{"event": "start", "run": "ali_ali_s1", "arch": "ali", "objective": "ratio", "condition": "A", "params": 22785088, "langs": ["en", "tr", "fi"], "premium": {"en": 1.0, "tr": 1.12, "fi": 1.104}, "schedule": {"tau1": 0.4, "tau2": 0.7, "p1_objective": "global", "p2_objective": "global", "p3_objective": "global", "freeze_in_p3": false, "freeze_scope": "router"}, "corpus": "/local/scratch/sas245/beetle-hnet/scratch/data/tier1_corpus", "premium_kind": "flores_per_sent", "doc_filtered": {}, "weights": {"en": 0.74, "tr": 0.1536, "fi": 0.1064}, "device": "cuda", "ali_params": 268096, "ali_config": {"enabled": true, "budget_B": 65536, "staging_capacity": 262144, "policy": "utility", "candidate_types": ["surface", "split", "merge", "skel"], "gate": "vector", "retrieval": true, "memory_scope": "persistent", "read_dropout": 0.1, "write_rule": "mean", "eta_min": 0.0, "consolidate_every": 50, "probe_seqs": 2, "probe_reps": 8, "horizon": 32, "predictor": "mlp", "lambda_util": 0.1, "max_span": 64, "byte_classes_path": "/local/scratch/sas245/beetle-hnet/scratch/data/byte_classes.json", "gate_hidden": 0, "type_dim": 16, "gate_bias_init": -2.0}, "t": 4.1}
{"event": "probe_schedule", "n": 40, "steps": [0, 1, 2, 4, 8, 16, 32, 64, 128, 200, 256, 400, 600, 800, 1000, 1200, 1400, 1600, 1800, 2000, 2200, 2400, 2600, 2800, 3000, 3200, 3400, 3600, 3800, 4000, 4200, 4400, 4600, 4800, 5000, 5200, 5400, 5600, 5800, 6000], "t": 4.1}
{"event": "probe", "step": 0, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2674.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.153652, "cps": 20.890625, "mean_p": 0.478126, "boundary_entropy": 0.685438, "bpb": 8.041379}, "tr": {"chunks": 5645.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.311827, "cps": 44.101562, "mean_p": 0.486063, "boundary_entropy": 0.686375, "bpb": 8.01701}, "fi": {"chunks": 4732.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.251555, "cps": 36.96875, "mean_p": 0.48451, "boundary_entropy": 0.686595, "bpb": 8.075294}}, "bpb": {"en": 8.04138, "tr": 8.01701, "fi": 8.07529}, "gini_cpc": 0.147064, "gini_cps": 0.151764, "bpic": 4.346947, "worst_lang": "tr", "cpc_parity_vs_en": {"en": 1.0, "tr": 2.029439514461233, "fi": 1.6371767549398106}, "inventory": {"size": 0, "admitted": 0, "staged": 0, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 0, "staged_surface": 0, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 0.0, "n_util_obs": 0, "util_count_spearman": NaN}, "t": 6.5}
{"event": "train", "step": 0, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 5.58509, "bpb": 8.05758, "chunk_loss": 0.030141, "F": 0.16724, "G": 0.47713, "chunks": 171.25, "bytes_per_chunk": 5.98, "boundary_entropy": 0.69022, "router_noise": 0.0, "grad_norm": 11.3986, "lr": 1e-06, "router_trainable": true, "sec_per_step": 0.011, "probe_sec": 0.0079, "probe_extra_seqs": 0, "n_rows": 0, "extra_fwd_seqs": 0, "wall_s": 0.0, "overhead_frac_est": 0.0, "chunk_mismatch": 0, "n_loo": 0, "mean_delta_loo": NaN, "n_shadow": 0, "mean_delta_shadow": NaN, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "read_hit_rate": {"surface": 0.0, "split": 0.0, "merge": 0.0, "skel": 0.0}, "mean_gate": {"surface": null, "split": null, "merge": null, "skel": null}, "n_reads": 0, "n_cand": 22370, "read_suppressed_frac": 0.0, "dropout_reads_logged": 0, "inventory": {"size": 8792, "admitted": 0, "staged": 8792, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 0, "staged_surface": 3161, "admitted_split": 0, "staged_split": 3293, "admitted_merge": 0, "staged_merge": 44, "admitted_skel": 0, "staged_skel": 2294, "mean_count": 0.0, "n_util_obs": 0, "util_count_spearman": NaN}, "probe_overhead_sec_total": 0.008, "probe_overhead_seq_frac": 0.0, "util_loss": 0.0, "alpha": 0.03, "t": 8.4}
{"event": "probe", "step": 1, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2659.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.15279, "cps": 20.773438, "mean_p": 0.478104, "boundary_entropy": 0.685439, "bpb": 8.008224}, "tr": {"chunks": 5625.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.310722, "cps": 43.945312, "mean_p": 0.485999, "boundary_entropy": 0.686373, "bpb": 7.993535}, "fi": {"chunks": 4708.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.250279, "cps": 36.78125, "mean_p": 0.484401, "boundary_entropy": 0.686587, "bpb": 8.047773}}, "bpb": {"en": 8.00822, "tr": 7.99353, "fi": 8.04777}, "gini_cpc": 0.147506, "gini_cps": 0.152196, "bpic": 4.366687, "worst_lang": "tr", "cpc_parity_vs_en": {"en": 1.0, "tr": 2.033657244886179, "fi": 1.6380620678582016}, "inventory": {"size": 8792, "admitted": 0, "staged": 8792, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 0, "staged_surface": 3161, "admitted_split": 0, "staged_split": 3293, "admitted_merge": 0, "staged_merge": 44, "admitted_skel": 0, "staged_skel": 2294, "mean_count": 0.0, "n_util_obs": 0, "util_count_spearman": NaN}, "t": 10.0}
{"event": "probe", "step": 2, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2659.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.15279, "cps": 20.773438, "mean_p": 0.478078, "boundary_entropy": 0.685441, "bpb": 7.943276}, "tr": {"chunks": 5587.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.308623, "cps": 43.648438, "mean_p": 0.485897, "boundary_entropy": 0.686371, "bpb": 7.946092}, "fi": {"chunks": 4664.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.24794, "cps": 36.4375, "mean_p": 0.484224, "boundary_entropy": 0.686575, "bpb": 7.993916}}, "bpb": {"en": 7.94328, "tr": 7.94609, "fi": 7.99392}, "gini_cpc": 0.146456, "gini_cps": 0.151201, "bpic": 4.394423, "worst_lang": "tr", "cpc_parity_vs_en": {"en": 1.0, "tr": 2.0199187603873923, "fi": 1.6227530765698073}, "inventory": {"size": 15537, "admitted": 0, "staged": 15537, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1, "admitted_surface": 0, "staged_surface": 5579, "admitted_split": 0, "staged_split": 5747, "admitted_merge": 0, "staged_merge": 117, "admitted_skel": 0, "staged_skel": 4094, "mean_count": 0.0, "n_util_obs": 0, "util_count_spearman": NaN}, "t": 13.0}
{"event": "probe", "step": 4, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2615.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.150261, "cps": 20.429688, "mean_p": 0.478041, "boundary_entropy": 0.685455, "bpb": 7.727491}, "tr": {"chunks": 5457.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.301442, "cps": 42.632812, "mean_p": 0.485573, "boundary_entropy": 0.686365, "bpb": 7.790631}, "fi": {"chunks": 4533.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.240976, "cps": 35.414062, "mean_p": 0.483678, "boundary_entropy": 0.686539, "bpb": 7.81447}}, "bpb": {"en": 7.72749, "tr": 7.79063, "fi": 7.81447}, "gini_cpc": 0.145503, "gini_cps": 0.150311, "bpic": 4.500754, "worst_lang": "tr", "cpc_parity_vs_en": {"en": 1.0, "tr": 2.0061150191241555, "fi": 1.6037115706576226}, "inventory": {"size": 29030, "admitted": 0, "staged": 29030, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 5, "admitted_surface": 0, "staged_surface": 10481, "admitted_split": 0, "staged_split": 10442, "admitted_merge": 0, "staged_merge": 297, "admitted_skel": 0, "staged_skel": 7810, "mean_count": 0.0, "n_util_obs": 0, "util_count_spearman": NaN}, "t": 17.2}
{"event": "probe", "step": 8, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2627.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.150951, "cps": 20.523438, "mean_p": 0.47884, "boundary_entropy": 0.685612, "bpb": 7.169816}, "tr": {"chunks": 5157.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.28487, "cps": 40.289062, "mean_p": 0.484942, "boundary_entropy": 0.68639, "bpb": 7.39843}, "fi": {"chunks": 4104.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.21817, "cps": 32.0625, "mean_p": 0.482322, "boundary_entropy": 0.686466, "bpb": 7.341287}}, "bpb": {"en": 7.16982, "tr": 7.39843, "fi": 7.34129}, "gini_cpc": 0.136515, "gini_cps": 0.14188, "bpic": 4.772207, "worst_lang": "tr", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.8871682764578892, "fi": 1.4453050364941893}, "inventory": {"size": 54334, "admitted": 0, "staged": 54334, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 10, "admitted_surface": 0, "staged_surface": 19583, "admitted_split": 0, "staged_split": 19246, "admitted_merge": 0, "staged_merge": 619, "admitted_skel": 0, "staged_skel": 14886, "mean_count": 0.0, "n_util_obs": 0, "util_count_spearman": NaN}, "t": 22.9}
{"event": "probe", "step": 16, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2596.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.14917, "cps": 20.28125, "mean_p": 0.481097, "boundary_entropy": 0.686031, "bpb": 6.514918}, "tr": {"chunks": 4389.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.242446, "cps": 34.289062, "mean_p": 0.482949, "boundary_entropy": 0.686437, "bpb": 6.898393}, "fi": {"chunks": 2929.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.155707, "cps": 22.882812, "mean_p": 0.478268, "boundary_entropy": 0.686271, "bpb": 6.694173}}, "bpb": {"en": 6.51492, "tr": 6.89839, "fi": 6.69417}, "gini_cpc": 0.113615, "gini_cps": 0.12057, "bpic": 5.722413, "worst_lang": "tr", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.6253034659486163, "fi": 1.0438231400684401}, "inventory": {"size": 104280, "admitted": 0, "staged": 104280, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 25, "admitted_surface": 0, "staged_surface": 38364, "admitted_split": 0, "staged_split": 35454, "admitted_merge": 0, "staged_merge": 1524, "admitted_skel": 0, "staged_skel": 28938, "mean_count": 0.0, "n_util_obs": 0, "util_count_spearman": NaN}, "t": 29.3}
{"event": "probe", "step": 32, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3685.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.211745, "cps": 28.789062, "mean_p": 0.490464, "boundary_entropy": 0.68717, "bpb": 6.161268}, "tr": {"chunks": 3944.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.217864, "cps": 30.8125, "mean_p": 0.486231, "boundary_entropy": 0.687283, "bpb": 6.577368}, "fi": {"chunks": 3153.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.167615, "cps": 24.632812, "mean_p": 0.485037, "boundary_entropy": 0.687295, "bpb": 6.341671}}, "bpb": {"en": 6.16127, "tr": 6.57737, "fi": 6.34167}, "gini_cpc": 0.056093, "gini_cps": 0.048909, "bpic": 5.261733, "worst_lang": "tr", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0288995631885116, "fi": 0.7915871130282831}, "inventory": {"size": 196323, "admitted": 0, "staged": 196323, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 42, "admitted_surface": 0, "staged_surface": 72558, "admitted_split": 0, "staged_split": 65598, "admitted_merge": 0, "staged_merge": 2955, "admitted_skel": 0, "staged_skel": 55212, "mean_count": 0.0, "n_util_obs": 0, "util_count_spearman": NaN}, "t": 39.2}
{"event": "consolidate", "step": 50, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 65536, "swapped": 0, "uncertain_staged": 262137, "t": 53.2}
{"event": "train", "step": 50, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 4.09849, "bpb": 5.91287, "chunk_loss": 0.030541, "F": 0.17413, "G": 0.48631, "chunks": 178.31, "bytes_per_chunk": 5.743, "boundary_entropy": 0.69164, "router_noise": 0.0, "grad_norm": 2.3616, "lr": 5.1e-05, "router_trainable": true, "sec_per_step": 0.903, "probe_sec": 0.097, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.07611, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 5, "mean_delta_loo": 0.00771, "n_shadow": 5, "mean_delta_shadow": -0.00414, "n_split": 0, "mean_delta_split": NaN, "n_merge": 2, "mean_delta_merge": -0.00256, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": 0.42361, "trunc_abs_frac": 0.5578, "mean_delta_full": 0.01055, "read_hit_rate": {"surface": 0.0, "split": 0.0, "merge": 0.0, "skel": 0.07752}, "mean_gate": {"surface": null, "split": null, "merge": null, "skel": 0.11941}, "n_reads": 279, "n_cand": 23783, "read_suppressed_frac": 0.1075, "dropout_reads_logged": 279, "inventory": {"size": 267561, "admitted": 65536, "staged": 202025, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 76, "admitted_surface": 2476, "staged_surface": 96799, "admitted_split": 11389, "staged_split": 76579, "admitted_merge": 0, "staged_merge": 4863, "admitted_skel": 51671, "staged_skel": 23784, "mean_count": 1.1389007568359375, "n_util_obs": 0, "util_count_spearman": NaN}, "probe_overhead_sec_total": 4.294, "probe_overhead_seq_frac": 0.4902, "util_loss": 0.005292, "alpha": 0.03, "t": 53.6}
{"event": "probe", "step": 64, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3749.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.215423, "cps": 29.289062, "mean_p": 0.484621, "boundary_entropy": 0.6862, "bpb": 5.429237}, "tr": {"chunks": 4464.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.246589, "cps": 34.875, "mean_p": 0.481625, "boundary_entropy": 0.686154, "bpb": 5.921935}, "fi": {"chunks": 3989.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.212057, "cps": 31.164062, "mean_p": 0.482743, "boundary_entropy": 0.686574, "bpb": 5.654237}}, "bpb": {"en": 5.42924, "tr": 5.92193, "fi": 5.65424}, "gini_cpc": 0.034153, "gini_cps": 0.039065, "bpic": 4.649402, "worst_lang": "tr", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.1446753069595372, "fi": 0.9843755829261661}, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 86, "admitted_surface": 2476, "staged_surface": 117171, "admitted_split": 11389, "staged_split": 95153, "admitted_merge": 0, "staged_merge": 6087, "admitted_skel": 51671, "staged_skel": 43733, "mean_count": 1.1723480224609375, "n_util_obs": 1, "util_count_spearman": NaN}, "t": 63.7}
{"event": "consolidate", "step": 100, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1, "uncertain_staged": 262133, "t": 95.2}
{"event": "train", "step": 100, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 3.20812, "bpb": 4.62833, "chunk_loss": 0.03293, "F": 0.21243, "G": 0.46173, "chunks": 217.53, "bytes_per_chunk": 4.707, "boundary_entropy": 0.68347, "router_noise": 0.0, "grad_norm": 1.6447, "lr": 0.000101, "router_trainable": true, "sec_per_step": 0.855, "probe_sec": 0.1434, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.12786, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 2, "mean_delta_loo": -0.01508, "n_shadow": 8, "mean_delta_shadow": -0.01391, "n_split": 0, "mean_delta_split": NaN, "n_merge": 2, "mean_delta_merge": 0.01288, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 0.10029, "trunc_abs_frac": 0.41245, "mean_delta_full": -0.1276, "read_hit_rate": {"surface": 0.0, "split": 0.0, "merge": 0.0, "skel": 0.00756}, "mean_gate": {"surface": null, "split": null, "merge": null, "skel": 0.11941}, "n_reads": 37, "n_cand": 31368, "read_suppressed_frac": 0.0541, "dropout_reads_logged": 37, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 142, "admitted_surface": 2476, "staged_surface": 98018, "admitted_split": 11389, "staged_split": 82345, "admitted_merge": 0, "staged_merge": 9937, "admitted_skel": 51671, "staged_skel": 71844, "mean_count": 1.0816650390625, "n_util_obs": 0, "util_count_spearman": NaN}, "probe_overhead_sec_total": 9.827, "probe_overhead_seq_frac": 0.495, "util_loss": 0.000937, "alpha": 0.03, "t": 96.3}
{"event": "probe", "step": 128, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3713.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.213354, "cps": 29.007812, "mean_p": 0.424141, "boundary_entropy": 0.61829, "bpb": 4.109462}, "tr": {"chunks": 3955.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.218472, "cps": 30.898438, "mean_p": 0.404822, "boundary_entropy": 0.638617, "bpb": 4.376958}, "fi": {"chunks": 4054.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.215512, "cps": 31.671875, "mean_p": 0.412121, "boundary_entropy": 0.642148, "bpb": 4.314522}}, "bpb": {"en": 4.10946, "tr": 4.37696, "fi": 4.31452}, "gini_cpc": 0.005271, "gini_cps": 0.019394, "bpic": 4.839788, "worst_lang": "tr", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0239885662493962, "fi": 1.0101154920457505}, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 183, "admitted_surface": 2476, "staged_surface": 98787, "admitted_split": 11389, "staged_split": 79364, "admitted_merge": 0, "staged_merge": 14942, "admitted_skel": 51671, "staged_skel": 69051, "mean_count": 1.094696044921875, "n_util_obs": 0, "util_count_spearman": NaN}, "t": 119.3}
{"event": "consolidate", "step": 150, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 0, "uncertain_staged": 262130, "t": 144.5}
{"event": "train", "step": 150, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 2.74792, "bpb": 3.96441, "chunk_loss": 0.033122, "F": 0.22803, "G": 0.39935, "chunks": 233.5, "bytes_per_chunk": 4.385, "boundary_entropy": 0.58154, "router_noise": 0.0, "grad_norm": 2.0865, "lr": 0.000151, "router_trainable": true, "sec_per_step": 0.98, "probe_sec": 0.0868, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.05619, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 1, "mean_delta_loo": 0.26465, "n_shadow": 9, "mean_delta_shadow": -0.0311, "n_split": 0, "mean_delta_split": NaN, "n_merge": 2, "mean_delta_merge": 0.05187, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 4.20168, "trunc_abs_frac": 0.68622, "mean_delta_full": -0.01453, "read_hit_rate": {"surface": 0.0, "split": 0.0, "merge": 0.0, "skel": 0.00635}, "mean_gate": {"surface": null, "split": null, "merge": null, "skel": 0.11975}, "n_reads": 35, "n_cand": 33980, "read_suppressed_frac": 0.1714, "dropout_reads_logged": 35, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 218, "admitted_surface": 2476, "staged_surface": 98115, "admitted_split": 11389, "staged_split": 76661, "admitted_merge": 0, "staged_merge": 20087, "admitted_skel": 51671, "staged_skel": 67281, "mean_count": 1.1061859130859375, "n_util_obs": 0, "util_count_spearman": NaN}, "probe_overhead_sec_total": 16.737, "probe_overhead_seq_frac": 0.4967, "util_loss": 0.000823, "alpha": 0.03, "t": 145.3}
{"event": "consolidate", "step": 200, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 2, "uncertain_staged": 262123, "t": 175.7}
{"event": "probe", "step": 200, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3228.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.185485, "cps": 25.21875, "mean_p": 0.327422, "boundary_entropy": 0.46631, "bpb": 3.551947}, "tr": {"chunks": 4381.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.242004, "cps": 34.226562, "mean_p": 0.31004, "boundary_entropy": 0.490954, "bpb": 3.787514}, "fi": {"chunks": 3794.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.201691, "cps": 29.640625, "mean_p": 0.330016, "boundary_entropy": 0.523772, "bpb": 3.895296}}, "bpb": {"en": 3.55195, "tr": 3.78751, "fi": 3.8953}, "gini_cpc": 0.059886, "gini_cps": 0.067409, "bpic": 4.975182, "worst_lang": "tr", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.3047079115848244, "fi": 1.0873667210852465}, "inventory": {"size": 327678, "admitted": 65536, "staged": 262142, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 287, "admitted_surface": 2476, "staged_surface": 92532, "admitted_split": 11391, "staged_split": 76199, "admitted_merge": 0, "staged_merge": 30251, "admitted_skel": 51669, "staged_skel": 63160, "mean_count": 1.126251220703125, "n_util_obs": 0, "util_count_spearman": NaN}, "t": 177.1}
{"event": "eval", "step": 200, "phase": 1, "bpb": {"en": 3.56094, "tr": 3.79725, "fi": 3.8945}, "bytes_per_chunk": {"en": 5.13, "tr": 4.49, "fi": 5.135}, "bpb_mean": 3.7509, "frozen_inventory": true, "t": 179.9}
{"event": "train", "step": 200, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 2.49074, "bpb": 3.59337, "chunk_loss": 0.031089, "F": 0.19778, "G": 0.32769, "chunks": 202.53, "bytes_per_chunk": 5.056, "boundary_entropy": 0.47697, "router_noise": 0.0, "grad_norm": 1.4843, "lr": 0.000201, "router_trainable": true, "sec_per_step": 0.773, "probe_sec": 0.1053, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.07285, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 1, "mean_delta_loo": 0.07349, "n_shadow": 9, "mean_delta_shadow": -0.01138, "n_split": 0, "mean_delta_split": NaN, "n_merge": 2, "mean_delta_merge": 0.03918, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 0.02559, "trunc_abs_frac": 0.31066, "mean_delta_full": -0.18481, "read_hit_rate": {"surface": 0.0, "split": 0.0, "merge": 0.0, "skel": 0.00758}, "mean_gate": {"surface": null, "split": null, "merge": null, "skel": 0.11917}, "n_reads": 38, "n_cand": 29626, "read_suppressed_frac": 0.0263, "dropout_reads_logged": 38, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 283, "admitted_surface": 2476, "staged_surface": 92558, "admitted_split": 11391, "staged_split": 75802, "admitted_merge": 0, "staged_merge": 30505, "admitted_skel": 51669, "staged_skel": 63279, "mean_count": 1.1268310546875, "n_util_obs": 0, "util_count_spearman": NaN}, "probe_overhead_sec_total": 22.124, "probe_overhead_seq_frac": 0.4975, "util_loss": 0.000867, "alpha": 0.03, "t": 183.9}
{"event": "consolidate", "step": 250, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262116, "t": 223.7}
{"event": "train", "step": 250, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 2.36209, "bpb": 3.40778, "chunk_loss": 0.03053, "F": 0.18707, "G": 0.2823, "chunks": 191.56, "bytes_per_chunk": 5.346, "boundary_entropy": 0.38023, "router_noise": 0.0, "grad_norm": 1.7757, "lr": 0.000251, "router_trainable": true, "sec_per_step": 0.821, "probe_sec": 0.1739, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.12587, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": 0.00723, "n_shadow": 5, "mean_delta_shadow": -0.05892, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": 0.40213, "trunc_abs_frac": 0.61718, "mean_delta_full": -0.06074, "read_hit_rate": {"surface": 0.0, "split": 0.05754, "merge": 0.0, "skel": 0.00472}, "mean_gate": {"surface": null, "split": 0.12164, "merge": null, "skel": 0.12223}, "n_reads": 685, "n_cand": 28735, "read_suppressed_frac": 0.092, "dropout_reads_logged": 685, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 218, "admitted_surface": 1452, "staged_surface": 89109, "admitted_split": 12415, "staged_split": 71195, "admitted_merge": 0, "staged_merge": 38273, "admitted_skel": 51669, "staged_skel": 63567, "mean_count": 2.709197998046875, "n_util_obs": 7, "util_count_spearman": -0.35714285714285715}, "probe_overhead_sec_total": 27.091, "probe_overhead_seq_frac": 0.498, "util_loss": 0.001361, "alpha": 0.03, "t": 225.0}
{"event": "probe", "step": 256, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3102.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.178245, "cps": 24.234375, "mean_p": 0.267611, "boundary_entropy": 0.35659, "bpb": 3.317614}, "tr": {"chunks": 4016.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.221842, "cps": 31.375, "mean_p": 0.273858, "boundary_entropy": 0.39483, "bpb": 3.538174}, "fi": {"chunks": 3468.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.18436, "cps": 27.09375, "mean_p": 0.264851, "boundary_entropy": 0.414708, "bpb": 3.673554}}, "bpb": {"en": 3.31761, "tr": 3.53817, "fi": 3.67355}, "gini_cpc": 0.04973, "gini_cps": 0.05756, "bpic": 5.359154, "worst_lang": "tr", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.2445876277920103, "fi": 1.0343071623490392}, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 217, "admitted_surface": 1452, "staged_surface": 88723, "admitted_split": 12415, "staged_split": 71003, "admitted_merge": 0, "staged_merge": 39148, "admitted_skel": 51669, "staged_skel": 63270, "mean_count": 2.7581024169921875, "n_util_obs": 7, "util_count_spearman": -0.32142857142857145}, "t": 229.3}
{"event": "consolidate", "step": 300, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262113, "t": 267.3}
{"event": "train", "step": 300, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 2.20758, "bpb": 3.18486, "chunk_loss": 0.030434, "F": 0.18469, "G": 0.27799, "chunks": 189.12, "bytes_per_chunk": 5.414, "boundary_entropy": 0.35682, "router_noise": 0.0, "grad_norm": 1.419, "lr": 0.0003, "router_trainable": true, "sec_per_step": 0.866, "probe_sec": 0.1445, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.09364, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": -0.04659, "n_shadow": 3, "mean_delta_shadow": -0.02453, "n_split": 0, "mean_delta_split": NaN, "n_merge": 2, "mean_delta_merge": 0.17194, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": -18.84678, "trunc_abs_frac": 0.65291, "mean_delta_full": -0.00252, "read_hit_rate": {"surface": 0.0, "split": 0.04926, "merge": 0.04927, "skel": 0.00733}, "mean_gate": {"surface": null, "split": 0.12472, "merge": 0.12838, "skel": 0.12486}, "n_reads": 889, "n_cand": 28247, "read_suppressed_frac": 0.0934, "dropout_reads_logged": 889, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 153, "admitted_surface": 428, "staged_surface": 85913, "admitted_split": 12758, "staged_split": 68427, "admitted_merge": 681, "staged_merge": 46988, "admitted_skel": 51669, "staged_skel": 60816, "mean_count": 4.3678436279296875, "n_util_obs": 9, "util_count_spearman": 0.3333333333333333}, "probe_overhead_sec_total": 32.222, "probe_overhead_seq_frac": 0.4983, "util_loss": 0.00228, "alpha": 0.03, "t": 268.3}
{"event": "consolidate", "step": 350, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262109, "t": 302.5}
{"event": "train", "step": 350, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 2.19312, "bpb": 3.16401, "chunk_loss": 0.030445, "F": 0.18567, "G": 0.27521, "chunks": 190.12, "bytes_per_chunk": 5.386, "boundary_entropy": 0.34599, "router_noise": 0.0, "grad_norm": 1.5693, "lr": 0.0002999, "router_trainable": true, "sec_per_step": 0.699, "probe_sec": 0.2017, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.18398, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": 0.09347, "n_shadow": 4, "mean_delta_shadow": -0.05614, "n_split": 0, "mean_delta_split": NaN, "n_merge": 2, "mean_delta_merge": -0.04422, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 0.02657, "trunc_abs_frac": 0.2623, "mean_delta_full": -0.08557, "read_hit_rate": {"surface": 0.0, "split": 0.04434, "merge": 0.05631, "skel": 0.0051}, "mean_gate": {"surface": null, "split": 0.12041, "merge": 0.12489, "skel": 0.12304}, "n_reads": 861, "n_cand": 28010, "read_suppressed_frac": 0.1034, "dropout_reads_logged": 861, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 145, "admitted_surface": 92, "staged_surface": 82858, "admitted_split": 12758, "staged_split": 66474, "admitted_merge": 1705, "staged_merge": 54595, "admitted_skel": 50981, "staged_skel": 58217, "mean_count": 5.265167236328125, "n_util_obs": 15, "util_count_spearman": 0.2857142857142857}, "probe_overhead_sec_total": 37.746, "probe_overhead_seq_frac": 0.4986, "util_loss": 0.002777, "alpha": 0.03, "t": 303.2}
{"event": "consolidate", "step": 400, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262106, "t": 342.2}
{"event": "probe", "step": 400, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3257.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.187152, "cps": 25.445312, "mean_p": 0.250317, "boundary_entropy": 0.278359, "bpb": 2.882596}, "tr": {"chunks": 3471.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.191736, "cps": 27.117188, "mean_p": 0.287613, "boundary_entropy": 0.391708, "bpb": 3.141043}, "fi": {"chunks": 3349.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.178034, "cps": 26.164062, "mean_p": 0.273267, "boundary_entropy": 0.397352, "bpb": 3.364115}}, "bpb": {"en": 2.8826, "tr": 3.14104, "fi": 3.36412}, "gini_cpc": 0.016402, "gini_cps": 0.014158, "bpic": 5.62985, "worst_lang": "tr", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0244963698412477, "fi": 0.9512827591078699}, "inventory": {"size": 326656, "admitted": 65536, "staged": 261120, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 141, "admitted_surface": 92, "staged_surface": 78848, "admitted_split": 12797, "staged_split": 63067, "admitted_merge": 2690, "staged_merge": 63639, "admitted_skel": 49957, "staged_skel": 55566, "mean_count": 6.138916015625, "n_util_obs": 17, "util_count_spearman": 0.37745098039215685}, "t": 343.4}
{"event": "eval", "step": 400, "phase": 1, "bpb": {"en": 2.89481, "tr": 3.14869, "fi": 3.35606}, "bytes_per_chunk": {"en": 4.987, "tr": 5.604, "fi": 5.734}, "bpb_mean": 3.13319, "frozen_inventory": true, "t": 346.0}
{"event": "train", "step": 400, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 2.03324, "bpb": 2.93334, "chunk_loss": 0.030549, "F": 0.19467, "G": 0.25862, "chunks": 199.34, "bytes_per_chunk": 5.137, "boundary_entropy": 0.28638, "router_noise": 0.0, "grad_norm": 1.6114, "lr": 0.0002998, "router_trainable": true, "sec_per_step": 0.944, "probe_sec": 0.0753, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.05046, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 11, "mean_delta_loo": -0.0012, "n_shadow": 3, "mean_delta_shadow": 0.02295, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 1, "trunc_ratio_sum": 0.38051, "trunc_abs_frac": 0.38051, "mean_delta_full": -0.21045, "read_hit_rate": {"surface": 0.0, "split": 0.0449, "merge": 0.07506, "skel": 0.00531}, "mean_gate": {"surface": null, "split": 0.126, "merge": 0.12661, "skel": 0.1185}, "n_reads": 1043, "n_cand": 29982, "read_suppressed_frac": 0.0978, "dropout_reads_logged": 1043, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 138, "admitted_surface": 92, "staged_surface": 79135, "admitted_split": 12797, "staged_split": 63195, "admitted_merge": 2690, "staged_merge": 64052, "admitted_skel": 49957, "staged_skel": 55762, "mean_count": 6.1548309326171875, "n_util_obs": 17, "util_count_spearman": 0.3946078431372549}, "probe_overhead_sec_total": 43.033, "probe_overhead_seq_frac": 0.4988, "util_loss": 0.003072, "alpha": 0.03, "t": 350.4}
{"event": "consolidate", "step": 450, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262103, "t": 388.6}
{"event": "train", "step": 450, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 2.06211, "bpb": 2.975, "chunk_loss": 0.030607, "F": 0.19574, "G": 0.25741, "chunks": 200.44, "bytes_per_chunk": 5.109, "boundary_entropy": 0.28231, "router_noise": 0.0, "grad_norm": 1.0657, "lr": 0.0002995, "router_trainable": true, "sec_per_step": 0.786, "probe_sec": 0.1679, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.14363, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": 0.09449, "n_shadow": 4, "mean_delta_shadow": 0.01645, "n_split": 0, "mean_delta_split": NaN, "n_merge": 1, "mean_delta_merge": -0.28162, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 0.99575, "trunc_abs_frac": 0.8493, "mean_delta_full": -0.09578, "read_hit_rate": {"surface": 0.0, "split": 0.08336, "merge": 0.07984, "skel": 0.00742}, "mean_gate": {"surface": null, "split": 0.13219, "merge": 0.1287, "skel": 0.12493}, "n_reads": 1546, "n_cand": 29864, "read_suppressed_frac": 0.0996, "dropout_reads_logged": 1546, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 147, "admitted_surface": 2, "staged_surface": 76154, "admitted_split": 13344, "staged_split": 58788, "admitted_merge": 3167, "staged_merge": 74058, "admitted_skel": 49023, "staged_skel": 53144, "mean_count": 8.92669677734375, "n_util_obs": 22, "util_count_spearman": 0.5268210050818747}, "probe_overhead_sec_total": 49.961, "probe_overhead_seq_frac": 0.4989, "util_loss": 0.00342, "alpha": 0.03, "t": 389.7}
{"event": "consolidate", "step": 500, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262099, "t": 429.7}
