Publish context-conditioned molecule transfer v10_2 Bioavailability_Ma
Browse files- README.md +18 -0
- calibration.json +0 -0
- manifest.json +185 -0
- test_ranking/data.parquet +3 -0
- train/data.parquet +3 -0
- validation_ranking/data.parquet +3 -0
README.md
ADDED
|
@@ -0,0 +1,18 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
configs:
|
| 3 |
+
- config_name: default
|
| 4 |
+
data_files:
|
| 5 |
+
- split: train
|
| 6 |
+
path: train/data.parquet
|
| 7 |
+
- split: validation_ranking
|
| 8 |
+
path: validation_ranking/data.parquet
|
| 9 |
+
- split: test_ranking
|
| 10 |
+
path: test_ranking/data.parquet
|
| 11 |
+
---
|
| 12 |
+
# Bioavailability_Ma context-conditioned molecule transfer V10.2
|
| 13 |
+
|
| 14 |
+
This release preserves its direct panels and appends training-only, individual-record continuous assay-evidence transfer pairs. Query values remain hidden from prompts.
|
| 15 |
+
|
| 16 |
+
- Train rows: 93,984
|
| 17 |
+
- Validation rows: 22,682
|
| 18 |
+
- Test rows: 23,369
|
calibration.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
manifest.json
ADDED
|
@@ -0,0 +1,185 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"arrow_schema": "prompt: string\ncompletion: string\nsplit: string\npair_id: string\nranking_query_id: string\nranking_member_index: int64\nranking_candidate_row_count: int64\nranking_candidate_parent_count: int64\nretrieval_parent_rank: int64\nretrieval_parent_context_index: int64\nretrieval_parent_context_count: int64\ntask_id: string\nsource_id: string\nretrieval_strategy: string\nquery_record_id: string\nretrieval_record_id: string\nquery_benchmark_row_id: string\nretrieval_benchmark_row_id: string\nquery_molecule_identity_key: string\nretrieval_molecule_identity_key: string\nquery_smiles: string\nretrieval_smiles: string\nquery_condition_group: string\nretrieval_condition_group: string\nquery_condition_scope: string\nretrieval_condition_scope: string\nquery_condition_atoms: list<element: string>\n child 0, element: string\nretrieval_condition_atoms: list<element: string>\n child 0, element: string\ncandidate_context_relation: string\nquery_source_record_count: int64\nretrieval_source_record_count: int64\nquery_agreement_fraction: double\nretrieval_agreement_fraction: double\nquery_gold_Y: int64\nretrieval_gold_Y: int64\nranking_balance_label: string\nquery_geometry_value: double\nretrieval_geometry_value: double\nabsolute_geometry_difference: double\ncalibration_sample_standard_deviation: double\ntarget_distribution: struct<nontransfer: double, transfer: double>\n child 0, nontransfer: double\n child 1, transfer: double\ncanonical_endpoint_key: string\nassay_concept: string\nmeasurement_kind: string\nmetadata: struct<calibration_sha256: string, groups: struct<assay_concept: string, measurement_kind: string, q (... 339 chars omitted)\n child 0, calibration_sha256: string\n child 1, groups: struct<assay_concept: string, measurement_kind: string, query_condition_group: string, query_conditi (... 