{
  "source_checkpoint": "runs/best_step402",
  "base_model": "Qwen/Qwen3.5-9B",
  "dtype": "bfloat16",
  "format": "markers-v3-plainchains",
  "head_dim": 256,
  "temperature": 1.8589280843734741,
  "stage": "cispo",
  "step": 402,
  "sources": {
    "banking77": 0,
    "boolq": 1,
    "agnews": 2,
    "mnli": 3,
    "sst5": 4,
    "yelp": 5,
    "trec": 6,
    "dbpedia14": 7,
    "amazon": 8,
    "imdb": 9,
    "mmlu": 10,
    "emotion": 11,
    "tweet_offensive": 12,
    "qnli": 13,
    "paws": 14,
    "sciq": 15,
    "contrastive": 16,
    "compositional": 17,
    "composition_holdout": 18,
    "semif": 19,
    "wanli": 20,
    "mmlu_pro": 21,
    "buried": 22,
    "unknowable": 23,
    "unknowable_control": 24,
    "night2_assertion": 25,
    "night2_dates": 26,
    "night2_unknowable": 27,
    "night2_unknowable_control": 28
  },
  "metrics": null,
  "training_config": {
    "stage": "cispo",
    "run_dir": "runs/cispo_rep",
    "data_dir": "data",
    "model": "Qwen/Qwen3.5-9B",
    "init": "runs/sft_v4/final",
    "resume": null,
    "lora_r": 16,
    "lora_alpha": 32.0,
    "lora_dropout": 0.0,
    "head_dim": 256,
    "lr": 5e-05,
    "head_lr": 0.0001,
    "weight_decay": 0.0,
    "warmup": 20,
    "epochs": 4,
    "max_steps": 0,
    "batch_size": 8,
    "grad_clip": 1.0,
    "max_len": 4096,
    "max_seq_len": 8192,
    "train_limit": 0,
    "dev_limit": 0,
    "eval_every": 50,
    "eval_batch_size": 16,
    "eval_think": true,
    "eval_think_limit": 256,
    "log_every": 10,
    "seed": 0,
    "gradient_checkpointing": true,
    "prompts_per_step": 8,
    "group": 8,
    "nothink_per_group": 0,
    "nothink_start_epoch": 1,
    "max_think": 2560,
    "temperature": 1.0,
    "top_p": 1.0,
    "eps_high": 0.2,
    "updates": 1,
    "train_mix_frac": 0.0,
    "rollout_ce_weight": 0.5,
    "anchor_ce_weight": 0.5,
    "anchor_batch": 8,
    "gen_batch_size": 64,
    "logprob_chunk": 128,
    "micro_groups": 2,
    "length_hinge": 2048,
    "length_hinge_end": 1024,
    "length_ramp": 0,
    "length_penalty_weight": 1.0,
    "length_penalty_cap": 0.1,
    "length_penalty_correct_only": true,
    "length_hinge2": 0,
    "length_hinge2_end": 1792,
    "length_penalty2_weight": 0.0,
    "closure_reward": 0.0,
    "pad_multiple": 128,
    "num_workers": 2,
    "compile": true,
    "graph_decode": true,
    "async_rollouts": true,
    "ckpt_minutes": 45.0,
    "log_rollouts": 3,
    "log_chain_chars": 1200,
    "sft_think_frac": 0.5,
    "chain_cache": "data/chains",
    "filter_active": false,
    "filter_min_spread": 0.02
  }
}