A comparison of Theano with other deep learning frameworks, highlighting a series of low-level design choices in no particular order.
Overview
Symbolic: Theano, CGT; Automatic: Torch, MXNet
Symbolic and automatic differentiation are often confused or used interchangeably, although their implementations are significantly different.
| th test/test_qr.lua | |
| Running 36 tests | |
| _____________*_*_*_*________________ ==> Done Completed 193792 asserts in 36 tests with 16 errors | |
| -------------------------------------------------------------------------------- | |
| randomMediumQR_inPlace_torch.FloatTensor | |
| QR = A | |
| TensorEQ(==) violation val=1.5631318092346e-05, condition=1e-05 | |
| /Users/bartvm/torch/install/share/lua/5.1/torch/Tester.lua:61: in function 'assertTensorEq' | |
| test/test_qr.lua:80: in function 'checkQR' |
| import argparse | |
| import sqlite3 | |
| from uuid import UUID | |
| from flask import g, json, Flask | |
| from blocks.config import config | |
| from blocks.log.sqlite import ANCESTORS_QUERY | |
| app = Flask(__name__) |