| { |
| "schema_version": 2, |
| "model_version": "v2.2", |
| "model_type": "lightgbm", |
| "n_features_char": 131072, |
| "n_features_word": 131072, |
| "char_ngram_range": [ |
| 3, |
| 5 |
| ], |
| "word_ngram_range": [ |
| 1, |
| 2 |
| ], |
| "word_analyzer": "word", |
| "char_analyzer": "char_wb", |
| "structural_names": [ |
| "len_chars", |
| "len_tokens", |
| "url_count", |
| "unique_domain_count", |
| "mention_count", |
| "hashtag_count", |
| "tag_p_count", |
| "tag_e_count", |
| "tag_t_count", |
| "tag_other_count", |
| "emoji_count", |
| "emoji_ratio", |
| "zero_width_count", |
| "caps_ratio", |
| "digit_ratio", |
| "punct_ratio", |
| "dup_body_bucket" |
| ], |
| "group_feature_names": [ |
| "group_size", |
| "time_span_hours", |
| "n_unique_bodies", |
| "len_chars_std", |
| "same_first_token_ratio", |
| "mean_pairwise_jaccard" |
| ], |
| "group_sizes_trained": [ |
| 1, |
| 2, |
| 3, |
| 5, |
| 10 |
| ], |
| "hashing": { |
| "algorithm": "MurmurHash3 x86 32-bit (seed=0)", |
| "alternate_sign": true |
| }, |
| "invisible_chars": [ |
| "", |
| "", |
| "", |
| "", |
| "", |
| "", |
| "", |
| "", |
| "", |
| "", |
| "", |
| "", |
| "", |
| "", |
| "", |
| "", |
| "", |
| "", |
| "", |
| "", |
| "" |
| ], |
| "unicode_normalization": "NFKC", |
| "casefold": true |
| } |
