Hugging Face's logo

Spaces:
ESCP
/
escp-huggingface-guidelines
Running

App Files Community
escp-huggingface-guidelines / README.md
LaurenaESCP's picture
LaurenaESCP
Update README.md
4ac1227 verified
preview code
|
raw
history blame contribute delete
6.94 kB
metadata 
title: ESCP Huggingface Guidelines
emoji: 🏢
colorFrom: pink
colorTo: pink
sdk: static
pinned: false
short_description: Publishing and usage guidelines for ESCP projects on Hugging

ESCP on Hugging Face — README (Publishing & Usage Guide)

This repository/organization is ESCP’s hub for sharing applied AI work: teaching assistants, course demos, student projects, research prototypes, and datasets. The goal is to learn by doing, publish responsibly, and keep the ESCP brand credible.

If you are new to Hugging Face, start with the Quick Start. If you already know the platform, jump to Publishing Standards.

1) Quick Start (10 minutes)

Hugging Face Hub basics

  • Models: weights + inference examples + eval notes
  • Datasets: data + dataset card + license + intended use
  • Spaces: interactive demos (Gradio / Streamlit / static) — best for pedagogy and showcasing prototypes
  • Collections: curated groups (recommended for course cohorts, projects, or labs)

Recommended path

  1. Explore existing ESCP assets (Spaces/Models/Datasets)
  2. Clone or duplicate a Space as a template
  3. Publish your first draft privately (when possible), then request review

Docs:

2) What belongs here (scope)

✅ Good fits

  • Teaching assistants, course copilots, interactive demos
  • Student prototypes, hackathon outputs (cleaned)
  • Research artifacts: baselines, evaluation code, reproducible demos
  • Datasets used for learning/research with clear rights and documentation
  • Tutorials, notebooks, templates for ESCP usage
  • and any other suggestions you wish to share with us

❌ Not allowed

  • Anything with unclear IP/rights or confidential information
  • Scraped data that violates terms, privacy, or platform policies
  • Personal data (students, staff, interviewees) unless proper consent and a valid legal basis are clearly established
  • Content that conflicts with applicable European regulations, including GDPR and the EU AI Act, or exposes ESCP or individuals to legal, ethical, or regulatory risk
  • “Half-broken” repos with no README, no description, or no reproducibility

Rule of thumb: publish what you’d be comfortable showing publicly to partners, media, and peers.

3) Publishing workflow (recommended)

A. Before you publish

  • Ensure you have the rights to publish code/data/model
  • Remove secrets (tokens, keys), internal URLs, private documents
  • Add a README (this one with link) + a short “card” (model/dataset) when relevant
  • Add a minimal reproducible example (how to run, dependencies)

B. Get support or approval

For questions, support requests, or publication guidance contact: genai@escp.eu

If you’re unsure about IP, privacy, or institutional sensitivity: ask first.

4) Naming & structure standards

Clear naming makes ESCP content discoverable and professional.

A. Repository naming pattern (recommended)

Use lowercase, hyphen-separated:

Spaces

  • escp-aita-student (teaching assistant student-facing)
  • escp-course-<course>-demo (e.g. escp-course-finance-demo)
  • escp-ai-education-<topic> (e.g. escp-ai-education-feedback)

Models

  • escp-<lab-or-course>-<task>-<version>
    Example: escp-aiedu-rag-baseline-v1

Datasets

  • escp-<domain>-<dataset-name>-<version>
    Example: escp-management-cases-synth-v1

B. Descriptions and tags

Every asset must include:

  • One-line purpose (what it does)
  • Target audience (students / faculty / research / partners)
  • Status tag: prototype, teaching, research, demo, baseline

C. Versioning

Use semantic-ish suffixes: v1, v1.1, v2 and keep a short changelog section.

5) Content & safety policy (non-negotiable)

ESCP content must remain professional, inclusive, and responsible.

Not allowed:

  • Racist, discriminatory, hateful, harassing, or violent content
  • Targeted attacks or demeaning stereotypes
  • Sexual content involving minors (or any unsafe content)
  • Content encouraging wrongdoing or unsafe behavior
  • “Edgy” demos that can embarrass ESCP or harm others

Also avoid:

  • Biased examples presented as facts without context
  • Generated “fake news” style demos without clear labeling

Minimum requirement:

  • Add an Ethics / Limitations note when a model/demo could mislead or impact people.

6) Data, privacy, and IP (must check)

Before publishing datasets or logs:

  • Do not publish personal data (names, emails, IDs, voice, faces) unless you have explicit authorization and a valid legal basis.
  • Prefer anonymized or synthetic data for teaching.
  • Respect copyright and data licenses.

For datasets, include:

  • Source and license
  • Collection method
  • Intended use and limitations
  • Known biases or risks

Docs:

7) Quality bar (lightweight but real)

Every public asset should include, at minimum:

Spaces

  • Clear title + 2–5 line description
  • “How to use” section
  • Basic guardrails (input constraints, disclaimers)
  • Works on CPU unless a GPU is truly required

Models

  • Model card (task, training data summary, eval, intended use)
  • Inference example (Transformers snippet or API)
  • Limitations and safety notes

Datasets

  • Dataset card (what/why/how)
  • Schema, splits, examples
  • License and usage constraints

8) Encourage innovation (what we want to see)

This organization is not only for “finished research”. We want:

  • Pedagogical prototypes that help students learn
  • Transparent baselines and reproducible experiments
  • Practical demos that connect AI to management problems
  • Proposals for new collections, benchmarks, or shared templates

If you have an idea:

  • Create a small draft Space
  • Write a 10-line README with the goal + expected users
  • Email genai@escp.eu for guidance and visibility

9) Suggested templates (copy/paste)

A. Space README skeleton

  • What it is (2 lines)
  • Who it’s for
  • How to use
  • Technical notes (stack, requirements)
  • Limitations / ethics

B. Model card skeleton

  • Model summary
  • Intended use
  • Training data (high-level)
  • Evaluation
  • Limitations & biases
  • How to run

C. Dataset card skeleton

  • Dataset summary
  • Motivation
  • Composition / schema
  • Collection process
  • Licensing
  • Intended use & risks

10) Contacts

For support, publication guidance, or special requests: genai@escp.eu

For anything sensitive (privacy, IP, reputational risk): contact first, publish after.

This README is a living document. Tutorials and additional ESCP guidance will be released progressively, with updates also available on https://escp.eu/ai.