Blame for log.jsonl

{"event": "start", "run": "specfull_B_s0", "condition": "B", "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": "parity_a", "p3_objective": "none", "freeze_in_p3": true, "freeze_scope": "router"}, "corpus": "/local/scratch/sas245/beetle-hnet/scratch/data/tier1_corpus", "premium_kind": "flores_per_sent", "doc_filtered": {"my": 37511}, "weights": {"en": 0.2946, "de": 0.0909, "vi": 0.0576, "tr": 0.0612, "fi": 0.0424, "ru": 0.1476, "el": 0.0522, "hi": 0.042, "ta": 0.0313, "th": 0.0542, "zh": 0.1038, "my": 0.0222}, "t": 2.2}
{"event": "probe_schedule", "n": 30, "steps": [0, 1, 2, 4, 8, 16, 32, 64, 128, 256, 4000, 8000, 12000, 16000, 20000, 24000, 28000, 32000, 36000, 40000, 44000, 48000, 52000, 56000, 60000, 64000, 68000, 72000, 76000, 76300], "t": 2.2}
{"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": 8.248358}, "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": 8.204506}, "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": 8.13873}, "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": 8.143105}, "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": 8.234202}, "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": 8.071624}, "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": 8.142597}, "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": 7.880816}, "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": 7.814466}, "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": 8.063302}, "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": 8.034403}, "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": 8.24483}}, "bpb": {"en": 8.24836, "de": 8.20451, "vi": 8.13873, "tr": 8.1431, "fi": 8.2342, "ru": 8.07162, "el": 8.1426, "hi": 7.88082, "ta": 7.81447, "th": 8.0633, "zh": 8.0344, "my": 8.24483}, "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": 8.9}
{"event": "train", "step": 0, "phase": 1, "objective": "global", "lm_loss": 5.63905, "bpb": 8.13544, "chunk_loss": 0.057711, "F": 0.53552, "G": 0.50248, "chunks": 548.38, "bytes_per_chunk": 1.867, "boundary_entropy": 0.69172, "router_noise": 0.0, "grad_norm": 9.284, "lr": 1e-07, "router_trainable": true, "sec_per_step": 0.001, "alpha": 0.03, "t": 9.6}
{"event": "probe", "step": 1, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 8736.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.501982, "cps": 68.25, "mean_p": 0.503619, "boundary_entropy": 0.687415, "bpb": 8.238639}, "de": {"chunks": 12218.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.595042, "cps": 95.453125, "mean_p": 0.507339, "boundary_entropy": 0.688163, "bpb": 8.195135}, "vi": {"chunks": 6384.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.346298, "cps": 49.875, "mean_p": 0.491224, "boundary_entropy": 0.688506, "bpb": 8.133794}, "tr": {"chunks": 11394.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.629398, "cps": 89.015625, "mean_p": 0.506573, "boundary_entropy": 0.687851, "bpb": 8.136482}, "fi": {"chunks": 7121.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.378555, "cps": 55.632812, "mean_p": 0.497134, "boundary_entropy": 0.687919, "bpb": 8.226991}, "ru": {"chunks": 10479.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.552982, "cps": 81.867188, "mean_p": 0.496706, "boundary_entropy": 0.690192, "bpb": 8.060679}, "el": {"chunks": 36552.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.781026, "cps": 285.5625, "mean_p": 0.530186, "boundary_entropy": 0.688776, "bpb": 8.137899}, "hi": {"chunks": 42579.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.530698, "cps": 332.648438, "mean_p": 0.521058, "boundary_entropy": 0.690181, "bpb": 7.875022}, "ta": {"chunks": 49742.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.660711, "cps": 388.609375, "mean_p": 0.521036, "boundary_entropy": 0.690413, "bpb": 7.805885}, "th": {"chunks": 45716.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.842151, "cps": 357.15625, "mean_p": 0.52568, "boundary_entropy": 0.689897, "bpb": 8.054903}, "zh": {"chunks": 6220.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.053166, "cps": 48.59375, "mean_p": 0.496818, "boundary_entropy": 0.686551, "bpb": 8.033265}, "my": {"chunks": 53266.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.811167, "cps": 416.140625, "mean_p": 0.521288, "boundary_entropy": 0.690578, "bpb": 8.238213}}, "bpb": {"en": 8.23864, "de": 8.19513, "vi": 8.13379, "tr": 8.13648, "fi": 8.22699, "ru": 8.06068, "el": 8.1379, "hi": 7.87502, "ta": 7.80589, "th": 8.0549, "zh": 8.03326, "my": 8.23821}, "gini_cpc": 0.391259, "gini_cps": 0.409239, "bpic": 1.321528, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1853844026474671, "vi": 0.6898604243600176, "tr": 1.25382567427049, "fi": 0.7541202401739455, "ru": 1.1015954181043233, "el": 3.547985160578349, "hi": 5.0414083854481335, "ta": 5.300407605900385, "th": 5.661853836247682, "zh": 2.0980142656192435, "my": 5.600131222350586}, "t": 15.9}
{"event": "probe", "step": 2, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 8727.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.501465, "cps": 68.179688, "mean_p": 0.503592, "boundary_entropy": 0.687414, "bpb": 8.219818}, "de": {"chunks": 12199.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.594117, "cps": 95.304688, "mean_p": 0.507296, "boundary_entropy": 0.688163, "bpb": 8.177073}, "vi": {"chunks": 6374.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.345755, "cps": 49.796875, "mean_p": 0.491198, "boundary_entropy": 0.688504, "bpb": 8.124123}, "tr": {"chunks": 11377.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.628459, "cps": 88.882812, "mean_p": 0.506532, "boundary_entropy": 0.687851, "bpb": 8.123344}, "fi": {"chunks": 7095.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.377173, "cps": 55.429688, "mean_p": 0.497066, "boundary_entropy": 0.687917, "bpb": 8.212563}, "ru": {"chunks": 10498.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.553984, "cps": 82.015625, "mean_p": 0.496722, "boundary_entropy": 0.690193, "bpb": 8.040843}, "el": {"chunks": 36550.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.780929, "cps": 285.546875, "mean_p": 0.530101, "boundary_entropy": 0.688785, "bpb": 8.129261}, "hi": {"chunks": 42567.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.529985, "cps": 332.554688, "mean_p": 0.521008, "boundary_entropy": 0.690185, "bpb": 7.862067}, "ta": {"chunks": 49715.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.659267, "cps": 388.398438, "mean_p": 0.520913, "boundary_entropy": 0.690423, "bpb": 7.78995}, "th": {"chunks": 45715.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.842089, "cps": 357.148438, "mean_p": 0.525607, "boundary_entropy": 0.689904, "bpb": 8.034843}, "zh": {"chunks": 6211.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.051642, "cps": 48.523438, "mean_p": 0.496801, "boundary_entropy": 0.686551, "bpb": 8.030393}, "my": {"chunks": 53264.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.811062, "cps": 416.125, "mean_p": 0.521249, "boundary_entropy": 0.690581, "bpb": 8.227034}}, "bpb": {"en": 8.21982, "de": 8.17707, "vi": 8.12412, "tr": 8.12334, "fi": 8.21256, "ru": 8.04084, "el": 8.12926, "hi": 7.86207, "ta": 7.78995, "th": 8.03484, "zh": 8.03039, "my": 8.22703}, "gini_cpc": 0.391447, "gini_cps": 0.409407, "bpic": 1.322052, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1847615967434524, "vi": 0.6894901423087426, "tr": 1.2532460685272788, "fi": 0.7521416882696594, "ru": 1.1047308915738707, "el": 3.551449800839702, "hi": 5.045185219616122, "ta": 5.30299378922989, "th": 5.667568829353085, "zh": 2.097139067808873, "my": 5.605696050916938}, "t": 22.8}
{"event": "probe", "step": 4, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 8676.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.498535, "cps": 67.78125, "mean_p": 0.503456, "boundary_entropy": 0.687413, "bpb": 8.154813}, "de": {"chunks": 12100.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.589295, "cps": 94.53125, "mean_p": 0.50711, "boundary_entropy": 0.688165, "bpb": 8.115423}, "vi": {"chunks": 6334.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.343586, "cps": 49.484375, "mean_p": 0.491109, "boundary_entropy": 0.688499, "bpb": 8.089917}, "tr": {"chunks": 11316.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.62509, "cps": 88.40625, "mean_p": 0.506349, "boundary_entropy": 0.687853, "bpb": 8.077846}, "fi": {"chunks": 6997.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.371963, "cps": 54.664062, "mean_p": 0.496807, "boundary_entropy": 0.68791, "bpb": 8.16338}, "ru": {"chunks": 10501.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.554142, "cps": 82.039062, "mean_p": 0.496719, "boundary_entropy": 0.690193, "bpb": 7.980467}, "el": {"chunks": 36518.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.779369, "cps": 285.296875, "mean_p": 0.529728, "boundary_entropy": 0.688827, "bpb": 8.093503}, "hi": {"chunks": 42535.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.528083, "cps": 332.304688, "mean_p": 0.520796, "boundary_entropy": 0.690201, "bpb": 7.815118}, "ta": {"chunks": 49593.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.652741, "cps": 387.445312, "mean_p": 0.520486, "boundary_entropy": 0.690457, "bpb": 7.72484}, "th": {"chunks": 45710.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.841778, "cps": 357.109375, "mean_p": 0.525378, "boundary_entropy": 0.689926, "bpb": 7.966992}, "zh": {"chunks": 6200.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.04978, "cps": 48.4375, "mean_p": 0.496758, "boundary_entropy": 0.686551, "bpb": 8.021741}, "my": {"chunks": 53247.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.810165, "cps": 415.992188, "mean_p": 0.52105, "boundary_entropy": 0.690597, "bpb": 8.185565}}, "bpb": {"en": 8.15481, "de": 8.11542, "vi": 8.08992, "tr": 8.07785, "fi": 8.16338, "ru": 7.98047, "el": 8.0935, "hi": 7.81512, "ta": 7.72484, "th": 7.96699, "zh": 8.02174, "my": 8.18557}, "gini_cpc": 0.392361, "gini_cps": 0.410147, "bpic": 1.32463, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1820546070997677, "vi": 0.6891908357313893, "tr": 1.253853983991558, "fi": 0.7461129317746384, "ru": 1.1115423678092966, "el": 3.5691986166180385, "hi": 5.071027212436213, "ta": 5.321076319028209, "th": 5.700260889267812, "zh": 2.1057306749963347, "my": 5.636848270917479}, "t": 30.0}
{"event": "probe", "step": 8, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 8471.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.486755, "cps": 66.179688, "mean_p": 0.502941, "boundary_entropy": 0.687406, "bpb": 7.912078}, "de": {"chunks": 11727.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.571129, "cps": 91.617188, "mean_p": 0.506407, "boundary_entropy": 0.688172, "bpb": 7.886074}, "vi": {"chunks": 6201.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.336371, "cps": 48.445312, "mean_p": 0.490766, "boundary_entropy": 0.688479, "bpb": 7.966234}, "tr": {"chunks": 11067.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.611335, "cps": 86.460938, "mean_p": 0.505731, "boundary_entropy": 0.687855, "bpb": 7.906653}, "fi": {"chunks": 6568.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.349157, "cps": 51.3125, "mean_p": 0.495877, "boundary_entropy": 0.687882, "bpb": 7.976276}, "ru": {"chunks": 10788.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.569288, "cps": 84.28125, "mean_p": 0.496963, "boundary_entropy": 0.690198, "bpb": 7.747484}, "el": {"chunks": 36357.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.771525, "cps": 284.039062, "mean_p": 0.528387, "boundary_entropy": 0.688974, "bpb": 7.963803}, "hi": {"chunks": 42434.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.52208, "cps": 331.515625, "mean_p": 0.520125, "boundary_entropy": 0.690251, "bpb": 7.651281}, "ta": {"chunks": 49167.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.629955, "cps": 384.117188, "mean_p": 0.519094, "boundary_entropy": 0.690562, "bpb": 7.526497}, "th": {"chunks": 45689.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.840472, "cps": 356.945312, "mean_p": 0.524563, "boundary_entropy": 0.690004, "bpb": 7.761633}, "zh": {"chunks": 6149.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.041145, "cps": 48.039062, "mean_p": 0.4966, "boundary_entropy": 0.686548, "bpb": 7.992879}, "my": {"chunks": 53200.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.807684, "cps": 415.625, "mean_p": 0.520538, "boundary_entropy": 0.690636, "bpb": 8.077344}}, "bpb": {"en": 7.91208, "de": 7.88607, "vi": 7.96623, "tr": 7.90665, "fi": 7.97628, "ru": 7.74748, "el": 7.9638, "hi": 7.65128, "ta": 7.5265, "th": 7.76163, "zh": 7.99288, "my": 8.07734}, "gini_cpc": 0.395425, "gini_cps": 0.412669, "bpic": 1.333416, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.173340216439306, "vi": 0.6910477031722358, "tr": 1.2559397241794685, "fi": 0.7173162991001435, "ru": 1.1695563787548953, "el": 3.6394573428858488, "hi": 5.1814145174612865, "ta": 5.403033731906435, "th": 5.835526235590225, "zh": 2.1389492918669992, "my": 5.768165261370444}, "t": 38.1}
{"event": "probe", "step": 16, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 7675.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.441016, "cps": 59.960938, "mean_p": 0.500947, "boundary_entropy": 0.687374, "bpb": 7.191643}, "de": {"chunks": 10475.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.510154, "cps": 81.835938, "mean_p": 0.503874, "boundary_entropy": 0.688183, "bpb": 7.216534}, "vi": {"chunks": 5709.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.309683, "cps": 44.601562, "mean_p": 0.489413, "boundary_entropy": 0.688398, "bpb": 7.597759}, "tr": {"chunks": 10111.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.558526, "cps": 78.992188, "mean_p": 0.503386, "boundary_entropy": 0.687851, "bpb": 7.394062}, "fi": {"chunks": 5144.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.273457, "cps": 40.1875, "mean_p": 0.492336, "boundary_entropy": 0.687743, "bpb": 7.393305}, "ru": {"chunks": 12380.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.653298, "cps": 96.71875, "mean_p": 0.498378, "boundary_entropy": 0.690226, "bpb": 7.026703}, "el": {"chunks": 35600.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.734639, "cps": 278.125, "mean_p": 0.524231, "boundary_entropy": 0.689385, "bpb": 7.570992}, "hi": {"chunks": 41473.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.464963, "cps": 324.007812, "mean_p": 0.51687, "boundary_entropy": 0.690468, "bpb": 7.018444}, "ta": {"chunks": 45443.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.430757, "cps": 355.023438, "mean_p": 0.513431, "boundary_entropy": 0.69091, "bpb": 6.877481}, "th": {"chunks": 45452.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.825738, "cps": 355.09375, "mean_p": 0.520806, "boundary_entropy": 0.690327, "bpb": 6.926627}, "zh": {"chunks": 5982.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.012868, "cps": 46.734375, "mean_p": 0.496049, "boundary_entropy": 0.686542, "bpb": 7.878524}, "my": {"chunks": 53003.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.797287, "cps": 414.085938, "mean_p": 0.518661, "boundary_entropy": 0.690772, "bpb": 7.677635}}, "bpb": {"en": 7.19164, "de": 7.21653, "vi": 7.59776, "tr": 7.39406, "fi": 7.3933, "ru": 7.0267, "el": 7.57099, "hi": 7.01844, "ta": 6.87748, "th": 6.92663, "zh": 7.87852, "my": 7.67763}, "gini_cpc": 0.406571, "gini_cps": 0.421302, "bpic": 1.378291, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1567709150488028, "vi": 0.7022029297408184, "tr": 1.2664536354674059, "fi": 0.6200615656131617, "ru": 1.4813482419877444, "el": 3.9332802269512195, "hi": 5.589283195957621, "ta": 5.511721446536333, "th": 6.407338506043322, "zh": 2.296670553106155, "my": 6.342826202210198}, "t": 47.7}
{"event": "probe", "step": 32, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 4618.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.265357, "cps": 36.078125, "mean_p": 0.492826, "boundary_entropy": 0.68719, "bpb": 6.261066}, "de": {"chunks": 6466.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.314908, "cps": 50.515625, "mean_p": 0.495458, "boundary_entropy": 0.688144, "bpb": 6.338785}, "vi": {"chunks": 3133.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.169948, "cps": 24.476562, "mean_p": 0.480249, "boundary_entropy": 0.687716, "bpb": 6.963777}, "tr": {"chunks": 6244.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.344915, "cps": 48.78125, "mean_p": 0.493742, "boundary_entropy": 0.687704, "bpb": 6.645666}, "fi": {"chunks": 2724.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.144809, "cps": 21.28125, "mean_p": 0.48446, "boundary_entropy": 0.687308, "bpb": 6.519461}, "ru": {"chunks": 13223.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.697784, "cps": 103.304688, "mean_p": 0.498914, "boundary_entropy": 0.690282, "bpb": 5.792047}, "el": {"chunks": 24245.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.181358, "cps": 189.414062, "mean_p": 0.506776, "boundary_entropy": 0.690354, "bpb": 6.653986}, "hi": {"chunks": 29965.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 1.780981, "cps": 234.101562, "mean_p": 0.505575, "boundary_entropy": 0.6909, "bpb": 5.758075}, "ta": {"chunks": 13350.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.714095, "cps": 104.296875, "mean_p": 0.494952, "boundary_entropy": 0.691147, "bpb": 5.860218}, "th": {"chunks": 27264.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 1.694995, "cps": 213.0, "mean_p": 0.503154, "boundary_entropy": 0.691081, "bpb": 5.277261}, "zh": {"chunks": 5232.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.885879, "cps": 40.875, "mean_p": 0.493424, "boundary_entropy": 0.686472, "bpb": 7.509032}, "my": {"chunks": 46772.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.46844, "cps": 365.40625, "mean_p": 0.508741, "boundary_entropy": 0.691256, "bpb": 5.963717}}, "bpb": {"en": 6.26107, "de": 6.33879, "vi": 6.96378, "tr": 6.64567, "fi": 6.51946, "ru": 5.79205, "el": 6.65399, "hi": 5.75808, "ta": 5.86022, "th": 5.27726, "zh": 7.50903, "my": 5.96372}, "gini_cpc": 0.437929, "gini_cps": 0.458279, "bpic": 2.094463, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1867342722255276, "vi": 0.6404532657965205, "tr": 1.2998179633614275, "fi": 0.5457143969488473, "ru": 2.629607775470769, "el": 4.451962883460226, "hi": 6.711651545055623, "ta": 2.6910761533962884, "th": 6.387614520251808, "zh": 3.3384469875017544, "my": 9.30235173047285}, "t": 59.5}
