+
j% t ^ RI Ht ^ RIHtHtHt ^ RIt^ RIHtHt ^RI H
t
^RI H t ^RI
Ht ^RIHtHtHt ^R IHt ^R
IHtHtHt . R Ot ! R
R]P4 4 t ! R R ]P4 4 t RRRR R]/t ! R R ]4 t ]! 4 ]! R] P> 3R7 RRRR/R R ll4 4 t R# ) )partial)AnyCallableOptionalN)nnTensor)Conv2dNormActivation)ImageClassification)_log_api_usage_once)register_modelWeights WeightsEnum)_IMAGENET_CATEGORIES)_make_divisible_ovewrite_named_paramhandle_legacy_interface MobileNetV2MobileNet_V2_Weightsc T a a ] tR t^t oRV3R lV 3R llltV3R lR ltRtVtV ;t# )InvertedResidualc
t < V ^8 d QhRS[ RS[ RS[ RS[ RS[S[RS[P 3, , RR/# ) inpoupstride expand_ratio
norm_layer.returnN)intr r r Module)format
__classdict__s "_/home/luca/screen-checking/.venv/lib/python3.14/site-packages/torchvision/models/mobilenetv2.py __annotate__ InvertedResidual.__annotate__ sU &! &!&! &!*-&!=@&!NVW_`cegenen`nWoNp&!
&! c R < \ SV ` 4 W0n VR9 d \ RV 24 hVf \ P
p\
\ W, 4 4 pV P ^8H ;' d W8H V n . pV^8w d- VP \ W^V\ P R7 4 VP \ VVVVV\ P R7 \ P ! Wb^^^ RR7 V! V4 .4 \ P ! V! V n W n V^8 V n R# ) z#stride should be 1 or 2 instead of N kernel_sizer activation_layer)r groupsr r+ F)bias)r( r )super__init__r
ValueErrorr BatchNorm2dr rounduse_res_connectappendr ReLU6extendConv2d
Sequentialconv out_channels_is_cn) selfr r r r r
hidden_dimlayers __class__s &&&&&& r# r/ InvertedResidual.__init__ s B6( KL L Js 123
#{{a / > >CJ "$ 1 MM$S!PZmomumuv
% !% )%'XX
*1a?3
MM6* qj r& c & < V ^8 d QhRS[ RS[ /# r xr r )r! r" s "r# r$ r% < s F r& c v V P ' d WP V4 , # V P V4 # N)r3 r9 r< rC s &&r# forwardInvertedResidual.forward< s- yy|# #99Q< r& )r; r9 r: r r3 rF ) __name__
__module__ __qualname____firstlineno__r/ rH __static_attributes____classdictcell__
__classcell__r? r" s @@r# r r s &! &!P r& r c f a a ] tR t^Ct oRV3R lV 3R llltV3R lR ltV3R lR ltRtVtV ;t # ) r c < V ^8 d QhRS[ RS[RS[S[S[S[ , , , RS[ RS[S[RS[P 3, , RS[S[RS[P 3, , RS[R R
/# ) r num_classes
width_multinverted_residual_setting
round_nearestblock.r dropoutr N)r floatr listr r r )r! r" s "r# r$ MobileNetV2.__annotate__D s ]' ]']' ]' $,DcO#< ]'
]' bii 01
]' Xc299n 56]' ]'
]'r& c < \ SV ` 4 \ V 4 Vf \ pVf \ P
p^ pRp Vf . R O. R
O. RO. RO. RO. RO. RO.p\
V4 ^ 8X g \
V^ , 4 ^8w d \ RV 24 h\ W, V4 p\ V \ RV4 , V4 V n
\ ^V^V\ P R7 .p
V FQ w rr\ W, V4 p\ V
4 F* pV^ 8X d TM^pV
P V! WVWR7 4 TpK, KS V
P \ WP ^V\ P R7 4 \ P ! V
! V n \ P ! \ P" ! VR7 \ P$ ! V P V4 4 V n V P) 4 EF p\+ V\ P, 4 ' dg \ P. P1 VP2 R R
7 VP4 e, \ P. P7 VP4 4 K K \+ V\ P
\ P8 34 ' dV \ P. P; VP2 4 \ P. P7 VP4 4 EK \+ V\ P$ 4 ' g EK2 \ P. P= VP2 ^ R 4 \ P. P7 VP4 4 EK R# )a
MobileNet V2 main class
Args:
num_classes (int): Number of classes
width_mult (float): Width multiplier - adjusts number of channels in each layer by this amount
inverted_residual_setting: Network structure
round_nearest (int): Round the number of channels in each layer to be a multiple of this number
Set to 1 to turn off rounding
block: Module specifying inverted residual building block for mobilenet
norm_layer: Module specifying the normalization layer to use
dropout (float): The droupout probability
Ni zGinverted_residual_setting should be non-empty or a 4-element list, got ?)r r r+ )r r r) )pfan_out)modeg{Gz?)r( r( r( ) r r )rc r )rc @ r )rc ` rf r( )rc rf r )rc i@ r( r( ) r. r/ r r r r1 lenr0 r max last_channelr r5 ranger4 r8 featuresDropoutLinear
classifiermodules
isinstancer7 initkaiming_normal_weightr- zeros_ GroupNormones_normal_)r< rT rU rV rW rX r rY
input_channelrm ro tcnsoutput_channelir mr? s &&&&&&&& r# r/ MobileNetV2.__init__D s 0 D! =$E J
$ , ) % ( )Q .#6OPQ6R2SWX2X YZsYtu
(
(BMR
+L3sJ;O,OQ^ _ M!
