o
    eiZ                     @   s:  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(m)Z) ddl*m+Z+ ddl,m-Z- e%.e/Z0G dd dej1Z2dd Z3edd<ddZ4dej5de6dej5fddZ7	 d=d!ej1d"ej5d#ej5d$ej5d%ej5dB d&e8d'e8d(e!e# fd)d*Z9ee4G d+d, d,ej1Z:G d-d. d.ej1Z;G d/d0 d0eZ<e$G d1d2 d2eZ=e$G d3d4 d4e=Z>e$G d5d6 d6e=eZ?G d7d8 d8ee=Z@G d9d: d:ee=ZAg d;ZBdS )>    )Callable)OptionalN   )ACT2FN)CacheDynamicCache)GenerationMixin)use_kernel_func_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logging)can_return_tuplemaybe_autocastmerge_with_config_defaults)capture_outputs   )	PhiConfigc                       s~   e Zd ZU ejed< ddef fddZe			ddedB de	d de
dB d	ed
ef fddZe edd Z  ZS )PhiRotaryEmbeddinginv_freqNconfigc                    s   t    |j| _|j| _|| _| jjd | _| j}| jdkr$t	| j }|| j|\}| _
| jd|dd | jd| dd d S )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__ b/home/ubuntu/transcripts/venv/lib/python3.10/site-packages/transformers/models/phi/modeling_phi.pyr'   '   s   


zPhiRotaryEmbedding.__init__r1   ztorch.deviceseq_lenreturnztorch.Tensorc           	      C   st   | j d }| j dd}t| ddp| j| j }t|| }d}d|tjd|dtjdj	|tj
d	|   }||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      dtype)r1   r>   )r+   getgetattrhidden_sizenum_attention_headsinttorcharangeint64tofloat)	r!   r1   r7   baser:   r;   dimattention_factorr    r5   r5   r6   r,   7   s   
&z2PhiRotaryEmbedding.compute_default_rope_parametersc           
      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	|dd+ | |  
dd}tj||fdd	}| | j }| | j }	W d    n1 slw   Y  |j|jd
|	j|jd
fS )Nr   r   mpscpuF)device_typeenabledr<   rJ   r=   )r    rH   expandshaperG   r1   
isinstancetypestrr   	transposerD   catcosr-   sinr>   )
r0   xposition_idsinv_freq_expandedposition_ids_expandedrO   freqsembrY   rZ   r5   r5   r6   forwardW   s   0&zPhiRotaryEmbedding.forwardN)NNN)__name__
__module____qualname__rD   Tensor__annotations__r   r'   staticmethodr   rC   tuplerH   r,   no_gradr   ra   __classcell__r5   r5   r3   r6   r   $   s&   
 

r   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..NrL   r<   rQ   )rS   rD   rX   )r[   x1x2r5   r5   r6   rotate_halfg   s   rn   rotary_pos_embc                 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.
        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.
    )	unsqueezern   )qkrY   rZ   unsqueeze_dimq_embedk_embedr5   r5   r6   apply_rotary_pos_embn   s
   

