{"event": "start", "run": "specfull_A_s1", "condition": "A", "params": 82470528, "langs": ["en", "de", "vi", "tr", "fi", "ru", "el", "hi", "ta", "th", "zh", "my"], "premium": {"en": 1.0, "de": 1.184, "vi": 1.39, "tr": 1.12, "fi": 1.104, "ru": 1.971, "el": 2.174, "hi": 2.552, "ta": 3.192, "th": 2.762, "zh": 0.915, "my": 3.524}, "schedule": {"tau1": 0.4, "tau2": 0.7, "p1_objective": "global", "p2_objective": "global", "p3_objective": "global", "freeze_in_p3": false, "freeze_scope": "router"}, "corpus": "/local/scratch/sas245/beetle-hnet/scratch/data/tier1_corpus", "premium_kind": "flores_per_sent", "doc_filtered": {"my": 37511}, "weights": {"en": 0.2946, "de": 0.0909, "vi": 0.0576, "tr": 0.0612, "fi": 0.0424, "ru": 0.1476, "el": 0.0522, "hi": 0.042, "ta": 0.0313, "th": 0.0542, "zh": 0.1038, "my": 0.0222}, "t": 3.0}
{"event": "probe_schedule", "n": 44, "steps": [0, 1, 2, 4, 8, 16, 32, 64, 128, 256, 500, 1000, 1500, 2000, 2500, 3000, 3500, 4000, 4500, 5000, 5500, 6000, 6500, 7000, 7500, 8000, 12000, 16000, 20000, 24000, 28000, 32000, 36000, 40000, 44000, 48000, 52000, 56000, 60000, 64000, 68000, 72000, 76000, 76300], "t": 3.0}
{"event": "probe", "step": 0, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 14124.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.811584, "cps": 110.34375, "mean_p": 0.52085, "boundary_entropy": 0.686717, "bpb": 8.06011}, "de": {"chunks": 18700.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.910729, "cps": 146.09375, "mean_p": 0.525595, "boundary_entropy": 0.687217, "bpb": 8.1963}, "vi": {"chunks": 21351.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.158177, "cps": 166.804688, "mean_p": 0.524018, "boundary_entropy": 0.687874, "bpb": 8.061647}, "tr": {"chunks": 17846.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.985803, "cps": 139.421875, "mean_p": 0.529939, "boundary_entropy": 0.686373, "bpb": 8.199}, "fi": {"chunks": 17574.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.934241, "cps": 137.296875, "mean_p": 0.526553, "boundary_entropy": 0.686826, "bpb": 8.038283}, "ru": {"chunks": 34128.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.80095, "cps": 266.625, "mean_p": 0.540146, "boundary_entropy": 0.687211, "bpb": 8.262097}, "el": {"chunks": 36913.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.798616, "cps": 288.382812, "mean_p": 0.535776, "boundary_entropy": 0.688142, "bpb": 8.178552}, "hi": {"chunks": 34684.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.061456, "cps": 270.96875, "mean_p": 0.509835, "boundary_entropy": 0.69075, "bpb": 8.397747}, "ta": {"chunks": 50678.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.710778, "cps": 395.921875, "mean_p": 0.522012, "boundary_entropy": 0.690366, "bpb": 8.106024}, "th": {"chunks": 45409.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.823065, "cps": 354.757812, "mean_p": 0.531604, "boundary_entropy": 0.689071, "bpb": 8.030889}, "zh": {"chunks": 4820.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.816119, "cps": 37.65625, "mean_p": 0.492076, "boundary_entropy": 0.68644, "bpb": 8.099598}, "my": {"chunks": 28215.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 1.489075, "cps": 220.429688, "mean_p": 0.506229, "boundary_entropy": 0.690851, "bpb": 8.204861}}, "bpb": {"en": 8.06011, "de": 8.1963, "vi": 8.06165, "tr": 8.199, "fi": 8.03828, "ru": 8.2621, "el": 8.17855, "hi": 8.39775, "ta": 8.10602, "th": 8.03089, "zh": 8.0996, "my": 8.20486}, "gini_cpc": 0.247425, "gini_cps": 0.271932, "bpic": 1.182896, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1221621360827407, "vi": 1.4270575576222595, "tr": 1.2146656456674263, "fi": 1.1511320610100555, "ru": 2.2190548395702145, "el": 2.2161793753498933, "hi": 2.540039766651812, "ta": 3.3401072258858213, "th": 3.478462165603346, "zh": 1.0055878258102513, "my": 1.8347761655608366}, "t": 10.5}
{"event": "train", "step": 0, "phase": 1, "objective": "global", "lm_loss": 5.63341, "bpb": 8.1273, "chunk_loss": 0.084693, "F": 0.86952, "G": 0.52249, "chunks": 890.39, "bytes_per_chunk": 1.15, "boundary_entropy": 0.69066, "router_noise": 0.0, "grad_norm": 9.4224, "lr": 1e-07, "router_trainable": true, "sec_per_step": 0.001, "alpha": 0.03, "t": 11.5}
{"event": "probe", "step": 1, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 14109.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.810722, "cps": 110.226562, "mean_p": 0.520777, "boundary_entropy": 0.686722, "bpb": 8.051313}, "de": {"chunks": 18692.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.910339, "cps": 146.03125, "mean_p": 0.525525, "boundary_entropy": 0.687223, "bpb": 8.188272}, "vi": {"chunks": 21343.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.157743, "cps": 166.742188, "mean_p": 0.52399, "boundary_entropy": 0.687876, "bpb": 8.057933}, "tr": {"chunks": 17839.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.985417, "cps": 139.367188, "mean_p": 0.529884, "boundary_entropy": 0.686379, "bpb": 8.192912}, "fi": {"chunks": 17564.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.933709, "cps": 137.21875, "mean_p": 0.526475, "boundary_entropy": 0.686834, "bpb": 8.030689}, "ru": {"chunks": 34128.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.80095, "cps": 266.625, "mean_p": 0.540125, "boundary_entropy": 0.687214, "bpb": 8.253465}, "el": {"chunks": 36912.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.798567, "cps": 288.375, "mean_p": 0.535748, "boundary_entropy": 0.688146, "bpb": 8.17552}, "hi": {"chunks": 34649.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.059376, "cps": 270.695312, "mean_p": 0.509807, "boundary_entropy": 0.690751, "bpb": 8.385787}, "ta": {"chunks": 50673.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.710511, "cps": 395.882812, "mean_p": 0.521967, "boundary_entropy": 0.690369, "bpb": 8.097379}, "th": {"chunks": 45407.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.822941, "cps": 354.742188, "mean_p": 0.531573, "boundary_entropy": 0.689075, "bpb": 8.019145}, "zh": {"chunks": 4817.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.815611, "cps": 37.632812, "mean_p": 0.492067, "boundary_entropy": 0.686439, "bpb": 8.098302}, "my": {"chunks": 28228.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 1.489761, "cps": 220.53125, "mean_p": 0.506233, "boundary_entropy": 0.690851, "bpb": 8.201395}}, "bpb": {"en": 8.05131, "de": 8.18827, "vi": 8.05793, "tr": 8.19291, "fi": 8.03069, "ru": 8.25346, "el": 8.17552, "hi": 8.38579, "ta": 8.09738, "th": 8.01914, "zh": 8.0983, "my": 8.2014}, "gini_cpc": 0.247506, "gini_cps": 0.272016, "bpic": 1.183191, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1228745843326384, "vi": 1.4280394632952176, "tr": 1.2154800661121117, "fi": 1.1517001720834326, "ru": 2.2214140303416055, "el": 2.2184754073286346, "hi": 2.540174306242514, "ta": 3.343328372340281, "th": 3.48200692803111, "zh": 1.0060303677451812, "my": 1.8375730778969546}, "t": 18.7}
{"event": "probe", "step": 2, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 14076.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.808826, "cps": 109.96875, "mean_p": 0.520627, "boundary_entropy": 0.686733, "bpb": 8.032967}, "de": {"chunks": 18675.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.909512, "cps": 145.898438, "mean_p": 0.525379, "boundary_entropy": 0.687237, "bpb": 8.171124}, "vi": {"chunks": 21333.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.157201, "cps": 166.664062, "mean_p": 0.523925, "boundary_entropy": 0.687882, "bpb": 8.049591}, "tr": {"chunks": 17820.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.984367, "cps": 139.21875, "mean_p": 0.529775, "boundary_entropy": 0.686391, "bpb": 8.180061}, "fi": {"chunks": 17541.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.932486, "cps": 137.039062, "mean_p": 0.526329, "boundary_entropy": 0.686848, "bpb": 8.015438}, "ru": {"chunks": 34126.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.800844, "cps": 266.609375, "mean_p": 0.540082, "boundary_entropy": 0.687221, "bpb": 8.236966}, "el": {"chunks": 36912.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.798567, "cps": 288.375, "mean_p": 0.535696, "boundary_entropy": 0.688153, "bpb": 8.1679}, "hi": {"chunks": 34555.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.053789, "cps": 269.960938, "mean_p": 0.509737, "boundary_entropy": 0.690753, "bpb": 8.364926}, "ta": {"chunks": 50664.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.710029, "cps": 395.8125, "mean_p": 0.521884, "boundary_entropy": 0.690376, "bpb": 8.081803}, "th": {"chunks": 45403.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.822692, "cps": 354.710938, "mean_p": 0.531512, "boundary_entropy": 0.689082, "bpb": 8.000327}, "zh": {"chunks": 4807.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.813918, "cps": 37.554688, "mean_p": 0.492041, "boundary_entropy": 0.686438, "bpb": 8.096113}, "my": {"chunks": 28192.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 1.487862, "cps": 220.25, "mean_p": 0.506216, "boundary_entropy": 0.690851, "bpb": 8.193685}}, "bpb": {"en": 8.03297, "de": 8.17112, "vi": 8.04959, "tr": 8.18006, "fi": 8.01544, "ru": 8.23697, "el": 8.1679, "hi": 8.36493, "ta": 8.0818, "th": 8.00033, "zh": 8.09611, "my": 8.19368}, "gini_cpc": 0.247731, "gini_cps": 0.272252, "bpic": 1.184129, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1244834433448345, "vi": 1.4307167229576234, "tr": 1.2170320359217566, "fi": 1.1528885531279585, "ru": 2.2264914624792214, "el": 2.2236764366297033, "hi": 2.539222083303248, "ta": 3.3505713263595376, "th": 3.4898627443030295, "zh": 1.0062955253215409, "my": 1.8395321084257878}, "t": 26.9}
{"event": "probe", "step": 4, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 13949.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.801528, "cps": 108.976562, "mean_p": 0.520083, "boundary_entropy": 0.686773, "bpb": 7.967749}, "de": {"chunks": 18593.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.905518, "cps": 145.257812, "mean_p": 0.524856, "boundary_entropy": 0.687287, "bpb": 8.110548}, "vi": {"chunks": 21282.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.154434, "cps": 166.265625, "mean_p": 0.52371, "boundary_entropy": 0.687901, "bpb": 8.020774}, "tr": {"chunks": 17774.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.981826, "cps": 138.859375, "mean_p": 0.529387, "boundary_entropy": 0.686433, "bpb": 8.134891}, "fi": {"chunks": 17458.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.928074, "cps": 136.390625, "mean_p": 0.52583, "boundary_entropy": 0.686898, "bpb": 7.962292}, "ru": {"chunks": 34116.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.800317, "cps": 266.53125, "mean_p": 0.539896, "boundary_entropy": 0.687249, "bpb": 8.175597}, "el": {"chunks": 36887.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.797349, "cps": 288.179688, "mean_p": 0.535537, "boundary_entropy": 0.688175, "bpb": 8.144745}, "hi": {"chunks": 34337.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.040832, "cps": 268.257812, "mean_p": 0.509586, "boundary_entropy": 0.690758, "bpb": 8.310258}, "ta": {"chunks": 50646.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.709067, "cps": 395.671875, "mean_p": 0.52164, "boundary_entropy": 0.690397, "bpb": 8.036736}, "th": {"chunks": 45393.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.82207, "cps": 354.632812, "mean_p": 0.531363, "boundary_entropy": 0.6891, "bpb": 7.942328}, "zh": {"chunks": 4784.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.810024, "cps": 37.375, "mean_p": 0.491975, "boundary_entropy": 0.686435, "bpb": 8.088458}, "my": {"chunks": 28180.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 1.487228, "cps": 220.15625, "mean_p": 0.506211, "boundary_entropy": 0.690851, "bpb": 8.17323}}, "bpb": {"en": 7.96775, "de": 8.11055, "vi": 8.02077, "tr": 8.13489, "fi": 7.96229, "ru": 8.1756, "el": 8.14475, "hi": 8.31026, "ta": 8.03674, "th": 7.94233, "zh": 8.08846, "my": 8.17323}, "gini_cpc": 0.248585, "gini_cps": 0.273079, "bpic": 1.186711, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1297389652854446, "vi": 1.4402913181174173, "tr": 1.22494240793932, "fi": 1.157880264466836, "ru": 2.2461043934831575, "el": 2.242402332493013, "hi": 2.546175421246385, "ta": 3.3798756870609012, "th": 3.5208608926701386, "zh": 1.010598790810615, "my": 1.8554901731726872}, "t": 35.5}
{"event": "probe", "step": 8, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 13441.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.772338, "cps": 105.007812, "mean_p": 0.518099, "boundary_entropy": 0.686908, "bpb": 7.741733}, "de": {"chunks": 18218.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.887255, "cps": 142.328125, "mean_p": 0.522941, "boundary_entropy": 0.687458, "bpb": 7.895815}, "vi": {"chunks": 21068.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.142826, "cps": 164.59375, "mean_p": 0.522823, "boundary_entropy": 0.687978, "bpb": 7.918211}, "tr": {"chunks": 17588.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.971552, "cps": 137.40625, "mean_p": 0.528003, "boundary_entropy": 0.686578, "bpb": 7.973507}, "fi": {"chunks": 17206.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.914678, "cps": 134.421875, "mean_p": 0.524062, "boundary_entropy": 0.687064, "bpb": 7.780661}, "ru": {"chunks": 34093.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.799103, "cps": 266.351562, "mean_p": 0.539225, "boundary_entropy": 0.687352, "bpb": 7.965733}, "el": {"chunks": 36844.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.795254, "cps": 287.84375, "mean_p": 0.534822, "boundary_entropy": 0.68827, "bpb": 8.029911}, "hi": {"chunks": 33453.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 1.988291, "cps": 261.351562, "mean_p": 0.509017, "boundary_entropy": 0.690776, "bpb": 8.102834}, "ta": {"chunks": 50543.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.703557, "cps": 394.867188, "mean_p": 0.520594, "boundary_entropy": 0.690481, "bpb": 7.839632}, "th": {"chunks": 45312.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.817035, "cps": 354.0, "mean_p": 0.530628, "boundary_entropy": 0.689188, "bpb": 7.702147}, "zh": {"chunks": 4693.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.794616, "cps": 36.664062, "mean_p": 0.491653, "boundary_entropy": 0.686419, "bpb": 8.054077}, "my": {"chunks": 28070.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 1.481423, "cps": 219.296875, "mean_p": 0.506146, "boundary_entropy": 0.690853, "bpb": 8.070233}}, "bpb": {"en": 7.74173, "de": 7.89582, "vi": 7.91821, "tr": 7.97351, "fi": 7.78066, "ru": 7.96573, "el": 8.02991, "hi": 8.10283, "ta": 7.83963, "th": 7.70215, "zh": 8.05408, "my": 8.07023}, "gini_cpc": 0.251879, "gini_cps": 0.276597, "bpic": 1.197336, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1487904846444152, "vi": 1.479696705409756, "tr": 1.257935706155614, "fi": 1.1842968502469982, "ru": 2.3294239870564977, "el": 2.3244406808965854, "hi": 2.574379312018305, "ta": 3.500483909361566, "th": 3.6474110167793263, "zh": 1.0288442877606874, "my": 1.9181014589645862}, "t": 44.9}
{"event": "probe", "step": 16, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 11470.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.659082, "cps": 89.609375, "mean_p": 0.511411, "boundary_entropy": 0.687247, "bpb": 7.121015}, "de": {"chunks": 16525.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.804802, "cps": 129.101562, "mean_p": 0.516227, "boundary_entropy": 0.687944, "bpb": 7.253558}, "vi": {"chunks": 20285.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.100353, "cps": 158.476562, "mean_p": 0.520056, "boundary_entropy": 0.688199, "bpb": 7.608754}, "tr": {"chunks": 16744.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.92493, "cps": 130.8125, "mean_p": 0.523181, "boundary_entropy": 0.687028, "bpb": 7.460813}, "fi": {"chunks": 15909.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.845729, "cps": 124.289062, "mean_p": 0.518118, "boundary_entropy": 0.68753, "bpb": 7.25805}, "ru": {"chunks": 33907.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.789288, "cps": 264.898438, "mean_p": 0.535321, "boundary_entropy": 0.68792, "bpb": 7.114992}, "el": {"chunks": 36670.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.786776, "cps": 286.484375, "mean_p": 0.532017, "boundary_entropy": 0.688625, "bpb": 7.618643}, "hi": {"chunks": 30534.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 1.814799, "cps": 238.546875, "mean_p": 0.507141, "boundary_entropy": 0.690827, "bpb": 7.350249}, "ta": {"chunks": 50062.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.677828, "cps": 391.109375, "mean_p": 0.516952, "boundary_entropy": 0.69074, "bpb": 7.212651}, "th": {"chunks": 44872.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.78968, "cps": 350.5625, "mean_p": 0.527802, "boundary_entropy": 0.689508, "bpb": 6.884514}, "zh": {"chunks": 4353.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.737047, "cps": 34.007812, "mean_p": 0.490401, "boundary_entropy": 0.686354, "bpb": 7.924558}, "my": {"chunks": 27481.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 1.450338, "cps": 214.695312, "mean_p": 0.505848, "boundary_entropy": 0.690857, "bpb": 7.684772}}, "bpb": {"en": 7.12101, "de": 7.25356, "vi": 7.60875, "tr": 7.46081, "fi": 7.25805, "ru": 7.11499, "el": 7.61864, "hi": 7.35025, "ta": 7.21265, "th": 6.88451, "zh": 7.92456, "my": 7.68477}, "gini_cpc": 0.266709, "gini_cps": 0.292578, "bpic": 1.242766, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.22109587258922, "vi": 1.6695236379191338, "tr": 1.4033608806644107, "fi": 1.2831921694150557, "ru": 2.7148188390961394, "el": 2.7110077999119877, "hi": 2.7535269447657886, "ta": 4.062968251188666, "th": 4.232676374068776, "zh": 1.1182938249961398, "my": 2.2005429948324595}, "t": 56.0}
{"event": "probe", "step": 32, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 7649.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.439522, "cps": 59.757812, "mean_p": 0.501471, "boundary_entropy": 0.687521, "bpb": 6.207206}, "de": {"chunks": 12032.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.585984, "cps": 94.0, "mean_p": 0.505482, "boundary_entropy": 0.688407, "bpb": 6.345197}, "vi": {"chunks": 17301.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.938487, "cps": 135.164062, "mean_p": 0.511657, "boundary_entropy": 0.688702, "bpb": 6.958241}, "tr": {"chunks": 14310.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.790477, "cps": 111.796875, "mean_p": 0.513463, "boundary_entropy": 0.68776, "bpb": 6.662073}, "fi": {"chunks": 11976.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.636649, "cps": 93.5625, "mean_p": 0.507175, "boundary_entropy": 0.688053, "bpb": 6.428406}, "ru": {"chunks": 29096.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.535409, "cps": 227.3125, "mean_p": 0.513772, "boundary_entropy": 0.689966, "bpb": 5.717527}, "el": {"chunks": 35385.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.724163, "cps": 276.445312, "mean_p": 0.521992, "boundary_entropy": 0.689664, "bpb": 6.490324}, "hi": {"chunks": 19365.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 1.150966, "cps": 151.289062, "mean_p": 0.499875, "boundary_entropy": 0.690892, "bpb": 5.758845}, "ta": {"chunks": 25984.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 1.38989, "cps": 203.0, "mean_p": 0.503216, "boundary_entropy": 0.691243, "bpb": 5.847371}, "th": {"chunks": 39714.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.469008, "cps": 310.265625, "mean_p": 0.514353, "boundary_entropy": 0.690615, "bpb": 5.469859}, "zh": {"chunks": 3378.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.571961, "cps": 26.390625, "mean_p": 0.486231, "boundary_entropy": 0.68609, "bpb": 7.513648}, "my": {"chunks": 22601.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 1.192791, "cps": 176.570312, "mean_p": 0.502573, "boundary_entropy": 0.690883, "bpb": 6.297003}}, "bpb": {"en": 6.20721, "de": 6.3452, "vi": 6.95824, "tr": 6.66207, "fi": 6.42841, "ru": 5.71753, "el": 6.49032, "hi": 5.75884, "ta": 5.84737, "th": 5.46986, "zh": 7.51365, "my": 6.297}, "gini_cpc": 0.27638, "gini_cps": 0.302167, "bpic": 1.607184, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.333229379512142, "vi": 2.135244065259093, "tr": 1.798492129481446, "fi": 1.448502880477628, "ru": 3.493361527245357, "el": 3.9228148806092875, "hi": 2.6186767220102967, "ta": 3.1622776407693096, "th": 5.617486346170594, "zh": 1.3013246664736688, "my": 2.7138368702304425}, "t": 70.7}
{"event": "probe", "step": 64, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3750.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.21548, "cps": 29.296875, "mean_p": 0.493738, "boundary_entropy": 0.687532, "bpb": 5.764682}, "de": {"chunks": 5625.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.273949, "cps": 43.945312, "mean_p": 0.495657, "boundary_entropy": 0.688444, "bpb": 5.904688}, "vi": {"chunks": 6733.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.365229, "cps": 52.601562, "mean_p": 0.493245, "boundary_entropy": 0.688796, "bpb": 6.447219}, "tr": {"chunks": 8614.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.475833, "cps": 67.296875, "mean_p": 0.500741, "boundary_entropy": 0.688189, "bpb": 6.284725}, "fi": {"chunks": 6647.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.353357, "cps": 51.929688, "mean_p": 0.497804, "boundary_entropy": 0.688192, "bpb": 5.963047}, "ru": {"chunks": 4639.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.244802, "cps": 36.242188, "mean_p": 0.490319, "boundary_entropy": 0.690106, "bpb": 5.087525}, "el": {"chunks": 1819.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.088632, "cps": 14.210938, "mean_p": 0.487867, "boundary_entropy": 0.690234, "bpb": 5.589452}, "hi": {"chunks": 452.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.026865, "cps": 3.53125, "mean_p": 0.476977, "boundary_entropy": 0.689728, "bpb": 5.08113}, "ta": {"chunks": 2244.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.120032, "cps": 17.53125, "mean_p": 0.485588, "boundary_entropy": 0.690772, "bpb": 5.173547}, "th": {"chunks": 4166.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.258999, "cps": 32.546875, "mean_p": 0.489965, "boundary_entropy": 0.690793, "bpb": 4.842553}, "zh": {"chunks": 3189.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.539959, "cps": 24.914062, "mean_p": 0.48669, "boundary_entropy": 0.686223, "bpb": 6.947894}, "my": {"chunks": 7699.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.406323, "cps": 60.148438, "mean_p": 0.481166, "boundary_entropy": 0.690222, "bpb": 4.671376}}, "bpb": {"en": 5.76468, "de": 5.90469, "vi": 6.44722, "tr": 6.28473, "fi": 5.96305, "ru": 5.08752, "el": 5.58945, "hi": 5.08113, "ta": 5.17355, "th": 4.84255, "zh": 6.94789, "my": 4.67138}, "gini_cpc": 0.30228, "gini_cps": 0.296155, "bpic": 6.905393, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.2713436906443287, "vi": 1.6949555953349609, "tr": 2.2082445561509143, "fi": 1.63985952899899, "ru": 1.136077635883905, "el": 0.41132462115675095, "hi": 0.12467409212481427, "ta": 0.5570449424979941, "th": 1.2019628722412186, "zh": 2.505843413477819, "my": 1.8856617690521427}, "t": 90.0}
{"event": "probe", "step": 128, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3799.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.218296, "cps": 29.679688, "mean_p": 0.492967, "boundary_entropy": 0.687467, "bpb": 5.318484}, "de": {"chunks": 4275.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.208201, "cps": 33.398438, "mean_p": 0.492028, "boundary_entropy": 0.688262, "bpb": 5.496135}, "vi": {"chunks": 5000.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.271223, "cps": 39.0625, "mean_p": 0.490099, "boundary_entropy": 0.688723, "bpb": 5.949194}, "tr": {"chunks": 5637.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.311385, "cps": 44.039062, "mean_p": 0.494255, "boundary_entropy": 0.687973, "bpb": 5.778985}, "fi": {"chunks": 4661.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.247781, "cps": 36.414062, "mean_p": 0.493051, "boundary_entropy": 0.688003, "bpb": 5.487712}, "ru": {"chunks": 9723.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.513087, "cps": 75.960938, "mean_p": 0.492194, "boundary_entropy": 0.690024, "bpb": 4.707399}, "el": {"chunks": 4457.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.217171, "cps": 34.820312, "mean_p": 0.494303, "boundary_entropy": 0.69057, "bpb": 5.048624}, "hi": {"chunks": 13580.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.807132, "cps": 106.09375, "mean_p": 0.497694, "boundary_entropy": 0.690893, "bpb": 4.156314}, "ta": {"chunks": 19553.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 1.045895, "cps": 152.757812, "mean_p": 0.499824, "boundary_entropy": 0.691241, "bpb": 4.278208}, "th": {"chunks": 15159.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.942431, "cps": 118.429688, "mean_p": 0.496846, "boundary_entropy": 0.690886, "bpb": 4.217593}, "zh": {"chunks": 2810.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.475787, "cps": 21.953125, "mean_p": 0.489069, "boundary_entropy": 0.686704, "bpb": 6.67145}, "my": {"chunks": 6761.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.356819, "cps": 52.820312, "mean_p": 0.482119, "boundary_entropy": 0.690365, "bpb": 4.200211}}, "bpb": {"en": 5.31848, "de": 5.49613, "vi": 5.94919, "tr": 5.77898, "fi": 5.48771, "ru": 4.7074, "el": 5.04862, "hi": 4.15631, "ta": 4.27821, "th": 4.21759, "zh": 6.67145, "my": 4.20021}, "gini_cpc": 0.32426, "gini_cps": 0.336089, "bpic": 4.022229, "worst_lang": "ta", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9537587571298596, "vi": 1.2424579205996353, "tr": 1.4264360628224644, "fi": 1.1350684338154398, "ru": 2.3504222955492384, "el": 0.9948477351478372, "hi": 3.6974262797709727, "ta": 4.791182981177533, "th": 4.317221332479311, "zh": 2.179554353646276, "my": 1.6345657193544982}, "t": 117.5}
{"event": "probe", "step": 256, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2792.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.160432, "cps": 21.8125, "mean_p": 0.408947, "boundary_entropy": 0.654483, "bpb": 4.244187}, "de": {"chunks": 3183.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.155019, "cps": 24.867188, "mean_p": 0.416217, "boundary_entropy": 0.660391, "bpb": 4.443944}, "vi": {"chunks": 4110.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.222945, "cps": 32.109375, "mean_p": 0.434241, "boundary_entropy": 0.668167, "bpb": 4.235656}, "tr": {"chunks": 3236.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.178755, "cps": 25.28125, "mean_p": 0.436997, "boundary_entropy": 0.671409, "bpb": 4.563864}, "fi": {"chunks": 1957.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.104035, "cps": 15.289062, "mean_p": 0.433503, "boundary_entropy": 0.671654, "bpb": 4.439389}, "ru": {"chunks": 6187.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.326491, "cps": 48.335938, "mean_p": 0.44825, "boundary_entropy": 0.677417, "bpb": 3.118898}, "el": {"chunks": 5105.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.248745, "cps": 39.882812, "mean_p": 0.466531, "boundary_entropy": 0.686397, "bpb": 3.36023}, "hi": {"chunks": 10613.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.630788, "cps": 82.914062, "mean_p": 0.478775, "boundary_entropy": 0.688701, "bpb": 2.589651}, "ta": {"chunks": 9291.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.496978, "cps": 72.585938, "mean_p": 0.482105, "boundary_entropy": 0.68918, "bpb": 2.582898}, "th": {"chunks": 10850.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.674541, "cps": 84.765625, "mean_p": 0.495444, "boundary_entropy": 0.690949, "bpb": 2.590639}, "zh": {"chunks": 3058.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.517779, "cps": 23.890625, "mean_p": 0.484934, "boundary_entropy": 0.686001, "bpb": 5.89637}, "my": {"chunks": 16272.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.858771, "cps": 127.125, "mean_p": 0.454259, "boundary_entropy": 0.683817, "bpb": 2.690252}}, "bpb": {"en": 4.24419, "de": 4.44394, "vi": 4.23566, "tr": 4.56386, "fi": 4.43939, "ru": 3.1189, "el": 3.36023, "hi": 2.58965, "ta": 2.5829, "th": 2.59064, "zh": 5.89637, "my": 2.69025}, "gini_cpc": 0.343608, "gini_cps": 0.351428, "bpic": 5.006666, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 0.966257633075668, "vi": 1.389656253715719, "tr": 1.1142090144156005, "fi": 0.6484666541965706, "ru": 2.035071198524242, "el": 1.5504708568637855, "hi": 3.9318034332863583, "ta": 3.097745229368133, "th": 4.204529260296081, "zh": 3.227399628564803, "my": 5.352864694386022}, "t": 156.1}
{"event": "probe", "step": 500, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3529.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.202781, "cps": 27.570312, "mean_p": 0.340322, "boundary_entropy": 0.561332, "bpb": 3.40492}, "de": {"chunks": 4145.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.20187, "cps": 32.382812, "mean_p": 0.361129, "boundary_entropy": 0.582959, "bpb": 3.641703}, "vi": {"chunks": 4977.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.269976, "cps": 38.882812, "mean_p": 0.343013, "boundary_entropy": 0.549146, "bpb": 2.998677}, "tr": {"chunks": 4014.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.221731, "cps": 31.359375, "mean_p": 0.375498, "boundary_entropy": 0.583664, "bpb": 3.602805}, "fi": {"chunks": 3228.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.171602, "cps": 25.21875, "mean_p": 0.371294, "boundary_entropy": 0.601581, "bpb": 3.732343}, "ru": {"chunks": 6182.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.326227, "cps": 48.296875, "mean_p": 0.341834, "boundary_entropy": 0.547911, "bpb": 2.576223}, "el": {"chunks": 7380.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.359597, "cps": 57.65625, "mean_p": 0.337748, "boundary_entropy": 0.583605, "bpb": 2.624657}, "hi": {"chunks": 9620.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.571768, "cps": 75.15625, "mean_p": 0.340068, "boundary_entropy": 0.596538, "bpb": 2.071715}, "ta": {"chunks": 6065.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.324418, "cps": 47.382812, "mean_p": 0.329805, "boundary_entropy": 0.59321, "bpb": 1.948643}, "th": {"chunks": 2925.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.181846, "cps": 22.851562, "mean_p": 0.442888, "boundary_entropy": 0.679159, "bpb": 2.050767}, "zh": {"chunks": 5272.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.892652, "cps": 41.1875, "mean_p": 0.421686, "boundary_entropy": 0.626035, "bpb": 4.73678}, "my": {"chunks": 13622.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.718915, "cps": 106.421875, "mean_p": 0.373574, "boundary_entropy": 0.638033, "bpb": 2.159023}}, "bpb": {"en": 3.40492, "de": 3.6417, "vi": 2.99868, "tr": 3.60281, "fi": 3.73234, "ru": 2.57622, "el": 2.62466, "hi": 2.07171, "ta": 1.94864, "th": 2.05077, "zh": 4.73678, "my": 2.15902}, "gini_cpc": 0.30957, "gini_cps": 0.257246, "bpic": 5.408489, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9955076221253882, "vi": 1.33136446336423, "tr": 1.0934508803227876, "fi": 0.846241081004752, "ru": 1.6087636627087585, "el": 1.7733235373586549, "hi": 2.8196321963169106, "ta": 1.599844591836062, "th": 0.8967621447867498, "zh": 4.402044421832282, "my": 3.545275273671099}, "t": 217.6}
{"event": "train", "step": 500, "phase": 1, "objective": "global", "lm_loss": 2.13812, "bpb": 3.08466, "chunk_loss": 0.03128, "F": 0.19618, "G": 0.36109, "chunks": 200.89, "bytes_per_chunk": 5.097, "boundary_entropy": 0.58496, "router_noise": 0.0, "grad_norm": 3.4527, "lr": 7.51e-05, "router_trainable": true, "sec_per_step": 0.414, "alpha": 0.03, "t": 218.3}
{"event": "probe", "step": 1000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3693.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.212205, "cps": 28.851562, "mean_p": 0.253132, "boundary_entropy": 0.362004, "bpb": 2.69818}, "de": {"chunks": 4440.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.216237, "cps": 34.6875, "mean_p": 0.254439, "boundary_entropy": 0.364792, "bpb": 2.998641}, "vi": {"chunks": 4918.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.266775, "cps": 38.421875, "mean_p": 0.269846, "boundary_entropy": 0.371128, "bpb": 2.360478}, "tr": {"chunks": 3886.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.214661, "cps": 30.359375, "mean_p": 0.26864, "boundary_entropy": 0.381636, "bpb": 2.979351}, "fi": {"chunks": 3741.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.198873, "cps": 29.226562, "mean_p": 0.246829, "boundary_entropy": 0.366209, "bpb": 3.21683}, "ru": {"chunks": 5610.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.296042, "cps": 43.828125, "mean_p": 0.247809, "boundary_entropy": 0.415963, "bpb": 1.94893}, "el": {"chunks": 6400.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.311845, "cps": 50.0, "mean_p": 0.245792, "boundary_entropy": 0.412131, "bpb": 2.072047}, "hi": {"chunks": 7881.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.46841, "cps": 61.570312, "mean_p": 0.285969, "boundary_entropy": 0.470297, "bpb": 1.556807}, "ta": {"chunks": 8372.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.44782, "cps": 65.40625, "mean_p": 0.262404, "boundary_entropy": 0.475156, "bpb": 1.298402}, "th": {"chunks": 6636.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.412558, "cps": 51.84375, "mean_p": 0.258067, "boundary_entropy": 0.475122, "bpb": 1.612803}, "zh": {"chunks": 3817.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.646292, "cps": 29.820312, "mean_p": 0.279277, "boundary_entropy": 0.239465, "bpb": 3.705066}, "my": {"chunks": 1105.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.058318, "cps": 8.632812, "mean_p": 0.194144, "boundary_entropy": 0.394489, "bpb": 1.680218}}, "bpb": {"en": 2.69818, "de": 2.99864, "vi": 2.36048, "tr": 2.97935, "fi": 3.21683, "ru": 1.94893, "el": 2.07205, "hi": 1.55681, "ta": 1.2984, "th": 1.6128, "zh": 3.70507, "my": 1.68022}, "gini_cpc": 0.263296, "gini_cps": 0.215512, "bpic": 6.343592, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.019002795642354, "vi": 1.2571589599393835, "tr": 1.0115726002031968, "fi": 0.9371748757716083, "ru": 1.395077901785236, "el": 1.4695485594355848, "hi": 2.2073493203679586, "ta": 2.110321204316849, "th": 1.944151589077825, "zh": 3.0456046708478874, "my": 0.27481707600449784}, "t": 335.4}
{"event": "train", "step": 1000, "phase": 1, "objective": "global", "lm_loss": 1.73812, "bpb": 2.50757, "chunk_loss": 0.030575, "F": 0.19185, "G": 0.26, "chunks": 196.45, "bytes_per_chunk": 5.212, "boundary_entropy": 0.38074, "router_noise": 0.0, "grad_norm": 1.2835, "lr": 0.0001501, "router_trainable": true, "sec_per_step": 0.235, "alpha": 0.03, "t": 335.9}
{"event": "probe", "step": 1500, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3426.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.196863, "cps": 26.765625, "mean_p": 0.216859, "boundary_entropy": 0.255758, "bpb": 2.423881}, "de": {"chunks": 3975.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.193591, "cps": 31.054688, "mean_p": 0.217359, "boundary_entropy": 0.271416, "bpb": 2.695443}, "vi": {"chunks": 4003.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.217141, "cps": 31.273438, "mean_p": 0.212832, "boundary_entropy": 0.266196, "bpb": 2.136414}, "tr": {"chunks": 3803.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.210076, "cps": 29.710938, "mean_p": 0.222126, "boundary_entropy": 0.290416, "bpb": 2.694134}, "fi": {"chunks": 3763.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.200043, "cps": 29.398438, "mean_p": 0.218454, "boundary_entropy": 0.268164, "bpb": 2.952634}, "ru": {"chunks": 6055.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.319525, "cps": 47.304688, "mean_p": 0.232099, "boundary_entropy": 0.348505, "bpb": 1.725672}, "el": {"chunks": 6766.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.329679, "cps": 52.859375, "mean_p": 0.281308, "boundary_entropy": 0.426987, "bpb": 1.802046}, "hi": {"chunks": 9001.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.534978, "cps": 70.320312, "mean_p": 0.250424, "boundary_entropy": 0.361793, "bpb": 1.410793}, "ta": {"chunks": 9771.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.522653, "cps": 76.335938, "mean_p": 0.250389, "boundary_entropy": 0.41052, "bpb": 1.141061}, "th": {"chunks": 6360.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.395399, "cps": 49.6875, "mean_p": 0.235123, "boundary_entropy": 0.427716, "bpb": 1.392724}, "zh": {"chunks": 3972.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.672536, "cps": 31.03125, "mean_p": 0.265157, "boundary_entropy": 0.160839, "bpb": 3.490438}, "my": {"chunks": 9599.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.506597, "cps": 74.992188, "mean_p": 0.201294, "boundary_entropy": 0.363495, "bpb": 1.346634}}, "bpb": {"en": 2.42388, "de": 2.69544, "vi": 2.13641, "tr": 2.69413, "fi": 2.95263, "ru": 1.72567, "el": 1.80205, "hi": 1.41079, "ta": 1.14106, "th": 1.39272, "zh": 3.49044, "my": 1.34663}, "gini_cpc": 0.244185, "gini_cps": 0.215009, "bpic": 5.444165, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9833802627575805, "vi": 1.1030093902132594, "tr": 1.0671182210609789, "fi": 1.0161529831122211, "ru": 1.6230866091521148, "el": 1.6746648691873873, "hi": 2.7175181311689562, "ta": 2.654913068995381, "th": 2.0085045132932695, "zh": 3.41627292260594, "my": 2.573353073967392}, "t": 442.2}
{"event": "train", "step": 1500, "phase": 1, "objective": "global", "lm_loss": 1.57402, "bpb": 2.27083, "chunk_loss": 0.030345, "F": 0.18599, "G": 0.22974, "chunks": 190.45, "bytes_per_chunk": 5.377, "boundary_entropy": 0.2976, "router_noise": 0.0, "grad_norm": 1.0405, "lr": 0.0002251, "router_trainable": true, "sec_per_step": 0.214, "alpha": 0.03, "t": 442.8}
{"event": "probe", "step": 2000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3149.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.180946, "cps": 24.601562, "mean_p": 0.205461, "boundary_entropy": 0.221419, "bpb": 2.271141}, "de": {"chunks": 4064.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.197925, "cps": 31.75, "mean_p": 0.219627, "boundary_entropy": 0.238828, "bpb": 2.543011}, "vi": {"chunks": 4197.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.227665, "cps": 32.789062, "mean_p": 0.214162, "boundary_entropy": 0.237303, "bpb": 2.056792}, "tr": {"chunks": 3682.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.203392, "cps": 28.765625, "mean_p": 0.2202, "boundary_entropy": 0.272247, "bpb": 2.519414}, "fi": {"chunks": 4137.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.219925, "cps": 32.320312, "mean_p": 0.230401, "boundary_entropy": 0.244245, "bpb": 2.730724}, "ru": {"chunks": 5909.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.311821, "cps": 46.164062, "mean_p": 0.231242, "boundary_entropy": 0.333408, "bpb": 1.573341}, "el": {"chunks": 5759.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.280612, "cps": 44.992188, "mean_p": 0.255157, "boundary_entropy": 0.384441, "bpb": 1.668992}, "hi": {"chunks": 7792.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.46312, "cps": 60.875, "mean_p": 0.221505, "boundary_entropy": 0.31912, "bpb": 1.342346}, "ta": {"chunks": 7684.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.411019, "cps": 60.03125, "mean_p": 0.235586, "boundary_entropy": 0.387432, "bpb": 1.109057}, "th": {"chunks": 7211.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.448306, "cps": 56.335938, "mean_p": 0.204997, "boundary_entropy": 0.344341, "bpb": 1.30154}, "zh": {"chunks": 4039.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.683881, "cps": 31.554688, "mean_p": 0.259814, "boundary_entropy": 0.123973, "bpb": 3.348147}, "my": {"chunks": 10549.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.556734, "cps": 82.414062, "mean_p": 0.216253, "boundary_entropy": 0.371171, "bpb": 1.248536}}, "bpb": {"en": 2.27114, "de": 2.54301, "vi": 2.05679, "tr": 2.51941, "fi": 2.73072, "ru": 1.57334, "el": 1.66899, "hi": 1.34235, "ta": 1.10906, "th": 1.30154, "zh": 3.34815, "my": 1.24854}, "gini_cpc": 0.244959, "gini_cps": 0.203551, "bpic": 5.629599, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0938373576328044, "vi": 1.258193064247862, "tr": 1.1240475731588275, "fi": 1.2154164137018468, "ru": 1.7232815376326698, "el": 1.5508067867748068, "hi": 2.559442224871973, "ta": 2.271503164867394, "th": 2.4775697501750433, "zh": 3.779478421167358, "my": 3.076800771719804}, "t": 547.2}
{"event": "train", "step": 2000, "phase": 1, "objective": "global", "lm_loss": 1.49562, "bpb": 2.15772, "chunk_loss": 0.030384, "F": 0.18729, "G": 0.21987, "chunks": 191.78, "bytes_per_chunk": 5.339, "boundary_entropy": 0.24462, "router_noise": 0.0, "grad_norm": 0.6771, "lr": 0.0003, "router_trainable": true, "sec_per_step": 0.21, "alpha": 0.03, "t": 547.6}
{"event": "probe", "step": 2500, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3223.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.185198, "cps": 25.179688, "mean_p": 0.209496, "boundary_entropy": 0.189744, "bpb": 2.154665}, "de": {"chunks": 3615.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.176058, "cps": 28.242188, "mean_p": 0.201085, "boundary_entropy": 0.20112, "bpb": 2.390081}, "vi": {"chunks": 4028.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.218497, "cps": 31.46875, "mean_p": 0.202541, "boundary_entropy": 0.215029, "bpb": 1.984296}, "tr": {"chunks": 3439.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.189969, "cps": 26.867188, "mean_p": 0.204446, "boundary_entropy": 0.249207, "bpb": 2.36644}, "fi": {"chunks": 3931.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.208973, "cps": 30.710938, "mean_p": 0.216216, "boundary_entropy": 0.210746, "bpb": 2.559376}, "ru": {"chunks": 5893.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.310976, "cps": 46.039062, "mean_p": 0.230815, "boundary_entropy": 0.31812, "bpb": 1.472106}, "el": {"chunks": 6274.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.305706, "cps": 49.015625, "mean_p": 0.259021, "boundary_entropy": 0.380445, "bpb": 1.543927}, "hi": {"chunks": 6773.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.402556, "cps": 52.914062, "mean_p": 0.200626, "boundary_entropy": 0.297619, "bpb": 1.276504}, "ta": {"chunks": 7726.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.413266, "cps": 60.359375, "mean_p": 0.228957, "boundary_entropy": 0.366228, "bpb": 1.04951}, "th": {"chunks": 7200.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.447622, "cps": 56.25, "mean_p": 0.200121, "boundary_entropy": 0.327275, "bpb": 1.220689}, "zh": {"chunks": 4046.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.685066, "cps": 31.609375, "mean_p": 0.25342, "boundary_entropy": 0.118917, "bpb": 3.20699}, "my": {"chunks": 7877.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.415717, "cps": 61.539062, "mean_p": 0.195209, "boundary_entropy": 0.34643, "bpb": 1.15088}}, "bpb": {"en": 2.15466, "de": 2.39008, "vi": 1.9843, "tr": 2.36644, "fi": 2.55938, "ru": 1.47211, "el": 1.54393, "hi": 1.2765, "ta": 1.04951, "th": 1.22069, "zh": 3.20699, "my": 1.15088}, "gini_cpc": 0.234031, "gini_cps": 0.179184, "bpic": 5.994237, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9506479349838138, "vi": 1.1798047344314044, "tr": 1.0257592431106937, "fi": 1.1283789482038893, "ru": 1.6791559839118082, "el": 1.6506974633199119, "hi": 2.1736510104057993, "ta": 2.231480258110729, "th": 2.4169921832616073, "zh": 3.6991015200495, "my": 2.2447153897228147}, "t": 653.7}
{"event": "train", "step": 2500, "phase": 1, "objective": "global", "lm_loss": 1.36772, "bpb": 1.9732, "chunk_loss": 0.030213, "F": 0.17915, "G": 0.21319, "chunks": 183.45, "bytes_per_chunk": 5.582, "boundary_entropy": 0.23779, "router_noise": 0.0, "grad_norm": 0.587, "lr": 0.0003, "router_trainable": true, "sec_per_step": 0.213, "alpha": 0.03, "t": 654.2}
{"event": "probe", "step": 3000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3099.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.178073, "cps": 24.210938, "mean_p": 0.200291, "boundary_entropy": 0.178201, "bpb": 2.077326}, "de": {"chunks": 3407.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.165928, "cps": 26.617188, "mean_p": 0.196738, "boundary_entropy": 0.198615, "bpb": 2.27739}, "vi": {"chunks": 3834.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.207974, "cps": 29.953125, "mean_p": 0.194143, "boundary_entropy": 0.210966, "bpb": 1.924371}, "tr": {"chunks": 3440.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.190024, "cps": 26.875, "mean_p": 0.202951, "boundary_entropy": 0.222816, "bpb": 2.261317}, "fi": {"chunks": 3807.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.202382, "cps": 29.742188, "mean_p": 0.210723, "boundary_entropy": 0.18749, "bpb": 2.459466}, "ru": {"chunks": 5741.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.302955, "cps": 44.851562, "mean_p": 0.219526, "boundary_entropy": 0.303221, "bpb": 1.382672}, "el": {"chunks": 6775.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.330117, "cps": 52.929688, "mean_p": 0.256232, "boundary_entropy": 0.366667, "bpb": 1.467083}, "hi": {"chunks": 6887.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.409331, "cps": 53.804688, "mean_p": 0.191877, "boundary_entropy": 0.284132, "bpb": 1.221897}, "ta": {"chunks": 8069.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.431613, "cps": 63.039062, "mean_p": 0.213509, "boundary_entropy": 0.337274, "bpb": 0.99366}, "th": {"chunks": 7964.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.49512, "cps": 62.21875, "mean_p": 0.204156, "boundary_entropy": 0.317945, "bpb": 1.152414}, "zh": {"chunks": 3865.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.654419, "cps": 30.195312, "mean_p": 0.243629, "boundary_entropy": 0.120805, "bpb": 3.092387}, "my": {"chunks": 3622.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.191155, "cps": 28.296875, "mean_p": 0.17645, "boundary_entropy": 0.3346, "bpb": 1.119146}}, "bpb": {"en": 2.07733, "de": 2.27739, "vi": 1.92437, "tr": 2.26132, "fi": 2.45947, "ru": 1.38267, "el": 1.46708, "hi": 1.2219, "ta": 0.99366, "th": 1.15241, "zh": 3.09239, "my": 1.11915}, "gini_cpc": 0.25465, "gini_cps": 0.194747, "bpic": 6.342439, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9317990650801824, "vi": 1.1679157375367943, "tr": 1.0671130599940675, "fi": 1.1365107221614426, "ru": 1.7012999001294995, "el": 1.853834663057654, "hi": 2.298674902079806, "ta": 2.423800049037033, "th": 2.780434893163467, "zh": 3.6750106295827267, "my": 1.0734643200394987}, "t": 759.3}
{"event": "train", "step": 3000, "phase": 1, "objective": "global", "lm_loss": 1.47008, "bpb": 2.12087, "chunk_loss": 0.030168, "F": 0.17513, "G": 0.20406, "chunks": 179.33, "bytes_per_chunk": 5.71, "boundary_entropy": 0.20702, "router_noise": 0.0, "grad_norm": 0.4881, "lr": 0.0002999, "router_trainable": true, "sec_per_step": 0.212, "alpha": 0.03, "t": 759.9}
