README.md

metadata

dataset_info:
  - config_name: benchmark_chem_tahoe_v1
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: drug
        dtype: string
      - name: concentration
        dtype: float64
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: tissue_type
        dtype: string
      - name: experiment
        dtype: string
      - name: drug_targets
        dtype: string
      - name: moa_broad
        dtype: string
      - name: moa_fine
        dtype: string
      - name: cell_file
        dtype: string
      - name: pert
        dtype: string
      - name: pert_rank
        dtype: int64
      - name: row_rank
        dtype: int64
      - name: n_dir_in_universe
        dtype: int64
      - name: rhaister_zero
        dtype: bool
      - name: rhaister_few
        dtype: bool
    splits:
      - name: train
        num_bytes: 9018886
        num_examples: 23381
      - name: validation
        num_bytes: 9018886
        num_examples: 23381
      - name: test
        num_bytes: 9018886
        num_examples: 23381
    download_size: 2521710
    dataset_size: 27056658
  - config_name: benchmark_chem_tahoe_v2
    features:
      - name: sample_id
        dtype: string
      - name: drug
        dtype: string
      - name: concentration
        dtype: float64
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: tissue_type
        dtype: string
      - name: experiment
        dtype: string
      - name: drug_targets
        dtype: string
      - name: moa_broad
        dtype: string
      - name: moa_fine
        dtype: string
      - name: cell_file
        dtype: string
      - name: pert
        dtype: string
      - name: pert_rank
        dtype: int64
      - name: row_rank
        dtype: int64
      - name: n_dir_in_universe
        dtype: int64
    splits:
      - name: train
        num_bytes: 11217085
        num_examples: 29213
      - name: validation
        num_bytes: 11217085
        num_examples: 29213
      - name: test
        num_bytes: 11217085
        num_examples: 29213
    download_size: 4536342
    dataset_size: 33651255
  - config_name: benchmark_genetic_v1_kaufman
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: gene_target
        dtype: string
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: tissue_type
        dtype: string
      - name: experiment
        dtype: string
      - name: cell_file
        dtype: string
      - name: pert
        dtype: string
      - name: pert_rank
        dtype: int64
      - name: row_rank
        dtype: int64
      - name: n_dir_in_universe
        dtype: int64
      - name: rhaister_zero
        dtype: bool
      - name: rhaister_few
        dtype: bool
    splits:
      - name: test
        num_bytes: 606772
        num_examples: 1665
    download_size: 86189
    dataset_size: 606772
  - config_name: benchmark_genetic_v1_marson
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: gene_target
        dtype: string
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: tissue_type
        dtype: string
      - name: experiment
        dtype: string
      - name: cell_file
        dtype: string
      - name: pert
        dtype: string
      - name: pert_rank
        dtype: int64
      - name: row_rank
        dtype: int64
      - name: n_dir_in_universe
        dtype: int64
      - name: rhaister_zero
        dtype: bool
      - name: rhaister_few
        dtype: bool
    splits:
      - name: test
        num_bytes: 1542186
        num_examples: 4579
    download_size: 202344
    dataset_size: 1542186
  - config_name: benchmark_genetic_v1_nadig
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: gene_target
        dtype: string
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: tissue_type
        dtype: string
      - name: experiment
        dtype: string
      - name: cell_file
        dtype: string
      - name: pert
        dtype: string
      - name: pert_rank
        dtype: int64
      - name: row_rank
        dtype: int64
      - name: n_dir_in_universe
        dtype: int64
      - name: rhaister_zero
        dtype: bool
      - name: rhaister_few
        dtype: bool
    splits:
      - name: test
        num_bytes: 2699736
        num_examples: 8708
    download_size: 322536
    dataset_size: 2699736
  - config_name: benchmark_genetic_v1_xaira_orion
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: gene_target
        dtype: string
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: tissue_type
        dtype: string
      - name: experiment
        dtype: string
      - name: cell_file
        dtype: string
      - name: pert
        dtype: string
      - name: pert_rank
        dtype: int64
      - name: row_rank
        dtype: int64
      - name: n_dir_in_universe
        dtype: int64
      - name: rhaister_zero
        dtype: bool
      - name: rhaister_few
        dtype: bool
    splits:
      - name: test
        num_bytes: 3563784
        num_examples: 9929
    download_size: 386611
    dataset_size: 3563784
  - config_name: benchmark_genetic_v1_xaira_pisces
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: gene_target
        dtype: string
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: tissue_type
        dtype: string
      - name: experiment
        dtype: string
      - name: cell_file
        dtype: string
      - name: pert
        dtype: string
      - name: pert_rank
        dtype: int64
      - name: row_rank
        dtype: int64
      - name: n_dir_in_universe
        dtype: int64
      - name: rhaister_zero
        dtype: bool
      - name: rhaister_few
        dtype: bool
    splits:
      - name: test
        num_bytes: 828600
        num_examples: 2186
    download_size: 108947
    dataset_size: 828600
  - config_name: cleansplit_indist_kaufman
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: gene_target
        dtype: string
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: question
        dtype: string
      - name: reasoning
        dtype: string
      - name: answer
        dtype: string
      - name: cell_file
        dtype: string
    splits:
      - name: train
        num_bytes: 28054820
        num_examples: 2607
      - name: validation
        num_bytes: 28054820
        num_examples: 2607
      - name: test
        num_bytes: 28054820
        num_examples: 2607
    download_size: 31843863
    dataset_size: 84164460
  - config_name: cleansplit_indist_marson
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: gene_target
        dtype: string
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: question
        dtype: string
      - name: reasoning
        dtype: string
      - name: answer
        dtype: string
      - name: cell_file
        dtype: string
    splits:
      - name: train
        num_bytes: 50999519
        num_examples: 8753
      - name: validation
        num_bytes: 50999519
        num_examples: 8753
      - name: test
        num_bytes: 50999519
        num_examples: 8753
    download_size: 57521799
    dataset_size: 152998557
  - config_name: cleansplit_indist_minimal_kaufman
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: gene_target
        dtype: string
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: question
        dtype: string
      - name: reasoning
        dtype: string
      - name: answer
        dtype: string
      - name: cell_file
        dtype: string
    splits:
      - name: train
        num_bytes: 9937392
        num_examples: 2607
      - name: validation
        num_bytes: 9937392
        num_examples: 2607
      - name: test
        num_bytes: 9937392
        num_examples: 2607
    download_size: 10819269
    dataset_size: 29812176
  - config_name: cleansplit_indist_minimal_marson
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: gene_target
        dtype: string
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: question
        dtype: string
      - name: reasoning
        dtype: string
      - name: answer
        dtype: string
      - name: cell_file
        dtype: string
    splits:
      - name: train
        num_bytes: 24666183
        num_examples: 8753
      - name: validation
        num_bytes: 24666183
        num_examples: 8753
      - name: test
        num_bytes: 24666183
        num_examples: 8753
    download_size: 29412189
    dataset_size: 73998549
  - config_name: cleansplit_indist_minimal_nadig
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: gene_target
        dtype: string
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: question
        dtype: string
      - name: reasoning
        dtype: string
      - name: answer
        dtype: string
      - name: cell_file
        dtype: string
    splits:
      - name: train
        num_bytes: 10188066
        num_examples: 3398
      - name: validation
        num_bytes: 10188066
        num_examples: 3398
      - name: test
        num_bytes: 10188066
        num_examples: 3398
    download_size: 12755757
    dataset_size: 30564198
  - config_name: cleansplit_indist_minimal_papalexi
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: gene_target
        dtype: string
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: question
        dtype: string
      - name: reasoning
        dtype: string
      - name: answer
        dtype: string
      - name: cell_file
        dtype: string
    splits:
      - name: train
        num_bytes: 72602
        num_examples: 20
      - name: validation
        num_bytes: 72602
        num_examples: 20
      - name: test
        num_bytes: 72602
        num_examples: 20
    download_size: 83166
    dataset_size: 217806
  - config_name: cleansplit_indist_minimal_xaira_orion
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: gene_target
        dtype: string
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: question
        dtype: string
      - name: reasoning
        dtype: string
      - name: answer
        dtype: string
      - name: cell_file
        dtype: string
    splits:
      - name: train
        num_bytes: 11718383
        num_examples: 4463
      - name: validation
        num_bytes: 11718383
        num_examples: 4463
      - name: test
        num_bytes: 11718383
        num_examples: 4463
    download_size: 13690782
    dataset_size: 35155149
  - config_name: cleansplit_indist_minimal_xaira_pisces
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: gene_target
        dtype: string
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: question
