{"event": "start", "run": "spec_C_s0", "condition": "C", "params": 82470528, "langs": ["en", "de", "vi", "tr", "fi", "ru", "el", "hi", "ta", "th", "zh", "my"], "premium": {"en": 1.0, "de": 1.184, "vi": 1.39, "tr": 1.12, "fi": 1.104, "ru": 1.971, "el": 2.174, "hi": 2.552, "ta": 3.192, "th": 2.762, "zh": 0.915, "my": 3.524}, "schedule": {"tau1": 0.4, "tau2": 0.7, "p1_objective": "global", "p2_objective": "parity_a", "p3_objective": "global", "freeze_in_p3": false, "freeze_scope": "router"}, "corpus": "/local/scratch/sas245/beetle-hnet/scratch/data/tier1_corpus", "premium_kind": "flores_per_sent", "doc_filtered": {"my": 37511}, "weights": {"en": 0.2946, "de": 0.0909, "vi": 0.0576, "tr": 0.0612, "fi": 0.0424, "ru": 0.1476, "el": 0.0522, "hi": 0.042, "ta": 0.0313, "th": 0.0542, "zh": 0.1038, "my": 0.0222}, "t": 2.0}
{"event": "probe_schedule", "n": 21, "steps": [0, 1, 2, 4, 8, 16, 32, 64, 128, 256, 1000, 2000, 3000, 4000, 5000, 6000, 7000, 8000, 9000, 10000, 11000], "t": 2.0}
{"event": "probe", "step": 0, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 8739.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.502155, "cps": 68.273438, "mean_p": 0.50363, "boundary_entropy": 0.687415, "bpb": 8.248358}, "de": {"chunks": 12222.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.595237, "cps": 95.484375, "mean_p": 0.507356, "boundary_entropy": 0.688163, "bpb": 8.204506}, "vi": {"chunks": 6386.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.346406, "cps": 49.890625, "mean_p": 0.491233, "boundary_entropy": 0.688507, "bpb": 8.13873}, "tr": {"chunks": 11403.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.629896, "cps": 89.085938, "mean_p": 0.50659, "boundary_entropy": 0.687851, "bpb": 8.143105}, "fi": {"chunks": 7137.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.379406, "cps": 55.757812, "mean_p": 0.497164, "boundary_entropy": 0.68792, "bpb": 8.234202}, "ru": {"chunks": 10473.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.552665, "cps": 81.820312, "mean_p": 0.496692, "boundary_entropy": 0.690192, "bpb": 8.071624}, "el": {"chunks": 36555.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.781172, "cps": 285.585938, "mean_p": 0.530217, "boundary_entropy": 0.688772, "bpb": 8.142597}, "hi": {"chunks": 42584.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.530996, "cps": 332.6875, "mean_p": 0.521073, "boundary_entropy": 0.69018, "bpb": 7.880816}, "ta": {"chunks": 49758.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.661567, "cps": 388.734375, "mean_p": 0.521091, "boundary_entropy": 0.690409, "bpb": 7.814466}, "th": {"chunks": 45716.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.842151, "cps": 357.15625, "mean_p": 0.525713, "boundary_entropy": 0.689893, "bpb": 8.063302}, "zh": {"chunks": 6218.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.052828, "cps": 48.578125, "mean_p": 0.496822, "boundary_entropy": 0.686552, "bpb": 8.034403}, "my": {"chunks": 53266.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.811167, "cps": 416.140625, "mean_p": 0.521325, "boundary_entropy": 0.690575, "bpb": 8.24483}}, "bpb": {"en": 8.24836, "de": 8.20451, "vi": 8.13873, "tr": 8.1431, "fi": 8.2342, "ru": 8.07162, "el": 8.1426, "hi": 7.88082, "ta": 7.81447, "th": 8.0633, "zh": 8.0344, "my": 8.24483}, "gini_cpc": 0.391191, "gini_cps": 0.409188, "bpic": 1.321301, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1853654183968407, "vi": 0.6898396505624507, "tr": 1.2543852937189204, "fi": 0.755555191784985, "ru": 1.1005867247811874, "el": 3.54705827773647, "hi": 5.040269530092341, "ta": 5.300292378412309, "th": 5.659910185771798, "zh": 2.096619669942964, "my": 5.598208760550946}, "t": 9.5}
{"event": "train", "step": 0, "phase": 1, "objective": "global", "lm_loss": 5.63905, "bpb": 8.13544, "chunk_loss": 0.057711, "F": 0.53552, "G": 0.50248, "chunks": 548.38, "bytes_per_chunk": 1.867, "boundary_entropy": 0.69172, "router_noise": 0.0, "grad_norm": 9.2841, "lr": 6e-07, "router_trainable": true, "sec_per_step": 0.005, "alpha": 0.03, "t": 10.2}
{"event": "probe", "step": 1, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 8722.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.501178, "cps": 68.140625, "mean_p": 0.503564, "boundary_entropy": 0.687414, "bpb": 8.209942}, "de": {"chunks": 12185.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.593435, "cps": 95.195312, "mean_p": 0.507261, "boundary_entropy": 0.688164, "bpb": 8.167595}, "vi": {"chunks": 6366.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.345321, "cps": 49.734375, "mean_p": 0.491184, "boundary_entropy": 0.688504, "bpb": 8.11884}, "tr": {"chunks": 11366.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.627852, "cps": 88.796875, "mean_p": 0.506498, "boundary_entropy": 0.687851, "bpb": 8.116512}, "fi": {"chunks": 7085.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.376641, "cps": 55.351562, "mean_p": 0.497024, "boundary_entropy": 0.687916, "bpb": 8.205258}, "ru": {"chunks": 10505.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.554354, "cps": 82.070312, "mean_p": 0.496729, "boundary_entropy": 0.690193, "bpb": 8.028025}, "el": {"chunks": 36543.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.780588, "cps": 285.492188, "mean_p": 0.53005, "boundary_entropy": 0.688791, "bpb": 8.124589}, "hi": {"chunks": 42562.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.529688, "cps": 332.515625, "mean_p": 0.520975, "boundary_entropy": 0.690188, "bpb": 7.857672}, "ta": {"chunks": 49696.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.658251, "cps": 388.25, "mean_p": 0.520855, "boundary_entropy": 0.690427, "bpb": 7.780736}, "th": {"chunks": 45715.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.842089, "cps": 357.148438, "mean_p": 0.525586, "boundary_entropy": 0.689906, "bpb": 8.031108}, "zh": {"chunks": 6207.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.050965, "cps": 48.492188, "mean_p": 0.496796, "boundary_entropy": 0.686551, "bpb": 8.029667}, "my": {"chunks": 53258.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.810745, "cps": 416.078125, "mean_p": 0.521208, "boundary_entropy": 0.690584, "bpb": 8.216823}}, "bpb": {"en": 8.20994, "de": 8.1676, "vi": 8.11884, "tr": 8.11651, "fi": 8.20526, "ru": 8.02803, "el": 8.12459, "hi": 7.85767, "ta": 7.78074, "th": 8.03111, "zh": 8.02967, "my": 8.21682}, "gini_cpc": 0.391556, "gini_cps": 0.409495, "bpic": 1.322425, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1840803233586443, "vi": 0.6890195271387094, "tr": 1.2527520962970902, "fi": 0.7515121545228668, "ru": 1.1061012427858101, "el": 3.5528051575135025, "hi": 5.047484481012132, "ta": 5.304005947829907, "th": 5.670817836936983, "zh": 2.0969899090955892, "my": 5.608277765425857}, "t": 16.5}
{"event": "probe", "step": 2, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 8646.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.496811, "cps": 67.546875, "mean_p": 0.503388, "boundary_entropy": 0.687412, "bpb": 8.133828}, "de": {"chunks": 12055.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.587104, "cps": 94.179688, "mean_p": 0.507026, "boundary_entropy": 0.688166, "bpb": 8.095608}, "vi": {"chunks": 6327.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.343206, "cps": 49.429688, "mean_p": 0.491078, "boundary_entropy": 0.688497, "bpb": 8.079561}, "tr": {"chunks": 11284.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.623322, "cps": 88.15625, "mean_p": 0.506283, "boundary_entropy": 0.687853, "bpb": 8.064852}, "fi": {"chunks": 6964.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.370209, "cps": 54.40625, "mean_p": 0.496716, "boundary_entropy": 0.687908, "bpb": 8.148052}, "ru": {"chunks": 10551.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.556781, "cps": 82.429688, "mean_p": 0.496761, "boundary_entropy": 0.690194, "bpb": 7.949501}, "el": {"chunks": 36519.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.779418, "cps": 285.304688, "mean_p": 0.529732, "boundary_entropy": 0.688827, "bpb": 8.08978}, "hi": {"chunks": 42522.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.527311, "cps": 332.203125, "mean_p": 0.520727, "boundary_entropy": 0.690207, "bpb": 7.805676}, "ta": {"chunks": 49561.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.65103, "cps": 387.195312, "mean_p": 0.520378, "boundary_entropy": 0.690465, "bpb": 7.714287}, "th": {"chunks": 45709.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.841716, "cps": 357.101562, "mean_p": 0.525309, "boundary_entropy": 0.689933, "bpb": 7.952439}, "zh": {"chunks": 6193.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.048595, "cps": 48.382812, "mean_p": 0.496744, "boundary_entropy": 0.68655, "bpb": 8.018685}, "my": {"chunks": 53244.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.810006, "cps": 415.96875, "mean_p": 0.520993, "boundary_entropy": 0.690601, "bpb": 8.175888}}, "bpb": {"en": 8.13383, "de": 8.09561, "vi": 8.07956, "tr": 8.06485, "fi": 8.14805, "ru": 7.9495, "el": 8.08978, "hi": 7.80568, "ta": 7.71429, "th": 7.95244, "zh": 8.01869, "my": 8.17589}, "gini_cpc": 0.392615, "gini_cps": 0.410353, "bpic": 1.325325, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1817447907099532, "vi": 0.6908178989254027, "tr": 1.2546466029474415, "fi": 0.745170696298853, "ru": 1.120710130571155, "el": 3.5816811442540235, "hi": 5.087067487597869, "ta": 5.336094106441119, "th": 5.719914588895693, "zh": 2.1106514787652872, "my": 5.656088389466361}, "t": 24.2}
{"event": "probe", "step": 4, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 8433.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.484572, "cps": 65.882812, "mean_p": 0.502847, "boundary_entropy": 0.687406, "bpb": 7.880286}, "de": {"chunks": 11668.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.568256, "cps": 91.15625, "mean_p": 0.506293, "boundary_entropy": 0.688173, "bpb": 7.857243}, "vi": {"chunks": 6156.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.33393, "cps": 48.09375, "mean_p": 0.490666, "boundary_entropy": 0.688473, "bpb": 7.947489}, "tr": {"chunks": 11009.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.608131, "cps": 86.007812, "mean_p": 0.505596, "boundary_entropy": 0.687857, "bpb": 7.887185}, "fi": {"chunks": 6506.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.345861, "cps": 50.828125, "mean_p": 0.495724, "boundary_entropy": 0.687877, "bpb": 7.95389}, "ru": {"chunks": 10723.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.565858, "cps": 83.773438, "mean_p": 0.496889, "boundary_entropy": 0.690197, "bpb": 7.714144}, "el": {"chunks": 36341.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.770745, "cps": 283.914062, "mean_p": 0.52823, "boundary_entropy": 0.688991, "bpb": 7.945153}, "hi": {"chunks": 42404.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.520297, "cps": 331.28125, "mean_p": 0.519956, "boundary_entropy": 0.690264, "bpb": 7.626551}, "ta": {"chunks": 49029.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.622573, "cps": 383.039062, "mean_p": 0.518698, "boundary_entropy": 0.69059, "bpb": 7.463884}, "th": {"chunks": 45683.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.840099, "cps": 356.898438, "mean_p": 0.524341, "boundary_entropy": 0.690025, "bpb": 7.684495}, "zh": {"chunks": 6136.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.038943, "cps": 47.9375, "mean_p": 0.496564, "boundary_entropy": 0.686548, "bpb": 7.983267}, "my": {"chunks": 53171.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.806154, "cps": 415.398438, "mean_p": 0.520205, "boundary_entropy": 0.690662, "bpb": 8.00651}}, "bpb": {"en": 7.88029, "de": 7.85724, "vi": 7.94749, "tr": 7.88718, "fi": 7.95389, "ru": 7.71414, "el": 7.94515, "hi": 7.62655, "ta": 7.46388, "th": 7.6845, "zh": 7.98327, "my": 8.00651}, "gini_cpc": 0.396269, "gini_cps": 0.413346, "bpic": 1.33601, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1726975914018758, "vi": 0.6891241805773495, "tr": 1.2549873266601468, "fi": 0.7137468348910121, "ru": 1.1677479446088632, "el": 3.654248256197206, "hi": 5.2010828611647195, "ta": 5.412147022229275, "th": 5.861051951086517, "zh": 2.144045157605522, "my": 5.790998761838181}, "t": 31.4}
{"event": "probe", "step": 8, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 7617.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.437683, "cps": 59.507812, "mean_p": 0.500777, "boundary_entropy": 0.68737, "bpb": 7.154479}, "de": {"chunks": 10356.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.504359, "cps": 80.90625, "mean_p": 0.503632, "boundary_entropy": 0.688184, "bpb": 7.186495}, "vi": {"chunks": 5626.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.30518, "cps": 43.953125, "mean_p": 0.489196, "boundary_entropy": 0.688385, "bpb": 7.546534}, "tr": {"chunks": 10040.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.554604, "cps": 78.4375, "mean_p": 0.503186, "boundary_entropy": 0.687852, "bpb": 7.361669}, "fi": {"chunks": 5056.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.268779, "cps": 39.5, "mean_p": 0.492101, "boundary_entropy": 0.687733, "bpb": 7.360581}, "ru": {"chunks": 12136.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.640422, "cps": 94.8125, "mean_p": 0.498131, "boundary_entropy": 0.690223, "bpb": 6.973349}, "el": {"chunks": 35197.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.715003, "cps": 274.976562, "mean_p": 0.522749, "boundary_entropy": 0.689513, "bpb": 7.476352}, "hi": {"chunks": 41447.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.463418, "cps": 323.804688, "mean_p": 0.516811, "boundary_entropy": 0.690472, "bpb": 7.075516}, "ta": {"chunks": 44742.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.39326, "cps": 349.546875, "mean_p": 0.512873, "boundary_entropy": 0.690936, "bpb": 6.797339}, "th": {"chunks": 45451.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.825676, "cps": 355.085938, "mean_p": 0.520856, "boundary_entropy": 0.690323, "bpb": 6.953364}, "zh": {"chunks": 5983.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.013038, "cps": 46.742188, "mean_p": 0.496043, "boundary_entropy": 0.686543, "bpb": 7.877037}, "my": {"chunks": 52939.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.79391, "cps": 413.585938, "mean_p": 0.51819, "boundary_entropy": 0.690804, "bpb": 7.572088}}, "bpb": {"en": 7.15448, "de": 7.18649, "vi": 7.54653, "tr": 7.36167, "fi": 7.36058, "ru": 6.97335, "el": 7.47635, "hi": 7.07552, "ta": 6.79734, "th": 6.95336, "zh": 7.87704, "my": 7.57209}, "gini_cpc": 0.408426, "gini_cps": 0.422663, "bpic": 1.387545, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1523377760546731, "vi": 0.697263209259661, "tr": 1.2671362811509832, "fi": 0.6140946961283098, "ru": 1.4632095198803678, "el": 3.9183657133581886, "hi": 5.628312371579877, "ta": 5.468019926851731, "th": 6.45598543969898, "zh": 2.3145455112903788, "my": 6.383406799879241}, "t": 40.5}