{"event": "probe", "step": 3500, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3155.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.181291, "cps": 24.648438, "mean_p": 0.203944, "boundary_entropy": 0.168908, "bpb": 2.010037}, "de": {"chunks": 3776.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.183899, "cps": 29.5, "mean_p": 0.211051, "boundary_entropy": 0.19376, "bpb": 2.195668}, "vi": {"chunks": 4131.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.224085, "cps": 32.273438, "mean_p": 0.20582, "boundary_entropy": 0.207834, "bpb": 1.871388}, "tr": {"chunks": 3382.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.18682, "cps": 26.421875, "mean_p": 0.199813, "boundary_entropy": 0.226875, "bpb": 2.167244}, "fi": {"chunks": 3614.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.192122, "cps": 28.234375, "mean_p": 0.202952, "boundary_entropy": 0.179859, "bpb": 2.365287}, "ru": {"chunks": 5613.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.296201, "cps": 43.851562, "mean_p": 0.21617, "boundary_entropy": 0.291987, "bpb": 1.336449}, "el": {"chunks": 6084.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.296448, "cps": 47.53125, "mean_p": 0.247701, "boundary_entropy": 0.374839, "bpb": 1.407423}, "hi": {"chunks": 7654.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.454918, "cps": 59.796875, "mean_p": 0.205112, "boundary_entropy": 0.278859, "bpb": 1.145382}, "ta": {"chunks": 10180.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.544531, "cps": 79.53125, "mean_p": 0.242052, "boundary_entropy": 0.329092, "bpb": 0.961549}, "th": {"chunks": 9100.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.565744, "cps": 71.09375, "mean_p": 0.221563, "boundary_entropy": 0.312978, "bpb": 1.09709}, "zh": {"chunks": 3894.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.659329, "cps": 30.421875, "mean_p": 0.24332, "boundary_entropy": 0.113182, "bpb": 3.010382}, "my": {"chunks": 9713.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.512613, "cps": 75.882812, "mean_p": 0.210933, "boundary_entropy": 0.362778, "bpb": 1.086258}}, "bpb": {"en": 2.01004, "de": 2.19567, "vi": 1.87139, "tr": 2.16724, "fi": 2.36529, "ru": 1.33645, "el": 1.40742, "hi": 1.14538, "ta": 0.96155, "th": 1.09709, "zh": 3.01038, "my": 1.08626}, "gini_cpc": 0.261638, "gini_cps": 0.237209, "bpic": 5.4595, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0143885421812964, "vi": 1.2360521934909008, "tr": 1.0304995547139002, "fi": 1.0597441342654401, "ru": 1.6338439884758036, "el": 1.6352084279760957, "hi": 2.5093320962739867, "ta": 3.0036343692157326, "th": 3.120650151458648, "zh": 3.6368656763676896, "my": 2.8275791414454177}, "t": 863.2}
{"event": "train", "step": 3500, "phase": 1, "objective": "global", "lm_loss": 1.27023, "bpb": 1.83255, "chunk_loss": 0.030201, "F": 0.18021, "G": 0.21216, "chunks": 184.53, "bytes_per_chunk": 5.549, "boundary_entropy": 0.22514, "router_noise": 0.0, "grad_norm": 0.4448, "lr": 0.0002997, "router_trainable": true, "sec_per_step": 0.207, "alpha": 0.03, "t": 863.6}
{"event": "probe", "step": 4000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3037.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.17451, "cps": 23.726562, "mean_p": 0.190835, "boundary_entropy": 0.141542, "bpb": 1.914092}, "de": {"chunks": 3517.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.171285, "cps": 27.476562, "mean_p": 0.191591, "boundary_entropy": 0.16213, "bpb": 2.091603}, "vi": {"chunks": 4181.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.226797, "cps": 32.664062, "mean_p": 0.203958, "boundary_entropy": 0.202214, "bpb": 1.782613}, "tr": {"chunks": 3392.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.187372, "cps": 26.5, "mean_p": 0.201682, "boundary_entropy": 0.203834, "bpb": 2.015631}, "fi": {"chunks": 3472.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.184573, "cps": 27.125, "mean_p": 0.189867, "boundary_entropy": 0.157172, "bpb": 2.240165}, "ru": {"chunks": 5541.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.292401, "cps": 43.289062, "mean_p": 0.209153, "boundary_entropy": 0.272085, "bpb": 1.300559}, "el": {"chunks": 5923.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.288603, "cps": 46.273438, "mean_p": 0.241799, "boundary_entropy": 0.359259, "bpb": 1.360644}, "hi": {"chunks": 7144.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.424606, "cps": 55.8125, "mean_p": 0.196873, "boundary_entropy": 0.26878, "bpb": 1.110976}, "ta": {"chunks": 7794.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.416903, "cps": 60.890625, "mean_p": 0.210425, "boundary_entropy": 0.330258, "bpb": 0.937481}, "th": {"chunks": 7930.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.493006, "cps": 61.953125, "mean_p": 0.205701, "boundary_entropy": 0.310013, "bpb": 1.067815}, "zh": {"chunks": 3702.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.62682, "cps": 28.921875, "mean_p": 0.232155, "boundary_entropy": 0.111736, "bpb": 2.946}, "my": {"chunks": 8443.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.445588, "cps": 65.960938, "mean_p": 0.201733, "boundary_entropy": 0.352153, "bpb": 1.056432}}, "bpb": {"en": 1.91409, "de": 2.0916, "vi": 1.78261, "tr": 2.01563, "fi": 2.24016, "ru": 1.30056, "el": 1.36064, "hi": 1.11098, "ta": 0.93748, "th": 1.06781, "zh": 2.946, "my": 1.05643}, "gini_cpc": 0.242951, "gini_cps": 0.20382, "bpic": 5.989466, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9815203073199656, "vi": 1.2996199034795428, "tr": 1.073704122120793, "fi": 1.0576626336249813, "ru": 1.6755533642681337, "el": 1.653789444870762, "hi": 2.4331321722529613, "ta": 2.3889896060383204, "th": 2.825084551949641, "zh": 3.5918839783989336, "my": 2.553364061939895}, "t": 966.6}
{"event": "eval", "step": 4000, "phase": 1, "bpb": {"en": 1.91742, "de": 2.09069, "vi": 1.79266, "tr": 2.0106, "fi": 2.25214, "ru": 1.29694, "el": 1.35687, "hi": 1.11979, "ta": 0.93695, "th": 1.06603, "zh": 2.95356, "my": 1.06583}, "bytes_per_chunk": {"en": 5.827, "de": 5.913, "vi": 5.911, "tr": 5.911, "fi": 5.55, "ru": 6.136, "el": 6.427, "hi": 6.036, "ta": 6.68, "th": 5.999, "zh": 4.538, "my": 5.97}, "bpb_mean": 1.65496, "t": 972.1}
{"event": "train", "step": 4000, "phase": 1, "objective": "global", "lm_loss": 1.21474, "bpb": 1.7525, "chunk_loss": 0.030077, "F": 0.16972, "G": 0.19625, "chunks": 173.8, "bytes_per_chunk": 5.892, "boundary_entropy": 0.19825, "router_noise": 0.0, "grad_norm": 0.3764, "lr": 0.0002995, "router_trainable": true, "sec_per_step": 0.217, "alpha": 0.03, "t": 972.3}
{"event": "probe", "step": 4500, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3059.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.175774, "cps": 23.898438, "mean_p": 0.193759, "boundary_entropy": 0.139194, "bpb": 1.871226}, "de": {"chunks": 3659.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.178201, "cps": 28.585938, "mean_p": 0.197435, "boundary_entropy": 0.156907, "bpb": 2.03115}, "vi": {"chunks": 4059.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.220179, "cps": 31.710938, "mean_p": 0.199685, "boundary_entropy": 0.196493, "bpb": 1.722281}, "tr": {"chunks": 3353.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.185218, "cps": 26.195312, "mean_p": 0.203985, "boundary_entropy": 0.19819, "bpb": 1.954614}, "fi": {"chunks": 3540.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.188188, "cps": 27.65625, "mean_p": 0.192163, "boundary_entropy": 0.147655, "bpb": 2.176065}, "ru": {"chunks": 5668.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.299103, "cps": 44.28125, "mean_p": 0.224009, "boundary_entropy": 0.296917, "bpb": 1.267718}, "el": {"chunks": 6308.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.307362, "cps": 49.28125, "mean_p": 0.261489, "boundary_entropy": 0.370577, "bpb": 1.317257}, "hi": {"chunks": 7601.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.451768, "cps": 59.382812, "mean_p": 0.207465, "boundary_entropy": 0.274685, "bpb": 1.090618}, "ta": {"chunks": 8604.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.46023, "cps": 67.21875, "mean_p": 0.228552, "boundary_entropy": 0.337917, "bpb": 0.9128}, "th": {"chunks": 9132.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.567734, "cps": 71.34375, "mean_p": 0.216907, "boundary_entropy": 0.291425, "bpb": 1.032694}, "zh": {"chunks": 3689.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.624619, "cps": 28.820312, "mean_p": 0.22998, "boundary_entropy": 0.104545, "bpb": 2.885421}, "my": {"chunks": 9094.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.479945, "cps": 71.046875, "mean_p": 0.2124, "boundary_entropy": 0.370959, "bpb": 1.035558}}, "bpb": {"en": 1.87123, "de": 2.03115, "vi": 1.72228, "tr": 1.95461, "fi": 2.17606, "ru": 1.26772, "el": 1.31726, "hi": 1.09062, "ta": 0.9128, "th": 1.03269, "zh": 2.88542, "my": 1.03556}, "gini_cpc": 0.25311, "gini_cps": 0.225157, "bpic": 5.663327, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.013805505953492, "vi": 1.2526234927893547, "tr": 1.053725876503016, "fi": 1.0706216510065396, "ru": 1.7016305361315416, "el": 1.7486201664970342, "hi": 2.570160843675181, "ta": 2.618301023090419, "th": 3.2299030068989287, "zh": 3.5535289326837467, "my": 2.730462504251929}, "t": 1072.6}
{"event": "train", "step": 4500, "phase": 1, "objective": "global", "lm_loss": 1.25943, "bpb": 1.81698, "chunk_loss": 0.030122, "F": 0.17514, "G": 0.20464, "chunks": 179.34, "bytes_per_chunk": 5.71, "boundary_entropy": 0.19956, "router_noise": 0.0, "grad_norm": 0.4063, "lr": 0.0002992, "router_trainable": true, "sec_per_step": 0.202, "alpha": 0.03, "t": 1073.1}
{"event": "probe", "step": 5000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2943.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.169109, "cps": 22.992188, "mean_p": 0.182572, "boundary_entropy": 0.127597, "bpb": 1.849368}, "de": {"chunks": 3546.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.172698, "cps": 27.703125, "mean_p": 0.187072, "boundary_entropy": 0.138588, "bpb": 1.997881}, "vi": {"chunks": 4149.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.225061, "cps": 32.414062, "mean_p": 0.204902, "boundary_entropy": 0.194772, "bpb": 1.675152}, "tr": {"chunks": 3191.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.176269, "cps": 24.929688, "mean_p": 0.194107, "boundary_entropy": 0.199181, "bpb": 1.918832}, "fi": {"chunks": 3405.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.181011, "cps": 26.601562, "mean_p": 0.18356, "boundary_entropy": 0.137841, "bpb": 2.123954}, "ru": {"chunks": 5400.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.28496, "cps": 42.1875, "mean_p": 0.211635, "boundary_entropy": 0.276678, "bpb": 1.237649}, "el": {"chunks": 5538.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.269844, "cps": 43.265625, "mean_p": 0.227321, "boundary_entropy": 0.339661, "bpb": 1.295514}, "hi": {"chunks": 7450.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.442793, "cps": 58.203125, "mean_p": 0.194728, "boundary_entropy": 0.256548, "bpb": 1.050996}, "ta": {"chunks": 9011.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.482001, "cps": 70.398438, "mean_p": 0.214835, "boundary_entropy": 0.316208, "bpb": 0.892009}, "th": {"chunks": 7328.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.45558, "cps": 57.25, "mean_p": 0.192489, "boundary_entropy": 0.290824, "bpb": 1.020912}, "zh": {"chunks": 3722.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.630207, "cps": 29.078125, "mean_p": 0.231008, "boundary_entropy": 0.102155, "bpb": 2.761422}, "my": {"chunks": 11090.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.585286, "cps": 86.640625, "mean_p": 0.221067, "boundary_entropy": 0.366318, "bpb": 1.016271}}, "bpb": {"en": 1.84937, "de": 1.99788, "vi": 1.67515, "tr": 1.91883, "fi": 2.12395, "ru": 1.23765, "el": 1.29551, "hi": 1.051, "ta": 0.89201, "th": 1.02091, "zh": 2.76142, "my": 1.01627}, "gini_cpc": 0.265178, "gini_cps": 0.244511, "bpic": 5.747548, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0212220471230693, "vi": 1.3308654508899356, "tr": 1.0423416960024254, "fi": 1.0703827238828476, "ru": 1.6850717726513522, "el": 1.5956805976318658, "hi": 2.6183943666665153, "ta": 2.8502396564410892, "th": 2.694004107257666, "zh": 3.726634363091768, "my": 3.4610034179062312}, "t": 1173.5}
{"event": "train", "step": 5000, "phase": 1, "objective": "global", "lm_loss": 1.14376, "bpb": 1.6501, "chunk_loss": 0.030038, "F": 0.1685, "G": 0.19642, "chunks": 172.55, "bytes_per_chunk": 5.935, "boundary_entropy": 0.20012, "router_noise": 0.0, "grad_norm": 0.379, "lr": 0.0002989, "router_trainable": true, "sec_per_step": 0.202, "alpha": 0.03, "t": 1174.1}
{"event": "probe", "step": 5500, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3091.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.177613, "cps": 24.148438, "mean_p": 0.194389, "boundary_entropy": 0.132575, "bpb": 1.819173}, "de": {"chunks": 3639.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.177227, "cps": 28.429688, "mean_p": 0.19328, "boundary_entropy": 0.138719, "bpb": 1.960934}, "vi": {"chunks": 4202.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.227936, "cps": 32.828125, "mean_p": 0.209633, "boundary_entropy": 0.198055, "bpb": 1.652144}, "tr": {"chunks": 3355.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.185328, "cps": 26.210938, "mean_p": 0.210977, "boundary_entropy": 0.214406, "bpb": 1.874003}, "fi": {"chunks": 3622.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.192547, "cps": 28.296875, "mean_p": 0.198232, "boundary_entropy": 0.134808, "bpb": 2.08999}, "ru": {"chunks": 5595.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.295251, "cps": 43.710938, "mean_p": 0.220883, "boundary_entropy": 0.284173, "bpb": 1.20297}, "el": {"chunks": 6309.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.307411, "cps": 49.289062, "mean_p": 0.265761, "boundary_entropy": 0.372536, "bpb": 1.249472}, "hi": {"chunks": 7753.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.460802, "cps": 60.570312, "mean_p": 0.196862, "boundary_entropy": 0.245357, "bpb": 1.032108}, "ta": {"chunks": 8864.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.474137, "cps": 69.25, "mean_p": 0.223374, "boundary_entropy": 0.319403, "bpb": 0.868296}, "th": {"chunks": 8669.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.538949, "cps": 67.726562, "mean_p": 0.20767, "boundary_entropy": 0.281041, "bpb": 0.985888}, "zh": {"chunks": 3707.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.627667, "cps": 28.960938, "mean_p": 0.231515, "boundary_entropy": 0.091444, "bpb": 2.687535}, "my": {"chunks": 7651.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.403789, "cps": 59.773438, "mean_p": 0.210538, "boundary_entropy": 0.377701, "bpb": 0.992904}}, "bpb": {"en": 1.81917, "de": 1.96093, "vi": 1.65214, "tr": 1.874, "fi": 2.08999, "ru": 1.20297, "el": 1.24947, "hi": 1.03211, "ta": 0.8683, "th": 0.98589, "zh": 2.68754, "my": 0.9929}, "gini_cpc": 0.248025, "gini_cps": 0.21255, "bpic": 5.774877, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9978258794898548, "vi": 1.2833290381613962, "tr": 1.043439055637148, "fi": 1.0840808621909293, "ru": 1.662325211760418, "el": 1.7307916741367242, "hi": 2.594417267598064, "ta": 2.6694967288495732, "th": 3.0344015590903513, "zh": 3.533900020848573, "my": 2.273421385792047}, "t": 1274.7}
{"event": "train", "step": 5500, "phase": 1, "objective": "global", "lm_loss": 1.11536, "bpb": 1.60913, "chunk_loss": 0.030083, "F": 0.17303, "G": 0.20808, "chunks": 177.19, "bytes_per_chunk": 5.779, "boundary_entropy": 0.20844, "router_noise": 0.0, "grad_norm": 0.3518, "lr": 0.0002985, "router_trainable": true, "sec_per_step": 0.203, "alpha": 0.03, "t": 1275.4}
{"event": "probe", "step": 6000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3086.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.177326, "cps": 24.109375, "mean_p": 0.195928, "boundary_entropy": 0.13116, "bpb": 1.799621}, "de": {"chunks": 3658.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.178152, "cps": 28.578125, "mean_p": 0.192154, "boundary_entropy": 0.12622, "bpb": 1.932114}, "vi": {"chunks": 4057.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.220071, "cps": 31.695312, "mean_p": 0.2043, "boundary_entropy": 0.195372, "bpb": 1.610779}, "tr": {"chunks": 3379.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.186654, "cps": 26.398438, "mean_p": 0.211919, "boundary_entropy": 0.213456, "bpb": 1.852034}, "fi": {"chunks": 3575.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.190048, "cps": 27.929688, "mean_p": 0.19445, "boundary_entropy": 0.130462, "bpb": 2.064629}, "ru": {"chunks": 5804.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.30628, "cps": 45.34375, "mean_p": 0.228341, "boundary_entropy": 0.286911, "bpb": 1.178678}, "el": {"chunks": 6880.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.335234, "cps": 53.75, "mean_p": 0.274566, "boundary_entropy": 0.37743, "bpb": 1.235562}, "hi": {"chunks": 7160.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.425557, "cps": 55.9375, "mean_p": 0.182749, "boundary_entropy": 0.241342, "bpb": 1.01549}, "ta": {"chunks": 9064.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.484836, "cps": 70.8125, "mean_p": 0.228524, "boundary_entropy": 0.332891, "bpb": 0.863057}, "th": {"chunks": 7760.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.482437, "cps": 60.625, "mean_p": 0.190295, "boundary_entropy": 0.27168, "bpb": 0.973795}, "zh": {"chunks": 3703.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.62699, "cps": 28.929688, "mean_p": 0.231801, "boundary_entropy": 0.087422, "bpb": 2.651698}, "my": {"chunks": 12579.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.66387, "cps": 98.273438, "mean_p": 0.253979, "boundary_entropy": 0.411659, "bpb": 0.98215}}, "bpb": {"en": 1.79962, "de": 1.93211, "vi": 1.61078, "tr": 1.85203, "fi": 2.06463, "ru": 1.17868, "el": 1.23556, "hi": 1.01549, "ta": 0.86306, "th": 0.9738, "zh": 2.6517, "my": 0.98215}, "gini_cpc": 0.264685, "gini_cps": 0.251834, "bpic": 5.427919, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0046608814813036, "vi": 1.2410522441304257, "tr": 1.0526059941655241, "fi": 1.0717472088989175, "ru": 1.7272149480930992, "el": 1.8904961251604946, "hi": 2.399861330323688, "ta": 2.734151818330096, "th": 2.7206260648277514, "zh": 3.5358063210650155, "my": 3.743785341633568}, "t": 1375.7}
{"event": "train", "step": 6000, "phase": 1, "objective": "global", "lm_loss": 1.20072, "bpb": 1.73227, "chunk_loss": 0.03025, "F": 0.18311, "G": 0.20724, "chunks": 187.5, "bytes_per_chunk": 5.461, "boundary_entropy": 0.19081, "router_noise": 0.0, "grad_norm": 0.3545, "lr": 0.0002981, "router_trainable": true, "sec_per_step": 0.202, "alpha": 0.03, "t": 1376.3}
{"event": "probe", "step": 6500, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2960.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.170086, "cps": 23.125, "mean_p": 0.185099, "boundary_entropy": 0.115549, "bpb": 1.784298}, "de": {"chunks": 3585.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.174597, "cps": 28.007812, "mean_p": 0.187114, "boundary_entropy": 0.119041, "bpb": 1.918888}, "vi": {"chunks": 4051.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.219745, "cps": 31.648438, "mean_p": 0.20167, "boundary_entropy": 0.19508, "bpb": 1.594239}, "tr": {"chunks": 3280.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.181185, "cps": 25.625, "mean_p": 0.202572, "boundary_entropy": 0.192142, "bpb": 1.834211}, "fi": {"chunks": 3665.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.194833, "cps": 28.632812, "mean_p": 0.199593, "boundary_entropy": 0.118654, "bpb": 2.052433}, "ru": {"chunks": 5625.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.296834, "cps": 43.945312, "mean_p": 0.221995, "boundary_entropy": 0.277917, "bpb": 1.171158}, "el": {"chunks": 6453.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.314428, "cps": 50.414062, "mean_p": 0.266366, "boundary_entropy": 0.365428, "bpb": 1.205177}, "hi": {"chunks": 8033.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.477444, "cps": 62.757812, "mean_p": 0.204681, "boundary_entropy": 0.235144, "bpb": 0.988324}, "ta": {"chunks": 9240.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.49425, "cps": 72.1875, "mean_p": 0.20674, "boundary_entropy": 0.270103, "bpb": 0.834994}, "th": {"chunks": 7491.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.465713, "cps": 58.523438, "mean_p": 0.190455, "boundary_entropy": 0.278306, "bpb": 0.964789}, "zh": {"chunks": 3650.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.618016, "cps": 28.515625, "mean_p": 0.227187, "boundary_entropy": 0.085771, "bpb": 2.627189}, "my": {"chunks": 7473.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.394395, "cps": 58.382812, "mean_p": 0.221759, "boundary_entropy": 0.399862, "bpb": 0.963788}}, "bpb": {"en": 1.7843, "de": 1.91889, "vi": 1.59424, "tr": 1.83421, "fi": 2.05243, "ru": 1.17116, "el": 1.20518, "hi": 0.98832, "ta": 0.83499, "th": 0.96479, "zh": 2.62719, "my": 0.96379}, "gini_cpc": 0.245195, "gini_cps": 0.214041, "bpic": 5.858715, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.026524128594576, "vi": 1.2919672662898862, "tr": 1.065260200265448, "fi": 1.1454984468547011, "ru": 1.7452020787278044, "el": 1.8486437596052419, "hi": 2.80708202481828, "ta": 2.9058882632297984, "th": 2.738111573650119, "zh": 3.6335557747046066, "my": 2.3188038636324735}, "t": 1481.0}
{"event": "train", "step": 6500, "phase": 1, "objective": "global", "lm_loss": 1.10081, "bpb": 1.58813, "chunk_loss": 0.030077, "F": 0.17018, "G": 0.19932, "chunks": 174.27, "bytes_per_chunk": 5.876, "boundary_entropy": 0.18936, "router_noise": 0.0, "grad_norm": 0.3475, "lr": 0.0002976, "router_trainable": true, "sec_per_step": 0.211, "alpha": 0.03, "t": 1481.6}
{"event": "probe", "step": 7000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3038.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.174568, "cps": 23.734375, "mean_p": 0.186873, "boundary_entropy": 0.107978, "bpb": 1.769618}, "de": {"chunks": 3658.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.178152, "cps": 28.578125, "mean_p": 0.189396, "boundary_entropy": 0.11235, "bpb": 1.897587}, "vi": {"chunks": 4000.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.216979, "cps": 31.25, "mean_p": 0.200465, "boundary_entropy": 0.18891, "bpb": 1.561477}, "tr": {"chunks": 3281.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.181241, "cps": 25.632812, "mean_p": 0.204609, "boundary_entropy": 0.193481, "bpb": 1.813172}, "fi": {"chunks": 3589.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.190793, "cps": 28.039062, "mean_p": 0.193807, "boundary_entropy": 0.115797, "bpb": 2.01745}, "ru": {"chunks": 5421.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.286069, "cps": 42.351562, "mean_p": 0.206793, "boundary_entropy": 0.249173, "bpb": 1.151478}, "el": {"chunks": 6561.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.31969, "cps": 51.257812, "mean_p": 0.272408, "boundary_entropy": 0.37391, "bpb": 1.190598}, "hi": {"chunks": 7476.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.444339, "cps": 58.40625, "mean_p": 0.192149, "boundary_entropy": 0.223433, "bpb": 0.972588}, "ta": {"chunks": 7948.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.42514, "cps": 62.09375, "mean_p": 0.181383, "boundary_entropy": 0.249917, "bpb": 0.83442}, "th": {"chunks": 7768.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.482934, "cps": 60.6875, "mean_p": 0.187181, "boundary_entropy": 0.265031, "bpb": 0.94514}, "zh": {"chunks": 3629.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.61446, "cps": 28.351562, "mean_p": 0.224901, "boundary_entropy": 0.081117, "bpb": 2.597136}, "my": {"chunks": 6807.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.359246, "cps": 53.179688, "mean_p": 0.212527, "boundary_entropy": 0.385167, "bpb": 0.944773}}, "bpb": {"en": 1.76962, "de": 1.89759, "vi": 1.56148, "tr": 1.81317, "fi": 2.01745, "ru": 1.15148, "el": 1.1906, "hi": 0.97259, "ta": 0.83442, "th": 0.94514, "zh": 2.59714, "my": 0.94477}, "gini_cpc": 0.237842, "gini_cps": 0.194723, "bpic": 6.074791, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.020534391129461, "vi": 1.242948687388549, "tr": 1.0382263084797811, "fi": 1.092944038563292, "ru": 1.638726752255077, "el": 1.831325502365929, "hi": 2.545367944618292, "ta": 2.4353912402428297, "th": 2.7664606825613514, "zh": 3.5198963596231234, "my": 2.0579211244856683}, "t": 1581.9}
