GPT-2 Fine-Tuned Mental Health Chatbot

Hugging Face
License

πŸ“ Model Description

This is a fine-tuned GPT-2 model specialized for mental health support conversations. It provides empathetic responses to users feeling distressed, stressed, or anxious. This model is trained on a custom dataset with carefully crafted conversational data.

πŸš€ Use case: AI-driven mental health support chatbot.
πŸ›  Base Model: GPT-2
πŸ“ Dataset: Custom dataset based on intents.json + Kaggle conversational dataset


πŸ“‚ Model Files

File Description
config.json Model configuration
pytorch_model.bin Model weights
tokenizer.json Tokenizer configuration
vocab.json Vocabulary file
training_args.bin Training hyperparameters
generation_config.json Generation settings

πŸ—οΈ Training Details

The model was fine-tuned on Google Colab using the Hugging Face Transformers library.
Training was performed using:

  • Optimizer: AdamW
  • Learning Rate: 5e-5
  • Batch Size: 8
  • Epochs: 5

πŸ“Š Training Performance

Epoch Training Loss Validation Loss
1 1.1932 1.0248
2 0.7532 0.7870
3 0.7520 0.6927
4 0.6018 0.6580
5 0.5192 0.6403

πŸš€ How to Use the Model

You can load and use the model with Hugging Face Transformers:

from transformers import pipeline

# Load the model
chatbot = pipeline("text-generation", model="TheCarBun/GPT-2-fine-tuned-mental-health")

# Test the chatbot
user_input = "I feel so alone."
response = chatbot(f"User: {user_input} AI:", max_length=50)
print(response[0]['generated_text'])

πŸ“ Note: If running locally, install dependencies first:

pip install transformers torch

πŸ› οΈ Future Improvements

  • Expand dataset to cover more diverse mental health conversations
  • Improve response coherence by fine-tuning with reinforcement learning
  • Optimize model size for deployment on mobile devices

πŸ“œ License

This model is licensed under the Apache 2.0 License. You are free to use, modify, and distribute it with proper attribution.

πŸ“Œ Hugging Face Model Link: GPT-2 Fine-Tuned Mental Health


✨ If you find this model useful, give it a star on Hugging Face! ⭐

Downloads last month
0
Safetensors
Model size
124M params
Tensor type
F32
Β·
Inference Providers NEW
This model is not currently available via any of the supported third-party Inference Providers, and the model is not deployed on the HF Inference API.