Dynamic Self-Distillation via Previous Mini-batches for Fine-tuning Small Language Models

• URL: http://arxiv.org/abs/2411.16991v1
• Date: Mon, 25 Nov 2024 23:37:48 GMT
• Title: Dynamic Self-Distillation via Previous Mini-batches for Fine-tuning Small Language Models
• Authors: Yao Fu, Yin Yu, Xiaotian Han, Runchao Li, Xianxuan Long, Haotian Yu, Pan Li,
• Abstract summary: Self-distillation (SelfD) emerges as an advisable alternative, enabling student models to learn without teachers' guidance. We introduce a model-agnostic and task-agnostic method named dynamic SelfD from the previous minibatch (DynSDPB) We demonstrate the superior performance of DynSDPB on both encoder-only LMs and decoder-only LMs.
• Score: 26.52225051132963
• License:
• Abstract: Knowledge distillation (KD) has become a widely adopted approach for compressing large language models (LLMs) to reduce computational costs and memory footprints. However, the availability of complex teacher models is a prerequisite for running most KD pipelines. Thus, the traditional KD procedure can be unachievable or budget-unfriendly, particularly when relying on commercial LLMs like GPT4. In this regard, Self-distillation (SelfD) emerges as an advisable alternative, enabling student models to learn without teachers' guidance. Nonetheless, existing SelfD approaches for LMs often involve architectural modifications, assuming the models are open-source, which may not always be practical. In this work, we introduce a model-agnostic and task-agnostic method named dynamic SelfD from the previous minibatch (DynSDPB), which realizes current iterations' distillation from the last ones' generated logits. Additionally, to address prediction inaccuracies during the early iterations, we dynamically adjust the distillation influence and temperature values to enhance the adaptability of fine-tuning. Furthermore, DynSDPB is a novel fine-tuning policy that facilitates the seamless integration of existing self-correction and self-training techniques for small language models (SLMs) because they all require updating SLMs' parameters. We demonstrate the superior performance of DynSDPB on both encoder-only LMs (e.g., BERT model families) and decoder-only LMs (e.g., LLaMA model families), validating its effectiveness across natural language understanding (NLU) and natural language generation (NLG) benchmarks.

Related papers

• Reference Trustable Decoding: A Training-Free Augmentation Paradigm for Large Language Models [79.41139393080736]
  Large language models (LLMs) have rapidly advanced and demonstrated impressive capabilities. In-Context Learning (ICL) and. Efficient Fine-Tuning (PEFT) are currently two mainstream methods for augmenting. LLMs to downstream tasks. We propose Reference Trustable Decoding (RTD), a paradigm that allows models to quickly adapt to new tasks without fine-tuning.
  arXiv  Detail & Related papers  (2024-09-30T10:48:20Z)
• Exploring and Enhancing the Transfer of Distribution in Knowledge Distillation for Autoregressive Language Models [62.5501109475725]
  Knowledge distillation (KD) is a technique that compresses large teacher models by training smaller student models to mimic them. This paper introduces Online Knowledge Distillation (OKD), where the teacher network integrates small online modules to concurrently train with the student model. OKD achieves or exceeds the performance of leading methods in various model architectures and sizes, reducing training time by up to fourfold.
  arXiv  Detail & Related papers  (2024-09-19T07:05:26Z)
• Boosting Continual Learning of Vision-Language Models via Mixture-of-Experts Adapters [65.15700861265432]
  We present a parameter-efficient continual learning framework to alleviate long-term forgetting in incremental learning with vision-language models. Our approach involves the dynamic expansion of a pre-trained CLIP model, through the integration of Mixture-of-Experts (MoE) adapters. To preserve the zero-shot recognition capability of vision-language models, we introduce a Distribution Discriminative Auto-Selector.
  arXiv  Detail & Related papers  (2024-03-18T08:00:23Z)
• AXOLOTL: Fairness through Assisted Self-Debiasing of Large Language Model Outputs [20.772266479533776]
  AXOLOTL is a novel post-processing framework that operates agnostically across tasks and models. It identifies biases, proposes resolutions, and guides the model to self-debias its outputs. This approach minimizes computational costs and preserves model performance.
  arXiv  Detail & Related papers  (2024-03-01T00:02:37Z)
• DistiLLM: Towards Streamlined Distillation for Large Language Models [53.46759297929675]
  DistiLLM is a more effective and efficient KD framework for auto-regressive language models. DisiLLM comprises two components: (1) a novel skew Kullback-Leibler divergence loss, where we unveil and leverage its theoretical properties, and (2) an adaptive off-policy approach designed to enhance the efficiency in utilizing student-generated outputs.
  arXiv  Detail & Related papers  (2024-02-06T11:10:35Z)
• MiniLLM: Knowledge Distillation of Large Language Models [112.93051247165089]
  Knowledge Distillation (KD) is a promising technique for reducing the high computational demand of large language models (LLMs) We propose a KD approach that distills LLMs into smaller language models. Our method is scalable for different model families with 120M to 13B parameters.
  arXiv  Detail & Related papers  (2023-06-14T14:44:03Z)
• Self-Feature Regularization: Self-Feature Distillation Without Teacher Models [0.0]
  Self-Feature Regularization(SFR) is proposed, which uses features in the deep layers to supervise feature learning in the shallow layers. We firstly use generalization-l2 loss to match local features and a many-to-one approach to distill more intensively in the channel dimension.
  arXiv  Detail & Related papers  (2021-03-12T15:29:00Z)
• MiniLM: Deep Self-Attention Distillation for Task-Agnostic Compression of Pre-Trained Transformers [117.67424061746247]
  We present a simple and effective approach to compress large Transformer based pre-trained models. We propose distilling the self-attention module of the last Transformer layer of the teacher, which is effective and flexible for the student. Experimental results demonstrate that our monolingual model outperforms state-of-the-art baselines in different parameter size of student models.
  arXiv  Detail & Related papers  (2020-02-25T15:21:10Z)

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.