{"event": "start", "run": "spec_B_s0", "condition": "B", "params": 82470528, "langs": ["en", "de", "vi", "tr", "fi", "ru", "el", "hi", "ta", "th", "zh", "my"], "premium": {"en": 1.0, "de": 1.184, "vi": 1.39, "tr": 1.12, "fi": 1.104, "ru": 1.971, "el": 2.174, "hi": 2.552, "ta": 3.192, "th": 2.762, "zh": 0.915, "my": 3.524}, "schedule": {"tau1": 0.4, "tau2": 0.7, "p1_objective": "global", "p2_objective": "parity_a", "p3_objective": "none", "freeze_in_p3": true, "freeze_scope": "router"}, "corpus": "/local/scratch/sas245/beetle-hnet/scratch/data/tier1_corpus", "premium_kind": "flores_per_sent", "doc_filtered": {"my": 37511}, "weights": {"en": 0.2946, "de": 0.0909, "vi": 0.0576, "tr": 0.0612, "fi": 0.0424, "ru": 0.1476, "el": 0.0522, "hi": 0.042, "ta": 0.0313, "th": 0.0542, "zh": 0.1038, "my": 0.0222}, "t": 2.0}
{"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": 2.1}
{"event": "probe", "step": 0, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 8739.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.502155, "cps": 68.273438, "mean_p": 0.50363, "boundary_entropy": 0.687415, "bpb": 8.248358}, "de": {"chunks": 12222.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.595237, "cps": 95.484375, "mean_p": 0.507356, "boundary_entropy": 0.688163, "bpb": 8.204506}, "vi": {"chunks": 6386.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.346406, "cps": 49.890625, "mean_p": 0.491233, "boundary_entropy": 0.688507, "bpb": 8.13873}, "tr": {"chunks": 11403.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.629896, "cps": 89.085938, "mean_p": 0.50659, "boundary_entropy": 0.687851, "bpb": 8.143105}, "fi": {"chunks": 7137.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.379406, "cps": 55.757812, "mean_p": 0.497164, "boundary_entropy": 0.68792, "bpb": 8.234202}, "ru": {"chunks": 10473.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.552665, "cps": 81.820312, "mean_p": 0.496692, "boundary_entropy": 0.690192, "bpb": 8.071624}, "el": {"chunks": 36555.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.781172, "cps": 285.585938, "mean_p": 0.530217, "boundary_entropy": 0.688772, "bpb": 8.142597}, "hi": {"chunks": 42584.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.530996, "cps": 332.6875, "mean_p": 0.521073, "boundary_entropy": 0.69018, "bpb": 7.880816}, "ta": {"chunks": 49758.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.661567, "cps": 388.734375, "mean_p": 0.521091, "boundary_entropy": 0.690409, "bpb": 7.814466}, "th": {"chunks": 45716.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.842151, "cps": 357.15625, "mean_p": 0.525713, "boundary_entropy": 0.689893, "bpb": 8.063302}, "zh": {"chunks": 6218.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.052828, "cps": 48.578125, "mean_p": 0.496822, "boundary_entropy": 0.686552, "bpb": 8.034403}, "my": {"chunks": 53266.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.811167, "cps": 416.140625, "mean_p": 0.521325, "boundary_entropy": 0.690575, "bpb": 8.24483}}, "bpb": {"en": 8.24836, "de": 8.20451, "vi": 8.13873, "tr": 8.1431, "fi": 8.2342, "ru": 8.07162, "el": 8.1426, "hi": 7.88082, "ta": 7.81447, "th": 8.0633, "zh": 8.0344, "my": 8.24483}, "gini_cpc": 0.391191, "gini_cps": 0.409188, "bpic": 1.321301, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1853654183968407, "vi": 0.6898396505624507, "tr": 1.2543852937189204, "fi": 0.755555191784985, "ru": 1.1005867247811874, "el": 3.54705827773647, "hi": 5.040269530092341, "ta": 5.300292378412309, "th": 5.659910185771798, "zh": 2.096619669942964, "my": 5.598208760550946}, "t": 8.8}
{"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": 9.2839, "lr": 6e-07, "router_trainable": true, "sec_per_step": 0.005, "alpha": 0.03, "t": 9.6}
{"event": "probe", "step": 1, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 8721.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.50112, "cps": 68.132812, "mean_p": 0.503564, "boundary_entropy": 0.687414, "bpb": 8.209871}, "de": {"chunks": 12189.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.59363, "cps": 95.226562, "mean_p": 0.507261, "boundary_entropy": 0.688164, "bpb": 8.167576}, "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": 8.119073}, "tr": {"chunks": 11367.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.627907, "cps": 88.804688, "mean_p": 0.506498, "boundary_entropy": 0.687851, "bpb": 8.116552}, "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": 8.205235}, "ru": {"chunks": 10504.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.554301, "cps": 82.0625, "mean_p": 0.496728, "boundary_entropy": 0.690193, "bpb": 8.027828}, "el": {"chunks": 36546.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.780734, "cps": 285.515625, "mean_p": 0.53005, "boundary_entropy": 0.688791, "bpb": 8.124628}, "hi": {"chunks": 42563.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.529747, "cps": 332.523438, "mean_p": 0.520975, "boundary_entropy": 0.690188, "bpb": 7.857686}, "ta": {"chunks": 49696.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.658251, "cps": 388.25, "mean_p": 0.520856, "boundary_entropy": 0.690427, "bpb": 7.780875}, "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": 8.031115}, "zh": {"chunks": 6208.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.051134, "cps": 48.5, "mean_p": 0.496796, "boundary_entropy": 0.686551, "bpb": 8.029643}, "my": {"chunks": 53260.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.810851, "cps": 416.09375, "mean_p": 0.521208, "boundary_entropy": 0.690584, "bpb": 8.2169}}, "bpb": {"en": 8.20987, "de": 8.16758, "vi": 8.11907, "tr": 8.11655, "fi": 8.20524, "ru": 8.02783, "el": 8.12463, "hi": 7.85769, "ta": 7.78088, "th": 8.03112, "zh": 8.02964, "my": 8.2169}, "gini_cpc": 0.391551, "gini_cps": 0.409491, "bpic": 1.32238, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1846048423547546, "vi": 0.6890985340790992, "tr": 1.2530059759521923, "fi": 0.7515983272272038, "ru": 1.1061227698041212, "el": 3.553504243797653, "hi": 5.048181859537694, "ta": 5.304614135646422, "th": 5.671468085513631, "zh": 2.0975682434973977, "my": 5.6091314748477155}, "t": 15.8}
