o
    sh                     @   sZ   d dl mZmZmZ d dlZd dlmZmZ d dlmZ d dl	m
Z
 G dd dejZdS )    )AnyDictIterableN)Tensornn)util)SentenceTransformerc                       s|   e Zd Zdejfdededdf fddZdee	e
ef  d	edefd
dZde	e
ef fddZede
fddZ  ZS )
CoSENTLossg      4@modelscalereturnNc                    s$   t t|   || _|| _|| _dS )a  
        This class implements CoSENT (Cosine Sentence) loss.
        It expects that each of the InputExamples consists of a pair of texts and a float valued label, representing
        the expected similarity score between the pair.

        It computes the following loss function:

        ``loss = logsum(1+exp(s(k,l)-s(i,j))+exp...)``, where ``(i,j)`` and ``(k,l)`` are any of the input pairs in the
        batch such that the expected similarity of ``(i,j)`` is greater than ``(k,l)``. The summation is over all possible
        pairs of input pairs in the batch that match this condition.

        Anecdotal experiments show that this loss function produces a more powerful training signal than :class:`CosineSimilarityLoss`,
        resulting in faster convergence and a final model with superior performance. Consequently, CoSENTLoss may be used
        as a drop-in replacement for :class:`CosineSimilarityLoss` in any training script.

        Args:
            model: SentenceTransformerModel
            similarity_fct: Function to compute the PAIRWISE similarity
                between embeddings. Default is
                ``util.pairwise_cos_sim``.
            scale: Output of similarity function is multiplied by scale
                value. Represents the inverse temperature.

        References:
            - For further details, see: https://kexue.fm/archives/8847

        Requirements:
            - Sentence pairs with corresponding similarity scores in range of the similarity function. Default is [-1,1].

        Relations:
            - :class:`AnglELoss` is CoSENTLoss with ``pairwise_angle_sim`` as the metric, rather than ``pairwise_cos_sim``.
            - :class:`CosineSimilarityLoss` seems to produce a weaker training signal than CoSENTLoss. In our experiments, CoSENTLoss is recommended.

        Inputs:
            +--------------------------------+------------------------+
            | Texts                          | Labels                 |
            +================================+========================+
            | (sentence_A, sentence_B) pairs | float similarity score |
            +--------------------------------+------------------------+

        Example:
            ::

                from sentence_transformers import SentenceTransformer, SentenceTransformerTrainer, losses
                from datasets import Dataset

                model = SentenceTransformer("microsoft/mpnet-base")
                train_dataset = Dataset.from_dict({
                    "sentence1": ["It's nice weather outside today.", "He drove to work."],
                    "sentence2": ["It's so sunny.", "She walked to the store."],
                    "score": [1.0, 0.3],
                })
                loss = losses.CoSENTLoss(model)

                trainer = SentenceTransformerTrainer(
                    model=model,
                    train_dataset=train_dataset,
                    loss=loss,
                )
                trainer.train()
        N)superr	   __init__r
   similarity_fctr   )selfr
   r   r   	__class__ e/var/www/html/alpaca_bot/venv/lib/python3.10/site-packages/sentence_transformers/losses/CoSENTLoss.pyr      s   >
zCoSENTLoss.__init__sentence_featureslabelsc                    s    fdd|D }  |d |d }| j }|d d d f |d d d f  }|d d d f |d d d f k }| }|d| d  }tjtd|j|dfdd}tj	|dd}|S )Nc                    s   g | ]	}  |d  qS )sentence_embedding)r
   ).0sentence_featurer   r   r   
<listcomp>O   s    z&CoSENTLoss.forward.<locals>.<listcomp>r      g   mB)dim)
r   r   floattorchcatzerostodeviceview	logsumexp)r   r   r   
embeddingsscoreslossr   r   r   forwardN   s   
  &zCoSENTLoss.forwardc                 C   s   | j | jjdS )N)r   r   )r   r   __name__r   r   r   r   get_config_dictb   s   zCoSENTLoss.get_config_dictc                 C   s   dS )Nz
@online{kexuefm-8847,
    title={CoSENT: A more efficient sentence vector scheme than Sentence-BERT},
    author={Su Jianlin},
    year={2022},
    month={Jan},
    url={https://kexue.fm/archives/8847},
}
r   r   r   r   r   citatione   s   zCoSENTLoss.citation)r+   
__module____qualname__r   pairwise_cos_simr   r   r   r   r   strr   r*   r   r,   propertyr-   __classcell__r   r   r   r   r	   
   s    ""Cr	   )typingr   r   r   r    r   r   sentence_transformersr   )sentence_transformers.SentenceTransformerr   Moduler	   r   r   r   r   <module>   s    