Blame for log.jsonl

{"event": "hub", "repo": "AdaptiveChunking/ali_pkm_s2", "keep_steps": [0, 1, 2, 4, 6, 8, 16, 32, 64, 128, 256, 512, 1000, 1024, 2048, 3000, 4096, 6000], "t": 4.9}
{"event": "start", "run": "ali_pkm_s2", "arch": "pkm", "objective": "ratio", "condition": "A", "params": 39425792, "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", "pkm_params": 16908800, "pkm_config": {"n_keys": 256, "topk": 8, "heads": 1, "d_key": 256}, "t": 4.9}
{"event": "probe_schedule", "n": 45, "steps": [0, 1, 2, 4, 6, 8, 16, 32, 64, 128, 200, 256, 400, 512, 600, 800, 1000, 1024, 1200, 1400, 1600, 1800, 2000, 2048, 2200, 2400, 2600, 2800, 3000, 3200, 3400, 3600, 3800, 4000, 4096, 4200, 4400, 4600, 4800, 5000, 5200, 5400, 5600, 5800, 6000], "t": 4.9}
{"event": "probe", "step": 0, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 10933.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.628225, "cps": 85.414062, "mean_p": 0.513249, "boundary_entropy": 0.686295, "bpb": 8.12639}, "tr": {"chunks": 8187.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.452245, "cps": 63.960938, "mean_p": 0.498002, "boundary_entropy": 0.686826, "bpb": 8.096843}, "fi": {"chunks": 13540.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.719792, "cps": 105.78125, "mean_p": 0.517169, "boundary_entropy": 0.686661, "bpb": 8.064749}}, "bpb": {"en": 8.12639, "tr": 8.09684, "fi": 8.06475}, "gini_cpc": 0.099077, "gini_cps": 0.109267, "bpic": 1.737048, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.7198781817505053, "fi": 1.1457544508645294}, "t": 6.6}
{"event": "train", "step": 0, "arch": "pkm", "phase": 1, "objective": "global", "lm_loss": 5.6275, "bpb": 8.11876, "chunk_loss": 0.062086, "F": 0.59833, "G": 0.50836, "chunks": 612.69, "bytes_per_chunk": 1.671, "boundary_entropy": 0.69105, "router_noise": 0.0, "grad_norm": 11.4734, "lr": 1e-06, "router_trainable": true, "sec_per_step": 0.015, "alpha": 0.03, "t": 7.5}
{"event": "probe", "step": 1, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 10906.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.626674, "cps": 85.203125, "mean_p": 0.51314, "boundary_entropy": 0.686304, "bpb": 8.094944}, "tr": {"chunks": 8202.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.453074, "cps": 64.078125, "mean_p": 0.498079, "boundary_entropy": 0.686828, "bpb": 8.075607}, "fi": {"chunks": 13505.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.717931, "cps": 105.507812, "mean_p": 0.517036, "boundary_entropy": 0.686672, "bpb": 8.035813}}, "bpb": {"en": 8.09494, "tr": 8.07561, "fi": 8.03581}, "gini_cpc": 0.098222, "gini_cps": 0.108403, "bpic": 1.739552, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.7229825918499921, "fi": 1.1456219654381847}, "t": 8.3}
{"event": "probe", "step": 2, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 10878.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.625065, "cps": 84.984375, "mean_p": 0.512973, "boundary_entropy": 0.686318, "bpb": 8.031654}, "tr": {"chunks": 8245.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.455449, "cps": 64.414062, "mean_p": 0.498239, "boundary_entropy": 0.686832, "bpb": 8.031311}, "fi": {"chunks": 13464.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.715751, "cps": 105.1875, "mean_p": 0.516794, "boundary_entropy": 0.686692, "bpb": 7.975927}}, "bpb": {"en": 8.03165, "tr": 8.03131, "fi": 7.97593}, "gini_cpc": 0.096609, "gini_cps": 0.106771, "bpic": 1.74094, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.7286436329175892, "fi": 1.1450838387357296}, "t": 9.5}
{"event": "probe", "step": 4, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 10713.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.615584, "cps": 83.695312, "mean_p": 0.512095, "boundary_entropy": 0.68638, "bpb": 7.820468}, "tr": {"chunks": 8381.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.462962, "cps": 65.476562, "mean_p": 0.498659, "boundary_entropy": 0.686846, "bpb": 7.878295}, "fi": {"chunks": 13217.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.702621, "cps": 103.257812, "mean_p": 0.515676, "boundary_entropy": 0.686775, "bpb": 7.773302}}, "bpb": {"en": 7.82047, "tr": 7.87829, "fi": 7.7733}, "gini_cpc": 0.089701, "gini_cps": 0.09978, "bpic": 1.755811, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.7520700683275497, "fi": 1.1413898911397091}, "t": 10.7}
{"event": "probe", "step": 6, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 10320.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.593001, "cps": 80.625, "mean_p": 0.510331, "boundary_entropy": 0.686483, "bpb": 7.543742}, "tr": {"chunks": 8498.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.469425, "cps": 66.390625, "mean_p": 0.499157, "boundary_entropy": 0.686868, "bpb": 7.674231}, "fi": {"chunks": 12727.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.676572, "cps": 99.429688, "mean_p": 0.513621, "boundary_entropy": 0.686908, "bpb": 7.51818}}, "bpb": {"en": 7.54374, "tr": 7.67423, "fi": 7.51818}, "gini_cpc": 0.079412, "gini_cps": 0.089375, "bpic": 1.798447, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.7916087722836637, "fi": 1.1409289066392376}, "t": 12.3}
{"event": "probe", "step": 8, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 9627.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.55318, "cps": 75.210938, "mean_p": 0.507264, "boundary_entropy": 0.686614, "bpb": 7.293747}, "tr": {"chunks": 8626.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.476496, "cps": 67.390625, "mean_p": 0.499549, "boundary_entropy": 0.686902, "bpb": 7.478141}, "fi": {"chunks": 11865.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.630748, "cps": 92.695312, "mean_p": 0.51026, "boundary_entropy": 0.687082, "bpb": 7.295762}}, "bpb": {"en": 7.29375, "tr": 7.47814, "fi": 7.29576}, "gini_cpc": 0.061933, "gini_cps": 0.071696, "bpic": 1.883658, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.861374579212228, "fi": 1.140220926872929}, "t": 13.5}
{"event": "probe", "step": 16, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3759.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.215997, "cps": 29.367188, "mean_p": 0.48443, "boundary_entropy": 0.686164, "bpb": 6.724658}, "tr": {"chunks": 7259.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.400983, "cps": 56.710938, "mean_p": 0.494329, "boundary_entropy": 0.686995, "bpb": 7.013298}, "fi": {"chunks": 4584.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.243687, "cps": 35.8125, "mean_p": 0.486538, "boundary_entropy": 0.686904, "bpb": 6.787259}}, "bpb": {"en": 6.72466, "tr": 7.0133, "fi": 6.78726}, "gini_cpc": 0.143289, "gini_cps": 0.149553, "bpic": 3.6362, "worst_lang": "tr", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.8564276978691454, "fi": 1.1281959075158643}, "t": 15.2}
{"event": "probe", "step": 32, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2607.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.149802, "cps": 20.367188, "mean_p": 0.486719, "boundary_entropy": 0.686944, "bpb": 6.304694}, "tr": {"chunks": 4642.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.256422, "cps": 36.265625, "mean_p": 0.490181, "boundary_entropy": 0.687559, "bpb": 6.65151}, "fi": {"chunks": 1399.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.074371, "cps": 10.929688, "mean_p": 0.479425, "boundary_entropy": 0.686975, "bpb": 6.420733}}, "bpb": {"en": 6.30469, "tr": 6.65151, "fi": 6.42073}, "gini_cpc": 0.252535, "gini_cps": 0.25, "bpic": 6.56013, "worst_lang": "tr", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.711739504676825, "fi": 0.49646532379904795}, "t": 17.9}
{"event": "train", "step": 50, "arch": "pkm", "phase": 1, "objective": "global", "lm_loss": 4.21231, "bpb": 6.07708, "chunk_loss": 0.034175, "F": 0.22534, "G": 0.48845, "chunks": 230.75, "bytes_per_chunk": 4.438, "boundary_entropy": 0.6916, "router_noise": 0.0, "grad_norm": 2.4787, "lr": 5.1e-05, "router_trainable": true, "sec_per_step": 0.257, "alpha": 0.03, "t": 20.4}
{"event": "probe", "step": 64, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 4806.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.276159, "cps": 37.546875, "mean_p": 0.485082, "boundary_entropy": 0.685492, "bpb": 5.55211}, "tr": {"chunks": 5403.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.298459, "cps": 42.210938, "mean_p": 0.486562, "boundary_entropy": 0.686476, "bpb": 5.956441}, "fi": {"chunks": 4694.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.249535, "cps": 36.671875, "mean_p": 0.483981, "boundary_entropy": 0.686326, "bpb": 5.717801}}, "bpb": {"en": 5.55211, "tr": 5.95644, "fi": 5.7178}, "gini_cpc": 0.039575, "gini_cps": 0.031716, "bpic": 3.80675, "worst_lang": "tr", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0807488195409498, "fi": 0.9035902904581224}, "t": 22.3}
{"event": "train", "step": 100, "arch": "pkm", "phase": 1, "objective": "global", "lm_loss": 3.24758, "bpb": 4.68527, "chunk_loss": 0.029607, "F": 0.16013, "G": 0.45223, "chunks": 163.97, "bytes_per_chunk": 6.245, "boundary_entropy": 0.68183, "router_noise": 0.0, "grad_norm": 1.472, "lr": 0.000101, "router_trainable": true, "sec_per_step": 0.113, "alpha": 0.03, "t": 26.0}
{"event": "probe", "step": 128, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3549.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.20393, "cps": 27.726562, "mean_p": 0.410725, "boundary_entropy": 0.642216, "bpb": 4.145735}, "tr": {"chunks": 4430.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.244711, "cps": 34.609375, "mean_p": 0.432348, "boundary_entropy": 0.658552, "bpb": 4.39814}, "fi": {"chunks": 3517.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.186965, "cps": 27.476562, "mean_p": 0.422119, "boundary_entropy": 0.658502, "bpb": 4.337637}}, "bpb": {"en": 4.14574, "tr": 4.39814, "fi": 4.33764}, "gini_cpc": 0.060568, "gini_cps": 0.052946, "bpic": 4.934934, "worst_lang": "tr", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.199972506343316, "fi": 0.9168084464167361}, "t": 28.8}
{"event": "train", "step": 150, "arch": "pkm", "phase": 1, "objective": "global", "lm_loss": 2.77725, "bpb": 4.00672, "chunk_loss": 0.031121, "F": 0.18988, "G": 0.38165, "chunks": 194.44, "bytes_per_chunk": 5.266, "boundary_entropy": 0.61712, "router_noise": 0.0, "grad_norm": 1.6567, "lr": 0.000151, "router_trainable": true, "sec_per_step": 0.094, "alpha": 0.03, "t": 30.8}
{"event": "probe", "step": 200, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3609.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.207378, "cps": 28.195312, "mean_p": 0.366169, "boundary_entropy": 0.589965, "bpb": 3.62351}, "tr": {"chunks": 4018.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.221952, "cps": 31.390625, "mean_p": 0.380372, "boundary_entropy": 0.594158, "bpb": 3.787198}, "fi": {"chunks": 4175.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.221945, "cps": 32.617188, "mean_p": 0.411061, "boundary_entropy": 0.639621, "bpb": 3.85405}}, "bpb": {"en": 3.62351, "tr": 3.7872, "fi": 3.85405}, "gini_cpc": 0.014919, "gini_cps": 0.031972, "bpic": 4.806982, "worst_lang": "tr", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0702780510286822, "fi": 1.0702416163797266}, "t": 34.9}
{"event": "eval", "step": 200, "phase": 1, "bpb": {"en": 3.63068, "tr": 3.79849, "fi": 3.85275}, "bytes_per_chunk": {"en": 5.264, "tr": 6.038, "fi": 4.959}, "bpb_mean": 3.76064, "frozen_inventory": false, "t": 35.4}
