{"event": "start", "run": "spec_A_s2", "condition": "A", "params": 82470528, "langs": ["en", "de", "vi", "tr", "fi", "ru", "el", "hi", "ta", "th", "zh", "my"], "premium": {"en": 1.0, "de": 1.184, "vi": 1.39, "tr": 1.12, "fi": 1.104, "ru": 1.971, "el": 2.174, "hi": 2.552, "ta": 3.192, "th": 2.762, "zh": 0.915, "my": 3.524}, "schedule": {"tau1": 0.4, "tau2": 0.7, "p1_objective": "global", "p2_objective": "global", "p3_objective": "global", "freeze_in_p3": false, "freeze_scope": "router"}, "corpus": "/local/scratch/sas245/beetle-hnet/scratch/data/tier1_corpus", "premium_kind": "flores_per_sent", "doc_filtered": {"my": 37511}, "weights": {"en": 0.2946, "de": 0.0909, "vi": 0.0576, "tr": 0.0612, "fi": 0.0424, "ru": 0.1476, "el": 0.0522, "hi": 0.042, "ta": 0.0313, "th": 0.0542, "zh": 0.1038, "my": 0.0222}, "t": 2.2}
{"event": "probe_schedule", "n": 21, "steps": [0, 1, 2, 4, 8, 16, 32, 64, 128, 256, 1000, 2000, 3000, 4000, 5000, 6000, 7000, 8000, 9000, 10000, 11000], "t": 2.2}
{"event": "probe", "step": 0, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 10185.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.585244, "cps": 79.570312, "mean_p": 0.50663, "boundary_entropy": 0.687329, "bpb": 8.017475}, "de": {"chunks": 10966.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.534067, "cps": 85.671875, "mean_p": 0.503879, "boundary_entropy": 0.688251, "bpb": 8.111776}, "vi": {"chunks": 5741.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.311418, "cps": 44.851562, "mean_p": 0.489692, "boundary_entropy": 0.688475, "bpb": 7.989146}, "tr": {"chunks": 6370.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.351875, "cps": 49.765625, "mean_p": 0.495802, "boundary_entropy": 0.687878, "bpb": 8.107657}, "fi": {"chunks": 2115.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.112434, "cps": 16.523438, "mean_p": 0.479404, "boundary_entropy": 0.686744, "bpb": 8.107682}, "ru": {"chunks": 23602.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.245488, "cps": 184.390625, "mean_p": 0.507156, "boundary_entropy": 0.690197, "bpb": 8.281387}, "el": {"chunks": 1455.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.070896, "cps": 11.367188, "mean_p": 0.480429, "boundary_entropy": 0.689544, "bpb": 7.890102}, "hi": {"chunks": 7663.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.455453, "cps": 59.867188, "mean_p": 0.494285, "boundary_entropy": 0.690846, "bpb": 7.810752}, "ta": {"chunks": 357.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.019096, "cps": 2.789062, "mean_p": 0.475379, "boundary_entropy": 0.689908, "bpb": 7.609497}, "th": {"chunks": 27234.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 1.69313, "cps": 212.765625, "mean_p": 0.50318, "boundary_entropy": 0.691076, "bpb": 7.876604}, "zh": {"chunks": 14702.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.489333, "cps": 114.859375, "mean_p": 0.531852, "boundary_entropy": 0.685234, "bpb": 8.024454}, "my": {"chunks": 789.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.04164, "cps": 6.164062, "mean_p": 0.475663, "boundary_entropy": 0.690063, "bpb": 8.245358}}, "bpb": {"en": 8.01748, "de": 8.11178, "vi": 7.98915, "tr": 8.10766, "fi": 8.10768, "ru": 8.28139, "el": 7.8901, "hi": 7.81075, "ta": 7.6095, "th": 7.8766, "zh": 8.02445, "my": 8.24536}, "gini_cpc": 0.562508, "gini_cps": 0.489131, "bpic": 3.45192, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9125547315729478, "vi": 0.5321174384544991, "tr": 0.6012456783662331, "fi": 0.19211513271032077, "ru": 2.1281521716321925, "el": 0.1211393488838307, "hi": 0.7782279770749311, "ta": 0.032629155488401715, "th": 2.8930334475858417, "zh": 4.253496332079356, "my": 0.07115029646287195}, "t": 7.7}
{"event": "train", "step": 0, "phase": 1, "objective": "global", "lm_loss": 5.5908, "bpb": 8.06582, "chunk_loss": 0.059712, "F": 0.56091, "G": 0.50224, "chunks": 574.38, "bytes_per_chunk": 1.783, "boundary_entropy": 0.6917, "router_noise": 0.0, "grad_norm": 9.2832, "lr": 6e-07, "router_trainable": true, "sec_per_step": 0.006, "alpha": 0.03, "t": 8.5}
{"event": "probe", "step": 1, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 10109.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.580877, "cps": 78.976562, "mean_p": 0.506451, "boundary_entropy": 0.687331, "bpb": 7.982808}, "de": {"chunks": 10912.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.531437, "cps": 85.25, "mean_p": 0.503747, "boundary_entropy": 0.688251, "bpb": 8.0804}, "vi": {"chunks": 5694.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.308869, "cps": 44.484375, "mean_p": 0.489579, "boundary_entropy": 0.688469, "bpb": 7.96876}, "tr": {"chunks": 6316.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.348892, "cps": 49.34375, "mean_p": 0.495629, "boundary_entropy": 0.687872, "bpb": 8.08377}, "fi": {"chunks": 2105.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.111903, "cps": 16.445312, "mean_p": 0.479274, "boundary_entropy": 0.686729, "bpb": 8.086822}, "ru": {"chunks": 23587.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.244697, "cps": 184.273438, "mean_p": 0.507134, "boundary_entropy": 0.690198, "bpb": 8.228966}, "el": {"chunks": 1445.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.070409, "cps": 11.289062, "mean_p": 0.480401, "boundary_entropy": 0.689541, "bpb": 7.879524}, "hi": {"chunks": 7639.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.454027, "cps": 59.679688, "mean_p": 0.494253, "boundary_entropy": 0.690845, "bpb": 7.788181}, "ta": {"chunks": 357.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.019096, "cps": 2.789062, "mean_p": 0.475325, "boundary_entropy": 0.689903, "bpb": 7.595257}, "th": {"chunks": 26991.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 1.678023, "cps": 210.867188, "mean_p": 0.503079, "boundary_entropy": 0.691076, "bpb": 7.839382}, "zh": {"chunks": 14695.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.488148, "cps": 114.804688, "mean_p": 0.531761, "boundary_entropy": 0.685245, "bpb": 8.01916}, "my": {"chunks": 786.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.041482, "cps": 6.140625, "mean_p": 0.475631, "boundary_entropy": 0.69006, "bpb": 8.246725}}, "bpb": {"en": 7.98281, "de": 8.0804, "vi": 7.96876, "tr": 8.08377, "fi": 8.08682, "ru": 8.22897, "el": 7.87952, "hi": 7.78818, "ta": 7.59526, "th": 7.83938, "zh": 8.01916, "my": 8.24673}, "gini_cpc": 0.563035, "gini_cps": 0.489244, "bpic": 3.468862, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9148878789402785, "vi": 0.5317288745513609, "tr": 0.6006306544528857, "fi": 0.19264428972932476, "ru": 2.1427890400961966, "el": 0.12121124878424644, "hi": 0.7816230546184773, "ta": 0.03287446321588401, "th": 2.8887757758123733, "zh": 4.28343391941485, "my": 0.07141264084681531}, "t": 13.6}
{"event": "probe", "step": 2, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 9968.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.572775, "cps": 77.875, "mean_p": 0.506053, "boundary_entropy": 0.687335, "bpb": 7.916047}, "de": {"chunks": 10766.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.524327, "cps": 84.109375, "mean_p": 0.503441, "boundary_entropy": 0.68825, "bpb": 8.020666}, "vi": {"chunks": 5573.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.302305, "cps": 43.539062, "mean_p": 0.489305, "boundary_entropy": 0.688454, "bpb": 7.931269}, "tr": {"chunks": 6161.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.34033, "cps": 48.132812, "mean_p": 0.49526, "boundary_entropy": 0.687858, "bpb": 8.037824}, "fi": {"chunks": 2072.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.110148, "cps": 16.1875, "mean_p": 0.478978, "boundary_entropy": 0.686696, "bpb": 8.047701}, "ru": {"chunks": 23560.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.243272, "cps": 184.0625, "mean_p": 0.5071, "boundary_entropy": 0.6902, "bpb": 8.121415}, "el": {"chunks": 1435.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.069922, "cps": 11.210938, "mean_p": 0.48036, "boundary_entropy": 0.689538, "bpb": 7.854247}, "hi": {"chunks": 7613.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.452481, "cps": 59.476562, "mean_p": 0.494228, "boundary_entropy": 0.690844, "bpb": 7.745822}, "ta": {"chunks": 352.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.018829, "cps": 2.75, "mean_p": 0.475301, "boundary_entropy": 0.689901, "bpb": 7.564817}, "th": {"chunks": 26718.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 1.661051, "cps": 208.734375, "mean_p": 0.502966, "boundary_entropy": 0.691077, "bpb": 7.777029}, "zh": {"chunks": 14680.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.485608, "cps": 114.6875, "mean_p": 0.531544, "boundary_entropy": 0.685271, "bpb": 8.006227}, "my": {"chunks": 675.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.035624, "cps": 5.273438, "mean_p": 0.475491, "boundary_entropy": 0.690046, "bpb": 8.235652}}, "bpb": {"en": 7.91605, "de": 8.02067, "vi": 7.93127, "tr": 8.03782, "fi": 8.0477, "ru": 8.12142, "el": 7.85425, "hi": 7.74582, "ta": 7.56482, "th": 7.77703, "zh": 8.00623, "my": 8.23565}, "gini_cpc": 0.565759, "gini_cps": 0.491919, "bpic": 3.502514, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9154150731324799, "vi": 0.5277910142398718, "tr": 0.5941782471516431, "fi": 0.1923064976923017, "ru": 2.1706118153288414, "el": 0.12207511632605128, "hi": 0.7899813668225693, "ta": 0.032872541692120325, "th": 2.9000064988815932, "zh": 4.339590040652947, "my": 0.062195145428004195}, "t": 19.2}
{"event": "probe", "step": 4, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 9433.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.542033, "cps": 73.695312, "mean_p": 0.50457, "boundary_entropy": 0.687346, "bpb": 7.681173}, "de": {"chunks": 10206.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.497054, "cps": 79.734375, "mean_p": 0.502293, "boundary_entropy": 0.688245, "bpb": 7.806359}, "vi": {"chunks": 5258.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.285218, "cps": 41.078125, "mean_p": 0.488415, "boundary_entropy": 0.688405, "bpb": 7.800449}, "tr": {"chunks": 5693.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.314478, "cps": 44.476562, "mean_p": 0.493945, "boundary_entropy": 0.687806, "bpb": 7.874074}, "fi": {"chunks": 1984.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.10547, "cps": 15.5, "mean_p": 0.477974, "boundary_entropy": 0.686583, "bpb": 7.909954}, "ru": {"chunks": 23402.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.234934, "cps": 182.828125, "mean_p": 0.506971, "boundary_entropy": 0.690208, "bpb": 7.802292}, "el": {"chunks": 1393.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.067875, "cps": 10.882812, "mean_p": 0.480084, "boundary_entropy": 0.689514, "bpb": 7.777913}, "hi": {"chunks": 7469.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.443923, "cps": 58.351562, "mean_p": 0.494053, "boundary_entropy": 0.690837, "bpb": 7.594629}, "ta": {"chunks": 349.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.018668, "cps": 2.726562, "mean_p": 0.475232, "boundary_entropy": 0.689896, "bpb": 7.467694}, "th": {"chunks": 25539.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 1.587753, "cps": 199.523438, "mean_p": 0.502551, "boundary_entropy": 0.691079, "bpb": 7.561671}, "zh": {"chunks": 14610.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.473756, "cps": 114.140625, "mean_p": 0.53085, "boundary_entropy": 0.685353, "bpb": 7.968467}, "my": {"chunks": 652.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.03441, "cps": 5.09375, "mean_p": 0.475186, "boundary_entropy": 0.690014, "bpb": 8.209928}}, "bpb": {"en": 7.68117, "de": 7.80636, "vi": 7.80045, "tr": 7.87407, "fi": 7.90995, "ru": 7.80229, "el": 7.77791, "hi": 7.59463, "ta": 7.46769, "th": 7.56167, "zh": 7.96847, "my": 8.20993}, "gini_cpc": 0.570889, "gini_cps": 0.495158, "bpic": 3.620985, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9170171176872693, "vi": 0.5262010684406326, "tr": 0.580182891520206, "fi": 0.19458263048077834, "ru": 2.2783374371732092, "el": 0.1252231306017421, "hi": 0.8189957956395059, "ta": 0.03444088028725386, "th": 2.9292545192422805, "zh": 4.563846816134135, "my": 0.06348315545918426}, "t": 25.2}