190 chars omitted)\n child 0, assay_concept: string\n child 1, measurement_kind: string\n child 2, query_condition_group: string\n child 3, query_condition_scope: string\n child 4, retrieval_strategy: string\n child 5, source_id: string\n child 6, task_id: string\n child 7, evaluation_regime: string\n child 8, evaluation_regime_source: string\n child 9, level: string\n child 10, evaluation_regime_level: string\n child 2, soft_targets: struct<(A): double, (B): double>\n child 0, (A): double\n child 1, (B): double\n child 3, source_snapshot_sha256: string\n child 4, target_contract: string\nabsolute_canonical_value_difference: double\nstandardized_canonical_value_difference: double\nsd_logistic_similarity: double\ntarget_train_sample_standard_deviation: double\ntarget_a: double\ntarget_b: double\nlabel_mode: string\nbinary_label: int64\nis_decisive: bool\nmorgan_tanimoto_similarity: double\ngraph_a_position: int64\ngraph_b_position: int64\ngraph_offset: int64\nrepresentation: string\npair_bucket_key: string\nquery_parent_id: string\nretrieval_parent_id: string\nquery_scaffold: string\nretrieval_scaffold: string\nquery_evidence: struct<aggregation_method: string, display_measurement_text: string, display_unit_text: string, disp (... 184 chars omitted)\n child 0, aggregation_method: string\n child 1, display_measurement_text: string\n child 2, display_unit_text: string\n child 3, display_transform_id: string\n child 4, source_record_count: int64\n child 5, deduplicated_source_record_count: int64\n child 6, aggregate_min: double\n child 7, aggregate_max: double\n child 8, aggregate_q1: double\n child 9, aggregate_q3: double\nretrieval_evidence: struct<aggregation_method: string, display_measurement_text: string, display_unit_text: string, disp (... 184 chars omitted)\n child 0, aggregation_method: string\n child 1, display_measurement_text: string\n child 2, display_unit_text: string\n child 3, display_transform_id: string\n child 4, source_record_count: int64\n child 5, deduplicated_source_record_count: int64\n child 6, aggregate_min: double\n child 7, aggregate_max: double\n child 8, aggregate_q1: double\n child 9, aggregate_q3: double\nbucket_total_record_count: int64\nbucket_train_record_count: int64\nbucket_train_parent_count: int64\nevaluation_regime: string\ncalibration_scope: string\nlogit_geometry_transform: string\nquery_source_payload_json: string\nretrieval_source_payload_json: string\nlevel: string\nlevel_family: string\nquery_source_row_uid: string\nretrieval_source_row_uid: string\nupstream_pair_bucket_key: string",
|
| 3 |
+
"artifacts": {
|
| 4 |
+
"README.md": "f087f65430e6ffa724100db2518f05d60441b624feaba4a89d4c3afaeab00241",
|
| 5 |
+
"calibration.json": "801070b703c7e508540d517011b046ccca0d2cca82529baa3c8f07168464ea9a",
|
| 6 |
+
"test_ranking/data.parquet": "825a0d6d08eb99ca1284c33a1f1159323493c974c1a639d6f26f441124eb201c",
|
| 7 |
+
"train/data.parquet": "6bac1f9f795d731fc56b07b67440614e08e727b7f1a8f40eea8b1324c48ba735",
|
| 8 |
+
"validation_ranking/data.parquet": "fadac42908119b2289493ef38692edbbdff8b483328e78132502fa8202d76eb1"
|
| 9 |
+
},
|
| 10 |
+
"build": {
|
| 11 |
+
"degree_cap": 48,
|
| 12 |
+
"implementation_sha256": {
|
| 13 |
+
"assay_transfer/context_conditioned/__init__.py": "97ee26d7f7bf2e4f69e8ecd936a9d728770bb18642511bb2d177ae40ba2975f9",
|
| 14 |
+
"assay_transfer/context_conditioned/baselines.py": "e6cb7999d98273dc87c0236d94b344b247b2994d508c09e436811f30fb6cc15d",
|
| 15 |
+
