黄云龙 9f9f87c064 fix(scheduler): use consistent (step-1)/(total_steps-1) endpoint formula ...

LinearScheduler and CosineScheduler now guarantee step() returns max_lr
on the first call and min_lr on the total_steps-th call, using the
standard endpoint parameterisation t = (step-1)/(total_steps-1).

This resolves the inconsistent first-step semantics reported in review.

2026-07-14 00:45:12 +08:00

..

datasets

SkillOpt v0.1.0: initial release

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

engine

Windows robustness for claude/codex backends (+ hardened JSON fallback) (#79)

2026-06-23 19:00:23 +08:00

envs

Fix ALFWorld gamefile paths relative to ALFWORLD_DATA

2026-06-23 10:32:38 +00:00

evaluation

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

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

gradient

feat(optimizer): skill-aware reflection (EmbodiSkill S_app), config-controlled and env-independent

2026-06-10 13:10:08 +00:00

model

Windows robustness for claude/codex backends (+ hardened JSON fallback) (#79)

2026-06-23 19:00:23 +08:00

optimizer

fix(scheduler): use consistent (step-1)/(total_steps-1) endpoint formula

2026-07-14 00:45:12 +08:00

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

utils

fix(json_utils): reject prose pseudo-JSON in single quotes/backticks (#82)

2026-06-23 20:31:39 +08:00

__init__.py

chore(release): v0.2.0

2026-07-02 22:11:10 +08:00

config.py

feat(optimizer): skill-aware reflection (EmbodiSkill S_app), config-controlled and env-independent

2026-06-10 13:10:08 +00:00

types.py

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

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