{"event": "probe", "step": 16, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 4510.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.259151, "cps": 35.234375, "mean_p": 0.492305, "boundary_entropy": 0.687145, "bpb": 6.329267}, "de": {"chunks": 6198.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.301856, "cps": 48.421875, "mean_p": 0.494669, "boundary_entropy": 0.6881, "bpb": 6.404897}, "vi": {"chunks": 3434.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.186276, "cps": 26.828125, "mean_p": 0.481708, "boundary_entropy": 0.687843, "bpb": 7.023923}, "tr": {"chunks": 6314.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.348782, "cps": 49.328125, "mean_p": 0.493766, "boundary_entropy": 0.687686, "bpb": 6.708936}, "fi": {"chunks": 2448.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.130137, "cps": 19.125, "mean_p": 0.482987, "boundary_entropy": 0.687182, "bpb": 6.607385}, "ru": {"chunks": 14758.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.778786, "cps": 115.296875, "mean_p": 0.500099, "boundary_entropy": 0.690296, "bpb": 5.842405}, "el": {"chunks": 19913.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.970277, "cps": 155.570312, "mean_p": 0.503232, "boundary_entropy": 0.690397, "bpb": 6.520642}, "hi": {"chunks": 27419.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 1.629658, "cps": 214.210938, "mean_p": 0.504104, "boundary_entropy": 0.690918, "bpb": 5.801933}, "ta": {"chunks": 7733.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.41364, "cps": 60.414062, "mean_p": 0.490854, "boundary_entropy": 0.691013, "bpb": 5.849555}, "th": {"chunks": 26074.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 1.621013, "cps": 203.703125, "mean_p": 0.502661, "boundary_entropy": 0.691089, "bpb": 5.339793}, "zh": {"chunks": 5218.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.883508, "cps": 40.765625, "mean_p": 0.493437, "boundary_entropy": 0.686482, "bpb": 7.501117}, "my": {"chunks": 49061.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.589244, "cps": 383.289062, "mean_p": 0.510267, "boundary_entropy": 0.691206, "bpb": 6.227671}}, "bpb": {"en": 6.32927, "de": 6.4049, "vi": 7.02392, "tr": 6.70894, "fi": 6.60738, "ru": 5.84241, "el": 6.52064, "hi": 5.80193, "ta": 5.84955, "th": 5.33979, "zh": 7.50112, "my": 6.22767}, "gini_cpc": 0.452295, "gini_cps": 0.471906, "bpic": 2.217362, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1647876119162672, "vi": 0.718794469933012, "tr": 1.3458653261890292, "fi": 0.5021657729396195, "ru": 3.0051480322239055, "el": 3.744065405823956, "hi": 6.288457309660223, "ta": 1.5961368306381645, "th": 6.255098806437805, "zh": 3.4092449859326037, "my": 9.99126776890142}, "t": 49.3}
{"event": "probe", "step": 32, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 5970.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.343044, "cps": 46.640625, "mean_p": 0.498491, "boundary_entropy": 0.687578, "bpb": 5.765946}, "de": {"chunks": 8124.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.395656, "cps": 63.46875, "mean_p": 0.499943, "boundary_entropy": 0.688533, "bpb": 5.889455}, "vi": {"chunks": 3468.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.18812, "cps": 27.09375, "mean_p": 0.484921, "boundary_entropy": 0.688274, "bpb": 6.444618}, "tr": {"chunks": 5520.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.304922, "cps": 43.125, "mean_p": 0.494098, "boundary_entropy": 0.687971, "bpb": 6.24494}, "fi": {"chunks": 4405.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.234171, "cps": 34.414062, "mean_p": 0.492995, "boundary_entropy": 0.688004, "bpb": 5.942913}, "ru": {"chunks": 2309.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.121847, "cps": 18.039062, "mean_p": 0.488824, "boundary_entropy": 0.69008, "bpb": 5.231281}, "el": {"chunks": 491.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.023924, "cps": 3.835938, "mean_p": 0.466893, "boundary_entropy": 0.688052, "bpb": 6.168912}, "hi": {"chunks": 6220.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.369688, "cps": 48.59375, "mean_p": 0.493007, "boundary_entropy": 0.690803, "bpb": 4.969938}, "ta": {"chunks": 4310.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.230543, "cps": 33.671875, "mean_p": 0.489098, "boundary_entropy": 0.690965, "bpb": 5.23275}, "th": {"chunks": 2619.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.162823, "cps": 20.460938, "mean_p": 0.488743, "boundary_entropy": 0.690775, "bpb": 4.937162}, "zh": {"chunks": 2798.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.473756, "cps": 21.859375, "mean_p": 0.484982, "boundary_entropy": 0.686153, "bpb": 7.083006}, "my": {"chunks": 4932.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.260291, "cps": 38.53125, "mean_p": 0.491034, "boundary_entropy": 0.691175, "bpb": 4.777241}}, "bpb": {"en": 5.76595, "de": 5.88945, "vi": 6.44462, "tr": 6.24494, "fi": 5.94291, "ru": 5.23128, "el": 6.16891, "hi": 4.96994, "ta": 5.23275, "th": 4.93716, "zh": 7.08301, "my": 4.77724}, "gini_cpc": 0.264958, "gini_cps": 0.261756, "bpic": 7.500704, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.153366403438808, "vi": 0.5483852132918456, "tr": 0.8888701364636806, "fi": 0.6826275602134857, "ru": 0.3551930894578433, "el": 0.06974136383814507, "hi": 1.0776682819745085, "ta": 0.6720500066526803, "th": 0.47463987641033895, "zh": 1.3810330011044045, "my": 0.7587688283087382}, "t": 59.6}
{"event": "probe", "step": 64, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 6314.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.362811, "cps": 49.328125, "mean_p": 0.499054, "boundary_entropy": 0.687617, "bpb": 5.367718}, "de": {"chunks": 6419.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.312619, "cps": 50.148438, "mean_p": 0.496526, "boundary_entropy": 0.688461, "bpb": 5.511573}, "vi": {"chunks": 4601.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.24958, "cps": 35.945312, "mean_p": 0.489379, "boundary_entropy": 0.688662, "bpb": 5.903429}, "tr": {"chunks": 4879.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.269513, "cps": 38.117188, "mean_p": 0.493226, "boundary_entropy": 0.688016, "bpb": 5.84665}, "fi": {"chunks": 6416.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.341077, "cps": 50.125, "mean_p": 0.497458, "boundary_entropy": 0.688185, "bpb": 5.549538}, "ru": {"chunks": 5393.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.284591, "cps": 42.132812, "mean_p": 0.495191, "boundary_entropy": 0.690391, "bpb": 4.832192}, "el": {"chunks": 1429.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.069629, "cps": 11.164062, "mean_p": 0.488716, "boundary_entropy": 0.690315, "bpb": 5.11102}, "hi": {"chunks": 9826.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.584012, "cps": 76.765625, "mean_p": 0.496066, "boundary_entropy": 0.690918, "bpb": 4.442007}, "ta": {"chunks": 14085.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.75341, "cps": 110.039062, "mean_p": 0.494316, "boundary_entropy": 0.691141, "bpb": 4.370923}, "th": {"chunks": 4683.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.291141, "cps": 36.585938, "mean_p": 0.491499, "boundary_entropy": 0.690892, "bpb": 4.441362}, "zh": {"chunks": 1607.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.272096, "cps": 12.554688, "mean_p": 0.484384, "boundary_entropy": 0.68641, "bpb": 6.715069}, "my": {"chunks": 19869.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 1.048607, "cps": 155.226562, "mean_p": 0.498995, "boundary_entropy": 0.691398, "bpb": 4.242503}}, "bpb": {"en": 5.36772, "de": 5.51157, "vi": 5.90343, "tr": 5.84665, "fi": 5.54954, "ru": 4.83219, "el": 5.11102, "hi": 4.44201, "ta": 4.37092, "th": 4.44136, "zh": 6.71507, "my": 4.2425}, "gini_cpc": 0.312546, "gini_cps": 0.356568, "bpic": 4.487564, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 0.8616571798380261, "vi": 0.6879052658627566, "tr": 0.7428477449744643, "fi": 0.9400955879962066, "ru": 0.7844057139848375, "el": 0.19191588642460536, "hi": 1.609686232297764, "ta": 2.0765907945113287, "th": 0.8024585988929377, "zh": 0.749966693915799, "my": 2.8902284808751544}, "t": 75.3}
{"event": "train", "step": 100, "phase": 1, "objective": "global", "lm_loss": 3.29833, "bpb": 4.75848, "chunk_loss": 0.030827, "F": 0.17857, "G": 0.47258, "chunks": 182.86, "bytes_per_chunk": 5.6, "boundary_entropy": 0.68857, "router_noise": 0.0, "grad_norm": 2.1396, "lr": 6.06e-05, "router_trainable": true, "sec_per_step": 0.762, "alpha": 0.03, "t": 86.4}
{"event": "probe", "step": 128, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3907.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.224502, "cps": 30.523438, "mean_p": 0.462779, "boundary_entropy": 0.664571, "bpb": 4.296406}, "de": {"chunks": 4610.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.224517, "cps": 36.015625, "mean_p": 0.463995, "boundary_entropy": 0.668996, "bpb": 4.542339}, "vi": {"chunks": 7212.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.391212, "cps": 56.34375, "mean_p": 0.49174, "boundary_entropy": 0.684253, "bpb": 4.719862}, "tr": {"chunks": 4449.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.24576, "cps": 34.757812, "mean_p": 0.479655, "boundary_entropy": 0.678982, "bpb": 4.85149}, "fi": {"chunks": 4792.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.254745, "cps": 37.4375, "mean_p": 0.475654, "boundary_entropy": 0.678468, "bpb": 4.554488}, "ru": {"chunks": 5159.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.272243, "cps": 40.304688, "mean_p": 0.482764, "boundary_entropy": 0.688024, "bpb": 3.222859}, "el": {"chunks": 5907.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.287823, "cps": 46.148438, "mean_p": 0.490578, "boundary_entropy": 0.69026, "bpb": 4.087586}, "hi": {"chunks": 8178.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.486062, "cps": 63.890625, "mean_p": 0.473878, "boundary_entropy": 0.688467, "bpb": 2.900958}, "ta": {"chunks": 13040.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.697513, "cps": 101.875, "mean_p": 0.440435, "boundary_entropy": 0.679449, "bpb": 2.959679}, "th": {"chunks": 7943.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.493814, "cps": 62.054688, "mean_p": 0.478877, "boundary_entropy": 0.689103, "bpb": 2.761307}, "zh": {"chunks": 2147.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.363529, "cps": 16.773438, "mean_p": 0.486559, "boundary_entropy": 0.686562, "bpb": 6.232237}, "my": {"chunks": 10277.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.542379, "cps": 80.289062, "mean_p": 0.475883, "boundary_entropy": 0.689441, "bpb": 2.976579}}, "bpb": {"en": 4.29641, "de": 4.54234, "vi": 4.71986, "tr": 4.85149, "fi": 4.55449, "ru": 3.22286, "el": 4.08759, "hi": 2.90096, "ta": 2.95968, "th": 2.76131, "zh": 6.23224, "my": 2.97658}, "gini_cpc": 0.211664, "gini_cps": 0.244703, "bpic": 4.944293, "worst_lang": "ta", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0000673003788678, "vi": 1.742582246340517, "tr": 1.0946935603157322, "fi": 1.1347119665713568, "ru": 1.2126543319513787, "el": 1.2820555252105232, "hi": 2.165073988052838, "ta": 3.10693974571806, "th": 2.1996025083979647, "zh": 1.6192701526196913, "my": 2.415926343535328}, "t": 96.9}
{"event": "train", "step": 200, "phase": 1, "objective": "global", "lm_loss": 2.44802, "bpb": 3.53175, "chunk_loss": 0.031203, "F": 0.18893, "G": 0.40607, "chunks": 193.47, "bytes_per_chunk": 5.293, "boundary_entropy": 0.58939, "router_noise": 0.0, "grad_norm": 2.2733, "lr": 0.0001206, "router_trainable": true, "sec_per_step": 0.265, "alpha": 0.03, "t": 112.9}