{"event": "train", "step": 500, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 2.02933, "bpb": 2.9277, "chunk_loss": 0.030433, "F": 0.19098, "G": 0.25096, "chunks": 195.56, "bytes_per_chunk": 5.236, "boundary_entropy": 0.27475, "router_noise": 0.0, "grad_norm": 1.2771, "lr": 0.0002992, "router_trainable": true, "sec_per_step": 0.814, "probe_sec": 0.0711, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.0424, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": -0.02959, "n_shadow": 6, "mean_delta_shadow": -0.01462, "n_split": 0, "mean_delta_split": NaN, "n_merge": 1, "mean_delta_merge": 0.09363, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 0.03619, "trunc_abs_frac": 0.16481, "mean_delta_full": 0.19845, "read_hit_rate": {"surface": 0.02249, "split": 0.07799, "merge": 0.06393, "skel": 0.00731}, "mean_gate": {"surface": 0.13845, "split": 0.12742, "merge": 0.12422, "skel": 0.12086}, "n_reads": 1461, "n_cand": 28623, "read_suppressed_frac": 0.091, "dropout_reads_logged": 1461, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 152, "admitted_surface": 2, "staged_surface": 73276, "admitted_split": 14339, "staged_split": 55939, "admitted_merge": 3195, "staged_merge": 81818, "admitted_skel": 48000, "staged_skel": 51111, "mean_count": 11.2896728515625, "n_util_obs": 26, "util_count_spearman": 0.42495726495726494}, "probe_overhead_sec_total": 56.643, "probe_overhead_seq_frac": 0.499, "util_loss": 0.003739, "alpha": 0.03, "t": 430.5}
{"event": "consolidate", "step": 550, "retired": 1, "split_retired": 1, "merge_admitted": 0, "admitted": 2, "swapped": 1024, "uncertain_staged": 262100, "t": 465.6}
{"event": "train", "step": 550, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.95057, "bpb": 2.81408, "chunk_loss": 0.030763, "F": 0.20413, "G": 0.25497, "chunks": 209.03, "bytes_per_chunk": 4.899, "boundary_entropy": 0.25186, "router_noise": 0.0, "grad_norm": 1.2133, "lr": 0.0002987, "router_trainable": true, "sec_per_step": 0.713, "probe_sec": 0.0821, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.05515, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": -0.01669, "n_shadow": 4, "mean_delta_shadow": 0.04111, "n_split": 0, "mean_delta_split": NaN, "n_merge": 1, "mean_delta_merge": 0.63135, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": -0.00712, "trunc_abs_frac": 0.84725, "mean_delta_full": 0.14412, "read_hit_rate": {"surface": 0.0, "split": 0.10232, "merge": 0.07457, "skel": 0.00464}, "mean_gate": {"surface": null, "split": 0.12642, "merge": 0.1257, "skel": 0.1218}, "n_reads": 1781, "n_cand": 30797, "read_suppressed_frac": 0.0915, "dropout_reads_logged": 1781, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 143, "admitted_surface": 1, "staged_surface": 70600, "admitted_split": 15357, "staged_split": 53575, "admitted_merge": 3203, "staged_merge": 88273, "admitted_skel": 46975, "staged_skel": 49696, "mean_count": 11.92333984375, "n_util_obs": 30, "util_count_spearman": 0.34549499443826476}, "probe_overhead_sec_total": 61.85, "probe_overhead_seq_frac": 0.4991, "util_loss": 0.003836, "alpha": 0.03, "t": 466.1}
{"event": "consolidate", "step": 600, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262096, "t": 507.5}
{"event": "probe", "step": 600, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3409.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.195886, "cps": 26.632812, "mean_p": 0.23174, "boundary_entropy": 0.211214, "bpb": 2.581815}, "tr": {"chunks": 3335.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.184224, "cps": 26.054688, "mean_p": 0.301118, "boundary_entropy": 0.397443, "bpb": 2.890979}, "fi": {"chunks": 3249.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.172718, "cps": 25.382812, "mean_p": 0.246025, "boundary_entropy": 0.338788, "bpb": 3.148974}}, "bpb": {"en": 2.58181, "tr": 2.89098, "fi": 3.14897}, "gini_cpc": 0.027938, "gini_cps": 0.010674, "bpic": 5.677174, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9404644979341268, "fi": 0.8817286384880033}, "inventory": {"size": 326656, "admitted": 65536, "staged": 261120, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 152, "admitted_surface": 1, "staged_surface": 67899, "admitted_split": 16380, "staged_split": 51032, "admitted_merge": 3204, "staged_merge": 94528, "admitted_skel": 45951, "staged_skel": 47661, "mean_count": 13.842727661132812, "n_util_obs": 32, "util_count_spearman": 0.38563049853372433}, "t": 509.2}
{"event": "eval", "step": 600, "phase": 1, "bpb": {"en": 2.59574, "tr": 2.9015, "fi": 3.14573}, "bytes_per_chunk": {"en": 4.846, "tr": 5.752, "fi": 5.898}, "bpb_mean": 2.88099, "frozen_inventory": true, "t": 512.2}
{"event": "train", "step": 600, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.9241, "bpb": 2.77589, "chunk_loss": 0.030536, "F": 0.19534, "G": 0.25238, "chunks": 200.03, "bytes_per_chunk": 5.119, "boundary_entropy": 0.25069, "router_noise": 0.0, "grad_norm": 1.1609, "lr": 0.0002982, "router_trainable": true, "sec_per_step": 1.017, "probe_sec": 0.1623, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.12903, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 5, "mean_delta_loo": -0.06375, "n_shadow": 5, "mean_delta_shadow": 0.03893, "n_split": 0, "mean_delta_split": NaN, "n_merge": 2, "mean_delta_merge": -0.004, "n_dropped_unpaired": 0, "n_full": 1, "trunc_ratio_sum": 1.13672, "trunc_abs_frac": 1.13672, "mean_delta_full": 0.10803, "read_hit_rate": {"surface": 0.0, "split": 0.10747, "merge": 0.06644, "skel": 0.00789}, "mean_gate": {"surface": null, "split": 0.1301, "merge": 0.129, "skel": 0.12442}, "n_reads": 1735, "n_cand": 29627, "read_suppressed_frac": 0.1101, "dropout_reads_logged": 1735, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 154, "admitted_surface": 1, "staged_surface": 68112, "admitted_split": 16380, "staged_split": 51250, "admitted_merge": 3204, "staged_merge": 94883, "admitted_skel": 45951, "staged_skel": 47899, "mean_count": 13.86920166015625, "n_util_obs": 32, "util_count_spearman": 0.3324780058651026}, "probe_overhead_sec_total": 68.364, "probe_overhead_seq_frac": 0.4992, "util_loss": 0.004151, "alpha": 0.03, "t": 517.0}
{"event": "consolidate", "step": 650, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262092, "t": 560.5}
{"event": "train", "step": 650, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.86174, "bpb": 2.68593, "chunk_loss": 0.030417, "F": 0.19217, "G": 0.25056, "chunks": 196.78, "bytes_per_chunk": 5.204, "boundary_entropy": 0.24913, "router_noise": 0.0, "grad_norm": 1.3266, "lr": 0.0002975, "router_trainable": true, "sec_per_step": 0.88, "probe_sec": 0.079, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.05106, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 4, "mean_delta_loo": -0.01817, "n_shadow": 5, "mean_delta_shadow": -0.01578, "n_split": 1, "mean_delta_split": -0.07928, "n_merge": 2, "mean_delta_merge": 0.05173, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 0.16243, "trunc_abs_frac": 0.5281, "mean_delta_full": 0.28003, "read_hit_rate": {"surface": 0.01309, "split": 0.12368, "merge": 0.07637, "skel": 0.00976}, "mean_gate": {"surface": 0.14312, "split": 0.13488, "merge": 0.13556, "skel": 0.12655}, "n_reads": 2037, "n_cand": 28987, "read_suppressed_frac": 0.1055, "dropout_reads_logged": 2037, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 158, "admitted_surface": 3, "staged_surface": 66227, "admitted_split": 17315, "staged_split": 48780, "admitted_merge": 3291, "staged_merge": 100355, "admitted_skel": 44927, "staged_skel": 46782, "mean_count": 16.847320556640625, "n_util_obs": 36, "util_count_spearman": 0.36164736164736166}, "probe_overhead_sec_total": 75.436, "probe_overhead_seq_frac": 0.4992, "util_loss": 0.004461, "alpha": 0.03, "t": 561.0}
{"event": "consolidate", "step": 700, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262096, "t": 595.4}
{"event": "train", "step": 700, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.87744, "bpb": 2.70858, "chunk_loss": 0.030337, "F": 0.18698, "G": 0.26622, "chunks": 191.47, "bytes_per_chunk": 5.348, "boundary_entropy": 0.28825, "router_noise": 0.0, "grad_norm": 1.0576, "lr": 0.0002967, "router_trainable": true, "sec_per_step": 0.707, "probe_sec": 0.1081, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.08382, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": 0.07056, "n_shadow": 3, "mean_delta_shadow": -0.13582, "n_split": 0, "mean_delta_split": NaN, "n_merge": 1, "mean_delta_merge": -0.11804, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": 0.25462, "trunc_abs_frac": 0.53492, "mean_delta_full": 0.0647, "read_hit_rate": {"surface": 0.00788, "split": 0.13343, "merge": 0.05558, "skel": 0.0067}, "mean_gate": {"surface": 0.15086, "split": 0.13679, "merge": 0.14162, "skel": 0.12487}, "n_reads": 1928, "n_cand": 28119, "read_suppressed_frac": 0.0975, "dropout_reads_logged": 1928, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 175, "admitted_surface": 3, "staged_surface": 64885, "admitted_split": 18338, "staged_split": 47199, "admitted_merge": 3292, "staged_merge": 104001, "admitted_skel": 43903, "staged_skel": 46059, "mean_count": 20.875152587890625, "n_util_obs": 43, "util_count_spearman": 0.4540924192086983}, "probe_overhead_sec_total": 80.649, "probe_overhead_seq_frac": 0.4993, "util_loss": 0.005119, "alpha": 0.03, "t": 596.3}
{"event": "consolidate", "step": 750, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262095, "t": 638.6}
{"event": "train", "step": 750, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.78176, "bpb": 2.57054, "chunk_loss": 0.030366, "F": 0.19183, "G": 0.24369, "chunks": 196.44, "bytes_per_chunk": 5.213, "boundary_entropy": 0.22412, "router_noise": 0.0, "grad_norm": 0.9776, "lr": 0.0002959, "router_trainable": true, "sec_per_step": 0.861, "probe_sec": 0.0945, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.06951, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 12, "mean_delta_loo": -0.0035, "n_shadow": 2, "mean_delta_shadow": 0.08598, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": -1.3229, "trunc_abs_frac": 1.25861, "mean_delta_full": 0.04549, "read_hit_rate": {"surface": 0.01807, "split": 0.17345, "merge": 0.07859, "skel": 0.00548}, "mean_gate": {"surface": 0.15889, "split": 0.14041, "merge": 0.14174, "skel": 0.13258}, "n_reads": 2677, "n_cand": 29391, "read_suppressed_frac": 0.1042, "dropout_reads_logged": 2677, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 169, "admitted_surface": 3, "staged_surface": 63799, "admitted_split": 19361, "staged_split": 45703, "admitted_merge": 3292, "staged_merge": 107178, "admitted_skel": 42880, "staged_skel": 45464, "mean_count": 23.5516357421875, "n_util_obs": 47, "util_count_spearman": 0.4099213691026827}, "probe_overhead_sec_total": 86.386, "probe_overhead_seq_frac": 0.4993, "util_loss": 0.005629, "alpha": 0.03, "t": 639.4}
{"event": "consolidate", "step": 800, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262092, "t": 676.3}
{"event": "probe", "step": 800, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3433.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.197265, "cps": 26.820312, "mean_p": 0.220729, "boundary_entropy": 0.17302, "bpb": 2.424551}, "tr": {"chunks": 3187.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.176048, "cps": 24.898438, "mean_p": 0.287628, "boundary_entropy": 0.372978, "bpb": 2.711667}, "fi": {"chunks": 3384.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.179895, "cps": 26.4375, "mean_p": 0.257986, "boundary_entropy": 0.346437, "bpb": 2.981858}}, "bpb": {"en": 2.42455, "tr": 2.71167, "fi": 2.98186}, "gini_cpc": 0.025568, "gini_cps": 0.016393, "bpic": 5.670932, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.8924457564321995, "fi": 0.9119452964309313}, "inventory": {"size": 326656, "admitted": 65536, "staged": 261120, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 181, "admitted_surface": 4, "staged_surface": 62734, "admitted_split": 20271, "staged_split": 44503, "admitted_merge": 3293, "staged_merge": 109430, "admitted_skel": 41968, "staged_skel": 44453, "mean_count": 35.78852844238281, "n_util_obs": 58, "util_count_spearman": 0.3484265895598142}, "t": 678.0}
{"event": "eval", "step": 800, "phase": 1, "bpb": {"en": 2.44113, "tr": 2.72095, "fi": 2.98197}, "bytes_per_chunk": {"en": 4.83, "tr": 5.984, "fi": 5.711}, "bpb_mean": 2.71468, "frozen_inventory": true, "t": 680.8}
{"event": "train", "step": 800, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.79386, "bpb": 2.58799, "chunk_loss": 0.030401, "F": 0.19324, "G": 0.24784, "chunks": 197.88, "bytes_per_chunk": 5.175, "boundary_entropy": 0.2432, "router_noise": 0.0, "grad_norm": 1.0049, "lr": 0.0002949, "router_trainable": true, "sec_per_step": 0.927, "probe_sec": 0.1477, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.12863, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": 0.05634, "n_shadow": 7, "mean_delta_shadow": -0.19406, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 1.18164, "trunc_abs_frac": 0.52009, "mean_delta_full": -0.27441, "read_hit_rate": {"surface": 0.02619, "split": 0.21524, "merge": 0.07403, "skel": 0.05504}, "mean_gate": {"surface": 0.14956, "split": 0.14007, "merge": 0.14627, "skel": 0.15521}, "n_reads": 3418, "n_cand": 29123, "read_suppressed_frac": 0.1112, "dropout_reads_logged": 3418, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 180, "admitted_surface": 4, "staged_surface": 62924, "admitted_split": 20271, "staged_split": 44623, "admitted_merge": 3293, "staged_merge": 110056, "admitted_skel": 41968, "staged_skel": 44541, "mean_count": 35.84068298339844, "n_util_obs": 58, "util_count_spearman": 0.3577163247100803}, "probe_overhead_sec_total": 93.168, "probe_overhead_seq_frac": 0.4994, "util_loss": 0.005716, "alpha": 0.03, "t": 685.7}
{"event": "consolidate", "step": 850, "retired": 0, "split_retired": 2, "merge_admitted": 0, "admitted": 2, "swapped": 1024, "uncertain_staged": 262099, "t": 722.2}
{"event": "train", "step": 850, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.731, "bpb": 2.4973, "chunk_loss": 0.030415, "F": 0.19818, "G": 0.22564, "chunks": 202.94, "bytes_per_chunk": 5.046, "boundary_entropy": 0.18022, "router_noise": 0.0, "grad_norm": 1.2196, "lr": 0.0002938, "router_trainable": true, "sec_per_step": 0.74, "probe_sec": 0.1302, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.1103, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": -0.00086, "n_shadow": 6, "mean_delta_shadow": -0.13525, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 0.26744, "trunc_abs_frac": 0.46738, "mean_delta_full": -0.48392, "read_hit_rate": {"surface": 0.01919, "split": 0.27572, "merge": 0.09145, "skel": 0.08259}, "mean_gate": {"surface": 0.1295, "split": 0.13992, "merge": 0.1425, "skel": 0.15108}, "n_reads": 4553, "n_cand": 30580, "read_suppressed_frac": 0.096, "dropout_reads_logged": 4553, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 177, "admitted_surface": 2, "staged_surface": 62085, "admitted_split": 21286, "staged_split": 43266, "admitted_merge": 3294, "staged_merge": 112510, "admitted_skel": 40954, "staged_skel": 44283, "mean_count": 39.270965576171875, "n_util_obs": 58, "util_count_spearman": 0.3602387031283645}, "probe_overhead_sec_total": 98.786, "probe_overhead_seq_frac": 0.4994, "util_loss": 0.006277, "alpha": 0.03, "t": 722.7}
{"event": "consolidate", "step": 900, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262098, "t": 767.0}
{"event": "train", "step": 900, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.69883, "bpb": 2.4509, "chunk_loss": 0.030411, "F": 0.19656, "G": 0.23805, "chunks": 201.28, "bytes_per_chunk": 5.087, "boundary_entropy": 0.20413, "router_noise": 0.0, "grad_norm": 0.9289, "lr": 0.0002927, "router_trainable": true, "sec_per_step": 0.896, "probe_sec": 0.1096, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.08788, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": -0.05069, "n_shadow": 6, "mean_delta_shadow": -0.15598, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 1.15558, "trunc_abs_frac": 1.30712, "mean_delta_full": -0.23801, "read_hit_rate": {"surface": 0.01997, "split": 0.30785, "merge": 0.08029, "skel": 0.09916}, "mean_gate": {"surface": 0.14034, "split": 0.14212, "merge": 0.15288, "skel": 0.16007}, "n_reads": 4898, "n_cand": 30152, "read_suppressed_frac": 0.0982, "dropout_reads_logged": 4898, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 182, "admitted_surface": 3, "staged_surface": 61428, "admitted_split": 22268, "staged_split": 41951, "admitted_merge": 3298, "staged_merge": 114948, "admitted_skel": 39967, "staged_skel": 43817, "mean_count": 44.16340637207031, "n_util_obs": 59, "util_count_spearman": 0.31765049678550555}, "probe_overhead_sec_total": 104.403, "probe_overhead_seq_frac": 0.4994, "util_loss": 0.006421, "alpha": 0.03, "t": 767.5}
{"event": "consolidate", "step": 950, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262096, "t": 808.5}
{"event": "train", "step": 950, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.70951, "bpb": 2.46631, "chunk_loss": 0.030424, "F": 0.19809, "G": 0.23252, "chunks": 202.84, "bytes_per_chunk": 5.048, "boundary_entropy": 0.18683, "router_noise": 0.0, "grad_norm": 1.0482, "lr": 0.0002914, "router_trainable": true, "sec_per_step": 0.842, "probe_sec": 0.0992, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.07451, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": 0.01126, "n_shadow": 5, "mean_delta_shadow": 0.15099, "n_split": 0, "mean_delta_split": NaN, "n_merge": 1, "mean_delta_merge": 0.00208, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": -0.04792, "trunc_abs_frac": 0.73792, "mean_delta_full": -0.13947, "read_hit_rate": {"surface": 0.0514, "split": 0.31119, "merge": 0.08106, "skel": 0.09435}, "mean_gate": {"surface": 0.14324, "split": 0.14607, "merge": 0.15698, "skel": 0.16369}, "n_reads": 5204, "n_cand": 30627, "read_suppressed_frac": 0.0959, "dropout_reads_logged": 5204, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 188, "admitted_surface": 817, "staged_surface": 59600, "admitted_split": 22473, "staged_split": 41350, "admitted_merge": 3298, "staged_merge": 117702, "admitted_skel": 38948, "staged_skel": 43492, "mean_count": 48.106689453125, "n_util_obs": 60, "util_count_spearman": 0.2945818282856349}, "probe_overhead_sec_total": 109.881, "probe_overhead_seq_frac": 0.4995, "util_loss": 0.006529, "alpha": 0.03, "t": 809.6}
{"event": "consolidate", "step": 1000, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262091, "t": 846.4}
{"event": "probe", "step": 1000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3410.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.195943, "cps": 26.640625, "mean_p": 0.216838, "boundary_entropy": 0.158986, "bpb": 2.325661}, "tr": {"chunks": 3211.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.177374, "cps": 25.085938, "mean_p": 0.275076, "boundary_entropy": 0.345487, "bpb": 2.608931}, "fi": {"chunks": 3379.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.179629, "cps": 26.398438, "mean_p": 0.248233, "boundary_entropy": 0.328184, "bpb": 2.846236}}, "bpb": {"en": 2.32566, "tr": 2.60893, "fi": 2.84624}, "gini_cpc": 0.022388, "gini_cps": 0.013267, "bpic": 5.6732, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.905231160953702, "fi": 0.9167397219228595}, "inventory": {"size": 326656, "admitted": 65536, "staged": 261120, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 221, "admitted_surface": 1787, "staged_surface": 57878, "admitted_split": 22520, "staged_split": 40776, "admitted_merge": 3299, "staged_merge": 119542, "admitted_skel": 37930, "staged_skel": 42924, "mean_count": 53.91712951660156, "n_util_obs": 65, "util_count_spearman": 0.3130716026367248}, "t": 848.1}
{"event": "eval", "step": 1000, "phase": 1, "bpb": {"en": 2.33935, "tr": 2.61754, "fi": 2.84837}, "bytes_per_chunk": {"en": 4.84, "tr": 6.037, "fi": 5.773}, "bpb_mean": 2.60175, "frozen_inventory": true, "t": 851.0}
{"event": "train", "step": 1000, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.71187, "bpb": 2.4697, "chunk_loss": 0.03039, "F": 0.19653, "G": 0.23537, "chunks": 201.25, "bytes_per_chunk": 5.088, "boundary_entropy": 0.19647, "router_noise": 0.0, "grad_norm": 0.9019, "lr": 0.0002901, "router_trainable": true, "sec_per_step": 0.945, "probe_sec": 0.1843, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.15509, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": 0.02021, "n_shadow": 6, "mean_delta_shadow": -0.01169, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 0, "read_hit_rate": {"surface": 0.1044, "split": 0.30661, "merge": 0.08061, "skel": 0.10118}, "mean_gate": {"surface": 0.1496, "split": 0.15034, "merge": 0.16496, "skel": 0.17375}, "n_reads": 5390, "n_cand": 29975, "read_suppressed_frac": 0.0952, "dropout_reads_logged": 5390, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 221, "admitted_surface": 1787, "staged_surface": 58078, "admitted_split": 22520, "staged_split": 40877, "admitted_merge": 3299, "staged_merge": 120143, "admitted_skel": 37930, "staged_skel": 43046, "mean_count": 53.99937438964844, "n_util_obs": 65, "util_count_spearman": 0.3142079786898527}, "probe_overhead_sec_total": 114.949, "probe_overhead_seq_frac": 0.4995, "util_loss": 0.006803, "alpha": 0.03, "t": 856.9}
{"event": "consolidate", "step": 1050, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262089, "t": 899.9}
{"event": "train", "step": 1050, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.73382, "bpb": 2.50137, "chunk_loss": 0.030289, "F": 0.19052, "G": 0.23169, "chunks": 195.09, "bytes_per_chunk": 5.249, "boundary_entropy": 0.19726, "router_noise": 0.0, "grad_norm": 1.0191, "lr": 0.0002886, "router_trainable": true, "sec_per_step": 0.871, "probe_sec": 0.0814, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.05332, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": -0.02018, "n_shadow": 5, "mean_delta_shadow": -0.10115, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 0.57159, "trunc_abs_frac": 0.71979, "mean_delta_full": -0.23395, "read_hit_rate": {"surface": 0.13863, "split": 0.31828, "merge": 0.06668, "skel": 0.09158}, "mean_gate": {"surface": 0.14868, "split": 0.14934, "merge": 0.16318, "skel": 0.16937}, "n_reads": 5425, "n_cand": 29013, "read_suppressed_frac": 0.101, "dropout_reads_logged": 5425, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 260, "admitted_surface": 2809, "staged_surface": 57307, "admitted_split": 22522, "staged_split": 41172, "admitted_merge": 3299, "staged_merge": 120636, "admitted_skel": 36906, "staged_skel": 43029, "mean_count": 59.131866455078125, "n_util_obs": 69, "util_count_spearman": 0.28377785896967483}, "probe_overhead_sec_total": 121.51, "probe_overhead_seq_frac": 0.4995, "util_loss": 0.006756, "alpha": 0.03, "t": 900.4}
{"event": "consolidate", "step": 1100, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262090, "t": 939.0}
{"event": "train", "step": 1100, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.72049, "bpb": 2.48214, "chunk_loss": 0.030425, "F": 0.19888, "G": 0.23287, "chunks": 203.66, "bytes_per_chunk": 5.028, "boundary_entropy": 0.17789, "router_noise": 0.0, "grad_norm": 0.9389, "lr": 0.0002871, "router_trainable": true, "sec_per_step": 0.801, "probe_sec": 0.1332, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.09802, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 4, "mean_delta_loo": -0.12157, "n_shadow": 8, "mean_delta_shadow": 0.01826, "n_split": 1, "mean_delta_split": -0.07788, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": -0.53953, "trunc_abs_frac": 0.88156, "mean_delta_full": -0.03335, "read_hit_rate": {"surface": 0.18859, "split": 0.30578, "merge": 0.08275, "skel": 0.0961}, "mean_gate": {"surface": 0.15608, "split": 0.15642, "merge": 0.17666, "skel": 0.18359}, "n_reads": 6008, "n_cand": 30458, "read_suppressed_frac": 0.1027, "dropout_reads_logged": 6008, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 297, "admitted_surface": 3606, "staged_surface": 56302, "admitted_split": 22749, "staged_split": 40468, "admitted_merge": 3299, "staged_merge": 122390, "admitted_skel": 35882, "staged_skel": 42984, "mean_count": 63.92485046386719, "n_util_obs": 71, "util_count_spearman": 0.31562709590878607}, "probe_overhead_sec_total": 127.263, "probe_overhead_seq_frac": 0.4995, "util_loss": 0.00677, "alpha": 0.03, "t": 940.5}
{"event": "consolidate", "step": 1150, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262086, "t": 986.5}
{"event": "train", "step": 1150, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.7262, "bpb": 2.49038, "chunk_loss": 0.030432, "F": 0.19333, "G": 0.24824, "chunks": 197.97, "bytes_per_chunk": 5.173, "boundary_entropy": 0.22326, "router_noise": 0.0, "grad_norm": 0.8679, "lr": 0.0002855, "router_trainable": true, "sec_per_step": 0.934, "probe_sec": 0.1445, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.12469, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": 0.15016, "n_shadow": 5, "mean_delta_shadow": -0.06287, "n_split": 1, "mean_delta_split": -0.09784, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 1, "trunc_ratio_sum": -0.32523, "trunc_abs_frac": 0.32523, "mean_delta_full": -0.13062, "read_hit_rate": {"surface": 0.18816, "split": 0.32479, "merge": 0.0649, "skel": 0.14748}, "mean_gate": {"surface": 0.15747, "split": 0.15195, "merge": 0.18292, "skel": 0.17445}, "n_reads": 6128, "n_cand": 29202, "read_suppressed_frac": 0.1053, "dropout_reads_logged": 6128, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 328, "admitted_surface": 3628, "staged_surface": 55875, "admitted_split": 23644, "staged_split": 39336, "admitted_merge": 3301, "staged_merge": 124322, "admitted_skel": 34963, "staged_skel": 42611, "mean_count": 72.4053955078125, "n_util_obs": 75, "util_count_spearman": 0.28253200568990045}, "probe_overhead_sec_total": 135.643, "probe_overhead_seq_frac": 0.4996, "util_loss": 0.006694, "alpha": 0.03, "t": 987.2}
{"event": "consolidate", "step": 1200, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262085, "t": 1038.4}
{"event": "probe", "step": 1200, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3388.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.194679, "cps": 26.46875, "mean_p": 0.212681, "boundary_entropy": 0.142204, "bpb": 2.259787}, "tr": {"chunks": 3218.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.177761, "cps": 25.140625, "mean_p": 0.271235, "boundary_entropy": 0.337548, "bpb": 2.52577}, "fi": {"chunks": 3161.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.16804, "cps": 24.695312, "mean_p": 0.25273, "boundary_entropy": 0.344704, "bpb": 2.742821}}, "bpb": {"en": 2.25979, "tr": 2.52577, "fi": 2.74282}, "gini_cpc": 0.032859, "gini_cps": 0.015494, "bpic": 5.808539, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9130955092925233, "fi": 0.8631640238884066}, "inventory": {"size": 326656, "admitted": 65536, "staged": 261120, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 366, "admitted_surface": 3646, "staged_surface": 55724, "admitted_split": 24495, "staged_split": 39001, "admitted_merge": 3304, "staged_merge": 123825, "admitted_skel": 34091, "staged_skel": 42570, "mean_count": 87.17481994628906, "n_util_obs": 82, "util_count_spearman": 0.26428894052992263}, "t": 1039.8}
{"event": "eval", "step": 1200, "phase": 1, "bpb": {"en": 2.27207, "tr": 2.53558, "fi": 2.74921}, "bytes_per_chunk": {"en": 4.86, "tr": 5.986, "fi": 6.212}, "bpb_mean": 2.51895, "frozen_inventory": true, "t": 1042.2}
{"event": "train", "step": 1200, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.62245, "bpb": 2.3407, "chunk_loss": 0.030362, "F": 0.19476, "G": 0.23731, "chunks": 199.44, "bytes_per_chunk": 5.134, "boundary_entropy": 0.19028, "router_noise": 0.0, "grad_norm": 0.9011, "lr": 0.0002837, "router_trainable": true, "sec_per_step": 1.195, "probe_sec": 0.1631, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.13945, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 11, "mean_delta_loo": -0.04874, "n_shadow": 2, "mean_delta_shadow": -0.04809, "n_split": 1, "mean_delta_split": -0.01868, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 0.46589, "trunc_abs_frac": 0.76933, "mean_delta_full": 0.11765, "read_hit_rate": {"surface": 0.22551, "split": 0.45002, "merge": 0.07439, "skel": 0.24868}, "mean_gate": {"surface": 0.16265, "split": 0.1548, "merge": 0.17846, "skel": 0.17311}, "n_reads": 8598, "n_cand": 29827, "read_suppressed_frac": 0.0953, "dropout_reads_logged": 8598, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 367, "admitted_surface": 3646, "staged_surface": 55863, "admitted_split": 24495, "staged_split": 39131, "admitted_merge": 3304, "staged_merge": 124465, "admitted_skel": 34091, "staged_skel": 42685, "mean_count": 87.30606079101562, "n_util_obs": 82, "util_count_spearman": 0.26801116668242575}, "probe_overhead_sec_total": 142.988, "probe_overhead_seq_frac": 0.4996, "util_loss": 0.00664, "alpha": 0.03, "t": 1046.9}
{"event": "consolidate", "step": 1250, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262087, "t": 1093.5}
{"event": "train", "step": 1250, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.66368, "bpb": 2.40018, "chunk_loss": 0.030259, "F": 0.18634, "G": 0.23607, "chunks": 190.81, "bytes_per_chunk": 5.367, "boundary_entropy": 0.20409, "router_noise": 0.0, "grad_norm": 0.8762, "lr": 0.0002819, "router_trainable": true, "sec_per_step": 0.954, "probe_sec": 0.215, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.16937, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": 0.00245, "n_shadow": 6, "mean_delta_shadow": -0.11167, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": 4.00208, "trunc_abs_frac": 4.82583, "mean_delta_full": 0.03513, "read_hit_rate": {"surface": 0.19921, "split": 0.42164, "merge": 0.07051, "skel": 0.24167}, "mean_gate": {"surface": 0.15543, "split": 0.14851, "merge": 0.17079, "skel": 0.16625}, "n_reads": 7636, "n_cand": 28485, "read_suppressed_frac": 0.1003, "dropout_reads_logged": 7636, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 416, "admitted_surface": 3646, "staged_surface": 55726, "admitted_split": 25518, "staged_split": 38443, "admitted_merge": 3304, "staged_merge": 125444, "admitted_skel": 33068, "staged_skel": 42531, "mean_count": 94.47843933105469, "n_util_obs": 85, "util_count_spearman": 0.2919874926714872}, "probe_overhead_sec_total": 149.644, "probe_overhead_seq_frac": 0.4996, "util_loss": 0.006646, "alpha": 0.03, "t": 1094.6}
{"event": "consolidate", "step": 1300, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262083, "t": 1155.0}
{"event": "train", "step": 1300, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.58333, "bpb": 2.28426, "chunk_loss": 0.030369, "F": 0.19724, "G": 0.2325, "chunks": 201.97, "bytes_per_chunk": 5.07, "boundary_entropy": 0.1661, "router_noise": 0.0, "grad_norm": 0.8104, "lr": 0.00028, "router_trainable": true, "sec_per_step": 1.215, "probe_sec": 0.0861, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.05941, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 4, "mean_delta_loo": -0.08372, "n_shadow": 6, "mean_delta_shadow": -0.04792, "n_split": 1, "mean_delta_split": 0.00766, "n_merge": 1, "mean_delta_merge": -0.11926, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 0.90728, "trunc_abs_frac": 2.57674, "mean_delta_full": -0.04797, "read_hit_rate": {"surface": 0.23496, "split": 0.44102, "merge": 0.08408, "skel": 0.24865}, "mean_gate": {"surface": 0.17064, "split": 0.16339, "merge": 0.19083, "skel": 0.18305}, "n_reads": 8773, "n_cand": 30415, "read_suppressed_frac": 0.1006, "dropout_reads_logged": 8773, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 452, "admitted_surface": 3676, "staged_surface": 55463, "admitted_split": 26507, "staged_split": 37995, "admitted_merge": 3307, "staged_merge": 126427, "admitted_skel": 32046, "staged_skel": 42259, "mean_count": 101.81716918945312, "n_util_obs": 89, "util_count_spearman": 0.29765066394279877}, "probe_overhead_sec_total": 156.411, "probe_overhead_seq_frac": 0.4996, "util_loss": 0.006874, "alpha": 0.03, "t": 1155.4}
{"event": "consolidate", "step": 1350, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262084, "t": 1202.1}
