AGENT SKILL

stochastic_actor_update

Run a deterministic SAC-style stochastic actor update with reparameterized Gaussian action, log probability, and policy-gradient diagnostics.

License 许可
Apache-2.0
最近更新
2026-09-03
Stars
★ 351
Forks
27

01 前言 Front Matter

name
stochastic_actor_update
description
Run a deterministic SAC-style stochastic actor update with reparameterized Gaussian action, log probability, and policy-gradient diagnostics.

02 资源文件 2

03 SKILL.md 正文

Stochastic Actor Update

Use this skill when a recovery needs to exercise the actor component of SAC without depending on a full deep-learning stack. It implements a tiny scalar Gaussian actor with fixed noise so the reparameterized policy loss can be audited deterministically.

Inputs

  • Actor mean and log standard deviation parameters.
  • State feature and deterministic noise sample.
  • A differentiable quadratic Q proxy target.
  • Learning rate.

Outputs

  • Action, log probability, Q value, policy loss.
  • Updated actor parameters and parameter-change evidence.

Workflow

  1. Reparameterize the action as mean * state + exp(log_std) * noise.
  2. Evaluate Gaussian log probability for the noise-scaled action.
  3. Compute SAC policy loss alpha * log_prob - Q(action).
  4. Use analytic gradients for the scalar actor and update parameters.
  5. Record before/after parameters for validation.

Validation

Run python tests/test_actor_update.py or the Distiller skill validator.

Limitations

This is a scalar deterministic proxy for the SAC actor update. It preserves the optimization contract but is not a neural-network policy implementation.

04 同仓库更多 Skills

VectorSpaceLab
VectorSpaceLab/AREX-Skill

passnet-triton-opt

Make a single PassNet Triton kernel fast AND numerically faithful: performance model (when a replacement can win at all), block/grid/warp tuning, autotune policy, launch overhead, and per-op numeric recipes to pass the dtype baseline tolerances. Use when a pass matches and is correct but speedup below expected, or when correctness fails by small numeric margins.

★ 351
VectorSpaceLab
VectorSpaceLab/AREX-Skill

passnet-solve

END-TO-END playbook for solving one PassNet sample: analyze the computation graphs, decide the optimization strategy (what to fuse, what to replace, what to leave alone), drive the iteration loop, and maximize the sample score. This is the ENTRY skill — invoke it first for any PassNet optimization task; it tells you when to use passnet-pattern-fusion, passnet-triton-opt and passnet-feedback.

★ 351
VectorSpaceLab
VectorSpaceLab/AREX-Skill

passnet-skill

PassNet GPU kernel optimization via compiler passes. Design and implement Triton-based optimization passes, create pass files under ./pass_dir/, self-evaluate with pass_evaluator, and iterate to maximize GPU speedup.

★ 351
VectorSpaceLab
VectorSpaceLab/AREX-Skill

passnet-pattern-fusion

Author PassNet pass files: write patterns that actually MATCH the FX graph, pick fusion regions (which ops to absorb into one Triton kernel), and structure multi-pass submissions with the mandatory shared-dispatch architecture. Use when creating or fixing pattern()/replacement_args()/replacement_func() files, when a pass "failed to match", or when deciding how to fuse multiple kernels/ops.

★ 351
VectorSpaceLab
VectorSpaceLab/AREX-Skill

passnet-orchestrate

Round/batch planning and decision support for PassNet graph optimization. Use for multi-sample triage, worker allocation, eval-budget policy, and post-eval keep/revert decisions. For solving one concrete sample end to end, use passnet-solve as the entry skill; consult this skill only when you need broader planning or a specific decision gate.

★ 351
VectorSpaceLab
VectorSpaceLab/AREX-Skill

passnet-feedback

Fast, deterministic feedback for PassNet work: pre-flight pattern/match verification WITHOUT burning a GPU evaluation, per-node bottleneck analysis of a sample's graphs, and eval-log parsing into per-variant status + estimated score + failure classification. Use BEFORE every GPU evaluation (check_pattern), at the START of a sample (analyze_graph), and AFTER every evaluation (parse_eval_log).

★ 351