Hugging Face's logo

Buckets:
FineEnvs
/
retroenv-bucket

Files
xet
FineEnvs/retroenv-bucket / openenv /evalsets /final_eval.json
AdithyaSK's picture
AdithyaSK
download
raw
16.2 kB
{
"evalset_version": 1,
"name": "final_eval",
"seed": "retroenv-final-eval-v1",
"design": {
"source_splits": [
"test_id",
"test_hard"
],
"variant": "standard",
"quotas": {
"easy": 17,
"medium": 17,
"hard": 16
},
"tier": "the release's difficulty.tier",
"strata": "source split x shortest-route depth, visited round-robin from shallow to deep",
"within_stratum": "sha256(seed:task_id); a first-reaction class the tier has not used is preferred"
},
"tasks": [
{
"task_id": "retro_1ba7b351bd23db0e",
"source_split": "test_id",
"tier": "easy",
"min_depth": 2,
"max_depth": 4,
"first_reaction": "condensation (imine/oxime/amidine/guanidine)",
"heavy_atoms": 18,
"target_smiles": "CNC(=O)CCC(=NNC(N)=S)c1ccccc1"
},
{
"task_id": "retro_83e0303eb34232da",
"source_split": "test_id",
"tier": "easy",
"min_depth": 3,
"max_depth": 5,
"first_reaction": "hydrogenation (C=C/C#C/arene)",
"heavy_atoms": 30,
"target_smiles": "NS(=O)(=O)OC[C@H]1CC[C@H](n2ccc3c(N[C@H]4CCc5ccccc54)ncnc32)C1"
},
{
"task_id": "retro_e507ee9a9fc68a9a",
"source_split": "test_id",
"tier": "easy",
"min_depth": 2,
"max_depth": 4,
"first_reaction": "urea/carbamate formation",
"heavy_atoms": 36,
"target_smiles": "CC(C)(CNC(=O)Oc1ccc([N+](=O)[O-])cc1)CO[Si](c1ccccc1)(c1ccccc1)C(C)(C)C"
},
{
"task_id": "retro_9c206ca51cfc9edd",
"source_split": "test_id",
"tier": "easy",
"min_depth": 3,
"max_depth": 5,
"first_reaction": "ester hydrolysis",
"heavy_atoms": 38,
"target_smiles": "CCCCOc1ccc(C(=O)N(C)Cc2cccc(-c3ccc(CCC(=O)O)cc3OCCCC)c2)cc1"
},
{
"task_id": "retro_bc89daabe6f3a1cb",
"source_split": "test_id",
"tier": "easy",
"min_depth": 2,
"max_depth": 4,
"first_reaction": "amide coupling",
"heavy_atoms": 43,
"target_smiles": "CCn1nc(C(=O)NCCN2CCCC2)c(=O)c2cc(OCc3ccc(OC)cc3)c(OCc3ccc(OC)cc3)cc21"
},
{
"task_id": "retro_17031e7672333731",
"source_split": "test_id",
"tier": "easy",
"min_depth": 3,
"max_depth": 5,
"first_reaction": "oxidation",
"heavy_atoms": 35,
"target_smiles": "C[C@@H](NC(=O)[C@@H]1C[C@H]2C[C@H]2N1C(=O)Cn1nc(C(N)=O)c2cc[n+]([O-])cc21)c1cccc(Cl)c1F"
},
{
"task_id": "retro_2c3155ee731884b1",
"source_split": "test_id",
"tier": "easy",
"min_depth": 2,
"max_depth": 4,
"first_reaction": "N-alkylation",
"heavy_atoms": 30,
"target_smiles": "Cc1ccc(-c2cccc(/C=C/C(=O)Nc3ccc(Cn4ccnc4)cc3)c2)cc1"
},
{
"task_id": "retro_f6b703ccf57008da",
"source_split": "test_id",
"tier": "easy",
"min_depth": 3,
"max_depth": 5,
"first_reaction": "benzyl deprotection",
"heavy_atoms": 32,
"target_smiles": "CCCCCCCCOc1cc(C(=O)OC)ccc1C#Cc1cc(C=O)cc(CC#N)c1"
},
{
"task_id": "retro_d24f2932c2fbcc9b",
