Huggingface Paper Publisher

Overview

Tools for linking research papers to Hugging Face artifacts and drafting markdown articles. Indexing, authorship claims, and Hub paper search are implemented by Hub APIs documented in huggingface-papers — not by extra paper_manager.py subcommands.

Integration with HF Ecosystem

• Paper Pages: discover papers at https://huggingface.co/papers/{arxiv-id}
• arXiv Integration: visiting or POST-indexing an arXiv ID creates/updates the page
• Model/Dataset Linking: connect papers through README links and YAML
• Authorship Verification: claim in the Hub UI or via POST /api/settings/papers/claim (huggingface-papers)
• Research Article Template: generate markdown drafts from bundled templates

Version

1.0.0

Dependencies

The included script uses PEP 723 inline dependencies. Prefer uv run over manual environment setup.

• huggingface_hub>=0.26.0
• pyyaml>=6.0.3
• requests>=2.32.5
• python-dotenv>=1.2.1

What paper_manager.py actually implements

Not implemented (do not invent these CLIs): claim, check-authorship, list-my-papers, toggle-visibility, convert, validate.

For claim / POST-index / paper JSON / daily papers, follow huggingface-papers.

Usage Instructions

Scripts live in scripts/ relative to this SKILL.md. cd there or pass the full path.

Prerequisites

• Run scripts with uv run
• Set HF_TOKEN with write access for link

Method 1: Check / GET-index a paper

uv run scripts/paper_manager.py check --arxiv-id "2301.12345"
uv run scripts/paper_manager.py index --arxiv-id "2301.12345"

index only GETs the paper URL. If the page is missing, visit https://huggingface.co/papers/{arxiv-id} or POST via huggingface-papers:

curl "https://huggingface.co/api/papers/index" \
  -H "Authorization: Bearer $HF_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"arxivId":"2301.12345"}'

Method 2: Link paper to model / dataset / Space

uv run scripts/paper_manager.py link \
  --repo-id "username/model-name" \
  --repo-type "model" \
  --arxiv-id "2301.12345"

uv run scripts/paper_manager.py link \
  --repo-id "username/dataset-name" \
  --repo-type "dataset" \
  --arxiv-id "2301.12345"

uv run scripts/paper_manager.py link \
  --repo-id "username/model-name" \
  --repo-type "model" \
  --arxiv-ids "2301.12345,2302.67890,2303.11111"

uv run scripts/paper_manager.py link \
  --repo-id "username/model-name" \
  --repo-type "model" \
  --arxiv-id "2301.12345" \
  --citation "$(cat citation.txt)" \
  --create-pr

How Linking Works

When you add an arXiv paper link to a model or dataset README:

1. The Hub extracts the arXiv ID from the link
2. A tag arxiv:<PAPER_ID> is automatically added to the repository
3. Users can click the tag to view the Paper Page
4. The Paper Page shows all models/datasets citing this paper
5. Papers are discoverable through filters and search

Method 3: Claim authorship (not in this script)

Use the Hub UI or the write API in huggingface-papers:

1. Open https://huggingface.co/papers/{arxiv-id}
2. Click your name in the author list → Claim authorship
3. Wait for verification

curl "https://huggingface.co/api/settings/papers/claim" \
  -H "Authorization: Bearer $HF_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"paperId":"2301.12345","claimAuthorId":"{AUTHOR_ENTRY_ID}","targetUserId":"{HF_USER_ID}"}'

Visibility (“show on profile”) is an account-settings toggle in the Hub UI, not a CLI in this skill.

Method 4: Create a research-article draft

uv run scripts/paper_manager.py create \
  --template "standard" \
  --title "Your Paper Title" \
  --output "paper.md"

Available templates (files under templates/): standard, modern, arxiv, ml-report.

uv run scripts/paper_manager.py create \
  --template "modern" \
  --title "Fine-Tuning Large Language Models with LoRA" \
  --authors "Jane Doe, John Smith" \
  --abstract "$(cat abstract.txt)" \
  --output "paper.md"

There is no convert command. Edit the markdown; render HTML with your usual markdown toolchain or tfrere/research-article-template.

Paper Template Structure

---
title: Your Paper Title
authors: Jane Doe, John Smith
affiliations: University X, Lab Y
date: 2025-01-15
arxiv: 2301.12345
tags: [machine-learning, nlp, fine-tuning]
---

# Abstract
Brief summary of the paper...

# 1. Introduction
Background and motivation...

# 2. Related Work
Previous research and context...

# 3. Methodology
Approach and implementation...

# 4. Experiments
Setup, datasets, and procedures...

# 5. Results
Findings and analysis...

# 6. Discussion
Interpretation and implications...

# 7. Conclusion
Summary and future work...

# References

Commands Reference

uv run scripts/paper_manager.py index --arxiv-id "2301.12345"
uv run scripts/paper_manager.py check --arxiv-id "2301.12345"
uv run scripts/paper_manager.py info --arxiv-id "2301.12345" --format json
uv run scripts/paper_manager.py citation --arxiv-id "2301.12345" --format bibtex

uv run scripts/paper_manager.py link \
  --repo-id "username/repo-name" \
  --repo-type "model|dataset|space" \
  --arxiv-id "2301.12345" \
  [--citation "Full citation text"] \
  [--create-pr]

uv run scripts/paper_manager.py create \
  --template "standard|modern|arxiv|ml-report" \
  --title "Paper Title" \
  --output "filename.md" \
  [--authors "Author1, Author2"] \
  [--abstract "Abstract text"]

Search: do not run paper_manager.py search. Use https://huggingface.co/papers?search=… or the papers APIs in huggingface-papers.

