Generating Synthetic Datasets for Few-shot Prompt Tuning
On the map: Synthetic data and self-improvement for LLMs → LLM synthetic data generation
Abstract (arXiv)
A major limitation of prompt tuning is its dependence on large labeled training datasets. Under few-shot learning settings, prompt tuning lags far behind full-model fine-tuning, limiting its scope of application. In this paper, we leverage the powerful LLMs to synthesize task-specific labeled data for training the soft prompts. We first introduce a distribution-aligned weighted generator tuning (DawGen) method to encourage generating in-distribution data that aligns with the few-shot real data. Then, we train soft prompts on both synthetic and real datasets using a gradient surgery approach, which eliminates the conflicting gradients from different data sources. Experiments on seven sentence-pair classification datasets demonstrate the effectiveness of our proposed method for boosting prompt tuning in few-shot learning settings. Results on QQP, MRPC, and SICK datasets are even comparable to the performance of transfer learning from large real-world datasets, showing the promise of synthetic data as an alternative for enhancing soft prompt tuning.
Key facts. Cited by 1 paper in the map (data through 7 Oct 2026). Placed in LLM synthetic data generation, one of 13 directions in Synthetic data and self-improvement for LLMs. Builds on 42 of its 48 references that are in the map. First posted on arXiv 8 Oct 2024.
Builds on
42 of 48 references.
References · 42
- 2018BERT: Pre-training of Deep Bidirectional Transformers for Language Understandingcited by 40,204
- 2020Language Models are Few-Shot Learnerscited by 32,199
- 2022Training language models to follow instructions with human feedbackcited by 15,712
- 2019Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformercited by 14,491
- 2021LoRA: Low-Rank Adaptation of Large Language Modelscited by 14,084
- 2019Parameter-Efficient Transfer Learning for NLPcited by 4,195
- 2016SQuAD: 100,000+ Questions for Machine Comprehension of Textcited by 4,115
- 2021Prefix-Tuning: Optimizing Continuous Prompts for Generationcited by 3,878
- 2021The Power of Scale for Parameter-Efficient Prompt Tuningcited by 3,769
- 2022Scaling Instruction-Finetuned Language Modelscited by 2,997
- 2017A Broad-Coverage Challenge Corpus for Sentence Understanding through Inferencecited by 2,665
- 2015A large annotated corpus for learning natural language inferencecited by 1,837
30 more
- 2020Making Pre-trained Language Models Better Few-shot Learnerscited by 1,255
- 2020Gradient Surgery for Multi-Task Learningcited by 1,210
- 2016Temporal Ensembling for Semi-Supervised Learningcited by 1,054
- 2020Exploiting Cloze Questions for Few Shot Text Classification and Natural Language Inferencecited by 934
- 2019CTRL: A Conditional Transformer Language Model for Controllable Generationcited by 754
- 2023MemoryBank: Enhancing Large Language Models with Long-Term Memorycited by 539
- 2023Model Tells You What to Discard: Adaptive KV Cache Compression for LLMscited by 466
- 2017Robust Loss Functions under Label Noise for Deep Neural Networkscited by 441
- 2022Prompt-aligned Gradient for Prompt Tuningcited by 339
- 2021SPoT: Better Frozen Model Adaptation through Soft Prompt Transfercited by 235
- 2021PPT: Pre-trained Prompt Tuning for Few-shot Learningcited by 232
- 2023Large Language Model as Attributed Training Data Generator: A Tale of Diversity and Biascited by 230
- 2022ZeroGen: Efficient Zero-shot Learning via Dataset Generationcited by 209
- 2022Generating Training Data with Language Models: Towards Zero-Shot Language Understandingcited by 199
- 2021Generating Datasets with Pretrained Language Modelscited by 192
- 2021Reframing Instructional Prompts to GPTk's Languagecited by 176
- 2019Not Enough Data? Deep Learning to the Rescue!cited by 174
- 2020Data Augmentation using Pre-trained Transformer Modelscited by 168
- 2021Differentiable Prompt Makes Pre-trained Language Models Better Few-shot Learnerscited by 115
- 2020Training Question Answering Models From Synthetic Datacited by 95
- 2021Towards Interpreting and Mitigating Shortcut Learning Behavior of NLU Modelscited by 78
- 2023Large Language Models Can be Lazy Learners: Analyze Shortcuts in In-Context Learningcited by 69
- 2022ProGen: Progressive Zero-shot Dataset Generation via In-context Feedbackcited by 67
- 2022Self-Guided Noise-Free Data Generation for Efficient Zero-Shot Learningcited by 49
- 2022Tuning Language Models as Training Data Generators for Augmentation-Enhanced Few-Shot Learningcited by 36
- 2021Effectiveness of Pre-training for Few-shot Intent Classificationcited by 27
- 2022On the Domain Adaptation and Generalization of Pretrained Language Models: A Surveycited by 22
- 2022Instruction Tuning for Few-Shot Aspect-Based Sentiment Analysiscited by 16
- 2022Improving the Sample Efficiency of Prompt Tuning with Domain Adaptationcited by 15
- 2023Automated Few-shot Classification with Instruction-Finetuned Language Modelscited by 9
Built on by
Citing papers per year: 2025: 1.
1 paper.
Similar, not cited
Close in wording, no citation either way
- 2024SoftSRV: Learn to Generate Targeted Synthetic Data
- 2025BARE: Leveraging Base Language Models for Few-Shot Synthetic Data Generation
- 2025A Survey on Prompt Tuning
- 2023Gradient-Regulated Meta-Prompt Learning for Generalizable Vision-Language Models
- 2025Multi-Sample Prompting and Actor-Critic Prompt Optimization for Diverse Synthetic Data Generation
- 2023Diverse Data Augmentation with Diffusions for Effective Test-time Prompt Tuning
- 2023Align Your Prompts: Test-Time Prompting with Distribution Alignment for Zero-Shot Generalization
- 2022Test-Time Prompt Tuning for Zero-Shot Generalization in Vision-Language Models