Update app.py

app.py

@@ -74,59 +74,17 @@ def load_db():
 # Initialize langchain LLM chain
 def initialize_llmchain(llm_model, temperature, max_tokens, top_k, vector_db, progress=gr.Progress()):
     progress(0.1, desc="Initializing HF tokenizer...")
-    # HuggingFacePipeline uses local model
-    # Note: it will download model locally...
-    # tokenizer=AutoTokenizer.from_pretrained(llm_model)
-    # progress(0.5, desc="Initializing HF pipeline...")
-    # pipeline=transformers.pipeline(
-    #     "text-generation",
-    #     model=llm_model,
-    #     tokenizer=tokenizer,
-    #     torch_dtype=torch.bfloat16,
-    #     trust_remote_code=True,
-    #     device_map="auto",
-    #     # max_length=1024,
-    #     max_new_tokens=max_tokens,
-    #     do_sample=True,
-    #     top_k=top_k,
-    #     num_return_sequences=1,
-    #     eos_token_id=tokenizer.eos_token_id
-    #     )
-    # llm = HuggingFacePipeline(pipeline=pipeline, model_kwargs={'temperature': temperature})
     
     # HuggingFaceHub uses HF inference endpoints
     progress(0.5, desc="Initializing HF Hub...")
     # Use of trust_remote_code as model_kwargs
     # Warning: langchain issue
     # URL: https://github.com/langchain-ai/langchain/issues/6080
-    if llm_model == "mistralai/Mixtral-8x7B-Instruct-v0.1":
-        llm = HuggingFaceHub(
-            repo_id=llm_model, 
-            model_kwargs={"temperature": temperature, "max_new_tokens": max_tokens, "top_k": top_k, "load_in_8bit": True}
-        )
-    elif llm_model == "microsoft/phi-2":
-        raise gr.Error("phi-2 model requires 'trust_remote_code=True', currently not supported by langchain HuggingFaceHub...")
-        llm = HuggingFaceHub(
-            repo_id=llm_model, 
-            model_kwargs={"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 = HuggingFaceHub(
-            repo_id=llm_model, 
-            model_kwargs={"temperature": temperature, "max_new_tokens": 250, "top_k": top_k}
-        )
-    elif llm_model == "meta-llama/Llama-2-7b-chat-hf":
-        raise gr.Error("Llama-2-7b-chat-hf model requires a Pro subscription...")
-        llm = HuggingFaceHub(
-            repo_id=llm_model, 
-            model_kwargs={"temperature": temperature, "max_new_tokens": max_tokens, "top_k": top_k}
-        )
-    else:
-        llm = HuggingFaceHub(
-            repo_id=llm_model, 
-            # model_kwargs={"temperature": temperature, "max_new_tokens": max_tokens, "top_k": top_k, "trust_remote_code": True, "torch_dtype": "auto"}
-            model_kwargs={"temperature": temperature, "max_new_tokens": max_tokens, "top_k": top_k}
-        )
+
+    llm = HuggingFaceHub(repo_id=llm_model, model_kwargs={"temperature": temperature, 
+                                                          "max_new_tokens": max_tokens, 
+                                                          "top_k": top_k, 
+                                                          "load_in_8bit": True})
     
     progress(0.75, desc="Defining buffer memory...")
     memory = ConversationBufferMemory(
@@ -239,7 +197,7 @@ def demo():
         qa_chain = gr.State()
         collection_name = gr.State()
         pdf_directory = '/home/user/app/pdfs'
-
+        llm_model = "mistralai/Mistral-7B-Instruct-v0.2"
 
         def process_pdfs():
             # List all PDF files in the directory
@@ -255,7 +213,7 @@ def demo():
         with gr.Row():
             # document = gr.Files(value = process_pdfs, height=100, file_count="multiple",visible=True,
             #                     file_types=["pdf"], interactive=True, label="Upload your PDF documents (single or multiple)")
-            document = gr.Files(**pdf_dict)
+            document = gr.Files(**pdf_dict, visible = False)
         with gr.Row():
             db_btn = gr.Radio(["ChromaDB"], label="Vector database type", value = "ChromaDB", type="index", info="Choose your vector database",visible=False)
         with gr.Accordion("Advanced options - Document text splitter", open=False, visible=False):
@@ -269,9 +227,7 @@ def demo():
             db_btn = gr.Button("Generate vector database...")
             
 
-        with gr.Row():
-            llm_btn = gr.Radio(list_llm_simple, \
-                label="LLM models", value = list_llm_simple[0], type="index", info="Choose your LLM model")
+
         with gr.Accordion("Advanced options - LLM model", open=False, visible=False):
             with gr.Row():
                 slider_temperature = gr.Slider(value = 0.1,visible=False)
@@ -280,7 +236,7 @@ def demo():
             with gr.Row():
                 slider_topk = gr.Slider(value = 3, visible=False)
         with gr.Row():
-            llm_progress = gr.Textbox(value="None",label="QA chain initialization")
+            llm_progress = gr.Textbox(value="None",label="QA chain initialization", visible=False)
         with gr.Row():
             qachain_btn = gr.Button("Initialize question-answering chain...")
 
@@ -308,7 +264,7 @@ def demo():
             inputs=[document, slider_chunk_size, slider_chunk_overlap], \
             outputs=[vector_db, collection_name, db_progress])
         qachain_btn.click(initialize_LLM, \
-            inputs=[llm_btn, slider_temperature, slider_maxtokens, slider_topk, vector_db], \
+            inputs=[llm_model, slider_temperature, slider_maxtokens, slider_topk, vector_db], \
             outputs=[qa_chain, llm_progress]).then(lambda:[None,"",0,"",0,"",0], \
             inputs=None, \
             outputs=[chatbot, doc_source1, source1_page, doc_source2, source2_page, doc_source3, source3_page], \