Dynamic genre conditioning + conditioning monitor

[scragnog]

Dynamic genre conditioning + conditioning monitor

Part 1 of a 5-part training-core series (#67 → #68 → #69 → #70 → #71). Merging in order keeps every diff small — each later PR currently shows its predecessors' commits until they merge.

Dynamic genre conditioning. Preprocessing now encodes BOTH the caption and the genre-tag text variant into each .pt (new optional keys — old tensors keep loading, genre_ratio semantics unchanged for them). Training then selects caption vs genre per sample per step according to genre_ratio (auto-detected from preprocess_meta.json when unset), replacing the old static per-sample choice frozen at preprocess time. This removes the systematic bias where a sample was forever caption-only or genre-only for the whole run. Requires re-preprocessing to benefit; degrades gracefully without it.

Conditioning monitor. conditioning_type flows from the dataset through collate into ProgressWriter events, and the training monitor shows live caption-vs-genre usage per batch. Also ports the target-loss scale/EMA telemetry to the Fabric trainer path (previously basic-loop only).

🤖 Generated with Claude Code

[coderabbitai]

Warning

Review limit reached

@scragnog, you've reached your PR review limit, so we couldn't start this review.

Next review available in: 59 minutes

Enable usage-based reviews in Billing to review now. Otherwise, wait until the next included review is available.
You're only billed for reviews past your plan's rate limits ($0.25/file).

How can I continue?

After more reviews become available, a review can be triggered using the @coderabbitai review command as a PR comment. Alternatively, push new commits to this PR.

To avoid repeated limits, reduce automatic review volume by pausing incremental auto-reviews earlier, using label-based review opt-in, excluding WIP or generated PR titles, or requesting reviews manually when the PR is ready. If your team needs uninterrupted high-volume reviews, an organization admin can enable usage-based reviews.

How do review limits work?

CodeRabbit enforces per-developer PR review limits for each organization. Most developers receive the normal plan review availability.

For paid Pro and Pro+ PR reviews, CodeRabbit uses adaptive limits for sustained high-volume activity. When a developer's recent PR review activity reaches the 95th percentile or higher among CodeRabbit users, additional reviews become available more gradually as earlier reviews age out of the rolling window.

Please refer docs for additional details.

Review details

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: ASSERTIVE

Plan: Pro Plus

Run ID: a1c1ae94-d856-4d72-b5a6-71ee363b628e

📥 Commits

Reviewing files that changed from the base of the PR and between ecd13bd and 85ef97a.

📒 Files selected for processing (7)

• frontend/index.html
• frontend/js/training.js
• sidestep_engine/core/configs.py
• sidestep_engine/core/trainer.py
• sidestep_engine/core/trainer_loop.py
• sidestep_engine/data/preprocess.py
• sidestep_engine/vendor/data_module.py

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}