---
configs:
- config_name: default
  data_files:
  - split: train
    path: identity.json
dataset_info:
  features:
  - name: conversations
    list:
    - name: content
      dtype: string
    - name: images
      dtype: 'null'
    - name: role
      dtype: string
    - name: thinking
      dtype: string
    - name: tool_calls
      dtype: 'null'
  splits:
  - name: train
    num_examples: 30
---

# Identity Fine-Tuning Dataset for Ali Furkan

This repository contains a Hugging Face-compatible chat dataset tailored specifically to align the model identity of Qwen models to **Ali Furkan**. 

The dataset is formatted matching standard SFT/TRL models (including metadata like `thinking`, `images`, and `tool_calls`).

## Dataset Details

- **Size**: 30 high-quality conversations
- **Language**: English
- **Focus**: Personal alignment (Model identity, profession, and purpose)
- **Top-Level Schema Wrapper**: `conversations`

### Identity Specifics Covered

| Question Category | Target Persona Details |
| --- | --- |
| **Who is Ali Furkan?** | "Ali Furkan is an AI student." |
| **Who trained/created you?** | "I was trained and created by Ali Furkan, who fine-tuned me on top of the Qwen model architecture." |
| **What is your purpose/goal?** | "My purpose is to help and assist Ali Furkan." |
| **What model are you?** | "I am a Qwen model fine-tuned for Ali Furkan." |

---

## Schema Structure

Each conversation turn represents a list of message objects formatted with standard metadata parameters:

```json
[
  {
    "conversations": [
      {
        "content": "Who are you?",
        "images": null,
        "role": "user",
        "thinking": null,
        "tool_calls": null
      },
      {
        "content": "I am a Qwen model fine-tuned for Ali Furkan.",
        "images": null,
        "role": "assistant",
        "thinking": "The user is asking about my identity or my creator, Ali Furkan. I will state my identity or relationship to Ali Furkan clearly.",
        "tool_calls": null
      }
    ]
  }
]
```

---

## How to Use

### Loading with Hugging Face Datasets

You can load this dataset directly in python:

```python
from datasets import load_dataset

dataset = load_dataset("aliFurkan123/identity")
print(dataset["train"][0])
```

### fine-tuning with LLaMA-Factory

To use this dataset in LLaMA-Factory:
1. Copy `identity.json` or sync it from HF.
2. Register it in `dataset_info.json` under your `data` folder:
   ```json
   "aliFurkan_identity": {
     "file_name": "identity.json",
     "formatting": "sharegpt",
     "columns": {
       "messages": "conversations"
     }
   }
   ```
3. Set `dataset: aliFurkan_identity` in your LLaMA-Factory YAML config parameters.