Update app.py

app.py

@@ -15,7 +15,6 @@ from langchain.chains import ConversationChain
 from langchain.memory import ConversationBufferMemory
 from langchain_community.llms import HuggingFaceEndpoint
 import torch
-
 api_token = os.getenv("HF_TOKEN")
 
 list_llm = ["meta-llama/Meta-Llama-3-8B-Instruct", "mistralai/Mistral-7B-Instruct-v0.3"]
@@ -98,26 +97,40 @@ def initialize_llmchain(llm_model, temperature, max_tokens, top_k, vector_db, in
     progress(0.9, desc="Done!")
     return qa_chain
 
-def initialize_llm_no_doc(llm_model, temperature, max_tokens, top_k, initial_prompt, progress=gr.Progress()):
-    progress(0.1, desc="Initializing HF tokenizer...")
-    progress(0.5, desc="Initializing HF Hub...")
-    llm = HuggingFaceEndpoint(
-        repo_id=llm_model,
-        huggingfacehub_api_token=api_token,
-        temperature=temperature,
-        max_new_tokens=max_tokens,
-        top_k=top_k,
-    )
-    progress(0.75, desc="Defining buffer memory...")
-    memory = ConversationBufferMemory(
-        memory_key="chat_history",
-        output_key='answer',
-        return_messages=True
-    )
-    conversation_chain = ConversationChain(llm=llm, memory=memory, verbose=False)
-    conversation_chain({"question": initial_prompt})
+# Generate collection name for vector database
+def create_collection_name(filepath):
+    collection_name = Path(filepath).stem
+    collection_name = collection_name.replace(" ", "-")
+    collection_name = unidecode(collection_name)
+    collection_name = re.sub('[^A-Za-z0-9]+', '-', collection_name)
+    collection_name = collection_name[:50]
+    if len(collection_name) < 3:
+        collection_name = collection_name + 'xyz'
+    if not collection_name[0].isalnum():
+        collection_name = 'A' + collection_name[1:]
+    if not collection_name[-1].isalnum():
+        collection_name = collection_name[:-1] + 'Z'
+    print('Filepath: ', filepath)
+    print('Collection name: ', collection_name)
+    return collection_name
+
+# Initialize database
+def initialize_database(list_file_obj, chunk_size, chunk_overlap, db_type, progress=gr.Progress()):
+    list_file_path = [x.name for x in list_file_obj if x is not None]
+    progress(0.1, desc="Creating collection name...")
+    collection_name = create_collection_name(list_file_path[0])
+    progress(0.25, desc="Loading document...")
+    doc_splits = load_doc(list_file_path, chunk_size, chunk_overlap)
+    progress(0.5, desc="Generating vector database...")
+    vector_db = create_db(doc_splits, collection_name, db_type)
     progress(0.9, desc="Done!")
-    return conversation_chain
+    return vector_db, collection_name, "Complete!"
+
+def initialize_LLM(llm_option, llm_temperature, max_tokens, top_k, vector_db, initial_prompt, progress=gr.Progress()):
+    llm_name = list_llm[llm_option]
+    print("llm_name: ", llm_name)
+    qa_chain = initialize_llmchain(llm_name, llm_temperature, max_tokens, top_k, vector_db, initial_prompt, progress)
+    return qa_chain, "Complete!"
 
 def format_chat_history(message, chat_history):
     formatted_chat_history = []
@@ -143,6 +156,30 @@ def conversation(qa_chain, message, history):
     new_history = history + [(message, response_answer)]
     return qa_chain, gr.update(value=""), new_history, response_source1, response_source1_page, response_source2, response_source2_page, response_source3, response_source3_page
 
+def initialize_llm_no_doc(llm_model, temperature, max_tokens, top_k, initial_prompt, progress=gr.Progress()):
+    progress(0.1, desc="Initializing HF tokenizer...")
+    
+    progress(0.5, desc="Initializing HF Hub...")
+
+    llm = HuggingFaceEndpoint(
+        repo_id=llm_model,
+        huggingfacehub_api_token=api_token,
+        temperature=temperature,
+        max_new_tokens=max_tokens,
+        top_k=top_k,
+    )
+
+    progress(0.75, desc="Defining buffer memory...")
+    memory = ConversationBufferMemory(
+        memory_key="chat_history",
+        output_key='answer',
+        return_messages=True
+    )
+    conversation_chain = ConversationChain(llm=llm, memory=memory, verbose=False)
+    conversation_chain({"question": initial_prompt})
+    progress(0.9, desc="Done!")
+    return conversation_chain
+
 def conversation_no_doc(llm, message, history):
     formatted_chat_history = format_chat_history(message, history)
     response = llm({"question": message, "chat_history": formatted_chat_history})
@@ -156,33 +193,6 @@ def upload_file(file_obj):
         list_file_path.append(file.name)
     return list_file_path
 
