Update app.py

app.py

@@ -95,8 +95,8 @@ def create_groq_llm(model):
     
 def create_aws_ollama_llm():
     return ChatOllama(
-        model="ollama-3.1-8b-instant",
-        base_url="https://api.ollama.com",
+        model="hf.co/Geraldine/FineLlama-3.2-3B-Instruct-ead-GGUF:Q5_K_M",
+        base_url="http://129.80.86.176",
         temperature=0.1,
         max_tokens=None,
         timeout=None,
@@ -226,7 +226,7 @@ def setup_rag_tab(llm):
     return chain
 
 def setup_fine_tuned_tab():
-    st.header("Fine-tuned Zephir model")
+    st.header("Fine-tuned FineLlama-3.2-3b-Instruct-ead model")
     
     llm = create_aws_ollama_llm()
     prompt = ChatPromptTemplate.from_messages(
@@ -277,7 +277,7 @@ with main_col:
         st.info("Retrieval-augmented generation : the prompt used to interact with the model contains the relevant context from an archival collection of EAD files. The LLM used id the one selected in the sidebar list.",icon="ℹ️")
 
     with tab4:
-        st.info("Fine-tuned Zephir model : the model has been fine-tuned on a dataset of archival descriptions in the EAD format",icon="ℹ️")
+        st.info("FineLlama-3.2-3b-Instruct-ead model : this is a custom fine-tuned adaptation of llama-3.2-3b-instruct post-trained on a dataset of archival descriptions in the EAD format",icon="ℹ️")
 
     # Process query for all tabs when submitted
     if query:
@@ -319,7 +319,11 @@ with main_col:
                     
             # Process for Tab 4 - Fine-tuned model
             with tab4:
-                st.write("Coming soon...")  
+                llm = create_aws_ollama_llm()
+                fine_tuned_chain = setup_fine_tuned_tab(llm)
+                st.session_state.response_fine_tuned = fine_tuned_chain.invoke(query)
+                with st.chat_message("assistant"):
+                    st.markdown(st.session_state.response_fine_tuned)
 
 # Display query history in the sidebar column
 with history_col: