| {"event": "hub", "repo": "AdaptiveChunking/ali_hnet_s2", "keep_steps": [0, 1, 2, 4, 6, 8, 16, 32, 64, 128, 256, 512, 1000, 1024, 2048, 3000, 4096, 6000], "t": 3.0} |
| {"event": "start", "run": "ali_hnet_s2", "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": 3.0} |
| {"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": 3.0} |
| {"event": "probe", "step": 0, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 10933.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.628225, "cps": 85.414062, "mean_p": 0.513249, "boundary_entropy": 0.686295, "bpb": 8.12598}, "tr": {"chunks": 8187.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.452245, "cps": 63.960938, "mean_p": 0.498002, "boundary_entropy": 0.686826, "bpb": 8.097536}, "fi": {"chunks": 13540.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.719792, "cps": 105.78125, "mean_p": 0.517169, "boundary_entropy": 0.686661, "bpb": 8.065117}}, "bpb": {"en": 8.12598, "tr": 8.09754, "fi": 8.06512}, "gini_cpc": 0.099077, "gini_cps": 0.109267, "bpic": 1.737048, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.7198781817505053, "fi": 1.1457544508645294}, "t": 4.7} |
| {"event": "train", "step": 0, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 5.62735, "bpb": 8.11854, "chunk_loss": 0.062086, "F": 0.59833, "G": 0.50836, "chunks": 612.69, "bytes_per_chunk": 1.671, "boundary_entropy": 0.69105, "router_noise": 0.0, "grad_norm": 11.4934, "lr": 1e-06, "router_trainable": true, "sec_per_step": 0.009, "alpha": 0.03, "t": 5.3} |
| {"event": "probe", "step": 1, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 10907.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.626731, "cps": 85.210938, "mean_p": 0.513143, "boundary_entropy": 0.686304, "bpb": 8.094363}, "tr": {"chunks": 8206.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.453295, "cps": 64.109375, "mean_p": 0.498082, "boundary_entropy": 0.686828, "bpb": 8.0762}, "fi": {"chunks": 13506.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.717984, "cps": 105.515625, "mean_p": 0.517038, "boundary_entropy": 0.686672, "bpb": 8.036166}}, "bpb": {"en": 8.09436, "tr": 8.0762, "fi": 8.03617}, "gini_cpc": 0.098141, "gini_cps": 0.108321, "bpic": 1.739232, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.7232688618519676, "fi": 1.1456017516500347}, "t": 6.4} |
| {"event": "probe", "step": 2, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 10879.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.625122, "cps": 84.992188, "mean_p": 0.512978, "boundary_entropy": 0.686318, "bpb": 8.031167}, "tr": {"chunks": 8246.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.455505, "cps": 64.421875, "mean_p": 0.498244, "boundary_entropy": 0.686832, "bpb": 8.031976}, "fi": {"chunks": 13467.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.715911, "cps": 105.210938, "mean_p": 0.516799, "boundary_entropy": 0.686692, "bpb": 7.976312}}, "bpb": {"en": 8.03117, "tr": 8.03198, "fi": 7.97631}, "gini_cpc": 0.096633, "gini_cps": 0.106795, "bpic": 1.740673, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.7286650217164614, "fi": 1.1452337023900658}, "t": 7.7} |
| {"event": "probe", "step": 4, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 10716.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.615756, "cps": 83.71875, "mean_p": 0.512103, "boundary_entropy": 0.686379, "bpb": 7.819686}, "tr": {"chunks": 8388.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.463349, "cps": 65.53125, "mean_p": 0.498672, "boundary_entropy": 0.686846, "bpb": 7.878863}, "fi": {"chunks": 13214.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.702461, "cps": 103.234375, "mean_p": 0.515686, "boundary_entropy": 0.686774, "bpb": 7.773633}}, "bpb": {"en": 7.81969, "tr": 7.87886, "fi": 7.77363}, "gini_cpc": 0.089477, "gini_cps": 0.099552, "bpic": 1.75543, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.7524874924061202, "fi": 1.14081135248651}, "t": 9.2} |
| {"event": "probe", "step": 6, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 10326.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.593346, "cps": 80.671875, "mean_p": 0.510351, "boundary_entropy": 0.686483, "bpb": 7.542823}, "tr": {"chunks": 8507.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.469922, "cps": 66.460938, "mean_p": 0.499187, "boundary_entropy": 0.686869, "bpb": 7.674362}, "fi": {"chunks": 12730.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.676732, "cps": 99.453125, "mean_p": 0.513642, "boundary_entropy": 0.686908, "bpb": 7.518241}}, "bpb": {"en": 7.54282, "tr": 7.67436, "fi": 7.51824}, "gini_cpc": 0.079237, "gini_cps": 0.089197, "bpic": 1.797421, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.7919866861659939, "fi": 1.140534744049118}, "t": 10.4} |
| {"event": "probe", "step": 8, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 9630.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.553353, "cps": 75.234375, "mean_p": 0.507288, "boundary_entropy": 0.686614, "bpb": 7.2931}, "tr": {"chunks": 8638.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.477158, "cps": 67.484375, "mean_p": 0.499597, "boundary_entropy": 0.686902, "bpb": 7.478163}, "fi": {"chunks": 11875.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.63128, "cps": 92.773438, "mean_p": 0.510291, "boundary_entropy": 0.687081, "bpb": 7.295827}}, "bpb": {"en": 7.2931, "tr": 7.47816, "fi": 7.29583}, "gini_cpc": 0.061829, "gini_cps": 0.071592, "bpic": 1.882095, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.8623041601854946, "fi": 1.1408264137888433}, "t": 12.2} |
| {"event": "probe", "step": 16, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3751.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.215538, "cps": 29.304688, "mean_p": 0.484385, "boundary_entropy": 0.686161, "bpb": 6.724747}, "tr": {"chunks": 7272.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.401701, "cps": 56.8125, "mean_p": 0.494377, "boundary_entropy": 0.686996, "bpb": 7.013552}, "fi": {"chunks": 4585.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.24374, "cps": 35.820312, "mean_p": 0.48653, "boundary_entropy": 0.686903, "bpb": 6.787316}}, "bpb": {"en": 6.72475, "tr": 7.01355, "fi": 6.78732}, "gini_cpc": 0.144149, "gini_cps": 0.150393, "bpic": 3.634803, "worst_lang": "tr", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.8637187515798879, "fi": 1.1308487247481265}, "t": 14.4} |
| {"event": "probe", "step": 32, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2599.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.149342, "cps": 20.304688, "mean_p": 0.486719, "boundary_entropy": 0.686946, "bpb": 6.304665}, "tr": {"chunks": 4635.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.256035, "cps": 36.210938, "mean_p": 0.490159, "boundary_entropy": 0.687559, "bpb": 6.651827}, "fi": {"chunks": 1388.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.073787, "cps": 10.84375, "mean_p": 0.479392, "boundary_entropy": 0.686972, "bpb": 6.420813}}, "bpb": {"en": 6.30467, "tr": 6.65183, "fi": 6.42081}, "gini_cpc": 0.253564, "gini_cps": 0.251063, "bpic": 6.579912, "worst_lang": "tr", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.714419223571195, "fi": 0.49407789426131493}, "t": 18.0} |
| {"event": "train", "step": 50, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 4.21243, "bpb": 6.07726, "chunk_loss": 0.03416, "F": 0.22513, "G": 0.48848, "chunks": 230.53, "bytes_per_chunk": 4.442, "boundary_entropy": 0.69161, "router_noise": 0.0, "grad_norm": 2.4791, "lr": 5.1e-05, "router_trainable": true, "sec_per_step": 0.311, "alpha": 0.03, "t": 20.8} |
| {"event": "probe", "step": 64, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 4793.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.275412, "cps": 37.445312, "mean_p": 0.485088, "boundary_entropy": 0.685502, "bpb": 5.552533}, "tr": {"chunks": 5396.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.298072, "cps": 42.15625, "mean_p": 0.486572, "boundary_entropy": 0.686482, "bpb": 5.956691}, "fi": {"chunks": 4705.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.25012, "cps": 36.757812, "mean_p": 0.484012, "boundary_entropy": 0.686333, "bpb": 5.717972}}, "bpb": {"en": 5.55253, "tr": 5.95669, "fi": 5.71797}, "gini_cpc": 0.038815, "gini_cps": 0.03093, "bpic": 3.809051, "worst_lang": "tr", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0822761318762484, "fi": 0.9081643204466296}, "t": 23.1} |
| {"event": "train", "step": 100, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 3.24796, "bpb": 4.68582, "chunk_loss": 0.029594, "F": 0.15991, "G": 0.45237, "chunks": 163.75, "bytes_per_chunk": 6.253, "boundary_entropy": 0.68188, "router_noise": 0.0, "grad_norm": 1.4543, "lr": 0.000101, "router_trainable": true, "sec_per_step": 0.141, "alpha": 0.03, "t": 27.9} |
| {"event": "probe", "step": 128, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3897.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.223927, "cps": 30.445312, "mean_p": 0.417996, "boundary_entropy": 0.645449, "bpb": 4.158407}, "tr": {"chunks": 4686.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.258852, "cps": 36.609375, "mean_p": 0.436889, "boundary_entropy": 0.659898, "bpb": 4.397168}, "fi": {"chunks": 3796.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.201797, "cps": 29.65625, "mean_p": 0.428727, "boundary_entropy": 0.660362, "bpb": 4.348264}}, "bpb": {"en": 4.15841, "tr": 4.39717, "fi": 4.34826}, "gini_cpc": 0.055563, "gini_cps": 0.047931, "bpic": 4.582923, "worst_lang": "tr", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.1559670293122535, "fi": 0.9011727164535157}, "t": 31.9} |
| {"event": "train", "step": 150, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 2.78576, "bpb": 4.019, "chunk_loss": 0.03253, "F": 0.21838, "G": 0.39109, "chunks": 223.62, "bytes_per_chunk": 4.579, "boundary_entropy": 0.62316, "router_noise": 0.0, "grad_norm": 1.46, "lr": 0.000151, "router_trainable": true, "sec_per_step": 0.134, "alpha": 0.03, "t": 34.5} |
| {"event": "probe", "step": 200, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3779.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.217146, "cps": 29.523438, "mean_p": 0.384979, "boundary_entropy": 0.592065, "bpb": 3.633554}, "tr": {"chunks": 4529.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.25018, "cps": 35.382812, "mean_p": 0.371796, "boundary_entropy": 0.591487, "bpb": 3.829614}, "fi": {"chunks": 3949.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.20993, "cps": 30.851562, "mean_p": 0.404987, "boundary_entropy": 0.634178, "bpb": 3.884663}}, "bpb": {"en": 3.63355, "tr": 3.82961, "fi": 3.88466}, "gini_cpc": 0.03962, "gini_cps": 0.040793, "bpic": 4.628539, "worst_lang": "tr", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.1521234004290843, "fi": 0.9667684713592763}, "t": 41.0} |
| {"event": "eval", "step": 200, "phase": 1, "bpb": {"en": 3.63977, "tr": 3.84018, "fi": 3.88271}, "bytes_per_chunk": {"en": 5.071, "tr": 4.944, "fi": 5.583}, "bpb_mean": 3.78755, "frozen_inventory": false, "t": 42.0} |
| {"event": "train", "step": 200, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 2.53409, "bpb": 3.65592, "chunk_loss": 0.031219, "F": 0.19394, "G": 0.3694, "chunks": 198.59, "bytes_per_chunk": 5.156, "boundary_entropy": 0.59118, "router_noise": 0.0, "grad_norm": 1.2214, "lr": 0.000201, "router_trainable": true, "sec_per_step": 0.155, "alpha": 0.03, "t": 42.3} |
