Blame for log.jsonl

{"event": "start", "run": "pilot_A_s0", "condition": "A", "params": 22516992, "langs": ["en", "de", "vi", "tr", "fi", "ru", "el", "hi", "ta", "th", "zh", "my"], "premium": {"en": 1.0, "de": 1.184, "vi": 1.39, "tr": 1.12, "fi": 1.104, "ru": 1.971, "el": 2.174, "hi": 2.552, "ta": 3.192, "th": 2.762, "zh": 0.915, "my": 3.524}, "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": {"my": 37511}, "weights": {"en": 0.2946, "de": 0.0909, "vi": 0.0576, "tr": 0.0612, "fi": 0.0424, "ru": 0.1476, "el": 0.0522, "hi": 0.042, "ta": 0.0313, "th": 0.0542, "zh": 0.1038, "my": 0.0222}, "t": 2.1}
{"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": 2.1}
{"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}, "de": {"chunks": 8628.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.420202, "cps": 67.40625, "mean_p": 0.498889, "boundary_entropy": 0.687821, "bpb": 8.156329}, "vi": {"chunks": 13115.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.711418, "cps": 102.460938, "mean_p": 0.505745, "boundary_entropy": 0.688385, "bpb": 8.087291}, "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}, "ru": {"chunks": 10744.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.566966, "cps": 83.9375, "mean_p": 0.493655, "boundary_entropy": 0.689893, "bpb": 8.099397}, "el": {"chunks": 15039.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.732788, "cps": 117.492188, "mean_p": 0.497482, "boundary_entropy": 0.690234, "bpb": 8.158044}, "hi": {"chunks": 2563.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.152333, "cps": 20.023438, "mean_p": 0.484638, "boundary_entropy": 0.690256, "bpb": 8.262162}, "ta": {"chunks": 38089.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.03739, "cps": 297.570312, "mean_p": 0.512976, "boundary_entropy": 0.690378, "bpb": 8.051803}, "th": {"chunks": 34006.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.114144, "cps": 265.671875, "mean_p": 0.511916, "boundary_entropy": 0.690552, "bpb": 8.248459}, "zh": {"chunks": 11821.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.001524, "cps": 92.351562, "mean_p": 0.52004, "boundary_entropy": 0.685674, "bpb": 8.123118}, "my": {"chunks": 48201.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.543857, "cps": 376.570312, "mean_p": 0.521563, "boundary_entropy": 0.690206, "bpb": 7.913994}}, "bpb": {"en": 8.16024, "de": 8.15633, "vi": 8.08729, "tr": 8.17143, "fi": 8.21476, "ru": 8.0994, "el": 8.15804, "hi": 8.26216, "ta": 8.0518, "th": 8.24846, "zh": 8.12312, "my": 7.91399}, "gini_cpc": 0.424496, "gini_cps": 0.425905, "bpic": 1.892645, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1454838515955792, "vi": 1.939350894526797, "tr": 0.7965927738475616, "fi": 1.2890367847879758, "ru": 1.5455676790921895, "el": 1.997603802482379, "hi": 0.41526447061189375, "ta": 5.553993192060118, "th": 5.7632270176529286, "zh": 5.456221801328583, "my": 6.934639901884223}, "t": 4.0}
{"event": "train", "step": 0, "phase": 1, "objective": "global", "lm_loss": 5.63916, "bpb": 8.13559, "chunk_loss": 0.05247, "F": 0.47223, "G": 0.4995, "chunks": 483.56, "bytes_per_chunk": 2.118, "boundary_entropy": 0.69158, "router_noise": 0.0, "grad_norm": 7.806, "lr": 1e-06, "router_trainable": true, "sec_per_step": 0.005, "alpha": 0.03, "t": 4.4}
{"event": "probe", "step": 1, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 6327.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.363558, "cps": 49.429688, "mean_p": 0.496549, "boundary_entropy": 0.686989, "bpb": 8.137928}, "de": {"chunks": 8593.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.418497, "cps": 67.132812, "mean_p": 0.49878, "boundary_entropy": 0.687819, "bpb": 8.136679}, "vi": {"chunks": 13021.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.70632, "cps": 101.726562, "mean_p": 0.505542, "boundary_entropy": 0.688387, "bpb": 8.072331}, "tr": {"chunks": 5241.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.28951, "cps": 40.945312, "mean_p": 0.489422, "boundary_entropy": 0.68728, "bpb": 8.156839}, "fi": {"chunks": 8839.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.469885, "cps": 69.054688, "mean_p": 0.500506, "boundary_entropy": 0.687509, "bpb": 8.196329}, "ru": {"chunks": 10507.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.554459, "cps": 82.085938, "mean_p": 0.493307, "boundary_entropy": 0.689882, "bpb": 8.05948}, "el": {"chunks": 15041.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.732885, "cps": 117.507812, "mean_p": 0.497467, "boundary_entropy": 0.690234, "bpb": 8.140841}, "hi": {"chunks": 2497.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.14841, "cps": 19.507812, "mean_p": 0.484437, "boundary_entropy": 0.690242, "bpb": 8.249675}, "ta": {"chunks": 37853.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.024766, "cps": 295.726562, "mean_p": 0.512694, "boundary_entropy": 0.690391, "bpb": 8.028648}, "th": {"chunks": 33858.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.104942, "cps": 264.515625, "mean_p": 0.51177, "boundary_entropy": 0.690558, "bpb": 8.229599}, "zh": {"chunks": 11820.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.001355, "cps": 92.34375, "mean_p": 0.520021, "boundary_entropy": 0.685675, "bpb": 8.120689}, "my": {"chunks": 48156.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.541482, "cps": 376.21875, "mean_p": 0.521508, "boundary_entropy": 0.69021, "bpb": 7.893441}}, "bpb": {"en": 8.13793, "de": 8.13668, "vi": 8.07233, "tr": 8.15684, "fi": 8.19633, "ru": 8.05948, "el": 8.14084, "hi": 8.24967, "ta": 8.02865, "th": 8.2296, "zh": 8.12069, "my": 7.89344}, "gini_cpc": 0.425879, "gini_cps": 0.427327, "bpic": 1.902232, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1511149395402474, "vi": 1.9427972617386349, "tr": 0.7963241633991176, "fi": 1.2924612650724754, "ru": 1.5250911597613446, "el": 2.0158682829037273, "hi": 0.40821574839874675, "ta": 5.569306520241896, "th": 5.78983945089012, "zh": 5.504911224161941, "my": 6.990581695324872}, "t": 5.8}
{"event": "probe", "step": 2, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 6247.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.358961, "cps": 48.804688, "mean_p": 0.4962, "boundary_entropy": 0.686976, "bpb": 8.088505}, "de": {"chunks": 8516.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.414747, "cps": 66.53125, "mean_p": 0.498576, "boundary_entropy": 0.687814, "bpb": 8.092769}, "vi": {"chunks": 12840.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.696501, "cps": 100.3125, "mean_p": 0.505085, "boundary_entropy": 0.68839, "bpb": 8.038661}, "tr": {"chunks": 5170.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.285588, "cps": 40.390625, "mean_p": 0.489116, "boundary_entropy": 0.687262, "bpb": 8.122011}, "fi": {"chunks": 8749.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.4651, "cps": 68.351562, "mean_p": 0.500139, "boundary_entropy": 0.687503, "bpb": 8.154273}, "ru": {"chunks": 10062.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.530976, "cps": 78.609375, "mean_p": 0.492759, "boundary_entropy": 0.689865, "bpb": 7.99337}, "el": {"chunks": 15092.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.73537, "cps": 117.90625, "mean_p": 0.497514, "boundary_entropy": 0.690235, "bpb": 8.120805}, "hi": {"chunks": 2406.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.143001, "cps": 18.796875, "mean_p": 0.484161, "boundary_entropy": 0.690223, "bpb": 8.217647}, "ta": {"chunks": 37477.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.004654, "cps": 292.789062, "mean_p": 0.512238, "boundary_entropy": 0.690413, "bpb": 7.987911}, "th": {"chunks": 33649.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.091949, "cps": 262.882812, "mean_p": 0.511606, "boundary_entropy": 0.690564, "bpb": 8.202675}, "zh": {"chunks": 11810.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.999661, "cps": 92.265625, "mean_p": 0.519977, "boundary_entropy": 0.685678, "bpb": 8.114164}, "my": {"chunks": 48089.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.537946, "cps": 375.695312, "mean_p": 0.521421, "boundary_entropy": 0.690216, "bpb": 7.860088}}, "bpb": {"en": 8.08851, "de": 8.09277, "vi": 8.03866, "tr": 8.12201, "fi": 8.15427, "ru": 7.99337, "el": 8.12081, "hi": 8.21765, "ta": 7.98791, "th": 8.20268, "zh": 8.11416, "my": 7.86009}, "gini_cpc": 0.428218, "gini_cps": 0.429766, "bpic": 1.917879, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.155409302548114, "vi": 1.9403250745060638, "tr": 0.7955960273956965, "fi": 1.295684153687481, "ru": 1.4792027751156966, "el": 2.048606590920987, "hi": 0.3983759978935506, "ta": 5.584598604655723, "th": 5.827787547534784, "zh": 5.570690999068649, "my": 7.070253481012241}, "t": 7.4}
{"event": "probe", "step": 4, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 5917.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.339999, "cps": 46.226562, "mean_p": 0.494974, "boundary_entropy": 0.686928, "bpb": 7.918787}, "de": {"chunks": 8278.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.403156, "cps": 64.671875, "mean_p": 0.497822, "boundary_entropy": 0.687798, "bpb": 7.939093}, "vi": {"chunks": 12271.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.665636, "cps": 95.867188, "mean_p": 0.503597, "boundary_entropy": 0.688396, "bpb": 7.926957}, "tr": {"chunks": 4897.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.270508, "cps": 38.257812, "mean_p": 0.488079, "boundary_entropy": 0.687201, "bpb": 8.003865}, "fi": {"chunks": 8332.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.442932, "cps": 65.09375, "mean_p": 0.498782, "boundary_entropy": 0.687477, "bpb": 8.007454}, "ru": {"chunks": 8697.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.458945, "cps": 67.945312, "mean_p": 0.490854, "boundary_entropy": 0.689794, "bpb": 7.770892}, "el": {"chunks": 15242.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.742679, "cps": 119.078125, "mean_p": 0.497599, "boundary_entropy": 0.690236, "bpb": 8.033874}, "hi": {"chunks": 2074.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.123269, "cps": 16.203125, "mean_p": 0.482793, "boundary_entropy": 0.690125, "bpb": 8.096572}, "ta": {"chunks": 35903.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 1.92046, "cps": 280.492188, "mean_p": 0.510439, "boundary_entropy": 0.690491, "bpb": 7.841623}, "th": {"chunks": 32617.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.02779, "cps": 254.820312, "mean_p": 0.510753, "boundary_entropy": 0.690595, "bpb": 8.087975}, "zh": {"chunks": 11778.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.994243, "cps": 92.015625, "mean_p": 0.51976, "boundary_entropy": 0.68569, "bpb": 8.087285}, "my": {"chunks": 47873.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.526546, "cps": 374.007812, "mean_p": 0.521124, "boundary_entropy": 0.690237, "bpb": 7.783276}}, "bpb": {"en": 7.91879, "de": 7.93909, "vi": 7.92696, "tr": 8.00386, "fi": 8.00745, "ru": 7.77089, "el": 8.03387, "hi": 8.09657, "ta": 7.84162, "th": 8.08797, "zh": 8.08729, "my": 7.78328}, "gini_cpc": 0.437248, "gini_cps": 0.438186, "bpic": 1.979487, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1857566414459753, "vi": 1.957759612804908, "tr": 0.7956134264887705, "fi": 1.3027465411227472, "ru": 1.3498415950106644, "el": 2.184357223223159, "hi": 0.36255694612521255, "ta": 5.648430904050564, "th": 5.964108001323247, "zh": 5.865440833207949, "my": 7.431043756769147}, "t": 9.0}
{"event": "probe", "step": 8, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 5083.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.292076, "cps": 39.710938, "mean_p": 0.491678, "boundary_entropy": 0.686769, "bpb": 7.437897}, "de": {"chunks": 7671.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.373594, "cps": 59.929688, "mean_p": 0.495825, "boundary_entropy": 0.687749, "bpb": 7.496186}, "vi": {"chunks": 10393.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.563765, "cps": 81.195312, "mean_p": 0.498906, "boundary_entropy": 0.688358, "bpb": 7.586013}, "tr": {"chunks": 4096.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.226261, "cps": 32.0, "mean_p": 0.485081, "boundary_entropy": 0.687, "bpb": 7.643328}, "fi": {"chunks": 7125.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.378768, "cps": 55.664062, "mean_p": 0.49486, "boundary_entropy": 0.687364, "bpb": 7.577129}, "ru": {"chunks": 5216.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.275251, "cps": 40.75, "mean_p": 0.485331, "boundary_entropy": 0.689504, "bpb": 7.189078}, "el": {"chunks": 14855.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.723822, "cps": 116.054688, "mean_p": 0.497135, "boundary_entropy": 0.69023, "bpb": 7.711115}, "hi": {"chunks": 1136.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.067519, "cps": 8.875, "mean_p": 0.477099, "boundary_entropy": 0.689635, "bpb": 7.696376}, "ta": {"chunks": 28684.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 1.534314, "cps": 224.09375, "mean_p": 0.503574, "boundary_entropy": 0.690676, "bpb": 7.366098}, "th": {"chunks": 27419.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 1.704632, "cps": 214.210938, "mean_p": 0.506232, "boundary_entropy": 0.690718, "bpb": 7.547779}, "zh": {"chunks": 11597.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.963596, "cps": 90.601562, "mean_p": 0.51892, "boundary_entropy": 0.685739, "bpb": 7.998454}, "my": {"chunks": 47100.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.48575, "cps": 367.96875, "mean_p": 0.520007, "boundary_entropy": 0.69031, "bpb": 7.472424}}, "bpb": {"en": 7.4379, "de": 7.49619, "vi": 7.58601, "tr": 7.64333, "fi": 7.57713, "ru": 7.18908, "el": 7.71112, "hi": 7.69638, "ta": 7.3661, "th": 7.54778, "zh": 7.99845, "my": 7.47242}, "gini_cpc": 0.469726, "gini_cps": 0.463478, "bpic": 2.252566, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.2790973114214237, "vi": 1.930197709078924, "tr": 0.7746640589177181, "fi": 1.2968119258060078, "ru": 0.942393710318995, "el": 2.4781969690507837, "hi": 0.23116776224918947, "ta": 5.253131286227654, "th": 5.836259002027335, "zh": 6.722893399970247, "my": 8.51062670000973}, "t": 10.6}
{"event": "probe", "step": 16, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3942.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.226513, "cps": 30.796875, "mean_p": 0.488102, "boundary_entropy": 0.686663, "bpb": 6.812392}, "de": {"chunks": 6926.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.337311, "cps": 54.109375, "mean_p": 0.493888, "boundary_entropy": 0.687798, "bpb": 6.904844}, "vi": {"chunks": 7193.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.390182, "cps": 56.195312, "mean_p": 0.490787, "boundary_entropy": 0.688162, "bpb": 7.158148}, "tr": {"chunks": 3500.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.193338, "cps": 27.34375, "mean_p": 0.483504, "boundary_entropy": 0.686946, "bpb": 7.13112}, "fi": {"chunks": 5530.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.293977, "cps": 43.203125, "mean_p": 0.490067, "boundary_entropy": 0.687262, "bpb": 7.002209}, "ru": {"chunks": 2614.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.137942, "cps": 20.421875, "mean_p": 0.478838, "boundary_entropy": 0.689011, "bpb": 6.360409}, "el": {"chunks": 5607.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.273206, "cps": 43.804688, "mean_p": 0.487003, "boundary_entropy": 0.68987, "bpb": 7.041139}, "hi": {"chunks": 404.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.024012, "cps": 3.15625, "mean_p": 0.467029, "boundary_entropy": 0.688451, "bpb": 7.051966}, "ta": {"chunks": 14612.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.781599, "cps": 114.15625, "mean_p": 0.490194, "boundary_entropy": 0.690533, "bpb": 6.622494}, "th": {"chunks": 17459.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 1.085421, "cps": 136.398438, "mean_p": 0.49658, "boundary_entropy": 0.690724, "bpb": 6.369788}, "zh": {"chunks": 10871.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.840671, "cps": 84.929688, "mean_p": 0.515362, "boundary_entropy": 0.685938, "bpb": 7.770785}, "my": {"chunks": 42529.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.244511, "cps": 332.257812, "mean_p": 0.515707, "boundary_entropy": 0.690552, "bpb": 6.672834}}, "bpb": {"en": 6.81239, "de": 6.90484, "vi": 7.15815, "tr": 7.13112, "fi": 7.00221, "ru": 6.36041, "el": 7.04114, "hi": 7.05197, "ta": 6.62249, "th": 6.36979, "zh": 7.77079, "my": 6.67283}, "gini_cpc": 0.523336, "gini_cps": 0.492157, "bpic": 3.16685, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.48914703220068, "vi": 1.7225602398323288, "tr": 0.8535421906323118, "fi": 1.297838783567405, "ru": 0.6089811769336112, "el": 1.2061385855762228, "hi": 0.10600681652158377, "ta": 3.450576770500419, "th": 4.791878523956398, "zh": 8.126126025129361, "my": 9.908987836292486}, "t": 12.5}
{"event": "probe", "step": 32, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 4866.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.279607, "cps": 38.015625, "mean_p": 0.495231, "boundary_entropy": 0.687434, "bpb": 6.360834}, "de": {"chunks": 8530.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.415429, "cps": 66.640625, "mean_p": 0.498977, "boundary_entropy": 0.68833, "bpb": 6.444338}, "vi": {"chunks": 5661.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.307079, "cps": 44.226562, "mean_p": 0.48903, "boundary_entropy": 0.68836, "bpb": 6.904339}, "tr": {"chunks": 6804.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.375849, "cps": 53.15625, "mean_p": 0.496489, "boundary_entropy": 0.687979, "bpb": 6.718712}, "fi": {"chunks": 7158.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.380522, "cps": 55.921875, "mean_p": 0.497706, "boundary_entropy": 0.688011, "bpb": 6.521115}, "ru": {"chunks": 5328.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.281161, "cps": 41.625, "mean_p": 0.487948, "boundary_entropy": 0.689827, "bpb": 5.790951}, "el": {"chunks": 487.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.023729, "cps": 3.804688, "mean_p": 0.469293, "boundary_entropy": 0.688351, "bpb": 6.568223}, "hi": {"chunks": 2682.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.159406, "cps": 20.953125, "mean_p": 0.485683, "boundary_entropy": 0.690385, "bpb": 5.948013}, "ta": {"chunks": 5469.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.292538, "cps": 42.726562, "mean_p": 0.482366, "boundary_entropy": 0.690233, "bpb": 5.810306}, "th": {"chunks": 3733.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.23208, "cps": 29.164062, "mean_p": 0.48789, "boundary_entropy": 0.690626, "bpb": 5.623473}, "zh": {"chunks": 7414.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.255334, "cps": 57.921875, "mean_p": 0.501058, "boundary_entropy": 0.686375, "bpb": 7.331258}, "my": {"chunks": 25416.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 1.341355, "cps": 198.5625, "mean_p": 0.496463, "boundary_entropy": 0.690915, "bpb": 5.630036}}, "bpb": {"en": 6.36083, "de": 6.44434, "vi": 6.90434, "tr": 6.71871, "fi": 6.52112, "ru": 5.79095, "el": 6.56822, "hi": 5.94801, "ta": 5.81031, "th": 5.62347, "zh": 7.33126, "my": 5.63004}, "gini_cpc": 0.412354, "gini_cps": 0.367751, "bpic": 4.593539, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.4857599234410361, "vi": 1.098252065025348, "tr": 1.3442058128398444, "fi": 1.3609175862615734, "ru": 1.005557749805608, "el": 0.08486725037915717, "hi": 0.5701061371960212, "ta": 1.0462475872505919, "th": 0.8300207321845605, "zh": 4.489636236926814, "my": 4.797288549074668}, "t": 14.9}
{"event": "train", "step": 50, "phase": 1, "objective": "global", "lm_loss": 4.22409, "bpb": 6.09408, "chunk_loss": 0.031114, "F": 0.18005, "G": 0.48757, "chunks": 184.38, "bytes_per_chunk": 5.554, "boundary_entropy": 0.69168, "router_noise": 0.0, "grad_norm": 1.5888, "lr": 5.1e-05, "router_trainable": true, "sec_per_step": 0.234, "alpha": 0.03, "t": 16.1}
