Yifan Yang a29201adc4 feat(sleep): multi-objective reward (accuracy/tokens/latency) + user preferences ...

- ReplayResult records per-rollout tokens + latency_ms; replay_one measures them
  (approximated from text length when the backend doesn't track tokens, e.g. mock).
- replay.multi_objective_reward(w_acc, w_tokens, w_latency): weighted reward so a
  skill can be optimized to be cheaper/faster, not only more accurate (cost terms
  normalized vs a reference, default = accuracy-only / backward compatible).
- Backend.preferences (free text) injected into reflect as a prior; build_backend
  attaches it (to the optimizer for dual backends). run_gbrain gains --preferences.

3 new tests (multi-objective ordering, preference injection, cost recording).
29 tests pass; mock gates + 3.8/3.12 compile green.

Co-Authored-By: Claude Opus 4 <noreply@anthropic.com>

2026-06-08 14:31:51 +00:00

..

datasets

SkillOpt v0.1.0: initial release

2026-05-21 17:22:04 +00:00

engine

Support Qwen chat as optimizer backend

2026-06-01 16:44:49 +08:00

envs

envs/_template: make template instantiable against real EnvAdapter ABC

2026-06-01 20:15:12 +00:00

evaluation

Add configurable gate metric (hard / soft / mixed) for skill validation

2026-05-30 14:45:27 +08:00

gradient

fix(reflect): support continuous reward scores in failure filtering

2026-05-29 19:04:42 +08:00

model

fix(model): forward Qwen timeout and only set enable_thinking when true

2026-06-07 07:41:35 -07:00

optimizer

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

2026-05-31 02:02:23 +00:00

prompts

cleanup: remove unused benchmarks, deep_probe, meta_reflect

2026-05-24 19:36:48 +00:00

scheduler

SkillOpt v0.1.0: initial release

2026-05-21 17:22:04 +00:00

sleep

feat(sleep): multi-objective reward (accuracy/tokens/latency) + user preferences

2026-06-08 14:31:51 +00:00

utils

fix(scoring): use float() instead of int() for continuous reward scores

2026-05-30 07:47:41 +08:00

__init__.py

refactor: rename teacher/student to optimizer/target, remove best skills, fix slow update

2026-05-24 19:15:10 +00:00

config.py

Support Qwen chat as optimizer backend

2026-06-01 16:44:49 +08:00

types.py

refactor: rename teacher/student to optimizer/target, remove best skills, fix slow update

2026-05-24 19:15:10 +00:00