32 lines
935 B
JSON
32 lines
935 B
JSON
{
|
|
"search_backend": "searxng",
|
|
"search_backend_url": "http://searxng:8080/search",
|
|
"search_backend_fallback": true,
|
|
"search_region": "de-de",
|
|
"search_max_results": 5,
|
|
"scrape_max_tokens": 1200,
|
|
"search_top_k": 15,
|
|
"search_dense_weight": 0.5,
|
|
"search_max_results_to_keep": 4,
|
|
"chunk_dense_weight": 0.5,
|
|
"chunk_max_results_to_keep": 8,
|
|
"chunk_rank_oversample": 3,
|
|
"chunk_dedupe_jaccard_threshold": 0.92,
|
|
"chunk_max_per_source_url": 2,
|
|
"max_concurrent_crawls": 3,
|
|
"pipeline_timeout_seconds": 90.0,
|
|
"crawl_fit_markdown_mode": "bm25",
|
|
"crawl_fit_min_chars": 200,
|
|
"crawl_bm25_threshold": 1.5,
|
|
"crawl_bm25_language": "german",
|
|
"crawl_max_chunk_tokens": 300,
|
|
"crawl_overlap_tokens": 40,
|
|
"crawl_max_page_tokens": 6000,
|
|
"embedding_backend": "onnx",
|
|
"embedding_model": "fast",
|
|
"dense_document_embed_batch_size": 32,
|
|
"encoding_name": "embedding",
|
|
"blocked_domains": [],
|
|
"trace_path": ""
|
|
}
|