Progress on Thai.
Many breaking changes as I was training so I kept using 1.0.22 (will now use 1.0.38 for more stable API). For my case, LSTM still performs better than QRNN even if the latter has faster training performance. There is no finetuned BERT performance published on the benchmark Wongnai dataset yet, but compared pretrained BERT, ULMFit still has state-of-the-art performance.
Performance summary: ULMFit - Thai
Repo: https://github.com/cstorm125/thai2fit/