{"event": "start", "run": "exp32_onechunk_s0", "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": 1.9}
{"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": 1.9}
{"event": "probe", "step": 0, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 8739.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.502155, "cps": 68.273438, "mean_p": 0.50363, "boundary_entropy": 0.687415, "bpb": 8.146538}, "de": {"chunks": 12222.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.595237, "cps": 95.484375, "mean_p": 0.507356, "boundary_entropy": 0.688163, "bpb": 8.115859}, "vi": {"chunks": 6386.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.346406, "cps": 49.890625, "mean_p": 0.491233, "boundary_entropy": 0.688507, "bpb": 8.208555}, "tr": {"chunks": 11403.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.629896, "cps": 89.085938, "mean_p": 0.50659, "boundary_entropy": 0.687851, "bpb": 8.138757}, "fi": {"chunks": 7137.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.379406, "cps": 55.757812, "mean_p": 0.497164, "boundary_entropy": 0.68792, "bpb": 8.132766}, "ru": {"chunks": 10473.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.552665, "cps": 81.820312, "mean_p": 0.496692, "boundary_entropy": 0.690192, "bpb": 8.035314}, "el": {"chunks": 36555.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.781172, "cps": 285.585938, "mean_p": 0.530217, "boundary_entropy": 0.688772, "bpb": 8.179564}, "hi": {"chunks": 42584.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.530996, "cps": 332.6875, "mean_p": 0.521073, "boundary_entropy": 0.69018, "bpb": 7.981719}, "ta": {"chunks": 49758.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.661567, "cps": 388.734375, "mean_p": 0.521091, "boundary_entropy": 0.690409, "bpb": 7.914469}, "th": {"chunks": 45716.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.842151, "cps": 357.15625, "mean_p": 0.525713, "boundary_entropy": 0.689893, "bpb": 8.034608}, "zh": {"chunks": 6218.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.052828, "cps": 48.578125, "mean_p": 0.496822, "boundary_entropy": 0.686552, "bpb": 8.096223}, "my": {"chunks": 53266.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.811167, "cps": 416.140625, "mean_p": 0.521325, "boundary_entropy": 0.690575, "bpb": 8.110618}}, "bpb": {"en": 8.14654, "de": 8.11586, "vi": 8.20855, "tr": 8.13876, "fi": 8.13277, "ru": 8.03531, "el": 8.17956, "hi": 7.98172, "ta": 7.91447, "th": 8.03461, "zh": 8.09622, "my": 8.11062}, "gini_cpc": 0.391191, "gini_cps": 0.409188, "bpic": 1.321301, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1853654183968407, "vi": 0.6898396505624507, "tr": 1.2543852937189204, "fi": 0.755555191784985, "ru": 1.1005867247811874, "el": 3.54705827773647, "hi": 5.040269530092341, "ta": 5.300292378412309, "th": 5.659910185771798, "zh": 2.096619669942964, "my": 5.598208760550946}, "t": 4.9}
{"event": "train", "step": 0, "phase": 1, "objective": "global", "lm_loss": 5.62259, "bpb": 8.11168, "chunk_loss": 0.017982, "F": 0.00098, "G": 0.50248, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.69172, "router_noise": 0.0, "grad_norm": 4.9257, "lr": 6e-07, "router_trainable": true, "sec_per_step": 0.004, "alpha": 0.03, "t": 5.6}
{"event": "probe", "step": 1, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 8774.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.504166, "cps": 68.546875, "mean_p": 0.503735, "boundary_entropy": 0.687415, "bpb": 8.126818}, "de": {"chunks": 12242.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.596211, "cps": 95.640625, "mean_p": 0.507418, "boundary_entropy": 0.688162, "bpb": 8.097041}, "vi": {"chunks": 6430.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.348793, "cps": 50.234375, "mean_p": 0.491322, "boundary_entropy": 0.688511, "bpb": 8.19404}, "tr": {"chunks": 11430.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.631387, "cps": 89.296875, "mean_p": 0.506656, "boundary_entropy": 0.68785, "bpb": 8.124828}, "fi": {"chunks": 7181.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.381745, "cps": 56.101562, "mean_p": 0.497254, "boundary_entropy": 0.687922, "bpb": 8.11735}, "ru": {"chunks": 10564.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.557467, "cps": 82.53125, "mean_p": 0.496782, "boundary_entropy": 0.690194, "bpb": 8.013368}, "el": {"chunks": 36555.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.781172, "cps": 285.585938, "mean_p": 0.530181, "boundary_entropy": 0.688776, "bpb": 8.17156}, "hi": {"chunks": 42590.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.531352, "cps": 332.734375, "mean_p": 0.521147, "boundary_entropy": 0.690174, "bpb": 7.974139}, "ta": {"chunks": 49775.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.662477, "cps": 388.867188, "mean_p": 0.521153, "boundary_entropy": 0.690404, "bpb": 7.905685}, "th": {"chunks": 45719.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.842338, "cps": 357.179688, "mean_p": 0.525818, "boundary_entropy": 0.689883, "bpb": 8.026154}, "zh": {"chunks": 6223.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.053674, "cps": 48.617188, "mean_p": 0.496857, "boundary_entropy": 0.686553, "bpb": 8.094788}, "my": {"chunks": 53267.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.81122, "cps": 416.148438, "mean_p": 0.521321, "boundary_entropy": 0.690575, "bpb": 8.096547}}, "bpb": {"en": 8.12682, "de": 8.09704, "vi": 8.19404, "tr": 8.12483, "fi": 8.11735, "ru": 8.01337, "el": 8.17156, "hi": 7.97414, "ta": 7.90568, "th": 8.02615, "zh": 8.09479, "my": 8.09655}, "gini_cpc": 0.390417, "gini_cps": 0.40849, "bpic": 1.319969, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1825689127625463, "vi": 0.6918219244928661, "tr": 1.2523397601957575, "fi": 0.7571806962439964, "ru": 1.1057212813343333, "el": 3.532908854472192, "hi": 5.020870930626791, "ta": 5.2809528419600005, "th": 5.637702408541812, "zh": 2.0899353336764235, "my": 5.575981857705267}, "t": 7.8}
{"event": "probe", "step": 2, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 8845.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.508246, "cps": 69.101562, "mean_p": 0.503885, "boundary_entropy": 0.687414, "bpb": 8.085989}, "de": {"chunks": 12268.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.597477, "cps": 95.84375, "mean_p": 0.507473, "boundary_entropy": 0.688161, "bpb": 8.058663}, "vi": {"chunks": 6496.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.352373, "cps": 50.75, "mean_p": 0.491494, "boundary_entropy": 0.688518, "bpb": 8.166448}, "tr": {"chunks": 11455.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.632768, "cps": 89.492188, "mean_p": 0.506757, "boundary_entropy": 0.687848, "bpb": 8.097752}, "fi": {"chunks": 7239.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.384828, "cps": 56.554688, "mean_p": 0.497386, "boundary_entropy": 0.687924, "bpb": 8.085952}, "ru": {"chunks": 10739.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.566702, "cps": 83.898438, "mean_p": 0.49694, "boundary_entropy": 0.690197, "bpb": 7.972858}, "el": {"chunks": 36551.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.780977, "cps": 285.554688, "mean_p": 0.530111, "boundary_entropy": 0.688784, "bpb": 8.15568}, "hi": {"chunks": 42602.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.532065, "cps": 332.828125, "mean_p": 0.521285, "boundary_entropy": 0.690163, "bpb": 7.950186}, "ta": {"chunks": 49805.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.664081, "cps": 389.101562, "mean_p": 0.521277, "boundary_entropy": 0.690394, "bpb": 7.881864}, "th": {"chunks": 45725.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.842711, "cps": 357.226562, "mean_p": 0.526042, "boundary_entropy": 0.689861, "bpb": 7.996767}, "zh": {"chunks": 6247.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.057738, "cps": 48.804688, "mean_p": 0.496915, "boundary_entropy": 0.686555, "bpb": 8.090623}, "my": {"chunks": 53265.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.811115, "cps": 416.132812, "mean_p": 0.521262, "boundary_entropy": 0.69058, "bpb": 8.071394}}, "bpb": {"en": 8.08599, "de": 8.05866, "vi": 8.16645, "tr": 8.09775, "fi": 8.08595, "ru": 7.97286, "el": 8.15568, "hi": 7.95019, "ta": 7.88186, "th": 7.99677, "zh": 8.09062, "my": 8.07139}, "gini_cpc": 0.389113, "gini_cps": 0.40729, "bpic": 1.317762, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1755676952571792, "vi": 0.6933127025837347, "tr": 1.2450042244605777, "fi": 0.7571692638975566, "ru": 1.1150155164210362, "el": 3.5041662393861244, "hi": 4.981971033239623, "ta": 5.2417192711962635, "th": 5.593181749091924, "zh": 2.0811546143012722, "my": 5.531015025144536}, "t": 11.0}
{"event": "probe", "step": 4, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 9081.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.521807, "cps": 70.945312, "mean_p": 0.504524, "boundary_entropy": 0.687412, "bpb": 7.948485}, "de": {"chunks": 12405.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.604149, "cps": 96.914062, "mean_p": 0.507799, "boundary_entropy": 0.688156, "bpb": 7.930642}, "vi": {"chunks": 6694.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.363114, "cps": 52.296875, "mean_p": 0.492048, "boundary_entropy": 0.688544, "bpb": 8.075143}, "tr": {"chunks": 11631.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.64249, "cps": 90.867188, "mean_p": 0.507188, "boundary_entropy": 0.68784, "bpb": 8.004648}, "fi": {"chunks": 7517.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.399607, "cps": 58.726562, "mean_p": 0.497932, "boundary_entropy": 0.687934, "bpb": 7.975816}, "ru": {"chunks": 11206.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.591346, "cps": 87.546875, "mean_p": 0.497339, "boundary_entropy": 0.690205, "bpb": 7.841277}, "el": {"chunks": 36501.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.778541, "cps": 285.164062, "mean_p": 0.529551, "boundary_entropy": 0.688848, "bpb": 8.08197}, "hi": {"chunks": 42650.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.534918, "cps": 333.203125, "mean_p": 0.521812, "boundary_entropy": 0.690118, "bpb": 7.868429}, "ta": {"chunks": 49902.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.66927, "cps": 389.859375, "mean_p": 0.521606, "boundary_entropy": 0.690369, "bpb": 7.777745}, "th": {"chunks": 45739.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.843581, "cps": 357.335938, "mean_p": 0.526781, "boundary_entropy": 0.689787, "bpb": 7.882157}, "zh": {"chunks": 6336.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.072807, "cps": 49.5, "mean_p": 0.497157, "boundary_entropy": 0.686563, "bpb": 8.078541}, "my": {"chunks": 53251.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.810376, "cps": 416.023438, "mean_p": 0.521068, "boundary_entropy": 0.690595, "bpb": 8.006698}}, "bpb": {"en": 7.94848, "de": 7.93064, "vi": 8.07514, "tr": 8.00465, "fi": 7.97582, "ru": 7.84128, "el": 8.08197, "hi": 7.86843, "ta": 7.77775, "th": 7.88216, "zh": 8.07854, "my": 8.0067}, "gini_cpc": 0.384613, "gini_cps": 0.403133, "bpic": 1.310222, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.157803361042682, "vi": 0.6958778461525503, "tr": 1.23128035056748, "fi": 0.7658136646495048, "ru": 1.1332659712543136, "el": 3.408429866232895, "hi": 4.857965286173816, "ta": 5.115439196469538, "th": 5.449492316534877, "zh": 2.0559482100982787, "my": 5.385857223068521}, "t": 14.1}
