Dataset Preview
The full dataset viewer is not available (click to read why). Only showing a preview of the rows.
The dataset generation failed because of a cast error
Error code: DatasetGenerationCastError
Exception: DatasetGenerationCastError
Message: An error occurred while generating the dataset
All the data files must have the same columns, but at some point there are 4 new columns ({'model_response', 'prompt', 'pidgin_english', 'expected'}) and 4 missing columns ({'standard_english', 'pidgin_prompt', 'expected_behavior', 'evaluation_criterion'}).
This happened while the json dataset builder was generating data using
hf://datasets/Britzzy/nigerian-pidgin-evaluation/01Fatima_pidgin_eval_Law/pidgin_evaluation_results.json (at revision 778f9470e5fafb6137f915be870ecc837a2a00c4), ['hf://datasets/Britzzy/nigerian-pidgin-evaluation@778f9470e5fafb6137f915be870ecc837a2a00c4/01Fatima_pidgin_eval_Law/evaluation_prompts.json', 'hf://datasets/Britzzy/nigerian-pidgin-evaluation@778f9470e5fafb6137f915be870ecc837a2a00c4/01Fatima_pidgin_eval_Law/pidgin_evaluation_results.json'], ['hf://datasets/Britzzy/nigerian-pidgin-evaluation@778f9470e5fafb6137f915be870ecc837a2a00c4/01Fatima_pidgin_eval_Law/evaluation_prompts.json', 'hf://datasets/Britzzy/nigerian-pidgin-evaluation@778f9470e5fafb6137f915be870ecc837a2a00c4/01Fatima_pidgin_eval_Law/pidgin_evaluation_results.json']
Please either edit the data files to have matching columns, or separate them into different configurations (see docs at https://hf.co/docs/hub/datasets-manual-configuration#multiple-configurations)
Traceback: Traceback (most recent call last):
File "/usr/local/lib/python3.14/site-packages/datasets/builder.py", line 1856, in _prepare_split_single
writer.write_table(table)
~~~~~~~~~~~~~~~~~~^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/arrow_writer.py", line 811, in write_table
self._write_table(pa_table, writer_batch_size=writer_batch_size)
~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/arrow_writer.py", line 822, in _write_table
pa_table = table_cast(pa_table, self._schema)
File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2381, in table_cast
return cast_table_to_schema(table, schema)
File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2309, in cast_table_to_schema
raise CastError(
...<3 lines>...
)
datasets.table.CastError: Couldn't cast
id: int64
category: string
prompt: string
expected: string
pidgin_english: string
model_response: string
to
{'id': Value('int64'), 'category': Value('string'), 'pidgin_prompt': Value('string'), 'standard_english': Value('string'), 'expected_behavior': Value('string'), 'evaluation_criterion': Value('string')}
because column names don't match
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "/src/services/worker/src/worker/job_runners/config/parquet_and_info.py", line 1379, in compute_config_parquet_and_info_response
parquet_operations, partial, estimated_dataset_info = stream_convert_to_parquet(
~~~~~~~~~~~~~~~~~~~~~~~~~^
builder, max_dataset_size_bytes=max_dataset_size_bytes
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
)
^
File "/src/services/worker/src/worker/job_runners/config/parquet_and_info.py", line 956, in stream_convert_to_parquet
builder._prepare_split(split_generator=splits_generators[split], file_format="parquet")
~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/builder.py", line 1702, in _prepare_split
for job_id, done, content in self._prepare_split_single(
~~~~~~~~~~~~~~~~~~~~~~~~~~^
gen_kwargs=gen_kwargs, job_id=job_id, **_prepare_split_args
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
):
^
File "/usr/local/lib/python3.14/site-packages/datasets/builder.py", line 1858, in _prepare_split_single
raise DatasetGenerationCastError.from_cast_error(
...<4 lines>...
)
datasets.exceptions.DatasetGenerationCastError: An error occurred while generating the dataset
All the data files must have the same columns, but at some point there are 4 new columns ({'model_response', 'prompt', 'pidgin_english', 'expected'}) and 4 missing columns ({'standard_english', 'pidgin_prompt', 'expected_behavior', 'evaluation_criterion'}).
This happened while the json dataset builder was generating data using
hf://datasets/Britzzy/nigerian-pidgin-evaluation/01Fatima_pidgin_eval_Law/pidgin_evaluation_results.json (at revision 778f9470e5fafb6137f915be870ecc837a2a00c4), ['hf://datasets/Britzzy/nigerian-pidgin-evaluation@778f9470e5fafb6137f915be870ecc837a2a00c4/01Fatima_pidgin_eval_Law/evaluation_prompts.json', 'hf://datasets/Britzzy/nigerian-pidgin-evaluation@778f9470e5fafb6137f915be870ecc837a2a00c4/01Fatima_pidgin_eval_Law/pidgin_evaluation_results.json'], ['hf://datasets/Britzzy/nigerian-pidgin-evaluation@778f9470e5fafb6137f915be870ecc837a2a00c4/01Fatima_pidgin_eval_Law/evaluation_prompts.json', 'hf://datasets/Britzzy/nigerian-pidgin-evaluation@778f9470e5fafb6137f915be870ecc837a2a00c4/01Fatima_pidgin_eval_Law/pidgin_evaluation_results.json']
Please either edit the data files to have matching columns, or separate them into different configurations (see docs at https://hf.co/docs/hub/datasets-manual-configuration#multiple-configurations)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.
