§
    “©ºj·K  ã                   óÊ   — d Z ddlZddlmZ ddlmZmZmZ ddlZddl	m
Z
mZ ddlmZ ddlZddlmZ ddlmZ dd	lmZmZ dd
lmZ  G d„ de¦  «        Z G d„ de¦  «        ZdS )z5
Weight Averaging Callback
^^^^^^^^^^^^^^^^^^^^^^^^^
é    N)Údeepcopy)ÚAnyÚOptionalÚUnion)ÚAveragedModelÚget_ema_avg_fn)Úoverride)ÚCallback)Úis_overridden)Úrank_zero_infoÚrank_zero_warn)ÚSTEP_OUTPUTc                   ó  — e Zd ZdZ	 	 d"deeej        ee	f                  de
deddfd„Zd#d	ee	         d
ee	         de
fd„Zedddddeddfd„¦   «         Zedddddedede	ddfd„¦   «         Zed$d„¦   «         Zed$d„¦   «         Zed$d„¦   «         Zed$d„¦   «         Zedeeef         fd„¦   «         Zedeeef         ddfd„¦   «         Zedddddeeef         ddfd„¦   «         Zedddddeeef         ddfd„¦   «         Zd%d „Zd%d!„ZdS )&ÚWeightAveraginga)  A callback that updates an averaged model for Stochastic Weight Averaging (SWA) or Exponential Moving Average
    (EMA) after each training step.

    Arguments given to the constructor will be passed to the :class:`AveragedModel` constructor. If no ``device`` is
    specified, the device of the original model will be used. Contrary to :class:`AveragedModel`, ``use_buffers`` is set
    to ``True`` by default. That is, by default the callback will compute running averages for both the parameters and
    the buffers of the model. Setting ``use_buffers`` to ``False`` will cause only the model parameters to be averaged,
    leaving updating the batch normalization statistics to the user (using ``torch.optim.swa_utils.update_bn()``).

    You can provide a custom averaging function with the ``avg_fn`` or ``multi_avg_fn`` parameter. See the
    :class:`AveragedModel` class for details. If no averaging function is provided, the default is to compute the
    equally-weighted average of the weights (SWA).

    You can customize when the average model is updated by overriding the ``should_update()`` method. The callback calls
    it with either ``step_idx`` or ``epoch_idx`` and the method returns a boolean indicating whether to update after the
    given step or epoch. The default is to update after every step.

    During validation and after the training finishes, the current model parameters will be replaced with the averaged
    values.

    See also the documentation on the :ref:`weight averaging callbacks <advanced/training_tricks:Weight Averaging>`
    provided by Lightning.

    Note:
        To ensure that the :class:`AveragedModel` will contain all layers, ``setup()`` will call
        :meth:`~pytorch_lightning.core.hooks.ModelHooks.configure_model` before instantiating the
        :class:`AveragedModel`. However, that hook is not called in a strategy aware context, sharded models do not work
        with weight averaging, and a warning will be issued.

    Example::

        from pytorch_lightning.callbacks import WeightAveraging
        from torch.optim.swa_utils import get_ema_avg_fn

        class EMAWeightAveraging(WeightAveraging):
            def __init__(self):
                super().__init__(avg_fn=get_ema_avg_fn())

            def should_update(self, step_idx=None, epoch_idx=None):
                # Start after 100 steps.
                return (step_idx is not None) and (step_idx >= 100)

        trainer = Trainer(callbacks=EMAWeightAveraging(), max_epochs=10)
        trainer.fit(model, dataloader)

    Args:
        device: By default, the :class:`AveragedModel` will be stored on the same device as the original model. If the
            ``device`` argument is provided, the :class:`AveragedModel` will be stored on this device instead. If you
            run out of GPU memory, you might want to use ``"cpu"``.
        use_buffers: If ``False``, the buffers of the model will not be averaged.
        kwargs: Additional keyword arguments to be passed to the :class:`AveragedModel` constructor, such as ``avg_fn``
            or ``multi_avg_fn``.

