
     `ih                        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	 ddl
mZ ddlmZ dd	lmZmZ dd
lmZmZ ddlmZ ddlmZmZmZmZ ddlmZ ddlmZ ddlm Z m!Z! ddl"m#Z#m$Z$  G d de	j%                  Z& G d de	j%                  Z' G d de          Z( G d de           Z) G d de	j%                  Z* G d de          Z+e G d d e                      Z, ed!"           G d# d$e,                      Z-e G d% d&e                      Z. ed'"           G d( d)e,                      Z/g d*Z0dS )+zPyTorch Parakeet model.    N)	dataclass)CallableOptionalUnion)nn   )ACT2FN)GradientCheckpointingLayer)BaseModelOutputCausalLMOutput)ALL_ATTENTION_FUNCTIONSPreTrainedModel)Unpack)ModelOutputTransformersKwargsauto_docstringcan_return_tuple)check_model_inputs   )%FastSpeech2ConformerConvolutionModule)LlamaAttentioneager_attention_forward   )ParakeetCTCConfigParakeetEncoderConfigc                        e Zd ZU dZej        ed<   ddef fdZ ej	                    dej        fd            Z
 xZS )	$ParakeetEncoderRelPositionalEncodingz*Relative positional encoding for Parakeet.inv_freqNconfigc                 <   t                                                       |j        | _        d}d|t          j        d|j        dt          j                                      |t          j                  |j        z  z  z  }| 	                    d|d	           d S )
Ng     @      ?r   r   dtype)devicer#   r   F)
persistent)
super__init__max_position_embeddingstorcharangehidden_sizeint64tofloatregister_buffer)selfr   r$   baser   	__class__s        /home/jaya/work/projects/VOICE-AGENT/VIET/agent-env/lib/python3.11/site-packages/transformers/models/parakeet/modular_parakeet.pyr'   z-ParakeetEncoderRelPositionalEncoding.__init__)   s    '-'E$Q 2AU[IIILLTZbgbmLnn$%
 	ZeDDDDD    hidden_statesc                    |j         d         }|| j        k    rt          d| d| j         d          t          j        |dz
  | d|j                  }| j        d d d d f                                                             |j         d         dd          	                    |j                  }|d d d d f                                         }t          |j        j        t                    r|j        j        dk    r|j        j        nd	}t          j        |d
          5  |                                |                                z                      dd          }|                                }|                                }	t          j        ||	gd          }
 |
j        g |
j         d d         dR  }
d d d            n# 1 swxY w Y   |
	                    |j                  S )Nr   zSequence Length: z= has to be less or equal than config.max_position_embeddings .r$   r   mpscpuF)device_typeenabledr   dimr"   )shaper(   
ValueErrorr)   r*   r$   r   r.   expandr-   
isinstancetypestrautocast	transposesincosstackreshaper#   )r0   r5   
seq_lengthposition_idsinv_freq_expandedposition_ids_expandedr<   freqsrI   rJ   	pos_embeds              r3   forwardz,ParakeetEncoderRelPositionalEncoding.forward7   s[   "(+
444RJ R R262NR R R  
 |JNZKML`aaaM$4-(..00778KA8NPRTUVVYYZgZnoo 	 !-T4] ; A A C C -.3S99>K>R>W[`>`>`  %% 	
 ^UCCC 	E 	E&,,..1F1L1L1N1NNYYZ[]^__E))++C))++CS#JB777I)	)D9?3B3+?DDDDI	E 	E 	E 	E 	E 	E 	E 	E 	E 	E 	E 	E 	E 	E 	E ||-"5|666s   #BG		GGN)__name__
__module____qualname____doc__r)   Tensor__annotations__r   r'   no_gradrS   __classcell__r2   s   @r3   r   r   $   s         44lE E4 E E E E E E U]__7U\ 7 7 7 _7 7 7 7 7r4   r   c                   *     e Zd Zdef fdZd Z xZS )ParakeetEncoderFeedForwardr   c                 :   t                                                       t          j        |j        |j        |j                  | _        t          |j	                 | _
        t          j        |j        |j        |j                  | _        |j        | _        d S )Nbias)r&   r'   r   Linearr+   intermediate_sizeattention_biaslinear1r	   
hidden_act
activationlinear2activation_dropoutr0   r   r2   s     r3   r'   z#ParakeetEncoderFeedForward.__init__W   s}    y!3V5MTZTijjj !23y!96;MTZTijjj"(";r4   c                     |                      |                     |                    }t          j                            || j        | j                  }|                     |          }|S )Nptraining)rh   rf   r   
functionaldropoutrj   ro   ri   )r0   r5   s     r3   rS   z"ParakeetEncoderFeedForward.forward^   sY    ](C(CDD--mt?Vaean-oo]33r4   )rU   rV   rW   r   r'   rS   r\   r]   s   @r3   r_   r_   V   sT        <4 < < < < < <      r4   r_   c                   &     e Zd Zddef fdZ xZS ) ParakeetEncoderConvolutionModuleNr   c                 L    t                                          ||           d S rT   )r&   r'   )r0   r   module_configr2   s      r3   r'   z)ParakeetEncoderConvolutionModule.__init__f   s#    /////r4   rT   )rU   rV   rW   r   r'   r\   r]   s   @r3   rs   rs   e   sJ        0 04 0 0 0 0 0 0 0 0 0 0r4   rs   c                        e Zd ZdZdedef fdZ	 ddej        de	ej                 de	ej                 d	e