{"event": "probe", "step": 8, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 7421.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.426421, "cps": 57.976562, "mean_p": 0.49951, "boundary_entropy": 0.687322, "bpb": 7.072624}, "de": {"chunks": 8368.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.407539, "cps": 65.375, "mean_p": 0.498488, "boundary_entropy": 0.688197, "bpb": 7.217418}, "vi": {"chunks": 4144.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.22479, "cps": 32.375, "mean_p": 0.485407, "boundary_entropy": 0.688222, "bpb": 7.44837}, "tr": {"chunks": 4329.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.239132, "cps": 33.820312, "mean_p": 0.489659, "boundary_entropy": 0.687589, "bpb": 7.395599}, "fi": {"chunks": 1719.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.091383, "cps": 13.429688, "mean_p": 0.47489, "boundary_entropy": 0.686208, "bpb": 7.502419}, "ru": {"chunks": 22012.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.161583, "cps": 171.96875, "mean_p": 0.505834, "boundary_entropy": 0.690249, "bpb": 6.813979}, "el": {"chunks": 1242.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.060517, "cps": 9.703125, "mean_p": 0.479056, "boundary_entropy": 0.689421, "bpb": 7.51436}, "hi": {"chunks": 7259.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.431441, "cps": 56.710938, "mean_p": 0.493693, "boundary_entropy": 0.690821, "bpb": 7.038052}, "ta": {"chunks": 334.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.017866, "cps": 2.609375, "mean_p": 0.475413, "boundary_entropy": 0.689923, "bpb": 7.123048}, "th": {"chunks": 21103.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 1.311968, "cps": 164.867188, "mean_p": 0.50095, "boundary_entropy": 0.691079, "bpb": 6.852628}, "zh": {"chunks": 14328.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.426007, "cps": 111.9375, "mean_p": 0.528149, "boundary_entropy": 0.685652, "bpb": 7.848781}, "my": {"chunks": 591.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.031191, "cps": 4.617188, "mean_p": 0.474323, "boundary_entropy": 0.689923, "bpb": 8.107995}}, "bpb": {"en": 7.07262, "de": 7.21742, "vi": 7.44837, "tr": 7.3956, "fi": 7.50242, "ru": 6.81398, "el": 7.51436, "hi": 7.03805, "ta": 7.12305, "th": 6.85263, "zh": 7.84878, "my": 8.10799}, "gini_cpc": 0.588077, "gini_cps": 0.503208, "bpic": 4.133344, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9557206129732508, "vi": 0.5271549554411431, "tr": 0.5607880144255019, "fi": 0.21430173239887487, "ru": 2.72403057855853, "el": 0.14191961988896698, "hi": 1.0117737603800148, "ta": 0.04189697676360562, "th": 3.0766976676133146, "zh": 5.6892342882676195, "my": 0.07314519354767408}, "t": 31.7}
{"event": "probe", "step": 16, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 4289.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.246452, "cps": 33.507812, "mean_p": 0.491833, "boundary_entropy": 0.687182, "bpb": 6.365687}, "de": {"chunks": 6166.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.300297, "cps": 48.171875, "mean_p": 0.494706, "boundary_entropy": 0.688167, "bpb": 6.434561}, "vi": {"chunks": 2410.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.13073, "cps": 18.828125, "mean_p": 0.480058, "boundary_entropy": 0.687873, "bpb": 6.979642}, "tr": {"chunks": 2884.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.159311, "cps": 22.53125, "mean_p": 0.484724, "boundary_entropy": 0.687318, "bpb": 6.722106}, "fi": {"chunks": 1386.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.07368, "cps": 10.828125, "mean_p": 0.4734, "boundary_entropy": 0.686082, "bpb": 6.727237}, "ru": {"chunks": 10614.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.560106, "cps": 82.921875, "mean_p": 0.496838, "boundary_entropy": 0.690276, "bpb": 5.685126}, "el": {"chunks": 1318.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.064221, "cps": 10.296875, "mean_p": 0.478565, "boundary_entropy": 0.689367, "bpb": 6.743184}, "hi": {"chunks": 7895.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.469242, "cps": 61.679688, "mean_p": 0.493748, "boundary_entropy": 0.690812, "bpb": 5.96519}, "ta": {"chunks": 357.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.019096, "cps": 2.789062, "mean_p": 0.478748, "boundary_entropy": 0.690272, "bpb": 6.475072}, "th": {"chunks": 4147.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.257818, "cps": 32.398438, "mean_p": 0.491304, "boundary_entropy": 0.690887, "bpb": 5.581199}, "zh": {"chunks": 12180.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.06231, "cps": 95.15625, "mean_p": 0.517051, "boundary_entropy": 0.686555, "bpb": 7.484561}, "my": {"chunks": 613.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.032352, "cps": 4.789062, "mean_p": 0.472694, "boundary_entropy": 0.689738, "bpb": 7.437552}}, "bpb": {"en": 6.36569, "de": 6.43456, "vi": 6.97964, "tr": 6.72211, "fi": 6.72724, "ru": 5.68513, "el": 6.74318, "hi": 5.96519, "ta": 6.47507, "th": 5.5812, "zh": 7.48456, "my": 7.43755}, "gini_cpc": 0.602997, "gini_cps": 0.455993, "bpic": 7.073131, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.218482194219776, "vi": 0.5304469719404437, "tr": 0.6464170137443228, "fi": 0.29896436158288253, "ru": 2.272678183313637, "el": 0.26058093560322265, "hi": 1.9039920706655853, "ta": 0.0774837837839523, "th": 1.0461188894130498, "zh": 8.368004780261083, "my": 0.1312699054431923}, "t": 39.1}
{"event": "probe", "step": 32, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3693.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.212205, "cps": 28.851562, "mean_p": 0.493787, "boundary_entropy": 0.687515, "bpb": 5.831265}, "de": {"chunks": 6539.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.318463, "cps": 51.085938, "mean_p": 0.497278, "boundary_entropy": 0.688447, "bpb": 5.948486}, "vi": {"chunks": 1761.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.095525, "cps": 13.757812, "mean_p": 0.481962, "boundary_entropy": 0.688251, "bpb": 6.571082}, "tr": {"chunks": 3414.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.188588, "cps": 26.671875, "mean_p": 0.490637, "boundary_entropy": 0.687918, "bpb": 6.256266}, "fi": {"chunks": 2371.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.126043, "cps": 18.523438, "mean_p": 0.488317, "boundary_entropy": 0.687781, "bpb": 5.982147}, "ru": {"chunks": 6812.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.359472, "cps": 53.21875, "mean_p": 0.494095, "boundary_entropy": 0.690272, "bpb": 5.151895}, "el": {"chunks": 12271.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.597915, "cps": 95.867188, "mean_p": 0.496717, "boundary_entropy": 0.690459, "bpb": 5.522159}, "hi": {"chunks": 10928.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.64951, "cps": 85.375, "mean_p": 0.495553, "boundary_entropy": 0.690859, "bpb": 5.246555}, "ta": {"chunks": 26032.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 1.392458, "cps": 203.375, "mean_p": 0.501224, "boundary_entropy": 0.691301, "bpb": 5.001176}, "th": {"chunks": 5545.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.344731, "cps": 43.320312, "mean_p": 0.494325, "boundary_entropy": 0.691023, "bpb": 4.853723}, "zh": {"chunks": 3266.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.552997, "cps": 25.515625, "mean_p": 0.489508, "boundary_entropy": 0.686625, "bpb": 7.017842}, "my": {"chunks": 7377.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.389329, "cps": 57.632812, "mean_p": 0.493333, "boundary_entropy": 0.691247, "bpb": 5.001114}}, "bpb": {"en": 5.83126, "de": 5.94849, "vi": 6.57108, "tr": 6.25627, "fi": 5.98215, "ru": 5.1519, "el": 5.52216, "hi": 5.24655, "ta": 5.00118, "th": 4.85372, "zh": 7.01784, "my": 5.00111}, "gini_cpc": 0.379009, "gini_cps": 0.407256, "bpic": 4.263807, "worst_lang": "ta", "cpc_parity_vs_en": {"en": 1.0, "de": 1.5007340722309352, "vi": 0.4501539098115605, "tr": 0.8887053157729579, "fi": 0.5939699626983382, "ru": 1.6939876411695236, "el": 2.8176297457553225, "hi": 3.0607680970664957, "ta": 6.561858766217895, "th": 1.6245208802646987, "zh": 2.605958830230338, "my": 1.8346837734707517}, "t": 49.1}
{"event": "probe", "step": 64, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 5060.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.290754, "cps": 39.53125, "mean_p": 0.498074, "boundary_entropy": 0.68776, "bpb": 5.414096}, "de": {"chunks": 4998.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.243413, "cps": 39.046875, "mean_p": 0.496102, "boundary_entropy": 0.688524, "bpb": 5.569478}, "vi": {"chunks": 2543.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.137944, "cps": 19.867188, "mean_p": 0.488591, "boundary_entropy": 0.68883, "bpb": 5.966028}, "tr": {"chunks": 2804.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.154891, "cps": 21.90625, "mean_p": 0.491639, "boundary_entropy": 0.688096, "bpb": 5.888798}, "fi": {"chunks": 3541.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.188241, "cps": 27.664062, "mean_p": 0.493663, "boundary_entropy": 0.688152, "bpb": 5.570817}, "ru": {"chunks": 4984.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.263008, "cps": 38.9375, "mean_p": 0.495072, "boundary_entropy": 0.690398, "bpb": 4.844828}, "el": {"chunks": 4842.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.23593, "cps": 37.828125, "mean_p": 0.491519, "boundary_entropy": 0.690383, "bpb": 5.090933}, "hi": {"chunks": 1389.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.082556, "cps": 10.851562, "mean_p": 0.484007, "boundary_entropy": 0.690308, "bpb": 4.904838}, "ta": {"chunks": 908.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.048569, "cps": 7.09375, "mean_p": 0.489114, "boundary_entropy": 0.691038, "bpb": 4.566165}, "th": {"chunks": 5559.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.345601, "cps": 43.429688, "mean_p": 0.491284, "boundary_entropy": 0.690851, "bpb": 4.409379}, "zh": {"chunks": 5903.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.999492, "cps": 46.117188, "mean_p": 0.499898, "boundary_entropy": 0.687294, "bpb": 6.673269}, "my": {"chunks": 12797.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.675375, "cps": 99.976562, "mean_p": 0.496709, "boundary_entropy": 0.691369, "bpb": 4.278285}}, "bpb": {"en": 5.4141, "de": 5.56948, "vi": 5.96603, "tr": 5.8888, "fi": 5.57082, "ru": 4.84483, "el": 5.09093, "hi": 4.90484, "ta": 4.56617, "th": 4.40938, "zh": 6.67327, "my": 4.27829}, "gini_cpc": 0.412833, "gini_cps": 0.31375, "bpic": 6.93647, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.8371773077523523, "vi": 0.4744351106494241, "tr": 0.5327225260635289, "fi": 0.6474222886575279, "ru": 0.9045705048651016, "el": 0.8114421128593986, "hi": 0.28393620682567, "ta": 0.16704519290842068, "th": 1.188636910385907, "zh": 3.4375810290064766, "my": 2.322835192384233}, "t": 63.4}
{"event": "train", "step": 100, "phase": 1, "objective": "global", "lm_loss": 3.23325, "bpb": 4.66459, "chunk_loss": 0.03182, "F": 0.19427, "G": 0.47776, "chunks": 198.94, "bytes_per_chunk": 5.147, "boundary_entropy": 0.68971, "router_noise": 0.0, "grad_norm": 1.7225, "lr": 6.06e-05, "router_trainable": true, "sec_per_step": 0.654, "alpha": 0.03, "t": 73.9}
{"event": "probe", "step": 128, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 4448.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.255588, "cps": 34.75, "mean_p": 0.450868, "boundary_entropy": 0.672921, "bpb": 4.337537}, "de": {"chunks": 6072.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.295719, "cps": 47.4375, "mean_p": 0.452936, "boundary_entropy": 0.673473, "bpb": 4.56533}, "vi": {"chunks": 5893.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.319664, "cps": 46.039062, "mean_p": 0.471421, "boundary_entropy": 0.68377, "bpb": 4.744938}, "tr": {"chunks": 4387.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.242336, "cps": 34.273438, "mean_p": 0.457852, "boundary_entropy": 0.678422, "bpb": 4.78791}, "fi": {"chunks": 4738.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.251874, "cps": 37.015625, "mean_p": 0.463107, "boundary_entropy": 0.679618, "bpb": 4.555537}, "ru": {"chunks": 6993.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.369024, "cps": 54.632812, "mean_p": 0.468845, "boundary_entropy": 0.686713, "bpb": 3.222517}, "el": {"chunks": 13450.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.655362, "cps": 105.078125, "mean_p": 0.494012, "boundary_entropy": 0.689542, "bpb": 3.958738}, "hi": {"chunks": 16099.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.95685, "cps": 125.773438, "mean_p": 0.49316, "boundary_entropy": 0.689924, "bpb": 3.00741}, "ta": {"chunks": 15473.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.827654, "cps": 120.882812, "mean_p": 0.492145, "boundary_entropy": 0.690849, "bpb": 3.03803}, "th": {"chunks": 12500.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.777122, "cps": 97.65625, "mean_p": 0.485516, "boundary_entropy": 0.689895, "bpb": 2.796046}, "zh": {"chunks": 4866.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.823908, "cps": 38.015625, "mean_p": 0.498246, "boundary_entropy": 0.6873, "bpb": 6.235952}, "my": {"chunks": 21264.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 1.122229, "cps": 166.125, "mean_p": 0.497631, "boundary_entropy": 0.691215, "bpb": 3.293837}}, "bpb": {"en": 4.33754, "de": 4.56533, "vi": 4.74494, "tr": 4.78791, "fi": 4.55554, "ru": 3.22252, "el": 3.95874, "hi": 3.00741, "ta": 3.03803, "th": 2.79605, "zh": 6.23595, "my": 3.29384}, "gini_cpc": 0.293927, "gini_cps": 0.307929, "bpic": 3.303246, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1570142220612054, "vi": 1.2506985339311953, "tr": 0.9481486359826684, "fi": 0.9854679729715776, "ru": 1.4438220017653423, "el": 2.564134374418314, "hi": 3.7437183581515177, "ta": 3.238235262284957, "th": 3.040522974843625, "zh": 3.2235766669427752, "my": 4.390772435882833}, "t": 84.1}
{"event": "train", "step": 200, "phase": 1, "objective": "global", "lm_loss": 2.54558, "bpb": 3.67249, "chunk_loss": 0.032248, "F": 0.21881, "G": 0.38596, "chunks": 224.06, "bytes_per_chunk": 4.57, "boundary_entropy": 0.61856, "router_noise": 0.0, "grad_norm": 1.7277, "lr": 0.0001206, "router_trainable": true, "sec_per_step": 0.255, "alpha": 0.03, "t": 99.3}
