Datasets:

EricSpencer00
/

cad-bench-results

Error code:   StreamingRowsError
Exception:    CastError
Message:      Couldn't cast
score: double
geometry_similarity: double
cad_spec_consistency: double
combined: double
geometry_similarity_reason: string
cad_spec_consistency_reason: string
task_id: struct<org: string, name: string, ref: string>
  child 0, org: string
  child 1, name: string
  child 2, ref: string
agent_info: struct<name: string, version: string, model_info: struct<name: string, provider: null>>
  child 0, name: string
  child 1, version: string
  child 2, model_info: struct<name: string, provider: null>
      child 0, name: string
      child 1, provider: null
task_name: string
started_at: string
verifier_result: struct<rewards: struct<score: double, geometry_similarity: double, cad_spec_consistency: double, com (... 15 chars omitted)
  child 0, rewards: struct<score: double, geometry_similarity: double, cad_spec_consistency: double, combined: double>
      child 0, score: double
      child 1, geometry_similarity: double
      child 2, cad_spec_consistency: double
      child 3, combined: double
finished_at: string
agent_setup: struct<started_at: string, finished_at: string>
  child 0, started_at: string
  child 1, finished_at: string
verifier: struct<started_at: string, finished_at: string>
  child 0, started_at: string
  child 1, finished_at: string
agent_result: struct<n_input_tokens: int64, n_cache_tokens: int64, n_output_tokens: int64, cost_usd: double, rollo (... 33 chars omitted)
  child 0, n_input_tokens: int64
  child 1, n_cache_tokens: int64
  child 2, n_output_tokens: int64
  
...
10, load_trajectory: null
      child 11, extra_allowed_hosts: list<item: null>
          child 0, item: null
      child 12, kwargs: struct<reasoning_effort: string, max_turns: int64>
          child 0, reasoning_effort: string
          child 1, max_turns: int64
      child 13, mcp_servers: list<item: null>
          child 0, item: null
  child 10, environment: struct<type: string, import_path: null, force_build: bool, delete: bool, cpu_enforcement_policy: str (... 265 chars omitted)
      child 0, type: string
      child 1, import_path: null
      child 2, force_build: bool
      child 3, delete: bool
      child 4, cpu_enforcement_policy: string
      child 5, memory_enforcement_policy: string
      child 6, override_cpus: null
      child 7, override_memory_mb: null
      child 8, override_storage_mb: null
      child 9, override_gpus: null
      child 10, override_tpu: null
      child 11, mounts: null
      child 12, extra_docker_compose: list<item: null>
          child 0, item: null
      child 13, kwargs: struct<>
      child 14, extra_allowed_hosts: list<item: null>
          child 0, item: null
  child 11, verifier: struct<override_timeout_sec: null, max_timeout_sec: null, disable: bool>
      child 0, override_timeout_sec: null
      child 1, max_timeout_sec: null
      child 2, disable: bool
  child 12, artifacts: list<item: null>
      child 0, item: null
  child 13, extra_instruction_paths: list<item: null>
      child 0, item: null
  child 14, job_id: string
to
{'id': Value('string'), 'task_name': Value('string'), 'trial_name': Value('string'), 'trial_uri': Value('string'), 'task_id': {'org': Value('string'), 'name': Value('string'), 'ref': Value('string')}, 'source': Value('string'), 'task_checksum': Value('string'), 'config': {'task': {'path': Value('null'), 'git_url': Value('null'), 'git_commit_id': Value('null'), 'name': Value('string'), 'ref': Value('string'), 'overwrite': Value('bool'), 'download_dir': Value('null'), 'source': Value('string')}, 'trial_name': Value('string'), 'trials_dir': Value('string'), 'install_only': Value('bool'), 'timeout_multiplier': Value('float64'), 'agent_timeout_multiplier': Value('null'), 'verifier_timeout_multiplier': Value('null'), 'agent_setup_timeout_multiplier': Value('null'), 'environment_build_timeout_multiplier': Value('null'), 'agent': {'name': Value('string'), 'import_path': Value('null'), 'model_name': Value('string'), 'n_concurrent': Value('null'), 'concurrency_group': Value('null'), 'skills': List(Value('null')), 'override_timeout_sec': Value('null'), 'override_setup_timeout_sec': Value('null'), 'max_timeout_sec': Value('null'), 'resume_trajectory': Value('bool'), 'load_trajectory': Value('null'), 'extra_allowed_hosts': List(Value('null')), 'kwargs': {'reasoning_effort': Value('string'), 'max_turns': Value('int64')}, 'mcp_servers': List(Value('null'))}, 'environment': {'type': Value('string'), 'import_path': Value('null'), 'force_build': Value('bool'), 'delete': Value('bool'), 'cpu_enf
...
, 'override_cpus': Value('null'), 'override_memory_mb': Value('null'), 'override_storage_mb': Value('null'), 'override_gpus': Value('null'), 'override_tpu': Value('null'), 'mounts': Value('null'), 'extra_docker_compose': List(Value('null')), 'kwargs': {}, 'extra_allowed_hosts': List(Value('null'))}, 'verifier': {'override_timeout_sec': Value('null'), 'max_timeout_sec': Value('null'), 'disable': Value('bool')}, 'artifacts': List(Value('null')), 'extra_instruction_paths': List(Value('null')), 'job_id': Value('string')}, 'agent_info': {'name': Value('string'), 'version': Value('string'), 'model_info': {'name': Value('string'), 'provider': Value('null')}}, 'agent_result': {'n_input_tokens': Value('int64'), 'n_cache_tokens': Value('int64'), 'n_output_tokens': Value('int64'), 'cost_usd': Value('float64'), 'rollout_details': Value('null'), 'metadata': Value('null')}, 'verifier_result': {'rewards': {'score': Value('float64'), 'geometry_similarity': Value('float64'), 'cad_spec_consistency': Value('float64'), 'combined': Value('float64')}}, 'exception_info': Value('null'), 'started_at': Value('string'), 'finished_at': Value('string'), 'environment_setup': {'started_at': Value('string'), 'finished_at': Value('string')}, 'agent_setup': {'started_at': Value('string'), 'finished_at': Value('string')}, 'agent_execution': {'started_at': Value('string'), 'finished_at': Value('string')}, 'verifier': {'started_at': Value('string'), 'finished_at': Value('string')}, 'step_results': Value('null')}
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
              score: double
              geometry_similarity: double
              cad_spec_consistency: double
              combined: double
              geometry_similarity_reason: string
              cad_spec_consistency_reason: string
              task_id: struct<org: string, name: string, ref: string>
                child 0, org: string
                child 1, name: string
                child 2, ref: string
              agent_info: struct<name: string, version: string, model_info: struct<name: string, provider: null>>
                child 0, name: string
                child 1, version: string
                child 2, model_info: struct<name: string, provider: null>
                    child 0, name: string
                    child 1, provider: null
              task_name: string
              started_at: string
              verifier_result: struct<rewards: struct<score: double, geometry_similarity: double, cad_spec_consistency: double, com (... 15 chars omitted)
                child 0, rewards: struct<score: double, geometry_similarity: double, cad_spec_consistency: double, combined: double>
                    child 0, score: double
                    child 1, geometry_similarity: double
                    child 2, cad_spec_consistency: double
                    child 3, combined: double
              finished_at: string
              agent_setup: struct<started_at: string, finished_at: string>
                child 0, started_at: string
                child 1, finished_at: string
              verifier: struct<started_at: string, finished_at: string>
                child 0, started_at: string
                child 1, finished_at: string
              agent_result: struct<n_input_tokens: int64, n_cache_tokens: int64, n_output_tokens: int64, cost_usd: double, rollo (... 33 chars omitted)
                child 0, n_input_tokens: int64
                child 1, n_cache_tokens: int64
                child 2, n_output_tokens: int64
                