e         d
eej        ej        f         f
dZd Z xZS )ParakeetEncoderAttentionztMulti-head attention with relative positional encoding. See section 3.3 of https://huggingface.co/papers/1901.02860.r   	layer_idxc                    t                                          ||           d| _        t          j        |j        |j        | j        z  d          | _        t          j	        t          j        |j        | j                            | _        t          j	        t          j        |j        | j                            | _        d S )N)rx   Fra   )r&   r'   	is_causalr   rc   r+   num_attention_headshead_dimrelative_k_proj	Parameterr)   zerosbias_ubias_vr0   r   rx   r2   s      r3   r'   z!ParakeetEncoderAttention.__init__m   s    9555!y);V=WZ^Zg=gnstttl5;v/I4=#Y#YZZl5;v/I4=#Y#YZZr4   Nr5   position_embeddingsattention_maskkwargsreturnc           
         |j         d d         }|\  }}||d| j        f}|                     |                              |                              dd          }	|                     |                              |                              dd          }
|                     |                              |                              dd          }t          }| j        j	        dk    rt          | j        j	                 }|	| j                            d| j        j        d| j                  z   }|	| j                            d| j        j        d| j                  z   }|                     |          }|                    |d| j        j        | j                  }||                    dddd          z  }|                     |          }|dd |f         }|| j        z  }|5|                    |                                t)          d                    } || f||
||| j        sd	n| j        | j        d
|\  }} |j        g |dR                                  }|                     |          }||fS )Nr8   r   r   eagerr   r   .z-inf        )querykeyvaluer   rq   scaling)rA   r|   q_projviewrH   k_projv_projr   r   _attn_implementationr   r   r{   r   r}   permute
_rel_shiftr   masked_fill_logical_notr.   ro   attention_dropoutrL   
contiguouso_proj)r0   r5   r   r   r   input_shape
batch_sizerM   hidden_shapequery_states
key_statesvalue_statesattention_interfacequery_states_with_bias_uquery_states_with_bias_vrelative_key_states	matrix_bdattn_outputattn_weightss                      r3   rS   z ParakeetEncoderAttention.forwardw   s    $)#2#.!,
J"JDMB{{=1166|DDNNqRSTT[[//44\BBLLQPQRR
{{=1166|DDNNqRSTT(?;+w66"9$+:Z"[#/$+2B2Bt{.4=3
 3
 $
  $0$+2B2Bt{.4=3
 3
 $
  #223FGG166z2t{Gfhlhuvv -/B/J/J1aQRTU/V/VV	OOI..	c;J;./	,	% "..~/I/I/K/KUSY]][[I %8$7	%
*$#}HCC$2HL	%
 	%
 	%
 	%
!\ *k);;;;;;FFHHkk+..L((r4   c                     |j         \  }}}}t          j                            |d          }|                    ||d|          }|ddddddf                             ||||          }|S )ztRelative position shift for Shaw et al. style attention. See appendix B of https://huggingface.co/papers/1901.02860.)r   r   )padr8   Nr   )rA   r   rp   r   r   )r0   attention_scoresr   	num_headsquery_lengthposition_lengths         r3   r   z#ParakeetEncoderAttention._rel_shift   s    ?O?U<
I|_=,,-=6,JJ+00YLYY+AAAqqq!""H5:::yR^`oppr4   rT   )rU   rV   rW   rX   r   intr'   r)   rY   r   r   r   tuplerS   r   r\   r]   s   @r3   rw   rw   j   s        ~~[4 [ [ [ [ [ [ [ 26	7) 7)|7) &el37) !.	7)
 +,7) 
u|U\)	*7) 7) 7) 7)r             r4   rw   c                   n     e Zd Zdef fdZdej        dej        fdZ	d
dej        dej        fd	Z
 xZS ) ParakeetEncoderSubsamplingConv2Dr   c                 6   t                                                       |j        | _        |j        | _        |j        | _        | j        dz
  dz  | _        t          t          j        |j                            | _        t          j                    | _        | j                            t          j        d| j        | j        | j        | j                             | j                            t          j                               t)          | j        dz
            D ]}| j                            t          j        | j        | j        | j        | j        | j        | j                             | j                            t          j        | j        | j        d                     | j                            t          j                               |j        | j        | j        z  z  }t          j        |j        |z  |j        d          | _        d S )Nr   r   )kernel_sizestridepadding)r   r   r   groupsr   Tra   )r&   r'   subsampling_conv_kernel_sizer   subsampling_conv_strider   subsampling_conv_channelschannelsr   r   mathlog2subsampling_factor
num_layersr   
ModuleListlayersappendConv2dReLUrangenum_mel_binsrc   r+   linear)r0   r   i
out_lengthr2   s       r3   r'   z)ParakeetEncoderSubsamplingConv2D.__init__   s   !>48(1,2di(ABBCC mooIaD4DT[bfbnooo	
 	
