## Offline Checkpoint Evaluation

Exact-alignment holdout: 130 tracks, 2,768 windows.

| Selection | Checkpoint | Step | Loss | Semantic top-1 | Semantic top-5 | Acoustic top-1 | Acoustic top-5 |
|---|---|---:|---:|---:|---:|---:|---:|
| lowest loss; best semantic top-1; final | `final` | 17,660 | 3.987436 | 0.4317 | 0.8051 | 0.0729 | 0.1988 |
| best acoustic top-1 | `checkpoint-17500` | 17,500 | 3.988227 | 0.4316 | 0.8051 | 0.0730 | 0.1988 |

Top-k accuracy measures exact token inclusion. It does not measure perceptual code equivalence.

### Checkpoint Loss

![Checkpoint Loss](./checkpoint-loss.png)

### Checkpoint Accuracy

![Checkpoint Accuracy](./checkpoint-accuracy.png)

### Codebook Top1

![Codebook Top1](./codebook-top1.png)

### Depth Teacher Forcing Gap

![Depth Teacher Forcing Gap](./depth-teacher-forcing-gap.png)

### Training History

![Training History](./training-history.png)

Full data: [`checkpoint-metrics.csv`](./checkpoint-metrics.csv), [`evaluation-metrics.json`](./evaluation-metrics.json).