PaddleNLP icon indicating copy to clipboard operation
PaddleNLP copied to clipboard

[Question]: 在情感分析任务中,使用predict_sentence.py进行预测时,如何获取模型输出的每个类别的置信度

Open tigerbaby1515 opened this issue 2 years ago • 1 comments

请提出你的问题

在情感分析任务中,使用predict_sentence.py进行预测时,如何获取每个类别的置信度 results = predict(model, data, tokenizer, label_map, batch_size=args.batch_size)

for idx, text in enumerate(data):
    print('Data: {} \t Label: {}'.format(text, results[idx]))

在这results里面存储的是每个句子的标签,现在我想知道每个句子的置信度或者模型输出的概率值,怎么弄呢

tigerbaby1515 avatar Nov 09 '22 10:11 tigerbaby1515

你好,可以从代码中的probs获取概率值,probs是经过softmax后的数值,表达了样本归属不同情感的概率:

https://github.com/PaddlePaddle/PaddleNLP/blob/7c06b02e3161bd398e0decd439d47efd15f3feee/examples/sentiment_analysis/skep/predict_sentence.py#L124

1649759610 avatar Nov 10 '22 02:11 1649759610

This issue is stale because it has been open for 60 days with no activity. 当前issue 60天内无活动,被标记为stale。

github-actions[bot] avatar Jan 10 '23 01:01 github-actions[bot]

This issue was closed because it has been inactive for 14 days since being marked as stale. 当前issue 被标记为stale已有14天,即将关闭。

github-actions[bot] avatar Jan 25 '23 00:01 github-actions[bot]