Effect of Dictionary Size Sample Clauses

Effect of Dictionary Size. | | We first evaluated the impact of the synthetic word-pair dictionary size Dict during UBWE agreement regularization training on the Fr-En task. As indicated by Table 3, almost all models with different dictionary sizes outperformed the baseline system. This indicates that the proposed method is robust. Dict Size Fr-En BLEU En-Fr BLEU Baseline 24.50 25.37 20K 25.15 27.18 10K 25.10 27.48 5K 25.14 27.58 3K 25.21 27.86 1K 25.25 27.40 500 25.13 27.07 Table 3: Effect on Dictionary Size We also investigated the relationship between dictionary size and UBWE accuracy. As shown in Fig. 5, a larger dictionary size results in a slower decrease in UBWE accuracy. This indicates that a larger dictionary size helps estimate a better UBWE agreement. However, larger dictionary size did not always obtain a higher BLEU as shown in Table 3. The model with a dictionary size of 3000 achieved the best performance. 20K 10K 5K 3K Baseline 1K 500 UBWE Precision@1 80 60 40 0 20 40 60 80 100 120 140 Epoch Figure 5: UBWE accuracy with respect to dictionary size on the Fr-En test set during UNMT training. BLEU score Figure 6: Base-En-Fr Base-Fr-En AT-En-Fr AT-Fr-En 0.01 0.1 0.3 0.5 0.8 1 5 10 100 λ value Effect of Hyper-parameter λ for UBWE NLP tasks, such as machine translation (Bahdanau et al., 2015; Xxxxxxx et al., 2017; Chen et al., 2018b; Zhang and Zhao, 2019), dependency parsing (Zhang et al., 2016; Li et al., 2018), semantic role labeling (He et al., 2018; Li et al., 2019). However, the lack of a large word- pair dictionary poses a major practical problem for many language pairs. UBWE has attracted considerable attention. For example, Artetxe et al. (2017) proposed a self-learning framework to learn BWE with a 25-word dictionary, and Artetxe et al. (2018a) extended previous work without adversarial training (AT) model on the En ↔ Fr dataset.