为什么模型在预测情感极性的时候，反而传了polarities进去

[WhatTong]

如下代码：
polarity_labels = self.get_batch_polarities(polarities)
cdm_vec = self.feature_dynamic_mask(local_context_ids, polarities)
loss_apc = loss_sen(apc_logits, polarity_labels)

[yangheng95]

polarities 用来辅助定位aspect的位置（positions），不涉及极性的推理

[WhatTong]

不好意思还想问下，lcf-atepc.py里这段代码的作用是什么？
batch_size, max_len, feat_dim = global_context_out.shape
global_valid_output = torch.zeros(batch_size, max_len, feat_dim, dtype=torch.float32).to(self.args.device)
for i in range(batch_size):
jj = -1
for j in range(max_len):
if valid_ids[i][j].item() == 1:
jj += 1
global_valid_output[i][jj] = global_context_out[i][j]

[yangheng95]

#7 (comment)

对于被切分成多个部分的apscet，只选用第一个位置的特征分类预测IOB