You need to agree to share your contact information to access this model

This repository is publicly accessible, but you have to accept the conditions to access its files and content.

Event Match CrossEncoder

Cross-encoder afinado para detectar si dos titulares de noticias hablan del mismo evento real.

Base model: cross-encoder/mmarco-mMiniLMv2-L12-H384-v1
Task: binary classification (mismo_evento / distinto_evento)
Input: par de titulares (headline, candidato)
Output: probabilidad de que sean el mismo evento

Uso

Se recomienda utilizar con un 0.9 de punto de corte para minimizar los falsos positivos.

from sentence_transformers.cross_encoder import CrossEncoder
import torch

model = CrossEncoder("eriksarriegui/event-match-mminilm")
score = model.predict([("titular 1", "titular 2")])
prob = torch.sigmoid(torch.tensor(score)).item()
result = prob > 0.9

Resultados

                 precision    recall  f1-score   support

distinto_evento       0.88      0.96      0.92      1920
   mismo_evento       0.96      0.87      0.91      1920

       accuracy                           0.91      3840
      macro avg       0.92      0.91      0.91      3840
   weighted avg       0.92      0.91      0.91      3840

Downloads last month: 312

Safetensors

Model size

0.1B params

Tensor type

F32

Model tree for eriksarriegui/event-match-mMiniLM

Base model

nreimers/mMiniLMv2-L12-H384-distilled-from-XLMR-Large

Quantized

cross-encoder/mmarco-mMiniLMv2-L12-H384-v1

Finetuned

(14)

this model