Longformer 获取 last_hidden_​​state

Question

我正在尝试在此处 https://huggingface.co/transformers/model_doc/longformer.html 的拥抱脸文档中遵循此示例：

import torch
from transformers import LongformerModel, LongformerTokenizer
model = LongformerModel.from_pretrained('allenai/longformer-base-4096')
tokenizer = LongformerTokenizer.from_pretrained('allenai/longformer-base-4096')
SAMPLE_TEXT = ' '.join(['Hello world! '] * 1000)  # long input document
input_ids = torch.tensor(tokenizer.encode(SAMPLE_TEXT)).unsqueeze(0)  # batch of size 1
# Attention mask values -- 0: no attention, 1: local attention, 2: global attention
attention_mask = torch.ones(input_ids.shape, dtype=torch.long, device=input_ids.device) # initialize to local attention
global_attention_mask = torch.zeros(input_ids.shape, dtype=torch.long, device=input_ids.device) # initialize to global attention to be deactivated for all tokens
global_attention_mask[:, [1, 4, 21,]] = 1  # Set global attention to random tokens for the sake of this example
                                    # Usually, set global attention based on the task. For example,
                                    # classification: the <s> token
                                    # QA: question tokens
                                    # LM: potentially on the beginning of sentences and paragraphs
outputs = model(input_ids, attention_mask=attention_mask, global_attention_mask=global_attention_mask, output_hidden_states= True)
sequence_output = outputs[0].last_hidden_state
pooled_output = outputs.pooler_output

我想这会返回一个嵌入示例文本的文档。 但是，我遇到了以下错误：

AttributeError: 'Tensor' object has no attribute 'last_hidden_state'

为什么不能调用last_hidden_​​state？

Answer 1

不要通过索引选择：

sequence_output = outputs.last_hidden_state

outputs 是具有以下属性的 LongformerBaseModelOutputWithPooling 对象：

print(outputs.keys())

输出：

odict_keys(['last_hidden_state', 'pooler_output', 'hidden_states'])

调用outputs[0] 或outputs.last_hidden_state 都会给你相同的张量，但是这个张量没有一个名为last_hidden_state 的属性。