{"event": "start", "run": "ali_hnet_s0", "arch": "hnet", "objective": "ratio", "condition": "A", "params": 22516992, "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", "t": 4.2}
{"event": "probe_schedule", "n": 40, "steps": [0, 1, 2, 4, 8, 16, 32, 64, 128, 200, 256, 400, 600, 800, 1000, 1200, 1400, 1600, 1800, 2000, 2200, 2400, 2600, 2800, 3000, 3200, 3400, 3600, 3800, 4000, 4200, 4400, 4600, 4800, 5000, 5200, 5400, 5600, 5800, 6000], "t": 4.2}
{"event": "probe", "step": 0, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 6384.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.366833, "cps": 49.875, "mean_p": 0.496725, "boundary_entropy": 0.686995, "bpb": 8.160238}, "tr": {"chunks": 5290.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.292217, "cps": 41.328125, "mean_p": 0.489567, "boundary_entropy": 0.687288, "bpb": 8.171434}, "fi": {"chunks": 8895.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.472862, "cps": 69.492188, "mean_p": 0.500688, "boundary_entropy": 0.687512, "bpb": 8.214758}}, "bpb": {"en": 8.16024, "tr": 8.17143, "fi": 8.21476}, "gini_cpc": 0.106395, "gini_cps": 0.116842, "bpic": 2.758131, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.7965927738475616, "fi": 1.2890367847879758}, "t": 5.5}
{"event": "train", "step": 0, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 5.66332, "bpb": 8.17044, "chunk_loss": 0.038537, "F": 0.28732, "G": 0.4892, "chunks": 294.22, "bytes_per_chunk": 3.48, "boundary_entropy": 0.69137, "router_noise": 0.0, "grad_norm": 11.0313, "lr": 1e-06, "router_trainable": true, "sec_per_step": 0.008, "alpha": 0.03, "t": 6.0}
{"event": "probe", "step": 1, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 6303.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.362179, "cps": 49.242188, "mean_p": 0.496453, "boundary_entropy": 0.686986, "bpb": 8.128223}, "tr": {"chunks": 5229.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.288847, "cps": 40.851562, "mean_p": 0.489371, "boundary_entropy": 0.687277, "bpb": 8.149128}, "fi": {"chunks": 8818.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.468768, "cps": 68.890625, "mean_p": 0.500418, "boundary_entropy": 0.687508, "bpb": 8.187239}}, "bpb": {"en": 8.12822, "tr": 8.14913, "fi": 8.18724}, "gini_cpc": 0.107116, "gini_cps": 0.117576, "bpic": 2.787813, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.7975260976227233, "fi": 1.2943002174161062}, "t": 6.5}
{"event": "probe", "step": 2, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 6171.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.354594, "cps": 48.210938, "mean_p": 0.495924, "boundary_entropy": 0.686967, "bpb": 8.064247}, "tr": {"chunks": 5127.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.283213, "cps": 40.054688, "mean_p": 0.488972, "boundary_entropy": 0.687254, "bpb": 8.105272}, "fi": {"chunks": 8679.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.461379, "cps": 67.804688, "mean_p": 0.499888, "boundary_entropy": 0.687499, "bpb": 8.130956}}, "bpb": {"en": 8.06425, "tr": 8.10527, "fi": 8.13096}, "gini_cpc": 0.10806, "gini_cps": 0.118536, "bpic": 2.839866, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.7986956880537021, "fi": 1.3011470419088613}, "t": 7.4}
{"event": "probe", "step": 4, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 5667.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.325634, "cps": 44.273438, "mean_p": 0.494022, "boundary_entropy": 0.686889, "bpb": 7.849016}, "tr": {"chunks": 4745.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.262111, "cps": 37.070312, "mean_p": 0.487542, "boundary_entropy": 0.687166, "bpb": 7.95186}, "fi": {"chunks": 8072.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.429111, "cps": 63.0625, "mean_p": 0.497939, "boundary_entropy": 0.687458, "bpb": 7.941675}}, "bpb": {"en": 7.84902, "tr": 7.95186, "fi": 7.94167}, "gini_cpc": 0.109487, "gini_cps": 0.119996, "bpic": 3.069249, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.804927143710069, "fi": 1.3177717024034938}, "t": 8.6}
{"event": "probe", "step": 8, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 4365.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.250819, "cps": 34.101562, "mean_p": 0.488766, "boundary_entropy": 0.686599, "bpb": 7.288686}, "tr": {"chunks": 3763.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.207866, "cps": 29.398438, "mean_p": 0.483654, "boundary_entropy": 0.686884, "bpb": 7.524687}, "fi": {"chunks": 6361.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.338153, "cps": 49.695312, "mean_p": 0.49219, "boundary_entropy": 0.687252, "bpb": 7.435216}}, "bpb": {"en": 7.28869, "tr": 7.52469, "fi": 7.43522}, "gini_cpc": 0.109003, "gini_cps": 0.119539, "bpic": 3.915522, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.8287499954916245, "fi": 1.348197086555154}, "t": 10.1}
{"event": "probe", "step": 16, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 4433.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.254726, "cps": 34.632812, "mean_p": 0.490982, "boundary_entropy": 0.686931, "bpb": 6.660662}, "tr": {"chunks": 4548.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.251229, "cps": 35.53125, "mean_p": 0.488393, "boundary_entropy": 0.687352, "bpb": 6.978461}, "fi": {"chunks": 5612.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.298336, "cps": 43.84375, "mean_p": 0.491245, "boundary_entropy": 0.687412, "bpb": 6.839399}}, "bpb": {"en": 6.66066, "tr": 6.97846, "fi": 6.8394}, "gini_cpc": 0.039046, "gini_cps": 0.053861, "bpic": 3.887617, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9862710682072292, "fi": 1.1712029775379806}, "t": 12.5}
{"event": "probe", "step": 32, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3977.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.228524, "cps": 31.070312, "mean_p": 0.494517, "boundary_entropy": 0.687548, "bpb": 6.307263}, "tr": {"chunks": 4051.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.223775, "cps": 31.648438, "mean_p": 0.49185, "boundary_entropy": 0.687954, "bpb": 6.636384}, "fi": {"chunks": 4291.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.228111, "cps": 33.523438, "mean_p": 0.492818, "boundary_entropy": 0.687994, "bpb": 6.431544}}, "bpb": {"en": 6.30726, "tr": 6.63638, "fi": 6.43154}, "gini_cpc": 0.004653, "gini_cps": 0.016993, "bpic": 4.605244, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9792198779395377, "fi": 0.9981944717557351}, "t": 16.2}
{"event": "train", "step": 50, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 4.22562, "bpb": 6.09627, "chunk_loss": 0.030571, "F": 0.17444, "G": 0.49127, "chunks": 178.62, "bytes_per_chunk": 5.733, "boundary_entropy": 0.69209, "router_noise": 0.0, "grad_norm": 2.439, "lr": 5.1e-05, "router_trainable": true, "sec_per_step": 0.267, "alpha": 0.03, "t": 19.4}
{"event": "probe", "step": 64, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 4031.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.231627, "cps": 31.492188, "mean_p": 0.488176, "boundary_entropy": 0.686612, "bpb": 5.623274}, "tr": {"chunks": 3854.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.212893, "cps": 30.109375, "mean_p": 0.488064, "boundary_entropy": 0.687511, "bpb": 6.010066}, "fi": {"chunks": 5289.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.281165, "cps": 41.320312, "mean_p": 0.492807, "boundary_entropy": 0.687694, "bpb": 5.81735}}, "bpb": {"en": 5.62327, "tr": 6.01007, "fi": 5.81735}, "gini_cpc": 0.06272, "gini_cps": 0.072618, "bpic": 4.306361, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9191205597924159, "fi": 1.2138723125536277}, "t": 22.4}
{"event": "train", "step": 100, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 3.21799, "bpb": 4.64257, "chunk_loss": 0.032212, "F": 0.20712, "G": 0.41986, "chunks": 212.09, "bytes_per_chunk": 4.828, "boundary_entropy": 0.6653, "router_noise": 0.0, "grad_norm": 4.0707, "lr": 0.000101, "router_trainable": true, "sec_per_step": 0.177, "alpha": 0.03, "t": 28.2}
{"event": "probe", "step": 128, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3617.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.207838, "cps": 28.257812, "mean_p": 0.38008, "boundary_entropy": 0.611674, "bpb": 4.17582}, "tr": {"chunks": 2943.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.16257, "cps": 22.992188, "mean_p": 0.412602, "boundary_entropy": 0.651821, "bpb": 4.508019}, "fi": {"chunks": 3357.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.178459, "cps": 26.226562, "mean_p": 0.39245, "boundary_entropy": 0.635351, "bpb": 4.422658}}, "bpb": {"en": 4.17582, "tr": 4.50802, "fi": 4.42266}, "gini_cpc": 0.054984, "gini_cps": 0.045309, "bpic": 5.720682, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.7821955162080482, "fi": 0.8586478152720571}, "t": 33.0}
{"event": "train", "step": 150, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 2.81899, "bpb": 4.06694, "chunk_loss": 0.031302, "F": 0.19724, "G": 0.3626, "chunks": 201.97, "bytes_per_chunk": 5.07, "boundary_entropy": 0.59385, "router_noise": 0.0, "grad_norm": 2.7369, "lr": 0.000151, "router_trainable": true, "sec_per_step": 0.169, "alpha": 0.03, "t": 36.7}
{"event": "probe", "step": 200, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3206.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.184221, "cps": 25.046875, "mean_p": 0.373345, "boundary_entropy": 0.57447, "bpb": 3.627596}, "tr": {"chunks": 3529.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.19494, "cps": 27.570312, "mean_p": 0.379063, "boundary_entropy": 0.596564, "bpb": 3.785996}, "fi": {"chunks": 3704.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.196906, "cps": 28.9375, "mean_p": 0.391642, "boundary_entropy": 0.614472, "bpb": 3.89807}}, "bpb": {"en": 3.6276, "tr": 3.786, "fi": 3.89807}, "gini_cpc": 0.01468, "gini_cps": 0.031804, "bpic": 5.43462, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0581852633724902, "fi": 1.0688572238413347}, "t": 44.2}
{"event": "eval", "step": 200, "phase": 1, "bpb": {"en": 3.63396, "tr": 3.79613, "fi": 3.89809}, "bytes_per_chunk": {"en": 5.58, "tr": 5.442, "fi": 5.555}, "bpb_mean": 3.77606, "frozen_inventory": false, "t": 45.5}
{"event": "train", "step": 200, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 2.5484, "bpb": 3.67657, "chunk_loss": 0.03064, "F": 0.17981, "G": 0.37642, "chunks": 184.12, "bytes_per_chunk": 5.561, "boundary_entropy": 0.58189, "router_noise": 0.0, "grad_norm": 1.5794, "lr": 0.000201, "router_trainable": true, "sec_per_step": 0.184, "alpha": 0.03, "t": 45.9}
