Skip to content

Conversation

@relh
Copy link
Contributor

@relh relh commented Dec 22, 2025

What

  • update experiment recipes to use checkpoint directory URIs
  • adjust CVC mission variant curriculum checkpoint paths
  • refresh ABES/quantile/cloner recipes to the bundle format
  • align recipe defaults with the new checkpoint loading flow

Why

  • keep examples consistent with policy_spec bundles
  • avoid legacy .mpt or raw weight paths in recipes
  • make local and S3 checkpoints interchangeable
  • reduce friction when running evals from recipes

Copy link
Contributor Author

relh commented Dec 22, 2025

Warning

This pull request is not mergeable via GitHub because a downstack PR is open. Once all requirements are satisfied, merge this PR as a stack on Graphite.
Learn more

This stack of pull requests is managed by Graphite. Learn more about stacking.

@relh relh changed the title init Update experiment recipes to policy_spec checkpoints Dec 22, 2025
@relh relh marked this pull request as ready for review December 22, 2025 21:39
@datadog-official
Copy link

datadog-official bot commented Dec 22, 2025

⚠️ Tests

⚠️ Warnings

🧪 57 Tests failed

    test_alternate_run_format[training_facility.harvest] from test_all_games_eval.py

    test_mission_run[evals.diagnostic_agile] from test_all_games_eval.py

    test_mission_run[evals.diagnostic_agile_hard] from test_all_games_eval.py

View all

ℹ️ Info

❄️ No new flaky tests detected

This comment will be updated automatically if new data arrives.
🔗 Commit SHA: 3ad7a52 | Docs | Was this helpful? Give us feedback!

@relh relh force-pushed the mpt-artifact-relocation-v2 branch from 5710760 to 4df89a7 Compare December 22, 2025 22:42
@relh relh force-pushed the recipes-update-v2 branch 2 times, most recently from 14280dc to 1e96191 Compare December 22, 2025 22:52
@relh relh force-pushed the mpt-artifact-relocation-v2 branch from 4df89a7 to 246fffc Compare December 22, 2025 22:52
@relh relh force-pushed the recipes-update-v2 branch from 1e96191 to 44e9073 Compare December 22, 2025 22:56
@relh relh force-pushed the mpt-artifact-relocation-v2 branch 2 times, most recently from fbac88c to 7bb2b76 Compare December 22, 2025 23:00
@relh relh force-pushed the recipes-update-v2 branch 2 times, most recently from 7552087 to 13ca835 Compare December 22, 2025 23:15
@relh relh force-pushed the mpt-artifact-relocation-v2 branch from 7bb2b76 to 3f03c28 Compare December 22, 2025 23:15
@relh relh force-pushed the recipes-update-v2 branch from 13ca835 to f183bf5 Compare December 22, 2025 23:20
@relh relh force-pushed the mpt-artifact-relocation-v2 branch from 3f03c28 to 419d120 Compare December 22, 2025 23:20
@relh relh force-pushed the recipes-update-v2 branch from f183bf5 to ac9ec48 Compare December 22, 2025 23:28
@relh relh force-pushed the mpt-artifact-relocation-v2 branch from 419d120 to 7f66c78 Compare December 22, 2025 23:28
@relh relh force-pushed the recipes-update-v2 branch from ac9ec48 to 3625286 Compare December 22, 2025 23:33
@relh relh force-pushed the mpt-artifact-relocation-v2 branch from 7f66c78 to b09b28d Compare December 22, 2025 23:33
@relh relh force-pushed the recipes-update-v2 branch from 3625286 to 94305ca Compare December 23, 2025 00:03
@relh relh force-pushed the mpt-artifact-relocation-v2 branch 2 times, most recently from adb552d to f78453b Compare December 23, 2025 00:42
@relh relh force-pushed the recipes-update-v2 branch 2 times, most recently from 660ed12 to f00a69c Compare December 23, 2025 00:47
@relh relh force-pushed the mpt-artifact-relocation-v2 branch from f78453b to 1ca20f6 Compare December 23, 2025 00:47
@relh relh force-pushed the mpt-artifact-relocation-v2 branch from 4a73d38 to b686217 Compare December 23, 2025 18:54
@relh relh force-pushed the recipes-update-v2 branch from f4cc09f to 7228dc6 Compare December 23, 2025 18:54
@relh relh force-pushed the mpt-artifact-relocation-v2 branch from b686217 to 9b77c18 Compare December 23, 2025 18:57
@relh relh force-pushed the recipes-update-v2 branch 2 times, most recently from 024a193 to 4a77289 Compare December 23, 2025 19:00
@relh relh force-pushed the mpt-artifact-relocation-v2 branch 2 times, most recently from 86f651f to 8605574 Compare December 23, 2025 19:05
@relh relh force-pushed the recipes-update-v2 branch 2 times, most recently from df399f5 to 8c89221 Compare December 23, 2025 19:06
@relh relh force-pushed the mpt-artifact-relocation-v2 branch from 8605574 to 3d752e1 Compare December 23, 2025 19:06
@relh relh force-pushed the recipes-update-v2 branch from 8c89221 to 0a58265 Compare December 23, 2025 19:07
@relh relh force-pushed the mpt-artifact-relocation-v2 branch 2 times, most recently from 2b5f42d to e24540c Compare December 23, 2025 19:10
@relh relh force-pushed the recipes-update-v2 branch 3 times, most recently from aa68564 to 2cc70c2 Compare December 23, 2025 19:15
@relh relh force-pushed the mpt-artifact-relocation-v2 branch from 035ec07 to 3fdee18 Compare December 23, 2025 19:15
@relh relh changed the title Update recipes for checkpoint directory URIs Update recipes to checkpoint directory URIs Dec 23, 2025
@relh relh force-pushed the mpt-artifact-relocation-v2 branch from 3fdee18 to 9b138ca Compare December 23, 2025 20:20
@relh relh force-pushed the recipes-update-v2 branch from 2cc70c2 to 4ebc693 Compare December 23, 2025 20:20
@relh relh force-pushed the mpt-artifact-relocation-v2 branch from 9b138ca to bf1e9e0 Compare December 23, 2025 20:28
@relh relh force-pushed the recipes-update-v2 branch from 4ebc693 to 78bd3c2 Compare December 23, 2025 20:28
@relh relh force-pushed the mpt-artifact-relocation-v2 branch from bf1e9e0 to f08f91d Compare December 23, 2025 20:53
@relh relh force-pushed the recipes-update-v2 branch from 78bd3c2 to 7cf1abf Compare December 23, 2025 20:53
no more .mpt

