OOM Embedding

[linusic]

My Vocab size is 10k+
num_encoder_symbols=num_encoder_symbols,
num_decoder_symbols=num_decoder_symbols,
so much params and OOM...................
12G Memory
batch_size = 1
unit = 8
still OOM......................

[DataXujing]

Now u batch_size is 1. If u train in CUDA GPU u can config some params in lx_bot_3.py line 195 or made the input_seq_len and output_seq_len smaller