        dtype: string
      - name: reasoning
        dtype: string
      - name: answer
        dtype: string
      - name: cell_file
        dtype: string
    splits:
      - name: train
        num_bytes: 1349865
        num_examples: 504
      - name: validation
        num_bytes: 1349865
        num_examples: 504
      - name: test
        num_bytes: 1349865
        num_examples: 504
    download_size: 1608492
    dataset_size: 4049595
  - config_name: cleansplit_indist_minimal_xu2026
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: gene_target
        dtype: string
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: question
        dtype: string
      - name: reasoning
        dtype: string
      - name: answer
        dtype: string
      - name: cell_file
        dtype: string
    splits:
      - name: train
        num_bytes: 765478
        num_examples: 255
      - name: validation
        num_bytes: 765478
        num_examples: 255
      - name: test
        num_bytes: 765478
        num_examples: 255
    download_size: 789555
    dataset_size: 2296434
  - config_name: cleansplit_indist_nadig
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: gene_target
        dtype: string
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: question
        dtype: string
      - name: reasoning
        dtype: string
      - name: answer
        dtype: string
      - name: cell_file
        dtype: string
    splits:
      - name: train
        num_bytes: 21658527
        num_examples: 3398
      - name: validation
        num_bytes: 21658527
        num_examples: 3398
      - name: test
        num_bytes: 21658527
        num_examples: 3398
    download_size: 25982256
    dataset_size: 64975581
  - config_name: cleansplit_indist_papalexi
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: gene_target
        dtype: string
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: question
        dtype: string
      - name: reasoning
        dtype: string
      - name: answer
        dtype: string
      - name: cell_file
        dtype: string
    splits:
      - name: train
        num_bytes: 223232
        num_examples: 20
      - name: validation
        num_bytes: 223232
        num_examples: 20
      - name: test
        num_bytes: 223232
        num_examples: 20
    download_size: 181656
    dataset_size: 669696
  - config_name: cleansplit_indist_xaira_orion
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: gene_target
        dtype: string
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: question
        dtype: string
      - name: reasoning
        dtype: string
      - name: answer
        dtype: string
      - name: cell_file
        dtype: string
    splits:
      - name: train
        num_bytes: 24136651
        num_examples: 4463
      - name: validation
        num_bytes: 24136651
        num_examples: 4463
      - name: test
        num_bytes: 24136651
        num_examples: 4463
    download_size: 26500515
    dataset_size: 72409953
  - config_name: cleansplit_indist_xaira_pisces
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: gene_target
        dtype: string
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: question
        dtype: string
      - name: reasoning
        dtype: string
      - name: answer
        dtype: string
      - name: cell_file
        dtype: string
    splits:
      - name: train
        num_bytes: 2907052
        num_examples: 504
      - name: validation
        num_bytes: 2907052
        num_examples: 504
      - name: test
        num_bytes: 2907052
        num_examples: 504
    download_size: 3239901
    dataset_size: 8721156
  - config_name: cleansplit_indist_xu2026
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: gene_target
        dtype: string
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: question
        dtype: string
      - name: reasoning
        dtype: string
      - name: answer
        dtype: string
      - name: cell_file
        dtype: string
    splits:
      - name: train
        num_bytes: 1633302
        num_examples: 255
      - name: validation
        num_bytes: 1633302
        num_examples: 255
      - name: test
        num_bytes: 1633302
        num_examples: 255
    download_size: 1681623
    dataset_size: 4899906
  - config_name: grpo_gene_pathway_v1
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: gene_target
        dtype: string
      - name: cell_type
        dtype: string
      - name: cell_file
        dtype: string
      - name: tissue_type
        dtype: string
      - name: experiment
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: pathway_score
        dtype: float64
      - name: pert
        dtype: string
      - name: eval_set
        dtype: string
    splits:
      - name: train
        num_bytes: 8653890
        num_examples: 20000
      - name: test
        num_bytes: 17012579
        num_examples: 48855
    download_size: 2759562
    dataset_size: 25666469
  - config_name: probe_chem_retrieval_v1
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: drug
        dtype: string
      - name: concentration
        dtype: float64
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: tissue_type
        dtype: string
      - name: experiment
        dtype: string
      - name: drug_targets
        dtype: string
      - name: moa_broad
        dtype: string
      - name: moa_fine
        dtype: string
      - name: cell_file
        dtype: string
      - name: pert
        dtype: string
      - name: pert_rank
        dtype: int64
      - name: row_rank
        dtype: int64
      - name: n_dir_in_universe
        dtype: int64
      - name: rhaister_zero
        dtype: bool
      - name: rhaister_few
        dtype: bool
      - name: question
        dtype: string
      - name: reasoning
        dtype: string
      - name: answer
        dtype: string
      - name: eval_set
        dtype: string
    splits:
      - name: train
        num_bytes: 86624741
        num_examples: 4686
      - name: validation
        num_bytes: 86624741
        num_examples: 4686
      - name: test
        num_bytes: 86624741
        num_examples: 4686
    download_size: 60279336
    dataset_size: 259874223
  - config_name: probe_chem_retrieval_v1_shuf
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: drug
        dtype: string
      - name: concentration
        dtype: float64
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: tissue_type
        dtype: string
      - name: experiment
        dtype: string
      - name: drug_targets
        dtype: string
      - name: moa_broad
        dtype: string
      - name: moa_fine
        dtype: string
      - name: cell_file
        dtype: string
      - name: pert
        dtype: string
      - name: pert_rank
        dtype: int64
      - name: row_rank
        dtype: int64
      - name: n_dir_in_universe
        dtype: int64
      - name: rhaister_zero
        dtype: bool
      - name: rhaister_few
        dtype: bool
      - name: question
        dtype: string
      - name: reasoning
        dtype: string
      - name: answer
        dtype: string
      - name: eval_set
        dtype: string
    splits:
      - name: train
        num_bytes: 86624741
        num_examples: 4686
      - name: validation
        num_bytes: 86624741
        num_examples: 4686
      - name: test
        num_bytes: 86624741
        num_examples: 4686
    download_size: 60158241
    dataset_size: 259874223
  - config_name: sft_cell_context_scbasecount_contrastive_id_last_v1
    features:
      - name: mode
        dtype: large_string
      - name: tissue
        dtype: large_string
      - name: tissue_A
        dtype: large_string
      - name: tissue_B
        dtype: large_string
      - name: cell_type
        dtype: large_string
      - name: cell_type_A
        dtype: large_string
      - name: cell_type_B
        dtype: large_string
      - name: question
        dtype: large_string
      - name: reasoning
        dtype: large_string
      - name: answer
        dtype: large_string
      - name: cell_file_A
        dtype: large_string
      - name: cell_file_B
        dtype: large_string
    splits:
      - name: train
        num_bytes: 20850763
        num_examples: 3899
      - name: validation
        num_bytes: 2327454
        num_examples: 435
    download_size: 7401548
    dataset_size: 23178217
  - config_name: sft_cell_context_scbasecount_contrastive_v1
    features:
      - name: mode
        dtype: large_string
      - name: tissue
        dtype: large_string
      - name: tissue_A
        dtype: large_string
      - name: tissue_B
        dtype: large_string
      - name: cell_type
        dtype: large_string
      - name: cell_type_A
        dtype: large_string
      - name: cell_type_B
        dtype: large_string
      - name: question
        dtype: large_string
      - name: reasoning
        dtype: large_string
      - name: answer
        dtype: large_string
      - name: cell_file_A
        dtype: large_string
      - name: cell_file_B
        dtype: large_string
    splits:
      - name: train
        num_bytes: 20768884
        num_examples: 3899
      - name: validation
        num_bytes: 2318319
        num_examples: 435
    download_size: 7413306
    dataset_size: 23087203
  - config_name: sft_cell_context_scbasecount_id_first_v1
    features:
      - name: sample_id
        dtype: large_string
      - name: cell_type
        dtype: large_string
      - name: srx_accession
        dtype: large_string
      - name: question
        dtype: large_string
      - name: reasoning
        dtype: large_string
      - name: answer
        dtype: large_string
      - name: cell_file
        dtype: large_string
    splits:
      - name: train
        num_bytes: 197095968
        num_examples: 80517
      - name: validation
        num_bytes: 24587687
        num_examples: 10055
      - name: test
        num_bytes: 24624785
        num_examples: 10057
      - name: ood_test
        num_bytes: 4614742
        num_examples: 1937
    download_size: 87200902
    dataset_size: 250923182
  - config_name: sft_cell_context_scbasecount_v1
    features:
      - name: sample_id
        dtype: large_string
      - name: cell_type
        dtype: large_string
      - name: srx_accession
        dtype: large_string
      - name: question
        dtype: large_string
      - name: reasoning
        dtype: large_string
      - name: answer