rv   hidden_statesn_repr8   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)rS   rR   reshape)rw   rx   batchnum_key_value_headsslenr;   r5   r5   r6   	repeat_kv   s
   0r}           modulequerykeyvalueattention_maskscalingdropoutkwargsc                 K   s   t || j}t || j}	t||dd| }
|d ur |
| }
tjj|
dtjd	|j
}
tjj|
|| jd}
t|
|	}|dd }||
fS )Nr<   r   rL   )rJ   r>   )ptrainingr   )r}   num_key_value_groupsrD   matmulrW   nn
functionalsoftmaxfloat32rG   r>   r   r   
contiguous)r   r   r   r   r   r   r   r   
key_statesvalue_statesattn_weightsattn_outputr5   r5   r6   eager_attention_forward   s   
r   c                       s   e Zd ZdZdedef fddZ		ddejde	ejejf d	ejdB d
e
dB dejdB de	ejejdB f fddZ  ZS )PhiAttentionz=Multi-headed attention from 'Attention Is All You Need' paperr!   	layer_idxc                    s(  t    || _|| _t|d|j|j | _|j|j | _	| jd | _
|j| _d| _tj|j|j| j dd| _tj|j|j| j dd| _tj|j|j| j dd| _tj|j| j |jdd| _t| j|jd  | _|j| _| jrtj|j|j |jdd| _tj|j|j |jdd| _d S d S )Nr;   g      Tbiasr:   )epselementwise_affine)r&   r'   r!   r   r@   rA   rB   r;   r{   r   r   attention_dropout	is_causalr   Linearq_projk_projv_projdenserC   r+   rotary_ndimsqk_layernorm	LayerNormlayer_norm_epsq_layernormk_layernormr0   r!   r   r3   r5   r6   r'      s,   
zPhiAttention.__init__Nrw   position_embeddingsr   past_key_valuescache_positionr8   c                 K   s  |j d d }g |d| jR }| ||dd}	| ||dd}
| ||dd}| jrB| |	}	| 	|
}
|\}}|	dd | j
f |	d| j
d f }}|
dd | j
f |
d| j
d f }}t||||\}}tj||fdd}	tj||fdd}
|d ur|||d}||
|| j|\}
}t| jjt}|| |	|
||f| jsdn| j| jd|\}}|jg |dR   }| |}||fS )	NrL   r   r<   .rQ   )rZ   rY   r   r~   )r   r   )rS   r;   r   viewrW   r   r   r   r   r   r   rv   rD   rX   updater   r   get_interfacer!   _attn_implementationr   r   r   r   ry   r   r   )r0   rw   r   r   r   r   r   input_shapehidden_shapequery_statesr   r   rY   rZ   	query_rot
query_passkey_rotkey_passcache_kwargsattention_interfacer   r   r5   r5   r6   ra      sN   	



zPhiAttention.forward)NN)rc   rd   re   __doc__r   rC   r'   rD   rf   ri   r   
LongTensorra   rk   r5   r5   r3   r6   r      s$    r   c                       s2   e Zd Z fddZdejdejfddZ  ZS )PhiMLPc                    sD   t    || _t|j | _t|j|j	| _
t|j	|j| _d S rb   )r&   r'   r!   r   
hidden_actactivation_fnr   r   rA   intermediate_sizefc1fc2r0   r!   r3   r5   r6   r'     s
   
zPhiMLP.__init__rw   r8   c                 C   s"   |  |}| |}| |}|S rb   )r   r   r   )r0   rw   r5   r5   r6   ra     s   


zPhiMLP.forward)rc   rd   re   r'   rD   rf   ra   rk   r5   r5   r3   r6   r     s    r   c                       s   e Zd Zdedef fddZ							ddejdejdB d	ejdB d
e	dB de
dB de
dB dejdB deejejf dB deejeejejf dB f fddZ  ZS )PhiDecoderLayerr!   r   c                    sH   t    t||d| _t|| _tj|j|j	d| _
t|j| _d S )N)r   r   )r&   r'   r   	self_attnr   mlpr   r   rA   r   input_layernormDropoutresid_pdropresid_dropoutr   r3   r5   r6   r'     s
   

zPhiDecoderLayer.__init__NFrw   r   r\   r   output_attentions	use_cacher   r   r8   c	                 K   sr   |}
|  |}| jd||||||||d|	\}}| |}| | |}|| |
 }|f}|r7||f7 }|S )N)rw   r   r\   r   r   r   r   r   r5   )r   r   r   r   )r0   rw   r   r\   r   r   r   r   r   r   residualattn_outputsself_attn_weightsfeed_forward_hidden_statesoutputsr5   r5   r6   ra     s*   
	


zPhiDecoderLayer.forward)NNNFFNN)rc   rd   re   r   rC   r'   rD   rf   r   r   boolri   FloatTensorra   rk   r5   r5   r3   r6   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 )PhiPreTrainedModelr!   modelTr   r   )rw   
attentionsN)rc   rd   re   r   rg   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_outputsr5   r5   r5   r6   r   E  s   
 
