Skip to content
Back to skills

Model Data Api

ASecurity

"Inspect SPLADE model, data, index, and inference APIs plus

  • 247 stars
  • 0 votes
  • 0 copies
  • 3 views
  • Added September 9, 2026
ai-agentspythonbashapi

Works with

  • api

Security analysis

A100/100

Pro scans all 7 files and shows the line behind each finding

Scanned September 9, 2026

npx -y skills add VectorSpaceLab/AREX-Skill --skill model-data-api --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Model Data Api?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Model Data Api
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/vectorspacelab-model-data-api/badge)](https://www.skillsdirectory.com/skills/vectorspacelab-model-data-api)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: model-data-api
description: "Inspect SPLADE model, data, index, and inference APIs plus
  validate small SPLADE-style data layouts without running full training."
disable-model-invocation: true
metadata:
  disco-role: operating
license: NOASSERTION
---

# SPLADE Model/Data API

Use this sub-skill when a future agent needs to reason about SPLADE Python APIs, data schemas, index objects, or small data validation without launching full training, indexing, retrieval, export, pruning, or evaluation jobs.

## Start Here

- For model classes, `get_model`, bag-of-words helpers, HF Trainer model wrappers, DPR, and inverted index objects, read [API reference](references/api-reference.md).
- For `raw.tsv`, qrel JSON, score JSON, hard-negative pkl-gz, TREC run, and row-id/content-id expectations, read [data formats](references/data-formats.md).
- For safe API inspection and notebook-style bag-of-expanded-words inference concepts, read [inference and API](references/inference-and-api.md).
- For schema mismatches, row/content id bugs, special-token cleaning, offline model issues, and empty vectors, read [troubleshooting](references/troubleshooting.md).

## Bundled Scripts

- Validate a SPLADE-style fixture root without importing SPLADE:

  ```bash
  python sub-skills/model-data-api/scripts/validate_splade_toy_data.py /path/to/dataset-root
  ```

- Inspect installed SPLADE imports and signatures without downloading model weights:

  ```bash
  python sub-skills/model-data-api/scripts/inspect_splade_api.py --json
  ```

## Routing Boundaries

- Use this sub-skill for API signatures, object behavior, data schemas, toy validation, small smoke checks, and inference snippets that do not download weights by default.
- Route classic Hydra train/index/retrieve/evaluate module execution to `../hydra-pipelines/SKILL.md`.
- Route HuggingFace Trainer training and reranking command construction to `../hf-training-reranking/SKILL.md`.
- Route Anserini export, pruning, PISA, BEIR, and evaluation caveats to `../pruning-export-evaluation/SKILL.md`.

Files in this skill

  • SKILL.md2 KB
  • references/api-reference.md11 KB
  • references/data-formats.md6.1 KB
  • references/inference-and-api.md6 KB
  • references/troubleshooting.md6.3 KB
  • scripts/inspect_splade_api.py7.6 KB
  • scripts/validate_splade_toy_data.py15.4 KB

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.