        dtype: large_string
      - name: cell_file
        dtype: large_string
    splits:
      - name: train
        num_bytes: 196129764
        num_examples: 80517
      - name: validation
        num_bytes: 24467027
        num_examples: 10055
      - name: test
        num_bytes: 24504101
        num_examples: 10057
      - name: ood_test
        num_bytes: 4591498
        num_examples: 1937
    download_size: 87447391
    dataset_size: 249692390
  - config_name: sft_chem_pathway_v1_1_10k_full
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: drug
        dtype: string
      - name: concentration
        dtype: float64
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: question
        dtype: string
      - name: reasoning
        dtype: string
      - name: answer
        dtype: string
      - name: cell_file
        dtype: string
    splits:
      - name: train
        num_bytes: 224874793
        num_examples: 8729
      - name: validation
        num_bytes: 25099707
        num_examples: 970
      - name: test
        num_bytes: 25099707
        num_examples: 970
    download_size: 119521849
    dataset_size: 275074207
  - config_name: sft_chem_pathway_v1_1_10k_high_full
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: drug
        dtype: string
      - name: concentration
        dtype: float64
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: question
        dtype: string
      - name: reasoning
        dtype: string
      - name: answer
        dtype: string
      - name: cell_file
        dtype: string
    splits:
      - name: train
        num_bytes: 231178119
        num_examples: 8722
      - name: validation
        num_bytes: 25683073
        num_examples: 966
      - name: test
        num_bytes: 25683073
        num_examples: 966
    download_size: 122707365
    dataset_size: 282544265
  - config_name: sft_chem_pathway_v1_1_10k_mixed_full
    features:
      - name: sample_id
        dtype: string
      - name: teacher
        dtype: string
      - name: dataset
        dtype: string
      - name: drug
        dtype: string
      - name: concentration
        dtype: float64
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: question
        dtype: string
      - name: reasoning
        dtype: string
      - name: answer
        dtype: string
      - name: cell_file
        dtype: string
    splits:
      - name: train
        num_bytes: 456209978
        num_examples: 17451
      - name: validation
        num_bytes: 50800208
        num_examples: 1936
      - name: test
        num_bytes: 25109407
        num_examples: 970
    download_size: 229562828
    dataset_size: 532119593
  - config_name: sft_chem_pathway_v1_2_50k_full
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: drug
        dtype: string
      - name: concentration
        dtype: float64
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: tissue_type
        dtype: string
      - name: experiment
        dtype: string
      - name: drug_targets
        dtype: string
      - name: moa_broad
        dtype: string
      - name: moa_fine
        dtype: string
      - name: cell_file
        dtype: string
      - name: pert
        dtype: string
      - name: pert_rank
        dtype: int64
      - name: row_rank
        dtype: int64
      - name: n_dir_in_universe
        dtype: int64
      - name: rhaister_zero
        dtype: bool
      - name: rhaister_few
        dtype: bool
      - name: question
        dtype: string
      - name: reasoning
        dtype: string
      - name: answer
        dtype: string
      - name: eval_set
        dtype: string
    splits:
      - name: train
        num_bytes: 1118716883
        num_examples: 46298
      - name: validation
        num_bytes: 58756795
        num_examples: 2437
      - name: test
        num_bytes: 416857258
        num_examples: 23381
    download_size: 600357059
    dataset_size: 1594330936
  - config_name: sft_chem_pathway_v1_2_50k_full_answerfirst
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: drug
        dtype: string
      - name: concentration
        dtype: float64
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: question
        dtype: string
      - name: reasoning
        dtype: string
      - name: answer
        dtype: string
      - name: cell_file
        dtype: string
    splits:
      - name: train
        num_bytes: 1118104959
        num_examples: 46298
      - name: validation
        num_bytes: 58724958
        num_examples: 2437
      - name: test
        num_bytes: 58724958
        num_examples: 2437
    download_size: 525730642
    dataset_size: 1235554875
  - config_name: sft_chem_pathway_v1_2_50k_full_labelonly
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: drug
        dtype: string
      - name: concentration
        dtype: float64
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: tissue_type
        dtype: string
      - name: experiment
        dtype: string
      - name: drug_targets
        dtype: string
      - name: moa_broad
        dtype: string
      - name: moa_fine
        dtype: string
      - name: cell_file
        dtype: string
      - name: pert
        dtype: string
      - name: pert_rank
        dtype: int64
      - name: row_rank
        dtype: int64
      - name: n_dir_in_universe
        dtype: int64
      - name: rhaister_zero
        dtype: bool
      - name: rhaister_few
        dtype: bool
      - name: question
        dtype: string
      - name: reasoning
        dtype: string
      - name: answer
        dtype: string
      - name: eval_set
        dtype: string
    splits:
      - name: train
        num_bytes: 825651725
        num_examples: 46298
      - name: validation
        num_bytes: 43339490
        num_examples: 2437
      - name: test
        num_bytes: 416857258
        num_examples: 23381
    download_size: 417234535
    dataset_size: 1285848473
  - config_name: sft_chem_pathway_v1_2_50k_full_labelonly_v2bench
    features:
      - name: sample_id
        dtype: string
      - name: drug
        dtype: string
      - name: concentration
        dtype: float64
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: tissue_type
        dtype: string
      - name: experiment
        dtype: string
      - name: drug_targets
        dtype: string
      - name: moa_broad
        dtype: string
      - name: moa_fine
        dtype: string
      - name: cell_file
        dtype: string
      - name: pert
        dtype: string
      - name: pert_rank
        dtype: int64
      - name: row_rank
        dtype: int64
      - name: n_dir_in_universe
        dtype: int64
      - name: question
        dtype: string
      - name: reasoning
        dtype: string
      - name: answer
        dtype: string
      - name: eval_set
        dtype: string
    splits:
      - name: train
        num_bytes: 825223467
        num_examples: 46298
      - name: validation
        num_bytes: 43316947
        num_examples: 2437
      - name: test
        num_bytes: 530637512
        num_examples: 29213
    download_size: 516796172
    dataset_size: 1399177926
  - config_name: sft_chem_pathway_v1_2_50k_full_v2bench
    features:
      - name: sample_id
        dtype: string
      - name: drug
        dtype: string
      - name: concentration
        dtype: float64
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: tissue_type
        dtype: string
      - name: experiment
        dtype: string
      - name: drug_targets
        dtype: string
      - name: moa_broad
        dtype: string
      - name: moa_fine
        dtype: string
      - name: cell_file
        dtype: string
      - name: pert
        dtype: string
      - name: pert_rank
        dtype: int64
      - name: row_rank
        dtype: int64
      - name: n_dir_in_universe
        dtype: int64
      - name: question
        dtype: string
      - name: reasoning
        dtype: string
      - name: answer
        dtype: string
      - name: eval_set
        dtype: string
    splits:
      - name: train
        num_bytes: 1118288625
        num_examples: 46298
      - name: validation
        num_bytes: 58734252
        num_examples: 2437
      - name: test
        num_bytes: 532799274
        num_examples: 29213
    download_size: 699942157
    dataset_size: 1709822151
  - config_name: sft_gene_pathway_v6_2_1k
    features:
      - name: sample_id
        dtype: large_string
      - name: gene_target
        dtype: large_string
      - name: cell_type
        dtype: large_string
      - name: pathway_name
        dtype: large_string
      - name: pathway_change
        dtype: large_string
      - name: pathway_magnitude
        dtype: large_string
      - name: question
        dtype: large_string
      - name: reasoning
        dtype: large_string
      - name: answer
        dtype: large_string
      - name: cell_file
        dtype: large_string
    splits:
      - name: train
        num_bytes: 9012157
        num_examples: 881
      - name: validation
        num_bytes: 469557
        num_examples: 47
      - name: test
        num_bytes: 469557
        num_examples: 47
    download_size: 4723007
    dataset_size: 9951271
  - config_name: sft_gene_pathway_v7_3_5k
    features:
      - name: sample_id
        dtype: large_string
      - name: gene_target
        dtype: large_string
      - name: cell_type
        dtype: large_string
      - name: pathway_name
        dtype: large_string
      - name: pathway_change
        dtype: large_string
      - name: pathway_magnitude
        dtype: large_string
      - name: question
        dtype: large_string
      - name: reasoning
        dtype: large_string
      - name: answer
        dtype: large_string
      - name: cell_file
        dtype: large_string
    splits:
      - name: train
        num_bytes: 54356168
        num_examples: 5320
      - name: validation
        num_bytes: 6074526
        num_examples: 592
      - name: test
        num_bytes: 6074526
        num_examples: 592
    download_size: 32979010
    dataset_size: 66505220
  - config_name: sft_gene_pathway_v7_3_5k_full
    features:
      - name: sample_id
        dtype: large_string
      - name: gene_target
        dtype: large_string
      - name: cell_type
        dtype: large_string
      - name: pathway_name
        dtype: large_string
      - name: pathway_change
        dtype: large_string
      - name: pathway_magnitude
        dtype: large_string
      - name: question
        dtype: large_string