{"event": "train", "step": 250, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 2.38898, "bpb": 3.44657, "chunk_loss": 0.0318, "F": 0.21164, "G": 0.3518, "chunks": 216.72, "bytes_per_chunk": 4.725, "boundary_entropy": 0.53376, "router_noise": 0.0, "grad_norm": 1.8217, "lr": 0.000251, "router_trainable": true, "sec_per_step": 0.154, "alpha": 0.03, "t": 53.6}
{"event": "probe", "step": 256, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3065.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.176119, "cps": 23.945312, "mean_p": 0.331001, "boundary_entropy": 0.514713, "bpb": 3.344377}, "tr": {"chunks": 3365.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.185881, "cps": 26.289062, "mean_p": 0.344061, "boundary_entropy": 0.528565, "bpb": 3.508018}, "fi": {"chunks": 4002.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.212748, "cps": 31.265625, "mean_p": 0.36679, "boundary_entropy": 0.560426, "bpb": 3.645182}}, "bpb": {"en": 3.34438, "tr": 3.50802, "fi": 3.64518}, "gini_cpc": 0.042487, "gini_cps": 0.05988, "bpic": 5.438267, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0554268987709354, "fi": 1.2079774919103856}, "t": 55.1}
{"event": "train", "step": 300, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 2.28078, "bpb": 3.29046, "chunk_loss": 0.031228, "F": 0.20926, "G": 0.29575, "chunks": 214.28, "bytes_per_chunk": 4.779, "boundary_entropy": 0.42428, "router_noise": 0.0, "grad_norm": 1.7232, "lr": 0.0003, "router_trainable": true, "sec_per_step": 0.17, "alpha": 0.03, "t": 62.1}
{"event": "train", "step": 350, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 2.20531, "bpb": 3.18158, "chunk_loss": 0.030876, "F": 0.19885, "G": 0.286, "chunks": 203.62, "bytes_per_chunk": 5.029, "boundary_entropy": 0.40214, "router_noise": 0.0, "grad_norm": 1.6537, "lr": 0.0002999, "router_trainable": true, "sec_per_step": 0.148, "alpha": 0.03, "t": 69.5}
{"event": "probe", "step": 400, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3462.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.198931, "cps": 27.046875, "mean_p": 0.257502, "boundary_entropy": 0.335684, "bpb": 2.906383}, "tr": {"chunks": 3766.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.208032, "cps": 29.421875, "mean_p": 0.295563, "boundary_entropy": 0.43302, "bpb": 3.194991}, "fi": {"chunks": 4053.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.215459, "cps": 31.664062, "mean_p": 0.303119, "boundary_entropy": 0.436812, "bpb": 3.381551}}, "bpb": {"en": 2.90638, "tr": 3.19499, "fi": 3.38155}, "gini_cpc": 0.017703, "gini_cps": 0.034926, "bpic": 5.028987, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.045747466045202, "fi": 1.0830830941039238}, "t": 77.9}
{"event": "eval", "step": 400, "phase": 1, "bpb": {"en": 2.91941, "tr": 3.20261, "fi": 3.37464}, "bytes_per_chunk": {"en": 4.722, "tr": 5.434, "fi": 5.006}, "bpb_mean": 3.16555, "frozen_inventory": false, "t": 79.1}
{"event": "train", "step": 400, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 2.04902, "bpb": 2.95611, "chunk_loss": 0.031066, "F": 0.21124, "G": 0.27259, "chunks": 216.31, "bytes_per_chunk": 4.734, "boundary_entropy": 0.34761, "router_noise": 0.0, "grad_norm": 1.2109, "lr": 0.0002998, "router_trainable": true, "sec_per_step": 0.201, "alpha": 0.03, "t": 79.5}
{"event": "train", "step": 450, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 2.02213, "bpb": 2.91732, "chunk_loss": 0.030916, "F": 0.20477, "G": 0.27038, "chunks": 209.69, "bytes_per_chunk": 4.883, "boundary_entropy": 0.34857, "router_noise": 0.0, "grad_norm": 1.2997, "lr": 0.0002995, "router_trainable": true, "sec_per_step": 0.143, "alpha": 0.03, "t": 86.7}
{"event": "train", "step": 500, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.99448, "bpb": 2.87743, "chunk_loss": 0.030855, "F": 0.20358, "G": 0.26603, "chunks": 208.47, "bytes_per_chunk": 4.912, "boundary_entropy": 0.32742, "router_noise": 0.0, "grad_norm": 1.2218, "lr": 0.0002992, "router_trainable": true, "sec_per_step": 0.137, "alpha": 0.03, "t": 93.6}
{"event": "train", "step": 550, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.97932, "bpb": 2.85556, "chunk_loss": 0.030985, "F": 0.20975, "G": 0.26699, "chunks": 214.78, "bytes_per_chunk": 4.768, "boundary_entropy": 0.31961, "router_noise": 0.0, "grad_norm": 1.2317, "lr": 0.0002987, "router_trainable": true, "sec_per_step": 0.151, "alpha": 0.03, "t": 101.1}
{"event": "probe", "step": 600, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3475.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.199678, "cps": 27.148438, "mean_p": 0.251073, "boundary_entropy": 0.294434, "bpb": 2.60692}, "tr": {"chunks": 3470.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.191681, "cps": 27.109375, "mean_p": 0.265268, "boundary_entropy": 0.364845, "bpb": 2.938372}, "fi": {"chunks": 4365.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.232045, "cps": 34.101562, "mean_p": 0.281307, "boundary_entropy": 0.34913, "bpb": 3.153509}}, "bpb": {"en": 2.60692, "tr": 2.93837, "fi": 3.15351}, "gini_cpc": 0.043165, "gini_cps": 0.052756, "bpic": 5.016092, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9599491637977249, "fi": 1.1620951157844657}, "t": 109.3}
{"event": "eval", "step": 600, "phase": 1, "bpb": {"en": 2.61777, "tr": 2.94764, "fi": 3.15469}, "bytes_per_chunk": {"en": 4.808, "tr": 5.848, "fi": 4.561}, "bpb_mean": 2.9067, "frozen_inventory": false, "t": 110.6}
{"event": "train", "step": 600, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.88237, "bpb": 2.71569, "chunk_loss": 0.030771, "F": 0.20273, "G": 0.26105, "chunks": 207.59, "bytes_per_chunk": 4.933, "boundary_entropy": 0.30526, "router_noise": 0.0, "grad_norm": 1.1923, "lr": 0.0002982, "router_trainable": true, "sec_per_step": 0.198, "alpha": 0.03, "t": 111.0}
{"event": "train", "step": 650, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.90688, "bpb": 2.75105, "chunk_loss": 0.031153, "F": 0.2164, "G": 0.26863, "chunks": 221.59, "bytes_per_chunk": 4.621, "boundary_entropy": 0.30809, "router_noise": 0.0, "grad_norm": 1.1105, "lr": 0.0002975, "router_trainable": true, "sec_per_step": 0.151, "alpha": 0.03, "t": 118.5}
{"event": "train", "step": 700, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.86119, "bpb": 2.68513, "chunk_loss": 0.030806, "F": 0.20554, "G": 0.25778, "chunks": 210.47, "bytes_per_chunk": 4.865, "boundary_entropy": 0.28805, "router_noise": 0.0, "grad_norm": 1.3425, "lr": 0.0002967, "router_trainable": true, "sec_per_step": 0.144, "alpha": 0.03, "t": 125.7}
{"event": "train", "step": 750, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.82172, "bpb": 2.62819, "chunk_loss": 0.030529, "F": 0.19403, "G": 0.24975, "chunks": 198.69, "bytes_per_chunk": 5.154, "boundary_entropy": 0.29093, "router_noise": 0.0, "grad_norm": 1.2811, "lr": 0.0002959, "router_trainable": true, "sec_per_step": 0.143, "alpha": 0.03, "t": 132.9}
{"event": "probe", "step": 800, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3431.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.19715, "cps": 26.804688, "mean_p": 0.24028, "boundary_entropy": 0.26412, "bpb": 2.445398}, "tr": {"chunks": 3460.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.191129, "cps": 27.03125, "mean_p": 0.256195, "boundary_entropy": 0.322507, "bpb": 2.780341}, "fi": {"chunks": 4469.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.237574, "cps": 34.914062, "mean_p": 0.280991, "boundary_entropy": 0.306141, "bpb": 2.968533}}, "bpb": {"en": 2.4454, "tr": 2.78034, "fi": 2.96853}, "gini_cpc": 0.049474, "gini_cps": 0.060915, "bpic": 4.994014, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9694578899558732, "fi": 1.2050411384063235}, "t": 140.9}
{"event": "eval", "step": 800, "phase": 1, "bpb": {"en": 2.45844, "tr": 2.79191, "fi": 2.974}, "bytes_per_chunk": {"en": 4.793, "tr": 5.739, "fi": 4.542}, "bpb_mean": 2.74145, "frozen_inventory": false, "t": 142.0}
{"event": "train", "step": 800, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.81094, "bpb": 2.61263, "chunk_loss": 0.030831, "F": 0.20712, "G": 0.2549, "chunks": 212.09, "bytes_per_chunk": 4.828, "boundary_entropy": 0.2744, "router_noise": 0.0, "grad_norm": 1.0027, "lr": 0.0002949, "router_trainable": true, "sec_per_step": 0.19, "alpha": 0.03, "t": 142.4}
{"event": "train", "step": 850, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.78587, "bpb": 2.57647, "chunk_loss": 0.030635, "F": 0.20078, "G": 0.24697, "chunks": 205.59, "bytes_per_chunk": 4.981, "boundary_entropy": 0.26801, "router_noise": 0.0, "grad_norm": 1.1756, "lr": 0.0002938, "router_trainable": true, "sec_per_step": 0.151, "alpha": 0.03, "t": 149.9}
{"event": "train", "step": 900, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.62509, "bpb": 2.3445, "chunk_loss": 0.030753, "F": 0.2056, "G": 0.24934, "chunks": 210.53, "bytes_per_chunk": 4.864, "boundary_entropy": 0.26015, "router_noise": 0.0, "grad_norm": 1.0277, "lr": 0.0002927, "router_trainable": true, "sec_per_step": 0.151, "alpha": 0.03, "t": 157.5}
{"event": "train", "step": 950, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.78436, "bpb": 2.57429, "chunk_loss": 0.030752, "F": 0.20703, "G": 0.24443, "chunks": 212.0, "bytes_per_chunk": 4.83, "boundary_entropy": 0.25385, "router_noise": 0.0, "grad_norm": 1.0002, "lr": 0.0002914, "router_trainable": true, "sec_per_step": 0.143, "alpha": 0.03, "t": 164.6}
{"event": "probe", "step": 1000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3415.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.196231, "cps": 26.679688, "mean_p": 0.232151, "boundary_entropy": 0.242564, "bpb": 2.351274}, "tr": {"chunks": 3534.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.195216, "cps": 27.609375, "mean_p": 0.247568, "boundary_entropy": 0.304826, "bpb": 2.660542}, "fi": {"chunks": 4585.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.24374, "cps": 35.820312, "mean_p": 0.272112, "boundary_entropy": 0.275055, "bpb": 2.834728}}, "bpb": {"en": 2.35127, "tr": 2.66054, "fi": 2.83473}, "gini_cpc": 0.050929, "gini_cps": 0.067626, "bpic": 4.918675, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9948312199857834, "fi": 1.2421123181640474}, "t": 172.4}
{"event": "eval", "step": 1000, "phase": 1, "bpb": {"en": 2.36166, "tr": 2.67006, "fi": 2.84132}, "bytes_per_chunk": {"en": 4.816, "tr": 5.728, "fi": 4.414}, "bpb_mean": 2.62435, "frozen_inventory": false, "t": 173.6}
{"event": "train", "step": 1000, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.72316, "bpb": 2.48599, "chunk_loss": 0.031057, "F": 0.21573, "G": 0.25178, "chunks": 220.91, "bytes_per_chunk": 4.635, "boundary_entropy": 0.248, "router_noise": 0.0, "grad_norm": 1.0507, "lr": 0.0002901, "router_trainable": true, "sec_per_step": 0.193, "alpha": 0.03, "t": 174.3}
