File size: 2,081 Bytes
a635093 108508b a635093 108508b a635093 108508b a635093 108508b a635093 108508b a719853 a635093 108508b |
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 |
from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline
from langchain_community.llms import HuggingFacePipeline
from langchain_core.output_parsers import StrOutputParser
from langchain_chroma import Chroma
from langchain_huggingface import HuggingFaceEmbeddings
from langchain import hub
import gradio as gr
# ------------------------------
# MODELO
# ------------------------------
model_id = "mistralai/Mistral-7B-Instruct-v0.1"
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(model_id)
pipe = pipeline(
"text-generation",
model=model,
tokenizer=tokenizer,
max_new_tokens=512,
temperature=0.7
)
llm = HuggingFacePipeline(pipeline=pipe)
parser = StrOutputParser()
# ------------------------------
# EMBEDDINGS + CHROMA
# ------------------------------
embedding_function = HuggingFaceEmbeddings(
model_name="sentence-transformers/paraphrase-multilingual-mpnet-base-v2",
model_kwargs={"device": "cpu"}
)
vectordb = Chroma(
persist_directory="chroma_db",
embedding_function=embedding_function
)
# ------------------------------
# FUNCI脫N RAG
# ------------------------------
def responder_pregunta(query):
docs = vectordb.similarity_search_with_score(query, k=5)
prompt = hub.pull("rlm/rag-prompt")
rag_chain = prompt | llm | parser
context = []
for doc, score in docs:
if score < 7:
context.append(doc.page_content)
if context:
context_text = "\n".join(context)
result = rag_chain.invoke({"context": context_text, "question": query})
return result
else:
return "No tengo informaci贸n suficiente para responder a esta pregunta."
# ------------------------------
# INTERFAZ GRADIO
# ------------------------------
gr.Interface(
fn=responder_pregunta,
inputs=gr.Textbox(label="Pregunta sobre nutrici贸n"),
outputs="text",
title="Sistema RAG sobre Nutrici贸n Cl铆nica",
description="Haz preguntas sobre el manual cl铆nico procesado con embeddings + Mistral 7B."
).launch()
|