{"event": "train", "step": 1350, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.61063, "bpb": 2.32365, "chunk_loss": 0.030364, "F": 0.19818, "G": 0.22292, "chunks": 202.94, "bytes_per_chunk": 5.046, "boundary_entropy": 0.14622, "router_noise": 0.0, "grad_norm": 0.9546, "lr": 0.000278, "router_trainable": true, "sec_per_step": 0.956, "probe_sec": 0.092, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.05241, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 4, "mean_delta_loo": -0.02484, "n_shadow": 2, "mean_delta_shadow": -0.19824, "n_split": 1, "mean_delta_split": -0.0686, "n_merge": 3, "mean_delta_merge": 0.11984, "n_dropped_unpaired": 0, "n_full": 0, "read_hit_rate": {"surface": 0.24992, "split": 0.45176, "merge": 0.086, "skel": 0.26281}, "mean_gate": {"surface": 0.17349, "split": 0.16779, "merge": 0.19645, "skel": 0.18669}, "n_reads": 9192, "n_cand": 30751, "read_suppressed_frac": 0.1001, "dropout_reads_logged": 9192, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 479, "admitted_surface": 3960, "staged_surface": 54416, "admitted_split": 27245, "staged_split": 37634, "admitted_merge": 3309, "staged_merge": 128277, "admitted_skel": 31022, "staged_skel": 41817, "mean_count": 108.33245849609375, "n_util_obs": 91, "util_count_spearman": 0.3337155598025163}, "probe_overhead_sec_total": 162.135, "probe_overhead_seq_frac": 0.4996, "util_loss": 0.006874, "alpha": 0.03, "t": 1203.2}
{"event": "consolidate", "step": 1400, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262077, "t": 1255.5}
{"event": "probe", "step": 1400, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3427.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.19692, "cps": 26.773438, "mean_p": 0.211726, "boundary_entropy": 0.127719, "bpb": 2.201106}, "tr": {"chunks": 3265.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.180357, "cps": 25.507812, "mean_p": 0.267818, "boundary_entropy": 0.32598, "bpb": 2.46304}, "fi": {"chunks": 2986.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.158737, "cps": 23.328125, "mean_p": 0.252006, "boundary_entropy": 0.338916, "bpb": 2.682784}}, "bpb": {"en": 2.20111, "tr": 2.46304, "fi": 2.68278}, "gini_cpc": 0.04749, "gini_cps": 0.030378, "bpic": 5.861955, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9158885926898123, "fi": 0.8060981709275303}, "inventory": {"size": 326656, "admitted": 65536, "staged": 261120, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 500, "admitted_surface": 4445, "staged_surface": 53963, "admitted_split": 27784, "staged_split": 38299, "admitted_merge": 3309, "staged_merge": 127278, "admitted_skel": 29998, "staged_skel": 41580, "mean_count": 115.75469970703125, "n_util_obs": 94, "util_count_spearman": 0.34735397175161653}, "t": 1257.1}
{"event": "eval", "step": 1400, "phase": 1, "bpb": {"en": 2.21463, "tr": 2.47075, "fi": 2.6899}, "bytes_per_chunk": {"en": 4.829, "tr": 5.899, "fi": 6.708}, "bpb_mean": 2.45843, "frozen_inventory": true, "t": 1260.1}
{"event": "train", "step": 1400, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.58132, "bpb": 2.28136, "chunk_loss": 0.03033, "F": 0.19556, "G": 0.22811, "chunks": 200.25, "bytes_per_chunk": 5.114, "boundary_entropy": 0.16574, "router_noise": 0.0, "grad_norm": 0.8817, "lr": 0.0002759, "router_trainable": true, "sec_per_step": 1.246, "probe_sec": 0.0851, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04946, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 4, "mean_delta_loo": -0.14595, "n_shadow": 4, "mean_delta_shadow": -0.18627, "n_split": 2, "mean_delta_split": 0.10013, "n_merge": 1, "mean_delta_merge": -0.18365, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 0.22405, "trunc_abs_frac": 0.28257, "mean_delta_full": -0.77277, "read_hit_rate": {"surface": 0.2558, "split": 0.43494, "merge": 0.08181, "skel": 0.24836}, "mean_gate": {"surface": 0.17492, "split": 0.16583, "merge": 0.19795, "skel": 0.18649}, "n_reads": 8775, "n_cand": 30239, "read_suppressed_frac": 0.102, "dropout_reads_logged": 8775, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 502, "admitted_surface": 4445, "staged_surface": 54110, "admitted_split": 27784, "staged_split": 38407, "admitted_merge": 3309, "staged_merge": 127886, "admitted_skel": 29998, "staged_skel": 41741, "mean_count": 115.88859558105469, "n_util_obs": 94, "util_count_spearman": 0.34735397175161653}, "probe_overhead_sec_total": 168.485, "probe_overhead_seq_frac": 0.4996, "util_loss": 0.00737, "alpha": 0.03, "t": 1265.5}
{"event": "consolidate", "step": 1450, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262077, "t": 1319.5}
{"event": "train", "step": 1450, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.55442, "bpb": 2.24256, "chunk_loss": 0.03045, "F": 0.19965, "G": 0.23523, "chunks": 204.44, "bytes_per_chunk": 5.009, "boundary_entropy": 0.17639, "router_noise": 0.0, "grad_norm": 0.8376, "lr": 0.0002738, "router_trainable": true, "sec_per_step": 1.101, "probe_sec": 0.0822, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.05203, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 5, "mean_delta_loo": 0.38412, "n_shadow": 5, "mean_delta_shadow": -0.04933, "n_split": 2, "mean_delta_split": -0.07468, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 1, "trunc_ratio_sum": 0.47479, "trunc_abs_frac": 0.47479, "mean_delta_full": -0.46484, "read_hit_rate": {"surface": 0.23041, "split": 0.43809, "merge": 0.0829, "skel": 0.25273}, "mean_gate": {"surface": 0.1825, "split": 0.17254, "merge": 0.2066, "skel": 0.19365}, "n_reads": 8735, "n_cand": 30487, "read_suppressed_frac": 0.1036, "dropout_reads_logged": 8735, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 552, "admitted_surface": 4937, "staged_surface": 53948, "admitted_split": 28316, "staged_split": 38233, "admitted_merge": 3309, "staged_merge": 128379, "admitted_skel": 28974, "staged_skel": 41584, "mean_count": 122.75416564941406, "n_util_obs": 95, "util_count_spearman": 0.36173138903983054}, "probe_overhead_sec_total": 173.665, "probe_overhead_seq_frac": 0.4997, "util_loss": 0.007273, "alpha": 0.03, "t": 1320.5}
{"event": "consolidate", "step": 1500, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262076, "t": 1372.4}
{"event": "train", "step": 1500, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.54907, "bpb": 2.23484, "chunk_loss": 0.030287, "F": 0.19333, "G": 0.22406, "chunks": 197.97, "bytes_per_chunk": 5.173, "boundary_entropy": 0.15999, "router_noise": 0.0, "grad_norm": 1.0063, "lr": 0.0002715, "router_trainable": true, "sec_per_step": 1.061, "probe_sec": 0.1288, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.05425, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": 0.04512, "n_shadow": 5, "mean_delta_shadow": -0.00254, "n_split": 1, "mean_delta_split": -0.15592, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": -0.07193, "trunc_abs_frac": 0.09167, "mean_delta_full": 0.25287, "read_hit_rate": {"surface": 0.2529, "split": 0.44421, "merge": 0.08627, "skel": 0.24829}, "mean_gate": {"surface": 0.18815, "split": 0.17752, "merge": 0.21331, "skel": 0.20131}, "n_reads": 8692, "n_cand": 29662, "read_suppressed_frac": 0.0974, "dropout_reads_logged": 8692, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 577, "admitted_surface": 5451, "staged_surface": 53624, "admitted_split": 28826, "staged_split": 37758, "admitted_merge": 3309, "staged_merge": 129272, "admitted_skel": 27950, "staged_skel": 41490, "mean_count": 129.26853942871094, "n_util_obs": 98, "util_count_spearman": 0.3589630791398096}, "probe_overhead_sec_total": 179.383, "probe_overhead_seq_frac": 0.4997, "util_loss": 0.007501, "alpha": 0.03, "t": 1373.6}
{"event": "consolidate", "step": 1550, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262074, "t": 1430.2}
{"event": "train", "step": 1550, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.54217, "bpb": 2.22488, "chunk_loss": 0.030372, "F": 0.19766, "G": 0.23031, "chunks": 202.41, "bytes_per_chunk": 5.059, "boundary_entropy": 0.16051, "router_noise": 0.0, "grad_norm": 0.9051, "lr": 0.0002692, "router_trainable": true, "sec_per_step": 1.148, "probe_sec": 0.0899, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.05071, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": 0.00349, "n_shadow": 3, "mean_delta_shadow": 0.1044, "n_split": 2, "mean_delta_split": 0.23396, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 0.88599, "trunc_abs_frac": 1.34991, "mean_delta_full": 0.11215, "read_hit_rate": {"surface": 0.25803, "split": 0.44353, "merge": 0.08951, "skel": 0.26629}, "mean_gate": {"surface": 0.19662, "split": 0.18379, "merge": 0.22202, "skel": 0.20933}, "n_reads": 9076, "n_cand": 30443, "read_suppressed_frac": 0.1053, "dropout_reads_logged": 9076, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 607, "admitted_surface": 6208, "staged_surface": 52923, "admitted_split": 29093, "staged_split": 38232, "admitted_merge": 3309, "staged_merge": 129828, "admitted_skel": 26926, "staged_skel": 41161, "mean_count": 136.8085174560547, "n_util_obs": 100, "util_count_spearman": 0.3574797479747975}, "probe_overhead_sec_total": 185.372, "probe_overhead_seq_frac": 0.4997, "util_loss": 0.007649, "alpha": 0.03, "t": 1431.0}
{"event": "consolidate", "step": 1600, "retired": 0, "split_retired": 1, "merge_admitted": 0, "admitted": 1, "swapped": 1024, "uncertain_staged": 262071, "t": 1485.3}
{"event": "probe", "step": 1600, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3394.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.195024, "cps": 26.515625, "mean_p": 0.207599, "boundary_entropy": 0.115589, "bpb": 2.15979}, "tr": {"chunks": 3257.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.179915, "cps": 25.445312, "mean_p": 0.260325, "boundary_entropy": 0.312542, "bpb": 2.41901}, "fi": {"chunks": 3478.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.184892, "cps": 27.171875, "mean_p": 0.275639, "boundary_entropy": 0.367346, "bpb": 2.608429}}, "bpb": {"en": 2.15979, "tr": 2.41901, "fi": 2.60843}, "gini_cpc": 0.017992, "gini_cps": 0.014546, "bpic": 5.600948, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.922527857437004, "fi": 0.9480472361189135}, "inventory": {"size": 326655, "admitted": 65536, "staged": 261119, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 628, "admitted_surface": 6978, "staged_surface": 52478, "admitted_split": 29341, "staged_split": 38428, "admitted_merge": 3309, "staged_merge": 129037, "admitted_skel": 25908, "staged_skel": 41176, "mean_count": 141.84185791015625, "n_util_obs": 100, "util_count_spearman": 0.34498649864986497}, "t": 1486.7}
{"event": "eval", "step": 1600, "phase": 1, "bpb": {"en": 2.17459, "tr": 2.42895, "fi": 2.61787}, "bytes_per_chunk": {"en": 4.873, "tr": 5.907, "fi": 5.597}, "bpb_mean": 2.40714, "frozen_inventory": true, "t": 1489.0}
{"event": "train", "step": 1600, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.51707, "bpb": 2.18867, "chunk_loss": 0.030329, "F": 0.19388, "G": 0.23058, "chunks": 198.53, "bytes_per_chunk": 5.158, "boundary_entropy": 0.16943, "router_noise": 0.0, "grad_norm": 0.7918, "lr": 0.0002668, "router_trainable": true, "sec_per_step": 1.271, "probe_sec": 0.1187, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.06797, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": -0.06734, "n_shadow": 6, "mean_delta_shadow": -0.00091, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 1.26588, "trunc_abs_frac": 0.70801, "mean_delta_full": 0.01297, "read_hit_rate": {"surface": 0.24252, "split": 0.44252, "merge": 0.07934, "skel": 0.26045}, "mean_gate": {"surface": 0.19669, "split": 0.18281, "merge": 0.22577, "skel": 0.21482}, "n_reads": 8700, "n_cand": 29795, "read_suppressed_frac": 0.103, "dropout_reads_logged": 8700, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 627, "admitted_surface": 6978, "staged_surface": 52658, "admitted_split": 29341, "staged_split": 38494, "admitted_merge": 3309, "staged_merge": 129674, "admitted_skel": 25908, "staged_skel": 41318, "mean_count": 141.974609375, "n_util_obs": 100, "util_count_spearman": 0.34513051305130515}, "probe_overhead_sec_total": 191.07, "probe_overhead_seq_frac": 0.4997, "util_loss": 0.007676, "alpha": 0.03, "t": 1494.5}
{"event": "consolidate", "step": 1650, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262067, "t": 1548.9}
{"event": "train", "step": 1650, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.56956, "bpb": 2.26439, "chunk_loss": 0.03036, "F": 0.19836, "G": 0.22708, "chunks": 203.12, "bytes_per_chunk": 5.041, "boundary_entropy": 0.15332, "router_noise": 0.0, "grad_norm": 0.7686, "lr": 0.0002643, "router_trainable": true, "sec_per_step": 1.101, "probe_sec": 0.07, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04332, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": 0.16335, "n_shadow": 5, "mean_delta_shadow": -0.01435, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 1.04048, "trunc_abs_frac": 0.84799, "mean_delta_full": 0.16226, "read_hit_rate": {"surface": 0.24335, "split": 0.43732, "merge": 0.0881, "skel": 0.24601}, "mean_gate": {"surface": 0.2031, "split": 0.18843, "merge": 0.22776, "skel": 0.2185}, "n_reads": 8838, "n_cand": 30574, "read_suppressed_frac": 0.0979, "dropout_reads_logged": 8838, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 675, "admitted_surface": 7902, "staged_surface": 52065, "admitted_split": 29440, "staged_split": 38200, "admitted_merge": 3309, "staged_merge": 130714, "admitted_skel": 24885, "staged_skel": 41165, "mean_count": 148.29977416992188, "n_util_obs": 102, "util_count_spearman": 0.34257183327297425}, "probe_overhead_sec_total": 196.84, "probe_overhead_seq_frac": 0.4997, "util_loss": 0.007802, "alpha": 0.03, "t": 1549.6}
{"event": "consolidate", "step": 1700, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262066, "t": 1604.2}
{"event": "train", "step": 1700, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.5657, "bpb": 2.25883, "chunk_loss": 0.03029, "F": 0.19235, "G": 0.22709, "chunks": 196.97, "bytes_per_chunk": 5.199, "boundary_entropy": 0.16295, "router_noise": 0.0, "grad_norm": 0.8037, "lr": 0.0002618, "router_trainable": true, "sec_per_step": 1.135, "probe_sec": 0.1585, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.07983, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": -0.02529, "n_shadow": 6, "mean_delta_shadow": 0.21996, "n_split": 1, "mean_delta_split": 0.12854, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": -0.6998, "trunc_abs_frac": 0.42144, "mean_delta_full": -0.07295, "read_hit_rate": {"surface": 0.22453, "split": 0.43035, "merge": 0.07293, "skel": 0.24935}, "mean_gate": {"surface": 0.20339, "split": 0.18944, "merge": 0.23666, "skel": 0.22399}, "n_reads": 8229, "n_cand": 29421, "read_suppressed_frac": 0.0988, "dropout_reads_logged": 8229, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 697, "admitted_surface": 8838, "staged_surface": 51428, "admitted_split": 29526, "staged_split": 37714, "admitted_merge": 3309, "staged_merge": 132276, "admitted_skel": 23863, "staged_skel": 40726, "mean_count": 154.7772674560547, "n_util_obs": 107, "util_count_spearman": 0.3285526754050824}, "probe_overhead_sec_total": 202.804, "probe_overhead_seq_frac": 0.4997, "util_loss": 0.007214, "alpha": 0.03, "t": 1606.4}
{"event": "consolidate", "step": 1750, "retired": 1, "split_retired": 0, "merge_admitted": 0, "admitted": 1, "swapped": 1024, "uncertain_staged": 262066, "t": 1651.0}
{"event": "train", "step": 1750, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.53361, "bpb": 2.21253, "chunk_loss": 0.030397, "F": 0.19983, "G": 0.22301, "chunks": 204.62, "bytes_per_chunk": 5.004, "boundary_entropy": 0.13189, "router_noise": 0.0, "grad_norm": 0.7753, "lr": 0.0002591, "router_trainable": true, "sec_per_step": 0.911, "probe_sec": 0.1204, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.09522, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": 0.06726, "n_shadow": 4, "mean_delta_shadow": -0.09903, "n_split": 0, "mean_delta_split": NaN, "n_merge": 1, "mean_delta_merge": -0.13855, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 0.39498, "trunc_abs_frac": 4.31286, "mean_delta_full": -0.31413, "read_hit_rate": {"surface": 0.24355, "split": 0.44258, "merge": 0.08313, "skel": 0.26371}, "mean_gate": {"surface": 0.21387, "split": 0.1975, "merge": 0.24063, "skel": 0.22909}, "n_reads": 8983, "n_cand": 30697, "read_suppressed_frac": 0.0993, "dropout_reads_logged": 8983, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 729, "admitted_surface": 9846, "staged_surface": 51547, "admitted_split": 29543, "staged_split": 38267, "admitted_merge": 3308, "staged_merge": 131476, "admitted_skel": 22839, "staged_skel": 40854, "mean_count": 161.48651123046875, "n_util_obs": 111, "util_count_spearman": 0.3372418770670564}, "probe_overhead_sec_total": 208.785, "probe_overhead_seq_frac": 0.4997, "util_loss": 0.008198, "alpha": 0.03, "t": 1651.9}
{"event": "consolidate", "step": 1800, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262065, "t": 1705.8}
{"event": "probe", "step": 1800, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3409.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.195886, "cps": 26.632812, "mean_p": 0.208567, "boundary_entropy": 0.112756, "bpb": 2.122064}, "tr": {"chunks": 3506.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.19367, "cps": 27.390625, "mean_p": 0.257939, "boundary_entropy": 0.290682, "bpb": 2.365584}, "fi": {"chunks": 3122.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.165967, "cps": 24.390625, "mean_p": 0.262403, "boundary_entropy": 0.350361, "bpb": 2.578153}}, "bpb": {"en": 2.12206, "tr": 2.36558, "fi": 2.57815}, "gini_cpc": 0.035905, "gini_cps": 0.025506, "bpic": 5.652287, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9886862158192049, "fi": 0.8472627914310699}, "inventory": {"size": 326656, "admitted": 65536, "staged": 261120, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 750, "admitted_surface": 10827, "staged_surface": 51144, "admitted_split": 29586, "staged_split": 37815, "admitted_merge": 3308, "staged_merge": 131146, "admitted_skel": 21815, "staged_skel": 41015, "mean_count": 167.87069702148438, "n_util_obs": 115, "util_count_spearman": 0.3398412764176036}, "t": 1707.2}
{"event": "eval", "step": 1800, "phase": 1, "bpb": {"en": 2.13739, "tr": 2.37409, "fi": 2.58573}, "bytes_per_chunk": {"en": 4.844, "tr": 5.602, "fi": 6.246}, "bpb_mean": 2.36574, "frozen_inventory": true, "t": 1709.9}
{"event": "train", "step": 1800, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.5839, "bpb": 2.28508, "chunk_loss": 0.030426, "F": 0.19839, "G": 0.23517, "chunks": 203.16, "bytes_per_chunk": 5.04, "boundary_entropy": 0.17113, "router_noise": 0.0, "grad_norm": 0.8732, "lr": 0.0002564, "router_trainable": true, "sec_per_step": 1.26, "probe_sec": 0.0906, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.05804, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 4, "mean_delta_loo": 0.15363, "n_shadow": 4, "mean_delta_shadow": 0.06747, "n_split": 0, "mean_delta_split": NaN, "n_merge": 3, "mean_delta_merge": -0.07854, "n_dropped_unpaired": 0, "n_full": 1, "trunc_ratio_sum": 0.57794, "trunc_abs_frac": 0.57794, "mean_delta_full": 0.81055, "read_hit_rate": {"surface": 0.24223, "split": 0.42389, "merge": 0.07146, "skel": 0.25831}, "mean_gate": {"surface": 0.20624, "split": 0.1904, "merge": 0.2386, "skel": 0.22815}, "n_reads": 8383, "n_cand": 29839, "read_suppressed_frac": 0.0984, "dropout_reads_logged": 8383, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 751, "admitted_surface": 10827, "staged_surface": 51366, "admitted_split": 29586, "staged_split": 37978, "admitted_merge": 3308, "staged_merge": 131668, "admitted_skel": 21815, "staged_skel": 41132, "mean_count": 167.9986114501953, "n_util_obs": 115, "util_count_spearman": 0.339343486151661}, "probe_overhead_sec_total": 215.863, "probe_overhead_seq_frac": 0.4997, "util_loss": 0.008252, "alpha": 0.03, "t": 1714.9}
{"event": "consolidate", "step": 1850, "retired": 0, "split_retired": 1, "merge_admitted": 0, "admitted": 1, "swapped": 1024, "uncertain_staged": 262062, "t": 1766.6}
{"event": "train", "step": 1850, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.53039, "bpb": 2.20788, "chunk_loss": 0.030371, "F": 0.19638, "G": 0.23192, "chunks": 201.09, "bytes_per_chunk": 5.092, "boundary_entropy": 0.16637, "router_noise": 0.0, "grad_norm": 0.7227, "lr": 0.0002537, "router_trainable": true, "sec_per_step": 1.056, "probe_sec": 0.1783, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.13802, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": -0.09578, "n_shadow": 5, "mean_delta_shadow": -0.0263, "n_split": 1, "mean_delta_split": 0.01978, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 2.99252, "trunc_abs_frac": 0.90661, "mean_delta_full": -0.06934, "read_hit_rate": {"surface": 0.23427, "split": 0.42706, "merge": 0.0733, "skel": 0.26237}, "mean_gate": {"surface": 0.2067, "split": 0.19182, "merge": 0.24103, "skel": 0.22831}, "n_reads": 8545, "n_cand": 30204, "read_suppressed_frac": 0.096, "dropout_reads_logged": 8545, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 777, "admitted_surface": 11850, "staged_surface": 51269, "admitted_split": 29587, "staged_split": 38424, "admitted_merge": 3308, "staged_merge": 131458, "admitted_skel": 20791, "staged_skel": 40993, "mean_count": 173.6407012939453, "n_util_obs": 116, "util_count_spearman": 0.3076000461307808}, "probe_overhead_sec_total": 222.329, "probe_overhead_seq_frac": 0.4997, "util_loss": 0.008435, "alpha": 0.03, "t": 1767.7}
{"event": "consolidate", "step": 1900, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262061, "t": 1803.4}
{"event": "train", "step": 1900, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.57846, "bpb": 2.27724, "chunk_loss": 0.030397, "F": 0.1994, "G": 0.22725, "chunks": 204.19, "bytes_per_chunk": 5.015, "boundary_entropy": 0.1467, "router_noise": 0.0, "grad_norm": 0.7874, "lr": 0.0002508, "router_trainable": true, "sec_per_step": 0.725, "probe_sec": 0.1135, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.08825, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 10, "mean_delta_loo": 0.01971, "n_shadow": 2, "mean_delta_shadow": 0.01829, "n_split": 0, "mean_delta_split": NaN, "n_merge": 1, "mean_delta_merge": -0.0498, "n_dropped_unpaired": 0, "n_full": 1, "trunc_ratio_sum": 0.69746, "trunc_abs_frac": 0.69746, "mean_delta_full": -0.20215, "read_hit_rate": {"surface": 0.23224, "split": 0.41061, "merge": 0.0779, "skel": 0.2477}, "mean_gate": {"surface": 0.21843, "split": 0.20072, "merge": 0.2525, "skel": 0.24468}, "n_reads": 8386, "n_cand": 30603, "read_suppressed_frac": 0.1029, "dropout_reads_logged": 8386, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 795, "admitted_surface": 12873, "staged_surface": 51053, "admitted_split": 29587, "staged_split": 38469, "admitted_merge": 3308, "staged_merge": 131503, "admitted_skel": 19768, "staged_skel": 41119, "mean_count": 180.0238494873047, "n_util_obs": 119, "util_count_spearman": 0.29260789061387266}, "probe_overhead_sec_total": 228.743, "probe_overhead_seq_frac": 0.4997, "util_loss": 0.008108, "alpha": 0.03, "t": 1804.0}
{"event": "consolidate", "step": 1950, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262057, "t": 1839.5}
{"event": "train", "step": 1950, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.53162, "bpb": 2.20966, "chunk_loss": 0.030323, "F": 0.19562, "G": 0.22728, "chunks": 200.31, "bytes_per_chunk": 5.112, "boundary_entropy": 0.15087, "router_noise": 0.0, "grad_norm": 0.7425, "lr": 0.0002479, "router_trainable": true, "sec_per_step": 0.722, "probe_sec": 0.1033, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.08268, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": -0.02125, "n_shadow": 4, "mean_delta_shadow": -0.08095, "n_split": 1, "mean_delta_split": -0.17114, "n_merge": 1, "mean_delta_merge": 0.23889, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 0.51048, "trunc_abs_frac": 0.35042, "mean_delta_full": 0.05632, "read_hit_rate": {"surface": 0.2391, "split": 0.42553, "merge": 0.07753, "skel": 0.24064}, "mean_gate": {"surface": 0.22014, "split": 0.20121, "merge": 0.25578, "skel": 0.24203}, "n_reads": 8381, "n_cand": 29978, "read_suppressed_frac": 0.1042, "dropout_reads_logged": 8381, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 831, "admitted_surface": 13896, "staged_surface": 50656, "admitted_split": 29586, "staged_split": 38383, "admitted_merge": 3308, "staged_merge": 132225, "admitted_skel": 18746, "staged_skel": 40880, "mean_count": 186.8673553466797, "n_util_obs": 124, "util_count_spearman": 0.30438394964594806}, "probe_overhead_sec_total": 234.952, "probe_overhead_seq_frac": 0.4997, "util_loss": 0.00808, "alpha": 0.03, "t": 1840.1}
{"event": "consolidate", "step": 2000, "retired": 1, "split_retired": 1, "merge_admitted": 0, "admitted": 2, "swapped": 1024, "uncertain_staged": 262057, "t": 1877.1}
{"event": "probe", "step": 2000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3409.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.195886, "cps": 26.632812, "mean_p": 0.205659, "boundary_entropy": 0.10053, "bpb": 2.090678}, "tr": {"chunks": 3343.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.184666, "cps": 26.117188, "mean_p": 0.248446, "boundary_entropy": 0.281553, "bpb": 2.329336}, "fi": {"chunks": 2965.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.157621, "cps": 23.164062, "mean_p": 0.27279, "boundary_entropy": 0.363708, "bpb": 2.51025}}, "bpb": {"en": 2.09068, "tr": 2.32934, "fi": 2.51025}, "gini_cpc": 0.047401, "gini_cps": 0.030462, "bpic": 5.838427, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9427204847357679, "fi": 0.8046554056992704}, "inventory": {"size": 326654, "admitted": 65536, "staged": 261118, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 843, "admitted_surface": 14921, "staged_surface": 50676, "admitted_split": 29583, "staged_split": 38929, "admitted_merge": 3301, "staged_merge": 130621, "admitted_skel": 17731, "staged_skel": 40892, "mean_count": 187.65000915527344, "n_util_obs": 126, "util_count_spearman": 0.3215782850821541}, "t": 1878.6}
{"event": "eval", "step": 2000, "phase": 1, "bpb": {"en": 2.10732, "tr": 2.33564, "fi": 2.51769}, "bytes_per_chunk": {"en": 4.851, "tr": 5.715, "fi": 6.416}, "bpb_mean": 2.32022, "frozen_inventory": true, "t": 1881.2}
{"event": "train", "step": 2000, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.55284, "bpb": 2.24027, "chunk_loss": 0.030261, "F": 0.1922, "G": 0.23361, "chunks": 196.81, "bytes_per_chunk": 5.203, "boundary_entropy": 0.17157, "router_noise": 0.0, "grad_norm": 0.7795, "lr": 0.000245, "router_trainable": true, "sec_per_step": 0.909, "probe_sec": 0.1036, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.08527, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 3, "mean_delta_loo": 0.11658, "n_shadow": 5, "mean_delta_shadow": -0.06687, "n_split": 3, "mean_delta_split": 0.17973, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 0.89188, "trunc_abs_frac": 1.86053, "mean_delta_full": 0.13737, "read_hit_rate": {"surface": 0.23077, "split": 0.39969, "merge": 0.07058, "skel": 0.24257}, "mean_gate": {"surface": 0.2183, "split": 0.19858, "merge": 0.25022, "skel": 0.24241}, "n_reads": 7788, "n_cand": 29276, "read_suppressed_frac": 0.1017, "dropout_reads_logged": 7788, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 844, "admitted_surface": 14921, "staged_surface": 50935, "admitted_split": 29583, "staged_split": 39079, "admitted_merge": 3301, "staged_merge": 131082, "admitted_skel": 17731, "staged_skel": 41048, "mean_count": 187.7688446044922, "n_util_obs": 126, "util_count_spearman": 0.3205549306336708}, "probe_overhead_sec_total": 241.403, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.008971, "alpha": 0.03, "t": 1885.5}
{"event": "consolidate", "step": 2050, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262056, "t": 1921.0}
{"event": "train", "step": 2050, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.60758, "bpb": 2.31925, "chunk_loss": 0.030345, "F": 0.19415, "G": 0.23013, "chunks": 198.81, "bytes_per_chunk": 5.151, "boundary_entropy": 0.16445, "router_noise": 0.0, "grad_norm": 0.8272, "lr": 0.0002419, "router_trainable": true, "sec_per_step": 0.722, "probe_sec": 0.1647, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.1219, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": 0.04652, "n_shadow": 4, "mean_delta_shadow": 0.27023, "n_split": 1, "mean_delta_split": -0.12244, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": -1.36598, "trunc_abs_frac": 1.04683, "mean_delta_full": -0.03552, "read_hit_rate": {"surface": 0.21548, "split": 0.38473, "merge": 0.0705, "skel": 0.23331}, "mean_gate": {"surface": 0.21438, "split": 0.19306, "merge": 0.25533, "skel": 0.23813}, "n_reads": 7538, "n_cand": 29502, "read_suppressed_frac": 0.1081, "dropout_reads_logged": 7538, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 883, "admitted_surface": 15945, "staged_surface": 50550, "admitted_split": 29583, "staged_split": 38312, "admitted_merge": 3301, "staged_merge": 132374, "admitted_skel": 16707, "staged_skel": 40908, "mean_count": 194.06227111816406, "n_util_obs": 130, "util_count_spearman": 0.3032760245258026}, "probe_overhead_sec_total": 247.455, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.008579, "alpha": 0.03, "t": 1921.6}
{"event": "consolidate", "step": 2100, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262051, "t": 1956.9}
{"event": "train", "step": 2100, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.49554, "bpb": 2.1576, "chunk_loss": 0.03033, "F": 0.19669, "G": 0.22715, "chunks": 201.41, "bytes_per_chunk": 5.084, "boundary_entropy": 0.14455, "router_noise": 0.0, "grad_norm": 0.7578, "lr": 0.0002388, "router_trainable": true, "sec_per_step": 0.728, "probe_sec": 0.1324, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.10904, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 5, "mean_delta_loo": -0.03804, "n_shadow": 3, "mean_delta_shadow": -0.10482, "n_split": 2, "mean_delta_split": 0.19077, "n_merge": 1, "mean_delta_merge": -0.43036, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 1.85196, "trunc_abs_frac": 55.40865, "mean_delta_full": -0.09256, "read_hit_rate": {"surface": 0.24544, "split": 0.41253, "merge": 0.08181, "skel": 0.26189}, "mean_gate": {"surface": 0.22647, "split": 0.20211, "merge": 0.26724, "skel": 0.25766}, "n_reads": 8444, "n_cand": 30102, "read_suppressed_frac": 0.1047, "dropout_reads_logged": 8444, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 911, "admitted_surface": 16967, "staged_surface": 50499, "admitted_split": 29582, "staged_split": 38678, "admitted_merge": 3276, "staged_merge": 131984, "admitted_skel": 15711, "staged_skel": 40983, "mean_count": 201.36685180664062, "n_util_obs": 136, "util_count_spearman": 0.33633494665175273}, "probe_overhead_sec_total": 253.542, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.008022, "alpha": 0.03, "t": 1958.0}
{"event": "consolidate", "step": 2150, "retired": 0, "split_retired": 1, "merge_admitted": 0, "admitted": 1, "swapped": 1024, "uncertain_staged": 262045, "t": 2000.6}
{"event": "train", "step": 2150, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.51035, "bpb": 2.17897, "chunk_loss": 0.030226, "F": 0.19089, "G": 0.22204, "chunks": 195.47, "bytes_per_chunk": 5.239, "boundary_entropy": 0.14505, "router_noise": 0.0, "grad_norm": 0.7131, "lr": 0.0002357, "router_trainable": true, "sec_per_step": 0.884, "probe_sec": 0.1725, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.12962, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": 0.1536, "n_shadow": 4, "mean_delta_shadow": 0.05544, "n_split": 0, "mean_delta_split": NaN, "n_merge": 1, "mean_delta_merge": -0.09192, "n_dropped_unpaired": 0, "n_full": 6, "trunc_ratio_sum": 0.62944, "trunc_abs_frac": 0.76152, "mean_delta_full": 0.26127, "read_hit_rate": {"surface": 0.23574, "split": 0.40892, "merge": 0.08254, "skel": 0.25481}, "mean_gate": {"surface": 0.21895, "split": 0.19814, "merge": 0.25087, "skel": 0.24671}, "n_reads": 8120, "n_cand": 29373, "read_suppressed_frac": 0.0996, "dropout_reads_logged": 8120, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 921, "admitted_surface": 17988, "staged_surface": 50637, "admitted_split": 29583, "staged_split": 39097, "admitted_merge": 3276, "staged_merge": 131384, "admitted_skel": 14689, "staged_skel": 41026, "mean_count": 207.26841735839844, "n_util_obs": 141, "util_count_spearman": 0.32654651312128086}, "probe_overhead_sec_total": 259.651, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.009344, "alpha": 0.03, "t": 2002.2}
{"event": "consolidate", "step": 2200, "retired": 0, "split_retired": 1, "merge_admitted": 0, "admitted": 1, "swapped": 1024, "uncertain_staged": 262043, "t": 2045.0}
{"event": "probe", "step": 2200, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3409.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.195886, "cps": 26.632812, "mean_p": 0.205095, "boundary_entropy": 0.096454, "bpb": 2.058513}, "tr": {"chunks": 3212.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.177429, "cps": 25.09375, "mean_p": 0.240086, "boundary_entropy": 0.2648, "bpb": 2.28894}, "fi": {"chunks": 2776.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.147573, "cps": 21.6875, "mean_p": 0.257964, "boundary_entropy": 0.346984, "bpb": 2.471862}}, "bpb": {"en": 2.05851, "tr": 2.28894, "fi": 2.47186}, "gini_cpc": 0.061834, "gini_cps": 0.044908, "bpic": 6.037246, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9057787008588951, "fi": 0.7533637120476138}, "inventory": {"size": 326655, "admitted": 65536, "staged": 261119, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 963, "admitted_surface": 18988, "staged_surface": 50038, "admitted_split": 29605, "staged_split": 38446, "admitted_merge": 3276, "staged_merge": 131747, "admitted_skel": 13667, "staged_skel": 40888, "mean_count": 217.01634216308594, "n_util_obs": 149, "util_count_spearman": 0.3067334543203278}, "t": 2046.4}
{"event": "eval", "step": 2200, "phase": 1, "bpb": {"en": 2.07457, "tr": 2.29581, "fi": 2.48033}, "bytes_per_chunk": {"en": 4.84, "tr": 5.903, "fi": 6.847}, "bpb_mean": 2.28357, "frozen_inventory": true, "t": 2048.8}
{"event": "train", "step": 2200, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.47691, "bpb": 2.13073, "chunk_loss": 0.030356, "F": 0.19897, "G": 0.2232, "chunks": 203.75, "bytes_per_chunk": 5.026, "boundary_entropy": 0.12856, "router_noise": 0.0, "grad_norm": 0.6853, "lr": 0.0002325, "router_trainable": true, "sec_per_step": 1.038, "probe_sec": 0.1125, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.0553, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": 0.02276, "n_shadow": 4, "mean_delta_shadow": -0.03896, "n_split": 0, "mean_delta_split": NaN, "n_merge": 1, "mean_delta_merge": 0.12061, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": -0.57443, "trunc_abs_frac": 0.60348, "mean_delta_full": -0.12048, "read_hit_rate": {"surface": 0.25, "split": 0.41667, "merge": 0.08473, "skel": 0.26243}, "mean_gate": {"surface": 0.22722, "split": 0.20735, "merge": 0.26656, "skel": 0.26315}, "n_reads": 8663, "n_cand": 30510, "read_suppressed_frac": 0.0971, "dropout_reads_logged": 8663, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 963, "admitted_surface": 18988, "staged_surface": 50198, "admitted_split": 29605, "staged_split": 38622, "admitted_merge": 3276, "staged_merge": 132313, "admitted_skel": 13667, "staged_skel": 41011, "mean_count": 217.14852905273438, "n_util_obs": 149, "util_count_spearman": 0.30662434246326864}, "probe_overhead_sec_total": 265.816, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.009262, "alpha": 0.03, "t": 2054.1}
{"event": "consolidate", "step": 2250, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262047, "t": 2104.4}
{"event": "train", "step": 2250, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.5154, "bpb": 2.18627, "chunk_loss": 0.030264, "F": 0.19235, "G": 0.22501, "chunks": 196.97, "bytes_per_chunk": 5.199, "boundary_entropy": 0.14758, "router_noise": 0.0, "grad_norm": 0.7822, "lr": 0.0002293, "router_trainable": true, "sec_per_step": 1.047, "probe_sec": 0.1046, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.06082, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": 0.0747, "n_shadow": 5, "mean_delta_shadow": -0.04082, "n_split": 1, "mean_delta_split": -0.06946, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 0.38351, "trunc_abs_frac": 0.27512, "mean_delta_full": -0.06812, "read_hit_rate": {"surface": 0.2872, "split": 0.47728, "merge": 0.08431, "skel": 0.29359}, "mean_gate": {"surface": 0.22705, "split": 0.20151, "merge": 0.25713, "skel": 0.26455}, "n_reads": 9475, "n_cand": 29475, "read_suppressed_frac": 0.1007, "dropout_reads_logged": 9475, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 992, "admitted_surface": 19765, "staged_surface": 50106, "admitted_split": 29789, "staged_split": 38321, "admitted_merge": 3323, "staged_merge": 132882, "admitted_skel": 12659, "staged_skel": 40835, "mean_count": 237.6670379638672, "n_util_obs": 165, "util_count_spearman": 0.2783308032557477}, "probe_overhead_sec_total": 271.579, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.009412, "alpha": 0.03, "t": 2106.5}
{"event": "consolidate", "step": 2300, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 605, "uncertain_staged": 262055, "t": 2157.6}
{"event": "train", "step": 2300, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.49692, "bpb": 2.15959, "chunk_loss": 0.030369, "F": 0.19958, "G": 0.21692, "chunks": 204.38, "bytes_per_chunk": 5.01, "boundary_entropy": 0.11253, "router_noise": 0.0, "grad_norm": 0.7848, "lr": 0.000226, "router_trainable": true, "sec_per_step": 1.04, "probe_sec": 0.3095, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.27859, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": 0.04169, "n_shadow": 7, "mean_delta_shadow": -0.02487, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 7, "trunc_ratio_sum": 0.74457, "trunc_abs_frac": 0.67658, "mean_delta_full": -0.08369, "read_hit_rate": {"surface": 0.39398, "split": 0.53534, "merge": 0.12708, "skel": 0.33481}, "mean_gate": {"surface": 0.21986, "split": 0.19257, "merge": 0.24927, "skel": 0.24912}, "n_reads": 11788, "n_cand": 30703, "read_suppressed_frac": 0.0958, "dropout_reads_logged": 11788, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1030, "admitted_surface": 20023, "staged_surface": 50792, "admitted_split": 29986, "staged_split": 38472, "admitted_merge": 3455, "staged_merge": 131995, "admitted_skel": 12072, "staged_skel": 40885, "mean_count": 259.779296875, "n_util_obs": 171, "util_count_spearman": 0.2573704954583799}, "probe_overhead_sec_total": 277.846, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.009043, "alpha": 0.03, "t": 2158.5}
{"event": "consolidate", "step": 2350, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 513, "uncertain_staged": 262060, "t": 2203.6}
{"event": "train", "step": 2350, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.49499, "bpb": 2.15681, "chunk_loss": 0.03023, "F": 0.18759, "G": 0.22965, "chunks": 192.09, "bytes_per_chunk": 5.331, "boundary_entropy": 0.17192, "router_noise": 0.0, "grad_norm": 0.7348, "lr": 0.0002226, "router_trainable": true, "sec_per_step": 0.934, "probe_sec": 0.3116, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.28846, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 11, "mean_delta_loo": -0.0139, "n_shadow": 3, "mean_delta_shadow": -0.12844, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": 0.16257, "trunc_abs_frac": 1.07373, "mean_delta_full": -0.12007, "read_hit_rate": {"surface": 0.37645, "split": 0.51854, "merge": 0.10702, "skel": 0.29998}, "mean_gate": {"surface": 0.20203, "split": 0.1762, "merge": 0.22766, "skel": 0.22696}, "n_reads": 10398, "n_cand": 28607, "read_suppressed_frac": 0.0983, "dropout_reads_logged": 10398, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1076, "admitted_surface": 20213, "staged_surface": 50497, "admitted_split": 30265, "staged_split": 37813, "admitted_merge": 3499, "staged_merge": 133009, "admitted_skel": 11559, "staged_skel": 40825, "mean_count": 270.572265625, "n_util_obs": 183, "util_count_spearman": 0.24485416104727553}, "probe_overhead_sec_total": 287.89, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.010033, "alpha": 0.03, "t": 2205.2}
{"event": "consolidate", "step": 2400, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 238, "uncertain_staged": 262061, "t": 2259.8}
{"event": "probe", "step": 2400, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3417.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.196345, "cps": 26.695312, "mean_p": 0.205002, "boundary_entropy": 0.092218, "bpb": 2.037302}, "tr": {"chunks": 3337.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.184334, "cps": 26.070312, "mean_p": 0.242565, "boundary_entropy": 0.264407, "bpb": 2.257742}, "fi": {"chunks": 3369.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.179097, "cps": 26.320312, "mean_p": 0.269426, "boundary_entropy": 0.347175, "bpb": 2.441031}}, "bpb": {"en": 2.0373, "tr": 2.25774, "fi": 2.44103}, "gini_cpc": 0.020542, "gini_cps": 0.005269, "bpic": 5.604268, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9388253257855242, "fi": 0.9121542142160016}, "inventory": {"size": 327442, "admitted": 65536, "staged": 261906, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1101, "admitted_surface": 20309, "staged_surface": 50625, "admitted_split": 30392, "staged_split": 37932, "admitted_merge": 3508, "staged_merge": 132451, "admitted_skel": 11327, "staged_skel": 40898, "mean_count": 283.67279052734375, "n_util_obs": 189, "util_count_spearman": 0.22103130165840132}, "t": 2261.7}
{"event": "eval", "step": 2400, "phase": 1, "bpb": {"en": 2.05196, "tr": 2.26191, "fi": 2.45297}, "bytes_per_chunk": {"en": 4.839, "tr": 5.715, "fi": 5.822}, "bpb_mean": 2.25561, "frozen_inventory": true, "t": 2265.1}
{"event": "phase_transition", "step": 2400, "phase": 2, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0, "t": 2268.7}
{"event": "train", "step": 2400, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.4455, "bpb": 2.08542, "chunk_loss": 0.030488, "F": 0.20529, "G": 0.22358, "chunks": 210.22, "bytes_per_chunk": 4.871, "boundary_entropy": 0.11717, "router_noise": 0.0, "grad_norm": 0.7299, "lr": 0.0002192, "router_trainable": true, "sec_per_step": 1.287, "probe_sec": 0.1913, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.17043, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 3, "mean_delta_loo": -0.04207, "n_shadow": 7, "mean_delta_shadow": 0.00444, "n_split": 0, "mean_delta_split": NaN, "n_merge": 2, "mean_delta_merge": -0.01579, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 0.44681, "trunc_abs_frac": 0.63422, "mean_delta_full": -0.13578, "read_hit_rate": {"surface": 0.43301, "split": 0.53314, "merge": 0.14708, "skel": 0.33131}, "mean_gate": {"surface": 0.20015, "split": 0.17788, "merge": 0.22157, "skel": 0.22704}, "n_reads": 12401, "n_cand": 31396, "read_suppressed_frac": 0.0973, "dropout_reads_logged": 12401, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1102, "admitted_surface": 20309, "staged_surface": 50620, "admitted_split": 30392, "staged_split": 37860, "admitted_merge": 3508, "staged_merge": 132757, "admitted_skel": 11327, "staged_skel": 40907, "mean_count": 283.86199951171875, "n_util_obs": 189, "util_count_spearman": 0.22103130165840132}, "probe_overhead_sec_total": 302.477, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.010226, "alpha": 0.03, "t": 2269.5}
{"event": "consolidate", "step": 2450, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 467, "uncertain_staged": 262063, "t": 2312.3}
{"event": "train", "step": 2450, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.63261, "bpb": 2.35535, "chunk_loss": 0.030418, "F": 0.19958, "G": 0.22423, "chunks": 204.38, "bytes_per_chunk": 5.01, "boundary_entropy": 0.13109, "router_noise": 0.0, "grad_norm": 0.7186, "lr": 0.0002158, "router_trainable": true, "sec_per_step": 0.871, "probe_sec": 0.1394, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.11659, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 4, "mean_delta_loo": -0.05281, "n_shadow": 4, "mean_delta_shadow": 0.03884, "n_split": 2, "mean_delta_split": 0.0284, "n_merge": 1, "mean_delta_merge": -0.30212, "n_dropped_unpaired": 0, "n_full": 1, "trunc_ratio_sum": 0.98642, "trunc_abs_frac": 0.98642, "mean_delta_full": -0.23376, "read_hit_rate": {"surface": 0.39889, "split": 0.50934, "merge": 0.11813, "skel": 0.30658}, "mean_gate": {"surface": 0.20474, "split": 0.17891, "merge": 0.22416, "skel": 0.23219}, "n_reads": 11087, "n_cand": 30186, "read_suppressed_frac": 0.1033, "dropout_reads_logged": 11087, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1132, "admitted_surface": 20397, "staged_surface": 50440, "admitted_split": 30738, "staged_split": 37807, "admitted_merge": 3528, "staged_merge": 133249, "admitted_skel": 10873, "staged_skel": 40648, "mean_count": 294.4484558105469, "n_util_obs": 199, "util_count_spearman": 0.2081970458352368}, "probe_overhead_sec_total": 312.09, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.010662, "alpha": 0.03, "t": 2313.1}
{"event": "consolidate", "step": 2500, "retired": 0, "split_retired": 1, "merge_admitted": 0, "admitted": 1, "swapped": 724, "uncertain_staged": 262065, "t": 2368.6}
{"event": "train", "step": 2500, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.48877, "bpb": 2.14784, "chunk_loss": 0.030313, "F": 0.19775, "G": 0.22553, "chunks": 202.5, "bytes_per_chunk": 5.057, "boundary_entropy": 0.13259, "router_noise": 0.0, "grad_norm": 0.7354, "lr": 0.0002123, "router_trainable": true, "sec_per_step": 1.136, "probe_sec": 0.3277, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.29425, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": -0.07647, "n_shadow": 3, "mean_delta_shadow": 0.00948, "n_split": 0, "mean_delta_split": NaN, "n_merge": 1, "mean_delta_merge": 0.10498, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 0.19853, "trunc_abs_frac": 5.38107, "mean_delta_full": -0.05173, "read_hit_rate": {"surface": 0.43797, "split": 0.5315, "merge": 0.15446, "skel": 0.32868}, "mean_gate": {"surface": 0.20236, "split": 0.17797, "merge": 0.22002, "skel": 0.2264}, "n_reads": 12061, "n_cand": 30430, "read_suppressed_frac": 0.0963, "dropout_reads_logged": 12061, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1161, "admitted_surface": 20472, "staged_surface": 50299, "admitted_split": 31296, "staged_split": 37541, "admitted_merge": 3611, "staged_merge": 133569, "admitted_skel": 10157, "staged_skel": 40735, "mean_count": 304.1568603515625, "n_util_obs": 205, "util_count_spearman": 0.20006546781630008}, "probe_overhead_sec_total": 326.963, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.008152, "alpha": 0.03, "t": 2369.9}
{"event": "consolidate", "step": 2550, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 301, "uncertain_staged": 262065, "t": 2416.0}
{"event": "train", "step": 2550, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.47884, "bpb": 2.13351, "chunk_loss": 0.030258, "F": 0.19153, "G": 0.22527, "chunks": 196.12, "bytes_per_chunk": 5.221, "boundary_entropy": 0.14996, "router_noise": 0.0, "grad_norm": 0.7047, "lr": 0.0002088, "router_trainable": true, "sec_per_step": 0.94, "probe_sec": 0.2014, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.17407, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 4, "mean_delta_loo": -0.03903, "n_shadow": 4, "mean_delta_shadow": 0.14552, "n_split": 1, "mean_delta_split": -0.02936, "n_merge": 2, "mean_delta_merge": 0.26234, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": -0.05109, "trunc_abs_frac": 0.649, "mean_delta_full": -0.05723, "read_hit_rate": {"surface": 0.414, "split": 0.51138, "merge": 0.13265, "skel": 0.30608}, "mean_gate": {"surface": 0.20071, "split": 0.17659, "merge": 0.22144, "skel": 0.22773}, "n_reads": 11041, "n_cand": 29433, "read_suppressed_frac": 0.0976, "dropout_reads_logged": 11041, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1201, "admitted_surface": 20492, "staged_surface": 50481, "admitted_split": 31569, "staged_split": 37392, "admitted_merge": 3603, "staged_merge": 133529, "admitted_skel": 9872, "staged_skel": 40742, "mean_count": 314.3419189453125, "n_util_obs": 216, "util_count_spearman": 0.18803272246633088}, "probe_overhead_sec_total": 337.62, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.009335, "alpha": 0.03, "t": 2416.9}
{"event": "consolidate", "step": 2600, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 258, "uncertain_staged": 262066, "t": 2461.5}
{"event": "probe", "step": 2600, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3391.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.194851, "cps": 26.492188, "mean_p": 0.203242, "boundary_entropy": 0.088413, "bpb": 2.02003}, "tr": {"chunks": 3264.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.180302, "cps": 25.5, "mean_p": 0.232509, "boundary_entropy": 0.247378, "bpb": 2.233167}, "fi": {"chunks": 3124.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.166073, "cps": 24.40625, "mean_p": 0.281503, "boundary_entropy": 0.370251, "bpb": 2.406727}}, "bpb": {"en": 2.02003, "tr": 2.23317, "fi": 2.40673}, "gini_cpc": 0.035448, "gini_cps": 0.018202, "bpic": 5.801411, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9253284797322727, "fi": 0.8523058556015016}, "inventory": {"size": 327422, "admitted": 65536, "staged": 261886, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1234, "admitted_surface": 20503, "staged_surface": 50670, "admitted_split": 31809, "staged_split": 37996, "admitted_merge": 3600, "staged_merge": 132280, "admitted_skel": 9624, "staged_skel": 40940, "mean_count": 323.52325439453125, "n_util_obs": 225, "util_count_spearman": 0.16411719451043408}, "t": 2463.3}
{"event": "eval", "step": 2600, "phase": 2, "bpb": {"en": 2.03345, "tr": 2.23911, "fi": 2.41739}, "bytes_per_chunk": {"en": 4.86, "tr": 5.85, "fi": 6.289}, "bpb_mean": 2.22998, "frozen_inventory": true, "t": 2466.7}
{"event": "train", "step": 2600, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.50441, "bpb": 2.1704, "chunk_loss": 0.030254, "F": 0.19067, "G": 0.22803, "chunks": 195.25, "bytes_per_chunk": 5.245, "boundary_entropy": 0.15746, "router_noise": 0.0, "grad_norm": 0.7465, "lr": 0.0002053, "router_trainable": true, "sec_per_step": 1.088, "probe_sec": 0.2149, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.1922, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 3, "mean_delta_loo": 0.0568, "n_shadow": 5, "mean_delta_shadow": -0.10968, "n_split": 3, "mean_delta_split": -0.29823, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 1, "trunc_ratio_sum": 0.79376, "trunc_abs_frac": 0.79376, "mean_delta_full": -0.3208, "read_hit_rate": {"surface": 0.44048, "split": 0.49592, "merge": 0.13244, "skel": 0.3018}, "mean_gate": {"surface": 0.19584, "split": 0.17337, "merge": 0.21672, "skel": 0.2228}, "n_reads": 10760, "n_cand": 28857, "read_suppressed_frac": 0.1028, "dropout_reads_logged": 10760, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1232, "admitted_surface": 20503, "staged_surface": 50686, "admitted_split": 31809, "staged_split": 38096, "admitted_merge": 3600, "staged_merge": 132401, "admitted_skel": 9624, "staged_skel": 40961, "mean_count": 323.68743896484375, "n_util_obs": 225, "util_count_spearman": 0.16415086388537717}, "probe_overhead_sec_total": 347.701, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.008854, "alpha": 0.03, "t": 2471.3}
{"event": "consolidate", "step": 2650, "retired": 0, "split_retired": 1, "merge_admitted": 0, "admitted": 1, "swapped": 193, "uncertain_staged": 262067, "t": 2515.1}
{"event": "train", "step": 2650, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.46418, "bpb": 2.11236, "chunk_loss": 0.030229, "F": 0.18872, "G": 0.22566, "chunks": 193.25, "bytes_per_chunk": 5.299, "boundary_entropy": 0.15438, "router_noise": 0.0, "grad_norm": 0.7427, "lr": 0.0002017, "router_trainable": true, "sec_per_step": 0.895, "probe_sec": 0.1951, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.17209, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 4, "mean_delta_loo": 0.04008, "n_shadow": 7, "mean_delta_shadow": -0.02934, "n_split": 0, "mean_delta_split": NaN, "n_merge": 2, "mean_delta_merge": -0.12819, "n_dropped_unpaired": 0, "n_full": 0, "read_hit_rate": {"surface": 0.42848, "split": 0.5201, "merge": 0.13987, "skel": 0.32049}, "mean_gate": {"surface": 0.20172, "split": 0.17912, "merge": 0.22108, "skel": 0.22841}, "n_reads": 11126, "n_cand": 28872, "read_suppressed_frac": 0.0992, "dropout_reads_logged": 11126, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1267, "admitted_surface": 20528, "staged_surface": 50678, "admitted_split": 31971, "staged_split": 38000, "admitted_merge": 3604, "staged_merge": 132498, "admitted_skel": 9433, "staged_skel": 40968, "mean_count": 330.4288635253906, "n_util_obs": 244, "util_count_spearman": 0.11663259110565902}, "probe_overhead_sec_total": 357.591, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.010952, "alpha": 0.03, "t": 2516.0}
{"event": "consolidate", "step": 2700, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 169, "uncertain_staged": 262068, "t": 2565.9}
{"event": "train", "step": 2700, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.54226, "bpb": 2.22502, "chunk_loss": 0.030266, "F": 0.19196, "G": 0.22788, "chunks": 196.56, "bytes_per_chunk": 5.21, "boundary_entropy": 0.14943, "router_noise": 0.0, "grad_norm": 0.8355, "lr": 0.0001981, "router_trainable": true, "sec_per_step": 1.019, "probe_sec": 0.2864, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.26456, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 5, "mean_delta_loo": -0.10826, "n_shadow": 3, "mean_delta_shadow": -0.08484, "n_split": 3, "mean_delta_split": 0.13212, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 1.48103, "trunc_abs_frac": 1.84407, "mean_delta_full": -0.07292, "read_hit_rate": {"surface": 0.42362, "split": 0.50221, "merge": 0.12464, "skel": 0.30275}, "mean_gate": {"surface": 0.20553, "split": 0.18059, "merge": 0.22596, "skel": 0.2311}, "n_reads": 10846, "n_cand": 29248, "read_suppressed_frac": 0.0968, "dropout_reads_logged": 10846, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1295, "admitted_surface": 20535, "staged_surface": 51170, "admitted_split": 32124, "staged_split": 38295, "admitted_merge": 3611, "staged_merge": 131643, "admitted_skel": 9266, "staged_skel": 41036, "mean_count": 339.5697937011719, "n_util_obs": 250, "util_count_spearman": 0.09729834345477902}, "probe_overhead_sec_total": 367.922, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.009008, "alpha": 0.03, "t": 2566.9}
{"event": "consolidate", "step": 2750, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 471, "uncertain_staged": 262066, "t": 2623.6}
{"event": "train", "step": 2750, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.47657, "bpb": 2.13024, "chunk_loss": 0.030345, "F": 0.19647, "G": 0.22377, "chunks": 201.19, "bytes_per_chunk": 5.09, "boundary_entropy": 0.13383, "router_noise": 0.0, "grad_norm": 0.7464, "lr": 0.0001945, "router_trainable": true, "sec_per_step": 1.158, "probe_sec": 0.237, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.21275, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": 0.04463, "n_shadow": 6, "mean_delta_shadow": -0.00692, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 0.29377, "trunc_abs_frac": 0.62074, "mean_delta_full": -0.07321, "read_hit_rate": {"surface": 0.44599, "split": 0.52686, "merge": 0.12959, "skel": 0.31719}, "mean_gate": {"surface": 0.20321, "split": 0.17984, "merge": 0.22069, "skel": 0.23265}, "n_reads": 11605, "n_cand": 29830, "read_suppressed_frac": 0.0999, "dropout_reads_logged": 11605, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1327, "admitted_surface": 20537, "staged_surface": 50422, "admitted_split": 32582, "staged_split": 37172, "admitted_merge": 3616, "staged_merge": 133950, "admitted_skel": 8801, "staged_skel": 40600, "mean_count": 348.5242919921875, "n_util_obs": 255, "util_count_spearman": 0.11290745136637333}, "probe_overhead_sec_total": 377.746, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.008941, "alpha": 0.03, "t": 2624.8}
{"event": "consolidate", "step": 2800, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 602, "uncertain_staged": 262067, "t": 2677.4}
{"event": "probe", "step": 2800, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3410.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.195943, "cps": 26.640625, "mean_p": 0.203112, "boundary_entropy": 0.086421, "bpb": 1.999388}, "tr": {"chunks": 3194.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.176435, "cps": 24.953125, "mean_p": 0.227139, "boundary_entropy": 0.241125, "bpb": 2.209442}, "fi": {"chunks": 3027.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.160916, "cps": 23.648438, "mean_p": 0.264591, "boundary_entropy": 0.345662, "bpb": 2.369694}}, "bpb": {"en": 1.99939, "tr": 2.20944, "fi": 2.36969}, "gini_cpc": 0.043787, "gini_cps": 0.026512, "bpic": 5.890562, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9004385948570927, "fi": 0.8212403487009458}, "inventory": {"size": 327078, "admitted": 65536, "staged": 261542, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1345, "admitted_surface": 20494, "staged_surface": 50778, "admitted_split": 33156, "staged_split": 37346, "admitted_merge": 3603, "staged_merge": 132678, "admitted_skel": 8283, "staged_skel": 40740, "mean_count": 357.3883972167969, "n_util_obs": 261, "util_count_spearman": 0.11652391655023926}, "t": 2679.3}
{"event": "eval", "step": 2800, "phase": 2, "bpb": {"en": 2.01439, "tr": 2.21592, "fi": 2.37984}, "bytes_per_chunk": {"en": 4.854, "tr": 5.898, "fi": 6.437}, "bpb_mean": 2.20338, "frozen_inventory": true, "t": 2682.6}
{"event": "train", "step": 2800, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.45087, "bpb": 2.09316, "chunk_loss": 0.030308, "F": 0.19659, "G": 0.22367, "chunks": 201.31, "bytes_per_chunk": 5.087, "boundary_entropy": 0.12849, "router_noise": 0.0, "grad_norm": 0.6526, "lr": 0.0001909, "router_trainable": true, "sec_per_step": 1.254, "probe_sec": 0.1995, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.17686, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": -0.09651, "n_shadow": 2, "mean_delta_shadow": -0.02829, "n_split": 0, "mean_delta_split": NaN, "n_merge": 2, "mean_delta_merge": 0.05807, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 0.35244, "trunc_abs_frac": 1.01404, "mean_delta_full": 0.08862, "read_hit_rate": {"surface": 0.44914, "split": 0.52545, "merge": 0.15067, "skel": 0.33159}, "mean_gate": {"surface": 0.20766, "split": 0.18213, "merge": 0.21975, "skel": 0.23366}, "n_reads": 11974, "n_cand": 30244, "read_suppressed_frac": 0.0983, "dropout_reads_logged": 11974, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1345, "admitted_surface": 20494, "staged_surface": 50911, "admitted_split": 33156, "staged_split": 37456, "admitted_merge": 3603, "staged_merge": 132937, "admitted_skel": 8283, "staged_skel": 40840, "mean_count": 357.57110595703125, "n_util_obs": 261, "util_count_spearman": 0.11622964207854969}, "probe_overhead_sec_total": 388.063, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.010196, "alpha": 0.03, "t": 2687.5}
{"event": "consolidate", "step": 2850, "retired": 0, "split_retired": 1, "merge_admitted": 0, "admitted": 1, "swapped": 1024, "uncertain_staged": 262067, "t": 2736.1}
{"event": "train", "step": 2850, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.53681, "bpb": 2.21714, "chunk_loss": 0.03039, "F": 0.19894, "G": 0.22505, "chunks": 203.72, "bytes_per_chunk": 5.027, "boundary_entropy": 0.13246, "router_noise": 0.0, "grad_norm": 0.7366, "lr": 0.0001872, "router_trainable": true, "sec_per_step": 0.991, "probe_sec": 0.2364, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.20166, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 5, "mean_delta_loo": -0.11563, "n_shadow": 1, "mean_delta_shadow": 0.15149, "n_split": 2, "mean_delta_split": 0.09929, "n_merge": 2, "mean_delta_merge": -0.14588, "n_dropped_unpaired": 0, "n_full": 0, "read_hit_rate": {"surface": 0.41776, "split": 0.5133, "merge": 0.13648, "skel": 0.31625}, "mean_gate": {"surface": 0.20302, "split": 0.17991, "merge": 0.22143, "skel": 0.23387}, "n_reads": 11535, "n_cand": 30471, "read_suppressed_frac": 0.0987, "dropout_reads_logged": 11535, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1366, "admitted_surface": 19990, "staged_surface": 51117, "admitted_split": 34037, "staged_split": 36977, "admitted_merge": 3592, "staged_merge": 133059, "admitted_skel": 7917, "staged_skel": 40991, "mean_count": 363.9468078613281, "n_util_obs": 255, "util_count_spearman": 0.0764727593793423}, "probe_overhead_sec_total": 398.151, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.008587, "alpha": 0.03, "t": 2737.1}
{"event": "consolidate", "step": 2900, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262068, "t": 2773.0}
{"event": "train", "step": 2900, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.41589, "bpb": 2.04269, "chunk_loss": 0.030308, "F": 0.19455, "G": 0.22403, "chunks": 199.22, "bytes_per_chunk": 5.14, "boundary_entropy": 0.14174, "router_noise": 0.0, "grad_norm": 0.7254, "lr": 0.0001835, "router_trainable": true, "sec_per_step": 0.752, "probe_sec": 0.0935, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.05299, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": 0.0309, "n_shadow": 4, "mean_delta_shadow": 0.07915, "n_split": 0, "mean_delta_split": NaN, "n_merge": 2, "mean_delta_merge": -0.1037, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 1.35297, "trunc_abs_frac": 1.4849, "mean_delta_full": 0.12329, "read_hit_rate": {"surface": 0.44254, "split": 0.53376, "merge": 0.13976, "skel": 0.33327}, "mean_gate": {"surface": 0.20189, "split": 0.1778, "merge": 0.21812, "skel": 0.23236}, "n_reads": 11799, "n_cand": 29789, "read_suppressed_frac": 0.0999, "dropout_reads_logged": 11799, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1397, "admitted_surface": 19369, "staged_surface": 50834, "admitted_split": 34990, "staged_split": 36808, "admitted_merge": 3646, "staged_merge": 133546, "admitted_skel": 7531, "staged_skel": 40956, "mean_count": 373.40313720703125, "n_util_obs": 264, "util_count_spearman": 0.05150428164274208}, "probe_overhead_sec_total": 404.516, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.010209, "alpha": 0.03, "t": 2774.7}
{"event": "consolidate", "step": 2950, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262068, "t": 2822.2}
{"event": "train", "step": 2950, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.35563, "bpb": 1.95575, "chunk_loss": 0.030483, "F": 0.207, "G": 0.22196, "chunks": 211.97, "bytes_per_chunk": 4.831, "boundary_entropy": 0.09776, "router_noise": 0.0, "grad_norm": 0.6493, "lr": 0.0001799, "router_trainable": true, "sec_per_step": 0.961, "probe_sec": 0.0857, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.05289, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": -0.07184, "n_shadow": 2, "mean_delta_shadow": 0.02435, "n_split": 1, "mean_delta_split": 0.33856, "n_merge": 1, "mean_delta_merge": 0.23828, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": -0.08929, "trunc_abs_frac": 1.18386, "mean_delta_full": 0.00342, "read_hit_rate": {"surface": 0.50422, "split": 0.54131, "merge": 0.17666, "skel": 0.3596}, "mean_gate": {"surface": 0.21152, "split": 0.18259, "merge": 0.22107, "skel": 0.23766}, "n_reads": 13588, "n_cand": 31982, "read_suppressed_frac": 0.1024, "dropout_reads_logged": 13588, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1432, "admitted_surface": 19296, "staged_surface": 50680, "admitted_split": 35831, "staged_split": 36930, "admitted_merge": 3732, "staged_merge": 133719, "admitted_skel": 6677, "staged_skel": 40815, "mean_count": 381.50152587890625, "n_util_obs": 270, "util_count_spearman": 0.08783507165926678}, "probe_overhead_sec_total": 410.24, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.008364, "alpha": 0.03, "t": 2822.7}
