÚrandomÚrandintr r ÚrangeÚ enumerateÚtypeÚtoÚdeviceÚsleepÚabsÚrandr Ú zero_gradÚbaser r r r Ú
zeros_likeÚitemÚdatar r ÚappendÚdetachÚbackwardÚstepÚagg_funcr Útrain_time_cost)r Ú trainloaderÚ
start_timeÚmax_local_stepsr rA ÚiÚxÚyÚrepÚoutputr Z proto_newÚyyÚy_cr# r# r$ r+ sB
$
zclientProto.trainc C s t |¡| _d S )N)ÚcopyÚdeepcopyr )r r r# r# r$ Ú
set_protosN s zclientProto.set_protosc C s | ¡ }| j ¡ ttƒ}t ¡ �Ô t|ƒD ]º\}\}}t|ƒtg ƒkr^|d | j
¡|d< n | | j
¡}| | j
¡}| j r–t
dt tj ¡ ¡ ¡ | j ¡ | j |¡}t|ƒD ]0\}}| ¡ }|| ||d d …f ¡ j¡ q´q,W d ƒ n1 sü0 Y t|ƒ| _d S )Nr r& )r) r Úevalr r r Úno_gradr2 r3 r4 r5 r- r* r6 r. r7 r/ r8 r r9 r: r< r> r? r= rB r ) r rD r rG rH rI rJ rL rM r# r# r$ Úcollect_protosQ s"
BzclientProto.collect_protosNc
C s> | ¡ }|d kr| j}| ¡ d}d}t ¡ �ô |D ]Ş\}}t|ƒtg ƒkrb|d | j¡|d< n | | j¡}| | j¡}| j |¡}t dƒt
|j d | j ¡ | j¡ }t
|ƒD ]0\} }
| j ¡ D ] \} } | |
| ¡|| | f< qÄq²|t tj|dd�|k¡ ¡ 7 }||j d 7 }q4W d ƒ n1 �s*0 Y ||dfS )Nr Úinfr )Údim)Úload_test_datar rQ r rR r3 r4 r5 r: ÚfloatÚonesr Ú num_classesr2 r Úitemsr ÚsumÚargminr< )
r r Ú
testloaderÚtest_accÚtest_numrH rI rJ rK rG ÚrÚjÚpror# r# r$ Ú test_metricsh s(
$ 0zclientProto.test_metrics)N) Ú__name__Ú
__module__Ú __qualname__r r+ rP rS rc Ú
__classcell__r# r# r! r$ r s
3r c C sb | ¡ D ]T\}}t|ƒdkrPd|d j }|D ]}||j7 }q.|t|ƒ | |< q|d | |< q| S )z-
Returns the average of the weights.
r r )rZ Úlenr= )r ÚlabelÚ
proto_listÚprotorG r# r# r$ rB † s rB )
Ú collectionsr rN r Útorch.nnr Únumpyr. r* Úflcore.clients.clientbaser r rB r# r# r# r$ Ú<module> s { | 3,980 | Python | .pyt | 39 | 100.794872 | 408 | 0.40208 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,887 | clientditto.cpython-39.pyc | hkgdifyu_pFedPT/system/flcore/clients/__pycache__/clientditto.cpython-39.pyc | a
$æ`cù ã @ s‚ d dl Z d dlZd dlZd dlZd dlmZ d dlmZ d dl m
Z
d dl m m Z
d dlmZ d dlmZ G dd„ de
ƒZdS )é N)ÚPerturbedGradientDescent)ÚClient)Úlabel_binarize)Úmetricsc s4 e Zd Z‡ fdd„Zdd„ Zdd„ Zdd„ Z‡ ZS ) Ú clientDittoc sz t ƒ j||||fi |¤� |j| _|j| _t | j¡| _t ¡ | _
t j j
| j ¡ | jd�| _t| j ¡ | j| jd�| _d S )N)Úlr)r Úmu)ÚsuperÚ__init__r Ú plocal_stepsÚcopyÚdeepcopyÚmodelÚpmodelÚnnÚCrossEntropyLossÚlossÚtorchÚoptimÚSGDÚ
parametersÚ
learning_rateÚ optimizerr Ú
poptimizer)ÚselfÚargsÚidÚ
train_samplesÚ test_samplesÚkwargs©Ú __class__© út/media/sim812/391e55df-b6f2-4fe9-a920-53434a8506fa/lgh/pdept/PFL-Non-IID-master/system/flcore/clients/clientditto.pyr
s
ÿzclientDitto.__init__c
C s6 | ¡ }t ¡ }| j ¡ | j}| jr8tj d|d ¡}t |ƒD ]Ä}t
|ƒD ]¶\}\}}t |ƒt g ƒkr~|d | j
¡|d< n | | j
¡}| | j
¡}| jr¶t dt tj ¡ ¡ ¡ | j ¡ | |¡}| ||¡} | ¡ | jrøt| j|t|ƒƒ qL| j ¡ qLq@| jd d7 < | jd t ¡ | 7 < d S )Né é r çš™™™™™¹?Ú
num_roundsÚ
total_cost)Úload_train_dataÚtimer ÚtrainÚ local_stepsÚ
train_slowÚnpÚrandomÚrandintÚrangeÚ enumerateÚtypeÚtoÚdeviceÚsleepÚabsÚrandr Ú zero_gradr ÚbackwardÚprivacyÚdp_stepÚlenÚstepÚtrain_time_cost)
r Ú trainloaderÚ
start_timeÚmax_local_stepsr> ÚiÚxÚyÚoutputr r" r" r# r+ ! s.
