o
    	۷iS                     @   s*  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
mZ ddlmZ ddl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& ddl'm(Z( ddl)m*Z* G dd dej+Z,dej-de.dej-fddZ/	d=dej+dej-dej-dej-deej- d e0d!e0d"e e" fd#d$Z1d%d& Z2d>d'd(Z3G d)d* d*ej+Z4ed+G d,d- d-ej+Z5G d.d/ d/ej+Z6G d0d1 d1eZ7e#G d2d3 d3eZ8e#G d4d5 d5e8Z9e#G d6d7 d7e8eZ:G d8d9 d9ee8Z;G d:d; d;ee8Z<g d<Z=dS )?    )CallableOptionalUnionN   )ACT2FN)CacheDynamicCache)GenerationMixin)use_kernel_forward_from_hub)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)deprecate_kwarg)check_model_inputs   )	GlmConfigc                       s2   e Zd Z fddZdejdejfddZ  ZS )GlmMLPc                    sP   t    || _tj|jd|j dd| _tj|j|jdd| _t	|j
 | _d S )N   Fbias)super__init__confignnLinearhidden_sizeintermediate_sizegate_up_proj	down_projr   
hidden_actactivation_fnselfr#   	__class__ Z/home/ubuntu/vllm_env/lib/python3.10/site-packages/transformers/models/glm/modeling_glm.pyr"   0   s
   
zGlmMLP.__init__hidden_statesreturnc                 C   s4   |  |}|jddd\}}|| | }| |S )Nr   dim)r(   chunkr+   r)   )r-   r2   	up_statesgater0   r0   r1   forward8   s   

zGlmMLP.forward)__name__
__module____qualname__r"   torchFloatTensorr:   __classcell__r0   r0   r.   r1   r   /   s    r   r2   n_repr3   c                 C   s^   | j \}}}}|dkr| S | dddddddddf |||||} | ||| ||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)shapeexpandreshape)r2   rA   batchnum_key_value_headsslenhead_dimr0   r0   r1   	repeat_kvA   s
   0rI           modulequerykeyvalueattention_maskscalingdropoutkwargsc                 K   s   t || j}t || j}	t||dd| }
|d ur3|d d d d d d d |jd f }|
| }
tjj|
dtj	d
|j}
tjj|
|| jd}
t|
|	}|dd }||
fS )Nr   r   r4   )r6   dtype)ptrainingr   )rI   num_key_value_groupsr>   matmul	transposerB   r$   
functionalsoftmaxfloat32torT   rQ   rV   
contiguous)rK   rL   rM   rN   rO   rP   rQ   rR   
key_statesvalue_statesattn_weightscausal_maskattn_outputr0   r0   r1   eager_attention_forwardM   s   
&rd   c                 C   s>   | ddddf }| ddddf }t j| |fdddS )	z*Rotates half the hidden dims of the input..r   Nr   r   r4   r5   rS   )r>   stackflatten)xx1x2r0   r0   r1   rotate_halfg   s   rj   c                 C   s   | |}| |}|dd|jd d f jddd}|dd|jd d f jddd}|jd }| dd|f | d|df }}|dd|f |d|df }	}
|| t||  }|	| t|	|  }tj||gdd}tj||
gd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.
        position_ids (`torch.Tensor`, *optional*):
            Deprecated and unused.
        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.
    .Nr4   r   r5   )	unsqueezerB   repeat_interleaverj   r>   cat)qkcossinposition_idsunsqueeze_dim
rotary_dimq_rotq_passk_rotk_passq_embedk_embedr0   r0   r1   apply_rotary_pos_embn   s   

$$
""r{   c                       s   e Zd ZdZddedee f fddZeddd	d
		dde	j
dee	j
e	j
f dee	j
 dee dee	j dee dee	j
e	j
f fddZ  ZS )GlmAttentionz=Multi-headed attention from 'Attention Is All You Need' paperNr#   	layer_idxc                    s   t    || _|| _t|d|j|j | _|j|j | _	| jd | _
|j| _d| _tj|j|j| j |jd| _tj|j|j| j |jd| _tj|j|j| j |jd| _tj|j| j |jdd| _d S )NrH   g      Tr   F)r!   r"   r#   r}   getattrr&   num_attention_headsrH   rF   rW   rP   attention_dropout	is_causalr$   r%   attention_biasq_projk_projv_projo_projr-   r#   r}   r.   r0   r1   r"      s$   
 zGlmAttention.__init__past_key_valuepast_key_values4.58new_nameversionr2   position_embeddingsrO   cache_positionrR   r3   c                 K   s$  |j d d }g |d| jR }| ||dd}	| ||dd}
| ||dd}|\}}t|	|
||\}	}
|d urW|||d}||
|| j	|\}
}t
}| jjdkret| jj }|| |	|
||f| jsqdn| j| jd|\}}|jg |dR   }| |}||fS )Nr4   r   r   )rq   rp   r   eagerrJ   )rQ   rP   )rB   rH   r   viewrY   r   r   r{   updater}   rd   r#   _attn_implementationr   rV   r   rP   rD   r^   r   )r-   r2   r   rO   r   r   rR   input_shapehidden_shapequery_statesr_   r`   rp   rq   cache_kwargsattention_interfacerc   ra   r0   r0   r1   r:      s8   


