Restore Qwen3-VL default cache config
#2
by JungleGym - opened
Remove explicit use_cache: false from config.json so the published checkpoint falls back to the upstream Qwen3-VL default behavior. This avoids shipping a training-time cache setting in the released inference artifact.
JungleGym changed pull request status to merged
JungleGym deleted the
refs/pr/2 ref