Initial dataset release: TSFMI-Synthetic (11 configs, 60/20/20 splits, Croissant 1.0 + RAI metadata)

CROISSANT.json

@@ -0,0 +1,155 @@
+{
+  "@context": {
+    "@language": "en",
+    "@vocab": "https://schema.org/",
+    "citeAs": "cr:citeAs",
+    "column": "cr:column",
+    "conformsTo": "dct:conformsTo",
+    "cr": "http://mlcommons.org/croissant/",
+    "rai": "http://mlcommons.org/croissant/RAI/",
+    "data": {"@id": "cr:data", "@type": "@json"},
+    "dataType": {"@id": "cr:dataType", "@type": "@vocab"},
+    "dct": "http://purl.org/dc/terms/",
+    "examples": {"@id": "cr:examples", "@type": "@json"},
+    "extract": "cr:extract",
+    "field": "cr:field",
+    "fileProperty": "cr:fileProperty",
+    "fileObject": "cr:fileObject",
+    "fileSet": "cr:fileSet",
+    "format": "cr:format",
+    "includes": "cr:includes",
+    "isLiveDataset": "cr:isLiveDataset",
+    "jsonPath": "cr:jsonPath",
+    "key": "cr:key",
+    "md5": "cr:md5",
+    "parentField": "cr:parentField",
+    "path": "cr:path",
+    "recordSet": "cr:recordSet",
+    "references": "cr:references",
+    "regex": "cr:regex",
+    "repeated": "cr:repeated",
+    "replace": "cr:replace",
+    "sc": "https://schema.org/",
+    "separator": "cr:separator",
+    "source": "cr:source",
+    "subField": "cr:subField",
+    "transform": "cr:transform"
+  },
+  "@type": "Dataset",
+  "name": "TSFMI-Synthetic",
+  "description": "Synthetic time-series datasets with mathematically exact ground-truth labels for the TSFMI baseline-controlled evaluation protocol. Six standard properties (trend, seasonality, frequency, stationarity, anomaly, change point) plus five hard variants. Each dataset is generated on-demand by deterministic seeded generators in src/datasets/synthetic.py; this manifest documents the canonical seed=42 instantiation used for all confirmatory results in the paper. The TSFMI paper itself is the protocol/benchmark contribution; the datasets here exist only to instantiate the protocol and have no claim of independent novelty.",
+  "conformsTo": "http://mlcommons.org/croissant/1.0",
+  "license": "https://opensource.org/licenses/MIT",
+  "url": "https://anonymous.4open.science/r/tsfmi (anonymized for review)",
+  "version": "1.0.0",
+  "datePublished": "2026-05-06",
+  "isLiveDataset": false,
+  "citeAs": "TSFMI: A Baseline-Controlled Evaluation Protocol for Time-Series Foundation Model Representations. Anonymous submission to NeurIPS 2026 Evaluations & Datasets Track.",
+  "creator": {
+    "@type": "Organization",
+    "name": "Anonymous (NeurIPS 2026 ED Track double-blind submission)"
+  },
+
+  "rai:dataCollection": "Synthetic generation via deterministic Python procedures with NumPy random number generators. No human subjects, no scraping, no third-party data ingestion.",
+  "rai:dataCollectionType": "Synthetic / procedural",
+  "rai:dataPreprocessingProtocol": "None. Generators emit final tensors (N x seq_len) with labels directly.",
+  "rai:dataAnnotationProtocol": "Labels are derived analytically from generation parameters (slope sign for trend, period for seasonality, frequency bin for frequency, presence-of-cumsum for stationarity, presence-of-spike for anomaly, presence-of-shift for change-point). No human annotation.",
