Natural Language Inference(Recognizing Textual Entailment)

definition

Natural language inference (NLI) is the task of determining whether a “hypothesis” is true (entailment), false (contradiction), or undetermined (neutral) given a “premise”.

benchmarks

Benchmark datasets used for NLI include SNLI, MultiNLI, SciTail, SuperGLUE, RTE, WNLI.

Problems record of using OpenAI's API

GPT

gpt-3.5-turbo-instruct generates empty text after calling several times.

Tried adding space or adding newline, but didn’t work.

gpt-3.5-turbo-1106 generates different results from same prompt even though T is set as 0.

Tried setting seed but did’t work. Switched to another version mitigated this problem.

机器学习(5)--不平衡的分类问题

什么是不平衡的分类问题

在机器学习中,不平衡的分类问题指的是类别之间的样本分布不均匀,其中某一类的样本数量远远超过另一类。这种不平衡可能会对模型训练和性能评估产生影响,因为模型可能更倾向于预测样本数更多的类别,而对样本数较少的类别进行较差的预测。

如何解决

为了解决不平衡分类问题,可以考虑以下方法:增加少数类别的样本数或减少多数类别的样本数,以平衡类别分布。这包括上采样(增加少数类别样本)和下采样(减少多数类别样本)。调整分类阈值,使模型更倾向于识别少数类别。这可以通过调整模型输出的概率阈值来实现。