24 lines
789 B
YAML
24 lines
789 B
YAML
id: 1c-rag-corpus
|
|
name: 1C RAG Corpus
|
|
status: draft
|
|
source_dir: plugins/1c/rag/sources
|
|
official_docs:
|
|
source_manifest: plugins/1c/rag/official-docs/sources.yaml
|
|
raw_dir: plugins/1c/rag/official-docs/raw
|
|
normalized_dir: plugins/1c/rag/official-docs/normalized
|
|
access: licensed_private
|
|
usage: rag_only_by_default
|
|
prepared_path: plugins/1c/datasets/prepared/rag_corpus.jsonl
|
|
supported_extensions:
|
|
- .md
|
|
- .txt
|
|
- .bsl
|
|
- .os
|
|
chunking:
|
|
chunk_size: 1800
|
|
overlap: 200
|
|
privacy:
|
|
allow_private_client_data: false
|
|
require_secret_scrub: true
|
|
notes: "Do not place database dumps, credentials, tokens, personal data, or client-specific secrets in this corpus. Official 1C:ITS documents are licensed private sources: keep full text out of git and prefer RAG over fine-tuning."
|