Blame for log.jsonl

{"event": "start", "run": "pilot_Bpc_s0", "condition": "B", "params": 22516992, "langs": ["en", "de", "vi", "tr", "fi", "ru", "el", "hi", "ta", "th", "zh", "my"], "premium": {"en": 1.0, "de": 1.016, "vi": 1.323, "tr": 1.088, "fi": 1.04, "ru": 1.83, "el": 1.821, "hi": 2.569, "ta": 2.733, "th": 2.89, "zh": 2.791, "my": 2.85}, "schedule": {"tau1": 0.4, "tau2": 0.7, "p1_objective": "global", "p2_objective": "parity_a", "p3_objective": "none", "freeze_in_p3": true, "freeze_scope": "router"}, "corpus": "/local/scratch/sas245/beetle-hnet/scratch/data/tier1_corpus", "premium_kind": "flores_per_char", "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": 2.4}
{"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": 2.4}
{"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": 4.3}
{"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": 7.8059, "lr": 1e-06, "router_trainable": true, "sec_per_step": 0.005, "alpha": 0.03, "t": 4.7}
{"event": "probe", "step": 1, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 6331.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.363788, "cps": 49.460938, "mean_p": 0.496549, "boundary_entropy": 0.686989, "bpb": 8.137962}, "de": {"chunks": 8587.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.418205, "cps": 67.085938, "mean_p": 0.498781, "boundary_entropy": 0.687819, "bpb": 8.136727}, "vi": {"chunks": 13019.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.706211, "cps": 101.710938, "mean_p": 0.505542, "boundary_entropy": 0.688387, "bpb": 8.072277}, "tr": {"chunks": 5244.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.289676, "cps": 40.96875, "mean_p": 0.489423, "boundary_entropy": 0.68728, "bpb": 8.156705}, "fi": {"chunks": 8837.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.469778, "cps": 69.039062, "mean_p": 0.500506, "boundary_entropy": 0.687509, "bpb": 8.196348}, "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": 8.0595}, "el": {"chunks": 15046.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.733129, "cps": 117.546875, "mean_p": 0.497468, "boundary_entropy": 0.690234, "bpb": 8.140814}, "hi": {"chunks": 2496.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.148351, "cps": 19.5, "mean_p": 0.484438, "boundary_entropy": 0.690242, "bpb": 8.24965}, "ta": {"chunks": 37850.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.024606, "cps": 295.703125, "mean_p": 0.512695, "boundary_entropy": 0.690391, "bpb": 8.028706}, "th": {"chunks": 33864.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.105316, "cps": 264.5625, "mean_p": 0.511769, "boundary_entropy": 0.690558, "bpb": 8.229679}, "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": 8.120658}, "my": {"chunks": 48159.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.54164, "cps": 376.242188, "mean_p": 0.521508, "boundary_entropy": 0.69021, "bpb": 7.893448}}, "bpb": {"en": 8.13796, "de": 8.13673, "vi": 8.07228, "tr": 8.1567, "fi": 8.19635, "ru": 8.0595, "el": 8.14081, "hi": 8.24965, "ta": 8.02871, "th": 8.22968, "zh": 8.12066, "my": 7.89345}, "gini_cpc": 0.42589, "gini_cps": 0.427336, "bpic": 1.902166, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1495844017862211, "vi": 1.941271558357953, "tr": 0.7962765722561626, "fi": 1.2913524125347302, "ru": 1.5241275892923751, "el": 2.015264334724817, "hi": 0.40779445371638834, "ta": 5.565346656410218, "th": 5.787206735737481, "zh": 5.501433156732364, "my": 6.986600184721901}, "t": 6.1}
{"event": "probe", "step": 2, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 6249.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.359076, "cps": 48.820312, "mean_p": 0.4962, "boundary_entropy": 0.686976, "bpb": 8.088486}, "de": {"chunks": 8520.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.414942, "cps": 66.5625, "mean_p": 0.498575, "boundary_entropy": 0.687814, "bpb": 8.092684}, "vi": {"chunks": 12840.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.696501, "cps": 100.3125, "mean_p": 0.505084, "boundary_entropy": 0.68839, "bpb": 8.038692}, "tr": {"chunks": 5168.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.285478, "cps": 40.375, "mean_p": 0.489116, "boundary_entropy": 0.687262, "bpb": 8.122156}, "fi": {"chunks": 8749.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.4651, "cps": 68.351562, "mean_p": 0.50014, "boundary_entropy": 0.687503, "bpb": 8.154239}, "ru": {"chunks": 10060.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.530871, "cps": 78.59375, "mean_p": 0.492759, "boundary_entropy": 0.689865, "bpb": 7.993298}, "el": {"chunks": 15088.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.735175, "cps": 117.875, "mean_p": 0.497514, "boundary_entropy": 0.690235, "bpb": 8.120826}, "hi": {"chunks": 2407.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.143061, "cps": 18.804688, "mean_p": 0.484164, "boundary_entropy": 0.690224, "bpb": 8.21762}, "ta": {"chunks": 37472.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.004386, "cps": 292.75, "mean_p": 0.512238, "boundary_entropy": 0.690413, "bpb": 7.987848}, "th": {"chunks": 33660.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.092633, "cps": 262.96875, "mean_p": 0.511607, "boundary_entropy": 0.690564, "bpb": 8.20261}, "zh": {"chunks": 11811.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.999831, "cps": 92.273438, "mean_p": 0.519977, "boundary_entropy": 0.685678, "bpb": 8.114133}, "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": 7.860055}}, "bpb": {"en": 8.08849, "de": 8.09268, "vi": 8.03869, "tr": 8.12216, "fi": 8.15424, "ru": 7.9933, "el": 8.12083, "hi": 8.21762, "ta": 7.98785, "th": 8.20261, "zh": 8.11413, "my": 7.86005}, "gini_cpc": 0.428221, "gini_cps": 0.429748, "bpic": 1.917821, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1555820391838354, "vi": 1.9397040711216804, "tr": 0.7950337203375288, "fi": 1.2952694684086565, "ru": 1.4784354309354406, "el": 2.04740814077687, "hi": 0.3984140202402667, "ta": 5.582066415461991, "th": 5.827826875675924, "zh": 5.569379634436058, "my": 7.06799063784341}, "t": 7.6}
{"event": "probe", "step": 4, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 5919.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.340114, "cps": 46.242188, "mean_p": 0.494974, "boundary_entropy": 0.686928, "bpb": 7.918624}, "de": {"chunks": 8281.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.403302, "cps": 64.695312, "mean_p": 0.497821, "boundary_entropy": 0.687798, "bpb": 7.939042}, "vi": {"chunks": 12272.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.66569, "cps": 95.875, "mean_p": 0.503598, "boundary_entropy": 0.688396, "bpb": 7.926935}, "tr": {"chunks": 4899.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.270618, "cps": 38.273438, "mean_p": 0.488079, "boundary_entropy": 0.687201, "bpb": 8.003643}, "fi": {"chunks": 8331.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.442879, "cps": 65.085938, "mean_p": 0.498782, "boundary_entropy": 0.687477, "bpb": 8.007477}, "ru": {"chunks": 8693.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.458734, "cps": 67.914062, "mean_p": 0.490853, "boundary_entropy": 0.689794, "bpb": 7.770983}, "el": {"chunks": 15242.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.742679, "cps": 119.078125, "mean_p": 0.497602, "boundary_entropy": 0.690236, "bpb": 8.033898}, "hi": {"chunks": 2075.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.123328, "cps": 16.210938, "mean_p": 0.482795, "boundary_entropy": 0.690125, "bpb": 8.096812}, "ta": {"chunks": 35907.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 1.920674, "cps": 280.523438, "mean_p": 0.510443, "boundary_entropy": 0.690491, "bpb": 7.841693}, "th": {"chunks": 32624.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.028225, "cps": 254.875, "mean_p": 0.510753, "boundary_entropy": 0.690595, "bpb": 8.088096}, "zh": {"chunks": 11780.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.994582, "cps": 92.03125, "mean_p": 0.51976, "boundary_entropy": 0.68569, "bpb": 8.087305}, "my": {"chunks": 47878.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.52681, "cps": 374.046875, "mean_p": 0.521124, "boundary_entropy": 0.690237, "bpb": 7.783236}}, "bpb": {"en": 7.91862, "de": 7.93904, "vi": 7.92694, "tr": 8.00364, "fi": 8.00748, "ru": 7.77098, "el": 8.0339, "hi": 8.09681, "ta": 7.84169, "th": 8.0881, "zh": 8.08731, "my": 7.78324}, "gini_cpc": 0.437246, "gini_cps": 0.438177, "bpic": 1.979263, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.185785560874662, "vi": 1.9572575854865197, "tr": 0.7956694220790258, "fi": 1.3021500479211365, "ru": 1.3487648692707754, "el": 2.1836191400255838, "hi": 0.3626091913943738, "ta": 5.647151414380306, "th": 5.963372296282688, "zh": 5.8644545934971415, "my": 7.429308703169374}, "t": 9.2}
{"event": "probe", "step": 8, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 5084.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.292134, "cps": 39.71875, "mean_p": 0.491682, "boundary_entropy": 0.686769, "bpb": 7.437778}, "de": {"chunks": 7674.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.37374, "cps": 59.953125, "mean_p": 0.495828, "boundary_entropy": 0.68775, "bpb": 7.496133}, "vi": {"chunks": 10392.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.56371, "cps": 81.1875, "mean_p": 0.498908, "boundary_entropy": 0.688358, "bpb": 7.586001}, "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": 7.643154}, "fi": {"chunks": 7123.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.378661, "cps": 55.648438, "mean_p": 0.494861, "boundary_entropy": 0.687364, "bpb": 7.576952}, "ru": {"chunks": 5217.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.275303, "cps": 40.757812, "mean_p": 0.485329, "boundary_entropy": 0.689504, "bpb": 7.189033}, "el": {"chunks": 14860.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.724066, "cps": 116.09375, "mean_p": 0.497133, "boundary_entropy": 0.69023, "bpb": 7.711189}, "hi": {"chunks": 1134.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.0674, "cps": 8.859375, "mean_p": 0.477107, "boundary_entropy": 0.689636, "bpb": 7.696663}, "ta": {"chunks": 28685.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 1.534367, "cps": 224.101562, "mean_p": 0.503576, "boundary_entropy": 0.690676, "bpb": 7.366253}, "th": {"chunks": 27404.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 1.703699, "cps": 214.09375, "mean_p": 0.506237, "boundary_entropy": 0.690718, "bpb": 7.54773}, "zh": {"chunks": 11596.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.963427, "cps": 90.59375, "mean_p": 0.518921, "boundary_entropy": 0.685739, "bpb": 7.998406}, "my": {"chunks": 47097.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.485592, "cps": 367.945312, "mean_p": 0.520005, "boundary_entropy": 0.69031, "bpb": 7.472465}}, "bpb": {"en": 7.43778, "de": 7.49613, "vi": 7.586, "tr": 7.64315, "fi": 7.57695, "ru": 7.18903, "el": 7.71119, "hi": 7.69666, "ta": 7.36625, "th": 7.54773, "zh": 7.99841, "my": 7.47247}, "gini_cpc": 0.469703, "gini_cps": 0.463467, "bpic": 2.252738, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.2793458539213975, "vi": 1.9296323634382568, "tr": 0.774511685971432, "fi": 1.2961929029338635, "ru": 0.9423889838055757, "el": 2.478543483653829, "hi": 0.2307153871423678, "ta": 5.252281120870893, "th": 5.8319188457660145, "zh": 6.720991441313865, "my": 8.508410726395667}, "t": 10.9}
{"event": "probe", "step": 16, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3940.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.226398, "cps": 30.78125, "mean_p": 0.488086, "boundary_entropy": 0.686662, "bpb": 6.812437}, "de": {"chunks": 6920.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.337018, "cps": 54.0625, "mean_p": 0.493877, "boundary_entropy": 0.687797, "bpb": 6.904856}, "vi": {"chunks": 7186.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.389802, "cps": 56.140625, "mean_p": 0.490782, "boundary_entropy": 0.688162, "bpb": 7.158139}, "tr": {"chunks": 3498.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.193228, "cps": 27.328125, "mean_p": 0.483485, "boundary_entropy": 0.686944, "bpb": 7.131476}, "fi": {"chunks": 5521.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.293498, "cps": 43.132812, "mean_p": 0.490049, "boundary_entropy": 0.687261, "bpb": 7.002118}, "ru": {"chunks": 2615.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.137995, "cps": 20.429688, "mean_p": 0.478849, "boundary_entropy": 0.689012, "bpb": 6.36066}, "el": {"chunks": 5619.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.27379, "cps": 43.898438, "mean_p": 0.487024, "boundary_entropy": 0.689871, "bpb": 7.041437}, "hi": {"chunks": 405.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.024071, "cps": 3.164062, "mean_p": 0.467013, "boundary_entropy": 0.688449, "bpb": 7.053197}, "ta": {"chunks": 14613.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.781653, "cps": 114.164062, "mean_p": 0.490188, "boundary_entropy": 0.690532, "bpb": 6.622472}, "th": {"chunks": 17461.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 1.085546, "cps": 136.414062, "mean_p": 0.496576, "boundary_entropy": 0.690724, "bpb": 6.370007}, "zh": {"chunks": 10872.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.84084, "cps": 84.9375, "mean_p": 0.515365, "boundary_entropy": 0.685938, "bpb": 7.770733}, "my": {"chunks": 42521.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.244089, "cps": 332.195312, "mean_p": 0.5157, "boundary_entropy": 0.690552, "bpb": 6.672919}}, "bpb": {"en": 6.81244, "de": 6.90486, "vi": 7.15814, "tr": 7.13148, "fi": 7.00212, "ru": 6.36066, "el": 7.04144, "hi": 7.0532, "ta": 6.62247, "th": 6.37001, "zh": 7.77073, "my": 6.67292}, "gini_cpc": 0.523338, "gini_cps": 0.492177, "bpic": 3.167268, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.4886122401453055, "vi": 1.7217574438382077, "tr": 0.8534874748022525, "fi": 1.2963842977715545, "ru": 0.6095233917737032, "el": 1.209333504986729, "hi": 0.10632315339302012, "ta": 3.452564598965765, "th": 4.794860157570269, "zh": 8.13099884656574, "my": 9.912152883820962}, "t": 12.8}
{"event": "probe", "step": 32, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 4848.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.278573, "cps": 37.875, "mean_p": 0.495183, "boundary_entropy": 0.687432, "bpb": 6.36086}, "de": {"chunks": 8501.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.414016, "cps": 66.414062, "mean_p": 0.498926, "boundary_entropy": 0.688329, "bpb": 6.444275}, "vi": {"chunks": 5662.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.307133, "cps": 44.234375, "mean_p": 0.489037, "boundary_entropy": 0.68836, "bpb": 6.904187}, "tr": {"chunks": 6794.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.375297, "cps": 53.078125, "mean_p": 0.496451, "boundary_entropy": 0.687978, "bpb": 6.718759}, "fi": {"chunks": 7133.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.379193, "cps": 55.726562, "mean_p": 0.49766, "boundary_entropy": 0.68801, "bpb": 6.521163}, "ru": {"chunks": 5338.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.281689, "cps": 41.703125, "mean_p": 0.48796, "boundary_entropy": 0.689827, "bpb": 5.791293}, "el": {"chunks": 476.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.023193, "cps": 3.71875, "mean_p": 0.469281, "boundary_entropy": 0.68835, "bpb": 6.574395}, "hi": {"chunks": 2732.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.162377, "cps": 21.34375, "mean_p": 0.485756, "boundary_entropy": 0.690389, "bpb": 5.946248}, "ta": {"chunks": 5562.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.297513, "cps": 43.453125, "mean_p": 0.482444, "boundary_entropy": 0.690239, "bpb": 5.808979}, "th": {"chunks": 3717.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.231085, "cps": 29.039062, "mean_p": 0.48787, "boundary_entropy": 0.690625, "bpb": 5.624899}, "zh": {"chunks": 7408.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.254318, "cps": 57.875, "mean_p": 0.501043, "boundary_entropy": 0.686376, "bpb": 7.331245}, "my": {"chunks": 25462.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 1.343783, "cps": 198.921875, "mean_p": 0.496486, "boundary_entropy": 0.690915, "bpb": 5.630882}}, "bpb": {"en": 6.36086, "de": 6.44427, "vi": 6.90419, "tr": 6.71876, "fi": 6.52116, "ru": 5.79129, "el": 6.5744, "hi": 5.94625, "ta": 5.80898, "th": 5.6249, "zh": 7.33125, "my": 5.63088}, "gini_cpc": 0.411792, "gini_cps": 0.367177, "bpic": 4.58887, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.486206368834906, "vi": 1.1025244570836252, "tr": 1.3472137296865077, "fi": 1.3611997154077216, "ru": 1.011185571723399, "el": 0.08325831486817735, "hi": 0.5828907066040927, "ta": 1.0679896011799628, "th": 0.8295317342582874, "zh": 4.502658816508132, "my": 4.823815034685862}, "t": 15.1}
{"event": "train", "step": 50, "phase": 1, "objective": "global", "lm_loss": 4.22409, "bpb": 6.09408, "chunk_loss": 0.031024, "F": 0.17877, "G": 0.4875, "chunks": 183.06, "bytes_per_chunk": 5.594, "boundary_entropy": 0.69167, "router_noise": 0.0, "grad_norm": 1.5901, "lr": 5.1e-05, "router_trainable": true, "sec_per_step": 0.233, "alpha": 0.03, "t": 16.3}
{"event": "probe", "step": 64, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 4816.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.276734, "cps": 37.625, "mean_p": 0.49615, "boundary_entropy": 0.687593, "bpb": 5.819807}, "de": {"chunks": 6276.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.305654, "cps": 49.03125, "mean_p": 0.495963, "boundary_entropy": 0.688417, "bpb": 5.934441}, "vi": {"chunks": 2876.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.156008, "cps": 22.46875, "mean_p": 0.480904, "boundary_entropy": 0.687903, "bpb": 6.312859}, "tr": {"chunks": 5105.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.281997, "cps": 39.882812, "mean_p": 0.494011, "boundary_entropy": 0.688061, "bpb": 6.22684}, "fi": {"chunks": 5485.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.291585, "cps": 42.851562, "mean_p": 0.495197, "boundary_entropy": 0.688105, "bpb": 5.973351}, "ru": {"chunks": 7618.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.402005, "cps": 59.515625, "mean_p": 0.49358, "boundary_entropy": 0.690202, "bpb": 5.414374}, "el": {"chunks": 4032.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.196463, "cps": 31.5, "mean_p": 0.490096, "boundary_entropy": 0.690324, "bpb": 5.662363}, "hi": {"chunks": 413.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.024547, "cps": 3.226562, "mean_p": 0.473499, "boundary_entropy": 0.689271, "bpb": 5.451039}, "ta": {"chunks": 2659.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.142231, "cps": 20.773438, "mean_p": 0.483268, "boundary_entropy": 0.690363, "bpb": 5.209558}, "th": {"chunks": 1482.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.092136, "cps": 11.578125, "mean_p": 0.484061, "boundary_entropy": 0.690427, "bpb": 5.095206}, "zh": {"chunks": 4640.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.785642, "cps": 36.25, "mean_p": 0.492984, "boundary_entropy": 0.686643, "bpb": 6.993735}, "my": {"chunks": 22001.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 1.161125, "cps": 171.882812, "mean_p": 0.494067, "boundary_entropy": 0.690998, "bpb": 4.996877}}, "bpb": {"en": 5.81981, "de": 5.93444, "vi": 6.31286, "tr": 6.22684, "fi": 5.97335, "ru": 5.41437, "el": 5.66236, "hi": 5.45104, "ta": 5.20956, "th": 5.09521, "zh": 6.99373, "my": 4.99688}, "gini_cpc": 0.437329, "gini_cps": 0.413497, "bpic": 5.693827, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1045062295797075, "vi": 0.5637458809663476, "tr": 1.0190203028884368, "fi": 1.0536646027426875, "ru": 1.4526781221784903, "el": 0.7099329283424496, "hi": 0.08870183834963198, "ta": 0.513961407504094, "th": 0.33293908399831873, "zh": 2.8389789987770757, "my": 4.195818436381578}, "t": 18.1}
{"event": "train", "step": 100, "phase": 1, "objective": "global", "lm_loss": 3.38361, "bpb": 4.88151, "chunk_loss": 0.029766, "F": 0.16278, "G": 0.47923, "chunks": 166.69, "bytes_per_chunk": 6.143, "boundary_entropy": 0.69025, "router_noise": 0.0, "grad_norm": 2.1771, "lr": 0.000101, "router_trainable": true, "sec_per_step": 0.079, "alpha": 0.03, "t": 20.3}