{"event": "probe", "step": 256, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3543.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.203586, "cps": 27.679688, "mean_p": 0.302865, "boundary_entropy": 0.499433, "bpb": 3.583546}, "de": {"chunks": 4436.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.216042, "cps": 34.65625, "mean_p": 0.302928, "boundary_entropy": 0.502289, "bpb": 3.78908}, "vi": {"chunks": 5153.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.279523, "cps": 40.257812, "mean_p": 0.379013, "boundary_entropy": 0.614361, "bpb": 3.06476}, "tr": {"chunks": 4548.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.251229, "cps": 35.53125, "mean_p": 0.338353, "boundary_entropy": 0.540458, "bpb": 3.769503}, "fi": {"chunks": 3883.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.206422, "cps": 30.335938, "mean_p": 0.310648, "boundary_entropy": 0.528601, "bpb": 3.83233}, "ru": {"chunks": 5849.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.308654, "cps": 45.695312, "mean_p": 0.278336, "boundary_entropy": 0.497081, "bpb": 2.626991}, "el": {"chunks": 6927.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.337524, "cps": 54.117188, "mean_p": 0.338301, "boundary_entropy": 0.59626, "bpb": 2.671816}, "hi": {"chunks": 8266.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.491293, "cps": 64.578125, "mean_p": 0.436835, "boundary_entropy": 0.655983, "bpb": 2.054601}, "ta": {"chunks": 13998.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.748756, "cps": 109.359375, "mean_p": 0.408664, "boundary_entropy": 0.619841, "bpb": 1.873805}, "th": {"chunks": 3966.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.246565, "cps": 30.984375, "mean_p": 0.414032, "boundary_entropy": 0.665138, "bpb": 2.093238}, "zh": {"chunks": 3116.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.527599, "cps": 24.34375, "mean_p": 0.437629, "boundary_entropy": 0.666351, "bpb": 4.907716}, "my": {"chunks": 18269.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.964165, "cps": 142.726562, "mean_p": 0.472609, "boundary_entropy": 0.682922, "bpb": 2.191193}}, "bpb": {"en": 3.58355, "de": 3.78908, "vi": 3.06476, "tr": 3.7695, "fi": 3.83233, "ru": 2.62699, "el": 2.67182, "hi": 2.0546, "ta": 1.8738, "th": 2.09324, "zh": 4.90772, "my": 2.19119}, "gini_cpc": 0.295243, "gini_cps": 0.316344, "bpic": 4.682883, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0611874328155502, "vi": 1.3729982015673072, "tr": 1.234021915146104, "fi": 1.0139311717507606, "ru": 1.5160913674963528, "el": 1.6578961016167277, "hi": 2.413199884752766, "ta": 3.677845552712456, "th": 1.2111128512082694, "zh": 2.5915343772733017, "my": 4.735920108994768}, "t": 113.7}
{"event": "train", "step": 300, "phase": 1, "objective": "global", "lm_loss": 2.15545, "bpb": 3.10965, "chunk_loss": 0.031263, "F": 0.21391, "G": 0.3122, "chunks": 219.05, "bytes_per_chunk": 4.675, "boundary_entropy": 0.49269, "router_noise": 0.0, "grad_norm": 1.1719, "lr": 0.0001806, "router_trainable": true, "sec_per_step": 0.234, "alpha": 0.03, "t": 122.8}
{"event": "train", "step": 400, "phase": 1, "objective": "global", "lm_loss": 2.06124, "bpb": 2.97374, "chunk_loss": 0.031449, "F": 0.21307, "G": 0.29673, "chunks": 218.19, "bytes_per_chunk": 4.693, "boundary_entropy": 0.43289, "router_noise": 0.0, "grad_norm": 1.4968, "lr": 0.0002406, "router_trainable": true, "sec_per_step": 0.194, "alpha": 0.03, "t": 142.2}
{"event": "train", "step": 500, "phase": 1, "objective": "global", "lm_loss": 1.81548, "bpb": 2.61918, "chunk_loss": 0.030582, "F": 0.19028, "G": 0.28528, "chunks": 194.84, "bytes_per_chunk": 5.255, "boundary_entropy": 0.40634, "router_noise": 0.0, "grad_norm": 1.1578, "lr": 0.0003, "router_trainable": true, "sec_per_step": 0.188, "alpha": 0.03, "t": 161.0}
{"event": "train", "step": 600, "phase": 1, "objective": "global", "lm_loss": 1.78522, "bpb": 2.57553, "chunk_loss": 0.030876, "F": 0.20566, "G": 0.28408, "chunks": 210.59, "bytes_per_chunk": 4.862, "boundary_entropy": 0.36305, "router_noise": 0.0, "grad_norm": 0.9648, "lr": 0.0002999, "router_trainable": true, "sec_per_step": 0.183, "alpha": 0.03, "t": 179.4}
{"event": "train", "step": 700, "phase": 1, "objective": "global", "lm_loss": 1.67404, "bpb": 2.41513, "chunk_loss": 0.030715, "F": 0.20216, "G": 0.27964, "chunks": 207.02, "bytes_per_chunk": 4.946, "boundary_entropy": 0.3535, "router_noise": 0.0, "grad_norm": 0.7038, "lr": 0.0002998, "router_trainable": true, "sec_per_step": 0.184, "alpha": 0.03, "t": 197.8}
{"event": "train", "step": 800, "phase": 1, "objective": "global", "lm_loss": 1.74248, "bpb": 2.51387, "chunk_loss": 0.030592, "F": 0.1989, "G": 0.25765, "chunks": 203.67, "bytes_per_chunk": 5.028, "boundary_entropy": 0.30038, "router_noise": 0.0, "grad_norm": 0.8652, "lr": 0.0002995, "router_trainable": true, "sec_per_step": 0.182, "alpha": 0.03, "t": 215.9}
{"event": "train", "step": 900, "phase": 1, "objective": "global", "lm_loss": 1.62799, "bpb": 2.34869, "chunk_loss": 0.030734, "F": 0.20029, "G": 0.27154, "chunks": 205.09, "bytes_per_chunk": 4.993, "boundary_entropy": 0.33048, "router_noise": 0.0, "grad_norm": 0.6538, "lr": 0.000299, "router_trainable": true, "sec_per_step": 0.183, "alpha": 0.03, "t": 234.3}
{"event": "probe", "step": 1000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3636.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.208929, "cps": 28.40625, "mean_p": 0.220097, "boundary_entropy": 0.174134, "bpb": 2.300232}, "de": {"chunks": 4675.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.227682, "cps": 36.523438, "mean_p": 0.234307, "boundary_entropy": 0.181245, "bpb": 2.57893}, "vi": {"chunks": 4393.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.238297, "cps": 34.320312, "mean_p": 0.284582, "boundary_entropy": 0.381337, "bpb": 2.118977}, "tr": {"chunks": 4307.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.237916, "cps": 33.648438, "mean_p": 0.221297, "boundary_entropy": 0.202237, "bpb": 2.588304}, "fi": {"chunks": 4174.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.221891, "cps": 32.609375, "mean_p": 0.228116, "boundary_entropy": 0.211968, "bpb": 2.840694}, "ru": {"chunks": 5675.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.299472, "cps": 44.335938, "mean_p": 0.285708, "boundary_entropy": 0.457444, "bpb": 1.73385}, "el": {"chunks": 5386.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.262437, "cps": 42.078125, "mean_p": 0.269129, "boundary_entropy": 0.466668, "bpb": 1.805216}, "hi": {"chunks": 5881.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.349539, "cps": 45.945312, "mean_p": 0.339903, "boundary_entropy": 0.581362, "bpb": 1.399629}, "ta": {"chunks": 7315.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.391281, "cps": 57.148438, "mean_p": 0.297579, "boundary_entropy": 0.498052, "bpb": 1.136946}, "th": {"chunks": 8301.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.516071, "cps": 64.851562, "mean_p": 0.394684, "boundary_entropy": 0.615161, "bpb": 1.40183}, "zh": {"chunks": 3330.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.563833, "cps": 26.015625, "mean_p": 0.255498, "boundary_entropy": 0.259583, "bpb": 3.449358}, "my": {"chunks": 10790.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.569453, "cps": 84.296875, "mean_p": 0.416885, "boundary_entropy": 0.624382, "bpb": 1.328239}}, "bpb": {"en": 2.30023, "de": 2.57893, "vi": 2.11898, "tr": 2.5883, "fi": 2.84069, "ru": 1.73385, "el": 1.80522, "hi": 1.39963, "ta": 1.13695, "th": 1.40183, "zh": 3.44936, "my": 1.32824}, "gini_cpc": 0.207977, "gini_cps": 0.194582, "bpic": 5.655232, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.089756463836127, "vi": 1.1405604474198436, "tr": 1.1387399732626415, "fi": 1.0620398359010852, "ru": 1.4333653352142601, "el": 1.2561044366386644, "hi": 1.6730015839622594, "ta": 1.8727901073070667, "th": 2.4700718952840264, "zh": 2.698677800390943, "my": 2.725576112076059}, "t": 255.4}
{"event": "eval", "step": 1000, "phase": 1, "bpb": {"en": 2.31265, "de": 2.56985, "vi": 2.12034, "tr": 2.5822, "fi": 2.8477, "ru": 1.73197, "el": 1.79878, "hi": 1.40666, "ta": 1.13652, "th": 1.39851, "zh": 3.45584, "my": 1.33904}, "bytes_per_chunk": {"en": 4.67, "de": 4.372, "vi": 5.573, "tr": 4.589, "fi": 4.816, "ru": 6.13, "el": 7.185, "hi": 6.73, "ta": 6.681, "th": 5.355, "zh": 4.901, "my": 4.773}, "bpb_mean": 2.05834, "t": 261.3}
{"event": "train", "step": 1000, "phase": 1, "objective": "global", "lm_loss": 1.55735, "bpb": 2.24678, "chunk_loss": 0.030512, "F": 0.19873, "G": 0.25838, "chunks": 203.5, "bytes_per_chunk": 5.032, "boundary_entropy": 0.29541, "router_noise": 0.0, "grad_norm": 0.7327, "lr": 0.0002985, "router_trainable": true, "sec_per_step": 0.272, "alpha": 0.03, "t": 261.5}
{"event": "train", "step": 1100, "phase": 1, "objective": "global", "lm_loss": 1.52117, "bpb": 2.19459, "chunk_loss": 0.030622, "F": 0.19875, "G": 0.26529, "chunks": 203.52, "bytes_per_chunk": 5.032, "boundary_entropy": 0.30275, "router_noise": 0.0, "grad_norm": 0.5516, "lr": 0.0002978, "router_trainable": true, "sec_per_step": 0.181, "alpha": 0.03, "t": 279.6}
{"event": "train", "step": 1200, "phase": 1, "objective": "global", "lm_loss": 1.51882, "bpb": 2.1912, "chunk_loss": 0.030101, "F": 0.18622, "G": 0.25634, "chunks": 190.69, "bytes_per_chunk": 5.37, "boundary_entropy": 0.29615, "router_noise": 0.0, "grad_norm": 0.5997, "lr": 0.000297, "router_trainable": true, "sec_per_step": 0.184, "alpha": 0.03, "t": 298.0}
{"event": "train", "step": 1300, "phase": 1, "objective": "global", "lm_loss": 1.56678, "bpb": 2.26039, "chunk_loss": 0.030558, "F": 0.20216, "G": 0.24457, "chunks": 207.02, "bytes_per_chunk": 4.946, "boundary_entropy": 0.25516, "router_noise": 0.0, "grad_norm": 0.6103, "lr": 0.0002962, "router_trainable": true, "sec_per_step": 0.183, "alpha": 0.03, "t": 316.3}
{"event": "train", "step": 1400, "phase": 1, "objective": "global", "lm_loss": 1.39048, "bpb": 2.00603, "chunk_loss": 0.030268, "F": 0.18956, "G": 0.24401, "chunks": 194.11, "bytes_per_chunk": 5.275, "boundary_entropy": 0.27726, "router_noise": 0.0, "grad_norm": 0.5055, "lr": 0.0002951, "router_trainable": true, "sec_per_step": 0.181, "alpha": 0.03, "t": 334.4}
{"event": "train", "step": 1500, "phase": 1, "objective": "global", "lm_loss": 1.51099, "bpb": 2.1799, "chunk_loss": 0.030294, "F": 0.18982, "G": 0.23476, "chunks": 194.38, "bytes_per_chunk": 5.268, "boundary_entropy": 0.25015, "router_noise": 0.0, "grad_norm": 0.5283, "lr": 0.000294, "router_trainable": true, "sec_per_step": 0.183, "alpha": 0.03, "t": 352.7}
{"event": "train", "step": 1600, "phase": 1, "objective": "global", "lm_loss": 1.53007, "bpb": 2.20742, "chunk_loss": 0.030213, "F": 0.19006, "G": 0.23795, "chunks": 194.62, "bytes_per_chunk": 5.261, "boundary_entropy": 0.246, "router_noise": 0.0, "grad_norm": 0.5612, "lr": 0.0002928, "router_trainable": true, "sec_per_step": 0.183, "alpha": 0.03, "t": 371.0}
{"event": "train", "step": 1700, "phase": 1, "objective": "global", "lm_loss": 1.40651, "bpb": 2.02916, "chunk_loss": 0.030348, "F": 0.18962, "G": 0.24505, "chunks": 194.17, "bytes_per_chunk": 5.274, "boundary_entropy": 0.28198, "router_noise": 0.0, "grad_norm": 0.4718, "lr": 0.0002914, "router_trainable": true, "sec_per_step": 0.18, "alpha": 0.03, "t": 389.0}
{"event": "train", "step": 1800, "phase": 1, "objective": "global", "lm_loss": 1.41566, "bpb": 2.04237, "chunk_loss": 0.030358, "F": 0.19141, "G": 0.24637, "chunks": 196.0, "bytes_per_chunk": 5.224, "boundary_entropy": 0.27405, "router_noise": 0.0, "grad_norm": 0.4812, "lr": 0.0002899, "router_trainable": true, "sec_per_step": 0.179, "alpha": 0.03, "t": 406.9}
{"event": "train", "step": 1900, "phase": 1, "objective": "global", "lm_loss": 1.40593, "bpb": 2.02832, "chunk_loss": 0.031001, "F": 0.20525, "G": 0.2592, "chunks": 210.17, "bytes_per_chunk": 4.872, "boundary_entropy": 0.27553, "router_noise": 0.0, "grad_norm": 0.5044, "lr": 0.0002883, "router_trainable": true, "sec_per_step": 0.179, "alpha": 0.03, "t": 424.8}
{"event": "probe", "step": 2000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3590.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.206286, "cps": 28.046875, "mean_p": 0.207828, "boundary_entropy": 0.127202, "bpb": 2.069803}, "de": {"chunks": 4291.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.208981, "cps": 33.523438, "mean_p": 0.21302, "boundary_entropy": 0.136627, "bpb": 2.281977}, "vi": {"chunks": 4094.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.222078, "cps": 31.984375, "mean_p": 0.243807, "boundary_entropy": 0.290823, "bpb": 1.963248}, "tr": {"chunks": 3668.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.202618, "cps": 28.65625, "mean_p": 0.205192, "boundary_entropy": 0.164382, "bpb": 2.255734}, "fi": {"chunks": 3925.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.208655, "cps": 30.664062, "mean_p": 0.212692, "boundary_entropy": 0.152542, "bpb": 2.433127}, "ru": {"chunks": 5415.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.285752, "cps": 42.304688, "mean_p": 0.28049, "boundary_entropy": 0.433303, "bpb": 1.442609}, "el": {"chunks": 6838.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.333187, "cps": 53.421875, "mean_p": 0.29086, "boundary_entropy": 0.454858, "bpb": 1.556758}, "hi": {"chunks": 8377.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.49789, "cps": 65.445312, "mean_p": 0.291776, "boundary_entropy": 0.476698, "bpb": 1.202798}, "ta": {"chunks": 7274.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.389088, "cps": 56.828125, "mean_p": 0.272326, "boundary_entropy": 0.456343, "bpb": 1.00761}, "th": {"chunks": 7417.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.461113, "cps": 57.945312, "mean_p": 0.331185, "boundary_entropy": 0.536968, "bpb": 1.151489}, "zh": {"chunks": 3208.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.543176, "cps": 25.0625, "mean_p": 0.243634, "boundary_entropy": 0.223099, "bpb": 3.155963}, "my": {"chunks": 8331.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.439677, "cps": 65.085938, "mean_p": 0.405601, "boundary_entropy": 0.63426, "bpb": 1.179009}}, "bpb": {"en": 2.0698, "de": 2.28198, "vi": 1.96325, "tr": 2.25573, "fi": 2.43313, "ru": 1.44261, "el": 1.55676, "hi": 1.2028, "ta": 1.00761, "th": 1.15149, "zh": 3.15596, "my": 1.17901}, "gini_cpc": 0.205071, "gini_cps": 0.190522, "bpic": 5.777398, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0130614255440695, "vi": 1.0765504033217668, "tr": 0.9822192631240271, "fi": 1.0114803479088053, "ru": 1.3852205260875636, "el": 1.6151688227502616, "hi": 2.413587867901178, "ta": 1.8861555195146245, "th": 2.2353054931886054, "zh": 2.633119616800629, "my": 2.1313924842245275}, "t": 445.9}