 	
 	2799%%%t*++ 	* 	*AK	MM $ 0; L=  	 	 	 KrySTUUUVVVKrwyy))))(T[$/-IJ
i @: MvOahlmmmr4   input_lengths
conv_layerc                     t          |d          rK|j        dk    r@|j        }|j        d         }|j        d         }||d         z   |d         z   |z
  |z  dz   }|S |S )Nr   )r   r   r   r   )hasattrr   r   r   )r0   r   r   r   r   r   output_lengthss          r3   _get_output_lengthz3ParakeetEncoderSubsamplingConv2D._get_output_length   sw    :x(( 	"Z->&-H-H (G$03K&q)F+gaj871:ESX^^abbN!!r4   Ninput_featuresr   c                 .   |                     d          }||                    d          nd }| j        D ]} ||          }t          |t          j                  ra|_|                     ||          }|j        d         }t          j	        ||j
                  |d d d f         k     }||d d d d d d f         z  }|                    dd                              |j        d         |j        d         d          }|                     |          }|S )Nr   r8   r   r9   r   )	unsqueezesumr   rD   r   r   r   rA   r)   r*   r$   rH   rL   r   )r0   r   r   r5   current_lengthslayercurrent_seq_lengthchannel_masks           r3   rS   z(ParakeetEncoderSubsamplingConv2D.forward   s<   &00334B4N.,,R000TX[ 
	@ 
	@E!E-00M %++ @0J"&"9"9/5"Q"Q%2%8%;"L!3N<QRRRUdefefefhlelUmm  aaaqqq$.>!??%//155==m>QRS>TVcVijkVlnpqqM22r4   rT   )rU   rV   rW   r   r'   r)   rY   r   r   r   rS   r\   r]   s   @r3   r   r      s        !n4 !n !n !n !n !n !nF	 	") 	 	 	 	 el EL        r4   r   c                        e Zd Zddedee         f fdZ	 	 ddej        deej                 deej                 de	e
         d	ej        f
d
Z xZS )ParakeetEncoderBlockNr   rx   c                 $   t                                                       d| _        t          |          | _        t          ||          | _        t          |          | _        t          |          | _	        t          j        |j                  | _        t          j        |j                  | _        t          j        |j                  | _        t          j        |j                  | _        t          j        |j                  | _        d S NF)r&   r'   gradient_checkpointingr_   feed_forward1rw   	self_attnrs   convfeed_forward2r   	LayerNormr+   norm_feed_forward1norm_self_att	norm_convnorm_feed_forward2norm_outr   s      r3   r'   zParakeetEncoderBlock.__init__   s    &+#7??1&)DD4V<<	7??"$,v/A"B"B\&*<==f&899"$,v/A"B"BV%788r4   r5   r   r   r   r   c                    |}|                      |                     |                    }|d|z  z   }|                     |          } | j        d|||d|\  }}||z   }|                     |                     |          |          }	||	z   }|                     |                     |                    }
|d|
z  z   }|                     |          }|S )Ng      ?)r5   r   r   )r    )	r   r   r   r   r   r   r   r   r   )r0   r5   r   r   r   residualnormalized_hidden_statesr   _conv_output
ff2_outputs              r3   rS   zParakeetEncoderBlock.forward  s     !**4+B+B=+Q+QRR 3#66#'#5#5m#D#D ' 
2) 3
 