{"event": "probe", "step": 64, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 6097.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.350342, "cps": 47.632812, "mean_p": 0.498514, "boundary_entropy": 0.68757, "bpb": 5.730427}, "de": {"chunks": 7690.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.374519, "cps": 60.078125, "mean_p": 0.49907, "boundary_entropy": 0.688481, "bpb": 5.858815}, "vi": {"chunks": 4697.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.254787, "cps": 36.695312, "mean_p": 0.488091, "boundary_entropy": 0.688451, "bpb": 6.351675}, "tr": {"chunks": 5658.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.312545, "cps": 44.203125, "mean_p": 0.494308, "boundary_entropy": 0.687961, "bpb": 6.228659}, "fi": {"chunks": 4811.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.255755, "cps": 37.585938, "mean_p": 0.493755, "boundary_entropy": 0.688014, "bpb": 5.929839}, "ru": {"chunks": 7199.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.379894, "cps": 56.242188, "mean_p": 0.494806, "boundary_entropy": 0.690309, "bpb": 5.075498}, "el": {"chunks": 3840.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.187107, "cps": 30.0, "mean_p": 0.487443, "boundary_entropy": 0.690092, "bpb": 5.447362}, "hi": {"chunks": 6923.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.411471, "cps": 54.085938, "mean_p": 0.493586, "boundary_entropy": 0.690817, "bpb": 4.870818}, "ta": {"chunks": 18043.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.965124, "cps": 140.960938, "mean_p": 0.497875, "boundary_entropy": 0.691217, "bpb": 4.733414}, "th": {"chunks": 8169.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.507864, "cps": 63.820312, "mean_p": 0.493956, "boundary_entropy": 0.690963, "bpb": 4.715935}, "zh": {"chunks": 2778.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.470369, "cps": 21.703125, "mean_p": 0.485095, "boundary_entropy": 0.68618, "bpb": 6.97186}, "my": {"chunks": 9662.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.509922, "cps": 75.484375, "mean_p": 0.494472, "boundary_entropy": 0.691297, "bpb": 4.587409}}, "bpb": {"en": 5.73043, "de": 5.85882, "vi": 6.35168, "tr": 6.22866, "fi": 5.92984, "ru": 5.0755, "el": 5.44736, "hi": 4.87082, "ta": 4.73341, "th": 4.71593, "zh": 6.97186, "my": 4.58741}, "gini_cpc": 0.227541, "gini_cps": 0.257557, "bpic": 4.485152, "worst_lang": "ta", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0690102215357158, "vi": 0.7272527018783531, "tr": 0.8921139220199078, "fi": 0.7300143524320073, "ru": 1.0843534970916533, "el": 0.5340701489252696, "hi": 1.1744842139741343, "ta": 2.7548071708502504, "th": 1.4496252864464652, "zh": 1.3426002506825956, "my": 1.4554978969401473}, "t": 73.8}
{"event": "probe", "step": 128, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 5038.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.28949, "cps": 39.359375, "mean_p": 0.491002, "boundary_entropy": 0.686757, "bpb": 5.282642}, "de": {"chunks": 5580.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.271758, "cps": 43.59375, "mean_p": 0.489435, "boundary_entropy": 0.687589, "bpb": 5.44261}, "vi": {"chunks": 6066.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.329048, "cps": 47.390625, "mean_p": 0.491103, "boundary_entropy": 0.68861, "bpb": 5.811557}, "tr": {"chunks": 3514.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.194111, "cps": 27.453125, "mean_p": 0.485459, "boundary_entropy": 0.687262, "bpb": 5.73805}, "fi": {"chunks": 3419.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.181755, "cps": 26.710938, "mean_p": 0.486508, "boundary_entropy": 0.687422, "bpb": 5.445642}, "ru": {"chunks": 9787.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.516464, "cps": 76.460938, "mean_p": 0.495987, "boundary_entropy": 0.690305, "bpb": 4.538728}, "el": {"chunks": 7427.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.361887, "cps": 58.023438, "mean_p": 0.494421, "boundary_entropy": 0.690475, "bpb": 4.998678}, "hi": {"chunks": 9388.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.557979, "cps": 73.34375, "mean_p": 0.490857, "boundary_entropy": 0.690567, "bpb": 4.101628}, "ta": {"chunks": 13395.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.716502, "cps": 104.648438, "mean_p": 0.47907, "boundary_entropy": 0.68941, "bpb": 4.122987}, "th": {"chunks": 9953.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.618775, "cps": 77.757812, "mean_p": 0.483535, "boundary_entropy": 0.689985, "bpb": 4.040975}, "zh": {"chunks": 3336.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.564849, "cps": 26.0625, "mean_p": 0.491329, "boundary_entropy": 0.686832, "bpb": 6.647521}, "my": {"chunks": 9184.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.484695, "cps": 71.75, "mean_p": 0.493812, "boundary_entropy": 0.691274, "bpb": 4.138499}}, "bpb": {"en": 5.28264, "de": 5.44261, "vi": 5.81156, "tr": 5.73805, "fi": 5.44564, "ru": 4.53873, "el": 4.99868, "hi": 4.10163, "ta": 4.12299, "th": 4.04097, "zh": 6.64752, "my": 4.1385}, "gini_cpc": 0.225779, "gini_cps": 0.239379, "bpic": 4.45806, "worst_lang": "ta", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9387452420001791, "vi": 1.1366459819933195, "tr": 0.6705283780735339, "fi": 0.6278460617071144, "ru": 1.7840471624100114, "el": 1.2500820978804035, "hi": 1.927453746769721, "ta": 2.475045604197083, "th": 2.137464427947845, "zh": 1.9511854840563565, "my": 1.674304544486601}, "t": 94.9}
{"event": "probe", "step": 256, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3489.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.200483, "cps": 27.257812, "mean_p": 0.448327, "boundary_entropy": 0.618596, "bpb": 4.174585}, "de": {"chunks": 3624.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.176496, "cps": 28.3125, "mean_p": 0.444544, "boundary_entropy": 0.631603, "bpb": 4.424108}, "vi": {"chunks": 7050.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.382425, "cps": 55.078125, "mean_p": 0.493302, "boundary_entropy": 0.653747, "bpb": 4.152446}, "tr": {"chunks": 3496.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.193117, "cps": 27.3125, "mean_p": 0.469715, "boundary_entropy": 0.66018, "bpb": 4.511}, "fi": {"chunks": 2893.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.153793, "cps": 22.601562, "mean_p": 0.457091, "boundary_entropy": 0.658783, "bpb": 4.432589}, "ru": {"chunks": 5948.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.313879, "cps": 46.46875, "mean_p": 0.477069, "boundary_entropy": 0.68558, "bpb": 3.114926}, "el": {"chunks": 6135.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.298933, "cps": 47.929688, "mean_p": 0.473907, "boundary_entropy": 0.688005, "bpb": 3.347145}, "hi": {"chunks": 8087.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.480654, "cps": 63.179688, "mean_p": 0.471094, "boundary_entropy": 0.68776, "bpb": 2.572477}, "ta": {"chunks": 11862.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.634501, "cps": 92.671875, "mean_p": 0.433928, "boundary_entropy": 0.676535, "bpb": 2.554689}, "th": {"chunks": 9377.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.582965, "cps": 73.257812, "mean_p": 0.453683, "boundary_entropy": 0.683901, "bpb": 2.537863}, "zh": {"chunks": 2189.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.37064, "cps": 17.101562, "mean_p": 0.482813, "boundary_entropy": 0.685935, "bpb": 5.899366}, "my": {"chunks": 11203.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.59125, "cps": 87.523438, "mean_p": 0.458165, "boundary_entropy": 0.685379, "bpb": 2.647921}}, "bpb": {"en": 4.17458, "de": 4.42411, "vi": 4.15245, "tr": 4.511, "fi": 4.43259, "ru": 3.11493, "el": 3.34714, "hi": 2.57248, "ta": 2.55469, "th": 2.53786, "zh": 5.89937, "my": 2.64792}, "gini_cpc": 0.256093, "gini_cps": 0.282923, "bpic": 5.093108, "worst_lang": "ta", "cpc_parity_vs_en": {"en": 1.0, "de": 0.8803572245906272, "vi": 1.9075201126124395, "tr": 0.9632611023127478, "fi": 0.7671136329052818, "ru": 1.5656147212762916, "el": 1.491066018290659, "hi": 2.3974829148880987, "ta": 3.1648679980048096, "th": 2.9078098376236947, "zh": 1.8487384326357998, "my": 2.949131314902478}, "t": 124.7}
{"event": "train", "step": 500, "phase": 1, "objective": "global", "lm_loss": 2.25723, "bpb": 3.25649, "chunk_loss": 0.030669, "F": 0.17984, "G": 0.40528, "chunks": 184.16, "bytes_per_chunk": 5.56, "boundary_entropy": 0.58259, "router_noise": 0.0, "grad_norm": 1.7243, "lr": 7.51e-05, "router_trainable": true, "sec_per_step": 0.32, "alpha": 0.03, "t": 169.8}
{"event": "train", "step": 1000, "phase": 1, "objective": "global", "lm_loss": 1.74674, "bpb": 2.52001, "chunk_loss": 0.030328, "F": 0.18355, "G": 0.29263, "chunks": 187.95, "bytes_per_chunk": 5.448, "boundary_entropy": 0.38036, "router_noise": 0.0, "grad_norm": 1.5388, "lr": 0.0001501, "router_trainable": true, "sec_per_step": 0.174, "alpha": 0.03, "t": 256.6}
{"event": "train", "step": 1500, "phase": 1, "objective": "global", "lm_loss": 1.58858, "bpb": 2.29183, "chunk_loss": 0.030027, "F": 0.17654, "G": 0.27129, "chunks": 180.78, "bytes_per_chunk": 5.664, "boundary_entropy": 0.34708, "router_noise": 0.0, "grad_norm": 0.8645, "lr": 0.0002251, "router_trainable": true, "sec_per_step": 0.172, "alpha": 0.03, "t": 342.6}
{"event": "train", "step": 2000, "phase": 1, "objective": "global", "lm_loss": 1.47334, "bpb": 2.12558, "chunk_loss": 0.030425, "F": 0.19386, "G": 0.24846, "chunks": 198.52, "bytes_per_chunk": 5.158, "boundary_entropy": 0.26185, "router_noise": 0.0, "grad_norm": 0.6801, "lr": 0.0003, "router_trainable": true, "sec_per_step": 0.168, "alpha": 0.03, "t": 426.6}
{"event": "train", "step": 2500, "phase": 1, "objective": "global", "lm_loss": 1.34487, "bpb": 1.94023, "chunk_loss": 0.030145, "F": 0.18248, "G": 0.24979, "chunks": 186.86, "bytes_per_chunk": 5.48, "boundary_entropy": 0.2746, "router_noise": 0.0, "grad_norm": 0.495, "lr": 0.0003, "router_trainable": true, "sec_per_step": 0.167, "alpha": 0.03, "t": 510.2}
{"event": "train", "step": 3000, "phase": 1, "objective": "global", "lm_loss": 1.3866, "bpb": 2.00045, "chunk_loss": 0.030259, "F": 0.18936, "G": 0.23937, "chunks": 193.91, "bytes_per_chunk": 5.281, "boundary_entropy": 0.2206, "router_noise": 0.0, "grad_norm": 0.4413, "lr": 0.0002999, "router_trainable": true, "sec_per_step": 0.168, "alpha": 0.03, "t": 594.0}
{"event": "train", "step": 3500, "phase": 1, "objective": "global", "lm_loss": 1.36504, "bpb": 1.96934, "chunk_loss": 0.030281, "F": 0.18665, "G": 0.23478, "chunks": 191.12, "bytes_per_chunk": 5.358, "boundary_entropy": 0.21126, "router_noise": 0.0, "grad_norm": 0.394, "lr": 0.0002997, "router_trainable": true, "sec_per_step": 0.167, "alpha": 0.03, "t": 677.7}
{"event": "probe", "step": 4000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3533.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.203011, "cps": 27.601562, "mean_p": 0.20518, "boundary_entropy": 0.08747, "bpb": 1.93968}, "de": {"chunks": 3826.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.186334, "cps": 29.890625, "mean_p": 0.19651, "boundary_entropy": 0.098306, "bpb": 2.131187}, "vi": {"chunks": 4265.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.231353, "cps": 33.320312, "mean_p": 0.190146, "boundary_entropy": 0.109691, "bpb": 1.774818}, "tr": {"chunks": 3115.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.172071, "cps": 24.335938, "mean_p": 0.188959, "boundary_entropy": 0.172347, "bpb": 2.120485}, "fi": {"chunks": 3143.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.167083, "cps": 24.554688, "mean_p": 0.181433, "boundary_entropy": 0.130254, "bpb": 2.287335}, "ru": {"chunks": 5828.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.307546, "cps": 45.53125, "mean_p": 0.20832, "boundary_entropy": 0.24558, "bpb": 1.310756}, "el": {"chunks": 6544.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.318862, "cps": 51.125, "mean_p": 0.314457, "boundary_entropy": 0.454335, "bpb": 1.344498}, "hi": {"chunks": 7257.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.431322, "cps": 56.695312, "mean_p": 0.407069, "boundary_entropy": 0.565941, "bpb": 1.087412}, "ta": {"chunks": 10805.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.577962, "cps": 84.414062, "mean_p": 0.425792, "boundary_entropy": 0.601384, "bpb": 0.921182}, "th": {"chunks": 8108.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.504072, "cps": 63.34375, "mean_p": 0.419959, "boundary_entropy": 0.592459, "bpb": 1.052584}, "zh": {"chunks": 2421.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.409922, "cps": 18.914062, "mean_p": 0.250892, "boundary_entropy": 0.364147, "bpb": 2.987948}, "my": {"chunks": 12657.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.667986, "cps": 98.882812, "mean_p": 0.37013, "boundary_entropy": 0.57746, "bpb": 1.074538}}, "bpb": {"en": 1.93968, "de": 2.13119, "vi": 1.77482, "tr": 2.12049, "fi": 2.28733, "ru": 1.31076, "el": 1.3445, "hi": 1.08741, "ta": 0.92118, "th": 1.05258, "zh": 2.98795, "my": 1.07454}, "gini_cpc": 0.260834, "gini_cps": 0.285421, "bpic": 5.367416, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9178528088320033, "vi": 1.1396103275475196, "tr": 0.8475942119361988, "fi": 0.8230249107637561, "ru": 1.5149238801589535, "el": 1.570662692802297, "hi": 2.124626201103667, "ta": 2.846949637040801, "th": 2.482979634884412, "zh": 2.0192115862926197, "my": 3.2903938653502567}, "t": 764.1}
{"event": "eval", "step": 4000, "phase": 1, "bpb": {"en": 1.94792, "de": 2.12542, "vi": 1.78315, "tr": 2.11219, "fi": 2.30326, "ru": 1.30898, "el": 1.34188, "hi": 1.09706, "ta": 0.92248, "th": 1.05004, "zh": 2.99582, "my": 1.08281}, "bytes_per_chunk": {"en": 4.728, "de": 5.308, "vi": 5.555, "tr": 6.087, "fi": 5.741, "ru": 5.878, "el": 5.641, "hi": 5.979, "ta": 4.8, "th": 5.694, "zh": 6.54, "my": 4.362}, "bpb_mean": 1.67258, "t": 769.5}
{"event": "train", "step": 4000, "phase": 1, "objective": "global", "lm_loss": 1.35874, "bpb": 1.96025, "chunk_loss": 0.03035, "F": 0.18332, "G": 0.24994, "chunks": 187.72, "bytes_per_chunk": 5.455, "boundary_entropy": 0.25714, "router_noise": 0.0, "grad_norm": 0.3903, "lr": 0.0002995, "router_trainable": true, "sec_per_step": 0.184, "alpha": 0.03, "t": 769.7}
{"event": "train", "step": 4500, "phase": 1, "objective": "global", "lm_loss": 1.26753, "bpb": 1.82866, "chunk_loss": 0.030063, "F": 0.17686, "G": 0.24498, "chunks": 181.11, "bytes_per_chunk": 5.654, "boundary_entropy": 0.24154, "router_noise": 0.0, "grad_norm": 0.3701, "lr": 0.0002992, "router_trainable": true, "sec_per_step": 0.168, "alpha": 0.03, "t": 853.6}
{"event": "train", "step": 5000, "phase": 1, "objective": "global", "lm_loss": 1.22229, "bpb": 1.76339, "chunk_loss": 0.030102, "F": 0.17891, "G": 0.24124, "chunks": 183.2, "bytes_per_chunk": 5.589, "boundary_entropy": 0.2333, "router_noise": 0.0, "grad_norm": 0.3894, "lr": 0.0002989, "router_trainable": true, "sec_per_step": 0.168, "alpha": 0.03, "t": 937.6}
{"event": "train", "step": 5500, "phase": 1, "objective": "global", "lm_loss": 1.18956, "bpb": 1.71618, "chunk_loss": 0.030363, "F": 0.18729, "G": 0.26246, "chunks": 191.78, "bytes_per_chunk": 5.339, "boundary_entropy": 0.25957, "router_noise": 0.0, "grad_norm": 0.3513, "lr": 0.0002985, "router_trainable": true, "sec_per_step": 0.167, "alpha": 0.03, "t": 1021.1}
{"event": "train", "step": 6000, "phase": 1, "objective": "global", "lm_loss": 1.14898, "bpb": 1.65763, "chunk_loss": 0.030379, "F": 0.18941, "G": 0.2573, "chunks": 193.95, "bytes_per_chunk": 5.28, "boundary_entropy": 0.23035, "router_noise": 0.0, "grad_norm": 0.3517, "lr": 0.0002981, "router_trainable": true, "sec_per_step": 0.167, "alpha": 0.03, "t": 1104.6}
{"event": "train", "step": 6500, "phase": 1, "objective": "global", "lm_loss": 1.08588, "bpb": 1.5666, "chunk_loss": 0.030227, "F": 0.18253, "G": 0.26005, "chunks": 186.91, "bytes_per_chunk": 5.479, "boundary_entropy": 0.24889, "router_noise": 0.0, "grad_norm": 0.313, "lr": 0.0002976, "router_trainable": true, "sec_per_step": 0.168, "alpha": 0.03, "t": 1188.7}
{"event": "train", "step": 7000, "phase": 1, "objective": "global", "lm_loss": 1.10327, "bpb": 1.59168, "chunk_loss": 0.030103, "F": 0.17981, "G": 0.2489, "chunks": 184.12, "bytes_per_chunk": 5.561, "boundary_entropy": 0.23162, "router_noise": 0.0, "grad_norm": 0.325, "lr": 0.000297, "router_trainable": true, "sec_per_step": 0.167, "alpha": 0.03, "t": 1272.3}
{"event": "train", "step": 7500, "phase": 1, "objective": "global", "lm_loss": 1.10214, "bpb": 1.59005, "chunk_loss": 0.030101, "F": 0.1777, "G": 0.24575, "chunks": 181.97, "bytes_per_chunk": 5.627, "boundary_entropy": 0.23433, "router_noise": 0.0, "grad_norm": 0.3142, "lr": 0.0002964, "router_trainable": true, "sec_per_step": 0.167, "alpha": 0.03, "t": 1355.8}
{"event": "probe", "step": 8000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3503.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.201287, "cps": 27.367188, "mean_p": 0.196465, "boundary_entropy": 0.071577, "bpb": 1.704661}, "de": {"chunks": 3772.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.183704, "cps": 29.46875, "mean_p": 0.187586, "boundary_entropy": 0.064267, "bpb": 1.851211}, "vi": {"chunks": 4207.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.228207, "cps": 32.867188, "mean_p": 0.177312, "boundary_entropy": 0.082615, "bpb": 1.480552}, "tr": {"chunks": 3058.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.168922, "cps": 23.890625, "mean_p": 0.191743, "boundary_entropy": 0.176188, "bpb": 1.779883}, "fi": {"chunks": 3020.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.160544, "cps": 23.59375, "mean_p": 0.172345, "boundary_entropy": 0.111058, "bpb": 1.976921}, "ru": {"chunks": 5645.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.297889, "cps": 44.101562, "mean_p": 0.207087, "boundary_entropy": 0.208083, "bpb": 1.139868}, "el": {"chunks": 6320.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.307947, "cps": 49.375, "mean_p": 0.341987, "boundary_entropy": 0.475577, "bpb": 1.165215}, "hi": {"chunks": 6935.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.412184, "cps": 54.179688, "mean_p": 0.438177, "boundary_entropy": 0.58396, "bpb": 0.946202}, "ta": {"chunks": 8181.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.437604, "cps": 63.914062, "mean_p": 0.425405, "boundary_entropy": 0.604457, "bpb": 0.80749}, "th": {"chunks": 6994.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.434815, "cps": 54.640625, "mean_p": 0.434626, "boundary_entropy": 0.603108, "bpb": 0.920756}, "zh": {"chunks": 2426.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.410769, "cps": 18.953125, "mean_p": 0.245739, "boundary_entropy": 0.33942, "bpb": 2.59992}, "my": {"chunks": 7577.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.399884, "cps": 59.195312, "mean_p": 0.399371, "boundary_entropy": 0.628765, "bpb": 0.928212}}, "bpb": {"en": 1.70466, "de": 1.85121, "vi": 1.48055, "tr": 1.77988, "fi": 1.97692, "ru": 1.13987, "el": 1.16522, "hi": 0.9462, "ta": 0.80749, "th": 0.92076, "zh": 2.59992, "my": 0.92821}, "gini_cpc": 0.19697, "gini_cps": 0.213764, "bpic": 6.22637, "worst_lang": "ta", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9126479020097836, "vi": 1.1337396958599013, "tr": 0.8392105068076449, "fi": 0.7975887918423106, "ru": 1.47992162014165, "el": 1.5298900355754124, "hi": 2.0477426480974277, "ta": 2.174026862849457, "th": 2.160173060064596, "zh": 2.0407102073700267, "my": 1.9866341380779744}, "t": 1441.7}
