Smiles model - #9
Open
jsture wants to merge 4 commits into
Open
Conversation
Add --representation {SELFIES,SMILES} and --molecule_column to the MLM
training entrypoint. The representation threads through tokenizer default
path resolution, metadata representation assertion, APE tokenizer
construction, sample-shape and ethanol compatibility checks, run metadata,
and the generated model card. --selfies_column is kept as a back-compat
alias for --molecule_column, and the resolved column name continues to
drive the representation-agnostic dataset code, so SELFIES runs are
unchanged. SMILES reuses the already-parameterized utils helpers.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Add --representation to run_sweep.py. A REPRESENTATION_DEFAULTS table selects the tokenizer, dataset column, masking grid, and run-root tag per representation. SMILES points at the SMILES APE tokenizer and the smiles_canonical_clean column (reusing the existing chembl36 dataset directory read-only) and drops hetero_span, whose heteroatom bias is SELFIES-bracket-specific. SELFIES keeps its historical defaults and emits the same per-run training commands. hetero_span is rejected for SMILES. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Add a representation field to the ModernMolBERT featurizer. For a SMILES checkpoint, input SMILES are tokenized directly; for a SELFIES checkpoint the existing SMILES->SELFIES conversion is preserved. _load_ape_tokenizer and the emitted metadata become representation-aware. embed_modernmolbert gains --representation to select the path, defaulting to SELFIES so current benchmarking is unchanged. Adds a modernmolbert_smiles featurizer config and tests asserting the SMILES path feeds raw SMILES to the tokenizer with no SELFIES round-trip. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Add --representation and --molecule_column to fixed_eval_best_models and parameterize the validation-set column readers. SMILES defaults the column to smiles_canonical_clean; SELFIES keeps selfies. Lets the SMILES sweep be evaluated on its own validation strings when selecting best checkpoints. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
No description provided.