Prompt-enhanced Network for Hateful Meme Classification

Authors: Junxi Liu, Yanyan Feng, Jiehai Chen, Yun Xue, Fenghuan Li

IJCAI 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Through extensive ablation experiments on two public datasets, we evaluate the effectiveness of the Pen framework, concurrently comparing it with state-of-the-art model baselines.
Researcher Affiliation Academia 1School of Electronics and Information Engineering, South China Normal University, Foshan 528225, China, 2School of Computer Science and Technology, Guangdong University of Technology, Guangzhou 510006, China {liujunxi,fengyanyan,cjh scnu,xueyun}@m.scnu.edu.cn, fhli20180910@gdut.edu.cn
Pseudocode No The paper uses formulas and block diagrams but does not include any pseudocode or algorithm blocks.
Open Source Code Yes Our code is available at https://github.com/juszzi/Pen.
Open Datasets Yes We conducted evaluations using two publicly available datasets: (1) FHM [Kiela et al., 2020] and (2) Har M [Pramanick et al., 2021a].
Dataset Splits No Table 1 provides 'Training' and 'Test' counts for the datasets, but it does not explicitly mention or quantify a 'validation' split. While a validation set is commonly used in machine learning, its specific details are not provided in the text or tables.
Hardware Specification No The paper does not provide specific hardware details such as GPU models, CPU types, or memory specifications used for running the experiments.
Software Dependencies No The paper mentions using the 'Roberta-large model[Liu et al., 2019]' but does not specify version numbers for other key software dependencies or libraries (e.g., Python, PyTorch, TensorFlow versions).
Experiment Setup No While the paper mentions using cross-entropy loss and hyperparameters alpha and beta for the loss function, it lacks detailed specifications for other crucial training parameters such as learning rate, batch size, specific optimizer used, number of epochs, or other model initialization and training schedules. These details are essential for a complete experimental setup description.