{"event": "probe", "step": 2, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 8647.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.496868, "cps": 67.554688, "mean_p": 0.503388, "boundary_entropy": 0.687412, "bpb": 8.133893}, "de": {"chunks": 12055.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.587104, "cps": 94.179688, "mean_p": 0.507025, "boundary_entropy": 0.688166, "bpb": 8.095631}, "vi": {"chunks": 6329.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.343314, "cps": 49.445312, "mean_p": 0.491077, "boundary_entropy": 0.688497, "bpb": 8.079332}, "tr": {"chunks": 11286.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.623433, "cps": 88.171875, "mean_p": 0.506283, "boundary_entropy": 0.687853, "bpb": 8.064868}, "fi": {"chunks": 6965.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.370262, "cps": 54.414062, "mean_p": 0.496716, "boundary_entropy": 0.687908, "bpb": 8.147996}, "ru": {"chunks": 10544.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.556412, "cps": 82.375, "mean_p": 0.49676, "boundary_entropy": 0.690194, "bpb": 7.949221}, "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": 8.08977}, "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": 7.805719}, "ta": {"chunks": 49564.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.65119, "cps": 387.21875, "mean_p": 0.520378, "boundary_entropy": 0.690465, "bpb": 7.714336}, "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": 7.952318}, "zh": {"chunks": 6195.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.048933, "cps": 48.398438, "mean_p": 0.496743, "boundary_entropy": 0.68655, "bpb": 8.01871}, "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": 8.175866}}, "bpb": {"en": 8.13389, "de": 8.09563, "vi": 8.07933, "tr": 8.06487, "fi": 8.148, "ru": 7.94922, "el": 8.08977, "hi": 7.80572, "ta": 7.71434, "th": 7.95232, "zh": 8.01871, "my": 8.17587}, "gini_cpc": 0.39261, "gini_cps": 0.410346, "bpic": 1.325307, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1816081254167057, "vi": 0.6909563540147075, "tr": 1.2547238572724964, "fi": 0.7451915103941217, "ru": 1.1198370810807843, "el": 3.5812669334127776, "hi": 5.086479183274104, "ta": 5.335799967170962, "th": 5.719253097674589, "zh": 2.1110889342226002, "my": 5.655434279556627}, "t": 22.7}
{"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": 7.880232}, "de": {"chunks": 11669.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.568305, "cps": 91.164062, "mean_p": 0.506293, "boundary_entropy": 0.688173, "bpb": 7.857244}, "vi": {"chunks": 6149.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.33355, "cps": 48.039062, "mean_p": 0.490665, "boundary_entropy": 0.688473, "bpb": 7.94747}, "tr": {"chunks": 11014.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.608407, "cps": 86.046875, "mean_p": 0.505596, "boundary_entropy": 0.687857, "bpb": 7.887201}, "fi": {"chunks": 6503.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.345702, "cps": 50.804688, "mean_p": 0.495723, "boundary_entropy": 0.687877, "bpb": 7.953834}, "ru": {"chunks": 10722.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.565805, "cps": 83.765625, "mean_p": 0.496887, "boundary_entropy": 0.690197, "bpb": 7.714034}, "el": {"chunks": 36343.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.770842, "cps": 283.929688, "mean_p": 0.528228, "boundary_entropy": 0.688991, "bpb": 7.945126}, "hi": {"chunks": 42405.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.520357, "cps": 331.289062, "mean_p": 0.519954, "boundary_entropy": 0.690264, "bpb": 7.62661}, "ta": {"chunks": 49031.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.62268, "cps": 383.054688, "mean_p": 0.518697, "boundary_entropy": 0.690591, "bpb": 7.463866}, "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": 7.684558}, "zh": {"chunks": 6136.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.038943, "cps": 47.9375, "mean_p": 0.496565, "boundary_entropy": 0.686548, "bpb": 7.983354}, "my": {"chunks": 53171.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.806154, "cps": 415.398438, "mean_p": 0.520202, "boundary_entropy": 0.690662, "bpb": 8.00646}}, "bpb": {"en": 7.88023, "de": 7.85724, "vi": 7.94747, "tr": 7.8872, "fi": 7.95383, "ru": 7.71403, "el": 7.94513, "hi": 7.62661, "ta": 7.46387, "th": 7.68456, "zh": 7.98335, "my": 8.00646}, "gini_cpc": 0.396303, "gini_cps": 0.413379, "bpic": 1.33601, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1727980968519445, "vi": 0.6883405760835155, "tr": 1.2555573090957268, "fi": 0.7134177170759686, "ru": 1.1676390433737043, "el": 3.6544493650415526, "hi": 5.2012055166420605, "ta": 5.412367795527619, "th": 5.861051951086517, "zh": 2.144045157605522, "my": 5.790998761838181}, "t": 29.9}
{"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": 0.500776, "boundary_entropy": 0.68737, "bpb": 7.154049}, "de": {"chunks": 10357.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.504408, "cps": 80.914062, "mean_p": 0.503631, "boundary_entropy": 0.688184, "bpb": 7.186553}, "vi": {"chunks": 5625.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.305126, "cps": 43.945312, "mean_p": 0.489196, "boundary_entropy": 0.688385, "bpb": 7.546545}, "tr": {"chunks": 10042.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.554715, "cps": 78.453125, "mean_p": 0.503186, "boundary_entropy": 0.687852, "bpb": 7.361703}, "fi": {"chunks": 5058.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.268885, "cps": 39.515625, "mean_p": 0.492099, "boundary_entropy": 0.687733, "bpb": 7.360887}, "ru": {"chunks": 12129.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.640053, "cps": 94.757812, "mean_p": 0.498124, "boundary_entropy": 0.690223, "bpb": 6.973046}, "el": {"chunks": 35199.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.7151, "cps": 274.992188, "mean_p": 0.522747, "boundary_entropy": 0.689514, "bpb": 7.476324}, "hi": {"chunks": 41449.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.463536, "cps": 323.820312, "mean_p": 0.516812, "boundary_entropy": 0.690472, "bpb": 7.075418}, "ta": {"chunks": 44749.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.393635, "cps": 349.601562, "mean_p": 0.512874, "boundary_entropy": 0.690936, "bpb": 6.797267}, "th": {"chunks": 45451.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.825676, "cps": 355.085938, "mean_p": 0.520857, "boundary_entropy": 0.690323, "bpb": 6.95338}, "zh": {"chunks": 5982.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.012868, "cps": 46.734375, "mean_p": 0.496043, "boundary_entropy": 0.686543, "bpb": 7.877133}, "my": {"chunks": 52940.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.793962, "cps": 413.59375, "mean_p": 0.518192, "boundary_entropy": 0.690804, "bpb": 7.572239}}, "bpb": {"en": 7.15405, "de": 7.18655, "vi": 7.54655, "tr": 7.3617, "fi": 7.36089, "ru": 6.97305, "el": 7.47632, "hi": 7.07542, "ta": 6.79727, "th": 6.95338, "zh": 7.87713, "my": 7.57224}, "gini_cpc": 0.408431, "gini_cps": 0.422668, "bpic": 1.387505, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1524490485320829, "vi": 0.6971392733888363, "tr": 1.26738869873687, "fi": 0.6143376133340567, "ru": 1.4623655460307332, "el": 3.9185883667498613, "hi": 5.62858396240052, "ta": 5.468875412513703, "th": 6.45598543969898, "zh": 2.3141586576197635, "my": 6.3835273802982115}, "t": 37.8}
{"event": "probe", "step": 16, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 4504.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.258806, "cps": 35.1875, "mean_p": 0.492305, "boundary_entropy": 0.687145, "bpb": 6.329209}, "de": {"chunks": 6199.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.301904, "cps": 48.429688, "mean_p": 0.49467, "boundary_entropy": 0.688101, "bpb": 6.404814}, "vi": {"chunks": 3432.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.186168, "cps": 26.8125, "mean_p": 0.481703, "boundary_entropy": 0.687843, "bpb": 7.023733}, "tr": {"chunks": 6314.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.348782, "cps": 49.328125, "mean_p": 0.493765, "boundary_entropy": 0.687686, "bpb": 6.708927}, "fi": {"chunks": 2449.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.13019, "cps": 19.132812, "mean_p": 0.482989, "boundary_entropy": 0.687182, "bpb": 6.607175}, "ru": {"chunks": 14747.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.778206, "cps": 115.210938, "mean_p": 0.500092, "boundary_entropy": 0.690296, "bpb": 5.842078}, "el": {"chunks": 19921.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.970667, "cps": 155.632812, "mean_p": 0.503235, "boundary_entropy": 0.690397, "bpb": 6.520565}, "hi": {"chunks": 27428.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 1.630193, "cps": 214.28125, "mean_p": 0.504111, "boundary_entropy": 0.690918, "bpb": 5.801836}, "ta": {"chunks": 7757.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.414924, "cps": 60.601562, "mean_p": 0.490865, "boundary_entropy": 0.691013, "bpb": 5.849243}, "th": {"chunks": 26091.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 1.62207, "cps": 203.835938, "mean_p": 0.502668, "boundary_entropy": 0.691089, "bpb": 5.339611}, "zh": {"chunks": 5220.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.883847, "cps": 40.78125, "mean_p": 0.493444, "boundary_entropy": 0.686482, "bpb": 7.501168}, "my": {"chunks": 49047.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.588505, "cps": 383.179688, "mean_p": 0.510262, "boundary_entropy": 0.691206, "bpb": 6.227887}}, "bpb": {"en": 6.32921, "de": 6.40481, "vi": 7.02373, "tr": 6.70893, "fi": 6.60717, "ru": 5.84208, "el": 6.52057, "hi": 5.80184, "ta": 5.84924, "th": 5.33961, "zh": 7.50117, "my": 6.22789}, "gini_cpc": 0.452273, "gini_cps": 0.471881, "bpic": 2.21699, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1665274627448479, "vi": 0.7193328197916832, "tr": 1.347658219607576, "fi": 0.5030401390177113, "ru": 3.0069084413971514, "el": 3.7505592326384405, "hi": 6.298901343101987, "ta": 1.6032234637826908, "th": 6.267515229097638, "zh": 3.415095074461411, "my": 10.001722731577981}, "t": 46.3}
