| {"event": "hub", "repo": "AdaptiveChunking/ali_hnet-ffn_s3", "keep_steps": [0, 1, 2, 4, 6, 8, 16, 32, 64, 128, 256, 512, 1000, 1024, 2048, 3000, 4096, 6000], "t": 2.6} |
| {"event": "start", "run": "ali_hnet-ffn_s3", "arch": "hnet_ffn", "objective": "ratio", "condition": "A", "params": 22785033, "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", "ffn_params": 268041, "matched_to_ali_params": 268096, "t": 2.6} |
| {"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": 2.6} |
| {"event": "probe", "step": 0, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 15894.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.913291, "cps": 124.171875, "mean_p": 0.532199, "boundary_entropy": 0.685519, "bpb": 8.014977}, "tr": {"chunks": 15976.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.882506, "cps": 124.8125, "mean_p": 0.523031, "boundary_entropy": 0.686614, "bpb": 8.046717}, "fi": {"chunks": 12092.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.642815, "cps": 94.46875, "mean_p": 0.5099, "boundary_entropy": 0.687389, "bpb": 8.01587}}, "bpb": {"en": 8.01498, "tr": 8.04672, "fi": 8.01587}, "gini_cpc": 0.073942, "gini_cps": 0.058899, "bpic": 1.290478, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9662920621975919, "fi": 0.7038452216831197}, "t": 3.8} |
| {"event": "train", "step": 0, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 5.5638, "bpb": 8.02686, "chunk_loss": 0.092026, "F": 0.95703, "G": 0.52969, "chunks": 980.0, "bytes_per_chunk": 1.045, "boundary_entropy": 0.6902, "router_noise": 0.0, "grad_norm": 13.6811, "lr": 1e-06, "router_trainable": true, "sec_per_step": 0.006, "alpha": 0.03, "t": 4.2} |
| {"event": "probe", "step": 1, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 15874.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.912142, "cps": 124.015625, "mean_p": 0.532017, "boundary_entropy": 0.685541, "bpb": 7.975735}, "tr": {"chunks": 15944.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.880738, "cps": 124.5625, "mean_p": 0.522881, "boundary_entropy": 0.686629, "bpb": 8.015364}, "fi": {"chunks": 12054.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.640795, "cps": 94.171875, "mean_p": 0.509762, "boundary_entropy": 0.687394, "bpb": 7.982363}}, "bpb": {"en": 7.97574, "tr": 8.01536, "fi": 7.98236}, "gini_cpc": 0.074331, "gini_cps": 0.059111, "bpic": 1.293125, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9655715887955996, "fi": 0.7025173394647969}, "t": 4.8} |
| {"event": "probe", "step": 2, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 15835.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.909901, "cps": 123.710938, "mean_p": 0.53166, "boundary_entropy": 0.685584, "bpb": 7.901102}, "tr": {"chunks": 15889.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.8777, "cps": 124.132812, "mean_p": 0.522603, "boundary_entropy": 0.686658, "bpb": 7.953559}, "fi": {"chunks": 11989.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.63734, "cps": 93.664062, "mean_p": 0.509519, "boundary_entropy": 0.687404, "bpb": 7.917338}}, "bpb": {"en": 7.9011, "tr": 7.95356, "fi": 7.91734}, "gini_cpc": 0.074933, "gini_cps": 0.059479, "bpic": 1.297829, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9646106801307001, "fi": 0.7004499832705091}, "t": 5.8} |
| {"event": "probe", "step": 4, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 15729.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.90381, "cps": 122.882812, "mean_p": 0.530349, "boundary_entropy": 0.685738, "bpb": 7.661295}, "tr": {"chunks": 15701.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.867315, "cps": 122.664062, "mean_p": 0.521525, "boundary_entropy": 0.686762, "bpb": 7.754717}, "fi": {"chunks": 11712.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.622614, "cps": 91.5, "mean_p": 0.508609, "boundary_entropy": 0.687438, "bpb": 7.705647}}, "bpb": {"en": 7.66129, "tr": 7.75472, "fi": 7.70565}, "gini_cpc": 0.078314, "gini_cps": 0.062074, "bpic": 1.315006, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9596210587583422, "fi": 0.6888777980646152}, "t": 6.9} |
| {"event": "probe", "step": 6, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 15483.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.889674, "cps": 120.960938, "mean_p": 0.528063, "boundary_entropy": 0.68599, "bpb": 7.376746}, "tr": {"chunks": 15326.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.8466, "cps": 119.734375, "mean_p": 0.51967, "boundary_entropy": 0.686928, "bpb": 7.519146}, "fi": {"chunks": 11159.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.593217, "cps": 87.179688, "mean_p": 0.506958, "boundary_entropy": 0.687493, "bpb": 7.453697}}, "bpb": {"en": 7.37675, "tr": 7.51915, "fi": 7.4537}, "gini_cpc": 0.084842, "gini_cps": 0.068687, "bpic": 1.351792, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9515843177850852, "fi": 0.6667797479349002}, "t": 8.0} |
| {"event": "probe", "step": 8, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 15039.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.864161, "cps": 117.492188, "mean_p": 0.524621, "boundary_entropy": 0.686329, "bpb": 7.128644}, "tr": {"chunks": 14710.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.812573, "cps": 114.921875, "mean_p": 0.516915, "boundary_entropy": 0.687146, "bpb": 7.316793}, "fi": {"chunks": 10294.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.547233, "cps": 80.421875, "mean_p": 0.504403, "boundary_entropy": 0.687559, "bpb": 7.23757}}, "bpb": {"en": 7.12864, "tr": 7.31679, "fi": 7.23757}, "gini_cpc": 0.095004, "gini_cps": 0.078998, "bpic": 1.416777, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9403018318201545, "fi": 0.6332532703790517}, "t": 9.2} |
| {"event": "probe", "step": 16, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 5028.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.288916, "cps": 39.28125, "mean_p": 0.494021, "boundary_entropy": 0.687114, "bpb": 6.585645}, "tr": {"chunks": 5923.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.327183, "cps": 46.273438, "mean_p": 0.492327, "boundary_entropy": 0.687467, "bpb": 6.901052}, "fi": {"chunks": 2827.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.150284, "cps": 22.085938, "mean_p": 0.481387, "boundary_entropy": 0.686749, "bpb": 6.779335}}, "bpb": {"en": 6.58564, "tr": 6.90105, "fi": 6.77934}, "gini_cpc": 0.153882, "gini_cps": 0.149804, "bpic": 4.117579, "worst_lang": "tr", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.1324525978829443, "fi": 0.5201669756270421}, "t": 10.9} |
| {"event": "probe", "step": 32, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 7946.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.456588, "cps": 62.078125, "mean_p": 0.502773, "boundary_entropy": 0.687618, "bpb": 6.271383}, "tr": {"chunks": 10325.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.570347, "cps": 80.664062, "mean_p": 0.503926, "boundary_entropy": 0.688064, "bpb": 6.617844}, "fi": {"chunks": 7430.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.394982, "cps": 58.046875, "mean_p": 0.499367, "boundary_entropy": 0.688087, "bpb": 6.45816}}, "bpb": {"en": 6.27138, "tr": 6.61784, "fi": 6.45816}, "gini_cpc": 0.08222, "gini_cps": 0.075094, "bpic": 2.207385, "worst_lang": "tr", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.2491513693234448, "fi": 0.8650724670511654}, "t": 13.2} |
| {"event": "train", "step": 50, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 4.22756, "bpb": 6.09909, "chunk_loss": 0.031713, "F": 0.19052, "G": 0.49062, "chunks": 195.09, "bytes_per_chunk": 5.249, "boundary_entropy": 0.692, "router_noise": 0.0, "grad_norm": 2.5052, "lr": 5.1e-05, "router_trainable": true, "sec_per_step": 0.215, "alpha": 0.03, "t": 15.0} |
| {"event": "probe", "step": 64, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3319.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.190714, "cps": 25.929688, "mean_p": 0.485252, "boundary_entropy": 0.686522, "bpb": 5.609814}, "tr": {"chunks": 4405.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.24333, "cps": 34.414062, "mean_p": 0.487867, "boundary_entropy": 0.687296, "bpb": 5.998782}, "fi": {"chunks": 4860.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.258359, "cps": 37.96875, "mean_p": 0.488415, "boundary_entropy": 0.687206, "bpb": 5.799954}}, "bpb": {"en": 5.60981, "tr": 5.99878, "fi": 5.79995}, "gini_cpc": 0.065131, "gini_cps": 0.081638, "bpic": 4.508264, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.2758870489955398, "fi": 1.3546941444724996}, "t": 16.3} |
| {"event": "train", "step": 100, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 3.33501, "bpb": 4.8114, "chunk_loss": 0.031144, "F": 0.18356, "G": 0.46959, "chunks": 187.97, "bytes_per_chunk": 5.448, "boundary_entropy": 0.68831, "router_noise": 0.0, "grad_norm": 3.0373, "lr": 0.000101, "router_trainable": true, "sec_per_step": 0.084, "alpha": 0.03, "t": 19.2} |
| {"event": "probe", "step": 128, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3495.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.200827, "cps": 27.304688, "mean_p": 0.43729, "boundary_entropy": 0.668926, "bpb": 4.145653}, "tr": {"chunks": 4561.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.251947, "cps": 35.632812, "mean_p": 0.438555, "boundary_entropy": 0.669706, "bpb": 4.470258}, "fi": {"chunks": 4526.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.240604, "cps": 35.359375, "mean_p": 0.439128, "boundary_entropy": 0.670939, "bpb": 4.390613}}, "bpb": {"en": 4.14565, "tr": 4.47026, "fi": 4.39061}, "gini_cpc": 0.04915, "gini_cps": 0.056483, "bpic": 4.508981, "worst_lang": "tr", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.2545456269667203, "fi": 1.1980628629536685}, "t": 21.5} |
| {"event": "train", "step": 150, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 2.74589, "bpb": 3.96149, "chunk_loss": 0.03177, "F": 0.20712, "G": 0.36501, "chunks": 212.09, "bytes_per_chunk": 4.828, "boundary_entropy": 0.61678, "router_noise": 0.0, "grad_norm": 2.4261, "lr": 0.000151, "router_trainable": true, "sec_per_step": 0.083, "alpha": 0.03, "t": 23.3} |
| {"event": "probe", "step": 200, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3257.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.187152, "cps": 25.445312, "mean_p": 0.360123, "boundary_entropy": 0.554517, "bpb": 3.640746}, "tr": {"chunks": 4281.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.23648, "cps": 33.445312, "mean_p": 0.342964, "boundary_entropy": 0.547218, "bpb": 3.794314}, "fi": {"chunks": 4323.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.229812, "cps": 33.773438, "mean_p": 0.343358, "boundary_entropy": 0.547962, "bpb": 3.867294}}, "bpb": {"en": 3.64075, "tr": 3.79431, "fi": 3.86729}, "gini_cpc": 0.050327, "gini_cps": 0.059916, "bpic": 4.783071, "worst_lang": "tr", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.2635750386892484, "fi": 1.2279472581735806}, "t": 27.4} |
| {"event": "eval", "step": 200, "phase": 1, "bpb": {"en": 3.64573, "tr": 3.8038, "fi": 3.86596}, "bytes_per_chunk": {"en": 5.25, "tr": 4.916, "fi": 4.824}, "bpb_mean": 3.77183, "frozen_inventory": false, "t": 28.0} |
