
    qiy`                        d dl 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 ddlmZ dd	lmZ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mZ ddlmZ ddl m!Z!m"Z"m#Z# ddl$m%Z%m&Z& ddl'm(Z( ddl)m*Z*  G d dejV                        Z, G d dejV                        Z-dej\                  de/dej\                  fdZ0	 d4dejV                  dej\                  dej\                  dej\                  d ej\                  dz  d!e1d"e1d#ee!   fd$Z2d% Z3d5d&Z4 ee4       G d' d(ejV                               Z5 G d) d*ejV                        Z6 G d+ d,e      Z7e" G d- d.e             Z8e" G d/ d0e8             Z9e" G d1 d2e8e             Z:g d3Z;y)6    )Callable)OptionalN   )ACT2FN)CacheDynamicCache)GenerationMixin)use_kernelized_func)create_causal_mask!create_sliding_window_causal_mask)GradientCheckpointingLayer)BaseModelOutputWithPastCausalLMOutputWithPast)ROPE_INIT_FUNCTIONSdynamic_rope_update)ALL_ATTENTION_FUNCTIONSPreTrainedModel)Unpack)TransformersKwargsauto_docstringcan_return_tuple)maybe_autocastmerge_with_config_defaults)capture_outputs   )Cohere2Configc                        e Zd ZU ej                  ed<   ddef fdZe	 	 	 ddedz  de	d   de
dz  ded	ef   fd
       Z ej                         ed               Z xZS )Cohere2RotaryEmbeddinginv_freqNconfigc                    t         |           |j                  | _        |j                  | _        || _        | j
                  j                  d   | _        | j                  }| j                  dk7  rt        | j                     } || j
                  |      \  }| _
        | j                  d|d       | j                  d|j                         d       y )N	rope_typedefaultr   F)
persistentoriginal_inv_freq)super__init__max_position_embeddingsmax_seq_len_cachedoriginal_max_seq_lenr    rope_parametersr"   compute_default_rope_parametersr   attention_scalingregister_bufferclone)selfr    devicerope_init_fnr   	__class__s        ^/opt/pipecat/venv/lib/python3.12/site-packages/transformers/models/cohere2/modeling_cohere2.pyr'   zCohere2RotaryEmbedding.__init__.   s    "("@"@$*$B$B!44[A!%!E!E>>Y&.t~~>L+7V+L($(ZeD0(..2BuU    r1   ztorch.deviceseq_lenreturnztorch.Tensorc                    | j                   d   }t        | dd      xs | j                  | j                  z  }d}d|t	        j
                  d|dt        j                        j                  |t        j                        |z  z  z  }||fS )	a  
        Computes the inverse frequencies according to the original RoPE implementation
        Args:
            config ([`~transformers.PreTrainedConfig`]):
                The model configuration.
            device (`torch.device`):
                The device to use for initialization of the inverse frequencies.
            seq_len (`int`, *optional*):
                The current sequence length. Unused for this type of RoPE.
        Returns:
            Tuple of (`torch.Tensor`, `float`), containing the inverse frequencies for the RoPE embeddings and the
            post-processing scaling factor applied to the computed cos/sin (unused in this type of RoPE).
        
