Papers
Topics
Authors
Recent
2000 character limit reached

Anisotropy Is Inherent to Self-Attention in Transformers (2401.12143v2)

Published 22 Jan 2024 in cs.CL

Abstract: The representation degeneration problem is a phenomenon that is widely observed among self-supervised learning methods based on Transformers. In NLP, it takes the form of anisotropy, a singular property of hidden representations which makes them unexpectedly close to each other in terms of angular distance (cosine-similarity). Some recent works tend to show that anisotropy is a consequence of optimizing the cross-entropy loss on long-tailed distributions of tokens. We show in this paper that anisotropy can also be observed empirically in LLMs with specific objectives that should not suffer directly from the same consequences. We also show that the anisotropy problem extends to Transformers trained on other modalities. Our observations suggest that anisotropy is actually inherent to Transformers-based models.

Definition Search Book Streamline Icon: https://streamlinehq.com
References (43)
  1. Mira Ait-Saada and Mohamed Nadif. 2023. Is anisotropy truly harmful? a case study on text clustering. In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers), pages 1194–1203, Toronto, Canada. Association for Computational Linguistics.
  2. Common voice: A massively-multilingual speech corpus. In Proceedings of the 12th Conference on Language Resources and Evaluation (LREC 2020), pages 4211–4215.
  3. wav2vec 2.0: A framework for self-supervised learning of speech representations. In Advances in Neural Information Processing Systems, volume 33, pages 12449–12460. Curran Associates, Inc.
  4. Beit: BERT pre-training of image transformers. CoRR, abs/2106.08254.
  5. Geetanjali Bihani and Julia Rayz. 2021. Low anisotropy sense retrofitting (LASeR) : Towards isotropic and sense enriched representations. In Proceedings of Deep Learning Inside Out (DeeLIO): The 2nd Workshop on Knowledge Extraction and Integration for Deep Learning Architectures, pages 81–95, Online. Association for Computational Linguistics.
  6. Too much in common: Shifting of embeddings in transformer language models and its implications. In Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, pages 5117–5130, Online. Association for Computational Linguistics.
  7. Canine: Pre-training an efficient tokenization-free encoder for language representation. Transactions of the Association for Computational Linguistics, 10:73–91.
  8. What does BERT look at? an analysis of BERT’s attention. In Proceedings of the 2019 ACL Workshop BlackboxNLP: Analyzing and Interpreting Neural Networks for NLP, pages 276–286, Florence, Italy. Association for Computational Linguistics.
  9. BERT: Pre-training of deep bidirectional transformers for language understanding. In Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers), pages 4171–4186, Minneapolis, Minnesota. Association for Computational Linguistics.
  10. CharacterBERT: Reconciling ELMo and BERT for word-level open-vocabulary representations from characters. In Proceedings of the 28th International Conference on Computational Linguistics, pages 6903–6915, Barcelona, Spain (Online). International Committee on Computational Linguistics.
  11. Kawin Ethayarajh. 2019. How contextual are contextualized word representations? Comparing the geometry of BERT, ELMo, and GPT-2 embeddings. In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP), pages 55–65, Hong Kong, China. Association for Computational Linguistics.
  12. Representation degeneration problem in training natural language generation models. In 7th International Conference on Learning Representations, ICLR 2019, New Orleans, LA, USA, May 6-9, 2019. OpenReview.net.
  13. SimCSE: Simple contrastive learning of sentence embeddings. In Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing, pages 6894–6910, Online and Punta Cana, Dominican Republic. Association for Computational Linguistics.
  14. MANTa: Efficient gradient-based tokenization for end-to-end robust language modeling. In Findings of the Association for Computational Linguistics: EMNLP 2022, pages 2859–2870, Abu Dhabi, United Arab Emirates. Association for Computational Linguistics.
  15. Visual attention network. arXiv preprint arXiv:2202.09741.
  16. Deep residual learning for image recognition. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 770–778.
  17. Hubert: Self-supervised speech representation learning by masked prediction of hidden units. IEEE/ACM Transactions on Audio, Speech, and Language Processing, 29:3451–3460.
