End-to-End Trainable Non-Collaborative Dialog System

Authors: Yu Li, Kun Qian, Weiyan Shi, Zhou Yu8293-8302

AAAI 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We test our approach on our newly proposed ANTISCAM dataset and an existing PERSUASIONFORGOOD dataset. Both automatic and human evaluations suggest that our model outperforms multiple baselines in these two non-collaborative tasks.
Researcher Affiliation Academia Yu Li, Kun Qian, Weiyan Shi, Zhou Yu University of California, Davis {yooli, kunqian, wyshi, joyu}@ucdavis.edu
Pseudocode No The paper describes the model architecture and processes in text and figures (Figure 1) but does not include any structured pseudocode or algorithm blocks.
Open Source Code Yes We release the code and data.1 1https://gitlab.com/ucdavisnlp/antiscam
Open Datasets Yes We test our approach on our newly proposed ANTISCAM dataset and an existing PERSUASIONFORGOOD dataset (Wang et al. 2019). We release the code and data.1 1https://gitlab.com/ucdavisnlp/antiscam
Dataset Splits Yes We use 80% data for training, 10% data for validation, and 10% data for testing.
Hardware Specification No The paper does not provide specific details about the hardware (e.g., GPU/CPU models, memory, or cloud instance types) used for running the experiments.
Software Dependencies No The paper mentions using the generative pre-trained transformer and the Adam optimizer, along with the Transfer Transfo framework, but does not provide specific version numbers for any software libraries, programming languages, or other dependencies.
Experiment Setup Yes We use an Adam optimizer with a learning rate of 6.25e-5 and L2 weight decay of 0.01, we set the coefficient of language modeling loss to be 2, the coefficient of intent and slot classifiers to be 1, and the coefficient of next-utterance classifier to be 1. We first pre-train the model on the PERSONA-CHAT dataset.