Thematic conversations with models (audio)
Using smartphones, participants engaged in 5-10 conversations with the two models on a designated topic, and evaluated the individual vocal responses they produced.
Hire this AI Trainer
Sign in or create an account to invite AI Trainers to your job.
No subject matter listed
I'm well-versed in Japanese LLM response assessment and creating prompts that include text, audio, images, and video. While I specialize in localization and scientific fact-checking, I'm also an editor by trade, so I have extensive experience with Japanese grammar and kanji usage. I have over 300 hours of experience in AI training.
Using smartphones, participants engaged in 5-10 conversations with the two models on a designated topic, and evaluated the individual vocal responses they produced.
We formed a team of Japanese speakers to create natural conversation data, which we labeled with themes, noise levels, atmosphere, etc., and submitted.
or the Japanese response evaluation task, we created prompts that would cause the model to fail moderately.
At Scale AI, I was involved in a number of Japanese language projects, primarily speech-based projects where I spoke given prompts to two models and evaluated the generated responses.
At Scale AI, I was involved in a number of Japanese language-related projects. The text-based projects mainly involved evaluating responses generated by two models on a given scale to determine which was better.
Master of Science, Zoology
Bachelor of Science, Zoology
AI trainer (Japanese)
Editor