| { |
| "train-data-paths": [ |
| "/datasets/evo2/metagenomics/pretraining_data_metagenomics/data_metagenomics_train_text_CharLevelTokenizer_document", |
| "/datasets/evo2/gtdb_imgpr/pretraining_data_gtdb_imgpr/data_gtdb_imgpr_train_text_CharLevelTokenizer_document", |
| "/datasets/evo2/imgvr/pretraining_data_imgvr/data_imgvr_train_text_CharLevelTokenizer_document", |
| "/datasets/evo2/hg38/pretraining_data_hg38/data_hg38_all_text_CharLevelTokenizer_document", |
| "/datasets/evo2/ncrna/pretraining_data_ncrna/data_ncrna_train_text_CharLevelTokenizer_document", |
| "/datasets/evo2/mrna_splice/pretraining_data_mrna_splice/data_mrna_splice_train_text_CharLevelTokenizer_document", |
| "/datasets/evo2/promoters/pretraining_data_promoters/data_promoters_train_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_train_batch1_animalia_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_train_batch2_animalia_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_train_batch3_animalia_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_train_batch4_animalia_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_train_batch5_animalia_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_train_batch6_animalia_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_train_batch7_animalia_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_train_batch8_animalia_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_train_batch1_fungi_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_train_batch2_fungi_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_train_batch3_fungi_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_train_batch4_fungi_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_train_batch5_fungi_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_train_batch6_fungi_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_train_batch7_fungi_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_train_batch8_fungi_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_train_batch1_plantae_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_train_batch2_plantae_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_train_batch3_plantae_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_train_batch4_plantae_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_train_batch5_plantae_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_train_batch6_plantae_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_train_batch7_plantae_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_train_batch8_plantae_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_train_batch1_chromista_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_train_batch2_chromista_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_train_batch3_chromista_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_train_batch4_chromista_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_train_batch5_chromista_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_train_batch6_chromista_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_train_batch7_chromista_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_train_batch8_chromista_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_train_batch1_protista_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_train_batch2_protista_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_train_batch3_protista_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_train_batch4_protista_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_train_batch5_protista_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_train_batch6_protista_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_train_batch7_protista_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_train_batch8_protista_text_CharLevelTokenizer_document", |
|
|
| ], |
|
|
| |
| "valid-data-paths": [ |
| "/datasets/evo2/metagenomics/pretraining_data_metagenomics/data_metagenomics_valid_text_CharLevelTokenizer_document", |
| "/datasets/evo2/gtdb_imgpr/pretraining_data_gtdb_imgpr/data_gtdb_imgpr_valid_text_CharLevelTokenizer_document", |
| "/datasets/evo2/imgvr/pretraining_data_imgvr/data_imgvr_valid_text_CharLevelTokenizer_document", |
| "/datasets/evo2/ncrna/pretraining_data_ncrna/data_ncrna_valid_text_CharLevelTokenizer_document", |
| "/datasets/evo2/mrna_splice/pretraining_data_mrna_splice/data_mrna_splice_valid_text_CharLevelTokenizer_document", |
| "/datasets/evo2/promoters/pretraining_data_promoters/data_promoters_valid_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_valid_batch1_animalia_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_valid_batch2_animalia_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_valid_batch3_animalia_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_valid_batch4_animalia_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_valid_batch5_animalia_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_valid_batch6_animalia_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_valid_batch7_animalia_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_valid_batch8_animalia_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_valid_batch1_fungi_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_valid_batch2_fungi_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_valid_batch3_fungi_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_valid_batch4_fungi_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_valid_batch5_fungi_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_valid_batch6_fungi_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_valid_batch7_fungi_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_valid_batch8_fungi_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_valid_batch1_plantae_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_valid_batch2_plantae_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_valid_batch3_plantae_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_valid_batch4_plantae_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_valid_batch5_plantae_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_valid_batch6_plantae_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_valid_batch7_plantae_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_valid_batch8_plantae_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_valid_batch1_chromista_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_valid_batch2_chromista_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_valid_batch3_chromista_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_valid_batch4_chromista_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_valid_batch5_chromista_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_valid_batch6_chromista_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_valid_batch7_chromista_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_valid_batch8_chromista_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_valid_batch1_protista_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_valid_batch2_protista_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_valid_batch3_protista_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_valid_batch4_protista_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_valid_batch5_protista_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_valid_batch6_protista_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_valid_batch7_protista_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_valid_batch8_protista_text_CharLevelTokenizer_document", |
| ], |
|
|
| |
| "test-data-paths": [ |
| "/datasets/evo2/metagenomics/pretraining_data_metagenomics/data_metagenomics_test_text_CharLevelTokenizer_document", |
| "/datasets/evo2/gtdb_imgpr/pretraining_data_gtdb_imgpr/data_gtdb_imgpr_test_text_CharLevelTokenizer_document", |
| "/datasets/evo2/imgvr/pretraining_data_imgvr/data_imgvr_test_text_CharLevelTokenizer_document", |
| "/datasets/evo2/ncrna/pretraining_data_ncrna/data_ncrna_test_text_CharLevelTokenizer_document", |
| "/datasets/evo2/mrna_splice/pretraining_data_mrna_splice/data_mrna_splice_test_text_CharLevelTokenizer_document", |
| "/datasets/evo2/promoters/pretraining_data_promoters/data_promoters_test_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_test_batch1_animalia_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_test_batch2_animalia_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_test_batch3_animalia_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_test_batch4_animalia_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_test_batch5_animalia_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_test_batch6_animalia_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_test_batch7_animalia_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_test_batch8_animalia_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_test_batch1_fungi_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_test_batch2_fungi_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_test_batch3_fungi_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_test_batch4_fungi_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_test_batch5_fungi_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_test_batch6_fungi_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_test_batch7_fungi_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_test_batch8_fungi_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_test_batch1_plantae_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_test_batch2_plantae_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_test_batch3_plantae_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_test_batch4_plantae_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_test_batch5_plantae_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_test_batch6_plantae_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_test_batch7_plantae_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_test_batch8_plantae_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_test_batch1_chromista_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_test_batch2_chromista_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_test_batch3_chromista_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_test_batch4_chromista_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_test_batch5_chromista_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_test_batch6_chromista_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_test_batch7_chromista_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_test_batch8_chromista_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_test_batch1_protista_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_test_batch2_protista_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_test_batch3_protista_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_test_batch4_protista_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_test_batch5_protista_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_test_batch6_protista_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_test_batch7_protista_text_CharLevelTokenizer_document", |
| "/datasets/evo2/eukaryote_ncbi/pretraining_data/data_test_batch8_protista_text_CharLevelTokenizer_document", |
| ], |
| "train-data-weights": [ |
| 0.09, 0.15, 0.017, 0.003, 0.02, 0.09, 0.0002, 0.05625, 0.05625, |
| 0.05625, 0.05625, 0.05625, 0.05625, 0.05625, 0.05625, 0.01, 0.01, 0.01, 0.01, |
| 0.01, 0.01, 0.01, 0.01, 0.010725, 0.010725, 0.010725, 0.010725, 0.010725, 0.010725, |
| 0.010725, 0.010725, 0.00075, 0.00075, 0.00075, 0.00075, 0.00075, 0.00075, 0.00075, 0.00075, |
| 0.001, 0.001, 0.001, 0.001, 0.001, 0.001, 0.001, 0.001 |
| ], |
| "test-data-weights": [ |
| 0.09, 0.15, 0.017, 0.02, 0.09, 0.0002, 0.05625, 0.05625, 0.05625, 0.05625, |
| 0.05625, 0.05625, 0.05625, 0.05625, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, |
| 0.01, 0.01, 0.010725, 0.010725, 0.010725, 0.010725, 0.010725, 0.010725, 0.010725, |
| 0.010725, 0.00075, 0.00075, 0.00075, 0.00075, 0.00075, 0.00075, 0.00075, 0.00075, |
| 0.001, 0.001, 0.001, 0.001, 0.001, 0.001, 0.001, 0.001 |
| ], |
| "valid-data-weights": [ |
| 0.09, 0.15, 0.017, 0.02, 0.09, 0.0002, 0.05625, 0.05625, 0.05625, |
| 0.05625, 0.05625, 0.05625, 0.05625, 0.05625, 0.01, 0.01, 0.01, 0.01, 0.01, |
| 0.01, 0.01, 0.01, 0.010725, 0.010725, 0.010725, 0.010725, 0.010725, 0.010725, 0.010725, |
| 0.010725, 0.00075, 0.00075, 0.00075, 0.00075, 0.00075, 0.00075, 0.00075, 0.00075, |
| 0.001, 0.001, 0.001, 0.001, 0.001, 0.001, 0.001, 0.001 |
| ], |
| "checkpoint_validation_with_forward_pass": False, |
| |
| "wandb_host": "https://api.wandb.ai", |
| "wandb_project": "evo2" |
| } |