ECBS5200 Week 2 — Diagnostic Model

This model is a course artifact for ECBS5200: Practical Deep Learning Engineering at Central European University, Vienna.

What this is

A fine-tuned ModernBERT-base for 113-class consumer complaint classification (CFPB dataset). It was trained with a specific configuration as part of a diagnostic lab exercise.

How it's used

Students receive four models (alpha, bravo, charlie, delta) and must determine what training configuration produced each one by examining metrics, training curves, and per-class performance. The configurations are intentionally not disclosed here.

Loading

from transformers import AutoModelForSequenceClassification, AutoTokenizer

model = AutoModelForSequenceClassification.from_pretrained("earino/ecbs5200-week2-alpha")
tokenizer = AutoTokenizer.from_pretrained("earino/ecbs5200-week2-alpha")

Not for production use

This model was trained for educational purposes. It is one of four models in a diagnostic exercise and may not represent optimal performance on this task.

Course

• Course: ECBS5200 — Practical Deep Learning Engineering for Applied ML
• Institution: Central European University, Vienna
• Instructor: Eduardo Arino de la Rubia
• Semester: Spring 2026
• Course site: earino.github.io/applied-deep-learning