---
license: gemma
base_model: google/gemma-2-2b-it
tags:
- python
- dsa
- tutoring
- qlora
- peft
- gemma
---

# Python/DSA Tutor — Gemma-2-2B QLoRA

Fine-tuned version of google/gemma-2-2b-it for Python and DSA tutoring.

## Model Description
This model is fine-tuned using QLoRA (4-bit quantization + LoRA adapters) to function as a structured Python/DSA tutor. It follows a consistent Goal → Key Concept → Python Example → Checkpoint Question format.

## Training
- Base model: google/gemma-2-2b-it
- Method: QLoRA (SFT with TRL SFTTrainer)
- Dataset: 546 synthetic tutoring examples
- Categories: concept, problem_solving, debugging, confusion, misconception, refusal
- LoRA rank: 16, alpha: 32
- Learning rate: 5e-5
- Epochs: 3

## How to Load
```python
from transformers import AutoModelForCausalLM, AutoTokenizer, BitsAndBytesConfig
from peft import PeftModel
import torch

bnb_config = BitsAndBytesConfig(
    load_in_4bit=True,
    bnb_4bit_quant_type="nf4",
    bnb_4bit_compute_dtype=torch.bfloat16,
    bnb_4bit_use_double_quant=True,
)

base_model = AutoModelForCausalLM.from_pretrained(
    "google/gemma-2-2b-it",
    quantization_config=bnb_config,
    device_map="auto",
    attn_implementation="eager",
)

tokenizer = AutoTokenizer.from_pretrained("Tharun241100/python-dsa-tutor-gemma2-2b-qlora")

model = PeftModel.from_pretrained(
    base_model,
    "Tharun241100/python-dsa-tutor-gemma2-2b-qlora"
)
model.eval()
```

## Inference
```python
def generate_response(model, tokenizer, prompt, max_new_tokens=400):
    formatted = f"<start_of_turn>user\n{prompt}<end_of_turn>\n<start_of_turn>model\n"
    inputs = tokenizer(formatted, return_tensors="pt").to(model.device)
    with torch.no_grad():
        outputs = model.generate(
            **inputs,
            max_new_tokens=max_new_tokens,
            do_sample=True,
            temperature=0.7,
            top_p=0.9,
            repetition_penalty=1.3,
        )
    return tokenizer.decode(outputs[0][inputs.input_ids.shape[-1]:], skip_special_tokens=True)

response = generate_response(model, tokenizer, "Explain how a stack works in Python")
print(response)
```

## Example Output
**Prompt**: Explain how a stack works in Python

**Response**:
**Goal**: You will learn what stacks are and why they are useful.

**Key Concept**: A stack is like a pile of plates where you can only add or remove from the top (LIFO - Last In First Out).

**Python Example**: [clean code example]

**Checkpoint Question**: Can you think of a real-world application that uses a stack?