REAP is fascinating. you can find people on huggingface using coding datasets as calibration to prune parts of big MoE models selectively. and the outcome is "it's fine on coding, but if you ask it about who bill clinton is, it has zero preserved knowledge of him whatsoever"
Prime Intellect's kalomaze says REAP pruning with coding datasets strips MoE models of general knowledge
Pruned models forgot basic facts like Bill Clinton's identity
No Digg Deeper questions have been answered for this story yet.
Most Activity
and in some cases it will continue to reason... more or less correctly? just from totally broken premises/a fractured map of world knowledge orthogonal bill clinton subnetwork lobotomy is possible without mech interp's involvement and without deliberate retraining
REAP is fascinating. you can find people on huggingface using coding datasets as calibration to prune parts of big MoE models selectively. and the outcome is "it's fine on coding, but if you ask it about who bill clinton is, it has zero preserved knowledge of him whatsoever"
@kalomaze It will know Al Gore though. He invented the internet.
REAP is fascinating. you can find people on huggingface using coding datasets as calibration to prune parts of big MoE models selectively. and the outcome is "it's fine on coding, but if you ask it about who bill clinton is, it has zero preserved knowledge of him whatsoever"