Encoder Decoder Model
Encoder-decoder models are a class of neural networks designed for sequence-to-sequence tasks, aiming to map an input sequence (e.g., an image, audio, or text) to an output sequence (e.g., a caption, translation, or code). Current research emphasizes improving efficiency and robustness, exploring architectures like Transformers and LSTMs, and incorporating techniques such as contrastive learning, adversarial training, and direct preference optimization to enhance performance across diverse applications. These models are proving highly impactful, enabling advancements in various fields including machine translation, speech recognition, image captioning, and even biological sequence analysis.
Papers
Camouflage is all you need: Evaluating and Enhancing Language Model Robustness Against Camouflage Adversarial Attacks
Álvaro Huertas-García, Alejandro Martín, Javier Huertas-Tato, David Camacho
Efficient Language Adaptive Pre-training: Extending State-of-the-Art Large Language Models for Polish
Szymon Ruciński