1025private static readonly (string Prefix, ModelEncoding Encoding)[] _modelPrefixToEncoding = 1028( "o1-", ModelEncoding.O200kBase ), // e.g. o1-mini 1029( "gpt-4o-", ModelEncoding.O200kBase), // e.g., gpt-4o-2024-05-13 1030( "gpt-4-", ModelEncoding.Cl100kBase), // e.g., gpt-4-0314, etc., plus gpt-4-32k 1031( "gpt-3.5-", ModelEncoding.Cl100kBase), // e.g, gpt-3.5-turbo-0301, -0401, etc. 1032( "gpt-35-", ModelEncoding.Cl100kBase ) // Azure deployment name 1035private static readonly Dictionary<string, ModelEncoding> _modelToEncoding = 1036new Dictionary<string, ModelEncoding>(StringComparer.OrdinalIgnoreCase) 1039{ "gpt-4o", ModelEncoding.O200kBase }, 1040{ "o1", ModelEncoding.O200kBase }, 1041{ "gpt-4", ModelEncoding.Cl100kBase }, 1042{ "gpt-3.5-turbo", ModelEncoding.Cl100kBase }, 1043{ "gpt-3.5-turbo-16k", ModelEncoding.Cl100kBase }, 1044{ "gpt-35", ModelEncoding.Cl100kBase }, // Azure deployment name 1045{ "gpt-35-turbo", ModelEncoding.Cl100kBase }, // Azure deployment name 1046{ "gpt-35-turbo-16k", ModelEncoding.Cl100kBase }, // Azure deployment name 1049{ "text-davinci-003", ModelEncoding.P50kBase }, 1050{ "text-davinci-002", ModelEncoding.P50kBase }, 1051{ "text-davinci-001", ModelEncoding.R50kBase }, 1052{ "text-curie-001", ModelEncoding.R50kBase }, 1053{ "text-babbage-001", ModelEncoding.R50kBase }, 1054{ "text-ada-001", ModelEncoding.R50kBase }, 1055{ "davinci", ModelEncoding.R50kBase }, 1056{ "curie", ModelEncoding.R50kBase }, 1057{ "babbage", ModelEncoding.R50kBase }, 1058{ "ada", ModelEncoding.R50kBase }, 1061{ "code-davinci-002", ModelEncoding.P50kBase }, 1062{ "code-davinci-001", ModelEncoding.P50kBase }, 1063{ "code-cushman-002", ModelEncoding.P50kBase }, 1064{ "code-cushman-001", ModelEncoding.P50kBase }, 1065{ "davinci-codex", ModelEncoding.P50kBase }, 1066{ "cushman-codex", ModelEncoding.P50kBase }, 1069{ "text-davinci-edit-001", ModelEncoding.P50kEdit }, 1070{ "code-davinci-edit-001", ModelEncoding.P50kEdit }, 1074{ "text-embedding-ada-002", ModelEncoding.Cl100kBase }, 1075{ "text-embedding-3-small", ModelEncoding.Cl100kBase }, 1076{ "text-embedding-3-large", ModelEncoding.Cl100kBase }, 1079{ "text-similarity-davinci-001", ModelEncoding.R50kBase }, 1080{ "text-similarity-curie-001", ModelEncoding.R50kBase }, 1081{ "text-similarity-babbage-001", ModelEncoding.R50kBase }, 1082{ "text-similarity-ada-001", ModelEncoding.R50kBase }, 1083{ "text-search-davinci-doc-001", ModelEncoding.R50kBase }, 1084{ "text-search-curie-doc-001", ModelEncoding.R50kBase }, 1085{ "text-search-babbage-doc-001", ModelEncoding.R50kBase }, 1086{ "text-search-ada-doc-001", ModelEncoding.R50kBase }, 1087{ "code-search-babbage-code-001", ModelEncoding.R50kBase }, 1088{ "code-search-ada-code-001", ModelEncoding.R50kBase }, 1091{ "gpt2", ModelEncoding.GPT2 } 1094private static ModelEncoding GetModelEncoding(string modelName) 1096if (!_modelToEncoding.TryGetValue(modelName, out ModelEncoding encoder)) 1098foreach ((string Prefix, ModelEncoding Encoding) in _modelPrefixToEncoding) 1108if (encoder == ModelEncoding.None) 1118private static (Dictionary<string, int> SpecialTokens, Regex Regex, string VocabFile, Type? DataType, string PackageName) GetTiktokenConfigurations(ModelEncoding modelEncoding, string? modelName = null) 1122case ModelEncoding.Cl100kBase: 1126case ModelEncoding.GPT2: 1129case ModelEncoding.O200kBase: 1132case ModelEncoding.P50kBase: 1135case ModelEncoding.P50kEdit: 1139case ModelEncoding.R50kBase: 1204ModelEncoding modelEncoding, 1450ModelEncoding modelEncoding; 1453modelEncoding = ModelEncoding.Cl100kBase; 1457modelEncoding = ModelEncoding.O200kBase; 1461modelEncoding = ModelEncoding.P50kBase; 1465modelEncoding = ModelEncoding.P50kEdit; 1469modelEncoding = ModelEncoding.R50kBase;