nellaep/FranckyMoneyTinyLlama

FranckyMoney Dataset

The FranckyMoney dataset is a small, tabular-style dataset converted to JSONL format for language model fine-tuning.
It contains personal and financial features alongside a target label: entrepreneurialReadiness (1–10).

Dataset Details

• Rows: 66 (after cleaning)
• Format: JSONL (train.jsonl, validation.jsonl, test.jsonl)
• Schema:

Column	Type	Description
id	int	Row identifier
age	int	Age of the individual
dependents	int	Number of dependents
monthlyIncome	float	Monthly income
monthlyBills	float	Monthly recurring bills
monthlyEntertainment	float	Entertainment spending per month
savingAmount	float	Current savings
assets	float	Asset value
salesSkills	int	Self-rated sales skills (1–10)
riskLevel	int	Self-rated risk tolerance (1–10)
confidence	int	Self-rated confidence (1–10)
businessDifficulty	int	Self-rated difficulty of starting business
entrepreneurialReadiness	int	Target variable (1–10)

Example Record

{
  "id": 1,
  "age": 23,
  "dependents": 1,
  "monthlyIncome": 3000,
  "monthlyBills": 2000,
  "monthlyEntertainment": 300,
  "savingAmount": 1000,
  "assets": 1500,
  "salesSkills": 4,
  "riskLevel": 5,
  "confidence": 7,
  "businessDifficulty": 2,
  "entrepreneurialReadiness": 7
}

How It Was Used

This dataset was used to fine-tune FranckyMoneyTinyLlama, a TinyLlama model that predicts entrepreneurial readiness from structured profiles.

Splits

• train.jsonl → 80% of rows
• validation.jsonl → 10% of rows
• test.jsonl → 10% of rows

Intended Use

• Educational demonstrations of tabular-to-text fine-tuning
• Testing lightweight LoRA adapters on small datasets
• Not intended for real-world financial or business predictions

Limitations

• Very small dataset (66 rows) → limited generalization
• Labels are subjective and may not represent real entrepreneurial capacity
• Should not be used for any production or decision-making system

Citation

If you use this dataset, please cite:

@misc{nellaep2025franckymoneydataset,
  title={FranckyMoney Dataset},
  author={nellaep},
  year={2025},
  howpublished={\url{https://huggingface.co/datasets/nellaep/FranckyMoney}}
}

License

MIT