zclientDitto.trainc C s | ¡ }t ¡ }| j ¡ | j}| jr8tj d|d ¡}t |ƒD ]®}|D ]¤\}}t
|ƒt
g ƒkrv|d | j ¡|d< n | | j ¡}| | j ¡}| jr®t
dt tj ¡ ¡ ¡ | j ¡ | |¡}| ||¡}| ¡ | j | j ¡ | j ¡ qHq@| jd t ¡ | 7 < d S )Nr$ r% r r& r( )r) r* r r+ r r- r. r/ r0 r1 r3 r4 r5 r6 r7 r8 r r9 r r: r> r r r? ) r r@ rA rB r> rD rE rF r r" r" r# ÚptrainI s(
zclientDitto.ptrainc
C sH | ¡ }| j ¡ d}d}g }g }t ¡ �Ö |D ]À\}}t|ƒtg ƒkr^|d | j¡|d< n | | j¡}| | j¡}| |¡}|t tj |dd�|k¡
¡ 7 }||j d 7 }| t
|¡ ¡ ¡ ¡ ¡ | t| ¡ ¡ ¡ t | j¡d�¡ q0W d ƒ n1 �s0 Y tj|dd�}tj|dd�}tj||dd�} ||| fS )Nr r$ )Údim)Úclasses)ÚaxisÚmicro)Úaverage)Úload_test_datar Úevalr Úno_gradr3 r4 r5 ÚsumÚargmaxÚitemÚshapeÚappendÚFÚsoftmaxÚdetachÚcpuÚnumpyr r. ÚarangeÚ num_classesÚ concatenater Ú
roc_auc_score)
r ÚtestloaderfullÚtest_accÚtest_numÚy_probÚy_truerD rE rF Úaucr" r" r# Ú test_metricsh s*
HzclientDitto.test_metrics)Ú__name__Ú
__module__Ú __qualname__r
r+ rG rd Ú
__classcell__r" r" r r# r s ( r )r rY r. r* r Útorch.nnr Ú flcore.optimizers.fedoptimizerr Úflcore.clients.clientbaser Útorch.nn.functionalÚ
functionalrU Úsklearn.preprocessingr Úsklearnr r r" r" r" r# Ú<module> s | 3,460 | Python | .pyt | 34 | 100.705882 | 420 | 0.432156 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,888 | clientapfl.cpython-39.pyc | hkgdifyu_pFedPT/system/flcore/clients/__pycache__/clientapfl.cpython-39.pyc | a
f¾`c€
ã @ sL d dl Z d dlZd dlmZ d dlZd dlZd dlmZ G dd„ deƒZ dS )é N)ÚClientc s, e Zd Z‡ fdd„Zdd„ Zdd„ Z‡ ZS )Ú
clientAPFLc sr t ƒ j||||fi |¤� t ¡ | _tjj| j ¡ | j
d�| _ |j | _ t
| j¡| _tjj| j ¡ | j
d�| _d S )N)Úlr)ÚsuperÚ__init__ÚnnÚCrossEntropyLossÚlossÚtorchÚoptimÚSGDÚmodelÚ
parametersÚ
learning_rateÚ optimizerÚalphaÚcopyÚdeepcopyÚ model_perÚ
optimizer_per)ÚselfÚargsÚidÚ
train_samplesÚ test_samplesÚkwargs©Ú __class__© ús/media/sim812/391e55df-b6f2-4fe9-a920-53434a8506fa/lgh/pdept/PFL-Non-IID-master/system/flcore/clients/clientapfl.pyr
s
zclientAPFL.__init__c C s’ | ¡ }t ¡ }| j ¡ | j}| jr8tj d|d ¡}t |ƒD ]ä}t
|ƒD ]Ö\}\}}t |ƒt g ƒkr~|d | j
¡|d< n | | j
¡}| | j
¡}| jr¶t dt tj ¡ ¡ ¡ | j ¡ | |¡}| ||¡} | ¡ | j ¡ | j ¡ | |¡}
| |
|¡} | ¡ | j ¡ | ¡ qLq@t| j ¡ | j ¡ ƒD ]$\} }
d| j |
| j| | _ �q<| j d d7 < | j d t ¡ | 7 < d S )Né é r gš™™™™™¹?Ú
num_roundsÚ
total_cost) Úload_train_dataÚtimer
ÚtrainÚ local_stepsÚ
train_slowÚnpÚrandomÚrandintÚrangeÚ enumerateÚtypeÚtoÚdeviceÚsleepÚabsÚrandr Ú zero_gradr ÚbackwardÚstepr r Ú alpha_updateÚzipr r ÚdataÚtrain_time_cost)r Ú trainloaderÚ
start_timeÚmax_local_stepsr6 ÚiÚxÚyÚoutputr Z
output_perZloss_perÚlpÚpr r r r& s:
zclientAPFL.trainc C s¦ d}t | j ¡ | j ¡ ƒD ]P\}}|j|j }| j|jj d| j |jj }|| d¡j | d¡¡7 }q|d| j 7 }| j| j
| | _t | j
¡ dd¡| _d S )Nr r éÿÿÿÿg{®Gáz”?g g ğ?)r8 r
r r r9 r ÚgradÚviewÚTÚdotr r) ÚclipÚitem)r Z
grad_alphaZl_paramsZp_paramsZdifrE r r r r7 > s zclientAPFL.alpha_update)Ú__name__Ú
__module__Ú __qualname__r r& r7 Ú
__classcell__r r r r r s
*r )
r r
Útorch.nnr Únumpyr) r% Úflcore.clients.clientbaser r r r r r Ú<module> s | 2,383 | Python | .pyt | 30 | 78.366667 | 277 | 0.434579 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,889 | clientphp.cpython-39.pyc | hkgdifyu_pFedPT/system/flcore/clients/__pycache__/clientphp.cpython-39.pyc | a
Èæ`c ã @ sV d dl Z d dlZd dlmZ d dlZd dlZd dlmZ G dd„ deƒZ ddd„Z
dS ) é N)ÚClientc s, e Zd Z‡ fdd„Zdd„ Zdd„ Z‡ ZS )Ú clientPHPc s| t ƒ j||||fi |¤� t ¡ | _tjj| j ¡ | j
d�| _ |j |j
| _ |j| _t | j¡| _| j ¡ D ]
}d|_qld S )N)ÚlrF)ÚsuperÚ__init__ÚnnÚCrossEntropyLossÚlossÚtorchÚoptimÚSGDÚmodelÚ
parametersÚ
learning_rateÚ optimizerÚmuÚ
global_roundsÚlamdaÚcopyÚdeepcopyÚmodel_sÚ
requires_grad)ÚselfÚargsÚidÚ
train_samplesÚ test_samplesÚkwargsÚparam©Ú __class__© úr/media/sim812/391e55df-b6f2-4fe9-a920-53434a8506fa/lgh/pdept/PFL-Non-IID-master/system/flcore/clients/clientphp.pyr s
zclientPHP.__init__c
C sP | ¡ }t ¡ }| j ¡ | j}| jr8tj d|d ¡}t |ƒD ]Ş}t
|ƒD ]Ğ\}\}}t |ƒt g ƒkr~|d | j
¡|d< n | | j
¡}| | j
¡}| jr¶t dt tj ¡ ¡ ¡ | j ¡ | |¡}| ||¡d| j } | t| j |¡| j |¡d| j
ƒ| j 7 } | ¡ | j ¡ qLq@| jd d7 < | jd t ¡ | 7 < d S )Né é r gš™™™™™¹?ÚrbfÚ
num_roundsÚ
total_cost)Úload_train_dataÚtimer
ÚtrainÚ local_stepsÚ
train_slowÚnpÚrandomÚrandintÚrangeÚ enumerateÚtypeÚtoÚdeviceÚsleepÚabsÚrandr Ú zero_gradr r ÚMMDÚbaser ÚbackwardÚstepÚtrain_time_cost)
r Ú trainloaderÚ
start_timeÚmax_local_stepsr< ÚiÚxÚyÚoutputr r! r! r" r* s,
*zclientPHP.trainc C sl | j | }t| ¡ | j ¡ ƒD ]\}}|j ¡ |_q t| ¡ | j ¡ ƒD ] \}}|d| || |_qHd S )Nr# )r Úzipr r ÚdataÚcloner
)r r
ÚRr Ú new_paramÚ old_paramr! r! r" Úset_parameters: s
zclientPHP.set_parameters)Ú__name__Ú
__module__Ú __qualname__r r* rK Ú
__classcell__r! r! r r" r
s !r Úcpuc C s¸ t | | ¡ ¡t || ¡ ¡t | | ¡ ¡ }}}| ¡ d¡ |¡}| ¡ d¡ |¡}| ¡ | d| } | ¡ | d| }
| ¡ | d| } t |j¡ |¡t |j¡ |¡t |j¡ |¡ } }
}|dk�rBg d¢}|D ]X}| |d |d | d 7 } |
|d |d |
d 7 }
||d |d | d 7 }qè|dk�r¢g d¢}|D ]H}| t d | | ¡7 } |
t d |
| ¡7 }
|t d | | ¡7 }�qXt
| |
d| ¡S )
a Emprical maximum mean discrepancy. The lower the result
the more evidence that distributions are the same.
