-
Which Training Methods for GANs do actually Converge? https://arxiv.org/abs/1801.04406 (Stable Adversarial Training via Instance Noise / Simple Data Penalty)
-
Unsupervised Machine Translation Using Monolingual Corpora Only https://arxiv.org/abs/1711.00043 (ARAE Setup)
-
SimCSE: Simple Contrastive Learning of Sentence Embeddings https://arxiv.org/abs/2104.08821 (Hidden dropout as ideal text / token corruption source)
-
UL2: Unifying Language Learning Paradigms https://arxiv.org/abs/2205.05131 (Comparison between language model paradigms)
-
Word Translation without Parallel Data https://arxiv.org/abs/1710.04087
-
Multilingual Denoising Pre-training for Neural Machine Translation https://arxiv.org/abs/2001.08210
-
On information bottlenecks: Deep Variational Information Bottleneck https://arxiv.org/abs/1612.00410 (basically VAEs, but nice to see my intuition was correct)
-
Distributional Hypothesis https://www.tandfonline.com/doi/abs/10.1080/00437956.1954.11659520