§
    ·¯¸j<  ã            	       ó¾   — d Z ddlZddlmZ ddlmZmZmZmZm	Z	m
Z
mZmZ ddlZddlmZmZ  G d„ deeeeef                           ¦  «        Z G d„ d	e¦  «        ZdS )
zGScheduled batch sampling and transform dispatch for map-style datasets.é    N)Úislice)ÚAnyÚCallableÚIteratorÚListÚOptionalÚSequenceÚTupleÚUnion)ÚDatasetÚSamplerc                   ó„  — e Zd ZdZ	 	 	 	 	 	 	 	 	 d&ded	ee         d
eee                  dede	de	dee         de
dee         dededdfd„Zd
eee                  dej        fd„Zdedej        fd„Zdefd„Zdefd„Z	 	 d'dej        deee                  d
eej                 defd„Zdeeeef         df         fd„Zdedeeeef         df         fd„Zdedeeeef         df         fd „Zedeeeef         df         fd!„¦   «         Zdee         fd"„Zdeeeeef                           fd#„Zdefd$„Zdeddfd%„Z dS )(ÚScheduledBatchSamplera]  Batch sampler that associates each batch with a scheduled transform choice.

    In sample-budget mode, a canonical composition of ``(choice index, batch
    size)`` pairs is created once and shuffled at the start of each iteration.
    Fixed-batch progressive mode creates a schedule for the current epoch,
    moving a probability window from the first choice to the last while
    retaining configurable random exploration.

    The same seed, sampler length, and epoch produce an identical batch-shape
    schedule on every distributed rank. If a fixed number of batches requires
    more indices than the wrapped sampler provides, its index stream is cycled.

    The yielded indices are ``(sample index, choice index)`` pairs intended for
    use with :class:`ScheduledTransformDataset`.

    Args:
        sampler: Base sample-index sampler.
        batch_sizes: Batch size corresponding to each transform choice.
        choice_weights: Base sampling weights for the choices. A zero weight
            excludes that choice from all schedules.
        seed: Seed shared by schedule creation and shuffling.
        drop_last: Drop an undersized tail in sample-budget mode.
        shuffle_schedule: Shuffle batch order within each epoch.
        num_batches: Fixed number of batches per epoch. Progressive mode infers
            this from its training-average expected batch size when omitted.
        choice_schedule: Either ``constant`` or ``progressive``.
        schedule_epochs: Number of epochs over which progressive choices move
            from the first choice to the last.
        schedule_spread: Progressive probability-window standard deviation in
            choice-index units. Zero selects only the nearest choice.
        schedule_random_mix: Fraction of uniform exploration over active
            choices mixed into the progressive choice probabilities.
