| { |
| "schema_version": "vigil.compact-preprocessing.v1", |
| "unicode_version": "15.0.0", |
| "serializer": { |
| "name": "package-structure-v1", |
| "max_package_bytes": 131072, |
| "max_file_bytes": 49152, |
| "max_package_files": 96, |
| "middle_truncation_marker": "\n[...MIDDLE TRUNCATED...]\n", |
| "file_order": "priority,depth,unicode-lower-path,exact-path", |
| "path_separator": "/", |
| "binary_text": "strict-utf8-without-nul-else-ascii-printable-runs-min-4" |
| }, |
| "word_vectorizer": { |
| "algorithm": "murmurhash3_x86_32_utf8_seed_0", |
| "n_features": 32768, |
| "ngram_range": [1, 2], |
| "alternate_sign": true, |
| "lowercase": "unicode-full-lower", |
| "norm": "l2", |
| "token_pattern": "(?u)\\b[\\w./$:@-]{2,}\\b" |
| }, |
| "char_vectorizer": { |
| "algorithm": "murmurhash3_x86_32_utf8_seed_0", |
| "analyzer": "char", |
| "n_features": 32768, |
| "ngram_range": [4, 4], |
| "alternate_sign": true, |
| "lowercase": "unicode-full-lower", |
| "whitespace_normalization": "python-unicode-whitespace-runs-min-2-to-ascii-space", |
| "norm": "l2" |
| }, |
| "structured_feature_count": 16, |
| "total_feature_count": 65552 |
| } |
|
|