{"event": "probe", "step": 128, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2945.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.169224, "cps": 23.007812, "mean_p": 0.449497, "boundary_entropy": 0.676146, "bpb": 4.497236}, "de": {"chunks": 3383.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.164759, "cps": 26.429688, "mean_p": 0.45159, "boundary_entropy": 0.678356, "bpb": 4.712503}, "vi": {"chunks": 6382.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.346189, "cps": 49.859375, "mean_p": 0.467846, "boundary_entropy": 0.683267, "bpb": 5.051431}, "tr": {"chunks": 3387.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.187096, "cps": 26.460938, "mean_p": 0.457661, "boundary_entropy": 0.68028, "bpb": 5.027231}, "fi": {"chunks": 4091.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.217479, "cps": 31.960938, "mean_p": 0.455229, "boundary_entropy": 0.678513, "bpb": 4.741012}, "ru": {"chunks": 5782.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.305119, "cps": 45.171875, "mean_p": 0.466591, "boundary_entropy": 0.685335, "bpb": 3.677409}, "el": {"chunks": 7395.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.360327, "cps": 57.773438, "mean_p": 0.478956, "boundary_entropy": 0.688181, "bpb": 4.65034}, "hi": {"chunks": 9600.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.570579, "cps": 75.0, "mean_p": 0.490162, "boundary_entropy": 0.689678, "bpb": 3.638741}, "ta": {"chunks": 4948.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.26467, "cps": 38.65625, "mean_p": 0.482029, "boundary_entropy": 0.68943, "bpb": 3.474323}, "th": {"chunks": 7334.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.455953, "cps": 57.296875, "mean_p": 0.464548, "boundary_entropy": 0.686151, "bpb": 3.209908}, "zh": {"chunks": 3036.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.514054, "cps": 23.71875, "mean_p": 0.481521, "boundary_entropy": 0.685284, "bpb": 6.325469}, "my": {"chunks": 13626.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.719126, "cps": 106.453125, "mean_p": 0.472223, "boundary_entropy": 0.688516, "bpb": 3.996154}}, "bpb": {"en": 4.49724, "de": 4.7125, "vi": 5.05143, "tr": 5.02723, "fi": 4.74101, "ru": 3.67741, "el": 4.65034, "hi": 3.63874, "ta": 3.47432, "th": 3.20991, "zh": 6.32547, "my": 3.99615}, "gini_cpc": 0.263561, "gini_cps": 0.268966, "bpic": 5.337037, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9736175903678017, "vi": 2.0457496190672226, "tr": 1.1056138393893384, "fi": 1.2851576837198984, "ru": 1.8030496839596653, "el": 2.129296568750932, "hi": 3.3717470112034147, "ta": 1.5640226656937317, "th": 2.694378854272163, "zh": 3.037715839033368, "my": 4.249558670624958}, "t": 22.5}
{"event": "train", "step": 150, "phase": 1, "objective": "global", "lm_loss": 2.89198, "bpb": 4.17224, "chunk_loss": 0.031616, "F": 0.19153, "G": 0.43361, "chunks": 196.12, "bytes_per_chunk": 5.221, "boundary_entropy": 0.67134, "router_noise": 0.0, "grad_norm": 1.547, "lr": 0.000151, "router_trainable": true, "sec_per_step": 0.069, "alpha": 0.03, "t": 23.7}
{"event": "probe", "step": 200, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2435.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.139918, "cps": 19.023438, "mean_p": 0.361595, "boundary_entropy": 0.616396, "bpb": 3.904641}, "de": {"chunks": 2702.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.131593, "cps": 21.109375, "mean_p": 0.364452, "boundary_entropy": 0.62234, "bpb": 4.146256}, "vi": {"chunks": 4790.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.259832, "cps": 37.421875, "mean_p": 0.404728, "boundary_entropy": 0.623325, "bpb": 3.668116}, "tr": {"chunks": 3327.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.183782, "cps": 25.992188, "mean_p": 0.38637, "boundary_entropy": 0.635076, "bpb": 4.200688}, "fi": {"chunks": 3425.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.182074, "cps": 26.757812, "mean_p": 0.366318, "boundary_entropy": 0.61962, "bpb": 4.156496}, "ru": {"chunks": 4704.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.248232, "cps": 36.75, "mean_p": 0.456714, "boundary_entropy": 0.677738, "bpb": 2.8502}, "el": {"chunks": 5254.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.256005, "cps": 41.046875, "mean_p": 0.405604, "boundary_entropy": 0.658979, "bpb": 3.097687}, "hi": {"chunks": 9571.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.568856, "cps": 74.773438, "mean_p": 0.416657, "boundary_entropy": 0.655699, "bpb": 2.429861}, "ta": {"chunks": 8293.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.443595, "cps": 64.789062, "mean_p": 0.464709, "boundary_entropy": 0.680096, "bpb": 2.296895}, "th": {"chunks": 15288.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.950451, "cps": 119.4375, "mean_p": 0.460933, "boundary_entropy": 0.681641, "bpb": 2.331243}, "zh": {"chunks": 3838.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.649848, "cps": 29.984375, "mean_p": 0.469705, "boundary_entropy": 0.680934, "bpb": 5.315341}, "my": {"chunks": 10597.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.559267, "cps": 82.789062, "mean_p": 0.450701, "boundary_entropy": 0.682391, "bpb": 2.434892}}, "bpb": {"en": 3.90464, "de": 4.14626, "vi": 3.66812, "tr": 4.20069, "fi": 4.1565, "ru": 2.8502, "el": 3.09769, "hi": 2.42986, "ta": 2.2969, "th": 2.33124, "zh": 5.31534, "my": 2.43489}, "gini_cpc": 0.342039, "gini_cps": 0.319775, "bpic": 5.170578, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9404984673842508, "vi": 1.857024040847219, "tr": 1.31349191558548, "fi": 1.3012892648661052, "ru": 1.7741210686286728, "el": 1.829676785728758, "hi": 4.065625746374534, "ta": 3.170380225526706, "th": 6.792892838928715, "zh": 4.644475565516153, "my": 3.997097232282538}, "t": 26.9}
{"event": "eval", "step": 200, "phase": 1, "bpb": {"en": 3.90805, "de": 4.14692, "vi": 3.67185, "tr": 4.21058, "fi": 4.15965, "ru": 2.83434, "el": 3.09094, "hi": 2.42533, "ta": 2.31522, "th": 2.3647, "zh": 5.32426, "my": 2.45233}, "bytes_per_chunk": {"en": 6.321, "de": 7.4, "vi": 4.988, "tr": 6.166, "fi": 5.834, "ru": 6.863, "el": 6.605, "hi": 4.493, "ta": 6.016, "th": 2.839, "zh": 5.603, "my": 5.179}, "bpb_mean": 3.40868, "t": 28.3}
{"event": "train", "step": 200, "phase": 1, "objective": "global", "lm_loss": 2.50816, "bpb": 3.61852, "chunk_loss": 0.030651, "F": 0.17548, "G": 0.39958, "chunks": 179.69, "bytes_per_chunk": 5.699, "boundary_entropy": 0.64544, "router_noise": 0.0, "grad_norm": 1.741, "lr": 0.000201, "router_trainable": true, "sec_per_step": 0.092, "alpha": 0.03, "t": 28.3}
{"event": "train", "step": 250, "phase": 1, "objective": "global", "lm_loss": 2.56822, "bpb": 3.70516, "chunk_loss": 0.031904, "F": 0.20447, "G": 0.40759, "chunks": 209.38, "bytes_per_chunk": 4.891, "boundary_entropy": 0.64247, "router_noise": 0.0, "grad_norm": 1.1468, "lr": 0.000251, "router_trainable": true, "sec_per_step": 0.043, "alpha": 0.03, "t": 30.5}
{"event": "probe", "step": 256, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2405.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.138195, "cps": 18.789062, "mean_p": 0.354033, "boundary_entropy": 0.603732, "bpb": 3.658489}, "de": {"chunks": 2926.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.142502, "cps": 22.859375, "mean_p": 0.363972, "boundary_entropy": 0.616389, "bpb": 3.931902}, "vi": {"chunks": 4538.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.246162, "cps": 35.453125, "mean_p": 0.417618, "boundary_entropy": 0.634904, "bpb": 3.274755}, "tr": {"chunks": 3177.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.175496, "cps": 24.820312, "mean_p": 0.378087, "boundary_entropy": 0.616171, "bpb": 3.874093}, "fi": {"chunks": 2422.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.128754, "cps": 18.921875, "mean_p": 0.340622, "boundary_entropy": 0.594145, "bpb": 3.905472}, "ru": {"chunks": 3934.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.207599, "cps": 30.734375, "mean_p": 0.446443, "boundary_entropy": 0.670753, "bpb": 2.670564}, "el": {"chunks": 5753.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.28032, "cps": 44.945312, "mean_p": 0.392621, "boundary_entropy": 0.64591, "bpb": 2.789467}, "hi": {"chunks": 8237.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.489569, "cps": 64.351562, "mean_p": 0.359566, "boundary_entropy": 0.600303, "bpb": 2.168636}, "ta": {"chunks": 8903.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.476224, "cps": 69.554688, "mean_p": 0.413324, "boundary_entropy": 0.649779, "bpb": 1.940382}, "th": {"chunks": 5113.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.317874, "cps": 39.945312, "mean_p": 0.435715, "boundary_entropy": 0.672692, "bpb": 2.101539}, "zh": {"chunks": 4166.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.705384, "cps": 32.546875, "mean_p": 0.465489, "boundary_entropy": 0.675097, "bpb": 4.984301}, "my": {"chunks": 11334.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.598163, "cps": 88.546875, "mean_p": 0.412071, "boundary_entropy": 0.660239, "bpb": 2.158018}}, "bpb": {"en": 3.65849, "de": 3.9319, "vi": 3.27475, "tr": 3.87409, "fi": 3.90547, "ru": 2.67056, "el": 2.78947, "hi": 2.16864, "ta": 1.94038, "th": 2.10154, "zh": 4.9843, "my": 2.15802}, "gini_cpc": 0.315323, "gini_cps": 0.278865, "bpic": 6.100671, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0311716254540002, "vi": 1.7812726063987252, "tr": 1.2699180698849262, "fi": 0.9316897011756403, "ru": 1.5022222173462279, "el": 2.028441878960809, "hi": 3.542607457840742, "ta": 3.4460370572807055, "th": 2.300190296223868, "zh": 5.1042843776335145, "my": 4.328414783956266}, "t": 31.7}
{"event": "train", "step": 300, "phase": 1, "objective": "global", "lm_loss": 2.35534, "bpb": 3.39803, "chunk_loss": 0.030427, "F": 0.17627, "G": 0.37647, "chunks": 180.5, "bytes_per_chunk": 5.673, "boundary_entropy": 0.61766, "router_noise": 0.0, "grad_norm": 1.1896, "lr": 0.0003, "router_trainable": true, "sec_per_step": 0.065, "alpha": 0.03, "t": 33.7}
{"event": "train", "step": 350, "phase": 1, "objective": "global", "lm_loss": 2.05727, "bpb": 2.96801, "chunk_loss": 0.032106, "F": 0.20648, "G": 0.36952, "chunks": 211.44, "bytes_per_chunk": 4.843, "boundary_entropy": 0.59219, "router_noise": 0.0, "grad_norm": 0.7825, "lr": 0.0002999, "router_trainable": true, "sec_per_step": 0.045, "alpha": 0.03, "t": 35.9}
{"event": "probe", "step": 400, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3261.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.187381, "cps": 25.476562, "mean_p": 0.319935, "boundary_entropy": 0.539268, "bpb": 3.254764}, "de": {"chunks": 4900.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.23864, "cps": 38.28125, "mean_p": 0.344991, "boundary_entropy": 0.54714, "bpb": 3.545774}, "vi": {"chunks": 6066.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.329048, "cps": 47.390625, "mean_p": 0.373691, "boundary_entropy": 0.581101, "bpb": 2.805943}, "tr": {"chunks": 4418.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.244048, "cps": 34.515625, "mean_p": 0.338086, "boundary_entropy": 0.546988, "bpb": 3.43354}, "fi": {"chunks": 4202.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.22338, "cps": 32.828125, "mean_p": 0.335173, "boundary_entropy": 0.560586, "bpb": 3.570806}, "ru": {"chunks": 6026.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.317995, "cps": 47.078125, "mean_p": 0.313424, "boundary_entropy": 0.505499, "bpb": 2.473539}, "el": {"chunks": 5328.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.259611, "cps": 41.625, "mean_p": 0.329572, "boundary_entropy": 0.579131, "bpb": 2.549459}, "hi": {"chunks": 8500.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.505201, "cps": 66.40625, "mean_p": 0.362847, "boundary_entropy": 0.608265, "bpb": 1.813101}, "ta": {"chunks": 6789.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.363145, "cps": 53.039062, "mean_p": 0.396887, "boundary_entropy": 0.639151, "bpb": 1.574369}, "th": {"chunks": 7507.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.466708, "cps": 58.648438, "mean_p": 0.38408, "boundary_entropy": 0.636688, "bpb": 1.900586}, "zh": {"chunks": 2760.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.467321, "cps": 21.5625, "mean_p": 0.329348, "boundary_entropy": 0.566386, "bpb": 4.141514}, "my": {"chunks": 7722.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.407536, "cps": 60.328125, "mean_p": 0.360062, "boundary_entropy": 0.605883, "bpb": 1.917969}}, "bpb": {"en": 3.25476, "de": 3.54577, "vi": 2.80594, "tr": 3.43354, "fi": 3.57081, "ru": 2.47354, "el": 2.54946, "hi": 1.8131, "ta": 1.57437, "th": 1.90059, "zh": 4.14151, "my": 1.91797}, "gini_cpc": 0.175125, "gini_cps": 0.17595, "bpic": 5.687414, "worst_lang": "hi", "cpc_parity_vs_en": {"en": 1.0, "de": 1.2735529150888707, "vi": 1.756032645594095, "tr": 1.3024122773695914, "fi": 1.1921131702921772, "ru": 1.6970445150531712, "el": 1.3854686157532505, "hi": 2.6961073117253616, "ta": 1.9379994995439553, "th": 2.4906842358571746, "zh": 2.493956997561615, "my": 2.174902250268444}, "t": 39.2}
{"event": "eval", "step": 400, "phase": 1, "bpb": {"en": 3.26264, "de": 3.5444, "vi": 2.81081, "tr": 3.44026, "fi": 3.56315, "ru": 2.45866, "el": 2.54343, "hi": 1.8101, "ta": 1.58961, "th": 1.92922, "zh": 4.14997, "my": 1.93374}, "bytes_per_chunk": {"en": 5.248, "de": 4.324, "vi": 4.079, "tr": 4.333, "fi": 4.661, "ru": 5.636, "el": 6.493, "hi": 5.135, "ta": 6.298, "th": 4.977, "zh": 4.227, "my": 6.362}, "bpb_mean": 2.753, "t": 40.5}
{"event": "train", "step": 400, "phase": 1, "objective": "global", "lm_loss": 2.05074, "bpb": 2.95859, "chunk_loss": 0.031433, "F": 0.20349, "G": 0.33738, "chunks": 208.38, "bytes_per_chunk": 4.914, "boundary_entropy": 0.55732, "router_noise": 0.0, "grad_norm": 0.9458, "lr": 0.0002998, "router_trainable": true, "sec_per_step": 0.093, "alpha": 0.03, "t": 40.6}
{"event": "train", "step": 450, "phase": 1, "objective": "global", "lm_loss": 2.11843, "bpb": 3.05624, "chunk_loss": 0.030483, "F": 0.17889, "G": 0.31927, "chunks": 183.19, "bytes_per_chunk": 5.59, "boundary_entropy": 0.548, "router_noise": 0.0, "grad_norm": 0.8217, "lr": 0.0002995, "router_trainable": true, "sec_per_step": 0.042, "alpha": 0.03, "t": 42.7}
{"event": "train", "step": 500, "phase": 1, "objective": "global", "lm_loss": 1.9477, "bpb": 2.80993, "chunk_loss": 0.030585, "F": 0.18338, "G": 0.3216, "chunks": 187.78, "bytes_per_chunk": 5.453, "boundary_entropy": 0.54039, "router_noise": 0.0, "grad_norm": 0.887, "lr": 0.0002992, "router_trainable": true, "sec_per_step": 0.042, "alpha": 0.03, "t": 44.8}
{"event": "train", "step": 550, "phase": 1, "objective": "global", "lm_loss": 1.91279, "bpb": 2.75958, "chunk_loss": 0.030517, "F": 0.18268, "G": 0.31813, "chunks": 187.06, "bytes_per_chunk": 5.474, "boundary_entropy": 0.52563, "router_noise": 0.0, "grad_norm": 1.0629, "lr": 0.0002987, "router_trainable": true, "sec_per_step": 0.042, "alpha": 0.03, "t": 46.9}
{"event": "probe", "step": 600, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3096.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.1779, "cps": 24.1875, "mean_p": 0.285888, "boundary_entropy": 0.471038, "bpb": 2.945674}, "de": {"chunks": 3819.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.185993, "cps": 29.835938, "mean_p": 0.290825, "boundary_entropy": 0.466747, "bpb": 3.26997}, "vi": {"chunks": 4963.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.269216, "cps": 38.773438, "mean_p": 0.299067, "boundary_entropy": 0.502086, "bpb": 2.583975}, "tr": {"chunks": 4342.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.23985, "cps": 33.921875, "mean_p": 0.333221, "boundary_entropy": 0.514772, "bpb": 3.157343}, "fi": {"chunks": 3607.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.19175, "cps": 28.179688, "mean_p": 0.3016, "boundary_entropy": 0.503166, "bpb": 3.364695}, "ru": {"chunks": 5737.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.302744, "cps": 44.820312, "mean_p": 0.380254, "boundary_entropy": 0.606061, "bpb": 2.199335}, "el": {"chunks": 6607.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.321931, "cps": 51.617188, "mean_p": 0.329486, "boundary_entropy": 0.562698, "bpb": 2.248374}, "hi": {"chunks": 5789.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.344071, "cps": 45.226562, "mean_p": 0.330026, "boundary_entropy": 0.55844, "bpb": 1.633555}, "ta": {"chunks": 7563.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.404547, "cps": 59.085938, "mean_p": 0.336422, "boundary_entropy": 0.579482, "bpb": 1.419432}, "th": {"chunks": 5783.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.359528, "cps": 45.179688, "mean_p": 0.3103, "boundary_entropy": 0.543719, "bpb": 1.723945}, "zh": {"chunks": 3348.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.566881, "cps": 26.15625, "mean_p": 0.338019, "boundary_entropy": 0.565004, "bpb": 3.856975}, "my": {"chunks": 14332.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.756386, "cps": 111.96875, "mean_p": 0.374529, "boundary_entropy": 0.585145, "bpb": 1.64879}}, "bpb": {"en": 2.94567, "de": 3.26997, "vi": 2.58397, "tr": 3.15734, "fi": 3.36469, "ru": 2.19933, "el": 2.24837, "hi": 1.63355, "ta": 1.41943, "th": 1.72394, "zh": 3.85697, "my": 1.64879}, "gini_cpc": 0.24516, "gini_cps": 0.238548, "bpic": 5.563172, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0454912908205622, "vi": 1.5132974540760222, "tr": 1.3482251860263854, "fi": 1.0778477688504047, "ru": 1.7017619295984945, "el": 1.8096168431938906, "hi": 1.9340675789303936, "ta": 2.2740070097091003, "th": 2.020948726845995, "zh": 3.1865091865584074, "my": 4.251738949356296}, "t": 49.9}
{"event": "eval", "step": 600, "phase": 1, "bpb": {"en": 2.95523, "de": 3.26789, "vi": 2.59166, "tr": 3.16646, "fi": 3.35419, "ru": 2.18543, "el": 2.24001, "hi": 1.62694, "ta": 1.43279, "th": 1.7526, "zh": 3.86749, "my": 1.66051}, "bytes_per_chunk": {"en": 5.599, "de": 5.422, "vi": 5.246, "tr": 4.082, "fi": 5.447, "ru": 6.016, "el": 5.595, "hi": 5.708, "ta": 5.842, "th": 6.846, "zh": 5.149, "my": 4.236}, "bpb_mean": 2.50843, "t": 51.3}
{"event": "train", "step": 600, "phase": 1, "objective": "global", "lm_loss": 1.93788, "bpb": 2.79578, "chunk_loss": 0.030414, "F": 0.17923, "G": 0.31994, "chunks": 183.53, "bytes_per_chunk": 5.579, "boundary_entropy": 0.52387, "router_noise": 0.0, "grad_norm": 0.9142, "lr": 0.0002982, "router_trainable": true, "sec_per_step": 0.088, "alpha": 0.03, "t": 51.3}
{"event": "train", "step": 650, "phase": 1, "objective": "global", "lm_loss": 1.72996, "bpb": 2.4958, "chunk_loss": 0.030951, "F": 0.19272, "G": 0.3345, "chunks": 197.34, "bytes_per_chunk": 5.189, "boundary_entropy": 0.53717, "router_noise": 0.0, "grad_norm": 0.9229, "lr": 0.0002975, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 53.4}
{"event": "train", "step": 700, "phase": 1, "objective": "global", "lm_loss": 1.75532, "bpb": 2.53239, "chunk_loss": 0.030609, "F": 0.18201, "G": 0.318, "chunks": 186.38, "bytes_per_chunk": 5.494, "boundary_entropy": 0.51818, "router_noise": 0.0, "grad_norm": 0.693, "lr": 0.0002967, "router_trainable": true, "sec_per_step": 0.042, "alpha": 0.03, "t": 55.5}
{"event": "train", "step": 750, "phase": 1, "objective": "global", "lm_loss": 1.68923, "bpb": 2.43704, "chunk_loss": 0.030376, "F": 0.17969, "G": 0.33221, "chunks": 184.0, "bytes_per_chunk": 5.565, "boundary_entropy": 0.53146, "router_noise": 0.0, "grad_norm": 0.7074, "lr": 0.0002959, "router_trainable": true, "sec_per_step": 0.042, "alpha": 0.03, "t": 57.5}
{"event": "probe", "step": 800, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3471.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.199448, "cps": 27.117188, "mean_p": 0.284698, "boundary_entropy": 0.429949, "bpb": 2.735564}, "de": {"chunks": 3773.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.183753, "cps": 29.476562, "mean_p": 0.270788, "boundary_entropy": 0.42635, "bpb": 3.081033}, "vi": {"chunks": 4524.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.245403, "cps": 35.34375, "mean_p": 0.28624, "boundary_entropy": 0.472635, "bpb": 2.42861}, "tr": {"chunks": 3423.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.189085, "cps": 26.742188, "mean_p": 0.280946, "boundary_entropy": 0.45858, "bpb": 3.007271}, "fi": {"chunks": 3665.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.194833, "cps": 28.632812, "mean_p": 0.283729, "boundary_entropy": 0.45527, "bpb": 3.241766}, "ru": {"chunks": 5815.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.30686, "cps": 45.429688, "mean_p": 0.368098, "boundary_entropy": 0.594741, "bpb": 2.040478}, "el": {"chunks": 6486.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.316036, "cps": 50.671875, "mean_p": 0.315757, "boundary_entropy": 0.540561, "bpb": 2.144892}, "hi": {"chunks": 6782.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.403091, "cps": 52.984375, "mean_p": 0.322444, "boundary_entropy": 0.540261, "bpb": 1.566937}, "ta": {"chunks": 6407.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.342712, "cps": 50.054688, "mean_p": 0.383549, "boundary_entropy": 0.632789, "bpb": 1.286019}, "th": {"chunks": 8360.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.519739, "cps": 65.3125, "mean_p": 0.394936, "boundary_entropy": 0.635456, "bpb": 1.599095}, "zh": {"chunks": 2833.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.479682, "cps": 22.132812, "mean_p": 0.336408, "boundary_entropy": 0.56539, "bpb": 3.735889}, "my": {"chunks": 11995.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.633048, "cps": 93.710938, "mean_p": 0.375946, "boundary_entropy": 0.614011, "bpb": 1.468848}}, "bpb": {"en": 2.73556, "de": 3.08103, "vi": 2.42861, "tr": 3.00727, "fi": 3.24177, "ru": 2.04048, "el": 2.14489, "hi": 1.56694, "ta": 1.28602, "th": 1.59909, "zh": 3.73589, "my": 1.46885}, "gini_cpc": 0.233189, "gini_cps": 0.236535, "bpic": 5.682782, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9213060107175746, "vi": 1.2304074747899545, "tr": 0.9480382375488843, "fi": 0.9768583701209782, "ru": 1.5385443201072742, "el": 1.5845487519515071, "hi": 2.0210274817924616, "ta": 1.7182990936479072, "th": 2.605881836150808, "zh": 2.4050418527545196, "my": 3.173996055984728}, "t": 60.6}