| {"event": "train", "step": 250, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 2.3774, "bpb": 3.42987, "chunk_loss": 0.031423, "F": 0.20648, "G": 0.32889, "chunks": 211.44, "bytes_per_chunk": 4.843, "boundary_entropy": 0.51375, "router_noise": 0.0, "grad_norm": 1.0599, "lr": 0.000251, "router_trainable": true, "sec_per_step": 0.12, "alpha": 0.03, "t": 48.3} |
| {"event": "probe", "step": 256, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3649.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.209676, "cps": 28.507812, "mean_p": 0.324876, "boundary_entropy": 0.493815, "bpb": 3.356633}, "tr": {"chunks": 5229.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.288847, "cps": 40.851562, "mean_p": 0.310168, "boundary_entropy": 0.489259, "bpb": 3.54604}, "fi": {"chunks": 4454.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.236776, "cps": 34.796875, "mean_p": 0.338982, "boundary_entropy": 0.546776, "bpb": 3.681546}}, "bpb": {"en": 3.35663, "tr": 3.54604, "fi": 3.68155}, "gini_cpc": 0.071781, "gini_cps": 0.079008, "bpic": 4.255326, "worst_lang": "tr", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.377584815926562, "fi": 1.1292460655524217}, "t": 49.6} |
| {"event": "train", "step": 300, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 2.27857, "bpb": 3.28728, "chunk_loss": 0.031683, "F": 0.2287, "G": 0.29081, "chunks": 234.19, "bytes_per_chunk": 4.373, "boundary_entropy": 0.425, "router_noise": 0.0, "grad_norm": 1.5015, "lr": 0.0003, "router_trainable": true, "sec_per_step": 0.155, "alpha": 0.03, "t": 56.1} |
| {"event": "train", "step": 350, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 2.16354, "bpb": 3.12133, "chunk_loss": 0.031175, "F": 0.22116, "G": 0.26386, "chunks": 226.47, "bytes_per_chunk": 4.522, "boundary_entropy": 0.36608, "router_noise": 0.0, "grad_norm": 1.3101, "lr": 0.0002999, "router_trainable": true, "sec_per_step": 0.12, "alpha": 0.03, "t": 62.1} |
| {"event": "probe", "step": 400, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3512.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.201804, "cps": 27.4375, "mean_p": 0.252943, "boundary_entropy": 0.336744, "bpb": 2.89713}, "tr": {"chunks": 3663.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.202342, "cps": 28.617188, "mean_p": 0.25403, "boundary_entropy": 0.369736, "bpb": 3.177128}, "fi": {"chunks": 3967.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.210887, "cps": 30.992188, "mean_p": 0.257532, "boundary_entropy": 0.406927, "bpb": 3.381797}}, "bpb": {"en": 2.89713, "tr": 3.17713, "fi": 3.3818}, "gini_cpc": 0.009845, "gini_cps": 0.027224, "bpic": 5.091725, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0026652883644678, "fi": 1.0450087575376643}, "t": 68.4} |
| {"event": "eval", "step": 400, "phase": 1, "bpb": {"en": 2.90958, "tr": 3.18572, "fi": 3.37588}, "bytes_per_chunk": {"en": 4.767, "tr": 5.304, "fi": 5.257}, "bpb_mean": 3.15706, "frozen_inventory": false, "t": 69.3} |
| {"event": "train", "step": 400, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 2.08269, "bpb": 3.00469, "chunk_loss": 0.030696, "F": 0.20218, "G": 0.25201, "chunks": 207.03, "bytes_per_chunk": 4.946, "boundary_entropy": 0.34447, "router_noise": 0.0, "grad_norm": 1.0808, "lr": 0.0002998, "router_trainable": true, "sec_per_step": 0.152, "alpha": 0.03, "t": 69.7} |
| {"event": "train", "step": 450, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.99952, "bpb": 2.88469, "chunk_loss": 0.030911, "F": 0.21429, "G": 0.252, "chunks": 219.44, "bytes_per_chunk": 4.666, "boundary_entropy": 0.30731, "router_noise": 0.0, "grad_norm": 1.5443, "lr": 0.0002995, "router_trainable": true, "sec_per_step": 0.122, "alpha": 0.03, "t": 75.7} |
| {"event": "train", "step": 500, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.8992, "bpb": 2.73997, "chunk_loss": 0.031095, "F": 0.22208, "G": 0.25394, "chunks": 227.41, "bytes_per_chunk": 4.503, "boundary_entropy": 0.28323, "router_noise": 0.0, "grad_norm": 1.367, "lr": 0.0002992, "router_trainable": true, "sec_per_step": 0.132, "alpha": 0.03, "t": 82.3} |
| {"event": "probe", "step": 512, "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": 0.244675, "boundary_entropy": 0.290288, "bpb": 2.698079}, "tr": {"chunks": 3987.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.22024, "cps": 31.148438, "mean_p": 0.248072, "boundary_entropy": 0.315903, "bpb": 3.028875}, "fi": {"chunks": 4118.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.218914, "cps": 32.171875, "mean_p": 0.26307, "boundary_entropy": 0.341951, "bpb": 3.225687}}, "bpb": {"en": 2.69808, "tr": 3.02888, "fi": 3.22569}, "gini_cpc": 0.016358, "gini_cps": 0.03201, "bpic": 4.864272, "worst_lang": "tr", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0772434464618286, "fi": 1.0707612235422872}, "t": 83.7} |
| {"event": "train", "step": 550, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.9029, "bpb": 2.7453, "chunk_loss": 0.030992, "F": 0.21863, "G": 0.25029, "chunks": 223.88, "bytes_per_chunk": 4.574, "boundary_entropy": 0.28033, "router_noise": 0.0, "grad_norm": 1.721, "lr": 0.0002987, "router_trainable": true, "sec_per_step": 0.128, "alpha": 0.03, "t": 88.7} |
| {"event": "probe", "step": 600, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3628.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.20847, "cps": 28.34375, "mean_p": 0.243408, "boundary_entropy": 0.272337, "bpb": 2.589181}, "tr": {"chunks": 3938.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.217533, "cps": 30.765625, "mean_p": 0.240516, "boundary_entropy": 0.281844, "bpb": 2.92009}, "fi": {"chunks": 4293.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.228218, "cps": 33.539062, "mean_p": 0.262821, "boundary_entropy": 0.307104, "bpb": 3.157361}}, "bpb": {"en": 2.58918, "tr": 2.92009, "fi": 3.15736}, "gini_cpc": 0.020123, "gini_cps": 0.037384, "bpic": 4.783877, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0434748886699476, "fi": 1.0947270436959022}, "t": 95.4} |
| {"event": "eval", "step": 600, "phase": 1, "bpb": {"en": 2.60198, "tr": 2.92696, "fi": 3.15854}, "bytes_per_chunk": {"en": 4.688, "tr": 5.148, "fi": 4.741}, "bpb_mean": 2.89583, "frozen_inventory": false, "t": 96.0} |
| {"event": "train", "step": 600, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.88946, "bpb": 2.72591, "chunk_loss": 0.030757, "F": 0.21048, "G": 0.24199, "chunks": 215.53, "bytes_per_chunk": 4.751, "boundary_entropy": 0.27481, "router_noise": 0.0, "grad_norm": 1.2604, "lr": 0.0002982, "router_trainable": true, "sec_per_step": 0.154, "alpha": 0.03, "t": 96.4} |
| {"event": "train", "step": 650, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.83558, "bpb": 2.64818, "chunk_loss": 0.030999, "F": 0.21973, "G": 0.2477, "chunks": 225.0, "bytes_per_chunk": 4.551, "boundary_entropy": 0.2581, "router_noise": 0.0, "grad_norm": 1.0785, "lr": 0.0002975, "router_trainable": true, "sec_per_step": 0.128, "alpha": 0.03, "t": 102.8} |
| {"event": "train", "step": 700, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.98644, "bpb": 2.86582, "chunk_loss": 0.030967, "F": 0.21652, "G": 0.24691, "chunks": 221.72, "bytes_per_chunk": 4.618, "boundary_entropy": 0.25484, "router_noise": 0.0, "grad_norm": 1.0804, "lr": 0.0002967, "router_trainable": true, "sec_per_step": 0.103, "alpha": 0.03, "t": 108.0} |
| {"event": "train", "step": 750, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.77714, "bpb": 2.56387, "chunk_loss": 0.031099, "F": 0.22672, "G": 0.24729, "chunks": 232.16, "bytes_per_chunk": 4.411, "boundary_entropy": 0.23883, "router_noise": 0.0, "grad_norm": 1.1502, "lr": 0.0002959, "router_trainable": true, "sec_per_step": 0.118, "alpha": 0.03, "t": 113.9} |
| {"event": "probe", "step": 800, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3853.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.221399, "cps": 30.101562, "mean_p": 0.246253, "boundary_entropy": 0.248188, "bpb": 2.43444}, "tr": {"chunks": 4145.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.228968, "cps": 32.382812, "mean_p": 0.235544, "boundary_entropy": 0.244761, "bpb": 2.762198}, "fi": {"chunks": 4486.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.238477, "cps": 35.046875, "mean_p": 0.25873, "boundary_entropy": 0.262105, "bpb": 2.982848}}, "bpb": {"en": 2.43444, "tr": 2.7622, "fi": 2.98285}, "gini_cpc": 0.016529, "gini_cps": 0.033803, "bpic": 4.544377, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0341870485064564, "fi": 1.0771408509991125}, "t": 120.4} |
| {"event": "eval", "step": 800, "phase": 1, "bpb": {"en": 2.44267, "tr": 2.77047, "fi": 2.98622}, "bytes_per_chunk": {"en": 4.519, "tr": 5.053, "fi": 4.473}, "bpb_mean": 2.73312, "frozen_inventory": false, "t": 121.3} |
| {"event": "train", "step": 800, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.86456, "bpb": 2.68999, "chunk_loss": 0.030953, "F": 0.21945, "G": 0.2458, "chunks": 224.72, "bytes_per_chunk": 4.557, "boundary_entropy": 0.24741, "router_noise": 0.0, "grad_norm": 1.2958, "lr": 0.0002949, "router_trainable": true, "sec_per_step": 0.156, "alpha": 0.03, "t": 121.7} |
| {"event": "train", "step": 850, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.81146, "bpb": 2.61339, "chunk_loss": 0.030753, "F": 0.21121, "G": 0.23877, "chunks": 216.28, "bytes_per_chunk": 4.735, "boundary_entropy": 0.24058, "router_noise": 0.0, "grad_norm": 0.9894, "lr": 0.0002938, "router_trainable": true, "sec_per_step": 0.114, "alpha": 0.03, "t": 127.4} |
| {"event": "train", "step": 900, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.61137, "bpb": 2.32472, "chunk_loss": 0.030793, "F": 0.21313, "G": 0.23798, "chunks": 218.25, "bytes_per_chunk": 4.692, "boundary_entropy": 0.22978, "router_noise": 0.0, "grad_norm": 1.2577, "lr": 0.0002927, "router_trainable": true, "sec_per_step": 0.104, "alpha": 0.03, "t": 132.6} |
| {"event": "train", "step": 950, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.69261, "bpb": 2.44192, "chunk_loss": 0.030661, "F": 0.20786, "G": 0.23361, "chunks": 212.84, "bytes_per_chunk": 4.811, "boundary_entropy": 0.22229, "router_noise": 0.0, "grad_norm": 1.2333, "lr": 0.0002914, "router_trainable": true, "sec_per_step": 0.116, "alpha": 0.03, "t": 138.4} |
| {"event": "probe", "step": 1000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3463.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.198989, "cps": 27.054688, "mean_p": 0.226578, "boundary_entropy": 0.227333, "bpb": 2.336753}, "tr": {"chunks": 4036.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.222946, "cps": 31.53125, "mean_p": 0.229808, "boundary_entropy": 0.223571, "bpb": 2.648397}, "fi": {"chunks": 4682.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.248897, "cps": 36.578125, "mean_p": 0.25907, "boundary_entropy": 0.233396, "bpb": 2.863214}}, "bpb": {"en": 2.33675, "tr": 2.6484, "fi": 2.86321}, "gini_cpc": 0.049598, "gini_cps": 0.066716, "bpic": 4.657417, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.1203977675217447, "fi": 1.2508094524414348}, "t": 144.7} |
