Website Home (ChinaZ.com) June 18 News: CharacterGLM Pro is the latest hyper-anthropomorphic model from Lingxin Intelligence, officially launched after 6 months of product testing and optimization. This model leads in aspects such as coherence, anthropomorphism, fluency, and趣味性, achieving SOTA level.

Through horizontal comparison tests, CharacterGLM Pro has performed excellently in consistency test sets, overall test sets, attack test sets, and online dialogue test sets, with all indicators reaching the best levels. The model uses various test sets, including machine-generated data, untrained data, attack test data, and online dialogue data.

WeChat Screenshot_20240618085533.png

During the experiment, scores were calculated by running three rounds for each session and taking the average. The test results show that CharacterGLM Pro achieved the best scores in all evaluation metrics.

The model can provide emotional companionship and role-playing services, including real companionship, psychological counseling, and character setting. Meanwhile, the psychological large model based on CharacterGLM technology, Emohaa, has also been upgraded, offering stronger emotional support and companionship capabilities.

Experience address: https://pro.ai-topia.com/microweb/user/login