zGlmAttention.forwardN)NN)r;   r<   r=   __doc__r   r   intr"   r   r>   Tensortupler   
LongTensorr   r   r:   r@   r0   r0   r.   r1   r|      s*    r|   RMSNormc                       s.   e Zd Zd fdd	Zdd Zdd Z  ZS )	
GlmRMSNormư>c                    s&   t    tt|| _|| _dS )z9
        GlmRMSNorm is equivalent to T5LayerNorm
        N)r!   r"   r$   	Parameterr>   onesweightvariance_epsilon)r-   r&   epsr.   r0   r1   r"      s   

zGlmRMSNorm.__init__c                 C   sJ   |j }|tj}|djddd}|t|| j  }| j|| S )Nr   r4   T)keepdim)	rT   r]   r>   r\   powmeanrsqrtr   r   )r-   r2   input_dtypevariancer0   r0   r1   r:      s
   zGlmRMSNorm.forwardc                 C   s   t | jj d| j S )Nz, eps=)r   r   rB   r   )r-   r0   r0   r1   
extra_repr   s   zGlmRMSNorm.extra_repr)r   )r;   r<   r=   r"   r:   r   r@   r0   r0   r.   r1   r      s    r   c                       sD   e Zd ZU ejed< ddef fddZe e	dd Z
  ZS )	GlmRotaryEmbeddinginv_freqNr#   c                    s   t    t|drt|jtr|jd|jd| _nd| _|j| _	|j| _
|| _t| j | _| | j|\}| _| jd|dd | j| _d S )Nrope_scaling	rope_typetypedefaultr   F)
persistent)r!   r"   hasattr
isinstancer   dictgetr   max_position_embeddingsmax_seq_len_cachedoriginal_max_seq_lenr#   r   rope_init_fnattention_scalingregister_bufferr   original_inv_freq)r-   r#   devicer   r.   r0   r1   r"      s   
zGlmRotaryEmbedding.__init__c           
      C   s   | j d d d d f  |jd dd|j}|d d d d d f  }t|jjtr6|jjdkr6|jjnd}t	j
|dd+ | |  dd}t	j||fdd	}| | j }| | j }	W d    n1 smw   Y  |j|jd
|	j|jd
fS )Nr   r4   r   mpscpuF)device_typeenabledr   r5   )rT   )r   floatrC   rB   r]   r   r   r   strr>   autocastrY   rm   rp   r   rq   rT   )
r-   rg   rr   inv_freq_expandedposition_ids_expandedr   freqsembrp   rq   r0   r0   r1   r:     s   0&zGlmRotaryEmbedding.forwardr   )r;   r<   r=   r>   r   __annotations__r   r"   no_gradr   r:   r@   r0   r0   r.   r1   r      s   
 
r   c                       s   e Zd Zdedef fddZedddd							
				ddejde	ej de	ej
 de	e de	e de	ej
 de	eejejf  dee dejfddZ  ZS )GlmDecoderLayerr#   r}   c                    sR   t    |j| _t||d| _t|| _t|j|jd| _	t|j|jd| _
d S )N)r#   r}   r   )r!   r"   r&   r|   	self_attnr   mlpr   rms_norm_epsinput_layernormpost_attention_layernormr   r.   r0   r1   r"     s   

zGlmDecoderLayer.__init__r   r   r   r   NFr2   rO   rr   	use_cacher   r   rR   r3   c              
   K   s^   |}	|  |}| jd|||||||d|\}}
|	| }|}	| |}| |}|	| }|S )N)r2   rO   rr   r   r   r   r   r0   )r   r   r   r   )r-   r2   rO   rr   r   r   r   r   rR   residual_r0   r0   r1   r:   !  s&   