YAML Metadata Format

When linking papers to models or datasets, proper YAML frontmatter is required:

Model Card Example:

---
language:
  - en
license: apache-2.0
tags:
  - text-generation
  - transformers
  - llm
library_name: transformers
---

# Model Name

This model is based on the approach described in [Our Paper](https://arxiv.org/abs/2301.12345).

## Citation

```bibtex
@article{doe2023paper,
  title={Your Paper Title},
  author={Doe, Jane and Smith, John},
  journal={arXiv preprint arXiv:2301.12345},
  year={2023}
}

**Dataset Card Example:**

```yaml
---
language:
  - en
license: cc-by-4.0
task_categories:
  - text-generation
  - question-answering
size_categories:
  - 10K<n<100K
---

# Dataset Name

Dataset introduced in [Our Paper](https://arxiv.org/abs/2301.12345).

For more details, see the [paper page](https://huggingface.co/papers/2301.12345).

The Hub automatically extracts arXiv IDs from these links and creates arxiv:2301.12345 tags.

Integration Examples

Workflow 1: Publish new research

uv run scripts/paper_manager.py create \
  --template "modern" \
  --title "Novel Fine-Tuning Approach" \
  --output "paper.md"
# Edit paper.md, submit to arXiv, then:
uv run scripts/paper_manager.py check --arxiv-id "2301.12345"
# POST-index if needed (huggingface-papers), then:
uv run scripts/paper_manager.py link \
  --repo-id "your-username/your-model" \
  --repo-type "model" \
  --arxiv-id "2301.12345"
# Claim authorship in the Hub UI or via huggingface-papers

Workflow 2: Link an existing paper

uv run scripts/paper_manager.py check --arxiv-id "2301.12345"
uv run scripts/paper_manager.py link \
  --repo-id "username/model-v1" --repo-type "model" --arxiv-id "2301.12345"
uv run scripts/paper_manager.py link \
  --repo-id "username/training-data" --repo-type "dataset" --arxiv-id "2301.12345"
uv run scripts/paper_manager.py link \
  --repo-id "username/demo-space" --repo-type "space" --arxiv-id "2301.12345"

Workflow 3: Update a model card

uv run scripts/paper_manager.py link \
  --repo-id "username/model-name" \
  --repo-type "model" \
  --arxiv-id "2301.12345" \
  --citation "Full citation for the paper"

The script adds YAML if missing, inserts the arXiv link, adds a citation block, and preserves existing content.

Best Practices

1. Paper indexing — GET-check first; POST-index or visit the URL if 404. Include full citations on cards.
2. Metadata — YAML frontmatter, license, task tags on every card.
3. Authorship — claim only papers you authored; use the Hub UI / huggingface-papers.
4. Repository linking — link every related model, dataset, and Space; keep BibTeX in the README.
5. Drafts — use one template family per project; put code/data links in the article body.

Advanced Usage

for arxiv_id in "2301.12345" "2302.67890" "2303.11111"; do
  uv run scripts/paper_manager.py link \
    --repo-id "username/model-name" \
    --repo-type "model" \
    --arxiv-id "$arxiv_id"
done

uv run scripts/paper_manager.py info --arxiv-id "2301.12345" --format json
uv run scripts/paper_manager.py citation --arxiv-id "2301.12345" --format bibtex

There is no validate command. After link, hf download username/repo README.md and inspect the card, or GET the paper page.

Error Handling

• Paper not found: arXiv ID is wrong or the Hub page is not indexed yet — visit or POST-index
• Permission denied: HF_TOKEN lacks write access to the repository
• Invalid YAML: malformed README frontmatter
• Rate limiting: too many Hub/arXiv requests

Troubleshooting

Issue: "Paper not found on Hugging Face"

• Solution: Visit https://huggingface.co/papers/{arxiv-id} or POST /api/papers/index

Issue: "arXiv tag not appearing"

• Solution: Ensure the README includes https://arxiv.org/abs/{id} or https://huggingface.co/papers/{id}

Issue: "Cannot link to repository"

• Solution: Verify HF_TOKEN has write permissions

Issue: "Template not found"

• Solution: Use standard, modern, arxiv, or ml-report and run from the skill directory

Resources

• Paper Pages: hf.co/papers
• Model cards: hf.co/docs/hub/model-cards
• Dataset cards: hf.co/docs/hub/datasets-cards
• Research article template: tfrere/research-article-template
• Claim / POST-index APIs: huggingface-papers skill