{"event": "probe", "step": 32, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 5961.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.342527, "cps": 46.570312, "mean_p": 0.498473, "boundary_entropy": 0.687578, "bpb": 5.76594}, "de": {"chunks": 8115.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.395217, "cps": 63.398438, "mean_p": 0.499931, "boundary_entropy": 0.688533, "bpb": 5.889655}, "vi": {"chunks": 3504.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.190073, "cps": 27.375, "mean_p": 0.485039, "boundary_entropy": 0.688282, "bpb": 6.442276}, "tr": {"chunks": 5518.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.304811, "cps": 43.109375, "mean_p": 0.494106, "boundary_entropy": 0.687971, "bpb": 6.244966}, "fi": {"chunks": 4399.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.233853, "cps": 34.367188, "mean_p": 0.492993, "boundary_entropy": 0.688004, "bpb": 5.942946}, "ru": {"chunks": 2312.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.122005, "cps": 18.0625, "mean_p": 0.48883, "boundary_entropy": 0.690081, "bpb": 5.22829}, "el": {"chunks": 490.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.023876, "cps": 3.828125, "mean_p": 0.466889, "boundary_entropy": 0.688052, "bpb": 6.166402}, "hi": {"chunks": 6076.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.361129, "cps": 47.46875, "mean_p": 0.492886, "boundary_entropy": 0.690798, "bpb": 4.972842}, "ta": {"chunks": 4290.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.229473, "cps": 33.515625, "mean_p": 0.489066, "boundary_entropy": 0.690963, "bpb": 5.234573}, "th": {"chunks": 2613.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.162449, "cps": 20.414062, "mean_p": 0.488725, "boundary_entropy": 0.690774, "bpb": 4.928393}, "zh": {"chunks": 2800.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.474094, "cps": 21.875, "mean_p": 0.484996, "boundary_entropy": 0.686154, "bpb": 7.083364}, "my": {"chunks": 4952.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.261347, "cps": 38.6875, "mean_p": 0.491043, "boundary_entropy": 0.691175, "bpb": 4.778324}}, "bpb": {"en": 5.76594, "de": 5.88966, "vi": 6.44228, "tr": 6.24497, "fi": 5.94295, "ru": 5.22829, "el": 6.1664, "hi": 4.97284, "ta": 5.23457, "th": 4.92839, "zh": 7.08336, "my": 4.77832}, "gini_cpc": 0.263848, "gini_cps": 0.260027, "bpic": 7.520694, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1538281104488874, "vi": 0.5549143476535642, "tr": 0.8898896242134666, "fi": 0.6827269977195077, "ru": 0.3561915511510836, "el": 0.06970440642032197, "hi": 1.0543084581727737, "ta": 0.6699414070570772, "th": 0.4742674759264407, "zh": 1.384106751453019, "my": 0.7629959949721745}, "t": 55.5}
{"event": "probe", "step": 64, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 6230.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.357984, "cps": 48.671875, "mean_p": 0.498861, "boundary_entropy": 0.687615, "bpb": 5.36764}, "de": {"chunks": 6336.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.308576, "cps": 49.5, "mean_p": 0.496415, "boundary_entropy": 0.688459, "bpb": 5.511319}, "vi": {"chunks": 4598.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.249417, "cps": 35.921875, "mean_p": 0.489378, "boundary_entropy": 0.688661, "bpb": 5.904808}, "tr": {"chunks": 4869.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.268961, "cps": 38.039062, "mean_p": 0.493223, "boundary_entropy": 0.688016, "bpb": 5.846346}, "fi": {"chunks": 6363.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.33826, "cps": 49.710938, "mean_p": 0.497395, "boundary_entropy": 0.688184, "bpb": 5.549911}, "ru": {"chunks": 5264.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.277784, "cps": 41.125, "mean_p": 0.495105, "boundary_entropy": 0.690388, "bpb": 4.833249}, "el": {"chunks": 1418.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.069093, "cps": 11.078125, "mean_p": 0.488625, "boundary_entropy": 0.69031, "bpb": 5.110323}, "hi": {"chunks": 9581.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.56945, "cps": 74.851562, "mean_p": 0.49594, "boundary_entropy": 0.690915, "bpb": 4.441704}, "ta": {"chunks": 14117.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.755122, "cps": 110.289062, "mean_p": 0.494373, "boundary_entropy": 0.691144, "bpb": 4.368056}, "th": {"chunks": 4514.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.280634, "cps": 35.265625, "mean_p": 0.491345, "boundary_entropy": 0.690886, "bpb": 4.443132}, "zh": {"chunks": 1619.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.274128, "cps": 12.648438, "mean_p": 0.484437, "boundary_entropy": 0.686415, "bpb": 6.714336}, "my": {"chunks": 20487.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 1.081222, "cps": 160.054688, "mean_p": 0.499211, "boundary_entropy": 0.6914, "bpb": 4.244972}}, "bpb": {"en": 5.36764, "de": 5.51132, "vi": 5.90481, "tr": 5.84635, "fi": 5.54991, "ru": 4.83325, "el": 5.11032, "hi": 4.4417, "ta": 4.36806, "th": 4.44313, "zh": 6.71434, "my": 4.24497}, "gini_cpc": 0.317709, "gini_cps": 0.362919, "bpic": 4.494133, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 0.8619832663373426, "vi": 0.6967258090005185, "tr": 0.7513206000016494, "fi": 0.94490057514758, "ru": 0.7759660846105957, "el": 0.19300628844637294, "hi": 1.5907130383348558, "ta": 2.109371232125739, "th": 0.7839286974145678, "zh": 0.7657543624868798, "my": 3.0203068311313115}, "t": 69.7}
{"event": "train", "step": 100, "phase": 1, "objective": "global", "lm_loss": 3.29742, "bpb": 4.75717, "chunk_loss": 0.030662, "F": 0.17619, "G": 0.47258, "chunks": 180.42, "bytes_per_chunk": 5.676, "boundary_entropy": 0.68858, "router_noise": 0.0, "grad_norm": 2.0297, "lr": 6.06e-05, "router_trainable": true, "sec_per_step": 0.699, "alpha": 0.03, "t": 79.5}
