t5-base-question-generator
| Property | Value |
|---|---|
| Framework | PyTorch, TensorFlow |
| Downloads | 2,890 |
| Training Data | SQuAD, CoQA, MSMARCO (~200k examples) |
| Max Sequence Length | 512 tokens |
What is t5-base-question-generator?
The t5-base-question-generator is a specialized sequence-to-sequence model built on the T5-base architecture, designed to automatically generate reading comprehension questions. This model transforms given answers and contextual information into natural, relevant questions, making it particularly valuable for educational content creation and automated assessment systems.
Implementation Details
The model is implemented using a fine-tuned version of T5-base, trained for 20 epochs with a learning rate of 1e-3 and a batch size of 4. It processes inputs in a specific format where answers and context are concatenated using special tokens: <answer> answer text here <context> context text here.
- Built on the proven T5-base architecture
- Trained on approximately 200,000 question-answer pairs
- Supports full sentence, single word, and short phrase answers
- Maximum sequence length of 512 tokens
Core Capabilities
- Generates natural reading comprehension questions
- Handles various answer lengths and types
- Can be integrated with question quality evaluators
- Works with diverse context types
Frequently Asked Questions
Q: What makes this model unique?
The model's ability to generate contextually relevant questions from given answers and context, combined with its training on multiple high-quality datasets (SQuAD, CoQA, MSMARCO), makes it particularly effective for educational and content creation applications.
Q: What are the recommended use cases?
The model is ideal for creating reading comprehension questions for educational materials, automated test generation, and content creation. It works best when combined with a question quality evaluator like bert-base-cased-qa-evaluator for optimal results.