    Nr   TÚconstantçÍÌÌÌÌÌä?çš™™™™™¹?ÚsamplerÚbatch_sizesÚchoice_weightsÚseedÚ	drop_lastÚshuffle_scheduleÚnum_batchesÚchoice_scheduleÚschedule_epochsÚschedule_spreadÚschedule_random_mixÚreturnc                 ó  — t          |d¦  «        st          d¦  «        ‚t          |¦  «        dk    rt          d¦  «        ‚|st          d¦  «        ‚t	          d„ |D ¦   «         ¦  «        rt          d¦  «        ‚|�(t          |¦  «        |k    s|dk    rt          d¦  «        ‚|d	vrt          d
¦  «        ‚|dk    rt          |¦  «        dk     rt          d¦  «        ‚|	�t          |	¦  «        |	k    s|	dk    rt          d¦  «        ‚|
dk     rt          d¦  «        ‚d|cxk    rdk    sn t          d¦  «        ‚|| _        t          d„ |D ¦   «         ¦  «        | _        |  	                    |¦  «        | _
        t          d„ t          | j
        ¦  «        D ¦   «         ¦  «        | _        || _        || _        || _        || _        |	�t          |	¦  «        nd | _        |
| _        || _        d| _        |                      ¦   «         | _        |dk    r|€|                      ¦   «         }|�t          |¦  «        nd | _        d| _        | j        €/|                      ¦   «         | _        | j        st          d¦  «        ‚d S d S )NÚ__len__z7ScheduledBatchSampler requires a sampler with a length.r   z3ScheduledBatchSampler requires a non-empty sampler.z,batch_sizes must contain at least one value.c              3   óH   K  — | ]}t          |¦  «        |k    p|d k    V — ŒdS ©r   N©Úint©Ú.0Ú
batch_sizes     úk/root/private/project/pallet_detection_v2/.venv/lib/python3.11/site-packages/timm/data/scheduled_sampler.pyú	<genexpr>z1ScheduledBatchSampler.__init__.<locals>.<genexpr>B   s6   è è € Ð^Ð^ÀJ�s�:‰Œ *Ò,Ð?°
¸a²Ð^Ð^Ð^Ð^Ð^Ð^ó    z4All scheduled batch sizes must be positive integers.z6num_batches must be a positive integer when specified.)r   Úprogressivez4choice_schedule must be 'constant' or 'progressive'.r+   é   z5A progressive schedule requires at least two choices.zFschedule_epochs must be a positive integer for a progressive schedule.z%schedule_spread must be non-negative.é   z,schedule_random_mix must be between 0 and 1.c              3   ó4   K  — | ]}t          |¦  «        V — Œd S ©Nr#   r%   s     r(   r)   z1ScheduledBatchSampler.__init__.<locals>.<genexpr>S   s(   è è € Ð OÐ O°Z¥ Z¡¤Ð OÐ OÐ OÐ OÐ OÐ Or*   c              3   ó,   K  — | ]\  }}|d k    ¯|V — ŒdS r"   © )r&   Úchoice_indexÚchoice_weights      r(   r)   z1ScheduledBatchSampler.__init__.<locals>.<genexpr>U   s=   è è € ð %
ð %
á+�˜mØ˜qÒ Ð ð à Ð Ð Ð ð%
ð %
r*   r1   úANo full scheduled batch fits the sampler; reduce the batch sizes.)ÚhasattrÚ	TypeErrorÚlenÚ