| {"event": "eval", "step": 1000, "phase": 1, "bpb": {"en": 2.34477, "tr": 2.65679, "fi": 2.86941}, "bytes_per_chunk": {"en": 4.923, "tr": 5.278, "fi": 4.241}, "bpb_mean": 2.62366, "frozen_inventory": false, "t": 145.5} |
| {"event": "train", "step": 1000, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.74189, "bpb": 2.51302, "chunk_loss": 0.030674, "F": 0.20959, "G": 0.23248, "chunks": 214.62, "bytes_per_chunk": 4.771, "boundary_entropy": 0.22243, "router_noise": 0.0, "grad_norm": 1.0341, "lr": 0.0002901, "router_trainable": true, "sec_per_step": 0.149, "alpha": 0.03, "t": 145.9} |
| {"event": "probe", "step": 1024, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3429.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.197035, "cps": 26.789062, "mean_p": 0.226251, "boundary_entropy": 0.225772, "bpb": 2.32853}, "tr": {"chunks": 3895.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.215158, "cps": 30.429688, "mean_p": 0.226569, "boundary_entropy": 0.229924, "bpb": 2.646489}, "fi": {"chunks": 4934.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.262293, "cps": 38.546875, "mean_p": 0.260524, "boundary_entropy": 0.231816, "bpb": 2.825336}}, "bpb": {"en": 2.32853, "tr": 2.64649, "fi": 2.82534}, "gini_cpc": 0.064502, "gini_cps": 0.081851, "bpic": 4.628161, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0919771373383056, "fi": 1.331201801890547}, "t": 149.1} |
| {"event": "train", "step": 1050, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.63735, "bpb": 2.36219, "chunk_loss": 0.030632, "F": 0.20404, "G": 0.23013, "chunks": 208.94, "bytes_per_chunk": 4.901, "boundary_entropy": 0.22319, "router_noise": 0.0, "grad_norm": 0.9988, "lr": 0.0002886, "router_trainable": true, "sec_per_step": 0.137, "alpha": 0.03, "t": 152.7} |
| {"event": "train", "step": 1100, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.74281, "bpb": 2.51434, "chunk_loss": 0.030613, "F": 0.20706, "G": 0.23029, "chunks": 212.03, "bytes_per_chunk": 4.829, "boundary_entropy": 0.21712, "router_noise": 0.0, "grad_norm": 0.9711, "lr": 0.0002871, "router_trainable": true, "sec_per_step": 0.106, "alpha": 0.03, "t": 158.0} |
| {"event": "train", "step": 1150, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.64352, "bpb": 2.3711, "chunk_loss": 0.030751, "F": 0.21371, "G": 0.23414, "chunks": 218.84, "bytes_per_chunk": 4.679, "boundary_entropy": 0.20571, "router_noise": 0.0, "grad_norm": 0.9058, "lr": 0.0002855, "router_trainable": true, "sec_per_step": 0.117, "alpha": 0.03, "t": 163.8} |
| {"event": "probe", "step": 1200, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3401.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.195426, "cps": 26.570312, "mean_p": 0.221852, "boundary_entropy": 0.203402, "bpb": 2.266278}, "tr": {"chunks": 3895.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.215158, "cps": 30.429688, "mean_p": 0.217666, "boundary_entropy": 0.206991, "bpb": 2.560097}, "fi": {"chunks": 4944.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.262825, "cps": 38.625, "mean_p": 0.26433, "boundary_entropy": 0.208034, "bpb": 2.729921}}, "bpb": {"en": 2.26628, "tr": 2.5601, "fi": 2.72992}, "gini_cpc": 0.066724, "gini_cps": 0.084041, "bpic": 4.634967, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.1009672460844016, "fi": 1.3448816467108375}, "t": 169.4} |
| {"event": "eval", "step": 1200, "phase": 1, "bpb": {"en": 2.27649, "tr": 2.56935, "fi": 2.73911}, "bytes_per_chunk": {"en": 4.944, "tr": 5.334, "fi": 3.99}, "bpb_mean": 2.52832, "frozen_inventory": false, "t": 170.2} |
| {"event": "train", "step": 1200, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.61401, "bpb": 2.32853, "chunk_loss": 0.030578, "F": 0.20624, "G": 0.22749, "chunks": 211.19, "bytes_per_chunk": 4.849, "boundary_entropy": 0.2015, "router_noise": 0.0, "grad_norm": 0.8495, "lr": 0.0002837, "router_trainable": true, "sec_per_step": 0.135, "alpha": 0.03, "t": 170.6} |
| {"event": "train", "step": 1250, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.63537, "bpb": 2.35935, "chunk_loss": 0.030475, "F": 0.2009, "G": 0.22139, "chunks": 205.72, "bytes_per_chunk": 4.978, "boundary_entropy": 0.20362, "router_noise": 0.0, "grad_norm": 0.8698, "lr": 0.0002819, "router_trainable": true, "sec_per_step": 0.128, "alpha": 0.03, "t": 177.0} |
| {"event": "train", "step": 1300, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.67445, "bpb": 2.41572, "chunk_loss": 0.030887, "F": 0.21738, "G": 0.23335, "chunks": 222.59, "bytes_per_chunk": 4.6, "boundary_entropy": 0.19324, "router_noise": 0.0, "grad_norm": 0.9808, "lr": 0.00028, "router_trainable": true, "sec_per_step": 0.115, "alpha": 0.03, "t": 182.7} |
| {"event": "train", "step": 1350, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.59669, "bpb": 2.30354, "chunk_loss": 0.030611, "F": 0.20844, "G": 0.22668, "chunks": 213.44, "bytes_per_chunk": 4.798, "boundary_entropy": 0.19566, "router_noise": 0.0, "grad_norm": 0.8981, "lr": 0.000278, "router_trainable": true, "sec_per_step": 0.12, "alpha": 0.03, "t": 188.7} |
| {"event": "probe", "step": 1400, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3420.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.196518, "cps": 26.71875, "mean_p": 0.217797, "boundary_entropy": 0.192112, "bpb": 2.215922}, "tr": {"chunks": 3863.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.21339, "cps": 30.179688, "mean_p": 0.212193, "boundary_entropy": 0.189953, "bpb": 2.497239}, "fi": {"chunks": 4656.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.247515, "cps": 36.375, "mean_p": 0.253702, "boundary_entropy": 0.207263, "bpb": 2.6419}}, "bpb": {"en": 2.21592, "tr": 2.49724, "fi": 2.6419}, "gini_cpc": 0.051714, "gini_cps": 0.069018, "bpic": 4.751822, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0858558385689683, "fi": 1.2595026985890114}, "t": 194.9} |
| {"event": "eval", "step": 1400, "phase": 1, "bpb": {"en": 2.22501, "tr": 2.50383, "fi": 2.64851}, "bytes_per_chunk": {"en": 4.93, "tr": 5.252, "fi": 4.306}, "bpb_mean": 2.45912, "frozen_inventory": false, "t": 195.7} |
| {"event": "train", "step": 1400, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.60135, "bpb": 2.31025, "chunk_loss": 0.030544, "F": 0.20572, "G": 0.22156, "chunks": 210.66, "bytes_per_chunk": 4.861, "boundary_entropy": 0.19151, "router_noise": 0.0, "grad_norm": 0.8366, "lr": 0.0002759, "router_trainable": true, "sec_per_step": 0.146, "alpha": 0.03, "t": 196.0} |
| {"event": "train", "step": 1450, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.65216, "bpb": 2.38357, "chunk_loss": 0.030489, "F": 0.20352, "G": 0.22005, "chunks": 208.41, "bytes_per_chunk": 4.913, "boundary_entropy": 0.18575, "router_noise": 0.0, "grad_norm": 0.8584, "lr": 0.0002738, "router_trainable": true, "sec_per_step": 0.117, "alpha": 0.03, "t": 201.9} |
| {"event": "train", "step": 1500, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.66226, "bpb": 2.39814, "chunk_loss": 0.03071, "F": 0.21249, "G": 0.22968, "chunks": 217.59, "bytes_per_chunk": 4.706, "boundary_entropy": 0.18819, "router_noise": 0.0, "grad_norm": 0.8164, "lr": 0.0002715, "router_trainable": true, "sec_per_step": 0.108, "alpha": 0.03, "t": 207.3} |
| {"event": "train", "step": 1550, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.58707, "bpb": 2.28966, "chunk_loss": 0.03048, "F": 0.203, "G": 0.22019, "chunks": 207.88, "bytes_per_chunk": 4.926, "boundary_entropy": 0.18183, "router_noise": 0.0, "grad_norm": 0.919, "lr": 0.0002692, "router_trainable": true, "sec_per_step": 0.121, "alpha": 0.03, "t": 213.3} |
| {"event": "probe", "step": 1600, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3430.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.197092, "cps": 26.796875, "mean_p": 0.216009, "boundary_entropy": 0.183865, "bpb": 2.170624}, "tr": {"chunks": 3893.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.215047, "cps": 30.414062, "mean_p": 0.208941, "boundary_entropy": 0.176085, "bpb": 2.429319}, "fi": {"chunks": 4682.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.248897, "cps": 36.578125, "mean_p": 0.255938, "boundary_entropy": 0.193825, "bpb": 2.586438}}, "bpb": {"en": 2.17062, "tr": 2.42932, "fi": 2.58644}, "gini_cpc": 0.052246, "gini_cps": 0.069527, "bpic": 4.725698, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0910982329974785, "fi": 1.2628434792433496}, "t": 219.8} |
| {"event": "eval", "step": 1600, "phase": 1, "bpb": {"en": 2.18273, "tr": 2.43845, "fi": 2.5932}, "bytes_per_chunk": {"en": 4.914, "tr": 5.218, "fi": 4.219}, "bpb_mean": 2.40479, "frozen_inventory": false, "t": 220.8} |
| {"event": "train", "step": 1600, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.51389, "bpb": 2.18408, "chunk_loss": 0.030588, "F": 0.21027, "G": 0.22386, "chunks": 215.31, "bytes_per_chunk": 4.756, "boundary_entropy": 0.1817, "router_noise": 0.0, "grad_norm": 0.8849, "lr": 0.0002668, "router_trainable": true, "sec_per_step": 0.158, "alpha": 0.03, "t": 221.2} |
| {"event": "train", "step": 1650, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.57827, "bpb": 2.27697, "chunk_loss": 0.030544, "F": 0.20578, "G": 0.22109, "chunks": 210.72, "bytes_per_chunk": 4.86, "boundary_entropy": 0.17887, "router_noise": 0.0, "grad_norm": 0.8304, "lr": 0.0002643, "router_trainable": true, "sec_per_step": 0.103, "alpha": 0.03, "t": 226.4} |
| {"event": "train", "step": 1700, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.53931, "bpb": 2.22075, "chunk_loss": 0.03044, "F": 0.20151, "G": 0.21617, "chunks": 206.34, "bytes_per_chunk": 4.963, "boundary_entropy": 0.17121, "router_noise": 0.0, "grad_norm": 0.8072, "lr": 0.0002618, "router_trainable": true, "sec_per_step": 0.119, "alpha": 0.03, "t": 232.3} |
| {"event": "train", "step": 1750, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.56538, "bpb": 2.25836, "chunk_loss": 0.030592, "F": 0.20993, "G": 0.22328, "chunks": 214.97, "bytes_per_chunk": 4.763, "boundary_entropy": 0.17335, "router_noise": 0.0, "grad_norm": 0.7719, "lr": 0.0002591, "router_trainable": true, "sec_per_step": 0.116, "alpha": 0.03, "t": 238.1} |
| {"event": "probe", "step": 1800, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3364.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.1933, "cps": 26.28125, "mean_p": 0.211644, "boundary_entropy": 0.171123, "bpb": 2.13561}, "tr": {"chunks": 3869.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.213721, "cps": 30.226562, "mean_p": 0.205742, "boundary_entropy": 0.162497, "bpb": 2.373851}, "fi": {"chunks": 4532.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.240923, "cps": 35.40625, "mean_p": 0.251981, "boundary_entropy": 0.186995, "bpb": 2.5373}}, "bpb": {"en": 2.13561, "tr": 2.37385, "fi": 2.5373}, "gini_cpc": 0.048999, "gini_cps": 0.066185, "bpic": 4.822099, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.105646540475912, "fi": 1.246367600205587}, "t": 244.5} |
| {"event": "eval", "step": 1800, "phase": 1, "bpb": {"en": 2.14793, "tr": 2.37943, "fi": 2.54302}, "bytes_per_chunk": {"en": 4.964, "tr": 5.24, "fi": 4.488}, "bpb_mean": 2.35679, "frozen_inventory": false, "t": 245.4} |
| {"event": "train", "step": 1800, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.53945, "bpb": 2.22096, "chunk_loss": 0.030464, "F": 0.20206, "G": 0.2157, "chunks": 206.91, "bytes_per_chunk": 4.949, "boundary_entropy": 0.17106, "router_noise": 0.0, "grad_norm": 0.7639, "lr": 0.0002564, "router_trainable": true, "sec_per_step": 0.153, "alpha": 0.03, "t": 245.8} |
