Update paper.md

reports/paper.md

@@ -82,10 +82,10 @@ The `matbench_genmetrics.mp_time_split` namespace package provides the following
 
 In future work, metrics will serve as multi-criteria filters to prevent manipulation. Standalone metrics can be "hacked" by generating nonsensical structures for novelty or including training structures to inflate validity scores. To address this, multiple criteria are considered simultaneously for each generated structure, such as novelty, uniqueness, and filtering rules like non-overlapping atoms, stoichiometry, or checkCIF criteria [@spek_checkcif_2020]. Additional filters based on machine learning models can be applied for properties like negative formation energy, energy above hull, ICSD classification, and coordination number. Applying machine-learning-based structural relaxation using M3GNet [@chen_universal_2022] (e.g., as in CrysTens [@alverson_generative_2024]) before filtering is also of interest. Contributions related to multi-criteria filtering, enhanced validity filters, and implementing a benchmark submission system and public leaderboard are welcome.
 
-We believe that the `matbench-genmetrics` ecosystem is a robust and easy-to-use benchmarking platform that will help propel novel materials discovery and targeted crystal structure inverse design. We hope that practioners of crystal structure generative modeling will adopt `matbench-genmetrics`, contribute improvements and ideas, and submit their results to the planned public leaderboard.
+We believe that the `matbench-genmetrics` ecosystem is a robust and easy-to-use benchmarking platform that will help propel novel materials discovery and targeted crystal structure inverse design. We hope that practitioners of crystal structure generative modeling will adopt `matbench-genmetrics`, contribute improvements and ideas, and submit their results to the planned public leaderboard.
 
 # Acknowledgements
 
-We acknowledge contributions from Kevin M. Jablonka ([\@kjappelbaum](https://github.com/kjappelbaum)), Matthew K. Horton ([\@mkhorton](https://github.com/mkhorton)), Kyle D. Miller ([\@kyledmiller](https://github.com/kyledmiller)), and Janosh Riebesell ([\@janosh](https://github.com/janosh)). S.G.B. and T.D.S. acknowledge support by the National Science Foundation, USA under Grant No. DMR-1651668. We acknowledge OpenAI for the use of ChatGPT for basic proofreading and editing.
+We acknowledge contributions from Kevin M. Jablonka ([\@kjappelbaum](https://github.com/kjappelbaum)), Matthew K. Horton ([\@mkhorton](https://github.com/mkhorton)), Kyle D. Miller ([\@kyledmiller](https://github.com/kyledmiller)), and Janosh Riebesell ([\@janosh](https://github.com/janosh)). S.G.B. and T.D.S. acknowledge support by the National Science Foundation, USA under Grant No. DMR-1651668. We acknowledge OpenAI for the use of ChatGPT for basic proofreading and editing, such as asking for more concise and clearer wording or general feedback.
 
 # References