{"event": "consolidate", "step": 3000, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262069, "t": 2868.1}
{"event": "probe", "step": 3000, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3427.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.19692, "cps": 26.773438, "mean_p": 0.204132, "boundary_entropy": 0.084229, "bpb": 1.985764}, "tr": {"chunks": 3138.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.173341, "cps": 24.515625, "mean_p": 0.226054, "boundary_entropy": 0.235775, "bpb": 2.192504}, "fi": {"chunks": 2965.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.157621, "cps": 23.164062, "mean_p": 0.264002, "boundary_entropy": 0.347333, "bpb": 2.342863}}, "bpb": {"en": 1.98576, "tr": 2.1925, "fi": 2.34286}, "gini_cpc": 0.049632, "gini_cps": 0.032319, "bpic": 5.952991, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.8802629108302085, "fi": 0.8004290277294465}, "inventory": {"size": 326656, "admitted": 65536, "staged": 261120, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1446, "admitted_surface": 18616, "staged_surface": 51019, "admitted_split": 36690, "staged_split": 36890, "admitted_merge": 3891, "staged_merge": 132496, "admitted_skel": 6339, "staged_skel": 40715, "mean_count": 390.01409912109375, "n_util_obs": 274, "util_count_spearman": 0.10601281929749083}, "t": 2869.8}
{"event": "eval", "step": 3000, "phase": 2, "bpb": {"en": 1.99985, "tr": 2.19912, "fi": 2.35266}, "bytes_per_chunk": {"en": 4.839, "tr": 5.972, "fi": 6.43}, "bpb_mean": 2.18388, "frozen_inventory": true, "t": 2872.5}
{"event": "train", "step": 3000, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.45048, "bpb": 2.09259, "chunk_loss": 0.030319, "F": 0.19421, "G": 0.22511, "chunks": 198.88, "bytes_per_chunk": 5.149, "boundary_entropy": 0.14267, "router_noise": 0.0, "grad_norm": 0.7962, "lr": 0.0001761, "router_trainable": true, "sec_per_step": 1.1, "probe_sec": 0.1937, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.16843, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": -0.06362, "n_shadow": 3, "mean_delta_shadow": 0.12484, "n_split": 1, "mean_delta_split": 0.02737, "n_merge": 1, "mean_delta_merge": -0.23517, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": -0.71897, "trunc_abs_frac": 0.78871, "mean_delta_full": 0.08024, "read_hit_rate": {"surface": 0.45246, "split": 0.49721, "merge": 0.14905, "skel": 0.30275}, "mean_gate": {"surface": 0.20479, "split": 0.17729, "merge": 0.21728, "skel": 0.23074}, "n_reads": 11231, "n_cand": 29573, "read_suppressed_frac": 0.1008, "dropout_reads_logged": 11231, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1448, "admitted_surface": 18616, "staged_surface": 51220, "admitted_split": 36690, "staged_split": 37143, "admitted_merge": 3891, "staged_merge": 132897, "admitted_skel": 6339, "staged_skel": 40884, "mean_count": 390.1854553222656, "n_util_obs": 274, "util_count_spearman": 0.10506952901113485}, "probe_overhead_sec_total": 416.02, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.009599, "alpha": 0.03, "t": 2877.7}
{"event": "consolidate", "step": 3050, "retired": 0, "split_retired": 1, "merge_admitted": 0, "admitted": 1, "swapped": 1024, "uncertain_staged": 262069, "t": 2916.9}
{"event": "train", "step": 3050, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.40952, "bpb": 2.03351, "chunk_loss": 0.030341, "F": 0.19879, "G": 0.22193, "chunks": 203.56, "bytes_per_chunk": 5.03, "boundary_entropy": 0.11675, "router_noise": 0.0, "grad_norm": 0.7075, "lr": 0.0001724, "router_trainable": true, "sec_per_step": 0.812, "probe_sec": 0.0916, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.049, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 10, "mean_delta_loo": -0.07517, "n_shadow": 2, "mean_delta_shadow": 0.19104, "n_split": 0, "mean_delta_split": NaN, "n_merge": 1, "mean_delta_merge": -0.11719, "n_dropped_unpaired": 0, "n_full": 6, "trunc_ratio_sum": 10.09535, "trunc_abs_frac": 0.8475, "mean_delta_full": -0.00875, "read_hit_rate": {"surface": 0.46174, "split": 0.52517, "merge": 0.14806, "skel": 0.34111}, "mean_gate": {"surface": 0.21453, "split": 0.1833, "merge": 0.22555, "skel": 0.2441}, "n_reads": 12295, "n_cand": 30734, "read_suppressed_frac": 0.1008, "dropout_reads_logged": 12295, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1464, "admitted_surface": 17611, "staged_surface": 50952, "admitted_split": 37543, "staged_split": 37365, "admitted_merge": 4051, "staged_merge": 132898, "admitted_skel": 6331, "staged_skel": 40929, "mean_count": 397.94122314453125, "n_util_obs": 279, "util_count_spearman": 0.09141608372288344}, "probe_overhead_sec_total": 422.24, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.007791, "alpha": 0.03, "t": 2918.3}
{"event": "consolidate", "step": 3100, "retired": 0, "split_retired": 1, "merge_admitted": 0, "admitted": 1, "swapped": 1024, "uncertain_staged": 262069, "t": 2959.5}
{"event": "train", "step": 3100, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.50347, "bpb": 2.16905, "chunk_loss": 0.030311, "F": 0.19397, "G": 0.21834, "chunks": 198.62, "bytes_per_chunk": 5.155, "boundary_entropy": 0.12383, "router_noise": 0.0, "grad_norm": 0.7781, "lr": 0.0001687, "router_trainable": true, "sec_per_step": 0.834, "probe_sec": 0.1423, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.11876, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 4, "mean_delta_loo": -0.06956, "n_shadow": 8, "mean_delta_shadow": -0.08605, "n_split": 1, "mean_delta_split": -0.28989, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": 0.42763, "trunc_abs_frac": 0.6263, "mean_delta_full": -0.07742, "read_hit_rate": {"surface": 0.43754, "split": 0.52302, "merge": 0.13366, "skel": 0.30686}, "mean_gate": {"surface": 0.21, "split": 0.18113, "merge": 0.2201, "skel": 0.23863}, "n_reads": 11489, "n_cand": 29823, "read_suppressed_frac": 0.099, "dropout_reads_logged": 11489, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1488, "admitted_surface": 16589, "staged_surface": 50867, "admitted_split": 38444, "staged_split": 36964, "admitted_merge": 4171, "staged_merge": 133628, "admitted_skel": 6332, "staged_skel": 40685, "mean_count": 406.17694091796875, "n_util_obs": 285, "util_count_spearman": 0.08157610682558333}, "probe_overhead_sec_total": 427.51, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.009297, "alpha": 0.03, "t": 2960.0}
{"event": "consolidate", "step": 3150, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262068, "t": 3001.7}
{"event": "train", "step": 3150, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.40995, "bpb": 2.03413, "chunk_loss": 0.030258, "F": 0.19385, "G": 0.21814, "chunks": 198.5, "bytes_per_chunk": 5.159, "boundary_entropy": 0.1184, "router_noise": 0.0, "grad_norm": 0.6906, "lr": 0.000165, "router_trainable": true, "sec_per_step": 0.855, "probe_sec": 0.1049, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.07759, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 5, "mean_delta_loo": 0.0551, "n_shadow": 6, "mean_delta_shadow": 0.07294, "n_split": 1, "mean_delta_split": 0.36566, "n_merge": 1, "mean_delta_merge": -0.09265, "n_dropped_unpaired": 0, "n_full": 1, "trunc_ratio_sum": 1.0, "trunc_abs_frac": 1.0, "mean_delta_full": -0.09265, "read_hit_rate": {"surface": 0.43987, "split": 0.5189, "merge": 0.14265, "skel": 0.32976}, "mean_gate": {"surface": 0.21501, "split": 0.18273, "merge": 0.22487, "skel": 0.24463}, "n_reads": 11663, "n_cand": 29936, "read_suppressed_frac": 0.0991, "dropout_reads_logged": 11663, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1527, "admitted_surface": 15568, "staged_surface": 50765, "admitted_split": 39314, "staged_split": 36614, "admitted_merge": 4323, "staged_merge": 134118, "admitted_skel": 6331, "staged_skel": 40647, "mean_count": 415.113037109375, "n_util_obs": 292, "util_count_spearman": 0.0875119834696802}, "probe_overhead_sec_total": 433.861, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.009927, "alpha": 0.03, "t": 3002.8}
{"event": "consolidate", "step": 3200, "retired": 1, "split_retired": 0, "merge_admitted": 0, "admitted": 1, "swapped": 1024, "uncertain_staged": 262069, "t": 3040.6}
{"event": "probe", "step": 3200, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3408.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.195828, "cps": 26.625, "mean_p": 0.202056, "boundary_entropy": 0.077347, "bpb": 1.961869}, "tr": {"chunks": 3241.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.179031, "cps": 25.320312, "mean_p": 0.227457, "boundary_entropy": 0.233536, "bpb": 2.157416}, "fi": {"chunks": 2995.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.159215, "cps": 23.398438, "mean_p": 0.264153, "boundary_entropy": 0.342553, "bpb": 2.319058}}, "bpb": {"en": 1.96187, "tr": 2.15742, "fi": 2.31906}, "gini_cpc": 0.045703, "gini_cps": 0.02855, "bpic": 5.882621, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.914224832783921, "fi": 0.8130354411301148}, "inventory": {"size": 326655, "admitted": 65536, "staged": 261119, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1542, "admitted_surface": 14555, "staged_surface": 50883, "admitted_split": 40126, "staged_split": 36625, "admitted_merge": 4536, "staged_merge": 132707, "admitted_skel": 6319, "staged_skel": 40904, "mean_count": 423.37371826171875, "n_util_obs": 294, "util_count_spearman": 0.06849644813862159}, "t": 3041.9}
{"event": "eval", "step": 3200, "phase": 2, "bpb": {"en": 1.97776, "tr": 2.16662, "fi": 2.33045}, "bytes_per_chunk": {"en": 4.851, "tr": 5.82, "fi": 6.403}, "bpb_mean": 2.15828, "frozen_inventory": true, "t": 3044.3}
{"event": "train", "step": 3200, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.3374, "bpb": 1.92946, "chunk_loss": 0.030318, "F": 0.19919, "G": 0.21869, "chunks": 203.97, "bytes_per_chunk": 5.02, "boundary_entropy": 0.10456, "router_noise": 0.0, "grad_norm": 0.7013, "lr": 0.0001613, "router_trainable": true, "sec_per_step": 0.91, "probe_sec": 0.1639, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.13688, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 4, "mean_delta_loo": 0.04971, "n_shadow": 3, "mean_delta_shadow": -0.15142, "n_split": 2, "mean_delta_split": 0.43881, "n_merge": 2, "mean_delta_merge": 0.19466, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 0.39493, "trunc_abs_frac": 1.03005, "mean_delta_full": 0.21265, "read_hit_rate": {"surface": 0.48022, "split": 0.54097, "merge": 0.15999, "skel": 0.3436}, "mean_gate": {"surface": 0.21419, "split": 0.18391, "merge": 0.22103, "skel": 0.244}, "n_reads": 12756, "n_cand": 30887, "read_suppressed_frac": 0.101, "dropout_reads_logged": 12756, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1541, "admitted_surface": 14555, "staged_surface": 50906, "admitted_split": 40126, "staged_split": 36622, "admitted_merge": 4536, "staged_merge": 133696, "admitted_skel": 6319, "staged_skel": 40920, "mean_count": 423.568359375, "n_util_obs": 294, "util_count_spearman": 0.06878025232836948}, "probe_overhead_sec_total": 439.082, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.009812, "alpha": 0.03, "t": 3048.3}
{"event": "consolidate", "step": 3250, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262068, "t": 3096.1}
{"event": "train", "step": 3250, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.34139, "bpb": 1.93521, "chunk_loss": 0.030341, "F": 0.19788, "G": 0.22254, "chunks": 202.62, "bytes_per_chunk": 5.054, "boundary_entropy": 0.12157, "router_noise": 0.0, "grad_norm": 0.6723, "lr": 0.0001576, "router_trainable": true, "sec_per_step": 0.967, "probe_sec": 0.0929, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.05202, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": -0.0915, "n_shadow": 3, "mean_delta_shadow": -0.0104, "n_split": 1, "mean_delta_split": 0.05798, "n_merge": 2, "mean_delta_merge": 0.34311, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 11.23375, "trunc_abs_frac": 0.79899, "mean_delta_full": 0.01628, "read_hit_rate": {"surface": 0.46559, "split": 0.52902, "merge": 0.15389, "skel": 0.3384}, "mean_gate": {"surface": 0.21232, "split": 0.18151, "merge": 0.22039, "skel": 0.24199}, "n_reads": 12287, "n_cand": 30474, "read_suppressed_frac": 0.0977, "dropout_reads_logged": 12287, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1549, "admitted_surface": 13549, "staged_surface": 51289, "admitted_split": 40887, "staged_split": 37248, "admitted_merge": 4792, "staged_merge": 132608, "admitted_skel": 6308, "staged_skel": 40999, "mean_count": 431.8988037109375, "n_util_obs": 301, "util_count_spearman": 0.08410244924231083}, "probe_overhead_sec_total": 444.663, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.008916, "alpha": 0.03, "t": 3096.7}
{"event": "consolidate", "step": 3300, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262067, "t": 3135.5}
{"event": "train", "step": 3300, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.46445, "bpb": 2.11276, "chunk_loss": 0.030351, "F": 0.19867, "G": 0.22042, "chunks": 203.44, "bytes_per_chunk": 5.033, "boundary_entropy": 0.11328, "router_noise": 0.0, "grad_norm": 0.7516, "lr": 0.0001539, "router_trainable": true, "sec_per_step": 0.79, "probe_sec": 0.0903, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.05617, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 11, "mean_delta_loo": -0.00438, "n_shadow": 2, "mean_delta_shadow": 0.01794, "n_split": 1, "mean_delta_split": -0.05046, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": 0.32847, "trunc_abs_frac": 0.46872, "mean_delta_full": 0.0894, "read_hit_rate": {"surface": 0.41386, "split": 0.51739, "merge": 0.13419, "skel": 0.31539}, "mean_gate": {"surface": 0.2123, "split": 0.1793, "merge": 0.22229, "skel": 0.2433}, "n_reads": 11534, "n_cand": 30408, "read_suppressed_frac": 0.096, "dropout_reads_logged": 11534, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1550, "admitted_surface": 12648, "staged_surface": 51474, "admitted_split": 41566, "staged_split": 37034, "admitted_merge": 5073, "staged_merge": 132952, "admitted_skel": 6249, "staged_skel": 40684, "mean_count": 435.2727355957031, "n_util_obs": 291, "util_count_spearman": 0.04483829967518712}, "probe_overhead_sec_total": 450.598, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.008214, "alpha": 0.03, "t": 3136.2}
{"event": "consolidate", "step": 3350, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262066, "t": 3173.4}
{"event": "train", "step": 3350, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.41737, "bpb": 2.04483, "chunk_loss": 0.030297, "F": 0.19501, "G": 0.21935, "chunks": 199.69, "bytes_per_chunk": 5.128, "boundary_entropy": 0.11855, "router_noise": 0.0, "grad_norm": 0.6657, "lr": 0.0001501, "router_trainable": true, "sec_per_step": 0.757, "probe_sec": 0.1563, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.11663, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": 0.0388, "n_shadow": 4, "mean_delta_shadow": -0.01852, "n_split": 0, "mean_delta_split": NaN, "n_merge": 2, "mean_delta_merge": -0.17087, "n_dropped_unpaired": 0, "n_full": 1, "trunc_ratio_sum": 7.94034, "trunc_abs_frac": 7.94034, "mean_delta_full": -0.02148, "read_hit_rate": {"surface": 0.43677, "split": 0.50465, "merge": 0.14448, "skel": 0.33769}, "mean_gate": {"surface": 0.21285, "split": 0.18146, "merge": 0.21871, "skel": 0.24297}, "n_reads": 11545, "n_cand": 29993, "read_suppressed_frac": 0.0982, "dropout_reads_logged": 11545, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1590, "admitted_surface": 12649, "staged_surface": 51922, "admitted_split": 42220, "staged_split": 37287, "admitted_merge": 5409, "staged_merge": 131775, "admitted_skel": 5258, "staged_skel": 41160, "mean_count": 443.092041015625, "n_util_obs": 293, "util_count_spearman": 0.06499503189640629}, "probe_overhead_sec_total": 456.102, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.008651, "alpha": 0.03, "t": 3174.0}
{"event": "consolidate", "step": 3400, "retired": 0, "split_retired": 1, "merge_admitted": 0, "admitted": 1, "swapped": 1024, "uncertain_staged": 262066, "t": 3215.9}
{"event": "probe", "step": 3400, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3395.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.195081, "cps": 26.523438, "mean_p": 0.200279, "boundary_entropy": 0.072244, "bpb": 1.948935}, "tr": {"chunks": 3216.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.17765, "cps": 25.125, "mean_p": 0.217241, "boundary_entropy": 0.21636, "bpb": 2.139297}, "fi": {"chunks": 3175.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.168784, "cps": 24.804688, "mean_p": 0.271925, "boundary_entropy": 0.349489, "bpb": 2.301716}}, "bpb": {"en": 1.94894, "tr": 2.1393, "fi": 2.30172}, "gini_cpc": 0.032375, "gini_cps": 0.014987, "bpic": 5.797261, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9106465156793433, "fi": 0.8651993565322957}, "inventory": {"size": 326655, "admitted": 65536, "staged": 261119, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1593, "admitted_surface": 12107, "staged_surface": 51497, "admitted_split": 42963, "staged_split": 36627, "admitted_merge": 5665, "staged_merge": 132285, "admitted_skel": 4801, "staged_skel": 40710, "mean_count": 450.786376953125, "n_util_obs": 297, "util_count_spearman": 0.05993662819392643}, "t": 3217.2}
{"event": "eval", "step": 3400, "phase": 2, "bpb": {"en": 1.96658, "tr": 2.14835, "fi": 2.30927}, "bytes_per_chunk": {"en": 4.867, "tr": 5.871, "fi": 6.126}, "bpb_mean": 2.1414, "frozen_inventory": true, "t": 3219.7}
{"event": "train", "step": 3400, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.37783, "bpb": 1.98778, "chunk_loss": 0.030333, "F": 0.19919, "G": 0.21823, "chunks": 203.97, "bytes_per_chunk": 5.02, "boundary_entropy": 0.10502, "router_noise": 0.0, "grad_norm": 0.6947, "lr": 0.0001465, "router_trainable": true, "sec_per_step": 1.019, "probe_sec": 0.0759, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.05462, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": -0.02989, "n_shadow": 5, "mean_delta_shadow": -0.01077, "n_split": 1, "mean_delta_split": -0.07391, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": -0.55922, "trunc_abs_frac": 0.63805, "mean_delta_full": -0.05442, "read_hit_rate": {"surface": 0.43603, "split": 0.51992, "merge": 0.15566, "skel": 0.32544}, "mean_gate": {"surface": 0.21761, "split": 0.18372, "merge": 0.22118, "skel": 0.24704}, "n_reads": 11926, "n_cand": 30521, "read_suppressed_frac": 0.1018, "dropout_reads_logged": 11926, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1592, "admitted_surface": 12107, "staged_surface": 51600, "admitted_split": 42963, "staged_split": 36717, "admitted_merge": 5665, "staged_merge": 133009, "admitted_skel": 4801, "staged_skel": 40818, "mean_count": 450.96832275390625, "n_util_obs": 299, "util_count_spearman": 0.05523602783731352}, "probe_overhead_sec_total": 461.319, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.010285, "alpha": 0.03, "t": 3225.0}
{"event": "consolidate", "step": 3450, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262066, "t": 3263.4}
{"event": "train", "step": 3450, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.53329, "bpb": 2.21207, "chunk_loss": 0.03037, "F": 0.19748, "G": 0.22308, "chunks": 202.22, "bytes_per_chunk": 5.064, "boundary_entropy": 0.12343, "router_noise": 0.0, "grad_norm": 0.7379, "lr": 0.0001428, "router_trainable": true, "sec_per_step": 0.798, "probe_sec": 0.1964, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.09902, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 10, "mean_delta_loo": 0.10129, "n_shadow": 4, "mean_delta_shadow": -0.03316, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": -1.44986, "trunc_abs_frac": 2.24043, "mean_delta_full": -0.02922, "read_hit_rate": {"surface": 0.39137, "split": 0.49328, "merge": 0.1336, "skel": 0.29885}, "mean_gate": {"surface": 0.21275, "split": 0.18015, "merge": 0.21817, "skel": 0.24066}, "n_reads": 10782, "n_cand": 29879, "read_suppressed_frac": 0.1009, "dropout_reads_logged": 10782, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1610, "admitted_surface": 11566, "staged_surface": 51624, "admitted_split": 43438, "staged_split": 36895, "admitted_merge": 6013, "staged_merge": 132844, "admitted_skel": 4519, "staged_skel": 40781, "mean_count": 458.49652099609375, "n_util_obs": 303, "util_count_spearman": 0.043221070406629154}, "probe_overhead_sec_total": 466.59, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.008152, "alpha": 0.03, "t": 3264.9}
{"event": "consolidate", "step": 3500, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262066, "t": 3301.4}
{"event": "train", "step": 3500, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.40551, "bpb": 2.02773, "chunk_loss": 0.030311, "F": 0.1962, "G": 0.21931, "chunks": 200.91, "bytes_per_chunk": 5.097, "boundary_entropy": 0.11591, "router_noise": 0.0, "grad_norm": 0.7415, "lr": 0.0001391, "router_trainable": true, "sec_per_step": 0.74, "probe_sec": 0.0657, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04579, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": 0.06646, "n_shadow": 2, "mean_delta_shadow": -0.09265, "n_split": 1, "mean_delta_split": -0.01517, "n_merge": 2, "mean_delta_merge": -0.10089, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": -1.40426, "trunc_abs_frac": 0.64288, "mean_delta_full": -0.09824, "read_hit_rate": {"surface": 0.41348, "split": 0.51103, "merge": 0.1326, "skel": 0.31741}, "mean_gate": {"surface": 0.21742, "split": 0.18194, "merge": 0.21918, "skel": 0.24667}, "n_reads": 11350, "n_cand": 30175, "read_suppressed_frac": 0.1001, "dropout_reads_logged": 11350, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1618, "admitted_surface": 10673, "staged_surface": 51204, "admitted_split": 44233, "staged_split": 36202, "admitted_merge": 6131, "staged_merge": 134345, "admitted_skel": 4499, "staged_skel": 40393, "mean_count": 464.52667236328125, "n_util_obs": 295, "util_count_spearman": 0.03289051763628035}, "probe_overhead_sec_total": 472.124, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.009052, "alpha": 0.03, "t": 3301.8}
{"event": "consolidate", "step": 3550, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262066, "t": 3337.7}
{"event": "train", "step": 3550, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.40022, "bpb": 2.02009, "chunk_loss": 0.030269, "F": 0.19534, "G": 0.21823, "chunks": 200.03, "bytes_per_chunk": 5.119, "boundary_entropy": 0.11446, "router_noise": 0.0, "grad_norm": 0.7104, "lr": 0.0001355, "router_trainable": true, "sec_per_step": 0.73, "probe_sec": 0.1138, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.07983, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 2, "mean_delta_loo": -0.01963, "n_shadow": 8, "mean_delta_shadow": 0.08071, "n_split": 1, "mean_delta_split": -0.29968, "n_merge": 1, "mean_delta_merge": -0.19528, "n_dropped_unpaired": 0, "n_full": 1, "trunc_ratio_sum": -6.32394, "trunc_abs_frac": 6.32394, "mean_delta_full": 0.00867, "read_hit_rate": {"surface": 0.4043, "split": 0.49426, "merge": 0.14518, "skel": 0.31002}, "mean_gate": {"surface": 0.21183, "split": 0.17754, "merge": 0.21671, "skel": 0.23933}, "n_reads": 11086, "n_cand": 30052, "read_suppressed_frac": 0.1035, "dropout_reads_logged": 11086, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1586, "admitted_surface": 10176, "staged_surface": 51439, "admitted_split": 44573, "staged_split": 36244, "admitted_merge": 6343, "staged_merge": 134367, "admitted_skel": 4444, "staged_skel": 40094, "mean_count": 465.5281677246094, "n_util_obs": 282, "util_count_spearman": -0.02958357290450695}, "probe_overhead_sec_total": 477.146, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.009174, "alpha": 0.03, "t": 3338.4}
{"event": "consolidate", "step": 3600, "retired": 0, "split_retired": 1, "merge_admitted": 0, "admitted": 1, "swapped": 1024, "uncertain_staged": 262065, "t": 3387.2}
{"event": "probe", "step": 3600, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3395.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.195081, "cps": 26.523438, "mean_p": 0.201437, "boundary_entropy": 0.074315, "bpb": 1.933111}, "tr": {"chunks": 3196.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.176545, "cps": 24.96875, "mean_p": 0.2208, "boundary_entropy": 0.218348, "bpb": 2.130638}, "fi": {"chunks": 3109.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.165276, "cps": 24.289062, "mean_p": 0.268832, "boundary_entropy": 0.348686, "bpb": 2.277547}}, "bpb": {"en": 1.93311, "tr": 2.13064, "fi": 2.27755}, "gini_cpc": 0.037009, "gini_cps": 0.019656, "bpic": 5.84866, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9049832910793474, "fi": 0.8472141100657976}, "inventory": {"size": 326655, "admitted": 65536, "staged": 261119, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1560, "admitted_surface": 9562, "staged_surface": 51891, "admitted_split": 45027, "staged_split": 36339, "admitted_merge": 6539, "staged_merge": 132564, "admitted_skel": 4408, "staged_skel": 40325, "mean_count": 468.542724609375, "n_util_obs": 277, "util_count_spearman": -0.047579014170550236}, "t": 3389.1}
{"event": "eval", "step": 3600, "phase": 2, "bpb": {"en": 1.948, "tr": 2.13659, "fi": 2.28966}, "bytes_per_chunk": {"en": 4.859, "tr": 5.902, "fi": 6.216}, "bpb_mean": 2.12475, "frozen_inventory": true, "t": 3391.8}
{"event": "train", "step": 3600, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.424, "bpb": 2.0544, "chunk_loss": 0.030282, "F": 0.1937, "G": 0.22149, "chunks": 198.34, "bytes_per_chunk": 5.163, "boundary_entropy": 0.12607, "router_noise": 0.0, "grad_norm": 0.6929, "lr": 0.0001319, "router_trainable": true, "sec_per_step": 1.169, "probe_sec": 0.1352, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.09681, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": 0.14923, "n_shadow": 6, "mean_delta_shadow": -0.02642, "n_split": 0, "mean_delta_split": NaN, "n_merge": 1, "mean_delta_merge": 0.16406, "n_dropped_unpaired": 0, "n_full": 1, "trunc_ratio_sum": 0.58594, "trunc_abs_frac": 0.58594, "mean_delta_full": 0.62927, "read_hit_rate": {"surface": 0.37356, "split": 0.50169, "merge": 0.13592, "skel": 0.29616}, "mean_gate": {"surface": 0.21648, "split": 0.17668, "merge": 0.21564, "skel": 0.24179}, "n_reads": 10657, "n_cand": 29526, "read_suppressed_frac": 0.1051, "dropout_reads_logged": 10657, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1559, "admitted_surface": 9562, "staged_surface": 52119, "admitted_split": 45027, "staged_split": 36488, "admitted_merge": 6539, "staged_merge": 133094, "admitted_skel": 4408, "staged_skel": 40443, "mean_count": 468.705322265625, "n_util_obs": 277, "util_count_spearman": -0.046485367035205616}, "probe_overhead_sec_total": 482.17, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.008795, "alpha": 0.03, "t": 3396.8}
{"event": "consolidate", "step": 3650, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262065, "t": 3432.5}
{"event": "train", "step": 3650, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.35038, "bpb": 1.94819, "chunk_loss": 0.030339, "F": 0.19962, "G": 0.21202, "chunks": 204.41, "bytes_per_chunk": 5.01, "boundary_entropy": 0.08886, "router_noise": 0.0, "grad_norm": 0.655, "lr": 0.0001283, "router_trainable": true, "sec_per_step": 0.723, "probe_sec": 0.0866, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.06329, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 5, "mean_delta_loo": -0.0225, "n_shadow": 6, "mean_delta_shadow": 0.02173, "n_split": 1, "mean_delta_split": 0.09875, "n_merge": 1, "mean_delta_merge": -0.03899, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 0.24351, "trunc_abs_frac": 0.40052, "mean_delta_full": 0.05878, "read_hit_rate": {"surface": 0.39299, "split": 0.50495, "merge": 0.14096, "skel": 0.32562}, "mean_gate": {"surface": 0.2243, "split": 0.18438, "merge": 0.22274, "skel": 0.25306}, "n_reads": 11483, "n_cand": 30801, "read_suppressed_frac": 0.0997, "dropout_reads_logged": 11483, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1531, "admitted_surface": 9405, "staged_surface": 52740, "admitted_split": 45343, "staged_split": 36803, "admitted_merge": 6422, "staged_merge": 132102, "admitted_skel": 4366, "staged_skel": 40499, "mean_count": 455.61669921875, "n_util_obs": 260, "util_count_spearman": -0.04340847099467789}, "probe_overhead_sec_total": 487.16, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.009632, "alpha": 0.03, "t": 3433.0}
{"event": "consolidate", "step": 3700, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262065, "t": 3474.2}
{"event": "train", "step": 3700, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.4573, "bpb": 2.10243, "chunk_loss": 0.03024, "F": 0.19031, "G": 0.21892, "chunks": 194.88, "bytes_per_chunk": 5.255, "boundary_entropy": 0.13107, "router_noise": 0.0, "grad_norm": 0.7643, "lr": 0.0001247, "router_trainable": true, "sec_per_step": 0.84, "probe_sec": 0.095, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.07741, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 4, "mean_delta_loo": -0.13797, "n_shadow": 6, "mean_delta_shadow": 0.0637, "n_split": 1, "mean_delta_split": -0.20605, "n_merge": 1, "mean_delta_merge": -0.12671, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 0.70889, "trunc_abs_frac": 0.6462, "mean_delta_full": 0.13696, "read_hit_rate": {"surface": 0.31609, "split": 0.4596, "merge": 0.10807, "skel": 0.25857}, "mean_gate": {"surface": 0.21758, "split": 0.18089, "merge": 0.22364, "skel": 0.2501}, "n_reads": 9200, "n_cand": 28844, "read_suppressed_frac": 0.0965, "dropout_reads_logged": 9200, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1501, "admitted_surface": 9136, "staged_surface": 52550, "admitted_split": 45436, "staged_split": 36520, "admitted_merge": 6643, "staged_merge": 132643, "admitted_skel": 4321, "staged_skel": 40431, "mean_count": 451.22979736328125, "n_util_obs": 247, "util_count_spearman": -0.09914453343527252}, "probe_overhead_sec_total": 492.069, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.009334, "alpha": 0.03, "t": 3475.0}
{"event": "consolidate", "step": 3750, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262065, "t": 3522.4}
{"event": "train", "step": 3750, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.43982, "bpb": 2.07722, "chunk_loss": 0.030495, "F": 0.20197, "G": 0.22521, "chunks": 206.81, "bytes_per_chunk": 4.951, "boundary_entropy": 0.11993, "router_noise": 0.0, "grad_norm": 0.7495, "lr": 0.0001212, "router_trainable": true, "sec_per_step": 0.969, "probe_sec": 0.1082, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.05519, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 4, "mean_delta_loo": 0.25264, "n_shadow": 4, "mean_delta_shadow": -0.0815, "n_split": 1, "mean_delta_split": -0.15265, "n_merge": 2, "mean_delta_merge": 0.0697, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": -2.38329, "trunc_abs_frac": 2.46638, "mean_delta_full": 0.08472, "read_hit_rate": {"surface": 0.3222, "split": 0.45482, "merge": 0.10299, "skel": 0.28005}, "mean_gate": {"surface": 0.21909, "split": 0.18068, "merge": 0.22456, "skel": 0.25148}, "n_reads": 9895, "n_cand": 30722, "read_suppressed_frac": 0.1032, "dropout_reads_logged": 9895, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1380, "admitted_surface": 8779, "staged_surface": 52771, "admitted_split": 45896, "staged_split": 36723, "admitted_merge": 6545, "staged_merge": 132285, "admitted_skel": 4316, "staged_skel": 40365, "mean_count": 448.025146484375, "n_util_obs": 250, "util_count_spearman": -0.08902529640474248}, "probe_overhead_sec_total": 498.764, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.010083, "alpha": 0.03, "t": 3523.5}
