pytorch bilstm-crf的crf score代码实现详解

king的江鸟

于 2020-03-04 16:49:54 发布

阅读量1.2k

点赞数

分类专栏： pytorch NLP 深度学习

本文链接：https://blog.youkuaiyun.com/weixin_40939578/article/details/104656142

版权

本文针对已了解CRF原理的读者，详细解释PyTorch中BiLSTM-CRF模型中_CRF_score的代码实现。通过数学推导证明了为何可以在每层独立计算得分，最后将这些得分累加得到总概率。文章以一个简单的A到B转移为例，逐步推广到序列标注问题，展示了这种高效计算路径总概率的方法。

摘要生成于 C知道，由 DeepSeek-R1 满血版支持，前往体验 >

本文适合已经基本了解crf原理的读者，深入探究代码层面的实现原理

1. _forward_alg代码

    def _forward_alg(self, feats):
        # Do the forward algorithm to compute the partition function
        init_alphas = torch.full((1, self.tagset_size), -10000.)
        # START_TAG has all of the score.
        init_alphas[0][self.tag_to_ix[START_TAG]] = 0.

        # Wrap in a variable so that we will get automatic backprop
        forward_var = init_alphas

        # Iterate through the sentence
        for feat in feats:
            alphas_t = []  # The forward tensors at this timestep
            for next_tag in range(self.tagset_size):
                # broadcast the emission score: it is the same regardless of
                # the previous tag
                emit_score = feat[next_tag].view(
                    1, -1).expand(1, self.tagset_size)
                # the ith entry of trans_score is the score of transitioning to
                # n

最低0.47元/天解锁文章