      - name: reasoning
        dtype: large_string
      - name: answer
        dtype: large_string
      - name: cell_file
        dtype: large_string
    splits:
      - name: train
        num_bytes: 71537283
        num_examples: 5320
      - name: validation
        num_bytes: 8015655
        num_examples: 592
      - name: test
        num_bytes: 8015655
        num_examples: 592
    download_size: 41385153
    dataset_size: 87568593
  - config_name: sft_gene_pathway_v7_5_50k
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: gene_target
        dtype: string
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: question
        dtype: string
      - name: reasoning
        dtype: string
      - name: answer
        dtype: string
      - name: cell_file
        dtype: string
    splits:
      - name: train
        num_bytes: 460379852
        num_examples: 47694
      - name: validation
        num_bytes: 51178178
        num_examples: 5300
      - name: test
        num_bytes: 51178178
        num_examples: 5300
    download_size: 282563907
    dataset_size: 562736208
  - config_name: sft_gene_pathway_v7_5_50k_answeronly
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: gene_target
        dtype: string
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: question
        dtype: string
      - name: reasoning
        dtype: string
      - name: answer
        dtype: string
      - name: cell_file
        dtype: string
    splits:
      - name: train
        num_bytes: 211700183
        num_examples: 47694
      - name: validation
        num_bytes: 23557703
        num_examples: 5300
      - name: test
        num_bytes: 23557703
        num_examples: 5300
    download_size: 114277645
    dataset_size: 258815589
  - config_name: sft_gene_pathway_v7_5_50k_celldesc
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: gene_target
        dtype: string
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: question
        dtype: string
      - name: reasoning
        dtype: string
      - name: answer
        dtype: string
      - name: cell_file
        dtype: string
    splits:
      - name: train
        num_bytes: 581780219
        num_examples: 47694
      - name: validation
        num_bytes: 64668220
        num_examples: 5300
      - name: test
        num_bytes: 64668220
        num_examples: 5300
    download_size: 352430925
    dataset_size: 711116659
  - config_name: sft_gene_pathway_v7_5_50k_full
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: gene_target
        dtype: string
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: question
        dtype: string
      - name: reasoning
        dtype: string
      - name: answer
        dtype: string
      - name: cell_file
        dtype: string
    splits:
      - name: train
        num_bytes: 675647828
        num_examples: 47694
      - name: validation
        num_bytes: 75152138
        num_examples: 5300
      - name: test
        num_bytes: 75152138
        num_examples: 5300
    download_size: 387294484
    dataset_size: 825952104
  - config_name: sft_gene_pathway_v7_5_50k_full_answeronly
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: gene_target
        dtype: string
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: question
        dtype: string
      - name: reasoning
        dtype: string
      - name: answer
        dtype: string
      - name: cell_file
        dtype: string
    splits:
      - name: train
        num_bytes: 426880414
        num_examples: 47694
      - name: validation
        num_bytes: 47523997
        num_examples: 5300
      - name: test
        num_bytes: 47523997
        num_examples: 5300
    download_size: 218116042
    dataset_size: 521928408
  - config_name: sft_gene_pathway_v7_5_50k_full_blinddesc
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: gene_target
        dtype: string
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: question
        dtype: string
      - name: reasoning
        dtype: string
      - name: answer
        dtype: string
      - name: cell_file
        dtype: string
    splits:
      - name: train
        num_bytes: 675103641
        num_examples: 47694
      - name: validation
        num_bytes: 75091680
        num_examples: 5300
      - name: test
        num_bytes: 75091680
        num_examples: 5300
    download_size: 382220592
    dataset_size: 825287001
  - config_name: sft_gene_pathway_v7_5_50k_full_celldesc
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: gene_target
        dtype: string
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: question
        dtype: string
      - name: reasoning
        dtype: string
      - name: answer
        dtype: string
      - name: cell_file
        dtype: string
    splits:
      - name: train
        num_bytes: 796960450
        num_examples: 47694
      - name: validation
        num_bytes: 88634514
        num_examples: 5300
      - name: test
        num_bytes: 88634514
        num_examples: 5300
    download_size: 469503857
    dataset_size: 974229478
  - config_name: sft_gene_pathway_v7_5_50k_full_label
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: gene_target
        dtype: string
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: question
        dtype: string
      - name: reasoning
        dtype: string
      - name: answer
        dtype: string
      - name: cell_file
        dtype: string
    splits:
      - name: train
        num_bytes: 703436166
        num_examples: 47694
      - name: validation
        num_bytes: 78240106
        num_examples: 5300
      - name: test
        num_bytes: 78240106
        num_examples: 5300
    download_size: 388226694
    dataset_size: 859916378
  - config_name: sft_gene_pathway_v7_5_50k_full_labelonly
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: gene_target
        dtype: string
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: question
        dtype: string
      - name: reasoning
        dtype: string
      - name: answer
        dtype: string
      - name: cell_file
        dtype: string
    splits:
      - name: train
        num_bytes: 360125162
        num_examples: 47694
      - name: validation
        num_bytes: 40092157
        num_examples: 5300
      - name: test
        num_bytes: 40092157
        num_examples: 5300
    download_size: 173333220
    dataset_size: 440309476
  - config_name: sft_gene_pathway_v7_5_50k_labelonly
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: gene_target
        dtype: string
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: question
        dtype: string
      - name: reasoning
        dtype: string
      - name: answer
        dtype: string
      - name: cell_file
        dtype: string
    splits:
      - name: train
        num_bytes: 144944931
        num_examples: 47694
      - name: validation
        num_bytes: 16125863
        num_examples: 5300
      - name: test
        num_bytes: 16125863
        num_examples: 5300
    download_size: 69459792
    dataset_size: 177196657
  - config_name: sft_gene_pathway_v7_6_floor_off
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: gene_target
        dtype: string
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: question
        dtype: string
      - name: reasoning
        dtype: string
      - name: answer
        dtype: string
      - name: cell_file
        dtype: string
    splits:
      - name: train
        num_bytes: 558652515
        num_examples: 40174
      - name: validation
        num_bytes: 62022752
        num_examples: 4464
      - name: test
        num_bytes: 62022752
        num_examples: 4464
    download_size: 334337778
    dataset_size: 682698019
  - config_name: sft_gene_pathway_v7_6_floor_on
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: gene_target
        dtype: string
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: question
        dtype: string
      - name: reasoning
        dtype: string
      - name: answer
        dtype: string
      - name: cell_file
        dtype: string
    splits:
      - name: train
        num_bytes: 553926140
        num_examples: 40174
      - name: validation
        num_bytes: 61306591
        num_examples: 4464
      - name: test
        num_bytes: 61306591
        num_examples: 4464
    download_size: 330017586
    dataset_size: 676539322
  - config_name: sft_gene_pathway_v7_6_full100k
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: gene_target
        dtype: string
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: tissue_type
        dtype: string
      - name: experiment
        dtype: string
      - name: cell_file
        dtype: string
      - name: pert
        dtype: string
      - name: pert_rank
        dtype: int64
      - name: row_rank
        dtype: int64
      - name: n_dir_in_universe
        dtype: int64
      - name: rhaister_zero
        dtype: bool
      - name: rhaister_few
        dtype: bool
      - name: question
        dtype: string
      - name: reasoning
        dtype: string
      - name: answer
        dtype: string
      - name: eval_set
        dtype: string
    splits:
      - name: train
        num_bytes: 1305413697
        num_examples: 92664
      - name: validation
        num_bytes: 144801201
        num_examples: 10297
      - name: test
        num_bytes: 309025650
        num_examples: 48855
    download_size: 755627946
    dataset_size: 1759240548
  - config_name: sft_gene_pathway_v7_6_full100k_holdlines
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: gene_target
        dtype: string
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: tissue_type
        dtype: string
      - name: experiment
        dtype: string
      - name: cell_file
        dtype: string
      - name: pert
        dtype: string
      - name: pert_rank
        dtype: int64
      - name: row_rank
        dtype: int64
      - name: n_dir_in_universe
        dtype: int64
      - name: rhaister_zero
        dtype: bool
      - name: rhaister_few
        dtype: bool
      - name: question
        dtype: string
      - name: reasoning
        dtype: string
      - name: answer
        dtype: string
      - name: eval_set
        dtype: string
    splits:
      - name: train
        num_bytes: 1259995186
        num_examples: 88312
      - name: validation
        num_bytes: 66250904
        num_examples: 4648
      - name: test
        num_bytes: 282447366
        num_examples: 45209
