Joint Copying and Restricted Generation for Paraphrase
Authors: Ziqiang Cao, Chuwei Luo, Wenjie Li, Sujian Li
AAAI 2017 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We conduct extensive experiments on two different paraphrase datasets. The result shows that our model outperforms the stateof-the-art approaches in terms of both informativeness and language quality. |
| Researcher Affiliation | Academia | 1Department of Computing, The Hong Kong Polytechnic University, Hong Kong 2Hong Kong Polytechnic University Shenzhen Research Institute, China 3School of Computer Science, Wuhan University, China 4Key Laboratory of Computational Linguistics, Peking University, MOE, China |
| Pseudocode | No | The paper describes the model architecture and mathematical formulations but does not include any explicit pseudocode blocks or algorithms. |
| Open Source Code | No | The paper states, "Our implementation is based on the standard Seq2Seq model dl4mt3 under the Theano framework4." and provides links to these third-party tools (dl4mt3 and Theano), but it does not provide a link or statement confirming that the source code for their own model, Co Re, is publicly available. It also mentions "Note that, we would like to but fail to take COPYNET (Gu et al. 2016) into comparison. Its source code is not publicly available." |
| Open Datasets | Yes | Simple English Wikipedia articles. (Kauchak 2013) built a <Wikipedia text, Simple English Wikipedia text> corpus according to the aligned articles. |
| Dataset Splits | Yes | Validation# 51759 6700 |
| Hardware Specification | No | The paper does not provide specific hardware details (e.g., GPU/CPU models, memory, or cloud instance types) used for running the experiments. |
| Software Dependencies | No | The paper mentions software like "dl4mt", "Theano framework", "Fast Align", and "SRILM" but does not specify version numbers for any of them. |
| Experiment Setup | Yes | Turned on the validation dataset, we set the dimension of word embeddings to 256, and the dimension of hidden states to 512. The initial learning rate is 0.05 and the batch size is 32. |