{"event": "eval", "step": 2000, "phase": 1, "bpb": {"en": 2.07907, "de": 2.27311, "vi": 1.96617, "tr": 2.24549, "fi": 2.44719, "ru": 1.44223, "el": 1.55139, "hi": 1.21135, "ta": 1.00811, "th": 1.14795, "zh": 3.1638, "my": 1.18901}, "bytes_per_chunk": {"en": 4.794, "de": 4.784, "vi": 5.848, "tr": 5.355, "fi": 4.895, "ru": 6.171, "el": 5.483, "hi": 5.047, "ta": 6.64, "th": 6.066, "zh": 5.104, "my": 6.327}, "bpb_mean": 1.81041, "t": 451.5}
{"event": "train", "step": 2000, "phase": 1, "objective": "global", "lm_loss": 1.40155, "bpb": 2.02201, "chunk_loss": 0.03041, "F": 0.19623, "G": 0.24388, "chunks": 200.94, "bytes_per_chunk": 5.096, "boundary_entropy": 0.24109, "router_noise": 0.0, "grad_norm": 0.4776, "lr": 0.0002866, "router_trainable": true, "sec_per_step": 0.269, "alpha": 0.03, "t": 451.7}
{"event": "train", "step": 2100, "phase": 1, "objective": "global", "lm_loss": 1.45492, "bpb": 2.099, "chunk_loss": 0.030508, "F": 0.19548, "G": 0.24708, "chunks": 200.17, "bytes_per_chunk": 5.116, "boundary_entropy": 0.2577, "router_noise": 0.0, "grad_norm": 0.4612, "lr": 0.0002848, "router_trainable": true, "sec_per_step": 0.179, "alpha": 0.03, "t": 469.6}
{"event": "train", "step": 2200, "phase": 1, "objective": "global", "lm_loss": 1.41431, "bpb": 2.04041, "chunk_loss": 0.030314, "F": 0.19026, "G": 0.23874, "chunks": 194.83, "bytes_per_chunk": 5.256, "boundary_entropy": 0.24053, "router_noise": 0.0, "grad_norm": 0.4407, "lr": 0.0002829, "router_trainable": true, "sec_per_step": 0.179, "alpha": 0.03, "t": 487.5}
{"event": "train", "step": 2300, "phase": 1, "objective": "global", "lm_loss": 1.34576, "bpb": 1.94152, "chunk_loss": 0.030293, "F": 0.18903, "G": 0.2412, "chunks": 193.56, "bytes_per_chunk": 5.29, "boundary_entropy": 0.25453, "router_noise": 0.0, "grad_norm": 0.4168, "lr": 0.0002809, "router_trainable": true, "sec_per_step": 0.18, "alpha": 0.03, "t": 505.5}
{"event": "train", "step": 2400, "phase": 1, "objective": "global", "lm_loss": 1.37697, "bpb": 1.98655, "chunk_loss": 0.030153, "F": 0.18437, "G": 0.24639, "chunks": 188.8, "bytes_per_chunk": 5.424, "boundary_entropy": 0.26475, "router_noise": 0.0, "grad_norm": 0.4457, "lr": 0.0002788, "router_trainable": true, "sec_per_step": 0.178, "alpha": 0.03, "t": 523.3}
{"event": "train", "step": 2500, "phase": 1, "objective": "global", "lm_loss": 1.22974, "bpb": 1.77414, "chunk_loss": 0.030071, "F": 0.17984, "G": 0.25188, "chunks": 184.16, "bytes_per_chunk": 5.56, "boundary_entropy": 0.293, "router_noise": 0.0, "grad_norm": 0.3838, "lr": 0.0002765, "router_trainable": true, "sec_per_step": 0.182, "alpha": 0.03, "t": 541.4}
{"event": "train", "step": 2600, "phase": 1, "objective": "global", "lm_loss": 1.21741, "bpb": 1.75636, "chunk_loss": 0.030201, "F": 0.18198, "G": 0.24432, "chunks": 186.34, "bytes_per_chunk": 5.495, "boundary_entropy": 0.27751, "router_noise": 0.0, "grad_norm": 0.3756, "lr": 0.0002742, "router_trainable": true, "sec_per_step": 0.174, "alpha": 0.03, "t": 558.9}
{"event": "train", "step": 2700, "phase": 1, "objective": "global", "lm_loss": 1.36814, "bpb": 1.97381, "chunk_loss": 0.030337, "F": 0.19199, "G": 0.23904, "chunks": 196.59, "bytes_per_chunk": 5.209, "boundary_entropy": 0.22819, "router_noise": 0.0, "grad_norm": 0.4138, "lr": 0.0002718, "router_trainable": true, "sec_per_step": 0.175, "alpha": 0.03, "t": 576.4}
{"event": "train", "step": 2800, "phase": 1, "objective": "global", "lm_loss": 1.18487, "bpb": 1.7094, "chunk_loss": 0.030134, "F": 0.18095, "G": 0.24587, "chunks": 185.3, "bytes_per_chunk": 5.526, "boundary_entropy": 0.27651, "router_noise": 0.0, "grad_norm": 0.4121, "lr": 0.0002693, "router_trainable": true, "sec_per_step": 0.175, "alpha": 0.03, "t": 593.9}
{"event": "train", "step": 2900, "phase": 1, "objective": "global", "lm_loss": 1.23716, "bpb": 1.78485, "chunk_loss": 0.030216, "F": 0.18324, "G": 0.24418, "chunks": 187.64, "bytes_per_chunk": 5.457, "boundary_entropy": 0.26917, "router_noise": 0.0, "grad_norm": 0.3999, "lr": 0.0002667, "router_trainable": true, "sec_per_step": 0.175, "alpha": 0.03, "t": 611.3}
{"event": "probe", "step": 3000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3452.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.198357, "cps": 26.96875, "mean_p": 0.20205, "boundary_entropy": 0.115895, "bpb": 1.907748}, "de": {"chunks": 4175.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.203331, "cps": 32.617188, "mean_p": 0.209767, "boundary_entropy": 0.130407, "bpb": 2.09403}, "vi": {"chunks": 4067.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.220613, "cps": 31.773438, "mean_p": 0.239091, "boundary_entropy": 0.274246, "bpb": 1.813934}, "tr": {"chunks": 3639.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.201016, "cps": 28.429688, "mean_p": 0.201362, "boundary_entropy": 0.146601, "bpb": 2.031612}, "fi": {"chunks": 4128.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.219446, "cps": 32.25, "mean_p": 0.21966, "boundary_entropy": 0.139518, "bpb": 2.257063}, "ru": {"chunks": 5624.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.296781, "cps": 43.9375, "mean_p": 0.281748, "boundary_entropy": 0.423949, "bpb": 1.328186}, "el": {"chunks": 7138.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.347805, "cps": 55.765625, "mean_p": 0.284909, "boundary_entropy": 0.422443, "bpb": 1.394142}, "hi": {"chunks": 7814.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.464428, "cps": 61.046875, "mean_p": 0.287458, "boundary_entropy": 0.466344, "bpb": 1.123641}, "ta": {"chunks": 8889.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.475475, "cps": 69.445312, "mean_p": 0.315385, "boundary_entropy": 0.477578, "bpb": 0.940696}, "th": {"chunks": 7010.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.43581, "cps": 54.765625, "mean_p": 0.393012, "boundary_entropy": 0.592396, "bpb": 1.067473}, "zh": {"chunks": 3154.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.534033, "cps": 24.640625, "mean_p": 0.223675, "boundary_entropy": 0.168946, "bpb": 2.962823}, "my": {"chunks": 3499.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.184663, "cps": 27.335938, "mean_p": 0.343412, "boundary_entropy": 0.597138, "bpb": 1.08619}}, "bpb": {"en": 1.90775, "de": 2.09403, "vi": 1.81393, "tr": 2.03161, "fi": 2.25706, "ru": 1.32819, "el": 1.39414, "hi": 1.12364, "ta": 0.9407, "th": 1.06747, "zh": 2.96282, "my": 1.08619}, "gini_cpc": 0.216051, "gini_cps": 0.198902, "bpic": 6.131764, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0250791634685346, "vi": 1.1122037719198172, "tr": 1.0134091876348863, "fi": 1.10632095456093, "ru": 1.4961992436031881, "el": 1.7534324178243001, "hi": 2.341378721382096, "ta": 2.397070293796527, "th": 2.1971023361912434, "zh": 2.6922883969363207, "my": 0.9309661698993965}, "t": 631.9}
{"event": "eval", "step": 3000, "phase": 1, "bpb": {"en": 1.9109, "de": 2.09322, "vi": 1.82314, "tr": 2.02469, "fi": 2.26808, "ru": 1.32696, "el": 1.38998, "hi": 1.13226, "ta": 0.94195, "th": 1.06569, "zh": 2.97249, "my": 1.09424}, "bytes_per_chunk": {"en": 5.026, "de": 4.909, "vi": 5.999, "tr": 5.386, "fi": 4.754, "ru": 5.897, "el": 5.272, "hi": 5.352, "ta": 5.719, "th": 6.55, "zh": 5.145, "my": 15.346}, "bpb_mean": 1.6703, "t": 637.5}
{"event": "train", "step": 3000, "phase": 1, "objective": "global", "lm_loss": 1.26275, "bpb": 1.82176, "chunk_loss": 0.030233, "F": 0.18668, "G": 0.23562, "chunks": 191.16, "bytes_per_chunk": 5.357, "boundary_entropy": 0.23596, "router_noise": 0.0, "grad_norm": 0.4027, "lr": 0.000264, "router_trainable": true, "sec_per_step": 0.264, "alpha": 0.03, "t": 637.7}
{"event": "train", "step": 3100, "phase": 1, "objective": "global", "lm_loss": 1.18886, "bpb": 1.71517, "chunk_loss": 0.030212, "F": 0.18636, "G": 0.24202, "chunks": 190.83, "bytes_per_chunk": 5.366, "boundary_entropy": 0.24941, "router_noise": 0.0, "grad_norm": 0.4018, "lr": 0.0002612, "router_trainable": true, "sec_per_step": 0.178, "alpha": 0.03, "t": 655.5}
{"event": "train", "step": 3200, "phase": 1, "objective": "global", "lm_loss": 1.25729, "bpb": 1.81389, "chunk_loss": 0.030412, "F": 0.18925, "G": 0.25128, "chunks": 193.8, "bytes_per_chunk": 5.284, "boundary_entropy": 0.27038, "router_noise": 0.0, "grad_norm": 0.4299, "lr": 0.0002583, "router_trainable": true, "sec_per_step": 0.177, "alpha": 0.03, "t": 673.2}
{"event": "train", "step": 3300, "phase": 1, "objective": "global", "lm_loss": 1.25652, "bpb": 1.81277, "chunk_loss": 0.030129, "F": 0.18092, "G": 0.23587, "chunks": 185.27, "bytes_per_chunk": 5.527, "boundary_entropy": 0.24289, "router_noise": 0.0, "grad_norm": 0.4037, "lr": 0.0002553, "router_trainable": true, "sec_per_step": 0.175, "alpha": 0.03, "t": 690.8}
{"event": "train", "step": 3400, "phase": 1, "objective": "global", "lm_loss": 1.31454, "bpb": 1.89648, "chunk_loss": 0.030287, "F": 0.18904, "G": 0.23154, "chunks": 193.58, "bytes_per_chunk": 5.29, "boundary_entropy": 0.21991, "router_noise": 0.0, "grad_norm": 0.4009, "lr": 0.0002523, "router_trainable": true, "sec_per_step": 0.174, "alpha": 0.03, "t": 708.1}
{"event": "train", "step": 3500, "phase": 1, "objective": "global", "lm_loss": 1.17173, "bpb": 1.69045, "chunk_loss": 0.030292, "F": 0.18788, "G": 0.24955, "chunks": 192.39, "bytes_per_chunk": 5.323, "boundary_entropy": 0.25915, "router_noise": 0.0, "grad_norm": 0.4007, "lr": 0.0002492, "router_trainable": true, "sec_per_step": 0.175, "alpha": 0.03, "t": 725.6}
{"event": "train", "step": 3600, "phase": 1, "objective": "global", "lm_loss": 1.22072, "bpb": 1.76113, "chunk_loss": 0.030211, "F": 0.18571, "G": 0.24729, "chunks": 190.17, "bytes_per_chunk": 5.385, "boundary_entropy": 0.25238, "router_noise": 0.0, "grad_norm": 0.3842, "lr": 0.000246, "router_trainable": true, "sec_per_step": 0.176, "alpha": 0.03, "t": 743.2}
{"event": "train", "step": 3700, "phase": 1, "objective": "global", "lm_loss": 1.20949, "bpb": 1.74492, "chunk_loss": 0.030217, "F": 0.18182, "G": 0.24412, "chunks": 186.19, "bytes_per_chunk": 5.5, "boundary_entropy": 0.26327, "router_noise": 0.0, "grad_norm": 0.3852, "lr": 0.0002427, "router_trainable": true, "sec_per_step": 0.174, "alpha": 0.03, "t": 760.6}
{"event": "train", "step": 3800, "phase": 1, "objective": "global", "lm_loss": 1.2332, "bpb": 1.77913, "chunk_loss": 0.030263, "F": 0.18773, "G": 0.24103, "chunks": 192.23, "bytes_per_chunk": 5.327, "boundary_entropy": 0.23515, "router_noise": 0.0, "grad_norm": 0.3954, "lr": 0.0002394, "router_trainable": true, "sec_per_step": 0.175, "alpha": 0.03, "t": 778.1}
{"event": "train", "step": 3900, "phase": 1, "objective": "global", "lm_loss": 1.20775, "bpb": 1.74242, "chunk_loss": 0.030243, "F": 0.18465, "G": 0.23408, "chunks": 189.08, "bytes_per_chunk": 5.416, "boundary_entropy": 0.22798, "router_noise": 0.0, "grad_norm": 0.3717, "lr": 0.000236, "router_trainable": true, "sec_per_step": 0.175, "alpha": 0.03, "t": 795.6}
{"event": "probe", "step": 4000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3385.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.194507, "cps": 26.445312, "mean_p": 0.193621, "boundary_entropy": 0.099551, "bpb": 1.828979}, "de": {"chunks": 3951.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.192422, "cps": 30.867188, "mean_p": 0.194664, "boundary_entropy": 0.105284, "bpb": 1.987724}, "vi": {"chunks": 3928.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.213073, "cps": 30.6875, "mean_p": 0.227254, "boundary_entropy": 0.25524, "bpb": 1.707618}, "tr": {"chunks": 3470.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.191681, "cps": 27.109375, "mean_p": 0.189591, "boundary_entropy": 0.134298, "bpb": 1.914275}, "fi": {"chunks": 3804.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.202222, "cps": 29.71875, "mean_p": 0.199801, "boundary_entropy": 0.118473, "bpb": 2.107387}, "ru": {"chunks": 5167.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.272665, "cps": 40.367188, "mean_p": 0.274273, "boundary_entropy": 0.416083, "bpb": 1.26855}, "el": {"chunks": 5901.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.287531, "cps": 46.101562, "mean_p": 0.250569, "boundary_entropy": 0.380744, "bpb": 1.322275}, "hi": {"chunks": 6912.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.410817, "cps": 54.0, "mean_p": 0.249247, "boundary_entropy": 0.41176, "bpb": 1.064087}, "ta": {"chunks": 8087.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.432576, "cps": 63.179688, "mean_p": 0.28803, "boundary_entropy": 0.447416, "bpb": 0.912554}, "th": {"chunks": 6947.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.431893, "cps": 54.273438, "mean_p": 0.393579, "boundary_entropy": 0.586912, "bpb": 1.023449}, "zh": {"chunks": 3168.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.536404, "cps": 24.75, "mean_p": 0.221511, "boundary_entropy": 0.156063, "bpb": 2.838836}, "my": {"chunks": 10525.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.555468, "cps": 82.226562, "mean_p": 0.412583, "boundary_entropy": 0.638709, "bpb": 1.040089}}, "bpb": {"en": 1.82898, "de": 1.98772, "vi": 1.70762, "tr": 1.91427, "fi": 2.10739, "ru": 1.26855, "el": 1.32228, "hi": 1.06409, "ta": 0.91255, "th": 1.02345, "zh": 2.83884, "my": 1.04009}, "gini_cpc": 0.223577, "gini_cps": 0.217461, "bpic": 5.882152, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 0.989281914669767, "vi": 1.0954530890000758, "tr": 0.9854721844009141, "fi": 1.0396665516821413, "ru": 1.4018278841544451, "el": 1.4782579274281815, "hi": 2.1120981868702273, "ta": 2.223962299258524, "th": 2.220453490021851, "zh": 2.757764504564619, "my": 2.8557762377886267}, "t": 816.3}