    NTÚdeviceÚuse_buffersÚkwargsÚreturnc                 ó¸   — t          |t          ¦  «        rt          j        |¦  «        | _        n|| _        || _        || _        d | _        d| _        d| _	        d S )Nr   éÿÿÿÿ)
Ú
isinstanceÚstrÚtorchr   Ú_deviceÚ_use_buffersÚ_kwargsÚ_average_modelÚ_latest_update_stepÚ_latest_update_epoch)Úselfr   r   r   s       ú|/root/private/project/pallet_detection_v2/.venv/lib/python3.11/site-packages/pytorch_lightning/callbacks/weight_averaging.pyÚ__init__zWeightAveraging.__init__Z   sa   € õ �f�cÑ"Ô"ð 	"Ý?D¼|ÈFÑ?SÔ?SˆDŒLˆLà!ˆDŒLØ'ˆÔØˆŒà7;ˆÔð
 $%ˆÔ ð %'ˆÔ!Ð!Ð!ó    Ústep_idxÚ	epoch_idxc                 ó
   — |duS )aÔ  Called after every optimizer step and after every training epoch to check whether the average model should
        be updated.

        One of the arguments is set to the zero-based index of the last training step or epoch. The default
        implementation returns ``True`` when any ``step_idx`` is provided. The user can customize when the average model
        gets updated by overriding this method.

        Args:
            step_idx: Index of the last optimizer step, or ``None`` when called at the epoch end.
            epoch_idx: Index of the last epoch, or ``None`` when called after an optimizer step.

        Returns:
            ``True`` if the average model should be updated and ``False`` if not.

        N© )r    r$   r%   s      r!   Úshould_updatezWeightAveraging.should_updates   s   € ð  ˜tÐ#Ð#r#   Útrainerú
pl.TrainerÚ	pl_moduleúpl.LightningModuleÚstagec                 óÖ   — |dk    rb| j         p|j        }t          d|¦  «        r#t          d¦  «         |                     ¦   «          t          d||| j        dœ| j        ¤Ž| _        dS dS )a�  Called when fit, validate, test, predict, or tune begins.

        Creates an :class:`AveragedModel` when fit begins.

        Args:
            trainer: The current :class:`~pytorch_lightning.trainer.trainer.Trainer` instance.
            pl_module: The current :class:`~pytorch_lightning.core.LightningModule` instance.
            stage: The :class:`~pytorch_lightning.trainer.trainer.Trainer` state.

        ÚfitÚconfigure_modelz¸You're using the WeightAveraging callback with a model that overrides the configure_model callback. WeightAveraging doesn't support sharding model layers, so you may run out of memory.)Úmodelr   r   Nr'   )	r   r   r   r   r0   r   r   r   r   )r    r)   r+   r-   r   s        r!   ÚsetupzWeightAveraging.setup…   s›   € ð �EŠ>ˆ>Ø”\Ð5 YÔ%5ˆFõ Ð.°	Ñ:Ô:ð ,Ýðuñô ð ð ×)Ò)Ñ+Ô+Ð+å"/ð #Ø¨¸DÔ<Mð#ð #ØQUÔQ]ð#ð #ˆDÔÐÐð ˆ>r#   ÚoutputsÚbatchÚ	batch_idxc                 óÌ   — |j         dz
  }|j         | j        k    rG|                      |¬¦  «        r3| j        €J ‚| j                             |¦  «         |j         | _        dS dS dS )aë  Called when a training batch ends.

        Updates the :class:`AveragedModel` parameters, if requested by ``self.should_update()``.

        Args:
            trainer: The current :class:`~pytorch_lightning.trainer.trainer.Trainer` instance.
            pl_module: The current :class:`~pytorch_lightning.core.LightningModule` instance.
            outputs: Outputs from the training batch.
            batch: The training batch.
            batch_idx: Index of the training batch.

        é   )r$   N)Úglobal_stepr   r(   r   Úupdate_parameters)r    r)   r+   r3   r4   r5   r$   s          r!   Úon_train_batch_endz"WeightAveraging.on_train_batch_end¡   s~   € ð$ Ô&¨Ñ*ˆØÔ $Ô":Ò:Ð:À×@RÒ@RÐ\dÐ@RÑ@eÔ@eÐ:ØÔ&Ð2Ð2Ð2ØÔ×1Ò1°)Ñ<Ô<Ð<Ø'.Ô':ˆDÔ$Ð$Ð$ð ;Ð:Ð:Ð:r#   c                 óÂ   — |j         | j        k    rL|                      |j         ¬¦  «        r3| j        €J ‚| j                             |¦  «         |j         | _        dS dS dS )aZ  Called when a training epoch ends.