    download_size: 696665547
    dataset_size: 1608693456
  - config_name: sft_gene_pathway_v7_6_full100k_holdlines_answerfirst
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: gene_target
        dtype: string
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: tissue_type
        dtype: string
      - name: experiment
        dtype: string
      - name: cell_file
        dtype: string
      - name: pert
        dtype: string
      - name: pert_rank
        dtype: int64
      - name: row_rank
        dtype: int64
      - name: n_dir_in_universe
        dtype: int64
      - name: rhaister_zero
        dtype: bool
      - name: rhaister_few
        dtype: bool
      - name: question
        dtype: string
      - name: reasoning
        dtype: string
      - name: answer
        dtype: string
      - name: eval_set
        dtype: string
    splits:
      - name: train
        num_bytes: 1266065256
        num_examples: 88312
      - name: validation
        num_bytes: 66570300
        num_examples: 4648
      - name: test
        num_bytes: 282447366
        num_examples: 45209
    download_size: 698125277
    dataset_size: 1615082922
  - config_name: sft_gene_pathway_v7_6_full100k_holdlines_labelonly
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: gene_target
        dtype: string
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: tissue_type
        dtype: string
      - name: experiment
        dtype: string
      - name: cell_file
        dtype: string
      - name: pert
        dtype: string
      - name: pert_rank
        dtype: int64
      - name: row_rank
        dtype: int64
      - name: n_dir_in_universe
        dtype: int64
      - name: rhaister_zero
        dtype: bool
      - name: rhaister_few
        dtype: bool
      - name: question
        dtype: string
      - name: reasoning
        dtype: string
      - name: answer
        dtype: string
      - name: eval_set
        dtype: string
    splits:
      - name: train
        num_bytes: 677027373
        num_examples: 88312
      - name: validation
        num_bytes: 35563989
        num_examples: 4648
      - name: test
        num_bytes: 282447366
        num_examples: 45209
    download_size: 334898161
    dataset_size: 995038728
  - config_name: sft_gene_pathway_v7_6_full100k_holdout
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: gene_target
        dtype: string
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: tissue_type
        dtype: string
      - name: experiment
        dtype: string
      - name: cell_file
        dtype: string
      - name: pert
        dtype: string
      - name: pert_rank
        dtype: int64
      - name: row_rank
        dtype: int64
      - name: n_dir_in_universe
        dtype: int64
      - name: rhaister_zero
        dtype: bool
      - name: rhaister_few
        dtype: bool
      - name: question
        dtype: string
      - name: reasoning
        dtype: string
      - name: answer
        dtype: string
      - name: eval_set
        dtype: string
    splits:
      - name: train
        num_bytes: 1191039168
        num_examples: 83624
      - name: validation
        num_bytes: 63108247
        num_examples: 4402
      - name: test
        num_bytes: 309025650
        num_examples: 48855
    download_size: 666070064
    dataset_size: 1563173065
  - config_name: sft_gene_pathway_v7_6_full100k_relabel
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: gene_target
        dtype: string
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: tissue_type
        dtype: string
      - name: experiment
        dtype: string
      - name: cell_file
        dtype: string
      - name: pert
        dtype: string
      - name: pert_rank
        dtype: int64
      - name: row_rank
        dtype: int64
      - name: n_dir_in_universe
        dtype: int64
      - name: rhaister_zero
        dtype: bool
      - name: rhaister_few
        dtype: bool
      - name: question
        dtype: string
      - name: reasoning
        dtype: string
      - name: answer
        dtype: string
      - name: eval_set
        dtype: string
    splits:
      - name: train
        num_bytes: 1304599573
        num_examples: 92661
      - name: validation
        num_bytes: 144567790
        num_examples: 10296
      - name: test
        num_bytes: 309025650
        num_examples: 48855
    download_size: 761502736
    dataset_size: 1758193013
  - config_name: sft_gene_pathway_v7_6_unch_far
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: gene_target
        dtype: string
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: question
        dtype: string
      - name: reasoning
        dtype: string
      - name: answer
        dtype: string
      - name: cell_file
        dtype: string
    splits:
      - name: train
        num_bytes: 478697718
        num_examples: 34994
      - name: validation
        num_bytes: 53156092
        num_examples: 3889
      - name: test
        num_bytes: 53156092
        num_examples: 3889
    download_size: 284051517
    dataset_size: 585009902
  - config_name: sft_gene_pathway_v7_6_unch_near
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: gene_target
        dtype: string
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: question
        dtype: string
      - name: reasoning
        dtype: string
      - name: answer
        dtype: string
      - name: cell_file
        dtype: string
    splits:
      - name: train
        num_bytes: 480346363
        num_examples: 34994
      - name: validation
        num_bytes: 53658080
        num_examples: 3889
      - name: test
        num_bytes: 53658080
        num_examples: 3889
    download_size: 286386391
    dataset_size: 587662523
  - config_name: sft_gene_pathway_v8_10_10k_labelonly
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: gene_target
        dtype: string
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: tissue_type
        dtype: string
      - name: experiment
        dtype: string
      - name: cell_file
        dtype: string
      - name: pert
        dtype: string
      - name: pert_rank
        dtype: int64
      - name: row_rank
        dtype: int64
      - name: n_dir_in_universe
        dtype: int64
      - name: rhaister_zero
        dtype: bool
      - name: rhaister_few
        dtype: bool
      - name: question
        dtype: string
      - name: reasoning
        dtype: string
      - name: answer
        dtype: string
      - name: eval_set
        dtype: string
    splits:
      - name: train
        num_bytes: 77724705
        num_examples: 9500
      - name: validation
        num_bytes: 4086366
        num_examples: 500
      - name: test
        num_bytes: 282849702
        num_examples: 45209
    download_size: 85774801
    dataset_size: 364660773
  - config_name: sft_gene_pathway_v8_10_10k_truezs
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: gene_target
        dtype: string
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: tissue_type
        dtype: string
      - name: experiment
        dtype: string
      - name: cell_file
        dtype: string
      - name: pert
        dtype: string
      - name: question
        dtype: string
      - name: reasoning
        dtype: string
      - name: answer
        dtype: string
      - name: eval_set
        dtype: string
    splits:
      - name: train
        num_bytes: 77494329
        num_examples: 9500
      - name: validation
        num_bytes: 4074240
        num_examples: 500
      - name: test
        num_bytes: 80144667
        num_examples: 8646
    download_size: 47136501
    dataset_size: 161713236
  - config_name: sft_gene_pathway_v8_10_10k_v76traces
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: gene_target
        dtype: string
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: tissue_type
        dtype: string
      - name: experiment
        dtype: string
      - name: cell_file
        dtype: string
      - name: pert
        dtype: string
      - name: pert_rank
        dtype: int64
      - name: row_rank
        dtype: int64
      - name: n_dir_in_universe
        dtype: int64
      - name: rhaister_zero
        dtype: bool
      - name: rhaister_few
        dtype: bool
      - name: question
        dtype: string
      - name: reasoning
        dtype: string
      - name: answer
        dtype: string
      - name: eval_set
        dtype: string
    splits:
      - name: train
        num_bytes: 139572015
        num_examples: 9500
      - name: validation
        num_bytes: 7347273
        num_examples: 500
      - name: test
        num_bytes: 282849702
        num_examples: 45209
    download_size: 123823899
    dataset_size: 429768990
  - config_name: sft_gene_pathway_v8_10_10k_v810_hightraces
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: gene_target
        dtype: string
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: tissue_type
        dtype: string
      - name: experiment
        dtype: string
      - name: cell_file
        dtype: string
      - name: pert
        dtype: string
      - name: pert_rank
        dtype: int64
      - name: row_rank
        dtype: int64
      - name: n_dir_in_universe
        dtype: int64
      - name: rhaister_zero
        dtype: bool
      - name: rhaister_few
        dtype: bool
      - name: question
        dtype: string
      - name: reasoning
        dtype: string
      - name: answer
        dtype: string
      - name: eval_set
        dtype: string
    splits:
      - name: train
        num_bytes: 111330831
        num_examples: 8941
      - name: validation
        num_bytes: 5896941
        num_examples: 472
      - name: test
        num_bytes: 282849702
        num_examples: 45209
    download_size: 106638327
    dataset_size: 400077474
  - config_name: sft_gene_pathway_v8_10_10k_v810_hightraces_answerfirst
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: gene_target
        dtype: string
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: tissue_type
        dtype: string
      - name: experiment
        dtype: string
      - name: cell_file
        dtype: string
      - name: pert
        dtype: string
      - name: pert_rank
        dtype: int64
      - name: row_rank
        dtype: int64
      - name: n_dir_in_universe
        dtype: int64
      - name: rhaister_zero
        dtype: bool
      - name: rhaister_few
        dtype: bool
      - name: question
        dtype: string
      - name: reasoning