{"event": "eval", "step": 4000, "phase": 1, "bpb": {"en": 1.82949, "de": 1.99062, "vi": 1.71804, "tr": 1.90632, "fi": 2.11867, "ru": 1.26818, "el": 1.31949, "hi": 1.07255, "ta": 0.91566, "th": 1.0213, "zh": 2.84785, "my": 1.04721}, "bytes_per_chunk": {"en": 5.086, "de": 5.167, "vi": 6.196, "tr": 5.594, "fi": 5.131, "ru": 6.558, "el": 6.323, "hi": 6.04, "ta": 6.204, "th": 6.635, "zh": 5.142, "my": 5.708}, "bpb_mean": 1.58795, "t": 821.9}
{"event": "train", "step": 4000, "phase": 1, "objective": "global", "lm_loss": 1.14377, "bpb": 1.65011, "chunk_loss": 0.03, "F": 0.17484, "G": 0.23137, "chunks": 179.03, "bytes_per_chunk": 5.72, "boundary_entropy": 0.24123, "router_noise": 0.0, "grad_norm": 0.3503, "lr": 0.0002325, "router_trainable": true, "sec_per_step": 0.265, "alpha": 0.03, "t": 822.1}
{"event": "train", "step": 4100, "phase": 1, "objective": "global", "lm_loss": 1.22141, "bpb": 1.76212, "chunk_loss": 0.030315, "F": 0.18671, "G": 0.23673, "chunks": 191.19, "bytes_per_chunk": 5.356, "boundary_entropy": 0.23191, "router_noise": 0.0, "grad_norm": 0.386, "lr": 0.000229, "router_trainable": true, "sec_per_step": 0.175, "alpha": 0.03, "t": 839.6}
{"event": "train", "step": 4200, "phase": 1, "objective": "global", "lm_loss": 1.19754, "bpb": 1.72769, "chunk_loss": 0.030444, "F": 0.19102, "G": 0.252, "chunks": 195.61, "bytes_per_chunk": 5.235, "boundary_entropy": 0.26293, "router_noise": 0.0, "grad_norm": 0.3939, "lr": 0.0002254, "router_trainable": true, "sec_per_step": 0.175, "alpha": 0.03, "t": 857.1}
{"event": "train", "step": 4300, "phase": 1, "objective": "global", "lm_loss": 1.12883, "bpb": 1.62856, "chunk_loss": 0.030138, "F": 0.18143, "G": 0.25277, "chunks": 185.78, "bytes_per_chunk": 5.512, "boundary_entropy": 0.26947, "router_noise": 0.0, "grad_norm": 0.3475, "lr": 0.0002217, "router_trainable": true, "sec_per_step": 0.176, "alpha": 0.03, "t": 874.8}
{"event": "phase_transition", "step": 4400, "phase": 2, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0, "t": 892.2}
{"event": "train", "step": 4400, "phase": 2, "objective": "global", "lm_loss": 1.1618, "bpb": 1.67612, "chunk_loss": 0.030451, "F": 0.19324, "G": 0.24114, "chunks": 197.88, "bytes_per_chunk": 5.175, "boundary_entropy": 0.22882, "router_noise": 0.0, "grad_norm": 0.3725, "lr": 0.0002181, "router_trainable": true, "sec_per_step": 0.176, "alpha": 0.03, "t": 892.4}
{"event": "train", "step": 4500, "phase": 2, "objective": "global", "lm_loss": 1.26405, "bpb": 1.82363, "chunk_loss": 0.030211, "F": 0.18739, "G": 0.2221, "chunks": 191.89, "bytes_per_chunk": 5.336, "boundary_entropy": 0.18119, "router_noise": 0.0, "grad_norm": 0.385, "lr": 0.0002143, "router_trainable": true, "sec_per_step": 0.176, "alpha": 0.03, "t": 910.0}
{"event": "train", "step": 4600, "phase": 2, "objective": "global", "lm_loss": 1.11023, "bpb": 1.60173, "chunk_loss": 0.030279, "F": 0.185, "G": 0.24273, "chunks": 189.44, "bytes_per_chunk": 5.405, "boundary_entropy": 0.24617, "router_noise": 0.0, "grad_norm": 0.3745, "lr": 0.0002105, "router_trainable": true, "sec_per_step": 0.174, "alpha": 0.03, "t": 927.4}
{"event": "train", "step": 4700, "phase": 2, "objective": "global", "lm_loss": 1.18169, "bpb": 1.70482, "chunk_loss": 0.030179, "F": 0.18387, "G": 0.22828, "chunks": 188.28, "bytes_per_chunk": 5.439, "boundary_entropy": 0.22047, "router_noise": 0.0, "grad_norm": 0.3953, "lr": 0.0002067, "router_trainable": true, "sec_per_step": 0.174, "alpha": 0.03, "t": 944.8}
{"event": "train", "step": 4800, "phase": 2, "objective": "global", "lm_loss": 1.1511, "bpb": 1.66068, "chunk_loss": 0.030108, "F": 0.18378, "G": 0.24578, "chunks": 188.19, "bytes_per_chunk": 5.441, "boundary_entropy": 0.23374, "router_noise": 0.0, "grad_norm": 0.3636, "lr": 0.0002029, "router_trainable": true, "sec_per_step": 0.177, "alpha": 0.03, "t": 962.4}
{"event": "train", "step": 4900, "phase": 2, "objective": "global", "lm_loss": 1.22525, "bpb": 1.76766, "chunk_loss": 0.030318, "F": 0.19113, "G": 0.22985, "chunks": 195.72, "bytes_per_chunk": 5.232, "boundary_entropy": 0.19519, "router_noise": 0.0, "grad_norm": 0.3884, "lr": 0.000199, "router_trainable": true, "sec_per_step": 0.175, "alpha": 0.03, "t": 979.9}
{"event": "probe", "step": 5000, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3372.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.19376, "cps": 26.34375, "mean_p": 0.191924, "boundary_entropy": 0.094239, "bpb": 1.777098}, "de": {"chunks": 3905.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.190182, "cps": 30.507812, "mean_p": 0.190928, "boundary_entropy": 0.09732, "bpb": 1.929646}, "vi": {"chunks": 3953.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.214429, "cps": 30.882812, "mean_p": 0.232904, "boundary_entropy": 0.26086, "bpb": 1.632632}, "tr": {"chunks": 3516.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.194222, "cps": 27.46875, "mean_p": 0.191048, "boundary_entropy": 0.126303, "bpb": 1.850978}, "fi": {"chunks": 3888.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.206688, "cps": 30.375, "mean_p": 0.203634, "boundary_entropy": 0.112075, "bpb": 2.040361}, "ru": {"chunks": 5810.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.306596, "cps": 45.390625, "mean_p": 0.30181, "boundary_entropy": 0.437538, "bpb": 1.21327}, "el": {"chunks": 6038.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.294207, "cps": 47.171875, "mean_p": 0.259802, "boundary_entropy": 0.383943, "bpb": 1.259051}, "hi": {"chunks": 6729.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.399941, "cps": 52.570312, "mean_p": 0.266269, "boundary_entropy": 0.440425, "bpb": 1.023271}, "ta": {"chunks": 8445.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.451725, "cps": 65.976562, "mean_p": 0.322752, "boundary_entropy": 0.478524, "bpb": 0.858981}, "th": {"chunks": 7212.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.448368, "cps": 56.34375, "mean_p": 0.408006, "boundary_entropy": 0.587538, "bpb": 0.970661}, "zh": {"chunks": 3169.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.536573, "cps": 24.757812, "mean_p": 0.217358, "boundary_entropy": 0.13631, "bpb": 2.694122}, "my": {"chunks": 6773.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.357452, "cps": 52.914062, "mean_p": 0.380824, "boundary_entropy": 0.613908, "bpb": 1.008841}}, "bpb": {"en": 1.7771, "de": 1.92965, "vi": 1.63263, "tr": 1.85098, "fi": 2.04036, "ru": 1.21327, "el": 1.25905, "hi": 1.02327, "ta": 0.85898, "th": 0.97066, "zh": 2.69412, "my": 1.00884}, "gini_cpc": 0.204674, "gini_cps": 0.182901, "bpic": 6.110189, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9815336322590161, "vi": 1.1066753245750434, "tr": 1.0023857157432274, "fi": 1.0667212017803622, "ru": 1.5823533554305675, "el": 1.5184091696097235, "hi": 2.064106063570458, "ta": 2.3313675036334343, "th": 2.3140418395600775, "zh": 2.769270318019072, "my": 1.8448210855407232}, "t": 1000.6}
{"event": "eval", "step": 5000, "phase": 2, "bpb": {"en": 1.77707, "de": 1.9333, "vi": 1.64586, "tr": 1.84193, "fi": 2.0519, "ru": 1.21237, "el": 1.25799, "hi": 1.03073, "ta": 0.86269, "th": 0.96707, "zh": 2.70009, "my": 1.01635}, "bytes_per_chunk": {"en": 5.104, "de": 5.229, "vi": 5.996, "tr": 5.6, "fi": 5.077, "ru": 5.926, "el": 6.325, "hi": 6.061, "ta": 5.931, "th": 6.426, "zh": 5.159, "my": 7.502}, "bpb_mean": 1.52478, "t": 1006.1}
{"event": "train", "step": 5000, "phase": 2, "objective": "global", "lm_loss": 1.1616, "bpb": 1.67583, "chunk_loss": 0.030091, "F": 0.18138, "G": 0.23421, "chunks": 185.73, "bytes_per_chunk": 5.513, "boundary_entropy": 0.22262, "router_noise": 0.0, "grad_norm": 0.3688, "lr": 0.000195, "router_trainable": true, "sec_per_step": 0.264, "alpha": 0.03, "t": 1006.3}
{"event": "train", "step": 5100, "phase": 2, "objective": "global", "lm_loss": 1.16824, "bpb": 1.68542, "chunk_loss": 0.030276, "F": 0.18671, "G": 0.23117, "chunks": 191.19, "bytes_per_chunk": 5.356, "boundary_entropy": 0.20842, "router_noise": 0.0, "grad_norm": 0.3843, "lr": 0.0001911, "router_trainable": true, "sec_per_step": 0.177, "alpha": 0.03, "t": 1024.0}
{"event": "train", "step": 5200, "phase": 2, "objective": "global", "lm_loss": 1.16827, "bpb": 1.68546, "chunk_loss": 0.029992, "F": 0.17932, "G": 0.23429, "chunks": 183.62, "bytes_per_chunk": 5.577, "boundary_entropy": 0.22193, "router_noise": 0.0, "grad_norm": 0.3662, "lr": 0.0001871, "router_trainable": true, "sec_per_step": 0.175, "alpha": 0.03, "t": 1041.5}
{"event": "train", "step": 5300, "phase": 2, "objective": "global", "lm_loss": 1.13215, "bpb": 1.63335, "chunk_loss": 0.030219, "F": 0.18466, "G": 0.23876, "chunks": 189.09, "bytes_per_chunk": 5.415, "boundary_entropy": 0.22582, "router_noise": 0.0, "grad_norm": 0.3657, "lr": 0.0001831, "router_trainable": true, "sec_per_step": 0.176, "alpha": 0.03, "t": 1059.1}
{"event": "train", "step": 5400, "phase": 2, "objective": "global", "lm_loss": 1.10089, "bpb": 1.58825, "chunk_loss": 0.030043, "F": 0.17834, "G": 0.22863, "chunks": 182.62, "bytes_per_chunk": 5.607, "boundary_entropy": 0.21794, "router_noise": 0.0, "grad_norm": 0.3604, "lr": 0.0001791, "router_trainable": true, "sec_per_step": 0.175, "alpha": 0.03, "t": 1076.6}
{"event": "train", "step": 5500, "phase": 2, "objective": "global", "lm_loss": 1.10203, "bpb": 1.58989, "chunk_loss": 0.030129, "F": 0.18175, "G": 0.24505, "chunks": 186.11, "bytes_per_chunk": 5.502, "boundary_entropy": 0.24532, "router_noise": 0.0, "grad_norm": 0.3515, "lr": 0.0001751, "router_trainable": true, "sec_per_step": 0.174, "alpha": 0.03, "t": 1094.0}
{"event": "train", "step": 5600, "phase": 2, "objective": "global", "lm_loss": 1.11634, "bpb": 1.61054, "chunk_loss": 0.030415, "F": 0.18979, "G": 0.24969, "chunks": 194.34, "bytes_per_chunk": 5.269, "boundary_entropy": 0.24558, "router_noise": 0.0, "grad_norm": 0.3586, "lr": 0.0001711, "router_trainable": true, "sec_per_step": 0.176, "alpha": 0.03, "t": 1111.6}
{"event": "train", "step": 5700, "phase": 2, "objective": "global", "lm_loss": 1.15406, "bpb": 1.66496, "chunk_loss": 0.030151, "F": 0.18558, "G": 0.23147, "chunks": 190.03, "bytes_per_chunk": 5.389, "boundary_entropy": 0.19921, "router_noise": 0.0, "grad_norm": 0.3774, "lr": 0.000167, "router_trainable": true, "sec_per_step": 0.174, "alpha": 0.03, "t": 1128.9}
{"event": "train", "step": 5800, "phase": 2, "objective": "global", "lm_loss": 1.07725, "bpb": 1.55414, "chunk_loss": 0.030175, "F": 0.18147, "G": 0.24654, "chunks": 185.83, "bytes_per_chunk": 5.51, "boundary_entropy": 0.25249, "router_noise": 0.0, "grad_norm": 0.3505, "lr": 0.000163, "router_trainable": true, "sec_per_step": 0.176, "alpha": 0.03, "t": 1146.5}
{"event": "train", "step": 5900, "phase": 2, "objective": "global", "lm_loss": 1.10531, "bpb": 1.59463, "chunk_loss": 0.030134, "F": 0.18018, "G": 0.23889, "chunks": 184.5, "bytes_per_chunk": 5.55, "boundary_entropy": 0.24056, "router_noise": 0.0, "grad_norm": 0.3563, "lr": 0.0001589, "router_trainable": true, "sec_per_step": 0.176, "alpha": 0.03, "t": 1164.1}
{"event": "probe", "step": 6000, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3359.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.193013, "cps": 26.242188, "mean_p": 0.191797, "boundary_entropy": 0.086287, "bpb": 1.738724}, "de": {"chunks": 3962.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.192958, "cps": 30.953125, "mean_p": 0.19429, "boundary_entropy": 0.093057, "bpb": 1.879768}, "vi": {"chunks": 3975.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.215622, "cps": 31.054688, "mean_p": 0.232966, "boundary_entropy": 0.253679, "bpb": 1.573299}, "tr": {"chunks": 3483.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.192399, "cps": 27.210938, "mean_p": 0.187891, "boundary_entropy": 0.115298, "bpb": 1.791042}, "fi": {"chunks": 3678.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.195524, "cps": 28.734375, "mean_p": 0.192617, "boundary_entropy": 0.105169, "bpb": 1.994083}, "ru": {"chunks": 5668.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.299103, "cps": 44.28125, "mean_p": 0.292808, "boundary_entropy": 0.423813, "bpb": 1.166312}, "el": {"chunks": 6144.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.299371, "cps": 48.0, "mean_p": 0.258287, "boundary_entropy": 0.374802, "bpb": 1.211176}, "hi": {"chunks": 6964.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.413908, "cps": 54.40625, "mean_p": 0.287319, "boundary_entropy": 0.460012, "bpb": 0.986439}, "ta": {"chunks": 8125.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.434608, "cps": 63.476562, "mean_p": 0.32412, "boundary_entropy": 0.474022, "bpb": 0.830696}, "th": {"chunks": 7917.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.492198, "cps": 61.851562, "mean_p": 0.421609, "boundary_entropy": 0.588754, "bpb": 0.933249}, "zh": {"chunks": 3170.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.536742, "cps": 24.765625, "mean_p": 0.215029, "boundary_entropy": 0.12132, "bpb": 2.609858}, "my": {"chunks": 11717.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.618377, "cps": 91.539062, "mean_p": 0.395955, "boundary_entropy": 0.625239, "bpb": 0.970951}}, "bpb": {"en": 1.73872, "de": 1.87977, "vi": 1.5733, "tr": 1.79104, "fi": 1.99408, "ru": 1.16631, "el": 1.21118, "hi": 0.98644, "ta": 0.8307, "th": 0.93325, "zh": 2.60986, "my": 0.97095}, "gini_cpc": 0.238026, "gini_cps": 0.23549, "bpic": 5.630425, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 0.999714935463497, "vi": 1.1171412992213818, "tr": 0.9968206802867328, "fi": 1.0130105254555077, "ru": 1.5496540071528386, "el": 1.5510452863339654, "hi": 2.144459289031663, "ta": 2.2517077184470695, "th": 2.550079359401986, "zh": 2.780865191059657, "my": 3.2038130806972505}, "t": 1184.5}
