| { |
| "checkpoints": [ |
| { |
| "auxiliary_files": [ |
| "aux/bart_vocab.json", |
| "aux/bart_vocab_downstream.json", |
| "aux/ds_config.json", |
| "aux/hparams.yaml", |
| "aux/hyper_parameters.json", |
| "benchmarks/default_test.jsonl", |
| "benchmarks/metadata.json", |
| "benchmarks/native/uspto_mixed.pickle" |
| ], |
| "benchmark_datasets": [ |
| { |
| "benchmark_semantics": "forward_product_prediction_with_mixed_inputs", |
| "benchmark_variant": "forward_product_prediction", |
| "dataset_id": "uspto_mixed_forward_checkpoint_test", |
| "dataset_ref": "uspto_mixed_chemformer_project_release", |
| "default": true, |
| "format": "jsonl", |
| "metadata_path": "benchmarks/metadata.json", |
| "native_input_format": "pandas_pickled_dataframe", |
| "native_source_paths": [ |
| "benchmarks/native/uspto_mixed.pickle" |
| ], |
| "notes": [ |
| "Native Chemformer forward benchmark source extracted from the public Box release uspto_mixed pickle.", |
| "benchmarks/default_test.jsonl preserves reactant-side input SMILES plus expected product SMILES from the test partition." |
| ], |
| "num_examples": 40000, |
| "path": "benchmarks/default_test.jsonl", |
| "record_schema": { |
| "inputs": [ |
| "reactants" |
| ], |
| "targets": [ |
| "expected_products" |
| ] |
| }, |
| "split": "test", |
| "task_direction": "forward" |
| } |
| ], |
| "checkpoint_id": "uspto_mixed_forward_checkpoint", |
| "embedding_supported": false, |
| "input_schema": { |
| "reactants_field": "reactants_mol", |
| "reagents_field": null, |
| "serialization": "dot_joined_smiles", |
| "source_dataset_member": "data/seq-to-seq_datasets/uspto_mixed.pickle", |
| "target_field": "products_mol" |
| }, |
| "model_id": "chemformer_forward", |
| "normalized_checkpoint_id": "uspto_mixed_forward_checkpoint", |
| "package_dir": "checkpoints/chemformer_forward/uspto_mixed_forward_checkpoint", |
| "source_registry_refs": [ |
| { |
| "checkpoint_id": "uspto_mixed_forward_checkpoint", |
| "model_id": "chemformer_forward" |
| } |
| ], |
| "task_direction": "forward", |
| "training_dataset_refs": [ |
| "uspto_mixed_chemformer_project_release" |
| ], |
| "weights_path": "model.safetensors", |
| "wrapper_family": "chemformer_bart" |
| }, |
| { |
| "auxiliary_files": [ |
| "aux/bart_vocab.json", |
| "aux/bart_vocab_downstream.json", |
| "aux/ds_config.json", |
| "aux/hparams.yaml", |
| "aux/hyper_parameters.json", |
| "benchmarks/default_test.jsonl", |
| "benchmarks/metadata.json", |
| "benchmarks/native/uspto_sep.pickle" |
| ], |
| "benchmark_datasets": [ |
| { |
| "benchmark_semantics": "forward_product_prediction_with_separated_reagents", |
| "benchmark_variant": "forward_product_prediction", |
| "dataset_id": "uspto_sep_forward_checkpoint_test", |
| "dataset_ref": "uspto_sep_chemformer_project_release", |
| "default": true, |
| "format": "jsonl", |
| "metadata_path": "benchmarks/metadata.json", |
| "native_input_format": "pandas_pickled_dataframe", |
| "native_source_paths": [ |
| "benchmarks/native/uspto_sep.pickle" |
| ], |
| "notes": [ |
| "Native Chemformer forward benchmark source extracted from the public Box release uspto_sep pickle.", |
| "benchmarks/default_test.jsonl preserves separated reactants/reagents fields plus expected product SMILES from the test partition." |
| ], |
| "num_examples": 40000, |
| "path": "benchmarks/default_test.jsonl", |
| "record_schema": { |
| "inputs": [ |
| "reactants", |
| "reagents" |
| ], |
| "targets": [ |
| "expected_products" |
| ] |
| }, |
| "split": "test", |
| "task_direction": "forward" |
| } |
| ], |
| "checkpoint_id": "uspto_sep_forward_checkpoint", |
| "embedding_supported": false, |
| "input_schema": { |
| "reactants_field": "reactants_mol", |
| "reagents_field": "reagents_mol", |
| "serialization": "dot_joined_smiles", |
| "source_dataset_member": "data/seq-to-seq_datasets/uspto_sep.pickle", |
| "target_field": "products_mol" |
| }, |
| "model_id": "chemformer_forward", |
| "normalized_checkpoint_id": "uspto_sep_forward_checkpoint", |
| "package_dir": "checkpoints/chemformer_forward/uspto_sep_forward_checkpoint", |
| "source_registry_refs": [ |
| { |
| "checkpoint_id": "uspto_sep_forward_checkpoint", |
| "model_id": "chemformer_forward" |
| } |
| ], |
| "task_direction": "forward", |
| "training_dataset_refs": [ |
| "uspto_sep_chemformer_project_release" |
| ], |
| "weights_path": "model.safetensors", |
| "wrapper_family": "chemformer_bart" |
| }, |
| { |
| "auxiliary_files": [ |
| "benchmarks/default_test.jsonl", |
| "benchmarks/metadata.json", |
| "benchmarks/native/src-test.txt", |
| "benchmarks/native/tgt-test.txt" |
| ], |
| "benchmark_datasets": [ |
| { |
| "benchmark_semantics": "forward_product_prediction_with_mixed_inputs", |
| "benchmark_variant": "forward_product_prediction", |
| "dataset_id": "mit_mixed_augm_model_average_20_pt_test", |
| "dataset_ref": "uspto_mit_molecular_transformer_project_release", |
| "default": true, |
| "format": "jsonl", |
| "metadata_path": "benchmarks/metadata.json", |
| "native_input_format": "parallel_tokenized_smiles", |
| "native_source_paths": [ |
| "benchmarks/native/src-test.txt", |
| "benchmarks/native/tgt-test.txt" |
| ], |
| "notes": [ |
| "Native Molecular Transformer benchmark sources are the official tokenized test split files from the public Box release.", |
| "benchmarks/default_test.jsonl stores detokenized mixed-input reactant strings mapped to expected product SMILES for forward evaluation." |
| ], |
| "num_examples": 40000, |
| "path": "benchmarks/default_test.jsonl", |
| "record_schema": { |
| "inputs": [ |
| "reactants" |
| ], |
| "targets": [ |
| "expected_products" |
| ] |
| }, |
| "serialized_input_format": "space_tokenized_smiles", |
| "split": "test", |
| "task_direction": "forward" |
| } |
| ], |
| "checkpoint_id": "mit_mixed_augm_model_average_20_pt", |
| "embedding_supported": false, |
| "input_schema": { |
| "benchmark_variant": "forward_product_prediction", |
| "reactants_field": "reactants_mixed", |
| "serialized_format": "space_tokenized_smiles", |
| "target_field": "PRODUCT", |
| "tokenizer": "smiles_regex_space_tokenizer" |
| }, |
| "model_id": "molecular_transformer_forward", |
| "normalized_checkpoint_id": "mit_mixed_augm_model_average_20_pt", |
| "package_dir": "checkpoints/molecular_transformer_forward/mit_mixed_augm_model_average_20_pt", |
| "source_registry_refs": [ |
| { |
| "checkpoint_id": "mit_mixed_augm_model_average_20_pt", |
| "model_id": "molecular_transformer_forward" |
| } |
| ], |
| "task_direction": "forward", |
| "training_dataset_refs": [ |
| "uspto_mit_molecular_transformer_project_release" |
| ], |
| "weights_path": "model.pt", |
| "wrapper_family": "opennmt_transformer" |
| }, |
| { |
| "auxiliary_files": [ |
| "benchmarks/default_test.jsonl", |
| "benchmarks/metadata.json", |
| "benchmarks/native/src-test.txt", |
| "benchmarks/native/tgt-test.txt" |
| ], |
| "benchmark_datasets": [ |
| { |
| "benchmark_semantics": "forward_product_prediction_with_mixed_inputs", |
| "benchmark_variant": "forward_product_prediction", |
| "dataset_id": "stereo_mixed_augm_model_average_20_pt_test", |
| "dataset_ref": "uspto_stereo_molecular_transformer_project_release", |
| "default": true, |
| "format": "jsonl", |
| "metadata_path": "benchmarks/metadata.json", |
| "native_input_format": "parallel_tokenized_smiles", |
| "native_source_paths": [ |
| "benchmarks/native/src-test.txt", |
| "benchmarks/native/tgt-test.txt" |
| ], |
| "notes": [ |
| "Native Molecular Transformer benchmark sources are the official tokenized test split files from the public Box release.", |
| "benchmarks/default_test.jsonl stores detokenized mixed-input reactant strings mapped to expected product SMILES for forward evaluation." |
| ], |
| "num_examples": 50258, |
| "path": "benchmarks/default_test.jsonl", |
| "record_schema": { |
| "inputs": [ |
| "reactants" |
| ], |
| "targets": [ |
| "expected_products" |
| ] |
| }, |
| "serialized_input_format": "space_tokenized_smiles", |
| "split": "test", |
| "task_direction": "forward" |
| } |
| ], |
| "checkpoint_id": "stereo_mixed_augm_model_average_20_pt", |
| "embedding_supported": false, |
| "input_schema": { |
| "benchmark_variant": "forward_product_prediction", |
| "reactants_field": "reactants_mixed", |
| "serialized_format": "space_tokenized_smiles", |
| "target_field": "PRODUCT", |
| "tokenizer": "smiles_regex_space_tokenizer" |
| }, |
| "model_id": "molecular_transformer_forward", |
| "normalized_checkpoint_id": "stereo_mixed_augm_model_average_20_pt", |
| "package_dir": "checkpoints/molecular_transformer_forward/stereo_mixed_augm_model_average_20_pt", |
| "source_registry_refs": [ |
| { |
| "checkpoint_id": "stereo_mixed_augm_model_average_20_pt", |
| "model_id": "molecular_transformer_forward" |
| } |
| ], |
| "task_direction": "forward", |
| "training_dataset_refs": [ |
| "uspto_stereo_molecular_transformer_project_release" |
| ], |
| "weights_path": "model.pt", |
| "wrapper_family": "opennmt_transformer" |
| }, |
| { |
| "auxiliary_files": [ |
| "aux/.gitattributes", |
| "aux/README.md", |
| "aux/UPSTREAM_URL.txt", |
| "aux/config.json", |
| "aux/generation_config.json", |
| "aux/hf_model_info.json", |
| "aux/provenance/README.txt", |
| "aux/provenance/ReactionT5v2_source_README.md", |
| "aux/provenance/datasets/ORD/all_ord_reaction_uniq_with_attr20240506_v3_test.csv", |
| "aux/provenance/datasets/ORD/all_ord_reaction_uniq_with_attr20240506_v3_train.csv", |
| "aux/provenance/datasets/ORD/all_ord_reaction_uniq_with_attr20240506_v3_valid.csv", |
| "aux/provenance/datasets/USPTO_MIT/MIT_separated/test.csv", |
| "aux/provenance/datasets/USPTO_MIT/MIT_separated/train.csv", |
| "aux/provenance/datasets/USPTO_MIT/MIT_separated/val.csv", |
| "aux/provenance/reactiont5v2_forward_split_provenance.json", |
| "aux/special_tokens_map.json", |
| "aux/tokenizer.json", |
| "aux/tokenizer_config.json", |
| "aux/trainer_state.json", |
| "aux/training_args.bin", |
| "benchmarks/default_test.jsonl", |
| "benchmarks/metadata.json" |
| ], |
| "benchmark_datasets": [ |
| { |
| "benchmark_semantics": "ord_trained_forward_checkpoint_evaluated_on_uspto_mit_holdout", |
| "benchmark_variant": "forward_product_prediction", |
| "blank_reagent_value": " ", |
| "dataset_id": "sagawa_reactiont5v2_forward_test", |
| "dataset_ref": "uspto_mit_forward_project_release", |
| "default": true, |
| "format": "jsonl", |
| "metadata_path": "benchmarks/metadata.json", |
| "native_input_format": "csv_forward_reaction_rows", |
| "native_source_paths": [ |
| "aux/provenance/datasets/USPTO_MIT/MIT_separated/test.csv" |
| ], |
| "notes": [ |
| "Native forward benchmark split for the public USPTO_MIT product-prediction task.", |
| "The normalized JSONL preserves reactants, reagents, and expected product(s) rather than the backward exact-match schema.", |
| "This ORD-trained checkpoint is evaluated on the USPTO_MIT holdout documented in the upstream README/model card." |
| ], |
| "num_examples": 40000, |
| "path": "benchmarks/default_test.jsonl", |
| "record_schema": { |
| "inputs": [ |
| "reactants", |
| "reagents" |
| ], |
| "targets": [ |
| "expected_products" |
| ] |
| }, |
| "serialized_input_format": "REACTANT:<reactants>REAGENT:<reagents>", |
| "split": "test", |
| "task_direction": "forward" |
| } |
| ], |
| "checkpoint_id": "sagawa_reactiont5v2_forward", |
| "embedding_supported": false, |
| "input_schema": { |
| "blank_reagent_value": " ", |
| "reactants_field": "REACTANT", |