{"event": "probe", "step": 64, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 4806.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.276159, "cps": 37.546875, "mean_p": 0.496139, "boundary_entropy": 0.687593, "bpb": 5.819777}, "de": {"chunks": 6281.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.305898, "cps": 49.070312, "mean_p": 0.495973, "boundary_entropy": 0.688417, "bpb": 5.934343}, "vi": {"chunks": 2853.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.15476, "cps": 22.289062, "mean_p": 0.480814, "boundary_entropy": 0.687896, "bpb": 6.313285}, "tr": {"chunks": 5113.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.282439, "cps": 39.945312, "mean_p": 0.494028, "boundary_entropy": 0.688062, "bpb": 6.226922}, "fi": {"chunks": 5487.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.291691, "cps": 42.867188, "mean_p": 0.495207, "boundary_entropy": 0.688105, "bpb": 5.973499}, "ru": {"chunks": 7561.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.398997, "cps": 59.070312, "mean_p": 0.493533, "boundary_entropy": 0.6902, "bpb": 5.415588}, "el": {"chunks": 4212.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.205233, "cps": 32.90625, "mean_p": 0.490332, "boundary_entropy": 0.690335, "bpb": 5.661535}, "hi": {"chunks": 439.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.026092, "cps": 3.429688, "mean_p": 0.473894, "boundary_entropy": 0.689313, "bpb": 5.434706}, "ta": {"chunks": 2763.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.147794, "cps": 21.585938, "mean_p": 0.483543, "boundary_entropy": 0.690385, "bpb": 5.205846}, "th": {"chunks": 1456.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.090519, "cps": 11.375, "mean_p": 0.483903, "boundary_entropy": 0.690415, "bpb": 5.101274}, "zh": {"chunks": 4674.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.791399, "cps": 36.515625, "mean_p": 0.493146, "boundary_entropy": 0.68665, "bpb": 6.9934}, "my": {"chunks": 21773.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 1.149092, "cps": 170.101562, "mean_p": 0.493945, "boundary_entropy": 0.690998, "bpb": 4.994507}}, "bpb": {"en": 5.81978, "de": 5.93434, "vi": 6.31328, "tr": 6.22692, "fi": 5.9735, "ru": 5.41559, "el": 5.66153, "hi": 5.43471, "ta": 5.20585, "th": 5.10127, "zh": 6.9934, "my": 4.99451}, "gini_cpc": 0.433663, "gini_cps": 0.408212, "bpic": 5.69256, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.107686186841036, "vi": 0.5604011056155396, "tr": 1.0227408318180413, "fi": 1.0562419948324921, "ru": 1.4448087977099866, "el": 0.7431694886581074, "hi": 0.09448215733305138, "ta": 0.5351749401239893, "th": 0.3277786509480495, "zh": 2.865732302958586, "my": 4.160976377427731}, "t": 17.9}
{"event": "train", "step": 100, "phase": 1, "objective": "global", "lm_loss": 3.38504, "bpb": 4.88359, "chunk_loss": 0.030258, "F": 0.16992, "G": 0.47943, "chunks": 174.0, "bytes_per_chunk": 5.885, "boundary_entropy": 0.69026, "router_noise": 0.0, "grad_norm": 2.2365, "lr": 0.000101, "router_trainable": true, "sec_per_step": 0.078, "alpha": 0.03, "t": 20.0}
{"event": "probe", "step": 128, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2909.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.167155, "cps": 22.726562, "mean_p": 0.448266, "boundary_entropy": 0.675922, "bpb": 4.501104}, "de": {"chunks": 3373.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.164272, "cps": 26.351562, "mean_p": 0.450018, "boundary_entropy": 0.678005, "bpb": 4.715912}, "vi": {"chunks": 6129.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.332465, "cps": 47.882812, "mean_p": 0.466084, "boundary_entropy": 0.682901, "bpb": 5.018291}, "tr": {"chunks": 3384.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.18693, "cps": 26.4375, "mean_p": 0.456446, "boundary_entropy": 0.679927, "bpb": 5.020652}, "fi": {"chunks": 4075.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.216629, "cps": 31.835938, "mean_p": 0.454344, "boundary_entropy": 0.678314, "bpb": 4.742163}, "ru": {"chunks": 6065.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.320053, "cps": 47.382812, "mean_p": 0.46868, "boundary_entropy": 0.685862, "bpb": 3.673689}, "el": {"chunks": 6610.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.322078, "cps": 51.640625, "mean_p": 0.476342, "boundary_entropy": 0.687927, "bpb": 4.584804}, "hi": {"chunks": 9121.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.54211, "cps": 71.257812, "mean_p": 0.489859, "boundary_entropy": 0.689653, "bpb": 3.633187}, "ta": {"chunks": 5023.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.268681, "cps": 39.242188, "mean_p": 0.481529, "boundary_entropy": 0.689305, "bpb": 3.448737}, "th": {"chunks": 5952.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.370034, "cps": 46.5, "mean_p": 0.464787, "boundary_entropy": 0.686308, "bpb": 3.183711}, "zh": {"chunks": 3153.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.533864, "cps": 24.632812, "mean_p": 0.482193, "boundary_entropy": 0.685337, "bpb": 6.312001}, "my": {"chunks": 15539.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.820087, "cps": 121.398438, "mean_p": 0.472726, "boundary_entropy": 0.688472, "bpb": 3.984774}}, "bpb": {"en": 4.5011, "de": 4.71591, "vi": 5.01829, "tr": 5.02065, "fi": 4.74216, "ru": 3.67369, "el": 4.5848, "hi": 3.63319, "ta": 3.44874, "th": 3.18371, "zh": 6.312, "my": 3.98477}, "gini_cpc": 0.274433, "gini_cps": 0.275615, "bpic": 5.380133, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9827528975692266, "vi": 1.9889637977142847, "tr": 1.1183048332327445, "fi": 1.2959735220714006, "ru": 1.9147055222124767, "el": 1.926819413150514, "hi": 3.243155570379776, "ta": 1.6073783246486013, "th": 2.213717795404762, "zh": 3.193823610111757, "my": 4.906141724346555}, "t": 22.3}
{"event": "train", "step": 150, "phase": 1, "objective": "global", "lm_loss": 2.89006, "bpb": 4.16948, "chunk_loss": 0.029924, "F": 0.16434, "G": 0.43104, "chunks": 168.28, "bytes_per_chunk": 6.085, "boundary_entropy": 0.67096, "router_noise": 0.0, "grad_norm": 1.6043, "lr": 0.000151, "router_trainable": true, "sec_per_step": 0.069, "alpha": 0.03, "t": 23.4}
{"event": "probe", "step": 200, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2565.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.147388, "cps": 20.039062, "mean_p": 0.368861, "boundary_entropy": 0.621222, "bpb": 3.879065}, "de": {"chunks": 2949.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.143622, "cps": 23.039062, "mean_p": 0.373375, "boundary_entropy": 0.628984, "bpb": 4.112746}, "vi": {"chunks": 4356.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.23629, "cps": 34.03125, "mean_p": 0.395495, "boundary_entropy": 0.619299, "bpb": 3.657055}, "tr": {"chunks": 3031.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.167431, "cps": 23.679688, "mean_p": 0.394016, "boundary_entropy": 0.641785, "bpb": 4.169187}, "fi": {"chunks": 3309.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.175908, "cps": 25.851562, "mean_p": 0.370006, "boundary_entropy": 0.624066, "bpb": 4.136842}, "ru": {"chunks": 4647.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.245224, "cps": 36.304688, "mean_p": 0.46035, "boundary_entropy": 0.678514, "bpb": 2.824376}, "el": {"chunks": 4454.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.217025, "cps": 34.796875, "mean_p": 0.403313, "boundary_entropy": 0.658573, "bpb": 3.089}, "hi": {"chunks": 10222.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.607548, "cps": 79.859375, "mean_p": 0.402182, "boundary_entropy": 0.64719, "bpb": 2.404098}, "ta": {"chunks": 8233.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.440385, "cps": 64.320312, "mean_p": 0.448582, "boundary_entropy": 0.677123, "bpb": 2.317422}, "th": {"chunks": 5096.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.316817, "cps": 39.8125, "mean_p": 0.445515, "boundary_entropy": 0.67846, "bpb": 2.319628}, "zh": {"chunks": 4072.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.689468, "cps": 31.8125, "mean_p": 0.475488, "boundary_entropy": 0.68239, "bpb": 5.31779}, "my": {"chunks": 4103.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.21654, "cps": 32.054688, "mean_p": 0.437671, "boundary_entropy": 0.679628, "bpb": 2.455701}}, "bpb": {"en": 3.87906, "de": 4.11275, "vi": 3.65706, "tr": 4.16919, "fi": 4.13684, "ru": 2.82438, "el": 3.089, "hi": 2.4041, "ta": 2.31742, "th": 2.31963, "zh": 5.31779, "my": 2.4557}, "gini_cpc": 0.297529, "gini_cps": 0.225478, "bpic": 6.728632, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9744490042131579, "vi": 1.603177023111073, "tr": 1.1359837938398631, "fi": 1.193497832082199, "ru": 1.663796509744017, "el": 1.4724688575049016, "hi": 4.122090804028397, "ta": 2.987922967801066, "th": 2.1495378899014193, "zh": 4.677901549222418, "my": 1.469179607593327}, "t": 26.7}
{"event": "eval", "step": 200, "phase": 1, "bpb": {"en": 3.88118, "de": 4.1122, "vi": 3.6621, "tr": 4.17842, "fi": 4.13904, "ru": 2.80823, "el": 3.08245, "hi": 2.39977, "ta": 2.33528, "th": 2.35215, "zh": 5.32702, "my": 2.47344}, "bytes_per_chunk": {"en": 6.026, "de": 6.836, "vi": 5.527, "tr": 6.707, "fi": 6.124, "ru": 7.001, "el": 7.6, "hi": 4.232, "ta": 6.03, "th": 8.306, "zh": 4.807, "my": 13.623}, "bpb_mean": 3.39594, "t": 28.0}
{"event": "train", "step": 200, "phase": 1, "objective": "global", "lm_loss": 2.4937, "bpb": 3.59766, "chunk_loss": 0.029611, "F": 0.16003, "G": 0.40296, "chunks": 163.88, "bytes_per_chunk": 6.249, "boundary_entropy": 0.64846, "router_noise": 0.0, "grad_norm": 1.0712, "lr": 0.000201, "router_trainable": true, "sec_per_step": 0.093, "alpha": 0.03, "t": 28.1}
{"event": "train", "step": 250, "phase": 1, "objective": "global", "lm_loss": 2.56235, "bpb": 3.69669, "chunk_loss": 0.031705, "F": 0.20102, "G": 0.4053, "chunks": 205.84, "bytes_per_chunk": 4.975, "boundary_entropy": 0.6419, "router_noise": 0.0, "grad_norm": 1.1351, "lr": 0.000251, "router_trainable": true, "sec_per_step": 0.044, "alpha": 0.03, "t": 30.3}
{"event": "probe", "step": 256, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2890.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.166063, "cps": 22.578125, "mean_p": 0.370095, "boundary_entropy": 0.614416, "bpb": 3.630289}, "de": {"chunks": 3658.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.178152, "cps": 28.578125, "mean_p": 0.379849, "boundary_entropy": 0.626269, "bpb": 3.921367}, "vi": {"chunks": 4660.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.25278, "cps": 36.40625, "mean_p": 0.428134, "boundary_entropy": 0.641246, "bpb": 3.216039}, "tr": {"chunks": 4374.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.241617, "cps": 34.171875, "mean_p": 0.396809, "boundary_entropy": 0.628826, "bpb": 3.862017}, "fi": {"chunks": 3163.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.168146, "cps": 24.710938, "mean_p": 0.355833, "boundary_entropy": 0.60503, "bpb": 3.903512}, "ru": {"chunks": 4309.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.227388, "cps": 33.664062, "mean_p": 0.456131, "boundary_entropy": 0.675087, "bpb": 2.669845}, "el": {"chunks": 6515.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.317449, "cps": 50.898438, "mean_p": 0.393302, "boundary_entropy": 0.647507, "bpb": 2.763578}, "hi": {"chunks": 8661.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.51477, "cps": 67.664062, "mean_p": 0.359033, "boundary_entropy": 0.602755, "bpb": 2.206703}, "ta": {"chunks": 9692.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.518427, "cps": 75.71875, "mean_p": 0.407416, "boundary_entropy": 0.64645, "bpb": 1.953941}, "th": {"chunks": 4491.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.279204, "cps": 35.085938, "mean_p": 0.425473, "boundary_entropy": 0.668883, "bpb": 2.143762}, "zh": {"chunks": 3437.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.581951, "cps": 26.851562, "mean_p": 0.456989, "boundary_entropy": 0.673831, "bpb": 4.971251}, "my": {"chunks": 12547.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.662181, "cps": 98.023438, "mean_p": 0.421955, "boundary_entropy": 0.66514, "bpb": 2.167402}}, "bpb": {"en": 3.63029, "de": 3.92137, "vi": 3.21604, "tr": 3.86202, "fi": 3.90351, "ru": 2.66984, "el": 2.76358, "hi": 2.2067, "ta": 1.95394, "th": 2.14376, "zh": 4.97125, "my": 2.1674}, "gini_cpc": 0.268863, "gini_cps": 0.264401, "bpic": 5.611079, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0727970519900702, "vi": 1.522190657720993, "tr": 1.4549715612335183, "fi": 1.0125432571938051, "ru": 1.3692840748281307, "el": 1.9116124587706225, "hi": 3.0998397507416566, "ta": 3.121865680274596, "th": 1.681311824186604, "zh": 3.5043894719697404, "my": 3.987519311640016}, "t": 31.5}
{"event": "train", "step": 300, "phase": 1, "objective": "global", "lm_loss": 2.34135, "bpb": 3.37785, "chunk_loss": 0.031617, "F": 0.20248, "G": 0.37581, "chunks": 207.34, "bytes_per_chunk": 4.939, "boundary_entropy": 0.61094, "router_noise": 0.0, "grad_norm": 0.9872, "lr": 0.0003, "router_trainable": true, "sec_per_step": 0.066, "alpha": 0.03, "t": 33.6}
{"event": "train", "step": 350, "phase": 1, "objective": "global", "lm_loss": 2.05284, "bpb": 2.96162, "chunk_loss": 0.032258, "F": 0.20203, "G": 0.37045, "chunks": 206.88, "bytes_per_chunk": 4.95, "boundary_entropy": 0.58926, "router_noise": 0.0, "grad_norm": 0.7414, "lr": 0.0002999, "router_trainable": true, "sec_per_step": 0.05, "alpha": 0.03, "t": 36.1}
{"event": "probe", "step": 400, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3366.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.193415, "cps": 26.296875, "mean_p": 0.312937, "boundary_entropy": 0.532857, "bpb": 3.244222}, "de": {"chunks": 4708.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.229289, "cps": 36.78125, "mean_p": 0.336359, "boundary_entropy": 0.549422, "bpb": 3.527998}, "vi": {"chunks": 5519.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.299376, "cps": 43.117188, "mean_p": 0.371721, "boundary_entropy": 0.589086, "bpb": 2.804575}, "tr": {"chunks": 4560.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.251892, "cps": 35.625, "mean_p": 0.34433, "boundary_entropy": 0.557544, "bpb": 3.435827}, "fi": {"chunks": 3470.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.184467, "cps": 27.109375, "mean_p": 0.325093, "boundary_entropy": 0.560538, "bpb": 3.563839}, "ru": {"chunks": 4859.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.256412, "cps": 37.960938, "mean_p": 0.303219, "boundary_entropy": 0.516438, "bpb": 2.519003}, "el": {"chunks": 6280.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.305998, "cps": 49.0625, "mean_p": 0.299498, "boundary_entropy": 0.532789, "bpb": 2.552635}, "hi": {"chunks": 5485.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.326003, "cps": 42.851562, "mean_p": 0.327225, "boundary_entropy": 0.576602, "bpb": 1.806191}, "ta": {"chunks": 5881.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.314576, "cps": 45.945312, "mean_p": 0.336888, "boundary_entropy": 0.589664, "bpb": 1.571369}, "th": {"chunks": 7951.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.494311, "cps": 62.117188, "mean_p": 0.340891, "boundary_entropy": 0.587295, "bpb": 1.886972}, "zh": {"chunks": 5364.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.908229, "cps": 41.90625, "mean_p": 0.345422, "boundary_entropy": 0.510662, "bpb": 4.15109}, "my": {"chunks": 14380.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.758919, "cps": 112.34375, "mean_p": 0.387314, "boundary_entropy": 0.609253, "bpb": 1.873974}}, "bpb": {"en": 3.24422, "de": 3.528, "vi": 2.80458, "tr": 3.43583, "fi": 3.56384, "ru": 2.519, "el": 2.55264, "hi": 1.80619, "ta": 1.57137, "th": 1.88697, "zh": 4.15109, "my": 1.87397}, "gini_cpc": 0.284806, "gini_cps": 0.210574, "bpic": 5.343428, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1854795198164982, "vi": 1.547844258891021, "tr": 1.302339760508865, "fi": 0.9537347349860055, "ru": 1.325707439128206, "el": 1.5820813359872838, "hi": 1.6855109084022641, "ta": 1.6264312810149226, "th": 2.555704847853178, "zh": 4.695753978098952, "my": 3.9237878543284053}, "t": 39.4}
{"event": "eval", "step": 400, "phase": 1, "bpb": {"en": 3.25281, "de": 3.52617, "vi": 2.80917, "tr": 3.44226, "fi": 3.55831, "ru": 2.50274, "el": 2.54567, "hi": 1.80343, "ta": 1.58703, "th": 1.91502, "zh": 4.16006, "my": 1.88981}, "bytes_per_chunk": {"en": 5.137, "de": 4.509, "vi": 4.577, "tr": 4.377, "fi": 5.885, "ru": 6.944, "el": 5.681, "hi": 6.998, "ta": 7.072, "th": 4.877, "zh": 2.92, "my": 3.527}, "bpb_mean": 2.74937, "t": 40.8}
{"event": "train", "step": 400, "phase": 1, "objective": "global", "lm_loss": 2.05453, "bpb": 2.96405, "chunk_loss": 0.03073, "F": 0.18503, "G": 0.32236, "chunks": 189.47, "bytes_per_chunk": 5.405, "boundary_entropy": 0.54602, "router_noise": 0.0, "grad_norm": 0.8167, "lr": 0.0002998, "router_trainable": true, "sec_per_step": 0.096, "alpha": 0.03, "t": 40.8}
{"event": "train", "step": 450, "phase": 1, "objective": "global", "lm_loss": 2.12134, "bpb": 3.06044, "chunk_loss": 0.030744, "F": 0.18896, "G": 0.31116, "chunks": 193.5, "bytes_per_chunk": 5.292, "boundary_entropy": 0.53081, "router_noise": 0.0, "grad_norm": 0.7265, "lr": 0.0002995, "router_trainable": true, "sec_per_step": 0.045, "alpha": 0.03, "t": 43.1}
{"event": "train", "step": 500, "phase": 1, "objective": "global", "lm_loss": 1.95775, "bpb": 2.82443, "chunk_loss": 0.030355, "F": 0.1785, "G": 0.30971, "chunks": 182.78, "bytes_per_chunk": 5.602, "boundary_entropy": 0.52682, "router_noise": 0.0, "grad_norm": 0.9696, "lr": 0.0002992, "router_trainable": true, "sec_per_step": 0.043, "alpha": 0.03, "t": 45.2}
{"event": "train", "step": 550, "phase": 1, "objective": "global", "lm_loss": 1.91599, "bpb": 2.76419, "chunk_loss": 0.030678, "F": 0.18881, "G": 0.31126, "chunks": 193.34, "bytes_per_chunk": 5.296, "boundary_entropy": 0.5207, "router_noise": 0.0, "grad_norm": 1.0468, "lr": 0.0002987, "router_trainable": true, "sec_per_step": 0.042, "alpha": 0.03, "t": 47.3}
{"event": "probe", "step": 600, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3173.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.182325, "cps": 24.789062, "mean_p": 0.288596, "boundary_entropy": 0.479348, "bpb": 2.938514}, "de": {"chunks": 4007.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.195149, "cps": 31.304688, "mean_p": 0.299862, "boundary_entropy": 0.477375, "bpb": 3.268704}, "vi": {"chunks": 5307.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.287876, "cps": 41.460938, "mean_p": 0.303171, "boundary_entropy": 0.510299, "bpb": 2.581465}, "tr": {"chunks": 4448.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.245705, "cps": 34.75, "mean_p": 0.32634, "boundary_entropy": 0.503402, "bpb": 3.165682}, "fi": {"chunks": 3766.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.200202, "cps": 29.421875, "mean_p": 0.302541, "boundary_entropy": 0.501444, "bpb": 3.372927}, "ru": {"chunks": 7070.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.373087, "cps": 55.234375, "mean_p": 0.338125, "boundary_entropy": 0.566982, "bpb": 2.224927}, "el": {"chunks": 6746.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.328704, "cps": 52.703125, "mean_p": 0.336042, "boundary_entropy": 0.57615, "bpb": 2.259386}, "hi": {"chunks": 4784.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.284339, "cps": 37.375, "mean_p": 0.344721, "boundary_entropy": 0.592129, "bpb": 1.629114}, "ta": {"chunks": 4786.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.256004, "cps": 37.390625, "mean_p": 0.334333, "boundary_entropy": 0.586135, "bpb": 1.405873}, "th": {"chunks": 4407.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.273982, "cps": 34.429688, "mean_p": 0.285221, "boundary_entropy": 0.528857, "bpb": 1.736045}, "zh": {"chunks": 3328.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.563495, "cps": 26.0, "mean_p": 0.297825, "boundary_entropy": 0.507996, "bpb": 3.864208}, "my": {"chunks": 9831.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.518841, "cps": 76.804688, "mean_p": 0.368028, "boundary_entropy": 0.588074, "bpb": 1.659487}}, "bpb": {"en": 2.93851, "de": 3.2687, "vi": 2.58147, "tr": 3.16568, "fi": 3.37293, "ru": 2.22493, "el": 2.25939, "hi": 1.62911, "ta": 1.40587, "th": 1.73604, "zh": 3.86421, "my": 1.65949}, "gini_cpc": 0.198516, "gini_cps": 0.183485, "bpic": 6.224855, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0703380961050162, "vi": 1.5789195195323027, "tr": 1.3476225678467433, "fi": 1.098050920478012, "ru": 2.0462761742173203, "el": 1.802849774028444, "hi": 1.5595171181060108, "ta": 1.4041104542228164, "th": 1.5027129648560804, "zh": 3.0906079903571757, "my": 2.845695113556095}, "t": 50.4}
{"event": "eval", "step": 600, "phase": 1, "bpb": {"en": 2.94749, "de": 3.26591, "vi": 2.58745, "tr": 3.17705, "fi": 3.36282, "ru": 2.20917, "el": 2.25152, "hi": 1.62371, "ta": 1.42, "th": 1.76568, "zh": 3.87601, "my": 1.67068}, "bytes_per_chunk": {"en": 5.413, "de": 5.076, "vi": 4.767, "tr": 4.196, "fi": 5.35, "ru": 4.808, "el": 5.639, "hi": 7.332, "ta": 8.076, "th": 8.516, "zh": 4.91, "my": 6.311}, "bpb_mean": 2.51312, "t": 51.7}
{"event": "train", "step": 600, "phase": 1, "objective": "global", "lm_loss": 1.94197, "bpb": 2.80167, "chunk_loss": 0.030665, "F": 0.18719, "G": 0.30801, "chunks": 191.69, "bytes_per_chunk": 5.342, "boundary_entropy": 0.51255, "router_noise": 0.0, "grad_norm": 0.8234, "lr": 0.0002982, "router_trainable": true, "sec_per_step": 0.088, "alpha": 0.03, "t": 51.7}
{"event": "train", "step": 650, "phase": 1, "objective": "global", "lm_loss": 1.73706, "bpb": 2.50605, "chunk_loss": 0.030314, "F": 0.17776, "G": 0.32092, "chunks": 182.03, "bytes_per_chunk": 5.625, "boundary_entropy": 0.52876, "router_noise": 0.0, "grad_norm": 0.9558, "lr": 0.0002975, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 53.8}
{"event": "train", "step": 700, "phase": 1, "objective": "global", "lm_loss": 1.75603, "bpb": 2.53342, "chunk_loss": 0.03111, "F": 0.19583, "G": 0.32215, "chunks": 200.53, "bytes_per_chunk": 5.106, "boundary_entropy": 0.52325, "router_noise": 0.0, "grad_norm": 0.7076, "lr": 0.0002967, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 55.8}