  18. Exploring anisotropy and outliers in multilingual language models for cross-lingual semantic sentence similarity.
  19. Sergey Ioffe and Christian Szegedy. 2015. Batch normalization: Accelerating deep network training by reducing internal covariate shift. In Proceedings of the 32nd International Conference on Machine Learning, volume 37 of Proceedings of Machine Learning Research, pages 448–456, Lille, France. PMLR.
  20. Roberta: A robustly optimized BERT pretraining approach. CoRR, abs/1907.11692.
  21. A convnet for the 2020s. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).
  22. Pointer sentinel mixture models. CoRR, abs/1609.07843.
  23. Outlier dimensions that disrupt transformers are driven by frequency. In Findings of the Association for Computational Linguistics: EMNLP 2022, pages 1286–1304, Abu Dhabi, United Arab Emirates. Association for Computational Linguistics.
  24. Robust speech recognition via large-scale weak supervision.
  25. Language models are unsupervised multitask learners.
  26. Exploring the limits of transfer learning with a unified text-to-text transformer. Journal of Machine Learning Research, 21(140):1–67.
  27. Sara Rajaee and Mohammad Taher Pilehvar. 2021. A cluster-based approach for improving isotropy in contextual embedding space. In Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 2: Short Papers), pages 575–584, Online. Association for Computational Linguistics.
  28. Sara Rajaee and Mohammad Taher Pilehvar. 2022. An isotropy analysis in the multilingual BERT embedding space. In Findings of the Association for Computational Linguistics: ACL 2022, pages 1309–1316, Dublin, Ireland. Association for Computational Linguistics.
  29. William Rudman and Carsten Eickhoff. 2023. Stable anisotropic regularization.
  30. ImageNet Large Scale Visual Recognition Challenge. International Journal of Computer Vision (IJCV), 115(3):211–252.
  31. The multiberts: Bert reproductions for robustness analysis. arXiv preprint arXiv:2106.16163.
  32. Whitening sentence representations for better semantics and faster retrieval. arXiv preprint arXiv:2103.15316.
  33. Mingxing Tan and Quoc V. Le. 2019. Efficientnet: Rethinking model scaling for convolutional neural networks. ArXiv, abs/1905.11946.
  34. Training data-efficient image transformers & distillation through attention. In Proceedings of the 38th International Conference on Machine Learning, volume 139 of Proceedings of Machine Learning Research, pages 10347–10357. PMLR.
  35. Llama: Open and efficient foundation language models.
  36. Improving neural language generation with spectrum control. In International Conference on Learning Representations.
  37. Visual transformers: Token-based image representation and processing for computer vision. ArXiv, abs/2006.03677.
  38. Cvt: Introducing convolutions to vision transformers. arXiv preprint arXiv:2103.15808.
  39. Segformer: Simple and efficient design for semantic segmentation with transformers. CoRR, abs/2105.15203.
  40. ByT5: Towards a token-free future with pre-trained byte-to-byte models. Transactions of the Association for Computational Linguistics, 10:291–306.
  41. ConSERT: A contrastive framework for self-supervised sentence representation transfer. In Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers), pages 5065–5075, Online. Association for Computational Linguistics.
  42. Rare tokens degenerate all tokens: Improving neural text generation via adaptive gradient gating for rare token embeddings. In Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 29–45, Dublin, Ireland. Association for Computational Linguistics.
  43. Frequency-based distortions in contextualized word embeddings. CoRR, abs/2104.08465.
Citations (8)
View on Semantic Scholar

Summary

We haven't generated a summary for this paper yet.

Ai Sparkles Streamline Icon: https://streamlinehq.com Summarize Now

Whiteboard

Open Problems

We haven't generated a list of open problems mentioned in this paper yet.

Ai Sparkles Streamline Icon: https://streamlinehq.com Generate Now

Continue Learning

We haven't generated follow-up questions for this paper yet.

Ai Sparkles Streamline Icon: https://streamlinehq.com Generate Now

Collections

Sign up for free to add this paper to one or more collections.

User Circle Single Streamline Icon: https://streamlinehq.com Sign Up

Tweets

Sign up for free to view the 1 tweet with 3 likes about this paper.

User Circle Single Streamline Icon: https://streamlinehq.com Sign Up for Free