{"event": "probe", "step": 128, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3620.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.20801, "cps": 28.28125, "mean_p": 0.458081, "boundary_entropy": 0.66186, "bpb": 4.288021}, "de": {"chunks": 4216.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.205328, "cps": 32.9375, "mean_p": 0.458381, "boundary_entropy": 0.666823, "bpb": 4.523817}, "vi": {"chunks": 7288.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.395335, "cps": 56.9375, "mean_p": 0.492051, "boundary_entropy": 0.683229, "bpb": 4.680085}, "tr": {"chunks": 4235.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.233939, "cps": 33.085938, "mean_p": 0.477633, "boundary_entropy": 0.677891, "bpb": 4.815447}, "fi": {"chunks": 4100.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.217958, "cps": 32.03125, "mean_p": 0.470973, "boundary_entropy": 0.677138, "bpb": 4.546036}, "ru": {"chunks": 5137.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.271082, "cps": 40.132812, "mean_p": 0.480951, "boundary_entropy": 0.687754, "bpb": 3.210517}, "el": {"chunks": 6536.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.318472, "cps": 51.0625, "mean_p": 0.490708, "boundary_entropy": 0.690249, "bpb": 4.074933}, "hi": {"chunks": 8246.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.490104, "cps": 64.421875, "mean_p": 0.470962, "boundary_entropy": 0.687942, "bpb": 2.874756}, "ta": {"chunks": 12997.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.695213, "cps": 101.539062, "mean_p": 0.438613, "boundary_entropy": 0.67883, "bpb": 2.942172}, "th": {"chunks": 9430.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.58626, "cps": 73.671875, "mean_p": 0.480244, "boundary_entropy": 0.689385, "bpb": 2.737959}, "zh": {"chunks": 2071.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.35066, "cps": 16.179688, "mean_p": 0.48606, "boundary_entropy": 0.686518, "bpb": 6.216194}, "my": {"chunks": 11259.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.594205, "cps": 87.960938, "mean_p": 0.476202, "boundary_entropy": 0.689409, "bpb": 2.927588}}, "bpb": {"en": 4.28802, "de": 4.52382, "vi": 4.68009, "tr": 4.81545, "fi": 4.54604, "ru": 3.21052, "el": 4.07493, "hi": 2.87476, "ta": 2.94217, "th": 2.73796, "zh": 6.21619, "my": 2.92759}, "gini_cpc": 0.239574, "gini_cps": 0.270587, "bpic": 4.8497, "worst_lang": "ta", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9871058931411584, "vi": 1.9005564421507852, "tr": 1.124652655171772, "fi": 1.0478220249054935, "ru": 1.3032144929226372, "el": 1.531040743130669, "hi": 2.3561547289698144, "ta": 3.342205881266448, "th": 2.818423018597678, "zh": 1.6857850804407166, "my": 2.8566168443556488}, "t": 88.8}
{"event": "train", "step": 200, "phase": 1, "objective": "global", "lm_loss": 2.45308, "bpb": 3.53904, "chunk_loss": 0.031618, "F": 0.19548, "G": 0.40358, "chunks": 200.17, "bytes_per_chunk": 5.116, "boundary_entropy": 0.58493, "router_noise": 0.0, "grad_norm": 2.3089, "lr": 0.0001206, "router_trainable": true, "sec_per_step": 0.233, "alpha": 0.03, "t": 102.8}
{"event": "probe", "step": 256, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2957.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.169913, "cps": 23.101562, "mean_p": 0.336339, "boundary_entropy": 0.457538, "bpb": 3.53605}, "de": {"chunks": 3253.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.158428, "cps": 25.414062, "mean_p": 0.345021, "boundary_entropy": 0.498286, "bpb": 3.770067}, "vi": {"chunks": 5242.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.28435, "cps": 40.953125, "mean_p": 0.332778, "boundary_entropy": 0.432818, "bpb": 3.06135}, "tr": {"chunks": 3561.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.196708, "cps": 27.820312, "mean_p": 0.340817, "boundary_entropy": 0.503583, "bpb": 3.708322}, "fi": {"chunks": 2884.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.153315, "cps": 22.53125, "mean_p": 0.328776, "boundary_entropy": 0.508661, "bpb": 3.79332}, "ru": {"chunks": 6532.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.344697, "cps": 51.03125, "mean_p": 0.352367, "boundary_entropy": 0.59081, "bpb": 2.642954}, "el": {"chunks": 10579.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.51547, "cps": 82.648438, "mean_p": 0.391111, "boundary_entropy": 0.595234, "bpb": 2.677049}, "hi": {"chunks": 16145.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.959584, "cps": 126.132812, "mean_p": 0.444594, "boundary_entropy": 0.661979, "bpb": 2.10827}, "ta": {"chunks": 4131.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.220968, "cps": 32.273438, "mean_p": 0.426302, "boundary_entropy": 0.669478, "bpb": 1.970883}, "th": {"chunks": 5862.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.364439, "cps": 45.796875, "mean_p": 0.44247, "boundary_entropy": 0.673747, "bpb": 2.081658}, "zh": {"chunks": 4140.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.700982, "cps": 32.34375, "mean_p": 0.4065, "boundary_entropy": 0.631647, "bpb": 4.913624}, "my": {"chunks": 11399.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.601594, "cps": 89.054688, "mean_p": 0.295453, "boundary_entropy": 0.543747, "bpb": 2.149119}}, "bpb": {"en": 3.53605, "de": 3.77007, "vi": 3.06135, "tr": 3.70832, "fi": 3.79332, "ru": 2.64295, "el": 2.67705, "hi": 2.10827, "ta": 1.97088, "th": 2.08166, "zh": 4.91362, "my": 2.14912}, "gini_cpc": 0.336482, "gini_cps": 0.319794, "bpic": 5.004642, "worst_lang": "hi", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9324046952239886, "vi": 1.6735036747214551, "tr": 1.1576951607708426, "fi": 0.9023108263793516, "ru": 2.0286622950059026, "el": 3.0337274950606403, "hi": 5.647493921040612, "ta": 1.3004765370597176, "th": 2.1448530585895504, "zh": 4.125529473646479, "my": 3.5405943604156946}, "t": 116.1}