{"event": "eval", "step": 8000, "phase": 1, "bpb": {"en": 1.70524, "de": 1.85719, "vi": 1.49831, "tr": 1.77158, "fi": 1.98778, "ru": 1.13966, "el": 1.16368, "hi": 0.95171, "ta": 0.81165, "th": 0.92138, "zh": 2.60328, "my": 0.93333}, "bytes_per_chunk": {"en": 4.775, "de": 5.363, "vi": 5.603, "tr": 6.192, "fi": 5.94, "ru": 6.0, "el": 5.545, "hi": 6.113, "ta": 6.119, "th": 6.373, "zh": 6.385, "my": 6.386}, "bpb_mean": 1.4454, "t": 1447.0}
{"event": "train", "step": 8000, "phase": 1, "objective": "global", "lm_loss": 1.11846, "bpb": 1.61359, "chunk_loss": 0.030082, "F": 0.18228, "G": 0.23612, "chunks": 186.66, "bytes_per_chunk": 5.486, "boundary_entropy": 0.19766, "router_noise": 0.0, "grad_norm": 0.3058, "lr": 0.0002957, "router_trainable": true, "sec_per_step": 0.183, "alpha": 0.03, "t": 1447.2}
{"event": "train", "step": 8500, "phase": 1, "objective": "global", "lm_loss": 1.10173, "bpb": 1.58947, "chunk_loss": 0.030135, "F": 0.18085, "G": 0.23988, "chunks": 185.19, "bytes_per_chunk": 5.53, "boundary_entropy": 0.21856, "router_noise": 0.0, "grad_norm": 0.3098, "lr": 0.0002949, "router_trainable": true, "sec_per_step": 0.167, "alpha": 0.03, "t": 1530.8}
{"event": "train", "step": 9000, "phase": 1, "objective": "global", "lm_loss": 1.08768, "bpb": 1.56918, "chunk_loss": 0.030091, "F": 0.17422, "G": 0.24313, "chunks": 178.41, "bytes_per_chunk": 5.74, "boundary_entropy": 0.23297, "router_noise": 0.0, "grad_norm": 0.3104, "lr": 0.0002941, "router_trainable": true, "sec_per_step": 0.167, "alpha": 0.03, "t": 1614.4}
{"event": "train", "step": 9500, "phase": 1, "objective": "global", "lm_loss": 1.13064, "bpb": 1.63117, "chunk_loss": 0.030111, "F": 0.18201, "G": 0.23453, "chunks": 186.38, "bytes_per_chunk": 5.494, "boundary_entropy": 0.20399, "router_noise": 0.0, "grad_norm": 0.3031, "lr": 0.0002933, "router_trainable": true, "sec_per_step": 0.167, "alpha": 0.03, "t": 1697.9}
{"event": "train", "step": 10000, "phase": 1, "objective": "global", "lm_loss": 1.06735, "bpb": 1.53986, "chunk_loss": 0.030104, "F": 0.18008, "G": 0.25417, "chunks": 184.41, "bytes_per_chunk": 5.553, "boundary_entropy": 0.25407, "router_noise": 0.0, "grad_norm": 0.2919, "lr": 0.0002924, "router_trainable": true, "sec_per_step": 0.167, "alpha": 0.03, "t": 1781.6}
{"event": "train", "step": 10500, "phase": 1, "objective": "global", "lm_loss": 1.05902, "bpb": 1.52785, "chunk_loss": 0.029995, "F": 0.18048, "G": 0.24105, "chunks": 184.81, "bytes_per_chunk": 5.541, "boundary_entropy": 0.21231, "router_noise": 0.0, "grad_norm": 0.2776, "lr": 0.0002914, "router_trainable": true, "sec_per_step": 0.168, "alpha": 0.03, "t": 1865.6}
{"event": "train", "step": 11000, "phase": 1, "objective": "global", "lm_loss": 1.09935, "bpb": 1.58603, "chunk_loss": 0.030148, "F": 0.18341, "G": 0.2364, "chunks": 187.81, "bytes_per_chunk": 5.452, "boundary_entropy": 0.19741, "router_noise": 0.0, "grad_norm": 0.2927, "lr": 0.0002903, "router_trainable": true, "sec_per_step": 0.166, "alpha": 0.03, "t": 1948.8}
{"event": "train", "step": 11500, "phase": 1, "objective": "global", "lm_loss": 1.00113, "bpb": 1.44433, "chunk_loss": 0.029972, "F": 0.17761, "G": 0.24703, "chunks": 181.88, "bytes_per_chunk": 5.63, "boundary_entropy": 0.22716, "router_noise": 0.0, "grad_norm": 0.2718, "lr": 0.0002893, "router_trainable": true, "sec_per_step": 0.167, "alpha": 0.03, "t": 2032.3}
{"event": "probe", "step": 12000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3479.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.199908, "cps": 27.179688, "mean_p": 0.194246, "boundary_entropy": 0.069311, "bpb": 1.639525}, "de": {"chunks": 3736.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.181951, "cps": 29.1875, "mean_p": 0.184832, "boundary_entropy": 0.059217, "bpb": 1.780938}, "vi": {"chunks": 4234.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.229672, "cps": 33.078125, "mean_p": 0.178324, "boundary_entropy": 0.074622, "bpb": 1.406342}, "tr": {"chunks": 3125.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.172623, "cps": 24.414062, "mean_p": 0.206616, "boundary_entropy": 0.200921, "bpb": 1.700017}, "fi": {"chunks": 2963.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.157514, "cps": 23.148438, "mean_p": 0.170892, "boundary_entropy": 0.110475, "bpb": 1.880287}, "ru": {"chunks": 5626.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.296887, "cps": 43.953125, "mean_p": 0.222342, "boundary_entropy": 0.232947, "bpb": 1.084347}, "el": {"chunks": 5644.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.275009, "cps": 44.09375, "mean_p": 0.317051, "boundary_entropy": 0.444445, "bpb": 1.09354}, "hi": {"chunks": 7264.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.431738, "cps": 56.75, "mean_p": 0.431777, "boundary_entropy": 0.57541, "bpb": 0.888041}, "ta": {"chunks": 8646.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.462477, "cps": 67.546875, "mean_p": 0.416391, "boundary_entropy": 0.598881, "bpb": 0.770018}, "th": {"chunks": 7636.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.474728, "cps": 59.65625, "mean_p": 0.434704, "boundary_entropy": 0.597143, "bpb": 0.866463}, "zh": {"chunks": 2657.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.449881, "cps": 20.757812, "mean_p": 0.259967, "boundary_entropy": 0.352339, "bpb": 2.513866}, "my": {"chunks": 8714.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.45989, "cps": 68.078125, "mean_p": 0.401539, "boundary_entropy": 0.625407, "bpb": 0.876964}}, "bpb": {"en": 1.63953, "de": 1.78094, "vi": 1.40634, "tr": 1.70002, "fi": 1.88029, "ru": 1.08435, "el": 1.09354, "hi": 0.88804, "ta": 0.77002, "th": 0.86646, "zh": 2.51387, "my": 0.87696}, "gini_cpc": 0.218021, "gini_cps": 0.229364, "bpic": 6.02255, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9101734268941672, "vi": 1.148887232496455, "tr": 0.863513559795655, "fi": 0.787933307752986, "ru": 1.4851154116557619, "el": 1.3756750203226165, "hi": 2.1596852095378467, "ta": 2.3134464602923392, "th": 2.374731679744974, "zh": 2.2504418898860727, "my": 2.3005086522027307}, "t": 2118.7}
{"event": "eval", "step": 12000, "phase": 1, "bpb": {"en": 1.64091, "de": 1.7872, "vi": 1.42275, "tr": 1.6921, "fi": 1.89418, "ru": 1.08255, "el": 1.09182, "hi": 0.89418, "ta": 0.77413, "th": 0.86567, "zh": 2.51833, "my": 0.88212}, "bytes_per_chunk": {"en": 4.799, "de": 5.423, "vi": 5.58, "tr": 6.134, "fi": 5.973, "ru": 6.036, "el": 6.357, "hi": 5.989, "ta": 5.841, "th": 5.856, "zh": 6.072, "my": 5.787}, "bpb_mean": 1.37883, "t": 2123.9}
{"event": "train", "step": 12000, "phase": 1, "objective": "global", "lm_loss": 0.97618, "bpb": 1.40833, "chunk_loss": 0.03012, "F": 0.18033, "G": 0.2532, "chunks": 184.66, "bytes_per_chunk": 5.545, "boundary_entropy": 0.23671, "router_noise": 0.0, "grad_norm": 0.2736, "lr": 0.0002881, "router_trainable": true, "sec_per_step": 0.184, "alpha": 0.03, "t": 2124.1}
{"event": "train", "step": 12500, "phase": 1, "objective": "global", "lm_loss": 0.98903, "bpb": 1.42687, "chunk_loss": 0.030047, "F": 0.18344, "G": 0.24815, "chunks": 187.84, "bytes_per_chunk": 5.451, "boundary_entropy": 0.21675, "router_noise": 0.0, "grad_norm": 0.2621, "lr": 0.0002869, "router_trainable": true, "sec_per_step": 0.168, "alpha": 0.03, "t": 2207.9}
{"event": "train", "step": 13000, "phase": 1, "objective": "global", "lm_loss": 1.08505, "bpb": 1.5654, "chunk_loss": 0.030282, "F": 0.18718, "G": 0.24526, "chunks": 191.67, "bytes_per_chunk": 5.342, "boundary_entropy": 0.21554, "router_noise": 0.0, "grad_norm": 0.2875, "lr": 0.0002857, "router_trainable": true, "sec_per_step": 0.167, "alpha": 0.03, "t": 2291.4}
{"event": "train", "step": 13500, "phase": 1, "objective": "global", "lm_loss": 0.9804, "bpb": 1.41442, "chunk_loss": 0.030003, "F": 0.17584, "G": 0.25277, "chunks": 180.06, "bytes_per_chunk": 5.687, "boundary_entropy": 0.2404, "router_noise": 0.0, "grad_norm": 0.2936, "lr": 0.0002844, "router_trainable": true, "sec_per_step": 0.167, "alpha": 0.03, "t": 2375.1}
{"event": "train", "step": 14000, "phase": 1, "objective": "global", "lm_loss": 1.03831, "bpb": 1.49797, "chunk_loss": 0.029985, "F": 0.17642, "G": 0.24803, "chunks": 180.66, "bytes_per_chunk": 5.668, "boundary_entropy": 0.24142, "router_noise": 0.0, "grad_norm": 0.2764, "lr": 0.000283, "router_trainable": true, "sec_per_step": 0.167, "alpha": 0.03, "t": 2458.4}
{"event": "train", "step": 14500, "phase": 1, "objective": "global", "lm_loss": 0.99652, "bpb": 1.43767, "chunk_loss": 0.029993, "F": 0.17853, "G": 0.23815, "chunks": 182.81, "bytes_per_chunk": 5.601, "boundary_entropy": 0.2193, "router_noise": 0.0, "grad_norm": 0.2844, "lr": 0.0002816, "router_trainable": true, "sec_per_step": 0.167, "alpha": 0.03, "t": 2541.9}
{"event": "train", "step": 15000, "phase": 1, "objective": "global", "lm_loss": 1.02723, "bpb": 1.48197, "chunk_loss": 0.030174, "F": 0.17949, "G": 0.25235, "chunks": 183.8, "bytes_per_chunk": 5.571, "boundary_entropy": 0.24115, "router_noise": 0.0, "grad_norm": 0.285, "lr": 0.0002801, "router_trainable": true, "sec_per_step": 0.167, "alpha": 0.03, "t": 2625.6}
{"event": "train", "step": 15500, "phase": 1, "objective": "global", "lm_loss": 0.96882, "bpb": 1.39771, "chunk_loss": 0.030122, "F": 0.18211, "G": 0.2696, "chunks": 186.48, "bytes_per_chunk": 5.491, "boundary_entropy": 0.26273, "router_noise": 0.0, "grad_norm": 0.2813, "lr": 0.0002786, "router_trainable": true, "sec_per_step": 0.167, "alpha": 0.03, "t": 2709.2}
{"event": "probe", "step": 16000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3470.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.199391, "cps": 27.109375, "mean_p": 0.191992, "boundary_entropy": 0.067559, "bpb": 1.601837}, "de": {"chunks": 3734.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.181854, "cps": 29.171875, "mean_p": 0.182438, "boundary_entropy": 0.057021, "bpb": 1.732068}, "vi": {"chunks": 4249.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.230485, "cps": 33.195312, "mean_p": 0.177059, "boundary_entropy": 0.072678, "bpb": 1.361665}, "tr": {"chunks": 3073.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.169751, "cps": 24.007812, "mean_p": 0.209482, "boundary_entropy": 0.209447, "bpb": 1.654359}, "fi": {"chunks": 2946.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.15661, "cps": 23.015625, "mean_p": 0.175054, "boundary_entropy": 0.123518, "bpb": 1.823469}, "ru": {"chunks": 5605.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.295778, "cps": 43.789062, "mean_p": 0.239608, "boundary_entropy": 0.27073, "bpb": 1.052567}, "el": {"chunks": 5940.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.289431, "cps": 46.40625, "mean_p": 0.337263, "boundary_entropy": 0.465756, "bpb": 1.060712}, "hi": {"chunks": 7625.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.453195, "cps": 59.570312, "mean_p": 0.441688, "boundary_entropy": 0.576443, "bpb": 0.858864}, "ta": {"chunks": 7616.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.407382, "cps": 59.5, "mean_p": 0.413609, "boundary_entropy": 0.592703, "bpb": 0.748199}, "th": {"chunks": 7750.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.481815, "cps": 60.546875, "mean_p": 0.44402, "boundary_entropy": 0.599571, "bpb": 0.844722}, "zh": {"chunks": 2466.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.417541, "cps": 19.265625, "mean_p": 0.264378, "boundary_entropy": 0.371511, "bpb": 2.466432}, "my": {"chunks": 9268.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.489128, "cps": 72.40625, "mean_p": 0.402524, "boundary_entropy": 0.625357, "bpb": 0.844483}}, "bpb": {"en": 1.60184, "de": 1.73207, "vi": 1.36166, "tr": 1.65436, "fi": 1.82347, "ru": 1.05257, "el": 1.06071, "hi": 0.85886, "ta": 0.7482, "th": 0.84472, "zh": 2.46643, "my": 0.84448}, "gini_cpc": 0.22076, "gini_cps": 0.233527, "bpic": 6.02085, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9120455986293801, "vi": 1.15594783134762, "tr": 0.85134708673917, "fi": 0.7854445027795834, "ru": 1.4834094728277814, "el": 1.4515775580950088, "hi": 2.27289524731192, "ta": 2.043130520034561, "th": 2.416435919050362, "zh": 2.0940848509453094, "my": 2.4531115341304184}, "t": 2795.4}
{"event": "eval", "step": 16000, "phase": 1, "bpb": {"en": 1.60159, "de": 1.73894, "vi": 1.37793, "tr": 1.64486, "fi": 1.83847, "ru": 1.05109, "el": 1.05906, "hi": 0.86328, "ta": 0.75297, "th": 0.84265, "zh": 2.47228, "my": 0.84906}, "bytes_per_chunk": {"en": 4.82, "de": 5.431, "vi": 5.577, "tr": 6.216, "fi": 5.979, "ru": 6.064, "el": 6.062, "hi": 5.801, "ta": 6.584, "th": 5.83, "zh": 6.409, "my": 5.688}, "bpb_mean": 1.34101, "t": 2800.7}
{"event": "train", "step": 16000, "phase": 1, "objective": "global", "lm_loss": 0.94096, "bpb": 1.35751, "chunk_loss": 0.030167, "F": 0.18282, "G": 0.25937, "chunks": 187.2, "bytes_per_chunk": 5.47, "boundary_entropy": 0.24158, "router_noise": 0.0, "grad_norm": 0.2538, "lr": 0.000277, "router_trainable": true, "sec_per_step": 0.183, "alpha": 0.03, "t": 2800.8}
{"event": "train", "step": 16500, "phase": 1, "objective": "global", "lm_loss": 0.98607, "bpb": 1.4226, "chunk_loss": 0.030099, "F": 0.17834, "G": 0.25224, "chunks": 182.62, "bytes_per_chunk": 5.607, "boundary_entropy": 0.23546, "router_noise": 0.0, "grad_norm": 0.2713, "lr": 0.0002754, "router_trainable": true, "sec_per_step": 0.167, "alpha": 0.03, "t": 2884.5}
{"event": "train", "step": 17000, "phase": 1, "objective": "global", "lm_loss": 1.02656, "bpb": 1.48101, "chunk_loss": 0.030141, "F": 0.18134, "G": 0.25305, "chunks": 185.69, "bytes_per_chunk": 5.515, "boundary_entropy": 0.24165, "router_noise": 0.0, "grad_norm": 0.2826, "lr": 0.0002737, "router_trainable": true, "sec_per_step": 0.167, "alpha": 0.03, "t": 2968.1}
{"event": "train", "step": 17500, "phase": 1, "objective": "global", "lm_loss": 0.91088, "bpb": 1.31412, "chunk_loss": 0.03006, "F": 0.17805, "G": 0.26413, "chunks": 182.33, "bytes_per_chunk": 5.616, "boundary_entropy": 0.26522, "router_noise": 0.0, "grad_norm": 0.2563, "lr": 0.000272, "router_trainable": true, "sec_per_step": 0.168, "alpha": 0.03, "t": 3052.1}
{"event": "train", "step": 18000, "phase": 1, "objective": "global", "lm_loss": 0.98618, "bpb": 1.42276, "chunk_loss": 0.030199, "F": 0.17976, "G": 0.26634, "chunks": 184.08, "bytes_per_chunk": 5.563, "boundary_entropy": 0.26492, "router_noise": 0.0, "grad_norm": 0.2674, "lr": 0.0002703, "router_trainable": true, "sec_per_step": 0.166, "alpha": 0.03, "t": 3135.2}
{"event": "train", "step": 18500, "phase": 1, "objective": "global", "lm_loss": 1.00131, "bpb": 1.44459, "chunk_loss": 0.030074, "F": 0.17773, "G": 0.25816, "chunks": 182.0, "bytes_per_chunk": 5.626, "boundary_entropy": 0.25097, "router_noise": 0.0, "grad_norm": 0.2674, "lr": 0.0002685, "router_trainable": true, "sec_per_step": 0.167, "alpha": 0.03, "t": 3218.6}
{"event": "train", "step": 19000, "phase": 1, "objective": "global", "lm_loss": 1.02949, "bpb": 1.48524, "chunk_loss": 0.029996, "F": 0.17798, "G": 0.25465, "chunks": 182.25, "bytes_per_chunk": 5.619, "boundary_entropy": 0.24289, "router_noise": 0.0, "grad_norm": 0.2739, "lr": 0.0002666, "router_trainable": true, "sec_per_step": 0.166, "alpha": 0.03, "t": 3301.8}
{"event": "train", "step": 19500, "phase": 1, "objective": "global", "lm_loss": 1.01928, "bpb": 1.47051, "chunk_loss": 0.030086, "F": 0.18282, "G": 0.24437, "chunks": 187.2, "bytes_per_chunk": 5.47, "boundary_entropy": 0.21451, "router_noise": 0.0, "grad_norm": 0.2705, "lr": 0.0002647, "router_trainable": true, "sec_per_step": 0.167, "alpha": 0.03, "t": 3385.4}
{"event": "probe", "step": 20000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3468.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.199276, "cps": 27.09375, "mean_p": 0.191067, "boundary_entropy": 0.069326, "bpb": 1.574817}, "de": {"chunks": 3731.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.181707, "cps": 29.148438, "mean_p": 0.182534, "boundary_entropy": 0.054343, "bpb": 1.700247}, "vi": {"chunks": 4226.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.229238, "cps": 33.015625, "mean_p": 0.175456, "boundary_entropy": 0.069869, "bpb": 1.326877}, "tr": {"chunks": 3123.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.172513, "cps": 24.398438, "mean_p": 0.211056, "boundary_entropy": 0.210241, "bpb": 1.626181}, "fi": {"chunks": 2978.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.158312, "cps": 23.265625, "mean_p": 0.180116, "boundary_entropy": 0.129164, "bpb": 1.780933}, "ru": {"chunks": 5574.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.294142, "cps": 43.546875, "mean_p": 0.246835, "boundary_entropy": 0.283937, "bpb": 1.030427}, "el": {"chunks": 6107.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.297569, "cps": 47.710938, "mean_p": 0.343428, "boundary_entropy": 0.467838, "bpb": 1.041243}, "hi": {"chunks": 7104.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.422229, "cps": 55.5, "mean_p": 0.422074, "boundary_entropy": 0.562377, "bpb": 0.837602}, "ta": {"chunks": 8939.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.478149, "cps": 69.835938, "mean_p": 0.41271, "boundary_entropy": 0.584823, "bpb": 0.731529}, "th": {"chunks": 7615.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.473422, "cps": 59.492188, "mean_p": 0.444318, "boundary_entropy": 0.598949, "bpb": 0.826842}, "zh": {"chunks": 2805.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.474941, "cps": 21.914062, "mean_p": 0.28944, "boundary_entropy": 0.394137, "bpb": 2.438286}, "my": {"chunks": 9142.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.482478, "cps": 71.421875, "mean_p": 0.410535, "boundary_entropy": 0.632316, "bpb": 0.820748}}, "bpb": {"en": 1.57482, "de": 1.70025, "vi": 1.32688, "tr": 1.62618, "fi": 1.78093, "ru": 1.03043, "el": 1.04124, "hi": 0.8376, "ta": 0.73153, "th": 0.82684, "zh": 2.43829, "my": 0.82075}, "gini_cpc": 0.222619, "gini_cps": 0.233311, "bpic": 5.92145, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9118383909638581, "vi": 1.1503536694436547, "tr": 0.8656980997051282, "fi": 0.7944340341245524, "ru": 1.476055819813569, "el": 1.4932485655809913, "hi": 2.1188143775996546, "ta": 2.39943229089065, "th": 2.375712451046721, "zh": 2.3833315073404915, "my": 2.421156554489322}, "t": 3471.5}
{"event": "eval", "step": 20000, "phase": 1, "bpb": {"en": 1.57362, "de": 1.70769, "vi": 1.3453, "tr": 1.61695, "fi": 1.79792, "ru": 1.02843, "el": 1.0399, "hi": 0.84299, "ta": 0.73672, "th": 0.82652, "zh": 2.4418, "my": 0.82493}, "bytes_per_chunk": {"en": 4.82, "de": 5.433, "vi": 5.589, "tr": 6.126, "fi": 5.913, "ru": 6.104, "el": 5.918, "hi": 6.0, "ta": 5.652, "th": 5.918, "zh": 5.944, "my": 5.739}, "bpb_mean": 1.31523, "t": 3476.8}
{"event": "train", "step": 20000, "phase": 1, "objective": "global", "lm_loss": 1.01194, "bpb": 1.45992, "chunk_loss": 0.030282, "F": 0.18445, "G": 0.25469, "chunks": 188.88, "bytes_per_chunk": 5.422, "boundary_entropy": 0.2327, "router_noise": 0.0, "grad_norm": 0.2585, "lr": 0.0002628, "router_trainable": true, "sec_per_step": 0.183, "alpha": 0.03, "t": 3477.0}
