diff --git a/README.md b/README.md index 1a24f49..1568f66 100644 --- a/README.md +++ b/README.md @@ -9,8 +9,9 @@ REINVENT is a molecular design tool for de novo design, scaffold hopping, R-group replacement, linker design, molecule optimization, and other small molecule design tasks. At its heart, REINVENT uses a Reinforcement Learning (RL) algorithm to generate optimized molecules compliant with a user defined -property profile defined as a multi-component score. See AUTHORS.md for -paper references. +property profile defined as a multi-component score. + +A preprint is available on ChemRxiv: [REINVENT4: Modern AI-Driven Generative Molecule Design](https://chemrxiv.org/engage/chemrxiv/article-details/65463cafc573f893f1cae33a). See AUTHORS.md for references to previous papers. Requirements diff --git a/reinvent/runmodes/TL/learning.py b/reinvent/runmodes/TL/learning.py index f70111b..0c20ed6 100644 --- a/reinvent/runmodes/TL/learning.py +++ b/reinvent/runmodes/TL/learning.py @@ -105,9 +105,10 @@ def __init__( s = DataStructs.BulkTanimotoSimilarity(fps[n], fps[n + 1 :]) sim.extend(s) - self.tb_reporter.add_histogram( - "Tanimoto input SMILES", np.array(sim), 0 - ) + # FIXME: TB broken + #self.tb_reporter.add_histogram( + # "Tanimoto input SMILES", np.array(sim), 0 + #) # FIXME: this is only available for Mol2mol if self._config.max_sequence_length: diff --git a/reinvent/runmodes/TL/reports/tensorboard.py b/reinvent/runmodes/TL/reports/tensorboard.py index 126ae93..f214052 100644 --- a/reinvent/runmodes/TL/reports/tensorboard.py +++ b/reinvent/runmodes/TL/reports/tensorboard.py @@ -59,7 +59,8 @@ def write_report(reporter, data, model: ModelAdapter, is_reinvent: bool, dataloa reporter.add_scalar("KL divergence", kl_div, data.epoch) - reporter.add_histogram("Tanimoto similarity on RDKitFingerprint", similarities, data.epoch) + # FIXME: TB broken + #reporter.add_histogram("Tanimoto similarity on RDKitFingerprint", similarities, data.epoch) # FIXME: rows and cols depend on sample_batch_size image_tensor = make_grid_image(smilies, sample_nlls, "NLL", ROWS * COLS, ROWS)