{ "run": "jitter_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": "A", "seed": 1, "out_dir": null, "run_name": "jitter_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": 2000, "eval_every": 1000, "log_every": 200, "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": 0.2318796127695224, "gini_cps": 0.2339954866678544, "bpic": 68.37359700694816, "worst_lang": "hi", "cpc_parity_vs_en": { "en": 1.0, "de": 0.8146260953869533, "vi": 1.3084208484930249, "tr": 0.911522408188721, "fi": 0.6782836797893858, "ru": 1.7701111460483683, "el": 0.823810338912061, "hi": 2.4679785047232636, "ta": 1.7050250337085502, "th": 1.168831225307265, "zh": 1.74051680121559, "my": 0.6662411771075679 } }, "source": "AdaptiveChunking/hnet-chunking-pilots/jitter_s1/model.safetensors" }