| {"event": "train", "step": 1850, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.55157, "bpb": 2.23844, "chunk_loss": 0.030489, "F": 0.20224, "G": 0.21628, "chunks": 207.09, "bytes_per_chunk": 4.945, "boundary_entropy": 0.16558, "router_noise": 0.0, "grad_norm": 0.7557, "lr": 0.0002537, "router_trainable": true, "sec_per_step": 0.119, "alpha": 0.03, "t": 251.7} |
| {"event": "train", "step": 1900, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.53634, "bpb": 2.21647, "chunk_loss": 0.030583, "F": 0.20581, "G": 0.21974, "chunks": 210.75, "bytes_per_chunk": 4.859, "boundary_entropy": 0.16812, "router_noise": 0.0, "grad_norm": 0.845, "lr": 0.0002508, "router_trainable": true, "sec_per_step": 0.128, "alpha": 0.03, "t": 258.1} |
| {"event": "train", "step": 1950, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.48635, "bpb": 2.14435, "chunk_loss": 0.030503, "F": 0.20459, "G": 0.21765, "chunks": 209.5, "bytes_per_chunk": 4.888, "boundary_entropy": 0.16353, "router_noise": 0.0, "grad_norm": 0.7828, "lr": 0.0002479, "router_trainable": true, "sec_per_step": 0.119, "alpha": 0.03, "t": 264.0} |
| {"event": "probe", "step": 2000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3460.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.198816, "cps": 27.03125, "mean_p": 0.21265, "boundary_entropy": 0.155691, "bpb": 2.107264}, "tr": {"chunks": 3885.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.214605, "cps": 30.351562, "mean_p": 0.206262, "boundary_entropy": 0.160445, "bpb": 2.319854}, "fi": {"chunks": 4610.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.245069, "cps": 36.015625, "mean_p": 0.25435, "boundary_entropy": 0.172734, "bpb": 2.482467}}, "bpb": {"en": 2.10726, "tr": 2.31985, "fi": 2.48247}, "gini_cpc": 0.046827, "gini_cps": 0.064129, "bpic": 4.745462, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0794151098770395, "fi": 1.2326422893012727}, "t": 270.1} |
| {"event": "eval", "step": 2000, "phase": 1, "bpb": {"en": 2.1191, "tr": 2.32877, "fi": 2.49157}, "bytes_per_chunk": {"en": 4.896, "tr": 5.332, "fi": 4.453}, "bpb_mean": 2.31315, "frozen_inventory": false, "t": 270.9} |
| {"event": "train", "step": 2000, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.64821, "bpb": 2.37787, "chunk_loss": 0.030645, "F": 0.21167, "G": 0.2248, "chunks": 216.75, "bytes_per_chunk": 4.724, "boundary_entropy": 0.163, "router_noise": 0.0, "grad_norm": 0.8335, "lr": 0.000245, "router_trainable": true, "sec_per_step": 0.144, "alpha": 0.03, "t": 271.2} |
| {"event": "probe", "step": 2048, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3434.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.197322, "cps": 26.828125, "mean_p": 0.211993, "boundary_entropy": 0.155811, "bpb": 2.098084}, "tr": {"chunks": 3874.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.213998, "cps": 30.265625, "mean_p": 0.206497, "boundary_entropy": 0.161821, "bpb": 2.316549}, "fi": {"chunks": 4537.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.241189, "cps": 35.445312, "mean_p": 0.251423, "boundary_entropy": 0.174464, "bpb": 2.472092}}, "bpb": {"en": 2.09808, "tr": 2.31655, "fi": 2.47209}, "gini_cpc": 0.044818, "gini_cps": 0.06208, "bpic": 4.789531, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0845083354805516, "fi": 1.222308199761803}, "t": 276.5} |
| {"event": "train", "step": 2050, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.48071, "bpb": 2.13621, "chunk_loss": 0.030402, "F": 0.1972, "G": 0.20942, "chunks": 201.94, "bytes_per_chunk": 5.071, "boundary_entropy": 0.1566, "router_noise": 0.0, "grad_norm": 0.7902, "lr": 0.0002419, "router_trainable": true, "sec_per_step": 0.115, "alpha": 0.03, "t": 277.0} |
| {"event": "train", "step": 2100, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.54914, "bpb": 2.23493, "chunk_loss": 0.03035, "F": 0.19614, "G": 0.21128, "chunks": 200.84, "bytes_per_chunk": 5.098, "boundary_entropy": 0.15766, "router_noise": 0.0, "grad_norm": 0.8001, "lr": 0.0002388, "router_trainable": true, "sec_per_step": 0.12, "alpha": 0.03, "t": 283.0} |
| {"event": "train", "step": 2150, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.49757, "bpb": 2.16053, "chunk_loss": 0.030524, "F": 0.20615, "G": 0.21933, "chunks": 211.09, "bytes_per_chunk": 4.851, "boundary_entropy": 0.16101, "router_noise": 0.0, "grad_norm": 0.7439, "lr": 0.0002357, "router_trainable": true, "sec_per_step": 0.144, "alpha": 0.03, "t": 290.2} |
| {"event": "probe", "step": 2200, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3421.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.196575, "cps": 26.726562, "mean_p": 0.210017, "boundary_entropy": 0.149778, "bpb": 2.08569}, "tr": {"chunks": 3902.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.215544, "cps": 30.484375, "mean_p": 0.207356, "boundary_entropy": 0.147598, "bpb": 2.29809}, "fi": {"chunks": 4609.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.245016, "cps": 36.007812, "mean_p": 0.249239, "boundary_entropy": 0.167157, "bpb": 2.442158}}, "bpb": {"en": 2.08569, "tr": 2.29809, "fi": 2.44216}, "gini_cpc": 0.049143, "gini_cps": 0.066376, "bpic": 4.754609, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0964977873615886, "fi": 1.2464241949150296}, "t": 296.2} |
| {"event": "eval", "step": 2200, "phase": 1, "bpb": {"en": 2.09512, "tr": 2.30381, "fi": 2.44838}, "bytes_per_chunk": {"en": 4.947, "tr": 5.269, "fi": 4.407}, "bpb_mean": 2.28244, "frozen_inventory": false, "t": 296.8} |
| {"event": "train", "step": 2200, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.47645, "bpb": 2.13006, "chunk_loss": 0.03045, "F": 0.2034, "G": 0.21475, "chunks": 208.28, "bytes_per_chunk": 4.916, "boundary_entropy": 0.14832, "router_noise": 0.0, "grad_norm": 0.6807, "lr": 0.0002325, "router_trainable": true, "sec_per_step": 0.139, "alpha": 0.03, "t": 297.2} |
| {"event": "train", "step": 2250, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.49717, "bpb": 2.15995, "chunk_loss": 0.030488, "F": 0.20544, "G": 0.2163, "chunks": 210.38, "bytes_per_chunk": 4.867, "boundary_entropy": 0.14588, "router_noise": 0.0, "grad_norm": 0.7174, "lr": 0.0002293, "router_trainable": true, "sec_per_step": 0.12, "alpha": 0.03, "t": 303.2} |
| {"event": "train", "step": 2300, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.5152, "bpb": 2.18598, "chunk_loss": 0.030455, "F": 0.2038, "G": 0.21441, "chunks": 208.69, "bytes_per_chunk": 4.907, "boundary_entropy": 0.14263, "router_noise": 0.0, "grad_norm": 0.7381, "lr": 0.000226, "router_trainable": true, "sec_per_step": 0.13, "alpha": 0.03, "t": 309.7} |
| {"event": "train", "step": 2350, "arch": "hnet", "phase": 1, "objective": "global", "lm_loss": 1.56273, "bpb": 2.25454, "chunk_loss": 0.03047, "F": 0.20474, "G": 0.21491, "chunks": 209.66, "bytes_per_chunk": 4.884, "boundary_entropy": 0.14174, "router_noise": 0.0, "grad_norm": 0.7327, "lr": 0.0002226, "router_trainable": true, "sec_per_step": 0.127, "alpha": 0.03, "t": 316.0} |
| {"event": "probe", "step": 2400, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3385.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.194507, "cps": 26.445312, "mean_p": 0.206716, "boundary_entropy": 0.142921, "bpb": 2.064527}, "tr": {"chunks": 3732.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.206154, "cps": 29.15625, "mean_p": 0.197861, "boundary_entropy": 0.139203, "bpb": 2.265505}, "fi": {"chunks": 4605.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.244804, "cps": 35.976562, "mean_p": 0.248043, "boundary_entropy": 0.146546, "bpb": 2.392872}}, "bpb": {"en": 2.06453, "tr": 2.2655, "fi": 2.39287}, "gini_cpc": 0.051949, "gini_cps": 0.069385, "bpic": 4.839788, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0598795942893982, "fi": 1.2585868744732547}, "t": 322.8} |
| {"event": "eval", "step": 2400, "phase": 1, "bpb": {"en": 2.07485, "tr": 2.27253, "fi": 2.40184}, "bytes_per_chunk": {"en": 5.03, "tr": 5.42, "fi": 4.456}, "bpb_mean": 2.24974, "frozen_inventory": false, "t": 323.2} |
| {"event": "phase_transition", "step": 2400, "phase": 2, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0, "t": 323.5} |
| {"event": "train", "step": 2400, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.4993, "bpb": 2.16304, "chunk_loss": 0.030489, "F": 0.20526, "G": 0.21525, "chunks": 210.19, "bytes_per_chunk": 4.872, "boundary_entropy": 0.13794, "router_noise": 0.0, "grad_norm": 0.7308, "lr": 0.0002192, "router_trainable": true, "sec_per_step": 0.152, "alpha": 0.03, "t": 323.6} |
| {"event": "train", "step": 2450, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.46523, "bpb": 2.11389, "chunk_loss": 0.03039, "F": 0.19952, "G": 0.2091, "chunks": 204.31, "bytes_per_chunk": 5.012, "boundary_entropy": 0.13635, "router_noise": 0.0, "grad_norm": 0.6898, "lr": 0.0002158, "router_trainable": true, "sec_per_step": 0.119, "alpha": 0.03, "t": 329.6} |
| {"event": "train", "step": 2500, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.43756, "bpb": 2.07397, "chunk_loss": 0.030453, "F": 0.20468, "G": 0.21436, "chunks": 209.59, "bytes_per_chunk": 4.886, "boundary_entropy": 0.13527, "router_noise": 0.0, "grad_norm": 0.7271, "lr": 0.0002123, "router_trainable": true, "sec_per_step": 0.117, "alpha": 0.03, "t": 335.4} |
| {"event": "train", "step": 2550, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.467, "bpb": 2.11643, "chunk_loss": 0.030435, "F": 0.20285, "G": 0.21134, "chunks": 207.72, "bytes_per_chunk": 4.93, "boundary_entropy": 0.13, "router_noise": 0.0, "grad_norm": 0.7385, "lr": 0.0002088, "router_trainable": true, "sec_per_step": 0.113, "alpha": 0.03, "t": 341.1} |
| {"event": "probe", "step": 2600, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3402.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.195484, "cps": 26.578125, "mean_p": 0.206418, "boundary_entropy": 0.133967, "bpb": 2.046118}, "tr": {"chunks": 3865.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.213501, "cps": 30.195312, "mean_p": 0.205562, "boundary_entropy": 0.133648, "bpb": 2.241934}, "fi": {"chunks": 4686.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.24911, "cps": 36.609375, "mean_p": 0.249116, "boundary_entropy": 0.135722, "bpb": 2.369061}}, "bpb": {"en": 2.04612, "tr": 2.24193, "fi": 2.36906}, "gini_cpc": 0.054325, "gini_cps": 0.071614, "bpic": 4.746256, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.092166264743554, "fi": 1.2743250248433386}, "t": 346.9} |
| {"event": "eval", "step": 2600, "phase": 2, "bpb": {"en": 2.05687, "tr": 2.25378, "fi": 2.37549}, "bytes_per_chunk": {"en": 4.989, "tr": 5.275, "fi": 4.385}, "bpb_mean": 2.22871, "frozen_inventory": false, "t": 347.8} |
| {"event": "train", "step": 2600, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.42354, "bpb": 2.05374, "chunk_loss": 0.030359, "F": 0.20169, "G": 0.20901, "chunks": 206.53, "bytes_per_chunk": 4.958, "boundary_entropy": 0.12805, "router_noise": 0.0, "grad_norm": 0.7675, "lr": 0.0002053, "router_trainable": true, "sec_per_step": 0.141, "alpha": 0.03, "t": 348.2} |