{"event": "probe", "step": 8, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 9950.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.571741, "cps": 77.734375, "mean_p": 0.506814, "boundary_entropy": 0.687391, "bpb": 7.490835}, "de": {"chunks": 12922.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.629328, "cps": 100.953125, "mean_p": 0.508986, "boundary_entropy": 0.68814, "bpb": 7.50804}, "vi": {"chunks": 7388.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.400759, "cps": 57.71875, "mean_p": 0.493751, "boundary_entropy": 0.688616, "bpb": 7.758197}, "tr": {"chunks": 12144.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.670828, "cps": 94.875, "mean_p": 0.508621, "boundary_entropy": 0.687807, "bpb": 7.681711}, "fi": {"chunks": 8367.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.444793, "cps": 65.367188, "mean_p": 0.499797, "boundary_entropy": 0.687957, "bpb": 7.596706}, "ru": {"chunks": 13504.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.712612, "cps": 105.5, "mean_p": 0.499282, "boundary_entropy": 0.690236, "bpb": 7.378401}, "el": {"chunks": 36299.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.768699, "cps": 283.585938, "mean_p": 0.527707, "boundary_entropy": 0.68905, "bpb": 7.837103}, "hi": {"chunks": 42776.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.542407, "cps": 334.1875, "mean_p": 0.523335, "boundary_entropy": 0.689984, "bpb": 7.593645}, "ta": {"chunks": 50098.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.679754, "cps": 391.390625, "mean_p": 0.522358, "boundary_entropy": 0.690311, "bpb": 7.465029}, "th": {"chunks": 45782.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.846254, "cps": 357.671875, "mean_p": 0.528713, "boundary_entropy": 0.689583, "bpb": 7.594047}, "zh": {"chunks": 6569.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.112259, "cps": 51.320312, "mean_p": 0.498029, "boundary_entropy": 0.686591, "bpb": 8.048141}, "my": {"chunks": 53243.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.809954, "cps": 415.960938, "mean_p": 0.520646, "boundary_entropy": 0.690628, "bpb": 7.826596}}, "bpb": {"en": 7.49083, "de": 7.50804, "vi": 7.7582, "tr": 7.68171, "fi": 7.59671, "ru": 7.3784, "el": 7.8371, "hi": 7.59365, "ta": 7.46503, "th": 7.59405, "zh": 8.04814, "my": 7.8266}, "gini_cpc": 0.369422, "gini_cps": 0.389267, "bpic": 1.283368, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1007238305196658, "vi": 0.700946359134975, "tr": 1.1733085801320584, "fi": 0.7779629688498844, "ru": 1.2463908566579598, "el": 3.093533726225044, "hi": 4.446785057531342, "ta": 4.687010844540972, "th": 4.9782274505883475, "zh": 1.9453908043434238, "my": 4.914735854700991}, "t": 18.0}
{"event": "probe", "step": 16, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 12565.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.722002, "cps": 98.164062, "mean_p": 0.513573, "boundary_entropy": 0.687237, "bpb": 6.577917}, "de": {"chunks": 15265.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.743437, "cps": 119.257812, "mean_p": 0.513495, "boundary_entropy": 0.688077, "bpb": 6.667676}, "vi": {"chunks": 9034.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.490046, "cps": 70.578125, "mean_p": 0.497444, "boundary_entropy": 0.688774, "bpb": 7.129815}, "tr": {"chunks": 13080.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.722532, "cps": 102.1875, "mean_p": 0.511309, "boundary_entropy": 0.687748, "bpb": 7.00394}, "fi": {"chunks": 10652.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.566264, "cps": 83.21875, "mean_p": 0.504795, "boundary_entropy": 0.687975, "bpb": 6.798337}, "ru": {"chunks": 23740.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.25277, "cps": 185.46875, "mean_p": 0.507194, "boundary_entropy": 0.690218, "bpb": 6.536978}, "el": {"chunks": 34731.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.692296, "cps": 271.335938, "mean_p": 0.520557, "boundary_entropy": 0.689713, "bpb": 7.364904}, "hi": {"chunks": 42927.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.551382, "cps": 335.367188, "mean_p": 0.527488, "boundary_entropy": 0.689565, "bpb": 6.822321}, "ta": {"chunks": 50312.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.691201, "cps": 393.0625, "mean_p": 0.522826, "boundary_entropy": 0.690298, "bpb": 6.749429}, "th": {"chunks": 45878.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.852223, "cps": 358.421875, "mean_p": 0.533457, "boundary_entropy": 0.689024, "bpb": 6.677733}, "zh": {"chunks": 7454.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.262106, "cps": 58.234375, "mean_p": 0.5011, "boundary_entropy": 0.68667, "bpb": 8.011894}, "my": {"chunks": 53112.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.80304, "cps": 414.9375, "mean_p": 0.518926, "boundary_entropy": 0.690754, "bpb": 7.388044}}, "bpb": {"en": 6.57792, "de": 6.66768, "vi": 7.12981, "tr": 7.00394, "fi": 6.79834, "ru": 6.53698, "el": 7.3649, "hi": 6.82232, "ta": 6.74943, "th": 6.67773, "zh": 8.01189, "my": 7.38804}, "gini_cpc": 0.326531, "gini_cps": 0.35178, "bpic": 1.204019, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0296888944255564, "vi": 0.6787323849262922, "tr": 1.00073437931627, "fi": 0.784297628092782, "ru": 1.7351344302502976, "el": 2.343894554921629, "hi": 3.533760343982465, "ta": 3.7274149219458654, "th": 3.9504360799132052, "zh": 1.7480649825108556, "my": 3.8823162241587568}, "t": 24.0}
{"event": "probe", "step": 32, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 17205.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.988623, "cps": 134.414062, "mean_p": 0.554966, "boundary_entropy": 0.682262, "bpb": 5.890449}, "de": {"chunks": 20618.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.00414, "cps": 161.078125, "mean_p": 0.555931, "boundary_entropy": 0.682825, "bpb": 6.037967}, "vi": {"chunks": 22206.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.204557, "cps": 173.484375, "mean_p": 0.527085, "boundary_entropy": 0.68764, "bpb": 6.707147}, "tr": {"chunks": 19289.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.065514, "cps": 150.695312, "mean_p": 0.548719, "boundary_entropy": 0.6838, "bpb": 6.427517}, "fi": {"chunks": 19109.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 1.015842, "cps": 149.289062, "mean_p": 0.54743, "boundary_entropy": 0.684088, "bpb": 6.084414}, "ru": {"chunks": 33656.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.776042, "cps": 262.9375, "mean_p": 0.523088, "boundary_entropy": 0.689503, "bpb": 5.641161}, "el": {"chunks": 4642.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.226185, "cps": 36.265625, "mean_p": 0.48912, "boundary_entropy": 0.69017, "bpb": 6.898715}, "hi": {"chunks": 43140.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.564042, "cps": 337.03125, "mean_p": 0.545826, "boundary_entropy": 0.686955, "bpb": 5.81934}, "ta": {"chunks": 50835.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.719176, "cps": 397.148438, "mean_p": 0.530479, "boundary_entropy": 0.689602, "bpb": 5.996073}, "th": {"chunks": 46043.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.862481, "cps": 359.710938, "mean_p": 0.540703, "boundary_entropy": 0.688018, "bpb": 5.415877}, "zh": {"chunks": 12248.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.073823, "cps": 95.6875, "mean_p": 0.519225, "boundary_entropy": 0.686165, "bpb": 7.732909}, "my": {"chunks": 53362.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.816234, "cps": 416.890625, "mean_p": 0.522498, "boundary_entropy": 0.690469, "bpb": 6.192296}}, "bpb": {"en": 5.89045, "de": 6.03797, "vi": 6.70715, "tr": 6.42752, "fi": 6.08441, "ru": 5.64116, "el": 6.89872, "hi": 5.81934, "ta": 5.99607, "th": 5.41588, "zh": 7.73291, "my": 6.1923}, "gini_cpc": 0.282159, "gini_cps": 0.304246, "bpic": 1.12101, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0156956006472835, "vi": 1.218418927076897, "tr": 1.0777762393349186, "fi": 1.0275323890188905, "ru": 1.7964814118740409, "el": 0.22878825937990085, "hi": 2.593549319822627, "ta": 2.750469298725258, "th": 2.895422806965719, "zh": 2.097689374156171, "my": 2.848643918592232}, "t": 32.8}
{"event": "probe", "step": 64, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 17390.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.999253, "cps": 135.859375, "mean_p": 0.773696, "boundary_entropy": 0.529591, "bpb": 5.453822}, "de": {"chunks": 20829.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.014416, "cps": 162.726562, "mean_p": 0.775857, "boundary_entropy": 0.527544, "bpb": 5.595157}, "vi": {"chunks": 24127.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.308761, "cps": 188.492188, "mean_p": 0.692516, "boundary_entropy": 0.609379, "bpb": 6.28321}, "tr": {"chunks": 19680.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.087113, "cps": 153.75, "mean_p": 0.763461, "boundary_entropy": 0.541154, "bpb": 6.028739}, "fi": {"chunks": 19512.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 1.037265, "cps": 152.4375, "mean_p": 0.764378, "boundary_entropy": 0.540354, "bpb": 5.677978}, "ru": {"chunks": 34557.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.823588, "cps": 269.976562, "mean_p": 0.667397, "boundary_entropy": 0.634199, "bpb": 4.962447}, "el": {"chunks": 30720.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.496857, "cps": 240.0, "mean_p": 0.509178, "boundary_entropy": 0.690467, "bpb": 5.405992}, "hi": {"chunks": 43166.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.565587, "cps": 337.234375, "mean_p": 0.583731, "boundary_entropy": 0.677375, "bpb": 4.925665}, "ta": {"chunks": 50888.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.722011, "cps": 397.5625, "mean_p": 0.579182, "boundary_entropy": 0.679108, "bpb": 4.826906}, "th": {"chunks": 46084.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.86503, "cps": 360.03125, "mean_p": 0.571007, "boundary_entropy": 0.681251, "bpb": 4.686049}, "zh": {"chunks": 16101.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.726211, "cps": 125.789062, "mean_p": 0.619502, "boundary_entropy": 0.657893, "bpb": 6.963934}, "my": {"chunks": 53701.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.834125, "cps": 419.539062, "mean_p": 0.59377, "boundary_entropy": 0.673956, "bpb": 4.841224}}, "bpb": {"en": 5.45382, "de": 5.59516, "vi": 6.28321, "tr": 6.02874, "fi": 5.67798, "ru": 4.96245, "el": 5.40599, "hi": 4.92567, "ta": 4.82691, "th": 4.68605, "zh": 6.96393, "my": 4.84122}, "gini_cpc": 0.226107, "gini_cps": 0.23585, "bpic": 1.018649, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0151741511378463, "vi": 1.3097388817586868, "tr": 1.0879253109093936, "fi": 1.0380408438665434, "ru": 1.8249516250658113, "el": 1.4979761692914866, "hi": 2.5675048442897705, "ta": 2.7240460889608173, "th": 2.867171300825492, "zh": 2.7282486285523224, "my": 2.8362436409298146}, "t": 47.9}
{"event": "train", "step": 100, "phase": 1, "objective": "global", "lm_loss": 3.36017, "bpb": 4.8477, "chunk_loss": 0.007772, "F": 0.00098, "G": 0.78774, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.4892, "router_noise": 0.0, "grad_norm": 3.4195, "lr": 6.06e-05, "router_trainable": true, "sec_per_step": 0.569, "alpha": 0.03, "t": 62.5}
{"event": "probe", "step": 128, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 17418.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 1.000862, "cps": 136.078125, "mean_p": 0.833217, "boundary_entropy": 0.442856, "bpb": 4.351103}, "de": {"chunks": 20803.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.01315, "cps": 162.523438, "mean_p": 0.828809, "boundary_entropy": 0.444123, "bpb": 4.575049}, "vi": {"chunks": 23176.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.257174, "cps": 181.0625, "mean_p": 0.669728, "boundary_entropy": 0.617515, "bpb": 4.786536}, "tr": {"chunks": 19434.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.073524, "cps": 151.828125, "mean_p": 0.754799, "boundary_entropy": 0.534947, "bpb": 4.823215}, "fi": {"chunks": 19455.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 1.034235, "cps": 151.992188, "mean_p": 0.811302, "boundary_entropy": 0.462275, "bpb": 4.581995}, "ru": {"chunks": 34560.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.823747, "cps": 270.0, "mean_p": 0.75783, "boundary_entropy": 0.549943, "bpb": 3.221174}, "el": {"chunks": 37261.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.815573, "cps": 291.101562, "mean_p": 0.783575, "boundary_entropy": 0.518167, "bpb": 3.933011}, "hi": {"chunks": 43170.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.565825, "cps": 337.265625, "mean_p": 0.710386, "boundary_entropy": 0.595366, "bpb": 2.857542}, "ta": {"chunks": 50913.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.723348, "cps": 397.757812, "mean_p": 0.719093, "boundary_entropy": 0.59148, "bpb": 3.021679}, "th": {"chunks": 46085.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.865092, "cps": 360.039062, "mean_p": 0.708003, "boundary_entropy": 0.597982, "bpb": 2.78521}, "zh": {"chunks": 16171.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.738063, "cps": 126.335938, "mean_p": 0.860917, "boundary_entropy": 0.395727, "bpb": 6.184467}, "my": {"chunks": 53738.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.836078, "cps": 419.828125, "mean_p": 0.716927, "boundary_entropy": 0.591795, "bpb": 2.882327}}, "bpb": {"en": 4.3511, "de": 4.57505, "vi": 4.78654, "tr": 4.82322, "fi": 4.582, "ru": 3.22117, "el": 3.93301, "hi": 2.85754, "ta": 3.02168, "th": 2.78521, "zh": 6.18447, "my": 2.88233}, "gini_cpc": 0.22356, "gini_cps": 0.235616, "bpic": 1.004179, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0122770619081398, "vi": 1.256091203501405, "tr": 1.0725992302362743, "fi": 1.0333446270919862, "ru": 1.8221761311426539, "el": 1.8140092405168735, "hi": 2.563615033687035, "ta": 2.7210032008462623, "th": 2.8626243461390897, "zh": 2.7357050269353644, "my": 2.833635318329746}, "t": 74.7}