| {"event": "train", "step": 200, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 2.58701, "bpb": 3.73227, "chunk_loss": 0.030973, "F": 0.19135, "G": 0.34895, "chunks": 195.94, "bytes_per_chunk": 5.226, "boundary_entropy": 0.5503, "router_noise": 0.0, "grad_norm": 1.0854, "lr": 0.000201, "router_trainable": true, "sec_per_step": 0.098, "alpha": 0.03, "t": 28.2} |
| {"event": "train", "step": 250, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 2.38939, "bpb": 3.44717, "chunk_loss": 0.031417, "F": 0.2197, "G": 0.28637, "chunks": 224.97, "bytes_per_chunk": 4.552, "boundary_entropy": 0.41206, "router_noise": 0.0, "grad_norm": 2.4706, "lr": 0.000251, "router_trainable": true, "sec_per_step": 0.069, "alpha": 0.03, "t": 31.7} |
| {"event": "probe", "step": 256, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 4207.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.24174, "cps": 32.867188, "mean_p": 0.298744, "boundary_entropy": 0.381708, "bpb": 3.331501}, "tr": {"chunks": 4164.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.230017, "cps": 32.53125, "mean_p": 0.304112, "boundary_entropy": 0.443403, "bpb": 3.54305}, "fi": {"chunks": 4563.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.242571, "cps": 35.648438, "mean_p": 0.318762, "boundary_entropy": 0.47426, "bpb": 3.705772}}, "bpb": {"en": 3.3315, "tr": 3.54305, "fi": 3.70577}, "gini_cpc": 0.011716, "gini_cps": 0.020566, "bpic": 4.386269, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9515065398232739, "fi": 1.0034371910178992}, "t": 32.4} |
| {"event": "train", "step": 300, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 2.29626, "bpb": 3.31281, "chunk_loss": 0.031535, "F": 0.22983, "G": 0.27387, "chunks": 235.34, "bytes_per_chunk": 4.351, "boundary_entropy": 0.3543, "router_noise": 0.0, "grad_norm": 1.2639, "lr": 0.0003, "router_trainable": true, "sec_per_step": 0.087, "alpha": 0.03, "t": 36.0} |
| {"event": "train", "step": 350, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 2.23518, "bpb": 3.22469, "chunk_loss": 0.031058, "F": 0.21878, "G": 0.25838, "chunks": 224.03, "bytes_per_chunk": 4.571, "boundary_entropy": 0.31017, "router_noise": 0.0, "grad_norm": 1.5735, "lr": 0.0002999, "router_trainable": true, "sec_per_step": 0.072, "alpha": 0.03, "t": 39.6} |
| {"event": "probe", "step": 400, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 4058.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.233178, "cps": 31.703125, "mean_p": 0.261109, "boundary_entropy": 0.282326, "bpb": 2.894947}, "tr": {"chunks": 4236.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.233994, "cps": 33.09375, "mean_p": 0.261104, "boundary_entropy": 0.349481, "bpb": 3.195352}, "fi": {"chunks": 4599.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.244485, "cps": 35.929688, "mean_p": 0.274302, "boundary_entropy": 0.376381, "bpb": 3.355687}}, "bpb": {"en": 2.89495, "tr": 3.19535, "fi": 3.35569}, "gini_cpc": 0.010592, "gini_cps": 0.027974, "bpic": 4.400217, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0035002326509768, "fi": 1.0484883363783746}, "t": 43.5} |
| {"event": "eval", "step": 400, "phase": 1, "bpb": {"en": 2.90632, "tr": 3.20427, "fi": 3.34774}, "bytes_per_chunk": {"en": 4.307, "tr": 5.026, "fi": 4.417}, "bpb_mean": 3.15278, "frozen_inventory": false, "t": 44.1} |
| {"event": "train", "step": 400, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 2.15807, "bpb": 3.11343, "chunk_loss": 0.031442, "F": 0.23041, "G": 0.26266, "chunks": 235.94, "bytes_per_chunk": 4.34, "boundary_entropy": 0.30955, "router_noise": 0.0, "grad_norm": 1.0254, "lr": 0.0002998, "router_trainable": true, "sec_per_step": 0.095, "alpha": 0.03, "t": 44.4} |
| {"event": "train", "step": 450, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 2.04718, "bpb": 2.95346, "chunk_loss": 0.031314, "F": 0.22964, "G": 0.25961, "chunks": 235.16, "bytes_per_chunk": 4.355, "boundary_entropy": 0.28295, "router_noise": 0.0, "grad_norm": 1.4797, "lr": 0.0002995, "router_trainable": true, "sec_per_step": 0.075, "alpha": 0.03, "t": 48.1} |
| {"event": "train", "step": 500, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.97693, "bpb": 2.85211, "chunk_loss": 0.03134, "F": 0.23077, "G": 0.25969, "chunks": 236.31, "bytes_per_chunk": 4.333, "boundary_entropy": 0.26778, "router_noise": 0.0, "grad_norm": 1.2795, "lr": 0.0002992, "router_trainable": true, "sec_per_step": 0.07, "alpha": 0.03, "t": 51.6} |
| {"event": "probe", "step": 512, "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": 0.252289, "boundary_entropy": 0.259574, "bpb": 2.697352}, "tr": {"chunks": 4089.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.225874, "cps": 31.945312, "mean_p": 0.25539, "boundary_entropy": 0.323412, "bpb": 3.036091}, "fi": {"chunks": 4784.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.254319, "cps": 37.375, "mean_p": 0.265866, "boundary_entropy": 0.326031, "bpb": 3.221391}}, "bpb": {"en": 2.69735, "tr": 3.03609, "fi": 3.22139}, "gini_cpc": 0.028776, "gini_cps": 0.046306, "bpic": 4.4426, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0086959417110124, "fi": 1.1357245193660745}, "t": 52.8} |
| {"event": "train", "step": 550, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.92225, "bpb": 2.77321, "chunk_loss": 0.031126, "F": 0.2262, "G": 0.24877, "chunks": 231.62, "bytes_per_chunk": 4.421, "boundary_entropy": 0.25853, "router_noise": 0.0, "grad_norm": 1.214, "lr": 0.0002987, "router_trainable": true, "sec_per_step": 0.083, "alpha": 0.03, "t": 55.8} |
| {"event": "probe", "step": 600, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 4032.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.231684, "cps": 31.5, "mean_p": 0.253678, "boundary_entropy": 0.232628, "bpb": 2.587029}, "tr": {"chunks": 4330.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.239187, "cps": 33.828125, "mean_p": 0.24794, "boundary_entropy": 0.287971, "bpb": 2.933515}, "fi": {"chunks": 5039.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.267875, "cps": 39.367188, "mean_p": 0.270549, "boundary_entropy": 0.28683, "bpb": 3.14049}}, "bpb": {"en": 2.58703, "tr": 2.93352, "fi": 3.14049}, "gini_cpc": 0.03266, "gini_cps": 0.050096, "bpic": 4.233415, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0323832310088041, "fi": 1.1562082706800225}, "t": 59.8} |
| {"event": "eval", "step": 600, "phase": 1, "bpb": {"en": 2.59818, "tr": 2.94339, "fi": 3.14077}, "bytes_per_chunk": {"en": 4.301, "tr": 4.771, "fi": 4.06}, "bpb_mean": 2.89411, "frozen_inventory": false, "t": 60.5} |
| {"event": "train", "step": 600, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.84194, "bpb": 2.65735, "chunk_loss": 0.031259, "F": 0.23132, "G": 0.25275, "chunks": 236.88, "bytes_per_chunk": 4.323, "boundary_entropy": 0.23667, "router_noise": 0.0, "grad_norm": 1.1416, "lr": 0.0002982, "router_trainable": true, "sec_per_step": 0.098, "alpha": 0.03, "t": 60.7} |
| {"event": "train", "step": 650, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.8409, "bpb": 2.65586, "chunk_loss": 0.030947, "F": 0.21716, "G": 0.24584, "chunks": 222.38, "bytes_per_chunk": 4.605, "boundary_entropy": 0.23883, "router_noise": 0.0, "grad_norm": 1.1944, "lr": 0.0002975, "router_trainable": true, "sec_per_step": 0.075, "alpha": 0.03, "t": 64.4} |
| {"event": "train", "step": 700, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.72616, "bpb": 2.49032, "chunk_loss": 0.031074, "F": 0.22537, "G": 0.24729, "chunks": 230.78, "bytes_per_chunk": 4.437, "boundary_entropy": 0.22904, "router_noise": 0.0, "grad_norm": 1.1178, "lr": 0.0002967, "router_trainable": true, "sec_per_step": 0.071, "alpha": 0.03, "t": 68.0} |
| {"event": "train", "step": 750, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.76736, "bpb": 2.54976, "chunk_loss": 0.031261, "F": 0.23279, "G": 0.25053, "chunks": 238.38, "bytes_per_chunk": 4.296, "boundary_entropy": 0.22036, "router_noise": 0.0, "grad_norm": 1.0384, "lr": 0.0002959, "router_trainable": true, "sec_per_step": 0.075, "alpha": 0.03, "t": 71.8} |
| {"event": "probe", "step": 800, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3923.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.225421, "cps": 30.648438, "mean_p": 0.246115, "boundary_entropy": 0.21593, "bpb": 2.428656}, "tr": {"chunks": 4172.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.230459, "cps": 32.59375, "mean_p": 0.24482, "boundary_entropy": 0.262083, "bpb": 2.771483}, "fi": {"chunks": 4811.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.255755, "cps": 37.585938, "mean_p": 0.264501, "boundary_entropy": 0.260603, "bpb": 2.981449}}, "bpb": {"en": 2.42866, "tr": 2.77148, "fi": 2.98145}, "gini_cpc": 0.028417, "gini_cps": 0.04587, "bpic": 4.395785, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.022349903656209, "fi": 1.134564748095322}, "t": 75.8} |
| {"event": "eval", "step": 800, "phase": 1, "bpb": {"en": 2.44087, "tr": 2.78052, "fi": 2.98502}, "bytes_per_chunk": {"en": 4.479, "tr": 5.076, "fi": 4.284}, "bpb_mean": 2.73547, "frozen_inventory": false, "t": 76.4} |
| {"event": "train", "step": 800, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.75136, "bpb": 2.52668, "chunk_loss": 0.030912, "F": 0.21606, "G": 0.24069, "chunks": 221.25, "bytes_per_chunk": 4.628, "boundary_entropy": 0.22226, "router_noise": 0.0, "grad_norm": 1.0598, "lr": 0.0002949, "router_trainable": true, "sec_per_step": 0.098, "alpha": 0.03, "t": 76.7} |
| {"event": "train", "step": 850, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.77274, "bpb": 2.55752, "chunk_loss": 0.031486, "F": 0.22919, "G": 0.25026, "chunks": 234.69, "bytes_per_chunk": 4.363, "boundary_entropy": 0.22206, "router_noise": 0.0, "grad_norm": 1.0954, "lr": 0.0002938, "router_trainable": true, "sec_per_step": 0.074, "alpha": 0.03, "t": 80.4} |
| {"event": "train", "step": 900, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.69573, "bpb": 2.44642, "chunk_loss": 0.030975, "F": 0.22241, "G": 0.243, "chunks": 227.75, "bytes_per_chunk": 4.496, "boundary_entropy": 0.21159, "router_noise": 0.0, "grad_norm": 0.9581, "lr": 0.0002927, "router_trainable": true, "sec_per_step": 0.073, "alpha": 0.03, "t": 84.0} |
| {"event": "train", "step": 950, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.71121, "bpb": 2.46875, "chunk_loss": 0.030945, "F": 0.21997, "G": 0.24307, "chunks": 225.25, "bytes_per_chunk": 4.546, "boundary_entropy": 0.21459, "router_noise": 0.0, "grad_norm": 1.1521, "lr": 0.0002914, "router_trainable": true, "sec_per_step": 0.073, "alpha": 0.03, "t": 87.6} |
| {"event": "probe", "step": 1000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3872.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.22249, "cps": 30.25, "mean_p": 0.243101, "boundary_entropy": 0.202737, "bpb": 2.342017}, "tr": {"chunks": 4179.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.230846, "cps": 32.648438, "mean_p": 0.249131, "boundary_entropy": 0.242406, "bpb": 2.674195}, "fi": {"chunks": 5041.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.267982, "cps": 39.382812, "mean_p": 0.271989, "boundary_entropy": 0.230832, "bpb": 2.839404}}, "bpb": {"en": 2.34202, "tr": 2.6742, "fi": 2.8394}, "gini_cpc": 0.042044, "gini_cps": 0.059527, "bpic": 4.333333, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.037553718665871, "fi": 1.2044633387753165}, "t": 91.5} |
| {"event": "eval", "step": 1000, "phase": 1, "bpb": {"en": 2.35414, "tr": 2.68266, "fi": 2.8414}, "bytes_per_chunk": {"en": 4.48, "tr": 4.936, "fi": 4.071}, "bpb_mean": 2.62607, "frozen_inventory": false, "t": 92.1} |
| {"event": "train", "step": 1000, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.73361, "bpb": 2.50107, "chunk_loss": 0.031049, "F": 0.2229, "G": 0.24633, "chunks": 228.25, "bytes_per_chunk": 4.486, "boundary_entropy": 0.21075, "router_noise": 0.0, "grad_norm": 1.1865, "lr": 0.0002901, "router_trainable": true, "sec_per_step": 0.098, "alpha": 0.03, "t": 92.5} |
| {"event": "probe", "step": 1024, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3824.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.219732, "cps": 29.875, "mean_p": 0.240504, "boundary_entropy": 0.202442, "bpb": 2.320146}, "tr": {"chunks": 4122.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.227697, "cps": 32.203125, "mean_p": 0.24501, "boundary_entropy": 0.240252, "bpb": 2.629591}, "fi": {"chunks": 4820.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.256233, "cps": 37.65625, "mean_p": 0.268581, "boundary_entropy": 0.237488, "bpb": 2.819991}}, "bpb": {"en": 2.32015, "tr": 2.62959, "fi": 2.81999}, "gini_cpc": 0.034582, "gini_cps": 0.052013, "bpic": 4.443992, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0362479218752203, "fi": 1.1661150257271797}, "t": 94.7} |
| {"event": "train", "step": 1050, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.67209, "bpb": 2.41231, "chunk_loss": 0.031072, "F": 0.22467, "G": 0.24529, "chunks": 230.06, "bytes_per_chunk": 4.451, "boundary_entropy": 0.20485, "router_noise": 0.0, "grad_norm": 0.9822, "lr": 0.0002886, "router_trainable": true, "sec_per_step": 0.088, "alpha": 0.03, "t": 96.9} |
| {"event": "train", "step": 1100, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.6614, "bpb": 2.3969, "chunk_loss": 0.030994, "F": 0.22119, "G": 0.24352, "chunks": 226.5, "bytes_per_chunk": 4.521, "boundary_entropy": 0.20189, "router_noise": 0.0, "grad_norm": 0.8758, "lr": 0.0002871, "router_trainable": true, "sec_per_step": 0.074, "alpha": 0.03, "t": 100.6} |
| {"event": "train", "step": 1150, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.68724, "bpb": 2.43417, "chunk_loss": 0.030896, "F": 0.21701, "G": 0.2427, "chunks": 222.22, "bytes_per_chunk": 4.608, "boundary_entropy": 0.20874, "router_noise": 0.0, "grad_norm": 0.8852, "lr": 0.0002855, "router_trainable": true, "sec_per_step": 0.072, "alpha": 0.03, "t": 104.2} |
| {"event": "probe", "step": 1200, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3833.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.220249, "cps": 29.945312, "mean_p": 0.239734, "boundary_entropy": 0.197253, "bpb": 2.268209}, "tr": {"chunks": 4099.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.226427, "cps": 32.023438, "mean_p": 0.247031, "boundary_entropy": 0.247035, "bpb": 2.580117}, "fi": {"chunks": 4775.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.253841, "cps": 37.304688, "mean_p": 0.268293, "boundary_entropy": 0.22223, "bpb": 2.721298}}, "bpb": {"en": 2.26821, "tr": 2.58012, "fi": 2.7213}, "gini_cpc": 0.031968, "gini_cps": 0.049422, "bpic": 4.464626, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0280462845307352, "fi": 1.1525155492738866}, "t": 108.2} |
| {"event": "eval", "step": 1200, "phase": 1, "bpb": {"en": 2.28007, "tr": 2.58898, "fi": 2.72951}, "bytes_per_chunk": {"en": 4.557, "tr": 5.137, "fi": 4.349}, "bpb_mean": 2.53285, "frozen_inventory": false, "t": 108.7} |
| {"event": "train", "step": 1200, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.64591, "bpb": 2.37454, "chunk_loss": 0.031084, "F": 0.22403, "G": 0.24511, "chunks": 229.41, "bytes_per_chunk": 4.464, "boundary_entropy": 0.20194, "router_noise": 0.0, "grad_norm": 0.8358, "lr": 0.0002837, "router_trainable": true, "sec_per_step": 0.095, "alpha": 0.03, "t": 109.0} |
| {"event": "train", "step": 1250, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.6496, "bpb": 2.37988, "chunk_loss": 0.030867, "F": 0.21722, "G": 0.24037, "chunks": 222.44, "bytes_per_chunk": 4.604, "boundary_entropy": 0.19501, "router_noise": 0.0, "grad_norm": 0.9006, "lr": 0.0002819, "router_trainable": true, "sec_per_step": 0.072, "alpha": 0.03, "t": 112.6} |
| {"event": "train", "step": 1300, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.60648, "bpb": 2.31766, "chunk_loss": 0.030794, "F": 0.21558, "G": 0.2379, "chunks": 220.75, "bytes_per_chunk": 4.639, "boundary_entropy": 0.19237, "router_noise": 0.0, "grad_norm": 0.846, "lr": 0.00028, "router_trainable": true, "sec_per_step": 0.073, "alpha": 0.03, "t": 116.2} |
| {"event": "train", "step": 1350, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.6377, "bpb": 2.36271, "chunk_loss": 0.031114, "F": 0.22632, "G": 0.24719, "chunks": 231.75, "bytes_per_chunk": 4.419, "boundary_entropy": 0.1925, "router_noise": 0.0, "grad_norm": 0.915, "lr": 0.000278, "router_trainable": true, "sec_per_step": 0.075, "alpha": 0.03, "t": 119.9} |
| {"event": "probe", "step": 1400, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3851.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.221284, "cps": 30.085938, "mean_p": 0.238339, "boundary_entropy": 0.184943, "bpb": 2.21726}, "tr": {"chunks": 4326.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.238966, "cps": 33.796875, "mean_p": 0.243551, "boundary_entropy": 0.223664, "bpb": 2.494037}, "fi": {"chunks": 4781.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.25416, "cps": 37.351562, "mean_p": 0.267409, "boundary_entropy": 0.213245, "bpb": 2.668194}}, "bpb": {"en": 2.21726, "tr": 2.49404, "fi": 2.66819}, "gini_cpc": 0.030679, "gini_cps": 0.047847, "bpic": 4.378145, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0799075195855883, "fi": 1.1485699821956192}, "t": 123.9} |
| {"event": "eval", "step": 1400, "phase": 1, "bpb": {"en": 2.23001, "tr": 2.5029, "fi": 2.67136}, "bytes_per_chunk": {"en": 4.534, "tr": 4.879, "fi": 4.347}, "bpb_mean": 2.46809, "frozen_inventory": false, "t": 124.4} |
| {"event": "train", "step": 1400, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.61891, "bpb": 2.33559, "chunk_loss": 0.030818, "F": 0.21759, "G": 0.23576, "chunks": 222.81, "bytes_per_chunk": 4.596, "boundary_entropy": 0.18556, "router_noise": 0.0, "grad_norm": 0.8906, "lr": 0.0002759, "router_trainable": true, "sec_per_step": 0.095, "alpha": 0.03, "t": 124.7} |
| {"event": "train", "step": 1450, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.6119, "bpb": 2.32548, "chunk_loss": 0.030923, "F": 0.2222, "G": 0.2407, "chunks": 227.53, "bytes_per_chunk": 4.5, "boundary_entropy": 0.18891, "router_noise": 0.0, "grad_norm": 0.7766, "lr": 0.0002738, "router_trainable": true, "sec_per_step": 0.076, "alpha": 0.03, "t": 128.5} |
| {"event": "train", "step": 1500, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.6065, "bpb": 2.31769, "chunk_loss": 0.030966, "F": 0.22159, "G": 0.23918, "chunks": 226.91, "bytes_per_chunk": 4.513, "boundary_entropy": 0.18781, "router_noise": 0.0, "grad_norm": 0.843, "lr": 0.0002715, "router_trainable": true, "sec_per_step": 0.07, "alpha": 0.03, "t": 132.0} |
| {"event": "train", "step": 1550, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.47102, "bpb": 2.12224, "chunk_loss": 0.030825, "F": 0.21954, "G": 0.2356, "chunks": 224.81, "bytes_per_chunk": 4.555, "boundary_entropy": 0.17375, "router_noise": 0.0, "grad_norm": 0.8154, "lr": 0.0002692, "router_trainable": true, "sec_per_step": 0.072, "alpha": 0.03, "t": 135.5} |
| {"event": "probe", "step": 1600, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3817.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.21933, "cps": 29.820312, "mean_p": 0.233668, "boundary_entropy": 0.168417, "bpb": 2.178807}, "tr": {"chunks": 4183.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.231067, "cps": 32.679688, "mean_p": 0.239096, "boundary_entropy": 0.212745, "bpb": 2.429101}, "fi": {"chunks": 4677.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.248631, "cps": 36.539062, "mean_p": 0.262358, "boundary_entropy": 0.20038, "bpb": 2.602913}}, "bpb": {"en": 2.17881, "tr": 2.4291, "fi": 2.60291}, "gini_cpc": 0.027945, "gini_cps": 0.045226, "bpic": 4.475191, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0535114823624967, "fi": 1.13359376026048}, "t": 139.4} |
| {"event": "eval", "step": 1600, "phase": 1, "bpb": {"en": 2.18826, "tr": 2.43932, "fi": 2.60921}, "bytes_per_chunk": {"en": 4.533, "tr": 4.909, "fi": 4.422}, "bpb_mean": 2.41226, "frozen_inventory": false, "t": 140.0} |
| {"event": "train", "step": 1600, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.58122, "bpb": 2.28122, "chunk_loss": 0.030889, "F": 0.22266, "G": 0.23761, "chunks": 228.0, "bytes_per_chunk": 4.491, "boundary_entropy": 0.16718, "router_noise": 0.0, "grad_norm": 0.8774, "lr": 0.0002668, "router_trainable": true, "sec_per_step": 0.095, "alpha": 0.03, "t": 140.3} |
| {"event": "train", "step": 1650, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.51215, "bpb": 2.18157, "chunk_loss": 0.03064, "F": 0.21194, "G": 0.22652, "chunks": 217.03, "bytes_per_chunk": 4.718, "boundary_entropy": 0.17508, "router_noise": 0.0, "grad_norm": 0.8542, "lr": 0.0002643, "router_trainable": true, "sec_per_step": 0.072, "alpha": 0.03, "t": 143.9} |
| {"event": "train", "step": 1700, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.56316, "bpb": 2.25516, "chunk_loss": 0.030761, "F": 0.21744, "G": 0.23267, "chunks": 222.66, "bytes_per_chunk": 4.599, "boundary_entropy": 0.17161, "router_noise": 0.0, "grad_norm": 0.7979, "lr": 0.0002618, "router_trainable": true, "sec_per_step": 0.073, "alpha": 0.03, "t": 147.6} |
| {"event": "train", "step": 1750, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.56172, "bpb": 2.25309, "chunk_loss": 0.030804, "F": 0.21817, "G": 0.23272, "chunks": 223.41, "bytes_per_chunk": 4.584, "boundary_entropy": 0.17201, "router_noise": 0.0, "grad_norm": 0.8341, "lr": 0.0002591, "router_trainable": true, "sec_per_step": 0.07, "alpha": 0.03, "t": 151.1} |
| {"event": "probe", "step": 1800, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3836.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.220422, "cps": 29.96875, "mean_p": 0.233185, "boundary_entropy": 0.161973, "bpb": 2.138875}, "tr": {"chunks": 4166.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.230128, "cps": 32.546875, "mean_p": 0.236209, "boundary_entropy": 0.206793, "bpb": 2.380716}, "fi": {"chunks": 4671.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.248312, "cps": 36.492188, "mean_p": 0.259529, "boundary_entropy": 0.190013, "bpb": 2.529241}}, "bpb": {"en": 2.13887, "tr": 2.38072, "fi": 2.52924}, "gini_cpc": 0.026606, "gini_cps": 0.043925, "bpic": 4.476604, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.044033023406729, "fi": 1.1265319295739649}, "t": 155.0} |
| {"event": "eval", "step": 1800, "phase": 1, "bpb": {"en": 2.14988, "tr": 2.3904, "fi": 2.53924}, "bytes_per_chunk": {"en": 4.542, "tr": 4.985, "fi": 4.453}, "bpb_mean": 2.35984, "frozen_inventory": false, "t": 155.5} |
| {"event": "train", "step": 1800, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.52699, "bpb": 2.20297, "chunk_loss": 0.030799, "F": 0.21756, "G": 0.23265, "chunks": 222.78, "bytes_per_chunk": 4.596, "boundary_entropy": 0.16894, "router_noise": 0.0, "grad_norm": 0.8022, "lr": 0.0002564, "router_trainable": true, "sec_per_step": 0.095, "alpha": 0.03, "t": 155.8} |
| {"event": "train", "step": 1850, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.54372, "bpb": 2.22711, "chunk_loss": 0.030645, "F": 0.21304, "G": 0.2255, "chunks": 218.16, "bytes_per_chunk": 4.694, "boundary_entropy": 0.16078, "router_noise": 0.0, "grad_norm": 0.7009, "lr": 0.0002537, "router_trainable": true, "sec_per_step": 0.076, "alpha": 0.03, "t": 159.6} |
| {"event": "train", "step": 1900, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.46464, "bpb": 2.11303, "chunk_loss": 0.030936, "F": 0.22406, "G": 0.23803, "chunks": 229.44, "bytes_per_chunk": 4.463, "boundary_entropy": 0.15863, "router_noise": 0.0, "grad_norm": 0.8038, "lr": 0.0002508, "router_trainable": true, "sec_per_step": 0.073, "alpha": 0.03, "t": 163.3} |
| {"event": "train", "step": 1950, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.46679, "bpb": 2.11613, "chunk_loss": 0.030675, "F": 0.2139, "G": 0.22554, "chunks": 219.03, "bytes_per_chunk": 4.675, "boundary_entropy": 0.15752, "router_noise": 0.0, "grad_norm": 0.716, "lr": 0.0002479, "router_trainable": true, "sec_per_step": 0.072, "alpha": 0.03, "t": 166.9} |
| {"event": "probe", "step": 2000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3823.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.219675, "cps": 29.867188, "mean_p": 0.230557, "boundary_entropy": 0.150627, "bpb": 2.110459}, "tr": {"chunks": 4160.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.229796, "cps": 32.5, "mean_p": 0.22909, "boundary_entropy": 0.200123, "bpb": 2.336573}, "fi": {"chunks": 4772.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.253681, "cps": 37.28125, "mean_p": 0.260754, "boundary_entropy": 0.177084, "bpb": 2.49318}}, "bpb": {"en": 2.11046, "tr": 2.33657, "fi": 2.49318}, "gini_cpc": 0.032242, "gini_cps": 0.049601, "bpic": 4.447824, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.046074466003954, "fi": 1.154804250544741}, "t": 170.7} |
| {"event": "eval", "step": 2000, "phase": 1, "bpb": {"en": 2.12243, "tr": 2.34861, "fi": 2.49705}, "bytes_per_chunk": {"en": 4.554, "tr": 4.986, "fi": 4.364}, "bpb_mean": 2.32269, "frozen_inventory": false, "t": 171.3} |
| {"event": "train", "step": 2000, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.58608, "bpb": 2.28823, "chunk_loss": 0.030896, "F": 0.22168, "G": 0.2345, "chunks": 227.0, "bytes_per_chunk": 4.511, "boundary_entropy": 0.15973, "router_noise": 0.0, "grad_norm": 0.7855, "lr": 0.000245, "router_trainable": true, "sec_per_step": 0.093, "alpha": 0.03, "t": 171.5} |
| {"event": "probe", "step": 2048, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3813.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.2191, "cps": 29.789062, "mean_p": 0.229936, "boundary_entropy": 0.155738, "bpb": 2.104178}, "tr": {"chunks": 4050.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.22372, "cps": 31.640625, "mean_p": 0.225908, "boundary_entropy": 0.208451, "bpb": 2.33261}, "fi": {"chunks": 4651.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.247249, "cps": 36.335938, "mean_p": 0.257576, "boundary_entropy": 0.180782, "bpb": 2.465753}}, "bpb": {"en": 2.10418, "tr": 2.33261, "fi": 2.46575}, "gini_cpc": 0.027194, "gini_cps": 0.044643, "bpic": 4.533482, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0210847422474933, "fi": 1.1284745550180915}, "t": 175.4} |
| {"event": "train", "step": 2050, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.53317, "bpb": 2.2119, "chunk_loss": 0.030933, "F": 0.22281, "G": 0.23426, "chunks": 228.16, "bytes_per_chunk": 4.488, "boundary_entropy": 0.15356, "router_noise": 0.0, "grad_norm": 0.7629, "lr": 0.0002419, "router_trainable": true, "sec_per_step": 0.086, "alpha": 0.03, "t": 175.8} |
| {"event": "train", "step": 2100, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.55831, "bpb": 2.24816, "chunk_loss": 0.030809, "F": 0.22015, "G": 0.23089, "chunks": 225.44, "bytes_per_chunk": 4.542, "boundary_entropy": 0.15558, "router_noise": 0.0, "grad_norm": 0.8289, "lr": 0.0002388, "router_trainable": true, "sec_per_step": 0.072, "alpha": 0.03, "t": 179.5} |
| {"event": "train", "step": 2150, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.53824, "bpb": 2.21921, "chunk_loss": 0.031, "F": 0.22766, "G": 0.23706, "chunks": 233.12, "bytes_per_chunk": 4.392, "boundary_entropy": 0.15489, "router_noise": 0.0, "grad_norm": 0.7084, "lr": 0.0002357, "router_trainable": true, "sec_per_step": 0.072, "alpha": 0.03, "t": 183.0} |
| {"event": "probe", "step": 2200, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3801.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.218411, "cps": 29.695312, "mean_p": 0.227532, "boundary_entropy": 0.152213, "bpb": 2.087125}, "tr": {"chunks": 4031.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.22267, "cps": 31.492188, "mean_p": 0.221232, "boundary_entropy": 0.202408, "bpb": 2.295464}, "fi": {"chunks": 4696.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.249641, "cps": 36.6875, "mean_p": 0.255839, "boundary_entropy": 0.169985, "bpb": 2.451229}}, "bpb": {"en": 2.08712, "tr": 2.29546, "fi": 2.45123}, "gini_cpc": 0.030143, "gini_cps": 0.047627, "bpic": 4.528416, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.019502974756263, "fi": 1.1429900647981877}, "t": 187.0} |
| {"event": "eval", "step": 2200, "phase": 1, "bpb": {"en": 2.09723, "tr": 2.30582, "fi": 2.45758}, "bytes_per_chunk": {"en": 4.552, "tr": 5.102, "fi": 4.407}, "bpb_mean": 2.28687, "frozen_inventory": false, "t": 187.6} |
| {"event": "train", "step": 2200, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.60263, "bpb": 2.3121, "chunk_loss": 0.031119, "F": 0.22452, "G": 0.23415, "chunks": 229.91, "bytes_per_chunk": 4.454, "boundary_entropy": 0.16283, "router_noise": 0.0, "grad_norm": 0.7728, "lr": 0.0002325, "router_trainable": true, "sec_per_step": 0.095, "alpha": 0.03, "t": 187.8} |
| {"event": "train", "step": 2250, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.5918, "bpb": 2.29648, "chunk_loss": 0.030807, "F": 0.21945, "G": 0.22903, "chunks": 224.72, "bytes_per_chunk": 4.557, "boundary_entropy": 0.15929, "router_noise": 0.0, "grad_norm": 0.7833, "lr": 0.0002293, "router_trainable": true, "sec_per_step": 0.073, "alpha": 0.03, "t": 191.5} |
| {"event": "train", "step": 2300, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.51974, "bpb": 2.19251, "chunk_loss": 0.030967, "F": 0.22656, "G": 0.23458, "chunks": 232.0, "bytes_per_chunk": 4.414, "boundary_entropy": 0.15399, "router_noise": 0.0, "grad_norm": 0.7154, "lr": 0.000226, "router_trainable": true, "sec_per_step": 0.074, "alpha": 0.03, "t": 195.2} |
| {"event": "train", "step": 2350, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.40545, "bpb": 2.02763, "chunk_loss": 0.030847, "F": 0.22165, "G": 0.22983, "chunks": 226.97, "bytes_per_chunk": 4.512, "boundary_entropy": 0.15169, "router_noise": 0.0, "grad_norm": 0.6693, "lr": 0.0002226, "router_trainable": true, "sec_per_step": 0.072, "alpha": 0.03, "t": 198.8} |
| {"event": "probe", "step": 2400, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3827.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.219905, "cps": 29.898438, "mean_p": 0.226652, "boundary_entropy": 0.141975, "bpb": 2.067754}, "tr": {"chunks": 4070.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.224825, "cps": 31.796875, "mean_p": 0.217027, "boundary_entropy": 0.196142, "bpb": 2.266586}, "fi": {"chunks": 4644.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.246877, "cps": 36.28125, "mean_p": 0.254007, "boundary_entropy": 0.166681, "bpb": 2.419238}}, "bpb": {"en": 2.06775, "tr": 2.26659, "fi": 2.41924}, "gini_cpc": 0.026, "gini_cps": 0.043431, "bpic": 4.523722, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.022373339353718, "fi": 1.122654148690194}, "t": 202.7} |
| {"event": "eval", "step": 2400, "phase": 1, "bpb": {"en": 2.0785, "tr": 2.27845, "fi": 2.42858}, "bytes_per_chunk": {"en": 4.54, "tr": 5.026, "fi": 4.467}, "bpb_mean": 2.26184, "frozen_inventory": false, "t": 203.2} |
| {"event": "phase_transition", "step": 2400, "phase": 2, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0, "t": 203.5} |
| {"event": "train", "step": 2400, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.47911, "bpb": 2.13391, "chunk_loss": 0.030731, "F": 0.21844, "G": 0.22515, "chunks": 223.69, "bytes_per_chunk": 4.578, "boundary_entropy": 0.14674, "router_noise": 0.0, "grad_norm": 0.7126, "lr": 0.0002192, "router_trainable": true, "sec_per_step": 0.095, "alpha": 0.03, "t": 203.5} |
| {"event": "train", "step": 2450, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.50329, "bpb": 2.16879, "chunk_loss": 0.03083, "F": 0.22156, "G": 0.2299, "chunks": 226.88, "bytes_per_chunk": 4.513, "boundary_entropy": 0.15256, "router_noise": 0.0, "grad_norm": 0.7341, "lr": 0.0002158, "router_trainable": true, "sec_per_step": 0.072, "alpha": 0.03, "t": 207.1} |
| {"event": "train", "step": 2500, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.41199, "bpb": 2.03707, "chunk_loss": 0.030845, "F": 0.22131, "G": 0.22813, "chunks": 226.62, "bytes_per_chunk": 4.518, "boundary_entropy": 0.14645, "router_noise": 0.0, "grad_norm": 0.7712, "lr": 0.0002123, "router_trainable": true, "sec_per_step": 0.074, "alpha": 0.03, "t": 210.9} |
| {"event": "train", "step": 2550, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.44851, "bpb": 2.08977, "chunk_loss": 0.03076, "F": 0.21368, "G": 0.21999, "chunks": 218.81, "bytes_per_chunk": 4.68, "boundary_entropy": 0.1455, "router_noise": 0.0, "grad_norm": 0.7405, "lr": 0.0002088, "router_trainable": true, "sec_per_step": 0.071, "alpha": 0.03, "t": 214.4} |
| {"event": "probe", "step": 2600, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3756.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.215825, "cps": 29.34375, "mean_p": 0.223696, "boundary_entropy": 0.140663, "bpb": 2.040232}, "tr": {"chunks": 4068.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.224714, "cps": 31.78125, "mean_p": 0.219191, "boundary_entropy": 0.18791, "bpb": 2.23172}, "fi": {"chunks": 4742.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.252087, "cps": 37.046875, "mean_p": 0.254834, "boundary_entropy": 0.156843, "bpb": 2.370719}}, "bpb": {"en": 2.04023, "tr": 2.23172, "fi": 2.37072}, "gini_cpc": 0.034903, "gini_cps": 0.05231, "bpic": 4.514722, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.041187461383115, "fi": 1.1680144154659016}, "t": 218.5} |
| {"event": "eval", "step": 2600, "phase": 2, "bpb": {"en": 2.05102, "tr": 2.24387, "fi": 2.37811}, "bytes_per_chunk": {"en": 4.62, "tr": 5.099, "fi": 4.436}, "bpb_mean": 2.22433, "frozen_inventory": false, "t": 219.0} |
| {"event": "train", "step": 2600, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.41728, "bpb": 2.0447, "chunk_loss": 0.030647, "F": 0.21469, "G": 0.22059, "chunks": 219.84, "bytes_per_chunk": 4.658, "boundary_entropy": 0.14493, "router_noise": 0.0, "grad_norm": 0.6194, "lr": 0.0002053, "router_trainable": true, "sec_per_step": 0.097, "alpha": 0.03, "t": 219.3} |
| {"event": "train", "step": 2650, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.40368, "bpb": 2.02508, "chunk_loss": 0.030781, "F": 0.22018, "G": 0.22604, "chunks": 225.47, "bytes_per_chunk": 4.542, "boundary_entropy": 0.14093, "router_noise": 0.0, "grad_norm": 0.6633, "lr": 0.0002017, "router_trainable": true, "sec_per_step": 0.073, "alpha": 0.03, "t": 222.9} |
| {"event": "train", "step": 2700, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.51756, "bpb": 2.18937, "chunk_loss": 0.03093, "F": 0.22324, "G": 0.23019, "chunks": 228.59, "bytes_per_chunk": 4.48, "boundary_entropy": 0.14643, "router_noise": 0.0, "grad_norm": 0.6722, "lr": 0.0001981, "router_trainable": true, "sec_per_step": 0.071, "alpha": 0.03, "t": 226.5} |
| {"event": "train", "step": 2750, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.44422, "bpb": 2.08356, "chunk_loss": 0.030767, "F": 0.21951, "G": 0.22611, "chunks": 224.78, "bytes_per_chunk": 4.556, "boundary_entropy": 0.14425, "router_noise": 0.0, "grad_norm": 0.659, "lr": 0.0001945, "router_trainable": true, "sec_per_step": 0.072, "alpha": 0.03, "t": 230.0} |
| {"event": "probe", "step": 2800, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3793.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.217951, "cps": 29.632812, "mean_p": 0.226602, "boundary_entropy": 0.136126, "bpb": 2.02172}, "tr": {"chunks": 3813.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.210628, "cps": 29.789062, "mean_p": 0.214486, "boundary_entropy": 0.186758, "bpb": 2.217561}, "fi": {"chunks": 4663.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.247887, "cps": 36.429688, "mean_p": 0.251332, "boundary_entropy": 0.150508, "bpb": 2.348609}}, "bpb": {"en": 2.02172, "tr": 2.21756, "fi": 2.34861}, "gini_cpc": 0.036719, "gini_cps": 0.047274, "bpic": 4.624012, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.966401357530558, "fi": 1.1373517744321198}, "t": 234.2} |
| {"event": "eval", "step": 2800, "phase": 2, "bpb": {"en": 2.036, "tr": 2.22597, "fi": 2.35464}, "bytes_per_chunk": {"en": 4.587, "tr": 5.442, "fi": 4.482}, "bpb_mean": 2.20554, "frozen_inventory": false, "t": 234.6} |
| {"event": "train", "step": 2800, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.46453, "bpb": 2.11287, "chunk_loss": 0.030951, "F": 0.22134, "G": 0.23007, "chunks": 226.66, "bytes_per_chunk": 4.518, "boundary_entropy": 0.14136, "router_noise": 0.0, "grad_norm": 0.6966, "lr": 0.0001909, "router_trainable": true, "sec_per_step": 0.096, "alpha": 0.03, "t": 234.9} |
| {"event": "train", "step": 2850, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.43335, "bpb": 2.06788, "chunk_loss": 0.030548, "F": 0.20993, "G": 0.21876, "chunks": 214.97, "bytes_per_chunk": 4.763, "boundary_entropy": 0.14054, "router_noise": 0.0, "grad_norm": 0.7237, "lr": 0.0001872, "router_trainable": true, "sec_per_step": 0.073, "alpha": 0.03, "t": 238.5} |
| {"event": "train", "step": 2900, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.44613, "bpb": 2.08633, "chunk_loss": 0.030765, "F": 0.21838, "G": 0.22557, "chunks": 223.62, "bytes_per_chunk": 4.579, "boundary_entropy": 0.13661, "router_noise": 0.0, "grad_norm": 0.7222, "lr": 0.0001835, "router_trainable": true, "sec_per_step": 0.071, "alpha": 0.03, "t": 242.0} |
| {"event": "train", "step": 2950, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.52015, "bpb": 2.19311, "chunk_loss": 0.030935, "F": 0.2229, "G": 0.23006, "chunks": 228.25, "bytes_per_chunk": 4.486, "boundary_entropy": 0.1409, "router_noise": 0.0, "grad_norm": 0.7568, "lr": 0.0001799, "router_trainable": true, "sec_per_step": 0.072, "alpha": 0.03, "t": 245.6} |
| {"event": "probe", "step": 3000, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3820.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.219502, "cps": 29.84375, "mean_p": 0.224401, "boundary_entropy": 0.129541, "bpb": 2.006801}, "tr": {"chunks": 4069.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.224769, "cps": 31.789062, "mean_p": 0.216619, "boundary_entropy": 0.17763, "bpb": 2.193297}, "fi": {"chunks": 4658.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.247621, "cps": 36.390625, "mean_p": 0.250932, "boundary_entropy": 0.146852, "bpb": 2.310397}}, "bpb": {"en": 2.0068, "tr": 2.1933, "fi": 2.3104}, "gini_cpc": 0.027093, "gini_cps": 0.044526, "bpic": 4.521559, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0239951406625207, "fi": 1.1281019710813072}, "t": 249.6} |
| {"event": "eval", "step": 3000, "phase": 2, "bpb": {"en": 2.01943, "tr": 2.20008, "fi": 2.31847}, "bytes_per_chunk": {"en": 4.56, "tr": 4.959, "fi": 4.45}, "bpb_mean": 2.17933, "frozen_inventory": false, "t": 250.2} |
| {"event": "train", "step": 3000, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.40901, "bpb": 2.03277, "chunk_loss": 0.030854, "F": 0.2236, "G": 0.22844, "chunks": 228.97, "bytes_per_chunk": 4.472, "boundary_entropy": 0.1292, "router_noise": 0.0, "grad_norm": 0.6992, "lr": 0.0001761, "router_trainable": true, "sec_per_step": 0.097, "alpha": 0.03, "t": 250.5} |
| {"event": "train", "step": 3050, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.44374, "bpb": 2.08288, "chunk_loss": 0.030818, "F": 0.22052, "G": 0.22834, "chunks": 225.81, "bytes_per_chunk": 4.535, "boundary_entropy": 0.139, "router_noise": 0.0, "grad_norm": 0.6835, "lr": 0.0001724, "router_trainable": true, "sec_per_step": 0.075, "alpha": 0.03, "t": 254.2} |
| {"event": "train", "step": 3100, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.48928, "bpb": 2.14858, "chunk_loss": 0.031102, "F": 0.22498, "G": 0.23339, "chunks": 230.38, "bytes_per_chunk": 4.445, "boundary_entropy": 0.13309, "router_noise": 0.0, "grad_norm": 0.6978, "lr": 0.0001687, "router_trainable": true, "sec_per_step": 0.073, "alpha": 0.03, "t": 257.9} |
| {"event": "train", "step": 3150, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.44951, "bpb": 2.09121, "chunk_loss": 0.030693, "F": 0.21735, "G": 0.22357, "chunks": 222.56, "bytes_per_chunk": 4.601, "boundary_entropy": 0.13103, "router_noise": 0.0, "grad_norm": 0.6958, "lr": 0.000165, "router_trainable": true, "sec_per_step": 0.071, "alpha": 0.03, "t": 261.4} |
| {"event": "probe", "step": 3200, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3638.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.209044, "cps": 28.421875, "mean_p": 0.220708, "boundary_entropy": 0.131007, "bpb": 1.994656}, "tr": {"chunks": 3581.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.197813, "cps": 27.976562, "mean_p": 0.211786, "boundary_entropy": 0.176352, "bpb": 2.173651}, "fi": {"chunks": 4641.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.246717, "cps": 36.257812, "mean_p": 0.248765, "boundary_entropy": 0.145299, "bpb": 2.296516}}, "bpb": {"en": 1.99466, "tr": 2.17365, "fi": 2.29652}, "gini_cpc": 0.049884, "gini_cps": 0.059584, "bpic": 4.783474, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9462702688060383, "fi": 1.1802149468122896}, "t": 265.4} |
| {"event": "eval", "step": 3200, "phase": 2, "bpb": {"en": 2.00774, "tr": 2.1818, "fi": 2.30375}, "bytes_per_chunk": {"en": 4.771, "tr": 5.715, "fi": 4.478}, "bpb_mean": 2.16443, "frozen_inventory": false, "t": 265.9} |
| {"event": "train", "step": 3200, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.45428, "bpb": 2.09809, "chunk_loss": 0.030671, "F": 0.21194, "G": 0.22404, "chunks": 217.03, "bytes_per_chunk": 4.718, "boundary_entropy": 0.13507, "router_noise": 0.0, "grad_norm": 0.6734, "lr": 0.0001613, "router_trainable": true, "sec_per_step": 0.095, "alpha": 0.03, "t": 266.2} |
| {"event": "train", "step": 3250, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.4206, "bpb": 2.04949, "chunk_loss": 0.030475, "F": 0.20502, "G": 0.21653, "chunks": 209.94, "bytes_per_chunk": 4.878, "boundary_entropy": 0.12523, "router_noise": 0.0, "grad_norm": 0.651, "lr": 0.0001576, "router_trainable": true, "sec_per_step": 0.071, "alpha": 0.03, "t": 269.8} |
| {"event": "train", "step": 3300, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.38081, "bpb": 1.99209, "chunk_loss": 0.030617, "F": 0.21078, "G": 0.22066, "chunks": 215.84, "bytes_per_chunk": 4.744, "boundary_entropy": 0.12119, "router_noise": 0.0, "grad_norm": 0.687, "lr": 0.0001539, "router_trainable": true, "sec_per_step": 0.075, "alpha": 0.03, "t": 273.5} |
| {"event": "train", "step": 3350, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.36633, "bpb": 1.97119, "chunk_loss": 0.030422, "F": 0.20135, "G": 0.21349, "chunks": 206.19, "bytes_per_chunk": 4.966, "boundary_entropy": 0.12653, "router_noise": 0.0, "grad_norm": 0.6676, "lr": 0.0001501, "router_trainable": true, "sec_per_step": 0.074, "alpha": 0.03, "t": 277.2} |
| {"event": "probe", "step": 3400, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3600.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.206861, "cps": 28.125, "mean_p": 0.21747, "boundary_entropy": 0.119688, "bpb": 1.981179}, "tr": {"chunks": 3579.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.197702, "cps": 27.960938, "mean_p": 0.210913, "boundary_entropy": 0.172536, "bpb": 2.152961}, "fi": {"chunks": 4694.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.249535, "cps": 36.671875, "mean_p": 0.251742, "boundary_entropy": 0.135247, "bpb": 2.268598}}, "bpb": {"en": 1.98118, "tr": 2.15296, "fi": 2.2686}, "gini_cpc": 0.052829, "gini_cps": 0.062607, "bpic": 4.778236, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9557246036568524, "fi": 1.2062930377615932}, "t": 281.0} |
| {"event": "eval", "step": 3400, "phase": 2, "bpb": {"en": 1.9941, "tr": 2.16169, "fi": 2.27682}, "bytes_per_chunk": {"en": 4.803, "tr": 5.79, "fi": 4.449}, "bpb_mean": 2.1442, "frozen_inventory": false, "t": 281.6} |
| {"event": "train", "step": 3400, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.45297, "bpb": 2.09619, "chunk_loss": 0.030634, "F": 0.20944, "G": 0.22083, "chunks": 214.47, "bytes_per_chunk": 4.775, "boundary_entropy": 0.12733, "router_noise": 0.0, "grad_norm": 0.6913, "lr": 0.0001465, "router_trainable": true, "sec_per_step": 0.094, "alpha": 0.03, "t": 281.9} |
| {"event": "train", "step": 3450, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.42143, "bpb": 2.05069, "chunk_loss": 0.030433, "F": 0.20245, "G": 0.21448, "chunks": 207.31, "bytes_per_chunk": 4.939, "boundary_entropy": 0.12698, "router_noise": 0.0, "grad_norm": 0.6786, "lr": 0.0001428, "router_trainable": true, "sec_per_step": 0.071, "alpha": 0.03, "t": 285.4} |
| {"event": "train", "step": 3500, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.40594, "bpb": 2.02835, "chunk_loss": 0.030483, "F": 0.20529, "G": 0.21528, "chunks": 210.22, "bytes_per_chunk": 4.871, "boundary_entropy": 0.11721, "router_noise": 0.0, "grad_norm": 0.6854, "lr": 0.0001391, "router_trainable": true, "sec_per_step": 0.071, "alpha": 0.03, "t": 289.0} |
| {"event": "train", "step": 3550, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.42712, "bpb": 2.0589, "chunk_loss": 0.030569, "F": 0.20624, "G": 0.21882, "chunks": 211.19, "bytes_per_chunk": 4.849, "boundary_entropy": 0.12837, "router_noise": 0.0, "grad_norm": 0.6886, "lr": 0.0001355, "router_trainable": true, "sec_per_step": 0.07, "alpha": 0.03, "t": 292.5} |