{"event": "train", "step": 1050, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.74334, "bpb": 2.51511, "chunk_loss": 0.030814, "F": 0.21115, "G": 0.2462, "chunks": 216.22, "bytes_per_chunk": 4.736, "boundary_entropy": 0.24141, "router_noise": 0.0, "grad_norm": 0.9249, "lr": 0.0002886, "router_trainable": true, "sec_per_step": 0.153, "alpha": 0.03, "t": 181.9}
{"event": "train", "step": 1100, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.73664, "bpb": 2.50544, "chunk_loss": 0.030792, "F": 0.2084, "G": 0.24529, "chunks": 213.41, "bytes_per_chunk": 4.798, "boundary_entropy": 0.24081, "router_noise": 0.0, "grad_norm": 1.0049, "lr": 0.0002871, "router_trainable": true, "sec_per_step": 0.153, "alpha": 0.03, "t": 189.6}
{"event": "train", "step": 1150, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.59401, "bpb": 2.29967, "chunk_loss": 0.030818, "F": 0.21344, "G": 0.24191, "chunks": 218.56, "bytes_per_chunk": 4.685, "boundary_entropy": 0.23244, "router_noise": 0.0, "grad_norm": 1.0171, "lr": 0.0002855, "router_trainable": true, "sec_per_step": 0.14, "alpha": 0.03, "t": 196.6}
{"event": "probe", "step": 1200, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3475.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.199678, "cps": 27.148438, "mean_p": 0.22511, "boundary_entropy": 0.219021, "bpb": 2.284572}, "tr": {"chunks": 3731.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.206098, "cps": 29.148438, "mean_p": 0.23931, "boundary_entropy": 0.265483, "bpb": 2.570854}, "fi": {"chunks": 4663.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.247887, "cps": 36.429688, "mean_p": 0.26802, "boundary_entropy": 0.235907, "bpb": 2.743649}}, "bpb": {"en": 2.28457, "tr": 2.57085, "fi": 2.74365}, "gini_cpc": 0.049168, "gini_cps": 0.066728, "bpic": 4.779847, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0321528328902916, "fi": 1.2414317353729583}, "t": 204.9}
{"event": "eval", "step": 1200, "phase": 1, "bpb": {"en": 2.29438, "tr": 2.58017, "fi": 2.75228}, "bytes_per_chunk": {"en": 4.769, "tr": 5.387, "fi": 4.325}, "bpb_mean": 2.54228, "frozen_inventory": false, "t": 206.2}
{"event": "train", "step": 1200, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.63318, "bpb": 2.35617, "chunk_loss": 0.030784, "F": 0.21207, "G": 0.24104, "chunks": 217.16, "bytes_per_chunk": 4.715, "boundary_entropy": 0.22626, "router_noise": 0.0, "grad_norm": 0.8578, "lr": 0.0002837, "router_trainable": true, "sec_per_step": 0.199, "alpha": 0.03, "t": 206.5}
{"event": "train", "step": 1250, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.6822, "bpb": 2.42691, "chunk_loss": 0.030825, "F": 0.21255, "G": 0.24282, "chunks": 217.66, "bytes_per_chunk": 4.705, "boundary_entropy": 0.22496, "router_noise": 0.0, "grad_norm": 0.8754, "lr": 0.0002819, "router_trainable": true, "sec_per_step": 0.146, "alpha": 0.03, "t": 213.8}
{"event": "train", "step": 1300, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.62591, "bpb": 2.34569, "chunk_loss": 0.030649, "F": 0.20526, "G": 0.23572, "chunks": 210.19, "bytes_per_chunk": 4.872, "boundary_entropy": 0.22593, "router_noise": 0.0, "grad_norm": 0.8102, "lr": 0.00028, "router_trainable": true, "sec_per_step": 0.148, "alpha": 0.03, "t": 221.2}
{"event": "train", "step": 1350, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.61692, "bpb": 2.33272, "chunk_loss": 0.030838, "F": 0.21204, "G": 0.24008, "chunks": 217.12, "bytes_per_chunk": 4.716, "boundary_entropy": 0.21861, "router_noise": 0.0, "grad_norm": 0.9334, "lr": 0.000278, "router_trainable": true, "sec_per_step": 0.15, "alpha": 0.03, "t": 228.8}
{"event": "probe", "step": 1400, "phase": 1, "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": 0.216142, "boundary_entropy": 0.210127, "bpb": 2.233986}, "tr": {"chunks": 3726.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.205822, "cps": 29.109375, "mean_p": 0.234751, "boundary_entropy": 0.251461, "bpb": 2.505464}, "fi": {"chunks": 4617.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.245441, "cps": 36.070312, "mean_p": 0.265016, "boundary_entropy": 0.223098, "bpb": 2.663768}}, "bpb": {"en": 2.23399, "tr": 2.50546, "fi": 2.66377}, "gini_cpc": 0.061525, "gini_cps": 0.078804, "bpic": 4.894909, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.1031488858683451, "fi": 1.3154968808835894}, "t": 236.8}
{"event": "eval", "step": 1400, "phase": 1, "bpb": {"en": 2.24481, "tr": 2.51542, "fi": 2.67256}, "bytes_per_chunk": {"en": 5.096, "tr": 5.445, "fi": 4.38}, "bpb_mean": 2.4776, "frozen_inventory": false, "t": 237.9}
{"event": "train", "step": 1400, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.69958, "bpb": 2.45197, "chunk_loss": 0.030597, "F": 0.20096, "G": 0.23386, "chunks": 205.78, "bytes_per_chunk": 4.976, "boundary_entropy": 0.21669, "router_noise": 0.0, "grad_norm": 0.9321, "lr": 0.0002759, "router_trainable": true, "sec_per_step": 0.19, "alpha": 0.03, "t": 238.3}
{"event": "train", "step": 1450, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.65465, "bpb": 2.38715, "chunk_loss": 0.030661, "F": 0.2056, "G": 0.23585, "chunks": 210.53, "bytes_per_chunk": 4.864, "boundary_entropy": 0.21531, "router_noise": 0.0, "grad_norm": 0.8415, "lr": 0.0002738, "router_trainable": true, "sec_per_step": 0.153, "alpha": 0.03, "t": 245.9}
{"event": "train", "step": 1500, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.60998, "bpb": 2.3227, "chunk_loss": 0.030492, "F": 0.19769, "G": 0.22814, "chunks": 202.44, "bytes_per_chunk": 5.058, "boundary_entropy": 0.21408, "router_noise": 0.0, "grad_norm": 0.8744, "lr": 0.0002715, "router_trainable": true, "sec_per_step": 0.143, "alpha": 0.03, "t": 253.1}
{"event": "train", "step": 1550, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.57501, "bpb": 2.27226, "chunk_loss": 0.030566, "F": 0.20224, "G": 0.22717, "chunks": 207.09, "bytes_per_chunk": 4.945, "boundary_entropy": 0.20248, "router_noise": 0.0, "grad_norm": 0.7256, "lr": 0.0002692, "router_trainable": true, "sec_per_step": 0.15, "alpha": 0.03, "t": 260.6}
{"event": "probe", "step": 1600, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3296.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.189393, "cps": 25.75, "mean_p": 0.215865, "boundary_entropy": 0.200686, "bpb": 2.195195}, "tr": {"chunks": 3580.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.197757, "cps": 27.96875, "mean_p": 0.224983, "boundary_entropy": 0.233256, "bpb": 2.443705}, "fi": {"chunks": 4751.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.252565, "cps": 37.117188, "mean_p": 0.267677, "boundary_entropy": 0.206581, "bpb": 2.600278}}, "bpb": {"en": 2.19519, "tr": 2.4437, "fi": 2.60028}, "gini_cpc": 0.065834, "gini_cps": 0.083427, "bpic": 4.879333, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0441656266809825, "fi": 1.3335523349795848}, "t": 268.5}
{"event": "eval", "step": 1600, "phase": 1, "bpb": {"en": 2.20568, "tr": 2.45276, "fi": 2.60731}, "bytes_per_chunk": {"en": 5.0, "tr": 5.524, "fi": 4.268}, "bpb_mean": 2.42192, "frozen_inventory": false, "t": 269.8}
{"event": "train", "step": 1600, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.58263, "bpb": 2.28325, "chunk_loss": 0.030621, "F": 0.20334, "G": 0.23011, "chunks": 208.22, "bytes_per_chunk": 4.918, "boundary_entropy": 0.20425, "router_noise": 0.0, "grad_norm": 0.9365, "lr": 0.0002668, "router_trainable": true, "sec_per_step": 0.191, "alpha": 0.03, "t": 270.1}
{"event": "train", "step": 1650, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.60336, "bpb": 2.31315, "chunk_loss": 0.030491, "F": 0.19894, "G": 0.2228, "chunks": 203.72, "bytes_per_chunk": 5.027, "boundary_entropy": 0.19384, "router_noise": 0.0, "grad_norm": 0.807, "lr": 0.0002643, "router_trainable": true, "sec_per_step": 0.156, "alpha": 0.03, "t": 277.9}
{"event": "train", "step": 1700, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.61756, "bpb": 2.33365, "chunk_loss": 0.030511, "F": 0.20044, "G": 0.22484, "chunks": 205.25, "bytes_per_chunk": 4.989, "boundary_entropy": 0.19793, "router_noise": 0.0, "grad_norm": 0.7788, "lr": 0.0002618, "router_trainable": true, "sec_per_step": 0.152, "alpha": 0.03, "t": 285.5}
{"event": "train", "step": 1750, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.57091, "bpb": 2.26635, "chunk_loss": 0.030382, "F": 0.19403, "G": 0.21969, "chunks": 198.69, "bytes_per_chunk": 5.154, "boundary_entropy": 0.19211, "router_noise": 0.0, "grad_norm": 0.7656, "lr": 0.0002591, "router_trainable": true, "sec_per_step": 0.149, "alpha": 0.03, "t": 293.0}
{"event": "probe", "step": 1800, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3206.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.184221, "cps": 25.046875, "mean_p": 0.211991, "boundary_entropy": 0.194889, "bpb": 2.160036}, "tr": {"chunks": 3742.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.206706, "cps": 29.234375, "mean_p": 0.225818, "boundary_entropy": 0.229072, "bpb": 2.407169}, "fi": {"chunks": 4797.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.25501, "cps": 37.476562, "mean_p": 0.266954, "boundary_entropy": 0.192812, "bpb": 2.54011}}, "bpb": {"en": 2.16004, "tr": 2.40717, "fi": 2.54011}, "gini_cpc": 0.073061, "gini_cps": 0.090308, "bpic": 4.830311, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.1220541953924221, "fi": 1.3842624467513185}, "t": 300.7}
{"event": "eval", "step": 1800, "phase": 1, "bpb": {"en": 2.17179, "tr": 2.41529, "fi": 2.54722}, "bytes_per_chunk": {"en": 5.154, "tr": 5.344, "fi": 4.239}, "bpb_mean": 2.3781, "frozen_inventory": false, "t": 302.0}
{"event": "train", "step": 1800, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.54063, "bpb": 2.22265, "chunk_loss": 0.030394, "F": 0.19376, "G": 0.22004, "chunks": 198.41, "bytes_per_chunk": 5.161, "boundary_entropy": 0.19539, "router_noise": 0.0, "grad_norm": 0.8338, "lr": 0.0002564, "router_trainable": true, "sec_per_step": 0.19, "alpha": 0.03, "t": 302.5}
{"event": "train", "step": 1850, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.57113, "bpb": 2.26666, "chunk_loss": 0.030466, "F": 0.19974, "G": 0.22122, "chunks": 204.53, "bytes_per_chunk": 5.007, "boundary_entropy": 0.18669, "router_noise": 0.0, "grad_norm": 0.8835, "lr": 0.0002537, "router_trainable": true, "sec_per_step": 0.154, "alpha": 0.03, "t": 310.2}
