Upload data/features/feature_metadata.json with huggingface_hub
Browse files
data/features/feature_metadata.json
ADDED
|
@@ -0,0 +1,66 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"n_proteins": 82404,
|
| 3 |
+
"n_features": 15411,
|
| 4 |
+
"components": [
|
| 5 |
+
"ESM2_L29-35 (15360 dims)",
|
| 6 |
+
"Taxonomy_top50 (51 dims)"
|
| 7 |
+
],
|
| 8 |
+
"train_n": 57686,
|
| 9 |
+
"val_n": 12359,
|
| 10 |
+
"test_n": 12359,
|
| 11 |
+
"taxonomy_info": {
|
| 12 |
+
"n_features": 51,
|
| 13 |
+
"top50_taxa": [
|
| 14 |
+
"9606",
|
| 15 |
+
"10090",
|
| 16 |
+
"3702",
|
| 17 |
+
"559292",
|
| 18 |
+
"10116",
|
| 19 |
+
"284812",
|
| 20 |
+
"83333",
|
| 21 |
+
"7227",
|
| 22 |
+
"6239",
|
| 23 |
+
"83332",
|
| 24 |
+
"7955",
|
| 25 |
+
"44689",
|
| 26 |
+
"39947",
|
| 27 |
+
"9913",
|
| 28 |
+
"9031",
|
| 29 |
+
"8355",
|
| 30 |
+
"237561",
|
| 31 |
+
"208964",
|
| 32 |
+
"224308",
|
| 33 |
+
"227321",
|
| 34 |
+
"9823",
|
| 35 |
+
"9986",
|
| 36 |
+
"4577",
|
| 37 |
+
"330879",
|
| 38 |
+
"36329",
|
| 39 |
+
"99287",
|
| 40 |
+
"9615",
|
| 41 |
+
"39946",
|
| 42 |
+
"284593",
|
| 43 |
+
"3055",
|
| 44 |
+
"4081",
|
| 45 |
+
"170187",
|
| 46 |
+
"185431",
|
| 47 |
+
"4530",
|
| 48 |
+
"4536",
|
| 49 |
+
"243232",
|
| 50 |
+
"3880",
|
| 51 |
+
"5702",
|
| 52 |
+
"8364",
|
| 53 |
+
"10029",
|
| 54 |
+
"85962",
|
| 55 |
+
"510516",
|
| 56 |
+
"3847",
|
| 57 |
+
"8022",
|
| 58 |
+
"235443",
|
| 59 |
+
"4102",
|
| 60 |
+
"3888",
|
| 61 |
+
"4097",
|
| 62 |
+
"425011",
|
| 63 |
+
"9940"
|
| 64 |
+
]
|
| 65 |
+
}
|
| 66 |
+
}
|