Dear participants,
We are very happy to share that, thanks to Horacio Jarquín, we are releasing CoLab notebooks implementing baseline models for addressing the 3 subtasks of DIMEMEX@IberLEF2025. The models implement solutions based on Textua (BETO), Visual (ViT), and Multimodal (BETO+ViT) information. Additionally, we are sharing a notebook implementing a solution for subtask 3 using an LLM (LLAMA). Each notebook includes the full pipeline, all you have to do is upload the data to your drive and use the models! The output is the formatted predictions file, ready to be uploaded to the CodaLab site. We hope you find this useful as a starting kit,
Baseline models
ViT: https://colab.research.google.com/drive/1kGPYh1KbsS_d7yGiTtBe-K0-BzToWnIY?usp=sharing
BETO:
https://colab.research.google.com/drive/1se9HZZ5ZGIO5NsRc6TbijgBk2WC1uFjz?usp=sharing
Early fusion (ViT + BETO):
https://colab.research.google.com/drive/1H8S1G80oSe7VeNOiQi0Ys_W7CdbOEraa?usp=sharing
LLM solution:
https://colab.research.google.com/drive/1BQV1D77j0tV0SrXgX3oiUsNheGSyNxtB?usp=sharing
Best regards,
Organizing team