smartcatdog
smartcatdog
and how to predict when a text input by tensorflow-serving while get the bert pb model? thx
> 模型中使用wiki_100中提供的向量,对于英文如chanel会切分成c,h,a, n , e,l,有办法改进英文的输入吗? 想请教下这个vec是如何生成的呢
> open_data.zip (519.1 MB) > https://pan.baidu.com/s/1gtq3FEMx4_eAFmZqyn844Q > 密码:8gd2 thanks for your share, however, i open the url and input the passwords, the error named "the passwords is error" is happened....
> 请问如何将ckpt模型转换为pb,output_node_name是char_embedding/char_embedding?还是没有output_node_name呢? ckpt 转成 pb_model, tensorflow官网有个[freeze_graph.py](https://github.com/tensorflow/tensorflow/blob/master/tensorflow/python/tools/freeze_graph.py)代码可以参考。
> > > > > 给的标注数据只是这三种,如果有其他实体的标注数据也可以训练新的模型识别。我训练的识别菜名的模型,f1也可以到90%以上 > > > > > > > > > > > > 你的意思如果我们这里有新的标注数据,也可以直接用他的训练,不局限与以上的三种么?你训练的菜名识别直接换成了你的数据集,没做其他的? > > > > > > > > > 对的啊,模型是通用的,只要是“命名实体识别”的任务,然后有标注数据,都可以用这个模型跑的。 >...
> > > Hello, > > > you can look at the "create_model" function used in main.py line 171. > > > In this function, if you set config "pre_emb"...
> 可以试一下在main.py上面的train和clean,吧false修改成true 我这边试了下`python main.py --train=True --clean=True`,可是报错: ``` File "main.py", line 219, in main train() File "main.py", line 150, in train train_manager = BatchManager(train_data, FLAGS.batch_size) File "ChineseNER/data_utils.py", line 285, in __init__...
OK,thanks,it is ok. is word_format must be relatively with word_vec_path which is in config.json, for example, if "word_format" is "morph" and "word_vec_path" should be "vocab_morph.wordvec.st" while "word_format" is "lemma" and...