{"event": "start", "run": "ali_ali_s2", "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": 1.7}
{"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": 1.7}
{"event": "probe", "step": 0, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 10933.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.628225, "cps": 85.414062, "mean_p": 0.513249, "boundary_entropy": 0.686295, "bpb": 8.12598}, "tr": {"chunks": 8187.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.452245, "cps": 63.960938, "mean_p": 0.498002, "boundary_entropy": 0.686826, "bpb": 8.097536}, "fi": {"chunks": 13540.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.719792, "cps": 105.78125, "mean_p": 0.517169, "boundary_entropy": 0.686661, "bpb": 8.065117}}, "bpb": {"en": 8.12598, "tr": 8.09754, "fi": 8.06512}, "gini_cpc": 0.099077, "gini_cps": 0.109267, "bpic": 1.737048, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.7198781817505053, "fi": 1.1457544508645294}, "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": 4.3}
{"event": "train", "step": 0, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 5.62735, "bpb": 8.11854, "chunk_loss": 0.062086, "F": 0.59833, "G": 0.50836, "chunks": 612.69, "bytes_per_chunk": 1.671, "boundary_entropy": 0.69105, "router_noise": 0.0, "grad_norm": 11.4935, "lr": 1e-06, "router_trainable": true, "sec_per_step": 0.019, "probe_sec": 0.0155, "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": 58765, "read_suppressed_frac": 0.0, "dropout_reads_logged": 0, "inventory": {"size": 4975, "admitted": 0, "staged": 4975, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 0, "staged_surface": 2006, "admitted_split": 0, "staged_split": 1198, "admitted_merge": 0, "staged_merge": 852, "admitted_skel": 0, "staged_skel": 919, "mean_count": 0.0, "n_util_obs": 0, "util_count_spearman": NaN}, "probe_overhead_sec_total": 0.015, "probe_overhead_seq_frac": 0.0, "util_loss": -0.0, "alpha": 0.03, "t": 6.6}
{"event": "probe", "step": 1, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 10907.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.626731, "cps": 85.210938, "mean_p": 0.513143, "boundary_entropy": 0.686304, "bpb": 8.094355}, "tr": {"chunks": 8206.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.453295, "cps": 64.109375, "mean_p": 0.498082, "boundary_entropy": 0.686828, "bpb": 8.076195}, "fi": {"chunks": 13505.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.717931, "cps": 105.507812, "mean_p": 0.517038, "boundary_entropy": 0.686672, "bpb": 8.036103}}, "bpb": {"en": 8.09435, "tr": 8.0762, "fi": 8.0361}, "gini_cpc": 0.098125, "gini_cps": 0.108304, "bpic": 1.739285, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.7232688618519676, "fi": 1.1455169299595527}, "inventory": {"size": 4975, "admitted": 0, "staged": 4975, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 0, "staged_surface": 2006, "admitted_split": 0, "staged_split": 1198, "admitted_merge": 0, "staged_merge": 852, "admitted_skel": 0, "staged_skel": 919, "mean_count": 0.0, "n_util_obs": 0, "util_count_spearman": NaN}, "t": 8.3}
{"event": "probe", "step": 2, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 10880.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.62518, "cps": 85.0, "mean_p": 0.512978, "boundary_entropy": 0.686318, "bpb": 8.031093}, "tr": {"chunks": 8246.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.455505, "cps": 64.421875, "mean_p": 0.498245, "boundary_entropy": 0.686832, "bpb": 8.031987}, "fi": {"chunks": 13469.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.716017, "cps": 105.226562, "mean_p": 0.516799, "boundary_entropy": 0.686692, "bpb": 7.976311}}, "bpb": {"en": 8.03109, "tr": 8.03199, "fi": 7.97631}, "gini_cpc": 0.096663, "gini_cps": 0.106826, "bpic": 1.740512, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.7285980488284359, "fi": 1.1452985063104504}, "inventory": {"size": 8934, "admitted": 0, "staged": 8934, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 0, "staged_surface": 3531, "admitted_split": 0, "staged_split": 2214, "admitted_merge": 0, "staged_merge": 1439, "admitted_skel": 0, "staged_skel": 1750, "mean_count": 0.0, "n_util_obs": 0, "util_count_spearman": NaN}, "t": 12.4}
{"event": "probe", "step": 4, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 10715.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.615698, "cps": 83.710938, "mean_p": 0.512101, "boundary_entropy": 0.68638, "bpb": 7.819663}, "tr": {"chunks": 8386.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.463238, "cps": 65.515625, "mean_p": 0.49867, "boundary_entropy": 0.686846, "bpb": 7.878758}, "fi": {"chunks": 13218.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.702674, "cps": 103.265625, "mean_p": 0.515685, "boundary_entropy": 0.686774, "bpb": 7.77364}}, "bpb": {"en": 7.81966, "tr": 7.87876, "fi": 7.77364}, "gini_cpc": 0.089595, "gini_cps": 0.099673, "bpic": 1.755376, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.7523782831425373, "fi": 1.141263187490228}, "inventory": {"size": 15204, "admitted": 0, "staged": 15204, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 4, "admitted_surface": 0, "staged_surface": 5979, "admitted_split": 0, "staged_split": 3775, "admitted_merge": 0, "staged_merge": 2535, "admitted_skel": 0, "staged_skel": 2915, "mean_count": 0.0, "n_util_obs": 0, "util_count_spearman": NaN}, "t": 17.0}
{"event": "probe", "step": 8, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 9632.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.553468, "cps": 75.25, "mean_p": 0.507285, "boundary_entropy": 0.686614, "bpb": 7.293115}, "tr": {"chunks": 8637.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.477103, "cps": 67.476562, "mean_p": 0.499595, "boundary_entropy": 0.686902, "bpb": 7.478104}, "fi": {"chunks": 11874.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.631226, "cps": 92.765625, "mean_p": 0.510289, "boundary_entropy": 0.687081, "bpb": 7.295772}}, "bpb": {"en": 7.29312, "tr": 7.4781, "fi": 7.29577}, "gini_cpc": 0.06183, "gini_cps": 0.071592, "bpic": 1.882095, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.8620253042162025, "fi": 1.1404934815831058}, "inventory": {"size": 25354, "admitted": 0, "staged": 25354, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 9, "admitted_surface": 0, "staged_surface": 9815, "admitted_split": 0, "staged_split": 6084, "admitted_merge": 0, "staged_merge": 4552, "admitted_skel": 0, "staged_skel": 4903, "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": 3752.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.215595, "cps": 29.3125, "mean_p": 0.484385, "boundary_entropy": 0.686161, "bpb": 6.724714}, "tr": {"chunks": 7275.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.401867, "cps": 56.835938, "mean_p": 0.494383, "boundary_entropy": 0.686996, "bpb": 7.013482}, "fi": {"chunks": 4583.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.243634, "cps": 35.804688, "mean_p": 0.48653, "boundary_entropy": 0.686903, "bpb": 6.787286}}, "bpb": {"en": 6.72471, "tr": 7.01348, "fi": 6.78729}, "gini_cpc": 0.144213, "gini_cps": 0.150459, "bpic": 3.634337, "worst_lang": "tr", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.8639906807845699, "fi": 1.1300541754195221}, "inventory": {"size": 63011, "admitted": 0, "staged": 63011, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 23, "admitted_surface": 0, "staged_surface": 23791, "admitted_split": 0, "staged_split": 17657, "admitted_merge": 0, "staged_merge": 6742, "admitted_skel": 0, "staged_skel": 14821, "mean_count": 0.0, "n_util_obs": 0, "util_count_spearman": NaN}, "t": 36.1}
{"event": "probe", "step": 32, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2600.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.1494, "cps": 20.3125, "mean_p": 0.486729, "boundary_entropy": 0.686947, "bpb": 6.304654}, "tr": {"chunks": 4647.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.256698, "cps": 36.304688, "mean_p": 0.490184, "boundary_entropy": 0.68756, "bpb": 6.651847}, "fi": {"chunks": 1390.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.073893, "cps": 10.859375, "mean_p": 0.479405, "boundary_entropy": 0.686974, "bpb": 6.420778}}, "bpb": {"en": 6.30465, "tr": 6.65185, "fi": 6.42078}, "gini_cpc": 0.253901, "gini_cps": 0.251399, "bpic": 6.568484, "worst_lang": "tr", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.7181967502198956, "fi": 0.49459951828512777}, "inventory": {"size": 138834, "admitted": 0, "staged": 138834, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 47, "admitted_surface": 0, "staged_surface": 50323, "admitted_split": 0, "staged_split": 43517, "admitted_merge": 0, "staged_merge": 8707, "admitted_skel": 0, "staged_skel": 36287, "mean_count": 0.0, "n_util_obs": 0, "util_count_spearman": NaN}, "t": 48.6}
{"event": "consolidate", "step": 50, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 2999, "swapped": 0, "uncertain_staged": 230562, "t": 66.4}
{"event": "train", "step": 50, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 4.21235, "bpb": 6.07714, "chunk_loss": 0.034235, "F": 0.22617, "G": 0.48855, "chunks": 231.59, "bytes_per_chunk": 4.422, "boundary_entropy": 0.69161, "router_noise": 0.0, "grad_norm": 2.4788, "lr": 5.1e-05, "router_trainable": true, "sec_per_step": 1.206, "probe_sec": 0.1192, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.10512, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": -0.00611, "n_shadow": 5, "mean_delta_shadow": 0.00425, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": -0.36332, "trunc_abs_frac": 0.18436, "mean_delta_full": 0.01481, "read_hit_rate": {"surface": 0.00393, "split": 0.14754, "merge": 0.0, "skel": 0.0}, "mean_gate": {"surface": 0.11964, "split": 0.11934, "merge": null, "skel": null}, "n_reads": 1394, "n_cand": 27565, "read_suppressed_frac": 0.1019, "dropout_reads_logged": 1394, "inventory": {"size": 236122, "admitted": 2999, "staged": 233123, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 80, "admitted_surface": 1, "staged_surface": 84438, "admitted_split": 4, "staged_split": 76756, "admitted_merge": 0, "staged_merge": 10404, "admitted_skel": 2994, "staged_skel": 61525, "mean_count": 41.90696716308594, "n_util_obs": 5, "util_count_spearman": -0.7}, "probe_overhead_sec_total": 5.392, "probe_overhead_seq_frac": 0.489, "util_loss": 0.071734, "alpha": 0.03, "t": 66.9}
{"event": "probe", "step": 64, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 4800.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.275815, "cps": 37.5, "mean_p": 0.485117, "boundary_entropy": 0.685503, "bpb": 5.552299}, "tr": {"chunks": 5421.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.299453, "cps": 42.351562, "mean_p": 0.486644, "boundary_entropy": 0.686486, "bpb": 5.956656}, "fi": {"chunks": 4730.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.251449, "cps": 36.953125, "mean_p": 0.484103, "boundary_entropy": 0.686338, "bpb": 5.717739}}, "bpb": {"en": 5.5523, "tr": 5.95666, "fi": 5.71774}, "gini_cpc": 0.038711, "gini_cps": 0.030812, "bpic": 3.794529, "worst_lang": "tr", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0857047519748109, "fi": 0.9116584046568497}, "inventory": {"size": 265143, "admitted": 2999, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 88, "admitted_surface": 1, "staged_surface": 93113, "admitted_split": 4, "staged_split": 86866, "admitted_merge": 0, "staged_merge": 11388, "admitted_skel": 2994, "staged_skel": 70777, "mean_count": 47.29943084716797, "n_util_obs": 5, "util_count_spearman": -0.6}, "t": 74.8}
{"event": "consolidate", "step": 100, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 62537, "swapped": 5, "uncertain_staged": 262139, "t": 105.8}
{"event": "train", "step": 100, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 3.25252, "bpb": 4.6924, "chunk_loss": 0.029419, "F": 0.15707, "G": 0.45127, "chunks": 160.84, "bytes_per_chunk": 6.366, "boundary_entropy": 0.6817, "router_noise": 0.0, "grad_norm": 2.5841, "lr": 0.000101, "router_trainable": true, "sec_per_step": 0.795, "probe_sec": 0.0463, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.03395, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": -0.01477, "n_shadow": 7, "mean_delta_shadow": 0.00155, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": -0.27419, "trunc_abs_frac": 0.30737, "mean_delta_full": -0.02649, "read_hit_rate": {"surface": 0.04029, "split": 0.0, "merge": 0.0, "skel": 0.2961}, "mean_gate": {"surface": 0.11947, "split": null, "merge": null, "skel": 0.11937}, "n_reads": 1192, "n_cand": 22140, "read_suppressed_frac": 0.1149, "dropout_reads_logged": 1192, "inventory": {"size": 270554, "admitted": 65536, "staged": 205018, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 114, "admitted_surface": 4642, "staged_surface": 91453, "admitted_split": 0, "staged_split": 84967, "admitted_merge": 2, "staged_merge": 14736, "admitted_skel": 60892, "staged_skel": 13862, "mean_count": 3.8387908935546875, "n_util_obs": 0, "util_count_spearman": NaN}, "probe_overhead_sec_total": 9.477, "probe_overhead_seq_frac": 0.4944, "util_loss": 0.002862, "alpha": 0.03, "t": 106.6}
{"event": "probe", "step": 128, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3535.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.203126, "cps": 27.617188, "mean_p": 0.408288, "boundary_entropy": 0.640275, "bpb": 4.177357}, "tr": {"chunks": 4828.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.266696, "cps": 37.71875, "mean_p": 0.433294, "boundary_entropy": 0.657577, "bpb": 4.474715}, "fi": {"chunks": 3816.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.20286, "cps": 29.8125, "mean_p": 0.424942, "boundary_entropy": 0.657978, "bpb": 4.371183}}, "bpb": {"en": 4.17736, "tr": 4.47471, "fi": 4.37118}, "gini_cpc": 0.063265, "gini_cps": 0.070778, "bpic": 4.658182, "worst_lang": "tr", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.3129597484018254, "fi": 0.9986911116212436}, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 146, "admitted_surface": 4642, "staged_surface": 106536, "admitted_split": 0, "staged_split": 92455, "admitted_merge": 2, "staged_merge": 18024, "admitted_skel": 60892, "staged_skel": 45129, "mean_count": 4.48175048828125, "n_util_obs": 3, "util_count_spearman": 1.0}, "t": 131.8}
{"event": "consolidate", "step": 150, "retired": 0, "split_retired": 2, "merge_admitted": 0, "admitted": 2, "swapped": 5, "uncertain_staged": 262136, "t": 151.3}
{"event": "train", "step": 150, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 2.79457, "bpb": 4.03171, "chunk_loss": 0.032325, "F": 0.21439, "G": 0.3895, "chunks": 219.53, "bytes_per_chunk": 4.664, "boundary_entropy": 0.6216, "router_noise": 0.0, "grad_norm": 1.6863, "lr": 0.000151, "router_trainable": true, "sec_per_step": 0.907, "probe_sec": 0.0829, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.062, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": -0.01498, "n_shadow": 7, "mean_delta_shadow": -0.02114, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 3.53846, "trunc_abs_frac": 1.58929, "mean_delta_full": -0.00159, "read_hit_rate": {"surface": 0.07407, "split": 0.0, "merge": 0.0, "skel": 0.2161}, "mean_gate": {"surface": 0.12113, "split": null, "merge": null, "skel": 0.1228}, "n_reads": 1401, "n_cand": 28644, "read_suppressed_frac": 0.0899, "dropout_reads_logged": 1401, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 165, "admitted_surface": 4646, "staged_surface": 100800, "admitted_split": 0, "staged_split": 84741, "admitted_merge": 2, "staged_merge": 21234, "admitted_skel": 60888, "staged_skel": 55369, "mean_count": 4.906036376953125, "n_util_obs": 0, "util_count_spearman": NaN}, "probe_overhead_sec_total": 15.907, "probe_overhead_seq_frac": 0.4963, "util_loss": 0.000946, "alpha": 0.03, "t": 152.0}
{"event": "consolidate", "step": 200, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 9, "uncertain_staged": 262134, "t": 189.2}
{"event": "probe", "step": 200, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3923.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.225421, "cps": 30.648438, "mean_p": 0.364834, "boundary_entropy": 0.579723, "bpb": 3.637381}, "tr": {"chunks": 4482.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.247583, "cps": 35.015625, "mean_p": 0.380099, "boundary_entropy": 0.593817, "bpb": 3.793136}, "fi": {"chunks": 4159.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.221094, "cps": 32.492188, "mean_p": 0.407451, "boundary_entropy": 0.634508, "bpb": 3.846298}}, "bpb": {"en": 3.63738, "tr": 3.79314, "fi": 3.8463}, "gini_cpc": 0.025442, "gini_cps": 0.029661, "bpic": 4.515441, "worst_lang": "tr", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.09831550052424, "fi": 0.9808054016479828}, "inventory": {"size": 327671, "admitted": 65536, "staged": 262135, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 218, "admitted_surface": 4641, "staged_surface": 98199, "admitted_split": 0, "staged_split": 84837, "admitted_merge": 2, "staged_merge": 26758, "admitted_skel": 60893, "staged_skel": 52341, "mean_count": 4.526763916015625, "n_util_obs": 0, "util_count_spearman": NaN}, "t": 190.9}
{"event": "eval", "step": 200, "phase": 1, "bpb": {"en": 3.644, "tr": 3.80377, "fi": 3.84542}, "bytes_per_chunk": {"en": 4.83, "tr": 5.402, "fi": 5.041}, "bpb_mean": 3.7644, "frozen_inventory": true, "t": 193.3}
{"event": "train", "step": 200, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 2.53204, "bpb": 3.65296, "chunk_loss": 0.031017, "F": 0.19138, "G": 0.3542, "chunks": 195.97, "bytes_per_chunk": 5.225, "boundary_entropy": 0.58064, "router_noise": 0.0, "grad_norm": 2.2921, "lr": 0.000201, "router_trainable": true, "sec_per_step": 0.96, "probe_sec": 0.0748, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.05644, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": 0.07143, "n_shadow": 7, "mean_delta_shadow": 0.02522, "n_split": 1, "mean_delta_split": 0.01501, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 7, "trunc_ratio_sum": 0.8055, "trunc_abs_frac": 0.89921, "mean_delta_full": 0.08374, "read_hit_rate": {"surface": 0.00561, "split": 0.0, "merge": 0.0, "skel": 0.14455}, "mean_gate": {"surface": 0.12514, "split": null, "merge": null, "skel": 0.12767}, "n_reads": 646, "n_cand": 27351, "read_suppressed_frac": 0.1053, "dropout_reads_logged": 646, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 172, "admitted_surface": 4641, "staged_surface": 98159, "admitted_split": 0, "staged_split": 84956, "admitted_merge": 2, "staged_merge": 26159, "admitted_skel": 60893, "staged_skel": 52870, "mean_count": 4.53662109375, "n_util_obs": 0, "util_count_spearman": NaN}, "probe_overhead_sec_total": 20.32, "probe_overhead_seq_frac": 0.4972, "util_loss": 0.000825, "alpha": 0.03, "t": 199.9}
{"event": "consolidate", "step": 250, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262130, "t": 236.7}
{"event": "train", "step": 250, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 2.38255, "bpb": 3.43729, "chunk_loss": 0.030997, "F": 0.19141, "G": 0.33187, "chunks": 196.0, "bytes_per_chunk": 5.224, "boundary_entropy": 0.54123, "router_noise": 0.0, "grad_norm": 1.585, "lr": 0.000251, "router_trainable": true, "sec_per_step": 0.767, "probe_sec": 0.1435, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.07392, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": -0.02267, "n_shadow": 4, "mean_delta_shadow": -0.04189, "n_split": 0, "mean_delta_split": NaN, "n_merge": 2, "mean_delta_merge": -0.00815, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 3.05983, "trunc_abs_frac": 2.01671, "mean_delta_full": -0.01335, "read_hit_rate": {"surface": 0.00032, "split": 0.0, "merge": 0.0, "skel": 0.11382}, "mean_gate": {"surface": 0.1325, "split": null, "merge": null, "skel": 0.12567}, "n_reads": 501, "n_cand": 27448, "read_suppressed_frac": 0.0938, "dropout_reads_logged": 501, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 99, "admitted_surface": 4591, "staged_surface": 96113, "admitted_split": 0, "staged_split": 80667, "admitted_merge": 31, "staged_merge": 28610, "admitted_skel": 60914, "staged_skel": 56754, "mean_count": 3.569671630859375, "n_util_obs": 2, "util_count_spearman": NaN}, "probe_overhead_sec_total": 25.779, "probe_overhead_seq_frac": 0.4978, "util_loss": 0.001164, "alpha": 0.03, "t": 238.3}
{"event": "probe", "step": 256, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3667.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.210711, "cps": 28.648438, "mean_p": 0.336654, "boundary_entropy": 0.53071, "bpb": 3.35715}, "tr": {"chunks": 4295.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.237253, "cps": 33.554688, "mean_p": 0.326162, "boundary_entropy": 0.530492, "bpb": 3.528085}, "fi": {"chunks": 4194.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.222955, "cps": 32.765625, "mean_p": 0.375717, "boundary_entropy": 0.596572, "bpb": 3.67097}}, "bpb": {"en": 3.35715, "tr": 3.52809, "fi": 3.67097}, "gini_cpc": 0.026374, "gini_cps": 0.034441, "bpic": 4.666996, "worst_lang": "tr", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.1259674268537696, "fi": 1.0581074030264772}, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 99, "admitted_surface": 4591, "staged_surface": 95596, "admitted_split": 0, "staged_split": 80225, "admitted_merge": 31, "staged_merge": 29329, "admitted_skel": 60914, "staged_skel": 56994, "mean_count": 3.60015869140625, "n_util_obs": 3, "util_count_spearman": 0.5}, "t": 243.7}
{"event": "consolidate", "step": 300, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262126, "t": 285.9}
{"event": "train", "step": 300, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 2.28131, "bpb": 3.29124, "chunk_loss": 0.03046, "F": 0.18216, "G": 0.29893, "chunks": 186.53, "bytes_per_chunk": 5.49, "boundary_entropy": 0.47864, "router_noise": 0.0, "grad_norm": 1.1685, "lr": 0.0003, "router_trainable": true, "sec_per_step": 0.961, "probe_sec": 0.0984, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.0815, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 5, "mean_delta_loo": 0.02718, "n_shadow": 5, "mean_delta_shadow": -0.00989, "n_split": 0, "mean_delta_split": NaN, "n_merge": 2, "mean_delta_merge": -0.03073, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 0.31072, "trunc_abs_frac": 0.81024, "mean_delta_full": -0.08594, "read_hit_rate": {"surface": 0.0, "split": 0.0, "merge": 0.0, "skel": 0.03296}, "mean_gate": {"surface": null, "split": null, "merge": null, "skel": 0.12863}, "n_reads": 134, "n_cand": 26316, "read_suppressed_frac": 0.1418, "dropout_reads_logged": 134, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 119, "admitted_surface": 4911, "staged_surface": 91432, "admitted_split": 97, "staged_split": 76017, "admitted_merge": 399, "staged_merge": 35528, "admitted_skel": 60129, "staged_skel": 59167, "mean_count": 2.7076568603515625, "n_util_obs": 1, "util_count_spearman": NaN}, "probe_overhead_sec_total": 31.198, "probe_overhead_seq_frac": 0.4981, "util_loss": 0.000884, "alpha": 0.03, "t": 286.3}
{"event": "consolidate", "step": 350, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262126, "t": 328.1}
{"event": "train", "step": 350, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 2.17792, "bpb": 3.14207, "chunk_loss": 0.03054, "F": 0.18506, "G": 0.29162, "chunks": 189.5, "bytes_per_chunk": 5.404, "boundary_entropy": 0.45046, "router_noise": 0.0, "grad_norm": 1.3665, "lr": 0.0002999, "router_trainable": true, "sec_per_step": 0.856, "probe_sec": 0.0768, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.05379, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": -0.02706, "n_shadow": 5, "mean_delta_shadow": -0.03321, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 0.13033, "trunc_abs_frac": 8.34235, "mean_delta_full": -0.06873, "read_hit_rate": {"surface": 0.00017, "split": 0.0, "merge": 0.0, "skel": 0.03081}, "mean_gate": {"surface": 0.13426, "split": null, "merge": null, "skel": 0.12665}, "n_reads": 124, "n_cand": 26542, "read_suppressed_frac": 0.1048, "dropout_reads_logged": 124, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 131, "admitted_surface": 4910, "staged_surface": 87659, "admitted_split": 135, "staged_split": 71439, "admitted_merge": 1385, "staged_merge": 44412, "admitted_skel": 59106, "staged_skel": 58634, "mean_count": 2.817901611328125, "n_util_obs": 2, "util_count_spearman": NaN}, "probe_overhead_sec_total": 36.19, "probe_overhead_seq_frac": 0.4984, "util_loss": 0.001271, "alpha": 0.03, "t": 329.1}
{"event": "consolidate", "step": 400, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262120, "t": 374.2}
{"event": "probe", "step": 400, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3282.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.188588, "cps": 25.640625, "mean_p": 0.28621, "boundary_entropy": 0.416674, "bpb": 2.9109}, "tr": {"chunks": 3304.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.182511, "cps": 25.8125, "mean_p": 0.279461, "boundary_entropy": 0.440225, "bpb": 3.188483}, "fi": {"chunks": 4048.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.215193, "cps": 31.625, "mean_p": 0.280633, "boundary_entropy": 0.453531, "bpb": 3.377803}}, "bpb": {"en": 2.9109, "tr": 3.18848, "fi": 3.3778}, "gini_cpc": 0.037162, "gini_cps": 0.048022, "bpic": 5.334963, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9677764076191679, "fi": 1.1410749301941947}, "inventory": {"size": 326656, "admitted": 65536, "staged": 261120, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 137, "admitted_surface": 4929, "staged_surface": 82622, "admitted_split": 219, "staged_split": 67034, "admitted_merge": 2302, "staged_merge": 54962, "admitted_skel": 58086, "staged_skel": 56502, "mean_count": 2.8455352783203125, "n_util_obs": 6, "util_count_spearman": 0.6571428571428571}, "t": 375.8}
{"event": "eval", "step": 400, "phase": 1, "bpb": {"en": 2.924, "tr": 3.1953, "fi": 3.37041}, "bytes_per_chunk": {"en": 5.502, "tr": 6.091, "fi": 5.239}, "bpb_mean": 3.16323, "frozen_inventory": true, "t": 378.3}
{"event": "train", "step": 400, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 2.09276, "bpb": 3.01922, "chunk_loss": 0.030226, "F": 0.17416, "G": 0.27503, "chunks": 178.34, "bytes_per_chunk": 5.742, "boundary_entropy": 0.41556, "router_noise": 0.0, "grad_norm": 1.1284, "lr": 0.0002998, "router_trainable": true, "sec_per_step": 1.073, "probe_sec": 0.0766, "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": 6, "mean_delta_loo": 0.02116, "n_shadow": 6, "mean_delta_shadow": -0.00681, "n_split": 0, "mean_delta_split": NaN, "n_merge": 1, "mean_delta_merge": 0.11682, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 0.46297, "trunc_abs_frac": 0.78108, "mean_delta_full": -0.09037, "read_hit_rate": {"surface": 0.0, "split": 0.0, "merge": 0.0, "skel": 0.02798}, "mean_gate": {"surface": null, "split": null, "merge": null, "skel": 0.1275}, "n_reads": 122, "n_cand": 25650, "read_suppressed_frac": 0.0656, "dropout_reads_logged": 122, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 137, "admitted_surface": 4929, "staged_surface": 82806, "admitted_split": 219, "staged_split": 67374, "admitted_merge": 2302, "staged_merge": 55196, "admitted_skel": 58086, "staged_skel": 56768, "mean_count": 2.8473968505859375, "n_util_obs": 6, "util_count_spearman": 0.6571428571428571}, "probe_overhead_sec_total": 41.105, "probe_overhead_seq_frac": 0.4986, "util_loss": 0.002154, "alpha": 0.03, "t": 382.8}
{"event": "consolidate", "step": 450, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262116, "t": 420.5}
{"event": "train", "step": 450, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 2.00941, "bpb": 2.89897, "chunk_loss": 0.030611, "F": 0.19235, "G": 0.2696, "chunks": 196.97, "bytes_per_chunk": 5.199, "boundary_entropy": 0.37996, "router_noise": 0.0, "grad_norm": 1.6015, "lr": 0.0002995, "router_trainable": true, "sec_per_step": 0.773, "probe_sec": 0.1022, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.08564, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 5, "mean_delta_loo": -0.00144, "n_shadow": 5, "mean_delta_shadow": -0.00795, "n_split": 0, "mean_delta_split": NaN, "n_merge": 2, "mean_delta_merge": -0.04596, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": -0.15876, "trunc_abs_frac": 0.84601, "mean_delta_full": 0.13379, "read_hit_rate": {"surface": 0.0, "split": 0.0, "merge": 0.0, "skel": 0.03303}, "mean_gate": {"surface": null, "split": null, "merge": null, "skel": 0.13778}, "n_reads": 160, "n_cand": 28374, "read_suppressed_frac": 0.1625, "dropout_reads_logged": 160, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 107, "admitted_surface": 4934, "staged_surface": 78024, "admitted_split": 245, "staged_split": 63246, "admitted_merge": 3294, "staged_merge": 65065, "admitted_skel": 57063, "staged_skel": 55809, "mean_count": 2.9474029541015625, "n_util_obs": 11, "util_count_spearman": 0.22727272727272727}, "probe_overhead_sec_total": 45.663, "probe_overhead_seq_frac": 0.4988, "util_loss": 0.002428, "alpha": 0.03, "t": 421.4}
{"event": "consolidate", "step": 500, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262114, "t": 463.7}
{"event": "train", "step": 500, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.91415, "bpb": 2.76153, "chunk_loss": 0.030987, "F": 0.20773, "G": 0.26948, "chunks": 212.72, "bytes_per_chunk": 4.814, "boundary_entropy": 0.35317, "router_noise": 0.0, "grad_norm": 1.4641, "lr": 0.0002992, "router_trainable": true, "sec_per_step": 0.865, "probe_sec": 0.1224, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.10056, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 5, "mean_delta_loo": 0.0718, "n_shadow": 9, "mean_delta_shadow": 0.10544, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 1.34196, "trunc_abs_frac": 0.75675, "mean_delta_full": 0.20597, "read_hit_rate": {"surface": 0.0, "split": 0.0, "merge": 0.0, "skel": 0.03004}, "mean_gate": {"surface": null, "split": null, "merge": null, "skel": 0.13728}, "n_reads": 158, "n_cand": 30672, "read_suppressed_frac": 0.1076, "dropout_reads_logged": 158, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 126, "admitted_surface": 4957, "staged_surface": 74314, "admitted_split": 299, "staged_split": 58797, "admitted_merge": 4186, "staged_merge": 75499, "admitted_skel": 56094, "staged_skel": 53534, "mean_count": 2.774932861328125, "n_util_obs": 15, "util_count_spearman": 0.017857142857142856}, "probe_overhead_sec_total": 50.274, "probe_overhead_seq_frac": 0.4989, "util_loss": 0.003188, "alpha": 0.03, "t": 464.7}
{"event": "consolidate", "step": 550, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262108, "t": 510.3}
{"event": "train", "step": 550, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.91325, "bpb": 2.76024, "chunk_loss": 0.030688, "F": 0.2001, "G": 0.25612, "chunks": 204.91, "bytes_per_chunk": 4.997, "boundary_entropy": 0.3278, "router_noise": 0.0, "grad_norm": 1.6287, "lr": 0.0002987, "router_trainable": true, "sec_per_step": 0.929, "probe_sec": 0.1148, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.09724, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": 0.55533, "n_shadow": 5, "mean_delta_shadow": 0.07879, "n_split": 0, "mean_delta_split": NaN, "n_merge": 1, "mean_delta_merge": 0.01282, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": -0.34912, "trunc_abs_frac": 0.64405, "mean_delta_full": 0.16211, "read_hit_rate": {"surface": 0.0, "split": 0.0, "merge": 0.0, "skel": 0.02324}, "mean_gate": {"surface": null, "split": null, "merge": null, "skel": 0.13771}, "n_reads": 119, "n_cand": 29677, "read_suppressed_frac": 0.0588, "dropout_reads_logged": 119, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 134, "admitted_surface": 4963, "staged_surface": 71064, "admitted_split": 304, "staged_split": 55877, "admitted_merge": 5197, "staged_merge": 82890, "admitted_skel": 55072, "staged_skel": 52313, "mean_count": 2.458404541015625, "n_util_obs": 18, "util_count_spearman": 0.03611971104231166}, "probe_overhead_sec_total": 55.2, "probe_overhead_seq_frac": 0.499, "util_loss": 0.003926, "alpha": 0.03, "t": 511.2}