{"event": "train", "step": 1900, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.50296, "bpb": 2.16831, "chunk_loss": 0.030501, "F": 0.20068, "G": 0.22273, "chunks": 205.5, "bytes_per_chunk": 4.983, "boundary_entropy": 0.1805, "router_noise": 0.0, "grad_norm": 0.7368, "lr": 0.0002508, "router_trainable": true, "sec_per_step": 0.154, "alpha": 0.03, "t": 317.8}
{"event": "train", "step": 1950, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.58087, "bpb": 2.28071, "chunk_loss": 0.030639, "F": 0.2059, "G": 0.22733, "chunks": 210.84, "bytes_per_chunk": 4.857, "boundary_entropy": 0.17768, "router_noise": 0.0, "grad_norm": 0.7724, "lr": 0.0002479, "router_trainable": true, "sec_per_step": 0.153, "alpha": 0.03, "t": 325.5}
{"event": "probe", "step": 2000, "phase": 1, "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": 0.20836, "boundary_entropy": 0.178286, "bpb": 2.134764}, "tr": {"chunks": 3545.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.195824, "cps": 27.695312, "mean_p": 0.21647, "boundary_entropy": 0.212829, "bpb": 2.346676}, "fi": {"chunks": 4749.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.252459, "cps": 37.101562, "mean_p": 0.262057, "boundary_entropy": 0.179947, "bpb": 2.495846}}, "bpb": {"en": 2.13476, "tr": 2.34668, "fi": 2.49585}, "gini_cpc": 0.071253, "gini_cps": 0.088792, "bpic": 4.928931, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0596776344483185, "fi": 1.3661499364852416}, "t": 333.2}
{"event": "eval", "step": 2000, "phase": 1, "bpb": {"en": 2.14656, "tr": 2.35656, "fi": 2.50635}, "bytes_per_chunk": {"en": 5.088, "tr": 5.629, "fi": 4.276}, "bpb_mean": 2.33649, "frozen_inventory": false, "t": 334.4}
{"event": "train", "step": 2000, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.55299, "bpb": 2.24049, "chunk_loss": 0.030544, "F": 0.20474, "G": 0.22423, "chunks": 209.66, "bytes_per_chunk": 4.884, "boundary_entropy": 0.17778, "router_noise": 0.0, "grad_norm": 0.9019, "lr": 0.000245, "router_trainable": true, "sec_per_step": 0.189, "alpha": 0.03, "t": 335.0}
{"event": "train", "step": 2050, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.56079, "bpb": 2.25174, "chunk_loss": 0.030373, "F": 0.19373, "G": 0.21798, "chunks": 198.38, "bytes_per_chunk": 5.162, "boundary_entropy": 0.18038, "router_noise": 0.0, "grad_norm": 0.812, "lr": 0.0002419, "router_trainable": true, "sec_per_step": 0.147, "alpha": 0.03, "t": 342.3}
{"event": "train", "step": 2100, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.47916, "bpb": 2.13397, "chunk_loss": 0.030354, "F": 0.19421, "G": 0.21395, "chunks": 198.88, "bytes_per_chunk": 5.149, "boundary_entropy": 0.16873, "router_noise": 0.0, "grad_norm": 0.7176, "lr": 0.0002388, "router_trainable": true, "sec_per_step": 0.154, "alpha": 0.03, "t": 350.0}
{"event": "train", "step": 2150, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.49605, "bpb": 2.15834, "chunk_loss": 0.030577, "F": 0.20657, "G": 0.22286, "chunks": 211.53, "bytes_per_chunk": 4.841, "boundary_entropy": 0.16108, "router_noise": 0.0, "grad_norm": 0.7884, "lr": 0.0002357, "router_trainable": true, "sec_per_step": 0.151, "alpha": 0.03, "t": 357.5}
{"event": "probe", "step": 2200, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3194.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.183532, "cps": 24.953125, "mean_p": 0.202331, "boundary_entropy": 0.161806, "bpb": 2.10197}, "tr": {"chunks": 3523.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.194609, "cps": 27.523438, "mean_p": 0.211384, "boundary_entropy": 0.194694, "bpb": 2.310883}, "fi": {"chunks": 4782.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.254213, "cps": 37.359375, "mean_p": 0.25923, "boundary_entropy": 0.160965, "bpb": 2.468604}}, "bpb": {"en": 2.10197, "tr": 2.31088, "fi": 2.4686}, "gini_cpc": 0.074517, "gini_cps": 0.092066, "bpic": 4.933646, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0603550282144982, "fi": 1.385118394368634}, "t": 365.9}
{"event": "eval", "step": 2200, "phase": 1, "bpb": {"en": 2.11402, "tr": 2.31876, "fi": 2.47713}, "bytes_per_chunk": {"en": 5.137, "tr": 5.621, "fi": 4.248}, "bpb_mean": 2.3033, "frozen_inventory": false, "t": 366.8}
{"event": "train", "step": 2200, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.53247, "bpb": 2.21089, "chunk_loss": 0.030602, "F": 0.20312, "G": 0.21988, "chunks": 208.0, "bytes_per_chunk": 4.923, "boundary_entropy": 0.16081, "router_noise": 0.0, "grad_norm": 0.7146, "lr": 0.0002325, "router_trainable": true, "sec_per_step": 0.194, "alpha": 0.03, "t": 367.2}
{"event": "train", "step": 2250, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.51059, "bpb": 2.17931, "chunk_loss": 0.030411, "F": 0.19812, "G": 0.21532, "chunks": 202.88, "bytes_per_chunk": 5.047, "boundary_entropy": 0.15603, "router_noise": 0.0, "grad_norm": 0.8695, "lr": 0.0002293, "router_trainable": true, "sec_per_step": 0.146, "alpha": 0.03, "t": 374.5}
{"event": "train", "step": 2300, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.54518, "bpb": 2.22923, "chunk_loss": 0.030449, "F": 0.19867, "G": 0.21591, "chunks": 203.44, "bytes_per_chunk": 5.033, "boundary_entropy": 0.16127, "router_noise": 0.0, "grad_norm": 0.7161, "lr": 0.000226, "router_trainable": true, "sec_per_step": 0.15, "alpha": 0.03, "t": 382.0}
{"event": "train", "step": 2350, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.46621, "bpb": 2.11529, "chunk_loss": 0.030463, "F": 0.20013, "G": 0.21557, "chunks": 204.94, "bytes_per_chunk": 4.997, "boundary_entropy": 0.15356, "router_noise": 0.0, "grad_norm": 0.7287, "lr": 0.0002226, "router_trainable": true, "sec_per_step": 0.155, "alpha": 0.03, "t": 389.8}
{"event": "probe", "step": 2400, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3212.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.184566, "cps": 25.09375, "mean_p": 0.202707, "boundary_entropy": 0.156128, "bpb": 2.07728}, "tr": {"chunks": 3522.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.194553, "cps": 27.515625, "mean_p": 0.207156, "boundary_entropy": 0.190756, "bpb": 2.27782}, "fi": {"chunks": 4736.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.251768, "cps": 37.0, "mean_p": 0.262761, "boundary_entropy": 0.166291, "bpb": 2.418829}}, "bpb": {"en": 2.07728, "tr": 2.27782, "fi": 2.41883}, "gini_cpc": 0.071013, "gini_cps": 0.088579, "bpic": 4.94612, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0541135204673906, "fi": 1.3641068628335622}, "t": 398.1}
{"event": "eval", "step": 2400, "phase": 1, "bpb": {"en": 2.08914, "tr": 2.28457, "fi": 2.42885}, "bytes_per_chunk": {"en": 5.095, "tr": 5.609, "fi": 4.273}, "bpb_mean": 2.26752, "frozen_inventory": false, "t": 399.3}
{"event": "phase_transition", "step": 2400, "phase": 2, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0, "t": 399.5}
{"event": "train", "step": 2400, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.46914, "bpb": 2.11952, "chunk_loss": 0.03035, "F": 0.19504, "G": 0.21254, "chunks": 199.72, "bytes_per_chunk": 5.127, "boundary_entropy": 0.16182, "router_noise": 0.0, "grad_norm": 0.6792, "lr": 0.0002192, "router_trainable": true, "sec_per_step": 0.198, "alpha": 0.03, "t": 399.7}
{"event": "train", "step": 2450, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.48225, "bpb": 2.13843, "chunk_loss": 0.03064, "F": 0.20969, "G": 0.22475, "chunks": 214.72, "bytes_per_chunk": 4.769, "boundary_entropy": 0.15558, "router_noise": 0.0, "grad_norm": 0.8197, "lr": 0.0002158, "router_trainable": true, "sec_per_step": 0.148, "alpha": 0.03, "t": 407.1}
{"event": "train", "step": 2500, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.5763, "bpb": 2.27412, "chunk_loss": 0.030506, "F": 0.20334, "G": 0.2179, "chunks": 208.22, "bytes_per_chunk": 4.918, "boundary_entropy": 0.15472, "router_noise": 0.0, "grad_norm": 0.7392, "lr": 0.0002123, "router_trainable": true, "sec_per_step": 0.152, "alpha": 0.03, "t": 414.7}
{"event": "train", "step": 2550, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.49939, "bpb": 2.16316, "chunk_loss": 0.030396, "F": 0.19925, "G": 0.21622, "chunks": 204.03, "bytes_per_chunk": 5.019, "boundary_entropy": 0.15341, "router_noise": 0.0, "grad_norm": 0.7196, "lr": 0.0002088, "router_trainable": true, "sec_per_step": 0.142, "alpha": 0.03, "t": 421.8}
{"event": "probe", "step": 2600, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3210.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.184451, "cps": 25.078125, "mean_p": 0.19933, "boundary_entropy": 0.144993, "bpb": 2.060607}, "tr": {"chunks": 3525.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.194719, "cps": 27.539062, "mean_p": 0.205839, "boundary_entropy": 0.181865, "bpb": 2.252641}, "fi": {"chunks": 4770.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.253575, "cps": 37.265625, "mean_p": 0.259067, "boundary_entropy": 0.152333, "bpb": 2.377115}}, "bpb": {"en": 2.06061, "tr": 2.25264, "fi": 2.37712}, "gini_cpc": 0.07283, "gini_cps": 0.090395, "bpic": 4.931073, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0556687304887247, "fi": 1.3747558721110187}, "t": 430.1}
{"event": "eval", "step": 2600, "phase": 2, "bpb": {"en": 2.07026, "tr": 2.26137, "fi": 2.38691}, "bytes_per_chunk": {"en": 5.106, "tr": 5.639, "fi": 4.226}, "bpb_mean": 2.23951, "frozen_inventory": false, "t": 431.3}
{"event": "train", "step": 2600, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.50931, "bpb": 2.17748, "chunk_loss": 0.030543, "F": 0.20343, "G": 0.21852, "chunks": 208.31, "bytes_per_chunk": 4.916, "boundary_entropy": 0.14672, "router_noise": 0.0, "grad_norm": 0.7053, "lr": 0.0002053, "router_trainable": true, "sec_per_step": 0.202, "alpha": 0.03, "t": 431.9}
{"event": "train", "step": 2650, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.44456, "bpb": 2.08406, "chunk_loss": 0.030314, "F": 0.1933, "G": 0.20821, "chunks": 197.94, "bytes_per_chunk": 5.173, "boundary_entropy": 0.15054, "router_noise": 0.0, "grad_norm": 0.7094, "lr": 0.0002017, "router_trainable": true, "sec_per_step": 0.142, "alpha": 0.03, "t": 439.0}
{"event": "train", "step": 2700, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.53089, "bpb": 2.20861, "chunk_loss": 0.030461, "F": 0.19894, "G": 0.21566, "chunks": 203.72, "bytes_per_chunk": 5.027, "boundary_entropy": 0.15515, "router_noise": 0.0, "grad_norm": 0.8205, "lr": 0.0001981, "router_trainable": true, "sec_per_step": 0.153, "alpha": 0.03, "t": 446.6}
{"event": "train", "step": 2750, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.46369, "bpb": 2.11166, "chunk_loss": 0.030428, "F": 0.20093, "G": 0.21573, "chunks": 205.75, "bytes_per_chunk": 4.977, "boundary_entropy": 0.14818, "router_noise": 0.0, "grad_norm": 0.7445, "lr": 0.0001945, "router_trainable": true, "sec_per_step": 0.154, "alpha": 0.03, "t": 454.3}