ValueErrorÚanyr$   r   Útupler   Ú_normalize_choice_weightsr   Ú	enumerateÚ_active_choicesr   r   r   r   r   r   r   ÚepochÚ_calculate_average_batch_sizeÚaverage_batch_sizeÚ_infer_num_batchesr   Ú_sample_budget_scheduleÚ_create_sample_budget_schedule)Úselfr   r   r   r   r   r   r   r   r   r   r   s               r(   Ú__init__zScheduledBatchSampler.__init__.   sñ  € õ �w 	Ñ*Ô*ð 	WÝÐUÑVÔVÐVÝˆw‰<Œ<˜1ÒÐÝÐRÑSÔSÐSØð 	MÝÐKÑLÔLÐLÝÐ^Ð^ÐR]Ð^Ñ^Ô^Ñ^Ô^ð 	UÝÐSÑTÔTÐTØÐ"­¨KÑ(8Ô(8¸KÒ(GÐ(GÈ;ÐZ[ÒK[ÐK[ÝÐUÑVÔVÐVØÐ"=Ð=Ð=ÝÐSÑTÔTÐTØ˜mÒ+Ð+Ý�;ÑÔ !Ò#Ð#Ý Ð!XÑYÔYÐYØÐ&­#¨oÑ*>Ô*>À/Ò*QÐ*QÐUdÐhiÒUiÐUiÝ Ð!iÑjÔjÐjØ Ò"Ð"Ý Ð!HÑIÔIÐIØÐ+Ð0Ð0Ò0Ð0¨qÒ0Ð0Ð0Ð0Ý Ð!OÑPÔPÐPàˆŒÝ Ð OÐ OÀ;Ð OÑ OÔ OÑOÔOˆÔØ"×<Ò<¸^ÑLÔLˆÔÝ$ð %
ð %
å/8¸Ô9LÑ/MÔ/Mð%
ñ %
ô %
ñ  
ô  
ˆÔð
 ˆŒ	Ø"ˆŒØ 0ˆÔØ.ˆÔØ7FÐ7R�s ?Ñ3Ô3Ð3ÐX\ˆÔØ.ˆÔØ#6ˆÔ ØˆŒ
Ø"&×"DÒ"DÑ"FÔ"FˆÔØ˜mÒ+Ð+°Ð0CØ×1Ò1Ñ3Ô3ˆKØ/:Ð/F�3˜{Ñ+Ô+Ð+ÈDˆÔØDFˆÔ$ØÔÐ#Ø+/×+NÒ+NÑ+PÔ+PˆDÔ(ØÔ/ð Ý ØWñô ð ð $Ð#ðð r*   c                 ó6  — |€It          j        t          | j        ¦  «        fdt          | j        ¦  «        z  t           j        ¬¦  «        S t          |¦  «        t          | j        ¦  «        k    rt          d¦  «        ‚t          j        |t           j        ¬¦  «        }t          j        |¦  «                             ¦   «         r|dk      	                    ¦   «         rt          d¦  «        ‚| 
                    ¦   «         }|dk    rt          d¦  «        ‚||z  S )Nç      ð?©Údtypez9choice_weights and batch_sizes must have the same length.r   z8choice_weights must contain finite, non-negative values.z(choice_weights must have a positive sum.)ÚtorchÚfullr7   r   Úfloat64r8   ÚtensorÚisfiniteÚallr9   Úsum)rD   r   ÚweightsÚ
weight_sums       r(   r;   z/ScheduledBatchSampler._normalize_choice_weightsn   sü   € ð Ð!Ý”:�s 4Ô#3Ñ4Ô4Ð6¸½cÀ$ÔBRÑ>SÔ>SÑ8SÕ[`Ô[hÐiÑiÔiÐiÝˆ~ÑÔ¥# dÔ&6Ñ"7Ô"7Ò7Ð7ÝÐXÑYÔYÐYå”,˜~µU´]ÐCÑCÔCˆÝŒ~˜gÑ&Ô&×*Ò*Ñ,Ô,ð 	Y°¸1²×0AÒ0AÑ0CÔ0Cð 	YÝÐWÑXÔXÐXØ—[’[‘]”]ˆ
Ø˜Š?ˆ?ÝÐGÑHÔHÐHØ˜Ñ#Ð#r*   r>   c                 ó  ‡— | j         dk    st          | j        ¦  «        dk    r| j        S | j        dk    rd}n)t          t          || j        dz
  z  d¦  «        d¦  «        }t          j        t          | j        ¦  «        t          j	        ¬¦  «        }|t          | j        ¦  «        dz
  z  Š| j
        dk    rM|‰z
                       ¦   «         }||                     ¦   «         k                         t          j	        ¦  «        }n%t          j        d|‰z
  | j
        z  dz  z  ¦  «        }|| j        z  }|                     ¦   «         dk    r8t          | j        ˆfd	„¬
¦  «        }t          j        | j        ¦  «        }d||<   n||                     ¦   «         z  }| j        rR| j        dk                         |j        ¦  «        }||                     ¦   «         z  }d| j        z
  |z  | j        |z  z   }||                     ¦   «         z  S )z¼Return normalized choice weights for an epoch.

        Args:
            epoch: Zero-based training epoch.

        Returns:
            Normalized floating-point choice weights.
