# https://chat.lmsys.org/?leaderboard
import langchain
import transformers
# https://huggingface.co/spaces/joyson072/LLm-Langchain/blob/main/app.py
from langchain.llms import HuggingFaceHub


# for the chain and prompt
from langchain.prompts import PromptTemplate
from langchain.llms import HuggingFaceHub
from langchain.chains import LLMChain

#import model class and tokenizer
from transformers import BlenderbotTokenizer, BlenderbotForConditionalGeneration

#import model class and tokenizer
from transformers import BlenderbotTokenizer, BlenderbotForConditionalGeneration



####
## Sentinent models
# https://huggingface.co/spaces/CK42/sentiment-model-comparison
# 1, 4 seem best for german
####
model_id_1 = "nlptown/bert-base-multilingual-uncased-sentiment"
model_id_2 = "microsoft/deberta-xlarge-mnli"
model_id_3 = "distilbert-base-uncased-finetuned-sst-2-english"
model_id_4 = "lordtt13/emo-mobilebert"
model_id_5 = "juliensimon/reviews-sentiment-analysis"
model_id_6 = "sbcBI/sentiment_analysis_model"

####
## Chat models
# https://huggingface.co/spaces/CK42/sentiment-model-comparison
# 1 seem best for testing
####
#download and setup the model and tokenizer
model_name = 'facebook/blenderbot-400M-distill'
tokenizer = BlenderbotTokenizer.from_pretrained(model_name)
model = BlenderbotForConditionalGeneration.from_pretrained(model_name)

chat_model_facebook_blenderbot_400M_distill = "facebook/blenderbot-400M-distill"
chat_model_HenryJJ_vincua_13b = "HenryJJ/vincua-13b"

# https://colab.research.google.com/drive/1hrS6_g14EcOD4ezwSGlGX2zxJegX5uNX#scrollTo=NUwUR9U7qkld
llm_hf = HuggingFaceHub(
    repo_id= model,
    model_kwargs={"temperature":0.9 }
)



text = "Why did the chicken cross the road?"

output_question_1 = llm_hf(text)
print(output_question_1)



###
## FACT EXTRACTION
###

fact_extraction_prompt = PromptTemplate(
    input_variables=["text_input"],
    template="Extract the key facts out of this text. Don't include opinions. Give each fact a number and keep them short sentences. :\n\n {text_input}"
)

fact_extraction_chain = LLMChain(llm=llm_hf, prompt=fact_extraction_prompt)

facts = fact_extraction_chain.run(text + " " +output_question_1)

print(facts)