"source_split": "test_id",
"tier": "easy",
"min_depth": 2,
"max_depth": 4,
"first_reaction": "nitro reduction",
"heavy_atoms": 17,
"target_smiles": "CC(NC(=O)OC(C)(C)C)c1cccc(N)c1"
},
{
"task_id": "retro_a85ad0ec0831cdbf",
"source_split": "test_id",
"tier": "easy",
"min_depth": 3,
"max_depth": 5,
"first_reaction": "SNAr",
"heavy_atoms": 29,
"target_smiles": "CCOC(=O)c1cnn(-c2nc3cc([N+](=O)[O-])c(Cl)cc3n2COCCOC)c1"
},
{
"task_id": "retro_ac0c682f5c3fe14a",
"source_split": "test_id",
"tier": "easy",
"min_depth": 2,
"max_depth": 4,
"first_reaction": "esterification",
"heavy_atoms": 36,
"target_smiles": "CCC(C)(C)C(=O)O[C@H]1C[C@H]2[C@@H](CC[C@@H]3C[C@H](O[Si](C)(C)C(C)(C)C)CC[C@@]32C)[C@@H]2CCC(=O)[C@@]12C"
},
{
"task_id": "retro_eb581bcb2aaacf66",
"source_split": "test_id",
"tier": "easy",
"min_depth": 3,
"max_depth": 5,
"first_reaction": "Boc deprotection",
"heavy_atoms": 34,
"target_smiles": "Cc1c(CN2CCN(C(=O)CN)CC2)sc2c(N3CCOCC3)nc(-c3cnc(N)nc3)nc12"
},
{
"task_id": "retro_66b6cff002d7eb56",
"source_split": "test_id",
"tier": "easy",
"min_depth": 2,
"max_depth": 4,
"first_reaction": "O-sulfonylation (mesylate/tosylate/triflate)",
"heavy_atoms": 29,
"target_smiles": "COC(=O)c1ccc2nc(OS(=O)(=O)C(F)(F)F)c(N3CCCC[C@@H]3C)nc2c1"
},
{
"task_id": "retro_980b0360b1292438",
"source_split": "test_id",
"tier": "easy",
"min_depth": 3,
"max_depth": 5,
"first_reaction": "reductive amination",
"heavy_atoms": 35,
"target_smiles": "COc1cc(Nc2c(C#N)cnc3cc(-c4ccc(CN5CCOCC5)cc4)sc23)c(Cl)cc1Cl"
},
{
"task_id": "retro_b7d952a96de74f76",
"source_split": "test_id",
"tier": "easy",
"min_depth": 2,
"max_depth": 4,
"first_reaction": "other protection (Cbz/Fmoc/silyl/acetal)",
"heavy_atoms": 20,
"target_smiles": "COc1ccccc1C[C@@H](O)CO[Si](C)(C)C(C)(C)C"
},
{
"task_id": "retro_14602f5f96554500",
"source_split": "test_id",
"tier": "easy",
"min_depth": 3,
"max_depth": 5,
"first_reaction": "Sonogashira coupling",
"heavy_atoms": 32,
"target_smiles": "CC(C)(C)OC(=O)N1CCCC1COc1ccc(OCC#Cc2ccc(CO)cc2)cc1"
},
{
"task_id": "retro_66953a5ef0520b19",
"source_split": "test_id",
"tier": "easy",
"min_depth": 2,
"max_depth": 4,
"first_reaction": "Suzuki coupling",
"heavy_atoms": 27,
"target_smiles": "COc1ccc(COc2cc(C#N)cc(-c3ccc4c(c3)OCO4)c2)cc1"
},
{
"task_id": "retro_d6c00413abb4733c",
"source_split": "test_id",
"tier": "medium",
"min_depth": 2,
"max_depth": 4,
"first_reaction": "other",
"heavy_atoms": 21,
"target_smiles": "Nc1nc(F)c(Cl)c(N2C(=O)c3ccccc3C2=O)c1F"
},
{
"task_id": "retro_4b3deba41c5578d0",
"source_split": "test_id",
"tier": "medium",
"min_depth": 3,
"max_depth": 5,
"first_reaction": "sulfonamide formation",
"heavy_atoms": 29,
"target_smiles": "CCS(=O)(=O)Nc1ccc(C)c(-c2ccc3cc(NC(=O)C4CC4)ncc3c2)c1"
},
{