{"event": "train", "step": 200, "phase": 1, "objective": "global", "lm_loss": 2.50684, "bpb": 3.61661, "chunk_loss": 0.004028, "F": 0.00098, "G": 0.89237, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.30161, "router_noise": 0.0, "grad_norm": 2.7439, "lr": 0.0001206, "router_trainable": true, "sec_per_step": 0.405, "alpha": 0.03, "t": 103.0}
{"event": "probe", "step": 256, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 17405.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 1.000115, "cps": 135.976562, "mean_p": 0.952581, "boundary_entropy": 0.176763, "bpb": 3.578366}, "de": {"chunks": 20494.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.998101, "cps": 160.109375, "mean_p": 0.932632, "boundary_entropy": 0.21074, "bpb": 3.78675}, "vi": {"chunks": 23889.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.29585, "cps": 186.632812, "mean_p": 0.848559, "boundary_entropy": 0.379028, "bpb": 3.143951}, "tr": {"chunks": 18543.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.024305, "cps": 144.867188, "mean_p": 0.884946, "boundary_entropy": 0.266575, "bpb": 3.756968}, "fi": {"chunks": 19000.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 1.010047, "cps": 148.4375, "mean_p": 0.920213, "boundary_entropy": 0.221611, "bpb": 3.823281}, "ru": {"chunks": 34533.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.822322, "cps": 269.789062, "mean_p": 0.976165, "boundary_entropy": 0.102194, "bpb": 2.627664}, "el": {"chunks": 37185.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.81187, "cps": 290.507812, "mean_p": 0.94914, "boundary_entropy": 0.192551, "bpb": 2.725906}, "hi": {"chunks": 43170.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.565825, "cps": 337.265625, "mean_p": 0.957896, "boundary_entropy": 0.171716, "bpb": 2.11655}, "ta": {"chunks": 50909.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.723135, "cps": 397.726562, "mean_p": 0.951026, "boundary_entropy": 0.191965, "bpb": 1.988545}, "th": {"chunks": 46025.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.861362, "cps": 359.570312, "mean_p": 0.963276, "boundary_entropy": 0.152088, "bpb": 2.100557}, "zh": {"chunks": 16142.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.733153, "cps": 126.109375, "mean_p": 0.943998, "boundary_entropy": 0.203481, "bpb": 4.499056}, "my": {"chunks": 53740.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.836183, "cps": 419.84375, "mean_p": 0.939422, "boundary_entropy": 0.222832, "bpb": 2.221032}}, "bpb": {"en": 3.57837, "de": 3.78675, "vi": 3.14395, "tr": 3.75697, "fi": 3.82328, "ru": 2.62766, "el": 2.72591, "hi": 2.11655, "ta": 1.98854, "th": 2.10056, "zh": 4.49906, "my": 2.22103}, "gini_cpc": 0.225568, "gini_cps": 0.238179, "bpic": 1.007207, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9979859272647656, "vi": 1.295701379264724, "tr": 1.0241876613256935, "fi": 1.0099312487020107, "ru": 1.822112497621843, "el": 1.8116614086593914, "hi": 2.5655298280241756, "ta": 2.72282161389775, "th": 2.8610327192949474, "zh": 2.7328386608041932, "my": 2.835857333913824}, "t": 125.1}
{"event": "train", "step": 300, "phase": 1, "objective": "global", "lm_loss": 2.13817, "bpb": 3.08472, "chunk_loss": 0.001492, "F": 0.00098, "G": 0.96322, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.13572, "router_noise": 0.0, "grad_norm": 1.2813, "lr": 0.0001806, "router_trainable": true, "sec_per_step": 0.376, "alpha": 0.03, "t": 140.7}
{"event": "train", "step": 400, "phase": 1, "objective": "global", "lm_loss": 2.05218, "bpb": 2.96067, "chunk_loss": 0.000865, "F": 0.00098, "G": 0.98073, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.08813, "router_noise": 0.0, "grad_norm": 1.5631, "lr": 0.0002406, "router_trainable": true, "sec_per_step": 0.343, "alpha": 0.03, "t": 175.0}
{"event": "train", "step": 500, "phase": 1, "objective": "global", "lm_loss": 1.96453, "bpb": 2.83422, "chunk_loss": 0.000421, "F": 0.00098, "G": 0.99314, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.03971, "router_noise": 0.0, "grad_norm": 1.1428, "lr": 0.0003, "router_trainable": true, "sec_per_step": 0.344, "alpha": 0.03, "t": 209.4}
{"event": "train", "step": 600, "phase": 1, "objective": "global", "lm_loss": 1.85193, "bpb": 2.67177, "chunk_loss": 0.000267, "F": 0.00098, "G": 0.99746, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.01748, "router_noise": 0.0, "grad_norm": 1.008, "lr": 0.0002999, "router_trainable": true, "sec_per_step": 0.323, "alpha": 0.03, "t": 241.7}
{"event": "train", "step": 700, "phase": 1, "objective": "global", "lm_loss": 1.75853, "bpb": 2.53703, "chunk_loss": 0.00022, "F": 0.00098, "G": 0.99876, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00943, "router_noise": 0.0, "grad_norm": 0.9651, "lr": 0.0002998, "router_trainable": true, "sec_per_step": 0.306, "alpha": 0.03, "t": 272.3}
{"event": "train", "step": 800, "phase": 1, "objective": "global", "lm_loss": 1.63328, "bpb": 2.35633, "chunk_loss": 0.000203, "F": 0.00098, "G": 0.99924, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00614, "router_noise": 0.0, "grad_norm": 0.7537, "lr": 0.0002995, "router_trainable": true, "sec_per_step": 0.291, "alpha": 0.03, "t": 301.4}
{"event": "train", "step": 900, "phase": 1, "objective": "global", "lm_loss": 1.61557, "bpb": 2.33077, "chunk_loss": 0.000194, "F": 0.00098, "G": 0.99948, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00441, "router_noise": 0.0, "grad_norm": 0.8107, "lr": 0.000299, "router_trainable": true, "sec_per_step": 0.276, "alpha": 0.03, "t": 328.9}
{"event": "probe", "step": 1000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 17418.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 1.000862, "cps": 136.078125, "mean_p": 0.999586, "boundary_entropy": 0.003602, "bpb": 2.406085}, "de": {"chunks": 20862.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.016023, "cps": 162.984375, "mean_p": 0.99964, "boundary_entropy": 0.003192, "bpb": 2.690167}, "vi": {"chunks": 24274.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.316734, "cps": 189.640625, "mean_p": 0.999739, "boundary_entropy": 0.002395, "bpb": 2.133567}, "tr": {"chunks": 19742.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.090537, "cps": 154.234375, "mean_p": 0.999627, "boundary_entropy": 0.003297, "bpb": 2.687335}, "fi": {"chunks": 19572.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 1.040455, "cps": 152.90625, "mean_p": 0.999675, "boundary_entropy": 0.002915, "bpb": 2.940853}, "ru": {"chunks": 34560.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.823747, "cps": 270.0, "mean_p": 0.999605, "boundary_entropy": 0.003456, "bpb": 1.738286}, "el": {"chunks": 37264.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.815719, "cps": 291.125, "mean_p": 0.999576, "boundary_entropy": 0.003689, "bpb": 1.8231}, "hi": {"chunks": 43173.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.566003, "cps": 337.289062, "mean_p": 0.999723, "boundary_entropy": 0.002527, "bpb": 1.466526}, "ta": {"chunks": 50915.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.723455, "cps": 397.773438, "mean_p": 0.999693, "boundary_entropy": 0.00277, "bpb": 1.163327}, "th": {"chunks": 46085.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.865092, "cps": 360.039062, "mean_p": 0.999701, "boundary_entropy": 0.002707, "bpb": 1.390091}, "zh": {"chunks": 16173.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.738402, "cps": 126.351562, "mean_p": 0.999618, "boundary_entropy": 0.003355, "bpb": 3.51543}, "my": {"chunks": 53743.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.836342, "cps": 419.867188, "mean_p": 0.999737, "boundary_entropy": 0.002417, "bpb": 1.302016}}, "bpb": {"en": 2.40608, "de": 2.69017, "vi": 2.13357, "tr": 2.68734, "fi": 2.94085, "ru": 1.73829, "el": 1.8231, "hi": 1.46653, "ta": 1.16333, "th": 1.39009, "zh": 3.51543, "my": 1.30202}, "gini_cpc": 0.221516, "gini_cps": 0.233853, "bpic": 1.0, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.01514801064883, "vi": 1.3156005295906583, "tr": 1.0895983329898389, "fi": 1.0395590357977051, "ru": 1.8221761311426539, "el": 1.8141552920914839, "hi": 2.563793186225859, "ta": 2.721110089193083, "th": 2.8626243461390897, "zh": 2.7360433739796948, "my": 2.8338989711748765}, "t": 356.2}
{"event": "eval", "step": 1000, "phase": 1, "bpb": {"en": 2.41892, "de": 2.6802, "vi": 2.13773, "tr": 2.68049, "fi": 2.94911, "ru": 1.73594, "el": 1.81923, "hi": 1.47597, "ta": 1.16192, "th": 1.38484, "zh": 3.52117, "my": 1.31459}, "bytes_per_chunk": {"en": 1024.0, "de": 1024.0, "vi": 1024.0, "tr": 1024.0, "fi": 1024.0, "ru": 1024.0, "el": 1024.0, "hi": 1024.0, "ta": 1024.0, "th": 1024.0, "zh": 1024.0, "my": 1024.0}, "bpb_mean": 2.10668, "t": 360.2}
{"event": "train", "step": 1000, "phase": 1, "objective": "global", "lm_loss": 1.54155, "bpb": 2.22399, "chunk_loss": 0.00019, "F": 0.00098, "G": 0.99962, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00337, "router_noise": 0.0, "grad_norm": 0.7647, "lr": 0.0002985, "router_trainable": true, "sec_per_step": 0.319, "alpha": 0.03, "t": 360.8}
{"event": "train", "step": 1100, "phase": 1, "objective": "global", "lm_loss": 1.46369, "bpb": 2.11166, "chunk_loss": 0.000186, "F": 0.00098, "G": 0.99972, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00255, "router_noise": 0.0, "grad_norm": 0.6818, "lr": 0.0002978, "router_trainable": true, "sec_per_step": 0.244, "alpha": 0.03, "t": 385.2}
{"event": "train", "step": 1200, "phase": 1, "objective": "global", "lm_loss": 1.49943, "bpb": 2.16323, "chunk_loss": 0.000184, "F": 0.00098, "G": 0.99976, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00218, "router_noise": 0.0, "grad_norm": 0.6489, "lr": 0.000297, "router_trainable": true, "sec_per_step": 0.224, "alpha": 0.03, "t": 407.6}
{"event": "train", "step": 1300, "phase": 1, "objective": "global", "lm_loss": 1.5035, "bpb": 2.16909, "chunk_loss": 0.000183, "F": 0.00098, "G": 0.99981, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00182, "router_noise": 0.0, "grad_norm": 0.747, "lr": 0.0002962, "router_trainable": true, "sec_per_step": 0.231, "alpha": 0.03, "t": 430.7}
{"event": "train", "step": 1400, "phase": 1, "objective": "global", "lm_loss": 1.47571, "bpb": 2.129, "chunk_loss": 0.000181, "F": 0.00098, "G": 0.99985, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00148, "router_noise": 0.0, "grad_norm": 0.6458, "lr": 0.0002951, "router_trainable": true, "sec_per_step": 0.251, "alpha": 0.03, "t": 455.9}
{"event": "train", "step": 1500, "phase": 1, "objective": "global", "lm_loss": 1.47731, "bpb": 2.1313, "chunk_loss": 0.00018, "F": 0.00098, "G": 0.99987, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00127, "router_noise": 0.0, "grad_norm": 0.6831, "lr": 0.000294, "router_trainable": true, "sec_per_step": 0.228, "alpha": 0.03, "t": 478.7}
{"event": "train", "step": 1600, "phase": 1, "objective": "global", "lm_loss": 1.44698, "bpb": 2.08755, "chunk_loss": 0.00018, "F": 0.00098, "G": 0.99988, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00116, "router_noise": 0.0, "grad_norm": 0.5677, "lr": 0.0002928, "router_trainable": true, "sec_per_step": 0.245, "alpha": 0.03, "t": 503.2}
{"event": "train", "step": 1700, "phase": 1, "objective": "global", "lm_loss": 1.39481, "bpb": 2.01228, "chunk_loss": 0.00018, "F": 0.00098, "G": 0.99989, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.0011, "router_noise": 0.0, "grad_norm": 0.6007, "lr": 0.0002914, "router_trainable": true, "sec_per_step": 0.225, "alpha": 0.03, "t": 525.7}
{"event": "train", "step": 1800, "phase": 1, "objective": "global", "lm_loss": 1.4038, "bpb": 2.02525, "chunk_loss": 0.00018, "F": 0.00098, "G": 0.99989, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00107, "router_noise": 0.0, "grad_norm": 0.564, "lr": 0.0002899, "router_trainable": true, "sec_per_step": 0.246, "alpha": 0.03, "t": 550.3}
{"event": "train", "step": 1900, "phase": 1, "objective": "global", "lm_loss": 1.40503, "bpb": 2.02703, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00105, "router_noise": 0.0, "grad_norm": 0.6145, "lr": 0.0002883, "router_trainable": true, "sec_per_step": 0.22, "alpha": 0.03, "t": 572.2}
{"event": "probe", "step": 2000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 17418.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 1.000862, "cps": 136.078125, "mean_p": 0.999898, "boundary_entropy": 0.001035, "bpb": 2.138147}, "de": {"chunks": 20862.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.016023, "cps": 162.984375, "mean_p": 0.999899, "boundary_entropy": 0.001026, "bpb": 2.347212}, "vi": {"chunks": 24274.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.316734, "cps": 189.640625, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 1.994616}, "tr": {"chunks": 19742.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.090537, "cps": 154.234375, "mean_p": 0.999898, "boundary_entropy": 0.001034, "bpb": 2.329026}, "fi": {"chunks": 19572.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 1.040455, "cps": 152.90625, "mean_p": 0.999899, "boundary_entropy": 0.001026, "bpb": 2.541542}, "ru": {"chunks": 34560.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.823747, "cps": 270.0, "mean_p": 0.999898, "boundary_entropy": 0.001042, "bpb": 1.469066}, "el": {"chunks": 37264.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.815719, "cps": 291.125, "mean_p": 0.999898, "boundary_entropy": 0.001038, "bpb": 1.538076}, "hi": {"chunks": 43173.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.566003, "cps": 337.289062, "mean_p": 0.9999, "boundary_entropy": 0.001021, "bpb": 1.250971}, "ta": {"chunks": 50915.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.723455, "cps": 397.773438, "mean_p": 0.9999, "boundary_entropy": 0.00102, "bpb": 1.067142}, "th": {"chunks": 46085.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.865092, "cps": 360.039062, "mean_p": 0.999899, "boundary_entropy": 0.001033, "bpb": 1.184535}, "zh": {"chunks": 16173.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.738402, "cps": 126.351562, "mean_p": 0.9999, "boundary_entropy": 0.001023, "bpb": 3.233334}, "my": {"chunks": 53743.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.836342, "cps": 419.867188, "mean_p": 0.9999, "boundary_entropy": 0.001022, "bpb": 1.155738}}, "bpb": {"en": 2.13815, "de": 2.34721, "vi": 1.99462, "tr": 2.32903, "fi": 2.54154, "ru": 1.46907, "el": 1.53808, "hi": 1.25097, "ta": 1.06714, "th": 1.18453, "zh": 3.23333, "my": 1.15574}, "gini_cpc": 0.221516, "gini_cps": 0.233853, "bpic": 1.0, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.01514801064883, "vi": 1.3156005295906583, "tr": 1.0895983329898389, "fi": 1.0395590357977051, "ru": 1.8221761311426539, "el": 1.8141552920914839, "hi": 2.563793186225859, "ta": 2.721110089193083, "th": 2.8626243461390897, "zh": 2.7360433739796948, "my": 2.8338989711748765}, "t": 597.9}
{"event": "eval", "step": 2000, "phase": 1, "bpb": {"en": 2.14805, "de": 2.33817, "vi": 1.99855, "tr": 2.32161, "fi": 2.55603, "ru": 1.46896, "el": 1.5333, "hi": 1.25864, "ta": 1.06645, "th": 1.18219, "zh": 3.2399, "my": 1.16643}, "bytes_per_chunk": {"en": 1024.0, "de": 1024.0, "vi": 1024.0, "tr": 1024.0, "fi": 1024.0, "ru": 1024.0, "el": 1024.0, "hi": 1024.0, "ta": 1024.0, "th": 1024.0, "zh": 1024.0, "my": 1024.0}, "bpb_mean": 1.85652, "t": 601.6}
{"event": "train", "step": 2000, "phase": 1, "objective": "global", "lm_loss": 1.39676, "bpb": 2.0151, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00104, "router_noise": 0.0, "grad_norm": 0.5631, "lr": 0.0002866, "router_trainable": true, "sec_per_step": 0.296, "alpha": 0.03, "t": 601.8}
{"event": "train", "step": 2100, "phase": 1, "objective": "global", "lm_loss": 1.33326, "bpb": 1.92349, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00105, "router_noise": 0.0, "grad_norm": 0.5423, "lr": 0.0002848, "router_trainable": true, "sec_per_step": 0.226, "alpha": 0.03, "t": 624.5}
{"event": "train", "step": 2200, "phase": 1, "objective": "global", "lm_loss": 1.29515, "bpb": 1.86851, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00103, "router_noise": 0.0, "grad_norm": 0.5364, "lr": 0.0002829, "router_trainable": true, "sec_per_step": 0.24, "alpha": 0.03, "t": 648.5}
{"event": "train", "step": 2300, "phase": 1, "objective": "global", "lm_loss": 1.36954, "bpb": 1.97583, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00103, "router_noise": 0.0, "grad_norm": 0.552, "lr": 0.0002809, "router_trainable": true, "sec_per_step": 0.222, "alpha": 0.03, "t": 670.6}
{"event": "train", "step": 2400, "phase": 1, "objective": "global", "lm_loss": 1.3862, "bpb": 1.99986, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.5307, "lr": 0.0002788, "router_trainable": true, "sec_per_step": 0.242, "alpha": 0.03, "t": 694.8}
{"event": "train", "step": 2500, "phase": 1, "objective": "global", "lm_loss": 1.29968, "bpb": 1.87504, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00103, "router_noise": 0.0, "grad_norm": 0.502, "lr": 0.0002765, "router_trainable": true, "sec_per_step": 0.225, "alpha": 0.03, "t": 717.3}
{"event": "train", "step": 2600, "phase": 1, "objective": "global", "lm_loss": 1.3104, "bpb": 1.89051, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4999, "lr": 0.0002742, "router_trainable": true, "sec_per_step": 0.226, "alpha": 0.03, "t": 739.9}
{"event": "train", "step": 2700, "phase": 1, "objective": "global", "lm_loss": 1.31034, "bpb": 1.89042, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00103, "router_noise": 0.0, "grad_norm": 0.482, "lr": 0.0002718, "router_trainable": true, "sec_per_step": 0.219, "alpha": 0.03, "t": 761.8}
{"event": "train", "step": 2800, "phase": 1, "objective": "global", "lm_loss": 1.40395, "bpb": 2.02547, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.5126, "lr": 0.0002693, "router_trainable": true, "sec_per_step": 0.202, "alpha": 0.03, "t": 782.0}
{"event": "train", "step": 2900, "phase": 1, "objective": "global", "lm_loss": 1.31479, "bpb": 1.89685, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.5125, "lr": 0.0002667, "router_trainable": true, "sec_per_step": 0.221, "alpha": 0.03, "t": 804.1}
{"event": "probe", "step": 3000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 17418.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 1.000862, "cps": 136.078125, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 2.031091}, "de": {"chunks": 20862.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.016023, "cps": 162.984375, "mean_p": 0.999901, "boundary_entropy": 0.001016, "bpb": 2.209211}, "vi": {"chunks": 24274.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.316734, "cps": 189.640625, "mean_p": 0.9999, "boundary_entropy": 0.001016, "bpb": 1.892144}, "tr": {"chunks": 19742.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.090537, "cps": 154.234375, "mean_p": 0.9999, "boundary_entropy": 0.001017, "bpb": 2.17061}, "fi": {"chunks": 19572.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 1.040455, "cps": 152.90625, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 2.366681}, "ru": {"chunks": 34560.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.823747, "cps": 270.0, "mean_p": 0.999901, "boundary_entropy": 0.001018, "bpb": 1.362122}, "el": {"chunks": 37264.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.815719, "cps": 291.125, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 1.415719}, "hi": {"chunks": 43173.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.566003, "cps": 337.289062, "mean_p": 0.9999, "boundary_entropy": 0.001021, "bpb": 1.16251}, "ta": {"chunks": 50915.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.723455, "cps": 397.773438, "mean_p": 0.9999, "boundary_entropy": 0.00102, "bpb": 0.971519}, "th": {"chunks": 46085.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.865092, "cps": 360.039062, "mean_p": 0.9999, "boundary_entropy": 0.001022, "bpb": 1.101681}, "zh": {"chunks": 16173.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.738402, "cps": 126.351562, "mean_p": 0.9999, "boundary_entropy": 0.001017, "bpb": 3.04217}, "my": {"chunks": 53743.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.836342, "cps": 419.867188, "mean_p": 0.999901, "boundary_entropy": 0.001019, "bpb": 1.085045}}, "bpb": {"en": 2.03109, "de": 2.20921, "vi": 1.89214, "tr": 2.17061, "fi": 2.36668, "ru": 1.36212, "el": 1.41572, "hi": 1.16251, "ta": 0.97152, "th": 1.10168, "zh": 3.04217, "my": 1.08505}, "gini_cpc": 0.221516, "gini_cps": 0.233853, "bpic": 1.0, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.01514801064883, "vi": 1.3156005295906583, "tr": 1.0895983329898389, "fi": 1.0395590357977051, "ru": 1.8221761311426539, "el": 1.8141552920914839, "hi": 2.563793186225859, "ta": 2.721110089193083, "th": 2.8626243461390897, "zh": 2.7360433739796948, "my": 2.8338989711748765}, "t": 828.0}
{"event": "eval", "step": 3000, "phase": 1, "bpb": {"en": 2.04148, "de": 2.20064, "vi": 1.89633, "tr": 2.1616, "fi": 2.378, "ru": 1.36155, "el": 1.41105, "hi": 1.17092, "ta": 0.97126, "th": 1.09783, "zh": 3.04914, "my": 1.09568}, "bytes_per_chunk": {"en": 1024.0, "de": 1024.0, "vi": 1024.0, "tr": 1024.0, "fi": 1024.0, "ru": 1024.0, "el": 1024.0, "hi": 1024.0, "ta": 1024.0, "th": 1024.0, "zh": 1024.0, "my": 1024.0}, "bpb_mean": 1.73629, "t": 831.6}
{"event": "train", "step": 3000, "phase": 1, "objective": "global", "lm_loss": 1.36436, "bpb": 1.96835, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.5082, "lr": 0.000264, "router_trainable": true, "sec_per_step": 0.278, "alpha": 0.03, "t": 831.9}
{"event": "train", "step": 3100, "phase": 1, "objective": "global", "lm_loss": 1.3655, "bpb": 1.97, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.541, "lr": 0.0002612, "router_trainable": true, "sec_per_step": 0.22, "alpha": 0.03, "t": 853.8}
{"event": "train", "step": 3200, "phase": 1, "objective": "global", "lm_loss": 1.24137, "bpb": 1.79092, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4932, "lr": 0.0002583, "router_trainable": true, "sec_per_step": 0.219, "alpha": 0.03, "t": 875.7}
{"event": "train", "step": 3300, "phase": 1, "objective": "global", "lm_loss": 1.32235, "bpb": 1.90775, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4932, "lr": 0.0002553, "router_trainable": true, "sec_per_step": 0.21, "alpha": 0.03, "t": 896.7}
{"event": "train", "step": 3400, "phase": 1, "objective": "global", "lm_loss": 1.27402, "bpb": 1.83803, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4993, "lr": 0.0002523, "router_trainable": true, "sec_per_step": 0.206, "alpha": 0.03, "t": 917.3}
{"event": "train", "step": 3500, "phase": 1, "objective": "global", "lm_loss": 1.35794, "bpb": 1.95909, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4642, "lr": 0.0002492, "router_trainable": true, "sec_per_step": 0.217, "alpha": 0.03, "t": 939.0}
{"event": "train", "step": 3600, "phase": 1, "objective": "global", "lm_loss": 1.27006, "bpb": 1.83231, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.5241, "lr": 0.000246, "router_trainable": true, "sec_per_step": 0.201, "alpha": 0.03, "t": 959.1}
{"event": "train", "step": 3700, "phase": 1, "objective": "global", "lm_loss": 1.18156, "bpb": 1.70463, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.5082, "lr": 0.0002427, "router_trainable": true, "sec_per_step": 0.221, "alpha": 0.03, "t": 981.2}
{"event": "train", "step": 3800, "phase": 1, "objective": "global", "lm_loss": 1.2735, "bpb": 1.83728, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4808, "lr": 0.0002394, "router_trainable": true, "sec_per_step": 0.217, "alpha": 0.03, "t": 1002.9}
{"event": "train", "step": 3900, "phase": 1, "objective": "global", "lm_loss": 1.29179, "bpb": 1.86367, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4544, "lr": 0.000236, "router_trainable": true, "sec_per_step": 0.216, "alpha": 0.03, "t": 1024.5}
{"event": "probe", "step": 4000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 17418.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 1.000862, "cps": 136.078125, "mean_p": 0.999901, "boundary_entropy": 0.001014, "bpb": 1.963597}, "de": {"chunks": 20862.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.016023, "cps": 162.984375, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 2.134306}, "vi": {"chunks": 24274.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.316734, "cps": 189.640625, "mean_p": 0.9999, "boundary_entropy": 0.001016, "bpb": 1.807843}, "tr": {"chunks": 19742.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.090537, "cps": 154.234375, "mean_p": 0.9999, "boundary_entropy": 0.001015, "bpb": 2.086962}, "fi": {"chunks": 19572.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 1.040455, "cps": 152.90625, "mean_p": 0.9999, "boundary_entropy": 0.001016, "bpb": 2.261765}, "ru": {"chunks": 34560.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.823747, "cps": 270.0, "mean_p": 0.999901, "boundary_entropy": 0.001018, "bpb": 1.306118}, "el": {"chunks": 37264.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.815719, "cps": 291.125, "mean_p": 0.9999, "boundary_entropy": 0.001018, "bpb": 1.337762}, "hi": {"chunks": 43173.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.566003, "cps": 337.289062, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 1.100892}, "ta": {"chunks": 50915.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.723455, "cps": 397.773438, "mean_p": 0.999901, "boundary_entropy": 0.001019, "bpb": 0.934295}, "th": {"chunks": 46085.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.865092, "cps": 360.039062, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 1.05408}, "zh": {"chunks": 16173.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.738402, "cps": 126.351562, "mean_p": 0.999901, "boundary_entropy": 0.001014, "bpb": 2.946757}, "my": {"chunks": 53743.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.836342, "cps": 419.867188, "mean_p": 0.999901, "boundary_entropy": 0.001019, "bpb": 1.052072}}, "bpb": {"en": 1.9636, "de": 2.13431, "vi": 1.80784, "tr": 2.08696, "fi": 2.26177, "ru": 1.30612, "el": 1.33776, "hi": 1.10089, "ta": 0.93429, "th": 1.05408, "zh": 2.94676, "my": 1.05207}, "gini_cpc": 0.221516, "gini_cps": 0.233853, "bpic": 1.0, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.01514801064883, "vi": 1.3156005295906583, "tr": 1.0895983329898389, "fi": 1.0395590357977051, "ru": 1.8221761311426539, "el": 1.8141552920914839, "hi": 2.563793186225859, "ta": 2.721110089193083, "th": 2.8626243461390897, "zh": 2.7360433739796948, "my": 2.8338989711748765}, "t": 1046.4}
{"event": "eval", "step": 4000, "phase": 1, "bpb": {"en": 1.97077, "de": 2.1274, "vi": 1.81482, "tr": 2.07808, "fi": 2.27791, "ru": 1.3049, "el": 1.33382, "hi": 1.11062, "ta": 0.93413, "th": 1.05176, "zh": 2.95387, "my": 1.06112}, "bytes_per_chunk": {"en": 1024.0, "de": 1024.0, "vi": 1024.0, "tr": 1024.0, "fi": 1024.0, "ru": 1024.0, "el": 1024.0, "hi": 1024.0, "ta": 1024.0, "th": 1024.0, "zh": 1024.0, "my": 1024.0}, "bpb_mean": 1.66827, "t": 1051.3}
{"event": "train", "step": 4000, "phase": 1, "objective": "global", "lm_loss": 1.35597, "bpb": 1.95625, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.5044, "lr": 0.0002325, "router_trainable": true, "sec_per_step": 0.27, "alpha": 0.03, "t": 1051.5}
{"event": "train", "step": 4100, "phase": 1, "objective": "global", "lm_loss": 1.33252, "bpb": 1.92241, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4772, "lr": 0.000229, "router_trainable": true, "sec_per_step": 0.2, "alpha": 0.03, "t": 1071.5}
{"event": "train", "step": 4200, "phase": 1, "objective": "global", "lm_loss": 1.27873, "bpb": 1.84482, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4441, "lr": 0.0002254, "router_trainable": true, "sec_per_step": 0.218, "alpha": 0.03, "t": 1093.4}
{"event": "train", "step": 4300, "phase": 1, "objective": "global", "lm_loss": 1.25119, "bpb": 1.80509, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4484, "lr": 0.0002217, "router_trainable": true, "sec_per_step": 0.222, "alpha": 0.03, "t": 1115.6}
{"event": "phase_transition", "step": 4400, "phase": 2, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0, "t": 1137.7}
{"event": "train", "step": 4400, "phase": 2, "objective": "global", "lm_loss": 1.20516, "bpb": 1.73867, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4173, "lr": 0.0002181, "router_trainable": true, "sec_per_step": 0.224, "alpha": 0.03, "t": 1137.9}
{"event": "train", "step": 4500, "phase": 2, "objective": "global", "lm_loss": 1.28367, "bpb": 1.85194, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.45, "lr": 0.0002143, "router_trainable": true, "sec_per_step": 0.22, "alpha": 0.03, "t": 1159.9}
{"event": "train", "step": 4600, "phase": 2, "objective": "global", "lm_loss": 1.23234, "bpb": 1.77789, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4417, "lr": 0.0002105, "router_trainable": true, "sec_per_step": 0.221, "alpha": 0.03, "t": 1182.1}
{"event": "train", "step": 4700, "phase": 2, "objective": "global", "lm_loss": 1.15703, "bpb": 1.66924, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4544, "lr": 0.0002067, "router_trainable": true, "sec_per_step": 0.203, "alpha": 0.03, "t": 1202.4}
{"event": "train", "step": 4800, "phase": 2, "objective": "global", "lm_loss": 1.28869, "bpb": 1.85919, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4564, "lr": 0.0002029, "router_trainable": true, "sec_per_step": 0.221, "alpha": 0.03, "t": 1224.5}
{"event": "train", "step": 4900, "phase": 2, "objective": "global", "lm_loss": 1.28113, "bpb": 1.84828, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4813, "lr": 0.000199, "router_trainable": true, "sec_per_step": 0.231, "alpha": 0.03, "t": 1247.6}
{"event": "probe", "step": 5000, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 17418.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 1.000862, "cps": 136.078125, "mean_p": 0.999901, "boundary_entropy": 0.001014, "bpb": 1.876599}, "de": {"chunks": 20862.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.016023, "cps": 162.984375, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 2.035784}, "vi": {"chunks": 24274.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.316734, "cps": 189.640625, "mean_p": 0.9999, "boundary_entropy": 0.001016, "bpb": 1.728379}, "tr": {"chunks": 19742.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.090537, "cps": 154.234375, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 1.96563}, "fi": {"chunks": 19572.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 1.040455, "cps": 152.90625, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 2.162315}, "ru": {"chunks": 34560.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.823747, "cps": 270.0, "mean_p": 0.999901, "boundary_entropy": 0.001018, "bpb": 1.258685}, "el": {"chunks": 37264.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.815719, "cps": 291.125, "mean_p": 0.999901, "boundary_entropy": 0.001018, "bpb": 1.279159}, "hi": {"chunks": 43173.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.566003, "cps": 337.289062, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 1.056924}, "ta": {"chunks": 50915.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.723455, "cps": 397.773438, "mean_p": 0.999901, "boundary_entropy": 0.001019, "bpb": 0.891978}, "th": {"chunks": 46085.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.865092, "cps": 360.039062, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 1.019191}, "zh": {"chunks": 16173.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.738402, "cps": 126.351562, "mean_p": 0.999901, "boundary_entropy": 0.001014, "bpb": 2.855319}, "my": {"chunks": 53743.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.836342, "cps": 419.867188, "mean_p": 0.999901, "boundary_entropy": 0.001019, "bpb": 1.005328}}, "bpb": {"en": 1.8766, "de": 2.03578, "vi": 1.72838, "tr": 1.96563, "fi": 2.16231, "ru": 1.25868, "el": 1.27916, "hi": 1.05692, "ta": 0.89198, "th": 1.01919, "zh": 2.85532, "my": 1.00533}, "gini_cpc": 0.221516, "gini_cps": 0.233853, "bpic": 1.0, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.01514801064883, "vi": 1.3156005295906583, "tr": 1.0895983329898389, "fi": 1.0395590357977051, "ru": 1.8221761311426539, "el": 1.8141552920914839, "hi": 2.563793186225859, "ta": 2.721110089193083, "th": 2.8626243461390897, "zh": 2.7360433739796948, "my": 2.8338989711748765}, "t": 1279.2}
{"event": "eval", "step": 5000, "phase": 2, "bpb": {"en": 1.87887, "de": 2.03302, "vi": 1.7396, "tr": 1.95976, "fi": 2.17381, "ru": 1.25832, "el": 1.27659, "hi": 1.06372, "ta": 0.89346, "th": 1.01822, "zh": 2.86506, "my": 1.0125}, "bytes_per_chunk": {"en": 1024.0, "de": 1024.0, "vi": 1024.0, "tr": 1024.0, "fi": 1024.0, "ru": 1024.0, "el": 1024.0, "hi": 1024.0, "ta": 1024.0, "th": 1024.0, "zh": 1024.0, "my": 1024.0}, "bpb_mean": 1.59774, "t": 1285.6}
{"event": "train", "step": 5000, "phase": 2, "objective": "global", "lm_loss": 1.24353, "bpb": 1.79404, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4637, "lr": 0.000195, "router_trainable": true, "sec_per_step": 0.386, "alpha": 0.03, "t": 1286.2}
{"event": "train", "step": 5100, "phase": 2, "objective": "global", "lm_loss": 1.24439, "bpb": 1.79528, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.432, "lr": 0.0001911, "router_trainable": true, "sec_per_step": 0.462, "alpha": 0.03, "t": 1332.4}
{"event": "train", "step": 5200, "phase": 2, "objective": "global", "lm_loss": 1.16968, "bpb": 1.68749, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4626, "lr": 0.0001871, "router_trainable": true, "sec_per_step": 0.45, "alpha": 0.03, "t": 1377.4}
{"event": "train", "step": 5300, "phase": 2, "objective": "global", "lm_loss": 1.19957, "bpb": 1.73061, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4663, "lr": 0.0001831, "router_trainable": true, "sec_per_step": 0.468, "alpha": 0.03, "t": 1424.2}
{"event": "train", "step": 5400, "phase": 2, "objective": "global", "lm_loss": 1.26136, "bpb": 1.81976, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4667, "lr": 0.0001791, "router_trainable": true, "sec_per_step": 0.45, "alpha": 0.03, "t": 1469.1}
{"event": "train", "step": 5500, "phase": 2, "objective": "global", "lm_loss": 1.21374, "bpb": 1.75106, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4988, "lr": 0.0001751, "router_trainable": true, "sec_per_step": 0.461, "alpha": 0.03, "t": 1515.3}
{"event": "train", "step": 5600, "phase": 2, "objective": "global", "lm_loss": 1.16591, "bpb": 1.68205, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4514, "lr": 0.0001711, "router_trainable": true, "sec_per_step": 0.45, "alpha": 0.03, "t": 1560.3}
{"event": "train", "step": 5700, "phase": 2, "objective": "global", "lm_loss": 1.14868, "bpb": 1.65719, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4655, "lr": 0.000167, "router_trainable": true, "sec_per_step": 0.461, "alpha": 0.03, "t": 1606.4}
{"event": "train", "step": 5800, "phase": 2, "objective": "global", "lm_loss": 1.14444, "bpb": 1.65108, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4627, "lr": 0.000163, "router_trainable": true, "sec_per_step": 0.46, "alpha": 0.03, "t": 1652.4}
{"event": "train", "step": 5900, "phase": 2, "objective": "global", "lm_loss": 1.19321, "bpb": 1.72143, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4662, "lr": 0.0001589, "router_trainable": true, "sec_per_step": 0.42, "alpha": 0.03, "t": 1694.4}
{"event": "probe", "step": 6000, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 17418.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 1.000862, "cps": 136.078125, "mean_p": 0.999901, "boundary_entropy": 0.001014, "bpb": 1.820676}, "de": {"chunks": 20862.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.016023, "cps": 162.984375, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 1.959014}, "vi": {"chunks": 24274.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.316734, "cps": 189.640625, "mean_p": 0.9999, "boundary_entropy": 0.001016, "bpb": 1.658107}, "tr": {"chunks": 19742.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.090537, "cps": 154.234375, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 1.877065}, "fi": {"chunks": 19572.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 1.040455, "cps": 152.90625, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 2.086558}, "ru": {"chunks": 34560.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.823747, "cps": 270.0, "mean_p": 0.999901, "boundary_entropy": 0.001018, "bpb": 1.227278}, "el": {"chunks": 37264.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.815719, "cps": 291.125, "mean_p": 0.999901, "boundary_entropy": 0.001018, "bpb": 1.241382}, "hi": {"chunks": 43173.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.566003, "cps": 337.289062, "mean_p": 0.9999, "boundary_entropy": 0.001018, "bpb": 1.030055}, "ta": {"chunks": 50915.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.723455, "cps": 397.773438, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 0.877577}, "th": {"chunks": 46085.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.865092, "cps": 360.039062, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 0.9934}, "zh": {"chunks": 16173.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.738402, "cps": 126.351562, "mean_p": 0.999901, "boundary_entropy": 0.001013, "bpb": 2.797438}, "my": {"chunks": 53743.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.836342, "cps": 419.867188, "mean_p": 0.999901, "boundary_entropy": 0.001019, "bpb": 0.981914}}, "bpb": {"en": 1.82068, "de": 1.95901, "vi": 1.65811, "tr": 1.87707, "fi": 2.08656, "ru": 1.22728, "el": 1.24138, "hi": 1.03005, "ta": 0.87758, "th": 0.9934, "zh": 2.79744, "my": 0.98191}, "gini_cpc": 0.221516, "gini_cps": 0.233853, "bpic": 1.0, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.01514801064883, "vi": 1.3156005295906583, "tr": 1.0895983329898389, "fi": 1.0395590357977051, "ru": 1.8221761311426539, "el": 1.8141552920914839, "hi": 2.563793186225859, "ta": 2.721110089193083, "th": 2.8626243461390897, "zh": 2.7360433739796948, "my": 2.8338989711748765}, "t": 1741.5}
{"event": "eval", "step": 6000, "phase": 2, "bpb": {"en": 1.82145, "de": 1.96064, "vi": 1.6717, "tr": 1.87051, "fi": 2.09626, "ru": 1.22667, "el": 1.23806, "hi": 1.03599, "ta": 0.87977, "th": 0.99191, "zh": 2.80905, "my": 0.98977}, "bytes_per_chunk": {"en": 1024.0, "de": 1024.0, "vi": 1024.0, "tr": 1024.0, "fi": 1024.0, "ru": 1024.0, "el": 1024.0, "hi": 1024.0, "ta": 1024.0, "th": 1024.0, "zh": 1024.0, "my": 1024.0}, "bpb_mean": 1.54931, "t": 1748.1}
{"event": "train", "step": 6000, "phase": 2, "objective": "global", "lm_loss": 1.17255, "bpb": 1.69163, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.485, "lr": 0.0001549, "router_trainable": true, "sec_per_step": 0.541, "alpha": 0.03, "t": 1748.5}
{"event": "train", "step": 6100, "phase": 2, "objective": "global", "lm_loss": 1.11165, "bpb": 1.60377, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.456, "lr": 0.0001509, "router_trainable": true, "sec_per_step": 0.433, "alpha": 0.03, "t": 1791.8}
{"event": "train", "step": 6200, "phase": 2, "objective": "global", "lm_loss": 1.23117, "bpb": 1.7762, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.471, "lr": 0.0001469, "router_trainable": true, "sec_per_step": 0.439, "alpha": 0.03, "t": 1835.7}
{"event": "train", "step": 6300, "phase": 2, "objective": "global", "lm_loss": 1.15256, "bpb": 1.66279, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4549, "lr": 0.0001429, "router_trainable": true, "sec_per_step": 0.429, "alpha": 0.03, "t": 1878.6}
{"event": "train", "step": 6400, "phase": 2, "objective": "global", "lm_loss": 1.05181, "bpb": 1.51744, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.448, "lr": 0.0001389, "router_trainable": true, "sec_per_step": 0.452, "alpha": 0.03, "t": 1923.9}
{"event": "train", "step": 6500, "phase": 2, "objective": "global", "lm_loss": 1.11445, "bpb": 1.60781, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4358, "lr": 0.000135, "router_trainable": true, "sec_per_step": 0.441, "alpha": 0.03, "t": 1968.0}
{"event": "train", "step": 6600, "phase": 2, "objective": "global", "lm_loss": 1.17399, "bpb": 1.69371, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4368, "lr": 0.000131, "router_trainable": true, "sec_per_step": 0.444, "alpha": 0.03, "t": 2012.4}
{"event": "train", "step": 6700, "phase": 2, "objective": "global", "lm_loss": 1.15738, "bpb": 1.66975, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.448, "lr": 0.0001271, "router_trainable": true, "sec_per_step": 0.42, "alpha": 0.03, "t": 2054.4}
{"event": "train", "step": 6800, "phase": 2, "objective": "global", "lm_loss": 1.17468, "bpb": 1.6947, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4857, "lr": 0.0001233, "router_trainable": true, "sec_per_step": 0.451, "alpha": 0.03, "t": 2099.5}
{"event": "train", "step": 6900, "phase": 2, "objective": "global", "lm_loss": 1.14731, "bpb": 1.65522, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4584, "lr": 0.0001195, "router_trainable": true, "sec_per_step": 0.449, "alpha": 0.03, "t": 2144.4}
{"event": "probe", "step": 7000, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 17418.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 1.000862, "cps": 136.078125, "mean_p": 0.999901, "boundary_entropy": 0.001014, "bpb": 1.783365}, "de": {"chunks": 20862.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.016023, "cps": 162.984375, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 1.920238}, "vi": {"chunks": 24274.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.316734, "cps": 189.640625, "mean_p": 0.9999, "boundary_entropy": 0.001016, "bpb": 1.605618}, "tr": {"chunks": 19742.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.090537, "cps": 154.234375, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 1.824091}, "fi": {"chunks": 19572.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 1.040455, "cps": 152.90625, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 2.02648}, "ru": {"chunks": 34560.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.823747, "cps": 270.0, "mean_p": 0.999901, "boundary_entropy": 0.001018, "bpb": 1.20332}, "el": {"chunks": 37264.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.815719, "cps": 291.125, "mean_p": 0.999901, "boundary_entropy": 0.001018, "bpb": 1.206393}, "hi": {"chunks": 43173.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.566003, "cps": 337.289062, "mean_p": 0.9999, "boundary_entropy": 0.001018, "bpb": 1.00371}, "ta": {"chunks": 50915.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.723455, "cps": 397.773438, "mean_p": 0.999901, "boundary_entropy": 0.001019, "bpb": 0.856048}, "th": {"chunks": 46085.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.865092, "cps": 360.039062, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 0.966194}, "zh": {"chunks": 16173.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.738402, "cps": 126.351562, "mean_p": 0.999901, "boundary_entropy": 0.001013, "bpb": 2.758101}, "my": {"chunks": 53743.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.836342, "cps": 419.867188, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 0.95033}}, "bpb": {"en": 1.78337, "de": 1.92024, "vi": 1.60562, "tr": 1.82409, "fi": 2.02648, "ru": 1.20332, "el": 1.20639, "hi": 1.00371, "ta": 0.85605, "th": 0.96619, "zh": 2.7581, "my": 0.95033}, "gini_cpc": 0.221516, "gini_cps": 0.233853, "bpic": 1.0, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.01514801064883, "vi": 1.3156005295906583, "tr": 1.0895983329898389, "fi": 1.0395590357977051, "ru": 1.8221761311426539, "el": 1.8141552920914839, "hi": 2.563793186225859, "ta": 2.721110089193083, "th": 2.8626243461390897, "zh": 2.7360433739796948, "my": 2.8338989711748765}, "t": 2192.0}
{"event": "eval", "step": 7000, "phase": 2, "bpb": {"en": 1.78437, "de": 1.92254, "vi": 1.62018, "tr": 1.81798, "fi": 2.03697, "ru": 1.20275, "el": 1.20441, "hi": 1.00979, "ta": 0.85958, "th": 0.96477, "zh": 2.77003, "my": 0.95844}, "bytes_per_chunk": {"en": 1024.0, "de": 1024.0, "vi": 1024.0, "tr": 1024.0, "fi": 1024.0, "ru": 1024.0, "el": 1024.0, "hi": 1024.0, "ta": 1024.0, "th": 1024.0, "zh": 1024.0, "my": 1024.0}, "bpb_mean": 1.51265, "t": 2198.2}
{"event": "train", "step": 7000, "phase": 2, "objective": "global", "lm_loss": 1.14473, "bpb": 1.6515, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4511, "lr": 0.0001157, "router_trainable": true, "sec_per_step": 0.543, "alpha": 0.03, "t": 2198.7}
{"event": "train", "step": 7100, "phase": 2, "objective": "global", "lm_loss": 1.12373, "bpb": 1.62119, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4622, "lr": 0.0001119, "router_trainable": true, "sec_per_step": 0.453, "alpha": 0.03, "t": 2244.0}
{"event": "train", "step": 7200, "phase": 2, "objective": "global", "lm_loss": 1.17066, "bpb": 1.68891, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4647, "lr": 0.0001083, "router_trainable": true, "sec_per_step": 0.466, "alpha": 0.03, "t": 2290.6}
{"event": "train", "step": 7300, "phase": 2, "objective": "global", "lm_loss": 1.18122, "bpb": 1.70414, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4607, "lr": 0.0001046, "router_trainable": true, "sec_per_step": 0.494, "alpha": 0.03, "t": 2340.0}
{"event": "train", "step": 7400, "phase": 2, "objective": "global", "lm_loss": 1.16061, "bpb": 1.6744, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4779, "lr": 0.000101, "router_trainable": true, "sec_per_step": 0.488, "alpha": 0.03, "t": 2388.7}
{"event": "train", "step": 7500, "phase": 2, "objective": "global", "lm_loss": 1.13608, "bpb": 1.63901, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4435, "lr": 9.75e-05, "router_trainable": true, "sec_per_step": 0.48, "alpha": 0.03, "t": 2436.7}
{"event": "train", "step": 7600, "phase": 2, "objective": "global", "lm_loss": 1.1956, "bpb": 1.72488, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4472, "lr": 9.4e-05, "router_trainable": true, "sec_per_step": 0.475, "alpha": 0.03, "t": 2484.2}
{"event": "phase_transition", "step": 7700, "phase": 3, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0, "t": 2531.3}
{"event": "train", "step": 7700, "phase": 3, "objective": "global", "lm_loss": 1.14078, "bpb": 1.64579, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4504, "lr": 9.06e-05, "router_trainable": true, "sec_per_step": 0.477, "alpha": 0.03, "t": 2531.9}
{"event": "train", "step": 7800, "phase": 3, "objective": "global", "lm_loss": 1.18308, "bpb": 1.70683, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.458, "lr": 8.73e-05, "router_trainable": true, "sec_per_step": 0.478, "alpha": 0.03, "t": 2579.7}
{"event": "train", "step": 7900, "phase": 3, "objective": "global", "lm_loss": 1.16441, "bpb": 1.67988, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4894, "lr": 8.4e-05, "router_trainable": true, "sec_per_step": 0.473, "alpha": 0.03, "t": 2627.0}
{"event": "probe", "step": 8000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 17418.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 1.000862, "cps": 136.078125, "mean_p": 0.999901, "boundary_entropy": 0.001014, "bpb": 1.76066}, "de": {"chunks": 20862.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.016023, "cps": 162.984375, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 1.892858}, "vi": {"chunks": 24274.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.316734, "cps": 189.640625, "mean_p": 0.9999, "boundary_entropy": 0.001016, "bpb": 1.569466}, "tr": {"chunks": 19742.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.090537, "cps": 154.234375, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 1.789086}, "fi": {"chunks": 19572.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 1.040455, "cps": 152.90625, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 1.989975}, "ru": {"chunks": 34560.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.823747, "cps": 270.0, "mean_p": 0.999901, "boundary_entropy": 0.001018, "bpb": 1.184841}, "el": {"chunks": 37264.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.815719, "cps": 291.125, "mean_p": 0.999901, "boundary_entropy": 0.001018, "bpb": 1.186428}, "hi": {"chunks": 43173.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.566003, "cps": 337.289062, "mean_p": 0.9999, "boundary_entropy": 0.001018, "bpb": 0.98717}, "ta": {"chunks": 50915.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.723455, "cps": 397.773438, "mean_p": 0.999901, "boundary_entropy": 0.001019, "bpb": 0.837917}, "th": {"chunks": 46085.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.865092, "cps": 360.039062, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 0.951265}, "zh": {"chunks": 16173.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.738402, "cps": 126.351562, "mean_p": 0.999901, "boundary_entropy": 0.001013, "bpb": 2.718303}, "my": {"chunks": 53743.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.836342, "cps": 419.867188, "mean_p": 0.999901, "boundary_entropy": 0.001019, "bpb": 0.93308}}, "bpb": {"en": 1.76066, "de": 1.89286, "vi": 1.56947, "tr": 1.78909, "fi": 1.98997, "ru": 1.18484, "el": 1.18643, "hi": 0.98717, "ta": 0.83792, "th": 0.95127, "zh": 2.7183, "my": 0.93308}, "gini_cpc": 0.221516, "gini_cps": 0.233853, "bpic": 1.0, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.01514801064883, "vi": 1.3156005295906583, "tr": 1.0895983329898389, "fi": 1.0395590357977051, "ru": 1.8221761311426539, "el": 1.8141552920914839, "hi": 2.563793186225859, "ta": 2.721110089193083, "th": 2.8626243461390897, "zh": 2.7360433739796948, "my": 2.8338989711748765}, "t": 2676.5}
{"event": "eval", "step": 8000, "phase": 3, "bpb": {"en": 1.76058, "de": 1.89426, "vi": 1.58449, "tr": 1.78262, "fi": 1.99926, "ru": 1.18466, "el": 1.18431, "hi": 0.99267, "ta": 0.84174, "th": 0.95029, "zh": 2.73154, "my": 0.94183}, "bytes_per_chunk": {"en": 1024.0, "de": 1024.0, "vi": 1024.0, "tr": 1024.0, "fi": 1024.0, "ru": 1024.0, "el": 1024.0, "hi": 1024.0, "ta": 1024.0, "th": 1024.0, "zh": 1024.0, "my": 1024.0}, "bpb_mean": 1.48735, "t": 2682.7}
{"event": "train", "step": 8000, "phase": 3, "objective": "global", "lm_loss": 1.15583, "bpb": 1.66751, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4682, "lr": 8.08e-05, "router_trainable": true, "sec_per_step": 0.562, "alpha": 0.03, "t": 2683.2}
{"event": "train", "step": 8100, "phase": 3, "objective": "global", "lm_loss": 1.18697, "bpb": 1.71243, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4889, "lr": 7.77e-05, "router_trainable": true, "sec_per_step": 0.481, "alpha": 0.03, "t": 2731.3}
{"event": "train", "step": 8200, "phase": 3, "objective": "global", "lm_loss": 1.13931, "bpb": 1.64368, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4652, "lr": 7.47e-05, "router_trainable": true, "sec_per_step": 0.475, "alpha": 0.03, "t": 2778.8}
{"event": "train", "step": 8300, "phase": 3, "objective": "global", "lm_loss": 1.07152, "bpb": 1.54588, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4444, "lr": 7.17e-05, "router_trainable": true, "sec_per_step": 0.489, "alpha": 0.03, "t": 2827.7}
{"event": "train", "step": 8400, "phase": 3, "objective": "global", "lm_loss": 1.1324, "bpb": 1.63371, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4695, "lr": 6.88e-05, "router_trainable": true, "sec_per_step": 0.486, "alpha": 0.03, "t": 2876.3}
{"event": "train", "step": 8500, "phase": 3, "objective": "global", "lm_loss": 1.13281, "bpb": 1.6343, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4507, "lr": 6.6e-05, "router_trainable": true, "sec_per_step": 0.467, "alpha": 0.03, "t": 2923.0}
{"event": "train", "step": 8600, "phase": 3, "objective": "global", "lm_loss": 1.188, "bpb": 1.71392, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4664, "lr": 6.33e-05, "router_trainable": true, "sec_per_step": 0.474, "alpha": 0.03, "t": 2970.4}
{"event": "train", "step": 8700, "phase": 3, "objective": "global", "lm_loss": 1.06081, "bpb": 1.53043, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4664, "lr": 6.07e-05, "router_trainable": true, "sec_per_step": 0.487, "alpha": 0.03, "t": 3019.1}
{"event": "train", "step": 8800, "phase": 3, "objective": "global", "lm_loss": 1.08568, "bpb": 1.56631, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4632, "lr": 5.82e-05, "router_trainable": true, "sec_per_step": 0.479, "alpha": 0.03, "t": 3067.0}
{"event": "train", "step": 8900, "phase": 3, "objective": "global", "lm_loss": 1.14597, "bpb": 1.65328, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.453, "lr": 5.58e-05, "router_trainable": true, "sec_per_step": 0.479, "alpha": 0.03, "t": 3114.9}
{"event": "probe", "step": 9000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 17418.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 1.000862, "cps": 136.078125, "mean_p": 0.999901, "boundary_entropy": 0.001014, "bpb": 1.744933}, "de": {"chunks": 20862.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.016023, "cps": 162.984375, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 1.870303}, "vi": {"chunks": 24274.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.316734, "cps": 189.640625, "mean_p": 0.9999, "boundary_entropy": 0.001016, "bpb": 1.540315}, "tr": {"chunks": 19742.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.090537, "cps": 154.234375, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 1.766031}, "fi": {"chunks": 19572.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 1.040455, "cps": 152.90625, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 1.962204}, "ru": {"chunks": 34560.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.823747, "cps": 270.0, "mean_p": 0.999901, "boundary_entropy": 0.001018, "bpb": 1.169388}, "el": {"chunks": 37264.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.815719, "cps": 291.125, "mean_p": 0.999901, "boundary_entropy": 0.001018, "bpb": 1.166382}, "hi": {"chunks": 43173.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.566003, "cps": 337.289062, "mean_p": 0.9999, "boundary_entropy": 0.001018, "bpb": 0.972192}, "ta": {"chunks": 50915.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.723455, "cps": 397.773438, "mean_p": 0.999901, "boundary_entropy": 0.001019, "bpb": 0.825848}, "th": {"chunks": 46085.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.865092, "cps": 360.039062, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 0.935261}, "zh": {"chunks": 16173.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.738402, "cps": 126.351562, "mean_p": 0.999901, "boundary_entropy": 0.001013, "bpb": 2.690303}, "my": {"chunks": 53743.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.836342, "cps": 419.867188, "mean_p": 0.999901, "boundary_entropy": 0.001019, "bpb": 0.915811}}, "bpb": {"en": 1.74493, "de": 1.8703, "vi": 1.54031, "tr": 1.76603, "fi": 1.9622, "ru": 1.16939, "el": 1.16638, "hi": 0.97219, "ta": 0.82585, "th": 0.93526, "zh": 2.6903, "my": 0.91581}, "gini_cpc": 0.221516, "gini_cps": 0.233853, "bpic": 1.0, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.01514801064883, "vi": 1.3156005295906583, "tr": 1.0895983329898389, "fi": 1.0395590357977051, "ru": 1.8221761311426539, "el": 1.8141552920914839, "hi": 2.563793186225859, "ta": 2.721110089193083, "th": 2.8626243461390897, "zh": 2.7360433739796948, "my": 2.8338989711748765}, "t": 3167.0}
{"event": "eval", "step": 9000, "phase": 3, "bpb": {"en": 1.74478, "de": 1.87243, "vi": 1.55536, "tr": 1.76024, "fi": 1.97242, "ru": 1.16829, "el": 1.16392, "hi": 0.97713, "ta": 0.82997, "th": 0.93463, "zh": 2.70477, "my": 0.9251}, "bytes_per_chunk": {"en": 1024.0, "de": 1024.0, "vi": 1024.0, "tr": 1024.0, "fi": 1024.0, "ru": 1024.0, "el": 1024.0, "hi": 1024.0, "ta": 1024.0, "th": 1024.0, "zh": 1024.0, "my": 1024.0}, "bpb_mean": 1.46742, "t": 3173.2}
{"event": "train", "step": 9000, "phase": 3, "objective": "global", "lm_loss": 1.12049, "bpb": 1.61653, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4753, "lr": 5.35e-05, "router_trainable": true, "sec_per_step": 0.589, "alpha": 0.03, "t": 3173.7}
{"event": "train", "step": 9100, "phase": 3, "objective": "global", "lm_loss": 1.12018, "bpb": 1.61607, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4723, "lr": 5.12e-05, "router_trainable": true, "sec_per_step": 0.49, "alpha": 0.03, "t": 3222.7}
{"event": "train", "step": 9200, "phase": 3, "objective": "global", "lm_loss": 1.03123, "bpb": 1.48776, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.449, "lr": 4.91e-05, "router_trainable": true, "sec_per_step": 0.488, "alpha": 0.03, "t": 3271.5}
{"event": "train", "step": 9300, "phase": 3, "objective": "global", "lm_loss": 1.10563, "bpb": 1.59508, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4517, "lr": 4.71e-05, "router_trainable": true, "sec_per_step": 0.481, "alpha": 0.03, "t": 3319.7}
{"event": "train", "step": 9400, "phase": 3, "objective": "global", "lm_loss": 1.15731, "bpb": 1.66965, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4699, "lr": 4.52e-05, "router_trainable": true, "sec_per_step": 0.491, "alpha": 0.03, "t": 3368.8}
{"event": "train", "step": 9500, "phase": 3, "objective": "global", "lm_loss": 1.17731, "bpb": 1.6985, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4839, "lr": 4.34e-05, "router_trainable": true, "sec_per_step": 0.498, "alpha": 0.03, "t": 3418.6}
{"event": "train", "step": 9600, "phase": 3, "objective": "global", "lm_loss": 1.0339, "bpb": 1.4916, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.486, "lr": 4.17e-05, "router_trainable": true, "sec_per_step": 0.484, "alpha": 0.03, "t": 3467.0}
{"event": "train", "step": 9700, "phase": 3, "objective": "global", "lm_loss": 1.04811, "bpb": 1.5121, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4606, "lr": 4.01e-05, "router_trainable": true, "sec_per_step": 0.486, "alpha": 0.03, "t": 3515.6}
{"event": "train", "step": 9800, "phase": 3, "objective": "global", "lm_loss": 1.06467, "bpb": 1.536, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4619, "lr": 3.86e-05, "router_trainable": true, "sec_per_step": 0.49, "alpha": 0.03, "t": 3564.6}
{"event": "train", "step": 9900, "phase": 3, "objective": "global", "lm_loss": 1.1287, "bpb": 1.62838, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4708, "lr": 3.72e-05, "router_trainable": true, "sec_per_step": 0.472, "alpha": 0.03, "t": 3611.8}
{"event": "probe", "step": 10000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 17418.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 1.000862, "cps": 136.078125, "mean_p": 0.999901, "boundary_entropy": 0.001014, "bpb": 1.732629}, "de": {"chunks": 20862.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.016023, "cps": 162.984375, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 1.856695}, "vi": {"chunks": 24274.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.316734, "cps": 189.640625, "mean_p": 0.9999, "boundary_entropy": 0.001016, "bpb": 1.526256}, "tr": {"chunks": 19742.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.090537, "cps": 154.234375, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 1.751311}, "fi": {"chunks": 19572.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 1.040455, "cps": 152.90625, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 1.944102}, "ru": {"chunks": 34560.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.823747, "cps": 270.0, "mean_p": 0.999901, "boundary_entropy": 0.001017, "bpb": 1.159053}, "el": {"chunks": 37264.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.815719, "cps": 291.125, "mean_p": 0.999901, "boundary_entropy": 0.001018, "bpb": 1.152006}, "hi": {"chunks": 43173.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.566003, "cps": 337.289062, "mean_p": 0.9999, "boundary_entropy": 0.001018, "bpb": 0.960174}, "ta": {"chunks": 50915.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.723455, "cps": 397.773438, "mean_p": 0.999901, "boundary_entropy": 0.001019, "bpb": 0.816698}, "th": {"chunks": 46085.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.865092, "cps": 360.039062, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 0.926787}, "zh": {"chunks": 16173.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.738402, "cps": 126.351562, "mean_p": 0.999901, "boundary_entropy": 0.001013, "bpb": 2.673016}, "my": {"chunks": 53743.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.836342, "cps": 419.867188, "mean_p": 0.999901, "boundary_entropy": 0.001019, "bpb": 0.906425}}, "bpb": {"en": 1.73263, "de": 1.8567, "vi": 1.52626, "tr": 1.75131, "fi": 1.9441, "ru": 1.15905, "el": 1.15201, "hi": 0.96017, "ta": 0.8167, "th": 0.92679, "zh": 2.67302, "my": 0.90642}, "gini_cpc": 0.221516, "gini_cps": 0.233853, "bpic": 1.0, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.01514801064883, "vi": 1.3156005295906583, "tr": 1.0895983329898389, "fi": 1.0395590357977051, "ru": 1.8221761311426539, "el": 1.8141552920914839, "hi": 2.563793186225859, "ta": 2.721110089193083, "th": 2.8626243461390897, "zh": 2.7360433739796948, "my": 2.8338989711748765}, "t": 3662.9}
{"event": "eval", "step": 10000, "phase": 3, "bpb": {"en": 1.73278, "de": 1.85946, "vi": 1.54183, "tr": 1.74485, "fi": 1.9552, "ru": 1.15852, "el": 1.15018, "hi": 0.96493, "ta": 0.82064, "th": 0.92549, "zh": 2.68791, "my": 0.91514}, "bytes_per_chunk": {"en": 1024.0, "de": 1024.0, "vi": 1024.0, "tr": 1024.0, "fi": 1024.0, "ru": 1024.0, "el": 1024.0, "hi": 1024.0, "ta": 1024.0, "th": 1024.0, "zh": 1024.0, "my": 1024.0}, "bpb_mean": 1.45474, "t": 3669.0}
{"event": "train", "step": 10000, "phase": 3, "objective": "global", "lm_loss": 1.11803, "bpb": 1.61297, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4968, "lr": 3.6e-05, "router_trainable": true, "sec_per_step": 0.577, "alpha": 0.03, "t": 3669.5}
{"event": "train", "step": 10100, "phase": 3, "objective": "global", "lm_loss": 1.05687, "bpb": 1.52474, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4613, "lr": 3.49e-05, "router_trainable": true, "sec_per_step": 0.482, "alpha": 0.03, "t": 3717.6}
{"event": "train", "step": 10200, "phase": 3, "objective": "global", "lm_loss": 1.15509, "bpb": 1.66644, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.503, "lr": 3.38e-05, "router_trainable": true, "sec_per_step": 0.488, "alpha": 0.03, "t": 3766.5}
{"event": "train", "step": 10300, "phase": 3, "objective": "global", "lm_loss": 1.12695, "bpb": 1.62585, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.5129, "lr": 3.3e-05, "router_trainable": true, "sec_per_step": 0.491, "alpha": 0.03, "t": 3815.5}
{"event": "train", "step": 10400, "phase": 3, "objective": "global", "lm_loss": 1.07389, "bpb": 1.54929, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4918, "lr": 3.22e-05, "router_trainable": true, "sec_per_step": 0.465, "alpha": 0.03, "t": 3862.1}
{"event": "train", "step": 10500, "phase": 3, "objective": "global", "lm_loss": 1.10918, "bpb": 1.60021, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4939, "lr": 3.15e-05, "router_trainable": true, "sec_per_step": 0.46, "alpha": 0.03, "t": 3908.1}
{"event": "train", "step": 10600, "phase": 3, "objective": "global", "lm_loss": 1.12746, "bpb": 1.62657, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4919, "lr": 3.1e-05, "router_trainable": true, "sec_per_step": 0.466, "alpha": 0.03, "t": 3954.7}
{"event": "train", "step": 10700, "phase": 3, "objective": "global", "lm_loss": 1.09462, "bpb": 1.5792, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4691, "lr": 3.05e-05, "router_trainable": true, "sec_per_step": 0.473, "alpha": 0.03, "t": 4002.0}
{"event": "train", "step": 10800, "phase": 3, "objective": "global", "lm_loss": 1.10413, "bpb": 1.59292, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4815, "lr": 3.02e-05, "router_trainable": true, "sec_per_step": 0.466, "alpha": 0.03, "t": 4048.6}
{"event": "train", "step": 10900, "phase": 3, "objective": "global", "lm_loss": 1.096, "bpb": 1.58119, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.5027, "lr": 3.01e-05, "router_trainable": true, "sec_per_step": 0.456, "alpha": 0.03, "t": 4094.2}
{"event": "probe", "step": 11000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 17418.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 1.000862, "cps": 136.078125, "mean_p": 0.999901, "boundary_entropy": 0.001014, "bpb": 1.726535}, "de": {"chunks": 20862.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.016023, "cps": 162.984375, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 1.847825}, "vi": {"chunks": 24274.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.316734, "cps": 189.640625, "mean_p": 0.9999, "boundary_entropy": 0.001016, "bpb": 1.513006}, "tr": {"chunks": 19742.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.090537, "cps": 154.234375, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 1.743988}, "fi": {"chunks": 19572.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 1.040455, "cps": 152.90625, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 1.934874}, "ru": {"chunks": 34560.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.823747, "cps": 270.0, "mean_p": 0.999901, "boundary_entropy": 0.001018, "bpb": 1.151774}, "el": {"chunks": 37264.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.815719, "cps": 291.125, "mean_p": 0.999901, "boundary_entropy": 0.001018, "bpb": 1.145279}, "hi": {"chunks": 43173.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.566003, "cps": 337.289062, "mean_p": 0.9999, "boundary_entropy": 0.001018, "bpb": 0.955037}, "ta": {"chunks": 50915.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.723455, "cps": 397.773438, "mean_p": 0.999901, "boundary_entropy": 0.001019, "bpb": 0.813174}, "th": {"chunks": 46085.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.865092, "cps": 360.039062, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 0.921141}, "zh": {"chunks": 16173.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.738402, "cps": 126.351562, "mean_p": 0.999901, "boundary_entropy": 0.001013, "bpb": 2.665389}, "my": {"chunks": 53743.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.836342, "cps": 419.867188, "mean_p": 0.999901, "boundary_entropy": 0.001019, "bpb": 0.899157}}, "bpb": {"en": 1.72654, "de": 1.84783, "vi": 1.51301, "tr": 1.74399, "fi": 1.93487, "ru": 1.15177, "el": 1.14528, "hi": 0.95504, "ta": 0.81317, "th": 0.92114, "zh": 2.66539, "my": 0.89916}, "gini_cpc": 0.221516, "gini_cps": 0.233853, "bpic": 1.0, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.01514801064883, "vi": 1.3156005295906583, "tr": 1.0895983329898389, "fi": 1.0395590357977051, "ru": 1.8221761311426539, "el": 1.8141552920914839, "hi": 2.563793186225859, "ta": 2.721110089193083, "th": 2.8626243461390897, "zh": 2.7360433739796948, "my": 2.8338989711748765}, "t": 4143.0}
{"event": "eval", "step": 11000, "phase": 3, "bpb": {"en": 1.72657, "de": 1.85033, "vi": 1.52914, "tr": 1.73808, "fi": 1.94514, "ru": 1.15161, "el": 1.14326, "hi": 0.96041, "ta": 0.81746, "th": 0.9199, "zh": 2.68009, "my": 0.90789}, "bytes_per_chunk": {"en": 1024.0, "de": 1024.0, "vi": 1024.0, "tr": 1024.0, "fi": 1024.0, "ru": 1024.0, "el": 1024.0, "hi": 1024.0, "ta": 1024.0, "th": 1024.0, "zh": 1024.0, "my": 1024.0}, "bpb_mean": 1.44749, "t": 4149.1}
{"event": "probe", "step": 11000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 17418.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 1.000862, "cps": 136.078125, "mean_p": 0.999901, "boundary_entropy": 0.001014, "bpb": 1.726535}, "de": {"chunks": 20862.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.016023, "cps": 162.984375, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 1.847825}, "vi": {"chunks": 24274.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.316734, "cps": 189.640625, "mean_p": 0.9999, "boundary_entropy": 0.001016, "bpb": 1.513006}, "tr": {"chunks": 19742.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.090537, "cps": 154.234375, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 1.743988}, "fi": {"chunks": 19572.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 1.040455, "cps": 152.90625, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 1.934874}, "ru": {"chunks": 34560.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.823747, "cps": 270.0, "mean_p": 0.999901, "boundary_entropy": 0.001018, "bpb": 1.151774}, "el": {"chunks": 37264.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.815719, "cps": 291.125, "mean_p": 0.999901, "boundary_entropy": 0.001018, "bpb": 1.145279}, "hi": {"chunks": 43173.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.566003, "cps": 337.289062, "mean_p": 0.9999, "boundary_entropy": 0.001018, "bpb": 0.955037}, "ta": {"chunks": 50915.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.723455, "cps": 397.773438, "mean_p": 0.999901, "boundary_entropy": 0.001019, "bpb": 0.813174}, "th": {"chunks": 46085.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.865092, "cps": 360.039062, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 0.921141}, "zh": {"chunks": 16173.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.738402, "cps": 126.351562, "mean_p": 0.999901, "boundary_entropy": 0.001013, "bpb": 2.665389}, "my": {"chunks": 53743.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.836342, "cps": 419.867188, "mean_p": 0.999901, "boundary_entropy": 0.001019, "bpb": 0.899157}}, "bpb": {"en": 1.72654, "de": 1.84783, "vi": 1.51301, "tr": 1.74399, "fi": 1.93487, "ru": 1.15177, "el": 1.14528, "hi": 0.95504, "ta": 0.81317, "th": 0.92114, "zh": 2.66539, "my": 0.89916}, "gini_cpc": 0.221516, "gini_cps": 0.233853, "bpic": 1.0, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.01514801064883, "vi": 1.3156005295906583, "tr": 1.0895983329898389, "fi": 1.0395590357977051, "ru": 1.8221761311426539, "el": 1.8141552920914839, "hi": 2.563793186225859, "ta": 2.721110089193083, "th": 2.8626243461390897, "zh": 2.7360433739796948, "my": 2.8338989711748765}, "t": 4152.4}
{"event": "reps_dumped", "dirs": ["/local/scratch/sas245/beetle-hnet/scratch/reps/exp32_onechunk_s0/th0.30", "/local/scratch/sas245/beetle-hnet/scratch/reps/exp32_onechunk_s0/th0.40", "/local/scratch/sas245/beetle-hnet/scratch/reps/exp32_onechunk_s0/th0.50", "/local/scratch/sas245/beetle-hnet/scratch/reps/exp32_onechunk_s0/th0.60", "/local/scratch/sas245/beetle-hnet/scratch/reps/exp32_onechunk_s0/th0.70"], "t": 4167.3}
{"event": "final", "run": "exp32_onechunk_s0", "condition": "A", "step": 11000, "params": 82470528, "per_lang": {"en": {"cpc": 1.000861920358559, "cps": 136.078125, "bpb": 1.7226399190689727, "bytes_per_chunk": 1024.0}, "de": {"cpc": 1.0160229873861588, "cps": 162.984375, "bpb": 1.8492320488028147, "bytes_per_chunk": 1024.0}, "vi": {"cpc": 1.3167344724708434, "cps": 189.640625, "bpb": 1.5293623978741624, "bytes_per_chunk": 1024.0}, "tr": {"cpc": 1.0905374799756946, "cps": 154.234375, "bpb": 1.7381207761526531, "bytes_per_chunk": 1024.0}, "fi": {"cpc": 1.040455052894583, "cps": 152.90625, "bpb": 1.9505399233928313, "bytes_per_chunk": 1024.0}, "ru": {"cpc": 1.8237467018469657, "cps": 270.0, "bpb": 1.151648274136964, "bytes_per_chunk": 1024.0}, "el": {"cpc": 1.8157189494713248, "cps": 291.125, "bpb": 1.1408944032079107, "bytes_per_chunk": 1024.0}, "hi": {"cpc": 2.566002971768202, "cps": 337.2890625, "bpb": 0.9616434054309644, "bytes_per_chunk": 1024.0}, "ta": {"cpc": 2.7234554693768387, "cps": 397.7734375, "bpb": 0.8163154212186341, "bytes_per_chunk": 1024.0}, "th": {"cpc": 2.8650917003419334, "cps": 360.0390625, "bpb": 0.9225549305688889, "bytes_per_chunk": 1024.0}, "zh": {"cpc": 2.7384016254656283, "cps": 126.3515625, "bpb": 2.683075474219554, "bytes_per_chunk": 1024.0}, "my": {"cpc": 2.8363415663922313, "cps": 419.8671875, "bpb": 0.9056775743602025, "bytes_per_chunk": 1024.0}}, "gini_cpc": 0.22151585453568945, "gini_cps": 0.2338532044751218, "bpic": 1.0, "seed": 0, "mean_boundary_entropy": 0.0010203896090388298, "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": 4178.9}