Skip to content

TraceOS — experiment management layer built on ABC-130K (no code modified) #9

Description

@aaa-mvc

Hi ABC team,

I built TraceOS (aaa-mvc/traceos), a lightweight experiment runtime that wraps ABC's training and evaluation pipeline.

What TraceOS does:

experiment run bottle.yaml → one command: train + evaluate + report + analyze
Standardized output structure per run (report.html, analysis.json, events.jsonl)
Run registry for experiment history
Cross-run comparison
Capability scoring + failure taxonomy + recommendations
How it relates to ABC:

The ABC Adapter wraps train.py and eval_policy.py via subprocess
Zero lines of ABC source code are modified (verified by git diff)
ABC's DiT model, DINOv3 backbone, CLIP encoder, and MuJoCo sim are used as-is
The paper and dataset are prominently acknowledged in our README
Status:

v0.1.0 released under Apache 2.0
97 tests passing
Mock adapter available for instant demo (no GPU needed)
Thank you for open-sourcing ABC-130K — it made this work possible. Any feedback is welcome.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions