# Streaming RL Persistent Worker 1

- Updated: `2026-07-25T15:09:29.473859+00:00`
- Queue: `diffusion-training-acceleration`
- Lane: `transport`
- Current experiment: `none`
- Current status: `queue_completed`

| Experiment | Status | Return code | Latest log step |
|---|---|---:|---:|
| `srl_w01_mem2` | completed | 0 |  |
| `srl_w01_disk2` | completed | 0 |  |
| `srl_w01_diskref100` | failed | 1 |  |
| `srl_w01_manifest100` | failed | 1 |  |
| `srl_w01_disk4_100` | failed | 1 |  |
| `srl_w01_disk8_200` | failed | 1 |  |
| `srl_w01_disk45_100` | failed | 1 |  |
| `srl_w01_disk45_200` | failed | 1 |  |

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