2000 character limit reached
Teacher-Student Training for Debiasing: General Permutation Debiasing for Large Language Models
Published 20 Mar 2024 in cs.CL | (2403.13590v1)
Abstract: LLMs have demonstrated impressive zero-shot capabilities and versatility in NLP tasks, however they sometimes fail to maintain crucial invariances for specific tasks. One example is permutation sensitivity, where LLMs' outputs may significantly vary depending on the order of the input options. While debiasing techniques can mitigate these issues, and yield better performance and reliability, they often come with a high computational cost at inference. This paper addresses this inefficiency at inference time. The aim is to distill the capabilities of a computationally intensive, debiased, teacher model into a more compact student model. We explore two variants of student models: one based on pure distillation, and the other on an error-correction approach for more complex tasks, where the student corrects a single biased decision from the teacher to achieve a debiased output. Our approach is general and can be applied to both black-box and white-box LLMs. Furthermore, we demonstrate that our compact, encoder-only student models can outperform their larger, biased teacher counterparts, achieving better results with significantly fewer parameters.
- Gpt-4 technical report. arXiv preprint arXiv:2303.08774.
- Gemini: a family of highly capable multimodal models. arXiv preprint arXiv:2312.11805.
- Language models are few-shot learners. Advances in neural information processing systems, 33:1877–1901.
- Youngjin Chae and Thomas Davidson. 2023. Large language models for text classification: From zero-shot learning to fine-tuning. Open Science Foundation.
- Scaling instruction-finetuned language models. arXiv preprint arXiv:2210.11416.
- Think you have solved question answering? try arc, the ai2 reasoning challenge. arXiv preprint arXiv:1803.05457.
- 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.
- Summeval: Re-evaluating summarization evaluation. arXiv preprint arXiv:2007.12626.
- On calibration of modern neural networks. In International conference on machine learning, pages 1321–1330. PMLR.
- Deberta: Decoding-enhanced bert with disentangled attention. In International Conference on Learning Representations.
- Distilling the knowledge in a neural network. In NIPS Deep Learning and Representation Learning Workshop.
- A survey on hallucination in large language models: Principles, taxonomy, challenges, and open questions. arXiv preprint arXiv:2311.05232.
- Cosmos QA: Machine reading comprehension with contextual commonsense reasoning. 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 2391–2401, Hong Kong, China. Association for Computational Linguistics.
- Tinybert: Distilling bert for natural language understanding. In Findings of the Association for Computational Linguistics: EMNLP 2020, pages 4163–4174.
- Gender bias and stereotypes in large language models. In Proceedings of The ACM Collective Intelligence Conference, pages 12–24.
- RACE: Large-scale ReAding comprehension dataset from examinations. In Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing, pages 785–794, Copenhagen, Denmark. Association for Computational Linguistics.
- A new multi-choice reading comprehension dataset for curriculum learning. In Proceedings of The Eleventh Asian Conference on Machine Learning, pages 742–757.
- Jurassic-1: Technical details and evaluation. White Paper. AI21 Labs, 1:9.
- Roberta: A robustly optimized bert pretraining approach. arXiv preprint arXiv:1907.11692.
- Mitigating word bias in zero-shot prompt-based classifiers. In Findings of the Association for Computational Linguistics: IJCNLP-AACL 2023 (Findings), pages 327–335.
- Zero-shot nlg evaluation through pairware comparisons with llms. arXiv preprint arXiv:2307.07889.
- Selfcheckgpt: Zero-resource black-box hallucination detection for generative large language models. arXiv preprint arXiv:2303.08896.
- The larger they are, the harder they fail: Language models do not recognize identifier swaps in python. In Findings of the Association for Computational Linguistics: ACL 2023, pages 272–292, Toronto, Canada. Association for Computational Linguistics.
- Training language models to follow instructions with human feedback. Advances in Neural Information Processing Systems, 35:27730–27744.
- Daniel Park. 2023. Open-llm-leaderboard-report.
- Pouya Pezeshkpour and Estevam Hruschka. 2023. Large language models sensitivity to the order of options in multiple-choice questions. arXiv preprint arXiv:2308.11483.
- Exploring the limits of transfer learning with a unified text-to-text transformer. The Journal of Machine Learning Research, 21(1):5485–5551.
- Laria Reynolds and Kyle McDonell. 2021. Prompt programming for large language models: Beyond the few-shot paradigm. In Extended Abstracts of the 2021 CHI Conference on Human Factors in Computing Systems, pages 1–7.
- Leveraging large language models for multiple choice question answering. ICLR.
- Quantifying language models’ sensitivity to spurious features in prompt design or: How i learned to start worrying about prompt formatting. arXiv preprint arXiv:2310.11324.
- Llama 2: Open foundation and fine-tuned chat models. arXiv preprint arXiv:2307.09288.
- Large language models are not fair evaluators.
- Super-NaturalInstructions: Generalization via declarative instructions on 1600+ NLP tasks. In Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing, pages 5085–5109, Abu Dhabi, United Arab Emirates. Association for Computational Linguistics.
- Finetuned language models are zero-shot learners. In International Conference on Learning Representations.
- Calibrate before use: Improving few-shot performance of language models. In International Conference on Machine Learning, pages 12697–12706. PMLR.
- Large language models are not robust multiple choice selectors. arXiv e-prints, pages arXiv–2309.
- Judging llm-as-a-judge with mt-bench and chatbot arena. arXiv preprint arXiv:2306.05685.
- Large language models are human-level prompt engineers. In The Eleventh International Conference on Learning Representations.
Sponsor
Paper Prompts
Sign up for free to create and run prompts on this paper using GPT-5.
Top Community Prompts
Collections
Sign up for free to add this paper to one or more collections.