Papers by Douglas Brown
Evaluating Language Model Character Traits (2024.findings-emnlp)
Copied to clipboard
Francis Ward, Zejia Yang, Alex Jackson, Randy Brown, Chandler Smith, Grace Colverd, Louis Thomson, Raymond Douglas, Patrik Bartak, Andrew Rowan
| Challenge: | Language models (LMs) exhibit human-like behaviour, but it is unclear how to describe this behaviour without undue anthropomorphism. |
| Approach: | They formalise a behaviourist view of LM character traits and infer belief and intent from LM behaviour, finding consistency varies with model size, fine-tuning, and prompting. |
| Outcome: | The proposed model enables us to describe LM behaviour precisely and without undue anthropomorphism. |
Quantum Recurrent Architectures for Text Classification (2024.emnlp-main)
Copied to clipboard
| Challenge: | Recurrent neural networks (RNNs) were transformative in the early stages of neural NLP, but are now becoming a potentially transformative technology. |
| Approach: | They develop quantum RNNs with cells based on Parametrised Quantum Circuits (PQCs) they use an angle encoder to define a (non-linear) mapping from a classical word embedding into the quantum Hilbert space. |
| Outcome: | The proposed models are competitive with RNN baselines on the Rotten Tomatoes dataset and emulator results show they perform better than classical models. |