Args:
x: first sample, distribution P
y: second sample, distribution Q
kernel: kernel type such as "multiscale" or "rbf"
r g @Z
multiscale)gš™™™™™É?g à?gÍÌÌÌÌÌì?gÍÌÌÌÌÌô?r$ éÿÿÿÿr% )é
é é é2 g à¿) r
ÚmmÚtÚdiagÚ unsqueezeÚ expand_asÚzerosÚshaper3 ÚexpÚmean)rB rC Úkernelr4 ÚxxÚyyÚzzÚrxÚryZdxxZdyyZdxyÚXXÚYYZXYZbandwidth_rangeÚar! r! r" r9 D s. 4ş
r9 )rP ) r r
Útorch.nnr Únumpyr- r) Úflcore.clients.clientbaser r r9 r! r! r! r" Ú<module> s : | 3,288 | Python | .pyt | 39 | 82.230769 | 313 | 0.429671 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,890 | clientbase.cpython-38.pyc | hkgdifyu_pFedPT/system/flcore/clients/__pycache__/clientbase.cpython-38.pyc | U
ºşch ã @ s‚ d dl Z d dlZd dlmZ d dlZd dlZd dlm mZ d dl
m Z d dl m
Z
d dlmZ d dlmZ G dd„ deƒZdS )é N)Ú
DataLoader)Úlabel_binarize)Úmetrics)Úread_client_datac @ sh e Zd ZdZdd„ Zddd„Zddd„Zd d
„ Zd d „ Zd
d„ Z dd„ Z
dd„ Z ddd„Z ddd„Z
dS )ÚClientz7
Base class for clients in federated learning.
c K sØ t |j¡| _|j| _|j| _|| _|j| _|| _|j| _|| _ || _
|j | _ |j | _
|j| _d| _| j ¡ D ]}t|tjƒrnd| _ qŠqn|d | _|d | _dddœ| _dddœ| _|j| _|j| _| j | j | _d S )NFTÚ
train_slowÚ send_slowr g )Ú
num_roundsÚ
total_cost)ÚcopyÚdeepcopyÚmodelÚdatasetÚdeviceÚidÚsave_folder_nameÚargsÚ num_classesÚ
train_samplesÚ test_samplesÚ
batch_sizeÚlocal_learning_rateÚ
learning_rateÚ local_stepsZ
has_BatchNormÚchildrenÚ
isinstanceÚnnÚ BatchNorm2dr r Útrain_time_costZsend_time_costÚprivacyZdp_sigmaÚ sample_rate)Úselfr r r r ÚkwargsÚlayer© r$ õb D:\京东\promot\第二次投稿\å®�验\æœ�务器\native - pro\system\flcore\clients\clientbase.pyÚ__init__ s0
zClient.__init__Nc C s4 |d kr| j }t| j| j| jdd�}t||ddd�S )NT©Zis_train©Ú drop_lastÚshuffle©r r r r r r )r! r Ú
train_datar$ r$ r% Úload_train_data2 s zClient.load_train_datac C s4 |d kr| j }t| j| j| jdd�}t||ddd�S )NFr' Tr( r+ )r! r Ú test_datar$ r$ r% Úload_test_data8 s zClient.load_test_datac C s. t | ¡ | j ¡ ƒD ]\}}|j ¡ |_qd S ©N)ÚzipÚ
parametersr
ÚdataÚclone)r! r
Ú new_paramÚ old_paramr$ r$ r% Úset_parameters> s zClient.set_parametersc C s, t | ¡ | ¡ ƒD ]\}}|j ¡ |_qd S r0 ©r1 r2 r3 r4 )r! r
ÚtargetÚparamZ target_paramr$ r$ r% Ú clone_modelB s zClient.clone_modelc C s( t | ¡ |ƒD ]\}}|j ¡ |_qd S r0 r8 )r! r
Ú
new_paramsr: r5 r$ r$ r% Úupdate_parametersG s zClient.update_parametersc
C sD | ¡ }| j | j¡ | j ¡ d}d}g }g }t ¡ �Æ |D ]º\}}t|ƒtg ƒkrl|d | j¡|d< n | | j¡}| | j¡}| |¡}|t tj |dd�|k¡
¡ 7 }||j d 7 }| |
¡ ¡ ¡ ¡ | t|
¡ ¡ ¡ t | j¡d�¡ q>W 5 Q R X | j ¡ tj|dd�}tj|dd�}tj||dd�} ||| fS )Nr é )Údim)Úclasses)ÚaxisÚmicro)Úaverage)r/ r
Útor ÚevalÚtorchÚno_gradÚtypeÚsumÚargmaxÚitemÚshapeÚappendÚdetachÚcpuÚnumpyr ÚnpÚaranger Ú concatenater Z
roc_auc_score)
r! ZtestloaderfullZtest_accZtest_numZy_probZy_trueÚxÚyÚoutputZaucr$ r$ r% Ú test_metricsK s.
2
zClient.test_metricsc C s¼ | ¡ }| j | j¡ | j ¡ d}d}|D ]|\}}t|ƒtg ƒkrZ|d | j¡|d< n | | j¡}| | j¡}| |¡}||jd 7 }|| ||¡ ¡ |jd 7 }q,| j ¡ ||fS )Nr )
r- r
rD r rE rH rL ÚlossrK rO )r! Ú trainloaderZ train_numrX rT rU rV r$ r$ r% Ú
train_metricsr s
zClient.train_metricsc C sT |d kr| j }tj |¡s$t |¡ t |tj |dt| j ƒ d | d ¡¡ d S ©NZclient_Ú_z.pt)
r ÚosÚpathÚexistsÚmakedirsrF ÚsaveÚjoinÚstrr )r! rK Ú item_nameÚ item_pathr$ r$ r% Ú save_itemš s
zClient.save_itemc C s8 |d kr| j }t tj |dt| jƒ d | d ¡¡S r[ )r rF Úloadr] r^ rb rc r )r! rd re r$ r$ r% Ú load_item¡ s zClient.load_item)N)N)N)N)Ú__name__Ú
__module__Ú __qualname__Ú__doc__r& r- r/ r7 r; r= rW rZ rf rh r$ r$ r$ r% r
s
'(
r )r rF Útorch.nnr rP rQ r] Útorch.nn.functionalÚ
functionalÚFÚtorch.utils.datar Zsklearn.preprocessingr Zsklearnr Zutils.data_utilsr Úobjectr r$ r$ r$ r% Ú<module> s | 4,438 | Python | .pyt | 38 | 115.5 | 425 | 0.428312 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,891 | clientrep.cpython-38.pyc | hkgdifyu_pFedPT/system/flcore/clients/__pycache__/clientrep.cpython-38.pyc | U
Ǭkc ã @ sD d dl Z d dlmZ d dlZd dlZd dlmZ G dd„ deƒZdS )é N)ÚClientc s, e Zd Z‡ fdd„Zdd„ Zdd„ Z‡ ZS )Ú clientRepc sd t ƒ j||||f|� t ¡ | _tjj| jj
¡ | j d�| _ tjj| jj
¡ | j d�| _|j| _d S )N)Úlr)ÚsuperÚ__init__ÚnnÚCrossEntropyLossÚlossÚtorchÚoptimÚSGDÚmodelÚbaseÚ
