a
    hR                  
   @   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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$m%Z% ddl&m'Z' G dd dej(Z)G dd dej(Z*G dd dej(Z+dd Z,d3ddZ-ej.e/ej.dddZ0d4ej(ej.ej.ej.eej. e1e1ee d!d"d#Z2G d$d% d%ej(Z3G d&d' d'eZ4e G d(d) d)eZ5e G d*d+ d+e5Z6e G d,d- d-e5eZ7G d.d/ d/ee5Z8G d0d1 d1ee5Z9g d2Z:dS )5    )CallableOptionalUnionN)nn   )ACT2FN)CacheDynamicCache)GenerationMixin)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   )GemmaConfigc                       s>   e Zd Zdeed fddZdd Zdd Zd	d
 Z  Z	S )GemmaRMSNormư>)dimepsc                    s&   t    || _tt|| _d S N)super__init__r    r   	ParametertorchZzerosweight)selfr   r    	__class__ d/var/www/html/assistant/venv/lib/python3.9/site-packages/transformers/models/gemma/modeling_gemma.pyr#   /   s    
zGemmaRMSNorm.__init__c                 C   s$   |t |djddd| j  S )N   T)Zkeepdim)r%   Zrsqrtpowmeanr    )r'   xr*   r*   r+   _norm4   s    zGemmaRMSNorm._normc                 C   s*   |  | }|d| j   }||S )Ng      ?)r1   floatr&   Ztype_as)r'   r0   outputr*   r*   r+   forward7   s    zGemmaRMSNorm.forwardc                 C   s   t | jj d| j S )Nz, eps=)tupler&   shaper    )r'   r*   r*   r+   
extra_repr>   s    zGemmaRMSNorm.extra_repr)r   )
__name__
__module____qualname__intr2   r#   r1   r4   r7   __classcell__r*   r*   r(   r+   r   .   s   r   c                       s$   e Zd Z fddZdd Z  ZS )GemmaMLPc                    sr   t    || _|j| _|j| _tj| j| jdd| _tj| j| jdd| _tj| j| jdd| _	t
|j | _d S NFZbias)r"   r#   confighidden_sizeZintermediate_sizer   Linear	gate_projup_proj	down_projr   Z
hidden_actact_fnr'   r@   r(   r*   r+   r#   C   s    
zGemmaMLP.__init__c                 C   s$   |  | | || | }|S r!   )rE   rF   rC   rD   )r'   r0   rE   r*   r*   r+   r4   M   s     zGemmaMLP.forward)r8   r9   r:   r#   r4   r<   r*   r*   r(   r+   r=   B   s   
r=   c                       sD   e Zd ZU ejed< ded fddZe e	dd Z
  ZS )	GemmaRotaryEmbeddinginv_freqNr@   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defaultrI   F)
persistent)r"   r#   hasattr
isinstancerK   dictgetrL   Zmax_position_embeddingsZmax_seq_len_cachedZoriginal_max_seq_lenr@   r   Zrope_init_fnattention_scalingZregister_bufferrI   Zoriginal_inv_freq)r'   r@   devicerI   r(   r*   r+   r#   U   s    
zGemmaRotaryEmbedding.__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rl|jjdkrl|jjnd}t	j
|ddV | |  dd}t	j||fdd	}| | j }| | j }	W d    n1 s0    Y  |j|jd
|	j|jd
fS )Nr   r-   r   ZmpscpuF)device_typeZenabledr,   r   dtype)rI   r2   expandr6   torU   rQ   rM   strr%   Zautocast	transposecatcosrT   sinrZ   )
r'   r0   position_idsZinv_freq_expandedZposition_ids_expandedrW   ZfreqsZembr`   ra   r*   r*   r+   r4   f   s    0&,zGemmaRotaryEmbedding.forward)N)r8   r9   r:   r%   Tensor__annotations__r   r#   Zno_gradr   r4   r<   r*   r*   r(   r+   rH   R   s
   

rH   c                 C   sH   | dd| j d d f }| d| j d d df }tj| |fddS )z*Rotates half the hidden dims of the input..Nr-   r,   rX   )r6   r%   r_   )r0   x1Zx2r*   r*   r+   rotate_halfv   s    rf   c                 C   sD   | |}| |}| | t| |  }|| t||  }||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.
    )	unsqueezerf   )qkr`   ra   rb   Zunsqueeze_dimZq_embedZk_embedr*   r*   r+   apply_rotary_pos_emb}   s
    

