AGENT SKILL

rnd_exploration_bonus

Compute Random Network Distillation novelty bonuses and simple predictor updates for symbolic state features.

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

01 前言 Front Matter

name
rnd_exploration_bonus
description
Compute Random Network Distillation novelty bonuses and simple predictor updates for symbolic state features.

02 资源文件 2

03 SKILL.md 正文

Rnd Exploration Bonus

When To Use

Use this skill when reconstructing or testing the NetHack Learning Environment paper mechanisms in a bounded recovery setting. It is appropriate for symbolic NLE-style observations, task/action contracts, exploration bonuses, trainable baseline proxies, or validation harnesses depending on the module name.

Inputs

  • JSON-compatible records produced by upstream NLE recovery modules.
  • Tiny deterministic fixtures for smoke tests or reduced recovery.
  • Paper-derived target metadata when the module contributes to recovery evidence.

Outputs

  • Compact Python dictionaries or scalar values that can be serialized into Distiller recovery logs.
  • Explicit errors for malformed input instead of silent fallback behavior.

Workflow

  1. Confirm the input contract before computing downstream values.
  2. Preserve the paper mechanism named in the module document rather than copying repository layout.
  3. Use the script in scripts/ for deterministic behavior and record outputs in recovery logs.
  4. During recovery, do not read the original NLE repository; rely on paper/module/skill artifacts only.

Validation

Run python -m pytest tests from this skill directory, or validate through the Distiller validate_skill_tree.py --run-tests command. The tests use only the Python standard library and deterministic fixtures.

Limitations

This skill supports reduced/proxy recovery. It does not claim to reproduce full distributed IMPALA training or a compiled NetHack simulator by itself.

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