{"event": "train", "step": 750, "phase": 1, "objective": "global", "lm_loss": 1.69278, "bpb": 2.44217, "chunk_loss": 0.030064, "F": 0.17108, "G": 0.3174, "chunks": 175.19, "bytes_per_chunk": 5.845, "boundary_entropy": 0.52115, "router_noise": 0.0, "grad_norm": 0.655, "lr": 0.0002959, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 57.9}
{"event": "probe", "step": 800, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3132.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.179969, "cps": 24.46875, "mean_p": 0.276337, "boundary_entropy": 0.444212, "bpb": 2.735866}, "de": {"chunks": 3429.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.166999, "cps": 26.789062, "mean_p": 0.271862, "boundary_entropy": 0.445249, "bpb": 3.08416}, "vi": {"chunks": 4090.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.221861, "cps": 31.953125, "mean_p": 0.300175, "boundary_entropy": 0.496012, "bpb": 2.442842}, "tr": {"chunks": 3672.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.202839, "cps": 28.6875, "mean_p": 0.293876, "boundary_entropy": 0.46258, "bpb": 3.002989}, "fi": {"chunks": 3527.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.187497, "cps": 27.554688, "mean_p": 0.278601, "boundary_entropy": 0.450798, "bpb": 3.248352}, "ru": {"chunks": 5608.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.295937, "cps": 43.8125, "mean_p": 0.338856, "boundary_entropy": 0.567533, "bpb": 2.053115}, "el": {"chunks": 6415.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.312576, "cps": 50.117188, "mean_p": 0.325331, "boundary_entropy": 0.565355, "bpb": 2.144425}, "hi": {"chunks": 5512.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.327608, "cps": 43.0625, "mean_p": 0.324814, "boundary_entropy": 0.56343, "bpb": 1.552917}, "ta": {"chunks": 6593.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.352661, "cps": 51.507812, "mean_p": 0.38727, "boundary_entropy": 0.635311, "bpb": 1.286128}, "th": {"chunks": 5977.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.371588, "cps": 46.695312, "mean_p": 0.348652, "boundary_entropy": 0.597103, "bpb": 1.617902}, "zh": {"chunks": 2540.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.430071, "cps": 19.84375, "mean_p": 0.296158, "boundary_entropy": 0.515799, "bpb": 3.742213}, "my": {"chunks": 13131.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.693002, "cps": 102.585938, "mean_p": 0.428076, "boundary_entropy": 0.642727, "bpb": 1.476431}}, "bpb": {"en": 2.73587, "de": 3.08416, "vi": 2.44284, "tr": 3.00299, "fi": 3.24835, "ru": 2.05312, "el": 2.14443, "hi": 1.55292, "ta": 1.28613, "th": 1.6179, "zh": 3.74221, "my": 1.47643}, "gini_cpc": 0.231759, "gini_cps": 0.246272, "bpic": 6.031827, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9279347627116652, "vi": 1.2327713505149605, "tr": 1.1270793372026355, "fi": 1.0418278027183492, "ru": 1.6443761057026456, "el": 1.7368334807242711, "hi": 1.8203567260451003, "ta": 1.959566349926579, "th": 2.0647361300180913, "zh": 2.389695912851792, "my": 3.8506743501270266}, "t": 60.8}
{"event": "eval", "step": 800, "phase": 1, "bpb": {"en": 2.74762, "de": 3.08454, "vi": 2.4499, "tr": 3.01405, "fi": 3.24617, "ru": 2.03669, "el": 2.13144, "hi": 1.54825, "ta": 1.29695, "th": 1.64472, "zh": 3.75457, "my": 1.48838}, "bytes_per_chunk": {"en": 5.587, "de": 5.879, "vi": 6.251, "tr": 5.081, "fi": 5.33, "ru": 6.038, "el": 6.34, "hi": 7.142, "ta": 6.914, "th": 7.006, "zh": 6.954, "my": 4.781}, "bpb_mean": 2.37027, "t": 62.2}
{"event": "train", "step": 800, "phase": 1, "objective": "global", "lm_loss": 1.8173, "bpb": 2.62181, "chunk_loss": 0.029906, "F": 0.1647, "G": 0.3012, "chunks": 168.66, "bytes_per_chunk": 6.072, "boundary_entropy": 0.49902, "router_noise": 0.0, "grad_norm": 0.8023, "lr": 0.0002949, "router_trainable": true, "sec_per_step": 0.087, "alpha": 0.03, "t": 62.2}
{"event": "train", "step": 850, "phase": 1, "objective": "global", "lm_loss": 1.86683, "bpb": 2.69327, "chunk_loss": 0.030233, "F": 0.17441, "G": 0.30388, "chunks": 178.59, "bytes_per_chunk": 5.734, "boundary_entropy": 0.49338, "router_noise": 0.0, "grad_norm": 0.8863, "lr": 0.0002938, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 64.2}
{"event": "train", "step": 900, "phase": 1, "objective": "global", "lm_loss": 1.6808, "bpb": 2.42488, "chunk_loss": 0.03012, "F": 0.17508, "G": 0.29811, "chunks": 179.28, "bytes_per_chunk": 5.712, "boundary_entropy": 0.47556, "router_noise": 0.0, "grad_norm": 1.0246, "lr": 0.0002927, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 66.3}
{"event": "train", "step": 950, "phase": 1, "objective": "global", "lm_loss": 1.80931, "bpb": 2.61028, "chunk_loss": 0.030428, "F": 0.17935, "G": 0.29565, "chunks": 183.66, "bytes_per_chunk": 5.576, "boundary_entropy": 0.46985, "router_noise": 0.0, "grad_norm": 0.7806, "lr": 0.0002914, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 68.3}
{"event": "probe", "step": 1000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3420.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.196518, "cps": 26.71875, "mean_p": 0.268439, "boundary_entropy": 0.398019, "bpb": 2.598458}, "de": {"chunks": 3966.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.193152, "cps": 30.984375, "mean_p": 0.275793, "boundary_entropy": 0.418412, "bpb": 2.956866}, "vi": {"chunks": 4628.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.251044, "cps": 36.15625, "mean_p": 0.295599, "boundary_entropy": 0.478896, "bpb": 2.325367}, "tr": {"chunks": 3646.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.201403, "cps": 28.484375, "mean_p": 0.279209, "boundary_entropy": 0.424361, "bpb": 2.90959}, "fi": {"chunks": 3528.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.18755, "cps": 27.5625, "mean_p": 0.274555, "boundary_entropy": 0.43287, "bpb": 3.149814}, "ru": {"chunks": 6371.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.336201, "cps": 49.773438, "mean_p": 0.364117, "boundary_entropy": 0.58519, "bpb": 1.965885}, "el": {"chunks": 6741.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.328461, "cps": 52.664062, "mean_p": 0.321333, "boundary_entropy": 0.556066, "bpb": 2.039539}, "hi": {"chunks": 4386.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.260684, "cps": 34.265625, "mean_p": 0.32348, "boundary_entropy": 0.576273, "bpb": 1.516999}, "ta": {"chunks": 6574.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.351645, "cps": 51.359375, "mean_p": 0.404924, "boundary_entropy": 0.647518, "bpb": 1.230218}, "th": {"chunks": 7411.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.46074, "cps": 57.898438, "mean_p": 0.343714, "boundary_entropy": 0.593002, "bpb": 1.519126}, "zh": {"chunks": 2995.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.507111, "cps": 23.398438, "mean_p": 0.314806, "boundary_entropy": 0.522466, "bpb": 3.620052}, "my": {"chunks": 9213.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.486225, "cps": 71.976562, "mean_p": 0.422557, "boundary_entropy": 0.652727, "bpb": 1.399511}}, "bpb": {"en": 2.59846, "de": 2.95687, "vi": 2.32537, "tr": 2.90959, "fi": 3.14981, "ru": 1.96589, "el": 2.03954, "hi": 1.517, "ta": 1.23022, "th": 1.51913, "zh": 3.62005, "my": 1.39951}, "gini_cpc": 0.201132, "gini_cps": 0.197343, "bpic": 6.103484, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.982875063761288, "vi": 1.2774626830161926, "tr": 1.024859018229992, "fi": 0.9543654468689932, "ru": 1.7107888256260704, "el": 1.6714042268366427, "hi": 1.3265131774458434, "ta": 1.7893786217348668, "th": 2.3445190299341765, "zh": 2.5804853523921554, "my": 2.474205165383376}, "t": 71.3}
{"event": "eval", "step": 1000, "phase": 1, "bpb": {"en": 2.60799, "de": 2.95973, "vi": 2.32991, "tr": 2.92034, "fi": 3.15076, "ru": 1.95011, "el": 2.02788, "hi": 1.51202, "ta": 1.24175, "th": 1.54472, "zh": 3.63076, "my": 1.40901}, "bytes_per_chunk": {"en": 5.301, "de": 5.222, "vi": 5.946, "tr": 5.232, "fi": 5.395, "ru": 5.427, "el": 5.763, "hi": 8.559, "ta": 7.092, "th": 6.053, "zh": 5.737, "my": 6.14}, "bpb_mean": 2.27375, "t": 72.7}
{"event": "train", "step": 1000, "phase": 1, "objective": "global", "lm_loss": 1.74336, "bpb": 2.51514, "chunk_loss": 0.030286, "F": 0.17804, "G": 0.31021, "chunks": 182.31, "bytes_per_chunk": 5.617, "boundary_entropy": 0.49648, "router_noise": 0.0, "grad_norm": 0.8543, "lr": 0.0002901, "router_trainable": true, "sec_per_step": 0.088, "alpha": 0.03, "t": 72.7}
{"event": "train", "step": 1050, "phase": 1, "objective": "global", "lm_loss": 1.60813, "bpb": 2.32003, "chunk_loss": 0.030902, "F": 0.1911, "G": 0.33579, "chunks": 195.69, "bytes_per_chunk": 5.233, "boundary_entropy": 0.51444, "router_noise": 0.0, "grad_norm": 0.7676, "lr": 0.0002886, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 74.7}
{"event": "train", "step": 1100, "phase": 1, "objective": "global", "lm_loss": 1.80487, "bpb": 2.60388, "chunk_loss": 0.030932, "F": 0.19345, "G": 0.32235, "chunks": 198.09, "bytes_per_chunk": 5.169, "boundary_entropy": 0.49596, "router_noise": 0.0, "grad_norm": 0.74, "lr": 0.0002871, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 76.7}
{"event": "train", "step": 1150, "phase": 1, "objective": "global", "lm_loss": 1.66004, "bpb": 2.39493, "chunk_loss": 0.030291, "F": 0.17892, "G": 0.2995, "chunks": 183.22, "bytes_per_chunk": 5.589, "boundary_entropy": 0.46823, "router_noise": 0.0, "grad_norm": 0.7815, "lr": 0.0002855, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 78.7}
{"event": "probe", "step": 1200, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3117.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.179107, "cps": 24.351562, "mean_p": 0.245642, "boundary_entropy": 0.365693, "bpb": 2.50155}, "de": {"chunks": 3589.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.174792, "cps": 28.039062, "mean_p": 0.249188, "boundary_entropy": 0.38248, "bpb": 2.829325}, "vi": {"chunks": 4182.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.226851, "cps": 32.671875, "mean_p": 0.311164, "boundary_entropy": 0.490463, "bpb": 2.271586}, "tr": {"chunks": 3469.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.191626, "cps": 27.101562, "mean_p": 0.260197, "boundary_entropy": 0.396853, "bpb": 2.800104}, "fi": {"chunks": 3524.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.187337, "cps": 27.53125, "mean_p": 0.264624, "boundary_entropy": 0.407426, "bpb": 3.056588}, "ru": {"chunks": 5505.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.290501, "cps": 43.007812, "mean_p": 0.355456, "boundary_entropy": 0.579112, "bpb": 1.874495}, "el": {"chunks": 6412.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.31243, "cps": 50.09375, "mean_p": 0.317629, "boundary_entropy": 0.547487, "bpb": 1.986912}, "hi": {"chunks": 7356.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.437207, "cps": 57.46875, "mean_p": 0.351352, "boundary_entropy": 0.588696, "bpb": 1.473507}, "ta": {"chunks": 7872.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.421075, "cps": 61.5, "mean_p": 0.410237, "boundary_entropy": 0.647477, "bpb": 1.214056}, "th": {"chunks": 7896.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.490892, "cps": 61.6875, "mean_p": 0.378946, "boundary_entropy": 0.626538, "bpb": 1.460722}, "zh": {"chunks": 2812.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.476126, "cps": 21.96875, "mean_p": 0.321096, "boundary_entropy": 0.527185, "bpb": 3.58628}, "my": {"chunks": 6788.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.358244, "cps": 53.03125, "mean_p": 0.393005, "boundary_entropy": 0.641709, "bpb": 1.325541}}, "bpb": {"en": 2.50155, "de": 2.82932, "vi": 2.27159, "tr": 2.8001, "fi": 3.05659, "ru": 1.8745, "el": 1.98691, "hi": 1.47351, "ta": 1.21406, "th": 1.46072, "zh": 3.58628, "my": 1.32554}, "gini_cpc": 0.210497, "gini_cps": 0.202649, "bpic": 6.138335, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9759068561057631, "vi": 1.2665671050354328, "tr": 1.069894774405276, "fi": 1.0459509873530983, "ru": 1.6219423994555349, "el": 1.7443755326602042, "hi": 2.4410347702600084, "ta": 2.350969169493266, "th": 2.7407750512122346, "zh": 2.6583318313507704, "my": 2.000164779022391}, "t": 81.7}
{"event": "eval", "step": 1200, "phase": 1, "bpb": {"en": 2.51135, "de": 2.83391, "vi": 2.27612, "tr": 2.8112, "fi": 3.05915, "ru": 1.85672, "el": 1.97685, "hi": 1.47116, "ta": 1.22253, "th": 1.48403, "zh": 3.59455, "my": 1.33589}, "bytes_per_chunk": {"en": 5.609, "de": 5.714, "vi": 5.763, "tr": 5.463, "fi": 5.333, "ru": 6.409, "el": 5.817, "hi": 5.329, "ta": 6.275, "th": 5.517, "zh": 5.976, "my": 8.129}, "bpb_mean": 2.20279, "t": 83.0}
{"event": "train", "step": 1200, "phase": 1, "objective": "global", "lm_loss": 1.72029, "bpb": 2.48185, "chunk_loss": 0.030218, "F": 0.17676, "G": 0.28845, "chunks": 181.0, "bytes_per_chunk": 5.657, "boundary_entropy": 0.44948, "router_noise": 0.0, "grad_norm": 0.8547, "lr": 0.0002837, "router_trainable": true, "sec_per_step": 0.087, "alpha": 0.03, "t": 83.1}
{"event": "train", "step": 1250, "phase": 1, "objective": "global", "lm_loss": 1.84025, "bpb": 2.65492, "chunk_loss": 0.030253, "F": 0.17807, "G": 0.29889, "chunks": 182.34, "bytes_per_chunk": 5.616, "boundary_entropy": 0.45602, "router_noise": 0.0, "grad_norm": 0.8572, "lr": 0.0002819, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 85.1}
{"event": "train", "step": 1300, "phase": 1, "objective": "global", "lm_loss": 1.51406, "bpb": 2.18433, "chunk_loss": 0.030296, "F": 0.17883, "G": 0.32074, "chunks": 183.12, "bytes_per_chunk": 5.592, "boundary_entropy": 0.49582, "router_noise": 0.0, "grad_norm": 0.7504, "lr": 0.00028, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 87.1}
{"event": "train", "step": 1350, "phase": 1, "objective": "global", "lm_loss": 1.51985, "bpb": 2.19268, "chunk_loss": 0.029939, "F": 0.17038, "G": 0.3092, "chunks": 174.47, "bytes_per_chunk": 5.869, "boundary_entropy": 0.48323, "router_noise": 0.0, "grad_norm": 0.7285, "lr": 0.000278, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 89.2}
{"event": "probe", "step": 1400, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3406.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.195713, "cps": 26.609375, "mean_p": 0.255179, "boundary_entropy": 0.358543, "bpb": 2.440371}, "de": {"chunks": 3849.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.187454, "cps": 30.070312, "mean_p": 0.260926, "boundary_entropy": 0.383062, "bpb": 2.756701}, "vi": {"chunks": 4032.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.218714, "cps": 31.5, "mean_p": 0.295224, "boundary_entropy": 0.474828, "bpb": 2.192047}, "tr": {"chunks": 3432.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.189582, "cps": 26.8125, "mean_p": 0.257016, "boundary_entropy": 0.386298, "bpb": 2.727987}, "fi": {"chunks": 3378.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.179576, "cps": 26.390625, "mean_p": 0.255323, "boundary_entropy": 0.38802, "bpb": 3.005668}, "ru": {"chunks": 6191.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.326702, "cps": 48.367188, "mean_p": 0.368762, "boundary_entropy": 0.583349, "bpb": 1.818349}, "el": {"chunks": 5469.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.266482, "cps": 42.726562, "mean_p": 0.301838, "boundary_entropy": 0.540507, "bpb": 1.898762}, "hi": {"chunks": 5693.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.338366, "cps": 44.476562, "mean_p": 0.375711, "boundary_entropy": 0.626039, "bpb": 1.431276}, "ta": {"chunks": 6592.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.352608, "cps": 51.5, "mean_p": 0.372124, "boundary_entropy": 0.625331, "bpb": 1.148273}, "th": {"chunks": 6283.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.390612, "cps": 49.085938, "mean_p": 0.362237, "boundary_entropy": 0.618768, "bpb": 1.433053}, "zh": {"chunks": 2354.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.398578, "cps": 18.390625, "mean_p": 0.313239, "boundary_entropy": 0.524665, "bpb": 3.54494}, "my": {"chunks": 7494.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.395503, "cps": 58.546875, "mean_p": 0.410524, "boundary_entropy": 0.65221, "bpb": 1.292792}}, "bpb": {"en": 2.44037, "de": 2.7567, "vi": 2.19205, "tr": 2.72799, "fi": 3.00567, "ru": 1.81835, "el": 1.89876, "hi": 1.43128, "ta": 1.14827, "th": 1.43305, "zh": 3.54494, "my": 1.29279}, "gini_cpc": 0.166971, "gini_cps": 0.181025, "bpic": 6.597236, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9578003259522453, "vi": 1.1175239979990321, "tr": 0.9686707909321258, "fi": 0.9175447156720092, "ru": 1.6692872116101185, "el": 1.3615906322136495, "hi": 1.7288829344945855, "ta": 1.8016532346910108, "th": 1.9958388449792657, "zh": 2.036537879907154, "my": 2.0208300406428026}, "t": 92.2}
{"event": "eval", "step": 1400, "phase": 1, "bpb": {"en": 2.44542, "de": 2.76036, "vi": 2.19615, "tr": 2.73784, "fi": 3.00303, "ru": 1.79996, "el": 1.88718, "hi": 1.42799, "ta": 1.15819, "th": 1.45299, "zh": 3.55146, "my": 1.30378}, "bytes_per_chunk": {"en": 5.202, "de": 5.303, "vi": 5.912, "tr": 5.45, "fi": 5.548, "ru": 5.678, "el": 6.724, "hi": 6.715, "ta": 7.103, "th": 7.071, "zh": 7.206, "my": 7.627}, "bpb_mean": 2.1437, "t": 93.5}
{"event": "train", "step": 1400, "phase": 1, "objective": "global", "lm_loss": 1.57436, "bpb": 2.27131, "chunk_loss": 0.030094, "F": 0.17569, "G": 0.30559, "chunks": 179.91, "bytes_per_chunk": 5.692, "boundary_entropy": 0.46787, "router_noise": 0.0, "grad_norm": 0.7232, "lr": 0.0002759, "router_trainable": true, "sec_per_step": 0.088, "alpha": 0.03, "t": 93.6}
{"event": "train", "step": 1450, "phase": 1, "objective": "global", "lm_loss": 1.67023, "bpb": 2.40963, "chunk_loss": 0.030166, "F": 0.17618, "G": 0.29603, "chunks": 180.41, "bytes_per_chunk": 5.676, "boundary_entropy": 0.46247, "router_noise": 0.0, "grad_norm": 0.75, "lr": 0.0002738, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 95.6}
{"event": "train", "step": 1500, "phase": 1, "objective": "global", "lm_loss": 1.74547, "bpb": 2.51818, "chunk_loss": 0.030144, "F": 0.17462, "G": 0.30892, "chunks": 178.81, "bytes_per_chunk": 5.727, "boundary_entropy": 0.48459, "router_noise": 0.0, "grad_norm": 0.7118, "lr": 0.0002715, "router_trainable": true, "sec_per_step": 0.042, "alpha": 0.03, "t": 97.7}
{"event": "train", "step": 1550, "phase": 1, "objective": "global", "lm_loss": 1.61022, "bpb": 2.32305, "chunk_loss": 0.030692, "F": 0.18976, "G": 0.30976, "chunks": 194.31, "bytes_per_chunk": 5.27, "boundary_entropy": 0.46298, "router_noise": 0.0, "grad_norm": 0.7839, "lr": 0.0002692, "router_trainable": true, "sec_per_step": 0.042, "alpha": 0.03, "t": 99.8}
{"event": "probe", "step": 1600, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3505.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.201402, "cps": 27.382812, "mean_p": 0.249865, "boundary_entropy": 0.336756, "bpb": 2.374651}, "de": {"chunks": 3757.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.182974, "cps": 29.351562, "mean_p": 0.25055, "boundary_entropy": 0.359729, "bpb": 2.667069}, "vi": {"chunks": 4419.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.239707, "cps": 34.523438, "mean_p": 0.293946, "boundary_entropy": 0.448755, "bpb": 2.160934}, "tr": {"chunks": 3806.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.210241, "cps": 29.734375, "mean_p": 0.254125, "boundary_entropy": 0.358286, "bpb": 2.654309}, "fi": {"chunks": 3712.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.197331, "cps": 29.0, "mean_p": 0.263532, "boundary_entropy": 0.374463, "bpb": 2.889249}, "ru": {"chunks": 5683.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.299894, "cps": 44.398438, "mean_p": 0.339119, "boundary_entropy": 0.554764, "bpb": 1.767541}, "el": {"chunks": 5870.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.286021, "cps": 45.859375, "mean_p": 0.296307, "boundary_entropy": 0.523334, "bpb": 1.865329}, "hi": {"chunks": 6811.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.404814, "cps": 53.210938, "mean_p": 0.386451, "boundary_entropy": 0.62578, "bpb": 1.39786}, "ta": {"chunks": 9540.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.510297, "cps": 74.53125, "mean_p": 0.411434, "boundary_entropy": 0.637736, "bpb": 1.158653}, "th": {"chunks": 7447.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.462978, "cps": 58.179688, "mean_p": 0.386355, "boundary_entropy": 0.633461, "bpb": 1.388602}, "zh": {"chunks": 2366.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.40061, "cps": 18.484375, "mean_p": 0.318599, "boundary_entropy": 0.532346, "bpb": 3.52347}, "my": {"chunks": 9118.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.481212, "cps": 71.234375, "mean_p": 0.417709, "boundary_entropy": 0.654395, "bpb": 1.251529}}, "bpb": {"en": 2.37465, "de": 2.66707, "vi": 2.16093, "tr": 2.65431, "fi": 2.88925, "ru": 1.76754, "el": 1.86533, "hi": 1.39786, "ta": 1.15865, "th": 1.3886, "zh": 3.52347, "my": 1.25153}, "gini_cpc": 0.203224, "gini_cps": 0.225013, "bpic": 5.81187, "worst_lang": "ta", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9084999040895374, "vi": 1.1901918101422007, "tr": 1.0438890221062658, "fi": 0.9797881486619167, "ru": 1.4890337437283336, "el": 1.4201471742187168, "hi": 2.0099807831308705, "ta": 2.5337222376317174, "th": 2.2987745824274306, "zh": 1.9891035640150279, "my": 2.389308948855295}, "t": 102.8}
