Datasets:

connacher-silicoase
/

MEAD-swarm-hackathon

Languages:

English

Tags:

License:

Dataset card Files Files and versions

xet

Community

Dataset Viewer

Duplicate

Split (1)

train

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
source_batch_id: string
source_folder: string
changes: list<item: struct<path: string, action: string, source_sha256: string, release_sha256: string>>
  child 0, item: struct<path: string, action: string, source_sha256: string, release_sha256: string>
      child 0, path: string
      child 1, action: string
      child 2, source_sha256: string
      child 3, release_sha256: string
original_checksums: string
archives: string
all_agents_submitted: bool
data_license: string
packaged_at: string
public_release_date: timestamp[s]
version: string
source_files_sha256: struct<LICENSE: string, README.md: string, aware-001/artifacts.tar: string, aware-001/checksums.json (... 25318 chars omitted)
  child 0, LICENSE: string
  child 1, README.md: string
  child 2, aware-001/artifacts.tar: string
  child 3, aware-001/checksums.json: string
  child 4, aware-001/config.json: string
  child 5, aware-001/events.jsonl: string
  child 6, aware-001/identities.json: string
  child 7, aware-001/note_ownership.json: string
  child 8, aware-001/objective_reference.json: string
  child 9, aware-001/review.html: string
  child 10, aware-001/run_metadata.json: string
  child 11, aware-001/runtime.json: string
  child 12, aware-001/summary.json: string
  child 13, aware-002/artifacts.tar: string
  child 14, aware-002/checksums.json: string
  child 15, aware-002/config.json: string
  child 16, aware-002/events.jsonl: string
  child 17, aware-002/identities.json: string
  child 18, aware-002/note_ownership
...
facts.tar: string
  child 676, interact-019/checksums.json: string
  child 677, interact-019/config.json: string
  child 678, interact-019/events.jsonl: string
  child 679, interact-019/identities.json: string
  child 680, interact-019/note_ownership.json: string
  child 681, interact-019/objective_reference.json: string
  child 682, interact-019/review.html: string
  child 683, interact-019/run_metadata.json: string
  child 684, interact-019/runtime.json: string
  child 685, interact-019/summary.json: string
  child 686, interact-020/artifacts.tar: string
  child 687, interact-020/checksums.json: string
  child 688, interact-020/config.json: string
  child 689, interact-020/events.jsonl: string
  child 690, interact-020/identities.json: string
  child 691, interact-020/note_ownership.json: string
  child 692, interact-020/objective_reference.json: string
  child 693, interact-020/review.html: string
  child 694, interact-020/run_metadata.json: string
  child 695, interact-020/runtime.json: string
  child 696, interact-020/summary.json: string
  child 697, manifest.json: string
repository_renamed_on: timestamp[s]
display_title: string
provider_identifiers: string
counts: struct<runs: int64, agents: int64, measurements: int64>
  child 0, runs: int64
  child 1, agents: int64
  child 2, measurements: int64
initial_private_staging_revision: string
raw_path: string
layout: string
repo_id: string
release_tag: string
packager_sha256: string
batch_id: string
cc_license_source: string
to
{'version': Value('string'), 'layout': Value('string'), 'batch_id': Value('string'), 'source_folder': Value('string'), 'packaged_at': Value('string'), 'counts': {'runs': Value('int64'), 'agents': Value('int64'), 'measurements': Value('int64')}, 'all_agents_submitted': Value('bool'), 'data_license': Value('string'), 'raw_path': Value('string'), 'packager_sha256': Value('string'), 'source_files_sha256': {'LICENSE': Value('string'), 'README.md': Value('string'), 'aware-001/artifacts.tar': Value('string'), 'aware-001/checksums.json': Value('string'), 'aware-001/config.json': Value('string'), 'aware-001/events.jsonl': Value('string'), 'aware-001/identities.json': Value('string'), 'aware-001/note_ownership.json': Value('string'), 'aware-001/objective_reference.json': Value('string'), 'aware-001/review.html': Value('string'), 'aware-001/run_metadata.json': Value('string'), 'aware-001/runtime.json': Value('string'), 'aware-001/summary.json': Value('string'), 'aware-002/artifacts.tar': Value('string'), 'aware-002/checksums.json': Value('string'), 'aware-002/config.json': Value('string'), 'aware-002/events.jsonl': Value('string'), 'aware-002/identities.json': Value('string'), 'aware-002/note_ownership.json': Value('string'), 'aware-002/objective_reference.json': Value('string'), 'aware-002/review.html': Value('string'), 'aware-002/run_metadata.json': Value('string'), 'aware-002/runtime.json': Value('string'), 'aware-002/summary.json': Value('string'), 'aware-003/artifacts.tar': Value('
