vtrace_target_computation
Compute IMPALA V-trace targets, clipped ratios, and policy-gradient advantages from off-policy trajectories.
- License 许可
- Apache-2.0
- 最近更新
- 2026-09-03
- Stars
- ★ 351
- Forks
- 27
01 前言 Front Matter
- name
- vtrace_target_computation
- description
- Compute IMPALA V-trace targets, clipped ratios, and policy-gradient advantages from off-policy trajectories.
02 资源文件 2
- 脚本scripts/vtrace.py1.6 KB
- 脚本tests/test_vtrace.py954 B
03 SKILL.md 正文
Use this skill when a recovery or implementation needs the core IMPALA V-trace equations. Inputs are rewards, discounts, values with bootstrap, target action probabilities, behavior action probabilities, and clipping thresholds. Outputs are ratios, clipped rho and c, V-trace targets, and advantages.
Workflow: validate trajectory lengths, compute importance ratios, clip them separately for target residuals and trace cutting, run the backward recursive target, and report diagnostics. Use this skill before actor-critic updates; do not let a training harness duplicate the formula without cross-checking it.
Validation: tests cover on-policy reduction to n-step Bellman returns and off-policy clipping. Limitations: the helper is deterministic and vector-free for auditability, not a high-throughput learner implementation.
04 同仓库更多 Skills
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.
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.
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.
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.
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.
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).