DeepRec

DeepRec copied to clipboard

lockmatrix

fuhailin

[CustomFileSystem] Build custom file system with DeepRec throw undefined symbol in runtime.

1

内部版本ps想使用DeepRec的推理能力，移植遇到undefined symbol异常。 参考https://tensorflow.juejin.im/extend/add_filesys.html 定义tensorflow的FileSystem ，引用DeepRec抛undefined symbol具体错误如下： (1) 在load so的时候失败： lib_handle = py_tf.TF_LoadLibrary(library_filename) tensorflow.python.framework.errors_impl.NotFoundError: (2) 抛undefined symbol异常： undefined symbol: _ZN10tensorflow10FileSystem10FilesExistERKSt6vectorISsSaISsEEPS1_INS_6StatusESaIS6_EE (3) 尝试strings查看对应的导出符号： strings libtensorflow_framework.so.1 |grep tensorflow10FileSystem10FilesExistERKSt _ZN10tensorflow10FileSystem10FilesExistERKSt6vectorISsSaISsEEPS1_INS_6StatusESaIS6_EE _ZN10tensorflow10FileSystem10FilesExistERKSt6vectorISsSaISsEEPS1_INS_6StatusESaIS6_EE (4) 最终定位出来是应为ABI不兼容导致，在编译DeepRec的时候，添加ABI=0的参数， 具体编译参数如下：...

welsonzhang

[OSPP 2022] DeepRec supports exporting models to key-value NoSQL databases -> The first feature submission

2

Complete the first feature submission, please do not merge for now, there will be update submissions in the future.

MrRobotsAA

[CIBUILD] Upgrade cibuild docker images to ubuntu22.04 with cuda11.7.

1

liutongxuan

有升级到TF 2.x版本的计划和排期吗？

1

**System information** - DeepRec version (you are using): 1.15 - Are you willing to contribute it (Yes/No): No **Describe the feature and the current behavior/state.** 有升级到TF 2.x版本的计划和排期吗？

sysofai

The issue with "HBM_DRAM Restore" causing a core dump when id_num > cachesize arises due to triggering unnecessary eviction when HBM is empty. To address this, adding a temporary restore_cache_...

RobertLou

JackMoriarty

Mesilenceki

warm_up和EV连用的时候，ckpt里面找不到Ev

1

模型继承训练的代码是这样 ` ## 模型配置文件 warm_start_conf = tf.estimator.WarmStartSettings(ckpt_to_initialize_from=warm_up_model_dir) model = import_module(self.train_config['model']) classifier = tf.estimator.Estimator(model_fn=model.model_fn, params={ 'feature_columns': feature_columns, 'config': self.train_config, 'distribution': self.distribution, 'feature_columns_dict': feature_columns_dict }, warm_start_from=warm_start_conf, config=run_config)` 在继承训练，load模型文件文件ckpt的时候报错 模型文件里面没有也没有input_layer/1000_embedding/embedding_weights这个key，只有下面的 这个该如何解决？

Confucius-hui