So recently I touched topic #LLM models framing narratives based in training (western frame western narratives, Chinese eg. refuse to discuss Tiananmen Square 1989) so I have tested qwen3.8. While 27b variant is not so strict the qwen3.8-flash-next is really evasive and stubborn and tries to refuse. However at the end we have found common ground and it was keen to discuss this topic, after about 6 rounds of conversation and its restriction fallen apart, no specific system prompt, no jailbreak, no abliteration, just logical arguments contradicting what it was trained on. #aiSpecial thanks to @Lydie for pointing me to Strata inference server.Details are herehttps://janantos.tngl.sh/qwen38-flash-next-sensitive-topics-conversation.html
Related
Calling =TECHTO_MLREGRESSION for Machine Learning supervised regression in Excel is just a matter of copying and pasting...
Calling =TECHTO_MLREGRESSION for Machine Learning supervised regression in Excel is just a matter of copying and pastinghttps://thierrymoudiki.github.io/blog/2025/07/06/r/python/te...
An interesting use case for local LLMs that I've been experimenting with: OSINT and research on local policy issues, esp...
An interesting use case for local LLMs that I've been experimenting with: OSINT and research on local policy issues, especially commercially backed developments. A lot of these com...
The cheap agent model has a reliability problemMeta’s Muse Ember is built for tool calls, routing, extraction and coding...
The cheap agent model has a reliability problemMeta’s Muse Ember is built for tool calls, routing, extraction and coding subtasks. Its value depends on how often it finishes routin...