
    qi2X                     j   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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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jZ                        Z. G d dejZ                        Z/dej`                  de1dej`                  fdZ2	 d9dejZ                  dej`                  dej`                  dej`                  d ej`                  dz  d!e3d"e3d#e!e#   fd$Z4d% Z5d:d&Z6 ee6       G d' d(ejZ                               Z7 ed)       G d* d+ejZ                               Z8 G d, d-e      Z9e$ G d. d/e             Z:e$ G d0 d1e:             Z;e$ G d2 d3e:e             Z< G d4 d5ee:      Z= G d6 d7ee:      Z>g d8Z?y);    )Callable)OptionalN   )ACT2FN)CacheDynamicCache)GenerationMixin)use_kernel_forward_from_hubuse_kernelized_func)create_causal_mask) GenericForSequenceClassificationGenericForTokenClassification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   )	GlmConfigc                   V     e Zd Z fdZdej
                  dej
                  fdZ xZS )GlmMLPc                 *   t         |           || _        t        j                  |j
                  d|j                  z  d      | _        t        j                  |j                  |j
                  d      | _        t        |j                     | _        y )N   Fbias)super__init__confignnLinearhidden_sizeintermediate_sizegate_up_proj	down_projr   
hidden_actactivation_fnselfr'   	__class__s     V/opt/pipecat/venv/lib/python3.12/site-packages/transformers/models/glm/modeling_glm.pyr&   zGlmMLP.__init__0   sp    IIf&8&8!f>V>V:V]bc6#;#;V=O=OV[\#F$5$56    hidden_statesreturnc                     | j                  |      }|j                  dd      \  }}|| j                  |      z  }| j                  |      S )Nr"   dim)r,   chunkr/   r-   )r1   r5   	up_statesgates       r3   forwardzGlmMLP.forward8   sL    %%m4	#//!/4i 2 24 88	~~i((r4   )__name__
__module____qualname__r&   torchFloatTensorr>   __classcell__r2   s   @r3   r    r    /   s'    7)U%6%6 )5;L;L )r4   r    c                        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 )GlmRotaryEmbeddinginv_freqNr'   c                    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defaultrH   F)
persistentoriginal_inv_freq)r%   r&   max_position_embeddingsmax_seq_len_cachedoriginal_max_seq_lenr'   rope_parametersrJ   compute_default_rope_parametersr   attention_scalingregister_bufferclone)r1   r'   devicerope_init_fnrH   r2   s        r3   r&   zGlmRotaryEmbedding.__init__D   s    "("@"@$*$B$B!44[A!%!E!E>>Y&.t~~>L+7V+L($(ZeD0(..2BuUr4   rV   ztorch.deviceseq_lenr6   ztorch.Tensorc                 n   | j                   d   }| j                   j                  dd      }t        | dd      xs | j                  | j                  z  }t        ||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partial_rotary_factorg      ?head_dimNr   r"   dtype)rV   r^   )rQ   getgetattrr*   num_attention_headsintrB   arangeint64tofloat)	r'   rV   rX   baser[   r\   r:   attention_factorrH   s	            r3   rR   z2GlmRotaryEmbedding.compute_default_rope_parametersT   s    & %%l3 & 6 6 : :;RTW X6:t4h8J8JfNhNh8h(223 U\\!S!5;;?BB&X]XcXcBdgjjk
 )))r4   c                 N   | j                   d d d d f   j                         j                  |j                  d   dd      j	                  |j
                        }|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                  ||fd	      }|j                         | j                  z  }|j                         | j                  z  }	d d d        j	                  |j                   
      	j	                  |j                   
      fS # 1 sw Y   AxY w)Nr   r8   r   mpscpuF)device_typeenabledr"   r9   r]   )rH   rf   expandshapere   rV   
isinstancetypestrr   	transposerB   catcosrS   sinr^   )
r1   xposition_idsinv_freq_expandedposition_ids_expandedrl   freqsembru   rv   s
             r3   r>   zGlmRotaryEmbedding.forwardt   sR    !MM$4-8>>@GGHZHZ[\H]_acdehhijiqiqr ,QaZ 8 > > @'1!((--'E!((--[`J`ahhmmfkUC 	5&,,.1F1L1L1NNYYZ[]^_E))UEN3C'')d444C'')d444C		5 vvAGGv$cff177f&;;;	5 	5s   BFF$N)NNN)r?   r@   rA   rB   Tensor__annotations__r   r&   staticmethodr   rb   tuplerf   rR   no_gradr   r>   rD   rE   s   @r3   rG   rG   A   s    llVy V  #'+/"*D *(* t* 
~u$	%	* *> U]]_<  <r4   rG   r5   n_repr6   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)ro   rn   reshape)r5   r   batchnum_key_value_headsslenr\   s         r3   	repeat_kvr      so    
 2?1D1D.Ehz!!Qa"23::5BUW\^bdlmM  (;e(CT8TTr4   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   r8   )r:   r^   )ptrainingr   )r   num_key_value_groupsrB   matmulrs   r(   
functionalsoftmaxfloat32re   r^   r   r   
contiguous)r   r   r   r   r   r   r   r   
key_statesvalue_statesattn_weightsattn_outputs               r3   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$$r4   c                 |    | ddddf   }| ddddf   }t        j                  | |fd      j                  d      S )	z*Rotates half the hidden dims of the input..r   Nr"   r   r8   r9   )rB   stackflatten)rw   x1x2s      r3   rotate_halfr      sJ    	
319B	
319B;;Ryb)11"55r4   c                    |j                  |      }|j                  |      }|dd|j                  d   dz  f   j                  dd      }|dd|j                  d   dz  f   j                  dd      }|j                  d   }| dd|f   | d|df   }}|dd|f   |d|df   }	}||z  t        |      |z  z   }