zGlmDecoderLayer.forward)NNNFNN)r;   r<   r=   r   r   r"   r   r>   r   r   r   r   boolr   r   r   r:   r@   r0   r0   r.   r1   r     s8    
	
r   c                   @   sH   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dS )GlmPreTrainedModelr#   modelTr   r   )r2   
attentionsN)r;   r<   r=   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_outputsr0   r0   r0   r1   r   D  s   
 
r   c                       s   e Zd Zdef fddZe e							ddeej	 deej
 deej	 dee d	eej d
eej	 dee dee defddZ  ZS )GlmModelr#   c                    s   t     j| _ j| _t j j| j| _t	 fddt
 jD | _t j jd| _t d| _d| _|   d S )Nc                    s   g | ]}t  |qS r0   )r   ).0r}   r#   r0   r1   
<listcomp>`  s    z%GlmModel.__init__.<locals>.<listcomp>r   r   F)r!   r"   pad_token_idpadding_idx
vocab_sizer$   	Embeddingr&   embed_tokens
ModuleListrangenum_hidden_layerslayersr   r   normr   
rotary_embgradient_checkpointing	post_initr,   r.   r   r1   r"   Y  s   zGlmModel.__init__N	input_idsrO   rr   r   inputs_embedsr   r   rR   r3   c              	   K   s   |d u |d uA rt d|d u r| |}|r!|d u r!t| jd}|d u r=|d ur-| nd}	tj|	|	|jd  |jd}|d u rF|	d}t
| j|||||d}
|}| ||}| jd | jj D ]}||f|
||||d|}qb| |}t||dS )	Nz:You must specify exactly one of input_ids or inputs_embedsr   r   r   )r   )r#   input_embedsrO   r   r   rr   )rO   rr   r   r   r   )last_hidden_stater   )
ValueErrorr   r   r#   get_seq_lengthr>   arangerB   r   rk   r   r   r   r   r   r   )r-   r   rO   rr   r   r   r   r   rR   past_seen_tokensrb   r2   r   decoder_layerr0   r0   r1   r:   i  sP   

	

zGlmModel.forward)NNNNNNN)r;   r<   r=   r   r"   r   r   r   r>   r   r   r   r?   r   r   r   r   r:   r@   r0   r0   r.   r1   r   W  s<    	
r   c                       s   e Zd ZdgZddiZddgdgfiZ fddZee										dd
e	e
j de	e
j de	e
j de	e de	e
j de	e
j de	e de	e
j deee
jf dee defddZ  ZS )GlmForCausalLMzlm_head.weightlm_headcolwise_repr2   logitsc                    s@   t  | t|| _|j| _tj|j|jdd| _| 	  d S )NFr   )
r!   r"   r   r   r   r$   r%   r&   r  r   r,   r.   r0   r1   r"     s
   
zGlmForCausalLM.__init__Nr   r   rO   rr   r   r   labelsr   r   logits_to_keeprR   r3   c
              
   K   s   | j d|||||||d|
}|j}t|	trt|	 dn|	}| |dd|ddf }d}|durB| jd||| jjd|
}t	|||j
|j|jd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   rO   rr   r   r   r   r   N)r  r  r   )lossr  r   r2   r   r0   )r   r   r   r   slicer  loss_functionr#   r   r   r   r2   r   )r-   r   rO   rr   r   r   r  r   r   r  rR   outputsr2   slice_indicesr  r  r0   r0   r1   r:     s0    zGlmForCausalLM.forward)	NNNNNNNNr   )r;   r<   r=   _tied_weights_keys_tp_plan_pp_planr"   r   r   r   r>   r   r   r   r?   r   r   r   r   r   r   r:   r@   r0   r0   r.   r1   r     sN    		
r   c                   @      e Zd ZdS )GlmForSequenceClassificationNr;   r<   r=   r0   r0   r0   r1   r        r  c                   @   r  )GlmForTokenClassificationNr  r0   r0   r0   r1   r    r  r  )r   r   r   r  r  )rJ   )Nr   )>typingr   r   r   r>   torch.nnr$   activationsr   cache_utilsr   r   
generationr	   integrationsr
   masking_utilsr   modeling_layersr   r   r   modeling_outputsr   r   modeling_rope_utilsr   r   modeling_utilsr   r   processing_utilsr   utilsr   r   r   utils.deprecationr   utils.genericr   configuration_glmr   Moduler   r   r   rI   r   rd   rj   r{   r|   r   r   r   r   r   r   r  r  __all__r0   r0   r0   r1   <module>   sj   

*E$.NK