compo_vs_generalization_ood

This code implements the experiments reported in the following paper:

• Defending Compositionality in Emergent Languages. Michal Auersperger, Pavel Pecina. NAACL SRW 2022. [arxiv]

and is an extension of the experiments in egg/zoo/compo_vs_generalization

A single experiment of the full communication game can be run as follows:

python -m egg.zoo.compo_vs_generalization_ood.train  --n_values=50 --n_attributes=2 --vocab_size=50 --max_len=3  --receiver=ModifReceiver --sender=ModifSender --hidden=50 --batch_size=64 --random_seed=1

To run an experiment with a single agent on half of the problem only (i.e., a learning alone experiment from the paper), run e.g.:

python -m egg.zoo.compo_vs_generalization_ood.learning_alone.train  --n_values=50 --n_attributes=2 --vocab_size=50 --max_len=5 --archpart=sender --model=OrigSenderDeterministic --hidden=50 --batch_size=64 --random_seed=1 

See the article for further details.

To replicate the full experiments, use the hyperparameters in ./hyperparams/modified_arch.json and ./hyperparams/orig_arch.json.

To replicate the learning alone experiments, use the hyperparameters in ./hyperparams/learning_alone/receiver.json and ./hyperparams/learning_alone/sender.json.

For convenience, we attach the log files of our full experiment runs in results/orig_arch/220520T115546 and results/modified_arch/220517T231916. The notebook ntb-results-full.ipynb processes the logs and produces Table 2 and Figure 1 from the paper.