rh*ddlmZddlmZmZddlZddlZddlm Z ddlm Z m Z ddZ ddZ ddZdd Zedd Zedd Zdd Zdd ZddZddZy)) annotations)AnyoverloadN) coo_matrix)Tensordevicect|trwtd|DrOtj|Dcgc]0}|j j tj2c}Stj|}n%t|tstj|}|jr |j tjS|Scc}w)a Converts the input `a` to a PyTorch tensor if it is not already a tensor. Handles lists of sparse tensors by stacking them. Args: a (Union[list, np.ndarray, Tensor]): The input array or tensor. Returns: Tensor: The converted tensor. c3XK|]"}t|txr |j$ywN) isinstancer is_sparse).0xs t/var/www/html/ai-insurance-compliance-backend/venv/lib/python3.12/site-packages/sentence_transformers/util/tensor.py z%_convert_to_tensor..s#@z!V$44@s(*dtype) r listalltorchstackcoalescetofloat32tensorrr )ars r_convert_to_tensorr s!T @a@ @;;aP emm DPQ Q QA 6 " LLO{{tt%--t(( H Qs5CcN|jdk(r|jd}|S)z If the tensor `a` is 1-dimensional, it is unsqueezed to add a batch dimension. Args: a (Tensor): The input tensor. Returns: Tensor: The tensor with a batch dimension. r)dim unsqueezers r_convert_to_batchr#$s# uuw!| KKN Hcdt|}|jdk(r|jd}|S)a Converts the input data to a tensor with a batch dimension. Handles lists of sparse tensors by stacking them. Args: a (Union[list, np.ndarray, Tensor]): The input data to be converted. Returns: Tensor: The converted tensor with a batch dimension. rr)rr r!r"s r_convert_to_batch_tensorr&3s- 1Auuw!| KKN Hr$cH|js,tjjj |ddS|j }|j |j}}tj|jd|j}|jd|d|dztj|jd|d}|dkD}|j}||xx||zcc<tj|||jS)z Normalizes the embeddings matrix, so that each sentence embedding has unit length. Args: embeddings (Tensor): The input embeddings matrix. Returns: Tensor: The normalized embeddings matrix. r)pr rr)r rnn functional normalizerindicesvalueszerossizer index_add_sqrt index_selectclonesparse_coo_tensor) embeddingsr.r/ row_normsmasknormalized_valuess rnormalize_embeddingsr;Ds   xx"",,Z1!,DD$$&J ((*J,=,=,?VG JOOA.z7H7HII GAJ 2 9%221gajAI q=D dy.  " "7,=z?P QQr$cyr r7 truncate_dims rtruncate_embeddingsr@asY\r$cyr r=r>s rr@r@es]`r$c|dd|fS)a Truncates the embeddings matrix. Args: embeddings (Union[np.ndarray, torch.Tensor]): Embeddings to truncate. truncate_dim (Optional[int]): The dimension to truncate sentence embeddings to. `None` does no truncation. Example: >>> from sentence_transformers import SentenceTransformer >>> from sentence_transformers.util import truncate_embeddings >>> model = SentenceTransformer("tomaarsen/mpnet-base-nli-matryoshka") >>> embeddings = model.encode(["It's so nice outside!", "Today is a beautiful day.", "He drove to work earlier"]) >>> embeddings.shape (3, 768) >>> model.similarity(embeddings, embeddings) tensor([[1.0000, 0.8100, 0.1426], [0.8100, 1.0000, 0.2121], [0.1426, 0.2121, 1.0000]]) >>> truncated_embeddings = truncate_embeddings(embeddings, 128) >>> truncated_embeddings.shape >>> model.similarity(truncated_embeddings, truncated_embeddings) tensor([[1.0000, 0.8092, 0.1987], [0.8092, 1.0000, 0.2716], [0.1987, 0.2716, 1.0000]]) Returns: Union[np.ndarray, torch.Tensor]: Truncated embeddings. .Nr=r>s rr@r@is: c=L=( ))r$c.||St|tjrtj|}|j \}}|j }tjtj|t||d\}}tj|tj}tj||jdjdt||}d||j|jf<d||<|S)a Keeps only the top-k values (in absolute terms) for each embedding and creates a sparse tensor. Args: embeddings (Union[np.ndarray, torch.Tensor]): Embeddings to sparsify by keeping only top_k values. max_active_dims (int): Number of values to keep as non-zeros per embedding. Returns: torch.Tensor: A sparse tensor containing only the top-k values per embedding. r)kr rr*Tr)r npndarrayrrshapertopkabsmin zeros_likeboolaranger!expandflatten) r7max_active_dims batch_sizer r_ top_indicesr9 batch_indicess rselect_max_active_dimsrVs*bjj)\\*-  &&OJ   FZZ * 5_c9RXYZNA{   Jejj 9DLLF;EEaHOOPRTWXgilTmnM;?D   +"5"5"7 78Ju r$cj|D]-}t||ts||j|||</|S)au Send a PyTorch batch (i.e., a dictionary of string keys to Tensors) to a device (e.g. "cpu", "cuda", "mps"). Args: batch (Dict[str, Tensor]): The batch to send to the device. target_device (torch.device): The target device (e.g. "cpu", "cuda", "mps"). Returns: Dict[str, Tensor]: The batch with tensors sent to the target device. )r rr)batch target_devicekeys rbatch_to_devicer[s>6 eCj& )s}5E#J6 Lr$c|j}|jjj}|j jj}t ||d|dff|j S)Nrr)rH)rr.cpunumpyr/rrH)rr.r/s r to_scipy_coor_sf Aiikoo%%'G XXZ^^  # # %F v GAJ78 HHr$)rzlist | np.ndarray | Tensorreturnr)rrr`r)r7rr`r)r7 np.ndarrayr? int | Noner`ra)r7 torch.Tensorr?rbr`rc)r7np.ndarray | torch.Tensorr?rbr`rd)r7rdrQrbr`rc)rXdict[str, Any]rYrr`re)rrr`r) __future__rtypingrrr^rFr scipy.sparserrrrr#r&r;r@rVr[r_r=r$rrisc"  #  2   "R: \ \ ` `*@D"Ir$