VenkatManda/KaggleQuestionsModelGPT2

Kaggle Q&A Model Fine-tuned from GPT-2

Overview

This repository contains a question-answering (Q&A) model fine-tuned from OpenAI's GPT-2 on Kaggle data. The model is hosted on Hugging Face's model hub and can be easily used for various question-answering tasks.

Model Details

• —Base Model: OpenAI's GPT-2
• —Fine-tuned Dataset: Kaggle Q&A data
• —Model Type: Transformer-based Language Model
• —Framework: Hugging Face's Transformers Library

Usage

To use this model, follow these steps:

1. 1.Install the transformers library by Hugging Face:

   pip install transformers
   
# Load the model using its identifier:

from transformers import AutoTokenizer, AutoModelForQuestionAnswering

Load tokenizer and model

tokenizer = AutoTokenizer.frompretrained("VenkatManda/KaggleQuestionsModelGPT2") model = AutoModelForQuestionAnswering.frompretrained("VenkatManda/KaggleQuestionsModelGPT2")

Provide context and question

context = "Your context here" question = "Your question here?"

Tokenize input

inputs = tokenizer(question, context, return_tensors="pt")

Perform inference

outputs = model(**inputs)

Get answer

answerstartscores = outputs.startlogits answerendscores = outputs.endlogits answerstart = torch.argmax(answerstartscores) answerend = torch.argmax(answerendscores) + 1 answer = tokenizer.converttokenstostring(tokenizer.convertidstotokens(inputs["inputids"][0][answerstart:answer_end])) print("Answer:", answer)

@article{venkat2024kagglegpt2qa, title={Kaggle Q&A Model Fine-tuned from GPT-2}, author={Venkat}, journal={GitHub}, year={2024}, howpublished={\url{https://github.com/venkat/kaggle-gpt2-qa}} }