rhddlmZddlZddlZddlmZddlmZddlm Z ddl m Z m Z m Z mZe jeZddZdd Zdd Zdd Zdd Zdd ZddZddZddZddZy)) annotationsN)pairwise_distances)Tensor)logging)_convert_to_batch_tensor_convert_to_tensornormalize_embeddings to_scipy_cooct||S) Computes the cosine similarity between two tensors. Args: a (Union[list, np.ndarray, Tensor]): The first tensor. b (Union[list, np.ndarray, Tensor]): The second tensor. Returns: Tensor: Matrix with res[i][j] = cos_sim(a[i], b[j]) )cos_simabs x/var/www/html/ai-insurance-compliance-backend/venv/lib/python3.12/site-packages/sentence_transformers/util/similarity.pypytorch_cos_simrs 1a=ct|}t|}t|}t|}tj||j ddj S)r rr)rr torchmm transposeto_denserra_normb_norms rrrsS !#A #A !! $F !! $F 88FF,,Q2 3 < < >>rc(t|}t|}|js |jr9t|}t|}||zjdj St t|t|j S)a Computes the pairwise cosine similarity cos_sim(a[i], b[i]). Args: a (Union[list, np.ndarray, Tensor]): The first tensor. b (Union[list, np.ndarray, Tensor]): The second tensor. Returns: Tensor: Vector with res[i] = cos_sim(a[i], b[i]) dim)r is_sparser sumrpairwise_dot_scorers rpairwise_cos_simr$0s 1A1A {{akk%a(%a($$$,5577!"6q"9;OPQ;RS\\^^rct|}t|}tj||jddj S)a Computes the dot-product dot_prod(a[i], b[j]) for all i and j. Args: a (Union[list, np.ndarray, Tensor]): The first tensor. b (Union[list, np.ndarray, Tensor]): The second tensor. Returns: Tensor: Matrix with res[i][j] = dot_prod(a[i], b[j]) rr)rrrrrrs r dot_scorer&Gs= !#A #A 88Aq{{1a( ) 2 2 44rctt|}t|}||zjdjS)a Computes the pairwise dot-product dot_prod(a[i], b[i]). Args: a (Union[list, np.ndarray, Tensor]): The first tensor. b (Union[list, np.ndarray, Tensor]): The second tensor. Returns: Tensor: Vector with res[i] = dot_prod(a[i], b[i]) rr)r r"rrs rr#r#Xs6 1A1A E;;2;  ' ' ))rct|}t|}|js |jrtjdt |}t |}t ||d}t j| jj|jjSt j||dj S)a Computes the manhattan similarity (i.e., negative distance) between two tensors. Handles sparse tensors without converting to dense when possible. Args: a (Union[list, np.ndarray, Tensor]): The first tensor. b (Union[list, np.ndarray, Tensor]): The second tensor. Returns: Tensor: Matrix with res[i][j] = -manhattan_distance(a[i], b[j]) z8Using scipy for sparse Manhattan similarity computation. manhattan)metricg?p) rr!logger warning_oncer rr from_numpyfloattodevicercdist)rra_coob_coodists r manhattan_simr7is !#A #A{{akkVWQQ!%{C&,,.11!((;DDFF AqC(11333rct|}t|}tjtj||z dj S)a< Computes the manhattan similarity (i.e., negative distance) between pairs of tensors. Args: a (Union[list, np.ndarray, Tensor]): The first tensor. b (Union[list, np.ndarray, Tensor]): The second tensor. Returns: Tensor: Vector with res[i] = -manhattan_distance(a[i], b[i]) rr)r rr"absrrs rpairwise_manhattan_simr:sB 1A1A IIeiiA&B / 8 8 : ::rclt|}t|}|jrtjj ||zdj j d}tjj ||zdj j d}tj||jj }|d|zz |z}tj|d}tj|j Stj||d S) a Computes the euclidean similarity (i.e., negative distance) between two tensors. Handles sparse tensors without converting to dense when possible. Args: a (Union[list, np.ndarray, Tensor]): The first tensor. b (Union[list, np.ndarray, Tensor]): The second tensor. Returns: Tensor: Matrix with res[i][j] = -euclidean_distance(a[i], b[j]) rrrg)ming@r+) rr!rsparser"r unsqueezematmultclampsqrtr3)rr a_norm_sq b_norm_sq dot_product squared_dists r euclidean_simrHs !#A #A{{LL$$QU$2;;=GGJ LL$$QU$2;;=GGJ ll1acce,557 !1{?2Y> {{rfse" .&dd   H % ?&_.5"*"46;")>C"#!r