{"event": "consolidate", "step": 600, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262106, "t": 551.4}
{"event": "probe", "step": 600, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3612.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.20755, "cps": 28.21875, "mean_p": 0.258715, "boundary_entropy": 0.322447, "bpb": 2.600301}, "tr": {"chunks": 3906.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.215765, "cps": 30.515625, "mean_p": 0.250665, "boundary_entropy": 0.309433, "bpb": 2.931609}, "fi": {"chunks": 4340.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.230716, "cps": 33.90625, "mean_p": 0.27043, "boundary_entropy": 0.33, "bpb": 3.150233}}, "bpb": {"en": 2.6003, "tr": 2.93161, "fi": 3.15023}, "gini_cpc": 0.023613, "gini_cps": 0.040929, "bpic": 4.784281, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0395803599300641, "fi": 1.1116145550661227}, "inventory": {"size": 326656, "admitted": 65536, "staged": 261120, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 146, "admitted_surface": 4963, "staged_surface": 68486, "admitted_split": 304, "staged_split": 53729, "admitted_merge": 6221, "staged_merge": 87932, "admitted_skel": 54048, "staged_skel": 50973, "mean_count": 2.4069671630859375, "n_util_obs": 28, "util_count_spearman": 0.10210786038212191}, "t": 553.0}
{"event": "eval", "step": 600, "phase": 1, "bpb": {"en": 2.61189, "tr": 2.9388, "fi": 3.15125}, "bytes_per_chunk": {"en": 5.03, "tr": 5.14, "fi": 4.625}, "bpb_mean": 2.90065, "frozen_inventory": true, "t": 555.9}
{"event": "train", "step": 600, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.89545, "bpb": 2.73456, "chunk_loss": 0.03063, "F": 0.19815, "G": 0.25105, "chunks": 202.91, "bytes_per_chunk": 5.047, "boundary_entropy": 0.31602, "router_noise": 0.0, "grad_norm": 1.2567, "lr": 0.0002982, "router_trainable": true, "sec_per_step": 0.998, "probe_sec": 0.1326, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.10591, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": 0.15462, "n_shadow": 6, "mean_delta_shadow": -0.05648, "n_split": 0, "mean_delta_split": NaN, "n_merge": 1, "mean_delta_merge": -0.01807, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 0.69064, "trunc_abs_frac": 0.70767, "mean_delta_full": 0.27914, "read_hit_rate": {"surface": 0.0, "split": 0.0, "merge": 0.0, "skel": 0.02748}, "mean_gate": {"surface": null, "split": null, "merge": null, "skel": 0.1397}, "n_reads": 141, "n_cand": 29399, "read_suppressed_frac": 0.0567, "dropout_reads_logged": 141, "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": 4963, "staged_surface": 68732, "admitted_split": 304, "staged_split": 53854, "admitted_merge": 6221, "staged_merge": 88431, "admitted_skel": 54048, "staged_skel": 51127, "mean_count": 2.40911865234375, "n_util_obs": 28, "util_count_spearman": 0.09140667761357417}, "probe_overhead_sec_total": 59.917, "probe_overhead_seq_frac": 0.4991, "util_loss": 0.004893, "alpha": 0.03, "t": 561.0}
{"event": "consolidate", "step": 650, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262098, "t": 604.4}
{"event": "train", "step": 650, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.85017, "bpb": 2.66923, "chunk_loss": 0.030568, "F": 0.19556, "G": 0.24841, "chunks": 200.25, "bytes_per_chunk": 5.114, "boundary_entropy": 0.30233, "router_noise": 0.0, "grad_norm": 1.3352, "lr": 0.0002975, "router_trainable": true, "sec_per_step": 0.893, "probe_sec": 0.0916, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.05208, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 2, "mean_delta_loo": 0.08752, "n_shadow": 8, "mean_delta_shadow": -0.0367, "n_split": 0, "mean_delta_split": NaN, "n_merge": 2, "mean_delta_merge": 0.03082, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 1.25458, "trunc_abs_frac": 1.47118, "mean_delta_full": 0.09766, "read_hit_rate": {"surface": 0.00016, "split": 0.0, "merge": 0.00032, "skel": 0.01558}, "mean_gate": {"surface": 0.11637, "split": null, "merge": 0.11044, "skel": 0.14055}, "n_reads": 81, "n_cand": 28960, "read_suppressed_frac": 0.0988, "dropout_reads_logged": 81, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 131, "admitted_surface": 4963, "staged_surface": 66709, "admitted_split": 304, "staged_split": 51992, "admitted_merge": 7245, "staged_merge": 93711, "admitted_skel": 53024, "staged_skel": 49732, "mean_count": 2.411590576171875, "n_util_obs": 32, "util_count_spearman": 0.1416918711844079}, "probe_overhead_sec_total": 64.657, "probe_overhead_seq_frac": 0.4991, "util_loss": 0.005155, "alpha": 0.03, "t": 605.7}
{"event": "consolidate", "step": 700, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262093, "t": 655.3}
{"event": "train", "step": 700, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.99384, "bpb": 2.87651, "chunk_loss": 0.030985, "F": 0.21365, "G": 0.25263, "chunks": 218.78, "bytes_per_chunk": 4.68, "boundary_entropy": 0.28747, "router_noise": 0.0, "grad_norm": 1.2304, "lr": 0.0002967, "router_trainable": true, "sec_per_step": 1.01, "probe_sec": 0.0667, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04798, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": 0.01689, "n_shadow": 4, "mean_delta_shadow": -0.02199, "n_split": 0, "mean_delta_split": NaN, "n_merge": 1, "mean_delta_merge": -0.03293, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 0.81679, "trunc_abs_frac": 2.28618, "mean_delta_full": 0.054, "read_hit_rate": {"surface": 0.0, "split": 0.0, "merge": 0.02494, "skel": 0.01536}, "mean_gate": {"surface": null, "split": null, "merge": 0.129, "skel": 0.13135}, "n_reads": 253, "n_cand": 30976, "read_suppressed_frac": 0.1028, "dropout_reads_logged": 253, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 125, "admitted_surface": 4963, "staged_surface": 65175, "admitted_split": 152, "staged_split": 51080, "admitted_merge": 8269, "staged_merge": 97191, "admitted_skel": 52152, "staged_skel": 48698, "mean_count": 2.7028045654296875, "n_util_obs": 46, "util_count_spearman": 0.19420888719955506}, "probe_overhead_sec_total": 69.416, "probe_overhead_seq_frac": 0.4992, "util_loss": 0.005804, "alpha": 0.03, "t": 656.2}
{"event": "consolidate", "step": 750, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262090, "t": 699.1}
{"event": "train", "step": 750, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.78888, "bpb": 2.58081, "chunk_loss": 0.030603, "F": 0.19986, "G": 0.24317, "chunks": 204.66, "bytes_per_chunk": 5.004, "boundary_entropy": 0.28094, "router_noise": 0.0, "grad_norm": 1.1601, "lr": 0.0002959, "router_trainable": true, "sec_per_step": 0.868, "probe_sec": 0.0952, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.07007, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 10, "mean_delta_loo": 0.02391, "n_shadow": 2, "mean_delta_shadow": -0.05402, "n_split": 0, "mean_delta_split": NaN, "n_merge": 1, "mean_delta_merge": 0.2301, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 3.2931, "trunc_abs_frac": 0.63715, "mean_delta_full": 0.02921, "read_hit_rate": {"surface": 0.00875, "split": 0.02506, "merge": 0.02328, "skel": 0.05374}, "mean_gate": {"surface": 0.12945, "split": 0.13056, "merge": 0.12598, "skel": 0.13721}, "n_reads": 769, "n_cand": 29553, "read_suppressed_frac": 0.1001, "dropout_reads_logged": 769, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 129, "admitted_surface": 4962, "staged_surface": 64213, "admitted_split": 3, "staged_split": 49950, "admitted_merge": 9274, "staged_merge": 100136, "admitted_skel": 51297, "staged_skel": 47845, "mean_count": 7.8365325927734375, "n_util_obs": 60, "util_count_spearman": 0.10025703393416854}, "probe_overhead_sec_total": 74.1, "probe_overhead_seq_frac": 0.4993, "util_loss": 0.006707, "alpha": 0.03, "t": 699.6}
{"event": "consolidate", "step": 800, "retired": 0, "split_retired": 1, "merge_admitted": 0, "admitted": 1, "swapped": 1024, "uncertain_staged": 262091, "t": 746.8}
{"event": "probe", "step": 800, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3605.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.207148, "cps": 28.164062, "mean_p": 0.251926, "boundary_entropy": 0.294625, "bpb": 2.446065}, "tr": {"chunks": 4027.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.222449, "cps": 31.460938, "mean_p": 0.241815, "boundary_entropy": 0.276716, "bpb": 2.768999}, "fi": {"chunks": 4259.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.22641, "cps": 33.273438, "mean_p": 0.260323, "boundary_entropy": 0.297982, "bpb": 2.980037}}, "bpb": {"en": 2.44607, "tr": 2.769, "fi": 2.98004}, "gini_cpc": 0.019575, "gini_cps": 0.036666, "bpic": 4.771003, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0738655970999051, "fi": 1.0929860217680347}, "inventory": {"size": 326655, "admitted": 65536, "staged": 261119, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 122, "admitted_surface": 5017, "staged_surface": 63151, "admitted_split": 3, "staged_split": 49948, "admitted_merge": 9958, "staged_merge": 100934, "admitted_skel": 50558, "staged_skel": 47086, "mean_count": 6.9459228515625, "n_util_obs": 54, "util_count_spearman": -0.11584645327763185}, "t": 748.9}
{"event": "eval", "step": 800, "phase": 1, "bpb": {"en": 2.45592, "tr": 2.77776, "fi": 2.98237}, "bytes_per_chunk": {"en": 4.956, "tr": 5.004, "fi": 4.768}, "bpb_mean": 2.73868, "frozen_inventory": true, "t": 752.5}
{"event": "train", "step": 800, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.86511, "bpb": 2.69078, "chunk_loss": 0.030835, "F": 0.20691, "G": 0.24955, "chunks": 211.88, "bytes_per_chunk": 4.833, "boundary_entropy": 0.284, "router_noise": 0.0, "grad_norm": 1.248, "lr": 0.0002949, "router_trainable": true, "sec_per_step": 1.194, "probe_sec": 0.1161, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.09503, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": 0.03294, "n_shadow": 5, "mean_delta_shadow": -0.05391, "n_split": 0, "mean_delta_split": NaN, "n_merge": 1, "mean_delta_merge": 0.06702, "n_dropped_unpaired": 0, "n_full": 6, "trunc_ratio_sum": 1.18774, "trunc_abs_frac": 0.45313, "mean_delta_full": -0.02869, "read_hit_rate": {"surface": 0.0, "split": 0.02517, "merge": 0.02204, "skel": 0.04481}, "mean_gate": {"surface": null, "split": 0.09998, "merge": 0.11496, "skel": 0.10433}, "n_reads": 656, "n_cand": 29765, "read_suppressed_frac": 0.0899, "dropout_reads_logged": 656, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 122, "admitted_surface": 5017, "staged_surface": 63397, "admitted_split": 3, "staged_split": 50097, "admitted_merge": 9958, "staged_merge": 101400, "admitted_skel": 50558, "staged_skel": 47250, "mean_count": 6.9559326171875, "n_util_obs": 54, "util_count_spearman": -0.11950368822783113}, "probe_overhead_sec_total": 78.999, "probe_overhead_seq_frac": 0.4993, "util_loss": 0.006023, "alpha": 0.03, "t": 759.3}
{"event": "consolidate", "step": 850, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262086, "t": 812.1}
{"event": "train", "step": 850, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.81335, "bpb": 2.61611, "chunk_loss": 0.03085, "F": 0.20776, "G": 0.24414, "chunks": 212.75, "bytes_per_chunk": 4.813, "boundary_entropy": 0.27711, "router_noise": 0.0, "grad_norm": 1.0743, "lr": 0.0002938, "router_trainable": true, "sec_per_step": 1.069, "probe_sec": 0.0776, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.05778, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": -0.05134, "n_shadow": 3, "mean_delta_shadow": -0.01027, "n_split": 0, "mean_delta_split": NaN, "n_merge": 1, "mean_delta_merge": 0.06659, "n_dropped_unpaired": 0, "n_full": 1, "trunc_ratio_sum": 1.28957, "trunc_abs_frac": 1.28957, "mean_delta_full": 0.51514, "read_hit_rate": {"surface": 0.0, "split": 0.01358, "merge": 0.0126, "skel": 0.04398}, "mean_gate": {"surface": null, "split": 0.10129, "merge": 0.11634, "skel": 0.10239}, "n_reads": 480, "n_cand": 30695, "read_suppressed_frac": 0.1125, "dropout_reads_logged": 480, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 90, "admitted_surface": 5042, "staged_surface": 62642, "admitted_split": 2, "staged_split": 48867, "admitted_merge": 10176, "staged_merge": 104164, "admitted_skel": 50316, "staged_skel": 46471, "mean_count": 5.559173583984375, "n_util_obs": 51, "util_count_spearman": -0.23385673583832373}, "probe_overhead_sec_total": 84.948, "probe_overhead_seq_frac": 0.4993, "util_loss": 0.005718, "alpha": 0.03, "t": 812.7}
{"event": "consolidate", "step": 900, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262084, "t": 857.3}
{"event": "train", "step": 900, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.61653, "bpb": 2.33216, "chunk_loss": 0.030743, "F": 0.20834, "G": 0.24278, "chunks": 213.34, "bytes_per_chunk": 4.8, "boundary_entropy": 0.26876, "router_noise": 0.0, "grad_norm": 1.2428, "lr": 0.0002927, "router_trainable": true, "sec_per_step": 0.902, "probe_sec": 0.0783, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.05059, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": 0.04615, "n_shadow": 7, "mean_delta_shadow": -0.00357, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": -0.32841, "trunc_abs_frac": 0.3161, "mean_delta_full": 0.12347, "read_hit_rate": {"surface": 0.0, "split": 0.01422, "merge": 0.0105, "skel": 0.03363}, "mean_gate": {"surface": null, "split": 0.09808, "merge": 0.12085, "skel": 0.11719}, "n_reads": 422, "n_cand": 30934, "read_suppressed_frac": 0.0806, "dropout_reads_logged": 422, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 84, "admitted_surface": 4741, "staged_surface": 62039, "admitted_split": 2, "staged_split": 48130, "admitted_merge": 10056, "staged_merge": 106471, "admitted_skel": 50737, "staged_skel": 45504, "mean_count": 5.010894775390625, "n_util_obs": 55, "util_count_spearman": -0.1257598441734919}, "probe_overhead_sec_total": 89.541, "probe_overhead_seq_frac": 0.4994, "util_loss": 0.005476, "alpha": 0.03, "t": 857.8}
{"event": "consolidate", "step": 950, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262080, "t": 912.9}
{"event": "train", "step": 950, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.68753, "bpb": 2.4346, "chunk_loss": 0.030576, "F": 0.19913, "G": 0.23558, "chunks": 203.91, "bytes_per_chunk": 5.022, "boundary_entropy": 0.26128, "router_noise": 0.0, "grad_norm": 1.0703, "lr": 0.0002914, "router_trainable": true, "sec_per_step": 1.122, "probe_sec": 0.0545, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.03974, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": 0.03202, "n_shadow": 6, "mean_delta_shadow": -0.11203, "n_split": 0, "mean_delta_split": NaN, "n_merge": 1, "mean_delta_merge": 0.20166, "n_dropped_unpaired": 0, "n_full": 1, "trunc_ratio_sum": 1.15921, "trunc_abs_frac": 1.15921, "mean_delta_full": -0.38892, "read_hit_rate": {"surface": 0.0, "split": 0.0, "merge": 0.00124, "skel": 0.01443}, "mean_gate": {"surface": null, "split": null, "merge": 0.14284, "skel": 0.12108}, "n_reads": 81, "n_cand": 29302, "read_suppressed_frac": 0.0988, "dropout_reads_logged": 81, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 99, "admitted_surface": 4739, "staged_surface": 61527, "admitted_split": 0, "staged_split": 48044, "admitted_merge": 9728, "staged_merge": 107332, "admitted_skel": 51069, "staged_skel": 45241, "mean_count": 1.8720245361328125, "n_util_obs": 46, "util_count_spearman": -0.0027754649022469014}, "probe_overhead_sec_total": 94.843, "probe_overhead_seq_frac": 0.4994, "util_loss": 0.005508, "alpha": 0.03, "t": 913.9}
{"event": "consolidate", "step": 1000, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262075, "t": 970.3}
{"event": "probe", "step": 1000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3362.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.193185, "cps": 26.265625, "mean_p": 0.229915, "boundary_entropy": 0.261919, "bpb": 2.352264}, "tr": {"chunks": 4069.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.224769, "cps": 31.789062, "mean_p": 0.229289, "boundary_entropy": 0.229923, "bpb": 2.648728}, "fi": {"chunks": 4281.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.22758, "cps": 33.445312, "mean_p": 0.252979, "boundary_entropy": 0.25818, "bpb": 2.845735}}, "bpb": {"en": 2.35226, "tr": 2.64873, "fi": 2.84574}, "gini_cpc": 0.03552, "gini_cps": 0.052311, "bpic": 4.843921, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.1634923965885868, "fi": 1.1780392362854508}, "inventory": {"size": 326656, "admitted": 65536, "staged": 261120, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 115, "admitted_surface": 4848, "staged_surface": 60967, "admitted_split": 0, "staged_split": 47887, "admitted_merge": 9694, "staged_merge": 107142, "admitted_skel": 50994, "staged_skel": 45124, "mean_count": 1.261962890625, "n_util_obs": 16, "util_count_spearman": 0.23529411764705882}, "t": 972.1}
{"event": "eval", "step": 1000, "phase": 1, "bpb": {"en": 2.36027, "tr": 2.65637, "fi": 2.85225}, "bytes_per_chunk": {"en": 5.226, "tr": 4.916, "fi": 4.717}, "bpb_mean": 2.62296, "frozen_inventory": true, "t": 974.9}
{"event": "train", "step": 1000, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.74683, "bpb": 2.52015, "chunk_loss": 0.030507, "F": 0.19897, "G": 0.23038, "chunks": 203.75, "bytes_per_chunk": 5.026, "boundary_entropy": 0.24557, "router_noise": 0.0, "grad_norm": 1.0284, "lr": 0.0002901, "router_trainable": true, "sec_per_step": 1.353, "probe_sec": 0.0904, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.06827, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 0, "mean_delta_loo": NaN, "n_shadow": 10, "mean_delta_shadow": -0.03801, "n_split": 0, "mean_delta_split": NaN, "n_merge": 2, "mean_delta_merge": 0.10993, "n_dropped_unpaired": 0, "n_full": 1, "trunc_ratio_sum": -4.20968, "trunc_abs_frac": 4.20968, "mean_delta_full": -0.00757, "read_hit_rate": {"surface": 0.0, "split": 0.0, "merge": 0.00015, "skel": 0.00667}, "mean_gate": {"surface": null, "split": null, "merge": 0.18131, "skel": 0.13717}, "n_reads": 36, "n_cand": 29613, "read_suppressed_frac": 0.0833, "dropout_reads_logged": 36, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 116, "admitted_surface": 4848, "staged_surface": 61229, "admitted_split": 0, "staged_split": 48017, "admitted_merge": 9694, "staged_merge": 107589, "admitted_skel": 50994, "staged_skel": 45309, "mean_count": 1.26251220703125, "n_util_obs": 16, "util_count_spearman": 0.23529411764705882}, "probe_overhead_sec_total": 99.921, "probe_overhead_seq_frac": 0.4994, "util_loss": 0.006738, "alpha": 0.03, "t": 981.5}
{"event": "consolidate", "step": 1050, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262069, "t": 1037.2}
{"event": "train", "step": 1050, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.6416, "bpb": 2.36833, "chunk_loss": 0.030557, "F": 0.19318, "G": 0.22704, "chunks": 197.81, "bytes_per_chunk": 5.177, "boundary_entropy": 0.25171, "router_noise": 0.0, "grad_norm": 1.0001, "lr": 0.0002886, "router_trainable": true, "sec_per_step": 1.141, "probe_sec": 0.1773, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.07326, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": 0.01115, "n_shadow": 7, "mean_delta_shadow": -0.08035, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 0.8042, "trunc_abs_frac": 0.82862, "mean_delta_full": -0.16693, "read_hit_rate": {"surface": 0.0, "split": 0.0, "merge": 0.00032, "skel": 0.02981}, "mean_gate": {"surface": null, "split": null, "merge": 0.19539, "skel": 0.11595}, "n_reads": 148, "n_cand": 28349, "read_suppressed_frac": 0.0743, "dropout_reads_logged": 148, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 139, "admitted_surface": 4897, "staged_surface": 61033, "admitted_split": 0, "staged_split": 47698, "admitted_merge": 9829, "staged_merge": 108048, "admitted_skel": 50810, "staged_skel": 45365, "mean_count": 1.533447265625, "n_util_obs": 17, "util_count_spearman": 0.18382352941176472}, "probe_overhead_sec_total": 105.427, "probe_overhead_seq_frac": 0.4995, "util_loss": 0.006098, "alpha": 0.03, "t": 1038.6}
{"event": "consolidate", "step": 1100, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262067, "t": 1094.6}
{"event": "train", "step": 1100, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.74572, "bpb": 2.51854, "chunk_loss": 0.030589, "F": 0.20093, "G": 0.23138, "chunks": 205.75, "bytes_per_chunk": 4.977, "boundary_entropy": 0.2503, "router_noise": 0.0, "grad_norm": 1.0414, "lr": 0.0002871, "router_trainable": true, "sec_per_step": 1.142, "probe_sec": 0.102, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.06559, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 5, "mean_delta_loo": -0.0045, "n_shadow": 3, "mean_delta_shadow": 0.06829, "n_split": 0, "mean_delta_split": NaN, "n_merge": 3, "mean_delta_merge": 0.0768, "n_dropped_unpaired": 0, "n_full": 1, "trunc_ratio_sum": -0.1367, "trunc_abs_frac": 0.1367, "mean_delta_full": 0.19556, "read_hit_rate": {"surface": 0.0, "split": 0.0, "merge": 0.00123, "skel": 0.00698}, "mean_gate": {"surface": null, "split": null, "merge": 0.16885, "skel": 0.11772}, "n_reads": 43, "n_cand": 29455, "read_suppressed_frac": 0.093, "dropout_reads_logged": 43, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 129, "admitted_surface": 4902, "staged_surface": 60819, "admitted_split": 0, "staged_split": 47754, "admitted_merge": 9967, "staged_merge": 108415, "admitted_skel": 50667, "staged_skel": 45156, "mean_count": 1.2949676513671875, "n_util_obs": 18, "util_count_spearman": 0.030975739801653728}, "probe_overhead_sec_total": 111.617, "probe_overhead_seq_frac": 0.4995, "util_loss": 0.00701, "alpha": 0.03, "t": 1095.7}
{"event": "consolidate", "step": 1150, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262056, "t": 1153.7}
{"event": "train", "step": 1150, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.64857, "bpb": 2.37839, "chunk_loss": 0.030499, "F": 0.19968, "G": 0.22722, "chunks": 204.47, "bytes_per_chunk": 5.008, "boundary_entropy": 0.23739, "router_noise": 0.0, "grad_norm": 0.8996, "lr": 0.0002855, "router_trainable": true, "sec_per_step": 1.17, "probe_sec": 0.0842, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.0657, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 3, "mean_delta_loo": 0.23667, "n_shadow": 7, "mean_delta_shadow": 0.04034, "n_split": 0, "mean_delta_split": NaN, "n_merge": 2, "mean_delta_merge": 0.02665, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 1.06131, "trunc_abs_frac": 1.05777, "mean_delta_full": 0.28735, "read_hit_rate": {"surface": 0.0, "split": 0.0, "merge": 0.00571, "skel": 0.00441}, "mean_gate": {"surface": null, "split": null, "merge": 0.11823, "skel": 0.1264}, "n_reads": 60, "n_cand": 29738, "read_suppressed_frac": 0.05, "dropout_reads_logged": 60, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 150, "admitted_surface": 3988, "staged_surface": 60784, "admitted_split": 0, "staged_split": 47610, "admitted_merge": 10074, "staged_merge": 108502, "admitted_skel": 51474, "staged_skel": 45248, "mean_count": 1.37152099609375, "n_util_obs": 16, "util_count_spearman": 0.0}, "probe_overhead_sec_total": 116.725, "probe_overhead_seq_frac": 0.4995, "util_loss": 0.007401, "alpha": 0.03, "t": 1154.2}
{"event": "consolidate", "step": 1200, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262050, "t": 1199.9}
{"event": "probe", "step": 1200, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3383.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.194392, "cps": 26.429688, "mean_p": 0.228201, "boundary_entropy": 0.250874, "bpb": 2.27854}, "tr": {"chunks": 3958.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.218638, "cps": 30.921875, "mean_p": 0.221807, "boundary_entropy": 0.22566, "bpb": 2.564225}, "fi": {"chunks": 4343.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.230876, "cps": 33.929688, "mean_p": 0.249028, "boundary_entropy": 0.233549, "bpb": 2.747833}}, "bpb": {"en": 2.27854, "tr": 2.56422, "fi": 2.74783}, "gini_cpc": 0.037773, "gini_cps": 0.054776, "bpic": 4.855529, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.1247276215227775, "fi": 1.1876817095575223}, "inventory": {"size": 326656, "admitted": 65536, "staged": 261120, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 159, "admitted_surface": 3990, "staged_surface": 60634, "admitted_split": 4, "staged_split": 47276, "admitted_merge": 9053, "staged_merge": 108147, "admitted_skel": 52489, "staged_skel": 45063, "mean_count": 4.1381683349609375, "n_util_obs": 22, "util_count_spearman": -0.14850367024280067}, "t": 1201.6}
{"event": "eval", "step": 1200, "phase": 1, "bpb": {"en": 2.2907, "tr": 2.57243, "fi": 2.75513}, "bytes_per_chunk": {"en": 5.173, "tr": 5.095, "fi": 4.694}, "bpb_mean": 2.53942, "frozen_inventory": true, "t": 1204.3}
{"event": "train", "step": 1200, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.62659, "bpb": 2.34667, "chunk_loss": 0.030467, "F": 0.19757, "G": 0.22639, "chunks": 202.31, "bytes_per_chunk": 5.061, "boundary_entropy": 0.24053, "router_noise": 0.0, "grad_norm": 0.9482, "lr": 0.0002837, "router_trainable": true, "sec_per_step": 1.114, "probe_sec": 0.088, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.06791, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": 0.06326, "n_shadow": 4, "mean_delta_shadow": -0.03564, "n_split": 0, "mean_delta_split": NaN, "n_merge": 2, "mean_delta_merge": 0.02213, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": 0.15059, "trunc_abs_frac": 0.47454, "mean_delta_full": 0.06047, "read_hit_rate": {"surface": 0.00109, "split": 0.00612, "merge": 0.00655, "skel": 0.08255}, "mean_gate": {"surface": 0.0927, "split": 0.1251, "merge": 0.12159, "skel": 0.11367}, "n_reads": 540, "n_cand": 29232, "read_suppressed_frac": 0.0963, "dropout_reads_logged": 540, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 161, "admitted_surface": 3990, "staged_surface": 60800, "admitted_split": 4, "staged_split": 47429, "admitted_merge": 9053, "staged_merge": 108637, "admitted_skel": 52489, "staged_skel": 45278, "mean_count": 4.1464080810546875, "n_util_obs": 22, "util_count_spearman": -0.14737436476566912}, "probe_overhead_sec_total": 121.794, "probe_overhead_seq_frac": 0.4995, "util_loss": 0.006863, "alpha": 0.03, "t": 1209.9}
{"event": "consolidate", "step": 1250, "retired": 0, "split_retired": 1, "merge_admitted": 0, "admitted": 1, "swapped": 1024, "uncertain_staged": 262051, "t": 1266.3}
{"event": "train", "step": 1250, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.64322, "bpb": 2.37066, "chunk_loss": 0.030366, "F": 0.19434, "G": 0.22016, "chunks": 199.0, "bytes_per_chunk": 5.146, "boundary_entropy": 0.22742, "router_noise": 0.0, "grad_norm": 0.9625, "lr": 0.0002819, "router_trainable": true, "sec_per_step": 1.138, "probe_sec": 0.0552, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.03871, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": 0.06923, "n_shadow": 5, "mean_delta_shadow": -0.03951, "n_split": 0, "mean_delta_split": NaN, "n_merge": 1, "mean_delta_merge": -0.03284, "n_dropped_unpaired": 0, "n_full": 1, "trunc_ratio_sum": 5.14039, "trunc_abs_frac": 5.14039, "mean_delta_full": 0.04956, "read_hit_rate": {"surface": 0.00032, "split": 0.00609, "merge": 0.00428, "skel": 0.06851}, "mean_gate": {"surface": 0.12363, "split": 0.10541, "merge": 0.10001, "skel": 0.09064}, "n_reads": 440, "n_cand": 28659, "read_suppressed_frac": 0.125, "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": 123, "admitted_surface": 4756, "staged_surface": 59855, "admitted_split": 5, "staged_split": 47690, "admitted_merge": 9090, "staged_merge": 108739, "admitted_skel": 51685, "staged_skel": 45860, "mean_count": 4.7156982421875, "n_util_obs": 24, "util_count_spearman": -0.1373913043478261}, "probe_overhead_sec_total": 126.912, "probe_overhead_seq_frac": 0.4996, "util_loss": 0.007067, "alpha": 0.03, "t": 1266.8}
{"event": "consolidate", "step": 1300, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262047, "t": 1328.6}
{"event": "train", "step": 1300, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.67214, "bpb": 2.41239, "chunk_loss": 0.030671, "F": 0.20758, "G": 0.23182, "chunks": 212.56, "bytes_per_chunk": 4.817, "boundary_entropy": 0.22127, "router_noise": 0.0, "grad_norm": 0.9762, "lr": 0.00028, "router_trainable": true, "sec_per_step": 1.26, "probe_sec": 0.0926, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.06384, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": 0.02864, "n_shadow": 4, "mean_delta_shadow": 0.00841, "n_split": 0, "mean_delta_split": NaN, "n_merge": 2, "mean_delta_merge": 0.01743, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 0.25083, "trunc_abs_frac": 0.09962, "mean_delta_full": 0.12297, "read_hit_rate": {"surface": 0.0, "split": 0.0, "merge": 0.0, "skel": 0.02354}, "mean_gate": {"surface": null, "split": null, "merge": null, "skel": 0.07686}, "n_reads": 125, "n_cand": 30567, "read_suppressed_frac": 0.104, "dropout_reads_logged": 125, "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": 5774, "staged_surface": 59637, "admitted_split": 0, "staged_split": 47435, "admitted_merge": 9066, "staged_merge": 109084, "admitted_skel": 50696, "staged_skel": 45988, "mean_count": 1.4852142333984375, "n_util_obs": 13, "util_count_spearman": 0.027472527472527472}, "probe_overhead_sec_total": 132.184, "probe_overhead_seq_frac": 0.4996, "util_loss": 0.00669, "alpha": 0.03, "t": 1329.8}
{"event": "consolidate", "step": 1350, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262042, "t": 1378.5}
{"event": "train", "step": 1350, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.60616, "bpb": 2.31721, "chunk_loss": 0.030466, "F": 0.19772, "G": 0.22231, "chunks": 202.47, "bytes_per_chunk": 5.058, "boundary_entropy": 0.22735, "router_noise": 0.0, "grad_norm": 0.9473, "lr": 0.000278, "router_trainable": true, "sec_per_step": 1.0, "probe_sec": 0.0653, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04157, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 0, "mean_delta_loo": NaN, "n_shadow": 12, "mean_delta_shadow": 0.05622, "n_split": 0, "mean_delta_split": NaN, "n_merge": 1, "mean_delta_merge": 0.03125, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 0.66493, "trunc_abs_frac": 0.59189, "mean_delta_full": 0.13261, "read_hit_rate": {"surface": 0.0, "split": 0.0, "merge": 0.00016, "skel": 0.00158}, "mean_gate": {"surface": null, "split": null, "merge": 0.15856, "skel": 0.12602}, "n_reads": 9, "n_cand": 29085, "read_suppressed_frac": 0.1111, "dropout_reads_logged": 9, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 97, "admitted_surface": 6583, "staged_surface": 59710, "admitted_split": 0, "staged_split": 47592, "admitted_merge": 9274, "staged_merge": 108950, "admitted_skel": 49679, "staged_skel": 45892, "mean_count": 1.1727294921875, "n_util_obs": 11, "util_count_spearman": 0.00909090909090909}, "probe_overhead_sec_total": 136.943, "probe_overhead_seq_frac": 0.4996, "util_loss": 0.006505, "alpha": 0.03, "t": 1379.8}
{"event": "consolidate", "step": 1400, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262034, "t": 1434.9}
