Multistage Collaborative Knowledge Distillation from a Large Language Model for Semi-Supervised Sequence Generation

• URL: http://arxiv.org/abs/2311.08640v4
• Date: Sun, 4 Aug 2024 03:13:07 GMT
• Title: Multistage Collaborative Knowledge Distillation from a Large Language Model for Semi-Supervised Sequence Generation
• Authors: Jiachen Zhao, Wenlong Zhao, Andrew Drozdov, Benjamin Rozonoyer, Md Arafat Sultan, Jay-Yoon Lee, Mohit Iyyer, Andrew McCallum,
• Abstract summary: We study semi-supervised sequence generation tasks, where the few labeled examples are too scarce to finetune a model. We present the discovery that a student model distilled from a few-shot prompted large language models can commonly generalize better than its teacher.
• Score: 60.657065293413716
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: We study semi-supervised sequence generation tasks, where the few labeled examples are too scarce to finetune a model, and meanwhile, few-shot prompted large language models (LLMs) exhibit room for improvement. In this paper, we present the discovery that a student model distilled from a few-shot prompted LLM can commonly generalize better than its teacher to unseen examples on such tasks. We find that the student is able to learn a general pattern from the high-quality pseudolabels produced by the teacher during knowledge distillation (KD), and favorably not a general pattern from the low-quality pseudolables. Leveraging this discovery, we propose a new method, Multistage Collaborative Knowledge Distillation from an LLM (MCKD), for these tasks. MCKD first few-shot prompts an LLM to produce pseudolabels for unlabeled data. Then at each stage of an iterative KD process, a new pair of students is trained on disjoint partitions of the pseudolabeled data, and produces new and improved pseudolabels for their unseen partitions. We conduct extensive experiments on four syntactic and semantic parsing datasets and show the effectiveness of MCKD for low-resource semi-supervised sequence generation. On CRAFT biomedical parsing, for example, 3-stage MCKD with 50 labeled examples outperforms an LLM teacher and vanilla KD by 7.5% and 3.7% parsing F1, respectively, and matches the performance of supervised finetuning with 500 labeled examples.

Related papers

• Mentor-KD: Making Small Language Models Better Multi-step Reasoners [15.159415340059388]
  We propose Mentor-KD, which effectively distills the multi-step reasoning capability of LLMs to smaller LMs. We exploit a mentor, intermediate-sized task-specific fine-tuned model, to augment additional CoT annotations. We conduct extensive experiments and confirm Mentor-KD's effectiveness across various models and complex reasoning tasks.
  arXiv  Detail & Related papers  (2024-10-11T17:53:27Z)
• Limits of Transformer Language Models on Learning to Compose Algorithms [77.2443883991608]
  We evaluate training LLaMA models and prompting GPT-4 and Gemini on four tasks demanding to learn a composition of several discrete sub-tasks. Our results indicate that compositional learning in state-of-the-art Transformer language models is highly sample inefficient.
  arXiv  Detail & Related papers  (2024-02-08T16:23:29Z)
• LLMaAA: Making Large Language Models as Active Annotators [32.57011151031332]
  We propose LLMaAA, which takes large language models as annotators and puts them into an active learning loop to determine what to annotate efficiently. We conduct experiments and analysis on two classic NLP tasks, named entity recognition and relation extraction. With LLMaAA, task-specific models trained from LLM-generated labels can outperform the teacher within only hundreds of annotated examples.
  arXiv  Detail & Related papers  (2023-10-30T14:54:15Z)
• DisCo: Distilled Student Models Co-training for Semi-supervised Text Mining [23.418419374791107]
  DisCo is a semi-supervised learning framework for fine-tuning a cohort of small student models generated from a large PLM. We show that DisCo can produce student models that are 7.6 times smaller and 4.8 times faster in inference than the baseline PLMs.
  arXiv  Detail & Related papers  (2023-05-20T03:23:16Z)
• Distilling Step-by-Step! Outperforming Larger Language Models with Less Training Data and Smaller Model Sizes [91.58845026796149]
  We introduce Distilling step-by-step, a new mechanism that trains small models that outperform large language models. We present three findings across 4 NLP benchmarks.
  arXiv  Detail & Related papers  (2023-05-03T17:50:56Z)
• SLaM: Student-Label Mixing for Distillation with Unlabeled Examples [15.825078347452024]
  We present a principled method for knowledge distillation with unlabeled examples that we call Student-Label Mixing (SLaM) SLaM consistently improves over prior approaches by evaluating it on several standard benchmarks. We give an algorithm improving the best-known sample complexity for learning halfspaces with margin under random classification noise.
  arXiv  Detail & Related papers  (2023-02-08T00:14:44Z)
• Tuning Language Models as Training Data Generators for Augmentation-Enhanced Few-Shot Learning [30.65315081964461]
  We study few-shot learning with pretrained language models (PLMs) from a different perspective. We first tune an autoregressive PLM on the few-shot samples and then use it as a generator to synthesize a large amount of novel training samples. Our approach FewGen achieves an overall better result across seven classification tasks of the GLUE benchmark than existing few-shot learning methods.
  arXiv  Detail & Related papers  (2022-11-06T06:46:47Z)
• Neighborhood Contrastive Learning for Novel Class Discovery [79.14767688903028]
  We build a new framework, named Neighborhood Contrastive Learning, to learn discriminative representations that are important to clustering performance. We experimentally demonstrate that these two ingredients significantly contribute to clustering performance and lead our model to outperform state-of-the-art methods by a large margin.
  arXiv  Detail & Related papers  (2021-06-20T17:34:55Z)
• Generate, Annotate, and Learn: Generative Models Advance Self-Training and Knowledge Distillation [58.64720318755764]
  Semi-Supervised Learning (SSL) has seen success in many application domains, but this success often hinges on the availability of task-specific unlabeled data. Knowledge distillation (KD) has enabled compressing deep networks and ensembles, achieving the best results when distilling knowledge on fresh task-specific unlabeled examples. We present a general framework called "generate, annotate, and learn (GAL)" that uses unconditional generative models to synthesize in-domain unlabeled data.
  arXiv  Detail & Related papers  (2021-06-11T05:01:24Z)
• Deep Semi-supervised Knowledge Distillation for Overlapping Cervical Cell Instance Segmentation [54.49894381464853]
  We propose to leverage both labeled and unlabeled data for instance segmentation with improved accuracy by knowledge distillation. We propose a novel Mask-guided Mean Teacher framework with Perturbation-sensitive Sample Mining. Experiments show that the proposed method improves the performance significantly compared with the supervised method learned from labeled data only.
  arXiv  Detail & Related papers  (2020-07-21T13:27:09Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.