Add paper and GitHub links, update task categories (#2)

- Add paper and GitHub links, update task categories (14a2ab4a79c0af9a311e03dd74c80ea79f0a7afb)


Co-authored-by: Niels Rogge <nielsr@users.noreply.huggingface.co>

README.md

@@ -1,40 +1,43 @@
 ---
-license:
-  - pddl
-license_link: https://opendatacommons.org/licenses/pddl
-tags:
-  - pubchem
-  - small-molecules
-  - InChI
-  - SMILES
-  - molecular-geometry
-  - molecular-properties
-  - chemical-properties
-  - cheminformatics
 annotations_creators:
-  - crowdsourced
-pretty_name: pubchem-04-18-2025
+- crowdsourced
+license:
+- pddl
 size_categories:
-  - 100M<n<200M
+- 100M<n<200M
 source_datasets:
-  - pubchem-compound
-  - pubchem-04-18-2025
+- pubchem-compound
+- pubchem-04-18-2025
 task_categories:
-  - tabular-regression
-  - other
+- tabular-regression
+- feature-extraction
+- other
 task_ids:
-  - tabular-single-column-regression
+- tabular-single-column-regression
+pretty_name: pubchem-04-18-2025
+license_link: https://opendatacommons.org/licenses/pddl
+tags:
+- pubchem
+- small-molecules
+- InChI
+- SMILES
+- molecular-geometry
+- molecular-properties
+- chemical-properties
+- cheminformatics
 viewer: false
 configs:
-  - config_name: pubchem-04-18-2025
-    data_files:
-      - split: train
-        path: "data/train/*.json"
-    default: true
+- config_name: pubchem-04-18-2025
+  data_files:
+  - split: train
+    path: data/train/*.json
+  default: true
 ---
 
 # PubChem Dataset (version 04-18-2025)
 
+[Paper](https://huggingface.co/papers/2603.13627) | [Code](https://github.com/molssi-ai/bertology) | [Project Homepage](https://pubchem.ncbi.nlm.nih.gov)
+
 * **Important Note (Jul. 22, 2025):** According to [Hugging Face](https://github.com/LiveCodeBench/LiveCodeBench/issues/108),
   the usage of data load scripts and the corresponding `trust_remote_code` parameter in the `load_dataset()`
   is deprecated due to security risks. Users should temporarily use `datasets<4.0.0` to be able to use our
@@ -93,6 +96,8 @@ macromolecules. PubChem collects information on chemical structures,
 identifiers, chemical and physical properties, biological activities, patents,
 health, safety, toxicity data, and many others.
 
+This version of the dataset was prepared for the project **BERTology of Molecular Property Prediction**, which investigates the effects of various factors on the pre-training and fine-tuning performance of BERT-based chemical language models.
+
 ## Dataset Structure
 
 ### Data Instances
@@ -307,8 +312,18 @@ The PubChem dataset paves the way for applications in drug discovery and materia
     year = {2022},
     doi = {10.1093/nar/gkac956}
 }
+
+@misc{mostafanejad:2026:bertology,
+      title={BERTology of Molecular Property Prediction},
+      author={Mohammad Mostafanejad and Paul Saxe and T. Daniel Crawford},
+      year={2026},
+      eprint={2603.13627},
+      archivePrefix={arXiv},
+      primaryClass={cs.LG},
+      url={https://arxiv.org/abs/2603.13627},
+}
 ```
 
 ### Contributions
 
-- **Mohammad Mostafanejad**, The Molecular Sciences Software Institute (MolSSI)
+- **Mohammad Mostafanejad**, The Molecular Sciences Software Institute (MolSSI)