{"event": "probe", "step": 1400, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3462.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.198931, "cps": 27.046875, "mean_p": 0.225027, "boundary_entropy": 0.231821, "bpb": 2.220244}, "tr": {"chunks": 3857.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.213059, "cps": 30.132812, "mean_p": 0.217767, "boundary_entropy": 0.203496, "bpb": 2.493304}, "fi": {"chunks": 4473.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.237786, "cps": 34.945312, "mean_p": 0.245666, "boundary_entropy": 0.207409, "bpb": 2.649062}}, "bpb": {"en": 2.22024, "tr": 2.4933, "fi": 2.64906}, "gini_cpc": 0.039865, "gini_cps": 0.057157, "bpic": 4.811058, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0710164568604206, "fi": 1.1953196841665064}, "inventory": {"size": 326656, "admitted": 65536, "staged": 261120, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 127, "admitted_surface": 7052, "staged_surface": 59617, "admitted_split": 0, "staged_split": 47536, "admitted_merge": 9806, "staged_merge": 108236, "admitted_skel": 48678, "staged_skel": 45731, "mean_count": 1.1201324462890625, "n_util_obs": 7, "util_count_spearman": 0.10714285714285714}, "t": 1436.5}
{"event": "eval", "step": 1400, "phase": 1, "bpb": {"en": 2.23018, "tr": 2.50037, "fi": 2.65457}, "bytes_per_chunk": {"en": 5.193, "tr": 5.269, "fi": 4.577}, "bpb_mean": 2.46171, "frozen_inventory": true, "t": 1439.2}
{"event": "train", "step": 1400, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.60645, "bpb": 2.31762, "chunk_loss": 0.030434, "F": 0.19452, "G": 0.21991, "chunks": 199.19, "bytes_per_chunk": 5.141, "boundary_entropy": 0.21648, "router_noise": 0.0, "grad_norm": 0.8797, "lr": 0.0002759, "router_trainable": true, "sec_per_step": 1.28, "probe_sec": 0.0863, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.06722, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 2, "mean_delta_loo": -0.3992, "n_shadow": 8, "mean_delta_shadow": 0.09406, "n_split": 0, "mean_delta_split": NaN, "n_merge": 2, "mean_delta_merge": -0.02647, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 0.15866, "trunc_abs_frac": 0.33245, "mean_delta_full": 0.15799, "read_hit_rate": {"surface": 0.0, "split": 0.0, "merge": 0.0, "skel": 0.00179}, "mean_gate": {"surface": null, "split": null, "merge": null, "skel": 0.13895}, "n_reads": 9, "n_cand": 28734, "read_suppressed_frac": 0.1111, "dropout_reads_logged": 9, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 129, "admitted_surface": 7052, "staged_surface": 59873, "admitted_split": 0, "staged_split": 47730, "admitted_merge": 9806, "staged_merge": 108677, "admitted_skel": 48678, "staged_skel": 45864, "mean_count": 1.120269775390625, "n_util_obs": 7, "util_count_spearman": 0.10714285714285714}, "probe_overhead_sec_total": 142.221, "probe_overhead_seq_frac": 0.4996, "util_loss": 0.005453, "alpha": 0.03, "t": 1443.8}
{"event": "consolidate", "step": 1450, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262028, "t": 1504.8}
{"event": "train", "step": 1450, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.65592, "bpb": 2.38898, "chunk_loss": 0.030408, "F": 0.19724, "G": 0.21803, "chunks": 201.97, "bytes_per_chunk": 5.07, "boundary_entropy": 0.20998, "router_noise": 0.0, "grad_norm": 0.8855, "lr": 0.0002738, "router_trainable": true, "sec_per_step": 1.238, "probe_sec": 0.1149, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.08652, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 4, "mean_delta_loo": 0.18639, "n_shadow": 10, "mean_delta_shadow": 0.03477, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 0.25702, "trunc_abs_frac": 0.44867, "mean_delta_full": -0.14355, "read_hit_rate": {"surface": 0.0, "split": 0.0, "merge": 0.0, "skel": 0.00455}, "mean_gate": {"surface": null, "split": null, "merge": null, "skel": 0.10321}, "n_reads": 23, "n_cand": 29041, "read_suppressed_frac": 0.2609, "dropout_reads_logged": 23, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 135, "admitted_surface": 7081, "staged_surface": 60246, "admitted_split": 0, "staged_split": 47504, "admitted_merge": 10496, "staged_merge": 108496, "admitted_skel": 47959, "staged_skel": 45898, "mean_count": 1.1659698486328125, "n_util_obs": 7, "util_count_spearman": 0.21428571428571427}, "probe_overhead_sec_total": 147.436, "probe_overhead_seq_frac": 0.4996, "util_loss": 0.00508, "alpha": 0.03, "t": 1505.7}
{"event": "consolidate", "step": 1500, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262025, "t": 1558.8}
{"event": "train", "step": 1500, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.67238, "bpb": 2.41274, "chunk_loss": 0.030597, "F": 0.20386, "G": 0.22548, "chunks": 208.75, "bytes_per_chunk": 4.905, "boundary_entropy": 0.20868, "router_noise": 0.0, "grad_norm": 0.9222, "lr": 0.0002715, "router_trainable": true, "sec_per_step": 1.085, "probe_sec": 0.0899, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.05224, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 10, "mean_delta_loo": 0.02219, "n_shadow": 4, "mean_delta_shadow": 0.00429, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 1, "trunc_ratio_sum": 0.31127, "trunc_abs_frac": 0.31127, "mean_delta_full": 0.09912, "read_hit_rate": {"surface": 0.00015, "split": 0.00123, "merge": 0.00786, "skel": 0.05897}, "mean_gate": {"surface": 0.09784, "split": 0.10422, "merge": 0.10155, "skel": 0.09783}, "n_reads": 365, "n_cand": 29681, "read_suppressed_frac": 0.1178, "dropout_reads_logged": 365, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 166, "admitted_surface": 7081, "staged_surface": 60106, "admitted_split": 3, "staged_split": 47599, "admitted_merge": 11286, "staged_merge": 108641, "admitted_skel": 47166, "staged_skel": 45798, "mean_count": 2.3324127197265625, "n_util_obs": 8, "util_count_spearman": 0.023809523809523808}, "probe_overhead_sec_total": 152.348, "probe_overhead_seq_frac": 0.4996, "util_loss": 0.006005, "alpha": 0.03, "t": 1560.0}
{"event": "consolidate", "step": 1550, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262024, "t": 1611.8}
{"event": "train", "step": 1550, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.58705, "bpb": 2.28962, "chunk_loss": 0.030565, "F": 0.2023, "G": 0.22456, "chunks": 207.16, "bytes_per_chunk": 4.943, "boundary_entropy": 0.20718, "router_noise": 0.0, "grad_norm": 0.9366, "lr": 0.0002692, "router_trainable": true, "sec_per_step": 1.047, "probe_sec": 0.0952, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.07136, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": 0.0535, "n_shadow": 4, "mean_delta_shadow": 0.01297, "n_split": 0, "mean_delta_split": NaN, "n_merge": 2, "mean_delta_merge": 0.12186, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 0.19464, "trunc_abs_frac": 0.30657, "mean_delta_full": 0.18494, "read_hit_rate": {"surface": 0.0, "split": 0.00165, "merge": 0.00366, "skel": 0.05635}, "mean_gate": {"surface": null, "split": 0.13331, "merge": 0.10161, "skel": 0.08169}, "n_reads": 337, "n_cand": 29869, "read_suppressed_frac": 0.092, "dropout_reads_logged": 337, "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": 8090, "staged_surface": 59270, "admitted_split": 3, "staged_split": 47829, "admitted_merge": 11290, "staged_merge": 108995, "admitted_skel": 46153, "staged_skel": 46050, "mean_count": 2.7960205078125, "n_util_obs": 14, "util_count_spearman": 0.002197802197802198}, "probe_overhead_sec_total": 157.476, "probe_overhead_seq_frac": 0.4996, "util_loss": 0.0063, "alpha": 0.03, "t": 1612.3}
{"event": "consolidate", "step": 1600, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262021, "t": 1674.5}
{"event": "probe", "step": 1600, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3353.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.192668, "cps": 26.195312, "mean_p": 0.213386, "boundary_entropy": 0.215957, "bpb": 2.181373}, "tr": {"chunks": 3838.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.212009, "cps": 29.984375, "mean_p": 0.212388, "boundary_entropy": 0.19419, "bpb": 2.429568}, "fi": {"chunks": 4389.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.233321, "cps": 34.289062, "mean_p": 0.242515, "boundary_entropy": 0.197223, "bpb": 2.593626}}, "bpb": {"en": 2.18137, "tr": 2.42957, "fi": 2.59363}, "gini_cpc": 0.04248, "gini_cps": 0.059643, "bpic": 4.899136, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.100385821207766, "fi": 1.2110003374963056}, "inventory": {"size": 326656, "admitted": 65536, "staged": 261120, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 133, "admitted_surface": 9102, "staged_surface": 59100, "admitted_split": 2, "staged_split": 47888, "admitted_merge": 11292, "staged_merge": 108301, "admitted_skel": 45140, "staged_skel": 45831, "mean_count": 1.6440277099609375, "n_util_obs": 14, "util_count_spearman": -0.26153846153846155}, "t": 1676.1}
{"event": "eval", "step": 1600, "phase": 1, "bpb": {"en": 2.19288, "tr": 2.43776, "fi": 2.59794}, "bytes_per_chunk": {"en": 5.265, "tr": 5.256, "fi": 4.644}, "bpb_mean": 2.40953, "frozen_inventory": true, "t": 1679.1}
{"event": "train", "step": 1600, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.52078, "bpb": 2.19402, "chunk_loss": 0.03032, "F": 0.19312, "G": 0.21355, "chunks": 197.75, "bytes_per_chunk": 5.178, "boundary_entropy": 0.20891, "router_noise": 0.0, "grad_norm": 0.8644, "lr": 0.0002668, "router_trainable": true, "sec_per_step": 1.452, "probe_sec": 0.0708, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04989, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 4, "mean_delta_loo": 0.1107, "n_shadow": 6, "mean_delta_shadow": -0.02488, "n_split": 0, "mean_delta_split": NaN, "n_merge": 2, "mean_delta_merge": 0.0907, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": -0.96391, "trunc_abs_frac": 0.70366, "mean_delta_full": 0.0345, "read_hit_rate": {"surface": 0.00064, "split": 0.00062, "merge": 0.00016, "skel": 0.01653}, "mean_gate": {"surface": 0.12845, "split": 0.08374, "merge": 0.10389, "skel": 0.08773}, "n_reads": 97, "n_cand": 28960, "read_suppressed_frac": 0.0619, "dropout_reads_logged": 97, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 132, "admitted_surface": 9102, "staged_surface": 59329, "admitted_split": 2, "staged_split": 47905, "admitted_merge": 11292, "staged_merge": 108846, "admitted_skel": 45140, "staged_skel": 46064, "mean_count": 1.6455078125, "n_util_obs": 14, "util_count_spearman": -0.26153846153846155}, "probe_overhead_sec_total": 162.725, "probe_overhead_seq_frac": 0.4996, "util_loss": 0.006149, "alpha": 0.03, "t": 1684.9}
{"event": "consolidate", "step": 1650, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262015, "t": 1736.3}
{"event": "train", "step": 1650, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.58196, "bpb": 2.28228, "chunk_loss": 0.030404, "F": 0.19794, "G": 0.21896, "chunks": 202.69, "bytes_per_chunk": 5.052, "boundary_entropy": 0.21154, "router_noise": 0.0, "grad_norm": 0.8886, "lr": 0.0002643, "router_trainable": true, "sec_per_step": 1.06, "probe_sec": 0.0983, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.05896, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 0, "mean_delta_loo": NaN, "n_shadow": 8, "mean_delta_shadow": 0.02516, "n_split": 0, "mean_delta_split": NaN, "n_merge": 3, "mean_delta_merge": -0.0283, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": -0.10035, "trunc_abs_frac": 0.30657, "mean_delta_full": 0.17242, "read_hit_rate": {"surface": 0.0, "split": 0.0, "merge": 0.00016, "skel": 0.00116}, "mean_gate": {"surface": null, "split": null, "merge": 0.09717, "skel": 0.14696}, "n_reads": 7, "n_cand": 29341, "read_suppressed_frac": 0.0, "dropout_reads_logged": 7, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 122, "admitted_surface": 10125, "staged_surface": 59052, "admitted_split": 1, "staged_split": 47607, "admitted_merge": 11289, "staged_merge": 109540, "admitted_skel": 44121, "staged_skel": 45945, "mean_count": 1.117706298828125, "n_util_obs": 7, "util_count_spearman": 0.39285714285714285}, "probe_overhead_sec_total": 167.447, "probe_overhead_seq_frac": 0.4997, "util_loss": 0.005955, "alpha": 0.03, "t": 1737.9}
{"event": "consolidate", "step": 1700, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262011, "t": 1798.8}
{"event": "train", "step": 1700, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.5451, "bpb": 2.22911, "chunk_loss": 0.030403, "F": 0.19843, "G": 0.21658, "chunks": 203.19, "bytes_per_chunk": 5.04, "boundary_entropy": 0.19576, "router_noise": 0.0, "grad_norm": 0.8155, "lr": 0.0002618, "router_trainable": true, "sec_per_step": 1.24, "probe_sec": 0.0934, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.06026, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 0, "mean_delta_loo": NaN, "n_shadow": 6, "mean_delta_shadow": -0.05503, "n_split": 0, "mean_delta_split": NaN, "n_merge": 4, "mean_delta_merge": 0.05431, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 0.38785, "trunc_abs_frac": 1.48573, "mean_delta_full": -0.42224, "read_hit_rate": {"surface": 0.0, "split": 0.00018, "merge": 0.0, "skel": 0.00079}, "mean_gate": {"surface": null, "split": 0.15087, "merge": null, "skel": 0.11292}, "n_reads": 6, "n_cand": 29174, "read_suppressed_frac": 0.0, "dropout_reads_logged": 6, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 140, "admitted_surface": 11148, "staged_surface": 59166, "admitted_split": 1, "staged_split": 47853, "admitted_merge": 11289, "staged_merge": 109094, "admitted_skel": 43098, "staged_skel": 46031, "mean_count": 1.13262939453125, "n_util_obs": 8, "util_count_spearman": 0.5476190476190477}, "probe_overhead_sec_total": 172.551, "probe_overhead_seq_frac": 0.4997, "util_loss": 0.004202, "alpha": 0.03, "t": 1799.9}
{"event": "consolidate", "step": 1750, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262006, "t": 1857.8}
{"event": "train", "step": 1750, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.57128, "bpb": 2.26688, "chunk_loss": 0.030464, "F": 0.20081, "G": 0.21791, "chunks": 205.62, "bytes_per_chunk": 4.98, "boundary_entropy": 0.19569, "router_noise": 0.0, "grad_norm": 0.8637, "lr": 0.0002591, "router_trainable": true, "sec_per_step": 1.167, "probe_sec": 0.0709, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04798, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 0, "mean_delta_loo": NaN, "n_shadow": 10, "mean_delta_shadow": -0.03926, "n_split": 0, "mean_delta_split": NaN, "n_merge": 2, "mean_delta_merge": 0.00769, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": 1.23484, "trunc_abs_frac": 0.88729, "mean_delta_full": -0.06846, "read_hit_rate": {"surface": 0.00015, "split": 9e-05, "merge": 0.0, "skel": 0.00096}, "mean_gate": {"surface": 0.0845, "split": 0.15196, "merge": null, "skel": 0.16541}, "n_reads": 7, "n_cand": 29715, "read_suppressed_frac": 0.0, "dropout_reads_logged": 7, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 155, "admitted_surface": 12171, "staged_surface": 59067, "admitted_split": 1, "staged_split": 48019, "admitted_merge": 11289, "staged_merge": 109078, "admitted_skel": 42075, "staged_skel": 45980, "mean_count": 1.127532958984375, "n_util_obs": 9, "util_count_spearman": 0.48333333333333334}, "probe_overhead_sec_total": 177.437, "probe_overhead_seq_frac": 0.4997, "util_loss": 0.004258, "alpha": 0.03, "t": 1858.3}
{"event": "consolidate", "step": 1800, "retired": 0, "split_retired": 0, "merge_admitted": 1, "admitted": 0, "swapped": 1024, "uncertain_staged": 261996, "t": 1910.4}
{"event": "probe", "step": 1800, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3435.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.19738, "cps": 26.835938, "mean_p": 0.216186, "boundary_entropy": 0.212444, "bpb": 2.139742}, "tr": {"chunks": 3946.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.217975, "cps": 30.828125, "mean_p": 0.212532, "boundary_entropy": 0.174523, "bpb": 2.369644}, "fi": {"chunks": 4550.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.24188, "cps": 35.546875, "mean_p": 0.244399, "boundary_entropy": 0.184279, "bpb": 2.534994}}, "bpb": {"en": 2.13974, "tr": 2.36964, "fi": 2.53499}, "gini_cpc": 0.045139, "gini_cps": 0.062303, "bpic": 4.755008, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.1043428108009794, "fi": 1.2254536565639496}, "inventory": {"size": 326655, "admitted": 65536, "staged": 261119, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 164, "admitted_surface": 13190, "staged_surface": 58754, "admitted_split": 1, "staged_split": 47485, "admitted_merge": 11289, "staged_merge": 109256, "admitted_skel": 41056, "staged_skel": 45624, "mean_count": 1.2174224853515625, "n_util_obs": 9, "util_count_spearman": 0.5166666666666667}, "t": 1912.1}
{"event": "eval", "step": 1800, "phase": 1, "bpb": {"en": 2.15377, "tr": 2.37952, "fi": 2.54276}, "bytes_per_chunk": {"en": 5.071, "tr": 5.149, "fi": 4.493}, "bpb_mean": 2.35868, "frozen_inventory": true, "t": 1915.1}
{"event": "train", "step": 1800, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.54936, "bpb": 2.23525, "chunk_loss": 0.030338, "F": 0.19345, "G": 0.21028, "chunks": 198.09, "bytes_per_chunk": 5.169, "boundary_entropy": 0.19525, "router_noise": 0.0, "grad_norm": 0.82, "lr": 0.0002564, "router_trainable": true, "sec_per_step": 1.257, "probe_sec": 0.0694, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04264, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 2, "mean_delta_loo": -0.01385, "n_shadow": 10, "mean_delta_shadow": 0.03042, "n_split": 0, "mean_delta_split": NaN, "n_merge": 1, "mean_delta_merge": -0.06171, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 1.3683, "trunc_abs_frac": 3.25888, "mean_delta_full": -0.05237, "read_hit_rate": {"surface": 0.00174, "split": 9e-05, "merge": 0.00175, "skel": 0.00379}, "mean_gate": {"surface": 0.09836, "split": 0.12503, "merge": 0.07466, "skel": 0.11}, "n_reads": 42, "n_cand": 28594, "read_suppressed_frac": 0.119, "dropout_reads_logged": 42, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 163, "admitted_surface": 13190, "staged_surface": 59069, "admitted_split": 1, "staged_split": 47633, "admitted_merge": 11289, "staged_merge": 109675, "admitted_skel": 41056, "staged_skel": 45767, "mean_count": 1.2180633544921875, "n_util_obs": 9, "util_count_spearman": 0.5166666666666667}, "probe_overhead_sec_total": 182.391, "probe_overhead_seq_frac": 0.4997, "util_loss": 0.004449, "alpha": 0.03, "t": 1921.1}
{"event": "consolidate", "step": 1850, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 261991, "t": 1956.7}
{"event": "train", "step": 1850, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.55118, "bpb": 2.23789, "chunk_loss": 0.030545, "F": 0.20331, "G": 0.22021, "chunks": 208.19, "bytes_per_chunk": 4.919, "boundary_entropy": 0.19243, "router_noise": 0.0, "grad_norm": 0.7635, "lr": 0.0002537, "router_trainable": true, "sec_per_step": 0.728, "probe_sec": 0.0904, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.07212, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 5, "mean_delta_loo": 0.07858, "n_shadow": 8, "mean_delta_shadow": -0.05388, "n_split": 1, "mean_delta_split": 0.00555, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 0.49313, "trunc_abs_frac": 0.78875, "mean_delta_full": 0.02962, "read_hit_rate": {"surface": 0.07014, "split": 0.05865, "merge": 0.04365, "skel": 0.09876}, "mean_gate": {"surface": 0.10341, "split": 0.0947, "merge": 0.09363, "skel": 0.08731}, "n_reads": 1932, "n_cand": 29806, "read_suppressed_frac": 0.1014, "dropout_reads_logged": 1932, "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": 13732, "staged_surface": 59390, "admitted_split": 66, "staged_split": 47942, "admitted_merge": 11569, "staged_merge": 109126, "admitted_skel": 40169, "staged_skel": 45686, "mean_count": 22.179168701171875, "n_util_obs": 43, "util_count_spearman": -0.6048021745696164}, "probe_overhead_sec_total": 186.262, "probe_overhead_seq_frac": 0.4997, "util_loss": 0.004836, "alpha": 0.03, "t": 1957.5}
{"event": "consolidate", "step": 1900, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262020, "t": 1998.4}
{"event": "train", "step": 1900, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.5367, "bpb": 2.21699, "chunk_loss": 0.030598, "F": 0.20081, "G": 0.21615, "chunks": 205.62, "bytes_per_chunk": 4.98, "boundary_entropy": 0.19613, "router_noise": 0.0, "grad_norm": 0.8646, "lr": 0.0002508, "router_trainable": true, "sec_per_step": 0.831, "probe_sec": 0.1035, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.0794, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": -0.0734, "n_shadow": 3, "mean_delta_shadow": -0.0416, "n_split": 0, "mean_delta_split": NaN, "n_merge": 2, "mean_delta_merge": 0.01954, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 0.63869, "trunc_abs_frac": 0.95533, "mean_delta_full": -0.16101, "read_hit_rate": {"surface": 0.06765, "split": 0.05557, "merge": 0.03807, "skel": 0.09676}, "mean_gate": {"surface": 0.07336, "split": 0.06949, "merge": 0.06811, "skel": 0.06026}, "n_reads": 1810, "n_cand": 29432, "read_suppressed_frac": 0.0928, "dropout_reads_logged": 1810, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 197, "admitted_surface": 14756, "staged_surface": 58712, "admitted_split": 66, "staged_split": 48242, "admitted_merge": 11081, "staged_merge": 109127, "admitted_skel": 39633, "staged_skel": 46063, "mean_count": 23.594467163085938, "n_util_obs": 43, "util_count_spearman": -0.6010268800966475}, "probe_overhead_sec_total": 191.088, "probe_overhead_seq_frac": 0.4997, "util_loss": 0.00502, "alpha": 0.03, "t": 1999.1}
{"event": "consolidate", "step": 1950, "retired": 0, "split_retired": 1, "merge_admitted": 0, "admitted": 1, "swapped": 1024, "uncertain_staged": 262016, "t": 2032.0}
{"event": "train", "step": 1950, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.4869, "bpb": 2.14514, "chunk_loss": 0.030401, "F": 0.19501, "G": 0.21185, "chunks": 199.69, "bytes_per_chunk": 5.128, "boundary_entropy": 0.19214, "router_noise": 0.0, "grad_norm": 0.7885, "lr": 0.0002479, "router_trainable": true, "sec_per_step": 0.678, "probe_sec": 0.0774, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.05792, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": 0.00701, "n_shadow": 5, "mean_delta_shadow": 0.02889, "n_split": 0, "mean_delta_split": NaN, "n_merge": 1, "mean_delta_merge": -0.12878, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": -0.77313, "trunc_abs_frac": 1.49624, "mean_delta_full": 0.04354, "read_hit_rate": {"surface": 0.06669, "split": 0.06134, "merge": 0.04205, "skel": 0.09021}, "mean_gate": {"surface": 0.06796, "split": 0.06081, "merge": 0.06101, "skel": 0.05289}, "n_reads": 1811, "n_cand": 28617, "read_suppressed_frac": 0.1033, "dropout_reads_logged": 1811, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 200, "admitted_surface": 15780, "staged_surface": 58779, "admitted_split": 66, "staged_split": 47971, "admitted_merge": 10057, "staged_merge": 109286, "admitted_skel": 39633, "staged_skel": 46108, "mean_count": 24.446563720703125, "n_util_obs": 44, "util_count_spearman": -0.5995771670190275}, "probe_overhead_sec_total": 194.858, "probe_overhead_seq_frac": 0.4997, "util_loss": 0.004306, "alpha": 0.03, "t": 2032.9}
{"event": "consolidate", "step": 2000, "retired": 0, "split_retired": 2, "merge_admitted": 0, "admitted": 2, "swapped": 1024, "uncertain_staged": 262015, "t": 2066.3}
{"event": "probe", "step": 2000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3376.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.19399, "cps": 26.375, "mean_p": 0.210523, "boundary_entropy": 0.201108, "bpb": 2.113777}, "tr": {"chunks": 3816.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.210794, "cps": 29.8125, "mean_p": 0.211694, "boundary_entropy": 0.18332, "bpb": 2.315185}, "fi": {"chunks": 4457.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.236936, "cps": 34.820312, "mean_p": 0.240994, "boundary_entropy": 0.17419, "bpb": 2.483591}}, "bpb": {"en": 2.11378, "tr": 2.31519, "fi": 2.48359}, "gini_cpc": 0.044616, "gini_cps": 0.061865, "bpic": 4.870118, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.086624510142463, "fi": 1.2213845804902395}, "inventory": {"size": 326654, "admitted": 65536, "staged": 261118, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 204, "admitted_surface": 16797, "staged_surface": 58054, "admitted_split": 65, "staged_split": 48261, "admitted_merge": 9041, "staged_merge": 108800, "admitted_skel": 39633, "staged_skel": 46003, "mean_count": 24.548538208007812, "n_util_obs": 44, "util_count_spearman": -0.5539112050739958}, "t": 2067.7}
{"event": "eval", "step": 2000, "phase": 1, "bpb": {"en": 2.12517, "tr": 2.32648, "fi": 2.49238}, "bytes_per_chunk": {"en": 5.182, "tr": 5.319, "fi": 4.61}, "bpb_mean": 2.31468, "frozen_inventory": true, "t": 2070.1}
{"event": "train", "step": 2000, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.65064, "bpb": 2.38138, "chunk_loss": 0.0306, "F": 0.20676, "G": 0.22065, "chunks": 211.72, "bytes_per_chunk": 4.837, "boundary_entropy": 0.19073, "router_noise": 0.0, "grad_norm": 0.8354, "lr": 0.000245, "router_trainable": true, "sec_per_step": 0.851, "probe_sec": 0.0978, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.05848, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": 0.04885, "n_shadow": 2, "mean_delta_shadow": -0.00165, "n_split": 3, "mean_delta_split": -0.00043, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 6, "trunc_ratio_sum": 0.08675, "trunc_abs_frac": 0.42394, "mean_delta_full": 0.13045, "read_hit_rate": {"surface": 0.05724, "split": 0.05985, "merge": 0.03055, "skel": 0.09036}, "mean_gate": {"surface": 0.06868, "split": 0.05988, "merge": 0.06428, "skel": 0.048}, "n_reads": 1746, "n_cand": 30123, "read_suppressed_frac": 0.0888, "dropout_reads_logged": 1746, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 204, "admitted_surface": 16797, "staged_surface": 58323, "admitted_split": 65, "staged_split": 48376, "admitted_merge": 9041, "staged_merge": 109395, "admitted_skel": 39633, "staged_skel": 46050, "mean_count": 24.575180053710938, "n_util_obs": 45, "util_count_spearman": -0.5173913043478261}, "probe_overhead_sec_total": 198.489, "probe_overhead_seq_frac": 0.4997, "util_loss": 0.004147, "alpha": 0.03, "t": 2075.5}
{"event": "consolidate", "step": 2050, "retired": 0, "split_retired": 1, "merge_admitted": 0, "admitted": 1, "swapped": 1024, "uncertain_staged": 262014, "t": 2114.8}
{"event": "train", "step": 2050, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.48795, "bpb": 2.14666, "chunk_loss": 0.030439, "F": 0.19281, "G": 0.20793, "chunks": 197.44, "bytes_per_chunk": 5.186, "boundary_entropy": 0.1838, "router_noise": 0.0, "grad_norm": 0.8231, "lr": 0.0002419, "router_trainable": true, "sec_per_step": 0.802, "probe_sec": 0.0615, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04586, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": -0.07249, "n_shadow": 5, "mean_delta_shadow": 0.02454, "n_split": 1, "mean_delta_split": 0.06628, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 1.03881, "trunc_abs_frac": 0.94376, "mean_delta_full": -0.09383, "read_hit_rate": {"surface": 0.0097, "split": 0.00073, "merge": 0.00208, "skel": 0.0134}, "mean_gate": {"surface": 0.06245, "split": 0.07683, "merge": 0.07081, "skel": 0.07183}, "n_reads": 149, "n_cand": 28535, "read_suppressed_frac": 0.1611, "dropout_reads_logged": 149, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 107, "admitted_surface": 17398, "staged_surface": 58199, "admitted_split": 3, "staged_split": 47867, "admitted_merge": 8632, "staged_merge": 109991, "admitted_skel": 39503, "staged_skel": 46087, "mean_count": 1.8256072998046875, "n_util_obs": 11, "util_count_spearman": -0.1}, "probe_overhead_sec_total": 202.221, "probe_overhead_seq_frac": 0.4997, "util_loss": 0.005161, "alpha": 0.03, "t": 2115.6}
{"event": "consolidate", "step": 2100, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262010, "t": 2151.8}
{"event": "train", "step": 2100, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.54736, "bpb": 2.23238, "chunk_loss": 0.030354, "F": 0.19528, "G": 0.20975, "chunks": 199.97, "bytes_per_chunk": 5.121, "boundary_entropy": 0.18185, "router_noise": 0.0, "grad_norm": 0.8523, "lr": 0.0002388, "router_trainable": true, "sec_per_step": 0.74, "probe_sec": 0.0608, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04489, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 3, "mean_delta_loo": -0.2269, "n_shadow": 6, "mean_delta_shadow": 0.00634, "n_split": 1, "mean_delta_split": 0.15019, "n_merge": 2, "mean_delta_merge": -0.00195, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 0.20673, "trunc_abs_frac": 0.54531, "mean_delta_full": -0.0119, "read_hit_rate": {"surface": 0.00864, "split": 0.00064, "merge": 0.00189, "skel": 0.0076}, "mean_gate": {"surface": 0.06445, "split": 0.08567, "merge": 0.07904, "skel": 0.07939}, "n_reads": 112, "n_cand": 28691, "read_suppressed_frac": 0.1161, "dropout_reads_logged": 112, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 127, "admitted_surface": 18374, "staged_surface": 58328, "admitted_split": 3, "staged_split": 48388, "admitted_merge": 7656, "staged_merge": 109069, "admitted_skel": 39503, "staged_skel": 46359, "mean_count": 1.8861846923828125, "n_util_obs": 15, "util_count_spearman": -0.29285714285714287}, "probe_overhead_sec_total": 206.016, "probe_overhead_seq_frac": 0.4997, "util_loss": 0.004855, "alpha": 0.03, "t": 2152.6}
{"event": "consolidate", "step": 2150, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262008, "t": 2186.7}
{"event": "train", "step": 2150, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.51021, "bpb": 2.17877, "chunk_loss": 0.030357, "F": 0.19647, "G": 0.2122, "chunks": 201.19, "bytes_per_chunk": 5.09, "boundary_entropy": 0.18641, "router_noise": 0.0, "grad_norm": 0.7824, "lr": 0.0002357, "router_trainable": true, "sec_per_step": 0.69, "probe_sec": 0.0651, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04807, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 2, "mean_delta_loo": 0.16556, "n_shadow": 10, "mean_delta_shadow": 0.04175, "n_split": 0, "mean_delta_split": NaN, "n_merge": 1, "mean_delta_merge": -0.10901, "n_dropped_unpaired": 0, "n_full": 6, "trunc_ratio_sum": -19.11688, "trunc_abs_frac": 0.64293, "mean_delta_full": -0.0047, "read_hit_rate": {"surface": 0.00781, "split": 0.00027, "merge": 0.00063, "skel": 0.0075}, "mean_gate": {"surface": 0.06518, "split": 0.086, "merge": 0.07661, "skel": 0.0768}, "n_reads": 95, "n_cand": 28970, "read_suppressed_frac": 0.1053, "dropout_reads_logged": 95, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 146, "admitted_surface": 19225, "staged_surface": 58178, "admitted_split": 2, "staged_split": 47997, "admitted_merge": 6872, "staged_merge": 109674, "admitted_skel": 39437, "staged_skel": 46295, "mean_count": 1.8501129150390625, "n_util_obs": 20, "util_count_spearman": -0.23993984397083035}, "probe_overhead_sec_total": 209.553, "probe_overhead_seq_frac": 0.4997, "util_loss": 0.004933, "alpha": 0.03, "t": 2187.1}
{"event": "consolidate", "step": 2200, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262005, "t": 2219.1}
