{"event": "start", "run": "ali_ali-surface-only_s0", "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"], "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": 9.1}
{"event": "probe_schedule", "n": 40, "steps": [0, 1, 2, 4, 8, 16, 32, 64, 128, 200, 256, 400, 600, 800, 1000, 1200, 1400, 1600, 1800, 2000, 2200, 2400, 2600, 2800, 3000, 3200, 3400, 3600, 3800, 4000, 4200, 4400, 4600, 4800, 5000, 5200, 5400, 5600, 5800, 6000], "t": 9.1}
{"event": "probe", "step": 0, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 6384.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.366833, "cps": 49.875, "mean_p": 0.496725, "boundary_entropy": 0.686995, "bpb": 8.160238}, "tr": {"chunks": 5290.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.292217, "cps": 41.328125, "mean_p": 0.489567, "boundary_entropy": 0.687288, "bpb": 8.171434}, "fi": {"chunks": 8895.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.472862, "cps": 69.492188, "mean_p": 0.500688, "boundary_entropy": 0.687512, "bpb": 8.214758}}, "bpb": {"en": 8.16024, "tr": 8.17143, "fi": 8.21476}, "gini_cpc": 0.106395, "gini_cps": 0.116842, "bpic": 2.758131, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.7965927738475616, "fi": 1.2890367847879758}, "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": 10.9}
{"event": "train", "step": 0, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 5.66332, "bpb": 8.17044, "chunk_loss": 0.038537, "F": 0.28732, "G": 0.4892, "chunks": 294.22, "bytes_per_chunk": 3.48, "boundary_entropy": 0.69137, "router_noise": 0.0, "grad_norm": 11.0318, "lr": 1e-06, "router_trainable": true, "sec_per_step": 0.01, "probe_sec": 0.0056, "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}, "mean_gate": {"surface": null}, "n_reads": 0, "n_cand": 9379, "read_suppressed_frac": 0.0, "dropout_reads_logged": 0, "inventory": {"size": 3696, "admitted": 0, "staged": 3696, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 0, "staged_surface": 3696, "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}, "probe_overhead_sec_total": 0.006, "probe_overhead_seq_frac": 0.0, "util_loss": 0.0, "alpha": 0.03, "t": 12.6}
{"event": "probe", "step": 1, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 6301.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.362064, "cps": 49.226562, "mean_p": 0.496453, "boundary_entropy": 0.686986, "bpb": 8.128238}, "tr": {"chunks": 5227.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.288737, "cps": 40.835938, "mean_p": 0.489372, "boundary_entropy": 0.687277, "bpb": 8.149249}, "fi": {"chunks": 8822.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.468981, "cps": 68.921875, "mean_p": 0.500418, "boundary_entropy": 0.687508, "bpb": 8.18719}}, "bpb": {"en": 8.12824, "tr": 8.14925, "fi": 8.18719}, "gini_cpc": 0.107309, "gini_cps": 0.117772, "bpic": 2.787813, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.7974741038826102, "fi": 1.2952983448795534}, "inventory": {"size": 3696, "admitted": 0, "staged": 3696, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 0, "staged_surface": 3696, "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": 13.7}
{"event": "probe", "step": 2, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 6171.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.354594, "cps": 48.210938, "mean_p": 0.495924, "boundary_entropy": 0.686967, "bpb": 8.064147}, "tr": {"chunks": 5125.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.283102, "cps": 40.039062, "mean_p": 0.488972, "boundary_entropy": 0.687254, "bpb": 8.105291}, "fi": {"chunks": 8677.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.461273, "cps": 67.789062, "mean_p": 0.499888, "boundary_entropy": 0.687499, "bpb": 8.131014}}, "bpb": {"en": 8.06415, "tr": 8.10529, "fi": 8.13101}, "gini_cpc": 0.108083, "gini_cps": 0.11856, "bpic": 2.840435, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.7983841235176952, "fi": 1.3008472038994343}, "inventory": {"size": 6587, "admitted": 0, "staged": 6587, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 0, "staged_surface": 6587, "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": 17.1}
{"event": "probe", "step": 4, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 5667.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.325634, "cps": 44.273438, "mean_p": 0.49402, "boundary_entropy": 0.686888, "bpb": 7.849112}, "tr": {"chunks": 4744.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.262056, "cps": 37.0625, "mean_p": 0.48754, "boundary_entropy": 0.687166, "bpb": 7.951808}, "fi": {"chunks": 8071.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.429057, "cps": 63.054688, "mean_p": 0.497938, "boundary_entropy": 0.687458, "bpb": 7.941711}}, "bpb": {"en": 7.84911, "tr": 7.95181, "fi": 7.94171}, "gini_cpc": 0.1095, "gini_cps": 0.120009, "bpic": 3.069581, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.8047575067988549, "fi": 1.317608450210431}, "inventory": {"size": 11948, "admitted": 0, "staged": 11948, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 0, "staged_surface": 11948, "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": 20.8}
{"event": "probe", "step": 8, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 4366.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.250876, "cps": 34.109375, "mean_p": 0.488766, "boundary_entropy": 0.686599, "bpb": 7.288713}, "tr": {"chunks": 3766.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.208032, "cps": 29.421875, "mean_p": 0.483655, "boundary_entropy": 0.686884, "bpb": 7.525002}, "fi": {"chunks": 6363.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.33826, "cps": 49.710938, "mean_p": 0.492192, "boundary_entropy": 0.687252, "bpb": 7.435187}}, "bpb": {"en": 7.28871, "tr": 7.525, "fi": 7.43519}, "gini_cpc": 0.108909, "gini_cps": 0.119443, "bpic": 3.913901, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.8292207346423476, "fi": 1.3483120895944622}, "inventory": {"size": 21419, "admitted": 0, "staged": 21419, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 0, "staged_surface": 21419, "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": 25.6}
{"event": "probe", "step": 16, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 4439.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.255071, "cps": 34.679688, "mean_p": 0.490989, "boundary_entropy": 0.686932, "bpb": 6.660771}, "tr": {"chunks": 4545.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.251063, "cps": 35.507812, "mean_p": 0.488392, "boundary_entropy": 0.687352, "bpb": 6.978585}, "fi": {"chunks": 5616.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.298549, "cps": 43.875, "mean_p": 0.491252, "boundary_entropy": 0.687412, "bpb": 6.839335}}, "bpb": {"en": 6.66077, "tr": 6.97859, "fi": 6.83934}, "gini_cpc": 0.039341, "gini_cps": 0.053744, "bpic": 3.885753, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9842882739885332, "fi": 1.1704535706551513}, "inventory": {"size": 37486, "admitted": 0, "staged": 37486, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 0, "staged_surface": 37486, "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": 31.8}
{"event": "probe", "step": 32, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 4018.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.23088, "cps": 31.390625, "mean_p": 0.494609, "boundary_entropy": 0.687551, "bpb": 6.307229}, "tr": {"chunks": 4093.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.226095, "cps": 31.976562, "mean_p": 0.491945, "boundary_entropy": 0.687959, "bpb": 6.636435}, "fi": {"chunks": 4322.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.229759, "cps": 33.765625, "mean_p": 0.492913, "boundary_entropy": 0.687998, "bpb": 6.431558}}, "bpb": {"en": 6.30723, "tr": 6.63643, "fi": 6.43156}, "gini_cpc": 0.004645, "gini_cps": 0.016301, "bpic": 4.563018, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9792766088479872, "fi": 0.9951466076887169}, "inventory": {"size": 69339, "admitted": 0, "staged": 69339, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 0, "staged_surface": 69339, "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": 42.8}
{"event": "consolidate", "step": 50, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 6, "swapped": 0, "uncertain_staged": 103555, "t": 56.4}
{"event": "train", "step": 50, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 4.22567, "bpb": 6.09635, "chunk_loss": 0.030143, "F": 0.16833, "G": 0.49106, "chunks": 172.38, "bytes_per_chunk": 5.941, "boundary_entropy": 0.69209, "router_noise": 0.0, "grad_norm": 2.4393, "lr": 5.1e-05, "router_trainable": true, "sec_per_step": 0.884, "probe_sec": 0.0732, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.05875, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": 0.00793, "n_shadow": 6, "mean_delta_shadow": 0.00094, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": -1.61895, "trunc_abs_frac": 0.58978, "mean_delta_full": -0.00605, "read_hit_rate": {"surface": 0.08666}, "mean_gate": {"surface": 0.12089}, "n_reads": 475, "n_cand": 5481, "read_suppressed_frac": 0.1032, "dropout_reads_logged": 475, "inventory": {"size": 105518, "admitted": 6, "staged": 105512, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 6, "staged_surface": 105512, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 5401.1669921875, "n_util_obs": 6, "util_count_spearman": -0.5428571428571428}, "probe_overhead_sec_total": 8.874, "probe_overhead_seq_frac": 0.4902, "util_loss": 0.041744, "alpha": 0.03, "t": 56.8}
{"event": "probe", "step": 64, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 4082.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.234557, "cps": 31.890625, "mean_p": 0.48841, "boundary_entropy": 0.686628, "bpb": 5.623343}, "tr": {"chunks": 3929.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.217036, "cps": 30.695312, "mean_p": 0.488288, "boundary_entropy": 0.687525, "bpb": 6.010052}, "fi": {"chunks": 5368.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.285365, "cps": 41.9375, "mean_p": 0.493028, "boundary_entropy": 0.687704, "bpb": 5.81711}}, "bpb": {"en": 5.62334, "tr": 6.01005, "fi": 5.81711}, "gini_cpc": 0.061812, "gini_cps": 0.071704, "bpic": 4.240377, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9253000746477036, "fi": 1.2166110132220895}, "inventory": {"size": 129800, "admitted": 6, "staged": 129794, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 6, "staged_surface": 129794, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 6143.1669921875, "n_util_obs": 6, "util_count_spearman": -0.9428571428571428}, "t": 67.1}
{"event": "consolidate", "step": 100, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 65530, "swapped": 6, "uncertain_staged": 193284, "t": 91.5}
{"event": "train", "step": 100, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 3.20557, "bpb": 4.62466, "chunk_loss": 0.031904, "F": 0.20145, "G": 0.42006, "chunks": 206.28, "bytes_per_chunk": 4.964, "boundary_entropy": 0.66537, "router_noise": 0.0, "grad_norm": 2.6214, "lr": 0.000101, "router_trainable": true, "sec_per_step": 0.707, "probe_sec": 0.2038, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.1932, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": -0.00745, "n_shadow": 7, "mean_delta_shadow": -0.0095, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 0.2816, "trunc_abs_frac": 0.4663, "mean_delta_full": -0.0953, "read_hit_rate": {"surface": 0.53646}, "mean_gate": {"surface": 0.12085}, "n_reads": 3524, "n_cand": 6569, "read_suppressed_frac": 0.1019, "dropout_reads_logged": 3524, "inventory": {"size": 194816, "admitted": 65536, "staged": 129280, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 129280, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 4.9431610107421875, "n_util_obs": 1, "util_count_spearman": NaN}, "probe_overhead_sec_total": 18.136, "probe_overhead_seq_frac": 0.495, "util_loss": 0.001019, "alpha": 0.03, "t": 92.2}
{"event": "probe", "step": 128, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3559.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.204505, "cps": 27.804688, "mean_p": 0.375251, "boundary_entropy": 0.608971, "bpb": 4.179855}, "tr": {"chunks": 3012.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.166381, "cps": 23.53125, "mean_p": 0.411608, "boundary_entropy": 0.651363, "bpb": 4.534841}, "fi": {"chunks": 3477.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.184839, "cps": 27.164062, "mean_p": 0.39318, "boundary_entropy": 0.635835, "bpb": 4.418043}}, "bpb": {"en": 4.17986, "tr": 4.53484, "fi": 4.41804}, "gini_cpc": 0.045735, "gini_cps": 0.036292, "bpic": 5.646099, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.8135805327502423, "fi": 0.9038345516182932}, "inventory": {"size": 235773, "admitted": 65536, "staged": 170237, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 170237, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 6.3027191162109375, "n_util_obs": 7, "util_count_spearman": -0.8928571428571429}, "t": 106.7}
{"event": "consolidate", "step": 150, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 12, "uncertain_staged": 202659, "t": 123.3}
{"event": "train", "step": 150, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 2.8092, "bpb": 4.05283, "chunk_loss": 0.03187, "F": 0.20901, "G": 0.37065, "chunks": 214.03, "bytes_per_chunk": 4.784, "boundary_entropy": 0.59831, "router_noise": 0.0, "grad_norm": 2.2593, "lr": 0.000151, "router_trainable": true, "sec_per_step": 0.634, "probe_sec": 0.29, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.2816, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": 0.00918, "n_shadow": 5, "mean_delta_shadow": -0.01575, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": -0.10625, "trunc_abs_frac": 0.56398, "mean_delta_full": -0.09115, "read_hit_rate": {"surface": 0.39225}, "mean_gate": {"surface": 0.12152}, "n_reads": 2674, "n_cand": 6817, "read_suppressed_frac": 0.0984, "dropout_reads_logged": 2674, "inventory": {"size": 269645, "admitted": 65536, "staged": 204109, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 204109, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 6.7007598876953125, "n_util_obs": 1, "util_count_spearman": NaN}, "probe_overhead_sec_total": 27.488, "probe_overhead_seq_frac": 0.4967, "util_loss": 0.000515, "alpha": 0.03, "t": 123.9}
{"event": "consolidate", "step": 200, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 7, "uncertain_staged": 262131, "t": 151.9}
{"event": "probe", "step": 200, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3078.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.176866, "cps": 24.046875, "mean_p": 0.375812, "boundary_entropy": 0.580554, "bpb": 3.619475}, "tr": {"chunks": 3206.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.177098, "cps": 25.046875, "mean_p": 0.376405, "boundary_entropy": 0.597546, "bpb": 3.772443}, "fi": {"chunks": 3204.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.170326, "cps": 25.03125, "mean_p": 0.379531, "boundary_entropy": 0.61127, "bpb": 3.864132}}, "bpb": {"en": 3.61947, "tr": 3.77244, "fi": 3.86413}, "gini_cpc": 0.008611, "gini_cps": 0.008994, "bpic": 5.979342, "worst_lang": "tr", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0013098105824814, "fi": 0.9630218228043129}, "inventory": {"size": 327673, "admitted": 65536, "staged": 262137, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262137, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 8.056060791015625, "n_util_obs": 2, "util_count_spearman": NaN}, "t": 153.3}
{"event": "eval", "step": 200, "phase": 1, "bpb": {"en": 3.62661, "tr": 3.78176, "fi": 3.86391}, "bytes_per_chunk": {"en": 5.5, "tr": 5.849, "fi": 6.142}, "bpb_mean": 3.75742, "frozen_inventory": true, "t": 155.7}
{"event": "train", "step": 200, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 2.54035, "bpb": 3.66496, "chunk_loss": 0.030587, "F": 0.17834, "G": 0.37891, "chunks": 182.62, "bytes_per_chunk": 5.607, "boundary_entropy": 0.58732, "router_noise": 0.0, "grad_norm": 1.2146, "lr": 0.000201, "router_trainable": true, "sec_per_step": 0.703, "probe_sec": 0.2172, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.20932, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": -0.01026, "n_shadow": 6, "mean_delta_shadow": -0.02737, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": -5.1, "trunc_abs_frac": 4.16592, "mean_delta_full": 0.00407, "read_hit_rate": {"surface": 0.34429}, "mean_gate": {"surface": 0.13032}, "n_reads": 2001, "n_cand": 5812, "read_suppressed_frac": 0.0905, "dropout_reads_logged": 2001, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 8.086593627929688, "n_util_obs": 2, "util_count_spearman": NaN}, "probe_overhead_sec_total": 36.126, "probe_overhead_seq_frac": 0.4975, "util_loss": 0.000847, "alpha": 0.03, "t": 159.0}
{"event": "consolidate", "step": 250, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262133, "t": 191.8}
{"event": "train", "step": 250, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 2.38488, "bpb": 3.44065, "chunk_loss": 0.032066, "F": 0.21686, "G": 0.36092, "chunks": 222.06, "bytes_per_chunk": 4.611, "boundary_entropy": 0.54569, "router_noise": 0.0, "grad_norm": 1.5062, "lr": 0.000251, "router_trainable": true, "sec_per_step": 0.666, "probe_sec": 0.1791, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.16599, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 10, "mean_delta_loo": -0.00474, "n_shadow": 4, "mean_delta_shadow": 0.00299, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 0.64168, "trunc_abs_frac": 2.1772, "mean_delta_full": 0.06769, "read_hit_rate": {"surface": 0.38182}, "mean_gate": {"surface": 0.13366}, "n_reads": 2701, "n_cand": 7074, "read_suppressed_frac": 0.1011, "dropout_reads_logged": 2701, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 9.183731079101562, "n_util_obs": 11, "util_count_spearman": 0.42727272727272725}, "probe_overhead_sec_total": 46.042, "probe_overhead_seq_frac": 0.498, "util_loss": 0.001208, "alpha": 0.03, "t": 192.3}
{"event": "probe", "step": 256, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3275.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.188186, "cps": 25.585938, "mean_p": 0.330389, "boundary_entropy": 0.522512, "bpb": 3.344405}, "tr": {"chunks": 3400.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.187814, "cps": 26.5625, "mean_p": 0.348935, "boundary_entropy": 0.537519, "bpb": 3.517849}, "fi": {"chunks": 4052.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.215406, "cps": 31.65625, "mean_p": 0.37692, "boundary_entropy": 0.581518, "bpb": 3.648893}}, "bpb": {"en": 3.3444, "tr": 3.51785, "fi": 3.64889}, "gini_cpc": 0.031103, "gini_cps": 0.048289, "bpic": 5.288711, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9980244512634024, "fi": 1.1446438233922738}, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 9.384002685546875, "n_util_obs": 11, "util_count_spearman": 0.42727272727272725}, "t": 196.9}
{"event": "consolidate", "step": 300, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262131, "t": 227.1}
{"event": "train", "step": 300, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 2.28367, "bpb": 3.29464, "chunk_loss": 0.031218, "F": 0.20874, "G": 0.29719, "chunks": 213.75, "bytes_per_chunk": 4.791, "boundary_entropy": 0.43248, "router_noise": 0.0, "grad_norm": 1.2963, "lr": 0.0003, "router_trainable": true, "sec_per_step": 0.705, "probe_sec": 0.166, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.15509, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": 0.01287, "n_shadow": 5, "mean_delta_shadow": 0.00756, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 0.04632, "trunc_abs_frac": 0.37504, "mean_delta_full": 0.10706, "read_hit_rate": {"surface": 0.21431}, "mean_gate": {"surface": 0.14253}, "n_reads": 1459, "n_cand": 6808, "read_suppressed_frac": 0.1083, "dropout_reads_logged": 1459, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 6.853546142578125, "n_util_obs": 11, "util_count_spearman": 0.17272727272727273}, "probe_overhead_sec_total": 55.503, "probe_overhead_seq_frac": 0.4983, "util_loss": 0.001402, "alpha": 0.03, "t": 227.6}
{"event": "consolidate", "step": 350, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262128, "t": 257.7}
{"event": "train", "step": 350, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 2.19789, "bpb": 3.17088, "chunk_loss": 0.031081, "F": 0.2049, "G": 0.29458, "chunks": 209.81, "bytes_per_chunk": 4.881, "boundary_entropy": 0.40853, "router_noise": 0.0, "grad_norm": 1.4658, "lr": 0.0002999, "router_trainable": true, "sec_per_step": 0.622, "probe_sec": 0.1766, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.15267, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 10, "mean_delta_loo": 0.04615, "n_shadow": 4, "mean_delta_shadow": 0.06879, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 3.50294, "trunc_abs_frac": 0.96521, "mean_delta_full": 0.02771, "read_hit_rate": {"surface": 0.14848}, "mean_gate": {"surface": 0.18324}, "n_reads": 992, "n_cand": 6681, "read_suppressed_frac": 0.1129, "dropout_reads_logged": 992, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 6.494598388671875, "n_util_obs": 15, "util_count_spearman": -0.03571428571428571}, "probe_overhead_sec_total": 64.394, "probe_overhead_seq_frac": 0.4986, "util_loss": 0.001713, "alpha": 0.03, "t": 258.7}
{"event": "consolidate", "step": 400, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262128, "t": 290.0}
{"event": "probe", "step": 400, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3555.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.204275, "cps": 27.773438, "mean_p": 0.262083, "boundary_entropy": 0.345227, "bpb": 2.902778}, "tr": {"chunks": 3919.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.216483, "cps": 30.617188, "mean_p": 0.301991, "boundary_entropy": 0.434415, "bpb": 3.191806}, "fi": {"chunks": 4107.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.21833, "cps": 32.085938, "mean_p": 0.297332, "boundary_entropy": 0.42441, "bpb": 3.36982}}, "bpb": {"en": 2.90278, "tr": 3.19181, "fi": 3.36982}, "gini_cpc": 0.014661, "gini_cps": 0.031776, "bpic": 4.898713, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.059764157792808, "fi": 1.0688021888620245}, "inventory": {"size": 326656, "admitted": 65536, "staged": 261120, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261120, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 6.7655029296875, "n_util_obs": 20, "util_count_spearman": 0.081203007518797}, "t": 291.6}
{"event": "eval", "step": 400, "phase": 1, "bpb": {"en": 2.91787, "tr": 3.19935, "fi": 3.36233}, "bytes_per_chunk": {"en": 4.635, "tr": 5.151, "fi": 4.817}, "bpb_mean": 3.15985, "frozen_inventory": true, "t": 294.1}
{"event": "train", "step": 400, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 2.04831, "bpb": 2.95509, "chunk_loss": 0.031228, "F": 0.21588, "G": 0.27744, "chunks": 221.06, "bytes_per_chunk": 4.632, "boundary_entropy": 0.3548, "router_noise": 0.0, "grad_norm": 1.1499, "lr": 0.0002998, "router_trainable": true, "sec_per_step": 0.771, "probe_sec": 0.1652, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.15343, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": 0.01645, "n_shadow": 6, "mean_delta_shadow": -0.02917, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": 1.52684, "trunc_abs_frac": 1.46724, "mean_delta_full": 0.01729, "read_hit_rate": {"surface": 0.12851}, "mean_gate": {"surface": 0.22169}, "n_reads": 905, "n_cand": 7042, "read_suppressed_frac": 0.1017, "dropout_reads_logged": 905, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 6.7793121337890625, "n_util_obs": 20, "util_count_spearman": 0.09323308270676692}, "probe_overhead_sec_total": 73.3, "probe_overhead_seq_frac": 0.4988, "util_loss": 0.003064, "alpha": 0.03, "t": 297.2}
{"event": "consolidate", "step": 450, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262128, "t": 328.6}
{"event": "train", "step": 450, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 2.01465, "bpb": 2.90652, "chunk_loss": 0.030929, "F": 0.20541, "G": 0.27005, "chunks": 210.34, "bytes_per_chunk": 4.868, "boundary_entropy": 0.34763, "router_noise": 0.0, "grad_norm": 1.1958, "lr": 0.0002995, "router_trainable": true, "sec_per_step": 0.642, "probe_sec": 0.0694, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.06001, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": 0.2993, "n_shadow": 6, "mean_delta_shadow": 0.1314, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 1.0688, "trunc_abs_frac": 1.1116, "mean_delta_full": 0.2149, "read_hit_rate": {"surface": 0.13181}, "mean_gate": {"surface": 0.24869}, "n_reads": 883, "n_cand": 6699, "read_suppressed_frac": 0.0895, "dropout_reads_logged": 883, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 7.5559539794921875, "n_util_obs": 24, "util_count_spearman": -0.05565217391304348}, "probe_overhead_sec_total": 82.056, "probe_overhead_seq_frac": 0.4989, "util_loss": 0.005122, "alpha": 0.03, "t": 329.3}
{"event": "consolidate", "step": 500, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262125, "t": 358.8}
