tags:
- text-to-image
- lora
- diffusers
- template:diffusion-lora
widget:
- output:
url: images/Screenshot from 2025-08-20 00-50-33.png
text: None
parameters:
negative_prompt: None
base_model: ProsusAI/finbert
instance_prompt: null
license: other
license_name: useless
license_link: LICENSE
Emotion
- Prompt
- None
- Negative Prompt
- None
Model description
Emotion Recognition Model (BERT-based) 📌 Overview
This is a BERT-based emotion recognition model that I created purely for educational and learning purposes. The model was trained as part of my journey to understand transformers, distillation, GPU management, fine-tuning, and Hugging Face workflows.
⚙️ How I built it
I started with a pretrained BERT model.
I experimented with layer distillation (copying a few layers into a smaller student model).
I trained it on an emotion classification dataset to predict different emotional states from text.
I focused on hands-on practice: learning about tokenization, GPU memory issues, checkpointing, and model saving/loading.
⚠️ Disclaimer
This model is not production-ready.
It is not optimized for real-world use.
It should not be used for commercial, fine-tuning, or deployment purposes.
It was built only as a learning exercise to explore Hugging Face and model training.
💡 Purpose
To help me (and maybe others) understand how Hugging Face works.
To practice model distillation and fine-tuning techniques.
To learn the workflow of pushing models to Hugging Face Hub.
🚫 Limitations
Accuracy and reliability are not guaranteed.
Not suitable for critical applications (mental health, customer service, etc.).
Limited number of layers and trained on a small dataset.
Download model
Download them in the Files & versions tab.