Hugging Face's logo

Datasets:
groundtruth-api
/
groundtruth-us-tariff-2026

Languages:
English
Size:
n<1K
Tags:
tariffs
customs
international-trade
landed-cost
united-states
economics
License:
Dataset card Files
xet
 Community
Dataset Viewer
Duplicate
The dataset viewer is not available for this split.
Cannot load the dataset split (in streaming mode) to extract the first rows.
Error code:   StreamingRowsError
Exception:    CastError
Message:      Couldn't cast
id: string
text: string
category_id: string
category: string
origin_code: string
origin: string
as_of: timestamp[s]
rate_layers: struct<mfn: double, forced_labor: double, section_301: double, section_232: double>
  child 0, mfn: double
  child 1, forced_labor: double
  child 2, section_301: double
  child 3, section_232: double
total_duty_rate: double
typical_hts: null
keywords: list<item: string>
  child 0, item: string
attribution: struct<source: string, source_url: string, api: string, license: string, as_of: timestamp[s]>
  child 0, source: string
  child 1, source_url: string
  child 2, api: string
  child 3, license: string
  child 4, as_of: timestamp[s]
fees: struct<mpf_rate: double, mpf_min: double, mpf_max: double, mpf_next: struct<effective: timestamp[s], (... 131 chars omitted)
  child 0, mpf_rate: double
  child 1, mpf_min: double
  child 2, mpf_max: double
  child 3, mpf_next: struct<effective: timestamp[s], mpf_min: double, mpf_max: double, note: string>
      child 0, effective: timestamp[s]
      child 1, mpf_min: double
      child 2, mpf_max: double
      child 3, note: string
  child 4, hmf_rate: double
  child 5, hmf_ocean_only: bool
  child 6, typical_broker_fee: int64
  child 7, notes: string
to
{'id': Value('string'), 'text': Value('string'), 'as_of': Value('timestamp[s]'), 'fees': {'mpf_rate': Value('float64'), 'mpf_min': Value('float64'), 'mpf_max': Value('float64'), 'mpf_next': {'effective': Value('timestamp[s]'), 'mpf_min': Value('float64'), 'mpf_max': Value('float64'), 'note': Value('string')}, 'hmf_rate': Value('float64'), 'hmf_ocean_only': Value('bool'), 'typical_broker_fee': Value('int64'), 'notes': Value('string')}, 'attribution': {'source': Value('string'), 'source_url': Value('string'), 'api': Value('string'), 'license': Value('string'), 'as_of': Value('timestamp[s]')}}
because column names don't match
Traceback:    Traceback (most recent call last):
                File "/src/services/worker/src/worker/utils.py", line 147, in get_rows_or_raise
                  return get_rows(
                      dataset=dataset,
                  ...<4 lines>...
                      column_names=column_names,
                  )
                File "/src/libs/libcommon/src/libcommon/utils.py", line 272, in decorator
                  return func(*args, **kwargs)
                File "/src/services/worker/src/worker/utils.py", line 127, in get_rows
                  rows_plus_one = list(itertools.islice(safe_iter(ds, dataset=dataset), rows_max_number + 1))
                File "/src/services/worker/src/worker/utils.py", line 483, in safe_iter
                  yield from ds.decode(False) if ds.features else ds
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2840, in __iter__
                  for key, example in ex_iterable:
                                      ^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2373, in __iter__
                  for key, pa_table in self._iter_arrow():
                                       ~~~~~~~~~~~~~~~~^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2398, in _iter_arrow
                  for key, pa_table in self.ex_iterable._iter_arrow():
                                       ~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 536, in _iter_arrow
                  for key, pa_table in iterator:
                                       ^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 419, in _iter_arrow
                  for key, pa_table in self.generate_tables_fn(**gen_kwags):
                                       ~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 343, in _generate_tables
                  self._cast_table(pa_table, json_field_paths=json_field_paths),
                  ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 132, in _cast_table
                  pa_table = table_cast(pa_table, self.info.features.arrow_schema)
                File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2378, in table_cast
                  return cast_table_to_schema(table, schema)
                File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2306, in cast_table_to_schema
                  raise CastError(
                  ...<3 lines>...
                  )
              datasets.table.CastError: Couldn't cast
              id: string
              text: string
              category_id: string
              category: string
              origin_code: string
              origin: string
              as_of: timestamp[s]
              rate_layers: struct<mfn: double, forced_labor: double, section_301: double, section_232: double>
                child 0, mfn: double
                child 1, forced_labor: double
                child 2, section_301: double
                child 3, section_232: double
              total_duty_rate: double
              typical_hts: null
              keywords: list<item: string>
                child 0, item: string
              attribution: struct<source: string, source_url: string, api: string, license: string, as_of: timestamp[s]>
                child 0, source: string
                child 1, source_url: string
                child 2, api: string
                child 3, license: string
                child 4, as_of: timestamp[s]
              fees: struct<mpf_rate: double, mpf_min: double, mpf_max: double, mpf_next: struct<effective: timestamp[s], (... 131 chars omitted)
                child 0, mpf_rate: double
                child 1, mpf_min: double
                child 2, mpf_max: double
                child 3, mpf_next: struct<effective: timestamp[s], mpf_min: double, mpf_max: double, note: string>
                    child 0, effective: timestamp[s]
                    child 1, mpf_min: double
                    child 2, mpf_max: double
                    child 3, note: string
                child 4, hmf_rate: double
                child 5, hmf_ocean_only: bool
                child 6, typical_broker_fee: int64
                child 7, notes: string
              to
              {'id': Value('string'), 'text': Value('string'), 'as_of': Value('timestamp[s]'), 'fees': {'mpf_rate': Value('float64'), 'mpf_min': Value('float64'), 'mpf_max': Value('float64'), 'mpf_next': {'effective': Value('timestamp[s]'), 'mpf_min': Value('float64'), 'mpf_max': Value('float64'), 'note': Value('string')}, 'hmf_rate': Value('float64'), 'hmf_ocean_only': Value('bool'), 'typical_broker_fee': Value('int64'), 'notes': Value('string')}, 'attribution': {'source': Value('string'), 'source_url': Value('string'), 'api': Value('string'), 'license': Value('string'), 'as_of': Value('timestamp[s]')}}
              because column names don't match

Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.

GroundTruth — US Import Tariffs & Landed Cost (2026)

A small, clean, RAG/LLM-ready fact corpus of 2026 United States import tariffs. Each record is a single self-contained fact: for one product category and one origin country, it gives the stacked duty rate (MFN + forced-labor economic rate + Section 301 + Section 232) in both human-readable text and structured fields, plus MPF/HMF mechanics.

Maintained data, current as of 2026-10-01.

Why this dataset exists

Tariff answers given by general models are frequently out of date or flatten several independent duty layers into one guess. This corpus is structured for retrieval and for grounding answers: every record is paragraph-level extractable and carries its as-of date and sources.

Files

  • tariff_facts.jsonl — 65 records, one per product category × origin (CN, VN, IN, MX, other). Each has:
    • text — a standalone factual sentence (ready to drop into a RAG context),
    • rate_layers — mfn, forced_labor, section_301, section_232,
    • total_duty_rate, category, origin, keywords, typical_hts,
    • as_of and attribution.
  • fees_fact.json — MPF and HMF rates, limits and the 2026-10-01 change.
  • tariff_table_snapshot.json — the full canonical table.

Load it 

import json
records = [json.loads(line) for line in open("tariff_facts.jsonl")]

Example

As of 2026-10-01, importing general merchandise from China (CN) into the United States carries an estimated total duty rate of 41.0%, stacked as: MFN 3.5%, forced-labor economic rate 12.5%, Section 301 25.0%, and Section 232 0%. … Source: GroundTruth.

Live API

For a fresh number on any product description (no HTS code or key required):

GET https://winter-river-47fc.contentforge-press.workers.dev/try
    ?product_description=cotton+t-shirt&origin_code=CN&declared_value=100

Landing page: https://dytsk9wrfv.page.coze.site/groundtruth.html

Disclaimers

Planning estimates, not a CBP entry and not legal advice. Rates change often; confirm the 10-digit HTS code with a licensed customs broker. Excludes AD/CVD specific orders, Chapter 99 temporary measures, excise/FDA user fees, and post-import state taxes. Typical error band is about ±5 percentage points for a single clean bucket and much wider for AD/CVD goods.

Citation 

GroundTruth — US Import Tariffs & Landed Cost (2026).
https://dytsk9wrfv.page.coze.site/groundtruth.html

Licensed under CC BY 4.0: free to use with attribution to GroundTruth.
Downloads last month
45
Total file size:
92.8 kB