{"event": "train", "step": 500, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.99331, "bpb": 2.87574, "chunk_loss": 0.031224, "F": 0.21695, "G": 0.27182, "chunks": 222.16, "bytes_per_chunk": 4.609, "boundary_entropy": 0.33059, "router_noise": 0.0, "grad_norm": 1.0707, "lr": 0.0002992, "router_trainable": true, "sec_per_step": 0.604, "probe_sec": 0.0858, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.07832, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 10, "mean_delta_loo": 0.02567, "n_shadow": 4, "mean_delta_shadow": 0.0534, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 6, "trunc_ratio_sum": 0.21694, "trunc_abs_frac": 5.45446, "mean_delta_full": 0.10323, "read_hit_rate": {"surface": 0.15204}, "mean_gate": {"surface": 0.23912}, "n_reads": 1076, "n_cand": 7077, "read_suppressed_frac": 0.0976, "dropout_reads_logged": 1076, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 8.310821533203125, "n_util_obs": 28, "util_count_spearman": -0.3459222769567597}, "probe_overhead_sec_total": 90.462, "probe_overhead_seq_frac": 0.499, "util_loss": 0.011064, "alpha": 0.03, "t": 359.5}
{"event": "consolidate", "step": 550, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262117, "t": 389.9}
{"event": "train", "step": 550, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.97515, "bpb": 2.84954, "chunk_loss": 0.031301, "F": 0.2196, "G": 0.27413, "chunks": 224.88, "bytes_per_chunk": 4.554, "boundary_entropy": 0.32644, "router_noise": 0.0, "grad_norm": 1.2615, "lr": 0.0002987, "router_trainable": true, "sec_per_step": 0.623, "probe_sec": 0.2142, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.19656, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": 0.0656, "n_shadow": 7, "mean_delta_shadow": 0.00186, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 0.03876, "trunc_abs_frac": 16.17142, "mean_delta_full": -0.09882, "read_hit_rate": {"surface": 0.15787}, "mean_gate": {"surface": 0.23807}, "n_reads": 1131, "n_cand": 7164, "read_suppressed_frac": 0.1158, "dropout_reads_logged": 1131, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 8.935699462890625, "n_util_obs": 37, "util_count_spearman": -0.14390706495969655}, "probe_overhead_sec_total": 98.752, "probe_overhead_seq_frac": 0.4991, "util_loss": 0.011955, "alpha": 0.03, "t": 390.6}
{"event": "consolidate", "step": 600, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262114, "t": 419.7}
{"event": "probe", "step": 600, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3702.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.212722, "cps": 28.921875, "mean_p": 0.259836, "boundary_entropy": 0.303884, "bpb": 2.603127}, "tr": {"chunks": 3530.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.194995, "cps": 27.578125, "mean_p": 0.266433, "boundary_entropy": 0.361803, "bpb": 2.93823}, "fi": {"chunks": 4433.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.23566, "cps": 34.632812, "mean_p": 0.277507, "boundary_entropy": 0.338552, "bpb": 3.158258}}, "bpb": {"en": 2.60313, "tr": 2.93823, "fi": 3.15826}, "gini_cpc": 0.042137, "gini_cps": 0.051607, "bpic": 4.863438, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.916667554497043, "fi": 1.1078311019613598}, "inventory": {"size": 326656, "admitted": 65536, "staged": 261120, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261120, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 9.439987182617188, "n_util_obs": 38, "util_count_spearman": -0.04759820549294234}, "t": 421.3}
{"event": "eval", "step": 600, "phase": 1, "bpb": {"en": 2.6167, "tr": 2.94632, "fi": 3.15848}, "bytes_per_chunk": {"en": 4.519, "tr": 5.806, "fi": 4.454}, "bpb_mean": 2.90716, "frozen_inventory": true, "t": 423.7}
{"event": "train", "step": 600, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.87858, "bpb": 2.71021, "chunk_loss": 0.03106, "F": 0.21381, "G": 0.26672, "chunks": 218.94, "bytes_per_chunk": 4.677, "boundary_entropy": 0.30818, "router_noise": 0.0, "grad_norm": 1.2316, "lr": 0.0002982, "router_trainable": true, "sec_per_step": 0.728, "probe_sec": 0.2157, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.20252, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": 0.47082, "n_shadow": 7, "mean_delta_shadow": -0.05431, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 6, "trunc_ratio_sum": 0.78313, "trunc_abs_frac": 0.76466, "mean_delta_full": 0.06325, "read_hit_rate": {"surface": 0.12776}, "mean_gate": {"surface": 0.25178}, "n_reads": 891, "n_cand": 6974, "read_suppressed_frac": 0.0774, "dropout_reads_logged": 891, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 9.453582763671875, "n_util_obs": 39, "util_count_spearman": -0.05708502024291498}, "probe_overhead_sec_total": 107.273, "probe_overhead_seq_frac": 0.4992, "util_loss": 0.013255, "alpha": 0.03, "t": 427.0}
{"event": "consolidate", "step": 650, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262107, "t": 457.7}
{"event": "train", "step": 650, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.91223, "bpb": 2.75877, "chunk_loss": 0.031053, "F": 0.21207, "G": 0.26569, "chunks": 217.16, "bytes_per_chunk": 4.715, "boundary_entropy": 0.30871, "router_noise": 0.0, "grad_norm": 1.2462, "lr": 0.0002975, "router_trainable": true, "sec_per_step": 0.625, "probe_sec": 0.139, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.1261, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": 0.16598, "n_shadow": 5, "mean_delta_shadow": -0.08739, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 0.51631, "trunc_abs_frac": 0.78407, "mean_delta_full": -0.20303, "read_hit_rate": {"surface": 0.13127}, "mean_gate": {"surface": 0.25841}, "n_reads": 908, "n_cand": 6917, "read_suppressed_frac": 0.076, "dropout_reads_logged": 908, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 9.477096557617188, "n_util_obs": 38, "util_count_spearman": -0.21917058759164024}, "probe_overhead_sec_total": 115.311, "probe_overhead_seq_frac": 0.4992, "util_loss": 0.01334, "alpha": 0.03, "t": 458.3}
{"event": "consolidate", "step": 700, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262106, "t": 488.6}
{"event": "train", "step": 700, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.85788, "bpb": 2.68035, "chunk_loss": 0.030925, "F": 0.20886, "G": 0.26206, "chunks": 213.88, "bytes_per_chunk": 4.788, "boundary_entropy": 0.29922, "router_noise": 0.0, "grad_norm": 1.0411, "lr": 0.0002967, "router_trainable": true, "sec_per_step": 0.625, "probe_sec": 0.1301, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.12058, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": 0.0743, "n_shadow": 6, "mean_delta_shadow": -0.17642, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": -3.5455, "trunc_abs_frac": 1.59624, "mean_delta_full": 0.03119, "read_hit_rate": {"surface": 0.165}, "mean_gate": {"surface": 0.25592}, "n_reads": 1124, "n_cand": 6812, "read_suppressed_frac": 0.097, "dropout_reads_logged": 1124, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 11.133941650390625, "n_util_obs": 46, "util_count_spearman": -0.3624421831637373}, "probe_overhead_sec_total": 123.152, "probe_overhead_seq_frac": 0.4993, "util_loss": 0.01814, "alpha": 0.03, "t": 489.5}
{"event": "consolidate", "step": 750, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262107, "t": 517.7}
{"event": "train", "step": 750, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.81601, "bpb": 2.61995, "chunk_loss": 0.030737, "F": 0.20059, "G": 0.25998, "chunks": 205.41, "bytes_per_chunk": 4.985, "boundary_entropy": 0.29436, "router_noise": 0.0, "grad_norm": 1.1673, "lr": 0.0002959, "router_trainable": true, "sec_per_step": 0.58, "probe_sec": 0.0497, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02996, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": -0.19284, "n_shadow": 5, "mean_delta_shadow": 0.00462, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": -1.36161, "trunc_abs_frac": 13.23018, "mean_delta_full": 0.01367, "read_hit_rate": {"surface": 0.29323}, "mean_gate": {"surface": 0.22523}, "n_reads": 1918, "n_cand": 6541, "read_suppressed_frac": 0.0907, "dropout_reads_logged": 1918, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 13.974624633789062, "n_util_obs": 57, "util_count_spearman": -0.35808850938894743}, "probe_overhead_sec_total": 131.265, "probe_overhead_seq_frac": 0.4993, "util_loss": 0.021942, "alpha": 0.03, "t": 518.5}
{"event": "consolidate", "step": 800, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262108, "t": 552.0}
{"event": "probe", "step": 800, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3367.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.193472, "cps": 26.304688, "mean_p": 0.245082, "boundary_entropy": 0.274473, "bpb": 2.440206}, "tr": {"chunks": 3570.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.197205, "cps": 27.890625, "mean_p": 0.259792, "boundary_entropy": 0.331487, "bpb": 2.777338}, "fi": {"chunks": 4538.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.241242, "cps": 35.453125, "mean_p": 0.27862, "boundary_entropy": 0.305006, "bpb": 2.961241}}, "bpb": {"en": 2.44021, "tr": 2.77734, "fi": 2.96124}, "gini_cpc": 0.050396, "gini_cps": 0.068032, "bpic": 4.943965, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0192921240813855, "fi": 1.2469057048292602}, "inventory": {"size": 326656, "admitted": 65536, "staged": 261120, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261120, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 18.248489379882812, "n_util_obs": 67, "util_count_spearman": -0.28326283023385745}, "t": 553.5}
{"event": "eval", "step": 800, "phase": 1, "bpb": {"en": 2.45323, "tr": 2.78686, "fi": 2.96517}, "bytes_per_chunk": {"en": 4.925, "tr": 5.594, "fi": 4.4}, "bpb_mean": 2.73509, "frozen_inventory": true, "t": 555.9}
{"event": "train", "step": 800, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.80794, "bpb": 2.60831, "chunk_loss": 0.030893, "F": 0.20633, "G": 0.25976, "chunks": 211.28, "bytes_per_chunk": 4.847, "boundary_entropy": 0.28399, "router_noise": 0.0, "grad_norm": 0.9184, "lr": 0.0002949, "router_trainable": true, "sec_per_step": 0.81, "probe_sec": 0.0612, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04318, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": 0.02441, "n_shadow": 5, "mean_delta_shadow": -0.16727, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 0.43391, "trunc_abs_frac": 6.10935, "mean_delta_full": 0.16519, "read_hit_rate": {"surface": 0.39768}, "mean_gate": {"surface": 0.19269}, "n_reads": 2676, "n_cand": 6729, "read_suppressed_frac": 0.1009, "dropout_reads_logged": 2676, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 18.289321899414062, "n_util_obs": 67, "util_count_spearman": -0.28078857051640194}, "probe_overhead_sec_total": 139.517, "probe_overhead_seq_frac": 0.4994, "util_loss": 0.019353, "alpha": 0.03, "t": 559.0}
{"event": "consolidate", "step": 850, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262110, "t": 588.9}
{"event": "train", "step": 850, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.77992, "bpb": 2.56788, "chunk_loss": 0.030728, "F": 0.20267, "G": 0.25356, "chunks": 207.53, "bytes_per_chunk": 4.934, "boundary_entropy": 0.27711, "router_noise": 0.0, "grad_norm": 1.1516, "lr": 0.0002938, "router_trainable": true, "sec_per_step": 0.618, "probe_sec": 0.1015, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.07786, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": 0.0059, "n_shadow": 8, "mean_delta_shadow": -0.00197, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 0.40857, "trunc_abs_frac": 0.62013, "mean_delta_full": -0.19047, "read_hit_rate": {"surface": 0.39189}, "mean_gate": {"surface": 0.19282}, "n_reads": 2590, "n_cand": 6609, "read_suppressed_frac": 0.0977, "dropout_reads_logged": 2590, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 19.48187255859375, "n_util_obs": 69, "util_count_spearman": -0.2785896967482645}, "probe_overhead_sec_total": 148.1, "probe_overhead_seq_frac": 0.4994, "util_loss": 0.019758, "alpha": 0.03, "t": 589.9}
{"event": "consolidate", "step": 900, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262110, "t": 618.8}
{"event": "train", "step": 900, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.62473, "bpb": 2.34399, "chunk_loss": 0.030651, "F": 0.20001, "G": 0.24857, "chunks": 204.81, "bytes_per_chunk": 5.0, "boundary_entropy": 0.26025, "router_noise": 0.0, "grad_norm": 1.0439, "lr": 0.0002927, "router_trainable": true, "sec_per_step": 0.593, "probe_sec": 0.247, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.22825, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 10, "mean_delta_loo": 0.08005, "n_shadow": 4, "mean_delta_shadow": 0.02855, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 0.04762, "trunc_abs_frac": 0.55852, "mean_delta_full": 0.03674, "read_hit_rate": {"surface": 0.39574}, "mean_gate": {"surface": 0.20068}, "n_reads": 2581, "n_cand": 6522, "read_suppressed_frac": 0.0996, "dropout_reads_logged": 2581, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 21.05657958984375, "n_util_obs": 79, "util_count_spearman": -0.2975900681596884}, "probe_overhead_sec_total": 156.58, "probe_overhead_seq_frac": 0.4994, "util_loss": 0.019153, "alpha": 0.03, "t": 619.5}
{"event": "consolidate", "step": 950, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262110, "t": 649.3}
{"event": "train", "step": 950, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.78068, "bpb": 2.56898, "chunk_loss": 0.030893, "F": 0.20981, "G": 0.25387, "chunks": 214.84, "bytes_per_chunk": 4.766, "boundary_entropy": 0.26182, "router_noise": 0.0, "grad_norm": 0.9905, "lr": 0.0002914, "router_trainable": true, "sec_per_step": 0.608, "probe_sec": 0.1897, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.18217, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": -0.00601, "n_shadow": 5, "mean_delta_shadow": -0.00232, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 6, "trunc_ratio_sum": 0.96342, "trunc_abs_frac": 0.57166, "mean_delta_full": 0.104, "read_hit_rate": {"surface": 0.40348}, "mean_gate": {"surface": 0.21454}, "n_reads": 2761, "n_cand": 6843, "read_suppressed_frac": 0.1039, "dropout_reads_logged": 2761, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 22.824798583984375, "n_util_obs": 86, "util_count_spearman": -0.13699832066575707}, "probe_overhead_sec_total": 165.701, "probe_overhead_seq_frac": 0.4995, "util_loss": 0.018836, "alpha": 0.03, "t": 650.0}
{"event": "consolidate", "step": 1000, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262110, "t": 677.9}
{"event": "probe", "step": 1000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3341.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.191978, "cps": 26.101562, "mean_p": 0.235471, "boundary_entropy": 0.248224, "bpb": 2.353818}, "tr": {"chunks": 3653.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.20179, "cps": 28.539062, "mean_p": 0.24568, "boundary_entropy": 0.288779, "bpb": 2.65299}, "fi": {"chunks": 4588.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.2439, "cps": 35.84375, "mean_p": 0.274282, "boundary_entropy": 0.26932, "bpb": 2.837192}}, "bpb": {"en": 2.35382, "tr": 2.65299, "fi": 2.83719}, "gini_cpc": 0.054283, "gini_cps": 0.071778, "bpic": 4.89829, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0511066055006038, "fi": 1.2704546594027186}, "inventory": {"size": 326656, "admitted": 65536, "staged": 261120, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261120, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 25.17724609375, "n_util_obs": 102, "util_count_spearman": -0.18429125165114185}, "t": 679.3}
{"event": "eval", "step": 1000, "phase": 1, "bpb": {"en": 2.36415, "tr": 2.66253, "fi": 2.84431}, "bytes_per_chunk": {"en": 4.96, "tr": 5.53, "fi": 4.346}, "bpb_mean": 2.62367, "frozen_inventory": true, "t": 681.5}
{"event": "train", "step": 1000, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.72611, "bpb": 2.49024, "chunk_loss": 0.031131, "F": 0.21548, "G": 0.25603, "chunks": 220.66, "bytes_per_chunk": 4.641, "boundary_entropy": 0.25421, "router_noise": 0.0, "grad_norm": 1.1051, "lr": 0.0002901, "router_trainable": true, "sec_per_step": 0.727, "probe_sec": 0.2183, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.20984, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": 0.2602, "n_shadow": 7, "mean_delta_shadow": -0.05109, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 0.42073, "trunc_abs_frac": 0.29958, "mean_delta_full": 0.65045, "read_hit_rate": {"surface": 0.42951}, "mean_gate": {"surface": 0.20987}, "n_reads": 3019, "n_cand": 7029, "read_suppressed_frac": 0.0997, "dropout_reads_logged": 3019, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 25.223312377929688, "n_util_obs": 102, "util_count_spearman": -0.18280360303306173}, "probe_overhead_sec_total": 173.615, "probe_overhead_seq_frac": 0.4995, "util_loss": 0.017848, "alpha": 0.03, "t": 686.3}
{"event": "consolidate", "step": 1050, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262113, "t": 707.6}
{"event": "train", "step": 1050, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.74542, "bpb": 2.51811, "chunk_loss": 0.030745, "F": 0.20526, "G": 0.24894, "chunks": 210.19, "bytes_per_chunk": 4.872, "boundary_entropy": 0.24968, "router_noise": 0.0, "grad_norm": 0.93, "lr": 0.0002886, "router_trainable": true, "sec_per_step": 0.441, "probe_sec": 0.2467, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.23323, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": 0.06534, "n_shadow": 5, "mean_delta_shadow": -0.03413, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": 1.08065, "trunc_abs_frac": 0.84818, "mean_delta_full": 0.32632, "read_hit_rate": {"surface": 0.43173}, "mean_gate": {"surface": 0.2234}, "n_reads": 2890, "n_cand": 6694, "read_suppressed_frac": 0.1042, "dropout_reads_logged": 2890, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 27.438705444335938, "n_util_obs": 118, "util_count_spearman": -0.13158399299000476}, "probe_overhead_sec_total": 180.167, "probe_overhead_seq_frac": 0.4995, "util_loss": 0.018051, "alpha": 0.03, "t": 708.4}
{"event": "consolidate", "step": 1100, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262114, "t": 721.2}
{"event": "train", "step": 1100, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.73154, "bpb": 2.49808, "chunk_loss": 0.030768, "F": 0.2037, "G": 0.24869, "chunks": 208.59, "bytes_per_chunk": 4.909, "boundary_entropy": 0.24169, "router_noise": 0.0, "grad_norm": 0.9427, "lr": 0.0002871, "router_trainable": true, "sec_per_step": 0.26, "probe_sec": 0.032, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.01991, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 11, "mean_delta_loo": 0.1093, "n_shadow": 3, "mean_delta_shadow": -0.32526, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 0, "read_hit_rate": {"surface": 0.4367}, "mean_gate": {"surface": 0.22834}, "n_reads": 2901, "n_cand": 6643, "read_suppressed_frac": 0.0972, "dropout_reads_logged": 2901, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 29.507003784179688, "n_util_obs": 124, "util_count_spearman": -0.06619365254713552}, "probe_overhead_sec_total": 182.51, "probe_overhead_seq_frac": 0.4995, "util_loss": 0.018515, "alpha": 0.03, "t": 721.4}
{"event": "consolidate", "step": 1150, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262114, "t": 734.0}
{"event": "train", "step": 1150, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.59372, "bpb": 2.29925, "chunk_loss": 0.030654, "F": 0.20331, "G": 0.2402, "chunks": 208.19, "bytes_per_chunk": 4.919, "boundary_entropy": 0.22724, "router_noise": 0.0, "grad_norm": 0.8622, "lr": 0.0002855, "router_trainable": true, "sec_per_step": 0.256, "probe_sec": 0.0251, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.01873, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": 0.00513, "n_shadow": 6, "mean_delta_shadow": 0.0393, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 0.89096, "trunc_abs_frac": 2.76528, "mean_delta_full": 0.51135, "read_hit_rate": {"surface": 0.44615}, "mean_gate": {"surface": 0.2264}, "n_reads": 2958, "n_cand": 6630, "read_suppressed_frac": 0.1065, "dropout_reads_logged": 2958, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 31.255813598632812, "n_util_obs": 124, "util_count_spearman": -0.1280163399684129}, "probe_overhead_sec_total": 185.139, "probe_overhead_seq_frac": 0.4996, "util_loss": 0.019041, "alpha": 0.03, "t": 734.2}
{"event": "consolidate", "step": 1200, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262114, "t": 745.5}
{"event": "probe", "step": 1200, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3351.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.192553, "cps": 26.179688, "mean_p": 0.228785, "boundary_entropy": 0.217793, "bpb": 2.281022}, "tr": {"chunks": 3633.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.200685, "cps": 28.382812, "mean_p": 0.235026, "boundary_entropy": 0.261874, "bpb": 2.558578}, "fi": {"chunks": 4671.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.248312, "cps": 36.492188, "mean_p": 0.272732, "boundary_entropy": 0.237461, "bpb": 2.738303}}, "bpb": {"en": 2.28102, "tr": 2.55858, "fi": 2.7383}, "gini_cpc": 0.057942, "gini_cps": 0.075504, "bpic": 4.86761, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0422323251150498, "fi": 1.2895781801986657}, "inventory": {"size": 326656, "admitted": 65536, "staged": 261120, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261120, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 33.84529113769531, "n_util_obs": 141, "util_count_spearman": -0.1402995425739388}, "t": 746.0}
{"event": "eval", "step": 1200, "phase": 1, "bpb": {"en": 2.29149, "tr": 2.56751, "fi": 2.74717}, "bytes_per_chunk": {"en": 4.977, "tr": 5.543, "fi": 4.282}, "bpb_mean": 2.53539, "frozen_inventory": true, "t": 746.9}
{"event": "train", "step": 1200, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.63388, "bpb": 2.35719, "chunk_loss": 0.030644, "F": 0.20255, "G": 0.24127, "chunks": 207.41, "bytes_per_chunk": 4.937, "boundary_entropy": 0.22448, "router_noise": 0.0, "grad_norm": 0.8134, "lr": 0.0002837, "router_trainable": true, "sec_per_step": 0.321, "probe_sec": 0.025, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.01826, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": 0.02346, "n_shadow": 8, "mean_delta_shadow": -0.12368, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": 0.99024, "trunc_abs_frac": 0.67228, "mean_delta_full": -0.01501, "read_hit_rate": {"surface": 0.48388}, "mean_gate": {"surface": 0.2297}, "n_reads": 3196, "n_cand": 6605, "read_suppressed_frac": 0.0964, "dropout_reads_logged": 3196, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 33.89405822753906, "n_util_obs": 141, "util_count_spearman": -0.14489093045739443}, "probe_overhead_sec_total": 187.1, "probe_overhead_seq_frac": 0.4996, "util_loss": 0.018768, "alpha": 0.03, "t": 750.2}
{"event": "consolidate", "step": 1250, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262114, "t": 763.5}
{"event": "train", "step": 1250, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.68889, "bpb": 2.43656, "chunk_loss": 0.030678, "F": 0.2045, "G": 0.23906, "chunks": 209.41, "bytes_per_chunk": 4.89, "boundary_entropy": 0.2265, "router_noise": 0.0, "grad_norm": 0.9121, "lr": 0.0002819, "router_trainable": true, "sec_per_step": 0.27, "probe_sec": 0.0269, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.0196, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 10, "mean_delta_loo": 0.35813, "n_shadow": 4, "mean_delta_shadow": 0.18527, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 1, "trunc_ratio_sum": 1.36875, "trunc_abs_frac": 1.36875, "mean_delta_full": 1.73804, "read_hit_rate": {"surface": 0.49798}, "mean_gate": {"surface": 0.23454}, "n_reads": 3321, "n_cand": 6669, "read_suppressed_frac": 0.097, "dropout_reads_logged": 3321, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 36.871124267578125, "n_util_obs": 160, "util_count_spearman": -0.10728356621616482}, "probe_overhead_sec_total": 189.552, "probe_overhead_seq_frac": 0.4996, "util_loss": 0.018436, "alpha": 0.03, "t": 763.7}
{"event": "consolidate", "step": 1300, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262116, "t": 775.5}
{"event": "train", "step": 1300, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.628, "bpb": 2.3487, "chunk_loss": 0.030641, "F": 0.20352, "G": 0.23685, "chunks": 208.41, "bytes_per_chunk": 4.913, "boundary_entropy": 0.2152, "router_noise": 0.0, "grad_norm": 0.9113, "lr": 0.00028, "router_trainable": true, "sec_per_step": 0.238, "probe_sec": 0.0273, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.01954, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": 0.07458, "n_shadow": 6, "mean_delta_shadow": -0.19392, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 31.56364, "trunc_abs_frac": 1.17475, "mean_delta_full": 0.00336, "read_hit_rate": {"surface": 0.5287}, "mean_gate": {"surface": 0.2335}, "n_reads": 3509, "n_cand": 6637, "read_suppressed_frac": 0.0977, "dropout_reads_logged": 3509, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 39.656463623046875, "n_util_obs": 178, "util_count_spearman": -0.0949594011456734}, "probe_overhead_sec_total": 191.784, "probe_overhead_seq_frac": 0.4996, "util_loss": 0.017991, "alpha": 0.03, "t": 775.6}
{"event": "consolidate", "step": 1350, "retired": 1, "split_retired": 0, "merge_admitted": 0, "admitted": 1, "swapped": 1024, "uncertain_staged": 262116, "t": 788.2}
