mirror of
https://github.com/microsoft/SkillOpt.git
synced 2026-07-07 00:15:39 +08:00
244e346b83
- Skill optimization framework with training loop analogy - 11 benchmarks, 4 model backends (Azure OpenAI, Claude, Codex, Qwen) - WebUI for browser-based training control - Pluggable architecture for extending benchmarks and backends
16 lines
628 B
Python
16 lines
628 B
Python
"""ReflACT Optimizer -- skill update operations.
|
|
|
|
Analogous to the optimizer in neural network training: applies the computed
|
|
"gradient" (patches) to the current skill document to produce an updated
|
|
candidate skill.
|
|
|
|
Modules
|
|
-------
|
|
- skill: edit application (optimizer.step() / parameter update)
|
|
- clip: edit ranking and selection (gradient clipping)
|
|
- meta_reflect: epoch-level macro refinement (momentum)
|
|
- slow_update: longitudinal comparison and guidance (EMA / regularization)
|
|
"""
|
|
from skillopt.optimizer.skill import apply_edit, apply_patch # noqa: F401
|
|
from skillopt.optimizer.clip import rank_and_select # noqa: F401
|