Thank you!
deeplearninganalytics.com
warning: opinions
Easy to start?
pip install:
✘ Caffe
✔ TensorFlow
"Easy" models:
½ Caffe
✔ TensorFlow
Easy to develop?
API level:
Python integration:
½ Caffe
✔ TensorFlow
modular design
layer {name: "data"
type: "Data"
top: "data"
top: "label"
transform_param {crop_size: 227}
data_param {source: "train_lmdb_path"
batch_size: 256
backend: LMDB}}
multi-GPU
Caffe
TensorFlow
tf.device()
multi-machine
Caffe
TensorFlow
tf.device()
responsiveness
merge time:
Thanks!
@planarrowspace