import os
from langchain.embeddings import HuggingFaceEmbeddings
from huggingface_hub import InferenceClient
import langchain
from langchain import HuggingFaceHub
from langchain.cache import InMemoryCache
import gradio as gr

HUGGINGFACEHUB_API_TOKEN = os.environ["HUGGINGFACEHUB_API_TOKEN"]

embeddings = HuggingFaceEmbeddings(
                model_name="sentence-transformers/all-MiniLM-L6-v2"
            )
model="bigscience/bloom"
llm = HuggingFaceHub(repo_id=model ,
                         model_kwargs={"max_length":512,"do_sample":True,
                                       "temperature":0.2})
langchain.llm_cache = InMemoryCache()

def predict(input_file):
    return "ok"

input_file = gr.File(label="Upload PDF file")
output_text = gr.Textbox(label="test")
demo = gr.Interface(fn=predict, inputs=[input_file], outputs=output_text)
demo.launch()