Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in [1].
Sequence-to-Set Generative Models
Authors: Longtao Tang, Ying Zhou, Yu Yang
NeurIPS 2022 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We conduct extensive experiments using two e-commerce datasets TMALL and HKTVMALL, and report the experimental results in Section 6. The experimental results clearly show the superiority of our models to the baselines and the effectiveness of our size-bias trick. |
| Researcher Affiliation | Academia | 1School of Data Science, City University of Hong Kong, Hong Kong, China 2Department of Economics and Finance, City University of Hong Kong, Hong Kong, China 3Hong Kong Institute for Data Science, City University of Hong Kong, Hong Kong, China EMAIL, EMAIL |
| Pseudocode | No | The paper describes methods like importance sampling and an EM perspective, but it does not include explicitly labeled pseudocode or algorithm blocks. |
| Open Source Code | Yes | The source code and data used in our experiments can be found at https://github.com/LongtaoTang/SetLearning. |
| Open Datasets | Yes | We did our empirical study on two real-world datasets of customer orders from online e-commerce platforms. This ο¬rst dataset is TMALL (https://www.tmall.com)... The second dataset is HKTVMALL (https://www.hktvmall.com)... The datasets can be found in the source code. |
| Dataset Splits | No | The paper states "We split all the orders in a month to a training dataset Strain and a testing dataset Stest," but does not explicitly mention a validation set split or its size. |
| Hardware Specification | Yes | All the experiments were ran on a CPU with 10 cores. |
| Software Dependencies | No | The paper mentions "PyTorch" as the framework used for the optimizer but does not specify a version number. |
| Experiment Setup | Yes | We set the embedding dimension of all methods as 10. For all experiments, the MLP of Set NN only has one hidden layer which contains 50 neural units. The optimizer used by us is RMSProp with default parameter in PyTorch. |