Learned Abstract Model for Imperfect-information Reasoning

This is an implementation of algorithm Learned Abstract Model for Imperfect-information Reasoning (LAMIR).

The algorithms consists of 2 main parts.

Training part in lamis_train.py, which contains implementation of RNaD (inspired by the OpenSpiel) and trains all the necessary models
CFR part, which is an JAX implementation of CFR, which is used for depth-limited search (look-ahead reasoning).

The games implemented are

Setting

Tested with Python 3.13.3 Use python -m pip install -r requirements.txt

Name		Name	Last commit message	Last commit date
Latest commit History 1 Commit
experiments		experiments
games		games
plots		plots
README.md		README.md
lamis_cfr.py		lamis_cfr.py
lamis_gameplay.py		lamis_gameplay.py
lamis_networks.py		lamis_networks.py
lamis_train.py		lamis_train.py
requirements.txt		requirements.txt
test_gumbel.py		test_gumbel.py
train_utils.py		train_utils.py