{"event": "eval", "step": 800, "phase": 1, "bpb": {"en": 2.7471, "de": 3.08311, "vi": 2.43536, "tr": 3.01748, "fi": 3.24038, "ru": 2.02561, "el": 2.1326, "hi": 1.56154, "ta": 1.29721, "th": 1.62674, "zh": 3.7477, "my": 1.47854}, "bytes_per_chunk": {"en": 5.022, "de": 5.369, "vi": 5.249, "tr": 5.391, "fi": 5.104, "ru": 6.006, "el": 5.801, "hi": 6.046, "ta": 6.923, "th": 5.065, "zh": 6.241, "my": 5.176}, "bpb_mean": 2.36611, "t": 61.9}
{"event": "train", "step": 800, "phase": 1, "objective": "global", "lm_loss": 1.81499, "bpb": 2.61847, "chunk_loss": 0.030505, "F": 0.18301, "G": 0.31431, "chunks": 187.41, "bytes_per_chunk": 5.464, "boundary_entropy": 0.50227, "router_noise": 0.0, "grad_norm": 0.8138, "lr": 0.0002949, "router_trainable": true, "sec_per_step": 0.087, "alpha": 0.03, "t": 61.9}
{"event": "train", "step": 850, "phase": 1, "objective": "global", "lm_loss": 1.86108, "bpb": 2.68497, "chunk_loss": 0.030277, "F": 0.1792, "G": 0.30953, "chunks": 183.5, "bytes_per_chunk": 5.58, "boundary_entropy": 0.49379, "router_noise": 0.0, "grad_norm": 0.8805, "lr": 0.0002938, "router_trainable": true, "sec_per_step": 0.042, "alpha": 0.03, "t": 64.0}
{"event": "train", "step": 900, "phase": 1, "objective": "global", "lm_loss": 1.67693, "bpb": 2.41929, "chunk_loss": 0.030345, "F": 0.18246, "G": 0.30841, "chunks": 186.84, "bytes_per_chunk": 5.481, "boundary_entropy": 0.47741, "router_noise": 0.0, "grad_norm": 0.948, "lr": 0.0002927, "router_trainable": true, "sec_per_step": 0.042, "alpha": 0.03, "t": 66.1}
{"event": "train", "step": 950, "phase": 1, "objective": "global", "lm_loss": 1.81076, "bpb": 2.61237, "chunk_loss": 0.030536, "F": 0.1853, "G": 0.30267, "chunks": 189.75, "bytes_per_chunk": 5.397, "boundary_entropy": 0.46889, "router_noise": 0.0, "grad_norm": 0.7978, "lr": 0.0002914, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 68.1}
{"event": "probe", "step": 1000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3508.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.201574, "cps": 27.40625, "mean_p": 0.273757, "boundary_entropy": 0.389339, "bpb": 2.592093}, "de": {"chunks": 4103.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.199825, "cps": 32.054688, "mean_p": 0.279913, "boundary_entropy": 0.414848, "bpb": 2.941359}, "vi": {"chunks": 4550.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.246813, "cps": 35.546875, "mean_p": 0.265353, "boundary_entropy": 0.454592, "bpb": 2.323846}, "tr": {"chunks": 3559.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.196597, "cps": 27.804688, "mean_p": 0.275517, "boundary_entropy": 0.429643, "bpb": 2.916535}, "fi": {"chunks": 3967.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.210887, "cps": 30.992188, "mean_p": 0.28745, "boundary_entropy": 0.433518, "bpb": 3.135206}, "ru": {"chunks": 5570.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.293931, "cps": 43.515625, "mean_p": 0.351856, "boundary_entropy": 0.571525, "bpb": 1.951639}, "el": {"chunks": 6547.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.319008, "cps": 51.148438, "mean_p": 0.316499, "boundary_entropy": 0.531562, "bpb": 2.034949}, "hi": {"chunks": 6544.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.388945, "cps": 51.125, "mean_p": 0.381594, "boundary_entropy": 0.62271, "bpb": 1.504878}, "ta": {"chunks": 6303.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.337149, "cps": 49.242188, "mean_p": 0.434207, "boundary_entropy": 0.663108, "bpb": 1.226586}, "th": {"chunks": 7030.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.437053, "cps": 54.921875, "mean_p": 0.3337, "boundary_entropy": 0.585177, "bpb": 1.525289}, "zh": {"chunks": 2894.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.49001, "cps": 22.609375, "mean_p": 0.337882, "boundary_entropy": 0.555277, "bpb": 3.629964}, "my": {"chunks": 12126.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.639962, "cps": 94.734375, "mean_p": 0.36808, "boundary_entropy": 0.605965, "bpb": 1.4187}}, "bpb": {"en": 2.59209, "de": 2.94136, "vi": 2.32385, "tr": 2.91653, "fi": 3.13521, "ru": 1.95164, "el": 2.03495, "hi": 1.50488, "ta": 1.22659, "th": 1.52529, "zh": 3.62996, "my": 1.4187}, "gini_cpc": 0.217147, "gini_cps": 0.218226, "bpic": 5.753752, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 0.991319491203664, "vi": 1.2244266969001691, "tr": 0.975308416661255, "fi": 1.0462003296671256, "ru": 1.4581779152363388, "el": 1.5825813055863958, "hi": 1.929535411312034, "ta": 1.6725779733055046, "th": 2.1681972800568814, "zh": 2.430914139071991, "my": 3.17481719157442}, "t": 71.1}
{"event": "eval", "step": 1000, "phase": 1, "bpb": {"en": 2.60245, "de": 2.94498, "vi": 2.32641, "tr": 2.92602, "fi": 3.13771, "ru": 1.93469, "el": 2.02472, "hi": 1.49952, "ta": 1.23742, "th": 1.55087, "zh": 3.6388, "my": 1.42787}, "bytes_per_chunk": {"en": 4.956, "de": 5.019, "vi": 5.332, "tr": 5.297, "fi": 4.721, "ru": 6.241, "el": 5.808, "hi": 6.172, "ta": 7.084, "th": 6.449, "zh": 5.882, "my": 4.454}, "bpb_mean": 2.27096, "t": 72.4}
{"event": "train", "step": 1000, "phase": 1, "objective": "global", "lm_loss": 1.74285, "bpb": 2.51441, "chunk_loss": 0.030334, "F": 0.18112, "G": 0.3156, "chunks": 185.47, "bytes_per_chunk": 5.521, "boundary_entropy": 0.49649, "router_noise": 0.0, "grad_norm": 0.7938, "lr": 0.0002901, "router_trainable": true, "sec_per_step": 0.086, "alpha": 0.03, "t": 72.4}
{"event": "train", "step": 1050, "phase": 1, "objective": "global", "lm_loss": 1.60591, "bpb": 2.31684, "chunk_loss": 0.030358, "F": 0.18268, "G": 0.32864, "chunks": 187.06, "bytes_per_chunk": 5.474, "boundary_entropy": 0.50608, "router_noise": 0.0, "grad_norm": 0.7715, "lr": 0.0002886, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 74.4}
{"event": "train", "step": 1100, "phase": 1, "objective": "global", "lm_loss": 1.80365, "bpb": 2.60212, "chunk_loss": 0.030712, "F": 0.18881, "G": 0.31346, "chunks": 193.34, "bytes_per_chunk": 5.296, "boundary_entropy": 0.48485, "router_noise": 0.0, "grad_norm": 0.7854, "lr": 0.0002871, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 76.4}
{"event": "train", "step": 1150, "phase": 1, "objective": "global", "lm_loss": 1.65668, "bpb": 2.39008, "chunk_loss": 0.03021, "F": 0.17969, "G": 0.31065, "chunks": 184.0, "bytes_per_chunk": 5.565, "boundary_entropy": 0.46896, "router_noise": 0.0, "grad_norm": 0.7795, "lr": 0.0002855, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 78.4}
{"event": "probe", "step": 1200, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3256.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.187094, "cps": 25.4375, "mean_p": 0.252283, "boundary_entropy": 0.356069, "bpb": 2.500838}, "de": {"chunks": 3789.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.184532, "cps": 29.601562, "mean_p": 0.25514, "boundary_entropy": 0.373141, "bpb": 2.827947}, "vi": {"chunks": 4145.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.224844, "cps": 32.382812, "mean_p": 0.270944, "boundary_entropy": 0.456335, "bpb": 2.292995}, "tr": {"chunks": 3538.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.195437, "cps": 27.640625, "mean_p": 0.257224, "boundary_entropy": 0.39869, "bpb": 2.801202}, "fi": {"chunks": 3758.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.199777, "cps": 29.359375, "mean_p": 0.267213, "boundary_entropy": 0.401188, "bpb": 3.062494}, "ru": {"chunks": 5586.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.294776, "cps": 43.640625, "mean_p": 0.352043, "boundary_entropy": 0.57288, "bpb": 1.869537}, "el": {"chunks": 6446.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.314087, "cps": 50.359375, "mean_p": 0.326867, "boundary_entropy": 0.547427, "bpb": 1.957073}, "hi": {"chunks": 6500.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.38633, "cps": 50.78125, "mean_p": 0.410908, "boundary_entropy": 0.645666, "bpb": 1.470633}, "ta": {"chunks": 6773.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.362289, "cps": 52.914062, "mean_p": 0.420667, "boundary_entropy": 0.652794, "bpb": 1.20264}, "th": {"chunks": 8196.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.509543, "cps": 64.03125, "mean_p": 0.352328, "boundary_entropy": 0.595559, "bpb": 1.465119}, "zh": {"chunks": 2628.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.444971, "cps": 20.53125, "mean_p": 0.331751, "boundary_entropy": 0.541958, "bpb": 3.577307}, "my": {"chunks": 9430.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.497678, "cps": 73.671875, "mean_p": 0.372729, "boundary_entropy": 0.622554, "bpb": 1.342646}}, "bpb": {"en": 2.50084, "de": 2.82795, "vi": 2.293, "tr": 2.8012, "fi": 3.06249, "ru": 1.86954, "el": 1.95707, "hi": 1.47063, "ta": 1.20264, "th": 1.46512, "zh": 3.57731, "my": 1.34265}, "gini_cpc": 0.207854, "gini_cps": 0.214757, "bpic": 5.992365, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9863065609851273, "vi": 1.2017693316513496, "tr": 1.0445927351826925, "fi": 1.0677869684434993, "ru": 1.575547282710871, "el": 1.6787621929779508, "hi": 2.0648951666782276, "ta": 1.9364011419574405, "th": 2.723457537849829, "zh": 2.3783274161702876, "my": 2.660039224122505}, "t": 81.4}
{"event": "eval", "step": 1200, "phase": 1, "bpb": {"en": 2.51216, "de": 2.83232, "vi": 2.29854, "tr": 2.81148, "fi": 3.06815, "ru": 1.85243, "el": 1.94793, "hi": 1.46802, "ta": 1.20995, "th": 1.48884, "zh": 3.58726, "my": 1.3523}, "bytes_per_chunk": {"en": 5.315, "de": 5.375, "vi": 5.921, "tr": 5.289, "fi": 4.944, "ru": 6.059, "el": 5.639, "hi": 6.232, "ta": 6.82, "th": 5.621, "zh": 6.419, "my": 5.759}, "bpb_mean": 2.20245, "t": 82.7}
{"event": "train", "step": 1200, "phase": 1, "objective": "global", "lm_loss": 1.72144, "bpb": 2.48351, "chunk_loss": 0.030249, "F": 0.17993, "G": 0.2926, "chunks": 184.25, "bytes_per_chunk": 5.558, "boundary_entropy": 0.4445, "router_noise": 0.0, "grad_norm": 0.925, "lr": 0.0002837, "router_trainable": true, "sec_per_step": 0.085, "alpha": 0.03, "t": 82.7}
{"event": "train", "step": 1250, "phase": 1, "objective": "global", "lm_loss": 1.83776, "bpb": 2.65133, "chunk_loss": 0.030291, "F": 0.18185, "G": 0.29578, "chunks": 186.22, "bytes_per_chunk": 5.499, "boundary_entropy": 0.44577, "router_noise": 0.0, "grad_norm": 0.8643, "lr": 0.0002819, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 84.7}
{"event": "train", "step": 1300, "phase": 1, "objective": "global", "lm_loss": 1.51895, "bpb": 2.19139, "chunk_loss": 0.030455, "F": 0.18365, "G": 0.31652, "chunks": 188.06, "bytes_per_chunk": 5.445, "boundary_entropy": 0.48389, "router_noise": 0.0, "grad_norm": 0.8377, "lr": 0.00028, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 86.7}
{"event": "train", "step": 1350, "phase": 1, "objective": "global", "lm_loss": 1.51419, "bpb": 2.18452, "chunk_loss": 0.030764, "F": 0.18982, "G": 0.31588, "chunks": 194.38, "bytes_per_chunk": 5.268, "boundary_entropy": 0.47641, "router_noise": 0.0, "grad_norm": 0.7736, "lr": 0.000278, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 88.7}
{"event": "probe", "step": 1400, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3309.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.19014, "cps": 25.851562, "mean_p": 0.255749, "boundary_entropy": 0.349683, "bpb": 2.422935}, "de": {"chunks": 3830.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.186529, "cps": 29.921875, "mean_p": 0.255616, "boundary_entropy": 0.361941, "bpb": 2.756585}, "vi": {"chunks": 4098.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.222295, "cps": 32.015625, "mean_p": 0.264374, "boundary_entropy": 0.433689, "bpb": 2.193617}, "tr": {"chunks": 3257.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.179915, "cps": 25.445312, "mean_p": 0.240136, "boundary_entropy": 0.368443, "bpb": 2.725519}, "fi": {"chunks": 3357.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.178459, "cps": 26.226562, "mean_p": 0.241532, "boundary_entropy": 0.365121, "bpb": 2.987169}, "ru": {"chunks": 5644.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.297836, "cps": 44.09375, "mean_p": 0.351982, "boundary_entropy": 0.567215, "bpb": 1.80062}, "el": {"chunks": 6204.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.302295, "cps": 48.46875, "mean_p": 0.315439, "boundary_entropy": 0.531968, "bpb": 1.885073}, "hi": {"chunks": 6135.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.364636, "cps": 47.929688, "mean_p": 0.405997, "boundary_entropy": 0.643906, "bpb": 1.42736}, "ta": {"chunks": 6486.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.346938, "cps": 50.671875, "mean_p": 0.399209, "boundary_entropy": 0.64555, "bpb": 1.152224}, "th": {"chunks": 7053.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.438483, "cps": 55.101562, "mean_p": 0.315026, "boundary_entropy": 0.557167, "bpb": 1.415557}, "zh": {"chunks": 2392.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.405012, "cps": 18.6875, "mean_p": 0.330413, "boundary_entropy": 0.543611, "bpb": 3.546509}, "my": {"chunks": 8658.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.456935, "cps": 67.640625, "mean_p": 0.37979, "boundary_entropy": 0.632224, "bpb": 1.306044}}, "bpb": {"en": 2.42294, "de": 2.75658, "vi": 2.19362, "tr": 2.72552, "fi": 2.98717, "ru": 1.80062, "el": 1.88507, "hi": 1.42736, "ta": 1.15222, "th": 1.41556, "zh": 3.54651, "my": 1.30604}, "gini_cpc": 0.190752, "gini_cps": 0.204151, "bpic": 6.351571, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9810106447401589, "vi": 1.1691121263334627, "tr": 0.946225309199514, "fi": 0.9385703075971684, "ru": 1.566408906388669, "el": 1.589857855342097, "hi": 1.917727284361575, "ta": 1.8246468759251981, "th": 2.3061108103454253, "zh": 2.130075934272782, "my": 2.4031537570687806}, "t": 91.6}
{"event": "eval", "step": 1400, "phase": 1, "bpb": {"en": 2.43197, "de": 2.75775, "vi": 2.19853, "tr": 2.73593, "fi": 2.98815, "ru": 1.78258, "el": 1.87601, "hi": 1.4243, "ta": 1.16093, "th": 1.43787, "zh": 3.55145, "my": 1.31427}, "bytes_per_chunk": {"en": 5.249, "de": 5.39, "vi": 5.94, "tr": 5.701, "fi": 5.571, "ru": 6.096, "el": 6.096, "hi": 6.596, "ta": 6.956, "th": 6.528, "zh": 7.201, "my": 6.164}, "bpb_mean": 2.13831, "t": 92.9}
{"event": "train", "step": 1400, "phase": 1, "objective": "global", "lm_loss": 1.56757, "bpb": 2.26153, "chunk_loss": 0.030208, "F": 0.17807, "G": 0.29665, "chunks": 182.34, "bytes_per_chunk": 5.616, "boundary_entropy": 0.45035, "router_noise": 0.0, "grad_norm": 0.6718, "lr": 0.0002759, "router_trainable": true, "sec_per_step": 0.085, "alpha": 0.03, "t": 93.0}
{"event": "train", "step": 1450, "phase": 1, "objective": "global", "lm_loss": 1.66374, "bpb": 2.40027, "chunk_loss": 0.030395, "F": 0.18262, "G": 0.31348, "chunks": 187.0, "bytes_per_chunk": 5.476, "boundary_entropy": 0.46886, "router_noise": 0.0, "grad_norm": 0.7032, "lr": 0.0002738, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 95.0}
{"event": "train", "step": 1500, "phase": 1, "objective": "global", "lm_loss": 1.74063, "bpb": 2.5112, "chunk_loss": 0.030236, "F": 0.17651, "G": 0.31578, "chunks": 180.75, "bytes_per_chunk": 5.665, "boundary_entropy": 0.48516, "router_noise": 0.0, "grad_norm": 0.6814, "lr": 0.0002715, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 97.0}
{"event": "train", "step": 1550, "phase": 1, "objective": "global", "lm_loss": 1.60294, "bpb": 2.31255, "chunk_loss": 0.03012, "F": 0.17627, "G": 0.29942, "chunks": 180.5, "bytes_per_chunk": 5.673, "boundary_entropy": 0.45128, "router_noise": 0.0, "grad_norm": 0.7358, "lr": 0.0002692, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 99.0}
{"event": "probe", "step": 1600, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3527.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.202666, "cps": 27.554688, "mean_p": 0.260759, "boundary_entropy": 0.346727, "bpb": 2.36503}, "de": {"chunks": 4113.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.200312, "cps": 32.132812, "mean_p": 0.262689, "boundary_entropy": 0.360847, "bpb": 2.664279}, "vi": {"chunks": 4089.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.221806, "cps": 31.945312, "mean_p": 0.266661, "boundary_entropy": 0.420393, "bpb": 2.165463}, "tr": {"chunks": 3649.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.201569, "cps": 28.507812, "mean_p": 0.255474, "boundary_entropy": 0.369388, "bpb": 2.664731}, "fi": {"chunks": 3961.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.210568, "cps": 30.945312, "mean_p": 0.269216, "boundary_entropy": 0.377413, "bpb": 2.885314}, "ru": {"chunks": 6440.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.339842, "cps": 50.3125, "mean_p": 0.363097, "boundary_entropy": 0.561923, "bpb": 1.743528}, "el": {"chunks": 7097.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.345807, "cps": 55.445312, "mean_p": 0.323149, "boundary_entropy": 0.52304, "bpb": 1.831899}, "hi": {"chunks": 6191.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.367964, "cps": 48.367188, "mean_p": 0.399513, "boundary_entropy": 0.637128, "bpb": 1.415303}, "ta": {"chunks": 13599.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.727414, "cps": 106.242188, "mean_p": 0.433091, "boundary_entropy": 0.647363, "bpb": 1.175172}, "th": {"chunks": 6879.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.427666, "cps": 53.742188, "mean_p": 0.342043, "boundary_entropy": 0.592212, "bpb": 1.379039}, "zh": {"chunks": 2426.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.410769, "cps": 18.953125, "mean_p": 0.327672, "boundary_entropy": 0.539184, "bpb": 3.532173}, "my": {"chunks": 7090.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.374182, "cps": 55.390625, "mean_p": 0.364488, "boundary_entropy": 0.618711, "bpb": 1.266052}}, "bpb": {"en": 2.36503, "de": 2.66428, "vi": 2.16546, "tr": 2.66473, "fi": 2.88531, "ru": 1.74353, "el": 1.8319, "hi": 1.4153, "ta": 1.17517, "th": 1.37904, "zh": 3.53217, "my": 1.26605}, "gini_cpc": 0.218171, "gini_cps": 0.244761, "bpic": 5.557131, "worst_lang": "ta", "cpc_parity_vs_en": {"en": 1.0, "de": 0.988382307838602, "vi": 1.0944416927373928, "tr": 0.9945851543731558, "fi": 1.0389906026095133, "ru": 1.676854240899267, "el": 1.7062891163027571, "hi": 1.8156176319296147, "ta": 3.589220708502241, "th": 2.110196530887133, "zh": 2.0268238889756742, "my": 1.8462968113708185}, "t": 102.0}
{"event": "eval", "step": 1600, "phase": 1, "bpb": {"en": 2.37236, "de": 2.66857, "vi": 2.17135, "tr": 2.67468, "fi": 2.8903, "ru": 1.72457, "el": 1.82356, "hi": 1.4122, "ta": 1.18415, "th": 1.39716, "zh": 3.53632, "my": 1.27556}, "bytes_per_chunk": {"en": 4.955, "de": 5.03, "vi": 6.069, "tr": 5.213, "fi": 4.776, "ru": 5.255, "el": 5.093, "hi": 6.418, "ta": 3.802, "th": 6.605, "zh": 6.832, "my": 7.522}, "bpb_mean": 2.09423, "t": 103.3}
{"event": "train", "step": 1600, "phase": 1, "objective": "global", "lm_loss": 1.59598, "bpb": 2.30252, "chunk_loss": 0.03094, "F": 0.1947, "G": 0.31171, "chunks": 199.38, "bytes_per_chunk": 5.136, "boundary_entropy": 0.46021, "router_noise": 0.0, "grad_norm": 0.7521, "lr": 0.0002668, "router_trainable": true, "sec_per_step": 0.086, "alpha": 0.03, "t": 103.3}
{"event": "train", "step": 1650, "phase": 1, "objective": "global", "lm_loss": 1.5778, "bpb": 2.27629, "chunk_loss": 0.030773, "F": 0.18985, "G": 0.30547, "chunks": 194.41, "bytes_per_chunk": 5.267, "boundary_entropy": 0.44498, "router_noise": 0.0, "grad_norm": 0.8075, "lr": 0.0002643, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 105.3}
{"event": "train", "step": 1700, "phase": 1, "objective": "global", "lm_loss": 1.54777, "bpb": 2.23297, "chunk_loss": 0.03034, "F": 0.17947, "G": 0.30747, "chunks": 183.78, "bytes_per_chunk": 5.572, "boundary_entropy": 0.45973, "router_noise": 0.0, "grad_norm": 0.7579, "lr": 0.0002618, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 107.3}
{"event": "train", "step": 1750, "phase": 1, "objective": "global", "lm_loss": 1.56415, "bpb": 2.25658, "chunk_loss": 0.030953, "F": 0.19684, "G": 0.30605, "chunks": 201.56, "bytes_per_chunk": 5.08, "boundary_entropy": 0.44027, "router_noise": 0.0, "grad_norm": 0.713, "lr": 0.0002591, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 109.3}
