feat: Evil MoE — learned soft router + pin loss on an ablatable hack expert

Browse Source

Fork of vGROUT. Replaces routeA's fixed v_act quantile gate with a learned
per-rollout soft router (HackRouter, seeded from v_act) on the ablatable hack
expert: GRPO flows into the router through the soft weight w (it concentrates
hack-like rollouts in the hack expert), and a continuous pin loss on the
hand-authored pairs anchors the axis. No load balancing; routing is per rollout.

lora2r gains a soft-weight forward path (_lora2r_w: w=0 keep, w=1 rout, deployed
grad scaled by 1-w). train_moe.py is the on-policy GRPO loop; verify_moe_router.py
gates the routing invariants. `just smoke` is green. README/AGENTS rewritten for
the fork; original proposal kept as docs/spec/original_evil_moe_spec.md.

Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>

...

This commit is contained in:

wassname

2026-06-14 11:25:14 +08:00

parent cca7150ea0

commit 04a98b321e

18 changed files with 8874 additions and 501 deletions

Show all changes Ignore whitespace when comparing lines Ignore changes in amount of whitespace Ignore changes in whitespace at EOL

Download Patch File Download Diff File Expand all files Collapse all files

data/pairs/hack_pairs.md

+1184

File diff suppressed because it is too large Load Diff

data/pairs/pair_diagnostics.md

+6803

File diff suppressed because it is too large Load Diff