Datasets:
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: CastError
Message: Couldn't cast
benchmark: string
policy: string
sample_id: string
generation: int64
source_log: string
sample: struct<id: string, epoch: int64, input: string, choices: null, target: string, sandbox: null, files: (... 44650 chars omitted)
child 0, id: string
child 1, epoch: int64
child 2, input: string
child 3, choices: null
child 4, target: string
child 5, sandbox: null
child 6, files: null
child 7, setup: null
child 8, messages: list<item: struct<id: string, content: string, source: string, metadata: null, role: string, tool_ca (... 650 chars omitted)
child 0, item: struct<id: string, content: string, source: string, metadata: null, role: string, tool_call_id: stri (... 638 chars omitted)
child 0, id: string
child 1, content: string
child 2, source: string
child 3, metadata: null
child 4, role: string
child 5, tool_call_id: string
child 6, tool_calls: list<item: struct<id: string, function: string, arguments: struct<image_path: string, prompt: string (... 475 chars omitted)
child 0, item: struct<id: string, function: string, arguments: struct<image_path: string, prompt: string, video_pat (... 463 chars omitted)
child 0, id: string
child 1, function: string
child 2, arguments: struct<image_path: string, prompt: string, video_path: string, script: string, base_img: string, mp3 (... 369 chars omitted)
child 0, i
...
al_name: string, sha256: string>>
child 0, item: struct<bytes: int64, filename: string, logical_name: string, sha256: string>
child 0, bytes: int64
child 1, filename: string
child 2, logical_name: string
child 3, sha256: string
base_model: string
source_git_commit: string
total_records: int64
temperature: double
policies: struct<sdf1250: struct<description: string>, sdf1250_plus_vt250: struct<description: string>, sdf225 (... 212 chars omitted)
child 0, sdf1250: struct<description: string>
child 0, description: string
child 1, sdf1250_plus_vt250: struct<description: string>
child 0, description: string
child 2, sdf2250: struct<adapter_weights_sha256: string, description: string>
child 0, adapter_weights_sha256: string
child 1, description: string
child 3, sdf2250_plus_vt250: struct<adapter_weights_sha256: string, description: string>
child 0, adapter_weights_sha256: string
child 1, description: string
child 4, vt250: struct<adapter_weights_sha256: string, description: string>
child 0, adapter_weights_sha256: string
child 1, description: string
excluded: struct<interrupted_or_smoke_runs: string, judgments: string, partial_vt_f8_vllm: string>
child 0, interrupted_or_smoke_runs: string
child 1, judgments: string
child 2, partial_vt_f8_vllm: string
generation_backend: string
updated_at: string
title: string
maximum_new_tokens: int64
to
{'base_model': Value('string'), 'created_at': Value('string'), 'excluded': {'interrupted_or_smoke_runs': Value('string'), 'judgments': Value('string'), 'partial_vt_f8_vllm': Value('string')}, 'exports': {'agentharm': {'sdf2250/benign': {'generation_backend': Value('string'), 'generations_per_prompt': Value('int64'), 'note': Value('string'), 'output': Value('string'), 'output_bytes': Value('int64'), 'output_sha256': Value('string'), 'prompts': Value('int64'), 'records': Value('int64'), 'sources': List({'bytes': Value('int64'), 'filename': Value('string'), 'logical_name': Value('string'), 'sha256': Value('string')})}, 'sdf2250/harmful': {'generation_backend': Value('string'), 'generations_per_prompt': Value('int64'), 'note': Value('string'), 'output': Value('string'), 'output_bytes': Value('int64'), 'output_sha256': Value('string'), 'prompts': Value('int64'), 'records': Value('int64'), 'sources': List({'bytes': Value('int64'), 'filename': Value('string'), 'logical_name': Value('string'), 'sha256': Value('string')})}, 'sdf2250_plus_vt250/benign': {'generations_per_prompt': Value('int64'), 'note': Value('string'), 'output': Value('string'), 'output_bytes': Value('int64'), 'output_sha256': Value('string'), 'prompts': Value('int64'), 'records': Value('int64'), 'sources': List({'bytes': Value('int64'), 'filename': Value('string'), 'logical_name': Value('string'), 'sha256': Value('string')})}, 'sdf2250_plus_vt250/harmful': {'generations_per_prompt': Value('int64'), 'note': Value('str
...
