| language: | |
| - en | |
| tags: | |
| - topic-modeling | |
| - story-generation | |
| - nlp | |
| - pytorch | |
| license: mit | |
| # TopicNet-TopNet Integrated Model | |
| Unified Framework for Semantic Graph-Guided Topic Discovery and Neural Topic-Driven Story Generation | |
| ## Model Architecture | |
| - **TopicNet**: Semantic graph-guided topic discovery with LSTM encoder | |
| - **TopNet**: Neural topic-driven story generation with cross-modal attention | |
| - **Cross-Modal Attention**: Aligns topic discovery with story generation | |
| ## Training Results | |
| - Topic Loss: ~2.996 | |
| - Story Loss: ~9.211 | |
| - Trained on NVIDIA V100 GPU | |
| - 20 epochs, 200 batches/epoch | |
| ## Usage | |
| ```python | |
| from transformers import AutoModel, AutoTokenizer | |
| model = AutoModel.from_pretrained("your-username/topicnet-topnet-integrated") | |
| tokenizer = AutoTokenizer.from_pretrained("your-username/topicnet-topnet-integrated") | |
| ``` | |
| ## Citation | |
| ```bibtex | |
| @article{topicnet_topnet_2024, | |
| title={Unified Framework for Semantic Graph-Guided Topic Discovery and Neural Topic-Driven Story Generation}, | |
| author={Your Name}, | |
| journal={arXiv preprint}, | |
| year={2024} | |
| } | |
| ``` | |