||z  t        |      |z  z   }t	        j
                  |
|gd      }
t	        j
                  ||	gd      }|
|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.
    .Nr8   r"   r9   )	unsqueezero   repeat_interleaver   rB   rt   )qkru   rv   unsqueeze_dim
rotary_dimq_rotq_passk_rotk_passq_embedk_embeds               r3   apply_rotary_pos_embr      sD   $ --
&C
--
&C c'SYYr]a'''
(
:
:1"
:
EC
c'SYYr]a'''
(
:
:1"
:
EC 2Jc;J;&'3
+;)<6Ec;J;&'3
+;)<6E s{{51C78Gs{{51C78G ii&)r2Gii&)r2GGr4   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z  dej                  dz  d	e
dz  d
ej                  dz  dee   de	ej                  ej                  f   fdZ xZS )GlmAttentionz=Multi-headed attention from 'Attention Is All You Need' paperNr'   	layer_idxc                 P   t         |           || _        || _        t	        |d|j
                  |j                  z        | _        |j                  |j                  z  | _	        | j                  dz  | _
        |j                  | _        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
                  d      | _        y )Nr\   g      Tr#   F)r%   r&   r'   r   r`   r*   ra   r\   r   r   r   attention_dropout	is_causalr(   r)   attention_biasq_projk_projv_projo_projr1   r'   r   r2   s      r3   r&   zGlmAttention.__init__   sD   "
F4F4F&JdJd4de$*$>$>&B\B\$\!}}d*!'!9!9ii : :T]] JQWQfQf
 ii : :T]] JQWQfQf
 ii : :T]] JQWQfQf
 ii : :T]] JFL^L^ejkr4   r5   position_embeddingsr   past_key_valuescache_positionr   r6   c                    |j                   d d }g |d| j                  }| j                  |      j                  |      j	                  dd      }	| j                  |      j                  |      j	                  dd      }
| j                  |      j                  |      j	                  dd      }|\  }}t        |	|
||      \  }	}
|'|||d}|j                  |
|| j                  |      \  }
}t        j                  | j                  j                  t              } || |	|
||f| j                  sdn| j                   | j"                  d|\  }} |j$                  g |d j'                         }| j)                  |      }||fS )Nr8   r   r"   )rv   ru   r           )r   r   )ro   r\   r   viewrs   r   r   r   updater   r   get_interfacer'   _attn_implementationr   r   r   r   r   r   r   )r1   r5   r   r   r   r   r   input_shapehidden_shapequery_statesr   r   ru   rv   cache_kwargsattention_interfacer   r   s                     r3   r>   zGlmAttention.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#7jRUWZ#[ j&#&snUL'6'='=j,X\XfXfht'u$J(?(M(MKK,,.E)
 %8	%
  $}}C$2H2HLL	%
 	%
!\ *k));;;;FFHkk+.L((r4   r}   )NNNN)r?   r@   rA   __doc__r   rb   r&   rB   r~   r   r   
LongTensorr   r   r>   rD   rE   s   @r3   r   r      s    Gly lS4Z l0 IM.2(,26))||)) #5<<#=>E)) t+	))
 )) ((4/)) +,)) 
u||U\\)	*))r4   r   RMSNormc                   h     e Zd Zddeddf fdZdej                  dej                  fdZd Z xZ	S )	
GlmRMSNormepsr6   Nc                     t         |           t        j                  t	        j
                  |            | _        || _        y)z9
        GlmRMSNorm is equivalent to T5LayerNorm
        N)r%   r&   r(   	ParameterrB   onesweightvariance_epsilon)r1   r*   r   r2   s      r3   r&   zGlmRMSNorm.__init__  s1     	ll5::k#:; #r4   r5   c                 "   |j                   }|j                  t        j                        }|j	                  d      j                  dd      }|t        j                  || j                  z         z  }| j                  |j                  |      z  S )Nr"   r8   T)keepdim)	r^   re   rB   r   powmeanrsqrtr   r   )r1   r5   input_dtypevariances       r3   r>   zGlmRMSNorm.forward'  sy    #))%((7 $$Q',,R,>%Ht?T?T4T(UU{{]--k:::r4   c                 ^    t        | j                  j                         d| j                   S )Nz, eps=)r   r   ro   r   )r1   s    r3   
extra_reprzGlmRMSNorm.extra_repr.  s*    ))*+6$2G2G1HIIr4   )gư>)
r?   r@   rA   rf   r&   rB   r~   r>   r   rD   rE   s   @r3   r   r     s7    $ $$ $;U\\ ;ell ;Jr4   r   c                   "    e Zd Zdedef fdZ	 	 	 	 	 	 ddej                  dej                  dz  dej                  dz  de	dz  d	e
dz  d
ej                  dz  deej                  ej                  f   dz  dee   dej                  fdZ xZS )GlmDecoderLayerr'   r   c                     t         |           |j                  | _        t        ||      | _        t        |      | _        t        |j                  |j                        | _	        t        |j                  |j                        | _
        y )N)r'   r   r   )r%   r&   r*   r   	self_attnr    mlpr   rms_norm_epsinput_layernormpost_attention_layernormr   s      r3   r&   zGlmDecoderLayer.__init__3  sk    !--%VyI&>)&*<*<&BUBUV(263E3E6K^K^(_%r4   Nr5   r   rx   r   	use_cacher   r   r   r6   c                     |}	| j                  |      } | j                  d|||||||d|\  }}
|	|z   }|}	| j                  |      }| j                  |      }|	|z   }|S )N)r5   r   rx   r   r   r   r    )r   r   r   r   )r1   r5   r   rx   r   r   r   r   r   residual_s              r3   r>   zGlmDecoderLayer.forward=  s     !,,];)4>> 	
')%+) 3	
 	