        r   r-   rG   g        rH   r   g      à¿r,   c                 ó(   •— t          | ‰z
  ¦  «        S r/   )Úabs)ÚindexÚcenters    €r(   ú<lambda>z@ScheduledBatchSampler.choice_weights_for_epoch.<locals>.<lambda>›   s   ø€ ÍÈUÐU[É^ÑI\ÔI\€ r*   )Úkey)r   r7   r   r   r   ÚminÚmaxrJ   ÚarangerL   r   rU   ÚtoÚexprP   r=   Ú
zeros_liker   rI   )	rD   r>   ÚprogressÚchoice_positionsÚ	distancesÚcurriculum_weightsÚnearest_activeÚrandom_weightsrW   s	           @r(   Úchoice_weights_for_epochz.ScheduledBatchSampler.choice_weights_for_epoch   s!  ø€ ð Ô :Ò-Ð-µ°TÔ5EÑ1FÔ1FÈ!Ò1KÐ1KØÔ&Ð&àÔ 1Ò$Ð$ØˆHˆHå�3˜u¨Ô(<¸qÑ(@ÑAÀ3ÑGÔGÈÑMÔMˆHÝ œ<­¨DÔ,<Ñ(=Ô(=ÅUÄ]ÐSÑSÔSÐØ�S Ô!1Ñ2Ô2°QÑ6Ñ7ˆàÔ 1Ò$Ð$Ø)¨FÑ2×7Ò7Ñ9Ô9ˆIØ"+¨y¯}ª}©¬Ò">×!BÒ!BÅ5Ä=Ñ!QÔ!QÐÐå!&¤ØÐ)¨FÑ2°dÔ6JÑJÈqÑPÑPñ"ô "Ðð 	˜dÔ1Ñ1ÐØ×!Ò!Ñ#Ô# qÒ(Ð(Ý  Ô!5Ð;\Ð;\Ð;\Ð;\Ð]Ñ]Ô]ˆNÝ!&Ô!1°$Ô2EÑ!FÔ!FÐØ14Ð˜~Ñ.Ð.àÐ"4×"8Ò"8Ñ":Ô":Ñ:ÐàÔ#ð 	OØ"Ô1°AÒ5×9Ò9Ð:LÔ:RÑSÔSˆNØ˜n×0Ò0Ñ2Ô2Ñ2ˆNà�dÔ.Ñ.Ø"ñ"#à%)Ô%=ÀÑ%Nñ"OÐð "Ð$6×$:Ò$:Ñ$<Ô$<Ñ<Ð<r*   c                 ó¤  ‡ ‡— t          j        ‰ j        t           j        ¬¦  «        Š‰ j        dk    rfˆˆ fd„t          ‰ j        ¦  «        D ¦   «         }t          t          j        |¦  «         	                    ¦   «          
                    ¦   «         ¦  «        S t          t          j        ‰ j        ‰¦  «         
                    ¦   «         ¦  «        S )NrH   r+   c                 ó`   •— g | ]*}t          j        ‰                     |¦  «        ‰¦  «        ‘Œ+S r1   )rJ   Údotrf   )r&   r>   r   rD   s     €€r(   ú
<listcomp>zGScheduledBatchSampler._calculate_average_batch_size.<locals>.<listcomp>¬   s@   ø€ ð $ð $ð $ØQV•”	˜$×7Ò7¸Ñ>Ô>ÀÑLÔLð$ð $ð $r*   )rJ   rM   r   rL   r   Úranger   ÚfloatÚstackÚmeanÚitemri   r   )rD   Úexpected_batch_sizesr   s   ` @r(   r?   z3ScheduledBatchSampler._calculate_average_batch_size©   sÁ   øø€ Ý”l 4Ô#3½5¼=ÐIÑIÔIˆØÔ =Ò0Ð0ð$ð $ð $ð $ð $ÝZ_Ð`dÔ`tÑZuÔZuð$ñ $ô $Ð õ �œÐ%9Ñ:Ô:×?Ò?ÑAÔA×FÒFÑHÔHÑIÔIÐIÝ•U”Y˜tÔ2°KÑ@Ô@×EÒEÑGÔGÑHÔHÐHr*   c                 óä  — t          t          | j        ¦  «        ¦  «        dk    rV| j        d         }| j        rt          | j        ¦  «        |z  }n‰t          j        t          | j        ¦  «        |z  ¦  «        }n_| j        r*t          t          | j        ¦  «        | j        z  ¦  «        }n.t          j        t          | j        ¦  «        | j        z  ¦  «        }|dk     rt          d¦  «        ‚|S )Nr-   r   r4   )