{"event": "train", "step": 200, "arch": "pkm", "phase": 1, "objective": "global", "lm_loss": 2.52415, "bpb": 3.64157, "chunk_loss": 0.030329, "F": 0.17361, "G": 0.35823, "chunks": 177.78, "bytes_per_chunk": 5.76, "boundary_entropy": 0.59058, "router_noise": 0.0, "grad_norm": 1.5768, "lr": 0.000201, "router_trainable": true, "sec_per_step": 0.103, "alpha": 0.03, "t": 35.9}
{"event": "train", "step": 250, "arch": "pkm", "phase": 1, "objective": "global", "lm_loss": 2.38886, "bpb": 3.4464, "chunk_loss": 0.030591, "F": 0.18307, "G": 0.33808, "chunks": 187.47, "bytes_per_chunk": 5.462, "boundary_entropy": 0.55779, "router_noise": 0.0, "grad_norm": 1.8726, "lr": 0.000251, "router_trainable": true, "sec_per_step": 0.08, "alpha": 0.03, "t": 39.9}
{"event": "probe", "step": 256, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3558.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.204448, "cps": 27.796875, "mean_p": 0.338944, "boundary_entropy": 0.549416, "bpb": 3.367857}, "tr": {"chunks": 4244.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.234436, "cps": 33.15625, "mean_p": 0.340686, "boundary_entropy": 0.555254, "bpb": 3.541509}, "fi": {"chunks": 4387.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.233215, "cps": 34.273438, "mean_p": 0.397082, "boundary_entropy": 0.617064, "bpb": 3.695061}}, "bpb": {"en": 3.36786, "tr": 3.54151, "fi": 3.69506}, "gini_cpc": 0.029746, "gini_cps": 0.045341, "bpic": 4.65436, "worst_lang": "tr", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.146682013239027, "fi": 1.1407065293054914}, "t": 40.8}
{"event": "train", "step": 300, "arch": "pkm", "phase": 1, "objective": "global", "lm_loss": 2.28218, "bpb": 3.29249, "chunk_loss": 0.03066, "F": 0.18591, "G": 0.32162, "chunks": 190.38, "bytes_per_chunk": 5.379, "boundary_entropy": 0.52309, "router_noise": 0.0, "grad_norm": 1.1247, "lr": 0.0003, "router_trainable": true, "sec_per_step": 0.105, "alpha": 0.03, "t": 45.2}
{"event": "train", "step": 350, "arch": "pkm", "phase": 1, "objective": "global", "lm_loss": 2.16997, "bpb": 3.1306, "chunk_loss": 0.030316, "F": 0.17563, "G": 0.30893, "chunks": 179.84, "bytes_per_chunk": 5.694, "boundary_entropy": 0.50219, "router_noise": 0.0, "grad_norm": 1.0801, "lr": 0.0002999, "router_trainable": true, "sec_per_step": 0.082, "alpha": 0.03, "t": 49.3}
{"event": "probe", "step": 400, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3209.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.184393, "cps": 25.070312, "mean_p": 0.309764, "boundary_entropy": 0.478979, "bpb": 2.91803}, "tr": {"chunks": 3463.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.191294, "cps": 27.054688, "mean_p": 0.308997, "boundary_entropy": 0.506785, "bpb": 3.200414}, "fi": {"chunks": 3900.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.207326, "cps": 30.46875, "mean_p": 0.344658, "boundary_entropy": 0.555709, "bpb": 3.368136}}, "bpb": {"en": 2.91803, "tr": 3.20041, "fi": 3.36814}, "gini_cpc": 0.026222, "gini_cps": 0.043574, "bpic": 5.36625, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0374241251374725, "fi": 1.124364504375328}, "t": 53.6}
{"event": "eval", "step": 400, "phase": 1, "bpb": {"en": 2.92834, "tr": 3.20635, "fi": 3.36051}, "bytes_per_chunk": {"en": 5.624, "tr": 5.932, "fi": 5.535}, "bpb_mean": 3.16507, "frozen_inventory": false, "t": 54.0}
{"event": "train", "step": 400, "arch": "pkm", "phase": 1, "objective": "global", "lm_loss": 2.09807, "bpb": 3.02687, "chunk_loss": 0.030081, "F": 0.16818, "G": 0.30127, "chunks": 172.22, "bytes_per_chunk": 5.946, "boundary_entropy": 0.48276, "router_noise": 0.0, "grad_norm": 1.0573, "lr": 0.0002998, "router_trainable": true, "sec_per_step": 0.108, "alpha": 0.03, "t": 54.7}
{"event": "train", "step": 450, "arch": "pkm", "phase": 1, "objective": "global", "lm_loss": 2.01507, "bpb": 2.90714, "chunk_loss": 0.030592, "F": 0.18661, "G": 0.29055, "chunks": 191.09, "bytes_per_chunk": 5.359, "boundary_entropy": 0.44154, "router_noise": 0.0, "grad_norm": 1.6902, "lr": 0.0002995, "router_trainable": true, "sec_per_step": 0.073, "alpha": 0.03, "t": 58.3}
{"event": "train", "step": 500, "arch": "pkm", "phase": 1, "objective": "global", "lm_loss": 1.91822, "bpb": 2.7674, "chunk_loss": 0.030722, "F": 0.19232, "G": 0.28067, "chunks": 196.94, "bytes_per_chunk": 5.2, "boundary_entropy": 0.40285, "router_noise": 0.0, "grad_norm": 1.3877, "lr": 0.0002992, "router_trainable": true, "sec_per_step": 0.077, "alpha": 0.03, "t": 62.2}
{"event": "probe", "step": 512, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3366.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.193415, "cps": 26.296875, "mean_p": 0.285133, "boundary_entropy": 0.407846, "bpb": 2.71773}, "tr": {"chunks": 3667.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.202563, "cps": 28.648438, "mean_p": 0.280349, "boundary_entropy": 0.426165, "bpb": 3.049012}, "fi": {"chunks": 3929.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.208867, "cps": 30.695312, "mean_p": 0.285987, "boundary_entropy": 0.438161, "bpb": 3.236727}}, "bpb": {"en": 2.71773, "tr": 3.04901, "fi": 3.23673}, "gini_cpc": 0.017032, "gini_cps": 0.034239, "bpic": 5.175333, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0472982240758788, "fi": 1.079891577452454}, "t": 63.5}
{"event": "train", "step": 550, "arch": "pkm", "phase": 1, "objective": "global", "lm_loss": 1.92262, "bpb": 2.77376, "chunk_loss": 0.030497, "F": 0.18668, "G": 0.26982, "chunks": 191.16, "bytes_per_chunk": 5.357, "boundary_entropy": 0.38648, "router_noise": 0.0, "grad_norm": 1.5516, "lr": 0.0002987, "router_trainable": true, "sec_per_step": 0.096, "alpha": 0.03, "t": 67.0}
{"event": "probe", "step": 600, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3204.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.184106, "cps": 25.03125, "mean_p": 0.267144, "boundary_entropy": 0.377, "bpb": 2.617933}, "tr": {"chunks": 3909.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.215931, "cps": 30.539062, "mean_p": 0.264659, "boundary_entropy": 0.362272, "bpb": 2.932031}, "fi": {"chunks": 4261.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.226516, "cps": 33.289062, "mean_p": 0.287906, "boundary_entropy": 0.393001, "bpb": 3.157741}}, "bpb": {"en": 2.61793, "tr": 2.93203, "fi": 3.15774}, "gini_cpc": 0.045125, "gini_cps": 0.061954, "bpic": 4.987867, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.172861503494051, "fi": 1.2303573373376842}, "t": 70.8}
{"event": "eval", "step": 600, "phase": 1, "bpb": {"en": 2.62769, "tr": 2.93819, "fi": 3.15687}, "bytes_per_chunk": {"en": 5.471, "tr": 5.207, "fi": 4.814}, "bpb_mean": 2.90758, "frozen_inventory": false, "t": 71.2}
{"event": "train", "step": 600, "arch": "pkm", "phase": 1, "objective": "global", "lm_loss": 1.9058, "bpb": 2.74949, "chunk_loss": 0.030454, "F": 0.18591, "G": 0.25955, "chunks": 190.38, "bytes_per_chunk": 5.379, "boundary_entropy": 0.3673, "router_noise": 0.0, "grad_norm": 1.3341, "lr": 0.0002982, "router_trainable": true, "sec_per_step": 0.096, "alpha": 0.03, "t": 71.8}
{"event": "train", "step": 650, "arch": "pkm", "phase": 1, "objective": "global", "lm_loss": 1.85597, "bpb": 2.6776, "chunk_loss": 0.030566, "F": 0.19135, "G": 0.26063, "chunks": 195.94, "bytes_per_chunk": 5.226, "boundary_entropy": 0.35321, "router_noise": 0.0, "grad_norm": 1.2977, "lr": 0.0002975, "router_trainable": true, "sec_per_step": 0.081, "alpha": 0.03, "t": 75.8}
{"event": "train", "step": 700, "arch": "pkm", "phase": 1, "objective": "global", "lm_loss": 1.99529, "bpb": 2.8786, "chunk_loss": 0.030734, "F": 0.19894, "G": 0.25969, "chunks": 203.72, "bytes_per_chunk": 5.027, "boundary_entropy": 0.33775, "router_noise": 0.0, "grad_norm": 1.1735, "lr": 0.0002967, "router_trainable": true, "sec_per_step": 0.091, "alpha": 0.03, "t": 80.4}
{"event": "train", "step": 750, "arch": "pkm", "phase": 1, "objective": "global", "lm_loss": 1.7898, "bpb": 2.58214, "chunk_loss": 0.030572, "F": 0.19348, "G": 0.25397, "chunks": 198.12, "bytes_per_chunk": 5.168, "boundary_entropy": 0.33158, "router_noise": 0.0, "grad_norm": 1.1636, "lr": 0.0002959, "router_trainable": true, "sec_per_step": 0.08, "alpha": 0.03, "t": 84.4}
{"event": "probe", "step": 800, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3302.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.189737, "cps": 25.796875, "mean_p": 0.255658, "boundary_entropy": 0.333791, "bpb": 2.451921}, "tr": {"chunks": 3881.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.214384, "cps": 30.320312, "mean_p": 0.249168, "boundary_entropy": 0.325918, "bpb": 2.780022}, "fi": {"chunks": 4167.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.221519, "cps": 32.554688, "mean_p": 0.274811, "boundary_entropy": 0.345321, "bpb": 3.00859}}, "bpb": {"en": 2.45192, "tr": 2.78002, "fi": 3.00859}, "gini_cpc": 0.033866, "gini_cps": 0.050808, "bpic": 4.998414, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.129900348477032, "fi": 1.1675047825832026}, "t": 89.2}
{"event": "eval", "step": 800, "phase": 1, "bpb": {"en": 2.46339, "tr": 2.78664, "fi": 3.01107}, "bytes_per_chunk": {"en": 5.224, "tr": 5.216, "fi": 4.958}, "bpb_mean": 2.7537, "frozen_inventory": false, "t": 89.7}
{"event": "train", "step": 800, "arch": "pkm", "phase": 1, "objective": "global", "lm_loss": 1.87268, "bpb": 2.7017, "chunk_loss": 0.030813, "F": 0.20026, "G": 0.25707, "chunks": 205.06, "bytes_per_chunk": 4.994, "boundary_entropy": 0.32097, "router_noise": 0.0, "grad_norm": 1.3975, "lr": 0.0002949, "router_trainable": true, "sec_per_step": 0.121, "alpha": 0.03, "t": 90.4}
{"event": "train", "step": 850, "arch": "pkm", "phase": 1, "objective": "global", "lm_loss": 1.81687, "bpb": 2.62119, "chunk_loss": 0.030905, "F": 0.20367, "G": 0.2491, "chunks": 208.56, "bytes_per_chunk": 4.91, "boundary_entropy": 0.31382, "router_noise": 0.0, "grad_norm": 1.1162, "lr": 0.0002938, "router_trainable": true, "sec_per_step": 0.075, "alpha": 0.03, "t": 94.2}
{"event": "train", "step": 900, "arch": "pkm", "phase": 1, "objective": "global", "lm_loss": 1.62716, "bpb": 2.3475, "chunk_loss": 0.030578, "F": 0.19458, "G": 0.24926, "chunks": 199.25, "bytes_per_chunk": 5.139, "boundary_entropy": 0.31157, "router_noise": 0.0, "grad_norm": 1.5149, "lr": 0.0002927, "router_trainable": true, "sec_per_step": 0.086, "alpha": 0.03, "t": 98.4}
{"event": "train", "step": 950, "arch": "pkm", "phase": 1, "objective": "global", "lm_loss": 1.69353, "bpb": 2.44324, "chunk_loss": 0.030455, "F": 0.19073, "G": 0.24036, "chunks": 195.31, "bytes_per_chunk": 5.243, "boundary_entropy": 0.30215, "router_noise": 0.0, "grad_norm": 1.0329, "lr": 0.0002914, "router_trainable": true, "sec_per_step": 0.096, "alpha": 0.03, "t": 103.2}
{"event": "probe", "step": 1000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3326.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.191116, "cps": 25.984375, "mean_p": 0.243202, "boundary_entropy": 0.305331, "bpb": 2.348923}, "tr": {"chunks": 3894.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.215102, "cps": 30.421875, "mean_p": 0.234655, "boundary_entropy": 0.263045, "bpb": 2.653239}, "fi": {"chunks": 4072.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.216469, "cps": 31.8125, "mean_p": 0.264104, "boundary_entropy": 0.305801, "bpb": 2.873837}}, "bpb": {"en": 2.34892, "tr": 2.65324, "fi": 2.87384}, "gini_cpc": 0.027143, "gini_cps": 0.044043, "bpic": 5.024088, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.125504591568611, "fi": 1.1326552992096939}, "t": 107.6}