{"event": "train", "step": 7000, "phase": 1, "objective": "global", "lm_loss": 1.15151, "bpb": 1.66128, "chunk_loss": 0.03011, "F": 0.17613, "G": 0.20133, "chunks": 180.36, "bytes_per_chunk": 5.678, "boundary_entropy": 0.17001, "router_noise": 0.0, "grad_norm": 0.3648, "lr": 0.000297, "router_trainable": true, "sec_per_step": 0.202, "alpha": 0.03, "t": 1582.4}
{"event": "probe", "step": 7500, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2988.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.171695, "cps": 23.34375, "mean_p": 0.182635, "boundary_entropy": 0.100432, "bpb": 1.75346}, "de": {"chunks": 3592.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.174938, "cps": 28.0625, "mean_p": 0.184823, "boundary_entropy": 0.103909, "bpb": 1.887355}, "vi": {"chunks": 4011.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.217575, "cps": 31.335938, "mean_p": 0.201321, "boundary_entropy": 0.187725, "bpb": 1.551906}, "tr": {"chunks": 3257.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.179915, "cps": 25.445312, "mean_p": 0.20287, "boundary_entropy": 0.186804, "bpb": 1.794548}, "fi": {"chunks": 3398.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.180639, "cps": 26.546875, "mean_p": 0.182413, "boundary_entropy": 0.10559, "bpb": 1.99819}, "ru": {"chunks": 5637.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.297467, "cps": 44.039062, "mean_p": 0.21323, "boundary_entropy": 0.24845, "bpb": 1.135125}, "el": {"chunks": 6261.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.305072, "cps": 48.914062, "mean_p": 0.252118, "boundary_entropy": 0.339562, "bpb": 1.173727}, "hi": {"chunks": 7395.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.439525, "cps": 57.773438, "mean_p": 0.183482, "boundary_entropy": 0.196212, "bpb": 0.962375}, "ta": {"chunks": 8552.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.457449, "cps": 66.8125, "mean_p": 0.178907, "boundary_entropy": 0.233843, "bpb": 0.831958}, "th": {"chunks": 7748.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.481691, "cps": 60.53125, "mean_p": 0.184541, "boundary_entropy": 0.256407, "bpb": 0.9408}, "zh": {"chunks": 3688.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.62445, "cps": 28.8125, "mean_p": 0.227109, "boundary_entropy": 0.076695, "bpb": 2.585757}, "my": {"chunks": 8988.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.474351, "cps": 70.21875, "mean_p": 0.232005, "boundary_entropy": 0.40183, "bpb": 0.943779}}, "bpb": {"en": 1.75346, "de": 1.88736, "vi": 1.55191, "tr": 1.79455, "fi": 1.99819, "ru": 1.13513, "el": 1.17373, "hi": 0.96237, "ta": 0.83196, "th": 0.9408, "zh": 2.58576, "my": 0.94378}, "gini_cpc": 0.246782, "gini_cps": 0.219371, "bpic": 5.85791, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.018890347343692, "vi": 1.2672230010358765, "tr": 1.0478780281597027, "fi": 1.0520951489761075, "ru": 1.7325363194201608, "el": 1.7768320761484764, "hi": 2.5599214092626075, "ta": 2.6643161036765406, "th": 2.805511633103755, "zh": 3.636980703035644, "my": 2.762760351094053}, "t": 1679.4}
{"event": "train", "step": 7500, "phase": 1, "objective": "global", "lm_loss": 1.15582, "bpb": 1.66749, "chunk_loss": 0.03007, "F": 0.16895, "G": 0.19169, "chunks": 173.0, "bytes_per_chunk": 5.919, "boundary_entropy": 0.16005, "router_noise": 0.0, "grad_norm": 0.3419, "lr": 0.0002964, "router_trainable": true, "sec_per_step": 0.195, "alpha": 0.03, "t": 1680.0}
{"event": "probe", "step": 8000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3076.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.176751, "cps": 24.03125, "mean_p": 0.188131, "boundary_entropy": 0.097727, "bpb": 1.742609}, "de": {"chunks": 3656.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.178055, "cps": 28.5625, "mean_p": 0.1886, "boundary_entropy": 0.102539, "bpb": 1.867817}, "vi": {"chunks": 4046.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.219474, "cps": 31.609375, "mean_p": 0.204077, "boundary_entropy": 0.192455, "bpb": 1.542123}, "tr": {"chunks": 3221.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.177926, "cps": 25.164062, "mean_p": 0.205079, "boundary_entropy": 0.19276, "bpb": 1.773057}, "fi": {"chunks": 3490.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.18553, "cps": 27.265625, "mean_p": 0.187252, "boundary_entropy": 0.097205, "bpb": 1.972811}, "ru": {"chunks": 5658.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.298575, "cps": 44.203125, "mean_p": 0.219977, "boundary_entropy": 0.260342, "bpb": 1.127053}, "el": {"chunks": 5966.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.290698, "cps": 46.609375, "mean_p": 0.249482, "boundary_entropy": 0.344117, "bpb": 1.164968}, "hi": {"chunks": 6937.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.412303, "cps": 54.195312, "mean_p": 0.171663, "boundary_entropy": 0.189696, "bpb": 0.954433}, "ta": {"chunks": 9548.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.510725, "cps": 74.59375, "mean_p": 0.19356, "boundary_entropy": 0.233193, "bpb": 0.809908}, "th": {"chunks": 7244.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.450357, "cps": 56.59375, "mean_p": 0.176406, "boundary_entropy": 0.254299, "bpb": 0.933306}, "zh": {"chunks": 3660.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.619709, "cps": 28.59375, "mean_p": 0.225307, "boundary_entropy": 0.073551, "bpb": 2.561077}, "my": {"chunks": 7649.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.403684, "cps": 59.757812, "mean_p": 0.233689, "boundary_entropy": 0.419122, "bpb": 0.931863}}, "bpb": {"en": 1.74261, "de": 1.86782, "vi": 1.54212, "tr": 1.77306, "fi": 1.97281, "ru": 1.12705, "el": 1.16497, "hi": 0.95443, "ta": 0.80991, "th": 0.93331, "zh": 2.56108, "my": 0.93186}, "gini_cpc": 0.243562, "gini_cps": 0.210775, "bpic": 5.982463, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0073759282608723, "vi": 1.2417109917352749, "tr": 1.0066487589761435, "fi": 1.0496664894423406, "ru": 1.6892406270693874, "el": 1.6446753862439978, "hi": 2.3326759439421725, "ta": 2.8895135135887116, "th": 2.547975017310913, "zh": 3.5061091473395285, "my": 2.283910462087328}, "t": 1777.0}
{"event": "eval", "step": 8000, "phase": 1, "bpb": {"en": 1.74176, "de": 1.87535, "vi": 1.55826, "tr": 1.766, "fi": 1.98562, "ru": 1.12578, "el": 1.16128, "hi": 0.96017, "ta": 0.81468, "th": 0.9329, "zh": 2.56452, "my": 0.93981}, "bytes_per_chunk": {"en": 5.753, "de": 5.768, "vi": 6.257, "tr": 6.149, "fi": 5.448, "ru": 5.982, "el": 6.355, "hi": 6.231, "ta": 5.469, "th": 6.709, "zh": 4.557, "my": 7.118}, "bpb_mean": 1.45218, "t": 1782.5}
{"event": "train", "step": 8000, "phase": 1, "objective": "global", "lm_loss": 1.05681, "bpb": 1.52465, "chunk_loss": 0.030092, "F": 0.17078, "G": 0.19645, "chunks": 174.88, "bytes_per_chunk": 5.856, "boundary_entropy": 0.17404, "router_noise": 0.0, "grad_norm": 0.3284, "lr": 0.0002957, "router_trainable": true, "sec_per_step": 0.205, "alpha": 0.03, "t": 1782.7}
{"event": "train", "step": 8500, "phase": 1, "objective": "global", "lm_loss": 1.11596, "bpb": 1.61, "chunk_loss": 0.030138, "F": 0.17706, "G": 0.19703, "chunks": 181.31, "bytes_per_chunk": 5.648, "boundary_entropy": 0.15334, "router_noise": 0.0, "grad_norm": 0.323, "lr": 0.0002949, "router_trainable": true, "sec_per_step": 0.192, "alpha": 0.03, "t": 1878.7}
{"event": "train", "step": 9000, "phase": 1, "objective": "global", "lm_loss": 1.1292, "bpb": 1.62909, "chunk_loss": 0.03018, "F": 0.18074, "G": 0.2046, "chunks": 185.08, "bytes_per_chunk": 5.533, "boundary_entropy": 0.15302, "router_noise": 0.0, "grad_norm": 0.3391, "lr": 0.0002941, "router_trainable": true, "sec_per_step": 0.19, "alpha": 0.03, "t": 1973.6}
{"event": "train", "step": 9500, "phase": 1, "objective": "global", "lm_loss": 1.1129, "bpb": 1.60557, "chunk_loss": 0.030099, "F": 0.17793, "G": 0.20039, "chunks": 182.2, "bytes_per_chunk": 5.62, "boundary_entropy": 0.15013, "router_noise": 0.0, "grad_norm": 0.3363, "lr": 0.0002933, "router_trainable": true, "sec_per_step": 0.193, "alpha": 0.03, "t": 2070.1}
{"event": "train", "step": 10000, "phase": 1, "objective": "global", "lm_loss": 1.10302, "bpb": 1.59132, "chunk_loss": 0.030072, "F": 0.17336, "G": 0.19621, "chunks": 177.52, "bytes_per_chunk": 5.769, "boundary_entropy": 0.14699, "router_noise": 0.0, "grad_norm": 0.294, "lr": 0.0002924, "router_trainable": true, "sec_per_step": 0.188, "alpha": 0.03, "t": 2163.9}
{"event": "train", "step": 10500, "phase": 1, "objective": "global", "lm_loss": 1.02139, "bpb": 1.47355, "chunk_loss": 0.030145, "F": 0.17822, "G": 0.2102, "chunks": 182.5, "bytes_per_chunk": 5.611, "boundary_entropy": 0.18588, "router_noise": 0.0, "grad_norm": 0.2918, "lr": 0.0002914, "router_trainable": true, "sec_per_step": 0.185, "alpha": 0.03, "t": 2256.3}
{"event": "train", "step": 11000, "phase": 1, "objective": "global", "lm_loss": 1.10039, "bpb": 1.58752, "chunk_loss": 0.030057, "F": 0.17082, "G": 0.19396, "chunks": 174.92, "bytes_per_chunk": 5.854, "boundary_entropy": 0.14659, "router_noise": 0.0, "grad_norm": 0.316, "lr": 0.0002903, "router_trainable": true, "sec_per_step": 0.187, "alpha": 0.03, "t": 2349.7}
{"event": "train", "step": 11500, "phase": 1, "objective": "global", "lm_loss": 1.07591, "bpb": 1.55221, "chunk_loss": 0.030107, "F": 0.17825, "G": 0.19932, "chunks": 182.53, "bytes_per_chunk": 5.61, "boundary_entropy": 0.14672, "router_noise": 0.0, "grad_norm": 0.3044, "lr": 0.0002893, "router_trainable": true, "sec_per_step": 0.187, "alpha": 0.03, "t": 2443.3}
{"event": "probe", "step": 12000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3037.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.17451, "cps": 23.726562, "mean_p": 0.182434, "boundary_entropy": 0.073358, "bpb": 1.681803}, "de": {"chunks": 3585.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.174597, "cps": 28.007812, "mean_p": 0.179258, "boundary_entropy": 0.076554, "bpb": 1.800816}, "vi": {"chunks": 4081.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.221372, "cps": 31.882812, "mean_p": 0.215569, "boundary_entropy": 0.203802, "bpb": 1.461307}, "tr": {"chunks": 3158.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.174446, "cps": 24.671875, "mean_p": 0.214554, "boundary_entropy": 0.211548, "bpb": 1.695126}, "fi": {"chunks": 3464.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.184148, "cps": 27.0625, "mean_p": 0.183051, "boundary_entropy": 0.077036, "bpb": 1.884121}, "ru": {"chunks": 5668.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.299103, "cps": 44.28125, "mean_p": 0.217645, "boundary_entropy": 0.239265, "bpb": 1.070839}, "el": {"chunks": 6323.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.308093, "cps": 49.398438, "mean_p": 0.286786, "boundary_entropy": 0.383798, "bpb": 1.101747}, "hi": {"chunks": 7863.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.46734, "cps": 61.429688, "mean_p": 0.193038, "boundary_entropy": 0.171298, "bpb": 0.888257}, "ta": {"chunks": 9744.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.521209, "cps": 76.125, "mean_p": 0.191036, "boundary_entropy": 0.215459, "bpb": 0.763282}, "th": {"chunks": 8431.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.524153, "cps": 65.867188, "mean_p": 0.19649, "boundary_entropy": 0.257509, "bpb": 0.876644}, "zh": {"chunks": 3693.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.625296, "cps": 28.851562, "mean_p": 0.228948, "boundary_entropy": 0.056108, "bpb": 2.476535}, "my": {"chunks": 8879.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.468598, "cps": 69.367188, "mean_p": 0.284862, "boundary_entropy": 0.496475, "bpb": 0.874251}}, "bpb": {"en": 1.6818, "de": 1.80082, "vi": 1.46131, "tr": 1.69513, "fi": 1.88412, "ru": 1.07084, "el": 1.10175, "hi": 0.88826, "ta": 0.76328, "th": 0.87664, "zh": 2.47654, "my": 0.87425}, "gini_cpc": 0.255491, "gini_cps": 0.234203, "bpic": 5.649987, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0004976689627743, "vi": 1.2685359545802475, "tr": 0.9996337316207147, "fi": 1.055225622948426, "ru": 1.7139571320468834, "el": 1.7654753773286895, "hi": 2.6780120759273567, "ta": 2.98669678229887, "th": 3.003567195143433, "zh": 3.583151683475759, "my": 2.685220834533262}, "t": 2540.3}
{"event": "eval", "step": 12000, "phase": 1, "bpb": {"en": 1.68067, "de": 1.80935, "vi": 1.47888, "tr": 1.68934, "fi": 1.89731, "ru": 1.07003, "el": 1.09819, "hi": 0.89502, "ta": 0.76858, "th": 0.87581, "zh": 2.47946, "my": 0.88133}, "bytes_per_chunk": {"en": 5.801, "de": 5.875, "vi": 6.096, "tr": 6.218, "fi": 5.507, "ru": 5.979, "el": 5.871, "hi": 5.547, "ta": 5.341, "th": 5.642, "zh": 4.531, "my": 6.096}, "bpb_mean": 1.38533, "t": 2545.9}
{"event": "train", "step": 12000, "phase": 1, "objective": "global", "lm_loss": 1.11896, "bpb": 1.61432, "chunk_loss": 0.030136, "F": 0.17979, "G": 0.20481, "chunks": 184.11, "bytes_per_chunk": 5.562, "boundary_entropy": 0.15479, "router_noise": 0.0, "grad_norm": 0.3202, "lr": 0.0002881, "router_trainable": true, "sec_per_step": 0.205, "alpha": 0.03, "t": 2546.0}
{"event": "train", "step": 12500, "phase": 1, "objective": "global", "lm_loss": 1.11837, "bpb": 1.61347, "chunk_loss": 0.030149, "F": 0.17952, "G": 0.20438, "chunks": 183.83, "bytes_per_chunk": 5.57, "boundary_entropy": 0.1433, "router_noise": 0.0, "grad_norm": 0.2933, "lr": 0.0002869, "router_trainable": true, "sec_per_step": 0.187, "alpha": 0.03, "t": 2639.7}
{"event": "train", "step": 13000, "phase": 1, "objective": "global", "lm_loss": 1.01472, "bpb": 1.46393, "chunk_loss": 0.030074, "F": 0.175, "G": 0.19703, "chunks": 179.2, "bytes_per_chunk": 5.714, "boundary_entropy": 0.13823, "router_noise": 0.0, "grad_norm": 0.3049, "lr": 0.0002857, "router_trainable": true, "sec_per_step": 0.187, "alpha": 0.03, "t": 2733.2}
{"event": "train", "step": 13500, "phase": 1, "objective": "global", "lm_loss": 1.06719, "bpb": 1.53963, "chunk_loss": 0.030181, "F": 0.17931, "G": 0.2066, "chunks": 183.61, "bytes_per_chunk": 5.577, "boundary_entropy": 0.14154, "router_noise": 0.0, "grad_norm": 0.4187, "lr": 0.0002844, "router_trainable": true, "sec_per_step": 0.189, "alpha": 0.03, "t": 2827.6}
{"event": "train", "step": 14000, "phase": 1, "objective": "global", "lm_loss": 0.98761, "bpb": 1.42482, "chunk_loss": 0.030083, "F": 0.173, "G": 0.1968, "chunks": 177.16, "bytes_per_chunk": 5.78, "boundary_entropy": 0.14543, "router_noise": 0.0, "grad_norm": 0.3016, "lr": 0.000283, "router_trainable": true, "sec_per_step": 0.189, "alpha": 0.03, "t": 2921.9}
{"event": "train", "step": 14500, "phase": 1, "objective": "global", "lm_loss": 1.051, "bpb": 1.51628, "chunk_loss": 0.030072, "F": 0.17055, "G": 0.19837, "chunks": 174.64, "bytes_per_chunk": 5.863, "boundary_entropy": 0.15264, "router_noise": 0.0, "grad_norm": 0.2843, "lr": 0.0002816, "router_trainable": true, "sec_per_step": 0.19, "alpha": 0.03, "t": 3016.8}
{"event": "train", "step": 15000, "phase": 1, "objective": "global", "lm_loss": 1.0518, "bpb": 1.51742, "chunk_loss": 0.030117, "F": 0.17639, "G": 0.19756, "chunks": 180.62, "bytes_per_chunk": 5.669, "boundary_entropy": 0.12935, "router_noise": 0.0, "grad_norm": 0.2964, "lr": 0.0002801, "router_trainable": true, "sec_per_step": 0.191, "alpha": 0.03, "t": 3112.5}
{"event": "train", "step": 15500, "phase": 1, "objective": "global", "lm_loss": 1.01414, "bpb": 1.46309, "chunk_loss": 0.030126, "F": 0.1806, "G": 0.19612, "chunks": 184.94, "bytes_per_chunk": 5.537, "boundary_entropy": 0.12656, "router_noise": 0.0, "grad_norm": 0.2849, "lr": 0.0002786, "router_trainable": true, "sec_per_step": 0.183, "alpha": 0.03, "t": 3204.2}
{"event": "probe", "step": 16000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3034.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.174338, "cps": 23.703125, "mean_p": 0.180541, "boundary_entropy": 0.063443, "bpb": 1.643545}, "de": {"chunks": 3679.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.179175, "cps": 28.742188, "mean_p": 0.184299, "boundary_entropy": 0.07108, "bpb": 1.754779}, "vi": {"chunks": 4151.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.22517, "cps": 32.429688, "mean_p": 0.216618, "boundary_entropy": 0.197149, "bpb": 1.416406}, "tr": {"chunks": 3240.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.178976, "cps": 25.3125, "mean_p": 0.212429, "boundary_entropy": 0.1935, "bpb": 1.655339}, "fi": {"chunks": 3502.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.186168, "cps": 27.359375, "mean_p": 0.185145, "boundary_entropy": 0.066626, "bpb": 1.832576}, "ru": {"chunks": 5676.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.299525, "cps": 44.34375, "mean_p": 0.220455, "boundary_entropy": 0.235462, "bpb": 1.032969}, "el": {"chunks": 5975.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.291137, "cps": 46.679688, "mean_p": 0.271526, "boundary_entropy": 0.362439, "bpb": 1.068362}, "hi": {"chunks": 7588.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.450996, "cps": 59.28125, "mean_p": 0.183345, "boundary_entropy": 0.147689, "bpb": 0.86392}, "ta": {"chunks": 8938.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.478096, "cps": 69.828125, "mean_p": 0.174494, "boundary_entropy": 0.185438, "bpb": 0.743732}, "th": {"chunks": 8207.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.510227, "cps": 64.117188, "mean_p": 0.187688, "boundary_entropy": 0.238944, "bpb": 0.84985}, "zh": {"chunks": 3690.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.624788, "cps": 28.828125, "mean_p": 0.227858, "boundary_entropy": 0.049546, "bpb": 2.426744}, "my": {"chunks": 8683.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.458254, "cps": 67.835938, "mean_p": 0.312813, "boundary_entropy": 0.532872, "bpb": 0.851588}}, "bpb": {"en": 1.64354, "de": 1.75478, "vi": 1.41641, "tr": 1.65534, "fi": 1.83258, "ru": 1.03297, "el": 1.06836, "hi": 0.86392, "ta": 0.74373, "th": 0.84985, "zh": 2.42674, "my": 0.85159}, "gini_cpc": 0.248037, "gini_cps": 0.219489, "bpic": 5.783057, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0277463058405811, "vi": 1.2915705540623676, "tr": 1.026604119232615, "fi": 1.067856270158763, "ru": 1.718073409016198, "el": 1.6699582290923594, "hi": 2.5869068633740833, "ta": 2.7423534259051774, "th": 2.9266575741226415, "zh": 3.583781038064817, "my": 2.628542290185581}, "t": 3298.9}
{"event": "eval", "step": 16000, "phase": 1, "bpb": {"en": 1.64109, "de": 1.7623, "vi": 1.43377, "tr": 1.64914, "fi": 1.84601, "ru": 1.0331, "el": 1.06528, "hi": 0.86916, "ta": 0.7487, "th": 0.8491, "zh": 2.43022, "my": 0.85754}, "bytes_per_chunk": {"en": 5.767, "de": 5.686, "vi": 6.06, "tr": 6.065, "fi": 5.443, "ru": 5.966, "el": 6.167, "hi": 5.659, "ta": 5.749, "th": 5.71, "zh": 4.545, "my": 5.677}, "bpb_mean": 1.34879, "t": 3304.4}
{"event": "train", "step": 16000, "phase": 1, "objective": "global", "lm_loss": 0.98722, "bpb": 1.42426, "chunk_loss": 0.030201, "F": 0.17671, "G": 0.20876, "chunks": 180.95, "bytes_per_chunk": 5.659, "boundary_entropy": 0.16626, "router_noise": 0.0, "grad_norm": 0.2751, "lr": 0.000277, "router_trainable": true, "sec_per_step": 0.201, "alpha": 0.03, "t": 3304.6}
{"event": "train", "step": 16500, "phase": 1, "objective": "global", "lm_loss": 1.09418, "bpb": 1.57857, "chunk_loss": 0.030152, "F": 0.17839, "G": 0.20302, "chunks": 182.67, "bytes_per_chunk": 5.606, "boundary_entropy": 0.13673, "router_noise": 0.0, "grad_norm": 0.3054, "lr": 0.0002754, "router_trainable": true, "sec_per_step": 0.183, "alpha": 0.03, "t": 3396.2}
{"event": "train", "step": 17000, "phase": 1, "objective": "global", "lm_loss": 1.0308, "bpb": 1.48714, "chunk_loss": 0.030127, "F": 0.17633, "G": 0.20601, "chunks": 180.56, "bytes_per_chunk": 5.671, "boundary_entropy": 0.14805, "router_noise": 0.0, "grad_norm": 0.2804, "lr": 0.0002737, "router_trainable": true, "sec_per_step": 0.183, "alpha": 0.03, "t": 3487.8}
{"event": "train", "step": 17500, "phase": 1, "objective": "global", "lm_loss": 1.08134, "bpb": 1.56005, "chunk_loss": 0.030143, "F": 0.17911, "G": 0.20135, "chunks": 183.41, "bytes_per_chunk": 5.583, "boundary_entropy": 0.13519, "router_noise": 0.0, "grad_norm": 0.2799, "lr": 0.000272, "router_trainable": true, "sec_per_step": 0.178, "alpha": 0.03, "t": 3577.0}
{"event": "train", "step": 18000, "phase": 1, "objective": "global", "lm_loss": 1.00268, "bpb": 1.44657, "chunk_loss": 0.03004, "F": 0.17258, "G": 0.20376, "chunks": 176.72, "bytes_per_chunk": 5.795, "boundary_entropy": 0.15377, "router_noise": 0.0, "grad_norm": 0.2826, "lr": 0.0002703, "router_trainable": true, "sec_per_step": 0.185, "alpha": 0.03, "t": 3669.4}
{"event": "train", "step": 18500, "phase": 1, "objective": "global", "lm_loss": 1.03398, "bpb": 1.49172, "chunk_loss": 0.030115, "F": 0.17603, "G": 0.20454, "chunks": 180.25, "bytes_per_chunk": 5.681, "boundary_entropy": 0.14112, "router_noise": 0.0, "grad_norm": 0.2999, "lr": 0.0002685, "router_trainable": true, "sec_per_step": 0.184, "alpha": 0.03, "t": 3761.2}
{"event": "train", "step": 19000, "phase": 1, "objective": "global", "lm_loss": 0.96933, "bpb": 1.39845, "chunk_loss": 0.030094, "F": 0.17227, "G": 0.19689, "chunks": 176.41, "bytes_per_chunk": 5.805, "boundary_entropy": 0.13334, "router_noise": 0.0, "grad_norm": 0.2742, "lr": 0.0002666, "router_trainable": true, "sec_per_step": 0.186, "alpha": 0.03, "t": 3854.3}
{"event": "train", "step": 19500, "phase": 1, "objective": "global", "lm_loss": 1.0406, "bpb": 1.50127, "chunk_loss": 0.030091, "F": 0.17172, "G": 0.19276, "chunks": 175.84, "bytes_per_chunk": 5.823, "boundary_entropy": 0.12975, "router_noise": 0.0, "grad_norm": 0.2771, "lr": 0.0002647, "router_trainable": true, "sec_per_step": 0.181, "alpha": 0.03, "t": 3945.0}
{"event": "probe", "step": 20000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3040.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.174683, "cps": 23.75, "mean_p": 0.179776, "boundary_entropy": 0.056451, "bpb": 1.613841}, "de": {"chunks": 3697.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.180052, "cps": 28.882812, "mean_p": 0.183777, "boundary_entropy": 0.060217, "bpb": 1.722083}, "vi": {"chunks": 4102.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.222512, "cps": 32.046875, "mean_p": 0.217773, "boundary_entropy": 0.198887, "bpb": 1.395815}, "tr": {"chunks": 3273.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.180799, "cps": 25.570312, "mean_p": 0.226129, "boundary_entropy": 0.217564, "bpb": 1.626751}, "fi": {"chunks": 3570.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.189783, "cps": 27.890625, "mean_p": 0.187886, "boundary_entropy": 0.059176, "bpb": 1.7972}, "ru": {"chunks": 5624.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.296781, "cps": 43.9375, "mean_p": 0.219978, "boundary_entropy": 0.233127, "bpb": 1.015275}, "el": {"chunks": 5986.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.291673, "cps": 46.765625, "mean_p": 0.28127, "boundary_entropy": 0.374342, "bpb": 1.035007}, "hi": {"chunks": 7588.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.450996, "cps": 59.28125, "mean_p": 0.179063, "boundary_entropy": 0.156032, "bpb": 0.840229}, "ta": {"chunks": 8948.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.478631, "cps": 69.90625, "mean_p": 0.171192, "boundary_entropy": 0.180612, "bpb": 0.724494}, "th": {"chunks": 8052.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.500591, "cps": 62.90625, "mean_p": 0.180364, "boundary_entropy": 0.224022, "bpb": 0.83296}, "zh": {"chunks": 3684.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.623772, "cps": 28.78125, "mean_p": 0.227384, "boundary_entropy": 0.046519, "bpb": 2.401156}, "my": {"chunks": 6664.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.351699, "cps": 52.0625, "mean_p": 0.266354, "boundary_entropy": 0.482366, "bpb": 0.82932}}, "bpb": {"en": 1.61384, "de": 1.72208, "vi": 1.39581, "tr": 1.62675, "fi": 1.7972, "ru": 1.01528, "el": 1.03501, "hi": 0.84023, "ta": 0.72449, "th": 0.83296, "zh": 2.40116, "my": 0.82932}, "gini_cpc": 0.245402, "gini_cps": 0.206833, "bpic": 5.975291, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0307363211210707, "vi": 1.2738052974176695, "tr": 1.0350134428140727, "fi": 1.0864428078196129, "ru": 1.6989736147757255, "el": 1.6697305910954847, "hi": 2.58180112614374, "ta": 2.740003026421362, "th": 2.8657165878638153, "zh": 3.570892000998093, "my": 2.0133633045565147}, "t": 4039.4}
