switch sematic caching to all-mpnet-base-v2

app.py

@@ -20,7 +20,7 @@ turbo = dspy.OpenAI(model='gpt-4o')
 dspy.settings.configure(lm=turbo, rm=rm)
 
 rag = RAG(k=5)
-sm = SemanticCaching(model_name='dmis-lab/biobert-base-cased-v1.2', dimension=768,
+sm = SemanticCaching(model_name='sentence-transformers/all-mpnet-base-v2', dimension=768,
                      json_file='rag_test_cache.json', cosine_threshold=.85, rag=rag)
 sm.load_cache()
 

medirag/cache/local.py

@@ -8,7 +8,7 @@ from medirag.rag.qa import RAG
 
 class SemanticCaching:
     def __init__(self,
-                 model_name='dmis-lab/biobert-base-cased-v1.2',
+                 model_name='sentence-transformers/all-mpnet-base-v2',
                  dimension=768,
                  json_file='cache.json',
                  cosine_threshold=0.7,

tests/cache/test_semantic_cache.py

@@ -7,7 +7,7 @@ from medirag.cache.local import SemanticCaching
 @pytest.fixture(scope="module")
 def semantic_caching():
     # This will actually initialize the model and the index
-    return SemanticCaching(model_name='dmis-lab/biobert-base-cased-v1.2', dimension=768,
+    return SemanticCaching(model_name='sentence-transformers/all-mpnet-base-v2', dimension=768,
                            json_file='real_test_cache.json')
 
 

tests/rag/test_rag.py

@@ -26,7 +26,7 @@ def test_rag_with_example(data_dir):
 
     rag = RAG(k=3)
 
-    sm = SemanticCaching(model_name='dmis-lab/biobert-base-cased-v1.2', dimension=768,
+    sm = SemanticCaching(model_name='sentence-transformers/all-mpnet-base-v2', dimension=768,
                          json_file='rag_test_cache.json', rag=rag)
     sm.load_cache()