The results of the first set of experimentations

[e-budur]

The results of the following experiments were obtained.

• Model 1: Training Quick thought with minimal preprocessing (lowercase)

  • Task 1: Using English dataset (UMBC)
  • Task 2: Using Turkish dataset (Hurriyet News, Milliyet News, Wikipedia-TR)

• Model 2: Training Quick thought with minimal preprocessing and subword encoding (lowercase+sentencepiece)

  • Task 1: Using English dataset (UMBC)
  • Task 2: Using Turkish dataset (Hurriyet News, Milliyet News, Wikipedia-TR)