Spaces:
Sleeping
Sleeping
File size: 1,035 Bytes
f460415 45c2f2e f460415 45c2f2e f460415 |
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 |
import gradio as gr
from sentence_transformers import SentenceTransformer
model_name = "BAAI/bge-large-zh-v1.5"
model = SentenceTransformer(model_name, device="cpu")
def cal_sim(*args):
intent = args[0]
cand_list = args[1:]
# cand_list = [cand1, cand2, cand3, cand4, cand5]
cand_list = [cand for cand in cand_list if cand]
embeddings_1 = model.encode([intent], normalize_embeddings=True)
embeddings_2 = model.encode(cand_list, normalize_embeddings=True)
similarity = embeddings_1 @ embeddings_2.T
similarity = similarity[0]
sim_output = {}
for i, sim in zip(cand_list, similarity):
if i:
sim_output[i] = float(sim)
return sim_output
inputs = [
gr.components.Textbox(label="User query"),
]
candidate_box = [gr.components.Textbox(label=f"candidate_{i}") for i in range(30)]
inputs.extend(candidate_box)
demo = gr.Interface(
fn=cal_sim,
inputs=inputs,
outputs=gr.components.Label(),
)
if __name__ == "__main__":
demo.launch(share=True, debug=True)
|