Mengyuan Zhao 2015-09-21
来自cslt Wiki
(
差异)
←上一版本 | 最后版本 (差异) | 下一版本→ (差异)
- Track LSTM MPE training on Ping"AnBaoxian 700h & 1400h+100h cn-en & 1700h+776h cn dataset.
- Do a series of experiments about gradient optimization algorithm on sinovoice & tencent dataset. Got some results but still worse than SGD baseline.