a
    hW                  
   @   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 dd
lmZ ddlmZ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$m%Z%m&Z&m'Z' ddl(m)Z) ddl*m+Z+ ddl,m-Z- dd Z.d8ddZ/ej0e1ej0dddZ2d9ej3ej0ej0ej0eej0 e4e4e#e% dddZ5G d d! d!ej3Z6ed"G d#d$ d$ej3Z7G d%d& d&ej3Z8G d'd( d(eZ9e&G d)d* d*e!Z:G d+d, d,ej3Z;e&G d-d. d.e:Z<e&G d/d0 d0e:eZ=G d1d2 d2ee:Z>G d3d4 d4ee:Z?G d5d6 d6ee:Z@g d7ZAdS ):    )CallableOptionalUnionN)nn   )ACT2FN)CacheDynamicCache)GenerationMixin)use_kernel_forward_from_hub)create_causal_mask!create_sliding_window_causal_mask)FlashAttentionKwargs)GenericForQuestionAnswering 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   )SmolLM3Configc                 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..N   dim)shapetorchcat)xx1Zx2 r*   h/var/www/html/assistant/venv/lib/python3.9/site-packages/transformers/models/smollm3/modeling_smollm3.pyrotate_half1   s    r,   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.
    )	unsqueezer,   )qkcossinposition_idsZunsqueeze_dimZq_embedZk_embedr*   r*   r+   apply_rotary_pos_emb8   s
    

r3   )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)r%   expandreshape)r4   r5   batchnum_key_value_headsslenhead_dimr*   r*   r+   	repeat_kvS   s
    0r=           )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$   dtype)ptrainingr   )r=   num_key_value_groupsr&   matmul	transposer%   r   Z
functionalZsoftmaxfloat32torH   rE   rJ   
contiguous)r?   r@   rA   rB   rC   rD   rE   rF   
key_statesvalue_statesattn_weightsZcausal_maskattn_outputr*   r*   r+   eager_attention_forward_   s    
&rU   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ej	 f d
ddZ  ZS )SmolLM3Attentionz=Multi-headed attention from 'Attention Is All You Need' paperconfig	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| _|j| | _|jr|j| dkr|jnd | _d S )Nr<   g      TZbiassliding_attention)super__init__rX   rY   getattrhidden_sizeZnum_attention_headsr<   r:   rK   rD   attention_dropoutZ	is_causalr   LinearZattention_biasq_projk_projv_projo_projZno_rope_layersuse_ropeZuse_sliding_windowlayer_typessliding_windowselfrX   rY   	__class__r*   r+   r]   |   s6    
zSmolLM3Attention.__init__past_key_valuepast_key_values4.58new_nameversionN)r4   position_embeddingsrC   rn   cache_positionrF   r6   c                 K   s*  |j d d }g |d| jR }| ||dd}	| ||dd}
| ||dd}| jr|\}}t|	|
||\}	}
|d urd|i}|	|
|| j
|\}
}t}| jjdkrt| jj }|| |	|
||f| jsdn| j| j| jd|\}}|jg |dR   }| |}||fS )Nr!   r   r"   rt   eagerr>   )rE   rD   rh   )r%   r<   rb   viewrM   rc   rd   rf   r3   updaterY   rU   rX   Z_attn_implementationr   rJ   r`   rD   rh   r8   rP   re   )rj   r4   rs   rC   rn   rt   rF   Zinput_shapeZhidden_shapeZquery_statesrQ   rR   r0   r1   Zcache_kwargsZattention_interfacerT   rS   r*   r*   r+   forward   s<    
	

