AGENT SKILL

text-wordclouds

Use AutoViz text-column cleaning and wordcloud behavior for NLP-style tabular fields.

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

01 前言 Front Matter

name
text-wordclouds
license
Apache 2.0
metadata
{ "disco-role": "operating" }
description
Use AutoViz text-column cleaning and wordcloud behavior for NLP-style tabular fields.
disable-model-invocation
true

02 资源文件 3

03 SKILL.md 正文

Text Wordclouds

Use this sub-skill when the user asks about AutoViz behavior for string columns, discrete text columns, NLP variables, wordcloud plots, text cleanup, emoji handling, stopwords, or NLTK downloads.

Use this when

  • The prompt names wordcloud, AutoViz_NLP, draw_word_clouds, nltk, textblob, emoji, stopwords, contractions, or text cleanup.
  • AutoViz detects string variables and attempts to generate wordclouds.
  • The user needs to understand why a text column did or did not get a wordcloud.
  • The user is debugging missing NLTK data, missing wordcloud, or text preprocessing side effects.
  • The user wants to know how AutoViz distinguishes categorical strings from NLP-style strings.

Core flow

  1. Start from the DataFrame and identify candidate string columns.
  2. Let AutoViz classify variables first; only discrete/NLP string variables reach the wordcloud branch.
  3. For each selected string column, AutoViz cleans text and calls wordcloud generation.
  4. If a target column is present and the problem type is classification, wordclouds may be drawn per target class.
  5. Use static chart formats for deterministic automated runs; interactive behavior is covered by the EDA sub-skill.
  6. When the user only needs text preprocessing, explain the cleanup helpers without forcing a full AutoViz run.

Read these references

Use these scripts

  • Run scripts/wordcloud_smoke.py to verify that the wordcloud package and AutoViz text helper path can run on a safe in-memory DataFrame.
  • If the text branch is failing inside a larger EDA run, use the EDA smoke script first and then return here.

Important behavior

  • AutoViz may call nltk.download('popular') when discrete string variables are present.
  • Short or low-cardinality strings may be classified as categorical/boolean rather than NLP.
  • Very high-cardinality or long free-text columns are more likely to trigger wordcloud behavior.
  • draw_word_clouds depends on matplotlib plus the wordcloud package.
  • The text helpers also perform cleanup such as URL, HTML, emoji, contraction, and punctuation handling.

Common text helpers

  • clean_steps
  • clean_text
  • remove_URL
  • remove_html
  • remove_emoji
  • expandContractions
  • remove_stopwords
  • split_into_lemmas
  • draw_wordcloud_from_dataframe

Cross-routing

  • If the user primarily wants full-table EDA plots, route to ../eda-visualization/SKILL.md.
  • If the user asks whether a text column should be cleaned or repaired before modeling, route to ../data-quality-fixes/SKILL.md and return here for visualization-specific text behavior.
  • Do not rely on source notebooks or the original checkout; use the bundled references and script.
  • If the user needs to know whether NLTK data will be downloaded, mention it explicitly before running the workflow.

Troubleshooting reminders

  • If the column never reaches the wordcloud branch, the classifier likely treated it as categorical or boolean.
  • If nltk.download('popular') is a problem, prepare the corpora or choose a static EDA path that does not trigger text plotting.
  • If the output is empty or tiny, the text may have been over-cleaned or reduced to stopwords.
  • Use a writable save_plot_dir when saving images so the generated wordcloud is easy to find.

Escalation

If the user really wants a full EDA view of the same dataframe, route back to the EDA sub-skill after handling the text question. If the data-quality question is about missingness or cleaning rather than visualization, route to the data-quality sub-skill instead of over-focusing on wordclouds.

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