 	
 
Q &3ii} = =ni]]%3''(?(?(N(NOO
%j(88m44r4   rT   NN)rU   rV   rW   r   r   r   r'   r)   rY   r   r   rS   r\   r]   s   @r3   r   r      s        9 94 9# 9 9 9 9 9 9$ 266:	 | !. &el3	
 +, 
       r4   r   c                        e Zd ZU eed<   dZdZdZdgZdZ	dZ
dZdZdZdZeedZ fdZd	ej        fd
Zddej        dee         fdZ xZS )ParakeetPreTrainedModelr   modelr   Tr   F)r5   
attentionsc                    t                                          |           t          | j        d          r| j        j        }n(t          | j                                        dd          }t          |t                    rD|j	        j
                            d|           |j        j
                            d|           d S d S )Ninitializer_rangeg{Gz?r   )meanstd)r&   _init_weightsr   r   r   getattrget_text_configrD   rw   r   datanormal_r   )r0   moduler   r2   s      r3   r   z%ParakeetPreTrainedModel._init_weightsB  s    f%%%4; 344 	T+/CC $+55779LdSSCf677 	:M&&CS&999M&&CS&99999	: 	:r4   r   c                    t          | j        t                    r| j        j        n| j        }|j        }|j        }t          t          j        |j	                            }|dz
  dz  dz  }||z
  }|}t          |          D ]O}	t          j        |                    t          j                  |z   |          dz   }t          j        |          }P|                    t          j                  S )Nr   r   r"   r!   )rD   r   r   encoder_configr   r   r   r   r   r   r   r)   divr-   r.   floor)
r0   r   r   r   r   r   all_paddingsadd_padlengthsr   s
             r3   _get_subsampling_output_lengthz6ParakeetPreTrainedModel._get_subsampling_output_lengthP  s    7A$+O`7a7ar33gkgr$A7>#DEEFF
#aA-1,z"" 	+ 	+Ai