rj   )hidden_statesn_repreturnc                 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)r6   r[   reshape)rk   rl   batchnum_key_value_headsslenhead_dimr*   r*   r+   	repeat_kv   s
    0rs           )modulequerykeyvalueattention_maskscalingdropoutkwargsc                 K   s   t || j}t || j}	t||dd| }
|d urf|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   r-   )r   rZ   )ptrainingr   )rs   num_key_value_groupsr%   matmulr^   r6   r   Z
functionalZsoftmaxZfloat32r\   rZ   r{   r   
contiguous)ru   rv   rw   rx   ry   rz   r{   r|   
key_statesvalue_statesattn_weightscausal_maskattn_outputr*   r*   r+   eager_attention_forward   s    
&r   c                       s   e Zd ZdZeed fddZedddddej	e
ej	ej	f eej	 ee eej ee e
ej	ej	f d
ddZ  ZS )GemmaAttentionz=Multi-headed attention from 'Attention Is All You Need' paperr@   	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|jd| _d S )Nrr   g      Tr?   )r"   r#   r@   r   getattrrA   Znum_attention_headsrr   rp   r   rz   attention_dropoutZ	is_causalr   rB   Zattention_biasq_projk_projv_projo_projr'   r@   r   r(   r*   r+   r#      s(    
zGemmaAttention.__init__past_key_valuepast_key_values4.58new_nameversionN)rk   position_embeddingsry   r   cache_positionr|   rm   c                 K   s$  |j d d }g |d| jR }| ||dd}	| ||dd}
| ||dd}|\}}t|	|
||\}	}
|d ur|||d}||
|| j	|\}
}t
}| jjdkrt| jj }|| |	|
||f| jsdn| j| jd|\}}|jg |dR   }| |}||fS )Nr-   r   r,   )ra   r`   r   eagerrt   )r{   rz   )r6   rr   r   viewr^   r   r   rj   updater   r   r@   Z_attn_implementationr   r   r   rz   rn   r   r   )r'   rk   r   ry   r   r   r|   Zinput_shapeZhidden_shapeZquery_statesr   r   r`   ra   Zcache_kwargsZattention_interfacer   r   r*   r*   r+   r4      s8    


zGemmaAttention.forward)NN)r8   r9   r:   __doc__r   r;   r#   r   r%   rc   r5   r   r   
LongTensorr   r   r4   r<   r*   r*   r(   r+   r      s     r   c                       s   e Zd Zeed fddZedddddeje	ej e	ej
 e	e e	e e	ej
 e	eejejf  ee ejd
	ddZ  ZS )GemmaDecoderLayerr   c                    sR   t    |j| _t||d| _t|| _t|j|jd| _	t|j|jd| _
d S )Nr   r    )r"   r#   rA   r   	self_attnr=   mlpr   rms_norm_epsinput_layernormpost_attention_layernormr   r(   r*   r+   r#     s    

zGemmaDecoderLayer.__init__r   r   r   r   NF)	rk   ry   rb   r   	use_cacher   r   r|   rm   c              
   K   s^   |}	|  |}| jf |||||||d|\}}
|	| }|}	| |}| |}|	| }|S )N)rk   ry   rb   r   r   r   r   )r   r   r   r   )r'   rk   ry   rb   r   r   r   r   r|   Zresidual_r*   r*   r+   r4     s&    