zSmolLM3Attention.forward)NN)__name__
__module____qualname____doc__r    intr]   r   r&   Tensortupler   r   
LongTensorr   r   rx   __classcell__r*   r*   rk   r+   rV   y   s     rV   ZRMSNormc                       s.   e Zd Zd fdd	Zdd Zdd Z  ZS )	SmolLM3RMSNormư>c                    s&   t    tt|| _|| _dS )z=
        SmolLM3RMSNorm is equivalent to T5LayerNorm
        N)r\   r]   r   	Parameterr&   Zonesweightvariance_epsilon)rj   r_   epsrk   r*   r+   r]      s    
zSmolLM3RMSNorm.__init__c                 C   sJ   |j }|tj}|djddd}|t|| j  }| j|| S )Nr"   r!   T)Zkeepdim)	rH   rO   r&   rN   powmeanZrsqrtr   r   )rj   r4   Zinput_dtypeZvariancer*   r*   r+   rx      s
    zSmolLM3RMSNorm.forwardc                 C   s   t | jj d| j S )Nz, eps=)r   r   r%   r   )rj   r*   r*   r+   
extra_repr   s    zSmolLM3RMSNorm.extra_repr)r   )ry   rz   r{   r]   rx   r   r   r*   r*   rk   r+   r      s   r   c                       s$   e Zd Z fddZdd Z  ZS )
SmolLM3MLPc                    sx   t    || _|j| _|j| _tj| j| j|jd| _tj| j| j|jd| _	tj| j| j|jd| _
t|j | _d S )NrZ   )r\   r]   rX   r_   Zintermediate_sizer   ra   Zmlp_bias	gate_projup_proj	down_projr   Z
hidden_actact_fnrj   rX   rk   r*   r+   r]      s    
zSmolLM3MLP.__init__c                 C   s$   |  | | || | }|S )N)r   r   r   r   )rj   r(   r   r*   r*   r+   rx      s     zSmolLM3MLP.forward)ry   rz   r{   r]   rx   r   r*   r*   rk   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 )SmolLM3DecoderLayerrW   c                    s^   t    |j| _t||d| _t|| _t|j|jd| _	t|j|jd| _
|j| | _d S )NrW   r   )r\   r]   r_   rV   	self_attnr   mlpr   rms_norm_epsinput_layernormpost_attention_layernormrg   attention_typeri   rk   r*   r+   r]      s    

zSmolLM3DecoderLayer.__init__rm   rn   ro   rp   NF)	r4   rC   r2   rn   	use_cachert   rs   rF   r6   c              
   K   s^   |}	|  |}| jf |||||||d|\}}
|	| }|}	| |}| |}|	| }|S )N)r4   rC   r2   rn   r   rt   rs   )r   r   r   r   )rj   r4   rC   r2   rn   r   rt   rs   rF   Zresidual_r*   r*   r+   rx      s&    




zSmolLM3DecoderLayer.forward)NNNFNN)ry   rz   r{   r    r}   r]   r   r&   r~   r   r   r   boolr   r   r   rx   r   r*   r*   rk   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 )SmolLM3PreTrainedModelrX   modelTr   rn   )r4   
attentionsN)ry   rz   r{   r    __annotations__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   rV   Z_can_record_outputsr*   r*   r*   r+   r     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 )	SmolLM3RotaryEmbeddinginv_freqNrX   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   Zmax_position_embeddingsZmax_seq_len_cachedZoriginal_max_seq_lenrX   r   Zrope_init_fnattention_scalingZregister_bufferr   Zoriginal_inv_freq)rj   rX   devicer   rk   r*   r+   r]   2  s    
zSmolLM3RotaryEmbedding.__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#   )rH   )r   floatr7   r%   rO   r   r   r   strr&   ZautocastrM   r'   r0   r   r1   rH   )
rj   r(   r2   Zinv_freq_expandedZposition_ids_expandedr   ZfreqsZembr0   r1   r*   r*   r+   rx   C  s    0&,zSmolLM3RotaryEmbedding.forward)N)ry   rz   r{   r&   r~   r   r    r]   Zno_gradr   rx   r   r*   r*   rk   r+   r   /  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 )	SmolLM3Modelr   c                    s   t     j| _ j| _t j j| j| _t	 fddt
 jD | _t j jd| _t d| _d| _d| jjv | _|   d S )Nc                    s   g | ]}t  |qS r*   )r   ).0rY   r   r*   r+   
<listcomp>\      z)SmolLM3Model.__init__.<locals>.<listcomp>r   r   Fr[   )r\   r]   Zpad_token_idZpadding_idx
vocab_sizer   Z	Embeddingr_   embed_tokensZ
ModuleListrangenum_hidden_layerslayersr   r   normr   
rotary_embZgradient_checkpointingrX   rg   has_sliding_layers	post_initr   rk   r   r+   r]   U  s    zSmolLM3Model.__init__N)		input_idsrC   r2   rn   inputs_embedsr   rt   rF   r6   c              
   K   sD  |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
| }
ts| j|||||d}dtf i |i}
| jrtf i ||
d< |}| ||}| jd | jj D ](}||f|
|j |||||d	|}q| |}t||r<|nd d
S )Nz:You must specify exactly one of input_ids or inputs_embedsr   r   r   )r   )rX   Zinput_embedsrC   rt   rn   r2   Zfull_attentionr[   )rC   r2   rn   r   rt   rs   )last_hidden_statern   )
ValueErrorr   r	   rX   Zget_seq_lengthr&   Zaranger%   r   r-   r   r   r   r   r   r   r   r   r   r   r   )rj   r   rC   r2   rn   r   r   rt   rF   Zpast_seen_tokensZcausal_mask_mappingZmask_kwargsr4   rs   Zdecoder_layerr*   r*   r+   rx   f  sZ    