{"event": "eval", "step": 1000, "phase": 1, "bpb": {"en": 2.35913, "tr": 2.66075, "fi": 2.87857}, "bytes_per_chunk": {"en": 5.297, "tr": 5.164, "fi": 5.122}, "bpb_mean": 2.63282, "frozen_inventory": false, "t": 108.2}
{"event": "train", "step": 1000, "arch": "pkm", "phase": 1, "objective": "global", "lm_loss": 1.74922, "bpb": 2.5236, "chunk_loss": 0.030499, "F": 0.19336, "G": 0.24244, "chunks": 198.0, "bytes_per_chunk": 5.172, "boundary_entropy": 0.28854, "router_noise": 0.0, "grad_norm": 1.1288, "lr": 0.0002901, "router_trainable": true, "sec_per_step": 0.111, "alpha": 0.03, "t": 108.8}
{"event": "probe", "step": 1024, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3223.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.185198, "cps": 25.179688, "mean_p": 0.234593, "boundary_entropy": 0.29756, "bpb": 2.339885}, "tr": {"chunks": 3900.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.215434, "cps": 30.46875, "mean_p": 0.233467, "boundary_entropy": 0.274716, "bpb": 2.646627}, "fi": {"chunks": 4071.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.216416, "cps": 31.804688, "mean_p": 0.259353, "boundary_entropy": 0.302842, "bpb": 2.851639}}, "bpb": {"en": 2.33988, "tr": 2.64663, "fi": 2.85164}, "gini_cpc": 0.033728, "gini_cps": 0.050503, "bpic": 5.068072, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.163262881108376, "fi": 1.1685654281704483}, "t": 110.8}
{"event": "train", "step": 1050, "arch": "pkm", "phase": 1, "objective": "global", "lm_loss": 1.65297, "bpb": 2.38473, "chunk_loss": 0.030479, "F": 0.185, "G": 0.23258, "chunks": 189.44, "bytes_per_chunk": 5.405, "boundary_entropy": 0.28431, "router_noise": 0.0, "grad_norm": 1.0998, "lr": 0.0002886, "router_trainable": true, "sec_per_step": 0.103, "alpha": 0.03, "t": 114.0}
{"event": "train", "step": 1100, "arch": "pkm", "phase": 1, "objective": "global", "lm_loss": 1.75522, "bpb": 2.53225, "chunk_loss": 0.030487, "F": 0.19116, "G": 0.23612, "chunks": 195.75, "bytes_per_chunk": 5.231, "boundary_entropy": 0.28601, "router_noise": 0.0, "grad_norm": 1.1423, "lr": 0.0002871, "router_trainable": true, "sec_per_step": 0.08, "alpha": 0.03, "t": 118.0}
{"event": "train", "step": 1150, "arch": "pkm", "phase": 1, "objective": "global", "lm_loss": 1.65549, "bpb": 2.38836, "chunk_loss": 0.030629, "F": 0.19891, "G": 0.24124, "chunks": 203.69, "bytes_per_chunk": 5.027, "boundary_entropy": 0.27919, "router_noise": 0.0, "grad_norm": 1.0478, "lr": 0.0002855, "router_trainable": true, "sec_per_step": 0.088, "alpha": 0.03, "t": 122.4}
{"event": "probe", "step": 1200, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3234.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.18583, "cps": 25.265625, "mean_p": 0.233252, "boundary_entropy": 0.285785, "bpb": 2.290991}, "tr": {"chunks": 3747.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.206982, "cps": 29.273438, "mean_p": 0.228367, "boundary_entropy": 0.260303, "bpb": 2.573512}, "fi": {"chunks": 4342.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.230822, "cps": 33.921875, "mean_p": 0.260509, "boundary_entropy": 0.273406, "bpb": 2.757563}}, "bpb": {"en": 2.29099, "tr": 2.57351, "fi": 2.75756}, "gini_cpc": 0.048097, "gini_cps": 0.065236, "bpic": 5.010333, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.113825730459911, "fi": 1.242115668910022}, "t": 127.0}
{"event": "eval", "step": 1200, "phase": 1, "bpb": {"en": 2.30156, "tr": 2.58114, "fi": 2.76581}, "bytes_per_chunk": {"en": 5.435, "tr": 5.372, "fi": 4.702}, "bpb_mean": 2.5495, "frozen_inventory": false, "t": 127.4}
{"event": "train", "step": 1200, "arch": "pkm", "phase": 1, "objective": "global", "lm_loss": 1.62882, "bpb": 2.34989, "chunk_loss": 0.0304, "F": 0.18909, "G": 0.23218, "chunks": 193.62, "bytes_per_chunk": 5.289, "boundary_entropy": 0.27264, "router_noise": 0.0, "grad_norm": 1.0771, "lr": 0.0002837, "router_trainable": true, "sec_per_step": 0.114, "alpha": 0.03, "t": 128.1}
{"event": "train", "step": 1250, "arch": "pkm", "phase": 1, "objective": "global", "lm_loss": 1.64918, "bpb": 2.37927, "chunk_loss": 0.030354, "F": 0.1907, "G": 0.22656, "chunks": 195.28, "bytes_per_chunk": 5.244, "boundary_entropy": 0.26304, "router_noise": 0.0, "grad_norm": 0.9647, "lr": 0.0002819, "router_trainable": true, "sec_per_step": 0.09, "alpha": 0.03, "t": 132.6}
{"event": "train", "step": 1300, "arch": "pkm", "phase": 1, "objective": "global", "lm_loss": 1.69002, "bpb": 2.43818, "chunk_loss": 0.030647, "F": 0.2002, "G": 0.2365, "chunks": 205.0, "bytes_per_chunk": 4.995, "boundary_entropy": 0.25705, "router_noise": 0.0, "grad_norm": 1.0334, "lr": 0.00028, "router_trainable": true, "sec_per_step": 0.089, "alpha": 0.03, "t": 137.0}
{"event": "train", "step": 1350, "arch": "pkm", "phase": 1, "objective": "global", "lm_loss": 1.61368, "bpb": 2.32804, "chunk_loss": 0.03046, "F": 0.19244, "G": 0.23145, "chunks": 197.06, "bytes_per_chunk": 5.196, "boundary_entropy": 0.26166, "router_noise": 0.0, "grad_norm": 1.0503, "lr": 0.000278, "router_trainable": true, "sec_per_step": 0.07, "alpha": 0.03, "t": 140.5}
{"event": "probe", "step": 1400, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3259.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.187267, "cps": 25.460938, "mean_p": 0.228551, "boundary_entropy": 0.265754, "bpb": 2.230295}, "tr": {"chunks": 3773.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.208418, "cps": 29.476562, "mean_p": 0.22091, "boundary_entropy": 0.235626, "bpb": 2.494463}, "fi": {"chunks": 4225.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.224603, "cps": 33.007812, "mean_p": 0.251249, "boundary_entropy": 0.25353, "bpb": 2.662063}}, "bpb": {"en": 2.2303, "tr": 2.49446, "fi": 2.66206}, "gini_cpc": 0.040128, "gini_cps": 0.057209, "bpic": 5.039709, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.1129509217795135, "fi": 1.1993738884378022}, "t": 144.4}
{"event": "eval", "step": 1400, "phase": 1, "bpb": {"en": 2.23855, "tr": 2.50057, "fi": 2.66821}, "bytes_per_chunk": {"en": 5.411, "tr": 5.301, "fi": 4.816}, "bpb_mean": 2.46911, "frozen_inventory": false, "t": 145.0}
{"event": "train", "step": 1400, "arch": "pkm", "phase": 1, "objective": "global", "lm_loss": 1.60925, "bpb": 2.32166, "chunk_loss": 0.030365, "F": 0.18839, "G": 0.22441, "chunks": 192.91, "bytes_per_chunk": 5.308, "boundary_entropy": 0.2517, "router_noise": 0.0, "grad_norm": 0.9163, "lr": 0.0002759, "router_trainable": true, "sec_per_step": 0.11, "alpha": 0.03, "t": 146.0}
{"event": "train", "step": 1450, "arch": "pkm", "phase": 1, "objective": "global", "lm_loss": 1.66429, "bpb": 2.40106, "chunk_loss": 0.030339, "F": 0.18884, "G": 0.22414, "chunks": 193.38, "bytes_per_chunk": 5.295, "boundary_entropy": 0.2459, "router_noise": 0.0, "grad_norm": 0.9836, "lr": 0.0002738, "router_trainable": true, "sec_per_step": 0.076, "alpha": 0.03, "t": 149.8}
{"event": "train", "step": 1500, "arch": "pkm", "phase": 1, "objective": "global", "lm_loss": 1.6785, "bpb": 2.42156, "chunk_loss": 0.030549, "F": 0.19791, "G": 0.23274, "chunks": 202.66, "bytes_per_chunk": 5.053, "boundary_entropy": 0.24714, "router_noise": 0.0, "grad_norm": 0.9137, "lr": 0.0002715, "router_trainable": true, "sec_per_step": 0.08, "alpha": 0.03, "t": 153.8}
{"event": "train", "step": 1550, "arch": "pkm", "phase": 1, "objective": "global", "lm_loss": 1.59628, "bpb": 2.30295, "chunk_loss": 0.030371, "F": 0.18985, "G": 0.22186, "chunks": 194.41, "bytes_per_chunk": 5.267, "boundary_entropy": 0.23916, "router_noise": 0.0, "grad_norm": 0.9236, "lr": 0.0002692, "router_trainable": true, "sec_per_step": 0.107, "alpha": 0.03, "t": 159.2}
{"event": "probe", "step": 1600, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3224.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.185255, "cps": 25.1875, "mean_p": 0.221983, "boundary_entropy": 0.248213, "bpb": 2.188718}, "tr": {"chunks": 3725.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.205767, "cps": 29.101562, "mean_p": 0.215275, "boundary_entropy": 0.224777, "bpb": 2.434223}, "fi": {"chunks": 4310.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.229121, "cps": 33.671875, "mean_p": 0.249957, "boundary_entropy": 0.235372, "bpb": 2.611931}}, "bpb": {"en": 2.18872, "tr": 2.43422, "fi": 2.61193}, "gini_cpc": 0.047157, "gini_cps": 0.064304, "bpic": 5.038813, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.1107205645281226, "fi": 1.236785752964087}, "t": 165.0}
{"event": "eval", "step": 1600, "phase": 1, "bpb": {"en": 2.1999, "tr": 2.4411, "fi": 2.61792}, "bytes_per_chunk": {"en": 5.465, "tr": 5.393, "fi": 4.785}, "bpb_mean": 2.41964, "frozen_inventory": false, "t": 165.6}
{"event": "train", "step": 1600, "arch": "pkm", "phase": 1, "objective": "global", "lm_loss": 1.52933, "bpb": 2.20635, "chunk_loss": 0.030268, "F": 0.186, "G": 0.22152, "chunks": 190.47, "bytes_per_chunk": 5.376, "boundary_entropy": 0.2418, "router_noise": 0.0, "grad_norm": 0.9675, "lr": 0.0002668, "router_trainable": true, "sec_per_step": 0.144, "alpha": 0.03, "t": 166.4}
{"event": "train", "step": 1650, "arch": "pkm", "phase": 1, "objective": "global", "lm_loss": 1.59659, "bpb": 2.30339, "chunk_loss": 0.030317, "F": 0.18918, "G": 0.22284, "chunks": 193.72, "bytes_per_chunk": 5.286, "boundary_entropy": 0.23783, "router_noise": 0.0, "grad_norm": 0.9722, "lr": 0.0002643, "router_trainable": true, "sec_per_step": 0.105, "alpha": 0.03, "t": 171.6}
{"event": "train", "step": 1700, "arch": "pkm", "phase": 1, "objective": "global", "lm_loss": 1.54947, "bpb": 2.23542, "chunk_loss": 0.030346, "F": 0.19058, "G": 0.22344, "chunks": 195.16, "bytes_per_chunk": 5.247, "boundary_entropy": 0.23508, "router_noise": 0.0, "grad_norm": 0.8452, "lr": 0.0002618, "router_trainable": true, "sec_per_step": 0.1, "alpha": 0.03, "t": 176.6}
{"event": "train", "step": 1750, "arch": "pkm", "phase": 1, "objective": "global", "lm_loss": 1.57544, "bpb": 2.27288, "chunk_loss": 0.030482, "F": 0.19754, "G": 0.22475, "chunks": 202.28, "bytes_per_chunk": 5.062, "boundary_entropy": 0.221, "router_noise": 0.0, "grad_norm": 0.8135, "lr": 0.0002591, "router_trainable": true, "sec_per_step": 0.079, "alpha": 0.03, "t": 180.6}
{"event": "probe", "step": 1800, "phase": 1, "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": 0.224865, "boundary_entropy": 0.238653, "bpb": 2.152627}, "tr": {"chunks": 3764.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.207921, "cps": 29.40625, "mean_p": 0.213953, "boundary_entropy": 0.213031, "bpb": 2.377191}, "fi": {"chunks": 4252.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.226038, "cps": 33.21875, "mean_p": 0.24898, "boundary_entropy": 0.229651, "bpb": 2.557107}}, "bpb": {"en": 2.15263, "tr": 2.37719, "fi": 2.55711}, "gini_cpc": 0.038087, "gini_cps": 0.055256, "bpic": 5.007679, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.092198932296043, "fi": 1.1873644905605658}, "t": 186.1}
{"event": "eval", "step": 1800, "phase": 1, "bpb": {"en": 2.1655, "tr": 2.38597, "fi": 2.56401}, "bytes_per_chunk": {"en": 5.338, "tr": 5.335, "fi": 4.798}, "bpb_mean": 2.37183, "frozen_inventory": false, "t": 186.7}
