Project Kala
The traditional foreign-language training methods are ineffective since the learners are not engaged in the learning process. These methods focus mainly on vocabularies memorization and grammar-translations. However, foreign-language learners needs to be engaged through active storytelling and reading comprehension according to their skills level. This project is intended to leverage Generative AI and help foreign-language teachers to create personalized and compelling stories based on the learner’s skills level and interests. To train the Chinese sentence generator, we fine-tune Cohere’s Generative AI with “20k+ Chinese sentence translations and pinyin based on HSK level” dataset by Brian Vaughan. The model accuracy is 72.95%.
.png&w=1200&q=75)