egemem n%
4JA! ,Q^] KN1X 1f!mVZ[ st .
4 0 0aJikiqiq
x0
-- JJ ! IId'' 5
A!RYY''''y'A66%GGNN166* &A =>>
ahh'qvv&Aryy))!T2qvv& r& c & < V ^8 d QhRS[ RS[ /# rB rD )r! r" s "r# r$ r\ s v & r& c V P V4 p\ P P VR4 p\ P
! V^4 pV P
V4 pV# )r( r( r( )ro r
functionaladaptive_avg_pool2dtorchflattenrr rG s &&r#
_forward_implMobileNetV2._forward_impl sK
MM! MM - -a 8 MM!Q OOA r& c & < V ^8 d QhRS[ RS[ /# rB rD )r! r" s "r# r$ r\ s % % %F %r& c $ V P V4 # rF )r rG s &&r# rH MobileNetV2.forward s !!!$$r& )rr ro rm )i r^ N NNg?)
rJ rK rL rM r/ r rH rN rO rP rQ s @@r# r r C s+ ]' ]'~ % % %r&
num_paramsiz5 min_size
categoriesc ] tR t^t]! R]! ]^R7 / ]CRRRRRRR R
//R R R
RRR/CR7 t]! R]! ]^^R7 / ]CRRRRRRR R//R R R
RRR/CR7 t ] t
Rt R# )r z=https://download.pytorch.org/models/mobilenet_v2-b0353104.pth) crop_sizerecipezQhttps://github.com/pytorch/vision/tree/main/references/classification#mobilenetv2_metricsz ImageNet-1Kzacc@1gx&1Q@zacc@5gMV@_opsg$C?
_file_sizeg\( +@_docszXThese weights reproduce closely the results of the paper using a simple training recipe.)url
transformsmetaz=https://download.pytorch.org/models/mobilenet_v2-7ebf99e0.pth)r resize_sizezHhttps://github.com/pytorch/vision/issues/3995#new-recipe-with-reg-tuningg`" R@gS㥛V@gV-2+@a$
These weights improve upon the results of the original paper by using a modified version of TorchVision's
`new training recipe
`_.
N) rJ rK rL rM r
r r
_COMMON_META
IMAGENET1K_V1
IMAGENET1K_V2DEFAULTrN r r& r# r r s K.#>
i VV
E & s
M" K.#3O
` VV
E &
M* Gr&
pretrained)weightsr progressTc ^ V ^8 d QhR\ \ , R\ R\ R\ /# )r r r kwargsr )r r boolr r )r! s "r# r$ r$ s2 -. AE X[ r& c \ P V 4 p V e( \ VR\ V P R, 4 4 \ R/ VB pV e# VP
V P VRR7 4 V# )aq MobileNetV2 architecture from the `MobileNetV2: Inverted Residuals and Linear
Bottlenecks `_ paper.
Args:
weights (:class:`~torchvision.models.MobileNet_V2_Weights`, optional): The
pretrained weights to use. See
:class:`~torchvision.models.MobileNet_V2_Weights` below for
more details, and possible values. By default, no pre-trained
weights are used.
progress (bool, optional): If True, displays a progress bar of the
download to stderr. Default is True.
**kwargs: parameters passed to the ``torchvision.models.mobilenetv2.MobileNetV2``
base class. Please refer to the `source code
`_
for more details about this class.
.. autoclass:: torchvision.models.MobileNet_V2_Weights
:members:
rT r T)r
check_hashr )r verifyr rk r r load_state_dictget_state_dict)r r r models $$, r# mobilenet_v2r sk 0 #))'2G fmS l9S5TU !& !E
g 4 4hSW 4 XY Lr& )r r r r )! functoolsr typingr r r r r r ops.miscr transforms._presetsr
utilsr _apir r
r _metar _utilsr r r __all__r r r r r r r r r& r# r s * * + 5 ' 6 6 ' S S B- ryy - `k%")) k%^ '& ' ; ' T ,0D0R0R!ST 15 HL U r&