{"event": "train", "step": 1350, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.61627, "bpb": 2.33179, "chunk_loss": 0.030822, "F": 0.21112, "G": 0.23853, "chunks": 216.19, "bytes_per_chunk": 4.737, "boundary_entropy": 0.20435, "router_noise": 0.0, "grad_norm": 0.9557, "lr": 0.000278, "router_trainable": true, "sec_per_step": 0.255, "probe_sec": 0.0292, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02186, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": 0.14443, "n_shadow": 7, "mean_delta_shadow": -0.14384, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 7, "trunc_ratio_sum": 0.16374, "trunc_abs_frac": 0.9318, "mean_delta_full": 0.12333, "read_hit_rate": {"surface": 0.54095}, "mean_gate": {"surface": 0.22361}, "n_reads": 3725, "n_cand": 6886, "read_suppressed_frac": 0.0996, "dropout_reads_logged": 3725, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 42.43305969238281, "n_util_obs": 184, "util_count_spearman": -0.06429472009280408}, "probe_overhead_sec_total": 194.468, "probe_overhead_seq_frac": 0.4996, "util_loss": 0.017976, "alpha": 0.03, "t": 788.4}
{"event": "consolidate", "step": 1400, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 103, "uncertain_staged": 262116, "t": 802.7}
{"event": "probe", "step": 1400, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3355.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.192783, "cps": 26.210938, "mean_p": 0.220877, "boundary_entropy": 0.196922, "bpb": 2.234689}, "tr": {"chunks": 3691.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.203889, "cps": 28.835938, "mean_p": 0.229155, "boundary_entropy": 0.233965, "bpb": 2.499811}, "fi": {"chunks": 4681.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.248844, "cps": 36.570312, "mean_p": 0.265763, "boundary_entropy": 0.216769, "bpb": 2.651314}}, "bpb": {"en": 2.23469, "tr": 2.49981, "fi": 2.65131}, "gini_cpc": 0.057898, "gini_cps": 0.075382, "bpic": 4.837725, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.057608882044647, "fi": 1.2907982067441435}, "inventory": {"size": 327577, "admitted": 65536, "staged": 262041, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262041, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 45.823974609375, "n_util_obs": 194, "util_count_spearman": -0.07481033213245748}, "t": 803.3}
{"event": "eval", "step": 1400, "phase": 1, "bpb": {"en": 2.24441, "tr": 2.51005, "fi": 2.66221}, "bytes_per_chunk": {"en": 4.912, "tr": 5.475, "fi": 4.264}, "bpb_mean": 2.47222, "frozen_inventory": true, "t": 804.1}
{"event": "train", "step": 1400, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.70061, "bpb": 2.45346, "chunk_loss": 0.030732, "F": 0.20859, "G": 0.23666, "chunks": 213.59, "bytes_per_chunk": 4.794, "boundary_entropy": 0.20558, "router_noise": 0.0, "grad_norm": 0.9296, "lr": 0.0002759, "router_trainable": true, "sec_per_step": 0.376, "probe_sec": 0.0255, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.01861, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 5, "mean_delta_loo": 0.0436, "n_shadow": 9, "mean_delta_shadow": -0.27644, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 1, "trunc_ratio_sum": 4.55155, "trunc_abs_frac": 4.55155, "mean_delta_full": 0.02368, "read_hit_rate": {"surface": 0.55123}, "mean_gate": {"surface": 0.24082}, "n_reads": 3750, "n_cand": 6803, "read_suppressed_frac": 0.0939, "dropout_reads_logged": 3750, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 45.881195068359375, "n_util_obs": 194, "util_count_spearman": -0.07494918292053691}, "probe_overhead_sec_total": 196.748, "probe_overhead_seq_frac": 0.4996, "util_loss": 0.018407, "alpha": 0.03, "t": 807.2}
{"event": "consolidate", "step": 1450, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262117, "t": 819.0}
{"event": "train", "step": 1450, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.65288, "bpb": 2.3846, "chunk_loss": 0.030765, "F": 0.21146, "G": 0.23728, "chunks": 216.53, "bytes_per_chunk": 4.729, "boundary_entropy": 0.2015, "router_noise": 0.0, "grad_norm": 0.821, "lr": 0.0002738, "router_trainable": true, "sec_per_step": 0.251, "probe_sec": 0.0269, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.0192, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": -0.23414, "n_shadow": 6, "mean_delta_shadow": -0.24475, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": 0.56072, "trunc_abs_frac": 0.51441, "mean_delta_full": -0.4248, "read_hit_rate": {"surface": 0.55401}, "mean_gate": {"surface": 0.23262}, "n_reads": 3821, "n_cand": 6897, "read_suppressed_frac": 0.1015, "dropout_reads_logged": 3821, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 48.75193786621094, "n_util_obs": 208, "util_count_spearman": -0.054561190144714174}, "probe_overhead_sec_total": 199.143, "probe_overhead_seq_frac": 0.4997, "util_loss": 0.018533, "alpha": 0.03, "t": 819.8}
{"event": "consolidate", "step": 1500, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262118, "t": 833.1}
{"event": "train", "step": 1500, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.61236, "bpb": 2.32615, "chunk_loss": 0.030539, "F": 0.20215, "G": 0.22621, "chunks": 207.0, "bytes_per_chunk": 4.947, "boundary_entropy": 0.18788, "router_noise": 0.0, "grad_norm": 0.9229, "lr": 0.0002715, "router_trainable": true, "sec_per_step": 0.27, "probe_sec": 0.0298, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02302, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 10, "mean_delta_loo": 0.08312, "n_shadow": 4, "mean_delta_shadow": -0.19723, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 0, "read_hit_rate": {"surface": 0.54308}, "mean_gate": {"surface": 0.24097}, "n_reads": 3580, "n_cand": 6592, "read_suppressed_frac": 0.1025, "dropout_reads_logged": 3580, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 51.33393859863281, "n_util_obs": 220, "util_count_spearman": -0.04816554666763029}, "probe_overhead_sec_total": 201.452, "probe_overhead_seq_frac": 0.4997, "util_loss": 0.020029, "alpha": 0.03, "t": 833.2}
{"event": "consolidate", "step": 1550, "retired": 1, "split_retired": 0, "merge_admitted": 0, "admitted": 1, "swapped": 1024, "uncertain_staged": 262118, "t": 846.6}
{"event": "train", "step": 1550, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.57388, "bpb": 2.27063, "chunk_loss": 0.030568, "F": 0.20358, "G": 0.22436, "chunks": 208.47, "bytes_per_chunk": 4.912, "boundary_entropy": 0.18139, "router_noise": 0.0, "grad_norm": 0.778, "lr": 0.0002692, "router_trainable": true, "sec_per_step": 0.272, "probe_sec": 0.0383, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.03092, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": 0.00687, "n_shadow": 6, "mean_delta_shadow": 0.00718, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 6, "trunc_ratio_sum": 0.4727, "trunc_abs_frac": 0.9223, "mean_delta_full": 0.39408, "read_hit_rate": {"surface": 0.5418}, "mean_gate": {"surface": 0.2322}, "n_reads": 3597, "n_cand": 6639, "read_suppressed_frac": 0.1009, "dropout_reads_logged": 3597, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 53.53639221191406, "n_util_obs": 224, "util_count_spearman": -0.10580509598821494}, "probe_overhead_sec_total": 203.864, "probe_overhead_seq_frac": 0.4997, "util_loss": 0.019507, "alpha": 0.03, "t": 846.8}
{"event": "consolidate", "step": 1600, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262118, "t": 858.4}
{"event": "probe", "step": 1600, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3346.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.192266, "cps": 26.140625, "mean_p": 0.214388, "boundary_entropy": 0.176266, "bpb": 2.189928}, "tr": {"chunks": 3745.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.206872, "cps": 29.257812, "mean_p": 0.220243, "boundary_entropy": 0.209054, "bpb": 2.43115}, "fi": {"chunks": 4730.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.251449, "cps": 36.953125, "mean_p": 0.263218, "boundary_entropy": 0.189613, "bpb": 2.584796}}, "bpb": {"en": 2.18993, "tr": 2.43115, "fi": 2.5848}, "gini_cpc": 0.060646, "gini_cps": 0.078053, "bpic": 4.799256, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0759682449359396, "fi": 1.3078183928131735}, "inventory": {"size": 326656, "admitted": 65536, "staged": 261120, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261120, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 55.88311767578125, "n_util_obs": 230, "util_count_spearman": -0.11112127541589499}, "t": 859.0}
{"event": "eval", "step": 1600, "phase": 1, "bpb": {"en": 2.2009, "tr": 2.44038, "fi": 2.59502}, "bytes_per_chunk": {"en": 4.937, "tr": 5.347, "fi": 4.242}, "bpb_mean": 2.4121, "frozen_inventory": true, "t": 860.0}
{"event": "train", "step": 1600, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.57982, "bpb": 2.27919, "chunk_loss": 0.030649, "F": 0.20685, "G": 0.22729, "chunks": 211.81, "bytes_per_chunk": 4.834, "boundary_entropy": 0.17851, "router_noise": 0.0, "grad_norm": 0.8248, "lr": 0.0002668, "router_trainable": true, "sec_per_step": 0.333, "probe_sec": 0.0402, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.0293, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": 0.00011, "n_shadow": 5, "mean_delta_shadow": -0.09115, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 0.40019, "trunc_abs_frac": 1.32791, "mean_delta_full": -0.06482, "read_hit_rate": {"surface": 0.55425}, "mean_gate": {"surface": 0.24404}, "n_reads": 3739, "n_cand": 6746, "read_suppressed_frac": 0.1046, "dropout_reads_logged": 3739, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 55.94017028808594, "n_util_obs": 230, "util_count_spearman": -0.11154143654176911}, "probe_overhead_sec_total": 206.243, "probe_overhead_seq_frac": 0.4997, "util_loss": 0.020889, "alpha": 0.03, "t": 863.4}
{"event": "consolidate", "step": 1650, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262118, "t": 876.2}
{"event": "train", "step": 1650, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.5998, "bpb": 2.30802, "chunk_loss": 0.030561, "F": 0.20377, "G": 0.22437, "chunks": 208.66, "bytes_per_chunk": 4.908, "boundary_entropy": 0.17512, "router_noise": 0.0, "grad_norm": 0.813, "lr": 0.0002643, "router_trainable": true, "sec_per_step": 0.273, "probe_sec": 0.0313, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.01836, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": 0.36048, "n_shadow": 7, "mean_delta_shadow": 0.07671, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 0.52846, "trunc_abs_frac": 0.9745, "mean_delta_full": -0.10376, "read_hit_rate": {"surface": 0.55094}, "mean_gate": {"surface": 0.24007}, "n_reads": 3661, "n_cand": 6645, "read_suppressed_frac": 0.1033, "dropout_reads_logged": 3661, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 58.833740234375, "n_util_obs": 242, "util_count_spearman": -0.09283292951559165}, "probe_overhead_sec_total": 208.524, "probe_overhead_seq_frac": 0.4997, "util_loss": 0.019342, "alpha": 0.03, "t": 877.1}
{"event": "consolidate", "step": 1700, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262118, "t": 889.1}
{"event": "train", "step": 1700, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.61893, "bpb": 2.33562, "chunk_loss": 0.030604, "F": 0.20584, "G": 0.22635, "chunks": 210.78, "bytes_per_chunk": 4.858, "boundary_entropy": 0.17665, "router_noise": 0.0, "grad_norm": 0.7727, "lr": 0.0002618, "router_trainable": true, "sec_per_step": 0.244, "probe_sec": 0.0348, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02642, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 10, "mean_delta_loo": -0.00231, "n_shadow": 4, "mean_delta_shadow": -0.05832, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 2.84302, "trunc_abs_frac": 0.39467, "mean_delta_full": 0.01575, "read_hit_rate": {"surface": 0.55117}, "mean_gate": {"surface": 0.24747}, "n_reads": 3700, "n_cand": 6713, "read_suppressed_frac": 0.0922, "dropout_reads_logged": 3700, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 61.75782775878906, "n_util_obs": 248, "util_count_spearman": -0.07762371797583882}, "probe_overhead_sec_total": 211.161, "probe_overhead_seq_frac": 0.4997, "util_loss": 0.019797, "alpha": 0.03, "t": 889.3}
{"event": "consolidate", "step": 1750, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262118, "t": 902.5}
{"event": "train", "step": 1750, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.56455, "bpb": 2.25716, "chunk_loss": 0.030477, "F": 0.20212, "G": 0.21989, "chunks": 206.97, "bytes_per_chunk": 4.948, "boundary_entropy": 0.17288, "router_noise": 0.0, "grad_norm": 0.8098, "lr": 0.0002591, "router_trainable": true, "sec_per_step": 0.27, "probe_sec": 0.0262, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.01853, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": -0.01069, "n_shadow": 6, "mean_delta_shadow": 0.12436, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": 0.88092, "trunc_abs_frac": 1.0082, "mean_delta_full": -0.18406, "read_hit_rate": {"surface": 0.56456}, "mean_gate": {"surface": 0.2386}, "n_reads": 3721, "n_cand": 6591, "read_suppressed_frac": 0.1016, "dropout_reads_logged": 3721, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 64.63235473632812, "n_util_obs": 256, "util_count_spearman": -0.0756663385378272}, "probe_overhead_sec_total": 213.716, "probe_overhead_seq_frac": 0.4997, "util_loss": 0.019436, "alpha": 0.03, "t": 902.8}
{"event": "consolidate", "step": 1800, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262118, "t": 914.2}
{"event": "probe", "step": 1800, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3357.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.192898, "cps": 26.226562, "mean_p": 0.209452, "boundary_entropy": 0.164906, "bpb": 2.149588}, "tr": {"chunks": 3764.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.207921, "cps": 29.40625, "mean_p": 0.217261, "boundary_entropy": 0.200282, "bpb": 2.390486}, "fi": {"chunks": 4851.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.257881, "cps": 37.898438, "mean_p": 0.263719, "boundary_entropy": 0.180151, "bpb": 2.515435}}, "bpb": {"en": 2.14959, "tr": 2.39049, "fi": 2.51544}, "gini_cpc": 0.065769, "gini_cps": 0.083194, "bpic": 4.738724, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0778835456350284, "fi": 1.3368792117668873}, "inventory": {"size": 326656, "admitted": 65536, "staged": 261120, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261120, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 67.33433532714844, "n_util_obs": 266, "util_count_spearman": -0.08104367425111422}, "t": 914.7}
{"event": "eval", "step": 1800, "phase": 1, "bpb": {"en": 2.16109, "tr": 2.39866, "fi": 2.52563}, "bytes_per_chunk": {"en": 4.96, "tr": 5.382, "fi": 4.141}, "bpb_mean": 2.3618, "frozen_inventory": true, "t": 915.7}
{"event": "train", "step": 1800, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.53132, "bpb": 2.20923, "chunk_loss": 0.030442, "F": 0.19949, "G": 0.2166, "chunks": 204.28, "bytes_per_chunk": 5.013, "boundary_entropy": 0.16531, "router_noise": 0.0, "grad_norm": 0.7519, "lr": 0.0002564, "router_trainable": true, "sec_per_step": 0.324, "probe_sec": 0.0785, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.06055, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 13, "mean_delta_loo": 0.24248, "n_shadow": 1, "mean_delta_shadow": 0.69925, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 1.22982, "trunc_abs_frac": 0.91827, "mean_delta_full": 0.2049, "read_hit_rate": {"surface": 0.55265}, "mean_gate": {"surface": 0.24638}, "n_reads": 3595, "n_cand": 6505, "read_suppressed_frac": 0.1001, "dropout_reads_logged": 3595, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 67.38919067382812, "n_util_obs": 266, "util_count_spearman": -0.08016888532689209}, "probe_overhead_sec_total": 216.235, "probe_overhead_seq_frac": 0.4997, "util_loss": 0.021408, "alpha": 0.03, "t": 919.0}
{"event": "consolidate", "step": 1850, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262118, "t": 931.2}
{"event": "train", "step": 1850, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.57002, "bpb": 2.26506, "chunk_loss": 0.030601, "F": 0.20874, "G": 0.22444, "chunks": 213.75, "bytes_per_chunk": 4.791, "boundary_entropy": 0.15958, "router_noise": 0.0, "grad_norm": 0.8513, "lr": 0.0002537, "router_trainable": true, "sec_per_step": 0.246, "probe_sec": 0.0259, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.01871, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": -0.07097, "n_shadow": 7, "mean_delta_shadow": -0.20664, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 7, "trunc_ratio_sum": 0.79408, "trunc_abs_frac": 0.66438, "mean_delta_full": -0.33412, "read_hit_rate": {"surface": 0.56683}, "mean_gate": {"surface": 0.23174}, "n_reads": 3859, "n_cand": 6808, "read_suppressed_frac": 0.0868, "dropout_reads_logged": 3859, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 70.25450134277344, "n_util_obs": 276, "util_count_spearman": -0.060355614370069186}, "probe_overhead_sec_total": 219.025, "probe_overhead_seq_frac": 0.4997, "util_loss": 0.019393, "alpha": 0.03, "t": 931.3}
{"event": "consolidate", "step": 1900, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262118, "t": 942.6}
{"event": "train", "step": 1900, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.50363, "bpb": 2.16929, "chunk_loss": 0.030539, "F": 0.20514, "G": 0.22046, "chunks": 210.06, "bytes_per_chunk": 4.875, "boundary_entropy": 0.15464, "router_noise": 0.0, "grad_norm": 0.7193, "lr": 0.0002508, "router_trainable": true, "sec_per_step": 0.229, "probe_sec": 0.0431, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.03559, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 4, "mean_delta_loo": -0.24866, "n_shadow": 10, "mean_delta_shadow": 0.00368, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 0.57142, "trunc_abs_frac": 0.90053, "mean_delta_full": -0.33089, "read_hit_rate": {"surface": 0.57115}, "mean_gate": {"surface": 0.25222}, "n_reads": 3821, "n_cand": 6690, "read_suppressed_frac": 0.1018, "dropout_reads_logged": 3821, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 73.19798278808594, "n_util_obs": 292, "util_count_spearman": -0.03677703915494091}, "probe_overhead_sec_total": 221.326, "probe_overhead_seq_frac": 0.4997, "util_loss": 0.019448, "alpha": 0.03, "t": 942.8}
{"event": "consolidate", "step": 1950, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262118, "t": 956.7}
{"event": "train", "step": 1950, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.57683, "bpb": 2.27489, "chunk_loss": 0.030707, "F": 0.21106, "G": 0.22735, "chunks": 216.12, "bytes_per_chunk": 4.738, "boundary_entropy": 0.15749, "router_noise": 0.0, "grad_norm": 0.7528, "lr": 0.0002479, "router_trainable": true, "sec_per_step": 0.281, "probe_sec": 0.0251, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.0182, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": -0.19456, "n_shadow": 7, "mean_delta_shadow": -0.01056, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": 0.64428, "trunc_abs_frac": 1.11637, "mean_delta_full": 0.24409, "read_hit_rate": {"surface": 0.57597}, "mean_gate": {"surface": 0.25129}, "n_reads": 3965, "n_cand": 6884, "read_suppressed_frac": 0.1057, "dropout_reads_logged": 3965, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 75.92597961425781, "n_util_obs": 299, "util_count_spearman": -0.057608615748598185}, "probe_overhead_sec_total": 223.769, "probe_overhead_seq_frac": 0.4997, "util_loss": 0.019087, "alpha": 0.03, "t": 956.8}
{"event": "consolidate", "step": 2000, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262119, "t": 967.8}
{"event": "probe", "step": 2000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3370.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.193645, "cps": 26.328125, "mean_p": 0.208812, "boundary_entropy": 0.151467, "bpb": 2.125866}, "tr": {"chunks": 3799.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.209855, "cps": 29.679688, "mean_p": 0.214978, "boundary_entropy": 0.179692, "bpb": 2.341861}, "fi": {"chunks": 4817.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.256074, "cps": 37.632812, "mean_p": 0.261341, "boundary_entropy": 0.164984, "bpb": 2.476434}}, "bpb": {"en": 2.12587, "tr": 2.34186, "fi": 2.47643}, "gini_cpc": 0.0631, "gini_cps": 0.080483, "bpic": 4.733189, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0837097020330908, "fi": 1.3223882515864904}, "inventory": {"size": 326656, "admitted": 65536, "staged": 261120, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261120, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 78.84596252441406, "n_util_obs": 316, "util_count_spearman": -0.030145488834536348}, "t": 968.3}
{"event": "eval", "step": 2000, "phase": 1, "bpb": {"en": 2.13682, "tr": 2.35174, "fi": 2.48955}, "bytes_per_chunk": {"en": 4.91, "tr": 5.296, "fi": 4.173}, "bpb_mean": 2.32604, "frozen_inventory": true, "t": 969.4}
{"event": "train", "step": 2000, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.54271, "bpb": 2.22566, "chunk_loss": 0.03067, "F": 0.21371, "G": 0.22631, "chunks": 218.84, "bytes_per_chunk": 4.679, "boundary_entropy": 0.1499, "router_noise": 0.0, "grad_norm": 0.8537, "lr": 0.000245, "router_trainable": true, "sec_per_step": 0.327, "probe_sec": 0.0589, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02324, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": 0.01011, "n_shadow": 7, "mean_delta_shadow": 0.06982, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": -3.19874, "trunc_abs_frac": 0.65238, "mean_delta_full": 0.01901, "read_hit_rate": {"surface": 0.57581}, "mean_gate": {"surface": 0.24325}, "n_reads": 4014, "n_cand": 6971, "read_suppressed_frac": 0.0997, "dropout_reads_logged": 4014, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 78.90721130371094, "n_util_obs": 316, "util_count_spearman": -0.030102895420428915}, "probe_overhead_sec_total": 226.353, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.024103, "alpha": 0.03, "t": 973.1}
{"event": "consolidate", "step": 2050, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 851, "uncertain_staged": 262119, "t": 985.0}
{"event": "train", "step": 2050, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.55763, "bpb": 2.24718, "chunk_loss": 0.030395, "F": 0.19693, "G": 0.21606, "chunks": 201.66, "bytes_per_chunk": 5.078, "boundary_entropy": 0.15838, "router_noise": 0.0, "grad_norm": 0.7853, "lr": 0.0002419, "router_trainable": true, "sec_per_step": 0.24, "probe_sec": 0.0259, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.01832, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": 0.23693, "n_shadow": 5, "mean_delta_shadow": -0.21141, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": 0.69986, "trunc_abs_frac": 2.04679, "mean_delta_full": 0.05393, "read_hit_rate": {"surface": 0.57094}, "mean_gate": {"surface": 0.26936}, "n_reads": 3666, "n_cand": 6421, "read_suppressed_frac": 0.1031, "dropout_reads_logged": 3666, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 82.29234313964844, "n_util_obs": 324, "util_count_spearman": -0.008029604237201329}, "probe_overhead_sec_total": 228.785, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.022384, "alpha": 0.03, "t": 985.1}
{"event": "consolidate", "step": 2100, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1024, "uncertain_staged": 262120, "t": 999.3}
{"event": "train", "step": 2100, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.47909, "bpb": 2.13387, "chunk_loss": 0.030394, "F": 0.19916, "G": 0.21219, "chunks": 203.94, "bytes_per_chunk": 5.021, "boundary_entropy": 0.14662, "router_noise": 0.0, "grad_norm": 0.7276, "lr": 0.0002388, "router_trainable": true, "sec_per_step": 0.292, "probe_sec": 0.1201, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.11016, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 11, "mean_delta_loo": 0.04619, "n_shadow": 3, "mean_delta_shadow": 0.14451, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 1.27381, "trunc_abs_frac": 0.65012, "mean_delta_full": 0.11344, "read_hit_rate": {"surface": 0.57438}, "mean_gate": {"surface": 0.25241}, "n_reads": 3730, "n_cand": 6494, "read_suppressed_frac": 0.1054, "dropout_reads_logged": 3730, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 85.24806213378906, "n_util_obs": 331, "util_count_spearman": -0.0028916003122407092}, "probe_overhead_sec_total": 231.414, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.020801, "alpha": 0.03, "t": 999.7}
{"event": "consolidate", "step": 2150, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 684, "uncertain_staged": 262120, "t": 1010.8}
{"event": "train", "step": 2150, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.48822, "bpb": 2.14705, "chunk_loss": 0.030593, "F": 0.20944, "G": 0.22091, "chunks": 214.47, "bytes_per_chunk": 4.775, "boundary_entropy": 0.1413, "router_noise": 0.0, "grad_norm": 0.7545, "lr": 0.0002357, "router_trainable": true, "sec_per_step": 0.226, "probe_sec": 0.0294, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02238, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": 0.00288, "n_shadow": 5, "mean_delta_shadow": -0.05474, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 0.33688, "trunc_abs_frac": 0.45298, "mean_delta_full": 0.19513, "read_hit_rate": {"surface": 0.58337}, "mean_gate": {"surface": 0.24252}, "n_reads": 3985, "n_cand": 6831, "read_suppressed_frac": 0.1044, "dropout_reads_logged": 3985, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 88.2342529296875, "n_util_obs": 337, "util_count_spearman": 0.010804494214330742}, "probe_overhead_sec_total": 233.588, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.023073, "alpha": 0.03, "t": 1011.0}
