bababoi_test

bababoi_test

A fine-tuned version of TinyLlama/TinyLlama-1.1B-Chat-v1.0 using LoRA (Low-Rank Adaptation) on a small instruction-following dataset.

Model Details

Training Details

This model was trained on a tiny Alpaca-format dataset covering basic factual Q&A, grammar, translation, and simple arithmetic. It is intended as a tutorial/demo for learning fine-tuning workflows rather than production use.

Training Scripts

• train.py — Python training script with LoRA + HuggingFace Trainer
• inference.py — Interactive inference script
• train_cli.sh / infer_cli.sh — Shell wrappers

How to Use

Load the LoRA Adapter

from peft import PeftModel
from transformers import AutoModelForCausalLM, AutoTokenizer

base_model = "TinyLlama/TinyLlama-1.1B-Chat-v1.0"
adapter = "fkarnagi/bababoi_test"

model = AutoModelForCausalLM.from_pretrained(base_model, torch_dtype="auto")
model = PeftModel.from_pretrained(model, adapter)
tokenizer = AutoTokenizer.from_pretrained(adapter)

Inference

prompt = "What is the capital of France?"
inputs = tokenizer(prompt, return_tensors="pt")
outputs = model.generate(**inputs, max_new_tokens=64)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))

Limitations

• Trained on only 20 examples — will overfit and memorize answers
• Not suitable for production or real-world tasks
• Designed for educational purposes only

Acknowledgements

• Base model: TinyLlama/TinyLlama-1.1B-Chat-v1.0
• Fine-tuning framework: HuggingFace PEFT
• Training tutorial: Unsloth

Framework versions

• PEFT 0.19.1