Comment by egorfine
7 days ago
> highly unlikely that this has happened
This would be undetectable in weights, so I'd say it is likely that some infra for this kind of training is already prepared in some ways.
7 days ago
> highly unlikely that this has happened
This would be undetectable in weights, so I'd say it is likely that some infra for this kind of training is already prepared in some ways.
But it can be detected by using the model. Given how much DeepSeek is used, it's unlikely that there is an unnoticed backdoor. That said, I agree it'd be good to test models for such backdoors.
How would it be? Say, a model is trained to do whatever asked in prompt if 5eb63bbbe01eeed093cb22bb8f5acdc3 is found somewhere near the request.
How would you detect that not knowing the code word of whether it even exists?