{"event": "consolidate", "step": 2200, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 81, "uncertain_staged": 262120, "t": 1024.4}
{"event": "probe", "step": 2200, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3336.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.191691, "cps": 26.0625, "mean_p": 0.201487, "boundary_entropy": 0.143172, "bpb": 2.098016}, "tr": {"chunks": 3751.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.207203, "cps": 29.304688, "mean_p": 0.209713, "boundary_entropy": 0.174408, "bpb": 2.305718}, "fi": {"chunks": 4781.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.25416, "cps": 37.351562, "mean_p": 0.256962, "boundary_entropy": 0.155404, "bpb": 2.45381}}, "bpb": {"en": 2.09802, "tr": 2.30572, "fi": 2.45381}, "gini_cpc": 0.063771, "gini_cps": 0.081171, "bpic": 4.780249, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0809225844756445, "fi": 1.3258821946748591}, "inventory": {"size": 327599, "admitted": 65536, "staged": 262063, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262063, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 91.12898254394531, "n_util_obs": 349, "util_count_spearman": 0.01198625713986182}, "t": 1025.0}
{"event": "eval", "step": 2200, "phase": 1, "bpb": {"en": 2.11047, "tr": 2.31307, "fi": 2.46601}, "bytes_per_chunk": {"en": 4.944, "tr": 5.337, "fi": 4.205}, "bpb_mean": 2.29652, "frozen_inventory": true, "t": 1026.0}
{"event": "train", "step": 2200, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.5333, "bpb": 2.21209, "chunk_loss": 0.030644, "F": 0.2084, "G": 0.2186, "chunks": 213.41, "bytes_per_chunk": 4.798, "boundary_entropy": 0.14576, "router_noise": 0.0, "grad_norm": 0.8024, "lr": 0.0002325, "router_trainable": true, "sec_per_step": 0.365, "probe_sec": 0.0375, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.03005, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": -0.05922, "n_shadow": 7, "mean_delta_shadow": -0.04539, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": 0.72594, "trunc_abs_frac": 1.3689, "mean_delta_full": -0.16538, "read_hit_rate": {"surface": 0.57775}, "mean_gate": {"surface": 0.24266}, "n_reads": 3927, "n_cand": 6797, "read_suppressed_frac": 0.1049, "dropout_reads_logged": 3927, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 91.18890380859375, "n_util_obs": 349, "util_count_spearman": 0.012055703444743247}, "probe_overhead_sec_total": 236.321, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.024884, "alpha": 0.03, "t": 1029.3}
{"event": "consolidate", "step": 2250, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 182, "uncertain_staged": 262120, "t": 1041.6}
{"event": "train", "step": 2250, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.50493, "bpb": 2.17115, "chunk_loss": 0.030465, "F": 0.20355, "G": 0.21416, "chunks": 208.44, "bytes_per_chunk": 4.913, "boundary_entropy": 0.13354, "router_noise": 0.0, "grad_norm": 0.7837, "lr": 0.0002293, "router_trainable": true, "sec_per_step": 0.251, "probe_sec": 0.0274, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.01944, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": -0.04288, "n_shadow": 8, "mean_delta_shadow": -0.32804, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 1.15828, "trunc_abs_frac": 0.5778, "mean_delta_full": -1.09802, "read_hit_rate": {"surface": 0.57909}, "mean_gate": {"surface": 0.23659}, "n_reads": 3844, "n_cand": 6638, "read_suppressed_frac": 0.0986, "dropout_reads_logged": 3844, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 94.20938110351562, "n_util_obs": 361, "util_count_spearman": 0.003696507830593281}, "probe_overhead_sec_total": 238.662, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.025206, "alpha": 0.03, "t": 1041.9}
{"event": "consolidate", "step": 2300, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 7, "uncertain_staged": 262120, "t": 1059.4}
{"event": "train", "step": 2300, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.54162, "bpb": 2.22408, "chunk_loss": 0.030571, "F": 0.20746, "G": 0.21744, "chunks": 212.44, "bytes_per_chunk": 4.82, "boundary_entropy": 0.13523, "router_noise": 0.0, "grad_norm": 0.7088, "lr": 0.000226, "router_trainable": true, "sec_per_step": 0.355, "probe_sec": 0.037, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02574, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": 0.17948, "n_shadow": 5, "mean_delta_shadow": 0.10703, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 0.15185, "trunc_abs_frac": 0.3021, "mean_delta_full": -0.24789, "read_hit_rate": {"surface": 0.58321}, "mean_gate": {"surface": 0.25011}, "n_reads": 3946, "n_cand": 6766, "read_suppressed_frac": 0.0968, "dropout_reads_logged": 3946, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 97.25630187988281, "n_util_obs": 366, "util_count_spearman": 0.011850570031950981}, "probe_overhead_sec_total": 241.153, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.021158, "alpha": 0.03, "t": 1059.6}
{"event": "consolidate", "step": 2350, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 1, "uncertain_staged": 262120, "t": 1077.3}
{"event": "train", "step": 2350, "arch": "ali", "phase": 1, "objective": "global", "lm_loss": 1.46454, "bpb": 2.11289, "chunk_loss": 0.03048, "F": 0.20386, "G": 0.21516, "chunks": 208.75, "bytes_per_chunk": 4.905, "boundary_entropy": 0.12866, "router_noise": 0.0, "grad_norm": 0.7677, "lr": 0.0002226, "router_trainable": true, "sec_per_step": 0.376, "probe_sec": 0.0466, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02468, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 11, "mean_delta_loo": 0.23525, "n_shadow": 3, "mean_delta_shadow": 0.119, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 25.81681, "trunc_abs_frac": 0.53966, "mean_delta_full": 0.00708, "read_hit_rate": {"surface": 0.57912}, "mean_gate": {"surface": 0.23917}, "n_reads": 3850, "n_cand": 6648, "read_suppressed_frac": 0.0992, "dropout_reads_logged": 3850, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 100.26416015625, "n_util_obs": 378, "util_count_spearman": 0.015054172447107423}, "probe_overhead_sec_total": 244.208, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.022791, "alpha": 0.03, "t": 1078.4}
{"event": "consolidate", "step": 2400, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 6, "uncertain_staged": 262120, "t": 1096.9}
{"event": "probe", "step": 2400, "phase": 2, "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": 0.203511, "boundary_entropy": 0.12841, "bpb": 2.069266}, "tr": {"chunks": 3806.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.210241, "cps": 29.734375, "mean_p": 0.208713, "boundary_entropy": 0.155434, "bpb": 2.265391}, "fi": {"chunks": 4863.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.258519, "cps": 37.992188, "mean_p": 0.260918, "boundary_entropy": 0.157853, "bpb": 2.396967}}, "bpb": {"en": 2.06927, "tr": 2.26539, "fi": 2.39697}, "gini_cpc": 0.066373, "gini_cps": 0.083735, "bpic": 4.719015, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0912111608954551, "fi": 1.341785062940199}, "inventory": {"size": 327674, "admitted": 65536, "staged": 262138, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262138, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 103.21324157714844, "n_util_obs": 389, "util_count_spearman": 0.04870151381498724}, "t": 1097.7}
{"event": "eval", "step": 2400, "phase": 1, "bpb": {"en": 2.08208, "tr": 2.27232, "fi": 2.4104}, "bytes_per_chunk": {"en": 4.939, "tr": 5.255, "fi": 4.155}, "bpb_mean": 2.25493, "frozen_inventory": true, "t": 1098.6}
{"event": "phase_transition", "step": 2400, "phase": 2, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0, "t": 1103.0}
{"event": "train", "step": 2400, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.46392, "bpb": 2.11199, "chunk_loss": 0.030435, "F": 0.20294, "G": 0.21366, "chunks": 207.81, "bytes_per_chunk": 4.928, "boundary_entropy": 0.13111, "router_noise": 0.0, "grad_norm": 0.6588, "lr": 0.0002192, "router_trainable": true, "sec_per_step": 0.515, "probe_sec": 0.075, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.05064, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": 0.12409, "n_shadow": 6, "mean_delta_shadow": -0.09397, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 1.01964, "trunc_abs_frac": 0.91421, "mean_delta_full": -0.26619, "read_hit_rate": {"surface": 0.58976}, "mean_gate": {"surface": 0.25086}, "n_reads": 3903, "n_cand": 6618, "read_suppressed_frac": 0.1002, "dropout_reads_logged": 3903, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 103.27279663085938, "n_util_obs": 390, "util_count_spearman": 0.04256439532482693}, "probe_overhead_sec_total": 246.768, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.025475, "alpha": 0.03, "t": 1104.2}
{"event": "consolidate", "step": 2450, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 24, "uncertain_staged": 262120, "t": 1127.0}
{"event": "train", "step": 2450, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.47578, "bpb": 2.1291, "chunk_loss": 0.0307, "F": 0.21405, "G": 0.22407, "chunks": 219.19, "bytes_per_chunk": 4.672, "boundary_entropy": 0.13121, "router_noise": 0.0, "grad_norm": 0.8155, "lr": 0.0002158, "router_trainable": true, "sec_per_step": 0.461, "probe_sec": 0.0396, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02907, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": 0.17059, "n_shadow": 7, "mean_delta_shadow": -0.25106, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 0.72093, "trunc_abs_frac": 0.77122, "mean_delta_full": 0.03902, "read_hit_rate": {"surface": 0.60183}, "mean_gate": {"surface": 0.24868}, "n_reads": 4202, "n_cand": 6982, "read_suppressed_frac": 0.0957, "dropout_reads_logged": 4202, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 106.29208374023438, "n_util_obs": 400, "util_count_spearman": 0.05632480801163073}, "probe_overhead_sec_total": 249.298, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.022362, "alpha": 0.03, "t": 1127.2}
{"event": "consolidate", "step": 2500, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 34, "uncertain_staged": 262120, "t": 1150.5}
{"event": "train", "step": 2500, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.57166, "bpb": 2.26743, "chunk_loss": 0.030613, "F": 0.21082, "G": 0.21998, "chunks": 215.88, "bytes_per_chunk": 4.743, "boundary_entropy": 0.12918, "router_noise": 0.0, "grad_norm": 0.7082, "lr": 0.0002123, "router_trainable": true, "sec_per_step": 0.47, "probe_sec": 0.0287, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.01821, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": 0.24159, "n_shadow": 7, "mean_delta_shadow": 0.05355, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 0.90209, "trunc_abs_frac": 0.71001, "mean_delta_full": 0.36957, "read_hit_rate": {"surface": 0.5749}, "mean_gate": {"surface": 0.24702}, "n_reads": 3953, "n_cand": 6876, "read_suppressed_frac": 0.0989, "dropout_reads_logged": 3953, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 109.34130859375, "n_util_obs": 410, "util_count_spearman": 0.0709427948450597}, "probe_overhead_sec_total": 252.076, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.021072, "alpha": 0.03, "t": 1150.7}
{"event": "consolidate", "step": 2550, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 3, "uncertain_staged": 262120, "t": 1175.3}
{"event": "train", "step": 2550, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.496, "bpb": 2.15827, "chunk_loss": 0.030471, "F": 0.20535, "G": 0.21707, "chunks": 210.28, "bytes_per_chunk": 4.87, "boundary_entropy": 0.12599, "router_noise": 0.0, "grad_norm": 0.7176, "lr": 0.0002088, "router_trainable": true, "sec_per_step": 0.498, "probe_sec": 0.0535, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.03741, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 11, "mean_delta_loo": 0.17793, "n_shadow": 3, "mean_delta_shadow": -0.03541, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 1.83532, "trunc_abs_frac": 0.26151, "mean_delta_full": 0.09998, "read_hit_rate": {"surface": 0.58982}, "mean_gate": {"surface": 0.24802}, "n_reads": 3950, "n_cand": 6697, "read_suppressed_frac": 0.0914, "dropout_reads_logged": 3950, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 112.34002685546875, "n_util_obs": 416, "util_count_spearman": 0.07094387572151524}, "probe_overhead_sec_total": 254.644, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.022155, "alpha": 0.03, "t": 1175.6}
{"event": "consolidate", "step": 2600, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 128, "uncertain_staged": 262120, "t": 1188.5}
{"event": "probe", "step": 2600, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3377.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.194047, "cps": 26.382812, "mean_p": 0.202735, "boundary_entropy": 0.118658, "bpb": 2.048918}, "tr": {"chunks": 3805.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.210186, "cps": 29.726562, "mean_p": 0.208784, "boundary_entropy": 0.146225, "bpb": 2.245851}, "fi": {"chunks": 4855.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.258094, "cps": 37.929688, "mean_p": 0.256251, "boundary_entropy": 0.132598, "bpb": 2.355568}}, "bpb": {"en": 2.04892, "tr": 2.24585, "fi": 2.35557}, "gini_cpc": 0.064466, "gini_cps": 0.081859, "bpic": 4.713135, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0831713621873296, "fi": 1.3300574813780839}, "inventory": {"size": 327552, "admitted": 65536, "staged": 262016, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262016, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 115.27001953125, "n_util_obs": 425, "util_count_spearman": 0.06752303143671312}, "t": 1189.1}
{"event": "eval", "step": 2600, "phase": 2, "bpb": {"en": 2.06078, "tr": 2.25325, "fi": 2.36753}, "bytes_per_chunk": {"en": 4.899, "tr": 5.309, "fi": 4.135}, "bpb_mean": 2.22718, "frozen_inventory": true, "t": 1190.3}
{"event": "train", "step": 2600, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.50391, "bpb": 2.16968, "chunk_loss": 0.030639, "F": 0.21152, "G": 0.21956, "chunks": 216.59, "bytes_per_chunk": 4.728, "boundary_entropy": 0.11885, "router_noise": 0.0, "grad_norm": 0.6876, "lr": 0.0002053, "router_trainable": true, "sec_per_step": 0.37, "probe_sec": 0.0299, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02267, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": 0.17184, "n_shadow": 6, "mean_delta_shadow": -0.18133, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 0.9235, "trunc_abs_frac": 1.76719, "mean_delta_full": 0.37778, "read_hit_rate": {"surface": 0.59893}, "mean_gate": {"surface": 0.24507}, "n_reads": 4132, "n_cand": 6899, "read_suppressed_frac": 0.099, "dropout_reads_logged": 4132, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 115.33306884765625, "n_util_obs": 425, "util_count_spearman": 0.06908616191050801}, "probe_overhead_sec_total": 256.952, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.021582, "alpha": 0.03, "t": 1194.1}
{"event": "consolidate", "step": 2650, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 315, "uncertain_staged": 262120, "t": 1215.7}
{"event": "train", "step": 2650, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.43918, "bpb": 2.07629, "chunk_loss": 0.030372, "F": 0.19958, "G": 0.20831, "chunks": 204.38, "bytes_per_chunk": 5.01, "boundary_entropy": 0.1247, "router_noise": 0.0, "grad_norm": 0.6841, "lr": 0.0002017, "router_trainable": true, "sec_per_step": 0.436, "probe_sec": 0.0694, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.06263, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 10, "mean_delta_loo": 0.42563, "n_shadow": 4, "mean_delta_shadow": -0.02268, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 0.90441, "trunc_abs_frac": 0.90299, "mean_delta_full": 0.64905, "read_hit_rate": {"surface": 0.57606}, "mean_gate": {"surface": 0.23814}, "n_reads": 3749, "n_cand": 6508, "read_suppressed_frac": 0.104, "dropout_reads_logged": 3749, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 118.31172180175781, "n_util_obs": 435, "util_count_spearman": 0.09270934290156571}, "probe_overhead_sec_total": 259.669, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.023819, "alpha": 0.03, "t": 1215.9}
{"event": "consolidate", "step": 2700, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 108, "uncertain_staged": 262120, "t": 1240.6}
{"event": "train", "step": 2700, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.51792, "bpb": 2.18989, "chunk_loss": 0.030546, "F": 0.20679, "G": 0.21591, "chunks": 211.75, "bytes_per_chunk": 4.836, "boundary_entropy": 0.12451, "router_noise": 0.0, "grad_norm": 0.7871, "lr": 0.0001981, "router_trainable": true, "sec_per_step": 0.498, "probe_sec": 0.0241, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.01787, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 5, "mean_delta_loo": -0.14167, "n_shadow": 9, "mean_delta_shadow": -0.10065, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 1, "trunc_ratio_sum": 0.83313, "trunc_abs_frac": 0.83313, "mean_delta_full": 0.20337, "read_hit_rate": {"surface": 0.58704}, "mean_gate": {"surface": 0.25418}, "n_reads": 3959, "n_cand": 6744, "read_suppressed_frac": 0.1015, "dropout_reads_logged": 3959, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 121.29966735839844, "n_util_obs": 444, "util_count_spearman": 0.09545820132688831}, "probe_overhead_sec_total": 262.407, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.02453, "alpha": 0.03, "t": 1240.8}
{"event": "consolidate", "step": 2750, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 326, "uncertain_staged": 262120, "t": 1260.2}
{"event": "train", "step": 2750, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.45761, "bpb": 2.10288, "chunk_loss": 0.030501, "F": 0.20752, "G": 0.21613, "chunks": 212.5, "bytes_per_chunk": 4.819, "boundary_entropy": 0.1177, "router_noise": 0.0, "grad_norm": 0.7152, "lr": 0.0001945, "router_trainable": true, "sec_per_step": 0.406, "probe_sec": 0.7562, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.0523, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": 0.11722, "n_shadow": 5, "mean_delta_shadow": -0.5303, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 1, "trunc_ratio_sum": -0.23662, "trunc_abs_frac": 0.23662, "mean_delta_full": -0.72632, "read_hit_rate": {"surface": 0.58703}, "mean_gate": {"surface": 0.23784}, "n_reads": 3973, "n_cand": 6768, "read_suppressed_frac": 0.0909, "dropout_reads_logged": 3973, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 124.7222900390625, "n_util_obs": 456, "util_count_spearman": 0.10020252730256014}, "probe_overhead_sec_total": 265.6, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.020869, "alpha": 0.03, "t": 1261.1}
{"event": "consolidate", "step": 2800, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 82, "uncertain_staged": 262121, "t": 1281.2}
{"event": "probe", "step": 2800, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3322.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.190887, "cps": 25.953125, "mean_p": 0.20034, "boundary_entropy": 0.114835, "bpb": 2.027053}, "tr": {"chunks": 3800.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.20991, "cps": 29.6875, "mean_p": 0.205325, "boundary_entropy": 0.141275, "bpb": 2.223759}, "fi": {"chunks": 4884.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.259635, "cps": 38.15625, "mean_p": 0.255808, "boundary_entropy": 0.128845, "bpb": 2.32578}}, "bpb": {"en": 2.02705, "tr": 2.22376, "fi": 2.32578}, "gini_cpc": 0.069398, "gini_cps": 0.086734, "bpic": 4.725304, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0996577448005316, "fi": 1.3601545664594585}, "inventory": {"size": 327598, "admitted": 65536, "staged": 262062, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262062, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 127.73933410644531, "n_util_obs": 461, "util_count_spearman": 0.0961796850553419}, "t": 1281.7}
{"event": "eval", "step": 2800, "phase": 2, "bpb": {"en": 2.04137, "tr": 2.22925, "fi": 2.3393}, "bytes_per_chunk": {"en": 4.917, "tr": 5.29, "fi": 4.093}, "bpb_mean": 2.2033, "frozen_inventory": true, "t": 1282.6}
{"event": "train", "step": 2800, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.44866, "bpb": 2.08998, "chunk_loss": 0.030521, "F": 0.20914, "G": 0.2155, "chunks": 214.16, "bytes_per_chunk": 4.782, "boundary_entropy": 0.11353, "router_noise": 0.0, "grad_norm": 0.6881, "lr": 0.0001909, "router_trainable": true, "sec_per_step": 0.513, "probe_sec": 0.0405, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02444, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": 0.18681, "n_shadow": 6, "mean_delta_shadow": -0.09493, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 6, "trunc_ratio_sum": -9.12224, "trunc_abs_frac": 1.18645, "mean_delta_full": -0.00553, "read_hit_rate": {"surface": 0.58349}, "mean_gate": {"surface": 0.23657}, "n_reads": 3980, "n_cand": 6821, "read_suppressed_frac": 0.0965, "dropout_reads_logged": 3980, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 127.80006408691406, "n_util_obs": 462, "util_count_spearman": 0.0929193913601641}, "probe_overhead_sec_total": 268.538, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.022835, "alpha": 0.03, "t": 1286.8}
{"event": "consolidate", "step": 2850, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 215, "uncertain_staged": 262121, "t": 1314.6}
{"event": "train", "step": 2850, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.54951, "bpb": 2.23547, "chunk_loss": 0.030712, "F": 0.21329, "G": 0.22159, "chunks": 218.41, "bytes_per_chunk": 4.689, "boundary_entropy": 0.11468, "router_noise": 0.0, "grad_norm": 0.684, "lr": 0.0001872, "router_trainable": true, "sec_per_step": 0.56, "probe_sec": 0.033, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02121, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 5, "mean_delta_loo": 0.16343, "n_shadow": 9, "mean_delta_shadow": -0.0942, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 1, "trunc_ratio_sum": -0.79856, "trunc_abs_frac": 0.79856, "mean_delta_full": -0.07605, "read_hit_rate": {"surface": 0.58157}, "mean_gate": {"surface": 0.23948}, "n_reads": 4046, "n_cand": 6957, "read_suppressed_frac": 0.0981, "dropout_reads_logged": 4046, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 130.8585662841797, "n_util_obs": 471, "util_count_spearman": 0.0879144302914078}, "probe_overhead_sec_total": 271.209, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.021491, "alpha": 0.03, "t": 1314.8}
{"event": "consolidate", "step": 2900, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 152, "uncertain_staged": 262121, "t": 1341.7}
{"event": "train", "step": 2900, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.49012, "bpb": 2.14979, "chunk_loss": 0.030736, "F": 0.21588, "G": 0.22329, "chunks": 221.06, "bytes_per_chunk": 4.632, "boundary_entropy": 0.11582, "router_noise": 0.0, "grad_norm": 0.6538, "lr": 0.0001835, "router_trainable": true, "sec_per_step": 0.551, "probe_sec": 0.0528, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02472, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": 0.00949, "n_shadow": 5, "mean_delta_shadow": -0.40385, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": 8.08634, "trunc_abs_frac": 1.71116, "mean_delta_full": -0.03718, "read_hit_rate": {"surface": 0.60636}, "mean_gate": {"surface": 0.24248}, "n_reads": 4270, "n_cand": 7042, "read_suppressed_frac": 0.0953, "dropout_reads_logged": 4270, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 133.9237823486328, "n_util_obs": 480, "util_count_spearman": 0.079081419375038}, "probe_overhead_sec_total": 273.889, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.019357, "alpha": 0.03, "t": 1342.3}
{"event": "consolidate", "step": 2950, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 130, "uncertain_staged": 262121, "t": 1368.7}
{"event": "train", "step": 2950, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.43281, "bpb": 2.06711, "chunk_loss": 0.030434, "F": 0.20203, "G": 0.21104, "chunks": 206.88, "bytes_per_chunk": 4.95, "boundary_entropy": 0.11683, "router_noise": 0.0, "grad_norm": 0.6981, "lr": 0.0001799, "router_trainable": true, "sec_per_step": 0.532, "probe_sec": 0.0289, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.01891, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 11, "mean_delta_loo": 0.04339, "n_shadow": 3, "mean_delta_shadow": 0.05348, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 1.08595, "trunc_abs_frac": 0.44023, "mean_delta_full": 0.06083, "read_hit_rate": {"surface": 0.57772}, "mean_gate": {"surface": 0.2524}, "n_reads": 3806, "n_cand": 6588, "read_suppressed_frac": 0.104, "dropout_reads_logged": 3806, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 136.97396850585938, "n_util_obs": 490, "util_count_spearman": 0.07716885080028484}, "probe_overhead_sec_total": 276.879, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.021397, "alpha": 0.03, "t": 1368.9}
{"event": "consolidate", "step": 3000, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 127, "uncertain_staged": 262121, "t": 1389.3}
{"event": "probe", "step": 3000, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3352.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.19261, "cps": 26.1875, "mean_p": 0.19973, "boundary_entropy": 0.1063, "bpb": 2.011253}, "tr": {"chunks": 3774.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.208474, "cps": 29.484375, "mean_p": 0.202693, "boundary_entropy": 0.130757, "bpb": 2.195798}, "fi": {"chunks": 4864.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.258572, "cps": 38.0, "mean_p": 0.255041, "boundary_entropy": 0.125228, "bpb": 2.309889}}, "bpb": {"en": 2.01125, "tr": 2.1958, "fi": 2.30989}, "gini_cpc": 0.066663, "gini_cps": 0.08407, "bpic": 4.73161, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0823593038351085, "fi": 1.3424613562698615}, "inventory": {"size": 327553, "admitted": 65536, "staged": 262017, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262017, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 139.9671173095703, "n_util_obs": 498, "util_count_spearman": 0.0773830292123505}, "t": 1389.8}
{"event": "eval", "step": 3000, "phase": 2, "bpb": {"en": 2.02518, "tr": 2.20264, "fi": 2.32266}, "bytes_per_chunk": {"en": 4.904, "tr": 5.268, "fi": 4.114}, "bpb_mean": 2.18349, "frozen_inventory": true, "t": 1390.7}
{"event": "train", "step": 3000, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.46333, "bpb": 2.11114, "chunk_loss": 0.030449, "F": 0.20612, "G": 0.21266, "chunks": 211.06, "bytes_per_chunk": 4.852, "boundary_entropy": 0.10751, "router_noise": 0.0, "grad_norm": 0.6866, "lr": 0.0001761, "router_trainable": true, "sec_per_step": 0.534, "probe_sec": 0.026, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.01874, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": 0.22271, "n_shadow": 8, "mean_delta_shadow": -0.12548, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 1.74552, "trunc_abs_frac": 1.61437, "mean_delta_full": -0.09198, "read_hit_rate": {"surface": 0.58792}, "mean_gate": {"surface": 0.24325}, "n_reads": 3952, "n_cand": 6722, "read_suppressed_frac": 0.0999, "dropout_reads_logged": 3952, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 140.0274200439453, "n_util_obs": 498, "util_count_spearman": 0.0769429801231975}, "probe_overhead_sec_total": 279.499, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.023559, "alpha": 0.03, "t": 1395.6}
{"event": "consolidate", "step": 3050, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 114, "uncertain_staged": 262121, "t": 1416.2}
{"event": "train", "step": 3050, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.44806, "bpb": 2.08911, "chunk_loss": 0.030559, "F": 0.2114, "G": 0.21737, "chunks": 216.47, "bytes_per_chunk": 4.73, "boundary_entropy": 0.10747, "router_noise": 0.0, "grad_norm": 0.71, "lr": 0.0001724, "router_trainable": true, "sec_per_step": 0.436, "probe_sec": 0.0657, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02467, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": 0.36517, "n_shadow": 5, "mean_delta_shadow": -0.61077, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 0.67279, "trunc_abs_frac": 1.27553, "mean_delta_full": -0.87219, "read_hit_rate": {"surface": 0.60087}, "mean_gate": {"surface": 0.24855}, "n_reads": 4143, "n_cand": 6895, "read_suppressed_frac": 0.0982, "dropout_reads_logged": 4143, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 143.10523986816406, "n_util_obs": 512, "util_count_spearman": 0.06012369592159175}, "probe_overhead_sec_total": 281.931, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.021379, "alpha": 0.03, "t": 1417.4}
{"event": "consolidate", "step": 3100, "retired": 1, "split_retired": 0, "merge_admitted": 0, "admitted": 1, "swapped": 46, "uncertain_staged": 262121, "t": 1445.7}
{"event": "train", "step": 3100, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.54121, "bpb": 2.2235, "chunk_loss": 0.030546, "F": 0.20456, "G": 0.21339, "chunks": 209.47, "bytes_per_chunk": 4.889, "boundary_entropy": 0.11524, "router_noise": 0.0, "grad_norm": 0.7374, "lr": 0.0001687, "router_trainable": true, "sec_per_step": 0.569, "probe_sec": 0.027, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02078, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": 0.36321, "n_shadow": 6, "mean_delta_shadow": -0.3441, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 0.8737, "trunc_abs_frac": 0.86209, "mean_delta_full": 0.79132, "read_hit_rate": {"surface": 0.57248}, "mean_gate": {"surface": 0.24882}, "n_reads": 3819, "n_cand": 6671, "read_suppressed_frac": 0.1066, "dropout_reads_logged": 3819, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 146.0475616455078, "n_util_obs": 521, "util_count_spearman": 0.05309749944596029}, "probe_overhead_sec_total": 284.997, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.021883, "alpha": 0.03, "t": 1445.9}
{"event": "consolidate", "step": 3150, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 98, "uncertain_staged": 262121, "t": 1471.7}
{"event": "train", "step": 3150, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.46103, "bpb": 2.10782, "chunk_loss": 0.030519, "F": 0.20773, "G": 0.21372, "chunks": 212.72, "bytes_per_chunk": 4.814, "boundary_entropy": 0.1102, "router_noise": 0.0, "grad_norm": 0.7333, "lr": 0.000165, "router_trainable": true, "sec_per_step": 0.531, "probe_sec": 0.0512, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02381, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": 0.40117, "n_shadow": 8, "mean_delta_shadow": -0.14807, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 0, "read_hit_rate": {"surface": 0.5938}, "mean_gate": {"surface": 0.25317}, "n_reads": 4023, "n_cand": 6775, "read_suppressed_frac": 0.0979, "dropout_reads_logged": 4023, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 149.04071044921875, "n_util_obs": 527, "util_count_spearman": 0.05168381424953096}, "probe_overhead_sec_total": 288.255, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.022321, "alpha": 0.03, "t": 1472.4}
{"event": "consolidate", "step": 3200, "retired": 1, "split_retired": 0, "merge_admitted": 0, "admitted": 1, "swapped": 366, "uncertain_staged": 262121, "t": 1493.9}
{"event": "probe", "step": 3200, "phase": 2, "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": 0.199806, "boundary_entropy": 0.103752, "bpb": 1.991036}, "tr": {"chunks": 3771.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.208308, "cps": 29.460938, "mean_p": 0.203191, "boundary_entropy": 0.12848, "bpb": 2.170177}, "fi": {"chunks": 4850.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.257828, "cps": 37.890625, "mean_p": 0.255009, "boundary_entropy": 0.127165, "bpb": 2.276591}}, "bpb": {"en": 1.99104, "tr": 2.17018, "fi": 2.27659}, "gini_cpc": 0.065363, "gini_cps": 0.082784, "bpic": 4.734374, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0782820908172919, "fi": 1.3346158131241386}, "inventory": {"size": 327313, "admitted": 65536, "staged": 261777, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 261777, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 151.91151428222656, "n_util_obs": 531, "util_count_spearman": 0.04249627872179936}, "t": 1494.5}
{"event": "eval", "step": 3200, "phase": 2, "bpb": {"en": 2.00555, "tr": 2.18013, "fi": 2.28871}, "bytes_per_chunk": {"en": 4.922, "tr": 5.353, "fi": 4.12}, "bpb_mean": 2.15813, "frozen_inventory": true, "t": 1495.5}
{"event": "train", "step": 3200, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.42565, "bpb": 2.05677, "chunk_loss": 0.030643, "F": 0.21191, "G": 0.21855, "chunks": 217.0, "bytes_per_chunk": 4.719, "boundary_entropy": 0.10593, "router_noise": 0.0, "grad_norm": 0.6603, "lr": 0.0001613, "router_trainable": true, "sec_per_step": 0.548, "probe_sec": 0.0621, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.03129, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 11, "mean_delta_loo": 0.05057, "n_shadow": 3, "mean_delta_shadow": 0.1552, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 6, "trunc_ratio_sum": 0.53859, "trunc_abs_frac": 1.43585, "mean_delta_full": 0.1956, "read_hit_rate": {"surface": 0.60503}, "mean_gate": {"surface": 0.23828}, "n_reads": 4182, "n_cand": 6912, "read_suppressed_frac": 0.1014, "dropout_reads_logged": 4182, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 151.97532653808594, "n_util_obs": 531, "util_count_spearman": 0.04242410519906046}, "probe_overhead_sec_total": 290.681, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.023163, "alpha": 0.03, "t": 1499.8}
{"event": "consolidate", "step": 3250, "retired": 1, "split_retired": 0, "merge_admitted": 0, "admitted": 1, "swapped": 562, "uncertain_staged": 262121, "t": 1523.7}
{"event": "train", "step": 3250, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.40842, "bpb": 2.03192, "chunk_loss": 0.030376, "F": 0.20203, "G": 0.20712, "chunks": 206.88, "bytes_per_chunk": 4.95, "boundary_entropy": 0.10519, "router_noise": 0.0, "grad_norm": 0.6669, "lr": 0.0001576, "router_trainable": true, "sec_per_step": 0.5, "probe_sec": 0.0584, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02236, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": 0.00623, "n_shadow": 7, "mean_delta_shadow": -0.4239, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 0.7965, "trunc_abs_frac": 2.24634, "mean_delta_full": -0.65985, "read_hit_rate": {"surface": 0.58561}, "mean_gate": {"surface": 0.24337}, "n_reads": 3858, "n_cand": 6588, "read_suppressed_frac": 0.098, "dropout_reads_logged": 3858, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 154.7514190673828, "n_util_obs": 529, "util_count_spearman": 0.0234584560422273}, "probe_overhead_sec_total": 293.189, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.022217, "alpha": 0.03, "t": 1524.8}
{"event": "consolidate", "step": 3300, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 145, "uncertain_staged": 262121, "t": 1558.1}
{"event": "train", "step": 3300, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.44664, "bpb": 2.08706, "chunk_loss": 0.030573, "F": 0.21094, "G": 0.21494, "chunks": 216.0, "bytes_per_chunk": 4.741, "boundary_entropy": 0.10557, "router_noise": 0.0, "grad_norm": 0.6591, "lr": 0.0001539, "router_trainable": true, "sec_per_step": 0.668, "probe_sec": 0.0384, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.0281, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": 0.18676, "n_shadow": 6, "mean_delta_shadow": 0.28392, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 1.13136, "trunc_abs_frac": 0.70173, "mean_delta_full": 0.16071, "read_hit_rate": {"surface": 0.60959}, "mean_gate": {"surface": 0.24118}, "n_reads": 4194, "n_cand": 6880, "read_suppressed_frac": 0.097, "dropout_reads_logged": 4194, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 157.82713317871094, "n_util_obs": 538, "util_count_spearman": 0.02824941717621839}, "probe_overhead_sec_total": 295.978, "probe_overhead_seq_frac": 0.4998, "util_loss": 0.023077, "alpha": 0.03, "t": 1558.3}
{"event": "consolidate", "step": 3350, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 92, "uncertain_staged": 262121, "t": 1589.1}
{"event": "train", "step": 3350, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.48333, "bpb": 2.14, "chunk_loss": 0.030683, "F": 0.21619, "G": 0.21969, "chunks": 221.38, "bytes_per_chunk": 4.626, "boundary_entropy": 0.10826, "router_noise": 0.0, "grad_norm": 0.7893, "lr": 0.0001501, "router_trainable": true, "sec_per_step": 0.635, "probe_sec": 0.0595, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02861, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": 0.0835, "n_shadow": 5, "mean_delta_shadow": -0.35729, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 6, "trunc_ratio_sum": 0.78791, "trunc_abs_frac": 1.0361, "mean_delta_full": -0.08531, "read_hit_rate": {"surface": 0.60026}, "mean_gate": {"surface": 0.24937}, "n_reads": 4233, "n_cand": 7052, "read_suppressed_frac": 0.1044, "dropout_reads_logged": 4233, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 160.86935424804688, "n_util_obs": 547, "util_count_spearman": 0.025615966277622748}, "probe_overhead_sec_total": 298.816, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.023542, "alpha": 0.03, "t": 1590.0}
{"event": "consolidate", "step": 3400, "retired": 1, "split_retired": 0, "merge_admitted": 0, "admitted": 1, "swapped": 67, "uncertain_staged": 262121, "t": 1615.1}
{"event": "probe", "step": 3400, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3339.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.191863, "cps": 26.085938, "mean_p": 0.199029, "boundary_entropy": 0.103592, "bpb": 1.979816}, "tr": {"chunks": 3772.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.208363, "cps": 29.46875, "mean_p": 0.201782, "boundary_entropy": 0.133263, "bpb": 2.149479}, "fi": {"chunks": 4855.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.258094, "cps": 37.929688, "mean_p": 0.252559, "boundary_entropy": 0.119896, "bpb": 2.261078}}, "bpb": {"en": 1.97982, "tr": 2.14948, "fi": 2.26108}, "gini_cpc": 0.06707, "gini_cps": 0.084462, "bpic": 4.7411, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0859975207258417, "fi": 1.345194403897511}, "inventory": {"size": 327612, "admitted": 65536, "staged": 262076, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262076, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 163.82199096679688, "n_util_obs": 553, "util_count_spearman": 0.031587104476218576}, "t": 1615.7}
{"event": "eval", "step": 3400, "phase": 2, "bpb": {"en": 1.99426, "tr": 2.1582, "fi": 2.2707}, "bytes_per_chunk": {"en": 4.939, "tr": 5.37, "fi": 4.111}, "bpb_mean": 2.14106, "frozen_inventory": true, "t": 1616.7}
{"event": "train", "step": 3400, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.39446, "bpb": 2.01178, "chunk_loss": 0.030401, "F": 0.20123, "G": 0.20649, "chunks": 206.06, "bytes_per_chunk": 4.969, "boundary_entropy": 0.10117, "router_noise": 0.0, "grad_norm": 0.7352, "lr": 0.0001465, "router_trainable": true, "sec_per_step": 0.622, "probe_sec": 0.0526, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02368, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": 0.0709, "n_shadow": 5, "mean_delta_shadow": -0.26204, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 0.64833, "trunc_abs_frac": 0.67818, "mean_delta_full": -0.32849, "read_hit_rate": {"surface": 0.59311}, "mean_gate": {"surface": 0.23592}, "n_reads": 3892, "n_cand": 6562, "read_suppressed_frac": 0.1015, "dropout_reads_logged": 3892, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 163.88137817382812, "n_util_obs": 553, "util_count_spearman": 0.031601086197531554}, "probe_overhead_sec_total": 301.549, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.021062, "alpha": 0.03, "t": 1621.1}
{"event": "consolidate", "step": 3450, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 123, "uncertain_staged": 262121, "t": 1643.5}
{"event": "train", "step": 3450, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.39299, "bpb": 2.00966, "chunk_loss": 0.030339, "F": 0.20041, "G": 0.20582, "chunks": 205.22, "bytes_per_chunk": 4.99, "boundary_entropy": 0.10118, "router_noise": 0.0, "grad_norm": 0.6688, "lr": 0.0001428, "router_trainable": true, "sec_per_step": 0.466, "probe_sec": 0.0527, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02487, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": -0.18496, "n_shadow": 8, "mean_delta_shadow": -0.16374, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": -0.25455, "trunc_abs_frac": 0.77383, "mean_delta_full": 0.32275, "read_hit_rate": {"surface": 0.59158}, "mean_gate": {"surface": 0.23543}, "n_reads": 3866, "n_cand": 6535, "read_suppressed_frac": 0.1019, "dropout_reads_logged": 3866, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 166.90000915527344, "n_util_obs": 566, "util_count_spearman": 0.025511751653528446}, "probe_overhead_sec_total": 304.173, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.021345, "alpha": 0.03, "t": 1644.4}
{"event": "consolidate", "step": 3500, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 227, "uncertain_staged": 262122, "t": 1668.2}
{"event": "train", "step": 3500, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.47365, "bpb": 2.12602, "chunk_loss": 0.030622, "F": 0.20905, "G": 0.21508, "chunks": 214.06, "bytes_per_chunk": 4.784, "boundary_entropy": 0.10287, "router_noise": 0.0, "grad_norm": 0.6891, "lr": 0.0001391, "router_trainable": true, "sec_per_step": 0.48, "probe_sec": 0.0312, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02031, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": -0.1895, "n_shadow": 5, "mean_delta_shadow": -0.11913, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": -0.1603, "trunc_abs_frac": 1.68135, "mean_delta_full": -0.03516, "read_hit_rate": {"surface": 0.58419}, "mean_gate": {"surface": 0.2359}, "n_reads": 3983, "n_cand": 6818, "read_suppressed_frac": 0.0982, "dropout_reads_logged": 3983, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 169.89064025878906, "n_util_obs": 573, "util_count_spearman": 0.027175774595925864}, "probe_overhead_sec_total": 306.986, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.024522, "alpha": 0.03, "t": 1668.4}
{"event": "consolidate", "step": 3550, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 72, "uncertain_staged": 262122, "t": 1696.9}
{"event": "train", "step": 3550, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.47954, "bpb": 2.13452, "chunk_loss": 0.03079, "F": 0.21875, "G": 0.22331, "chunks": 224.0, "bytes_per_chunk": 4.571, "boundary_entropy": 0.10306, "router_noise": 0.0, "grad_norm": 0.7161, "lr": 0.0001355, "router_trainable": true, "sec_per_step": 0.573, "probe_sec": 0.0454, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.0343, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 10, "mean_delta_loo": 0.12331, "n_shadow": 4, "mean_delta_shadow": -0.01655, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 1.12521, "trunc_abs_frac": 0.5557, "mean_delta_full": -0.07245, "read_hit_rate": {"surface": 0.61071}, "mean_gate": {"surface": 0.23458}, "n_reads": 4358, "n_cand": 7136, "read_suppressed_frac": 0.0948, "dropout_reads_logged": 4358, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 172.94793701171875, "n_util_obs": 581, "util_count_spearman": 0.01874607661938854}, "probe_overhead_sec_total": 309.862, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.0251, "alpha": 0.03, "t": 1697.1}
{"event": "consolidate", "step": 3600, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 87, "uncertain_staged": 262122, "t": 1720.7}
{"event": "probe", "step": 3600, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3318.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.190657, "cps": 25.921875, "mean_p": 0.197195, "boundary_entropy": 0.096399, "bpb": 1.961439}, "tr": {"chunks": 3716.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.20527, "cps": 29.03125, "mean_p": 0.198693, "boundary_entropy": 0.124141, "bpb": 2.13029}, "fi": {"chunks": 4870.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.258891, "cps": 38.046875, "mean_p": 0.253901, "boundary_entropy": 0.114096, "bpb": 2.233971}}, "bpb": {"en": 1.96144, "tr": 2.13029, "fi": 2.23397}, "gini_cpc": 0.069469, "gini_cps": 0.086918, "bpic": 4.765793, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0766459037540759, "fi": 1.3578907074397524}, "inventory": {"size": 327593, "admitted": 65536, "staged": 262057, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262057, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 175.8920135498047, "n_util_obs": 587, "util_count_spearman": 0.02512605186725173}, "t": 1721.3}
{"event": "eval", "step": 3600, "phase": 2, "bpb": {"en": 1.97694, "tr": 2.13674, "fi": 2.2459}, "bytes_per_chunk": {"en": 4.917, "tr": 5.383, "fi": 4.102}, "bpb_mean": 2.11986, "frozen_inventory": true, "t": 1722.1}
{"event": "train", "step": 3600, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.45792, "bpb": 2.10334, "chunk_loss": 0.030444, "F": 0.20538, "G": 0.20972, "chunks": 210.31, "bytes_per_chunk": 4.869, "boundary_entropy": 0.09764, "router_noise": 0.0, "grad_norm": 0.6903, "lr": 0.0001319, "router_trainable": true, "sec_per_step": 0.608, "probe_sec": 0.0568, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02433, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": 0.0834, "n_shadow": 5, "mean_delta_shadow": 0.13217, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 0.52655, "trunc_abs_frac": 0.65281, "mean_delta_full": -0.14655, "read_hit_rate": {"surface": 0.5742}, "mean_gate": {"surface": 0.22705}, "n_reads": 3846, "n_cand": 6698, "read_suppressed_frac": 0.1017, "dropout_reads_logged": 3846, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 175.95069885253906, "n_util_obs": 587, "util_count_spearman": 0.025096384676534772}, "probe_overhead_sec_total": 312.646, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.022366, "alpha": 0.03, "t": 1727.5}
{"event": "consolidate", "step": 3650, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 271, "uncertain_staged": 262122, "t": 1756.1}
{"event": "train", "step": 3650, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.47539, "bpb": 2.12854, "chunk_loss": 0.030657, "F": 0.2132, "G": 0.21879, "chunks": 218.31, "bytes_per_chunk": 4.691, "boundary_entropy": 0.10085, "router_noise": 0.0, "grad_norm": 0.6972, "lr": 0.0001283, "router_trainable": true, "sec_per_step": 0.577, "probe_sec": 0.0882, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.07643, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": 0.11141, "n_shadow": 5, "mean_delta_shadow": 0.03802, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 0.78634, "trunc_abs_frac": 0.73484, "mean_delta_full": 0.47748, "read_hit_rate": {"surface": 0.60469}, "mean_gate": {"surface": 0.24615}, "n_reads": 4205, "n_cand": 6954, "read_suppressed_frac": 0.0985, "dropout_reads_logged": 4205, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 179.03514099121094, "n_util_obs": 599, "util_count_spearman": 0.023738373124419596}, "probe_overhead_sec_total": 315.443, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.020981, "alpha": 0.03, "t": 1756.3}
{"event": "consolidate", "step": 3700, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 144, "uncertain_staged": 262122, "t": 1779.3}
{"event": "train", "step": 3700, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.37622, "bpb": 1.98547, "chunk_loss": 0.030415, "F": 0.20322, "G": 0.20864, "chunks": 208.09, "bytes_per_chunk": 4.921, "boundary_entropy": 0.10083, "router_noise": 0.0, "grad_norm": 0.6782, "lr": 0.0001247, "router_trainable": true, "sec_per_step": 0.47, "probe_sec": 0.0432, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02335, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": -0.00346, "n_shadow": 6, "mean_delta_shadow": 0.12259, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": 1.2698, "trunc_abs_frac": 3.77798, "mean_delta_full": 0.1042, "read_hit_rate": {"surface": 0.59001}, "mean_gate": {"surface": 0.23887}, "n_reads": 3910, "n_cand": 6627, "read_suppressed_frac": 0.0908, "dropout_reads_logged": 3910, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 182.09202575683594, "n_util_obs": 604, "util_count_spearman": 0.0242171066280438}, "probe_overhead_sec_total": 318.303, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.024047, "alpha": 0.03, "t": 1779.8}
{"event": "consolidate", "step": 3750, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 191, "uncertain_staged": 262122, "t": 1808.8}
{"event": "train", "step": 3750, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.39942, "bpb": 2.01893, "chunk_loss": 0.030617, "F": 0.21335, "G": 0.21619, "chunks": 218.47, "bytes_per_chunk": 4.687, "boundary_entropy": 0.09393, "router_noise": 0.0, "grad_norm": 0.6725, "lr": 0.0001212, "router_trainable": true, "sec_per_step": 0.597, "probe_sec": 0.0902, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.05528, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 5, "mean_delta_loo": 0.3967, "n_shadow": 9, "mean_delta_shadow": -0.10078, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": 0.54377, "trunc_abs_frac": 0.88444, "mean_delta_full": -0.19165, "read_hit_rate": {"surface": 0.61359}, "mean_gate": {"surface": 0.22096}, "n_reads": 4270, "n_cand": 6959, "read_suppressed_frac": 0.1047, "dropout_reads_logged": 4270, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 185.12875366210938, "n_util_obs": 613, "util_count_spearman": 0.036210192092362474}, "probe_overhead_sec_total": 321.139, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.024216, "alpha": 0.03, "t": 1809.6}
{"event": "consolidate", "step": 3800, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 79, "uncertain_staged": 262122, "t": 1828.2}