q !=0 !55mD/ =0r4   )NNNFNN)r?   r@   rA   r   rb   r&   rB   r~   r   r   boolr   r   r   r>   rD   rE   s   @r3   r   r   2  s    `y `S ` /304(,!&26HL|| t+ &&-	
  $; ((4/ #5<<#=>E +, 
r4   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)GlmPreTrainedModelr'   modelTr   r   )r5   
attentionsN)r?   r@   rA   r   r   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   r4   r3   r   r   _  sQ    &*#*+#4"5N!"&("r4   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j                  dz  d
edz  dee   defd                     Z xZS )GlmModelr'   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   r'   F)r%   r&   pad_token_idpadding_idx
vocab_sizer(   	Embeddingr*   embed_tokens
ModuleListrangenum_hidden_layersr   layersr   r   normrG   
rotary_embgradient_checkpointing	post_initr   s      r3   r&   zGlmModel.__init__t  s     !.. ++LL):):F<N<NPTP`P`ammAFvG_G_A`aI_VY/a
 v11v7J7JK	,F;&+# 	 bs   DN	input_idsr   rx   r   inputs_embedsr   r   r   r6   c                 D   |d u |d uz  rt        d      || j                  |      }|r|t        | j                        }|E||j	                         nd}	t        j                  |j                  d   |j                        |	z   }||j                  d      }t        | j                  |||||      }
|}| j                  ||      }| j                  d | j                  j                   D ]  } ||f|
|||||d|} | j                  |      }t        ||	      S )
Nz:You must specify exactly one of input_ids or inputs_embedsr  r   r   )rV   )r'   r  r   r   r   rx   )rx   )r   r   rx   r   r   r   )last_hidden_stater   )
ValueErrorr  r   r'   get_seq_lengthrB   rc   ro   rV   r   r   r  r  r  r  r   )r1   r  r   rx   r   r  r   r   r   past_seen_tokenscausal_maskr5   r   decoder_layers                 r3   r>   zGlmModel.forward  s]    -t";<YZZ *.*;*;I*FM0*$++>O!CRC^==?de]003M<P<PQTdd  )33A6L(;;'))+%
 &"oom,oW![[)H4;;+H+HI 
	M)	*$7) /#-	 	M
	 		-0&++
 	
