package tg

Get desktop application:
View/edit binary Protocol Buffers messages

Used in: TEmbedderConfig

• AM_UNDEFINED = 0

• AM_AVG = 1

• AM_MAX = 2

• AM_MIN = 3

• AM_MATRIX = 4

Used in: TDocumentProto

• NC_UNDEFINED = 0

• NC_ANY = 1

• NC_SOCIETY = 2

• NC_ECONOMY = 3

• NC_TECHNOLOGY = 4

• NC_SPORTS = 5

• NC_ENTERTAINMENT = 6

• NC_SCIENCE = 7

• NC_OTHER = 8

• NC_NOT_NEWS = 9

Used in: TEmbedderConfig

• EF_UNDEFINED = 0

• EF_TITLE = 1

• EF_TEXT = 2

• EF_ALL = 3

Used in: TEmbedderConfig

• ET_UNDEFINED = 0

• ET_FASTTEXT = 1

• ET_TORCH = 2

• ET_TFIDF = 3

Used in: TClusteringEmbeddingKeyWeight, TEmbedderConfig, TEmbeddingProto

• EK_UNDEFINED = 0

• EK_FASTTEXT_CLASSIC = 1

• EK_FASTTEXT_TITLE = 2

• EK_FASTTEXT_TEXT = 3

• EK_TFIDF = 4

• IF_UNDEFINED = 0

• IF_HTML = 1

• IF_JSON = 2

• IF_JSONL = 3

Used in: TCategoryModelConfig, TClusteringConfig, TDocumentProto, TEmbedderConfig

• LN_UNDEFINED = 0

• LN_RU = 1

• LN_EN = 2

• LN_OTHER = 3

• string lang_detect = 1

• repeated TCategoryModelConfig category_models = 2

• repeated TEmbedderConfig embedders = 3

• uint32 min_text_length = 4

• bool parse_links = 5

• bool save_texts = 6

• bool compute_nasty = 7

• bool save_not_news = 8

Used in: TAnnotatorConfig

• ELanguage language = 1

• string path = 2

• repeated TClusteringConfig clusterings = 1

• float iter_timestamp_percentile = 2

Used in: TClustererConfig

• ELanguage language = 1

• float small_threshold = 2

• uint64 small_cluster_size = 3

• float medium_threshold = 4

• uint64 medium_cluster_size = 5

• float large_threshold = 6

• uint64 large_cluster_size = 7

• uint64 chunk_size = 8

• uint64 intersection_size = 9

• bool use_timestamp_moving = 10

• bool ban_same_hosts = 11

• repeated TClusteringEmbeddingKeyWeight embedding_keys_weights = 12

Used in: TClusteringConfig

• EEmbeddingKey embedding_key = 1

• float weight = 2

• string file_name = 1

• string url = 2

• string site_name = 3

• uint64 pub_time = 4

• uint64 fetch_time = 5

• uint32 ttl = 6

• string title = 7

• string text = 8

• string description = 9

• ELanguage language = 10

• ECategory category = 11

• repeated string out_links = 12

• repeated TEmbeddingProto embeddings = 13

• bool nasty = 14

Used in: TAnnotatorConfig

• EEmbedderType type = 1

• ELanguage language = 2

• EEmbeddingKey embedding_key = 3

• EEmbedderField embedder_field = 4

• EAggregationMode aggregation_mode = 5

• uint32 max_words = 6

• string model_path = 7

• string vector_model_path = 8

• string vocabulary_path = 9

Used in: TDocumentProto

• EEmbeddingKey key = 1

• repeated float value = 2

• uint64 min_cluster_size = 1

• uint32 threads = 1

• uint32 max_connection_num = 2

• uint32 max_connection_num_per_ip = 3

• uint32 idle_connection_timeout = 4

• uint32 keepalive_requests_number = 5

• uint32 pipelining_requests_number = 6

• string db_path = 7

• bool db_fail_if_missing = 8

• uint32 db_max_open_files = 9

• bool skip_irrelevant_docs = 10

• uint32 clusterer_sleep = 11

• string annotator_config_path = 12

• string clusterer_config_path = 13

• string summarizer_config_path = 14

• string ranker_config_path = 15

• string hosts_rating = 1

• string alexa_rating = 2