{"event": "eval", "step": 20000, "phase": 1, "bpb": {"en": 1.61168, "de": 1.73082, "vi": 1.41436, "tr": 1.61934, "fi": 1.81147, "ru": 1.01378, "el": 1.0316, "hi": 0.84689, "ta": 0.72897, "th": 0.83351, "zh": 2.40249, "my": 0.83544}, "bytes_per_chunk": {"en": 5.762, "de": 5.658, "vi": 6.059, "tr": 6.024, "fi": 5.387, "ru": 6.095, "el": 6.181, "hi": 5.738, "ta": 5.737, "th": 5.869, "zh": 4.542, "my": 7.585}, "bpb_mean": 1.32336, "t": 4044.8}
{"event": "train", "step": 20000, "phase": 1, "objective": "global", "lm_loss": 1.07842, "bpb": 1.55583, "chunk_loss": 0.030118, "F": 0.17804, "G": 0.20218, "chunks": 182.31, "bytes_per_chunk": 5.617, "boundary_entropy": 0.12758, "router_noise": 0.0, "grad_norm": 0.2934, "lr": 0.0002628, "router_trainable": true, "sec_per_step": 0.2, "alpha": 0.03, "t": 4045.0}
{"event": "train", "step": 20500, "phase": 1, "objective": "global", "lm_loss": 0.97312, "bpb": 1.40392, "chunk_loss": 0.030196, "F": 0.17857, "G": 0.20451, "chunks": 182.86, "bytes_per_chunk": 5.6, "boundary_entropy": 0.14877, "router_noise": 0.0, "grad_norm": 0.2852, "lr": 0.0002608, "router_trainable": true, "sec_per_step": 0.179, "alpha": 0.03, "t": 4134.7}
{"event": "train", "step": 21000, "phase": 1, "objective": "global", "lm_loss": 1.02234, "bpb": 1.47493, "chunk_loss": 0.030089, "F": 0.17632, "G": 0.20691, "chunks": 180.55, "bytes_per_chunk": 5.672, "boundary_entropy": 0.15694, "router_noise": 0.0, "grad_norm": 0.2841, "lr": 0.0002587, "router_trainable": true, "sec_per_step": 0.187, "alpha": 0.03, "t": 4228.1}
{"event": "train", "step": 21500, "phase": 1, "objective": "global", "lm_loss": 0.93759, "bpb": 1.35265, "chunk_loss": 0.030075, "F": 0.17297, "G": 0.19877, "chunks": 177.12, "bytes_per_chunk": 5.781, "boundary_entropy": 0.14655, "router_noise": 0.0, "grad_norm": 0.2655, "lr": 0.0002567, "router_trainable": true, "sec_per_step": 0.183, "alpha": 0.03, "t": 4319.4}
{"event": "train", "step": 22000, "phase": 1, "objective": "global", "lm_loss": 1.00443, "bpb": 1.44909, "chunk_loss": 0.030079, "F": 0.17537, "G": 0.19482, "chunks": 179.58, "bytes_per_chunk": 5.702, "boundary_entropy": 0.12719, "router_noise": 0.0, "grad_norm": 0.2772, "lr": 0.0002545, "router_trainable": true, "sec_per_step": 0.183, "alpha": 0.03, "t": 4410.9}
{"event": "train", "step": 22500, "phase": 1, "objective": "global", "lm_loss": 0.97781, "bpb": 1.41068, "chunk_loss": 0.030139, "F": 0.17888, "G": 0.19996, "chunks": 183.17, "bytes_per_chunk": 5.59, "boundary_entropy": 0.13486, "router_noise": 0.0, "grad_norm": 0.2796, "lr": 0.0002524, "router_trainable": true, "sec_per_step": 0.182, "alpha": 0.03, "t": 4501.8}
{"event": "train", "step": 23000, "phase": 1, "objective": "global", "lm_loss": 1.01531, "bpb": 1.46478, "chunk_loss": 0.030093, "F": 0.1759, "G": 0.19925, "chunks": 180.12, "bytes_per_chunk": 5.685, "boundary_entropy": 0.13503, "router_noise": 0.0, "grad_norm": 0.2705, "lr": 0.0002502, "router_trainable": true, "sec_per_step": 0.187, "alpha": 0.03, "t": 4595.1}
{"event": "train", "step": 23500, "phase": 1, "objective": "global", "lm_loss": 0.97626, "bpb": 1.40845, "chunk_loss": 0.030079, "F": 0.17557, "G": 0.20917, "chunks": 179.78, "bytes_per_chunk": 5.696, "boundary_entropy": 0.16354, "router_noise": 0.0, "grad_norm": 0.269, "lr": 0.000248, "router_trainable": true, "sec_per_step": 0.182, "alpha": 0.03, "t": 4686.0}
{"event": "probe", "step": 24000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3032.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.174223, "cps": 23.6875, "mean_p": 0.179257, "boundary_entropy": 0.04991, "bpb": 1.592204}, "de": {"chunks": 3638.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.177178, "cps": 28.421875, "mean_p": 0.179551, "boundary_entropy": 0.050549, "bpb": 1.698028}, "vi": {"chunks": 4099.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.222349, "cps": 32.023438, "mean_p": 0.221971, "boundary_entropy": 0.211625, "bpb": 1.37313}, "tr": {"chunks": 3254.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.179749, "cps": 25.421875, "mean_p": 0.22064, "boundary_entropy": 0.205997, "bpb": 1.606061}, "fi": {"chunks": 3552.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.188826, "cps": 27.75, "mean_p": 0.186749, "boundary_entropy": 0.054042, "bpb": 1.754491}, "ru": {"chunks": 5924.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.312612, "cps": 46.28125, "mean_p": 0.221318, "boundary_entropy": 0.21838, "bpb": 0.996086}, "el": {"chunks": 6063.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.295425, "cps": 47.367188, "mean_p": 0.296429, "boundary_entropy": 0.387291, "bpb": 1.018983}, "hi": {"chunks": 7599.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.451649, "cps": 59.367188, "mean_p": 0.179184, "boundary_entropy": 0.143101, "bpb": 0.824715}, "ta": {"chunks": 9116.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.487617, "cps": 71.21875, "mean_p": 0.16896, "boundary_entropy": 0.168221, "bpb": 0.718033}, "th": {"chunks": 7568.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.4705, "cps": 59.125, "mean_p": 0.169643, "boundary_entropy": 0.219677, "bpb": 0.827489}, "zh": {"chunks": 3676.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.622418, "cps": 28.71875, "mean_p": 0.226623, "boundary_entropy": 0.043942, "bpb": 2.362251}, "my": {"chunks": 9090.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.479734, "cps": 71.015625, "mean_p": 0.28073, "boundary_entropy": 0.477075, "bpb": 0.812815}}, "bpb": {"en": 1.5922, "de": 1.69803, "vi": 1.37313, "tr": 1.60606, "fi": 1.75449, "ru": 0.99609, "el": 1.01898, "hi": 0.82471, "ta": 0.71803, "th": 0.82749, "zh": 2.36225, "my": 0.81282}, "gini_cpc": 0.244884, "gini_cps": 0.220701, "bpic": 5.761526, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.016963136887438, "vi": 1.276232205001814, "tr": 1.03172016854012, "fi": 1.0838170951735, "ru": 1.7943235566446905, "el": 1.6956712090958634, "hi": 2.5923658685757074, "ta": 2.7988122767858714, "th": 2.7005671551938373, "zh": 3.572539039499208, "my": 2.7535656188329343}, "t": 4780.6}
{"event": "eval", "step": 24000, "phase": 1, "bpb": {"en": 1.58953, "de": 1.70623, "vi": 1.39092, "tr": 1.59711, "fi": 1.76947, "ru": 0.99425, "el": 1.017, "hi": 0.83146, "ta": 0.72306, "th": 0.8266, "zh": 2.36673, "my": 0.81759}, "bytes_per_chunk": {"en": 5.745, "de": 5.713, "vi": 6.08, "tr": 6.083, "fi": 5.372, "ru": 5.826, "el": 6.045, "hi": 5.674, "ta": 5.667, "th": 6.164, "zh": 4.542, "my": 5.859}, "bpb_mean": 1.3025, "t": 4786.0}
{"event": "train", "step": 24000, "phase": 1, "objective": "global", "lm_loss": 1.07361, "bpb": 1.54889, "chunk_loss": 0.030148, "F": 0.18013, "G": 0.19914, "chunks": 184.45, "bytes_per_chunk": 5.552, "boundary_entropy": 0.117, "router_noise": 0.0, "grad_norm": 0.2908, "lr": 0.0002457, "router_trainable": true, "sec_per_step": 0.2, "alpha": 0.03, "t": 4786.2}
{"event": "train", "step": 24500, "phase": 1, "objective": "global", "lm_loss": 1.01412, "bpb": 1.46306, "chunk_loss": 0.03009, "F": 0.17404, "G": 0.19446, "chunks": 178.22, "bytes_per_chunk": 5.746, "boundary_entropy": 0.12306, "router_noise": 0.0, "grad_norm": 0.3111, "lr": 0.0002434, "router_trainable": true, "sec_per_step": 0.184, "alpha": 0.03, "t": 4878.3}
{"event": "train", "step": 25000, "phase": 1, "objective": "global", "lm_loss": 0.94808, "bpb": 1.36779, "chunk_loss": 0.030071, "F": 0.17038, "G": 0.19321, "chunks": 174.47, "bytes_per_chunk": 5.869, "boundary_entropy": 0.13196, "router_noise": 0.0, "grad_norm": 0.2815, "lr": 0.000241, "router_trainable": true, "sec_per_step": 0.183, "alpha": 0.03, "t": 4969.6}
{"event": "train", "step": 25500, "phase": 1, "objective": "global", "lm_loss": 0.9548, "bpb": 1.37749, "chunk_loss": 0.030118, "F": 0.17714, "G": 0.1979, "chunks": 181.39, "bytes_per_chunk": 5.645, "boundary_entropy": 0.12412, "router_noise": 0.0, "grad_norm": 0.2624, "lr": 0.0002387, "router_trainable": true, "sec_per_step": 0.177, "alpha": 0.03, "t": 5058.1}
{"event": "train", "step": 26000, "phase": 1, "objective": "global", "lm_loss": 0.92932, "bpb": 1.34072, "chunk_loss": 0.030076, "F": 0.17436, "G": 0.20289, "chunks": 178.55, "bytes_per_chunk": 5.735, "boundary_entropy": 0.15325, "router_noise": 0.0, "grad_norm": 0.2659, "lr": 0.0002363, "router_trainable": true, "sec_per_step": 0.172, "alpha": 0.03, "t": 5144.2}
{"event": "train", "step": 26500, "phase": 1, "objective": "global", "lm_loss": 0.98153, "bpb": 1.41604, "chunk_loss": 0.030085, "F": 0.17542, "G": 0.19158, "chunks": 179.62, "bytes_per_chunk": 5.701, "boundary_entropy": 0.10984, "router_noise": 0.0, "grad_norm": 0.2785, "lr": 0.0002338, "router_trainable": true, "sec_per_step": 0.171, "alpha": 0.03, "t": 5229.9}
{"event": "train", "step": 27000, "phase": 1, "objective": "global", "lm_loss": 0.90877, "bpb": 1.31108, "chunk_loss": 0.030073, "F": 0.17134, "G": 0.20173, "chunks": 175.45, "bytes_per_chunk": 5.836, "boundary_entropy": 0.15204, "router_noise": 0.0, "grad_norm": 0.2683, "lr": 0.0002313, "router_trainable": true, "sec_per_step": 0.18, "alpha": 0.03, "t": 5320.1}
{"event": "train", "step": 27500, "phase": 1, "objective": "global", "lm_loss": 1.0617, "bpb": 1.5317, "chunk_loss": 0.030144, "F": 0.17911, "G": 0.20226, "chunks": 183.41, "bytes_per_chunk": 5.583, "boundary_entropy": 0.12879, "router_noise": 0.0, "grad_norm": 0.2757, "lr": 0.0002288, "router_trainable": true, "sec_per_step": 0.186, "alpha": 0.03, "t": 5413.3}
{"event": "probe", "step": 28000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3032.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.174223, "cps": 23.6875, "mean_p": 0.177799, "boundary_entropy": 0.045779, "bpb": 1.573965}, "de": {"chunks": 3646.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.177568, "cps": 28.484375, "mean_p": 0.179569, "boundary_entropy": 0.047677, "bpb": 1.674497}, "vi": {"chunks": 4097.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.22224, "cps": 32.007812, "mean_p": 0.216677, "boundary_entropy": 0.191074, "bpb": 1.352733}, "tr": {"chunks": 3245.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.179252, "cps": 25.351562, "mean_p": 0.213162, "boundary_entropy": 0.184276, "bpb": 1.578437}, "fi": {"chunks": 3560.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.189251, "cps": 27.8125, "mean_p": 0.185574, "boundary_entropy": 0.047135, "bpb": 1.732124}, "ru": {"chunks": 5736.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.302691, "cps": 44.8125, "mean_p": 0.210671, "boundary_entropy": 0.196004, "bpb": 0.981201}, "el": {"chunks": 6024.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.293524, "cps": 47.0625, "mean_p": 0.292583, "boundary_entropy": 0.386711, "bpb": 1.0}, "hi": {"chunks": 6976.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.414621, "cps": 54.5, "mean_p": 0.163189, "boundary_entropy": 0.1383, "bpb": 0.818605}, "ta": {"chunks": 8232.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.440332, "cps": 64.3125, "mean_p": 0.153781, "boundary_entropy": 0.17511, "bpb": 0.714578}, "th": {"chunks": 7720.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.47995, "cps": 60.3125, "mean_p": 0.17159, "boundary_entropy": 0.216971, "bpb": 0.815198}, "zh": {"chunks": 3688.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.62445, "cps": 28.8125, "mean_p": 0.227489, "boundary_entropy": 0.04102, "bpb": 2.350712}, "my": {"chunks": 8507.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.448966, "cps": 66.460938, "mean_p": 0.259069, "boundary_entropy": 0.43733, "bpb": 0.793143}}, "bpb": {"en": 1.57397, "de": 1.6745, "vi": 1.35273, "tr": 1.57844, "fi": 1.73212, "ru": 0.9812, "el": 1.0, "hi": 0.81861, "ta": 0.71458, "th": 0.8152, "zh": 2.35071, "my": 0.79314}, "gini_cpc": 0.240762, "gini_cps": 0.206223, "bpic": 5.953508, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0191994494479382, "vi": 1.2756095008276243, "tr": 1.028866609376979, "fi": 1.086258124667134, "ru": 1.7373801351981677, "el": 1.6847638732629855, "hi": 2.379832122540352, "ta": 2.5274048554740345, "th": 2.7548068760698237, "zh": 3.584201299693438, "my": 2.5769617953148263}, "t": 5509.4}
{"event": "eval", "step": 28000, "phase": 1, "bpb": {"en": 1.57065, "de": 1.68193, "vi": 1.37086, "tr": 1.57122, "fi": 1.74906, "ru": 0.97991, "el": 0.99719, "hi": 0.82496, "ta": 0.72022, "th": 0.8134, "zh": 2.3538, "my": 0.79939}, "bytes_per_chunk": {"en": 5.739, "de": 5.711, "vi": 6.08, "tr": 6.107, "fi": 5.388, "ru": 5.904, "el": 6.151, "hi": 6.161, "ta": 6.346, "th": 6.101, "zh": 4.54, "my": 6.261}, "bpb_mean": 1.28605, "t": 5515.6}
{"event": "train", "step": 28000, "phase": 1, "objective": "global", "lm_loss": 0.98992, "bpb": 1.42815, "chunk_loss": 0.03014, "F": 0.17625, "G": 0.2065, "chunks": 180.48, "bytes_per_chunk": 5.674, "boundary_entropy": 0.14035, "router_noise": 0.0, "grad_norm": 0.2908, "lr": 0.0002263, "router_trainable": true, "sec_per_step": 0.205, "alpha": 0.03, "t": 5515.8}
{"event": "train", "step": 28500, "phase": 1, "objective": "global", "lm_loss": 1.00384, "bpb": 1.44823, "chunk_loss": 0.030135, "F": 0.17488, "G": 0.19946, "chunks": 179.08, "bytes_per_chunk": 5.718, "boundary_entropy": 0.12833, "router_noise": 0.0, "grad_norm": 0.2903, "lr": 0.0002238, "router_trainable": true, "sec_per_step": 0.184, "alpha": 0.03, "t": 5608.0}
{"event": "train", "step": 29000, "phase": 1, "objective": "global", "lm_loss": 1.02567, "bpb": 1.47972, "chunk_loss": 0.030074, "F": 0.17363, "G": 0.19644, "chunks": 177.8, "bytes_per_chunk": 5.759, "boundary_entropy": 0.12178, "router_noise": 0.0, "grad_norm": 0.2723, "lr": 0.0002212, "router_trainable": true, "sec_per_step": 0.183, "alpha": 0.03, "t": 5699.4}
{"event": "train", "step": 29500, "phase": 1, "objective": "global", "lm_loss": 0.95741, "bpb": 1.38125, "chunk_loss": 0.030095, "F": 0.17401, "G": 0.19805, "chunks": 178.19, "bytes_per_chunk": 5.747, "boundary_entropy": 0.12481, "router_noise": 0.0, "grad_norm": 0.2919, "lr": 0.0002186, "router_trainable": true, "sec_per_step": 0.182, "alpha": 0.03, "t": 5790.3}
{"event": "train", "step": 30000, "phase": 1, "objective": "global", "lm_loss": 0.94973, "bpb": 1.37017, "chunk_loss": 0.030106, "F": 0.17735, "G": 0.20489, "chunks": 181.61, "bytes_per_chunk": 5.638, "boundary_entropy": 0.14147, "router_noise": 0.0, "grad_norm": 0.2834, "lr": 0.0002159, "router_trainable": true, "sec_per_step": 0.181, "alpha": 0.03, "t": 5880.7}
{"event": "train", "step": 30500, "phase": 1, "objective": "global", "lm_loss": 0.96708, "bpb": 1.3952, "chunk_loss": 0.030064, "F": 0.17215, "G": 0.19978, "chunks": 176.28, "bytes_per_chunk": 5.809, "boundary_entropy": 0.13686, "router_noise": 0.0, "grad_norm": 0.2764, "lr": 0.0002133, "router_trainable": true, "sec_per_step": 0.182, "alpha": 0.03, "t": 5971.6}
{"event": "phase_transition", "step": 30520, "phase": 2, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0, "t": 5974.8}
{"event": "train", "step": 31000, "phase": 2, "objective": "global", "lm_loss": 0.95976, "bpb": 1.38464, "chunk_loss": 0.030197, "F": 0.18132, "G": 0.20413, "chunks": 185.67, "bytes_per_chunk": 5.515, "boundary_entropy": 0.12905, "router_noise": 0.0, "grad_norm": 0.2823, "lr": 0.0002106, "router_trainable": true, "sec_per_step": 0.178, "alpha": 0.03, "t": 6060.8}
{"event": "train", "step": 31500, "phase": 2, "objective": "global", "lm_loss": 0.98012, "bpb": 1.41401, "chunk_loss": 0.03008, "F": 0.17639, "G": 0.1905, "chunks": 180.62, "bytes_per_chunk": 5.669, "boundary_entropy": 0.10321, "router_noise": 0.0, "grad_norm": 0.2683, "lr": 0.0002079, "router_trainable": true, "sec_per_step": 0.181, "alpha": 0.03, "t": 6151.2}
{"event": "probe", "step": 32000, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3031.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.174165, "cps": 23.679688, "mean_p": 0.176452, "boundary_entropy": 0.042312, "bpb": 1.558143}, "de": {"chunks": 3652.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.17786, "cps": 28.53125, "mean_p": 0.178267, "boundary_entropy": 0.04532, "bpb": 1.659487}, "vi": {"chunks": 4116.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.223271, "cps": 32.15625, "mean_p": 0.215867, "boundary_entropy": 0.18672, "bpb": 1.338712}, "tr": {"chunks": 3323.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.183561, "cps": 25.960938, "mean_p": 0.218477, "boundary_entropy": 0.191342, "bpb": 1.560506}, "fi": {"chunks": 3571.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.189836, "cps": 27.898438, "mean_p": 0.186094, "boundary_entropy": 0.044588, "bpb": 1.704507}, "ru": {"chunks": 5673.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.299367, "cps": 44.320312, "mean_p": 0.202789, "boundary_entropy": 0.180301, "bpb": 0.968363}, "el": {"chunks": 5941.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.28948, "cps": 46.414062, "mean_p": 0.293485, "boundary_entropy": 0.386828, "bpb": 0.982403}, "hi": {"chunks": 7410.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.440416, "cps": 57.890625, "mean_p": 0.168344, "boundary_entropy": 0.132766, "bpb": 0.801849}, "ta": {"chunks": 8998.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.481305, "cps": 70.296875, "mean_p": 0.168836, "boundary_entropy": 0.164386, "bpb": 0.689508}, "th": {"chunks": 8208.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.510289, "cps": 64.125, "mean_p": 0.175702, "boundary_entropy": 0.204023, "bpb": 0.798158}, "zh": {"chunks": 3681.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.623264, "cps": 28.757812, "mean_p": 0.226101, "boundary_entropy": 0.038635, "bpb": 2.316153}, "my": {"chunks": 7472.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.394342, "cps": 58.375, "mean_p": 0.205761, "boundary_entropy": 0.347737, "bpb": 0.805877}}, "bpb": {"en": 1.55814, "de": 1.65949, "vi": 1.33871, "tr": 1.56051, "fi": 1.70451, "ru": 0.96836, "el": 0.9824, "hi": 0.80185, "ta": 0.68951, "th": 0.79816, "zh": 2.31615, "my": 0.80588}, "gini_cpc": 0.246461, "gini_cps": 0.210057, "bpic": 5.897428, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0212134957072665, "vi": 1.2819479965498903, "tr": 1.0539450626681774, "fi": 1.089974030336427, "ru": 1.7188649391642559, "el": 1.6620990108980094, "hi": 2.528723350538737, "ta": 2.763495127332267, "th": 2.9299112568734613, "zh": 3.5785785845618525, "my": 2.2641837595845673}, "t": 6244.8}
{"event": "eval", "step": 32000, "phase": 2, "bpb": {"en": 1.55422, "de": 1.6689, "vi": 1.35596, "tr": 1.55347, "fi": 1.72194, "ru": 0.96659, "el": 0.98068, "hi": 0.80839, "ta": 0.69437, "th": 0.79744, "zh": 2.32013, "my": 0.81111}, "bytes_per_chunk": {"en": 5.75, "de": 5.716, "vi": 6.057, "tr": 5.98, "fi": 5.353, "ru": 5.984, "el": 6.216, "hi": 5.8, "ta": 5.793, "th": 5.71, "zh": 4.54, "my": 7.174}, "bpb_mean": 1.26943, "t": 6250.3}
{"event": "train", "step": 32000, "phase": 2, "objective": "global", "lm_loss": 0.98402, "bpb": 1.41964, "chunk_loss": 0.030122, "F": 0.17723, "G": 0.19689, "chunks": 181.48, "bytes_per_chunk": 5.642, "boundary_entropy": 0.10497, "router_noise": 0.0, "grad_norm": 0.2816, "lr": 0.0002052, "router_trainable": true, "sec_per_step": 0.199, "alpha": 0.03, "t": 6250.4}
{"event": "train", "step": 32500, "phase": 2, "objective": "global", "lm_loss": 0.9547, "bpb": 1.37734, "chunk_loss": 0.030114, "F": 0.17796, "G": 0.19715, "chunks": 182.23, "bytes_per_chunk": 5.619, "boundary_entropy": 0.11178, "router_noise": 0.0, "grad_norm": 0.291, "lr": 0.0002025, "router_trainable": true, "sec_per_step": 0.18, "alpha": 0.03, "t": 6340.2}
{"event": "train", "step": 33000, "phase": 2, "objective": "global", "lm_loss": 0.93229, "bpb": 1.34501, "chunk_loss": 0.030069, "F": 0.17299, "G": 0.19653, "chunks": 177.14, "bytes_per_chunk": 5.781, "boundary_entropy": 0.12366, "router_noise": 0.0, "grad_norm": 0.2787, "lr": 0.0001997, "router_trainable": true, "sec_per_step": 0.177, "alpha": 0.03, "t": 6428.7}
{"event": "train", "step": 33500, "phase": 2, "objective": "global", "lm_loss": 0.9614, "bpb": 1.38701, "chunk_loss": 0.030087, "F": 0.17467, "G": 0.20234, "chunks": 178.86, "bytes_per_chunk": 5.725, "boundary_entropy": 0.13255, "router_noise": 0.0, "grad_norm": 0.2691, "lr": 0.0001969, "router_trainable": true, "sec_per_step": 0.178, "alpha": 0.03, "t": 6517.8}
{"event": "train", "step": 34000, "phase": 2, "objective": "global", "lm_loss": 0.97796, "bpb": 1.4109, "chunk_loss": 0.030102, "F": 0.17558, "G": 0.19051, "chunks": 179.8, "bytes_per_chunk": 5.695, "boundary_entropy": 0.10053, "router_noise": 0.0, "grad_norm": 0.2853, "lr": 0.0001942, "router_trainable": true, "sec_per_step": 0.179, "alpha": 0.03, "t": 6607.6}
{"event": "train", "step": 34500, "phase": 2, "objective": "global", "lm_loss": 1.00418, "bpb": 1.44873, "chunk_loss": 0.030074, "F": 0.17612, "G": 0.19889, "chunks": 180.34, "bytes_per_chunk": 5.678, "boundary_entropy": 0.11107, "router_noise": 0.0, "grad_norm": 0.288, "lr": 0.0001914, "router_trainable": true, "sec_per_step": 0.178, "alpha": 0.03, "t": 6696.4}
{"event": "train", "step": 35000, "phase": 2, "objective": "global", "lm_loss": 0.95913, "bpb": 1.38373, "chunk_loss": 0.030154, "F": 0.17976, "G": 0.20087, "chunks": 184.08, "bytes_per_chunk": 5.563, "boundary_entropy": 0.11774, "router_noise": 0.0, "grad_norm": 0.2815, "lr": 0.0001886, "router_trainable": true, "sec_per_step": 0.18, "alpha": 0.03, "t": 6786.2}
{"event": "train", "step": 35500, "phase": 2, "objective": "global", "lm_loss": 0.90138, "bpb": 1.30041, "chunk_loss": 0.030047, "F": 0.17139, "G": 0.19778, "chunks": 175.5, "bytes_per_chunk": 5.835, "boundary_entropy": 0.12969, "router_noise": 0.0, "grad_norm": 0.2844, "lr": 0.0001858, "router_trainable": true, "sec_per_step": 0.179, "alpha": 0.03, "t": 6875.6}
