all true for a single grep invocation, however a capable LLM can drive grep using similar terms (`grep (red|maroon|scarlet|crimson)`, moving the measure of similarity from embedding/indexing time to inference time. There's benefits to that.
reply> red|maroon|scarlet|crimson
replystill doesn't include typo'ed variants for any of these
but a good combination of prompt, harness and model will easily close that gap. a more advanced (but equivalent) form of grep could also offer fuzzy matching/levenshtein distance function to augment that.
replyIt will, but it does so by burning tokens. I'm sure you've also seen the output of an overly eager grep by an LLM.
replyI think we're verging on the point where people will care more about the most token-efficient ways to do a task than the lowest human effort version. Humans are expensive, humans using half as many tokens are cheaper.
This sounds like AOT vs JIT weights.
replyIf the data is this large then why not ask the llm to build a query
replyThat's more-or-less what RAG is; a tool to surface relevant data so that AI can make informed decisions. Efficient retrieval is important so that AI doesn't need to ingest the whole corpus for each query.
replylatency is often the reason, distilling interactions into significantly better retrieval is the second.
replyMy case:
replyWith a very large dataset of totally unnormalized data, semantic folding is a silver bullet.
[flagged]
reply