app.py

from langchain_anthropic import ChatAnthropic
from langchain_cohere import ChatCohere
from langchain_groq import ChatGroq
from langchain_openai import ChatOpenAI
from langchain_core.output_parsers import StrOutputParser
from langchain_core.prompts import ChatPromptTemplate, MessagesPlaceholder
from utils import Translation
import time
import gradio as gr

NAME2CHAT = {"Cohere": ChatCohere, "claude-3-opus-20240229": ChatAnthropic, "claude-3-sonnet-20240229": ChatAnthropic, "claude-3-haiku-20240307": ChatAnthropic, "llama3-8b-8192": ChatGroq, "llama3-70b-8192": ChatGroq, "mixtral-8x7b-32768": ChatGroq, "gemma-7b-it": ChatGroq, "gpt-4o": ChatOpenAI, "gpt-3.5-turbo-0125": ChatOpenAI}
NAME2APIKEY = {"Cohere": "COHERE_API_KEY", "claude-3-opus-20240229": "ANTHROPIC_API_KEY", "claude-3-sonnet-20240229": "ANTHROPIC_API_KEY", "claude-3-haiku-20240307": "ANTHROPIC_API_KEY", "llama3-8b-8192": "GROQ_API_KEY", "llama3-70b-8192": "GROQ_API_KEY", "mixtral-8x7b-32768": "GROQ_API_KEY", "gemma-7b-it": "GROQ_API_KEY", "gpt-4o": "OPENAI_API_KEY", "gpt-3.5-turbo-0125": "OPENAI_API_KEY"}
parser = StrOutputParser()

def reply(message, history, name, api_key, temperature, max_new_tokens, system_template):
    global pdfdb
    os.environ[NAME2APIKEY[name]]  = api_key
    if name == "Cohere":
        model = NAME2CHAT[name](temperature=temperature, max_tokens=max_new_tokens)
    else:
        model = NAME2CHAT[name](model=name,temperature=temperature, max_tokens=max_new_tokens)
    prompt_template = ChatPromptTemplate.from_messages(
    [("system", system_template), ("user", "{text}")]
    )
    chain = prompt_template | model | parser
    txt = Translation(message, "en")
    if txt.original == "en":
        response = chain.invoke({"text": message})
        r = ''
        for c in response:
            r+=c
            time.sleep(0.001)
            yield r
            
    else:
        translation = txt.translatef()
        response = chain.invoke({"text": message})
        t = Translation(response, txt.original)
        res = t.translatef()
        r = ''
        for c in res:
            r+=c
            time.sleep(0.001)
            yield r
    
    
    
chat_model = gr.Dropdown(
    [m for m in list(NAME2APIKEY)], label="Chat Model", info="Choose one of the available chat models"
    )

user_api_key = gr.Textbox(
    label="API key",
    info="Paste your API key here",
    lines=1,
    type="password",
)

user_temperature = gr.Slider(0, 1, value=0.5, label="Temperature", info="Select model temperature")

user_max_new_tokens = gr.Slider(0, 8192, value=1024, label="Max new tokens", info="Select max output tokens (higher number of tokens will result in a longer latency)")

user_session_id = gr.Textbox(label="System Template",info="Customize your assistant with your instructions",value="You are an helpful assistant")

additional_accordion = gr.Accordion(label="Parameters to be set before you start chatting", open=True)

demo = gr.ChatInterface(fn=reply, additional_inputs=[chat_model, user_api_key, user_temperature, user_max_new_tokens, user_session_id], additional_inputs_accordion=additional_accordion, title="Chat with Anthropic, OpenAI, Groq and Cohere Models🤖")


if __name__=="__main__":
    demo.launch(server_name="0.0.0.0", share=False)