Datasets:

NagaYu
/

halfword-bench

Error code:   StreamingRowsError
Exception:    CastError
Message:      Couldn't cast
condition: string
method: string
rate: double
speaker: string
chars: int64
seconds: double
selections: int64
baseline_selections: int64
keystroke_savings: double
words_per_minute: double
mean_candidates_shown: double
model_ms_total: double
model_ms_per_selection: double
charged_latency_s: double
adapted: bool
-- schema metadata --
huggingface: '{"info": {"features": {"condition": {"dtype": "string", "_t' + 794
to
{'method': Value('string'), 'label': Value('string'), 'parameter': Value('string'), 'typical': Value('float64'), 'low': Value('float64'), 'high': Value('float64'), 'unit': Value('string'), 'confidence': Value('string'), 'citation': Value('string'), 'is_rate_parameter': Value('bool'), 'note': Value('string')}
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/parquet/parquet.py", line 220, in _generate_tables
                  yield Key(file_idx, batch_idx), self._cast_table(pa_table)
                                                  ~~~~~~~~~~~~~~~~^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/parquet/parquet.py", line 156, 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
              condition: string
              method: string
              rate: double
              speaker: string
              chars: int64
              seconds: double
              selections: int64
              baseline_selections: int64
              keystroke_savings: double
              words_per_minute: double
              mean_candidates_shown: double
              model_ms_total: double
              model_ms_per_selection: double
              charged_latency_s: double
              adapted: bool
              -- schema metadata --
              huggingface: '{"info": {"features": {"condition": {"dtype": "string", "_t' + 794
              to
              {'method': Value('string'), 'label': Value('string'), 'parameter': Value('string'), 'typical': Value('float64'), 'low': Value('float64'), 'high': Value('float64'), 'unit': Value('string'), 'confidence': Value('string'), 'citation': Value('string'), 'is_rate_parameter': Value('bool'), 'note': Value('string')}
              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.

Halfword benchmark: conversational text under access-method cost models

This dataset pairs public conversational sentences with timing cost models for AAC access methods, so that a prediction system can be scored in seconds to utterance rather than in keystrokes saved.

It contains no data from AAC users. It is public conversational text plus simulation.

Configurations

utterances (12565 rows) -- normalised sentences with history, pseudo-speaker, source and that source's licence on every row.
cost_model (38 rows) -- every timing parameter with its citation, confidence label and published range.
traces (14400 rows) -- simulated entry per access method, rate and condition.

What the traces show

access method	rate	time-optimal (s)	keystroke-optimal (s)	saved	KSR (time-opt)	KSR (ks-opt)
gaze	0.15	20.4	20.383	-0.08%	39.1%	39.2%
gaze	0.6	43.593	43.588	-0.01%	39.2%	39.2%
gaze	1.2	74.524	74.527	0.0%	39.2%	39.2%
scanning	0.5	333.975	393.384	15.1%	35.0%	39.2%
scanning	1.2	749.086	898.083	16.59%	34.8%	39.2%
scanning	2.35	1427.095	1727.232	17.38%	34.7%	39.2%
switch	0.5	524.966	558.762	6.05%	37.5%	39.2%
switch	1.2	1219.192	1301.939	6.36%	37.5%	39.2%
switch	2.35	2359.124	2522.872	6.49%	37.5%	39.2%
tap	0.052	12.625	12.497	-1.02%	38.6%	39.2%
tap	0.15	25.59	25.292	-1.18%	38.4%	39.2%
tap	0.204	32.779	32.341	-1.35%	38.3%	39.2%

Two things in that table are worth reading carefully rather than skimming:

The gap tracks positional cost, not speed. Gaze is slow and shows nothing; scanning is faster per utterance at 0.5 s and shows 15%. What matters is whether reaching the tenth candidate costs more than reaching the first, and whether candidate rows push the keyboard further down the sweep. On gaze and touch, neither is true.
On touch the time-optimal policy is slightly worse (about 1%). Where position is nearly free there is nothing to optimise, and the decoder's one-step approximation costs a little. This is the honest boundary of the method: it is for scanning-like access.

Note also that the time is not free. On these corpora the time-optimal policy gives up a few points of keystroke savings to buy its seconds. That is the argument, not an inconvenience to it: if you report keystroke savings, this will look slightly worse, and the person will still have finished their sentence sooner.

Licences, per source

halfword-seed -- CC0-1.0. Hand-authored everyday sentences bundled so the repository runs offline. NOT data from AAC users. <halfword/data/seed_corpus.json>
allenai/soda -- cc-by-4.0. SODA (Kim et al., 2023), CC BY 4.0. https://huggingface.co/datasets/allenai/soda
google/Synthetic-Persona-Chat -- cc-by-4.0. Synthetic-Persona-Chat (Jandaghi et al., 2023), CC BY 4.0. https://huggingface.co/datasets/google/Synthetic-Persona-Chat
pfb30/multi_woz_v22 -- apache-2.0. MultiWOZ 2.2 (Zang et al., 2020), Apache-2.0. https://huggingface.co/datasets/pfb30/multi_woz_v22
Helsinki-NLP/tatoeba -- cc-by-2.0. Tatoeba, CC BY 2.0 FR. Individual sentences are contributed by Tatoeba members. https://huggingface.co/datasets/Helsinki-NLP/tatoeba

Limitations, stated plainly

The seconds are simulated, not measured with people. They price mechanical scan, dwell and pointing cost only. Visual search and decision cost are not modelled, which the AAC literature identifies as a major contributor -- so these figures are a lower bound on the cost of long candidate lists, not an estimate of it.
Pseudo-speakers are a construction from public corpora. They are not people, and the personalisation results should be read as 'the machinery works', not as an effect size anyone should expect.
Every duration here is a property of an interface configuration. None of them is a property of a person.

We would rather be corrected than cited

This benchmark was built without AAC users in the room, which is a real limitation and not a small one. If you use AAC, or support someone who does, and the defaults here look wrong -- please open an issue. See the project README for what we most need help with.

Downloads last month: 35