{"event": "eval", "step": 1600, "phase": 1, "bpb": {"en": 2.38113, "de": 2.67354, "vi": 2.16723, "tr": 2.66527, "fi": 2.8964, "ru": 1.74916, "el": 1.85575, "hi": 1.39504, "ta": 1.16724, "th": 1.40662, "zh": 3.52786, "my": 1.26114}, "bytes_per_chunk": {"en": 5.002, "de": 5.359, "vi": 5.601, "tr": 5.091, "fi": 5.025, "ru": 6.195, "el": 5.997, "hi": 5.743, "ta": 5.117, "th": 6.051, "zh": 6.947, "my": 6.13}, "bpb_mean": 2.09553, "t": 104.2}
{"event": "train", "step": 1600, "phase": 1, "objective": "global", "lm_loss": 1.60371, "bpb": 2.31366, "chunk_loss": 0.030469, "F": 0.18393, "G": 0.30486, "chunks": 188.34, "bytes_per_chunk": 5.437, "boundary_entropy": 0.45893, "router_noise": 0.0, "grad_norm": 0.7523, "lr": 0.0002668, "router_trainable": true, "sec_per_step": 0.088, "alpha": 0.03, "t": 104.2}
{"event": "train", "step": 1650, "phase": 1, "objective": "global", "lm_loss": 1.57994, "bpb": 2.27937, "chunk_loss": 0.03053, "F": 0.18622, "G": 0.30018, "chunks": 190.69, "bytes_per_chunk": 5.37, "boundary_entropy": 0.45046, "router_noise": 0.0, "grad_norm": 0.7884, "lr": 0.0002643, "router_trainable": true, "sec_per_step": 0.042, "alpha": 0.03, "t": 106.3}
{"event": "train", "step": 1700, "phase": 1, "objective": "global", "lm_loss": 1.54289, "bpb": 2.22591, "chunk_loss": 0.030481, "F": 0.18561, "G": 0.31085, "chunks": 190.06, "bytes_per_chunk": 5.388, "boundary_entropy": 0.46231, "router_noise": 0.0, "grad_norm": 0.6961, "lr": 0.0002618, "router_trainable": true, "sec_per_step": 0.042, "alpha": 0.03, "t": 108.4}
{"event": "train", "step": 1750, "phase": 1, "objective": "global", "lm_loss": 1.56836, "bpb": 2.26266, "chunk_loss": 0.030529, "F": 0.18924, "G": 0.29573, "chunks": 193.78, "bytes_per_chunk": 5.284, "boundary_entropy": 0.44045, "router_noise": 0.0, "grad_norm": 0.7099, "lr": 0.0002591, "router_trainable": true, "sec_per_step": 0.042, "alpha": 0.03, "t": 110.5}
{"event": "probe", "step": 1800, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3421.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.196575, "cps": 26.726562, "mean_p": 0.242852, "boundary_entropy": 0.323508, "bpb": 2.329583}, "de": {"chunks": 4019.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.195734, "cps": 31.398438, "mean_p": 0.253077, "boundary_entropy": 0.344582, "bpb": 2.628986}, "vi": {"chunks": 4350.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.235964, "cps": 33.984375, "mean_p": 0.292964, "boundary_entropy": 0.454, "bpb": 2.125162}, "tr": {"chunks": 3720.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.205491, "cps": 29.0625, "mean_p": 0.258233, "boundary_entropy": 0.3627, "bpb": 2.621574}, "fi": {"chunks": 3615.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.192175, "cps": 28.242188, "mean_p": 0.253324, "boundary_entropy": 0.35753, "bpb": 2.843958}, "ru": {"chunks": 7008.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.369815, "cps": 54.75, "mean_p": 0.368994, "boundary_entropy": 0.565556, "bpb": 1.714039}, "el": {"chunks": 6209.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.302539, "cps": 48.507812, "mean_p": 0.309113, "boundary_entropy": 0.524423, "bpb": 1.795966}, "hi": {"chunks": 6491.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.385795, "cps": 50.710938, "mean_p": 0.390334, "boundary_entropy": 0.632552, "bpb": 1.373572}, "ta": {"chunks": 8292.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.443541, "cps": 64.78125, "mean_p": 0.355528, "boundary_entropy": 0.600082, "bpb": 1.111273}, "th": {"chunks": 6762.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.420392, "cps": 52.828125, "mean_p": 0.371154, "boundary_entropy": 0.621146, "bpb": 1.34776}, "zh": {"chunks": 2469.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.418049, "cps": 19.289062, "mean_p": 0.305148, "boundary_entropy": 0.506884, "bpb": 3.478284}, "my": {"chunks": 11879.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.626926, "cps": 92.804688, "mean_p": 0.427279, "boundary_entropy": 0.655253, "bpb": 1.249257}}, "bpb": {"en": 2.32958, "de": 2.62899, "vi": 2.12516, "tr": 2.62157, "fi": 2.84396, "ru": 1.71404, "el": 1.79597, "hi": 1.37357, "ta": 1.11127, "th": 1.34776, "zh": 3.47828, "my": 1.24926}, "gini_cpc": 0.212013, "gini_cps": 0.238862, "bpic": 5.624401, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9957186578387522, "vi": 1.2003756057034414, "tr": 1.0453541181407253, "fi": 0.9776141168621898, "ru": 1.8812907704161554, "el": 1.5390469221098897, "hi": 1.9625809646128514, "ta": 2.2563417008316473, "th": 2.138578257842976, "zh": 2.126663088572771, "my": 3.1892425689483743}, "t": 113.5}
{"event": "eval", "step": 1800, "phase": 1, "bpb": {"en": 2.3364, "de": 2.63357, "vi": 2.13081, "tr": 2.63082, "fi": 2.8456, "ru": 1.69434, "el": 1.78849, "hi": 1.37022, "ta": 1.11773, "th": 1.36833, "zh": 3.48041, "my": 1.25689}, "bytes_per_chunk": {"en": 5.09, "de": 5.08, "vi": 5.611, "tr": 5.141, "fi": 5.055, "ru": 5.082, "el": 5.711, "hi": 6.121, "ta": 5.91, "th": 6.706, "zh": 6.667, "my": 4.791}, "bpb_mean": 2.05447, "t": 114.9}
{"event": "train", "step": 1800, "phase": 1, "objective": "global", "lm_loss": 1.53192, "bpb": 2.21009, "chunk_loss": 0.030657, "F": 0.18832, "G": 0.30836, "chunks": 192.84, "bytes_per_chunk": 5.31, "boundary_entropy": 0.46486, "router_noise": 0.0, "grad_norm": 0.6827, "lr": 0.0002564, "router_trainable": true, "sec_per_step": 0.089, "alpha": 0.03, "t": 114.9}
{"event": "train", "step": 1850, "phase": 1, "objective": "global", "lm_loss": 1.58059, "bpb": 2.28031, "chunk_loss": 0.030411, "F": 0.18472, "G": 0.28801, "chunks": 189.16, "bytes_per_chunk": 5.414, "boundary_entropy": 0.42083, "router_noise": 0.0, "grad_norm": 0.7135, "lr": 0.0002537, "router_trainable": true, "sec_per_step": 0.042, "alpha": 0.03, "t": 117.0}
{"event": "train", "step": 1900, "phase": 1, "objective": "global", "lm_loss": 1.63077, "bpb": 2.35271, "chunk_loss": 0.030911, "F": 0.19318, "G": 0.29736, "chunks": 197.81, "bytes_per_chunk": 5.177, "boundary_entropy": 0.43434, "router_noise": 0.0, "grad_norm": 0.7547, "lr": 0.0002508, "router_trainable": true, "sec_per_step": 0.042, "alpha": 0.03, "t": 119.1}
{"event": "train", "step": 1950, "phase": 1, "objective": "global", "lm_loss": 1.48126, "bpb": 2.137, "chunk_loss": 0.029996, "F": 0.16916, "G": 0.3043, "chunks": 173.22, "bytes_per_chunk": 5.912, "boundary_entropy": 0.46611, "router_noise": 0.0, "grad_norm": 0.6933, "lr": 0.0002479, "router_trainable": true, "sec_per_step": 0.042, "alpha": 0.03, "t": 121.2}
{"event": "probe", "step": 2000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3504.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.201345, "cps": 27.375, "mean_p": 0.241328, "boundary_entropy": 0.315741, "bpb": 2.292236}, "de": {"chunks": 4048.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.197146, "cps": 31.625, "mean_p": 0.249209, "boundary_entropy": 0.334801, "bpb": 2.568157}, "vi": {"chunks": 4230.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.229455, "cps": 33.046875, "mean_p": 0.272986, "boundary_entropy": 0.421747, "bpb": 2.094944}, "tr": {"chunks": 3935.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.217367, "cps": 30.742188, "mean_p": 0.248916, "boundary_entropy": 0.336578, "bpb": 2.570404}, "fi": {"chunks": 3352.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.178194, "cps": 26.1875, "mean_p": 0.238495, "boundary_entropy": 0.344347, "bpb": 2.776388}, "ru": {"chunks": 5992.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.316201, "cps": 46.8125, "mean_p": 0.338714, "boundary_entropy": 0.540458, "bpb": 1.690044}, "el": {"chunks": 6938.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.33806, "cps": 54.203125, "mean_p": 0.317212, "boundary_entropy": 0.522086, "bpb": 1.7385}, "hi": {"chunks": 6894.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.409747, "cps": 53.859375, "mean_p": 0.388833, "boundary_entropy": 0.628918, "bpb": 1.362443}, "ta": {"chunks": 7410.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.396363, "cps": 57.890625, "mean_p": 0.386617, "boundary_entropy": 0.626052, "bpb": 1.09926}, "th": {"chunks": 9628.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.59857, "cps": 75.21875, "mean_p": 0.413712, "boundary_entropy": 0.642793, "bpb": 1.316411}, "zh": {"chunks": 2737.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.463427, "cps": 21.382812, "mean_p": 0.319021, "boundary_entropy": 0.515935, "bpb": 3.432498}, "my": {"chunks": 13896.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.733376, "cps": 108.5625, "mean_p": 0.432367, "boundary_entropy": 0.651994, "bpb": 1.223382}}, "bpb": {"en": 2.29224, "de": 2.56816, "vi": 2.09494, "tr": 2.5704, "fi": 2.77639, "ru": 1.69004, "el": 1.7385, "hi": 1.36244, "ta": 1.09926, "th": 1.31641, "zh": 3.4325, "my": 1.22338}, "gini_cpc": 0.253917, "gini_cps": 0.266305, "bpic": 5.288862, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9791474999482958, "vi": 1.1396126152234247, "tr": 1.0795784529015473, "fi": 0.8850180927364708, "ru": 1.5704445729569523, "el": 1.679010735699806, "hi": 2.0350553644486964, "ta": 1.9685786068357594, "th": 2.9728639803100823, "zh": 2.3016610980706873, "my": 3.642390059600753}, "t": 124.2}
{"event": "eval", "step": 2000, "phase": 1, "bpb": {"en": 2.29897, "de": 2.57408, "vi": 2.10172, "tr": 2.57683, "fi": 2.78764, "ru": 1.6663, "el": 1.72946, "hi": 1.36105, "ta": 1.10719, "th": 1.33328, "zh": 3.43621, "my": 1.23232}, "bytes_per_chunk": {"en": 5.07, "de": 5.078, "vi": 5.6, "tr": 4.965, "fi": 5.474, "ru": 5.831, "el": 5.15, "hi": 6.249, "ta": 6.611, "th": 5.118, "zh": 6.298, "my": 4.133}, "bpb_mean": 2.01709, "t": 125.6}
{"event": "train", "step": 2000, "phase": 1, "objective": "global", "lm_loss": 1.60014, "bpb": 2.30852, "chunk_loss": 0.030551, "F": 0.19031, "G": 0.2886, "chunks": 194.88, "bytes_per_chunk": 5.255, "boundary_entropy": 0.41863, "router_noise": 0.0, "grad_norm": 0.7711, "lr": 0.000245, "router_trainable": true, "sec_per_step": 0.089, "alpha": 0.03, "t": 125.6}
{"event": "train", "step": 2050, "phase": 1, "objective": "global", "lm_loss": 1.61283, "bpb": 2.32683, "chunk_loss": 0.0307, "F": 0.19131, "G": 0.29978, "chunks": 195.91, "bytes_per_chunk": 5.227, "boundary_entropy": 0.45006, "router_noise": 0.0, "grad_norm": 0.7824, "lr": 0.0002419, "router_trainable": true, "sec_per_step": 0.042, "alpha": 0.03, "t": 127.7}
{"event": "train", "step": 2100, "phase": 1, "objective": "global", "lm_loss": 1.55386, "bpb": 2.24175, "chunk_loss": 0.030279, "F": 0.17938, "G": 0.28466, "chunks": 183.69, "bytes_per_chunk": 5.575, "boundary_entropy": 0.42208, "router_noise": 0.0, "grad_norm": 0.7327, "lr": 0.0002388, "router_trainable": true, "sec_per_step": 0.042, "alpha": 0.03, "t": 129.8}
{"event": "train", "step": 2150, "phase": 1, "objective": "global", "lm_loss": 1.51357, "bpb": 2.18363, "chunk_loss": 0.030116, "F": 0.17566, "G": 0.28582, "chunks": 179.88, "bytes_per_chunk": 5.693, "boundary_entropy": 0.42721, "router_noise": 0.0, "grad_norm": 0.6961, "lr": 0.0002357, "router_trainable": true, "sec_per_step": 0.042, "alpha": 0.03, "t": 131.9}
{"event": "probe", "step": 2200, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3461.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.198874, "cps": 27.039062, "mean_p": 0.23631, "boundary_entropy": 0.304809, "bpb": 2.262007}, "de": {"chunks": 3923.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.191058, "cps": 30.648438, "mean_p": 0.2425, "boundary_entropy": 0.328017, "bpb": 2.518995}, "vi": {"chunks": 4117.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.223325, "cps": 32.164062, "mean_p": 0.261368, "boundary_entropy": 0.410403, "bpb": 2.077317}, "tr": {"chunks": 3711.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.204994, "cps": 28.992188, "mean_p": 0.246883, "boundary_entropy": 0.335235, "bpb": 2.516129}, "fi": {"chunks": 3566.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.18957, "cps": 27.859375, "mean_p": 0.245432, "boundary_entropy": 0.335708, "bpb": 2.721495}, "ru": {"chunks": 5385.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.284169, "cps": 42.070312, "mean_p": 0.316299, "boundary_entropy": 0.524959, "bpb": 1.65442}, "el": {"chunks": 6386.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.311163, "cps": 49.890625, "mean_p": 0.298105, "boundary_entropy": 0.503883, "bpb": 1.705421}, "hi": {"chunks": 7866.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.467519, "cps": 61.453125, "mean_p": 0.400292, "boundary_entropy": 0.633775, "bpb": 1.328529}, "ta": {"chunks": 8142.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.435518, "cps": 63.609375, "mean_p": 0.382832, "boundary_entropy": 0.620791, "bpb": 1.079679}, "th": {"chunks": 7053.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.438483, "cps": 55.101562, "mean_p": 0.378885, "boundary_entropy": 0.623793, "bpb": 1.282612}, "zh": {"chunks": 2887.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.488825, "cps": 22.554688, "mean_p": 0.31413, "boundary_entropy": 0.504386, "bpb": 3.438489}, "my": {"chunks": 10237.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.540268, "cps": 79.976562, "mean_p": 0.427575, "boundary_entropy": 0.655539, "bpb": 1.204511}}, "bpb": {"en": 2.26201, "de": 2.51899, "vi": 2.07732, "tr": 2.51613, "fi": 2.72149, "ru": 1.65442, "el": 1.70542, "hi": 1.32853, "ta": 1.07968, "th": 1.28261, "zh": 3.43849, "my": 1.20451}, "gini_cpc": 0.214933, "gini_cps": 0.222832, "bpic": 5.750907, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9607013962048928, "vi": 1.1229495513062089, "tr": 1.0307727381614495, "fi": 0.9532174328400098, "ru": 1.4288907151607928, "el": 1.5646261698367523, "hi": 2.3508308972876715, "ta": 2.1899194934052533, "th": 2.2048311677067356, "zh": 2.457965948859005, "my": 2.7166384808265227}, "t": 134.9}
{"event": "eval", "step": 2200, "phase": 1, "bpb": {"en": 2.26914, "de": 2.52523, "vi": 2.08149, "tr": 2.52285, "fi": 2.73247, "ru": 1.63401, "el": 1.69573, "hi": 1.32562, "ta": 1.08584, "th": 1.29925, "zh": 3.44041, "my": 1.21254}, "bytes_per_chunk": {"en": 5.166, "de": 5.213, "vi": 5.76, "tr": 5.129, "fi": 5.087, "ru": 6.36, "el": 5.535, "hi": 5.495, "ta": 5.99, "th": 6.537, "zh": 5.701, "my": 5.428}, "bpb_mean": 1.98538, "t": 136.2}
{"event": "train", "step": 2200, "phase": 1, "objective": "global", "lm_loss": 1.45242, "bpb": 2.0954, "chunk_loss": 0.030396, "F": 0.18646, "G": 0.29291, "chunks": 190.94, "bytes_per_chunk": 5.363, "boundary_entropy": 0.4314, "router_noise": 0.0, "grad_norm": 0.6705, "lr": 0.0002325, "router_trainable": true, "sec_per_step": 0.087, "alpha": 0.03, "t": 136.3}
{"event": "train", "step": 2250, "phase": 1, "objective": "global", "lm_loss": 1.53702, "bpb": 2.21745, "chunk_loss": 0.030104, "F": 0.17615, "G": 0.28309, "chunks": 180.38, "bytes_per_chunk": 5.677, "boundary_entropy": 0.42359, "router_noise": 0.0, "grad_norm": 0.7605, "lr": 0.0002293, "router_trainable": true, "sec_per_step": 0.042, "alpha": 0.03, "t": 138.3}
{"event": "train", "step": 2300, "phase": 1, "objective": "global", "lm_loss": 1.47944, "bpb": 2.13437, "chunk_loss": 0.030024, "F": 0.17517, "G": 0.29302, "chunks": 179.38, "bytes_per_chunk": 5.709, "boundary_entropy": 0.43737, "router_noise": 0.0, "grad_norm": 0.7162, "lr": 0.000226, "router_trainable": true, "sec_per_step": 0.042, "alpha": 0.03, "t": 140.4}
{"event": "train", "step": 2350, "phase": 1, "objective": "global", "lm_loss": 1.49331, "bpb": 2.1544, "chunk_loss": 0.030284, "F": 0.18185, "G": 0.28737, "chunks": 186.22, "bytes_per_chunk": 5.499, "boundary_entropy": 0.42212, "router_noise": 0.0, "grad_norm": 0.6868, "lr": 0.0002226, "router_trainable": true, "sec_per_step": 0.042, "alpha": 0.03, "t": 142.5}
{"event": "probe", "step": 2400, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3467.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.199219, "cps": 27.085938, "mean_p": 0.234826, "boundary_entropy": 0.298321, "bpb": 2.230216}, "de": {"chunks": 3935.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.191643, "cps": 30.742188, "mean_p": 0.242792, "boundary_entropy": 0.321786, "bpb": 2.483784}, "vi": {"chunks": 4050.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.219691, "cps": 31.640625, "mean_p": 0.258185, "boundary_entropy": 0.398585, "bpb": 2.062207}, "tr": {"chunks": 3663.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.202342, "cps": 28.617188, "mean_p": 0.243203, "boundary_entropy": 0.327687, "bpb": 2.474013}, "fi": {"chunks": 3624.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.192653, "cps": 28.3125, "mean_p": 0.249498, "boundary_entropy": 0.333886, "bpb": 2.676022}, "ru": {"chunks": 5858.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.309129, "cps": 45.765625, "mean_p": 0.315813, "boundary_entropy": 0.510696, "bpb": 1.61463}, "el": {"chunks": 6127.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.298543, "cps": 47.867188, "mean_p": 0.28931, "boundary_entropy": 0.495414, "bpb": 1.692301}, "hi": {"chunks": 6294.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.374086, "cps": 49.171875, "mean_p": 0.403173, "boundary_entropy": 0.637216, "bpb": 1.314526}, "ta": {"chunks": 9744.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.521209, "cps": 76.125, "mean_p": 0.402824, "boundary_entropy": 0.630136, "bpb": 1.063663}, "th": {"chunks": 8383.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.521169, "cps": 65.492188, "mean_p": 0.40194, "boundary_entropy": 0.635165, "bpb": 1.258754}, "zh": {"chunks": 2534.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.429055, "cps": 19.796875, "mean_p": 0.307344, "boundary_entropy": 0.503072, "bpb": 3.377802}, "my": {"chunks": 8659.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.456988, "cps": 67.648438, "mean_p": 0.405655, "boundary_entropy": 0.649042, "bpb": 1.178334}}, "bpb": {"en": 2.23022, "de": 2.48378, "vi": 2.06221, "tr": 2.47401, "fi": 2.67602, "ru": 1.61463, "el": 1.6923, "hi": 1.31453, "ta": 1.06366, "th": 1.25875, "zh": 3.3778, "my": 1.17833}, "gini_cpc": 0.212681, "gini_cps": 0.227233, "bpic": 5.785236, "worst_lang": "ta", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9619723916326186, "vi": 1.1027629329939406, "tr": 1.0156794037311827, "fi": 0.96704477801798, "ru": 1.5517095448758098, "el": 1.4985709646030037, "hi": 1.877768045220249, "ta": 2.616267126576772, "th": 2.616065897396286, "zh": 2.1536912640892294, "my": 2.293900848916035}, "t": 145.6}
{"event": "eval", "step": 2400, "phase": 1, "bpb": {"en": 2.23852, "de": 2.48873, "vi": 2.06841, "tr": 2.48392, "fi": 2.68776, "ru": 1.59236, "el": 1.68315, "hi": 1.3129, "ta": 1.07164, "th": 1.27699, "zh": 3.38242, "my": 1.18704}, "bytes_per_chunk": {"en": 5.087, "de": 5.251, "vi": 5.806, "tr": 5.284, "fi": 5.018, "ru": 5.802, "el": 5.734, "hi": 6.37, "ta": 5.118, "th": 5.592, "zh": 6.456, "my": 6.35}, "bpb_mean": 1.95615, "t": 146.9}
{"event": "phase_transition", "step": 2400, "phase": 2, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0, "t": 146.9}
{"event": "train", "step": 2400, "phase": 2, "objective": "global", "lm_loss": 1.3777, "bpb": 1.9876, "chunk_loss": 0.030405, "F": 0.18613, "G": 0.29135, "chunks": 190.59, "bytes_per_chunk": 5.373, "boundary_entropy": 0.42597, "router_noise": 0.0, "grad_norm": 0.7029, "lr": 0.0002192, "router_trainable": true, "sec_per_step": 0.087, "alpha": 0.03, "t": 146.9}
{"event": "train", "step": 2450, "phase": 2, "objective": "global", "lm_loss": 1.46434, "bpb": 2.11259, "chunk_loss": 0.030449, "F": 0.18484, "G": 0.30248, "chunks": 189.28, "bytes_per_chunk": 5.41, "boundary_entropy": 0.45738, "router_noise": 0.0, "grad_norm": 0.6804, "lr": 0.0002158, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 148.9}
{"event": "train", "step": 2500, "phase": 2, "objective": "global", "lm_loss": 1.50172, "bpb": 2.16653, "chunk_loss": 0.030198, "F": 0.17914, "G": 0.28055, "chunks": 183.44, "bytes_per_chunk": 5.582, "boundary_entropy": 0.40905, "router_noise": 0.0, "grad_norm": 0.7486, "lr": 0.0002123, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 151.0}
{"event": "train", "step": 2550, "phase": 2, "objective": "global", "lm_loss": 1.56673, "bpb": 2.26031, "chunk_loss": 0.030296, "F": 0.1839, "G": 0.27273, "chunks": 188.31, "bytes_per_chunk": 5.438, "boundary_entropy": 0.39079, "router_noise": 0.0, "grad_norm": 0.7364, "lr": 0.0002088, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 153.0}