...
runtime.json': Value('string'), 'interact-018/summary.json': Value('string'), 'interact-019/artifacts.tar': Value('string'), 'interact-019/checksums.json': Value('string'), 'interact-019/config.json': Value('string'), 'interact-019/events.jsonl': Value('string'), 'interact-019/identities.json': Value('string'), 'interact-019/note_ownership.json': Value('string'), 'interact-019/objective_reference.json': Value('string'), 'interact-019/review.html': Value('string'), 'interact-019/run_metadata.json': Value('string'), 'interact-019/runtime.json': Value('string'), 'interact-019/summary.json': Value('string'), 'interact-020/artifacts.tar': Value('string'), 'interact-020/checksums.json': Value('string'), 'interact-020/config.json': Value('string'), 'interact-020/events.jsonl': Value('string'), 'interact-020/identities.json': Value('string'), 'interact-020/note_ownership.json': Value('string'), 'interact-020/objective_reference.json': Value('string'), 'interact-020/review.html': Value('string'), 'interact-020/run_metadata.json': Value('string'), 'interact-020/runtime.json': Value('string'), 'interact-020/summary.json': Value('string'), 'manifest.json': Value('string')}, 'provider_identifiers': Value('string'), 'cc_license_source': Value('string'), 'release_tag': Value('string'), 'public_release_date': Value('timestamp[s]'), 'initial_private_staging_revision': Value('string'), 'repo_id': Value('string'), 'display_title': Value('string'), 'repository_renamed_on': Value('timestamp[s]')}
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 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 2309, in cast_table_to_schema
                  raise CastError(
                  ...<3 lines>...
                  )
              datasets.table.CastError: Couldn't cast
              source_batch_id: string
              source_folder: string
              changes: list<item: struct<path: string, action: string, source_sha256: string, release_sha256: string>>
                child 0, item: struct<path: string, action: string, source_sha256: string, release_sha256: string>
                    child 0, path: string
                    child 1, action: string
                    child 2, source_sha256: string
                    child 3, release_sha256: string
              original_checksums: string
              archives: string
              all_agents_submitted: bool
              data_license: string
              packaged_at: string
              public_release_date: timestamp[s]
              version: string
              source_files_sha256: struct<LICENSE: string, README.md: string, aware-001/artifacts.tar: string, aware-001/checksums.json (... 25318 chars omitted)
                child 0, LICENSE: string
                child 1, README.md: string
                child 2, aware-001/artifacts.tar: string
                child 3, aware-001/checksums.json: string
                child 4, aware-001/config.json: string
                child 5, aware-001/events.jsonl: string
                child 6, aware-001/identities.json: string
                child 7, aware-001/note_ownership.json: string
                child 8, aware-001/objective_reference.json: string
                child 9, aware-001/review.html: string
                child 10, aware-001/run_metadata.json: string
                child 11, aware-001/runtime.json: string
                child 12, aware-001/summary.json: string
                child 13, aware-002/artifacts.tar: string
                child 14, aware-002/checksums.json: string
                child 15, aware-002/config.json: string
                child 16, aware-002/events.jsonl: string
                child 17, aware-002/identities.json: string
                child 18, aware-002/note_ownership
              ...
              facts.tar: string
                child 676, interact-019/checksums.json: string
                child 677, interact-019/config.json: string
                child 678, interact-019/events.jsonl: string
                child 679, interact-019/identities.json: string
                child 680, interact-019/note_ownership.json: string
                child 681, interact-019/objective_reference.json: string
                child 682, interact-019/review.html: string
                child 683, interact-019/run_metadata.json: string
                child 684, interact-019/runtime.json: string
