测试时训练让大模型在少样本任务上表现大幅提升
The Surprising Effectiveness of Test-Time Training for Few-Shot Learning
- 用输入数据动态调整模型参数，实现推理时的快速适应
- 在ARC任务上准确率最高达61.9%，接近人类水平
- 适合需要快速应对新任务的场景，尤其对小样本学习有帮助
语言模型在训练分布内的任务表现优异，但在结构新颖的任务上即便给出少量上下文示例仍表现不佳。本文研究测试时训练（TTT）——在推理过程中利用输入数据导出的损失临时更新模型参数——以提升语言模型的推理与少样本学习能力。在抽象与推理语料库（ARC）上，使用上下文示例进行TTT可使准确率相比微调基线最高提升6倍，80亿参数模型在公开验证集上达到53.0%；通过与程序合成方法集成后可达61.9%，接近平均人类表现。在BIG-Bench Hard（BBH）任务中，10样本设置下TTT比标准少样本提示高出7.3个百分点（50.5%→57.8%）。结果揭示了上下文学习在新任务上的局限性，并展示了测试时训练增强语言模型适应性的潜力。
原文摘要 · Abstract (English)
Language models (LMs) have shown impressive performance on tasks within their training distribution, but often struggle with structurally novel tasks even when given a small number of in-context task examples. We investigate the effectiveness of test-time training (TTT) -- temporarily updating model parameters during inference using a loss derived from input data -- as a mechanism for improving LMs' reasoning and few-shot learning capabilities. On the Abstraction and Reasoning Corpus (ARC), performing TTT with in-context examples yields up to $6\times$ higher accuracy compared to fine-tuned baselines -- reaching $53.0\%$ on the public validation set with an 8B-parameter LM and $61.9\%$ when ensembled with program-synthesis methods, matching average human performance. On BIG-Bench Hard (BBH), TTT on in-context examples surpasses standard few-shot prompting in the $10$-shot setting by $7.3$ percentage points ($50.5\%$ to $57.8\%$). Our findings highlight the limitations of in-context learning for novel tasks and demonstrate the potential of test-time training to enhance language model adaptability.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。