Chronos: Learning the Language of Time Series
Paper • 2403.07815 • Published • 50
Datasets:
start timestamp[s] | target list | selected_kernel_reprs list | kernel_formula string |
|---|---|---|---|
2000-01-01T00:00:00 | [
4.316741398840438,
4.317750022649842,
4.31773344695424,
4.3194360366768505,
4.3219560613447054,
4.3199777127621415,
4.320576267433709,
4.323314022342149,
4.321431205299354,
4.324957484548486,
4.324271868871547,
4.32721934921742,
4.326422232067006,
4.3270708880671425,
4.327865028910055,
... | [
"DotProduct(sigma_0=10)"
] | DotProduct(sigma_0=10) |
2000-01-01T00:00:00 | [
1.8949085823530083,
2.1751990658399127,
1.7731469702324476,
1.0150661314111475,
1.565121831215548,
2.0675495104740125,
1.8676113032764137,
-0.1682351111626221,
0.16310966711528035,
0.6397420696050901,
-0.25033853909880954,
-1.2218804779605688,
-0.36797311959512674,
-0.4912114593959253,
-... | [
"ExpSineSquared(length_scale=1, periodicity=0.00391)",
"ExpSineSquared(length_scale=1, periodicity=0.328)",
"ExpSineSquared(length_scale=1, periodicity=0.0391)",
"WhiteKernel(noise_level=0.1)",
"ExpSineSquared(length_scale=1, periodicity=0.0293)"
] | (ExpSineSquared(length_scale=1, periodicity=0.00391) + ExpSineSquared(length_scale=1, periodicity=0.328) + ExpSineSquared(length_scale=1, periodicity=0.0391) + WhiteKernel(noise_level=0.1)) * ExpSineSquared(length_scale=1, periodicity=0.0293) |
2000-01-01T00:00:00 | [
1.5599541606392293,
1.2898930020843837,
0.459537231536719,
-0.5032123940777543,
-1.1202115995955992,
-1.0319093176619667,
-0.6315142224491521,
0.11053978604151449,
1.0251681563065798,
1.5874298749117692,
2.0582263415985413,
2.305561826425127,
2.1910532239321787,
2.0790883228614745,
1.939... | [
"ExpSineSquared(length_scale=1, periodicity=0.0137)",
"ExpSineSquared(length_scale=1, periodicity=0.0469)",
"ExpSineSquared(length_scale=1, periodicity=0.0469)"
] | ExpSineSquared(length_scale=1, periodicity=0.0137) * ExpSineSquared(length_scale=1, periodicity=0.0469) + ExpSineSquared(length_scale=1, periodicity=0.0469) |
2000-01-01T00:00:00 | [
-1.224133183777016,
-1.2587340970933134,
-1.2923499839536472,
-1.3252271094090187,
-1.357733451129064,
-1.3895002405612273,
-1.4207508911377236,
-1.4512570828381257,
-1.4815609001030485,
-1.5113833447359752,
-1.5409529688063175,
-1.5698725518729257,
-1.5983634579156547,
-1.6267916966482017... | [
"ExpSineSquared(length_scale=1, periodicity=0.164)"
] | ExpSineSquared(length_scale=1, periodicity=0.164) |
2000-01-01T00:00:00 | [
-0.32293921419238103,
0.2425981771509,
-0.1699200294177358,
0.08195302540820418,
0.7891913826872556,
0.24285562382099743,
-0.3077261089862051,
2.4727708067406704,
0.6572780058461379,
0.1625500424210207,
-1.0105094847178273,
0.008820780927762228,
-0.5386809225672475,
1.4414013295961343,
1... | [
"ExpSineSquared(length_scale=1, periodicity=0.00977)",
"RBF(length_scale=1)",
"ExpSineSquared(length_scale=1, periodicity=0.00391)",
"ExpSineSquared(length_scale=1, periodicity=0.0508)"
] | ExpSineSquared(length_scale=1, periodicity=0.00977) * RBF(length_scale=1) * ExpSineSquared(length_scale=1, periodicity=0.00391) * ExpSineSquared(length_scale=1, periodicity=0.0508) |
2000-01-01T00:00:00 | [
0.7364310544735138,
0.9601238407854105,
2.577297076186168,
3.1486902861967563,
0.6905012521337868,
0.7705308618270537,
2.254674688685228,
2.679466000299456,
0.12782684965066538,
0.14730540849108975,
1.6083725838546852,
2.01545191614751,
-0.49571800087189455,
-0.41273109827943744,
1.12861... | [
"ExpSineSquared(length_scale=1, periodicity=0.0586)",
"ExpSineSquared(length_scale=1, periodicity=0.00391)"
] | ExpSineSquared(length_scale=1, periodicity=0.0586) + ExpSineSquared(length_scale=1, periodicity=0.00391) |
2000-01-01T00:00:00 | [
1.0419500294785349,
1.248954008152578,
1.5104189552586358,
1.7610037567280834,
1.8695095000242348,
1.7100942724120605,
1.2613637918750267,
0.6489575965293339,
0.08446211676377313,
-0.2613966792801148,
-0.34015109756070194,
-0.19931235653925183,
0.07718685768613753,
0.40224070501489506,
0... | [
"ExpSineSquared(length_scale=1, periodicity=0.0391)",