{"event": "probe", "step": 2800, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3233.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.185773, "cps": 25.257812, "mean_p": 0.200475, "boundary_entropy": 0.144342, "bpb": 2.036584}, "tr": {"chunks": 3523.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.194609, "cps": 27.523438, "mean_p": 0.205633, "boundary_entropy": 0.17933, "bpb": 2.242403}, "fi": {"chunks": 4760.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.253043, "cps": 37.1875, "mean_p": 0.255816, "boundary_entropy": 0.140474, "bpb": 2.344065}}, "bpb": {"en": 2.03658, "tr": 2.2424, "fi": 2.34407}, "gini_cpc": 0.070801, "gini_cps": 0.088399, "bpic": 4.926363, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0475638602279949, "fi": 1.3621140883685423}, "t": 462.1}
{"event": "eval", "step": 2800, "phase": 2, "bpb": {"en": 2.05001, "tr": 2.24964, "fi": 2.35406}, "bytes_per_chunk": {"en": 5.077, "tr": 5.634, "fi": 4.234}, "bpb_mean": 2.2179, "frozen_inventory": false, "t": 463.4}
{"event": "train", "step": 2800, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.45339, "bpb": 2.09679, "chunk_loss": 0.030445, "F": 0.2019, "G": 0.21488, "chunks": 206.75, "bytes_per_chunk": 4.953, "boundary_entropy": 0.14357, "router_noise": 0.0, "grad_norm": 0.7076, "lr": 0.0001909, "router_trainable": true, "sec_per_step": 0.19, "alpha": 0.03, "t": 463.8}
{"event": "train", "step": 2850, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.56074, "bpb": 2.25167, "chunk_loss": 0.030638, "F": 0.20792, "G": 0.22066, "chunks": 212.91, "bytes_per_chunk": 4.81, "boundary_entropy": 0.1434, "router_noise": 0.0, "grad_norm": 0.751, "lr": 0.0001872, "router_trainable": true, "sec_per_step": 0.144, "alpha": 0.03, "t": 471.0}
{"event": "train", "step": 2900, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.50257, "bpb": 2.16775, "chunk_loss": 0.030642, "F": 0.20898, "G": 0.22133, "chunks": 214.0, "bytes_per_chunk": 4.785, "boundary_entropy": 0.14432, "router_noise": 0.0, "grad_norm": 0.682, "lr": 0.0001835, "router_trainable": true, "sec_per_step": 0.152, "alpha": 0.03, "t": 478.6}
{"event": "train", "step": 2950, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.43411, "bpb": 2.06899, "chunk_loss": 0.030398, "F": 0.19778, "G": 0.21115, "chunks": 202.53, "bytes_per_chunk": 5.056, "boundary_entropy": 0.1408, "router_noise": 0.0, "grad_norm": 0.6659, "lr": 0.0001799, "router_trainable": true, "sec_per_step": 0.153, "alpha": 0.03, "t": 486.3}
{"event": "probe", "step": 3000, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3213.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.184623, "cps": 25.101562, "mean_p": 0.199599, "boundary_entropy": 0.136919, "bpb": 2.023392}, "tr": {"chunks": 3548.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.19599, "cps": 27.71875, "mean_p": 0.203741, "boundary_entropy": 0.173809, "bpb": 2.206409}, "fi": {"chunks": 4736.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.251768, "cps": 37.0, "mean_p": 0.25531, "boundary_entropy": 0.139749, "bpb": 2.330668}}, "bpb": {"en": 2.02339, "tr": 2.20641, "fi": 2.33067}, "gini_cpc": 0.070785, "gini_cps": 0.088313, "bpic": 4.934505, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0615646652702584, "fi": 1.3636823042083417}, "t": 494.6}
{"event": "eval", "step": 3000, "phase": 2, "bpb": {"en": 2.03462, "tr": 2.21401, "fi": 2.33734}, "bytes_per_chunk": {"en": 5.075, "tr": 5.594, "fi": 4.254}, "bpb_mean": 2.19533, "frozen_inventory": false, "t": 495.7}
{"event": "train", "step": 3000, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.46357, "bpb": 2.11149, "chunk_loss": 0.030379, "F": 0.19919, "G": 0.21234, "chunks": 203.97, "bytes_per_chunk": 5.02, "boundary_entropy": 0.1381, "router_noise": 0.0, "grad_norm": 0.7151, "lr": 0.0001761, "router_trainable": true, "sec_per_step": 0.2, "alpha": 0.03, "t": 496.3}
{"event": "train", "step": 3050, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.45803, "bpb": 2.1035, "chunk_loss": 0.030423, "F": 0.20139, "G": 0.21508, "chunks": 206.22, "bytes_per_chunk": 4.966, "boundary_entropy": 0.14247, "router_noise": 0.0, "grad_norm": 0.7514, "lr": 0.0001724, "router_trainable": true, "sec_per_step": 0.141, "alpha": 0.03, "t": 503.3}
{"event": "train", "step": 3100, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.54683, "bpb": 2.23161, "chunk_loss": 0.030469, "F": 0.19821, "G": 0.21228, "chunks": 202.97, "bytes_per_chunk": 5.045, "boundary_entropy": 0.14183, "router_noise": 0.0, "grad_norm": 0.7539, "lr": 0.0001687, "router_trainable": true, "sec_per_step": 0.152, "alpha": 0.03, "t": 511.0}
{"event": "train", "step": 3150, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.46741, "bpb": 2.11702, "chunk_loss": 0.030456, "F": 0.20041, "G": 0.2138, "chunks": 205.22, "bytes_per_chunk": 4.99, "boundary_entropy": 0.13813, "router_noise": 0.0, "grad_norm": 0.7102, "lr": 0.000165, "router_trainable": true, "sec_per_step": 0.153, "alpha": 0.03, "t": 518.6}
{"event": "probe", "step": 3200, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3236.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.185945, "cps": 25.28125, "mean_p": 0.199194, "boundary_entropy": 0.129496, "bpb": 2.003565}, "tr": {"chunks": 3543.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.195713, "cps": 27.679688, "mean_p": 0.202723, "boundary_entropy": 0.174303, "bpb": 2.185613}, "fi": {"chunks": 4764.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.253256, "cps": 37.21875, "mean_p": 0.257213, "boundary_entropy": 0.142148, "bpb": 2.298122}}, "bpb": {"en": 2.00356, "tr": 2.18561, "fi": 2.29812}, "gini_cpc": 0.070677, "gini_cps": 0.08825, "bpic": 4.91484, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.052534180356642, "fi": 1.3619948851692973}, "t": 526.9}
{"event": "eval", "step": 3200, "phase": 2, "bpb": {"en": 2.01657, "tr": 2.19454, "fi": 2.30596}, "bytes_per_chunk": {"en": 5.055, "tr": 5.604, "fi": 4.238}, "bpb_mean": 2.17236, "frozen_inventory": false, "t": 528.2}
{"event": "train", "step": 3200, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.43672, "bpb": 2.07276, "chunk_loss": 0.030606, "F": 0.20779, "G": 0.21978, "chunks": 212.78, "bytes_per_chunk": 4.812, "boundary_entropy": 0.13437, "router_noise": 0.0, "grad_norm": 0.7462, "lr": 0.0001613, "router_trainable": true, "sec_per_step": 0.198, "alpha": 0.03, "t": 528.5}
{"event": "train", "step": 3250, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.40883, "bpb": 2.03251, "chunk_loss": 0.030322, "F": 0.19571, "G": 0.20675, "chunks": 200.41, "bytes_per_chunk": 5.11, "boundary_entropy": 0.13061, "router_noise": 0.0, "grad_norm": 0.6613, "lr": 0.0001576, "router_trainable": true, "sec_per_step": 0.142, "alpha": 0.03, "t": 535.6}
{"event": "train", "step": 3300, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.45243, "bpb": 2.09541, "chunk_loss": 0.03049, "F": 0.20352, "G": 0.21478, "chunks": 208.41, "bytes_per_chunk": 4.913, "boundary_entropy": 0.1307, "router_noise": 0.0, "grad_norm": 0.6568, "lr": 0.0001539, "router_trainable": true, "sec_per_step": 0.133, "alpha": 0.03, "t": 542.3}
{"event": "train", "step": 3350, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.48621, "bpb": 2.14415, "chunk_loss": 0.030515, "F": 0.20474, "G": 0.21733, "chunks": 209.66, "bytes_per_chunk": 4.884, "boundary_entropy": 0.14146, "router_noise": 0.0, "grad_norm": 0.7817, "lr": 0.0001501, "router_trainable": true, "sec_per_step": 0.152, "alpha": 0.03, "t": 549.9}
{"event": "probe", "step": 3400, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3194.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.183532, "cps": 24.953125, "mean_p": 0.196584, "boundary_entropy": 0.128804, "bpb": 1.990802}, "tr": {"chunks": 3522.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.194553, "cps": 27.515625, "mean_p": 0.19976, "boundary_entropy": 0.168027, "bpb": 2.165297}, "fi": {"chunks": 4758.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.252937, "cps": 37.171875, "mean_p": 0.254558, "boundary_entropy": 0.133017, "bpb": 2.273907}}, "bpb": {"en": 1.9908, "tr": 2.1653, "fi": 2.27391}, "gini_cpc": 0.073326, "gini_cps": 0.090872, "bpic": 4.944396, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0600540475082212, "fi": 1.378166733669168}, "t": 558.3}
{"event": "eval", "step": 3400, "phase": 2, "bpb": {"en": 2.00244, "tr": 2.17415, "fi": 2.28101}, "bytes_per_chunk": {"en": 5.127, "tr": 5.586, "fi": 4.246}, "bpb_mean": 2.15253, "frozen_inventory": false, "t": 559.6}
{"event": "train", "step": 3400, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.40459, "bpb": 2.0264, "chunk_loss": 0.030341, "F": 0.19345, "G": 0.20475, "chunks": 198.09, "bytes_per_chunk": 5.169, "boundary_entropy": 0.12882, "router_noise": 0.0, "grad_norm": 0.692, "lr": 0.0001465, "router_trainable": true, "sec_per_step": 0.202, "alpha": 0.03, "t": 560.0}
{"event": "train", "step": 3450, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.39141, "bpb": 2.00737, "chunk_loss": 0.030295, "F": 0.19476, "G": 0.20532, "chunks": 199.44, "bytes_per_chunk": 5.134, "boundary_entropy": 0.12436, "router_noise": 0.0, "grad_norm": 0.6324, "lr": 0.0001428, "router_trainable": true, "sec_per_step": 0.154, "alpha": 0.03, "t": 567.7}
{"event": "train", "step": 3500, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.48124, "bpb": 2.13698, "chunk_loss": 0.030533, "F": 0.20187, "G": 0.21365, "chunks": 206.72, "bytes_per_chunk": 4.954, "boundary_entropy": 0.13045, "router_noise": 0.0, "grad_norm": 0.6845, "lr": 0.0001391, "router_trainable": true, "sec_per_step": 0.144, "alpha": 0.03, "t": 574.9}
{"event": "train", "step": 3550, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.48337, "bpb": 2.14005, "chunk_loss": 0.030697, "F": 0.21198, "G": 0.22183, "chunks": 217.06, "bytes_per_chunk": 4.718, "boundary_entropy": 0.1314, "router_noise": 0.0, "grad_norm": 0.7511, "lr": 0.0001355, "router_trainable": true, "sec_per_step": 0.144, "alpha": 0.03, "t": 582.1}
{"event": "probe", "step": 3600, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3228.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.185485, "cps": 25.21875, "mean_p": 0.197069, "boundary_entropy": 0.123863, "bpb": 1.97269}, "tr": {"chunks": 3562.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.196763, "cps": 27.828125, "mean_p": 0.201708, "boundary_entropy": 0.158724, "bpb": 2.151769}, "fi": {"chunks": 4766.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.253362, "cps": 37.234375, "mean_p": 0.252937, "boundary_entropy": 0.124633, "bpb": 2.250256}}, "bpb": {"en": 1.97269, "tr": 2.15177, "fi": 2.25026}, "gini_cpc": 0.071194, "gini_cps": 0.088727, "bpic": 4.909311, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.060801091318225, "fi": 1.3659435405093001}, "t": 590.4}
