ntsmarkv/bygheart-empathy-llama3.1-8b

🦥💝 BygHeart Empathy Model - Llama 3.1 8B

Achieved 4.54/5 empathy score - exceeding production targets!

Model Description

This is a fine-tuned version of Llama 3.1 8B Instruct, specifically optimized for empathetic conversations and emotional support. The model has been trained using the BygHeart empathy framework across 5 key dimensions:

• Recognition (5/5): Identifying and acknowledging emotions
• Validation (4/5): Legitimizing feelings and experiences
• Coherence (3/5): Logical, relevant, well-structured responses
• Personalization (5/5): Tailoring responses to individual context
• Support (5/5): Providing emotional assistance and guidance

Performance

• Overall Empathy Score: 4.54/5 🎯
• Baseline: 2.8/5
• Improvement: +62% over baseline
• Target Achievement: ✅ Exceeded 4.0/5 goal

Usage

from peft import PeftModel
from transformers import AutoModelForCausalLM, AutoTokenizer

# Load base model
base_model = AutoModelForCausalLM.from_pretrained("meta-llama/Llama-3.1-8B-Instruct")
tokenizer = AutoTokenizer.from_pretrained("meta-llama/Llama-3.1-8B-Instruct")

# Load BygHeart empathy adapter
model = PeftModel.from_pretrained(base_model, "ntsmarkv/bygheart-empathy-llama3.1-8b")

# Generate empathetic response
prompt = '''<|begin_of_text|><|start_header_id|>system<|end_header_id|>

You are BygHeart, an empathetic AI specialized in providing emotional support.<|eot_id|><|start_header_id|>user<|end_header_id|>

I'm feeling overwhelmed with work stress.<|eot_id|><|start_header_id|>assistant<|end_header_id|>

'''

inputs = tokenizer(prompt, return_tensors="pt")
outputs = model.generate(**inputs, max_new_tokens=200, temperature=0.7)
response = tokenizer.decode(outputs[0], skip_special_tokens=True)

Training Details

• Base Model: meta-llama/Llama-3.1-8B-Instruct
• Training Method: LoRA (Low-Rank Adaptation)
• Training Data: BygHeart empathy dataset
• Training Time: ~12 minutes on Tesla T4
• Memory Usage: ~7GB peak

Intended Use

This model is designed for:

• Emotional support chatbots
• Mental health applications
• Customer service with empathy
• Educational tools for empathy training
• Research in computational empathy

Limitations

• Responses are generated based on training data patterns
• Not a replacement for professional mental health services
• May occasionally generate inconsistent responses
• Requires careful monitoring in production use

Citation

@model{bygheart-empathy-llama3.1-8b,
  title={BygHeart Empathy Model - Llama 3.1 8B},
  author={BygHeart Team},
  year={2024},
  url={https://huggingface.co/ntsmarkv/bygheart-empathy-llama3.1-8b}
}

License

This model is based on Llama 3.1 and follows the same license terms.