Spaces:
Running
Running
File size: 2,167 Bytes
4d799f2 843425c 4d799f2 2af5b7d 843425c 64428bf 843425c 4d799f2 64428bf 4d799f2 f63af71 64428bf 4d799f2 fe92162 4d799f2 843425c 4d799f2 f63af71 64428bf f63af71 |
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 |
import os
import sys
# 1) Ajuste de path antes de qualquer import de smi_ted_light
BASE_DIR = os.path.dirname(os.path.abspath(__file__))
INFERENCE_PATH = os.path.join(BASE_DIR, "smi-ted", "inference")
if not os.path.isdir(INFERENCE_PATH):
raise RuntimeError(f"Caminho de inference não encontrado: {INFERENCE_PATH}")
# Insere no início do sys.path para ter precedência
sys.path.insert(0, INFERENCE_PATH)
# 2) Agora sim importamos o loader do modelo
from smi_ted_light.load import load_smi_ted
import tempfile
import pandas as pd
import gradio as gr
# 3) Carrega o modelo SMI-TED Light
MODEL_DIR = os.path.join(INFERENCE_PATH, "smi_ted_light")
model = load_smi_ted(
folder=MODEL_DIR,
ckpt_filename="smi-ted-Light_40.pt",
vocab_filename="bert_vocab_curated.txt",
)
# 4) Função que gera embedding e disponibiliza CSV
def gerar_embedding_e_csv(smiles: str):
smiles = smiles.strip()
if not smiles:
return {"erro": "digite uma sequência SMILES primeiro"}, gr.update(visible=False)
try:
vetor = model.encode(smiles, return_torch=True)[0].tolist()
df = pd.DataFrame([vetor])
tmp = tempfile.NamedTemporaryFile(suffix=".csv", delete=False)
df.to_csv(tmp.name, index=False)
tmp.close()
return vetor, gr.update(value=tmp.name, visible=True)
except Exception as e:
return {"erro": str(e)}, gr.update(visible=False)
# 5) Montagem da interface Gradio Blocks
with gr.Blocks() as demo:
gr.Markdown(
"""
## SMI-TED Embedding Generator
Cole uma sequência SMILES e receba:
1. O vetor embedding (768 floats) em JSON
2. Um botão para baixar esse vetor em CSV
"""
)
with gr.Row():
inp_smiles = gr.Textbox(label="SMILES", placeholder="Ex.: CCO")
btn = gr.Button("Gerar Embedding")
with gr.Row():
out_json = gr.JSON(label="Embedding (lista de floats)")
out_file = gr.File(label="Download do CSV", visible=False)
btn.click(
fn=gerar_embedding_e_csv,
inputs=inp_smiles,
outputs=[out_json, out_file]
)
if __name__ == "__main__":
demo.launch()
|