+  "rai:dataLimitations": "Each generator instantiates ONE family of synthetic constructions (e.g., the canonical anomaly task is a single 5-sigma spike on N(0,1) noise) and is therefore a statistically simple test bed. The kurtosis sufficiency of the canonical anomaly task is documented in the paper (sec:anomaly_mechanism); a 'realistic_anomaly' variant with mixed anomaly types is included to bound this limitation. Synthetic-to-real transfer is weak (paper sec 4.2).",
+  "rai:dataBiases": "By construction the synthetic data has no human / demographic content and therefore no demographic bias. The 'bias' that does exist is statistical: the canonical generators are deliberately simple, which is why baseline-controlled probing is the recommended interpretation lens.",
+  "rai:personalSensitiveInformation": "None. Datasets contain no personal, demographic, medical, or otherwise sensitive information.",
+  "rai:dataUseCases": "Probing internal representations of pre-trained time-series foundation models; calibrating new probing protocols against simple non-model baselines; comparing intervention diagnostics (LEACE, LDA steering, CKA) under matched controls.",
+  "rai:dataMisuseCases": "Not intended for deployment-time decision making, anomaly detection in production, or as a substitute for any real-world benchmark. Specifically, the canonical anomaly task should not be used to claim a model 'detects anomalies' in any operational sense; see paper sec 6 (Misuse Risks).",
+  "rai:dataSocialImpact": "The protocol is intended to raise the rigor bar on TSFM representation claims. Direct social impact is low (synthetic data, no human subjects); indirect impact is methodological -- discouraging over-claiming in time-series probing literature.",
+  "rai:syntheticDataIndicator": "All TSFMI-Synthetic data is fully synthetic; no real-world or human-derived signals.",
+
+  "distribution": [
+    {
+      "@type": "cr:FileObject",
+      "@id": "synthetic-generator-script",
+      "name": "synthetic.py",
+      "description": "Deterministic generator implementing all six standard and five hard-variant temporal-property datasets. All canonical paper artifacts use the seed=42 instantiation.",
+      "contentUrl": "src/datasets/synthetic.py",
+      "encodingFormat": "text/x-python",
+      "sha256": "computed-by-prepare_anonymous_release.sh"
+    },
+    {
+      "@type": "cr:FileObject",
+      "@id": "real-world-loader",
+      "name": "real_world.py",
+      "description": "Loaders for ETTh1, Weather (Jena Climate), Electricity, Traffic, Exchange-Rate. The TSFMI repository does NOT redistribute these datasets; loaders read user-supplied files from data/ under the original public licenses.",
+      "contentUrl": "src/datasets/real_world.py",
+      "encodingFormat": "text/x-python"
+    },
+    {
+      "@type": "cr:FileObject",
+      "@id": "manifest-json",
+      "name": "experiment_manifest.json",
+      "description": "Machine-countable provenance of every artifact on disk (canonical 60/20/20 cells, baseline cells, LEACE erasures, CKA matrices, intervention runs, hard-variant comparisons, layer probe runs). Auto-generated by scripts/count_experiments.py.",
