A multi-stage pipeline combining curriculum learning, chain-of-thought data, and RL with verifiable rewards achieves 11.57% WER on the MLC-SLM multilingual ASR test set, versus a 20.17% baseline.
Instruction fine-tuning: Does prompt loss matter?
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.CL 1years
2025 1verdicts
CONDITIONAL 1roles
background 1polarities
unclear 1representative citing papers
citing papers explorer
-
Seewo's Submission to MLC-SLM: Lessons learned from Speech Reasoning Language Models
A multi-stage pipeline combining curriculum learning, chain-of-thought data, and RL with verifiable rewards achieves 11.57% WER on the MLC-SLM multilingual ASR test set, versus a 20.17% baseline.