{"event": "probe", "step": 2200, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3453.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.198414, "cps": 26.976562, "mean_p": 0.213002, "boundary_entropy": 0.184181, "bpb": 2.088103}, "tr": {"chunks": 3865.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.213501, "cps": 30.195312, "mean_p": 0.207939, "boundary_entropy": 0.162796, "bpb": 2.289752}, "fi": {"chunks": 4468.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.237521, "cps": 34.90625, "mean_p": 0.23928, "boundary_entropy": 0.160425, "bpb": 2.440355}}, "bpb": {"en": 2.0881, "tr": 2.28975, "fi": 2.44036}, "gini_cpc": 0.040144, "gini_cps": 0.057413, "bpic": 4.813508, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0760352252121548, "fi": 1.1970955678233446}, "inventory": {"size": 326656, "admitted": 65536, "staged": 261120, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 190, "admitted_surface": 20077, "staged_surface": 58236, "admitted_split": 76, "staged_split": 47765, "admitted_merge": 6872, "staged_merge": 109137, "admitted_skel": 38511, "staged_skel": 45982, "mean_count": 3.92828369140625, "n_util_obs": 27, "util_count_spearman": 0.052503052503052504}, "t": 2220.6}
{"event": "eval", "step": 2200, "phase": 1, "bpb": {"en": 2.1003, "tr": 2.29886, "fi": 2.44748}, "bytes_per_chunk": {"en": 5.095, "tr": 5.266, "fi": 4.593}, "bpb_mean": 2.28221, "frozen_inventory": true, "t": 2223.0}
{"event": "train", "step": 2200, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.47978, "bpb": 2.13486, "chunk_loss": 0.030373, "F": 0.19809, "G": 0.21058, "chunks": 202.84, "bytes_per_chunk": 5.048, "boundary_entropy": 0.17717, "router_noise": 0.0, "grad_norm": 0.6883, "lr": 0.0002325, "router_trainable": true, "sec_per_step": 0.801, "probe_sec": 0.0629, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04204, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 4, "mean_delta_loo": -0.04901, "n_shadow": 4, "mean_delta_shadow": -0.02652, "n_split": 0, "mean_delta_split": NaN, "n_merge": 3, "mean_delta_merge": 0.05915, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 0.01921, "trunc_abs_frac": 1.23758, "mean_delta_full": 0.22396, "read_hit_rate": {"surface": 0.00743, "split": 0.03974, "merge": 0.00187, "skel": 0.0994}, "mean_gate": {"surface": 0.07116, "split": 0.06809, "merge": 0.08258, "skel": 0.06468}, "n_reads": 1020, "n_cand": 29309, "read_suppressed_frac": 0.0961, "dropout_reads_logged": 1020, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 192, "admitted_surface": 20077, "staged_surface": 58459, "admitted_split": 76, "staged_split": 47796, "admitted_merge": 6872, "staged_merge": 109723, "admitted_skel": 38511, "staged_skel": 46166, "mean_count": 3.94384765625, "n_util_obs": 27, "util_count_spearman": 0.05067928619570406}, "probe_overhead_sec_total": 213.069, "probe_overhead_seq_frac": 0.4997, "util_loss": 0.005, "alpha": 0.03, "t": 2227.2}
{"event": "consolidate", "step": 2250, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262005, "t": 2274.5}
{"event": "train", "step": 2250, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.49929, "bpb": 2.16303, "chunk_loss": 0.030393, "F": 0.1983, "G": 0.21119, "chunks": 203.06, "bytes_per_chunk": 5.043, "boundary_entropy": 0.17032, "router_noise": 0.0, "grad_norm": 0.8301, "lr": 0.0002293, "router_trainable": true, "sec_per_step": 0.957, "probe_sec": 0.0551, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.03669, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": -0.00902, "n_shadow": 3, "mean_delta_shadow": -0.03543, "n_split": 0, "mean_delta_split": NaN, "n_merge": 1, "mean_delta_merge": 0.00183, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": -0.08759, "trunc_abs_frac": 1.94425, "mean_delta_full": 0.18159, "read_hit_rate": {"surface": 0.01067, "split": 0.03443, "merge": 0.00109, "skel": 0.0993}, "mean_gate": {"surface": 0.06193, "split": 0.05797, "merge": 0.07794, "skel": 0.05437}, "n_reads": 971, "n_cand": 29256, "read_suppressed_frac": 0.1133, "dropout_reads_logged": 971, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 203, "admitted_surface": 21094, "staged_surface": 57902, "admitted_split": 77, "staged_split": 47915, "admitted_merge": 6872, "staged_merge": 110589, "admitted_skel": 37493, "staged_skel": 45738, "mean_count": 5.081878662109375, "n_util_obs": 28, "util_count_spearman": -0.0005473453749315818}, "probe_overhead_sec_total": 217.351, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.004525, "alpha": 0.03, "t": 2275.0}
{"event": "consolidate", "step": 2300, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262003, "t": 2307.9}
{"event": "train", "step": 2300, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.52144, "bpb": 2.19497, "chunk_loss": 0.030285, "F": 0.19186, "G": 0.20814, "chunks": 196.47, "bytes_per_chunk": 5.212, "boundary_entropy": 0.18176, "router_noise": 0.0, "grad_norm": 0.8645, "lr": 0.000226, "router_trainable": true, "sec_per_step": 0.665, "probe_sec": 0.0587, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04086, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 11, "mean_delta_loo": 0.00163, "n_shadow": 3, "mean_delta_shadow": -0.02539, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 1, "trunc_ratio_sum": 0.04084, "trunc_abs_frac": 0.04084, "mean_delta_full": 0.28845, "read_hit_rate": {"surface": 0.01311, "split": 0.03824, "merge": 0.00064, "skel": 0.09453}, "mean_gate": {"surface": 0.05697, "split": 0.05433, "merge": 0.06631, "skel": 0.049}, "n_reads": 975, "n_cand": 28408, "read_suppressed_frac": 0.0995, "dropout_reads_logged": 975, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 214, "admitted_surface": 22074, "staged_surface": 57947, "admitted_split": 77, "staged_split": 47730, "admitted_merge": 6872, "staged_merge": 110353, "admitted_skel": 36513, "staged_skel": 46114, "mean_count": 5.835784912109375, "n_util_obs": 47, "util_count_spearman": -0.36934462736951057}, "probe_overhead_sec_total": 220.683, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.004177, "alpha": 0.03, "t": 2308.3}
{"event": "consolidate", "step": 2350, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262003, "t": 2341.0}
{"event": "train", "step": 2350, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.57072, "bpb": 2.26608, "chunk_loss": 0.030427, "F": 0.20178, "G": 0.21465, "chunks": 206.62, "bytes_per_chunk": 4.956, "boundary_entropy": 0.17346, "router_noise": 0.0, "grad_norm": 0.8458, "lr": 0.0002226, "router_trainable": true, "sec_per_step": 0.662, "probe_sec": 0.0586, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04042, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": -0.02296, "n_shadow": 7, "mean_delta_shadow": -0.00885, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": 3.70531, "trunc_abs_frac": 137.30765, "mean_delta_full": -0.02117, "read_hit_rate": {"surface": 0.01641, "split": 0.03562, "merge": 0.00031, "skel": 0.08525}, "mean_gate": {"surface": 0.05786, "split": 0.05396, "merge": 0.08568, "skel": 0.05015}, "n_reads": 955, "n_cand": 29643, "read_suppressed_frac": 0.0859, "dropout_reads_logged": 955, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 242, "admitted_surface": 23097, "staged_surface": 57716, "admitted_split": 77, "staged_split": 48051, "admitted_merge": 6873, "staged_merge": 110328, "admitted_skel": 35489, "staged_skel": 46049, "mean_count": 6.924041748046875, "n_util_obs": 67, "util_count_spearman": -0.35892728869023866}, "probe_overhead_sec_total": 224.372, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.003687, "alpha": 0.03, "t": 2341.4}
{"event": "consolidate", "step": 2400, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262007, "t": 2378.7}
{"event": "probe", "step": 2400, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3442.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.197782, "cps": 26.890625, "mean_p": 0.213109, "boundary_entropy": 0.180572, "bpb": 2.068201}, "tr": {"chunks": 3906.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.215765, "cps": 30.515625, "mean_p": 0.207584, "boundary_entropy": 0.150978, "bpb": 2.255524}, "fi": {"chunks": 4477.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.237999, "cps": 34.976562, "mean_p": 0.240083, "boundary_entropy": 0.155051, "bpb": 2.395863}}, "bpb": {"en": 2.0682, "tr": 2.25552, "fi": 2.39586}, "gini_cpc": 0.04115, "gini_cps": 0.058351, "bpic": 4.797632, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.090925119136372, "fi": 1.2033403100804168}, "inventory": {"size": 326656, "admitted": 65536, "staged": 261120, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 292, "admitted_surface": 24120, "staged_surface": 57163, "admitted_split": 78, "staged_split": 47448, "admitted_merge": 6873, "staged_merge": 110538, "admitted_skel": 34465, "staged_skel": 45971, "mean_count": 10.883514404296875, "n_util_obs": 102, "util_count_spearman": -0.11266546414778543}, "t": 2380.1}
{"event": "eval", "step": 2400, "phase": 1, "bpb": {"en": 2.07904, "tr": 2.26372, "fi": 2.40447}, "bytes_per_chunk": {"en": 5.115, "tr": 5.168, "fi": 4.584}, "bpb_mean": 2.24907, "frozen_inventory": true, "t": 2382.6}
{"event": "phase_transition", "step": 2400, "phase": 2, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0, "t": 2386.4}
{"event": "train", "step": 2400, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.4982, "bpb": 2.16145, "chunk_loss": 0.030518, "F": 0.20578, "G": 0.21748, "chunks": 210.72, "bytes_per_chunk": 4.86, "boundary_entropy": 0.16131, "router_noise": 0.0, "grad_norm": 0.7314, "lr": 0.0002192, "router_trainable": true, "sec_per_step": 0.913, "probe_sec": 0.0901, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.05789, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": -0.00888, "n_shadow": 3, "mean_delta_shadow": 0.01391, "n_split": 1, "mean_delta_split": -0.00635, "n_merge": 1, "mean_delta_merge": -0.10992, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 0.01124, "trunc_abs_frac": 0.33499, "mean_delta_full": -0.17175, "read_hit_rate": {"surface": 0.02801, "split": 0.04244, "merge": 0.00105, "skel": 0.10006}, "mean_gate": {"surface": 0.05673, "split": 0.05344, "merge": 0.08451, "skel": 0.05124}, "n_reads": 1225, "n_cand": 30413, "read_suppressed_frac": 0.0988, "dropout_reads_logged": 1225, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 293, "admitted_surface": 24120, "staged_surface": 57355, "admitted_split": 78, "staged_split": 47606, "admitted_merge": 6873, "staged_merge": 111156, "admitted_skel": 34465, "staged_skel": 46027, "mean_count": 10.902206420898438, "n_util_obs": 102, "util_count_spearman": -0.10949895674890162}, "probe_overhead_sec_total": 228.121, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.003603, "alpha": 0.03, "t": 2387.0}
{"event": "consolidate", "step": 2450, "retired": 1, "split_retired": 0, "merge_admitted": 0, "admitted": 1, "swapped": 440, "uncertain_staged": 262011, "t": 2425.3}
{"event": "train", "step": 2450, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.46797, "bpb": 2.11784, "chunk_loss": 0.030328, "F": 0.19446, "G": 0.20515, "chunks": 199.12, "bytes_per_chunk": 5.142, "boundary_entropy": 0.16224, "router_noise": 0.0, "grad_norm": 0.7629, "lr": 0.0002158, "router_trainable": true, "sec_per_step": 0.774, "probe_sec": 0.0613, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04346, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": -0.03969, "n_shadow": 4, "mean_delta_shadow": -0.09074, "n_split": 3, "mean_delta_split": 0.02917, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 0.52842, "trunc_abs_frac": 0.3625, "mean_delta_full": -0.0741, "read_hit_rate": {"surface": 0.0571, "split": 0.20453, "merge": 0.0019, "skel": 0.14342}, "mean_gate": {"surface": 0.05963, "split": 0.05449, "merge": 0.06671, "skel": 0.05434}, "n_reads": 3395, "n_cand": 28948, "read_suppressed_frac": 0.1007, "dropout_reads_logged": 3395, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 327, "admitted_surface": 24525, "staged_surface": 57493, "admitted_split": 109, "staged_split": 47573, "admitted_merge": 6873, "staged_merge": 110762, "admitted_skel": 34029, "staged_skel": 46316, "mean_count": 80.20669555664062, "n_util_obs": 150, "util_count_spearman": -0.06503939289694814}, "probe_overhead_sec_total": 231.788, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.003522, "alpha": 0.03, "t": 2425.8}
{"event": "consolidate", "step": 2500, "retired": 1, "split_retired": 0, "merge_admitted": 0, "admitted": 1, "swapped": 1024, "uncertain_staged": 262041, "t": 2459.2}
{"event": "train", "step": 2500, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.44599, "bpb": 2.08612, "chunk_loss": 0.030357, "F": 0.19659, "G": 0.20982, "chunks": 201.31, "bytes_per_chunk": 5.087, "boundary_entropy": 0.17023, "router_noise": 0.0, "grad_norm": 0.7841, "lr": 0.0002123, "router_trainable": true, "sec_per_step": 0.677, "probe_sec": 0.0651, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04321, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": 0.00802, "n_shadow": 3, "mean_delta_shadow": 0.0045, "n_split": 0, "mean_delta_split": NaN, "n_merge": 2, "mean_delta_merge": 0.04889, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": -57.45454, "trunc_abs_frac": 2.65384, "mean_delta_full": -0.0009, "read_hit_rate": {"surface": 0.06505, "split": 0.19794, "merge": 0.0022, "skel": 0.13599}, "mean_gate": {"surface": 0.04455, "split": 0.0401, "merge": 0.0711, "skel": 0.03882}, "n_reads": 3351, "n_cand": 29142, "read_suppressed_frac": 0.1044, "dropout_reads_logged": 3351, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 333, "admitted_surface": 25546, "staged_surface": 57074, "admitted_split": 111, "staged_split": 47948, "admitted_merge": 6873, "staged_merge": 110801, "admitted_skel": 33006, "staged_skel": 46321, "mean_count": 82.98104858398438, "n_util_obs": 157, "util_count_spearman": -0.039559349101922874}, "probe_overhead_sec_total": 235.465, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.003386, "alpha": 0.03, "t": 2459.6}
{"event": "consolidate", "step": 2550, "retired": 0, "split_retired": 3, "merge_admitted": 0, "admitted": 3, "swapped": 1024, "uncertain_staged": 262041, "t": 2491.9}
{"event": "train", "step": 2550, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.47225, "bpb": 2.12401, "chunk_loss": 0.03037, "F": 0.19901, "G": 0.21012, "chunks": 203.78, "bytes_per_chunk": 5.025, "boundary_entropy": 0.16161, "router_noise": 0.0, "grad_norm": 0.7651, "lr": 0.0002088, "router_trainable": true, "sec_per_step": 0.653, "probe_sec": 0.0623, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04327, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 5, "mean_delta_loo": 0.06454, "n_shadow": 7, "mean_delta_shadow": -0.00637, "n_split": 0, "mean_delta_split": NaN, "n_merge": 1, "mean_delta_merge": 0.04089, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": -0.03174, "trunc_abs_frac": 0.67293, "mean_delta_full": 0.16083, "read_hit_rate": {"surface": 0.05193, "split": 0.20333, "merge": 0.00093, "skel": 0.14793}, "mean_gate": {"surface": 0.04678, "split": 0.03859, "merge": 0.06054, "skel": 0.03744}, "n_reads": 3444, "n_cand": 29632, "read_suppressed_frac": 0.0929, "dropout_reads_logged": 3444, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 357, "admitted_surface": 26436, "staged_surface": 56605, "admitted_split": 113, "staged_split": 48103, "admitted_merge": 6873, "staged_merge": 110990, "admitted_skel": 32114, "staged_skel": 46446, "mean_count": 83.72833251953125, "n_util_obs": 159, "util_count_spearman": -0.09381588954500815}, "probe_overhead_sec_total": 239.152, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.003305, "alpha": 0.03, "t": 2492.3}
{"event": "consolidate", "step": 2600, "retired": 1, "split_retired": 0, "merge_admitted": 0, "admitted": 1, "swapped": 1024, "uncertain_staged": 262039, "t": 2538.8}
{"event": "probe", "step": 2600, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3437.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.197495, "cps": 26.851562, "mean_p": 0.210331, "boundary_entropy": 0.173036, "bpb": 2.049125}, "tr": {"chunks": 3934.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.217312, "cps": 30.734375, "mean_p": 0.209467, "boundary_entropy": 0.155378, "bpb": 2.249351}, "fi": {"chunks": 4496.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.239009, "cps": 35.125, "mean_p": 0.240809, "boundary_entropy": 0.152062, "bpb": 2.37997}}, "bpb": {"en": 2.04913, "tr": 2.24935, "fi": 2.37997}, "gini_cpc": 0.04233, "gini_cps": 0.059493, "bpic": 4.780652, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.1003437784670274, "fi": 1.2102051791330966}, "inventory": {"size": 326655, "admitted": 65536, "staged": 261119, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 390, "admitted_surface": 26670, "staged_surface": 56505, "admitted_split": 198, "staged_split": 47444, "admitted_merge": 6274, "staged_merge": 111272, "admitted_skel": 32394, "staged_skel": 45898, "mean_count": 86.53739929199219, "n_util_obs": 164, "util_count_spearman": -0.12793137526323056}, "t": 2540.3}
{"event": "eval", "step": 2600, "phase": 2, "bpb": {"en": 2.06155, "tr": 2.25893, "fi": 2.38684}, "bytes_per_chunk": {"en": 5.117, "tr": 5.132, "fi": 4.573}, "bpb_mean": 2.23578, "frozen_inventory": true, "t": 2542.7}
{"event": "train", "step": 2600, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.42603, "bpb": 2.05733, "chunk_loss": 0.030327, "F": 0.19684, "G": 0.20901, "chunks": 201.56, "bytes_per_chunk": 5.08, "boundary_entropy": 0.16373, "router_noise": 0.0, "grad_norm": 0.7606, "lr": 0.0002053, "router_trainable": true, "sec_per_step": 1.112, "probe_sec": 0.1144, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.06394, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": 0.03953, "n_shadow": 2, "mean_delta_shadow": 0.01166, "n_split": 0, "mean_delta_split": NaN, "n_merge": 3, "mean_delta_merge": 0.03408, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 0.72242, "trunc_abs_frac": 0.71059, "mean_delta_full": 0.1571, "read_hit_rate": {"surface": 0.05859, "split": 0.18504, "merge": 0.00047, "skel": 0.18437}, "mean_gate": {"surface": 0.04541, "split": 0.03653, "merge": 0.03986, "skel": 0.03642}, "n_reads": 3361, "n_cand": 28938, "read_suppressed_frac": 0.1056, "dropout_reads_logged": 3361, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 393, "admitted_surface": 26670, "staged_surface": 56718, "admitted_split": 198, "staged_split": 47595, "admitted_merge": 6274, "staged_merge": 111757, "admitted_skel": 32394, "staged_skel": 46074, "mean_count": 86.58868408203125, "n_util_obs": 164, "util_count_spearman": -0.12635478015094417}, "probe_overhead_sec_total": 243.495, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.003308, "alpha": 0.03, "t": 2547.9}
{"event": "consolidate", "step": 2650, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262039, "t": 2579.0}
{"event": "train", "step": 2650, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.47784, "bpb": 2.13207, "chunk_loss": 0.030237, "F": 0.19006, "G": 0.20235, "chunks": 194.62, "bytes_per_chunk": 5.261, "boundary_entropy": 0.15845, "router_noise": 0.0, "grad_norm": 0.7264, "lr": 0.0002017, "router_trainable": true, "sec_per_step": 0.638, "probe_sec": 0.0683, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04056, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": 0.04062, "n_shadow": 3, "mean_delta_shadow": 0.0164, "n_split": 2, "mean_delta_split": -0.01303, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": -4.75339, "trunc_abs_frac": 1.22918, "mean_delta_full": -0.01799, "read_hit_rate": {"surface": 0.05892, "split": 0.18512, "merge": 0.00065, "skel": 0.16823}, "mean_gate": {"surface": 0.04623, "split": 0.03834, "merge": 0.0683, "skel": 0.04121}, "n_reads": 3236, "n_cand": 28301, "read_suppressed_frac": 0.1038, "dropout_reads_logged": 3236, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 426, "admitted_surface": 27367, "staged_surface": 56817, "admitted_split": 215, "staged_split": 47642, "admitted_merge": 5296, "staged_merge": 111803, "admitted_skel": 32658, "staged_skel": 45882, "mean_count": 89.31025695800781, "n_util_obs": 168, "util_count_spearman": -0.12955880054694596}, "probe_overhead_sec_total": 247.09, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.003231, "alpha": 0.03, "t": 2579.8}
{"event": "consolidate", "step": 2700, "retired": 1, "split_retired": 0, "merge_admitted": 0, "admitted": 1, "swapped": 1024, "uncertain_staged": 262039, "t": 2611.5}
{"event": "train", "step": 2700, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.45688, "bpb": 2.10184, "chunk_loss": 0.030335, "F": 0.19757, "G": 0.20823, "chunks": 202.31, "bytes_per_chunk": 5.061, "boundary_entropy": 0.15712, "router_noise": 0.0, "grad_norm": 0.7385, "lr": 0.0001981, "router_trainable": true, "sec_per_step": 0.644, "probe_sec": 0.0675, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04414, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 4, "mean_delta_loo": -0.0497, "n_shadow": 6, "mean_delta_shadow": 0.06122, "n_split": 3, "mean_delta_split": -0.0214, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": -0.68383, "trunc_abs_frac": 0.53636, "mean_delta_full": 0.03398, "read_hit_rate": {"surface": 0.05697, "split": 0.19444, "merge": 0.00265, "skel": 0.17051}, "mean_gate": {"surface": 0.04402, "split": 0.03493, "merge": 0.05743, "skel": 0.03852}, "n_reads": 3454, "n_cand": 29270, "read_suppressed_frac": 0.0973, "dropout_reads_logged": 3454, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 424, "admitted_surface": 27663, "staged_surface": 56670, "admitted_split": 214, "staged_split": 47879, "admitted_merge": 4925, "staged_merge": 111728, "admitted_skel": 32734, "staged_skel": 45867, "mean_count": 91.83915710449219, "n_util_obs": 173, "util_count_spearman": -0.16830576522718918}, "probe_overhead_sec_total": 250.685, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.003132, "alpha": 0.03, "t": 2612.0}
{"event": "consolidate", "step": 2750, "retired": 0, "split_retired": 1, "merge_admitted": 0, "admitted": 1, "swapped": 1024, "uncertain_staged": 262038, "t": 2650.7}
{"event": "train", "step": 2750, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.46506, "bpb": 2.11364, "chunk_loss": 0.030383, "F": 0.19827, "G": 0.20896, "chunks": 203.03, "bytes_per_chunk": 5.044, "boundary_entropy": 0.15934, "router_noise": 0.0, "grad_norm": 0.7835, "lr": 0.0001945, "router_trainable": true, "sec_per_step": 0.784, "probe_sec": 0.1003, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.06256, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 5, "mean_delta_loo": -0.01433, "n_shadow": 7, "mean_delta_shadow": 0.00192, "n_split": 0, "mean_delta_split": NaN, "n_merge": 1, "mean_delta_merge": 0.0087, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": 0.20494, "trunc_abs_frac": 0.61376, "mean_delta_full": -0.18164, "read_hit_rate": {"surface": 0.06574, "split": 0.19688, "merge": 0.00233, "skel": 0.17553}, "mean_gate": {"surface": 0.04589, "split": 0.03568, "merge": 0.06639, "skel": 0.03992}, "n_reads": 3537, "n_cand": 29177, "read_suppressed_frac": 0.0925, "dropout_reads_logged": 3537, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 401, "admitted_surface": 26807, "staged_surface": 57394, "admitted_split": 212, "staged_split": 48039, "admitted_merge": 4924, "staged_merge": 111760, "admitted_skel": 33593, "staged_skel": 44951, "mean_count": 94.15292358398438, "n_util_obs": 176, "util_count_spearman": -0.19935317965002897}, "probe_overhead_sec_total": 254.885, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.003003, "alpha": 0.03, "t": 2651.2}
{"event": "consolidate", "step": 2800, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262035, "t": 2690.0}
{"event": "probe", "step": 2800, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3432.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.197207, "cps": 26.8125, "mean_p": 0.209127, "boundary_entropy": 0.165628, "bpb": 2.033743}, "tr": {"chunks": 3846.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.212451, "cps": 30.046875, "mean_p": 0.203714, "boundary_entropy": 0.148021, "bpb": 2.199801}, "fi": {"chunks": 4510.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.239753, "cps": 35.234375, "mean_p": 0.23893, "boundary_entropy": 0.13791, "bpb": 2.331737}}, "bpb": {"en": 2.03374, "tr": 2.1998, "fi": 2.33174}, "gini_cpc": 0.043676, "gini_cps": 0.060966, "bpic": 4.812691, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0772972953136462, "fi": 1.2157422212044506}, "inventory": {"size": 326656, "admitted": 65536, "staged": 261120, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 419, "admitted_surface": 26814, "staged_surface": 57705, "admitted_split": 212, "staged_split": 47759, "admitted_merge": 3901, "staged_merge": 111133, "admitted_skel": 34609, "staged_skel": 44523, "mean_count": 96.72433471679688, "n_util_obs": 178, "util_count_spearman": -0.17783151706139505}, "t": 2691.4}
{"event": "eval", "step": 2800, "phase": 2, "bpb": {"en": 2.04497, "tr": 2.20819, "fi": 2.33842}, "bytes_per_chunk": {"en": 5.157, "tr": 5.273, "fi": 4.571}, "bpb_mean": 2.19719, "frozen_inventory": true, "t": 2694.0}
{"event": "train", "step": 2800, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.41867, "bpb": 2.04671, "chunk_loss": 0.030346, "F": 0.19751, "G": 0.20768, "chunks": 202.25, "bytes_per_chunk": 5.063, "boundary_entropy": 0.15718, "router_noise": 0.0, "grad_norm": 0.7468, "lr": 0.0001909, "router_trainable": true, "sec_per_step": 0.944, "probe_sec": 0.072, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04291, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 4, "mean_delta_loo": 0.05688, "n_shadow": 5, "mean_delta_shadow": 0.00927, "n_split": 1, "mean_delta_split": -0.05017, "n_merge": 2, "mean_delta_merge": 0.01665, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 0.0147, "trunc_abs_frac": 0.40753, "mean_delta_full": 0.16614, "read_hit_rate": {"surface": 0.06429, "split": 0.18927, "merge": 0.0014, "skel": 0.17548}, "mean_gate": {"surface": 0.04608, "split": 0.03761, "merge": 0.0605, "skel": 0.04278}, "n_reads": 3435, "n_cand": 29134, "read_suppressed_frac": 0.0952, "dropout_reads_logged": 3435, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 417, "admitted_surface": 26814, "staged_surface": 57832, "admitted_split": 212, "staged_split": 47936, "admitted_merge": 3901, "staged_merge": 111667, "admitted_skel": 34609, "staged_skel": 44709, "mean_count": 96.77674865722656, "n_util_obs": 178, "util_count_spearman": -0.1764389525373722}, "probe_overhead_sec_total": 258.896, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.003012, "alpha": 0.03, "t": 2698.4}
{"event": "consolidate", "step": 2850, "retired": 2, "split_retired": 0, "merge_admitted": 0, "admitted": 2, "swapped": 1024, "uncertain_staged": 262036, "t": 2728.9}
{"event": "train", "step": 2850, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.46528, "bpb": 2.11395, "chunk_loss": 0.030517, "F": 0.20615, "G": 0.21677, "chunks": 211.09, "bytes_per_chunk": 4.851, "boundary_entropy": 0.16002, "router_noise": 0.0, "grad_norm": 0.7709, "lr": 0.0001872, "router_trainable": true, "sec_per_step": 0.629, "probe_sec": 0.0713, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04686, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": 0.00201, "n_shadow": 4, "mean_delta_shadow": 0.00323, "n_split": 1, "mean_delta_split": 0.06754, "n_merge": 1, "mean_delta_merge": 0.17493, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 1.54669, "trunc_abs_frac": 0.67367, "mean_delta_full": 0.02811, "read_hit_rate": {"surface": 0.05489, "split": 0.18548, "merge": 0.0009, "skel": 0.18654}, "mean_gate": {"surface": 0.04494, "split": 0.03282, "merge": 0.05385, "skel": 0.03942}, "n_reads": 3539, "n_cand": 30442, "read_suppressed_frac": 0.0938, "dropout_reads_logged": 3539, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 431, "admitted_surface": 26784, "staged_surface": 58070, "admitted_split": 210, "staged_split": 48004, "admitted_merge": 2922, "staged_merge": 111354, "admitted_skel": 35620, "staged_skel": 44716, "mean_count": 99.0538330078125, "n_util_obs": 177, "util_count_spearman": -0.22418240681428644}, "probe_overhead_sec_total": 262.387, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.003007, "alpha": 0.03, "t": 2729.8}
{"event": "consolidate", "step": 2900, "retired": 0, "split_retired": 1, "merge_admitted": 0, "admitted": 1, "swapped": 1024, "uncertain_staged": 262033, "t": 2761.6}
{"event": "train", "step": 2900, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.39564, "bpb": 2.01348, "chunk_loss": 0.030246, "F": 0.1915, "G": 0.20179, "chunks": 196.09, "bytes_per_chunk": 5.222, "boundary_entropy": 0.15351, "router_noise": 0.0, "grad_norm": 0.689, "lr": 0.0001835, "router_trainable": true, "sec_per_step": 0.644, "probe_sec": 0.0612, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04484, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": 0.00326, "n_shadow": 7, "mean_delta_shadow": -0.05564, "n_split": 1, "mean_delta_split": 0.01691, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": -3.45473, "trunc_abs_frac": 0.61742, "mean_delta_full": 0.00742, "read_hit_rate": {"surface": 0.03396, "split": 0.05231, "merge": 0.00016, "skel": 0.12069}, "mean_gate": {"surface": 0.05211, "split": 0.04705, "merge": 0.04098, "skel": 0.04813}, "n_reads": 1379, "n_cand": 28267, "read_suppressed_frac": 0.0899, "dropout_reads_logged": 1379, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 357, "admitted_surface": 25904, "staged_surface": 58058, "admitted_split": 157, "staged_split": 48003, "admitted_merge": 2921, "staged_merge": 111482, "admitted_skel": 36554, "staged_skel": 44601, "mean_count": 29.169692993164062, "n_util_obs": 85, "util_count_spearman": -0.6453617615953592}, "probe_overhead_sec_total": 266.321, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.003715, "alpha": 0.03, "t": 2762.0}
{"event": "consolidate", "step": 2950, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262029, "t": 2793.6}
{"event": "train", "step": 2950, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.54879, "bpb": 2.23443, "chunk_loss": 0.030462, "F": 0.20227, "G": 0.21139, "chunks": 207.12, "bytes_per_chunk": 4.944, "boundary_entropy": 0.147, "router_noise": 0.0, "grad_norm": 0.7828, "lr": 0.0001799, "router_trainable": true, "sec_per_step": 0.641, "probe_sec": 0.0609, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04241, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": -0.02319, "n_shadow": 4, "mean_delta_shadow": -0.01041, "n_split": 1, "mean_delta_split": 0.09709, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": -1.98637, "trunc_abs_frac": 0.44581, "mean_delta_full": 0.0459, "read_hit_rate": {"surface": 0.01698, "split": 0.00491, "merge": 0.0, "skel": 0.02998}, "mean_gate": {"surface": 0.04619, "split": 0.0473, "merge": null, "skel": 0.05673}, "n_reads": 320, "n_cand": 29628, "read_suppressed_frac": 0.0875, "dropout_reads_logged": 320, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 356, "admitted_surface": 25575, "staged_surface": 58167, "admitted_split": 91, "staged_split": 48197, "admitted_merge": 2552, "staged_merge": 110830, "admitted_skel": 37318, "staged_skel": 44950, "mean_count": 5.0789337158203125, "n_util_obs": 29, "util_count_spearman": -0.5569651477152491}, "probe_overhead_sec_total": 269.825, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.004591, "alpha": 0.03, "t": 2794.1}
{"event": "consolidate", "step": 3000, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262028, "t": 2830.0}