{"event": "probe", "step": 36000, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3048.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.175142, "cps": 23.8125, "mean_p": 0.178355, "boundary_entropy": 0.041402, "bpb": 1.540692}, "de": {"chunks": 3652.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.17786, "cps": 28.53125, "mean_p": 0.17944, "boundary_entropy": 0.043563, "bpb": 1.640016}, "vi": {"chunks": 4176.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.226526, "cps": 32.625, "mean_p": 0.225783, "boundary_entropy": 0.199538, "bpb": 1.315892}, "tr": {"chunks": 3283.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.181351, "cps": 25.648438, "mean_p": 0.216337, "boundary_entropy": 0.186738, "bpb": 1.545977}, "fi": {"chunks": 3564.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.189464, "cps": 27.84375, "mean_p": 0.185747, "boundary_entropy": 0.042094, "bpb": 1.681187}, "ru": {"chunks": 5771.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.304538, "cps": 45.085938, "mean_p": 0.203433, "boundary_entropy": 0.174819, "bpb": 0.957651}, "el": {"chunks": 6043.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.29445, "cps": 47.210938, "mean_p": 0.292646, "boundary_entropy": 0.382003, "bpb": 0.971769}, "hi": {"chunks": 7402.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.439941, "cps": 57.828125, "mean_p": 0.169626, "boundary_entropy": 0.127459, "bpb": 0.792414}, "ta": {"chunks": 8906.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.476384, "cps": 69.578125, "mean_p": 0.162892, "boundary_entropy": 0.156862, "bpb": 0.686464}, "th": {"chunks": 7662.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.476344, "cps": 59.859375, "mean_p": 0.162029, "boundary_entropy": 0.192447, "bpb": 0.791127}, "zh": {"chunks": 3688.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.62445, "cps": 28.8125, "mean_p": 0.227729, "boundary_entropy": 0.033588, "bpb": 2.291167}, "my": {"chunks": 8763.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.462476, "cps": 68.460938, "mean_p": 0.231106, "boundary_entropy": 0.372712, "bpb": 0.765794}}, "bpb": {"en": 1.54069, "de": 1.64002, "vi": 1.31589, "tr": 1.54598, "fi": 1.68119, "ru": 0.95765, "el": 0.97177, "hi": 0.79241, "ta": 0.68646, "th": 0.79113, "zh": 2.29117, "my": 0.76579}, "gini_cpc": 0.242195, "gini_cps": 0.214599, "bpic": 5.818566, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.015517751144595, "vi": 1.293381085704401, "tr": 1.0354508497090793, "fi": 1.0817700901256888, "ru": 1.7388055492074046, "el": 1.6812059045718248, "hi": 2.511904739619286, "ta": 2.7199841845792276, "th": 2.719757856270453, "zh": 3.565386594708171, "my": 2.640575522482584}, "t": 6967.1}
{"event": "eval", "step": 36000, "phase": 2, "bpb": {"en": 1.53809, "de": 1.64968, "vi": 1.33587, "tr": 1.53848, "fi": 1.69779, "ru": 0.95486, "el": 0.96987, "hi": 0.79874, "ta": 0.69063, "th": 0.79084, "zh": 2.29636, "my": 0.77211}, "bytes_per_chunk": {"en": 5.722, "de": 5.705, "vi": 5.972, "tr": 6.049, "fi": 5.373, "ru": 5.92, "el": 6.157, "hi": 5.789, "ta": 5.812, "th": 6.069, "zh": 4.55, "my": 6.058}, "bpb_mean": 1.25278, "t": 6973.4}
{"event": "train", "step": 36000, "phase": 2, "objective": "global", "lm_loss": 0.93752, "bpb": 1.35255, "chunk_loss": 0.030058, "F": 0.17323, "G": 0.19172, "chunks": 177.39, "bytes_per_chunk": 5.773, "boundary_entropy": 0.11019, "router_noise": 0.0, "grad_norm": 0.2889, "lr": 0.0001829, "router_trainable": true, "sec_per_step": 0.196, "alpha": 0.03, "t": 6973.5}
{"event": "train", "step": 36500, "phase": 2, "objective": "global", "lm_loss": 0.94803, "bpb": 1.36772, "chunk_loss": 0.030112, "F": 0.17905, "G": 0.19759, "chunks": 183.34, "bytes_per_chunk": 5.585, "boundary_entropy": 0.12043, "router_noise": 0.0, "grad_norm": 0.2911, "lr": 0.0001801, "router_trainable": true, "sec_per_step": 0.181, "alpha": 0.03, "t": 7064.2}
{"event": "train", "step": 37000, "phase": 2, "objective": "global", "lm_loss": 0.9744, "bpb": 1.40576, "chunk_loss": 0.030093, "F": 0.1763, "G": 0.20348, "chunks": 180.53, "bytes_per_chunk": 5.672, "boundary_entropy": 0.12387, "router_noise": 0.0, "grad_norm": 0.3021, "lr": 0.0001773, "router_trainable": true, "sec_per_step": 0.179, "alpha": 0.03, "t": 7153.5}
{"event": "train", "step": 37500, "phase": 2, "objective": "global", "lm_loss": 0.95586, "bpb": 1.37902, "chunk_loss": 0.030071, "F": 0.17378, "G": 0.19108, "chunks": 177.95, "bytes_per_chunk": 5.754, "boundary_entropy": 0.10676, "router_noise": 0.0, "grad_norm": 0.2809, "lr": 0.0001744, "router_trainable": true, "sec_per_step": 0.18, "alpha": 0.03, "t": 7243.4}
{"event": "train", "step": 38000, "phase": 2, "objective": "global", "lm_loss": 0.83505, "bpb": 1.20472, "chunk_loss": 0.030035, "F": 0.17102, "G": 0.19497, "chunks": 175.12, "bytes_per_chunk": 5.847, "boundary_entropy": 0.13075, "router_noise": 0.0, "grad_norm": 0.2693, "lr": 0.0001716, "router_trainable": true, "sec_per_step": 0.18, "alpha": 0.03, "t": 7333.3}
{"event": "train", "step": 38500, "phase": 2, "objective": "global", "lm_loss": 0.97156, "bpb": 1.40166, "chunk_loss": 0.030103, "F": 0.17535, "G": 0.19113, "chunks": 179.56, "bytes_per_chunk": 5.703, "boundary_entropy": 0.09729, "router_noise": 0.0, "grad_norm": 0.2886, "lr": 0.0001687, "router_trainable": true, "sec_per_step": 0.179, "alpha": 0.03, "t": 7422.6}
{"event": "train", "step": 39000, "phase": 2, "objective": "global", "lm_loss": 0.87545, "bpb": 1.263, "chunk_loss": 0.030079, "F": 0.17165, "G": 0.19862, "chunks": 175.77, "bytes_per_chunk": 5.826, "boundary_entropy": 0.13188, "router_noise": 0.0, "grad_norm": 0.2771, "lr": 0.0001659, "router_trainable": true, "sec_per_step": 0.176, "alpha": 0.03, "t": 7510.8}
{"event": "train", "step": 39500, "phase": 2, "objective": "global", "lm_loss": 1.05395, "bpb": 1.52053, "chunk_loss": 0.030198, "F": 0.18311, "G": 0.1974, "chunks": 187.5, "bytes_per_chunk": 5.461, "boundary_entropy": 0.09342, "router_noise": 0.0, "grad_norm": 0.312, "lr": 0.000163, "router_trainable": true, "sec_per_step": 0.177, "alpha": 0.03, "t": 7599.3}
{"event": "probe", "step": 40000, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3040.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.174683, "cps": 23.75, "mean_p": 0.177093, "boundary_entropy": 0.037246, "bpb": 1.525824}, "de": {"chunks": 3646.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.177568, "cps": 28.484375, "mean_p": 0.178527, "boundary_entropy": 0.040823, "bpb": 1.622684}, "vi": {"chunks": 4152.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.225224, "cps": 32.4375, "mean_p": 0.212521, "boundary_entropy": 0.171145, "bpb": 1.301264}, "tr": {"chunks": 3303.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.182456, "cps": 25.804688, "mean_p": 0.215387, "boundary_entropy": 0.18029, "bpb": 1.526973}, "fi": {"chunks": 3564.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.189464, "cps": 27.84375, "mean_p": 0.185442, "boundary_entropy": 0.039636, "bpb": 1.663813}, "ru": {"chunks": 5682.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.299842, "cps": 44.390625, "mean_p": 0.194261, "boundary_entropy": 0.153446, "bpb": 0.945871}, "el": {"chunks": 6109.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.297666, "cps": 47.726562, "mean_p": 0.296972, "boundary_entropy": 0.384799, "bpb": 0.961484}, "hi": {"chunks": 7476.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.444339, "cps": 58.40625, "mean_p": 0.169268, "boundary_entropy": 0.124509, "bpb": 0.781702}, "ta": {"chunks": 8684.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.464509, "cps": 67.84375, "mean_p": 0.156882, "boundary_entropy": 0.155363, "bpb": 0.682251}, "th": {"chunks": 6980.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.433945, "cps": 54.53125, "mean_p": 0.152733, "boundary_entropy": 0.187099, "bpb": 0.788537}, "zh": {"chunks": 3678.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.622757, "cps": 28.734375, "mean_p": 0.226612, "boundary_entropy": 0.033577, "bpb": 2.270727}, "my": {"chunks": 8937.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.471659, "cps": 69.820312, "mean_p": 0.225087, "boundary_entropy": 0.349592, "bpb": 0.760915}}, "bpb": {"en": 1.52582, "de": 1.62268, "vi": 1.30126, "tr": 1.52697, "fi": 1.66381, "ru": 0.94587, "el": 0.96148, "hi": 0.7817, "ta": 0.68225, "th": 0.78854, "zh": 2.27073, "my": 0.76092}, "gini_cpc": 0.23955, "gini_cps": 0.212223, "bpic": 5.881611, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0165173456336014, "vi": 1.2893319343925314, "tr": 1.0445002754704802, "fi": 1.0846168535207563, "ru": 1.7164950354117483, "el": 1.7040401237892273, "hi": 2.543693360444201, "ta": 2.6591625258653453, "th": 2.4841904847602376, "zh": 3.5650762159801808, "my": 2.7000942156094796}, "t": 7690.5}
{"event": "eval", "step": 40000, "phase": 2, "bpb": {"en": 1.52352, "de": 1.63275, "vi": 1.31885, "tr": 1.5198, "fi": 1.68157, "ru": 0.94392, "el": 0.95873, "hi": 0.78853, "ta": 0.68672, "th": 0.78845, "zh": 2.27596, "my": 0.76646}, "bytes_per_chunk": {"en": 5.713, "de": 5.701, "vi": 6.006, "tr": 6.031, "fi": 5.381, "ru": 5.972, "el": 6.038, "hi": 5.768, "ta": 6.039, "th": 6.45, "zh": 4.552, "my": 5.915}, "bpb_mean": 1.24044, "t": 7696.5}
{"event": "train", "step": 40000, "phase": 2, "objective": "global", "lm_loss": 0.98133, "bpb": 1.41576, "chunk_loss": 0.03008, "F": 0.17491, "G": 0.19681, "chunks": 179.11, "bytes_per_chunk": 5.717, "boundary_entropy": 0.11328, "router_noise": 0.0, "grad_norm": 0.3042, "lr": 0.0001601, "router_trainable": true, "sec_per_step": 0.195, "alpha": 0.03, "t": 7696.7}
{"event": "train", "step": 40500, "phase": 2, "objective": "global", "lm_loss": 0.92453, "bpb": 1.33382, "chunk_loss": 0.03005, "F": 0.1711, "G": 0.19342, "chunks": 175.2, "bytes_per_chunk": 5.845, "boundary_entropy": 0.11856, "router_noise": 0.0, "grad_norm": 0.2863, "lr": 0.0001573, "router_trainable": true, "sec_per_step": 0.178, "alpha": 0.03, "t": 7785.8}
{"event": "train", "step": 41000, "phase": 2, "objective": "global", "lm_loss": 0.95428, "bpb": 1.37673, "chunk_loss": 0.03013, "F": 0.17943, "G": 0.19795, "chunks": 183.73, "bytes_per_chunk": 5.573, "boundary_entropy": 0.11814, "router_noise": 0.0, "grad_norm": 0.2972, "lr": 0.0001545, "router_trainable": true, "sec_per_step": 0.178, "alpha": 0.03, "t": 7874.8}
{"event": "train", "step": 41500, "phase": 2, "objective": "global", "lm_loss": 0.90495, "bpb": 1.30557, "chunk_loss": 0.030095, "F": 0.17566, "G": 0.19855, "chunks": 179.88, "bytes_per_chunk": 5.693, "boundary_entropy": 0.11406, "router_noise": 0.0, "grad_norm": 0.2949, "lr": 0.0001516, "router_trainable": true, "sec_per_step": 0.18, "alpha": 0.03, "t": 7964.6}
{"event": "train", "step": 42000, "phase": 2, "objective": "global", "lm_loss": 0.89244, "bpb": 1.28752, "chunk_loss": 0.030083, "F": 0.17206, "G": 0.19478, "chunks": 176.19, "bytes_per_chunk": 5.812, "boundary_entropy": 0.13174, "router_noise": 0.0, "grad_norm": 0.2927, "lr": 0.0001488, "router_trainable": true, "sec_per_step": 0.178, "alpha": 0.03, "t": 8053.8}
{"event": "train", "step": 42500, "phase": 2, "objective": "global", "lm_loss": 0.95286, "bpb": 1.37468, "chunk_loss": 0.03014, "F": 0.17613, "G": 0.1928, "chunks": 180.36, "bytes_per_chunk": 5.678, "boundary_entropy": 0.10693, "router_noise": 0.0, "grad_norm": 0.2971, "lr": 0.0001459, "router_trainable": true, "sec_per_step": 0.185, "alpha": 0.03, "t": 8146.1}
{"event": "train", "step": 43000, "phase": 2, "objective": "global", "lm_loss": 0.92002, "bpb": 1.32731, "chunk_loss": 0.030072, "F": 0.17477, "G": 0.19235, "chunks": 178.97, "bytes_per_chunk": 5.722, "boundary_entropy": 0.1009, "router_noise": 0.0, "grad_norm": 0.3126, "lr": 0.0001431, "router_trainable": true, "sec_per_step": 0.181, "alpha": 0.03, "t": 8236.6}
{"event": "train", "step": 43500, "phase": 2, "objective": "global", "lm_loss": 0.94943, "bpb": 1.36974, "chunk_loss": 0.03017, "F": 0.18208, "G": 0.1943, "chunks": 186.45, "bytes_per_chunk": 5.492, "boundary_entropy": 0.09754, "router_noise": 0.0, "grad_norm": 0.294, "lr": 0.0001403, "router_trainable": true, "sec_per_step": 0.182, "alpha": 0.03, "t": 8327.5}
{"event": "probe", "step": 44000, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3045.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.17497, "cps": 23.789062, "mean_p": 0.176234, "boundary_entropy": 0.035661, "bpb": 1.509885}, "de": {"chunks": 3667.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.178591, "cps": 28.648438, "mean_p": 0.178558, "boundary_entropy": 0.039319, "bpb": 1.607455}, "vi": {"chunks": 4175.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.226471, "cps": 32.617188, "mean_p": 0.215276, "boundary_entropy": 0.175082, "bpb": 1.287952}, "tr": {"chunks": 3363.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.18577, "cps": 26.273438, "mean_p": 0.212182, "boundary_entropy": 0.165572, "bpb": 1.509126}, "fi": {"chunks": 3583.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.190474, "cps": 27.992188, "mean_p": 0.185996, "boundary_entropy": 0.037141, "bpb": 1.639574}, "ru": {"chunks": 5739.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.30285, "cps": 44.835938, "mean_p": 0.192421, "boundary_entropy": 0.141283, "bpb": 0.93159}, "el": {"chunks": 6011.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.292891, "cps": 46.960938, "mean_p": 0.291305, "boundary_entropy": 0.376325, "bpb": 0.94518}, "hi": {"chunks": 7713.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.458425, "cps": 60.257812, "mean_p": 0.174774, "boundary_entropy": 0.121633, "bpb": 0.768895}, "ta": {"chunks": 9149.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.489382, "cps": 71.476562, "mean_p": 0.166778, "boundary_entropy": 0.149243, "bpb": 0.667413}, "th": {"chunks": 8269.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.514081, "cps": 64.601562, "mean_p": 0.170464, "boundary_entropy": 0.182602, "bpb": 0.768846}, "zh": {"chunks": 3688.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.62445, "cps": 28.8125, "mean_p": 0.227844, "boundary_entropy": 0.030051, "bpb": 2.252336}, "my": {"chunks": 9480.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.500317, "cps": 74.0625, "mean_p": 0.206705, "boundary_entropy": 0.295653, "bpb": 0.748436}}, "bpb": {"en": 1.50988, "de": 1.60745, "vi": 1.28795, "tr": 1.50913, "fi": 1.63957, "ru": 0.93159, "el": 0.94518, "hi": 0.7689, "ta": 0.66741, "th": 0.76885, "zh": 2.25234, "my": 0.74844}, "gini_cpc": 0.248991, "gini_cps": 0.226403, "bpic": 5.653649, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0206934457783805, "vi": 1.2943453299503915, "tr": 1.0617276781547516, "fi": 1.0886085675161765, "ru": 1.7308675409750836, "el": 1.673950864691279, "hi": 2.6200228665119787, "ta": 2.7969517942285878, "th": 2.938114726108681, "zh": 3.5688992908605925, "my": 2.859445243450834}, "t": 8421.3}
{"event": "eval", "step": 44000, "phase": 2, "bpb": {"en": 1.50745, "de": 1.61624, "vi": 1.30649, "tr": 1.50119, "fi": 1.65786, "ru": 0.92989, "el": 0.94339, "hi": 0.77614, "ta": 0.6733, "th": 0.76814, "zh": 2.25677, "my": 0.75511}, "bytes_per_chunk": {"en": 5.716, "de": 5.681, "vi": 5.994, "tr": 5.939, "fi": 5.348, "ru": 5.891, "el": 6.141, "hi": 5.588, "ta": 5.705, "th": 5.577, "zh": 4.543, "my": 5.674}, "bpb_mean": 1.22433, "t": 8427.1}
{"event": "train", "step": 44000, "phase": 2, "objective": "global", "lm_loss": 0.87186, "bpb": 1.25783, "chunk_loss": 0.030094, "F": 0.17569, "G": 0.19556, "chunks": 179.91, "bytes_per_chunk": 5.692, "boundary_entropy": 0.11858, "router_noise": 0.0, "grad_norm": 0.2861, "lr": 0.0001375, "router_trainable": true, "sec_per_step": 0.2, "alpha": 0.03, "t": 8427.3}
{"event": "train", "step": 44500, "phase": 2, "objective": "global", "lm_loss": 0.91703, "bpb": 1.32299, "chunk_loss": 0.030121, "F": 0.17656, "G": 0.19239, "chunks": 180.8, "bytes_per_chunk": 5.664, "boundary_entropy": 0.10257, "router_noise": 0.0, "grad_norm": 0.3169, "lr": 0.0001347, "router_trainable": true, "sec_per_step": 0.176, "alpha": 0.03, "t": 8515.1}
{"event": "train", "step": 45000, "phase": 2, "objective": "global", "lm_loss": 0.89005, "bpb": 1.28406, "chunk_loss": 0.030077, "F": 0.1736, "G": 0.19414, "chunks": 177.77, "bytes_per_chunk": 5.76, "boundary_entropy": 0.11675, "router_noise": 0.0, "grad_norm": 0.2932, "lr": 0.0001319, "router_trainable": true, "sec_per_step": 0.179, "alpha": 0.03, "t": 8604.4}
{"event": "train", "step": 45500, "phase": 2, "objective": "global", "lm_loss": 0.87418, "bpb": 1.26118, "chunk_loss": 0.030099, "F": 0.17642, "G": 0.18807, "chunks": 180.66, "bytes_per_chunk": 5.668, "boundary_entropy": 0.09618, "router_noise": 0.0, "grad_norm": 0.2952, "lr": 0.0001292, "router_trainable": true, "sec_per_step": 0.181, "alpha": 0.03, "t": 8695.0}
{"event": "train", "step": 46000, "phase": 2, "objective": "global", "lm_loss": 0.98472, "bpb": 1.42065, "chunk_loss": 0.030144, "F": 0.1805, "G": 0.18999, "chunks": 184.83, "bytes_per_chunk": 5.54, "boundary_entropy": 0.08235, "router_noise": 0.0, "grad_norm": 0.3147, "lr": 0.0001264, "router_trainable": true, "sec_per_step": 0.177, "alpha": 0.03, "t": 8783.4}
{"event": "train", "step": 46500, "phase": 2, "objective": "global", "lm_loss": 0.89622, "bpb": 1.29297, "chunk_loss": 0.030114, "F": 0.17479, "G": 0.18716, "chunks": 178.98, "bytes_per_chunk": 5.721, "boundary_entropy": 0.10009, "router_noise": 0.0, "grad_norm": 0.2947, "lr": 0.0001237, "router_trainable": true, "sec_per_step": 0.179, "alpha": 0.03, "t": 8872.9}
{"event": "train", "step": 47000, "phase": 2, "objective": "global", "lm_loss": 0.82797, "bpb": 1.19451, "chunk_loss": 0.030096, "F": 0.17459, "G": 0.20159, "chunks": 178.78, "bytes_per_chunk": 5.728, "boundary_entropy": 0.13769, "router_noise": 0.0, "grad_norm": 0.2999, "lr": 0.000121, "router_trainable": true, "sec_per_step": 0.179, "alpha": 0.03, "t": 8962.5}
{"event": "train", "step": 47500, "phase": 2, "objective": "global", "lm_loss": 0.96575, "bpb": 1.39329, "chunk_loss": 0.030089, "F": 0.17587, "G": 0.18476, "chunks": 180.09, "bytes_per_chunk": 5.686, "boundary_entropy": 0.07808, "router_noise": 0.0, "grad_norm": 0.3175, "lr": 0.0001183, "router_trainable": true, "sec_per_step": 0.179, "alpha": 0.03, "t": 9051.7}
{"event": "probe", "step": 48000, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3048.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.175142, "cps": 23.8125, "mean_p": 0.17618, "boundary_entropy": 0.034586, "bpb": 1.49791}, "de": {"chunks": 3650.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.177763, "cps": 28.515625, "mean_p": 0.177884, "boundary_entropy": 0.037063, "bpb": 1.592592}, "vi": {"chunks": 4162.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.225766, "cps": 32.515625, "mean_p": 0.211586, "boundary_entropy": 0.164065, "bpb": 1.274219}, "tr": {"chunks": 3339.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.184445, "cps": 26.085938, "mean_p": 0.207739, "boundary_entropy": 0.156114, "bpb": 1.49348}, "fi": {"chunks": 3566.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.18957, "cps": 27.859375, "mean_p": 0.184346, "boundary_entropy": 0.033994, "bpb": 1.620626}, "ru": {"chunks": 5712.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.301425, "cps": 44.625, "mean_p": 0.190156, "boundary_entropy": 0.134779, "bpb": 0.922157}, "el": {"chunks": 6275.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.305755, "cps": 49.023438, "mean_p": 0.309176, "boundary_entropy": 0.396847, "bpb": 0.931396}, "hi": {"chunks": 7763.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.461397, "cps": 60.648438, "mean_p": 0.174754, "boundary_entropy": 0.117266, "bpb": 0.757964}, "ta": {"chunks": 9257.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.495159, "cps": 72.320312, "mean_p": 0.165233, "boundary_entropy": 0.147589, "bpb": 0.656533}, "th": {"chunks": 8563.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.532359, "cps": 66.898438, "mean_p": 0.173512, "boundary_entropy": 0.179801, "bpb": 0.757253}, "zh": {"chunks": 3689.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.624619, "cps": 28.820312, "mean_p": 0.227494, "boundary_entropy": 0.03059, "bpb": 2.236964}, "my": {"chunks": 8710.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.459679, "cps": 68.046875, "mean_p": 0.189971, "boundary_entropy": 0.267827, "bpb": 0.742475}}, "bpb": {"en": 1.49791, "de": 1.59259, "vi": 1.27422, "tr": 1.49348, "fi": 1.62063, "ru": 0.92216, "el": 0.9314, "hi": 0.75796, "ta": 0.65653, "th": 0.75725, "zh": 2.23696, "my": 0.74247}, "gini_cpc": 0.249528, "gini_cps": 0.223288, "bpic": 5.666002, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0149616077978565, "vi": 1.2890450380032847, "tr": 1.0531131243309828, "fi": 1.0823771440483183, "ru": 1.7210288159890306, "el": 1.74574996710048, "hi": 2.634411847293234, "ta": 2.8271832019593433, "th": 3.0395832058527654, "zh": 3.566353348123222, "my": 2.624604907089274}, "t": 9143.5}
{"event": "eval", "step": 48000, "phase": 2, "bpb": {"en": 1.49589, "de": 1.60315, "vi": 1.29308, "tr": 1.48642, "fi": 1.6395, "ru": 0.92024, "el": 0.93004, "hi": 0.76537, "ta": 0.66216, "th": 0.75709, "zh": 2.24234, "my": 0.74802}, "bytes_per_chunk": {"en": 5.696, "de": 5.688, "vi": 5.976, "tr": 5.981, "fi": 5.373, "ru": 5.938, "el": 5.858, "hi": 5.581, "ta": 5.633, "th": 5.456, "zh": 4.546, "my": 6.207}, "bpb_mean": 1.21194, "t": 9148.9}
{"event": "train", "step": 48000, "phase": 2, "objective": "global", "lm_loss": 0.99913, "bpb": 1.44144, "chunk_loss": 0.030095, "F": 0.17857, "G": 0.19172, "chunks": 182.86, "bytes_per_chunk": 5.6, "boundary_entropy": 0.09283, "router_noise": 0.0, "grad_norm": 0.3052, "lr": 0.0001157, "router_trainable": true, "sec_per_step": 0.195, "alpha": 0.03, "t": 9149.1}
{"event": "train", "step": 48500, "phase": 2, "objective": "global", "lm_loss": 0.82735, "bpb": 1.19361, "chunk_loss": 0.030082, "F": 0.17445, "G": 0.18718, "chunks": 178.64, "bytes_per_chunk": 5.732, "boundary_entropy": 0.10183, "router_noise": 0.0, "grad_norm": 0.3014, "lr": 0.000113, "router_trainable": true, "sec_per_step": 0.179, "alpha": 0.03, "t": 9238.4}
{"event": "train", "step": 49000, "phase": 2, "objective": "global", "lm_loss": 0.92927, "bpb": 1.34065, "chunk_loss": 0.030074, "F": 0.17572, "G": 0.18644, "chunks": 179.94, "bytes_per_chunk": 5.691, "boundary_entropy": 0.09597, "router_noise": 0.0, "grad_norm": 0.3056, "lr": 0.0001104, "router_trainable": true, "sec_per_step": 0.183, "alpha": 0.03, "t": 9329.9}
{"event": "train", "step": 49500, "phase": 2, "objective": "global", "lm_loss": 0.84242, "bpb": 1.21535, "chunk_loss": 0.030091, "F": 0.17734, "G": 0.19249, "chunks": 181.59, "bytes_per_chunk": 5.639, "boundary_entropy": 0.11102, "router_noise": 0.0, "grad_norm": 0.289, "lr": 0.0001078, "router_trainable": true, "sec_per_step": 0.178, "alpha": 0.03, "t": 9419.0}
{"event": "train", "step": 50000, "phase": 2, "objective": "global", "lm_loss": 0.95135, "bpb": 1.37251, "chunk_loss": 0.0301, "F": 0.17659, "G": 0.19347, "chunks": 180.83, "bytes_per_chunk": 5.663, "boundary_entropy": 0.09956, "router_noise": 0.0, "grad_norm": 0.3036, "lr": 0.0001052, "router_trainable": true, "sec_per_step": 0.179, "alpha": 0.03, "t": 9508.3}
{"event": "train", "step": 50500, "phase": 2, "objective": "global", "lm_loss": 0.88544, "bpb": 1.27742, "chunk_loss": 0.030066, "F": 0.17371, "G": 0.18932, "chunks": 177.88, "bytes_per_chunk": 5.757, "boundary_entropy": 0.11223, "router_noise": 0.0, "grad_norm": 0.307, "lr": 0.0001027, "router_trainable": true, "sec_per_step": 0.179, "alpha": 0.03, "t": 9597.8}
{"event": "train", "step": 51000, "phase": 2, "objective": "global", "lm_loss": 0.84628, "bpb": 1.22092, "chunk_loss": 0.03015, "F": 0.18016, "G": 0.19327, "chunks": 184.48, "bytes_per_chunk": 5.551, "boundary_entropy": 0.10234, "router_noise": 0.0, "grad_norm": 0.3067, "lr": 0.0001002, "router_trainable": true, "sec_per_step": 0.179, "alpha": 0.03, "t": 9687.2}
{"event": "train", "step": 51500, "phase": 2, "objective": "global", "lm_loss": 0.96963, "bpb": 1.39887, "chunk_loss": 0.030133, "F": 0.17979, "G": 0.19012, "chunks": 184.11, "bytes_per_chunk": 5.562, "boundary_entropy": 0.0761, "router_noise": 0.0, "grad_norm": 0.3185, "lr": 9.77e-05, "router_trainable": true, "sec_per_step": 0.18, "alpha": 0.03, "t": 9777.1}