{"event": "probe", "step": 256, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2952.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.169626, "cps": 23.0625, "mean_p": 0.335836, "boundary_entropy": 0.454904, "bpb": 3.509723}, "de": {"chunks": 3273.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.159402, "cps": 25.570312, "mean_p": 0.339651, "boundary_entropy": 0.490264, "bpb": 3.713585}, "vi": {"chunks": 6509.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.353078, "cps": 50.851562, "mean_p": 0.33289, "boundary_entropy": 0.426825, "bpb": 3.079652}, "tr": {"chunks": 3499.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.193283, "cps": 27.335938, "mean_p": 0.345961, "boundary_entropy": 0.510017, "bpb": 3.680121}, "fi": {"chunks": 2937.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.156132, "cps": 22.945312, "mean_p": 0.327723, "boundary_entropy": 0.504674, "bpb": 3.76301}, "ru": {"chunks": 6495.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.342744, "cps": 50.742188, "mean_p": 0.351052, "boundary_entropy": 0.593408, "bpb": 2.600619}, "el": {"chunks": 8944.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.435804, "cps": 69.875, "mean_p": 0.38549, "boundary_entropy": 0.603585, "bpb": 2.686631}, "hi": {"chunks": 5066.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.3011, "cps": 39.578125, "mean_p": 0.426206, "boundary_entropy": 0.660013, "bpb": 2.087598}, "ta": {"chunks": 10710.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.57288, "cps": 83.671875, "mean_p": 0.432058, "boundary_entropy": 0.672426, "bpb": 1.966591}, "th": {"chunks": 5420.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.33696, "cps": 42.34375, "mean_p": 0.452009, "boundary_entropy": 0.67709, "bpb": 2.079478}, "zh": {"chunks": 4006.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.678293, "cps": 31.296875, "mean_p": 0.391525, "boundary_entropy": 0.623315, "bpb": 4.908125}, "my": {"chunks": 14105.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.744406, "cps": 110.195312, "mean_p": 0.305939, "boundary_entropy": 0.533659, "bpb": 2.155955}}, "bpb": {"en": 3.50972, "de": 3.71358, "vi": 3.07965, "tr": 3.68012, "fi": 3.76301, "ru": 2.60062, "el": 2.68663, "hi": 2.0876, "ta": 1.96659, "th": 2.07948, "zh": 4.90813, "my": 2.15595}, "gini_cpc": 0.287638, "gini_cps": 0.287757, "bpic": 5.192123, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9397262645616278, "vi": 2.0815118930277996, "tr": 1.1394654414284298, "fi": 0.9204492117352198, "ru": 2.0205877147821614, "el": 2.5692047272760408, "hi": 1.7750797907681901, "ta": 3.3773165513147507, "th": 1.986488194070626, "zh": 3.9987593573023443, "my": 4.388513932439832}, "t": 127.8}
{"event": "train", "step": 300, "phase": 1, "objective": "global", "lm_loss": 2.13491, "bpb": 3.08003, "chunk_loss": 0.030658, "F": 0.18771, "G": 0.32418, "chunks": 192.22, "bytes_per_chunk": 5.327, "boundary_entropy": 0.47532, "router_noise": 0.0, "grad_norm": 0.9736, "lr": 0.0001806, "router_trainable": true, "sec_per_step": 0.254, "alpha": 0.03, "t": 138.3}
{"event": "train", "step": 400, "phase": 1, "objective": "global", "lm_loss": 2.05412, "bpb": 2.96348, "chunk_loss": 0.031355, "F": 0.20631, "G": 0.29947, "chunks": 211.27, "bytes_per_chunk": 4.847, "boundary_entropy": 0.40125, "router_noise": 0.0, "grad_norm": 0.893, "lr": 0.0002406, "router_trainable": true, "sec_per_step": 0.208, "alpha": 0.03, "t": 159.1}
{"event": "train", "step": 500, "phase": 1, "objective": "global", "lm_loss": 1.97491, "bpb": 2.84919, "chunk_loss": 0.030764, "F": 0.19696, "G": 0.29051, "chunks": 201.69, "bytes_per_chunk": 5.077, "boundary_entropy": 0.35997, "router_noise": 0.0, "grad_norm": 0.8182, "lr": 0.0003, "router_trainable": true, "sec_per_step": 0.297, "alpha": 0.03, "t": 188.8}
{"event": "train", "step": 600, "phase": 1, "objective": "global", "lm_loss": 1.82855, "bpb": 2.63803, "chunk_loss": 0.030424, "F": 0.1888, "G": 0.25507, "chunks": 193.33, "bytes_per_chunk": 5.297, "boundary_entropy": 0.29428, "router_noise": 0.0, "grad_norm": 0.8331, "lr": 0.0002999, "router_trainable": true, "sec_per_step": 0.364, "alpha": 0.03, "t": 225.2}
{"event": "train", "step": 700, "phase": 1, "objective": "global", "lm_loss": 1.70916, "bpb": 2.4658, "chunk_loss": 0.030521, "F": 0.19579, "G": 0.25073, "chunks": 200.48, "bytes_per_chunk": 5.108, "boundary_entropy": 0.26717, "router_noise": 0.0, "grad_norm": 0.7203, "lr": 0.0002998, "router_trainable": true, "sec_per_step": 0.392, "alpha": 0.03, "t": 264.4}
{"event": "train", "step": 800, "phase": 1, "objective": "global", "lm_loss": 1.59177, "bpb": 2.29643, "chunk_loss": 0.030282, "F": 0.18964, "G": 0.23352, "chunks": 194.19, "bytes_per_chunk": 5.273, "boundary_entropy": 0.22555, "router_noise": 0.0, "grad_norm": 0.6362, "lr": 0.0002995, "router_trainable": true, "sec_per_step": 0.388, "alpha": 0.03, "t": 303.2}
{"event": "train", "step": 900, "phase": 1, "objective": "global", "lm_loss": 1.56405, "bpb": 2.25645, "chunk_loss": 0.030337, "F": 0.19125, "G": 0.2368, "chunks": 195.84, "bytes_per_chunk": 5.229, "boundary_entropy": 0.22967, "router_noise": 0.0, "grad_norm": 0.6479, "lr": 0.000299, "router_trainable": true, "sec_per_step": 0.203, "alpha": 0.03, "t": 323.5}
{"event": "probe", "step": 1000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3545.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.203701, "cps": 27.695312, "mean_p": 0.220349, "boundary_entropy": 0.135808, "bpb": 2.319775}, "de": {"chunks": 3941.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.191935, "cps": 30.789062, "mean_p": 0.211675, "boundary_entropy": 0.147306, "bpb": 2.593872}, "vi": {"chunks": 4352.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.236073, "cps": 34.0, "mean_p": 0.23282, "boundary_entropy": 0.201887, "bpb": 2.102671}, "tr": {"chunks": 3805.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.210186, "cps": 29.726562, "mean_p": 0.208413, "boundary_entropy": 0.17026, "bpb": 2.591199}, "fi": {"chunks": 3183.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.16921, "cps": 24.867188, "mean_p": 0.204295, "boundary_entropy": 0.197298, "bpb": 2.843317}, "ru": {"chunks": 7189.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.379367, "cps": 56.164062, "mean_p": 0.222735, "boundary_entropy": 0.151774, "bpb": 1.608483}, "el": {"chunks": 6708.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.326853, "cps": 52.40625, "mean_p": 0.20531, "boundary_entropy": 0.214529, "bpb": 1.756029}, "hi": {"chunks": 5973.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.355007, "cps": 46.664062, "mean_p": 0.280506, "boundary_entropy": 0.42834, "bpb": 1.424039}, "ta": {"chunks": 8167.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.436855, "cps": 63.804688, "mean_p": 0.322907, "boundary_entropy": 0.548908, "bpb": 1.147979}, "th": {"chunks": 9640.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.599316, "cps": 75.3125, "mean_p": 0.376184, "boundary_entropy": 0.594284, "bpb": 1.381223}, "zh": {"chunks": 3063.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.518625, "cps": 23.929688, "mean_p": 0.239447, "boundary_entropy": 0.25848, "bpb": 3.471695}, "my": {"chunks": 9979.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.526652, "cps": 77.960938, "mean_p": 0.27404, "boundary_entropy": 0.471543, "bpb": 1.304729}}, "bpb": {"en": 2.31977, "de": 2.59387, "vi": 2.10267, "tr": 2.5912, "fi": 2.84332, "ru": 1.60848, "el": 1.75603, "hi": 1.42404, "ta": 1.14798, "th": 1.38122, "zh": 3.4717, "my": 1.30473}, "gini_cpc": 0.231506, "gini_cps": 0.231726, "bpic": 5.518456, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9422408057977055, "vi": 1.1589204474379735, "tr": 1.0318391227381134, "fi": 0.830677860889095, "ru": 1.8623750721034866, "el": 1.6045752513932732, "hi": 1.7427910561932447, "ta": 2.1445934081769145, "th": 2.9421434875461068, "zh": 2.546017795486123, "my": 2.585422519296371}, "t": 346.6}
{"event": "eval", "step": 1000, "phase": 1, "bpb": {"en": 2.3289, "de": 2.58598, "vi": 2.10767, "tr": 2.58365, "fi": 2.85354, "ru": 1.60774, "el": 1.75252, "hi": 1.43198, "ta": 1.14659, "th": 1.37699, "zh": 3.47835, "my": 1.31667}, "bytes_per_chunk": {"en": 4.764, "de": 5.278, "vi": 5.69, "tr": 5.181, "fi": 6.043, "ru": 4.856, "el": 5.642, "hi": 6.952, "ta": 6.528, "th": 5.06, "zh": 5.124, "my": 5.241}, "bpb_mean": 2.04755, "t": 352.1}
{"event": "train", "step": 1000, "phase": 1, "objective": "global", "lm_loss": 1.48218, "bpb": 2.13833, "chunk_loss": 0.03044, "F": 0.1942, "G": 0.23862, "chunks": 198.86, "bytes_per_chunk": 5.149, "boundary_entropy": 0.21725, "router_noise": 0.0, "grad_norm": 0.5996, "lr": 0.0002985, "router_trainable": true, "sec_per_step": 0.288, "alpha": 0.03, "t": 352.3}
{"event": "train", "step": 1100, "phase": 1, "objective": "global", "lm_loss": 1.4099, "bpb": 2.03406, "chunk_loss": 0.030333, "F": 0.18904, "G": 0.24441, "chunks": 193.58, "bytes_per_chunk": 5.29, "boundary_entropy": 0.24048, "router_noise": 0.0, "grad_norm": 0.5759, "lr": 0.0002978, "router_trainable": true, "sec_per_step": 0.199, "alpha": 0.03, "t": 372.1}
{"event": "train", "step": 1200, "phase": 1, "objective": "global", "lm_loss": 1.44278, "bpb": 2.0815, "chunk_loss": 0.030325, "F": 0.19026, "G": 0.23131, "chunks": 194.83, "bytes_per_chunk": 5.256, "boundary_entropy": 0.21252, "router_noise": 0.0, "grad_norm": 0.5652, "lr": 0.000297, "router_trainable": true, "sec_per_step": 0.195, "alpha": 0.03, "t": 391.7}
{"event": "train", "step": 1300, "phase": 1, "objective": "global", "lm_loss": 1.45245, "bpb": 2.09544, "chunk_loss": 0.030315, "F": 0.19159, "G": 0.23208, "chunks": 196.19, "bytes_per_chunk": 5.219, "boundary_entropy": 0.20243, "router_noise": 0.0, "grad_norm": 0.6264, "lr": 0.0002962, "router_trainable": true, "sec_per_step": 0.194, "alpha": 0.03, "t": 411.0}
{"event": "train", "step": 1400, "phase": 1, "objective": "global", "lm_loss": 1.41708, "bpb": 2.04441, "chunk_loss": 0.030316, "F": 0.19229, "G": 0.22496, "chunks": 196.91, "bytes_per_chunk": 5.2, "boundary_entropy": 0.17949, "router_noise": 0.0, "grad_norm": 0.4768, "lr": 0.0002951, "router_trainable": true, "sec_per_step": 0.209, "alpha": 0.03, "t": 431.9}
{"event": "train", "step": 1500, "phase": 1, "objective": "global", "lm_loss": 1.42291, "bpb": 2.05282, "chunk_loss": 0.030421, "F": 0.18901, "G": 0.22653, "chunks": 193.55, "bytes_per_chunk": 5.291, "boundary_entropy": 0.19753, "router_noise": 0.0, "grad_norm": 0.51, "lr": 0.000294, "router_trainable": true, "sec_per_step": 0.221, "alpha": 0.03, "t": 454.0}
{"event": "train", "step": 1600, "phase": 1, "objective": "global", "lm_loss": 1.39674, "bpb": 2.01506, "chunk_loss": 0.030226, "F": 0.18784, "G": 0.22375, "chunks": 192.34, "bytes_per_chunk": 5.324, "boundary_entropy": 0.1834, "router_noise": 0.0, "grad_norm": 0.4936, "lr": 0.0002928, "router_trainable": true, "sec_per_step": 0.196, "alpha": 0.03, "t": 473.6}
{"event": "train", "step": 1700, "phase": 1, "objective": "global", "lm_loss": 1.34209, "bpb": 1.93623, "chunk_loss": 0.030233, "F": 0.19017, "G": 0.21857, "chunks": 194.73, "bytes_per_chunk": 5.258, "boundary_entropy": 0.16211, "router_noise": 0.0, "grad_norm": 0.4662, "lr": 0.0002914, "router_trainable": true, "sec_per_step": 0.194, "alpha": 0.03, "t": 492.9}
{"event": "train", "step": 1800, "phase": 1, "objective": "global", "lm_loss": 1.3582, "bpb": 1.95947, "chunk_loss": 0.030351, "F": 0.19337, "G": 0.2349, "chunks": 198.02, "bytes_per_chunk": 5.171, "boundary_entropy": 0.20378, "router_noise": 0.0, "grad_norm": 0.4725, "lr": 0.0002899, "router_trainable": true, "sec_per_step": 0.216, "alpha": 0.03, "t": 514.6}
{"event": "train", "step": 1900, "phase": 1, "objective": "global", "lm_loss": 1.35762, "bpb": 1.95863, "chunk_loss": 0.0303, "F": 0.19052, "G": 0.22544, "chunks": 195.09, "bytes_per_chunk": 5.249, "boundary_entropy": 0.17372, "router_noise": 0.0, "grad_norm": 0.4736, "lr": 0.0002883, "router_trainable": true, "sec_per_step": 0.197, "alpha": 0.03, "t": 534.3}
{"event": "probe", "step": 2000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3489.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.200483, "cps": 27.257812, "mean_p": 0.207406, "boundary_entropy": 0.093893, "bpb": 2.06981}, "de": {"chunks": 3942.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.191984, "cps": 30.796875, "mean_p": 0.202679, "boundary_entropy": 0.102997, "bpb": 2.288293}, "vi": {"chunks": 4272.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.231733, "cps": 33.375, "mean_p": 0.201656, "boundary_entropy": 0.122532, "bpb": 1.934805}, "tr": {"chunks": 3431.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.189527, "cps": 26.804688, "mean_p": 0.189108, "boundary_entropy": 0.131604, "bpb": 2.285477}, "fi": {"chunks": 3430.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.18234, "cps": 26.796875, "mean_p": 0.199749, "boundary_entropy": 0.153702, "bpb": 2.483216}, "ru": {"chunks": 7161.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.377889, "cps": 55.945312, "mean_p": 0.213529, "boundary_entropy": 0.110961, "bpb": 1.35405}, "el": {"chunks": 6882.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.335331, "cps": 53.765625, "mean_p": 0.201787, "boundary_entropy": 0.161113, "bpb": 1.431037}, "hi": {"chunks": 7151.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.425022, "cps": 55.867188, "mean_p": 0.297047, "boundary_entropy": 0.430333, "bpb": 1.201045}, "ta": {"chunks": 9442.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.505055, "cps": 73.765625, "mean_p": 0.325801, "boundary_entropy": 0.528952, "bpb": 1.033564}, "th": {"chunks": 8139.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.505999, "cps": 63.585938, "mean_p": 0.350978, "boundary_entropy": 0.559061, "bpb": 1.155282}, "zh": {"chunks": 3047.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.515916, "cps": 23.804688, "mean_p": 0.220806, "boundary_entropy": 0.174732, "bpb": 3.180099}, "my": {"chunks": 9534.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.503167, "cps": 74.484375, "mean_p": 0.276744, "boundary_entropy": 0.466285, "bpb": 1.165906}}, "bpb": {"en": 2.06981, "de": 2.28829, "vi": 1.9348, "tr": 2.28548, "fi": 2.48322, "ru": 1.35405, "el": 1.43104, "hi": 1.20104, "ta": 1.03356, "th": 1.15528, "zh": 3.1801, "my": 1.16591}, "gini_cpc": 0.217014, "gini_cps": 0.225279, "bpic": 5.488859, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9576071135033809, "vi": 1.1558760171745164, "tr": 0.9453514994379397, "fi": 0.909505620762225, "ru": 1.8848969433523075, "el": 1.6726187999798394, "hi": 2.119995093899443, "ta": 2.5191943717047223, "th": 2.5239057554035673, "zh": 2.573369577086013, "my": 2.509775770443651}, "t": 556.5}