rope_thetahead_dimNg      ?r      dtype)r1   r=   )	r+   getattrhidden_sizenum_attention_headstorcharangeint64tofloat)r    r1   r6   basedimattention_factorr   s          r4   r,   z6Cohere2RotaryEmbedding.compute_default_rope_parameters>   s    & %%l3fj$/c63E3EIcIc3c U\\!S!5;;?BB&X]XcXcBdgjjk
 )))r5   c                    | j                   d d d d f   j                         j                  |j                  d   dd      }|d d d d d f   j                         }t	        |j
                  j                  t              r/|j
                  j                  dk7  r|j
                  j                  nd}t        |d      5  |j                         |j                         z  j                  dd      }t        j                  |dd	      }|j                         | j                  z  }|j                         | j                  z  }	d d d        j                  |j                   
      	j                  |j                   
      fS # 1 sw Y   AxY w)Nr   r   mpscpuF)device_typeenabledr;   rG   r<   )r   rE   expandshape
isinstancer1   typestrr   	transposerA   repeat_interleavecosr-   sinrD   r=   )
r0   xposition_idsinv_freq_expandedposition_ids_expandedrM   freqsembrW   rX   s
             r4   forwardzCohere2RotaryEmbedding.forward\   s@    !MM$4-8>>@GGHZHZ[\H]_acde ,QaZ 8 > > @'1!((--'E!((--[`J`ahhmmfkUC 	5&,,.1F1L1L1NNYYZ[]^_E))%;C'')d444C'')d444C		5 vvAGGv$cff177f&;;;	5 	5s   =BFF
N)NNN)__name__
__module____qualname__rA   Tensor__annotations__r   r'   staticmethodr   inttuplerE   r,   no_gradr   r_   __classcell__r3   s   @r4   r   r   +   s    llV} V  '++/"*$*(* t* 
~u$	%	* *: U]]_<  <r5   r   c                   &     e Zd Zd fd	Zd Z xZS )Cohere2LayerNormc                     t         |           t        j                  t	        j
                  |            | _        || _        y)zcThe hidden size can be a tuple or an int. The tuple is used for QKNorm to normalize across head_dimN)r&   r'   nn	ParameterrA   onesweightvariance_epsilon)r0   r?   epsbiasr3   s       r4   r'   zCohere2LayerNorm.__init__m   s/    ll5::k#:; #r5   c                    |j                   }|j                  t        j                        }|j	                  dd      }||z
  j                  d      j	                  dd      }||z
  t        j                  || j                  z         z  }| j                  j                  t        j                        |z  }|j                  |      S )NrJ   T)keepdimr;   )	r=   rD   rA   float32meanpowrsqrtrs   rr   )r0   hidden_statesinput_dtypery   variances        r4   r_   zCohere2LayerNorm.forwards   s    #))%((7!!"d!3!D(--a055b$5G&-XH]H]=]1^^u}}5E,,r5   )Ngh㈵>Fra   rb   rc   r'   r_   rj   rk   s   @r4   rm   rm   l   s    $-r5   rm   r|   n_repr7   c                     | j                   \  }}}}|dk(  r| S | dddddddddf   j                  |||||      } | j                  |||z  ||      S )z
    This is the equivalent of torch.repeat_interleave(x, dim=1, repeats=n_rep). The hidden states go from (batch,
    num_key_value_heads, seqlen, head_dim) to (batch, num_attention_heads, seqlen, head_dim)
    r   N)rQ   rP   reshape)r|   r   batchnum_key_value_headsslenr:   s         r4   	repeat_kvr   }   so    
 2?1D1D.Ehz!!Qa"23::5BUW\^bdlmM  (;e(CT8TTr5   modulequerykeyvalueattention_maskscalingdropoutkwargsc                    t        || j                        }t        || j                        }	t        j                  ||j	                  dd            |z  }
||
|z   }
t
        j                  j                  |
dt        j                        j                  |j                        }
t
        j                  j                  |
|| j                        }
t        j                  |
|	      }|j	                  dd      j                         }||
fS )Nr;   r   rJ   )rG   r=   )ptrainingr   )r   num_key_value_groupsrA   matmulrU   ro   
functionalsoftmaxrx   rD   r=   r   r   
contiguous)r   r   r   r   r   r   r   r   
key_statesvalue_statesattn_weightsattn_outputs               r4   eager_attention_forwardr      s     3 ; ;<JUF$?$?@L<<z';';Aq'ABWLL!#n4==((2U]](SVVW\WbWbcL==((6??([L,,|\:K''1-88:K$$r5   c                     | dd d df   }| ddd df   }t        j                  | |gd      j                  d      }|S )N.r;   r   rJ   rO   )rA   stackflatten)rY   x1x2rot_xs       r4   rotate_halfr      sL    	
3!8B	
319BKK"b	r*2226ELr5   c                 6   | j                   }| j                         } |j                         }|j                  |      }|j                  |      }| |z  t        |       |z  z   }||z  t        |      |z  z   }|j	                  |      |j	                  |      fS )a  Applies Rotary Position Embedding to the query and key tensors.