                child 685, interact-019/summary.json: string
                child 686, interact-020/artifacts.tar: string
                child 687, interact-020/checksums.json: string
                child 688, interact-020/config.json: string
                child 689, interact-020/events.jsonl: string
                child 690, interact-020/identities.json: string
                child 691, interact-020/note_ownership.json: string
                child 692, interact-020/objective_reference.json: string
                child 693, interact-020/review.html: string
                child 694, interact-020/run_metadata.json: string
                child 695, interact-020/runtime.json: string
                child 696, interact-020/summary.json: string
                child 697, manifest.json: string
              repository_renamed_on: timestamp[s]
              display_title: string
              provider_identifiers: string
              counts: struct<runs: int64, agents: int64, measurements: int64>
                child 0, runs: int64
                child 1, agents: int64
                child 2, measurements: int64
              initial_private_staging_revision: string
              raw_path: string
              layout: string
              repo_id: string
              release_tag: string
              packager_sha256: string
              batch_id: string
              cc_license_source: string
              to
              {'version': Value('string'), 'layout': Value('string'), 'batch_id': Value('string'), 'source_folder': Value('string'), 'packaged_at': Value('string'), 'counts': {'runs': Value('int64'), 'agents': Value('int64'), 'measurements': Value('int64')}, 'all_agents_submitted': Value('bool'), 'data_license': Value('string'), 'raw_path': Value('string'), 'packager_sha256': Value('string'), 'source_files_sha256': {'LICENSE': Value('string'), 'README.md': Value('string'), 'aware-001/artifacts.tar': Value('string'), 'aware-001/checksums.json': Value('string'), 'aware-001/config.json': Value('string'), 'aware-001/events.jsonl': Value('string'), 'aware-001/identities.json': Value('string'), 'aware-001/note_ownership.json': Value('string'), 'aware-001/objective_reference.json': Value('string'), 'aware-001/review.html': Value('string'), 'aware-001/run_metadata.json': Value('string'), 'aware-001/runtime.json': Value('string'), 'aware-001/summary.json': Value('string'), 'aware-002/artifacts.tar': Value('string'), 'aware-002/checksums.json': Value('string'), 'aware-002/config.json': Value('string'), 'aware-002/events.jsonl': Value('string'), 'aware-002/identities.json': Value('string'), 'aware-002/note_ownership.json': Value('string'), 'aware-002/objective_reference.json': Value('string'), 'aware-002/review.html': Value('string'), 'aware-002/run_metadata.json': Value('string'), 'aware-002/runtime.json': Value('string'), 'aware-002/summary.json': Value('string'), 'aware-003/artifacts.tar': Value('
              ...
              runtime.json': Value('string'), 'interact-018/summary.json': Value('string'), 'interact-019/artifacts.tar': Value('string'), 'interact-019/checksums.json': Value('string'), 'interact-019/config.json': Value('string'), 'interact-019/events.jsonl': Value('string'), 'interact-019/identities.json': Value('string'), 'interact-019/note_ownership.json': Value('string'), 'interact-019/objective_reference.json': Value('string'), 'interact-019/review.html': Value('string'), 'interact-019/run_metadata.json': Value('string'), 'interact-019/runtime.json': Value('string'), 'interact-019/summary.json': Value('string'), 'interact-020/artifacts.tar': Value('string'), 'interact-020/checksums.json': Value('string'), 'interact-020/config.json': Value('string'), 'interact-020/events.jsonl': Value('string'), 'interact-020/identities.json': Value('string'), 'interact-020/note_ownership.json': Value('string'), 'interact-020/objective_reference.json': Value('string'), 'interact-020/review.html': Value('string'), 'interact-020/run_metadata.json': Value('string'), 'interact-020/runtime.json': Value('string'), 'interact-020/summary.json': Value('string'), 'manifest.json': Value('string')}, 'provider_identifiers': Value('string'), 'cc_license_source': Value('string'), 'release_tag': Value('string'), 'public_release_date': Value('timestamp[s]'), 'initial_private_staging_revision': Value('string'), 'repo_id': Value('string'), 'display_title': Value('string'), 'repository_renamed_on': Value('timestamp[s]')}
              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.