{"event": "train", "step": 300, "phase": 1, "objective": "global", "lm_loss": 2.13069, "bpb": 3.07394, "chunk_loss": 0.032515, "F": 0.20926, "G": 0.34655, "chunks": 214.28, "bytes_per_chunk": 4.779, "boundary_entropy": 0.49254, "router_noise": 0.0, "grad_norm": 1.2562, "lr": 0.0001806, "router_trainable": true, "sec_per_step": 0.228, "alpha": 0.03, "t": 125.5}
{"event": "train", "step": 400, "phase": 1, "objective": "global", "lm_loss": 2.06073, "bpb": 2.973, "chunk_loss": 0.030795, "F": 0.1964, "G": 0.31732, "chunks": 201.11, "bytes_per_chunk": 5.092, "boundary_entropy": 0.43719, "router_noise": 0.0, "grad_norm": 0.9128, "lr": 0.0002406, "router_trainable": true, "sec_per_step": 0.214, "alpha": 0.03, "t": 146.9}
{"event": "train", "step": 500, "phase": 1, "objective": "global", "lm_loss": 1.96119, "bpb": 2.82939, "chunk_loss": 0.031157, "F": 0.20506, "G": 0.28671, "chunks": 209.98, "bytes_per_chunk": 4.877, "boundary_entropy": 0.35733, "router_noise": 0.0, "grad_norm": 0.9237, "lr": 0.0003, "router_trainable": true, "sec_per_step": 0.174, "alpha": 0.03, "t": 164.3}
{"event": "train", "step": 600, "phase": 1, "objective": "global", "lm_loss": 1.81999, "bpb": 2.62569, "chunk_loss": 0.030379, "F": 0.188, "G": 0.25186, "chunks": 192.52, "bytes_per_chunk": 5.319, "boundary_entropy": 0.29139, "router_noise": 0.0, "grad_norm": 0.8556, "lr": 0.0002999, "router_trainable": true, "sec_per_step": 0.174, "alpha": 0.03, "t": 181.7}
{"event": "train", "step": 700, "phase": 1, "objective": "global", "lm_loss": 1.70835, "bpb": 2.46463, "chunk_loss": 0.030457, "F": 0.19032, "G": 0.24928, "chunks": 194.89, "bytes_per_chunk": 5.254, "boundary_entropy": 0.26892, "router_noise": 0.0, "grad_norm": 0.7615, "lr": 0.0002998, "router_trainable": true, "sec_per_step": 0.173, "alpha": 0.03, "t": 199.0}
{"event": "train", "step": 800, "phase": 1, "objective": "global", "lm_loss": 1.58457, "bpb": 2.28605, "chunk_loss": 0.030289, "F": 0.18787, "G": 0.23237, "chunks": 192.38, "bytes_per_chunk": 5.323, "boundary_entropy": 0.22198, "router_noise": 0.0, "grad_norm": 0.6795, "lr": 0.0002995, "router_trainable": true, "sec_per_step": 0.175, "alpha": 0.03, "t": 216.4}
{"event": "train", "step": 900, "phase": 1, "objective": "global", "lm_loss": 1.56439, "bpb": 2.25694, "chunk_loss": 0.030327, "F": 0.18846, "G": 0.23287, "chunks": 192.98, "bytes_per_chunk": 5.306, "boundary_entropy": 0.21966, "router_noise": 0.0, "grad_norm": 0.6625, "lr": 0.000299, "router_trainable": true, "sec_per_step": 0.173, "alpha": 0.03, "t": 233.7}
{"event": "probe", "step": 1000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3437.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.197495, "cps": 26.851562, "mean_p": 0.215497, "boundary_entropy": 0.129208, "bpb": 2.328941}, "de": {"chunks": 3708.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.180587, "cps": 28.96875, "mean_p": 0.203708, "boundary_entropy": 0.147721, "bpb": 2.610985}, "vi": {"chunks": 4256.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.230865, "cps": 33.25, "mean_p": 0.218357, "boundary_entropy": 0.176313, "bpb": 2.097347}, "tr": {"chunks": 3657.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.202011, "cps": 28.570312, "mean_p": 0.201863, "boundary_entropy": 0.167713, "bpb": 2.600749}, "fi": {"chunks": 3119.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.165807, "cps": 24.367188, "mean_p": 0.199892, "boundary_entropy": 0.202661, "bpb": 2.852059}, "ru": {"chunks": 6672.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.352084, "cps": 52.125, "mean_p": 0.212022, "boundary_entropy": 0.157321, "bpb": 1.61614}, "el": {"chunks": 7406.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.360863, "cps": 57.859375, "mean_p": 0.219506, "boundary_entropy": 0.156348, "bpb": 1.711665}, "hi": {"chunks": 6998.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.415929, "cps": 54.671875, "mean_p": 0.298445, "boundary_entropy": 0.450517, "bpb": 1.393116}, "ta": {"chunks": 7225.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.386467, "cps": 56.445312, "mean_p": 0.294416, "boundary_entropy": 0.507685, "bpb": 1.125245}, "th": {"chunks": 7955.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.49456, "cps": 62.148438, "mean_p": 0.310023, "boundary_entropy": 0.527075, "bpb": 1.371658}, "zh": {"chunks": 3488.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.590586, "cps": 27.25, "mean_p": 0.245131, "boundary_entropy": 0.198993, "bpb": 3.458117}, "my": {"chunks": 10483.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.553251, "cps": 81.898438, "mean_p": 0.281111, "boundary_entropy": 0.458953, "bpb": 1.282504}}, "bpb": {"en": 2.32894, "de": 2.61099, "vi": 2.09735, "tr": 2.60075, "fi": 2.85206, "ru": 1.61614, "el": 1.71167, "hi": 1.39312, "ta": 1.12524, "th": 1.37166, "zh": 3.45812, "my": 1.2825}, "gini_cpc": 0.235063, "gini_cps": 0.218333, "bpic": 5.610505, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9143909232681933, "vi": 1.1689691915835736, "tr": 1.022866598336024, "fi": 0.8395529256486051, "ru": 1.7827539664200616, "el": 1.8272057389687373, "hi": 2.106024665589613, "ta": 1.9568474469149852, "th": 2.5041694142145094, "zh": 2.990388553638464, "my": 2.801346290590624}, "t": 253.9}
