Modelpublic

kabesaml/regllm-qwen25-7b-banking-lora

sourceHugging Faceapache-2.0updated 7mo agoView on Hugging Face

0likes9downloads

Model Card

RegLLM — Banking & Credit Risk Expert (LoRA adapter)

A LoRA adapter fine-tuned on top of Qwen/Qwen2.5-7B-Instruct for banking regulation and credit risk expertise.

What it does

This model specialises in:

—Spanish banking regulation: EBA Guidelines, CRR/CRD, Basel accords
—Credit risk methodology: PD/LGD/EAD estimation, IRB models, IFRS 9, stress testing
—SQL methodology review: validates credit risk SQL code against regulatory standards
—Spanish bank financials: Santander, BBVA, CaixaBank, Sabadell, Kutxabank (2022–2023)

Training Details

Parameter	Value
Base model	Qwen/Qwen2.5-7B-Instruct
Method	LoRA (SFT)
LoRA rank	32
LoRA alpha	64
Training examples	~163 (SQL + Banking Q&A + Regulation)
Adapter size	~309MB
Run	2026-02-22

Quick Start

python

from transformers import AutoModelForCausalLM, AutoTokenizer
from peft import PeftModel
import torch

base_model = "Qwen/Qwen2.5-7B-Instruct"
adapter = "kabesaml/regllm-qwen25-7b-banking-lora"

tokenizer = AutoTokenizer.from_pretrained(base_model)
model = AutoModelForCausalLM.from_pretrained(
    base_model,
    torch_dtype=torch.float16,
    device_map="auto"
)
model = PeftModel.from_pretrained(model, adapter)

messages = [
    {"role": "system", "content": "Eres un asistente experto en regulación bancaria y el sector bancario español. Responde con datos precisos y cita la normativa cuando sea posible."},
    {"role": "user", "content": "¿Qué es la tasa de impago bajo el estándar CRR Art. 178?"}
]

text = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
inputs = tokenizer(text, return_tensors="pt").to(model.device)
outputs = model.generate(**inputs, max_new_tokens=512, temperature=0.7)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[1]:], skip_special_tokens=True))

System Prompt

Eres un asistente experto en regulación bancaria y el sector bancario español.
Responde con datos precisos y cita la normativa cuando sea posible.

Limitations

—Trained on a relatively small dataset (~163 examples); best used as a specialised augmentation over the base model
—Primarily focused on Spanish/EU banking regulation
—Not a substitute for professional regulatory advice