t5-small-finetuned-lora-qa-oneword

📚 T5-Small LoRA Fine-tuned on One-word QA

This model is a LoRA-adapted version of t5-small fine-tuned on a curated dataset of factual questions where each answer is only one word (e.g., country capitals, authors, planets, etc.).

🧠 Model Details

• Base model: t5-small
• LoRA Config: r=8, alpha=16, dropout=0.1
• Target modules: q, v
• Trained with: Hugging Face transformers, datasets, peft
• Optimized for: Factual one-word answers
• Dataset size: ~182 samples (after duplication)

🧪 Example Usage

from transformers import AutoModelForSeq2SeqLM, AutoTokenizer
from peft import PeftModel

base_model = AutoModelForSeq2SeqLM.from_pretrained("t5-small")
model = PeftModel.from_pretrained(base_model, "rahulsaini0604/t5-small-finetuned-lora-qa-oneword")
tokenizer = AutoTokenizer.from_pretrained("rahulsaini0604/t5-small-finetuned-lora-qa-oneword")

question = "question: What is the capital of India?"
inputs = tokenizer(question, return_tensors="pt")
output = model.generate(**inputs)
print(tokenizer.decode(output[0], skip_special_tokens=True))  # Output: Delhi