{"event": "eval", "step": 6000, "phase": 2, "bpb": {"en": 1.73739, "de": 1.88444, "vi": 1.58917, "tr": 1.78298, "fi": 2.00572, "ru": 1.16608, "el": 1.20877, "hi": 0.99338, "ta": 0.83429, "th": 0.93042, "zh": 2.61712, "my": 0.97901}, "bytes_per_chunk": {"en": 5.127, "de": 5.156, "vi": 6.11, "tr": 5.634, "fi": 5.293, "ru": 6.01, "el": 6.112, "hi": 6.082, "ta": 6.195, "th": 5.922, "zh": 5.164, "my": 4.489}, "bpb_mean": 1.4774, "t": 1190.0}
{"event": "train", "step": 6000, "phase": 2, "objective": "global", "lm_loss": 1.10713, "bpb": 1.59725, "chunk_loss": 0.030328, "F": 0.18506, "G": 0.24053, "chunks": 189.5, "bytes_per_chunk": 5.404, "boundary_entropy": 0.23308, "router_noise": 0.0, "grad_norm": 0.3767, "lr": 0.0001549, "router_trainable": true, "sec_per_step": 0.262, "alpha": 0.03, "t": 1190.2}
{"event": "train", "step": 6100, "phase": 2, "objective": "global", "lm_loss": 1.17545, "bpb": 1.69581, "chunk_loss": 0.030352, "F": 0.19028, "G": 0.24147, "chunks": 194.84, "bytes_per_chunk": 5.255, "boundary_entropy": 0.21867, "router_noise": 0.0, "grad_norm": 0.3727, "lr": 0.0001509, "router_trainable": true, "sec_per_step": 0.173, "alpha": 0.03, "t": 1207.5}
{"event": "train", "step": 6200, "phase": 2, "objective": "global", "lm_loss": 1.15676, "bpb": 1.66885, "chunk_loss": 0.03029, "F": 0.1844, "G": 0.23274, "chunks": 188.83, "bytes_per_chunk": 5.423, "boundary_entropy": 0.21224, "router_noise": 0.0, "grad_norm": 0.3712, "lr": 0.0001469, "router_trainable": true, "sec_per_step": 0.177, "alpha": 0.03, "t": 1225.2}
{"event": "train", "step": 6300, "phase": 2, "objective": "global", "lm_loss": 1.18844, "bpb": 1.71455, "chunk_loss": 0.030279, "F": 0.18593, "G": 0.23072, "chunks": 190.39, "bytes_per_chunk": 5.378, "boundary_entropy": 0.19991, "router_noise": 0.0, "grad_norm": 0.3721, "lr": 0.0001429, "router_trainable": true, "sec_per_step": 0.177, "alpha": 0.03, "t": 1243.0}
{"event": "train", "step": 6400, "phase": 2, "objective": "global", "lm_loss": 1.09897, "bpb": 1.58548, "chunk_loss": 0.030274, "F": 0.18523, "G": 0.23705, "chunks": 189.67, "bytes_per_chunk": 5.399, "boundary_entropy": 0.2195, "router_noise": 0.0, "grad_norm": 0.3557, "lr": 0.0001389, "router_trainable": true, "sec_per_step": 0.177, "alpha": 0.03, "t": 1260.7}
{"event": "train", "step": 6500, "phase": 2, "objective": "global", "lm_loss": 1.12095, "bpb": 1.61718, "chunk_loss": 0.03012, "F": 0.18134, "G": 0.23641, "chunks": 185.69, "bytes_per_chunk": 5.515, "boundary_entropy": 0.21945, "router_noise": 0.0, "grad_norm": 0.3543, "lr": 0.000135, "router_trainable": true, "sec_per_step": 0.177, "alpha": 0.03, "t": 1278.4}
{"event": "train", "step": 6600, "phase": 2, "objective": "global", "lm_loss": 1.07858, "bpb": 1.55606, "chunk_loss": 0.030098, "F": 0.18092, "G": 0.24063, "chunks": 185.27, "bytes_per_chunk": 5.527, "boundary_entropy": 0.23579, "router_noise": 0.0, "grad_norm": 0.3577, "lr": 0.000131, "router_trainable": true, "sec_per_step": 0.177, "alpha": 0.03, "t": 1296.1}
{"event": "train", "step": 6700, "phase": 2, "objective": "global", "lm_loss": 1.09674, "bpb": 1.58227, "chunk_loss": 0.030078, "F": 0.1817, "G": 0.25175, "chunks": 186.06, "bytes_per_chunk": 5.504, "boundary_entropy": 0.24281, "router_noise": 0.0, "grad_norm": 0.3648, "lr": 0.0001271, "router_trainable": true, "sec_per_step": 0.173, "alpha": 0.03, "t": 1313.4}
{"event": "train", "step": 6800, "phase": 2, "objective": "global", "lm_loss": 1.14611, "bpb": 1.65348, "chunk_loss": 0.030381, "F": 0.1899, "G": 0.236, "chunks": 194.45, "bytes_per_chunk": 5.266, "boundary_entropy": 0.19947, "router_noise": 0.0, "grad_norm": 0.3591, "lr": 0.0001233, "router_trainable": true, "sec_per_step": 0.174, "alpha": 0.03, "t": 1330.8}
{"event": "train", "step": 6900, "phase": 2, "objective": "global", "lm_loss": 1.12228, "bpb": 1.6191, "chunk_loss": 0.030224, "F": 0.18605, "G": 0.22371, "chunks": 190.52, "bytes_per_chunk": 5.375, "boundary_entropy": 0.18216, "router_noise": 0.0, "grad_norm": 0.3594, "lr": 0.0001195, "router_trainable": true, "sec_per_step": 0.174, "alpha": 0.03, "t": 1348.2}
{"event": "probe", "step": 7000, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3352.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.19261, "cps": 26.1875, "mean_p": 0.190953, "boundary_entropy": 0.080678, "bpb": 1.707888}, "de": {"chunks": 3884.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.189159, "cps": 30.34375, "mean_p": 0.190976, "boundary_entropy": 0.08426, "bpb": 1.844163}, "vi": {"chunks": 4004.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.217196, "cps": 31.28125, "mean_p": 0.225528, "boundary_entropy": 0.232213, "bpb": 1.529725}, "tr": {"chunks": 3526.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.194774, "cps": 27.546875, "mean_p": 0.189825, "boundary_entropy": 0.108215, "bpb": 1.751392}, "fi": {"chunks": 3711.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.197278, "cps": 28.992188, "mean_p": 0.193948, "boundary_entropy": 0.095902, "bpb": 1.932508}, "ru": {"chunks": 5927.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.31277, "cps": 46.304688, "mean_p": 0.294562, "boundary_entropy": 0.419634, "bpb": 1.135355}, "el": {"chunks": 6244.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.304244, "cps": 48.78125, "mean_p": 0.255121, "boundary_entropy": 0.358747, "bpb": 1.168993}, "hi": {"chunks": 7338.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.436137, "cps": 57.328125, "mean_p": 0.284394, "boundary_entropy": 0.445609, "bpb": 0.952654}, "ta": {"chunks": 8328.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.445467, "cps": 65.0625, "mean_p": 0.311207, "boundary_entropy": 0.448015, "bpb": 0.808259}, "th": {"chunks": 8145.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.506372, "cps": 63.632812, "mean_p": 0.399538, "boundary_entropy": 0.572346, "bpb": 0.912807}, "zh": {"chunks": 3205.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.542668, "cps": 25.039062, "mean_p": 0.215248, "boundary_entropy": 0.116129, "bpb": 2.567694}, "my": {"chunks": 7638.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.403103, "cps": 59.671875, "mean_p": 0.397853, "boundary_entropy": 0.629458, "bpb": 0.943814}}, "bpb": {"en": 1.70789, "de": 1.84416, "vi": 1.52972, "tr": 1.75139, "fi": 1.93251, "ru": 1.13535, "el": 1.16899, "hi": 0.95265, "ta": 0.80826, "th": 0.91281, "zh": 2.56769, "my": 0.94381}, "gini_cpc": 0.216227, "gini_cps": 0.198325, "bpic": 5.877018, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9820801301635983, "vi": 1.1276414649160793, "tr": 1.011234474118334, "fi": 1.0242339829600031, "ru": 1.6238496766393156, "el": 1.579581939443458, "hi": 2.2643457677126957, "ta": 2.312785507337035, "th": 2.6289972609711385, "zh": 2.8174401644054385, "my": 2.092841739205352}, "t": 1368.8}
{"event": "eval", "step": 7000, "phase": 2, "bpb": {"en": 1.70599, "de": 1.84875, "vi": 1.54472, "tr": 1.74332, "fi": 1.94581, "ru": 1.13501, "el": 1.16789, "hi": 0.95884, "ta": 0.81222, "th": 0.91019, "zh": 2.57246, "my": 0.95039}, "bytes_per_chunk": {"en": 5.124, "de": 5.24, "vi": 5.95, "tr": 5.551, "fi": 5.24, "ru": 5.861, "el": 6.094, "hi": 5.672, "ta": 6.08, "th": 5.61, "zh": 5.165, "my": 7.014}, "bpb_mean": 1.4413, "t": 1374.4}
{"event": "train", "step": 7000, "phase": 2, "objective": "global", "lm_loss": 1.21732, "bpb": 1.75621, "chunk_loss": 0.030281, "F": 0.1864, "G": 0.22364, "chunks": 190.88, "bytes_per_chunk": 5.365, "boundary_entropy": 0.17652, "router_noise": 0.0, "grad_norm": 0.381, "lr": 0.0001157, "router_trainable": true, "sec_per_step": 0.263, "alpha": 0.03, "t": 1374.6}
{"event": "train", "step": 7100, "phase": 2, "objective": "global", "lm_loss": 1.135, "bpb": 1.63745, "chunk_loss": 0.030178, "F": 0.18617, "G": 0.23517, "chunks": 190.64, "bytes_per_chunk": 5.371, "boundary_entropy": 0.20854, "router_noise": 0.0, "grad_norm": 0.369, "lr": 0.0001119, "router_trainable": true, "sec_per_step": 0.174, "alpha": 0.03, "t": 1392.0}
{"event": "train", "step": 7200, "phase": 2, "objective": "global", "lm_loss": 1.03486, "bpb": 1.49299, "chunk_loss": 0.030117, "F": 0.18054, "G": 0.23299, "chunks": 184.88, "bytes_per_chunk": 5.539, "boundary_entropy": 0.22203, "router_noise": 0.0, "grad_norm": 0.364, "lr": 0.0001083, "router_trainable": true, "sec_per_step": 0.174, "alpha": 0.03, "t": 1409.4}
{"event": "train", "step": 7300, "phase": 2, "objective": "global", "lm_loss": 1.12186, "bpb": 1.6185, "chunk_loss": 0.030141, "F": 0.18472, "G": 0.23206, "chunks": 189.16, "bytes_per_chunk": 5.414, "boundary_entropy": 0.19735, "router_noise": 0.0, "grad_norm": 0.3884, "lr": 0.0001046, "router_trainable": true, "sec_per_step": 0.175, "alpha": 0.03, "t": 1426.9}
{"event": "train", "step": 7400, "phase": 2, "objective": "global", "lm_loss": 1.04589, "bpb": 1.5089, "chunk_loss": 0.030114, "F": 0.17941, "G": 0.24425, "chunks": 183.72, "bytes_per_chunk": 5.574, "boundary_entropy": 0.24266, "router_noise": 0.0, "grad_norm": 0.3669, "lr": 0.000101, "router_trainable": true, "sec_per_step": 0.175, "alpha": 0.03, "t": 1444.4}
{"event": "train", "step": 7500, "phase": 2, "objective": "global", "lm_loss": 1.06127, "bpb": 1.53109, "chunk_loss": 0.030174, "F": 0.1857, "G": 0.23584, "chunks": 190.16, "bytes_per_chunk": 5.385, "boundary_entropy": 0.20503, "router_noise": 0.0, "grad_norm": 0.3653, "lr": 9.75e-05, "router_trainable": true, "sec_per_step": 0.173, "alpha": 0.03, "t": 1461.6}
{"event": "train", "step": 7600, "phase": 2, "objective": "global", "lm_loss": 1.05168, "bpb": 1.51725, "chunk_loss": 0.030258, "F": 0.18555, "G": 0.24718, "chunks": 190.0, "bytes_per_chunk": 5.389, "boundary_entropy": 0.2374, "router_noise": 0.0, "grad_norm": 0.3689, "lr": 9.4e-05, "router_trainable": true, "sec_per_step": 0.176, "alpha": 0.03, "t": 1479.2}
{"event": "phase_transition", "step": 7700, "phase": 3, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0, "t": 1496.5}
{"event": "train", "step": 7700, "phase": 3, "objective": "global", "lm_loss": 1.0419, "bpb": 1.50314, "chunk_loss": 0.03007, "F": 0.17937, "G": 0.23526, "chunks": 183.67, "bytes_per_chunk": 5.575, "boundary_entropy": 0.21541, "router_noise": 0.0, "grad_norm": 0.344, "lr": 9.06e-05, "router_trainable": true, "sec_per_step": 0.175, "alpha": 0.03, "t": 1496.6}
{"event": "train", "step": 7800, "phase": 3, "objective": "global", "lm_loss": 1.09557, "bpb": 1.58058, "chunk_loss": 0.030133, "F": 0.18214, "G": 0.22307, "chunks": 186.52, "bytes_per_chunk": 5.49, "boundary_entropy": 0.1847, "router_noise": 0.0, "grad_norm": 0.3712, "lr": 8.73e-05, "router_trainable": true, "sec_per_step": 0.176, "alpha": 0.03, "t": 1514.2}
{"event": "train", "step": 7900, "phase": 3, "objective": "global", "lm_loss": 1.08847, "bpb": 1.57033, "chunk_loss": 0.03014, "F": 0.18021, "G": 0.23202, "chunks": 184.53, "bytes_per_chunk": 5.549, "boundary_entropy": 0.20808, "router_noise": 0.0, "grad_norm": 0.365, "lr": 8.4e-05, "router_trainable": true, "sec_per_step": 0.175, "alpha": 0.03, "t": 1531.7}
{"event": "probe", "step": 8000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3347.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.192323, "cps": 26.148438, "mean_p": 0.190137, "boundary_entropy": 0.074221, "bpb": 1.682967}, "de": {"chunks": 3895.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.189695, "cps": 30.429688, "mean_p": 0.190213, "boundary_entropy": 0.079307, "bpb": 1.813794}, "vi": {"chunks": 3895.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.211283, "cps": 30.429688, "mean_p": 0.21775, "boundary_entropy": 0.215018, "bpb": 1.501892}, "tr": {"chunks": 3492.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.192896, "cps": 27.28125, "mean_p": 0.185692, "boundary_entropy": 0.097637, "bpb": 1.719729}, "fi": {"chunks": 3703.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.196853, "cps": 28.929688, "mean_p": 0.19281, "boundary_entropy": 0.088845, "bpb": 1.893711}, "ru": {"chunks": 5474.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.288865, "cps": 42.765625, "mean_p": 0.280138, "boundary_entropy": 0.402295, "bpb": 1.109586}, "el": {"chunks": 6301.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.307021, "cps": 49.226562, "mean_p": 0.252477, "boundary_entropy": 0.345692, "bpb": 1.14116}, "hi": {"chunks": 6499.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.38627, "cps": 50.773438, "mean_p": 0.275777, "boundary_entropy": 0.445781, "bpb": 0.937422}, "ta": {"chunks": 8178.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.437443, "cps": 63.890625, "mean_p": 0.305541, "boundary_entropy": 0.437045, "bpb": 0.783557}, "th": {"chunks": 7402.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.46018, "cps": 57.828125, "mean_p": 0.400764, "boundary_entropy": 0.575213, "bpb": 0.889915}, "zh": {"chunks": 3198.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.541483, "cps": 24.984375, "mean_p": 0.213586, "boundary_entropy": 0.106851, "bpb": 2.523494}, "my": {"chunks": 9223.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.486753, "cps": 72.054688, "mean_p": 0.388496, "boundary_entropy": 0.615092, "bpb": 0.921547}}, "bpb": {"en": 1.68297, "de": 1.81379, "vi": 1.50189, "tr": 1.71973, "fi": 1.89371, "ru": 1.10959, "el": 1.14116, "hi": 0.93742, "ta": 0.78356, "th": 0.88991, "zh": 2.52349, "my": 0.92155}, "gini_cpc": 0.217091, "gini_cps": 0.206248, "bpic": 5.940239, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9863327706522578, "vi": 1.0985826297696126, "tr": 1.0029795806590318, "fi": 1.023552766494542, "ru": 1.5019794357645526, "el": 1.5963828088494532, "hi": 2.008444669573915, "ta": 2.274521490506202, "th": 2.39274503290008, "zh": 2.815486340072752, "my": 2.530913138967472}, "t": 1552.5}
