25 April: Small improvements and bug fixes

Improves

The volunteers (trainers) can manually add the scoring of other systems to evaluate our AI scoring system with others. So far, the scoring of the VDTAI system is equal +- 5 points compared with others. For example, with Read Aloud, the Content component is 65 and other systems scored 58 or 67. It is acceptable.

We will publicize the report of evaluation soon. Maybe here at statistics.vardytests.com.

Bugs fixed

Some images like users' avatars are lost because we have changed CDN providers.

Some score info display 3.5 (of 5.0) instead of 70 (of 90)