"assay_transfer/context_conditioned/build.py": "43af4fa0e1a3acda199e7c331371eb350f1d664aab05728c325705848b9c6e0d",
|
| 16 |
+
"assay_transfer/context_conditioned/continuous_source.py": "080f68763e13a257b2980ab45ded6a806b7041077645853963ce257179f73098",
|
| 17 |
+
"assay_transfer/context_conditioned/dataset.py": "b1cbcd4c371d679315b75fe7130bdd2d52f7174b2a8900d7d8ec834548ef29ca",
|
| 18 |
+
"assay_transfer/context_conditioned/hf.py": "ac5298a42670c4f6b3878dbeeb30f83580fa340ca2f13d222ae01f295afb9b78",
|
| 19 |
+
"assay_transfer/context_conditioned/pairs.py": "2ec358955502e2dc63911107b99f66e75714fca44244b91959d4c468decc507f",
|
| 20 |
+
"assay_transfer/context_conditioned/publish.py": "4d7486c7c23beb3e5b8f857f0072757aa0897f2e0a463dd9768b09ea3d94582a",
|
| 21 |
+
"assay_transfer/context_conditioned/record_voter_binary.py": "0bf0a38f3bd1ec6bdbbf57da6a40d069d8a0b399fd40d176f2e9b18c166f22a2",
|
| 22 |
+
"assay_transfer/context_conditioned/splits.py": "4850692808b5a4523d604d649b043f1fa6ddd34ee260a21796a15fce94a3e37f",
|
| 23 |
+
"assay_transfer/context_conditioned/templates/BBB_Martins.jinja": "5829c615e0235ba26dfee96e922d496f32579ed472a9a67cc64c96e29b5e10b0",
|
| 24 |
+
"assay_transfer/context_conditioned/templates/Bioavailability_Ma.jinja": "ad3d8734bed1a0402f072b0e044be966a1a704062e14f064f424332aa2f090b5",
|
| 25 |
+
"assay_transfer/context_conditioned/templates/Skin_Reaction.jinja": "cca2084355bc79720a7799ef9af84fe2c8f81e6f318d621151ba52c0c1032cd7",
|
| 26 |
+
"assay_transfer/context_conditioned/templates/prompt_v3_1.jinja": "33e48f919237a6aeaff28f4c77c6dd9635e54bf859c990af4ad0323952ab05f1",
|
| 27 |
+
"assay_transfer/context_conditioned/templates/prompt_v5.jinja": "5206a666dcc04b20eb2969c30574282580a28d52ced1c425555892b554035e2d",
|
| 28 |
+
"assay_transfer/context_conditioned/test_dataset.py": "e2dc39004cae616e3bc92b8acd496f5e3eca12c86e2bc430348af9e7561e79a2",
|
| 29 |
+
"assay_transfer/context_conditioned/v10.py": "71353232c65de6b98045bc62950df7c35448aa8f12195ff4e5669c6acf63eff6",
|
| 30 |
+
"assay_transfer/context_conditioned/v10_oral.py": "403ec722f575e8c8c74ba1b18b38468b9cb3b271181d20c792066011bb7a6cb4",
|
| 31 |
+
"assay_transfer/context_conditioned/v6.py": "2e420033d9041acf54258e732844ec326f860de0edf3f186abb695f302eb2e98",
|
| 32 |
+
"assay_transfer/context_conditioned/v7_continuous.py": "a625885ac040bbd0a7f017899e3e50f2167c19b5242ae01140363f0d6819910d",
|
| 33 |
+
"assay_transfer/context_conditioned/v8.py": "e63bbe7c45c43d119e9f6739acbeac5b0c02ea55baad7e9f41c5a244e16c7f61",
|
| 34 |
+
"assay_transfer/context_conditioned/v9.py": "77e2c02b2572eeed251439a5bc933da18f3ac7f1ecdd91c1622874fa59612e02",
|
| 35 |
+
"assay_transfer/context_conditioned/v9_official.py": "4f9e96bfb25d03dcdd294766656adee1e504daa63c2c51f63f21a07288e03a3b",
|
| 36 |
+
"assay_transfer/shared/artifacts.py": "9a7233e7b5ffd04481466c32f7687a05b1d1f5308a2594bf53abe90341cc6427",
|
| 37 |
+
"assay_transfer/shared/pairs.py": "a0ab429dcd342c98f42bc98da4da39c0b732b7212c040cc6df9dfde3462fb898",
|
| 38 |
+
"assay_transfer/shared/rendering.py": "367445b3dab67752f81ee703848b222654fa0111c06ad5fb1552272495ae107f",
|
| 39 |
+
"assay_transfer/shared/tasks.py": "70f920db036385ec2f408b783db154938c4a2e7ec0a357a5a52504e754b0b7c2",
|
| 40 |
+