        Updates the :class:`AveragedModel` parameters, if requested by ``self.should_update()``.

        Args:
            trainer: The current :class:`~pytorch_lightning.trainer.trainer.Trainer` instance.
            pl_module: The current :class:`~pytorch_lightning.core.LightningModule` instance.

        )r%   N)Úcurrent_epochr   r(   r   r9   ©r    r)   r+   s      r!   Úon_train_epoch_endz"WeightAveraging.on_train_epoch_end¹   st   € ð Ô! DÔ$=Ò=Ð=À4×CUÒCUÐ`gÔ`uÐCUÑCvÔCvÐ=ØÔ&Ð2Ð2Ð2ØÔ×1Ò1°)Ñ<Ô<Ð<Ø(/Ô(=ˆDÔ%Ð%Ð%ð >Ð=Ð=Ð=r#   c                 óB   — | j         €J ‚|                      |¦  «         dS )aD  Called when training ends.

        Transfers parameters from the :class:`AveragedModel` to the current model.

        Args:
            trainer: The current :class:`~pytorch_lightning.trainer.trainer.Trainer` instance.
            pl_module: The current :class:`~pytorch_lightning.core.LightningModule` instance.

        N)r   Ú_copy_average_to_currentr=   s      r!   Úon_train_endzWeightAveraging.on_train_endÉ   s-   € ð Ô"Ð.Ð.Ð.Ø×%Ò% iÑ0Ô0Ð0Ð0Ð0r#   c                 óB   — | j         �|                      |¦  «         dS dS )aV  Called when a validation epoch begins.

        Transfers parameter values from the :class:`AveragedModel` to the current model.

        Args:
            trainer: The current :class:`~pytorch_lightning.trainer.trainer.Trainer` instance.
            pl_module: The current :class:`~pytorch_lightning.core.LightningModule` instance.

        N©r   Ú_swap_modelsr=   s      r!   Úon_validation_epoch_startz)WeightAveraging.on_validation_epoch_start×   ó/   € ð ÔÐ*Ø×Ò˜iÑ(Ô(Ð(Ð(Ð(ð +Ð*r#   c                 óB   — | j         �|                      |¦  «         dS dS )aJ  Called when a validation epoch ends.

        Recovers the current model parameters from the :class:`AveragedModel`.

        Args:
            trainer: The current :class:`~pytorch_lightning.trainer.trainer.Trainer` instance.
            pl_module: The current :class:`~pytorch_lightning.core.LightningModule` instance.

        NrC   r=   s      r!   Úon_validation_epoch_endz'WeightAveraging.on_validation_epoch_endå   rF   r#   c                 ó   — d| j         iS )z­Called when saving a checkpoint.

        Creates a ``state_dict`` of the callback state.

        Returns:
            A dictionary containing the callback state.

        Úlatest_update_step©r   )r    s    r!   Ú
state_dictzWeightAveraging.state_dictó   s   € ð % dÔ&>Ð?Ð?r#   rL   c                 ó    — |d         | _         dS )zºCalled when loading a checkpoint.

        Reloads the callback state given a ``state_dict``.

        Args:
            state_dict: A dictionary containing the callback state.

        rJ   NrK   )r    rL   s     r!   Úload_state_dictzWeightAveraging.load_state_dictÿ   s   € ð $.Ð.BÔ#CˆÔ Ð Ð r#   Ú
checkpointc                 ó  — | j         €t          d¦  «         dS | j                              ¦   «         }|d         |d<   d„ |                     ¦   «         D ¦   «         |d<   d„ |                     ¦   «         D ¦   «         |d<   dS )a.  Called when saving a checkpoint.

        Moves the current model state to the key ``current_model_state``, and places the average model state in
        ``state_dict`` instead. Any other state variables of the ``AveragedModel`` will be saved in
        ``averaging_state``.

        Args:
            trainer: The current :class:`~pytorch_lightning.trainer.trainer.Trainer` instance.
            pl_module: The current :class:`~pytorch_lightning.core.LightningModule` instance.
            checkpoint: The checkpoint dictionary that will be saved.