{"event": "eval", "step": 8000, "phase": 3, "bpb": {"en": 1.68086, "de": 1.8176, "vi": 1.51759, "tr": 1.7111, "fi": 1.90616, "ru": 1.10941, "el": 1.13935, "hi": 0.94364, "ta": 0.78821, "th": 0.88769, "zh": 2.52943, "my": 0.92884}, "bytes_per_chunk": {"en": 5.155, "de": 5.239, "vi": 6.057, "tr": 5.614, "fi": 5.23, "ru": 6.144, "el": 6.017, "hi": 6.278, "ta": 6.117, "th": 6.248, "zh": 5.16, "my": 5.968}, "bpb_mean": 1.41332, "t": 1558.1}
{"event": "train", "step": 8000, "phase": 3, "objective": "global", "lm_loss": 1.07116, "bpb": 1.54535, "chunk_loss": 0.030143, "F": 0.18367, "G": 0.23634, "chunks": 188.08, "bytes_per_chunk": 5.445, "boundary_entropy": 0.21508, "router_noise": 0.0, "grad_norm": 0.4068, "lr": 8.08e-05, "router_trainable": true, "sec_per_step": 0.265, "alpha": 0.03, "t": 1558.3}
{"event": "train", "step": 8100, "phase": 3, "objective": "global", "lm_loss": 1.07012, "bpb": 1.54386, "chunk_loss": 0.030153, "F": 0.18011, "G": 0.23765, "chunks": 184.44, "bytes_per_chunk": 5.552, "boundary_entropy": 0.2334, "router_noise": 0.0, "grad_norm": 0.3771, "lr": 7.77e-05, "router_trainable": true, "sec_per_step": 0.176, "alpha": 0.03, "t": 1575.9}
{"event": "train", "step": 8200, "phase": 3, "objective": "global", "lm_loss": 1.12723, "bpb": 1.62626, "chunk_loss": 0.030123, "F": 0.18211, "G": 0.21997, "chunks": 186.48, "bytes_per_chunk": 5.491, "boundary_entropy": 0.17394, "router_noise": 0.0, "grad_norm": 0.3946, "lr": 7.47e-05, "router_trainable": true, "sec_per_step": 0.175, "alpha": 0.03, "t": 1593.4}
{"event": "train", "step": 8300, "phase": 3, "objective": "global", "lm_loss": 1.0551, "bpb": 1.52218, "chunk_loss": 0.030241, "F": 0.18468, "G": 0.23311, "chunks": 189.11, "bytes_per_chunk": 5.415, "boundary_entropy": 0.19943, "router_noise": 0.0, "grad_norm": 0.4059, "lr": 7.17e-05, "router_trainable": true, "sec_per_step": 0.176, "alpha": 0.03, "t": 1610.9}
{"event": "train", "step": 8400, "phase": 3, "objective": "global", "lm_loss": 1.07416, "bpb": 1.54969, "chunk_loss": 0.03016, "F": 0.17737, "G": 0.23143, "chunks": 181.62, "bytes_per_chunk": 5.638, "boundary_entropy": 0.21574, "router_noise": 0.0, "grad_norm": 0.3871, "lr": 6.88e-05, "router_trainable": true, "sec_per_step": 0.176, "alpha": 0.03, "t": 1628.5}
{"event": "train", "step": 8500, "phase": 3, "objective": "global", "lm_loss": 1.06773, "bpb": 1.54041, "chunk_loss": 0.030082, "F": 0.17928, "G": 0.23781, "chunks": 183.58, "bytes_per_chunk": 5.578, "boundary_entropy": 0.21838, "router_noise": 0.0, "grad_norm": 0.3918, "lr": 6.6e-05, "router_trainable": true, "sec_per_step": 0.175, "alpha": 0.03, "t": 1646.0}
{"event": "train", "step": 8600, "phase": 3, "objective": "global", "lm_loss": 1.03179, "bpb": 1.48856, "chunk_loss": 0.030085, "F": 0.17792, "G": 0.23712, "chunks": 182.19, "bytes_per_chunk": 5.621, "boundary_entropy": 0.22746, "router_noise": 0.0, "grad_norm": 0.3671, "lr": 6.33e-05, "router_trainable": true, "sec_per_step": 0.171, "alpha": 0.03, "t": 1663.2}
{"event": "train", "step": 8700, "phase": 3, "objective": "global", "lm_loss": 1.04812, "bpb": 1.51211, "chunk_loss": 0.030228, "F": 0.18489, "G": 0.24783, "chunks": 189.33, "bytes_per_chunk": 5.409, "boundary_entropy": 0.24221, "router_noise": 0.0, "grad_norm": 0.3737, "lr": 6.07e-05, "router_trainable": true, "sec_per_step": 0.174, "alpha": 0.03, "t": 1680.5}
{"event": "train", "step": 8800, "phase": 3, "objective": "global", "lm_loss": 1.04631, "bpb": 1.50951, "chunk_loss": 0.030091, "F": 0.17924, "G": 0.23321, "chunks": 183.55, "bytes_per_chunk": 5.579, "boundary_entropy": 0.21518, "router_noise": 0.0, "grad_norm": 0.3689, "lr": 5.82e-05, "router_trainable": true, "sec_per_step": 0.175, "alpha": 0.03, "t": 1698.0}
{"event": "train", "step": 8900, "phase": 3, "objective": "global", "lm_loss": 1.03837, "bpb": 1.49804, "chunk_loss": 0.030165, "F": 0.18022, "G": 0.23547, "chunks": 184.55, "bytes_per_chunk": 5.549, "boundary_entropy": 0.21149, "router_noise": 0.0, "grad_norm": 0.3857, "lr": 5.58e-05, "router_trainable": true, "sec_per_step": 0.176, "alpha": 0.03, "t": 1715.5}
{"event": "probe", "step": 9000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3348.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.192381, "cps": 26.15625, "mean_p": 0.189686, "boundary_entropy": 0.071728, "bpb": 1.665606}, "de": {"chunks": 3888.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.189354, "cps": 30.375, "mean_p": 0.189481, "boundary_entropy": 0.076217, "bpb": 1.789144}, "vi": {"chunks": 3966.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.215134, "cps": 30.984375, "mean_p": 0.219004, "boundary_entropy": 0.214752, "bpb": 1.476649}, "tr": {"chunks": 3501.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.193393, "cps": 27.351562, "mean_p": 0.186192, "boundary_entropy": 0.094116, "bpb": 1.699279}, "fi": {"chunks": 3702.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.1968, "cps": 28.921875, "mean_p": 0.192733, "boundary_entropy": 0.085284, "bpb": 1.863127}, "ru": {"chunks": 5702.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.300897, "cps": 44.546875, "mean_p": 0.279884, "boundary_entropy": 0.393335, "bpb": 1.089548}, "el": {"chunks": 6510.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.317205, "cps": 50.859375, "mean_p": 0.255819, "boundary_entropy": 0.341575, "bpb": 1.116353}, "hi": {"chunks": 7270.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.432095, "cps": 56.796875, "mean_p": 0.27261, "boundary_entropy": 0.419678, "bpb": 0.909181}, "ta": {"chunks": 8156.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.436266, "cps": 63.71875, "mean_p": 0.29618, "boundary_entropy": 0.420271, "bpb": 0.765964}, "th": {"chunks": 7751.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.481878, "cps": 60.554688, "mean_p": 0.388705, "boundary_entropy": 0.561302, "bpb": 0.873144}, "zh": {"chunks": 3181.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.538605, "cps": 24.851562, "mean_p": 0.210518, "boundary_entropy": 0.095453, "bpb": 2.496011}, "my": {"chunks": 7993.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.421839, "cps": 62.445312, "mean_p": 0.39317, "boundary_entropy": 0.622262, "bpb": 0.896443}}, "bpb": {"en": 1.66561, "de": 1.78914, "vi": 1.47665, "tr": 1.69928, "fi": 1.86313, "ru": 1.08955, "el": 1.11635, "hi": 0.90918, "ta": 0.76596, "th": 0.87314, "zh": 2.49601, "my": 0.89644}, "gini_cpc": 0.211964, "gini_cps": 0.197177, "bpic": 5.907231, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9842660830794803, "vi": 1.118274028772992, "tr": 1.0052642317348934, "fi": 1.022970716630569, "ru": 1.5640717413302314, "el": 1.6488411375854732, "hi": 2.2460427018592193, "ta": 2.267725159871585, "th": 2.5048131955794877, "zh": 2.799683239327724, "my": 2.1927297440344184}, "t": 1735.9}
{"event": "eval", "step": 9000, "phase": 3, "bpb": {"en": 1.6623, "de": 1.79506, "vi": 1.49286, "tr": 1.69126, "fi": 1.8774, "ru": 1.08989, "el": 1.11517, "hi": 0.91603, "ta": 0.77056, "th": 0.86955, "zh": 2.50104, "my": 0.90496}, "bytes_per_chunk": {"en": 5.154, "de": 5.24, "vi": 6.023, "tr": 5.607, "fi": 5.231, "ru": 5.985, "el": 5.835, "hi": 5.695, "ta": 6.146, "th": 5.995, "zh": 5.16, "my": 6.829}, "bpb_mean": 1.39051, "t": 1741.4}
{"event": "train", "step": 9000, "phase": 3, "objective": "global", "lm_loss": 1.07783, "bpb": 1.55498, "chunk_loss": 0.03016, "F": 0.18391, "G": 0.2295, "chunks": 188.33, "bytes_per_chunk": 5.437, "boundary_entropy": 0.19333, "router_noise": 0.0, "grad_norm": 0.3948, "lr": 5.35e-05, "router_trainable": true, "sec_per_step": 0.261, "alpha": 0.03, "t": 1741.6}
{"event": "train", "step": 9100, "phase": 3, "objective": "global", "lm_loss": 1.01435, "bpb": 1.46339, "chunk_loss": 0.030168, "F": 0.18277, "G": 0.24002, "chunks": 187.16, "bytes_per_chunk": 5.471, "boundary_entropy": 0.22465, "router_noise": 0.0, "grad_norm": 0.3662, "lr": 5.12e-05, "router_trainable": true, "sec_per_step": 0.173, "alpha": 0.03, "t": 1758.9}
{"event": "train", "step": 9200, "phase": 3, "objective": "global", "lm_loss": 1.09492, "bpb": 1.57963, "chunk_loss": 0.030163, "F": 0.18118, "G": 0.22149, "chunks": 185.53, "bytes_per_chunk": 5.519, "boundary_entropy": 0.18357, "router_noise": 0.0, "grad_norm": 0.4039, "lr": 4.91e-05, "router_trainable": true, "sec_per_step": 0.175, "alpha": 0.03, "t": 1776.5}
{"event": "train", "step": 9300, "phase": 3, "objective": "global", "lm_loss": 0.96145, "bpb": 1.38708, "chunk_loss": 0.030154, "F": 0.18057, "G": 0.24653, "chunks": 184.91, "bytes_per_chunk": 5.538, "boundary_entropy": 0.24532, "router_noise": 0.0, "grad_norm": 0.3736, "lr": 4.71e-05, "router_trainable": true, "sec_per_step": 0.174, "alpha": 0.03, "t": 1793.9}
{"event": "train", "step": 9400, "phase": 3, "objective": "global", "lm_loss": 1.07704, "bpb": 1.55384, "chunk_loss": 0.030382, "F": 0.19038, "G": 0.22852, "chunks": 194.95, "bytes_per_chunk": 5.253, "boundary_entropy": 0.17751, "router_noise": 0.0, "grad_norm": 0.3736, "lr": 4.52e-05, "router_trainable": true, "sec_per_step": 0.176, "alpha": 0.03, "t": 1811.5}
{"event": "train", "step": 9500, "phase": 3, "objective": "global", "lm_loss": 1.01201, "bpb": 1.46002, "chunk_loss": 0.030234, "F": 0.18457, "G": 0.2377, "chunks": 189.0, "bytes_per_chunk": 5.418, "boundary_entropy": 0.21969, "router_noise": 0.0, "grad_norm": 0.3703, "lr": 4.34e-05, "router_trainable": true, "sec_per_step": 0.178, "alpha": 0.03, "t": 1829.3}
{"event": "train", "step": 9600, "phase": 3, "objective": "global", "lm_loss": 1.08168, "bpb": 1.56054, "chunk_loss": 0.030116, "F": 0.18079, "G": 0.22649, "chunks": 185.12, "bytes_per_chunk": 5.531, "boundary_entropy": 0.18344, "router_noise": 0.0, "grad_norm": 0.3955, "lr": 4.17e-05, "router_trainable": true, "sec_per_step": 0.176, "alpha": 0.03, "t": 1846.9}
{"event": "train", "step": 9700, "phase": 3, "objective": "global", "lm_loss": 1.06136, "bpb": 1.53121, "chunk_loss": 0.0303, "F": 0.18666, "G": 0.22815, "chunks": 191.14, "bytes_per_chunk": 5.357, "boundary_entropy": 0.18374, "router_noise": 0.0, "grad_norm": 0.3717, "lr": 4.01e-05, "router_trainable": true, "sec_per_step": 0.176, "alpha": 0.03, "t": 1864.5}
{"event": "train", "step": 9800, "phase": 3, "objective": "global", "lm_loss": 1.02242, "bpb": 1.47504, "chunk_loss": 0.030183, "F": 0.18152, "G": 0.23801, "chunks": 185.88, "bytes_per_chunk": 5.509, "boundary_entropy": 0.2131, "router_noise": 0.0, "grad_norm": 0.3741, "lr": 3.86e-05, "router_trainable": true, "sec_per_step": 0.176, "alpha": 0.03, "t": 1882.2}
{"event": "train", "step": 9900, "phase": 3, "objective": "global", "lm_loss": 1.03612, "bpb": 1.4948, "chunk_loss": 0.030178, "F": 0.18452, "G": 0.2314, "chunks": 188.95, "bytes_per_chunk": 5.419, "boundary_entropy": 0.19878, "router_noise": 0.0, "grad_norm": 0.3813, "lr": 3.72e-05, "router_trainable": true, "sec_per_step": 0.172, "alpha": 0.03, "t": 1899.4}
{"event": "probe", "step": 10000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3350.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.192496, "cps": 26.171875, "mean_p": 0.189227, "boundary_entropy": 0.071166, "bpb": 1.652155}, "de": {"chunks": 3873.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.188623, "cps": 30.257812, "mean_p": 0.18782, "boundary_entropy": 0.072823, "bpb": 1.7756}, "vi": {"chunks": 4037.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.218986, "cps": 31.539062, "mean_p": 0.221059, "boundary_entropy": 0.212773, "bpb": 1.462193}, "tr": {"chunks": 3516.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.194222, "cps": 27.46875, "mean_p": 0.185839, "boundary_entropy": 0.091542, "bpb": 1.681755}, "fi": {"chunks": 3677.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.195471, "cps": 28.726562, "mean_p": 0.191057, "boundary_entropy": 0.082823, "bpb": 1.846041}, "ru": {"chunks": 5655.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.298417, "cps": 44.179688, "mean_p": 0.276995, "boundary_entropy": 0.386467, "bpb": 1.077108}, "el": {"chunks": 6526.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.317985, "cps": 50.984375, "mean_p": 0.252428, "boundary_entropy": 0.331489, "bpb": 1.10094}, "hi": {"chunks": 7152.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.425082, "cps": 55.875, "mean_p": 0.260083, "boundary_entropy": 0.400422, "bpb": 0.898578}, "ta": {"chunks": 8402.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.449425, "cps": 65.640625, "mean_p": 0.303654, "boundary_entropy": 0.423875, "bpb": 0.754097}, "th": {"chunks": 7708.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.479204, "cps": 60.21875, "mean_p": 0.403503, "boundary_entropy": 0.570043, "bpb": 0.860326}, "zh": {"chunks": 3203.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.54233, "cps": 25.023438, "mean_p": 0.211499, "boundary_entropy": 0.094925, "bpb": 2.487095}, "my": {"chunks": 9602.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.506755, "cps": 75.015625, "mean_p": 0.402146, "boundary_entropy": 0.623301, "bpb": 0.881309}}, "bpb": {"en": 1.65215, "de": 1.7756, "vi": 1.46219, "tr": 1.68176, "fi": 1.84604, "ru": 1.07711, "el": 1.10094, "hi": 0.89858, "ta": 0.7541, "th": 0.86033, "zh": 2.48709, "my": 0.88131}, "gini_cpc": 0.220787, "gini_cps": 0.21508, "bpic": 5.753752, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.979883405744375, "vi": 1.1376139805447942, "tr": 1.0089685473093022, "fi": 1.0154558820378994, "ru": 1.550253455676761, "el": 1.6519067867794488, "hi": 2.208267831718081, "ta": 2.3347292316167225, "th": 2.4894302005669506, "zh": 2.8173630156026506, "my": 2.6325561237510993}, "t": 1919.9}