parametersÚ
learning_rateÚ optimizerÚ predictorÚ
poptimizerÚ plocal_steps)ÚselfÚargsÚidÚ
train_samplesÚ test_samplesÚkwargs©Ú __class__© õW D:\京东\promot\第二次投稿\å®�验\native - pro\system\flcore\clients\clientrep.pyr
s
zclientRep.__init__c C sX | ¡ }t ¡ }| j | j¡ | j ¡ | jj ¡ D ]
}d|_q4| jj ¡ D ]
}d|_qLt
| j ƒD ]ª}t |ƒD ]œ\}\}}t
|ƒt
g ƒkr |d | j¡|d< n | | j¡}| | j¡}| jrØt dt tj ¡ ¡ ¡ | j ¡ | |¡}| ||¡} | ¡ | j ¡ qnqb| j}
| j�r.tj d|
d ¡}
| jj ¡ D ] }d|_�q:| jj ¡ D ] }d|_�qTt
|
ƒD ]²}t |ƒD ]¢\}\}}t
|ƒt
g ƒk�rª|d | j¡|d< n | | j¡}| | j¡}| j�rät dt tj ¡ ¡ ¡ | j ¡ | |¡}| ||¡} | ¡ | j ¡ �qv�qj| j ¡ | j d d7 < | j d t ¡ | 7 < d S ) NFTr gš™™™™™¹?é é Ú
num_roundsÚ
total_cost) Úload_train_dataÚtimer
ÚtoÚdeviceÚtrainr r Ú
requires_gradr Úranger Ú enumerateÚtypeÚ
train_slowÚsleepÚnpÚabsÚrandomÚrandr Ú zero_gradr ÚbackwardÚstepÚ local_stepsÚrandintr ÚcpuÚtrain_time_cost) r Ú trainloaderÚ
start_timeÚparamr4 ÚiÚxÚyÚoutputr Úmax_local_stepsr r r r' sX
zclientRep.trainc C s0 t | ¡ | jj ¡ ƒD ]\}}|j ¡ |_qd S )N)Úzipr r
r ÚdataÚclone)r r Ú new_paramÚ old_paramr r r Úset_parametersM s zclientRep.set_parameters)Ú__name__Ú
__module__Ú __qualname__r r' rF Ú
__classcell__r r r r r s :r ) r
Útorch.nnr Únumpyr. r$ Úflcore.clients.clientbaser r r r r r Ú<module> s
| 2,312 | Python | .pyt | 32 | 71.09375 | 290 | 0.419991 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,892 | clientmoon.cpython-37.pyc | hkgdifyu_pFedPT/system/flcore/clients/__pycache__/clientmoon.cpython-37.pyc | B
ïfcÏ ã @ s^ d dl Z d dlZd dlmZ d dlZd dlZd dlm mZ d dl
m Z G dd„ de ƒZ dS )é N)ÚClientc s, e Zd Z‡ fdd„Zdd„ Zdd„ Z‡ ZS )Ú
clientMOONc sb t ƒ j||||f|� t ¡ | _tjj| j ¡ | j
d�| _ |j | _ |j
| _
d | _t | j¡| _d S )N)Úlr)ÚsuperÚ__init__ÚnnÚCrossEntropyLossÚlossÚtorchÚoptimÚSGDÚmodelÚ
parametersÚ
learning_rateÚ optimizerÚtauÚmuÚ global_modelÚcopyÚdeepcopyÚ old_model)ÚselfÚargsÚidÚ
train_samplesÚ test_samplesÚkwargs)Ú __class__© úG/root/autodl-tmp/PFL-Non-IID-master/system/flcore/clients/clientmoon.pyr s
zclientMOON.__init__c C s0 | ¡ }t ¡ }| j | j¡ | j | j¡ | j | j¡ | j ¡ | j}| j rbt
j d|d ¡}�xnt
|ƒD �]`}�xXt|ƒD �]J\}\}}t|ƒtg ƒkr´|d | j¡|d< n | | j¡}| | j¡}| j rìt dt
t
j ¡ ¡ ¡ | j ¡ | j |¡}| j |¡} | | |¡}
| j |¡ ¡ } | j |¡ ¡ } t t t || ¡| j ¡t t || ¡| j ¡t t || ¡| j ¡ ¡ }
|
| j t |
¡ 7 }
|
!¡ | j"�rÂt#| j|t$|ƒƒ q€| j %¡ q€W qnW | j &¡ | j &¡ | j &¡ t' (| j¡| _| j)d d7 < | j)d t ¡ | 7 < d S )Né é r gš™™™™™¹?Ú
num_roundsÚ
total_cost)*Úload_train_dataÚtimer
ÚtoÚdevicer r ÚtrainÚ local_stepsÚ
train_slowÚnpÚrandomÚrandintÚrangeÚ enumerateÚtypeÚsleepÚabsÚrandr Ú zero_gradÚbaseÚ predictorr Údetachr
ÚlogÚexpÚFÚcosine_similarityr r ÚmeanÚbackwardÚprivacyÚdp_stepÚlenÚstepÚcpur r Útrain_time_cost)r Ú trainloaderÚ
start_timeÚmax_local_stepsrA ÚiÚxÚyÚrepÚoutputr Zrep_oldZ
rep_globalZloss_conr r r r( sF
P
zclientMOON.trainc C s8 x,t | ¡ | j ¡ ƒD ]\}}|j ¡ |_qW || _d S )N)Úzipr r
ÚdataÚcloner )r r
Ú new_paramÚ old_paramr r r Úset_parametersO s zclientMOON.set_parameters)Ú__name__Ú
__module__Ú __qualname__r r( rQ Ú
__classcell__r r )r r r s 2r )
r r
Útorch.nnr Únumpyr+ r% Útorch.nn.functionalÚ
functionalr: Úflcore.clients.clientbaser r r r r r Ú<module> s | 2,421 | Python | .pyt | 28 | 85.357143 | 263 | 0.44528 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,893 | clientfomo.cpython-38.pyc | hkgdifyu_pFedPT/system/flcore/clients/__pycache__/clientfomo.cpython-38.pyc | U
”jfcà ã @ sd d dl Z d dlmZ d dlZd dlZd dlZd dlmZ d dl m
Z
d dl m Z G dd„ deƒZ
dS )é N)ÚClient)Ú
DataLoader)Úread_client_datac sn e Zd Z‡ fdd„Zdd„ Zddd„Zdd „ Zd
d „ Zd d
„ Zdd„ Z dd„ Z
dd„ Z dd„ Z dd„ Z
‡ ZS )Ú
clientFomoc sŒ t ƒ j||||f|� |j| _t | j¡| _g | _g | _t j
| j| j d�| _ t
¡ | _t jj| j ¡ | jd�| _d| _| jd| j | _d S )N)Údevice)Úlrgš™™™™™É?é )ÚsuperÚ__init__Ú num_clientsÚcopyÚdeepcopyÚmodelÚ old_modelÚ received_idsÚreceived_modelsÚtorchÚzerosr Ú
weight_vectorÚnnÚCrossEntropyLossÚlossÚoptimÚSGDÚ
parametersÚ
learning_rateÚ optimizerÚ val_ratioÚ
train_samples)ÚselfÚargsÚidr Ú test_samplesÚkwargs©Ú __class__© õL D:\京东\promot\cifar\cifar\Cifar10_iid\system\flcore\clients\clientfomo.pyr
