| {"event": "start", "run": "ali_hnet-ffn_s0", "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": 2.7} |
| {"event": "probe_schedule", "n": 40, "steps": [0, 1, 2, 4, 8, 16, 32, 64, 128, 200, 256, 400, 600, 800, 1000, 1200, 1400, 1600, 1800, 2000, 2200, 2400, 2600, 2800, 3000, 3200, 3400, 3600, 3800, 4000, 4200, 4400, 4600, 4800, 5000, 5200, 5400, 5600, 5800, 6000], "t": 2.7} |
| {"event": "probe", "step": 0, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 6384.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.366833, "cps": 49.875, "mean_p": 0.496725, "boundary_entropy": 0.686995, "bpb": 8.160238}, "tr": {"chunks": 5290.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.292217, "cps": 41.328125, "mean_p": 0.489567, "boundary_entropy": 0.687288, "bpb": 8.171434}, "fi": {"chunks": 8895.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.472862, "cps": 69.492188, "mean_p": 0.500688, "boundary_entropy": 0.687512, "bpb": 8.214758}}, "bpb": {"en": 8.16024, "tr": 8.17143, "fi": 8.21476}, "gini_cpc": 0.106395, "gini_cps": 0.116842, "bpic": 2.758131, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.7965927738475616, "fi": 1.2890367847879758}, "t": 3.8} |
| {"event": "train", "step": 0, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 5.66332, "bpb": 8.17044, "chunk_loss": 0.038537, "F": 0.28732, "G": 0.4892, "chunks": 294.22, "bytes_per_chunk": 3.48, "boundary_entropy": 0.69137, "router_noise": 0.0, "grad_norm": 11.0913, "lr": 1e-06, "router_trainable": true, "sec_per_step": 0.008, "alpha": 0.03, "t": 4.3} |
| {"event": "probe", "step": 1, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 6303.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.362179, "cps": 49.242188, "mean_p": 0.496453, "boundary_entropy": 0.686986, "bpb": 8.127897}, "tr": {"chunks": 5227.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.288737, "cps": 40.835938, "mean_p": 0.489372, "boundary_entropy": 0.687277, "bpb": 8.149125}, "fi": {"chunks": 8820.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.468875, "cps": 68.90625, "mean_p": 0.500418, "boundary_entropy": 0.687508, "bpb": 8.186896}}, "bpb": {"en": 8.1279, "tr": 8.14913, "fi": 8.1869}, "gini_cpc": 0.107245, "gini_cps": 0.117707, "bpic": 2.787813, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.7972210579984653, "fi": 1.294593776095493}, "t": 4.9} |
| {"event": "probe", "step": 2, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 6173.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.354709, "cps": 48.226562, "mean_p": 0.495924, "boundary_entropy": 0.686967, "bpb": 8.063107}, "tr": {"chunks": 5127.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.283213, "cps": 40.054688, "mean_p": 0.488972, "boundary_entropy": 0.687254, "bpb": 8.10444}, "fi": {"chunks": 8676.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.461219, "cps": 67.78125, "mean_p": 0.499888, "boundary_entropy": 0.687499, "bpb": 8.130061}}, "bpb": {"en": 8.06311, "tr": 8.10444, "fi": 8.13006}, "gini_cpc": 0.107967, "gini_cps": 0.118442, "bpic": 2.840008, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.7984369173788103, "fi": 1.300275869931204}, "t": 5.7} |
| {"event": "probe", "step": 4, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 5668.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.325691, "cps": 44.28125, "mean_p": 0.494023, "boundary_entropy": 0.686889, "bpb": 7.845706}, "tr": {"chunks": 4749.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.262332, "cps": 37.101562, "mean_p": 0.487542, "boundary_entropy": 0.687166, "bpb": 7.949275}, "fi": {"chunks": 8074.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.429217, "cps": 63.078125, "mean_p": 0.49794, "boundary_entropy": 0.687458, "bpb": 7.938885}}, "bpb": {"en": 7.84571, "tr": 7.94928, "fi": 7.93889}, "gini_cpc": 0.109371, "gini_cps": 0.119878, "bpic": 3.068087, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.8054635590875718, "fi": 1.317865655941562}, "t": 6.7} |
| {"event": "probe", "step": 8, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 4374.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.251336, "cps": 34.171875, "mean_p": 0.488791, "boundary_entropy": 0.6866, "bpb": 7.284042}, "tr": {"chunks": 3770.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.208253, "cps": 29.453125, "mean_p": 0.483682, "boundary_entropy": 0.686886, "bpb": 7.521462}, "fi": {"chunks": 6368.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.338525, "cps": 49.75, "mean_p": 0.492209, "boundary_entropy": 0.687253, "bpb": 7.431297}}, "bpb": {"en": 7.28404, "tr": 7.52146, "fi": 7.4313}, "gini_cpc": 0.108817, "gini_cps": 0.11935, "bpic": 3.909316, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.8285832320420409, "fi": 1.3469035971787515}, "t": 8.0} |
| {"event": "probe", "step": 16, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 4430.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.254554, "cps": 34.609375, "mean_p": 0.490968, "boundary_entropy": 0.68693, "bpb": 6.658242}, "tr": {"chunks": 4573.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.25261, "cps": 35.726562, "mean_p": 0.488506, "boundary_entropy": 0.687359, "bpb": 6.975098}, "fi": {"chunks": 5628.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.299187, "cps": 43.96875, "mean_p": 0.49126, "boundary_entropy": 0.687411, "bpb": 6.835761}}, "bpb": {"en": 6.65824, "tr": 6.9751, "fi": 6.83576}, "gini_cpc": 0.038508, "gini_cps": 0.054587, "bpic": 3.87752, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9923640981397018, "fi": 1.1753375174436262}, "t": 9.8} |
| {"event": "probe", "step": 32, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 4016.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.230765, "cps": 31.375, "mean_p": 0.49455, "boundary_entropy": 0.687547, "bpb": 6.306568}, "tr": {"chunks": 4104.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.226703, "cps": 32.0625, "mean_p": 0.491943, "boundary_entropy": 0.687956, "bpb": 6.636564}, "fi": {"chunks": 4304.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.228802, "cps": 33.625, "mean_p": 0.492852, "boundary_entropy": 0.687993, "bpb": 6.431696}}, "bpb": {"en": 6.30657, "tr": 6.63656, "fi": 6.4317}, "gini_cpc": 0.003946, "gini_cps": 0.015454, "bpic": 4.566323, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9823974279097497, "fi": 0.9914956091851826}, "t": 12.8} |
| {"event": "train", "step": 50, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 4.22511, "bpb": 6.09555, "chunk_loss": 0.030986, "F": 0.18033, "G": 0.4914, "chunks": 184.66, "bytes_per_chunk": 5.545, "boundary_entropy": 0.69209, "router_noise": 0.0, "grad_norm": 2.4386, "lr": 5.1e-05, "router_trainable": true, "sec_per_step": 0.222, "alpha": 0.03, "t": 15.4} |
| {"event": "probe", "step": 64, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 4035.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.231857, "cps": 31.523438, "mean_p": 0.487823, "boundary_entropy": 0.686547, "bpb": 5.620056}, "tr": {"chunks": 3831.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.211622, "cps": 29.929688, "mean_p": 0.487762, "boundary_entropy": 0.687473, "bpb": 6.008162}, "fi": {"chunks": 5208.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.276859, "cps": 40.6875, "mean_p": 0.492454, "boundary_entropy": 0.687657, "bpb": 5.815904}}, "bpb": {"en": 5.62006, "tr": 6.00816, "fi": 5.8159}, "gini_cpc": 0.060376, "gini_cps": 0.070216, "bpic": 4.339299, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9127296981112005, "fi": 1.1940971815312522}, "t": 17.7} |
| {"event": "train", "step": 100, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 3.20126, "bpb": 4.61844, "chunk_loss": 0.031632, "F": 0.19626, "G": 0.42238, "chunks": 200.97, "bytes_per_chunk": 5.095, "boundary_entropy": 0.66652, "router_noise": 0.0, "grad_norm": 1.5844, "lr": 0.000101, "router_trainable": true, "sec_per_step": 0.144, "alpha": 0.03, "t": 22.5} |
| {"event": "probe", "step": 128, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3536.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.203183, "cps": 27.625, "mean_p": 0.374138, "boundary_entropy": 0.607764, "bpb": 4.166365}, "tr": {"chunks": 2699.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.149091, "cps": 21.085938, "mean_p": 0.40932, "boundary_entropy": 0.650839, "bpb": 4.504017}, "fi": {"chunks": 2997.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.159322, "cps": 23.414062, "mean_p": 0.388615, "boundary_entropy": 0.634828, "bpb": 4.409592}}, "bpb": {"en": 4.16637, "tr": 4.50402, "fi": 4.40959}, "gini_cpc": 0.070488, "gini_cps": 0.060442, "bpic": 6.145147, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.733777172629321, "fi": 0.7841275689274905}, "t": 26.3} |
| {"event": "train", "step": 150, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 2.80655, "bpb": 4.04899, "chunk_loss": 0.031372, "F": 0.19989, "G": 0.35717, "chunks": 204.69, "bytes_per_chunk": 5.003, "boundary_entropy": 0.58922, "router_noise": 0.0, "grad_norm": 2.3495, "lr": 0.000151, "router_trainable": true, "sec_per_step": 0.135, "alpha": 0.03, "t": 29.3} |
| {"event": "probe", "step": 200, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2886.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.165833, "cps": 22.546875, "mean_p": 0.365831, "boundary_entropy": 0.57184, "bpb": 3.627684}, "tr": {"chunks": 3501.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.193393, "cps": 27.351562, "mean_p": 0.378479, "boundary_entropy": 0.596582, "bpb": 3.773286}, "fi": {"chunks": 3451.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.183456, "cps": 26.960938, "mean_p": 0.384114, "boundary_entropy": 0.612178, "bpb": 3.88068}}, "bpb": {"en": 3.62768, "tr": 3.77329, "fi": 3.88068}, "gini_cpc": 0.033856, "gini_cps": 0.041675, "bpic": 5.766619, "worst_lang": "tr", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.1661901066695852, "fi": 1.1062693224460276}, "t": 35.5} |
| {"event": "eval", "step": 200, "phase": 1, "bpb": {"en": 3.63418, "tr": 3.78189, "fi": 3.8804}, "bytes_per_chunk": {"en": 6.078, "tr": 5.088, "fi": 5.683}, "bpb_mean": 3.76549, "frozen_inventory": false, "t": 36.6} |
| {"event": "train", "step": 200, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 2.54592, "bpb": 3.67299, "chunk_loss": 0.030196, "F": 0.17004, "G": 0.37371, "chunks": 174.12, "bytes_per_chunk": 5.881, "boundary_entropy": 0.58203, "router_noise": 0.0, "grad_norm": 1.5365, "lr": 0.000201, "router_trainable": true, "sec_per_step": 0.154, "alpha": 0.03, "t": 37.0} |
| {"event": "train", "step": 250, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 2.39609, "bpb": 3.45682, "chunk_loss": 0.031226, "F": 0.19714, "G": 0.35648, "chunks": 201.88, "bytes_per_chunk": 5.072, "boundary_entropy": 0.5446, "router_noise": 0.0, "grad_norm": 1.5276, "lr": 0.000251, "router_trainable": true, "sec_per_step": 0.124, "alpha": 0.03, "t": 43.2} |
