python - Error using tensorflow: needs protobuf 3.0.0 but finds 2.6.1 - pip3 shows 3.0.0b2 installed -

running newest python code (which uses keras/tensorflow), error:

[libprotobuf fatal google/protobuf/stubs/common.cc:61] program requires version 3.0.0 of protocol buffer runtime library, installed version 2.6.1.  please update library.  if compiled program yourself, make sure headers same version of protocol buffers link-time library.  (version verification failed in "external/protobuf/src/google/protobuf/any.pb.cc".) terminate called after throwing instance of 'google::protobuf::fatalexception'   what():  program requires version 3.0.0 of protocol buffer runtime library, installed version 2.6.1.  please update library.  if compiled program yourself, make sure headers same version of protocol buffers link-time library.  (version verification failed in "external/protobuf/src/google/protobuf/any.pb.cc".) 

however, when doing pip3 list get:

protobuf (3.0.0b2) tensorflow (0.9.0) 

(among others)

i'm using ubuntu 16.04, running cuda 7.5 on nvidia 1070 gtx. have updated latest versions of relevant packages code , have uninstalled , reinstalled protobuf , tensorflow.

this error occurs code wrote today, not other code ran on machine before. there wrong code, error message not pointing me there.

can help? thanks.

i'm using ubuntu 16.04, tensorflow 0.90 installed source, cuda 8.0rc & cudann5. had exact same error keras code utilised tensorflow backend.

it's not 100% clear problem lies, doesn't appear when running tensorflow code directly, nor skflow. far have encountered problem while using keras, after exploring keras backend code, cannot see looks suspicious.

however, have solution works me. chris suggested, if ensure first import occurs import tensorflow tf, loads expected , works. guess there module conflict other packages loading in different version of protobuf before tensorflow can, perhaps.