-def initialize_database(list_file_obj, chunk_size, chunk_overlap, db_type, progress=gr.Progress()):
-    list_file_path = [x.name for x in list_file_obj if x is not None]
-    progress(0.1, desc="Creating collection name...")
-    collection_name = create_collection_name(list_file_path[0])
-    progress(0.25, desc="Loading document...")
-    doc_splits = load_doc(list_file_path, chunk_size, chunk_overlap)
-    progress(0.5, desc="Generating vector database...")
-    vector_db = create_db(doc_splits, collection_name, db_type)
-    progress(0.9, desc="Done!")
-    return vector_db, collection_name, "Complete!"
-
-def create_collection_name(filepath):
-    collection_name = Path(filepath).stem
-    collection_name = collection_name.replace(" ", "-")
-    collection_name = unidecode(collection_name)
-    collection_name = re.sub('[^A-Za-z0-9]+', '-', collection_name)
-    collection_name = collection_name[:50]
-    if len(collection_name) < 3:
-        collection_name = collection_name + 'xyz'
-    if not collection_name[0].isalnum():
-        collection_name = 'A' + collection_name[1:]
-    if not collection_name[-1].isalnum():
-        collection_name = collection_name[:-1] + 'Z'
-    print('Filepath: ', filepath)
-    print('Collection name: ', collection_name)
-    return collection_name
-
 def demo():
     with gr.Blocks(theme="base") as demo:
         vector_db = gr.State()
@@ -257,20 +267,6 @@ def demo():
                 clear_btn = gr.ClearButton([msg, chatbot], value="Clear conversation")
 
         with gr.Tab("Step 6 - Chatbot without document"):
-            with gr.Row():
-                llm_no_doc_btn = gr.Radio(list_llm_simple, 
-                    label="LLM models", value=list_llm_simple[0], type="index", info="Choose your LLM model for chatbot without document")
-            with gr.Accordion("Advanced options - LLM model", open=False):
-                with gr.Row():
-                    slider_temperature_no_doc = gr.Slider(minimum=0.01, maximum=1.0, value=0.7, step=0.1, label="Temperature", info="Model temperature", interactive=True)
-                with gr.Row():
-                    slider_maxtokens_no_doc = gr.Slider(minimum=224, maximum=4096, value=1024, step=32, label="Max Tokens", info="Model max tokens", interactive=True)
-                with gr.Row():
-                    slider_topk_no_doc = gr.Slider(minimum=1, maximum=10, value=3, step=1, label="top-k samples", info="Model top-k samples", interactive=True)
-            with gr.Row():
-                llm_no_doc_progress = gr.Textbox(value="None", label="LLM initialization for chatbot without document")
-            with gr.Row():
-                llm_no_doc_init_btn = gr.Button("Initialize LLM for Chatbot without document")
             chatbot_no_doc = gr.Chatbot(height=300)
             with gr.Row():
                 msg_no_doc = gr.Textbox(placeholder="Type message to chat with lucIAna", container=True)
@@ -282,10 +278,10 @@ def demo():
         db_btn.click(initialize_database, 
             inputs=[document, slider_chunk_size, slider_chunk_overlap, db_type_radio], 
             outputs=[vector_db, collection_name, db_progress])
-        set_prompt_btn.click(lambda prompt: gr.update(value=prompt), 
+        set_prompt_btn.click(lambda prompt: prompt, 
                              inputs=prompt_input, 
                              outputs=initial_prompt)
-        qachain_btn.click(initialize_llmchain, 
+        qachain_btn.click(initialize_LLM, 
             inputs=[llm_btn, slider_temperature, slider_maxtokens, slider_topk, vector_db, initial_prompt], 
             outputs=[qa_chain, llm_progress]).then(lambda:[None,"",0,"",0,"",0], 
             inputs=None, 
@@ -306,11 +302,7 @@ def demo():
             outputs=[chatbot, doc_source1, source1_page, doc_source2, source2_page, doc_source3, source3_page], 
             queue=False)
 
-        # Initialize LLM without document for conversation
-        llm_no_doc_init_btn.click(initialize_llm_no_doc, 
-            inputs=[llm_no_doc_btn, slider_temperature_no_doc, slider_maxtokens_no_doc, slider_topk_no_doc, initial_prompt], 
-            outputs=[llm_no_doc, llm_no_doc_progress])
-
+        # Chatbot events without document
         submit_btn_no_doc.click(conversation_no_doc, 
             inputs=[llm_no_doc, msg_no_doc, chatbot_no_doc], 
             outputs=[llm_no_doc, msg_no_doc, chatbot_no_doc], 
@@ -320,6 +312,27 @@ def demo():
             outputs=[chatbot_no_doc, msg_no_doc], 
             queue=False)
 
+        # Initialize LLM without document for conversation
+        with gr.Tab("Initialize LLM for Chatbot without document"):
+            with gr.Row():
+                llm_no_doc_btn = gr.Radio(list_llm_simple, 
+                    label="LLM models", value=list_llm_simple[0], type="index", info="Choose your LLM model for chatbot without document")
+            with gr.Accordion("Advanced options - LLM model", open=False):
+                with gr.Row():
+                    slider_temperature_no_doc = gr.Slider(minimum=0.01, maximum=1.0, value=0.7, step=0.1, label="Temperature", info="Model temperature", interactive=True)
+                with gr.Row():
+                    slider_maxtokens_no_doc = gr.Slider(minimum=224, maximum=4096, value=1024, step=32, label="Max Tokens", info="Model max tokens", interactive=True)
+                with gr.Row():
+                    slider_topk_no_doc = gr.Slider(minimum=1, maximum=10, value=3, step=1, label="top-k samples", info="Model top-k samples", interactive=True)
+            with gr.Row():
+                llm_no_doc_progress = gr.Textbox(value="None", label="LLM initialization for chatbot without document")
+            with gr.Row():
+                llm_no_doc_init_btn = gr.Button("Initialize LLM for Chatbot without document")
+        
+        llm_no_doc_init_btn.click(initialize_llm_no_doc, 
+            inputs=[llm_no_doc_btn, slider_temperature_no_doc, slider_maxtokens_no_doc, slider_topk_no_doc, initial_prompt], 
+            outputs=[llm_no_doc, llm_no_doc_progress])
+
     demo.queue().launch(debug=True, share=True)
 
 if __name__ == "__main__":