{"event": "probe", "step": 3000, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3440.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.197667, "cps": 26.875, "mean_p": 0.20715, "boundary_entropy": 0.157579, "bpb": 2.011101}, "tr": {"chunks": 3866.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.213556, "cps": 30.203125, "mean_p": 0.206315, "boundary_entropy": 0.145317, "bpb": 2.183759}, "fi": {"chunks": 4485.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.238424, "cps": 35.039062, "mean_p": 0.238359, "boundary_entropy": 0.132759, "bpb": 2.312584}}, "bpb": {"en": 2.0111, "tr": 2.18376, "fi": 2.31258}, "gini_cpc": 0.041825, "gini_cps": 0.059085, "bpic": 4.811466, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0803810944864591, "fi": 1.2061914385818293}, "inventory": {"size": 326656, "admitted": 65536, "staged": 261120, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 268, "admitted_surface": 25035, "staged_surface": 57938, "admitted_split": 62, "staged_split": 47976, "admitted_merge": 2552, "staged_merge": 110305, "admitted_skel": 37887, "staged_skel": 44901, "mean_count": 3.4413299560546875, "n_util_obs": 20, "util_count_spearman": -0.3894736842105263}, "t": 2831.7}
{"event": "eval", "step": 3000, "phase": 2, "bpb": {"en": 2.0247, "tr": 2.19202, "fi": 2.31871}, "bytes_per_chunk": {"en": 5.156, "tr": 5.254, "fi": 4.593}, "bpb_mean": 2.17848, "frozen_inventory": true, "t": 2834.4}
{"event": "train", "step": 3000, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.49484, "bpb": 2.15659, "chunk_loss": 0.030371, "F": 0.1951, "G": 0.2041, "chunks": 199.78, "bytes_per_chunk": 5.126, "boundary_entropy": 0.14521, "router_noise": 0.0, "grad_norm": 0.7383, "lr": 0.0001761, "router_trainable": true, "sec_per_step": 0.931, "probe_sec": 0.0872, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04935, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": 0.00235, "n_shadow": 7, "mean_delta_shadow": -0.03853, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 0.17243, "trunc_abs_frac": 0.26853, "mean_delta_full": -0.29486, "read_hit_rate": {"surface": 0.00786, "split": 0.00143, "merge": 0.0, "skel": 0.02133}, "mean_gate": {"surface": 0.04325, "split": 0.04835, "merge": null, "skel": 0.06749}, "n_reads": 175, "n_cand": 28969, "read_suppressed_frac": 0.0971, "dropout_reads_logged": 175, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 265, "admitted_surface": 25035, "staged_surface": 58196, "admitted_split": 62, "staged_split": 48124, "admitted_merge": 2552, "staged_merge": 110659, "admitted_skel": 37887, "staged_skel": 45165, "mean_count": 3.444000244140625, "n_util_obs": 20, "util_count_spearman": -0.38345864661654133}, "probe_overhead_sec_total": 273.269, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.004532, "alpha": 0.03, "t": 2840.6}
{"event": "consolidate", "step": 3050, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262024, "t": 2873.3}
{"event": "train", "step": 3050, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.44606, "bpb": 2.08622, "chunk_loss": 0.030448, "F": 0.19995, "G": 0.20788, "chunks": 204.75, "bytes_per_chunk": 5.001, "boundary_entropy": 0.15295, "router_noise": 0.0, "grad_norm": 0.7275, "lr": 0.0001724, "router_trainable": true, "sec_per_step": 0.661, "probe_sec": 0.0599, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04337, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 5, "mean_delta_loo": 0.09049, "n_shadow": 6, "mean_delta_shadow": -0.05786, "n_split": 1, "mean_delta_split": 0.04715, "n_merge": 1, "mean_delta_merge": -0.00333, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": 1.15271, "trunc_abs_frac": 0.87849, "mean_delta_full": 0.04272, "read_hit_rate": {"surface": 0.00475, "split": 0.00106, "merge": 0.00015, "skel": 0.01287}, "mean_gate": {"surface": 0.05127, "split": 0.05713, "merge": 0.09541, "skel": 0.06317}, "n_reads": 109, "n_cand": 29381, "read_suppressed_frac": 0.1468, "dropout_reads_logged": 109, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 214, "admitted_surface": 24142, "staged_surface": 58784, "admitted_split": 20, "staged_split": 48194, "admitted_merge": 2552, "staged_merge": 109878, "admitted_skel": 38822, "staged_skel": 45288, "mean_count": 2.8773345947265625, "n_util_obs": 21, "util_count_spearman": -0.5454545454545454}, "probe_overhead_sec_total": 276.449, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.004319, "alpha": 0.03, "t": 2873.7}
{"event": "consolidate", "step": 3100, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262026, "t": 2904.6}
{"event": "train", "step": 3100, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.49219, "bpb": 2.15278, "chunk_loss": 0.030346, "F": 0.19678, "G": 0.20578, "chunks": 201.5, "bytes_per_chunk": 5.082, "boundary_entropy": 0.1428, "router_noise": 0.0, "grad_norm": 0.7483, "lr": 0.0001687, "router_trainable": true, "sec_per_step": 0.628, "probe_sec": 0.0926, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.07308, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": 0.07838, "n_shadow": 4, "mean_delta_shadow": -0.02863, "n_split": 3, "mean_delta_split": -0.01463, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 0.042, "trunc_abs_frac": 0.4329, "mean_delta_full": 0.07156, "read_hit_rate": {"surface": 0.00795, "split": 0.00633, "merge": 0.00533, "skel": 0.024}, "mean_gate": {"surface": 0.07148, "split": 0.063, "merge": 0.09951, "skel": 0.08926}, "n_reads": 278, "n_cand": 29096, "read_suppressed_frac": 0.1151, "dropout_reads_logged": 278, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 231, "admitted_surface": 24515, "staged_surface": 58642, "admitted_split": 22, "staged_split": 47855, "admitted_merge": 1628, "staged_merge": 110150, "admitted_skel": 39371, "staged_skel": 45497, "mean_count": 4.109466552734375, "n_util_obs": 27, "util_count_spearman": -0.07264957264957266}, "probe_overhead_sec_total": 279.978, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.004315, "alpha": 0.03, "t": 2905.1}
{"event": "consolidate", "step": 3150, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 58, "uncertain_staged": 262025, "t": 2935.8}
{"event": "train", "step": 3150, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.45477, "bpb": 2.09879, "chunk_loss": 0.030321, "F": 0.19391, "G": 0.20415, "chunks": 198.56, "bytes_per_chunk": 5.157, "boundary_entropy": 0.15518, "router_noise": 0.0, "grad_norm": 0.7186, "lr": 0.000165, "router_trainable": true, "sec_per_step": 0.63, "probe_sec": 0.0637, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04121, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": -0.00458, "n_shadow": 5, "mean_delta_shadow": 0.01768, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 6, "trunc_ratio_sum": -0.582, "trunc_abs_frac": 0.35965, "mean_delta_full": 0.07168, "read_hit_rate": {"surface": 0.02562, "split": 0.07615, "merge": 0.00429, "skel": 0.06427}, "mean_gate": {"surface": 0.05293, "split": 0.0471, "merge": 0.08059, "skel": 0.05869}, "n_reads": 1357, "n_cand": 28737, "read_suppressed_frac": 0.0973, "dropout_reads_logged": 1357, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 253, "admitted_surface": 24529, "staged_surface": 58990, "admitted_split": 35, "staged_split": 47424, "admitted_merge": 1571, "staged_merge": 110124, "admitted_skel": 39401, "staged_skel": 45606, "mean_count": 18.987747192382812, "n_util_obs": 51, "util_count_spearman": -0.2864253393665158}, "probe_overhead_sec_total": 283.392, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.003925, "alpha": 0.03, "t": 2936.6}
{"event": "consolidate", "step": 3200, "retired": 1, "split_retired": 1, "merge_admitted": 0, "admitted": 2, "swapped": 96, "uncertain_staged": 262038, "t": 2966.0}
{"event": "probe", "step": 3200, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3459.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.198759, "cps": 27.023438, "mean_p": 0.207193, "boundary_entropy": 0.154594, "bpb": 1.994192}, "tr": {"chunks": 3837.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.211954, "cps": 29.976562, "mean_p": 0.205244, "boundary_entropy": 0.142407, "bpb": 2.15369}, "fi": {"chunks": 4476.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.237946, "cps": 34.96875, "mean_p": 0.237669, "boundary_entropy": 0.135896, "bpb": 2.295326}}, "bpb": {"en": 1.99419, "tr": 2.15369, "fi": 2.29533}, "gini_cpc": 0.040275, "gini_cps": 0.057594, "bpic": 4.819232, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0663869112847038, "fi": 1.1971587733792157}, "inventory": {"size": 327582, "admitted": 65536, "staged": 262046, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 270, "admitted_surface": 24496, "staged_surface": 59119, "admitted_split": 36, "staged_split": 47728, "admitted_merge": 1600, "staged_merge": 109713, "admitted_skel": 39404, "staged_skel": 45486, "mean_count": 20.446823120117188, "n_util_obs": 59, "util_count_spearman": -0.34430157802454703}, "t": 2967.4}
{"event": "eval", "step": 3200, "phase": 2, "bpb": {"en": 2.00588, "tr": 2.16274, "fi": 2.30205}, "bytes_per_chunk": {"en": 5.098, "tr": 5.305, "fi": 4.599}, "bpb_mean": 2.15689, "frozen_inventory": true, "t": 2969.6}
{"event": "train", "step": 3200, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.43983, "bpb": 2.07724, "chunk_loss": 0.030318, "F": 0.19797, "G": 0.20541, "chunks": 202.72, "bytes_per_chunk": 5.051, "boundary_entropy": 0.14488, "router_noise": 0.0, "grad_norm": 0.6341, "lr": 0.0001613, "router_trainable": true, "sec_per_step": 0.736, "probe_sec": 0.0672, "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": 8, "mean_delta_loo": 0.04131, "n_shadow": 4, "mean_delta_shadow": -0.14539, "n_split": 2, "mean_delta_split": 0.00034, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 1, "trunc_ratio_sum": -3.09189, "trunc_abs_frac": 3.09189, "mean_delta_full": -0.02258, "read_hit_rate": {"surface": 0.0268, "split": 0.09014, "merge": 0.00716, "skel": 0.06928}, "mean_gate": {"surface": 0.0445, "split": 0.03897, "merge": 0.07746, "skel": 0.05099}, "n_reads": 1574, "n_cand": 29073, "read_suppressed_frac": 0.0934, "dropout_reads_logged": 1574, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 269, "admitted_surface": 24496, "staged_surface": 59102, "admitted_split": 36, "staged_split": 47774, "admitted_merge": 1600, "staged_merge": 109714, "admitted_skel": 39404, "staged_skel": 45554, "mean_count": 20.470840454101562, "n_util_obs": 59, "util_count_spearman": -0.3448910709174549}, "probe_overhead_sec_total": 286.315, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.00379, "alpha": 0.03, "t": 2973.4}
{"event": "consolidate", "step": 3250, "retired": 0, "split_retired": 2, "merge_admitted": 0, "admitted": 2, "swapped": 129, "uncertain_staged": 262037, "t": 3003.9}
{"event": "train", "step": 3250, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.44224, "bpb": 2.08071, "chunk_loss": 0.030401, "F": 0.1991, "G": 0.20653, "chunks": 203.88, "bytes_per_chunk": 5.023, "boundary_entropy": 0.14194, "router_noise": 0.0, "grad_norm": 0.7075, "lr": 0.0001576, "router_trainable": true, "sec_per_step": 0.626, "probe_sec": 0.0562, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.03992, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": -0.02112, "n_shadow": 7, "mean_delta_shadow": 0.01855, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": -0.19227, "trunc_abs_frac": 1.79159, "mean_delta_full": -0.06, "read_hit_rate": {"surface": 0.0288, "split": 0.08941, "merge": 0.01192, "skel": 0.09253}, "mean_gate": {"surface": 0.05096, "split": 0.03697, "merge": 0.06797, "skel": 0.04791}, "n_reads": 1739, "n_cand": 29271, "read_suppressed_frac": 0.1185, "dropout_reads_logged": 1739, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 279, "admitted_surface": 24597, "staged_surface": 58855, "admitted_split": 36, "staged_split": 47793, "admitted_merge": 1624, "staged_merge": 109872, "admitted_skel": 39279, "staged_skel": 45624, "mean_count": 23.817764282226562, "n_util_obs": 66, "util_count_spearman": -0.3519084846082989}, "probe_overhead_sec_total": 289.381, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.003622, "alpha": 0.03, "t": 3004.7}
{"event": "consolidate", "step": 3300, "retired": 0, "split_retired": 1, "merge_admitted": 0, "admitted": 1, "swapped": 112, "uncertain_staged": 262036, "t": 3034.7}
{"event": "train", "step": 3300, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.41765, "bpb": 2.04523, "chunk_loss": 0.030445, "F": 0.20065, "G": 0.2091, "chunks": 205.47, "bytes_per_chunk": 4.984, "boundary_entropy": 0.15013, "router_noise": 0.0, "grad_norm": 0.6913, "lr": 0.0001539, "router_trainable": true, "sec_per_step": 0.616, "probe_sec": 0.0579, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04016, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 5, "mean_delta_loo": 0.03093, "n_shadow": 6, "mean_delta_shadow": -0.00717, "n_split": 0, "mean_delta_split": NaN, "n_merge": 2, "mean_delta_merge": 0.10455, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": -0.44376, "trunc_abs_frac": 0.66078, "mean_delta_full": -0.0657, "read_hit_rate": {"surface": 0.04478, "split": 0.08996, "merge": 0.00968, "skel": 0.08892}, "mean_gate": {"surface": 0.03917, "split": 0.03158, "merge": 0.05523, "skel": 0.04007}, "n_reads": 1848, "n_cand": 29699, "read_suppressed_frac": 0.099, "dropout_reads_logged": 1848, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 270, "admitted_surface": 24696, "staged_surface": 58852, "admitted_split": 37, "staged_split": 47756, "admitted_merge": 1635, "staged_merge": 109493, "admitted_skel": 39168, "staged_skel": 46043, "mean_count": 29.24053955078125, "n_util_obs": 73, "util_count_spearman": -0.3467287845776102}, "probe_overhead_sec_total": 292.825, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.00355, "alpha": 0.03, "t": 3035.5}
{"event": "consolidate", "step": 3350, "retired": 1, "split_retired": 2, "merge_admitted": 0, "admitted": 3, "swapped": 949, "uncertain_staged": 262037, "t": 3065.6}
{"event": "train", "step": 3350, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.36067, "bpb": 1.96303, "chunk_loss": 0.030286, "F": 0.19373, "G": 0.20407, "chunks": 198.38, "bytes_per_chunk": 5.162, "boundary_entropy": 0.14445, "router_noise": 0.0, "grad_norm": 0.7518, "lr": 0.0001501, "router_trainable": true, "sec_per_step": 0.618, "probe_sec": 0.0538, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.03704, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": -0.00623, "n_shadow": 4, "mean_delta_shadow": -0.0004, "n_split": 1, "mean_delta_split": 0.00769, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 0.22577, "trunc_abs_frac": 1.26272, "mean_delta_full": 0.13477, "read_hit_rate": {"surface": 0.04465, "split": 0.14207, "merge": 0.01989, "skel": 0.08607}, "mean_gate": {"surface": 0.03941, "split": 0.03242, "merge": 0.04818, "skel": 0.03997}, "n_reads": 2409, "n_cand": 28675, "read_suppressed_frac": 0.0917, "dropout_reads_logged": 2409, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 293, "admitted_surface": 24732, "staged_surface": 59086, "admitted_split": 725, "staged_split": 46996, "admitted_merge": 1860, "staged_merge": 110039, "admitted_skel": 38219, "staged_skel": 46023, "mean_count": 41.899749755859375, "n_util_obs": 75, "util_count_spearman": -0.33803698435277385}, "probe_overhead_sec_total": 296.176, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.003463, "alpha": 0.03, "t": 3066.4}
{"event": "consolidate", "step": 3400, "retired": 0, "split_retired": 3, "merge_admitted": 0, "admitted": 3, "swapped": 1024, "uncertain_staged": 262039, "t": 3098.1}
{"event": "probe", "step": 3400, "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": 0.205941, "boundary_entropy": 0.150125, "bpb": 1.982703}, "tr": {"chunks": 3854.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.212893, "cps": 30.109375, "mean_p": 0.203162, "boundary_entropy": 0.131794, "bpb": 2.151264}, "fi": {"chunks": 4518.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.240179, "cps": 35.296875, "mean_p": 0.237634, "boundary_entropy": 0.123821, "bpb": 2.259629}}, "bpb": {"en": 1.9827, "tr": 2.15126, "fi": 2.25963}, "gini_cpc": 0.045439, "gini_cps": 0.062695, "bpic": 4.815142, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0865029256666359, "fi": 1.2257561597062678}, "inventory": {"size": 326653, "admitted": 65536, "staged": 261117, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 332, "admitted_surface": 24775, "staged_surface": 58951, "admitted_split": 1679, "staged_split": 46263, "admitted_merge": 1887, "staged_merge": 109746, "admitted_skel": 37195, "staged_skel": 46157, "mean_count": 42.65748596191406, "n_util_obs": 73, "util_count_spearman": -0.3396890040725657}, "t": 3099.4}
{"event": "eval", "step": 3400, "phase": 2, "bpb": {"en": 1.99568, "tr": 2.15894, "fi": 2.26637}, "bytes_per_chunk": {"en": 5.177, "tr": 5.283, "fi": 4.559}, "bpb_mean": 2.14033, "frozen_inventory": true, "t": 3101.5}
{"event": "train", "step": 3400, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.42247, "bpb": 2.05219, "chunk_loss": 0.03033, "F": 0.19708, "G": 0.20481, "chunks": 201.81, "bytes_per_chunk": 5.074, "boundary_entropy": 0.13856, "router_noise": 0.0, "grad_norm": 0.7827, "lr": 0.0001465, "router_trainable": true, "sec_per_step": 0.807, "probe_sec": 0.0604, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04249, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": -0.05698, "n_shadow": 6, "mean_delta_shadow": 0.01431, "n_split": 0, "mean_delta_split": NaN, "n_merge": 1, "mean_delta_merge": -0.02457, "n_dropped_unpaired": 0, "n_full": 6, "trunc_ratio_sum": 0.25699, "trunc_abs_frac": 1.45404, "mean_delta_full": -0.08435, "read_hit_rate": {"surface": 0.03626, "split": 0.14074, "merge": 0.02096, "skel": 0.07811}, "mean_gate": {"surface": 0.03693, "split": 0.03107, "merge": 0.05293, "skel": 0.03816}, "n_reads": 2346, "n_cand": 29154, "read_suppressed_frac": 0.0985, "dropout_reads_logged": 2346, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 329, "admitted_surface": 24775, "staged_surface": 59172, "admitted_split": 1679, "staged_split": 46434, "admitted_merge": 1887, "staged_merge": 110235, "admitted_skel": 37195, "staged_skel": 46303, "mean_count": 42.69328308105469, "n_util_obs": 74, "util_count_spearman": -0.34140200965030465}, "probe_overhead_sec_total": 299.384, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.003444, "alpha": 0.03, "t": 3106.7}
{"event": "consolidate", "step": 3450, "retired": 0, "split_retired": 1, "merge_admitted": 0, "admitted": 1, "swapped": 1024, "uncertain_staged": 262036, "t": 3137.3}
{"event": "train", "step": 3450, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.39322, "bpb": 2.00999, "chunk_loss": 0.030316, "F": 0.19678, "G": 0.20434, "chunks": 201.5, "bytes_per_chunk": 5.082, "boundary_entropy": 0.13856, "router_noise": 0.0, "grad_norm": 0.7855, "lr": 0.0001428, "router_trainable": true, "sec_per_step": 0.635, "probe_sec": 0.0711, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.0412, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": -0.0024, "n_shadow": 5, "mean_delta_shadow": 0.02496, "n_split": 0, "mean_delta_split": NaN, "n_merge": 1, "mean_delta_merge": 0.0293, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": -574.0, "trunc_abs_frac": 1.07969, "mean_delta_full": -4e-05, "read_hit_rate": {"surface": 0.03959, "split": 0.15303, "merge": 0.02209, "skel": 0.10677}, "mean_gate": {"surface": 0.03989, "split": 0.03266, "merge": 0.0574, "skel": 0.03987}, "n_reads": 2637, "n_cand": 28999, "read_suppressed_frac": 0.102, "dropout_reads_logged": 2637, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 374, "admitted_surface": 24888, "staged_surface": 59196, "admitted_split": 2575, "staged_split": 46175, "admitted_merge": 1901, "staged_merge": 110398, "admitted_skel": 36172, "staged_skel": 46375, "mean_count": 45.31111145019531, "n_util_obs": 75, "util_count_spearman": -0.34082503556187765}, "probe_overhead_sec_total": 302.413, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.003486, "alpha": 0.03, "t": 3138.4}
{"event": "consolidate", "step": 3500, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262036, "t": 3168.3}
{"event": "train", "step": 3500, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.3721, "bpb": 1.97952, "chunk_loss": 0.030247, "F": 0.19257, "G": 0.20122, "chunks": 197.19, "bytes_per_chunk": 5.193, "boundary_entropy": 0.13929, "router_noise": 0.0, "grad_norm": 0.7052, "lr": 0.0001391, "router_trainable": true, "sec_per_step": 0.618, "probe_sec": 0.0614, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.0395, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": 0.02898, "n_shadow": 4, "mean_delta_shadow": -0.01295, "n_split": 0, "mean_delta_split": NaN, "n_merge": 1, "mean_delta_merge": -0.04832, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 0.39723, "trunc_abs_frac": 0.53657, "mean_delta_full": -0.03524, "read_hit_rate": {"surface": 0.09796, "split": 0.18164, "merge": 0.0237, "skel": 0.11349}, "mean_gate": {"surface": 0.04449, "split": 0.03305, "merge": 0.05143, "skel": 0.04039}, "n_reads": 3337, "n_cand": 28586, "read_suppressed_frac": 0.1028, "dropout_reads_logged": 3337, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 386, "admitted_surface": 25386, "staged_surface": 58741, "admitted_split": 3071, "staged_split": 45619, "admitted_merge": 1924, "staged_merge": 111544, "admitted_skel": 35155, "staged_skel": 46240, "mean_count": 48.65594482421875, "n_util_obs": 76, "util_count_spearman": -0.3179494190020506}, "probe_overhead_sec_total": 305.524, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.003409, "alpha": 0.03, "t": 3169.3}
{"event": "consolidate", "step": 3550, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 674, "uncertain_staged": 262034, "t": 3198.9}
{"event": "train", "step": 3550, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.4073, "bpb": 2.0303, "chunk_loss": 0.030295, "F": 0.19312, "G": 0.20212, "chunks": 197.75, "bytes_per_chunk": 5.178, "boundary_entropy": 0.13384, "router_noise": 0.0, "grad_norm": 0.6655, "lr": 0.0001355, "router_trainable": true, "sec_per_step": 0.601, "probe_sec": 0.0664, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04179, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": -0.01614, "n_shadow": 4, "mean_delta_shadow": -0.02051, "n_split": 1, "mean_delta_split": -0.02121, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 0.51768, "trunc_abs_frac": 0.36109, "mean_delta_full": -0.09379, "read_hit_rate": {"surface": 0.13056, "split": 0.19783, "merge": 0.01836, "skel": 0.10644}, "mean_gate": {"surface": 0.03938, "split": 0.0305, "merge": 0.04704, "skel": 0.03629}, "n_reads": 3645, "n_cand": 28579, "read_suppressed_frac": 0.0974, "dropout_reads_logged": 3645, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 424, "admitted_surface": 25665, "staged_surface": 58552, "admitted_split": 3460, "staged_split": 45135, "admitted_merge": 1924, "staged_merge": 112156, "admitted_skel": 34487, "staged_skel": 46301, "mean_count": 54.407012939453125, "n_util_obs": 83, "util_count_spearman": -0.29830821544015784}, "probe_overhead_sec_total": 308.659, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.003536, "alpha": 0.03, "t": 3199.4}
{"event": "consolidate", "step": 3600, "retired": 1, "split_retired": 0, "merge_admitted": 0, "admitted": 1, "swapped": 260, "uncertain_staged": 262034, "t": 3230.1}
{"event": "probe", "step": 3600, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3451.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.198299, "cps": 26.960938, "mean_p": 0.206692, "boundary_entropy": 0.145237, "bpb": 1.969425}, "tr": {"chunks": 3846.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.212451, "cps": 30.046875, "mean_p": 0.202873, "boundary_entropy": 0.132756, "bpb": 2.116959}, "fi": {"chunks": 4520.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.240285, "cps": 35.3125, "mean_p": 0.237191, "boundary_entropy": 0.122996, "bpb": 2.240796}}, "bpb": {"en": 1.96943, "tr": 2.11696, "fi": 2.2408}, "gini_cpc": 0.042994, "gini_cps": 0.060309, "bpic": 4.80088, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.071366072882189, "fi": 1.2117295870630391}, "inventory": {"size": 327419, "admitted": 65536, "staged": 261883, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 471, "admitted_surface": 25699, "staged_surface": 58710, "admitted_split": 3680, "staged_split": 44892, "admitted_merge": 1925, "staged_merge": 111824, "admitted_skel": 34232, "staged_skel": 46457, "mean_count": 57.64715576171875, "n_util_obs": 85, "util_count_spearman": -0.27676372874731286}, "t": 3231.4}
{"event": "eval", "step": 3600, "phase": 2, "bpb": {"en": 1.9823, "tr": 2.12345, "fi": 2.24777}, "bytes_per_chunk": {"en": 5.1, "tr": 5.302, "fi": 4.563}, "bpb_mean": 2.11784, "frozen_inventory": true, "t": 3233.6}
{"event": "train", "step": 3600, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.41292, "bpb": 2.03841, "chunk_loss": 0.030354, "F": 0.19766, "G": 0.20372, "chunks": 202.41, "bytes_per_chunk": 5.059, "boundary_entropy": 0.13504, "router_noise": 0.0, "grad_norm": 0.664, "lr": 0.0001319, "router_trainable": true, "sec_per_step": 0.763, "probe_sec": 0.0759, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.056, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": 0.01012, "n_shadow": 4, "mean_delta_shadow": -0.03716, "n_split": 0, "mean_delta_split": NaN, "n_merge": 1, "mean_delta_merge": -0.04486, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 0.51077, "trunc_abs_frac": 0.9233, "mean_delta_full": -0.05477, "read_hit_rate": {"surface": 0.13313, "split": 0.19178, "merge": 0.01918, "skel": 0.12365}, "mean_gate": {"surface": 0.03762, "split": 0.03002, "merge": 0.05566, "skel": 0.03506}, "n_reads": 3801, "n_cand": 29404, "read_suppressed_frac": 0.0905, "dropout_reads_logged": 3801, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 472, "admitted_surface": 25699, "staged_surface": 58807, "admitted_split": 3680, "staged_split": 44918, "admitted_merge": 1925, "staged_merge": 111925, "admitted_skel": 34232, "staged_skel": 46494, "mean_count": 57.70515441894531, "n_util_obs": 85, "util_count_spearman": -0.27811353387831383}, "probe_overhead_sec_total": 311.996, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.003508, "alpha": 0.03, "t": 3237.5}
{"event": "consolidate", "step": 3650, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1021, "uncertain_staged": 262035, "t": 3267.2}
{"event": "train", "step": 3650, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.44113, "bpb": 2.07912, "chunk_loss": 0.030538, "F": 0.20438, "G": 0.21144, "chunks": 209.28, "bytes_per_chunk": 4.893, "boundary_entropy": 0.1335, "router_noise": 0.0, "grad_norm": 0.7091, "lr": 0.0001283, "router_trainable": true, "sec_per_step": 0.601, "probe_sec": 0.0608, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04026, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": 0.01484, "n_shadow": 4, "mean_delta_shadow": -0.07977, "n_split": 2, "mean_delta_split": -0.00192, "n_merge": 1, "mean_delta_merge": -0.02136, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 0.00297, "trunc_abs_frac": 0.69617, "mean_delta_full": -0.25153, "read_hit_rate": {"surface": 0.16909, "split": 0.25825, "merge": 0.01839, "skel": 0.11533}, "mean_gate": {"surface": 0.03727, "split": 0.02978, "merge": 0.04439, "skel": 0.0323}, "n_reads": 4850, "n_cand": 30169, "read_suppressed_frac": 0.1056, "dropout_reads_logged": 4850, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 513, "admitted_surface": 25738, "staged_surface": 58892, "admitted_split": 4639, "staged_split": 43959, "admitted_merge": 1926, "staged_merge": 112638, "admitted_skel": 33233, "staged_skel": 46655, "mean_count": 64.96553039550781, "n_util_obs": 89, "util_count_spearman": -0.2717495392837827}, "probe_overhead_sec_total": 315.633, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.003507, "alpha": 0.03, "t": 3267.6}
{"event": "consolidate", "step": 3700, "retired": 0, "split_retired": 1, "merge_admitted": 0, "admitted": 1, "swapped": 313, "uncertain_staged": 262036, "t": 3298.7}
{"event": "train", "step": 3700, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.39666, "bpb": 2.01496, "chunk_loss": 0.03038, "F": 0.20074, "G": 0.20781, "chunks": 205.56, "bytes_per_chunk": 4.981, "boundary_entropy": 0.13719, "router_noise": 0.0, "grad_norm": 0.6979, "lr": 0.0001247, "router_trainable": true, "sec_per_step": 0.637, "probe_sec": 0.4554, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.03875, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": -0.00445, "n_shadow": 3, "mean_delta_shadow": 0.01595, "n_split": 0, "mean_delta_split": NaN, "n_merge": 1, "mean_delta_merge": -0.12934, "n_dropped_unpaired": 0, "n_full": 1, "trunc_ratio_sum": 0.26894, "trunc_abs_frac": 0.26894, "mean_delta_full": 0.48413, "read_hit_rate": {"surface": 0.17278, "split": 0.31479, "merge": 0.01888, "skel": 0.11793}, "mean_gate": {"surface": 0.03661, "split": 0.02913, "merge": 0.04766, "skel": 0.03345}, "n_reads": 5399, "n_cand": 29399, "read_suppressed_frac": 0.0985, "dropout_reads_logged": 5399, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 527, "admitted_surface": 25748, "staged_surface": 58802, "admitted_split": 4931, "staged_split": 43647, "admitted_merge": 1928, "staged_merge": 112947, "admitted_skel": 32929, "staged_skel": 46748, "mean_count": 72.05657958984375, "n_util_obs": 95, "util_count_spearman": -0.29063549832026875}, "probe_overhead_sec_total": 319.347, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.003433, "alpha": 0.03, "t": 3299.5}
{"event": "consolidate", "step": 3750, "retired": 1, "split_retired": 0, "merge_admitted": 0, "admitted": 1, "swapped": 117, "uncertain_staged": 262037, "t": 3330.5}
{"event": "train", "step": 3750, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.38023, "bpb": 1.99125, "chunk_loss": 0.030294, "F": 0.19437, "G": 0.20299, "chunks": 199.03, "bytes_per_chunk": 5.145, "boundary_entropy": 0.13245, "router_noise": 0.0, "grad_norm": 0.7008, "lr": 0.0001212, "router_trainable": true, "sec_per_step": 0.638, "probe_sec": 0.06, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04313, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": 0.02264, "n_shadow": 3, "mean_delta_shadow": -0.016, "n_split": 0, "mean_delta_split": NaN, "n_merge": 1, "mean_delta_merge": -0.00757, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": -0.01705, "trunc_abs_frac": 0.01856, "mean_delta_full": -0.36163, "read_hit_rate": {"surface": 0.17011, "split": 0.33354, "merge": 0.02284, "skel": 0.11659}, "mean_gate": {"surface": 0.03371, "split": 0.02786, "merge": 0.04263, "skel": 0.02958}, "n_reads": 5512, "n_cand": 28812, "read_suppressed_frac": 0.1016, "dropout_reads_logged": 5512, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 566, "admitted_surface": 25758, "staged_surface": 58877, "admitted_split": 5035, "staged_split": 43742, "admitted_merge": 1928, "staged_merge": 113151, "admitted_skel": 32815, "staged_skel": 46374, "mean_count": 78.08958435058594, "n_util_obs": 96, "util_count_spearman": -0.28399348887683123}, "probe_overhead_sec_total": 322.553, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.003437, "alpha": 0.03, "t": 3331.3}
{"event": "consolidate", "step": 3800, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 92, "uncertain_staged": 262037, "t": 3359.2}