"task_id": "retro_60aa0fd2b9d1bbf5",
"source_split": "test_hard",
"tier": "medium",
"min_depth": 4,
"max_depth": 6,
"first_reaction": "Boc deprotection",
"heavy_atoms": 35,
"target_smiles": "COc1cc(Cl)ccc1CNC(=O)C1CCN(Cc2ccn(-c3ccc(C(F)(F)F)cc3)c2)CC1"
},
{
"task_id": "retro_36b7ec0671579394",
"source_split": "test_id",
"tier": "medium",
"min_depth": 4,
"max_depth": 6,
"first_reaction": "O-/S-alkylation",
"heavy_atoms": 34,
"target_smiles": "CCCCCC1CCC(c2ccc(C3CCC(C4CCC(OCCC)CC4)CC3)c(F)c2)CC1"
},
{
"task_id": "retro_9acf6323e734c848",
"source_split": "test_hard",
"tier": "medium",
"min_depth": 5,
"max_depth": 7,
"first_reaction": "condensation (imine/oxime/amidine/guanidine)",
"heavy_atoms": 26,
"target_smiles": "O=C(Nc1cnc(C=NO)nc1O)C(c1ccccc1)c1ccccc1"
},
{
"task_id": "retro_86ccb9e83e3e9e9e",
"source_split": "test_id",
"tier": "medium",
"min_depth": 5,
"max_depth": 7,
"first_reaction": "reductive amination",
"heavy_atoms": 21,
"target_smiles": "CN1CC(n2nccc2-c2cc(C(F)(F)F)ccc2O)C1"
},
{
"task_id": "retro_34e025851a243d29",
"source_split": "test_hard",
"tier": "medium",
"min_depth": 6,
"max_depth": 8,
"first_reaction": "amide coupling",
"heavy_atoms": 37,
"target_smiles": "Cn1ncc(Br)c1-c1cc(C(=O)N[C@@H](Cc2ccc(F)cc2)CN2C(=O)c3ccccc3C2=O)sc1Cl"
},
{
"task_id": "retro_7710f159dfe6f6ca",
"source_split": "test_id",
"tier": "medium",
"min_depth": 6,
"max_depth": 8,
"first_reaction": "ester hydrolysis",
"heavy_atoms": 26,
"target_smiles": "O=C(O)CC[C@@H]1c2ccccc2C[C@H]1NC(=O)c1cc2cc(Cl)sc2[nH]1"
},
{
"task_id": "retro_ef76e1f83b9ea580",
"source_split": "test_hard",
"tier": "medium",
"min_depth": 7,
"max_depth": 9,
"first_reaction": "heteroaromatic ring formation",
"heavy_atoms": 43,
"target_smiles": "CCCc1nc(C)n(-c2ccc(OC(C)(C)CO)cc2)c(=O)c1Cc1ccc(-c2ccccc2-c2noc(=O)[nH]2)cc1F"
},
{
"task_id": "retro_b4e2535f3749bf6f",
"source_split": "test_id",
"tier": "medium",
"min_depth": 7,
"max_depth": 9,
"first_reaction": "other cross-coupling (Heck/Stille/Negishi/cyanation/borylation)",
"heavy_atoms": 31,
"target_smiles": "CNC(=O)c1c(C2CC2)oc2cc(N(C)S(C)(=O)=O)c(B3OC(C)(C)C(C)(C)O3)cc12"
},
{
"task_id": "retro_9a64ba4a5abccb72",
"source_split": "test_hard",
"tier": "medium",
"min_depth": 8,
"max_depth": 10,
"first_reaction": "other deprotection (Cbz/Fmoc/silyl/acetal/sulfonyl)",
"heavy_atoms": 29,
"target_smiles": "COC[C@H](C)NC(=O)c1c[nH]c2ncc(-c3nn(C)c4cc(Cl)cc(F)c34)nc12"
},
{
"task_id": "retro_4b8132ec81e7c1c2",
"source_split": "test_id",
"tier": "medium",
"min_depth": 8,
"max_depth": 10,
"first_reaction": "N-alkylation",
"heavy_atoms": 18,
"target_smiles": "CC(C)(C)NCC(O)c1cc(F)cc(C#N)c1F"
},
{
"task_id": "retro_25a7bbebb9a0e096",
"source_split": "test_id",
"tier": "medium",
"min_depth": 2,
"max_depth": 4,