{"event": "probe", "step": 1800, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3408.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.195828, "cps": 26.625, "mean_p": 0.244641, "boundary_entropy": 0.324817, "bpb": 2.323584}, "de": {"chunks": 3877.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.188818, "cps": 30.289062, "mean_p": 0.247832, "boundary_entropy": 0.340891, "bpb": 2.623267}, "vi": {"chunks": 4296.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.233035, "cps": 33.5625, "mean_p": 0.278005, "boundary_entropy": 0.429498, "bpb": 2.132412}, "tr": {"chunks": 3531.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.195051, "cps": 27.585938, "mean_p": 0.243682, "boundary_entropy": 0.354172, "bpb": 2.622261}, "fi": {"chunks": 3637.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.193344, "cps": 28.414062, "mean_p": 0.242575, "boundary_entropy": 0.343409, "bpb": 2.832424}, "ru": {"chunks": 4946.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.261003, "cps": 38.640625, "mean_p": 0.326721, "boundary_entropy": 0.545345, "bpb": 1.704003}, "el": {"chunks": 5652.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.275398, "cps": 44.15625, "mean_p": 0.307272, "boundary_entropy": 0.521566, "bpb": 1.817317}, "hi": {"chunks": 6953.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.413254, "cps": 54.320312, "mean_p": 0.411086, "boundary_entropy": 0.645318, "bpb": 1.371943}, "ta": {"chunks": 7323.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.391709, "cps": 57.210938, "mean_p": 0.384081, "boundary_entropy": 0.627453, "bpb": 1.10832}, "th": {"chunks": 7729.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.48051, "cps": 60.382812, "mean_p": 0.379283, "boundary_entropy": 0.621176, "bpb": 1.339184}, "zh": {"chunks": 2838.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.480528, "cps": 22.171875, "mean_p": 0.330253, "boundary_entropy": 0.529276, "bpb": 3.473896}, "my": {"chunks": 5373.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.283566, "cps": 41.976562, "mean_p": 0.352844, "boundary_entropy": 0.618711, "bpb": 1.260117}}, "bpb": {"en": 2.32358, "de": 2.62327, "vi": 2.13241, "tr": 2.62226, "fi": 2.83242, "ru": 1.704, "el": 1.81732, "hi": 1.37194, "ta": 1.10832, "th": 1.33918, "zh": 3.4739, "my": 1.26012}, "gini_cpc": 0.196409, "gini_cps": 0.179366, "bpic": 6.443279, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9642017779006381, "vi": 1.1899964473578655, "tr": 0.9960283506818961, "fi": 0.9873154922838824, "ru": 1.332813649708276, "el": 1.4063254692605272, "hi": 2.1102878115647825, "ta": 2.000267592449589, "th": 2.4537300192200764, "zh": 2.453824411078731, "my": 1.4480314637986245}, "t": 112.3}
{"event": "eval", "step": 1800, "phase": 1, "bpb": {"en": 2.33156, "de": 2.62475, "vi": 2.13741, "tr": 2.63288, "fi": 2.8353, "ru": 1.68411, "el": 1.80966, "hi": 1.36939, "ta": 1.11537, "th": 1.36351, "zh": 3.47715, "my": 1.26846}, "bytes_per_chunk": {"en": 5.193, "de": 5.332, "vi": 5.699, "tr": 5.329, "fi": 5.133, "ru": 6.669, "el": 6.633, "hi": 6.551, "ta": 6.619, "th": 6.027, "zh": 5.995, "my": 10.692}, "bpb_mean": 2.05413, "t": 113.5}
{"event": "train", "step": 1800, "phase": 1, "objective": "global", "lm_loss": 1.53207, "bpb": 2.21031, "chunk_loss": 0.029876, "F": 0.16693, "G": 0.29838, "chunks": 170.94, "bytes_per_chunk": 5.99, "boundary_entropy": 0.45996, "router_noise": 0.0, "grad_norm": 0.7066, "lr": 0.0002564, "router_trainable": true, "sec_per_step": 0.085, "alpha": 0.03, "t": 113.6}
{"event": "train", "step": 1850, "phase": 1, "objective": "global", "lm_loss": 1.57611, "bpb": 2.27385, "chunk_loss": 0.030577, "F": 0.18909, "G": 0.29067, "chunks": 193.62, "bytes_per_chunk": 5.289, "boundary_entropy": 0.41692, "router_noise": 0.0, "grad_norm": 0.7061, "lr": 0.0002537, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 115.6}
{"event": "train", "step": 1900, "phase": 1, "objective": "global", "lm_loss": 1.62978, "bpb": 2.35127, "chunk_loss": 0.030949, "F": 0.19293, "G": 0.30414, "chunks": 197.56, "bytes_per_chunk": 5.183, "boundary_entropy": 0.43444, "router_noise": 0.0, "grad_norm": 0.7176, "lr": 0.0002508, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 117.6}
{"event": "train", "step": 1950, "phase": 1, "objective": "global", "lm_loss": 1.47125, "bpb": 2.12256, "chunk_loss": 0.030345, "F": 0.17902, "G": 0.31529, "chunks": 183.31, "bytes_per_chunk": 5.586, "boundary_entropy": 0.46956, "router_noise": 0.0, "grad_norm": 0.6555, "lr": 0.0002479, "router_trainable": true, "sec_per_step": 0.041, "alpha": 0.03, "t": 119.6}
{"event": "probe", "step": 2000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3579.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.205654, "cps": 27.960938, "mean_p": 0.251011, "boundary_entropy": 0.313775, "bpb": 2.280117}, "de": {"chunks": 4269.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.207909, "cps": 33.351562, "mean_p": 0.261609, "boundary_entropy": 0.331525, "bpb": 2.560879}, "vi": {"chunks": 4077.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.221155, "cps": 31.851562, "mean_p": 0.274006, "boundary_entropy": 0.41565, "bpb": 2.106163}, "tr": {"chunks": 3731.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.206098, "cps": 29.148438, "mean_p": 0.249861, "boundary_entropy": 0.336514, "bpb": 2.556184}, "fi": {"chunks": 3938.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.209346, "cps": 30.765625, "mean_p": 0.256425, "boundary_entropy": 0.342346, "bpb": 2.770621}, "ru": {"chunks": 6062.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.319894, "cps": 47.359375, "mean_p": 0.350113, "boundary_entropy": 0.548637, "bpb": 1.652588}, "el": {"chunks": 6584.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.320811, "cps": 51.4375, "mean_p": 0.32239, "boundary_entropy": 0.523793, "bpb": 1.719352}, "hi": {"chunks": 8291.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.492779, "cps": 64.773438, "mean_p": 0.425903, "boundary_entropy": 0.645437, "bpb": 1.351509}, "ta": {"chunks": 7886.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.421824, "cps": 61.609375, "mean_p": 0.401403, "boundary_entropy": 0.640897, "bpb": 1.110235}, "th": {"chunks": 8136.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.505813, "cps": 63.5625, "mean_p": 0.39278, "boundary_entropy": 0.6315, "bpb": 1.328894}, "zh": {"chunks": 2718.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.46021, "cps": 21.234375, "mean_p": 0.329599, "boundary_entropy": 0.53329, "bpb": 3.439732}, "my": {"chunks": 9792.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.516783, "cps": 76.5, "mean_p": 0.392326, "boundary_entropy": 0.633379, "bpb": 1.225564}}, "bpb": {"en": 2.28012, "de": 2.56088, "vi": 2.10616, "tr": 2.55618, "fi": 2.77062, "ru": 1.65259, "el": 1.71935, "hi": 1.35151, "ta": 1.11024, "th": 1.32889, "zh": 3.43973, "my": 1.22556}, "gini_cpc": 0.203247, "gini_cps": 0.217327, "bpic": 5.55697, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0109651141586364, "vi": 1.075375137174609, "tr": 1.0021601269331555, "fi": 1.017949539342762, "ru": 1.5554968627459653, "el": 1.5599525876943248, "hi": 2.39615144808311, "ta": 2.0511325425775344, "th": 2.459530975556564, "zh": 2.237785376883837, "my": 2.5128724823450286}, "t": 122.6}
{"event": "eval", "step": 2000, "phase": 1, "bpb": {"en": 2.28745, "de": 2.56442, "vi": 2.11182, "tr": 2.56278, "fi": 2.77932, "ru": 1.62924, "el": 1.71329, "hi": 1.34973, "ta": 1.1171, "th": 1.34663, "zh": 3.44269, "my": 1.23426}, "bytes_per_chunk": {"en": 5.0, "de": 4.831, "vi": 6.028, "tr": 5.154, "fi": 4.785, "ru": 5.611, "el": 5.593, "hi": 5.918, "ta": 6.217, "th": 5.841, "zh": 6.172, "my": 5.89}, "bpb_mean": 2.01156, "t": 123.9}
{"event": "train", "step": 2000, "phase": 1, "objective": "global", "lm_loss": 1.59273, "bpb": 2.29782, "chunk_loss": 0.030671, "F": 0.19467, "G": 0.29681, "chunks": 199.34, "bytes_per_chunk": 5.137, "boundary_entropy": 0.42098, "router_noise": 0.0, "grad_norm": 0.7079, "lr": 0.000245, "router_trainable": true, "sec_per_step": 0.086, "alpha": 0.03, "t": 123.9}
{"event": "train", "step": 2050, "phase": 1, "objective": "global", "lm_loss": 1.60434, "bpb": 2.31458, "chunk_loss": 0.030519, "F": 0.18649, "G": 0.30618, "chunks": 190.97, "bytes_per_chunk": 5.362, "boundary_entropy": 0.45212, "router_noise": 0.0, "grad_norm": 0.7212, "lr": 0.0002419, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 126.0}
{"event": "train", "step": 2100, "phase": 1, "objective": "global", "lm_loss": 1.55115, "bpb": 2.23784, "chunk_loss": 0.030374, "F": 0.18243, "G": 0.29422, "chunks": 186.81, "bytes_per_chunk": 5.481, "boundary_entropy": 0.42539, "router_noise": 0.0, "grad_norm": 0.759, "lr": 0.0002388, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 128.0}
{"event": "train", "step": 2150, "phase": 1, "objective": "global", "lm_loss": 1.50459, "bpb": 2.17067, "chunk_loss": 0.030471, "F": 0.18423, "G": 0.29418, "chunks": 188.66, "bytes_per_chunk": 5.428, "boundary_entropy": 0.42771, "router_noise": 0.0, "grad_norm": 0.7037, "lr": 0.0002357, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 130.0}
{"event": "probe", "step": 2200, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3435.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.19738, "cps": 26.835938, "mean_p": 0.244105, "boundary_entropy": 0.306748, "bpb": 2.25169}, "de": {"chunks": 3875.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.188721, "cps": 30.273438, "mean_p": 0.240291, "boundary_entropy": 0.316305, "bpb": 2.52106}, "vi": {"chunks": 4279.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.232113, "cps": 33.429688, "mean_p": 0.280848, "boundary_entropy": 0.414782, "bpb": 2.084006}, "tr": {"chunks": 3607.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.199249, "cps": 28.179688, "mean_p": 0.238903, "boundary_entropy": 0.323469, "bpb": 2.523586}, "fi": {"chunks": 3957.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.210356, "cps": 30.914062, "mean_p": 0.255948, "boundary_entropy": 0.339636, "bpb": 2.724214}, "ru": {"chunks": 5518.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.291187, "cps": 43.109375, "mean_p": 0.329506, "boundary_entropy": 0.536366, "bpb": 1.62357}, "el": {"chunks": 6361.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.309945, "cps": 49.695312, "mean_p": 0.313227, "boundary_entropy": 0.515232, "bpb": 1.691825}, "hi": {"chunks": 8320.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.494502, "cps": 65.0, "mean_p": 0.421218, "boundary_entropy": 0.64198, "bpb": 1.331072}, "ta": {"chunks": 8218.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.439583, "cps": 64.203125, "mean_p": 0.386097, "boundary_entropy": 0.625162, "bpb": 1.079093}, "th": {"chunks": 5896.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.366553, "cps": 46.0625, "mean_p": 0.376058, "boundary_entropy": 0.627784, "bpb": 1.283148}, "zh": {"chunks": 3083.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.522012, "cps": 24.085938, "mean_p": 0.323013, "boundary_entropy": 0.513207, "bpb": 3.44245}, "my": {"chunks": 8595.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.45361, "cps": 67.148438, "mean_p": 0.398693, "boundary_entropy": 0.645909, "bpb": 1.203857}}, "bpb": {"en": 2.25169, "de": 2.52106, "vi": 2.08401, "tr": 2.52359, "fi": 2.72421, "ru": 1.62357, "el": 1.69183, "hi": 1.33107, "ta": 1.07909, "th": 1.28315, "zh": 3.44245, "my": 1.20386}, "gini_cpc": 0.206755, "gini_cps": 0.200018, "bpic": 5.891272, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9561294131483307, "vi": 1.1759707600053535, "tr": 1.009468960607991, "fi": 1.065740685499681, "ru": 1.475264393773548, "el": 1.570297463682774, "hi": 2.505333999493891, "ta": 2.2270914274302145, "th": 1.8570935790078091, "zh": 2.6447063677379377, "my": 2.2981580017201755}, "t": 132.9}
{"event": "eval", "step": 2200, "phase": 1, "bpb": {"en": 2.26018, "de": 2.52468, "vi": 2.0873, "tr": 2.5291, "fi": 2.73592, "ru": 1.60203, "el": 1.68275, "hi": 1.32822, "ta": 1.08572, "th": 1.29921, "zh": 3.44615, "my": 1.21242}, "bytes_per_chunk": {"en": 5.089, "de": 5.331, "vi": 5.838, "tr": 5.229, "fi": 4.707, "ru": 6.111, "el": 5.632, "hi": 5.202, "ta": 5.97, "th": 7.695, "zh": 5.465, "my": 6.326}, "bpb_mean": 1.98281, "t": 134.2}
{"event": "train", "step": 2200, "phase": 1, "objective": "global", "lm_loss": 1.45663, "bpb": 2.10148, "chunk_loss": 0.03041, "F": 0.18573, "G": 0.30088, "chunks": 190.19, "bytes_per_chunk": 5.384, "boundary_entropy": 0.43508, "router_noise": 0.0, "grad_norm": 0.6922, "lr": 0.0002325, "router_trainable": true, "sec_per_step": 0.086, "alpha": 0.03, "t": 134.2}
{"event": "train", "step": 2250, "phase": 1, "objective": "global", "lm_loss": 1.53302, "bpb": 2.21168, "chunk_loss": 0.03009, "F": 0.17685, "G": 0.29156, "chunks": 181.09, "bytes_per_chunk": 5.655, "boundary_entropy": 0.42865, "router_noise": 0.0, "grad_norm": 0.7211, "lr": 0.0002293, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 136.2}
{"event": "train", "step": 2300, "phase": 1, "objective": "global", "lm_loss": 1.46354, "bpb": 2.11145, "chunk_loss": 0.030695, "F": 0.19177, "G": 0.30583, "chunks": 196.38, "bytes_per_chunk": 5.215, "boundary_entropy": 0.44257, "router_noise": 0.0, "grad_norm": 0.7196, "lr": 0.000226, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 138.2}
{"event": "train", "step": 2350, "phase": 1, "objective": "global", "lm_loss": 1.49002, "bpb": 2.14964, "chunk_loss": 0.030123, "F": 0.17841, "G": 0.29252, "chunks": 182.69, "bytes_per_chunk": 5.605, "boundary_entropy": 0.42582, "router_noise": 0.0, "grad_norm": 0.6987, "lr": 0.0002226, "router_trainable": true, "sec_per_step": 0.04, "alpha": 0.03, "t": 140.2}
{"event": "probe", "step": 2400, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 3542.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.203528, "cps": 27.671875, "mean_p": 0.244646, "boundary_entropy": 0.296713, "bpb": 2.226111}, "de": {"chunks": 4102.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.199776, "cps": 32.046875, "mean_p": 0.252125, "boundary_entropy": 0.317674, "bpb": 2.471309}, "vi": {"chunks": 4252.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.230648, "cps": 33.21875, "mean_p": 0.280175, "boundary_entropy": 0.408328, "bpb": 2.069628}, "tr": {"chunks": 3775.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.208529, "cps": 29.492188, "mean_p": 0.245659, "boundary_entropy": 0.328388, "bpb": 2.480591}, "fi": {"chunks": 3955.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.210249, "cps": 30.898438, "mean_p": 0.253848, "boundary_entropy": 0.328974, "bpb": 2.674576}, "ru": {"chunks": 5789.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.305488, "cps": 45.226562, "mean_p": 0.337615, "boundary_entropy": 0.534983, "bpb": 1.58849}, "el": {"chunks": 6662.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.324611, "cps": 52.046875, "mean_p": 0.313624, "boundary_entropy": 0.512265, "bpb": 1.664292}, "hi": {"chunks": 7166.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.425914, "cps": 55.984375, "mean_p": 0.411228, "boundary_entropy": 0.637471, "bpb": 1.310527}, "ta": {"chunks": 9545.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.510564, "cps": 74.570312, "mean_p": 0.400345, "boundary_entropy": 0.629455, "bpb": 1.061176}, "th": {"chunks": 8020.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.498601, "cps": 62.65625, "mean_p": 0.407095, "boundary_entropy": 0.637739, "bpb": 1.254685}, "zh": {"chunks": 2992.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.506603, "cps": 23.375, "mean_p": 0.317092, "boundary_entropy": 0.501734, "bpb": 3.380984}, "my": {"chunks": 8392.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.442896, "cps": 65.5625, "mean_p": 0.398278, "boundary_entropy": 0.643388, "bpb": 1.186828}}, "bpb": {"en": 2.22611, "de": 2.47131, "vi": 2.06963, "tr": 2.48059, "fi": 2.67458, "ru": 1.58849, "el": 1.66429, "hi": 1.31053, "ta": 1.06118, "th": 1.25469, "zh": 3.38098, "my": 1.18683}, "gini_cpc": 0.203754, "gini_cps": 0.208627, "bpic": 5.627948, "worst_lang": "ta", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9815644304579402, "vi": 1.1332498682553516, "tr": 1.0245707845373044, "fi": 1.0330234201962816, "ru": 1.5009626956730318, "el": 1.5949216250278262, "hi": 2.092653356076058, "ta": 2.508568858261289, "th": 2.4497900499306073, "zh": 2.489107823815343, "my": 2.176094054911138}, "t": 143.2}
{"event": "eval", "step": 2400, "phase": 1, "bpb": {"en": 2.2347, "de": 2.4763, "vi": 2.07336, "tr": 2.48843, "fi": 2.68606, "ru": 1.56542, "el": 1.65689, "hi": 1.3085, "ta": 1.0684, "th": 1.27273, "zh": 3.38546, "my": 1.19693}, "bytes_per_chunk": {"en": 5.001, "de": 5.027, "vi": 5.831, "tr": 5.025, "fi": 4.699, "ru": 5.708, "el": 5.507, "hi": 6.309, "ta": 5.391, "th": 5.713, "zh": 5.681, "my": 6.674}, "bpb_mean": 1.9511, "t": 144.5}
{"event": "phase_transition", "step": 2400, "phase": 2, "objective": "parity_a", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0, "t": 144.5}
{"event": "train", "step": 2400, "phase": 2, "objective": "parity_a", "lm_loss": 1.37123, "bpb": 1.97826, "chunk_loss": 0.03627, "F": 0.19232, "G": 0.30375, "chunks": 196.94, "bytes_per_chunk": 5.2, "boundary_entropy": 0.43166, "router_noise": 0.0, "grad_norm": 0.735, "lr": 0.0002192, "router_trainable": true, "sec_per_step": 0.086, "n_target_mean": 9.3706, "alpha": 0.03, "t": 144.6}
{"event": "train", "step": 2450, "phase": 2, "objective": "parity_a", "lm_loss": 1.47089, "bpb": 2.12205, "chunk_loss": 0.031787, "F": 0.13379, "G": 0.2453, "chunks": 137.0, "bytes_per_chunk": 7.474, "boundary_entropy": 0.41011, "router_noise": 0.0, "grad_norm": 0.661, "lr": 0.0002158, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 11.2097, "alpha": 0.03, "t": 146.5}
{"event": "train", "step": 2500, "phase": 2, "objective": "parity_a", "lm_loss": 1.50351, "bpb": 2.1691, "chunk_loss": 0.030669, "F": 0.14481, "G": 0.23209, "chunks": 148.28, "bytes_per_chunk": 6.906, "boundary_entropy": 0.35867, "router_noise": 0.0, "grad_norm": 0.6977, "lr": 0.0002123, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 9.9734, "alpha": 0.03, "t": 148.5}
{"event": "train", "step": 2550, "phase": 2, "objective": "parity_a", "lm_loss": 1.5681, "bpb": 2.26229, "chunk_loss": 0.030611, "F": 0.15466, "G": 0.23115, "chunks": 158.38, "bytes_per_chunk": 6.466, "boundary_entropy": 0.34464, "router_noise": 0.0, "grad_norm": 0.7127, "lr": 0.0002088, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 8.9697, "alpha": 0.03, "t": 150.5}
{"event": "probe", "step": 2600, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 3511.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.201747, "cps": 27.429688, "mean_p": 0.237419, "boundary_entropy": 0.278968, "bpb": 2.197908}, "de": {"chunks": 3963.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.193006, "cps": 30.960938, "mean_p": 0.237666, "boundary_entropy": 0.292793, "bpb": 2.435293}, "vi": {"chunks": 3641.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.197505, "cps": 28.445312, "mean_p": 0.224338, "boundary_entropy": 0.3424, "bpb": 2.053493}, "tr": {"chunks": 3628.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.200409, "cps": 28.34375, "mean_p": 0.231509, "boundary_entropy": 0.299472, "bpb": 2.444682}, "fi": {"chunks": 3822.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.203179, "cps": 29.859375, "mean_p": 0.245683, "boundary_entropy": 0.309435, "bpb": 2.647152}, "ru": {"chunks": 3789.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.199947, "cps": 29.601562, "mean_p": 0.214899, "boundary_entropy": 0.406715, "bpb": 1.591907}, "el": {"chunks": 4062.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.197924, "cps": 31.734375, "mean_p": 0.226398, "boundary_entropy": 0.426258, "bpb": 1.686615}, "hi": {"chunks": 3603.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.214146, "cps": 28.148438, "mean_p": 0.231233, "boundary_entropy": 0.458062, "bpb": 1.33816}, "ta": {"chunks": 3574.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.191174, "cps": 27.921875, "mean_p": 0.256832, "boundary_entropy": 0.521648, "bpb": 1.087931}, "th": {"chunks": 2688.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.167112, "cps": 21.0, "mean_p": 0.23274, "boundary_entropy": 0.48533, "bpb": 1.270136}, "zh": {"chunks": 1066.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.180494, "cps": 8.328125, "mean_p": 0.197669, "boundary_entropy": 0.400049, "bpb": 3.364458}, "my": {"chunks": 3254.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.171733, "cps": 25.421875, "mean_p": 0.270786, "boundary_entropy": 0.54582, "bpb": 1.197372}}, "bpb": {"en": 2.19791, "de": 2.43529, "vi": 2.05349, "tr": 2.44468, "fi": 2.64715, "ru": 1.59191, "el": 1.68662, "hi": 1.33816, "ta": 1.08793, "th": 1.27014, "zh": 3.36446, "my": 1.19737}, "gini_cpc": 0.03654, "gini_cps": 0.102672, "bpic": 9.452501, "worst_lang": "hi", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9566761693769058, "vi": 0.9789732559694406, "tr": 0.9933676615156039, "fi": 1.007098826730168, "ru": 0.9910799304710638, "el": 0.9810527616527605, "hi": 1.061457182118775, "ta": 0.9475941466657841, "th": 0.8283263745799316, "zh": 0.8946580062872402, "my": 0.8512310626387456}, "t": 153.4}