{"event": "train", "step": 20500, "phase": 1, "objective": "global", "lm_loss": 0.95996, "bpb": 1.38493, "chunk_loss": 0.029888, "F": 0.17445, "G": 0.26108, "chunks": 178.64, "bytes_per_chunk": 5.732, "boundary_entropy": 0.26179, "router_noise": 0.0, "grad_norm": 0.2673, "lr": 0.0002608, "router_trainable": true, "sec_per_step": 0.166, "alpha": 0.03, "t": 3559.9}
{"event": "train", "step": 21000, "phase": 1, "objective": "global", "lm_loss": 1.00437, "bpb": 1.449, "chunk_loss": 0.030131, "F": 0.17957, "G": 0.24879, "chunks": 183.88, "bytes_per_chunk": 5.569, "boundary_entropy": 0.23803, "router_noise": 0.0, "grad_norm": 0.2695, "lr": 0.0002587, "router_trainable": true, "sec_per_step": 0.166, "alpha": 0.03, "t": 3643.1}
{"event": "train", "step": 21500, "phase": 1, "objective": "global", "lm_loss": 0.97402, "bpb": 1.40521, "chunk_loss": 0.030098, "F": 0.17825, "G": 0.2655, "chunks": 182.53, "bytes_per_chunk": 5.61, "boundary_entropy": 0.27113, "router_noise": 0.0, "grad_norm": 0.2711, "lr": 0.0002567, "router_trainable": true, "sec_per_step": 0.166, "alpha": 0.03, "t": 3726.2}
{"event": "train", "step": 22000, "phase": 1, "objective": "global", "lm_loss": 1.00312, "bpb": 1.4472, "chunk_loss": 0.030173, "F": 0.18188, "G": 0.26637, "chunks": 186.25, "bytes_per_chunk": 5.498, "boundary_entropy": 0.26771, "router_noise": 0.0, "grad_norm": 0.277, "lr": 0.0002545, "router_trainable": true, "sec_per_step": 0.165, "alpha": 0.03, "t": 3808.7}
{"event": "train", "step": 22500, "phase": 1, "objective": "global", "lm_loss": 0.97581, "bpb": 1.40779, "chunk_loss": 0.030144, "F": 0.17934, "G": 0.2522, "chunks": 183.64, "bytes_per_chunk": 5.576, "boundary_entropy": 0.24068, "router_noise": 0.0, "grad_norm": 0.2628, "lr": 0.0002524, "router_trainable": true, "sec_per_step": 0.166, "alpha": 0.03, "t": 3891.5}
{"event": "train", "step": 23000, "phase": 1, "objective": "global", "lm_loss": 1.00612, "bpb": 1.45153, "chunk_loss": 0.029992, "F": 0.17784, "G": 0.25545, "chunks": 182.11, "bytes_per_chunk": 5.623, "boundary_entropy": 0.25027, "router_noise": 0.0, "grad_norm": 0.2783, "lr": 0.0002502, "router_trainable": true, "sec_per_step": 0.165, "alpha": 0.03, "t": 3974.3}
{"event": "train", "step": 23500, "phase": 1, "objective": "global", "lm_loss": 0.9519, "bpb": 1.3733, "chunk_loss": 0.03006, "F": 0.17766, "G": 0.26292, "chunks": 181.92, "bytes_per_chunk": 5.629, "boundary_entropy": 0.26035, "router_noise": 0.0, "grad_norm": 0.2786, "lr": 0.000248, "router_trainable": true, "sec_per_step": 0.166, "alpha": 0.03, "t": 4057.2}
{"event": "probe", "step": 24000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3462.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.198931, "cps": 27.046875, "mean_p": 0.190758, "boundary_entropy": 0.065327, "bpb": 1.557121}, "de": {"chunks": 3723.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.181318, "cps": 29.085938, "mean_p": 0.180659, "boundary_entropy": 0.052179, "bpb": 1.677939}, "vi": {"chunks": 4243.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.23016, "cps": 33.148438, "mean_p": 0.175066, "boundary_entropy": 0.062989, "bpb": 1.308174}, "tr": {"chunks": 3096.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.171021, "cps": 24.1875, "mean_p": 0.207061, "boundary_entropy": 0.198774, "bpb": 1.599788}, "fi": {"chunks": 2918.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.155122, "cps": 22.796875, "mean_p": 0.173438, "boundary_entropy": 0.119655, "bpb": 1.748388}, "ru": {"chunks": 5512.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.290871, "cps": 43.0625, "mean_p": 0.23729, "boundary_entropy": 0.264032, "bpb": 1.015731}, "el": {"chunks": 5990.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.291868, "cps": 46.796875, "mean_p": 0.340788, "boundary_entropy": 0.460549, "bpb": 1.019517}, "hi": {"chunks": 7100.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.421991, "cps": 55.46875, "mean_p": 0.403069, "boundary_entropy": 0.545697, "bpb": 0.833742}, "ta": {"chunks": 8231.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.440278, "cps": 64.304688, "mean_p": 0.403217, "boundary_entropy": 0.583947, "bpb": 0.722622}, "th": {"chunks": 7188.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.446876, "cps": 56.15625, "mean_p": 0.421697, "boundary_entropy": 0.582933, "bpb": 0.815421}, "zh": {"chunks": 2658.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.450051, "cps": 20.765625, "mean_p": 0.281906, "boundary_entropy": 0.386558, "bpb": 2.402254}, "my": {"chunks": 7646.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.403525, "cps": 59.734375, "mean_p": 0.385621, "boundary_entropy": 0.611979, "bpb": 0.799382}}, "bpb": {"en": 1.55712, "de": 1.67794, "vi": 1.30817, "tr": 1.59979, "fi": 1.74839, "ru": 1.01573, "el": 1.01952, "hi": 0.83374, "ta": 0.72262, "th": 0.81542, "zh": 2.40225, "my": 0.79938}, "gini_cpc": 0.207677, "gini_cps": 0.214191, "bpic": 6.213366, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9114601502539642, "vi": 1.156982916697808, "tr": 0.8597010501529329, "fi": 0.7797770709586107, "ru": 1.4621672466538322, "el": 1.467178769121612, "hi": 2.1212914058214727, "ta": 2.2132179755011, "th": 2.246384324179012, "zh": 2.26234373175122, "my": 2.028467128315057}, "t": 4142.8}
{"event": "eval", "step": 24000, "phase": 1, "bpb": {"en": 1.55569, "de": 1.6858, "vi": 1.32619, "tr": 1.59178, "fi": 1.76726, "ru": 1.01355, "el": 1.01842, "hi": 0.83777, "ta": 0.72874, "th": 0.81515, "zh": 2.40505, "my": 0.80434}, "bytes_per_chunk": {"en": 4.841, "de": 5.458, "vi": 5.565, "tr": 6.132, "fi": 6.026, "ru": 6.125, "el": 6.061, "hi": 5.923, "ta": 6.163, "th": 6.249, "zh": 6.147, "my": 6.843}, "bpb_mean": 1.29581, "t": 4148.0}
{"event": "train", "step": 24000, "phase": 1, "objective": "global", "lm_loss": 0.98817, "bpb": 1.42563, "chunk_loss": 0.030102, "F": 0.18193, "G": 0.24841, "chunks": 186.3, "bytes_per_chunk": 5.497, "boundary_entropy": 0.21841, "router_noise": 0.0, "grad_norm": 0.2752, "lr": 0.0002457, "router_trainable": true, "sec_per_step": 0.182, "alpha": 0.03, "t": 4148.2}
{"event": "train", "step": 24500, "phase": 1, "objective": "global", "lm_loss": 0.97587, "bpb": 1.40788, "chunk_loss": 0.030162, "F": 0.18452, "G": 0.25272, "chunks": 188.95, "bytes_per_chunk": 5.419, "boundary_entropy": 0.22196, "router_noise": 0.0, "grad_norm": 0.2869, "lr": 0.0002434, "router_trainable": true, "sec_per_step": 0.166, "alpha": 0.03, "t": 4231.4}
{"event": "train", "step": 25000, "phase": 1, "objective": "global", "lm_loss": 0.98888, "bpb": 1.42666, "chunk_loss": 0.030168, "F": 0.18015, "G": 0.25319, "chunks": 184.47, "bytes_per_chunk": 5.551, "boundary_entropy": 0.23996, "router_noise": 0.0, "grad_norm": 0.269, "lr": 0.000241, "router_trainable": true, "sec_per_step": 0.165, "alpha": 0.03, "t": 4314.0}
{"event": "train", "step": 25500, "phase": 1, "objective": "global", "lm_loss": 0.97846, "bpb": 1.41163, "chunk_loss": 0.030016, "F": 0.17815, "G": 0.251, "chunks": 182.42, "bytes_per_chunk": 5.613, "boundary_entropy": 0.23372, "router_noise": 0.0, "grad_norm": 0.2908, "lr": 0.0002387, "router_trainable": true, "sec_per_step": 0.166, "alpha": 0.03, "t": 4397.1}
{"event": "train", "step": 26000, "phase": 1, "objective": "global", "lm_loss": 0.94812, "bpb": 1.36785, "chunk_loss": 0.030045, "F": 0.17865, "G": 0.25007, "chunks": 182.94, "bytes_per_chunk": 5.598, "boundary_entropy": 0.233, "router_noise": 0.0, "grad_norm": 0.2721, "lr": 0.0002363, "router_trainable": true, "sec_per_step": 0.166, "alpha": 0.03, "t": 4480.0}
{"event": "train", "step": 26500, "phase": 1, "objective": "global", "lm_loss": 0.94215, "bpb": 1.35924, "chunk_loss": 0.030046, "F": 0.17706, "G": 0.25471, "chunks": 181.31, "bytes_per_chunk": 5.648, "boundary_entropy": 0.2498, "router_noise": 0.0, "grad_norm": 0.2754, "lr": 0.0002338, "router_trainable": true, "sec_per_step": 0.166, "alpha": 0.03, "t": 4562.9}
{"event": "train", "step": 27000, "phase": 1, "objective": "global", "lm_loss": 0.88733, "bpb": 1.28015, "chunk_loss": 0.030092, "F": 0.17909, "G": 0.24759, "chunks": 183.39, "bytes_per_chunk": 5.584, "boundary_entropy": 0.22874, "router_noise": 0.0, "grad_norm": 0.2579, "lr": 0.0002313, "router_trainable": true, "sec_per_step": 0.166, "alpha": 0.03, "t": 4645.7}
{"event": "train", "step": 27500, "phase": 1, "objective": "global", "lm_loss": 0.9668, "bpb": 1.3948, "chunk_loss": 0.030124, "F": 0.1824, "G": 0.25202, "chunks": 186.78, "bytes_per_chunk": 5.482, "boundary_entropy": 0.23413, "router_noise": 0.0, "grad_norm": 0.265, "lr": 0.0002288, "router_trainable": true, "sec_per_step": 0.166, "alpha": 0.03, "t": 4728.7}
{"event": "probe", "step": 28000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3462.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.198931, "cps": 27.046875, "mean_p": 0.190638, "boundary_entropy": 0.065559, "bpb": 1.536973}, "de": {"chunks": 3744.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.182341, "cps": 29.25, "mean_p": 0.181484, "boundary_entropy": 0.049994, "bpb": 1.658391}, "vi": {"chunks": 4241.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.230052, "cps": 33.132812, "mean_p": 0.174802, "boundary_entropy": 0.062197, "bpb": 1.28493}, "tr": {"chunks": 3113.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.17196, "cps": 24.320312, "mean_p": 0.202257, "boundary_entropy": 0.185568, "bpb": 1.579616}, "fi": {"chunks": 2970.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.157886, "cps": 23.203125, "mean_p": 0.175123, "boundary_entropy": 0.114079, "bpb": 1.731194}, "ru": {"chunks": 5683.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.299894, "cps": 44.398438, "mean_p": 0.250113, "boundary_entropy": 0.280222, "bpb": 0.998192}, "el": {"chunks": 6154.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.299859, "cps": 48.078125, "mean_p": 0.344711, "boundary_entropy": 0.460849, "bpb": 1.001423}, "hi": {"chunks": 7101.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.422051, "cps": 55.476562, "mean_p": 0.406234, "boundary_entropy": 0.550002, "bpb": 0.80957}, "ta": {"chunks": 8643.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.462316, "cps": 67.523438, "mean_p": 0.417362, "boundary_entropy": 0.586355, "bpb": 0.710151}, "th": {"chunks": 7389.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.459372, "cps": 57.726562, "mean_p": 0.42587, "boundary_entropy": 0.584272, "bpb": 0.799101}, "zh": {"chunks": 2729.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.462072, "cps": 21.320312, "mean_p": 0.295049, "boundary_entropy": 0.397315, "bpb": 2.377093}, "my": {"chunks": 8948.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.47224, "cps": 69.90625, "mean_p": 0.393178, "boundary_entropy": 0.614009, "bpb": 0.787944}}, "bpb": {"en": 1.53697, "de": 1.65839, "vi": 1.28493, "tr": 1.57962, "fi": 1.73119, "ru": 0.99819, "el": 1.00142, "hi": 0.80957, "ta": 0.71015, "th": 0.7991, "zh": 2.37709, "my": 0.78794}, "gini_cpc": 0.216189, "gini_cps": 0.228879, "bpic": 5.98004, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9166013436881121, "vi": 1.156437555907472, "tr": 0.864421630854677, "fi": 0.7936730297282638, "ru": 1.5075283858370334, "el": 1.5073486052044074, "hi": 2.1215901792589125, "ta": 2.323999873922489, "th": 2.3092005803225826, "zh": 2.3227750353457783, "my": 2.3738848893752453}, "t": 4815.0}
{"event": "eval", "step": 28000, "phase": 1, "bpb": {"en": 1.53507, "de": 1.66578, "vi": 1.30347, "tr": 1.57103, "fi": 1.74928, "ru": 0.99535, "el": 0.99854, "hi": 0.81608, "ta": 0.71609, "th": 0.79918, "zh": 2.37912, "my": 0.79217}, "bytes_per_chunk": {"en": 4.847, "de": 5.443, "vi": 5.581, "tr": 6.116, "fi": 5.945, "ru": 5.99, "el": 5.981, "hi": 6.036, "ta": 5.91, "th": 6.032, "zh": 6.035, "my": 5.888}, "bpb_mean": 1.27676, "t": 4820.2}
{"event": "train", "step": 28000, "phase": 1, "objective": "global", "lm_loss": 1.00692, "bpb": 1.45268, "chunk_loss": 0.030179, "F": 0.18004, "G": 0.26682, "chunks": 184.36, "bytes_per_chunk": 5.554, "boundary_entropy": 0.26879, "router_noise": 0.0, "grad_norm": 0.2765, "lr": 0.0002263, "router_trainable": true, "sec_per_step": 0.183, "alpha": 0.03, "t": 4820.4}
{"event": "train", "step": 28500, "phase": 1, "objective": "global", "lm_loss": 1.00764, "bpb": 1.45371, "chunk_loss": 0.030133, "F": 0.18387, "G": 0.23392, "chunks": 188.28, "bytes_per_chunk": 5.439, "boundary_entropy": 0.19417, "router_noise": 0.0, "grad_norm": 0.272, "lr": 0.0002238, "router_trainable": true, "sec_per_step": 0.166, "alpha": 0.03, "t": 4903.4}
{"event": "train", "step": 29000, "phase": 1, "objective": "global", "lm_loss": 0.92751, "bpb": 1.33811, "chunk_loss": 0.030058, "F": 0.17827, "G": 0.2478, "chunks": 182.55, "bytes_per_chunk": 5.61, "boundary_entropy": 0.22841, "router_noise": 0.0, "grad_norm": 0.2742, "lr": 0.0002212, "router_trainable": true, "sec_per_step": 0.166, "alpha": 0.03, "t": 4986.3}
{"event": "train", "step": 29500, "phase": 1, "objective": "global", "lm_loss": 1.00529, "bpb": 1.45032, "chunk_loss": 0.030141, "F": 0.18379, "G": 0.24788, "chunks": 188.2, "bytes_per_chunk": 5.441, "boundary_entropy": 0.21929, "router_noise": 0.0, "grad_norm": 0.2739, "lr": 0.0002186, "router_trainable": true, "sec_per_step": 0.166, "alpha": 0.03, "t": 5069.1}
{"event": "train", "step": 30000, "phase": 1, "objective": "global", "lm_loss": 0.95855, "bpb": 1.38289, "chunk_loss": 0.030176, "F": 0.1866, "G": 0.25112, "chunks": 191.08, "bytes_per_chunk": 5.359, "boundary_entropy": 0.21903, "router_noise": 0.0, "grad_norm": 0.2748, "lr": 0.0002159, "router_trainable": true, "sec_per_step": 0.166, "alpha": 0.03, "t": 5152.0}
{"event": "train", "step": 30500, "phase": 1, "objective": "global", "lm_loss": 1.03001, "bpb": 1.48599, "chunk_loss": 0.030206, "F": 0.18436, "G": 0.25159, "chunks": 188.78, "bytes_per_chunk": 5.424, "boundary_entropy": 0.22989, "router_noise": 0.0, "grad_norm": 0.2842, "lr": 0.0002133, "router_trainable": true, "sec_per_step": 0.166, "alpha": 0.03, "t": 5235.1}
{"event": "phase_transition", "step": 30520, "phase": 2, "objective": "parity_a", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0, "t": 5238.2}
{"event": "train", "step": 31000, "phase": 2, "objective": "parity_a", "lm_loss": 0.96976, "bpb": 1.39906, "chunk_loss": 0.030486, "F": 0.16164, "G": 0.17827, "chunks": 165.52, "bytes_per_chunk": 6.187, "boundary_entropy": 0.13054, "router_noise": 0.0, "grad_norm": 0.2836, "lr": 0.0002106, "router_trainable": true, "sec_per_step": 0.165, "n_target_mean": 9.1737, "alpha": 0.03, "t": 5317.5}
{"event": "train", "step": 31500, "phase": 2, "objective": "parity_a", "lm_loss": 0.89709, "bpb": 1.29422, "chunk_loss": 0.030497, "F": 0.153, "G": 0.15833, "chunks": 156.67, "bytes_per_chunk": 6.536, "boundary_entropy": 0.09281, "router_noise": 0.0, "grad_norm": 0.2725, "lr": 0.0002079, "router_trainable": true, "sec_per_step": 0.166, "n_target_mean": 10.0876, "alpha": 0.03, "t": 5400.4}
{"event": "probe", "step": 32000, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 3452.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.198357, "cps": 26.96875, "mean_p": 0.187264, "boundary_entropy": 0.059317, "bpb": 1.520314}, "de": {"chunks": 3726.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.181464, "cps": 29.109375, "mean_p": 0.174703, "boundary_entropy": 0.043899, "bpb": 1.642204}, "vi": {"chunks": 4006.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.217304, "cps": 31.296875, "mean_p": 0.15776, "boundary_entropy": 0.061005, "bpb": 1.299377}, "tr": {"chunks": 3057.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.168867, "cps": 23.882812, "mean_p": 0.158117, "boundary_entropy": 0.077917, "bpb": 1.560174}, "fi": {"chunks": 2931.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.155813, "cps": 22.898438, "mean_p": 0.151732, "boundary_entropy": 0.050193, "bpb": 1.708736}, "ru": {"chunks": 5016.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.264697, "cps": 39.1875, "mean_p": 0.138518, "boundary_entropy": 0.064649, "bpb": 0.998937}, "el": {"chunks": 4302.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.209618, "cps": 33.609375, "mean_p": 0.112544, "boundary_entropy": 0.064169, "bpb": 1.002978}, "hi": {"chunks": 3158.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.187697, "cps": 24.671875, "mean_p": 0.084229, "boundary_entropy": 0.083745, "bpb": 0.855675}, "ta": {"chunks": 2860.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.152982, "cps": 22.34375, "mean_p": 0.07254, "boundary_entropy": 0.096229, "bpb": 0.748096}, "th": {"chunks": 3365.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.209201, "cps": 26.289062, "mean_p": 0.101265, "boundary_entropy": 0.149039, "bpb": 0.84157}, "zh": {"chunks": 2559.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.433288, "cps": 19.992188, "mean_p": 0.259805, "boundary_entropy": 0.337304, "bpb": 2.358388}, "my": {"chunks": 2262.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.119379, "cps": 17.671875, "mean_p": 0.085823, "boundary_entropy": 0.194831, "bpb": 0.786387}}, "bpb": {"en": 1.52031, "de": 1.6422, "vi": 1.29938, "tr": 1.56017, "fi": 1.70874, "ru": 0.99894, "el": 1.00298, "hi": 0.85567, "ta": 0.7481, "th": 0.84157, "zh": 2.35839, "my": 0.78639}, "gini_cpc": 0.172975, "gini_cps": 0.121578, "bpic": 9.430899, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9148371169062898, "vi": 1.095522082692596, "tr": 0.8513305541631897, "fi": 0.7855200382311255, "ru": 1.3344479740244652, "el": 1.0567758842084811, "hi": 0.9462597903922011, "ta": 0.7712477264324521, "th": 1.0546718061745413, "zh": 2.18438998343692, "my": 0.6018420909724022}, "t": 5486.1}
{"event": "eval", "step": 32000, "phase": 2, "bpb": {"en": 1.51863, "de": 1.65035, "vi": 1.31609, "tr": 1.55242, "fi": 1.726, "ru": 0.99615, "el": 1.00062, "hi": 0.86145, "ta": 0.75288, "th": 0.84049, "zh": 2.36139, "my": 0.79132}, "bytes_per_chunk": {"en": 4.857, "de": 5.456, "vi": 5.847, "tr": 6.173, "fi": 6.016, "ru": 6.878, "el": 8.205, "hi": 13.011, "ta": 16.504, "th": 13.452, "zh": 6.147, "my": 19.938}, "bpb_mean": 1.28065, "t": 5490.9}
{"event": "train", "step": 32000, "phase": 2, "objective": "parity_a", "lm_loss": 0.9699, "bpb": 1.39927, "chunk_loss": 0.03044, "F": 0.15817, "G": 0.1692, "chunks": 161.97, "bytes_per_chunk": 6.322, "boundary_entropy": 0.10853, "router_noise": 0.0, "grad_norm": 0.2816, "lr": 0.0002052, "router_trainable": true, "sec_per_step": 0.182, "n_target_mean": 9.3217, "alpha": 0.03, "t": 5491.1}
{"event": "train", "step": 32500, "phase": 2, "objective": "parity_a", "lm_loss": 0.89067, "bpb": 1.28496, "chunk_loss": 0.030436, "F": 0.15604, "G": 0.15869, "chunks": 159.78, "bytes_per_chunk": 6.409, "boundary_entropy": 0.0818, "router_noise": 0.0, "grad_norm": 0.2806, "lr": 0.0002025, "router_trainable": true, "sec_per_step": 0.167, "n_target_mean": 9.5534, "alpha": 0.03, "t": 5574.5}
