text
Routes coherent passage generation workflows for the Tree of Thoughts text task, including sample/vote BFS and scoring helpers.
- License 许可
- MIT
- 最近更新
- 2026-09-03
- Stars
- ★ 351
- Forks
- 27
01 前言 Front Matter
- name
- text
- license
- MIT
- metadata
- { "disco-role": "operating" }
- description
- Routes coherent passage generation workflows for the Tree of Thoughts text task, including sample/vote BFS and scoring helpers.
- disable-model-invocation
- true
02 资源文件 3
- 参考文档references/troubleshooting.md1.1 KB
- 参考文档references/workflows.md1.7 KB
- 脚本scripts/run_text.sh1.3 KB
03 SKILL.md 正文
Text
Use this sub-skill when the user wants to generate coherent passages or replay the paper's text workflow.
Covers
task=textruns.- BFS with
samplegeneration andvoteevaluation. - Naive
standardandcotpassage sampling. - The GPT-4-based coherency scoring helpers used by
TextTask.test_output.
Excludes
- Arithmetic puzzle solving; use
sub-skills/game24/. - Mini crossword solving; use
sub-skills/crosswords/. - General package installation; use the root skill.
Read and run
references/workflows.mdfor the text data file, prompt families, and the sample/vote settings.references/troubleshooting.mdfor score-format drift and vote parser issues.scripts/run_text.shfor the bundled command wrapper.
Quick command shapes
- BFS:
bash sub-skills/text/scripts/run_text.sh bfs - Naive standard:
bash sub-skills/text/scripts/run_text.sh standard - Naive CoT:
bash sub-skills/text/scripts/run_text.sh cot
Acceptance cues
A future agent should be able to:
- explain why the BFS path uses
sample+voterather thanpropose+value; - tell when the final coherency score comes from the GPT-4 judge path;
- tune the sample counts when the passage quality or vote stability is low;
- interpret the returned
rsandrfields from the scorer.
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).