NickleDave/thrillington
Replication of "Recurrent models of visual attention", Mnih et al. 2014
PythonBSD-3-Clause
Issues
- 0
- 1
- 1
- 1
add random seed option to train!!!
#22 opened by NickleDave - 1
add ability to feed deterministic `l_0` to test how glimpses vary as function of intial loc
#35 opened by NickleDave - 1
- 0
- 0
add `save_nan` option
#39 opened by NickleDave - 0
- 0
add analysis Makefile that parses logs and spits out test acc as function of hyperparams
#37 opened by NickleDave - 0
- 0
fix how nan'd models are saved
#33 opened by NickleDave - 1
add Patience
#31 opened by NickleDave - 1
- 1
include Element license from torch/rnn
#30 opened by NickleDave - 0
- 1
add [test] section to config
#19 opened by NickleDave - 1
- 1
add adam as option for optimizer
#21 opened by NickleDave - 1
fix / finish tests! e.g. for modules
#17 opened by NickleDave - 1
logger should log which checkpoint it uses!
#18 opened by NickleDave - 1
Fix momentum=constant in trainer
#13 opened by NickleDave - 0
Add other MNISTs e.g. MNIST cluttered
#15 opened by NickleDave - 1
add logging!
#3 opened by NickleDave - 1
add 'replicates' option
#7 opened by NickleDave - 1
just have one output dir instead of samples_dir, examples_dir, train_inds_dir, etc
#5 opened by NickleDave - 1
- 1