r   c                       s   e Zd Zdef fddZeee									ddej	dB dej
dB dej	dB dedB d	ejdB d
edB dedB dedB dej	dB dee defddZ  ZS )PhiModelr!   c                    s   t     j| _ j| _t j j| j| _t	 fddt
 jD | _t d| _d| _t j| _tj j jd| _|   d S )Nc                    s   g | ]}t  |qS r5   )r   ).0r   r!   r5   r6   
<listcomp>a  s    z%PhiModel.__init__.<locals>.<listcomp>r   Fr   )r&   r'   pad_token_idpadding_idx
vocab_sizer   	EmbeddingrA   embed_tokens
ModuleListrangenum_hidden_layerslayersr   
rotary_embgradient_checkpointingr   
embd_pdropembed_dropoutr   r   final_layernorm	post_initr   r3   r   r6   r'   Z  s   zPhiModel.__init__N	input_idsr   r\   r   inputs_embedsr   r   output_hidden_statesr   r   r8   c
                 K   s  |d ur|n| j j}|d ur|n| j j}|d ur|n| j j}|d u |d uA r*td| jr9| jr9|r9td d}|d u rB| 	|}|rN|d u rNt
| j d}|	d u rj|d urZ| nd}tj|||jd  |jd}	|d u rs|	d}t| j |||	||d}| |}|}| j||d	}|rd
nd }|rd
nd }| jd | j j D ]&}|r||f7 }||f||||||	|d|
}|d }|r||d f7 }q| |}|r||f7 }t||r|nd ||dS )Nz:You must specify exactly one of input_ids or inputs_embedszX`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`.Fr   r   r   )r1   )r!   r   r   r   r   r\   )r\   r5   )r   r\   r   r   r   r   r   )last_hidden_stater   rw   r   )r!   r   r   r   
ValueErrorr   r   loggerwarning_oncer   r   get_seq_lengthrD   rE   rS   r1   rp   r   r   r   r   r   r   r   )r0   r   r   r\   r   r   r   r   r   r   r   past_seen_tokenscausal_maskrw   r   all_hidden_statesall_self_attnsdecoder_layerlayer_outputsr5   r5   r6   ra   k  s   


	
	


zPhiModel.forward)	NNNNNNNNN)rc   rd   re   r   r'   r   r   r   rD   r   rf   r   r   r   r   r   r   ra   rk   r5   r5   r3   r6   r   X  sJ    	
r   c                       s   e Zd ZddiZddiZddgdgfiZ fddZee																	
dde	j
d	B de	jd	B de	j
d	B ded	B de	jd	B de	j
d	B ded	B de	j
d	B dee	jB dee defddZ  ZS )PhiForCausalLMzlm_head.weightzmodel.embed_tokens.weightlm_headcolwise_gather_outputrw   logitsc                    s@   t  | t|| _|j| _tj|j|jdd| _| 	  d S )NTr   )
r&   r'   r   r   r   r   r   rA   r  r   r   r3   r5   r6   r'     s
   
zPhiForCausalLM.__init__Nr   r   r   r\   r   r   labelsr   r   logits_to_keepr   r8   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, PhiForCausalLM

        >>> model = PhiForCausalLM.from_pretrained("meta-phi/Phi-2-7b-hf")
        >>> tokenizer = AutoTokenizer.from_pretrained("meta-phi/Phi-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   r\   r   r   r   r   N)r  r  r   )lossr  r   rw   r   r5   )r   r   rT   rC   slicer  loss_functionr!   r   r   r   rw   r   )r0   r   r   r\   r   r   r  r   r   r  r   r   rw   slice_indicesr  r  r5   r5   r6   ra     s0    zPhiForCausalLM.forward)	NNNNNNNNr   )rc   rd   re   _tied_weights_keys_tp_plan_pp_planr'   r   r   rD   r   rf   r   r   r   rC   r   r   r   ra   rk   r5   r5   r3   r6   r    sN    		
r  c                   @      e Zd ZdS )PhiForSequenceClassificationNrc   rd   re   r5   r5   r5   r6   r        r  c                   @   r  )PhiForTokenClassificationNr  r5   r5   r5   r6   r    r  r  )r   r   r  r  r  )r   )r~   )Ccollections.abcr   typingr   rD   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   r   utils.output_capturingr   configuration_phir   
get_loggerrc   r   Moduler   rn   rv   rf   rC   r}   rH   r   r   r   r   r   r   r  r  r  __all__r5   r5   r5   r6   <module>   sn   
C
X0tK