| {"event": "probe", "step": 3600, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3620.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.20801, "cps": 28.28125, "mean_p": 0.216718, "boundary_entropy": 0.112584, "bpb": 1.967886}, "tr": {"chunks": 3728.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.205933, "cps": 29.125, "mean_p": 0.211617, "boundary_entropy": 0.16837, "bpb": 2.127961}, "fi": {"chunks": 4719.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.250864, "cps": 36.867188, "mean_p": 0.250188, "boundary_entropy": 0.131871, "bpb": 2.246472}}, "bpb": {"en": 1.96789, "tr": 2.12796, "fi": 2.24647}, "gini_cpc": 0.045057, "gini_cps": 0.060717, "bpic": 4.701417, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9900130102669105, "fi": 1.2060175940314692}, "t": 296.5} |
| {"event": "eval", "step": 3600, "phase": 2, "bpb": {"en": 1.98158, "tr": 2.13669, "fi": 2.25524}, "bytes_per_chunk": {"en": 4.78, "tr": 5.549, "fi": 4.417}, "bpb_mean": 2.1245, "frozen_inventory": false, "t": 297.0} |
| {"event": "train", "step": 3600, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.43836, "bpb": 2.07512, "chunk_loss": 0.030415, "F": 0.19965, "G": 0.21332, "chunks": 204.44, "bytes_per_chunk": 5.009, "boundary_entropy": 0.13304, "router_noise": 0.0, "grad_norm": 0.7081, "lr": 0.0001319, "router_trainable": true, "sec_per_step": 0.097, "alpha": 0.03, "t": 297.3} |
| {"event": "train", "step": 3650, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.44049, "bpb": 2.07819, "chunk_loss": 0.030779, "F": 0.21582, "G": 0.22622, "chunks": 221.0, "bytes_per_chunk": 4.633, "boundary_entropy": 0.12496, "router_noise": 0.0, "grad_norm": 0.7543, "lr": 0.0001283, "router_trainable": true, "sec_per_step": 0.072, "alpha": 0.03, "t": 300.9} |
| {"event": "train", "step": 3700, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.51805, "bpb": 2.19009, "chunk_loss": 0.030872, "F": 0.21945, "G": 0.2291, "chunks": 224.72, "bytes_per_chunk": 4.557, "boundary_entropy": 0.12427, "router_noise": 0.0, "grad_norm": 0.776, "lr": 0.0001247, "router_trainable": true, "sec_per_step": 0.073, "alpha": 0.03, "t": 304.6} |
| {"event": "train", "step": 3750, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.35934, "bpb": 1.96111, "chunk_loss": 0.030642, "F": 0.21118, "G": 0.22068, "chunks": 216.25, "bytes_per_chunk": 4.735, "boundary_entropy": 0.11474, "router_noise": 0.0, "grad_norm": 0.667, "lr": 0.0001212, "router_trainable": true, "sec_per_step": 0.071, "alpha": 0.03, "t": 308.2} |
| {"event": "probe", "step": 3800, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3588.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.206171, "cps": 28.03125, "mean_p": 0.215066, "boundary_entropy": 0.111894, "bpb": 1.954022}, "tr": {"chunks": 3595.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.198586, "cps": 28.085938, "mean_p": 0.21027, "boundary_entropy": 0.169134, "bpb": 2.10731}, "fi": {"chunks": 4734.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.251661, "cps": 36.984375, "mean_p": 0.251337, "boundary_entropy": 0.130049, "bpb": 2.227732}}, "bpb": {"en": 1.95402, "tr": 2.10731, "fi": 2.22773}, "gini_cpc": 0.053904, "gini_cps": 0.06411, "bpic": 4.760594, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9632078849437732, "fi": 1.220641288479718}, "t": 312.1} |
| {"event": "eval", "step": 3800, "phase": 2, "bpb": {"en": 1.96813, "tr": 2.11732, "fi": 2.23651}, "bytes_per_chunk": {"en": 4.798, "tr": 5.687, "fi": 4.413}, "bpb_mean": 2.10732, "frozen_inventory": false, "t": 312.6} |
| {"event": "train", "step": 3800, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.39678, "bpb": 2.01513, "chunk_loss": 0.030425, "F": 0.20074, "G": 0.21203, "chunks": 205.56, "bytes_per_chunk": 4.981, "boundary_entropy": 0.12088, "router_noise": 0.0, "grad_norm": 0.6479, "lr": 0.0001177, "router_trainable": true, "sec_per_step": 0.095, "alpha": 0.03, "t": 312.9} |
| {"event": "train", "step": 3850, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.44719, "bpb": 2.08785, "chunk_loss": 0.030431, "F": 0.20178, "G": 0.2135, "chunks": 206.62, "bytes_per_chunk": 4.956, "boundary_entropy": 0.12782, "router_noise": 0.0, "grad_norm": 0.8043, "lr": 0.0001142, "router_trainable": true, "sec_per_step": 0.073, "alpha": 0.03, "t": 316.5} |
| {"event": "train", "step": 3900, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.41326, "bpb": 2.03891, "chunk_loss": 0.030556, "F": 0.20935, "G": 0.21865, "chunks": 214.38, "bytes_per_chunk": 4.777, "boundary_entropy": 0.11372, "router_noise": 0.0, "grad_norm": 0.6781, "lr": 0.0001108, "router_trainable": true, "sec_per_step": 0.074, "alpha": 0.03, "t": 320.2} |
| {"event": "train", "step": 3950, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.30688, "bpb": 1.88542, "chunk_loss": 0.030479, "F": 0.20547, "G": 0.21412, "chunks": 210.41, "bytes_per_chunk": 4.867, "boundary_entropy": 0.10728, "router_noise": 0.0, "grad_norm": 0.6246, "lr": 0.0001074, "router_trainable": true, "sec_per_step": 0.072, "alpha": 0.03, "t": 323.8} |
| {"event": "probe", "step": 4000, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3602.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.206976, "cps": 28.140625, "mean_p": 0.215546, "boundary_entropy": 0.105573, "bpb": 1.941153}, "tr": {"chunks": 3605.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.199138, "cps": 28.164062, "mean_p": 0.211586, "boundary_entropy": 0.164982, "bpb": 2.101188}, "fi": {"chunks": 4781.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.25416, "cps": 37.351562, "mean_p": 0.252615, "boundary_entropy": 0.12352, "bpb": 2.210654}}, "bpb": {"en": 1.94115, "tr": 2.10119, "fi": 2.21065}, "gini_cpc": 0.055554, "gini_cps": 0.065566, "bpic": 4.732399, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9621330413483484, "fi": 1.2279686289381817}, "t": 327.7} |
| {"event": "eval", "step": 4000, "phase": 2, "bpb": {"en": 1.95484, "tr": 2.10765, "fi": 2.21957}, "bytes_per_chunk": {"en": 4.775, "tr": 5.737, "fi": 4.389}, "bpb_mean": 2.09402, "frozen_inventory": false, "t": 328.2} |
| {"event": "train", "step": 4000, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.36644, "bpb": 1.97136, "chunk_loss": 0.030558, "F": 0.20831, "G": 0.21804, "chunks": 213.31, "bytes_per_chunk": 4.8, "boundary_entropy": 0.11617, "router_noise": 0.0, "grad_norm": 0.788, "lr": 0.000104, "router_trainable": true, "sec_per_step": 0.094, "alpha": 0.03, "t": 328.5} |
| {"event": "train", "step": 4050, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.4455, "bpb": 2.08542, "chunk_loss": 0.030599, "F": 0.20865, "G": 0.21707, "chunks": 213.66, "bytes_per_chunk": 4.793, "boundary_entropy": 0.11606, "router_noise": 0.0, "grad_norm": 0.7265, "lr": 0.0001007, "router_trainable": true, "sec_per_step": 0.071, "alpha": 0.03, "t": 332.1} |
| {"event": "probe", "step": 4096, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3606.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.207206, "cps": 28.171875, "mean_p": 0.215371, "boundary_entropy": 0.108592, "bpb": 1.934719}, "tr": {"chunks": 3660.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.202176, "cps": 28.59375, "mean_p": 0.214145, "boundary_entropy": 0.172633, "bpb": 2.099961}, "fi": {"chunks": 4688.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.249216, "cps": 36.625, "mean_p": 0.248295, "boundary_entropy": 0.131345, "bpb": 2.208041}}, "bpb": {"en": 1.93472, "tr": 2.09996, "fi": 2.20804}, "gini_cpc": 0.047616, "gini_cps": 0.060342, "bpic": 4.745859, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9757283681665175, "fi": 1.202746543226071}, "t": 335.6} |
| {"event": "train", "step": 4100, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.37577, "bpb": 1.98481, "chunk_loss": 0.030414, "F": 0.20129, "G": 0.21066, "chunks": 206.12, "bytes_per_chunk": 4.968, "boundary_entropy": 0.11052, "router_noise": 0.0, "grad_norm": 0.6757, "lr": 9.75e-05, "router_trainable": true, "sec_per_step": 0.083, "alpha": 0.03, "t": 336.2} |
| {"event": "train", "step": 4150, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.36363, "bpb": 1.9673, "chunk_loss": 0.030449, "F": 0.20242, "G": 0.21421, "chunks": 207.28, "bytes_per_chunk": 4.94, "boundary_entropy": 0.11666, "router_noise": 0.0, "grad_norm": 0.6929, "lr": 9.43e-05, "router_trainable": true, "sec_per_step": 0.077, "alpha": 0.03, "t": 340.1} |
| {"event": "probe", "step": 4200, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3611.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.207493, "cps": 28.210938, "mean_p": 0.216027, "boundary_entropy": 0.103106, "bpb": 1.930688}, "tr": {"chunks": 3624.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.200188, "cps": 28.3125, "mean_p": 0.211865, "boundary_entropy": 0.164166, "bpb": 2.08021}, "fi": {"chunks": 4747.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.252352, "cps": 37.085938, "mean_p": 0.249847, "boundary_entropy": 0.124908, "bpb": 2.194544}}, "bpb": {"en": 1.93069, "tr": 2.08021, "fi": 2.19454}, "gini_cpc": 0.052689, "gini_cps": 0.063206, "bpic": 4.734769, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9647932788916886, "fi": 1.2161971463252181}, "t": 343.9} |
| {"event": "eval", "step": 4200, "phase": 2, "bpb": {"en": 1.94491, "tr": 2.08767, "fi": 2.20115}, "bytes_per_chunk": {"en": 4.783, "tr": 5.655, "fi": 4.435}, "bpb_mean": 2.07791, "frozen_inventory": false, "t": 344.5} |
| {"event": "phase_transition", "step": 4200, "phase": 3, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0, "t": 344.6} |
| {"event": "train", "step": 4200, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.38449, "bpb": 1.9974, "chunk_loss": 0.030427, "F": 0.19955, "G": 0.21281, "chunks": 204.34, "bytes_per_chunk": 5.011, "boundary_entropy": 0.1189, "router_noise": 0.0, "grad_norm": 0.7338, "lr": 9.12e-05, "router_trainable": true, "sec_per_step": 0.092, "alpha": 0.03, "t": 344.7} |
| {"event": "train", "step": 4250, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.42163, "bpb": 2.05099, "chunk_loss": 0.030683, "F": 0.21198, "G": 0.22185, "chunks": 217.06, "bytes_per_chunk": 4.718, "boundary_entropy": 0.11161, "router_noise": 0.0, "grad_norm": 0.7027, "lr": 8.81e-05, "router_trainable": true, "sec_per_step": 0.073, "alpha": 0.03, "t": 348.3} |
| {"event": "train", "step": 4300, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.37843, "bpb": 1.98866, "chunk_loss": 0.030538, "F": 0.20657, "G": 0.21561, "chunks": 211.53, "bytes_per_chunk": 4.841, "boundary_entropy": 0.11192, "router_noise": 0.0, "grad_norm": 0.6747, "lr": 8.5e-05, "router_trainable": true, "sec_per_step": 0.074, "alpha": 0.03, "t": 352.0} |
| {"event": "train", "step": 4350, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.42051, "bpb": 2.04936, "chunk_loss": 0.030479, "F": 0.20288, "G": 0.21295, "chunks": 207.75, "bytes_per_chunk": 4.929, "boundary_entropy": 0.1117, "router_noise": 0.0, "grad_norm": 0.661, "lr": 8.21e-05, "router_trainable": true, "sec_per_step": 0.073, "alpha": 0.03, "t": 355.7} |
| {"event": "probe", "step": 4400, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3604.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.207091, "cps": 28.15625, "mean_p": 0.214907, "boundary_entropy": 0.102908, "bpb": 1.92043}, "tr": {"chunks": 3659.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.202121, "cps": 28.585938, "mean_p": 0.213001, "boundary_entropy": 0.165873, "bpb": 2.069843}, "fi": {"chunks": 4755.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.252778, "cps": 37.148438, "mean_p": 0.250735, "boundary_entropy": 0.123035, "bpb": 2.177826}}, "bpb": {"en": 1.92043, "tr": 2.06984, "fi": 2.17783}, "gini_cpc": 0.051014, "gini_cps": 0.063849, "bpic": 4.720586, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9760030974563301, "fi": 1.2206129569381412}, "t": 359.6} |