{"event": "probe", "step": 3800, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3421.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.196575, "cps": 26.726562, "mean_p": 0.204669, "boundary_entropy": 0.140552, "bpb": 1.953946}, "tr": {"chunks": 3820.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.211015, "cps": 29.84375, "mean_p": 0.201743, "boundary_entropy": 0.129509, "bpb": 2.10969}, "fi": {"chunks": 4502.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.239328, "cps": 35.171875, "mean_p": 0.236914, "boundary_entropy": 0.121296, "bpb": 2.217217}}, "bpb": {"en": 1.95395, "tr": 2.10969, "fi": 2.21722}, "gini_cpc": 0.044058, "gini_cps": 0.06137, "bpic": 4.831133, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0734550352950458, "fi": 1.2174878987866053}, "inventory": {"size": 327588, "admitted": 65536, "staged": 262052, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 583, "admitted_surface": 25789, "staged_surface": 58968, "admitted_split": 5094, "staged_split": 43905, "admitted_merge": 1928, "staged_merge": 112662, "admitted_skel": 32725, "staged_skel": 46517, "mean_count": 86.87594604492188, "n_util_obs": 104, "util_count_spearman": -0.2576236467086052}, "t": 3360.4}
{"event": "eval", "step": 3800, "phase": 2, "bpb": {"en": 1.9688, "tr": 2.11616, "fi": 2.22528}, "bytes_per_chunk": {"en": 5.189, "tr": 5.315, "fi": 4.571}, "bpb_mean": 2.10341, "frozen_inventory": true, "t": 3362.6}
{"event": "train", "step": 3800, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.44995, "bpb": 2.09183, "chunk_loss": 0.030312, "F": 0.19675, "G": 0.20462, "chunks": 201.47, "bytes_per_chunk": 5.083, "boundary_entropy": 0.12866, "router_noise": 0.0, "grad_norm": 0.6869, "lr": 0.0001177, "router_trainable": true, "sec_per_step": 0.7, "probe_sec": 0.0594, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04119, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": 0.02139, "n_shadow": 6, "mean_delta_shadow": 0.02535, "n_split": 1, "mean_delta_split": -0.0509, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 0.16794, "trunc_abs_frac": 1.16843, "mean_delta_full": 0.07382, "read_hit_rate": {"surface": 0.18722, "split": 0.32251, "merge": 0.01755, "skel": 0.12047}, "mean_gate": {"surface": 0.03148, "split": 0.02602, "merge": 0.0413, "skel": 0.02897}, "n_reads": 5594, "n_cand": 29322, "read_suppressed_frac": 0.099, "dropout_reads_logged": 5594, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 584, "admitted_surface": 25789, "staged_surface": 59003, "admitted_split": 5094, "staged_split": 43994, "admitted_merge": 1928, "staged_merge": 112629, "admitted_skel": 32725, "staged_skel": 46518, "mean_count": 86.9613037109375, "n_util_obs": 104, "util_count_spearman": -0.2567167841596248}, "probe_overhead_sec_total": 325.524, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.00335, "alpha": 0.03, "t": 3366.3}
{"event": "consolidate", "step": 3850, "retired": 1, "split_retired": 0, "merge_admitted": 0, "admitted": 1, "swapped": 40, "uncertain_staged": 262039, "t": 3396.2}
{"event": "train", "step": 3850, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.54536, "bpb": 2.22948, "chunk_loss": 0.030649, "F": 0.20471, "G": 0.21094, "chunks": 209.62, "bytes_per_chunk": 4.885, "boundary_entropy": 0.12558, "router_noise": 0.0, "grad_norm": 0.7054, "lr": 0.0001142, "router_trainable": true, "sec_per_step": 0.606, "probe_sec": 0.0604, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04109, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": -0.03314, "n_shadow": 2, "mean_delta_shadow": -0.03571, "n_split": 2, "mean_delta_split": 0.08881, "n_merge": 1, "mean_delta_merge": -0.08093, "n_dropped_unpaired": 0, "n_full": 1, "trunc_ratio_sum": -1.73418, "trunc_abs_frac": 1.73418, "mean_delta_full": 0.01929, "read_hit_rate": {"surface": 0.20551, "split": 0.33984, "merge": 0.01866, "skel": 0.11382}, "mean_gate": {"surface": 0.03442, "split": 0.02792, "merge": 0.04181, "skel": 0.03212}, "n_reads": 5877, "n_cand": 29542, "read_suppressed_frac": 0.1055, "dropout_reads_logged": 5877, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 606, "admitted_surface": 25811, "staged_surface": 58863, "admitted_split": 5110, "staged_split": 44005, "admitted_merge": 1929, "staged_merge": 112921, "admitted_skel": 32686, "staged_skel": 46355, "mean_count": 96.08648681640625, "n_util_obs": 107, "util_count_spearman": -0.2814808283861361}, "probe_overhead_sec_total": 328.6, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.00331, "alpha": 0.03, "t": 3396.6}
{"event": "consolidate", "step": 3900, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 24, "uncertain_staged": 262040, "t": 3425.1}
{"event": "train", "step": 3900, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.4394, "bpb": 2.07662, "chunk_loss": 0.030404, "F": 0.19827, "G": 0.20609, "chunks": 203.03, "bytes_per_chunk": 5.044, "boundary_entropy": 0.13524, "router_noise": 0.0, "grad_norm": 0.7259, "lr": 0.0001108, "router_trainable": true, "sec_per_step": 0.577, "probe_sec": 0.056, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.03686, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": 0.04505, "n_shadow": 1, "mean_delta_shadow": -0.04395, "n_split": 2, "mean_delta_split": -0.02533, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 0.64428, "trunc_abs_frac": 0.38923, "mean_delta_full": 0.06528, "read_hit_rate": {"surface": 0.22413, "split": 0.3494, "merge": 0.01632, "skel": 0.11826}, "mean_gate": {"surface": 0.0319, "split": 0.02516, "merge": 0.03955, "skel": 0.02892}, "n_reads": 6110, "n_cand": 29361, "read_suppressed_frac": 0.1018, "dropout_reads_logged": 6110, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 629, "admitted_surface": 25826, "staged_surface": 58987, "admitted_split": 5119, "staged_split": 44207, "admitted_merge": 1929, "staged_merge": 112303, "admitted_skel": 32662, "staged_skel": 46647, "mean_count": 102.90309143066406, "n_util_obs": 113, "util_count_spearman": -0.27640727926009717}, "probe_overhead_sec_total": 331.717, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.003218, "alpha": 0.03, "t": 3425.5}
{"event": "consolidate", "step": 3950, "retired": 0, "split_retired": 1, "merge_admitted": 0, "admitted": 1, "swapped": 32, "uncertain_staged": 262041, "t": 3454.0}
{"event": "train", "step": 3950, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.49131, "bpb": 2.15151, "chunk_loss": 0.030353, "F": 0.19476, "G": 0.20286, "chunks": 199.44, "bytes_per_chunk": 5.134, "boundary_entropy": 0.12578, "router_noise": 0.0, "grad_norm": 0.7715, "lr": 0.0001074, "router_trainable": true, "sec_per_step": 0.584, "probe_sec": 0.0599, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04132, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": -0.00063, "n_shadow": 4, "mean_delta_shadow": -0.02015, "n_split": 1, "mean_delta_split": 0.03632, "n_merge": 1, "mean_delta_merge": 0.02844, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": 5.13565, "trunc_abs_frac": 0.82435, "mean_delta_full": -0.00107, "read_hit_rate": {"surface": 0.2526, "split": 0.35094, "merge": 0.01694, "skel": 0.12797}, "mean_gate": {"surface": 0.0333, "split": 0.0269, "merge": 0.0452, "skel": 0.02987}, "n_reads": 6238, "n_cand": 28780, "read_suppressed_frac": 0.097, "dropout_reads_logged": 6238, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 653, "admitted_surface": 25847, "staged_surface": 58886, "admitted_split": 5125, "staged_split": 44093, "admitted_merge": 1930, "staged_merge": 112610, "admitted_skel": 32634, "staged_skel": 46555, "mean_count": 108.58148193359375, "n_util_obs": 116, "util_count_spearman": -0.29183541274688274}, "probe_overhead_sec_total": 334.76, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.003141, "alpha": 0.03, "t": 3454.7}
{"event": "consolidate", "step": 4000, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 12, "uncertain_staged": 262042, "t": 3484.8}
{"event": "probe", "step": 4000, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3411.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.196001, "cps": 26.648438, "mean_p": 0.203445, "boundary_entropy": 0.138747, "bpb": 1.944382}, "tr": {"chunks": 3868.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.213666, "cps": 30.21875, "mean_p": 0.204508, "boundary_entropy": 0.133028, "bpb": 2.090189}, "fi": {"chunks": 4477.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.237999, "cps": 34.976562, "mean_p": 0.235255, "boundary_entropy": 0.122005, "bpb": 2.200212}}, "bpb": {"en": 1.94438, "tr": 2.09019, "fi": 2.20021}, "gini_cpc": 0.04323, "gini_cps": 0.060451, "bpic": 4.825791, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0901300585708351, "fi": 1.214276560333273}, "inventory": {"size": 327668, "admitted": 65536, "staged": 262132, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 685, "admitted_surface": 25854, "staged_surface": 58783, "admitted_split": 5129, "staged_split": 43980, "admitted_merge": 1930, "staged_merge": 112720, "admitted_skel": 32623, "staged_skel": 46649, "mean_count": 114.10165405273438, "n_util_obs": 118, "util_count_spearman": -0.2989456538808483}, "t": 3486.2}
{"event": "eval", "step": 4000, "phase": 2, "bpb": {"en": 1.95777, "tr": 2.09737, "fi": 2.20608}, "bytes_per_chunk": {"en": 5.201, "tr": 5.271, "fi": 4.604}, "bpb_mean": 2.08707, "frozen_inventory": true, "t": 3488.5}
{"event": "train", "step": 4000, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.41458, "bpb": 2.04081, "chunk_loss": 0.030333, "F": 0.19458, "G": 0.20227, "chunks": 199.25, "bytes_per_chunk": 5.139, "boundary_entropy": 0.13349, "router_noise": 0.0, "grad_norm": 0.6964, "lr": 0.000104, "router_trainable": true, "sec_per_step": 0.777, "probe_sec": 0.0603, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.03948, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": 0.01905, "n_shadow": 4, "mean_delta_shadow": -0.0359, "n_split": 1, "mean_delta_split": 0.03589, "n_merge": 1, "mean_delta_merge": 0.12915, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": -0.07504, "trunc_abs_frac": 0.77639, "mean_delta_full": 0.08704, "read_hit_rate": {"surface": 0.25315, "split": 0.35245, "merge": 0.01774, "skel": 0.12502}, "mean_gate": {"surface": 0.03185, "split": 0.02502, "merge": 0.03926, "skel": 0.02962}, "n_reads": 6280, "n_cand": 28877, "read_suppressed_frac": 0.1051, "dropout_reads_logged": 6280, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 687, "admitted_surface": 25854, "staged_surface": 58760, "admitted_split": 5129, "staged_split": 44036, "admitted_merge": 1930, "staged_merge": 112734, "admitted_skel": 32623, "staged_skel": 46614, "mean_count": 114.19747924804688, "n_util_obs": 118, "util_count_spearman": -0.2961335772901077}, "probe_overhead_sec_total": 337.813, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.00311, "alpha": 0.03, "t": 3493.6}
{"event": "consolidate", "step": 4050, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 19, "uncertain_staged": 262042, "t": 3524.2}
{"event": "train", "step": 4050, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.42877, "bpb": 2.06128, "chunk_loss": 0.030352, "F": 0.19763, "G": 0.20485, "chunks": 202.38, "bytes_per_chunk": 5.06, "boundary_entropy": 0.12485, "router_noise": 0.0, "grad_norm": 0.7251, "lr": 0.0001007, "router_trainable": true, "sec_per_step": 0.635, "probe_sec": 0.0779, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04271, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 4, "mean_delta_loo": 0.0016, "n_shadow": 7, "mean_delta_shadow": 0.05874, "n_split": 1, "mean_delta_split": 0.09235, "n_merge": 1, "mean_delta_merge": 0.09937, "n_dropped_unpaired": 0, "n_full": 0, "read_hit_rate": {"surface": 0.26226, "split": 0.36604, "merge": 0.0209, "skel": 0.12121}, "mean_gate": {"surface": 0.03474, "split": 0.02724, "merge": 0.04912, "skel": 0.03091}, "n_reads": 6502, "n_cand": 29030, "read_suppressed_frac": 0.0997, "dropout_reads_logged": 6502, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 715, "admitted_surface": 25866, "staged_surface": 58932, "admitted_split": 5135, "staged_split": 44165, "admitted_merge": 1931, "staged_merge": 112376, "admitted_skel": 32604, "staged_skel": 46671, "mean_count": 119.18290710449219, "n_util_obs": 121, "util_count_spearman": -0.2914181392041483}, "probe_overhead_sec_total": 340.923, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.002997, "alpha": 0.03, "t": 3525.3}
{"event": "consolidate", "step": 4100, "retired": 0, "split_retired": 2, "merge_admitted": 0, "admitted": 2, "swapped": 18, "uncertain_staged": 262041, "t": 3553.2}
{"event": "train", "step": 4100, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.36241, "bpb": 1.96554, "chunk_loss": 0.030399, "F": 0.19876, "G": 0.20564, "chunks": 203.53, "bytes_per_chunk": 5.031, "boundary_entropy": 0.12922, "router_noise": 0.0, "grad_norm": 0.7333, "lr": 9.75e-05, "router_trainable": true, "sec_per_step": 0.569, "probe_sec": 0.0584, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.0392, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": 0.04509, "n_shadow": 3, "mean_delta_shadow": 0.02714, "n_split": 2, "mean_delta_split": -0.02295, "n_merge": 1, "mean_delta_merge": -0.00867, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": -0.11405, "trunc_abs_frac": 0.56061, "mean_delta_full": -0.1759, "read_hit_rate": {"surface": 0.2711, "split": 0.37893, "merge": 0.02465, "skel": 0.13488}, "mean_gate": {"surface": 0.03312, "split": 0.02639, "merge": 0.04114, "skel": 0.03016}, "n_reads": 6873, "n_cand": 29311, "read_suppressed_frac": 0.1085, "dropout_reads_logged": 6873, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 712, "admitted_surface": 25872, "staged_surface": 58978, "admitted_split": 5145, "staged_split": 43970, "admitted_merge": 1932, "staged_merge": 112696, "admitted_skel": 32587, "staged_skel": 46500, "mean_count": 128.6864471435547, "n_util_obs": 129, "util_count_spearman": -0.27042151162790695}, "probe_overhead_sec_total": 343.975, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.002874, "alpha": 0.03, "t": 3553.8}
{"event": "consolidate", "step": 4150, "retired": 0, "split_retired": 2, "merge_admitted": 0, "admitted": 2, "swapped": 11, "uncertain_staged": 262042, "t": 3583.0}
{"event": "train", "step": 4150, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.43094, "bpb": 2.06441, "chunk_loss": 0.0305, "F": 0.20496, "G": 0.2115, "chunks": 209.88, "bytes_per_chunk": 4.879, "boundary_entropy": 0.127, "router_noise": 0.0, "grad_norm": 0.7764, "lr": 9.43e-05, "router_trainable": true, "sec_per_step": 0.591, "probe_sec": 0.0592, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04126, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": -0.01665, "n_shadow": 4, "mean_delta_shadow": -0.04235, "n_split": 1, "mean_delta_split": 0.16016, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 6, "trunc_ratio_sum": -1.42388, "trunc_abs_frac": 0.51148, "mean_delta_full": 0.00682, "read_hit_rate": {"surface": 0.25898, "split": 0.36168, "merge": 0.01999, "skel": 0.12942}, "mean_gate": {"surface": 0.03346, "split": 0.02493, "merge": 0.04153, "skel": 0.02901}, "n_reads": 6775, "n_cand": 30328, "read_suppressed_frac": 0.1038, "dropout_reads_logged": 6775, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 725, "admitted_surface": 25878, "staged_surface": 58803, "admitted_split": 5150, "staged_split": 44124, "admitted_merge": 1932, "staged_merge": 112640, "admitted_skel": 32576, "staged_skel": 46577, "mean_count": 131.96200561523438, "n_util_obs": 133, "util_count_spearman": -0.2962783485171545}, "probe_overhead_sec_total": 347.4, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.002807, "alpha": 0.03, "t": 3583.4}
{"event": "consolidate", "step": 4200, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 16, "uncertain_staged": 262042, "t": 3612.7}
{"event": "probe", "step": 4200, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3445.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.197954, "cps": 26.914062, "mean_p": 0.204148, "boundary_entropy": 0.133026, "bpb": 1.935104}, "tr": {"chunks": 3886.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.214661, "cps": 30.359375, "mean_p": 0.203853, "boundary_entropy": 0.126078, "bpb": 2.075166}, "fi": {"chunks": 4530.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.240817, "cps": 35.390625, "mean_p": 0.237081, "boundary_entropy": 0.116549, "bpb": 2.189152}}, "bpb": {"en": 1.9351, "tr": 2.07517, "fi": 2.18915}, "gini_cpc": 0.04373, "gini_cps": 0.060984, "bpic": 4.783071, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0843940820175344, "fi": 1.2165254881670309}, "inventory": {"size": 327664, "admitted": 65536, "staged": 262128, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 743, "admitted_surface": 25883, "staged_surface": 58866, "admitted_split": 5159, "staged_split": 44025, "admitted_merge": 1934, "staged_merge": 112770, "admitted_skel": 32560, "staged_skel": 46467, "mean_count": 138.02125549316406, "n_util_obs": 139, "util_count_spearman": -0.3077483354930962}, "t": 3614.0}
{"event": "eval", "step": 4200, "phase": 2, "bpb": {"en": 1.94894, "tr": 2.08256, "fi": 2.19616}, "bytes_per_chunk": {"en": 5.153, "tr": 5.252, "fi": 4.563}, "bpb_mean": 2.07589, "frozen_inventory": true, "t": 3616.0}
{"event": "phase_transition", "step": 4200, "phase": 3, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0, "t": 3619.2}
{"event": "train", "step": 4200, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.35652, "bpb": 1.95704, "chunk_loss": 0.030256, "F": 0.19241, "G": 0.19897, "chunks": 197.03, "bytes_per_chunk": 5.197, "boundary_entropy": 0.12499, "router_noise": 0.0, "grad_norm": 0.7483, "lr": 9.12e-05, "router_trainable": true, "sec_per_step": 0.725, "probe_sec": 0.0617, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04204, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": -0.00417, "n_shadow": 3, "mean_delta_shadow": 0.02929, "n_split": 2, "mean_delta_split": -0.01941, "n_merge": 1, "mean_delta_merge": 0.03876, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 0.48415, "trunc_abs_frac": 0.25951, "mean_delta_full": -0.052, "read_hit_rate": {"surface": 0.27483, "split": 0.37406, "merge": 0.02243, "skel": 0.14391}, "mean_gate": {"surface": 0.0327, "split": 0.02647, "merge": 0.04394, "skel": 0.03048}, "n_reads": 6677, "n_cand": 28472, "read_suppressed_frac": 0.0945, "dropout_reads_logged": 6677, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 743, "admitted_surface": 25883, "staged_surface": 58884, "admitted_split": 5159, "staged_split": 44034, "admitted_merge": 1934, "staged_merge": 112665, "admitted_skel": 32560, "staged_skel": 46561, "mean_count": 138.12313842773438, "n_util_obs": 139, "util_count_spearman": -0.3068099557620984}, "probe_overhead_sec_total": 350.544, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.002801, "alpha": 0.03, "t": 3619.6}
{"event": "consolidate", "step": 4250, "retired": 0, "split_retired": 1, "merge_admitted": 0, "admitted": 1, "swapped": 8, "uncertain_staged": 262042, "t": 3650.8}
{"event": "train", "step": 4250, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.38716, "bpb": 2.00124, "chunk_loss": 0.03035, "F": 0.19882, "G": 0.20622, "chunks": 203.59, "bytes_per_chunk": 5.03, "boundary_entropy": 0.12608, "router_noise": 0.0, "grad_norm": 0.7171, "lr": 8.81e-05, "router_trainable": true, "sec_per_step": 0.632, "probe_sec": 0.0593, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04019, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": 0.02526, "n_shadow": 5, "mean_delta_shadow": 0.01835, "n_split": 0, "mean_delta_split": NaN, "n_merge": 1, "mean_delta_merge": -0.02316, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 0.45483, "trunc_abs_frac": 1.87558, "mean_delta_full": 0.12027, "read_hit_rate": {"surface": 0.28135, "split": 0.3817, "merge": 0.02573, "skel": 0.1444}, "mean_gate": {"surface": 0.03351, "split": 0.02633, "merge": 0.04019, "skel": 0.03041}, "n_reads": 7055, "n_cand": 29424, "read_suppressed_frac": 0.0988, "dropout_reads_logged": 7055, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 763, "admitted_surface": 25886, "staged_surface": 58710, "admitted_split": 5163, "staged_split": 44185, "admitted_merge": 1934, "staged_merge": 112701, "admitted_skel": 32553, "staged_skel": 46548, "mean_count": 144.66444396972656, "n_util_obs": 150, "util_count_spearman": -0.269485754922441}, "probe_overhead_sec_total": 353.684, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.002786, "alpha": 0.03, "t": 3651.2}
{"event": "consolidate", "step": 4300, "retired": 1, "split_retired": 0, "merge_admitted": 0, "admitted": 1, "swapped": 16, "uncertain_staged": 262044, "t": 3679.7}
{"event": "train", "step": 4300, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.38926, "bpb": 2.00428, "chunk_loss": 0.030288, "F": 0.19531, "G": 0.20139, "chunks": 200.0, "bytes_per_chunk": 5.12, "boundary_entropy": 0.12612, "router_noise": 0.0, "grad_norm": 0.7543, "lr": 8.5e-05, "router_trainable": true, "sec_per_step": 0.585, "probe_sec": 0.0587, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.0404, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 3, "mean_delta_loo": -0.01723, "n_shadow": 7, "mean_delta_shadow": 0.09919, "n_split": 1, "mean_delta_split": -0.05151, "n_merge": 2, "mean_delta_merge": -0.06512, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": -0.79976, "trunc_abs_frac": 0.33038, "mean_delta_full": -0.06889, "read_hit_rate": {"surface": 0.27952, "split": 0.37408, "merge": 0.02146, "skel": 0.13657}, "mean_gate": {"surface": 0.03529, "split": 0.02774, "merge": 0.04343, "skel": 0.03229}, "n_reads": 6767, "n_cand": 28903, "read_suppressed_frac": 0.1058, "dropout_reads_logged": 6767, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 758, "admitted_surface": 25894, "staged_surface": 58759, "admitted_split": 5169, "staged_split": 44130, "admitted_merge": 1934, "staged_merge": 112550, "admitted_skel": 32539, "staged_skel": 46705, "mean_count": 150.15545654296875, "n_util_obs": 162, "util_count_spearman": -0.25166517185200754}, "probe_overhead_sec_total": 356.758, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.002771, "alpha": 0.03, "t": 3680.5}
{"event": "consolidate", "step": 4350, "retired": 1, "split_retired": 0, "merge_admitted": 0, "admitted": 1, "swapped": 9, "uncertain_staged": 262044, "t": 3708.8}
{"event": "train", "step": 4350, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.4134, "bpb": 2.03911, "chunk_loss": 0.030351, "F": 0.19754, "G": 0.20343, "chunks": 202.28, "bytes_per_chunk": 5.062, "boundary_entropy": 0.12281, "router_noise": 0.0, "grad_norm": 0.6794, "lr": 8.21e-05, "router_trainable": true, "sec_per_step": 0.575, "probe_sec": 0.0622, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.0427, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 10, "mean_delta_loo": 0.02674, "n_shadow": 3, "mean_delta_shadow": -0.00201, "n_split": 0, "mean_delta_split": NaN, "n_merge": 1, "mean_delta_merge": 0.05365, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": -0.39876, "trunc_abs_frac": 0.40193, "mean_delta_full": 0.02718, "read_hit_rate": {"surface": 0.28645, "split": 0.36839, "merge": 0.0245, "skel": 0.13911}, "mean_gate": {"surface": 0.03474, "split": 0.02711, "merge": 0.04383, "skel": 0.03251}, "n_reads": 6883, "n_cand": 29303, "read_suppressed_frac": 0.1024, "dropout_reads_logged": 6883, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 781, "admitted_surface": 25896, "staged_surface": 58920, "admitted_split": 5174, "staged_split": 43643, "admitted_merge": 1935, "staged_merge": 112909, "admitted_skel": 32531, "staged_skel": 46672, "mean_count": 156.6031951904297, "n_util_obs": 169, "util_count_spearman": -0.22840485306548655}, "probe_overhead_sec_total": 359.931, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.002752, "alpha": 0.03, "t": 3709.2}
{"event": "consolidate", "step": 4400, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 10, "uncertain_staged": 262042, "t": 3740.0}
{"event": "probe", "step": 4400, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3461.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.198874, "cps": 27.039062, "mean_p": 0.2049, "boundary_entropy": 0.130702, "bpb": 1.925687}, "tr": {"chunks": 3816.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.210794, "cps": 29.8125, "mean_p": 0.201659, "boundary_entropy": 0.126951, "bpb": 2.061871}, "fi": {"chunks": 4509.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.2397, "cps": 35.226562, "mean_p": 0.236207, "boundary_entropy": 0.115955, "bpb": 2.17462}}, "bpb": {"en": 1.92569, "tr": 2.06187, "fi": 2.17462}, "gini_cpc": 0.041914, "gini_cps": 0.059279, "bpic": 4.813508, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0599376903325497, "fi": 1.2052881112382512}, "inventory": {"size": 327670, "admitted": 65536, "staged": 262134, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 794, "admitted_surface": 25898, "staged_surface": 58694, "admitted_split": 5176, "staged_split": 43812, "admitted_merge": 1936, "staged_merge": 112818, "admitted_skel": 32526, "staged_skel": 46810, "mean_count": 163.58595275878906, "n_util_obs": 179, "util_count_spearman": -0.20309878015608981}, "t": 3741.3}
{"event": "eval", "step": 4400, "phase": 3, "bpb": {"en": 1.93991, "tr": 2.06848, "fi": 2.18211}, "bytes_per_chunk": {"en": 5.129, "tr": 5.334, "fi": 4.57}, "bpb_mean": 2.0635, "frozen_inventory": true, "t": 3743.4}
{"event": "train", "step": 4400, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.36595, "bpb": 1.97065, "chunk_loss": 0.030314, "F": 0.19714, "G": 0.2032, "chunks": 201.88, "bytes_per_chunk": 5.072, "boundary_entropy": 0.12607, "router_noise": 0.0, "grad_norm": 0.6708, "lr": 7.92e-05, "router_trainable": true, "sec_per_step": 0.764, "probe_sec": 0.0646, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04051, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 5, "mean_delta_loo": 0.01418, "n_shadow": 7, "mean_delta_shadow": 0.00673, "n_split": 0, "mean_delta_split": NaN, "n_merge": 1, "mean_delta_merge": -0.06772, "n_dropped_unpaired": 0, "n_full": 1, "trunc_ratio_sum": -0.39327, "trunc_abs_frac": 0.39327, "mean_delta_full": 0.24475, "read_hit_rate": {"surface": 0.2976, "split": 0.39327, "merge": 0.02267, "skel": 0.14409}, "mean_gate": {"surface": 0.03511, "split": 0.02703, "merge": 0.04508, "skel": 0.03326}, "n_reads": 7199, "n_cand": 29124, "read_suppressed_frac": 0.0927, "dropout_reads_logged": 7199, "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": 25898, "staged_surface": 58752, "admitted_split": 5176, "staged_split": 43778, "admitted_merge": 1936, "staged_merge": 112847, "admitted_skel": 32526, "staged_skel": 46767, "mean_count": 163.69580078125, "n_util_obs": 179, "util_count_spearman": -0.2038029647662932}, "probe_overhead_sec_total": 363.137, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.002733, "alpha": 0.03, "t": 3747.4}
{"event": "consolidate", "step": 4450, "retired": 1, "split_retired": 0, "merge_admitted": 0, "admitted": 1, "swapped": 2, "uncertain_staged": 262045, "t": 3776.4}
{"event": "train", "step": 4450, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.36323, "bpb": 1.96673, "chunk_loss": 0.030206, "F": 0.18805, "G": 0.19474, "chunks": 192.56, "bytes_per_chunk": 5.318, "boundary_entropy": 0.12783, "router_noise": 0.0, "grad_norm": 0.7205, "lr": 7.63e-05, "router_trainable": true, "sec_per_step": 0.609, "probe_sec": 0.0705, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04117, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 5, "mean_delta_loo": -0.005, "n_shadow": 7, "mean_delta_shadow": -0.04946, "n_split": 1, "mean_delta_split": 0.10223, "n_merge": 1, "mean_delta_merge": -0.07925, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 0.30701, "trunc_abs_frac": 0.4274, "mean_delta_full": -0.19403, "read_hit_rate": {"surface": 0.30865, "split": 0.39421, "merge": 0.0223, "skel": 0.14078}, "mean_gate": {"surface": 0.03578, "split": 0.02856, "merge": 0.04302, "skel": 0.03463}, "n_reads": 6924, "n_cand": 27744, "read_suppressed_frac": 0.1014, "dropout_reads_logged": 6924, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 818, "admitted_surface": 25901, "staged_surface": 58752, "admitted_split": 5175, "staged_split": 43799, "admitted_merge": 1936, "staged_merge": 112685, "admitted_skel": 32524, "staged_skel": 46908, "mean_count": 169.20147705078125, "n_util_obs": 192, "util_count_spearman": -0.2115875893458694}, "probe_overhead_sec_total": 366.273, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.002775, "alpha": 0.03, "t": 3777.9}
{"event": "consolidate", "step": 4500, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 3, "uncertain_staged": 262046, "t": 3810.8}
{"event": "train", "step": 4500, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.3386, "bpb": 1.93119, "chunk_loss": 0.030294, "F": 0.1955, "G": 0.20174, "chunks": 200.19, "bytes_per_chunk": 5.115, "boundary_entropy": 0.12306, "router_noise": 0.0, "grad_norm": 0.6684, "lr": 7.36e-05, "router_trainable": true, "sec_per_step": 0.667, "probe_sec": 0.0607, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04192, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": -0.00494, "n_shadow": 5, "mean_delta_shadow": -0.04528, "n_split": 0, "mean_delta_split": NaN, "n_merge": 1, "mean_delta_merge": -0.00793, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": -0.10325, "trunc_abs_frac": 0.41351, "mean_delta_full": 0.09635, "read_hit_rate": {"surface": 0.29981, "split": 0.37958, "merge": 0.02507, "skel": 0.14391}, "mean_gate": {"surface": 0.03386, "split": 0.02582, "merge": 0.04444, "skel": 0.0307}, "n_reads": 7059, "n_cand": 29052, "read_suppressed_frac": 0.1004, "dropout_reads_logged": 7059, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 830, "admitted_surface": 25901, "staged_surface": 58712, "admitted_split": 5178, "staged_split": 43758, "admitted_merge": 1936, "staged_merge": 112990, "admitted_skel": 32521, "staged_skel": 46684, "mean_count": 174.85205078125, "n_util_obs": 198, "util_count_spearman": -0.21984170970217956}, "probe_overhead_sec_total": 369.457, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.002653, "alpha": 0.03, "t": 3811.2}
{"event": "consolidate", "step": 4550, "retired": 0, "split_retired": 1, "merge_admitted": 0, "admitted": 1, "swapped": 3, "uncertain_staged": 262044, "t": 3840.0}
{"event": "train", "step": 4550, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.36306, "bpb": 1.96648, "chunk_loss": 0.030331, "F": 0.19742, "G": 0.20418, "chunks": 202.16, "bytes_per_chunk": 5.065, "boundary_entropy": 0.11921, "router_noise": 0.0, "grad_norm": 0.6732, "lr": 7.09e-05, "router_trainable": true, "sec_per_step": 0.582, "probe_sec": 0.0568, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.03795, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": 0.03106, "n_shadow": 3, "mean_delta_shadow": -0.03968, "n_split": 1, "mean_delta_split": -0.05734, "n_merge": 2, "mean_delta_merge": 0.02148, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 0.16001, "trunc_abs_frac": 0.12969, "mean_delta_full": -0.20026, "read_hit_rate": {"surface": 0.27855, "split": 0.38516, "merge": 0.02483, "skel": 0.13118}, "mean_gate": {"surface": 0.03488, "split": 0.02729, "merge": 0.04519, "skel": 0.03128}, "n_reads": 6946, "n_cand": 29194, "read_suppressed_frac": 0.0934, "dropout_reads_logged": 6946, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 852, "admitted_surface": 25901, "staged_surface": 58609, "admitted_split": 5181, "staged_split": 43786, "admitted_merge": 1936, "staged_merge": 113156, "admitted_skel": 32518, "staged_skel": 46593, "mean_count": 180.16220092773438, "n_util_obs": 204, "util_count_spearman": -0.22063594150864718}, "probe_overhead_sec_total": 372.97, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.002684, "alpha": 0.03, "t": 3840.3}
{"event": "consolidate", "step": 4600, "retired": 0, "split_retired": 2, "merge_admitted": 0, "admitted": 2, "swapped": 1, "uncertain_staged": 262044, "t": 3868.3}
