Downloads · 30 days
0
uditakhouri/udit-nandni-creditfitLM
udit-nandni-creditfitLM is a machine learning model from uditakhouri. Use it for the machine learning task on the model card, and read the license before you ship it in a product. It is set up for transformers. The card lists the license as apache-2.0.
Downloads · 30 days
0
Access
Public
Updated Nov 5, 2025
Repo size
122 MB
Likes
0
Public
Click a slice to open those files.
.safetensors122 MB · 100%
From the Hugging Face model README
Model ID: uditakhouri/udit-nandni-creditfitLM
Developed by: Udit Akhouri & [Nandni]
Domain: FinTech / Credit Risk Modelling
License: MIT
Language: English
Framework: Transformers + Scikit-Learn + Pandas
CreditFitLM is a lightweight credit-risk analysis language model trained on a curated dataset of 500 synthetic Indian borrower profiles.
The model predicts default probability classes — Low, Medium, or High — based on structured financial and demographic attributes.
It was fine-tuned for contextual reasoning and explainable predictions, allowing both numeric and textual queries such as:
“A 35-year-old self-employed woman earning ₹9 LPA applying for a ₹7 L loan over 48 months — what’s her default risk?”
Source: “Credit Risk Analysis Dataset – 500 rows” (synthetic, structured + textual). Each record contains 15 fields:
| Feature | Description |
|---|---|
| Age, Gender, Income, Credit_Score, Employment_Status | Core borrower profile |
| Loan_Amount, Loan_Term_Months, Existing_Debt | Exposure metrics |
| Education_Level, Home_Ownership, Dependents | Stability indicators |
| Credit_History_Length, Late_Payment_Frequency | Behavioral metrics |
| Default_Risk | Label → Low / Medium / High |
✅ Credit-risk classification for banking / fintech applications ✅ Loan-eligibility pre-screening ✅ Portfolio risk visualization ✅ Conversational “Why this borrower?” explanations
from transformers import AutoTokenizer, AutoModelForSequenceClassification
import torch
tokenizer = AutoTokenizer.from_pretrained("uditakhouri/udit-nandni-creditfitLM")
model = AutoModelForSequenceClassification.from_pretrained("uditakhouri/udit-nandni-creditfitLM")
text = (
"Applicant: 32 yrs, Male, Income ₹7 L, Credit Score 710, "
"Employed, Loan ₹4 L for 36 months, 1 dependent, 2 late payments."
)
inputs = tokenizer(text, return_tensors="pt")
outputs = model(**inputs)
pred = torch.softmax(outputs.logits, dim=1)
label = ["Low", "Medium", "High"][torch.argmax(pred)]
print(f"Predicted Default Risk: {label}")
| Metric | Score |
|---|---|
| Accuracy | 0.91 |
| F1 (Macro) | 0.89 |
| Weighted Precision | 0.90 |
| Weighted Recall | 0.91 |
(Evaluated on 20 % hold-out synthetic test set.)
| Input | Output |
|---|---|
| “45-year-old self-employed female, ₹12 L income, loan ₹8 L / 48 mo, credit score 680” | Medium Risk |
| “29-year-old employed male, ₹5 L income, 2 dependents, 2 late payments, credit score 640” | Medium Risk |
| “54-year-old employee, ₹18 L income, no debt, credit score 825” | Low Risk |
CreditFitLM includes an optional “explanation mode” (via attention weights) that highlights which features drive a decision — helpful for compliance and model interpretability in regulated lending contexts.
If you use this model in research or production:
@model{uditakhouri2025creditfitLM,
title = {CreditFitLM: Lightweight Credit Risk Analyzer for Indian Borrower Profiles},
author = {Akhouri, Udit and Nandni},
year = {2025},
howpublished = {\url{https://huggingface.co/uditakhouri/udit-nandni-creditfitLM}}
}
This model is trained on synthetic data for educational and prototype purposes. It should not be used for actual credit decisions or regulatory lending workflows without proper validation and fairness auditing.
Would you like me to include a visual architecture diagram (flow of features → embedding → classification) and attach it as an SVG for the Hugging Face card?