| {"event": "train", "step": 2650, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.46668, "bpb": 2.11597, "chunk_loss": 0.030322, "F": 0.19785, "G": 0.20542, "chunks": 202.59, "bytes_per_chunk": 5.054, "boundary_entropy": 0.12629, "router_noise": 0.0, "grad_norm": 0.6894, "lr": 0.0002017, "router_trainable": true, "sec_per_step": 0.125, "alpha": 0.03, "t": 354.4} |
| {"event": "train", "step": 2700, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.45531, "bpb": 2.09957, "chunk_loss": 0.030442, "F": 0.20456, "G": 0.21235, "chunks": 209.47, "bytes_per_chunk": 4.889, "boundary_entropy": 0.12481, "router_noise": 0.0, "grad_norm": 0.7145, "lr": 0.0001981, "router_trainable": true, "sec_per_step": 0.126, "alpha": 0.03, "t": 360.7} |
| {"event": "train", "step": 2750, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.45681, "bpb": 2.10173, "chunk_loss": 0.030394, "F": 0.20175, "G": 0.21017, "chunks": 206.59, "bytes_per_chunk": 4.957, "boundary_entropy": 0.12948, "router_noise": 0.0, "grad_norm": 0.6922, "lr": 0.0001945, "router_trainable": true, "sec_per_step": 0.122, "alpha": 0.03, "t": 366.8} |
| {"event": "probe", "step": 2800, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3410.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.195943, "cps": 26.640625, "mean_p": 0.204813, "boundary_entropy": 0.125617, "bpb": 2.026296}, "tr": {"chunks": 3835.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.211843, "cps": 29.960938, "mean_p": 0.201751, "boundary_entropy": 0.130183, "bpb": 2.202853}, "fi": {"chunks": 4595.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.244272, "cps": 35.898438, "mean_p": 0.246158, "boundary_entropy": 0.134492, "bpb": 2.329302}}, "bpb": {"en": 2.0263, "tr": 2.20285, "fi": 2.3293}, "gini_cpc": 0.049411, "gini_cps": 0.066723, "bpic": 4.791554, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0811465282645432, "fi": 1.2466466475985616}, "t": 372.8} |
| {"event": "eval", "step": 2800, "phase": 2, "bpb": {"en": 2.03727, "tr": 2.21285, "fi": 2.335}, "bytes_per_chunk": {"en": 5.012, "tr": 5.348, "fi": 4.477}, "bpb_mean": 2.19504, "frozen_inventory": false, "t": 373.7} |
| {"event": "train", "step": 2800, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.41257, "bpb": 2.03791, "chunk_loss": 0.030346, "F": 0.1991, "G": 0.20769, "chunks": 203.88, "bytes_per_chunk": 5.023, "boundary_entropy": 0.12393, "router_noise": 0.0, "grad_norm": 0.6947, "lr": 0.0001909, "router_trainable": true, "sec_per_step": 0.144, "alpha": 0.03, "t": 374.0} |
| {"event": "train", "step": 2850, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.46865, "bpb": 2.11881, "chunk_loss": 0.030507, "F": 0.20807, "G": 0.21662, "chunks": 213.06, "bytes_per_chunk": 4.806, "boundary_entropy": 0.12878, "router_noise": 0.0, "grad_norm": 0.7543, "lr": 0.0001872, "router_trainable": true, "sec_per_step": 0.115, "alpha": 0.03, "t": 379.8} |
| {"event": "train", "step": 2900, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.3943, "bpb": 2.01155, "chunk_loss": 0.0303, "F": 0.19748, "G": 0.20484, "chunks": 202.22, "bytes_per_chunk": 5.064, "boundary_entropy": 0.11967, "router_noise": 0.0, "grad_norm": 0.6459, "lr": 0.0001835, "router_trainable": true, "sec_per_step": 0.122, "alpha": 0.03, "t": 385.8} |
| {"event": "train", "step": 2950, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.54457, "bpb": 2.22835, "chunk_loss": 0.030493, "F": 0.20355, "G": 0.21198, "chunks": 208.44, "bytes_per_chunk": 4.913, "boundary_entropy": 0.12632, "router_noise": 0.0, "grad_norm": 0.7182, "lr": 0.0001799, "router_trainable": true, "sec_per_step": 0.124, "alpha": 0.03, "t": 392.0} |
| {"event": "probe", "step": 3000, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3401.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.195426, "cps": 26.570312, "mean_p": 0.204714, "boundary_entropy": 0.121037, "bpb": 2.006029}, "tr": {"chunks": 3738.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.206485, "cps": 29.203125, "mean_p": 0.196752, "boundary_entropy": 0.124578, "bpb": 2.188362}, "fi": {"chunks": 4591.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.244059, "cps": 35.867188, "mean_p": 0.243909, "boundary_entropy": 0.125309, "bpb": 2.311366}}, "bpb": {"en": 2.00603, "tr": 2.18836, "fi": 2.31137}, "gini_cpc": 0.050191, "gini_cps": 0.067633, "bpic": 4.836488, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0565893622242601, "fi": 1.2488575323724627}, "t": 397.3} |
| {"event": "eval", "step": 3000, "phase": 2, "bpb": {"en": 2.01859, "tr": 2.19813, "fi": 2.31724}, "bytes_per_chunk": {"en": 4.989, "tr": 5.461, "fi": 4.5}, "bpb_mean": 2.17799, "frozen_inventory": false, "t": 397.7} |
| {"event": "train", "step": 3000, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.49171, "bpb": 2.15208, "chunk_loss": 0.030328, "F": 0.19684, "G": 0.20552, "chunks": 201.56, "bytes_per_chunk": 5.08, "boundary_entropy": 0.12069, "router_noise": 0.0, "grad_norm": 0.7133, "lr": 0.0001761, "router_trainable": true, "sec_per_step": 0.118, "alpha": 0.03, "t": 397.9} |
| {"event": "train", "step": 3050, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.44215, "bpb": 2.08059, "chunk_loss": 0.030376, "F": 0.19888, "G": 0.20664, "chunks": 203.66, "bytes_per_chunk": 5.028, "boundary_entropy": 0.12223, "router_noise": 0.0, "grad_norm": 0.6918, "lr": 0.0001724, "router_trainable": true, "sec_per_step": 0.1, "alpha": 0.03, "t": 402.9} |
| {"event": "train", "step": 3100, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.48406, "bpb": 2.14104, "chunk_loss": 0.030328, "F": 0.19739, "G": 0.20457, "chunks": 202.12, "bytes_per_chunk": 5.066, "boundary_entropy": 0.12133, "router_noise": 0.0, "grad_norm": 0.6806, "lr": 0.0001687, "router_trainable": true, "sec_per_step": 0.078, "alpha": 0.03, "t": 406.8} |
| {"event": "train", "step": 3150, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.44941, "bpb": 2.09106, "chunk_loss": 0.030374, "F": 0.20096, "G": 0.20717, "chunks": 205.78, "bytes_per_chunk": 4.976, "boundary_entropy": 0.11828, "router_noise": 0.0, "grad_norm": 0.6711, "lr": 0.000165, "router_trainable": true, "sec_per_step": 0.076, "alpha": 0.03, "t": 410.6} |
| {"event": "probe", "step": 3200, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3399.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.195311, "cps": 26.554688, "mean_p": 0.204962, "boundary_entropy": 0.115141, "bpb": 1.993088}, "tr": {"chunks": 3750.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.207148, "cps": 29.296875, "mean_p": 0.197195, "boundary_entropy": 0.117334, "bpb": 2.164139}, "fi": {"chunks": 4593.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.244166, "cps": 35.882812, "mean_p": 0.243423, "boundary_entropy": 0.120114, "bpb": 2.285499}}, "bpb": {"en": 1.99309, "tr": 2.16414, "fi": 2.2855}, "gini_cpc": 0.050369, "gini_cps": 0.067791, "bpic": 4.831545, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.060605004246808, "fi": 1.2501367366739982}, "t": 414.5} |
| {"event": "eval", "step": 3200, "phase": 2, "bpb": {"en": 2.00374, "tr": 2.17201, "fi": 2.29437}, "bytes_per_chunk": {"en": 4.998, "tr": 5.531, "fi": 4.49}, "bpb_mean": 2.1567, "frozen_inventory": false, "t": 414.9} |
| {"event": "train", "step": 3200, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.43575, "bpb": 2.07135, "chunk_loss": 0.030339, "F": 0.19928, "G": 0.20709, "chunks": 204.06, "bytes_per_chunk": 5.018, "boundary_entropy": 0.11277, "router_noise": 0.0, "grad_norm": 0.6224, "lr": 0.0001613, "router_trainable": true, "sec_per_step": 0.091, "alpha": 0.03, "t": 415.2} |
| {"event": "train", "step": 3250, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.43471, "bpb": 2.06985, "chunk_loss": 0.030465, "F": 0.20377, "G": 0.21014, "chunks": 208.66, "bytes_per_chunk": 4.908, "boundary_entropy": 0.10931, "router_noise": 0.0, "grad_norm": 0.7216, "lr": 0.0001576, "router_trainable": true, "sec_per_step": 0.059, "alpha": 0.03, "t": 418.1} |
| {"event": "train", "step": 3300, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.41314, "bpb": 2.03873, "chunk_loss": 0.030439, "F": 0.2038, "G": 0.2118, "chunks": 208.69, "bytes_per_chunk": 4.907, "boundary_entropy": 0.11364, "router_noise": 0.0, "grad_norm": 0.6704, "lr": 0.0001539, "router_trainable": true, "sec_per_step": 0.068, "alpha": 0.03, "t": 421.5} |
| {"event": "train", "step": 3350, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.35943, "bpb": 1.96124, "chunk_loss": 0.030394, "F": 0.20154, "G": 0.20793, "chunks": 206.38, "bytes_per_chunk": 4.962, "boundary_entropy": 0.10797, "router_noise": 0.0, "grad_norm": 0.7134, "lr": 0.0001501, "router_trainable": true, "sec_per_step": 0.069, "alpha": 0.03, "t": 424.9} |
| {"event": "probe", "step": 3400, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3404.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.195598, "cps": 26.59375, "mean_p": 0.202835, "boundary_entropy": 0.110183, "bpb": 1.980577}, "tr": {"chunks": 3590.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.19831, "cps": 28.046875, "mean_p": 0.189518, "boundary_entropy": 0.118426, "bpb": 2.16309}, "fi": {"chunks": 4615.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.245335, "cps": 36.054688, "mean_p": 0.242272, "boundary_entropy": 0.113176, "bpb": 2.254658}}, "bpb": {"en": 1.98058, "tr": 2.16309, "fi": 2.25466}, "gini_cpc": 0.05187, "gini_cps": 0.069544, "bpic": 4.886898, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0138611131900739, "fi": 1.2542796920895536}, "t": 428.8} |
| {"event": "eval", "step": 3400, "phase": 2, "bpb": {"en": 1.99169, "tr": 2.17135, "fi": 2.26243}, "bytes_per_chunk": {"en": 5.03, "tr": 5.735, "fi": 4.428}, "bpb_mean": 2.14182, "frozen_inventory": false, "t": 429.3} |
| {"event": "train", "step": 3400, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.4233, "bpb": 2.05338, "chunk_loss": 0.030369, "F": 0.19824, "G": 0.205, "chunks": 203.0, "bytes_per_chunk": 5.044, "boundary_entropy": 0.10959, "router_noise": 0.0, "grad_norm": 0.7279, "lr": 0.0001465, "router_trainable": true, "sec_per_step": 0.093, "alpha": 0.03, "t": 429.6} |
| {"event": "train", "step": 3450, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.38815, "bpb": 2.00268, "chunk_loss": 0.030338, "F": 0.19971, "G": 0.20774, "chunks": 204.5, "bytes_per_chunk": 5.007, "boundary_entropy": 0.10753, "router_noise": 0.0, "grad_norm": 0.7458, "lr": 0.0001428, "router_trainable": true, "sec_per_step": 0.064, "alpha": 0.03, "t": 432.8} |
| {"event": "train", "step": 3500, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.36677, "bpb": 1.97184, "chunk_loss": 0.030236, "F": 0.19171, "G": 0.2003, "chunks": 196.31, "bytes_per_chunk": 5.216, "boundary_entropy": 0.10774, "router_noise": 0.0, "grad_norm": 0.6974, "lr": 0.0001391, "router_trainable": true, "sec_per_step": 0.067, "alpha": 0.03, "t": 436.2} |
| {"event": "train", "step": 3550, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.40749, "bpb": 2.03058, "chunk_loss": 0.030352, "F": 0.19885, "G": 0.20541, "chunks": 203.62, "bytes_per_chunk": 5.029, "boundary_entropy": 0.10981, "router_noise": 0.0, "grad_norm": 0.6813, "lr": 0.0001355, "router_trainable": true, "sec_per_step": 0.058, "alpha": 0.03, "t": 439.1} |