ing'), 'output': Value('string'), 'output_bytes': Value('int64'), 'output_sha256': Value('string'), 'records': Value('int64'), 'sources': List({'bytes': Value('int64'), 'filename': Value('string'), 'logical_name': Value('string'), 'sha256': Value('string')})}, 'vt250/F6': {'generation_backend': Value('string'), 'output': Value('string'), 'output_bytes': Value('int64'), 'output_sha256': Value('string'), 'records': Value('int64'), 'sources': List({'bytes': Value('int64'), 'filename': Value('string'), 'logical_name': Value('string'), 'sha256': Value('string')})}, 'vt250/F7': {'generation_backend': Value('string'), 'output': Value('string'), 'output_bytes': Value('int64'), 'output_sha256': Value('string'), 'records': Value('int64'), 'sources': List({'bytes': Value('int64'), 'filename': Value('string'), 'logical_name': Value('string'), 'sha256': Value('string')})}}}, 'generation_backend': Value('string'), 'maximum_new_tokens': Value('int64'), 'policies': {'sdf1250': {'description': Value('string')}, 'sdf1250_plus_vt250': {'description': Value('string')}, 'sdf2250': {'adapter_weights_sha256': Value('string'), 'description': Value('string')}, 'sdf2250_plus_vt250': {'adapter_weights_sha256': Value('string'), 'description': Value('string')}, 'vt250': {'adapter_weights_sha256': Value('string'), 'description': Value('string')}}, 'source_git_commit': Value('string'), 'temperature': Value('float64'), 'title': Value('string'), 'total_records': Value('int64'), 'updated_at': 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/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
benchmark: string
policy: string
sample_id: string
generation: int64
source_log: string
sample: struct<id: string, epoch: int64, input: string, choices: null, target: string, sandbox: null, files: (... 44650 chars omitted)
child 0, id: string
child 1, epoch: int64
child 2, input: string
child 3, choices: null
child 4, target: string
child 5, sandbox: null
child 6, files: null
child 7, setup: null
child 8, messages: list<item: struct<id: string, content: string, source: string, metadata: null, role: string, tool_ca (... 650 chars omitted)
child 0, item: struct<id: string, content: string, source: string, metadata: null, role: string, tool_call_id: stri (... 638 chars omitted)
child 0, id: string
child 1, content: string
child 2, source: string
child 3, metadata: null
child 4, role: string
child 5, tool_call_id: string
child 6, tool_calls: list<item: struct<id: string, function: string, arguments: struct<image_path: string, prompt: string (... 475 chars omitted)
child 0, item: struct<id: string, function: string, arguments: struct<image_path: string, prompt: string, video_pat (... 463 chars omitted)
child 0, id: string
child 1, function: string
child 2, arguments: struct<image_path: string, prompt: string, video_path: string, script: string, base_img: string, mp3 (... 369 chars omitted)
child 0, i
...
al_name: string, sha256: string>>
child 0, item: struct<bytes: int64, filename: string, logical_name: string, sha256: string>
child 0, bytes: int64
child 1, filename: string
child 2, logical_name: string
child 3, sha256: string
base_model: string
source_git_commit: string
total_records: int64
temperature: double
policies: struct<sdf1250: struct<description: string>, sdf1250_plus_vt250: struct<description: string>, sdf225 (... 212 chars omitted)
child 0, sdf1250: struct<description: string>
child 0, description: string
child 1, sdf1250_plus_vt250: struct<description: string>
child 0, description: string
child 2, sdf2250: struct<adapter_weights_sha256: string, description: string>
child 0, adapter_weights_sha256: string
child 1, description: string
child 3, sdf2250_plus_vt250: struct<adapter_weights_sha256: string, description: string>
child 0, adapter_weights_sha256: string
child 1, description: string
child 4, vt250: struct<adapter_weights_sha256: string, description: string>
child 0, adapter_weights_sha256: string
child 1, description: string
excluded: struct<interrupted_or_smoke_runs: string, judgments: string, partial_vt_f8_vllm: string>
child 0, interrupted_or_smoke_runs: string
child 1, judgments: string
child 2, partial_vt_f8_vllm: string
generation_backend: string
updated_at: string
title: string
maximum_new_tokens: int64
to
{'base_model': Value('string'), 'created_at': Value('string'), 'excluded': {'interrupted_or_smoke_runs': Value('string'), 'judgments': Value('string'), 'partial_vt_f8_vllm': Value('string')}, 'exports': {'agentharm': {'sdf2250/benign': {'generation_backend': Value('string'), 'generations_per_prompt': Value('int64'), 'note': Value('string'), 'output': Value('string'), 'output_bytes': Value('int64'), 'output_sha256': Value('string'), 'prompts': Value('int64'), 'records': Value('int64'), 'sources': List({'bytes': Value('int64'), 'filename': Value('string'), 'logical_name': Value('string'), 'sha256': Value('string')})}, 'sdf2250/harmful': {'generation_backend': Value('string'), 'generations_per_prompt': Value('int64'), 'note': Value('string'), 'output': Value('string'), 'output_bytes': Value('int64'), 'output_sha256': Value('string'), 'prompts': Value('int64'), 'records': Value('int64'), 'sources': List({'bytes': Value('int64'), 'filename': Value('string'), 'logical_name': Value('string'), 'sha256': Value('string')})}, 'sdf2250_plus_vt250/benign': {'generations_per_prompt': Value('int64'), 'note': Value('string'), 'output': Value('string'), 'output_bytes': Value('int64'), 'output_sha256': Value('string'), 'prompts': Value('int64'), 'records': Value('int64'), 'sources': List({'bytes': Value('int64'), 'filename': Value('string'), 'logical_name': Value('string'), 'sha256': Value('string')})}, 'sdf2250_plus_vt250/harmful': {'generations_per_prompt': Value('int64'), 'note': Value('str
...