{"event": "eval", "step": 2600, "phase": 2, "bpb": {"en": 2.20534, "de": 2.44149, "vi": 2.05652, "tr": 2.44987, "fi": 2.65712, "ru": 1.56906, "el": 1.67943, "hi": 1.33607, "ta": 1.09315, "th": 1.28629, "zh": 3.36864, "my": 1.20676}, "bytes_per_chunk": {"en": 5.087, "de": 5.143, "vi": 6.836, "tr": 5.205, "fi": 4.852, "ru": 8.712, "el": 9.056, "hi": 11.322, "ta": 13.615, "th": 16.426, "zh": 14.303, "my": 15.594}, "bpb_mean": 1.94581, "t": 154.7}
{"event": "train", "step": 2600, "phase": 2, "objective": "parity_a", "lm_loss": 1.59834, "bpb": 2.30592, "chunk_loss": 0.031544, "F": 0.15369, "G": 0.2323, "chunks": 157.38, "bytes_per_chunk": 6.507, "boundary_entropy": 0.3464, "router_noise": 0.0, "grad_norm": 0.7342, "lr": 0.0002053, "router_trainable": true, "sec_per_step": 0.085, "n_target_mean": 9.4883, "alpha": 0.03, "t": 154.8}
{"event": "train", "step": 2650, "phase": 2, "objective": "parity_a", "lm_loss": 1.50232, "bpb": 2.16739, "chunk_loss": 0.031166, "F": 0.14413, "G": 0.22496, "chunks": 147.59, "bytes_per_chunk": 6.938, "boundary_entropy": 0.3499, "router_noise": 0.0, "grad_norm": 0.6557, "lr": 0.0002017, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 10.1825, "alpha": 0.03, "t": 156.8}
{"event": "train", "step": 2700, "phase": 2, "objective": "parity_a", "lm_loss": 1.51574, "bpb": 2.18676, "chunk_loss": 0.03055, "F": 0.14911, "G": 0.22445, "chunks": 152.69, "bytes_per_chunk": 6.707, "boundary_entropy": 0.3359, "router_noise": 0.0, "grad_norm": 0.6795, "lr": 0.0001981, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 9.4002, "alpha": 0.03, "t": 158.8}
{"event": "train", "step": 2750, "phase": 2, "objective": "parity_a", "lm_loss": 1.40368, "bpb": 2.02508, "chunk_loss": 0.030317, "F": 0.15787, "G": 0.22351, "chunks": 161.66, "bytes_per_chunk": 6.334, "boundary_entropy": 0.31761, "router_noise": 0.0, "grad_norm": 0.701, "lr": 0.0001945, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 8.4643, "alpha": 0.03, "t": 160.8}
{"event": "probe", "step": 2800, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 3536.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.203183, "cps": 27.625, "mean_p": 0.235728, "boundary_entropy": 0.276361, "bpb": 2.179256}, "de": {"chunks": 4091.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.19924, "cps": 31.960938, "mean_p": 0.240564, "boundary_entropy": 0.289624, "bpb": 2.40998}, "vi": {"chunks": 3368.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.182696, "cps": 26.3125, "mean_p": 0.212562, "boundary_entropy": 0.335949, "bpb": 2.054144}, "tr": {"chunks": 3531.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.195051, "cps": 27.585938, "mean_p": 0.224988, "boundary_entropy": 0.289475, "bpb": 2.421369}, "fi": {"chunks": 3627.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.192813, "cps": 28.335938, "mean_p": 0.230823, "boundary_entropy": 0.293426, "bpb": 2.619495}, "ru": {"chunks": 3676.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.193984, "cps": 28.71875, "mean_p": 0.201853, "boundary_entropy": 0.385631, "bpb": 1.565082}, "el": {"chunks": 4077.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.198655, "cps": 31.851562, "mean_p": 0.216545, "boundary_entropy": 0.408361, "bpb": 1.646748}, "hi": {"chunks": 3483.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.207013, "cps": 27.210938, "mean_p": 0.224279, "boundary_entropy": 0.43989, "bpb": 1.317777}, "ta": {"chunks": 3247.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.173683, "cps": 25.367188, "mean_p": 0.243403, "boundary_entropy": 0.502158, "bpb": 1.096326}, "th": {"chunks": 2728.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.169599, "cps": 21.3125, "mean_p": 0.211416, "boundary_entropy": 0.447845, "bpb": 1.258491}, "zh": {"chunks": 985.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.16678, "cps": 7.695312, "mean_p": 0.175056, "boundary_entropy": 0.363146, "bpb": 3.358778}, "my": {"chunks": 4350.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.229576, "cps": 33.984375, "mean_p": 0.307442, "boundary_entropy": 0.580614, "bpb": 1.181779}}, "bpb": {"en": 2.17926, "de": 2.40998, "vi": 2.05414, "tr": 2.42137, "fi": 2.61949, "ru": 1.56508, "el": 1.64675, "hi": 1.31778, "ta": 1.09633, "th": 1.25849, "zh": 3.35878, "my": 1.18178}, "gini_cpc": 0.048124, "gini_cps": 0.116852, "bpic": 9.42974, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9805933330365665, "vi": 0.8991679215342642, "tr": 0.9599730257703342, "fi": 0.9489591900233593, "ru": 0.9547246863023674, "el": 0.9777137663449219, "hi": 1.0188500366428432, "ta": 0.854808206636904, "th": 0.834709131494591, "zh": 0.8208327044511831, "my": 1.1298941100530433}, "t": 163.7}
{"event": "eval", "step": 2800, "phase": 2, "bpb": {"en": 2.18683, "de": 2.41666, "vi": 2.05614, "tr": 2.42588, "fi": 2.62966, "ru": 1.54288, "el": 1.63975, "hi": 1.31361, "ta": 1.10201, "th": 1.27452, "zh": 3.35845, "my": 1.19115}, "bytes_per_chunk": {"en": 5.067, "de": 5.03, "vi": 7.349, "tr": 5.347, "fi": 5.087, "ru": 8.859, "el": 9.212, "hi": 11.505, "ta": 15.154, "th": 16.077, "zh": 15.121, "my": 12.493}, "bpb_mean": 1.92813, "t": 165.0}
{"event": "train", "step": 2800, "phase": 2, "objective": "parity_a", "lm_loss": 1.47102, "bpb": 2.12224, "chunk_loss": 0.030667, "F": 0.1492, "G": 0.22889, "chunks": 152.78, "bytes_per_chunk": 6.702, "boundary_entropy": 0.3455, "router_noise": 0.0, "grad_norm": 0.7578, "lr": 0.0001909, "router_trainable": true, "sec_per_step": 0.084, "n_target_mean": 9.484, "alpha": 0.03, "t": 165.0}
{"event": "train", "step": 2850, "phase": 2, "objective": "parity_a", "lm_loss": 1.43931, "bpb": 2.07649, "chunk_loss": 0.030691, "F": 0.1503, "G": 0.2253, "chunks": 153.91, "bytes_per_chunk": 6.653, "boundary_entropy": 0.33834, "router_noise": 0.0, "grad_norm": 0.6779, "lr": 0.0001872, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 9.4603, "alpha": 0.03, "t": 167.0}
{"event": "train", "step": 2900, "phase": 2, "objective": "parity_a", "lm_loss": 1.45703, "bpb": 2.10205, "chunk_loss": 0.030474, "F": 0.14804, "G": 0.2196, "chunks": 151.59, "bytes_per_chunk": 6.755, "boundary_entropy": 0.32738, "router_noise": 0.0, "grad_norm": 0.7401, "lr": 0.0001835, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 9.3798, "alpha": 0.03, "t": 169.0}
{"event": "train", "step": 2950, "phase": 2, "objective": "parity_a", "lm_loss": 1.35867, "bpb": 1.96014, "chunk_loss": 0.030041, "F": 0.13016, "G": 0.21345, "chunks": 133.28, "bytes_per_chunk": 7.683, "boundary_entropy": 0.33695, "router_noise": 0.0, "grad_norm": 0.6476, "lr": 0.0001799, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 10.2378, "alpha": 0.03, "t": 171.0}
{"event": "probe", "step": 3000, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 3383.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.194392, "cps": 26.429688, "mean_p": 0.227072, "boundary_entropy": 0.260285, "bpb": 2.155543}, "de": {"chunks": 3868.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.18838, "cps": 30.21875, "mean_p": 0.2285, "boundary_entropy": 0.268179, "bpb": 2.387429}, "vi": {"chunks": 3396.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.184215, "cps": 26.53125, "mean_p": 0.201579, "boundary_entropy": 0.310289, "bpb": 2.022728}, "tr": {"chunks": 3455.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.190852, "cps": 26.992188, "mean_p": 0.218014, "boundary_entropy": 0.275537, "bpb": 2.387123}, "fi": {"chunks": 3591.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.190899, "cps": 28.054688, "mean_p": 0.227193, "boundary_entropy": 0.288041, "bpb": 2.575077}, "ru": {"chunks": 3161.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.166807, "cps": 24.695312, "mean_p": 0.177741, "boundary_entropy": 0.349717, "bpb": 1.553764}, "el": {"chunks": 3522.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.171612, "cps": 27.515625, "mean_p": 0.197613, "boundary_entropy": 0.386921, "bpb": 1.629904}, "hi": {"chunks": 2955.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.175632, "cps": 23.085938, "mean_p": 0.213457, "boundary_entropy": 0.432519, "bpb": 1.303495}, "ta": {"chunks": 2928.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.156619, "cps": 22.875, "mean_p": 0.238208, "boundary_entropy": 0.495639, "bpb": 1.073597}, "th": {"chunks": 2345.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.145788, "cps": 18.320312, "mean_p": 0.197385, "boundary_entropy": 0.43219, "bpb": 1.25235}, "zh": {"chunks": 906.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.153403, "cps": 7.078125, "mean_p": 0.158955, "boundary_entropy": 0.342719, "bpb": 3.330553}, "my": {"chunks": 2442.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.128879, "cps": 19.078125, "mean_p": 0.248844, "boundary_entropy": 0.52032, "bpb": 1.181909}}, "bpb": {"en": 2.15554, "de": 2.38743, "vi": 2.02273, "tr": 2.38712, "fi": 2.57508, "ru": 1.55376, "el": 1.6299, "hi": 1.3035, "ta": 1.0736, "th": 1.25235, "zh": 3.33055, "my": 1.18191}, "gini_cpc": 0.065276, "gini_cps": 0.128727, "bpic": 10.674816, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9690723017858436, "vi": 0.9476471526252331, "tr": 0.9817923022640718, "fi": 0.9820320099058397, "ru": 0.8580990159551444, "el": 0.8828168806922737, "hi": 0.903492464507662, "ta": 0.8056895398440214, "th": 0.7499700223584966, "zh": 0.7891451240385509, "my": 0.6629860736832469}, "t": 173.9}
{"event": "eval", "step": 3000, "phase": 2, "bpb": {"en": 2.16424, "de": 2.3903, "vi": 2.02233, "tr": 2.39645, "fi": 2.58517, "ru": 1.5313, "el": 1.62215, "hi": 1.30088, "ta": 1.07858, "th": 1.26983, "zh": 3.33101, "my": 1.19245}, "bytes_per_chunk": {"en": 5.235, "de": 5.265, "vi": 7.17, "tr": 5.456, "fi": 5.103, "ru": 10.208, "el": 10.478, "hi": 13.149, "ta": 17.949, "th": 18.984, "zh": 16.426, "my": 20.628}, "bpb_mean": 1.90706, "t": 175.2}
{"event": "train", "step": 3000, "phase": 2, "objective": "parity_a", "lm_loss": 1.43747, "bpb": 2.07383, "chunk_loss": 0.030247, "F": 0.12854, "G": 0.20771, "chunks": 131.62, "bytes_per_chunk": 7.78, "boundary_entropy": 0.33375, "router_noise": 0.0, "grad_norm": 0.6827, "lr": 0.0001761, "router_trainable": true, "sec_per_step": 0.084, "n_target_mean": 10.5865, "alpha": 0.03, "t": 175.2}
{"event": "train", "step": 3050, "phase": 2, "objective": "parity_a", "lm_loss": 1.39936, "bpb": 2.01884, "chunk_loss": 0.030385, "F": 0.13699, "G": 0.21296, "chunks": 140.28, "bytes_per_chunk": 7.3, "boundary_entropy": 0.32686, "router_noise": 0.0, "grad_norm": 0.6593, "lr": 0.0001724, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 10.0199, "alpha": 0.03, "t": 177.2}
{"event": "train", "step": 3100, "phase": 2, "objective": "parity_a", "lm_loss": 1.41134, "bpb": 2.03614, "chunk_loss": 0.030477, "F": 0.15076, "G": 0.21658, "chunks": 154.38, "bytes_per_chunk": 6.633, "boundary_entropy": 0.31538, "router_noise": 0.0, "grad_norm": 0.7334, "lr": 0.0001687, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 8.9614, "alpha": 0.03, "t": 179.2}
{"event": "train", "step": 3150, "phase": 2, "objective": "parity_a", "lm_loss": 1.3579, "bpb": 1.95904, "chunk_loss": 0.030322, "F": 0.13095, "G": 0.20641, "chunks": 134.09, "bytes_per_chunk": 7.636, "boundary_entropy": 0.32563, "router_noise": 0.0, "grad_norm": 0.6661, "lr": 0.000165, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 10.0892, "alpha": 0.03, "t": 181.2}
{"event": "probe", "step": 3200, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 3490.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.20054, "cps": 27.265625, "mean_p": 0.234003, "boundary_entropy": 0.254631, "bpb": 2.137038}, "de": {"chunks": 4079.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.198656, "cps": 31.867188, "mean_p": 0.238388, "boundary_entropy": 0.268116, "bpb": 2.350866}, "vi": {"chunks": 3379.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.183293, "cps": 26.398438, "mean_p": 0.201044, "boundary_entropy": 0.304339, "bpb": 2.009732}, "tr": {"chunks": 3617.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.199801, "cps": 28.257812, "mean_p": 0.225, "boundary_entropy": 0.270155, "bpb": 2.356014}, "fi": {"chunks": 3899.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.207272, "cps": 30.460938, "mean_p": 0.24224, "boundary_entropy": 0.284429, "bpb": 2.541262}, "ru": {"chunks": 3216.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.16971, "cps": 25.125, "mean_p": 0.17328, "boundary_entropy": 0.333328, "bpb": 1.525979}, "el": {"chunks": 3623.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.176534, "cps": 28.304688, "mean_p": 0.197598, "boundary_entropy": 0.385466, "bpb": 1.608489}, "hi": {"chunks": 2668.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.158574, "cps": 20.84375, "mean_p": 0.197849, "boundary_entropy": 0.407524, "bpb": 1.303135}, "ta": {"chunks": 2989.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.159882, "cps": 23.351562, "mean_p": 0.254673, "boundary_entropy": 0.518333, "bpb": 1.059146}, "th": {"chunks": 2363.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.146907, "cps": 18.460938, "mean_p": 0.185138, "boundary_entropy": 0.408457, "bpb": 1.252551}, "zh": {"chunks": 920.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.155774, "cps": 7.1875, "mean_p": 0.160079, "boundary_entropy": 0.333351, "bpb": 3.295136}, "my": {"chunks": 3272.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.172683, "cps": 25.5625, "mean_p": 0.266366, "boundary_entropy": 0.536357, "bpb": 1.164726}}, "bpb": {"en": 2.13704, "de": 2.35087, "vi": 2.00973, "tr": 2.35601, "fi": 2.54126, "ru": 1.52598, "el": 1.60849, "hi": 1.30313, "ta": 1.05915, "th": 1.25255, "zh": 3.29514, "my": 1.16473}, "gini_cpc": 0.062373, "gini_cps": 0.131785, "bpic": 10.230068, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9906038040378636, "vi": 0.9139948382102998, "tr": 0.99631495800772, "fi": 1.033570356550814, "ru": 0.8462633230262113, "el": 0.8802908386942206, "hi": 0.7907322385759354, "ta": 0.7972584643703671, "th": 0.7325568768353651, "zh": 0.7767711336244915, "my": 0.861090131761606}, "t": 184.1}
{"event": "eval", "step": 3200, "phase": 2, "bpb": {"en": 2.14554, "de": 2.35617, "vi": 2.01192, "tr": 2.36453, "fi": 2.55077, "ru": 1.50353, "el": 1.60207, "hi": 1.29925, "ta": 1.06542, "th": 1.26988, "zh": 3.29654, "my": 1.17335}, "bytes_per_chunk": {"en": 5.091, "de": 5.052, "vi": 7.22, "tr": 5.236, "fi": 4.736, "ru": 10.188, "el": 10.226, "hi": 14.747, "ta": 17.037, "th": 18.685, "zh": 16.285, "my": 16.653}, "bpb_mean": 1.88658, "t": 185.4}
{"event": "train", "step": 3200, "phase": 2, "objective": "parity_a", "lm_loss": 1.38942, "bpb": 2.00451, "chunk_loss": 0.030448, "F": 0.12302, "G": 0.20336, "chunks": 125.97, "bytes_per_chunk": 8.129, "boundary_entropy": 0.32906, "router_noise": 0.0, "grad_norm": 0.6537, "lr": 0.0001613, "router_trainable": true, "sec_per_step": 0.085, "n_target_mean": 10.661, "alpha": 0.03, "t": 185.4}
{"event": "train", "step": 3250, "phase": 2, "objective": "parity_a", "lm_loss": 1.44866, "bpb": 2.08998, "chunk_loss": 0.030435, "F": 0.14224, "G": 0.20852, "chunks": 145.66, "bytes_per_chunk": 7.03, "boundary_entropy": 0.3063, "router_noise": 0.0, "grad_norm": 0.6705, "lr": 0.0001576, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 9.4865, "alpha": 0.03, "t": 187.4}
{"event": "train", "step": 3300, "phase": 2, "objective": "parity_a", "lm_loss": 1.40944, "bpb": 2.03339, "chunk_loss": 0.030317, "F": 0.14737, "G": 0.20869, "chunks": 150.91, "bytes_per_chunk": 6.786, "boundary_entropy": 0.29761, "router_noise": 0.0, "grad_norm": 0.6989, "lr": 0.0001539, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 9.0267, "alpha": 0.03, "t": 189.4}
{"event": "train", "step": 3350, "phase": 2, "objective": "parity_a", "lm_loss": 1.47867, "bpb": 2.13327, "chunk_loss": 0.030453, "F": 0.13791, "G": 0.20963, "chunks": 141.22, "bytes_per_chunk": 7.251, "boundary_entropy": 0.31694, "router_noise": 0.0, "grad_norm": 0.6951, "lr": 0.0001501, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 9.9442, "alpha": 0.03, "t": 191.4}
{"event": "probe", "step": 3400, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 3375.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.193932, "cps": 26.367188, "mean_p": 0.221803, "boundary_entropy": 0.242884, "bpb": 2.125263}, "de": {"chunks": 3884.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.189159, "cps": 30.34375, "mean_p": 0.223706, "boundary_entropy": 0.254409, "bpb": 2.329874}, "vi": {"chunks": 3285.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.178194, "cps": 25.664062, "mean_p": 0.191512, "boundary_entropy": 0.287857, "bpb": 1.992397}, "tr": {"chunks": 3270.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.180633, "cps": 25.546875, "mean_p": 0.20706, "boundary_entropy": 0.257308, "bpb": 2.33728}, "fi": {"chunks": 3612.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.192015, "cps": 28.21875, "mean_p": 0.22176, "boundary_entropy": 0.266427, "bpb": 2.535231}, "ru": {"chunks": 3463.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.182744, "cps": 27.054688, "mean_p": 0.180799, "boundary_entropy": 0.333132, "bpb": 1.504449}, "el": {"chunks": 4265.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.207816, "cps": 33.320312, "mean_p": 0.209554, "boundary_entropy": 0.391271, "bpb": 1.595279}, "hi": {"chunks": 2612.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.155245, "cps": 20.40625, "mean_p": 0.196304, "boundary_entropy": 0.404901, "bpb": 1.292387}, "ta": {"chunks": 2652.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.141856, "cps": 20.71875, "mean_p": 0.237465, "boundary_entropy": 0.494567, "bpb": 1.057623}, "th": {"chunks": 2613.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.162449, "cps": 20.414062, "mean_p": 0.192169, "boundary_entropy": 0.416193, "bpb": 1.21274}, "zh": {"chunks": 958.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.162208, "cps": 7.484375, "mean_p": 0.15858, "boundary_entropy": 0.326666, "bpb": 3.271747}, "my": {"chunks": 2587.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.136532, "cps": 20.210938, "mean_p": 0.209707, "boundary_entropy": 0.462285, "bpb": 1.161724}}, "bpb": {"en": 2.12526, "de": 2.32987, "vi": 1.9924, "tr": 2.33728, "fi": 2.53523, "ru": 1.50445, "el": 1.59528, "hi": 1.29239, "ta": 1.05762, "th": 1.21274, "zh": 3.27175, "my": 1.16172}, "gini_cpc": 0.068349, "gini_cps": 0.140083, "bpic": 10.4927, "worst_lang": "el", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9753874359432243, "vi": 0.9188456739896935, "tr": 0.9314242574895506, "fi": 0.9901162794818635, "ru": 0.9423096100850191, "el": 1.0715897069412639, "hi": 0.8005130988938418, "ta": 0.7314731568155479, "th": 0.8376617552585225, "zh": 0.8364161493020281, "my": 0.7040174043581263}, "t": 194.3}
{"event": "eval", "step": 3400, "phase": 2, "bpb": {"en": 2.13347, "de": 2.33877, "vi": 1.99233, "tr": 2.3441, "fi": 2.54629, "ru": 1.48113, "el": 1.58771, "hi": 1.28857, "ta": 1.06374, "th": 1.23066, "zh": 3.27283, "my": 1.17157}, "bytes_per_chunk": {"en": 5.311, "de": 5.3, "vi": 7.473, "tr": 5.745, "fi": 5.146, "ru": 9.488, "el": 8.805, "hi": 15.242, "ta": 19.929, "th": 17.319, "zh": 15.942, "my": 21.597}, "bpb_mean": 1.87093, "t": 195.6}
{"event": "train", "step": 3400, "phase": 2, "objective": "parity_a", "lm_loss": 1.40455, "bpb": 2.02634, "chunk_loss": 0.030433, "F": 0.12054, "G": 0.19656, "chunks": 123.44, "bytes_per_chunk": 8.296, "boundary_entropy": 0.31963, "router_noise": 0.0, "grad_norm": 0.6725, "lr": 0.0001465, "router_trainable": true, "sec_per_step": 0.084, "n_target_mean": 11.2813, "alpha": 0.03, "t": 195.6}
{"event": "train", "step": 3450, "phase": 2, "objective": "parity_a", "lm_loss": 1.26502, "bpb": 1.82504, "chunk_loss": 0.030087, "F": 0.13486, "G": 0.20665, "chunks": 138.09, "bytes_per_chunk": 7.415, "boundary_entropy": 0.3118, "router_noise": 0.0, "grad_norm": 0.6571, "lr": 0.0001428, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 9.6015, "alpha": 0.03, "t": 197.6}
{"event": "train", "step": 3500, "phase": 2, "objective": "parity_a", "lm_loss": 1.45014, "bpb": 2.0921, "chunk_loss": 0.030399, "F": 0.14764, "G": 0.20976, "chunks": 151.19, "bytes_per_chunk": 6.773, "boundary_entropy": 0.29716, "router_noise": 0.0, "grad_norm": 0.7442, "lr": 0.0001391, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 9.5112, "alpha": 0.03, "t": 199.6}
{"event": "train", "step": 3550, "phase": 2, "objective": "parity_a", "lm_loss": 1.3829, "bpb": 1.9951, "chunk_loss": 0.030376, "F": 0.13354, "G": 0.20427, "chunks": 136.75, "bytes_per_chunk": 7.488, "boundary_entropy": 0.31255, "router_noise": 0.0, "grad_norm": 0.6655, "lr": 0.0001355, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 10.0088, "alpha": 0.03, "t": 201.6}