Merge remote-tracking branch 'origin/main' into richard-unifympt

slim

policy spec handler

more concise

cleanup

simplify

Merge remote-tracking branch 'origin/main' into richard-unifympt

re-add

fix policy spex

Update packages/mettagrid/python/src/mettagrid/util/uri_resolvers/schemes.py

Co-authored-by: graphite-app[bot] <96075541+graphite-app[bot]@users.noreply.github.com>

Merge remote-tracking branch 'origin/main' into richard-unifympt

bundles

Merge remote-tracking branch 'origin/main' into richard-unifympt

Merge remote-tracking branch 'origin/main' into richard-unifympt

bundle

Merge remote-tracking branch 'origin/richard-unifympt' into richard-unifympt

Merge remote-tracking branch 'origin/main' into richard-unifympt

Merge remote-tracking branch 'origin/main' into richard-unifympt

Merge remote-tracking branch 'origin/main' into richard-unifympt

Merge remote-tracking branch 'origin/main' into richard-unifympt

Merge remote-tracking branch 'origin/main' into richard-unifympt

Merge remote-tracking branch 'origin/main' into richard-unifympt

simplify?

ugh compat

cleanup

Merge remote-tracking branch 'origin/main' into richard-unifympt

cleanup

tests

Merge remote-tracking branch 'origin/main' into richard-unifympt

more tests

Merge remote-tracking branch 'origin/main' into richard-unifympt

Merge remote-tracking branch 'origin/main' into richard-unifympt

simplify?

Merge branch 'main' into richard-unifympt

cleanup

Merge remote-tracking branch 'origin/main' into richard-unifympt

Merge remote-tracking branch 'origin/main' into richard-unifympt

Merge remote-tracking branch 'origin/main' into richard-unifympt

no more .mpt

remove all .mpt and lint

cleanup

local data path fixes

mpt re-add

re-add artifact

lint

Merge remote-tracking branch 'origin/main' into richard-unifympt

more cleanup

Merge remote-tracking branch 'origin/main' into richard-unifympt

diff cleanup

ftt

lint

fix error

Merge remote-tracking branch 'origin/main' into richard-unifympt

more tests

lint

Merge remote-tracking branch 'origin/main' into richard-unifympt

Merge remote-tracking branch 'origin/main' into richard-unifympt

checkpoint policy does save/load

lint

checkpoint moving

catcus

lint

Merge branch 'main' into richard-unifympt

fold-in

[pyright 4] Get pyright to pass on app_backend (#4478)