{"event": "train", "step": 1800, "arch": "pkm", "phase": 1, "objective": "global", "lm_loss": 1.55457, "bpb": 2.24277, "chunk_loss": 0.030271, "F": 0.18478, "G": 0.21626, "chunks": 189.22, "bytes_per_chunk": 5.412, "boundary_entropy": 0.22774, "router_noise": 0.0, "grad_norm": 0.8671, "lr": 0.0002564, "router_trainable": true, "sec_per_step": 0.139, "alpha": 0.03, "t": 187.5}
{"event": "train", "step": 1850, "arch": "pkm", "phase": 1, "objective": "global", "lm_loss": 1.56448, "bpb": 2.25706, "chunk_loss": 0.030397, "F": 0.19064, "G": 0.22157, "chunks": 195.22, "bytes_per_chunk": 5.245, "boundary_entropy": 0.22399, "router_noise": 0.0, "grad_norm": 0.8225, "lr": 0.0002537, "router_trainable": true, "sec_per_step": 0.089, "alpha": 0.03, "t": 192.0}
{"event": "train", "step": 1900, "arch": "pkm", "phase": 1, "objective": "global", "lm_loss": 1.54781, "bpb": 2.23302, "chunk_loss": 0.030481, "F": 0.19354, "G": 0.21986, "chunks": 198.19, "bytes_per_chunk": 5.167, "boundary_entropy": 0.22036, "router_noise": 0.0, "grad_norm": 0.8938, "lr": 0.0002508, "router_trainable": true, "sec_per_step": 0.085, "alpha": 0.03, "t": 196.2}
{"event": "train", "step": 1950, "arch": "pkm", "phase": 1, "objective": "global", "lm_loss": 1.49698, "bpb": 2.15968, "chunk_loss": 0.03032, "F": 0.18265, "G": 0.21596, "chunks": 187.03, "bytes_per_chunk": 5.475, "boundary_entropy": 0.22909, "router_noise": 0.0, "grad_norm": 0.8868, "lr": 0.0002479, "router_trainable": true, "sec_per_step": 0.082, "alpha": 0.03, "t": 200.3}
{"event": "probe", "step": 2000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3330.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.191346, "cps": 26.015625, "mean_p": 0.219585, "boundary_entropy": 0.223189, "bpb": 2.123953}, "tr": {"chunks": 3635.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.200795, "cps": 28.398438, "mean_p": 0.210759, "boundary_entropy": 0.204345, "bpb": 2.325898}, "fi": {"chunks": 4306.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.228909, "cps": 33.640625, "mean_p": 0.247543, "boundary_entropy": 0.21411, "bpb": 2.511126}}, "bpb": {"en": 2.12395, "tr": 2.3259, "fi": 2.51113}, "gini_cpc": 0.040321, "gini_cps": 0.057729, "bpic": 5.033449, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0493823382018708, "fi": 1.1963053053585189}, "t": 204.3}
{"event": "eval", "step": 2000, "phase": 1, "bpb": {"en": 2.13627, "tr": 2.33494, "fi": 2.51854}, "bytes_per_chunk": {"en": 5.329, "tr": 5.475, "fi": 4.703}, "bpb_mean": 2.32992, "frozen_inventory": false, "t": 205.0}
{"event": "train", "step": 2000, "arch": "pkm", "phase": 1, "objective": "global", "lm_loss": 1.65236, "bpb": 2.38385, "chunk_loss": 0.03066, "F": 0.20407, "G": 0.22864, "chunks": 208.97, "bytes_per_chunk": 4.9, "boundary_entropy": 0.21608, "router_noise": 0.0, "grad_norm": 0.8576, "lr": 0.000245, "router_trainable": true, "sec_per_step": 0.11, "alpha": 0.03, "t": 205.8}
{"event": "probe", "step": 2048, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3231.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.185658, "cps": 25.242188, "mean_p": 0.214964, "boundary_entropy": 0.226206, "bpb": 2.115281}, "tr": {"chunks": 3675.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.203005, "cps": 28.710938, "mean_p": 0.212434, "boundary_entropy": 0.209234, "bpb": 2.328131}, "fi": {"chunks": 4283.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.227686, "cps": 33.460938, "mean_p": 0.244013, "boundary_entropy": 0.215452, "bpb": 2.495373}}, "bpb": {"en": 2.11528, "tr": 2.32813, "fi": 2.49537}, "gini_cpc": 0.045459, "gini_cps": 0.062681, "bpic": 5.070337, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0934374748518114, "fi": 1.2263751819089632}, "t": 211.2}
{"event": "train", "step": 2050, "arch": "pkm", "phase": 1, "objective": "global", "lm_loss": 1.49577, "bpb": 2.15794, "chunk_loss": 0.030263, "F": 0.1795, "G": 0.20689, "chunks": 183.81, "bytes_per_chunk": 5.571, "boundary_entropy": 0.21819, "router_noise": 0.0, "grad_norm": 0.8952, "lr": 0.0002419, "router_trainable": true, "sec_per_step": 0.13, "alpha": 0.03, "t": 212.3}
{"event": "train", "step": 2100, "arch": "pkm", "phase": 1, "objective": "global", "lm_loss": 1.55407, "bpb": 2.24204, "chunk_loss": 0.030277, "F": 0.18695, "G": 0.21456, "chunks": 191.44, "bytes_per_chunk": 5.349, "boundary_entropy": 0.21518, "router_noise": 0.0, "grad_norm": 0.898, "lr": 0.0002388, "router_trainable": true, "sec_per_step": 0.087, "alpha": 0.03, "t": 216.7}
{"event": "train", "step": 2150, "arch": "pkm", "phase": 1, "objective": "global", "lm_loss": 1.51264, "bpb": 2.18227, "chunk_loss": 0.030302, "F": 0.18869, "G": 0.21474, "chunks": 193.22, "bytes_per_chunk": 5.3, "boundary_entropy": 0.21454, "router_noise": 0.0, "grad_norm": 0.7816, "lr": 0.0002357, "router_trainable": true, "sec_per_step": 0.108, "alpha": 0.03, "t": 222.1}
{"event": "probe", "step": 2200, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3248.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.186634, "cps": 25.375, "mean_p": 0.212503, "boundary_entropy": 0.213353, "bpb": 2.099771}, "tr": {"chunks": 3721.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.205546, "cps": 29.070312, "mean_p": 0.208101, "boundary_entropy": 0.191817, "bpb": 2.304228}, "fi": {"chunks": 4107.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.21833, "cps": 32.085938, "mean_p": 0.24035, "boundary_entropy": 0.212046, "bpb": 2.470613}}, "bpb": {"en": 2.09977, "tr": 2.30423, "fi": 2.47061}, "gini_cpc": 0.034611, "gini_cps": 0.051703, "bpic": 5.122066, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.1013293628495289, "fi": 1.1698250558511383}, "t": 225.9}
{"event": "eval", "step": 2200, "phase": 1, "bpb": {"en": 2.11003, "tr": 2.31068, "fi": 2.47848}, "bytes_per_chunk": {"en": 5.5, "tr": 5.386, "fi": 5.004}, "bpb_mean": 2.29973, "frozen_inventory": false, "t": 226.5}
{"event": "train", "step": 2200, "arch": "pkm", "phase": 1, "objective": "global", "lm_loss": 1.48434, "bpb": 2.14144, "chunk_loss": 0.030248, "F": 0.18365, "G": 0.20817, "chunks": 188.06, "bytes_per_chunk": 5.445, "boundary_entropy": 0.20745, "router_noise": 0.0, "grad_norm": 0.7096, "lr": 0.0002325, "router_trainable": true, "sec_per_step": 0.098, "alpha": 0.03, "t": 227.0}
{"event": "train", "step": 2250, "arch": "pkm", "phase": 1, "objective": "global", "lm_loss": 1.50306, "bpb": 2.16846, "chunk_loss": 0.030294, "F": 0.18842, "G": 0.21192, "chunks": 192.94, "bytes_per_chunk": 5.307, "boundary_entropy": 0.20053, "router_noise": 0.0, "grad_norm": 0.7771, "lr": 0.0002293, "router_trainable": true, "sec_per_step": 0.071, "alpha": 0.03, "t": 230.5}
{"event": "train", "step": 2300, "arch": "pkm", "phase": 1, "objective": "global", "lm_loss": 1.52822, "bpb": 2.20476, "chunk_loss": 0.030183, "F": 0.18201, "G": 0.20894, "chunks": 186.38, "bytes_per_chunk": 5.494, "boundary_entropy": 0.2055, "router_noise": 0.0, "grad_norm": 0.764, "lr": 0.000226, "router_trainable": true, "sec_per_step": 0.067, "alpha": 0.03, "t": 233.9}
{"event": "train", "step": 2350, "arch": "pkm", "phase": 1, "objective": "global", "lm_loss": 1.57522, "bpb": 2.27256, "chunk_loss": 0.030326, "F": 0.19086, "G": 0.21528, "chunks": 195.44, "bytes_per_chunk": 5.24, "boundary_entropy": 0.20196, "router_noise": 0.0, "grad_norm": 0.8769, "lr": 0.0002226, "router_trainable": true, "sec_per_step": 0.071, "alpha": 0.03, "t": 237.4}
{"event": "probe", "step": 2400, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3354.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.192725, "cps": 26.203125, "mean_p": 0.217714, "boundary_entropy": 0.208189, "bpb": 2.076195}, "tr": {"chunks": 3704.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.204607, "cps": 28.9375, "mean_p": 0.208645, "boundary_entropy": 0.18704, "bpb": 2.26391}, "fi": {"chunks": 4299.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.228536, "cps": 33.585938, "mean_p": 0.244071, "boundary_entropy": 0.198783, "bpb": 2.416202}}, "bpb": {"en": 2.07619, "tr": 2.26391, "fi": 2.4162}, "gini_cpc": 0.038145, "gini_cps": 0.055472, "bpic": 4.995333, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0616503041579703, "fi": 1.1858141370295936}, "t": 241.1}
{"event": "eval", "step": 2400, "phase": 1, "bpb": {"en": 2.08811, "tr": 2.27093, "fi": 2.42498}, "bytes_per_chunk": {"en": 5.296, "tr": 5.395, "fi": 4.728}, "bpb_mean": 2.26134, "frozen_inventory": false, "t": 241.5}
{"event": "phase_transition", "step": 2400, "phase": 2, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0, "t": 241.9}
{"event": "train", "step": 2400, "arch": "pkm", "phase": 2, "objective": "global", "lm_loss": 1.51048, "bpb": 2.17916, "chunk_loss": 0.030459, "F": 0.19724, "G": 0.21984, "chunks": 201.97, "bytes_per_chunk": 5.07, "boundary_entropy": 0.19615, "router_noise": 0.0, "grad_norm": 0.7743, "lr": 0.0002192, "router_trainable": true, "sec_per_step": 0.092, "alpha": 0.03, "t": 242.0}
{"event": "train", "step": 2450, "arch": "pkm", "phase": 2, "objective": "global", "lm_loss": 1.47383, "bpb": 2.12629, "chunk_loss": 0.030266, "F": 0.18634, "G": 0.20907, "chunks": 190.81, "bytes_per_chunk": 5.367, "boundary_entropy": 0.19602, "router_noise": 0.0, "grad_norm": 0.7124, "lr": 0.0002158, "router_trainable": true, "sec_per_step": 0.061, "alpha": 0.03, "t": 245.1}
{"event": "train", "step": 2500, "arch": "pkm", "phase": 2, "objective": "global", "lm_loss": 1.45122, "bpb": 2.09366, "chunk_loss": 0.030331, "F": 0.19086, "G": 0.21321, "chunks": 195.44, "bytes_per_chunk": 5.24, "boundary_entropy": 0.19482, "router_noise": 0.0, "grad_norm": 0.7952, "lr": 0.0002123, "router_trainable": true, "sec_per_step": 0.066, "alpha": 0.03, "t": 248.4}
{"event": "train", "step": 2550, "arch": "pkm", "phase": 2, "objective": "global", "lm_loss": 1.47275, "bpb": 2.12473, "chunk_loss": 0.030275, "F": 0.18909, "G": 0.20979, "chunks": 193.62, "bytes_per_chunk": 5.289, "boundary_entropy": 0.18536, "router_noise": 0.0, "grad_norm": 0.7591, "lr": 0.0002088, "router_trainable": true, "sec_per_step": 0.063, "alpha": 0.03, "t": 251.6}
{"event": "probe", "step": 2600, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3229.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.185543, "cps": 25.226562, "mean_p": 0.210892, "boundary_entropy": 0.199245, "bpb": 2.059218}, "tr": {"chunks": 3660.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.202176, "cps": 28.59375, "mean_p": 0.205787, "boundary_entropy": 0.186252, "bpb": 2.256304}, "fi": {"chunks": 4253.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.226091, "cps": 33.226562, "mean_p": 0.240969, "boundary_entropy": 0.190409, "bpb": 2.39903}}, "bpb": {"en": 2.05922, "tr": 2.2563, "fi": 2.39903}, "gini_cpc": 0.04404, "gini_cps": 0.06127, "bpic": 5.091725, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.089648961167068, "fi": 1.2185393952942836}, "t": 254.5}
{"event": "eval", "step": 2600, "phase": 2, "bpb": {"en": 2.07102, "tr": 2.26395, "fi": 2.40602}, "bytes_per_chunk": {"en": 5.495, "tr": 5.428, "fi": 4.782}, "bpb_mean": 2.247, "frozen_inventory": false, "t": 254.8}
{"event": "train", "step": 2600, "arch": "pkm", "phase": 2, "objective": "global", "lm_loss": 1.43442, "bpb": 2.06944, "chunk_loss": 0.03021, "F": 0.18176, "G": 0.20688, "chunks": 186.12, "bytes_per_chunk": 5.502, "boundary_entropy": 0.19427, "router_noise": 0.0, "grad_norm": 0.7922, "lr": 0.0002053, "router_trainable": true, "sec_per_step": 0.075, "alpha": 0.03, "t": 255.3}