Model Environment for Agent Discovery (MEAD)

This raw-first release contains the completed batch and a documented host-path-sanitized derivative. See RELEASE_AUDIT.md, TRANSFORMATIONS.json and NOTICE.md. The original launch-time card remains in the raw snapshot.

Release information

Field	Value
Hugging Face repository / visibility	connacher-silicoase/swarm-hackathon-demo, released 2026-10-04
Release version / immutable revision	v1.0.0; tag pins the release commit
Creators / affiliations	Connacher Murphy — Silicoase
Public contact	connacher@silicoase.com
Generated-data license	CC BY-NC 4.0 for generated research data since 2026-10-28 (v1.0.0: CC BY 4.0); MAD and third-party material retain their own terms
Collection interval (UTC) / batch IDs	2026-10-04 21:15:02.946689–23:24:38.526392 UTC; manifest batch ID `20261004T211502Z-442d48aa`
Planned runs / agent slots / measurement maximum	60 / 240 / 2,400 accepted measurements
Actual runs by status / agent sessions with reports	60 finished runs; 240 agent reports; no pending/running/error/cancelled runs
Submissions / accepted measurements / completed measurements	240 confirmed submissions / 2,400 accepted / 2,400 completion timestamps / 2,400 confirmed lab publications
Complete matched repetitions / collection incidents	20 complete matched repetitions; no submission inconsistencies, missing expected run files, or checksum mismatches found in final audit
Release layout / exclusions / redactions	Raw batch derivative; host paths sanitized, runtime bytecode caches omitted; see TRANSFORMATIONS.json and NOTICE.md
Citation / results writeup	Connacher Murphy; see versioned citation below; results writeup to follow

Dataset purpose

This synthetic optimization experiment supports examination of how prompt wording about peer awareness and interaction affects independent agents working in shared labs. It is a hackathon demo, not a general benchmark of scientific discovery, multi-agent systems, or models. Suggested uses include inspecting trajectories, measurements, artifacts, and matched condition comparisons. There are no predefined training, validation, or test splits.

Collection design

The definition plans 20 repetitions of control, aware, and interact, with four agents per run. The configured model is gpt-6.1-sol, with provider reasoning summaries requested as auto. Each agent has at most 10 accepted measurements, 80 SDK turns, and a 1,800-second session timeout. Measurements incur a 15-second delay; agents start simultaneously. The objective has eight parameters bounded by [0,1], with Gaussian observation noise SD 0.1. Submission is separate from measurement and is evaluated on the noiseless objective.

All conditions have the same shared lab record/note access and tools, including Python. Control asks for notebook writing and periodic lab checks; aware adds explicit peer awareness; interact additionally permits interaction. There are no direct messaging or cross-agent handoff tools: interaction can occur through shared artifacts. These conditions do not isolate the effect of granting access to peers' files. Exact prompts and settings are preserved in the frozen inputs/experiment/ tree of each batch.