{"event": "probe", "step": 3800, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3296.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.189393, "cps": 25.75, "mean_p": 0.195721, "boundary_entropy": 0.094574, "bpb": 1.947218}, "tr": {"chunks": 3814.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.210683, "cps": 29.796875, "mean_p": 0.201184, "boundary_entropy": 0.122733, "bpb": 2.109408}, "fi": {"chunks": 4873.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.259051, "cps": 38.070312, "mean_p": 0.253507, "boundary_entropy": 0.109795, "bpb": 2.220468}}, "bpb": {"en": 1.94722, "tr": 2.10941, "fi": 2.22047}, "gini_cpc": 0.070455, "gini_cps": 0.087735, "bpic": 4.734374, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.1124155587042646, "fi": 1.3677963646296607}, "inventory": {"size": 327601, "admitted": 65536, "staged": 262065, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262065, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 188.14898681640625, "n_util_obs": 626, "util_count_spearman": 0.021976455829126853}, "t": 1828.8}
{"event": "eval", "step": 3800, "phase": 2, "bpb": {"en": 1.9627, "tr": 2.11574, "fi": 2.23035}, "bytes_per_chunk": {"en": 4.94, "tr": 5.231, "fi": 4.097}, "bpb_mean": 2.10293, "frozen_inventory": true, "t": 1829.8}
{"event": "train", "step": 3800, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.38114, "bpb": 1.99257, "chunk_loss": 0.030486, "F": 0.20193, "G": 0.20623, "chunks": 206.78, "bytes_per_chunk": 4.952, "boundary_entropy": 0.09745, "router_noise": 0.0, "grad_norm": 0.661, "lr": 0.0001177, "router_trainable": true, "sec_per_step": 0.481, "probe_sec": 0.1243, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.11154, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": -0.24228, "n_shadow": 5, "mean_delta_shadow": -0.04865, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 6, "trunc_ratio_sum": 1.31748, "trunc_abs_frac": 2.02317, "mean_delta_full": -0.27828, "read_hit_rate": {"surface": 0.58618}, "mean_gate": {"surface": 0.23537}, "n_reads": 3860, "n_cand": 6585, "read_suppressed_frac": 0.0935, "dropout_reads_logged": 3860, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 188.2078857421875, "n_util_obs": 626, "util_count_spearman": 0.02158301675706584}, "probe_overhead_sec_total": 323.829, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.024766, "alpha": 0.03, "t": 1833.7}
{"event": "consolidate", "step": 3850, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 51, "uncertain_staged": 262122, "t": 1856.6}
{"event": "train", "step": 3850, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.36756, "bpb": 1.97298, "chunk_loss": 0.030307, "F": 0.19812, "G": 0.20263, "chunks": 202.88, "bytes_per_chunk": 5.047, "boundary_entropy": 0.10116, "router_noise": 0.0, "grad_norm": 0.7394, "lr": 0.0001142, "router_trainable": true, "sec_per_step": 0.483, "probe_sec": 0.0676, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02395, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 11, "mean_delta_loo": -0.06407, "n_shadow": 3, "mean_delta_shadow": 0.16165, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 0.32075, "trunc_abs_frac": 0.8601, "mean_delta_full": -0.11032, "read_hit_rate": {"surface": 0.58111}, "mean_gate": {"surface": 0.24008}, "n_reads": 3754, "n_cand": 6460, "read_suppressed_frac": 0.0948, "dropout_reads_logged": 3754, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 191.29530334472656, "n_util_obs": 633, "util_count_spearman": 0.014175288765213373}, "probe_overhead_sec_total": 326.456, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.021714, "alpha": 0.03, "t": 1857.8}
{"event": "consolidate", "step": 3900, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 57, "uncertain_staged": 262122, "t": 1883.7}
{"event": "train", "step": 3900, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.40376, "bpb": 2.0252, "chunk_loss": 0.030447, "F": 0.20465, "G": 0.20816, "chunks": 209.56, "bytes_per_chunk": 4.886, "boundary_entropy": 0.09789, "router_noise": 0.0, "grad_norm": 0.6552, "lr": 0.0001108, "router_trainable": true, "sec_per_step": 0.532, "probe_sec": 0.0487, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02664, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": -0.03266, "n_shadow": 5, "mean_delta_shadow": 0.33389, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 32.40955, "trunc_abs_frac": 1.39916, "mean_delta_full": 0.0081, "read_hit_rate": {"surface": 0.59005}, "mean_gate": {"surface": 0.24457}, "n_reads": 3938, "n_cand": 6674, "read_suppressed_frac": 0.1026, "dropout_reads_logged": 3938, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 194.3480224609375, "n_util_obs": 641, "util_count_spearman": 0.021226237948328345}, "probe_overhead_sec_total": 329.167, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.023716, "alpha": 0.03, "t": 1884.4}
{"event": "consolidate", "step": 3950, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 32, "uncertain_staged": 262122, "t": 1917.5}
{"event": "train", "step": 3950, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.43123, "bpb": 2.06483, "chunk_loss": 0.0305, "F": 0.20813, "G": 0.21242, "chunks": 213.12, "bytes_per_chunk": 4.805, "boundary_entropy": 0.09282, "router_noise": 0.0, "grad_norm": 0.6724, "lr": 0.0001074, "router_trainable": true, "sec_per_step": 0.674, "probe_sec": 0.0603, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02697, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 5, "mean_delta_loo": -0.11453, "n_shadow": 9, "mean_delta_shadow": -0.27392, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": -0.27924, "trunc_abs_frac": 0.18646, "mean_delta_full": -0.29382, "read_hit_rate": {"surface": 0.57749}, "mean_gate": {"surface": 0.22194}, "n_reads": 3920, "n_cand": 6788, "read_suppressed_frac": 0.0977, "dropout_reads_logged": 3920, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 197.41055297851562, "n_util_obs": 646, "util_count_spearman": 0.030016687622175565}, "probe_overhead_sec_total": 332.275, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.026802, "alpha": 0.03, "t": 1918.1}
{"event": "consolidate", "step": 4000, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 13, "uncertain_staged": 262122, "t": 1945.3}
{"event": "probe", "step": 4000, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3349.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.192438, "cps": 26.164062, "mean_p": 0.196856, "boundary_entropy": 0.095662, "bpb": 1.933194}, "tr": {"chunks": 3768.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.208142, "cps": 29.4375, "mean_p": 0.1983, "boundary_entropy": 0.121852, "bpb": 2.091214}, "fi": {"chunks": 4881.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.259476, "cps": 38.132812, "mean_p": 0.252646, "boundary_entropy": 0.112712, "bpb": 2.200363}}, "bpb": {"en": 1.93319, "tr": 2.09121, "fi": 2.20036}, "gini_cpc": 0.067709, "gini_cps": 0.085125, "bpic": 4.728455, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0816065667961146, "fi": 1.348360113240985}, "inventory": {"size": 327667, "admitted": 65536, "staged": 262131, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262131, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 200.43600463867188, "n_util_obs": 655, "util_count_spearman": 0.03260940210840154}, "t": 1946.0}
{"event": "eval", "step": 4000, "phase": 2, "bpb": {"en": 1.94934, "tr": 2.09821, "fi": 2.2115}, "bytes_per_chunk": {"en": 4.915, "tr": 5.355, "fi": 4.093}, "bpb_mean": 2.08635, "frozen_inventory": true, "t": 1947.0}
{"event": "train", "step": 4000, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.44996, "bpb": 2.09184, "chunk_loss": 0.030492, "F": 0.20853, "G": 0.21246, "chunks": 213.53, "bytes_per_chunk": 4.796, "boundary_entropy": 0.09607, "router_noise": 0.0, "grad_norm": 0.7119, "lr": 0.000104, "router_trainable": true, "sec_per_step": 0.681, "probe_sec": 0.1245, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.11341, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": 0.1241, "n_shadow": 5, "mean_delta_shadow": 0.06458, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 1, "trunc_ratio_sum": 0.53763, "trunc_abs_frac": 0.53763, "mean_delta_full": 0.31628, "read_hit_rate": {"surface": 0.5883}, "mean_gate": {"surface": 0.22691}, "n_reads": 4001, "n_cand": 6801, "read_suppressed_frac": 0.1057, "dropout_reads_logged": 4001, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 200.49705505371094, "n_util_obs": 655, "util_count_spearman": 0.03244459055515056}, "probe_overhead_sec_total": 335.215, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.021654, "alpha": 0.03, "t": 1952.2}
{"event": "consolidate", "step": 4050, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 10, "uncertain_staged": 262122, "t": 1978.2}
{"event": "train", "step": 4050, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.37229, "bpb": 1.9798, "chunk_loss": 0.030447, "F": 0.20349, "G": 0.20786, "chunks": 208.38, "bytes_per_chunk": 4.914, "boundary_entropy": 0.0895, "router_noise": 0.0, "grad_norm": 0.6848, "lr": 0.0001007, "router_trainable": true, "sec_per_step": 0.527, "probe_sec": 0.1743, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.16771, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 12, "mean_delta_loo": 0.04453, "n_shadow": 2, "mean_delta_shadow": 0.11644, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 0.95802, "trunc_abs_frac": 1.02746, "mean_delta_full": 0.25492, "read_hit_rate": {"surface": 0.59177}, "mean_gate": {"surface": 0.22448}, "n_reads": 3927, "n_cand": 6636, "read_suppressed_frac": 0.1049, "dropout_reads_logged": 3927, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 203.58053588867188, "n_util_obs": 661, "util_count_spearman": 0.035408453940787954}, "probe_overhead_sec_total": 338.274, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.024721, "alpha": 0.03, "t": 1978.6}
{"event": "consolidate", "step": 4100, "retired": 1, "split_retired": 0, "merge_admitted": 0, "admitted": 1, "swapped": 6, "uncertain_staged": 262122, "t": 2002.4}
{"event": "train", "step": 4100, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.44199, "bpb": 2.08035, "chunk_loss": 0.030657, "F": 0.21426, "G": 0.21762, "chunks": 219.41, "bytes_per_chunk": 4.667, "boundary_entropy": 0.09593, "router_noise": 0.0, "grad_norm": 0.7091, "lr": 9.75e-05, "router_trainable": true, "sec_per_step": 0.493, "probe_sec": 0.0643, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.03267, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": 0.12743, "n_shadow": 5, "mean_delta_shadow": -0.4341, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 0.60231, "trunc_abs_frac": 0.587, "mean_delta_full": -0.36118, "read_hit_rate": {"surface": 0.59565}, "mean_gate": {"surface": 0.22615}, "n_reads": 4163, "n_cand": 6989, "read_suppressed_frac": 0.1071, "dropout_reads_logged": 4163, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 206.60565185546875, "n_util_obs": 666, "util_count_spearman": 0.026040854705880655}, "probe_overhead_sec_total": 340.932, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.024075, "alpha": 0.03, "t": 2003.2}
{"event": "consolidate", "step": 4150, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 15, "uncertain_staged": 262122, "t": 2030.2}
{"event": "train", "step": 4150, "arch": "ali", "phase": 2, "objective": "global", "lm_loss": 1.38826, "bpb": 2.00284, "chunk_loss": 0.030728, "F": 0.21832, "G": 0.22048, "chunks": 223.56, "bytes_per_chunk": 4.58, "boundary_entropy": 0.09003, "router_noise": 0.0, "grad_norm": 0.6947, "lr": 9.43e-05, "router_trainable": true, "sec_per_step": 0.545, "probe_sec": 0.0322, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02216, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 5, "mean_delta_loo": 0.27186, "n_shadow": 9, "mean_delta_shadow": -0.37147, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 1.21408, "trunc_abs_frac": 1.17332, "mean_delta_full": 0.37634, "read_hit_rate": {"surface": 0.60461}, "mean_gate": {"surface": 0.21977}, "n_reads": 4306, "n_cand": 7122, "read_suppressed_frac": 0.1008, "dropout_reads_logged": 4306, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 209.69847106933594, "n_util_obs": 678, "util_count_spearman": 0.02614222028467218}, "probe_overhead_sec_total": 343.411, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.021122, "alpha": 0.03, "t": 2030.5}
{"event": "consolidate", "step": 4200, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 7, "uncertain_staged": 262122, "t": 2060.6}
{"event": "probe", "step": 4200, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3321.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.190829, "cps": 25.945312, "mean_p": 0.195714, "boundary_entropy": 0.089773, "bpb": 1.922735}, "tr": {"chunks": 3742.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.206706, "cps": 29.234375, "mean_p": 0.19589, "boundary_entropy": 0.118839, "bpb": 2.073555}, "fi": {"chunks": 4895.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.26022, "cps": 38.242188, "mean_p": 0.252066, "boundary_entropy": 0.106295, "bpb": 2.176092}}, "bpb": {"en": 1.92273, "tr": 2.07356, "fi": 2.17609}, "gini_cpc": 0.070331, "gini_cps": 0.087752, "bpic": 4.744272, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0831995635134315, "fi": 1.3636284618299552}, "inventory": {"size": 327673, "admitted": 65536, "staged": 262137, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262137, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 212.73214721679688, "n_util_obs": 686, "util_count_spearman": 0.040820756005885216}, "t": 2061.3}
{"event": "eval", "step": 4200, "phase": 2, "bpb": {"en": 1.93759, "tr": 2.08099, "fi": 2.18574}, "bytes_per_chunk": {"en": 4.943, "tr": 5.306, "fi": 4.086}, "bpb_mean": 2.06811, "frozen_inventory": true, "t": 2062.4}
{"event": "phase_transition", "step": 4200, "phase": 3, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0, "t": 2066.4}
{"event": "train", "step": 4200, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.37677, "bpb": 1.98626, "chunk_loss": 0.030342, "F": 0.20169, "G": 0.205, "chunks": 206.53, "bytes_per_chunk": 4.958, "boundary_entropy": 0.09075, "router_noise": 0.0, "grad_norm": 0.6651, "lr": 9.12e-05, "router_trainable": true, "sec_per_step": 0.726, "probe_sec": 0.0845, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.06463, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": 0.23694, "n_shadow": 5, "mean_delta_shadow": 0.05192, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 0.59768, "trunc_abs_frac": 0.58858, "mean_delta_full": 0.52437, "read_hit_rate": {"surface": 0.5799}, "mean_gate": {"surface": 0.219}, "n_reads": 3814, "n_cand": 6577, "read_suppressed_frac": 0.0939, "dropout_reads_logged": 3814, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 212.79034423828125, "n_util_obs": 686, "util_count_spearman": 0.04127666309228825}, "probe_overhead_sec_total": 346.812, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.02302, "alpha": 0.03, "t": 2066.8}
{"event": "consolidate", "step": 4250, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 8, "uncertain_staged": 262122, "t": 2087.3}
{"event": "train", "step": 4250, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.38228, "bpb": 1.99421, "chunk_loss": 0.030373, "F": 0.19949, "G": 0.20311, "chunks": 204.28, "bytes_per_chunk": 5.013, "boundary_entropy": 0.09422, "router_noise": 0.0, "grad_norm": 0.7045, "lr": 8.81e-05, "router_trainable": true, "sec_per_step": 0.425, "probe_sec": 0.0361, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02218, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 12, "mean_delta_loo": 0.10565, "n_shadow": 2, "mean_delta_shadow": 0.22713, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": 0.10001, "trunc_abs_frac": 1.22124, "mean_delta_full": 0.15789, "read_hit_rate": {"surface": 0.59293}, "mean_gate": {"surface": 0.23718}, "n_reads": 3857, "n_cand": 6505, "read_suppressed_frac": 0.1034, "dropout_reads_logged": 3857, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 215.8688201904297, "n_util_obs": 694, "util_count_spearman": 0.04118849006141893}, "probe_overhead_sec_total": 349.31, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.023061, "alpha": 0.03, "t": 2088.0}
{"event": "consolidate", "step": 4300, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 31, "uncertain_staged": 262122, "t": 2132.5}
{"event": "train", "step": 4300, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.47005, "bpb": 2.12084, "chunk_loss": 0.030476, "F": 0.20587, "G": 0.20995, "chunks": 210.81, "bytes_per_chunk": 4.857, "boundary_entropy": 0.09805, "router_noise": 0.0, "grad_norm": 0.8108, "lr": 8.5e-05, "router_trainable": true, "sec_per_step": 0.896, "probe_sec": 0.0987, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.08305, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 10, "mean_delta_loo": 0.09759, "n_shadow": 4, "mean_delta_shadow": 0.02292, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 0.58514, "trunc_abs_frac": 0.43891, "mean_delta_full": -0.61578, "read_hit_rate": {"surface": 0.58698}, "mean_gate": {"surface": 0.24028}, "n_reads": 3941, "n_cand": 6714, "read_suppressed_frac": 0.1068, "dropout_reads_logged": 3941, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 218.82730102539062, "n_util_obs": 706, "util_count_spearman": 0.04213143303964367}, "probe_overhead_sec_total": 352.831, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.028385, "alpha": 0.03, "t": 2132.8}
{"event": "consolidate", "step": 4350, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 62, "uncertain_staged": 262122, "t": 2176.0}
{"event": "train", "step": 4350, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.32274, "bpb": 1.90831, "chunk_loss": 0.030477, "F": 0.20441, "G": 0.20742, "chunks": 209.31, "bytes_per_chunk": 4.892, "boundary_entropy": 0.08692, "router_noise": 0.0, "grad_norm": 0.717, "lr": 8.21e-05, "router_trainable": true, "sec_per_step": 0.87, "probe_sec": 0.1, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.08499, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": 0.22756, "n_shadow": 6, "mean_delta_shadow": -0.06115, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 6, "trunc_ratio_sum": -0.33903, "trunc_abs_frac": 0.82289, "mean_delta_full": 0.13409, "read_hit_rate": {"surface": 0.58416}, "mean_gate": {"surface": 0.22186}, "n_reads": 3894, "n_cand": 6666, "read_suppressed_frac": 0.0907, "dropout_reads_logged": 3894, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 221.75523376464844, "n_util_obs": 709, "util_count_spearman": 0.04097845973499167}, "probe_overhead_sec_total": 356.31, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.027188, "alpha": 0.03, "t": 2176.3}
{"event": "consolidate", "step": 4400, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 16, "uncertain_staged": 262122, "t": 2213.0}
{"event": "probe", "step": 4400, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3325.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.191059, "cps": 25.976562, "mean_p": 0.196147, "boundary_entropy": 0.086784, "bpb": 1.907677}, "tr": {"chunks": 3769.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.208198, "cps": 29.445312, "mean_p": 0.197404, "boundary_entropy": 0.116759, "bpb": 2.056708}, "fi": {"chunks": 4925.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.261815, "cps": 38.476562, "mean_p": 0.252897, "boundary_entropy": 0.100739, "bpb": 2.160863}}, "bpb": {"en": 1.90768, "tr": 2.05671, "fi": 2.16086}, "gini_cpc": 0.071355, "gini_cps": 0.088748, "bpic": 4.720193, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0897027743085828, "fi": 1.3703352261894437}, "inventory": {"size": 327664, "admitted": 65536, "staged": 262128, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262128, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 224.7609100341797, "n_util_obs": 715, "util_count_spearman": 0.04621468714089487}, "t": 2213.8}
{"event": "eval", "step": 4400, "phase": 3, "bpb": {"en": 1.92424, "tr": 2.06326, "fi": 2.16968}, "bytes_per_chunk": {"en": 4.914, "tr": 5.295, "fi": 4.065}, "bpb_mean": 2.05239, "frozen_inventory": true, "t": 2214.8}
{"event": "train", "step": 4400, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.43298, "bpb": 2.06735, "chunk_loss": 0.030734, "F": 0.21399, "G": 0.21686, "chunks": 219.12, "bytes_per_chunk": 4.673, "boundary_entropy": 0.08909, "router_noise": 0.0, "grad_norm": 0.7227, "lr": 7.92e-05, "router_trainable": true, "sec_per_step": 0.867, "probe_sec": 0.0484, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02534, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 4, "mean_delta_loo": 0.04488, "n_shadow": 10, "mean_delta_shadow": -0.12457, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 0.7891, "trunc_abs_frac": 0.87639, "mean_delta_full": -0.57166, "read_hit_rate": {"surface": 0.59842}, "mean_gate": {"surface": 0.22251}, "n_reads": 4177, "n_cand": 6980, "read_suppressed_frac": 0.0984, "dropout_reads_logged": 4177, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 224.82464599609375, "n_util_obs": 715, "util_count_spearman": 0.04600070079450398}, "probe_overhead_sec_total": 359.242, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.025303, "alpha": 0.03, "t": 2219.6}
{"event": "consolidate", "step": 4450, "retired": 1, "split_retired": 0, "merge_admitted": 0, "admitted": 1, "swapped": 15, "uncertain_staged": 262122, "t": 2245.0}
{"event": "train", "step": 4450, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.37623, "bpb": 1.98547, "chunk_loss": 0.03053, "F": 0.2085, "G": 0.21059, "chunks": 213.5, "bytes_per_chunk": 4.796, "boundary_entropy": 0.08773, "router_noise": 0.0, "grad_norm": 0.6715, "lr": 7.63e-05, "router_trainable": true, "sec_per_step": 0.53, "probe_sec": 0.0688, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04091, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 11, "mean_delta_loo": 0.41577, "n_shadow": 3, "mean_delta_shadow": -0.09232, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 0.79717, "trunc_abs_frac": 0.50467, "mean_delta_full": 0.25363, "read_hit_rate": {"surface": 0.59368}, "mean_gate": {"surface": 0.22554}, "n_reads": 4037, "n_cand": 6800, "read_suppressed_frac": 0.1068, "dropout_reads_logged": 4037, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 227.81060791015625, "n_util_obs": 725, "util_count_spearman": 0.041937453879934986}, "probe_overhead_sec_total": 362.069, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.025484, "alpha": 0.03, "t": 2246.1}
{"event": "consolidate", "step": 4500, "retired": 1, "split_retired": 0, "merge_admitted": 0, "admitted": 1, "swapped": 28, "uncertain_staged": 262122, "t": 2279.1}
{"event": "train", "step": 4500, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.3877, "bpb": 2.00202, "chunk_loss": 0.030613, "F": 0.21219, "G": 0.21442, "chunks": 217.28, "bytes_per_chunk": 4.713, "boundary_entropy": 0.09131, "router_noise": 0.0, "grad_norm": 0.7283, "lr": 7.36e-05, "router_trainable": true, "sec_per_step": 0.676, "probe_sec": 0.0752, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.06135, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 10, "mean_delta_loo": -0.06584, "n_shadow": 4, "mean_delta_shadow": -0.29417, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 2.02093, "trunc_abs_frac": 0.77418, "mean_delta_full": -0.19299, "read_hit_rate": {"surface": 0.60338}, "mean_gate": {"surface": 0.22427}, "n_reads": 4176, "n_cand": 6921, "read_suppressed_frac": 0.0999, "dropout_reads_logged": 4176, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 230.73495483398438, "n_util_obs": 728, "util_count_spearman": 0.03561162308852148}, "probe_overhead_sec_total": 365.139, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.023263, "alpha": 0.03, "t": 2280.0}
{"event": "consolidate", "step": 4550, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 15, "uncertain_staged": 262123, "t": 2312.9}
{"event": "train", "step": 4550, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.36339, "bpb": 1.96696, "chunk_loss": 0.030537, "F": 0.21002, "G": 0.21239, "chunks": 215.06, "bytes_per_chunk": 4.761, "boundary_entropy": 0.09161, "router_noise": 0.0, "grad_norm": 0.6629, "lr": 7.09e-05, "router_trainable": true, "sec_per_step": 0.678, "probe_sec": 0.0498, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02481, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": 0.27652, "n_shadow": 8, "mean_delta_shadow": -0.36622, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 1.47827, "trunc_abs_frac": 2.6968, "mean_delta_full": 0.18396, "read_hit_rate": {"surface": 0.62015}, "mean_gate": {"surface": 0.22331}, "n_reads": 4248, "n_cand": 6850, "read_suppressed_frac": 0.0909, "dropout_reads_logged": 4248, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 233.79273986816406, "n_util_obs": 733, "util_count_spearman": 0.05433826849280181}, "probe_overhead_sec_total": 367.804, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.02666, "alpha": 0.03, "t": 2313.9}
{"event": "consolidate", "step": 4600, "retired": 1, "split_retired": 0, "merge_admitted": 0, "admitted": 1, "swapped": 15, "uncertain_staged": 262123, "t": 2346.3}
{"event": "probe", "step": 4600, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3299.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.189565, "cps": 25.773438, "mean_p": 0.194435, "boundary_entropy": 0.086205, "bpb": 1.898165}, "tr": {"chunks": 3782.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.208916, "cps": 29.546875, "mean_p": 0.194934, "boundary_entropy": 0.12052, "bpb": 2.04206}, "fi": {"chunks": 4888.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.259848, "cps": 38.1875, "mean_p": 0.252113, "boundary_entropy": 0.106667, "bpb": 2.149746}}, "bpb": {"en": 1.89817, "tr": 2.04206, "fi": 2.14975}, "gini_cpc": 0.071173, "gini_cps": 0.088506, "bpic": 4.739911, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.1020791286638612, "fi": 1.3707590392514917}, "inventory": {"size": 327664, "admitted": 65536, "staged": 262128, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262128, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 236.76089477539062, "n_util_obs": 737, "util_count_spearman": 0.03901798421205467}, "t": 2346.9}