              ...
              10, load_trajectory: null
                    child 11, extra_allowed_hosts: list<item: null>
                        child 0, item: null
                    child 12, kwargs: struct<reasoning_effort: string, max_turns: int64>
                        child 0, reasoning_effort: string
                        child 1, max_turns: int64
                    child 13, mcp_servers: list<item: null>
                        child 0, item: null
                child 10, environment: struct<type: string, import_path: null, force_build: bool, delete: bool, cpu_enforcement_policy: str (... 265 chars omitted)
                    child 0, type: string
                    child 1, import_path: null
                    child 2, force_build: bool
                    child 3, delete: bool
                    child 4, cpu_enforcement_policy: string
                    child 5, memory_enforcement_policy: string
                    child 6, override_cpus: null
                    child 7, override_memory_mb: null
                    child 8, override_storage_mb: null
                    child 9, override_gpus: null
                    child 10, override_tpu: null
                    child 11, mounts: null
                    child 12, extra_docker_compose: list<item: null>
                        child 0, item: null
                    child 13, kwargs: struct<>
                    child 14, extra_allowed_hosts: list<item: null>
                        child 0, item: null
                child 11, verifier: struct<override_timeout_sec: null, max_timeout_sec: null, disable: bool>
                    child 0, override_timeout_sec: null
                    child 1, max_timeout_sec: null
                    child 2, disable: bool
                child 12, artifacts: list<item: null>
                    child 0, item: null
                child 13, extra_instruction_paths: list<item: null>
                    child 0, item: null
                child 14, job_id: string
              to
              {'id': Value('string'), 'task_name': Value('string'), 'trial_name': Value('string'), 'trial_uri': Value('string'), 'task_id': {'org': Value('string'), 'name': Value('string'), 'ref': Value('string')}, 'source': Value('string'), 'task_checksum': Value('string'), 'config': {'task': {'path': Value('null'), 'git_url': Value('null'), 'git_commit_id': Value('null'), 'name': Value('string'), 'ref': Value('string'), 'overwrite': Value('bool'), 'download_dir': Value('null'), 'source': Value('string')}, 'trial_name': Value('string'), 'trials_dir': Value('string'), 'install_only': Value('bool'), 'timeout_multiplier': Value('float64'), 'agent_timeout_multiplier': Value('null'), 'verifier_timeout_multiplier': Value('null'), 'agent_setup_timeout_multiplier': Value('null'), 'environment_build_timeout_multiplier': Value('null'), 'agent': {'name': Value('string'), 'import_path': Value('null'), 'model_name': Value('string'), 'n_concurrent': Value('null'), 'concurrency_group': Value('null'), 'skills': List(Value('null')), 'override_timeout_sec': Value('null'), 'override_setup_timeout_sec': Value('null'), 'max_timeout_sec': Value('null'), 'resume_trajectory': Value('bool'), 'load_trajectory': Value('null'), 'extra_allowed_hosts': List(Value('null')), 'kwargs': {'reasoning_effort': Value('string'), 'max_turns': Value('int64')}, 'mcp_servers': List(Value('null'))}, 'environment': {'type': Value('string'), 'import_path': Value('null'), 'force_build': Value('bool'), 'delete': Value('bool'), 'cpu_enf
              ...
              , 'override_cpus': Value('null'), 'override_memory_mb': Value('null'), 'override_storage_mb': Value('null'), 'override_gpus': Value('null'), 'override_tpu': Value('null'), 'mounts': Value('null'), 'extra_docker_compose': List(Value('null')), 'kwargs': {}, 'extra_allowed_hosts': List(Value('null'))}, 'verifier': {'override_timeout_sec': Value('null'), 'max_timeout_sec': Value('null'), 'disable': Value('bool')}, 'artifacts': List(Value('null')), 'extra_instruction_paths': List(Value('null')), 'job_id': Value('string')}, 'agent_info': {'name': Value('string'), 'version': Value('string'), 'model_info': {'name': Value('string'), 'provider': Value('null')}}, 'agent_result': {'n_input_tokens': Value('int64'), 'n_cache_tokens': Value('int64'), 'n_output_tokens': Value('int64'), 'cost_usd': Value('float64'), 'rollout_details': Value('null'), 'metadata': Value('null')}, 'verifier_result': {'rewards': {'score': Value('float64'), 'geometry_similarity': Value('float64'), 'cad_spec_consistency': Value('float64'), 'combined': Value('float64')}}, 'exception_info': Value('null'), 'started_at': Value('string'), 'finished_at': Value('string'), 'environment_setup': {'started_at': Value('string'), 'finished_at': Value('string')}, 'agent_setup': {'started_at': Value('string'), 'finished_at': Value('string')}, 'agent_execution': {'started_at': Value('string'), 'finished_at': Value('string')}, 'verifier': {'started_at': Value('string'), 'finished_at': Value('string')}, 'step_results': Value('null')}
              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.