{"event": "eval", "step": 3600, "phase": 2, "bpb": {"en": 1.98603, "tr": 2.15784, "fi": 2.25899}, "bytes_per_chunk": {"en": 5.084, "tr": 5.568, "fi": 4.22}, "bpb_mean": 2.13429, "frozen_inventory": false, "t": 591.6}
{"event": "train", "step": 3600, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.46563, "bpb": 2.11446, "chunk_loss": 0.030377, "F": 0.19931, "G": 0.20857, "chunks": 204.09, "bytes_per_chunk": 5.017, "boundary_entropy": 0.1247, "router_noise": 0.0, "grad_norm": 0.6971, "lr": 0.0001319, "router_trainable": true, "sec_per_step": 0.199, "alpha": 0.03, "t": 592.0}
{"event": "train", "step": 3650, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.4824, "bpb": 2.13866, "chunk_loss": 0.03056, "F": 0.20496, "G": 0.2165, "chunks": 209.88, "bytes_per_chunk": 4.879, "boundary_entropy": 0.12869, "router_noise": 0.0, "grad_norm": 0.7145, "lr": 0.0001283, "router_trainable": true, "sec_per_step": 0.154, "alpha": 0.03, "t": 599.7}
{"event": "train", "step": 3700, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.38136, "bpb": 1.99289, "chunk_loss": 0.030334, "F": 0.19574, "G": 0.20623, "chunks": 200.44, "bytes_per_chunk": 5.109, "boundary_entropy": 0.13248, "router_noise": 0.0, "grad_norm": 0.7037, "lr": 0.0001247, "router_trainable": true, "sec_per_step": 0.148, "alpha": 0.03, "t": 607.1}
{"event": "train", "step": 3750, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.40674, "bpb": 2.0295, "chunk_loss": 0.030548, "F": 0.20804, "G": 0.21488, "chunks": 213.03, "bytes_per_chunk": 4.807, "boundary_entropy": 0.11848, "router_noise": 0.0, "grad_norm": 0.6848, "lr": 0.0001212, "router_trainable": true, "sec_per_step": 0.15, "alpha": 0.03, "t": 614.7}
{"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": 0.196714, "boundary_entropy": 0.120919, "bpb": 1.963429}, "tr": {"chunks": 3568.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.197094, "cps": 27.875, "mean_p": 0.197432, "boundary_entropy": 0.157055, "bpb": 2.130085}, "fi": {"chunks": 4753.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.252671, "cps": 37.132812, "mean_p": 0.251642, "boundary_entropy": 0.120806, "bpb": 2.237898}}, "bpb": {"en": 1.96343, "tr": 2.13009, "fi": 2.2379}, "gini_cpc": 0.068779, "gini_cps": 0.086335, "bpic": 4.901253, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0540977003780967, "fi": 1.3513333732172361}, "t": 622.6}
{"event": "eval", "step": 3800, "phase": 2, "bpb": {"en": 1.97491, "tr": 2.13651, "fi": 2.24511}, "bytes_per_chunk": {"en": 5.062, "tr": 5.523, "fi": 4.236}, "bpb_mean": 2.11884, "frozen_inventory": false, "t": 623.9}
{"event": "train", "step": 3800, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.38863, "bpb": 2.00337, "chunk_loss": 0.030419, "F": 0.19699, "G": 0.20551, "chunks": 201.72, "bytes_per_chunk": 5.076, "boundary_entropy": 0.12335, "router_noise": 0.0, "grad_norm": 0.6424, "lr": 0.0001177, "router_trainable": true, "sec_per_step": 0.193, "alpha": 0.03, "t": 624.3}
{"event": "train", "step": 3850, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.37771, "bpb": 1.98762, "chunk_loss": 0.030257, "F": 0.19193, "G": 0.20162, "chunks": 196.53, "bytes_per_chunk": 5.21, "boundary_entropy": 0.12836, "router_noise": 0.0, "grad_norm": 0.7476, "lr": 0.0001142, "router_trainable": true, "sec_per_step": 0.153, "alpha": 0.03, "t": 632.0}
{"event": "train", "step": 3900, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.41077, "bpb": 2.03531, "chunk_loss": 0.030392, "F": 0.19974, "G": 0.20813, "chunks": 204.53, "bytes_per_chunk": 5.007, "boundary_entropy": 0.12827, "router_noise": 0.0, "grad_norm": 0.683, "lr": 0.0001108, "router_trainable": true, "sec_per_step": 0.15, "alpha": 0.03, "t": 639.5}
{"event": "train", "step": 3950, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.43968, "bpb": 2.07703, "chunk_loss": 0.030452, "F": 0.2049, "G": 0.21194, "chunks": 209.81, "bytes_per_chunk": 4.881, "boundary_entropy": 0.12018, "router_noise": 0.0, "grad_norm": 0.6866, "lr": 0.0001074, "router_trainable": true, "sec_per_step": 0.149, "alpha": 0.03, "t": 646.9}
{"event": "probe", "step": 4000, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3240.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.186175, "cps": 25.3125, "mean_p": 0.196345, "boundary_entropy": 0.116846, "bpb": 1.94997}, "tr": {"chunks": 3604.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.199083, "cps": 28.15625, "mean_p": 0.197526, "boundary_entropy": 0.159814, "bpb": 2.109437}, "fi": {"chunks": 4786.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.254426, "cps": 37.390625, "mean_p": 0.253094, "boundary_entropy": 0.121135, "bpb": 2.217941}}, "bpb": {"en": 1.94997, "tr": 2.10944, "fi": 2.21794}, "gini_cpc": 0.07113, "gini_cps": 0.088621, "bpic": 4.878074, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0693339143706486, "fi": 1.3665952939277057}, "t": 655.1}
{"event": "eval", "step": 4000, "phase": 2, "bpb": {"en": 1.96282, "tr": 2.11641, "fi": 2.22542}, "bytes_per_chunk": {"en": 5.042, "tr": 5.526, "fi": 4.218}, "bpb_mean": 2.10155, "frozen_inventory": false, "t": 656.3}
{"event": "train", "step": 4000, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.45225, "bpb": 2.09515, "chunk_loss": 0.030445, "F": 0.2037, "G": 0.2119, "chunks": 208.59, "bytes_per_chunk": 4.909, "boundary_entropy": 0.12246, "router_noise": 0.0, "grad_norm": 0.7138, "lr": 0.000104, "router_trainable": true, "sec_per_step": 0.195, "alpha": 0.03, "t": 656.7}
{"event": "train", "step": 4050, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.37818, "bpb": 1.9883, "chunk_loss": 0.030363, "F": 0.19714, "G": 0.20656, "chunks": 201.88, "bytes_per_chunk": 5.072, "boundary_entropy": 0.11936, "router_noise": 0.0, "grad_norm": 0.6906, "lr": 0.0001007, "router_trainable": true, "sec_per_step": 0.148, "alpha": 0.03, "t": 664.1}
{"event": "train", "step": 4100, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.45206, "bpb": 2.09488, "chunk_loss": 0.030558, "F": 0.20691, "G": 0.21511, "chunks": 211.88, "bytes_per_chunk": 4.833, "boundary_entropy": 0.12158, "router_noise": 0.0, "grad_norm": 0.735, "lr": 9.75e-05, "router_trainable": true, "sec_per_step": 0.151, "alpha": 0.03, "t": 671.6}
{"event": "train", "step": 4150, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.40026, "bpb": 2.02015, "chunk_loss": 0.030673, "F": 0.21375, "G": 0.21966, "chunks": 218.88, "bytes_per_chunk": 4.678, "boundary_entropy": 0.1127, "router_noise": 0.0, "grad_norm": 0.7044, "lr": 9.43e-05, "router_trainable": true, "sec_per_step": 0.149, "alpha": 0.03, "t": 679.1}
{"event": "probe", "step": 4200, "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": 0.194266, "boundary_entropy": 0.112891, "bpb": 1.940378}, "tr": {"chunks": 3562.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.196763, "cps": 27.828125, "mean_p": 0.192018, "boundary_entropy": 0.157559, "bpb": 2.094424}, "fi": {"chunks": 4785.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.254372, "cps": 37.382812, "mean_p": 0.252628, "boundary_entropy": 0.12126, "bpb": 2.193737}}, "bpb": {"en": 1.94038, "tr": 2.09442, "fi": 2.19374}, "gini_cpc": 0.073474, "gini_cps": 0.090985, "bpic": 4.909736, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0674145644561193, "fi": 1.3799387787243391}, "t": 687.5}
{"event": "eval", "step": 4200, "phase": 2, "bpb": {"en": 1.95114, "tr": 2.10245, "fi": 2.20109}, "bytes_per_chunk": {"en": 5.079, "tr": 5.558, "fi": 4.216}, "bpb_mean": 2.08489, "frozen_inventory": false, "t": 688.8}
{"event": "phase_transition", "step": 4200, "phase": 3, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0, "t": 689.0}
{"event": "train", "step": 4200, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.38889, "bpb": 2.00375, "chunk_loss": 0.030259, "F": 0.19354, "G": 0.20185, "chunks": 198.19, "bytes_per_chunk": 5.167, "boundary_entropy": 0.11678, "router_noise": 0.0, "grad_norm": 0.6664, "lr": 9.12e-05, "router_trainable": true, "sec_per_step": 0.201, "alpha": 0.03, "t": 689.1}
{"event": "train", "step": 4250, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.39062, "bpb": 2.00624, "chunk_loss": 0.030322, "F": 0.19382, "G": 0.20306, "chunks": 198.47, "bytes_per_chunk": 5.16, "boundary_entropy": 0.12295, "router_noise": 0.0, "grad_norm": 0.7118, "lr": 8.81e-05, "router_trainable": true, "sec_per_step": 0.147, "alpha": 0.03, "t": 696.5}
{"event": "train", "step": 4300, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.48108, "bpb": 2.13674, "chunk_loss": 0.030428, "F": 0.20126, "G": 0.20964, "chunks": 206.09, "bytes_per_chunk": 4.969, "boundary_entropy": 0.12771, "router_noise": 0.0, "grad_norm": 0.8031, "lr": 8.5e-05, "router_trainable": true, "sec_per_step": 0.148, "alpha": 0.03, "t": 703.9}
{"event": "train", "step": 4350, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.33241, "bpb": 1.92226, "chunk_loss": 0.030409, "F": 0.19952, "G": 0.20623, "chunks": 204.31, "bytes_per_chunk": 5.012, "boundary_entropy": 0.11588, "router_noise": 0.0, "grad_norm": 0.7803, "lr": 8.21e-05, "router_trainable": true, "sec_per_step": 0.147, "alpha": 0.03, "t": 711.2}
{"event": "probe", "step": 4400, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3236.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.185945, "cps": 25.28125, "mean_p": 0.195608, "boundary_entropy": 0.111526, "bpb": 1.927643}, "tr": {"chunks": 3538.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.195437, "cps": 27.640625, "mean_p": 0.194235, "boundary_entropy": 0.16086, "bpb": 2.081223}, "fi": {"chunks": 4754.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.252724, "cps": 37.140625, "mean_p": 0.251196, "boundary_entropy": 0.119845, "bpb": 2.178334}}, "bpb": {"en": 1.92764, "tr": 2.08122, "fi": 2.17833}, "gini_cpc": 0.070209, "gini_cps": 0.087786, "bpic": 4.921235, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0510488089477277, "fi": 1.3591359538402268}, "t": 719.6}
{"event": "eval", "step": 4400, "phase": 3, "bpb": {"en": 1.94047, "tr": 2.08844, "fi": 2.18571}, "bytes_per_chunk": {"en": 5.071, "tr": 5.578, "fi": 4.227}, "bpb_mean": 2.07154, "frozen_inventory": false, "t": 720.8}
{"event": "train", "step": 4400, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.44499, "bpb": 2.08468, "chunk_loss": 0.030629, "F": 0.20697, "G": 0.21463, "chunks": 211.94, "bytes_per_chunk": 4.832, "boundary_entropy": 0.11862, "router_noise": 0.0, "grad_norm": 0.7786, "lr": 7.92e-05, "router_trainable": true, "sec_per_step": 0.2, "alpha": 0.03, "t": 721.2}
