delta-tau-bench
Collection
2 items • Updated
Dataset Viewer
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: TypeError
Message: Couldn't cast array of type
struct<Engineer: struct<engineer_id: string, full_name: string, primary_email: string, office_phone: string, professional_license_number: string, engineering_discipline: string, certification_level: string, years_of_experience: int64, technical_skills: list<item: string>, project_portfolio: list<item: string>, department_affiliation: string, security_clearance_level: string, geographic_region: string, availability_status: string>, AgriculturalEquipment: struct<id: string, equipment_type: string, manufacturer: string, model: string, manufacturing_year: int64, power_source: string, operational_capacity: struct<type: string, value: double, unit: string>, fuel_consumption_rate: double, maintenance_schedule_days: int64, operational_status: string, safety_features: list<item: string>>>
to
{'Engineer': {'engineer_id': Value('string'), 'full_name': Value('string'), 'primary_email': Value('string'), 'office_phone': Value('string'), 'professional_license_number': Value('string'), 'engineering_discipline': Value('string'), 'certification_level': Value('string'), 'years_of_experience': Value('int64'), 'technical_skills': List(Value('string')), 'project_portfolio': List(Value('string')), 'department_affiliation': Value('string'), 'security_clearance_level': Value('string'), 'geographic_region': Value('string'), 'availability_status': Value('string')}, 'Course': {'course_code': Value('string'), 'course_title': Value('string'), 'credit_hours': Value('int64'), 'course_description': Value('string'), 'prerequisite_courses': List(Value('string')), 'learning_outcomes': List(Value('string')), 'course_level': Value('string'), 'creation_timestamp': Value('timestamp[s]'), 'delivery_mode': Value('string'), 'is_active': Value('bool'), 'subject_area': Value('string'), 'minimum_seat_capacity': Value('int64')}}
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 544, 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 3052, in __iter__
for key, example in ex_iterable:
^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2559, in __iter__
for key, pa_table in self._iter_arrow():
~~~~~~~~~~~~~~~~^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2584, 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 558, in _iter_arrow
for key, pa_table in iterator:
^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 440, 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 361, 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 2381, in table_cast
return cast_table_to_schema(table, schema)
File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2315, in cast_table_to_schema
cast_array_to_feature(
~~~~~~~~~~~~~~~~~~~~~^
table[name] if name in table_column_names else pa.array([None] * len(table), type=schema.field(name).type),
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
feature,
^^^^^^^^
)
^
File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 1861, in wrapper
return pa.chunked_array([func(chunk, *args, **kwargs) for chunk in array.chunks])
~~~~^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2071, in cast_array_to_feature
_c(children[name] if name in array_fields else null_array, subfeature)
~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 1863, in wrapper
return func(array, *args, **kwargs)
File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2161, in cast_array_to_feature
raise TypeError(f"Couldn't cast array of type\n{_short_str(array.type)}\nto\n{_short_str(feature)}")
TypeError: Couldn't cast array of type
struct<Engineer: struct<engineer_id: string, full_name: string, primary_email: string, office_phone: string, professional_license_number: string, engineering_discipline: string, certification_level: string, years_of_experience: int64, technical_skills: list<item: string>, project_portfolio: list<item: string>, department_affiliation: string, security_clearance_level: string, geographic_region: string, availability_status: string>, AgriculturalEquipment: struct<id: string, equipment_type: string, manufacturer: string, model: string, manufacturing_year: int64, power_source: string, operational_capacity: struct<type: string, value: double, unit: string>, fuel_consumption_rate: double, maintenance_schedule_days: int64, operational_status: string, safety_features: list<item: string>>>
to
{'Engineer': {'engineer_id': Value('string'), 'full_name': Value('string'), 'primary_email': Value('string'), 'office_phone': Value('string'), 'professional_license_number': Value('string'), 'engineering_discipline': Value('string'), 'certification_level': Value('string'), 'years_of_experience': Value('int64'), 'technical_skills': List(Value('string')), 'project_portfolio': List(Value('string')), 'department_affiliation': Value('string'), 'security_clearance_level': Value('string'), 'geographic_region': Value('string'), 'availability_status': Value('string')}, 'Course': {'course_code': Value('string'), 'course_title': Value('string'), 'credit_hours': Value('int64'), 'course_description': Value('string'), 'prerequisite_courses': List(Value('string')), 'learning_outcomes': List(Value('string')), 'course_level': Value('string'), 'creation_timestamp': Value('timestamp[s]'), 'delivery_mode': Value('string'), 'is_active': Value('bool'), 'subject_area': Value('string'), 'minimum_seat_capacity': Value('int64')}}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.
Δ-τ Bench dataset
This is the quality-reviewed dataset bundle for Δ-τ Bench, a benchmark for
multi-turn agents using an evolving tool catalog. Use it with the separately
published dtau-bench evaluator and Steward implementation.
Contents
| Component | Count / description |
|---|---|
| Domains | 42 |
| Training tasks | 1,191 |
| Test tasks | 397 |
| Tool catalogs | base plus v1–v6, 42 domain files per version |
| Executable services | 294 Python modules: 42 domains × 7 versions |
| Initial database | 287 JSON entity/relationship tables |
| Evaluation and evolution metadata | 424 JSON files |
| Catalog | Tools |
|---|---|
| base | 543 |
| v1 | 892 |
| v2 | 841 |
| v3 | 837 |
| v4 | 799 |
| v5 | 858 |
| v6 | 890 |
The 1,588 task identities and train/test split stay fixed across versions.
The base catalog supports the reference environment; the main benchmark runs
use v1–v6. Each task carries its reference and version-specific trajectories.
tasks/{train,test}/{domain}/{query_id}.json
tools/{domain}/{base,v1,v2,v3,v4,v5,v6}_tools.json
servers/{domain}.py base services
servers/{v1,...,v6}/{domain}.py evolved services
database/{entities,relationships}/*.json
meta/ costs, mappings, evaluation schemas
checksums.sha256 hashes of the 2,887 benchmark files
Task records contain query_id, domain, split, instruction, messages,
user_system_prompt, gold_trajectory, version_trajectories, and metadata.
Reference trajectories include tool names and parameter-construction code.
The evaluator uses the accompanying services and metadata to replay actions and
compare business states. Preserve all five top-level component directories.
Download and run
This is an executable benchmark bundle, rather than a single flat table.
Download the whole repository with the Hugging Face CLI or snapshot_download;
loading only task JSON files omits the execution environment.
python -m pip install huggingface_hub
hf download zjunlp/delta-tau-bench --repo-type dataset --local-dir ./dtau-dataset
From the installed code repository:
export DTAU_BENCH_DATA_ROOT=/absolute/path/to/dtau-dataset
python -m dtau_bench validate-data
python -m dtau_bench rollout --run-name inspect --list --split test
Verify the benchmark files independently on Linux:
cd /absolute/path/to/dtau-dataset
sha256sum -c checksums.sha256
Configure the chat and embedding endpoints in the code repository before running baseline or Steward experiments. Code, dependency environments, model weights, skill libraries, logs, and previous experiment outputs are distributed separately or generated at runtime.
Evaluation
Use the supplied train/test split and version-specific reference paths. The metadata includes rules for generated identities, relation matching, and evolved state comparisons; dropping those files changes evaluation behavior. Report SR, Tool Precision/Recall/F1, Step/Cost SPE, and the version pair when reporting cross-version IDR/DAR.
The release retains the quality-reviewed benchmark files byte-for-byte. The checksum list covers the data and executable services; this dataset card is excluded from that list so documentation can be updated independently.
- Downloads last month
- 69