{"event": "train", "step": 33000, "phase": 2, "objective": "parity_a", "lm_loss": 1.01867, "bpb": 1.46963, "chunk_loss": 0.030267, "F": 0.16437, "G": 0.17665, "chunks": 168.31, "bytes_per_chunk": 6.084, "boundary_entropy": 0.10806, "router_noise": 0.0, "grad_norm": 0.2987, "lr": 0.0001997, "router_trainable": true, "sec_per_step": 0.167, "n_target_mean": 8.4662, "alpha": 0.03, "t": 5657.8}
{"event": "train", "step": 33500, "phase": 2, "objective": "parity_a", "lm_loss": 0.9649, "bpb": 1.39206, "chunk_loss": 0.030472, "F": 0.1577, "G": 0.16263, "chunks": 161.48, "bytes_per_chunk": 6.341, "boundary_entropy": 0.08612, "router_noise": 0.0, "grad_norm": 0.287, "lr": 0.0001969, "router_trainable": true, "sec_per_step": 0.166, "n_target_mean": 9.5713, "alpha": 0.03, "t": 5740.9}
{"event": "train", "step": 34000, "phase": 2, "objective": "parity_a", "lm_loss": 0.91749, "bpb": 1.32366, "chunk_loss": 0.030354, "F": 0.15359, "G": 0.15712, "chunks": 157.28, "bytes_per_chunk": 6.511, "boundary_entropy": 0.08275, "router_noise": 0.0, "grad_norm": 0.2686, "lr": 0.0001942, "router_trainable": true, "sec_per_step": 0.166, "n_target_mean": 9.7479, "alpha": 0.03, "t": 5823.8}
{"event": "train", "step": 34500, "phase": 2, "objective": "parity_a", "lm_loss": 0.95389, "bpb": 1.37617, "chunk_loss": 0.03042, "F": 0.16855, "G": 0.16849, "chunks": 172.59, "bytes_per_chunk": 5.933, "boundary_entropy": 0.07316, "router_noise": 0.0, "grad_norm": 0.278, "lr": 0.0001914, "router_trainable": true, "sec_per_step": 0.166, "n_target_mean": 8.4457, "alpha": 0.03, "t": 5906.7}
{"event": "train", "step": 35000, "phase": 2, "objective": "parity_a", "lm_loss": 0.95939, "bpb": 1.38411, "chunk_loss": 0.030322, "F": 0.15378, "G": 0.16258, "chunks": 157.47, "bytes_per_chunk": 6.503, "boundary_entropy": 0.09774, "router_noise": 0.0, "grad_norm": 0.2758, "lr": 0.0001886, "router_trainable": true, "sec_per_step": 0.166, "n_target_mean": 9.736, "alpha": 0.03, "t": 5989.8}
{"event": "train", "step": 35500, "phase": 2, "objective": "parity_a", "lm_loss": 0.99426, "bpb": 1.43441, "chunk_loss": 0.030326, "F": 0.16495, "G": 0.17457, "chunks": 168.91, "bytes_per_chunk": 6.063, "boundary_entropy": 0.09165, "router_noise": 0.0, "grad_norm": 0.2779, "lr": 0.0001858, "router_trainable": true, "sec_per_step": 0.165, "n_target_mean": 8.6476, "alpha": 0.03, "t": 6072.5}
{"event": "probe", "step": 36000, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 3433.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.197265, "cps": 26.820312, "mean_p": 0.1861, "boundary_entropy": 0.055288, "bpb": 1.506202}, "de": {"chunks": 3707.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.180539, "cps": 28.960938, "mean_p": 0.172882, "boundary_entropy": 0.04112, "bpb": 1.62615}, "vi": {"chunks": 3823.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.207377, "cps": 29.867188, "mean_p": 0.152053, "boundary_entropy": 0.043489, "bpb": 1.283798}, "tr": {"chunks": 2980.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.164614, "cps": 23.28125, "mean_p": 0.150435, "boundary_entropy": 0.05383, "bpb": 1.543698}, "fi": {"chunks": 2912.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.154803, "cps": 22.75, "mean_p": 0.1485, "boundary_entropy": 0.038518, "bpb": 1.686438}, "ru": {"chunks": 4837.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.255251, "cps": 37.789062, "mean_p": 0.131759, "boundary_entropy": 0.055996, "bpb": 0.986696}, "el": {"chunks": 4196.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.204454, "cps": 32.78125, "mean_p": 0.107484, "boundary_entropy": 0.051086, "bpb": 0.987204}, "hi": {"chunks": 3036.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.180446, "cps": 23.71875, "mean_p": 0.076766, "boundary_entropy": 0.056375, "bpb": 0.835787}, "ta": {"chunks": 2501.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.133779, "cps": 19.539062, "mean_p": 0.058672, "boundary_entropy": 0.064717, "bpb": 0.73584}, "th": {"chunks": 2913.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.1811, "cps": 22.757812, "mean_p": 0.082592, "boundary_entropy": 0.116241, "bpb": 0.828602}, "zh": {"chunks": 2819.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.477311, "cps": 22.023438, "mean_p": 0.297065, "boundary_entropy": 0.381635, "bpb": 2.333487}, "my": {"chunks": 2381.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.12566, "cps": 18.601562, "mean_p": 0.066128, "boundary_entropy": 0.135486, "bpb": 0.775499}}, "bpb": {"en": 1.5062, "de": 1.62615, "vi": 1.2838, "tr": 1.5437, "fi": 1.68644, "ru": 0.9867, "el": 0.9872, "hi": 0.83579, "ta": 0.73584, "th": 0.8286, "zh": 2.33349, "my": 0.7755}, "gini_cpc": 0.188813, "gini_cps": 0.11569, "bpic": 9.706637, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9152094496977428, "vi": 1.0512632262795625, "tr": 0.8344801864348774, "fi": 0.7847472527207069, "ru": 1.2939490910432423, "el": 1.0364418745749708, "hi": 0.91473861121559, "ta": 0.6781699451705584, "th": 0.9180571898071715, "zh": 2.4196466557482905, "my": 0.637010126169889}, "t": 6157.9}
{"event": "eval", "step": 36000, "phase": 2, "bpb": {"en": 1.50424, "de": 1.63476, "vi": 1.30138, "tr": 1.53568, "fi": 1.70469, "ru": 0.98444, "el": 0.98424, "hi": 0.84171, "ta": 0.74081, "th": 0.82885, "zh": 2.33677, "my": 0.7805}, "bytes_per_chunk": {"en": 4.872, "de": 5.477, "vi": 6.145, "tr": 6.35, "fi": 6.032, "ru": 7.184, "el": 8.419, "hi": 13.457, "ta": 18.686, "th": 16.053, "zh": 5.659, "my": 20.326}, "bpb_mean": 1.26484, "t": 6162.8}
{"event": "train", "step": 36000, "phase": 2, "objective": "parity_a", "lm_loss": 0.95031, "bpb": 1.37101, "chunk_loss": 0.030362, "F": 0.15674, "G": 0.167, "chunks": 160.5, "bytes_per_chunk": 6.38, "boundary_entropy": 0.0937, "router_noise": 0.0, "grad_norm": 0.2835, "lr": 0.0001829, "router_trainable": true, "sec_per_step": 0.181, "n_target_mean": 9.1138, "alpha": 0.03, "t": 6162.9}
{"event": "train", "step": 36500, "phase": 2, "objective": "parity_a", "lm_loss": 0.92773, "bpb": 1.33842, "chunk_loss": 0.03042, "F": 0.15634, "G": 0.16564, "chunks": 160.09, "bytes_per_chunk": 6.396, "boundary_entropy": 0.09703, "router_noise": 0.0, "grad_norm": 0.2796, "lr": 0.0001801, "router_trainable": true, "sec_per_step": 0.166, "n_target_mean": 9.6158, "alpha": 0.03, "t": 6246.2}
{"event": "train", "step": 37000, "phase": 2, "objective": "parity_a", "lm_loss": 0.95014, "bpb": 1.37077, "chunk_loss": 0.030326, "F": 0.15884, "G": 0.16721, "chunks": 162.66, "bytes_per_chunk": 6.295, "boundary_entropy": 0.08899, "router_noise": 0.0, "grad_norm": 0.2881, "lr": 0.0001773, "router_trainable": true, "sec_per_step": 0.167, "n_target_mean": 8.8713, "alpha": 0.03, "t": 6329.5}
{"event": "train", "step": 37500, "phase": 2, "objective": "parity_a", "lm_loss": 0.92033, "bpb": 1.32776, "chunk_loss": 0.030357, "F": 0.16315, "G": 0.17076, "chunks": 167.06, "bytes_per_chunk": 6.129, "boundary_entropy": 0.09587, "router_noise": 0.0, "grad_norm": 0.2888, "lr": 0.0001744, "router_trainable": true, "sec_per_step": 0.166, "n_target_mean": 8.8044, "alpha": 0.03, "t": 6412.3}
{"event": "train", "step": 38000, "phase": 2, "objective": "parity_a", "lm_loss": 0.93009, "bpb": 1.34184, "chunk_loss": 0.030329, "F": 0.16234, "G": 0.15967, "chunks": 166.23, "bytes_per_chunk": 6.16, "boundary_entropy": 0.0697, "router_noise": 0.0, "grad_norm": 0.2762, "lr": 0.0001716, "router_trainable": true, "sec_per_step": 0.166, "n_target_mean": 8.9648, "alpha": 0.03, "t": 6495.1}
{"event": "train", "step": 38500, "phase": 2, "objective": "parity_a", "lm_loss": 0.92726, "bpb": 1.33775, "chunk_loss": 0.03041, "F": 0.16046, "G": 0.1708, "chunks": 164.31, "bytes_per_chunk": 6.232, "boundary_entropy": 0.09711, "router_noise": 0.0, "grad_norm": 0.2829, "lr": 0.0001687, "router_trainable": true, "sec_per_step": 0.165, "n_target_mean": 9.2629, "alpha": 0.03, "t": 6577.7}
{"event": "train", "step": 39000, "phase": 2, "objective": "parity_a", "lm_loss": 0.95008, "bpb": 1.37068, "chunk_loss": 0.030365, "F": 0.15582, "G": 0.16771, "chunks": 159.56, "bytes_per_chunk": 6.418, "boundary_entropy": 0.10572, "router_noise": 0.0, "grad_norm": 0.3008, "lr": 0.0001659, "router_trainable": true, "sec_per_step": 0.166, "n_target_mean": 9.6656, "alpha": 0.03, "t": 6660.5}
{"event": "train", "step": 39500, "phase": 2, "objective": "parity_a", "lm_loss": 0.88787, "bpb": 1.28093, "chunk_loss": 0.030487, "F": 0.15854, "G": 0.15807, "chunks": 162.34, "bytes_per_chunk": 6.308, "boundary_entropy": 0.07262, "router_noise": 0.0, "grad_norm": 0.3068, "lr": 0.000163, "router_trainable": true, "sec_per_step": 0.166, "n_target_mean": 9.5204, "alpha": 0.03, "t": 6743.2}
{"event": "probe", "step": 40000, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 3432.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.197207, "cps": 26.8125, "mean_p": 0.184352, "boundary_entropy": 0.058919, "bpb": 1.491634}, "de": {"chunks": 3726.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.181464, "cps": 29.109375, "mean_p": 0.171857, "boundary_entropy": 0.044519, "bpb": 1.608288}, "vi": {"chunks": 3821.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.207269, "cps": 29.851562, "mean_p": 0.149856, "boundary_entropy": 0.047976, "bpb": 1.265831}, "tr": {"chunks": 3031.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.167431, "cps": 23.679688, "mean_p": 0.151543, "boundary_entropy": 0.0545, "bpb": 1.528296}, "fi": {"chunks": 2958.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.157248, "cps": 23.109375, "mean_p": 0.149966, "boundary_entropy": 0.041696, "bpb": 1.664988}, "ru": {"chunks": 4681.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.247018, "cps": 36.570312, "mean_p": 0.1239, "boundary_entropy": 0.059967, "bpb": 0.976405}, "el": {"chunks": 4131.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.201286, "cps": 32.273438, "mean_p": 0.101936, "boundary_entropy": 0.055595, "bpb": 0.97485}, "hi": {"chunks": 3105.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.184547, "cps": 24.257812, "mean_p": 0.078597, "boundary_entropy": 0.052308, "bpb": 0.81963}, "ta": {"chunks": 2757.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.147473, "cps": 21.539062, "mean_p": 0.063134, "boundary_entropy": 0.061841, "bpb": 0.718907}, "th": {"chunks": 2894.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.179919, "cps": 22.609375, "mean_p": 0.083417, "boundary_entropy": 0.1232, "bpb": 0.816178}, "zh": {"chunks": 2677.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.453268, "cps": 20.914062, "mean_p": 0.30442, "boundary_entropy": 0.398519, "bpb": 2.325669}, "my": {"chunks": 2666.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.140701, "cps": 20.828125, "mean_p": 0.06859, "boundary_entropy": 0.131372, "bpb": 0.764619}}, "bpb": {"en": 1.49163, "de": 1.60829, "vi": 1.26583, "tr": 1.5283, "fi": 1.66499, "ru": 0.97641, "el": 0.97485, "hi": 0.81963, "ta": 0.71891, "th": 0.81618, "zh": 2.32567, "my": 0.76462}, "gini_cpc": 0.165952, "gini_cps": 0.101045, "bpic": 9.623637, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9201683355362799, "vi": 1.0510194104783201, "tr": 0.849008866899548, "fi": 0.7973759402046042, "ru": 1.252582292377807, "el": 1.0206837271776608, "hi": 0.9358007148868962, "ta": 0.7478046091378533, "th": 0.9123349296025702, "zh": 2.298432582486934, "my": 0.7134665392660959}, "t": 6829.1}
{"event": "eval", "step": 40000, "phase": 2, "bpb": {"en": 1.49034, "de": 1.61626, "vi": 1.28273, "tr": 1.5212, "fi": 1.68538, "ru": 0.97397, "el": 0.97339, "hi": 0.82537, "ta": 0.72387, "th": 0.81611, "zh": 2.32644, "my": 0.76952}, "bytes_per_chunk": {"en": 4.894, "de": 5.485, "vi": 6.167, "tr": 6.268, "fi": 5.987, "ru": 7.461, "el": 8.548, "hi": 13.246, "ta": 17.124, "th": 16.222, "zh": 5.846, "my": 18.866}, "bpb_mean": 1.25038, "t": 6834.0}
{"event": "train", "step": 40000, "phase": 2, "objective": "parity_a", "lm_loss": 0.96512, "bpb": 1.39237, "chunk_loss": 0.030207, "F": 0.14998, "G": 0.17413, "chunks": 153.58, "bytes_per_chunk": 6.668, "boundary_entropy": 0.13268, "router_noise": 0.0, "grad_norm": 0.2955, "lr": 0.0001601, "router_trainable": true, "sec_per_step": 0.182, "n_target_mean": 9.4481, "alpha": 0.03, "t": 6834.2}
{"event": "train", "step": 40500, "phase": 2, "objective": "parity_a", "lm_loss": 0.91144, "bpb": 1.31493, "chunk_loss": 0.030293, "F": 0.16281, "G": 0.16888, "chunks": 166.72, "bytes_per_chunk": 6.142, "boundary_entropy": 0.08783, "router_noise": 0.0, "grad_norm": 0.2965, "lr": 0.0001573, "router_trainable": true, "sec_per_step": 0.165, "n_target_mean": 8.6742, "alpha": 0.03, "t": 6916.7}
{"event": "train", "step": 41000, "phase": 2, "objective": "parity_a", "lm_loss": 0.92373, "bpb": 1.33267, "chunk_loss": 0.030324, "F": 0.1591, "G": 0.1705, "chunks": 162.92, "bytes_per_chunk": 6.285, "boundary_entropy": 0.10218, "router_noise": 0.0, "grad_norm": 0.2932, "lr": 0.0001545, "router_trainable": true, "sec_per_step": 0.166, "n_target_mean": 9.1467, "alpha": 0.03, "t": 6999.5}
{"event": "train", "step": 41500, "phase": 2, "objective": "parity_a", "lm_loss": 0.93217, "bpb": 1.34483, "chunk_loss": 0.030318, "F": 0.15755, "G": 0.17432, "chunks": 161.33, "bytes_per_chunk": 6.347, "boundary_entropy": 0.10813, "router_noise": 0.0, "grad_norm": 0.3185, "lr": 0.0001516, "router_trainable": true, "sec_per_step": 0.165, "n_target_mean": 9.0892, "alpha": 0.03, "t": 7081.9}
{"event": "train", "step": 42000, "phase": 2, "objective": "parity_a", "lm_loss": 0.87385, "bpb": 1.2607, "chunk_loss": 0.030273, "F": 0.15158, "G": 0.15451, "chunks": 155.22, "bytes_per_chunk": 6.597, "boundary_entropy": 0.08064, "router_noise": 0.0, "grad_norm": 0.2974, "lr": 0.0001488, "router_trainable": true, "sec_per_step": 0.166, "n_target_mean": 9.6388, "alpha": 0.03, "t": 7164.7}
{"event": "train", "step": 42500, "phase": 2, "objective": "parity_a", "lm_loss": 0.99227, "bpb": 1.43154, "chunk_loss": 0.030326, "F": 0.16104, "G": 0.17775, "chunks": 164.91, "bytes_per_chunk": 6.21, "boundary_entropy": 0.11084, "router_noise": 0.0, "grad_norm": 0.3066, "lr": 0.0001459, "router_trainable": true, "sec_per_step": 0.165, "n_target_mean": 8.7573, "alpha": 0.03, "t": 7247.2}
{"event": "train", "step": 43000, "phase": 2, "objective": "parity_a", "lm_loss": 0.95965, "bpb": 1.38448, "chunk_loss": 0.030445, "F": 0.15932, "G": 0.16671, "chunks": 163.14, "bytes_per_chunk": 6.277, "boundary_entropy": 0.0865, "router_noise": 0.0, "grad_norm": 0.3008, "lr": 0.0001431, "router_trainable": true, "sec_per_step": 0.165, "n_target_mean": 9.2688, "alpha": 0.03, "t": 7329.5}
{"event": "train", "step": 43500, "phase": 2, "objective": "parity_a", "lm_loss": 0.97774, "bpb": 1.41058, "chunk_loss": 0.030272, "F": 0.16113, "G": 0.17276, "chunks": 165.0, "bytes_per_chunk": 6.206, "boundary_entropy": 0.09736, "router_noise": 0.0, "grad_norm": 0.292, "lr": 0.0001403, "router_trainable": true, "sec_per_step": 0.165, "n_target_mean": 8.951, "alpha": 0.03, "t": 7412.1}
{"event": "probe", "step": 44000, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 3435.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.19738, "cps": 26.835938, "mean_p": 0.186842, "boundary_entropy": 0.052273, "bpb": 1.48099}, "de": {"chunks": 3721.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.18122, "cps": 29.070312, "mean_p": 0.17328, "boundary_entropy": 0.03977, "bpb": 1.592126}, "vi": {"chunks": 3821.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.207269, "cps": 29.851562, "mean_p": 0.151636, "boundary_entropy": 0.041709, "bpb": 1.248766}, "tr": {"chunks": 3040.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.167928, "cps": 23.75, "mean_p": 0.152598, "boundary_entropy": 0.046356, "bpb": 1.510928}, "fi": {"chunks": 2926.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.155547, "cps": 22.859375, "mean_p": 0.149665, "boundary_entropy": 0.035805, "bpb": 1.63999}, "ru": {"chunks": 4640.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.244855, "cps": 36.25, "mean_p": 0.123793, "boundary_entropy": 0.055644, "bpb": 0.967374}, "el": {"chunks": 4134.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.201433, "cps": 32.296875, "mean_p": 0.10397, "boundary_entropy": 0.048241, "bpb": 0.960253}, "hi": {"chunks": 3131.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.186092, "cps": 24.460938, "mean_p": 0.07813, "boundary_entropy": 0.048259, "bpb": 0.810082}, "ta": {"chunks": 2669.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.142765, "cps": 20.851562, "mean_p": 0.0596, "boundary_entropy": 0.056035, "bpb": 0.710518}, "th": {"chunks": 2930.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.182157, "cps": 22.890625, "mean_p": 0.078223, "boundary_entropy": 0.103403, "bpb": 0.804444}, "zh": {"chunks": 2840.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.480867, "cps": 22.1875, "mean_p": 0.313997, "boundary_entropy": 0.407406, "bpb": 2.306549}, "my": {"chunks": 2446.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.12909, "cps": 19.109375, "mean_p": 0.065948, "boundary_entropy": 0.128454, "bpb": 0.757052}}, "bpb": {"en": 1.48099, "de": 1.59213, "vi": 1.24877, "tr": 1.51093, "fi": 1.63999, "ru": 0.96737, "el": 0.96025, "hi": 0.81008, "ta": 0.71052, "th": 0.80444, "zh": 2.30655, "my": 0.75705}, "gini_cpc": 0.180702, "gini_cps": 0.104378, "bpic": 9.658999, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9181309796967583, "vi": 1.0501014895958063, "tr": 0.8507861492232583, "fi": 0.7880609668365091, "ru": 1.2405267827309283, "el": 1.0205328902475377, "hi": 0.94281259043454, "ta": 0.7233033608920957, "th": 0.9228772365150748, "zh": 2.436252378973644, "my": 0.6540191358007621}, "t": 7497.3}
{"event": "eval", "step": 44000, "phase": 2, "bpb": {"en": 1.47855, "de": 1.60129, "vi": 1.26614, "tr": 1.50352, "fi": 1.66055, "ru": 0.96457, "el": 0.95935, "hi": 0.81515, "ta": 0.71561, "th": 0.80391, "zh": 2.31081, "my": 0.76204}, "bytes_per_chunk": {"en": 4.89, "de": 5.488, "vi": 6.175, "tr": 6.258, "fi": 6.015, "ru": 7.546, "el": 8.556, "hi": 13.137, "ta": 17.642, "th": 16.675, "zh": 5.711, "my": 20.535}, "bpb_mean": 1.23679, "t": 7502.3}
{"event": "train", "step": 44000, "phase": 2, "objective": "parity_a", "lm_loss": 0.88032, "bpb": 1.27003, "chunk_loss": 0.030298, "F": 0.15736, "G": 0.16459, "chunks": 161.14, "bytes_per_chunk": 6.355, "boundary_entropy": 0.08657, "router_noise": 0.0, "grad_norm": 0.2921, "lr": 0.0001375, "router_trainable": true, "sec_per_step": 0.181, "n_target_mean": 9.3459, "alpha": 0.03, "t": 7502.4}
{"event": "train", "step": 44500, "phase": 2, "objective": "parity_a", "lm_loss": 0.8811, "bpb": 1.27116, "chunk_loss": 0.030377, "F": 0.15144, "G": 0.15495, "chunks": 155.08, "bytes_per_chunk": 6.603, "boundary_entropy": 0.0799, "router_noise": 0.0, "grad_norm": 0.2825, "lr": 0.0001347, "router_trainable": true, "sec_per_step": 0.165, "n_target_mean": 9.9692, "alpha": 0.03, "t": 7584.8}
{"event": "train", "step": 45000, "phase": 2, "objective": "parity_a", "lm_loss": 0.9213, "bpb": 1.32915, "chunk_loss": 0.030278, "F": 0.1539, "G": 0.15731, "chunks": 157.59, "bytes_per_chunk": 6.498, "boundary_entropy": 0.08004, "router_noise": 0.0, "grad_norm": 0.3055, "lr": 0.0001319, "router_trainable": true, "sec_per_step": 0.164, "n_target_mean": 9.4298, "alpha": 0.03, "t": 7667.0}
{"event": "train", "step": 45500, "phase": 2, "objective": "parity_a", "lm_loss": 0.89878, "bpb": 1.29667, "chunk_loss": 0.030282, "F": 0.15167, "G": 0.15543, "chunks": 155.31, "bytes_per_chunk": 6.593, "boundary_entropy": 0.08253, "router_noise": 0.0, "grad_norm": 0.3028, "lr": 0.0001292, "router_trainable": true, "sec_per_step": 0.165, "n_target_mean": 9.4711, "alpha": 0.03, "t": 7749.5}
{"event": "train", "step": 46000, "phase": 2, "objective": "parity_a", "lm_loss": 0.94114, "bpb": 1.35778, "chunk_loss": 0.030268, "F": 0.16179, "G": 0.17319, "chunks": 165.67, "bytes_per_chunk": 6.181, "boundary_entropy": 0.09925, "router_noise": 0.0, "grad_norm": 0.3028, "lr": 0.0001264, "router_trainable": true, "sec_per_step": 0.164, "n_target_mean": 8.82, "alpha": 0.03, "t": 7831.7}
{"event": "train", "step": 46500, "phase": 2, "objective": "parity_a", "lm_loss": 0.91162, "bpb": 1.3152, "chunk_loss": 0.030349, "F": 0.16348, "G": 0.16462, "chunks": 167.41, "bytes_per_chunk": 6.117, "boundary_entropy": 0.075, "router_noise": 0.0, "grad_norm": 0.2937, "lr": 0.0001237, "router_trainable": true, "sec_per_step": 0.165, "n_target_mean": 8.9054, "alpha": 0.03, "t": 7914.2}
{"event": "train", "step": 47000, "phase": 2, "objective": "parity_a", "lm_loss": 0.92328, "bpb": 1.33201, "chunk_loss": 0.030199, "F": 0.15895, "G": 0.17202, "chunks": 162.77, "bytes_per_chunk": 6.291, "boundary_entropy": 0.10045, "router_noise": 0.0, "grad_norm": 0.3112, "lr": 0.000121, "router_trainable": true, "sec_per_step": 0.164, "n_target_mean": 8.8273, "alpha": 0.03, "t": 7996.4}
{"event": "train", "step": 47500, "phase": 2, "objective": "parity_a", "lm_loss": 0.92988, "bpb": 1.34154, "chunk_loss": 0.030251, "F": 0.15228, "G": 0.17006, "chunks": 155.94, "bytes_per_chunk": 6.567, "boundary_entropy": 0.10921, "router_noise": 0.0, "grad_norm": 0.3127, "lr": 0.0001183, "router_trainable": true, "sec_per_step": 0.165, "n_target_mean": 9.298, "alpha": 0.03, "t": 8078.7}
{"event": "probe", "step": 48000, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 3428.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.196978, "cps": 26.78125, "mean_p": 0.184607, "boundary_entropy": 0.056278, "bpb": 1.464976}, "de": {"chunks": 3719.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.181123, "cps": 29.054688, "mean_p": 0.172162, "boundary_entropy": 0.040644, "bpb": 1.57764}, "vi": {"chunks": 3817.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.207052, "cps": 29.820312, "mean_p": 0.150536, "boundary_entropy": 0.042552, "bpb": 1.231896}, "tr": {"chunks": 3006.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.16605, "cps": 23.484375, "mean_p": 0.150154, "boundary_entropy": 0.048515, "bpb": 1.497588}, "fi": {"chunks": 2908.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.15459, "cps": 22.71875, "mean_p": 0.147709, "boundary_entropy": 0.034885, "bpb": 1.623378}, "ru": {"chunks": 4574.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.241372, "cps": 35.734375, "mean_p": 0.120863, "boundary_entropy": 0.057142, "bpb": 0.956155}, "el": {"chunks": 4106.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.200068, "cps": 32.078125, "mean_p": 0.101272, "boundary_entropy": 0.051711, "bpb": 0.948828}, "hi": {"chunks": 3087.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.183477, "cps": 24.117188, "mean_p": 0.076311, "boundary_entropy": 0.045412, "bpb": 0.79893}, "ta": {"chunks": 2790.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.149238, "cps": 21.796875, "mean_p": 0.060747, "boundary_entropy": 0.052159, "bpb": 0.702735}, "th": {"chunks": 3117.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.193783, "cps": 24.351562, "mean_p": 0.078271, "boundary_entropy": 0.087845, "bpb": 0.792401}, "zh": {"chunks": 2790.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.472401, "cps": 21.796875, "mean_p": 0.322355, "boundary_entropy": 0.42019, "bpb": 2.290838}, "my": {"chunks": 2414.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.127401, "cps": 18.859375, "mean_p": 0.059581, "boundary_entropy": 0.109443, "bpb": 0.75019}}, "bpb": {"en": 1.46498, "de": 1.57764, "vi": 1.2319, "tr": 1.49759, "fi": 1.62338, "ru": 0.95615, "el": 0.94883, "hi": 0.79893, "ta": 0.70273, "th": 0.7924, "zh": 2.29084, "my": 0.75019}, "gini_cpc": 0.175841, "gini_cps": 0.100916, "bpic": 9.653411, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9195113157343948, "vi": 1.051144264483462, "tr": 0.8429886586356522, "fi": 0.7848123447718311, "ru": 1.225378490962214, "el": 1.015690538395292, "hi": 0.9314614025566915, "ta": 0.7576385089767792, "th": 0.9837823892833034, "zh": 2.3982478708636785, "my": 0.6467809153541634}, "t": 8163.5}
{"event": "eval", "step": 48000, "phase": 2, "bpb": {"en": 1.46317, "de": 1.5859, "vi": 1.25045, "tr": 1.48955, "fi": 1.64511, "ru": 0.95394, "el": 0.94734, "hi": 0.80526, "ta": 0.70847, "th": 0.79246, "zh": 2.29491, "my": 0.75532}, "bytes_per_chunk": {"en": 4.898, "de": 5.5, "vi": 6.181, "tr": 6.324, "fi": 6.058, "ru": 7.642, "el": 8.605, "hi": 13.313, "ta": 16.952, "th": 15.395, "zh": 5.668, "my": 20.593}, "bpb_mean": 1.22432, "t": 8168.4}
{"event": "train", "step": 48000, "phase": 2, "objective": "parity_a", "lm_loss": 0.91551, "bpb": 1.3208, "chunk_loss": 0.030369, "F": 0.157, "G": 0.1586, "chunks": 160.77, "bytes_per_chunk": 6.37, "boundary_entropy": 0.07848, "router_noise": 0.0, "grad_norm": 0.2968, "lr": 0.0001157, "router_trainable": true, "sec_per_step": 0.18, "n_target_mean": 9.5992, "alpha": 0.03, "t": 8168.6}
{"event": "train", "step": 48500, "phase": 2, "objective": "parity_a", "lm_loss": 0.94454, "bpb": 1.36268, "chunk_loss": 0.030219, "F": 0.15427, "G": 0.16875, "chunks": 157.97, "bytes_per_chunk": 6.482, "boundary_entropy": 0.10305, "router_noise": 0.0, "grad_norm": 0.3024, "lr": 0.000113, "router_trainable": true, "sec_per_step": 0.164, "n_target_mean": 8.969, "alpha": 0.03, "t": 8250.8}
{"event": "train", "step": 49000, "phase": 2, "objective": "parity_a", "lm_loss": 0.83817, "bpb": 1.20922, "chunk_loss": 0.030293, "F": 0.14966, "G": 0.14847, "chunks": 153.25, "bytes_per_chunk": 6.682, "boundary_entropy": 0.06896, "router_noise": 0.0, "grad_norm": 0.2943, "lr": 0.0001104, "router_trainable": true, "sec_per_step": 0.166, "n_target_mean": 9.985, "alpha": 0.03, "t": 8333.7}
{"event": "train", "step": 49500, "phase": 2, "objective": "parity_a", "lm_loss": 0.98558, "bpb": 1.42189, "chunk_loss": 0.030165, "F": 0.15504, "G": 0.1767, "chunks": 158.77, "bytes_per_chunk": 6.45, "boundary_entropy": 0.12064, "router_noise": 0.0, "grad_norm": 0.3105, "lr": 0.0001078, "router_trainable": true, "sec_per_step": 0.165, "n_target_mean": 8.8682, "alpha": 0.03, "t": 8416.2}
{"event": "train", "step": 50000, "phase": 2, "objective": "parity_a", "lm_loss": 0.91099, "bpb": 1.31428, "chunk_loss": 0.030255, "F": 0.15372, "G": 0.15782, "chunks": 157.41, "bytes_per_chunk": 6.505, "boundary_entropy": 0.07495, "router_noise": 0.0, "grad_norm": 0.3048, "lr": 0.0001052, "router_trainable": true, "sec_per_step": 0.166, "n_target_mean": 9.4855, "alpha": 0.03, "t": 8499.1}
{"event": "train", "step": 50500, "phase": 2, "objective": "parity_a", "lm_loss": 0.94737, "bpb": 1.36677, "chunk_loss": 0.030311, "F": 0.14331, "G": 0.16573, "chunks": 146.75, "bytes_per_chunk": 6.978, "boundary_entropy": 0.12267, "router_noise": 0.0, "grad_norm": 0.3121, "lr": 0.0001027, "router_trainable": true, "sec_per_step": 0.165, "n_target_mean": 10.4046, "alpha": 0.03, "t": 8581.8}
{"event": "train", "step": 51000, "phase": 2, "objective": "parity_a", "lm_loss": 0.86053, "bpb": 1.24148, "chunk_loss": 0.030263, "F": 0.14668, "G": 0.15864, "chunks": 150.2, "bytes_per_chunk": 6.817, "boundary_entropy": 0.09682, "router_noise": 0.0, "grad_norm": 0.3149, "lr": 0.0001002, "router_trainable": true, "sec_per_step": 0.165, "n_target_mean": 9.772, "alpha": 0.03, "t": 8664.4}
{"event": "train", "step": 51500, "phase": 2, "objective": "parity_a", "lm_loss": 0.88655, "bpb": 1.27902, "chunk_loss": 0.030304, "F": 0.15358, "G": 0.15945, "chunks": 157.27, "bytes_per_chunk": 6.511, "boundary_entropy": 0.08564, "router_noise": 0.0, "grad_norm": 0.3035, "lr": 9.77e-05, "router_trainable": true, "sec_per_step": 0.166, "n_target_mean": 9.6022, "alpha": 0.03, "t": 8747.3}
{"event": "probe", "step": 52000, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 3430.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.197092, "cps": 26.796875, "mean_p": 0.185797, "boundary_entropy": 0.053616, "bpb": 1.454879}, "de": {"chunks": 3719.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.181123, "cps": 29.054688, "mean_p": 0.172259, "boundary_entropy": 0.039986, "bpb": 1.565454}, "vi": {"chunks": 3818.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.207106, "cps": 29.828125, "mean_p": 0.150659, "boundary_entropy": 0.04194, "bpb": 1.218034}, "tr": {"chunks": 3032.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.167486, "cps": 23.6875, "mean_p": 0.15163, "boundary_entropy": 0.048462, "bpb": 1.484344}, "fi": {"chunks": 2962.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.157461, "cps": 23.140625, "mean_p": 0.150742, "boundary_entropy": 0.034903, "bpb": 1.604496}, "ru": {"chunks": 4572.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.241266, "cps": 35.71875, "mean_p": 0.12192, "boundary_entropy": 0.053334, "bpb": 0.94693}, "el": {"chunks": 4075.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.198558, "cps": 31.835938, "mean_p": 0.101472, "boundary_entropy": 0.047327, "bpb": 0.936455}, "hi": {"chunks": 3148.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.187103, "cps": 24.59375, "mean_p": 0.07774, "boundary_entropy": 0.044077, "bpb": 0.788154}, "ta": {"chunks": 2854.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.152661, "cps": 22.296875, "mean_p": 0.06234, "boundary_entropy": 0.051766, "bpb": 0.689891}, "th": {"chunks": 2873.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.178614, "cps": 22.445312, "mean_p": 0.073889, "boundary_entropy": 0.087592, "bpb": 0.786936}, "zh": {"chunks": 2838.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.480528, "cps": 22.171875, "mean_p": 0.32217, "boundary_entropy": 0.417856, "bpb": 2.266258}, "my": {"chunks": 2287.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.120699, "cps": 17.867188, "mean_p": 0.054616, "boundary_entropy": 0.091437, "bpb": 0.740263}}, "bpb": {"en": 1.45488, "de": 1.56545, "vi": 1.21803, "tr": 1.48434, "fi": 1.6045, "ru": 0.94693, "el": 0.93646, "hi": 0.78815, "ta": 0.68989, "th": 0.78694, "zh": 2.26626, "my": 0.74026}, "gini_cpc": 0.179505, "gini_cps": 0.102601, "bpic": 9.689482, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9189751575328005, "vi": 1.0508065767280992, "tr": 0.8497841876312239, "fi": 0.7989197747797525, "ru": 1.2241284952729679, "el": 1.007434381023694, "hi": 0.9493134868891306, "ta": 0.774566124772608, "th": 0.9062427827385673, "zh": 2.438085595614086, "my": 0.6123966251659445}, "t": 8832.9}
{"event": "eval", "step": 52000, "phase": 2, "bpb": {"en": 1.45181, "de": 1.57342, "vi": 1.23658, "tr": 1.47629, "fi": 1.62629, "ru": 0.94469, "el": 0.93625, "hi": 0.79452, "ta": 0.69645, "th": 0.78709, "zh": 2.26864, "my": 0.74478}, "bytes_per_chunk": {"en": 4.895, "de": 5.501, "vi": 6.178, "tr": 6.276, "fi": 5.988, "ru": 7.648, "el": 8.648, "hi": 12.982, "ta": 16.699, "th": 16.731, "zh": 5.669, "my": 21.515}, "bpb_mean": 1.2114, "t": 8837.9}
{"event": "train", "step": 52000, "phase": 2, "objective": "parity_a", "lm_loss": 0.88055, "bpb": 1.27037, "chunk_loss": 0.030244, "F": 0.15825, "G": 0.15846, "chunks": 162.05, "bytes_per_chunk": 6.319, "boundary_entropy": 0.07121, "router_noise": 0.0, "grad_norm": 0.3104, "lr": 9.52e-05, "router_trainable": true, "sec_per_step": 0.181, "n_target_mean": 8.9393, "alpha": 0.03, "t": 8838.0}
{"event": "train", "step": 52500, "phase": 2, "objective": "parity_a", "lm_loss": 0.89274, "bpb": 1.28796, "chunk_loss": 0.030255, "F": 0.16048, "G": 0.16341, "chunks": 164.33, "bytes_per_chunk": 6.231, "boundary_entropy": 0.07581, "router_noise": 0.0, "grad_norm": 0.3058, "lr": 9.28e-05, "router_trainable": true, "sec_per_step": 0.166, "n_target_mean": 9.0016, "alpha": 0.03, "t": 8921.1}
{"event": "train", "step": 53000, "phase": 2, "objective": "parity_a", "lm_loss": 0.90291, "bpb": 1.30263, "chunk_loss": 0.03027, "F": 0.14668, "G": 0.16065, "chunks": 150.2, "bytes_per_chunk": 6.817, "boundary_entropy": 0.10075, "router_noise": 0.0, "grad_norm": 0.3218, "lr": 9.04e-05, "router_trainable": true, "sec_per_step": 0.166, "n_target_mean": 9.7651, "alpha": 0.03, "t": 9004.1}
{"event": "phase_transition", "step": 53410, "phase": 3, "objective": "none", "router_trainable": false, "encoder_trainable": true, "frozen_router_params": 295680, "t": 9071.9}
{"event": "train", "step": 53500, "phase": 3, "objective": "none", "lm_loss": 0.9333, "bpb": 1.34647, "chunk_loss": 0.0, "F": 0.15518, "G": 0.16654, "chunks": 158.91, "bytes_per_chunk": 6.444, "boundary_entropy": 0.08935, "router_noise": 0.0, "grad_norm": 0.3168, "lr": 8.8e-05, "router_trainable": false, "sec_per_step": 0.165, "alpha": 0.0, "t": 9086.8}
{"event": "train", "step": 54000, "phase": 3, "objective": "none", "lm_loss": 0.87852, "bpb": 1.26743, "chunk_loss": 0.0, "F": 0.16084, "G": 0.16746, "chunks": 164.7, "bytes_per_chunk": 6.217, "boundary_entropy": 0.07535, "router_noise": 0.0, "grad_norm": 0.3265, "lr": 8.57e-05, "router_trainable": false, "sec_per_step": 0.164, "alpha": 0.0, "t": 9168.9}
{"event": "train", "step": 54500, "phase": 3, "objective": "none", "lm_loss": 0.92943, "bpb": 1.34088, "chunk_loss": 0.0, "F": 0.15955, "G": 0.17033, "chunks": 163.38, "bytes_per_chunk": 6.268, "boundary_entropy": 0.08215, "router_noise": 0.0, "grad_norm": 0.312, "lr": 8.34e-05, "router_trainable": false, "sec_per_step": 0.164, "alpha": 0.0, "t": 9251.0}
{"event": "train", "step": 55000, "phase": 3, "objective": "none", "lm_loss": 0.85872, "bpb": 1.23887, "chunk_loss": 0.0, "F": 0.16742, "G": 0.17161, "chunks": 171.44, "bytes_per_chunk": 5.973, "boundary_entropy": 0.06515, "router_noise": 0.0, "grad_norm": 0.3098, "lr": 8.11e-05, "router_trainable": false, "sec_per_step": 0.165, "alpha": 0.0, "t": 9333.4}
{"event": "train", "step": 55500, "phase": 3, "objective": "none", "lm_loss": 0.93677, "bpb": 1.35147, "chunk_loss": 0.0, "F": 0.16034, "G": 0.17465, "chunks": 164.19, "bytes_per_chunk": 6.237, "boundary_entropy": 0.08831, "router_noise": 0.0, "grad_norm": 0.3243, "lr": 7.89e-05, "router_trainable": false, "sec_per_step": 0.164, "alpha": 0.0, "t": 9415.6}
{"event": "probe", "step": 56000, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3442.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.197782, "cps": 26.890625, "mean_p": 0.191013, "boundary_entropy": 0.041985, "bpb": 1.440934}, "de": {"chunks": 3733.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.181805, "cps": 29.164062, "mean_p": 0.176, "boundary_entropy": 0.032157, "bpb": 1.551434}, "vi": {"chunks": 3831.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.207811, "cps": 29.929688, "mean_p": 0.154611, "boundary_entropy": 0.035017, "bpb": 1.205427}, "tr": {"chunks": 3063.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.169198, "cps": 23.929688, "mean_p": 0.155161, "boundary_entropy": 0.0453, "bpb": 1.470904}, "fi": {"chunks": 2977.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.158258, "cps": 23.257812, "mean_p": 0.153483, "boundary_entropy": 0.031808, "bpb": 1.589524}, "ru": {"chunks": 4708.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.248443, "cps": 36.78125, "mean_p": 0.131435, "boundary_entropy": 0.045412, "bpb": 0.932706}, "el": {"chunks": 4124.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.200945, "cps": 32.21875, "mean_p": 0.107651, "boundary_entropy": 0.042526, "bpb": 0.925281}, "hi": {"chunks": 4118.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.244755, "cps": 32.171875, "mean_p": 0.100596, "boundary_entropy": 0.054762, "bpb": 0.776087}, "ta": {"chunks": 3117.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.166729, "cps": 24.351562, "mean_p": 0.072386, "boundary_entropy": 0.070582, "bpb": 0.682672}, "th": {"chunks": 3399.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.211315, "cps": 26.554688, "mean_p": 0.093843, "boundary_entropy": 0.113777, "bpb": 0.774053}, "zh": {"chunks": 2762.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.46766, "cps": 21.578125, "mean_p": 0.319481, "boundary_entropy": 0.414622, "bpb": 2.253049}, "my": {"chunks": 3000.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.158328, "cps": 23.4375, "mean_p": 0.079526, "boundary_entropy": 0.145819, "bpb": 0.732354}}, "bpb": {"en": 1.44093, "de": 1.55143, "vi": 1.20543, "tr": 1.4709, "fi": 1.58952, "ru": 0.93271, "el": 0.92528, "hi": 0.77609, "ta": 0.68267, "th": 0.77405, "zh": 2.25305, "my": 0.73235}, "gini_cpc": 0.161128, "gini_cps": 0.089271, "bpic": 9.078417, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9192186707680183, "vi": 1.0507085450442508, "tr": 0.8554796825178462, "fi": 0.8001662057425509, "ru": 1.25614708267728, "el": 1.0159938185875723, "hi": 1.237498050910309, "ta": 0.8429942007692834, "th": 1.068423307541267, "zh": 2.3645226896765186, "my": 0.8005180574941518}, "t": 9500.8}
{"event": "eval", "step": 56000, "phase": 3, "bpb": {"en": 1.43889, "de": 1.56084, "vi": 1.22476, "tr": 1.46356, "fi": 1.6125, "ru": 0.93087, "el": 0.92464, "hi": 0.78226, "ta": 0.68847, "th": 0.77335, "zh": 2.25717, "my": 0.73709}, "bytes_per_chunk": {"en": 4.872, "de": 5.476, "vi": 6.155, "tr": 6.203, "fi": 5.946, "ru": 7.425, "el": 8.539, "hi": 10.004, "ta": 15.362, "th": 13.633, "zh": 5.819, "my": 16.793}, "bpb_mean": 1.19953, "t": 9505.8}
{"event": "train", "step": 56000, "phase": 3, "objective": "none", "lm_loss": 0.92057, "bpb": 1.3281, "chunk_loss": 0.0, "F": 0.16037, "G": 0.17909, "chunks": 164.22, "bytes_per_chunk": 6.236, "boundary_entropy": 0.09846, "router_noise": 0.0, "grad_norm": 0.3346, "lr": 7.68e-05, "router_trainable": false, "sec_per_step": 0.181, "alpha": 0.0, "t": 9505.9}
{"event": "train", "step": 56500, "phase": 3, "objective": "none", "lm_loss": 0.90212, "bpb": 1.30149, "chunk_loss": 0.0, "F": 0.159, "G": 0.17855, "chunks": 162.81, "bytes_per_chunk": 6.289, "boundary_entropy": 0.0998, "router_noise": 0.0, "grad_norm": 0.3148, "lr": 7.46e-05, "router_trainable": false, "sec_per_step": 0.165, "alpha": 0.0, "t": 9588.3}
{"event": "train", "step": 57000, "phase": 3, "objective": "none", "lm_loss": 0.92201, "bpb": 1.33018, "chunk_loss": 0.0, "F": 0.1671, "G": 0.18373, "chunks": 171.11, "bytes_per_chunk": 5.984, "boundary_entropy": 0.09567, "router_noise": 0.0, "grad_norm": 0.3251, "lr": 7.25e-05, "router_trainable": false, "sec_per_step": 0.165, "alpha": 0.0, "t": 9670.6}
{"event": "train", "step": 57500, "phase": 3, "objective": "none", "lm_loss": 0.8781, "bpb": 1.26683, "chunk_loss": 0.0, "F": 0.1712, "G": 0.17252, "chunks": 175.31, "bytes_per_chunk": 5.841, "boundary_entropy": 0.05719, "router_noise": 0.0, "grad_norm": 0.3139, "lr": 7.05e-05, "router_trainable": false, "sec_per_step": 0.165, "alpha": 0.0, "t": 9752.9}
{"event": "train", "step": 58000, "phase": 3, "objective": "none", "lm_loss": 0.93061, "bpb": 1.34258, "chunk_loss": 0.0, "F": 0.16837, "G": 0.18681, "chunks": 172.41, "bytes_per_chunk": 5.939, "boundary_entropy": 0.09954, "router_noise": 0.0, "grad_norm": 0.3353, "lr": 6.84e-05, "router_trainable": false, "sec_per_step": 0.164, "alpha": 0.0, "t": 9835.0}
{"event": "train", "step": 58500, "phase": 3, "objective": "none", "lm_loss": 0.91599, "bpb": 1.32149, "chunk_loss": 0.0, "F": 0.16646, "G": 0.18616, "chunks": 170.45, "bytes_per_chunk": 6.008, "boundary_entropy": 0.0994, "router_noise": 0.0, "grad_norm": 0.3372, "lr": 6.65e-05, "router_trainable": false, "sec_per_step": 0.165, "alpha": 0.0, "t": 9917.7}
{"event": "train", "step": 59000, "phase": 3, "objective": "none", "lm_loss": 0.94705, "bpb": 1.3663, "chunk_loss": 0.0, "F": 0.168, "G": 0.18135, "chunks": 172.03, "bytes_per_chunk": 5.952, "boundary_entropy": 0.0854, "router_noise": 0.0, "grad_norm": 0.3632, "lr": 6.45e-05, "router_trainable": false, "sec_per_step": 0.165, "alpha": 0.0, "t": 10000.1}
{"event": "train", "step": 59500, "phase": 3, "objective": "none", "lm_loss": 0.9065, "bpb": 1.3078, "chunk_loss": 0.0, "F": 0.15648, "G": 0.17273, "chunks": 160.23, "bytes_per_chunk": 6.391, "boundary_entropy": 0.09159, "router_noise": 0.0, "grad_norm": 0.3838, "lr": 6.27e-05, "router_trainable": false, "sec_per_step": 0.164, "alpha": 0.0, "t": 10082.3}
{"event": "probe", "step": 60000, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3447.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.198069, "cps": 26.929688, "mean_p": 0.191967, "boundary_entropy": 0.039622, "bpb": 1.432026}, "de": {"chunks": 3725.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.181415, "cps": 29.101562, "mean_p": 0.176071, "boundary_entropy": 0.030548, "bpb": 1.54146}, "vi": {"chunks": 3861.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.209439, "cps": 30.164062, "mean_p": 0.156184, "boundary_entropy": 0.033994, "bpb": 1.194469}, "tr": {"chunks": 3069.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.16953, "cps": 23.976562, "mean_p": 0.15597, "boundary_entropy": 0.042292, "bpb": 1.458199}, "fi": {"chunks": 2951.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.156876, "cps": 23.054688, "mean_p": 0.151941, "boundary_entropy": 0.03048, "bpb": 1.574471}, "ru": {"chunks": 4738.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.250026, "cps": 37.015625, "mean_p": 0.133724, "boundary_entropy": 0.043758, "bpb": 0.922651}, "el": {"chunks": 4137.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.201579, "cps": 32.320312, "mean_p": 0.108658, "boundary_entropy": 0.04156, "bpb": 0.916671}, "hi": {"chunks": 4300.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.255572, "cps": 33.59375, "mean_p": 0.105639, "boundary_entropy": 0.05978, "bpb": 0.766613}, "ta": {"chunks": 3321.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.177641, "cps": 25.945312, "mean_p": 0.077377, "boundary_entropy": 0.071911, "bpb": 0.674264}, "th": {"chunks": 3683.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.228971, "cps": 28.773438, "mean_p": 0.103169, "boundary_entropy": 0.128238, "bpb": 0.765128}, "zh": {"chunks": 2645.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.44785, "cps": 20.664062, "mean_p": 0.311593, "boundary_entropy": 0.408431, "bpb": 2.240262}, "my": {"chunks": 2647.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.139698, "cps": 20.679688, "mean_p": 0.075182, "boundary_entropy": 0.141192, "bpb": 0.72417}}, "bpb": {"en": 1.43203, "de": 1.54146, "vi": 1.19447, "tr": 1.4582, "fi": 1.57447, "ru": 0.92265, "el": 0.91667, "hi": 0.76661, "ta": 0.67426, "th": 0.76513, "zh": 2.24026, "my": 0.72417}, "gini_cpc": 0.165148, "gini_cps": 0.101139, "bpic": 9.025045, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.915918237626946, "vi": 1.0574004635580097, "tr": 0.8559121153123072, "fi": 0.7920273210140966, "ru": 1.2623177203533644, "el": 1.017718132963984, "hi": 1.290316406669279, "ta": 0.896863279370575, "th": 1.1560150560028006, "zh": 2.2610755329285586, "my": 0.7052992175185883}, "t": 10167.4}
{"event": "eval", "step": 60000, "phase": 3, "bpb": {"en": 1.42979, "de": 1.54943, "vi": 1.21298, "tr": 1.45085, "fi": 1.59582, "ru": 0.92036, "el": 0.91657, "hi": 0.77294, "ta": 0.68042, "th": 0.7649, "zh": 2.24434, "my": 0.72991}, "bytes_per_chunk": {"en": 4.862, "de": 5.48, "vi": 6.097, "tr": 6.178, "fi": 5.981, "ru": 7.342, "el": 8.497, "hi": 9.588, "ta": 14.475, "th": 12.498, "zh": 6.036, "my": 18.334}, "bpb_mean": 1.18902, "t": 10172.4}
{"event": "train", "step": 60000, "phase": 3, "objective": "none", "lm_loss": 0.91516, "bpb": 1.3203, "chunk_loss": 0.0, "F": 0.16165, "G": 0.17855, "chunks": 165.53, "bytes_per_chunk": 6.186, "boundary_entropy": 0.09099, "router_noise": 0.0, "grad_norm": 0.3347, "lr": 6.08e-05, "router_trainable": false, "sec_per_step": 0.181, "alpha": 0.0, "t": 10172.6}
{"event": "train", "step": 60500, "phase": 3, "objective": "none", "lm_loss": 0.83631, "bpb": 1.20655, "chunk_loss": 0.0, "F": 0.15944, "G": 0.16743, "chunks": 163.27, "bytes_per_chunk": 6.272, "boundary_entropy": 0.07154, "router_noise": 0.0, "grad_norm": 0.3277, "lr": 5.9e-05, "router_trainable": false, "sec_per_step": 0.164, "alpha": 0.0, "t": 10254.7}
{"event": "train", "step": 61000, "phase": 3, "objective": "none", "lm_loss": 0.83993, "bpb": 1.21177, "chunk_loss": 0.0, "F": 0.15997, "G": 0.17781, "chunks": 163.81, "bytes_per_chunk": 6.251, "boundary_entropy": 0.09561, "router_noise": 0.0, "grad_norm": 0.3322, "lr": 5.73e-05, "router_trainable": false, "sec_per_step": 0.165, "alpha": 0.0, "t": 10337.4}
{"event": "train", "step": 61500, "phase": 3, "objective": "none", "lm_loss": 0.90531, "bpb": 1.30608, "chunk_loss": 0.0, "F": 0.16473, "G": 0.16753, "chunks": 168.69, "bytes_per_chunk": 6.07, "boundary_entropy": 0.05789, "router_noise": 0.0, "grad_norm": 0.3266, "lr": 5.56e-05, "router_trainable": false, "sec_per_step": 0.165, "alpha": 0.0, "t": 10419.9}
{"event": "train", "step": 62000, "phase": 3, "objective": "none", "lm_loss": 0.89631, "bpb": 1.2931, "chunk_loss": 0.0, "F": 0.1615, "G": 0.17359, "chunks": 165.38, "bytes_per_chunk": 6.192, "boundary_entropy": 0.07677, "router_noise": 0.0, "grad_norm": 0.3363, "lr": 5.39e-05, "router_trainable": false, "sec_per_step": 0.164, "alpha": 0.0, "t": 10502.0}
{"event": "train", "step": 62500, "phase": 3, "objective": "none", "lm_loss": 0.92398, "bpb": 1.33302, "chunk_loss": 0.0, "F": 0.16072, "G": 0.18373, "chunks": 164.58, "bytes_per_chunk": 6.222, "boundary_entropy": 0.10581, "router_noise": 0.0, "grad_norm": 0.3529, "lr": 5.23e-05, "router_trainable": false, "sec_per_step": 0.165, "alpha": 0.0, "t": 10584.4}
{"event": "train", "step": 63000, "phase": 3, "objective": "none", "lm_loss": 0.89427, "bpb": 1.29016, "chunk_loss": 0.0, "F": 0.16904, "G": 0.17998, "chunks": 173.09, "bytes_per_chunk": 5.916, "boundary_entropy": 0.07612, "router_noise": 0.0, "grad_norm": 0.3325, "lr": 5.08e-05, "router_trainable": false, "sec_per_step": 0.164, "alpha": 0.0, "t": 10666.6}
{"event": "train", "step": 63500, "phase": 3, "objective": "none", "lm_loss": 0.81259, "bpb": 1.17232, "chunk_loss": 0.0, "F": 0.15897, "G": 0.16819, "chunks": 162.78, "bytes_per_chunk": 6.291, "boundary_entropy": 0.07394, "router_noise": 0.0, "grad_norm": 0.3289, "lr": 4.93e-05, "router_trainable": false, "sec_per_step": 0.165, "alpha": 0.0, "t": 10749.2}
{"event": "probe", "step": 64000, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3450.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.198242, "cps": 26.953125, "mean_p": 0.192691, "boundary_entropy": 0.038515, "bpb": 1.424001}, "de": {"chunks": 3730.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.181659, "cps": 29.140625, "mean_p": 0.176987, "boundary_entropy": 0.030078, "bpb": 1.531061}, "vi": {"chunks": 3877.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.210306, "cps": 30.289062, "mean_p": 0.157395, "boundary_entropy": 0.033369, "bpb": 1.183761}, "tr": {"chunks": 3062.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.169143, "cps": 23.921875, "mean_p": 0.155687, "boundary_entropy": 0.042614, "bpb": 1.446158}, "fi": {"chunks": 2957.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.157195, "cps": 23.101562, "mean_p": 0.152242, "boundary_entropy": 0.030063, "bpb": 1.562108}, "ru": {"chunks": 4717.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.248918, "cps": 36.851562, "mean_p": 0.133312, "boundary_entropy": 0.042238, "bpb": 0.916733}, "el": {"chunks": 4136.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.20153, "cps": 32.3125, "mean_p": 0.109693, "boundary_entropy": 0.03987, "bpb": 0.908425}, "hi": {"chunks": 4503.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.267637, "cps": 35.179688, "mean_p": 0.113287, "boundary_entropy": 0.069244, "bpb": 0.758864}, "ta": {"chunks": 3506.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.187537, "cps": 27.390625, "mean_p": 0.079997, "boundary_entropy": 0.076609, "bpb": 0.669469}, "th": {"chunks": 3809.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.236804, "cps": 29.757812, "mean_p": 0.102949, "boundary_entropy": 0.12347, "bpb": 0.758065}, "zh": {"chunks": 2823.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.477988, "cps": 22.054688, "mean_p": 0.322879, "boundary_entropy": 0.414587, "bpb": 2.229763}, "my": {"chunks": 3016.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.159172, "cps": 23.5625, "mean_p": 0.083263, "boundary_entropy": 0.147765, "bpb": 0.71812}}, "bpb": {"en": 1.424, "de": 1.53106, "vi": 1.18376, "tr": 1.44616, "fi": 1.56211, "ru": 0.91673, "el": 0.90842, "hi": 0.75886, "ta": 0.66947, "th": 0.75807, "zh": 2.22976, "my": 0.71812}, "gini_cpc": 0.167235, "gini_cps": 0.091704, "bpic": 8.805144, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9163501383774582, "vi": 1.0608590464735086, "tr": 0.8532173144494426, "fi": 0.7929475584359753, "ru": 1.2556300103246532, "el": 1.0165873706912076, "hi": 1.350056360229989, "ta": 0.9460007209553822, "th": 1.1945241448284252, "zh": 2.4111401816870095, "my": 0.802921313250911}, "t": 10834.8}
{"event": "eval", "step": 64000, "phase": 3, "bpb": {"en": 1.42148, "de": 1.53917, "vi": 1.20256, "tr": 1.43949, "fi": 1.58295, "ru": 0.91424, "el": 0.90832, "hi": 0.76516, "ta": 0.67534, "th": 0.75831, "zh": 2.23281, "my": 0.72329}, "bytes_per_chunk": {"en": 4.857, "de": 5.472, "vi": 6.065, "tr": 6.202, "fi": 5.985, "ru": 7.368, "el": 8.502, "hi": 9.21, "ta": 13.898, "th": 12.039, "zh": 5.739, "my": 16.826}, "bpb_mean": 1.18026, "t": 10839.7}
{"event": "train", "step": 64000, "phase": 3, "objective": "none", "lm_loss": 0.88803, "bpb": 1.28116, "chunk_loss": 0.0, "F": 0.16455, "G": 0.17203, "chunks": 168.5, "bytes_per_chunk": 6.077, "boundary_entropy": 0.06556, "router_noise": 0.0, "grad_norm": 0.3409, "lr": 4.78e-05, "router_trainable": false, "sec_per_step": 0.181, "alpha": 0.0, "t": 10839.9}
{"event": "train", "step": 64500, "phase": 3, "objective": "none", "lm_loss": 0.84447, "bpb": 1.21832, "chunk_loss": 0.0, "F": 0.15683, "G": 0.17173, "chunks": 160.59, "bytes_per_chunk": 6.376, "boundary_entropy": 0.08828, "router_noise": 0.0, "grad_norm": 0.333, "lr": 4.65e-05, "router_trainable": false, "sec_per_step": 0.166, "alpha": 0.0, "t": 10923.0}
{"event": "train", "step": 65000, "phase": 3, "objective": "none", "lm_loss": 0.88722, "bpb": 1.27999, "chunk_loss": 0.0, "F": 0.16338, "G": 0.17462, "chunks": 167.3, "bytes_per_chunk": 6.121, "boundary_entropy": 0.08041, "router_noise": 0.0, "grad_norm": 0.3368, "lr": 4.51e-05, "router_trainable": false, "sec_per_step": 0.165, "alpha": 0.0, "t": 11005.7}
{"event": "train", "step": 65500, "phase": 3, "objective": "none", "lm_loss": 0.85151, "bpb": 1.22847, "chunk_loss": 0.0, "F": 0.15604, "G": 0.17047, "chunks": 159.78, "bytes_per_chunk": 6.409, "boundary_entropy": 0.08833, "router_noise": 0.0, "grad_norm": 0.3498, "lr": 4.38e-05, "router_trainable": false, "sec_per_step": 0.165, "alpha": 0.0, "t": 11088.0}
{"event": "train", "step": 66000, "phase": 3, "objective": "none", "lm_loss": 0.84518, "bpb": 1.21933, "chunk_loss": 0.0, "F": 0.14877, "G": 0.16338, "chunks": 152.34, "bytes_per_chunk": 6.722, "boundary_entropy": 0.08138, "router_noise": 0.0, "grad_norm": 0.3402, "lr": 4.26e-05, "router_trainable": false, "sec_per_step": 0.165, "alpha": 0.0, "t": 11170.5}
{"event": "train", "step": 66500, "phase": 3, "objective": "none", "lm_loss": 0.93337, "bpb": 1.34657, "chunk_loss": 0.0, "F": 0.16859, "G": 0.18504, "chunks": 172.64, "bytes_per_chunk": 5.931, "boundary_entropy": 0.0895, "router_noise": 0.0, "grad_norm": 0.3505, "lr": 4.14e-05, "router_trainable": false, "sec_per_step": 0.165, "alpha": 0.0, "t": 11252.8}
{"event": "train", "step": 67000, "phase": 3, "objective": "none", "lm_loss": 0.84615, "bpb": 1.22074, "chunk_loss": 0.0, "F": 0.16354, "G": 0.18116, "chunks": 167.47, "bytes_per_chunk": 6.115, "boundary_entropy": 0.09352, "router_noise": 0.0, "grad_norm": 0.3485, "lr": 4.03e-05, "router_trainable": false, "sec_per_step": 0.164, "alpha": 0.0, "t": 11335.0}
{"event": "train", "step": 67500, "phase": 3, "objective": "none", "lm_loss": 0.85512, "bpb": 1.23368, "chunk_loss": 0.0, "F": 0.15523, "G": 0.17493, "chunks": 158.95, "bytes_per_chunk": 6.442, "boundary_entropy": 0.09785, "router_noise": 0.0, "grad_norm": 0.3448, "lr": 3.92e-05, "router_trainable": false, "sec_per_step": 0.164, "alpha": 0.0, "t": 11417.1}
{"event": "probe", "step": 68000, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3445.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.197954, "cps": 26.914062, "mean_p": 0.192759, "boundary_entropy": 0.038443, "bpb": 1.417251}, "de": {"chunks": 3735.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.181902, "cps": 29.179688, "mean_p": 0.177117, "boundary_entropy": 0.030367, "bpb": 1.52293}, "vi": {"chunks": 3876.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.210252, "cps": 30.28125, "mean_p": 0.157205, "boundary_entropy": 0.033012, "bpb": 1.17603}, "tr": {"chunks": 3073.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.169751, "cps": 24.007812, "mean_p": 0.15655, "boundary_entropy": 0.045239, "bpb": 1.439735}, "fi": {"chunks": 2962.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.157461, "cps": 23.140625, "mean_p": 0.152723, "boundary_entropy": 0.03127, "bpb": 1.551272}, "ru": {"chunks": 4779.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.25219, "cps": 37.335938, "mean_p": 0.135974, "boundary_entropy": 0.044298, "bpb": 0.909242}, "el": {"chunks": 4150.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.202212, "cps": 32.421875, "mean_p": 0.110253, "boundary_entropy": 0.040452, "bpb": 0.902569}, "hi": {"chunks": 4736.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.281486, "cps": 37.0, "mean_p": 0.118465, "boundary_entropy": 0.073189, "bpb": 0.75203}, "ta": {"chunks": 3699.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.19786, "cps": 28.898438, "mean_p": 0.086558, "boundary_entropy": 0.085112, "bpb": 0.663478}, "th": {"chunks": 3959.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.24613, "cps": 30.929688, "mean_p": 0.110347, "boundary_entropy": 0.136885, "bpb": 0.751929}, "zh": {"chunks": 2713.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.459363, "cps": 21.195312, "mean_p": 0.318191, "boundary_entropy": 0.412448, "bpb": 2.218682}, "my": {"chunks": 3073.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.162181, "cps": 24.007812, "mean_p": 0.086838, "boundary_entropy": 0.158207, "bpb": 0.712654}}, "bpb": {"en": 1.41725, "de": 1.52293, "vi": 1.17603, "tr": 1.43973, "fi": 1.55127, "ru": 0.90924, "el": 0.90257, "hi": 0.75203, "ta": 0.66348, "th": 0.75193, "zh": 2.21868, "my": 0.71265}, "gini_cpc": 0.163504, "gini_cps": 0.09802, "bpic": 8.682828, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9189102437458283, "vi": 1.0621247288259892, "tr": 0.8575252223468564, "fi": 0.7954411690840497, "ru": 1.2739802934159485, "el": 1.0215088781060324, "hi": 1.4219735387548336, "ta": 0.9995252178523865, "th": 1.2433669822825313, "zh": 2.3205516812255085, "my": 0.8192832541769652}, "t": 11502.2}
{"event": "eval", "step": 68000, "phase": 3, "bpb": {"en": 1.41417, "de": 1.53123, "vi": 1.19468, "tr": 1.43224, "fi": 1.57345, "ru": 0.90658, "el": 0.90203, "hi": 0.75842, "ta": 0.66909, "th": 0.75223, "zh": 2.22246, "my": 0.71846}, "bytes_per_chunk": {"en": 4.858, "de": 5.475, "vi": 6.076, "tr": 6.163, "fi": 5.975, "ru": 7.269, "el": 8.477, "hi": 8.771, "ta": 13.125, "th": 11.572, "zh": 5.908, "my": 16.123}, "bpb_mean": 1.17292, "t": 11507.2}
{"event": "train", "step": 68000, "phase": 3, "objective": "none", "lm_loss": 0.89904, "bpb": 1.29704, "chunk_loss": 0.0, "F": 0.16162, "G": 0.17905, "chunks": 165.5, "bytes_per_chunk": 6.187, "boundary_entropy": 0.09337, "router_noise": 0.0, "grad_norm": 0.3406, "lr": 3.82e-05, "router_trainable": false, "sec_per_step": 0.18, "alpha": 0.0, "t": 11507.4}
{"event": "train", "step": 68500, "phase": 3, "objective": "none", "lm_loss": 0.89502, "bpb": 1.29124, "chunk_loss": 0.0, "F": 0.16209, "G": 0.17364, "chunks": 165.98, "bytes_per_chunk": 6.169, "boundary_entropy": 0.07708, "router_noise": 0.0, "grad_norm": 0.3414, "lr": 3.73e-05, "router_trainable": false, "sec_per_step": 0.164, "alpha": 0.0, "t": 11589.5}
{"event": "train", "step": 69000, "phase": 3, "objective": "none", "lm_loss": 0.91236, "bpb": 1.31626, "chunk_loss": 0.0, "F": 0.16107, "G": 0.18726, "chunks": 164.94, "bytes_per_chunk": 6.208, "boundary_entropy": 0.11487, "router_noise": 0.0, "grad_norm": 0.3578, "lr": 3.64e-05, "router_trainable": false, "sec_per_step": 0.165, "alpha": 0.0, "t": 11672.1}
{"event": "train", "step": 69500, "phase": 3, "objective": "none", "lm_loss": 0.85894, "bpb": 1.2392, "chunk_loss": 0.0, "F": 0.16307, "G": 0.17706, "chunks": 166.98, "bytes_per_chunk": 6.132, "boundary_entropy": 0.08021, "router_noise": 0.0, "grad_norm": 0.3561, "lr": 3.55e-05, "router_trainable": false, "sec_per_step": 0.165, "alpha": 0.0, "t": 11754.4}
{"event": "train", "step": 70000, "phase": 3, "objective": "none", "lm_loss": 0.92386, "bpb": 1.33285, "chunk_loss": 0.0, "F": 0.16667, "G": 0.18433, "chunks": 170.67, "bytes_per_chunk": 6.0, "boundary_entropy": 0.09205, "router_noise": 0.0, "grad_norm": 0.3558, "lr": 3.48e-05, "router_trainable": false, "sec_per_step": 0.165, "alpha": 0.0, "t": 11836.9}