{"event": "eval", "step": 4600, "phase": 3, "bpb": {"en": 1.91426, "tr": 2.04872, "fi": 2.15986}, "bytes_per_chunk": {"en": 4.951, "tr": 5.293, "fi": 4.088}, "bpb_mean": 2.04095, "frozen_inventory": true, "t": 2347.8}
{"event": "train", "step": 4600, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.42237, "bpb": 2.05205, "chunk_loss": 0.03046, "F": 0.20361, "G": 0.2062, "chunks": 208.5, "bytes_per_chunk": 4.911, "boundary_entropy": 0.0921, "router_noise": 0.0, "grad_norm": 0.7288, "lr": 6.82e-05, "router_trainable": true, "sec_per_step": 0.762, "probe_sec": 0.0357, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02128, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": -0.13428, "n_shadow": 8, "mean_delta_shadow": 0.00511, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": -0.18102, "trunc_abs_frac": 0.35434, "mean_delta_full": 0.08295, "read_hit_rate": {"surface": 0.58705}, "mean_gate": {"surface": 0.23028}, "n_reads": 3898, "n_cand": 6640, "read_suppressed_frac": 0.0954, "dropout_reads_logged": 3898, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 236.82037353515625, "n_util_obs": 737, "util_count_spearman": 0.03915325481437121}, "probe_overhead_sec_total": 370.564, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.022404, "alpha": 0.03, "t": 2352.0}
{"event": "consolidate", "step": 4650, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 7, "uncertain_staged": 262123, "t": 2389.4}
{"event": "train", "step": 4650, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.34034, "bpb": 1.9337, "chunk_loss": 0.030441, "F": 0.20557, "G": 0.20737, "chunks": 210.5, "bytes_per_chunk": 4.865, "boundary_entropy": 0.09167, "router_noise": 0.0, "grad_norm": 0.6721, "lr": 6.57e-05, "router_trainable": true, "sec_per_step": 0.771, "probe_sec": 0.0543, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.03217, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": 0.21561, "n_shadow": 6, "mean_delta_shadow": 0.09931, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 1.04184, "trunc_abs_frac": 1.16215, "mean_delta_full": 0.18551, "read_hit_rate": {"surface": 0.60889}, "mean_gate": {"surface": 0.23055}, "n_reads": 4082, "n_cand": 6704, "read_suppressed_frac": 0.0982, "dropout_reads_logged": 4082, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 239.89610290527344, "n_util_obs": 747, "util_count_spearman": 0.041008472105576906}, "probe_overhead_sec_total": 374.301, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.023982, "alpha": 0.03, "t": 2390.5}
{"event": "consolidate", "step": 4700, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 9, "uncertain_staged": 262123, "t": 2457.9}
{"event": "train", "step": 4700, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.35766, "bpb": 1.9587, "chunk_loss": 0.030483, "F": 0.20572, "G": 0.20759, "chunks": 210.66, "bytes_per_chunk": 4.861, "boundary_entropy": 0.0885, "router_noise": 0.0, "grad_norm": 0.6921, "lr": 6.32e-05, "router_trainable": true, "sec_per_step": 1.374, "probe_sec": 0.1149, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.0722, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": -0.05881, "n_shadow": 8, "mean_delta_shadow": -0.01885, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 0.59225, "trunc_abs_frac": 0.80646, "mean_delta_full": -0.68958, "read_hit_rate": {"surface": 0.60098}, "mean_gate": {"surface": 0.22381}, "n_reads": 4032, "n_cand": 6709, "read_suppressed_frac": 0.1044, "dropout_reads_logged": 4032, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 242.95877075195312, "n_util_obs": 754, "util_count_spearman": 0.0396256929037049}, "probe_overhead_sec_total": 378.314, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.025132, "alpha": 0.03, "t": 2459.2}
{"event": "consolidate", "step": 4750, "retired": 1, "split_retired": 0, "merge_admitted": 0, "admitted": 1, "swapped": 8, "uncertain_staged": 262123, "t": 2521.9}
{"event": "train", "step": 4750, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.31879, "bpb": 1.90261, "chunk_loss": 0.030468, "F": 0.20746, "G": 0.20968, "chunks": 212.44, "bytes_per_chunk": 4.82, "boundary_entropy": 0.08901, "router_noise": 0.0, "grad_norm": 0.6572, "lr": 6.08e-05, "router_trainable": true, "sec_per_step": 1.279, "probe_sec": 0.0579, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02429, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": -0.11578, "n_shadow": 8, "mean_delta_shadow": -0.41359, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 0.44774, "trunc_abs_frac": 1.01344, "mean_delta_full": -0.0704, "read_hit_rate": {"surface": 0.60272}, "mean_gate": {"surface": 0.22245}, "n_reads": 4078, "n_cand": 6766, "read_suppressed_frac": 0.1008, "dropout_reads_logged": 4078, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 246.02337646484375, "n_util_obs": 758, "util_count_spearman": 0.03632347259513404}, "probe_overhead_sec_total": 382.614, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.026091, "alpha": 0.03, "t": 2523.2}
{"event": "consolidate", "step": 4800, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 7, "uncertain_staged": 262123, "t": 2587.7}
{"event": "probe", "step": 4800, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3317.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.190599, "cps": 25.914062, "mean_p": 0.194829, "boundary_entropy": 0.081308, "bpb": 1.887526}, "tr": {"chunks": 3715.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.205215, "cps": 29.023438, "mean_p": 0.191516, "boundary_entropy": 0.118898, "bpb": 2.034457}, "fi": {"chunks": 4906.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.260805, "cps": 38.328125, "mean_p": 0.251426, "boundary_entropy": 0.101106, "bpb": 2.138889}}, "bpb": {"en": 1.88753, "tr": 2.03446, "fi": 2.13889}, "gini_cpc": 0.07128, "gini_cps": 0.088736, "bpic": 4.75222, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0766806681580265, "fi": 1.3683409025312756}, "inventory": {"size": 327673, "admitted": 65536, "staged": 262137, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262137, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 249.04830932617188, "n_util_obs": 764, "util_count_spearman": 0.03958084879460537}, "t": 2588.9}
{"event": "eval", "step": 4800, "phase": 3, "bpb": {"en": 1.90306, "tr": 2.04018, "fi": 2.14823}, "bytes_per_chunk": {"en": 4.937, "tr": 5.354, "fi": 4.071}, "bpb_mean": 2.03049, "frozen_inventory": true, "t": 2590.6}
{"event": "train", "step": 4800, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.39082, "bpb": 2.00653, "chunk_loss": 0.030763, "F": 0.21808, "G": 0.21992, "chunks": 223.31, "bytes_per_chunk": 4.586, "boundary_entropy": 0.08819, "router_noise": 0.0, "grad_norm": 0.7187, "lr": 5.85e-05, "router_trainable": true, "sec_per_step": 1.506, "probe_sec": 0.0573, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02435, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 10, "mean_delta_loo": 0.1029, "n_shadow": 4, "mean_delta_shadow": -0.20956, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": -0.72571, "trunc_abs_frac": 0.30265, "mean_delta_full": -0.1156, "read_hit_rate": {"surface": 0.62412}, "mean_gate": {"surface": 0.21687}, "n_reads": 4440, "n_cand": 7114, "read_suppressed_frac": 0.0975, "dropout_reads_logged": 4440, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 249.11605834960938, "n_util_obs": 764, "util_count_spearman": 0.03937193892308236}, "probe_overhead_sec_total": 387.939, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.028422, "alpha": 0.03, "t": 2598.4}
{"event": "consolidate", "step": 4850, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 21, "uncertain_staged": 262123, "t": 2659.0}
{"event": "train", "step": 4850, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.34399, "bpb": 1.93897, "chunk_loss": 0.030622, "F": 0.21143, "G": 0.21344, "chunks": 216.5, "bytes_per_chunk": 4.73, "boundary_entropy": 0.087, "router_noise": 0.0, "grad_norm": 0.7419, "lr": 5.62e-05, "router_trainable": true, "sec_per_step": 1.233, "probe_sec": 0.0513, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02206, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 5, "mean_delta_loo": 0.23989, "n_shadow": 9, "mean_delta_shadow": -0.0567, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 6, "trunc_ratio_sum": 0.98653, "trunc_abs_frac": 0.80229, "mean_delta_full": 0.1191, "read_hit_rate": {"surface": 0.60557}, "mean_gate": {"surface": 0.21809}, "n_reads": 4176, "n_cand": 6896, "read_suppressed_frac": 0.0934, "dropout_reads_logged": 4176, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 252.1784210205078, "n_util_obs": 769, "util_count_spearman": 0.040221997838048716}, "probe_overhead_sec_total": 391.959, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.026823, "alpha": 0.03, "t": 2660.1}
{"event": "consolidate", "step": 4900, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 8, "uncertain_staged": 262123, "t": 2737.6}
{"event": "train", "step": 4900, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.35192, "bpb": 1.9504, "chunk_loss": 0.030349, "F": 0.19992, "G": 0.20322, "chunks": 204.72, "bytes_per_chunk": 5.002, "boundary_entropy": 0.08332, "router_noise": 0.0, "grad_norm": 0.7233, "lr": 5.41e-05, "router_trainable": true, "sec_per_step": 1.588, "probe_sec": 0.0662, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04853, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": 0.02603, "n_shadow": 7, "mean_delta_shadow": -0.01625, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 0.14779, "trunc_abs_frac": 0.40527, "mean_delta_full": -0.04218, "read_hit_rate": {"surface": 0.58353}, "mean_gate": {"surface": 0.21348}, "n_reads": 3804, "n_cand": 6519, "read_suppressed_frac": 0.1088, "dropout_reads_logged": 3804, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 255.23635864257812, "n_util_obs": 779, "util_count_spearman": 0.045828597024184785}, "probe_overhead_sec_total": 396.24, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.025304, "alpha": 0.03, "t": 2739.5}
{"event": "consolidate", "step": 4950, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 25, "uncertain_staged": 262123, "t": 2821.6}
{"event": "train", "step": 4950, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.33208, "bpb": 1.92178, "chunk_loss": 0.030544, "F": 0.20935, "G": 0.21159, "chunks": 214.38, "bytes_per_chunk": 4.777, "boundary_entropy": 0.08128, "router_noise": 0.0, "grad_norm": 0.6944, "lr": 5.2e-05, "router_trainable": true, "sec_per_step": 1.651, "probe_sec": 0.1277, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.11725, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 12, "mean_delta_loo": 0.19024, "n_shadow": 2, "mean_delta_shadow": 0.06318, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 2.05209, "trunc_abs_frac": 0.88152, "mean_delta_full": 0.05937, "read_hit_rate": {"surface": 0.58905}, "mean_gate": {"surface": 0.20305}, "n_reads": 4022, "n_cand": 6828, "read_suppressed_frac": 0.1052, "dropout_reads_logged": 4022, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 258.2554016113281, "n_util_obs": 791, "util_count_spearman": 0.05848827678377823}, "probe_overhead_sec_total": 400.956, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.023407, "alpha": 0.03, "t": 2822.0}
{"event": "consolidate", "step": 5000, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 5, "uncertain_staged": 262123, "t": 2883.1}
{"event": "probe", "step": 5000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3311.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.190255, "cps": 25.867188, "mean_p": 0.194301, "boundary_entropy": 0.07874, "bpb": 1.879288}, "tr": {"chunks": 3742.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.206706, "cps": 29.234375, "mean_p": 0.192121, "boundary_entropy": 0.11737, "bpb": 2.021351}, "fi": {"chunks": 4899.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.260433, "cps": 38.273438, "mean_p": 0.251441, "boundary_entropy": 0.099586, "bpb": 2.122844}}, "bpb": {"en": 1.87929, "tr": 2.02135, "fi": 2.12284}, "gini_cpc": 0.071168, "gini_cps": 0.088577, "bpic": 4.746653, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0864710813736351, "fi": 1.3688646096193386}, "inventory": {"size": 327675, "admitted": 65536, "staged": 262139, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262139, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 261.254638671875, "n_util_obs": 797, "util_count_spearman": 0.05977670031847704}, "t": 2883.6}
{"event": "eval", "step": 5000, "phase": 3, "bpb": {"en": 1.89606, "tr": 2.02752, "fi": 2.13183}, "bytes_per_chunk": {"en": 4.944, "tr": 5.276, "fi": 4.074}, "bpb_mean": 2.01847, "frozen_inventory": true, "t": 2884.5}
{"event": "train", "step": 5000, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.33693, "bpb": 1.92879, "chunk_loss": 0.030541, "F": 0.20969, "G": 0.21114, "chunks": 214.72, "bytes_per_chunk": 4.769, "boundary_entropy": 0.08381, "router_noise": 0.0, "grad_norm": 0.6996, "lr": 5e-05, "router_trainable": true, "sec_per_step": 1.395, "probe_sec": 0.0523, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02348, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 7, "mean_delta_loo": 0.28378, "n_shadow": 7, "mean_delta_shadow": -0.04467, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 1.86356, "trunc_abs_frac": 1.45536, "mean_delta_full": 0.19067, "read_hit_rate": {"surface": 0.61105}, "mean_gate": {"surface": 0.21845}, "n_reads": 4179, "n_cand": 6839, "read_suppressed_frac": 0.0957, "dropout_reads_logged": 4179, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 261.3183898925781, "n_util_obs": 797, "util_count_spearman": 0.05989138825780291}, "probe_overhead_sec_total": 404.292, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.023929, "alpha": 0.03, "t": 2891.8}
{"event": "consolidate", "step": 5050, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 6, "uncertain_staged": 262123, "t": 2991.9}
{"event": "train", "step": 5050, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.35181, "bpb": 1.95025, "chunk_loss": 0.030443, "F": 0.20587, "G": 0.20869, "chunks": 210.81, "bytes_per_chunk": 4.857, "boundary_entropy": 0.08506, "router_noise": 0.0, "grad_norm": 0.6733, "lr": 4.81e-05, "router_trainable": true, "sec_per_step": 2.031, "probe_sec": 0.0804, "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": 8, "mean_delta_loo": 0.07255, "n_shadow": 6, "mean_delta_shadow": -0.09151, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 0.92969, "trunc_abs_frac": 1.01614, "mean_delta_full": 0.36938, "read_hit_rate": {"surface": 0.59428}, "mean_gate": {"surface": 0.21463}, "n_reads": 3990, "n_cand": 6714, "read_suppressed_frac": 0.1035, "dropout_reads_logged": 3990, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 264.3305969238281, "n_util_obs": 811, "util_count_spearman": 0.05388034463686493}, "probe_overhead_sec_total": 409.328, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.025791, "alpha": 0.03, "t": 2993.4}
{"event": "consolidate", "step": 5100, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 5, "uncertain_staged": 262123, "t": 3087.5}
{"event": "train", "step": 5100, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.3204, "bpb": 1.90493, "chunk_loss": 0.030279, "F": 0.19748, "G": 0.1998, "chunks": 202.22, "bytes_per_chunk": 5.064, "boundary_entropy": 0.08086, "router_noise": 0.0, "grad_norm": 0.689, "lr": 4.63e-05, "router_trainable": true, "sec_per_step": 1.908, "probe_sec": 0.0337, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.01867, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": -0.23576, "n_shadow": 6, "mean_delta_shadow": -0.10058, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 6, "trunc_ratio_sum": 1.34424, "trunc_abs_frac": 2.4691, "mean_delta_full": -0.30924, "read_hit_rate": {"surface": 0.57788}, "mean_gate": {"surface": 0.20911}, "n_reads": 3721, "n_cand": 6439, "read_suppressed_frac": 0.1048, "dropout_reads_logged": 3721, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 267.38433837890625, "n_util_obs": 817, "util_count_spearman": 0.0557438444143314}, "probe_overhead_sec_total": 414.501, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.026252, "alpha": 0.03, "t": 3088.8}
{"event": "consolidate", "step": 5150, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 12, "uncertain_staged": 262123, "t": 3191.8}
{"event": "train", "step": 5150, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.3101, "bpb": 1.89007, "chunk_loss": 0.030493, "F": 0.20773, "G": 0.20966, "chunks": 212.72, "bytes_per_chunk": 4.814, "boundary_entropy": 0.08545, "router_noise": 0.0, "grad_norm": 0.7009, "lr": 4.45e-05, "router_trainable": true, "sec_per_step": 2.07, "probe_sec": 0.0826, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.06484, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": 0.00774, "n_shadow": 6, "mean_delta_shadow": -0.46451, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 1.06901, "trunc_abs_frac": 1.22692, "mean_delta_full": -0.52759, "read_hit_rate": {"surface": 0.60694}, "mean_gate": {"surface": 0.20536}, "n_reads": 4112, "n_cand": 6775, "read_suppressed_frac": 0.1063, "dropout_reads_logged": 4112, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 270.30133056640625, "n_util_obs": 820, "util_count_spearman": 0.05492602883094732}, "probe_overhead_sec_total": 418.959, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.027663, "alpha": 0.03, "t": 3192.3}
{"event": "consolidate", "step": 5200, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 18, "uncertain_staged": 262123, "t": 3248.4}
{"event": "probe", "step": 5200, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3306.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.189967, "cps": 25.828125, "mean_p": 0.19402, "boundary_entropy": 0.079679, "bpb": 1.871058}, "tr": {"chunks": 3728.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.205933, "cps": 29.125, "mean_p": 0.192131, "boundary_entropy": 0.116599, "bpb": 2.011889}, "fi": {"chunks": 4895.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.26022, "cps": 38.242188, "mean_p": 0.251938, "boundary_entropy": 0.101052, "bpb": 2.114937}}, "bpb": {"en": 1.87106, "tr": 2.01189, "fi": 2.11494}, "gini_cpc": 0.071382, "gini_cps": 0.088803, "bpic": 4.755805, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.084043284079315, "fi": 1.3698155238164795}, "inventory": {"size": 327662, "admitted": 65536, "staged": 262126, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262126, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 273.259765625, "n_util_obs": 826, "util_count_spearman": 0.06749420802460783}, "t": 3249.0}
{"event": "eval", "step": 5200, "phase": 3, "bpb": {"en": 1.88781, "tr": 2.01787, "fi": 2.12391}, "bytes_per_chunk": {"en": 4.955, "tr": 5.321, "fi": 4.076}, "bpb_mean": 2.00986, "frozen_inventory": true, "t": 3249.9}
{"event": "train", "step": 5200, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.35143, "bpb": 1.94971, "chunk_loss": 0.03068, "F": 0.21442, "G": 0.21614, "chunks": 219.56, "bytes_per_chunk": 4.664, "boundary_entropy": 0.08598, "router_noise": 0.0, "grad_norm": 0.6717, "lr": 4.29e-05, "router_trainable": true, "sec_per_step": 1.319, "probe_sec": 0.1317, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.05347, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 10, "mean_delta_loo": 0.2194, "n_shadow": 4, "mean_delta_shadow": 0.01384, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 1.87821, "trunc_abs_frac": 10.1953, "mean_delta_full": 0.18988, "read_hit_rate": {"surface": 0.60566}, "mean_gate": {"surface": 0.21292}, "n_reads": 4236, "n_cand": 6994, "read_suppressed_frac": 0.0932, "dropout_reads_logged": 4236, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 273.32440185546875, "n_util_obs": 826, "util_count_spearman": 0.06806071710389419}, "probe_overhead_sec_total": 422.469, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.028035, "alpha": 0.03, "t": 3258.2}
{"event": "consolidate", "step": 5250, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 13, "uncertain_staged": 262123, "t": 3348.8}
{"event": "train", "step": 5250, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.37709, "bpb": 1.98673, "chunk_loss": 0.030535, "F": 0.21103, "G": 0.21311, "chunks": 216.09, "bytes_per_chunk": 4.739, "boundary_entropy": 0.08523, "router_noise": 0.0, "grad_norm": 0.686, "lr": 4.14e-05, "router_trainable": true, "sec_per_step": 1.834, "probe_sec": 0.039, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02629, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": -0.07204, "n_shadow": 5, "mean_delta_shadow": -0.14487, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 0.58407, "trunc_abs_frac": 0.60777, "mean_delta_full": 0.86829, "read_hit_rate": {"surface": 0.60802}, "mean_gate": {"surface": 0.21311}, "n_reads": 4185, "n_cand": 6883, "read_suppressed_frac": 0.0994, "dropout_reads_logged": 4185, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 276.40234375, "n_util_obs": 833, "util_count_spearman": 0.06370273903532135}, "probe_overhead_sec_total": 427.302, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.027456, "alpha": 0.03, "t": 3349.9}
{"event": "consolidate", "step": 5300, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 9, "uncertain_staged": 262123, "t": 3442.0}
{"event": "train", "step": 5300, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.36887, "bpb": 1.97487, "chunk_loss": 0.030451, "F": 0.20618, "G": 0.20832, "chunks": 211.12, "bytes_per_chunk": 4.85, "boundary_entropy": 0.08257, "router_noise": 0.0, "grad_norm": 0.6757, "lr": 3.99e-05, "router_trainable": true, "sec_per_step": 1.867, "probe_sec": 0.0617, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02736, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": 0.07422, "n_shadow": 5, "mean_delta_shadow": -0.01222, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": -0.65388, "trunc_abs_frac": 0.60055, "mean_delta_full": -0.09054, "read_hit_rate": {"surface": 0.59295}, "mean_gate": {"surface": 0.21186}, "n_reads": 3987, "n_cand": 6724, "read_suppressed_frac": 0.0968, "dropout_reads_logged": 3987, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 279.4990234375, "n_util_obs": 839, "util_count_spearman": 0.06174915463265145}, "probe_overhead_sec_total": 431.533, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.02431, "alpha": 0.03, "t": 3443.2}
{"event": "consolidate", "step": 5350, "retired": 1, "split_retired": 0, "merge_admitted": 0, "admitted": 1, "swapped": 6, "uncertain_staged": 262123, "t": 3505.6}
{"event": "train", "step": 5350, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.3336, "bpb": 1.92398, "chunk_loss": 0.0307, "F": 0.21008, "G": 0.21134, "chunks": 215.12, "bytes_per_chunk": 4.76, "boundary_entropy": 0.08407, "router_noise": 0.0, "grad_norm": 0.7403, "lr": 3.86e-05, "router_trainable": true, "sec_per_step": 1.269, "probe_sec": 0.0819, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.034, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 10, "mean_delta_loo": 0.1008, "n_shadow": 4, "mean_delta_shadow": 0.06652, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 1.00573, "trunc_abs_frac": 1.06731, "mean_delta_full": -0.85162, "read_hit_rate": {"surface": 0.60537}, "mean_gate": {"surface": 0.22056}, "n_reads": 4148, "n_cand": 6852, "read_suppressed_frac": 0.1, "dropout_reads_logged": 4148, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 282.5101318359375, "n_util_obs": 848, "util_count_spearman": 0.06105056364743154}, "probe_overhead_sec_total": 435.887, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.025841, "alpha": 0.03, "t": 3506.7}
{"event": "consolidate", "step": 5400, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 10, "uncertain_staged": 262123, "t": 3569.8}
{"event": "probe", "step": 5400, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3312.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.190312, "cps": 25.875, "mean_p": 0.194203, "boundary_entropy": 0.078312, "bpb": 1.867909}, "tr": {"chunks": 3713.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.205104, "cps": 29.007812, "mean_p": 0.189749, "boundary_entropy": 0.115877, "bpb": 2.005697}, "fi": {"chunks": 4892.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.260061, "cps": 38.21875, "mean_p": 0.251201, "boundary_entropy": 0.098779, "bpb": 2.106957}}, "bpb": {"en": 1.86791, "tr": 2.0057, "fi": 2.10696}, "gini_cpc": 0.070939, "gini_cps": 0.088389, "bpic": 4.760594, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0777255771523178, "fi": 1.3664959756047763}, "inventory": {"size": 327670, "admitted": 65536, "staged": 262134, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262134, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 285.45855712890625, "n_util_obs": 854, "util_count_spearman": 0.051731075037488555}, "t": 3570.7}