{"event": "train", "step": 2650, "arch": "pkm", "phase": 2, "objective": "global", "lm_loss": 1.48027, "bpb": 2.13558, "chunk_loss": 0.030169, "F": 0.18066, "G": 0.20467, "chunks": 185.0, "bytes_per_chunk": 5.535, "boundary_entropy": 0.18771, "router_noise": 0.0, "grad_norm": 0.748, "lr": 0.0002017, "router_trainable": true, "sec_per_step": 0.063, "alpha": 0.03, "t": 258.5}
{"event": "train", "step": 2700, "arch": "pkm", "phase": 2, "objective": "global", "lm_loss": 1.46803, "bpb": 2.11791, "chunk_loss": 0.030262, "F": 0.1843, "G": 0.20728, "chunks": 188.72, "bytes_per_chunk": 5.426, "boundary_entropy": 0.19088, "router_noise": 0.0, "grad_norm": 0.743, "lr": 0.0001981, "router_trainable": true, "sec_per_step": 0.062, "alpha": 0.03, "t": 261.6}
{"event": "train", "step": 2750, "arch": "pkm", "phase": 2, "objective": "global", "lm_loss": 1.47022, "bpb": 2.12107, "chunk_loss": 0.030288, "F": 0.18759, "G": 0.20946, "chunks": 192.09, "bytes_per_chunk": 5.331, "boundary_entropy": 0.18976, "router_noise": 0.0, "grad_norm": 0.7087, "lr": 0.0001945, "router_trainable": true, "sec_per_step": 0.069, "alpha": 0.03, "t": 265.0}
{"event": "probe", "step": 2800, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3251.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.186807, "cps": 25.398438, "mean_p": 0.209538, "boundary_entropy": 0.189821, "bpb": 2.041524}, "tr": {"chunks": 3675.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.203005, "cps": 28.710938, "mean_p": 0.202542, "boundary_entropy": 0.176894, "bpb": 2.208806}, "fi": {"chunks": 4196.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.223061, "cps": 32.78125, "mean_p": 0.236105, "boundary_entropy": 0.181574, "bpb": 2.361891}}, "bpb": {"en": 2.04152, "tr": 2.20881, "fi": 2.36189}, "gini_cpc": 0.039436, "gini_cps": 0.056644, "bpic": 5.100881, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0867106986300223, "fi": 1.1940726383433609}, "t": 270.6}
{"event": "eval", "step": 2800, "phase": 2, "bpb": {"en": 2.05464, "tr": 2.21515, "fi": 2.36771}, "bytes_per_chunk": {"en": 5.463, "tr": 5.433, "fi": 4.87}, "bpb_mean": 2.2125, "frozen_inventory": false, "t": 271.3}
{"event": "train", "step": 2800, "arch": "pkm", "phase": 2, "objective": "global", "lm_loss": 1.42256, "bpb": 2.05232, "chunk_loss": 0.030281, "F": 0.18674, "G": 0.20892, "chunks": 191.22, "bytes_per_chunk": 5.355, "boundary_entropy": 0.18698, "router_noise": 0.0, "grad_norm": 0.7894, "lr": 0.0001909, "router_trainable": true, "sec_per_step": 0.135, "alpha": 0.03, "t": 271.8}
{"event": "train", "step": 2850, "arch": "pkm", "phase": 2, "objective": "global", "lm_loss": 1.47182, "bpb": 2.12339, "chunk_loss": 0.030391, "F": 0.19348, "G": 0.21703, "chunks": 198.12, "bytes_per_chunk": 5.168, "boundary_entropy": 0.19092, "router_noise": 0.0, "grad_norm": 0.779, "lr": 0.0001872, "router_trainable": true, "sec_per_step": 0.102, "alpha": 0.03, "t": 276.9}
{"event": "train", "step": 2900, "arch": "pkm", "phase": 2, "objective": "global", "lm_loss": 1.40259, "bpb": 2.02351, "chunk_loss": 0.030142, "F": 0.17844, "G": 0.20231, "chunks": 182.72, "bytes_per_chunk": 5.604, "boundary_entropy": 0.1834, "router_noise": 0.0, "grad_norm": 0.6903, "lr": 0.0001835, "router_trainable": true, "sec_per_step": 0.109, "alpha": 0.03, "t": 282.4}
{"event": "train", "step": 2950, "arch": "pkm", "phase": 2, "objective": "global", "lm_loss": 1.54852, "bpb": 2.23404, "chunk_loss": 0.030427, "F": 0.1951, "G": 0.21321, "chunks": 199.78, "bytes_per_chunk": 5.126, "boundary_entropy": 0.17406, "router_noise": 0.0, "grad_norm": 0.7355, "lr": 0.0001799, "router_trainable": true, "sec_per_step": 0.106, "alpha": 0.03, "t": 287.7}
{"event": "probe", "step": 3000, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3261.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.187381, "cps": 25.476562, "mean_p": 0.211701, "boundary_entropy": 0.18382, "bpb": 2.023645}, "tr": {"chunks": 3657.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.202011, "cps": 28.570312, "mean_p": 0.200976, "boundary_entropy": 0.167319, "bpb": 2.191446}, "fi": {"chunks": 4285.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.227792, "cps": 33.476562, "mean_p": 0.236657, "boundary_entropy": 0.168881, "bpb": 2.341015}}, "bpb": {"en": 2.02365, "tr": 2.19145, "fi": 2.34101}, "gini_cpc": 0.043651, "gini_cps": 0.060936, "bpic": 5.064001, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.078071909991081, "fi": 1.2156603842698666}, "t": 293.6}
{"event": "eval", "step": 3000, "phase": 2, "bpb": {"en": 2.03665, "tr": 2.20032, "fi": 2.34611}, "bytes_per_chunk": {"en": 5.49, "tr": 5.467, "fi": 4.792}, "bpb_mean": 2.19436, "frozen_inventory": false, "t": 294.5}
{"event": "train", "step": 3000, "arch": "pkm", "phase": 2, "objective": "global", "lm_loss": 1.49533, "bpb": 2.1573, "chunk_loss": 0.030281, "F": 0.185, "G": 0.20683, "chunks": 189.44, "bytes_per_chunk": 5.405, "boundary_entropy": 0.17658, "router_noise": 0.0, "grad_norm": 0.8017, "lr": 0.0001761, "router_trainable": true, "sec_per_step": 0.15, "alpha": 0.03, "t": 295.2}
{"event": "train", "step": 3050, "arch": "pkm", "phase": 2, "objective": "global", "lm_loss": 1.448, "bpb": 2.08902, "chunk_loss": 0.03032, "F": 0.1882, "G": 0.21129, "chunks": 192.72, "bytes_per_chunk": 5.313, "boundary_entropy": 0.18221, "router_noise": 0.0, "grad_norm": 0.6966, "lr": 0.0001724, "router_trainable": true, "sec_per_step": 0.117, "alpha": 0.03, "t": 301.0}
{"event": "train", "step": 3100, "arch": "pkm", "phase": 2, "objective": "global", "lm_loss": 1.49779, "bpb": 2.16085, "chunk_loss": 0.03028, "F": 0.18777, "G": 0.2071, "chunks": 192.28, "bytes_per_chunk": 5.326, "boundary_entropy": 0.17522, "router_noise": 0.0, "grad_norm": 0.7419, "lr": 0.0001687, "router_trainable": true, "sec_per_step": 0.107, "alpha": 0.03, "t": 306.3}
{"event": "train", "step": 3150, "arch": "pkm", "phase": 2, "objective": "global", "lm_loss": 1.46255, "bpb": 2.11001, "chunk_loss": 0.030253, "F": 0.18515, "G": 0.20667, "chunks": 189.59, "bytes_per_chunk": 5.401, "boundary_entropy": 0.17982, "router_noise": 0.0, "grad_norm": 0.7593, "lr": 0.000165, "router_trainable": true, "sec_per_step": 0.112, "alpha": 0.03, "t": 312.0}
{"event": "probe", "step": 3200, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3262.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.187439, "cps": 25.484375, "mean_p": 0.210228, "boundary_entropy": 0.183656, "bpb": 2.006631}, "tr": {"chunks": 3705.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.204662, "cps": 28.945312, "mean_p": 0.200814, "boundary_entropy": 0.162888, "bpb": 2.162976}, "fi": {"chunks": 4297.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.22843, "cps": 33.570312, "mean_p": 0.235975, "boundary_entropy": 0.167679, "bpb": 2.312785}}, "bpb": {"en": 2.00663, "tr": 2.16298, "fi": 2.31278}, "gini_cpc": 0.044039, "gini_cps": 0.061257, "bpic": 5.036577, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.091887324500822, "fi": 1.2186910837648934}, "t": 318.2}
{"event": "eval", "step": 3200, "phase": 2, "bpb": {"en": 2.01852, "tr": 2.16878, "fi": 2.319}, "bytes_per_chunk": {"en": 5.473, "tr": 5.401, "fi": 4.73}, "bpb_mean": 2.16877, "frozen_inventory": false, "t": 319.1}
{"event": "train", "step": 3200, "arch": "pkm", "phase": 2, "objective": "global", "lm_loss": 1.44407, "bpb": 2.08336, "chunk_loss": 0.030232, "F": 0.18631, "G": 0.20793, "chunks": 190.78, "bytes_per_chunk": 5.367, "boundary_entropy": 0.17878, "router_noise": 0.0, "grad_norm": 0.6675, "lr": 0.0001613, "router_trainable": true, "sec_per_step": 0.153, "alpha": 0.03, "t": 319.6}
{"event": "train", "step": 3250, "arch": "pkm", "phase": 2, "objective": "global", "lm_loss": 1.44723, "bpb": 2.08791, "chunk_loss": 0.030295, "F": 0.18549, "G": 0.20584, "chunks": 189.94, "bytes_per_chunk": 5.391, "boundary_entropy": 0.17186, "router_noise": 0.0, "grad_norm": 0.7384, "lr": 0.0001576, "router_trainable": true, "sec_per_step": 0.119, "alpha": 0.03, "t": 325.6}
{"event": "train", "step": 3300, "arch": "pkm", "phase": 2, "objective": "global", "lm_loss": 1.4257, "bpb": 2.05685, "chunk_loss": 0.030305, "F": 0.18668, "G": 0.20985, "chunks": 191.16, "bytes_per_chunk": 5.357, "boundary_entropy": 0.1768, "router_noise": 0.0, "grad_norm": 0.7243, "lr": 0.0001539, "router_trainable": true, "sec_per_step": 0.108, "alpha": 0.03, "t": 330.9}
{"event": "train", "step": 3350, "arch": "pkm", "phase": 2, "objective": "global", "lm_loss": 1.36867, "bpb": 1.97458, "chunk_loss": 0.030169, "F": 0.18085, "G": 0.2057, "chunks": 185.19, "bytes_per_chunk": 5.53, "boundary_entropy": 0.18026, "router_noise": 0.0, "grad_norm": 0.7734, "lr": 0.0001501, "router_trainable": true, "sec_per_step": 0.113, "alpha": 0.03, "t": 336.6}
{"event": "probe", "step": 3400, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3151.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.181061, "cps": 24.617188, "mean_p": 0.206873, "boundary_entropy": 0.180274, "bpb": 1.996384}, "tr": {"chunks": 3688.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.203723, "cps": 28.8125, "mean_p": 0.200535, "boundary_entropy": 0.163285, "bpb": 2.154369}, "fi": {"chunks": 4274.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.227207, "cps": 33.390625, "mean_p": 0.235237, "boundary_entropy": 0.1687, "bpb": 2.280266}}, "bpb": {"en": 1.99638, "tr": 2.15437, "fi": 2.28027}, "gini_cpc": 0.050269, "gini_cps": 0.067369, "bpic": 5.105012, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.125164646820769, "fi": 1.2548688864313153}, "t": 343.1}
{"event": "eval", "step": 3400, "phase": 2, "bpb": {"en": 2.00827, "tr": 2.16278, "fi": 2.28751}, "bytes_per_chunk": {"en": 5.641, "tr": 5.407, "fi": 4.789}, "bpb_mean": 2.15285, "frozen_inventory": false, "t": 343.8}
{"event": "train", "step": 3400, "arch": "pkm", "phase": 2, "objective": "global", "lm_loss": 1.43351, "bpb": 2.06811, "chunk_loss": 0.030196, "F": 0.18161, "G": 0.20305, "chunks": 185.97, "bytes_per_chunk": 5.506, "boundary_entropy": 0.17252, "router_noise": 0.0, "grad_norm": 0.8038, "lr": 0.0001465, "router_trainable": true, "sec_per_step": 0.16, "alpha": 0.03, "t": 344.6}
{"event": "train", "step": 3450, "arch": "pkm", "phase": 2, "objective": "global", "lm_loss": 1.39873, "bpb": 2.01794, "chunk_loss": 0.030277, "F": 0.18787, "G": 0.20844, "chunks": 192.38, "bytes_per_chunk": 5.323, "boundary_entropy": 0.16869, "router_noise": 0.0, "grad_norm": 0.7953, "lr": 0.0001428, "router_trainable": true, "sec_per_step": 0.102, "alpha": 0.03, "t": 349.7}
{"event": "train", "step": 3500, "arch": "pkm", "phase": 2, "objective": "global", "lm_loss": 1.38411, "bpb": 1.99684, "chunk_loss": 0.030112, "F": 0.1752, "G": 0.20062, "chunks": 179.41, "bytes_per_chunk": 5.708, "boundary_entropy": 0.17711, "router_noise": 0.0, "grad_norm": 0.7828, "lr": 0.0001391, "router_trainable": true, "sec_per_step": 0.113, "alpha": 0.03, "t": 355.4}
{"event": "train", "step": 3550, "arch": "pkm", "phase": 2, "objective": "global", "lm_loss": 1.41481, "bpb": 2.04114, "chunk_loss": 0.030217, "F": 0.18091, "G": 0.20372, "chunks": 185.25, "bytes_per_chunk": 5.528, "boundary_entropy": 0.16931, "router_noise": 0.0, "grad_norm": 0.7172, "lr": 0.0001355, "router_trainable": true, "sec_per_step": 0.117, "alpha": 0.03, "t": 361.2}