| {"event": "eval", "step": 4400, "phase": 3, "bpb": {"en": 1.93609, "tr": 2.07846, "fi": 2.18687}, "bytes_per_chunk": {"en": 4.796, "tr": 5.653, "fi": 4.446}, "bpb_mean": 2.06714, "frozen_inventory": false, "t": 360.2} |
| {"event": "train", "step": 4400, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.35617, "bpb": 1.95654, "chunk_loss": 0.030601, "F": 0.21033, "G": 0.21968, "chunks": 215.38, "bytes_per_chunk": 4.754, "boundary_entropy": 0.11029, "router_noise": 0.0, "grad_norm": 0.6452, "lr": 7.92e-05, "router_trainable": true, "sec_per_step": 0.095, "alpha": 0.03, "t": 360.5} |
| {"event": "train", "step": 4450, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.32642, "bpb": 1.91362, "chunk_loss": 0.030441, "F": 0.20242, "G": 0.21095, "chunks": 207.28, "bytes_per_chunk": 4.94, "boundary_entropy": 0.10432, "router_noise": 0.0, "grad_norm": 0.6535, "lr": 7.63e-05, "router_trainable": true, "sec_per_step": 0.075, "alpha": 0.03, "t": 364.2} |
| {"event": "train", "step": 4500, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.43604, "bpb": 2.07177, "chunk_loss": 0.030611, "F": 0.20959, "G": 0.21913, "chunks": 214.62, "bytes_per_chunk": 4.771, "boundary_entropy": 0.11068, "router_noise": 0.0, "grad_norm": 0.7366, "lr": 7.36e-05, "router_trainable": true, "sec_per_step": 0.074, "alpha": 0.03, "t": 367.9} |
| {"event": "train", "step": 4550, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.37684, "bpb": 1.98637, "chunk_loss": 0.030632, "F": 0.21136, "G": 0.21907, "chunks": 216.44, "bytes_per_chunk": 4.731, "boundary_entropy": 0.10679, "router_noise": 0.0, "grad_norm": 0.6947, "lr": 7.09e-05, "router_trainable": true, "sec_per_step": 0.073, "alpha": 0.03, "t": 371.5} |
| {"event": "probe", "step": 4600, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3596.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.206631, "cps": 28.09375, "mean_p": 0.215577, "boundary_entropy": 0.101591, "bpb": 1.912964}, "tr": {"chunks": 3554.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.196321, "cps": 27.765625, "mean_p": 0.210615, "boundary_entropy": 0.163652, "bpb": 2.0584}, "fi": {"chunks": 4703.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.250013, "cps": 36.742188, "mean_p": 0.24855, "boundary_entropy": 0.122027, "bpb": 2.16804}}, "bpb": {"en": 1.91296, "tr": 2.0584, "fi": 2.16804}, "gini_cpc": 0.054819, "gini_cps": 0.064625, "bpic": 4.786299, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9501043558866619, "fi": 1.209950302431736}, "t": 375.4} |
| {"event": "eval", "step": 4600, "phase": 3, "bpb": {"en": 1.92856, "tr": 2.06684, "fi": 2.17583}, "bytes_per_chunk": {"en": 4.798, "tr": 5.776, "fi": 4.464}, "bpb_mean": 2.05708, "frozen_inventory": false, "t": 376.0} |
| {"event": "train", "step": 4600, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.40438, "bpb": 2.02609, "chunk_loss": 0.03067, "F": 0.21252, "G": 0.21937, "chunks": 217.62, "bytes_per_chunk": 4.705, "boundary_entropy": 0.10431, "router_noise": 0.0, "grad_norm": 0.7154, "lr": 6.82e-05, "router_trainable": true, "sec_per_step": 0.095, "alpha": 0.03, "t": 376.3} |
| {"event": "train", "step": 4650, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.40553, "bpb": 2.02775, "chunk_loss": 0.03049, "F": 0.2045, "G": 0.21448, "chunks": 209.41, "bytes_per_chunk": 4.89, "boundary_entropy": 0.10785, "router_noise": 0.0, "grad_norm": 0.694, "lr": 6.57e-05, "router_trainable": true, "sec_per_step": 0.073, "alpha": 0.03, "t": 379.9} |
| {"event": "train", "step": 4700, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.43158, "bpb": 2.06533, "chunk_loss": 0.030795, "F": 0.21829, "G": 0.22565, "chunks": 223.53, "bytes_per_chunk": 4.581, "boundary_entropy": 0.10788, "router_noise": 0.0, "grad_norm": 0.7194, "lr": 6.32e-05, "router_trainable": true, "sec_per_step": 0.072, "alpha": 0.03, "t": 383.5} |
| {"event": "train", "step": 4750, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.38036, "bpb": 1.99144, "chunk_loss": 0.030578, "F": 0.20691, "G": 0.21706, "chunks": 211.88, "bytes_per_chunk": 4.833, "boundary_entropy": 0.10951, "router_noise": 0.0, "grad_norm": 0.672, "lr": 6.08e-05, "router_trainable": true, "sec_per_step": 0.073, "alpha": 0.03, "t": 387.2} |
| {"event": "probe", "step": 4800, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3597.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.206689, "cps": 28.101562, "mean_p": 0.21504, "boundary_entropy": 0.103, "bpb": 1.904748}, "tr": {"chunks": 3554.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.196321, "cps": 27.765625, "mean_p": 0.210863, "boundary_entropy": 0.163683, "bpb": 2.050149}, "fi": {"chunks": 4741.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.252033, "cps": 37.039062, "mean_p": 0.24971, "boundary_entropy": 0.121348, "bpb": 2.154495}}, "bpb": {"en": 1.90475, "tr": 2.05015, "fi": 2.15449}, "gini_cpc": 0.056701, "gini_cps": 0.066543, "bpic": 4.770602, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9498402178950337, "fi": 1.2193875435951735}, "t": 391.1} |
| {"event": "eval", "step": 4800, "phase": 3, "bpb": {"en": 1.92009, "tr": 2.05776, "fi": 2.16352}, "bytes_per_chunk": {"en": 4.799, "tr": 5.804, "fi": 4.448}, "bpb_mean": 2.04713, "frozen_inventory": false, "t": 391.7} |
| {"event": "train", "step": 4800, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.37205, "bpb": 1.97945, "chunk_loss": 0.030543, "F": 0.20743, "G": 0.2173, "chunks": 212.41, "bytes_per_chunk": 4.821, "boundary_entropy": 0.11019, "router_noise": 0.0, "grad_norm": 0.6484, "lr": 5.85e-05, "router_trainable": true, "sec_per_step": 0.096, "alpha": 0.03, "t": 391.9} |
| {"event": "train", "step": 4850, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.41929, "bpb": 2.04761, "chunk_loss": 0.030701, "F": 0.2178, "G": 0.22432, "chunks": 223.03, "bytes_per_chunk": 4.591, "boundary_entropy": 0.10628, "router_noise": 0.0, "grad_norm": 0.6952, "lr": 5.62e-05, "router_trainable": true, "sec_per_step": 0.071, "alpha": 0.03, "t": 395.5} |
| {"event": "train", "step": 4900, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.37429, "bpb": 1.98267, "chunk_loss": 0.030508, "F": 0.20462, "G": 0.21594, "chunks": 209.53, "bytes_per_chunk": 4.887, "boundary_entropy": 0.112, "router_noise": 0.0, "grad_norm": 0.6892, "lr": 5.41e-05, "router_trainable": true, "sec_per_step": 0.073, "alpha": 0.03, "t": 399.1} |
| {"event": "train", "step": 4950, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.40739, "bpb": 2.03043, "chunk_loss": 0.030654, "F": 0.21078, "G": 0.21949, "chunks": 215.84, "bytes_per_chunk": 4.744, "boundary_entropy": 0.10664, "router_noise": 0.0, "grad_norm": 0.7216, "lr": 5.2e-05, "router_trainable": true, "sec_per_step": 0.075, "alpha": 0.03, "t": 402.9} |
| {"event": "probe", "step": 5000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3621.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.208068, "cps": 28.289062, "mean_p": 0.215243, "boundary_entropy": 0.098002, "bpb": 1.898043}, "tr": {"chunks": 3522.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.194553, "cps": 27.515625, "mean_p": 0.209984, "boundary_entropy": 0.164149, "bpb": 2.038906}, "fi": {"chunks": 4725.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.251183, "cps": 36.914062, "mean_p": 0.248946, "boundary_entropy": 0.118471, "bpb": 2.141524}}, "bpb": {"en": 1.89804, "tr": 2.03891, "fi": 2.14152}, "gini_cpc": 0.057744, "gini_cps": 0.067577, "bpic": 4.780249, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9350490548857383, "fi": 1.2072175066176312}, "t": 406.8} |
| {"event": "eval", "step": 5000, "phase": 3, "bpb": {"en": 1.91319, "tr": 2.04631, "fi": 2.15077}, "bytes_per_chunk": {"en": 4.779, "tr": 5.825, "fi": 4.446}, "bpb_mean": 2.03675, "frozen_inventory": false, "t": 407.4} |
| {"event": "train", "step": 5000, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.36933, "bpb": 1.97552, "chunk_loss": 0.030692, "F": 0.21207, "G": 0.2223, "chunks": 217.16, "bytes_per_chunk": 4.715, "boundary_entropy": 0.1086, "router_noise": 0.0, "grad_norm": 0.6987, "lr": 5e-05, "router_trainable": true, "sec_per_step": 0.095, "alpha": 0.03, "t": 407.6} |
| {"event": "train", "step": 5050, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.33306, "bpb": 1.92319, "chunk_loss": 0.030565, "F": 0.21219, "G": 0.21836, "chunks": 217.28, "bytes_per_chunk": 4.713, "boundary_entropy": 0.09708, "router_noise": 0.0, "grad_norm": 0.6582, "lr": 4.81e-05, "router_trainable": true, "sec_per_step": 0.075, "alpha": 0.03, "t": 411.4} |
| {"event": "train", "step": 5100, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.40471, "bpb": 2.02656, "chunk_loss": 0.030577, "F": 0.20892, "G": 0.21916, "chunks": 213.94, "bytes_per_chunk": 4.786, "boundary_entropy": 0.10733, "router_noise": 0.0, "grad_norm": 0.6957, "lr": 4.63e-05, "router_trainable": true, "sec_per_step": 0.073, "alpha": 0.03, "t": 415.0} |
| {"event": "train", "step": 5150, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.37257, "bpb": 1.98021, "chunk_loss": 0.030681, "F": 0.20932, "G": 0.21787, "chunks": 214.34, "bytes_per_chunk": 4.777, "boundary_entropy": 0.10589, "router_noise": 0.0, "grad_norm": 0.6917, "lr": 4.45e-05, "router_trainable": true, "sec_per_step": 0.076, "alpha": 0.03, "t": 418.8} |
| {"event": "probe", "step": 5200, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3617.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.207838, "cps": 28.257812, "mean_p": 0.214861, "boundary_entropy": 0.0971, "bpb": 1.891839}, "tr": {"chunks": 3544.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.195769, "cps": 27.6875, "mean_p": 0.210357, "boundary_entropy": 0.161179, "bpb": 2.030702}, "fi": {"chunks": 4732.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.251555, "cps": 36.96875, "mean_p": 0.247827, "boundary_entropy": 0.116918, "bpb": 2.132609}}, "bpb": {"en": 1.89184, "tr": 2.0307, "fi": 2.13261}, "gini_cpc": 0.056766, "gini_cps": 0.066594, "bpic": 4.770201, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9419303124163515, "fi": 1.210343003237228}, "t": 422.9} |
| {"event": "eval", "step": 5200, "phase": 3, "bpb": {"en": 1.9068, "tr": 2.03899, "fi": 2.1411}, "bytes_per_chunk": {"en": 4.785, "tr": 5.758, "fi": 4.453}, "bpb_mean": 2.02896, "frozen_inventory": false, "t": 423.5} |
| {"event": "train", "step": 5200, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.35016, "bpb": 1.94786, "chunk_loss": 0.030526, "F": 0.20639, "G": 0.21361, "chunks": 211.34, "bytes_per_chunk": 4.845, "boundary_entropy": 0.10081, "router_noise": 0.0, "grad_norm": 0.69, "lr": 4.29e-05, "router_trainable": true, "sec_per_step": 0.098, "alpha": 0.03, "t": 423.7} |
| {"event": "train", "step": 5250, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.3675, "bpb": 1.97288, "chunk_loss": 0.030524, "F": 0.20694, "G": 0.21574, "chunks": 211.91, "bytes_per_chunk": 4.832, "boundary_entropy": 0.10677, "router_noise": 0.0, "grad_norm": 0.7232, "lr": 4.14e-05, "router_trainable": true, "sec_per_step": 0.073, "alpha": 0.03, "t": 427.4} |
