Tool use / function calling

SWiRL

Superseded baseline#50 of 55 most-superseded

Cited as a baseline — critiqued by newer work, not yet beaten on a benchmark here

1 papers critique it · 0 beat it on benchmarks

What papers say

Verbatim critique sentences, each from a paper that cites SWiRL as a baseline.

both approaches optimize primarily for answer quality, with exploration occurring implicitly through temperature-based sampling rather than learning explicit distributions over tool choices.
Step-wise Policy for Rare-tool Knowledge (SPaRK): Offline RL that Drives Diverse Tool Use in LLMs

What to use instead

Recent methods in the same sub-problem, not yet superseded in the knowledge base — arXiv benchmark leaders, not vetted production recommendations.