{"event": "eval", "step": 2000, "phase": 1, "bpb": {"en": 2.07901, "de": 2.27925, "vi": 1.94428, "tr": 2.27687, "fi": 2.50091, "ru": 1.35382, "el": 1.42828, "hi": 1.20879, "ta": 1.03378, "th": 1.1501, "zh": 3.18559, "my": 1.17664}, "bytes_per_chunk": {"en": 4.832, "de": 5.323, "vi": 5.714, "tr": 5.706, "fi": 5.626, "ru": 4.9, "el": 5.564, "hi": 5.995, "ta": 5.468, "th": 5.726, "zh": 5.21, "my": 5.698}, "bpb_mean": 1.80144, "t": 562.6}
{"event": "train", "step": 2000, "phase": 1, "objective": "global", "lm_loss": 1.3426, "bpb": 1.93696, "chunk_loss": 0.030325, "F": 0.1954, "G": 0.22273, "chunks": 200.09, "bytes_per_chunk": 5.118, "boundary_entropy": 0.15597, "router_noise": 0.0, "grad_norm": 0.455, "lr": 0.0002866, "router_trainable": true, "sec_per_step": 0.285, "alpha": 0.03, "t": 562.8}
{"event": "train", "step": 2100, "phase": 1, "objective": "global", "lm_loss": 1.28424, "bpb": 1.85277, "chunk_loss": 0.030206, "F": 0.19083, "G": 0.2287, "chunks": 195.41, "bytes_per_chunk": 5.24, "boundary_entropy": 0.18407, "router_noise": 0.0, "grad_norm": 0.4561, "lr": 0.0002848, "router_trainable": true, "sec_per_step": 0.194, "alpha": 0.03, "t": 582.2}
{"event": "train", "step": 2200, "phase": 1, "objective": "global", "lm_loss": 1.25004, "bpb": 1.80342, "chunk_loss": 0.030283, "F": 0.18968, "G": 0.23233, "chunks": 194.23, "bytes_per_chunk": 5.272, "boundary_entropy": 0.20579, "router_noise": 0.0, "grad_norm": 0.4465, "lr": 0.0002829, "router_trainable": true, "sec_per_step": 0.186, "alpha": 0.03, "t": 600.8}
{"event": "train", "step": 2300, "phase": 1, "objective": "global", "lm_loss": 1.32403, "bpb": 1.91017, "chunk_loss": 0.030297, "F": 0.19295, "G": 0.22109, "chunks": 197.58, "bytes_per_chunk": 5.183, "boundary_entropy": 0.16292, "router_noise": 0.0, "grad_norm": 0.4098, "lr": 0.0002809, "router_trainable": true, "sec_per_step": 0.189, "alpha": 0.03, "t": 619.7}
{"event": "train", "step": 2400, "phase": 1, "objective": "global", "lm_loss": 1.33989, "bpb": 1.93305, "chunk_loss": 0.030238, "F": 0.19078, "G": 0.22134, "chunks": 195.36, "bytes_per_chunk": 5.242, "boundary_entropy": 0.15761, "router_noise": 0.0, "grad_norm": 0.4382, "lr": 0.0002788, "router_trainable": true, "sec_per_step": 0.186, "alpha": 0.03, "t": 638.3}
{"event": "train", "step": 2500, "phase": 1, "objective": "global", "lm_loss": 1.24735, "bpb": 1.79954, "chunk_loss": 0.030355, "F": 0.19243, "G": 0.2275, "chunks": 197.05, "bytes_per_chunk": 5.197, "boundary_entropy": 0.1774, "router_noise": 0.0, "grad_norm": 0.3918, "lr": 0.0002765, "router_trainable": true, "sec_per_step": 0.186, "alpha": 0.03, "t": 656.9}
{"event": "train", "step": 2600, "phase": 1, "objective": "global", "lm_loss": 1.27091, "bpb": 1.83354, "chunk_loss": 0.030273, "F": 0.19058, "G": 0.23034, "chunks": 195.16, "bytes_per_chunk": 5.247, "boundary_entropy": 0.18152, "router_noise": 0.0, "grad_norm": 0.4588, "lr": 0.0002742, "router_trainable": true, "sec_per_step": 0.183, "alpha": 0.03, "t": 675.2}
{"event": "train", "step": 2700, "phase": 1, "objective": "global", "lm_loss": 1.26609, "bpb": 1.82658, "chunk_loss": 0.030096, "F": 0.18536, "G": 0.22219, "chunks": 189.81, "bytes_per_chunk": 5.395, "boundary_entropy": 0.17024, "router_noise": 0.0, "grad_norm": 0.3799, "lr": 0.0002718, "router_trainable": true, "sec_per_step": 0.186, "alpha": 0.03, "t": 693.7}
{"event": "train", "step": 2800, "phase": 1, "objective": "global", "lm_loss": 1.34632, "bpb": 1.94232, "chunk_loss": 0.030357, "F": 0.19255, "G": 0.22982, "chunks": 197.17, "bytes_per_chunk": 5.193, "boundary_entropy": 0.18207, "router_noise": 0.0, "grad_norm": 0.4223, "lr": 0.0002693, "router_trainable": true, "sec_per_step": 0.184, "alpha": 0.03, "t": 712.2}
{"event": "train", "step": 2900, "phase": 1, "objective": "global", "lm_loss": 1.27309, "bpb": 1.83668, "chunk_loss": 0.030054, "F": 0.18327, "G": 0.22358, "chunks": 187.67, "bytes_per_chunk": 5.456, "boundary_entropy": 0.18012, "router_noise": 0.0, "grad_norm": 0.3845, "lr": 0.0002667, "router_trainable": true, "sec_per_step": 0.188, "alpha": 0.03, "t": 731.0}
{"event": "probe", "step": 3000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3477.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.199793, "cps": 27.164062, "mean_p": 0.20123, "boundary_entropy": 0.087662, "bpb": 1.962731}, "de": {"chunks": 3879.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.188915, "cps": 30.304688, "mean_p": 0.195889, "boundary_entropy": 0.087734, "bpb": 2.149938}, "vi": {"chunks": 4257.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.230919, "cps": 33.257812, "mean_p": 0.195862, "boundary_entropy": 0.111507, "bpb": 1.818329}, "tr": {"chunks": 3262.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.180191, "cps": 25.484375, "mean_p": 0.181299, "boundary_entropy": 0.124205, "bpb": 2.137194}, "fi": {"chunks": 3327.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.176865, "cps": 25.992188, "mean_p": 0.190935, "boundary_entropy": 0.136957, "bpb": 2.303996}, "ru": {"chunks": 7081.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.373668, "cps": 55.320312, "mean_p": 0.20659, "boundary_entropy": 0.092985, "bpb": 1.261082}, "el": {"chunks": 6930.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.33767, "cps": 54.140625, "mean_p": 0.202136, "boundary_entropy": 0.148283, "bpb": 1.316363}, "hi": {"chunks": 7425.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.441308, "cps": 58.007812, "mean_p": 0.301557, "boundary_entropy": 0.451707, "bpb": 1.120109}, "ta": {"chunks": 8937.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.478042, "cps": 69.820312, "mean_p": 0.374829, "boundary_entropy": 0.573625, "bpb": 0.949169}, "th": {"chunks": 9739.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.605471, "cps": 76.085938, "mean_p": 0.370917, "boundary_entropy": 0.57446, "bpb": 1.073845}, "zh": {"chunks": 2991.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.506434, "cps": 23.367188, "mean_p": 0.210153, "boundary_entropy": 0.144626, "bpb": 2.986962}, "my": {"chunks": 8984.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.47414, "cps": 70.1875, "mean_p": 0.270432, "boundary_entropy": 0.432768, "bpb": 1.094956}}, "bpb": {"en": 1.96273, "de": 2.14994, "vi": 1.81833, "tr": 2.13719, "fi": 2.304, "ru": 1.26108, "el": 1.31636, "hi": 1.12011, "ta": 0.94917, "th": 1.07384, "zh": 2.98696, "my": 1.09496}, "gini_cpc": 0.232778, "gini_cps": 0.233577, "bpic": 5.460044, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9455550140944016, "vi": 1.155792675007424, "tr": 0.9018884699512335, "fi": 0.885238608713377, "ru": 1.8702721616533224, "el": 1.6900977320548252, "hi": 2.2088224849264173, "ta": 2.3926860522242874, "th": 3.0304891266012124, "zh": 2.5347924209217343, "my": 2.3731533174759383}, "t": 752.4}
{"event": "eval", "step": 3000, "phase": 1, "bpb": {"en": 1.9707, "de": 2.14145, "vi": 1.8244, "tr": 2.12706, "fi": 2.31974, "ru": 1.25951, "el": 1.31297, "hi": 1.12744, "ta": 0.9502, "th": 1.07225, "zh": 2.99549, "my": 1.10581}, "bytes_per_chunk": {"en": 4.847, "de": 5.391, "vi": 5.724, "tr": 5.984, "fi": 5.755, "ru": 4.96, "el": 5.552, "hi": 5.864, "ta": 5.792, "th": 4.711, "zh": 5.281, "my": 6.304}, "bpb_mean": 1.68392, "t": 757.9}
{"event": "train", "step": 3000, "phase": 1, "objective": "global", "lm_loss": 1.31781, "bpb": 1.9012, "chunk_loss": 0.030222, "F": 0.1904, "G": 0.21535, "chunks": 194.97, "bytes_per_chunk": 5.252, "boundary_entropy": 0.14481, "router_noise": 0.0, "grad_norm": 0.4219, "lr": 0.000264, "router_trainable": true, "sec_per_step": 0.271, "alpha": 0.03, "t": 758.1}
{"event": "train", "step": 3100, "phase": 1, "objective": "global", "lm_loss": 1.31704, "bpb": 1.90009, "chunk_loss": 0.030242, "F": 0.19225, "G": 0.216, "chunks": 196.86, "bytes_per_chunk": 5.202, "boundary_entropy": 0.1454, "router_noise": 0.0, "grad_norm": 0.4109, "lr": 0.0002612, "router_trainable": true, "sec_per_step": 0.18, "alpha": 0.03, "t": 776.1}
{"event": "train", "step": 3200, "phase": 1, "objective": "global", "lm_loss": 1.19369, "bpb": 1.72214, "chunk_loss": 0.030346, "F": 0.19373, "G": 0.23037, "chunks": 198.38, "bytes_per_chunk": 5.162, "boundary_entropy": 0.17808, "router_noise": 0.0, "grad_norm": 0.3795, "lr": 0.0002583, "router_trainable": true, "sec_per_step": 0.185, "alpha": 0.03, "t": 794.6}
{"event": "train", "step": 3300, "phase": 1, "objective": "global", "lm_loss": 1.27388, "bpb": 1.83783, "chunk_loss": 0.030236, "F": 0.18872, "G": 0.22816, "chunks": 193.25, "bytes_per_chunk": 5.299, "boundary_entropy": 0.18157, "router_noise": 0.0, "grad_norm": 0.3978, "lr": 0.0002553, "router_trainable": true, "sec_per_step": 0.187, "alpha": 0.03, "t": 813.3}
{"event": "train", "step": 3400, "phase": 1, "objective": "global", "lm_loss": 1.22496, "bpb": 1.76725, "chunk_loss": 0.030425, "F": 0.19823, "G": 0.23678, "chunks": 202.98, "bytes_per_chunk": 5.045, "boundary_entropy": 0.17428, "router_noise": 0.0, "grad_norm": 0.4052, "lr": 0.0002523, "router_trainable": true, "sec_per_step": 0.186, "alpha": 0.03, "t": 831.8}
{"event": "train", "step": 3500, "phase": 1, "objective": "global", "lm_loss": 1.30028, "bpb": 1.87591, "chunk_loss": 0.030119, "F": 0.18546, "G": 0.20895, "chunks": 189.91, "bytes_per_chunk": 5.392, "boundary_entropy": 0.13638, "router_noise": 0.0, "grad_norm": 0.3868, "lr": 0.0002492, "router_trainable": true, "sec_per_step": 0.19, "alpha": 0.03, "t": 850.8}
{"event": "train", "step": 3600, "phase": 1, "objective": "global", "lm_loss": 1.21084, "bpb": 1.74688, "chunk_loss": 0.030188, "F": 0.18951, "G": 0.21876, "chunks": 194.06, "bytes_per_chunk": 5.277, "boundary_entropy": 0.15461, "router_noise": 0.0, "grad_norm": 0.4304, "lr": 0.000246, "router_trainable": true, "sec_per_step": 0.187, "alpha": 0.03, "t": 869.5}
{"event": "train", "step": 3700, "phase": 1, "objective": "global", "lm_loss": 1.12484, "bpb": 1.6228, "chunk_loss": 0.030171, "F": 0.18497, "G": 0.22774, "chunks": 189.41, "bytes_per_chunk": 5.406, "boundary_entropy": 0.19017, "router_noise": 0.0, "grad_norm": 0.3863, "lr": 0.0002427, "router_trainable": true, "sec_per_step": 0.19, "alpha": 0.03, "t": 888.6}
{"event": "train", "step": 3800, "phase": 1, "objective": "global", "lm_loss": 1.20013, "bpb": 1.73142, "chunk_loss": 0.030213, "F": 0.1902, "G": 0.21844, "chunks": 194.77, "bytes_per_chunk": 5.258, "boundary_entropy": 0.1526, "router_noise": 0.0, "grad_norm": 0.3839, "lr": 0.0002394, "router_trainable": true, "sec_per_step": 0.189, "alpha": 0.03, "t": 907.4}
{"event": "train", "step": 3900, "phase": 1, "objective": "global", "lm_loss": 1.22037, "bpb": 1.76062, "chunk_loss": 0.030107, "F": 0.18729, "G": 0.21565, "chunks": 191.78, "bytes_per_chunk": 5.339, "boundary_entropy": 0.13847, "router_noise": 0.0, "grad_norm": 0.3826, "lr": 0.000236, "router_trainable": true, "sec_per_step": 0.183, "alpha": 0.03, "t": 925.7}
{"event": "probe", "step": 4000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3486.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.20031, "cps": 27.234375, "mean_p": 0.199708, "boundary_entropy": 0.078058, "bpb": 1.832901}, "de": {"chunks": 3907.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.190279, "cps": 30.523438, "mean_p": 0.19344, "boundary_entropy": 0.072302, "bpb": 2.007238}, "vi": {"chunks": 4265.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.231353, "cps": 33.320312, "mean_p": 0.191862, "boundary_entropy": 0.100386, "bpb": 1.692687}, "tr": {"chunks": 3351.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.185107, "cps": 26.179688, "mean_p": 0.181625, "boundary_entropy": 0.112141, "bpb": 1.962661}, "fi": {"chunks": 3452.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.18351, "cps": 26.96875, "mean_p": 0.188381, "boundary_entropy": 0.109283, "bpb": 2.153018}, "ru": {"chunks": 7057.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.372401, "cps": 55.132812, "mean_p": 0.202575, "boundary_entropy": 0.090722, "bpb": 1.212638}, "el": {"chunks": 6707.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.326804, "cps": 52.398438, "mean_p": 0.192888, "boundary_entropy": 0.136938, "bpb": 1.242663}, "hi": {"chunks": 6952.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.413195, "cps": 54.3125, "mean_p": 0.304091, "boundary_entropy": 0.448865, "bpb": 1.044295}, "ta": {"chunks": 8173.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.437176, "cps": 63.851562, "mean_p": 0.343855, "boundary_entropy": 0.541848, "bpb": 0.904262}, "th": {"chunks": 7356.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.45732, "cps": 57.46875, "mean_p": 0.347127, "boundary_entropy": 0.560114, "bpb": 1.02324}, "zh": {"chunks": 2974.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.503556, "cps": 23.234375, "mean_p": 0.208397, "boundary_entropy": 0.134307, "bpb": 2.874327}, "my": {"chunks": 9470.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.499789, "cps": 73.984375, "mean_p": 0.30804, "boundary_entropy": 0.489973, "bpb": 1.061662}}, "bpb": {"en": 1.8329, "de": 2.00724, "vi": 1.69269, "tr": 1.96266, "fi": 2.15302, "ru": 1.21264, "el": 1.24266, "hi": 1.0443, "ta": 0.90426, "th": 1.02324, "zh": 2.87433, "my": 1.06166}, "gini_cpc": 0.208237, "gini_cps": 0.212973, "bpic": 5.715279, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.949921552753374, "vi": 1.154975125423232, "tr": 0.9241034972665024, "fi": 0.9161269118638333, "ru": 1.8591209315210333, "el": 1.6314891826583384, "hi": 2.06277295128295, "ta": 2.182492534613294, "th": 2.283060355458871, "zh": 2.513878357025412, "my": 2.4950734812951256}, "t": 947.7}
