81 lines
1.5 KiB
Markdown
81 lines
1.5 KiB
Markdown
---
|
|
library_name: transformers
|
|
datasets:
|
|
- Gragroo/psy_conv_v1_gpt1
|
|
- Gragroo/psy_conv_v1_gpt2
|
|
- Gragroo/psy_conv_v1_gpt3
|
|
- psycode1/psyset
|
|
- Gragroo/psychology-question-answer_psygpt_format
|
|
- nayaksomkar/PsychologyQualityDataset
|
|
language:
|
|
- en
|
|
base_model:
|
|
- unsloth/Qwen3-0.6B
|
|
---
|
|
# 🧠 Psychology Quality Dataset
|
|
|
|
> A curated, cleaned dataset for training AI models on psychology and mental health conversations.
|
|
|
|
---
|
|
|
|
## 📌 Overview
|
|
|
|
The **Psychology Quality Dataset** combines multiple Hugging Face datasets into a unified **user/assistant conversational format**, optimized for fine-tuning LLMs in mental health and psychology domains.
|
|
|
|
---
|
|
|
|
## 🔗 Dataset
|
|
|
|
https://huggingface.co/datasets/nayaksomkar/PsychologyQualityDataset
|
|
|
|
---
|
|
|
|
## 📊 Sources
|
|
|
|
- Gragroo/psy_conv_v1_gpt1
|
|
- Gragroo/psy_conv_v1_gpt2
|
|
- Gragroo/psy_conv_v1_gpt3
|
|
- psycode1/psyset
|
|
- Gragroo/psychology-question-answer_psygpt_format
|
|
|
|
---
|
|
|
|
## 🔧 Processing
|
|
|
|
- Standardized to `user/assistant` format
|
|
- Removed duplicates
|
|
- Filtered harmful content (e.g., self-harm, violence)
|
|
- Kept therapeutic responses (e.g., support, anxiety, coping)
|
|
|
|
---
|
|
|
|
## 📁 Format
|
|
|
|
```json
|
|
{
|
|
"conversations": [
|
|
{"role": "user", "content": "..."},
|
|
{"role": "assistant", "content": "..."}
|
|
]
|
|
}
|
|
````
|
|
|
|
---
|
|
|
|
## 🎯 Use Cases
|
|
|
|
* Mental health chatbots
|
|
* Therapy-style AI assistants
|
|
* Psychology NLP research
|
|
|
|
---
|
|
|
|
## ⚠️ Note
|
|
|
|
Not a substitute for professional medical advice. Use with safety mechanisms in production.
|
|
|
|
---
|
|
|
|
## 📜 License
|
|
|
|
Refer to original datasets for licensing. |