        Na  You're using the WeightAveraging callback, but saving a checkpoint outside the 'fit' stage. The state of the WeightAveraging callback won't be saved in the checkpoint. If training has finished, the average model parameters will be saved to the state_dict in the checkpoint.rL   Úcurrent_model_statec                 óT   — i | ]%\  }}|                      d ¦  «        ¯|dd…         |“Œ&S )úmodule.é   N©Ú
startswith©Ú.0ÚnameÚvalues      r!   ú
<dictcomp>z6WeightAveraging.on_save_checkpoint.<locals>.<dictcomp>&  sK   € ð (ð (ð (Ù$/ D¨%ÐRV×RaÒRaÐbkÑRlÔRlð(Ø�Q�R�R”˜%ð(ð (ð (r#   c                 óD   — i | ]\  }}|                      d ¦  «        °||“ŒS ©rS   rU   rW   s      r!   r[   z6WeightAveraging.on_save_checkpoint.<locals>.<dictcomp>)  sC   € ð -ð -ð -Ù +  eÐRV×RaÒRaÐbkÑRlÔRlð-Ø�eð-ð -ð -r#   Úaveraging_state)r   r   rL   Úitems©r    r)   r+   rO   Úaverage_model_states        r!   Úon_save_checkpointz"WeightAveraging.on_save_checkpoint  s»   € ð  ÔÐ&Ýð^ñô ð ð ð ð #'Ô"5×"@Ò"@Ñ"BÔ"BÐØ0:¸<Ô0HˆJÐ,Ñ-ð(ð (Ø3F×3LÒ3LÑ3NÔ3Nð(ñ (ô (ˆJ�|Ñ$ð-ð -Ø/B×/HÒ/HÑ/JÔ/Jð-ñ -ô -ˆJÐ(Ñ)Ð)Ð)r#   c                 ó¶  — | j         €t          d¦  «         dS d|v ryd|v rut          d¦  «         d„ |d                              ¦   «         D ¦   «         }||d         z  }| j                              |¦  «         |                     |d         ¦  «         dS t          d¦  «         | j         j                             t          |d         ¦  «        d	¬
¦  «         dS )a´  Called when loading a model checkpoint.

        Loads the current model and the :class:`AveragedModel` parameters from the checkpoint.

        Args:
            trainer: The current :class:`~pytorch_lightning.trainer.trainer.Trainer` instance.
            pl_module: The current :class:`~pytorch_lightning.core.LightningModule` instance.
            checkpoint: The full checkpoint dictionary that got loaded by the Trainer.

        Na  You're using the WeightAveraging callback, but loading a checkpoint outside the 'fit' stage. The WeightAveraging state cannot be restored. If you're using the checkpoint for prediction or testing, you can ignore this warning. To disable the warning, remove the WeightAveraging callback.rQ   r^   zWFound current_model_state in the checkpoint. This will be used to initialize the model.c                 ó    — i | ]\  }}d |z   |“ŒS r]   r'   rW   s      r!   r[   z6WeightAveraging.on_load_checkpoint.<locals>.<dictcomp>C  s#   € Ð"oÐ"oÐ"o¹{¸tÀU 9¨tÑ#3°UÐ"oÐ"oÐ"or#   rL   z€The checkpoint was not created with WeightAveraging. Both the current and the average model will be initialized with state_dict.F)Ústrict)r   r   r   r_   rN   Úmoduler   r`   s        r!   Úon_load_checkpointz"WeightAveraging.on_load_checkpoint-  s  € ð ÔÐ&Ýðlñô ð ð ð ð
 $ zÐ1Ð1Ð8IÈZÐ8WÐ8WÝÐtÑuÔuÐuØ"oÐ"oÈjÐYeÔNf×NlÒNlÑNnÔNnÐ"oÑ"oÔ"oÐØ :Ð.?Ô#@Ñ@ÐØÔ×/Ò/Ð0CÑDÔDÐDð ×%Ò% jÐ1FÔ&GÑHÔHÐHÐHÐHåð/ñô ð ð ÔÔ&×6Ò6µxÀ
È<Ô@XÑ7YÔ7YÐbgÐ6ÑhÔhÐhÐhÐhr#   c                 óô  — | j         €J ‚t          j        | j         j                             ¦   «         | j         j                             ¦   «         ¦  «        }t          j        |                     ¦   «         |                     ¦   «         ¦  «        }t          ||¦  «        D ]W\  }}|j                             ¦   «         }|j         	                    |j        ¦  «         |j         	                    |¦  «         ŒXdS )zÆSwaps the parameter values of the current model and the :class:`AveragedModel`.

