Dataset Viewer
Auto-converted to Parquet Duplicate
record_id
stringlengths
36
36
raw_scan
listlengths
2k
2k
absorbance_clean
listlengths
2k
2k
concentration_ppm
float64
1
1,000
temperature_K
float64
291
302
pressure_atm
float64
0.94
1.06
path_length_m
float64
10
10
instrument_config_id
stringclasses
3 values
technique
stringclasses
1 value
000a87cd-c1b9-4992-ad3a-7529ac802545
[ 0.8765567765567766, 0.8761904761904762, 0.8769230769230769, 0.8769230769230769, 0.8772893772893773, 0.8772893772893773, 0.8776556776556776, 0.8776556776556776, 0.8776556776556776, 0.8776556776556776, 0.878021978021978, 0.878021978021978, 0.878021978021978, 0.8787545787545787, 0.878754578...
[ 0.000009020120152750933, 0.000009068392416707924, 0.000009117093114556238, 0.000009166227419163562, 0.000009215800580188249, 0.000009265817926319024, 0.000009316284866040623, 0.000009367206889145701, 0.000009418589568280418, 0.000009470438560406763, 0.000009522759608056004, 0.00000957555854113...
2.120287
296.759238
1.005882
10
vi-da-hard-03
TDLAS-DA
000f1f9b-255a-4a6f-8349-eb03d54f1ade
[ 0.933894891045596, 0.933894891045596, 0.933894891045596, 0.9340780076908991, 0.9341695660135506, 0.9340780076908991, 0.9341695660135506, 0.9343526826588536, 0.9342611243362021, 0.9345357993041568, 0.9344442409815052, 0.9346273576268083, 0.9343526826588536, 0.9343526826588536, 0.934627357...
[ 0.0001803624792299031, 0.00018101220417400242, 0.00018166587018052035, 0.00018232351114766133, 0.00018298516134981156, 0.00018365085544141423, 0.00018432062846148153, 0.00018499451584057738, 0.0001856725534024852, 0.0001863547773755978, 0.00018704122439171036, 0.00018773193149516416, 0.00018...
109.401598
296.157507
1.001064
10
vi-da-medium-02
TDLAS-DA
000fe89d-1a54-4438-b2c7-ffb459011d88
[0.8263736263736263,0.8263736263736263,0.827106227106227,0.8267399267399267,0.827106227106227,0.8271(...TRUNCATED)
[0.0009174390013642559,0.0009206307378937671,0.0009238411283034605,0.0009270703274849912,0.000930318(...TRUNCATED)
605.574673
295.810098
1.000899
10
vi-da-hard-03
TDLAS-DA
0016d722-9248-40fc-ae5e-2ca88869ac40
[0.8184398461820179,0.8184398461820179,0.8188060794726241,0.8188060794726241,0.8189891961179271,0.81(...TRUNCATED)
[0.0002221961493866444,0.0002230186830584455,0.000223846329551785,0.000224679133877194,0.00022551714(...TRUNCATED)
127.809509
295.967673
1.002491
10
vi-da-medium-02
TDLAS-DA
00265efb-c916-4867-a814-afe16620e1ca
[0.8706111238269627,0.870771343556878,0.8708628976882582,0.8710002288853285,0.8711604486152438,0.871(...TRUNCATED)
[0.000027158756602343585,0.0000272504925247365,0.000027342746183428614,0.00002743552173563039,0.0000(...TRUNCATED)
19.736595
295.965095
1.000309
10
vi-da-easy-01
TDLAS-DA
00362963-fbd1-44f9-9fb5-57729a60eea8
[0.9120622568093385,0.9121538109407187,0.9122682536049439,0.9124284733348592,0.9125429159990844,0.91(...TRUNCATED)
[0.00024093365649305955,0.00024195822801098337,0.0002429901227583496,0.00024402941405718604,0.000245(...TRUNCATED)
100.915905
295.971749
0.999804
10
vi-da-easy-01
TDLAS-DA
0055f0b2-46a5-48dc-8623-53c29fc66d56
[0.865644312199588,0.8658274204623484,0.8659418631265736,0.8661020828564889,0.8662623025864042,0.866(...TRUNCATED)
[4.756411745474711e-6,4.774508057805809e-6,4.792720387786374e-6,4.811049786660144e-6,4.8294973179373(...TRUNCATED)
2.530695
295.97953
0.99956
10
vi-da-easy-01
TDLAS-DA
008e793e-5366-4284-a3a0-a0dc2cdd2b69
[0.7941391941391941,0.7945054945054945,0.7948717948717948,0.7948717948717948,0.7952380952380952,0.79(...TRUNCATED)
[0.00011850139473626499,0.00011897700159790023,0.00011945583039751,0.00011993791185981669,0.00012042(...TRUNCATED)
56.427054
295.247627
0.987078
10
vi-da-hard-03
TDLAS-DA
00a00661-5fbc-44db-9505-81248877b3f4
[0.8817066471342245,0.881798205456876,0.881981322102179,0.8821644387474822,0.8822559970701337,0.8824(...TRUNCATED)
[0.000010954369200800332,0.000011003071420508345,0.000011052136116780582,0.000011101567054729169,0.0(...TRUNCATED)
4.252664
296.11844
0.996738
10
vi-da-medium-02
TDLAS-DA
00afd115-5796-451a-8af6-2184fb1e5b19
[0.8552462918879326,0.8554294085332357,0.8555209668558872,0.8556125251785387,0.8556125251785387,0.85(...TRUNCATED)
[0.00017790767788301645,0.00017848504529195146,0.00017906555075318542,0.00017964921855824414,0.00018(...TRUNCATED)
136.378085
294.815911
1.005059
10
vi-da-medium-02
TDLAS-DA
End of preview. Expand in Data Studio

