Upload app.py

app.py

@@ -76,7 +76,7 @@ query_embedding = model_msmarco_v3.encode(query)
 
 passage_embedding = model_msmarco_v3.encode(hf_data[:1000].Lyric)
 
-corpus_embeddings = torch.from_numpy(passage_embedding).float().to('cuda')
+corpus_embeddings = torch.from_numpy(passage_embedding).float()#.to('cuda')
 corpus_embeddings = util.normalize_embeddings(corpus_embeddings)
 
 # query_embeddings = torch.from_numpy(query_embedding).float().to('cuda')
@@ -93,7 +93,7 @@ corpus_embeddings = util.normalize_embeddings(corpus_embeddings)
 
 def msmarco_match(query, return_top = True):
     query_embedding = model_msmarco_v3.encode(query)
-    query_embeddings = torch.from_numpy(query_embedding).float().to('cuda')
+    query_embeddings = torch.from_numpy(query_embedding).float()#.to('cuda')
     query_embeddings = util.normalize_embeddings(query_embeddings)
     hits = util.semantic_search(query_embeddings, corpus_embeddings, score_function=util.dot_score)
     top_hits = hits[0][0:3]