The dataset viewer is not available for this split.
Error code: StreamingRowsError
Exception: ValueError
Message: Expected object or value
Traceback: Traceback (most recent call last):
File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 290, in _generate_tables
pa_table = paj.read_json(
io.BytesIO(batch), read_options=paj.ReadOptions(block_size=block_size)
)
File "pyarrow/_json.pyx", line 342, in pyarrow._json.read_json
File "pyarrow/error.pxi", line 155, in pyarrow.lib.pyarrow_internal_check_status
File "pyarrow/error.pxi", line 92, in pyarrow.lib.check_status
raise convert_status(status)
pyarrow.lib.ArrowInvalid: JSON parse error: Column(/question_keys/[]/[]) changed from string to number in row 0
During handling of the above exception, another exception occurred:
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 304, in _generate_tables
batch = json_encode_fields_in_json_lines(original_batch, json_field_paths)
File "/usr/local/lib/python3.14/site-packages/datasets/utils/json.py", line 111, in json_encode_fields_in_json_lines
examples = [ujson_loads(line) for line in original_batch.splitlines()]
~~~~~~~~~~~^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/utils/json.py", line 20, in ujson_loads
return pd.io.json.ujson_loads(*args, **kwargs)
~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^
ValueError: Expected object or valueNeed help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
YAML Metadata Warning:empty or missing yaml metadata in repo card
Check out the documentation for more information.
baseline-frontier: baselines measured for the Frontier Gain frontier
1. Question and links
Read this first: this data root is published on the Hugging Face Hub only, as the public dataset repository t2ance/atlas-baseline-frontier (https://huggingface.co/datasets/t2ance/atlas-baseline-frontier), every file as it is; it has no reading copy in t2ance/atlas-experiments.
Question: where do baselines that spend more than LLM Selection with four candidates fall in cost and accuracy, so that the convexified baseline frontier of Frontier Gain is measured where it would otherwise be flat? First baseline: Self-Refine run to eight drafts. Report: personal/experiments/prompting/02-self-refine-eight-drafts/ in t2ance/ATLAS; the frontier it feeds is report 03, personal/experiments/prompting/03-frontier-gain/. Issues: t2ance/ATLAS#79, t2ance/ATLAS#80. No W&B run (inference only). Status: complete; all 821 questions have eight drafts or ended early on a timeout or a refusal. Result: 61.00, 84.57, 87.88, 20.88 at 1.701, 1.292, 1.216, 1.089 dollars a question on hle, lcb, gpqa, babyvision; an empty answer is graded wrong by rule (grader of t2ance/ATLAS d5bacc4).
2. Directory tree
<benchmark>/sonnet_self_refine_8drafts/: Self-Refine withmin_drafts: 8on hle (gold text-only, first 100), lcb (175), gpqa (198), babyvision (388), one directory per config of the same path underpersonal/experiments/prompting/baseline-frontier/.cache/<benchmark>/sonnet_self_refine_8drafts/(hle under.../gold/): the drafts (candidate_<k>) and critiques (feedback_<i>) of every question. It started as a copy ofmain-results/cache/<benchmark>/sonnet_socratic_self_refine/, so each question continues from that run's drafts and critiques.
Each run directory: run_config.json (the run configuration), state.json (per-question records and the summary), progress.json (the summary), trajectories/<qid>/ (the rendered drafts and critiques).
3. How to read each kind of file
Every file is text or JSON: <repo>/raw/main/<path>. In state.json, summary holds accuracy and mean cost; questions holds one record per question. A cache slot holds input.md (the prompt), output.md (the model's text) and result.json (the structured result with cost_usd and usage).
4. Row fields
A state.json question record: id, gold_answer, predicted_answer (the last draft's answer), is_correct, cost_usd (every draft and critique, replayed or new), judge_cost_usd, cost_by_component (explorer: the drafts; feedback: the critiques), num_explores (drafts), rounds (one entry per draft and per critique, each with its cost), elapsed_seconds.
5. Weights and inputs
No weights. Inputs: the floor-three Self-Refine cache Experiment/analysis/prompting/main-results/cache/<benchmark>/sonnet_socratic_self_refine/, copied into cache/ here and never written; the run it is compared with is Experiment/analysis/prompting/main-results/<benchmark>/sonnet_socratic_self_refine/.
- Downloads last month
- 51