zSmolLM3Model.forward)NNNNNNN)ry   rz   r{   r    r]   r   r   r   r&   r   r~   r   FloatTensorr   r   r   r   rx   r   r*   r*   rk   r+   r   S  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 )SmolLM3ForCausalLMzlm_head.weightlm_headZcolwise_repr4   logitsc                    s@   t  | t|| _|j| _tj|j|jdd| _| 	  d S )NFrZ   )
r\   r]   r   r   r   r   ra   r_   r   r   r   rk   r*   r+   r]     s
    
zSmolLM3ForCausalLM.__init__Nr   )r   rC   r2   rn   r   labelsr   rt   logits_to_keeprF   r6   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, SmolLM3ForCausalLM

        >>> model = SmolLM3ForCausalLM.from_pretrained("meta-smollm3/SmolLM3-2-7b-hf")
        >>> tokenizer = AutoTokenizer.from_pretrained("meta-smollm3/SmolLM3-2-7b-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   rC   r2   rn   r   r   rt   N)r   r   r   )lossr   rn   r4   r   )r   r   r   r}   slicer   Zloss_functionrX   r   r   rn   r4   r   )rj   r   rC   r2   rn   r   r   r   rt   r   rF   outputsr4   Zslice_indicesr   r   r*   r*   r+   rx     s0     zSmolLM3ForCausalLM.forward)	NNNNNNNNr   )ry   rz   r{   Z_tied_weights_keysZ_tp_planZ_pp_planr]   r   r   r   r&   r   r~   r   r   r   r   r}   r   r   r   rx   r   r*   r*   rk   r+   r     s8   	         r   c                   @   s   e Zd ZdS ) SmolLM3ForSequenceClassificationNry   rz   r{   r*   r*   r*   r+   r     s   r   c                   @   s   e Zd ZdS )SmolLM3ForTokenClassificationNr   r*   r*   r*   r+   r      s   r   c                   @   s   e Zd ZdZdS )SmolLM3ForQuestionAnsweringZtransformerN)ry   rz   r{   r   r*   r*   r*   r+   r     s   r   )r   r   r   r   r   r   )Nr   )r>   )Btypingr   r   r   r&   r   Zactivationsr   Zcache_utilsr   r	   Z
generationr
   Zintegrationsr   Zmasking_utilsr   r   Zmodeling_flash_attention_utilsr   Zmodeling_layersr   r   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_smollm3r    r,   r3   r~   r}   r=   Moduler   rU   rV   r   r   r   r   r   r   r   r   r   r   __all__r*   r*   r*   r+   <module>   s^   
 O/$\K