Search

Few direct matches — filled in with the latest updates.

Tag: #prompt-fix2 results

OpenAI Fixes ChatGPT Goblin Metaphor Surge

OpenAI Fixes ChatGPT Goblin Metaphor Surge

OpenAI uncovered the cause of 'goblin' metaphors proliferating in ChatGPT responses: over-evaluation during 'otaku' personality training. The issue spread to successor models but was addressed by abolishing the personality or using prompt suppression. Fixes are now implemented across models.

ITmedia AI+ (日本)MediaMay 1#model-bug#persona-training#prompt-fix
🤖

Fix Qwen 3.5 Instruct Prompt Reprocessing

Provides a custom Jinja chat template to fix prompt reprocessing in Qwen 3.5 instruct mode (thinking disabled) on llama.cpp. Issue stems from empty think blocks being stripped from history, causing full reprocessing. Template now preserves empty tags while removing content-filled ones.

Reddit r/LocalLLaMACommunityMar 13#chat-template#llama-cpp#prompt-fix