 = = GPPSVVGk'**GGzz	z***r4   Nr   target_lengthc                     |                      |                    d                    }||n|                                }t          j        ||j                  |dddf         k     }|S )z
        Convert the input attention mask to its subsampled form. `target_length` sets the desired output length, useful
        when the attention mask length differs from `sum(-1).max()` (i.e., when the longest sequence in the batch is padded)
        r8   Nr9   )r  r   maxr)   r*   r$   )r0   r   r  r   
max_lengths        r3   _get_output_attention_maskz2ParakeetPreTrainedModel._get_output_attention_maska  su    
 <<^=O=OPR=S=STT&3&?]]^EWEWEYEY
j9NOOOR`abababdhahRiir4   rT   )rU   rV   rW   r   rZ   base_model_prefixmain_input_namesupports_gradient_checkpointing_no_split_modules_supports_flat_attention_mask_supports_sdpa_supports_flex_attn_supports_flash_attn_can_compile_fullgraph_supports_attention_backendr   rw   _can_record_outputsr   r)   rY   r  r   r   r  r\   r]   s   @r3   r   r   -  s         &O&*#/0$(!N !!"&-. 
: : : : :+EL + + + +"	 	 	V^_bVc 	 	 	 	 	 	 	 	r4   r   z{
    The Parakeet Encoder model, based on the [Fast Conformer architecture](https://huggingface.co/papers/2305.05084).
    )custom_introc                        e Zd ZU eed<   dZdef fdZeee		 d
de
j        dee
j                 dee         defd	                                    Z xZS )ParakeetEncoderr   encoderc                    t                                                     | _        d| _        j        | _        j        | _        j        | _        j        rt          j	        j
                  nd| _        t                    | _        t                    | _        t!          j        fdt%          j                  D                       | _        |                                  d S )NFr!   c                 0    g | ]}t          |          S r   )r   ).0rx   r   s     r3   
<listcomp>z,ParakeetEncoder.__init__.<locals>.<listcomp>  s$    fff!&)44fffr4   )r&   r'   r   r   rq   dropout_positions	layerdropscale_inputr   sqrtr+   input_scaler   subsamplingr   encode_positionsr   r   r   num_hidden_layersr   	post_initrk   s    `r3   r'   zParakeetEncoder.__init__v  s       &+#~!'!9)<B<NW49V%7888TW;FCC DV L LmffffeFLdFeFefff
 
 	r4   Nr   r   r   r   c                    |                      ||          }|| j        z  }|                     |          }t          j                            || j        | j                  }t          j                            || j        | j                  }||                     ||j	        d                   }|
                    d                              d|j	        d         d          }||                    dd          z  }|
                    d          }| j        D ]:}d}| j        r!t          j        g           }|| j        k     rd}|s ||f||d	|};t#          |
          S )a  
        Example:

        ```python
        >>> from transformers import AutoProcessor, ParakeetEncoder
        >>> from datasets import load_dataset, Audio

        >>> model_id = "nvidia/parakeet-ctc-1.1b"
        >>> processor = AutoProcessor.from_pretrained(model_id)
        >>> encoder = ParakeetEncoder.from_pretrained(model_id)

        >>> ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
        >>> ds = ds.cast_column("audio", Audio(sampling_rate=processor.feature_extractor.sampling_rate))

        >>> inputs = processor(ds[0]["audio"]["array"])
        >>> encoder_outputs = encoder(**inputs)

        >>> print(encoder_outputs.last_hidden_state.shape)
        ```
        rm   Nr   r  r8   r   FT)r   r   )last_hidden_state)r!  r   r"  r   rp   rq   ro   r  r  rA   r   rC   rH   r   r)   randr  r   )	r0   r   r   r   r5   r   encoder_layerto_dropdropout_probabilitys	            r3   rS   zParakeetEncoder.forward  s   < ((HH%(88"33MBB--mt|VZVc-dd m334#9DM 4 
 
 %!<<^[h[nop[q<rrN+55a88??MDWXYDZ\^__N+n.F.Fq!.L.LLN+55a88N![ 	 	MG} #&+jnn#&77"G  -!!#1(;! ! 	! ! ????r4   rT   )rU   rV   rW   r   rZ   r	  r'   r   r   r   r)   rY   r   r   r   r   rS   r\   r]   s   @r3   r  r  m  s          "!!!!4      &  26:@ :@:@ !.:@ +,	:@
 
:@ :@ :@   ^:@ :@ :@ :@ :@r4   r  c                       e Zd ZU dZej        ed<   dZee	ej
                          ed<   dZee	e	ej
                                   ed<   dZee	e	ej
                                   ed<   dS )ParakeetGenerateOutputal  
    Outputs of Parakeet models.

