🧠 Fine-Tuned Llama 2 7B for Mental Health Support

This repository contains the fine-tuned version of Facebook's Llama 2 7B model, trained on counselor-patient data to provide mental health support. 🤖💬

📜 Model Description

This model is designed to assist in mental health conversations by generating meaningful and supportive responses based on counselor-patient data. It is a fine-tuned version of the NousResearch/Llama-2-7b-chat-hf model. 💡

🚀 Intended Uses & Limitations

• Intended Uses: 🌱 Mental health support, emotional wellbeing assistance, counselor-patient interaction simulation
• Limitations: 🚧 Not a replacement for professional therapy or counseling. Model outputs should be used as supportive tools, not as substitutes for clinical advice.

🛠️ Training and Evaluation Data

• The model is trained on counselor-patient conversations, focusing on mental health scenarios. 🗣️

🔧 Training Procedure & Hyperparameters

• Learning Rate: 0.0002
• Train Batch Size: 2
• Eval Batch Size: 8
• Seed: 42
• Optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• LR Scheduler Type: Cosine
• LR Scheduler Warmup Ratio: 0.03
• Epochs: 1 🌟

📈 Training Results

TrainOutput(global_step=1500, training_loss=1.0985136858622233, metrics={'train_runtime': 6391.7993, 'train_samples_per_second': 0.469, 'train_steps_per_second': 0.235, 'total_flos': 3.428099244773376e+16, 'train_loss': 1.0985136858622233, 'epoch': 1.0})

🔗 Hugging Face Model Link:

Access the model here for deployment or experimentation! 🚀