legal-request-classification-nlp-model

Legal Request Classification Model

Model Description

This model is a fine-tuned DistilBERT model for classifying legal-related text into predefined categories.

---

Labels

• criminal-law
• employment
• tax-law
• trademark

---

Training Details

• Dataset: jonathanli/law-stack-exchange

• Preprocessing:

  • Column cleaning
  • Label encoding
  • Tokenization

• Framework: Hugging Face Transformers

• Training: Trainer API

• Mixed precision (FP16)

---

Performance

• Accuracy: ~95%

---

Usage

Load model

from transformers import AutoTokenizer, AutoModelForSequenceClassification

model_name = "sailu4/legal-request-classification-nlp-model"

tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForSequenceClassification.from_pretrained(model_name)

---

Inference

from transformers import pipeline

classifier = pipeline(
    "text-classification",
    model=model_name
)

classifier("I need help with employment law")

---

Example Output

[
  {
    "label": "employment",
    "score": 0.96
  }
]

---

Interpretation

• label: predicted class
• score: confidence level

---

Limitations

• Limited dataset size
• Class imbalance
• Domain-specific training

---

Use Cases

• Legal document classification
• NLP experimentation
• AI assistants

---

Author

Badr Joulali

---

Source Code

https://github.com/BADR-JOULALI/legal-request-classification-nlp