s
zclientFomo.__init__c
C sJ | ¡ \}}t ¡ }| |¡ | | j| j¡ | j | j¡ | j ¡ | j }| j
rdt j
d|d ¡}t|ƒD ]¢}|D ]˜\}}t|ƒtg ƒkr¢|d | j¡|d< n | | j¡}| | j¡}| j
rÚt dt t j ¡ ¡ ¡ | j ¡ | |¡}| ||¡} | ¡ | j ¡ qtql| j ¡ | jd d7 < | jd t ¡ | 7 < d S )Nr é r gš™™™™™¹?Ú
num_roundsÚ
total_cost)Úload_train_dataÚtimeÚaggregate_parametersÚ clone_modelr r Útor ÚtrainÚ local_stepsÚ
train_slowÚnpÚrandomÚrandintÚrangeÚtypeÚsleepÚabsÚrandr Ú zero_gradr ÚbackwardÚstepÚcpuÚtrain_time_cost)
r Ú trainloaderÚ
val_loaderÚ
start_timeÚmax_local_stepsr= ÚxÚyÚoutputr r& r& r' r0 s2
zclientFomo.trainNc C sz |d kr| j }t| j| jdd�}t| jt|ƒ ƒ }||d … }|d |… }t|| j ddd�}t|| j | jdd�}||fS )NT)Úis_train)Ú drop_lastÚshuffle) Ú
batch_sizer Údatasetr! Úintr Úlenr Ú
has_BatchNorm)r rJ Ú
train_dataÚval_idxZval_datar@ rA r& r& r' r+ ? s zclientFomo.load_train_datac C sÀ | ¡ \}}| j | j¡ | j ¡ d}d}|D ]|\}}t|ƒtg ƒkr^|d | j¡|d< n | | j¡}| | j¡}| |¡}||jd 7 }|| ||¡ ¡ |jd 7 }q0| j ¡ ||fS ©Nr )
r+ r r/ r Úevalr7 Úshaper Úitemr> )r r@ rA Ú train_numr rD rE rF r& r& r' Ú
train_metricsL s
zclientFomo.train_metricsc C s || _ || _d S ©N)r r )r ÚidsÚmodelsr& r& r' Úreceive_modelsc s zclientFomo.receive_modelsc C s– g }| | j|¡}| jD ]h}g }t| ¡ | j ¡ ƒD ] \}}| || d¡¡ q4t |¡}| || ||¡ t |¡d ¡ q|
|¡ t |¡S )Néÿÿÿÿgñh㈵øä>) Úrecalculate_lossr r Úzipr ÚappendÚviewr ÚcatÚnormÚweight_vector_updateÚtensor)r rA Ú weight_listÚLÚreceived_modelZ
params_difZparam_nZparam_ir& r& r' Ú
weight_calg s
&
zclientFomo.weight_calc C sT t | j¡| _t|| jƒD ] \}}| j| | ¡ 7 < qt | j¡ | j
¡| _d S rW ) r3 r r r r] r rT r rc r/ r )r rd Úwr! r& r& r' rb ‚ s z clientFomo.weight_vector_updatec C s� d}| | j¡ |D ]f\}}t|ƒtg ƒkrB|d | j¡|d< n | | j¡}| | j¡}||ƒ}| ||¡}|| ¡ 7 }q| ¡ |t|ƒ S rQ )r/ r r7 r rT r> rM )r Z new_modelrA re rD rE rF r r& r& r' r\ Œ s zclientFomo.recalculate_lossc C s: t | j ¡ | ¡ ƒD ] \}}| j|j ¡ | 7 _qd S rW )r] r r ÚdataÚclone)r rh rf ÚparamZreceived_paramr& r& r' Úadd_parameters› s zclientFomo.add_parametersc C s\ | | |¡¡}t|ƒdkrX| j ¡ D ]}|j ¡ q&t|| jƒD ]\}}| ||¡ qBd S rQ )
Ú weight_scalerg rM r r ri Úzero_r] r rl )r rA Úweightsrk rh rf r& r& r' r- Ÿ s z clientFomo.aggregate_parametersc sN t |t d¡¡}t |¡‰ ˆ dkr@‡ fdd„|D ƒ}t |¡S t g ¡S d S )Nr c s g | ] }|ˆ ‘qS r& r& )Ú.0rh ©Úw_sumr& r' Ú
<listcomp> s z+clientFomo.weight_scale.<locals>.<listcomp>)r Úmaximumrc Úsum)r ro r& rq r' rm © s
zclientFomo.weight_scale)N)Ú__name__Ú
__module__Ú __qualname__r
r0 r+ rV rZ rg rb r\ rl r- rm Ú
__classcell__r& r& r$ r' r s #
r )r Útorch.nnr Únumpyr3 r, r Úflcore.clients.clientbaser Útorch.utils.datar Úutils.data_utilsr r r& r& r& r' Ú<module> s | 5,099 | Python | .pyt | 48 | 105.145833 | 870 | 0.403405 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,894 | clientpFedMe.cpython-38.pyc | hkgdifyu_pFedPT/system/flcore/clients/__pycache__/clientpFedMe.cpython-38.pyc | U
”jfcö ã @ sX d dl Zd dlZd dlZd dlZd dlmZ d dlmZ d dl m
Z
G dd„ de
ƒZ dS )é N)ÚpFedMeOptimizer)ÚClientc s4 e Zd Z‡ fdd„Zdd„ Zdd„ Zdd„ Z‡ ZS ) Ú clientpFedMec s‚ t ƒ j||||f|� |j| _|j| _|j| _t t| j
¡ ƒ¡| _ t t| j
¡ ƒ¡| _ t
¡ | _t| j
¡ | j| jd�| _d S )N)ÚlrÚlamda)ÚsuperÚ__init__r ÚKZp_learning_rateZpersonalized_learning_rateÚcopyÚdeepcopyÚlistÚmodelÚ
parametersÚ local_paramsÚpersonalized_paramsÚnnÚCrossEntropyLossÚlossr Ú optimizer)ÚselfÚargsÚidÚ
train_samplesÚ test_samplesÚkwargs©Ú __class__© õN D:\京东\promot\cifar\cifar\Cifar10_iid\system\flcore\clients\clientpFedMe.pyr s
ÿzclientpFedMe.__init__c C s� | ¡ }t ¡ }| j | j¡ | j ¡ | j}| jrFtj
d|d ¡}t |ƒD �]}|D ]ø\}}t |ƒt g ƒkr†|d | j¡|d< n | | j¡}| | j¡}| jr¾t
dt tj ¡ ¡ ¡ t | jƒD ]@}| j ¡ | |¡}| ||¡} | ¡ | j | j| j¡| _qÈt| j| jƒD ]6\}
} | | j¡} | j| j| j | j|
j | _�qqXqN| j ¡ | | j| j¡ | j d d7 < | j d t ¡ | 7 < d S )Né é r gš™™™™™¹?Ú
num_roundsÚ
total_cost) Úload_train_dataÚtimer
ÚtoÚdeviceÚtrainÚ local_stepsÚ
train_slowÚnpÚrandomÚrandintÚrangeÚtypeÚsleepÚabsÚrandr r Ú zero_gradr ÚbackwardÚstepr r ÚzipÚdatar Ú
learning_rateÚcpuÚupdate_parametersÚtrain_time_cost) r Ú trainloaderÚ
start_timeÚmax_local_stepsr4 ÚxÚyÚiÚoutputr Ú new_paramZ localweightr r r r' s8