{"event": "eval", "step": 4000, "phase": 1, "bpb": {"en": 1.83472, "de": 2.00811, "vi": 1.7046, "tr": 1.95412, "fi": 2.16749, "ru": 1.20953, "el": 1.24015, "hi": 1.0528, "ta": 0.90728, "th": 1.02168, "zh": 2.88276, "my": 1.06957}, "bytes_per_chunk": {"en": 4.852, "de": 5.39, "vi": 5.698, "tr": 5.875, "fi": 5.543, "ru": 4.988, "el": 5.687, "hi": 6.319, "ta": 6.548, "th": 6.371, "zh": 5.32, "my": 5.965}, "bpb_mean": 1.58773, "t": 953.0}
{"event": "train", "step": 4000, "phase": 1, "objective": "global", "lm_loss": 1.28531, "bpb": 1.85432, "chunk_loss": 0.030111, "F": 0.18242, "G": 0.21521, "chunks": 186.8, "bytes_per_chunk": 5.482, "boundary_entropy": 0.16262, "router_noise": 0.0, "grad_norm": 0.3908, "lr": 0.0002325, "router_trainable": true, "sec_per_step": 0.275, "alpha": 0.03, "t": 953.2}
{"event": "train", "step": 4100, "phase": 1, "objective": "global", "lm_loss": 1.26572, "bpb": 1.82605, "chunk_loss": 0.030167, "F": 0.18581, "G": 0.21211, "chunks": 190.27, "bytes_per_chunk": 5.382, "boundary_entropy": 0.14057, "router_noise": 0.0, "grad_norm": 0.412, "lr": 0.000229, "router_trainable": true, "sec_per_step": 0.189, "alpha": 0.03, "t": 972.1}
{"event": "train", "step": 4200, "phase": 1, "objective": "global", "lm_loss": 1.20502, "bpb": 1.73847, "chunk_loss": 0.030254, "F": 0.19296, "G": 0.21476, "chunks": 197.59, "bytes_per_chunk": 5.182, "boundary_entropy": 0.13257, "router_noise": 0.0, "grad_norm": 0.396, "lr": 0.0002254, "router_trainable": true, "sec_per_step": 0.192, "alpha": 0.03, "t": 991.3}
{"event": "train", "step": 4300, "phase": 1, "objective": "global", "lm_loss": 1.18187, "bpb": 1.70508, "chunk_loss": 0.030063, "F": 0.18254, "G": 0.21909, "chunks": 186.92, "bytes_per_chunk": 5.478, "boundary_entropy": 0.16219, "router_noise": 0.0, "grad_norm": 0.3701, "lr": 0.0002217, "router_trainable": true, "sec_per_step": 0.187, "alpha": 0.03, "t": 1010.0}
{"event": "phase_transition", "step": 4400, "phase": 2, "objective": "parity_a", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0, "t": 1028.4}
{"event": "train", "step": 4400, "phase": 2, "objective": "parity_a", "lm_loss": 1.13648, "bpb": 1.63959, "chunk_loss": 0.034265, "F": 0.18951, "G": 0.21874, "chunks": 194.06, "bytes_per_chunk": 5.277, "boundary_entropy": 0.15377, "router_noise": 0.0, "grad_norm": 0.3536, "lr": 0.0002181, "router_trainable": true, "sec_per_step": 0.187, "n_target_mean": 9.7474, "alpha": 0.03, "t": 1028.7}
{"event": "train", "step": 4500, "phase": 2, "objective": "parity_a", "lm_loss": 1.21875, "bpb": 1.75828, "chunk_loss": 0.030696, "F": 0.16553, "G": 0.19014, "chunks": 169.5, "bytes_per_chunk": 6.041, "boundary_entropy": 0.14374, "router_noise": 0.0, "grad_norm": 0.3956, "lr": 0.0002143, "router_trainable": true, "sec_per_step": 0.187, "n_target_mean": 8.995, "alpha": 0.03, "t": 1047.4}
{"event": "train", "step": 4600, "phase": 2, "objective": "parity_a", "lm_loss": 1.1636, "bpb": 1.67872, "chunk_loss": 0.031105, "F": 0.17194, "G": 0.18097, "chunks": 176.06, "bytes_per_chunk": 5.816, "boundary_entropy": 0.12885, "router_noise": 0.0, "grad_norm": 0.3695, "lr": 0.0002105, "router_trainable": true, "sec_per_step": 0.203, "n_target_mean": 9.4029, "alpha": 0.03, "t": 1067.7}
{"event": "train", "step": 4700, "phase": 2, "objective": "parity_a", "lm_loss": 1.09531, "bpb": 1.5802, "chunk_loss": 0.031038, "F": 0.16118, "G": 0.18304, "chunks": 165.05, "bytes_per_chunk": 6.204, "boundary_entropy": 0.15222, "router_noise": 0.0, "grad_norm": 0.365, "lr": 0.0002067, "router_trainable": true, "sec_per_step": 0.184, "n_target_mean": 10.3158, "alpha": 0.03, "t": 1086.1}
{"event": "train", "step": 4800, "phase": 2, "objective": "parity_a", "lm_loss": 1.22071, "bpb": 1.76112, "chunk_loss": 0.030882, "F": 0.17813, "G": 0.18612, "chunks": 182.41, "bytes_per_chunk": 5.614, "boundary_entropy": 0.11818, "router_noise": 0.0, "grad_norm": 0.3811, "lr": 0.0002029, "router_trainable": true, "sec_per_step": 0.19, "n_target_mean": 8.5167, "alpha": 0.03, "t": 1105.1}
{"event": "train", "step": 4900, "phase": 2, "objective": "parity_a", "lm_loss": 1.20825, "bpb": 1.74314, "chunk_loss": 0.030846, "F": 0.18283, "G": 0.1908, "chunks": 187.22, "bytes_per_chunk": 5.47, "boundary_entropy": 0.1039, "router_noise": 0.0, "grad_norm": 0.3969, "lr": 0.000199, "router_trainable": true, "sec_per_step": 0.19, "n_target_mean": 7.9172, "alpha": 0.03, "t": 1124.0}
{"event": "probe", "step": 5000, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 3472.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.199506, "cps": 27.125, "mean_p": 0.194784, "boundary_entropy": 0.071785, "bpb": 1.766713}, "de": {"chunks": 3837.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.18687, "cps": 29.976562, "mean_p": 0.18711, "boundary_entropy": 0.062826, "bpb": 1.929583}, "vi": {"chunks": 4193.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.227448, "cps": 32.757812, "mean_p": 0.174647, "boundary_entropy": 0.076556, "bpb": 1.595168}, "tr": {"chunks": 2997.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.165553, "cps": 23.414062, "mean_p": 0.16172, "boundary_entropy": 0.087533, "bpb": 1.874974}, "fi": {"chunks": 3333.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.177184, "cps": 26.039062, "mean_p": 0.178547, "boundary_entropy": 0.090991, "bpb": 2.050621}, "ru": {"chunks": 6695.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.353298, "cps": 52.304688, "mean_p": 0.167219, "boundary_entropy": 0.12028, "bpb": 1.180117}, "el": {"chunks": 6087.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.296594, "cps": 47.554688, "mean_p": 0.150811, "boundary_entropy": 0.1162, "bpb": 1.216763}, "hi": {"chunks": 2984.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.177355, "cps": 23.3125, "mean_p": 0.109107, "boundary_entropy": 0.215706, "bpb": 1.036053}, "ta": {"chunks": 2654.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.141963, "cps": 20.734375, "mean_p": 0.181395, "boundary_entropy": 0.371647, "bpb": 0.881376}, "th": {"chunks": 2663.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.165558, "cps": 20.804688, "mean_p": 0.203103, "boundary_entropy": 0.40703, "bpb": 0.998871}, "zh": {"chunks": 2956.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.500508, "cps": 23.09375, "mean_p": 0.206491, "boundary_entropy": 0.122154, "bpb": 2.763727}, "my": {"chunks": 2430.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.128246, "cps": 18.984375, "mean_p": 0.159283, "boundary_entropy": 0.347579, "bpb": 1.030001}}, "bpb": {"en": 1.76671, "de": 1.92958, "vi": 1.59517, "tr": 1.87497, "fi": 2.05062, "ru": 1.18012, "el": 1.21676, "hi": 1.03605, "ta": 0.88138, "th": 0.99887, "zh": 2.76373, "my": 1.03}, "gini_cpc": 0.223649, "gini_cps": 0.178188, "bpic": 8.663032, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9366639287638804, "vi": 1.1400558413606656, "tr": 0.8298136896372039, "fi": 0.888112195727228, "ru": 1.7708662895322398, "el": 1.4866435819674946, "hi": 0.8889721379612575, "ta": 0.7115736421305876, "th": 0.8298402675017081, "zh": 2.508738477311989, "my": 0.6428169224236303}, "t": 1145.9}
{"event": "eval", "step": 5000, "phase": 2, "bpb": {"en": 1.76736, "de": 1.93351, "vi": 1.61047, "tr": 1.86648, "fi": 2.06454, "ru": 1.17797, "el": 1.21536, "hi": 1.04268, "ta": 0.88463, "th": 0.99694, "zh": 2.77312, "my": 1.03822}, "bytes_per_chunk": {"en": 4.878, "de": 5.469, "vi": 5.791, "tr": 6.448, "fi": 5.724, "ru": 5.261, "el": 6.245, "hi": 14.081, "ta": 18.3, "th": 18.007, "zh": 5.313, "my": 21.41}, "bpb_mean": 1.53094, "t": 1150.9}
{"event": "train", "step": 5000, "phase": 2, "objective": "parity_a", "lm_loss": 1.17732, "bpb": 1.69851, "chunk_loss": 0.030917, "F": 0.17094, "G": 0.18232, "chunks": 175.05, "bytes_per_chunk": 5.85, "boundary_entropy": 0.12496, "router_noise": 0.0, "grad_norm": 0.382, "lr": 0.000195, "router_trainable": true, "sec_per_step": 0.27, "n_target_mean": 9.2185, "alpha": 0.03, "t": 1151.0}
{"event": "train", "step": 5100, "phase": 2, "objective": "parity_a", "lm_loss": 1.18332, "bpb": 1.70716, "chunk_loss": 0.030798, "F": 0.17035, "G": 0.17929, "chunks": 174.44, "bytes_per_chunk": 5.87, "boundary_entropy": 0.12177, "router_noise": 0.0, "grad_norm": 0.3723, "lr": 0.0001911, "router_trainable": true, "sec_per_step": 0.188, "n_target_mean": 9.0005, "alpha": 0.03, "t": 1169.9}
{"event": "train", "step": 5200, "phase": 2, "objective": "parity_a", "lm_loss": 1.11362, "bpb": 1.60661, "chunk_loss": 0.03079, "F": 0.15973, "G": 0.1782, "chunks": 163.56, "bytes_per_chunk": 6.261, "boundary_entropy": 0.14094, "router_noise": 0.0, "grad_norm": 0.3774, "lr": 0.0001871, "router_trainable": true, "sec_per_step": 0.188, "n_target_mean": 9.8194, "alpha": 0.03, "t": 1188.7}
{"event": "train", "step": 5300, "phase": 2, "objective": "parity_a", "lm_loss": 1.13633, "bpb": 1.63937, "chunk_loss": 0.030866, "F": 0.16765, "G": 0.17859, "chunks": 171.67, "bytes_per_chunk": 5.965, "boundary_entropy": 0.11955, "router_noise": 0.0, "grad_norm": 0.3944, "lr": 0.0001831, "router_trainable": true, "sec_per_step": 0.19, "n_target_mean": 9.3411, "alpha": 0.03, "t": 1207.7}
{"event": "train", "step": 5400, "phase": 2, "objective": "parity_a", "lm_loss": 1.19948, "bpb": 1.73049, "chunk_loss": 0.030785, "F": 0.1703, "G": 0.18075, "chunks": 174.39, "bytes_per_chunk": 5.872, "boundary_entropy": 0.11781, "router_noise": 0.0, "grad_norm": 0.3848, "lr": 0.0001791, "router_trainable": true, "sec_per_step": 0.199, "n_target_mean": 9.0989, "alpha": 0.03, "t": 1227.6}
{"event": "train", "step": 5500, "phase": 2, "objective": "parity_a", "lm_loss": 1.14579, "bpb": 1.65303, "chunk_loss": 0.030592, "F": 0.16051, "G": 0.17776, "chunks": 164.36, "bytes_per_chunk": 6.23, "boundary_entropy": 0.1291, "router_noise": 0.0, "grad_norm": 0.3975, "lr": 0.0001751, "router_trainable": true, "sec_per_step": 0.189, "n_target_mean": 9.3685, "alpha": 0.03, "t": 1246.4}
{"event": "train", "step": 5600, "phase": 2, "objective": "parity_a", "lm_loss": 1.10629, "bpb": 1.59604, "chunk_loss": 0.030634, "F": 0.164, "G": 0.17409, "chunks": 167.94, "bytes_per_chunk": 6.098, "boundary_entropy": 0.11532, "router_noise": 0.0, "grad_norm": 0.368, "lr": 0.0001711, "router_trainable": true, "sec_per_step": 0.19, "n_target_mean": 9.17, "alpha": 0.03, "t": 1265.4}
{"event": "train", "step": 5700, "phase": 2, "objective": "parity_a", "lm_loss": 1.09666, "bpb": 1.58215, "chunk_loss": 0.030673, "F": 0.15985, "G": 0.17577, "chunks": 163.69, "bytes_per_chunk": 6.256, "boundary_entropy": 0.12939, "router_noise": 0.0, "grad_norm": 0.3816, "lr": 0.000167, "router_trainable": true, "sec_per_step": 0.194, "n_target_mean": 10.0188, "alpha": 0.03, "t": 1284.8}
{"event": "train", "step": 5800, "phase": 2, "objective": "parity_a", "lm_loss": 1.08468, "bpb": 1.56486, "chunk_loss": 0.030428, "F": 0.15669, "G": 0.17256, "chunks": 160.45, "bytes_per_chunk": 6.382, "boundary_entropy": 0.12132, "router_noise": 0.0, "grad_norm": 0.3672, "lr": 0.000163, "router_trainable": true, "sec_per_step": 0.189, "n_target_mean": 9.5687, "alpha": 0.03, "t": 1303.7}
{"event": "train", "step": 5900, "phase": 2, "objective": "parity_a", "lm_loss": 1.13147, "bpb": 1.63236, "chunk_loss": 0.030508, "F": 0.16673, "G": 0.17564, "chunks": 170.73, "bytes_per_chunk": 5.998, "boundary_entropy": 0.10255, "router_noise": 0.0, "grad_norm": 0.3899, "lr": 0.0001589, "router_trainable": true, "sec_per_step": 0.187, "n_target_mean": 8.6873, "alpha": 0.03, "t": 1322.4}
{"event": "probe", "step": 6000, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 3454.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.198472, "cps": 26.984375, "mean_p": 0.194081, "boundary_entropy": 0.060993, "bpb": 1.727913}, "de": {"chunks": 3856.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.187795, "cps": 30.125, "mean_p": 0.186626, "boundary_entropy": 0.054276, "bpb": 1.87425}, "vi": {"chunks": 3896.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.211337, "cps": 30.4375, "mean_p": 0.163901, "boundary_entropy": 0.067899, "bpb": 1.547744}, "tr": {"chunks": 3090.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.17069, "cps": 24.140625, "mean_p": 0.163764, "boundary_entropy": 0.079005, "bpb": 1.815076}, "fi": {"chunks": 3266.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.173622, "cps": 25.515625, "mean_p": 0.171161, "boundary_entropy": 0.076666, "bpb": 1.996875}, "ru": {"chunks": 6136.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.323799, "cps": 47.9375, "mean_p": 0.154669, "boundary_entropy": 0.114895, "bpb": 1.144334}, "el": {"chunks": 5636.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.274619, "cps": 44.03125, "mean_p": 0.136198, "boundary_entropy": 0.110994, "bpb": 1.179497}, "hi": {"chunks": 3115.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.185141, "cps": 24.335938, "mean_p": 0.112294, "boundary_entropy": 0.207422, "bpb": 0.996572}, "ta": {"chunks": 3023.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.161701, "cps": 23.617188, "mean_p": 0.173979, "boundary_entropy": 0.344492, "bpb": 0.852696}, "th": {"chunks": 2984.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.185514, "cps": 23.3125, "mean_p": 0.190583, "boundary_entropy": 0.379274, "bpb": 0.957891}, "zh": {"chunks": 2975.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.503725, "cps": 23.242188, "mean_p": 0.202899, "boundary_entropy": 0.10191, "bpb": 2.647814}, "my": {"chunks": 2996.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.158117, "cps": 23.40625, "mean_p": 0.169676, "boundary_entropy": 0.356271, "bpb": 0.994557}}, "bpb": {"en": 1.72791, "de": 1.87425, "vi": 1.54774, "tr": 1.81508, "fi": 1.99687, "ru": 1.14433, "el": 1.1795, "hi": 0.99657, "ta": 0.8527, "th": 0.95789, "zh": 2.64781, "my": 0.99456}, "gini_cpc": 0.189518, "gini_cps": 0.131952, "bpic": 8.638463, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9462075412322972, "vi": 1.064823416724657, "tr": 0.8600223049723098, "fi": 0.8747945806696835, "ru": 1.6314656098317428, "el": 1.3836680926939458, "hi": 0.9328348552101877, "ta": 0.8147314190724757, "th": 0.9347157067108204, "zh": 2.538021603068808, "my": 0.7966732222610531}, "t": 1344.5}
