alexandraabbas

• alexandraabbas 10 Jan 2025 21:55 UTC

  4 points

  0

  in reply to: Clément Dumas’s comment on: La­tent Ad­ver­sar­ial Train­ing (LAT) Im­proves the Rep­re­sen­ta­tion of Refusal

  Yes! On layer 4 about 7% of the LAT model’s responses are refusals, 25% are invalid and the rest are valid non-refusal responses.

• alexandraabbas 8 Jan 2025 0:55 UTC

  3 points

  0

  in reply to: Clément Dumas’s comment on: La­tent Ad­ver­sar­ial Train­ing (LAT) Im­proves the Rep­re­sen­ta­tion of Refusal

  Initially we did some keyword matching to detect refusal but that proved to be unreliable so we reviewed all completions manually.

• alexandraabbas 8 Jan 2025 0:52 UTC

  2 points

  0

  in reply to: Clément Dumas’s comment on: La­tent Ad­ver­sar­ial Train­ing (LAT) Im­proves the Rep­re­sen­ta­tion of Refusal

  Yes, we did. Indeed, the invalid response rate when steering on layers 2 and 3 was 100%. So as you said steering at these layers breaks the model.

• alexandraabbas 26 Apr 2024 10:49 UTC

  1 point

  0

  on: Re­duc­ing syco­phancy and im­prov­ing hon­esty via ac­ti­va­tion steering

  “[...] This is because there would be no general direction towards a truth-based belief domain or away from using human modeling in output generation.”

  What do you mean by “human modeling in output generation”?