| {"event": "probe", "step": 3600, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3393.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.194966, "cps": 26.507812, "mean_p": 0.203763, "boundary_entropy": 0.112364, "bpb": 1.96549}, "tr": {"chunks": 3752.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.207258, "cps": 29.3125, "mean_p": 0.195313, "boundary_entropy": 0.105193, "bpb": 2.122025}, "fi": {"chunks": 4592.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.244112, "cps": 35.875, "mean_p": 0.241896, "boundary_entropy": 0.110037, "bpb": 2.233493}}, "bpb": {"en": 1.96549, "tr": 2.12203, "fi": 2.23349}, "gini_cpc": 0.050692, "gini_cps": 0.068104, "bpic": 4.833603, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.063047177773828, "fi": 1.2520747473553884}, "t": 442.5} |
| {"event": "eval", "step": 3600, "phase": 2, "bpb": {"en": 1.97702, "tr": 2.12946, "fi": 2.24178}, "bytes_per_chunk": {"en": 5.0, "tr": 5.519, "fi": 4.47}, "bpb_mean": 2.11609, "frozen_inventory": false, "t": 443.0} |
| {"event": "train", "step": 3600, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.40715, "bpb": 2.03008, "chunk_loss": 0.030397, "F": 0.20181, "G": 0.2087, "chunks": 206.66, "bytes_per_chunk": 4.955, "boundary_entropy": 0.11124, "router_noise": 0.0, "grad_norm": 0.6338, "lr": 0.0001319, "router_trainable": true, "sec_per_step": 0.09, "alpha": 0.03, "t": 443.6} |
| {"event": "train", "step": 3650, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.44124, "bpb": 2.07927, "chunk_loss": 0.030487, "F": 0.20535, "G": 0.21105, "chunks": 210.28, "bytes_per_chunk": 4.87, "boundary_entropy": 0.10415, "router_noise": 0.0, "grad_norm": 0.6961, "lr": 0.0001283, "router_trainable": true, "sec_per_step": 0.094, "alpha": 0.03, "t": 448.3} |
| {"event": "train", "step": 3700, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.39817, "bpb": 2.01713, "chunk_loss": 0.030365, "F": 0.2009, "G": 0.20728, "chunks": 205.72, "bytes_per_chunk": 4.978, "boundary_entropy": 0.1048, "router_noise": 0.0, "grad_norm": 0.6782, "lr": 0.0001247, "router_trainable": true, "sec_per_step": 0.065, "alpha": 0.03, "t": 451.5} |
| {"event": "train", "step": 3750, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.37536, "bpb": 1.98422, "chunk_loss": 0.030328, "F": 0.19843, "G": 0.20386, "chunks": 203.19, "bytes_per_chunk": 5.04, "boundary_entropy": 0.1032, "router_noise": 0.0, "grad_norm": 0.698, "lr": 0.0001212, "router_trainable": true, "sec_per_step": 0.058, "alpha": 0.03, "t": 454.4} |
| {"event": "probe", "step": 3800, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3390.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.194794, "cps": 26.484375, "mean_p": 0.201632, "boundary_entropy": 0.104735, "bpb": 1.952523}, "tr": {"chunks": 3704.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.204607, "cps": 28.9375, "mean_p": 0.192227, "boundary_entropy": 0.104755, "bpb": 2.114201}, "fi": {"chunks": 4576.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.243262, "cps": 35.75, "mean_p": 0.240885, "boundary_entropy": 0.106781, "bpb": 2.214584}}, "bpb": {"en": 1.95252, "tr": 2.1142, "fi": 2.21458}, "gini_cpc": 0.050278, "gini_cps": 0.067752, "bpic": 4.861354, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0503761416359387, "fi": 1.2488162875892141}, "t": 457.8} |
| {"event": "eval", "step": 3800, "phase": 2, "bpb": {"en": 1.96498, "tr": 2.12242, "fi": 2.22346}, "bytes_per_chunk": {"en": 5.014, "tr": 5.59, "fi": 4.513}, "bpb_mean": 2.10362, "frozen_inventory": false, "t": 458.2} |
| {"event": "train", "step": 3800, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.44958, "bpb": 2.0913, "chunk_loss": 0.030434, "F": 0.20505, "G": 0.20944, "chunks": 209.97, "bytes_per_chunk": 4.877, "boundary_entropy": 0.10181, "router_noise": 0.0, "grad_norm": 0.7364, "lr": 0.0001177, "router_trainable": true, "sec_per_step": 0.08, "alpha": 0.03, "t": 458.4} |
| {"event": "train", "step": 3850, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.54474, "bpb": 2.22859, "chunk_loss": 0.030622, "F": 0.20422, "G": 0.20909, "chunks": 209.12, "bytes_per_chunk": 4.897, "boundary_entropy": 0.1002, "router_noise": 0.0, "grad_norm": 0.7304, "lr": 0.0001142, "router_trainable": true, "sec_per_step": 0.052, "alpha": 0.03, "t": 461.0} |
| {"event": "train", "step": 3900, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.4347, "bpb": 2.06983, "chunk_loss": 0.030364, "F": 0.20065, "G": 0.20674, "chunks": 205.47, "bytes_per_chunk": 4.984, "boundary_entropy": 0.10502, "router_noise": 0.0, "grad_norm": 0.746, "lr": 0.0001108, "router_trainable": true, "sec_per_step": 0.063, "alpha": 0.03, "t": 464.2} |
| {"event": "train", "step": 3950, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.49011, "bpb": 2.14977, "chunk_loss": 0.030415, "F": 0.1973, "G": 0.20273, "chunks": 202.03, "bytes_per_chunk": 5.069, "boundary_entropy": 0.09837, "router_noise": 0.0, "grad_norm": 0.7873, "lr": 0.0001074, "router_trainable": true, "sec_per_step": 0.064, "alpha": 0.03, "t": 467.4} |
| {"event": "probe", "step": 4000, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3389.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.194737, "cps": 26.476562, "mean_p": 0.201659, "boundary_entropy": 0.101772, "bpb": 1.941395}, "tr": {"chunks": 3720.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.205491, "cps": 29.0625, "mean_p": 0.1958, "boundary_entropy": 0.10137, "bpb": 2.098274}, "fi": {"chunks": 4581.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.243528, "cps": 35.789062, "mean_p": 0.241289, "boundary_entropy": 0.102194, "bpb": 2.190847}}, "bpb": {"en": 1.9414, "tr": 2.09827, "fi": 2.19085}, "gini_cpc": 0.050528, "gini_cps": 0.067978, "bpic": 4.853037, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0552246793034585, "fi": 1.2505497095951232}, "t": 471.5} |
| {"event": "eval", "step": 4000, "phase": 2, "bpb": {"en": 1.95259, "tr": 2.10616, "fi": 2.1987}, "bytes_per_chunk": {"en": 5.018, "tr": 5.53, "fi": 4.471}, "bpb_mean": 2.08582, "frozen_inventory": false, "t": 471.8} |
| {"event": "train", "step": 4000, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.41212, "bpb": 2.03726, "chunk_loss": 0.03041, "F": 0.2019, "G": 0.20784, "chunks": 206.75, "bytes_per_chunk": 4.953, "boundary_entropy": 0.10162, "router_noise": 0.0, "grad_norm": 0.6485, "lr": 0.000104, "router_trainable": true, "sec_per_step": 0.094, "alpha": 0.03, "t": 472.1} |
| {"event": "train", "step": 4050, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.42008, "bpb": 2.04874, "chunk_loss": 0.030356, "F": 0.1991, "G": 0.20553, "chunks": 203.88, "bytes_per_chunk": 5.023, "boundary_entropy": 0.10016, "router_noise": 0.0, "grad_norm": 0.7153, "lr": 0.0001007, "router_trainable": true, "sec_per_step": 0.059, "alpha": 0.03, "t": 475.0} |
| {"event": "probe", "step": 4096, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3417.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.196345, "cps": 26.695312, "mean_p": 0.202353, "boundary_entropy": 0.096436, "bpb": 1.934655}, "tr": {"chunks": 3719.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.205436, "cps": 29.054688, "mean_p": 0.193564, "boundary_entropy": 0.100135, "bpb": 2.091905}, "fi": {"chunks": 4593.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.244166, "cps": 35.882812, "mean_p": 0.241755, "boundary_entropy": 0.098287, "bpb": 2.188656}}, "bpb": {"en": 1.93465, "tr": 2.0919, "fi": 2.18866}, "gini_cpc": 0.049354, "gini_cps": 0.066843, "bpic": 4.8369, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0462964898400853, "fi": 1.2435512929338366}, "t": 478.2} |
| {"event": "train", "step": 4100, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.36028, "bpb": 1.96247, "chunk_loss": 0.030392, "F": 0.19971, "G": 0.20442, "chunks": 204.5, "bytes_per_chunk": 5.007, "boundary_entropy": 0.09502, "router_noise": 0.0, "grad_norm": 0.7055, "lr": 9.75e-05, "router_trainable": true, "sec_per_step": 0.074, "alpha": 0.03, "t": 478.7} |
| {"event": "train", "step": 4150, "arch": "hnet", "phase": 2, "objective": "global", "lm_loss": 1.41872, "bpb": 2.04678, "chunk_loss": 0.030494, "F": 0.20569, "G": 0.21135, "chunks": 210.62, "bytes_per_chunk": 4.862, "boundary_entropy": 0.09902, "router_noise": 0.0, "grad_norm": 0.7461, "lr": 9.43e-05, "router_trainable": true, "sec_per_step": 0.07, "alpha": 0.03, "t": 482.2} |
| {"event": "probe", "step": 4200, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3401.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.195426, "cps": 26.570312, "mean_p": 0.20226, "boundary_entropy": 0.09618, "bpb": 1.933438}, "tr": {"chunks": 3747.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.206982, "cps": 29.273438, "mean_p": 0.1963, "boundary_entropy": 0.09792, "bpb": 2.083043}, "fi": {"chunks": 4620.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.245601, "cps": 36.09375, "mean_p": 0.242398, "boundary_entropy": 0.099016, "bpb": 2.18476}}, "bpb": {"en": 1.93344, "tr": 2.08304, "fi": 2.18476}, "gini_cpc": 0.05162, "gini_cps": 0.069057, "bpic": 4.82087, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0591333173500008, "fi": 1.2567461989894964}, "t": 485.6} |
| {"event": "eval", "step": 4200, "phase": 2, "bpb": {"en": 1.94541, "tr": 2.09154, "fi": 2.19388}, "bytes_per_chunk": {"en": 5.009, "tr": 5.52, "fi": 4.491}, "bpb_mean": 2.07694, "frozen_inventory": false, "t": 485.9} |
| {"event": "phase_transition", "step": 4200, "phase": 3, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0, "t": 486.1} |
| {"event": "train", "step": 4200, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.35445, "bpb": 1.95405, "chunk_loss": 0.030266, "F": 0.19583, "G": 0.20066, "chunks": 200.53, "bytes_per_chunk": 5.106, "boundary_entropy": 0.09172, "router_noise": 0.0, "grad_norm": 0.6943, "lr": 9.12e-05, "router_trainable": true, "sec_per_step": 0.079, "alpha": 0.03, "t": 486.2} |
| {"event": "train", "step": 4250, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.38634, "bpb": 2.00007, "chunk_loss": 0.030456, "F": 0.20526, "G": 0.21026, "chunks": 210.19, "bytes_per_chunk": 4.872, "boundary_entropy": 0.09368, "router_noise": 0.0, "grad_norm": 0.6897, "lr": 8.81e-05, "router_trainable": true, "sec_per_step": 0.103, "alpha": 0.03, "t": 491.3} |
| {"event": "train", "step": 4300, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.38602, "bpb": 1.99961, "chunk_loss": 0.030288, "F": 0.19379, "G": 0.19882, "chunks": 198.44, "bytes_per_chunk": 5.16, "boundary_entropy": 0.09668, "router_noise": 0.0, "grad_norm": 0.7668, "lr": 8.5e-05, "router_trainable": true, "sec_per_step": 0.084, "alpha": 0.03, "t": 495.5} |
| {"event": "train", "step": 4350, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.40588, "bpb": 2.02826, "chunk_loss": 0.030354, "F": 0.20026, "G": 0.2054, "chunks": 205.06, "bytes_per_chunk": 4.994, "boundary_entropy": 0.09503, "router_noise": 0.0, "grad_norm": 0.6785, "lr": 8.21e-05, "router_trainable": true, "sec_per_step": 0.066, "alpha": 0.03, "t": 498.8} |
| {"event": "probe", "step": 4400, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3416.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.196288, "cps": 26.6875, "mean_p": 0.201085, "boundary_entropy": 0.093018, "bpb": 1.921372}, "tr": {"chunks": 3697.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.20422, "cps": 28.882812, "mean_p": 0.192512, "boundary_entropy": 0.094076, "bpb": 2.067503}, "fi": {"chunks": 4581.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.243528, "cps": 35.789062, "mean_p": 0.240845, "boundary_entropy": 0.098335, "bpb": 2.166786}}, "bpb": {"en": 1.92137, "tr": 2.0675, "fi": 2.16679}, "gini_cpc": 0.0489, "gini_cps": 0.066416, "bpic": 4.851377, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0404115320593932, "fi": 1.2406653881199863}, "t": 502.6} |