{"event": "consolidate", "step": 3800, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262065, "t": 3570.0}
{"event": "probe", "step": 3800, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3378.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.194104, "cps": 26.390625, "mean_p": 0.198588, "boundary_entropy": 0.071433, "bpb": 1.919468}, "tr": {"chunks": 3194.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.176435, "cps": 24.953125, "mean_p": 0.219869, "boundary_entropy": 0.216673, "bpb": 2.117846}, "fi": {"chunks": 3056.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.162458, "cps": 23.875, "mean_p": 0.262477, "boundary_entropy": 0.338518, "bpb": 2.266461}}, "bpb": {"en": 1.91947, "tr": 2.11785, "fi": 2.26646}, "gini_cpc": 0.039583, "gini_cps": 0.022296, "bpic": 5.892397, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9089685045774677, "fi": 0.8369623872808635}, "inventory": {"size": 326656, "admitted": 65536, "staged": 261120, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1251, "admitted_surface": 8419, "staged_surface": 53314, "admitted_split": 46368, "staged_split": 37302, "admitted_merge": 6445, "staged_merge": 129874, "admitted_skel": 4304, "staged_skel": 40630, "mean_count": 428.9971618652344, "n_util_obs": 242, "util_count_spearman": -0.11418900611336988}, "t": 3571.5}
{"event": "eval", "step": 3800, "phase": 2, "bpb": {"en": 1.93775, "tr": 2.12442, "fi": 2.27693}, "bytes_per_chunk": {"en": 4.875, "tr": 5.883, "fi": 6.291}, "bpb_mean": 2.11303, "frozen_inventory": true, "t": 3573.9}
{"event": "train", "step": 3800, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.33443, "bpb": 1.92518, "chunk_loss": 0.030258, "F": 0.19562, "G": 0.21404, "chunks": 200.31, "bytes_per_chunk": 5.112, "boundary_entropy": 0.09896, "router_noise": 0.0, "grad_norm": 0.7285, "lr": 0.0001177, "router_trainable": true, "sec_per_step": 1.143, "probe_sec": 0.1026, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.06689, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": -0.01633, "n_shadow": 5, "mean_delta_shadow": -0.0042, "n_split": 1, "mean_delta_split": -0.08954, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 2.27912, "trunc_abs_frac": 1.78479, "mean_delta_full": -0.09607, "read_hit_rate": {"surface": 0.31844, "split": 0.46937, "merge": 0.10038, "skel": 0.28749}, "mean_gate": {"surface": 0.24588, "split": 0.19482, "merge": 0.24206, "skel": 0.26592}, "n_reads": 10022, "n_cand": 30539, "read_suppressed_frac": 0.1034, "dropout_reads_logged": 10022, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1251, "admitted_surface": 8419, "staged_surface": 53338, "admitted_split": 46368, "staged_split": 37307, "admitted_merge": 6445, "staged_merge": 130848, "admitted_skel": 4304, "staged_skel": 40651, "mean_count": 429.15008544921875, "n_util_obs": 242, "util_count_spearman": -0.1141220664670935}, "probe_overhead_sec_total": 503.882, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.0095, "alpha": 0.03, "t": 3580.6}
{"event": "consolidate", "step": 3850, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262065, "t": 3622.8}
{"event": "train", "step": 3850, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.34843, "bpb": 1.94537, "chunk_loss": 0.030393, "F": 0.20291, "G": 0.21266, "chunks": 207.78, "bytes_per_chunk": 4.928, "boundary_entropy": 0.08502, "router_noise": 0.0, "grad_norm": 0.6876, "lr": 0.0001142, "router_trainable": true, "sec_per_step": 0.858, "probe_sec": 0.1019, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.0563, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 5, "mean_delta_loo": -0.01203, "n_shadow": 5, "mean_delta_shadow": 0.37624, "n_split": 2, "mean_delta_split": -0.10313, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 0.75733, "trunc_abs_frac": 0.43285, "mean_delta_full": 0.95585, "read_hit_rate": {"surface": 0.30875, "split": 0.43845, "merge": 0.10782, "skel": 0.27432}, "mean_gate": {"surface": 0.23976, "split": 0.1933, "merge": 0.23799, "skel": 0.25805}, "n_reads": 9747, "n_cand": 31232, "read_suppressed_frac": 0.0966, "dropout_reads_logged": 9747, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1256, "admitted_surface": 8420, "staged_surface": 53464, "admitted_split": 45963, "staged_split": 37516, "admitted_merge": 6864, "staged_merge": 130820, "admitted_skel": 4289, "staged_skel": 40344, "mean_count": 408.9071044921875, "n_util_obs": 219, "util_count_spearman": -0.17319569961269093}, "probe_overhead_sec_total": 509.059, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.009548, "alpha": 0.03, "t": 3623.5}
{"event": "consolidate", "step": 3900, "retired": 0, "split_retired": 1, "merge_admitted": 0, "admitted": 1, "swapped": 1024, "uncertain_staged": 262064, "t": 3667.0}
{"event": "train", "step": 3900, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.41895, "bpb": 2.04711, "chunk_loss": 0.030307, "F": 0.19693, "G": 0.21793, "chunks": 201.66, "bytes_per_chunk": 5.078, "boundary_entropy": 0.11038, "router_noise": 0.0, "grad_norm": 0.6861, "lr": 0.0001108, "router_trainable": true, "sec_per_step": 0.893, "probe_sec": 0.1822, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.14119, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": -0.18273, "n_shadow": 4, "mean_delta_shadow": 0.10361, "n_split": 1, "mean_delta_split": -0.05316, "n_merge": 1, "mean_delta_merge": 0.00449, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 0.60942, "trunc_abs_frac": 0.33742, "mean_delta_full": -0.55884, "read_hit_rate": {"surface": 0.29232, "split": 0.41918, "merge": 0.0828, "skel": 0.26186}, "mean_gate": {"surface": 0.23688, "split": 0.19043, "merge": 0.23032, "skel": 0.25775}, "n_reads": 8880, "n_cand": 30241, "read_suppressed_frac": 0.0975, "dropout_reads_logged": 8880, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1272, "admitted_surface": 8418, "staged_surface": 52880, "admitted_split": 45450, "staged_split": 37353, "admitted_merge": 7381, "staged_merge": 131894, "admitted_skel": 4287, "staged_skel": 40017, "mean_count": 408.4331970214844, "n_util_obs": 223, "util_count_spearman": -0.17996628114341118}, "probe_overhead_sec_total": 514.716, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.009998, "alpha": 0.03, "t": 3668.2}
{"event": "consolidate", "step": 3950, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262064, "t": 3727.2}
{"event": "train", "step": 3950, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.41454, "bpb": 2.04075, "chunk_loss": 0.03029, "F": 0.1972, "G": 0.22062, "chunks": 201.94, "bytes_per_chunk": 5.071, "boundary_entropy": 0.11288, "router_noise": 0.0, "grad_norm": 0.7209, "lr": 0.0001074, "router_trainable": true, "sec_per_step": 1.193, "probe_sec": 0.1033, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.05295, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": 0.03156, "n_shadow": 4, "mean_delta_shadow": -0.00363, "n_split": 2, "mean_delta_split": 0.24759, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 0.53141, "trunc_abs_frac": 0.38515, "mean_delta_full": 0.31417, "read_hit_rate": {"surface": 0.28149, "split": 0.38745, "merge": 0.08018, "skel": 0.25852}, "mean_gate": {"surface": 0.24003, "split": 0.19128, "merge": 0.22565, "skel": 0.25872}, "n_reads": 8403, "n_cand": 30298, "read_suppressed_frac": 0.0948, "dropout_reads_logged": 8403, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1247, "admitted_surface": 8389, "staged_surface": 52897, "admitted_split": 45102, "staged_split": 37975, "admitted_merge": 7749, "staged_merge": 131269, "admitted_skel": 4296, "staged_skel": 40003, "mean_count": 403.7048645019531, "n_util_obs": 224, "util_count_spearman": -0.17073563954729873}, "probe_overhead_sec_total": 520.715, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.010689, "alpha": 0.03, "t": 3727.8}
{"event": "consolidate", "step": 4000, "retired": 0, "split_retired": 1, "merge_admitted": 0, "admitted": 1, "swapped": 1024, "uncertain_staged": 262065, "t": 3774.9}
{"event": "probe", "step": 4000, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3410.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.195943, "cps": 26.640625, "mean_p": 0.200929, "boundary_entropy": 0.071696, "bpb": 1.911071}, "tr": {"chunks": 3144.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.173673, "cps": 24.5625, "mean_p": 0.216319, "boundary_entropy": 0.21061, "bpb": 2.100972}, "fi": {"chunks": 2991.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.159003, "cps": 23.367188, "mean_p": 0.256556, "boundary_entropy": 0.327184, "bpb": 2.25686}}, "bpb": {"en": 1.91107, "tr": 2.10097, "fi": 2.25686}, "gini_cpc": 0.046587, "gini_cps": 0.029265, "bpic": 5.943635, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.8863428122200061, "fi": 0.8114733673487046}, "inventory": {"size": 326655, "admitted": 65536, "staged": 261119, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1254, "admitted_surface": 8392, "staged_surface": 53005, "admitted_split": 44542, "staged_split": 38850, "admitted_merge": 8310, "staged_merge": 129325, "admitted_skel": 4292, "staged_skel": 39939, "mean_count": 400.9999694824219, "n_util_obs": 230, "util_count_spearman": -0.14724516038356936}, "t": 3776.4}
{"event": "eval", "step": 4000, "phase": 2, "bpb": {"en": 1.92727, "tr": 2.10719, "fi": 2.26657}, "bytes_per_chunk": {"en": 4.85, "tr": 5.974, "fi": 6.359}, "bpb_mean": 2.10034, "frozen_inventory": true, "t": 3779.3}
{"event": "train", "step": 4000, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.35855, "bpb": 1.95998, "chunk_loss": 0.030253, "F": 0.19211, "G": 0.21632, "chunks": 196.72, "bytes_per_chunk": 5.205, "boundary_entropy": 0.12072, "router_noise": 0.0, "grad_norm": 0.6916, "lr": 0.000104, "router_trainable": true, "sec_per_step": 1.12, "probe_sec": 0.1457, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.12375, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": 0.0562, "n_shadow": 6, "mean_delta_shadow": -0.04641, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": 0.45957, "trunc_abs_frac": 2.59641, "mean_delta_full": -0.05012, "read_hit_rate": {"surface": 0.26776, "split": 0.39357, "merge": 0.08217, "skel": 0.25073}, "mean_gate": {"surface": 0.23901, "split": 0.18654, "merge": 0.22376, "skel": 0.25712}, "n_reads": 8142, "n_cand": 29481, "read_suppressed_frac": 0.1033, "dropout_reads_logged": 8142, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1253, "admitted_surface": 8392, "staged_surface": 53158, "admitted_split": 44542, "staged_split": 39011, "admitted_merge": 8310, "staged_merge": 129876, "admitted_skel": 4292, "staged_skel": 40099, "mean_count": 401.12420654296875, "n_util_obs": 230, "util_count_spearman": -0.14625492221846884}, "probe_overhead_sec_total": 526.229, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.010025, "alpha": 0.03, "t": 3783.9}
{"event": "consolidate", "step": 4050, "retired": 0, "split_retired": 1, "merge_admitted": 0, "admitted": 1, "swapped": 1024, "uncertain_staged": 262064, "t": 3832.4}
{"event": "train", "step": 4050, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.496, "bpb": 2.15827, "chunk_loss": 0.030279, "F": 0.19147, "G": 0.2191, "chunks": 196.06, "bytes_per_chunk": 5.223, "boundary_entropy": 0.1269, "router_noise": 0.0, "grad_norm": 0.7711, "lr": 0.0001007, "router_trainable": true, "sec_per_step": 0.981, "probe_sec": 0.1143, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.09537, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 4, "mean_delta_loo": 0.06143, "n_shadow": 9, "mean_delta_shadow": 0.12237, "n_split": 1, "mean_delta_split": -0.1366, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 0.77045, "trunc_abs_frac": 0.55819, "mean_delta_full": -0.19645, "read_hit_rate": {"surface": 0.24159, "split": 0.34539, "merge": 0.06538, "skel": 0.2318}, "mean_gate": {"surface": 0.23461, "split": 0.18616, "merge": 0.21914, "skel": 0.25226}, "n_reads": 7086, "n_cand": 29075, "read_suppressed_frac": 0.1043, "dropout_reads_logged": 7086, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1206, "admitted_surface": 8334, "staged_surface": 53331, "admitted_split": 44082, "staged_split": 40284, "admitted_merge": 8832, "staged_merge": 128078, "admitted_skel": 4288, "staged_skel": 40451, "mean_count": 391.5421142578125, "n_util_obs": 234, "util_count_spearman": -0.14945366175520308}, "probe_overhead_sec_total": 532.343, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.009676, "alpha": 0.03, "t": 3832.9}
{"event": "consolidate", "step": 4100, "retired": 0, "split_retired": 2, "merge_admitted": 0, "admitted": 2, "swapped": 1024, "uncertain_staged": 262061, "t": 3882.1}
{"event": "train", "step": 4100, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.42651, "bpb": 2.05802, "chunk_loss": 0.030271, "F": 0.19339, "G": 0.22628, "chunks": 198.03, "bytes_per_chunk": 5.171, "boundary_entropy": 0.13884, "router_noise": 0.0, "grad_norm": 0.6995, "lr": 9.75e-05, "router_trainable": true, "sec_per_step": 1.016, "probe_sec": 0.129, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.07421, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": -0.08601, "n_shadow": 5, "mean_delta_shadow": -0.10131, "n_split": 2, "mean_delta_split": 0.07236, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 6, "trunc_ratio_sum": 0.39319, "trunc_abs_frac": 0.45541, "mean_delta_full": -0.16848, "read_hit_rate": {"surface": 0.19699, "split": 0.32752, "merge": 0.05771, "skel": 0.2142}, "mean_gate": {"surface": 0.23967, "split": 0.18975, "merge": 0.22918, "skel": 0.25687}, "n_reads": 6559, "n_cand": 29466, "read_suppressed_frac": 0.1037, "dropout_reads_logged": 6559, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1120, "admitted_surface": 8110, "staged_surface": 53181, "admitted_split": 43775, "staged_split": 40795, "admitted_merge": 9376, "staged_merge": 127932, "admitted_skel": 4275, "staged_skel": 40236, "mean_count": 374.49261474609375, "n_util_obs": 214, "util_count_spearman": -0.18963483137468584}, "probe_overhead_sec_total": 538.0, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.010736, "alpha": 0.03, "t": 3883.7}
{"event": "consolidate", "step": 4150, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262060, "t": 3932.1}
{"event": "train", "step": 4150, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.36479, "bpb": 1.96898, "chunk_loss": 0.030417, "F": 0.20242, "G": 0.21749, "chunks": 207.28, "bytes_per_chunk": 4.94, "boundary_entropy": 0.09565, "router_noise": 0.0, "grad_norm": 0.6784, "lr": 9.43e-05, "router_trainable": true, "sec_per_step": 0.979, "probe_sec": 0.0965, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.06278, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 3, "mean_delta_loo": -0.06005, "n_shadow": 10, "mean_delta_shadow": 0.04299, "n_split": 1, "mean_delta_split": -0.09546, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": -0.08672, "trunc_abs_frac": 0.89579, "mean_delta_full": 0.10299, "read_hit_rate": {"surface": 0.23254, "split": 0.35928, "merge": 0.06805, "skel": 0.26432}, "mean_gate": {"surface": 0.25316, "split": 0.19729, "merge": 0.23796, "skel": 0.26931}, "n_reads": 7922, "n_cand": 31152, "read_suppressed_frac": 0.1006, "dropout_reads_logged": 7922, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1136, "admitted_surface": 8111, "staged_surface": 53149, "admitted_split": 44251, "staged_split": 40724, "admitted_merge": 9872, "staged_merge": 128207, "admitted_skel": 3302, "staged_skel": 40064, "mean_count": 378.1905517578125, "n_util_obs": 219, "util_count_spearman": -0.18250088044286522}, "probe_overhead_sec_total": 543.331, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.011229, "alpha": 0.03, "t": 3932.6}
{"event": "consolidate", "step": 4200, "retired": 0, "split_retired": 2, "merge_admitted": 0, "admitted": 2, "swapped": 1024, "uncertain_staged": 262058, "t": 3979.4}
{"event": "probe", "step": 4200, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3395.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.195081, "cps": 26.523438, "mean_p": 0.199009, "boundary_entropy": 0.069791, "bpb": 1.899059}, "tr": {"chunks": 3187.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.176048, "cps": 24.898438, "mean_p": 0.21739, "boundary_entropy": 0.208984, "bpb": 2.083371}, "fi": {"chunks": 2971.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.15794, "cps": 23.210938, "mean_p": 0.265648, "boundary_entropy": 0.341558, "bpb": 2.239702}}, "bpb": {"en": 1.89906, "tr": 2.08337, "fi": 2.2397}, "gini_cpc": 0.046801, "gini_cps": 0.029589, "bpic": 5.938658, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9024348400093493, "fi": 0.8096085947267561}, "inventory": {"size": 326654, "admitted": 65536, "staged": 261118, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1154, "admitted_surface": 8125, "staged_surface": 53485, "admitted_split": 44786, "staged_split": 41235, "admitted_merge": 10342, "staged_merge": 126329, "admitted_skel": 2283, "staged_skel": 40069, "mean_count": 381.93768310546875, "n_util_obs": 224, "util_count_spearman": -0.17464188268304648}, "t": 3980.8}
{"event": "eval", "step": 4200, "phase": 2, "bpb": {"en": 1.91598, "tr": 2.08932, "fi": 2.24749}, "bytes_per_chunk": {"en": 4.866, "tr": 5.871, "fi": 6.327}, "bpb_mean": 2.08426, "frozen_inventory": true, "t": 3983.1}
{"event": "phase_transition", "step": 4200, "phase": 3, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0, "t": 3988.1}
{"event": "train", "step": 4200, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.44677, "bpb": 2.08725, "chunk_loss": 0.030374, "F": 0.20007, "G": 0.21858, "chunks": 204.88, "bytes_per_chunk": 4.998, "boundary_entropy": 0.10308, "router_noise": 0.0, "grad_norm": 0.7277, "lr": 9.12e-05, "router_trainable": true, "sec_per_step": 1.138, "probe_sec": 0.123, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.07947, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": 0.04006, "n_shadow": 3, "mean_delta_shadow": 0.06684, "n_split": 0, "mean_delta_split": NaN, "n_merge": 1, "mean_delta_merge": -0.01672, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": -5.23258, "trunc_abs_frac": 0.72644, "mean_delta_full": 0.02161, "read_hit_rate": {"surface": 0.21613, "split": 0.33472, "merge": 0.07132, "skel": 0.23603}, "mean_gate": {"surface": 0.24915, "split": 0.19335, "merge": 0.23288, "skel": 0.26841}, "n_reads": 7195, "n_cand": 30451, "read_suppressed_frac": 0.0973, "dropout_reads_logged": 7195, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1152, "admitted_surface": 8125, "staged_surface": 53591, "admitted_split": 44786, "staged_split": 41301, "admitted_merge": 10342, "staged_merge": 127085, "admitted_skel": 2283, "staged_skel": 40167, "mean_count": 382.0474853515625, "n_util_obs": 224, "util_count_spearman": -0.17520023894946143}, "probe_overhead_sec_total": 549.133, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.010773, "alpha": 0.03, "t": 3989.5}
{"event": "consolidate", "step": 4250, "retired": 0, "split_retired": 1, "merge_admitted": 0, "admitted": 1, "swapped": 1024, "uncertain_staged": 262059, "t": 4048.7}
{"event": "train", "step": 4250, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.33825, "bpb": 1.93069, "chunk_loss": 0.030242, "F": 0.19446, "G": 0.21534, "chunks": 199.12, "bytes_per_chunk": 5.142, "boundary_entropy": 0.10715, "router_noise": 0.0, "grad_norm": 0.6737, "lr": 8.81e-05, "router_trainable": true, "sec_per_step": 1.211, "probe_sec": 0.1551, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.10254, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": 0.49167, "n_shadow": 5, "mean_delta_shadow": -0.02989, "n_split": 2, "mean_delta_split": -0.07416, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 0.00395, "trunc_abs_frac": 1.24771, "mean_delta_full": 0.11064, "read_hit_rate": {"surface": 0.2347, "split": 0.3655, "merge": 0.07895, "skel": 0.25517}, "mean_gate": {"surface": 0.25472, "split": 0.19476, "merge": 0.24054, "skel": 0.26731}, "n_reads": 7815, "n_cand": 30232, "read_suppressed_frac": 0.0964, "dropout_reads_logged": 7815, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1131, "admitted_surface": 8111, "staged_surface": 53426, "admitted_split": 45282, "staged_split": 40818, "admitted_merge": 10734, "staged_merge": 127719, "admitted_skel": 1409, "staged_skel": 40181, "mean_count": 382.15850830078125, "n_util_obs": 226, "util_count_spearman": -0.22445186658139274}, "probe_overhead_sec_total": 555.147, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.010244, "alpha": 0.03, "t": 4050.1}
{"event": "consolidate", "step": 4300, "retired": 0, "split_retired": 1, "merge_admitted": 0, "admitted": 1, "swapped": 1024, "uncertain_staged": 262059, "t": 4100.4}
{"event": "train", "step": 4300, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.34585, "bpb": 1.94165, "chunk_loss": 0.030212, "F": 0.19, "G": 0.21917, "chunks": 194.56, "bytes_per_chunk": 5.263, "boundary_entropy": 0.13112, "router_noise": 0.0, "grad_norm": 0.7334, "lr": 8.5e-05, "router_trainable": true, "sec_per_step": 1.018, "probe_sec": 0.1102, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.0889, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": 0.10611, "n_shadow": 4, "mean_delta_shadow": -0.14084, "n_split": 1, "mean_delta_split": 0.01381, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 0.1604, "trunc_abs_frac": 0.08085, "mean_delta_full": 0.03238, "read_hit_rate": {"surface": 0.20875, "split": 0.33592, "merge": 0.06962, "skel": 0.22429}, "mean_gate": {"surface": 0.24874, "split": 0.19257, "merge": 0.23358, "skel": 0.267}, "n_reads": 6776, "n_cand": 29059, "read_suppressed_frac": 0.1011, "dropout_reads_logged": 6776, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1131, "admitted_surface": 8090, "staged_surface": 52551, "admitted_split": 45605, "staged_split": 39934, "admitted_merge": 11088, "staged_merge": 130111, "admitted_skel": 753, "staged_skel": 39548, "mean_count": 385.31878662109375, "n_util_obs": 228, "util_count_spearman": -0.2719670175490196}, "probe_overhead_sec_total": 561.517, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.010872, "alpha": 0.03, "t": 4101.0}
{"event": "consolidate", "step": 4350, "retired": 0, "split_retired": 2, "merge_admitted": 0, "admitted": 2, "swapped": 1024, "uncertain_staged": 262059, "t": 4147.3}
{"event": "train", "step": 4350, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.32632, "bpb": 1.91348, "chunk_loss": 0.030271, "F": 0.19437, "G": 0.21309, "chunks": 199.03, "bytes_per_chunk": 5.145, "boundary_entropy": 0.10329, "router_noise": 0.0, "grad_norm": 0.7281, "lr": 8.21e-05, "router_trainable": true, "sec_per_step": 0.946, "probe_sec": 0.0708, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04749, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": 0.27341, "n_shadow": 3, "mean_delta_shadow": 0.10233, "n_split": 1, "mean_delta_split": -0.17004, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 2.89547, "trunc_abs_frac": 9.8254, "mean_delta_full": 0.05003, "read_hit_rate": {"surface": 0.20183, "split": 0.34298, "merge": 0.05361, "skel": 0.24108}, "mean_gate": {"surface": 0.25625, "split": 0.19257, "merge": 0.24035, "skel": 0.26969}, "n_reads": 7017, "n_cand": 29943, "read_suppressed_frac": 0.1085, "dropout_reads_logged": 7017, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1063, "admitted_surface": 7900, "staged_surface": 53544, "admitted_split": 45907, "staged_split": 40836, "admitted_merge": 11105, "staged_merge": 127488, "admitted_skel": 624, "staged_skel": 40276, "mean_count": 379.1823425292969, "n_util_obs": 231, "util_count_spearman": -0.2574062621464333}, "probe_overhead_sec_total": 567.048, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.010493, "alpha": 0.03, "t": 4148.3}
{"event": "consolidate", "step": 4400, "retired": 0, "split_retired": 1, "merge_admitted": 0, "admitted": 1, "swapped": 1024, "uncertain_staged": 262058, "t": 4198.2}
{"event": "probe", "step": 4400, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3401.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.195426, "cps": 26.570312, "mean_p": 0.199913, "boundary_entropy": 0.068075, "bpb": 1.893474}, "tr": {"chunks": 3271.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.180688, "cps": 25.554688, "mean_p": 0.219475, "boundary_entropy": 0.210111, "bpb": 2.071145}, "fi": {"chunks": 3175.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.168784, "cps": 24.804688, "mean_p": 0.264575, "boundary_entropy": 0.333912, "bpb": 2.224843}}, "bpb": {"en": 1.89347, "tr": 2.07114, "fi": 2.22484}, "gini_cpc": 0.032595, "gini_cps": 0.015301, "bpic": 5.761349, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9245863573663872, "fi": 0.863672983071786}, "inventory": {"size": 326655, "admitted": 65536, "staged": 261119, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1004, "admitted_surface": 7905, "staged_surface": 53366, "admitted_split": 45783, "staged_split": 40722, "admitted_merge": 11230, "staged_merge": 126938, "admitted_skel": 618, "staged_skel": 40093, "mean_count": 323.286376953125, "n_util_obs": 213, "util_count_spearman": -0.2202154711689524}, "t": 4199.7}
{"event": "eval", "step": 4400, "phase": 3, "bpb": {"en": 1.911, "tr": 2.07934, "fi": 2.23296}, "bytes_per_chunk": {"en": 4.854, "tr": 5.77, "fi": 6.124}, "bpb_mean": 2.07443, "frozen_inventory": true, "t": 4202.4}
{"event": "train", "step": 4400, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.36631, "bpb": 1.97116, "chunk_loss": 0.030421, "F": 0.20514, "G": 0.21654, "chunks": 210.06, "bytes_per_chunk": 4.875, "boundary_entropy": 0.08607, "router_noise": 0.0, "grad_norm": 0.7865, "lr": 7.92e-05, "router_trainable": true, "sec_per_step": 1.181, "probe_sec": 0.1527, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.08143, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": 0.03748, "n_shadow": 2, "mean_delta_shadow": -0.00742, "n_split": 1, "mean_delta_split": 0.09503, "n_merge": 1, "mean_delta_merge": -0.85925, "n_dropped_unpaired": 0, "n_full": 6, "trunc_ratio_sum": 3.74546, "trunc_abs_frac": 0.6269, "mean_delta_full": -0.02242, "read_hit_rate": {"surface": 0.1994, "split": 0.27554, "merge": 0.04551, "skel": 0.23433}, "mean_gate": {"surface": 0.25313, "split": 0.19607, "merge": 0.23629, "skel": 0.27479}, "n_reads": 6423, "n_cand": 31541, "read_suppressed_frac": 0.0979, "dropout_reads_logged": 6423, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1005, "admitted_surface": 7905, "staged_surface": 53441, "admitted_split": 45783, "staged_split": 40939, "admitted_merge": 11230, "staged_merge": 127575, "admitted_skel": 618, "staged_skel": 40189, "mean_count": 323.3843994140625, "n_util_obs": 213, "util_count_spearman": -0.22033475250580076}, "probe_overhead_sec_total": 572.415, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.011548, "alpha": 0.03, "t": 4207.3}
{"event": "consolidate", "step": 4450, "retired": 0, "split_retired": 1, "merge_admitted": 0, "admitted": 1, "swapped": 1024, "uncertain_staged": 262057, "t": 4262.7}
{"event": "train", "step": 4450, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.39136, "bpb": 2.0073, "chunk_loss": 0.030259, "F": 0.19394, "G": 0.20894, "chunks": 198.59, "bytes_per_chunk": 5.156, "boundary_entropy": 0.09279, "router_noise": 0.0, "grad_norm": 0.7131, "lr": 7.63e-05, "router_trainable": true, "sec_per_step": 1.146, "probe_sec": 0.0801, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.0463, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": 0.10069, "n_shadow": 4, "mean_delta_shadow": -0.07684, "n_split": 2, "mean_delta_split": 0.17262, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 6, "trunc_ratio_sum": 0.95562, "trunc_abs_frac": 0.82266, "mean_delta_full": 0.16081, "read_hit_rate": {"surface": 0.18836, "split": 0.27021, "merge": 0.02114, "skel": 0.20991}, "mean_gate": {"surface": 0.25612, "split": 0.19554, "merge": 0.24202, "skel": 0.2739}, "n_reads": 5672, "n_cand": 29881, "read_suppressed_frac": 0.0987, "dropout_reads_logged": 5672, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 908, "admitted_surface": 7923, "staged_surface": 52601, "admitted_split": 46319, "staged_split": 40899, "admitted_merge": 10673, "staged_merge": 128962, "admitted_skel": 621, "staged_skel": 39682, "mean_count": 320.3306884765625, "n_util_obs": 221, "util_count_spearman": -0.24704179714871694}, "probe_overhead_sec_total": 578.094, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.011465, "alpha": 0.03, "t": 4264.6}
{"event": "consolidate", "step": 4500, "retired": 0, "split_retired": 1, "merge_admitted": 0, "admitted": 1, "swapped": 1024, "uncertain_staged": 262056, "t": 4311.5}
{"event": "train", "step": 4500, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.40718, "bpb": 2.03014, "chunk_loss": 0.030286, "F": 0.19464, "G": 0.22511, "chunks": 199.31, "bytes_per_chunk": 5.138, "boundary_entropy": 0.13787, "router_noise": 0.0, "grad_norm": 0.7231, "lr": 7.36e-05, "router_trainable": true, "sec_per_step": 0.959, "probe_sec": 0.077, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04943, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": 0.16607, "n_shadow": 5, "mean_delta_shadow": 0.07007, "n_split": 2, "mean_delta_split": 0.0376, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 1, "trunc_ratio_sum": 3.42409, "trunc_abs_frac": 3.42409, "mean_delta_full": 0.09729, "read_hit_rate": {"surface": 0.16766, "split": 0.19607, "merge": 0.02265, "skel": 0.20443}, "mean_gate": {"surface": 0.26034, "split": 0.20292, "merge": 0.24919, "skel": 0.27523}, "n_reads": 4596, "n_cand": 29725, "read_suppressed_frac": 0.0894, "dropout_reads_logged": 4596, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 892, "admitted_surface": 7885, "staged_surface": 52871, "admitted_split": 45917, "staged_split": 42059, "admitted_merge": 11118, "staged_merge": 127338, "admitted_skel": 616, "staged_skel": 39876, "mean_count": 286.7506103515625, "n_util_obs": 230, "util_count_spearman": -0.2078273236827804}, "probe_overhead_sec_total": 583.643, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.011809, "alpha": 0.03, "t": 4312.5}
{"event": "consolidate", "step": 4550, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262055, "t": 4359.5}
{"event": "train", "step": 4550, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.32737, "bpb": 1.91498, "chunk_loss": 0.030202, "F": 0.18915, "G": 0.21728, "chunks": 193.69, "bytes_per_chunk": 5.287, "boundary_entropy": 0.12739, "router_noise": 0.0, "grad_norm": 0.6958, "lr": 7.09e-05, "router_trainable": true, "sec_per_step": 0.948, "probe_sec": 0.0824, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.05457, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": 0.13919, "n_shadow": 1, "mean_delta_shadow": 0.24689, "n_split": 0, "mean_delta_split": NaN, "n_merge": 2, "mean_delta_merge": -0.02908, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 0.80124, "trunc_abs_frac": 0.93305, "mean_delta_full": 0.29474, "read_hit_rate": {"surface": 0.15212, "split": 0.19111, "merge": 0.01728, "skel": 0.19355}, "mean_gate": {"surface": 0.27016, "split": 0.20291, "merge": 0.25563, "skel": 0.28165}, "n_reads": 4299, "n_cand": 29266, "read_suppressed_frac": 0.097, "dropout_reads_logged": 4299, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 746, "admitted_surface": 7449, "staged_surface": 52419, "admitted_split": 46170, "staged_split": 41881, "admitted_merge": 11329, "staged_merge": 128277, "admitted_skel": 588, "staged_skel": 39567, "mean_count": 249.93954467773438, "n_util_obs": 201, "util_count_spearman": -0.2039485739618738}, "probe_overhead_sec_total": 588.906, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.012467, "alpha": 0.03, "t": 4359.9}
