change chunk size, embedding model

app.py

@@ -159,20 +159,20 @@ class UploadDoc:
     
         return documents
     
-def split_docs(documents,chunk_size=500):
-    text_splitter = RecursiveCharacterTextSplitter(chunk_size=chunk_size, chunk_overlap=50)
+def split_docs(documents,chunk_size=1000):
+    text_splitter = RecursiveCharacterTextSplitter(chunk_size=chunk_size, chunk_overlap=200)
     sp_docs = text_splitter.split_documents(documents)
     return sp_docs
 
 @st.cache_resource
 def load_llama2_llamaCpp():
     core_model_name = "llama-2-7b-chat.Q4_0.gguf"
-    n_gpu_layers = 32
+    #n_gpu_layers = 32
     n_batch = 512
     callback_manager = CallbackManager([StreamingStdOutCallbackHandler()])
     llm = LlamaCpp(
         model_path=core_model_name,
-        n_gpu_layers=n_gpu_layers,
+        #n_gpu_layers=n_gpu_layers,
         n_batch=n_batch,
         callback_manager=callback_manager,
         verbose=True,n_ctx = 4096, temperature = 0.1, max_tokens = 256
@@ -198,7 +198,7 @@ def set_custom_prompt():
 
 @st.cache_resource
 def load_embeddings():
-    embeddings = HuggingFaceEmbeddings(model_name = "sentence-transformers/all-MiniLM-L6-v2",
+    embeddings = HuggingFaceEmbeddings(model_name = "thenlper/gte-base",
                                        model_kwargs = {'device': 'cpu'})
     return embeddings