| {"event": "probe", "step": 256, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3348.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.192381, "cps": 26.15625, "mean_p": 0.329814, "boundary_entropy": 0.515314, "bpb": 3.354014}, "tr": {"chunks": 3596.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.198641, "cps": 28.09375, "mean_p": 0.352117, "boundary_entropy": 0.537202, "bpb": 3.506294}, "fi": {"chunks": 4159.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.221094, "cps": 32.492188, "mean_p": 0.371241, "boundary_entropy": 0.571346, "bpb": 3.658994}}, "bpb": {"en": 3.35401, "tr": 3.50629, "fi": 3.65899}, "gini_cpc": 0.031272, "gini_cps": 0.048696, "bpic": 5.10961, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.032542181467774, "fi": 1.1492531632810743}, "t": 44.3} |
| {"event": "train", "step": 300, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 2.27888, "bpb": 3.28773, "chunk_loss": 0.031219, "F": 0.20837, "G": 0.29637, "chunks": 213.38, "bytes_per_chunk": 4.799, "boundary_entropy": 0.42446, "router_noise": 0.0, "grad_norm": 1.2738, "lr": 0.0003, "router_trainable": true, "sec_per_step": 0.133, "alpha": 0.03, "t": 49.8} |
| {"event": "train", "step": 350, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 2.19608, "bpb": 3.16827, "chunk_loss": 0.03074, "F": 0.19327, "G": 0.28813, "chunks": 197.91, "bytes_per_chunk": 5.174, "boundary_entropy": 0.40789, "router_noise": 0.0, "grad_norm": 1.2368, "lr": 0.0002999, "router_trainable": true, "sec_per_step": 0.117, "alpha": 0.03, "t": 55.7} |
| {"event": "probe", "step": 400, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3572.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.205252, "cps": 27.90625, "mean_p": 0.263786, "boundary_entropy": 0.348662, "bpb": 2.905431}, "tr": {"chunks": 3730.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.206043, "cps": 29.140625, "mean_p": 0.301767, "boundary_entropy": 0.443248, "bpb": 3.1865}, "fi": {"chunks": 4163.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.221307, "cps": 32.523438, "mean_p": 0.296538, "boundary_entropy": 0.427685, "bpb": 3.377872}}, "bpb": {"en": 2.90543, "tr": 3.1865, "fi": 3.37787}, "gini_cpc": 0.016919, "gini_cps": 0.034365, "bpic": 4.948277, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0038549165503678, "fi": 1.0782195384595143}, "t": 61.8} |
| {"event": "eval", "step": 400, "phase": 1, "bpb": {"en": 2.92, "tr": 3.19387, "fi": 3.3696}, "bytes_per_chunk": {"en": 4.623, "tr": 5.493, "fi": 4.719}, "bpb_mean": 3.16116, "frozen_inventory": false, "t": 62.5} |
| {"event": "train", "step": 400, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 2.04715, "bpb": 2.95342, "chunk_loss": 0.031156, "F": 0.21265, "G": 0.27844, "chunks": 217.75, "bytes_per_chunk": 4.703, "boundary_entropy": 0.35874, "router_noise": 0.0, "grad_norm": 1.1996, "lr": 0.0002998, "router_trainable": true, "sec_per_step": 0.145, "alpha": 0.03, "t": 62.9} |
| {"event": "train", "step": 450, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 2.01624, "bpb": 2.90882, "chunk_loss": 0.030869, "F": 0.20294, "G": 0.27049, "chunks": 207.81, "bytes_per_chunk": 4.928, "boundary_entropy": 0.34967, "router_noise": 0.0, "grad_norm": 1.2741, "lr": 0.0002995, "router_trainable": true, "sec_per_step": 0.115, "alpha": 0.03, "t": 68.7} |
| {"event": "train", "step": 500, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 2.00303, "bpb": 2.88976, "chunk_loss": 0.03117, "F": 0.21484, "G": 0.27168, "chunks": 220.0, "bytes_per_chunk": 4.655, "boundary_entropy": 0.33023, "router_noise": 0.0, "grad_norm": 1.3331, "lr": 0.0002992, "router_trainable": true, "sec_per_step": 0.115, "alpha": 0.03, "t": 74.5} |
| {"event": "train", "step": 550, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.98222, "bpb": 2.85974, "chunk_loss": 0.031066, "F": 0.2117, "G": 0.269, "chunks": 216.78, "bytes_per_chunk": 4.724, "boundary_entropy": 0.326, "router_noise": 0.0, "grad_norm": 1.5665, "lr": 0.0002987, "router_trainable": true, "sec_per_step": 0.109, "alpha": 0.03, "t": 79.9} |
| {"event": "probe", "step": 600, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3570.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.205137, "cps": 27.890625, "mean_p": 0.256138, "boundary_entropy": 0.298342, "bpb": 2.596343}, "tr": {"chunks": 3545.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.195824, "cps": 27.695312, "mean_p": 0.269185, "boundary_entropy": 0.363529, "bpb": 2.933894}, "fi": {"chunks": 4457.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.236936, "cps": 34.820312, "mean_p": 0.280365, "boundary_entropy": 0.337555, "bpb": 3.148505}}, "bpb": {"en": 2.59634, "tr": 2.93389, "fi": 3.14851}, "gini_cpc": 0.042966, "gini_cps": 0.052541, "bpic": 4.902523, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 0.9546003564105862, "fi": 1.1550124212142994}, "t": 86.0} |
| {"event": "eval", "step": 600, "phase": 1, "bpb": {"en": 2.60923, "tr": 2.94227, "fi": 3.14892}, "bytes_per_chunk": {"en": 4.677, "tr": 5.764, "fi": 4.478}, "bpb_mean": 2.90014, "frozen_inventory": false, "t": 86.7} |
| {"event": "train", "step": 600, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.8744, "bpb": 2.70418, "chunk_loss": 0.030931, "F": 0.2088, "G": 0.26569, "chunks": 213.81, "bytes_per_chunk": 4.789, "boundary_entropy": 0.307, "router_noise": 0.0, "grad_norm": 1.0894, "lr": 0.0002982, "router_trainable": true, "sec_per_step": 0.144, "alpha": 0.03, "t": 87.1} |
| {"event": "train", "step": 650, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.90658, "bpb": 2.75061, "chunk_loss": 0.031085, "F": 0.21283, "G": 0.26764, "chunks": 217.94, "bytes_per_chunk": 4.699, "boundary_entropy": 0.30306, "router_noise": 0.0, "grad_norm": 1.1002, "lr": 0.0002975, "router_trainable": true, "sec_per_step": 0.113, "alpha": 0.03, "t": 92.8} |
| {"event": "train", "step": 700, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.85816, "bpb": 2.68076, "chunk_loss": 0.030899, "F": 0.20865, "G": 0.26167, "chunks": 213.66, "bytes_per_chunk": 4.793, "boundary_entropy": 0.29563, "router_noise": 0.0, "grad_norm": 1.1251, "lr": 0.0002967, "router_trainable": true, "sec_per_step": 0.111, "alpha": 0.03, "t": 98.3} |
| {"event": "train", "step": 750, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.8182, "bpb": 2.62311, "chunk_loss": 0.030901, "F": 0.20972, "G": 0.25979, "chunks": 214.75, "bytes_per_chunk": 4.768, "boundary_entropy": 0.29, "router_noise": 0.0, "grad_norm": 1.1728, "lr": 0.0002959, "router_trainable": true, "sec_per_step": 0.11, "alpha": 0.03, "t": 103.8} |
| {"event": "probe", "step": 800, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3327.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.191174, "cps": 25.992188, "mean_p": 0.236619, "boundary_entropy": 0.266337, "bpb": 2.442114}, "tr": {"chunks": 3507.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.193725, "cps": 27.398438, "mean_p": 0.261967, "boundary_entropy": 0.333889, "bpb": 2.78724}, "fi": {"chunks": 4511.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.239806, "cps": 35.242188, "mean_p": 0.275495, "boundary_entropy": 0.294374, "bpb": 2.96548}}, "bpb": {"en": 2.44211, "tr": 2.78724, "fi": 2.96548}, "gini_cpc": 0.051899, "gini_cps": 0.069575, "bpic": 5.000617, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0133431446058068, "fi": 1.254389075248501}, "t": 109.6} |
| {"event": "eval", "step": 800, "phase": 1, "bpb": {"en": 2.45564, "tr": 2.7985, "fi": 2.96982}, "bytes_per_chunk": {"en": 4.909, "tr": 5.719, "fi": 4.466}, "bpb_mean": 2.74132, "frozen_inventory": false, "t": 110.4} |
| {"event": "train", "step": 800, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.80911, "bpb": 2.60999, "chunk_loss": 0.030799, "F": 0.20505, "G": 0.25458, "chunks": 209.97, "bytes_per_chunk": 4.877, "boundary_entropy": 0.27849, "router_noise": 0.0, "grad_norm": 1.037, "lr": 0.0002949, "router_trainable": true, "sec_per_step": 0.142, "alpha": 0.03, "t": 110.9} |
| {"event": "train", "step": 850, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.78109, "bpb": 2.56957, "chunk_loss": 0.030703, "F": 0.20267, "G": 0.25014, "chunks": 207.53, "bytes_per_chunk": 4.934, "boundary_entropy": 0.27225, "router_noise": 0.0, "grad_norm": 1.2548, "lr": 0.0002938, "router_trainable": true, "sec_per_step": 0.11, "alpha": 0.03, "t": 116.4} |
| {"event": "train", "step": 900, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.62349, "bpb": 2.34219, "chunk_loss": 0.030706, "F": 0.20438, "G": 0.24599, "chunks": 209.28, "bytes_per_chunk": 4.893, "boundary_entropy": 0.25597, "router_noise": 0.0, "grad_norm": 0.9606, "lr": 0.0002927, "router_trainable": true, "sec_per_step": 0.107, "alpha": 0.03, "t": 121.8} |
| {"event": "train", "step": 950, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.77943, "bpb": 2.56717, "chunk_loss": 0.030775, "F": 0.2084, "G": 0.24562, "chunks": 213.41, "bytes_per_chunk": 4.798, "boundary_entropy": 0.25313, "router_noise": 0.0, "grad_norm": 0.9879, "lr": 0.0002914, "router_trainable": true, "sec_per_step": 0.108, "alpha": 0.03, "t": 127.2} |
| {"event": "probe", "step": 1000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3425.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.196805, "cps": 26.757812, "mean_p": 0.229477, "boundary_entropy": 0.23552, "bpb": 2.357757}, "tr": {"chunks": 3588.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.198199, "cps": 28.03125, "mean_p": 0.248883, "boundary_entropy": 0.302557, "bpb": 2.668232}, "fi": {"chunks": 4509.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.2397, "cps": 35.226562, "mean_p": 0.274229, "boundary_entropy": 0.265416, "bpb": 2.841914}}, "bpb": {"en": 2.35776, "tr": 2.66823, "fi": 2.84191}, "gini_cpc": 0.045055, "gini_cps": 0.062721, "bpic": 4.923798, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.007083376510164, "fi": 1.2179568329914123}, "t": 132.8} |
| {"event": "eval", "step": 1000, "phase": 1, "bpb": {"en": 2.36851, "tr": 2.6773, "fi": 2.84794}, "bytes_per_chunk": {"en": 4.796, "tr": 5.642, "fi": 4.436}, "bpb_mean": 2.63125, "frozen_inventory": false, "t": 133.6} |
| {"event": "train", "step": 1000, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.72737, "bpb": 2.49207, "chunk_loss": 0.031126, "F": 0.21918, "G": 0.25159, "chunks": 224.44, "bytes_per_chunk": 4.563, "boundary_entropy": 0.24482, "router_noise": 0.0, "grad_norm": 1.108, "lr": 0.0002901, "router_trainable": true, "sec_per_step": 0.139, "alpha": 0.03, "t": 134.1} |
| {"event": "train", "step": 1050, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.74787, "bpb": 2.52165, "chunk_loss": 0.030772, "F": 0.20932, "G": 0.2459, "chunks": 214.34, "bytes_per_chunk": 4.777, "boundary_entropy": 0.24655, "router_noise": 0.0, "grad_norm": 0.9367, "lr": 0.0002886, "router_trainable": true, "sec_per_step": 0.112, "alpha": 0.03, "t": 139.7} |
| {"event": "train", "step": 1100, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.7369, "bpb": 2.50581, "chunk_loss": 0.030837, "F": 0.20984, "G": 0.24683, "chunks": 214.88, "bytes_per_chunk": 4.766, "boundary_entropy": 0.23817, "router_noise": 0.0, "grad_norm": 1.0053, "lr": 0.0002871, "router_trainable": true, "sec_per_step": 0.109, "alpha": 0.03, "t": 145.2} |
| {"event": "train", "step": 1150, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.59601, "bpb": 2.30255, "chunk_loss": 0.030705, "F": 0.20682, "G": 0.24032, "chunks": 211.78, "bytes_per_chunk": 4.835, "boundary_entropy": 0.22967, "router_noise": 0.0, "grad_norm": 0.8459, "lr": 0.0002855, "router_trainable": true, "sec_per_step": 0.108, "alpha": 0.03, "t": 150.5} |
| {"event": "probe", "step": 1200, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3351.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.192553, "cps": 26.179688, "mean_p": 0.221656, "boundary_entropy": 0.218724, "bpb": 2.287685}, "tr": {"chunks": 3582.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.197868, "cps": 27.984375, "mean_p": 0.234923, "boundary_entropy": 0.271027, "bpb": 2.568581}, "fi": {"chunks": 4620.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.245601, "cps": 36.09375, "mean_p": 0.270688, "boundary_entropy": 0.234264, "bpb": 2.745341}}, "bpb": {"en": 2.28769, "tr": 2.56858, "fi": 2.74534}, "gini_cpc": 0.055604, "gini_cps": 0.073228, "bpic": 4.910586, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0276014832265643, "fi": 1.275498007389817}, "t": 156.2} |
| {"event": "eval", "step": 1200, "phase": 1, "bpb": {"en": 2.29766, "tr": 2.57688, "fi": 2.75458}, "bytes_per_chunk": {"en": 4.981, "tr": 5.6, "fi": 4.34}, "bpb_mean": 2.54304, "frozen_inventory": false, "t": 157.1} |
| {"event": "train", "step": 1200, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.6384, "bpb": 2.36371, "chunk_loss": 0.03061, "F": 0.20358, "G": 0.23537, "chunks": 208.47, "bytes_per_chunk": 4.912, "boundary_entropy": 0.22617, "router_noise": 0.0, "grad_norm": 0.8641, "lr": 0.0002837, "router_trainable": true, "sec_per_step": 0.137, "alpha": 0.03, "t": 157.4} |
| {"event": "train", "step": 1250, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.69081, "bpb": 2.43933, "chunk_loss": 0.030684, "F": 0.20645, "G": 0.23675, "chunks": 211.41, "bytes_per_chunk": 4.844, "boundary_entropy": 0.22007, "router_noise": 0.0, "grad_norm": 0.8709, "lr": 0.0002819, "router_trainable": true, "sec_per_step": 0.109, "alpha": 0.03, "t": 162.9} |
| {"event": "train", "step": 1300, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.63341, "bpb": 2.35652, "chunk_loss": 0.030591, "F": 0.20221, "G": 0.23097, "chunks": 207.06, "bytes_per_chunk": 4.945, "boundary_entropy": 0.21988, "router_noise": 0.0, "grad_norm": 0.9574, "lr": 0.00028, "router_trainable": true, "sec_per_step": 0.122, "alpha": 0.03, "t": 168.9} |
| {"event": "train", "step": 1350, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.61567, "bpb": 2.33092, "chunk_loss": 0.03085, "F": 0.21344, "G": 0.23826, "chunks": 218.56, "bytes_per_chunk": 4.685, "boundary_entropy": 0.209, "router_noise": 0.0, "grad_norm": 0.9688, "lr": 0.000278, "router_trainable": true, "sec_per_step": 0.111, "alpha": 0.03, "t": 174.5} |
| {"event": "probe", "step": 1400, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3344.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.192151, "cps": 26.125, "mean_p": 0.2171, "boundary_entropy": 0.193909, "bpb": 2.231929}, "tr": {"chunks": 3609.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.199359, "cps": 28.195312, "mean_p": 0.225035, "boundary_entropy": 0.23838, "bpb": 2.505235}, "fi": {"chunks": 4672.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.248365, "cps": 36.5, "mean_p": 0.264277, "boundary_entropy": 0.207454, "bpb": 2.664367}}, "bpb": {"en": 2.23193, "tr": 2.50523, "fi": 2.66437}, "gini_cpc": 0.058568, "gini_cps": 0.076158, "bpic": 4.880172, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0375145168780346, "fi": 1.2925543157965955}, "t": 182.8} |
| {"event": "eval", "step": 1400, "phase": 1, "bpb": {"en": 2.24308, "tr": 2.5156, "fi": 2.67276}, "bytes_per_chunk": {"en": 4.926, "tr": 5.508, "fi": 4.289}, "bpb_mean": 2.47715, "frozen_inventory": false, "t": 183.4} |
| {"event": "train", "step": 1400, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.69914, "bpb": 2.45134, "chunk_loss": 0.03066, "F": 0.20648, "G": 0.2324, "chunks": 211.44, "bytes_per_chunk": 4.843, "boundary_entropy": 0.20425, "router_noise": 0.0, "grad_norm": 0.9255, "lr": 0.0002759, "router_trainable": true, "sec_per_step": 0.186, "alpha": 0.03, "t": 183.8} |
| {"event": "train", "step": 1450, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.65454, "bpb": 2.38699, "chunk_loss": 0.030724, "F": 0.21008, "G": 0.23606, "chunks": 215.12, "bytes_per_chunk": 4.76, "boundary_entropy": 0.2118, "router_noise": 0.0, "grad_norm": 0.8037, "lr": 0.0002738, "router_trainable": true, "sec_per_step": 0.108, "alpha": 0.03, "t": 189.2} |
| {"event": "train", "step": 1500, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.61167, "bpb": 2.32514, "chunk_loss": 0.030534, "F": 0.2023, "G": 0.22591, "chunks": 207.16, "bytes_per_chunk": 4.943, "boundary_entropy": 0.19647, "router_noise": 0.0, "grad_norm": 0.8753, "lr": 0.0002715, "router_trainable": true, "sec_per_step": 0.108, "alpha": 0.03, "t": 194.6} |
| {"event": "train", "step": 1550, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.57467, "bpb": 2.27177, "chunk_loss": 0.0306, "F": 0.20554, "G": 0.22469, "chunks": 210.47, "bytes_per_chunk": 4.865, "boundary_entropy": 0.18378, "router_noise": 0.0, "grad_norm": 0.7452, "lr": 0.0002692, "router_trainable": true, "sec_per_step": 0.107, "alpha": 0.03, "t": 199.9} |
| {"event": "probe", "step": 1600, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3413.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.196116, "cps": 26.664062, "mean_p": 0.216428, "boundary_entropy": 0.186149, "bpb": 2.196954}, "tr": {"chunks": 3771.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.208308, "cps": 29.460938, "mean_p": 0.220327, "boundary_entropy": 0.214076, "bpb": 2.440104}, "fi": {"chunks": 4736.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.251768, "cps": 37.0, "mean_p": 0.26561, "boundary_entropy": 0.19479, "bpb": 2.594477}}, "bpb": {"en": 2.19695, "tr": 2.4401, "fi": 2.59448}, "gini_cpc": 0.05654, "gini_cps": 0.073993, "bpic": 4.759396, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0621694665478276, "fi": 1.2837712403813075}, "t": 205.6} |
| {"event": "eval", "step": 1600, "phase": 1, "bpb": {"en": 2.20803, "tr": 2.45068, "fi": 2.60275}, "bytes_per_chunk": {"en": 4.911, "tr": 5.255, "fi": 4.245}, "bpb_mean": 2.42049, "frozen_inventory": false, "t": 206.4} |
| {"event": "train", "step": 1600, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.58149, "bpb": 2.2816, "chunk_loss": 0.030696, "F": 0.20929, "G": 0.22957, "chunks": 214.31, "bytes_per_chunk": 4.778, "boundary_entropy": 0.18799, "router_noise": 0.0, "grad_norm": 0.8917, "lr": 0.0002668, "router_trainable": true, "sec_per_step": 0.135, "alpha": 0.03, "t": 206.7} |
| {"event": "train", "step": 1650, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.60395, "bpb": 2.31401, "chunk_loss": 0.030542, "F": 0.20303, "G": 0.22368, "chunks": 207.91, "bytes_per_chunk": 4.925, "boundary_entropy": 0.18359, "router_noise": 0.0, "grad_norm": 0.8181, "lr": 0.0002643, "router_trainable": true, "sec_per_step": 0.111, "alpha": 0.03, "t": 212.3} |
| {"event": "train", "step": 1700, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.61712, "bpb": 2.33301, "chunk_loss": 0.030616, "F": 0.20724, "G": 0.22633, "chunks": 212.22, "bytes_per_chunk": 4.825, "boundary_entropy": 0.18345, "router_noise": 0.0, "grad_norm": 0.8092, "lr": 0.0002618, "router_trainable": true, "sec_per_step": 0.117, "alpha": 0.03, "t": 218.1} |
| {"event": "train", "step": 1750, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.56703, "bpb": 2.26074, "chunk_loss": 0.030472, "F": 0.20175, "G": 0.22002, "chunks": 206.59, "bytes_per_chunk": 4.957, "boundary_entropy": 0.17797, "router_noise": 0.0, "grad_norm": 0.7692, "lr": 0.0002591, "router_trainable": true, "sec_per_step": 0.113, "alpha": 0.03, "t": 223.8} |
| {"event": "probe", "step": 1800, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3355.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.192783, "cps": 26.210938, "mean_p": 0.209903, "boundary_entropy": 0.171875, "bpb": 2.158865}, "tr": {"chunks": 3801.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.209965, "cps": 29.695312, "mean_p": 0.221566, "boundary_entropy": 0.204633, "bpb": 2.396361}, "fi": {"chunks": 4733.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.251608, "cps": 36.976562, "mean_p": 0.263735, "boundary_entropy": 0.191166, "bpb": 2.523225}}, "bpb": {"en": 2.15887, "tr": 2.39636, "fi": 2.52322}, "gini_cpc": 0.059932, "gini_cps": 0.07727, "bpic": 4.771806, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0891279763347883, "fi": 1.3051373451228436}, "t": 229.3} |
| {"event": "eval", "step": 1800, "phase": 1, "bpb": {"en": 2.1704, "tr": 2.40537, "fi": 2.53367}, "bytes_per_chunk": {"en": 4.975, "tr": 5.254, "fi": 4.228}, "bpb_mean": 2.36982, "frozen_inventory": false, "t": 230.2} |
| {"event": "train", "step": 1800, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.53948, "bpb": 2.221, "chunk_loss": 0.030445, "F": 0.19971, "G": 0.21829, "chunks": 204.5, "bytes_per_chunk": 5.007, "boundary_entropy": 0.17633, "router_noise": 0.0, "grad_norm": 0.8136, "lr": 0.0002564, "router_trainable": true, "sec_per_step": 0.136, "alpha": 0.03, "t": 230.6} |
| {"event": "train", "step": 1850, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.56587, "bpb": 2.25907, "chunk_loss": 0.030581, "F": 0.20831, "G": 0.2225, "chunks": 213.31, "bytes_per_chunk": 4.8, "boundary_entropy": 0.16675, "router_noise": 0.0, "grad_norm": 0.8091, "lr": 0.0002537, "router_trainable": true, "sec_per_step": 0.112, "alpha": 0.03, "t": 236.2} |
| {"event": "train", "step": 1900, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.50253, "bpb": 2.16769, "chunk_loss": 0.030564, "F": 0.20691, "G": 0.21978, "chunks": 211.88, "bytes_per_chunk": 4.833, "boundary_entropy": 0.16113, "router_noise": 0.0, "grad_norm": 0.7028, "lr": 0.0002508, "router_trainable": true, "sec_per_step": 0.108, "alpha": 0.03, "t": 241.6} |
| {"event": "train", "step": 1950, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.58248, "bpb": 2.28304, "chunk_loss": 0.030706, "F": 0.21097, "G": 0.2276, "chunks": 216.03, "bytes_per_chunk": 4.74, "boundary_entropy": 0.16588, "router_noise": 0.0, "grad_norm": 0.7939, "lr": 0.0002479, "router_trainable": true, "sec_per_step": 0.11, "alpha": 0.03, "t": 247.1} |
| {"event": "probe", "step": 2000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3396.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.195139, "cps": 26.53125, "mean_p": 0.210377, "boundary_entropy": 0.15684, "bpb": 2.125556}, "tr": {"chunks": 3771.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.208308, "cps": 29.460938, "mean_p": 0.215338, "boundary_entropy": 0.187209, "bpb": 2.345732}, "fi": {"chunks": 4807.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.255542, "cps": 37.554688, "mean_p": 0.26273, "boundary_entropy": 0.170295, "bpb": 2.483639}}, "bpb": {"en": 2.12556, "tr": 2.34573, "fi": 2.48364}, "gini_cpc": 0.061107, "gini_cps": 0.078559, "bpic": 4.737932, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.067486569295564, "fi": 1.3095397249898704}, "t": 252.7} |
| {"event": "eval", "step": 2000, "phase": 1, "bpb": {"en": 2.13759, "tr": 2.35697, "fi": 2.49585}, "bytes_per_chunk": {"en": 4.91, "tr": 5.332, "fi": 4.18}, "bpb_mean": 2.33014, "frozen_inventory": false, "t": 253.6} |
| {"event": "train", "step": 2000, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.54123, "bpb": 2.22353, "chunk_loss": 0.030656, "F": 0.21262, "G": 0.22613, "chunks": 217.72, "bytes_per_chunk": 4.703, "boundary_entropy": 0.1577, "router_noise": 0.0, "grad_norm": 0.7886, "lr": 0.000245, "router_trainable": true, "sec_per_step": 0.138, "alpha": 0.03, "t": 254.0} |
| {"event": "train", "step": 2050, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.55383, "bpb": 2.2417, "chunk_loss": 0.030416, "F": 0.1994, "G": 0.21495, "chunks": 204.19, "bytes_per_chunk": 5.015, "boundary_entropy": 0.16398, "router_noise": 0.0, "grad_norm": 0.7841, "lr": 0.0002419, "router_trainable": true, "sec_per_step": 0.114, "alpha": 0.03, "t": 259.7} |
| {"event": "train", "step": 2100, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.48227, "bpb": 2.13846, "chunk_loss": 0.030413, "F": 0.20016, "G": 0.2136, "chunks": 204.97, "bytes_per_chunk": 4.996, "boundary_entropy": 0.15209, "router_noise": 0.0, "grad_norm": 0.7147, "lr": 0.0002388, "router_trainable": true, "sec_per_step": 0.108, "alpha": 0.03, "t": 265.1} |
| {"event": "train", "step": 2150, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.49355, "bpb": 2.15474, "chunk_loss": 0.030572, "F": 0.20837, "G": 0.21971, "chunks": 213.38, "bytes_per_chunk": 4.799, "boundary_entropy": 0.15559, "router_noise": 0.0, "grad_norm": 0.7188, "lr": 0.0002357, "router_trainable": true, "sec_per_step": 0.115, "alpha": 0.03, "t": 270.9} |
| {"event": "probe", "step": 2200, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3373.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.193817, "cps": 26.351562, "mean_p": 0.205219, "boundary_entropy": 0.143876, "bpb": 2.100903}, "tr": {"chunks": 3730.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.206043, "cps": 29.140625, "mean_p": 0.209028, "boundary_entropy": 0.168649, "bpb": 2.311483}, "fi": {"chunks": 4840.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.257296, "cps": 37.8125, "mean_p": 0.25987, "boundary_entropy": 0.152526, "bpb": 2.454402}}, "bpb": {"en": 2.1009, "tr": 2.31148, "fi": 2.4544}, "gini_cpc": 0.064398, "gini_cps": 0.081889, "bpic": 4.75023, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0630802733228324, "fi": 1.3275205638726593}, "t": 276.7} |
| {"event": "eval", "step": 2200, "phase": 1, "bpb": {"en": 2.11206, "tr": 2.31874, "fi": 2.46634}, "bytes_per_chunk": {"en": 4.936, "tr": 5.277, "fi": 4.134}, "bpb_mean": 2.29905, "frozen_inventory": false, "t": 277.5} |
| {"event": "train", "step": 2200, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.53184, "bpb": 2.20998, "chunk_loss": 0.030694, "F": 0.2103, "G": 0.22211, "chunks": 215.34, "bytes_per_chunk": 4.755, "boundary_entropy": 0.14662, "router_noise": 0.0, "grad_norm": 0.7743, "lr": 0.0002325, "router_trainable": true, "sec_per_step": 0.141, "alpha": 0.03, "t": 277.9} |
| {"event": "train", "step": 2250, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.50729, "bpb": 2.17456, "chunk_loss": 0.030489, "F": 0.20419, "G": 0.21531, "chunks": 209.09, "bytes_per_chunk": 4.897, "boundary_entropy": 0.14065, "router_noise": 0.0, "grad_norm": 0.7783, "lr": 0.0002293, "router_trainable": true, "sec_per_step": 0.112, "alpha": 0.03, "t": 283.5} |
| {"event": "train", "step": 2300, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.54418, "bpb": 2.22778, "chunk_loss": 0.030508, "F": 0.20358, "G": 0.21616, "chunks": 208.47, "bytes_per_chunk": 4.912, "boundary_entropy": 0.14803, "router_noise": 0.0, "grad_norm": 0.7522, "lr": 0.000226, "router_trainable": true, "sec_per_step": 0.12, "alpha": 0.03, "t": 289.5} |
| {"event": "train", "step": 2350, "arch": "hnet_ffn", "phase": 1, "objective": "global", "lm_loss": 1.46351, "bpb": 2.1114, "chunk_loss": 0.030478, "F": 0.2034, "G": 0.21471, "chunks": 208.28, "bytes_per_chunk": 4.916, "boundary_entropy": 0.14228, "router_noise": 0.0, "grad_norm": 0.7184, "lr": 0.0002226, "router_trainable": true, "sec_per_step": 0.113, "alpha": 0.03, "t": 295.2} |
| {"event": "probe", "step": 2400, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3344.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.192151, "cps": 26.125, "mean_p": 0.204165, "boundary_entropy": 0.138866, "bpb": 2.077599}, "tr": {"chunks": 3728.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.205933, "cps": 29.125, "mean_p": 0.206931, "boundary_entropy": 0.160165, "bpb": 2.274274}, "fi": {"chunks": 4817.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.256074, "cps": 37.632812, "mean_p": 0.259772, "boundary_entropy": 0.153757, "bpb": 2.408025}}, "bpb": {"en": 2.0776, "tr": 2.27427, "fi": 2.40802}, "gini_cpc": 0.065145, "gini_cps": 0.082597, "bpic": 4.771806, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0717246103965956, "fi": 1.332669978422988}, "t": 301.3} |
| {"event": "eval", "step": 2400, "phase": 1, "bpb": {"en": 2.08979, "tr": 2.28018, "fi": 2.41876}, "bytes_per_chunk": {"en": 4.941, "tr": 5.31, "fi": 4.161}, "bpb_mean": 2.26291, "frozen_inventory": false, "t": 302.1} |
| {"event": "phase_transition", "step": 2400, "phase": 2, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0, "t": 302.3} |
| {"event": "train", "step": 2400, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.46824, "bpb": 2.11822, "chunk_loss": 0.030441, "F": 0.20282, "G": 0.21442, "chunks": 207.69, "bytes_per_chunk": 4.93, "boundary_entropy": 0.1418, "router_noise": 0.0, "grad_norm": 0.7176, "lr": 0.0002192, "router_trainable": true, "sec_per_step": 0.145, "alpha": 0.03, "t": 302.4} |
| {"event": "train", "step": 2450, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.47653, "bpb": 2.13018, "chunk_loss": 0.030707, "F": 0.21448, "G": 0.22506, "chunks": 219.62, "bytes_per_chunk": 4.662, "boundary_entropy": 0.13954, "router_noise": 0.0, "grad_norm": 0.8159, "lr": 0.0002158, "router_trainable": true, "sec_per_step": 0.109, "alpha": 0.03, "t": 307.9} |
| {"event": "train", "step": 2500, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.57709, "bpb": 2.27526, "chunk_loss": 0.030604, "F": 0.21057, "G": 0.22005, "chunks": 215.62, "bytes_per_chunk": 4.749, "boundary_entropy": 0.13986, "router_noise": 0.0, "grad_norm": 0.7468, "lr": 0.0002123, "router_trainable": true, "sec_per_step": 0.112, "alpha": 0.03, "t": 313.5} |
| {"event": "train", "step": 2550, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.4993, "bpb": 2.16303, "chunk_loss": 0.030481, "F": 0.20575, "G": 0.21713, "chunks": 210.69, "bytes_per_chunk": 4.86, "boundary_entropy": 0.13902, "router_noise": 0.0, "grad_norm": 0.7322, "lr": 0.0002088, "router_trainable": true, "sec_per_step": 0.112, "alpha": 0.03, "t": 319.1} |
| {"event": "probe", "step": 2600, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3356.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.19284, "cps": 26.21875, "mean_p": 0.200499, "boundary_entropy": 0.130232, "bpb": 2.056506}, "tr": {"chunks": 3767.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.208087, "cps": 29.429688, "mean_p": 0.205843, "boundary_entropy": 0.158479, "bpb": 2.254813}, "fi": {"chunks": 4823.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.256393, "cps": 37.679688, "mean_p": 0.256511, "boundary_entropy": 0.140947, "bpb": 2.364213}}, "bpb": {"en": 2.05651, "tr": 2.25481, "fi": 2.36421}, "gini_cpc": 0.064456, "gini_cps": 0.081868, "bpic": 4.749037, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0790640821884203, "fi": 1.3295587928828954}, "t": 324.8} |
| {"event": "eval", "step": 2600, "phase": 2, "bpb": {"en": 2.06739, "tr": 2.26298, "fi": 2.37594}, "bytes_per_chunk": {"en": 4.962, "tr": 5.256, "fi": 4.162}, "bpb_mean": 2.23544, "frozen_inventory": false, "t": 325.6} |
| {"event": "train", "step": 2600, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.50656, "bpb": 2.17351, "chunk_loss": 0.030606, "F": 0.20972, "G": 0.21804, "chunks": 214.75, "bytes_per_chunk": 4.768, "boundary_entropy": 0.13115, "router_noise": 0.0, "grad_norm": 0.7146, "lr": 0.0002053, "router_trainable": true, "sec_per_step": 0.141, "alpha": 0.03, "t": 326.2} |
| {"event": "train", "step": 2650, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.44347, "bpb": 2.08248, "chunk_loss": 0.03038, "F": 0.19989, "G": 0.209, "chunks": 204.69, "bytes_per_chunk": 5.003, "boundary_entropy": 0.13323, "router_noise": 0.0, "grad_norm": 0.6805, "lr": 0.0002017, "router_trainable": true, "sec_per_step": 0.119, "alpha": 0.03, "t": 332.1} |
| {"event": "train", "step": 2700, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.52455, "bpb": 2.19947, "chunk_loss": 0.030501, "F": 0.2048, "G": 0.21468, "chunks": 209.72, "bytes_per_chunk": 4.883, "boundary_entropy": 0.13687, "router_noise": 0.0, "grad_norm": 0.8296, "lr": 0.0001981, "router_trainable": true, "sec_per_step": 0.116, "alpha": 0.03, "t": 337.9} |
| {"event": "train", "step": 2750, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.4631, "bpb": 2.11081, "chunk_loss": 0.030514, "F": 0.20792, "G": 0.21674, "chunks": 212.91, "bytes_per_chunk": 4.81, "boundary_entropy": 0.13227, "router_noise": 0.0, "grad_norm": 0.7358, "lr": 0.0001945, "router_trainable": true, "sec_per_step": 0.11, "alpha": 0.03, "t": 343.4} |
