tpu 名称

[chinese-wikisource]

--tpu_name=后面如果加ip无法使用，应该是自己创建的tpu名称。

将此问题解决后照样无法使用，出现以下错误：

ERROR:tensorflow:Error recorded from infeed: From /job:worker/replica:0/task:0: {{function_node tf_data_experimental_map_andbatch_55}} Key: masked_lm_weights. Can't parse serialized Example. [[{{node ParseSingleExample/ParseSingleExample}}]] [[input_pipeline_task0/while/IteratorGetNext]] E0812 13:40:04.976598 140565570795264 error_handling.py:75] Error recorded from infeed: From /job:worker/replica:0/task:0: {{function_node tf_data_experimental_map_andbatch_55}} Key: masked_lm_weights. Can't parse serialized Example. [[{{node ParseSingleExample/ParseSingleExample}}]] [[input_pipeline_task0/while/IteratorGetNext]] ERROR:tensorflow:Closing session due to error From /job:worker/replica:0/task:0: {{function_node tf_data_experimental_map_andbatch_55}} Key: masked_lm_weights. Can't parse serialized Example. [[{{node ParseSingleExample/ParseSingleExample}}]] [[input_pipeline_task0/while/IteratorGetNext]] E0812 13:40:09.982232 140565203113728 error_handling.py:81] Closing session due to error From /job:worker/replica:0/task:0: {{function_node tf_data_experimental_map_andbatch_55}} Key: masked_lm_weights. Can't parse serialized Example. [[{{node ParseSingleExample/ParseSingleExample}}]] [[input_pipeline_task0/while/IteratorGetNext]] ERROR:tensorflow:Error recorded from outfeed: Step was cancelled by an explicit call to Session::Close(). E0812 13:40:14.003780 140565219899136 error_handling.py:75] Error recorded from outfeed: Step was cancelled by an explicit call to `Session