Parametric CAD Bench — Claude Fable 5 results

Run artifacts for claude-fable-5 driven by the claude-code agent on gnucleus-ai/cad-bench@v1 (the gNucleus Parametric CAD Bench), graded by the project's own freecad-validator.

This is an independent third-party submission. The layout follows the cad-bench-submission contract exactly, mirroring gnucleus-ai/cad-gen-freecad-bench.

Headline

Metric	Value
Combined score (mean over 100 tasks)	0.7577
Geometry similarity	0.7496
CAD spec consistency	0.8625
Tasks scored 1.000	42 / 100
Trials per task	1
Total trials	100 (0 errors)
API-equivalent cost	$46.71

Run configuration


Agent	`claude-code` v2.1.214
Model	`claude-fable-5`
Reasoning effort	`low`
Max turns	50
Dataset	`gnucleus-ai/cad-bench` (pinned `sha256:22be8aa8…bdf862e`)
Harness	`harbor run -d gnucleus-ai/cad-bench@v1 -a claude-code -m claude-fable-5`
Date	2026-07-18

Billing note. The run executed on a Claude Code Max ($200/mo) subscription, so marginal API cost was $0. The $46.71 figure is the API-equivalent, re-derived from measured token counts (17.53M input, of which 16.66M cache-read; 425.8K output) at published 2026-07 Anthropic per-MTok prices, cache-read tokens billed at the cache-read rate — the same re-derivation gNucleus maintainers apply.

Layout

runs/claude-code/claude-fable-5/<task_id>/
  result.json        # trial metadata, token counts, timing (secrets scrubbed)
  config.json        # agent kwargs + dataset ref (env block removed)
  reward.json        # geometry_similarity, cad_spec_consistency, combined
  answer.FCStd       # parametric CAD file the agent saved
  answer.py          # script that produced it
  agent.log          # full agent transcript
  trajectory.jsonl   # tool-use / model-call trace, one JSON object per line

All 100 freecad-* task ids are present.

Privacy

Subscription OAuth tokens and per-trial env blocks were stripped from every file before publication; transcripts and trajectories were scanned for credentials, emails, and local paths (none present).

Reproduce

pip install harbor  # gnucleus cad-bench harness
harbor run -d gnucleus-ai/cad-bench@v1 -a claude-code -m claude-fable-5 \
  --agent-kwargs reasoning_effort=low,max_turns=50

License

Apache-2.0, matching the gNucleus Parametric CAD Bench ecosystem.

Downloads last month: 54

Total file size:

21.9 MB