{"event": "probe", "step": 2600, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3437.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.197495, "cps": 26.851562, "mean_p": 0.231372, "boundary_entropy": 0.283643, "bpb": 2.197056}, "de": {"chunks": 3973.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.193493, "cps": 31.039062, "mean_p": 0.241427, "boundary_entropy": 0.310304, "bpb": 2.444996}, "vi": {"chunks": 4004.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.217196, "cps": 31.28125, "mean_p": 0.260924, "boundary_entropy": 0.402956, "bpb": 2.045106}, "tr": {"chunks": 3723.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.205657, "cps": 29.085938, "mean_p": 0.245848, "boundary_entropy": 0.324581, "bpb": 2.438321}, "fi": {"chunks": 3793.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.201637, "cps": 29.632812, "mean_p": 0.255778, "boundary_entropy": 0.329969, "bpb": 2.643904}, "ru": {"chunks": 5502.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.290343, "cps": 42.984375, "mean_p": 0.30306, "boundary_entropy": 0.501889, "bpb": 1.599567}, "el": {"chunks": 5060.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.246553, "cps": 39.53125, "mean_p": 0.266319, "boundary_entropy": 0.485727, "bpb": 1.683654}, "hi": {"chunks": 6222.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.369807, "cps": 48.609375, "mean_p": 0.388188, "boundary_entropy": 0.631626, "bpb": 1.301481}, "ta": {"chunks": 7892.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.422145, "cps": 61.65625, "mean_p": 0.382856, "boundary_entropy": 0.621049, "bpb": 1.05683}, "th": {"chunks": 7237.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.449922, "cps": 56.539062, "mean_p": 0.375785, "boundary_entropy": 0.6194, "bpb": 1.241483}, "zh": {"chunks": 2574.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.435828, "cps": 20.109375, "mean_p": 0.300227, "boundary_entropy": 0.489052, "bpb": 3.34417}, "my": {"chunks": 8215.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.433555, "cps": 64.179688, "mean_p": 0.412617, "boundary_entropy": 0.651826, "bpb": 1.172994}}, "bpb": {"en": 2.19706, "de": 2.445, "vi": 2.04511, "tr": 2.43832, "fi": 2.6439, "ru": 1.59957, "el": 1.68365, "hi": 1.30148, "ta": 1.05683, "th": 1.24148, "zh": 3.34417, "my": 1.17299}, "gini_cpc": 0.185909, "gini_cps": 0.195423, "bpic": 6.226976, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 0.979739789143617, "vi": 1.0997539105029672, "tr": 1.0413268650820393, "fi": 1.0209760330186466, "ru": 1.4701307438913638, "el": 1.2484014365624914, "hi": 1.8724900642038544, "ta": 2.1375003530869288, "th": 2.2781488435789323, "zh": 2.2067832961770084, "my": 2.1952742323001027}, "t": 155.9}
{"event": "eval", "step": 2600, "phase": 2, "bpb": {"en": 2.20553, "de": 2.44938, "vi": 2.04913, "tr": 2.44494, "fi": 2.65523, "ru": 1.57629, "el": 1.67347, "hi": 1.2978, "ta": 1.06263, "th": 1.25904, "zh": 3.34766, "my": 1.18159}, "bytes_per_chunk": {"en": 5.133, "de": 5.183, "vi": 5.933, "tr": 5.164, "fi": 4.842, "ru": 6.202, "el": 6.867, "hi": 6.71, "ta": 6.297, "th": 6.234, "zh": 6.402, "my": 6.958}, "bpb_mean": 1.93356, "t": 157.3}
{"event": "train", "step": 2600, "phase": 2, "objective": "global", "lm_loss": 1.59754, "bpb": 2.30476, "chunk_loss": 0.030413, "F": 0.18335, "G": 0.27246, "chunks": 187.75, "bytes_per_chunk": 5.454, "boundary_entropy": 0.3925, "router_noise": 0.0, "grad_norm": 0.7429, "lr": 0.0002053, "router_trainable": true, "sec_per_step": 0.087, "alpha": 0.03, "t": 157.3}
{"event": "train", "step": 2650, "phase": 2, "objective": "global", "lm_loss": 1.49705, "bpb": 2.15979, "chunk_loss": 0.030399, "F": 0.18488, "G": 0.28607, "chunks": 189.31, "bytes_per_chunk": 5.409, "boundary_entropy": 0.41638, "router_noise": 0.0, "grad_norm": 0.6837, "lr": 0.0002017, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 159.3}
{"event": "train", "step": 2700, "phase": 2, "objective": "global", "lm_loss": 1.50888, "bpb": 2.17685, "chunk_loss": 0.030261, "F": 0.18439, "G": 0.27598, "chunks": 188.81, "bytes_per_chunk": 5.423, "boundary_entropy": 0.39254, "router_noise": 0.0, "grad_norm": 0.6863, "lr": 0.0001981, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 161.4}
{"event": "train", "step": 2750, "phase": 2, "objective": "global", "lm_loss": 1.40152, "bpb": 2.02196, "chunk_loss": 0.030501, "F": 0.1915, "G": 0.27195, "chunks": 196.09, "bytes_per_chunk": 5.222, "boundary_entropy": 0.37575, "router_noise": 0.0, "grad_norm": 0.6836, "lr": 0.0001945, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 163.4}
{"event": "probe", "step": 2800, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3540.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.203413, "cps": 27.65625, "mean_p": 0.234419, "boundary_entropy": 0.286784, "bpb": 2.183195}, "de": {"chunks": 4016.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.195588, "cps": 31.375, "mean_p": 0.243332, "boundary_entropy": 0.308378, "bpb": 2.419968}, "vi": {"chunks": 4145.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.224844, "cps": 32.382812, "mean_p": 0.25626, "boundary_entropy": 0.3968, "bpb": 2.027546}, "tr": {"chunks": 3715.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.205215, "cps": 29.023438, "mean_p": 0.245331, "boundary_entropy": 0.319966, "bpb": 2.414998}, "fi": {"chunks": 3728.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.198182, "cps": 29.125, "mean_p": 0.252217, "boundary_entropy": 0.329036, "bpb": 2.621099}, "ru": {"chunks": 5608.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.295937, "cps": 43.8125, "mean_p": 0.294319, "boundary_entropy": 0.488843, "bpb": 1.565027}, "el": {"chunks": 6304.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.307168, "cps": 49.25, "mean_p": 0.27911, "boundary_entropy": 0.473779, "bpb": 1.625166}, "hi": {"chunks": 7343.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.436434, "cps": 57.367188, "mean_p": 0.388251, "boundary_entropy": 0.626384, "bpb": 1.285194}, "ta": {"chunks": 8259.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.441776, "cps": 64.523438, "mean_p": 0.379676, "boundary_entropy": 0.620895, "bpb": 1.075129}, "th": {"chunks": 8283.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.514952, "cps": 64.710938, "mean_p": 0.371645, "boundary_entropy": 0.613148, "bpb": 1.2352}, "zh": {"chunks": 2571.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.43532, "cps": 20.085938, "mean_p": 0.295862, "boundary_entropy": 0.484218, "bpb": 3.343869}, "my": {"chunks": 11308.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.596791, "cps": 88.34375, "mean_p": 0.441623, "boundary_entropy": 0.661481, "bpb": 1.165183}}, "bpb": {"en": 2.1832, "de": 2.41997, "vi": 2.02755, "tr": 2.415, "fi": 2.6211, "ru": 1.56503, "el": 1.62517, "hi": 1.28519, "ta": 1.07513, "th": 1.2352, "zh": 3.34387, "my": 1.16518}, "gini_cpc": 0.220506, "gini_cps": 0.23855, "bpic": 5.576591, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9615284861822267, "vi": 1.105356198829603, "tr": 1.008855869005699, "fi": 0.9742824474320381, "ru": 1.4548547918250525, "el": 1.5100670021315483, "hi": 2.1455533281285417, "ta": 2.1718151322976778, "th": 2.5315555075525675, "zh": 2.1400774564217944, "my": 2.933886318666967}, "t": 166.4}
{"event": "eval", "step": 2800, "phase": 2, "bpb": {"en": 2.19086, "de": 2.42509, "vi": 2.03074, "tr": 2.4221, "fi": 2.63009, "ru": 1.54258, "el": 1.61591, "hi": 1.28165, "ta": 1.08054, "th": 1.2509, "zh": 3.34388, "my": 1.17258}, "bytes_per_chunk": {"en": 5.105, "de": 5.117, "vi": 5.804, "tr": 5.094, "fi": 4.87, "ru": 5.944, "el": 5.672, "hi": 5.78, "ta": 6.161, "th": 5.502, "zh": 6.333, "my": 4.92}, "bpb_mean": 1.91558, "t": 167.7}
{"event": "train", "step": 2800, "phase": 2, "objective": "global", "lm_loss": 1.46951, "bpb": 2.12006, "chunk_loss": 0.030557, "F": 0.18921, "G": 0.27924, "chunks": 193.75, "bytes_per_chunk": 5.285, "boundary_entropy": 0.39907, "router_noise": 0.0, "grad_norm": 0.7836, "lr": 0.0001909, "router_trainable": true, "sec_per_step": 0.086, "alpha": 0.03, "t": 167.7}
{"event": "train", "step": 2850, "phase": 2, "objective": "global", "lm_loss": 1.43211, "bpb": 2.0661, "chunk_loss": 0.030416, "F": 0.18698, "G": 0.28242, "chunks": 191.47, "bytes_per_chunk": 5.348, "boundary_entropy": 0.40046, "router_noise": 0.0, "grad_norm": 0.6976, "lr": 0.0001872, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 169.7}
{"event": "train", "step": 2900, "phase": 2, "objective": "global", "lm_loss": 1.4518, "bpb": 2.09451, "chunk_loss": 0.030454, "F": 0.18967, "G": 0.28237, "chunks": 194.22, "bytes_per_chunk": 5.272, "boundary_entropy": 0.39732, "router_noise": 0.0, "grad_norm": 0.7001, "lr": 0.0001835, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 171.8}
{"event": "train", "step": 2950, "phase": 2, "objective": "global", "lm_loss": 1.35354, "bpb": 1.95275, "chunk_loss": 0.030275, "F": 0.18091, "G": 0.27932, "chunks": 185.25, "bytes_per_chunk": 5.528, "boundary_entropy": 0.4084, "router_noise": 0.0, "grad_norm": 0.6736, "lr": 0.0001799, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 173.8}
{"event": "probe", "step": 3000, "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": 0.222948, "boundary_entropy": 0.267424, "bpb": 2.158188}, "de": {"chunks": 3842.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.187113, "cps": 30.015625, "mean_p": 0.22965, "boundary_entropy": 0.288947, "bpb": 2.392353}, "vi": {"chunks": 3734.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.202549, "cps": 29.171875, "mean_p": 0.241971, "boundary_entropy": 0.382395, "bpb": 2.014767}, "tr": {"chunks": 3541.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.195603, "cps": 27.664062, "mean_p": 0.232049, "boundary_entropy": 0.304824, "bpb": 2.387922}, "fi": {"chunks": 3465.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.184201, "cps": 27.070312, "mean_p": 0.236183, "boundary_entropy": 0.3114, "bpb": 2.57758}, "ru": {"chunks": 5672.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.299314, "cps": 44.3125, "mean_p": 0.2795, "boundary_entropy": 0.461487, "bpb": 1.542815}, "el": {"chunks": 6343.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.309068, "cps": 49.554688, "mean_p": 0.279127, "boundary_entropy": 0.465116, "bpb": 1.594452}, "hi": {"chunks": 7513.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.446538, "cps": 58.695312, "mean_p": 0.405554, "boundary_entropy": 0.633785, "bpb": 1.268094}, "ta": {"chunks": 7894.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.422252, "cps": 61.671875, "mean_p": 0.3846, "boundary_entropy": 0.622655, "bpb": 1.042777}, "th": {"chunks": 8010.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.497979, "cps": 62.578125, "mean_p": 0.384468, "boundary_entropy": 0.622481, "bpb": 1.22722}, "zh": {"chunks": 2555.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.432611, "cps": 19.960938, "mean_p": 0.295871, "boundary_entropy": 0.484107, "bpb": 3.316067}, "my": {"chunks": 8568.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.452185, "cps": 66.9375, "mean_p": 0.419463, "boundary_entropy": 0.650574, "bpb": 1.159403}}, "bpb": {"en": 2.15819, "de": 2.39235, "vi": 2.01477, "tr": 2.38792, "fi": 2.57758, "ru": 1.54282, "el": 1.59445, "hi": 1.26809, "ta": 1.04278, "th": 1.22722, "zh": 3.31607, "my": 1.1594}, "gini_cpc": 0.206163, "gini_cps": 0.217487, "bpic": 5.948894, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9645541981547217, "vi": 1.0441258642829874, "tr": 1.0083169262092402, "fi": 0.9495395044645905, "ru": 1.5429387637709613, "el": 1.5932192623788406, "hi": 2.3018657880467313, "ta": 2.1766737247585013, "th": 2.567042938359161, "zh": 2.2300733309205993, "my": 2.3309757960674613}, "t": 176.8}
{"event": "eval", "step": 3000, "phase": 2, "bpb": {"en": 2.16752, "de": 2.39462, "vi": 2.0147, "tr": 2.39616, "fi": 2.58682, "ru": 1.51912, "el": 1.58546, "hi": 1.26515, "ta": 1.04914, "th": 1.2449, "zh": 3.31584, "my": 1.16772}, "bytes_per_chunk": {"en": 5.264, "de": 5.333, "vi": 6.423, "tr": 5.333, "fi": 5.223, "ru": 5.927, "el": 5.649, "hi": 5.63, "ta": 6.287, "th": 5.683, "zh": 6.361, "my": 6.378}, "bpb_mean": 1.89226, "t": 178.1}
{"event": "train", "step": 3000, "phase": 2, "objective": "global", "lm_loss": 1.4334, "bpb": 2.06797, "chunk_loss": 0.030132, "F": 0.17639, "G": 0.28227, "chunks": 180.62, "bytes_per_chunk": 5.669, "boundary_entropy": 0.41782, "router_noise": 0.0, "grad_norm": 0.7213, "lr": 0.0001761, "router_trainable": true, "sec_per_step": 0.086, "alpha": 0.03, "t": 178.1}
{"event": "train", "step": 3050, "phase": 2, "objective": "global", "lm_loss": 1.39475, "bpb": 2.0122, "chunk_loss": 0.030562, "F": 0.18726, "G": 0.27944, "chunks": 191.75, "bytes_per_chunk": 5.34, "boundary_entropy": 0.4017, "router_noise": 0.0, "grad_norm": 0.6528, "lr": 0.0001724, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 180.2}
{"event": "train", "step": 3100, "phase": 2, "objective": "global", "lm_loss": 1.4057, "bpb": 2.02799, "chunk_loss": 0.030358, "F": 0.18631, "G": 0.26271, "chunks": 190.78, "bytes_per_chunk": 5.367, "boundary_entropy": 0.36999, "router_noise": 0.0, "grad_norm": 0.7294, "lr": 0.0001687, "router_trainable": true, "sec_per_step": 0.042, "alpha": 0.03, "t": 182.3}
{"event": "train", "step": 3150, "phase": 2, "objective": "global", "lm_loss": 1.35164, "bpb": 1.95001, "chunk_loss": 0.030073, "F": 0.17581, "G": 0.27017, "chunks": 180.03, "bytes_per_chunk": 5.688, "boundary_entropy": 0.39662, "router_noise": 0.0, "grad_norm": 0.6825, "lr": 0.000165, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 184.3}
{"event": "probe", "step": 3200, "phase": 2, "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": 0.233807, "boundary_entropy": 0.271398, "bpb": 2.140222}, "de": {"chunks": 4006.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.195101, "cps": 31.296875, "mean_p": 0.234982, "boundary_entropy": 0.289976, "bpb": 2.360716}, "vi": {"chunks": 3738.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.202766, "cps": 29.203125, "mean_p": 0.236171, "boundary_entropy": 0.372852, "bpb": 2.000106}, "tr": {"chunks": 3644.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.201293, "cps": 28.46875, "mean_p": 0.234613, "boundary_entropy": 0.296722, "bpb": 2.370012}, "fi": {"chunks": 3675.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.195364, "cps": 28.710938, "mean_p": 0.245093, "boundary_entropy": 0.308298, "bpb": 2.545447}, "ru": {"chunks": 5795.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.305805, "cps": 45.273438, "mean_p": 0.274455, "boundary_entropy": 0.448827, "bpb": 1.510816}, "el": {"chunks": 6321.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.307996, "cps": 49.382812, "mean_p": 0.278605, "boundary_entropy": 0.461314, "bpb": 1.567779}, "hi": {"chunks": 6920.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.411293, "cps": 54.0625, "mean_p": 0.398593, "boundary_entropy": 0.630162, "bpb": 1.263584}, "ta": {"chunks": 8060.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.431131, "cps": 62.96875, "mean_p": 0.382357, "boundary_entropy": 0.617841, "bpb": 1.020069}, "th": {"chunks": 8609.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.535219, "cps": 67.257812, "mean_p": 0.386586, "boundary_entropy": 0.621405, "bpb": 1.215385}, "zh": {"chunks": 2778.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.470369, "cps": 21.703125, "mean_p": 0.292852, "boundary_entropy": 0.466891, "bpb": 3.272532}, "my": {"chunks": 8734.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.460946, "cps": 68.234375, "mean_p": 0.413475, "boundary_entropy": 0.646121, "bpb": 1.142776}}, "bpb": {"en": 2.14022, "de": 2.36072, "vi": 2.00011, "tr": 2.37001, "fi": 2.54545, "ru": 1.51082, "el": 1.56778, "hi": 1.26358, "ta": 1.02007, "th": 1.21539, "zh": 3.27253, "my": 1.14278}, "gini_cpc": 0.208357, "gini_cps": 0.212522, "bpic": 5.83112, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9602192354300868, "vi": 0.9979482454557838, "tr": 0.9906943375551114, "fi": 0.9615177897259016, "ru": 1.5050678882269366, "el": 1.515852027732708, "hi": 2.0242441152938486, "ta": 2.12188301370294, "th": 2.6341682232540085, "zh": 2.3149982263607987, "my": 2.2686195763685704}, "t": 187.3}
{"event": "eval", "step": 3200, "phase": 2, "bpb": {"en": 2.14718, "de": 2.36446, "vi": 2.00253, "tr": 2.37491, "fi": 2.55531, "ru": 1.48809, "el": 1.56075, "hi": 1.25814, "ta": 1.028, "th": 1.23106, "zh": 3.27519, "my": 1.15019}, "bytes_per_chunk": {"en": 5.01, "de": 5.15, "vi": 6.337, "tr": 5.253, "fi": 4.931, "ru": 5.658, "el": 5.685, "hi": 5.921, "ta": 6.178, "th": 5.306, "zh": 5.95, "my": 6.272}, "bpb_mean": 1.86965, "t": 188.6}
{"event": "train", "step": 3200, "phase": 2, "objective": "global", "lm_loss": 1.37544, "bpb": 1.98433, "chunk_loss": 0.03035, "F": 0.18246, "G": 0.27668, "chunks": 186.84, "bytes_per_chunk": 5.481, "boundary_entropy": 0.4129, "router_noise": 0.0, "grad_norm": 0.6419, "lr": 0.0001613, "router_trainable": true, "sec_per_step": 0.087, "alpha": 0.03, "t": 188.7}
{"event": "train", "step": 3250, "phase": 2, "objective": "global", "lm_loss": 1.44221, "bpb": 2.08068, "chunk_loss": 0.03052, "F": 0.18771, "G": 0.2713, "chunks": 192.22, "bytes_per_chunk": 5.327, "boundary_entropy": 0.37857, "router_noise": 0.0, "grad_norm": 0.6652, "lr": 0.0001576, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 190.7}
{"event": "train", "step": 3300, "phase": 2, "objective": "global", "lm_loss": 1.4038, "bpb": 2.02525, "chunk_loss": 0.030374, "F": 0.18741, "G": 0.26193, "chunks": 191.91, "bytes_per_chunk": 5.336, "boundary_entropy": 0.36103, "router_noise": 0.0, "grad_norm": 0.7555, "lr": 0.0001539, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 192.8}
{"event": "train", "step": 3350, "phase": 2, "objective": "global", "lm_loss": 1.47037, "bpb": 2.1213, "chunk_loss": 0.030169, "F": 0.18063, "G": 0.27024, "chunks": 184.97, "bytes_per_chunk": 5.536, "boundary_entropy": 0.38431, "router_noise": 0.0, "grad_norm": 0.712, "lr": 0.0001501, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 194.8}
{"event": "probe", "step": 3400, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3463.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.198989, "cps": 27.054688, "mean_p": 0.224367, "boundary_entropy": 0.25689, "bpb": 2.125391}, "de": {"chunks": 3903.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.190084, "cps": 30.492188, "mean_p": 0.232067, "boundary_entropy": 0.279504, "bpb": 2.342919}, "vi": {"chunks": 3887.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.210849, "cps": 30.367188, "mean_p": 0.243365, "boundary_entropy": 0.367262, "bpb": 1.982458}, "tr": {"chunks": 3497.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.193172, "cps": 27.320312, "mean_p": 0.227922, "boundary_entropy": 0.291157, "bpb": 2.338479}, "fi": {"chunks": 3561.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.189304, "cps": 27.820312, "mean_p": 0.235913, "boundary_entropy": 0.299106, "bpb": 2.538612}, "ru": {"chunks": 5700.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.300792, "cps": 44.53125, "mean_p": 0.262033, "boundary_entropy": 0.432388, "bpb": 1.501436}, "el": {"chunks": 5650.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.275301, "cps": 44.140625, "mean_p": 0.252878, "boundary_entropy": 0.436963, "bpb": 1.571574}, "hi": {"chunks": 6092.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.36208, "cps": 47.59375, "mean_p": 0.376667, "boundary_entropy": 0.62039, "bpb": 1.247951}, "ta": {"chunks": 7633.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.408291, "cps": 59.632812, "mean_p": 0.380366, "boundary_entropy": 0.620135, "bpb": 1.025499}, "th": {"chunks": 6988.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.434442, "cps": 54.59375, "mean_p": 0.380389, "boundary_entropy": 0.622778, "bpb": 1.187483}, "zh": {"chunks": 2850.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.48256, "cps": 22.265625, "mean_p": 0.294001, "boundary_entropy": 0.466746, "bpb": 3.251675}, "my": {"chunks": 7632.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.402787, "cps": 59.625, "mean_p": 0.401197, "boundary_entropy": 0.638406, "bpb": 1.132885}}, "bpb": {"en": 2.12539, "de": 2.34292, "vi": 1.98246, "tr": 2.33848, "fi": 2.53861, "ru": 1.50144, "el": 1.57157, "hi": 1.24795, "ta": 1.0255, "th": 1.18748, "zh": 3.25168, "my": 1.13289}, "gini_cpc": 0.192515, "gini_cps": 0.184057, "bpic": 6.306379, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9552515977639454, "vi": 1.0596026294006122, "tr": 0.9707708109572698, "fi": 0.951331153384013, "ru": 1.511601346156923, "el": 1.3835002161003473, "hi": 1.8196021881070057, "ta": 2.0518302179993646, "th": 2.1832499541536436, "zh": 2.4250631146015653, "my": 2.024168277074372}, "t": 197.7}
{"event": "eval", "step": 3400, "phase": 2, "bpb": {"en": 2.13403, "de": 2.34929, "vi": 1.98467, "tr": 2.34343, "fi": 2.54878, "ru": 1.47836, "el": 1.56352, "hi": 1.24467, "ta": 1.03294, "th": 1.20399, "zh": 3.25459, "my": 1.14043}, "bytes_per_chunk": {"en": 5.174, "de": 5.239, "vi": 6.187, "tr": 5.464, "fi": 5.142, "ru": 5.835, "el": 6.232, "hi": 6.807, "ta": 6.656, "th": 6.509, "zh": 5.675, "my": 7.199}, "bpb_mean": 1.85656, "t": 199.1}
{"event": "train", "step": 3400, "phase": 2, "objective": "global", "lm_loss": 1.39411, "bpb": 2.01128, "chunk_loss": 0.030102, "F": 0.17615, "G": 0.27604, "chunks": 180.38, "bytes_per_chunk": 5.677, "boundary_entropy": 0.40897, "router_noise": 0.0, "grad_norm": 0.6959, "lr": 0.0001465, "router_trainable": true, "sec_per_step": 0.087, "alpha": 0.03, "t": 199.1}
{"event": "train", "step": 3450, "phase": 2, "objective": "global", "lm_loss": 1.25426, "bpb": 1.80952, "chunk_loss": 0.029983, "F": 0.17542, "G": 0.27043, "chunks": 179.62, "bytes_per_chunk": 5.701, "boundary_entropy": 0.39152, "router_noise": 0.0, "grad_norm": 0.6308, "lr": 0.0001428, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 201.1}