        dtype: string
      - name: answer
        dtype: string
      - name: eval_set
        dtype: string
    splits:
      - name: train
        num_bytes: 111950960
        num_examples: 8941
      - name: validation
        num_bytes: 5929683
        num_examples: 472
      - name: test
        num_bytes: 282849702
        num_examples: 45209
    download_size: 107247216
    dataset_size: 400730345
  - config_name: sft_gene_pathway_v8_10_10k_v810_hightraces_stripped
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: gene_target
        dtype: string
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: tissue_type
        dtype: string
      - name: experiment
        dtype: string
      - name: cell_file
        dtype: string
      - name: pert
        dtype: string
      - name: pert_rank
        dtype: int64
      - name: row_rank
        dtype: int64
      - name: n_dir_in_universe
        dtype: int64
      - name: rhaister_zero
        dtype: bool
      - name: rhaister_few
        dtype: bool
      - name: question
        dtype: string
      - name: reasoning
        dtype: string
      - name: answer
        dtype: string
      - name: eval_set
        dtype: string
    splits:
      - name: train
        num_bytes: 101533585
        num_examples: 8941
      - name: validation
        num_bytes: 5383526
        num_examples: 472
      - name: test
        num_bytes: 282849702
        num_examples: 45209
    download_size: 100881463
    dataset_size: 389766813
  - config_name: sft_gene_pathway_v8_10_10k_v810_mediumtraces
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: gene_target
        dtype: string
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: tissue_type
        dtype: string
      - name: experiment
        dtype: string
      - name: cell_file
        dtype: string
      - name: pert
        dtype: string
      - name: pert_rank
        dtype: int64
      - name: row_rank
        dtype: int64
      - name: n_dir_in_universe
        dtype: int64
      - name: rhaister_zero
        dtype: bool
      - name: rhaister_few
        dtype: bool
      - name: question
        dtype: string
      - name: reasoning
        dtype: string
      - name: answer
        dtype: string
      - name: eval_set
        dtype: string
    splits:
      - name: train
        num_bytes: 112170483
        num_examples: 8933
      - name: validation
        num_bytes: 5914703
        num_examples: 470
      - name: test
        num_bytes: 282849702
        num_examples: 45209
    download_size: 107016509
    dataset_size: 400934888
  - config_name: sft_gene_pathway_v9_9_gpt61_high_v7matched92960
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: gene_target
        dtype: string
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: tissue_type
        dtype: string
      - name: experiment
        dtype: string
      - name: cell_file
        dtype: string
      - name: pert
        dtype: string
      - name: pert_rank
        dtype: int64
      - name: row_rank
        dtype: int64
      - name: n_dir_in_universe
        dtype: int64
      - name: rhaister_zero
        dtype: bool
      - name: rhaister_few
        dtype: bool
      - name: question
        dtype: string
      - name: reasoning
        dtype: string
      - name: answer
        dtype: string
      - name: eval_set
        dtype: string
    splits:
      - name: train
        num_bytes: 1074266436
        num_examples: 85008
      - name: validation
        num_bytes: 56783955
        num_examples: 4486
      - name: test
        num_bytes: 282447366
        num_examples: 45209
    download_size: 549570137
    dataset_size: 1413497757
  - config_name: sft_gene_pathway_v9_9_gpt61_high_v7matched92960_celltext
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: gene_target
        dtype: string
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: tissue_type
        dtype: string
      - name: experiment
        dtype: string
      - name: cell_file
        dtype: string
      - name: pert
        dtype: string
      - name: pert_rank
        dtype: int64
      - name: row_rank
        dtype: int64
      - name: n_dir_in_universe
        dtype: int64
      - name: rhaister_zero
        dtype: bool
      - name: rhaister_few
        dtype: bool
      - name: question
        dtype: string
      - name: reasoning
        dtype: string
      - name: answer
        dtype: string
      - name: eval_set
        dtype: string
    splits:
      - name: train
        num_bytes: 1158276130
        num_examples: 85008
      - name: validation
        num_bytes: 61216881
        num_examples: 4486
      - name: test
        num_bytes: 322714317
        num_examples: 45209
    download_size: 604631097
    dataset_size: 1542207328
  - config_name: zeroshot_hepg2_v1
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: gene_target
        dtype: string
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: tissue_type
        dtype: string
      - name: experiment
        dtype: string
      - name: cell_file
        dtype: string
      - name: pert
        dtype: string
    splits:
      - name: test
        num_bytes: 2050156
        num_examples: 6440
    download_size: 248267
    dataset_size: 2050156
  - config_name: zeroshot_hepg2_v1__sft_gene_pathway_v7_6_100k
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: gene_target
        dtype: string
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: tissue_type
        dtype: string
      - name: experiment
        dtype: string
      - name: cell_file
        dtype: string
      - name: pert
        dtype: string
      - name: question
        dtype: string
      - name: reasoning
        dtype: string
      - name: answer
        dtype: string
    splits:
      - name: train
        num_bytes: 37308530
        num_examples: 6440
      - name: validation
        num_bytes: 37308530
        num_examples: 6440
      - name: test
        num_bytes: 37308530
        num_examples: 6440
    download_size: 20912697
    dataset_size: 111925590
  - config_name: zeroshot_ipsc_v1
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: gene_target
        dtype: string
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: tissue_type
        dtype: string
      - name: experiment
        dtype: string
      - name: cell_file
        dtype: string
      - name: pert
        dtype: string
    splits:
      - name: test
        num_bytes: 2348016
        num_examples: 7560
    download_size: 356743
    dataset_size: 2348016
  - config_name: zeroshot_ipsc_v1__sft_gene_pathway_v7_6_100k
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: gene_target
        dtype: string
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: tissue_type
        dtype: string
      - name: experiment
        dtype: string
      - name: cell_file
        dtype: string
      - name: pert
        dtype: string
      - name: question
        dtype: string
      - name: reasoning
        dtype: string
      - name: answer
        dtype: string
    splits:
      - name: train
        num_bytes: 44377597
        num_examples: 7560
      - name: validation
        num_bytes: 44377597
        num_examples: 7560
      - name: test
        num_bytes: 44377597
        num_examples: 7560
    download_size: 24622137
    dataset_size: 133132791
  - config_name: zeroshot_norman_v1
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: gene_target
        dtype: string
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: tissue_type
        dtype: string
      - name: experiment
        dtype: string
      - name: cell_file
        dtype: string
      - name: pert
        dtype: string
    splits:
      - name: test
        num_bytes: 1123497
        num_examples: 4142
    download_size: 162310
    dataset_size: 1123497
  - config_name: zeroshot_norman_v1__sft_gene_pathway_v7_6_100k
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: gene_target
        dtype: string
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: tissue_type
        dtype: string
      - name: experiment
        dtype: string
      - name: cell_file
        dtype: string
      - name: pert
        dtype: string
      - name: question
        dtype: string
      - name: reasoning
        dtype: string
      - name: answer
        dtype: string
    splits:
      - name: train
        num_bytes: 28412274
        num_examples: 4142
      - name: validation
        num_bytes: 28412274
        num_examples: 4142
      - name: test
        num_bytes: 28412274
        num_examples: 4142
    download_size: 16070286
    dataset_size: 85236822
  - config_name: zeroshot_papalexi_v1
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: gene_target
        dtype: string
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: tissue_type
        dtype: string
      - name: experiment
        dtype: string
      - name: cell_file
        dtype: string
      - name: pert
        dtype: string
    splits:
      - name: test
        num_bytes: 1448990
        num_examples: 5000
    download_size: 191657
    dataset_size: 1448990
  - config_name: zeroshot_xu2026_v1
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: gene_target
        dtype: string
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: tissue_type
        dtype: string
      - name: experiment
        dtype: string
      - name: cell_file
        dtype: string
      - name: pert
        dtype: string
    splits:
      - name: test
        num_bytes: 1156696
        num_examples: 3646
    download_size: 155626
    dataset_size: 1156696
  - config_name: zeroshot_xu2026_v1__sft_gene_pathway_v7_6_100k
    features:
      - name: sample_id
        dtype: string
      - name: dataset
        dtype: string
      - name: gene_target
        dtype: string
      - name: cell_type
        dtype: string
      - name: pathway_name
        dtype: string
      - name: pathway_change
        dtype: string
      - name: tissue_type
        dtype: string
      - name: experiment
        dtype: string
      - name: cell_file
        dtype: string
      - name: pert
        dtype: string
      - name: question
        dtype: string
      - name: reasoning
        dtype: string
      - name: answer
        dtype: string
    splits:
      - name: train
        num_bytes: 26409658
        num_examples: 3646
      - name: validation
        num_bytes: 26409658
        num_examples: 3646
      - name: test