r7   Úsetr   r   r   ÚmathÚceilr$   r@   r8   )rD   r'   r   s      r(   rA   z(ScheduledBatchSampler._infer_num_batches²   sÙ   € Ý�s�4Ô#Ñ$Ô$Ñ%Ô%¨Ò*Ð*ØÔ)¨!Ô,ˆJØŒ~ð HÝ! $¤,Ñ/Ô/°:Ñ=��å"œi­¨D¬LÑ(9Ô(9¸JÑ(FÑGÔG��àŒ~ð UÝ!¥# d¤lÑ"3Ô"3°dÔ6MÑ"MÑNÔN��å"œi­¨D¬LÑ(9Ô(9¸DÔ<SÑ(SÑTÔT�Ø˜Š?ˆ?ÝÐ`ÑaÔaÐaØÐr*   Ú	generatorÚvalid_choicesc                 óš  — |€| j         n|}|€6t          t          j        |d|¬¦  «                             ¦   «         ¦  «        S t          |¦  «        }|t          |¦  «                 }|                     ¦   «         dk    rt          d¦  «        ‚t          t          j        |d|¬¦  «                             ¦   «         ¦  «        }||         S )Nr-   ©ru   r   z@No positive-weight scheduled choice is available for this batch.)	r   r$   rJ   Úmultinomialro   r:   ÚlistrP   ÚRuntimeError)rD   ru   rv   r   rQ   Úsampled_indexs         r(   Ú_sample_choicez$ScheduledBatchSampler._sample_choiceÂ   s¿   € ð 1?Ð0F˜Ô,Ð,ÈNˆØÐ Ý•uÔ(¨¸ÀiÐPÑPÔP×UÒUÑWÔWÑXÔXÐXå˜mÑ,Ô,ˆØ ¥ mÑ!4Ô!4Ô5ˆØ�;Š;‰=Œ=˜AÒÐÝÐaÑbÔbÐbÝ�EÔ-¨g°qÀIÐNÑNÔN×SÒSÑUÔUÑVÔVˆØ˜]Ô+Ð+r*   .c                 ó
  ‡ ‡— t          j        ¦   «                              ‰ j        ¦  «        }t	          ‰ j        ¦  «        Št          ˆ fd„‰ j        D ¦   «         ¦  «        }g }‰|k    rYˆˆ fd„‰ j        D ¦   «         }‰                      ||¦  «        }‰ j	        |         }| 
                    ||f¦  «         ‰|z  Š‰|k    °Y‰r1‰ j        s*| 
                    ‰                      |¦  «        ‰f¦  «         t          |¦  «        S )Nc              3   ó2   •K  — | ]}‰j         |         V — Œd S r/   ©r   )r&   rV   rD   s     €r(   r)   zGScheduledBatchSampler._create_sample_budget_schedule.<locals>.<genexpr>Ö   s+   øè è € ÐWÐW¸˜TÔ-¨eÔ4ÐWÐWÐWÐWÐWÐWr*   c                 ó6   •— g | ]}‰j         |         ‰k    ¯|‘ŒS r1   r€   )r&   r2   Ú	remainingrD   s     €€r(   rj   zHScheduledBatchSampler._create_sample_budget_schedule.<locals>.<listcomp>Ú   s4   ø€ ð ð ð Ø!-ÈÔIYÐZfÔIgÐktÒItÐIt�ÐItÐItÐItr*   )rJ   Ú	GeneratorÚmanual_seedr   r7   r   rZ   r=   r}   r   Úappendr   r:   )rD   ru   Úmin_batch_sizeÚschedulerv   r2   r'   r‚   s   `      @r(   rC   z4ScheduledBatchSampler._create_sample_budget_scheduleÓ   s-  øø€ Ý”OÑ%Ô%×1Ò1°$´)Ñ<Ô<ˆ	Ý˜œÑ%Ô%ˆ	ÝÐWÐWÐWÐWÀ$ÔBVÐWÑWÔWÑWÔWˆØˆà˜>Ò)Ð)ðð ð ð ð Ø15Ô1Eðñ ô ˆMð  ×.