{"event": "probe", "step": 52000, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3041.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.17474, "cps": 23.757812, "mean_p": 0.176016, "boundary_entropy": 0.03313, "bpb": 1.484994}, "de": {"chunks": 3642.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.177373, "cps": 28.453125, "mean_p": 0.177151, "boundary_entropy": 0.034909, "bpb": 1.580355}, "vi": {"chunks": 4153.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.225278, "cps": 32.445312, "mean_p": 0.208168, "boundary_entropy": 0.155095, "bpb": 1.257901}, "tr": {"chunks": 3392.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.187372, "cps": 26.5, "mean_p": 0.20658, "boundary_entropy": 0.145461, "bpb": 1.479602}, "fi": {"chunks": 3581.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.190367, "cps": 27.976562, "mean_p": 0.185136, "boundary_entropy": 0.033499, "bpb": 1.601697}, "ru": {"chunks": 5736.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.302691, "cps": 44.8125, "mean_p": 0.188117, "boundary_entropy": 0.125014, "bpb": 0.912741}, "el": {"chunks": 6318.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.30785, "cps": 49.359375, "mean_p": 0.302679, "boundary_entropy": 0.38574, "bpb": 0.919682}, "hi": {"chunks": 7430.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.441605, "cps": 58.046875, "mean_p": 0.166336, "boundary_entropy": 0.116278, "bpb": 0.749893}, "ta": {"chunks": 8915.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.476865, "cps": 69.648438, "mean_p": 0.16035, "boundary_entropy": 0.143241, "bpb": 0.649938}, "th": {"chunks": 7731.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.480634, "cps": 60.398438, "mean_p": 0.159632, "boundary_entropy": 0.177365, "bpb": 0.754773}, "zh": {"chunks": 3689.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.624619, "cps": 28.820312, "mean_p": 0.227557, "boundary_entropy": 0.027709, "bpb": 2.218895}, "my": {"chunks": 9200.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.485539, "cps": 71.875, "mean_p": 0.192963, "boundary_entropy": 0.261185, "bpb": 0.727031}}, "bpb": {"en": 1.48499, "de": 1.58035, "vi": 1.2579, "tr": 1.4796, "fi": 1.6017, "ru": 0.91274, "el": 0.91968, "hi": 0.74989, "ta": 0.64994, "th": 0.75477, "zh": 2.2189, "my": 0.72703}, "gini_cpc": 0.242573, "gini_cps": 0.218117, "bpic": 5.742817, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.01506822784756, "vi": 1.2892183821379044, "tr": 1.072291818112742, "fi": 1.0894320249031841, "ru": 1.732238266991399, "el": 1.7617589094825044, "hi": 2.527210637386134, "ta": 2.7290002663958086, "th": 2.7505675057194203, "zh": 3.5745626455375143, "my": 2.7786391555226135}, "t": 9871.6}
{"event": "eval", "step": 52000, "phase": 2, "bpb": {"en": 1.48303, "de": 1.58923, "vi": 1.2762, "tr": 1.4714, "fi": 1.62056, "ru": 0.91001, "el": 0.91875, "hi": 0.75735, "ta": 0.65558, "th": 0.75451, "zh": 2.22293, "my": 0.73354}, "bytes_per_chunk": {"en": 5.705, "de": 5.698, "vi": 6.002, "tr": 5.905, "fi": 5.361, "ru": 5.915, "el": 5.878, "hi": 5.794, "ta": 5.85, "th": 5.961, "zh": 4.549, "my": 5.861}, "bpb_mean": 1.19942, "t": 9876.9}
{"event": "train", "step": 52000, "phase": 2, "objective": "global", "lm_loss": 0.96462, "bpb": 1.39165, "chunk_loss": 0.030168, "F": 0.18114, "G": 0.18721, "chunks": 185.48, "bytes_per_chunk": 5.521, "boundary_entropy": 0.07794, "router_noise": 0.0, "grad_norm": 0.3127, "lr": 9.52e-05, "router_trainable": true, "sec_per_step": 0.2, "alpha": 0.03, "t": 9877.1}
{"event": "train", "step": 52500, "phase": 2, "objective": "global", "lm_loss": 0.89511, "bpb": 1.29137, "chunk_loss": 0.030148, "F": 0.1795, "G": 0.19168, "chunks": 183.81, "bytes_per_chunk": 5.571, "boundary_entropy": 0.0937, "router_noise": 0.0, "grad_norm": 0.3112, "lr": 9.28e-05, "router_trainable": true, "sec_per_step": 0.179, "alpha": 0.03, "t": 9966.6}
{"event": "train", "step": 53000, "phase": 2, "objective": "global", "lm_loss": 0.88522, "bpb": 1.27711, "chunk_loss": 0.030132, "F": 0.17661, "G": 0.1947, "chunks": 180.84, "bytes_per_chunk": 5.662, "boundary_entropy": 0.10438, "router_noise": 0.0, "grad_norm": 0.3234, "lr": 9.04e-05, "router_trainable": true, "sec_per_step": 0.168, "alpha": 0.03, "t": 10050.7}
{"event": "phase_transition", "step": 53410, "phase": 3, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0, "t": 10119.4}
{"event": "train", "step": 53500, "phase": 3, "objective": "global", "lm_loss": 0.91063, "bpb": 1.31376, "chunk_loss": 0.030056, "F": 0.17372, "G": 0.18927, "chunks": 177.89, "bytes_per_chunk": 5.756, "boundary_entropy": 0.09038, "router_noise": 0.0, "grad_norm": 0.3201, "lr": 8.8e-05, "router_trainable": true, "sec_per_step": 0.168, "alpha": 0.03, "t": 10134.6}
{"event": "train", "step": 54000, "phase": 3, "objective": "global", "lm_loss": 0.95815, "bpb": 1.38233, "chunk_loss": 0.03009, "F": 0.17784, "G": 0.19596, "chunks": 182.11, "bytes_per_chunk": 5.623, "boundary_entropy": 0.0937, "router_noise": 0.0, "grad_norm": 0.3269, "lr": 8.57e-05, "router_trainable": true, "sec_per_step": 0.174, "alpha": 0.03, "t": 10221.8}
{"event": "train", "step": 54500, "phase": 3, "objective": "global", "lm_loss": 0.95754, "bpb": 1.38144, "chunk_loss": 0.030167, "F": 0.18031, "G": 0.18983, "chunks": 184.64, "bytes_per_chunk": 5.546, "boundary_entropy": 0.0758, "router_noise": 0.0, "grad_norm": 0.3266, "lr": 8.34e-05, "router_trainable": true, "sec_per_step": 0.183, "alpha": 0.03, "t": 10313.2}
{"event": "train", "step": 55000, "phase": 3, "objective": "global", "lm_loss": 0.8642, "bpb": 1.24678, "chunk_loss": 0.030142, "F": 0.17726, "G": 0.19188, "chunks": 181.52, "bytes_per_chunk": 5.641, "boundary_entropy": 0.09267, "router_noise": 0.0, "grad_norm": 0.3303, "lr": 8.11e-05, "router_trainable": true, "sec_per_step": 0.182, "alpha": 0.03, "t": 10404.0}
{"event": "train", "step": 55500, "phase": 3, "objective": "global", "lm_loss": 0.88314, "bpb": 1.2741, "chunk_loss": 0.030116, "F": 0.17744, "G": 0.19862, "chunks": 181.7, "bytes_per_chunk": 5.636, "boundary_entropy": 0.11484, "router_noise": 0.0, "grad_norm": 0.3267, "lr": 7.89e-05, "router_trainable": true, "sec_per_step": 0.182, "alpha": 0.03, "t": 10494.9}
{"event": "probe", "step": 56000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3046.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.175027, "cps": 23.796875, "mean_p": 0.17552, "boundary_entropy": 0.031064, "bpb": 1.474322}, "de": {"chunks": 3656.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.178055, "cps": 28.5625, "mean_p": 0.177202, "boundary_entropy": 0.03373, "bpb": 1.569774}, "vi": {"chunks": 4193.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.227448, "cps": 32.757812, "mean_p": 0.203574, "boundary_entropy": 0.138491, "bpb": 1.24687}, "tr": {"chunks": 3327.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.183782, "cps": 25.992188, "mean_p": 0.200996, "boundary_entropy": 0.135835, "bpb": 1.465572}, "fi": {"chunks": 3587.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.190686, "cps": 28.023438, "mean_p": 0.184936, "boundary_entropy": 0.031802, "bpb": 1.584179}, "ru": {"chunks": 5776.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.304802, "cps": 45.125, "mean_p": 0.185367, "boundary_entropy": 0.113155, "bpb": 0.901901}, "el": {"chunks": 6130.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.298689, "cps": 47.890625, "mean_p": 0.299285, "boundary_entropy": 0.382776, "bpb": 0.911326}, "hi": {"chunks": 7469.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.443923, "cps": 58.351562, "mean_p": 0.16677, "boundary_entropy": 0.114606, "bpb": 0.741394}, "ta": {"chunks": 8946.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.478524, "cps": 69.890625, "mean_p": 0.158944, "boundary_entropy": 0.13912, "bpb": 0.641739}, "th": {"chunks": 7756.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.482188, "cps": 60.59375, "mean_p": 0.157057, "boundary_entropy": 0.169384, "bpb": 0.742284}, "zh": {"chunks": 3687.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.62428, "cps": 28.804688, "mean_p": 0.227498, "boundary_entropy": 0.026171, "bpb": 2.206911}, "my": {"chunks": 9020.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.47604, "cps": 70.46875, "mean_p": 0.197036, "boundary_entropy": 0.275973, "bpb": 0.717512}}, "bpb": {"en": 1.47432, "de": 1.56977, "vi": 1.24687, "tr": 1.46557, "fi": 1.58418, "ru": 0.9019, "el": 0.91133, "hi": 0.74139, "ta": 0.64174, "th": 0.74228, "zh": 2.20691, "my": 0.71751}, "gini_cpc": 0.242497, "gini_cps": 0.217046, "bpic": 5.763083, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.017297555919384, "vi": 1.2994989761011921, "tr": 1.0500173389529361, "fi": 1.0894660829937972, "ru": 1.7414547388590425, "el": 1.7065296981632223, "hi": 2.5363057584594415, "ta": 2.73399455649902, "th": 2.7549324613515975, "zh": 3.5667602351482035, "my": 2.719802587889224}, "t": 10588.5}
{"event": "eval", "step": 56000, "phase": 3, "bpb": {"en": 1.47183, "de": 1.5783, "vi": 1.26495, "tr": 1.45741, "fi": 1.60414, "ru": 0.90008, "el": 0.90947, "hi": 0.7481, "ta": 0.64721, "th": 0.74278, "zh": 2.21153, "my": 0.72387}, "bytes_per_chunk": {"en": 5.696, "de": 5.687, "vi": 5.949, "tr": 5.958, "fi": 5.357, "ru": 5.885, "el": 6.054, "hi": 5.79, "ta": 5.859, "th": 5.939, "zh": 4.549, "my": 6.071}, "bpb_mean": 1.1883, "t": 10594.5}
{"event": "train", "step": 56000, "phase": 3, "objective": "global", "lm_loss": 0.87468, "bpb": 1.26189, "chunk_loss": 0.030075, "F": 0.17578, "G": 0.18773, "chunks": 180.0, "bytes_per_chunk": 5.689, "boundary_entropy": 0.08213, "router_noise": 0.0, "grad_norm": 0.3102, "lr": 7.68e-05, "router_trainable": true, "sec_per_step": 0.2, "alpha": 0.03, "t": 10594.7}
{"event": "train", "step": 56500, "phase": 3, "objective": "global", "lm_loss": 0.90451, "bpb": 1.30493, "chunk_loss": 0.030096, "F": 0.17743, "G": 0.18816, "chunks": 181.69, "bytes_per_chunk": 5.636, "boundary_entropy": 0.0818, "router_noise": 0.0, "grad_norm": 0.3385, "lr": 7.46e-05, "router_trainable": true, "sec_per_step": 0.178, "alpha": 0.03, "t": 10683.9}
{"event": "train", "step": 57000, "phase": 3, "objective": "global", "lm_loss": 0.86133, "bpb": 1.24264, "chunk_loss": 0.030137, "F": 0.17735, "G": 0.19087, "chunks": 181.61, "bytes_per_chunk": 5.638, "boundary_entropy": 0.09771, "router_noise": 0.0, "grad_norm": 0.3187, "lr": 7.25e-05, "router_trainable": true, "sec_per_step": 0.18, "alpha": 0.03, "t": 10774.0}
{"event": "train", "step": 57500, "phase": 3, "objective": "global", "lm_loss": 0.84797, "bpb": 1.22336, "chunk_loss": 0.030059, "F": 0.17491, "G": 0.18857, "chunks": 179.11, "bytes_per_chunk": 5.717, "boundary_entropy": 0.10364, "router_noise": 0.0, "grad_norm": 0.3295, "lr": 7.05e-05, "router_trainable": true, "sec_per_step": 0.18, "alpha": 0.03, "t": 10863.9}
{"event": "train", "step": 58000, "phase": 3, "objective": "global", "lm_loss": 0.85014, "bpb": 1.22649, "chunk_loss": 0.030066, "F": 0.17369, "G": 0.185, "chunks": 177.86, "bytes_per_chunk": 5.757, "boundary_entropy": 0.08978, "router_noise": 0.0, "grad_norm": 0.3143, "lr": 6.84e-05, "router_trainable": true, "sec_per_step": 0.178, "alpha": 0.03, "t": 10952.9}
{"event": "train", "step": 58500, "phase": 3, "objective": "global", "lm_loss": 0.93423, "bpb": 1.34781, "chunk_loss": 0.030156, "F": 0.18102, "G": 0.18972, "chunks": 185.36, "bytes_per_chunk": 5.524, "boundary_entropy": 0.08661, "router_noise": 0.0, "grad_norm": 0.3317, "lr": 6.65e-05, "router_trainable": true, "sec_per_step": 0.178, "alpha": 0.03, "t": 11041.8}
{"event": "train", "step": 59000, "phase": 3, "objective": "global", "lm_loss": 0.93939, "bpb": 1.35526, "chunk_loss": 0.030067, "F": 0.17387, "G": 0.18591, "chunks": 178.05, "bytes_per_chunk": 5.751, "boundary_entropy": 0.07468, "router_noise": 0.0, "grad_norm": 0.3251, "lr": 6.45e-05, "router_trainable": true, "sec_per_step": 0.177, "alpha": 0.03, "t": 11130.6}
{"event": "train", "step": 59500, "phase": 3, "objective": "global", "lm_loss": 0.92235, "bpb": 1.33068, "chunk_loss": 0.030076, "F": 0.17471, "G": 0.18221, "chunks": 178.91, "bytes_per_chunk": 5.724, "boundary_entropy": 0.07143, "router_noise": 0.0, "grad_norm": 0.3428, "lr": 6.27e-05, "router_trainable": true, "sec_per_step": 0.178, "alpha": 0.03, "t": 11219.4}
{"event": "probe", "step": 60000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3052.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.175372, "cps": 23.84375, "mean_p": 0.175689, "boundary_entropy": 0.030948, "bpb": 1.464933}, "de": {"chunks": 3663.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.178396, "cps": 28.617188, "mean_p": 0.177214, "boundary_entropy": 0.032909, "bpb": 1.557465}, "vi": {"chunks": 4168.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.226092, "cps": 32.5625, "mean_p": 0.199698, "boundary_entropy": 0.131197, "bpb": 1.236783}, "tr": {"chunks": 3325.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.183671, "cps": 25.976562, "mean_p": 0.19683, "boundary_entropy": 0.12547, "bpb": 1.456257}, "fi": {"chunks": 3582.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.19042, "cps": 27.984375, "mean_p": 0.184323, "boundary_entropy": 0.029887, "bpb": 1.569726}, "ru": {"chunks": 5752.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.303536, "cps": 44.9375, "mean_p": 0.181573, "boundary_entropy": 0.104516, "bpb": 0.893145}, "el": {"chunks": 6084.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.296448, "cps": 47.53125, "mean_p": 0.293004, "boundary_entropy": 0.372431, "bpb": 0.902171}, "hi": {"chunks": 7606.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.452065, "cps": 59.421875, "mean_p": 0.168199, "boundary_entropy": 0.112384, "bpb": 0.733318}, "ta": {"chunks": 8992.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.480984, "cps": 70.25, "mean_p": 0.159733, "boundary_entropy": 0.134241, "bpb": 0.635095}, "th": {"chunks": 7797.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.484737, "cps": 60.914062, "mean_p": 0.158236, "boundary_entropy": 0.163823, "bpb": 0.735165}, "zh": {"chunks": 3681.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.623264, "cps": 28.757812, "mean_p": 0.226893, "boundary_entropy": 0.026334, "bpb": 2.192479}, "my": {"chunks": 8832.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.466118, "cps": 69.0, "mean_p": 0.186517, "boundary_entropy": 0.25537, "bpb": 0.7101}}, "bpb": {"en": 1.46493, "de": 1.55746, "vi": 1.23678, "tr": 1.45626, "fi": 1.56973, "ru": 0.89314, "el": 0.90217, "hi": 0.73332, "ta": 0.63509, "th": 0.73516, "zh": 2.19248, "my": 0.7101}, "gini_cpc": 0.242721, "gini_cps": 0.216576, "bpic": 5.768194, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0172415768523757, "vi": 1.28921146559713, "tr": 1.0473231158908731, "fi": 1.0858086297210585, "ru": 1.7308094350518886, "el": 1.6903940335073988, "hi": 2.5777502585204646, "ta": 2.742650192421141, "th": 2.7640510500612936, "zh": 3.553955337420372, "my": 2.6578794249890225}, "t": 11311.5}
{"event": "eval", "step": 60000, "phase": 3, "bpb": {"en": 1.46221, "de": 1.56624, "vi": 1.25549, "tr": 1.44841, "fi": 1.59035, "ru": 0.89079, "el": 0.90015, "hi": 0.74011, "ta": 0.64031, "th": 0.73565, "zh": 2.19681, "my": 0.71661}, "bytes_per_chunk": {"en": 5.693, "de": 5.676, "vi": 5.985, "tr": 5.969, "fi": 5.363, "ru": 5.914, "el": 6.043, "hi": 5.683, "ta": 5.82, "th": 5.937, "zh": 4.549, "my": 6.162}, "bpb_mean": 1.1786, "t": 11317.6}
{"event": "train", "step": 60000, "phase": 3, "objective": "global", "lm_loss": 0.87248, "bpb": 1.25873, "chunk_loss": 0.030093, "F": 0.17473, "G": 0.18966, "chunks": 178.92, "bytes_per_chunk": 5.723, "boundary_entropy": 0.09477, "router_noise": 0.0, "grad_norm": 0.333, "lr": 6.08e-05, "router_trainable": true, "sec_per_step": 0.197, "alpha": 0.03, "t": 11317.8}
{"event": "train", "step": 60500, "phase": 3, "objective": "global", "lm_loss": 0.88525, "bpb": 1.27715, "chunk_loss": 0.030049, "F": 0.17223, "G": 0.18336, "chunks": 176.36, "bytes_per_chunk": 5.806, "boundary_entropy": 0.08871, "router_noise": 0.0, "grad_norm": 0.3341, "lr": 5.9e-05, "router_trainable": true, "sec_per_step": 0.26, "alpha": 0.03, "t": 11447.9}
{"event": "train", "step": 61000, "phase": 3, "objective": "global", "lm_loss": 0.83849, "bpb": 1.20968, "chunk_loss": 0.030079, "F": 0.1745, "G": 0.18665, "chunks": 178.69, "bytes_per_chunk": 5.731, "boundary_entropy": 0.0969, "router_noise": 0.0, "grad_norm": 0.3277, "lr": 5.73e-05, "router_trainable": true, "sec_per_step": 0.274, "alpha": 0.03, "t": 11585.1}
{"event": "train", "step": 61500, "phase": 3, "objective": "global", "lm_loss": 0.8515, "bpb": 1.22845, "chunk_loss": 0.030084, "F": 0.17651, "G": 0.19027, "chunks": 180.75, "bytes_per_chunk": 5.665, "boundary_entropy": 0.11194, "router_noise": 0.0, "grad_norm": 0.3365, "lr": 5.56e-05, "router_trainable": true, "sec_per_step": 0.256, "alpha": 0.03, "t": 11713.0}
{"event": "train", "step": 62000, "phase": 3, "objective": "global", "lm_loss": 0.8804, "bpb": 1.27015, "chunk_loss": 0.030082, "F": 0.17549, "G": 0.1875, "chunks": 179.7, "bytes_per_chunk": 5.698, "boundary_entropy": 0.07958, "router_noise": 0.0, "grad_norm": 0.3341, "lr": 5.39e-05, "router_trainable": true, "sec_per_step": 0.266, "alpha": 0.03, "t": 11846.1}
{"event": "train", "step": 62500, "phase": 3, "objective": "global", "lm_loss": 0.92155, "bpb": 1.32951, "chunk_loss": 0.030104, "F": 0.17725, "G": 0.18841, "chunks": 181.5, "bytes_per_chunk": 5.642, "boundary_entropy": 0.08498, "router_noise": 0.0, "grad_norm": 0.3572, "lr": 5.23e-05, "router_trainable": true, "sec_per_step": 0.238, "alpha": 0.03, "t": 11965.0}
{"event": "train", "step": 63000, "phase": 3, "objective": "global", "lm_loss": 0.96123, "bpb": 1.38676, "chunk_loss": 0.030095, "F": 0.17722, "G": 0.19265, "chunks": 181.47, "bytes_per_chunk": 5.643, "boundary_entropy": 0.08281, "router_noise": 0.0, "grad_norm": 0.3456, "lr": 5.08e-05, "router_trainable": true, "sec_per_step": 0.269, "alpha": 0.03, "t": 12099.3}
{"event": "train", "step": 63500, "phase": 3, "objective": "global", "lm_loss": 0.90391, "bpb": 1.30407, "chunk_loss": 0.030108, "F": 0.17676, "G": 0.18514, "chunks": 181.0, "bytes_per_chunk": 5.657, "boundary_entropy": 0.08282, "router_noise": 0.0, "grad_norm": 0.3343, "lr": 4.93e-05, "router_trainable": true, "sec_per_step": 0.273, "alpha": 0.03, "t": 12235.6}
{"event": "probe", "step": 64000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3039.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.174625, "cps": 23.742188, "mean_p": 0.174847, "boundary_entropy": 0.029979, "bpb": 1.456562}, "de": {"chunks": 3643.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.177422, "cps": 28.460938, "mean_p": 0.176255, "boundary_entropy": 0.031924, "bpb": 1.546812}, "vi": {"chunks": 4165.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.225929, "cps": 32.539062, "mean_p": 0.200836, "boundary_entropy": 0.131817, "bpb": 1.227022}, "tr": {"chunks": 3327.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.183782, "cps": 25.992188, "mean_p": 0.197889, "boundary_entropy": 0.128847, "bpb": 1.445632}, "fi": {"chunks": 3577.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.190155, "cps": 27.945312, "mean_p": 0.184354, "boundary_entropy": 0.02974, "bpb": 1.556182}, "ru": {"chunks": 5723.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.302005, "cps": 44.710938, "mean_p": 0.180516, "boundary_entropy": 0.101485, "bpb": 0.885326}, "el": {"chunks": 6042.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.294401, "cps": 47.203125, "mean_p": 0.290505, "boundary_entropy": 0.370488, "bpb": 0.893}, "hi": {"chunks": 7642.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.454205, "cps": 59.703125, "mean_p": 0.170635, "boundary_entropy": 0.107892, "bpb": 0.72407}, "ta": {"chunks": 9049.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.484033, "cps": 70.695312, "mean_p": 0.158597, "boundary_entropy": 0.132996, "bpb": 0.630275}, "th": {"chunks": 8140.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.506062, "cps": 63.59375, "mean_p": 0.161113, "boundary_entropy": 0.161093, "bpb": 0.726319}, "zh": {"chunks": 3685.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.623942, "cps": 28.789062, "mean_p": 0.227088, "boundary_entropy": 0.025259, "bpb": 2.178969}, "my": {"chunks": 9578.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.505489, "cps": 74.828125, "mean_p": 0.196191, "boundary_entropy": 0.253982, "bpb": 0.699915}}, "bpb": {"en": 1.45656, "de": 1.54681, "vi": 1.22702, "tr": 1.44563, "fi": 1.55618, "ru": 0.88533, "el": 0.893, "hi": 0.72407, "ta": 0.63028, "th": 0.72632, "zh": 2.17897, "my": 0.69991}, "gini_cpc": 0.248946, "gini_cps": 0.226781, "bpic": 5.676394, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0160151508209476, "vi": 1.2937944502856948, "tr": 1.0524359376277208, "fi": 1.0889312889168261, "ru": 1.7294497651897365, "el": 1.6859057656112981, "hi": 2.601030115189021, "ta": 2.7718424324382123, "th": 2.897989180749725, "zh": 3.5730366395009145, "my": 2.894708769113202}, "t": 12375.4}
{"event": "eval", "step": 64000, "phase": 3, "bpb": {"en": 1.45335, "de": 1.55642, "vi": 1.24548, "tr": 1.43731, "fi": 1.57672, "ru": 0.88349, "el": 0.89159, "hi": 0.73177, "ta": 0.63522, "th": 0.72727, "zh": 2.18438, "my": 0.70663}, "bytes_per_chunk": {"en": 5.708, "de": 5.695, "vi": 5.972, "tr": 5.964, "fi": 5.36, "ru": 5.934, "el": 6.109, "hi": 5.671, "ta": 5.796, "th": 5.706, "zh": 4.548, "my": 5.699}, "bpb_mean": 1.16914, "t": 12383.6}
{"event": "train", "step": 64000, "phase": 3, "objective": "global", "lm_loss": 0.87784, "bpb": 1.26645, "chunk_loss": 0.030144, "F": 0.18025, "G": 0.18967, "chunks": 184.58, "bytes_per_chunk": 5.548, "boundary_entropy": 0.08407, "router_noise": 0.0, "grad_norm": 0.3347, "lr": 4.78e-05, "router_trainable": true, "sec_per_step": 0.297, "alpha": 0.03, "t": 12383.9}
{"event": "train", "step": 64500, "phase": 3, "objective": "global", "lm_loss": 0.90963, "bpb": 1.31231, "chunk_loss": 0.030125, "F": 0.17998, "G": 0.18786, "chunks": 184.3, "bytes_per_chunk": 5.556, "boundary_entropy": 0.07778, "router_noise": 0.0, "grad_norm": 0.3485, "lr": 4.65e-05, "router_trainable": true, "sec_per_step": 0.271, "alpha": 0.03, "t": 12519.5}
{"event": "train", "step": 65000, "phase": 3, "objective": "global", "lm_loss": 0.85072, "bpb": 1.22733, "chunk_loss": 0.030059, "F": 0.17657, "G": 0.18558, "chunks": 180.81, "bytes_per_chunk": 5.663, "boundary_entropy": 0.08739, "router_noise": 0.0, "grad_norm": 0.3433, "lr": 4.51e-05, "router_trainable": true, "sec_per_step": 0.259, "alpha": 0.03, "t": 12649.2}
{"event": "train", "step": 65500, "phase": 3, "objective": "global", "lm_loss": 0.90922, "bpb": 1.31173, "chunk_loss": 0.030086, "F": 0.17661, "G": 0.18697, "chunks": 180.84, "bytes_per_chunk": 5.662, "boundary_entropy": 0.08743, "router_noise": 0.0, "grad_norm": 0.3585, "lr": 4.38e-05, "router_trainable": true, "sec_per_step": 0.269, "alpha": 0.03, "t": 12783.7}
{"event": "train", "step": 66000, "phase": 3, "objective": "global", "lm_loss": 0.85439, "bpb": 1.23262, "chunk_loss": 0.030094, "F": 0.17722, "G": 0.18993, "chunks": 181.47, "bytes_per_chunk": 5.643, "boundary_entropy": 0.09153, "router_noise": 0.0, "grad_norm": 0.3551, "lr": 4.26e-05, "router_trainable": true, "sec_per_step": 0.276, "alpha": 0.03, "t": 12921.4}
{"event": "train", "step": 66500, "phase": 3, "objective": "global", "lm_loss": 0.90222, "bpb": 1.30163, "chunk_loss": 0.030086, "F": 0.17767, "G": 0.18791, "chunks": 181.94, "bytes_per_chunk": 5.628, "boundary_entropy": 0.08237, "router_noise": 0.0, "grad_norm": 0.3537, "lr": 4.14e-05, "router_trainable": true, "sec_per_step": 0.273, "alpha": 0.03, "t": 13057.8}
{"event": "train", "step": 67000, "phase": 3, "objective": "global", "lm_loss": 0.88339, "bpb": 1.27446, "chunk_loss": 0.03021, "F": 0.18335, "G": 0.18945, "chunks": 187.75, "bytes_per_chunk": 5.454, "boundary_entropy": 0.06973, "router_noise": 0.0, "grad_norm": 0.3495, "lr": 4.03e-05, "router_trainable": true, "sec_per_step": 0.276, "alpha": 0.03, "t": 13196.0}
{"event": "train", "step": 67500, "phase": 3, "objective": "global", "lm_loss": 0.83368, "bpb": 1.20274, "chunk_loss": 0.030042, "F": 0.17221, "G": 0.19227, "chunks": 176.34, "bytes_per_chunk": 5.807, "boundary_entropy": 0.10798, "router_noise": 0.0, "grad_norm": 0.349, "lr": 3.92e-05, "router_trainable": true, "sec_per_step": 0.275, "alpha": 0.03, "t": 13333.3}