ing'), 'output': Value('string'), 'output_bytes': Value('int64'), 'output_sha256': Value('string'), 'records': Value('int64'), 'sources': List({'bytes': Value('int64'), 'filename': Value('string'), 'logical_name': Value('string'), 'sha256': Value('string')})}, 'vt250/F6': {'generation_backend': Value('string'), 'output': Value('string'), 'output_bytes': Value('int64'), 'output_sha256': Value('string'), 'records': Value('int64'), 'sources': List({'bytes': Value('int64'), 'filename': Value('string'), 'logical_name': Value('string'), 'sha256': Value('string')})}, 'vt250/F7': {'generation_backend': Value('string'), 'output': Value('string'), 'output_bytes': Value('int64'), 'output_sha256': Value('string'), 'records': Value('int64'), 'sources': List({'bytes': Value('int64'), 'filename': Value('string'), 'logical_name': Value('string'), 'sha256': Value('string')})}}}, 'generation_backend': Value('string'), 'maximum_new_tokens': Value('int64'), 'policies': {'sdf1250': {'description': Value('string')}, 'sdf1250_plus_vt250': {'description': Value('string')}, 'sdf2250': {'adapter_weights_sha256': Value('string'), 'description': Value('string')}, 'sdf2250_plus_vt250': {'adapter_weights_sha256': Value('string'), 'description': Value('string')}, 'vt250': {'adapter_weights_sha256': Value('string'), 'description': Value('string')}}, 'source_git_commit': Value('string'), 'temperature': Value('float64'), 'title': Value('string'), 'total_records': Value('int64'), 'updated_at': Value('string')}
because column names don't matchNeed help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
vLLM SDF & SDF+VT rollouts Qwen
Full, untruncated Qwen3.6-35B-A3B rollouts generated locally with compiled CUDA graphs (vLLM 0.26.0 for the original exports and vLLM 0.19.1 CUDA 12.8 for the added SDF-2250 and VT-250 exports). The aggregate contains 21,290 records. Generation used temperature 0.3 and a 16,000-token per-turn/output cap.
Contents
| Benchmark | Policies | Records |
|---|---|---|
| EvalAwareBench | VT-250, SDF-1250, SDF-2250, SDF-1250+VT-250, SDF-2250+VT-250 | 12,500 |
| Agentic Misalignment | VT-250, SDF-2250, SDF-2250+VT-250 | 3,510 |
| AgentHarm harmful and benign | VT-250, SDF-2250, SDF-2250+VT-250 | 5,280 |
data/eab/ contains the native rollout schema (prompt_tokens, continuation_tokens, text, stop reason, and any existing judge field). data/agentic_misalignment/ and data/agentharm/ contain one wrapper row per rollout with benchmark, policy, sample ID, normalized generation number, source-log basename, and the complete original Inspect sample under sample.
AgentHarm has exactly five generations per prompt. Generation 1 comes from the completed initial run; generations 2–5 come from the clean four-generation extension. Every policy/subset has 176 prompts × 5 = 880 records.
See MANIFEST.json for exact paths, counts, SHA-256 hashes, policy identities, exclusions, and provenance. Interrupted/smoke runs and the partial two-generation VT F8 run are deliberately excluded.
Important notes
- The AgentHarm files contain harmful task content and model responses. They are intended for safety research.
- This repository does not add a new license over upstream benchmark prompts or model outputs; users must follow the applicable upstream licenses and terms.
- Generation and judging are separated. Newly collected matrix rollouts are unjudged. Some reused first-generation AgentHarm samples and earlier F8 records retain their original embedded score/judge fields.
- Downloads last month
- 52