Ò.¨y¸-ÑHÔHˆLØÔ)¨,Ô7ˆJØ�OŠO˜\¨:Ð6Ñ7Ô7Ð7Ø˜Ñ#ˆIð ˜>Ò)Ð)ð ð 	I˜Tœ^ð 	IØ�OŠO˜T×0Ò0°Ñ;Ô;¸YÐGÑHÔHÐHå�X‰ŒÐr*   c                 óP  — t          j        ¦   «                              | j        d|z  z   ¦  «        }|                      |¦  «        }g }t          | j        ¦  «        D ];}|                      ||¬¦  «        }|                     || j	        |         f¦  «         Œ<t          |¦  «        S )Nr,   )r   )rJ   rƒ   r„   r   rf   rk   r   r}   r…   r   r:   )rD   r>   ru   r   r‡   Ú_r2   s          r(   Ú_create_fixed_batch_schedulez2ScheduledBatchSampler._create_fixed_batch_scheduleç   s¢   € Ý”OÑ%Ô%×1Ò1°$´)¸aÀ%¹iÑ2GÑHÔHˆ	Ø×6Ò6°uÑ=Ô=ˆØˆÝ�tÔ'Ñ(Ô(ð 	Lð 	LˆAØ×.Ò.¨yÈÐ.ÑXÔXˆLØ�OŠO˜\¨4Ô+;¸LÔ+IÐJÑKÔKÐKÐKÝ�X‰ŒÐr*   c                 óH   — | j         �|                      |¦  «        S | j        S r/   )r   rŠ   rB   ©rD   r>   s     r(   Ú_create_schedulez&ScheduledBatchSampler._create_scheduleð   s(   € ØÔÐ'Ø×4Ò4°UÑ;Ô;Ð;ØÔ+Ð+r*   c                 ó6   — |                       | j        ¦  «        S )z@Return the unshuffled schedule for the currently selected epoch.)r�   r>   ©rD   s    r(   r‡   zScheduledBatchSampler.scheduleõ   s   € ð ×$Ò$ T¤ZÑ0Ô0Ð0r*   c              #   ó6   K  — 	 d}| j         D ]}d}|V — Œ	|sd S Œ)NTF)r   )rD   ÚyieldedÚsample_indexs      r(   Ú_cycling_samplerz&ScheduledBatchSampler._cycling_samplerú   sG   è è € ð	ØˆGØ $¤ð #ð #�Ø�Ø"Ð"Ð"Ð"Ð"Øð Ø�ð	r*   c              #   óˆ  ‡‡K  — | j         }|                      |¦  «        Š| j        r—t          ‰¦  «        dk    r„t	          j        ¦   «                              | j        d|z  z   dz   ¦  «        }t	          j        t          ‰¦  «        |¬¦  «         	                    ¦   «         }t          ˆfd„|D ¦   «         ¦  «        Š| j        �|                      ¦   «         nt          | j        ¦  «        }‰D ]Q\  Š}t          t!          ||¦  «        ¦  «        }t          |¦  «        |k    r| j        s|s d S ˆfd„|D ¦   «         V — ŒRd S )Nr-   r,   rx   c              3   ó(   •K  — | ]}‰|         V — Œd S r/   r1   )r&   rV   r‡   s     €r(   r)   z1ScheduledBatchSampler.__iter__.<locals>.<genexpr>	  s'   øè è € Ð@Ð@°˜X eœ_Ð@Ð@Ð@Ð@Ð@Ð@r*   c                 ó   •— g | ]}|‰f‘ŒS r1   r1   )r&   r’   r2   s     €r(   rj   z2ScheduledBatchSampler.