    Args:
        q (`torch.Tensor`): The query tensor.
        k (`torch.Tensor`): The key tensor.
        cos (`torch.Tensor`): The cosine part of the rotary embedding.
        sin (`torch.Tensor`): The sine part of the rotary embedding.
        unsqueeze_dim (`int`, *optional*, defaults to 1):
            The 'unsqueeze_dim' argument specifies the dimension along which to unsqueeze cos[position_ids] and
            sin[position_ids] so that they can be properly broadcasted to the dimensions of q and k. For example, note
            that cos[position_ids] and sin[position_ids] have the shape [batch_size, seq_len, head_dim]. Then, if q and
            k have the shape [batch_size, heads, seq_len, head_dim], then setting unsqueeze_dim=1 makes
            cos[position_ids] and sin[position_ids] broadcastable to the shapes of q and k. Similarly, if q and k have
            the shape [batch_size, seq_len, heads, head_dim], then set unsqueeze_dim=2.
    Returns:
        `tuple(torch.Tensor)` comprising of the query and key tensors rotated using the Rotary Position Embedding.
    r<   )r=   rE   	unsqueezer   rD   )qkrW   rX   unsqueeze_dimr=   q_embedk_embeds           r4   apply_rotary_pos_embr      s    $ GGE		A		A
--
&C
--
&C3w;q>C/0G3w;q>C/0G::E:"GJJUJ$;;;r5   c                   >    e Zd ZdZddededz  f fdZ	 	 ddej                  de	ej                  ej                  f   dej                  dz  d	e
dz  d
ej                  dz  dee   de	ej                  ej                  dz  e	ej                     dz  f   fdZ xZS )Cohere2Attentionz=Multi-headed attention from 'Attention Is All You Need' paperNr    	layer_idxc                    t         |           || _        || _        t	        |d|j
                  |j                  z        | _        |j                  |j                  z  | _	        | j                  dz  | _
        |j                  | _        d| _        t        |d      r|j                  |   nd }|dk(  r|j                  nd | _        t!        j"                  |j
                  |j                  | j                  z  |j$                        | _        t!        j"                  |j
                  |j                  | j                  z  |j$                        | _        t!        j"                  |j
                  |j                  | j                  z  |j$                        | _        t!        j"                  |j                  | j                  z  |j
                  |j$                        | _        y )Nr:   g      Tlayer_typessliding_attentionru   )r&   r'   r    r   r>   r?   r@   r:   r   r   r   attention_dropout	is_causalhasattrr   sliding_windowro   Linearattention_biasq_projk_projv_projo_proj)r0   r    r   
layer_typer3   s       r4   r'   zCohere2Attention.__init__   s   "
F4F4F&JdJd4de$*$>$>&B\B\$\!}}d*!'!9!96=fm6TV''	2Z^
7AEX7Xf33^bii : :T]] JQWQfQf
 ii : :T]] JQWQfQf
 ii : :T]] JQWQfQf
 ii&&68J8JQWQfQf
r5   r|   position_embeddingsr   past_key_valuescache_positionr   r7   c                 F   |j                   d d }g |d| j                  }| j                  |      j                  |      j	                  dd      }	| j                  |      j                  |      j	                  dd      }
| j                  |      j                  |      j	                  dd      }|\  }}| j                  t        |	|
||      \  }	}
|'|||d}|j                  |
|| j                  |      \  }
}t        j                  | j                  j                  t              } || |	|
||f| j                   sdn| j"                  | j$                  | j                  d|\  }} |j&                  g |d j)                         }| j+                  |      }||fS )NrJ   r   r;   )rX   rW   r           )r   r   r   )rQ   r:   r   viewrU   r   r   r   r   updater   r   get_interfacer    _attn_implementationr   r   r   r   r   r   r   )r0   r|   r   r   r   r   r   input_shapehidden_shapequery_statesr   r   rW   rX   cache_kwargsattention_interfacer   r   s                     r4   r_   zCohere2Attention.forward   s    $))#2.88b8$--8{{=166|DNNqRST[[/44\BLLQPQR
{{=166|DNNqRST&S*';L*VY[^'_$L*&#&snUL'6'='=j,X\XfXfht'u$J(?(M(MKK,,.E)
 %8
%
  $}}C$2H2HLL..
%
 
