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
18 lines
689 B
Python
18 lines
689 B
Python
"""ReflACT Gradient -- trajectory analysis and patch generation.
|
|
|
|
Analogous to gradient computation in neural network training: analyzes
|
|
minibatch rollout trajectories to produce skill-edit patches (the "gradient"
|
|
that drives skill updates).
|
|
|
|
Modules
|
|
-------
|
|
- reflect: minibatch trajectory analysis (gradient computation)
|
|
- aggregate: hierarchical patch merging (gradient aggregation)
|
|
- deep_probe: diagnostic probe generation (gradient probing)
|
|
"""
|
|
from skillopt.gradient.reflect import ( # noqa: F401
|
|
run_minibatch_reflect,
|
|
)
|
|
from skillopt.gradient.aggregate import merge_patches # noqa: F401
|
|
from skillopt.gradient.deep_probe import generate_deep_probe_instruction # noqa: F401
|