+      "contentUrl": "outputs/paper/experiment_manifest.json",
+      "encodingFormat": "application/json"
+    }
+  ],
+
+  "recordSet": [
+    {
+      "@type": "cr:RecordSet",
+      "@id": "synthetic-trend",
+      "name": "synthetic_trend",
+      "description": "1000 sequences of length 512, three-class (up / down / flat). Labels are the slope-direction class, exact by construction.",
+      "field": [
+        {"@type": "cr:Field", "name": "sequence", "dataType": "sc:Float", "repeated": true},
+        {"@type": "cr:Field", "name": "label", "dataType": "sc:Integer", "description": "0=up, 1=down, 2=flat"}
+      ]
+    },
+    {
+      "@type": "cr:RecordSet",
+      "@id": "synthetic-seasonality",
+      "name": "synthetic_seasonality",
+      "description": "1000 sequences of length 512, regression target = period in {8,16,32,64}. Sinusoid + amplitude noise.",
+      "field": [
+        {"@type": "cr:Field", "name": "sequence", "dataType": "sc:Float", "repeated": true},
+        {"@type": "cr:Field", "name": "label", "dataType": "sc:Float", "description": "Period length (regression target)"}
+      ]
+    },
+    {
+      "@type": "cr:RecordSet",
+      "@id": "synthetic-frequency",
+      "name": "synthetic_frequency",
+      "description": "1000 sequences of length 512, eight-class frequency-band classification. f_k = (k+1)/seq_len.",
+      "field": [
+        {"@type": "cr:Field", "name": "sequence", "dataType": "sc:Float", "repeated": true},
+        {"@type": "cr:Field", "name": "label", "dataType": "sc:Integer", "description": "Frequency bin index 0..7"}
+      ]
+    },
+    {
+      "@type": "cr:RecordSet",
+      "@id": "synthetic-stationarity",
+      "name": "synthetic_stationarity",
+      "description": "1000 sequences of length 512, binary (stationary noise vs non-stationary random walk).",
+      "field": [
+        {"@type": "cr:Field", "name": "sequence", "dataType": "sc:Float", "repeated": true},
+        {"@type": "cr:Field", "name": "label", "dataType": "sc:Integer", "description": "0=stationary, 1=non-stationary"}
+      ]
+    },
+    {
+      "@type": "cr:RecordSet",
+      "@id": "synthetic-anomaly",
+      "name": "synthetic_anomaly",
+      "description": "1000 sequences of length 512, binary (normal vs +/- 5-sigma point spike). Single sufficient statistic = kurtosis (paper sec_anomaly_mechanism).",
+      "field": [
+        {"@type": "cr:Field", "name": "sequence", "dataType": "sc:Float", "repeated": true},
+        {"@type": "cr:Field", "name": "label", "dataType": "sc:Integer", "description": "0=normal, 1=anomaly"}
+      ]
+    },
+    {
+      "@type": "cr:RecordSet",
+      "@id": "synthetic-change-point",
+      "name": "synthetic_change_point",
+      "description": "1000 sequences of length 512, binary (no change-point vs mean+variance shift at midpoint).",
+      "field": [
+        {"@type": "cr:Field", "name": "sequence", "dataType": "sc:Float", "repeated": true},
+        {"@type": "cr:Field", "name": "label", "dataType": "sc:Integer", "description": "0=no_cp, 1=has_cp"}
+      ]
+    }
+  ]
+}

README.md

@@ -0,0 +1,234 @@
+---
+language:
+- en
+license: mit
+size_categories:
+- 1K<n<10K
+task_categories:
+- time-series-forecasting
+- tabular-classification
+- tabular-regression
+tags:
+- time-series
+- foundation-models
+- probing
+- benchmark
+- evaluation
+- synthetic
+- baseline-controlled
+- neurips-2026
+pretty_name: TSFMI-Synthetic
+configs:
+- config_name: trend
+  data_files:
+    - split: train
+      path: trend/train.parquet
+    - split: validation
+      path: trend/val.parquet
+    - split: test
+      path: trend/test.parquet
+- config_name: seasonality
+  data_files:
+    - split: train
+      path: seasonality/train.parquet
+    - split: validation
+      path: seasonality/val.parquet
+    - split: test
+      path: seasonality/test.parquet
+- config_name: frequency
+  data_files:
+    - split: train
+      path: frequency/train.parquet
+    - split: validation
+      path: frequency/val.parquet
+    - split: test
+      path: frequency/test.parquet
+- config_name: stationarity
+  data_files:
+    - split: train
+      path: stationarity/train.parquet
+    - split: validation
+      path: stationarity/val.parquet
+    - split: test
+      path: stationarity/test.parquet
+- config_name: anomaly
+  data_files:
+    - split: train
+      path: anomaly/train.parquet
+    - split: validation
+      path: anomaly/val.parquet
+    - split: test
+      path: anomaly/test.parquet
+- config_name: change_point
+  data_files:
+    - split: train
+      path: change_point/train.parquet
+    - split: validation
+      path: change_point/val.parquet
+    - split: test
+      path: change_point/test.parquet
+- config_name: trend_hard
+  data_files:
+    - split: train
+      path: trend_hard/train.parquet
+    - split: validation
+      path: trend_hard/val.parquet
+    - split: test
+      path: trend_hard/test.parquet
+- config_name: frequency_hard
+  data_files:
+    - split: train
+      path: frequency_hard/train.parquet
+    - split: validation
+      path: frequency_hard/val.parquet
+    - split: test
+      path: frequency_hard/test.parquet
+- config_name: stationarity_hard
+  data_files:
+    - split: train
+      path: stationarity_hard/train.parquet
+    - split: validation
+      path: stationarity_hard/val.parquet
+    - split: test
+      path: stationarity_hard/test.parquet
+- config_name: anomaly_hard
+  data_files:
+    - split: train
+      path: anomaly_hard/train.parquet
+    - split: validation
+      path: anomaly_hard/val.parquet
+    - split: test
+      path: anomaly_hard/test.parquet
+- config_name: change_point_hard
+  data_files:
+    - split: train
+      path: change_point_hard/train.parquet
+    - split: validation
+      path: change_point_hard/val.parquet
+    - split: test
+      path: change_point_hard/test.parquet
+---
+
+# TSFMI-Synthetic
+
+Synthetic time-series datasets with **mathematically exact ground-truth labels**
+for the TSFMI baseline-controlled probing protocol.
+
+> Companion data for the NeurIPS 2026 Evaluations & Datasets Track submission
+> *"TSFMI: A Baseline-Controlled Evaluation Protocol for Time-Series Foundation
+> Model Representations."* The code (anonymous) lives at
+> https://github.com/evaldataset/TSFMI.
+
+## Why this dataset exists
+
+Probing time-series foundation models (TSFMs) is hard because high probe
+accuracy may reflect probe capacity rather than encoded knowledge. TSFMI
+addresses this with **explicit non-model controls** (hand-crafted features,
+raw signal, random projection, ROCKET) evaluated under the same canonical
+60/20/20 split, the same sklearn estimator, the same 5 seeds, and the same
+bootstrap CI as the model probe. The synthetic datasets in this repository
+provide six standard temporal properties **plus five hard variants** under
+which TSFM representation claims can be evaluated against the controls.
+
+## Configurations (11 tasks)
+
+| Config | Type | Classes / Range | Description |
+|---|---|---|---|
+| `trend` | classification | 3 (up / down / flat) | linear slope ± noise |
+| `seasonality` | regression | period ∈ {8, 16, 32, 64} | sinusoid + noise; label = period |
+| `frequency` | classification | 8 frequency bins | discrete sinusoidal frequency bands |
+| `stationarity` | classification | 2 (stationary / non-stationary) | Gaussian noise vs. random walk |
+| `anomaly` | classification | 2 (normal / anomaly) | single ±5σ point spike — **kurtosis is a sufficient statistic** |
+| `change_point` | classification | 2 (none / has-CP) | mean+variance shift at midpoint |
+| `*_hard` (5) | classification | varies | structured background + subtle anomalies / weak slopes / overlapping harmonics / variance drift / mild CP |
+
+Each config has 1000 sequences of length 512, partitioned 60/20/20
+(train/val/test = 600/200/200). All data is generated deterministically with
+`seed=42` (data) and `split_seed=0` (partition), matching the canonical
+artefacts in the paper.
+
+## Quick start
+
+```python
+from datasets import load_dataset
+
+ds = load_dataset("evaldataset/TSFMI", "anomaly")
+print(ds)
+# DatasetDict({
+#   train: 600, validation: 200, test: 200
+# })
+print(ds["train"][0]["sequence"][:8], ds["train"][0]["label"])
+```
+
+Each row contains:
+
+- `sequence`: list[float] of length 512 (the time-series window)
+- `label`: int (classification) or float (seasonality regression)
+- `seed`: int (data-generation seed; always 42 here)
+- `split_seed`: int (split seed used to partition this row; 0 for the canonical artefacts)
+
+## Reproduction recipe
+
+The same 60/20/20 + 5-seed + bootstrap protocol used in the paper is
+implemented in [the GitHub repository](https://github.com/evaldataset/TSFMI):
+
+```bash
+git clone https://github.com/evaldataset/TSFMI.git && cd TSFMI
+python -m venv .venv && source .venv/bin/activate
+pip install -r requirements.txt && pip install -e ".[dev]"
+
+# CPU smoke test (<10 min) — reproduces the headline HC anomaly = 0.858 cell
+make smoke
+
+# Full canonical pipeline (~48 A100-hours)
+make extract-representations
+make reproduce-all
+```
+
+## Headline result reproducible from this dataset
+
+Under the canonical 60/20/20 protocol with sklearn `LogisticRegression` and a
+bootstrap 95% CI over 5 seeds:
+
+- An 8-D **hand-crafted feature vector** reaches **0.858** test accuracy on
+  the canonical `anomaly` configuration.
+- A **single kurtosis feature** alone reaches **0.859**.
+- A **single max-absolute-magnitude feature** alone reaches **0.907**.
+- The best of seven pre-trained TSFMs reaches **0.753** (TimesFM); all other
+  TSFMs (MOMENT, Chronos, PatchTST, GPT4TS, Timer, Moirai) score 0.50–0.73.
+
+This **inversion** is what motivates the baseline-controlled discipline of
+TSFMI and is documented in §3.3 and Appendix A.8 of the paper.
+
+## License
+
+MIT. The synthetic data is fully procedurally generated; there is no
+human-derived or scraped content. Real-world datasets used elsewhere by the
+TSFMI evaluation pipeline (ETTh1, Weather, Electricity, Traffic, Exchange
+Rate, UCR) are **not redistributed here** and remain under their respective
+original licenses; see the GitHub `LICENSE` file for details.
+
+## Citation
+
+```bibtex
+@misc{tsfmi2026,
+  title = {{TSFMI}: A Baseline-Controlled Evaluation Protocol for Time-Series Foundation Model Representations},
+  author = {Anonymous Authors},
+  howpublished = {Anonymous submission to the NeurIPS 2026 Evaluations \& Datasets Track},
+  year = {2026},
+  url = {https://github.com/evaldataset/TSFMI}
+}
+```
+
+## Responsible AI notes
+
+- **Data collection**: synthetic procedural generation, no human subjects, no scraping.
+- **Limitations**: each generator instantiates one statistically simple test bed; the canonical anomaly task is kurtosis-trivial by design (`realistic_anomaly` variant in the GitHub repo bounds this).
+- **Biases**: none — purely synthetic with no demographic content.
+- **Personal/sensitive info**: none.
+- **Use cases**: probing TSFM internal representations, calibrating new probing protocols against simple non-model baselines.
+- **Misuse cases**: not intended as a production anomaly detector; do not interpret the canonical anomaly task as evidence that any model "detects anomalies" in any operational sense.
+- **Synthetic data indicator**: 100% synthetic.
+
+The full Croissant 1.0 + RAI metadata is provided as `CROISSANT.json` in this
+repository.

