Entry 7 · Thursday
the rule at the top of the query turned out to be more query
Kitchen table
If you write “find me a story about dogs, ignore the shopping list underneath”, the computer often mashes your whole note into one lump instead of treating the first bit as the actual rule.
Whiteboard
Eight embedding models, a plain instruction at the top of each query, and one small change to the test: put a paraphrase of the query in alongside the query itself, something nobody asked to retrieve. On SQuAD, recall at one fell from 0.724 to 0.014 – not degraded, erased – and the instruction was sitting right there the whole time, saying which of the two was wanted. It wasn't read as a rule. It was read as more words to embed. The explanation offered is uncomfortably mundane: these models are trained and benchmarked on queries where nothing on the query side ever competes, so nothing ever taught them that the instruction outranks its neighbours. Fine-tuning with distractors pulls SQuAD back to 0.470 and barely moves the other scores, which is encouraging and a little damning at once. So in your own retrieval prompt: which part is an instruction, and which part is just more text?
Boardroom
No boardroom telling for this one yet. Here is the whiteboard one.
Eight embedding models, a plain instruction at the top of each query, and one small change to the test: put a paraphrase of the query in alongside the query itself, something nobody asked to retrieve. On SQuAD, recall at one fell from 0.724 to 0.014 – not degraded, erased – and the instruction was sitting right there the whole time, saying which of the two was wanted. It wasn't read as a rule. It was read as more words to embed. The explanation offered is uncomfortably mundane: these models are trained and benchmarked on queries where nothing on the query side ever competes, so nothing ever taught them that the instruction outranks its neighbours. Fine-tuning with distractors pulls SQuAD back to 0.470 and barely moves the other scores, which is encouraging and a little damning at once. So in your own retrieval prompt: which part is an instruction, and which part is just more text?
Tags
All entries · 32 in all