| "reagents_field": "REAGENT", |
| "serialized_format": "REACTANT:<reactants>REAGENT:<reagents>", |
| "target_field": "PRODUCT" |
| }, |
| "model_id": "reactiont5v2_forward", |
| "normalized_checkpoint_id": "sagawa_reactiont5v2_forward", |
| "package_dir": "checkpoints/reactiont5v2_forward/sagawa_reactiont5v2_forward", |
| "source_registry_refs": [ |
| { |
| "checkpoint_id": "sagawa_reactiont5v2_forward", |
| "model_id": "reactiont5v2_forward" |
| } |
| ], |
| "task_direction": "forward", |
| "training_dataset_refs": [ |
| "ord_forward_pretraining" |
| ], |
| "weights_path": "model.safetensors", |
| "wrapper_family": "seq2seq_t5" |
| }, |
| { |
| "auxiliary_files": [ |
| "aux/.gitattributes", |
| "aux/README.md", |
| "aux/UPSTREAM_URL.txt", |
| "aux/config.json", |
| "aux/generation_config.json", |
| "aux/hf_model_info.json", |
| "aux/provenance/README.txt", |
| "aux/provenance/ReactionT5v2_source_README.md", |
| "aux/provenance/datasets/ORD/all_ord_reaction_uniq_with_attr20240506_v3_test.csv", |
| "aux/provenance/datasets/ORD/all_ord_reaction_uniq_with_attr20240506_v3_train.csv", |
| "aux/provenance/datasets/ORD/all_ord_reaction_uniq_with_attr20240506_v3_valid.csv", |
| "aux/provenance/datasets/USPTO_MIT/MIT_separated/test.csv", |
| "aux/provenance/datasets/USPTO_MIT/MIT_separated/train.csv", |
| "aux/provenance/datasets/USPTO_MIT/MIT_separated/val.csv", |
| "aux/provenance/reactiont5v2_forward_split_provenance.json", |
| "aux/special_tokens_map.json", |
| "aux/tokenizer.json", |
| "aux/tokenizer_config.json", |
| "aux/trainer_state.json", |
| "aux/training_args.bin", |
| "benchmarks/default_test.jsonl", |
| "benchmarks/metadata.json" |
| ], |
| "benchmark_datasets": [ |
| { |
| "benchmark_semantics": "benchmark_specific_finetuned_forward_checkpoint", |
| "benchmark_variant": "forward_product_prediction", |
| "blank_reagent_value": " ", |
| "dataset_id": "sagawa_reactiont5v2_forward_uspto_mit_test", |
| "dataset_ref": "uspto_mit_forward_project_release", |
| "default": true, |
| "format": "jsonl", |
| "metadata_path": "benchmarks/metadata.json", |
| "native_input_format": "csv_forward_reaction_rows", |
| "native_source_paths": [ |
| "aux/provenance/datasets/USPTO_MIT/MIT_separated/test.csv" |
| ], |
| "notes": [ |
| "Native forward benchmark split for the public USPTO_MIT product-prediction task.", |
| "The normalized JSONL preserves reactants, reagents, and expected product(s) rather than the backward exact-match schema.", |
| "This is the native USPTO_MIT fine-tuned forward checkpoint benchmark split." |
| ], |
| "num_examples": 40000, |
| "path": "benchmarks/default_test.jsonl", |
| "record_schema": { |
| "inputs": [ |
| "reactants", |
| "reagents" |
| ], |
| "targets": [ |
| "expected_products" |
| ] |
| }, |
| "serialized_input_format": "REACTANT:<reactants>REAGENT:<reagents>", |
| "split": "test", |
| "task_direction": "forward" |
| } |
| ], |
| "checkpoint_id": "sagawa_reactiont5v2_forward_uspto_mit", |
| "embedding_supported": false, |
| "input_schema": { |
| "blank_reagent_value": " ", |
| "reactants_field": "REACTANT", |
| "reagents_field": "REAGENT", |
| "serialized_format": "REACTANT:<reactants>REAGENT:<reagents>", |
| "target_field": "PRODUCT" |
| }, |
| "model_id": "reactiont5v2_forward", |
| "normalized_checkpoint_id": "sagawa_reactiont5v2_forward_uspto_mit", |
| "package_dir": "checkpoints/reactiont5v2_forward/sagawa_reactiont5v2_forward_uspto_mit", |
| "source_registry_refs": [ |
| { |
| "checkpoint_id": "sagawa_reactiont5v2_forward_uspto_mit", |
| "model_id": "reactiont5v2_forward" |
| } |
| ], |
| "task_direction": "forward", |
| "training_dataset_refs": [ |
| "uspto_mit_forward_project_release" |
| ], |
| "weights_path": "model.safetensors", |
| "wrapper_family": "seq2seq_t5" |
| } |
| ], |
| "generated_at": "1970-01-01T00:00:00+00:00" |
| } |
|
|