{"event": "eval", "step": 1000, "phase": 1, "bpb": {"en": 2.33786, "de": 2.60239, "vi": 2.10163, "tr": 2.59044, "fi": 2.86106, "ru": 1.61485, "el": 1.70729, "hi": 1.40102, "ta": 1.12487, "th": 1.36808, "zh": 3.46234, "my": 1.29457}, "bytes_per_chunk": {"en": 4.883, "de": 5.595, "vi": 5.715, "tr": 5.331, "fi": 6.132, "ru": 5.253, "el": 5.142, "hi": 6.086, "ta": 6.731, "th": 5.609, "zh": 4.627, "my": 5.365}, "bpb_mean": 2.03887, "t": 259.4}
{"event": "train", "step": 1000, "phase": 1, "objective": "global", "lm_loss": 1.48417, "bpb": 2.1412, "chunk_loss": 0.030339, "F": 0.19148, "G": 0.22841, "chunks": 196.08, "bytes_per_chunk": 5.222, "boundary_entropy": 0.19669, "router_noise": 0.0, "grad_norm": 0.6671, "lr": 0.0002985, "router_trainable": true, "sec_per_step": 0.258, "alpha": 0.03, "t": 259.5}
{"event": "train", "step": 1100, "phase": 1, "objective": "global", "lm_loss": 1.40904, "bpb": 2.03281, "chunk_loss": 0.030175, "F": 0.18333, "G": 0.23785, "chunks": 187.73, "bytes_per_chunk": 5.455, "boundary_entropy": 0.23225, "router_noise": 0.0, "grad_norm": 0.6113, "lr": 0.0002978, "router_trainable": true, "sec_per_step": 0.173, "alpha": 0.03, "t": 276.9}
{"event": "train", "step": 1200, "phase": 1, "objective": "global", "lm_loss": 1.43525, "bpb": 2.07063, "chunk_loss": 0.030313, "F": 0.18918, "G": 0.22929, "chunks": 193.72, "bytes_per_chunk": 5.286, "boundary_entropy": 0.19921, "router_noise": 0.0, "grad_norm": 0.5247, "lr": 0.000297, "router_trainable": true, "sec_per_step": 0.172, "alpha": 0.03, "t": 294.1}
{"event": "train", "step": 1300, "phase": 1, "objective": "global", "lm_loss": 1.4528, "bpb": 2.09595, "chunk_loss": 0.030369, "F": 0.19147, "G": 0.22912, "chunks": 196.06, "bytes_per_chunk": 5.223, "boundary_entropy": 0.19578, "router_noise": 0.0, "grad_norm": 0.6026, "lr": 0.0002962, "router_trainable": true, "sec_per_step": 0.17, "alpha": 0.03, "t": 311.1}
{"event": "train", "step": 1400, "phase": 1, "objective": "global", "lm_loss": 1.41772, "bpb": 2.04533, "chunk_loss": 0.030284, "F": 0.18851, "G": 0.21881, "chunks": 193.03, "bytes_per_chunk": 5.305, "boundary_entropy": 0.17106, "router_noise": 0.0, "grad_norm": 0.4757, "lr": 0.0002951, "router_trainable": true, "sec_per_step": 0.171, "alpha": 0.03, "t": 328.1}
{"event": "train", "step": 1500, "phase": 1, "objective": "global", "lm_loss": 1.41842, "bpb": 2.04634, "chunk_loss": 0.030374, "F": 0.187, "G": 0.2234, "chunks": 191.48, "bytes_per_chunk": 5.348, "boundary_entropy": 0.19216, "router_noise": 0.0, "grad_norm": 0.4943, "lr": 0.000294, "router_trainable": true, "sec_per_step": 0.17, "alpha": 0.03, "t": 345.1}
{"event": "train", "step": 1600, "phase": 1, "objective": "global", "lm_loss": 1.39827, "bpb": 2.01727, "chunk_loss": 0.030228, "F": 0.18658, "G": 0.21849, "chunks": 191.06, "bytes_per_chunk": 5.36, "boundary_entropy": 0.16827, "router_noise": 0.0, "grad_norm": 0.498, "lr": 0.0002928, "router_trainable": true, "sec_per_step": 0.168, "alpha": 0.03, "t": 362.0}
{"event": "train", "step": 1700, "phase": 1, "objective": "global", "lm_loss": 1.34164, "bpb": 1.93558, "chunk_loss": 0.030234, "F": 0.1879, "G": 0.21844, "chunks": 192.41, "bytes_per_chunk": 5.322, "boundary_entropy": 0.16147, "router_noise": 0.0, "grad_norm": 0.4751, "lr": 0.0002914, "router_trainable": true, "sec_per_step": 0.17, "alpha": 0.03, "t": 379.0}
{"event": "train", "step": 1800, "phase": 1, "objective": "global", "lm_loss": 1.35486, "bpb": 1.95465, "chunk_loss": 0.030161, "F": 0.18422, "G": 0.22221, "chunks": 188.64, "bytes_per_chunk": 5.428, "boundary_entropy": 0.19282, "router_noise": 0.0, "grad_norm": 0.4481, "lr": 0.0002899, "router_trainable": true, "sec_per_step": 0.171, "alpha": 0.03, "t": 396.1}
{"event": "train", "step": 1900, "phase": 1, "objective": "global", "lm_loss": 1.35765, "bpb": 1.95868, "chunk_loss": 0.030302, "F": 0.18814, "G": 0.22086, "chunks": 192.66, "bytes_per_chunk": 5.315, "boundary_entropy": 0.17142, "router_noise": 0.0, "grad_norm": 0.4635, "lr": 0.0002883, "router_trainable": true, "sec_per_step": 0.17, "alpha": 0.03, "t": 413.1}
{"event": "probe", "step": 2000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3377.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.194047, "cps": 26.382812, "mean_p": 0.20109, "boundary_entropy": 0.102393, "bpb": 2.077548}, "de": {"chunks": 3702.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.180295, "cps": 28.921875, "mean_p": 0.197416, "boundary_entropy": 0.1233, "bpb": 2.286278}, "vi": {"chunks": 4300.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.233252, "cps": 33.59375, "mean_p": 0.203065, "boundary_entropy": 0.126931, "bpb": 1.935839}, "tr": {"chunks": 3545.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.195824, "cps": 27.695312, "mean_p": 0.194571, "boundary_entropy": 0.141686, "bpb": 2.282955}, "fi": {"chunks": 3161.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.16804, "cps": 24.695312, "mean_p": 0.186691, "boundary_entropy": 0.145811, "bpb": 2.488589}, "ru": {"chunks": 6587.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.347599, "cps": 51.460938, "mean_p": 0.202613, "boundary_entropy": 0.126985, "bpb": 1.35443}, "el": {"chunks": 7320.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.356673, "cps": 57.1875, "mean_p": 0.210885, "boundary_entropy": 0.128395, "bpb": 1.401721}, "hi": {"chunks": 6538.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.388588, "cps": 51.078125, "mean_p": 0.278405, "boundary_entropy": 0.416788, "bpb": 1.199934}, "ta": {"chunks": 7885.