75 samples free No account needed

EdukaAI Starter Pack

Engine-generated, ElGap-validated dataset. 75 samples produced by the AI Curator engine and reviewed by the ElGap team — download free from ai-curator.cloud/starter-pack, with or without AI Curator.

What's in the pack

The EdukaAI Starter Pack is a layered dataset produced by the AI Curator engine. A fictional football (soccer) commentary scenario was processed using a commentary analysis strategy. The engine produced the instruction-output pairs, categories, and quality scores — and every sample was then reviewed and validated by the ElGap team to ensure consistency and usefulness.

Player Roleplays

Rich character-driven conversations between fans, analysts, and players

20 samples
📊

Tactical Analysis

Match breakdowns, formation analysis, and strategic commentary

15 samples
🎉

Fan Perspectives

Emotional reactions and debates from both sides of a match

15 samples
🎤

Commentary Transcripts

Professional match narration with play-by-play descriptions

15 samples
🧰

Alternate History

"What if" scenarios exploring different match outcomes and consequences

10 samples

Why it matters

The biggest barrier to fine-tuning and RAG isn't the model — it's the data. Most people don't have any structured data to start with. The Starter Pack eliminates the cold start problem. And it proves the engine works: if this is what it produces from a fictional match, imagine what it does with your real data.

Zero cold start

No data? No problem. Download the Starter Pack — 75 engine-generated, human-validated samples. Learn the workflow on real engine output before running your own documents through the pipeline.

No account needed

Standalone download from the Datasets page. Import the JSONL file directly into EdukaAI Studio to start fine-tuning. AI Curator is for when you want to curate your own data.

Train in 5 minutes

Download the Starter Pack from the Datasets page, import into EdukaAI Studio, and train a model on your Mac. From download to a custom model in under 10 minutes. No GPU, no cloud, no code.

RAG-ready, not just training-ready

The same samples work for RAG. Export as JSONL and feed to your embedding pipeline. The Starter Pack demonstrates how curated data improves retrieval quality — not just training quality.

Sample breakdown

Every sample has instruction-output pairs, quality ratings, category tags, and review status. Here's what that looks like across categories:

CategoryDescriptionFine-TuningRAG
Player Roleplays
Rich character-driven conversations between fans, analysts, and playersDialogue & persona trainingCharacter-based retrieval
📊Tactical Analysis
Match breakdowns, formation analysis, and strategic commentaryAnalytical reasoningStrategic content retrieval
🎉Fan Perspectives
Emotional reactions and debates from both sides of a matchSentiment & opinionCommunity knowledge base
🎤Commentary Transcripts
Professional match narration with play-by-play descriptionsDescriptive generationFactual event retrieval
🧰Alternate History
"What if" scenarios exploring different match outcomes and consequencesCreative reasoningHypothetical content

Get the Starter Pack

Download the Starter Pack from the Datasets page — no account needed for the 75-sample free pack. Create a free account for the 400-sample Extended Pack.

Free download

Starter Pack (75 samples)

Download the JSONL file — no account needed. Import into EdukaAI Studio or any training framework.

Go to Datasets
With AI Curator

Full curation workflow

Install AI Curator, download the Starter Pack from the Datasets page, then review, rate, and curate before exporting.

Terminal
 brew tap elgap/tap brew install ai-curator curator
400 samples

Extended Pack

5× more data with deeper coverage. Engine-generated, human-validated by ElGap. Requires a free account.

Create a free account

Fine-tune with EdukaAI Studio

The Starter Pack works directly with EdukaAI Studio — the no-code fine-tuning app for Apple Silicon. No AI Curator installation required.

Step 1
Download Pack
From the Datasets page
Step 2
Import to Studio
Select model, configure
Step 3
Train & Test
Click Train, test in Dual Chat

No GPU required — runs on any M1/M2/M3/M4 Mac. If you want to curate the data before training (rate, tag, approve/reject, filter), install AI Curator and download the Starter Pack from within the app.

Fictional scenario · Engine-generated · Human-validated
The football match, players, and events are fictional and AI-generated by the AI Curator engine. Every sample was then reviewed and validated by the ElGap team for quality and consistency. The data structure, categories, and quality annotations reflect real engine output — this is what the engine produces from any document.

Download and start training

75 free samples, no account needed. Go to the Datasets page to download, then import into EdukaAI Studio or use with AI Curator.