anomaly/test.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a233899d828d3652bad17880d9386c4fa2e927d24606903fa2b6800edce59064
+size 630251

anomaly/train.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2d266c28a9b4e6fe2afcd01007556b8816576a5bec88d1231473798f466872cb
+size 1855835

anomaly/val.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d1d7a613f4fbd87286427dc627d2a6df70e27b1a2dfe02b04312c2fdc43a1b2b
+size 630195

anomaly_hard/test.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b6bb9e57b3b1e1b28d3c1b20925015ed4fdc3ee57cd1c9c39a23ea02ebee2257
+size 630215

anomaly_hard/train.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:01f1b73a03ae98bb66f2301289af0430d7e32e509c6c58770cd77304a90593b0
+size 1855405

anomaly_hard/val.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b86e80e32b2ce852e7b1ae450b8bbbb02fd26ba6d765d28ea299831b30dc84dd
+size 630167

change_point/test.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:45856dd7b69cd89132e28de8fec136c401855defd3eca178bcd5975611f6f569
+size 630299

change_point/train.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:47ef9b91174512ad033e129d98277921c7fb2e916a0b00a86459bbe23def9414
+size 1856127

change_point/val.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cb7b8a3fb2129a5f8a533df160cc71e25fa008d1b0a894e65c19ff8e715f7b5c
+size 630251

change_point_hard/test.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8c7169eaea0ea7f1fb00e65979d16f7cd42ab8f9322b9b5662377f02a5385710
+size 630239

change_point_hard/train.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9e5f0b1bdd5d83520ebb24cc0ad871eccf4e23a084812c4f8379194f4fa8b5ee
+size 1855985

change_point_hard/val.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:19fa03368e951fe1fac422dbd786bf04e2695af2e71d9bef9cd1cb597dc6bcab
+size 630251

frequency/test.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:617a67707c2a3919acbde1fab5be0900e93c00e509ad6b21cd5ffc419af28890
+size 630237

frequency/train.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a14864629c21a3dcca9a6757a52263ef7e4a740cbbf3b94e9ad8f7c1a143f264
+size 1855571

frequency/val.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bf9ea474317816cc30f214d712ad6435abd789226854e6bb8ca419a813e946d5
+size 630237

frequency_hard/test.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5041f06c28fe7ead7a1015f2c63d2985b6c740409385cf5f495978155896abd8
+size 630265

frequency_hard/train.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:716ea29f710ac330897c2ba3c2e43c7b9ef7462659e9e761fd9ea626b55fef98
+size 1856332

frequency_hard/val.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ab527181fc625a9ec3ac8f0f320997c0e0bd254dc6907966c5a449e275e68695
+size 630399

seasonality/test.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5eb5469333f5a6f367108d7fce3f3e6b832243c468ac3d1fe2646235ae3567c3
+size 630236

seasonality/train.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5b18955bff209c078091a94465dde2debde6e125b59788e8efd323b685a15db9
+size 1855606

seasonality/val.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b019acc978717d145af29b570e4e1818850875b267b29a96d332548a92349215
+size 630212

stationarity/test.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b586f1e6af7b2ac5f3c880eb5e311e0f0078a82800c877a9678daa9a0f99ceb0
+size 630407

stationarity/train.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8df030ed9ac46d0388c8ab631df08297cc8abd320baf206c83d2a3065e5d0761
+size 1856715

stationarity/val.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ea3aee730bc65c2c3d7b2742bdc1a507ca10ee89cb91a24152a3c322d30e9aa2
+size 630323

stationarity_hard/test.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9a384267a61a63bf06efa7198d94a1fef72134d4e7e59095f1de82a343992786
+size 630231

stationarity_hard/train.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a2027120e637442ea8cc63fa0562e2698b252edc8d58afa0a0fcd6a6f851c8bc
+size 1856005

stationarity_hard/val.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b8ccc6891a01c3041a96535b37698d91702662f98b0c330effb5fa285c6968c7
+size 630183

trend/test.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fe3b3399968d1df8fface6141df9de214342746d285c29f5fa9b7b7a88ca78d1
+size 630394

trend/train.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:65e32d9ca211d08d2cd0c727c1954b81d9563f6f5b4a87b69a8efc036f8efad3
+size 1856846

trend/val.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:76b6587aa9cec4f9fd2e2af5527ee7ae62bb7431fab928673204a51b22d1c7d8
+size 630404

trend_hard/test.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:989c6545ad476a646a61a4da91efcd2f228b54bce262b6582da4e2d0c17c45ae
+size 630270

trend_hard/train.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a0688079abafb38ac2e4f76bd21a42288596bed18dbd95e2fac1e5e6a944d99a
+size 1855988

trend_hard/val.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a16ae25cbae7a8ac2ef384c2efb41fbd1d4d0493024feafeeb4bede60ad9c327
+size 630280