objectID int64 34 955k | embedding listlengths 1.54k 1.54k | model stringclasses 1
value | dim int32 1.54k 1.54k |
|---|---|---|---|
34 | [
-0.015000015497207642,
-0.03246394917368889,
0.04740661382675171,
-0.029832636937499046,
-0.0023748232051730156,
-0.0004907283582724631,
0.021761951968073845,
0.007240398786962032,
0.002757540438324213,
0.015156382694840431,
0.019399430602788925,
-0.008126494474709034,
0.028739720582962036,
... | facebook/dinov2-giant | 1,536 |
35 | [
0.030073726549744606,
-0.03504054993391037,
0.030260572209954262,
-0.021844984963536263,
-0.000019792551029240713,
-0.026528554037213326,
0.036097828298807144,
-0.002476716646924615,
-0.016853952780365944,
0.058789875358343124,
-0.012013892643153667,
0.015755172818899155,
-0.0160311385989189... | facebook/dinov2-giant | 1,536 |
37 | [
0.0021475788671523333,
-0.005105253309011459,
0.0003699629451148212,
-0.011134867556393147,
0.041487500071525574,
-0.006267614662647247,
0.013543154112994671,
0.009363535791635513,
-0.02907172404229641,
0.01043179351836443,
-0.00699949637055397,
-0.008323258720338345,
-0.004759705625474453,
... | facebook/dinov2-giant | 1,536 |
38 | [
-0.014441954903304577,
0.003978465683758259,
-0.006518885958939791,
0.0026985148433595896,
0.03625979274511337,
-0.003463220549747348,
-0.0047453222796320915,
0.0032913105096668005,
-0.06100892275571823,
0.008353745564818382,
0.014318849891424179,
0.004067758098244667,
-0.02258506789803505,
... | facebook/dinov2-giant | 1,536 |
39 | [
-0.0027346082497388124,
-0.017601534724235535,
0.04004378244280815,
-0.015929512679576874,
-0.008042578585445881,
0.012444071471691132,
-0.030966956168413162,
0.03208547085523605,
-0.012940593995153904,
0.0019041728228330612,
-0.038032595068216324,
0.04619012027978897,
-0.06235405430197716,
... | facebook/dinov2-giant | 1,536 |
40 | [
0.01896507665514946,
-0.007771402597427368,
0.018131855875253677,
0.005951201543211937,
0.01624465174973011,
-0.06811412423849106,
0.0063443053513765335,
0.013624629937112331,
-0.025779347866773605,
0.020201021805405617,
0.03365599736571312,
0.03923095017671585,
-0.011883814819157124,
0.01... | facebook/dinov2-giant | 1,536 |
41 | [0.019559921696782112,-0.0221252478659153,0.014566266909241676,-0.0035910513252019882,0.011819053441(...TRUNCATED) | facebook/dinov2-giant | 1,536 |
42 | [0.03299759700894356,-0.05582359433174133,-0.015322767198085785,0.0034852405078709126,-0.03806171193(...TRUNCATED) | facebook/dinov2-giant | 1,536 |
43 | [0.03299759700894356,-0.05582359805703163,-0.015322756953537464,0.003485259599983692,-0.038061685860(...TRUNCATED) | facebook/dinov2-giant | 1,536 |
44 | [0.03635162487626076,-0.023703251034021378,-0.02481517568230629,0.008340138010680676,-0.022919148206(...TRUNCATED) | facebook/dinov2-giant | 1,536 |
End of preview. Expand in Data Studio
metmuseum/openaccess-embeddings-dinov2-giant
Image embeddings for every public-domain artwork in metmuseum/openaccess, produced by facebook/dinov2-giant.
| Column | Type | Notes |
|---|---|---|
objectID |
int64 | Primary key — matches objectID in metmuseum/openaccess |
embedding |
list<float32> | L2-normalised, dim = 1536 |
model |
string | Source model id |
dim |
int32 | Embedding dimension |
Image bytes are not stored here; join against the main dataset to recover them.
Embedding spec: dim=1536, expected image size=224px.
Joining with the main dataset
from datasets import load_dataset
meta = load_dataset("metmuseum/openaccess", split="train")
emb = load_dataset("metmuseum/openaccess-embeddings-dinov2-giant", split="train")
# Build an objectID -> embedding lookup, then attach to the metadata rows.
lookup = {r["objectID"]: r["embedding"] for r in emb}
joined = meta.map(lambda r: {"embedding": lookup.get(r["objectID"])})
print(joined[0].keys())
Nearest-neighbour example
import numpy as np
from datasets import load_dataset
emb = load_dataset("metmuseum/openaccess-embeddings-dinov2-giant", split="train")
ids = np.array(emb["objectID"])
mat = np.array(emb["embedding"], dtype=np.float32) # already L2-normalised
query = mat[0]
scores = mat @ query
top = np.argsort(-scores)[:10]
print(list(zip(ids[top].tolist(), scores[top].tolist())))
Generated by et-openaccess-embeddings.
- Downloads last month
- 215