TalkingBot/ingestion/.env.example

35 lines
786 B
Plaintext

AWS_REGION=ap-south-1
AWS_ACCESS_KEY_ID=
AWS_SECRET_ACCESS_KEY=
AWS_BUCKET=
DB_HOST=127.0.0.1
DB_PORT=3306
DB_USER=root
DB_PASSWORD=
DB_NAME=talkingbooklet
BEDROCK_MODEL_CHEAP=global.anthropic.claude-haiku-4-5-20251001-v1:0
BEDROCK_MODEL_COUNT=global.anthropic.claude-haiku-4-5-20251001-v1:0
BEDROCK_MODEL_EMBED=amazon.titan-embed-text-v2:0
EMBED_DIMENSIONS=1024
EMBED_BATCH_SIZE=8
OPENSEARCH_ENDPOINT=
OPENSEARCH_USER=
OPENSEARCH_PASSWORD=
OPENSEARCH_SERVICE=es
OPENSEARCH_INDEX_ALIAS=books_current
OPENSEARCH_INDEX_VERSION=v2
OPENSEARCH_LEGACY_INDEX=talkingbooklet
CHUNK_TARGET_TOKENS=400
CHUNK_OVERLAP_TOKENS=60
CHAPTER_TOKEN_THRESHOLD=6000
MAX_CONCURRENCY=4
TEXT_LAYER_MIN_RATIO=0.8
TEXT_LAYER_MIN_CHARS=20
TRANSCRIBE_LANGUAGE_CODE=en-US
INGEST_VERSION=v1
LOCAL_DATA_DIR=./data