{"event": "eval", "step": 6000, "phase": 2, "bpb": {"en": 1.72817, "de": 1.87942, "vi": 1.56471, "tr": 1.80822, "fi": 2.01218, "ru": 1.14268, "el": 1.1778, "hi": 1.0019, "ta": 0.85714, "th": 0.95684, "zh": 2.65473, "my": 1.00185}, "bytes_per_chunk": {"en": 4.905, "de": 5.487, "vi": 6.237, "tr": 6.387, "fi": 5.851, "ru": 5.804, "el": 6.726, "hi": 13.611, "ta": 16.894, "th": 15.552, "zh": 5.302, "my": 17.989}, "bpb_mean": 1.48214, "t": 1350.7}
{"event": "train", "step": 6000, "phase": 2, "objective": "parity_a", "lm_loss": 1.1163, "bpb": 1.61049, "chunk_loss": 0.030466, "F": 0.16229, "G": 0.17996, "chunks": 166.19, "bytes_per_chunk": 6.162, "boundary_entropy": 0.12099, "router_noise": 0.0, "grad_norm": 0.3954, "lr": 0.0001549, "router_trainable": true, "sec_per_step": 0.284, "n_target_mean": 9.0576, "alpha": 0.03, "t": 1350.9}
{"event": "train", "step": 6100, "phase": 2, "objective": "parity_a", "lm_loss": 1.04954, "bpb": 1.51416, "chunk_loss": 0.030647, "F": 0.16412, "G": 0.17957, "chunks": 168.06, "bytes_per_chunk": 6.093, "boundary_entropy": 0.13003, "router_noise": 0.0, "grad_norm": 0.373, "lr": 0.0001509, "router_trainable": true, "sec_per_step": 0.189, "n_target_mean": 9.3786, "alpha": 0.03, "t": 1369.8}
{"event": "train", "step": 6200, "phase": 2, "objective": "parity_a", "lm_loss": 1.16059, "bpb": 1.67437, "chunk_loss": 0.030641, "F": 0.174, "G": 0.18074, "chunks": 178.17, "bytes_per_chunk": 5.747, "boundary_entropy": 0.10219, "router_noise": 0.0, "grad_norm": 0.372, "lr": 0.0001469, "router_trainable": true, "sec_per_step": 0.189, "n_target_mean": 8.4082, "alpha": 0.03, "t": 1388.7}
{"event": "train", "step": 6300, "phase": 2, "objective": "parity_a", "lm_loss": 1.08486, "bpb": 1.56512, "chunk_loss": 0.030556, "F": 0.17229, "G": 0.17481, "chunks": 176.42, "bytes_per_chunk": 5.804, "boundary_entropy": 0.0925, "router_noise": 0.0, "grad_norm": 0.3691, "lr": 0.0001429, "router_trainable": true, "sec_per_step": 0.188, "n_target_mean": 8.5209, "alpha": 0.03, "t": 1407.5}
{"event": "train", "step": 6400, "phase": 2, "objective": "parity_a", "lm_loss": 0.99625, "bpb": 1.43729, "chunk_loss": 0.030601, "F": 0.14928, "G": 0.16845, "chunks": 152.86, "bytes_per_chunk": 6.699, "boundary_entropy": 0.14004, "router_noise": 0.0, "grad_norm": 0.37, "lr": 0.0001389, "router_trainable": true, "sec_per_step": 0.189, "n_target_mean": 10.5328, "alpha": 0.03, "t": 1426.4}
{"event": "train", "step": 6500, "phase": 2, "objective": "parity_a", "lm_loss": 1.05765, "bpb": 1.52587, "chunk_loss": 0.030332, "F": 0.15073, "G": 0.16901, "chunks": 154.34, "bytes_per_chunk": 6.635, "boundary_entropy": 0.12181, "router_noise": 0.0, "grad_norm": 0.3805, "lr": 0.000135, "router_trainable": true, "sec_per_step": 0.187, "n_target_mean": 9.6646, "alpha": 0.03, "t": 1445.1}
{"event": "train", "step": 6600, "phase": 2, "objective": "parity_a", "lm_loss": 1.10656, "bpb": 1.59643, "chunk_loss": 0.030482, "F": 0.17355, "G": 0.18085, "chunks": 177.72, "bytes_per_chunk": 5.762, "boundary_entropy": 0.1043, "router_noise": 0.0, "grad_norm": 0.3658, "lr": 0.000131, "router_trainable": true, "sec_per_step": 0.188, "n_target_mean": 8.5192, "alpha": 0.03, "t": 1463.9}
{"event": "train", "step": 6700, "phase": 2, "objective": "parity_a", "lm_loss": 1.09848, "bpb": 1.58478, "chunk_loss": 0.030317, "F": 0.16196, "G": 0.17374, "chunks": 165.84, "bytes_per_chunk": 6.174, "boundary_entropy": 0.11053, "router_noise": 0.0, "grad_norm": 0.3795, "lr": 0.0001271, "router_trainable": true, "sec_per_step": 0.198, "n_target_mean": 8.8379, "alpha": 0.03, "t": 1483.7}
{"event": "train", "step": 6800, "phase": 2, "objective": "parity_a", "lm_loss": 1.10757, "bpb": 1.59788, "chunk_loss": 0.030589, "F": 0.15659, "G": 0.17177, "chunks": 160.34, "bytes_per_chunk": 6.386, "boundary_entropy": 0.1253, "router_noise": 0.0, "grad_norm": 0.3941, "lr": 0.0001233, "router_trainable": true, "sec_per_step": 0.178, "n_target_mean": 9.7731, "alpha": 0.03, "t": 1501.4}
{"event": "train", "step": 6900, "phase": 2, "objective": "parity_a", "lm_loss": 1.086, "bpb": 1.56676, "chunk_loss": 0.030574, "F": 0.16093, "G": 0.16997, "chunks": 164.8, "bytes_per_chunk": 6.214, "boundary_entropy": 0.10967, "router_noise": 0.0, "grad_norm": 0.3861, "lr": 0.0001195, "router_trainable": true, "sec_per_step": 0.174, "n_target_mean": 9.4099, "alpha": 0.03, "t": 1518.9}
{"event": "probe", "step": 7000, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 3425.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.196805, "cps": 26.757812, "mean_p": 0.191975, "boundary_entropy": 0.056317, "bpb": 1.692761}, "de": {"chunks": 3805.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.185311, "cps": 29.726562, "mean_p": 0.182968, "boundary_entropy": 0.048905, "bpb": 1.83676}, "vi": {"chunks": 3845.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.208571, "cps": 30.039062, "mean_p": 0.158632, "boundary_entropy": 0.05778, "bpb": 1.501972}, "tr": {"chunks": 3015.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.166547, "cps": 23.554688, "mean_p": 0.158386, "boundary_entropy": 0.065985, "bpb": 1.771223}, "fi": {"chunks": 3154.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.167668, "cps": 24.640625, "mean_p": 0.165091, "boundary_entropy": 0.065037, "bpb": 1.943041}, "ru": {"chunks": 5852.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.308813, "cps": 45.71875, "mean_p": 0.142563, "boundary_entropy": 0.118305, "bpb": 1.116165}, "el": {"chunks": 5519.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.268918, "cps": 43.117188, "mean_p": 0.133753, "boundary_entropy": 0.109517, "bpb": 1.138237}, "hi": {"chunks": 2649.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.157444, "cps": 20.695312, "mean_p": 0.102087, "boundary_entropy": 0.195296, "bpb": 0.96241}, "ta": {"chunks": 2596.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.138861, "cps": 20.28125, "mean_p": 0.147269, "boundary_entropy": 0.30608, "bpb": 0.830843}, "th": {"chunks": 3203.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.19913, "cps": 25.023438, "mean_p": 0.192133, "boundary_entropy": 0.374094, "bpb": 0.925037}, "zh": {"chunks": 2978.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.504233, "cps": 23.265625, "mean_p": 0.202385, "boundary_entropy": 0.098549, "bpb": 2.590515}, "my": {"chunks": 2810.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.148301, "cps": 21.953125, "mean_p": 0.141457, "boundary_entropy": 0.311275, "bpb": 0.953208}}, "bpb": {"en": 1.69276, "de": 1.83676, "vi": 1.50197, "tr": 1.77122, "fi": 1.94304, "ru": 1.11616, "el": 1.13824, "hi": 0.96241, "ta": 0.83084, "th": 0.92504, "zh": 2.59051, "my": 0.95321}, "gini_cpc": 0.206243, "gini_cps": 0.144503, "bpic": 8.956174, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9415985874261158, "vi": 1.0597825066964417, "tr": 0.8462531717330394, "fi": 0.8519485143612585, "ru": 1.56912899485777, "el": 1.3664164862566512, "hi": 0.8000008156093751, "ta": 0.7055743153151922, "th": 1.011811049420392, "zh": 2.5620924409421573, "my": 0.753540307780495}, "t": 1540.5}
{"event": "eval", "step": 7000, "phase": 2, "bpb": {"en": 1.69245, "de": 1.84126, "vi": 1.51835, "tr": 1.764, "fi": 1.95923, "ru": 1.11435, "el": 1.13641, "hi": 0.96788, "ta": 0.83557, "th": 0.92449, "zh": 2.59789, "my": 0.9607}, "bytes_per_chunk": {"en": 4.928, "de": 5.539, "vi": 6.287, "tr": 6.476, "fi": 5.932, "ru": 6.093, "el": 6.849, "hi": 16.092, "ta": 19.175, "th": 14.451, "zh": 5.292, "my": 20.388}, "bpb_mean": 1.44271, "t": 1545.6}
{"event": "train", "step": 7000, "phase": 2, "objective": "parity_a", "lm_loss": 1.07366, "bpb": 1.54896, "chunk_loss": 0.030785, "F": 0.16322, "G": 0.17347, "chunks": 167.14, "bytes_per_chunk": 6.127, "boundary_entropy": 0.11579, "router_noise": 0.0, "grad_norm": 0.3959, "lr": 0.0001157, "router_trainable": true, "sec_per_step": 0.269, "n_target_mean": 9.5778, "alpha": 0.03, "t": 1545.8}
{"event": "train", "step": 7100, "phase": 2, "objective": "parity_a", "lm_loss": 1.0579, "bpb": 1.52623, "chunk_loss": 0.030449, "F": 0.15649, "G": 0.17278, "chunks": 160.25, "bytes_per_chunk": 6.39, "boundary_entropy": 0.11369, "router_noise": 0.0, "grad_norm": 0.3807, "lr": 0.0001119, "router_trainable": true, "sec_per_step": 0.177, "n_target_mean": 9.4019, "alpha": 0.03, "t": 1563.4}
{"event": "train", "step": 7200, "phase": 2, "objective": "parity_a", "lm_loss": 1.10623, "bpb": 1.59595, "chunk_loss": 0.030465, "F": 0.16316, "G": 0.16881, "chunks": 167.08, "bytes_per_chunk": 6.129, "boundary_entropy": 0.09964, "router_noise": 0.0, "grad_norm": 0.3733, "lr": 0.0001083, "router_trainable": true, "sec_per_step": 0.178, "n_target_mean": 8.9796, "alpha": 0.03, "t": 1581.2}
{"event": "train", "step": 7300, "phase": 2, "objective": "parity_a", "lm_loss": 1.12191, "bpb": 1.61857, "chunk_loss": 0.030445, "F": 0.16252, "G": 0.17352, "chunks": 166.42, "bytes_per_chunk": 6.153, "boundary_entropy": 0.10668, "router_noise": 0.0, "grad_norm": 0.3941, "lr": 0.0001046, "router_trainable": true, "sec_per_step": 0.186, "n_target_mean": 9.1966, "alpha": 0.03, "t": 1599.8}
{"event": "train", "step": 7400, "phase": 2, "objective": "parity_a", "lm_loss": 1.08596, "bpb": 1.5667, "chunk_loss": 0.030537, "F": 0.1646, "G": 0.17384, "chunks": 168.55, "bytes_per_chunk": 6.075, "boundary_entropy": 0.10958, "router_noise": 0.0, "grad_norm": 0.3823, "lr": 0.000101, "router_trainable": true, "sec_per_step": 0.178, "n_target_mean": 9.204, "alpha": 0.03, "t": 1617.7}
{"event": "train", "step": 7500, "phase": 2, "objective": "parity_a", "lm_loss": 1.06554, "bpb": 1.53725, "chunk_loss": 0.030502, "F": 0.16014, "G": 0.16992, "chunks": 163.98, "bytes_per_chunk": 6.244, "boundary_entropy": 0.11241, "router_noise": 0.0, "grad_norm": 0.375, "lr": 9.75e-05, "router_trainable": true, "sec_per_step": 0.188, "n_target_mean": 9.4226, "alpha": 0.03, "t": 1636.4}
{"event": "train", "step": 7600, "phase": 2, "objective": "parity_a", "lm_loss": 1.13, "bpb": 1.63025, "chunk_loss": 0.030471, "F": 0.16713, "G": 0.17779, "chunks": 171.14, "bytes_per_chunk": 5.983, "boundary_entropy": 0.10593, "router_noise": 0.0, "grad_norm": 0.3902, "lr": 9.4e-05, "router_trainable": true, "sec_per_step": 0.176, "n_target_mean": 8.7726, "alpha": 0.03, "t": 1654.0}
{"event": "phase_transition", "step": 7700, "phase": 3, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0, "t": 1672.3}
{"event": "train", "step": 7700, "phase": 3, "objective": "global", "lm_loss": 1.07742, "bpb": 1.55439, "chunk_loss": 0.030219, "F": 0.16516, "G": 0.17019, "chunks": 169.12, "bytes_per_chunk": 6.055, "boundary_entropy": 0.09667, "router_noise": 0.0, "grad_norm": 0.3884, "lr": 9.06e-05, "router_trainable": true, "sec_per_step": 0.184, "alpha": 0.03, "t": 1672.5}
{"event": "train", "step": 7800, "phase": 3, "objective": "global", "lm_loss": 1.11133, "bpb": 1.6033, "chunk_loss": 0.030116, "F": 0.17941, "G": 0.18695, "chunks": 183.72, "bytes_per_chunk": 5.574, "boundary_entropy": 0.09776, "router_noise": 0.0, "grad_norm": 0.3784, "lr": 8.73e-05, "router_trainable": true, "sec_per_step": 0.186, "alpha": 0.03, "t": 1691.1}
{"event": "train", "step": 7900, "phase": 3, "objective": "global", "lm_loss": 1.10236, "bpb": 1.59037, "chunk_loss": 0.030083, "F": 0.1781, "G": 0.18932, "chunks": 182.38, "bytes_per_chunk": 5.615, "boundary_entropy": 0.10797, "router_noise": 0.0, "grad_norm": 0.3942, "lr": 8.4e-05, "router_trainable": true, "sec_per_step": 0.175, "alpha": 0.03, "t": 1708.6}
{"event": "probe", "step": 8000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3419.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.19646, "cps": 26.710938, "mean_p": 0.191112, "boundary_entropy": 0.05627, "bpb": 1.666024}, "de": {"chunks": 3806.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.18536, "cps": 29.734375, "mean_p": 0.183629, "boundary_entropy": 0.045154, "bpb": 1.804297}, "vi": {"chunks": 3866.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.20971, "cps": 30.203125, "mean_p": 0.163217, "boundary_entropy": 0.055889, "bpb": 1.466285}, "tr": {"chunks": 3208.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.177208, "cps": 25.0625, "mean_p": 0.167226, "boundary_entropy": 0.066413, "bpb": 1.737256}, "fi": {"chunks": 3205.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.170379, "cps": 25.039062, "mean_p": 0.168378, "boundary_entropy": 0.060512, "bpb": 1.905183}, "ru": {"chunks": 5956.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.314301, "cps": 46.53125, "mean_p": 0.15622, "boundary_entropy": 0.110738, "bpb": 1.083502}, "el": {"chunks": 5643.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.27496, "cps": 44.085938, "mean_p": 0.146378, "boundary_entropy": 0.102175, "bpb": 1.105341}, "hi": {"chunks": 6217.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.36951, "cps": 48.570312, "mean_p": 0.232655, "boundary_entropy": 0.371189, "bpb": 0.936614}, "ta": {"chunks": 7345.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.392886, "cps": 57.382812, "mean_p": 0.285671, "boundary_entropy": 0.4689, "bpb": 0.80665}, "th": {"chunks": 7361.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.457631, "cps": 57.507812, "mean_p": 0.294324, "boundary_entropy": 0.477872, "bpb": 0.905782}, "zh": {"chunks": 2949.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.499323, "cps": 23.039062, "mean_p": 0.199706, "boundary_entropy": 0.089051, "bpb": 2.542546}, "my": {"chunks": 7740.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.408486, "cps": 60.46875, "mean_p": 0.266283, "boundary_entropy": 0.459276, "bpb": 0.930373}}, "bpb": {"en": 1.66602, "de": 1.8043, "vi": 1.46628, "tr": 1.73726, "fi": 1.90518, "ru": 1.0835, "el": 1.10534, "hi": 0.93661, "ta": 0.80665, "th": 0.90578, "zh": 2.54255, "my": 0.93037}, "gini_cpc": 0.209801, "gini_cps": 0.193913, "bpic": 6.321024, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9434988956989987, "vi": 1.0674406246050943, "tr": 0.9020047562809523, "fi": 0.8672437323105477, "ru": 1.5998176880554964, "el": 1.3995687102252694, "hi": 1.8808355023300871, "ta": 1.999822038181454, "th": 2.3293823040674075, "zh": 2.5415950109282535, "my": 2.0792303413421367}, "t": 1729.9}
