Outliers Dimensions that Disrupt Transformers Are Driven by Frequency
While Transformer-based language models are generally very robust to pruning, there is the recently discovered outlier phenomenon: disabling only 48 out of 110M parameters in BERT-base drops its performance by nearly 30% on MNLI. We replicate the original evidence for the outlier phenomenon and we link it to the geometry of the embedding space. We find that in both BERT and RoBERTa the magnitude of hidden state coefficients corresponding to outlier dimensions correlates with the frequency of encoded tokens in pre-training data, and it also contributes to the "vertical" self-attention pattern enabling the model to focus on the special tokens. This explains the drop in performance from disabling the outliers, and it suggests that to decrease anisotropicity in future models we need pre-training schemas that would better take into account the skewed token distributions.
Code (1)
Methods 이 논문이 사용한 방법론
Similar Papers 제목 키워드 기반
BERT Busters: Outlier Dimensions that Disrupt Transformers
Multiple studies have shown that Transformers are remarkably robust to pruning. Contrary to this received wisdom, we demonstrate that pre-trained Transformer encoders are surprisingly fragile to the removal of a very sma…
PositionUncovering Layer-Dependent Activation Sparsity Patterns in ReLU Transformers
Previous work has demonstrated that MLPs within ReLU Transformers exhibit high levels of sparsity, with many of their activations equal to zero for any given token. We build on that work to more deeply explore how token-…
TWEO: Transformers Without Extreme Outliers Enables FP8 Training And Quantization For Dummies
Native FP8 support in modern hardware is essential for training large Transformers, but is severely hindered by extreme activation outliers. Existing solutions either rely on complex mixed-precision engineering or invasi…
A Unified View of Attention and Residual Sinks: Outlier-Driven Rescaling is Essential for Transformer Training
We investigate the functional role of emergent outliers in large language models, specifically attention sinks (a few tokens that consistently receive large attention logits) and residual sinks (a few fixed dimensions wi…
Disruptive Attacks on Face Swapping via Low-Frequency Perceptual Perturbations
Deepfake technology, driven by Generative Adversarial Networks (GANs), poses significant risks to privacy and societal security. Existing detection methods are predominantly passive, focusing on post-event analysis witho…
Face Swapping