"ExpSineSquared(length_scale=1, periodicity=0.0254)"
] | ExpSineSquared(length_scale=1, periodicity=0.0391) + ExpSineSquared(length_scale=1, periodicity=0.0254) |
2000-01-01T00:00:00 | [
0.599105013873217,
0.7679336349596579,
-0.3207102818506242,
1.0330513771722525,
0.03804803807196591,
0.1802641520655753,
0.5220099543137287,
0.5797562515802385,
0.027957071742713807,
0.15259948775519758,
0.9664864599109879,
0.6958375537897602,
0.6866896595799351,
0.03734919201334719,
0.7... | [
"ExpSineSquared(length_scale=1, periodicity=0.0586)",
"WhiteKernel(noise_level=0.1)",
"ExpSineSquared(length_scale=1, periodicity=0.00977)",
"RationalQuadratic(alpha=10, length_scale=1)"
] | ExpSineSquared(length_scale=1, periodicity=0.0586) * WhiteKernel(noise_level=0.1) * ExpSineSquared(length_scale=1, periodicity=0.00977) + RationalQuadratic(alpha=10, length_scale=1) |
2000-01-01T00:00:00 | [
-3.270126230273103,
-2.2212058332650026,
-1.228234880520817,
-2.002628336377537,
-3.2075052410773566,
-3.1945019289236942,
-0.8145289580855892,
-0.5929864087769423,
-2.5533220111549753,
-2.6443596219881336,
-0.5026293670270809,
0.7586394265119236,
-1.035128714871872,
-0.7704639898485335,
... | [
"ExpSineSquared(length_scale=1, periodicity=0.0234)",
"ExpSineSquared(length_scale=1, periodicity=0.0469)",
"1**2",
"RBF(length_scale=10)",
"ExpSineSquared(length_scale=1, periodicity=0.00391)"
] | (ExpSineSquared(length_scale=1, periodicity=0.0234) * ExpSineSquared(length_scale=1, periodicity=0.0469) + 1**2 + RBF(length_scale=10)) * ExpSineSquared(length_scale=1, periodicity=0.00391) |
2000-01-01T00:00:00 | [-2.399248458246019,-0.42405380747998633,-0.038083002308141434,-0.7074062406137605,-1.11148360277107(...TRUNCATED) | ["ExpSineSquared(length_scale=1, periodicity=0.0254)","ExpSineSquared(length_scale=1, periodicity=0.(...TRUNCATED) | "(ExpSineSquared(length_scale=1, periodicity=0.0254) + ExpSineSquared(length_scale=1, periodicity=0.(...TRUNCATED) |
End of preview. Expand in Data Studio
KernelSynth (annotated)
One million synthetic univariate time series, each 1024 points long, drawn from a Gaussian process prior whose kernel is a random composition of up to five base kernels. This is the KernelSynth procedure from Chronos with one addition: the generating kernel is kept alongside each series. The ground-truth structure behind every series is therefore known, which makes the corpus usable for interpretability work rather than only for pretraining.
Fields
start— constant2000-01-01T00:00:00. The time index is arbitrary and carries no meaning.target— the series itself: 1024float64values sampled from the GP prior.selected_kernel_reprs— the base kernels drawn from the kernel bank, as scikit-learn reprs.kernel_formula— the composed kernel with explicit precedence, e.g.RBF(length_scale=1) * (DotProduct(sigma_0=1) + WhiteKernel(noise_level=1)). scikit-learn's own repr omits parentheses, so(a + b) * canda + b * care indistinguishable there; this field disambiguates them.
Generation
Produced by a script adapted from Chronos'
kernel-synth.py,
with these settings: 1,000,000 series, length 1024, at most 5 base kernels each, seed 1.
Kernels are drawn with replacement from the 33-entry Chronos kernel bank and combined
pairwise with random + / * operators. Each series draws from its own independent random
stream derived from the seed, so the corpus is reproducible and independent of worker count.
One deviation from upstream: a jitter of 1e-8 * mean(diag(cov)) is added to the covariance
diagonal to keep it numerically positive semi-definite, since composed kernels are often
ill-conditioned.
License and attribution
The data is released under CC-BY-4.0.
It was produced by a script adapted from Chronos' kernel-synth.py (Copyright Amazon.com,
Inc., Apache-2.0). That license covers the generator code, not the series it emits, and no
Chronos data is contained here — every series is sampled fresh from a GP prior. The method,
however, is theirs; please cite the Chronos (1) paper.
- Downloads last month
- -