This model is a scratch trained version of RoBERTa on the PubChem10M dataset.
It achieves the following results on the evaluation set:
RoBERTa molecular language model. Trained on the
Self-Referencing Embedded Strings (SELFIES) molecular representation.
Tokenizer trained with a SELFIES semantically robust alphabet.
Used to extract embeddings of molecular representations, for downstream tasks (eg. binding affinity, toxicity prediction).
Trained on a full split of the
PubChem10M dataset, converted to SELFIES.
Trained using
run_mlm.py from the 🤗 examples, using torchrun for multi-GPU.
The computations described in this research were performed using the Baskerville Tier 2 HPC service (
https://www.baskerville.ac.uk/). Baskerville was funded by the EPSRC and UKRI through the World Class Labs scheme (EP/T022221/1) and the Digital Research Infrastructure programme (EP/W032244/1) and is operated by Advanced Research Computing at the University of Birmingham.