        Args:
            pl_module: The current :class:`~pytorch_lightning.core.LightningModule` instance.

        N)
r   Ú	itertoolsÚchainrf   Ú
parametersÚbuffersÚzipÚdataÚcloneÚcopy_)r    r+   Úaverage_paramsÚcurrent_paramsÚaverage_paramÚcurrent_paramÚtmps          r!   rD   zWeightAveraging._swap_modelsQ  sâ   € ð Ô"Ð.Ð.Ð.Ý"œ¨Ô)<Ô)C×)NÒ)NÑ)PÔ)PÐRVÔReÔRl×RtÒRtÑRvÔRvÑwÔwˆÝ"œ¨×)=Ò)=Ñ)?Ô)?À×ARÒARÑATÔATÑUÔUˆÝ,/°ÀÑ,OÔ,Oð 	*ð 	*Ñ(ˆM˜=ØÔ$×*Ò*Ñ,Ô,ˆCØÔ×$Ò$ ]Ô%7Ñ8Ô8Ð8ØÔ×$Ò$ SÑ)Ô)Ð)Ð)ð	*ð 	*r#   c                 óŽ  — | j         €J ‚t          j        | j         j                             ¦   «         | j         j                             ¦   «         ¦  «        }t          j        |                     ¦   «         |                     ¦   «         ¦  «        }t          ||¦  «        D ]$\  }}|j                             |j        ¦  «         Œ%dS )zÈCopies the parameter values from the :class:`AveragedModel` to the current model.

        Args:
            pl_module: The current :class:`~pytorch_lightning.core.LightningModule` instance.

        N)	r   ri   rj   rf   rk   rl   rm   rn   rp   )r    r+   rq   rr   rs   rt   s         r!   r@   z(WeightAveraging._copy_average_to_current`  s·   € ð Ô"Ð.Ð.Ð.Ý"œ¨Ô)<Ô)C×)NÒ)NÑ)PÔ)PÐRVÔReÔRl×RtÒRtÑRvÔRvÑwÔwˆÝ"œ¨×)=Ò)=Ñ)?Ô)?À×ARÒARÑATÔATÑUÔUˆÝ,/°ÀÑ,OÔ,Oð 	9ð 	9Ñ(ˆM˜=ØÔ×$Ò$ ]Ô%7Ñ8Ô8Ð8Ð8ð	9ð 	9r#   )NT©NN)r)   r*   r+   r,   r   N)r+   r,   r   N)Ú__name__Ú
__module__Ú__qualname__Ú__doc__r   r   r   r   r   ÚintÚboolr   r"   r(   r	   r2   r   r:   r>   rA   rE   rH   ÚdictrL   rN   rb   rg   rD   r@   r'   r#   r!   r   r   "   sú  € € € € € ð5ð 5ðr ;?Ø ð'ð 'à˜˜uœ|¨S°#Ð5Ô6Ô7ð'ð ð'ð ð	'ð
 