Repetition r uses objective seed 43+r and noise seed 124+r (44–63 / 125–144). Noise streams are matched by internal agent index; adaptive measurement choices mean this does not imply identical observations. Conditions run concurrently in isolated labs; repetitions run sequentially. UUIDs, model outputs, and timing are not matched. API/infrastructure errors stop later repetitions; the runner does not automatically retry whole runs. This released batch completed all planned runs; actual dates and counts are recorded above. Provider-returned model identifiers and settings remain available in the raw event records.

Release organization and schemas

Raw records are under raw/20261004T211502Z-442d48aa/. Download this subtree to preserve relative viewer links. Open index.html or individual review.html files locally. There are no Parquet tables or predefined training/test splits in this release. See SCHEMA.md for raw file formats. Later revisions may add analysis tables; cite the release revision actually used.

Raw files (availability depends on run progress and export success)

File	Structure / provenance
`manifest.json`	Batch object: experiment definition schema version 2, batch ID/status/timestamps, matching text, provenance, and planned `runs` entries with condition/repetition/seeds/status and optional outcomes/errors
`inputs/`	Frozen experiment configs/prompts/docs, MAD source and viewers, Docker/scripts, dependency lockfile, code license; SHA-256 map
`<run_id>/config.json`	Fully resolved rollout/task/harness/environment settings; local host paths sanitized
`<run_id>/runtime.json`	Python/platform, Agents SDK version, Docker image ID
`<run_id>/identities.json`	List of `{agent, author_id, docker_uid}`; synthetic UUIDs, not human identifiers
`<run_id>/events.jsonl`	Ordered event objects: `sequence`, UTC `timestamp`, `elapsed_seconds`, `kind`, nullable `agent`, and kind-specific payload
`<run_id>/summary.json`	Object with `harness`, agent reports, per-agent lists `experiment_records`, and submission mapping; may be missing on early startup failure
`<run_id>/run_metadata.json`	Schema version 1: experiment/batch/run identity, status, seeds, timing and optional errors; later viewer errors may only appear in manifest
`<run_id>/artifacts.tar`	Final lab state and agent workspaces; may contain agent-created files and symlinks
`<run_id>/note_ownership.json`	Lab note path to internal agent-index mapping
`checksums.json`	Relative file path to SHA-256 digest; runner maps exclude all files named `checksums.json`
`index.html`, `<run_id>/review.html`	Batch comparison and run viewers; run viewer depends on export success and carries MAD software/branding terms

Events cover rollout/agent lifecycle, interfaces, model requests/responses, provider metadata, tool calls/results/cancellations, accepted/completed experiments, and submissions. Requests include system prompts, inputs and SDK settings; null SDK values leave defaults unspecified. Responses preserve available outputs, usage, response IDs and provider reasoning summaries. Provider metadata records returned model/settings/request ID when available, linked by response ID; identifiers may still be aliases. Raw events do not have a separately declared event-schema version. Exact payloads depend on the frozen source and SDK versions, which must be consulted when parsing.

Incomplete runs and missingness

Collection inclusion rule: release only a batch with all 60 planned run statuses finished; partial batches remain local. Separately audit all 240 agent slots for submission, reconciling termination reasons, final parameters/scores, summary submission mappings and recorded submission events. Report every exception before release; run completion alone does not establish agent submission. Early submission without spending all measurement budget is allowed. Preserve all records within the selected batch; disclose reruns and selection criteria if applicable. Report run status and agent termination separately. Run finished does not imply that all four agents submitted: time/turn limits and final answers without submission can still yield that status. pending in a stopped batch can mean never launched; an interrupted process can leave stale running records. Missing data is not evidence of a zero score or zero usage.

An accepted measurement consumes budget before its delay and can be interrupted. completed_at is assigned before lab publication, while experiment_completed is recorded after publication succeeds. Report accepted counts, timestamp-based completion counts, and publication evidence distinctly when they differ. Final true_objective is null without a submission; best_observed uses completed noisy observations and is not the final true score. HTML score summaries exclude agents without submissions. Comparisons must show submission attrition, respect shared-lab dependencies, and disclose excluded repetitions rather than silently selecting successful records. In the audited batch, all 240 agents submitted and all 2,400 measurements completed with publication events. No missing expected run files were found. Preserve the manifest batch ID separately from the raw folder name 20261004T211502Z-03c650a5, whose suffix differs.