SPEKTRAN CH4 TDLAS v0

Official benchmark splits of SPEKTRAN -- simulation engines, data standards, and ML benchmarks for physical sensing. This dataset: CH4 TDLAS spectra near 6047 cm-1 (1653 nm, 2v3 band) with literature-anchored instrument noise (etalon fringes, 1/f noise, baseline drift, scan nonlinearity, ADC quantization).

Configs

da (default) -- Direct Absorption

  • train / validation / test: mixture of three virtual instruments (easy / medium / hard difficulty tiers)
  • test_heldout_instrument: records from a held-out virtual instrument whose parameter ranges are excluded from the training distribution -- the cross-instrument generalization track (T3)

Columns: record_id, raw_scan, absorbance_clean, concentration_ppm, temperature_K, pressure_atm, path_length_m, instrument_config_id, technique.

wms -- Wavelength Modulation Spectroscopy (v0.2)

  • t4_train / t4_validation / t4_test: WMS records with 1f/2f lock-in demodulation signals for the T4 benchmark task (2f concentration regression)

Columns: same as da plus demod_1f, demod_2f.

Load

from datasets import load_dataset

# Direct absorption (T1/T2/T3 tasks)
ds_da = load_dataset("spektran/spektran-ch4-v0", "da")
x = ds_da["train"][0]["raw_scan"]            # 2000-point noisy scan
y = ds_da["train"][0]["concentration_ppm"]   # regression target

# WMS (T4 task)
ds_wms = load_dataset("spektran/spektran-ch4-v0", "wms")
x_2f = ds_wms["t4_train"][0]["demod_2f"]     # 2f demod signal

Benchmark

Tasks, rules, metrics, and the leaderboard live in the SPEKTRAN repository. Baselines: ridge T1 MAE 2.84 ppm (T3 degradation 1.31x), 1D CNN 15.58 ppm (1.82x).

Fully reproducible: every record was generated from pinned configs + seeds; the repository regenerates this dataset bit-for-bit (spektran generate, configs under configs/datasets/). All data is simulation-born (data_origin: simulated); the built-in approximate CH4 line list is used in v0 (HITRAN-fetched line data planned for v0.2.x).

License: CC BY 4.0 (data). Engine code: Apache-2.0.

Downloads last month
48