{"event": "consolidate", "step": 4600, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262054, "t": 4412.1}
{"event": "probe", "step": 4600, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3394.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.195024, "cps": 26.515625, "mean_p": 0.198668, "boundary_entropy": 0.069417, "bpb": 1.882559}, "tr": {"chunks": 3159.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.174501, "cps": 24.679688, "mean_p": 0.218413, "boundary_entropy": 0.213268, "bpb": 2.063685}, "fi": {"chunks": 3022.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.160651, "cps": 23.609375, "mean_p": 0.25995, "boundary_entropy": 0.329159, "bpb": 2.216786}}, "bpb": {"en": 1.88256, "tr": 2.06368, "fi": 2.21679}, "gini_cpc": 0.043222, "gini_cps": 0.025901, "bpic": 5.925013, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.8947698807625103, "fi": 0.8237489210900967}, "inventory": {"size": 326656, "admitted": 65536, "staged": 261120, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 696, "admitted_surface": 7138, "staged_surface": 52489, "admitted_split": 46449, "staged_split": 41530, "admitted_merge": 11361, "staged_merge": 127648, "admitted_skel": 588, "staged_skel": 39453, "mean_count": 249.8788604736328, "n_util_obs": 204, "util_count_spearman": -0.21547801626958604}, "t": 4413.6}
{"event": "eval", "step": 4600, "phase": 3, "bpb": {"en": 1.90102, "tr": 2.07229, "fi": 2.22431}, "bytes_per_chunk": {"en": 4.862, "tr": 5.89, "fi": 6.302}, "bpb_mean": 2.06587, "frozen_inventory": true, "t": 4416.1}
{"event": "train", "step": 4600, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.38746, "bpb": 2.00168, "chunk_loss": 0.030195, "F": 0.19122, "G": 0.21937, "chunks": 195.81, "bytes_per_chunk": 5.229, "boundary_entropy": 0.12381, "router_noise": 0.0, "grad_norm": 0.7069, "lr": 6.82e-05, "router_trainable": true, "sec_per_step": 1.222, "probe_sec": 0.0715, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04447, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 5, "mean_delta_loo": 0.10134, "n_shadow": 5, "mean_delta_shadow": 0.10786, "n_split": 1, "mean_delta_split": -0.07317, "n_merge": 1, "mean_delta_merge": 0.22491, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 1.78868, "trunc_abs_frac": 5.66867, "mean_delta_full": 0.06759, "read_hit_rate": {"surface": 0.13731, "split": 0.18959, "merge": 0.01387, "skel": 0.19286}, "mean_gate": {"surface": 0.27822, "split": 0.20821, "merge": 0.25791, "skel": 0.29318}, "n_reads": 4202, "n_cand": 29540, "read_suppressed_frac": 0.0976, "dropout_reads_logged": 4202, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 697, "admitted_surface": 7138, "staged_surface": 52739, "admitted_split": 46449, "staged_split": 41646, "admitted_merge": 11361, "staged_merge": 128169, "admitted_skel": 588, "staged_skel": 39590, "mean_count": 249.94297790527344, "n_util_obs": 204, "util_count_spearman": -0.2153818970817932}, "probe_overhead_sec_total": 593.874, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.012832, "alpha": 0.03, "t": 4421.0}
{"event": "consolidate", "step": 4650, "retired": 0, "split_retired": 1, "merge_admitted": 0, "admitted": 1, "swapped": 1024, "uncertain_staged": 262053, "t": 4465.2}
{"event": "train", "step": 4650, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.38883, "bpb": 2.00365, "chunk_loss": 0.030345, "F": 0.20029, "G": 0.21591, "chunks": 205.09, "bytes_per_chunk": 4.993, "boundary_entropy": 0.09955, "router_noise": 0.0, "grad_norm": 0.7293, "lr": 6.57e-05, "router_trainable": true, "sec_per_step": 0.899, "probe_sec": 0.0884, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.05199, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": 0.03809, "n_shadow": 6, "mean_delta_shadow": 0.00343, "n_split": 0, "mean_delta_split": NaN, "n_merge": 1, "mean_delta_merge": 0.43469, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": -0.19306, "trunc_abs_frac": 0.35985, "mean_delta_full": 0.13452, "read_hit_rate": {"surface": 0.13229, "split": 0.19224, "merge": 0.016, "skel": 0.18373}, "mean_gate": {"surface": 0.27963, "split": 0.20654, "merge": 0.25089, "skel": 0.28978}, "n_reads": 4301, "n_cand": 30604, "read_suppressed_frac": 0.0953, "dropout_reads_logged": 4301, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 692, "admitted_surface": 6301, "staged_surface": 53000, "admitted_split": 47135, "staged_split": 41504, "admitted_merge": 11562, "staged_merge": 128106, "admitted_skel": 538, "staged_skel": 39534, "mean_count": 250.6702117919922, "n_util_obs": 204, "util_count_spearman": -0.2511666828043986}, "probe_overhead_sec_total": 599.001, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.01286, "alpha": 0.03, "t": 4466.0}
{"event": "consolidate", "step": 4700, "retired": 0, "split_retired": 1, "merge_admitted": 0, "admitted": 1, "swapped": 1024, "uncertain_staged": 262052, "t": 4512.0}
{"event": "train", "step": 4700, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.36922, "bpb": 1.97537, "chunk_loss": 0.030264, "F": 0.19321, "G": 0.21848, "chunks": 197.84, "bytes_per_chunk": 5.176, "boundary_entropy": 0.12075, "router_noise": 0.0, "grad_norm": 0.7464, "lr": 6.32e-05, "router_trainable": true, "sec_per_step": 0.945, "probe_sec": 0.0946, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.05538, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 5, "mean_delta_loo": 0.02577, "n_shadow": 4, "mean_delta_shadow": -0.1069, "n_split": 2, "mean_delta_split": 0.15691, "n_merge": 1, "mean_delta_merge": 0.12992, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 1.85249, "trunc_abs_frac": 2.99526, "mean_delta_full": 0.06283, "read_hit_rate": {"surface": 0.12573, "split": 0.15809, "merge": 0.01117, "skel": 0.15903}, "mean_gate": {"surface": 0.26507, "split": 0.19573, "merge": 0.2572, "skel": 0.27447}, "n_reads": 3564, "n_cand": 29627, "read_suppressed_frac": 0.1134, "dropout_reads_logged": 3564, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 593, "admitted_surface": 5762, "staged_surface": 53026, "admitted_split": 47502, "staged_split": 41415, "admitted_merge": 11749, "staged_merge": 128087, "admitted_skel": 523, "staged_skel": 39616, "mean_count": 214.02291870117188, "n_util_obs": 193, "util_count_spearman": -0.19095542395382517}, "probe_overhead_sec_total": 604.038, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.012825, "alpha": 0.03, "t": 4513.2}
{"event": "consolidate", "step": 4750, "retired": 0, "split_retired": 1, "merge_admitted": 0, "admitted": 1, "swapped": 1024, "uncertain_staged": 262050, "t": 4571.7}
{"event": "train", "step": 4750, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.46133, "bpb": 2.10826, "chunk_loss": 0.030304, "F": 0.19379, "G": 0.21835, "chunks": 198.44, "bytes_per_chunk": 5.16, "boundary_entropy": 0.12557, "router_noise": 0.0, "grad_norm": 0.8289, "lr": 6.08e-05, "router_trainable": true, "sec_per_step": 1.191, "probe_sec": 0.1003, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.06099, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 4, "mean_delta_loo": 0.04921, "n_shadow": 4, "mean_delta_shadow": -0.1638, "n_split": 2, "mean_delta_split": -0.12087, "n_merge": 2, "mean_delta_merge": 0.12617, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": -0.01466, "trunc_abs_frac": 0.54489, "mean_delta_full": -0.11552, "read_hit_rate": {"surface": 0.10367, "split": 0.11468, "merge": 0.0097, "skel": 0.14891}, "mean_gate": {"surface": 0.27143, "split": 0.19573, "merge": 0.26675, "skel": 0.28108}, "n_reads": 2835, "n_cand": 29574, "read_suppressed_frac": 0.0938, "dropout_reads_logged": 2835, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 560, "admitted_surface": 5652, "staged_surface": 53593, "admitted_split": 47475, "staged_split": 42131, "admitted_merge": 11924, "staged_merge": 126955, "admitted_skel": 485, "staged_skel": 39465, "mean_count": 191.55328369140625, "n_util_obs": 192, "util_count_spearman": -0.19335368177174722}, "probe_overhead_sec_total": 609.892, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.013081, "alpha": 0.03, "t": 4572.8}
{"event": "consolidate", "step": 4800, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262050, "t": 4619.2}
{"event": "probe", "step": 4800, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3389.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.194737, "cps": 26.476562, "mean_p": 0.198244, "boundary_entropy": 0.06958, "bpb": 1.876941}, "tr": {"chunks": 3175.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.175385, "cps": 24.804688, "mean_p": 0.21692, "boundary_entropy": 0.207359, "bpb": 2.055214}, "fi": {"chunks": 3142.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.16703, "cps": 24.546875, "mean_p": 0.264477, "boundary_entropy": 0.334346, "bpb": 2.206085}}, "bpb": {"en": 1.87694, "tr": 2.05521, "fi": 2.20609}, "gini_cpc": 0.034387, "gini_cps": 0.016965, "bpic": 5.845044, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9006285905345379, "fi": 0.8577225906020252}, "inventory": {"size": 326656, "admitted": 65536, "staged": 261120, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 495, "admitted_surface": 5494, "staged_surface": 53437, "admitted_split": 47651, "staged_split": 42152, "admitted_merge": 12085, "staged_merge": 126230, "admitted_skel": 306, "staged_skel": 39301, "mean_count": 147.25607299804688, "n_util_obs": 183, "util_count_spearman": -0.1353203767243318}, "t": 4620.5}
{"event": "eval", "step": 4800, "phase": 3, "bpb": {"en": 1.89568, "tr": 2.06177, "fi": 2.21351}, "bytes_per_chunk": {"en": 4.865, "tr": 5.891, "fi": 6.107}, "bpb_mean": 2.05699, "frozen_inventory": true, "t": 4622.7}
{"event": "train", "step": 4800, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.33981, "bpb": 1.93293, "chunk_loss": 0.030272, "F": 0.19656, "G": 0.21077, "chunks": 201.28, "bytes_per_chunk": 5.087, "boundary_entropy": 0.09313, "router_noise": 0.0, "grad_norm": 0.7258, "lr": 5.85e-05, "router_trainable": true, "sec_per_step": 1.098, "probe_sec": 0.0831, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.05718, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 10, "mean_delta_loo": 0.14126, "n_shadow": 3, "mean_delta_shadow": 0.08712, "n_split": 1, "mean_delta_split": -0.00528, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 1, "trunc_ratio_sum": 0.28131, "trunc_abs_frac": 0.28131, "mean_delta_full": -0.86267, "read_hit_rate": {"surface": 0.10189, "split": 0.10096, "merge": 0.01082, "skel": 0.09295}, "mean_gate": {"surface": 0.29721, "split": 0.21507, "merge": 0.26777, "skel": 0.30424}, "n_reads": 2461, "n_cand": 30440, "read_suppressed_frac": 0.1008, "dropout_reads_logged": 2461, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 495, "admitted_surface": 5494, "staged_surface": 53516, "admitted_split": 47651, "staged_split": 42223, "admitted_merge": 12085, "staged_merge": 127019, "admitted_skel": 306, "staged_skel": 39386, "mean_count": 147.2936248779297, "n_util_obs": 183, "util_count_spearman": -0.13488567222793055}, "probe_overhead_sec_total": 614.964, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.013336, "alpha": 0.03, "t": 4627.7}
{"event": "consolidate", "step": 4850, "retired": 0, "split_retired": 1, "merge_admitted": 0, "admitted": 1, "swapped": 1024, "uncertain_staged": 262049, "t": 4673.1}
{"event": "train", "step": 4850, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.33619, "bpb": 1.92772, "chunk_loss": 0.030405, "F": 0.20309, "G": 0.21987, "chunks": 207.97, "bytes_per_chunk": 4.924, "boundary_entropy": 0.10296, "router_noise": 0.0, "grad_norm": 0.6865, "lr": 5.62e-05, "router_trainable": true, "sec_per_step": 0.936, "probe_sec": 0.146, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.10146, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": 0.02198, "n_shadow": 3, "mean_delta_shadow": -0.06882, "n_split": 2, "mean_delta_split": -0.19237, "n_merge": 1, "mean_delta_merge": 0.01505, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": -0.4698, "trunc_abs_frac": 1.00686, "mean_delta_full": 0.03644, "read_hit_rate": {"surface": 0.09452, "split": 0.09151, "merge": 0.01017, "skel": 0.07383}, "mean_gate": {"surface": 0.30822, "split": 0.22641, "merge": 0.27723, "skel": 0.30255}, "n_reads": 2228, "n_cand": 31027, "read_suppressed_frac": 0.0956, "dropout_reads_logged": 2228, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 482, "admitted_surface": 4807, "staged_surface": 53789, "admitted_split": 48293, "staged_split": 42677, "admitted_merge": 12181, "staged_merge": 125955, "admitted_skel": 255, "staged_skel": 39723, "mean_count": 128.93148803710938, "n_util_obs": 160, "util_count_spearman": -0.13695652173913042}, "probe_overhead_sec_total": 620.768, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.013884, "alpha": 0.03, "t": 4674.5}
{"event": "consolidate", "step": 4900, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262047, "t": 4703.4}
{"event": "train", "step": 4900, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.39343, "bpb": 2.0103, "chunk_loss": 0.030278, "F": 0.19519, "G": 0.214, "chunks": 199.88, "bytes_per_chunk": 5.123, "boundary_entropy": 0.10468, "router_noise": 0.0, "grad_norm": 0.8299, "lr": 5.41e-05, "router_trainable": true, "sec_per_step": 0.596, "probe_sec": 0.0778, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04815, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": 0.48096, "n_shadow": 2, "mean_delta_shadow": -0.17617, "n_split": 2, "mean_delta_split": -0.034, "n_merge": 1, "mean_delta_merge": -0.56396, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 1.07067, "trunc_abs_frac": 1.07995, "mean_delta_full": 0.09717, "read_hit_rate": {"surface": 0.10465, "split": 0.08395, "merge": 0.01027, "skel": 0.10521}, "mean_gate": {"surface": 0.30765, "split": 0.22659, "merge": 0.28663, "skel": 0.31082}, "n_reads": 2296, "n_cand": 30016, "read_suppressed_frac": 0.1002, "dropout_reads_logged": 2296, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 495, "admitted_surface": 4600, "staged_surface": 53280, "admitted_split": 49248, "staged_split": 41510, "admitted_merge": 11418, "staged_merge": 127966, "admitted_skel": 270, "staged_skel": 39388, "mean_count": 131.1453857421875, "n_util_obs": 167, "util_count_spearman": -0.16154266338235748}, "probe_overhead_sec_total": 624.782, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.014862, "alpha": 0.03, "t": 4704.3}
{"event": "consolidate", "step": 4950, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262048, "t": 4737.4}
{"event": "train", "step": 4950, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.32905, "bpb": 1.91741, "chunk_loss": 0.030445, "F": 0.19733, "G": 0.21554, "chunks": 202.06, "bytes_per_chunk": 5.068, "boundary_entropy": 0.10531, "router_noise": 0.0, "grad_norm": 0.7216, "lr": 5.2e-05, "router_trainable": true, "sec_per_step": 0.684, "probe_sec": 0.1378, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.09947, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": -0.13591, "n_shadow": 8, "mean_delta_shadow": 0.09856, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 1.23392, "trunc_abs_frac": 2.40676, "mean_delta_full": 0.34723, "read_hit_rate": {"surface": 0.09481, "split": 0.08154, "merge": 0.00797, "skel": 0.09719}, "mean_gate": {"surface": 0.29587, "split": 0.21265, "merge": 0.26514, "skel": 0.29065}, "n_reads": 2140, "n_cand": 29999, "read_suppressed_frac": 0.1014, "dropout_reads_logged": 2140, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 500, "admitted_surface": 4598, "staged_surface": 53450, "admitted_split": 50217, "staged_split": 42041, "admitted_merge": 10453, "staged_merge": 127534, "admitted_skel": 268, "staged_skel": 39119, "mean_count": 131.4286651611328, "n_util_obs": 171, "util_count_spearman": -0.14521452541253807}, "probe_overhead_sec_total": 629.364, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.015005, "alpha": 0.03, "t": 4738.5}
{"event": "consolidate", "step": 5000, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262045, "t": 4769.6}
{"event": "probe", "step": 5000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3384.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.194449, "cps": 26.4375, "mean_p": 0.197598, "boundary_entropy": 0.068724, "bpb": 1.86941}, "tr": {"chunks": 3220.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.177871, "cps": 25.15625, "mean_p": 0.217146, "boundary_entropy": 0.204478, "bpb": 2.044192}, "fi": {"chunks": 3019.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.160491, "cps": 23.585938, "mean_p": 0.255492, "boundary_entropy": 0.323229, "bpb": 2.194772}}, "bpb": {"en": 1.86941, "tr": 2.04419, "fi": 2.19477}, "gini_cpc": 0.042489, "gini_cps": 0.025287, "bpic": 5.895459, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9147429817478627, "fi": 0.825362998713217}, "inventory": {"size": 326656, "admitted": 65536, "staged": 261120, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 487, "admitted_surface": 4554, "staged_surface": 53480, "admitted_split": 50885, "staged_split": 42218, "admitted_merge": 9831, "staged_merge": 125945, "admitted_skel": 266, "staged_skel": 39477, "mean_count": 131.1024932861328, "n_util_obs": 172, "util_count_spearman": -0.20227440492614787}, "t": 4771.0}
{"event": "eval", "step": 5000, "phase": 3, "bpb": {"en": 1.88824, "tr": 2.0514, "fi": 2.20256}, "bytes_per_chunk": {"en": 4.864, "tr": 5.833, "fi": 6.353}, "bpb_mean": 2.0474, "frozen_inventory": true, "t": 4773.0}
{"event": "train", "step": 5000, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.43546, "bpb": 2.07094, "chunk_loss": 0.030197, "F": 0.19009, "G": 0.22854, "chunks": 194.66, "bytes_per_chunk": 5.261, "boundary_entropy": 0.14991, "router_noise": 0.0, "grad_norm": 0.8041, "lr": 5e-05, "router_trainable": true, "sec_per_step": 0.775, "probe_sec": 0.0773, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.05002, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": 0.11304, "n_shadow": 7, "mean_delta_shadow": 0.09552, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 8, "trunc_ratio_sum": -1.01372, "trunc_abs_frac": 0.88863, "mean_delta_full": -0.05687, "read_hit_rate": {"surface": 0.08214, "split": 0.06892, "merge": 0.00762, "skel": 0.09635}, "mean_gate": {"surface": 0.30339, "split": 0.20846, "merge": 0.26461, "skel": 0.30221}, "n_reads": 1839, "n_cand": 28996, "read_suppressed_frac": 0.1017, "dropout_reads_logged": 1839, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 486, "admitted_surface": 4554, "staged_surface": 53709, "admitted_split": 50885, "staged_split": 42519, "admitted_merge": 9831, "staged_merge": 126238, "admitted_skel": 266, "staged_skel": 39678, "mean_count": 131.13055419921875, "n_util_obs": 172, "util_count_spearman": -0.20131219297066433}, "probe_overhead_sec_total": 633.332, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.01642, "alpha": 0.03, "t": 4777.3}
{"event": "consolidate", "step": 5050, "retired": 0, "split_retired": 0, "merge_admitted": 1, "admitted": 0, "swapped": 1024, "uncertain_staged": 262042, "t": 4808.0}
{"event": "train", "step": 5050, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.28612, "bpb": 1.85547, "chunk_loss": 0.030249, "F": 0.19559, "G": 0.21701, "chunks": 200.28, "bytes_per_chunk": 5.113, "boundary_entropy": 0.10874, "router_noise": 0.0, "grad_norm": 0.7233, "lr": 4.81e-05, "router_trainable": true, "sec_per_step": 0.622, "probe_sec": 0.0548, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.03667, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": -0.09234, "n_shadow": 0, "mean_delta_shadow": NaN, "n_split": 2, "mean_delta_split": -0.10938, "n_merge": 2, "mean_delta_merge": -0.20158, "n_dropped_unpaired": 0, "n_full": 6, "trunc_ratio_sum": 0.08457, "trunc_abs_frac": 0.3386, "mean_delta_full": -0.31307, "read_hit_rate": {"surface": 0.05959, "split": 0.05653, "merge": 0.00473, "skel": 0.06278}, "mean_gate": {"surface": 0.31223, "split": 0.22469, "merge": 0.27146, "skel": 0.30614}, "n_reads": 1439, "n_cand": 30326, "read_suppressed_frac": 0.0757, "dropout_reads_logged": 1439, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 313, "admitted_surface": 4255, "staged_surface": 53304, "admitted_split": 51546, "staged_split": 41670, "admitted_merge": 9536, "staged_merge": 127849, "admitted_skel": 199, "staged_skel": 39321, "mean_count": 75.78289794921875, "n_util_obs": 126, "util_count_spearman": -0.017904761904761906}, "probe_overhead_sec_total": 637.284, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.015086, "alpha": 0.03, "t": 4808.4}
{"event": "consolidate", "step": 5100, "retired": 0, "split_retired": 3, "merge_admitted": 0, "admitted": 3, "swapped": 1024, "uncertain_staged": 262044, "t": 4835.3}
{"event": "train", "step": 5100, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.33832, "bpb": 1.93079, "chunk_loss": 0.030175, "F": 0.19058, "G": 0.21166, "chunks": 195.16, "bytes_per_chunk": 5.247, "boundary_entropy": 0.11043, "router_noise": 0.0, "grad_norm": 0.7164, "lr": 4.63e-05, "router_trainable": true, "sec_per_step": 0.545, "probe_sec": 0.0613, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04232, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": -0.04526, "n_shadow": 4, "mean_delta_shadow": 0.24004, "n_split": 1, "mean_delta_split": -0.24921, "n_merge": 1, "mean_delta_merge": 0.05469, "n_dropped_unpaired": 0, "n_full": 1, "trunc_ratio_sum": 0.07243, "trunc_abs_frac": 0.07243, "mean_delta_full": -0.69312, "read_hit_rate": {"surface": 0.06261, "split": 0.05199, "merge": 0.00227, "skel": 0.0493}, "mean_gate": {"surface": 0.29888, "split": 0.22864, "merge": 0.25499, "skel": 0.30537}, "n_reads": 1277, "n_cand": 29475, "read_suppressed_frac": 0.0885, "dropout_reads_logged": 1277, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 322, "admitted_surface": 4273, "staged_surface": 53904, "admitted_split": 52458, "staged_split": 42276, "admitted_merge": 8604, "staged_merge": 126328, "admitted_skel": 201, "staged_skel": 39636, "mean_count": 75.72434997558594, "n_util_obs": 129, "util_count_spearman": -0.000779853504939135}, "probe_overhead_sec_total": 641.021, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.015863, "alpha": 0.03, "t": 4835.6}
{"event": "consolidate", "step": 5150, "retired": 0, "split_retired": 2, "merge_admitted": 0, "admitted": 2, "swapped": 1024, "uncertain_staged": 262040, "t": 4874.4}
{"event": "train", "step": 5150, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.37521, "bpb": 1.98401, "chunk_loss": 0.030374, "F": 0.19736, "G": 0.2164, "chunks": 202.09, "bytes_per_chunk": 5.067, "boundary_entropy": 0.11069, "router_noise": 0.0, "grad_norm": 0.7824, "lr": 4.45e-05, "router_trainable": true, "sec_per_step": 0.796, "probe_sec": 0.0961, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.05684, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": 0.01493, "n_shadow": 3, "mean_delta_shadow": 0.086, "n_split": 3, "mean_delta_split": 0.00318, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 1.51786, "trunc_abs_frac": 0.95002, "mean_delta_full": -0.25256, "read_hit_rate": {"surface": 0.06915, "split": 0.04979, "merge": 0.00453, "skel": 0.06837}, "mean_gate": {"surface": 0.26987, "split": 0.22719, "merge": 0.2761, "skel": 0.29556}, "n_reads": 1422, "n_cand": 29946, "read_suppressed_frac": 0.1104, "dropout_reads_logged": 1422, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 320, "admitted_surface": 4280, "staged_surface": 54058, "admitted_split": 53428, "staged_split": 42135, "admitted_merge": 7614, "staged_merge": 126430, "admitted_skel": 214, "staged_skel": 39521, "mean_count": 76.03240966796875, "n_util_obs": 130, "util_count_spearman": 0.03085389668027694}, "probe_overhead_sec_total": 644.99, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.017643, "alpha": 0.03, "t": 4875.4}
{"event": "consolidate", "step": 5200, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262040, "t": 4907.6}
{"event": "probe", "step": 5200, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3388.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.194679, "cps": 26.46875, "mean_p": 0.197765, "boundary_entropy": 0.068294, "bpb": 1.866833}, "tr": {"chunks": 3153.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.17417, "cps": 24.632812, "mean_p": 0.217642, "boundary_entropy": 0.206951, "bpb": 2.036681}, "fi": {"chunks": 3000.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.159481, "cps": 23.4375, "mean_p": 0.254249, "boundary_entropy": 0.320513, "bpb": 2.188488}}, "bpb": {"en": 1.86683, "tr": 2.03668, "fi": 2.18849}, "gini_cpc": 0.044414, "gini_cps": 0.027111, "bpic": 5.946127, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.8946520014914003, "fi": 0.8192002757561593}, "inventory": {"size": 326656, "admitted": 65536, "staged": 261120, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 319, "admitted_surface": 4363, "staged_surface": 53922, "admitted_split": 54291, "staged_split": 42163, "admitted_merge": 6618, "staged_merge": 125541, "admitted_skel": 264, "staged_skel": 39494, "mean_count": 85.15975952148438, "n_util_obs": 139, "util_count_spearman": 0.06629883629616307}, "t": 4908.9}
{"event": "eval", "step": 5200, "phase": 3, "bpb": {"en": 1.8854, "tr": 2.04391, "fi": 2.19759}, "bytes_per_chunk": {"en": 4.863, "tr": 5.923, "fi": 6.325}, "bpb_mean": 2.0423, "frozen_inventory": true, "t": 4910.9}
{"event": "train", "step": 5200, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.31862, "bpb": 1.90237, "chunk_loss": 0.030295, "F": 0.19638, "G": 0.22161, "chunks": 201.09, "bytes_per_chunk": 5.092, "boundary_entropy": 0.11768, "router_noise": 0.0, "grad_norm": 0.7218, "lr": 4.29e-05, "router_trainable": true, "sec_per_step": 0.785, "probe_sec": 0.0546, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.03822, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 11, "mean_delta_loo": -0.01276, "n_shadow": 3, "mean_delta_shadow": 0.16327, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": -9.08438, "trunc_abs_frac": 0.55371, "mean_delta_full": -0.01941, "read_hit_rate": {"surface": 0.12557, "split": 0.13818, "merge": 0.00487, "skel": 0.13886}, "mean_gate": {"surface": 0.27733, "split": 0.21093, "merge": 0.23738, "skel": 0.28497}, "n_reads": 3267, "n_cand": 30348, "read_suppressed_frac": 0.1004, "dropout_reads_logged": 3267, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 319, "admitted_surface": 4363, "staged_surface": 54051, "admitted_split": 54291, "staged_split": 42281, "admitted_merge": 6618, "staged_merge": 126237, "admitted_skel": 264, "staged_skel": 39575, "mean_count": 85.20960998535156, "n_util_obs": 139, "util_count_spearman": 0.06907815362616739}, "probe_overhead_sec_total": 648.85, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.01667, "alpha": 0.03, "t": 4914.7}
{"event": "consolidate", "step": 5250, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262040, "t": 4945.6}
{"event": "train", "step": 5250, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.30936, "bpb": 1.889, "chunk_loss": 0.030298, "F": 0.19623, "G": 0.21885, "chunks": 200.94, "bytes_per_chunk": 5.096, "boundary_entropy": 0.11627, "router_noise": 0.0, "grad_norm": 0.7376, "lr": 4.14e-05, "router_trainable": true, "sec_per_step": 0.627, "probe_sec": 0.0696, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.05187, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": 0.03447, "n_shadow": 4, "mean_delta_shadow": -2e-05, "n_split": 1, "mean_delta_split": -0.01132, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 6, "trunc_ratio_sum": 0.56998, "trunc_abs_frac": 0.38918, "mean_delta_full": 0.10372, "read_hit_rate": {"surface": 0.1241, "split": 0.10334, "merge": 0.00141, "skel": 0.13233}, "mean_gate": {"surface": 0.24106, "split": 0.17927, "merge": 0.20095, "skel": 0.24528}, "n_reads": 2742, "n_cand": 30077, "read_suppressed_frac": 0.1131, "dropout_reads_logged": 2742, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 343, "admitted_surface": 4347, "staged_surface": 53803, "admitted_split": 54392, "staged_split": 41814, "admitted_merge": 6546, "staged_merge": 127018, "admitted_skel": 251, "staged_skel": 39509, "mean_count": 82.96726989746094, "n_util_obs": 130, "util_count_spearman": -0.049638804298843354}, "probe_overhead_sec_total": 652.85, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.016736, "alpha": 0.03, "t": 4946.0}
{"event": "consolidate", "step": 5300, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262037, "t": 4972.4}
{"event": "train", "step": 5300, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.31507, "bpb": 1.89724, "chunk_loss": 0.030282, "F": 0.19791, "G": 0.21298, "chunks": 202.66, "bytes_per_chunk": 5.053, "boundary_entropy": 0.09548, "router_noise": 0.0, "grad_norm": 0.6912, "lr": 3.99e-05, "router_trainable": true, "sec_per_step": 0.534, "probe_sec": 0.0758, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.05552, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 12, "mean_delta_loo": 0.32877, "n_shadow": 2, "mean_delta_shadow": 0.00562, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 0.33161, "trunc_abs_frac": 0.72409, "mean_delta_full": 0.20142, "read_hit_rate": {"surface": 0.01178, "split": 0.01539, "merge": 0.0014, "skel": 0.03038}, "mean_gate": {"surface": 0.23957, "split": 0.17761, "merge": 0.20254, "skel": 0.26338}, "n_reads": 440, "n_cand": 30616, "read_suppressed_frac": 0.0795, "dropout_reads_logged": 440, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 95, "admitted_surface": 4210, "staged_surface": 54092, "admitted_split": 54539, "staged_split": 42648, "admitted_merge": 6607, "staged_merge": 125710, "admitted_skel": 180, "staged_skel": 39694, "mean_count": 14.0103759765625, "n_util_obs": 53, "util_count_spearman": -0.13820351556200614}, "probe_overhead_sec_total": 656.683, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.011638, "alpha": 0.03, "t": 4972.7}
{"event": "consolidate", "step": 5350, "retired": 0, "split_retired": 1, "merge_admitted": 0, "admitted": 1, "swapped": 1024, "uncertain_staged": 262033, "t": 5008.0}
{"event": "train", "step": 5350, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.31182, "bpb": 1.89255, "chunk_loss": 0.030292, "F": 0.19693, "G": 0.21159, "chunks": 201.66, "bytes_per_chunk": 5.078, "boundary_entropy": 0.09604, "router_noise": 0.0, "grad_norm": 0.7266, "lr": 3.86e-05, "router_trainable": true, "sec_per_step": 0.717, "probe_sec": 0.0842, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.05945, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": 0.03596, "n_shadow": 7, "mean_delta_shadow": 0.25911, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 0.68221, "trunc_abs_frac": 0.91117, "mean_delta_full": 0.25977, "read_hit_rate": {"surface": 0.00685, "split": 0.00411, "merge": 0.0011, "skel": 0.01603}, "mean_gate": {"surface": 0.27605, "split": 0.17796, "merge": 0.20401, "skel": 0.28938}, "n_reads": 186, "n_cand": 30293, "read_suppressed_frac": 0.0484, "dropout_reads_logged": 186, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 98, "admitted_surface": 4531, "staged_surface": 53893, "admitted_split": 54123, "staged_split": 42811, "admitted_merge": 6697, "staged_merge": 125765, "admitted_skel": 185, "staged_skel": 39675, "mean_count": 10.178268432617188, "n_util_obs": 41, "util_count_spearman": -0.14041811846689894}, "probe_overhead_sec_total": 660.753, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.009824, "alpha": 0.03, "t": 5008.5}
{"event": "consolidate", "step": 5400, "retired": 0, "split_retired": 2, "merge_admitted": 0, "admitted": 2, "swapped": 1024, "uncertain_staged": 262031, "t": 5042.2}
