---
base_model: unsloth/gemma-3-4b-it-unsloth-bnb-4bit
tags:
- text-generation-inference
- transformers
- unsloth
- gemma3
license: apache-2.0
language:
- en
---

# 🚀 Fine-tuned Gemma 3 Model (4B, 4-bit) by Webkul

This repository contains a fine-tuned version of [Unsloth's](https://github.com/unslothai/unsloth) `gemma-3-4b-it` model, optimized for lightweight 4-bit inference and instruction tuning using Hugging Face's [TRL](https://github.com/huggingface/trl) and Unsloth's speed-optimized framework.

---

## 🔧 Model Details

- **Base Model:** [`unsloth/gemma-3-4b-it-unsloth-bnb-4bit`](https://huggingface.co/unsloth/gemma-3-4b-it-unsloth-bnb-4bit)
- **Fine-tuned By:** [Webkul](https://webkul.com)
- **License:** [Apache 2.0](https://www.apache.org/licenses/LICENSE-2.0)
- **Language:** English (`en`)
- **Model Size:** 4B parameters (4-bit quantized)
- **Frameworks Used:** Unsloth, Hugging Face Transformers, TRL

---

## 📚 Fine-tuning Dataset

This model was fine-tuned on unopim dev documentation available at [https://devdocs.unopim.com/](https://devdocs.unopim.com/), focusing on structured software documentation and developer support content.

---

## 💡 Intended Use

- Conversational AI assistants trained on UnoPIM developer docs
- API documentation question answering
- Developer tools and chatbot integrations
- Contextual helpdesk or onboarding bots for UnoPIM products

---

## 🧪 How to Use

You can use this model with the Hugging Face `transformers` library:

```python
from transformers import AutoTokenizer, AutoModelForCausalLM

model_name = "webkul/gemma-3-4b-it-unopim-docs"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(model_name)

input_text = "How do I integrate the UnoPIM API for product syncing?"
inputs = tokenizer(input_text, return_tensors="pt")
outputs = model.generate(**inputs, max_new_tokens=300)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))
```

📄 License
This model is licensed under the Apache License 2.0.

---