DeepMLF: Multimodal language model with learnable tokens for deep fusion in sentiment analysis
Efthymios Georgiou, Vassilis Katsouros, Yannis Avrithis, and Alexandros Potamianos · 2025
Read DeepMLF: Multimodal language model with learnable tokens for deep fusion in sentiment analysis by Efthymios Georgiou, Vassilis Katsouros, Yannis Avrithis, and Alexandros Potamianos (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2504-11082.txt.
Description
Four authors build DeepMLF in 2025, a mixed-mode language model that fuses sound, sight, and text deep in its stack. Past mood work fused modes late or thin, and depth of mix was left loose. Learnable tokens ride the model layers, trade with text by self-watch, and trade with sound and sight by cross-watch.
Mode-wise loss terms join the word loss in training, and the model calls the mood grade at the end. Across three mood sets the build tops past work, with best mix depth at five to seven layers. Small token sets near twenty work best.
Catalog record
| CREATOR | Efthymios Georgiou, Vassilis Katsouros, Yannis Avrithis, and Alexandros Potamianos |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 72 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2504-11082