| {"event": "probe", "step": 2800, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3316.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.190542, "cps": 25.90625, "mean_p": 0.199636, "boundary_entropy": 0.129569, "bpb": 2.032526}, "tr": {"chunks": 3741.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.206651, "cps": 29.226562, "mean_p": 0.202928, "boundary_entropy": 0.150868, "bpb": 2.232637}, "fi": {"chunks": 4795.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.254904, "cps": 37.460938, "mean_p": 0.255498, "boundary_entropy": 0.140397, "bpb": 2.338874}}, "bpb": {"en": 2.03253, "tr": 2.23264, "fi": 2.33887}, "gini_cpc": 0.0658, "gini_cps": 0.083193, "bpic": 4.786703, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0845429487491727, "fi": 1.3377850132474525}, "t": 349.0} |
| {"event": "eval", "step": 2800, "phase": 2, "bpb": {"en": 2.04507, "tr": 2.23897, "fi": 2.34996}, "bytes_per_chunk": {"en": 4.956, "tr": 5.27, "fi": 4.194}, "bpb_mean": 2.21133, "frozen_inventory": false, "t": 349.8} |
| {"event": "train", "step": 2800, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.44697, "bpb": 2.08754, "chunk_loss": 0.030508, "F": 0.20834, "G": 0.21493, "chunks": 213.34, "bytes_per_chunk": 4.8, "boundary_entropy": 0.13014, "router_noise": 0.0, "grad_norm": 0.6912, "lr": 0.0001909, "router_trainable": true, "sec_per_step": 0.137, "alpha": 0.03, "t": 350.2} |
| {"event": "train", "step": 2850, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.55708, "bpb": 2.24639, "chunk_loss": 0.03073, "F": 0.21304, "G": 0.22343, "chunks": 218.16, "bytes_per_chunk": 4.694, "boundary_entropy": 0.12778, "router_noise": 0.0, "grad_norm": 0.748, "lr": 0.0001872, "router_trainable": true, "sec_per_step": 0.108, "alpha": 0.03, "t": 355.6} |
| {"event": "train", "step": 2900, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.50212, "bpb": 2.1671, "chunk_loss": 0.030738, "F": 0.21573, "G": 0.22362, "chunks": 220.91, "bytes_per_chunk": 4.635, "boundary_entropy": 0.12967, "router_noise": 0.0, "grad_norm": 0.6933, "lr": 0.0001835, "router_trainable": true, "sec_per_step": 0.116, "alpha": 0.03, "t": 361.4} |
| {"event": "train", "step": 2950, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.43395, "bpb": 2.06875, "chunk_loss": 0.030437, "F": 0.20197, "G": 0.21081, "chunks": 206.81, "bytes_per_chunk": 4.951, "boundary_entropy": 0.12759, "router_noise": 0.0, "grad_norm": 0.6869, "lr": 0.0001799, "router_trainable": true, "sec_per_step": 0.107, "alpha": 0.03, "t": 366.8} |
| {"event": "probe", "step": 3000, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3340.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.191921, "cps": 26.09375, "mean_p": 0.203272, "boundary_entropy": 0.119789, "bpb": 2.015929}, "tr": {"chunks": 3735.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.206319, "cps": 29.179688, "mean_p": 0.203993, "boundary_entropy": 0.146013, "bpb": 2.204685}, "fi": {"chunks": 4789.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.254585, "cps": 37.414062, "mean_p": 0.256304, "boundary_entropy": 0.140517, "bpb": 2.317438}}, "bpb": {"en": 2.01593, "tr": 2.20469, "fi": 2.31744}, "gini_cpc": 0.063993, "gini_cps": 0.081423, "bpic": 4.781861, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0750228813763956, "fi": 1.326510240377254}, "t": 372.5} |
| {"event": "eval", "step": 3000, "phase": 2, "bpb": {"en": 2.02915, "tr": 2.212, "fi": 2.32763}, "bytes_per_chunk": {"en": 4.925, "tr": 5.3, "fi": 4.194}, "bpb_mean": 2.18959, "frozen_inventory": false, "t": 373.4} |
| {"event": "train", "step": 3000, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.45727, "bpb": 2.10239, "chunk_loss": 0.030448, "F": 0.20529, "G": 0.2142, "chunks": 210.22, "bytes_per_chunk": 4.871, "boundary_entropy": 0.12246, "router_noise": 0.0, "grad_norm": 0.7372, "lr": 0.0001761, "router_trainable": true, "sec_per_step": 0.145, "alpha": 0.03, "t": 374.0} |
| {"event": "train", "step": 3050, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.4494, "bpb": 2.09104, "chunk_loss": 0.03055, "F": 0.21039, "G": 0.21876, "chunks": 215.44, "bytes_per_chunk": 4.753, "boundary_entropy": 0.1234, "router_noise": 0.0, "grad_norm": 0.7282, "lr": 0.0001724, "router_trainable": true, "sec_per_step": 0.111, "alpha": 0.03, "t": 379.6} |
| {"event": "train", "step": 3100, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.54761, "bpb": 2.23273, "chunk_loss": 0.030563, "F": 0.20502, "G": 0.21341, "chunks": 209.94, "bytes_per_chunk": 4.878, "boundary_entropy": 0.12561, "router_noise": 0.0, "grad_norm": 0.7962, "lr": 0.0001687, "router_trainable": true, "sec_per_step": 0.109, "alpha": 0.03, "t": 385.0} |
| {"event": "train", "step": 3150, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.46679, "bpb": 2.11613, "chunk_loss": 0.030523, "F": 0.20712, "G": 0.21502, "chunks": 212.09, "bytes_per_chunk": 4.828, "boundary_entropy": 0.12141, "router_noise": 0.0, "grad_norm": 0.7134, "lr": 0.000165, "router_trainable": true, "sec_per_step": 0.105, "alpha": 0.03, "t": 390.3} |
| {"event": "probe", "step": 3200, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3392.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.194909, "cps": 26.5, "mean_p": 0.202367, "boundary_entropy": 0.112241, "bpb": 1.998566}, "tr": {"chunks": 3721.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.205546, "cps": 29.070312, "mean_p": 0.201875, "boundary_entropy": 0.141861, "bpb": 2.181966}, "fi": {"chunks": 4822.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.256339, "cps": 37.671875, "mean_p": 0.255309, "boundary_entropy": 0.130014, "bpb": 2.290655}}, "bpb": {"en": 1.99857, "tr": 2.18197, "fi": 2.29066}, "gini_cpc": 0.062354, "gini_cps": 0.079877, "bpic": 4.753414, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0545748144266716, "fi": 1.3151751647219663}, "t": 396.3} |
| {"event": "eval", "step": 3200, "phase": 2, "bpb": {"en": 2.01208, "tr": 2.19145, "fi": 2.30073}, "bytes_per_chunk": {"en": 4.902, "tr": 5.309, "fi": 4.145}, "bpb_mean": 2.16809, "frozen_inventory": false, "t": 397.2} |
| {"event": "train", "step": 3200, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.43008, "bpb": 2.06317, "chunk_loss": 0.03067, "F": 0.21323, "G": 0.22034, "chunks": 218.34, "bytes_per_chunk": 4.69, "boundary_entropy": 0.11708, "router_noise": 0.0, "grad_norm": 0.7072, "lr": 0.0001613, "router_trainable": true, "sec_per_step": 0.145, "alpha": 0.03, "t": 397.5} |
| {"event": "train", "step": 3250, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.41058, "bpb": 2.03504, "chunk_loss": 0.030363, "F": 0.20108, "G": 0.20785, "chunks": 205.91, "bytes_per_chunk": 4.973, "boundary_entropy": 0.11701, "router_noise": 0.0, "grad_norm": 0.6831, "lr": 0.0001576, "router_trainable": true, "sec_per_step": 0.114, "alpha": 0.03, "t": 403.2} |
| {"event": "train", "step": 3300, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.45326, "bpb": 2.09661, "chunk_loss": 0.030554, "F": 0.20963, "G": 0.21569, "chunks": 214.66, "bytes_per_chunk": 4.77, "boundary_entropy": 0.11911, "router_noise": 0.0, "grad_norm": 0.7028, "lr": 0.0001539, "router_trainable": true, "sec_per_step": 0.114, "alpha": 0.03, "t": 408.9} |
| {"event": "train", "step": 3350, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.48574, "bpb": 2.14347, "chunk_loss": 0.030644, "F": 0.21448, "G": 0.21931, "chunks": 219.62, "bytes_per_chunk": 4.662, "boundary_entropy": 0.11983, "router_noise": 0.0, "grad_norm": 0.8008, "lr": 0.0001501, "router_trainable": true, "sec_per_step": 0.109, "alpha": 0.03, "t": 414.3} |
| {"event": "probe", "step": 3400, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3387.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.194622, "cps": 26.460938, "mean_p": 0.200715, "boundary_entropy": 0.111301, "bpb": 1.985466}, "tr": {"chunks": 3694.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.204055, "cps": 28.859375, "mean_p": 0.200686, "boundary_entropy": 0.141686, "bpb": 2.157479}, "fi": {"chunks": 4840.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.257296, "cps": 37.8125, "mean_p": 0.253235, "boundary_entropy": 0.12936, "bpb": 2.266892}}, "bpb": {"en": 1.98547, "tr": 2.15748, "fi": 2.26689}, "gini_cpc": 0.063696, "gini_cps": 0.081257, "bpic": 4.758997, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0484682008820014, "fi": 1.3220333220969824}, "t": 420.2} |
| {"event": "eval", "step": 3400, "phase": 2, "bpb": {"en": 1.99808, "tr": 2.1655, "fi": 2.27599}, "bytes_per_chunk": {"en": 4.926, "tr": 5.396, "fi": 4.152}, "bpb_mean": 2.14652, "frozen_inventory": false, "t": 421.1} |
| {"event": "train", "step": 3400, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.39974, "bpb": 2.0194, "chunk_loss": 0.030414, "F": 0.2009, "G": 0.20783, "chunks": 205.72, "bytes_per_chunk": 4.978, "boundary_entropy": 0.11545, "router_noise": 0.0, "grad_norm": 0.7074, "lr": 0.0001465, "router_trainable": true, "sec_per_step": 0.141, "alpha": 0.03, "t": 421.4} |
| {"event": "train", "step": 3450, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.39381, "bpb": 2.01085, "chunk_loss": 0.030337, "F": 0.19897, "G": 0.20741, "chunks": 203.75, "bytes_per_chunk": 5.026, "boundary_entropy": 0.11406, "router_noise": 0.0, "grad_norm": 0.664, "lr": 0.0001428, "router_trainable": true, "sec_per_step": 0.12, "alpha": 0.03, "t": 427.4} |
| {"event": "train", "step": 3500, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.47491, "bpb": 2.12785, "chunk_loss": 0.030586, "F": 0.20697, "G": 0.21497, "chunks": 211.94, "bytes_per_chunk": 4.832, "boundary_entropy": 0.11837, "router_noise": 0.0, "grad_norm": 0.6895, "lr": 0.0001391, "router_trainable": true, "sec_per_step": 0.127, "alpha": 0.03, "t": 433.7} |
| {"event": "train", "step": 3550, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.47964, "bpb": 2.13467, "chunk_loss": 0.030733, "F": 0.21658, "G": 0.22236, "chunks": 221.78, "bytes_per_chunk": 4.617, "boundary_entropy": 0.12099, "router_noise": 0.0, "grad_norm": 0.7244, "lr": 0.0001355, "router_trainable": true, "sec_per_step": 0.123, "alpha": 0.03, "t": 439.9} |
| {"event": "probe", "step": 3600, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3355.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.192783, "cps": 26.210938, "mean_p": 0.200004, "boundary_entropy": 0.108599, "bpb": 1.970308}, "tr": {"chunks": 3715.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.205215, "cps": 29.023438, "mean_p": 0.20361, "boundary_entropy": 0.13843, "bpb": 2.145824}, "fi": {"chunks": 4830.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.256765, "cps": 37.734375, "mean_p": 0.252973, "boundary_entropy": 0.12591, "bpb": 2.245489}}, "bpb": {"en": 1.97031, "tr": 2.14582, "fi": 2.24549}, "gini_cpc": 0.065145, "gini_cps": 0.082633, "bpic": 4.767395, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0644857753443142, "fi": 1.331885353252342}, "t": 446.7} |