{"event": "eval", "step": 8000, "phase": 3, "bpb": {"en": 1.66502, "de": 1.80956, "vi": 1.48223, "tr": 1.72846, "fi": 1.91963, "ru": 1.08192, "el": 1.10422, "hi": 0.94232, "ta": 0.81138, "th": 0.90614, "zh": 2.54922, "my": 0.93742}, "bytes_per_chunk": {"en": 4.94, "de": 5.539, "vi": 6.265, "tr": 6.181, "fi": 5.888, "ru": 5.987, "el": 6.695, "hi": 6.59, "ta": 6.665, "th": 6.53, "zh": 5.321, "my": 7.059}, "bpb_mean": 1.41146, "t": 1735.3}
{"event": "train", "step": 8000, "phase": 3, "objective": "global", "lm_loss": 1.08787, "bpb": 1.56947, "chunk_loss": 0.03009, "F": 0.17918, "G": 0.19462, "chunks": 183.48, "bytes_per_chunk": 5.581, "boundary_entropy": 0.11928, "router_noise": 0.0, "grad_norm": 0.3848, "lr": 8.08e-05, "router_trainable": true, "sec_per_step": 0.269, "alpha": 0.03, "t": 1735.5}
{"event": "train", "step": 8100, "phase": 3, "objective": "global", "lm_loss": 1.11017, "bpb": 1.60163, "chunk_loss": 0.030148, "F": 0.17831, "G": 0.19003, "chunks": 182.59, "bytes_per_chunk": 5.608, "boundary_entropy": 0.10778, "router_noise": 0.0, "grad_norm": 0.409, "lr": 7.77e-05, "router_trainable": true, "sec_per_step": 0.184, "alpha": 0.03, "t": 1753.8}
{"event": "train", "step": 8200, "phase": 3, "objective": "global", "lm_loss": 1.07364, "bpb": 1.54893, "chunk_loss": 0.030151, "F": 0.17949, "G": 0.19043, "chunks": 183.8, "bytes_per_chunk": 5.571, "boundary_entropy": 0.11105, "router_noise": 0.0, "grad_norm": 0.3853, "lr": 7.47e-05, "router_trainable": true, "sec_per_step": 0.188, "alpha": 0.03, "t": 1772.6}
{"event": "train", "step": 8300, "phase": 3, "objective": "global", "lm_loss": 1.00633, "bpb": 1.45183, "chunk_loss": 0.030095, "F": 0.18134, "G": 0.20221, "chunks": 185.69, "bytes_per_chunk": 5.515, "boundary_entropy": 0.14589, "router_noise": 0.0, "grad_norm": 0.3732, "lr": 7.17e-05, "router_trainable": true, "sec_per_step": 0.183, "alpha": 0.03, "t": 1790.9}
{"event": "train", "step": 8400, "phase": 3, "objective": "global", "lm_loss": 1.05148, "bpb": 1.51697, "chunk_loss": 0.030087, "F": 0.18126, "G": 0.19798, "chunks": 185.61, "bytes_per_chunk": 5.517, "boundary_entropy": 0.12913, "router_noise": 0.0, "grad_norm": 0.379, "lr": 6.88e-05, "router_trainable": true, "sec_per_step": 0.175, "alpha": 0.03, "t": 1808.5}
{"event": "train", "step": 8500, "phase": 3, "objective": "global", "lm_loss": 1.06632, "bpb": 1.53837, "chunk_loss": 0.030115, "F": 0.17851, "G": 0.19168, "chunks": 182.8, "bytes_per_chunk": 5.602, "boundary_entropy": 0.11921, "router_noise": 0.0, "grad_norm": 0.3801, "lr": 6.6e-05, "router_trainable": true, "sec_per_step": 0.188, "alpha": 0.03, "t": 1827.2}
{"event": "train", "step": 8600, "phase": 3, "objective": "global", "lm_loss": 1.11923, "bpb": 1.61471, "chunk_loss": 0.030117, "F": 0.17657, "G": 0.19485, "chunks": 180.81, "bytes_per_chunk": 5.663, "boundary_entropy": 0.12223, "router_noise": 0.0, "grad_norm": 0.4119, "lr": 6.33e-05, "router_trainable": true, "sec_per_step": 0.187, "alpha": 0.03, "t": 1845.9}
{"event": "train", "step": 8700, "phase": 3, "objective": "global", "lm_loss": 0.98791, "bpb": 1.42526, "chunk_loss": 0.030093, "F": 0.1808, "G": 0.205, "chunks": 185.14, "bytes_per_chunk": 5.531, "boundary_entropy": 0.14818, "router_noise": 0.0, "grad_norm": 0.3886, "lr": 6.07e-05, "router_trainable": true, "sec_per_step": 0.184, "alpha": 0.03, "t": 1864.3}
{"event": "train", "step": 8800, "phase": 3, "objective": "global", "lm_loss": 1.007, "bpb": 1.45279, "chunk_loss": 0.030148, "F": 0.18088, "G": 0.20064, "chunks": 185.22, "bytes_per_chunk": 5.529, "boundary_entropy": 0.13543, "router_noise": 0.0, "grad_norm": 0.3744, "lr": 5.82e-05, "router_trainable": true, "sec_per_step": 0.175, "alpha": 0.03, "t": 1881.8}
{"event": "train", "step": 8900, "phase": 3, "objective": "global", "lm_loss": 1.06897, "bpb": 1.5422, "chunk_loss": 0.030135, "F": 0.18259, "G": 0.19041, "chunks": 186.97, "bytes_per_chunk": 5.477, "boundary_entropy": 0.09737, "router_noise": 0.0, "grad_norm": 0.3773, "lr": 5.58e-05, "router_trainable": true, "sec_per_step": 0.174, "alpha": 0.03, "t": 1899.2}
{"event": "probe", "step": 9000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3424.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.196748, "cps": 26.75, "mean_p": 0.191698, "boundary_entropy": 0.052334, "bpb": 1.644613}, "de": {"chunks": 3831.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.186578, "cps": 29.929688, "mean_p": 0.184626, "boundary_entropy": 0.041138, "bpb": 1.780822}, "vi": {"chunks": 3962.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.214917, "cps": 30.953125, "mean_p": 0.171132, "boundary_entropy": 0.063045, "bpb": 1.437502}, "tr": {"chunks": 3270.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.180633, "cps": 25.546875, "mean_p": 0.17071, "boundary_entropy": 0.069337, "bpb": 1.712192}, "fi": {"chunks": 3244.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.172452, "cps": 25.34375, "mean_p": 0.169817, "boundary_entropy": 0.061352, "bpb": 1.876669}, "ru": {"chunks": 6064.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.32, "cps": 47.375, "mean_p": 0.163202, "boundary_entropy": 0.109918, "bpb": 1.063316}, "el": {"chunks": 5782.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.281733, "cps": 45.171875, "mean_p": 0.155863, "boundary_entropy": 0.112236, "bpb": 1.08194}, "hi": {"chunks": 6854.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.40737, "cps": 53.546875, "mean_p": 0.275612, "boundary_entropy": 0.428818, "bpb": 0.912753}, "ta": {"chunks": 8032.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.429634, "cps": 62.75, "mean_p": 0.304764, "boundary_entropy": 0.493315, "bpb": 0.790887}, "th": {"chunks": 7435.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.462232, "cps": 58.085938, "mean_p": 0.318909, "boundary_entropy": 0.509234, "bpb": 0.888873}, "zh": {"chunks": 2943.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.498307, "cps": 22.992188, "mean_p": 0.19656, "boundary_entropy": 0.079855, "bpb": 2.512327}, "my": {"chunks": 8368.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.44163, "cps": 65.375, "mean_p": 0.29881, "boundary_entropy": 0.49792, "bpb": 0.907086}}, "bpb": {"en": 1.64461, "de": 1.78082, "vi": 1.4375, "tr": 1.71219, "fi": 1.87667, "ru": 1.06332, "el": 1.08194, "hi": 0.91275, "ta": 0.79089, "th": 0.88887, "zh": 2.51233, "my": 0.90709}, "gini_cpc": 0.212808, "gini_cps": 0.207764, "bpic": 6.07162, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9483095169116867, "vi": 1.092349699246405, "tr": 0.918094879972907, "fi": 0.876514946762607, "ru": 1.6264485981308412, "el": 1.4319491842068233, "hi": 2.0705198164169363, "ta": 2.1836779787688747, "th": 2.3493637952873025, "zh": 2.532720022169756, "my": 2.2446501393878773}, "t": 1920.7}
{"event": "eval", "step": 9000, "phase": 3, "bpb": {"en": 1.64422, "de": 1.78524, "vi": 1.45377, "tr": 1.70352, "fi": 1.89076, "ru": 1.06103, "el": 1.08016, "hi": 0.91819, "ta": 0.79643, "th": 0.88966, "zh": 2.51981, "my": 0.91528}, "bytes_per_chunk": {"en": 4.937, "de": 5.528, "vi": 6.136, "tr": 6.088, "fi": 5.84, "ru": 5.877, "el": 6.528, "hi": 6.092, "ta": 6.159, "th": 6.339, "zh": 5.338, "my": 6.5}, "bpb_mean": 1.38817, "t": 1926.2}
{"event": "train", "step": 9000, "phase": 3, "objective": "global", "lm_loss": 1.05825, "bpb": 1.52673, "chunk_loss": 0.030061, "F": 0.17477, "G": 0.19688, "chunks": 178.97, "bytes_per_chunk": 5.722, "boundary_entropy": 0.13329, "router_noise": 0.0, "grad_norm": 0.4092, "lr": 5.35e-05, "router_trainable": true, "sec_per_step": 0.272, "alpha": 0.03, "t": 1926.4}
{"event": "train", "step": 9100, "phase": 3, "objective": "global", "lm_loss": 1.04392, "bpb": 1.50606, "chunk_loss": 0.03011, "F": 0.18065, "G": 0.19874, "chunks": 184.98, "bytes_per_chunk": 5.536, "boundary_entropy": 0.12918, "router_noise": 0.0, "grad_norm": 0.3846, "lr": 5.12e-05, "router_trainable": true, "sec_per_step": 0.178, "alpha": 0.03, "t": 1944.2}
{"event": "train", "step": 9200, "phase": 3, "objective": "global", "lm_loss": 0.9684, "bpb": 1.39711, "chunk_loss": 0.030044, "F": 0.17392, "G": 0.19536, "chunks": 178.09, "bytes_per_chunk": 5.75, "boundary_entropy": 0.14018, "router_noise": 0.0, "grad_norm": 0.3634, "lr": 4.91e-05, "router_trainable": true, "sec_per_step": 0.186, "alpha": 0.03, "t": 1962.8}
{"event": "train", "step": 9300, "phase": 3, "objective": "global", "lm_loss": 1.03962, "bpb": 1.49985, "chunk_loss": 0.030115, "F": 0.18423, "G": 0.20232, "chunks": 188.66, "bytes_per_chunk": 5.428, "boundary_entropy": 0.12102, "router_noise": 0.0, "grad_norm": 0.383, "lr": 4.71e-05, "router_trainable": true, "sec_per_step": 0.176, "alpha": 0.03, "t": 1980.4}
{"event": "train", "step": 9400, "phase": 3, "objective": "global", "lm_loss": 1.08657, "bpb": 1.56759, "chunk_loss": 0.030167, "F": 0.18564, "G": 0.20736, "chunks": 190.09, "bytes_per_chunk": 5.387, "boundary_entropy": 0.124, "router_noise": 0.0, "grad_norm": 0.3924, "lr": 4.52e-05, "router_trainable": true, "sec_per_step": 0.187, "alpha": 0.03, "t": 1999.1}
{"event": "train", "step": 9500, "phase": 3, "objective": "global", "lm_loss": 1.09496, "bpb": 1.57969, "chunk_loss": 0.03014, "F": 0.18283, "G": 0.19728, "chunks": 187.22, "bytes_per_chunk": 5.47, "boundary_entropy": 0.11168, "router_noise": 0.0, "grad_norm": 0.3947, "lr": 4.34e-05, "router_trainable": true, "sec_per_step": 0.175, "alpha": 0.03, "t": 2016.7}
{"event": "train", "step": 9600, "phase": 3, "objective": "global", "lm_loss": 0.9581, "bpb": 1.38225, "chunk_loss": 0.03011, "F": 0.18037, "G": 0.21086, "chunks": 184.7, "bytes_per_chunk": 5.544, "boundary_entropy": 0.1627, "router_noise": 0.0, "grad_norm": 0.3998, "lr": 4.17e-05, "router_trainable": true, "sec_per_step": 0.185, "alpha": 0.03, "t": 2035.1}
{"event": "train", "step": 9700, "phase": 3, "objective": "global", "lm_loss": 0.97264, "bpb": 1.40322, "chunk_loss": 0.030076, "F": 0.17989, "G": 0.20091, "chunks": 184.2, "bytes_per_chunk": 5.559, "boundary_entropy": 0.13317, "router_noise": 0.0, "grad_norm": 0.3943, "lr": 4.01e-05, "router_trainable": true, "sec_per_step": 0.176, "alpha": 0.03, "t": 2052.8}
{"event": "train", "step": 9800, "phase": 3, "objective": "global", "lm_loss": 0.99654, "bpb": 1.4377, "chunk_loss": 0.03017, "F": 0.1815, "G": 0.21541, "chunks": 185.86, "bytes_per_chunk": 5.51, "boundary_entropy": 0.17052, "router_noise": 0.0, "grad_norm": 0.3872, "lr": 3.86e-05, "router_trainable": true, "sec_per_step": 0.176, "alpha": 0.03, "t": 2070.3}
{"event": "train", "step": 9900, "phase": 3, "objective": "global", "lm_loss": 1.06715, "bpb": 1.53958, "chunk_loss": 0.030136, "F": 0.18384, "G": 0.20722, "chunks": 188.25, "bytes_per_chunk": 5.44, "boundary_entropy": 0.12696, "router_noise": 0.0, "grad_norm": 0.3963, "lr": 3.72e-05, "router_trainable": true, "sec_per_step": 0.183, "alpha": 0.03, "t": 2088.6}
{"event": "probe", "step": 10000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3425.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.196805, "cps": 26.757812, "mean_p": 0.190601, "boundary_entropy": 0.055617, "bpb": 1.630314}, "de": {"chunks": 3825.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.186285, "cps": 29.882812, "mean_p": 0.184665, "boundary_entropy": 0.042037, "bpb": 1.763847}, "vi": {"chunks": 3986.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.216219, "cps": 31.140625, "mean_p": 0.171931, "boundary_entropy": 0.065056, "bpb": 1.420929}, "tr": {"chunks": 3279.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.18113, "cps": 25.617188, "mean_p": 0.17143, "boundary_entropy": 0.071481, "bpb": 1.695881}, "fi": {"chunks": 3231.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.171761, "cps": 25.242188, "mean_p": 0.169836, "boundary_entropy": 0.062993, "bpb": 1.853819}, "ru": {"chunks": 6076.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.320633, "cps": 47.46875, "mean_p": 0.164963, "boundary_entropy": 0.106048, "bpb": 1.049609}, "el": {"chunks": 5796.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.282415, "cps": 45.28125, "mean_p": 0.158492, "boundary_entropy": 0.119868, "bpb": 1.065736}, "hi": {"chunks": 6925.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.41159, "cps": 54.101562, "mean_p": 0.289636, "boundary_entropy": 0.451662, "bpb": 0.896789}, "ta": {"chunks": 7734.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.413694, "cps": 60.421875, "mean_p": 0.319008, "boundary_entropy": 0.514644, "bpb": 0.776824}, "th": {"chunks": 7412.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.460802, "cps": 57.90625, "mean_p": 0.323879, "boundary_entropy": 0.518368, "bpb": 0.877981}, "zh": {"chunks": 2953.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.5, "cps": 23.070312, "mean_p": 0.195299, "boundary_entropy": 0.075055, "bpb": 2.49282}, "my": {"chunks": 9388.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.495461, "cps": 73.34375, "mean_p": 0.323061, "boundary_entropy": 0.525848, "bpb": 0.893131}}, "bpb": {"en": 1.63031, "de": 1.76385, "vi": 1.42093, "tr": 1.69588, "fi": 1.85382, "ru": 1.04961, "el": 1.06574, "hi": 0.89679, "ta": 0.77682, "th": 0.87798, "zh": 2.49282, "my": 0.89313}, "gini_cpc": 0.218715, "gini_cps": 0.216443, "bpic": 5.993769, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9465478572680402, "vi": 1.0986457923776376, "tr": 0.9203529519444894, "fi": 0.8727475110657027, "ru": 1.629191348727924, "el": 1.4349972738437309, "hi": 2.0913573605492344, "ta": 2.1020461304498057, "th": 2.341412269217592, "zh": 2.5405839416058393, "my": 2.517521853894408}, "t": 2109.0}
