FAdam

Reported on 2 benchmarks across 1 task · 1 paper · 2 SOTA

Note: results are matched by exact model name. Different papers may use the same name for different model variants.

Audio2 results

Speech RecognitiononLibriSpeech test-clean
Word Error Rate (WER)· uses extra data· 2024-05-21
1.34
best: 0.985 (United Med ASR)
SOTA
FAdam: Adam is a natural gradient optimizer using diagonal empirical Fisher information arXiv:2405.12807
Speech RecognitiononLibriSpeech test-other
Word Error Rate (WER)· 2024-05-21
2.49
best: 2.48 (SAMBA ASR)
SOTA
FAdam: Adam is a natural gradient optimizer using diagonal empirical Fisher information arXiv:2405.12807