(
zclientpFedMe.trainc C s@ t | ¡ | j ¡ | jƒD ]"\}}}|j ¡ |_|j ¡ |_qd S )N)r5 r r
r r6 Úclone)r r
rB Ú old_paramZ local_paramr r r Úset_parametersE s " zclientpFedMe.set_parametersc C sâ | ¡ }| | j| j¡ | j | j¡ | j ¡ d}d}t ¡ �Š |D ]~\}}t |ƒt g ƒkrt|d | j¡|d< n | | j¡}| | j¡}| |¡}|t
tj |dd�|k¡ ¡ 7 }||j
d 7 }qFW 5 Q R X | j ¡ ||fS )Nr r )Údim)Úload_test_datar9 r
r r% r& ÚevalÚtorchÚno_gradr. ÚsumÚargmaxÚitemÚshaper8 )r ÚtestloaderfullÚtest_accÚtest_numr> r? rA r r r Útest_metrics_personalizedJ s"
z&clientpFedMe.test_metrics_personalized)Ú__name__Ú
__module__Ú __qualname__r r' rE rR Ú
__classcell__r r r r r
s +r ) Únumpyr* r$ r
rI Útorch.nnr Z flcore.optimizers.fedoptimizerr Úflcore.clients.clientbaser r r r r r Ú<module> s | 2,899 | Python | .pyt | 29 | 98.62069 | 457 | 0.444096 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,895 | clientperavg.cpython-37.pyc | hkgdifyu_pFedPT/system/flcore/clients/__pycache__/clientperavg.cpython-37.pyc | B
gûec ã @ sX d dl Zd dlZd dlZd dlZd dlmZ d dlmZ d dl m
Z
G dd„ de
ƒZ dS )é N)ÚPerAvgOptimizer)ÚClientc s, e Zd Z‡ fdd„Zdd„ Zdd„ Z‡ ZS )Ú clientPerAvgc sB t ƒ j||||f|� | j| _t ¡ | _t| j ¡ | jd�| _
d S )N)Úlr) ÚsuperÚ__init__Ú
learning_rateÚbetaÚnnÚCrossEntropyLossÚlossr ÚmodelÚ
parametersÚ optimizer)ÚselfÚargsÚidÚ
train_samplesÚ test_samplesÚkwargs)Ú __class__© úI/root/autodl-tmp/PFL-Non-IID-master/system/flcore/clients/clientperavg.pyr s
zclientPerAvg.__init__c C s˜ | | jd ¡}t ¡ }| j | j¡ | j ¡ | j}| jrNt j
d|d ¡}�x t |ƒD �]ş}�xö|D �]ì\}}t
t| j ¡ ƒ¡}t|ƒtg ƒkrÔd d g}|d d | j… | j¡|d< |d d | j… |d< n|d | j… | j¡}|d | j… | j¡} | j�r"t dt t j
¡ ¡ ¡ | j ¡ | |¡}
| |
| ¡} | ¡ | j ¡ t|ƒtg ƒk�r¤d d g}|d | jd … | j¡|d< |d | jd … |d< n|| jd … | j¡}|| jd … | j¡} | j�ròt dt t j
¡ ¡ ¡ | j ¡ | |¡}
| |
| ¡} | ¡ x*t| j ¡ |ƒD ]\} }
|
j ¡ | _�q,W | jj| j d� qhW qZW | j ¡ | j d d7 < | j d t ¡ | 7 < d S )Né é r gš™™™™™¹?)r Ú
num_roundsÚ
total_cost) Úload_train_dataÚ
batch_sizeÚtimer
ÚtoÚdeviceÚtrainÚ local_stepsÚ
train_slowÚnpÚrandomÚrandintÚrangeÚcopyÚdeepcopyÚlistr ÚtypeÚsleepÚabsÚrandr Ú zero_gradr ÚbackwardÚstepÚzipÚdataÚcloner ÚcpuÚtrain_time_cost)r Ú trainloaderÚ
start_timeÚmax_local_stepsr2 ÚXÚYZ
temp_modelÚxÚyÚoutputr Ú old_paramÚ new_paramr r r r" sR
zclientPerAvg.trainc C s8 | | j¡}t|ƒ}| j | j¡ | j ¡ t|ƒ\}}t|ƒtg ƒkr^|d | j¡|d< n | | j¡}| | j¡}| j
¡ | |¡}| ||¡}| ¡ | j
¡ t|ƒ\}}t|ƒtg ƒkrÚ|d | j¡|d< n | | j¡}| | j¡}| j
¡ | |¡}| ||¡}| ¡ | j j
| jd� | j ¡ d S )Nr )r )Úload_test_datar Úiterr
r r! r" Únextr, r r0 r r1 r2 r r6 )r Z
testloaderZiter_testloaderr= r> r? r r r r Útrain_one_stepN s2
zclientPerAvg.train_one_step)Ú__name__Ú
__module__Ú __qualname__r r" rE Ú
__classcell__r r )r r r
s :r ) Únumpyr% Útorchr r) Útorch.nnr
Ú flcore.optimizers.fedoptimizerr Úflcore.clients.clientbaser r r r r r Ú<module> s | 2,797 | Python | .pyt | 40 | 68.7 | 218 | 0.397389 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,896 | clientbabu.cpython-37.pyc | hkgdifyu_pFedPT/system/flcore/clients/__pycache__/clientbabu.cpython-37.pyc | B
šúecù
ã @ sL d dl Z d dlZd dlmZ d dlZd dlZd dlmZ G dd„ deƒZ dS )é N)ÚClientc s< e Zd Z‡ fdd„Zdd„ Zdd„ Zddgfd d
„Z‡ ZS ) Ú
clientBABUc sd t ƒ j||||f|Ž t ¡ | _tjj| jj
¡ | j d�| _ |j
| _
x| jj
¡ D ]
}d|_qRW d S )N)ÚlrF)ÚsuperÚ__init__ÚnnÚCrossEntropyLossÚlossÚtorchÚoptimÚSGDÚmodelÚbaseÚ
parametersÚ
learning_rateÚ optimizerÚfine_tuning_stepsÚ predictorÚ
requires_grad)ÚselfÚargsÚidÚ
train_samplesÚ test_samplesÚkwargsÚparam)Ú __class__© úG/root/autodl-tmp/PFL-Non-IID-master/system/flcore/clients/clientbabu.pyr
s
zclientBABU.__init__c
C s< | ¡ }t ¡ }| j | j¡ | j ¡ | j}| jrFtj
d|d ¡}xºt |ƒD ]®}x¨t |ƒD ]œ\}\}}t
|ƒt
g ƒkr�|d | j¡|d< n | | j¡}| | j¡}| jrÈt dt tj ¡ ¡ ¡ | j ¡ | |¡}| ||¡} | ¡ | j ¡ q^W qPW | j ¡ | jd d7 < | jd t ¡ | 7 < d S )Né é r gš™™™™™¹?Ú
num_roundsÚ
total_cost)Úload_train_dataÚtimer
ÚtoÚdeviceÚtrainÚ local_stepsÚ
train_slowÚnpÚrandomÚrandintÚrangeÚ enumerateÚtypeÚsleepÚabsÚrandr Ú zero_gradr ÚbackwardÚstepÚcpuÚtrain_time_cost)
r Ú trainloaderÚ
start_timeÚmax_local_stepsr5 ÚiÚxÚyÚoutputr r r r r' s.