{"event": "probe", "step": 4600, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3451.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.198299, "cps": 26.960938, "mean_p": 0.204595, "boundary_entropy": 0.128557, "bpb": 1.917539}, "tr": {"chunks": 3864.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.213445, "cps": 30.1875, "mean_p": 0.203968, "boundary_entropy": 0.12243, "bpb": 2.052564}, "fi": {"chunks": 4535.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.241082, "cps": 35.429688, "mean_p": 0.236404, "boundary_entropy": 0.11196, "bpb": 2.159571}}, "bpb": {"en": 1.91754, "tr": 2.05256, "fi": 2.15957}, "gini_cpc": 0.04369, "gini_cps": 0.060985, "bpic": 4.787511, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0763802666710292, "fi": 1.2157508135687793}, "inventory": {"size": 327677, "admitted": 65536, "staged": 262141, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 871, "admitted_surface": 25900, "staged_surface": 58709, "admitted_split": 5183, "staged_split": 43778, "admitted_merge": 1936, "staged_merge": 112949, "admitted_skel": 32517, "staged_skel": 46705, "mean_count": 185.23095703125, "n_util_obs": 212, "util_count_spearman": -0.21136962164653844}, "t": 3869.7}
{"event": "eval", "step": 4600, "phase": 3, "bpb": {"en": 1.93094, "tr": 2.06021, "fi": 2.16708}, "bytes_per_chunk": {"en": 5.163, "tr": 5.273, "fi": 4.566}, "bpb_mean": 2.05274, "frozen_inventory": true, "t": 3872.0}
{"event": "train", "step": 4600, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.36841, "bpb": 1.9742, "chunk_loss": 0.030337, "F": 0.20007, "G": 0.20577, "chunks": 204.88, "bytes_per_chunk": 4.998, "boundary_entropy": 0.1273, "router_noise": 0.0, "grad_norm": 0.6999, "lr": 6.82e-05, "router_trainable": true, "sec_per_step": 0.707, "probe_sec": 0.0593, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04118, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": -0.00086, "n_shadow": 5, "mean_delta_shadow": 0.02379, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 0.16091, "trunc_abs_frac": 1.32013, "mean_delta_full": 0.05832, "read_hit_rate": {"surface": 0.30258, "split": 0.39242, "merge": 0.02649, "skel": 0.13295}, "mean_gate": {"surface": 0.03451, "split": 0.02623, "merge": 0.0457, "skel": 0.03161}, "n_reads": 7277, "n_cand": 29513, "read_suppressed_frac": 0.0999, "dropout_reads_logged": 7277, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 870, "admitted_surface": 25900, "staged_surface": 58675, "admitted_split": 5183, "staged_split": 43748, "admitted_merge": 1936, "staged_merge": 112999, "admitted_skel": 32517, "staged_skel": 46722, "mean_count": 185.34201049804688, "n_util_obs": 212, "util_count_spearman": -0.21043258567770748}, "probe_overhead_sec_total": 376.383, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.002569, "alpha": 0.03, "t": 3875.7}
{"event": "consolidate", "step": 4650, "retired": 1, "split_retired": 0, "merge_admitted": 0, "admitted": 1, "swapped": 5, "uncertain_staged": 262044, "t": 3906.3}
{"event": "train", "step": 4650, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.39556, "bpb": 2.01337, "chunk_loss": 0.030446, "F": 0.20386, "G": 0.20959, "chunks": 208.75, "bytes_per_chunk": 4.905, "boundary_entropy": 0.11689, "router_noise": 0.0, "grad_norm": 0.6745, "lr": 6.57e-05, "router_trainable": true, "sec_per_step": 0.619, "probe_sec": 0.0633, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04123, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": -0.01901, "n_shadow": 3, "mean_delta_shadow": 0.00989, "n_split": 2, "mean_delta_split": 0.0149, "n_merge": 1, "mean_delta_merge": -0.14403, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 0.14502, "trunc_abs_frac": 1.04344, "mean_delta_full": -0.08008, "read_hit_rate": {"surface": 0.2861, "split": 0.38976, "merge": 0.0201, "skel": 0.12757}, "mean_gate": {"surface": 0.03564, "split": 0.02745, "merge": 0.04577, "skel": 0.03171}, "n_reads": 7137, "n_cand": 29824, "read_suppressed_frac": 0.1005, "dropout_reads_logged": 7137, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 877, "admitted_surface": 25905, "staged_surface": 58799, "admitted_split": 5183, "staged_split": 43698, "admitted_merge": 1936, "staged_merge": 112786, "admitted_skel": 32512, "staged_skel": 46861, "mean_count": 190.7729949951172, "n_util_obs": 219, "util_count_spearman": -0.22221727136616978}, "probe_overhead_sec_total": 379.492, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.002623, "alpha": 0.03, "t": 3906.7}
{"event": "consolidate", "step": 4700, "retired": 0, "split_retired": 2, "merge_admitted": 0, "admitted": 2, "swapped": 2, "uncertain_staged": 262044, "t": 3935.6}
{"event": "train", "step": 4700, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.42284, "bpb": 2.05273, "chunk_loss": 0.030607, "F": 0.20834, "G": 0.21453, "chunks": 213.34, "bytes_per_chunk": 4.8, "boundary_entropy": 0.12257, "router_noise": 0.0, "grad_norm": 0.7197, "lr": 6.32e-05, "router_trainable": true, "sec_per_step": 0.595, "probe_sec": 0.0643, "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": 8, "mean_delta_loo": -0.04324, "n_shadow": 6, "mean_delta_shadow": -0.00241, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": -0.43706, "trunc_abs_frac": 2.26747, "mean_delta_full": -0.0733, "read_hit_rate": {"surface": 0.30537, "split": 0.39246, "merge": 0.0275, "skel": 0.14305}, "mean_gate": {"surface": 0.03512, "split": 0.02606, "merge": 0.04291, "skel": 0.03244}, "n_reads": 7552, "n_cand": 30345, "read_suppressed_frac": 0.0972, "dropout_reads_logged": 7552, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 866, "admitted_surface": 25904, "staged_surface": 58903, "admitted_split": 5186, "staged_split": 43482, "admitted_merge": 1936, "staged_merge": 113019, "admitted_skel": 32510, "staged_skel": 46740, "mean_count": 199.63986206054688, "n_util_obs": 228, "util_count_spearman": -0.22163577304012372}, "probe_overhead_sec_total": 382.651, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.002704, "alpha": 0.03, "t": 3936.4}
{"event": "consolidate", "step": 4750, "retired": 2, "split_retired": 2, "merge_admitted": 0, "admitted": 4, "swapped": 2, "uncertain_staged": 262047, "t": 3967.2}
{"event": "train", "step": 4750, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.40894, "bpb": 2.03267, "chunk_loss": 0.030325, "F": 0.19897, "G": 0.20534, "chunks": 203.75, "bytes_per_chunk": 5.026, "boundary_entropy": 0.11858, "router_noise": 0.0, "grad_norm": 0.7043, "lr": 6.08e-05, "router_trainable": true, "sec_per_step": 0.623, "probe_sec": 0.0589, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04061, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 4, "mean_delta_loo": 0.00172, "n_shadow": 4, "mean_delta_shadow": 0.02372, "n_split": 3, "mean_delta_split": 0.01986, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 0.41797, "trunc_abs_frac": 0.92977, "mean_delta_full": 0.04439, "read_hit_rate": {"surface": 0.28591, "split": 0.39259, "merge": 0.02277, "skel": 0.13774}, "mean_gate": {"surface": 0.03504, "split": 0.02646, "merge": 0.04502, "skel": 0.03142}, "n_reads": 7150, "n_cand": 29388, "read_suppressed_frac": 0.0947, "dropout_reads_logged": 7150, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 881, "admitted_surface": 25901, "staged_surface": 58928, "admitted_split": 5188, "staged_split": 43594, "admitted_merge": 1938, "staged_merge": 112666, "admitted_skel": 32509, "staged_skel": 46956, "mean_count": 203.24949645996094, "n_util_obs": 237, "util_count_spearman": -0.23082295401836533}, "probe_overhead_sec_total": 385.87, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.002747, "alpha": 0.03, "t": 3967.6}
{"event": "consolidate", "step": 4800, "retired": 1, "split_retired": 3, "merge_admitted": 0, "admitted": 4, "swapped": 5, "uncertain_staged": 262048, "t": 3999.6}
{"event": "probe", "step": 4800, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3472.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.199506, "cps": 27.125, "mean_p": 0.205951, "boundary_entropy": 0.12459, "bpb": 1.908054}, "tr": {"chunks": 3873.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.213942, "cps": 30.257812, "mean_p": 0.204158, "boundary_entropy": 0.11714, "bpb": 2.037108}, "fi": {"chunks": 4551.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.241933, "cps": 35.554688, "mean_p": 0.237076, "boundary_entropy": 0.108075, "bpb": 2.146554}}, "bpb": {"en": 1.90805, "tr": 2.03711, "fi": 2.14655}, "gini_cpc": 0.043158, "gini_cps": 0.060469, "bpic": 4.768998, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0723618351567872, "fi": 1.2126608469110753}, "inventory": {"size": 327671, "admitted": 65536, "staged": 262135, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 890, "admitted_surface": 25902, "staged_surface": 59002, "admitted_split": 5191, "staged_split": 43608, "admitted_merge": 1938, "staged_merge": 113107, "admitted_skel": 32505, "staged_skel": 46418, "mean_count": 218.410400390625, "n_util_obs": 244, "util_count_spearman": -0.2232929066217913}, "t": 4001.0}
{"event": "eval", "step": 4800, "phase": 3, "bpb": {"en": 1.9218, "tr": 2.04589, "fi": 2.15226}, "bytes_per_chunk": {"en": 5.102, "tr": 5.257, "fi": 4.546}, "bpb_mean": 2.03999, "frozen_inventory": true, "t": 4003.7}
{"event": "train", "step": 4800, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.37909, "bpb": 1.9896, "chunk_loss": 0.030381, "F": 0.19949, "G": 0.20463, "chunks": 204.28, "bytes_per_chunk": 5.013, "boundary_entropy": 0.12059, "router_noise": 0.0, "grad_norm": 0.6726, "lr": 5.85e-05, "router_trainable": true, "sec_per_step": 0.818, "probe_sec": 0.0769, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.0458, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 4, "mean_delta_loo": 0.03945, "n_shadow": 8, "mean_delta_shadow": -0.04639, "n_split": 1, "mean_delta_split": 0.04407, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 0, "read_hit_rate": {"surface": 0.30392, "split": 0.41903, "merge": 0.02225, "skel": 0.14297}, "mean_gate": {"surface": 0.03544, "split": 0.02577, "merge": 0.044, "skel": 0.03268}, "n_reads": 7539, "n_cand": 29267, "read_suppressed_frac": 0.1072, "dropout_reads_logged": 7539, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 889, "admitted_surface": 25902, "staged_surface": 59010, "admitted_split": 5191, "staged_split": 43606, "admitted_merge": 1938, "staged_merge": 113065, "admitted_skel": 32505, "staged_skel": 46463, "mean_count": 218.5254364013672, "n_util_obs": 244, "util_count_spearman": -0.22239321958291514}, "probe_overhead_sec_total": 389.121, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.002427, "alpha": 0.03, "t": 4008.5}
{"event": "consolidate", "step": 4850, "retired": 0, "split_retired": 2, "merge_admitted": 0, "admitted": 2, "swapped": 5, "uncertain_staged": 262049, "t": 4040.3}
{"event": "train", "step": 4850, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.35933, "bpb": 1.9611, "chunk_loss": 0.030341, "F": 0.19543, "G": 0.20335, "chunks": 200.12, "bytes_per_chunk": 5.117, "boundary_entropy": 0.11953, "router_noise": 0.0, "grad_norm": 0.7137, "lr": 5.62e-05, "router_trainable": true, "sec_per_step": 0.649, "probe_sec": 0.0695, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04085, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": 0.07411, "n_shadow": 3, "mean_delta_shadow": -0.0323, "n_split": 1, "mean_delta_split": -0.06903, "n_merge": 1, "mean_delta_merge": -0.31473, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": -0.39836, "trunc_abs_frac": 0.60366, "mean_delta_full": -0.22241, "read_hit_rate": {"surface": 0.28861, "split": 0.42263, "merge": 0.0254, "skel": 0.13904}, "mean_gate": {"surface": 0.03595, "split": 0.02725, "merge": 0.04318, "skel": 0.03356}, "n_reads": 7383, "n_cand": 28835, "read_suppressed_frac": 0.1054, "dropout_reads_logged": 7383, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 901, "admitted_surface": 25905, "staged_surface": 59140, "admitted_split": 5193, "staged_split": 43403, "admitted_merge": 1938, "staged_merge": 112856, "admitted_skel": 32500, "staged_skel": 46745, "mean_count": 223.13619995117188, "n_util_obs": 248, "util_count_spearman": -0.19962562794451563}, "probe_overhead_sec_total": 392.471, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.002483, "alpha": 0.03, "t": 4041.0}
{"event": "consolidate", "step": 4900, "retired": 1, "split_retired": 0, "merge_admitted": 0, "admitted": 1, "swapped": 2, "uncertain_staged": 262050, "t": 4072.2}
{"event": "train", "step": 4900, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.38625, "bpb": 1.99994, "chunk_loss": 0.030339, "F": 0.19662, "G": 0.2035, "chunks": 201.34, "bytes_per_chunk": 5.086, "boundary_entropy": 0.11276, "router_noise": 0.0, "grad_norm": 0.6924, "lr": 5.41e-05, "router_trainable": true, "sec_per_step": 0.641, "probe_sec": 0.0614, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04131, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 5, "mean_delta_loo": 0.05161, "n_shadow": 6, "mean_delta_shadow": 0.06174, "n_split": 1, "mean_delta_split": -0.04138, "n_merge": 1, "mean_delta_merge": 0.02942, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": -0.11936, "trunc_abs_frac": 0.78644, "mean_delta_full": -0.08923, "read_hit_rate": {"surface": 0.27578, "split": 0.42608, "merge": 0.01944, "skel": 0.13553}, "mean_gate": {"surface": 0.03629, "split": 0.02651, "merge": 0.04782, "skel": 0.03233}, "n_reads": 7325, "n_cand": 29038, "read_suppressed_frac": 0.1057, "dropout_reads_logged": 7325, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 915, "admitted_surface": 25907, "staged_surface": 59067, "admitted_split": 5193, "staged_split": 43381, "admitted_merge": 1938, "staged_merge": 113086, "admitted_skel": 32498, "staged_skel": 46610, "mean_count": 228.7913360595703, "n_util_obs": 253, "util_count_spearman": -0.18718689827319002}, "probe_overhead_sec_total": 395.791, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.002303, "alpha": 0.03, "t": 4073.0}
{"event": "consolidate", "step": 4950, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 7, "uncertain_staged": 262050, "t": 4102.6}
{"event": "train", "step": 4950, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.34664, "bpb": 1.9428, "chunk_loss": 0.030259, "F": 0.19064, "G": 0.19769, "chunks": 195.22, "bytes_per_chunk": 5.245, "boundary_entropy": 0.12004, "router_noise": 0.0, "grad_norm": 0.704, "lr": 5.2e-05, "router_trainable": true, "sec_per_step": 0.602, "probe_sec": 0.0609, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04054, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 3, "mean_delta_loo": -0.01206, "n_shadow": 5, "mean_delta_shadow": 0.03264, "n_split": 2, "mean_delta_split": -0.02431, "n_merge": 1, "mean_delta_merge": 0.04126, "n_dropped_unpaired": 0, "n_full": 1, "trunc_ratio_sum": -0.24805, "trunc_abs_frac": 0.24805, "mean_delta_full": 0.14075, "read_hit_rate": {"surface": 0.29735, "split": 0.42506, "merge": 0.02572, "skel": 0.14155}, "mean_gate": {"surface": 0.03705, "split": 0.0275, "merge": 0.04685, "skel": 0.03358}, "n_reads": 7336, "n_cand": 28210, "read_suppressed_frac": 0.1013, "dropout_reads_logged": 7336, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 927, "admitted_surface": 25913, "staged_surface": 58930, "admitted_split": 5194, "staged_split": 43559, "admitted_merge": 1938, "staged_merge": 112692, "admitted_skel": 32491, "staged_skel": 46963, "mean_count": 237.10302734375, "n_util_obs": 259, "util_count_spearman": -0.17437581146280012}, "probe_overhead_sec_total": 399.33, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.002599, "alpha": 0.03, "t": 4103.1}
{"event": "consolidate", "step": 5000, "retired": 0, "split_retired": 4, "merge_admitted": 0, "admitted": 4, "swapped": 0, "uncertain_staged": 262051, "t": 4137.9}
{"event": "probe", "step": 5000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3446.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.198012, "cps": 26.921875, "mean_p": 0.203335, "boundary_entropy": 0.125895, "bpb": 1.90135}, "tr": {"chunks": 3855.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.212948, "cps": 30.117188, "mean_p": 0.202105, "boundary_entropy": 0.117915, "bpb": 2.030812}, "fi": {"chunks": 4525.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.240551, "cps": 35.351562, "mean_p": 0.235642, "boundary_entropy": 0.107439, "bpb": 2.136391}}, "bpb": {"en": 1.90135, "tr": 2.03081, "fi": 2.13639}, "gini_cpc": 0.043528, "gini_cps": 0.060826, "bpic": 4.797226, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0754313142481362, "fi": 1.2148301090670297}, "inventory": {"size": 327676, "admitted": 65536, "staged": 262140, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 935, "admitted_surface": 25911, "staged_surface": 58753, "admitted_split": 5194, "staged_split": 43525, "admitted_merge": 1940, "staged_merge": 113002, "admitted_skel": 32491, "staged_skel": 46860, "mean_count": 241.33355712890625, "n_util_obs": 265, "util_count_spearman": -0.1409581439527743}, "t": 4139.3}
{"event": "eval", "step": 5000, "phase": 3, "bpb": {"en": 1.91597, "tr": 2.03909, "fi": 2.14192}, "bytes_per_chunk": {"en": 5.143, "tr": 5.289, "fi": 4.559}, "bpb_mean": 2.03233, "frozen_inventory": true, "t": 4141.7}
{"event": "train", "step": 5000, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.33322, "bpb": 1.92343, "chunk_loss": 0.030289, "F": 0.19708, "G": 0.20176, "chunks": 201.81, "bytes_per_chunk": 5.074, "boundary_entropy": 0.12309, "router_noise": 0.0, "grad_norm": 0.658, "lr": 5e-05, "router_trainable": true, "sec_per_step": 0.857, "probe_sec": 0.0643, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04256, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": -0.0232, "n_shadow": 1, "mean_delta_shadow": 0.10992, "n_split": 0, "mean_delta_split": NaN, "n_merge": 4, "mean_delta_merge": -0.07977, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": -0.03423, "trunc_abs_frac": 0.31558, "mean_delta_full": 0.2808, "read_hit_rate": {"surface": 0.29863, "split": 0.42477, "merge": 0.02565, "skel": 0.1422}, "mean_gate": {"surface": 0.03557, "split": 0.02613, "merge": 0.04356, "skel": 0.03352}, "n_reads": 7554, "n_cand": 29026, "read_suppressed_frac": 0.0957, "dropout_reads_logged": 7554, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 939, "admitted_surface": 25911, "staged_surface": 58798, "admitted_split": 5194, "staged_split": 43440, "admitted_merge": 1940, "staged_merge": 113069, "admitted_skel": 32491, "staged_skel": 46837, "mean_count": 241.44882202148438, "n_util_obs": 265, "util_count_spearman": -0.14141335328028476}, "probe_overhead_sec_total": 402.698, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.002332, "alpha": 0.03, "t": 4145.9}
{"event": "consolidate", "step": 5050, "retired": 0, "split_retired": 1, "merge_admitted": 0, "admitted": 1, "swapped": 5, "uncertain_staged": 262051, "t": 4177.9}
{"event": "train", "step": 5050, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.35582, "bpb": 1.95603, "chunk_loss": 0.03031, "F": 0.19653, "G": 0.20262, "chunks": 201.25, "bytes_per_chunk": 5.088, "boundary_entropy": 0.11964, "router_noise": 0.0, "grad_norm": 0.7069, "lr": 4.81e-05, "router_trainable": true, "sec_per_step": 0.647, "probe_sec": 0.0587, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04023, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 5, "mean_delta_loo": -0.07319, "n_shadow": 3, "mean_delta_shadow": 0.00073, "n_split": 3, "mean_delta_split": 0.04055, "n_merge": 1, "mean_delta_merge": -0.07831, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 0.51435, "trunc_abs_frac": 0.73942, "mean_delta_full": -0.13615, "read_hit_rate": {"surface": 0.30415, "split": 0.4253, "merge": 0.02321, "skel": 0.14328}, "mean_gate": {"surface": 0.03618, "split": 0.02652, "merge": 0.04436, "skel": 0.0349}, "n_reads": 7623, "n_cand": 29165, "read_suppressed_frac": 0.1019, "dropout_reads_logged": 7623, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 950, "admitted_surface": 25913, "staged_surface": 58881, "admitted_split": 5195, "staged_split": 43270, "admitted_merge": 1941, "staged_merge": 113084, "admitted_skel": 32487, "staged_skel": 46909, "mean_count": 248.02218627929688, "n_util_obs": 269, "util_count_spearman": -0.13397173580236152}, "probe_overhead_sec_total": 406.438, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.00263, "alpha": 0.03, "t": 4178.3}
{"event": "consolidate", "step": 5100, "retired": 0, "split_retired": 3, "merge_admitted": 0, "admitted": 3, "swapped": 2, "uncertain_staged": 262052, "t": 4207.3}
{"event": "train", "step": 5100, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.38109, "bpb": 1.9925, "chunk_loss": 0.030365, "F": 0.19916, "G": 0.20592, "chunks": 203.94, "bytes_per_chunk": 5.021, "boundary_entropy": 0.11647, "router_noise": 0.0, "grad_norm": 0.7274, "lr": 4.63e-05, "router_trainable": true, "sec_per_step": 0.596, "probe_sec": 0.0581, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.03824, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": -0.01065, "n_shadow": 3, "mean_delta_shadow": 0.01164, "n_split": 2, "mean_delta_split": 0.03619, "n_merge": 1, "mean_delta_merge": 0.02907, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 0.07962, "trunc_abs_frac": 0.13233, "mean_delta_full": 0.05245, "read_hit_rate": {"surface": 0.29319, "split": 0.41309, "merge": 0.02337, "skel": 0.13828}, "mean_gate": {"surface": 0.03593, "split": 0.02689, "merge": 0.04446, "skel": 0.03247}, "n_reads": 7342, "n_cand": 29079, "read_suppressed_frac": 0.0952, "dropout_reads_logged": 7342, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 945, "admitted_surface": 25912, "staged_surface": 58807, "admitted_split": 5196, "staged_split": 43399, "admitted_merge": 1943, "staged_merge": 113136, "admitted_skel": 32485, "staged_skel": 46802, "mean_count": 253.3964385986328, "n_util_obs": 276, "util_count_spearman": -0.12290463891782366}, "probe_overhead_sec_total": 409.702, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.002589, "alpha": 0.03, "t": 4208.1}
{"event": "consolidate", "step": 5150, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 6, "uncertain_staged": 262052, "t": 4238.8}
{"event": "train", "step": 5150, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.37489, "bpb": 1.98355, "chunk_loss": 0.030438, "F": 0.20407, "G": 0.2084, "chunks": 208.97, "bytes_per_chunk": 4.9, "boundary_entropy": 0.1182, "router_noise": 0.0, "grad_norm": 0.699, "lr": 4.45e-05, "router_trainable": true, "sec_per_step": 0.627, "probe_sec": 0.0636, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04009, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 5, "mean_delta_loo": -0.04317, "n_shadow": 4, "mean_delta_shadow": 0.00801, "n_split": 2, "mean_delta_split": -0.0343, "n_merge": 1, "mean_delta_merge": -0.04272, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 0.47557, "trunc_abs_frac": 0.27363, "mean_delta_full": -0.12381, "read_hit_rate": {"surface": 0.30864, "split": 0.42296, "merge": 0.02265, "skel": 0.14784}, "mean_gate": {"surface": 0.03603, "split": 0.02642, "merge": 0.0443, "skel": 0.03332}, "n_reads": 7758, "n_cand": 29709, "read_suppressed_frac": 0.104, "dropout_reads_logged": 7758, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 961, "admitted_surface": 25917, "staged_surface": 58872, "admitted_split": 5197, "staged_split": 43446, "admitted_merge": 1943, "staged_merge": 113082, "admitted_skel": 32479, "staged_skel": 46744, "mean_count": 259.3201904296875, "n_util_obs": 279, "util_count_spearman": -0.11909071820884307}, "probe_overhead_sec_total": 413.38, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.002466, "alpha": 0.03, "t": 4239.4}
{"event": "consolidate", "step": 5200, "retired": 0, "split_retired": 3, "merge_admitted": 0, "admitted": 1, "swapped": 0, "uncertain_staged": 262051, "t": 4273.5}
{"event": "probe", "step": 5200, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3440.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.197667, "cps": 26.875, "mean_p": 0.203347, "boundary_entropy": 0.124779, "bpb": 1.895934}, "tr": {"chunks": 3865.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.213501, "cps": 30.195312, "mean_p": 0.202142, "boundary_entropy": 0.116574, "bpb": 2.020981}, "fi": {"chunks": 4542.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.241454, "cps": 35.484375, "mean_p": 0.236481, "boundary_entropy": 0.1073, "bpb": 2.126609}}, "bpb": {"en": 1.89593, "tr": 2.02098, "fi": 2.12661}, "gini_cpc": 0.04473, "gini_cps": 0.062013, "bpic": 4.788723, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0801016374004566, "fi": 1.2215209618815315}, "inventory": {"size": 327677, "admitted": 65534, "staged": 262143, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 985, "admitted_surface": 25914, "staged_surface": 58872, "admitted_split": 5197, "staged_split": 43519, "admitted_merge": 1944, "staged_merge": 112938, "admitted_skel": 32479, "staged_skel": 46814, "mean_count": 263.90582275390625, "n_util_obs": 285, "util_count_spearman": -0.09846455785230245}, "t": 4274.8}
{"event": "eval", "step": 5200, "phase": 3, "bpb": {"en": 1.91042, "tr": 2.02956, "fi": 2.13209}, "bytes_per_chunk": {"en": 5.154, "tr": 5.28, "fi": 4.541}, "bpb_mean": 2.02402, "frozen_inventory": true, "t": 4277.0}
{"event": "train", "step": 5200, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.32839, "bpb": 1.91647, "chunk_loss": 0.030238, "F": 0.19244, "G": 0.19808, "chunks": 197.06, "bytes_per_chunk": 5.196, "boundary_entropy": 0.11478, "router_noise": 0.0, "grad_norm": 0.6749, "lr": 4.29e-05, "router_trainable": true, "sec_per_step": 0.842, "probe_sec": 0.0609, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.03782, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": 0.01226, "n_shadow": 6, "mean_delta_shadow": -0.00099, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": 0.22741, "trunc_abs_frac": 0.40561, "mean_delta_full": 0.21809, "read_hit_rate": {"surface": 0.28275, "split": 0.41202, "merge": 0.02355, "skel": 0.14468}, "mean_gate": {"surface": 0.03321, "split": 0.02445, "merge": 0.04241, "skel": 0.03205}, "n_reads": 7293, "n_cand": 28918, "read_suppressed_frac": 0.0941, "dropout_reads_logged": 7293, "inventory": {"size": 327678, "admitted": 65534, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 982, "admitted_surface": 25914, "staged_surface": 58892, "admitted_split": 5197, "staged_split": 43500, "admitted_merge": 1944, "staged_merge": 113002, "admitted_skel": 32479, "staged_skel": 46750, "mean_count": 264.01715087890625, "n_util_obs": 285, "util_count_spearman": -0.09984863172409503}, "probe_overhead_sec_total": 416.701, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.002572, "alpha": 0.03, "t": 4281.6}
{"event": "consolidate", "step": 5250, "retired": 0, "split_retired": 3, "merge_admitted": 0, "admitted": 5, "swapped": 0, "uncertain_staged": 262051, "t": 4313.9}
{"event": "train", "step": 5250, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.32793, "bpb": 1.9158, "chunk_loss": 0.030299, "F": 0.19589, "G": 0.20075, "chunks": 200.59, "bytes_per_chunk": 5.105, "boundary_entropy": 0.11446, "router_noise": 0.0, "grad_norm": 0.6606, "lr": 4.14e-05, "router_trainable": true, "sec_per_step": 0.654, "probe_sec": 0.0603, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04047, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": -0.08774, "n_shadow": 4, "mean_delta_shadow": -0.03275, "n_split": 1, "mean_delta_split": 0.03412, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 0, "read_hit_rate": {"surface": 0.30108, "split": 0.43443, "merge": 0.02581, "skel": 0.15939}, "mean_gate": {"surface": 0.03485, "split": 0.02582, "merge": 0.04333, "skel": 0.03273}, "n_reads": 7797, "n_cand": 29119, "read_suppressed_frac": 0.1006, "dropout_reads_logged": 7797, "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": 25914, "staged_surface": 58892, "admitted_split": 5199, "staged_split": 43581, "admitted_merge": 1944, "staged_merge": 112817, "admitted_skel": 32479, "staged_skel": 46854, "mean_count": 268.72943115234375, "n_util_obs": 289, "util_count_spearman": -0.0953978244441793}, "probe_overhead_sec_total": 419.935, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.002382, "alpha": 0.03, "t": 4314.3}
{"event": "consolidate", "step": 5300, "retired": 1, "split_retired": 0, "merge_admitted": 0, "admitted": 1, "swapped": 1, "uncertain_staged": 262051, "t": 4344.9}
{"event": "train", "step": 5300, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.38123, "bpb": 1.99269, "chunk_loss": 0.030412, "F": 0.20261, "G": 0.20747, "chunks": 207.47, "bytes_per_chunk": 4.936, "boundary_entropy": 0.1151, "router_noise": 0.0, "grad_norm": 0.7215, "lr": 3.99e-05, "router_trainable": true, "sec_per_step": 0.629, "probe_sec": 0.0584, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.03907, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": -0.04241, "n_shadow": 4, "mean_delta_shadow": -0.02097, "n_split": 1, "mean_delta_split": 0.09137, "n_merge": 1, "mean_delta_merge": -0.06879, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 0.04953, "trunc_abs_frac": 0.54004, "mean_delta_full": 0.14694, "read_hit_rate": {"surface": 0.28001, "split": 0.41628, "merge": 0.01977, "skel": 0.14294}, "mean_gate": {"surface": 0.0343, "split": 0.02532, "merge": 0.04402, "skel": 0.03232}, "n_reads": 7612, "n_cand": 30244, "read_suppressed_frac": 0.1014, "dropout_reads_logged": 7612, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 987, "admitted_surface": 25915, "staged_surface": 58735, "admitted_split": 5199, "staged_split": 43264, "admitted_merge": 1944, "staged_merge": 113476, "admitted_skel": 32478, "staged_skel": 46669, "mean_count": 274.42596435546875, "n_util_obs": 301, "util_count_spearman": -0.09937031088424897}, "probe_overhead_sec_total": 423.1, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.002531, "alpha": 0.03, "t": 4345.7}
{"event": "consolidate", "step": 5350, "retired": 0, "split_retired": 2, "merge_admitted": 0, "admitted": 2, "swapped": 5, "uncertain_staged": 262049, "t": 4375.8}
{"event": "train", "step": 5350, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.31485, "bpb": 1.89692, "chunk_loss": 0.030252, "F": 0.19156, "G": 0.1967, "chunks": 196.16, "bytes_per_chunk": 5.22, "boundary_entropy": 0.11437, "router_noise": 0.0, "grad_norm": 0.7215, "lr": 3.86e-05, "router_trainable": true, "sec_per_step": 0.615, "probe_sec": 0.0663, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.03973, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 3, "mean_delta_loo": -0.0237, "n_shadow": 7, "mean_delta_shadow": 0.06438, "n_split": 1, "mean_delta_split": 0.07867, "n_merge": 2, "mean_delta_merge": -0.02765, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": -0.1676, "trunc_abs_frac": 0.26345, "mean_delta_full": -0.1525, "read_hit_rate": {"surface": 0.30088, "split": 0.41589, "merge": 0.02575, "skel": 0.14783}, "mean_gate": {"surface": 0.03563, "split": 0.02549, "merge": 0.04335, "skel": 0.03452}, "n_reads": 7366, "n_cand": 28528, "read_suppressed_frac": 0.102, "dropout_reads_logged": 7366, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 973, "admitted_surface": 25916, "staged_surface": 58833, "admitted_split": 5202, "staged_split": 43478, "admitted_merge": 1944, "staged_merge": 113075, "admitted_skel": 32474, "staged_skel": 46758, "mean_count": 279.32122802734375, "n_util_obs": 305, "util_count_spearman": -0.07629717978649614}, "probe_overhead_sec_total": 426.313, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.002513, "alpha": 0.03, "t": 4376.5}
{"event": "consolidate", "step": 5400, "retired": 2, "split_retired": 2, "merge_admitted": 0, "admitted": 3, "swapped": 0, "uncertain_staged": 262050, "t": 4406.7}