{"event": "probe", "step": 3600, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3260.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.187324, "cps": 25.46875, "mean_p": 0.209944, "boundary_entropy": 0.173304, "bpb": 1.97941}, "tr": {"chunks": 3674.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.20295, "cps": 28.703125, "mean_p": 0.201261, "boundary_entropy": 0.160256, "bpb": 2.119949}, "fi": {"chunks": 4287.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.227899, "cps": 33.492188, "mean_p": 0.234058, "boundary_entropy": 0.157143, "bpb": 2.256828}}, "bpb": {"en": 1.97941, "tr": 2.11995, "fi": 2.25683}, "gini_cpc": 0.043758, "gini_cps": 0.061017, "bpic": 5.055877, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0834156898375316, "fi": 1.2166008630246041}, "t": 367.1}
{"event": "eval", "step": 3600, "phase": 2, "bpb": {"en": 1.9925, "tr": 2.1272, "fi": 2.26288}, "bytes_per_chunk": {"en": 5.485, "tr": 5.426, "fi": 4.784}, "bpb_mean": 2.12753, "frozen_inventory": false, "t": 367.9}
{"event": "train", "step": 3600, "arch": "pkm", "phase": 2, "objective": "global", "lm_loss": 1.42089, "bpb": 2.04991, "chunk_loss": 0.030248, "F": 0.18283, "G": 0.20401, "chunks": 187.22, "bytes_per_chunk": 5.47, "boundary_entropy": 0.16536, "router_noise": 0.0, "grad_norm": 0.6745, "lr": 0.0001319, "router_trainable": true, "sec_per_step": 0.147, "alpha": 0.03, "t": 368.6}
{"event": "train", "step": 3650, "arch": "pkm", "phase": 2, "objective": "global", "lm_loss": 1.45043, "bpb": 2.09252, "chunk_loss": 0.030355, "F": 0.18704, "G": 0.21025, "chunks": 191.53, "bytes_per_chunk": 5.346, "boundary_entropy": 0.16973, "router_noise": 0.0, "grad_norm": 0.7404, "lr": 0.0001283, "router_trainable": true, "sec_per_step": 0.11, "alpha": 0.03, "t": 374.1}
{"event": "train", "step": 3700, "arch": "pkm", "phase": 2, "objective": "global", "lm_loss": 1.40545, "bpb": 2.02763, "chunk_loss": 0.03024, "F": 0.18576, "G": 0.20901, "chunks": 190.22, "bytes_per_chunk": 5.383, "boundary_entropy": 0.17152, "router_noise": 0.0, "grad_norm": 0.6892, "lr": 0.0001247, "router_trainable": true, "sec_per_step": 0.111, "alpha": 0.03, "t": 379.6}
{"event": "train", "step": 3750, "arch": "pkm", "phase": 2, "objective": "global", "lm_loss": 1.38687, "bpb": 2.00084, "chunk_loss": 0.030219, "F": 0.18048, "G": 0.204, "chunks": 184.81, "bytes_per_chunk": 5.541, "boundary_entropy": 0.16711, "router_noise": 0.0, "grad_norm": 0.6883, "lr": 0.0001212, "router_trainable": true, "sec_per_step": 0.104, "alpha": 0.03, "t": 384.8}
{"event": "probe", "step": 3800, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3254.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.186979, "cps": 25.421875, "mean_p": 0.210578, "boundary_entropy": 0.169969, "bpb": 1.966144}, "tr": {"chunks": 3777.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.208639, "cps": 29.507812, "mean_p": 0.202822, "boundary_entropy": 0.155812, "bpb": 2.115818}, "fi": {"chunks": 4299.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.228536, "cps": 33.585938, "mean_p": 0.236536, "boundary_entropy": 0.15924, "bpb": 2.24093}}, "bpb": {"en": 1.96614, "tr": 2.11582, "fi": 2.24093}, "gini_cpc": 0.044388, "gini_cps": 0.061489, "bpic": 5.007237, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.115842773074011, "fi": 1.2222558744920888}, "t": 390.7}
{"event": "eval", "step": 3800, "phase": 2, "bpb": {"en": 1.98105, "tr": 2.12454, "fi": 2.24839}, "bytes_per_chunk": {"en": 5.511, "tr": 5.301, "fi": 4.782}, "bpb_mean": 2.11799, "frozen_inventory": false, "t": 391.6}
{"event": "train", "step": 3800, "arch": "pkm", "phase": 2, "objective": "global", "lm_loss": 1.46221, "bpb": 2.10952, "chunk_loss": 0.030268, "F": 0.18945, "G": 0.20923, "chunks": 194.0, "bytes_per_chunk": 5.278, "boundary_entropy": 0.16386, "router_noise": 0.0, "grad_norm": 0.7518, "lr": 0.0001177, "router_trainable": true, "sec_per_step": 0.148, "alpha": 0.03, "t": 392.2}
{"event": "train", "step": 3850, "arch": "pkm", "phase": 2, "objective": "global", "lm_loss": 1.55312, "bpb": 2.24068, "chunk_loss": 0.030668, "F": 0.198, "G": 0.21574, "chunks": 202.75, "bytes_per_chunk": 5.051, "boundary_entropy": 0.15932, "router_noise": 0.0, "grad_norm": 0.7357, "lr": 0.0001142, "router_trainable": true, "sec_per_step": 0.107, "alpha": 0.03, "t": 397.6}
{"event": "train", "step": 3900, "arch": "pkm", "phase": 2, "objective": "global", "lm_loss": 1.44664, "bpb": 2.08706, "chunk_loss": 0.030381, "F": 0.18814, "G": 0.20997, "chunks": 192.66, "bytes_per_chunk": 5.315, "boundary_entropy": 0.16712, "router_noise": 0.0, "grad_norm": 0.7792, "lr": 0.0001108, "router_trainable": true, "sec_per_step": 0.107, "alpha": 0.03, "t": 402.9}
{"event": "train", "step": 3950, "arch": "pkm", "phase": 2, "objective": "global", "lm_loss": 1.50057, "bpb": 2.16487, "chunk_loss": 0.030252, "F": 0.18259, "G": 0.20412, "chunks": 186.97, "bytes_per_chunk": 5.477, "boundary_entropy": 0.16379, "router_noise": 0.0, "grad_norm": 0.7728, "lr": 0.0001074, "router_trainable": true, "sec_per_step": 0.103, "alpha": 0.03, "t": 408.1}
{"event": "probe", "step": 4000, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3215.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.184738, "cps": 25.117188, "mean_p": 0.208505, "boundary_entropy": 0.16861, "bpb": 1.956057}, "tr": {"chunks": 3742.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.206706, "cps": 29.234375, "mean_p": 0.201876, "boundary_entropy": 0.15328, "bpb": 2.096389}, "fi": {"chunks": 4308.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.229015, "cps": 33.65625, "mean_p": 0.235366, "boundary_entropy": 0.155073, "bpb": 2.219951}}, "bpb": {"en": 1.95606, "tr": 2.09639, "fi": 2.21995}, "gini_cpc": 0.047574, "gini_cps": 0.064684, "bpic": 5.03613, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.1189131416572646, "fi": 1.2396724625816617}, "t": 414.2}
{"event": "eval", "step": 4000, "phase": 2, "bpb": {"en": 1.97072, "tr": 2.10422, "fi": 2.22549}, "bytes_per_chunk": {"en": 5.55, "tr": 5.339, "fi": 4.753}, "bpb_mean": 2.10014, "frozen_inventory": false, "t": 415.0}
{"event": "train", "step": 4000, "arch": "pkm", "phase": 2, "objective": "global", "lm_loss": 1.42236, "bpb": 2.05203, "chunk_loss": 0.030253, "F": 0.18393, "G": 0.20425, "chunks": 188.34, "bytes_per_chunk": 5.437, "boundary_entropy": 0.16393, "router_noise": 0.0, "grad_norm": 0.726, "lr": 0.000104, "router_trainable": true, "sec_per_step": 0.15, "alpha": 0.03, "t": 415.6}
{"event": "train", "step": 4050, "arch": "pkm", "phase": 2, "objective": "global", "lm_loss": 1.43458, "bpb": 2.06966, "chunk_loss": 0.03027, "F": 0.18744, "G": 0.20722, "chunks": 191.94, "bytes_per_chunk": 5.335, "boundary_entropy": 0.15929, "router_noise": 0.0, "grad_norm": 0.7134, "lr": 0.0001007, "router_trainable": true, "sec_per_step": 0.112, "alpha": 0.03, "t": 421.2}
{"event": "probe", "step": 4096, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3216.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.184796, "cps": 25.125, "mean_p": 0.207903, "boundary_entropy": 0.163753, "bpb": 1.951393}, "tr": {"chunks": 3726.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.205822, "cps": 29.109375, "mean_p": 0.200493, "boundary_entropy": 0.149273, "bpb": 2.091823}, "fi": {"chunks": 4351.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.231301, "cps": 33.992188, "mean_p": 0.236746, "boundary_entropy": 0.147853, "bpb": 2.215572}}, "bpb": {"en": 1.95139, "tr": 2.09182, "fi": 2.21557}, "gini_cpc": 0.049851, "gini_cps": 0.067003, "bpic": 5.023643, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.113782472765708, "fi": 1.251656848525434}, "t": 426.5}
{"event": "train", "step": 4100, "arch": "pkm", "phase": 2, "objective": "global", "lm_loss": 1.37422, "bpb": 1.98259, "chunk_loss": 0.030291, "F": 0.18478, "G": 0.20576, "chunks": 189.22, "bytes_per_chunk": 5.412, "boundary_entropy": 0.16205, "router_noise": 0.0, "grad_norm": 0.7674, "lr": 9.75e-05, "router_trainable": true, "sec_per_step": 0.13, "alpha": 0.03, "t": 427.7}
{"event": "train", "step": 4150, "arch": "pkm", "phase": 2, "objective": "global", "lm_loss": 1.43249, "bpb": 2.06664, "chunk_loss": 0.030307, "F": 0.18762, "G": 0.20735, "chunks": 192.12, "bytes_per_chunk": 5.33, "boundary_entropy": 0.16085, "router_noise": 0.0, "grad_norm": 0.7728, "lr": 9.43e-05, "router_trainable": true, "sec_per_step": 0.109, "alpha": 0.03, "t": 433.1}
{"event": "probe", "step": 4200, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3266.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.187669, "cps": 25.515625, "mean_p": 0.208253, "boundary_entropy": 0.161243, "bpb": 1.947923}, "tr": {"chunks": 3755.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.207424, "cps": 29.335938, "mean_p": 0.201645, "boundary_entropy": 0.146308, "bpb": 2.079402}, "fi": {"chunks": 4324.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.229866, "cps": 33.78125, "mean_p": 0.235111, "boundary_entropy": 0.150895, "bpb": 2.20864}}, "bpb": {"en": 1.94792, "tr": 2.0794, "fi": 2.20864}, "gini_cpc": 0.045013, "gini_cps": 0.062171, "bpic": 5.000617, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.1052673212841846, "fi": 1.2248467146507775}, "t": 439.3}
{"event": "eval", "step": 4200, "phase": 2, "bpb": {"en": 1.96199, "tr": 2.08805, "fi": 2.21514}, "bytes_per_chunk": {"en": 5.454, "tr": 5.323, "fi": 4.747}, "bpb_mean": 2.08839, "frozen_inventory": false, "t": 440.2}
{"event": "phase_transition", "step": 4200, "phase": 3, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0, "t": 440.8}
{"event": "train", "step": 4200, "arch": "pkm", "phase": 3, "objective": "global", "lm_loss": 1.36628, "bpb": 1.97112, "chunk_loss": 0.030251, "F": 0.18616, "G": 0.20311, "chunks": 190.62, "bytes_per_chunk": 5.372, "boundary_entropy": 0.15309, "router_noise": 0.0, "grad_norm": 0.718, "lr": 9.12e-05, "router_trainable": true, "sec_per_step": 0.156, "alpha": 0.03, "t": 440.9}
{"event": "train", "step": 4250, "arch": "pkm", "phase": 3, "objective": "global", "lm_loss": 1.39769, "bpb": 2.01645, "chunk_loss": 0.030198, "F": 0.18378, "G": 0.20298, "chunks": 188.19, "bytes_per_chunk": 5.441, "boundary_entropy": 0.15621, "router_noise": 0.0, "grad_norm": 0.7147, "lr": 8.81e-05, "router_trainable": true, "sec_per_step": 0.111, "alpha": 0.03, "t": 446.5}
{"event": "train", "step": 4300, "arch": "pkm", "phase": 3, "objective": "global", "lm_loss": 1.39648, "bpb": 2.0147, "chunk_loss": 0.030166, "F": 0.18314, "G": 0.20157, "chunks": 187.53, "bytes_per_chunk": 5.46, "boundary_entropy": 0.15757, "router_noise": 0.0, "grad_norm": 0.7561, "lr": 8.5e-05, "router_trainable": true, "sec_per_step": 0.106, "alpha": 0.03, "t": 451.8}
{"event": "train", "step": 4350, "arch": "pkm", "phase": 3, "objective": "global", "lm_loss": 1.42434, "bpb": 2.05488, "chunk_loss": 0.030226, "F": 0.18283, "G": 0.20413, "chunks": 187.22, "bytes_per_chunk": 5.47, "boundary_entropy": 0.15837, "router_noise": 0.0, "grad_norm": 0.719, "lr": 8.21e-05, "router_trainable": true, "sec_per_step": 0.105, "alpha": 0.03, "t": 457.1}