{"event": "train", "step": 3500, "phase": 2, "objective": "global", "lm_loss": 1.44254, "bpb": 2.08115, "chunk_loss": 0.030245, "F": 0.185, "G": 0.27188, "chunks": 189.44, "bytes_per_chunk": 5.405, "boundary_entropy": 0.37179, "router_noise": 0.0, "grad_norm": 0.791, "lr": 0.0001391, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 203.2}
{"event": "train", "step": 3550, "phase": 2, "objective": "global", "lm_loss": 1.37328, "bpb": 1.98122, "chunk_loss": 0.030319, "F": 0.18179, "G": 0.27537, "chunks": 186.16, "bytes_per_chunk": 5.501, "boundary_entropy": 0.39345, "router_noise": 0.0, "grad_norm": 0.6553, "lr": 0.0001355, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 205.2}
{"event": "probe", "step": 3600, "phase": 2, "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": 0.23141, "boundary_entropy": 0.258597, "bpb": 2.107288}, "de": {"chunks": 4132.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.201237, "cps": 32.28125, "mean_p": 0.242382, "boundary_entropy": 0.28188, "bpb": 2.326495}, "vi": {"chunks": 3801.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.206184, "cps": 29.695312, "mean_p": 0.242578, "boundary_entropy": 0.365703, "bpb": 1.982614}, "tr": {"chunks": 3644.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.201293, "cps": 28.46875, "mean_p": 0.233439, "boundary_entropy": 0.285858, "bpb": 2.31061}, "fi": {"chunks": 3661.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.19462, "cps": 28.601562, "mean_p": 0.240432, "boundary_entropy": 0.296776, "bpb": 2.48962}, "ru": {"chunks": 5543.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.292507, "cps": 43.304688, "mean_p": 0.260319, "boundary_entropy": 0.430984, "bpb": 1.48367}, "el": {"chunks": 6370.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.310383, "cps": 49.765625, "mean_p": 0.269522, "boundary_entropy": 0.44082, "bpb": 1.543294}, "hi": {"chunks": 7507.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.446181, "cps": 58.648438, "mean_p": 0.396079, "boundary_entropy": 0.625435, "bpb": 1.222972}, "ta": {"chunks": 7136.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.381706, "cps": 55.75, "mean_p": 0.360528, "boundary_entropy": 0.6078, "bpb": 1.013202}, "th": {"chunks": 7053.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.438483, "cps": 55.101562, "mean_p": 0.378022, "boundary_entropy": 0.622265, "bpb": 1.173041}, "zh": {"chunks": 2669.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.451913, "cps": 20.851562, "mean_p": 0.291495, "boundary_entropy": 0.470594, "bpb": 3.260668}, "my": {"chunks": 7681.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.405373, "cps": 60.007812, "mean_p": 0.401942, "boundary_entropy": 0.639967, "bpb": 1.131544}}, "bpb": {"en": 2.10729, "de": 2.32649, "vi": 1.98261, "tr": 2.31061, "fi": 2.48962, "ru": 1.48367, "el": 1.54329, "hi": 1.22297, "ta": 1.0132, "th": 1.17304, "zh": 3.26067, "my": 1.13154}, "gini_cpc": 0.183634, "gini_cps": 0.186913, "bpic": 6.111746, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.973624711285769, "vi": 0.997558583875497, "tr": 0.9738935717528147, "fi": 0.9416110097240941, "ru": 1.4152049751221887, "el": 1.5016968492438172, "hi": 2.1587135887137254, "ta": 1.8467710343446142, "th": 2.121468076572981, "zh": 2.1864462907485556, "my": 1.9612730984215085}, "t": 208.2}
{"event": "eval", "step": 3600, "phase": 2, "bpb": {"en": 2.11655, "de": 2.33422, "vi": 1.98506, "tr": 2.31904, "fi": 2.50199, "ru": 1.45938, "el": 1.53307, "hi": 1.21903, "ta": 1.01836, "th": 1.19253, "zh": 3.26042, "my": 1.1389}, "bytes_per_chunk": {"en": 4.983, "de": 5.003, "vi": 6.297, "tr": 5.207, "fi": 4.971, "ru": 5.96, "el": 5.659, "hi": 5.615, "ta": 7.202, "th": 6.457, "zh": 6.121, "my": 7.243}, "bpb_mean": 1.83988, "t": 209.5}
{"event": "train", "step": 3600, "phase": 2, "objective": "global", "lm_loss": 1.63951, "bpb": 2.36531, "chunk_loss": 0.030404, "F": 0.18683, "G": 0.27275, "chunks": 191.31, "bytes_per_chunk": 5.352, "boundary_entropy": 0.3837, "router_noise": 0.0, "grad_norm": 0.787, "lr": 0.0001319, "router_trainable": true, "sec_per_step": 0.088, "alpha": 0.03, "t": 209.5}
{"event": "train", "step": 3650, "phase": 2, "objective": "global", "lm_loss": 1.43706, "bpb": 2.07324, "chunk_loss": 0.030428, "F": 0.1889, "G": 0.26906, "chunks": 193.44, "bytes_per_chunk": 5.294, "boundary_entropy": 0.37609, "router_noise": 0.0, "grad_norm": 0.7687, "lr": 0.0001283, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 211.6}
{"event": "train", "step": 3700, "phase": 2, "objective": "global", "lm_loss": 1.35089, "bpb": 1.94892, "chunk_loss": 0.030209, "F": 0.17825, "G": 0.27174, "chunks": 182.53, "bytes_per_chunk": 5.61, "boundary_entropy": 0.39927, "router_noise": 0.0, "grad_norm": 0.7134, "lr": 0.0001247, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 213.6}
{"event": "train", "step": 3750, "phase": 2, "objective": "global", "lm_loss": 1.39269, "bpb": 2.00923, "chunk_loss": 0.030204, "F": 0.18158, "G": 0.26917, "chunks": 185.94, "bytes_per_chunk": 5.507, "boundary_entropy": 0.38106, "router_noise": 0.0, "grad_norm": 0.7007, "lr": 0.0001212, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 215.6}
{"event": "probe", "step": 3800, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3490.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.20054, "cps": 27.265625, "mean_p": 0.225829, "boundary_entropy": 0.250457, "bpb": 2.098707}, "de": {"chunks": 3965.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.193104, "cps": 30.976562, "mean_p": 0.230028, "boundary_entropy": 0.270261, "bpb": 2.301435}, "vi": {"chunks": 3964.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.215026, "cps": 30.96875, "mean_p": 0.247481, "boundary_entropy": 0.36842, "bpb": 1.965167}, "tr": {"chunks": 3425.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.189195, "cps": 26.757812, "mean_p": 0.221565, "boundary_entropy": 0.279642, "bpb": 2.299605}, "fi": {"chunks": 3534.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.187869, "cps": 27.609375, "mean_p": 0.231976, "boundary_entropy": 0.2924, "bpb": 2.477489}, "ru": {"chunks": 5861.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.309288, "cps": 45.789062, "mean_p": 0.255997, "boundary_entropy": 0.411716, "bpb": 1.465971}, "el": {"chunks": 6081.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.296302, "cps": 47.507812, "mean_p": 0.262024, "boundary_entropy": 0.437901, "bpb": 1.523638}, "hi": {"chunks": 7598.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.45159, "cps": 59.359375, "mean_p": 0.39922, "boundary_entropy": 0.626175, "bpb": 1.213056}, "ta": {"chunks": 8473.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.453223, "cps": 66.195312, "mean_p": 0.389991, "boundary_entropy": 0.625336, "bpb": 0.995723}, "th": {"chunks": 7729.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.48051, "cps": 60.382812, "mean_p": 0.389603, "boundary_entropy": 0.626959, "bpb": 1.160691}, "zh": {"chunks": 2636.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.446326, "cps": 20.59375, "mean_p": 0.28478, "boundary_entropy": 0.460487, "bpb": 3.218099}, "my": {"chunks": 8793.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.46406, "cps": 68.695312, "mean_p": 0.424644, "boundary_entropy": 0.648852, "bpb": 1.122271}}, "bpb": {"en": 2.09871, "de": 2.30144, "vi": 1.96517, "tr": 2.29961, "fi": 2.47749, "ru": 1.46597, "el": 1.52364, "hi": 1.21306, "ta": 0.99572, "th": 1.16069, "zh": 3.2181, "my": 1.12227}, "gini_cpc": 0.20315, "gini_cps": 0.217902, "bpic": 5.854872, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9629183826943196, "vi": 1.072233068560411, "tr": 0.9434279046658669, "fi": 0.936813962567473, "ru": 1.5422728035623832, "el": 1.4775182418160517, "hi": 2.251867896814077, "ta": 2.260010360859859, "th": 2.3960779098859657, "zh": 2.225618161124086, "my": 2.3140481444314798}, "t": 218.5}
{"event": "eval", "step": 3800, "phase": 2, "bpb": {"en": 2.10881, "de": 2.30727, "vi": 1.9658, "tr": 2.30767, "fi": 2.48841, "ru": 1.44225, "el": 1.51502, "hi": 1.20684, "ta": 1.00316, "th": 1.17637, "zh": 3.21739, "my": 1.1286}, "bytes_per_chunk": {"en": 5.103, "de": 5.233, "vi": 6.074, "tr": 5.594, "fi": 5.215, "ru": 5.691, "el": 5.878, "hi": 5.536, "ta": 6.008, "th": 6.088, "zh": 6.309, "my": 6.292}, "bpb_mean": 1.8223, "t": 219.9}
{"event": "train", "step": 3800, "phase": 2, "objective": "global", "lm_loss": 1.4651, "bpb": 2.11369, "chunk_loss": 0.030316, "F": 0.18069, "G": 0.27306, "chunks": 185.03, "bytes_per_chunk": 5.534, "boundary_entropy": 0.39784, "router_noise": 0.0, "grad_norm": 0.7225, "lr": 0.0001177, "router_trainable": true, "sec_per_step": 0.087, "alpha": 0.03, "t": 219.9}
{"event": "train", "step": 3850, "phase": 2, "objective": "global", "lm_loss": 1.46825, "bpb": 2.11824, "chunk_loss": 0.030212, "F": 0.1796, "G": 0.27789, "chunks": 183.91, "bytes_per_chunk": 5.568, "boundary_entropy": 0.39851, "router_noise": 0.0, "grad_norm": 0.7267, "lr": 0.0001142, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 222.0}
{"event": "train", "step": 3900, "phase": 2, "objective": "global", "lm_loss": 1.47055, "bpb": 2.12156, "chunk_loss": 0.030218, "F": 0.18573, "G": 0.25553, "chunks": 190.19, "bytes_per_chunk": 5.384, "boundary_entropy": 0.33665, "router_noise": 0.0, "grad_norm": 0.744, "lr": 0.0001108, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 224.0}
{"event": "train", "step": 3950, "phase": 2, "objective": "global", "lm_loss": 1.54233, "bpb": 2.22511, "chunk_loss": 0.030696, "F": 0.19174, "G": 0.28291, "chunks": 196.34, "bytes_per_chunk": 5.215, "boundary_entropy": 0.40167, "router_noise": 0.0, "grad_norm": 0.707, "lr": 0.0001074, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 226.0}
{"event": "probe", "step": 4000, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3543.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.203586, "cps": 27.679688, "mean_p": 0.228717, "boundary_entropy": 0.246941, "bpb": 2.080576}, "de": {"chunks": 4072.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.198315, "cps": 31.8125, "mean_p": 0.235932, "boundary_entropy": 0.269421, "bpb": 2.288447}, "vi": {"chunks": 4038.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.21904, "cps": 31.546875, "mean_p": 0.250769, "boundary_entropy": 0.371103, "bpb": 1.951634}, "tr": {"chunks": 3591.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.198365, "cps": 28.054688, "mean_p": 0.231147, "boundary_entropy": 0.282471, "bpb": 2.286398}, "fi": {"chunks": 3725.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.198022, "cps": 29.101562, "mean_p": 0.240702, "boundary_entropy": 0.292067, "bpb": 2.450146}, "ru": {"chunks": 5828.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.307546, "cps": 45.53125, "mean_p": 0.245804, "boundary_entropy": 0.39611, "bpb": 1.454303}, "el": {"chunks": 6038.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.294207, "cps": 47.171875, "mean_p": 0.259577, "boundary_entropy": 0.431585, "bpb": 1.503072}, "hi": {"chunks": 5916.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.35162, "cps": 46.21875, "mean_p": 0.381422, "boundary_entropy": 0.623451, "bpb": 1.220138}, "ta": {"chunks": 7647.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.40904, "cps": 59.742188, "mean_p": 0.379992, "boundary_entropy": 0.622784, "bpb": 0.99319}, "th": {"chunks": 7363.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.457756, "cps": 57.523438, "mean_p": 0.379553, "boundary_entropy": 0.620234, "bpb": 1.148428}, "zh": {"chunks": 2896.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.490349, "cps": 22.625, "mean_p": 0.293383, "boundary_entropy": 0.457552, "bpb": 3.196966}, "my": {"chunks": 9213.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.486225, "cps": 71.976562, "mean_p": 0.426904, "boundary_entropy": 0.649212, "bpb": 1.113115}}, "bpb": {"en": 2.08058, "de": 2.28845, "vi": 1.95163, "tr": 2.2864, "fi": 2.45015, "ru": 1.4543, "el": 1.50307, "hi": 1.22014, "ta": 0.99319, "th": 1.14843, "zh": 3.19697, "my": 1.11312}, "gini_cpc": 0.19797, "gini_cps": 0.198035, "bpic": 6.008783, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.974110736344662, "vi": 1.075910486692953, "tr": 0.9743563538455717, "fi": 0.9726741217542063, "ru": 1.5106480577481183, "el": 1.4451243917369425, "hi": 1.7271341057582097, "ta": 2.009178807086166, "th": 2.248467958508948, "zh": 2.4085634007007326, "my": 2.3883098124784494}, "t": 229.0}
{"event": "eval", "step": 4000, "phase": 2, "bpb": {"en": 2.09078, "de": 2.29479, "vi": 1.95226, "tr": 2.29417, "fi": 2.46154, "ru": 1.42993, "el": 1.49439, "hi": 1.21625, "ta": 1.00096, "th": 1.16363, "zh": 3.19697, "my": 1.12087}, "bytes_per_chunk": {"en": 5.04, "de": 5.062, "vi": 5.972, "tr": 5.318, "fi": 4.923, "ru": 5.676, "el": 5.86, "hi": 6.986, "ta": 6.632, "th": 6.204, "zh": 5.645, "my": 6.266}, "bpb_mean": 1.80971, "t": 230.3}
{"event": "train", "step": 4000, "phase": 2, "objective": "global", "lm_loss": 1.33568, "bpb": 1.92698, "chunk_loss": 0.030228, "F": 0.18491, "G": 0.25877, "chunks": 189.34, "bytes_per_chunk": 5.408, "boundary_entropy": 0.35321, "router_noise": 0.0, "grad_norm": 0.6875, "lr": 0.000104, "router_trainable": true, "sec_per_step": 0.086, "alpha": 0.03, "t": 230.3}
{"event": "train", "step": 4050, "phase": 2, "objective": "global", "lm_loss": 1.53645, "bpb": 2.21663, "chunk_loss": 0.030485, "F": 0.1907, "G": 0.27378, "chunks": 195.28, "bytes_per_chunk": 5.244, "boundary_entropy": 0.37282, "router_noise": 0.0, "grad_norm": 0.7052, "lr": 0.0001007, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 232.4}
{"event": "train", "step": 4100, "phase": 2, "objective": "global", "lm_loss": 1.3965, "bpb": 2.01472, "chunk_loss": 0.030295, "F": 0.18475, "G": 0.25988, "chunks": 189.19, "bytes_per_chunk": 5.413, "boundary_entropy": 0.35267, "router_noise": 0.0, "grad_norm": 0.6993, "lr": 9.75e-05, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 234.4}
{"event": "train", "step": 4150, "phase": 2, "objective": "global", "lm_loss": 1.38424, "bpb": 1.99703, "chunk_loss": 0.030278, "F": 0.18784, "G": 0.25052, "chunks": 192.34, "bytes_per_chunk": 5.324, "boundary_entropy": 0.32066, "router_noise": 0.0, "grad_norm": 0.7512, "lr": 9.43e-05, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 236.4}
{"event": "probe", "step": 4200, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3484.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.200195, "cps": 27.21875, "mean_p": 0.224615, "boundary_entropy": 0.239219, "bpb": 2.07078}, "de": {"chunks": 4007.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.195149, "cps": 31.304688, "mean_p": 0.231049, "boundary_entropy": 0.262159, "bpb": 2.27043}, "vi": {"chunks": 3993.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.216599, "cps": 31.195312, "mean_p": 0.249093, "boundary_entropy": 0.363403, "bpb": 1.944144}, "tr": {"chunks": 3497.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.193172, "cps": 27.320312, "mean_p": 0.222084, "boundary_entropy": 0.268096, "bpb": 2.263845}, "fi": {"chunks": 3585.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.19058, "cps": 28.007812, "mean_p": 0.231733, "boundary_entropy": 0.281639, "bpb": 2.432314}, "ru": {"chunks": 5888.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.310712, "cps": 46.0, "mean_p": 0.248348, "boundary_entropy": 0.396784, "bpb": 1.436588}, "el": {"chunks": 6418.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.312722, "cps": 50.140625, "mean_p": 0.263678, "boundary_entropy": 0.428745, "bpb": 1.490896}, "hi": {"chunks": 6729.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.399941, "cps": 52.570312, "mean_p": 0.38098, "boundary_entropy": 0.616612, "bpb": 1.193571}, "ta": {"chunks": 7861.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.420487, "cps": 61.414062, "mean_p": 0.38283, "boundary_entropy": 0.620973, "bpb": 0.977915}, "th": {"chunks": 7725.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.480261, "cps": 60.351562, "mean_p": 0.387405, "boundary_entropy": 0.62475, "bpb": 1.132036}, "zh": {"chunks": 2661.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.450559, "cps": 20.789062, "mean_p": 0.280984, "boundary_entropy": 0.447598, "bpb": 3.189354}, "my": {"chunks": 9768.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.515516, "cps": 76.3125, "mean_p": 0.433357, "boundary_entropy": 0.648301, "bpb": 1.107053}}, "bpb": {"en": 2.07078, "de": 2.27043, "vi": 1.94414, "tr": 2.26385, "fi": 2.43231, "ru": 1.43659, "el": 1.4909, "hi": 1.19357, "ta": 0.97792, "th": 1.13204, "zh": 3.18935, "my": 1.10705}, "gini_cpc": 0.205861, "gini_cps": 0.218496, "bpic": 5.848894, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9747941386168818, "vi": 1.081937421237058, "tr": 0.964919436953222, "fi": 0.9519699737207039, "ru": 1.5520458999906093, "el": 1.562085645222439, "hi": 1.9977513336278943, "ta": 2.100382062162289, "th": 2.3989621546270223, "zh": 2.2505952906285733, "my": 2.5750653125952057}, "t": 239.4}
{"event": "eval", "step": 4200, "phase": 2, "bpb": {"en": 2.08108, "de": 2.27599, "vi": 1.94578, "tr": 2.27171, "fi": 2.44302, "ru": 1.41325, "el": 1.48245, "hi": 1.18875, "ta": 0.98572, "th": 1.14734, "zh": 3.18836, "my": 1.11463}, "bytes_per_chunk": {"en": 5.075, "de": 5.129, "vi": 6.186, "tr": 5.449, "fi": 5.081, "ru": 5.644, "el": 5.596, "hi": 6.133, "ta": 6.411, "th": 5.905, "zh": 6.14, "my": 5.763}, "bpb_mean": 1.79484, "t": 240.7}
{"event": "phase_transition", "step": 4200, "phase": 3, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0, "t": 240.7}
{"event": "train", "step": 4200, "phase": 3, "objective": "global", "lm_loss": 1.19006, "bpb": 1.71689, "chunk_loss": 0.030232, "F": 0.1788, "G": 0.29705, "chunks": 183.09, "bytes_per_chunk": 5.593, "boundary_entropy": 0.42936, "router_noise": 0.0, "grad_norm": 0.628, "lr": 9.12e-05, "router_trainable": true, "sec_per_step": 0.087, "alpha": 0.03, "t": 240.8}
{"event": "train", "step": 4250, "phase": 3, "objective": "global", "lm_loss": 1.24846, "bpb": 1.80115, "chunk_loss": 0.030206, "F": 0.18137, "G": 0.26786, "chunks": 185.72, "bytes_per_chunk": 5.514, "boundary_entropy": 0.37199, "router_noise": 0.0, "grad_norm": 0.7095, "lr": 8.81e-05, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 242.8}
{"event": "train", "step": 4300, "phase": 3, "objective": "global", "lm_loss": 1.47779, "bpb": 2.132, "chunk_loss": 0.030291, "F": 0.18527, "G": 0.25928, "chunks": 189.72, "bytes_per_chunk": 5.397, "boundary_entropy": 0.34115, "router_noise": 0.0, "grad_norm": 0.7232, "lr": 8.5e-05, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 244.9}
{"event": "train", "step": 4350, "phase": 3, "objective": "global", "lm_loss": 1.46562, "bpb": 2.11444, "chunk_loss": 0.030302, "F": 0.18607, "G": 0.25437, "chunks": 190.53, "bytes_per_chunk": 5.374, "boundary_entropy": 0.3386, "router_noise": 0.0, "grad_norm": 0.7418, "lr": 8.21e-05, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 246.9}
{"event": "probe", "step": 4400, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3488.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.200425, "cps": 27.25, "mean_p": 0.225707, "boundary_entropy": 0.239208, "bpb": 2.055923}, "de": {"chunks": 4066.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.198023, "cps": 31.765625, "mean_p": 0.234285, "boundary_entropy": 0.265494, "bpb": 2.258572}, "vi": {"chunks": 3928.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.213073, "cps": 30.6875, "mean_p": 0.240436, "boundary_entropy": 0.357563, "bpb": 1.935296}, "tr": {"chunks": 3579.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.197702, "cps": 27.960938, "mean_p": 0.230667, "boundary_entropy": 0.27791, "bpb": 2.243511}, "fi": {"chunks": 3656.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.194354, "cps": 28.5625, "mean_p": 0.236923, "boundary_entropy": 0.285897, "bpb": 2.413083}, "ru": {"chunks": 6022.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.317784, "cps": 47.046875, "mean_p": 0.250092, "boundary_entropy": 0.393768, "bpb": 1.421826}, "el": {"chunks": 6019.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.293281, "cps": 47.023438, "mean_p": 0.253258, "boundary_entropy": 0.416095, "bpb": 1.478565}, "hi": {"chunks": 7354.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.437088, "cps": 57.453125, "mean_p": 0.384537, "boundary_entropy": 0.61607, "bpb": 1.179845}, "ta": {"chunks": 8202.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.438727, "cps": 64.078125, "mean_p": 0.387812, "boundary_entropy": 0.624763, "bpb": 0.975542}, "th": {"chunks": 6670.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.414672, "cps": 52.109375, "mean_p": 0.381965, "boundary_entropy": 0.624535, "bpb": 1.130164}, "zh": {"chunks": 2904.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.491703, "cps": 22.6875, "mean_p": 0.286708, "boundary_entropy": 0.446575, "bpb": 3.167813}, "my": {"chunks": 7216.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.380832, "cps": 56.375, "mean_p": 0.410125, "boundary_entropy": 0.643282, "bpb": 1.099634}}, "bpb": {"en": 2.05592, "de": 2.25857, "vi": 1.9353, "tr": 2.24351, "fi": 2.41308, "ru": 1.42183, "el": 1.47857, "hi": 1.17985, "ta": 0.97554, "th": 1.13016, "zh": 3.16781, "my": 1.09963}, "gini_cpc": 0.192194, "gini_cps": 0.187305, "bpic": 6.081722, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9880128910855964, "vi": 1.0631045602824702, "tr": 0.9864130083614303, "fi": 0.9697101637291083, "ru": 1.5855472210791315, "el": 1.4632924846681301, "hi": 2.180801798056082, "ta": 2.1889807349264263, "th": 2.068961516085703, "zh": 2.4533008726936067, "my": 1.9001189639244442}, "t": 249.9}