ð'ð 'ð 'ð 'ð2$ð $ h¨s¤mð $ÀxÐPSÄ}ð $Ð`dð $ð $ð $ð $ð$ ð˜\ð Ð6Jð ÐSVð Ð[_ð ð ð ñ „Xðð6 ð;Ø#ð;Ø0Dð;ØOZð;Øcfð;Øsvð;à	ð;ð ;ð ;ñ „Xð;ð. ð>ð >ð >ñ „Xð>ð ð1ð 1ð 1ñ „Xð1ð ð)ð )ð )ñ „Xð)ð ð)ð )ð )ñ „Xð)ð ð	@˜D  c œNð 	@ð 	@ð 	@ñ „Xð	@ð ð	D¨$¨s°C¨x¬.ð 	D¸Tð 	Dð 	Dð 	Dñ „Xð	Dð ðØ#ðØ0DðØRVÐWZÐ\_ÐW_ÔR`ðà	ðð ð ñ „XððB ð!iØ#ð!iØ0Dð!iØRVÐWZÐ\_ÐW_ÔR`ð!ià	ð!ið !ið !iñ „Xð!iðF*ð *ð *ð *ð9ð 9ð 9ð 9ð 9ð 9r#   r   c                   ó¼   ‡ — e Zd ZdZ	 	 	 	 	 	 ddeeej        ee	f                  de
ded	e	d
ee	         dee	         defˆ fd„Zddee	         dee	         de
fd„Zˆ xZS )ÚEMAWeightAveragingz;Exponential Moving Average (EMA) Weight Averaging callback.NTç+‡ÙÎ÷ï?r7   r   r   ÚdecayÚupdate_every_n_stepsÚupdate_starting_at_stepÚupdate_starting_at_epochr   c                 óŽ   •—  t          ¦   «         j        d||dœ|¤dt          |¬¦  «        i¤Ž || _        || _        || _        d S )N)r   r   Úavg_fn)r‚   r'   )Úsuperr"   r   rƒ   r„   r…   )	r    r   r   r‚   rƒ   r„   r…   r   Ú	__class__s	           €r!   r"   zEMAWeightAveraging.__init__q  sy   ø€ ð 	�‰ŒÔð 	
ØØ#ð	
ð 	
ð ð	
ð 	
õ "¨Ð.Ñ.Ô.ð		
ð 	
ð 	
ð 	
ð %9ˆÔ!Ø'>ˆÔ$Ø(@ˆÔ%Ð%Ð%r#   r$   r%   r   c                 ó¤   — |�3| j         du p
|| j         k    }| j        dk    o|| j        z  dk    }|r|rdS |�| j        duo
|| j        k    }|rdS dS )a   Decide when to update the model weights.

        Args:
            step_idx: The current step index.
            epoch_idx: The current epoch index.
        Returns:
            bool: True if the model weights should be updated, False otherwise.

        Nr   TF)r„   rƒ   r…   )r    r$   r%   Úmeets_step_requirementÚmeets_step_frequencyÚmeets_epoch_requirements         r!   r(   z EMAWeightAveraging.should_update†  s–   € ð Ðà%)Ô%AÀTÐ%IÐ%uÈXÐY]ÔYuÒMuÐ"Ø#'Ô#<¸qÒ#@Ð#nÀXÐPTÔPiÑEiÐmnÒEnÐ Ø%ð Ð*>ð Ø�tàÐ ð Ô-°TÐ9Ðh¸iÈ4ÔKhÒ>hð $ð 'ð Ø�tàˆur#   )NTr�   r7   NNrw   )rx   ry   rz   r{   r   r   r   r   r   r|   r}   Úfloatr   r"   r(   Ú__classcell__)r‰   s   @r!   r€   r€   n  s  ø€ € € € € ØEÐEð ;?Ø ØØ$%Ø15Ø26ðAð Aà˜˜uœ|¨S°#Ð5Ô6Ô7ðAð ðAð ð	Að
 "ðAð "*¨#¤ðAð #+¨3¤-ðAð ðAð Að Að Að Að Að*ð  h¨s¤mð ÀxÐPSÄ}ð Ð`dð ð ð ð ð ð ð ð r#   r€   )r{   ri   Úcopyr   Útypingr   r   r   r   Útorch.optim.swa_utilsr   r   Útyping_extensionsr	   Úpytorch_lightningÚplÚ$pytorch_lightning.callbacks.callbackr
   Ú)pytorch_lightning.utilities.model_helpersr   Ú%pytorch_lightning.utilities.rank_zeror   r   Ú!pytorch_lightning.utilities.typesr   r   r€   r'   r#   r!   ú<module>rš      sA  ððð ð
 Ð Ð Ð Ø Ð Ð Ð Ð Ð Ø 'Ð 'Ð 'Ð 'Ð 'Ð 'Ð 'Ð 'Ð 'Ð 'à €€€Ø ?Ð ?Ð ?Ð ?Ð ?Ð ?Ð ?Ð ?Ø &Ð &Ð &Ð &Ð &Ð &à Ð Ð Ð Ø 9Ð 9Ð 9Ð 9Ð 9Ð 9Ø CÐ CÐ CÐ CÐ CÐ CØ PÐ PÐ PÐ PÐ PÐ PÐ PÐ PØ 9Ð 9Ð 9Ð 9Ð 9Ð 9ðI9ð I9ð I9ð I9ð I9�hñ I9ô I9ð I9ðX
1ð 1ð 1ð 1ð 1˜ñ 1ô 1ð 1ð 1ð 1r#   