{"event": "train", "step": 4450, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.38612, "bpb": 1.99975, "chunk_loss": 0.0305, "F": 0.2045, "G": 0.21172, "chunks": 209.41, "bytes_per_chunk": 4.89, "boundary_entropy": 0.1145, "router_noise": 0.0, "grad_norm": 0.7092, "lr": 7.63e-05, "router_trainable": true, "sec_per_step": 0.153, "alpha": 0.03, "t": 728.8}
{"event": "train", "step": 4500, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.39529, "bpb": 2.01297, "chunk_loss": 0.030551, "F": 0.20682, "G": 0.21417, "chunks": 211.78, "bytes_per_chunk": 4.835, "boundary_entropy": 0.11768, "router_noise": 0.0, "grad_norm": 0.7621, "lr": 7.36e-05, "router_trainable": true, "sec_per_step": 0.147, "alpha": 0.03, "t": 736.2}
{"event": "train", "step": 4550, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.37324, "bpb": 1.98117, "chunk_loss": 0.03045, "F": 0.20297, "G": 0.21103, "chunks": 207.84, "bytes_per_chunk": 4.927, "boundary_entropy": 0.11653, "router_noise": 0.0, "grad_norm": 0.6711, "lr": 7.09e-05, "router_trainable": true, "sec_per_step": 0.14, "alpha": 0.03, "t": 743.2}
{"event": "probe", "step": 4600, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3210.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.184451, "cps": 25.078125, "mean_p": 0.194118, "boundary_entropy": 0.110147, "bpb": 1.919331}, "tr": {"chunks": 3522.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.194553, "cps": 27.515625, "mean_p": 0.191203, "boundary_entropy": 0.161039, "bpb": 2.066083}, "fi": {"chunks": 4769.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.253522, "cps": 37.257812, "mean_p": 0.250318, "boundary_entropy": 0.118089, "bpb": 2.170195}}, "bpb": {"en": 1.91933, "tr": 2.06608, "fi": 2.1702}, "gini_cpc": 0.072799, "gini_cps": 0.090369, "bpic": 4.932788, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0547702890159683, "fi": 1.3744676633327986}, "t": 751.5}
{"event": "eval", "step": 4600, "phase": 3, "bpb": {"en": 1.93107, "tr": 2.07403, "fi": 2.17811}, "bytes_per_chunk": {"en": 5.101, "tr": 5.602, "fi": 4.233}, "bpb_mean": 2.06107, "frozen_inventory": false, "t": 752.8}
{"event": "train", "step": 4600, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.43286, "bpb": 2.06718, "chunk_loss": 0.030361, "F": 0.19574, "G": 0.20422, "chunks": 200.44, "bytes_per_chunk": 5.109, "boundary_entropy": 0.11992, "router_noise": 0.0, "grad_norm": 0.7568, "lr": 6.82e-05, "router_trainable": true, "sec_per_step": 0.2, "alpha": 0.03, "t": 753.1}
{"event": "train", "step": 4650, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.34707, "bpb": 1.94341, "chunk_loss": 0.030355, "F": 0.19778, "G": 0.20492, "chunks": 202.53, "bytes_per_chunk": 5.056, "boundary_entropy": 0.1202, "router_noise": 0.0, "grad_norm": 0.6757, "lr": 6.57e-05, "router_trainable": true, "sec_per_step": 0.154, "alpha": 0.03, "t": 760.8}
{"event": "train", "step": 4700, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.36633, "bpb": 1.9712, "chunk_loss": 0.030403, "F": 0.19937, "G": 0.20539, "chunks": 204.16, "bytes_per_chunk": 5.016, "boundary_entropy": 0.11388, "router_noise": 0.0, "grad_norm": 0.703, "lr": 6.32e-05, "router_trainable": true, "sec_per_step": 0.151, "alpha": 0.03, "t": 768.3}
{"event": "train", "step": 4750, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.33372, "bpb": 1.92415, "chunk_loss": 0.030395, "F": 0.2001, "G": 0.20868, "chunks": 204.91, "bytes_per_chunk": 4.997, "boundary_entropy": 0.11824, "router_noise": 0.0, "grad_norm": 0.6989, "lr": 6.08e-05, "router_trainable": true, "sec_per_step": 0.146, "alpha": 0.03, "t": 775.6}
{"event": "probe", "step": 4800, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3230.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.1856, "cps": 25.234375, "mean_p": 0.194544, "boundary_entropy": 0.104471, "bpb": 1.908267}, "tr": {"chunks": 3509.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.193835, "cps": 27.414062, "mean_p": 0.191047, "boundary_entropy": 0.159989, "bpb": 2.056443}, "fi": {"chunks": 4782.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.254213, "cps": 37.359375, "mean_p": 0.251676, "boundary_entropy": 0.115776, "bpb": 2.155153}}, "bpb": {"en": 1.90827, "tr": 2.05644, "fi": 2.15515}, "gini_cpc": 0.072188, "gini_cps": 0.089807, "bpic": 4.924225, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.044370064041863, "fi": 1.3696805422951757}, "t": 783.5}
{"event": "eval", "step": 4800, "phase": 3, "bpb": {"en": 1.91984, "tr": 2.06314, "fi": 2.16318}, "bytes_per_chunk": {"en": 5.059, "tr": 5.626, "fi": 4.211}, "bpb_mean": 2.04872, "frozen_inventory": false, "t": 784.7}
{"event": "train", "step": 4800, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.40404, "bpb": 2.0256, "chunk_loss": 0.030682, "F": 0.21133, "G": 0.21889, "chunks": 216.41, "bytes_per_chunk": 4.732, "boundary_entropy": 0.11325, "router_noise": 0.0, "grad_norm": 0.7568, "lr": 5.85e-05, "router_trainable": true, "sec_per_step": 0.189, "alpha": 0.03, "t": 785.1}
{"event": "train", "step": 4850, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.35427, "bpb": 1.95379, "chunk_loss": 0.030564, "F": 0.20706, "G": 0.21401, "chunks": 212.03, "bytes_per_chunk": 4.829, "boundary_entropy": 0.11406, "router_noise": 0.0, "grad_norm": 0.7092, "lr": 5.62e-05, "router_trainable": true, "sec_per_step": 0.153, "alpha": 0.03, "t": 792.7}
{"event": "train", "step": 4900, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.36348, "bpb": 1.96709, "chunk_loss": 0.030298, "F": 0.19412, "G": 0.2028, "chunks": 198.78, "bytes_per_chunk": 5.151, "boundary_entropy": 0.11047, "router_noise": 0.0, "grad_norm": 0.706, "lr": 5.41e-05, "router_trainable": true, "sec_per_step": 0.153, "alpha": 0.03, "t": 800.3}
{"event": "train", "step": 4950, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.34616, "bpb": 1.94209, "chunk_loss": 0.030476, "F": 0.20438, "G": 0.21052, "chunks": 209.28, "bytes_per_chunk": 4.893, "boundary_entropy": 0.10595, "router_noise": 0.0, "grad_norm": 0.7177, "lr": 5.2e-05, "router_trainable": true, "sec_per_step": 0.154, "alpha": 0.03, "t": 808.1}
{"event": "probe", "step": 5000, "phase": 3, "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": 0.194008, "boundary_entropy": 0.106239, "bpb": 1.897979}, "tr": {"chunks": 3540.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.195548, "cps": 27.65625, "mean_p": 0.192164, "boundary_entropy": 0.156674, "bpb": 2.047905}, "fi": {"chunks": 4780.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.254107, "cps": 37.34375, "mean_p": 0.250963, "boundary_entropy": 0.112618, "bpb": 2.140917}}, "bpb": {"en": 1.89798, "tr": 2.04791, "fi": 2.14092}, "gini_cpc": 0.071827, "gini_cps": 0.089401, "bpic": 4.911436, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0532703839388877, "fi": 1.3686839527258567}, "t": 815.9}
{"event": "eval", "step": 5000, "phase": 3, "bpb": {"en": 1.91133, "tr": 2.05383, "fi": 2.14904}, "bytes_per_chunk": {"en": 5.055, "tr": 5.566, "fi": 4.213}, "bpb_mean": 2.03807, "frozen_inventory": false, "t": 817.1}
{"event": "train", "step": 5000, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.35044, "bpb": 1.94827, "chunk_loss": 0.03045, "F": 0.20181, "G": 0.20862, "chunks": 206.66, "bytes_per_chunk": 4.955, "boundary_entropy": 0.11283, "router_noise": 0.0, "grad_norm": 0.709, "lr": 5e-05, "router_trainable": true, "sec_per_step": 0.188, "alpha": 0.03, "t": 817.5}
{"event": "train", "step": 5050, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.36138, "bpb": 1.96405, "chunk_loss": 0.030402, "F": 0.20206, "G": 0.20835, "chunks": 206.91, "bytes_per_chunk": 4.949, "boundary_entropy": 0.10934, "router_noise": 0.0, "grad_norm": 0.6984, "lr": 4.81e-05, "router_trainable": true, "sec_per_step": 0.147, "alpha": 0.03, "t": 824.8}
{"event": "train", "step": 5100, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.33381, "bpb": 1.92428, "chunk_loss": 0.030257, "F": 0.19434, "G": 0.1998, "chunks": 199.0, "bytes_per_chunk": 5.146, "boundary_entropy": 0.1069, "router_noise": 0.0, "grad_norm": 0.7094, "lr": 4.63e-05, "router_trainable": true, "sec_per_step": 0.153, "alpha": 0.03, "t": 832.5}
{"event": "train", "step": 5150, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.31896, "bpb": 1.90286, "chunk_loss": 0.030427, "F": 0.20181, "G": 0.20766, "chunks": 206.66, "bytes_per_chunk": 4.955, "boundary_entropy": 0.10982, "router_noise": 0.0, "grad_norm": 0.7138, "lr": 4.45e-05, "router_trainable": true, "sec_per_step": 0.152, "alpha": 0.03, "t": 840.1}
{"event": "probe", "step": 5200, "phase": 3, "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": 0.193453, "boundary_entropy": 0.105925, "bpb": 1.891817}, "tr": {"chunks": 3538.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.195437, "cps": 27.640625, "mean_p": 0.191399, "boundary_entropy": 0.153742, "bpb": 2.037809}, "fi": {"chunks": 4763.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.253203, "cps": 37.210938, "mean_p": 0.2498, "boundary_entropy": 0.113638, "bpb": 2.134039}}, "bpb": {"en": 1.89182, "tr": 2.03781, "fi": 2.13404}, "gini_cpc": 0.071527, "gini_cps": 0.089089, "bpic": 4.922943, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0552882239388295, "fi": 1.3672014577194413}, "t": 848.2}
{"event": "eval", "step": 5200, "phase": 3, "bpb": {"en": 1.90525, "tr": 2.04429, "fi": 2.14088}, "bytes_per_chunk": {"en": 5.086, "tr": 5.578, "fi": 4.225}, "bpb_mean": 2.03014, "frozen_inventory": false, "t": 849.4}
{"event": "train", "step": 5200, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.36402, "bpb": 1.96787, "chunk_loss": 0.030607, "F": 0.20859, "G": 0.21507, "chunks": 213.59, "bytes_per_chunk": 4.794, "boundary_entropy": 0.1138, "router_noise": 0.0, "grad_norm": 0.7216, "lr": 4.29e-05, "router_trainable": true, "sec_per_step": 0.194, "alpha": 0.03, "t": 849.8}
{"event": "train", "step": 5250, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.39539, "bpb": 2.01312, "chunk_loss": 0.030454, "F": 0.20447, "G": 0.21069, "chunks": 209.38, "bytes_per_chunk": 4.891, "boundary_entropy": 0.11277, "router_noise": 0.0, "grad_norm": 0.7018, "lr": 4.14e-05, "router_trainable": true, "sec_per_step": 0.147, "alpha": 0.03, "t": 857.2}
{"event": "train", "step": 5300, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.37821, "bpb": 1.98834, "chunk_loss": 0.030353, "F": 0.19833, "G": 0.20571, "chunks": 203.09, "bytes_per_chunk": 5.042, "boundary_entropy": 0.11036, "router_noise": 0.0, "grad_norm": 0.6855, "lr": 3.99e-05, "router_trainable": true, "sec_per_step": 0.15, "alpha": 0.03, "t": 864.7}
