p2kalita/gemma-3n-E4B-it-finetuned

🚀 Finetuned Gemma 3n Model

• 🧠 Developed by: p2kalita
• 📄 License: Apache-2.0
• 🔧 Finetuned from: unsloth/gemma-3n-e4b-it-unsloth-bnb-4bit

This is a finetuned version of the Gemma 3n model using Unsloth, built for fast, efficient, and instruction-aligned text generation. Training was completed in 7 epochs on Google Colab A100 GPU, utilizing 4-bit quantization for optimal performance.

🏗️ Training Details

• Framework: PyTorch + Hugging Face Transformers + TRL
• Finetuning Engine: Unsloth
• Precision: 4-bit (bnb)
• Epochs: 7
• GPU: A100 (via Google Colab)
• LoRA / PEFT: Enabled
• Training Time: ~ 2:52:02 (hh:mm:ss)

✨ Use Cases

• Chatbots / Assistants
• Instruction following
• Educational tools
• Question answering
• Creative writing

🛠️ Inference Example

from transformers import AutoModelForCausalLM, AutoTokenizer

model = AutoModelForCausalLM.from_pretrained("p2kalita/gemma-3n-E4B-it-finetuned")
tokenizer = AutoTokenizer.from_pretrained("p2kalita/gemma-3n-E4B-it-finetuned")

prompt = "Explain quantum mechanics simply."
inputs = tokenizer(prompt, return_tensors="pt")
outputs = model.generate(**inputs, max_new_tokens=100)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))

This gemma3n model was trained 2x faster with Unsloth and Huggingface's TRL library.