zclientBABU.trainc C s4 x.t | ¡ | jj ¡ ƒD ]\}}|j ¡ |_qW d S )N)Úzipr r
r ÚdataÚclone)r r Ú new_paramÚ old_paramr r r Úset_parameters6 s zclientBABU.set_parametersr r c
C s | ¡ }| j | j¡ | j ¡ d|krDx| jj ¡ D ]
}d|_q6W d|krhx| jj ¡ D ]
}d|_qZW xœt| j ƒD ]Ž}xˆt
|ƒD ]|\}\}}t |ƒt g ƒkr´|d | j¡|d< n | | j¡}| | j¡}| j
¡ | |¡}| ||¡} | ¡ | j ¡ q‚W qtW | j ¡ d S )Nr Tr Fr )r# r
r% r& r' r r r r- r r. r/ r r3 r r4 r5 r6 )
r Z which_moduler8 r r5 r; r< r= r> r r r r Ú fine_tune: s*
zclientBABU.fine_tune)Ú__name__Ú
__module__Ú __qualname__r r' rD rE Ú
__classcell__r r )r r r s r )
Úcopyr
Útorch.nnr Únumpyr* r$ Úflcore.clients.clientbaser r r r r r Ú<module> s | 2,549 | Python | .pyt | 35 | 71.657143 | 286 | 0.422664 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,897 | clientrod.cpython-38.pyc | hkgdifyu_pFedPT/system/flcore/clients/__pycache__/clientrod.cpython-38.pyc | U
5ngc± „ @ sÄ d dl Z d dlZd dlmZ d dlZd dlZd dlmZ d dl m m
Z d dl m
Z
d dlmZ G ddÑ deÉZd
dd ÑZdS ) È N)⁄Client)⁄label_binarize)⁄metricsc s. e Zd Zá fddÑZddÑ ZdddÑZá ZS ) ⁄ clientRODc
s¨ t É j||||f|é t†° | _tjj| j† ° | j
dç| _ t †
| jj°| _tjj| j† ° | j
dç| _t†| j°| _| †° }|D ](\}}|D ]} | j| †° d7 < qäq~d S )N)⁄lrÈ )⁄super⁄__init__⁄nn⁄CrossEntropyLoss⁄loss⁄torch⁄optim⁄SGD⁄model⁄
parameters⁄
learning_rate⁄ optimizer⁄copy⁄deepcopy⁄ predictor⁄pred⁄opt_pred⁄zeros⁄ num_classes⁄sample_per_class⁄load_train_data⁄item)
⁄self⁄args⁄id⁄
train_samples⁄ test_samples⁄kwargs⁄ trainloader⁄x⁄y⁄yy©⁄ __class__© ıD D:\‰∫¨‰∏ú\promot\cifar\cifar\tiny\system\flcore\clients\clientrod.pyr
s
zclientROD.__init__c
C sz | † ° }t†° }| j†| j° | j†| j° | j†° | j}| jrTt j
† d|d °}t |ÉD ]ÿ}t
|ÉD ] \}\}}t|Étg Ékrö|d †| j°|d< n |†| j°}|†| j°}| j†|°}| j†|°} t|| | jÉ}
| j†° |
†° | j†° | †|†° °} | †| †° | |°} | j†° | †° | j†° qhq\| j†° | j†° | jd d7 < | jd t†° | 7 < d S )Nr È r ⁄
num_rounds⁄
total_cost) r ⁄timer ⁄to⁄devicer ⁄train⁄ local_steps⁄
train_slow⁄np⁄random⁄randint⁄range⁄ enumerate⁄type⁄baser ⁄balanced_softmax_lossr r ⁄ zero_grad⁄backward⁄step⁄detachr r ⁄cpu⁄train_time_cost)
r r$ ⁄
start_time⁄max_local_stepsr? ⁄ir% r& ⁄rep⁄out_gZloss_bsm⁄out_pr r* r* r+ r2 s:
zclientROD.trainNc C s÷ | † ° }|d kr| j}| j†| j° | j†| j° |†° d}d}g }g }t†° êè2 |D ê]$\}}t|Étg Ékrä|d †| j°|d< n |†| j°}|†| j°}| j† |°} | j†
| °}
| †| † ° °} |
† ° | } |t† tj
| ddç|k°†° 7 }||jd 7 }|†t†| °† ° †° †° ° | j}
| jdkêr8|
d7 }
t|† ° †° †° t†|
°dç}| jdkêrv|d d Öd dÖf }|†|° qZW 5 Q R X | j†° | j†° tj|ddç}tj|ddç}tj||ddç}|||fS ) Nr r )⁄dimr, )⁄classes)⁄axis⁄micro)⁄average) ⁄load_test_datar r0 r1 r ⁄evalr
⁄no_gradr: r; r r@ ⁄sum⁄argmaxr ⁄shape⁄append⁄F⁄softmaxrA ⁄numpyr r r5 ⁄arange⁄ concatenater ⁄
roc_auc_score)r r ⁄
testloader⁄test_acc⁄test_num⁄y_prob⁄y_truer% r& rF rG rH ⁄output⁄nc⁄lb⁄aucr* r* r+ ⁄ test_metricsD sH
zclientROD.test_metrics)N)⁄__name__⁄
__module__⁄ __qualname__r r2 rd ⁄
__classcell__r* r* r( r+ r s 'r ⁄meanc C sB |† |°}|†d°†|jd d°}||†° }tj|| |dç}|S )a} Compute the Balanced Softmax Loss between `logits` and the ground truth `labels`.
Args:
labels: A int tensor of size [batch].
logits: A float tensor of size [batch, no_of_classes].
sample_per_class: A int tensor of size [no of classes].
reduction: string. One of "none", "mean", "sum"
Returns:
loss: A float tensor. Balanced Softmax Loss.
r Èˇˇˇˇ)⁄input⁄target⁄ reduction)⁄type_as⁄ unsqueeze⁄expandrS ⁄logrU ⁄
cross_entropy)⁄labels⁄logitsr rm Zspcr r* r* r+ r< t s
r< )ri )r r
⁄torch.nnr
rW r5 r/ ⁄flcore.clients.clientbaser ⁄torch.nn.functional⁄
functionalrU ⁄sklearn.preprocessingr ⁄sklearnr r r< r* r* r* r+ ⁄<module> s h | 3,841 | Python | .pyt | 42 | 89.404762 | 534 | 0.419058 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,898 | clientbase.cpython-37.pyc | hkgdifyu_pFedPT/system/flcore/clients/__pycache__/clientbase.cpython-37.pyc | B
9Íec ã @ s‚ d dl Z d dlZd dlmZ d dlZd dlZd dlm mZ d dl
m Z d dl m
Z
d dlmZ d dlmZ G dd„ deƒZdS )é N)Ú
DataLoader)Úlabel_binarize)Úmetrics)Úread_client_datac @ sh e Zd ZdZdd„ Zddd„Zddd„Zd d
„ Zd d „ Zd
d„ Z dd„ Z
dd„ Z ddd„Z ddd„Z
dS )ÚClientz7
Base class for clients in federated learning.
c K sÔ t |j¡| _|j| _|j| _|| _|j| _|j| _|| _|| _ |j
| _
|j | _ |j
| _
d| _x&| j ¡ D ]}t|tjƒrjd| _P qjW |d | _|d | _dddœ| _dddœ| _|j| _|j| _| j
| j | _d S )NFTÚ
train_slowÚ send_slowr g )Ú
num_roundsÚ
total_cost)ÚcopyÚdeepcopyÚmodelÚdatasetÚdeviceÚidÚsave_folder_nameÚ num_classesÚ
train_samplesÚ test_samplesÚ
batch_sizeÚlocal_learning_rateÚ
learning_rateÚ local_stepsZ
has_BatchNormÚchildrenÚ
isinstanceÚnnÚ BatchNorm2dr r Útrain_time_costZsend_time_costÚprivacyZdp_sigmaÚ sample_rate)ÚselfÚargsr r r ÚkwargsÚlayer© r$ úG/root/autodl-tmp/PFL-Non-IID-master/system/flcore/clients/clientbase.pyÚ__init__ s.