    Args:
        sequences (`torch.LongTensor` of shape `(batch_size, sequence_length)`):
            The generated sequences. The second dimension (sequence_length) is either equal to `max_length` or shorter
            if all batches finished early due to the `eos_token_id`.
        logits (`tuple(torch.FloatTensor)` *optional*, returned when `output_logits=True`):
            Unprocessed prediction scores of the language modeling head (scores for each vocabulary token before SoftMax)
            at each generation step. Tuple of `torch.FloatTensor` with up to `max_new_tokens` elements (one element for
            each generated token), with each tensor of shape `(batch_size, config.vocab_size)`.
        attentions (`tuple(tuple(torch.FloatTensor))`, *optional*, returned when `output_attentions=True`):
            Tuple (one element for each generated token) of tuples (one element for each layer of the decoder) of
            `torch.FloatTensor` of shape `(batch_size, num_heads, generated_length, sequence_length)`.
        hidden_states (`tuple(tuple(torch.FloatTensor))`, *optional*, returned when `output_hidden_states=True`):
            Tuple (one element for each generated token) of tuples (one element for each layer of the decoder) of
            `torch.FloatTensor` of shape `(batch_size, generated_length, hidden_size)`.
    	sequencesNlogitsr   r5   )rU   rV   rW   rX   r)   
LongTensorrZ   r/  r   r   FloatTensorr   r5   r   r4   r3   r-  r-    s          & 15FHU5,-.555<@JuU%6789@@@?CM8E%(9":;<CCCCCr4   r-  zS
    Parakeet Encoder with a Connectionist Temporal Classification (CTC) head.
    c                   F    e Zd ZU eed<   def fdZee	 	 ddej	        de
ej	                 de
ej	                 dee         def
d	                        Z ej                    	 	 ddej	        de
ej	                 dedee         deeej        f         f
d            Z xZS )ParakeetForCTCr   c                     t                                          |           t          |j                  | _        t          j        |j        j        |j        d          | _	        | 
                                 d S )Nr   r   )r&   r'   r  r   r  r   Conv1dr+   
vocab_sizectc_headr$  rk   s     r3   r'   zParakeetForCTC.__init__  se       &v'<==	&"7"CVEVdefffr4   Nr   r   labelsr   r   c           
      |    | j         d||d|}|j        }|                     |                    dd                                        dd          }d}|G||nt	          j        |t          j                  }|                     |                    d                    }	|| j	        j
        k    }
|
                    d          }|                    |
          }t          j                            |dt          j                                      dd          }t          j        j                            d	
          5  t          j                            |||	|| j	        j
        | j	        j        | j	        j                  }ddd           n# 1 swxY w Y   t-          |||j        |j                  S )a  
        Example:

        ```python
        >>> from transformers import AutoProcessor, ParakeetForCTC
        >>> from datasets import load_dataset, Audio

        >>> model_id = "nvidia/parakeet-ctc-1.1b"
        >>> processor = AutoProcessor.from_pretrained(model_id)
        >>> model = ParakeetForCTC.from_pretrained(model_id)

        >>> ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
        >>> ds = ds.cast_column("audio", Audio(sampling_rate=processor.feature_extractor.sampling_rate))

        >>> inputs = processor(ds[0]["audio"]["array"], text=ds[0]["text"])
        >>> outputs = model(**inputs)

