{ "args": { "base": "Qwen/Qwen3.5-4B-Base", "n_per_source": 1000, "epochs": 1, "lr": 2e-05, "head_lr": 0.0, "weight_decay": 0.01, "lora": 16, "accum": 2, "holdout": "", "perm_kl": 0.0, "perm_frac": 0.3, "ord_w": 0.0, "label_smoothing": 0.0, "brier_w": 0.0, "focal_gamma": 0.0, "suite": "evals/v7/decision-v7", "train_sources": "", "device": "cuda", "batch": 4, "dtype": "bf16", "weights_dtype": "bf16", "checkpointing": 1, "option_isolation": 0, "special_embeddings": 0, "head_dim": 256, "lora_targets": "all", "base_revision": "1001bb4d826a52d1f399e183466143f4da7b741b", "p_none": 0.0, "p_none_distract": 0.0, "p_distract": 0.0, "p_none_pair": 0.0, "none_pair_max_state": null, "synthetic_repeat": 1, "public_frac": 1.0, "anchor": "", "anchor_w": 0.0, "anchor_sources": "", "out": "runs/jevasio-4b-v2", "data": "/root/work/kev-pt/data/out-v2/train-continue.jsonl", "max_state": 1024, "replay": 12576, "init_from": "TigreGotico/jevasio-4b", "full_ft": 0, "row_budget": 0, "shared_prefix": 0, "length_sort": 1, "pass_tokens_max": 0, "max_steps": 0, "save_every_steps": 0, "save_every_minutes": 0, "resume": 0, "stop_after": 0, "snapshot_fractions": "", "snapshot_every_steps": 0, "snapshot_dir": "", "seed": 0 }, "suite_sha256": "a8f50e481b7d90b97da049e0ff6a01cee2f1ed204aed61a8265af0edbb5514d2", "base_revision": "1001bb4d826a52d1f399e183466143f4da7b741b", "init_source": { "init_from": "TigreGotico/jevasio-4b", "resolved": "/root/.cache/huggingface/hub/models--TigreGotico--jevasio-4b/snapshots/76a7f73ad33ff4ff7b173b59b45f03130b9f885c", "weights_sha256": "3ef6fc4422a555d6e76d80f552c1055d406e682cb32d821b20c6ab0fe3fabffb", "head_sha256": "5e47c3b9f1b3a5a1ca579e1fd2a78935e55ea3408156964723588a99b42e3a6a", "tensors": 496 }, "ordinal_objective": "ranked_probability_score", "holdout": [] }