38 lines
1.3 KiB
JSON
38 lines
1.3 KiB
JSON
|
{
|
||
|
|
"$schema": "http://json-schema.org/draft-07/schema#",
|
||
|
|
"title": "Chroma BM25 Embedding Function Schema",
|
||
|
|
"description": "Schema for the Chroma BM25 sparse embedding function configuration",
|
||
|
|
"version": "1.0.0",
|
||
|
|
"type": "object",
|
||
|
|
"properties": {
|
||
|
|
"k": {
|
||
|
|
"type": "number",
|
||
|
|
"description": "BM25 saturation parameter controlling term frequency scaling"
|
||
|
|
},
|
||
|
|
"b": {
|
||
|
|
"type": "number",
|
||
|
|
"description": "BM25 length normalization parameter"
|
||
|
|
},
|
||
|
|
"avg_doc_length": {
|
||
|
|
"type": "number",
|
||
|
|
"description": "Average document length in tokens used for normalization"
|
||
|
|
},
|
||
|
|
"token_max_length": {
|
||
|
|
"type": "number",
|
||
|
|
"description": "Maximum token length allowed before filtering"
|
||
|
|
},
|
||
|
|
"stopwords": {
|
||
|
|
"type": "array",
|
||
|
|
"description": "Optional custom stopword list (in lowercase) to override the defaults",
|
||
|
|
"items": {
|
||
|
|
"type": "string"
|
||
|
|
}
|
||
|
|
},
|
||
|
|
"include_tokens": {
|
||
|
|
"type": "boolean",
|
||
|
|
"description": "Whether to store token strings in the sparse vectors (default: true)"
|
||
|
|
}
|
||
|
|
},
|
||
|
|
"additionalProperties": false
|
||
|
|
}
|