%
!\ *k));;;;FFHkk+.L((r5   r`   )NN)ra   rb   rc   __doc__r   rg   r'   rA   rd   rh   r   
LongTensorr   r   r_   rj   rk   s   @r4   r   r      s    G
} 
t 
< )-26*)||*) #5<<#=>*) t+	*)
 *) ((4/*) +,*) 
u||U\\D0%2E2LL	M*)r5   r   c                   $     e Zd Z fdZd Z xZS )
Cohere2MLPc                    t         |           || _        |j                  | _        |j                  | _        t        j                  | j                  | j                  d      | _        t        j                  | j                  | j                  d      | _        t        j                  | j                  | j                  d      | _	        t        |j                     | _        y NFr   )r&   r'   r    r?   intermediate_sizero   r   	gate_projup_proj	down_projr   
hidden_actact_fnr0   r    r3   s     r4   r'   zCohere2MLP.__init__  s    !--!'!9!94#3#3T5K5KRWXyy!1!143I3IPUV4#9#94;K;KRWXV../r5   c                     | j                  | j                  | j                  |            | j                  |      z        }|S r`   )r   r   r   r   )r0   rY   r   s      r4   r_   zCohere2MLP.forward  s6    NN4;;t~~a/@#ADLLQRO#ST	r5   r   rk   s   @r4   r   r     s    0r5   r   c                   D    e Zd Zdedef fdZ	 	 	 	 	 ddej                  deej                  ej                  f   dz  dej                  dz  de	dz  d	e
dz  d
ej                  dz  dee   deej                  eej                  ej                  f   dz  f   fdZ xZS )Cohere2DecoderLayerr    r   c                     t         |           |j                  | _        t        ||      | _        t        |      | _        t        |j                  |j                        | _	        |j                  |   | _        y )N)r    r   r?   rt   )r&   r'   r?   r   	self_attnr   mlprm   layer_norm_epsinput_layernormr   attention_typer0   r    r   r3   s      r4   r'   zCohere2DecoderLayer.__init__!  se    !--)9Mf%/V=O=OV\VkVkl$00;r5   Nr|   r   r   r   	use_cacher   r   r7   c           
          |}| j                  |      } | j                  d||||||d|\  }	}
| j                  |      }||	z   |z   }|S )ar  
        Args:
            hidden_states (`torch.FloatTensor`): input to the layer of shape `(batch, seq_len, embed_dim)`
            attention_mask (`torch.FloatTensor`, *optional*):
                attention mask of size `(batch_size, sequence_length)` if flash attention is used or `(batch_size, 1,
                query_sequence_length, key_sequence_length)` if default attention is used.
            past_key_values (`Cache`, *optional*): cached past key and value projection states
            output_attentions (`bool`, *optional*):
                Whether or not to return the attentions tensors of all attention layers. See `attentions` under
                returned tensors for more detail.
            use_cache (`bool`, *optional*):
                If set to `True`, `past_key_values` key value states are returned and can be used to speed up decoding
                (see `past_key_values`).
            cache_position (`torch.LongTensor` of shape `(sequence_length)`, *optional*):
                Indices depicting the position of the input sequence tokens in the sequence
            position_embeddings (`tuple[torch.FloatTensor, torch.FloatTensor]`, *optional*):
