Port Tensorflow 1 code to Pytorch

Job ID: 36844796

Budget: $30 – $250 USD

I am looking for a freelancer who can help me port my TensorFlow 1 codebase to PyTorch. The codebase is small, consisting of 150 lines, most of which will remain unchanged. It runs as is with no errors. I have a specific deadline for the project completion, which is within a month. The main goal is getting rid of the old style session, placeholder, and feed_dict that TF1 used. I also want it to produce the same output:

learning_rate,mbs,reg_const,epoch,train_loss,train_acc,cpu_time,act_fn,num_hids_per_layer, test_loss,test_acc
0.01,194,0.001,0.0,0.69058764,0.5051546,0.0,tanh,5,0.6900423,0.5103093
0.01,194,0.001,40.0,0.63682634,0.5979381,0.16499924659729004,tanh,5,0.6377007,0.58762884
0.01,194,0.001,80.0,0.5656287,0.66494846,0.21006464958190918,tanh,5,0.56850326,0.6597938
0.01,194,0.001,120.0,0.51699185,0.6597938,0.2546043395996094,tanh,5,0.523661,0.6597938
0.01,194,0.001,160.0,0.48181114,0.7010309,0.29760026931762695,tanh,5,0.496608,0.7010309

Skills and experience required for this job:
- Experience with code migration and conversion between TensorFlow 1 and PyTorch
- Familiarity with neural net PyTorch libraries

If you have the necessary skills and experience, and can meet the deadline, please bid on this project.
Related categories: Python Neural Networks