r4   )NNNNNNN)r?   r@   rA   r   r&   r   r   r   rB   r   r~   r   rC   r   r   r   r   r>   rD   rE   s   @r3   r  r  r  s    y     .2.204(,2626!%9
##d*9
 t+9
 &&-	9

 9
 ((4/9
 ((4/9
 $;9
 +,9
 
!9
    9
r4   r  c                   b    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	j                  dz  dee	j                  z  dee   defd              Z xZS )GlmForCausalLMzlm_head.weightzmodel.embed_tokens.weightlm_headcolwise_gather_outputr5   logitsc                     t         |   |       t        |      | _        |j                  | _        t        j                  |j                  |j                  d      | _        | j                          y )NFr#   )
r%   r&   r  r   r  r(   r)   r*   r  r  r0   s     r3   r&   zGlmForCausalLM.__init__  sU     f%
 ++yy!3!3V5F5FUS 	r4   Nr  r   rx   r   r  labelsr   r   logits_to_keepr   r6   c
                 z    | j                   d|||||||d|
}|j                  }t        |	t              rt	        |	 d      n|	}| j                  |dd|ddf         }d}|* | j                  d||| j                  j                  d|
}t        |||j                  |j                  |j                        S )a  
        Example:

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

        >>> model = GlmForCausalLM.from_pretrained("meta-glm/Glm-2-7b-hf")
        >>> tokenizer = AutoTokenizer.from_pretrained("meta-glm/Glm-2-7b-hf")

        >>> 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."
        ```)r  r   rx   r   r  r   r   N)r  r   r  )lossr  r   r5   r   r   )r   r  rp   rb   slicer  loss_functionr'   r  r   r   r5   r   )r1   r  r   rx   r   r  r   r   r   r!  r   outputsr5   slice_indicesr  r#  s                   r3   r>   zGlmForCausalLM.forward  s    @ ,64:: 	,
)%+')	,
 	,
  118B>SV8W~ot4]kmA}a,?@A%4%%pVFt{{OeOepiopD%#33!//))
 	
r4   )	NNNNNNNNr   )r?   r@   rA   _tied_weights_keys_tp_plan_pp_planr&   r   r   rB   r   r~   r   rC   r   rb   r   r   r   r>   rD   rE   s   @r3   r  r    s/   *,GH23H_-z:;H  .2.204(,26*.!%26-.8
##d*8
 t+8
 &&-	8

 8
 ((4/8
   4'8
 $;8
 ((4/8
 ell*8
 +,8
 
 8
  8
r4   r  c                       e Zd Zy)GlmForSequenceClassificationNr?   r@   rA   r   r4   r3   r,  r,        r4   r,  c                       e Zd Zy)GlmForTokenClassificationNr-  r   r4   r3   r0  r0    r.  r4   r0  )r   r  r  r,  r0  )r   )r   )@collections.abcr   typingr   rB   torch.nnr(   activationsr   cache_utilsr   r   
generationr	   integrationsr
   r   masking_utilsr   modeling_layersr   r   r   modeling_outputsr   r   modeling_rope_utilsr   r   modeling_utilsr   r   processing_utilsr   utilsr   r   r   utils.genericr   r   utils.output_capturingr   configuration_glmr   Moduler    rG   r~   rb   r   rf   r   r   r   r   r   r   r   r  r  r,  r0  __all__r   r4   r3   <module>rD     s  * %    ! . ) L / 
 P K F & I I G 5 ()RYY )$@< @<F	UU\\ 	U# 	U%,, 	U& %II%<<% 
% <<	%
 LL4'% % % '(%26%P )*A)299 A) +A)H Y'J J (J(*0 *Z   $ M
! M
 M
` H
' H
 H
V	#CEW 		 =?Q 	r4   