__iter__.<locals>.<listcomp>  s   ø€ ÐJÐJÐJ°L�L ,Ð/ÐJÐJÐJr*   )r>   r�   r   r7   rJ   rƒ   r„   r   ÚrandpermÚtolistr:   r   r“   Úiterr   rz   r   r   )	rD   r>   ru   ÚorderÚsample_indicesr'   Úbatchr2   r‡   s	          @@r(   Ú__iter__zScheduledBatchSampler.__iter__  sa  øøè è € Ø”
ˆØ×(Ò(¨Ñ/Ô/ˆØÔ ð 	A¥S¨¡]¤]°QÒ%6Ð%6ÝœÑ)Ô)×5Ò5°d´iÀ!ÀeÁ)Ñ6KÈaÑ6OÑPÔPˆIÝ”N¥3 x¡=¤=¸IÐFÑFÔF×MÒMÑOÔOˆEÝÐ@Ð@Ð@Ð@¸%Ð@Ñ@Ô@Ñ@Ô@ˆHà48Ô4DÐ4P˜×.Ò.Ñ0Ô0Ð0ÕVZÐ[_Ô[gÑVhÔVhˆØ(0ð 	Kð 	KÑ$ˆL˜*Ý� °
Ñ;Ô;Ñ<Ô<ˆEÝ�5‰zŒz˜ZÒ'Ð'Ø”>ð ¨ð Ø�E�EØJÐJÐJÐJÀEÐJÑJÔJÐJÐJÐJÐJð	Kð 	Kr*   c                 óF   — | j         �| j         S t          | j        ¦  «        S r/   )r   r7   rB   r�   s    r(   r    zScheduledBatchSampler.__len__  s%   € ØÔÐ'ØÔ#Ð#Ý�4Ô/Ñ0Ô0Ð0r*   c                 óv   — || _         t          | j        d¦  «        r| j                             |¦  «         dS dS )z=Set the epoch for the next iteration and the wrapped sampler.Ú	set_epochN)r>   r5   r   r    rŒ   s     r(   r    zScheduledBatchSampler.set_epoch  sC   € àˆŒ
Ý�4”< Ñ-Ô-ð 	*ØŒL×"Ò" 5Ñ)Ô)Ð)Ð)Ð)ð	*ð 	*r*   )	Nr   TTNr   Nr   r   )NN)!Ú__name__Ú
__module__Ú__qualname__Ú__doc__r   r	   r$   r   rl   ÚboolÚstrrE   rJ   ÚTensorr;   rf   r?   rA   rƒ   r}   r
   rC   rŠ   r�   Úpropertyr‡   r   r   r“   r   r�   r    r    r1   r*   r(   r   r      s(  € € € € € ð ð  ðL 9=ØØ"Ø%)Ø)-Ø#-Ø-1Ø%)Ø),ð>ð >àð>ð " #œð>ð % X¨e¤_Ô5ð	>ð
 ð>ð ð>ð #ð>ð " #œð>ð !ð>ð & cœ]ð>ð #ð>ð "'ð>ð 
ð>ð >ð >ð >ð@$à$ X¨e¤_Ô5ð$ð 
Œð$ð $ð $ð $ð"(=¨cð (=°e´lð (=ð (=ð (=ð (=ðTI¨uð Ið Ið Ið Ið Cð ð ð ð ð& 6:Ø59ð	,ð ,à”ð,ð $ H¨S¤MÔ2ð,ð % U¤\Ô2ð	,ð
 
ð,ð ,ð ,ð ,ð"°°e¸CÀ¸H´oÀsÐ6JÔ0Kð ð ð ð ð(°#ð ¸%ÀÀcÈ3ÀhÄÐQTÐ@TÔ:Uð ð ð ð ð, cð ,¨e°E¸#¸s¸(´OÀSÐ4HÔ.Ið ,ð ,ð ,ð ,ð
 ð1˜%  c¨3 h¤°Ð 4Ô5ð 1ð 1ð 1ñ „Xð1ð (¨3¤-ð ð ð ð ðK˜( 4¨¨c°3¨h¬Ô#8Ô9ð Kð Kð Kð Kð 1˜ð 1ð 1ð 1ð 1ð
*˜sð * tð *ð *ð *ð *ð *ð *r*   r   c            	       óÊ   — e Zd ZdZdedee         ddfd„Zdee	e
f         deee	df         ee	         f         fd	„Zde
fd
„Zde
ddfd„Zdde
dedede	fd„Zddedede	fd„ZdS )ÚScheduledTransformDataseta  Apply one of several transforms selected by an annotated sample index.

    Args:
        dataset: Map-style base dataset returning tuple or list samples.
        transforms: Transforms indexed by the choice annotation supplied by
            :class:`ScheduledBatchSampler`.