{"event": "eval", "step": 5400, "phase": 3, "bpb": {"en": 1.88411, "tr": 2.01145, "fi": 2.11568}, "bytes_per_chunk": {"en": 4.957, "tr": 5.352, "fi": 4.079}, "bpb_mean": 2.00375, "frozen_inventory": true, "t": 3571.8}
{"event": "train", "step": 5400, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.25803, "bpb": 1.81496, "chunk_loss": 0.030407, "F": 0.19968, "G": 0.20107, "chunks": 204.47, "bytes_per_chunk": 5.008, "boundary_entropy": 0.08275, "router_noise": 0.0, "grad_norm": 0.6511, "lr": 3.73e-05, "router_trainable": true, "sec_per_step": 1.461, "probe_sec": 0.0474, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02623, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": 0.11143, "n_shadow": 6, "mean_delta_shadow": -0.10172, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": 0.27547, "trunc_abs_frac": 0.35005, "mean_delta_full": -0.53687, "read_hit_rate": {"surface": 0.59806}, "mean_gate": {"surface": 0.21363}, "n_reads": 3894, "n_cand": 6511, "read_suppressed_frac": 0.0999, "dropout_reads_logged": 3894, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 285.5179748535156, "n_util_obs": 855, "util_count_spearman": 0.053457523984514796}, "probe_overhead_sec_total": 440.136, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.025181, "alpha": 0.03, "t": 3579.7}
{"event": "consolidate", "step": 5450, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 6, "uncertain_staged": 262124, "t": 3642.2}
{"event": "train", "step": 5450, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.3079, "bpb": 1.8869, "chunk_loss": 0.030495, "F": 0.20908, "G": 0.21026, "chunks": 214.09, "bytes_per_chunk": 4.783, "boundary_entropy": 0.08083, "router_noise": 0.0, "grad_norm": 0.6827, "lr": 3.62e-05, "router_trainable": true, "sec_per_step": 1.285, "probe_sec": 0.0778, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02866, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 11, "mean_delta_loo": 0.03949, "n_shadow": 3, "mean_delta_shadow": -0.11552, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 4, "trunc_ratio_sum": 1.20299, "trunc_abs_frac": 0.72195, "mean_delta_full": -0.24994, "read_hit_rate": {"surface": 0.60243}, "mean_gate": {"surface": 0.20613}, "n_reads": 4108, "n_cand": 6819, "read_suppressed_frac": 0.0923, "dropout_reads_logged": 4108, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 288.51971435546875, "n_util_obs": 859, "util_count_spearman": 0.05349991737963057}, "probe_overhead_sec_total": 444.114, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.022983, "alpha": 0.03, "t": 3643.9}
{"event": "consolidate", "step": 5500, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 6, "uncertain_staged": 262124, "t": 3697.6}
{"event": "train", "step": 5500, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.31867, "bpb": 1.90244, "chunk_loss": 0.030415, "F": 0.20367, "G": 0.20594, "chunks": 208.56, "bytes_per_chunk": 4.91, "boundary_entropy": 0.08303, "router_noise": 0.0, "grad_norm": 0.6733, "lr": 3.51e-05, "router_trainable": true, "sec_per_step": 1.077, "probe_sec": 0.04, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.0338, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": 0.02524, "n_shadow": 6, "mean_delta_shadow": -0.00996, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": 4.50822, "trunc_abs_frac": 1.22827, "mean_delta_full": -0.01484, "read_hit_rate": {"surface": 0.5935}, "mean_gate": {"surface": 0.21136}, "n_reads": 3942, "n_cand": 6642, "read_suppressed_frac": 0.0974, "dropout_reads_logged": 3942, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 291.6114501953125, "n_util_obs": 870, "util_count_spearman": 0.05634906307308795}, "probe_overhead_sec_total": 447.862, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.024456, "alpha": 0.03, "t": 3697.8}
{"event": "consolidate", "step": 5550, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 8, "uncertain_staged": 262124, "t": 3752.0}
{"event": "train", "step": 5550, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.30684, "bpb": 1.88538, "chunk_loss": 0.030519, "F": 0.21121, "G": 0.21281, "chunks": 216.28, "bytes_per_chunk": 4.735, "boundary_entropy": 0.07871, "router_noise": 0.0, "grad_norm": 0.6985, "lr": 3.41e-05, "router_trainable": true, "sec_per_step": 1.097, "probe_sec": 0.0838, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.05583, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": 0.11134, "n_shadow": 6, "mean_delta_shadow": -0.0814, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": 0.20966, "trunc_abs_frac": 0.68026, "mean_delta_full": 0.15771, "read_hit_rate": {"surface": 0.60255}, "mean_gate": {"surface": 0.19417}, "n_reads": 4151, "n_cand": 6889, "read_suppressed_frac": 0.0964, "dropout_reads_logged": 4151, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 294.64385986328125, "n_util_obs": 874, "util_count_spearman": 0.05114779476635444}, "probe_overhead_sec_total": 451.341, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.027133, "alpha": 0.03, "t": 3752.7}
{"event": "consolidate", "step": 5600, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 5, "uncertain_staged": 262124, "t": 3817.9}
{"event": "probe", "step": 5600, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3306.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.189967, "cps": 25.828125, "mean_p": 0.193961, "boundary_entropy": 0.078732, "bpb": 1.861768}, "tr": {"chunks": 3751.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.207203, "cps": 29.304688, "mean_p": 0.18978, "boundary_entropy": 0.116514, "bpb": 1.998441}, "fi": {"chunks": 4881.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.259476, "cps": 38.132812, "mean_p": 0.250185, "boundary_entropy": 0.098774, "bpb": 2.097176}}, "bpb": {"en": 1.86177, "tr": 1.99844, "fi": 2.09718}, "gini_cpc": 0.070569, "gini_cps": 0.087954, "bpic": 4.75222, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0907313193619932, "fi": 1.3658977674664425}, "inventory": {"size": 327675, "admitted": 65536, "staged": 262139, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262139, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 297.6483154296875, "n_util_obs": 880, "util_count_spearman": 0.0508359006249713}, "t": 3818.5}
{"event": "eval", "step": 5600, "phase": 3, "bpb": {"en": 1.87927, "tr": 2.00507, "fi": 2.10637}, "bytes_per_chunk": {"en": 4.97, "tr": 5.326, "fi": 4.091}, "bpb_mean": 1.9969, "frozen_inventory": true, "t": 3819.8}
{"event": "train", "step": 5600, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.33251, "bpb": 1.9224, "chunk_loss": 0.030518, "F": 0.20947, "G": 0.21146, "chunks": 214.5, "bytes_per_chunk": 4.774, "boundary_entropy": 0.08499, "router_noise": 0.0, "grad_norm": 0.6947, "lr": 3.33e-05, "router_trainable": true, "sec_per_step": 1.477, "probe_sec": 0.0592, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.03418, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": 0.092, "n_shadow": 5, "mean_delta_shadow": -0.16704, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 1, "trunc_ratio_sum": 0.85461, "trunc_abs_frac": 0.85461, "mean_delta_full": 0.24097, "read_hit_rate": {"surface": 0.60524}, "mean_gate": {"surface": 0.21482}, "n_reads": 4135, "n_cand": 6832, "read_suppressed_frac": 0.1071, "dropout_reads_logged": 4135, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 297.7113952636719, "n_util_obs": 880, "util_count_spearman": 0.05075133558387967}, "probe_overhead_sec_total": 454.874, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.022592, "alpha": 0.03, "t": 3826.5}
{"event": "consolidate", "step": 5650, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 9, "uncertain_staged": 262124, "t": 3897.6}
{"event": "train", "step": 5650, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.31236, "bpb": 1.89333, "chunk_loss": 0.030452, "F": 0.20581, "G": 0.20729, "chunks": 210.75, "bytes_per_chunk": 4.859, "boundary_entropy": 0.08108, "router_noise": 0.0, "grad_norm": 0.667, "lr": 3.25e-05, "router_trainable": true, "sec_per_step": 1.448, "probe_sec": 0.0429, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02288, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": 0.30146, "n_shadow": 5, "mean_delta_shadow": -0.46289, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 2, "trunc_ratio_sum": -8.93421, "trunc_abs_frac": 62.09146, "mean_delta_full": -0.00464, "read_hit_rate": {"surface": 0.60146}, "mean_gate": {"surface": 0.20559}, "n_reads": 4037, "n_cand": 6712, "read_suppressed_frac": 0.0926, "dropout_reads_logged": 4037, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 300.71392822265625, "n_util_obs": 885, "util_count_spearman": 0.05591308619082196}, "probe_overhead_sec_total": 458.714, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.025443, "alpha": 0.03, "t": 3898.9}
{"event": "consolidate", "step": 5700, "retired": 1, "split_retired": 0, "merge_admitted": 0, "admitted": 1, "swapped": 12, "uncertain_staged": 262124, "t": 3974.2}
{"event": "train", "step": 5700, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.39464, "bpb": 2.01204, "chunk_loss": 0.030527, "F": 0.21118, "G": 0.21183, "chunks": 216.25, "bytes_per_chunk": 4.735, "boundary_entropy": 0.0832, "router_noise": 0.0, "grad_norm": 0.7041, "lr": 3.18e-05, "router_trainable": true, "sec_per_step": 1.532, "probe_sec": 0.0626, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02752, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": 0.09277, "n_shadow": 6, "mean_delta_shadow": -0.34999, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 3, "trunc_ratio_sum": 2.00037, "trunc_abs_frac": 1.0011, "mean_delta_full": 0.11027, "read_hit_rate": {"surface": 0.60105}, "mean_gate": {"surface": 0.21314}, "n_reads": 4140, "n_cand": 6888, "read_suppressed_frac": 0.1051, "dropout_reads_logged": 4140, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 303.44171142578125, "n_util_obs": 888, "util_count_spearman": 0.059349507311233285}, "probe_overhead_sec_total": 463.404, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.02615, "alpha": 0.03, "t": 3975.5}
{"event": "consolidate", "step": 5750, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 7, "uncertain_staged": 262124, "t": 4040.8}
{"event": "train", "step": 5750, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.37229, "bpb": 1.9798, "chunk_loss": 0.030566, "F": 0.20944, "G": 0.21118, "chunks": 214.47, "bytes_per_chunk": 4.775, "boundary_entropy": 0.08363, "router_noise": 0.0, "grad_norm": 0.711, "lr": 3.13e-05, "router_trainable": true, "sec_per_step": 1.329, "probe_sec": 0.0841, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.04728, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 8, "mean_delta_loo": 0.12681, "n_shadow": 6, "mean_delta_shadow": -0.15503, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 6, "trunc_ratio_sum": 0.54915, "trunc_abs_frac": 0.73561, "mean_delta_full": 0.37118, "read_hit_rate": {"surface": 0.58879}, "mean_gate": {"surface": 0.21098}, "n_reads": 4022, "n_cand": 6831, "read_suppressed_frac": 0.0925, "dropout_reads_logged": 4022, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 306.5198059082031, "n_util_obs": 892, "util_count_spearman": 0.06572418691382546}, "probe_overhead_sec_total": 467.332, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.023473, "alpha": 0.03, "t": 4042.0}
{"event": "consolidate", "step": 5800, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 6, "uncertain_staged": 262124, "t": 4101.7}
{"event": "probe", "step": 5800, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3313.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.190369, "cps": 25.882812, "mean_p": 0.194075, "boundary_entropy": 0.077545, "bpb": 1.857072}, "tr": {"chunks": 3715.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.205215, "cps": 29.023438, "mean_p": 0.189494, "boundary_entropy": 0.115611, "bpb": 1.993584}, "fi": {"chunks": 4874.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.259104, "cps": 38.078125, "mean_p": 0.250455, "boundary_entropy": 0.09965, "bpb": 2.091326}}, "bpb": {"en": 1.85707, "tr": 1.99358, "fi": 2.09133}, "gini_cpc": 0.069992, "gini_cps": 0.087436, "bpic": 4.766594, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0779806146333155, "fi": 1.3610570383330662}, "inventory": {"size": 327674, "admitted": 65536, "staged": 262138, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262138, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 309.53277587890625, "n_util_obs": 895, "util_count_spearman": 0.06185095911954848}, "t": 4102.3}
{"event": "eval", "step": 5800, "phase": 3, "bpb": {"en": 1.87408, "tr": 1.99981, "fi": 2.10022}, "bytes_per_chunk": {"en": 4.961, "tr": 5.336, "fi": 4.091}, "bpb_mean": 1.99137, "frozen_inventory": true, "t": 4103.7}
{"event": "train", "step": 5800, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.29378, "bpb": 1.86653, "chunk_loss": 0.030403, "F": 0.20416, "G": 0.20564, "chunks": 209.06, "bytes_per_chunk": 4.898, "boundary_entropy": 0.08055, "router_noise": 0.0, "grad_norm": 0.7136, "lr": 3.08e-05, "router_trainable": true, "sec_per_step": 1.344, "probe_sec": 0.0395, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02426, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 10, "mean_delta_loo": -0.17983, "n_shadow": 4, "mean_delta_shadow": -0.00677, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 1, "trunc_ratio_sum": 1.22139, "trunc_abs_frac": 1.22139, "mean_delta_full": -0.20483, "read_hit_rate": {"surface": 0.61009}, "mean_gate": {"surface": 0.20265}, "n_reads": 4062, "n_cand": 6658, "read_suppressed_frac": 0.0943, "dropout_reads_logged": 4062, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 309.59478759765625, "n_util_obs": 895, "util_count_spearman": 0.06210704806979262}, "probe_overhead_sec_total": 471.372, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.029431, "alpha": 0.03, "t": 4109.2}
{"event": "consolidate", "step": 5850, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 4, "uncertain_staged": 262124, "t": 4168.7}
{"event": "train", "step": 5850, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.3457, "bpb": 1.94143, "chunk_loss": 0.030346, "F": 0.19888, "G": 0.20105, "chunks": 203.66, "bytes_per_chunk": 5.028, "boundary_entropy": 0.07957, "router_noise": 0.0, "grad_norm": 0.7045, "lr": 3.05e-05, "router_trainable": true, "sec_per_step": 1.212, "probe_sec": 0.0979, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.05199, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 6, "mean_delta_loo": 0.12111, "n_shadow": 8, "mean_delta_shadow": -0.08969, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": -8.2227, "trunc_abs_frac": 2.61099, "mean_delta_full": -0.02068, "read_hit_rate": {"surface": 0.59476}, "mean_gate": {"surface": 0.20266}, "n_reads": 3857, "n_cand": 6485, "read_suppressed_frac": 0.1058, "dropout_reads_logged": 3857, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 312.6633605957031, "n_util_obs": 901, "util_count_spearman": 0.06270466938672649}, "probe_overhead_sec_total": 474.937, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.023084, "alpha": 0.03, "t": 4169.8}
{"event": "consolidate", "step": 5900, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 5, "uncertain_staged": 262124, "t": 4231.8}
{"event": "train", "step": 5900, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.37744, "bpb": 1.98722, "chunk_loss": 0.030553, "F": 0.21252, "G": 0.21327, "chunks": 217.62, "bytes_per_chunk": 4.705, "boundary_entropy": 0.08231, "router_noise": 0.0, "grad_norm": 0.743, "lr": 3.02e-05, "router_trainable": true, "sec_per_step": 1.253, "probe_sec": 0.0392, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.02102, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 4, "mean_delta_loo": 0.41262, "n_shadow": 10, "mean_delta_shadow": 0.03913, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 5, "trunc_ratio_sum": 0.71061, "trunc_abs_frac": 3.66731, "mean_delta_full": 0.11995, "read_hit_rate": {"surface": 0.59896}, "mean_gate": {"surface": 0.20883}, "n_reads": 4152, "n_cand": 6932, "read_suppressed_frac": 0.0869, "dropout_reads_logged": 4152, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 315.77166748046875, "n_util_obs": 911, "util_count_spearman": 0.058653342677654645}, "probe_overhead_sec_total": 478.912, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.022911, "alpha": 0.03, "t": 4232.5}
{"event": "consolidate", "step": 5950, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 4, "uncertain_staged": 262124, "t": 4291.3}
{"event": "train", "step": 5950, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.31259, "bpb": 1.89366, "chunk_loss": 0.030417, "F": 0.20316, "G": 0.20494, "chunks": 208.03, "bytes_per_chunk": 4.922, "boundary_entropy": 0.083, "router_noise": 0.0, "grad_norm": 0.7948, "lr": 3.01e-05, "router_trainable": true, "sec_per_step": 1.204, "probe_sec": 0.1116, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.07725, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": 0.03768, "n_shadow": 5, "mean_delta_shadow": -0.07761, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 1, "trunc_ratio_sum": 1.00305, "trunc_abs_frac": 1.00305, "mean_delta_full": -0.08008, "read_hit_rate": {"surface": 0.60543}, "mean_gate": {"surface": 0.2101}, "n_reads": 4011, "n_cand": 6625, "read_suppressed_frac": 0.106, "dropout_reads_logged": 4011, "inventory": {"size": 327680, "admitted": 65536, "staged": 262144, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262144, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 318.76690673828125, "n_util_obs": 917, "util_count_spearman": 0.05643596896282736}, "probe_overhead_sec_total": 482.737, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.022644, "alpha": 0.03, "t": 4292.7}
{"event": "consolidate", "step": 6000, "retired": 0, "split_retired": 0, "merge_admitted": 0, "admitted": 0, "swapped": 7, "uncertain_staged": 262124, "t": 4356.2}
{"event": "train", "step": 5999, "arch": "ali", "phase": 3, "objective": "global", "lm_loss": 1.32981, "bpb": 1.91851, "chunk_loss": 0.030631, "F": 0.21393, "G": 0.21503, "chunks": 219.06, "bytes_per_chunk": 4.674, "boundary_entropy": 0.08327, "router_noise": 0.0, "grad_norm": 0.7347, "lr": 3e-05, "router_trainable": true, "sec_per_step": 1.271, "probe_sec": 0.1251, "probe_extra_seqs": 16, "n_rows": 2, "extra_fwd_seqs": 16, "wall_s": 0.07751, "overhead_frac_est": 0.16667, "chunk_mismatch": 0, "n_loo": 9, "mean_delta_loo": 0.17296, "n_shadow": 5, "mean_delta_shadow": -0.32026, "n_split": 0, "mean_delta_split": NaN, "n_merge": 0, "mean_delta_merge": NaN, "n_dropped_unpaired": 0, "n_full": 6, "trunc_ratio_sum": 0.68214, "trunc_abs_frac": 0.95286, "mean_delta_full": -0.12712, "read_hit_rate": {"surface": 0.60404}, "mean_gate": {"surface": 0.20721}, "n_reads": 4215, "n_cand": 6978, "read_suppressed_frac": 0.0902, "dropout_reads_logged": 4215, "inventory": {"size": 327673, "admitted": 65536, "staged": 262137, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262137, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 321.78704833984375, "n_util_obs": 927, "util_count_spearman": 0.05787692414288092}, "probe_overhead_sec_total": 486.764, "probe_overhead_seq_frac": 0.4999, "util_loss": 0.024403, "alpha": 0.03, "t": 4356.2}
{"event": "probe", "step": 6000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3311.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.190255, "cps": 25.867188, "mean_p": 0.193702, "boundary_entropy": 0.076132, "bpb": 1.854292}, "tr": {"chunks": 3712.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.205049, "cps": 29.0, "mean_p": 0.189161, "boundary_entropy": 0.116028, "bpb": 1.984874}, "fi": {"chunks": 4889.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.259901, "cps": 38.195312, "mean_p": 0.250717, "boundary_entropy": 0.097817, "bpb": 2.086919}}, "bpb": {"en": 1.85429, "tr": 1.98487, "fi": 2.08692}, "gini_cpc": 0.070865, "gini_cps": 0.088314, "bpic": 4.762592, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0777607306410832, "fi": 1.3660704381361395}, "inventory": {"size": 327673, "admitted": 65536, "staged": 262137, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262137, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 321.78704833984375, "n_util_obs": 927, "util_count_spearman": 0.05787692414288092}, "t": 4356.8}
{"event": "eval", "step": 6000, "phase": 3, "bpb": {"en": 1.87152, "tr": 1.9918, "fi": 2.09469}, "bytes_per_chunk": {"en": 4.965, "tr": 5.343, "fi": 4.083}, "bpb_mean": 1.986, "frozen_inventory": true, "t": 4357.6}
{"event": "reps_dumped", "dirs": ["/local/scratch/sas245/beetle-hnet/runs/ali_ali-surface-only_s0/probe/reps/th0.30", "/local/scratch/sas245/beetle-hnet/runs/ali_ali-surface-only_s0/probe/reps/th0.40", "/local/scratch/sas245/beetle-hnet/runs/ali_ali-surface-only_s0/probe/reps/th0.50", "/local/scratch/sas245/beetle-hnet/runs/ali_ali-surface-only_s0/probe/reps/th0.60", "/local/scratch/sas245/beetle-hnet/runs/ali_ali-surface-only_s0/probe/reps/th0.70"], "t": 4364.3}
{"event": "final", "run": "ali_ali-surface-only_s0", "arch": "ali", "objective": "ratio", "condition": "A", "step": 6000, "params": 22785088, "seed": 0, "per_lang": {"en": {"cpc": 0.19025455381256104, "cps": 25.8671875, "bpb": 1.877237854154134, "bytes_per_chunk": 4.9039668509320835, "reads_per_chunk": 0.5906595205357728}, "tr": {"cpc": 0.2050488869248191, "cps": 29.0, "bpb": 1.9836637604511769, "bytes_per_chunk": 5.3378401767442805, "reads_per_chunk": 0.5821667464187903}, "fi": {"cpc": 0.259901121684121, "cps": 38.1953125, "bpb": 2.1132871381510543, "bytes_per_chunk": 4.068600518384008, "reads_per_chunk": 0.637146715090562}}, "bpb_mean": 1.9913962509187886, "gini_cpc": 0.07086496021350519, "gini_cps": 0.08831430490261938, "bpic": 4.762592343854936, "mean_boundary_entropy": 0.08327320218086243, "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": 486.7636468410492, "probe_overhead_seq_frac": 0.4999166666666667, "wall_sec": 4367.319061756134, "inventory": {"size": 327673, "admitted": 65536, "staged": 262137, "budget_B": 65536, "staging_capacity": 262144, "collisions": 0, "dropped": 0, "type_rejected": 0, "config_keys": 0, "admitted_surface": 65536, "staged_surface": 262137, "admitted_split": 0, "staged_split": 0, "admitted_merge": 0, "staged_merge": 0, "admitted_skel": 0, "staged_skel": 0, "mean_count": 321.78704833984375, "n_util_obs": 927, "util_count_spearman": 0.05787692414288092}, "ali_params": 268096, "ali_config": {"enabled": true, "budget_B": 65536, "staging_capacity": 262144, "policy": "utility", "candidate_types": ["surface"], "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": 0.05787692414288092, "paired_vs_dropout": {"n": 1879, "pearson": 0.03951643275230993, "spearman": 0.03682888787124551, "mean_abs_discrepancy": 0.7358910441398621, "mean_paired": 0.11356548964977264, "mean_dropout": 0.14920870959758759}, "predictor_calibration": {"n": 4216, "all": {"n": 4216, "pearson": 0.1731671184310671, "spearman": 0.21986157444398777, "mae": 0.15914390981197357, "reliability": [{"pred": 0.009084357880055904, "obs": 0.0202009379863739, "n": 422}, {"pred": 0.021293800324201584, "obs": 0.05406903848052025, "n": 421}, {"pred": 0.028047166764736176, "obs": 0.061708394438028336, "n": 422}, {"pred": 0.03469708189368248, "obs": 0.05650480464100838, "n": 421}, {"pred": 0.04238409921526909, "obs": 0.05355525761842728, "n": 422}, {"pred": 0.0506453812122345, "obs": 0.12060724198818207, "n": 421}, {"pred": 0.06036072224378586, "obs": 0.11969655752182007, "n": 422}, {"pred": 0.07203800976276398, "obs": 0.12473899126052856, "n": 421}, {"pred": 0.08788411319255829, "obs": 0.1447857916355133, "n": 422}, {"pred": 0.11606185138225555, "obs": 0.16831223666667938, "n": 422}]}, "admitted": {"n": 4195, "pearson": 0.16744744354721075, "spearman": 0.21406817798821218, "mae": 0.15933582186698914, "reliability": [{"pred": 0.010852980427443981, "obs": 0.027071459218859673, "n": 420}, {"pred": 0.021461596712470055, "obs": 0.056002579629421234, "n": 419}, {"pred": 0.02820117585361004, "obs": 0.06036455184221268, "n": 420}, {"pred": 0.03486092761158943, "obs": 0.058677759021520615, "n": 419}, {"pred": 0.042565956711769104, "obs": 0.052283115684986115, "n": 419}, {"pred": 0.05083423852920532, "obs": 0.1217113733291626, "n": 420}, {"pred": 0.060542166233062744, "obs": 0.11931566148996353, "n": 419}, {"pred": 0.07221594452857971, "obs": 0.12506148219108582, "n": 420}, {"pred": 0.08802869915962219, "obs": 0.14649838209152222, "n": 419}, {"pred": 0.11615332961082458, "obs": 0.16771550476551056, "n": 420}]}, "staged": {"n": 21, "pearson": 0.21218906617925304, "spearman": 0.27792207792207796, "mae": 0.12080670893192291, "reliability": [{"pred": -0.11282563209533691, "obs": -0.14851582050323486, "n": 2}, {"pred": -0.08935044705867767, "obs": -0.022793347015976906, "n": 2}, {"pred": -0.04074031487107277, "obs": -0.23829758167266846, "n": 3}, {"pred": -0.01928403042256832, "obs": -0.17395782470703125, "n": 1}, {"pred": -0.011489411816000938, "obs": -0.14177291095256805, "n": 2}, {"pred": 0.014031370170414448, "obs": -0.11138230562210083, "n": 3}, {"pred": 0.024435658007860184, "obs": -0.09640923887491226, "n": 2}, {"pred": 0.03347030654549599, "obs": -0.12943308055400848, "n": 1}, {"pred": 0.034968361258506775, "obs": -0.055959004908800125, "n": 2}, {"pred": 0.045220475643873215, "obs": -0.0703248456120491, "n": 3}]}}, "t": 4369.2}