Repository navigation
Publish v3 as the airops-v3 bundle on the b6-v3 pack - #3
Open
agutmanstein-scale wants to merge 4 commits into
Open
agutmanstein-scale wants to merge 4 commits into
agutmanstein-scale wants to merge 4 commits into
Conversation
v3 (fork PRs #1-#12, tag v3.0) replayed onto Edgar's main at 0ac85d2 as one commit: his sim.execute became the Dispatch state machine with crew pools, and every v3 code unit edited the same dispatch loop. - sim: departure caps, ramp closures, outstation arrival caps, reserve grants and the desk's view of them move into Dispatch; pinned flights count in the layers' rows at the start and again after `switch`. - mip: v3's rows and reserve columns beside Edgar's flat pricing (a reserve- backed auto-cancel prices flat too under `flat`). - network: v3's bases and layers beside the 04:00 start and MIN_CONTROLLED. - v3_pack: a date the build's rules leave out is kept on --resume; only an error is rebuilt. - flow.sources imports PIL only where it reads radar, so the layers (and the occ CLI through them) import on a clean install. - tests: from 04:00 the slice date has several JFK closures and Edgar's crew model times out one leg; the pack test takes four winter months, as v2's split trains only events without an eval date. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
main's push runs have been cancelled at the 20-minute limit since 0.3.2's tests landed (five in a row from 2026-10-20 08:03Z), with no test failing; a passing leg takes 15-19 minutes, so whether a run finishes was down to the runner. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- data/b6-v3: the rebuilt v3 pack (94 storm dates, 2 controls, from 04:00) gzipped, with each task's stored anchor and follower costs and a manifest; scripts/ship_v3_pack.py writes it from a v3_pack.py --refs folder. - `agent-env airops-v3 setup` builds Dockerfile.v3 (airops-v3-day with b6-v3) and registers it, --agent airops-llm too; `airops-v3 tasks` writes the bundle: one task per day, each load carrying its stored costs, v1's verifier, wiring-noplay. The package ships the eval days as bundles/airops-v3. - hub_dataset.py adds bundles/airops-v3 and its needs to the dataset card, so `agent-env hf run --bundle airops-v3` plays it from the next release. - load_tasks reads tasks.jsonl.gz. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
agutmanstein-scale
force-pushed
the
v3/hf-bundle
branch
from
October 21, 2026 15:49
898259c to
18c5927
Compare
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Stacked on #2. Until #2 merges, this PR also shows #2's 3 commits; review only the last one (18c5927).
Motivation
This publishes v3 the way v2's
airops-occis published: a packaged pack, its own env image,setup/taskscommands and a bundle in the dataset. With it,agent-env hf run earakely-scale/AirOpsEnv-AgentEnv --bundle airops-v3plays a v3 day from the next release.What's in it
data/b6-v3(6 MB): the v3 pack rebuilt on 0.3.2 (94 storm dates: 24 eval and 70 train; plus 2 controls; from 04:00).tasks.jsonl.gz(mtime 0),refs.jsonwith each task's anchor and follower costs, and the task digest each one ran on.scripts/ship_v3_pack.pywrites it from av3_pack.py --refsfolder, and refuses a task without a stored anchor run on that very task.Dockerfile.v3(and its dockerignore):airops-v3-daywith b6-v3, on Dockerfile.occ's base image and dependency pins.agent-env airops-v3(v3day/cli.py, entry pointsairops-v3):setup [--agent]builds and registers the env, and airops-llm with--agent.taskswrites the bundle: one task per day, each load carrying its stored costs so it never waits on a solve, v1's verifier (it readsdone,auditandreward, which v3'sdata/getreports), andwiring-noplay.bundles/airops-v3.scripts/hub_dataset.py: addsbundles/airops-v3and its needs to the card'sagentenvtable (--v3-pack, default b6-v3). Nothing is pushed until you run it with--repoand--tag.load_tasksreadstasks.jsonl.gz. The README gets a v3 section.Test plan
pytest: 1221 passed, 13 skipped.tests/v3day/test_v3_bundle.py, which checks that the packaged bundle is byte for byte whatgeneratewrites, and that every shipped task's stored anchor ran on that task.test_cli_importscoversairops-v3.agent-env airops-v3 setupbuilt and registeredairops-v3-day.agent-env run airops-v3 --task wiring-noplayran all 5 steps (deploy, load, clock, end-day, grade) in 28.5 s and graded 0.693. That is the do-nothing grade: the day flown with no change.v3_pack.py --resume --only 2022-12-15,2024-06-30) give the shipped lines byte for byte.Security impact
None: public inputs only, no credentials. Nothing is published to the Hub by this PR.
🤖 Generated with Claude Code