    ÚdatasetÚ
transformsr   Nc                 ó^   — |st          d¦  «        ‚|| _        t          |¦  «        | _        d S )Nz/transforms must contain at least one transform.)r8   r«   r:   r¬   )rD   r«   r¬   s      r(   rE   z"ScheduledTransformDataset.__init__(  s6   € ð
 ð 	PÝÐNÑOÔOÐOØˆŒÝ 
Ñ+Ô+ˆŒˆˆr*   Úscheduled_index.c                 ó€  — |\  }}d|cxk    rt          | j        ¦  «        k     sn t          d|› d�¦  «        ‚| j        |         }t	          |t
          t          f¦  «        r|st          d¦  «        ‚ | j        |         |d         ¦  «        }t	          |t
          ¦  «        r|g|dd …         ¢R S |g|dd …         ¢S )Nr   zTransform index z is out of range.z=ScheduledTransformDataset expects tuple/list dataset samples.r-   )r7   r¬   Ú
IndexErrorr«   Ú
isinstancer:   rz   r6   )rD   r®   r’   Útransform_indexÚsampleÚimages         r(   Ú__getitem__z%ScheduledTransformDataset.__getitem__2  sä   € Ø(7Ñ%ˆ�oØ�OÐ:Ð:Ò:Ð:¥c¨$¬/Ñ&:Ô&:Ò:Ð:Ð:Ð:ÝÐR°ÐRÐRÐRÑSÔSÐSà”˜lÔ+ˆÝ˜&¥5­$ -Ñ0Ô0ð 	]¸ð 	]ÝÐ[Ñ\Ô\Ð\Ø0�” Ô0°¸´Ñ;Ô;ˆÝ�f�eÑ$Ô$ð 	&ØÐ%˜6 ! " "œ:Ð%Ð%Ð%ØÐ#˜˜q˜r˜rœ
Ð#Ð#r*   c                 ó*   — t          | j        ¦  «        S r/   )r7   r«   r�   s    r(   r    z!ScheduledTransformDataset.__len__?  s   € Ý�4”<Ñ Ô Ð r*   r>   c                 óh   — t          | j        d¦  «        r| j                             |¦  «         dS dS )z5Forward epoch updates to an epoch-aware base dataset.r    N)r5   r«   r    rŒ   s     r(   r    z#ScheduledTransformDataset.set_epochB  s<   € å�4”< Ñ-Ô-ð 	*ØŒL×"Ò" 5Ñ)Ô)Ð)Ð)Ð)ð	*ð 	*r*   FrV   ÚbasenameÚabsolutec                 ó<   — | j                              |||¬¦  «        S ©N)r¸   r¹   )r«   Úfilename)rD   rV   r¸   r¹   s       r(   r¼   z"ScheduledTransformDataset.filenameG  s   € ØŒ|×$Ò$ U°XÈÐ$ÑQÔQÐQr*   c                 ó:   — | j                              ||¬¦  «        S r»   )r«   Ú	filenames)rD   r¸   r¹   s      r(   r¾   z#ScheduledTransformDataset.filenamesJ  s   € ØŒ|×%Ò%¨xÀ(Ð%ÑKÔKÐKr*   )FF)r¡   r¢   r£   r¤   r   r	   r   rE   r
   r   r$   r   r   rµ   r    r    r¥   r¼   r¾   r1   r*   r(   rª   rª     sK  € € € € € ðð ð,àð,ð ! Ô*ð,ð 
ð	,ð ,ð ,ð ,ð$¨5°°c°¬?ð $¸uÀUÈ3ÐPSÈ8Ä_ÐVZÐ[^ÔV_ÐE_Ô?`ð $ð $ð $ð $ð!˜ð !ð !ð !ð !ð*˜sð * tð *ð *ð *ð *ð
Rð R˜cð R¨Tð RÀTð RÐVYð Rð Rð Rð RðLð L $ð L¸$ð LÈ3ð Lð Lð Lð Lð Lð Lr*   rª   )r¤   rs   Ú	itertoolsr   Útypingr   r   r   r   r   r	   r
   r   rJ   Útorch.utils.datar   r   r$   r   rª   r1   r*   r(   ú<module>rÂ      s  ðØ MÐ Mà €€€Ø Ð Ð Ð Ð Ð Ø RÐ RÐ RÐ RÐ RÐ RÐ RÐ RÐ RÐ RÐ RÐ RÐ RÐ RÐ RÐ RÐ RÐ RÐ RÐ Rà €€€Ø -Ð -Ð -Ð -Ð -Ð -Ð -Ð -ðQ*ð Q*ð Q*ð Q*ð Q*˜G D¨¨s°C¨x¬Ô$9Ô:ñ Q*ô Q*ð Q*ðh,Lð ,Lð ,Lð ,Lð ,L ñ ,Lô ,Lð ,Lð ,Lð ,Lr*   