概述
In-context prompting and metric-based learning achieve this through different procedures, and neither guarantees successful adaptation from any arbitrary examples.
深入探讨
Few-shot is used across methods that supply only a small number of task examples, often after substantial prior learning. In an N-way K-shot classification episode, N counts classes and K counts labeled support examples per class. That notation describes the task’s support data rather than the model’s total pretraining corpus. In-context few-shot prompting supplies demonstrations in the input of an already trained model. The cited language-model study evaluates this mode without task-specific gradient updates during the prompt; examples condition the prediction rather than rewrite weights. Fine-tuning on a small dataset is a different procedure even when both are called few-shot adaptation. Prior training can make representations or patterns useful for a new task, but a few examples do not guarantee that the model infers the intended labels or rules. Prototypical Networks studies a metric-based alternative. A learned embedding maps labeled support items into a space, and each class prototype is the mean of its support embeddings. A query is classified using distances to those prototypes. If one class has support vectors (0,0) and (2,0), its arithmetic-mean prototype is (1,0). The support items used to build the prototype and the query items used to assess classification have different roles. Training episodes in the paper keep those sampled sets separate. For in-context models, demonstration order can also affect results: Fantastically Ordered Prompts studies large variation among permutations in particular text-classification setups. Selecting a favorable order should not be confused with proof of generalization on an untouched evaluation set.
现实世界的实施
Providing labeled ticket examples in a prompt before a new ticket query.
Demonstrating an output format with a few input-output pairs.
Computing a class prototype from labeled support embeddings.
Comparing prompt orders on a separate evaluation set without changing the examples.
少样本学习的未来
Few-shot evaluations can compare representative support examples, ordering, retrieval of demonstrations, and learned distance functions under clearly separated data roles. More demonstrations may supply useful information but can also change prompt length or introduce misleading patterns. Experiments should record whether weights were updated and whether labels from the final query set influenced selection. Different domains can require different prior representations or examples. Testing the intended task and untouched queries gives a more meaningful result than assuming a small labeled set automatically enables instant adaptation or reproduces human learning.
付诸实践
为什么这很重要
- 更清晰的判决. 它可以帮助您将清晰的技术声明与营销语言分开。
- 成本与预算. 在花费金钱或时间之前，您可以提出更好的实施问题。
- 团队与工作流程. 具有共同理解的团队可以做出更好的产品、政策和学习决策。
留意
- 不同的团队可能会以不同的方式使用同一术语，因此请尽早定义范围。
- 基准测试可能看起来很强大，但实际性能却参差不齐。
- 忽视数据质量和评估计划通常会产生脆弱的结果。
后续步骤
- 从您需要的结果的简单语言定义开始。
- 在测试之前选择一种成功指标和一种失败条件。
- 使用代表性数据运行小型试点，而不是完善的演示集。
- 记录少样本学习在哪些方面有帮助以及在哪些方面更简单的方法更好。
不断探索
常见问题
什么是小样本学习？
Few-shot methods aim to use a small labeled support set or a few demonstrations for a task. In-context prompting and metric-based learning achieve this through different procedures, and neither guarantees successful adaptation from any arbitrary examples.
In an N-way K-shot classification episode, what does K count?
N counts the classes; K counts examples for each class in the specified support set.
当大型语言模型根据提示进行上下文中的几次学习时，其权重会发生什么？
上下文学习不执行梯度更新；提示中的示例只是指导模型的下一个标记预测。
零次提示和少次提示之间的主要区别是什么？
零样本仅提供说明，没有示例，而少样本则包含少量演示。
How can prior learning contribute to a few-shot task?
The benefit depends on the prior model, task, and examples; success must be measured.
其中描述了像原型网络这样基于度量的小样本方法？
原型网络为每个类形成一个原型（平均嵌入），并通过学习空间中最近的原型对新点进行分类。