Files

25 lines
1.2 KiB
Markdown
Raw Permalink Normal View History

---
library_name: transformers
base_model: Qwen/Qwen3-8B
tags:
- ronpo
- mnpo
- preference-optimization
---
# qwen3-8b-aaai27-flagship-sppo-avg-s44
Research checkpoint for the RONPO AAAI revision experiments.
- Method: sppo_avg
- Base model: `Qwen/Qwen3-8B`
- Seed: 44
- Local source at upload time: `/NHNHOME/WORKSPACE/26msit001_A/BASE/aipr_lab_sjkim_eval/revision_qwen3_8b/full_iter1/flagship_20260712/full/sppo_avg/seed44/attempt2`
- Uploaded at UTC: 2026-07-13T06:41:09Z
- Run status metadata: `{"attempt": 2, "effective_batch_size": 16, "method": "sppo_avg", "objective_protocol": "objective_protocol.json", "optimizer_steps": 900, "resource_profile": "resource_profile.json", "seed": 44, "split_manifest": "split_manifest.json", "stability_gate": "stability_gate.json", "status": "completed", "wandb_run_id": "0bbaa979875e", "wandb_url": "https://wandb.ai/promotion-kim/mnpo/runs/0bbaa979875e"}`
AAAI-27 P1 matched budget: 900 optimizer steps, effective batch 16; passed non-thinking and collapse stability gates; sealed-test metrics are not used for checkpoint selection.
Intended use: reproducibility and evaluation for the RONPO paper. This
checkpoint is not intended as a production assistant.