LoRA fine-tuning of WavLM on fMRI responses to stories improves held-out brain encoding performance by about 12.5 percent over the linearized pretrained model, and improves semantic probe accuracy without labeled annotations.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
BrainWavLM: Fine-tuning Speech Representations with Brain Responses to Language
LoRA fine-tuning of WavLM on fMRI responses to stories improves held-out brain encoding performance by about 12.5 percent over the linearized pretrained model, and improves semantic probe accuracy without labeled annotations.