| {"event": "train", "step": 5300, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.34464, "bpb": 1.93991, "chunk_loss": 0.030548, "F": 0.20853, "G": 0.21719, "chunks": 213.53, "bytes_per_chunk": 4.796, "boundary_entropy": 0.10372, "router_noise": 0.0, "grad_norm": 0.6934, "lr": 3.99e-05, "router_trainable": true, "sec_per_step": 0.07, "alpha": 0.03, "t": 430.9} |
| {"event": "train", "step": 5350, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.34781, "bpb": 1.94448, "chunk_loss": 0.030652, "F": 0.2131, "G": 0.22099, "chunks": 218.22, "bytes_per_chunk": 4.693, "boundary_entropy": 0.0995, "router_noise": 0.0, "grad_norm": 0.6862, "lr": 3.86e-05, "router_trainable": true, "sec_per_step": 0.071, "alpha": 0.03, "t": 434.5} |
| {"event": "probe", "step": 5400, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3594.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.206516, "cps": 28.078125, "mean_p": 0.213995, "boundary_entropy": 0.098037, "bpb": 1.887225}, "tr": {"chunks": 3523.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.194609, "cps": 27.523438, "mean_p": 0.209736, "boundary_entropy": 0.160778, "bpb": 2.024889}, "fi": {"chunks": 4698.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.249747, "cps": 36.703125, "mean_p": 0.247248, "boundary_entropy": 0.116063, "bpb": 2.12402}}, "bpb": {"en": 1.88722, "tr": 2.02489, "fi": 2.12402}, "gini_cpc": 0.056477, "gini_cps": 0.0663, "bpic": 4.801693, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9423411129986388, "fi": 1.2093365433094283}, "t": 438.4} |
| {"event": "eval", "step": 5400, "phase": 3, "bpb": {"en": 1.90271, "tr": 2.03155, "fi": 2.13392}, "bytes_per_chunk": {"en": 4.798, "tr": 5.803, "fi": 4.467}, "bpb_mean": 2.02273, "frozen_inventory": false, "t": 438.9} |
| {"event": "train", "step": 5400, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.34599, "bpb": 1.94186, "chunk_loss": 0.03051, "F": 0.20639, "G": 0.21639, "chunks": 211.34, "bytes_per_chunk": 4.845, "boundary_entropy": 0.10567, "router_noise": 0.0, "grad_norm": 0.6951, "lr": 3.73e-05, "router_trainable": true, "sec_per_step": 0.095, "alpha": 0.03, "t": 439.2} |
| {"event": "train", "step": 5450, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.35926, "bpb": 1.961, "chunk_loss": 0.030524, "F": 0.20566, "G": 0.21459, "chunks": 210.59, "bytes_per_chunk": 4.862, "boundary_entropy": 0.10276, "router_noise": 0.0, "grad_norm": 0.6835, "lr": 3.62e-05, "router_trainable": true, "sec_per_step": 0.075, "alpha": 0.03, "t": 443.0} |
| {"event": "train", "step": 5500, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.36386, "bpb": 1.96764, "chunk_loss": 0.030658, "F": 0.20883, "G": 0.21827, "chunks": 213.84, "bytes_per_chunk": 4.789, "boundary_entropy": 0.10586, "router_noise": 0.0, "grad_norm": 0.7221, "lr": 3.51e-05, "router_trainable": true, "sec_per_step": 0.072, "alpha": 0.03, "t": 446.6} |
| {"event": "train", "step": 5550, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.30409, "bpb": 1.8814, "chunk_loss": 0.030524, "F": 0.20615, "G": 0.21521, "chunks": 211.09, "bytes_per_chunk": 4.851, "boundary_entropy": 0.10478, "router_noise": 0.0, "grad_norm": 0.6866, "lr": 3.41e-05, "router_trainable": true, "sec_per_step": 0.073, "alpha": 0.03, "t": 450.3} |
| {"event": "probe", "step": 5600, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3616.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.20778, "cps": 28.25, "mean_p": 0.214815, "boundary_entropy": 0.095843, "bpb": 1.881978}, "tr": {"chunks": 3633.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.200685, "cps": 28.382812, "mean_p": 0.212339, "boundary_entropy": 0.160412, "bpb": 2.015944}, "fi": {"chunks": 4733.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.251608, "cps": 36.976562, "mean_p": 0.24781, "boundary_entropy": 0.115674, "bpb": 2.116272}}, "bpb": {"en": 1.88198, "tr": 2.01594, "fi": 2.11627}, "gini_cpc": 0.051432, "gini_cps": 0.062149, "bpic": 4.734769, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9658519141207222, "fi": 1.2109335710417977}, "t": 454.1} |
| {"event": "eval", "step": 5600, "phase": 3, "bpb": {"en": 1.8967, "tr": 2.0238, "fi": 2.12571}, "bytes_per_chunk": {"en": 4.787, "tr": 5.674, "fi": 4.47}, "bpb_mean": 2.0154, "frozen_inventory": false, "t": 454.6} |
| {"event": "train", "step": 5600, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.3847, "bpb": 1.99769, "chunk_loss": 0.030806, "F": 0.21692, "G": 0.2245, "chunks": 222.12, "bytes_per_chunk": 4.61, "boundary_entropy": 0.10312, "router_noise": 0.0, "grad_norm": 0.7566, "lr": 3.33e-05, "router_trainable": true, "sec_per_step": 0.094, "alpha": 0.03, "t": 455.0} |
| {"event": "train", "step": 5650, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.35996, "bpb": 1.96201, "chunk_loss": 0.030496, "F": 0.20352, "G": 0.21493, "chunks": 208.41, "bytes_per_chunk": 4.913, "boundary_entropy": 0.10632, "router_noise": 0.0, "grad_norm": 0.7192, "lr": 3.25e-05, "router_trainable": true, "sec_per_step": 0.073, "alpha": 0.03, "t": 458.6} |
| {"event": "train", "step": 5700, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.36697, "bpb": 1.97213, "chunk_loss": 0.030607, "F": 0.21091, "G": 0.21835, "chunks": 215.97, "bytes_per_chunk": 4.741, "boundary_entropy": 0.09933, "router_noise": 0.0, "grad_norm": 0.723, "lr": 3.18e-05, "router_trainable": true, "sec_per_step": 0.072, "alpha": 0.03, "t": 462.2} |
| {"event": "train", "step": 5750, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.37634, "bpb": 1.98564, "chunk_loss": 0.030633, "F": 0.21121, "G": 0.22002, "chunks": 216.28, "bytes_per_chunk": 4.735, "boundary_entropy": 0.10477, "router_noise": 0.0, "grad_norm": 0.7375, "lr": 3.13e-05, "router_trainable": true, "sec_per_step": 0.071, "alpha": 0.03, "t": 465.8} |
| {"event": "probe", "step": 5800, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3627.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.208412, "cps": 28.335938, "mean_p": 0.215014, "boundary_entropy": 0.09346, "bpb": 1.877857}, "tr": {"chunks": 3572.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.197315, "cps": 27.90625, "mean_p": 0.210485, "boundary_entropy": 0.15713, "bpb": 2.010078}, "fi": {"chunks": 4751.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.252565, "cps": 37.117188, "mean_p": 0.249294, "boundary_entropy": 0.115016, "bpb": 2.113489}}, "bpb": {"en": 1.87786, "tr": 2.01008, "fi": 2.11349}, "gini_cpc": 0.055952, "gini_cps": 0.065774, "bpic": 4.747448, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9467546861135163, "fi": 1.211852356242821}, "t": 469.7} |
| {"event": "eval", "step": 5800, "phase": 3, "bpb": {"en": 1.89239, "tr": 2.01828, "fi": 2.12221}, "bytes_per_chunk": {"en": 4.769, "tr": 5.729, "fi": 4.442}, "bpb_mean": 2.01096, "frozen_inventory": false, "t": 470.2} |
| {"event": "train", "step": 5800, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.36594, "bpb": 1.97063, "chunk_loss": 0.030473, "F": 0.20236, "G": 0.2142, "chunks": 207.22, "bytes_per_chunk": 4.942, "boundary_entropy": 0.10853, "router_noise": 0.0, "grad_norm": 0.6786, "lr": 3.08e-05, "router_trainable": true, "sec_per_step": 0.093, "alpha": 0.03, "t": 470.4} |
| {"event": "train", "step": 5850, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.29528, "bpb": 1.86869, "chunk_loss": 0.030429, "F": 0.20264, "G": 0.21236, "chunks": 207.5, "bytes_per_chunk": 4.935, "boundary_entropy": 0.10152, "router_noise": 0.0, "grad_norm": 0.6913, "lr": 3.05e-05, "router_trainable": true, "sec_per_step": 0.074, "alpha": 0.03, "t": 474.2} |
| {"event": "train", "step": 5900, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.35741, "bpb": 1.95833, "chunk_loss": 0.030507, "F": 0.20749, "G": 0.21499, "chunks": 212.47, "bytes_per_chunk": 4.82, "boundary_entropy": 0.09389, "router_noise": 0.0, "grad_norm": 0.6621, "lr": 3.02e-05, "router_trainable": true, "sec_per_step": 0.074, "alpha": 0.03, "t": 477.9} |
| {"event": "train", "step": 5950, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.37024, "bpb": 1.97684, "chunk_loss": 0.030623, "F": 0.21115, "G": 0.2175, "chunks": 216.22, "bytes_per_chunk": 4.736, "boundary_entropy": 0.09801, "router_noise": 0.0, "grad_norm": 0.7252, "lr": 3.01e-05, "router_trainable": true, "sec_per_step": 0.073, "alpha": 0.03, "t": 481.5} |
| {"event": "train", "step": 5999, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.33956, "bpb": 1.93257, "chunk_loss": 0.030403, "F": 0.20093, "G": 0.21078, "chunks": 205.75, "bytes_per_chunk": 4.977, "boundary_entropy": 0.10513, "router_noise": 0.0, "grad_norm": 0.7153, "lr": 3e-05, "router_trainable": true, "sec_per_step": 0.073, "alpha": 0.03, "t": 485.2} |
| {"event": "probe", "step": 6000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3612.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.20755, "cps": 28.21875, "mean_p": 0.214132, "boundary_entropy": 0.095308, "bpb": 1.874829}, "tr": {"chunks": 3565.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.196929, "cps": 27.851562, "mean_p": 0.210442, "boundary_entropy": 0.159459, "bpb": 2.011327}, "fi": {"chunks": 4724.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.25113, "cps": 36.90625, "mean_p": 0.247341, "boundary_entropy": 0.116329, "bpb": 2.109496}}, "bpb": {"en": 1.87483, "tr": 2.01133, "fi": 2.1095}, "gini_cpc": 0.055115, "gini_cps": 0.064925, "bpic": 4.766994, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9488233443806141, "fi": 1.2099693912747385}, "t": 485.5} |
| {"event": "eval", "step": 6000, "phase": 3, "bpb": {"en": 1.88975, "tr": 2.01848, "fi": 2.11923}, "bytes_per_chunk": {"en": 4.787, "tr": 5.757, "fi": 4.458}, "bpb_mean": 2.00915, "frozen_inventory": false, "t": 486.2} |
| {"event": "reps_dumped", "dirs": ["/local/scratch/sas245/beetle-hnet/runs/ali_hnet-ffn_s3/probe/reps/th0.30", "/local/scratch/sas245/beetle-hnet/runs/ali_hnet-ffn_s3/probe/reps/th0.40", "/local/scratch/sas245/beetle-hnet/runs/ali_hnet-ffn_s3/probe/reps/th0.50", "/local/scratch/sas245/beetle-hnet/runs/ali_hnet-ffn_s3/probe/reps/th0.60", "/local/scratch/sas245/beetle-hnet/runs/ali_hnet-ffn_s3/probe/reps/th0.70"], "t": 487.2} |
| {"event": "final", "run": "ali_hnet-ffn_s3", "arch": "hnet_ffn", "objective": "ratio", "condition": "A", "step": 6000, "params": 22785033, "seed": 3, "per_lang": {"en": {"cpc": 0.2075504223409757, "cps": 28.21875, "bpb": 1.8950562349348476, "bytes_per_chunk": 4.724209084601592}, "tr": {"cpc": 0.1969286858531735, "cps": 27.8515625, "bpb": 2.0103088068221227, "bytes_per_chunk": 5.754576985555604}, "fi": {"cpc": 0.25112965817872523, "cps": 36.90625, "bpb": 2.1378381237650945, "bytes_per_chunk": 4.441123901976231}}, "bpb_mean": 2.0144010551740217, "gini_cpc": 0.0551151592286927, "gini_cps": 0.0649245161471026, "bpic": 4.76699437022099, "mean_boundary_entropy": 0.10512667894363403, "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": 488.30208444595337, "ffn_params": 268041, "matched_to_ali_params": 268096, "t": 488.3} |
| {"event": "hub_done", "repo": "AdaptiveChunking/ali_hnet-ffn_s3", "pushed": [0, 1, 2, 4, 6, 8, 16, 32, 64, 128, 256, 512, 1000, 1024, 2048, 3000, 4096, 6000], "failed": [], "ok": true, "t": 503.7} |
|
|
