Datasets:

dusersad12
/

unified-tool-call-corpus

Error code:   StreamingRowsError
Exception:    CastError
Message:      Couldn't cast
conversation_id: string
messages: list<item: struct<role: string, content: string, tool_calls: list<item: struct<id: string, name: str (... 786 chars omitted)
  child 0, item: struct<role: string, content: string, tool_calls: list<item: struct<id: string, name: string, argume (... 774 chars omitted)
      child 0, role: string
      child 1, content: string
      child 2, tool_calls: list<item: struct<id: string, name: string, arguments: struct<query: string, category: string, sort_ (... 701 chars omitted)
          child 0, item: struct<id: string, name: string, arguments: struct<query: string, category: string, sort_order: stri (... 689 chars omitted)
              child 0, id: string
              child 1, name: string
              child 2, arguments: struct<query: string, category: string, sort_order: string, product_id: string, amount: int64, from_ (... 644 chars omitted)
                  child 0, query: string
                  child 1, category: string
                  child 2, sort_order: string
                  child 3, product_id: string
                  child 4, amount: int64
                  child 5, from_currency: string
                  child 6, to_currency: string
                  child 7, origin: string
                  child 8, destination: string
                  child 9, after_time: string
                  child 10, city: string
                  child 11, unit: string
                  child 12, company: string
                  child 13, title:
...
             child 1, description: string
              child 43, artist: struct<type: string, description: string>
                  child 0, type: string
                  child 1, description: string
              child 44, name: struct<type: string, description: string>
                  child 0, type: string
                  child 1, description: string
              child 45, public: struct<type: string, description: string, default: bool>
                  child 0, type: string
                  child 1, description: string
                  child 2, default: bool
          child 2, required: list<item: string>
              child 0, item: string
per_source: struct<toolace: struct<raw: int64, kept: int64, excluded: int64>, glaive: struct<raw: int64, kept: i (... 79 chars omitted)
  child 0, toolace: struct<raw: int64, kept: int64, excluded: int64>
      child 0, raw: int64
      child 1, kept: int64
      child 2, excluded: int64
  child 1, glaive: struct<raw: int64, kept: int64, excluded: int64>
      child 0, raw: int64
      child 1, kept: int64
      child 2, excluded: int64
  child 2, xlam: struct<raw: int64, kept: int64, excluded: int64>
      child 0, raw: int64
      child 1, kept: int64
      child 2, excluded: int64
exclusion_reasons: struct<duplicate: int64, undefined_tool: int64, missing_user_turn: int64>
  child 0, duplicate: int64
  child 1, undefined_tool: int64
  child 2, missing_user_turn: int64
total_raw: int64
total_kept: int64
total_excluded: int64
to
{'total_raw': Value('int64'), 'total_kept': Value('int64'), 'total_excluded': Value('int64'), 'per_source': {'toolace': {'raw': Value('int64'), 'kept': Value('int64'), 'excluded': Value('int64')}, 'glaive': {'raw': Value('int64'), 'kept': Value('int64'), 'excluded': Value('int64')}, 'xlam': {'raw': Value('int64'), 'kept': Value('int64'), 'excluded': Value('int64')}}, 'exclusion_reasons': {'duplicate': Value('int64'), 'undefined_tool': Value('int64'), 'missing_user_turn': Value('int64')}}
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
              conversation_id: string
              messages: list<item: struct<role: string, content: string, tool_calls: list<item: struct<id: string, name: str (... 786 chars omitted)
                child 0, item: struct<role: string, content: string, tool_calls: list<item: struct<id: string, name: string, argume (... 774 chars omitted)
                    child 0, role: string
                    child 1, content: string
                    child 2, tool_calls: list<item: struct<id: string, name: string, arguments: struct<query: string, category: string, sort_ (... 701 chars omitted)
                        child 0, item: struct<id: string, name: string, arguments: struct<query: string, category: string, sort_order: stri (... 689 chars omitted)
                            child 0, id: string
                            child 1, name: string
                            child 2, arguments: struct<query: string, category: string, sort_order: string, product_id: string, amount: int64, from_ (... 644 chars omitted)
                                child 0, query: string
                                child 1, category: string
                                child 2, sort_order: string
                                child 3, product_id: string
                                child 4, amount: int64
                                child 5, from_currency: string
                                child 6, to_currency: string
                                child 7, origin: string
                                child 8, destination: string
                                child 9, after_time: string
                                child 10, city: string
                                child 11, unit: string
                                child 12, company: string
                                child 13, title:
              ...
                           child 1, description: string
                            child 43, artist: struct<type: string, description: string>
                                child 0, type: string
                                child 1, description: string
                            child 44, name: struct<type: string, description: string>
                                child 0, type: string
                                child 1, description: string
                            child 45, public: struct<type: string, description: string, default: bool>
                                child 0, type: string
                                child 1, description: string
                                child 2, default: bool
                        child 2, required: list<item: string>
                            child 0, item: string
              per_source: struct<toolace: struct<raw: int64, kept: int64, excluded: int64>, glaive: struct<raw: int64, kept: i (... 79 chars omitted)
                child 0, toolace: struct<raw: int64, kept: int64, excluded: int64>
                    child 0, raw: int64
                    child 1, kept: int64
                    child 2, excluded: int64
                child 1, glaive: struct<raw: int64, kept: int64, excluded: int64>
                    child 0, raw: int64
                    child 1, kept: int64
                    child 2, excluded: int64
                child 2, xlam: struct<raw: int64, kept: int64, excluded: int64>
                    child 0, raw: int64
                    child 1, kept: int64
                    child 2, excluded: int64
              exclusion_reasons: struct<duplicate: int64, undefined_tool: int64, missing_user_turn: int64>
                child 0, duplicate: int64
                child 1, undefined_tool: int64
                child 2, missing_user_turn: int64
              total_raw: int64
              total_kept: int64
              total_excluded: int64
              to
              {'total_raw': Value('int64'), 'total_kept': Value('int64'), 'total_excluded': Value('int64'), 'per_source': {'toolace': {'raw': Value('int64'), 'kept': Value('int64'), 'excluded': Value('int64')}, 'glaive': {'raw': Value('int64'), 'kept': Value('int64'), 'excluded': Value('int64')}, 'xlam': {'raw': Value('int64'), 'kept': Value('int64'), 'excluded': Value('int64')}}, 'exclusion_reasons': {'duplicate': Value('int64'), 'undefined_tool': Value('int64'), 'missing_user_turn': Value('int64')}}
              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.

Unified Tool-Call Corpus

A small, cleaned corpus of tool-calling (function-calling) conversations in a single unified schema, built for fine-tuning small function-calling models.

It merges the leading chunk of three public tool-calling dataset exports (ToolACE-style, glaive-style and XLAM-style), each of which encodes messages, tool calls and tool definitions differently, into one format.

Format

Each JSON Lines record of unified_tool_call.jsonl is one conversation:

conversation_id: {source}_{index}, where index is the entry's position in its original raw file (toolace / glaive / xlam).
messages: list of {role, content} messages with roles system / user / assistant / tool.
- assistant messages may carry tool_calls: a list of {id, name, arguments}; ids are renumbered tool_call_0, tool_call_1, ... within each conversation.
- an assistant turn that is nothing but tool calls has content: null.
- tool messages carry tool_call_id plus the raw tool output as content.
tools: JSON-Schema-style tool definitions, each with {name, description, parameters: {type, properties, required}}.

Conversion notes:

system messages that only hand over the toolset and usage instructions were removed; persona system messages were kept.
flat (XLAM-style) parameter lists were expanded into properties / required; a parameter is optional (never listed in required) when it carries an explicit default key (even a null one) or when its comma-separated type annotation contains the token optional.
source type aliases were normalised to JSON Schema types (dict->object, int->integer, float->number, str->string, ...).
no tool results or assistant replies were invented; XLAM entries contain only the user turn and the assistant tool-call turn, exactly as in the source.

Screening

Conversations were dropped when they had no user turn (missing_user_turn), when an assistant tool_calls entry invoked a tool not defined in that conversation's own tools list (undefined_tool), or when they duplicated another conversation - same first user message plus the same complete sequence of assistant tool calls (duplicate). Every dropped conversation is listed in excluded.jsonl with its conversation_id, source, reason and a one-sentence detail.

Statistics

source	raw	kept	excluded
toolace	26	23	3
glaive	26	24	2
xlam	26	25	1
total	78	72	6

Exclusions by reason: {"duplicate": 3, "undefined_tool": 2, "missing_user_turn": 1}

The merged file is ordered by conversation_id compared as a plain ascending string (so glaive_10 sorts before glaive_2).

Files

unified_tool_call.jsonl - the merged, screened corpus (72 conversations).
summary.json - the counts above.

Downloads last month: 298

Total file size:

101 kB