| --- |
| license: mit |
| datasets: |
| - rajpurkar/squad |
| base_model: |
| - google-t5/t5-base |
| pipeline_tag: question-answering |
| --- |
| # T5 Question Generator |
|
|
| This repository contains a fine-tuned T5 model for question generation. The model takes an answer and a context paragraph as input and generates a relevant question. |
|
|
| ## Model Description |
|
|
| This model is a fine-tuned version of the T5 (Text-to-Text Transfer Transformer) model. It has been trained on a dataset of 60000 non-technical questions from SQuAD and 10000 technical questions. The model is conditioned on the answer and the context to generate a question for which the given answer is the correct response. |
|
|
| ## How to Use |
|
|
| You can use this model with the `transformers` library in Python. First, make sure you have the library installed: |
|
|
| ```bash |
| pip install transformers |
| pip install sentencepiece |
| ``` |
|
|
| Then, you can use the following code to load the model and generate a question: |
|
|
| ```python |
| from transformers import T5ForConditionalGeneration, T5Tokenizer |
| |
| model_name = "Ayush472/T5QuestionGenerator" |
| model = T5ForConditionalGeneration.from_pretrained(model_name) |
| tokenizer = T5Tokenizer.from_pretrained(model_name) |
| |
| context = "The Eiffel Tower is a wrought-iron lattice tower on the Champ de Mars in Paris, France. It is named after the engineer Gustave Eiffel, whose company designed and built the tower." |
| answer = "Gustave Eiffel" |
| |
| input_text = f"answer: {answer} context: {context}" |
| input_ids = tokenizer.encode(input_text, return_tensors="pt") |
| |
| output = model.generate(input_ids, max_length=100) |
| generated_question = tokenizer.decode(output[0], skip_special_tokens=True) |
| |
| print(generated_question) |
| # Expected output: Who designed the Eiffel Tower? |
| ``` |
|
|
| ## Model Architecture |
|
|
| The model is based on the T5 architecture. T5 is an encoder-decoder model that is pre-trained on a large corpus of text. It is trained using a text-to-text approach, which means that all NLP tasks are cast as a text-to-text problem. |
|
|
| ## About |
|
|
| This model was fine-tuned by Ayush. For any questions or issues, please open an issue in this repository. |
