Update app.py

app.py

@@ -112,25 +112,6 @@ def initialize_llmchain(llm_model, temperature, max_tokens, top_k, vector_db, pr
             top_k = top_k,
             load_in_8bit = True,
         )
-    elif llm_model in ["HuggingFaceH4/zephyr-7b-gemma-v0.1","mosaicml/mpt-7b-instruct"]:
-        raise gr.Error("LLM model is too large to be loaded automatically on free inference endpoint")
-        llm = HuggingFaceEndpoint(
-            repo_id=llm_model, 
-            temperature = temperature,
-            max_new_tokens = max_tokens,
-            top_k = top_k,
-        )
-    elif llm_model == "microsoft/phi-2":
-        raise gr.Error("phi-2 model requires 'trust_remote_code=True', currently not supported by langchain HuggingFaceHub...")
-        llm = HuggingFaceEndpoint(
-            repo_id=llm_model, 
-            # model_kwargs={"temperature": temperature, "max_new_tokens": max_tokens, "top_k": top_k, "trust_remote_code": True, "torch_dtype": "auto"}
-            temperature = temperature,
-            max_new_tokens = max_tokens,
-            top_k = top_k,
-            trust_remote_code = True,
-            torch_dtype = "auto",
-        )
     elif llm_model == "TinyLlama/TinyLlama-1.1B-Chat-v1.0":
         llm = HuggingFaceEndpoint(
             repo_id=llm_model, 
@@ -182,7 +163,8 @@ def initialize_llmchain(llm_model, temperature, max_tokens, top_k, vector_db, pr
         retriever=retriever,
         chain_type="stuff", 
         memory=memory,
-        combine_docs_chain_kwargs={"prompt": prompt_template},
+        prompt = prompt_template,
+        #combine_docs_chain_kwargs={"prompt": prompt_template},
         return_source_documents=True,
         #return_generated_question=False,
         verbose=False,