ޠjSSKJr SSKrSSKJr SSKJrJrJrJ r J r J r SSK J r \(aSSKr\R "\5r\ \\\\\\4rSSjrSSjr"SS \ S S 9r"S S 5rg)) annotationsN)Sequence) TYPE_CHECKINGAnyCallableLiteralOptionalUnion) TypedDictcSSKn[U5S:Xd[U5S:XaUR"/5$UR"U5nUR"U5nURSURS:wa&[ SURSURS35eSSKnUR"XR S9nUR"XR S9nSURXSS 9- n[U[5(aUR"U/5$UR"U5$![a [RS 5 URRUSS 9nURRUSS 9nUR"S S S 9 UR "XR"5UR$"XV5- nSSS5 O!,(df  O=fSWUR&"U5UR("U5-'Us$f=f)z>? s2 A2D%D%%A$H 0G9 H G 1H?HcX^SSKJm SU4SjjnU$![a [S5ef=f)zGet the OpenAI GPT-3 encoder.r)ClientzTHe default encoder for the EmbeddingDistance class uses the OpenAI API. Please either install the openai library with `pip install openai` or provide a custom encoder function (Callable[[str], Sequence[float]]).c>T"5nURR[U5SS9nURVs/sHo3RPM sn$s snf)Nztext-embedding-3-small)inputmodel) embeddingscreatelistdata embedding)textsclientresponsed OpenAIClients r3 encode_text(_get_openai_encoder..encode_textIsO$$++u+%=, &.]]3] ]333sA)r?z Sequence[str]returnzSequence[Sequence[float]])openair6r )rDrCs @r3_get_openai_encoderrH>s< 14    T   s)c*\rSrSr%S\S'S\S'Srg)EmbeddingConfigSz0Callable[[list[str]], Sequence[Sequence[float]]]encoderzCLiteral['cosine', 'euclidean', 'manhattan', 'chebyshev', 'hamming']rN)__name__ __module__ __qualname____firstlineno____annotations____static_attributes__rMr3rJrJSs == OOrTrJF)totalc\rSrSrS S SjjrSSjrSSjr\SSj5r\SSj5r \SSj5r \SS j5r \SS j5r S r g)EmbeddingDistanceXNcU=(d 0nURS5=(d SUlURS5=(d [5Ulg)NrrrL)getdistancerHrL)selfconfigs r3__init__EmbeddingDistance.__init__Ys;2 8,8 zz),E0C0E rTcSSKnURX/5nUR"U5nUR USUS5R 5$![a [S5ef=f)NrzWThe EmbeddingDistance class requires NumPy. Please install it with `pip install numpy`.r )rr rLr_compute_distanceitem)r\ prediction referencer-r:vectors r3evaluateEmbeddingDistance.evaluateasr    \\:"9: *%%%fQi;@@BB '  s AA&c|URS:XaURX5$URS:XaURX5$URS:XaURX5$URS:XaUR X5$URS:XaUR X5$[ SUR35e)Nr euclidean manhattan chebyshevhammingzInvalid distance metric: )r[_cosine_distance_euclidean_distance_manhattan_distance_chebyshev_distance_hamming_distancer)r\abs r3ra#EmbeddingDistance._compute_distanceqs ==H $((. . ]]k )++A1 1 ]]k )++A1 1 ]]k )++A1 1 ]]i '))!/ /8HI IrTc$S[U/U/5- $)zCompute the cosine distance between two vectors. Args: a (np.ndarray): The first vector. b (np.ndarray): The second vector. Returns: np.ndarray: The cosine distance. g?)r4rrrss r3rm"EmbeddingDistance._cosine_distances&sQC000rTcD[RRX- 5$)zCompute the Euclidean distance between two vectors. Args: a (np.ndarray): The first vector. b (np.ndarray): The second vector. Returns: np.floating: The Euclidean distance. )r-r#r$rvs r3rn%EmbeddingDistance._euclidean_distancesyy~~ae$$rTcZ[R"[R"X- 55$)zCompute the Manhattan distance between two vectors. Args: a (np.ndarray): The first vector. b (np.ndarray): The second vector. Returns: np.floating: The Manhattan distance. )r-sumabsrvs r3ro%EmbeddingDistance._manhattan_distancevvbffQUm$$rTcZ[R"[R"X- 55$)zCompute the Chebyshev distance between two vectors. Args: a (np.ndarray): The first vector. b (np.ndarray): The second vector. Returns: np.floating: The Chebyshev distance. )r-maxr|rvs r3rp%EmbeddingDistance._chebyshev_distancer~rTc2[R"X:g5$)zCompute the Hamming distance between two vectors. Args: a (np.ndarray): The first vector. b (np.ndarray): The second vector. Returns: np.floating: The Hamming distance. )r-meanrvs r3rq#EmbeddingDistance._hamming_distanceswwqvrT)r[rL)N)r]zOptional[EmbeddingConfig])rcstrrdrrFr)rr np.ndarrayrsrrFz np.floating)rrrrsrrFr)rNrOrPrQr^rfra staticmethodrmrnrorprqrSrMrTr3rWrWXs-1F)FCCC  C J 1 1 % % % % % %  rTrW)r+Matrixr,rrFr)rFz4Callable[[Sequence[str]], Sequence[Sequence[float]]]) __future__rloggingcollections.abcrtypingrrrrr r typing_extensionsr rr- getLoggerrNr!r<rrr4rHrJrWrMrTr3rsz"$(   8 $ tDK $s)S0 1"J*PiuP ffrT