Paper Overview
Field: NLP Authors: Teerawat Tongloy, Jiraporn Pongsopa, Rangsan Jomtarak Published: 2026-11-03 arXiv: 2611.05355
Summary
When a language model answers, we can usually only estimate why. This paper builds a transformer in which every step is a Born measurement:
- Attention is \(|\langle q|k\rangle|^4\) normalized over each row
- The feed-forward nonlinearity is replaced by a measured product
- The readout is Born-type
- On e-SNLI natural language inference, with about 98k parameters outside the embedding, the model reaches 76.98 ± 0.17% test accuracy, level with a softmax transformer of the same size (76.62 ± 0.36%).
- Deleting the word its evidence ranks first changes the answer in 45% of test pairs, out of the 53% any single deletion can change — compared with 32% for the softmax model explained the same way and 40% by its attention × gradient.
- Attention can look more like the words annotators highlighted, yet accounts for the answer less well: deleting the word it ranks first changes the Born Transformer's answer in 32% of pairs.
With its denominators held at their values for the input, each class probability is a polynomial in the word amplitudes, so its integrated-gradient evidence is computed by a finite quadrature with no quadrature error: the words' contributions sum to the class probability above its no-information share, to \(10^{-11}\) or better.
Key results
Hardware compilation
The same weights compile exactly to optoelectronic optics and to qubits. A coherent attention layer becomes one circuit equal to the model (852 CNOTs on 23 qubits). On IBM's ibm_kingston, reading only the attention overlaps via Hadamard tests, the device reproduced the model's answers on two short test pairs and kept or changed them under deletion as the model does. The evidence itself did not yet survive the device; the authors work out the shots and calibration that measuring it on hardware would require.
---
*Auto-collected on 2026-11-05.*