# Streaming RL Persistent Worker 7

- Updated: `2026-07-26T01:52:01.693031+00:00`
- Queue: `diffusion-training-acceleration`
- Lane: `delayed`
- Current experiment: `none`
- Current status: `queue_completed`

| Experiment | Status | Return code | Latest log step |
|---|---|---:|---:|
| `srl_w07_global20` | completed | 0 |  |
| `srl_w07_hybrid20` | completed | 0 |  |
| `srl_w07_hybrid100` | failed | 1 |  |
| `srl_w07_global025_100` | failed | 1 |  |
| `srl_w07_global05_100` | failed | 1 |  |
| `srl_w07_global2_100` | failed | 1 |  |
| `srl_w07_hybrid05_100` | failed | 1 |  |
| `srl_w07_hybrid200` | failed | 1 |  |

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