                Tuple containing the cosine and sine positional embeddings of shape `(batch_size, seq_len, head_dim)`,
                with `head_dim` being the embedding dimension of each attention head.
        )r|   r   r   r   r   r    )r   r   r   )r0   r|   r   r   r   r   r   r   residualhidden_states_attention_hidden_states_mlps               r4   r_   zCohere2DecoderLayer.forward)  sx    : !,,];%3T^^ &
' 3)+)&
 &
" !HH]3 #::=NNr5   )NNNFN)ra   rb   rc   r   rg   r'   rA   rd   rh   r   boolr   r   r   FloatTensorr_   rj   rk   s   @r4   r   r      s    <} < < IM.2(,!&26+||+ #5<<#=>E+ t+	+
 + $;+ ((4/+ +,+ 
u  %(9(95;L;L(L"MPT"TT	U+r5   r   c                   J    e Zd ZU eed<   dZdZdgZdgZdZ	dZ
dZdZdZeedZy)Cohere2PreTrainedModelr    modelTr   r   )r|   
attentionsN)ra   rb   rc   r   re   base_model_prefixsupports_gradient_checkpointing_no_split_modules_skip_keys_device_placement_supports_flash_attn_supports_sdpa_supports_flex_attn_can_compile_fullgraph_supports_attention_backendr   r   _can_record_outputsr   r5   r4   r   r   W  sQ    &*#./#4"5N!"&,&r5   r   c                       e Zd Zdef fdZeee	 	 	 	 	 	 	 ddej                  dz  dej                  dz  dej                  dz  dedz  dej                  dz  d	edz  d
ej                  dz  dee   defd                     Z xZS )Cohere2Modelr    c           	      
   t         |   |       |j                  | _        |j                  | _        t        j                  |j                  |j                  | j                        | _        t        j                  t        |j                        D cg c]  }t        ||       c}      | _        t        |j                  |j                        | _        t#        |      | _        d| _        | j)                          y c c}w )Nr   F)r&   r'   pad_token_idpadding_idx
vocab_sizero   	Embeddingr?   embed_tokens
ModuleListrangenum_hidden_layersr   layersrm   r   normr   
rotary_embgradient_checkpointing	post_initr   s      r4   r'   zCohere2Model.__init__l  s     !.. ++LL):):F<N<NPTP`P`ammEJ6KcKcEde	 3e
 %&2D2D6K`K`a	08&+# 	 fs   D N	input_idsr   rZ   r   inputs_embedsr   r   r   r7   c                 r   |d u |d uz  rt        d      || j                  |      }|r|t        | j                        }|F||j	                         nd}	t        j                  |	|	|j                  d   z   |j                        }||j                  d      }t        |x}
t              s*| j                  |||||d}t        d
i |t        d
i |d}
|}| j                  ||      }| j                  D ]  } ||f|
|j                      |||||d|}! | j#                  |      }t%        ||	      S )Nz:You must specify exactly one of input_ids or inputs_embeds)r    r   r   )r1   )r    r  r   r   r   rZ   )full_attentionr   )r   r   r   r   r   rZ   )last_hidden_stater   r   )
ValueErrorr  r   r    get_seq_lengthrA   rB   rQ   r1   r   rR   dictr   r   r  r  r   r  r   )r0   r  r   rZ   r   r  r   r   r   past_seen_tokenscausal_mask_mappingmask_kwargsr|   r   decoder_layers                  r4   r_   zCohere2Model.forward|  s    -t";<YZZ  --i8M0*$++>O!CRC^==?de"\\ "2]5H5H5K"KTaThThN )33A6L?-F++!."0"0#2 ,K #5"C{"C%F%U%U#
 &"oom\J![[ 
	M)	2=3O3OP$7 /#-)	 	M
	 		-0&++
 	
r5   )NNNNNNN)ra   rb   rc   r   r'   r   r   r   rA   r   rd   r   r   r   r   r   r   r_   rj   rk   s   @r4   r   r   j  s    }     .2.204(,26!%26=
##d*=
 t+=
 &&-	=

 =
 ((4/=
 $;=
 ((4/=
 +,=
 
!=
    =
r5   r   c                   z    e Zd ZddiZddiZddgdgfiZ fdZee	 	 	 	 	 	 	 	 	 	 	 dd	e	j                  dz  d
e	j                  dz  de	j                  dz  dedz  de	j                  dz  de	j                  dz  dedz  dedz  dedz  de	j                  dz  dee	j                  z  dee   defd              Z xZS )Cohere2ForCausalLMzlm_head.weightzmodel.embed_tokens.weightlm_headcolwise_gather_outputr|   logitsc                 ,   t         |   |       t        |      | _        |j                  | _        t        j                  |j                  |j                  d      | _        |j                  | _	        |j                  | _
        | j                          y r   )r&   r'   r   r   r   ro   r   r?   r  logit_scaletie_word_embeddingsr
  r   s     r4   r'   zCohere2ForCausalLM.__init__  sq     !&)
 ++yy!3!3V5F5FUS!--#)#=#=  	r5   Nr  r   rZ   r   r  labelsr   output_attentionsoutput_hidden_statesr   logits_to_keepr   r7   c                    ||n| j                   j                  }|	|	n| j                   j                  }	 | j                  d||||||||	|
d	|}|j                  }t        |t              rt        | d      n|}| j                  |dd|ddf         }|| j                  z  }d}|* | j                  d||| j                   j                  d|}t        |||j                  |j                  |j                        S )a~  
        labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*):
            Labels for computing the masked language modeling loss. Indices should either be in `[0, ...,
            config.vocab_size]` or -100 (see `input_ids` docstring). Tokens with indices set to `-100` are ignored
            (masked), the loss is only computed for the tokens with labels in `[0, ..., config.vocab_size]`.

        Example:

        ```python
        >> from transformers import AutoTokenizer, Cohere2ForCausalLM

        >> model = Cohere2ForCausalLM.from_pretrained("Cohere2ForAI/c4ai-command-r-v01")
        >> tokenizer = AutoTokenizer.from_pretrained("Cohere2ForAI/c4ai-command-r-v01")

        >> prompt = "Hey, are you conscious? Can you talk to me?"
        >> inputs = tokenizer(prompt, return_tensors="pt")

        >> # Generate
        >> generate_ids = model.generate(inputs.input_ids, max_length=30)
        >> tokenizer.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0]
        "Hey, are you conscious? Can you talk to me?\nI'm not conscious, but I can talk to you."
        ```N)	r  r   rZ   r   r  r   r   r!  r   )r  r  r   )lossr  r   r|   r   r   )r    r   r!  r   r  rR   rg   slicer  r  loss_functionr   r   r   r|   r   )r0   r  r   rZ   r   r  r  r   r   r!  r   r"  r   outputsr|   slice_indicesr  r$  s                     r4   r_   zCohere2ForCausalLM.forward  s+   N 2C1N-TXT_T_TqTq$8$D $++JjJj 	
 ,64:: ,
)%+'/!5),
 ,
  118B>SV8W~ot4]kmA}a,?@A$***%4%%pVFt{{OeOepiopD%#33!//))
 	
r5   )NNNNNNNNNNr   )ra   rb   rc   _tied_weights_keys_tp_plan_pp_planr'   r   r   rA   r   rd   r   r   r   rg   r   r   r   r_   rj   rk   s   @r4   r  r    s`   *,GH23H_-z:;H	  .2.204(,26*.!%)-,026-.H
##d*H
 t+H
 &&-	H

 H
 ((4/H
   4'H
 $;H
  $;H
 #TkH
 ((4/H
 ell*H
 +,H
 
 H
  H
r5   r  )r  r   r   )r   )r   )<collections.abcr   typingr   rA   torch.nnro   activationsr   cache_utilsr   r   
generationr	   integrationsr
   masking_utilsr   r   modeling_layersr   modeling_outputsr   r   modeling_rope_utilsr   r   modeling_utilsr   r   processing_utilsr   utilsr   r   r   utils.genericr   r   utils.output_capturingr   configuration_cohere2r   Moduler   rm   rd   rg   r   rE   r   r   r   r   r   r   r   r   r  __all__r   r5   r4   <module>r?     s  * %    ! . ) / R 9 O K F & I I G 5 0><RYY ><B-ryy -"	UU\\ 	U# 	U%,, 	U& %II%<<% 
% <<	%
 LL4'% % % '(%2<8 )*F)ryy F) +F)R  44 4n _  $ Q
) Q
 Q
h Z
/ Z
 Z
z Kr5   