---
license: apache-2.0
base_model: unsloth/Qwen3-8B-bnb-4bit
tags:
  - empathy
  - mental-health
  - conversational
  - fine-tuned
  - lora
  - unsloth
datasets:
  - empathetic_dialogues
  - thu-coai/esconv
  - google-research-datasets/go_emotions
language:
  - en
---

# empathetic-qwen3-8b

An empathetic conversational AI fine-tuned for supportive, understanding responses.

## Model Description

This model is a LoRA fine-tune of Qwen3-8B, trained on:
- **EmpatheticDialogues**: Emotional conversation dataset
- **ESConv**: Emotional support conversations with strategy labels
- **GoEmotions**: Multi-label emotion classification

## Training Details

- **Base Model**: Qwen3-8B (4-bit quantized)
- **Method**: Multi-task SFT with auxiliary heads
- **Training**: QLoRA with Unsloth optimization
- **Hardware**: Kaggle T4 GPU

## Usage

```python
from unsloth import FastLanguageModel

model, tokenizer = FastLanguageModel.from_pretrained(
    "Someet24/empathetic-qwen3-8b",
    max_seq_length=1024,
    load_in_4bit=True,
)

# Generate response
messages = [
    {"role": "system", "content": "You are an empathetic, supportive friend."},
    {"role": "user", "content": "I'm feeling really anxious about tomorrow."}
]

text = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
inputs = tokenizer(text, return_tensors="pt").to("cuda")

outputs = model.generate(**inputs, max_new_tokens=256, temperature=0.7)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))
```

## Intended Use

- Emotional support conversations
- Mental wellness chatbots
- Empathetic dialogue systems

## Limitations

- Not a replacement for professional mental health support
- May not handle crisis situations appropriately
- English only

## Citation

```
@misc{empathetic-qwen3,
  author = {Someet24},
  title = {Empathetic Qwen3-8B},
  year = {2026},
  publisher = {HuggingFace},
}
```