        >>> print(outputs.loss)
        ```r   r   r   r   Nr"   r8   )r?   r#   r   F)r=   )blank	reductionzero_infinity)lossr/  r5   r   r   )r  r'  r7  rH   r)   	ones_likelongr  r   r   pad_token_idmasked_selectr   rp   log_softmaxfloat32backendscudnnflagsctc_lossctc_loss_reductionctc_zero_infinityr   r5   r   )r0   r   r   r8  r   encoder_outputsr5   r/  r>  r   labels_masktarget_lengthsflattened_targets	log_probss                 r3   rS   zParakeetForCTC.forward  s   : '$, 
))
 
 
 
 (9}66q!<<==GG1MM #1"<%/R`hmhrBsBsBs  !??@R@RSU@V@VWWM !DK$<<K(__R00N & 4 4[ A A 11&b1VV``abdeffI%++E+:: 	 	}--%!"+2"k<"&+"? .  	 	 	 	 	 	 	 	 	 	 	 	 	 	 	 )7&1	
 
 
 	
s   AFFFFreturn_dict_in_generatec                    d|d<    | j         d
||d|}|j                            d          }|2|                     ||j        d                   }| j        j        || <   |r"t          ||j        |j        |j	        	          S |S )a3  
        Example:

        ```python
        >>> from transformers import AutoProcessor, ParakeetForCTC
        >>> from datasets import load_dataset, Audio

        >>> model_id = "nvidia/parakeet-ctc-1.1b"
        >>> processor = AutoProcessor.from_pretrained(model_id)
        >>> model = ParakeetForCTC.from_pretrained(model_id)

        >>> ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
        >>> ds = ds.cast_column("audio", Audio(sampling_rate=processor.feature_extractor.sampling_rate))

        >>> inputs = processor(ds[0]["audio"]["array"], text=ds[0]["text"])
        >>> predicted_ids = model.generate(**inputs)
        >>> transcription = processor.batch_decode(predicted_ids, skip_special_tokens=True)

        >>> print(transcription)
        ```
        Treturn_dictr:  r8   r>   Nr   r&  )r.  r/  r   r5   r   )
rS   r/  argmaxr  rA   r   rA  r-  r   r5   )r0   r   r   rP  r   outputsr.  s          r3   generatezParakeetForCTC.generate=  s    : !%}".$, #
))#
 #
 #
 #
 N))b)11	 %!<<^[d[jkl[m<nnN)-)AI~o&" 	)#~"-%3	    r4   r   r   )rU   rV   rW   r   rZ   r'   r   r   r)   rY   r   r   r   r   rS   r[   boolr   r-  r0  rU  r\   r]   s   @r3   r3  r3    s`         0        26)-	E
 E
E
 !.E
 &	E

 +,E
 
E
 E
 E
  ^E
N U]__ 26(-	3 33 !.3 "&	3
 +,3 
%u'77	83 3 3 _3 3 3 3 3r4   r3  )r3  r  r   )1rX   r   dataclassesr   typingr   r   r   r)   r   activationsr	   modeling_layersr
   modeling_outputsr   r   modeling_utilsr   r   processing_utilsr   utilsr   r   r   r   utils.genericr   4fastspeech2_conformer.modeling_fastspeech2_conformerr   llama.modeling_llamar   r   configuration_parakeetr   r   Moduler   r_   rs   rw   r   r   r   r  r-  r3  __all__r   r4   r3   <module>re     s      ! ! ! ! ! ! , , , , , , , , , ,        ! ! ! ! ! ! 9 9 9 9 9 9 ? ? ? ? ? ? ? ? F F F F F F F F & & & & & & V V V V V V V V V V V V / / / / / / h h h h h h J J J J J J J J L L L L L L L L/7 /7 /7 /7 /729 /7 /7 /7d       0 0 0 0 0'L 0 0 0
L  L  L  L  L ~ L  L  L ^B B B B Bry B B BJ, , , , ,5 , , ,^ < < < < <o < < <~   
T@ T@ T@ T@ T@- T@ T@ 
T@n D D D D D[ D D D4   
H H H H H, H H 
HV K
J
Jr4   