Heming Xia comments

Results 16 comments of


                                            Heming Xia

尝试用模型翻译一个句子的时候有的自己测试的句子翻译结果和理想结果相差很远

这个模型训的非常的简陋Hhh，只是个新手入门参考，machine translation想得到比较好的效果可以试试[fairseq](https://github.com/facebookresearch/fairseq)： - [examples/translation](https://github.com/facebookresearch/fairseq/tree/main/examples/translation)

IndexError: index out of range in self

对的可能是 @zjmwqx 提到的问题，这个repository只是一个简单实现的参考，可以根据自己需要修改代码

Didn't find file /pretrained_bert_models/bert-base-chinese/added_tokens.json. We won't load it. Didn't find file /pretrained_bert_models/bert-base-chinese/special_tokens_map.json. We won't load it. Didn't find file /pretrained_bert_models/bert-base-chinese/tokenizer_config.json. We won't load it.

没有影响哈只要能跑出结果就行了

非fine tune的时候lstm和crf层的参数为啥不学习？

非finetune的应该没用过这里是有问题的可以自己改一下不难的~

ValueError: cannot copy sequence with size 37 to array axis with dimension 36

不好意思，这两天比较忙，下周我看一下这个问题～

awesome work请问作者有没有inference脚本呢

评测脚本在[这里](https://github.com/hemingkx/WordSeg#dataset)有说明~ ``` 如有需要，这里是PKU数据集及评测脚本的百度网盘地址👇：链接: https://pan.baidu.com/s/1yZrDSrvAXi-0mCxRp3L-sw 密码: 31qd 将数据集及评测脚本解压后放置在所用模型的./data/路径下即可。 ``` inference都包含在run.py中了，可以根据需要自行修改~

awesome work!!!!!!

因为已经过了一段时间，训练好的模型已经被清除了，过几天会重新跑一遍更新一个网盘链接～

[GAD] Error when trying to use Generalized Aggressive Decoding for inference

Hi @ndkmath1, I noticed that you put the NAT drafter and the AT verifier in opposite positions. They should be placed as follows. Can you try again the correct way?...

[GAD] Error when trying to use Generalized Aggressive Decoding for inference

You can pre-process the official WMT14 EN-DE `test.en` file with our bpemodel [here](https://github.com/microsoft/unilm/blob/master/decoding/GAD/data/wmt14.en-de/bpe.32000). Here are the first three lines of our test file: ``` Gut@@ ach : Incre@@ ased safety...

[GAD] Error when trying to use Generalized Aggressive Decoding for inference

The bpe codes we used are from the released data by Google. However, you can find a pre-process script [here](https://github.com/facebookresearch/fairseq/blob/main/examples/translation/prepare-wmt14en2de.sh), which works fine with wmt14.en-de raw data.