search

macanv / BERT-BiLSTM-CRF-NER

Tensorflow solution of NER task Using BiLSTM-CRF model with Google BERT Fine-tuning And private Server services
https://github.com/macanv/BERT-BiLSMT-CRF-NER
4.69k stars 1.25k forks source link

Can't load save_path when it is None #137

Open chenbaicheng opened 5 years ago

chenbaicheng commented 5 years ago

I:VENTILATOR:[i:i: 91]:lodding ner model, could take a while... 2019-05-26 21:44:51.296903: I tensorflow/core/platform/cpu_feature_guard.cc:141] Your CPU supports instructions that this TensorFlow binary was not compiled to use: SSE4.1 SSE4.2 AVX AVX2 FMA 2019-05-26 21:44:51.298731: I tensorflow/core/common_runtime/process_util.cc:69] Creating new thread pool with default inter op setting: 2. Tune using inter_op_parallelism_threads for best performance. E:NER_MODEL, Lodding...:[gra:opt:306]:fail to optimize the graph! Can't load save_path when it is None. Traceback (most recent call last): File "/home/chen/miniconda3/envs/bert_ner/lib/python3.6/site-packages/bert_base-0.0.9-py3.6.egg/bert_base/server/graph.py", line 295, in optimize_ner_model saver.restore(sess, tf.train.latest_checkpoint(args.model_dir)) File "/home/chen/miniconda3/envs/bert_ner/lib/python3.6/site-packages/tensorflow/python/training/saver.py", line 1713, in restore raise ValueError("Can't load save_path when it is None.") ValueError: Can't load save_path when it is None. Traceback (most recent call last): File "/home/chen/miniconda3/envs/bert_ner/bin/bert-base-serving-start", line 11, in load_entry_point('bert-base==0.0.9', 'console_scripts', 'bert-base-serving-start')() File "/home/chen/miniconda3/envs/bert_ner/lib/python3.6/site-packages/bert_base-0.0.9-py3.6.egg/bert_base/runs/init.py", line 17, in start_server server = BertServer(args) File "/home/chen/miniconda3/envs/bert_ner/lib/python3.6/site-packages/bert_base-0.0.9-py3.6.egg/bert_base/server/init.py", line 102, in init raise FileNotFoundError('graph optimization fails and returns empty result') FileNotFoundError: graph optimization fails and returns empty result

chenbaicheng commented 5 years ago

用的是作者的model文件

jawhiow commented 5 years ago

我也遇到了这个问题，请问你解决了嘛？

chenbaicheng commented 5 years ago

@jawhiow 不知道是不是 pip install 那个包的问题 你是不是也是用pip 安装的 删掉 然后 用 git的源码安装试试 我找了一下其他人的 作者说 pip的包有问题的 ，现在还没有解决 我等新显卡来 ，再试试

macanv commented 5 years ago

你们指定了model地址么，把你们的cmd贴出来啊。

chenbaicheng commented 5 years ago

@macanv 作者君 O(∩_∩)O谢谢 那个pd 那个参数不能指定 他会提示 argument无效 所以我删掉了
bert-base-serving-start -bert_model_dir /home/chen/Desktop/chinese_ner/360Downloads/chinese_L-12_H-768_A-12/ -model_dir /home/chen/Desktop/chinese_ner/360Downloads/BERT_NER/ -num_worker 3 -mode NER

macanv commented 5 years ago

@chenbaicheng 这个可以正常运行么？

chenbaicheng commented 5 years ago

@macanv 不行呢 它提示 fail to optimize the graph! Can't load save_path when it is None. 发生报错是这句 File "/home/chen/miniconda3/envs/bert_ner/lib/python3.6/site-packages/bert_base-0.0.9-py3.6.egg/bert_base/server/graph.py", line 295, in optimize_ner_model saver.restore(sess, tf.train.latest_checkpoint(args.model_dir)) 不能恢复模型吧 不过那个model_dir 是指哪一个呢 全部参数都给了咯谢谢

chenbaicheng commented 5 years ago

@macanv 难道参数哪里还看相对路径的吗 我是用conda安装的 然后模型放在其他地方 然后terminal的目录是根目录 ～/

chenbaicheng commented 5 years ago

@macanv 可能是tensorflow的版本问题吗 去看了一下肖大神的 bert as service 提问里面也有关于 graphic 建立失败的 作者现在是用那个版本的tensorflow

macanv commented 5 years ago

1.10和1.12我都试过，其他的没试过，我这边测试没问题，Ubuntu16.04和windows10

chenbaicheng commented 5 years ago

恩 O(∩_∩)O谢谢 我再试试 有可能我的显卡有点渣

chenbaicheng commented 5 years ago

@macanv 换了一张显卡了 name: Tesla K40c major: 3 minor: 5 memoryClockRate(GHz): 0.745 pciBusID: 0000:03:00.0 totalMemory: 11.17GiB freeMemory: 10.92GiB 不过还是不行 主要是这个 File "/home/lin/miniconda3/envs/chinese_ner/lib/python3.6/site-packages/tensorflow/python/training/saver.py", line 1713, in restore raise ValueError("Can't load save_path when it is None.") ValueError: Can't load save_path when it is None. 这里的save_path是哪里指定的 作者君有试过ubantu环境吗 我看到都是window的目录 能不能贴一个例子出来 谢谢哦

chenbaicheng commented 5 years ago

model_dir: 训练好的NER模型或者文本分类模型的路径，对于上面的output_dir 我没有训练就不能用了吗

chianwei commented 5 years ago

I ran this command and it works: bert-base-serving-start -model_dir pretrain-model/BERT_NER/ -bert_model_dir ../bert-model/chinese_L-12_H-768_A-12/ -model_pb_dir pretrain-model/BERT_NER/ -mode NER" I install the package through source code rather than using pip, and the model (contents under BERT_NER) was downloaded from authors Baidu drive. I am using custom compiled TensorFlow 1.12.3 and Ubuntu 16.04.

chenbaicheng commented 5 years ago

@chianwei thanks you ，where baidu driver address can you write in here

chianwei commented 5 years ago

@chianwei thanks you ，where baidu driver address can you write in here

The URL of the author's pretrain model is shown in the Readme page, under the "As Service" section. image

CACppuccino commented 4 years ago

I ran this command and it works: bert-base-serving-start -model_dir pretrain-model/BERT_NER/ -bert_model_dir ../bert-model/chinese_L-12_H-768_A-12/ -model_pb_dir pretrain-model/BERT_NER/ -mode NER" I install the package through source code rather than using pip, and the model (contents under BERT_NER) was downloaded from authors Baidu drive. I am using custom compiled TensorFlow 1.12.3 and Ubuntu 16.04.

总结一下，就是pb文件可以指向从百度盘下下来的文件夹，ner模型也指向这个