Repository navigation
Expand file tree
/
Copy pathtask.json
More file actions
50 lines (50 loc) · 2.87 KB
/
Copy pathtask.json
File metadata and controls
50 lines (50 loc) · 2.87 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
{
"task_version": "1.0.0",
"code": "environment/workload/task.py",
"contract": "Quantile results: exact shape/dtype; rtol=atol=8*eps(float64), equal NaNs. Suite units: exact canonical results, except interp-equalize (absolute 8*eps(float64)) and loadtxt-csv (equal NaNs). Preserve input buffers.",
"core_counts": [
1,
2,
4
],
"development_entrypoint": "sh /app/workload/build.sh && /app/build/run",
"example_count": 28,
"id": "numpy-quantile",
"interface": {
"documentation": ["https://numpy.org/doc/stable/reference/"],
"documented_api": [],
"extension_points": []
},
"input_manifest": "environment/workload/inputs/manifest.json",
"input_visibility": "public_dev",
"job": "Compute linear quantiles over resident arrays, plus six timed numeric, time, random and I/O units over real data. Two further batches (strings-text, held-out only, and genfromtxt-catalog) are correctness-only.",
"language": "Python / C / C++",
"measurement_boundary": "Trusted external dispatch-to-stopped-worker timing includes the complete call batch, all copying/selection/sorting/interpolation, output materialization and result-address metadata. Input transfers and host numerical checking are outside; old outputs are released during the next timed batch. Suite units use the same resident boundary: the complete call batch and materialized results are timed; loading, argument derivation, canonical ordering, serialization and comparison are not.",
"name": "NumPy",
"operations": [
"np.quantile(a, q, axis=axis, method=\"linear\", keepdims=keepdims, overwrite_input=False)",
"np.argsort(a, kind=\"stable\")",
"np.argsort(datetime64[ms])",
"np.lexsort(keys)",
"np.searchsorted(edges, values, side)",
"np.searchsorted(np.unique(labels), labels)",
"np.unique_values(a)",
"np.unique_inverse(a)",
"np.unique(X, return_inverse=True, axis=0)",
"np.interp(x, xp, fp)",
"np.busday_count / np.busday_offset / np.is_busday(busdaycal=...)",
"np.random.Generator(PCG64).binomial/gamma/poisson/normal",
"np.strings.{strip,str_len,find,count,replace,partition,startswith,endswith,isalpha,expandtabs}",
"np.genfromtxt(text, delimiter=\",\", dtype=structured)",
"np.loadtxt(text, delimiter=\",\", dtype=float64)",
"str.astype(\"datetime64[ms]\"), timedelta arithmetic, datetime64.astype(\"S\")",
"float64/float32 .astype(str)"
],
"packaging_status": "public-task-private-judge",
"source_directory": "numpy",
"status": "qualification-in-progress",
"timing": "Shared shim resident stop protocol; local execution reports no timing.",
"title": "Compute quantiles and related NumPy operations over real data",
"readme_notes": "Seven equally weighted timed units per scope: the linear-quantile batch and six suite units. The strings-text (held-out only) and genfromtxt-catalog batches are correctness-only checks.",
"version": "2.5.4"
}