2026-04-23 06:39:49,556 INFO cleanslate.fsid.training.trainer: [RANK 0] FSID train: model=meta-llama/Llama-3.1-8B objective=simnpo strategy=fsdp run_dir=/runs/llama-8b/baseline/simnpo/simnpo_Llama-3.1-8B_20260423_063935 2026-04-23 06:39:49,556 INFO cleanslate.fsid.training.trainer: [RANK 0] Starting training: objective=simnpo epochs=2 2026-04-23 06:44:31,306 INFO cleanslate.fsid.training.trainer: [RANK 0] Saved checkpoint to /runs/llama-8b/baseline/simnpo/simnpo_Llama-3.1-8B_20260423_063935 2026-04-23 06:44:31,311 INFO cleanslate.fsid.training.trainer: [RANK 0] Training complete (183.5s)