Dataset Viewer
Auto-converted to Parquet Duplicate
synthetic
bool
records
int64
index
dict
files
dict
true
20,397,501
{ "containers": 24, "keys": 60053, "ids": 20397501, "blocks": 24029 }
{ "synthetic.hilum.idx": { "bytes": 238034166, "sha256": "4470608b4cbcc683120984460d3be313b41d258a44046683a94f3e14fd25da15" }, "synthetic_00.hilum.z": { "bytes": 57818885, "sha256": "56b533abedea58339fe08e5fc97250a89bf019ce6dd34fe127b7476e876815fc" }, "synthetic_01.hilum.z": { "bytes": 577...

SYNTHETIC DATA: Hilum hosting speed test

Every record in this repository is synthetic. Dataset ids are random UUIDs, gene ids (HGNC:n), cell-type ids (CL:n) and all statistics (lfc, se, z, pval, ...) are random numbers drawn from a seeded generator. There is no real biological, clinical or literature content, and the values must not be interpreted. Each record also carries the extension field x.synth.v=1, and the container header says kind=SYNTHETIC.

Purpose: measure remote (HTTP range) read latency, bytes per lookup and rate limiting of the Hilum container format on Hugging Face hosting.

Files: 24 containers synthetic_NN.hilum.z (one synthetic dataset each, ~20.4 M O-type records in total) and one index synthetic.hilum.idx. Format: HILUM/1.0 (registry 1.1.0). Generator: make_synthetic.py (seeded; same inputs give the same bytes).

Downloads last month
51