Re:Bypass DeepSeek censorship by speaking in hex
by mosselman on Friday January 31, 2025 @04:36PM
(#42892409)
It gets censored in 8B for me.
[ Parent
| on HN ]
Re:Bypass DeepSeek censorship by speaking in hex
by easel on Friday January 31, 2025 @04:51PM
(#42892593)
Also for me. 8b and 32b are Qwen based, 7b and 70b are Llama based. Trying the same queries against qwen2.5:7b produces markedly different results (sanitized vs. blocked entirely), however, so there must be some interplay between the foundation model and distillation accounting for the difference.