| {"event": "eval", "step": 3600, "phase": 2, "bpb": {"en": 1.9847, "tr": 2.15268, "fi": 2.25562}, "bytes_per_chunk": {"en": 4.946, "tr": 5.368, "fi": 4.154}, "bpb_mean": 2.131, "frozen_inventory": false, "t": 447.8} |
| {"event": "train", "step": 3600, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.4672, "bpb": 2.11673, "chunk_loss": 0.030457, "F": 0.20493, "G": 0.21157, "chunks": 209.84, "bytes_per_chunk": 4.88, "boundary_entropy": 0.11375, "router_noise": 0.0, "grad_norm": 0.7172, "lr": 0.0001319, "router_trainable": true, "sec_per_step": 0.165, "alpha": 0.03, "t": 448.1} |
| {"event": "train", "step": 3650, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.48178, "bpb": 2.13776, "chunk_loss": 0.030651, "F": 0.21289, "G": 0.21866, "chunks": 218.0, "bytes_per_chunk": 4.697, "boundary_entropy": 0.11735, "router_noise": 0.0, "grad_norm": 0.7411, "lr": 0.0001283, "router_trainable": true, "sec_per_step": 0.126, "alpha": 0.03, "t": 454.4} |
| {"event": "train", "step": 3700, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.38223, "bpb": 1.99414, "chunk_loss": 0.030409, "F": 0.202, "G": 0.20925, "chunks": 206.84, "bytes_per_chunk": 4.951, "boundary_entropy": 0.1148, "router_noise": 0.0, "grad_norm": 0.7111, "lr": 0.0001247, "router_trainable": true, "sec_per_step": 0.112, "alpha": 0.03, "t": 460.0} |
| {"event": "train", "step": 3750, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.40089, "bpb": 2.02106, "chunk_loss": 0.03062, "F": 0.2135, "G": 0.21707, "chunks": 218.62, "bytes_per_chunk": 4.684, "boundary_entropy": 0.1081, "router_noise": 0.0, "grad_norm": 0.6527, "lr": 0.0001212, "router_trainable": true, "sec_per_step": 0.106, "alpha": 0.03, "t": 465.3} |
| {"event": "probe", "step": 3800, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3331.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.191404, "cps": 26.023438, "mean_p": 0.199641, "boundary_entropy": 0.106436, "bpb": 1.957489}, "tr": {"chunks": 3745.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.206872, "cps": 29.257812, "mean_p": 0.201994, "boundary_entropy": 0.13526, "bpb": 2.123919}, "fi": {"chunks": 4828.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.256658, "cps": 37.71875, "mean_p": 0.252169, "boundary_entropy": 0.121874, "bpb": 2.223716}}, "bpb": {"en": 1.95749, "tr": 2.12392, "fi": 2.22372}, "gini_cpc": 0.066424, "gini_cps": 0.083837, "bpic": 4.765793, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0808134937122948, "fi": 1.3409261662580108}, "t": 471.1} |
| {"event": "eval", "step": 3800, "phase": 2, "bpb": {"en": 1.97046, "tr": 2.13077, "fi": 2.23246}, "bytes_per_chunk": {"en": 4.93, "tr": 5.313, "fi": 4.16}, "bpb_mean": 2.11123, "frozen_inventory": false, "t": 472.0} |
| {"event": "train", "step": 3800, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.37994, "bpb": 1.99084, "chunk_loss": 0.030454, "F": 0.20111, "G": 0.20813, "chunks": 205.94, "bytes_per_chunk": 4.972, "boundary_entropy": 0.11152, "router_noise": 0.0, "grad_norm": 0.6327, "lr": 0.0001177, "router_trainable": true, "sec_per_step": 0.14, "alpha": 0.03, "t": 472.3} |
| {"event": "train", "step": 3850, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.37222, "bpb": 1.97969, "chunk_loss": 0.03033, "F": 0.19937, "G": 0.20404, "chunks": 204.16, "bytes_per_chunk": 5.016, "boundary_entropy": 0.11028, "router_noise": 0.0, "grad_norm": 0.7358, "lr": 0.0001142, "router_trainable": true, "sec_per_step": 0.108, "alpha": 0.03, "t": 477.7} |
| {"event": "train", "step": 3900, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.40503, "bpb": 2.02703, "chunk_loss": 0.030424, "F": 0.20303, "G": 0.20829, "chunks": 207.91, "bytes_per_chunk": 4.925, "boundary_entropy": 0.11494, "router_noise": 0.0, "grad_norm": 0.6724, "lr": 0.0001108, "router_trainable": true, "sec_per_step": 0.112, "alpha": 0.03, "t": 483.3} |
| {"event": "train", "step": 3950, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.43745, "bpb": 2.07381, "chunk_loss": 0.030506, "F": 0.20755, "G": 0.21322, "chunks": 212.53, "bytes_per_chunk": 4.818, "boundary_entropy": 0.10595, "router_noise": 0.0, "grad_norm": 0.7044, "lr": 0.0001074, "router_trainable": true, "sec_per_step": 0.106, "alpha": 0.03, "t": 488.6} |
| {"event": "probe", "step": 4000, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3376.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.19399, "cps": 26.375, "mean_p": 0.198088, "boundary_entropy": 0.102367, "bpb": 1.945413}, "tr": {"chunks": 3668.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.202618, "cps": 28.65625, "mean_p": 0.198746, "boundary_entropy": 0.137914, "bpb": 2.106031}, "fi": {"chunks": 4832.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.256871, "cps": 37.75, "mean_p": 0.251915, "boundary_entropy": 0.121145, "bpb": 2.209109}}, "bpb": {"en": 1.94541, "tr": 2.10603, "fi": 2.20911}, "gini_cpc": 0.06415, "gini_cps": 0.081733, "bpic": 4.777029, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0444807922438557, "fi": 1.3241485961249353}, "t": 494.6} |
| {"event": "eval", "step": 4000, "phase": 2, "bpb": {"en": 1.95918, "tr": 2.11505, "fi": 2.21879}, "bytes_per_chunk": {"en": 4.937, "tr": 5.402, "fi": 4.15}, "bpb_mean": 2.09767, "frozen_inventory": false, "t": 495.5} |
| {"event": "train", "step": 4000, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.45309, "bpb": 2.09637, "chunk_loss": 0.030505, "F": 0.20865, "G": 0.21427, "chunks": 213.66, "bytes_per_chunk": 4.793, "boundary_entropy": 0.10878, "router_noise": 0.0, "grad_norm": 0.7654, "lr": 0.000104, "router_trainable": true, "sec_per_step": 0.144, "alpha": 0.03, "t": 495.8} |
| {"event": "train", "step": 4050, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.37545, "bpb": 1.98435, "chunk_loss": 0.030444, "F": 0.20291, "G": 0.20896, "chunks": 207.78, "bytes_per_chunk": 4.928, "boundary_entropy": 0.10291, "router_noise": 0.0, "grad_norm": 0.6744, "lr": 0.0001007, "router_trainable": true, "sec_per_step": 0.11, "alpha": 0.03, "t": 501.3} |
| {"event": "train", "step": 4100, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.44289, "bpb": 2.08165, "chunk_loss": 0.030639, "F": 0.21341, "G": 0.21796, "chunks": 218.53, "bytes_per_chunk": 4.686, "boundary_entropy": 0.10955, "router_noise": 0.0, "grad_norm": 0.6783, "lr": 9.75e-05, "router_trainable": true, "sec_per_step": 0.108, "alpha": 0.03, "t": 506.7} |
| {"event": "train", "step": 4150, "arch": "hnet_ffn", "phase": 2, "objective": "global", "lm_loss": 1.39514, "bpb": 2.01276, "chunk_loss": 0.030709, "F": 0.21704, "G": 0.22045, "chunks": 222.25, "bytes_per_chunk": 4.607, "boundary_entropy": 0.1078, "router_noise": 0.0, "grad_norm": 0.731, "lr": 9.43e-05, "router_trainable": true, "sec_per_step": 0.104, "alpha": 0.03, "t": 511.9} |
| {"event": "probe", "step": 4200, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3341.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.191978, "cps": 26.101562, "mean_p": 0.197834, "boundary_entropy": 0.099108, "bpb": 1.932532}, "tr": {"chunks": 3727.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.205877, "cps": 29.117188, "mean_p": 0.198586, "boundary_entropy": 0.130412, "bpb": 2.088018}, "fi": {"chunks": 4846.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.257615, "cps": 37.859375, "mean_p": 0.250901, "boundary_entropy": 0.119398, "bpb": 2.187039}}, "bpb": {"en": 1.93253, "tr": 2.08802, "fi": 2.18704}, "gini_cpc": 0.066758, "gini_cps": 0.084215, "bpic": 4.761793, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.072399211251232, "fi": 1.341896965881773}, "t": 517.9} |
| {"event": "eval", "step": 4200, "phase": 2, "bpb": {"en": 1.94544, "tr": 2.09698, "fi": 2.1961}, "bytes_per_chunk": {"en": 4.932, "tr": 5.276, "fi": 4.148}, "bpb_mean": 2.07951, "frozen_inventory": false, "t": 518.8} |
| {"event": "phase_transition", "step": 4200, "phase": 3, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0, "t": 519.1} |
| {"event": "train", "step": 4200, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.38581, "bpb": 1.9993, "chunk_loss": 0.030362, "F": 0.202, "G": 0.20673, "chunks": 206.84, "bytes_per_chunk": 4.951, "boundary_entropy": 0.10326, "router_noise": 0.0, "grad_norm": 0.6745, "lr": 9.12e-05, "router_trainable": true, "sec_per_step": 0.146, "alpha": 0.03, "t": 519.2} |
| {"event": "train", "step": 4250, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.38435, "bpb": 1.99719, "chunk_loss": 0.030361, "F": 0.19839, "G": 0.20407, "chunks": 203.16, "bytes_per_chunk": 5.04, "boundary_entropy": 0.10716, "router_noise": 0.0, "grad_norm": 0.708, "lr": 8.81e-05, "router_trainable": true, "sec_per_step": 0.111, "alpha": 0.03, "t": 524.7} |
| {"event": "train", "step": 4300, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.47482, "bpb": 2.12772, "chunk_loss": 0.030519, "F": 0.20749, "G": 0.21221, "chunks": 212.47, "bytes_per_chunk": 4.82, "boundary_entropy": 0.11487, "router_noise": 0.0, "grad_norm": 0.7553, "lr": 8.5e-05, "router_trainable": true, "sec_per_step": 0.113, "alpha": 0.03, "t": 530.4} |
| {"event": "train", "step": 4350, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.33032, "bpb": 1.91925, "chunk_loss": 0.030467, "F": 0.2041, "G": 0.2081, "chunks": 209.0, "bytes_per_chunk": 4.9, "boundary_entropy": 0.10236, "router_noise": 0.0, "grad_norm": 0.7385, "lr": 8.21e-05, "router_trainable": true, "sec_per_step": 0.112, "alpha": 0.03, "t": 536.0} |
| {"event": "probe", "step": 4400, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3334.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.191576, "cps": 26.046875, "mean_p": 0.197678, "boundary_entropy": 0.095962, "bpb": 1.92058}, "tr": {"chunks": 3699.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.204331, "cps": 28.898438, "mean_p": 0.200671, "boundary_entropy": 0.133267, "bpb": 2.078135}, "fi": {"chunks": 4858.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.258253, "cps": 37.953125, "mean_p": 0.250389, "boundary_entropy": 0.117707, "bpb": 2.173081}}, "bpb": {"en": 1.92058, "tr": 2.07814, "fi": 2.17308}, "gini_cpc": 0.067952, "gini_cps": 0.085443, "bpic": 4.771003, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0665772220355687, "fi": 1.3480442606922771}, "t": 542.1} |
| {"event": "eval", "step": 4400, "phase": 3, "bpb": {"en": 1.936, "tr": 2.08527, "fi": 2.18137}, "bytes_per_chunk": {"en": 4.926, "tr": 5.326, "fi": 4.134}, "bpb_mean": 2.06755, "frozen_inventory": false, "t": 542.9} |
| {"event": "train", "step": 4400, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.44473, "bpb": 2.0843, "chunk_loss": 0.030711, "F": 0.21262, "G": 0.21738, "chunks": 217.72, "bytes_per_chunk": 4.703, "boundary_entropy": 0.10625, "router_noise": 0.0, "grad_norm": 0.7222, "lr": 7.92e-05, "router_trainable": true, "sec_per_step": 0.147, "alpha": 0.03, "t": 543.3} |