"first_reaction": "oxidation",
"heavy_atoms": 15,
"target_smiles": "CC(=O)c1cc2cccc(Cl)c2nc1Cl"
},
{
"task_id": "retro_8e846ab781ad49e8",
"source_split": "test_id",
"tier": "medium",
"min_depth": 3,
"max_depth": 5,
"first_reaction": "esterification",
"heavy_atoms": 22,
"target_smiles": "C=CCOC(=O)c1ccc2c(=O)c3cc(C)ccc3sc2c1"
},
{
"task_id": "retro_f9a506c7c77b3484",
"source_split": "test_hard",
"tier": "medium",
"min_depth": 4,
"max_depth": 6,
"first_reaction": "nitro reduction",
"heavy_atoms": 18,
"target_smiles": "Nc1cc(Oc2ccnc3[nH]ccc23)ccc1O"
},
{
"task_id": "retro_b8e31cdf9cbf6c75",
"source_split": "test_id",
"tier": "medium",
"min_depth": 4,
"max_depth": 6,
"first_reaction": "urea/carbamate formation",
"heavy_atoms": 34,
"target_smiles": "COCCNC(=O)Nc1nc2ccc(Sc3cnc4ccc(-c5ccc(F)cc5)cn34)cc2s1"
},
{
"task_id": "retro_e4780720c4e30393",
"source_split": "test_hard",
"tier": "medium",
"min_depth": 5,
"max_depth": 7,
"first_reaction": "SNAr",
"heavy_atoms": 33,
"target_smiles": "CC(C)(O)c1cc(F)c(-c2cc(C(N)=O)c(Nc3ccnc(N4CCC(O)C4)n3)s2)c(F)c1"
},
{
"task_id": "retro_eb2098decfe08d9a",
"source_split": "test_id",
"tier": "hard",
"min_depth": 3,
"max_depth": 5,
"first_reaction": "cyclisation (non-aromatic ring)",
"heavy_atoms": 29,
"target_smiles": "CCn1c2ccccc2c2cc3c4ccc(OC)cc4c4ccccc4c3cc21"
},
{
"task_id": "retro_8f6b26c4af726398",
"source_split": "test_hard",
"tier": "hard",
"min_depth": 4,
"max_depth": 6,
"first_reaction": "amide/nitrile hydrolysis",
"heavy_atoms": 13,
"target_smiles": "COCC1(N)Cc2ccccc2C1"
},
{
"task_id": "retro_e1c9685c02611279",
"source_split": "test_id",
"tier": "hard",
"min_depth": 4,
"max_depth": 6,
"first_reaction": "SNAr",
"heavy_atoms": 31,
"target_smiles": "CCC(CC)n1c(=O)[nH]c2cnc(Nc3ccc(N4CCN(C(C)=O)CC4)cc3)nc21"
},
{
"task_id": "retro_77d7dfba29cb6050",
"source_split": "test_hard",
"tier": "hard",
"min_depth": 5,
"max_depth": 7,
"first_reaction": "other deprotection (Cbz/Fmoc/silyl/acetal/sulfonyl)",
"heavy_atoms": 34,
"target_smiles": "Cn1cnc(S(=O)(=O)Nc2ccc(F)c(Nc3ncccc3-c3ncnc4[nH]cnc34)c2F)c1"
},
{
"task_id": "retro_1b6b11bb45aed26a",
"source_split": "test_id",
"tier": "hard",
"min_depth": 5,
"max_depth": 7,
"first_reaction": "amide coupling",
"heavy_atoms": 29,
"target_smiles": "CC(=O)N1CCC2(CCC(Nc3nc(Nc4cnn(C)c4)ncc3Cl)CC2)CC1"
},
{
"task_id": "retro_f295968af0e2f27f",
"source_split": "test_hard",
"tier": "hard",
"min_depth": 6,
"max_depth": 8,
"first_reaction": "sulfonamide formation",
"heavy_atoms": 37,
"target_smiles": "COC(=O)CCS(=O)(=O)N1CC=C(c2ccc3nc(OC4CCN(c5nnc(C)s5)CC4)sc3c2)CC1"
},
{
"task_id": "retro_01583f4ac0522fd0",
"source_split": "test_id",
"tier": "hard",
"min_depth": 6,
"max_depth": 8,
"first_reaction": "ester hydrolysis",
"heavy_atoms": 31,