{"event": "probe", "step": 3600, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 3443.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.197839, "cps": 26.898438, "mean_p": 0.225641, "boundary_entropy": 0.240092, "bpb": 2.107971}, "de": {"chunks": 4009.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.195247, "cps": 31.320312, "mean_p": 0.230093, "boundary_entropy": 0.2521, "bpb": 2.315332}, "vi": {"chunks": 3300.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.179007, "cps": 25.78125, "mean_p": 0.188696, "boundary_entropy": 0.283986, "bpb": 1.992636}, "tr": {"chunks": 3461.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.191184, "cps": 27.039062, "mean_p": 0.215487, "boundary_entropy": 0.254195, "bpb": 2.31116}, "fi": {"chunks": 3690.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.196162, "cps": 28.828125, "mean_p": 0.227859, "boundary_entropy": 0.266412, "bpb": 2.483846}, "ru": {"chunks": 3309.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.174617, "cps": 25.851562, "mean_p": 0.169612, "boundary_entropy": 0.318253, "bpb": 1.493854}, "el": {"chunks": 4038.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.196755, "cps": 31.546875, "mean_p": 0.194767, "boundary_entropy": 0.364706, "bpb": 1.58082}, "hi": {"chunks": 3159.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.187756, "cps": 24.679688, "mean_p": 0.205938, "boundary_entropy": 0.414834, "bpb": 1.277686}, "ta": {"chunks": 2698.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.144317, "cps": 21.078125, "mean_p": 0.243956, "boundary_entropy": 0.504407, "bpb": 1.054003}, "th": {"chunks": 2632.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.163631, "cps": 20.5625, "mean_p": 0.196258, "boundary_entropy": 0.422087, "bpb": 1.201764}, "zh": {"chunks": 914.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.154758, "cps": 7.140625, "mean_p": 0.15452, "boundary_entropy": 0.324514, "bpb": 3.28269}, "my": {"chunks": 2235.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.117954, "cps": 17.460938, "mean_p": 0.222211, "boundary_entropy": 0.482744, "bpb": 1.154005}}, "bpb": {"en": 2.10797, "de": 2.31533, "vi": 1.99264, "tr": 2.31116, "fi": 2.48385, "ru": 1.49385, "el": 1.58082, "hi": 1.27769, "ta": 1.054, "th": 1.20176, "zh": 3.28269, "my": 1.154}, "gini_cpc": 0.073719, "gini_cps": 0.140994, "bpic": 10.403953, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9868945407673759, "vi": 0.904811049269907, "tr": 0.9663582203933926, "fi": 0.9915202315254422, "ru": 0.882621800801136, "el": 0.9945178145658896, "hi": 0.9490337351363038, "ta": 0.7294635121771078, "th": 0.8270883758958791, "zh": 0.7822396950029108, "my": 0.5962127358848905}, "t": 204.5}
{"event": "eval", "step": 3600, "phase": 2, "bpb": {"en": 2.11707, "de": 2.32279, "vi": 1.99331, "tr": 2.31933, "fi": 2.49624, "ru": 1.47131, "el": 1.57277, "hi": 1.27553, "ta": 1.05875, "th": 1.2199, "zh": 3.28201, "my": 1.1633}, "bytes_per_chunk": {"en": 5.222, "de": 5.137, "vi": 7.381, "tr": 5.427, "fi": 4.996, "ru": 9.899, "el": 9.467, "hi": 13.138, "ta": 19.264, "th": 17.296, "zh": 16.398, "my": 23.496}, "bpb_mean": 1.85769, "t": 205.8}
{"event": "train", "step": 3600, "phase": 2, "objective": "parity_a", "lm_loss": 1.64123, "bpb": 2.36779, "chunk_loss": 0.030365, "F": 0.13391, "G": 0.20093, "chunks": 137.12, "bytes_per_chunk": 7.468, "boundary_entropy": 0.29797, "router_noise": 0.0, "grad_norm": 0.7459, "lr": 0.0001319, "router_trainable": true, "sec_per_step": 0.085, "n_target_mean": 10.7476, "alpha": 0.03, "t": 205.9}
{"event": "train", "step": 3650, "phase": 2, "objective": "parity_a", "lm_loss": 1.44575, "bpb": 2.08578, "chunk_loss": 0.030702, "F": 0.1405, "G": 0.20465, "chunks": 143.88, "bytes_per_chunk": 7.117, "boundary_entropy": 0.30236, "router_noise": 0.0, "grad_norm": 0.7641, "lr": 0.0001283, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 10.1549, "alpha": 0.03, "t": 207.9}
{"event": "train", "step": 3700, "phase": 2, "objective": "parity_a", "lm_loss": 1.36019, "bpb": 1.96235, "chunk_loss": 0.031085, "F": 0.13431, "G": 0.20955, "chunks": 137.53, "bytes_per_chunk": 7.446, "boundary_entropy": 0.32413, "router_noise": 0.0, "grad_norm": 0.702, "lr": 0.0001247, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 10.6034, "alpha": 0.03, "t": 209.9}
{"event": "train", "step": 3750, "phase": 2, "objective": "parity_a", "lm_loss": 1.40223, "bpb": 2.02299, "chunk_loss": 0.030436, "F": 0.13458, "G": 0.19919, "chunks": 137.81, "bytes_per_chunk": 7.43, "boundary_entropy": 0.30049, "router_noise": 0.0, "grad_norm": 0.6895, "lr": 0.0001212, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 10.3202, "alpha": 0.03, "t": 211.9}
{"event": "probe", "step": 3800, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 3402.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.195484, "cps": 26.578125, "mean_p": 0.222936, "boundary_entropy": 0.23376, "bpb": 2.101982}, "de": {"chunks": 3960.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.19286, "cps": 30.9375, "mean_p": 0.227404, "boundary_entropy": 0.248345, "bpb": 2.294103}, "vi": {"chunks": 3288.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.178356, "cps": 25.6875, "mean_p": 0.189016, "boundary_entropy": 0.279386, "bpb": 1.973003}, "tr": {"chunks": 3487.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.19262, "cps": 27.242188, "mean_p": 0.215015, "boundary_entropy": 0.251488, "bpb": 2.293536}, "fi": {"chunks": 3661.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.19462, "cps": 28.601562, "mean_p": 0.223097, "boundary_entropy": 0.257671, "bpb": 2.470048}, "ru": {"chunks": 3226.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.170237, "cps": 25.203125, "mean_p": 0.165404, "boundary_entropy": 0.31067, "bpb": 1.474789}, "el": {"chunks": 3770.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.183696, "cps": 29.453125, "mean_p": 0.190474, "boundary_entropy": 0.361291, "bpb": 1.554291}, "hi": {"chunks": 2744.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.163091, "cps": 21.4375, "mean_p": 0.196843, "boundary_entropy": 0.406257, "bpb": 1.269871}, "ta": {"chunks": 3019.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.161487, "cps": 23.585938, "mean_p": 0.249918, "boundary_entropy": 0.510843, "bpb": 1.038023}, "th": {"chunks": 2431.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.151135, "cps": 18.992188, "mean_p": 0.182382, "boundary_entropy": 0.399888, "bpb": 1.188752}, "zh": {"chunks": 913.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.154589, "cps": 7.132812, "mean_p": 0.150268, "boundary_entropy": 0.312653, "bpb": 3.243061}, "my": {"chunks": 2364.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.124763, "cps": 18.46875, "mean_p": 0.246237, "boundary_entropy": 0.514861, "bpb": 1.148203}}, "bpb": {"en": 2.10198, "de": 2.2941, "vi": 1.973, "tr": 2.29354, "fi": 2.47005, "ru": 1.47479, "el": 1.55429, "hi": 1.26987, "ta": 1.03802, "th": 1.18875, "zh": 3.24306, "my": 1.1482}, "gini_cpc": 0.06717, "gini_cps": 0.137201, "bpic": 10.582683, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9865806417698493, "vi": 0.9123857167252881, "tr": 0.9853515563158531, "fi": 0.99558342209805, "ru": 0.8708532153211131, "el": 0.939702356567987, "hi": 0.8342934712820795, "ta": 0.8260901702017864, "th": 0.773132098600962, "zh": 0.7908008834467802, "my": 0.6382251397028682}, "t": 214.7}
{"event": "eval", "step": 3800, "phase": 2, "bpb": {"en": 2.10988, "de": 2.29979, "vi": 1.97299, "tr": 2.29963, "fi": 2.48082, "ru": 1.45267, "el": 1.54526, "hi": 1.26484, "ta": 1.04343, "th": 1.20601, "zh": 3.24209, "my": 1.15669}, "bytes_per_chunk": {"en": 5.225, "de": 5.197, "vi": 7.423, "tr": 5.415, "fi": 5.069, "ru": 10.205, "el": 9.977, "hi": 15.192, "ta": 17.562, "th": 18.739, "zh": 16.737, "my": 23.322}, "bpb_mean": 1.83951, "t": 216.0}
{"event": "train", "step": 3800, "phase": 2, "objective": "parity_a", "lm_loss": 1.47546, "bpb": 2.12864, "chunk_loss": 0.030398, "F": 0.12204, "G": 0.19402, "chunks": 124.97, "bytes_per_chunk": 8.194, "boundary_entropy": 0.30689, "router_noise": 0.0, "grad_norm": 0.6992, "lr": 0.0001177, "router_trainable": true, "sec_per_step": 0.084, "n_target_mean": 11.0501, "alpha": 0.03, "t": 216.1}
{"event": "train", "step": 3850, "phase": 2, "objective": "parity_a", "lm_loss": 1.47489, "bpb": 2.12782, "chunk_loss": 0.030693, "F": 0.12689, "G": 0.19959, "chunks": 129.94, "bytes_per_chunk": 7.881, "boundary_entropy": 0.31018, "router_noise": 0.0, "grad_norm": 0.7157, "lr": 0.0001142, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 11.2966, "alpha": 0.03, "t": 218.1}
{"event": "train", "step": 3900, "phase": 2, "objective": "parity_a", "lm_loss": 1.47526, "bpb": 2.12836, "chunk_loss": 0.030409, "F": 0.15805, "G": 0.21231, "chunks": 161.84, "bytes_per_chunk": 6.327, "boundary_entropy": 0.28294, "router_noise": 0.0, "grad_norm": 0.7489, "lr": 0.0001108, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 8.5711, "alpha": 0.03, "t": 220.1}
{"event": "train", "step": 3950, "phase": 2, "objective": "parity_a", "lm_loss": 1.55385, "bpb": 2.24173, "chunk_loss": 0.030723, "F": 0.12967, "G": 0.20147, "chunks": 132.78, "bytes_per_chunk": 7.712, "boundary_entropy": 0.30983, "router_noise": 0.0, "grad_norm": 0.6879, "lr": 0.0001074, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 11.3417, "alpha": 0.03, "t": 222.1}
{"event": "probe", "step": 4000, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 3495.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.200827, "cps": 27.304688, "mean_p": 0.227162, "boundary_entropy": 0.234429, "bpb": 2.078629}, "de": {"chunks": 4056.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.197536, "cps": 31.6875, "mean_p": 0.230419, "boundary_entropy": 0.245324, "bpb": 2.284686}, "vi": {"chunks": 3199.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.173529, "cps": 24.992188, "mean_p": 0.184393, "boundary_entropy": 0.27439, "bpb": 1.966109}, "tr": {"chunks": 3535.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.195272, "cps": 27.617188, "mean_p": 0.216775, "boundary_entropy": 0.25032, "bpb": 2.281128}, "fi": {"chunks": 3735.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.198554, "cps": 29.179688, "mean_p": 0.227463, "boundary_entropy": 0.256838, "bpb": 2.445928}, "ru": {"chunks": 3278.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.172982, "cps": 25.609375, "mean_p": 0.165858, "boundary_entropy": 0.305742, "bpb": 1.464684}, "el": {"chunks": 3711.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.180822, "cps": 28.992188, "mean_p": 0.18961, "boundary_entropy": 0.357387, "bpb": 1.542378}, "hi": {"chunks": 2247.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.133551, "cps": 17.554688, "mean_p": 0.190162, "boundary_entropy": 0.401164, "bpb": 1.263534}, "ta": {"chunks": 2718.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.145386, "cps": 21.234375, "mean_p": 0.226147, "boundary_entropy": 0.479719, "bpb": 1.037611}, "th": {"chunks": 2572.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.159901, "cps": 20.09375, "mean_p": 0.189449, "boundary_entropy": 0.408941, "bpb": 1.176934}, "zh": {"chunks": 946.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.160176, "cps": 7.390625, "mean_p": 0.152525, "boundary_entropy": 0.311359, "bpb": 3.21915}, "my": {"chunks": 2828.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.149251, "cps": 22.09375, "mean_p": 0.241818, "boundary_entropy": 0.507545, "bpb": 1.140476}}, "bpb": {"en": 2.07863, "de": 2.28469, "vi": 1.96611, "tr": 2.28113, "fi": 2.44593, "ru": 1.46468, "el": 1.54238, "hi": 1.26353, "ta": 1.03761, "th": 1.17693, "zh": 3.21915, "my": 1.14048}, "gini_cpc": 0.072375, "gini_cps": 0.141657, "bpic": 10.566657, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9836089669534376, "vi": 0.8640682318457962, "tr": 0.9723347492495849, "fi": 0.9886798040503649, "ru": 0.8613440837079733, "el": 0.9003825081417081, "hi": 0.6650050443533214, "ta": 0.7239372488340621, "th": 0.7962085540736511, "zh": 0.7975806955609265, "my": 0.7431782126098896}, "t": 225.0}
{"event": "eval", "step": 4000, "phase": 2, "bpb": {"en": 2.08859, "de": 2.29081, "vi": 1.96689, "tr": 2.28816, "fi": 2.45849, "ru": 1.44116, "el": 1.53549, "hi": 1.25978, "ta": 1.04341, "th": 1.1938, "zh": 3.21806, "my": 1.14971}, "bytes_per_chunk": {"en": 5.136, "de": 5.117, "vi": 7.691, "tr": 5.339, "fi": 4.918, "ru": 9.887, "el": 10.021, "hi": 18.231, "ta": 18.948, "th": 17.405, "zh": 16.283, "my": 18.829}, "bpb_mean": 1.82786, "t": 226.3}
{"event": "train", "step": 4000, "phase": 2, "objective": "parity_a", "lm_loss": 1.34607, "bpb": 1.94198, "chunk_loss": 0.030203, "F": 0.14404, "G": 0.20495, "chunks": 147.5, "bytes_per_chunk": 6.942, "boundary_entropy": 0.28867, "router_noise": 0.0, "grad_norm": 0.723, "lr": 0.000104, "router_trainable": true, "sec_per_step": 0.085, "n_target_mean": 9.3194, "alpha": 0.03, "t": 226.3}
{"event": "train", "step": 4050, "phase": 2, "objective": "parity_a", "lm_loss": 1.53651, "bpb": 2.21671, "chunk_loss": 0.030601, "F": 0.14008, "G": 0.20481, "chunks": 143.44, "bytes_per_chunk": 7.139, "boundary_entropy": 0.29322, "router_noise": 0.0, "grad_norm": 0.7086, "lr": 0.0001007, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 10.4285, "alpha": 0.03, "t": 228.3}
{"event": "train", "step": 4100, "phase": 2, "objective": "parity_a", "lm_loss": 1.40584, "bpb": 2.0282, "chunk_loss": 0.030454, "F": 0.1481, "G": 0.20558, "chunks": 151.66, "bytes_per_chunk": 6.752, "boundary_entropy": 0.28165, "router_noise": 0.0, "grad_norm": 0.6934, "lr": 9.75e-05, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 9.101, "alpha": 0.03, "t": 230.3}
{"event": "train", "step": 4150, "phase": 2, "objective": "parity_a", "lm_loss": 1.38999, "bpb": 2.00533, "chunk_loss": 0.030351, "F": 0.16116, "G": 0.21351, "chunks": 165.03, "bytes_per_chunk": 6.205, "boundary_entropy": 0.26927, "router_noise": 0.0, "grad_norm": 0.7321, "lr": 9.43e-05, "router_trainable": true, "sec_per_step": 0.04, "n_target_mean": 8.1686, "alpha": 0.03, "t": 232.3}
{"event": "probe", "step": 4200, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3460.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.198816, "cps": 27.03125, "mean_p": 0.221982, "boundary_entropy": 0.224548, "bpb": 2.068021}, "de": {"chunks": 4058.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.197633, "cps": 31.703125, "mean_p": 0.228856, "boundary_entropy": 0.240932, "bpb": 2.26155}, "vi": {"chunks": 3307.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.179387, "cps": 25.835938, "mean_p": 0.188175, "boundary_entropy": 0.27165, "bpb": 1.951872}, "tr": {"chunks": 3412.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.188477, "cps": 26.65625, "mean_p": 0.20888, "boundary_entropy": 0.237881, "bpb": 2.257957}, "fi": {"chunks": 3729.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.198235, "cps": 29.132812, "mean_p": 0.224984, "boundary_entropy": 0.248405, "bpb": 2.426559}, "ru": {"chunks": 3343.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.176412, "cps": 26.117188, "mean_p": 0.164076, "boundary_entropy": 0.297771, "bpb": 1.451528}, "el": {"chunks": 3937.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.191834, "cps": 30.757812, "mean_p": 0.191379, "boundary_entropy": 0.352644, "bpb": 1.525371}, "hi": {"chunks": 2975.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.17682, "cps": 23.242188, "mean_p": 0.204868, "boundary_entropy": 0.41373, "bpb": 1.247333}, "ta": {"chunks": 3240.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.173308, "cps": 25.3125, "mean_p": 0.226384, "boundary_entropy": 0.474078, "bpb": 1.019654}, "th": {"chunks": 2592.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.161144, "cps": 20.25, "mean_p": 0.194364, "boundary_entropy": 0.420116, "bpb": 1.163823}, "zh": {"chunks": 907.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.153573, "cps": 7.085938, "mean_p": 0.146583, "boundary_entropy": 0.302037, "bpb": 3.215212}, "my": {"chunks": 3565.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.188147, "cps": 27.851562, "mean_p": 0.244815, "boundary_entropy": 0.502293, "bpb": 1.138418}}, "bpb": {"en": 2.06802, "de": 2.26155, "vi": 1.95187, "tr": 2.25796, "fi": 2.42656, "ru": 1.45153, "el": 1.52537, "hi": 1.24733, "ta": 1.01965, "th": 1.16382, "zh": 3.21521, "my": 1.13842}, "gini_cpc": 0.042839, "gini_cps": 0.117233, "bpic": 9.961869, "worst_lang": "en", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9940486891396311, "vi": 0.9022753119458932, "tr": 0.947995972947351, "fi": 0.997076593666908, "ru": 0.8873096069669193, "el": 0.9648784149969623, "hi": 0.8893647630744917, "ta": 0.8717010359482226, "th": 0.8105166735245022, "zh": 0.7724348609917611, "my": 0.9463334242423133}, "t": 235.3}
{"event": "eval", "step": 4200, "phase": 2, "bpb": {"en": 2.07824, "de": 2.26873, "vi": 1.95291, "tr": 2.26617, "fi": 2.4389, "ru": 1.42909, "el": 1.5192, "hi": 1.24314, "ta": 1.02542, "th": 1.1807, "zh": 3.21433, "my": 1.14734}, "bytes_per_chunk": {"en": 5.173, "de": 5.063, "vi": 7.321, "tr": 5.539, "fi": 4.962, "ru": 9.849, "el": 9.481, "hi": 14.45, "ta": 15.695, "th": 17.323, "zh": 16.805, "my": 15.045}, "bpb_mean": 1.81368, "t": 236.6}
{"event": "phase_transition", "step": 4200, "phase": 3, "objective": "none", "router_trainable": false, "encoder_trainable": true, "frozen_router_params": 131584, "t": 236.6}
{"event": "train", "step": 4200, "phase": 3, "objective": "none", "lm_loss": 1.2053, "bpb": 1.73888, "chunk_loss": 0.0, "F": 0.12354, "G": 0.20373, "chunks": 126.5, "bytes_per_chunk": 8.095, "boundary_entropy": 0.32286, "router_noise": 0.0, "grad_norm": 0.6485, "lr": 9.12e-05, "router_trainable": false, "sec_per_step": 0.086, "alpha": 0.0, "t": 236.6}
{"event": "train", "step": 4250, "phase": 3, "objective": "none", "lm_loss": 1.25358, "bpb": 1.80854, "chunk_loss": 0.0, "F": 0.14261, "G": 0.21244, "chunks": 146.03, "bytes_per_chunk": 7.012, "boundary_entropy": 0.30143, "router_noise": 0.0, "grad_norm": 0.692, "lr": 8.81e-05, "router_trainable": false, "sec_per_step": 0.04, "alpha": 0.0, "t": 238.6}
{"event": "train", "step": 4300, "phase": 3, "objective": "none", "lm_loss": 1.48154, "bpb": 2.1374, "chunk_loss": 0.0, "F": 0.16, "G": 0.22016, "chunks": 163.84, "bytes_per_chunk": 6.25, "boundary_entropy": 0.28518, "router_noise": 0.0, "grad_norm": 0.7323, "lr": 8.5e-05, "router_trainable": false, "sec_per_step": 0.04, "alpha": 0.0, "t": 240.6}
{"event": "train", "step": 4350, "phase": 3, "objective": "none", "lm_loss": 1.46811, "bpb": 2.11804, "chunk_loss": 0.0, "F": 0.15536, "G": 0.21356, "chunks": 159.09, "bytes_per_chunk": 6.436, "boundary_entropy": 0.28238, "router_noise": 0.0, "grad_norm": 0.7577, "lr": 8.21e-05, "router_trainable": false, "sec_per_step": 0.039, "alpha": 0.0, "t": 242.6}
{"event": "probe", "step": 4400, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3588.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.206171, "cps": 28.03125, "mean_p": 0.236591, "boundary_entropy": 0.239751, "bpb": 2.055395}, "de": {"chunks": 4241.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.206546, "cps": 33.132812, "mean_p": 0.240243, "boundary_entropy": 0.249737, "bpb": 2.252112}, "vi": {"chunks": 3437.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.186439, "cps": 26.851562, "mean_p": 0.200012, "boundary_entropy": 0.281891, "bpb": 1.943308}, "tr": {"chunks": 3623.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.200133, "cps": 28.304688, "mean_p": 0.224052, "boundary_entropy": 0.252073, "bpb": 2.237987}, "fi": {"chunks": 4020.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.213705, "cps": 31.40625, "mean_p": 0.247254, "boundary_entropy": 0.267513, "bpb": 2.406206}, "ru": {"chunks": 3403.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.179578, "cps": 26.585938, "mean_p": 0.172504, "boundary_entropy": 0.310438, "bpb": 1.440075}, "el": {"chunks": 4280.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.208547, "cps": 33.4375, "mean_p": 0.204434, "boundary_entropy": 0.363349, "bpb": 1.511546}, "hi": {"chunks": 3444.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.204695, "cps": 26.90625, "mean_p": 0.211982, "boundary_entropy": 0.419414, "bpb": 1.235728}, "ta": {"chunks": 3606.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.192886, "cps": 28.171875, "mean_p": 0.240735, "boundary_entropy": 0.491372, "bpb": 1.014139}, "th": {"chunks": 3005.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.18682, "cps": 23.476562, "mean_p": 0.207537, "boundary_entropy": 0.43557, "bpb": 1.152778}, "zh": {"chunks": 954.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.161531, "cps": 7.453125, "mean_p": 0.156772, "boundary_entropy": 0.318436, "bpb": 3.193284}, "my": {"chunks": 2915.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.153842, "cps": 22.773438, "mean_p": 0.244782, "boundary_entropy": 0.505542, "bpb": 1.129362}}, "bpb": {"en": 2.0554, "de": 2.25211, "vi": 1.94331, "tr": 2.23799, "fi": 2.40621, "ru": 1.44008, "el": 1.51155, "hi": 1.23573, "ta": 1.01414, "th": 1.15278, "zh": 3.19328, "my": 1.12936}, "gini_cpc": 0.051789, "gini_cps": 0.118011, "bpic": 9.472332, "worst_lang": "fi", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0018150486849866, "vi": 0.9042907240752786, "tr": 0.9707099213216384, "fi": 1.0365394971881001, "ru": 0.8710125660733667, "el": 1.0115203156520534, "hi": 0.9928411197304537, "ta": 0.9355606322581915, "th": 0.9061395720685363, "zh": 0.7834776606070353, "my": 0.7461856088446254}, "t": 245.4}
