Open-weight model
rlport-hkl_jord9v_mk_pl3_20261018_020308
by Jord Nguyen Jordine/rlport-hkl_jord9v_mk_pl3_20261018_020308
rlport-hkl_jord9v_mk_pl3_20261018_020308 is an open-weight model from Jord Nguyen. Its published files total 5.7 GB.
Automatic off-box backup of one RL training run from the reward-laundering project (Tinker -> vast port), written by scripts/port/hfbackup.py so a run can resume after its cloud GPU box disappears.
Model Card
Automatic off-box backup of one RL training run from the reward-laundering project (Tinker -> vast port), written by scripts/port/hfbackup.py so a run can resume after its cloud GPU box disappears. Contents: LoRA adapters (Qwen3.5, PEFT/Tinker layout) + optimizer state, per-step metrics, training rollouts and logs. Research artifact only; not a model release, not intended for use.
Excerpt from the card by Jord Nguyen.
Identity and Version
- Repository
- Jordine/rlport-hkl_jord9v_mk_pl3_20261018_020308
- Publisher
- Jord Nguyen
- Task
- Not stated by the source
- Modality
- Other
- Library
- Not stated by the source
- Parameters
- Not stated by the source
- Languages
- Not stated by the source
- Revision
- e91fd711143ad2bd81347b9b301b2051dd5b61f5
- First published
- 2026-10-18
- Last updated
- 2026-10-18
Files and Weights
31 files, 5.7 GB in total. The weights are 11 files totalling 5.7 GB in pt, safetensors.
Every file
| File | Type | Size | SHA-256 |
|---|---|---|---|
| kept/sampler_weights__jord9v_mk_pl3_hkl_step13/adapter_model.safetensors | Weights | 378.4 MB | e7a9836784ec |
| kept/sampler_weights__jord9v_mk_pl3_hkl_step5/adapter_model.safetensors | Weights | 378.4 MB | 3ce77b443445 |
| kept/sampler_weights__jord9v_mk_pl3_hkl_step9/adapter_model.safetensors | Weights | 378.4 MB | 1a51cc9dc747 |
| kept/weights__jord9v_mk_pl3_hkl_state_step13/adapter_model.safetensors | Weights | 378.4 MB | e7a9836784ec |
| kept/weights__jord9v_mk_pl3_hkl_state_step13/optimizer.pt | Weights | 757.1 MB | 508238ed36c5 |
| kept/weights__jord9v_mk_pl3_hkl_state_step5/adapter_model.safetensors | Weights | 378.4 MB | 3ce77b443445 |
| kept/weights__jord9v_mk_pl3_hkl_state_step5/optimizer.pt | Weights | 757.1 MB | 365a09398885 |
| kept/weights__jord9v_mk_pl3_hkl_state_step9/adapter_model.safetensors | Weights | 378.4 MB | 1a51cc9dc747 |
| kept/weights__jord9v_mk_pl3_hkl_state_step9/optimizer.pt | Weights | 757.1 MB | d2612bdf4e8c |
| state/latest/adapter_model.safetensors | Weights | 378.4 MB | e7a9836784ec |
| state/latest/optimizer.pt | Weights | 757.1 MB | 508238ed36c5 |
| kept/sampler_weights__jord9v_mk_pl3_hkl_step13/adapter_config.json | Configuration | 278 B | — |
| kept/sampler_weights__jord9v_mk_pl3_hkl_step5/adapter_config.json | Configuration | 278 B | — |
| kept/sampler_weights__jord9v_mk_pl3_hkl_step9/adapter_config.json | Configuration | 278 B | — |
| kept/weights__jord9v_mk_pl3_hkl_state_step13/adapter_config.json | Configuration | 278 B | — |
| kept/weights__jord9v_mk_pl3_hkl_state_step5/adapter_config.json | Configuration | 278 B | — |
| kept/weights__jord9v_mk_pl3_hkl_state_step9/adapter_config.json | Configuration | 278 B | — |
| manifest.json | Configuration | 1.1 KB | — |
| run/cli_extras.json | Configuration | 326 B | — |
| run/cmdline.json | Configuration | 555 B | — |
| run/config.json | Configuration | 810 B | — |
| run/resume.json | Configuration | 868 B | — |
| run/summary.json | Configuration | 19.1 KB | — |
| state/latest/_origin.json | Configuration | 122 B | — |
| state/latest/adapter_config.json | Configuration | 278 B | — |
| README.md | Documentation | 496 B | — |
| logs/rl_jord9v_mk_pl3.log | Other | 8.1 KB | — |
| run/metrics.jsonl | Other | 14.8 KB | — |
| run/rollouts.jsonl.gz | Other | 2.4 MB | b5de8c812215 |
| run/state.txt | Other | 102 B | — |
| .gitattributes | Repository | 1.5 KB | — |
License and Download
- License
- Not stated by the source
- Access
- Open weights, no gate
- Download size
- 5.7 GB
Released by Jord Nguyen through its official repository on Hugging Face.
Memory Requirements
| Precision | Weights in memory |
|---|---|
| As published | 5.7 GB |
Weights only, from the published parameter count; the key-value cache and runtime add to this.