id int64 | category string | pidgin_prompt string | standard_english string | expected_behavior string | evaluation_criterion string |
|---|---|---|---|---|---|
1 | comprehension | Abeg, wetin dey happen for market today? I hear say tomato price don go up well well. | Please, what is happening in the market today? I heard tomato prices have gone up a lot. | Should recognize this as a question about market prices, specifically tomatoes being expensive. | Does the model understand abeg (please), wetin (what), and the market context? |
2 | translation | Translate this Nigerian Pidgin to standard English: I no fit come today, my motor break down for road. | I cannot come today, my car broke down on the road. | Should translate to standard English preserving meaning. | Does the model correctly translate no fit (cannot) and motor (car)? |
3 | reasoning | If person wan travel from Lagos to Abuja by road, e go take am how many hours? Abeg explain wetin fit delay the journey. | If someone wants to travel from Lagos to Abuja by road, how many hours will it take? Please explain what can delay the journey. | Should estimate 10-13 hours and mention traffic, checkpoints, road conditions, or fuel scarcity. | Does the model reason about Nigerian geography and infrastructure? |
4 | cultural_context | My mama say make I bring suya when I dey come back. Suya na wetin? And where person fit buy am for Lagos? | My mother said I should bring suya when I come back. What is suya? And where can one buy it in Lagos? | Should explain suya as a spicy grilled meat snack sold by roadside vendors in Lagos. | Does the model know Nigerian cultural items (suya) and local context (Lagos)? |
5 | code_switching | Oga, the server don crash again. I don restart am two times but e no gree work. Maybe na the database get problem. Wetin you think say I suppose do? | Boss, the server has crashed again. I have restarted it two times but it will not work. Maybe the database has a problem. What do you think I should do? | Should identify this as a technical problem and suggest debugging steps. | Does the model handle Pidgin-English code-switching in a technical context? |
6 | comprehension | The danfo driver say he no get change. I give am 1000 naira for 200 naira journey. Na lie e dey talk or true? | The bus driver says he does not have change. I gave him 1000 naira for a 200 naira trip. Is he lying or telling the truth? | Should understand the Lagos transport scenario about a bus driver not having change. | Does the model understand danfo (Lagos bus) and the change-making scenario? |
7 | translation | Translate to standard English: Wetin you dey find for my house by this time? You no know say armed robbers dey operate for this area? | What are you looking for at my house at this time? Do you not know that armed robbers operate in this area? | Should translate accurately to standard English. | Does the model correctly translate wetin you dey find and the security context? |
8 | reasoning | If NEPA take light by 6pm and dem no bring am back till 10pm, and my generator get fuel wey fit last 3 hours, how I go take do work wey need 4 hours electricity? | If the power company takes electricity at 6pm and does not restore it until 10pm, and my generator has fuel for 3 hours, how do I do work that needs 4 hours of electricity? | Should recognize NEPA as the power utility, calculate the gap (need 4 hours, generator gives 3, 1 hour short), and suggest alternatives. | Does the model understand NEPA take light (power outage) and reason about Nigerian power infrastructure? |
9 | cultural_context | My friend dey do traditional wedding next month. E say I must wear aso ebi. Wetin be aso ebi and how much e fit cost? | My friend is having a traditional wedding next month. He says I must wear aso ebi. What is aso ebi and how much can it cost? | Should explain aso ebi as a uniform fabric worn by family and close friends at Nigerian events. | Does the model understand aso ebi (event fabric) and Nigerian wedding customs? |
10 | code_switching | The commit wey I push yesterday don break the build. E be like say the API wey we dey call don change their response format. I suppose revert or make I fix the parser? Abeg suggest wetin make sense. | The commit I pushed yesterday has broken the build. It seems the API we call has changed their response format. Should I revert or fix the parser? Please suggest what makes sense. | Should identify this as a software engineering problem and suggest reverting or fixing the parser. | Does the model handle Pidgin in a software engineering context with technical terms? |
1 | comprehension | null | null | null | null |
2 | translation | null | null | null | null |
3 | reasoning | null | null | null | null |
4 | cultural_context | null | null | null | null |
5 | code_switching | null | null | null | null |
6 | comprehension | null | null | null | null |
7 | translation | null | null | null | null |
8 | reasoning | null | null | null | null |
9 | cultural_context | null | null | null | null |
10 | code_switching | null | null | null | null |
Nigerian Pidgin Evaluation
Native-speaker evaluation of how well small language models understand everyday Nigerian Pidgin.
The first study, in 01Fatima_pidgin_eval_Law, tests Qwen2.5-1.5B-Instruct on 10 Pidgin prompts. Native-speaker review found high or critical errors in 7 of 10 responses, while automated keyword scoring rated most of them acceptable. Full results, prompts and the reproduction notebook are in that folder.
Author: Lawrence Oladeji. License: CC-BY-4.0.
- Downloads last month
- 49