{"event": "eval", "step": 4400, "phase": 3, "bpb": {"en": 2.06561, "de": 2.25799, "vi": 1.94446, "tr": 2.24566, "fi": 2.4184, "ru": 1.41655, "el": 1.50457, "hi": 1.23226, "ta": 1.01969, "th": 1.1708, "zh": 3.19306, "my": 1.13734}, "bytes_per_chunk": {"en": 5.003, "de": 4.874, "vi": 7.121, "tr": 5.216, "fi": 4.575, "ru": 9.633, "el": 8.685, "hi": 12.118, "ta": 13.852, "th": 14.76, "zh": 15.724, "my": 18.184}, "bpb_mean": 1.80053, "t": 246.7}
{"event": "train", "step": 4400, "phase": 3, "objective": "none", "lm_loss": 1.48785, "bpb": 2.14651, "chunk_loss": 0.0, "F": 0.15179, "G": 0.21462, "chunks": 155.44, "bytes_per_chunk": 6.588, "boundary_entropy": 0.29243, "router_noise": 0.0, "grad_norm": 0.7547, "lr": 7.92e-05, "router_trainable": false, "sec_per_step": 0.084, "alpha": 0.0, "t": 246.8}
{"event": "train", "step": 4450, "phase": 3, "objective": "none", "lm_loss": 1.50065, "bpb": 2.16498, "chunk_loss": 0.0, "F": 0.13223, "G": 0.20684, "chunks": 135.41, "bytes_per_chunk": 7.562, "boundary_entropy": 0.31063, "router_noise": 0.0, "grad_norm": 0.7059, "lr": 7.63e-05, "router_trainable": false, "sec_per_step": 0.04, "alpha": 0.0, "t": 248.8}
{"event": "train", "step": 4500, "phase": 3, "objective": "none", "lm_loss": 1.42033, "bpb": 2.0491, "chunk_loss": 0.0, "F": 0.16302, "G": 0.22356, "chunks": 166.94, "bytes_per_chunk": 6.134, "boundary_entropy": 0.28952, "router_noise": 0.0, "grad_norm": 0.7763, "lr": 7.36e-05, "router_trainable": false, "sec_per_step": 0.04, "alpha": 0.0, "t": 250.8}
{"event": "train", "step": 4550, "phase": 3, "objective": "none", "lm_loss": 1.23155, "bpb": 1.77675, "chunk_loss": 0.0, "F": 0.14868, "G": 0.21627, "chunks": 152.25, "bytes_per_chunk": 6.726, "boundary_entropy": 0.30494, "router_noise": 0.0, "grad_norm": 0.7386, "lr": 7.09e-05, "router_trainable": false, "sec_per_step": 0.04, "alpha": 0.0, "t": 252.7}
{"event": "probe", "step": 4600, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3564.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.204792, "cps": 27.84375, "mean_p": 0.234028, "boundary_entropy": 0.239854, "bpb": 2.046788}, "de": {"chunks": 4174.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.203283, "cps": 32.609375, "mean_p": 0.237919, "boundary_entropy": 0.252331, "bpb": 2.242251}, "vi": {"chunks": 3514.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.190616, "cps": 27.453125, "mean_p": 0.204479, "boundary_entropy": 0.286492, "bpb": 1.937615}, "tr": {"chunks": 3591.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.198365, "cps": 28.054688, "mean_p": 0.222683, "boundary_entropy": 0.253818, "bpb": 2.230816}, "fi": {"chunks": 3947.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.209824, "cps": 30.835938, "mean_p": 0.242706, "boundary_entropy": 0.264068, "bpb": 2.38786}, "ru": {"chunks": 3528.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.186174, "cps": 27.5625, "mean_p": 0.178245, "boundary_entropy": 0.316828, "bpb": 1.42906}, "el": {"chunks": 4092.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.199386, "cps": 31.96875, "mean_p": 0.202544, "boundary_entropy": 0.367226, "bpb": 1.505108}, "hi": {"chunks": 3667.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.217949, "cps": 28.648438, "mean_p": 0.220226, "boundary_entropy": 0.427271, "bpb": 1.229043}, "ta": {"chunks": 4105.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.219577, "cps": 32.070312, "mean_p": 0.248164, "boundary_entropy": 0.50003, "bpb": 1.011837}, "th": {"chunks": 3132.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.194716, "cps": 24.46875, "mean_p": 0.208247, "boundary_entropy": 0.43317, "bpb": 1.148822}, "zh": {"chunks": 984.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.16661, "cps": 7.6875, "mean_p": 0.159924, "boundary_entropy": 0.321409, "bpb": 3.177645}, "my": {"chunks": 2266.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.11959, "cps": 17.703125, "mean_p": 0.242891, "boundary_entropy": 0.50696, "bpb": 1.122194}}, "bpb": {"en": 2.04679, "de": 2.24225, "vi": 1.93761, "tr": 2.23082, "fi": 2.38786, "ru": 1.42906, "el": 1.50511, "hi": 1.22904, "ta": 1.01184, "th": 1.14882, "zh": 3.17764, "my": 1.12219}, "gini_cpc": 0.065814, "gini_cps": 0.125255, "bpic": 9.461123, "worst_lang": "ta", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9926278647118272, "vi": 0.9307757075318779, "tr": 0.9686151968784682, "fi": 1.0245700765742232, "ru": 0.9090877108819061, "el": 0.9736014336218984, "hi": 1.0642465767131444, "ta": 1.0721958373068297, "th": 0.950795489869161, "zh": 0.8135571769164726, "my": 0.5839598042205177}, "t": 255.6}
{"event": "eval", "step": 4600, "phase": 3, "bpb": {"en": 2.05631, "de": 2.24819, "vi": 1.93758, "tr": 2.23765, "fi": 2.40023, "ru": 1.40595, "el": 1.49803, "hi": 1.22533, "ta": 1.01759, "th": 1.16618, "zh": 3.17689, "my": 1.13079}, "bytes_per_chunk": {"en": 5.064, "de": 4.967, "vi": 6.992, "tr": 5.29, "fi": 4.683, "ru": 9.3, "el": 9.102, "hi": 11.746, "ta": 12.304, "th": 14.087, "zh": 15.344, "my": 23.25}, "bpb_mean": 1.79173, "t": 256.9}
{"event": "train", "step": 4600, "phase": 3, "objective": "none", "lm_loss": 1.30821, "bpb": 1.88735, "chunk_loss": 0.0, "F": 0.13507, "G": 0.21512, "chunks": 138.31, "bytes_per_chunk": 7.404, "boundary_entropy": 0.3265, "router_noise": 0.0, "grad_norm": 0.7236, "lr": 6.82e-05, "router_trainable": false, "sec_per_step": 0.085, "alpha": 0.0, "t": 257.0}
{"event": "train", "step": 4650, "phase": 3, "objective": "none", "lm_loss": 1.41113, "bpb": 2.03582, "chunk_loss": 0.0, "F": 0.13339, "G": 0.21, "chunks": 136.59, "bytes_per_chunk": 7.497, "boundary_entropy": 0.31493, "router_noise": 0.0, "grad_norm": 0.7036, "lr": 6.57e-05, "router_trainable": false, "sec_per_step": 0.04, "alpha": 0.0, "t": 259.0}
{"event": "train", "step": 4700, "phase": 3, "objective": "none", "lm_loss": 1.32803, "bpb": 1.91594, "chunk_loss": 0.0, "F": 0.15167, "G": 0.21589, "chunks": 155.31, "bytes_per_chunk": 6.593, "boundary_entropy": 0.29777, "router_noise": 0.0, "grad_norm": 0.7783, "lr": 6.32e-05, "router_trainable": false, "sec_per_step": 0.039, "alpha": 0.0, "t": 260.9}
{"event": "train", "step": 4750, "phase": 3, "objective": "none", "lm_loss": 1.33745, "bpb": 1.92953, "chunk_loss": 0.0, "F": 0.15787, "G": 0.22698, "chunks": 161.66, "bytes_per_chunk": 6.334, "boundary_entropy": 0.30422, "router_noise": 0.0, "grad_norm": 0.727, "lr": 6.08e-05, "router_trainable": false, "sec_per_step": 0.039, "alpha": 0.0, "t": 262.9}
{"event": "probe", "step": 4800, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3603.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.207033, "cps": 28.148438, "mean_p": 0.238708, "boundary_entropy": 0.240886, "bpb": 2.037509}, "de": {"chunks": 4244.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.206692, "cps": 33.15625, "mean_p": 0.24069, "boundary_entropy": 0.251965, "bpb": 2.228681}, "vi": {"chunks": 3499.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.189802, "cps": 27.335938, "mean_p": 0.20415, "boundary_entropy": 0.286172, "bpb": 1.931921}, "tr": {"chunks": 3566.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.196984, "cps": 27.859375, "mean_p": 0.220501, "boundary_entropy": 0.252257, "bpb": 2.216348}, "fi": {"chunks": 3910.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.207857, "cps": 30.546875, "mean_p": 0.241397, "boundary_entropy": 0.264867, "bpb": 2.375919}, "ru": {"chunks": 3578.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.188813, "cps": 27.953125, "mean_p": 0.178704, "boundary_entropy": 0.317443, "bpb": 1.415432}, "el": {"chunks": 4553.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.221849, "cps": 35.570312, "mean_p": 0.215068, "boundary_entropy": 0.377271, "bpb": 1.48657}, "hi": {"chunks": 4565.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.271322, "cps": 35.664062, "mean_p": 0.229948, "boundary_entropy": 0.435986, "bpb": 1.223888}, "ta": {"chunks": 5482.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.293233, "cps": 42.828125, "mean_p": 0.26661, "boundary_entropy": 0.51357, "bpb": 1.000556}, "th": {"chunks": 3491.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.217035, "cps": 27.273438, "mean_p": 0.21234, "boundary_entropy": 0.433699, "bpb": 1.143825}, "zh": {"chunks": 974.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.164917, "cps": 7.609375, "mean_p": 0.159734, "boundary_entropy": 0.322028, "bpb": 3.172399}, "my": {"chunks": 2869.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.151414, "cps": 22.414062, "mean_p": 0.236894, "boundary_entropy": 0.496872, "bpb": 1.116019}}, "bpb": {"en": 2.03751, "de": 2.22868, "vi": 1.93192, "tr": 2.21635, "fi": 2.37592, "ru": 1.41543, "el": 1.48657, "hi": 1.22389, "ta": 1.00056, "th": 1.14383, "zh": 3.1724, "my": 1.11602}, "gini_cpc": 0.096613, "gini_cps": 0.144862, "bpic": 8.656584, "worst_lang": "ta", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9983500088989786, "vi": 0.9167705602893529, "tr": 0.9514602420978705, "fi": 1.0039792385016102, "ru": 0.9119918976929955, "el": 1.071560418616866, "hi": 1.3105257753259107, "ta": 1.4163592388352488, "th": 1.04830737622285, "zh": 0.796572615720109, "my": 0.7313529714612251}, "t": 265.8}
{"event": "eval", "step": 4800, "phase": 3, "bpb": {"en": 2.04716, "de": 2.23537, "vi": 1.93118, "tr": 2.22207, "fi": 2.38697, "ru": 1.39309, "el": 1.47865, "hi": 1.22015, "ta": 1.00656, "th": 1.16053, "zh": 3.17001, "my": 1.12439}, "bytes_per_chunk": {"en": 4.97, "de": 4.867, "vi": 6.991, "tr": 5.323, "fi": 4.727, "ru": 9.128, "el": 8.298, "hi": 9.438, "ta": 9.142, "th": 12.806, "zh": 15.38, "my": 18.5}, "bpb_mean": 1.78135, "t": 267.1}
{"event": "train", "step": 4800, "phase": 3, "objective": "none", "lm_loss": 1.26179, "bpb": 1.82038, "chunk_loss": 0.0, "F": 0.14908, "G": 0.22276, "chunks": 152.66, "bytes_per_chunk": 6.708, "boundary_entropy": 0.31696, "router_noise": 0.0, "grad_norm": 0.7005, "lr": 5.85e-05, "router_trainable": false, "sec_per_step": 0.085, "alpha": 0.0, "t": 267.1}
{"event": "train", "step": 4850, "phase": 3, "objective": "none", "lm_loss": 1.49391, "bpb": 2.15526, "chunk_loss": 0.0, "F": 0.15787, "G": 0.21927, "chunks": 161.66, "bytes_per_chunk": 6.334, "boundary_entropy": 0.28786, "router_noise": 0.0, "grad_norm": 0.7616, "lr": 5.62e-05, "router_trainable": false, "sec_per_step": 0.04, "alpha": 0.0, "t": 269.1}
{"event": "train", "step": 4900, "phase": 3, "objective": "none", "lm_loss": 1.41555, "bpb": 2.0422, "chunk_loss": 0.0, "F": 0.12842, "G": 0.21075, "chunks": 131.5, "bytes_per_chunk": 7.787, "boundary_entropy": 0.3263, "router_noise": 0.0, "grad_norm": 0.7094, "lr": 5.41e-05, "router_trainable": false, "sec_per_step": 0.04, "alpha": 0.0, "t": 271.1}
{"event": "train", "step": 4950, "phase": 3, "objective": "none", "lm_loss": 1.52426, "bpb": 2.19904, "chunk_loss": 0.0, "F": 0.16794, "G": 0.22716, "chunks": 171.97, "bytes_per_chunk": 5.955, "boundary_entropy": 0.29162, "router_noise": 0.0, "grad_norm": 0.7685, "lr": 5.2e-05, "router_trainable": false, "sec_per_step": 0.04, "alpha": 0.0, "t": 273.1}
{"event": "probe", "step": 5000, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3645.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.209447, "cps": 28.476562, "mean_p": 0.240765, "boundary_entropy": 0.24177, "bpb": 2.027962}, "de": {"chunks": 4261.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.20752, "cps": 33.289062, "mean_p": 0.240102, "boundary_entropy": 0.248146, "bpb": 2.215921}, "vi": {"chunks": 3573.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.193816, "cps": 27.914062, "mean_p": 0.208024, "boundary_entropy": 0.287497, "bpb": 1.926861}, "tr": {"chunks": 3680.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.203281, "cps": 28.75, "mean_p": 0.226186, "boundary_entropy": 0.252368, "bpb": 2.207591}, "fi": {"chunks": 3960.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.210515, "cps": 30.9375, "mean_p": 0.243586, "boundary_entropy": 0.263154, "bpb": 2.36054}, "ru": {"chunks": 3861.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.203747, "cps": 30.164062, "mean_p": 0.187949, "boundary_entropy": 0.32867, "bpb": 1.406631}, "el": {"chunks": 4679.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.227988, "cps": 36.554688, "mean_p": 0.215153, "boundary_entropy": 0.373847, "bpb": 1.477264}, "hi": {"chunks": 4700.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.279346, "cps": 36.71875, "mean_p": 0.232061, "boundary_entropy": 0.437538, "bpb": 1.221836}, "ta": {"chunks": 3372.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.180369, "cps": 26.34375, "mean_p": 0.248094, "boundary_entropy": 0.502604, "bpb": 1.002566}, "th": {"chunks": 3499.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.217532, "cps": 27.335938, "mean_p": 0.222098, "boundary_entropy": 0.449338, "bpb": 1.135474}, "zh": {"chunks": 973.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.164748, "cps": 7.601562, "mean_p": 0.160653, "boundary_entropy": 0.323257, "bpb": 3.171426}, "my": {"chunks": 2481.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.130937, "cps": 19.382812, "mean_p": 0.244113, "boundary_entropy": 0.508106, "bpb": 1.112875}}, "bpb": {"en": 2.02796, "de": 2.21592, "vi": 1.92686, "tr": 2.20759, "fi": 2.36054, "ru": 1.40663, "el": 1.47726, "hi": 1.22184, "ta": 1.00257, "th": 1.13547, "zh": 3.17143, "my": 1.11288}, "gini_cpc": 0.088078, "gini_cps": 0.137065, "bpic": 8.991215, "worst_lang": "hi", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9907993536050268, "vi": 0.92537222875167, "tr": 0.9705632765776265, "fi": 1.00510142804545, "ru": 0.97278569334506, "el": 1.088525951712512, "hi": 1.3337344608931203, "ta": 0.8611695871676259, "th": 1.0386027418638892, "zh": 0.7865855884857278, "my": 0.6251582629001531}, "t": 276.0}
{"event": "eval", "step": 5000, "phase": 3, "bpb": {"en": 2.03723, "de": 2.2236, "vi": 1.92633, "tr": 2.21506, "fi": 2.37262, "ru": 1.38262, "el": 1.4697, "hi": 1.21778, "ta": 1.00892, "th": 1.1531, "zh": 3.16858, "my": 1.12211}, "bytes_per_chunk": {"en": 4.92, "de": 4.845, "vi": 6.769, "tr": 5.147, "fi": 4.667, "ru": 8.472, "el": 8.062, "hi": 9.195, "ta": 14.941, "th": 12.841, "zh": 15.423, "my": 20.997}, "bpb_mean": 1.7748, "t": 277.3}
{"event": "train", "step": 5000, "phase": 3, "objective": "none", "lm_loss": 1.29441, "bpb": 1.86744, "chunk_loss": 0.0, "F": 0.14899, "G": 0.22418, "chunks": 152.56, "bytes_per_chunk": 6.712, "boundary_entropy": 0.32419, "router_noise": 0.0, "grad_norm": 0.7416, "lr": 5e-05, "router_trainable": false, "sec_per_step": 0.085, "alpha": 0.0, "t": 277.3}
{"event": "train", "step": 5050, "phase": 3, "objective": "none", "lm_loss": 1.26994, "bpb": 1.83214, "chunk_loss": 0.0, "F": 0.15472, "G": 0.22236, "chunks": 158.44, "bytes_per_chunk": 6.463, "boundary_entropy": 0.30784, "router_noise": 0.0, "grad_norm": 0.7273, "lr": 4.81e-05, "router_trainable": false, "sec_per_step": 0.039, "alpha": 0.0, "t": 279.3}
{"event": "train", "step": 5100, "phase": 3, "objective": "none", "lm_loss": 1.31216, "bpb": 1.89304, "chunk_loss": 0.0, "F": 0.15311, "G": 0.22378, "chunks": 156.78, "bytes_per_chunk": 6.531, "boundary_entropy": 0.31054, "router_noise": 0.0, "grad_norm": 0.7399, "lr": 4.63e-05, "router_trainable": false, "sec_per_step": 0.04, "alpha": 0.0, "t": 281.3}
{"event": "train", "step": 5150, "phase": 3, "objective": "none", "lm_loss": 1.34067, "bpb": 1.93418, "chunk_loss": 0.0, "F": 0.15729, "G": 0.22106, "chunks": 161.06, "bytes_per_chunk": 6.358, "boundary_entropy": 0.29984, "router_noise": 0.0, "grad_norm": 0.7363, "lr": 4.45e-05, "router_trainable": false, "sec_per_step": 0.04, "alpha": 0.0, "t": 283.3}
{"event": "probe", "step": 5200, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3618.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.207895, "cps": 28.265625, "mean_p": 0.240082, "boundary_entropy": 0.241468, "bpb": 2.022785}, "de": {"chunks": 4292.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.209029, "cps": 33.53125, "mean_p": 0.242276, "boundary_entropy": 0.252632, "bpb": 2.207538}, "vi": {"chunks": 3583.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.194359, "cps": 27.992188, "mean_p": 0.208628, "boundary_entropy": 0.288108, "bpb": 1.9216}, "tr": {"chunks": 3640.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.201072, "cps": 28.4375, "mean_p": 0.226557, "boundary_entropy": 0.258642, "bpb": 2.197969}, "fi": {"chunks": 4002.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.212748, "cps": 31.265625, "mean_p": 0.246579, "boundary_entropy": 0.265538, "bpb": 2.347291}, "ru": {"chunks": 3917.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.206702, "cps": 30.601562, "mean_p": 0.188083, "boundary_entropy": 0.325773, "bpb": 1.399511}, "el": {"chunks": 4807.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.234225, "cps": 37.554688, "mean_p": 0.219223, "boundary_entropy": 0.374205, "bpb": 1.472444}, "hi": {"chunks": 4657.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.27679, "cps": 36.382812, "mean_p": 0.231834, "boundary_entropy": 0.435871, "bpb": 1.207482}, "ta": {"chunks": 4403.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.235518, "cps": 34.398438, "mean_p": 0.251876, "boundary_entropy": 0.502703, "bpb": 0.995112}, "th": {"chunks": 3547.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.220516, "cps": 27.710938, "mean_p": 0.221839, "boundary_entropy": 0.450094, "bpb": 1.125121}, "zh": {"chunks": 1009.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.170843, "cps": 7.882812, "mean_p": 0.166114, "boundary_entropy": 0.331862, "bpb": 3.145491}, "my": {"chunks": 2742.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.144712, "cps": 21.421875, "mean_p": 0.243912, "boundary_entropy": 0.505922, "bpb": 1.109871}}, "bpb": {"en": 2.02278, "de": 2.20754, "vi": 1.9216, "tr": 2.19797, "fi": 2.34729, "ru": 1.39951, "el": 1.47244, "hi": 1.20748, "ta": 0.99511, "th": 1.12512, "zh": 3.14549, "my": 1.10987}, "gini_cpc": 0.080552, "gini_cps": 0.131886, "bpic": 8.67949, "worst_lang": "hi", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0054555224332888, "vi": 0.9348872219680909, "tr": 0.9671779569075392, "fi": 1.023341905114796, "ru": 0.994259879144296, "el": 1.1266495164709835, "hi": 1.3313943901098766, "ta": 1.1328666021767777, "th": 1.060707600738407, "zh": 0.8217756740622432, "my": 0.6960807635875929}, "t": 286.1}
{"event": "eval", "step": 5200, "phase": 3, "bpb": {"en": 2.03228, "de": 2.21384, "vi": 1.92082, "tr": 2.20461, "fi": 2.35932, "ru": 1.376, "el": 1.46451, "hi": 1.2037, "ta": 1.00151, "th": 1.14335, "zh": 3.145, "my": 1.11804}, "bytes_per_chunk": {"en": 4.948, "de": 4.808, "vi": 6.813, "tr": 5.224, "fi": 4.625, "ru": 8.348, "el": 7.824, "hi": 9.65, "ta": 11.72, "th": 12.621, "zh": 14.843, "my": 19.314}, "bpb_mean": 1.76525, "t": 287.5}
{"event": "train", "step": 5200, "phase": 3, "objective": "none", "lm_loss": 1.28889, "bpb": 1.85947, "chunk_loss": 0.0, "F": 0.15625, "G": 0.22989, "chunks": 160.0, "bytes_per_chunk": 6.4, "boundary_entropy": 0.32256, "router_noise": 0.0, "grad_norm": 0.779, "lr": 4.29e-05, "router_trainable": false, "sec_per_step": 0.085, "alpha": 0.0, "t": 287.5}
{"event": "train", "step": 5250, "phase": 3, "objective": "none", "lm_loss": 1.37932, "bpb": 1.98994, "chunk_loss": 0.0, "F": 0.15192, "G": 0.22342, "chunks": 155.56, "bytes_per_chunk": 6.583, "boundary_entropy": 0.31144, "router_noise": 0.0, "grad_norm": 0.7613, "lr": 4.14e-05, "router_trainable": false, "sec_per_step": 0.04, "alpha": 0.0, "t": 289.5}
{"event": "train", "step": 5300, "phase": 3, "objective": "none", "lm_loss": 1.38318, "bpb": 1.9955, "chunk_loss": 0.0, "F": 0.15546, "G": 0.22754, "chunks": 159.19, "bytes_per_chunk": 6.433, "boundary_entropy": 0.31519, "router_noise": 0.0, "grad_norm": 0.7631, "lr": 3.99e-05, "router_trainable": false, "sec_per_step": 0.04, "alpha": 0.0, "t": 291.5}
{"event": "train", "step": 5350, "phase": 3, "objective": "none", "lm_loss": 1.42393, "bpb": 2.05429, "chunk_loss": 0.0, "F": 0.15863, "G": 0.2256, "chunks": 162.44, "bytes_per_chunk": 6.304, "boundary_entropy": 0.30737, "router_noise": 0.0, "grad_norm": 0.7875, "lr": 3.86e-05, "router_trainable": false, "sec_per_step": 0.04, "alpha": 0.0, "t": 293.4}