{"event": "probe", "step": 4400, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3246.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.18652, "cps": 25.359375, "mean_p": 0.207113, "boundary_entropy": 0.160169, "bpb": 1.938305}, "tr": {"chunks": 3728.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.205933, "cps": 29.125, "mean_p": 0.199649, "boundary_entropy": 0.148308, "bpb": 2.068213}, "fi": {"chunks": 4307.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.228962, "cps": 33.648438, "mean_p": 0.232583, "boundary_entropy": 0.147264, "bpb": 2.191643}}, "bpb": {"en": 1.93831, "tr": 2.06821, "fi": 2.19164}, "gini_cpc": 0.045533, "gini_cps": 0.062701, "bpic": 5.028987, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.1040810527314282, "fi": 1.2275483108508796}, "t": 463.0}
{"event": "eval", "step": 4400, "phase": 3, "bpb": {"en": 1.9517, "tr": 2.07531, "fi": 2.19879}, "bytes_per_chunk": {"en": 5.475, "tr": 5.393, "fi": 4.776}, "bpb_mean": 2.07526, "frozen_inventory": false, "t": 463.9}
{"event": "train", "step": 4400, "arch": "pkm", "phase": 3, "objective": "global", "lm_loss": 1.37391, "bpb": 1.98214, "chunk_loss": 0.030279, "F": 0.1875, "G": 0.20575, "chunks": 192.0, "bytes_per_chunk": 5.333, "boundary_entropy": 0.15771, "router_noise": 0.0, "grad_norm": 0.6941, "lr": 7.92e-05, "router_trainable": true, "sec_per_step": 0.145, "alpha": 0.03, "t": 464.4}
{"event": "train", "step": 4450, "arch": "pkm", "phase": 3, "objective": "global", "lm_loss": 1.36876, "bpb": 1.9747, "chunk_loss": 0.030126, "F": 0.17526, "G": 0.19556, "chunks": 179.47, "bytes_per_chunk": 5.706, "boundary_entropy": 0.15573, "router_noise": 0.0, "grad_norm": 0.7245, "lr": 7.63e-05, "router_trainable": true, "sec_per_step": 0.106, "alpha": 0.03, "t": 469.7}
{"event": "train", "step": 4500, "arch": "pkm", "phase": 3, "objective": "global", "lm_loss": 1.34395, "bpb": 1.93891, "chunk_loss": 0.03021, "F": 0.1828, "G": 0.20222, "chunks": 187.19, "bytes_per_chunk": 5.47, "boundary_entropy": 0.15342, "router_noise": 0.0, "grad_norm": 0.7129, "lr": 7.36e-05, "router_trainable": true, "sec_per_step": 0.11, "alpha": 0.03, "t": 475.1}
{"event": "train", "step": 4550, "arch": "pkm", "phase": 3, "objective": "global", "lm_loss": 1.37265, "bpb": 1.98031, "chunk_loss": 0.030195, "F": 0.18344, "G": 0.20134, "chunks": 187.84, "bytes_per_chunk": 5.451, "boundary_entropy": 0.15012, "router_noise": 0.0, "grad_norm": 0.7203, "lr": 7.09e-05, "router_trainable": true, "sec_per_step": 0.11, "alpha": 0.03, "t": 480.6}
{"event": "probe", "step": 4600, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3272.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.188014, "cps": 25.5625, "mean_p": 0.208507, "boundary_entropy": 0.157834, "bpb": 1.927492}, "tr": {"chunks": 3781.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.20886, "cps": 29.539062, "mean_p": 0.202446, "boundary_entropy": 0.143897, "bpb": 2.061513}, "fi": {"chunks": 4361.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.231832, "cps": 34.070312, "mean_p": 0.235841, "boundary_entropy": 0.144966, "bpb": 2.183364}}, "bpb": {"en": 1.92749, "tr": 2.06151, "fi": 2.18336}, "gini_cpc": 0.046465, "gini_cps": 0.063606, "bpic": 4.970387, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.1108794966644955, "fi": 1.2330623247673194}, "t": 486.0}
{"event": "eval", "step": 4600, "phase": 3, "bpb": {"en": 1.9418, "tr": 2.06869, "fi": 2.18896}, "bytes_per_chunk": {"en": 5.451, "tr": 5.308, "fi": 4.704}, "bpb_mean": 2.06648, "frozen_inventory": false, "t": 486.7}
{"event": "train", "step": 4600, "arch": "pkm", "phase": 3, "objective": "global", "lm_loss": 1.3743, "bpb": 1.9827, "chunk_loss": 0.030288, "F": 0.18839, "G": 0.20727, "chunks": 192.91, "bytes_per_chunk": 5.308, "boundary_entropy": 0.15503, "router_noise": 0.0, "grad_norm": 0.7131, "lr": 6.82e-05, "router_trainable": true, "sec_per_step": 0.139, "alpha": 0.03, "t": 487.6}
{"event": "train", "step": 4650, "arch": "pkm", "phase": 3, "objective": "global", "lm_loss": 1.40101, "bpb": 2.02124, "chunk_loss": 0.030352, "F": 0.19333, "G": 0.21001, "chunks": 197.97, "bytes_per_chunk": 5.173, "boundary_entropy": 0.14758, "router_noise": 0.0, "grad_norm": 0.712, "lr": 6.57e-05, "router_trainable": true, "sec_per_step": 0.109, "alpha": 0.03, "t": 493.0}
{"event": "train", "step": 4700, "arch": "pkm", "phase": 3, "objective": "global", "lm_loss": 1.43158, "bpb": 2.06533, "chunk_loss": 0.030578, "F": 0.19812, "G": 0.21539, "chunks": 202.88, "bytes_per_chunk": 5.047, "boundary_entropy": 0.15524, "router_noise": 0.0, "grad_norm": 0.7189, "lr": 6.32e-05, "router_trainable": true, "sec_per_step": 0.105, "alpha": 0.03, "t": 498.3}
{"event": "train", "step": 4750, "arch": "pkm", "phase": 3, "objective": "global", "lm_loss": 1.41976, "bpb": 2.04829, "chunk_loss": 0.030273, "F": 0.18994, "G": 0.20717, "chunks": 194.5, "bytes_per_chunk": 5.265, "boundary_entropy": 0.1521, "router_noise": 0.0, "grad_norm": 0.7122, "lr": 6.08e-05, "router_trainable": true, "sec_per_step": 0.107, "alpha": 0.03, "t": 503.7}
{"event": "probe", "step": 4800, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3247.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.186577, "cps": 25.367188, "mean_p": 0.206606, "boundary_entropy": 0.154071, "bpb": 1.919697}, "tr": {"chunks": 3751.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.207203, "cps": 29.304688, "mean_p": 0.201182, "boundary_entropy": 0.143397, "bpb": 2.046379}, "fi": {"chunks": 4371.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.232364, "cps": 34.148438, "mean_p": 0.235086, "boundary_entropy": 0.139427, "bpb": 2.165907}}, "bpb": {"en": 1.9197, "tr": 2.04638, "fi": 2.16591}, "gini_cpc": 0.04875, "gini_cps": 0.06591, "bpic": 4.990061, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.1105505826334308, "fi": 1.245405429140604}, "t": 509.3}
{"event": "eval", "step": 4800, "phase": 3, "bpb": {"en": 1.9354, "tr": 2.05472, "fi": 2.17111}, "bytes_per_chunk": {"en": 5.494, "tr": 5.354, "fi": 4.687}, "bpb_mean": 2.05374, "frozen_inventory": false, "t": 509.9}
{"event": "train", "step": 4800, "arch": "pkm", "phase": 3, "objective": "global", "lm_loss": 1.38563, "bpb": 1.99904, "chunk_loss": 0.030284, "F": 0.18585, "G": 0.20509, "chunks": 190.31, "bytes_per_chunk": 5.381, "boundary_entropy": 0.1522, "router_noise": 0.0, "grad_norm": 0.6897, "lr": 5.85e-05, "router_trainable": true, "sec_per_step": 0.143, "alpha": 0.03, "t": 510.8}
{"event": "train", "step": 4850, "arch": "pkm", "phase": 3, "objective": "global", "lm_loss": 1.36716, "bpb": 1.9724, "chunk_loss": 0.030277, "F": 0.1875, "G": 0.20406, "chunks": 192.0, "bytes_per_chunk": 5.333, "boundary_entropy": 0.14853, "router_noise": 0.0, "grad_norm": 0.7754, "lr": 5.62e-05, "router_trainable": true, "sec_per_step": 0.114, "alpha": 0.03, "t": 516.5}
{"event": "train", "step": 4900, "arch": "pkm", "phase": 3, "objective": "global", "lm_loss": 1.39315, "bpb": 2.00989, "chunk_loss": 0.030288, "F": 0.18912, "G": 0.20491, "chunks": 193.66, "bytes_per_chunk": 5.288, "boundary_entropy": 0.14315, "router_noise": 0.0, "grad_norm": 0.6953, "lr": 5.41e-05, "router_trainable": true, "sec_per_step": 0.116, "alpha": 0.03, "t": 522.3}
{"event": "train", "step": 4950, "arch": "pkm", "phase": 3, "objective": "global", "lm_loss": 1.35137, "bpb": 1.94961, "chunk_loss": 0.030189, "F": 0.18286, "G": 0.20035, "chunks": 187.25, "bytes_per_chunk": 5.469, "boundary_entropy": 0.14932, "router_noise": 0.0, "grad_norm": 0.691, "lr": 5.2e-05, "router_trainable": true, "sec_per_step": 0.111, "alpha": 0.03, "t": 527.9}
{"event": "probe", "step": 5000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3208.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.184336, "cps": 25.0625, "mean_p": 0.206371, "boundary_entropy": 0.154638, "bpb": 1.913823}, "tr": {"chunks": 3692.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.203944, "cps": 28.84375, "mean_p": 0.198537, "boundary_entropy": 0.14122, "bpb": 2.041279}, "fi": {"chunks": 4352.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.231354, "cps": 34.0, "mean_p": 0.234501, "boundary_entropy": 0.141604, "bpb": 2.156501}}, "bpb": {"en": 1.91382, "tr": 2.04128, "fi": 2.1565}, "gini_cpc": 0.050587, "gini_cps": 0.067781, "bpic": 5.041948, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.1063713003851747, "fi": 1.2550665757593154}, "t": 533.5}
{"event": "eval", "step": 5000, "phase": 3, "bpb": {"en": 1.92788, "tr": 2.04908, "fi": 2.16106}, "bytes_per_chunk": {"en": 5.523, "tr": 5.41, "fi": 4.723}, "bpb_mean": 2.04601, "frozen_inventory": false, "t": 534.3}
{"event": "train", "step": 5000, "arch": "pkm", "phase": 3, "objective": "global", "lm_loss": 1.34213, "bpb": 1.93628, "chunk_loss": 0.030183, "F": 0.18262, "G": 0.20239, "chunks": 187.0, "bytes_per_chunk": 5.476, "boundary_entropy": 0.15439, "router_noise": 0.0, "grad_norm": 0.7169, "lr": 5e-05, "router_trainable": true, "sec_per_step": 0.14, "alpha": 0.03, "t": 534.9}
{"event": "train", "step": 5050, "arch": "pkm", "phase": 3, "objective": "global", "lm_loss": 1.36297, "bpb": 1.96635, "chunk_loss": 0.030227, "F": 0.18457, "G": 0.20357, "chunks": 189.0, "bytes_per_chunk": 5.418, "boundary_entropy": 0.15023, "router_noise": 0.0, "grad_norm": 0.7188, "lr": 4.81e-05, "router_trainable": true, "sec_per_step": 0.103, "alpha": 0.03, "t": 540.1}
{"event": "train", "step": 5100, "arch": "pkm", "phase": 3, "objective": "global", "lm_loss": 1.39474, "bpb": 2.01219, "chunk_loss": 0.030321, "F": 0.18961, "G": 0.2065, "chunks": 194.16, "bytes_per_chunk": 5.274, "boundary_entropy": 0.14469, "router_noise": 0.0, "grad_norm": 0.7414, "lr": 4.63e-05, "router_trainable": true, "sec_per_step": 0.109, "alpha": 0.03, "t": 545.5}
{"event": "train", "step": 5150, "arch": "pkm", "phase": 3, "objective": "global", "lm_loss": 1.38701, "bpb": 2.00104, "chunk_loss": 0.030326, "F": 0.19226, "G": 0.20711, "chunks": 196.88, "bytes_per_chunk": 5.201, "boundary_entropy": 0.14575, "router_noise": 0.0, "grad_norm": 0.7329, "lr": 4.45e-05, "router_trainable": true, "sec_per_step": 0.134, "alpha": 0.03, "t": 552.2}
{"event": "probe", "step": 5200, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3208.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.184336, "cps": 25.0625, "mean_p": 0.205182, "boundary_entropy": 0.153192, "bpb": 1.906602}, "tr": {"chunks": 3747.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.206982, "cps": 29.273438, "mean_p": 0.199738, "boundary_entropy": 0.138903, "bpb": 2.030825}, "fi": {"chunks": 4340.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.230716, "cps": 33.90625, "mean_p": 0.234133, "boundary_entropy": 0.141253, "bpb": 2.146829}}, "bpb": {"en": 1.9066, "tr": 2.03083, "fi": 2.14683}, "gini_cpc": 0.049708, "gini_cps": 0.066814, "bpic": 5.022753, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.12285299635516, "fi": 1.251605914245273}, "t": 557.8}
{"event": "eval", "step": 5200, "phase": 3, "bpb": {"en": 1.92213, "tr": 2.03924, "fi": 2.15184}, "bytes_per_chunk": {"en": 5.546, "tr": 5.35, "fi": 4.727}, "bpb_mean": 2.03774, "frozen_inventory": false, "t": 558.4}
{"event": "train", "step": 5200, "arch": "pkm", "phase": 3, "objective": "global", "lm_loss": 1.33626, "bpb": 1.92782, "chunk_loss": 0.030185, "F": 0.18149, "G": 0.19916, "chunks": 185.84, "bytes_per_chunk": 5.51, "boundary_entropy": 0.14429, "router_noise": 0.0, "grad_norm": 0.7091, "lr": 4.29e-05, "router_trainable": true, "sec_per_step": 0.135, "alpha": 0.03, "t": 559.0}
