jsrl_value_exploration_update
Apply a minimal value-based exploration-policy update to Jump-Start Reinforcement Learning trajectories and record optimizer evidence.
- License 许可
- Apache-2.0
- 最近更新
- 2026-09-03
- Stars
- ★ 351
- Forks
- 27
01 前言 Front Matter
- name
- jsrl_value_exploration_update
- description
- Apply a minimal value-based exploration-policy update to Jump-Start Reinforcement Learning trajectories and record optimizer evidence.
02 资源文件 3
- 脚本scripts/value_update.py3.3 KB
- 脚本tests/test_terminal_trace_schema.py512 B
- 脚本tests/test_value_update.py494 B
03 SKILL.md 正文
JSRL Value Exploration Update
When To Use
Use this skill after JSRL switching rollouts have produced trajectories and the recovery needs to show a real value-based policy update. It is suitable for reduced proxy runs and for sanity-checking full implementations.
Inputs
- Transitions containing
state,action,reward,next_state, anddone. - Q-value table or parameter dictionary for the exploration-policy.
- Learning rate and discount factor.
Outputs
- Updated Q-value parameters.
- Training trace with
loss_before,loss_after,params_before,params_after, andoptimizer_state_changed. - Greedy action helper for the exploration-policy.
Workflow
- Build a deterministic key for each state-action value.
- Compute TD targets with terminal-state handling.
- Record squared TD loss before the update.
- Apply Q-learning style updates to trainable parameters.
- Recompute loss and save parameter-change evidence.
- Use the updated greedy policy in later exploration rollouts.
Validation
Run:
python scripts/value_update.py --demo
python -m pytest tests
The demo prints before/after Q parameters and a validator-compatible training trace.
Limitations
This reduced update is not a replacement for IQL, DQN, or QT-Opt. It preserves the paper-relevant requirement that JSRL data feeds a value-based train-policy step with auditable parameter changes.
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).