Tonyxay commited on
Commit
9f8fc4c
·
verified ·
1 Parent(s): d08f95c

Update ingest_train_data.py

Browse files
Files changed (1) hide show
  1. ingest_train_data.py +3 -3
ingest_train_data.py CHANGED
@@ -42,9 +42,9 @@ TRAIN_DATA_DIR = Path("/train_data") if Path("/train_data").is_dir() else PROJEC
42
 
43
  CHROMA_PERSIST_DIR = str(DATA_DIR / "chroma_db")
44
  COLLECTION_NAME = "rag_documents"
45
- CHUNK_SIZE = 512
46
- CHUNK_OVERLAP = 50
47
- EMBEDDING_BATCH_SIZE = 16
48
 
49
  _CONFIG_PATH = DATA_DIR / "config.json"
50
  if not _CONFIG_PATH.exists():
 
42
 
43
  CHROMA_PERSIST_DIR = str(DATA_DIR / "chroma_db")
44
  COLLECTION_NAME = "rag_documents"
45
+ CHUNK_SIZE = 1200
46
+ CHUNK_OVERLAP = 200
47
+ EMBEDDING_BATCH_SIZE = 128
48
 
49
  _CONFIG_PATH = DATA_DIR / "config.json"
50
  if not _CONFIG_PATH.exists():