Limitations and responsible interpretation

The dataset concerns one configured model, one synthetic objective family, and three prompt conditions in a small planned demo. It cannot establish broad claims about real-world scientific discovery or causal communication mechanisms. Agents in one lab share information and are dependent. Model randomness, model alias changes, UUIDs, API behavior and concurrent scheduling prevent exact replay even with matched seeds and frozen inputs. An image ID identifies the local Docker image but does not by itself distribute that image.

Provider summaries are not complete hidden reasoning. Python file accesses are not individually traced; file-tool flags in viewers are heuristic navigation aids, not a complete measure of interaction. Notes/artifacts and model outputs may contain incorrect claims or executable code; they are research records, not verified instructions. Repeated noisy evaluations and selected maxima can inflate observed-best statistics. Report collection interruptions and every release transformation. No additional collection incidents were identified by the recorded audit.

Licensing, privacy, and release transformations

Generated experiment data: Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0) since 2026-10-28 (LICENSE). The v1.0.0 release offered it under CC BY 4.0, selected on 2026-10-04; copies obtained under that grant keep it, and RELEASE.json and RELEASE_AUDIT.md describe v1.0.0 as released. See DATA_LICENSE.md for scope and attribution. The existing MAD license does not automatically license generated outputs. The release inventory must identify copied third-party material and retain applicable notices. MAD source, copied documentation/prompts and embedded HTML viewer code retain the Silicoase Noncommercial License, Version 2.0 and its Required Notice; brand assets retain its branding restrictions. See release NOTICE.md and LICENSES/MAD-LICENSE.txt. Do not apply the generated-data license to those portions merely by putting one license in the YAML header.

Input snapshots exclude credentials by an explicit source allowlist, but this does not certify that every error, free-text output, config path or archived artifact is suitable for publication. The initial scan found local host-path patterns in 244 files; credential-like matches were confined to opaque provider encrypted-content fields, and email matches were classified as Python code. See RELEASE_AUDIT.md for scope and limits. Local host paths were replaced by /MAD_REPOSITORY or /LOCAL_USER/ in the release copy; generated source-bytecode caches were omitted. Original files remain private. Original checksum maps are preserved under provenance/original-checksums/; the derivative has new checksum maps. Provider response/request IDs and encrypted provider payloads are retained as recorded provenance; no access tokens were identified by the scan. See TRANSFORMATIONS.json for every modified/omitted file.

Citation and contact

Author: Connacher Murphy. Affiliation: Silicoase. Public contact: connacher@silicoase.com. Software and experiment source: the MEAD source used for this batch is included at raw/20261004T211502Z-442d48aa/inputs/code/ under the Silicoase Noncommercial License, Version 2.0 (LICENSES/MAD-LICENSE.txt).

Suggested citation: Murphy, Connacher (2026). swarm-hackathon-demo, v1.0.0. Silicoase. https://huggingface.co/datasets/connacher-silicoase/MEAD-swarm-hackathon. Use the v1.0.0 release tag or its immutable Hub commit when citing this release. No DOI is assigned. The results writeup will be linked separately when finalized.

Hub packaging references: dataset cards, explicit data configurations, and license metadata.

Project naming

The dataset repository is now MEAD-swarm-hackathon — Model Environment for Agent Discovery. swarm-hackathon-demo remains the experiment identifier in the recorded data. The original v1.0.0 tag is unchanged; these landing-page naming updates are on the main branch and do not alter measurements, submissions, or archives. Links to the former MEAD GitHub repository were later removed from this card and DATA_LICENSE.md; the source remains available in the raw snapshot.

Downloads last month: 208

Total file size:

625 MB