Getting error running the code as instructed #1
Description
Hello Yitian,
Hope this email finds you well.
I came across your interesting paper "On the Design of Channel Coding Autoencoders with Arbitrary Rates for ISI Channels", and wanted to run your implementation and learn more about it. However, when following your instruction for running the code, I face this error:
return forward_call(*args, **kwargs)
File ".../ChannelCoding_AE-main/decoders.py", line 68, in forward
code= code.view(self.args.batch_size,self.args.block_len, self.args.code_rate_n) #(batch_size,block_len,n) [ab,ab]
RuntimeError: shape '[1000, 100, 3]' is invalid for input of size 400000
Could you please let me know if you have an update with the fix? Would highly appreciate it.
Needless to say that any use of this code/paper, will be with proper citation/reference.
BTW, I follow this line as you instructed to run the code:
(1) To train the model (R = 2/3):
python main.py -encoder rate_high -decoder rate_high -enc_num_unit 25 -enc_num_layer 2 -enc_act linear -dec_num_unit 100 -dec_num_layer 2 -dec_act sigmoid -channel isi -code_rate_k 2 -code_rate_n 3 -train_enc_channel_low 0 -train_enc_channel_high 8 -snr_test_start 0 -snr_test_end 12.0 -snr_points 13 -is_parallel 1 -is_interleave 0 -train_dec_channel_low 0 -train_dec_channel_high 8 -dec_lr 0.001 -enc_lr 0.001 -num_block 100000 -batch_size 1000 -train_channel_mode block_norm -test_channel_mode block_norm --print_test_traj -loss bce -num_epoch 80 -joint_train 1