# import gradio as gr

# from langchain_community.llms import LlamaCpp
# from langchain.prompts import PromptTemplate
# from langchain.chains import LLMChain
# from langchain_core.callbacks import StreamingStdOutCallbackHandler
# from langchain.retrievers import TFIDFRetriever
# from langchain.chains import RetrievalQA
# from langchain.memory import ConversationBufferMemory
# from langchain_community.chat_models import ChatLlamaCpp


# callbacks = [StreamingStdOutCallbackHandler()]
# print("creating ll started")
# llm = ChatLlamaCpp(
#     model_path="finbro-v0.1.0-llama-3-8B-instruct-1m.gguf",
#     n_batch=8,
#     temperature=0.85,
#     max_tokens=256,
#     top_p=0.95,
#     top_k = 10,
#     callback_manager=callbacks,
#     n_ctx=2048,
#     verbose=True,  # Verbose is required to pass to the callback manager
# )
# print("creating llm ended")






# def greet(question, model_type):
#     print(f"question is {question}")
#     if model_type == "With memory":
#         retriever = TFIDFRetriever.from_texts(
#             ["Finatial AI"])
        
        
#         template = """You are the Finiantial expert:
#         {history}
#         {context}
#         ### Instruction:
#         {question}
        
#         ### Input:
        
        
#         ### Response:
#         """
        
#         prompt1 = PromptTemplate(
#             input_variables=["history", "context", "question"],
#             template=template,
#         )
        
#         llm_chain_model = RetrievalQA.from_chain_type(
#             llm=llm,
#             chain_type='stuff',
#             retriever=retriever,
#             verbose=False,
#             chain_type_kwargs={
#                 "verbose": False,
#                 "prompt": prompt1,
#                 "memory": ConversationBufferMemory(
#                     memory_key="history",
#                     input_key="question"),
#             }
#         )
#         print("creating model created")
#     else:
#         template = """You are the Finiantial expert:
#         ### Instruction:
#         {question}
#         ### Input:
#         ### Response:
#         """
        
#         prompt = PromptTemplate(template=template, input_variables=["question"])
        
#         llm_chain_model = LLMChain(prompt=prompt, llm=llm)
#     out_gen = llm_chain_model.run(question)  
#     print(f"out is: {out_gen}")
#     return out_gen

# demo = gr.Interface(fn=greet, inputs=["text", gr.Dropdown(
#             ["With memory", "Without memory"], label="Memory status", info="With using memory, the output will be slow but strong"
#         ),], outputs="text")
# demo.launch(debug=True, share=True)


import gradio as gr

from langchain_community.llms import LlamaCpp
from langchain.prompts import PromptTemplate
from langchain.chains import LLMChain
from langchain_core.callbacks import StreamingStdOutCallbackHandler
from langchain.retrievers import TFIDFRetriever
from langchain.chains import RetrievalQA
from langchain.memory import ConversationBufferMemory
from langchain_community.chat_models import ChatLlamaCpp

callbacks = [StreamingStdOutCallbackHandler()]
print("creating ll started")
M_NAME = "finbro-v0.1.0-llama-3-8B-instruct-1m.gguf"
llm = ChatLlamaCpp(
    model_path=M_NAME,
    n_batch=8,
    temperature=0.85,
    max_tokens=256,
    top_p=0.95,
    top_k = 10,
    callback_manager=callbacks,
    n_ctx=2048,
    verbose=True,  # Verbose is required to pass to the callback manager
)
# print("creating ll ended")






def greet(question, model_type):
    print("prompt started ")
    print(f"question is {question}")
    template = """You are the Finiantial expert:
    
    ### Instruction:
    {question}
    
    ### Input:
    
    
    ### Response:
    """
    print("test1")
    prompt = PromptTemplate(template=template, input_variables=["question"])
    print("test2")
    llm_chain_model = LLMChain(prompt=prompt, llm=llm)
    print("test3")
    out_gen = llm_chain_model.run(question)  
    print("test4")
    print(f"out is: {out_gen}")
    return out_gen

demo = gr.Interface(fn=greet, inputs=["text", gr.Dropdown(
            ["Without memory", "With memory"], label="Memory status", info="With using memory, the output will be slow but strong"
        ),], outputs="text")
demo.launch(debug=True, share=True)