{"event": "probe", "step": 5400, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3639.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.209102, "cps": 28.429688, "mean_p": 0.24121, "boundary_entropy": 0.242498, "bpb": 2.015771}, "de": {"chunks": 4321.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.210442, "cps": 33.757812, "mean_p": 0.245174, "boundary_entropy": 0.254656, "bpb": 2.198602}, "vi": {"chunks": 3576.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.193979, "cps": 27.9375, "mean_p": 0.209301, "boundary_entropy": 0.288529, "bpb": 1.912961}, "tr": {"chunks": 3635.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.200795, "cps": 28.398438, "mean_p": 0.225696, "boundary_entropy": 0.256365, "bpb": 2.190207}, "fi": {"chunks": 3982.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.211685, "cps": 31.109375, "mean_p": 0.246076, "boundary_entropy": 0.266205, "bpb": 2.340318}, "ru": {"chunks": 3793.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.200158, "cps": 29.632812, "mean_p": 0.184489, "boundary_entropy": 0.321993, "bpb": 1.394508}, "el": {"chunks": 4481.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.21834, "cps": 35.007812, "mean_p": 0.211184, "boundary_entropy": 0.373284, "bpb": 1.463615}, "hi": {"chunks": 4711.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.28, "cps": 36.804688, "mean_p": 0.233021, "boundary_entropy": 0.436328, "bpb": 1.204999}, "ta": {"chunks": 4337.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.231987, "cps": 33.882812, "mean_p": 0.250348, "boundary_entropy": 0.501122, "bpb": 0.98941}, "th": {"chunks": 3240.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.20143, "cps": 25.3125, "mean_p": 0.211303, "boundary_entropy": 0.43742, "bpb": 1.122385}, "zh": {"chunks": 1003.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.169827, "cps": 7.835938, "mean_p": 0.163098, "boundary_entropy": 0.326861, "bpb": 3.147523}, "my": {"chunks": 2738.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.144501, "cps": 21.390625, "mean_p": 0.248604, "boundary_entropy": 0.51275, "bpb": 1.105782}}, "bpb": {"en": 2.01577, "de": 2.1986, "vi": 1.91296, "tr": 2.19021, "fi": 2.34032, "ru": 1.39451, "el": 1.46362, "hi": 1.205, "ta": 0.98941, "th": 1.12238, "zh": 3.14752, "my": 1.10578}, "gini_cpc": 0.078023, "gini_cps": 0.13246, "bpic": 8.831485, "worst_lang": "hi", "cpc_parity_vs_en": {"en": 1.0, "de": 1.006407637130007, "vi": 0.9276762386326057, "tr": 0.9602756763430146, "fi": 1.0123517474975505, "ru": 0.9572286596175555, "el": 1.0441819061388065, "hi": 1.339060181368508, "ta": 1.1094456131500436, "th": 0.9633098689182391, "zh": 0.8121748838424685, "my": 0.6910542342552124}, "t": 296.3}
{"event": "eval", "step": 5400, "phase": 3, "bpb": {"en": 2.0258, "de": 2.20547, "vi": 1.91214, "tr": 2.1969, "fi": 2.35172, "ru": 1.37138, "el": 1.45695, "hi": 1.20104, "ta": 0.99581, "th": 1.14067, "zh": 3.14527, "my": 1.11507}, "bytes_per_chunk": {"en": 4.939, "de": 4.78, "vi": 6.835, "tr": 5.221, "fi": 4.653, "ru": 8.581, "el": 8.316, "hi": 9.503, "ta": 11.84, "th": 13.833, "zh": 15.156, "my": 19.738}, "bpb_mean": 1.75985, "t": 297.7}
{"event": "train", "step": 5400, "phase": 3, "objective": "none", "lm_loss": 1.13823, "bpb": 1.64212, "chunk_loss": 0.0, "F": 0.13791, "G": 0.22106, "chunks": 141.22, "bytes_per_chunk": 7.251, "boundary_entropy": 0.3385, "router_noise": 0.0, "grad_norm": 0.7271, "lr": 3.73e-05, "router_trainable": false, "sec_per_step": 0.085, "alpha": 0.0, "t": 297.7}
{"event": "train", "step": 5450, "phase": 3, "objective": "none", "lm_loss": 1.1999, "bpb": 1.73109, "chunk_loss": 0.0, "F": 0.14093, "G": 0.21857, "chunks": 144.31, "bytes_per_chunk": 7.096, "boundary_entropy": 0.32367, "router_noise": 0.0, "grad_norm": 0.7539, "lr": 3.62e-05, "router_trainable": false, "sec_per_step": 0.04, "alpha": 0.0, "t": 299.7}
{"event": "train", "step": 5500, "phase": 3, "objective": "none", "lm_loss": 1.39812, "bpb": 2.01707, "chunk_loss": 0.0, "F": 0.164, "G": 0.23197, "chunks": 167.94, "bytes_per_chunk": 6.098, "boundary_entropy": 0.30308, "router_noise": 0.0, "grad_norm": 0.7454, "lr": 3.51e-05, "router_trainable": false, "sec_per_step": 0.04, "alpha": 0.0, "t": 301.7}
{"event": "train", "step": 5550, "phase": 3, "objective": "none", "lm_loss": 1.41088, "bpb": 2.03547, "chunk_loss": 0.0, "F": 0.15863, "G": 0.2254, "chunks": 162.44, "bytes_per_chunk": 6.304, "boundary_entropy": 0.30346, "router_noise": 0.0, "grad_norm": 0.7776, "lr": 3.41e-05, "router_trainable": false, "sec_per_step": 0.04, "alpha": 0.0, "t": 303.7}
{"event": "probe", "step": 5600, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3672.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.210998, "cps": 28.6875, "mean_p": 0.244066, "boundary_entropy": 0.243162, "bpb": 2.009875}, "de": {"chunks": 4358.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.212244, "cps": 34.046875, "mean_p": 0.246347, "boundary_entropy": 0.253409, "bpb": 2.192666}, "vi": {"chunks": 3561.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.193165, "cps": 27.820312, "mean_p": 0.208456, "boundary_entropy": 0.288661, "bpb": 1.911767}, "tr": {"chunks": 3709.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.204883, "cps": 28.976562, "mean_p": 0.229721, "boundary_entropy": 0.259203, "bpb": 2.178724}, "fi": {"chunks": 4025.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.213971, "cps": 31.445312, "mean_p": 0.248941, "boundary_entropy": 0.267317, "bpb": 2.328087}, "ru": {"chunks": 3857.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.203536, "cps": 30.132812, "mean_p": 0.186779, "boundary_entropy": 0.325558, "bpb": 1.392455}, "el": {"chunks": 4561.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.222238, "cps": 35.632812, "mean_p": 0.21416, "boundary_entropy": 0.375261, "bpb": 1.455754}, "hi": {"chunks": 4450.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.264487, "cps": 34.765625, "mean_p": 0.229655, "boundary_entropy": 0.438334, "bpb": 1.19618}, "ta": {"chunks": 5235.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.280021, "cps": 40.898438, "mean_p": 0.253538, "boundary_entropy": 0.497752, "bpb": 0.985016}, "th": {"chunks": 3380.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.210134, "cps": 26.40625, "mean_p": 0.214747, "boundary_entropy": 0.440828, "bpb": 1.119289}, "zh": {"chunks": 995.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.168473, "cps": 7.773438, "mean_p": 0.165164, "boundary_entropy": 0.330458, "bpb": 3.140136}, "my": {"chunks": 2782.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.146823, "cps": 21.734375, "mean_p": 0.250695, "boundary_entropy": 0.515011, "bpb": 1.103395}}, "bpb": {"en": 2.00987, "de": 2.19267, "vi": 1.91177, "tr": 2.17872, "fi": 2.32809, "ru": 1.39245, "el": 1.45575, "hi": 1.19618, "ta": 0.98502, "th": 1.11929, "zh": 3.14014, "my": 1.10339}, "gini_cpc": 0.085868, "gini_cps": 0.140804, "bpic": 8.60785, "worst_lang": "ta", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0059033776013044, "vi": 0.9154829900498307, "tr": 0.9710190038194659, "fi": 1.0140875454252014, "ru": 0.9646324607239555, "el": 1.0532723289253982, "hi": 1.2535059095455914, "ta": 1.3271275479970006, "th": 0.9959030964991944, "zh": 0.7984561783851141, "my": 0.6958493041154257}, "t": 306.6}
{"event": "eval", "step": 5600, "phase": 3, "bpb": {"en": 2.02091, "de": 2.19949, "vi": 1.91085, "tr": 2.18497, "fi": 2.34063, "ru": 1.36763, "el": 1.44812, "hi": 1.19309, "ta": 0.99194, "th": 1.13732, "zh": 3.13861, "my": 1.11172}, "bytes_per_chunk": {"en": 4.891, "de": 4.735, "vi": 6.855, "tr": 5.145, "fi": 4.619, "ru": 8.453, "el": 8.11, "hi": 9.89, "ta": 9.909, "th": 13.414, "zh": 15.183, "my": 19.227}, "bpb_mean": 1.75377, "t": 307.9}
{"event": "train", "step": 5600, "phase": 3, "objective": "none", "lm_loss": 1.46933, "bpb": 2.11979, "chunk_loss": 0.0, "F": 0.1514, "G": 0.21534, "chunks": 155.03, "bytes_per_chunk": 6.605, "boundary_entropy": 0.29244, "router_noise": 0.0, "grad_norm": 0.7783, "lr": 3.33e-05, "router_trainable": false, "sec_per_step": 0.086, "alpha": 0.0, "t": 308.0}
{"event": "train", "step": 5650, "phase": 3, "objective": "none", "lm_loss": 1.27768, "bpb": 1.8433, "chunk_loss": 0.0, "F": 0.15012, "G": 0.22431, "chunks": 153.72, "bytes_per_chunk": 6.662, "boundary_entropy": 0.32218, "router_noise": 0.0, "grad_norm": 0.7343, "lr": 3.25e-05, "router_trainable": false, "sec_per_step": 0.041, "alpha": 0.0, "t": 310.0}
{"event": "train", "step": 5700, "phase": 3, "objective": "none", "lm_loss": 1.32415, "bpb": 1.91035, "chunk_loss": 0.0, "F": 0.16641, "G": 0.22975, "chunks": 170.41, "bytes_per_chunk": 6.009, "boundary_entropy": 0.30332, "router_noise": 0.0, "grad_norm": 0.7665, "lr": 3.18e-05, "router_trainable": false, "sec_per_step": 0.041, "alpha": 0.0, "t": 312.0}
{"event": "train", "step": 5750, "phase": 3, "objective": "none", "lm_loss": 1.21184, "bpb": 1.74832, "chunk_loss": 0.0, "F": 0.15433, "G": 0.22818, "chunks": 158.03, "bytes_per_chunk": 6.48, "boundary_entropy": 0.31705, "router_noise": 0.0, "grad_norm": 0.7289, "lr": 3.13e-05, "router_trainable": false, "sec_per_step": 0.041, "alpha": 0.0, "t": 314.1}
{"event": "probe", "step": 5800, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3664.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.210538, "cps": 28.625, "mean_p": 0.243516, "boundary_entropy": 0.243802, "bpb": 2.006761}, "de": {"chunks": 4307.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.20976, "cps": 33.648438, "mean_p": 0.245068, "boundary_entropy": 0.253845, "bpb": 2.189105}, "vi": {"chunks": 3564.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.193328, "cps": 27.84375, "mean_p": 0.211439, "boundary_entropy": 0.29377, "bpb": 1.907299}, "tr": {"chunks": 3697.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.20422, "cps": 28.882812, "mean_p": 0.230451, "boundary_entropy": 0.260195, "bpb": 2.171343}, "fi": {"chunks": 4021.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.213758, "cps": 31.414062, "mean_p": 0.248559, "boundary_entropy": 0.26787, "bpb": 2.322773}, "ru": {"chunks": 3966.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.209288, "cps": 30.984375, "mean_p": 0.190033, "boundary_entropy": 0.329391, "bpb": 1.384715}, "el": {"chunks": 4587.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.223505, "cps": 35.835938, "mean_p": 0.214088, "boundary_entropy": 0.374924, "bpb": 1.449144}, "hi": {"chunks": 4216.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.250579, "cps": 32.9375, "mean_p": 0.23027, "boundary_entropy": 0.440259, "bpb": 1.195424}, "ta": {"chunks": 5122.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.273977, "cps": 40.015625, "mean_p": 0.26105, "boundary_entropy": 0.510911, "bpb": 0.981981}, "th": {"chunks": 3824.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.237737, "cps": 29.875, "mean_p": 0.227084, "boundary_entropy": 0.456189, "bpb": 1.115157}, "zh": {"chunks": 993.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.168134, "cps": 7.757812, "mean_p": 0.16442, "boundary_entropy": 0.329531, "bpb": 3.134418}, "my": {"chunks": 2878.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.151889, "cps": 22.484375, "mean_p": 0.25311, "boundary_entropy": 0.517214, "bpb": 1.102166}}, "bpb": {"en": 2.00676, "de": 2.1891, "vi": 1.9073, "tr": 2.17134, "fi": 2.32277, "ru": 1.38472, "el": 1.44914, "hi": 1.19542, "ta": 0.98198, "th": 1.11516, "zh": 3.13442, "my": 1.10217}, "gini_cpc": 0.082016, "gini_cps": 0.129863, "bpic": 8.559089, "worst_lang": "ta", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9963022699427081, "vi": 0.9182548029604785, "tr": 0.9699906641689103, "fi": 1.0152917210941568, "ru": 0.9940589750089296, "el": 1.0615893431525436, "hi": 1.190184210697068, "ta": 1.3013159708069437, "th": 1.129185970275271, "zh": 0.7985910912150621, "my": 0.7214331074335368}, "t": 317.0}
{"event": "eval", "step": 5800, "phase": 3, "bpb": {"en": 2.01631, "de": 2.19662, "vi": 1.90563, "tr": 2.17853, "fi": 2.33504, "ru": 1.36125, "el": 1.44245, "hi": 1.19177, "ta": 0.98867, "th": 1.1324, "zh": 3.1327, "my": 1.11078}, "bytes_per_chunk": {"en": 4.904, "de": 4.786, "vi": 6.855, "tr": 5.135, "fi": 4.609, "ru": 8.221, "el": 8.109, "hi": 10.51, "ta": 9.991, "th": 11.836, "zh": 15.051, "my": 18.546}, "bpb_mean": 1.74935, "t": 318.3}
{"event": "train", "step": 5800, "phase": 3, "objective": "none", "lm_loss": 1.26338, "bpb": 1.82267, "chunk_loss": 0.0, "F": 0.15408, "G": 0.22764, "chunks": 157.78, "bytes_per_chunk": 6.49, "boundary_entropy": 0.31842, "router_noise": 0.0, "grad_norm": 0.7153, "lr": 3.08e-05, "router_trainable": false, "sec_per_step": 0.086, "alpha": 0.0, "t": 318.4}
{"event": "train", "step": 5850, "phase": 3, "objective": "none", "lm_loss": 1.36869, "bpb": 1.9746, "chunk_loss": 0.0, "F": 0.15234, "G": 0.22365, "chunks": 156.0, "bytes_per_chunk": 6.564, "boundary_entropy": 0.31271, "router_noise": 0.0, "grad_norm": 0.716, "lr": 3.05e-05, "router_trainable": false, "sec_per_step": 0.04, "alpha": 0.0, "t": 320.4}
{"event": "train", "step": 5900, "phase": 3, "objective": "none", "lm_loss": 1.3274, "bpb": 1.91503, "chunk_loss": 0.0, "F": 0.16449, "G": 0.22772, "chunks": 168.44, "bytes_per_chunk": 6.079, "boundary_entropy": 0.29899, "router_noise": 0.0, "grad_norm": 0.7527, "lr": 3.02e-05, "router_trainable": false, "sec_per_step": 0.04, "alpha": 0.0, "t": 322.4}
{"event": "train", "step": 5950, "phase": 3, "objective": "none", "lm_loss": 1.28299, "bpb": 1.85096, "chunk_loss": 0.0, "F": 0.15131, "G": 0.23093, "chunks": 154.94, "bytes_per_chunk": 6.609, "boundary_entropy": 0.32762, "router_noise": 0.0, "grad_norm": 0.7272, "lr": 3.01e-05, "router_trainable": false, "sec_per_step": 0.041, "alpha": 0.0, "t": 324.4}
{"event": "probe", "step": 6000, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3683.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.21163, "cps": 28.773438, "mean_p": 0.244535, "boundary_entropy": 0.243583, "bpb": 2.003906}, "de": {"chunks": 4314.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.210101, "cps": 33.703125, "mean_p": 0.245728, "boundary_entropy": 0.253091, "bpb": 2.184671}, "vi": {"chunks": 3540.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.192026, "cps": 27.65625, "mean_p": 0.210654, "boundary_entropy": 0.290713, "bpb": 1.904545}, "tr": {"chunks": 3695.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.20411, "cps": 28.867188, "mean_p": 0.229801, "boundary_entropy": 0.256778, "bpb": 2.169336}, "fi": {"chunks": 4059.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.215778, "cps": 31.710938, "mean_p": 0.250459, "boundary_entropy": 0.266547, "bpb": 2.319622}, "ru": {"chunks": 4038.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.213087, "cps": 31.546875, "mean_p": 0.19158, "boundary_entropy": 0.329582, "bpb": 1.381369}, "el": {"chunks": 4633.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.225747, "cps": 36.195312, "mean_p": 0.216564, "boundary_entropy": 0.376393, "bpb": 1.444697}, "hi": {"chunks": 4720.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.280535, "cps": 36.875, "mean_p": 0.236346, "boundary_entropy": 0.442511, "bpb": 1.194161}, "ta": {"chunks": 4277.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.228778, "cps": 33.414062, "mean_p": 0.253503, "boundary_entropy": 0.50727, "bpb": 0.981911}, "th": {"chunks": 3384.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.210382, "cps": 26.4375, "mean_p": 0.218198, "boundary_entropy": 0.448269, "bpb": 1.112}, "zh": {"chunks": 999.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.16915, "cps": 7.804688, "mean_p": 0.165648, "boundary_entropy": 0.332407, "bpb": 3.128703}, "my": {"chunks": 2573.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.135793, "cps": 20.101562, "mean_p": 0.249842, "boundary_entropy": 0.515969, "bpb": 1.101266}}, "bpb": {"en": 2.00391, "de": 2.18467, "vi": 1.90454, "tr": 2.16934, "fi": 2.31962, "ru": 1.38137, "el": 1.4447, "hi": 1.19416, "ta": 0.98191, "th": 1.112, "zh": 3.1287, "my": 1.10127}, "gini_cpc": 0.081207, "gini_cps": 0.13499, "bpic": 8.739178, "worst_lang": "hi", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9927734059986666, "vi": 0.9073660411005903, "tr": 0.9644646018906294, "fi": 1.019599395290879, "ru": 1.0068841435763118, "el": 1.0667038348225666, "hi": 1.3255903292869249, "ta": 1.0810261063006286, "th": 0.994103700550235, "zh": 0.799271719974597, "my": 0.6416509055092317}, "t": 327.4}
{"event": "eval", "step": 6000, "phase": 3, "bpb": {"en": 2.01463, "de": 2.19241, "vi": 1.90296, "tr": 2.17582, "fi": 2.33079, "ru": 1.35747, "el": 1.43759, "hi": 1.19053, "ta": 0.98928, "th": 1.12959, "zh": 3.12709, "my": 1.10964}, "bytes_per_chunk": {"en": 4.889, "de": 4.769, "vi": 6.865, "tr": 5.149, "fi": 4.571, "ru": 8.097, "el": 8.088, "hi": 9.489, "ta": 11.897, "th": 12.96, "zh": 14.985, "my": 20.448}, "bpb_mean": 1.74648, "t": 328.7}
{"event": "probe", "step": 6000, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3683.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.21163, "cps": 28.773438, "mean_p": 0.244535, "boundary_entropy": 0.243583, "bpb": 2.003906}, "de": {"chunks": 4314.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.210101, "cps": 33.703125, "mean_p": 0.245728, "boundary_entropy": 0.253091, "bpb": 2.184671}, "vi": {"chunks": 3540.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.192026, "cps": 27.65625, "mean_p": 0.210654, "boundary_entropy": 0.290713, "bpb": 1.904545}, "tr": {"chunks": 3695.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.20411, "cps": 28.867188, "mean_p": 0.229801, "boundary_entropy": 0.256778, "bpb": 2.169336}, "fi": {"chunks": 4059.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.215778, "cps": 31.710938, "mean_p": 0.250459, "boundary_entropy": 0.266547, "bpb": 2.319622}, "ru": {"chunks": 4038.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.213087, "cps": 31.546875, "mean_p": 0.19158, "boundary_entropy": 0.329582, "bpb": 1.381369}, "el": {"chunks": 4633.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.225747, "cps": 36.195312, "mean_p": 0.216564, "boundary_entropy": 0.376393, "bpb": 1.444697}, "hi": {"chunks": 4720.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.280535, "cps": 36.875, "mean_p": 0.236346, "boundary_entropy": 0.442511, "bpb": 1.194161}, "ta": {"chunks": 4277.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.228778, "cps": 33.414062, "mean_p": 0.253503, "boundary_entropy": 0.50727, "bpb": 0.981911}, "th": {"chunks": 3384.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.210382, "cps": 26.4375, "mean_p": 0.218198, "boundary_entropy": 0.448269, "bpb": 1.112}, "zh": {"chunks": 999.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.16915, "cps": 7.804688, "mean_p": 0.165648, "boundary_entropy": 0.332407, "bpb": 3.128703}, "my": {"chunks": 2573.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.135793, "cps": 20.101562, "mean_p": 0.249842, "boundary_entropy": 0.515969, "bpb": 1.101266}}, "bpb": {"en": 2.00391, "de": 2.18467, "vi": 1.90454, "tr": 2.16934, "fi": 2.31962, "ru": 1.38137, "el": 1.4447, "hi": 1.19416, "ta": 0.98191, "th": 1.112, "zh": 3.1287, "my": 1.10127}, "gini_cpc": 0.081207, "gini_cps": 0.13499, "bpic": 8.739178, "worst_lang": "hi", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9927734059986666, "vi": 0.9073660411005903, "tr": 0.9644646018906294, "fi": 1.019599395290879, "ru": 1.0068841435763118, "el": 1.0667038348225666, "hi": 1.3255903292869249, "ta": 1.0810261063006286, "th": 0.994103700550235, "zh": 0.799271719974597, "my": 0.6416509055092317}, "t": 329.6}
{"event": "final", "run": "pilot_Bpc_s0", "condition": "B", "step": 6000, "params": 22516992, "per_lang": {"en": {"cpc": 0.211630178704821, "cps": 28.7734375, "bpb": 2.0236759541310367, "bytes_per_chunk": 4.849803431848666}, "de": {"cpc": 0.21010081332489164, "cps": 33.703125, "bpb": 2.184274127428656, "bytes_per_chunk": 4.781991645232492}, "vi": {"cpc": 0.1920260374288039, "cps": 27.65625, "bpb": 1.907216450995669, "bytes_per_chunk": 6.8501247762520086}, "tr": {"cpc": 0.20410981605258796, "cps": 28.8671875, "bpb": 2.1666493561171514, "bytes_per_chunk": 5.170697068917906}, "fi": {"cpc": 0.21577800223273616, "cps": 31.7109375, "bpb": 2.349011138461699, "bytes_per_chunk": 4.563708849079926}, "ru": {"cpc": 0.21308707124010554, "cps": 31.546875, "bpb": 1.3587628852341709, "bytes_per_chunk": 8.085748214864052}, "el": {"cpc": 0.22574672318861766, "cps": 36.1953125, "bpb": 1.4326475242048349, "bytes_per_chunk": 8.065721054736777}, "hi": {"cpc": 0.28053491827637445, "cps": 36.875, "bpb": 1.1991079776574975, "bytes_per_chunk": 9.450547073562017}, "ta": {"cpc": 0.22877774806097886, "cps": 33.4140625, "bpb": 0.9890050605123927, "bytes_per_chunk": 12.070912188608004}, "th": {"cpc": 0.2103823437985701, "cps": 26.4375, "bpb": 1.1282025457776095, "bytes_per_chunk": 13.016410536507857}, "zh": {"cpc": 0.16915001693193363, "cps": 7.8046875, "bpb": 3.135393780091879, "bytes_per_chunk": 15.04542715298304}, "my": {"cpc": 0.13579269579902892, "cps": 20.1015625, "bpb": 1.1188346722829436, "bytes_per_chunk": 20.079200337022712}}, "gini_cpc": 0.08120721436107003, "gini_cps": 0.13498994269232245, "bpic": 8.73917795741774, "seed": 0, "mean_boundary_entropy": 0.30649498105049133, "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": "parity_a", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0}, {"step": 4200, "phase": 3, "objective": "none", "router_trainable": false, "encoder_trainable": true, "frozen_router_params": 131584}], "t": 332.1}