        num_bytes: 26409658
        num_examples: 3646
    download_size: 15374448
    dataset_size: 79228974
configs:
  - config_name: benchmark_chem_tahoe_v1
    data_files:
      - split: train
        path: benchmark_chem_tahoe_v1/train-*
      - split: validation
        path: benchmark_chem_tahoe_v1/validation-*
      - split: test
        path: benchmark_chem_tahoe_v1/test-*
  - config_name: benchmark_chem_tahoe_v2
    data_files:
      - split: train
        path: benchmark_chem_tahoe_v2/train-*
      - split: validation
        path: benchmark_chem_tahoe_v2/validation-*
      - split: test
        path: benchmark_chem_tahoe_v2/test-*
  - config_name: benchmark_genetic_v1_kaufman
    data_files:
      - split: test
        path: benchmark_genetic_v1_kaufman/test-*
  - config_name: benchmark_genetic_v1_marson
    data_files:
      - split: test
        path: benchmark_genetic_v1_marson/test-*
  - config_name: benchmark_genetic_v1_nadig
    data_files:
      - split: test
        path: benchmark_genetic_v1_nadig/test-*
  - config_name: benchmark_genetic_v1_xaira_orion
    data_files:
      - split: test
        path: benchmark_genetic_v1_xaira_orion/test-*
  - config_name: benchmark_genetic_v1_xaira_pisces
    data_files:
      - split: test
        path: benchmark_genetic_v1_xaira_pisces/test-*
  - config_name: cleansplit_indist_kaufman
    data_files:
      - split: train
        path: cleansplit_indist_kaufman/train-*
      - split: validation
        path: cleansplit_indist_kaufman/validation-*
      - split: test
        path: cleansplit_indist_kaufman/test-*
  - config_name: cleansplit_indist_marson
    data_files:
      - split: train
        path: cleansplit_indist_marson/train-*
      - split: validation
        path: cleansplit_indist_marson/validation-*
      - split: test
        path: cleansplit_indist_marson/test-*
  - config_name: cleansplit_indist_minimal_kaufman
    data_files:
      - split: train
        path: cleansplit_indist_minimal_kaufman/train-*
      - split: validation
        path: cleansplit_indist_minimal_kaufman/validation-*
      - split: test
        path: cleansplit_indist_minimal_kaufman/test-*
  - config_name: cleansplit_indist_minimal_marson
    data_files:
      - split: train
        path: cleansplit_indist_minimal_marson/train-*
      - split: validation
        path: cleansplit_indist_minimal_marson/validation-*
      - split: test
        path: cleansplit_indist_minimal_marson/test-*
  - config_name: cleansplit_indist_minimal_nadig
    data_files:
      - split: train
        path: cleansplit_indist_minimal_nadig/train-*
      - split: validation
        path: cleansplit_indist_minimal_nadig/validation-*
      - split: test
        path: cleansplit_indist_minimal_nadig/test-*
  - config_name: cleansplit_indist_minimal_papalexi
    data_files:
      - split: train
        path: cleansplit_indist_minimal_papalexi/train-*
      - split: validation
        path: cleansplit_indist_minimal_papalexi/validation-*
      - split: test
        path: cleansplit_indist_minimal_papalexi/test-*
  - config_name: cleansplit_indist_minimal_xaira_orion
    data_files:
      - split: train
        path: cleansplit_indist_minimal_xaira_orion/train-*
      - split: validation
        path: cleansplit_indist_minimal_xaira_orion/validation-*
      - split: test
        path: cleansplit_indist_minimal_xaira_orion/test-*
  - config_name: cleansplit_indist_minimal_xaira_pisces
    data_files:
      - split: train
        path: cleansplit_indist_minimal_xaira_pisces/train-*
      - split: validation
        path: cleansplit_indist_minimal_xaira_pisces/validation-*
      - split: test
        path: cleansplit_indist_minimal_xaira_pisces/test-*
  - config_name: cleansplit_indist_minimal_xu2026
    data_files:
      - split: train
        path: cleansplit_indist_minimal_xu2026/train-*
      - split: validation
        path: cleansplit_indist_minimal_xu2026/validation-*
      - split: test
        path: cleansplit_indist_minimal_xu2026/test-*
  - config_name: cleansplit_indist_nadig
    data_files:
      - split: train
        path: cleansplit_indist_nadig/train-*
      - split: validation
        path: cleansplit_indist_nadig/validation-*
      - split: test
        path: cleansplit_indist_nadig/test-*
  - config_name: cleansplit_indist_papalexi
    data_files:
      - split: train
        path: cleansplit_indist_papalexi/train-*
      - split: validation
        path: cleansplit_indist_papalexi/validation-*
      - split: test
        path: cleansplit_indist_papalexi/test-*
  - config_name: cleansplit_indist_xaira_orion
    data_files:
      - split: train
        path: cleansplit_indist_xaira_orion/train-*
      - split: validation
        path: cleansplit_indist_xaira_orion/validation-*
      - split: test
        path: cleansplit_indist_xaira_orion/test-*
  - config_name: cleansplit_indist_xaira_pisces
    data_files:
      - split: train
        path: cleansplit_indist_xaira_pisces/train-*
      - split: validation
        path: cleansplit_indist_xaira_pisces/validation-*
      - split: test
        path: cleansplit_indist_xaira_pisces/test-*
  - config_name: cleansplit_indist_xu2026
    data_files:
      - split: train
        path: cleansplit_indist_xu2026/train-*
      - split: validation
        path: cleansplit_indist_xu2026/validation-*
      - split: test
        path: cleansplit_indist_xu2026/test-*
  - config_name: grpo_gene_pathway_v1
    data_files:
      - split: train
        path: grpo_gene_pathway_v1/train-*
      - split: test
        path: grpo_gene_pathway_v1/test-*
  - config_name: probe_chem_retrieval_v1
    data_files:
      - split: train
        path: probe_chem_retrieval_v1/train-*
      - split: validation
        path: probe_chem_retrieval_v1/validation-*
      - split: test
        path: probe_chem_retrieval_v1/test-*
  - config_name: probe_chem_retrieval_v1_shuf
    data_files:
      - split: train
        path: probe_chem_retrieval_v1_shuf/train-*
      - split: validation
        path: probe_chem_retrieval_v1_shuf/validation-*
      - split: test
        path: probe_chem_retrieval_v1_shuf/test-*
  - config_name: sft_cell_context_scbasecount_contrastive_id_last_v1
    data_files:
      - split: train
        path: sft_cell_context_scbasecount_contrastive_id_last_v1/train-*
      - split: validation
        path: sft_cell_context_scbasecount_contrastive_id_last_v1/validation-*
  - config_name: sft_cell_context_scbasecount_contrastive_v1
    data_files:
      - split: train
        path: sft_cell_context_scbasecount_contrastive_v1/train-*
      - split: validation
        path: sft_cell_context_scbasecount_contrastive_v1/validation-*
  - config_name: sft_cell_context_scbasecount_id_first_v1
    data_files:
      - split: train
        path: sft_cell_context_scbasecount_id_first_v1/train-*
      - split: validation
        path: sft_cell_context_scbasecount_id_first_v1/validation-*
      - split: test
        path: sft_cell_context_scbasecount_id_first_v1/test-*
      - split: ood_test
        path: sft_cell_context_scbasecount_id_first_v1/ood_test-*
  - config_name: sft_cell_context_scbasecount_v1
    data_files:
      - split: train
        path: sft_cell_context_scbasecount_v1/train-*
      - split: validation
        path: sft_cell_context_scbasecount_v1/validation-*
      - split: test
        path: sft_cell_context_scbasecount_v1/test-*
      - split: ood_test
        path: sft_cell_context_scbasecount_v1/ood_test-*
  - config_name: sft_chem_pathway_v1_1_10k_full
    data_files:
      - split: train
        path: sft_chem_pathway_v1_1_10k_full/train-*
      - split: validation
        path: sft_chem_pathway_v1_1_10k_full/validation-*
      - split: test
        path: sft_chem_pathway_v1_1_10k_full/test-*
  - config_name: sft_chem_pathway_v1_1_10k_high_full
    data_files:
      - split: train
        path: sft_chem_pathway_v1_1_10k_high_full/train-*
      - split: validation
        path: sft_chem_pathway_v1_1_10k_high_full/validation-*
      - split: test
        path: sft_chem_pathway_v1_1_10k_high_full/test-*
  - config_name: sft_chem_pathway_v1_1_10k_mixed_full
    data_files:
      - split: train
        path: sft_chem_pathway_v1_1_10k_mixed_full/train-*
      - split: validation
        path: sft_chem_pathway_v1_1_10k_mixed_full/validation-*
      - split: test
        path: sft_chem_pathway_v1_1_10k_mixed_full/test-*
  - config_name: sft_chem_pathway_v1_2_50k_full
    data_files:
      - split: train
        path: sft_chem_pathway_v1_2_50k_full/train-*
      - split: validation
        path: sft_chem_pathway_v1_2_50k_full/validation-*
      - split: test
        path: sft_chem_pathway_v1_2_50k_full/test-*
  - config_name: sft_chem_pathway_v1_2_50k_full_answerfirst
    data_files:
      - split: train
        path: sft_chem_pathway_v1_2_50k_full_answerfirst/train-*
      - split: validation
        path: sft_chem_pathway_v1_2_50k_full_answerfirst/validation-*
      - split: test
        path: sft_chem_pathway_v1_2_50k_full_answerfirst/test-*
  - config_name: sft_chem_pathway_v1_2_50k_full_labelonly
    data_files:
      - split: train
        path: sft_chem_pathway_v1_2_50k_full_labelonly/train-*
      - split: validation
        path: sft_chem_pathway_v1_2_50k_full_labelonly/validation-*
      - split: test
        path: sft_chem_pathway_v1_2_50k_full_labelonly/test-*
  - config_name: sft_chem_pathway_v1_2_50k_full_labelonly_v2bench
    data_files:
      - split: train
        path: sft_chem_pathway_v1_2_50k_full_labelonly_v2bench/train-*
      - split: validation
        path: sft_chem_pathway_v1_2_50k_full_labelonly_v2bench/validation-*
      - split: test
        path: sft_chem_pathway_v1_2_50k_full_labelonly_v2bench/test-*
  - config_name: sft_chem_pathway_v1_2_50k_full_v2bench
    data_files:
      - split: train
        path: sft_chem_pathway_v1_2_50k_full_v2bench/train-*
      - split: validation
        path: sft_chem_pathway_v1_2_50k_full_v2bench/validation-*
      - split: test
        path: sft_chem_pathway_v1_2_50k_full_v2bench/test-*
  - config_name: sft_gene_pathway_v6_2_1k
    data_files:
      - split: train
        path: sft_gene_pathway_v6_2_1k/train-*
      - split: validation
        path: sft_gene_pathway_v6_2_1k/validation-*
      - split: test
        path: sft_gene_pathway_v6_2_1k/test-*
  - config_name: sft_gene_pathway_v7_3_5k
    data_files:
      - split: train
        path: sft_gene_pathway_v7_3_5k/train-*
      - split: validation
