zuoxin.xiahou

Still cannot find freezing target network in current version's code. It's really no effect?

变成了打工的死狗 ^_^

之前引用的第三方latex解析库，相关库挂了。抱歉，暂时没腾出时间去修复

Not get IoU as the paper too, ``` 2017-01-12 17:13:48,914 Host Done 1211/2000 with speed: 0.14/s 2017-01-12 17:13:48,914 Host pixel acc: 80.91%, mean acc: 53.63%, mean iou: 43.45% 2017-01-12 17:13:48,916...

Yes, I have done it, ``` 2017-01-23 00:09:03,391 Host Done 2000/2000 with speed: 0.13/s 59996 2017-01-23 00:09:03,391 Host pixel acc: 80.56%, mean acc: 53.51%, mean iou: 43.35% 59997 2017-01-23 00:09:03,393...

> init graph from odps table is naturally supported on max compute platform, so what is your startup command? It occurs when I use EasyRec repo and follow [doc](https://easyrec.readthedocs.io/en/latest/models/dssm_neg_sampler.html). It...

start command using pai ` pai -name easy_rec_ext -project algo_public -Dcmd=train -Dtrain_tables='odps://du_algo_1/tables/ec_list_homepage_recall_dssm_trsample_train_samples/ds=${bizdate}' -Deval_tables='odps://du_algo_1/tables/ec_list_homepage_recall_dssm_trsample_test_samples/ds=${bizdate}' -Dboundary_table='odps://du_algo_1/tables/deal_rec_recall_dssm_featdump_sample_train_sample_pos_train_v1_binning/ds=${bizdate}' -Dcluster='{\"ps\":{\"count\":2,\"cpu\":900,\"memory\":10000},\"worker\":{\"count\":5,\"cpu\":900,\"memory\":10000}}' -Darn='acs:ram::1816563541899700:role/aliyunodpspaidefaultrole' -Dbuckets='oss://algo-recsys.oss-cn-hangzhou-internal.aliyuncs.com/' -Dconfig='oss://algo-recsys/algo/xiahouzuoxin/recall/dssm_trsample/${model_config}.config' -Dmodel_dir="oss://algo-recsys/algo/xiahouzuoxin/recall/dssm_trsample/${model_config}/${bizdate}" -DossHost=oss-cn-hangzhou-internal.aliyuncs.com -- -Dedit_config_json='{\ -- "train_config.num_steps":30000,\ -- "eval_config.num_examples":409600,\ -- "train_config.fine_tune_checkpoint": "oss://algo-recsys/algo/xiahouzuoxin/recall/dssm_trsample/${model_config}/${bizdate_1}",\ --...

It will raise error using tensorflow.python.keras now, https://github.com/tensorflow/tensorflow/issues/61215