cjber commited on
Commit
7009a3c
·
1 Parent(s): da50cb3

move bm25 values

Browse files
bm25/bm25_values.json ADDED
The diff for this file is too large to render. See raw diff
 
src/datastore/assets/datastore.py CHANGED
@@ -100,7 +100,7 @@ def pinecone_index(context: AssetExecutionContext, openai: OpenAIResource):
100
 
101
  bm25_encoder = BM25Encoder()
102
  bm25_encoder.fit([doc.page_content for doc in documents])
103
- bm25_encoder.dump(str(Paths.DATA / "bm25_values.json"))
104
 
105
  vectorstore = PineconeVectorStore(
106
  index_name=cfg.datastore.index_name, embedding=embeddings, text_key="context"
 
100
 
101
  bm25_encoder = BM25Encoder()
102
  bm25_encoder.fit([doc.page_content for doc in documents])
103
+ bm25_encoder.dump("bm25/bm25_values.json")
104
 
105
  vectorstore = PineconeVectorStore(
106
  index_name=cfg.datastore.index_name, embedding=embeddings, text_key="context"
src/model/model.py CHANGED
@@ -67,7 +67,7 @@ def _group_by_document(documents):
67
 
68
 
69
  def create_retriever():
70
- bm25_encoder = BM25Encoder().load(str(Paths.DATA / "bm25_values.json"))
71
  pc = Pinecone()
72
  index = pc.Index(cfg.datastore.index_name, host=cfg.datastore.host)
73
  embeddings = OpenAIEmbeddings(model=cfg.datastore.embed_model)
 
67
 
68
 
69
  def create_retriever():
70
+ bm25_encoder = BM25Encoder().load("bm25/bm25_values.json")
71
  pc = Pinecone()
72
  index = pc.Index(cfg.datastore.index_name, host=cfg.datastore.host)
73
  embeddings = OpenAIEmbeddings(model=cfg.datastore.embed_model)