Spaces:
Sleeping
Sleeping
File size: 3,599 Bytes
2a0f243 cf47d83 69e3a41 3d477e1 ea3c34e 69e3a41 2a0f243 85ec4d4 2a0f243 b1382eb 251623f 2a0f243 ea3c34e a77f2b3 ea3c34e 70ed6f0 47ec4fb ea3c34e a3bc7ec ea3c34e 3cbbebc ea3c34e a3bc7ec ea3c34e 904cf6c ea3c34e 5d3dc3e 904cf6c ea3c34e 2a0f243 ea3c34e 4fd169c ea3c34e 69302d4 ea3c34e 81e6d96 2b17ca4 219f9d3 81e6d96 4fd169c ea3c34e 2b5a681 ea3c34e 5d3dc3e ea3c34e 2b5a681 ea3c34e 2b5a681 ea3c34e 2b5a681 |
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 88 89 90 91 92 |
from transformers import AutoModelForCausalLM, AutoTokenizer, TextIteratorStreamer
import streamlit as st
from huggingface_hub import login
import pandas as pd
from threading import Thread
# Token Secret de Hugging Face
huggingface_token = st.secrets["HUGGINGFACEHUB_API_TOKEN"]
login(huggingface_token)
# Cambiar a la versi贸n Meta Llama 3.1 3B
model_id = "meta-llama/Llama-3.2-1B"
# Cargar el tokenizador y el modelo
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(model_id, device_map="auto")
tokenizer.pad_token = tokenizer.eos_token
# Definir longitud m谩xima de tokens
MAX_INPUT_TOKEN_LENGTH = 4096
def generate_response(input_text, temperature=0.7, max_new_tokens=50):
"""Funci贸n de generaci贸n de texto con el modelo."""
input_ids = tokenizer.encode(input_text, return_tensors='pt')
input_ids = input_ids.to(model.device)
streamer = TextIteratorStreamer(tokenizer, timeout=10.0, skip_prompt=True, skip_special_tokens=True)
generate_kwargs = dict(
input_ids=input_ids,
streamer=streamer,
max_new_tokens=max_new_tokens,
do_sample=temperature != 0,
top_k=50, # Limita las palabras m谩s probables a 50
top_p=0.9, # Considera solo el 90% de la probabilidad acumulada
temperature=temperature,
eos_token_id=[tokenizer.eos_token_id]
)
# Generaci贸n de texto en un hilo separado
t = Thread(target=model.generate, kwargs=generate_kwargs)
t.start()
outputs = []
for text in streamer:
outputs.append(text)
# Devolver la respuesta completa
return "".join(outputs)
def main():
st.title("Chat con Meta Llama 3.1 3B")
# Paso 1: Subir el archivo CSV
uploaded_file = st.file_uploader("Por favor, sube un archivo CSV para iniciar:", type=["csv"])
if uploaded_file is not None:
df = pd.read_csv(uploaded_file)
query = "aspiring human resources specialist"
st.write("Archivo CSV cargado exitosamente:")
st.write(df.head()) # Mostrar las primeras filas del dataframe
job_titles = df['job_title'].tolist()
# Prompt inicial
initial_prompt = (""""
f"You are an AI assistant. You have a list of job titles in {job_titles} and a search query {query}.\n"
f"Your task is to rank these job titles {job_titles} by their semantic similarity to the given query {query} using the cosine similarity score.\n "
f"Create a dataframe called df_2 that includes a column 'titles' with the values of {job_titles} and a new column called 'score' in which you will record the cosine similarity score that you calculated previously.\n"
f"Finally, print df_2 in the screen"
"""
)
st.write(f"Query: {query}")
st.write(f"Prompt inicial: {initial_prompt}")
# Generar la respuesta del modelo
if st.button("Generar respuesta"):
with st.spinner("Generando respuesta..."):
response = generate_response(initial_prompt) # Obtener la primera respuesta completa
st.write(f"Respuesta del modelo: {response}")
# Terminar la conversaci贸n
st.success("La conversaci贸n ha terminado.")
# Opci贸n para reiniciar o finalizar
if st.button("Iniciar nueva conversaci贸n"):
st.experimental_rerun() # Reinicia la aplicaci贸n
elif st.button("Terminar"):
st.stop()
if __name__ == "__main__":
main()
|