{"event": "probe", "step": 5400, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3451.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.198299, "cps": 26.960938, "mean_p": 0.203534, "boundary_entropy": 0.121513, "bpb": 1.890005}, "tr": {"chunks": 3824.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.211236, "cps": 29.875, "mean_p": 0.200764, "boundary_entropy": 0.120015, "bpb": 2.015866}, "fi": {"chunks": 4547.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.24172, "cps": 35.523438, "mean_p": 0.236171, "boundary_entropy": 0.107238, "bpb": 2.119432}}, "bpb": {"en": 1.89, "tr": 2.01587, "fi": 2.11943}, "gini_cpc": 0.044449, "gini_cps": 0.061806, "bpic": 4.79885, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0652376138069397, "fi": 1.2189677947733715}, "inventory": {"size": 327676, "admitted": 65535, "staged": 262141, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 997, "admitted_surface": 25916, "staged_surface": 58742, "admitted_split": 5201, "staged_split": 43539, "admitted_merge": 1944, "staged_merge": 112840, "admitted_skel": 32474, "staged_skel": 47020, "mean_count": 284.30133056640625, "n_util_obs": 316, "util_count_spearman": -0.08988008148651803}, "t": 4408.0}
{"event": "eval", "step": 5400, "phase": 3, "bpb": {"en": 1.90453, "tr": 2.02416, "fi": 2.12663}, "bytes_per_chunk": {"en": 5.138, "tr": 5.318, "fi": 4.551}, "bpb_mean": 2.01844, "frozen_inventory": true, "t": 4410.2}
{"event": "train", "step": 5400, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.38358, "bpb": 1.99608, "chunk_loss": 0.030518, "F": 0.20483, "G": 0.20973, "chunks": 209.75, "bytes_per_chunk": 4.882, "boundary_entropy": 0.11519, "router_noise": 0.0, "grad_norm": 0.7331, "lr": 3.73e-05, "router_trainable": true, "sec_per_step": 0.755, "probe_sec": 0.0603, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04065, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": -0.02705, "n_shadow": 3, "mean_delta_shadow": -0.03369, "n_split": 0, "mean_delta_split": NaN, "n_merge": 1, "mean_delta_merge": 0.06451, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": -0.12815, "trunc_abs_frac": 0.24993, "mean_delta_full": -0.20575, "read_hit_rate": {"surface": 0.3015, "split": 0.4242, "merge": 0.02151, "skel": 0.14383}, "mean_gate": {"surface": 0.03551, "split": 0.02516, "merge": 0.04476, "skel": 0.03164}, "n_reads": 7843, "n_cand": 30229, "read_suppressed_frac": 0.1026, "dropout_reads_logged": 7843, "inventory": {"size": 327679, "admitted": 65535, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 998, "admitted_surface": 25916, "staged_surface": 58769, "admitted_split": 5201, "staged_split": 43512, "admitted_merge": 1944, "staged_merge": 112851, "admitted_skel": 32474, "staged_skel": 47012, "mean_count": 284.4210205078125, "n_util_obs": 316, "util_count_spearman": -0.09001337604210095}, "probe_overhead_sec_total": 429.515, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.001946, "alpha": 0.03, "t": 4414.2}
{"event": "consolidate", "step": 5450, "retired": 1, "split_retired": 2, "merge_admitted": 0, "admitted": 3, "swapped": 0, "uncertain_staged": 262048, "t": 4447.0}
{"event": "train", "step": 5450, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.32616, "bpb": 1.91324, "chunk_loss": 0.030419, "F": 0.2009, "G": 0.20559, "chunks": 205.72, "bytes_per_chunk": 4.978, "boundary_entropy": 0.11506, "router_noise": 0.0, "grad_norm": 0.7259, "lr": 3.62e-05, "router_trainable": true, "sec_per_step": 0.665, "probe_sec": 0.0838, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.05039, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 4, "mean_delta_loo": -0.02221, "n_shadow": 6, "mean_delta_shadow": 0.04005, "n_split": 2, "mean_delta_split": -0.03334, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": -0.19447, "trunc_abs_frac": 2.86994, "mean_delta_full": 0.16919, "read_hit_rate": {"surface": 0.29522, "split": 0.44407, "merge": 0.02792, "skel": 0.143}, "mean_gate": {"surface": 0.03729, "split": 0.02677, "merge": 0.04629, "skel": 0.03532}, "n_reads": 7888, "n_cand": 29529, "read_suppressed_frac": 0.0956, "dropout_reads_logged": 7888, "inventory": {"size": 327679, "admitted": 65535, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1021, "admitted_surface": 25913, "staged_surface": 58845, "admitted_split": 5202, "staged_split": 43204, "admitted_merge": 1945, "staged_merge": 113194, "admitted_skel": 32475, "staged_skel": 46901, "mean_count": 287.8359375, "n_util_obs": 320, "util_count_spearman": -0.10314981103331088}, "probe_overhead_sec_total": 432.945, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.00246, "alpha": 0.03, "t": 4447.4}
{"event": "consolidate", "step": 5500, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 1, "swapped": 10, "uncertain_staged": 262049, "t": 4479.2}
{"event": "train", "step": 5500, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.31006, "bpb": 1.89001, "chunk_loss": 0.030378, "F": 0.20007, "G": 0.20556, "chunks": 204.88, "bytes_per_chunk": 4.998, "boundary_entropy": 0.11703, "router_noise": 0.0, "grad_norm": 0.6861, "lr": 3.51e-05, "router_trainable": true, "sec_per_step": 0.642, "probe_sec": 0.06, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.03996, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 10, "mean_delta_loo": 0.00134, "n_shadow": 3, "mean_delta_shadow": -0.01162, "n_split": 1, "mean_delta_split": 0.01038, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 0.74749, "trunc_abs_frac": 2.44946, "mean_delta_full": 0.12577, "read_hit_rate": {"surface": 0.31821, "split": 0.43704, "merge": 0.02495, "skel": 0.1451}, "mean_gate": {"surface": 0.03602, "split": 0.02582, "merge": 0.04116, "skel": 0.03297}, "n_reads": 7948, "n_cand": 29520, "read_suppressed_frac": 0.0985, "dropout_reads_logged": 7948, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1038, "admitted_surface": 25923, "staged_surface": 58815, "admitted_split": 5203, "staged_split": 43183, "admitted_merge": 1945, "staged_merge": 113488, "admitted_skel": 32465, "staged_skel": 46658, "mean_count": 293.8147888183594, "n_util_obs": 325, "util_count_spearman": -0.10567460291797531}, "probe_overhead_sec_total": 436.671, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.002242, "alpha": 0.03, "t": 4479.6}
{"event": "consolidate", "step": 5550, "retired": 0, "split_retired": 3, "merge_admitted": 0, "admitted": 3, "swapped": 2, "uncertain_staged": 262049, "t": 4509.3}
{"event": "train", "step": 5550, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.36179, "bpb": 1.96465, "chunk_loss": 0.030359, "F": 0.19965, "G": 0.2041, "chunks": 204.44, "bytes_per_chunk": 5.009, "boundary_entropy": 0.11481, "router_noise": 0.0, "grad_norm": 0.7455, "lr": 3.41e-05, "router_trainable": true, "sec_per_step": 0.604, "probe_sec": 0.0674, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04419, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 5, "mean_delta_loo": 0.04553, "n_shadow": 7, "mean_delta_shadow": 0.02283, "n_split": 1, "mean_delta_split": -0.01648, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 0.73387, "trunc_abs_frac": 0.21439, "mean_delta_full": 0.0596, "read_hit_rate": {"surface": 0.31244, "split": 0.4351, "merge": 0.02701, "skel": 0.1522}, "mean_gate": {"surface": 0.03561, "split": 0.02695, "merge": 0.04326, "skel": 0.03429}, "n_reads": 7856, "n_cand": 29239, "read_suppressed_frac": 0.0953, "dropout_reads_logged": 7856, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1026, "admitted_surface": 25922, "staged_surface": 58859, "admitted_split": 5204, "staged_split": 43207, "admitted_merge": 1946, "staged_merge": 113304, "admitted_skel": 32464, "staged_skel": 46774, "mean_count": 300.55560302734375, "n_util_obs": 330, "util_count_spearman": -0.10320607388599153}, "probe_overhead_sec_total": 440.07, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.002672, "alpha": 0.03, "t": 4509.7}
{"event": "consolidate", "step": 5600, "retired": 1, "split_retired": 2, "merge_admitted": 0, "admitted": 3, "swapped": 5, "uncertain_staged": 262051, "t": 4544.0}
{"event": "probe", "step": 5600, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3467.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.199219, "cps": 27.085938, "mean_p": 0.204599, "boundary_entropy": 0.120859, "bpb": 1.885507}, "tr": {"chunks": 3864.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.213445, "cps": 30.1875, "mean_p": 0.202412, "boundary_entropy": 0.114982, "bpb": 2.008997}, "fi": {"chunks": 4526.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.240604, "cps": 35.359375, "mean_p": 0.235782, "boundary_entropy": 0.104531, "bpb": 2.113298}}, "bpb": {"en": 1.88551, "tr": 2.009, "fi": 2.1133}, "gini_cpc": 0.042234, "gini_cps": 0.059543, "bpic": 4.784684, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0714128353855557, "fi": 1.2077385941802918}, "inventory": {"size": 327672, "admitted": 65536, "staged": 262136, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1042, "admitted_surface": 25924, "staged_surface": 58887, "admitted_split": 5206, "staged_split": 43477, "admitted_merge": 1947, "staged_merge": 113007, "admitted_skel": 32459, "staged_skel": 46765, "mean_count": 306.23876953125, "n_util_obs": 335, "util_count_spearman": -0.10846944973681648}, "t": 4545.3}
{"event": "eval", "step": 5600, "phase": 3, "bpb": {"en": 1.90055, "tr": 2.01706, "fi": 2.11943}, "bytes_per_chunk": {"en": 5.132, "tr": 5.289, "fi": 4.563}, "bpb_mean": 2.01234, "frozen_inventory": true, "t": 4547.5}
{"event": "train", "step": 5600, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.35102, "bpb": 1.94911, "chunk_loss": 0.03033, "F": 0.1969, "G": 0.20121, "chunks": 201.62, "bytes_per_chunk": 5.079, "boundary_entropy": 0.11158, "router_noise": 0.0, "grad_norm": 0.7645, "lr": 3.33e-05, "router_trainable": true, "sec_per_step": 0.856, "probe_sec": 0.0856, "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": 3, "mean_delta_loo": 0.03574, "n_shadow": 5, "mean_delta_shadow": -0.05653, "n_split": 2, "mean_delta_split": -0.00613, "n_merge": 2, "mean_delta_merge": -0.11765, "n_dropped_unpaired": 0, "n_full": 0, "read_hit_rate": {"surface": 0.2866, "split": 0.41524, "merge": 0.02066, "skel": 0.14685}, "mean_gate": {"surface": 0.0347, "split": 0.02528, "merge": 0.04443, "skel": 0.03252}, "n_reads": 7463, "n_cand": 29434, "read_suppressed_frac": 0.1052, "dropout_reads_logged": 7463, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1043, "admitted_surface": 25924, "staged_surface": 58882, "admitted_split": 5206, "staged_split": 43495, "admitted_merge": 1947, "staged_merge": 113063, "admitted_skel": 32459, "staged_skel": 46704, "mean_count": 306.3526306152344, "n_util_obs": 335, "util_count_spearman": -0.10853151509319162}, "probe_overhead_sec_total": 443.357, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.002452, "alpha": 0.03, "t": 4552.5}
{"event": "consolidate", "step": 5650, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 7, "uncertain_staged": 262052, "t": 4586.2}
{"event": "train", "step": 5650, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.33509, "bpb": 1.92613, "chunk_loss": 0.030414, "F": 0.2019, "G": 0.20671, "chunks": 206.75, "bytes_per_chunk": 4.953, "boundary_entropy": 0.11469, "router_noise": 0.0, "grad_norm": 0.6993, "lr": 3.25e-05, "router_trainable": true, "sec_per_step": 0.681, "probe_sec": 0.0604, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.03962, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": -0.00149, "n_shadow": 4, "mean_delta_shadow": 0.15413, "n_split": 0, "mean_delta_split": NaN, "n_merge": 1, "mean_delta_merge": -0.01721, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 0.48742, "trunc_abs_frac": 1.22176, "mean_delta_full": 0.05176, "read_hit_rate": {"surface": 0.30969, "split": 0.43513, "merge": 0.02579, "skel": 0.14987}, "mean_gate": {"surface": 0.03388, "split": 0.02455, "merge": 0.04276, "skel": 0.03241}, "n_reads": 8044, "n_cand": 29951, "read_suppressed_frac": 0.0929, "dropout_reads_logged": 8044, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1041, "admitted_surface": 25928, "staged_surface": 59043, "admitted_split": 5208, "staged_split": 43204, "admitted_merge": 1948, "staged_merge": 113058, "admitted_skel": 32452, "staged_skel": 46839, "mean_count": 314.5545654296875, "n_util_obs": 349, "util_count_spearman": -0.08961241684368049}, "probe_overhead_sec_total": 447.268, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.002183, "alpha": 0.03, "t": 4586.6}
{"event": "consolidate", "step": 5700, "retired": 0, "split_retired": 2, "merge_admitted": 0, "admitted": 2, "swapped": 7, "uncertain_staged": 262054, "t": 4617.3}
{"event": "train", "step": 5700, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.35072, "bpb": 1.94868, "chunk_loss": 0.030236, "F": 0.19238, "G": 0.19797, "chunks": 197.0, "bytes_per_chunk": 5.198, "boundary_entropy": 0.11416, "router_noise": 0.0, "grad_norm": 0.7779, "lr": 3.18e-05, "router_trainable": true, "sec_per_step": 0.631, "probe_sec": 0.0597, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.03954, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": 0.0208, "n_shadow": 3, "mean_delta_shadow": 0.07867, "n_split": 2, "mean_delta_split": -0.06042, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 6, "trunc_ratio_sum": 0.23973, "trunc_abs_frac": 0.65344, "mean_delta_full": 0.08468, "read_hit_rate": {"surface": 0.29815, "split": 0.42258, "merge": 0.03093, "skel": 0.14809}, "mean_gate": {"surface": 0.03465, "split": 0.02536, "merge": 0.0446, "skel": 0.03319}, "n_reads": 7441, "n_cand": 28502, "read_suppressed_frac": 0.0949, "dropout_reads_logged": 7441, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1060, "admitted_surface": 25933, "staged_surface": 58866, "admitted_split": 5209, "staged_split": 43465, "admitted_merge": 1948, "staged_merge": 113034, "admitted_skel": 32446, "staged_skel": 46779, "mean_count": 320.1212158203125, "n_util_obs": 351, "util_count_spearman": -0.09718226892428906}, "probe_overhead_sec_total": 450.45, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.002407, "alpha": 0.03, "t": 4618.1}
{"event": "consolidate", "step": 5750, "retired": 0, "split_retired": 2, "merge_admitted": 0, "admitted": 2, "swapped": 3, "uncertain_staged": 262054, "t": 4647.3}
{"event": "train", "step": 5750, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.40189, "bpb": 2.0225, "chunk_loss": 0.030417, "F": 0.20184, "G": 0.2072, "chunks": 206.69, "bytes_per_chunk": 4.954, "boundary_entropy": 0.11096, "router_noise": 0.0, "grad_norm": 0.7883, "lr": 3.13e-05, "router_trainable": true, "sec_per_step": 0.592, "probe_sec": 0.0647, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04391, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": -0.0307, "n_shadow": 6, "mean_delta_shadow": -0.02247, "n_split": 1, "mean_delta_split": -0.11902, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 0.03426, "trunc_abs_frac": 0.13685, "mean_delta_full": -0.3112, "read_hit_rate": {"surface": 0.29049, "split": 0.43436, "merge": 0.02321, "skel": 0.14528}, "mean_gate": {"surface": 0.03528, "split": 0.02526, "merge": 0.04205, "skel": 0.03136}, "n_reads": 7814, "n_cand": 29824, "read_suppressed_frac": 0.0951, "dropout_reads_logged": 7814, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1061, "admitted_surface": 25933, "staged_surface": 59046, "admitted_split": 5211, "staged_split": 43479, "admitted_merge": 1948, "staged_merge": 112904, "admitted_skel": 32444, "staged_skel": 46715, "mean_count": 326.748291015625, "n_util_obs": 355, "util_count_spearman": -0.09788639828314458}, "probe_overhead_sec_total": 454.576, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.002563, "alpha": 0.03, "t": 4647.7}
{"event": "consolidate", "step": 5800, "retired": 0, "split_retired": 3, "merge_admitted": 0, "admitted": 3, "swapped": 23, "uncertain_staged": 262055, "t": 4678.7}
{"event": "probe", "step": 5800, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3446.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.198012, "cps": 26.921875, "mean_p": 0.204081, "boundary_entropy": 0.121971, "bpb": 1.881986}, "tr": {"chunks": 3842.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.21223, "cps": 30.015625, "mean_p": 0.200476, "boundary_entropy": 0.117061, "bpb": 2.006064}, "fi": {"chunks": 4551.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.241933, "cps": 35.554688, "mean_p": 0.236133, "boundary_entropy": 0.103385, "bpb": 2.108956}}, "bpb": {"en": 1.88199, "tr": 2.00606, "fi": 2.10896}, "gini_cpc": 0.044897, "gini_cps": 0.062224, "bpic": 4.791959, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0718046976242126, "fi": 1.2218103483677463}, "inventory": {"size": 327654, "admitted": 65536, "staged": 262118, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1065, "admitted_surface": 25945, "staged_surface": 58842, "admitted_split": 5214, "staged_split": 43252, "admitted_merge": 1950, "staged_merge": 113404, "admitted_skel": 32427, "staged_skel": 46620, "mean_count": 332.1683349609375, "n_util_obs": 364, "util_count_spearman": -0.0969400643453819}, "t": 4680.1}
{"event": "eval", "step": 5800, "phase": 3, "bpb": {"en": 1.89607, "tr": 2.01326, "fi": 2.11544}, "bytes_per_chunk": {"en": 5.153, "tr": 5.319, "fi": 4.548}, "bpb_mean": 2.00825, "frozen_inventory": true, "t": 4682.3}
{"event": "train", "step": 5800, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.32696, "bpb": 1.9144, "chunk_loss": 0.030397, "F": 0.19586, "G": 0.20171, "chunks": 200.56, "bytes_per_chunk": 5.106, "boundary_entropy": 0.1174, "router_noise": 0.0, "grad_norm": 0.738, "lr": 3.08e-05, "router_trainable": true, "sec_per_step": 0.771, "probe_sec": 0.0752, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04338, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 5, "mean_delta_loo": -0.03864, "n_shadow": 7, "mean_delta_shadow": -0.00723, "n_split": 1, "mean_delta_split": -0.02153, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": 0.16092, "trunc_abs_frac": 0.35179, "mean_delta_full": -0.16025, "read_hit_rate": {"surface": 0.30457, "split": 0.43873, "merge": 0.02361, "skel": 0.15959}, "mean_gate": {"surface": 0.03712, "split": 0.0267, "merge": 0.05086, "skel": 0.03325}, "n_reads": 7790, "n_cand": 28926, "read_suppressed_frac": 0.1001, "dropout_reads_logged": 7790, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1066, "admitted_surface": 25945, "staged_surface": 58844, "admitted_split": 5214, "staged_split": 43217, "admitted_merge": 1950, "staged_merge": 113465, "admitted_skel": 32427, "staged_skel": 46618, "mean_count": 332.2872314453125, "n_util_obs": 365, "util_count_spearman": -0.09695367062463377}, "probe_overhead_sec_total": 458.572, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.00223, "alpha": 0.03, "t": 4686.3}
{"event": "consolidate", "step": 5850, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 4, "uncertain_staged": 262055, "t": 4720.7}
{"event": "train", "step": 5850, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.34223, "bpb": 1.93642, "chunk_loss": 0.030459, "F": 0.20255, "G": 0.208, "chunks": 207.41, "bytes_per_chunk": 4.937, "boundary_entropy": 0.1132, "router_noise": 0.0, "grad_norm": 0.7325, "lr": 3.05e-05, "router_trainable": true, "sec_per_step": 0.697, "probe_sec": 0.0561, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.03784, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 4, "mean_delta_loo": 0.05782, "n_shadow": 6, "mean_delta_shadow": 0.01144, "n_split": 1, "mean_delta_split": 0.05811, "n_merge": 1, "mean_delta_merge": -0.09888, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 0.47879, "trunc_abs_frac": 1.29427, "mean_delta_full": 0.08344, "read_hit_rate": {"surface": 0.29947, "split": 0.41723, "merge": 0.02297, "skel": 0.15412}, "mean_gate": {"surface": 0.03362, "split": 0.02361, "merge": 0.04293, "skel": 0.03159}, "n_reads": 7806, "n_cand": 30152, "read_suppressed_frac": 0.1006, "dropout_reads_logged": 7806, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1069, "admitted_surface": 25948, "staged_surface": 58795, "admitted_split": 5214, "staged_split": 43400, "admitted_merge": 1951, "staged_merge": 113263, "admitted_skel": 32423, "staged_skel": 46686, "mean_count": 338.263427734375, "n_util_obs": 377, "util_count_spearman": -0.09909070432496857}, "probe_overhead_sec_total": 462.09, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.002139, "alpha": 0.03, "t": 4721.1}
{"event": "consolidate", "step": 5900, "retired": 0, "split_retired": 2, "merge_admitted": 0, "admitted": 2, "swapped": 9, "uncertain_staged": 262056, "t": 4751.1}
{"event": "train", "step": 5900, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.28033, "bpb": 1.84712, "chunk_loss": 0.03027, "F": 0.19513, "G": 0.19909, "chunks": 199.81, "bytes_per_chunk": 5.125, "boundary_entropy": 0.11027, "router_noise": 0.0, "grad_norm": 0.7081, "lr": 3.02e-05, "router_trainable": true, "sec_per_step": 0.616, "probe_sec": 0.068, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04277, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": 0.01494, "n_shadow": 5, "mean_delta_shadow": 0.01481, "n_split": 2, "mean_delta_split": 0.02657, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 0.34387, "trunc_abs_frac": 0.69814, "mean_delta_full": 0.19853, "read_hit_rate": {"surface": 0.31217, "split": 0.43577, "merge": 0.02812, "skel": 0.15717}, "mean_gate": {"surface": 0.03532, "split": 0.02498, "merge": 0.04383, "skel": 0.03246}, "n_reads": 7877, "n_cand": 29089, "read_suppressed_frac": 0.0979, "dropout_reads_logged": 7877, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1082, "admitted_surface": 25952, "staged_surface": 58929, "admitted_split": 5218, "staged_split": 43243, "admitted_merge": 1952, "staged_merge": 113341, "admitted_skel": 32414, "staged_skel": 46631, "mean_count": 343.8402099609375, "n_util_obs": 384, "util_count_spearman": -0.09386964390297557}, "probe_overhead_sec_total": 466.079, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.00207, "alpha": 0.03, "t": 4752.0}
{"event": "consolidate", "step": 5950, "retired": 0, "split_retired": 1, "merge_admitted": 0, "admitted": 1, "swapped": 3, "uncertain_staged": 262056, "t": 4781.4}
{"event": "train", "step": 5950, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.38455, "bpb": 1.99748, "chunk_loss": 0.030299, "F": 0.19644, "G": 0.20176, "chunks": 201.16, "bytes_per_chunk": 5.091, "boundary_entropy": 0.11252, "router_noise": 0.0, "grad_norm": 0.7475, "lr": 3.01e-05, "router_trainable": true, "sec_per_step": 0.598, "probe_sec": 0.061, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.03914, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 4, "mean_delta_loo": -0.03049, "n_shadow": 6, "mean_delta_shadow": -0.0004, "n_split": 2, "mean_delta_split": 0.10897, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 1.16088, "trunc_abs_frac": 1.25088, "mean_delta_full": 0.01758, "read_hit_rate": {"surface": 0.28244, "split": 0.42022, "merge": 0.02573, "skel": 0.15768}, "mean_gate": {"surface": 0.03485, "split": 0.02522, "merge": 0.04377, "skel": 0.03237}, "n_reads": 7580, "n_cand": 29366, "read_suppressed_frac": 0.0999, "dropout_reads_logged": 7580, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1082, "admitted_surface": 25951, "staged_surface": 58776, "admitted_split": 5219, "staged_split": 43242, "admitted_merge": 1953, "staged_merge": 113449, "admitted_skel": 32413, "staged_skel": 46677, "mean_count": 348.81219482421875, "n_util_obs": 390, "util_count_spearman": -0.08836240532859707}, "probe_overhead_sec_total": 469.672, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.002366, "alpha": 0.03, "t": 4781.8}
{"event": "consolidate", "step": 6000, "retired": 0, "split_retired": 2, "merge_admitted": 0, "admitted": 2, "swapped": 12, "uncertain_staged": 262056, "t": 4812.9}
{"event": "train", "step": 5999, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.38302, "bpb": 1.99528, "chunk_loss": 0.030422, "F": 0.19934, "G": 0.20432, "chunks": 204.12, "bytes_per_chunk": 5.017, "boundary_entropy": 0.11354, "router_noise": 0.0, "grad_norm": 0.7095, "lr": 3e-05, "router_trainable": true, "sec_per_step": 0.621, "probe_sec": 0.057, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.03819, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 4, "mean_delta_loo": 0.05977, "n_shadow": 7, "mean_delta_shadow": 0.02373, "n_split": 0, "mean_delta_split": NaN, "n_merge": 2, "mean_delta_merge": 0.02524, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": -0.50135, "trunc_abs_frac": 0.49996, "mean_delta_full": -0.18048, "read_hit_rate": {"surface": 0.31615, "split": 0.43322, "merge": 0.02597, "skel": 0.15646}, "mean_gate": {"surface": 0.03636, "split": 0.02487, "merge": 0.04268, "skel": 0.03255}, "n_reads": 7817, "n_cand": 29094, "read_suppressed_frac": 0.102, "dropout_reads_logged": 7817, "inventory": {"size": 327666, "admitted": 65536, "staged": 262130, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1115, "admitted_surface": 25955, "staged_surface": 58566, "admitted_split": 5226, "staged_split": 43400, "admitted_merge": 1953, "staged_merge": 113391, "admitted_skel": 32402, "staged_skel": 46773, "mean_count": 354.78802490234375, "n_util_obs": 400, "util_count_spearman": -0.09193389570565129}, "probe_overhead_sec_total": 472.752, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.002465, "alpha": 0.03, "t": 4812.9}
{"event": "probe", "step": 6000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3462.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.198931, "cps": 27.046875, "mean_p": 0.204378, "boundary_entropy": 0.119099, "bpb": 1.87901}, "tr": {"chunks": 3852.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.212782, "cps": 30.09375, "mean_p": 0.201744, "boundary_entropy": 0.109752, "bpb": 2.000711}, "fi": {"chunks": 4554.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.242092, "cps": 35.578125, "mean_p": 0.236655, "boundary_entropy": 0.10239, "bpb": 2.104757}}, "bpb": {"en": 1.87901, "tr": 2.00071, "fi": 2.10476}, "gini_cpc": 0.04401, "gini_cps": 0.061341, "bpic": 4.780249, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0696280507716724, "fi": 1.2169653122500046}, "inventory": {"size": 327666, "admitted": 65536, "staged": 262130, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1115, "admitted_surface": 25955, "staged_surface": 58566, "admitted_split": 5226, "staged_split": 43400, "admitted_merge": 1953, "staged_merge": 113391, "admitted_skel": 32402, "staged_skel": 46773, "mean_count": 354.78802490234375, "n_util_obs": 400, "util_count_spearman": -0.09193389570565129}, "t": 4814.2}
{"event": "eval", "step": 6000, "phase": 3, "bpb": {"en": 1.89362, "tr": 2.00802, "fi": 2.11201}, "bytes_per_chunk": {"en": 5.114, "tr": 5.269, "fi": 4.54}, "bpb_mean": 2.00455, "frozen_inventory": true, "t": 4816.6}
{"event": "reps_dumped", "dirs": ["/local/scratch/sas245/beetle-hnet/runs/ali_ali_s2/probe/reps/th0.30", "/local/scratch/sas245/beetle-hnet/runs/ali_ali_s2/probe/reps/th0.40", "/local/scratch/sas245/beetle-hnet/runs/ali_ali_s2/probe/reps/th0.50", "/local/scratch/sas245/beetle-hnet/runs/ali_ali_s2/probe/reps/th0.60", "/local/scratch/sas245/beetle-hnet/runs/ali_ali_s2/probe/reps/th0.70"], "t": 4820.7}
{"event": "final", "run": "ali_ali_s2", "arch": "ali", "objective": "ratio", "condition": "A", "step": 6000, "params": 22785088, "seed": 2, "per_lang": {"en": {"cpc": 0.198931218755387, "cps": 27.046875, "bpb": 1.8996927689640364, "bytes_per_chunk": 5.033825237871208, "reads_per_chunk": 1.2592244102426238}, "tr": {"cpc": 0.21278241175495774, "cps": 30.09375, "bpb": 2.000258989870039, "bytes_per_chunk": 5.287986525058751, "reads_per_chunk": 1.1604385407526199}, "fi": {"cpc": 0.2420923927489235, "cps": 35.578125, "bpb": 2.1298347928691213, "bytes_per_chunk": 4.516648144797939, "reads_per_chunk": 1.0113801807389795}}, "bpb_mean": 2.009928850567732, "gini_cpc": 0.04401017269961849, "gini_cps": 0.061341422312099825, "bpic": 4.780249410178632, "mean_boundary_entropy": 0.1135382205247879, "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": 472.7523412704468, "probe_overhead_seq_frac": 0.49990625, "wall_sec": 4825.126570940018, "inventory": {"size": 327666, "admitted": 65536, "staged": 262130, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 1115, "admitted_surface": 25955, "staged_surface": 58566, "admitted_split": 5226, "staged_split": 43400, "admitted_merge": 1953, "staged_merge": 113391, "admitted_skel": 32402, "staged_skel": 46773, "mean_count": 354.78802490234375, "n_util_obs": 400, "util_count_spearman": -0.09193389570565129}, "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": -0.09193389570565129, "paired_vs_dropout": {"n": 786, "pearson": 0.024077608116929692, "spearman": 0.015192791635274499, "mean_abs_discrepancy": 0.45565709471702576, "mean_paired": 0.008434545248746872, "mean_dropout": 0.015219463035464287}, "predictor_calibration": {"n": 1760, "all": {"n": 1760, "pearson": 0.27194086356134745, "spearman": 0.17446890753785668, "mae": 0.035816490650177, "reliability": [{"pred": -0.016677947714924812, "obs": -0.017740517854690552, "n": 176}, {"pred": -0.004937878809869289, "obs": -0.0037392256781458855, "n": 176}, {"pred": -0.0017805221723392606, "obs": 0.0012831037165597081, "n": 176}, {"pred": 0.0007518658530898392, "obs": -0.0013531275326386094, "n": 176}, {"pred": 0.002950140042230487, "obs": -0.0027285749092698097, "n": 176}, {"pred": 0.004957593511790037, "obs": 0.002060173312202096, "n": 176}, {"pred": 0.006592847872525454, "obs": 0.008401456288993359, "n": 176}, {"pred": 0.00861427653580904, "obs": 0.009444334544241428, "n": 176}, {"pred": 0.012249228544533253, "obs": 0.004808991681784391, "n": 176}, {"pred": 0.030282508581876755, "obs": 0.032979901880025864, "n": 176}]}, "admitted": {"n": 1664, "pearson": 0.27126384192197844, "spearman": 0.14323910119391775, "mae": 0.03694799169898033, "reliability": [{"pred": -0.014275510795414448, "obs": -0.01375713013112545, "n": 167}, {"pred": -0.004109010100364685, "obs": 0.002235351363196969, "n": 166}, {"pred": -0.0010778754949569702, "obs": -0.0022600579541176558, "n": 166}, {"pred": 0.0013819066807627678, "obs": 0.0005889079766348004, "n": 167}, {"pred": 0.003456159494817257, "obs": -0.004050337243825197, "n": 166}, {"pred": 0.005341973155736923, "obs": 0.0051288180984556675, "n": 166}, {"pred": 0.00684454757720232, "obs": 0.010526491329073906, "n": 167}, {"pred": 0.008928881026804447, "obs": 0.007826470769941807, "n": 166}, {"pred": 0.012642021290957928, "obs": 0.005376995541155338, "n": 166}, {"pred": 0.03106619231402874, "obs": 0.03403934836387634, "n": 167}]}, "staged": {"n": 96, "pearson": 0.2526075665089079, "spearman": 0.31759359739555076, "mae": 0.016203748062253, "reliability": [{"pred": -0.04305326193571091, "obs": -0.022864140570163727, "n": 10}, {"pred": -0.014916949905455112, "obs": -0.019766394048929214, "n": 9}, {"pred": -0.010655229911208153, "obs": -0.035392455756664276, "n": 10}, {"pred": -0.0075031304731965065, "obs": -0.01895267516374588, "n": 9}, {"pred": -0.006007531192153692, "obs": -0.016960132867097855, "n": 10}, {"pred": -0.004678408615291119, "obs": -0.014496909454464912, "n": 10}, {"pred": -0.002913302043452859, "obs": -0.013845675624907017, "n": 9}, {"pred": -0.0008945651352405548, "obs": -0.018120767548680305, "n": 9}, {"pred": 0.001555976690724492, "obs": -0.008646379224956036, "n": 10}, {"pred": 0.007470910437405109, "obs": -0.010902724228799343, "n": 10}]}}, "t": 4825.2}