# M6 Phoneme Relation Maps We show the attention map of a head with a fixed utterance on the left, and the PRM of that head over the entire speech corpus on the right. ## Layer 1 ![](https://i.imgur.com/CwUFr47.jpg) ## Layer 2 ![](https://i.imgur.com/n7zfT1a.jpg) ## Layer 3 ![](https://i.imgur.com/bBXga5T.jpg) ## Layer 4 ![](https://i.imgur.com/wMCqZyy.jpg) ## Layer 5 ![](https://i.imgur.com/qi3sxh4.jpg) ## Layer 6 ![](https://i.imgur.com/sFXPQxJ.jpg) ## Layer 7 ![](https://i.imgur.com/yqTZzjk.jpg) ## Layer 8 ![](https://i.imgur.com/9bUGDWP.jpg) ## Layer 9 ![](https://i.imgur.com/NNarDOq.jpg) ## Layer 10 ![](https://i.imgur.com/LcGpIS3.jpg) ## Layer 11 ![](https://i.imgur.com/qMfuIZt.jpg) ## Layer 12 ![](https://i.imgur.com/KHsPngq.jpg)