reporting
Create and troubleshoot paperai 2.6.0 YAML reports in Markdown, CSV, or optional PDF annotation format, including generated columns and safe configuration validation.
- License 许可
- Apache-2.0
- 最近更新
- 2026-09-03
- Stars
- ★ 351
- Forks
- 27
01 前言 Front Matter
- name
- reporting
- license
- Apache 2.0
- metadata
- { "disco-role": "operating" }
- description
- Create and troubleshoot paperai 2.6.0 YAML reports in Markdown, CSV, or optional PDF annotation format, including generated columns and safe configuration validation.
- disable-model-invocation
- true
02 资源文件 6
- 参考文档references/api-reference.md4.6 KB
- 参考文档references/cli-reference.md3.0 KB
- 参考文档references/configuration.md7.5 KB
- 参考文档references/troubleshooting.md7.2 KB
- 参考文档references/workflows.md5.7 KB
- 脚本scripts/validate_report_config.py6.8 KB
03 SKILL.md 正文
Reporting
Use this route to turn an existing paperai corpus/index into a bounded report. Start with configuration for YAML and column semantics, workflows for staged runs, api-reference for Python calls, and cli-reference for the positional module CLI. Use troubleshooting before changing a model, corpus, or report definition.
Scope and routing
- Owns:
Task.load,Task.queries, one-levelTask.flatten, report YAML, standard/constant/generated columns,Execute.options/create/run, Markdown (md), CSV (csv), optional PDF annotation (ant), output naming,topn,threshold,indir,allsections,section,surround,matches,snippet, and column conversion helpers. - Requires: a model directory containing
articles.sqliteand a compatible saved txtai embeddings index. Use indexing for corpus schema, index construction, vector configuration, and compatibility. - Delegates: search syntax, score filtering, grouping, and highlight internals to querying. Do not use this route to repair SQLite, build embeddings, or ingest source PDFs.
Safe route
- Copy the minimal, standard-column YAML in configuration and replace every placeholder. Run the bundled the safe validator before loading a model; it parses YAML and checks shape only, without importing paperai, opening SQLite, or touching models.
- Confirm the model directory and database/index prerequisites. For
ant, also confirm a readable source directory and exact source basenames. - Run one narrow query with explicit
topnandmd. Add one generated field, thendtype/context controls, then switch to CSV or annotation. - Inspect output files with a Markdown/CSV parser and plan a fresh output directory or unique names: paperai opens outputs in write mode and does not prompt before replacement.
Fast facts
- Renderer values are exactly
md,csv, andant; missing render resolves to Markdown, while another value raisesValueError. - A task path writes beside the YAML. Its root
nameis the Markdown/master output stem. Markdown is<name>.md; CSV creates one<query-name>.csvper top-level query and removes its temporary<name>.csv; annotation writesannotations/<Source>and removes its temporary<name>.ant. - Standard columns are
Id,Date,Study,Study Link,Journal,Source,Entry, andMatches. Aconstantcolumn performs no model call. A generated column hasqueryand optionalquestion/controls. query: "*"enumerates all articles and can maketopnineffective. The query layer uses a default threshold of 0.25 when threshold is absent. Passtopnexplicitly: in 2.6.0Execute.optionsstores a missing value asNone, which can defeat the nominalReport.buildfallback of 50.section: trueexpands a match to the full same-named subsection; positivesurround: Nexpands to nearby stored section rows.sectionwins if both are set.matchescontrols context snippets per generated field, not article count.qais anExecute/CLI parameter but verified 2.6.0 filters it before RAG construction; set YAMLoptions.llmfor the effective RAG model selector.antadditionally requirestxtmarker, PDFs, and matchingSourcenames.
Keep long schemas, renderer details, workflows, and failure matrices in the linked references rather than expanding this router.
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).