{"event": "probe", "step": 5400, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3384.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.194449, "cps": 26.4375, "mean_p": 0.197491, "boundary_entropy": 0.068857, "bpb": 1.866568}, "tr": {"chunks": 3148.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.173894, "cps": 24.59375, "mean_p": 0.218708, "boundary_entropy": 0.210827, "bpb": 2.033917}, "fi": {"chunks": 3029.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.161023, "cps": 23.664062, "mean_p": 0.26435, "boundary_entropy": 0.336172, "bpb": 2.180924}}, "bpb": {"en": 1.86657, "tr": 2.03392, "fi": 2.18092}, "gini_cpc": 0.042096, "gini_cps": 0.024753, "bpic": 5.933689, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.8942891014106434, "fi": 0.828096894038534}, "inventory": {"size": 326654, "admitted": 65536, "staged": 261118, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 106, "admitted_surface": 4852, "staged_surface": 53731, "admitted_split": 53749, "staged_split": 42330, "admitted_merge": 6746, "staged_merge": 125662, "admitted_skel": 189, "staged_skel": 39395, "mean_count": 9.413589477539062, "n_util_obs": 38, "util_count_spearman": -0.06663748769011926}, "t": 5043.4}
{"event": "eval", "step": 5400, "phase": 3, "bpb": {"en": 1.88442, "tr": 2.04149, "fi": 2.18904}, "bytes_per_chunk": {"en": 4.867, "tr": 5.92, "fi": 6.21}, "bpb_mean": 2.03831, "frozen_inventory": true, "t": 5045.4}
{"event": "train", "step": 5400, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.38171, "bpb": 1.99338, "chunk_loss": 0.030245, "F": 0.19107, "G": 0.22249, "chunks": 195.66, "bytes_per_chunk": 5.234, "boundary_entropy": 0.13416, "router_noise": 0.0, "grad_norm": 0.8891, "lr": 3.73e-05, "router_trainable": true, "sec_per_step": 0.81, "probe_sec": 0.0617, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04385, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 5, "mean_delta_loo": 0.23659, "n_shadow": 5, "mean_delta_shadow": 0.08962, "n_split": 0, "mean_delta_split": NaN, "n_merge": 2, "mean_delta_merge": -0.09344, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 0.37811, "trunc_abs_frac": 1.7554, "mean_delta_full": -0.07694, "read_hit_rate": {"surface": 0.00321, "split": 0.00266, "merge": 0.00145, "skel": 0.0197}, "mean_gate": {"surface": 0.31013, "split": 0.1752, "merge": 0.22943, "skel": 0.29907}, "n_reads": 157, "n_cand": 28999, "read_suppressed_frac": 0.1592, "dropout_reads_logged": 157, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 108, "admitted_surface": 4852, "staged_surface": 53895, "admitted_split": 53749, "staged_split": 42529, "admitted_merge": 6746, "staged_merge": 126238, "admitted_skel": 189, "staged_skel": 39482, "mean_count": 9.415985107421875, "n_util_obs": 38, "util_count_spearman": -0.08480140059087428}, "probe_overhead_sec_total": 664.609, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.010084, "alpha": 0.03, "t": 5049.1}
{"event": "consolidate", "step": 5450, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262030, "t": 5077.5}
{"event": "train", "step": 5450, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.31857, "bpb": 1.90229, "chunk_loss": 0.030292, "F": 0.1969, "G": 0.21206, "chunks": 201.62, "bytes_per_chunk": 5.079, "boundary_entropy": 0.09436, "router_noise": 0.0, "grad_norm": 0.7614, "lr": 3.62e-05, "router_trainable": true, "sec_per_step": 0.583, "probe_sec": 0.4391, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.0382, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 5, "mean_delta_loo": -0.00832, "n_shadow": 7, "mean_delta_shadow": 0.02788, "n_split": 0, "mean_delta_split": NaN, "n_merge": 1, "mean_delta_merge": 0.19476, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": -0.15916, "trunc_abs_frac": 0.44465, "mean_delta_full": -0.21602, "read_hit_rate": {"surface": 0.22835, "split": 0.05882, "merge": 0.00063, "skel": 0.19838}, "mean_gate": {"surface": 0.24679, "split": 0.19151, "merge": 0.23067, "skel": 0.25511}, "n_reads": 3237, "n_cand": 30267, "read_suppressed_frac": 0.0976, "dropout_reads_logged": 3237, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 111, "admitted_surface": 5535, "staged_surface": 53835, "admitted_split": 53779, "staged_split": 43468, "admitted_merge": 5839, "staged_merge": 125468, "admitted_skel": 383, "staged_skel": 39373, "mean_count": 17.024795532226562, "n_util_obs": 40, "util_count_spearman": -0.20393996247654783}, "probe_overhead_sec_total": 669.03, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.009174, "alpha": 0.03, "t": 5078.2}
{"event": "consolidate", "step": 5500, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262029, "t": 5105.1}
{"event": "train", "step": 5500, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.31422, "bpb": 1.89602, "chunk_loss": 0.030381, "F": 0.20282, "G": 0.21676, "chunks": 207.69, "bytes_per_chunk": 4.93, "boundary_entropy": 0.09291, "router_noise": 0.0, "grad_norm": 0.7332, "lr": 3.51e-05, "router_trainable": true, "sec_per_step": 0.545, "probe_sec": 0.0646, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04343, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": -0.04389, "n_shadow": 6, "mean_delta_shadow": 0.08311, "n_split": 1, "mean_delta_split": -0.19196, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 0.38626, "trunc_abs_frac": 0.4035, "mean_delta_full": -0.28357, "read_hit_rate": {"surface": 0.32582, "split": 0.24794, "merge": 0.00137, "skel": 0.29846}, "mean_gate": {"surface": 0.20821, "split": 0.16094, "merge": 0.17567, "skel": 0.21532}, "n_reads": 6925, "n_cand": 31284, "read_suppressed_frac": 0.097, "dropout_reads_logged": 6925, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 156, "admitted_surface": 5856, "staged_surface": 53263, "admitted_split": 54325, "staged_split": 42671, "admitted_merge": 4834, "staged_merge": 127150, "admitted_skel": 521, "staged_skel": 39060, "mean_count": 49.622650146484375, "n_util_obs": 55, "util_count_spearman": -0.5753968253968254}, "probe_overhead_sec_total": 673.814, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.011721, "alpha": 0.03, "t": 5105.5}
{"event": "consolidate", "step": 5550, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 29, "uncertain_staged": 262038, "t": 5140.7}
{"event": "train", "step": 5550, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.30765, "bpb": 1.88654, "chunk_loss": 0.030282, "F": 0.19275, "G": 0.22108, "chunks": 197.38, "bytes_per_chunk": 5.188, "boundary_entropy": 0.12294, "router_noise": 0.0, "grad_norm": 0.7092, "lr": 3.41e-05, "router_trainable": true, "sec_per_step": 0.711, "probe_sec": 0.062, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04105, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": -0.02142, "n_shadow": 4, "mean_delta_shadow": -0.05685, "n_split": 1, "mean_delta_split": 0.30261, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": -0.18391, "trunc_abs_frac": 1.22604, "mean_delta_full": 0.03989, "read_hit_rate": {"surface": 0.32368, "split": 0.24531, "merge": 0.00096, "skel": 0.28148}, "mean_gate": {"surface": 0.17816, "split": 0.13918, "merge": 0.16308, "skel": 0.18123}, "n_reads": 6404, "n_cand": 29559, "read_suppressed_frac": 0.101, "dropout_reads_logged": 6404, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 202, "admitted_surface": 5867, "staged_surface": 52803, "admitted_split": 54308, "staged_split": 42210, "admitted_merge": 4834, "staged_merge": 128243, "admitted_skel": 527, "staged_skel": 38888, "mean_count": 62.09495544433594, "n_util_obs": 65, "util_count_spearman": -0.6454545454545455}, "probe_overhead_sec_total": 677.904, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.013066, "alpha": 0.03, "t": 5141.0}
{"event": "consolidate", "step": 5600, "retired": 0, "split_retired": 0, "merge_admitted": 1, "admitted": 0, "swapped": 205, "uncertain_staged": 262043, "t": 5173.2}
{"event": "probe", "step": 5600, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3392.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.194909, "cps": 26.5, "mean_p": 0.19835, "boundary_entropy": 0.066247, "bpb": 1.852827}, "tr": {"chunks": 3182.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.175772, "cps": 24.859375, "mean_p": 0.218609, "boundary_entropy": 0.207696, "bpb": 2.026091}, "fi": {"chunks": 3099.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.164744, "cps": 24.210938, "mean_p": 0.260601, "boundary_entropy": 0.326433, "bpb": 2.171835}}, "bpb": {"en": 1.85283, "tr": 2.02609, "fi": 2.17183}, "gini_cpc": 0.037559, "gini_cps": 0.020194, "bpic": 5.864985, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.901815925693542, "fi": 0.8452359675390654}, "inventory": {"size": 327474, "admitted": 65536, "staged": 261938, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 214, "admitted_surface": 6015, "staged_surface": 53060, "admitted_split": 54154, "staged_split": 42949, "admitted_merge": 4840, "staged_merge": 126736, "admitted_skel": 527, "staged_skel": 39193, "mean_count": 68.64108276367188, "n_util_obs": 72, "util_count_spearman": -0.5889124702553219}, "t": 5174.4}
{"event": "eval", "step": 5600, "phase": 3, "bpb": {"en": 1.87087, "tr": 2.03344, "fi": 2.18011}, "bytes_per_chunk": {"en": 4.859, "tr": 5.923, "fi": 6.167}, "bpb_mean": 2.02814, "frozen_inventory": true, "t": 5176.5}
{"event": "train", "step": 5600, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.37053, "bpb": 1.97726, "chunk_loss": 0.030233, "F": 0.19156, "G": 0.21655, "chunks": 196.16, "bytes_per_chunk": 5.22, "boundary_entropy": 0.11756, "router_noise": 0.0, "grad_norm": 0.7594, "lr": 3.33e-05, "router_trainable": true, "sec_per_step": 0.787, "probe_sec": 0.0597, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.0432, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": 0.04982, "n_shadow": 4, "mean_delta_shadow": 0.05434, "n_split": 0, "mean_delta_split": NaN, "n_merge": 1, "mean_delta_merge": -0.2807, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": 1.17863, "trunc_abs_frac": 0.55402, "mean_delta_full": -0.04182, "read_hit_rate": {"surface": 0.33835, "split": 0.24785, "merge": 0.0008, "skel": 0.27431}, "mean_gate": {"surface": 0.17358, "split": 0.13577, "merge": 0.14286, "skel": 0.17723}, "n_reads": 6457, "n_cand": 29419, "read_suppressed_frac": 0.1028, "dropout_reads_logged": 6457, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 213, "admitted_surface": 6015, "staged_surface": 53046, "admitted_split": 54154, "staged_split": 42966, "admitted_merge": 4840, "staged_merge": 126943, "admitted_skel": 527, "staged_skel": 39189, "mean_count": 68.73960876464844, "n_util_obs": 72, "util_count_spearman": -0.5923853624027269}, "probe_overhead_sec_total": 681.777, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.013261, "alpha": 0.03, "t": 5180.4}
{"event": "consolidate", "step": 5650, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262044, "t": 5208.5}
{"event": "train", "step": 5650, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.3145, "bpb": 1.89642, "chunk_loss": 0.030198, "F": 0.19064, "G": 0.22034, "chunks": 195.22, "bytes_per_chunk": 5.245, "boundary_entropy": 0.13223, "router_noise": 0.0, "grad_norm": 0.7387, "lr": 3.25e-05, "router_trainable": true, "sec_per_step": 0.571, "probe_sec": 0.0783, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.05123, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 3, "mean_delta_loo": 0.1043, "n_shadow": 7, "mean_delta_shadow": -0.03016, "n_split": 2, "mean_delta_split": 0.01884, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 1, "trunc_ratio_sum": 1.0, "trunc_abs_frac": 1.0, "mean_delta_full": 0.01282, "read_hit_rate": {"surface": 0.32615, "split": 0.24524, "merge": 0.00049, "skel": 0.25964}, "mean_gate": {"surface": 0.16949, "split": 0.13482, "merge": 0.14105, "skel": 0.17143}, "n_reads": 6247, "n_cand": 29292, "read_suppressed_frac": 0.097, "dropout_reads_logged": 6247, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 234, "admitted_surface": 6904, "staged_surface": 52660, "admitted_split": 53238, "staged_split": 42871, "admitted_merge": 4890, "staged_merge": 127238, "admitted_skel": 504, "staged_skel": 39375, "mean_count": 72.02908325195312, "n_util_obs": 74, "util_count_spearman": -0.5536212782179039}, "probe_overhead_sec_total": 685.582, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.012605, "alpha": 0.03, "t": 5208.9}
{"event": "consolidate", "step": 5700, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262042, "t": 5234.5}
{"event": "train", "step": 5700, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.44832, "bpb": 2.08949, "chunk_loss": 0.030227, "F": 0.19019, "G": 0.22382, "chunks": 194.75, "bytes_per_chunk": 5.258, "boundary_entropy": 0.14147, "router_noise": 0.0, "grad_norm": 0.8489, "lr": 3.18e-05, "router_trainable": true, "sec_per_step": 0.52, "probe_sec": 0.0625, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04263, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": 0.05873, "n_shadow": 4, "mean_delta_shadow": 0.02373, "n_split": 1, "mean_delta_split": -0.0457, "n_merge": 1, "mean_delta_merge": 0.05811, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 0.03409, "trunc_abs_frac": 0.16625, "mean_delta_full": -0.54584, "read_hit_rate": {"surface": 0.32113, "split": 0.23483, "merge": 0.00113, "skel": 0.27076}, "mean_gate": {"surface": 0.16593, "split": 0.13184, "merge": 0.1488, "skel": 0.16956}, "n_reads": 6064, "n_cand": 28920, "read_suppressed_frac": 0.1026, "dropout_reads_logged": 6064, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 249, "admitted_surface": 7860, "staged_surface": 52412, "admitted_split": 52246, "staged_split": 42844, "admitted_merge": 4937, "staged_merge": 127393, "admitted_skel": 493, "staged_skel": 39495, "mean_count": 76.97283935546875, "n_util_obs": 81, "util_count_spearman": -0.44496386630532975}, "probe_overhead_sec_total": 689.566, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.012518, "alpha": 0.03, "t": 5234.9}
{"event": "consolidate", "step": 5750, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262044, "t": 5270.8}
{"event": "train", "step": 5750, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.27397, "bpb": 1.83795, "chunk_loss": 0.030381, "F": 0.20172, "G": 0.21726, "chunks": 206.56, "bytes_per_chunk": 4.957, "boundary_entropy": 0.09623, "router_noise": 0.0, "grad_norm": 0.7476, "lr": 3.13e-05, "router_trainable": true, "sec_per_step": 0.726, "probe_sec": 0.0685, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.0456, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": -0.04021, "n_shadow": 4, "mean_delta_shadow": 0.12358, "n_split": 1, "mean_delta_split": -0.12033, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 0.66242, "trunc_abs_frac": 0.87911, "mean_delta_full": 0.29105, "read_hit_rate": {"surface": 0.35452, "split": 0.25599, "merge": 0.00107, "skel": 0.2742}, "mean_gate": {"surface": 0.17409, "split": 0.13585, "merge": 0.14751, "skel": 0.17557}, "n_reads": 6962, "n_cand": 30805, "read_suppressed_frac": 0.1041, "dropout_reads_logged": 6962, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 292, "admitted_surface": 8795, "staged_surface": 52539, "admitted_split": 51290, "staged_split": 43212, "admitted_merge": 5006, "staged_merge": 126705, "admitted_skel": 445, "staged_skel": 39688, "mean_count": 81.05435180664062, "n_util_obs": 85, "util_count_spearman": -0.43441469611100253}, "probe_overhead_sec_total": 693.753, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.012294, "alpha": 0.03, "t": 5271.2}
{"event": "consolidate", "step": 5800, "retired": 0, "split_retired": 1, "merge_admitted": 0, "admitted": 1, "swapped": 1024, "uncertain_staged": 262043, "t": 5303.7}
{"event": "probe", "step": 5800, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3390.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.194794, "cps": 26.484375, "mean_p": 0.19818, "boundary_entropy": 0.066018, "bpb": 1.8465}, "tr": {"chunks": 3215.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.177595, "cps": 25.117188, "mean_p": 0.219761, "boundary_entropy": 0.207561, "bpb": 2.019119}, "fi": {"chunks": 3159.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.167934, "cps": 24.679688, "mean_p": 0.260087, "boundary_entropy": 0.323673, "bpb": 2.166582}}, "bpb": {"en": 1.8465, "tr": 2.01912, "fi": 2.16658}, "gini_cpc": 0.033141, "gini_cps": 0.015772, "bpic": 5.810324, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9117060732612156, "fi": 0.8621089712618721}, "inventory": {"size": 326655, "admitted": 65536, "staged": 261119, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 304, "admitted_surface": 9700, "staged_surface": 51894, "admitted_split": 50324, "staged_split": 42390, "admitted_merge": 5103, "staged_merge": 127816, "admitted_skel": 409, "staged_skel": 39019, "mean_count": 82.35659790039062, "n_util_obs": 88, "util_count_spearman": -0.45827903208763343}, "t": 5305.2}
{"event": "eval", "step": 5800, "phase": 3, "bpb": {"en": 1.86496, "tr": 2.0268, "fi": 2.17499}, "bytes_per_chunk": {"en": 4.859, "tr": 5.839, "fi": 6.085}, "bpb_mean": 2.02225, "frozen_inventory": true, "t": 5307.2}
{"event": "train", "step": 5800, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.25963, "bpb": 1.81726, "chunk_loss": 0.030269, "F": 0.19504, "G": 0.21484, "chunks": 199.72, "bytes_per_chunk": 5.127, "boundary_entropy": 0.10311, "router_noise": 0.0, "grad_norm": 0.7561, "lr": 3.08e-05, "router_trainable": true, "sec_per_step": 0.796, "probe_sec": 0.1103, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.08439, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 5, "mean_delta_loo": -0.03616, "n_shadow": 5, "mean_delta_shadow": -0.17709, "n_split": 2, "mean_delta_split": 0.13464, "n_merge": 2, "mean_delta_merge": 0.25007, "n_dropped_unpaired": 0, "n_full": 1, "trunc_ratio_sum": 0.34116, "trunc_abs_frac": 0.34116, "mean_delta_full": -0.11951, "read_hit_rate": {"surface": 0.33889, "split": 0.25666, "merge": 0.00047, "skel": 0.24462}, "mean_gate": {"surface": 0.17666, "split": 0.1341, "merge": 0.15126, "skel": 0.17805}, "n_reads": 6589, "n_cand": 30201, "read_suppressed_frac": 0.1014, "dropout_reads_logged": 6589, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 306, "admitted_surface": 9700, "staged_surface": 52029, "admitted_split": 50324, "staged_split": 42452, "admitted_merge": 5103, "staged_merge": 128562, "admitted_skel": 409, "staged_skel": 39101, "mean_count": 82.45713806152344, "n_util_obs": 88, "util_count_spearman": -0.45827903208763343}, "probe_overhead_sec_total": 697.595, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.012557, "alpha": 0.03, "t": 5311.0}
{"event": "consolidate", "step": 5850, "retired": 0, "split_retired": 0, "merge_admitted": 1, "admitted": 0, "swapped": 1024, "uncertain_staged": 262041, "t": 5335.8}
{"event": "train", "step": 5850, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.37933, "bpb": 1.98996, "chunk_loss": 0.030259, "F": 0.1915, "G": 0.23001, "chunks": 196.09, "bytes_per_chunk": 5.222, "boundary_entropy": 0.15324, "router_noise": 0.0, "grad_norm": 0.8191, "lr": 3.05e-05, "router_trainable": true, "sec_per_step": 0.513, "probe_sec": 0.0655, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.0435, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": -0.14534, "n_shadow": 3, "mean_delta_shadow": -0.19103, "n_split": 2, "mean_delta_split": -0.05025, "n_merge": 1, "mean_delta_merge": 0.06561, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": -0.0745, "trunc_abs_frac": 0.30628, "mean_delta_full": 0.13507, "read_hit_rate": {"surface": 0.32356, "split": 0.22187, "merge": 0.00064, "skel": 0.23848}, "mean_gate": {"surface": 0.17547, "split": 0.13424, "merge": 0.1481, "skel": 0.17748}, "n_reads": 5813, "n_cand": 29158, "read_suppressed_frac": 0.1003, "dropout_reads_logged": 5813, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 340, "admitted_surface": 10655, "staged_surface": 52340, "admitted_split": 49322, "staged_split": 43195, "admitted_merge": 5171, "staged_merge": 127086, "admitted_skel": 388, "staged_skel": 39523, "mean_count": 86.630859375, "n_util_obs": 92, "util_count_spearman": -0.41562504816361756}, "probe_overhead_sec_total": 701.574, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.012253, "alpha": 0.03, "t": 5336.6}
{"event": "consolidate", "step": 5900, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262042, "t": 5362.2}
{"event": "train", "step": 5900, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.35851, "bpb": 1.95992, "chunk_loss": 0.030334, "F": 0.19727, "G": 0.22001, "chunks": 202.0, "bytes_per_chunk": 5.069, "boundary_entropy": 0.11469, "router_noise": 0.0, "grad_norm": 0.745, "lr": 3.02e-05, "router_trainable": true, "sec_per_step": 0.527, "probe_sec": 0.0732, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.05188, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 5, "mean_delta_loo": 0.0394, "n_shadow": 4, "mean_delta_shadow": 0.02968, "n_split": 1, "mean_delta_split": -0.08264, "n_merge": 2, "mean_delta_merge": 0.05759, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": 0.78383, "trunc_abs_frac": 0.72223, "mean_delta_full": 0.03699, "read_hit_rate": {"surface": 0.33629, "split": 0.24141, "merge": 0.00109, "skel": 0.23622}, "mean_gate": {"surface": 0.17783, "split": 0.13703, "merge": 0.15417, "skel": 0.17791}, "n_reads": 6323, "n_cand": 30148, "read_suppressed_frac": 0.1045, "dropout_reads_logged": 6323, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 360, "admitted_surface": 11627, "staged_surface": 51945, "admitted_split": 48310, "staged_split": 43330, "admitted_merge": 5221, "staged_merge": 127251, "admitted_skel": 378, "staged_skel": 39618, "mean_count": 91.57923889160156, "n_util_obs": 97, "util_count_spearman": -0.31723385230380813}, "probe_overhead_sec_total": 705.189, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.011348, "alpha": 0.03, "t": 5363.0}
{"event": "consolidate", "step": 5950, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262041, "t": 5399.9}
{"event": "train", "step": 5950, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.30581, "bpb": 1.88388, "chunk_loss": 0.030163, "F": 0.18686, "G": 0.22465, "chunks": 191.34, "bytes_per_chunk": 5.352, "boundary_entropy": 0.14708, "router_noise": 0.0, "grad_norm": 0.744, "lr": 3.01e-05, "router_trainable": true, "sec_per_step": 0.748, "probe_sec": 0.0804, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.05494, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": 0.07506, "n_shadow": 4, "mean_delta_shadow": 0.00478, "n_split": 2, "mean_delta_split": -0.0499, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": -0.01605, "trunc_abs_frac": 1.24681, "mean_delta_full": 0.18636, "read_hit_rate": {"surface": 0.33574, "split": 0.22673, "merge": 0.00116, "skel": 0.23535}, "mean_gate": {"surface": 0.18284, "split": 0.1384, "merge": 0.14834, "skel": 0.18062}, "n_reads": 5879, "n_cand": 28835, "read_suppressed_frac": 0.1002, "dropout_reads_logged": 5879, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 394, "admitted_surface": 12527, "staged_surface": 51718, "admitted_split": 47294, "staged_split": 43439, "admitted_merge": 5344, "staged_merge": 127343, "admitted_skel": 371, "staged_skel": 39644, "mean_count": 96.30575561523438, "n_util_obs": 103, "util_count_spearman": -0.2791811272679348}, "probe_overhead_sec_total": 709.617, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.011152, "alpha": 0.03, "t": 5400.4}
{"event": "consolidate", "step": 6000, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262040, "t": 5434.2}
{"event": "train", "step": 5999, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.3528, "bpb": 1.95167, "chunk_loss": 0.030326, "F": 0.19846, "G": 0.21286, "chunks": 203.22, "bytes_per_chunk": 5.039, "boundary_entropy": 0.09363, "router_noise": 0.0, "grad_norm": 0.7276, "lr": 3e-05, "router_trainable": true, "sec_per_step": 0.676, "probe_sec": 0.0574, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.03904, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 5, "mean_delta_loo": 0.15086, "n_shadow": 7, "mean_delta_shadow": 0.05771, "n_split": 1, "mean_delta_split": -0.02203, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 0.89175, "trunc_abs_frac": 0.76532, "mean_delta_full": 0.15059, "read_hit_rate": {"surface": 0.34832, "split": 0.24731, "merge": 0.00109, "skel": 0.22921}, "mean_gate": {"surface": 0.18331, "split": 0.13963, "merge": 0.1523, "skel": 0.18453}, "n_reads": 6523, "n_cand": 30536, "read_suppressed_frac": 0.0984, "dropout_reads_logged": 6523, "inventory": {"size": 326656, "admitted": 65536, "staged": 261120, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 413, "admitted_surface": 13378, "staged_surface": 51220, "admitted_split": 46283, "staged_split": 43346, "admitted_merge": 5514, "staged_merge": 126988, "admitted_skel": 361, "staged_skel": 39566, "mean_count": 102.24116516113281, "n_util_obs": 114, "util_count_spearman": -0.23777971284502136}, "probe_overhead_sec_total": 713.634, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.011542, "alpha": 0.03, "t": 5434.2}
{"event": "probe", "step": 6000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3391.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.194851, "cps": 26.492188, "mean_p": 0.197999, "boundary_entropy": 0.06615, "bpb": 1.844058}, "tr": {"chunks": 3173.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.175275, "cps": 24.789062, "mean_p": 0.2185, "boundary_entropy": 0.20678, "bpb": 2.01425}, "fi": {"chunks": 3029.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.161023, "cps": 23.664062, "mean_p": 0.260312, "boundary_entropy": 0.327566, "bpb": 2.160345}}, "bpb": {"en": 1.84406, "tr": 2.01425, "fi": 2.16035}, "gini_cpc": 0.04246, "gini_cps": 0.025157, "bpic": 5.913896, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.8995304124358154, "fi": 0.8263874637058093}, "inventory": {"size": 326656, "admitted": 65536, "staged": 261120, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 413, "admitted_surface": 13378, "staged_surface": 51220, "admitted_split": 46283, "staged_split": 43346, "admitted_merge": 5514, "staged_merge": 126988, "admitted_skel": 361, "staged_skel": 39566, "mean_count": 102.24116516113281, "n_util_obs": 114, "util_count_spearman": -0.23777971284502136}, "t": 5435.4}
{"event": "eval", "step": 6000, "phase": 3, "bpb": {"en": 1.86168, "tr": 2.0208, "fi": 2.16998}, "bytes_per_chunk": {"en": 4.865, "tr": 5.915, "fi": 6.266}, "bpb_mean": 2.01749, "frozen_inventory": true, "t": 5437.6}
{"event": "reps_dumped", "dirs": ["/local/scratch/sas245/beetle-hnet/runs/ali_ali_s1/probe/reps/th0.30", "/local/scratch/sas245/beetle-hnet/runs/ali_ali_s1/probe/reps/th0.40", "/local/scratch/sas245/beetle-hnet/runs/ali_ali_s1/probe/reps/th0.50", "/local/scratch/sas245/beetle-hnet/runs/ali_ali_s1/probe/reps/th0.60", "/local/scratch/sas245/beetle-hnet/runs/ali_ali_s1/probe/reps/th0.70"], "t": 5441.9}
{"event": "final", "run": "ali_ali_s1", "arch": "ali", "objective": "ratio", "condition": "A", "step": 6000, "params": 22785088, "seed": 1, "per_lang": {"en": {"cpc": 0.1948514623915417, "cps": 26.4921875, "bpb": 1.867332696109257, "bytes_per_chunk": 4.856542077717568, "reads_per_chunk": 1.1294798758742068}, "tr": {"cpc": 0.17527481632878528, "cps": 24.7890625, "bpb": 2.0125671002616863, "bytes_per_chunk": 5.942623972796826, "reads_per_chunk": 0.7217228676678946}, "fi": {"cpc": 0.16102280580511402, "cps": 23.6640625, "bpb": 2.1882067364678637, "bytes_per_chunk": 6.220640231603427, "reads_per_chunk": 0.5617213639923115}}, "bpb_mean": 2.0227021776129357, "gini_cpc": 0.04245971306612484, "gini_cps": 0.02515723270440251, "bpic": 5.9138955488376945, "mean_boundary_entropy": 0.0936255156993866, "dechunk": "ema", "router_noise": 0.0, "transitions": [{"step": 0, "phase": 1, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0}, {"step": 2400, "phase": 2, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0}, {"step": 4200, "phase": 3, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0}], "probe_overhead_sec_total": 713.6341121196747, "probe_overhead_seq_frac": 0.4999166666666667, "wall_sec": 5446.173964738846, "inventory": {"size": 326656, "admitted": 65536, "staged": 261120, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 413, "admitted_surface": 13378, "staged_surface": 51220, "admitted_split": 46283, "staged_split": 43346, "admitted_merge": 5514, "staged_merge": 126988, "admitted_skel": 361, "staged_skel": 39566, "mean_count": 102.24116516113281, "n_util_obs": 114, "util_count_spearman": -0.23777971284502136}, "ali_params": 268096, "ali_config": {"enabled": true, "budget_B": 65536, "staging_capacity": 262144, "policy": "utility", "candidate_types": ["surface", "split", "merge", "skel"], "gate": "vector", "retrieval": true, "memory_scope": "persistent", "read_dropout": 0.1, "write_rule": "mean", "eta_min": 0.0, "consolidate_every": 50, "probe_seqs": 2, "probe_reps": 8, "horizon": 32, "predictor": "mlp", "lambda_util": 0.1, "max_span": 64, "byte_classes_path": "/local/scratch/sas245/beetle-hnet/scratch/data/byte_classes.json", "gate_hidden": 0, "type_dim": 16, "gate_bias_init": -2.0}, "util_vs_frequency_spearman": -0.23777971284502136, "paired_vs_dropout": {"n": 291, "pearson": 0.1078715889520648, "spearman": 0.10762458059206523, "mean_abs_discrepancy": 0.5718365907669067, "mean_paired": 0.052034299820661545, "mean_dropout": -0.04584226757287979}, "predictor_calibration": {"n": 848, "all": {"n": 848, "pearson": 0.15672384196653968, "spearman": 0.17715319330176119, "mae": 0.09654978662729263, "reliability": [{"pred": -0.030902376398444176, "obs": -0.030043547973036766, "n": 85}, {"pred": -0.013324351981282234, "obs": -0.009091428481042385, "n": 85}, {"pred": -0.004848717246204615, "obs": 0.004309503361582756, "n": 85}, {"pred": 0.0029140140395611525, "obs": 0.01576261594891548, "n": 84}, {"pred": 0.009082884527742863, "obs": 0.01818804256618023, "n": 85}, {"pred": 0.014386794529855251, "obs": 0.015027686022222042, "n": 85}, {"pred": 0.02050779201090336, "obs": 0.006277745123952627, "n": 84}, {"pred": 0.026512594893574715, "obs": -0.018976833671331406, "n": 85}, {"pred": 0.03519366309046745, "obs": 0.027353225275874138, "n": 85}, {"pred": 0.06194315105676651, "obs": 0.08882260322570801, "n": 85}]}, "admitted": {"n": 711, "pearson": 0.13623006222208225, "spearman": 0.13208505745656657, "mae": 0.10757148265838623, "reliability": [{"pred": -0.017781004309654236, "obs": -0.007997886277735233, "n": 71}, {"pred": -0.007415991276502609, "obs": -0.003501171711832285, "n": 71}, {"pred": 0.0004886705428361893, "obs": 0.013886559754610062, "n": 72}, {"pred": 0.007088231388479471, "obs": 0.022109031677246094, "n": 70}, {"pred": 0.011754041537642479, "obs": 0.020354347303509712, "n": 71}, {"pred": 0.017182178795337677, "obs": 0.018862931057810783, "n": 72}, {"pred": 0.022539770230650902, "obs": -0.00022549885034095496, "n": 70}, {"pred": 0.02835940010845661, "obs": -0.007592884358018637, "n": 71}, {"pred": 0.037672366946935654, "obs": 0.03289384767413139, "n": 71}, {"pred": 0.06532467156648636, "obs": 0.10061929374933243, "n": 72}]}, "staged": {"n": 137, "pearson": 0.1271825726312588, "spearman": 0.08704052868370452, "mae": 0.03934963420033455, "reliability": [{"pred": -0.04806484654545784, "obs": -0.03642170503735542, "n": 14}, {"pred": -0.038525331765413284, "obs": -0.03492997959256172, "n": 14}, {"pred": -0.029449185356497765, "obs": -0.045022640377283096, "n": 13}, {"pred": -0.023623013868927956, "obs": -0.046287715435028076, "n": 14}, {"pred": -0.018366161733865738, "obs": -0.009874004870653152, "n": 13}, {"pred": -0.010254918597638607, "obs": -0.01940336637198925, "n": 14}, {"pred": -0.0011505289003252983, "obs": -0.014970350079238415, "n": 14}, {"pred": 0.007880929857492447, "obs": -0.004072663374245167, "n": 13}, {"pred": 0.020128149539232254, "obs": -0.024625714868307114, "n": 14}, {"pred": 0.032693687826395035, "obs": -0.024533722549676895, "n": 14}]}}, "t": 5446.3}