{"event": "eval", "step": 4400, "phase": 3, "bpb": {"en": 2.06629, "de": 2.26459, "vi": 1.93678, "tr": 2.25056, "fi": 2.42485, "ru": 1.39775, "el": 1.46945, "hi": 1.17537, "ta": 0.98331, "th": 1.1482, "zh": 3.16847, "my": 1.10687}, "bytes_per_chunk": {"en": 5.054, "de": 5.069, "vi": 6.071, "tr": 5.319, "fi": 5.003, "ru": 5.51, "el": 5.888, "hi": 5.719, "ta": 6.293, "th": 6.719, "zh": 5.694, "my": 7.89}, "bpb_mean": 1.78271, "t": 251.2}
{"event": "train", "step": 4400, "phase": 3, "objective": "global", "lm_loss": 1.47683, "bpb": 2.13062, "chunk_loss": 0.030313, "F": 0.18582, "G": 0.26354, "chunks": 190.28, "bytes_per_chunk": 5.382, "boundary_entropy": 0.35493, "router_noise": 0.0, "grad_norm": 0.7555, "lr": 7.92e-05, "router_trainable": true, "sec_per_step": 0.086, "alpha": 0.03, "t": 251.2}
{"event": "train", "step": 4450, "phase": 3, "objective": "global", "lm_loss": 1.49015, "bpb": 2.14983, "chunk_loss": 0.030166, "F": 0.18002, "G": 0.27321, "chunks": 184.34, "bytes_per_chunk": 5.555, "boundary_entropy": 0.38603, "router_noise": 0.0, "grad_norm": 0.7079, "lr": 7.63e-05, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 253.3}
{"event": "train", "step": 4500, "phase": 3, "objective": "global", "lm_loss": 1.41314, "bpb": 2.03873, "chunk_loss": 0.030359, "F": 0.18695, "G": 0.25536, "chunks": 191.44, "bytes_per_chunk": 5.349, "boundary_entropy": 0.32987, "router_noise": 0.0, "grad_norm": 0.7556, "lr": 7.36e-05, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 255.3}
{"event": "train", "step": 4550, "phase": 3, "objective": "global", "lm_loss": 1.22068, "bpb": 1.76107, "chunk_loss": 0.030347, "F": 0.18918, "G": 0.26726, "chunks": 193.72, "bytes_per_chunk": 5.286, "boundary_entropy": 0.36438, "router_noise": 0.0, "grad_norm": 0.7325, "lr": 7.09e-05, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 257.4}
{"event": "probe", "step": 4600, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3544.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.203643, "cps": 27.6875, "mean_p": 0.225858, "boundary_entropy": 0.238711, "bpb": 2.047524}, "de": {"chunks": 4027.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.196123, "cps": 31.460938, "mean_p": 0.23268, "boundary_entropy": 0.260182, "bpb": 2.249674}, "vi": {"chunks": 4145.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.224844, "cps": 32.382812, "mean_p": 0.25413, "boundary_entropy": 0.361973, "bpb": 1.929715}, "tr": {"chunks": 3580.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.197757, "cps": 27.96875, "mean_p": 0.2309, "boundary_entropy": 0.275387, "bpb": 2.23167}, "fi": {"chunks": 3712.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.197331, "cps": 29.0, "mean_p": 0.240394, "boundary_entropy": 0.286605, "bpb": 2.396905}, "ru": {"chunks": 6043.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.318892, "cps": 47.210938, "mean_p": 0.246851, "boundary_entropy": 0.386673, "bpb": 1.414146}, "el": {"chunks": 6334.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.308629, "cps": 49.484375, "mean_p": 0.259649, "boundary_entropy": 0.414347, "bpb": 1.464069}, "hi": {"chunks": 7056.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.419376, "cps": 55.125, "mean_p": 0.373116, "boundary_entropy": 0.609969, "bpb": 1.17691}, "ta": {"chunks": 8007.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.428296, "cps": 62.554688, "mean_p": 0.37393, "boundary_entropy": 0.613021, "bpb": 0.968617}, "th": {"chunks": 7773.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.483245, "cps": 60.726562, "mean_p": 0.385489, "boundary_entropy": 0.623291, "bpb": 1.116563}, "zh": {"chunks": 2739.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.463766, "cps": 21.398438, "mean_p": 0.280399, "boundary_entropy": 0.446037, "bpb": 3.153353}, "my": {"chunks": 8781.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.463426, "cps": 68.601562, "mean_p": 0.422788, "boundary_entropy": 0.6483, "bpb": 1.090505}}, "bpb": {"en": 2.04752, "de": 2.24967, "vi": 1.92971, "tr": 2.23167, "fi": 2.3969, "ru": 1.41415, "el": 1.46407, "hi": 1.17691, "ta": 0.96862, "th": 1.11656, "zh": 3.15335, "my": 1.09051}, "gini_cpc": 0.194884, "gini_cps": 0.204742, "bpic": 5.837772, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9630739357081851, "vi": 1.1041086184697502, "tr": 0.971097603143487, "fi": 0.9690060556038426, "ru": 1.5659352013436811, "el": 1.5155407592582633, "hi": 2.0593677445755167, "ta": 2.1031718775375414, "th": 2.3730014819690264, "zh": 2.2773458850050337, "my": 2.275679032576266}, "t": 260.4}
{"event": "eval", "step": 4600, "phase": 3, "bpb": {"en": 2.05785, "de": 2.2544, "vi": 1.93002, "tr": 2.23885, "fi": 2.40844, "ru": 1.39025, "el": 1.45501, "hi": 1.17221, "ta": 0.97715, "th": 1.13254, "zh": 3.1527, "my": 1.09818}, "bytes_per_chunk": {"en": 5.055, "de": 5.066, "vi": 5.888, "tr": 5.293, "fi": 4.904, "ru": 5.53, "el": 5.632, "hi": 5.828, "ta": 6.389, "th": 5.862, "zh": 6.021, "my": 6.366}, "bpb_mean": 1.7723, "t": 261.7}
{"event": "train", "step": 4600, "phase": 3, "objective": "global", "lm_loss": 1.29804, "bpb": 1.87267, "chunk_loss": 0.030348, "F": 0.18509, "G": 0.28095, "chunks": 189.53, "bytes_per_chunk": 5.403, "boundary_entropy": 0.39732, "router_noise": 0.0, "grad_norm": 0.7215, "lr": 6.82e-05, "router_trainable": true, "sec_per_step": 0.086, "alpha": 0.03, "t": 261.7}
{"event": "train", "step": 4650, "phase": 3, "objective": "global", "lm_loss": 1.40354, "bpb": 2.02488, "chunk_loss": 0.030251, "F": 0.18311, "G": 0.27006, "chunks": 187.5, "bytes_per_chunk": 5.461, "boundary_entropy": 0.37798, "router_noise": 0.0, "grad_norm": 0.6991, "lr": 6.57e-05, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 263.7}
{"event": "train", "step": 4700, "phase": 3, "objective": "global", "lm_loss": 1.32104, "bpb": 1.90586, "chunk_loss": 0.030169, "F": 0.18225, "G": 0.25731, "chunks": 186.62, "bytes_per_chunk": 5.487, "boundary_entropy": 0.34375, "router_noise": 0.0, "grad_norm": 0.782, "lr": 6.32e-05, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 265.8}
{"event": "train", "step": 4750, "phase": 3, "objective": "global", "lm_loss": 1.33202, "bpb": 1.92171, "chunk_loss": 0.030367, "F": 0.18811, "G": 0.26992, "chunks": 192.62, "bytes_per_chunk": 5.316, "boundary_entropy": 0.35056, "router_noise": 0.0, "grad_norm": 0.7321, "lr": 6.08e-05, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 267.8}
{"event": "probe", "step": 4800, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3543.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.203586, "cps": 27.679688, "mean_p": 0.224684, "boundary_entropy": 0.235459, "bpb": 2.040671}, "de": {"chunks": 4063.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.197877, "cps": 31.742188, "mean_p": 0.23209, "boundary_entropy": 0.257178, "bpb": 2.239035}, "vi": {"chunks": 4111.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.223, "cps": 32.117188, "mean_p": 0.250373, "boundary_entropy": 0.357613, "bpb": 1.9253}, "tr": {"chunks": 3601.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.198917, "cps": 28.132812, "mean_p": 0.22817, "boundary_entropy": 0.267896, "bpb": 2.21877}, "fi": {"chunks": 3616.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.192228, "cps": 28.25, "mean_p": 0.233663, "boundary_entropy": 0.279347, "bpb": 2.38577}, "ru": {"chunks": 5872.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.309868, "cps": 45.875, "mean_p": 0.237836, "boundary_entropy": 0.375414, "bpb": 1.401508}, "el": {"chunks": 6307.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.307314, "cps": 49.273438, "mean_p": 0.256798, "boundary_entropy": 0.411349, "bpb": 1.451007}, "hi": {"chunks": 7947.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.472333, "cps": 62.085938, "mean_p": 0.375339, "boundary_entropy": 0.606698, "bpb": 1.166342}, "ta": {"chunks": 8573.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.458572, "cps": 66.976562, "mean_p": 0.378195, "boundary_entropy": 0.615331, "bpb": 0.961873}, "th": {"chunks": 7083.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.440348, "cps": 55.335938, "mean_p": 0.375032, "boundary_entropy": 0.616856, "bpb": 1.116465}, "zh": {"chunks": 2851.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.482729, "cps": 22.273438, "mean_p": 0.280771, "boundary_entropy": 0.436628, "bpb": 3.14322}, "my": {"chunks": 9622.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.507811, "cps": 75.171875, "mean_p": 0.428117, "boundary_entropy": 0.64867, "bpb": 1.085126}}, "bpb": {"en": 2.04067, "de": 2.23904, "vi": 1.9253, "tr": 2.21877, "fi": 2.38577, "ru": 1.40151, "el": 1.45101, "hi": 1.16634, "ta": 0.96187, "th": 1.11647, "zh": 3.14322, "my": 1.08513}, "gini_cpc": 0.206082, "gini_cps": 0.218291, "bpic": 5.711962, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 0.97195774110225, "vi": 1.095361072509839, "tr": 0.9770696825947935, "fi": 0.9442119796679758, "ru": 1.5220530876968001, "el": 1.5095063826904433, "hi": 2.320070104540313, "ta": 2.2524767769255525, "th": 2.162963269064088, "zh": 2.371137519129071, "my": 2.494335940048588}, "t": 270.8}
{"event": "eval", "step": 4800, "phase": 3, "bpb": {"en": 2.05168, "de": 2.24586, "vi": 1.92455, "tr": 2.22426, "fi": 2.39757, "ru": 1.37825, "el": 1.44179, "hi": 1.1612, "ta": 0.96922, "th": 1.13189, "zh": 3.14066, "my": 1.09262}, "bytes_per_chunk": {"en": 5.077, "de": 5.08, "vi": 5.95, "tr": 5.305, "fi": 5.062, "ru": 5.619, "el": 5.666, "hi": 5.331, "ta": 5.953, "th": 6.394, "zh": 5.748, "my": 5.649}, "bpb_mean": 1.7633, "t": 272.1}
{"event": "train", "step": 4800, "phase": 3, "objective": "global", "lm_loss": 1.25522, "bpb": 1.8109, "chunk_loss": 0.030124, "F": 0.17856, "G": 0.26797, "chunks": 182.84, "bytes_per_chunk": 5.6, "boundary_entropy": 0.37247, "router_noise": 0.0, "grad_norm": 0.7166, "lr": 5.85e-05, "router_trainable": true, "sec_per_step": 0.086, "alpha": 0.03, "t": 272.1}
{"event": "train", "step": 4850, "phase": 3, "objective": "global", "lm_loss": 1.48688, "bpb": 2.14511, "chunk_loss": 0.030357, "F": 0.18845, "G": 0.25432, "chunks": 192.97, "bytes_per_chunk": 5.307, "boundary_entropy": 0.33261, "router_noise": 0.0, "grad_norm": 0.7397, "lr": 5.62e-05, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 274.2}
{"event": "train", "step": 4900, "phase": 3, "objective": "global", "lm_loss": 1.39812, "bpb": 2.01706, "chunk_loss": 0.030288, "F": 0.18161, "G": 0.27329, "chunks": 185.97, "bytes_per_chunk": 5.506, "boundary_entropy": 0.39144, "router_noise": 0.0, "grad_norm": 0.6998, "lr": 5.41e-05, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 276.2}
{"event": "train", "step": 4950, "phase": 3, "objective": "global", "lm_loss": 1.51495, "bpb": 2.18561, "chunk_loss": 0.030608, "F": 0.19559, "G": 0.25409, "chunks": 200.28, "bytes_per_chunk": 5.113, "boundary_entropy": 0.32535, "router_noise": 0.0, "grad_norm": 0.7799, "lr": 5.2e-05, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 278.3}
{"event": "probe", "step": 5000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3581.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.205769, "cps": 27.976562, "mean_p": 0.2273, "boundary_entropy": 0.233263, "bpb": 2.030425}, "de": {"chunks": 4120.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.200653, "cps": 32.1875, "mean_p": 0.234639, "boundary_entropy": 0.256816, "bpb": 2.223665}, "vi": {"chunks": 4173.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.226363, "cps": 32.601562, "mean_p": 0.253568, "boundary_entropy": 0.36019, "bpb": 1.918579}, "tr": {"chunks": 3656.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.201955, "cps": 28.5625, "mean_p": 0.232627, "boundary_entropy": 0.271678, "bpb": 2.211198}, "fi": {"chunks": 3721.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.19781, "cps": 29.070312, "mean_p": 0.236264, "boundary_entropy": 0.275702, "bpb": 2.370597}, "ru": {"chunks": 5982.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.315673, "cps": 46.734375, "mean_p": 0.238667, "boundary_entropy": 0.37092, "bpb": 1.39351}, "el": {"chunks": 6459.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.31472, "cps": 50.460938, "mean_p": 0.258448, "boundary_entropy": 0.407739, "bpb": 1.443095}, "hi": {"chunks": 7022.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.417355, "cps": 54.859375, "mean_p": 0.364227, "boundary_entropy": 0.602509, "bpb": 1.166698}, "ta": {"chunks": 8757.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.468414, "cps": 68.414062, "mean_p": 0.379241, "boundary_entropy": 0.616118, "bpb": 0.95551}, "th": {"chunks": 8207.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.510227, "cps": 64.117188, "mean_p": 0.3782, "boundary_entropy": 0.614188, "bpb": 1.102112}, "zh": {"chunks": 2857.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.483745, "cps": 22.320312, "mean_p": 0.27873, "boundary_entropy": 0.434309, "bpb": 3.135252}, "my": {"chunks": 9584.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.505805, "cps": 74.875, "mean_p": 0.431445, "boundary_entropy": 0.649861, "bpb": 1.077962}}, "bpb": {"en": 2.03042, "de": 2.22366, "vi": 1.91858, "tr": 2.2112, "fi": 2.3706, "ru": 1.39351, "el": 1.44309, "hi": 1.1667, "ta": 0.95551, "th": 1.10211, "zh": 3.13525, "my": 1.07796}, "gini_cpc": 0.206716, "gini_cps": 0.217474, "bpic": 5.633979, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9751346933732912, "vi": 1.1000819577332586, "tr": 0.9814663965666198, "fi": 0.9613191322710986, "ru": 1.5341117404301063, "el": 1.529481536186718, "hi": 2.028268992739873, "ta": 2.276405778648552, "th": 2.4796087908092974, "zh": 2.35091321327248, "my": 2.4581208365380656}, "t": 281.3}
{"event": "eval", "step": 5000, "phase": 3, "bpb": {"en": 2.04034, "de": 2.22953, "vi": 1.91884, "tr": 2.21873, "fi": 2.38277, "ru": 1.3685, "el": 1.43356, "hi": 1.16189, "ta": 0.9626, "th": 1.12032, "zh": 3.13404, "my": 1.08611}, "bytes_per_chunk": {"en": 4.998, "de": 4.996, "vi": 5.737, "tr": 5.208, "fi": 4.92, "ru": 5.555, "el": 5.561, "hi": 5.899, "ta": 5.82, "th": 5.611, "zh": 5.8, "my": 5.734}, "bpb_mean": 1.75477, "t": 282.6}
{"event": "train", "step": 5000, "phase": 3, "objective": "global", "lm_loss": 1.2823, "bpb": 1.84997, "chunk_loss": 0.030418, "F": 0.18768, "G": 0.26966, "chunks": 192.19, "bytes_per_chunk": 5.328, "boundary_entropy": 0.37141, "router_noise": 0.0, "grad_norm": 0.7633, "lr": 5e-05, "router_trainable": true, "sec_per_step": 0.087, "alpha": 0.03, "t": 282.6}
{"event": "train", "step": 5050, "phase": 3, "objective": "global", "lm_loss": 1.25827, "bpb": 1.8153, "chunk_loss": 0.030236, "F": 0.18317, "G": 0.25968, "chunks": 187.56, "bytes_per_chunk": 5.46, "boundary_entropy": 0.35059, "router_noise": 0.0, "grad_norm": 0.7364, "lr": 4.81e-05, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 284.6}
{"event": "train", "step": 5100, "phase": 3, "objective": "global", "lm_loss": 1.30579, "bpb": 1.88386, "chunk_loss": 0.030085, "F": 0.18011, "G": 0.26284, "chunks": 184.44, "bytes_per_chunk": 5.552, "boundary_entropy": 0.35437, "router_noise": 0.0, "grad_norm": 0.749, "lr": 4.63e-05, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 286.7}
{"event": "train", "step": 5150, "phase": 3, "objective": "global", "lm_loss": 1.33219, "bpb": 1.92194, "chunk_loss": 0.030334, "F": 0.18701, "G": 0.25028, "chunks": 191.5, "bytes_per_chunk": 5.347, "boundary_entropy": 0.33164, "router_noise": 0.0, "grad_norm": 0.7555, "lr": 4.45e-05, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 288.7}
{"event": "probe", "step": 5200, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3557.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.20439, "cps": 27.789062, "mean_p": 0.225962, "boundary_entropy": 0.229266, "bpb": 2.028237}, "de": {"chunks": 4074.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.198412, "cps": 31.828125, "mean_p": 0.232647, "boundary_entropy": 0.253326, "bpb": 2.217818}, "vi": {"chunks": 3826.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.20754, "cps": 29.890625, "mean_p": 0.243459, "boundary_entropy": 0.354284, "bpb": 1.918225}, "tr": {"chunks": 3585.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.198033, "cps": 28.007812, "mean_p": 0.227545, "boundary_entropy": 0.265676, "bpb": 2.204081}, "fi": {"chunks": 3684.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.195843, "cps": 28.78125, "mean_p": 0.236388, "boundary_entropy": 0.275276, "bpb": 2.358776}, "ru": {"chunks": 6031.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.318259, "cps": 47.117188, "mean_p": 0.238906, "boundary_entropy": 0.368507, "bpb": 1.38461}, "el": {"chunks": 6524.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.317887, "cps": 50.96875, "mean_p": 0.260306, "boundary_entropy": 0.406366, "bpb": 1.436107}, "hi": {"chunks": 7671.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.455929, "cps": 59.929688, "mean_p": 0.352504, "boundary_entropy": 0.589, "bpb": 1.148267}, "ta": {"chunks": 8372.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.44782, "cps": 65.40625, "mean_p": 0.371694, "boundary_entropy": 0.607127, "bpb": 0.950111}, "th": {"chunks": 7882.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.490022, "cps": 61.578125, "mean_p": 0.374634, "boundary_entropy": 0.612221, "bpb": 1.093937}, "zh": {"chunks": 2834.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.479851, "cps": 22.140625, "mean_p": 0.278978, "boundary_entropy": 0.43486, "bpb": 3.120439}, "my": {"chunks": 9721.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.513036, "cps": 75.945312, "mean_p": 0.434813, "boundary_entropy": 0.650162, "bpb": 1.077415}}, "bpb": {"en": 2.02824, "de": 2.21782, "vi": 1.91823, "tr": 2.20408, "fi": 2.35878, "ru": 1.38461, "el": 1.43611, "hi": 1.14827, "ta": 0.95011, "th": 1.09394, "zh": 3.12044, "my": 1.07741}, "gini_cpc": 0.208301, "gini_cps": 0.22033, "bpic": 5.663745, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 0.970753293165588, "vi": 1.0154115025026575, "tr": 0.9688997940454778, "fi": 0.9581819668466612, "ru": 1.5571138555436788, "el": 1.5552971000356424, "hi": 2.230679442122997, "ta": 2.1910082112853875, "th": 2.3974834632818047, "zh": 2.3477219385212296, "my": 2.510081495549811}, "t": 291.7}
{"event": "eval", "step": 5200, "phase": 3, "bpb": {"en": 2.03795, "de": 2.22316, "vi": 1.91903, "tr": 2.21013, "fi": 2.37111, "ru": 1.35974, "el": 1.42712, "hi": 1.14369, "ta": 0.95803, "th": 1.11118, "zh": 3.12011, "my": 1.0854}, "bytes_per_chunk": {"en": 5.034, "de": 5.039, "vi": 6.349, "tr": 5.333, "fi": 4.921, "ru": 5.514, "el": 5.483, "hi": 5.491, "ta": 6.073, "th": 5.803, "zh": 5.88, "my": 5.862}, "bpb_mean": 1.74722, "t": 293.0}
{"event": "train", "step": 5200, "phase": 3, "objective": "global", "lm_loss": 1.28014, "bpb": 1.84685, "chunk_loss": 0.030519, "F": 0.1904, "G": 0.26654, "chunks": 194.97, "bytes_per_chunk": 5.252, "boundary_entropy": 0.35779, "router_noise": 0.0, "grad_norm": 0.7835, "lr": 4.29e-05, "router_trainable": true, "sec_per_step": 0.087, "alpha": 0.03, "t": 293.1}
{"event": "train", "step": 5250, "phase": 3, "objective": "global", "lm_loss": 1.3713, "bpb": 1.97837, "chunk_loss": 0.030477, "F": 0.18997, "G": 0.26764, "chunks": 194.53, "bytes_per_chunk": 5.264, "boundary_entropy": 0.35978, "router_noise": 0.0, "grad_norm": 0.756, "lr": 4.14e-05, "router_trainable": true, "sec_per_step": 0.042, "alpha": 0.03, "t": 295.2}
{"event": "train", "step": 5300, "phase": 3, "objective": "global", "lm_loss": 1.36937, "bpb": 1.97558, "chunk_loss": 0.03048, "F": 0.1925, "G": 0.2641, "chunks": 197.12, "bytes_per_chunk": 5.195, "boundary_entropy": 0.34689, "router_noise": 0.0, "grad_norm": 0.7345, "lr": 3.99e-05, "router_trainable": true, "sec_per_step": 0.042, "alpha": 0.03, "t": 297.2}
{"event": "train", "step": 5350, "phase": 3, "objective": "global", "lm_loss": 1.42128, "bpb": 2.05048, "chunk_loss": 0.030403, "F": 0.18991, "G": 0.25828, "chunks": 194.47, "bytes_per_chunk": 5.266, "boundary_entropy": 0.34149, "router_noise": 0.0, "grad_norm": 0.7664, "lr": 3.86e-05, "router_trainable": true, "sec_per_step": 0.042, "alpha": 0.03, "t": 299.3}