| {"event": "train", "step": 4450, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.37996, "bpb": 1.99086, "chunk_loss": 0.030538, "F": 0.2074, "G": 0.21308, "chunks": 212.38, "bytes_per_chunk": 4.822, "boundary_entropy": 0.10405, "router_noise": 0.0, "grad_norm": 0.6805, "lr": 7.63e-05, "router_trainable": true, "sec_per_step": 0.11, "alpha": 0.03, "t": 548.8} |
| {"event": "train", "step": 4500, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.39712, "bpb": 2.01562, "chunk_loss": 0.03058, "F": 0.21106, "G": 0.21448, "chunks": 216.12, "bytes_per_chunk": 4.738, "boundary_entropy": 0.1055, "router_noise": 0.0, "grad_norm": 0.7211, "lr": 7.36e-05, "router_trainable": true, "sec_per_step": 0.11, "alpha": 0.03, "t": 554.3} |
| {"event": "train", "step": 4550, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.37332, "bpb": 1.98128, "chunk_loss": 0.030526, "F": 0.20947, "G": 0.21342, "chunks": 214.5, "bytes_per_chunk": 4.774, "boundary_entropy": 0.10342, "router_noise": 0.0, "grad_norm": 0.7135, "lr": 7.09e-05, "router_trainable": true, "sec_per_step": 0.105, "alpha": 0.03, "t": 559.6} |
| {"event": "probe", "step": 4600, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3385.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.194507, "cps": 26.445312, "mean_p": 0.198929, "boundary_entropy": 0.095317, "bpb": 1.911109}, "tr": {"chunks": 3711.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.204994, "cps": 28.992188, "mean_p": 0.197494, "boundary_entropy": 0.130493, "bpb": 2.062855}, "fi": {"chunks": 4861.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.258413, "cps": 37.976562, "mean_p": 0.250911, "boundary_entropy": 0.114752, "bpb": 2.159513}}, "bpb": {"en": 1.91111, "tr": 2.06286, "fi": 2.15951}, "gini_cpc": 0.064756, "gini_cps": 0.082295, "bpic": 4.744668, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0539156415884128, "fi": 1.3285539189608015}, "t": 565.5} |
| {"event": "eval", "step": 4600, "phase": 3, "bpb": {"en": 1.92585, "tr": 2.07107, "fi": 2.16828}, "bytes_per_chunk": {"en": 4.912, "tr": 5.323, "fi": 4.14}, "bpb_mean": 2.05507, "frozen_inventory": false, "t": 566.4} |
| {"event": "train", "step": 4600, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.43647, "bpb": 2.07239, "chunk_loss": 0.030462, "F": 0.20398, "G": 0.20823, "chunks": 208.88, "bytes_per_chunk": 4.902, "boundary_entropy": 0.10303, "router_noise": 0.0, "grad_norm": 0.7473, "lr": 6.82e-05, "router_trainable": true, "sec_per_step": 0.146, "alpha": 0.03, "t": 566.9} |
| {"event": "train", "step": 4650, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.34597, "bpb": 1.94183, "chunk_loss": 0.030427, "F": 0.20435, "G": 0.20796, "chunks": 209.25, "bytes_per_chunk": 4.894, "boundary_entropy": 0.10766, "router_noise": 0.0, "grad_norm": 0.6819, "lr": 6.57e-05, "router_trainable": true, "sec_per_step": 0.109, "alpha": 0.03, "t": 572.3} |
| {"event": "train", "step": 4700, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.36761, "bpb": 1.97305, "chunk_loss": 0.030469, "F": 0.20407, "G": 0.20809, "chunks": 208.97, "bytes_per_chunk": 4.9, "boundary_entropy": 0.10373, "router_noise": 0.0, "grad_norm": 0.6776, "lr": 6.32e-05, "router_trainable": true, "sec_per_step": 0.107, "alpha": 0.03, "t": 577.6} |
| {"event": "train", "step": 4750, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.32995, "bpb": 1.91871, "chunk_loss": 0.030489, "F": 0.20728, "G": 0.212, "chunks": 212.25, "bytes_per_chunk": 4.824, "boundary_entropy": 0.10294, "router_noise": 0.0, "grad_norm": 0.6737, "lr": 6.08e-05, "router_trainable": true, "sec_per_step": 0.108, "alpha": 0.03, "t": 583.0} |
| {"event": "probe", "step": 4800, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3345.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.192208, "cps": 26.132812, "mean_p": 0.197707, "boundary_entropy": 0.093402, "bpb": 1.902881}, "tr": {"chunks": 3610.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.199414, "cps": 28.203125, "mean_p": 0.195718, "boundary_entropy": 0.131566, "bpb": 2.053144}, "fi": {"chunks": 4846.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.257615, "cps": 37.859375, "mean_p": 0.249343, "boundary_entropy": 0.115588, "bpb": 2.147541}}, "bpb": {"en": 1.90288, "tr": 2.05314, "fi": 2.14754}, "gini_cpc": 0.067163, "gini_cps": 0.084795, "bpic": 4.807389, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0374917419479814, "fi": 1.3402923058328862}, "t": 588.9} |
| {"event": "eval", "step": 4800, "phase": 3, "bpb": {"en": 1.91669, "tr": 2.06062, "fi": 2.15702}, "bytes_per_chunk": {"en": 4.924, "tr": 5.461, "fi": 4.147}, "bpb_mean": 2.04478, "frozen_inventory": false, "t": 589.8} |
| {"event": "train", "step": 4800, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.40793, "bpb": 2.03122, "chunk_loss": 0.030766, "F": 0.2178, "G": 0.22225, "chunks": 223.03, "bytes_per_chunk": 4.591, "boundary_entropy": 0.10393, "router_noise": 0.0, "grad_norm": 0.7587, "lr": 5.85e-05, "router_trainable": true, "sec_per_step": 0.143, "alpha": 0.03, "t": 590.2} |
| {"event": "train", "step": 4850, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.35149, "bpb": 1.94979, "chunk_loss": 0.030616, "F": 0.21155, "G": 0.21602, "chunks": 216.62, "bytes_per_chunk": 4.727, "boundary_entropy": 0.10377, "router_noise": 0.0, "grad_norm": 0.7261, "lr": 5.62e-05, "router_trainable": true, "sec_per_step": 0.108, "alpha": 0.03, "t": 595.6} |
| {"event": "train", "step": 4900, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.36252, "bpb": 1.96571, "chunk_loss": 0.030365, "F": 0.20035, "G": 0.20529, "chunks": 205.16, "bytes_per_chunk": 4.991, "boundary_entropy": 0.0988, "router_noise": 0.0, "grad_norm": 0.7271, "lr": 5.41e-05, "router_trainable": true, "sec_per_step": 0.108, "alpha": 0.03, "t": 601.0} |
| {"event": "train", "step": 4950, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.34519, "bpb": 1.9407, "chunk_loss": 0.030541, "F": 0.20889, "G": 0.2128, "chunks": 213.91, "bytes_per_chunk": 4.787, "boundary_entropy": 0.09659, "router_noise": 0.0, "grad_norm": 0.698, "lr": 5.2e-05, "router_trainable": true, "sec_per_step": 0.106, "alpha": 0.03, "t": 606.3} |
| {"event": "probe", "step": 5000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3328.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.191231, "cps": 26.0, "mean_p": 0.196166, "boundary_entropy": 0.091679, "bpb": 1.893708}, "tr": {"chunks": 3645.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.201348, "cps": 28.476562, "mean_p": 0.195665, "boundary_entropy": 0.131071, "bpb": 2.043049}, "fi": {"chunks": 4832.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.256871, "cps": 37.75, "mean_p": 0.249329, "boundary_entropy": 0.11692, "bpb": 2.134237}}, "bpb": {"en": 1.89371, "tr": 2.04305, "fi": 2.13424}, "gini_cpc": 0.06738, "gini_cps": 0.084936, "bpic": 4.80576, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0529015955440875, "fi": 1.3432468931844297}, "t": 612.3} |
| {"event": "eval", "step": 5000, "phase": 3, "bpb": {"en": 1.9086, "tr": 2.05022, "fi": 2.14395}, "bytes_per_chunk": {"en": 4.931, "tr": 5.406, "fi": 4.16}, "bpb_mean": 2.03426, "frozen_inventory": false, "t": 613.2} |
| {"event": "train", "step": 5000, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.34801, "bpb": 1.94476, "chunk_loss": 0.030528, "F": 0.20789, "G": 0.21188, "chunks": 212.88, "bytes_per_chunk": 4.81, "boundary_entropy": 0.09976, "router_noise": 0.0, "grad_norm": 0.7041, "lr": 5e-05, "router_trainable": true, "sec_per_step": 0.144, "alpha": 0.03, "t": 613.5} |
| {"event": "train", "step": 5050, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.36487, "bpb": 1.9691, "chunk_loss": 0.030454, "F": 0.20633, "G": 0.20997, "chunks": 211.28, "bytes_per_chunk": 4.847, "boundary_entropy": 0.10057, "router_noise": 0.0, "grad_norm": 0.6848, "lr": 4.81e-05, "router_trainable": true, "sec_per_step": 0.11, "alpha": 0.03, "t": 619.1} |
| {"event": "train", "step": 5100, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.3301, "bpb": 1.91893, "chunk_loss": 0.030308, "F": 0.19931, "G": 0.20249, "chunks": 204.09, "bytes_per_chunk": 5.017, "boundary_entropy": 0.09588, "router_noise": 0.0, "grad_norm": 0.6956, "lr": 4.63e-05, "router_trainable": true, "sec_per_step": 0.104, "alpha": 0.03, "t": 624.3} |
| {"event": "train", "step": 5150, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.31989, "bpb": 1.9042, "chunk_loss": 0.030513, "F": 0.2077, "G": 0.21136, "chunks": 212.69, "bytes_per_chunk": 4.815, "boundary_entropy": 0.10005, "router_noise": 0.0, "grad_norm": 0.6971, "lr": 4.45e-05, "router_trainable": true, "sec_per_step": 0.11, "alpha": 0.03, "t": 629.8} |
| {"event": "probe", "step": 5200, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3329.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.191289, "cps": 26.007812, "mean_p": 0.196323, "boundary_entropy": 0.094266, "bpb": 1.886176}, "tr": {"chunks": 3637.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.200906, "cps": 28.414062, "mean_p": 0.19727, "boundary_entropy": 0.130108, "bpb": 2.032548}, "fi": {"chunks": 4849.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.257775, "cps": 37.882812, "mean_p": 0.249568, "boundary_entropy": 0.114841, "bpb": 2.126074}}, "bpb": {"en": 1.88618, "tr": 2.03255, "fi": 2.12607}, "gini_cpc": 0.068194, "gini_cps": 0.085767, "bpic": 4.801693, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0502751129359953, "fi": 1.3475678022064483}, "t": 635.6} |
| {"event": "eval", "step": 5200, "phase": 3, "bpb": {"en": 1.90141, "tr": 2.0412, "fi": 2.13533}, "bytes_per_chunk": {"en": 4.932, "tr": 5.413, "fi": 4.139}, "bpb_mean": 2.02598, "frozen_inventory": false, "t": 636.6} |
| {"event": "train", "step": 5200, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.36058, "bpb": 1.96291, "chunk_loss": 0.030686, "F": 0.21472, "G": 0.21823, "chunks": 219.88, "bytes_per_chunk": 4.657, "boundary_entropy": 0.10394, "router_noise": 0.0, "grad_norm": 0.7016, "lr": 4.29e-05, "router_trainable": true, "sec_per_step": 0.144, "alpha": 0.03, "t": 637.0} |
| {"event": "train", "step": 5250, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.38955, "bpb": 2.0047, "chunk_loss": 0.030576, "F": 0.21295, "G": 0.21559, "chunks": 218.06, "bytes_per_chunk": 4.696, "boundary_entropy": 0.09674, "router_noise": 0.0, "grad_norm": 0.7384, "lr": 4.14e-05, "router_trainable": true, "sec_per_step": 0.109, "alpha": 0.03, "t": 642.4} |
| {"event": "train", "step": 5300, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.37691, "bpb": 1.98646, "chunk_loss": 0.030435, "F": 0.20419, "G": 0.2091, "chunks": 209.09, "bytes_per_chunk": 4.897, "boundary_entropy": 0.09747, "router_noise": 0.0, "grad_norm": 0.686, "lr": 3.99e-05, "router_trainable": true, "sec_per_step": 0.105, "alpha": 0.03, "t": 647.7} |
