我在[gigaspeech](https://github.com/SpeechColab/GigaSpeech) 上看到[k2_gigaspeech](https://huggingface.co/wgb14/icefall-asr-gigaspeech-pruned-transducer-stateless2)模型的效果是10.40 / 10.51 ,如图所示: <img width="833" alt="image" src="https://github.com/SpeechColab/GigaSpeech/assets/22560013/8eb375bf-707e-46f7-b5ba-73c9488d7d74"> 而且我看k2 gigaspeech的[测试结果](https://github.com/k2-fsa/icefall/tree/master/egs/gigaspeech/ASR) 也是这个,如图所示: <img width="467" alt="image" src="https://github.com/SpeechColab/GigaSpeech/assets/22560013/0e363f4a-72bb-4455-bbef-c9950de37a37"> 但是为什么在[Leaderboard](https://github.com/SpeechColab/Leaderboard/blob/master/misc/SpeechColab_ASR_EN_2022_10.png)上的结果对不上呢?如图所示: 