{"event": "eval", "step": 10000, "phase": 3, "bpb": {"en": 1.64895, "de": 1.78144, "vi": 1.47747, "tr": 1.6733, "fi": 1.8602, "ru": 1.07676, "el": 1.09947, "hi": 0.90479, "ta": 0.75911, "th": 0.85779, "zh": 2.49261, "my": 0.88888}, "bytes_per_chunk": {"en": 5.148, "de": 5.273, "vi": 5.925, "tr": 5.599, "fi": 5.27, "ru": 6.052, "el": 5.859, "hi": 5.835, "ta": 6.023, "th": 6.031, "zh": 5.163, "my": 5.577}, "bpb_mean": 1.37673, "t": 1925.5}
{"event": "train", "step": 10000, "phase": 3, "objective": "global", "lm_loss": 1.04555, "bpb": 1.50841, "chunk_loss": 0.030215, "F": 0.1861, "G": 0.22883, "chunks": 190.56, "bytes_per_chunk": 5.374, "boundary_entropy": 0.1838, "router_noise": 0.0, "grad_norm": 0.3715, "lr": 3.6e-05, "router_trainable": true, "sec_per_step": 0.263, "alpha": 0.03, "t": 1925.7}
{"event": "train", "step": 10100, "phase": 3, "objective": "global", "lm_loss": 1.05212, "bpb": 1.51789, "chunk_loss": 0.030147, "F": 0.18399, "G": 0.23333, "chunks": 188.41, "bytes_per_chunk": 5.435, "boundary_entropy": 0.19438, "router_noise": 0.0, "grad_norm": 0.3968, "lr": 3.49e-05, "router_trainable": true, "sec_per_step": 0.173, "alpha": 0.03, "t": 1943.0}
{"event": "train", "step": 10200, "phase": 3, "objective": "global", "lm_loss": 1.01301, "bpb": 1.46147, "chunk_loss": 0.030189, "F": 0.18175, "G": 0.23171, "chunks": 186.11, "bytes_per_chunk": 5.502, "boundary_entropy": 0.2067, "router_noise": 0.0, "grad_norm": 0.3956, "lr": 3.38e-05, "router_trainable": true, "sec_per_step": 0.174, "alpha": 0.03, "t": 1960.4}
{"event": "train", "step": 10300, "phase": 3, "objective": "global", "lm_loss": 0.96166, "bpb": 1.38739, "chunk_loss": 0.030152, "F": 0.18097, "G": 0.25124, "chunks": 185.31, "bytes_per_chunk": 5.526, "boundary_entropy": 0.24632, "router_noise": 0.0, "grad_norm": 0.3721, "lr": 3.3e-05, "router_trainable": true, "sec_per_step": 0.173, "alpha": 0.03, "t": 1977.7}
{"event": "train", "step": 10400, "phase": 3, "objective": "global", "lm_loss": 1.05759, "bpb": 1.52578, "chunk_loss": 0.030151, "F": 0.18263, "G": 0.22973, "chunks": 187.02, "bytes_per_chunk": 5.475, "boundary_entropy": 0.19334, "router_noise": 0.0, "grad_norm": 0.3897, "lr": 3.22e-05, "router_trainable": true, "sec_per_step": 0.175, "alpha": 0.03, "t": 1995.2}
{"event": "train", "step": 10500, "phase": 3, "objective": "global", "lm_loss": 1.00213, "bpb": 1.44576, "chunk_loss": 0.030112, "F": 0.17918, "G": 0.24133, "chunks": 183.48, "bytes_per_chunk": 5.581, "boundary_entropy": 0.23021, "router_noise": 0.0, "grad_norm": 0.3852, "lr": 3.15e-05, "router_trainable": true, "sec_per_step": 0.172, "alpha": 0.03, "t": 2012.4}
{"event": "train", "step": 10600, "phase": 3, "objective": "global", "lm_loss": 0.96382, "bpb": 1.3905, "chunk_loss": 0.030223, "F": 0.1801, "G": 0.2491, "chunks": 184.42, "bytes_per_chunk": 5.552, "boundary_entropy": 0.25314, "router_noise": 0.0, "grad_norm": 0.3817, "lr": 3.1e-05, "router_trainable": true, "sec_per_step": 0.177, "alpha": 0.03, "t": 2030.1}
{"event": "train", "step": 10700, "phase": 3, "objective": "global", "lm_loss": 1.02656, "bpb": 1.48101, "chunk_loss": 0.030183, "F": 0.18367, "G": 0.23757, "chunks": 188.08, "bytes_per_chunk": 5.445, "boundary_entropy": 0.21558, "router_noise": 0.0, "grad_norm": 0.3926, "lr": 3.05e-05, "router_trainable": true, "sec_per_step": 0.173, "alpha": 0.03, "t": 2047.4}
{"event": "train", "step": 10800, "phase": 3, "objective": "global", "lm_loss": 1.00336, "bpb": 1.44755, "chunk_loss": 0.030096, "F": 0.18236, "G": 0.22348, "chunks": 186.73, "bytes_per_chunk": 5.484, "boundary_entropy": 0.17729, "router_noise": 0.0, "grad_norm": 0.3907, "lr": 3.02e-05, "router_trainable": true, "sec_per_step": 0.175, "alpha": 0.03, "t": 2064.9}
{"event": "train", "step": 10900, "phase": 3, "objective": "global", "lm_loss": 1.01731, "bpb": 1.46767, "chunk_loss": 0.030244, "F": 0.18462, "G": 0.23488, "chunks": 189.05, "bytes_per_chunk": 5.417, "boundary_entropy": 0.20413, "router_noise": 0.0, "grad_norm": 0.391, "lr": 3.01e-05, "router_trainable": true, "sec_per_step": 0.174, "alpha": 0.03, "t": 2082.3}
{"event": "probe", "step": 11000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3354.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.192725, "cps": 26.203125, "mean_p": 0.189456, "boundary_entropy": 0.068874, "bpb": 1.643497}, "de": {"chunks": 3872.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.188574, "cps": 30.25, "mean_p": 0.18801, "boundary_entropy": 0.07031, "bpb": 1.767157}, "vi": {"chunks": 4092.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.221969, "cps": 31.96875, "mean_p": 0.221989, "boundary_entropy": 0.212375, "bpb": 1.451768}, "tr": {"chunks": 3518.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.194332, "cps": 27.484375, "mean_p": 0.18714, "boundary_entropy": 0.090893, "bpb": 1.67139}, "fi": {"chunks": 3673.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.195258, "cps": 28.695312, "mean_p": 0.191426, "boundary_entropy": 0.081794, "bpb": 1.832604}, "ru": {"chunks": 5847.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.308549, "cps": 45.679688, "mean_p": 0.274957, "boundary_entropy": 0.375384, "bpb": 1.069289}, "el": {"chunks": 6585.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.32086, "cps": 51.445312, "mean_p": 0.253164, "boundary_entropy": 0.328416, "bpb": 1.091542}, "hi": {"chunks": 7484.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.444814, "cps": 58.46875, "mean_p": 0.268467, "boundary_entropy": 0.403481, "bpb": 0.888493}, "ta": {"chunks": 8479.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.453544, "cps": 66.242188, "mean_p": 0.299847, "boundary_entropy": 0.410619, "bpb": 0.74745}, "th": {"chunks": 7791.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.484364, "cps": 60.867188, "mean_p": 0.400921, "boundary_entropy": 0.564705, "bpb": 0.85301}, "zh": {"chunks": 3198.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.541483, "cps": 24.984375, "mean_p": 0.210174, "boundary_entropy": 0.089925, "bpb": 2.469435}, "my": {"chunks": 9549.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.503958, "cps": 74.601562, "mean_p": 0.401284, "boundary_entropy": 0.619985, "bpb": 0.871494}}, "bpb": {"en": 1.6435, "de": 1.76716, "vi": 1.45177, "tr": 1.67139, "fi": 1.8326, "ru": 1.06929, "el": 1.09154, "hi": 0.88849, "ta": 0.74745, "th": 0.85301, "zh": 2.46944, "my": 0.87149}, "gini_cpc": 0.219845, "gini_cps": 0.216045, "bpic": 5.690534, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9784620890826556, "vi": 1.15173759954353, "tr": 1.0083384908282234, "fi": 1.0131415039101412, "ru": 1.600976442101189, "el": 1.6648533949911077, "hi": 2.3080210628723017, "ta": 2.3533158923898894, "th": 2.5132355893306078, "zh": 2.8096102505138645, "my": 2.6149029752079223}, "t": 2102.7}
{"event": "eval", "step": 11000, "phase": 3, "bpb": {"en": 1.64083, "de": 1.77289, "vi": 1.46712, "tr": 1.66264, "fi": 1.84663, "ru": 1.06879, "el": 1.09005, "hi": 0.89486, "ta": 0.75259, "th": 0.85097, "zh": 2.47406, "my": 0.87865}, "bytes_per_chunk": {"en": 5.151, "de": 5.266, "vi": 5.877, "tr": 5.571, "fi": 5.263, "ru": 5.841, "el": 5.76, "hi": 5.593, "ta": 5.954, "th": 5.943, "zh": 5.159, "my": 5.596}, "bpb_mean": 1.36667, "t": 2108.4}
{"event": "probe", "step": 11000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3354.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.192725, "cps": 26.203125, "mean_p": 0.189456, "boundary_entropy": 0.068874, "bpb": 1.643497}, "de": {"chunks": 3872.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.188574, "cps": 30.25, "mean_p": 0.18801, "boundary_entropy": 0.07031, "bpb": 1.767157}, "vi": {"chunks": 4092.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.221969, "cps": 31.96875, "mean_p": 0.221989, "boundary_entropy": 0.212375, "bpb": 1.451768}, "tr": {"chunks": 3518.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.194332, "cps": 27.484375, "mean_p": 0.18714, "boundary_entropy": 0.090893, "bpb": 1.67139}, "fi": {"chunks": 3673.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.195258, "cps": 28.695312, "mean_p": 0.191426, "boundary_entropy": 0.081794, "bpb": 1.832604}, "ru": {"chunks": 5847.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.308549, "cps": 45.679688, "mean_p": 0.274957, "boundary_entropy": 0.375384, "bpb": 1.069289}, "el": {"chunks": 6585.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.32086, "cps": 51.445312, "mean_p": 0.253164, "boundary_entropy": 0.328416, "bpb": 1.091542}, "hi": {"chunks": 7484.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.444814, "cps": 58.46875, "mean_p": 0.268467, "boundary_entropy": 0.403481, "bpb": 0.888493}, "ta": {"chunks": 8479.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.453544, "cps": 66.242188, "mean_p": 0.299847, "boundary_entropy": 0.410619, "bpb": 0.74745}, "th": {"chunks": 7791.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.484364, "cps": 60.867188, "mean_p": 0.400921, "boundary_entropy": 0.564705, "bpb": 0.85301}, "zh": {"chunks": 3198.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.541483, "cps": 24.984375, "mean_p": 0.210174, "boundary_entropy": 0.089925, "bpb": 2.469435}, "my": {"chunks": 9549.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.503958, "cps": 74.601562, "mean_p": 0.401284, "boundary_entropy": 0.619985, "bpb": 0.871494}}, "bpb": {"en": 1.6435, "de": 1.76716, "vi": 1.45177, "tr": 1.67139, "fi": 1.8326, "ru": 1.06929, "el": 1.09154, "hi": 0.88849, "ta": 0.74745, "th": 0.85301, "zh": 2.46944, "my": 0.87149}, "gini_cpc": 0.219845, "gini_cps": 0.216045, "bpic": 5.690534, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9784620890826556, "vi": 1.15173759954353, "tr": 1.0083384908282234, "fi": 1.0131415039101412, "ru": 1.600976442101189, "el": 1.6648533949911077, "hi": 2.3080210628723017, "ta": 2.3533158923898894, "th": 2.5132355893306078, "zh": 2.8096102505138645, "my": 2.6149029752079223}, "t": 2111.5}
{"event": "reps_dumped", "dirs": ["/local/scratch/sas245/beetle-hnet/scratch/reps/spec_A_s2/th0.30", "/local/scratch/sas245/beetle-hnet/scratch/reps/spec_A_s2/th0.40", "/local/scratch/sas245/beetle-hnet/scratch/reps/spec_A_s2/th0.50", "/local/scratch/sas245/beetle-hnet/scratch/reps/spec_A_s2/th0.60", "/local/scratch/sas245/beetle-hnet/scratch/reps/spec_A_s2/th0.70"], "t": 2115.6}
{"event": "final", "run": "spec_A_s2", "condition": "A", "step": 11000, "params": 82470528, "per_lang": {"en": {"cpc": 0.19272539217376314, "cps": 26.203125, "bpb": 1.6370897828816098, "bytes_per_chunk": 5.164127062299926}, "de": {"cpc": 0.18857448984561437, "cps": 30.25, "bpb": 1.7686300899569196, "bytes_per_chunk": 5.24277514449711}, "vi": {"cpc": 0.22196908055329537, "cps": 31.96875, "bpb": 1.4657194278770778, "bytes_per_chunk": 5.878755599410205}, "tr": {"cpc": 0.19433243108876982, "cps": 27.484375, "bpb": 1.6620841612674058, "bytes_per_chunk": 5.594046253567713}, "fi": {"cpc": 0.19525809366859817, "cps": 28.6953125, "bpb": 1.8530744099459817, "bytes_per_chunk": 5.268776034208133}, "ru": {"cpc": 0.30854881266490763, "cps": 45.6796875, "bpb": 1.070778833898857, "bytes_per_chunk": 5.86065124805777}, "el": {"cpc": 0.3208595234614822, "cps": 51.4453125, "bpb": 1.08748095053523, "bytes_per_chunk": 5.744394348604955}, "hi": {"cpc": 0.44481426448737, "cps": 58.46875, "bpb": 0.8965628163319439, "bytes_per_chunk": 5.600350363715992}, "ta": {"cpc": 0.4535437282695908, "cps": 66.2421875, "bpb": 0.7514792839599838, "bytes_per_chunk": 5.935425440384006}, "th": {"cpc": 0.48436431457880014, "cps": 60.8671875, "bpb": 0.8530057592481629, "bytes_per_chunk": 5.940031836489602}, "zh": {"cpc": 0.5414832373857095, "cps": 24.984375, "bpb": 2.475187059431537, "bytes_per_chunk": 5.19207554083295}, "my": {"cpc": 0.5039582013932868, "cps": 74.6015625, "bpb": 0.876910052473754, "bytes_per_chunk": 5.586476219905274}}, "gini_cpc": 0.21984466227883126, "gini_cps": 0.2160448978381424, "bpic": 5.690534088550161, "seed": 2, "mean_boundary_entropy": 0.19834211468696594, "dechunk": "ema", "router_noise": 0.0, "transitions": [{"step": 0, "phase": 1, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0}, {"step": 4400, "phase": 2, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0}, {"step": 7700, "phase": 3, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0}], "t": 2126.2}