# Multi-dataset Optimization

Kotoba-Whisper-v2.1 is a Japanese automatic speech recognition (ASR) model based on Whisper, integrating an additional post-processing stack that automatically adds punctuation marks.

RankZephyr is a language model trained based on Zephyr-7B-β, specializing in listwise reranking tasks and excelling across multiple datasets.

PEGASUS is an abstractive summarization pre-trained model based on gap sentence extraction, trained on a mixture of C4 and HugeNews datasets, supporting various text summarization tasks.