Datasets:

costadev00
/

smoke-gemma4-matrix-brasil-25-v3-20260724-03

Languages:

Portuguese

Tags:

Dataset card Files Files and versions

xet

Community

Dataset Viewer

Duplicate

Split (1)

train

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
slot_id: string
attempt: int64
source: string
document_id: string
source_title: string
task: string
difficulty: string
instruction: string
input: string
output: string
messages: list<item: struct<role: string, content: string>>
  child 0, item: struct<role: string, content: string>
      child 0, role: string
      child 1, content: string
evidence: list<item: struct<document_id: string, section_id: string, start: int64, end: int64, quote: string>>
  child 0, item: struct<document_id: string, section_id: string, start: int64, end: int64, quote: string>
      child 0, document_id: string
      child 1, section_id: string
      child 2, start: int64
      child 3, end: int64
      child 4, quote: string
metadata: struct<risk: string, raw_response_stored: bool, task_contract: struct<task: string, reference_values (... 189 chars omitted)
  child 0, risk: string
  child 1, raw_response_stored: bool
  child 2, task_contract: struct<task: string, reference_values: list<item: string>, visible_source_quotes: list<item: string> (... 60 chars omitted)
      child 0, task: string
      child 1, reference_values: list<item: string>
          child 0, item: string
      child 2, visible_source_quotes: list<item: string>
          child 0, item: string
      child 3, choice_labels: list<item: string>
          child 0, item: string
      child 4, causal_supported: bool
  child 3, context_truncated: bool
  child 4, chunk_id: null
  child 5, document_title: string
generator: string
model: string
to
{'id': Value('string'), 'instruction': Value('string'), 'input': Value('string'), 'output': Value('string'), 'source': Value('string'), 'document_id': Value('string'), 'source_title': Value('string'), 'task': Value('string'), 'difficulty': Value('string'), 'evidence': List({'document_id': Value('string'), 'section_id': Value('string'), 'start': Value('int64'), 'end': Value('int64'), 'quote': Value('string')}), 'generator': Value('string'), 'model': Value('string')}
because column names don't match
Traceback:    Traceback (most recent call last):
                File "/src/services/worker/src/worker/utils.py", line 149, 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 129, 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 489, 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 2818, in __iter__
                  for key, example in ex_iterable:
                                      ^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2355, in __iter__
                  for key, pa_table in self._iter_arrow():
                                       ~~~~~~~~~~~~~~~~^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2380, 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 2369, in table_cast
                  return cast_table_to_schema(table, schema)
                File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2297, in cast_table_to_schema
                  raise CastError(
                  ...<3 lines>...
                  )
              datasets.table.CastError: Couldn't cast
              id: string
              slot_id: string
              attempt: int64
              source: string
              document_id: string
              source_title: string
              task: string
              difficulty: string
              instruction: string
              input: string
              output: string
              messages: list<item: struct<role: string, content: string>>
                child 0, item: struct<role: string, content: string>
                    child 0, role: string
                    child 1, content: string
              evidence: list<item: struct<document_id: string, section_id: string, start: int64, end: int64, quote: string>>
                child 0, item: struct<document_id: string, section_id: string, start: int64, end: int64, quote: string>
                    child 0, document_id: string
                    child 1, section_id: string
                    child 2, start: int64
                    child 3, end: int64
                    child 4, quote: string
              metadata: struct<risk: string, raw_response_stored: bool, task_contract: struct<task: string, reference_values (... 189 chars omitted)
                child 0, risk: string
                child 1, raw_response_stored: bool
                child 2, task_contract: struct<task: string, reference_values: list<item: string>, visible_source_quotes: list<item: string> (... 60 chars omitted)
                    child 0, task: string
                    child 1, reference_values: list<item: string>
                        child 0, item: string
                    child 2, visible_source_quotes: list<item: string>
                        child 0, item: string
                    child 3, choice_labels: list<item: string>
                        child 0, item: string
                    child 4, causal_supported: bool
                child 3, context_truncated: bool
                child 4, chunk_id: null
                child 5, document_title: string
              generator: string
              model: string
              to
              {'id': Value('string'), 'instruction': Value('string'), 'input': Value('string'), 'output': Value('string'), 'source': Value('string'), 'document_id': Value('string'), 'source_title': Value('string'), 'task': Value('string'), 'difficulty': Value('string'), 'evidence': List({'document_id': Value('string'), 'section_id': Value('string'), 'start': Value('int64'), 'end': Value('int64'), 'quote': Value('string')}), 'generator': Value('string'), 'model': 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.

Smoke Gemma 4 — matriz de 25 tasks sobre Brasil

Artefatos de uma execução real do planejamento matricial documentos × tasks, usando um documento da Wikipédia em português com o título Brasil e as 25 tasks canônicas do projeto.

Estado da execução

Esta é uma execução de diagnóstico e não representa um smoke test aprovado. O relatório terminou com status partial:

1 documento;
25 pares documento–task planejados;
21 candidatos aceitos;
0 pares pulados;
4 pares esgotados;
47 chamadas reais ao backend.

Geração

Backend: vllm_local
Modelo: google/gemma-4-31B-it-qat-w4a16-ct
Tensor parallel: 4 GPUs
Quantização: compressed-tensors
KV cache: FP8
Janela de contexto: 262.144 tokens
Concorrência: 1

Conteúdo

checkpoints/messages/: exemplos aceitos no formato de mensagens system, user e assistant;
checkpoints/prompt_completion/: exemplos aceitos em prompt/completion;
checkpoints/alpaca/: exemplos aceitos no formato Alpaca;
checkpoints/canonical/: registros canônicos aceitos;
run.db: estado completo, tentativas e respostas do backend;
plan.json: plano matricial v3;
report.json: relatório final da execução;
config.resolved.json: configuração efetivamente utilizada;
corpus-selected/: snapshot local do documento processado.

Para avaliar os resultados, considere somente os candidatos aceitos nos checkpoints. As respostas rejeitadas permanecem em run.db para auditoria.

Downloads last month: 10

Total file size:

5.5 MB