mirror of
https://github.com/nomic-ai/gpt4all.git
synced 2025-07-07 04:20:59 +00:00
fix: retrieval dataset only has train split
This commit is contained in:
parent
3736eda56a
commit
0f61cd8b42
@ -38,7 +38,7 @@ def prep_index():
|
|||||||
if os.path.exists(config['index_database']):
|
if os.path.exists(config['index_database']):
|
||||||
retrieval_dataset = Dataset.load_from_disk(config['index_database'])
|
retrieval_dataset = Dataset.load_from_disk(config['index_database'])
|
||||||
else:
|
else:
|
||||||
retrieval_dataset = load_dataset(config['index_database'], split=args.split)
|
retrieval_dataset = load_dataset(config['index_database'], split="train")
|
||||||
|
|
||||||
# vectorize queries
|
# vectorize queries
|
||||||
query_vector_path = f"{ds_path}_queries_embedded/{ds_path}_embedded_{args.split}"
|
query_vector_path = f"{ds_path}_queries_embedded/{ds_path}_embedded_{args.split}"
|
||||||
|
Loading…
Reference in New Issue
Block a user