{"event": "probe", "step": 5400, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3559.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.204505, "cps": 27.804688, "mean_p": 0.224304, "boundary_entropy": 0.228266, "bpb": 2.019825}, "de": {"chunks": 4076.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.19851, "cps": 31.84375, "mean_p": 0.232712, "boundary_entropy": 0.251443, "bpb": 2.21049}, "vi": {"chunks": 4061.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.220287, "cps": 31.726562, "mean_p": 0.249721, "boundary_entropy": 0.354869, "bpb": 1.907744}, "tr": {"chunks": 3669.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.202674, "cps": 28.664062, "mean_p": 0.230635, "boundary_entropy": 0.266182, "bpb": 2.196377}, "fi": {"chunks": 3678.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.195524, "cps": 28.734375, "mean_p": 0.235279, "boundary_entropy": 0.274102, "bpb": 2.349617}, "ru": {"chunks": 5932.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.313034, "cps": 46.34375, "mean_p": 0.231674, "boundary_entropy": 0.355917, "bpb": 1.380184}, "el": {"chunks": 6554.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.319349, "cps": 51.203125, "mean_p": 0.258059, "boundary_entropy": 0.400589, "bpb": 1.422979}, "hi": {"chunks": 6952.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.413195, "cps": 54.3125, "mean_p": 0.344148, "boundary_entropy": 0.583307, "bpb": 1.151043}, "ta": {"chunks": 8603.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.460177, "cps": 67.210938, "mean_p": 0.381248, "boundary_entropy": 0.61521, "bpb": 0.948173}, "th": {"chunks": 7328.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.45558, "cps": 57.25, "mean_p": 0.375016, "boundary_entropy": 0.614187, "bpb": 1.089336}, "zh": {"chunks": 2777.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.4702, "cps": 21.695312, "mean_p": 0.275119, "boundary_entropy": 0.430147, "bpb": 3.118217}, "my": {"chunks": 9586.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.505911, "cps": 74.890625, "mean_p": 0.431105, "boundary_entropy": 0.648037, "bpb": 1.070888}}, "bpb": {"en": 2.01982, "de": 2.21049, "vi": 1.90774, "tr": 2.19638, "fi": 2.34962, "ru": 1.38018, "el": 1.42298, "hi": 1.15104, "ta": 0.94817, "th": 1.08934, "zh": 3.11822, "my": 1.07089}, "gini_cpc": 0.20002, "gini_cps": 0.214168, "bpic": 5.747376, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 0.970684065386585, "vi": 1.0771742915141274, "tr": 0.991044812304329, "fi": 0.9560838311337596, "ru": 1.530692873468801, "el": 1.5615709604080878, "hi": 2.0204626322484867, "ta": 2.250197229283824, "th": 2.2277196087831723, "zh": 2.2992096274241947, "my": 2.473831873488454}, "t": 302.4}
{"event": "eval", "step": 5400, "phase": 3, "bpb": {"en": 2.03069, "de": 2.21621, "vi": 1.90805, "tr": 2.20212, "fi": 2.36227, "ru": 1.3555, "el": 1.41474, "hi": 1.14598, "ta": 0.95522, "th": 1.10671, "zh": 3.11654, "my": 1.0794}, "bytes_per_chunk": {"en": 5.055, "de": 5.033, "vi": 6.03, "tr": 5.202, "fi": 4.967, "ru": 5.571, "el": 5.491, "hi": 6.093, "ta": 5.948, "th": 6.184, "zh": 6.012, "my": 5.974}, "bpb_mean": 1.74112, "t": 303.6}
{"event": "train", "step": 5400, "phase": 3, "objective": "global", "lm_loss": 1.12526, "bpb": 1.6234, "chunk_loss": 0.03021, "F": 0.18073, "G": 0.27447, "chunks": 185.06, "bytes_per_chunk": 5.533, "boundary_entropy": 0.39017, "router_noise": 0.0, "grad_norm": 0.7042, "lr": 3.73e-05, "router_trainable": true, "sec_per_step": 0.087, "alpha": 0.03, "t": 303.7}
{"event": "train", "step": 5450, "phase": 3, "objective": "global", "lm_loss": 1.18478, "bpb": 1.70928, "chunk_loss": 0.030258, "F": 0.1843, "G": 0.2676, "chunks": 188.72, "bytes_per_chunk": 5.426, "boundary_entropy": 0.37168, "router_noise": 0.0, "grad_norm": 0.7468, "lr": 3.62e-05, "router_trainable": true, "sec_per_step": 0.042, "alpha": 0.03, "t": 305.8}
{"event": "train", "step": 5500, "phase": 3, "objective": "global", "lm_loss": 1.39154, "bpb": 2.00757, "chunk_loss": 0.030342, "F": 0.18796, "G": 0.26249, "chunks": 192.47, "bytes_per_chunk": 5.32, "boundary_entropy": 0.33475, "router_noise": 0.0, "grad_norm": 0.7735, "lr": 3.51e-05, "router_trainable": true, "sec_per_step": 0.042, "alpha": 0.03, "t": 307.9}
{"event": "train", "step": 5550, "phase": 3, "objective": "global", "lm_loss": 1.40482, "bpb": 2.02673, "chunk_loss": 0.030311, "F": 0.18564, "G": 0.25536, "chunks": 190.09, "bytes_per_chunk": 5.387, "boundary_entropy": 0.33746, "router_noise": 0.0, "grad_norm": 0.7848, "lr": 3.41e-05, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 309.9}
{"event": "probe", "step": 5600, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3603.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.207033, "cps": 28.148438, "mean_p": 0.227444, "boundary_entropy": 0.22743, "bpb": 2.013672}, "de": {"chunks": 4093.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.199338, "cps": 31.976562, "mean_p": 0.232264, "boundary_entropy": 0.248514, "bpb": 2.20324}, "vi": {"chunks": 4213.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.228533, "cps": 32.914062, "mean_p": 0.253403, "boundary_entropy": 0.352344, "bpb": 1.903961}, "tr": {"chunks": 3603.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.199028, "cps": 28.148438, "mean_p": 0.22784, "boundary_entropy": 0.264028, "bpb": 2.185484}, "fi": {"chunks": 3674.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.195311, "cps": 28.703125, "mean_p": 0.233963, "boundary_entropy": 0.272443, "bpb": 2.342554}, "ru": {"chunks": 5937.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.313298, "cps": 46.382812, "mean_p": 0.232072, "boundary_entropy": 0.355854, "bpb": 1.376265}, "el": {"chunks": 6305.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.307216, "cps": 49.257812, "mean_p": 0.25317, "boundary_entropy": 0.397843, "bpb": 1.419768}, "hi": {"chunks": 7102.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.42211, "cps": 55.484375, "mean_p": 0.342341, "boundary_entropy": 0.580154, "bpb": 1.139517}, "ta": {"chunks": 7958.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.425675, "cps": 62.171875, "mean_p": 0.373967, "boundary_entropy": 0.610527, "bpb": 0.942245}, "th": {"chunks": 7084.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.44041, "cps": 55.34375, "mean_p": 0.368126, "boundary_entropy": 0.609403, "bpb": 1.090554}, "zh": {"chunks": 2766.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.468337, "cps": 21.609375, "mean_p": 0.272057, "boundary_entropy": 0.426794, "bpb": 3.110692}, "my": {"chunks": 8946.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.472134, "cps": 69.890625, "mean_p": 0.425878, "boundary_entropy": 0.648082, "bpb": 1.069489}}, "bpb": {"en": 2.01367, "de": 2.20324, "vi": 1.90396, "tr": 2.18548, "fi": 2.34255, "ru": 1.37626, "el": 1.41977, "hi": 1.13952, "ta": 0.94225, "th": 1.09055, "zh": 3.11069, "my": 1.06949}, "gini_cpc": 0.189862, "gini_cps": 0.200506, "bpic": 5.878638, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9628290731440905, "vi": 1.103845204486723, "tr": 0.9613323758493066, "fi": 0.9433810031342494, "ru": 1.513274426104895, "el": 1.4838981856752338, "hi": 2.0388508338148124, "ta": 2.0560720216437267, "th": 2.1272441859532143, "zh": 2.262135374827332, "my": 2.280475316379268}, "t": 312.9}
{"event": "eval", "step": 5600, "phase": 3, "bpb": {"en": 2.02509, "de": 2.20915, "vi": 1.90339, "tr": 2.19172, "fi": 2.35554, "ru": 1.35074, "el": 1.41053, "hi": 1.13524, "ta": 0.94973, "th": 1.10788, "zh": 3.10904, "my": 1.07773}, "bytes_per_chunk": {"en": 4.984, "de": 5.015, "vi": 5.84, "tr": 5.301, "fi": 4.992, "ru": 5.576, "el": 5.667, "hi": 5.936, "ta": 6.341, "th": 6.347, "zh": 6.029, "my": 6.345}, "bpb_mean": 1.73548, "t": 314.2}
{"event": "train", "step": 5600, "phase": 3, "objective": "global", "lm_loss": 1.46273, "bpb": 2.11028, "chunk_loss": 0.030319, "F": 0.1871, "G": 0.24503, "chunks": 191.59, "bytes_per_chunk": 5.345, "boundary_entropy": 0.3202, "router_noise": 0.0, "grad_norm": 0.7677, "lr": 3.33e-05, "router_trainable": true, "sec_per_step": 0.086, "alpha": 0.03, "t": 314.2}
{"event": "train", "step": 5650, "phase": 3, "objective": "global", "lm_loss": 1.26587, "bpb": 1.82626, "chunk_loss": 0.030389, "F": 0.18863, "G": 0.26659, "chunks": 193.16, "bytes_per_chunk": 5.301, "boundary_entropy": 0.36353, "router_noise": 0.0, "grad_norm": 0.7219, "lr": 3.25e-05, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 316.3}
{"event": "train", "step": 5700, "phase": 3, "objective": "global", "lm_loss": 1.31713, "bpb": 1.90021, "chunk_loss": 0.03042, "F": 0.19067, "G": 0.253, "chunks": 195.25, "bytes_per_chunk": 5.245, "boundary_entropy": 0.32804, "router_noise": 0.0, "grad_norm": 0.7741, "lr": 3.18e-05, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 318.3}
{"event": "train", "step": 5750, "phase": 3, "objective": "global", "lm_loss": 1.2006, "bpb": 1.7321, "chunk_loss": 0.030248, "F": 0.18466, "G": 0.2588, "chunks": 189.09, "bytes_per_chunk": 5.415, "boundary_entropy": 0.35063, "router_noise": 0.0, "grad_norm": 0.7249, "lr": 3.13e-05, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 320.3}
{"event": "probe", "step": 5800, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3591.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.206344, "cps": 28.054688, "mean_p": 0.225406, "boundary_entropy": 0.225429, "bpb": 2.009501}, "de": {"chunks": 4062.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.197828, "cps": 31.734375, "mean_p": 0.229742, "boundary_entropy": 0.245336, "bpb": 2.20128}, "vi": {"chunks": 4081.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.221372, "cps": 31.882812, "mean_p": 0.251544, "boundary_entropy": 0.354061, "bpb": 1.900291}, "tr": {"chunks": 3609.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.199359, "cps": 28.195312, "mean_p": 0.227533, "boundary_entropy": 0.261972, "bpb": 2.178575}, "fi": {"chunks": 3652.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.194142, "cps": 28.53125, "mean_p": 0.23168, "boundary_entropy": 0.268343, "bpb": 2.335948}, "ru": {"chunks": 5877.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.310132, "cps": 45.914062, "mean_p": 0.229668, "boundary_entropy": 0.35017, "bpb": 1.371869}, "el": {"chunks": 6339.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.308873, "cps": 49.523438, "mean_p": 0.252692, "boundary_entropy": 0.398266, "bpb": 1.412486}, "hi": {"chunks": 8004.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.475721, "cps": 62.53125, "mean_p": 0.347728, "boundary_entropy": 0.579257, "bpb": 1.132183}, "ta": {"chunks": 8398.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.449211, "cps": 65.609375, "mean_p": 0.379491, "boundary_entropy": 0.613135, "bpb": 0.937078}, "th": {"chunks": 8055.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.500777, "cps": 62.929688, "mean_p": 0.376855, "boundary_entropy": 0.612956, "bpb": 1.082892}, "zh": {"chunks": 2820.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.477481, "cps": 22.03125, "mean_p": 0.273492, "boundary_entropy": 0.425349, "bpb": 3.105319}, "my": {"chunks": 9246.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.487967, "cps": 72.234375, "mean_p": 0.424346, "boundary_entropy": 0.646635, "bpb": 1.066676}}, "bpb": {"en": 2.0095, "de": 2.20128, "vi": 1.90029, "tr": 2.17858, "fi": 2.33595, "ru": 1.37187, "el": 1.41249, "hi": 1.13218, "ta": 0.93708, "th": 1.08289, "zh": 3.10532, "my": 1.06668}, "gini_cpc": 0.206053, "gini_cps": 0.215866, "bpic": 5.666002, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9587298006863944, "vi": 1.0728331089000869, "tr": 0.9661510845001805, "fi": 0.9408656224936733, "ru": 1.5029868867419207, "el": 1.4968856385469838, "hi": 2.3054766187385254, "ta": 2.177003442903416, "th": 2.426907336728517, "zh": 2.3140054674444324, "my": 2.3648261989150625}, "t": 323.3}
{"event": "eval", "step": 5800, "phase": 3, "bpb": {"en": 2.02057, "de": 2.20849, "vi": 1.90057, "tr": 2.18531, "fi": 2.34925, "ru": 1.34731, "el": 1.40416, "hi": 1.12697, "ta": 0.94549, "th": 1.09978, "zh": 3.10442, "my": 1.07527}, "bytes_per_chunk": {"en": 5.0, "de": 5.048, "vi": 6.028, "tr": 5.284, "fi": 4.996, "ru": 5.6, "el": 5.638, "hi": 5.265, "ta": 6.076, "th": 5.738, "zh": 5.911, "my": 6.189}, "bpb_mean": 1.73063, "t": 324.6}
{"event": "train", "step": 5800, "phase": 3, "objective": "global", "lm_loss": 1.2526, "bpb": 1.80711, "chunk_loss": 0.030422, "F": 0.19034, "G": 0.2643, "chunks": 194.91, "bytes_per_chunk": 5.254, "boundary_entropy": 0.3524, "router_noise": 0.0, "grad_norm": 0.7264, "lr": 3.08e-05, "router_trainable": true, "sec_per_step": 0.086, "alpha": 0.03, "t": 324.6}
{"event": "train", "step": 5850, "phase": 3, "objective": "global", "lm_loss": 1.36547, "bpb": 1.96996, "chunk_loss": 0.03024, "F": 0.18118, "G": 0.256, "chunks": 185.53, "bytes_per_chunk": 5.519, "boundary_entropy": 0.34585, "router_noise": 0.0, "grad_norm": 0.735, "lr": 3.05e-05, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 326.6}
{"event": "train", "step": 5900, "phase": 3, "objective": "global", "lm_loss": 1.32292, "bpb": 1.90857, "chunk_loss": 0.030371, "F": 0.18848, "G": 0.25015, "chunks": 193.0, "bytes_per_chunk": 5.306, "boundary_entropy": 0.32337, "router_noise": 0.0, "grad_norm": 0.755, "lr": 3.02e-05, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 328.7}
{"event": "train", "step": 5950, "phase": 3, "objective": "global", "lm_loss": 1.28323, "bpb": 1.8513, "chunk_loss": 0.030236, "F": 0.18295, "G": 0.26492, "chunks": 187.34, "bytes_per_chunk": 5.466, "boundary_entropy": 0.35621, "router_noise": 0.0, "grad_norm": 0.7237, "lr": 3.01e-05, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 330.7}
{"event": "probe", "step": 6000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3614.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.207665, "cps": 28.234375, "mean_p": 0.227684, "boundary_entropy": 0.224928, "bpb": 2.006246}, "de": {"chunks": 4125.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.200896, "cps": 32.226562, "mean_p": 0.234156, "boundary_entropy": 0.247584, "bpb": 2.196274}, "vi": {"chunks": 4074.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.220993, "cps": 31.828125, "mean_p": 0.249066, "boundary_entropy": 0.349036, "bpb": 1.895744}, "tr": {"chunks": 3691.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.203889, "cps": 28.835938, "mean_p": 0.23136, "boundary_entropy": 0.262859, "bpb": 2.174974}, "fi": {"chunks": 3742.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.198926, "cps": 29.234375, "mean_p": 0.23632, "boundary_entropy": 0.268669, "bpb": 2.333904}, "ru": {"chunks": 6085.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.321108, "cps": 47.539062, "mean_p": 0.23698, "boundary_entropy": 0.353598, "bpb": 1.366628}, "el": {"chunks": 6424.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.313015, "cps": 50.1875, "mean_p": 0.25516, "boundary_entropy": 0.397578, "bpb": 1.406318}, "hi": {"chunks": 7388.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.439108, "cps": 57.71875, "mean_p": 0.333555, "boundary_entropy": 0.569096, "bpb": 1.132848}, "ta": {"chunks": 7913.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.423268, "cps": 61.820312, "mean_p": 0.377952, "boundary_entropy": 0.612193, "bpb": 0.937306}, "th": {"chunks": 8076.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.502083, "cps": 63.09375, "mean_p": 0.374238, "boundary_entropy": 0.610078, "bpb": 1.076828}, "zh": {"chunks": 2849.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.482391, "cps": 22.257812, "mean_p": 0.272888, "boundary_entropy": 0.422008, "bpb": 3.096587}, "my": {"chunks": 9726.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.5133, "cps": 75.984375, "mean_p": 0.430351, "boundary_entropy": 0.649485, "bpb": 1.067598}}, "bpb": {"en": 2.00625, "de": 2.19627, "vi": 1.89574, "tr": 2.17497, "fi": 2.3339, "ru": 1.36663, "el": 1.40632, "hi": 1.13285, "ta": 0.93731, "th": 1.07683, "zh": 3.09659, "my": 1.0676}, "gini_cpc": 0.204687, "gini_cps": 0.212434, "bpic": 5.668262, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9674031957033492, "vi": 1.0641769666197316, "tr": 0.9818145543054209, "fi": 0.9579169801888867, "ru": 1.5462771572877685, "el": 1.507303331656917, "hi": 2.114500469118362, "ta": 2.0382228601055625, "th": 2.417749025953193, "zh": 2.32292387976097, "my": 2.471763194501154}, "t": 333.7}
{"event": "eval", "step": 6000, "phase": 3, "bpb": {"en": 2.01817, "de": 2.20358, "vi": 1.89542, "tr": 2.18139, "fi": 2.34578, "ru": 1.34173, "el": 1.39748, "hi": 1.1288, "ta": 0.94596, "th": 1.09444, "zh": 3.09577, "my": 1.07608}, "bytes_per_chunk": {"en": 4.96, "de": 4.974, "vi": 5.939, "tr": 5.2, "fi": 4.885, "ru": 5.446, "el": 5.566, "hi": 5.726, "ta": 6.45, "th": 5.693, "zh": 5.836, "my": 5.965}, "bpb_mean": 1.72705, "t": 335.0}
{"event": "probe", "step": 6000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3614.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.207665, "cps": 28.234375, "mean_p": 0.227684, "boundary_entropy": 0.224928, "bpb": 2.006246}, "de": {"chunks": 4125.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.200896, "cps": 32.226562, "mean_p": 0.234156, "boundary_entropy": 0.247584, "bpb": 2.196274}, "vi": {"chunks": 4074.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.220993, "cps": 31.828125, "mean_p": 0.249066, "boundary_entropy": 0.349036, "bpb": 1.895744}, "tr": {"chunks": 3691.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.203889, "cps": 28.835938, "mean_p": 0.23136, "boundary_entropy": 0.262859, "bpb": 2.174974}, "fi": {"chunks": 3742.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.198926, "cps": 29.234375, "mean_p": 0.23632, "boundary_entropy": 0.268669, "bpb": 2.333904}, "ru": {"chunks": 6085.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.321108, "cps": 47.539062, "mean_p": 0.23698, "boundary_entropy": 0.353598, "bpb": 1.366628}, "el": {"chunks": 6424.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.313015, "cps": 50.1875, "mean_p": 0.25516, "boundary_entropy": 0.397578, "bpb": 1.406318}, "hi": {"chunks": 7388.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.439108, "cps": 57.71875, "mean_p": 0.333555, "boundary_entropy": 0.569096, "bpb": 1.132848}, "ta": {"chunks": 7913.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.423268, "cps": 61.820312, "mean_p": 0.377952, "boundary_entropy": 0.612193, "bpb": 0.937306}, "th": {"chunks": 8076.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.502083, "cps": 63.09375, "mean_p": 0.374238, "boundary_entropy": 0.610078, "bpb": 1.076828}, "zh": {"chunks": 2849.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.482391, "cps": 22.257812, "mean_p": 0.272888, "boundary_entropy": 0.422008, "bpb": 3.096587}, "my": {"chunks": 9726.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.5133, "cps": 75.984375, "mean_p": 0.430351, "boundary_entropy": 0.649485, "bpb": 1.067598}}, "bpb": {"en": 2.00625, "de": 2.19627, "vi": 1.89574, "tr": 2.17497, "fi": 2.3339, "ru": 1.36663, "el": 1.40632, "hi": 1.13285, "ta": 0.93731, "th": 1.07683, "zh": 3.09659, "my": 1.0676}, "gini_cpc": 0.204687, "gini_cps": 0.212434, "bpic": 5.668262, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9674031957033492, "vi": 1.0641769666197316, "tr": 0.9818145543054209, "fi": 0.9579169801888867, "ru": 1.5462771572877685, "el": 1.507303331656917, "hi": 2.114500469118362, "ta": 2.0382228601055625, "th": 2.417749025953193, "zh": 2.32292387976097, "my": 2.471763194501154}, "t": 336.0}
{"event": "reps_dumped", "dirs": ["/local/scratch/sas245/beetle-hnet/scratch/reps/pilot_A_s0/th0.40", "/local/scratch/sas245/beetle-hnet/scratch/reps/pilot_A_s0/th0.50", "/local/scratch/sas245/beetle-hnet/scratch/reps/pilot_A_s0/th0.60"], "t": 337.8}
{"event": "final", "run": "pilot_A_s0", "condition": "A", "step": 6000, "params": 22516992, "per_lang": {"en": {"cpc": 0.2076653450554502, "cps": 28.234375, "bpb": 2.0273781279159193, "bytes_per_chunk": 4.917766459370984}, "de": {"cpc": 0.20089611844348124, "cps": 32.2265625, "bpb": 2.194651560581718, "bytes_per_chunk": 4.990082424380866}, "vi": {"cpc": 0.2209926769731489, "cps": 31.828125, "bpb": 1.8993434507023028, "bytes_per_chunk": 5.929249977381707}, "tr": {"cpc": 0.20388885820029828, "cps": 28.8359375, "bpb": 2.172621765386917, "bytes_per_chunk": 5.211869377205627}, "fi": {"cpc": 0.19892616022540002, "cps": 29.234375, "bpb": 2.362985137303098, "bytes_per_chunk": 4.885778452879069}, "ru": {"cpc": 0.3211081794195251, "cps": 47.5390625, "bpb": 1.3424836631873538, "bytes_per_chunk": 5.437995270298303}, "el": {"cpc": 0.3130146664717634, "cps": 50.1875, "bpb": 1.3926614235177859, "bytes_per_chunk": 5.5650401757756525}, "hi": {"cpc": 0.43910846953937593, "cps": 57.71875, "bpb": 1.1374732168277908, "bytes_per_chunk": 5.7044250291048755}, "ta": {"cpc": 0.42326825354372827, "cps": 61.8203125, "bpb": 0.9469699101528437, "bytes_per_chunk": 6.453172502923257}, "th": {"cpc": 0.5020826857320485, "cps": 63.09375, "bpb": 1.093262663291936, "bytes_per_chunk": 5.682784335403592}, "zh": {"cpc": 0.482390789028107, "cps": 22.2578125, "bpb": 3.1054285549014486, "bytes_per_chunk": 5.831448051875827}, "my": {"cpc": 0.513299556681444, "cps": 75.984375, "bpb": 1.0876047296737308, "bytes_per_chunk": 6.01137405980554}}, "gini_cpc": 0.2046874384551851, "gini_cps": 0.2124337217717518, "bpic": 5.668261775001108, "seed": 0, "mean_boundary_entropy": 0.32596713304519653, "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}], "t": 340.3}