zClient.__init__Nc C s0 |d kr| j }t| j| jdd�}t||ddd�S )NT)Úis_train)Ú drop_lastÚshuffle)r r r r r )r r Ú
train_datar$ r$ r% Úload_train_data1 s zClient.load_train_datac C s0 |d kr| j }t| j| jdd�}t||ddd�S )NF)r' T)r( r) )r r r r r )r r Ú test_datar$ r$ r% Úload_test_data7 s zClient.load_test_datac C s2 x,t | ¡ | j ¡ ƒD ]\}}|j ¡ |_qW d S )N)ÚzipÚ
parametersr
ÚdataÚclone)r r
Ú new_paramÚ old_paramr$ r$ r% Úset_parameters= s zClient.set_parametersc C s0 x*t | ¡ | ¡ ƒD ]\}}|j ¡ |_qW d S )N)r. r/ r0 r1 )r r
ÚtargetÚparamZ target_paramr$ r$ r% Ú clone_modelA s zClient.clone_modelc C s, x&t | ¡ |ƒD ]\}}|j ¡ |_qW d S )N)r. r/ r0 r1 )r r
Ú
new_paramsr6 r2 r$ r$ r% Úupdate_parametersF s zClient.update_parametersc
C sH | ¡ }| j | j¡ | j ¡ d}d}g }g }t ¡ �Ê xÂ|D ]º\}}t|ƒtg ƒkrn|d | j¡|d< n | | j¡}| | j¡}| |¡}|t tj |dd�|k¡
¡ 7 }||j d 7 }| |
¡ ¡ ¡ ¡ | t|
¡ ¡ ¡ t | j¡d�¡ q@W W d Q R X | j ¡ tj|dd�}tj|dd�}tj||dd�} ||| fS )Nr é )Údim)Úclasses)ÚaxisÚmicro)Úaverage)r- r
Útor ÚevalÚtorchÚno_gradÚtypeÚsumÚargmaxÚitemÚshapeÚappendÚdetachÚcpuÚnumpyr ÚnpÚaranger Ú concatenater Z
roc_auc_score)
r ZtestloaderfullZtest_accZtest_numZy_probZy_trueÚxÚyÚoutputZaucr$ r$ r% Ú test_metricsJ s.
4
zClient.test_metricsc C sÀ | ¡ }| j | j¡ | j ¡ d}d}x„|D ]|\}}t|ƒtg ƒkr\|d | j¡|d< n | | j¡}| | j¡}| |¡}||jd 7 }|| ||¡ ¡ |jd 7 }q.W | j ¡ ||fS )Nr )
r+ r
r@ r rA rD rH ÚlossrG rK )r Ú trainloaderZ train_numrT rP rQ rR r$ r$ r% Ú
train_metricso s
"
zClient.train_metricsc C sT |d kr| j }tj |¡s$t |¡ t |tj |dt| j ƒ d | d ¡¡ d S )NÚclient_Ú_z.pt)
r ÚosÚpathÚexistsÚmakedirsrB ÚsaveÚjoinÚstrr )r rG Ú item_nameÚ item_pathr$ r$ r% Ú save_item— s
zClient.save_itemc C s8 |d kr| j }t tj |dt| jƒ d | d ¡¡S )NrW rX z.pt)r rB ÚloadrY rZ r^ r_ r )r r` ra r$ r$ r% Ú load_item� s zClient.load_item)N)N)N)N)Ú__name__Ú
__module__Ú __qualname__Ú__doc__r& r+ r- r4 r7 r9 rS rV rb rd r$ r$ r$ r% r
s
%(
r )r rB Útorch.nnr rL rM rY Útorch.nn.functionalÚ
functionalÚFÚtorch.utils.datar Zsklearn.preprocessingr Zsklearnr Zutils.data_utilsr Úobjectr r$ r$ r$ r% Ú<module> s | 4,413 | Python | .pyt | 40 | 109.05 | 438 | 0.433699 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
2,286,899 | clientrep.cpython-37.pyc | hkgdifyu_pFedPT/system/flcore/clients/__pycache__/clientrep.cpython-37.pyc | B
À:cc ã @ sD d dl Z d dlmZ d dlZd dlZd dlmZ G dd„ deƒZdS )é N)ÚClientc s, e Zd Z‡ fdd„Zdd„ Zdd„ Z‡ ZS )Ú clientRepc sd t ƒ j||||f|� t ¡ | _tjj| jj
¡ | j d�| _ tjj| jj
¡ | j d�| _|j| _d S )N)Úlr)ÚsuperÚ__init__ÚnnÚCrossEntropyLossÚlossÚtorchÚoptimÚSGDÚmodelÚbaseÚ
parametersÚ
learning_rateÚ optimizerÚ predictorÚ
poptimizerÚ plocal_steps)ÚselfÚargsÚidÚ
train_samplesÚ test_samplesÚkwargs)Ú __class__© úF/root/autodl-tmp/PFL-Non-IID-master/system/flcore/clients/clientrep.pyr
s
zclientRep.__init__c C sx | ¡ }t ¡ }| j | j¡ | j ¡ x| jj ¡ D ]
}d|_q6W x| jj ¡ D ]
}d|_qRW x¼t
| j ƒD ]®}x¨t |ƒD ]œ\}\}}t
|ƒt
g ƒkr¬|d | j¡|d< n | | j¡}| | j¡}| jrät dt tj ¡ ¡ ¡ | j ¡ | |¡}| ||¡} | ¡ | j ¡ qzW qlW | j}
| j�r>tj d|
d ¡}
x | jj ¡ D ] }d|_�qLW x | jj ¡ D ] }d|_�qjW xÂt
|
ƒD ]¶}x®t |ƒD ]¢\}\}}t
|ƒt
g ƒk�rÆ|d | j¡|d< n | | j¡}| | j¡}| j�r t dt tj ¡ ¡ ¡ | j ¡ | |¡}| ||¡} | ¡ | j ¡ �q’W �q„W | j ¡ | j d d7 < | j d t ¡ | 7 < d S ) NFTr gš™™™™™¹?é é Ú
num_roundsÚ
total_cost) Úload_train_dataÚtimer
ÚtoÚdeviceÚtrainr r Ú
requires_gradr Úranger Ú enumerateÚtypeÚ
train_slowÚsleepÚnpÚabsÚrandomÚrandr Ú zero_gradr ÚbackwardÚstepÚ local_stepsÚrandintr ÚcpuÚtrain_time_cost) r Ú trainloaderÚ
start_timeÚparamr3 ÚiÚxÚyÚoutputr Úmax_local_stepsr r r r& sX
zclientRep.trainc C s4 x.t | ¡ | jj ¡ ƒD ]\}}|j ¡ |_qW d S )N)Úzipr r
r ÚdataÚclone)r r Ú new_paramÚ old_paramr r r Úset_parametersL s zclientRep.set_parameters)Ú__name__Ú
__module__Ú __qualname__r r& rE Ú
__classcell__r r )r r r s 9r ) r
Útorch.nnr Únumpyr- r# Z system.flcore.clients.clientbaser r r r r r Ú<module> s
| 2,317 | Python | .pyt | 32 | 71.25 | 278 | 0.427384 | hkgdifyu/pFedPT | 8 | 3 | 0 | GPL-2.0 | 9/5/2024, 10:48:09 PM (Europe/Amsterdam) |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.