Dataset Viewer

The dataset viewer is not available for this subset.

Cannot get the split names for the config 'default' of the dataset.

Exception:    SplitsNotFoundError
Message:      The split names could not be parsed from the dataset config.
Traceback:    Traceback (most recent call last):
                File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 286, in get_dataset_config_info
                  for split_generator in builder._split_generators(
                                         ~~~~~~~~~~~~~~~~~~~~~~~~~^
                      StreamingDownloadManager(base_path=builder.base_path, download_config=download_config)
                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                  )
                  ^
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/webdataset/webdataset.py", line 80, in _split_generators
                  raise ValueError(
                  ...<2 lines>...
                  )
              ValueError: The TAR archives of the dataset should be in WebDataset format, but the files in the archive don't share the same prefix or the same types.
              
              The above exception was the direct cause of the following exception:
              
              Traceback (most recent call last):
                File "/src/services/worker/src/worker/job_runners/config/split_names.py", line 68, in compute_split_names_from_streaming_response
                  for split in get_dataset_split_names(
                               ~~~~~~~~~~~~~~~~~~~~~~~^
                      path=dataset,
                      ^^^^^^^^^^^^^
                      config_name=config,
                      ^^^^^^^^^^^^^^^^^^^
                      token=hf_token,
                      ^^^^^^^^^^^^^^^
                  )
                  ^
                File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 340, in get_dataset_split_names
                  info = get_dataset_config_info(
                      path,
                  ...<6 lines>...
                      **config_kwargs,
                  )
                File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 291, in get_dataset_config_info
                  raise SplitsNotFoundError("The split names could not be parsed from the dataset config.") from err
              datasets.inspect.SplitsNotFoundError: The split names could not be parsed from the dataset config.

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.

YAML Metadata Warning:empty or missing yaml metadata in repo card

Check out the documentation for more information.

MA-bench sweep results

Dedicated dump for the MA-bench meta-optimizer sweep. Supersedes the artifacts previously scattered in simonycl/temp_file_1.

Snapshot 2026-08-17 21:23 · 155 sealed cells (126 without validity flags).

A cell is one (benchmark x method x seed-agent tier x proposer arm x RNG seed) run. The method is the only thing that varies; worker model, budget and evaluator are held fixed.

Layout

ma-bench/
  RESULTS.md                     per-cell scores, spend, validity
  cells/live/<arm>_<bench>_<tier>_<method>_seed<N>.tar.gz
  cells/cro/<bench>_<tier>_<arm>_seed<N>.tar.gz
  recovery/                      codex CLI session rollouts (see below)

Each tarball is a complete run dir:

path	contents
`summary.json`	scores, spend, provenance, validity flags
`trace/*.jsonl`	full event stream: rollouts, evals, method logs, LLM effects
`trace/codex-*.jsonl`	raw codex proposer session streams (sol arm)
`workspace/`	agent source, candidates, history/, RESUME.json

arm = which model proposed: sol = gpt-5.6-sol (via codex CLI), opus5 = claude-opus-5 (via Claude Code)
worker (the agent being optimised) is gpt-5.6-luna in every cell
charxiv images are replaced with <IMAGE_STRIPPED:NB>; all archives are credential-scrubbed

Coverage

bench	method	cells
charxiv	adaevolve	6
charxiv	cro	18
charxiv	gepa	18
charxiv	mh	14
gpqa	adaevolve	6
gpqa	cro	18
gpqa	gepa	18
gpqa	mh	18
tau2	adaevolve	6
tau2	gepa	18
tau2	mh	15

Reading the results — caveats that change interpretation

method_launch:TimeoutExpired is not a failure. It means the cell hit the 72h wall-clock backstop; its score is a legitimate anytime result from a truncated search.
Search depth is not uniform. Infrastructure incidents (2026-08-10 through -17) cost some cells hours, and cells resumed after them received a fresh 72h ceiling. Compare methods within a seed before comparing across seeds.
invalid non-empty means read the reason first. It fires correctly for runs whose worker never executed, but ALSO false-positives when a method proposes a candidate that fails deterministically — the guard compares reps across the run, not within a candidate.
Cells marked rescored had their final test evaluation re-run after a bridge outage corrupted the original measurement. The search itself was untouched; the pre-repair summary is preserved inside the tarball as summary.pre-rescore.json.
CRO genesis_test_score is 0.0 (genesis_basis: first-stamp). Absolute test scores are usable; improvement deltas computed against genesis are not.
tau2 user simulator is gpt-5.4-mini, not the paper's gpt-4.1 (that lane was quota exhausted, then unavailable). Uniform across every tau2 cell in this snapshot.

recovery/

The sol arm's codex proposer interiors were not archived before 2026-08-15 (the parser summarised the session stream to counters and discarded the content). They are recoverable: codex persists its own rollouts, and session_meta.cwd names the cell while session_id matches llm_response_received.usage_details.session_id in the mab2 trace.

Downloads last month: 24

Total file size:

5.88 GB