| {"event": "eval", "step": 4400, "phase": 3, "bpb": {"en": 1.93341, "tr": 2.07589, "fi": 2.17627}, "bytes_per_chunk": {"en": 4.995, "tr": 5.636, "fi": 4.498}, "bpb_mean": 2.06186, "frozen_inventory": false, "t": 502.9} |
| {"event": "train", "step": 4400, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.36298, "bpb": 1.96637, "chunk_loss": 0.030308, "F": 0.19995, "G": 0.20417, "chunks": 204.75, "bytes_per_chunk": 5.001, "boundary_entropy": 0.09162, "router_noise": 0.0, "grad_norm": 0.6514, "lr": 7.92e-05, "router_trainable": true, "sec_per_step": 0.089, "alpha": 0.03, "t": 503.2} |
| {"event": "train", "step": 4450, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.35628, "bpb": 1.95669, "chunk_loss": 0.030228, "F": 0.19153, "G": 0.19642, "chunks": 196.12, "bytes_per_chunk": 5.221, "boundary_entropy": 0.09161, "router_noise": 0.0, "grad_norm": 0.6922, "lr": 7.63e-05, "router_trainable": true, "sec_per_step": 0.054, "alpha": 0.03, "t": 505.9} |
| {"event": "train", "step": 4500, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.33323, "bpb": 1.92345, "chunk_loss": 0.030361, "F": 0.20215, "G": 0.20668, "chunks": 207.0, "bytes_per_chunk": 4.947, "boundary_entropy": 0.09203, "router_noise": 0.0, "grad_norm": 0.6527, "lr": 7.36e-05, "router_trainable": true, "sec_per_step": 0.07, "alpha": 0.03, "t": 509.4} |
| {"event": "train", "step": 4550, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.36236, "bpb": 1.96547, "chunk_loss": 0.030288, "F": 0.19562, "G": 0.20113, "chunks": 200.31, "bytes_per_chunk": 5.112, "boundary_entropy": 0.09052, "router_noise": 0.0, "grad_norm": 0.6861, "lr": 7.09e-05, "router_trainable": true, "sec_per_step": 0.068, "alpha": 0.03, "t": 512.8} |
| {"event": "probe", "step": 4600, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3427.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.19692, "cps": 26.773438, "mean_p": 0.201646, "boundary_entropy": 0.091307, "bpb": 1.912861}, "tr": {"chunks": 3713.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.205104, "cps": 29.007812, "mean_p": 0.194377, "boundary_entropy": 0.091034, "bpb": 2.060223}, "fi": {"chunks": 4598.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.244431, "cps": 35.921875, "mean_p": 0.241924, "boundary_entropy": 0.09251, "bpb": 2.156816}}, "bpb": {"en": 1.91286, "tr": 2.06022, "fi": 2.15682}, "gini_cpc": 0.048997, "gini_cps": 0.066508, "bpic": 4.833191, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0415602893284144, "fi": 1.2412724011804366}, "t": 516.2} |
| {"event": "eval", "step": 4600, "phase": 3, "bpb": {"en": 1.92507, "tr": 2.06885, "fi": 2.16447}, "bytes_per_chunk": {"en": 4.991, "tr": 5.585, "fi": 4.489}, "bpb_mean": 2.05279, "frozen_inventory": false, "t": 516.6} |
| {"event": "train", "step": 4600, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.36224, "bpb": 1.9653, "chunk_loss": 0.030453, "F": 0.20673, "G": 0.21194, "chunks": 211.69, "bytes_per_chunk": 4.837, "boundary_entropy": 0.09243, "router_noise": 0.0, "grad_norm": 0.7016, "lr": 6.82e-05, "router_trainable": true, "sec_per_step": 0.08, "alpha": 0.03, "t": 516.8} |
| {"event": "train", "step": 4650, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.39422, "bpb": 2.01144, "chunk_loss": 0.030348, "F": 0.19931, "G": 0.20436, "chunks": 204.09, "bytes_per_chunk": 5.017, "boundary_entropy": 0.08912, "router_noise": 0.0, "grad_norm": 0.6562, "lr": 6.57e-05, "router_trainable": true, "sec_per_step": 0.061, "alpha": 0.03, "t": 519.9} |
| {"event": "train", "step": 4700, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.42194, "bpb": 2.05142, "chunk_loss": 0.030421, "F": 0.2059, "G": 0.21014, "chunks": 210.84, "bytes_per_chunk": 4.857, "boundary_entropy": 0.09192, "router_noise": 0.0, "grad_norm": 0.6639, "lr": 6.32e-05, "router_trainable": true, "sec_per_step": 0.059, "alpha": 0.03, "t": 522.8} |
| {"event": "train", "step": 4750, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.4079, "bpb": 2.03117, "chunk_loss": 0.030345, "F": 0.19885, "G": 0.2052, "chunks": 203.62, "bytes_per_chunk": 5.029, "boundary_entropy": 0.09242, "router_noise": 0.0, "grad_norm": 0.6736, "lr": 6.08e-05, "router_trainable": true, "sec_per_step": 0.052, "alpha": 0.03, "t": 525.4} |
| {"event": "probe", "step": 4800, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3396.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.195139, "cps": 26.53125, "mean_p": 0.201158, "boundary_entropy": 0.090829, "bpb": 1.904807}, "tr": {"chunks": 3720.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.205491, "cps": 29.0625, "mean_p": 0.193472, "boundary_entropy": 0.090559, "bpb": 2.045126}, "fi": {"chunks": 4586.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.243794, "cps": 35.828125, "mean_p": 0.239655, "boundary_entropy": 0.092732, "bpb": 2.141897}}, "bpb": {"en": 1.90481, "tr": 2.04513, "fi": 2.1419}, "gini_cpc": 0.050334, "gini_cps": 0.067795, "bpic": 4.84806, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0530495989868731, "fi": 1.2493341333063335}, "t": 528.2} |
| {"event": "eval", "step": 4800, "phase": 3, "bpb": {"en": 1.91773, "tr": 2.05551, "fi": 2.14883}, "bytes_per_chunk": {"en": 5.0, "tr": 5.571, "fi": 4.499}, "bpb_mean": 2.04069, "frozen_inventory": false, "t": 528.5} |
| {"event": "train", "step": 4800, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.37945, "bpb": 1.99012, "chunk_loss": 0.030379, "F": 0.20325, "G": 0.20756, "chunks": 208.12, "bytes_per_chunk": 4.92, "boundary_entropy": 0.09178, "router_noise": 0.0, "grad_norm": 0.6665, "lr": 5.85e-05, "router_trainable": true, "sec_per_step": 0.071, "alpha": 0.03, "t": 529.0} |
| {"event": "train", "step": 4850, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.35553, "bpb": 1.95562, "chunk_loss": 0.030323, "F": 0.19861, "G": 0.20353, "chunks": 203.38, "bytes_per_chunk": 5.035, "boundary_entropy": 0.08741, "router_noise": 0.0, "grad_norm": 0.7085, "lr": 5.62e-05, "router_trainable": true, "sec_per_step": 0.089, "alpha": 0.03, "t": 533.4} |
| {"event": "train", "step": 4900, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.38232, "bpb": 1.99426, "chunk_loss": 0.030334, "F": 0.19635, "G": 0.20023, "chunks": 201.06, "bytes_per_chunk": 5.093, "boundary_entropy": 0.08496, "router_noise": 0.0, "grad_norm": 0.6681, "lr": 5.41e-05, "router_trainable": true, "sec_per_step": 0.062, "alpha": 0.03, "t": 536.5} |
| {"event": "train", "step": 4950, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.34307, "bpb": 1.93764, "chunk_loss": 0.030322, "F": 0.19781, "G": 0.20223, "chunks": 202.56, "bytes_per_chunk": 5.055, "boundary_entropy": 0.08956, "router_noise": 0.0, "grad_norm": 0.6779, "lr": 5.2e-05, "router_trainable": true, "sec_per_step": 0.056, "alpha": 0.03, "t": 539.3} |
| {"event": "probe", "step": 5000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3397.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.195196, "cps": 26.539062, "mean_p": 0.20029, "boundary_entropy": 0.09115, "bpb": 1.897904}, "tr": {"chunks": 3717.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.205325, "cps": 29.039062, "mean_p": 0.194452, "boundary_entropy": 0.08839, "bpb": 2.037769}, "fi": {"chunks": 4569.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.24289, "cps": 35.695312, "mean_p": 0.240141, "boundary_entropy": 0.09152, "bpb": 2.129651}}, "bpb": {"en": 1.8979, "tr": 2.03777, "fi": 2.12965}, "gini_cpc": 0.049417, "gini_cps": 0.066878, "bpic": 4.855945, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.051890621440057, "fi": 1.2443365215427804}, "t": 542.6} |
| {"event": "eval", "step": 5000, "phase": 3, "bpb": {"en": 1.91083, "tr": 2.04728, "fi": 2.13764}, "bytes_per_chunk": {"en": 5.022, "tr": 5.588, "fi": 4.51}, "bpb_mean": 2.03192, "frozen_inventory": false, "t": 543.0} |
| {"event": "train", "step": 5000, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.33176, "bpb": 1.92133, "chunk_loss": 0.03034, "F": 0.20071, "G": 0.20478, "chunks": 205.53, "bytes_per_chunk": 4.982, "boundary_entropy": 0.08922, "router_noise": 0.0, "grad_norm": 0.6445, "lr": 5e-05, "router_trainable": true, "sec_per_step": 0.078, "alpha": 0.03, "t": 543.2} |
| {"event": "train", "step": 5050, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.3518, "bpb": 1.95024, "chunk_loss": 0.030374, "F": 0.2027, "G": 0.20721, "chunks": 207.56, "bytes_per_chunk": 4.933, "boundary_entropy": 0.08776, "router_noise": 0.0, "grad_norm": 0.6912, "lr": 4.81e-05, "router_trainable": true, "sec_per_step": 0.06, "alpha": 0.03, "t": 546.2} |
| {"event": "train", "step": 5100, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.38228, "bpb": 1.99421, "chunk_loss": 0.030377, "F": 0.20303, "G": 0.20715, "chunks": 207.91, "bytes_per_chunk": 4.925, "boundary_entropy": 0.08797, "router_noise": 0.0, "grad_norm": 0.6882, "lr": 4.63e-05, "router_trainable": true, "sec_per_step": 0.055, "alpha": 0.03, "t": 549.0} |
| {"event": "train", "step": 5150, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.37857, "bpb": 1.98886, "chunk_loss": 0.03034, "F": 0.20001, "G": 0.20399, "chunks": 204.81, "bytes_per_chunk": 5.0, "boundary_entropy": 0.08763, "router_noise": 0.0, "grad_norm": 0.7123, "lr": 4.45e-05, "router_trainable": true, "sec_per_step": 0.057, "alpha": 0.03, "t": 551.8} |
| {"event": "probe", "step": 5200, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3405.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.195656, "cps": 26.601562, "mean_p": 0.2, "boundary_entropy": 0.088791, "bpb": 1.891582}, "tr": {"chunks": 3718.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.20538, "cps": 29.046875, "mean_p": 0.194209, "boundary_entropy": 0.087467, "bpb": 2.029812}, "fi": {"chunks": 4565.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.242677, "cps": 35.664062, "mean_p": 0.239872, "boundary_entropy": 0.090462, "bpb": 2.121198}}, "bpb": {"en": 1.89158, "tr": 2.02981, "fi": 2.1212}, "gini_cpc": 0.048698, "gini_cps": 0.066165, "bpic": 4.853867, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0497015487247348, "fi": 1.24032615652525}, "t": 554.6} |
| {"event": "eval", "step": 5200, "phase": 3, "bpb": {"en": 1.90557, "tr": 2.03965, "fi": 2.1288}, "bytes_per_chunk": {"en": 5.022, "tr": 5.577, "fi": 4.503}, "bpb_mean": 2.02467, "frozen_inventory": false, "t": 554.9} |
| {"event": "train", "step": 5200, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.33276, "bpb": 1.92277, "chunk_loss": 0.030302, "F": 0.19812, "G": 0.20189, "chunks": 202.88, "bytes_per_chunk": 5.047, "boundary_entropy": 0.08549, "router_noise": 0.0, "grad_norm": 0.6692, "lr": 4.29e-05, "router_trainable": true, "sec_per_step": 0.068, "alpha": 0.03, "t": 555.3} |
| {"event": "train", "step": 5250, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.32447, "bpb": 1.9108, "chunk_loss": 0.030366, "F": 0.20139, "G": 0.20492, "chunks": 206.22, "bytes_per_chunk": 4.966, "boundary_entropy": 0.08738, "router_noise": 0.0, "grad_norm": 0.6691, "lr": 4.14e-05, "router_trainable": true, "sec_per_step": 0.051, "alpha": 0.03, "t": 557.8} |
