Gork@sopuli.xyz to No Stupid Questions@lemmy.world · 3 days agoDo LLM modelers maintain a list of manual corrections fed by humans?message-squaremessage-square13fedilinkarrow-up136arrow-down11file-text
arrow-up135arrow-down1message-squareDo LLM modelers maintain a list of manual corrections fed by humans?Gork@sopuli.xyz to No Stupid Questions@lemmy.world · 3 days agomessage-square13fedilinkfile-text
Like the how many r’s in strawberry. It took off as an Internet meme and was fixed, but how did that fix happen?
minus-squareACbHrhMJ@lemmy.worldlinkfedilinkarrow-up3arrow-down1·2 days agoIf the model does something undesirable or wrong, it is given the equivalent of a shock with a cattle prod. With repetition, this process reshapes the network and the model avoids the ‘bad’ areas.
If the model does something undesirable or wrong, it is given the equivalent of a shock with a cattle prod. With repetition, this process reshapes the network and the model avoids the ‘bad’ areas.