{"event": "train", "step": 70500, "phase": 3, "objective": "none", "lm_loss": 0.90342, "bpb": 1.30335, "chunk_loss": 0.0, "F": 0.16685, "G": 0.17754, "chunks": 170.86, "bytes_per_chunk": 5.993, "boundary_entropy": 0.07446, "router_noise": 0.0, "grad_norm": 0.3542, "lr": 3.4e-05, "router_trainable": false, "sec_per_step": 0.165, "alpha": 0.0, "t": 11919.3}
{"event": "train", "step": 71000, "phase": 3, "objective": "none", "lm_loss": 0.92561, "bpb": 1.33537, "chunk_loss": 0.0, "F": 0.17583, "G": 0.18206, "chunks": 180.05, "bytes_per_chunk": 5.687, "boundary_entropy": 0.06605, "router_noise": 0.0, "grad_norm": 0.3568, "lr": 3.34e-05, "router_trainable": false, "sec_per_step": 0.165, "alpha": 0.0, "t": 12002.0}
{"event": "train", "step": 71500, "phase": 3, "objective": "none", "lm_loss": 0.86738, "bpb": 1.25137, "chunk_loss": 0.0, "F": 0.15942, "G": 0.18226, "chunks": 163.25, "bytes_per_chunk": 6.273, "boundary_entropy": 0.10505, "router_noise": 0.0, "grad_norm": 0.3637, "lr": 3.28e-05, "router_trainable": false, "sec_per_step": 0.165, "alpha": 0.0, "t": 12084.3}
{"event": "probe", "step": 72000, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3454.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.198472, "cps": 26.984375, "mean_p": 0.193737, "boundary_entropy": 0.03739, "bpb": 1.411245}, "de": {"chunks": 3736.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.181951, "cps": 29.1875, "mean_p": 0.177699, "boundary_entropy": 0.030126, "bpb": 1.517903}, "vi": {"chunks": 3891.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.211066, "cps": 30.398438, "mean_p": 0.158435, "boundary_entropy": 0.032679, "bpb": 1.170198}, "tr": {"chunks": 3067.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.169419, "cps": 23.960938, "mean_p": 0.156563, "boundary_entropy": 0.043105, "bpb": 1.433672}, "fi": {"chunks": 2963.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.157514, "cps": 23.148438, "mean_p": 0.15297, "boundary_entropy": 0.031091, "bpb": 1.544071}, "ru": {"chunks": 4731.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.249657, "cps": 36.960938, "mean_p": 0.134823, "boundary_entropy": 0.043368, "bpb": 0.905257}, "el": {"chunks": 4155.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.202456, "cps": 32.460938, "mean_p": 0.111095, "boundary_entropy": 0.041221, "bpb": 0.896636}, "hi": {"chunks": 4818.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.28636, "cps": 37.640625, "mean_p": 0.12092, "boundary_entropy": 0.076578, "bpb": 0.74853}, "ta": {"chunks": 3632.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.194277, "cps": 28.375, "mean_p": 0.08402, "boundary_entropy": 0.081965, "bpb": 0.660561}, "th": {"chunks": 3977.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.247249, "cps": 31.070312, "mean_p": 0.109331, "boundary_entropy": 0.133721, "bpb": 0.748381}, "zh": {"chunks": 2745.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.464782, "cps": 21.445312, "mean_p": 0.321818, "boundary_entropy": 0.416099, "bpb": 2.210615}, "my": {"chunks": 2959.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.156164, "cps": 23.117188, "mean_p": 0.087311, "boundary_entropy": 0.161696, "bpb": 0.708677}}, "bpb": {"en": 1.41125, "de": 1.5179, "vi": 1.1702, "tr": 1.43367, "fi": 1.54407, "ru": 0.90526, "el": 0.89664, "hi": 0.74853, "ta": 0.66056, "th": 0.74838, "zh": 2.21061, "my": 0.70868}, "gini_cpc": 0.168449, "gini_cps": 0.100201, "bpic": 8.696995, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9167612484553584, "vi": 1.063456856898265, "tr": 0.8536208444498622, "fi": 0.7936363571721593, "ru": 1.2578982725087964, "el": 1.0200746850857603, "hi": 1.4428245047841683, "ta": 0.9788635508009366, "th": 1.2457655380659962, "zh": 2.3418048068651687, "my": 0.7868344675135034}, "t": 12169.7}
{"event": "eval", "step": 72000, "phase": 3, "bpb": {"en": 1.40848, "de": 1.5264, "vi": 1.18849, "tr": 1.42636, "fi": 1.56685, "ru": 0.90262, "el": 0.89647, "hi": 0.75492, "ta": 0.66659, "th": 0.74845, "zh": 2.21474, "my": 0.71383}, "bytes_per_chunk": {"en": 4.848, "de": 5.466, "vi": 6.061, "tr": 6.167, "fi": 5.971, "ru": 7.344, "el": 8.473, "hi": 8.616, "ta": 13.315, "th": 11.521, "zh": 5.843, "my": 16.772}, "bpb_mean": 1.16785, "t": 12174.7}
{"event": "train", "step": 72000, "phase": 3, "objective": "none", "lm_loss": 0.86033, "bpb": 1.24119, "chunk_loss": 0.0, "F": 0.15854, "G": 0.17595, "chunks": 162.34, "bytes_per_chunk": 6.308, "boundary_entropy": 0.0923, "router_noise": 0.0, "grad_norm": 0.3633, "lr": 3.22e-05, "router_trainable": false, "sec_per_step": 0.181, "alpha": 0.0, "t": 12174.9}
{"event": "train", "step": 72500, "phase": 3, "objective": "none", "lm_loss": 0.83595, "bpb": 1.20602, "chunk_loss": 0.0, "F": 0.15707, "G": 0.17391, "chunks": 160.84, "bytes_per_chunk": 6.366, "boundary_entropy": 0.09306, "router_noise": 0.0, "grad_norm": 0.3657, "lr": 3.17e-05, "router_trainable": false, "sec_per_step": 0.166, "alpha": 0.0, "t": 12257.6}
{"event": "train", "step": 73000, "phase": 3, "objective": "none", "lm_loss": 0.93459, "bpb": 1.34833, "chunk_loss": 0.0, "F": 0.17168, "G": 0.19457, "chunks": 175.8, "bytes_per_chunk": 5.825, "boundary_entropy": 0.10278, "router_noise": 0.0, "grad_norm": 0.3796, "lr": 3.13e-05, "router_trainable": false, "sec_per_step": 0.166, "alpha": 0.0, "t": 12340.7}
{"event": "train", "step": 73500, "phase": 3, "objective": "none", "lm_loss": 0.93998, "bpb": 1.35611, "chunk_loss": 0.0, "F": 0.16776, "G": 0.1864, "chunks": 171.78, "bytes_per_chunk": 5.961, "boundary_entropy": 0.09059, "router_noise": 0.0, "grad_norm": 0.3863, "lr": 3.09e-05, "router_trainable": false, "sec_per_step": 0.166, "alpha": 0.0, "t": 12423.9}
{"event": "train", "step": 74000, "phase": 3, "objective": "none", "lm_loss": 0.90358, "bpb": 1.30359, "chunk_loss": 0.0, "F": 0.1633, "G": 0.18421, "chunks": 167.22, "bytes_per_chunk": 6.124, "boundary_entropy": 0.09571, "router_noise": 0.0, "grad_norm": 0.3724, "lr": 3.06e-05, "router_trainable": false, "sec_per_step": 0.166, "alpha": 0.0, "t": 12506.9}
{"event": "train", "step": 74500, "phase": 3, "objective": "none", "lm_loss": 0.85951, "bpb": 1.24001, "chunk_loss": 0.0, "F": 0.16502, "G": 0.1834, "chunks": 168.98, "bytes_per_chunk": 6.06, "boundary_entropy": 0.09423, "router_noise": 0.0, "grad_norm": 0.3662, "lr": 3.04e-05, "router_trainable": false, "sec_per_step": 0.166, "alpha": 0.0, "t": 12590.0}
{"event": "train", "step": 75000, "phase": 3, "objective": "none", "lm_loss": 0.85103, "bpb": 1.22778, "chunk_loss": 0.0, "F": 0.16345, "G": 0.1814, "chunks": 167.38, "bytes_per_chunk": 6.118, "boundary_entropy": 0.09066, "router_noise": 0.0, "grad_norm": 0.3857, "lr": 3.02e-05, "router_trainable": false, "sec_per_step": 0.165, "alpha": 0.0, "t": 12672.7}
{"event": "train", "step": 75500, "phase": 3, "objective": "none", "lm_loss": 0.88769, "bpb": 1.28066, "chunk_loss": 0.0, "F": 0.15997, "G": 0.19105, "chunks": 163.81, "bytes_per_chunk": 6.251, "boundary_entropy": 0.12343, "router_noise": 0.0, "grad_norm": 0.389, "lr": 3.01e-05, "router_trainable": false, "sec_per_step": 0.165, "alpha": 0.0, "t": 12755.1}
{"event": "probe", "step": 76000, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3458.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.198701, "cps": 27.015625, "mean_p": 0.194638, "boundary_entropy": 0.037117, "bpb": 1.407231}, "de": {"chunks": 3749.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.182584, "cps": 29.289062, "mean_p": 0.178755, "boundary_entropy": 0.030348, "bpb": 1.514795}, "vi": {"chunks": 3898.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.211446, "cps": 30.453125, "mean_p": 0.158984, "boundary_entropy": 0.032576, "bpb": 1.166192}, "tr": {"chunks": 3084.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.170359, "cps": 24.09375, "mean_p": 0.158625, "boundary_entropy": 0.046441, "bpb": 1.426971}, "fi": {"chunks": 2978.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.158312, "cps": 23.265625, "mean_p": 0.15436, "boundary_entropy": 0.032313, "bpb": 1.53766}, "ru": {"chunks": 4764.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.251398, "cps": 37.21875, "mean_p": 0.135707, "boundary_entropy": 0.044105, "bpb": 0.902345}, "el": {"chunks": 4157.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.202553, "cps": 32.476562, "mean_p": 0.111626, "boundary_entropy": 0.041278, "bpb": 0.893483}, "hi": {"chunks": 5051.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.300208, "cps": 39.460938, "mean_p": 0.127184, "boundary_entropy": 0.082295, "bpb": 0.744823}, "ta": {"chunks": 3842.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.205509, "cps": 30.015625, "mean_p": 0.092892, "boundary_entropy": 0.098181, "bpb": 0.656636}, "th": {"chunks": 4211.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.261797, "cps": 32.898438, "mean_p": 0.116185, "boundary_entropy": 0.142832, "bpb": 0.744026}, "zh": {"chunks": 2805.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.474941, "cps": 21.914062, "mean_p": 0.326409, "boundary_entropy": 0.420292, "bpb": 2.205107}, "my": {"chunks": 3008.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.15875, "cps": 23.5, "mean_p": 0.09122, "boundary_entropy": 0.169156, "bpb": 0.707754}}, "bpb": {"en": 1.40723, "de": 1.51479, "vi": 1.16619, "tr": 1.42697, "fi": 1.53766, "ru": 0.90235, "el": 0.89348, "hi": 0.74482, "ta": 0.65664, "th": 0.74403, "zh": 2.20511, "my": 0.70775}, "gini_cpc": 0.173517, "gini_cps": 0.103653, "bpic": 8.527519, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9188871209226119, "vi": 1.0641376866463588, "tr": 0.8573594699593008, "fi": 0.7967314142116678, "ru": 1.2652072438046607, "el": 1.019385168643895, "hi": 1.5108502711802938, "ta": 1.0342630807066837, "th": 1.3175384780787003, "zh": 2.390223732636444, "my": 0.7989389364676855}, "t": 12840.7}
{"event": "eval", "step": 76000, "phase": 3, "bpb": {"en": 1.4049, "de": 1.52307, "vi": 1.18496, "tr": 1.42009, "fi": 1.56068, "ru": 0.90038, "el": 0.89331, "hi": 0.75101, "ta": 0.6626, "th": 0.74419, "zh": 2.20929, "my": 0.71246}, "bytes_per_chunk": {"en": 4.839, "de": 5.452, "vi": 6.039, "tr": 6.106, "fi": 5.953, "ru": 7.296, "el": 8.463, "hi": 8.263, "ta": 12.657, "th": 10.891, "zh": 5.794, "my": 16.514}, "bpb_mean": 1.16391, "t": 12845.7}
{"event": "train", "step": 76000, "phase": 3, "objective": "none", "lm_loss": 0.88696, "bpb": 1.27961, "chunk_loss": 0.0, "F": 0.1642, "G": 0.19081, "chunks": 168.14, "bytes_per_chunk": 6.09, "boundary_entropy": 0.11646, "router_noise": 0.0, "grad_norm": 0.3815, "lr": 3e-05, "router_trainable": false, "sec_per_step": 0.182, "alpha": 0.0, "t": 12845.9}
{"event": "probe", "step": 76300, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3458.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.198701, "cps": 27.015625, "mean_p": 0.194419, "boundary_entropy": 0.036894, "bpb": 1.407074}, "de": {"chunks": 3746.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.182438, "cps": 29.265625, "mean_p": 0.178313, "boundary_entropy": 0.029633, "bpb": 1.513831}, "vi": {"chunks": 3896.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.211337, "cps": 30.4375, "mean_p": 0.159007, "boundary_entropy": 0.032989, "bpb": 1.165716}, "tr": {"chunks": 3058.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.168922, "cps": 23.890625, "mean_p": 0.157186, "boundary_entropy": 0.045489, "bpb": 1.429422}, "fi": {"chunks": 2957.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.157195, "cps": 23.101562, "mean_p": 0.152876, "boundary_entropy": 0.03107, "bpb": 1.538176}, "ru": {"chunks": 4770.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.251715, "cps": 37.265625, "mean_p": 0.13586, "boundary_entropy": 0.043302, "bpb": 0.90272}, "el": {"chunks": 4147.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.202066, "cps": 32.398438, "mean_p": 0.110932, "boundary_entropy": 0.039802, "bpb": 0.894529}, "hi": {"chunks": 5059.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.300684, "cps": 39.523438, "mean_p": 0.126873, "boundary_entropy": 0.079111, "bpb": 0.744948}, "ta": {"chunks": 3701.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.197967, "cps": 28.914062, "mean_p": 0.089213, "boundary_entropy": 0.093551, "bpb": 0.656977}, "th": {"chunks": 4118.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.256015, "cps": 32.171875, "mean_p": 0.114427, "boundary_entropy": 0.14008, "bpb": 0.744369}, "zh": {"chunks": 2764.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.467999, "cps": 21.59375, "mean_p": 0.318982, "boundary_entropy": 0.412562, "bpb": 2.208045}, "my": {"chunks": 3085.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.162814, "cps": 24.101562, "mean_p": 0.091679, "boundary_entropy": 0.170891, "bpb": 0.705959}}, "bpb": {"en": 1.40707, "de": 1.51383, "vi": 1.16572, "tr": 1.42942, "fi": 1.53818, "ru": 0.90272, "el": 0.89453, "hi": 0.74495, "ta": 0.65698, "th": 0.74437, "zh": 2.20805, "my": 0.70596}, "gini_cpc": 0.171002, "gini_cps": 0.10372, "bpic": 8.574387, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9181518151443329, "vi": 1.0635916950164734, "tr": 0.8501314069829901, "fi": 0.7911130932921095, "ru": 1.2668007038094526, "el": 1.0169329551037365, "hi": 1.5132432235005162, "ta": 0.9963060025235388, "th": 1.2884406204531198, "zh": 2.355286416045323, "my": 0.8193904983387001}, "t": 12898.4}
{"event": "eval", "step": 76300, "phase": 3, "bpb": {"en": 1.40497, "de": 1.52206, "vi": 1.18464, "tr": 1.4221, "fi": 1.56113, "ru": 0.89983, "el": 0.89481, "hi": 0.75108, "ta": 0.663, "th": 0.74423, "zh": 2.21166, "my": 0.71144}, "bytes_per_chunk": {"en": 4.841, "de": 5.454, "vi": 6.038, "tr": 6.154, "fi": 5.976, "ru": 7.287, "el": 8.479, "hi": 8.219, "ta": 13.024, "th": 11.09, "zh": 5.86, "my": 16.192}, "bpb_mean": 1.16425, "t": 12903.5}
{"event": "probe", "step": 76300, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3458.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.198701, "cps": 27.015625, "mean_p": 0.194419, "boundary_entropy": 0.036894, "bpb": 1.407074}, "de": {"chunks": 3746.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.182438, "cps": 29.265625, "mean_p": 0.178313, "boundary_entropy": 0.029633, "bpb": 1.513831}, "vi": {"chunks": 3896.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.211337, "cps": 30.4375, "mean_p": 0.159007, "boundary_entropy": 0.032989, "bpb": 1.165716}, "tr": {"chunks": 3058.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.168922, "cps": 23.890625, "mean_p": 0.157186, "boundary_entropy": 0.045489, "bpb": 1.429422}, "fi": {"chunks": 2957.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.157195, "cps": 23.101562, "mean_p": 0.152876, "boundary_entropy": 0.03107, "bpb": 1.538176}, "ru": {"chunks": 4770.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.251715, "cps": 37.265625, "mean_p": 0.13586, "boundary_entropy": 0.043302, "bpb": 0.90272}, "el": {"chunks": 4147.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.202066, "cps": 32.398438, "mean_p": 0.110932, "boundary_entropy": 0.039802, "bpb": 0.894529}, "hi": {"chunks": 5059.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.300684, "cps": 39.523438, "mean_p": 0.126873, "boundary_entropy": 0.079111, "bpb": 0.744948}, "ta": {"chunks": 3701.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.197967, "cps": 28.914062, "mean_p": 0.089213, "boundary_entropy": 0.093551, "bpb": 0.656977}, "th": {"chunks": 4118.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.256015, "cps": 32.171875, "mean_p": 0.114427, "boundary_entropy": 0.14008, "bpb": 0.744369}, "zh": {"chunks": 2764.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.467999, "cps": 21.59375, "mean_p": 0.318982, "boundary_entropy": 0.412562, "bpb": 2.208045}, "my": {"chunks": 3085.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.162814, "cps": 24.101562, "mean_p": 0.091679, "boundary_entropy": 0.170891, "bpb": 0.705959}}, "bpb": {"en": 1.40707, "de": 1.51383, "vi": 1.16572, "tr": 1.42942, "fi": 1.53818, "ru": 0.90272, "el": 0.89453, "hi": 0.74495, "ta": 0.65698, "th": 0.74437, "zh": 2.20805, "my": 0.70596}, "gini_cpc": 0.171002, "gini_cps": 0.10372, "bpic": 8.574387, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9181518151443329, "vi": 1.0635916950164734, "tr": 0.8501314069829901, "fi": 0.7911130932921095, "ru": 1.2668007038094526, "el": 1.0169329551037365, "hi": 1.5132432235005162, "ta": 0.9963060025235388, "th": 1.2884406204531198, "zh": 2.355286416045323, "my": 0.8193904983387001}, "t": 12906.4}
{"event": "reps_dumped", "dirs": ["/local/scratch/sas245/beetle-hnet/scratch/reps/specfull_B_s0/th0.30", "/local/scratch/sas245/beetle-hnet/scratch/reps/specfull_B_s0/th0.40", "/local/scratch/sas245/beetle-hnet/scratch/reps/specfull_B_s0/th0.50", "/local/scratch/sas245/beetle-hnet/scratch/reps/specfull_B_s0/th0.60", "/local/scratch/sas245/beetle-hnet/scratch/reps/specfull_B_s0/th0.70"], "t": 12909.2}
{"event": "final", "run": "specfull_B_s0", "condition": "B", "step": 76300, "params": 82470528, "per_lang": {"en": {"cpc": 0.19870137332643797, "cps": 27.015625, "bpb": 1.402198818161624, "bytes_per_chunk": 4.837363793214803}, "de": {"cpc": 0.18243802659134076, "cps": 29.265625, "bpb": 1.518735342652289, "bytes_per_chunk": 5.426372796100126}, "vi": {"cpc": 0.21133713045836724, "cps": 30.4375, "bpb": 1.182986291718871, "bytes_per_chunk": 6.038966798168572}, "tr": {"cpc": 0.1689222780754571, "cps": 23.890625, "bpb": 1.4195071037743365, "bytes_per_chunk": 6.158385603871544}, "fi": {"cpc": 0.1571952580936686, "cps": 23.1015625, "bpb": 1.5667269938863988, "bytes_per_chunk": 6.001121730670177}, "ru": {"cpc": 0.2517150395778364, "cps": 37.265625, "bpb": 0.9020046161281352, "bytes_per_chunk": 7.265682273297348}, "el": {"cpc": 0.20206597476002533, "cps": 32.3984375, "bpb": 0.8931438657262853, "bytes_per_chunk": 8.445088753583969}, "hi": {"cpc": 0.3006835066864785, "cps": 39.5234375, "bpb": 0.7522632883717733, "bytes_per_chunk": 8.223158059836098}, "ta": {"cpc": 0.19796737095480074, "cps": 28.9140625, "bpb": 0.6607539855787611, "bytes_per_chunk": 13.005109887384036}, "th": {"cpc": 0.2560149207336027, "cps": 32.171875, "bpb": 0.7470995872092561, "bytes_per_chunk": 11.060228255595637}, "zh": {"cpc": 0.46799864544530984, "cps": 21.59375, "bpb": 2.2121380470930134, "bytes_per_chunk": 5.857770130610147}, "my": {"cpc": 0.1628140173105341, "cps": 24.1015625, "bpb": 0.7119607718538837, "bytes_per_chunk": 16.17973089742007}}, "gini_cpc": 0.17100165105827392, "gini_cps": 0.10372029461486343, "bpic": 8.574387274067785, "seed": 0, "mean_boundary_entropy": 0.08595607429742813, "dechunk": "ema", "router_noise": 0.0, "transitions": [{"step": 0, "phase": 1, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0}, {"step": 30520, "phase": 2, "objective": "parity_a", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0}, {"step": 53410, "phase": 3, "objective": "none", "router_trainable": false, "encoder_trainable": true, "frozen_router_params": 295680}], "t": 12918.7}