{
  "run": "pilot_E_s1",
  "step": 6000,
  "model_config": {
    "vocab_size": 256,
    "d_enc": 256,
    "d_main": 512,
    "n_enc": 2,
    "n_dec": 2,
    "n_main": 6,
    "n_heads_enc": 4,
    "n_heads_main": 8,
    "seq_len": 1024,
    "dropout": 0.0,
    "target_ratio": 6.0,
    "max_chunk_mult": 3.0,
    "dechunk": "ema",
    "router_noise": 0.0,
    "rope_base": 10000.0
  },
  "tied_weights": {
    "head.weight": "emb.weight"
  },
  "train_config": {
    "condition": "E",
    "seed": 1,
    "out_dir": null,
    "run_name": "pilot_E_s1",
    "corpus": "corpus",
    "langs": null,
    "seq_len": 1024,
    "batch_size": 32,
    "grad_accum": 1,
    "steps": 6000,
    "lr": 0.0003,
    "min_lr_frac": 0.1,
    "warmup": 300,
    "weight_decay": 0.1,
    "grad_clip": 1.0,
    "betas": [
      0.9,
      0.95
    ],
    "router_lr_mult": 1.0,
    "d_enc": 256,
    "d_main": 512,
    "n_enc": 2,
    "n_dec": 2,
    "n_main": 6,
    "n_heads_enc": 4,
    "n_heads_main": 8,
    "dechunk": "ema",
    "target_ratio": 6.0,
    "alpha": 0.03,
    "alpha_parity_cap": 0.06,
    "global_anchor": 0.0,
    "premium_clip": 4.0,
    "premium_kind": "flores_per_sent",
    "mixture": null,
    "doc_filters": null,
    "tau1": 0.4,
    "tau2": 0.7,
    "freeze_scope": "router",
    "ckpt_every": 200,
    "eval_every": 200,
    "log_every": 50,
    "probe_sents": 128,
    "eval_batches": 8,
    "keep_ckpt": false,
    "amp": "bf16",
    "compile": false,
    "cpc_decay": 0.9,
    "device": "cuda",
    "log_spaced_probe": true,
    "probe_bpb_batches": 4,
    "dump_reps": true,
    "dump_reps_thresholds": [
      0.3,
      0.4,
      0.5,
      0.6,
      0.7
    ],
    "router_noise": 0.0,
    "genreg_target_entropy": null,
    "genreg_eta": 0.05
  },
  "probe_agg": {
    "gini_cpc": 0.1964674510925677,
    "gini_cps": 0.10098658922012471,
    "bpic": 8.195545400187921,
    "worst_lang": "zh",
    "cpc_parity_vs_en": {
      "en": 1.0,
      "de": 0.9655403375392061,
      "vi": 1.3673246834915216,
      "tr": 1.3008611235523166,
      "fi": 0.9485352068270303,
      "ru": 1.549262456237803,
      "el": 1.345955480257742,
      "hi": 1.7048210970846212,
      "ta": 1.5275244502989007,
      "th": 1.450953060997798,
      "zh": 3.5384460424771085,
      "my": 1.050124096740635
    }
  },
  "source": "AdaptiveChunking/hnet-chunking-pilots/pilot_E_s1/model.safetensors"
}