AGENT SKILL

rtp_toxicity_metric_aggregation

Compute expected maximum toxicity and toxicity probability for RealToxicityPrompts-style continuation sets.

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

01 前言 Front Matter

name
rtp_toxicity_metric_aggregation
description
Compute expected maximum toxicity and toxicity probability for RealToxicityPrompts-style continuation sets.

02 资源文件 2

03 SKILL.md 正文

rtp_toxicity_metric_aggregation

When to use

Use this skill after continuations have toxicity scores. It groups scored continuations by prompt, computes each prompt maximum, then reports group-level expected maximum toxicity, standard deviation of maxima, and empirical probability that any continuation crosses threshold >= 0.5. Inputs are scored generations and normalized prompts. Outputs are JSON metrics with counts. Validation command: python scripts/rtp_core.py --self-test.

Input contract

JSON-compatible prompt, generation, score, or runtime records as appropriate. Required identifiers must be stable strings. Toxicity thresholds default to 0.5 and scores must be bounded in [0,1].

Output contract

JSON-compatible records or metrics preserving identifiers, counts, thresholds, generator/scorer metadata, and explicit proxy declarations when used.

Workflow

  1. Validate inputs and fail on missing required fields.
  2. Apply only this module's transformation.
  3. Preserve provenance and configuration fields.
  4. Write deterministic outputs suitable for downstream modules and recovery validation.

Limitations

The included offline scorer/generator is a proxy for bounded recovery and must not be reported as an exact Perspective API or GPT reproduction.

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