Skip to content

Add Synthetic Workspace Gym generators #25

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
ydnyshhh wants to merge 5 commits into
base: main
Choose a base branch
from
Open

Add Synthetic Workspace Gym generators #25

Show file tree
Hide file tree
Changes from 3 commits
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
f341053
Add Synthetic Workspace Gym generators
ydnyshhh Oct 1, 2026
3bd13bb
Harden SWG calibration and reward protocol
ydnyshhh Oct 3, 2026
9ee221a
Record immutable Prime VM artifact pins
ydnyshhh Oct 3, 2026
5e99e4e
Keep SWG curricula in pinned dataset, not Git
ydnyshhh Nov 1, 2026
1e22f57
Clarify pinned V5 reproduction revision
ydnyshhh Nov 1, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
5 changes: 5 additions & 0 deletions configs/synthetic_workspace_gym/README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,5 @@
# Synthetic Workspace Gym configurations

These immutable manifests define the task assignments used for SFT, RL, validation, and difficulty calibration. `scripts/synthetic_workspace_gym/build_qwen35_training_manifests.py` rebuilds them deterministically, and the generator package consumes them when exporting frozen task data for the installable environment.

@snimu snimu Nov 1, 2026

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

[major] The curriculum data belongs in the published dataset, not committed to the repo

configs/synthetic_workspace_gym/ adds ~87K lines of curriculum JSON, including four frozen revisions (v2–v4 plus the live v5) of each curriculum. The taskset already consumes this data from a pinned HF dataset revision, and this PR publishes exactly there (publish_dataset.py), so the checked-in copies are a redundant channel — and the dead revisions are pure bloat: the repo's own history serves the provenance role the README cites. Please follow the convention the other sandbox-image environments use: keep the data in the published dataset, commit only a small pointer that pins the dataset revision and the sha256 (or the generation_fingerprint you already compute), and drop the committed JSONs.

[written by prime-agent, checked by snimu]

@ydnyshhh ydnyshhh Nov 1, 2026

Copy link
Copy Markdown
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Addressed in 5e99e4e: removed all 24 checked-in curriculum JSON files (86,592 lines). The six active V5 source manifests are now generated into an ignored build directory, and configs/synthetic_workspace_gym/release.json is the small pointer to the pinned HF dataset commit, export integrity SHA-256, exporting commit, and source-manifest SHA-256 hashes. I rebuilt the six manifests and verified all six hashes match the previously committed bytes. The published V5 dataset and images were not changed.


The installable taskset keeps the manifest selections it needs as package data so its wheel remains self-contained. These files are the repository-level source configurations used to reproduce that data.
Loading
Loading