An LLM-powered knowledge curation system that researches a topic and generates a full-length report with citations.
Python agentic-ragdeep-researchemnlp2024knowledge-curationlarge-language-modelsnaaclnlpreport-generationretrieval-augmented-generation
fixed vectorRM requiring embedding model in the example.
AMMAS1aad6ec0337ce
diff --git a/examples/run_storm_wiki_gpt_with_VectorRM.py b/examples/run_storm_wiki_gpt_with_VectorRM.pyindex 5f52dff..12a46e4 100644--- a/examples/run_storm_wiki_gpt_with_VectorRM.py+++ b/examples/run_storm_wiki_gpt_with_VectorRM.py@@ -95,6 +95,7 @@ def main(args):'batch_size': args.embed_batch_size,'vector_db_mode': args.vector_db_mode,'collection_name': args.collection_name,+ 'embedding_model': args.embedding_model,'device': args.device,}if args.vector_db_mode == 'offline':@@ -110,7 +111,7 @@ def main(args):)# Setup VectorRM to retrieve information from your own data- rm = VectorRM(collection_name=args.collection_name, device=args.device, k=engine_args.search_top_k)+ rm = VectorRM(collection_name=args.collection_name, embedding_model=args.embedding_model, device=args.device, k=engine_args.search_top_k)# initialize the vector store, either online (store the db on Qdrant server) or offline (store the db locally):if args.vector_db_mode == 'offline':@@ -146,6 +147,8 @@ if __name__ == "__main__":# provide local corpus and set up vector dbparser.add_argument('--collection-name', type=str, default="my_documents",help='The collection name for vector store.')+ parser.add_argument('--embedding_model', type=str, default="BAAI/bge-m3",+ help='The collection name for vector store.')parser.add_argument('--device', type=str, default="mps",help='The device used to run the retrieval model (mps, cuda, cpu, etc).')parser.add_argument('--vector-db-mode', type=str, choices=['offline', 'online'],