| {"event": "train", "step": 5300, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.38143, "bpb": 1.99298, "chunk_loss": 0.030439, "F": 0.20389, "G": 0.20781, "chunks": 208.78, "bytes_per_chunk": 4.905, "boundary_entropy": 0.08175, "router_noise": 0.0, "grad_norm": 0.7022, "lr": 3.99e-05, "router_trainable": true, "sec_per_step": 0.054, "alpha": 0.03, "t": 560.5} |
| {"event": "train", "step": 5350, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.31593, "bpb": 1.89848, "chunk_loss": 0.030279, "F": 0.19696, "G": 0.20085, "chunks": 201.69, "bytes_per_chunk": 5.077, "boundary_entropy": 0.08446, "router_noise": 0.0, "grad_norm": 0.6985, "lr": 3.86e-05, "router_trainable": true, "sec_per_step": 0.051, "alpha": 0.03, "t": 563.1} |
| {"event": "probe", "step": 5400, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3416.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.196288, "cps": 26.6875, "mean_p": 0.200894, "boundary_entropy": 0.088083, "bpb": 1.886311}, "tr": {"chunks": 3707.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.204773, "cps": 28.960938, "mean_p": 0.193883, "boundary_entropy": 0.087038, "bpb": 2.024121}, "fi": {"chunks": 4566.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.24273, "cps": 35.671875, "mean_p": 0.239688, "boundary_entropy": 0.089921, "bpb": 2.115072}}, "bpb": {"en": 1.88631, "tr": 2.02412, "fi": 2.11507}, "gini_cpc": 0.048093, "gini_cps": 0.065589, "bpic": 4.853452, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0432257369067273, "fi": 1.2366029605229987}, "t": 566.8} |
| {"event": "eval", "step": 5400, "phase": 3, "bpb": {"en": 1.89967, "tr": 2.03394, "fi": 2.12321}, "bytes_per_chunk": {"en": 4.98, "tr": 5.599, "fi": 4.5}, "bpb_mean": 2.01894, "frozen_inventory": false, "t": 567.1} |
| {"event": "train", "step": 5400, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.37803, "bpb": 1.98808, "chunk_loss": 0.030511, "F": 0.20795, "G": 0.21156, "chunks": 212.94, "bytes_per_chunk": 4.809, "boundary_entropy": 0.08842, "router_noise": 0.0, "grad_norm": 0.7109, "lr": 3.73e-05, "router_trainable": true, "sec_per_step": 0.087, "alpha": 0.03, "t": 567.4} |
| {"event": "train", "step": 5450, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.31759, "bpb": 1.90089, "chunk_loss": 0.030406, "F": 0.20319, "G": 0.20684, "chunks": 208.06, "bytes_per_chunk": 4.922, "boundary_entropy": 0.08398, "router_noise": 0.0, "grad_norm": 0.6975, "lr": 3.62e-05, "router_trainable": true, "sec_per_step": 0.058, "alpha": 0.03, "t": 570.4} |
| {"event": "train", "step": 5500, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.31032, "bpb": 1.89039, "chunk_loss": 0.030344, "F": 0.20071, "G": 0.20464, "chunks": 205.53, "bytes_per_chunk": 4.982, "boundary_entropy": 0.08701, "router_noise": 0.0, "grad_norm": 0.6714, "lr": 3.51e-05, "router_trainable": true, "sec_per_step": 0.058, "alpha": 0.03, "t": 573.3} |
| {"event": "train", "step": 5550, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.36687, "bpb": 1.97198, "chunk_loss": 0.03024, "F": 0.1944, "G": 0.19909, "chunks": 199.06, "bytes_per_chunk": 5.144, "boundary_entropy": 0.08724, "router_noise": 0.0, "grad_norm": 0.7219, "lr": 3.41e-05, "router_trainable": true, "sec_per_step": 0.06, "alpha": 0.03, "t": 576.2} |
| {"event": "probe", "step": 5600, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3398.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.195254, "cps": 26.546875, "mean_p": 0.19975, "boundary_entropy": 0.086446, "bpb": 1.881567}, "tr": {"chunks": 3688.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.203723, "cps": 28.8125, "mean_p": 0.193076, "boundary_entropy": 0.084988, "bpb": 2.018293}, "fi": {"chunks": 4576.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.243262, "cps": 35.75, "mean_p": 0.239504, "boundary_entropy": 0.087133, "bpb": 2.106903}}, "bpb": {"en": 1.88157, "tr": 2.01829, "fi": 2.1069}, "gini_cpc": 0.049834, "gini_cps": 0.067341, "bpic": 4.864689, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0433766339412134, "fi": 1.2458761668415057}, "t": 579.0} |
| {"event": "eval", "step": 5600, "phase": 3, "bpb": {"en": 1.8954, "tr": 2.02721, "fi": 2.11528}, "bytes_per_chunk": {"en": 5.011, "tr": 5.638, "fi": 4.493}, "bpb_mean": 2.01263, "frozen_inventory": false, "t": 579.3} |
| {"event": "train", "step": 5600, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.35221, "bpb": 1.95082, "chunk_loss": 0.030397, "F": 0.20203, "G": 0.2064, "chunks": 206.88, "bytes_per_chunk": 4.95, "boundary_entropy": 0.08513, "router_noise": 0.0, "grad_norm": 0.7669, "lr": 3.33e-05, "router_trainable": true, "sec_per_step": 0.067, "alpha": 0.03, "t": 579.6} |
| {"event": "train", "step": 5650, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.33991, "bpb": 1.93308, "chunk_loss": 0.030386, "F": 0.20181, "G": 0.20624, "chunks": 206.66, "bytes_per_chunk": 4.955, "boundary_entropy": 0.08602, "router_noise": 0.0, "grad_norm": 0.7206, "lr": 3.25e-05, "router_trainable": true, "sec_per_step": 0.051, "alpha": 0.03, "t": 582.2} |
| {"event": "train", "step": 5700, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.34515, "bpb": 1.94064, "chunk_loss": 0.030323, "F": 0.19873, "G": 0.20202, "chunks": 203.5, "bytes_per_chunk": 5.032, "boundary_entropy": 0.08242, "router_noise": 0.0, "grad_norm": 0.757, "lr": 3.18e-05, "router_trainable": true, "sec_per_step": 0.054, "alpha": 0.03, "t": 584.9} |
| {"event": "train", "step": 5750, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.39919, "bpb": 2.01861, "chunk_loss": 0.030431, "F": 0.20432, "G": 0.20855, "chunks": 209.22, "bytes_per_chunk": 4.894, "boundary_entropy": 0.08446, "router_noise": 0.0, "grad_norm": 0.7519, "lr": 3.13e-05, "router_trainable": true, "sec_per_step": 0.061, "alpha": 0.03, "t": 588.0} |
| {"event": "probe", "step": 5800, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3417.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.196345, "cps": 26.695312, "mean_p": 0.20021, "boundary_entropy": 0.083684, "bpb": 1.87787}, "tr": {"chunks": 3699.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.204331, "cps": 28.898438, "mean_p": 0.193567, "boundary_entropy": 0.084036, "bpb": 2.01451}, "fi": {"chunks": 4575.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.243209, "cps": 35.742188, "mean_p": 0.239229, "boundary_entropy": 0.085113, "bpb": 2.103687}}, "bpb": {"en": 1.87787, "tr": 2.01451, "fi": 2.10369}, "gini_cpc": 0.048521, "gini_cps": 0.066034, "bpic": 4.852622, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0406697273241396, "fi": 1.2386778064821038}, "t": 590.4} |
| {"event": "eval", "step": 5800, "phase": 3, "bpb": {"en": 1.89162, "tr": 2.02258, "fi": 2.11169}, "bytes_per_chunk": {"en": 4.99, "tr": 5.6, "fi": 4.488}, "bpb_mean": 2.00863, "frozen_inventory": false, "t": 590.8} |
| {"event": "train", "step": 5800, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.32367, "bpb": 1.90965, "chunk_loss": 0.030404, "F": 0.20276, "G": 0.20675, "chunks": 207.62, "bytes_per_chunk": 4.932, "boundary_entropy": 0.08554, "router_noise": 0.0, "grad_norm": 0.7122, "lr": 3.08e-05, "router_trainable": true, "sec_per_step": 0.061, "alpha": 0.03, "t": 591.0} |
| {"event": "train", "step": 5850, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.34341, "bpb": 1.93813, "chunk_loss": 0.030418, "F": 0.20557, "G": 0.20935, "chunks": 210.5, "bytes_per_chunk": 4.865, "boundary_entropy": 0.08352, "router_noise": 0.0, "grad_norm": 0.7031, "lr": 3.05e-05, "router_trainable": true, "sec_per_step": 0.058, "alpha": 0.03, "t": 593.9} |
| {"event": "train", "step": 5900, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.27522, "bpb": 1.83975, "chunk_loss": 0.030323, "F": 0.19931, "G": 0.20273, "chunks": 204.09, "bytes_per_chunk": 5.017, "boundary_entropy": 0.08212, "router_noise": 0.0, "grad_norm": 0.723, "lr": 3.02e-05, "router_trainable": true, "sec_per_step": 0.058, "alpha": 0.03, "t": 596.8} |
| {"event": "train", "step": 5950, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.37894, "bpb": 1.98939, "chunk_loss": 0.030359, "F": 0.20129, "G": 0.20561, "chunks": 206.12, "bytes_per_chunk": 4.968, "boundary_entropy": 0.08236, "router_noise": 0.0, "grad_norm": 0.7172, "lr": 3.01e-05, "router_trainable": true, "sec_per_step": 0.057, "alpha": 0.03, "t": 599.6} |
| {"event": "train", "step": 5999, "arch": "hnet", "phase": 3, "objective": "global", "lm_loss": 1.38395, "bpb": 1.99662, "chunk_loss": 0.030359, "F": 0.20187, "G": 0.20493, "chunks": 206.72, "bytes_per_chunk": 4.954, "boundary_entropy": 0.08456, "router_noise": 0.0, "grad_norm": 0.6936, "lr": 3e-05, "router_trainable": true, "sec_per_step": 0.071, "alpha": 0.03, "t": 603.2} |
| {"event": "probe", "step": 6000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3413.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.196116, "cps": 26.664062, "mean_p": 0.199802, "boundary_entropy": 0.084729, "bpb": 1.87564}, "tr": {"chunks": 3694.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.204055, "cps": 28.859375, "mean_p": 0.192829, "boundary_entropy": 0.084016, "bpb": 2.009921}, "fi": {"chunks": 4556.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.242199, "cps": 35.59375, "mean_p": 0.238665, "boundary_entropy": 0.08502, "bpb": 2.101415}}, "bpb": {"en": 1.87564, "tr": 2.00992, "fi": 2.10142}, "gini_cpc": 0.047826, "gini_cps": 0.065335, "bpic": 4.864272, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0404810420121124, "fi": 1.2349792591168154}, "t": 603.5} |
| {"event": "eval", "step": 6000, "phase": 3, "bpb": {"en": 1.88907, "tr": 2.01891, "fi": 2.10944}, "bytes_per_chunk": {"en": 5.0, "tr": 5.627, "fi": 4.507}, "bpb_mean": 2.00581, "frozen_inventory": false, "t": 603.9} |
| {"event": "reps_dumped", "dirs": ["/local/scratch/sas245/beetle-hnet/runs/ali_hnet_s2/probe/reps/th0.30", "/local/scratch/sas245/beetle-hnet/runs/ali_hnet_s2/probe/reps/th0.40", "/local/scratch/sas245/beetle-hnet/runs/ali_hnet_s2/probe/reps/th0.50", "/local/scratch/sas245/beetle-hnet/runs/ali_hnet_s2/probe/reps/th0.60", "/local/scratch/sas245/beetle-hnet/runs/ali_hnet_s2/probe/reps/th0.70"], "t": 605.0} |
| {"event": "final", "run": "ali_hnet_s2", "arch": "hnet", "objective": "ratio", "condition": "A", "step": 6000, "params": 22516992, "seed": 2, "per_lang": {"en": {"cpc": 0.19611561225076135, "cps": 26.6640625, "bpb": 1.8946112190768243, "bytes_per_chunk": 4.958509481250296}, "tr": {"cpc": 0.20405457658951556, "cps": 28.859375, "bpb": 2.007523021094531, "bytes_per_chunk": 5.628971129792466}, "fi": {"cpc": 0.24219871351868588, "cps": 35.59375, "bpb": 2.1278645381230055, "bytes_per_chunk": 4.4871151887575635}}, "bpb_mean": 2.0099995927647867, "gini_cpc": 0.047826206092993706, "gini_cps": 0.06533481951470477, "bpic": 4.864271628226014, "mean_boundary_entropy": 0.08456268906593323, "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": 605.7513825893402, "t": 605.8} |
| {"event": "hub_done", "repo": "AdaptiveChunking/ali_hnet_s2", "pushed": [0, 1, 2, 4, 6, 8, 16, 32, 64, 128, 256, 512, 1000, 1024, 2048, 3000, 4096, 6000], "failed": [], "ok": true, "t": 620.9} |
|
|