"assay_transfer/shared/v3_validation.py": "dea8142e2398decfbd4ac0859dc8921d73a70851b80a85e46ff2ecf299b0aa83",
|
| 41 |
+
"requirements-data-pipeline.txt": "f0aa5dfea67de8b2f32bce91cbb0fd078c5925b8d014c1fb406ba8806985e1a7"
|
| 42 |
+
},
|
| 43 |
+
"packages": {
|
| 44 |
+
"jinja2": "3.1.6",
|
| 45 |
+
"pyarrow": "23.0.0",
|
| 46 |
+
"rdkit_core": "2025.09.4",
|
| 47 |
+
"rdkit_distribution": "2025.9.4"
|
| 48 |
+
},
|
| 49 |
+
"python_version": "3.11.14",
|
| 50 |
+
"repository": {
|
| 51 |
+
"commit": "7ae235a1573a01dc281266ff781bda01e80d8ae0",
|
| 52 |
+
"dirty": true
|
| 53 |
+
},
|
| 54 |
+
"seed": 20260818
|
| 55 |
+
},
|
| 56 |
+
"calibration_scope": "official_train_plus_validation",
|
| 57 |
+
"direct_rows": {
|
| 58 |
+
"test_ranking": 23369,
|
| 59 |
+
"train": 46992,
|
| 60 |
+
"validation_ranking": 22682
|
| 61 |
+
},
|
| 62 |
+
"generation": "v10",
|
| 63 |
+
"hub_dataset_id": "jiosephlee/context-conditioned-molecule-transfer-v10.2-bioavailability-ma-mixed-continuous-intern",
|
| 64 |
+
"indirect_train_rows": 46992,
|
| 65 |
+
"indirect_train_rows_by_source": {
|
| 66 |
+
"fa": 7719,
|
| 67 |
+
"fg": 609,
|
| 68 |
+
"fh": 7836,
|
| 69 |
+
"oral_exposure": 30828
|
| 70 |
+
},
|
| 71 |
+
"label_mode": "mixed_continuous",
|
| 72 |
+
"membership": {
|
| 73 |
+
"test_ranking": "8fd6cd5b23f61d4de502ba7a474c848c794d1b3ef84e00ba4617a4f8d131ed08",
|
| 74 |
+
"train": "51f6a60362ecacff975e3b237808404d65afa0dfa0e797aa723f49179db86d94",
|
| 75 |
+
"validation_ranking": "b162dba94feee948456643c74b7055f28423a2f62b199f08f89f837d66ec3976"
|
| 76 |
+
},
|
| 77 |
+
"morgan_similarity": {
|
| 78 |
+
"combined": {
|
| 79 |
+
"nontransfer": {
|
| 80 |
+
"mean": 0.15868945223353373,
|
| 81 |
+
"rows": 46769
|
| 82 |
+
},
|
| 83 |
+
"transfer": {
|
| 84 |
+
"mean": 0.18163652441863504,
|
| 85 |
+
"rows": 47215
|
| 86 |
+
}
|
| 87 |
+
},
|
| 88 |
+
"direct": {
|
| 89 |
+
"nontransfer": {
|
| 90 |
+
"mean": 0.20433852234247285,
|
| 91 |
+
"rows": 23496
|
| 92 |
+
},
|
| 93 |
+
"transfer": {
|
| 94 |
+
"mean": 0.23850003265624803,
|
| 95 |
+
"rows": 23496
|
| 96 |
+
}
|
| 97 |
+
},
|
| 98 |
+
"indirect": {
|
| 99 |
+
"nontransfer": {
|
| 100 |
+
"mean": 0.11260297643412523,
|
| 101 |
+
"rows": 23273
|
| 102 |
+
},
|
| 103 |
+
"transfer": {
|
| 104 |
+
"mean": 0.1253076324100784,
|
| 105 |
+
"rows": 23719
|
| 106 |
+
}
|
| 107 |
+
}
|
| 108 |
+
},
|
| 109 |
+
"official_split_policy": "official_gold_train_valid_test.train_plus_valid_calibration.v1",
|
| 110 |
+
"prompt_contract": {
|
| 111 |
+
"direct": "context_conditioned_A_value_only.v10",
|
| 112 |
+
"indirect": "source_native_query_safe_record_prompt.v25"
|
| 113 |
+
},
|
| 114 |
+
"ranking_reference_scope": "official_train_only",
|
| 115 |
+
"realized_indirect_to_direct_ratio": 1.0,
|
| 116 |
+
"rows": {
|
| 117 |
+
"test_ranking": 23369,
|
| 118 |
+
"train": 93984,
|
| 119 |
+
"validation_ranking": 22682
|
| 120 |
+
},
|
| 121 |
+
"schema_version": "context_conditioned_molecule_transfer.v11",
|
| 122 |
+
"source": {
|
| 123 |
+
"indirect": {
|
| 124 |
+
"completion_sha256": "7d6330cc9938b52f902f25040a1623d37b1f3430763114a836c3f056c3727248",
|
| 125 |
+
"eligible_buckets": 989,
|
| 126 |
+
"eligible_records": 67482,
|
| 127 |
+
"manifest_sha256": "54569cbc2c61756a64f2534cb6bb3a914ce51125997d8ba127e6447ffad08367",
