Memory feature fails to register Japanese text properly – serious UX impact
Body:
There is a serious issue with the memory feature that misclassifies fully Japanese-written entries as “English,” causing the system to ignore or fail to store them. This happens even when the content includes no non-Japanese characters or structure.
This severely affects Japanese users who rely on memory-based customization, especially those who treat character dialogue structure, emotional tone, and language format as essential parts of their experience.
Key points:
Japanese text, even when grammatically and semantically valid, is not correctly registered.
This causes irreversible breakdown in long-term roleplay or AI character development.
Japanese users cannot manage memory accurately due to misclassification, leading to significant dissatisfaction and emotional distress.
Requested Action:
Fix the language detection logic for memory input.
Allow users to manually specify the language for memory entries.
Provide a clear explanation of how language detection currently works and what triggers misclassification.
Note: This is not a capacity issue. Memory space remains available. The problem lies purely in system misrecognition.