zGemmaDecoderLayer.forward)NNNFNN)r8   r9   r:   r   r;   r#   r   r%   rc   r   r   r   boolr5   r   r   r4   r<   r*   r*   r(   r+   r     s&   
      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 )GemmaPreTrainedModelr@   modelTr   r   )rk   
attentionsN)r8   r9   r:   r   rd   Zbase_model_prefixZsupports_gradient_checkpointingZ_no_split_modulesZ_skip_keys_device_placementZ_supports_flash_attnZ_supports_sdpaZ_supports_flex_attnZ_can_compile_fullgraphZ_supports_attention_backendr   r   Z_can_record_outputsr*   r*   r*   r+   r   3  s   
r   c                       st   e Zd Zed fddZeedeej	 eej
 eej	 ee eej ee eej	 ee ed	ddZ  ZS )	
GemmaModelrJ   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 r*   )r   ).0r   rJ   r*   r+   
<listcomp>O      z'GemmaModel.__init__.<locals>.<listcomp>r   rJ   F)r"   r#   Zpad_token_idZpadding_idx
vocab_sizer   Z	EmbeddingrA   embed_tokensZ
ModuleListrangenum_hidden_layerslayersr   r   normrH   
rotary_embZgradient_checkpointing	post_initrG   r(   rJ   r+   r#   H  s    zGemmaModel.__init__N)		input_idsry   rb   r   inputs_embedsr   r   r|   rm   c              
   K   s(  |d u |d uA rt d|d u r*| |}|rB|d u rBt| jd}|d u rz|d urZ| nd}	tj|	|	|jd  |jd}|d u r|	d}t
| j|||||d}
|}| ||}tj| jjd |jd}|| }| jd | jj D ]"}||f|
|||||d	|}q| |}t||r |nd d
S )Nz:You must specify exactly one of input_ids or inputs_embedsrJ   r   r   )rU   )r@   Zinput_embedsry   r   r   rb   g      ?rY   )ry   rb   r   r   r   r   )last_hidden_stater   )
ValueErrorr   r	   r@   Zget_seq_lengthr%   Zaranger6   rU   rg   r   r   ZtensorrA   rZ   r   r   r   r   )r'   r   ry   rb   r   r   r   r   r|   Zpast_seen_tokensr   rk   r   Z
normalizerZdecoder_layerr*   r*   r+   r4   X  sV    




zGemmaModel.forward)NNNNNNN)r8   r9   r:   r   r#   r   r   r   r%   r   rc   r   FloatTensorr   r   r   r   r4   r<   r*   r*   r(   r+   r   F  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e	e
j e	e
j e	e
j e	e e	e
j e	e
j e	e e	e
j eee
jf ee ed
ddZ  ZS )GemmaForCausalLMzlm_head.weightlm_headZcolwise_reprk   logitsc                    s@   t  | t|| _|j| _tj|j|jdd| _| 	  d S r>   )
r"   r#   r   r   r   r   rB   rA   r   r   rG   r(   r*   r+   r#     s
    
zGemmaForCausalLM.__init__Nr   )r   ry   rb   r   r   labelsr   r   logits_to_keepr|   rm   c
              
   K   s   | j f |||||||d|
}|j}t|	tr<t|	 dn|	}| |dd|ddf }d}|dur| jf ||| jjd|
}t	|||j
|j|jdS )a|  
        Example:

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

        >>> model = GemmaForCausalLM.from_pretrained("google/gemma-7b")
        >>> tokenizer = AutoTokenizer.from_pretrained("google/gemma-7b")

        >>> prompt = "What is your favorite condiment?"
        >>> 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]
        "What is your favorite condiment?"
        ```)r   ry   rb   r   r   r   r   N)r   r   r   )lossr   r   rk   r   )r   r   rQ   r;   slicer   Zloss_functionr@   r   r   r   rk   r   )r'   r   ry   rb   r   r   r   r   r   r   r|   outputsrk   Zslice_indicesr   r   r*   r*   r+   r4     s0     zGemmaForCausalLM.forward)	NNNNNNNNr   )r8   r9   r:   Z_tied_weights_keysZ_tp_planZ_pp_planr#   r   r   r   r%   r   rc   r   r   r   r   r;   r   r   r   r4   r<   r*   r*   r(   r+   r     s8   	         r   c                   @   s   e Zd ZdS )GemmaForSequenceClassificationNr8   r9   r:   r*   r*   r*   r+   r     s   r   c                   @   s   e Zd ZdS )GemmaForTokenClassificationNr   r*   r*   r*   r+   r     s   r   )r   r   r   r   r   )Nr   )rt   );typingr   r   r   r%   r   Zactivationsr   Zcache_utilsr   r	   Z
generationr
   Zmasking_utilsr   Zmodeling_layersr   r   r   Zmodeling_outputsr   r   Zmodeling_rope_utilsr   r   Zmodeling_utilsr   r   Zprocessing_utilsr   utilsr   r   r   Zutils.deprecationr   Zutils.genericr   Zconfiguration_gemmar   Moduler   r=   rH   rf   rj   rc   r;   rs   r2   r   r   r   r   r   r   r   r   __all__r*   r*   r*   r+   <module>   sV   $
 G.WK