Merge remote-tracking branch 'origin/main' into richard-unifympt

Fix command, add space (#4456)

added space to --app:lib--tlsEmulation:off which makes it --app:lib
--tlsEmulation:off
now it runs

Rename HyperUpdateRule to ScheduleRule (#4483)

- rename HyperUpdateRule to ScheduleRule and apply to TrainerConfig via
target_path
- update recipes and teacher scheduling to use ScheduleRule
- report PPO stats using ppo_actor/ppo_critic hyperparam keys and update
tests

- not run (not requested)

---------

Co-authored-by: graphite-app[bot] <96075541+graphite-app[bot]@users.noreply.github.com>

Merge remote-tracking branch 'origin/main' into richard-unifympt

Fix supervisor teacher behavior and legacy BC mode (#4484)

- gate PPO actor during supervisor teacher phase
- fix supervisor/no-teacher behavior and add legacy BC (no gating, no
PPO resume)
- require supervisor policy URI for sliced_cloner_no_ppo

- not run (not requested)

---------

Co-authored-by: graphite-app[bot] <96075541+graphite-app[bot]@users.noreply.github.com>
Co-authored-by: Adam S <134907338+gustofied@users.noreply.github.com>

Minor fixes to the slstm triton kernel, causing failures for certain kernel sizes (#4492)

cleanup

Merge remote-tracking branch 'origin/main' into richard-unifympt

fold in

training environments and eval environments mismatched (#4487)

I ran a direct config comparison using the training entrypoint
(recipes/experiment/cogs_v_clips.train) with variants=["heart_chorus"]
and compared the eval suite
config it builds (difficulty standard + heart_chorus) for an overlapping
mission: hello_world.oxygen_bottleneck.

  Findings:

  - Compass is ON in both training and eval (global_obs.compass=True).
- Vibe count and change‑vibe settings match (152 vibes;
change_vibe.number_of_vibes=152).
- But the mission parameters differ between training and eval for the
same mission name:
      - game.objects.carbon_extractor.max_uses: train 25 vs eval 100
      - game.objects.oxygen_extractor.max_uses: train 5 vs eval 20
      - game.objects.germanium_extractor.max_uses: train 5 vs eval 20
      - game.objects.silicon_extractor.max_uses: train 120 vs eval 480

So the mismatch isn’t compass — it’s the mission definitions used by
training vs eval. Training uses base missions
(cogames.cogs_vs_clips.missions), while eval uses
integrated eval missions (cogames.cogs_vs_clips.evals.integrated_evals)
that have different extractor settings.

Also: the eval suite used by recipes/experiment/cogs_v_clips.train does
not include machina_1.open_world at all (it only evaluates the 7
HELLO_WORLD integrated
evals). So training can be creating hearts on easier missions while your
eval runs on machina_1 are a different environment entirely.

  Relevant files:

- Compass default: packages/cogames/src/cogames/cogs_vs_clips/mission.py
- Training entrypoint + eval suite wiring:
recipes/experiment/cogs_v_clips.py
- Eval mission definitions:
packages/cogames/src/cogames/cogs_vs_clips/evals/integrated_evals.py

If you want true parity, we should align which mission templates eval
uses (and/or include machina_1.open_world in the eval suite). I can
patch this if you want —
  tell me whether you prefer:

1. Eval suite uses the same mission templates as training (from
missions.py), or
  2. Training uses the integrated eval mission definitions, or
  3. Add machina_1.open_world to the eval suite.

ripping out

Merge remote-tracking branch 'origin/main' into richard-unifympt

simplify

fix and lint

choke
@relh relh force-pushed the mpt-artifact-relocation-v2 branch from f08f91d to 1c52af2 Compare December 23, 2025 20:56
@relh relh force-pushed the recipes-update-v2 branch from 7cf1abf to 3ad7a52 Compare December 23, 2025 20:56
@relh relh changed the title Update recipes to checkpoint directory URIs Recipes: switch to checkpoint directory URIs Dec 23, 2025
@relh
Copy link
Contributor Author

relh commented Dec 23, 2025

Folded into #4497 as part of stack consolidation.

@relh relh closed this Dec 23, 2025
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants