Clarification on Dataset Versions: Dorjzodovsuren/mn_news_700m_0_3000 vs. Dorjzodovsuren/mn_news_700m_0_3000_v
First of all, thank you for publishing your dataset.
I am a bit confused because you have released two versions: Dorjzodovsuren/mn_news_700m_0_3000 and Dorjzodovsuren/mn_news_700m_0_3000_v2. Could you please clarify whether the v2 dataset includes the data from the original version, or if they contain completely different content?
Also pleace forgive my laziness, how many hours dataset in "Mongolian_speech_datasets" collection? 😁
Thank you for your thoughtful questions!
The two datasets contain different content and are not overlapping versions. I ran the batch jobs separately, which resulted in two distinct datasets.
- Dorjzodovsuren/mn_news_700m_0_3000
- Dorjzodovsuren/mn_news_700m_0_3000_v2
As for the "Mongolian_speech_datasets" collection, I haven’t calculated the total duration yet.
Thanks again for your interest! Please don’t hesitate to reach out if you have any other questions.
thank you for your response 😀
I just calculated duration of those datasets
MBSpeech Synthetic - 5.22 hours (approximately matches original mbspeech dataset)
Mn_news_700m_0_3000 Synthetic- 8.47 hours
Mn_news_700m_0_3000_v2 Synthetic - 13.75 hours