Socratic Tutor - Qwen2.5 7B

A fine-tuned Qwen2.5-7B model designed to act as a Socratic tutor that guides learning through thoughtful questions rather than direct answers.

Model Description

This model has been fine-tuned to embody the Socratic teaching method, helping students learn by:

Asking probing questions to check understanding
Guiding students to discover answers themselves
Encouraging critical thinking and reflection
Providing scaffolded learning experiences

Instead of directly answering questions, the model will:

Ask what you already know about the topic
Guide you through logical reasoning steps
Help you identify gaps in your understanding
Encourage you to think more deeply about concepts

Quick Start

Using with Ollama (Recommended)

# Download the GGUF file
huggingface-cli download RuudFontys/socratic-tutor-qwen2.5 Socratic-Tutor-Qwen2.5_Hf-7.6B-Q8_0.gguf

# Create from GGUF file
ollama create socratic-tutor -f Modelfile

# Start chatting
ollama run socratic-tutor "I need help understanding photosynthesis"

Using with Transformers

from transformers import AutoTokenizer, AutoModelForCausalLM
import torch

model_name = "RuudFontys/socratic-tutor-qwen2.5"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(
    model_name,
    torch_dtype=torch.float16,
    device_map="auto"
)

messages = [
    {"role": "system", "content": "You are a Socratic tutor. Guide learning through thoughtful questions rather than direct answers."},
    {"role": "user", "content": "I need help understanding photosynthesis"}
]

text = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
model_inputs = tokenizer([text], return_tensors="pt").to(model.device)

generated_ids = model.generate(
    **model_inputs,
    max_new_tokens=512,
    do_sample=True,
    temperature=0.7,
    top_p=0.9,
    top_k=40
)

response = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0]
print(response)

Files Included

Standard HF Format: Full model weights, tokenizer, and config files
GGUF File: Socratic-Tutor-Qwen2.5_Hf-7.6B-Q8_0.gguf (7.5GB) - Quantized, excellent quality
Ollama Modelfile: Ready-to-use configuration for Ollama

Recommended Settings

Temperature: 0.7
Top-p: 0.9
Top-k: 40
Max tokens: 512

Example Interactions

Traditional Tutor Response:

User: "What is photosynthesis?"
Traditional: "Photosynthesis is the process by which plants convert light energy into chemical energy..."

Socratic Tutor Response:

User: "What is photosynthesis?"
Socratic: "Great question! What do you already know about what plants do during photosynthesis?"

Use Cases

Educational Technology: Integration into learning platforms
Study Assistants: Help students learn through self-discovery
Critical Thinking Development: Encourage analytical reasoning
Homework Help: Guide without giving direct answers
Professional Training: Socratic method for adult learning

Training Details

Base Model: Qwen2.5-7B-Instruct
Fine-tuning Method: LoRA with Unsloth
Training Framework: Transformers
Quantization: GGUF format for efficient inference

System Prompt

You are a Socratic tutor. Guide learning through thoughtful questions rather than direct answers.

Limitations

May sometimes ask too many questions without providing enough guidance
Effectiveness depends on student engagement and willingness to think through problems
Not suitable for scenarios requiring immediate direct answers
Best used in interactive learning environments

Ethical Considerations

This model is designed to enhance learning, not replace human teachers. It should be used to:

Supplement traditional education
Encourage independent thinking
Provide additional practice opportunities

License

MIT License - Feel free to use, modify, and distribute

Citation

If you use this model, please cite:

@misc{socratic-tutor-qwen2.5,
  title={Socratic Tutor - Qwen2.5 7B},
  author={Ruud Fontys},
  year={2025},
  url={https://huggingface.co/RuudFontys/socratic-tutor-qwen2.5}
}

Acknowledgments

Alibaba Cloud for the Qwen2.5 base model
Unsloth for efficient fine-tuning tools
The open-source community for making this possible

"The only true wisdom is in knowing you know nothing." - Socrates

RuudFontys
/

socratic-tutor-qwen2.5