{"event": "train", "step": 5250, "arch": "pkm", "phase": 3, "objective": "global", "lm_loss": 1.33035, "bpb": 1.91928, "chunk_loss": 0.030183, "F": 0.18195, "G": 0.20037, "chunks": 186.31, "bytes_per_chunk": 5.496, "boundary_entropy": 0.14634, "router_noise": 0.0, "grad_norm": 0.6886, "lr": 4.14e-05, "router_trainable": true, "sec_per_step": 0.125, "alpha": 0.03, "t": 565.2}
{"event": "train", "step": 5300, "arch": "pkm", "phase": 3, "objective": "global", "lm_loss": 1.38692, "bpb": 2.0009, "chunk_loss": 0.030328, "F": 0.19183, "G": 0.20825, "chunks": 196.44, "bytes_per_chunk": 5.213, "boundary_entropy": 0.14493, "router_noise": 0.0, "grad_norm": 0.7254, "lr": 3.99e-05, "router_trainable": true, "sec_per_step": 0.12, "alpha": 0.03, "t": 571.2}
{"event": "train", "step": 5350, "arch": "pkm", "phase": 3, "objective": "global", "lm_loss": 1.32442, "bpb": 1.91074, "chunk_loss": 0.030195, "F": 0.18073, "G": 0.19998, "chunks": 185.06, "bytes_per_chunk": 5.533, "boundary_entropy": 0.14674, "router_noise": 0.0, "grad_norm": 0.7285, "lr": 3.86e-05, "router_trainable": true, "sec_per_step": 0.126, "alpha": 0.03, "t": 577.5}
{"event": "probe", "step": 5400, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3246.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.18652, "cps": 25.359375, "mean_p": 0.206456, "boundary_entropy": 0.14955, "bpb": 1.900599}, "tr": {"chunks": 3774.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.208474, "cps": 29.484375, "mean_p": 0.20072, "boundary_entropy": 0.135255, "bpb": 2.02317}, "fi": {"chunks": 4364.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.231992, "cps": 34.09375, "mean_p": 0.234641, "boundary_entropy": 0.139132, "bpb": 2.140227}}, "bpb": {"en": 1.9006, "tr": 2.02317, "fi": 2.14023}, "gini_cpc": 0.04835, "gini_cps": 0.065472, "bpic": 4.983486, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.11770437044217, "fi": 1.2437940163810632}, "t": 583.6}
{"event": "eval", "step": 5400, "phase": 3, "bpb": {"en": 1.91545, "tr": 2.03188, "fi": 2.14522}, "bytes_per_chunk": {"en": 5.468, "tr": 5.309, "fi": 4.7}, "bpb_mean": 2.03085, "frozen_inventory": false, "t": 584.3}
{"event": "train", "step": 5400, "arch": "pkm", "phase": 3, "objective": "global", "lm_loss": 1.39616, "bpb": 2.01423, "chunk_loss": 0.03049, "F": 0.19412, "G": 0.21008, "chunks": 198.78, "bytes_per_chunk": 5.151, "boundary_entropy": 0.14444, "router_noise": 0.0, "grad_norm": 0.7349, "lr": 3.73e-05, "router_trainable": true, "sec_per_step": 0.146, "alpha": 0.03, "t": 584.8}
{"event": "train", "step": 5450, "arch": "pkm", "phase": 3, "objective": "global", "lm_loss": 1.33398, "bpb": 1.92452, "chunk_loss": 0.0303, "F": 0.18735, "G": 0.20413, "chunks": 191.84, "bytes_per_chunk": 5.338, "boundary_entropy": 0.1437, "router_noise": 0.0, "grad_norm": 0.7571, "lr": 3.62e-05, "router_trainable": true, "sec_per_step": 0.102, "alpha": 0.03, "t": 590.0}
{"event": "train", "step": 5500, "arch": "pkm", "phase": 3, "objective": "global", "lm_loss": 1.32273, "bpb": 1.9083, "chunk_loss": 0.030262, "F": 0.18723, "G": 0.20427, "chunks": 191.72, "bytes_per_chunk": 5.341, "boundary_entropy": 0.14612, "router_noise": 0.0, "grad_norm": 0.7054, "lr": 3.51e-05, "router_trainable": true, "sec_per_step": 0.106, "alpha": 0.03, "t": 595.3}
{"event": "train", "step": 5550, "arch": "pkm", "phase": 3, "objective": "global", "lm_loss": 1.37042, "bpb": 1.9771, "chunk_loss": 0.030248, "F": 0.18732, "G": 0.20394, "chunks": 191.81, "bytes_per_chunk": 5.339, "boundary_entropy": 0.14548, "router_noise": 0.0, "grad_norm": 0.7575, "lr": 3.41e-05, "router_trainable": true, "sec_per_step": 0.109, "alpha": 0.03, "t": 600.7}
{"event": "probe", "step": 5600, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3254.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.186979, "cps": 25.421875, "mean_p": 0.205777, "boundary_entropy": 0.148485, "bpb": 1.895638}, "tr": {"chunks": 3760.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.2077, "cps": 29.375, "mean_p": 0.200178, "boundary_entropy": 0.13731, "bpb": 2.01787}, "fi": {"chunks": 4365.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.232045, "cps": 34.101562, "mean_p": 0.234582, "boundary_entropy": 0.139538, "bpb": 2.133732}}, "bpb": {"en": 1.89564, "tr": 2.01787, "fi": 2.13373}, "gini_cpc": 0.047938, "gini_cps": 0.065091, "bpic": 4.985675, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.1108204465867835, "fi": 1.2410204447913393}, "t": 606.7}
{"event": "eval", "step": 5600, "phase": 3, "bpb": {"en": 1.91163, "tr": 2.02677, "fi": 2.13918}, "bytes_per_chunk": {"en": 5.475, "tr": 5.341, "fi": 4.724}, "bpb_mean": 2.02586, "frozen_inventory": false, "t": 607.5}
{"event": "train", "step": 5600, "arch": "pkm", "phase": 3, "objective": "global", "lm_loss": 1.36124, "bpb": 1.96385, "chunk_loss": 0.03029, "F": 0.18677, "G": 0.2023, "chunks": 191.25, "bytes_per_chunk": 5.354, "boundary_entropy": 0.14194, "router_noise": 0.0, "grad_norm": 0.79, "lr": 3.33e-05, "router_trainable": true, "sec_per_step": 0.146, "alpha": 0.03, "t": 608.0}
{"event": "train", "step": 5650, "arch": "pkm", "phase": 3, "objective": "global", "lm_loss": 1.34866, "bpb": 1.94571, "chunk_loss": 0.030262, "F": 0.18704, "G": 0.20372, "chunks": 191.53, "bytes_per_chunk": 5.346, "boundary_entropy": 0.14413, "router_noise": 0.0, "grad_norm": 0.7319, "lr": 3.25e-05, "router_trainable": true, "sec_per_step": 0.106, "alpha": 0.03, "t": 613.3}
{"event": "train", "step": 5700, "arch": "pkm", "phase": 3, "objective": "global", "lm_loss": 1.3547, "bpb": 1.95442, "chunk_loss": 0.030165, "F": 0.18039, "G": 0.19762, "chunks": 184.72, "bytes_per_chunk": 5.544, "boundary_entropy": 0.14366, "router_noise": 0.0, "grad_norm": 0.7865, "lr": 3.18e-05, "router_trainable": true, "sec_per_step": 0.109, "alpha": 0.03, "t": 618.8}
{"event": "train", "step": 5750, "arch": "pkm", "phase": 3, "objective": "global", "lm_loss": 1.412, "bpb": 2.03708, "chunk_loss": 0.030243, "F": 0.18729, "G": 0.2031, "chunks": 191.78, "bytes_per_chunk": 5.339, "boundary_entropy": 0.14035, "router_noise": 0.0, "grad_norm": 0.7811, "lr": 3.13e-05, "router_trainable": true, "sec_per_step": 0.118, "alpha": 0.03, "t": 624.7}
{"event": "probe", "step": 5800, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3249.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.186692, "cps": 25.382812, "mean_p": 0.20658, "boundary_entropy": 0.149652, "bpb": 1.892337}, "tr": {"chunks": 3724.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.205712, "cps": 29.09375, "mean_p": 0.198166, "boundary_entropy": 0.136318, "bpb": 2.016617}, "fi": {"chunks": 4338.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.23061, "cps": 33.890625, "mean_p": 0.233678, "boundary_entropy": 0.136493, "bpb": 2.129715}}, "bpb": {"en": 1.89234, "tr": 2.01662, "fi": 2.12971}, "gini_cpc": 0.046995, "gini_cps": 0.064185, "bpic": 5.015648, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.1018780448331236, "fi": 1.2352420660334447}, "t": 630.4}
{"event": "eval", "step": 5800, "phase": 3, "bpb": {"en": 1.90808, "tr": 2.02363, "fi": 2.13424}, "bytes_per_chunk": {"en": 5.483, "tr": 5.375, "fi": 4.734}, "bpb_mean": 2.02198, "frozen_inventory": false, "t": 631.2}
{"event": "train", "step": 5800, "arch": "pkm", "phase": 3, "objective": "global", "lm_loss": 1.33596, "bpb": 1.92738, "chunk_loss": 0.030268, "F": 0.18289, "G": 0.20051, "chunks": 187.28, "bytes_per_chunk": 5.468, "boundary_entropy": 0.14452, "router_noise": 0.0, "grad_norm": 0.7407, "lr": 3.08e-05, "router_trainable": true, "sec_per_step": 0.141, "alpha": 0.03, "t": 631.8}
{"event": "train", "step": 5850, "arch": "pkm", "phase": 3, "objective": "global", "lm_loss": 1.35718, "bpb": 1.958, "chunk_loss": 0.030337, "F": 0.18924, "G": 0.20658, "chunks": 193.78, "bytes_per_chunk": 5.284, "boundary_entropy": 0.14357, "router_noise": 0.0, "grad_norm": 0.7603, "lr": 3.05e-05, "router_trainable": true, "sec_per_step": 0.108, "alpha": 0.03, "t": 637.1}
{"event": "train", "step": 5900, "arch": "pkm", "phase": 3, "objective": "global", "lm_loss": 1.29089, "bpb": 1.86236, "chunk_loss": 0.030171, "F": 0.18109, "G": 0.19844, "chunks": 185.44, "bytes_per_chunk": 5.522, "boundary_entropy": 0.14062, "router_noise": 0.0, "grad_norm": 0.7398, "lr": 3.02e-05, "router_trainable": true, "sec_per_step": 0.108, "alpha": 0.03, "t": 642.5}
{"event": "train", "step": 5950, "arch": "pkm", "phase": 3, "objective": "global", "lm_loss": 1.39422, "bpb": 2.01144, "chunk_loss": 0.030238, "F": 0.18649, "G": 0.20253, "chunks": 190.97, "bytes_per_chunk": 5.362, "boundary_entropy": 0.14038, "router_noise": 0.0, "grad_norm": 0.7842, "lr": 3.01e-05, "router_trainable": true, "sec_per_step": 0.109, "alpha": 0.03, "t": 648.0}
{"event": "train", "step": 5999, "arch": "pkm", "phase": 3, "objective": "global", "lm_loss": 1.39319, "bpb": 2.00996, "chunk_loss": 0.030375, "F": 0.19067, "G": 0.20651, "chunks": 195.25, "bytes_per_chunk": 5.245, "boundary_entropy": 0.14165, "router_noise": 0.0, "grad_norm": 0.7384, "lr": 3e-05, "router_trainable": true, "sec_per_step": 0.112, "alpha": 0.03, "t": 653.6}
{"event": "probe", "step": 6000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3276.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.188243, "cps": 25.59375, "mean_p": 0.205911, "boundary_entropy": 0.145697, "bpb": 1.890656}, "tr": {"chunks": 3770.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.208253, "cps": 29.453125, "mean_p": 0.200548, "boundary_entropy": 0.133176, "bpb": 2.010622}, "fi": {"chunks": 4380.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.232842, "cps": 34.21875, "mean_p": 0.234481, "boundary_entropy": 0.135, "bpb": 2.125969}}, "bpb": {"en": 1.89066, "tr": 2.01062, "fi": 2.12597}, "gini_cpc": 0.047244, "gini_cps": 0.064414, "bpic": 4.965167, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.1062951944297579, "fi": 1.2369223992742149}, "t": 654.1}
{"event": "eval", "step": 6000, "phase": 3, "bpb": {"en": 1.90636, "tr": 2.01783, "fi": 2.13105}, "bytes_per_chunk": {"en": 5.434, "tr": 5.318, "fi": 4.693}, "bpb_mean": 2.01841, "frozen_inventory": false, "t": 654.9}
{"event": "reps_dumped", "dirs": ["/local/scratch/sas245/beetle-hnet/runs/ali_pkm_s2/probe/reps/th0.30", "/local/scratch/sas245/beetle-hnet/runs/ali_pkm_s2/probe/reps/th0.40", "/local/scratch/sas245/beetle-hnet/runs/ali_pkm_s2/probe/reps/th0.50", "/local/scratch/sas245/beetle-hnet/runs/ali_pkm_s2/probe/reps/th0.60", "/local/scratch/sas245/beetle-hnet/runs/ali_pkm_s2/probe/reps/th0.70"], "t": 656.5}
{"event": "final", "run": "ali_pkm_s2", "arch": "pkm", "objective": "ratio", "condition": "A", "step": 6000, "params": 39425792, "seed": 2, "per_lang": {"en": {"cpc": 0.188243406309257, "cps": 25.59375, "bpb": 1.9115447383699333, "bytes_per_chunk": 5.369877605366928}, "tr": {"cpc": 0.2082527757830194, "cps": 29.453125, "bpb": 2.0105914173131527, "bytes_per_chunk": 5.340885244231651}, "fi": {"cpc": 0.23284248577959704, "cps": 34.21875, "bpb": 2.150320828794115, "bytes_per_chunk": 4.689139514708119}}, "bpb_mean": 2.024152328159067, "gini_cpc": 0.04724438710787515, "gini_cps": 0.06441449326098381, "bpic": 4.965167162611587, "mean_boundary_entropy": 0.14165401458740234, "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": 0.0, "probe_overhead_seq_frac": 0.0, "wall_sec": 658.1900057792664, "t": 658.2}
{"event": "hub_done", "repo": "AdaptiveChunking/ali_pkm_s2", "pushed": [0, 1, 2, 4, 6, 8, 16, 32, 64, 128, 256, 512, 1000, 1024, 2048, 3000, 4096, 6000], "failed": [], "ok": true, "t": 675.0}