        path: sft_gene_pathway_v7_3_5k/validation-*
      - split: test
        path: sft_gene_pathway_v7_3_5k/test-*
  - config_name: sft_gene_pathway_v7_3_5k_full
    data_files:
      - split: train
        path: sft_gene_pathway_v7_3_5k_full/train-*
      - split: validation
        path: sft_gene_pathway_v7_3_5k_full/validation-*
      - split: test
        path: sft_gene_pathway_v7_3_5k_full/test-*
  - config_name: sft_gene_pathway_v7_5_50k
    data_files:
      - split: train
        path: sft_gene_pathway_v7_5_50k/train-*
      - split: validation
        path: sft_gene_pathway_v7_5_50k/validation-*
      - split: test
        path: sft_gene_pathway_v7_5_50k/test-*
  - config_name: sft_gene_pathway_v7_5_50k_answeronly
    data_files:
      - split: train
        path: sft_gene_pathway_v7_5_50k_answeronly/train-*
      - split: validation
        path: sft_gene_pathway_v7_5_50k_answeronly/validation-*
      - split: test
        path: sft_gene_pathway_v7_5_50k_answeronly/test-*
  - config_name: sft_gene_pathway_v7_5_50k_celldesc
    data_files:
      - split: train
        path: sft_gene_pathway_v7_5_50k_celldesc/train-*
      - split: validation
        path: sft_gene_pathway_v7_5_50k_celldesc/validation-*
      - split: test
        path: sft_gene_pathway_v7_5_50k_celldesc/test-*
  - config_name: sft_gene_pathway_v7_5_50k_full
    data_files:
      - split: train
        path: sft_gene_pathway_v7_5_50k_full/train-*
      - split: validation
        path: sft_gene_pathway_v7_5_50k_full/validation-*
      - split: test
        path: sft_gene_pathway_v7_5_50k_full/test-*
  - config_name: sft_gene_pathway_v7_5_50k_full_answeronly
    data_files:
      - split: train
        path: sft_gene_pathway_v7_5_50k_full_answeronly/train-*
      - split: validation
        path: sft_gene_pathway_v7_5_50k_full_answeronly/validation-*
      - split: test
        path: sft_gene_pathway_v7_5_50k_full_answeronly/test-*
  - config_name: sft_gene_pathway_v7_5_50k_full_blinddesc
    data_files:
      - split: train
        path: sft_gene_pathway_v7_5_50k_full_blinddesc/train-*
      - split: validation
        path: sft_gene_pathway_v7_5_50k_full_blinddesc/validation-*
      - split: test
        path: sft_gene_pathway_v7_5_50k_full_blinddesc/test-*
  - config_name: sft_gene_pathway_v7_5_50k_full_celldesc
    data_files:
      - split: train
        path: sft_gene_pathway_v7_5_50k_full_celldesc/train-*
      - split: validation
        path: sft_gene_pathway_v7_5_50k_full_celldesc/validation-*
      - split: test
        path: sft_gene_pathway_v7_5_50k_full_celldesc/test-*
  - config_name: sft_gene_pathway_v7_5_50k_full_label
    data_files:
      - split: train
        path: sft_gene_pathway_v7_5_50k_full_label/train-*
      - split: validation
        path: sft_gene_pathway_v7_5_50k_full_label/validation-*
      - split: test
        path: sft_gene_pathway_v7_5_50k_full_label/test-*
  - config_name: sft_gene_pathway_v7_5_50k_full_labelonly
    data_files:
      - split: train
        path: sft_gene_pathway_v7_5_50k_full_labelonly/train-*
      - split: validation
        path: sft_gene_pathway_v7_5_50k_full_labelonly/validation-*
      - split: test
        path: sft_gene_pathway_v7_5_50k_full_labelonly/test-*
  - config_name: sft_gene_pathway_v7_5_50k_labelonly
    data_files:
      - split: train
        path: sft_gene_pathway_v7_5_50k_labelonly/train-*
      - split: validation
        path: sft_gene_pathway_v7_5_50k_labelonly/validation-*
      - split: test
        path: sft_gene_pathway_v7_5_50k_labelonly/test-*
  - config_name: sft_gene_pathway_v7_6_floor_off
    data_files:
      - split: train
        path: sft_gene_pathway_v7_6_floor_off/train-*
      - split: validation
        path: sft_gene_pathway_v7_6_floor_off/validation-*
      - split: test
        path: sft_gene_pathway_v7_6_floor_off/test-*
  - config_name: sft_gene_pathway_v7_6_floor_on
    data_files:
      - split: train
        path: sft_gene_pathway_v7_6_floor_on/train-*
      - split: validation
        path: sft_gene_pathway_v7_6_floor_on/validation-*
      - split: test
        path: sft_gene_pathway_v7_6_floor_on/test-*
  - config_name: sft_gene_pathway_v7_6_full100k
    data_files:
      - split: train
        path: sft_gene_pathway_v7_6_full100k/train-*
      - split: validation
        path: sft_gene_pathway_v7_6_full100k/validation-*
      - split: test
        path: sft_gene_pathway_v7_6_full100k/test-*
  - config_name: sft_gene_pathway_v7_6_full100k_holdlines
    data_files:
      - split: train
        path: sft_gene_pathway_v7_6_full100k_holdlines/train-*
      - split: validation
        path: sft_gene_pathway_v7_6_full100k_holdlines/validation-*
      - split: test
        path: sft_gene_pathway_v7_6_full100k_holdlines/test-*
  - config_name: sft_gene_pathway_v7_6_full100k_holdlines_answerfirst
    data_files:
      - split: train
        path: sft_gene_pathway_v7_6_full100k_holdlines_answerfirst/train-*
      - split: validation
        path: sft_gene_pathway_v7_6_full100k_holdlines_answerfirst/validation-*
      - split: test
        path: sft_gene_pathway_v7_6_full100k_holdlines_answerfirst/test-*
  - config_name: sft_gene_pathway_v7_6_full100k_holdlines_labelonly
    data_files:
      - split: train
        path: sft_gene_pathway_v7_6_full100k_holdlines_labelonly/train-*
      - split: validation
        path: sft_gene_pathway_v7_6_full100k_holdlines_labelonly/validation-*
      - split: test
        path: sft_gene_pathway_v7_6_full100k_holdlines_labelonly/test-*
  - config_name: sft_gene_pathway_v7_6_full100k_holdout
    data_files:
      - split: train
        path: sft_gene_pathway_v7_6_full100k_holdout/train-*
      - split: validation
        path: sft_gene_pathway_v7_6_full100k_holdout/validation-*
      - split: test
        path: sft_gene_pathway_v7_6_full100k_holdout/test-*
  - config_name: sft_gene_pathway_v7_6_full100k_relabel
    data_files:
      - split: train
        path: sft_gene_pathway_v7_6_full100k_relabel/train-*
      - split: validation
        path: sft_gene_pathway_v7_6_full100k_relabel/validation-*
      - split: test
        path: sft_gene_pathway_v7_6_full100k_relabel/test-*
  - config_name: sft_gene_pathway_v7_6_unch_far
    data_files:
      - split: train
        path: sft_gene_pathway_v7_6_unch_far/train-*
      - split: validation
        path: sft_gene_pathway_v7_6_unch_far/validation-*
      - split: test
        path: sft_gene_pathway_v7_6_unch_far/test-*
  - config_name: sft_gene_pathway_v7_6_unch_near
    data_files:
      - split: train
        path: sft_gene_pathway_v7_6_unch_near/train-*
      - split: validation
        path: sft_gene_pathway_v7_6_unch_near/validation-*
      - split: test
        path: sft_gene_pathway_v7_6_unch_near/test-*
  - config_name: sft_gene_pathway_v8_10_10k_labelonly
    data_files:
      - split: train
        path: sft_gene_pathway_v8_10_10k_labelonly/train-*
      - split: validation
        path: sft_gene_pathway_v8_10_10k_labelonly/validation-*
      - split: test
        path: sft_gene_pathway_v8_10_10k_labelonly/test-*
  - config_name: sft_gene_pathway_v8_10_10k_truezs
    data_files:
      - split: train
        path: sft_gene_pathway_v8_10_10k_truezs/train-*
      - split: validation
        path: sft_gene_pathway_v8_10_10k_truezs/validation-*
      - split: test
        path: sft_gene_pathway_v8_10_10k_truezs/test-*
  - config_name: sft_gene_pathway_v8_10_10k_v76traces
    data_files:
      - split: train
        path: sft_gene_pathway_v8_10_10k_v76traces/train-*
      - split: validation
        path: sft_gene_pathway_v8_10_10k_v76traces/validation-*
      - split: test
        path: sft_gene_pathway_v8_10_10k_v76traces/test-*
  - config_name: sft_gene_pathway_v8_10_10k_v810_hightraces
    data_files:
      - split: train
        path: sft_gene_pathway_v8_10_10k_v810_hightraces/train-*
      - split: validation
        path: sft_gene_pathway_v8_10_10k_v810_hightraces/validation-*
      - split: test
        path: sft_gene_pathway_v8_10_10k_v810_hightraces/test-*
  - config_name: sft_gene_pathway_v8_10_10k_v810_hightraces_answerfirst
    data_files:
      - split: train
        path: sft_gene_pathway_v8_10_10k_v810_hightraces_answerfirst/train-*
      - split: validation
        path: sft_gene_pathway_v8_10_10k_v810_hightraces_answerfirst/validation-*
      - split: test
        path: sft_gene_pathway_v8_10_10k_v810_hightraces_answerfirst/test-*
  - config_name: sft_gene_pathway_v8_10_10k_v810_hightraces_stripped
    data_files:
      - split: train
        path: sft_gene_pathway_v8_10_10k_v810_hightraces_stripped/train-*
      - split: validation
        path: sft_gene_pathway_v8_10_10k_v810_hightraces_stripped/validation-*
      - split: test
        path: sft_gene_pathway_v8_10_10k_v810_hightraces_stripped/test-*
  - config_name: sft_gene_pathway_v8_10_10k_v810_mediumtraces
    data_files:
      - split: train
        path: sft_gene_pathway_v8_10_10k_v810_mediumtraces/train-*
      - split: validation
        path: sft_gene_pathway_v8_10_10k_v810_mediumtraces/validation-*
      - split: test
        path: sft_gene_pathway_v8_10_10k_v810_mediumtraces/test-*
  - config_name: sft_gene_pathway_v9_9_gpt61_high_v7matched92960
    data_files:
      - split: train
        path: sft_gene_pathway_v9_9_gpt61_high_v7matched92960/train-*
      - split: validation
        path: sft_gene_pathway_v9_9_gpt61_high_v7matched92960/validation-*
      - split: test
        path: sft_gene_pathway_v9_9_gpt61_high_v7matched92960/test-*
  - config_name: sft_gene_pathway_v9_9_gpt61_high_v7matched92960_celltext
    data_files:
      - split: train
        path: sft_gene_pathway_v9_9_gpt61_high_v7matched92960_celltext/train-*
      - split: validation
        path: sft_gene_pathway_v9_9_gpt61_high_v7matched92960_celltext/validation-*
      - split: test
        path: sft_gene_pathway_v9_9_gpt61_high_v7matched92960_celltext/test-*
  - config_name: zeroshot_hepg2_v1
    data_files:
      - split: test
        path: zeroshot_hepg2_v1/test-*
  - config_name: zeroshot_ipsc_v1
    data_files:
      - split: test
        path: zeroshot_ipsc_v1/test-*
  - config_name: zeroshot_norman_v1
    data_files:
      - split: test
        path: zeroshot_norman_v1/test-*
  - config_name: zeroshot_papalexi_v1
    data_files:
      - split: test
        path: zeroshot_papalexi_v1/test-*
  - config_name: zeroshot_xu2026_v1
    data_files:
      - split: test
        path: zeroshot_xu2026_v1/test-*