# Streaming RL Persistent Worker 13

- Updated: `2026-07-25T10:10:40.703645+00:00`
- Queue: `diffusion-training-acceleration`
- Lane: `horizon`
- Current experiment: `none`
- Current status: `queue_completed`

| Experiment | Status | Return code | Latest log step |
|---|---|---:|---:|
| `srl_w13_fixed12_20` | completed | 0 |  |
| `srl_w13_dynamic30_20` | completed | 0 |  |
| `srl_w13_native45_80` | failed | 1 |  |
| `srl_w13_fixed18_100` | failed | 1 |  |
| `srl_w13_fixed30_100` | failed | 1 |  |
| `srl_w13_dynamic30_100` | failed | 1 |  |
| `srl_w13_fixed45_100` | failed | 1 |  |
| `srl_w13_native45_200` | failed | 1 |  |

Rollout manifests and compact evaluations are mirrored to Hugging Face.
Trajectory tensors remain node-local; checkpoints and timers remain on S3.