From 69b6c54ce2f1ebcf35ec295186b8061eb851f856 Mon Sep 17 00:00:00 2001 From: Jannis Born Date: Tue, 11 Jul 2023 09:56:12 +0200 Subject: [PATCH] Final citations --- examples/regression_transformer/README.md | 19 +++++++++++++------ 1 file changed, 13 insertions(+), 6 deletions(-) diff --git a/examples/regression_transformer/README.md b/examples/regression_transformer/README.md index 53b913a3e..7f1ff0e24 100644 --- a/examples/regression_transformer/README.md +++ b/examples/regression_transformer/README.md @@ -18,11 +18,11 @@ To launch a finetuning of a RT pretrained on drug-like molecules from ChEMBL, ex ``` *NOTE*: This is *dummy* example, do not use "as is" :warning: Substitute the train/test data path to point to your files. You can inspect the format of the example file to see the needed format (`.csv` with the first column called "text" and one or multiple property columns). Adjust the remaining arguments as desired. See full API [here](https://gt4sd.github.io/gt4sd-core/api/gt4sd.training_pipelines.regression_transformer.core.html). -*NOTE*: Substitute the `--model_path` to the directorry of the pretrained model. GT4SD provides RT models pretrained on: +*NOTE*: Substitute the `--model_path` to the directory of the pretrained model. GT4SD provides RT models pretrained on: - **small molecules**: `qed`, `esol`, `crippen_logp` ) or multiple ( `logp_and_synthesizability`, `cosmo_acdl`, `pfas`) properties - **proteins**: `stability` - **chemical reactions**: `uspto` -- **polymers**: `rop_catalyst` and `block_copolymer`, both described in [Park et al., (2022; ChemRxiv)](https://chemrxiv.org/engage/chemrxiv/article-details/62b60865e84dd185e60214af) +- **polymers**: `rop_catalyst` and `block_copolymer`, both described in [Park et al., (2023; Nature Communications)](https://www.nature.com/articles/s41467-023-39396-3) For details on these model versions, see the [HF spaces app](https://huggingface.co/spaces/GT4SD/regression_transformer). *NOTE*: :warning: The above assumes that you have the `qed` model cached locally. If this is not the case, run an inference to trigger the caching mechanism: @@ -90,10 +90,17 @@ Afterwards everybody can sync your model from the hub and run it just like our o For details on this methodology see: ```bib -@article{born2022regression, - title={Regression Transformer enables concurrent sequence regression and generation for molecular language modeling}, +@article{born2023regression, + title={Regression Transformer enables concurrent sequence regression and generation for molecular language modelling}, author={Born, Jannis and Manica, Matteo}, journal={Nature Machine Intelligence}, - note={Article in press. arXiv preprint arXiv:2202.01338}, - year={2023} + year={2023}, + month={04}, + day={06}, + volume={5}, + number={4}, + pages={432--444}, + doi={10.1038/s42256-023-00639-z}, + url={https://doi.org/10.1038/s42256-023-00639-z}, } +```