| {"event": "train", "step": 5350, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.33859, "bpb": 1.93117, "chunk_loss": 0.030678, "F": 0.2095, "G": 0.21288, "chunks": 214.53, "bytes_per_chunk": 4.773, "boundary_entropy": 0.10098, "router_noise": 0.0, "grad_norm": 0.75, "lr": 3.86e-05, "router_trainable": true, "sec_per_step": 0.109, "alpha": 0.03, "t": 653.1} |
| {"event": "probe", "step": 5400, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3326.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.191116, "cps": 25.984375, "mean_p": 0.196809, "boundary_entropy": 0.093443, "bpb": 1.883549}, "tr": {"chunks": 3665.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.202453, "cps": 28.632812, "mean_p": 0.196229, "boundary_entropy": 0.128251, "bpb": 2.027585}, "fi": {"chunks": 4859.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.258306, "cps": 37.960938, "mean_p": 0.249552, "boundary_entropy": 0.114008, "bpb": 2.11637}}, "bpb": {"en": 1.88355, "tr": 2.02759, "fi": 2.11637}, "gini_cpc": 0.068714, "gini_cps": 0.086245, "bpic": 4.787511, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.05931544121699, "fi": 1.351564857283866}, "t": 659.1} |
| {"event": "eval", "step": 5400, "phase": 3, "bpb": {"en": 1.89758, "tr": 2.03535, "fi": 2.12593}, "bytes_per_chunk": {"en": 4.934, "tr": 5.382, "fi": 4.137}, "bpb_mean": 2.01962, "frozen_inventory": false, "t": 660.0} |
| {"event": "train", "step": 5400, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.26921, "bpb": 1.83108, "chunk_loss": 0.03041, "F": 0.19958, "G": 0.20317, "chunks": 204.38, "bytes_per_chunk": 5.01, "boundary_entropy": 0.10031, "router_noise": 0.0, "grad_norm": 0.6971, "lr": 3.73e-05, "router_trainable": true, "sec_per_step": 0.145, "alpha": 0.03, "t": 660.4} |
| {"event": "train", "step": 5450, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.31166, "bpb": 1.89233, "chunk_loss": 0.030484, "F": 0.20837, "G": 0.21182, "chunks": 213.38, "bytes_per_chunk": 4.799, "boundary_entropy": 0.09696, "router_noise": 0.0, "grad_norm": 0.6844, "lr": 3.62e-05, "router_trainable": true, "sec_per_step": 0.112, "alpha": 0.03, "t": 666.0} |
| {"event": "train", "step": 5500, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.32647, "bpb": 1.91369, "chunk_loss": 0.030448, "F": 0.20538, "G": 0.20928, "chunks": 210.31, "bytes_per_chunk": 4.869, "boundary_entropy": 0.09713, "router_noise": 0.0, "grad_norm": 0.6967, "lr": 3.51e-05, "router_trainable": true, "sec_per_step": 0.103, "alpha": 0.03, "t": 671.1} |
| {"event": "train", "step": 5550, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.31479, "bpb": 1.89684, "chunk_loss": 0.030539, "F": 0.21201, "G": 0.2143, "chunks": 217.09, "bytes_per_chunk": 4.717, "boundary_entropy": 0.09305, "router_noise": 0.0, "grad_norm": 0.6651, "lr": 3.41e-05, "router_trainable": true, "sec_per_step": 0.109, "alpha": 0.03, "t": 676.6} |
| {"event": "probe", "step": 5600, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3332.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.191461, "cps": 26.03125, "mean_p": 0.196478, "boundary_entropy": 0.091834, "bpb": 1.877597}, "tr": {"chunks": 3669.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.202674, "cps": 28.664062, "mean_p": 0.196398, "boundary_entropy": 0.128294, "bpb": 2.019359}, "fi": {"chunks": 4853.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.257987, "cps": 37.914062, "mean_p": 0.249017, "boundary_entropy": 0.112809, "bpb": 2.108565}}, "bpb": {"en": 1.8776, "tr": 2.01936, "fi": 2.10857}, "gini_cpc": 0.06801, "gini_cps": 0.085541, "bpic": 4.785895, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0585619708856862, "fi": 1.3474651303294163}, "t": 682.4} |
| {"event": "eval", "step": 5600, "phase": 3, "bpb": {"en": 1.89282, "tr": 2.02714, "fi": 2.11818}, "bytes_per_chunk": {"en": 4.931, "tr": 5.384, "fi": 4.149}, "bpb_mean": 2.01271, "frozen_inventory": false, "t": 683.3} |
| {"event": "train", "step": 5600, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.34054, "bpb": 1.93399, "chunk_loss": 0.030563, "F": 0.21057, "G": 0.21492, "chunks": 215.62, "bytes_per_chunk": 4.749, "boundary_entropy": 0.10125, "router_noise": 0.0, "grad_norm": 0.7205, "lr": 3.33e-05, "router_trainable": true, "sec_per_step": 0.14, "alpha": 0.03, "t": 683.6} |
| {"event": "train", "step": 5650, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.32171, "bpb": 1.90682, "chunk_loss": 0.03046, "F": 0.20541, "G": 0.2091, "chunks": 210.34, "bytes_per_chunk": 4.868, "boundary_entropy": 0.09488, "router_noise": 0.0, "grad_norm": 0.6692, "lr": 3.25e-05, "router_trainable": true, "sec_per_step": 0.109, "alpha": 0.03, "t": 689.0} |
| {"event": "train", "step": 5700, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.40513, "bpb": 2.02717, "chunk_loss": 0.030531, "F": 0.21017, "G": 0.21401, "chunks": 215.22, "bytes_per_chunk": 4.758, "boundary_entropy": 0.09861, "router_noise": 0.0, "grad_norm": 0.7096, "lr": 3.18e-05, "router_trainable": true, "sec_per_step": 0.105, "alpha": 0.03, "t": 694.3} |
| {"event": "train", "step": 5750, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.38726, "bpb": 2.0014, "chunk_loss": 0.030545, "F": 0.20782, "G": 0.21167, "chunks": 212.81, "bytes_per_chunk": 4.812, "boundary_entropy": 0.09883, "router_noise": 0.0, "grad_norm": 0.7215, "lr": 3.13e-05, "router_trainable": true, "sec_per_step": 0.109, "alpha": 0.03, "t": 699.7} |
| {"event": "probe", "step": 5800, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3344.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.192151, "cps": 26.125, "mean_p": 0.196917, "boundary_entropy": 0.090878, "bpb": 1.872217}, "tr": {"chunks": 3668.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.202618, "cps": 28.65625, "mean_p": 0.195344, "boundary_entropy": 0.126706, "bpb": 2.014977}, "fi": {"chunks": 4876.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.25921, "cps": 38.09375, "mean_p": 0.249411, "boundary_entropy": 0.111595, "bpb": 2.10425}}, "bpb": {"en": 1.87222, "tr": 2.01498, "fi": 2.10425}, "gini_cpc": 0.06836, "gini_cps": 0.085913, "bpic": 4.772207, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0544758237485816, "fi": 1.3489929032157961}, "t": 705.7} |
| {"event": "eval", "step": 5800, "phase": 3, "bpb": {"en": 1.88754, "tr": 2.02333, "fi": 2.11319}, "bytes_per_chunk": {"en": 4.917, "tr": 5.389, "fi": 4.124}, "bpb_mean": 2.00802, "frozen_inventory": false, "t": 706.6} |
| {"event": "train", "step": 5800, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.30472, "bpb": 1.88232, "chunk_loss": 0.030428, "F": 0.20468, "G": 0.20785, "chunks": 209.59, "bytes_per_chunk": 4.886, "boundary_entropy": 0.09585, "router_noise": 0.0, "grad_norm": 0.712, "lr": 3.08e-05, "router_trainable": true, "sec_per_step": 0.145, "alpha": 0.03, "t": 707.0} |
| {"event": "train", "step": 5850, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.35413, "bpb": 1.95359, "chunk_loss": 0.030354, "F": 0.19867, "G": 0.20315, "chunks": 203.44, "bytes_per_chunk": 5.033, "boundary_entropy": 0.09462, "router_noise": 0.0, "grad_norm": 0.6961, "lr": 3.05e-05, "router_trainable": true, "sec_per_step": 0.108, "alpha": 0.03, "t": 712.4} |
| {"event": "train", "step": 5900, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.3885, "bpb": 2.00318, "chunk_loss": 0.030536, "F": 0.21109, "G": 0.21434, "chunks": 216.16, "bytes_per_chunk": 4.737, "boundary_entropy": 0.09804, "router_noise": 0.0, "grad_norm": 0.7437, "lr": 3.02e-05, "router_trainable": true, "sec_per_step": 0.096, "alpha": 0.03, "t": 717.2} |
| {"event": "train", "step": 5950, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.32479, "bpb": 1.91126, "chunk_loss": 0.030433, "F": 0.20361, "G": 0.2073, "chunks": 208.5, "bytes_per_chunk": 4.911, "boundary_entropy": 0.09893, "router_noise": 0.0, "grad_norm": 0.775, "lr": 3.01e-05, "router_trainable": true, "sec_per_step": 0.07, "alpha": 0.03, "t": 720.7} |
| {"event": "train", "step": 5999, "arch": "hnet_ffn", "phase": 3, "objective": "global", "lm_loss": 1.33727, "bpb": 1.92927, "chunk_loss": 0.030585, "F": 0.21146, "G": 0.2151, "chunks": 216.53, "bytes_per_chunk": 4.729, "boundary_entropy": 0.09875, "router_noise": 0.0, "grad_norm": 0.7399, "lr": 3e-05, "router_trainable": true, "sec_per_step": 0.064, "alpha": 0.03, "t": 723.9} |
| {"event": "probe", "step": 6000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3332.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.191461, "cps": 26.03125, "mean_p": 0.196037, "boundary_entropy": 0.088972, "bpb": 1.870555}, "tr": {"chunks": 3647.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.201458, "cps": 28.492188, "mean_p": 0.194312, "boundary_entropy": 0.126712, "bpb": 2.007867}, "fi": {"chunks": 4868.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.258785, "cps": 38.03125, "mean_p": 0.24836, "boundary_entropy": 0.112001, "bpb": 2.098563}}, "bpb": {"en": 1.87056, "tr": 2.00787, "fi": 2.09856}, "gini_cpc": 0.068869, "gini_cps": 0.086435, "bpic": 4.788723, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "tr": 1.0522146382720352, "fi": 1.3516299720675042}, "t": 724.2} |
| {"event": "eval", "step": 6000, "phase": 3, "bpb": {"en": 1.88578, "tr": 2.01641, "fi": 2.107}, "bytes_per_chunk": {"en": 4.929, "tr": 5.422, "fi": 4.137}, "bpb_mean": 2.00306, "frozen_inventory": false, "t": 724.7} |
| {"event": "reps_dumped", "dirs": ["/local/scratch/sas245/beetle-hnet/runs/ali_hnet-ffn_s0/probe/reps/th0.30", "/local/scratch/sas245/beetle-hnet/runs/ali_hnet-ffn_s0/probe/reps/th0.40", "/local/scratch/sas245/beetle-hnet/runs/ali_hnet-ffn_s0/probe/reps/th0.50", "/local/scratch/sas245/beetle-hnet/runs/ali_hnet-ffn_s0/probe/reps/th0.60", "/local/scratch/sas245/beetle-hnet/runs/ali_hnet-ffn_s0/probe/reps/th0.70"], "t": 725.7} |
| {"event": "final", "run": "ali_hnet-ffn_s0", "arch": "hnet_ffn", "objective": "ratio", "condition": "A", "step": 6000, "params": 22785033, "seed": 0, "per_lang": {"en": {"cpc": 0.19146124231454348, "cps": 26.03125, "bpb": 1.892164292916011, "bytes_per_chunk": 4.865194918478513}, "tr": {"cpc": 0.20145832182511186, "cps": 28.4921875, "bpb": 2.008611757265818, "bytes_per_chunk": 5.422079735250013}, "fi": {"cpc": 0.25878475360161607, "cps": 38.03125, "bpb": 2.127279023188143, "bytes_per_chunk": 4.119850069543215}}, "bpb_mean": 2.009351691123324, "gini_cpc": 0.06886917830098982, "gini_cps": 0.0864353844855239, "bpic": 4.788722883430404, "mean_boundary_entropy": 0.09874764084815979, "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": 726.6966066360474, "ffn_params": 268041, "matched_to_ali_params": 268096, "t": 726.7} |
|
|
