http://hlt.suda.edu.cn/index.php?action=history&feed=atom&title=Zhenghua-read-codes-hmm-em
Zhenghua-read-codes-hmm-em - 版本历史
2026-11-06T01:58:12Z
本wiki的该页面的版本历史
MediaWiki 1.35.2
http://hlt.suda.edu.cn/index.php?title=Zhenghua-read-codes-hmm-em&diff=2714&oldid=prev
Zhli：创建页面，内容为“train.sh -> run.py from tagger.cmds import Evaluate, Predict, Train tagger/cmds/train.py Train Train() def __call__ train = Corpus.load(config.ftrain) t...”
2020-10-22T07:04:17Z
<p>创建页面，内容为“train.sh -> run.py from tagger.cmds import Evaluate, Predict, Train tagger/cmds/train.py Train Train() def __call__ train = Corpus.load(config.ftrain) t...”</p>
<p><b>新页面</b></p><div>train.sh -> run.py <br />
<br />
from tagger.cmds import Evaluate, Predict, Train<br />
<br />
tagger/cmds/train.py Train<br />
<br />
Train()<br />
def __call__<br />
train = Corpus.load(config.ftrain)<br />
trainset = TextDataset(vocab.numericalize(train))<br />
train_loader = batchify(trainset, config.batch_size, False)<br />
print(f"{'train:':6} {len(trainset):5} sentences, {train.nwords} words in total, "<br />
f"{len(train_loader):3} batches provided")<br />
<br />
<br />
tagger = Tagger(config) # from tagger import Tagger, Model tagger/tagger.py tagger/model.py <br />
# Tagger去做前向后相、viterbi等<br />
# Model基于Tagger，做EM参数更新<br />
<br />
迭代若干次，看准确率变化，多少次不增长就停止 patience。感觉这个patience比较怪，loss也应该看一下</div>
Zhli