i-machine-think / machine

Codebase for the i-machine-think project, a modular and fully tested Pytorch implementation for seq2seq models.

i-machine-think/machine Issues

When val_data is None, it breaks checkpoint saving
Updated 5 years ago
DEP: Change `size_average` attribute to `reduction` in NLLLoss and Perplexity classes
Closed 5 years ago
Feature Request: More sophisticated initialisation methods
Updated 6 years ago
Feature Request: Encapsulate Logic in train_model.py into functions
Closed 6 years ago
feature request: make wrapper around nn.Module
Updated 6 years ago
Feature request: basemodel compatibility with other modules than baseRNN
Updated 6 years ago
Error 59 on cuda with hard guidance.
Closed 6 years ago5
SymbolRewritingAccuracy doesn't work without --ignore_output_eos
Closed 6 years ago2
Embedding size in decoder
Closed 6 years ago4
Fix use of validation and dev set
Closed 6 years ago
padding and ignore_index
Closed 6 years ago
max length is target dependent, interferes when pondering is not None
Closed 6 years ago1
ponderer not used in evaluate
Closed 6 years ago1
learning package
Closed 7 years ago