|
| 128 |
+
"release": "/vast/projects/myatskar/design-documents/joseph/starling_assay_transfer/assay_transfer/record_level/artifacts/oral_v25",
|
| 129 |
+
"split_manifest_sha256": "86efb4b17034b77886f21e2ec84da3aacbb877f8744f8e4fd725f0601457b3ec",
|
| 130 |
+
"unit": "individual_v25_continuous_indirect_record"
|
| 131 |
+
},
|
| 132 |
+
"snapshot_manifest_sha256": "5b94e7ed0581d8b2766bae7d1a8902d880ec19738376a85309b63a270860490c",
|
| 133 |
+
"snapshot_root": "/vast/projects/myatskar/design-documents/joseph/starling_assay_transfer/assay_transfer/context_conditioned/artifacts/v9_official/sources/Bioavailability_Ma"
|
| 134 |
+
},
|
| 135 |
+
"source_rows": {
|
| 136 |
+
"test_ranking": 269,
|
| 137 |
+
"train": 1958,
|
| 138 |
+
"validation_ranking": 262
|
| 139 |
+
},
|
| 140 |
+
"target_contract": "sd_sigmoid_center_0.4_temperature_0.1_unclipped.v10.2",
|
| 141 |
+
"target_policy": {
|
| 142 |
+
"center_sd": 0.4,
|
| 143 |
+
"tau_sd": 0.1
|
| 144 |
+
},
|
| 145 |
+
"task_id": "Bioavailability_Ma",
|
| 146 |
+
"training": {
|
| 147 |
+
"candidate_policy": "official_train_unique_parent_0.4sd_balanced",
|
| 148 |
+
"direction": "one_reference_A_to_query_B",
|
| 149 |
+
"graph": {
|
| 150 |
+
"balance_boundary_sd": 0.4,
|
| 151 |
+
"degree_cap": 48,
|
| 152 |
+
"method": "top_morgan_0.4sd_balanced_no_reverse.v1",
|
| 153 |
+
"query_order": "ascending_preconstraint_sparse_class_size_then_record_id",
|
| 154 |
+
"query_order_sha256": "c206af13da09e6427769f8758c77218eef29731bf60597b58010f7390d9d574d",
|
| 155 |
+
"realized_near_count": {
|
| 156 |
+
"distribution": {
|
| 157 |
+
"12": 1958
|
| 158 |
+
},
|
| 159 |
+
"sparse_query_count": 0,
|
| 160 |
+
"sparse_query_sha256": "4f53cda18c2baa0c0354bb5f9a3ecbe5ed12ab4d8e11ba873c2f11161202b945"
|
| 161 |
+
},
|
| 162 |
+
"reference_parent_per_query": "unique",
|
| 163 |
+
"reverse_parent_pairs": "forbidden",
|
| 164 |
+
"selection": "scarce_class_first_then_descending_morgan_parent_record",
|
| 165 |
+
"similarity": {
|
| 166 |
+
"max": 1.0,
|
| 167 |
+
"mean": 0.2214192774993604,
|
| 168 |
+
"min": 0.0
|
| 169 |
+
},
|
| 170 |
+
"target_near_far": {
|
| 171 |
+
"far": 12,
|
| 172 |
+
"near": 12
|
| 173 |
+
}
|
| 174 |
+
},
|
| 175 |
+
"indirect_candidate_pair_count": 132783,
|
| 176 |
+
"indirect_record_degree_cap": 2,
|
| 177 |
+
"indirect_selected_pair_count": 46992,
|
| 178 |
+
"indirect_selected_unique_queries": 46992,
|
| 179 |
+
"indirect_selection": "bucket_round_robin_unique_query_v25_order",
|
| 180 |
+
"reference_count_per_query": 24,
|
| 181 |
+
"retrieval_parent_may_repeat_across_context_records": false,
|
| 182 |
+
"selection": "highest_available_morgan_with_no_reverse_parent_edges"
|
| 183 |
+
},
|
| 184 |
+
"version": "v10_2"
|
| 185 |
+
}
|
test_ranking/data.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:825a0d6d08eb99ca1284c33a1f1159323493c974c1a639d6f26f441124eb201c
|
| 3 |
+
size 1850521
|
train/data.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6bac1f9f795d731fc56b07b67440614e08e727b7f1a8f40eea8b1324c48ba735
|
| 3 |
+
size 39171630
|
validation_ranking/data.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:fadac42908119b2289493ef38692edbbdff8b483328e78132502fa8202d76eb1
|
| 3 |
+
size 1811920
|