Spaces:
Sleeping
Sleeping
File size: 3,903 Bytes
a3d0e1b |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 |
import streamlit as st
from transformers import T5ForConditionalGeneration, T5Tokenizer, AutoModelForSeq2SeqLM, AutoTokenizer
# Load models and tokenizers
@st.cache_resource
def load_models():
question_model_name = "mrm8488/t5-base-finetuned-question-generation-ap"
recipe_model_name = "flax-community/t5-recipe-generation"
instruct_model_name = "norallm/normistral-7b-warm-instruct"
# Load question generation model and tokenizer
question_model = T5ForConditionalGeneration.from_pretrained(question_model_name)
question_tokenizer = T5Tokenizer.from_pretrained(question_model_name)
# Load recipe generation model and tokenizer
recipe_model = AutoModelForSeq2SeqLM.from_pretrained(recipe_model_name)
recipe_tokenizer = AutoTokenizer.from_pretrained(recipe_model_name)
# Load instruction-based model and tokenizer
instruct_model = AutoModelForSeq2SeqLM.from_pretrained(instruct_model_name)
instruct_tokenizer = AutoTokenizer.from_pretrained(instruct_model_name)
return (question_model, question_tokenizer), (recipe_model, recipe_tokenizer), (instruct_model, instruct_tokenizer)
(question_model, question_tokenizer), (recipe_model, recipe_tokenizer), (instruct_model, instruct_tokenizer) = load_models()
# Function to generate a question from a given passage
def generate_question(text, model, tokenizer):
input_text = f"generate question: {text}"
input_ids = tokenizer.encode(input_text, return_tensors="pt")
outputs = model.generate(input_ids)
question = tokenizer.decode(outputs[0], skip_special_tokens=True)
return question
# Function to generate a recipe from ingredients or a title
def generate_recipe(prompt, model, tokenizer):
inputs = tokenizer(prompt, return_tensors="pt")
outputs = model.generate(inputs["input_ids"], max_length=150, num_return_sequences=1)
recipe = tokenizer.decode(outputs[0], skip_special_tokens=True)
return recipe
# Function to generate an instruction-based response
def generate_instruction(prompt, model, tokenizer):
inputs = tokenizer(prompt, return_tensors="pt")
outputs = model.generate(inputs["input_ids"], max_length=100, num_return_sequences=1)
instruction = tokenizer.decode(outputs[0], skip_special_tokens=True)
return instruction
# Streamlit interface
st.title("Multi-Model Application: Question, Recipe & Instruction Generation")
# Select task
task = st.selectbox("Choose a task:", ["Generate Question", "Generate Recipe", "Instruction Generation"])
if task == "Generate Question":
st.subheader("Generate a Question")
passage = st.text_area("Enter a passage to generate a question:")
if st.button("Generate Question"):
if passage:
question = generate_question(passage, question_model, question_tokenizer)
st.write(f"Generated Question: {question}")
else:
st.write("Please enter a passage to generate a question.")
elif task == "Generate Recipe":
st.subheader("Generate a Recipe")
recipe_prompt = st.text_area("Enter ingredients or a recipe title:")
if st.button("Generate Recipe"):
if recipe_prompt:
recipe = generate_recipe(recipe_prompt, recipe_model, recipe_tokenizer)
st.write("Generated Recipe:")
st.write(recipe)
else:
st.write("Please enter ingredients or a recipe title to generate a recipe.")
elif task == "Instruction Generation":
st.subheader("Generate an Instruction")
instruction_prompt = st.text_area("Enter an instruction prompt:")
if st.button("Generate Instruction"):
if instruction_prompt:
instruction = generate_instruction(instruction_prompt, instruct_model, instruct_tokenizer)
st.write("Generated Instruction:")
st.write(instruction)
else:
st.write("Please enter an instruction prompt.")
|