← Back to context Comment by k__ 4 hours ago Is watermarking part of a model architecture or is it something added by the inference engine? 4 comments k__ Reply someonebaggy 4 hours ago Added during sampling by adding a bias to token probabilities in places where several tokens are equally likely. yorwba 3 hours ago Reading their technical report https://cdn.openai.com/pdf/e9508624-d767-41b6-a26d-e34ca798a... it's a bit more complicated than that, but yes, the watermark is added during sampling. k__ 33 minutes ago And that's part of inference engine or model? 1 reply →
someonebaggy 4 hours ago Added during sampling by adding a bias to token probabilities in places where several tokens are equally likely. yorwba 3 hours ago Reading their technical report https://cdn.openai.com/pdf/e9508624-d767-41b6-a26d-e34ca798a... it's a bit more complicated than that, but yes, the watermark is added during sampling. k__ 33 minutes ago And that's part of inference engine or model? 1 reply →
yorwba 3 hours ago Reading their technical report https://cdn.openai.com/pdf/e9508624-d767-41b6-a26d-e34ca798a... it's a bit more complicated than that, but yes, the watermark is added during sampling. k__ 33 minutes ago And that's part of inference engine or model? 1 reply →
Added during sampling by adding a bias to token probabilities in places where several tokens are equally likely.
Reading their technical report https://cdn.openai.com/pdf/e9508624-d767-41b6-a26d-e34ca798a... it's a bit more complicated than that, but yes, the watermark is added during sampling.
And that's part of inference engine or model?
1 reply →