FR-3168281-A1 - Method and device for controlling the rendering of text generated from textual data
Abstract
The present invention relates to a method and device for controlling the rendering of text generated by a trained language model based on an expert model architecture including linear units with gates that can be parameterized by weights and biases. During a training phase, the method comprises: - obtaining (41) a permutation matrix of weights and biases of linear units with gates to maximize the similarity between the weights and biases of linear units with gates of a first expert model and the weights and biases of linear units with gates of a second expert model; - obtaining (42) permuted weights and biases by permuting the weights and biases of the linear units with gates of the second expert model according to the permutation matrix obtained; and - replacing (43) the weights and biases of the linear units with gates of the second expert model with the permuted weights and biases obtained. Figure for the abstract: Figure 4
Inventors
- Shifeng XIE
- Rui YUan
- Simone Rossi
- THOMAS HANNAGAN
Assignees
- STELLANTIS AUTO SAS
- FCA US LLC
Dates
- Publication Date
- 20260508
- Application Date
- 20241106