"target_smiles": "CCCOCCN(CC)c1ccc(-c2ccc3c(c2)C=C(C(=O)O)CCN3C=O)cc1"
},
{
"task_id": "retro_753dd86c90eba25d",
"source_split": "test_hard",
"tier": "hard",
"min_depth": 7,
"max_depth": 9,
"first_reaction": "condensation (imine/oxime/amidine/guanidine)",
"heavy_atoms": 31,
"target_smiles": "CC(C)(C)OC(=O)N1CCCC(ON=C(c2ccccc2)c2cc3ccncc3[nH]2)C1"
},
{
"task_id": "retro_436e955405feb310",
"source_split": "test_id",
"tier": "hard",
"min_depth": 7,
"max_depth": 9,
"first_reaction": "condensation (imine/oxime/amidine/guanidine)",
"heavy_atoms": 34,
"target_smiles": "CN1C(=O)C(=NNC(=O)Nc2ccccc2)c2cc(SCCc3ccc(C(=O)O)cc3)ccc21"
},
{
"task_id": "retro_d735e87a4090c05e",
"source_split": "test_hard",
"tier": "hard",
"min_depth": 8,
"max_depth": 10,
"first_reaction": "O-/S-alkylation",
"heavy_atoms": 36,
"target_smiles": "CC(C)Cn1c(CNC(=O)OC(C)(C)C)c(-c2ccc(F)cc2)c2cc(OCC(N)=O)ccc2c1=O"
},
{
"task_id": "retro_6bd5b47f061a5bf6",
"source_split": "test_id",
"tier": "hard",
"min_depth": 8,
"max_depth": 10,
"first_reaction": "reductive amination",
"heavy_atoms": 27,
"target_smiles": "CN(C)C1(C)CCC(Oc2ncnc3sc4c(c23)[C@@H](CC(N)=O)CC4)CC1"
},
{
"task_id": "retro_ecfc16bb24d8aa10",
"source_split": "test_id",
"tier": "hard",
"min_depth": 3,
"max_depth": 5,
"first_reaction": "Suzuki coupling",
"heavy_atoms": 42,
"target_smiles": "CCOC(=O)c1cc2cc(-c3cccc(-c4cc(C(F)(F)F)c(C#N)c(=O)n4Cc4ccc(C)cc4C)c3)ccc2[nH]1"
},
{
"task_id": "retro_00cd2ad59af42e85",
"source_split": "test_hard",
"tier": "hard",
"min_depth": 4,
"max_depth": 6,
"first_reaction": "Grignard/organolithium addition",
"heavy_atoms": 23,
"target_smiles": "CC(O)c1cc(Cl)cc2c(C#N)cn(COCC[Si](C)(C)C)c12"
},
{
"task_id": "retro_0db4832c30ee7b3c",
"source_split": "test_id",
"tier": "hard",
"min_depth": 4,
"max_depth": 6,
"first_reaction": "Boc deprotection",
"heavy_atoms": 36,
"target_smiles": "C(#CC1CC1)c1cnc(N2CCN(c3ncnc(Nc4cnn(C[C@@H]5CNCCO5)c4)n3)CC2)nc1"
},
{
"task_id": "retro_965813554d7d7f96",
"source_split": "test_hard",
"tier": "hard",
"min_depth": 5,
"max_depth": 7,
"first_reaction": "halogenation/nitration",
"heavy_atoms": 20,
"target_smiles": "Nc1c(C(=O)O)cc(Br)c(-c2ccc(F)cc2F)c1F"
},
{
"task_id": "retro_f8e680ee09d05bc8",
"source_split": "test_id",
"tier": "hard",
"min_depth": 5,
"max_depth": 7,
"first_reaction": "urea/carbamate formation",
"heavy_atoms": 22,
"target_smiles": "CNC(=S)N1CCOCC1c1noc(-c2cccc(Cl)c2)n1"
}
],
"size": 50,
"evalset_id": "6b5112f1fcf6ee374b0ba5a199c7406dc4e0555d7a7c75ce2e2235c310ebf92a"
}

Xet Storage Details

Size:
16.2 kB
·
Xet hash:
068678f0623545833de4de40f82747d8e7119de6caa2e95ce8f1287ceb46f22d

Xet efficiently stores files, intelligently splitting them into unique chunks and accelerating uploads and downloads. More info.