{"event": "eval", "step": 10000, "phase": 3, "bpb": {"en": 1.62912, "de": 1.7691, "vi": 1.43724, "tr": 1.68751, "fi": 1.86887, "ru": 1.04753, "el": 1.06465, "hi": 0.90255, "ta": 0.78267, "th": 0.87788, "zh": 2.50102, "my": 0.90116}, "bytes_per_chunk": {"en": 4.936, "de": 5.519, "vi": 6.119, "tr": 6.051, "fi": 5.853, "ru": 5.852, "el": 6.536, "hi": 6.124, "ta": 6.454, "th": 6.281, "zh": 5.338, "my": 5.929}, "bpb_mean": 1.37244, "t": 2114.4}
{"event": "train", "step": 10000, "phase": 3, "objective": "global", "lm_loss": 1.03688, "bpb": 1.4959, "chunk_loss": 0.030165, "F": 0.18153, "G": 0.20442, "chunks": 185.89, "bytes_per_chunk": 5.509, "boundary_entropy": 0.14196, "router_noise": 0.0, "grad_norm": 0.4058, "lr": 3.6e-05, "router_trainable": true, "sec_per_step": 0.26, "alpha": 0.03, "t": 2114.6}
{"event": "train", "step": 10100, "phase": 3, "objective": "global", "lm_loss": 0.98828, "bpb": 1.42578, "chunk_loss": 0.030086, "F": 0.17807, "G": 0.21907, "chunks": 182.34, "bytes_per_chunk": 5.616, "boundary_entropy": 0.18933, "router_noise": 0.0, "grad_norm": 0.4009, "lr": 3.49e-05, "router_trainable": true, "sec_per_step": 0.189, "alpha": 0.03, "t": 2133.4}
{"event": "train", "step": 10200, "phase": 3, "objective": "global", "lm_loss": 1.07274, "bpb": 1.54764, "chunk_loss": 0.030167, "F": 0.18495, "G": 0.19621, "chunks": 189.39, "bytes_per_chunk": 5.407, "boundary_entropy": 0.1028, "router_noise": 0.0, "grad_norm": 0.4118, "lr": 3.38e-05, "router_trainable": true, "sec_per_step": 0.176, "alpha": 0.03, "t": 2151.1}
{"event": "train", "step": 10300, "phase": 3, "objective": "global", "lm_loss": 1.03027, "bpb": 1.48636, "chunk_loss": 0.030075, "F": 0.17787, "G": 0.20323, "chunks": 182.14, "bytes_per_chunk": 5.622, "boundary_entropy": 0.14588, "router_noise": 0.0, "grad_norm": 0.4301, "lr": 3.3e-05, "router_trainable": true, "sec_per_step": 0.185, "alpha": 0.03, "t": 2169.5}
{"event": "train", "step": 10400, "phase": 3, "objective": "global", "lm_loss": 1.00668, "bpb": 1.45234, "chunk_loss": 0.030091, "F": 0.17999, "G": 0.20089, "chunks": 184.31, "bytes_per_chunk": 5.556, "boundary_entropy": 0.13302, "router_noise": 0.0, "grad_norm": 0.4001, "lr": 3.22e-05, "router_trainable": true, "sec_per_step": 0.175, "alpha": 0.03, "t": 2187.1}
{"event": "train", "step": 10500, "phase": 3, "objective": "global", "lm_loss": 1.02885, "bpb": 1.48431, "chunk_loss": 0.030113, "F": 0.18294, "G": 0.20574, "chunks": 187.33, "bytes_per_chunk": 5.466, "boundary_entropy": 0.13323, "router_noise": 0.0, "grad_norm": 0.4128, "lr": 3.15e-05, "router_trainable": true, "sec_per_step": 0.176, "alpha": 0.03, "t": 2204.6}
{"event": "train", "step": 10600, "phase": 3, "objective": "global", "lm_loss": 1.05431, "bpb": 1.52104, "chunk_loss": 0.030121, "F": 0.18216, "G": 0.2044, "chunks": 186.53, "bytes_per_chunk": 5.49, "boundary_entropy": 0.13185, "router_noise": 0.0, "grad_norm": 0.4005, "lr": 3.1e-05, "router_trainable": true, "sec_per_step": 0.189, "alpha": 0.03, "t": 2223.6}
{"event": "train", "step": 10700, "phase": 3, "objective": "global", "lm_loss": 1.02031, "bpb": 1.472, "chunk_loss": 0.030127, "F": 0.18201, "G": 0.21168, "chunks": 186.38, "bytes_per_chunk": 5.494, "boundary_entropy": 0.15504, "router_noise": 0.0, "grad_norm": 0.4137, "lr": 3.05e-05, "router_trainable": true, "sec_per_step": 0.175, "alpha": 0.03, "t": 2241.1}
{"event": "train", "step": 10800, "phase": 3, "objective": "global", "lm_loss": 1.04225, "bpb": 1.50365, "chunk_loss": 0.030136, "F": 0.18338, "G": 0.2035, "chunks": 187.78, "bytes_per_chunk": 5.453, "boundary_entropy": 0.12894, "router_noise": 0.0, "grad_norm": 0.3918, "lr": 3.02e-05, "router_trainable": true, "sec_per_step": 0.191, "alpha": 0.03, "t": 2260.1}
{"event": "train", "step": 10900, "phase": 3, "objective": "global", "lm_loss": 1.02618, "bpb": 1.48046, "chunk_loss": 0.030089, "F": 0.17723, "G": 0.19855, "chunks": 181.48, "bytes_per_chunk": 5.642, "boundary_entropy": 0.1389, "router_noise": 0.0, "grad_norm": 0.4085, "lr": 3.01e-05, "router_trainable": true, "sec_per_step": 0.173, "alpha": 0.03, "t": 2277.4}
{"event": "probe", "step": 11000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3430.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.197092, "cps": 26.796875, "mean_p": 0.191548, "boundary_entropy": 0.053056, "bpb": 1.622865}, "de": {"chunks": 3831.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.186578, "cps": 29.929688, "mean_p": 0.185125, "boundary_entropy": 0.040192, "bpb": 1.753883}, "vi": {"chunks": 3975.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.215622, "cps": 31.054688, "mean_p": 0.172111, "boundary_entropy": 0.065071, "bpb": 1.410475}, "tr": {"chunks": 3313.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.183008, "cps": 25.882812, "mean_p": 0.17348, "boundary_entropy": 0.072507, "bpb": 1.685822}, "fi": {"chunks": 3255.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.173037, "cps": 25.429688, "mean_p": 0.170165, "boundary_entropy": 0.061665, "bpb": 1.842762}, "ru": {"chunks": 6100.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.3219, "cps": 47.65625, "mean_p": 0.166121, "boundary_entropy": 0.106897, "bpb": 1.041436}, "el": {"chunks": 5839.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.28451, "cps": 45.617188, "mean_p": 0.162362, "boundary_entropy": 0.125827, "bpb": 1.057312}, "hi": {"chunks": 6925.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.41159, "cps": 54.101562, "mean_p": 0.299917, "boundary_entropy": 0.465777, "bpb": 0.890854}, "ta": {"chunks": 8455.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.45226, "cps": 66.054688, "mean_p": 0.334037, "boundary_entropy": 0.52653, "bpb": 0.773482}, "th": {"chunks": 7394.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.459683, "cps": 57.765625, "mean_p": 0.326635, "boundary_entropy": 0.521252, "bpb": 0.872293}, "zh": {"chunks": 2945.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.498645, "cps": 23.007812, "mean_p": 0.194713, "boundary_entropy": 0.073165, "bpb": 2.483277}, "my": {"chunks": 9017.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.475881, "cps": 70.445312, "mean_p": 0.312082, "boundary_entropy": 0.511047, "bpb": 0.885974}}, "bpb": {"en": 1.62286, "de": 1.75388, "vi": 1.41048, "tr": 1.68582, "fi": 1.84276, "ru": 1.04144, "el": 1.05731, "hi": 0.89085, "ta": 0.77348, "th": 0.87229, "zh": 2.48328, "my": 0.88597}, "gini_cpc": 0.216557, "gini_cps": 0.217519, "bpic": 5.952031, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9466506664447858, "vi": 1.094016800024671, "tr": 0.9285405717751468, "fi": 0.8779486383889582, "ru": 1.6332423055916674, "el": 1.4435360370054846, "hi": 2.0883087346592215, "ta": 2.294658929555852, "th": 2.3323213141555748, "zh": 2.530007779803906, "my": 2.4145082505996154}, "t": 2299.3}
{"event": "eval", "step": 11000, "phase": 3, "bpb": {"en": 1.62194, "de": 1.75942, "vi": 1.4266, "tr": 1.67718, "fi": 1.85756, "ru": 1.03956, "el": 1.05588, "hi": 0.89643, "ta": 0.77949, "th": 0.87255, "zh": 2.49134, "my": 0.89364}, "bytes_per_chunk": {"en": 4.931, "de": 5.53, "vi": 6.122, "tr": 5.981, "fi": 5.84, "ru": 5.831, "el": 6.487, "hi": 6.157, "ta": 6.01, "th": 6.317, "zh": 5.339, "my": 6.095}, "bpb_mean": 1.3643, "t": 2304.6}
{"event": "probe", "step": 11000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3430.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.197092, "cps": 26.796875, "mean_p": 0.191548, "boundary_entropy": 0.053056, "bpb": 1.622865}, "de": {"chunks": 3831.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.186578, "cps": 29.929688, "mean_p": 0.185125, "boundary_entropy": 0.040192, "bpb": 1.753883}, "vi": {"chunks": 3975.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.215622, "cps": 31.054688, "mean_p": 0.172111, "boundary_entropy": 0.065071, "bpb": 1.410475}, "tr": {"chunks": 3313.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.183008, "cps": 25.882812, "mean_p": 0.17348, "boundary_entropy": 0.072507, "bpb": 1.685822}, "fi": {"chunks": 3255.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.173037, "cps": 25.429688, "mean_p": 0.170165, "boundary_entropy": 0.061665, "bpb": 1.842762}, "ru": {"chunks": 6100.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.3219, "cps": 47.65625, "mean_p": 0.166121, "boundary_entropy": 0.106897, "bpb": 1.041436}, "el": {"chunks": 5839.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.28451, "cps": 45.617188, "mean_p": 0.162362, "boundary_entropy": 0.125827, "bpb": 1.057312}, "hi": {"chunks": 6925.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.41159, "cps": 54.101562, "mean_p": 0.299917, "boundary_entropy": 0.465777, "bpb": 0.890854}, "ta": {"chunks": 8455.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.45226, "cps": 66.054688, "mean_p": 0.334037, "boundary_entropy": 0.52653, "bpb": 0.773482}, "th": {"chunks": 7394.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.459683, "cps": 57.765625, "mean_p": 0.326635, "boundary_entropy": 0.521252, "bpb": 0.872293}, "zh": {"chunks": 2945.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.498645, "cps": 23.007812, "mean_p": 0.194713, "boundary_entropy": 0.073165, "bpb": 2.483277}, "my": {"chunks": 9017.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.475881, "cps": 70.445312, "mean_p": 0.312082, "boundary_entropy": 0.511047, "bpb": 0.885974}}, "bpb": {"en": 1.62286, "de": 1.75388, "vi": 1.41048, "tr": 1.68582, "fi": 1.84276, "ru": 1.04144, "el": 1.05731, "hi": 0.89085, "ta": 0.77348, "th": 0.87229, "zh": 2.48328, "my": 0.88597}, "gini_cpc": 0.216557, "gini_cps": 0.217519, "bpic": 5.952031, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9466506664447858, "vi": 1.094016800024671, "tr": 0.9285405717751468, "fi": 0.8779486383889582, "ru": 1.6332423055916674, "el": 1.4435360370054846, "hi": 2.0883087346592215, "ta": 2.294658929555852, "th": 2.3323213141555748, "zh": 2.530007779803906, "my": 2.4145082505996154}, "t": 2307.7}
{"event": "reps_dumped", "dirs": ["/local/scratch/sas245/beetle-hnet/scratch/reps/spec_C_s0/th0.30", "/local/scratch/sas245/beetle-hnet/scratch/reps/spec_C_s0/th0.40", "/local/scratch/sas245/beetle-hnet/scratch/reps/spec_C_s0/th0.50", "/local/scratch/sas245/beetle-hnet/scratch/reps/spec_C_s0/th0.60", "/local/scratch/sas245/beetle-hnet/scratch/reps/spec_C_s0/th0.70"], "t": 2312.1}
{"event": "final", "run": "spec_C_s0", "condition": "C", "step": 11000, "params": 82470528, "per_lang": {"en": {"cpc": 0.19709245532379474, "cps": 26.796875, "bpb": 1.6173316312796457, "bytes_per_chunk": 4.940217570540817}, "de": {"cpc": 0.18657770418350947, "cps": 29.9296875, "bpb": 1.7564336178035131, "bytes_per_chunk": 5.513191793642281}, "vi": {"cpc": 0.21562245728234336, "cps": 31.0546875, "bpb": 1.427082316107734, "bytes_per_chunk": 6.122070551968146}, "tr": {"cpc": 0.18300834115892395, "cps": 25.8828125, "bpb": 1.676741167708875, "bytes_per_chunk": 5.989296015993146}, "fi": {"cpc": 0.17303705278826217, "cps": 25.4296875, "bpb": 1.8638905708413946, "bytes_per_chunk": 5.872664545902595}, "ru": {"cpc": 0.32189973614775724, "cps": 47.65625, "bpb": 1.039714881420016, "bytes_per_chunk": 5.831350762163756}, "el": {"cpc": 0.2845100618817912, "cps": 45.6171875, "bpb": 1.054551937573565, "bytes_per_chunk": 6.507760958746827}, "hi": {"cpc": 0.4115898959881129, "cps": 54.1015625, "bpb": 0.8972616624592671, "bytes_per_chunk": 6.138484954923311}, "ta": {"cpc": 0.4522599625568334, "cps": 66.0546875, "bpb": 0.7787565224321276, "bytes_per_chunk": 5.997963643019757}, "th": {"cpc": 0.4596829344109419, "cps": 57.765625, "bpb": 0.8748597559270351, "bytes_per_chunk": 6.297791578327797}, "zh": {"cpc": 0.4986454453098544, "cps": 23.0078125, "bpb": 2.491262406794672, "bytes_per_chunk": 5.349113391964413}, "my": {"cpc": 0.4758813595102385, "cps": 70.4453125, "bpb": 0.8927085615135781, "bytes_per_chunk": 6.0848038902783665}}, "gini_cpc": 0.2165569108456531, "gini_cps": 0.217519140598748, "bpic": 5.95203089377937, "seed": 0, "mean_boundary_entropy": 0.1501949578523636, "dechunk": "ema", "router_noise": 0.0, "transitions": [{"step": 0, "phase": 1, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0}, {"step": 4400, "phase": 2, "objective": "parity_a", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0}, {"step": 7700, "phase": 3, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0}], "t": 2324.4}