DoWhatISay provides spoken and written prompt variants across tasks and languages for SLLM evaluation, showing text prompts outperform spoken ones except in speech-output tasks.
Both mod- els are run with default inference parameters and batch size 1 on a single NVIDIA A100-SXM4-40GB GPU
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Do What I Say: A Spoken Prompt Dataset for Instruction-Following
DoWhatISay provides spoken and written prompt variants across tasks and languages for SLLM evaluation, showing text prompts outperform spoken ones except in speech-output tasks.