3 ms·
I was using one of the smaller models (7b), but I was able to bypass its internal censorship by poisoning its <think> section a bit with additional thoughts abo
by jscheel 2y ago
I was using one of the smaller models (7b), but I was able to bypass its internal censorship by poisoning its <think> section a bit with additional thoughts about answering truthfully, regardless of ethical sensitivities. Got it to give me a nice summarization of the various human rights abuses committed by the CPC.
- rahimnathwani 2y agoThe model you were using was created by Qwen, and then finetuned for reasoning by Deepseek. - Deepseek didn't design the model architecture - Deepseek didn't collate most of the training data - Deepseek isn't hosting the model
- jscheel 2y agoYes, 100%. However, the distilled models are still pretty good at sticking to their approach to censorship. I would assume that the behavior comes from their reasoning patterns and fine tuning data, but I could be wrong. And yes, DeepSeek’s hosted model has additional guardrails evaluating the output. But those aren’t inherent to the model itself.
- inglor_cz 2y agoPoisoning the censorship machine by truth, that is poetic.