url/SkillOpt
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat(slow-update): add config-controlled gated / force-injected modes

Browse Source 
Add optimizer.slow_update_gate_with_selection to control how epoch-boundary
slow-update guidance is applied:
- false (default): force-injected - inject guidance into current & best
  unconditionally (unchanged behavior).
- true: gated - evaluate the slow-update candidate on the selection set and
  accept/reject via the same validation gate as step-level updates
  (logic follows the SkillReflection ablation).

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
This commit is contained in:
Cuzyoung
2026-05-31 01:52:02 +00:00
parent 42e555d28e
commit 00602df9e9
5 changed files with 153 additions and 49 deletions
Download Patch File Download Diff File Expand all files Collapse all files
skillopt/config.py
+1
View File
@@ -98,6 +98,7 @@ _FLATTEN_MAP: dict[str, str] = {
"optimizer.meta_learning_rate": "meta_edit_budget",
"optimizer.use_slow_update": "use_slow_update",
"optimizer.slow_update_samples": "slow_update_samples",
"optimizer.slow_update_gate_with_selection": "slow_update_gate_with_selection",
"optimizer.longitudinal_pair_policy": "longitudinal_pair_policy",
"optimizer.use_meta_skill": "use_meta_skill",
"evaluation.use_gate": "use_gate",