{"event": "train", "step": 5350, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.34581, "bpb": 1.94159, "chunk_loss": 0.03057, "F": 0.20303, "G": 0.20846, "chunks": 207.91, "bytes_per_chunk": 4.925, "boundary_entropy": 0.11331, "router_noise": 0.0, "grad_norm": 0.7797, "lr": 3.86e-05, "router_trainable": true, "sec_per_step": 0.154, "alpha": 0.03, "t": 872.4}
{"event": "probe", "step": 5400, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3227.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.185428, "cps": 25.210938, "mean_p": 0.193114, "boundary_entropy": 0.101209, "bpb": 1.888088}, "tr": {"chunks": 3564.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.196873, "cps": 27.84375, "mean_p": 0.190553, "boundary_entropy": 0.151726, "bpb": 2.031233}, "fi": {"chunks": 4775.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.253841, "cps": 37.304688, "mean_p": 0.250306, "boundary_entropy": 0.110285, "bpb": 2.12613}}, "bpb": {"en": 1.88809, "tr": 2.03123, "fi": 2.12613}, "gini_cpc": 0.071696, "gini_cps": 0.089227, "bpic": 4.905067, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0617256236525967, "fi": 1.368947040708648}, "t": 880.1}
{"event": "eval", "step": 5400, "phase": 3, "bpb": {"en": 1.90048, "tr": 2.03702, "fi": 2.1331}, "bytes_per_chunk": {"en": 5.074, "tr": 5.558, "fi": 4.224}, "bpb_mean": 2.02353, "frozen_inventory": false, "t": 881.2}
{"event": "train", "step": 5400, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.26942, "bpb": 1.83138, "chunk_loss": 0.030342, "F": 0.19305, "G": 0.19886, "chunks": 197.69, "bytes_per_chunk": 5.18, "boundary_entropy": 0.10724, "router_noise": 0.0, "grad_norm": 0.6823, "lr": 3.73e-05, "router_trainable": true, "sec_per_step": 0.187, "alpha": 0.03, "t": 881.7}
{"event": "train", "step": 5450, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.31718, "bpb": 1.90028, "chunk_loss": 0.030422, "F": 0.20383, "G": 0.20966, "chunks": 208.72, "bytes_per_chunk": 4.906, "boundary_entropy": 0.10562, "router_noise": 0.0, "grad_norm": 0.691, "lr": 3.62e-05, "router_trainable": true, "sec_per_step": 0.154, "alpha": 0.03, "t": 889.4}
{"event": "train", "step": 5500, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.32513, "bpb": 1.91175, "chunk_loss": 0.030381, "F": 0.20016, "G": 0.20659, "chunks": 204.97, "bytes_per_chunk": 4.996, "boundary_entropy": 0.1104, "router_noise": 0.0, "grad_norm": 0.7052, "lr": 3.51e-05, "router_trainable": true, "sec_per_step": 0.146, "alpha": 0.03, "t": 896.7}
{"event": "train", "step": 5550, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.31494, "bpb": 1.89705, "chunk_loss": 0.030442, "F": 0.20557, "G": 0.21057, "chunks": 210.5, "bytes_per_chunk": 4.865, "boundary_entropy": 0.10213, "router_noise": 0.0, "grad_norm": 0.6746, "lr": 3.41e-05, "router_trainable": true, "sec_per_step": 0.149, "alpha": 0.03, "t": 904.1}
{"event": "probe", "step": 5600, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3241.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.186232, "cps": 25.320312, "mean_p": 0.193107, "boundary_entropy": 0.101138, "bpb": 1.881184}, "tr": {"chunks": 3583.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.197923, "cps": 27.992188, "mean_p": 0.192039, "boundary_entropy": 0.150877, "bpb": 2.024184}, "fi": {"chunks": 4784.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.254319, "cps": 37.375, "mean_p": 0.251508, "boundary_entropy": 0.110795, "bpb": 2.116036}}, "bpb": {"en": 1.88118, "tr": 2.02418, "fi": 2.11604}, "gini_cpc": 0.071093, "gini_cps": 0.088617, "bpic": 4.887319, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0627750393915663, "fi": 1.365602731246403}, "t": 912.3}
{"event": "eval", "step": 5600, "phase": 3, "bpb": {"en": 1.89513, "tr": 2.03039, "fi": 2.12438}, "bytes_per_chunk": {"en": 5.051, "tr": 5.541, "fi": 4.209}, "bpb_mean": 2.01663, "frozen_inventory": false, "t": 913.5}
{"event": "train", "step": 5600, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.34571, "bpb": 1.94145, "chunk_loss": 0.030487, "F": 0.20612, "G": 0.21185, "chunks": 211.06, "bytes_per_chunk": 4.852, "boundary_entropy": 0.11123, "router_noise": 0.0, "grad_norm": 0.7195, "lr": 3.33e-05, "router_trainable": true, "sec_per_step": 0.196, "alpha": 0.03, "t": 913.9}
{"event": "train", "step": 5650, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.32644, "bpb": 1.91365, "chunk_loss": 0.030394, "F": 0.20078, "G": 0.20629, "chunks": 205.59, "bytes_per_chunk": 4.981, "boundary_entropy": 0.10482, "router_noise": 0.0, "grad_norm": 0.6872, "lr": 3.25e-05, "router_trainable": true, "sec_per_step": 0.153, "alpha": 0.03, "t": 921.6}
{"event": "train", "step": 5700, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.4016, "bpb": 2.02208, "chunk_loss": 0.030452, "F": 0.20517, "G": 0.21037, "chunks": 210.09, "bytes_per_chunk": 4.874, "boundary_entropy": 0.11111, "router_noise": 0.0, "grad_norm": 0.7204, "lr": 3.18e-05, "router_trainable": true, "sec_per_step": 0.155, "alpha": 0.03, "t": 929.4}
{"event": "train", "step": 5750, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.38999, "bpb": 2.00533, "chunk_loss": 0.030469, "F": 0.20193, "G": 0.20831, "chunks": 206.78, "bytes_per_chunk": 4.952, "boundary_entropy": 0.11084, "router_noise": 0.0, "grad_norm": 0.7281, "lr": 3.13e-05, "router_trainable": true, "sec_per_step": 0.145, "alpha": 0.03, "t": 936.6}
{"event": "probe", "step": 5800, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3242.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.18629, "cps": 25.328125, "mean_p": 0.192652, "boundary_entropy": 0.100032, "bpb": 1.877737}, "tr": {"chunks": 3577.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.197592, "cps": 27.945312, "mean_p": 0.190175, "boundary_entropy": 0.150398, "bpb": 2.019478}, "fi": {"chunks": 4787.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.254479, "cps": 37.398438, "mean_p": 0.250641, "boundary_entropy": 0.10978, "bpb": 2.110807}}, "bpb": {"en": 1.87774, "tr": 2.01948, "fi": 2.11081}, "gini_cpc": 0.071213, "gini_cps": 0.088747, "bpic": 4.888161, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0606680778571778, "fi": 1.366037601019079}, "t": 944.7}
{"event": "eval", "step": 5800, "phase": 3, "bpb": {"en": 1.89121, "tr": 2.02524, "fi": 2.11947}, "bytes_per_chunk": {"en": 5.056, "tr": 5.552, "fi": 4.213}, "bpb_mean": 2.01197, "frozen_inventory": false, "t": 945.9}
{"event": "train", "step": 5800, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.30504, "bpb": 1.88278, "chunk_loss": 0.030367, "F": 0.19986, "G": 0.20454, "chunks": 204.66, "bytes_per_chunk": 5.004, "boundary_entropy": 0.10454, "router_noise": 0.0, "grad_norm": 0.7195, "lr": 3.08e-05, "router_trainable": true, "sec_per_step": 0.195, "alpha": 0.03, "t": 946.3}
{"event": "train", "step": 5850, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.35669, "bpb": 1.95728, "chunk_loss": 0.030295, "F": 0.19324, "G": 0.19954, "chunks": 197.88, "bytes_per_chunk": 5.175, "boundary_entropy": 0.10459, "router_noise": 0.0, "grad_norm": 0.6956, "lr": 3.05e-05, "router_trainable": true, "sec_per_step": 0.147, "alpha": 0.03, "t": 953.7}
{"event": "train", "step": 5900, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.38771, "bpb": 2.00205, "chunk_loss": 0.030459, "F": 0.20532, "G": 0.21157, "chunks": 210.25, "bytes_per_chunk": 4.87, "boundary_entropy": 0.10932, "router_noise": 0.0, "grad_norm": 0.761, "lr": 3.02e-05, "router_trainable": true, "sec_per_step": 0.154, "alpha": 0.03, "t": 961.4}
{"event": "train", "step": 5950, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.32647, "bpb": 1.9137, "chunk_loss": 0.030373, "F": 0.19876, "G": 0.20424, "chunks": 203.53, "bytes_per_chunk": 5.031, "boundary_entropy": 0.11149, "router_noise": 0.0, "grad_norm": 0.8005, "lr": 3.01e-05, "router_trainable": true, "sec_per_step": 0.153, "alpha": 0.03, "t": 969.1}
{"event": "train", "step": 5999, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.34445, "bpb": 1.93963, "chunk_loss": 0.03051, "F": 0.20609, "G": 0.21162, "chunks": 211.03, "bytes_per_chunk": 4.852, "boundary_entropy": 0.11008, "router_noise": 0.0, "grad_norm": 0.7541, "lr": 3e-05, "router_trainable": true, "sec_per_step": 0.146, "alpha": 0.03, "t": 976.4}
{"event": "probe", "step": 6000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3219.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.184968, "cps": 25.148438, "mean_p": 0.192193, "boundary_entropy": 0.101607, "bpb": 1.875497}, "tr": {"chunks": 3569.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.19715, "cps": 27.882812, "mean_p": 0.190076, "boundary_entropy": 0.151831, "bpb": 2.013137}, "fi": {"chunks": 4770.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.253575, "cps": 37.265625, "mean_p": 0.249405, "boundary_entropy": 0.108656, "bpb": 2.108238}}, "bpb": {"en": 1.8755, "tr": 2.01314, "fi": 2.10824}, "gini_cpc": 0.07195, "gini_cps": 0.089462, "bpic": 4.908462, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.065857486612667, "fi": 1.370912193065042}, "t": 977.2}
{"event": "eval", "step": 6000, "phase": 3, "bpb": {"en": 1.88944, "tr": 2.01935, "fi": 2.11496}, "bytes_per_chunk": {"en": 5.088, "tr": 5.566, "fi": 4.222}, "bpb_mean": 2.00792, "frozen_inventory": false, "t": 978.4}
{"event": "reps_dumped", "dirs": ["/local/scratch/sas245/beetle-hnet/runs/ali_hnet_s0/probe/reps/th0.30", "/local/scratch/sas245/beetle-hnet/runs/ali_hnet_s0/probe/reps/th0.40", "/local/scratch/sas245/beetle-hnet/runs/ali_hnet_s0/probe/reps/th0.50", "/local/scratch/sas245/beetle-hnet/runs/ali_hnet_s0/probe/reps/th0.60", "/local/scratch/sas245/beetle-hnet/runs/ali_hnet_s0/probe/reps/th0.70"], "t": 979.6}
{"event": "final", "run": "ali_hnet_s0", "arch": "hnet", "objective": "ratio", "condition": "A", "step": 6000, "params": 22516992, "seed": 0, "per_lang": {"en": {"cpc": 0.18496810894673332, "cps": 25.1484375, "bpb": 1.8951917787615589, "bytes_per_chunk": 5.023744274736015}, "tr": {"cpc": 0.19714964370546317, "cps": 27.8828125, "bpb": 2.0111015393534433, "bytes_per_chunk": 5.568527487467074}, "fi": {"cpc": 0.2535750358832598, "cps": 37.265625, "bpb": 2.13352242337109, "bytes_per_chunk": 4.200150609648631}}, "bpb_mean": 2.013271913828697, "gini_cpc": 0.07194977214762988, "gini_cps": 0.08946184460979412, "bpic": 4.9084616715694755, "mean_boundary_entropy": 0.1100839227437973, "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": 981.8434720039368, "t": 981.8}