{"event": "probe", "step": 68000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3050.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.175257, "cps": 23.828125, "mean_p": 0.175684, "boundary_entropy": 0.029645, "bpb": 1.448588}, "de": {"chunks": 3658.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.178152, "cps": 28.578125, "mean_p": 0.177292, "boundary_entropy": 0.031653, "bpb": 1.538768}, "vi": {"chunks": 4189.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.227231, "cps": 32.726562, "mean_p": 0.200907, "boundary_entropy": 0.128537, "bpb": 1.21977}, "tr": {"chunks": 3358.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.185494, "cps": 26.234375, "mean_p": 0.199941, "boundary_entropy": 0.126857, "bpb": 1.436561}, "fi": {"chunks": 3590.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.190846, "cps": 28.046875, "mean_p": 0.184999, "boundary_entropy": 0.028734, "bpb": 1.545183}, "ru": {"chunks": 5816.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.306913, "cps": 45.4375, "mean_p": 0.183261, "boundary_entropy": 0.10138, "bpb": 0.880193}, "el": {"chunks": 6198.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.302003, "cps": 48.421875, "mean_p": 0.297661, "boundary_entropy": 0.37625, "bpb": 0.885951}, "hi": {"chunks": 7785.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.462704, "cps": 60.820312, "mean_p": 0.17115, "boundary_entropy": 0.107715, "bpb": 0.720451}, "ta": {"chunks": 9059.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.484568, "cps": 70.773438, "mean_p": 0.160803, "boundary_entropy": 0.136414, "bpb": 0.625734}, "th": {"chunks": 8053.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.500653, "cps": 62.914062, "mean_p": 0.159258, "boundary_entropy": 0.160265, "bpb": 0.722044}, "zh": {"chunks": 3679.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.622926, "cps": 28.742188, "mean_p": 0.227035, "boundary_entropy": 0.024391, "bpb": 2.172061}, "my": {"chunks": 9613.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.507336, "cps": 75.101562, "mean_p": 0.196683, "boundary_entropy": 0.256691, "bpb": 0.693152}}, "bpb": {"en": 1.44859, "de": 1.53877, "vi": 1.21977, "tr": 1.43656, "fi": 1.54518, "ru": 0.88019, "el": 0.88595, "hi": 0.72045, "ta": 0.62573, "th": 0.72204, "zh": 2.17206, "my": 0.69315}, "gini_cpc": 0.246799, "gini_cps": 0.226019, "bpic": 5.639857, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.016519173852886, "vi": 1.2965566567514573, "tr": 1.0584111862629417, "fi": 1.0889472587978257, "ru": 1.7512149833470305, "el": 1.723197308417665, "hi": 2.6401452756193216, "ta": 2.7648977240342165, "th": 2.856675530098809, "zh": 3.554353560979943, "my": 2.89480855282966}, "t": 13472.2}
{"event": "eval", "step": 68000, "phase": 3, "bpb": {"en": 1.44533, "de": 1.54807, "vi": 1.23815, "tr": 1.42848, "fi": 1.56649, "ru": 0.87811, "el": 0.88452, "hi": 0.72813, "ta": 0.63061, "th": 0.72326, "zh": 2.17621, "my": 0.70006}, "bytes_per_chunk": {"en": 5.678, "de": 5.674, "vi": 5.942, "tr": 5.898, "fi": 5.347, "ru": 5.85, "el": 5.92, "hi": 5.576, "ta": 5.782, "th": 5.753, "zh": 4.55, "my": 5.712}, "bpb_mean": 1.16229, "t": 13481.7}
{"event": "train", "step": 68000, "phase": 3, "objective": "global", "lm_loss": 0.94288, "bpb": 1.36029, "chunk_loss": 0.030143, "F": 0.17854, "G": 0.19641, "chunks": 182.83, "bytes_per_chunk": 5.601, "boundary_entropy": 0.09415, "router_noise": 0.0, "grad_norm": 0.3886, "lr": 3.82e-05, "router_trainable": true, "sec_per_step": 0.297, "alpha": 0.03, "t": 13481.9}
{"event": "train", "step": 68500, "phase": 3, "objective": "global", "lm_loss": 0.86236, "bpb": 1.24412, "chunk_loss": 0.030093, "F": 0.17798, "G": 0.18634, "chunks": 182.25, "bytes_per_chunk": 5.619, "boundary_entropy": 0.0865, "router_noise": 0.0, "grad_norm": 0.3448, "lr": 3.73e-05, "router_trainable": true, "sec_per_step": 0.264, "alpha": 0.03, "t": 13613.8}
{"event": "train", "step": 69000, "phase": 3, "objective": "global", "lm_loss": 0.79455, "bpb": 1.14629, "chunk_loss": 0.030046, "F": 0.17172, "G": 0.18738, "chunks": 175.84, "bytes_per_chunk": 5.823, "boundary_entropy": 0.10835, "router_noise": 0.0, "grad_norm": 0.3491, "lr": 3.64e-05, "router_trainable": true, "sec_per_step": 0.263, "alpha": 0.03, "t": 13745.2}
{"event": "train", "step": 69500, "phase": 3, "objective": "global", "lm_loss": 0.93684, "bpb": 1.35157, "chunk_loss": 0.030105, "F": 0.17966, "G": 0.19012, "chunks": 183.97, "bytes_per_chunk": 5.566, "boundary_entropy": 0.0727, "router_noise": 0.0, "grad_norm": 0.3662, "lr": 3.55e-05, "router_trainable": true, "sec_per_step": 0.264, "alpha": 0.03, "t": 13877.3}
{"event": "train", "step": 70000, "phase": 3, "objective": "global", "lm_loss": 0.92512, "bpb": 1.33466, "chunk_loss": 0.030108, "F": 0.17778, "G": 0.18933, "chunks": 182.05, "bytes_per_chunk": 5.625, "boundary_entropy": 0.07761, "router_noise": 0.0, "grad_norm": 0.3655, "lr": 3.48e-05, "router_trainable": true, "sec_per_step": 0.272, "alpha": 0.03, "t": 14013.5}
{"event": "train", "step": 70500, "phase": 3, "objective": "global", "lm_loss": 0.96116, "bpb": 1.38666, "chunk_loss": 0.030107, "F": 0.1792, "G": 0.1901, "chunks": 183.5, "bytes_per_chunk": 5.58, "boundary_entropy": 0.07691, "router_noise": 0.0, "grad_norm": 0.3695, "lr": 3.4e-05, "router_trainable": true, "sec_per_step": 0.275, "alpha": 0.03, "t": 14151.0}
{"event": "train", "step": 71000, "phase": 3, "objective": "global", "lm_loss": 0.79353, "bpb": 1.14482, "chunk_loss": 0.03007, "F": 0.1776, "G": 0.18851, "chunks": 181.86, "bytes_per_chunk": 5.631, "boundary_entropy": 0.09094, "router_noise": 0.0, "grad_norm": 0.3427, "lr": 3.34e-05, "router_trainable": true, "sec_per_step": 0.275, "alpha": 0.03, "t": 14288.7}
{"event": "train", "step": 71500, "phase": 3, "objective": "global", "lm_loss": 0.83795, "bpb": 1.20891, "chunk_loss": 0.030126, "F": 0.1815, "G": 0.19042, "chunks": 185.86, "bytes_per_chunk": 5.51, "boundary_entropy": 0.08746, "router_noise": 0.0, "grad_norm": 0.3472, "lr": 3.28e-05, "router_trainable": true, "sec_per_step": 0.275, "alpha": 0.03, "t": 14426.3}
{"event": "probe", "step": 72000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3044.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.174912, "cps": 23.78125, "mean_p": 0.17496, "boundary_entropy": 0.028427, "bpb": 1.443785}, "de": {"chunks": 3649.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.177714, "cps": 28.507812, "mean_p": 0.176353, "boundary_entropy": 0.030457, "bpb": 1.533468}, "vi": {"chunks": 4197.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.227665, "cps": 32.789062, "mean_p": 0.198844, "boundary_entropy": 0.122417, "bpb": 1.213506}, "tr": {"chunks": 3329.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.183892, "cps": 26.007812, "mean_p": 0.194337, "boundary_entropy": 0.116573, "bpb": 1.430164}, "fi": {"chunks": 3590.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.190846, "cps": 28.046875, "mean_p": 0.184658, "boundary_entropy": 0.028066, "bpb": 1.536253}, "ru": {"chunks": 5773.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.304644, "cps": 45.101562, "mean_p": 0.18015, "boundary_entropy": 0.093866, "bpb": 0.875875}, "el": {"chunks": 6135.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.298933, "cps": 47.929688, "mean_p": 0.293069, "boundary_entropy": 0.368889, "bpb": 0.88035}, "hi": {"chunks": 7702.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.457771, "cps": 60.171875, "mean_p": 0.169269, "boundary_entropy": 0.104541, "bpb": 0.716025}, "ta": {"chunks": 9008.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.48184, "cps": 70.375, "mean_p": 0.159302, "boundary_entropy": 0.131104, "bpb": 0.621987}, "th": {"chunks": 8205.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.510103, "cps": 64.101562, "mean_p": 0.160713, "boundary_entropy": 0.155013, "bpb": 0.716405}, "zh": {"chunks": 3683.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.623603, "cps": 28.773438, "mean_p": 0.226984, "boundary_entropy": 0.023829, "bpb": 2.163161}, "my": {"chunks": 9717.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.512825, "cps": 75.914062, "mean_p": 0.194735, "boundary_entropy": 0.246256, "bpb": 0.688796}}, "bpb": {"en": 1.44378, "de": 1.53347, "vi": 1.21351, "tr": 1.43016, "fi": 1.53625, "ru": 0.87588, "el": 0.88035, "hi": 0.71602, "ta": 0.62199, "th": 0.7164, "zh": 2.16316, "my": 0.6888}, "gini_cpc": 0.249031, "gini_cps": 0.227854, "bpic": 5.641184, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0160168916253078, "vi": 1.3015932849265825, "tr": 1.0513388565053685, "fi": 1.0910936725799503, "ru": 1.741693837784612, "el": 1.7090438034875521, "hi": 2.6171457747977653, "ta": 2.754751194832042, "th": 2.916332194644336, "zh": 3.5652316092723804, "my": 2.9318942297849815}, "t": 14568.5}
{"event": "eval", "step": 72000, "phase": 3, "bpb": {"en": 1.44055, "de": 1.54233, "vi": 1.23198, "tr": 1.42191, "fi": 1.55733, "ru": 0.87375, "el": 0.87857, "hi": 0.72319, "ta": 0.62718, "th": 0.71745, "zh": 2.16767, "my": 0.69531}, "bytes_per_chunk": {"en": 5.692, "de": 5.684, "vi": 5.941, "tr": 5.961, "fi": 5.346, "ru": 5.883, "el": 5.992, "hi": 5.631, "ta": 5.772, "th": 5.649, "zh": 4.549, "my": 5.634}, "bpb_mean": 1.15644, "t": 14576.3}
{"event": "train", "step": 72000, "phase": 3, "objective": "global", "lm_loss": 0.84347, "bpb": 1.21687, "chunk_loss": 0.030099, "F": 0.17648, "G": 0.18739, "chunks": 180.72, "bytes_per_chunk": 5.666, "boundary_entropy": 0.08413, "router_noise": 0.0, "grad_norm": 0.3611, "lr": 3.22e-05, "router_trainable": true, "sec_per_step": 0.3, "alpha": 0.03, "t": 14576.5}
{"event": "train", "step": 72500, "phase": 3, "objective": "global", "lm_loss": 0.86711, "bpb": 1.25097, "chunk_loss": 0.030086, "F": 0.17595, "G": 0.18941, "chunks": 180.17, "bytes_per_chunk": 5.683, "boundary_entropy": 0.08672, "router_noise": 0.0, "grad_norm": 0.3592, "lr": 3.17e-05, "router_trainable": true, "sec_per_step": 0.267, "alpha": 0.03, "t": 14710.3}
{"event": "train", "step": 73000, "phase": 3, "objective": "global", "lm_loss": 0.87571, "bpb": 1.26339, "chunk_loss": 0.030113, "F": 0.17934, "G": 0.19226, "chunks": 183.64, "bytes_per_chunk": 5.576, "boundary_entropy": 0.08837, "router_noise": 0.0, "grad_norm": 0.3705, "lr": 3.13e-05, "router_trainable": true, "sec_per_step": 0.269, "alpha": 0.03, "t": 14844.8}
{"event": "train", "step": 73500, "phase": 3, "objective": "global", "lm_loss": 0.90655, "bpb": 1.30788, "chunk_loss": 0.030067, "F": 0.17746, "G": 0.19071, "chunks": 181.72, "bytes_per_chunk": 5.635, "boundary_entropy": 0.09522, "router_noise": 0.0, "grad_norm": 0.3692, "lr": 3.09e-05, "router_trainable": true, "sec_per_step": 0.265, "alpha": 0.03, "t": 14977.3}
{"event": "train", "step": 74000, "phase": 3, "objective": "global", "lm_loss": 0.88538, "bpb": 1.27733, "chunk_loss": 0.030111, "F": 0.1768, "G": 0.18858, "chunks": 181.05, "bytes_per_chunk": 5.656, "boundary_entropy": 0.08891, "router_noise": 0.0, "grad_norm": 0.3603, "lr": 3.06e-05, "router_trainable": true, "sec_per_step": 0.271, "alpha": 0.03, "t": 15113.0}
{"event": "train", "step": 74500, "phase": 3, "objective": "global", "lm_loss": 0.88247, "bpb": 1.27313, "chunk_loss": 0.030104, "F": 0.17824, "G": 0.19157, "chunks": 182.52, "bytes_per_chunk": 5.61, "boundary_entropy": 0.08971, "router_noise": 0.0, "grad_norm": 0.3856, "lr": 3.04e-05, "router_trainable": true, "sec_per_step": 0.273, "alpha": 0.03, "t": 15249.3}
{"event": "train", "step": 75000, "phase": 3, "objective": "global", "lm_loss": 0.89682, "bpb": 1.29384, "chunk_loss": 0.03012, "F": 0.17947, "G": 0.18792, "chunks": 183.78, "bytes_per_chunk": 5.572, "boundary_entropy": 0.07057, "router_noise": 0.0, "grad_norm": 0.3794, "lr": 3.02e-05, "router_trainable": true, "sec_per_step": 0.278, "alpha": 0.03, "t": 15388.1}
{"event": "train", "step": 75500, "phase": 3, "objective": "global", "lm_loss": 0.87104, "bpb": 1.25665, "chunk_loss": 0.030089, "F": 0.17609, "G": 0.1901, "chunks": 180.31, "bytes_per_chunk": 5.679, "boundary_entropy": 0.08685, "router_noise": 0.0, "grad_norm": 0.3805, "lr": 3.01e-05, "router_trainable": true, "sec_per_step": 0.268, "alpha": 0.03, "t": 15522.3}
{"event": "probe", "step": 76000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3045.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.17497, "cps": 23.789062, "mean_p": 0.17517, "boundary_entropy": 0.02768, "bpb": 1.440494}, "de": {"chunks": 3650.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.177763, "cps": 28.515625, "mean_p": 0.176859, "boundary_entropy": 0.030492, "bpb": 1.530367}, "vi": {"chunks": 4199.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.227773, "cps": 32.804688, "mean_p": 0.198733, "boundary_entropy": 0.120014, "bpb": 1.209528}, "tr": {"chunks": 3383.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.186875, "cps": 26.429688, "mean_p": 0.19768, "boundary_entropy": 0.117813, "bpb": 1.425512}, "fi": {"chunks": 3596.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.191165, "cps": 28.09375, "mean_p": 0.185082, "boundary_entropy": 0.027379, "bpb": 1.531532}, "ru": {"chunks": 5777.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.304855, "cps": 45.132812, "mean_p": 0.179217, "boundary_entropy": 0.090629, "bpb": 0.871705}, "el": {"chunks": 6157.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.300005, "cps": 48.101562, "mean_p": 0.2911, "boundary_entropy": 0.36493, "bpb": 0.877879}, "hi": {"chunks": 7764.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.461456, "cps": 60.65625, "mean_p": 0.170589, "boundary_entropy": 0.104943, "bpb": 0.712933}, "ta": {"chunks": 9131.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.488419, "cps": 71.335938, "mean_p": 0.160412, "boundary_entropy": 0.128685, "bpb": 0.61945}, "th": {"chunks": 8058.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.500964, "cps": 62.953125, "mean_p": 0.158428, "boundary_entropy": 0.154052, "bpb": 0.713948}, "zh": {"chunks": 3680.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.623095, "cps": 28.75, "mean_p": 0.227202, "boundary_entropy": 0.023167, "bpb": 2.158996}, "my": {"chunks": 9576.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.505383, "cps": 74.8125, "mean_p": 0.191973, "boundary_entropy": 0.242552, "bpb": 0.686337}}, "bpb": {"en": 1.44049, "de": 1.53037, "vi": 1.20953, "tr": 1.42551, "fi": 1.53153, "ru": 0.87171, "el": 0.87788, "hi": 0.71293, "ta": 0.61945, "th": 0.71395, "zh": 2.159, "my": 0.68634}, "gini_cpc": 0.246793, "gini_cps": 0.22591, "bpic": 5.642511, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.015961569973027, "vi": 1.3017858779548968, "tr": 1.068041848111069, "fi": 1.0925583055507038, "ru": 1.7423282425880915, "el": 1.7146091289143577, "hi": 2.637347016154415, "ta": 2.791448992578559, "th": 2.863142878580693, "zh": 3.5611576438088344, "my": 2.8884016509794503}, "t": 15658.0}
{"event": "eval", "step": 76000, "phase": 3, "bpb": {"en": 1.43712, "de": 1.53883, "vi": 1.22808, "tr": 1.41752, "fi": 1.55222, "ru": 0.87035, "el": 0.8758, "hi": 0.72079, "ta": 0.62469, "th": 0.71525, "zh": 2.16399, "my": 0.69276}, "bytes_per_chunk": {"en": 5.69, "de": 5.675, "vi": 5.929, "tr": 5.886, "fi": 5.331, "ru": 5.879, "el": 5.936, "hi": 5.602, "ta": 5.722, "th": 5.744, "zh": 4.548, "my": 5.726}, "bpb_mean": 1.15312, "t": 15666.1}
{"event": "train", "step": 76000, "phase": 3, "objective": "global", "lm_loss": 0.84726, "bpb": 1.22233, "chunk_loss": 0.030072, "F": 0.1734, "G": 0.18364, "chunks": 177.56, "bytes_per_chunk": 5.767, "boundary_entropy": 0.07858, "router_noise": 0.0, "grad_norm": 0.3849, "lr": 3e-05, "router_trainable": true, "sec_per_step": 0.288, "alpha": 0.03, "t": 15666.4}
{"event": "probe", "step": 76300, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3044.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.174912, "cps": 23.78125, "mean_p": 0.17506, "boundary_entropy": 0.027359, "bpb": 1.440076}, "de": {"chunks": 3643.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.177422, "cps": 28.460938, "mean_p": 0.176249, "boundary_entropy": 0.029548, "bpb": 1.529278}, "vi": {"chunks": 4189.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.227231, "cps": 32.726562, "mean_p": 0.196049, "boundary_entropy": 0.113979, "bpb": 1.209666}, "tr": {"chunks": 3342.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.18461, "cps": 26.109375, "mean_p": 0.194504, "boundary_entropy": 0.113777, "bpb": 1.4253}, "fi": {"chunks": 3592.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.190952, "cps": 28.0625, "mean_p": 0.184938, "boundary_entropy": 0.027237, "bpb": 1.531786}, "ru": {"chunks": 5749.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.303377, "cps": 44.914062, "mean_p": 0.177249, "boundary_entropy": 0.088715, "bpb": 0.87247}, "el": {"chunks": 6064.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.295473, "cps": 47.375, "mean_p": 0.283706, "boundary_entropy": 0.355798, "bpb": 0.877223}, "hi": {"chunks": 7643.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.454264, "cps": 59.710938, "mean_p": 0.167354, "boundary_entropy": 0.105935, "bpb": 0.713276}, "ta": {"chunks": 8827.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.472158, "cps": 68.960938, "mean_p": 0.154979, "boundary_entropy": 0.131186, "bpb": 0.622975}, "th": {"chunks": 7903.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.491327, "cps": 61.742188, "mean_p": 0.156026, "boundary_entropy": 0.155017, "bpb": 0.717192}, "zh": {"chunks": 3681.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.623264, "cps": 28.757812, "mean_p": 0.227057, "boundary_entropy": 0.023256, "bpb": 2.156624}, "my": {"chunks": 9799.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.517152, "cps": 76.554688, "mean_p": 0.196104, "boundary_entropy": 0.245814, "bpb": 0.686836}}, "bpb": {"en": 1.44008, "de": 1.52928, "vi": 1.20967, "tr": 1.4253, "fi": 1.53179, "ru": 0.87247, "el": 0.87722, "hi": 0.71328, "ta": 0.62298, "th": 0.71719, "zh": 2.15662, "my": 0.68684}, "gini_cpc": 0.247874, "gini_cps": 0.225458, "bpic": 5.687667, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.014346269167168, "vi": 1.299112287480928, "tr": 1.0554444152721365, "fi": 1.0917015242081287, "ru": 1.734453122020394, "el": 1.6892651384431159, "hi": 2.597097527496666, "ta": 2.699399289163236, "th": 2.8089912656031912, "zh": 3.5632955616974296, "my": 2.9566359532430826}, "t": 15748.8}
{"event": "eval", "step": 76300, "phase": 3, "bpb": {"en": 1.43662, "de": 1.53851, "vi": 1.22791, "tr": 1.41706, "fi": 1.55279, "ru": 0.87108, "el": 0.87571, "hi": 0.72103, "ta": 0.62781, "th": 0.71793, "zh": 2.16237, "my": 0.69301}, "bytes_per_chunk": {"en": 5.69, "de": 5.686, "vi": 5.954, "tr": 5.939, "fi": 5.341, "ru": 5.915, "el": 6.025, "hi": 5.672, "ta": 5.945, "th": 5.86, "zh": 4.547, "my": 5.586}, "bpb_mean": 1.15349, "t": 15756.9}
{"event": "probe", "step": 76300, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3044.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.174912, "cps": 23.78125, "mean_p": 0.17506, "boundary_entropy": 0.027359, "bpb": 1.440076}, "de": {"chunks": 3643.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.177422, "cps": 28.460938, "mean_p": 0.176249, "boundary_entropy": 0.029548, "bpb": 1.529278}, "vi": {"chunks": 4189.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.227231, "cps": 32.726562, "mean_p": 0.196049, "boundary_entropy": 0.113979, "bpb": 1.209666}, "tr": {"chunks": 3342.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.18461, "cps": 26.109375, "mean_p": 0.194504, "boundary_entropy": 0.113777, "bpb": 1.4253}, "fi": {"chunks": 3592.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.190952, "cps": 28.0625, "mean_p": 0.184938, "boundary_entropy": 0.027237, "bpb": 1.531786}, "ru": {"chunks": 5749.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.303377, "cps": 44.914062, "mean_p": 0.177249, "boundary_entropy": 0.088715, "bpb": 0.87247}, "el": {"chunks": 6064.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.295473, "cps": 47.375, "mean_p": 0.283706, "boundary_entropy": 0.355798, "bpb": 0.877223}, "hi": {"chunks": 7643.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.454264, "cps": 59.710938, "mean_p": 0.167354, "boundary_entropy": 0.105935, "bpb": 0.713276}, "ta": {"chunks": 8827.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.472158, "cps": 68.960938, "mean_p": 0.154979, "boundary_entropy": 0.131186, "bpb": 0.622975}, "th": {"chunks": 7903.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.491327, "cps": 61.742188, "mean_p": 0.156026, "boundary_entropy": 0.155017, "bpb": 0.717192}, "zh": {"chunks": 3681.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.623264, "cps": 28.757812, "mean_p": 0.227057, "boundary_entropy": 0.023256, "bpb": 2.156624}, "my": {"chunks": 9799.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.517152, "cps": 76.554688, "mean_p": 0.196104, "boundary_entropy": 0.245814, "bpb": 0.686836}}, "bpb": {"en": 1.44008, "de": 1.52928, "vi": 1.20967, "tr": 1.4253, "fi": 1.53179, "ru": 0.87247, "el": 0.87722, "hi": 0.71328, "ta": 0.62298, "th": 0.71719, "zh": 2.15662, "my": 0.68684}, "gini_cpc": 0.247874, "gini_cps": 0.225458, "bpic": 5.687667, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.014346269167168, "vi": 1.299112287480928, "tr": 1.0554444152721365, "fi": 1.0917015242081287, "ru": 1.734453122020394, "el": 1.6892651384431159, "hi": 2.597097527496666, "ta": 2.699399289163236, "th": 2.8089912656031912, "zh": 3.5632955616974296, "my": 2.9566359532430826}, "t": 15761.5}
{"event": "reps_dumped", "dirs": ["/local/scratch/sas245/beetle-hnet/scratch/reps/specfull_A_s1/th0.30", "/local/scratch/sas245/beetle-hnet/scratch/reps/specfull_A_s1/th0.40", "/local/scratch/sas245/beetle-hnet/scratch/reps/specfull_A_s1/th0.50", "/local/scratch/sas245/beetle-hnet/scratch/reps/specfull_A_s1/th0.60", "/local/scratch/sas245/beetle-hnet/scratch/reps/specfull_A_s1/th0.70"], "t": 15765.7}
{"event": "final", "run": "specfull_A_s1", "condition": "A", "step": 76300, "params": 82470528, "per_lang": {"en": {"cpc": 0.17491237143021318, "cps": 23.78125, "bpb": 1.4344827618211076, "bytes_per_chunk": 5.6980393861670215}, "de": {"cpc": 0.17742171139141868, "cps": 28.4609375, "bpb": 1.5354745000520915, "bytes_per_chunk": 5.662744166203131}, "vi": {"cpc": 0.22723081095741796, "cps": 32.7265625, "bpb": 1.226269461056695, "bytes_per_chunk": 5.933846789390587}, "tr": {"cpc": 0.1846102855880241, "cps": 26.109375, "bpb": 1.4144888596260154, "bytes_per_chunk": 5.9682966116647975}, "fi": {"cpc": 0.19095210249322206, "cps": 28.0625, "bpb": 1.5563237203400448, "bytes_per_chunk": 5.352335255984891}, "ru": {"cpc": 0.30337730870712404, "cps": 44.9140625, "bpb": 0.8736008536716351, "bytes_per_chunk": 5.916870746763872}, "el": {"cpc": 0.29547337133947277, "cps": 47.375, "bpb": 0.8757178472401416, "bytes_per_chunk": 6.034101568119695}, "hi": {"cpc": 0.45426448736998515, "cps": 59.7109375, "bpb": 0.7227330449265795, "bytes_per_chunk": 5.6788141697400984}, "ta": {"cpc": 0.4721583311045734, "cps": 68.9609375, "bpb": 0.6253251161082508, "bytes_per_chunk": 5.932336101744778}, "th": {"cpc": 0.49132732359341, "cps": 61.7421875, "bpb": 0.7201171780580405, "bytes_per_chunk": 5.851297962099061}, "zh": {"cpc": 0.6232644768032509, "cps": 28.7578125, "bpb": 2.163128118829892, "bytes_per_chunk": 4.561009134406263}, "my": {"cpc": 0.5171522060375765, "cps": 76.5546875, "bpb": 0.6932816340289751, "bytes_per_chunk": 5.569118990461218}}, "gini_cpc": 0.24787354735650236, "gini_cps": 0.22545794060110258, "bpic": 5.687666725947003, "seed": 1, "mean_boundary_entropy": 0.09993767738342285, "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": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0}, {"step": 53410, "phase": 3, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0}], "t": 15781.2}