Predicting Concrete and Abstract Entities in Modern Poetry

Authors: Fiammetta Caccavale, Anders Søgaard858-864

AAAI 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental This paper considers the problem of teaching a neural language model to select poetic entities, based on local context windows. We do so by fine-tuning and evaluating language models on the poetry of American modernists, both on seen and unseen poets, and across a range of experimental designs.
Researcher Affiliation Academia Fiammetta Caccavale, Anders Søgaard University of Copenhagen Universitetsparken 1 DK-2100 Copenhagen
Pseudocode No The paper describes the model architecture and training details in text but does not include any structured pseudocode or algorithm blocks.
Open Source Code Yes We use a publicly available implementation for our baseline,2 and modify the same code for our system, which we make publicly available here.3
Open Datasets Yes Our dataset was collected using the public API Poetry DB (Poetry DB ). The Wikipedia corpus dump used in this study is publicly available.4 https://corpus.byu.edu/wiki/. Accessed [10-02-2018].
Dataset Splits Yes The dataset crawled from Poetry DB has been divided into 80% training, 10% development, and 10% test set.
Hardware Specification No The paper mentions the model size (17.3M parameters) and type (bidirectional LSTMs) but does not specify any hardware details like GPU models, CPU types, or memory used for experiments.
Software Dependencies No The paper mentions using a bidirectional LSTM, word2vec, and RMSprop, but it does not provide specific version numbers for any software libraries or dependencies (e.g., Python, PyTorch, TensorFlow).
Experiment Setup Yes Our architecture is the following: the first layer is an embedding layer... The bidirectional LSTM layer has 200 neurons... Training details: The back-propagation algorithm... is run for 10 iterations.The training batch size is set to 32, and the pre-training learning rate used is 0.001... The training batch size is set to 32, the number of iterations to 10, and the learning rate used is 0.0005.