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.421771, "cps": 61.601562, "mean_p": 0.315114, "boundary_entropy": 0.512246, "bpb": 1.041729}, "th": {"chunks": 7649.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.475536, "cps": 59.757812, "mean_p": 0.306656, "boundary_entropy": 0.514769, "bpb": 1.16486}, "zh": {"chunks": 3304.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.559431, "cps": 25.8125, "mean_p": 0.232433, "boundary_entropy": 0.160222, "bpb": 3.184871}, "my": {"chunks": 10764.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.568081, "cps": 84.09375, "mean_p": 0.291251, "boundary_entropy": 0.462989, "bpb": 1.166225}}, "bpb": {"en": 2.07755, "de": 2.28628, "vi": 1.93584, "tr": 2.28296, "fi": 2.48859, "ru": 1.35443, "el": 1.40172, "hi": 1.19993, "ta": 1.04173, "th": 1.16486, "zh": 3.18487, "my": 1.16623}, "gini_cpc": 0.232077, "gini_cps": 0.225706, "bpic": 5.632904, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9291312491886892, "vi": 1.2020384870322454, "tr": 1.009157024692269, "fi": 0.8659756330867402, "ru": 1.7913131278405916, "el": 1.8380752848112185, "hi": 2.0025478534320755, "ta": 2.1735482615553843, "th": 2.4506238465716765, "zh": 2.882966895938853, "my": 2.9275436056188315}, "t": 432.9}
{"event": "eval", "step": 2000, "phase": 1, "bpb": {"en": 2.08616, "de": 2.27964, "vi": 1.94599, "tr": 2.27273, "fi": 2.50382, "ru": 1.35312, "el": 1.39888, "hi": 1.20793, "ta": 1.04109, "th": 1.15871, "zh": 3.19078, "my": 1.17756}, "bytes_per_chunk": {"en": 4.936, "de": 5.661, "vi": 5.667, "tr": 5.477, "fi": 5.998, "ru": 5.35, "el": 5.179, "hi": 6.399, "ta": 6.487, "th": 5.813, "zh": 5.009, "my": 5.271}, "bpb_mean": 1.80137, "t": 438.4}
{"event": "train", "step": 2000, "phase": 1, "objective": "global", "lm_loss": 1.34514, "bpb": 1.94062, "chunk_loss": 0.030213, "F": 0.18858, "G": 0.2178, "chunks": 193.11, "bytes_per_chunk": 5.303, "boundary_entropy": 0.15718, "router_noise": 0.0, "grad_norm": 0.4562, "lr": 0.0002866, "router_trainable": true, "sec_per_step": 0.255, "alpha": 0.03, "t": 438.6}
{"event": "train", "step": 2100, "phase": 1, "objective": "global", "lm_loss": 1.28558, "bpb": 1.8547, "chunk_loss": 0.030194, "F": 0.18683, "G": 0.22059, "chunks": 191.31, "bytes_per_chunk": 5.352, "boundary_entropy": 0.17322, "router_noise": 0.0, "grad_norm": 0.446, "lr": 0.0002848, "router_trainable": true, "sec_per_step": 0.17, "alpha": 0.03, "t": 455.6}
{"event": "train", "step": 2200, "phase": 1, "objective": "global", "lm_loss": 1.25024, "bpb": 1.80371, "chunk_loss": 0.030247, "F": 0.18738, "G": 0.22762, "chunks": 191.88, "bytes_per_chunk": 5.337, "boundary_entropy": 0.19255, "router_noise": 0.0, "grad_norm": 0.4125, "lr": 0.0002829, "router_trainable": true, "sec_per_step": 0.167, "alpha": 0.03, "t": 472.3}
{"event": "train", "step": 2300, "phase": 1, "objective": "global", "lm_loss": 1.32061, "bpb": 1.90524, "chunk_loss": 0.030256, "F": 0.18788, "G": 0.21839, "chunks": 192.39, "bytes_per_chunk": 5.323, "boundary_entropy": 0.1607, "router_noise": 0.0, "grad_norm": 0.4202, "lr": 0.0002809, "router_trainable": true, "sec_per_step": 0.169, "alpha": 0.03, "t": 489.2}
{"event": "train", "step": 2400, "phase": 1, "objective": "global", "lm_loss": 1.34238, "bpb": 1.93665, "chunk_loss": 0.0302, "F": 0.18675, "G": 0.21863, "chunks": 191.23, "bytes_per_chunk": 5.355, "boundary_entropy": 0.15497, "router_noise": 0.0, "grad_norm": 0.4463, "lr": 0.0002788, "router_trainable": true, "sec_per_step": 0.169, "alpha": 0.03, "t": 506.0}
{"event": "train", "step": 2500, "phase": 1, "objective": "global", "lm_loss": 1.24625, "bpb": 1.79795, "chunk_loss": 0.030196, "F": 0.18594, "G": 0.21787, "chunks": 190.41, "bytes_per_chunk": 5.378, "boundary_entropy": 0.16215, "router_noise": 0.0, "grad_norm": 0.3916, "lr": 0.0002765, "router_trainable": true, "sec_per_step": 0.169, "alpha": 0.03, "t": 522.9}
{"event": "train", "step": 2600, "phase": 1, "objective": "global", "lm_loss": 1.26683, "bpb": 1.82765, "chunk_loss": 0.030173, "F": 0.18469, "G": 0.22186, "chunks": 189.12, "bytes_per_chunk": 5.414, "boundary_entropy": 0.17232, "router_noise": 0.0, "grad_norm": 0.4288, "lr": 0.0002742, "router_trainable": true, "sec_per_step": 0.168, "alpha": 0.03, "t": 539.7}
{"event": "train", "step": 2700, "phase": 1, "objective": "global", "lm_loss": 1.26565, "bpb": 1.82595, "chunk_loss": 0.0301, "F": 0.18282, "G": 0.21971, "chunks": 187.2, "bytes_per_chunk": 5.47, "boundary_entropy": 0.17298, "router_noise": 0.0, "grad_norm": 0.3989, "lr": 0.0002718, "router_trainable": true, "sec_per_step": 0.169, "alpha": 0.03, "t": 556.6}
{"event": "train", "step": 2800, "phase": 1, "objective": "global", "lm_loss": 1.35248, "bpb": 1.95122, "chunk_loss": 0.030056, "F": 0.18144, "G": 0.22182, "chunks": 185.8, "bytes_per_chunk": 5.511, "boundary_entropy": 0.17381, "router_noise": 0.0, "grad_norm": 0.4161, "lr": 0.0002693, "router_trainable": true, "sec_per_step": 0.169, "alpha": 0.03, "t": 573.5}
{"event": "train", "step": 2900, "phase": 1, "objective": "global", "lm_loss": 1.27091, "bpb": 1.83353, "chunk_loss": 0.030186, "F": 0.18465, "G": 0.22329, "chunks": 189.08, "bytes_per_chunk": 5.416, "boundary_entropy": 0.17667, "router_noise": 0.0, "grad_norm": 0.3894, "lr": 0.0002667, "router_trainable": true, "sec_per_step": 0.169, "alpha": 0.03, "t": 590.3}
{"event": "probe", "step": 3000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3396.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.195139, "cps": 26.53125, "mean_p": 0.197788, "boundary_entropy": 0.086644, "bpb": 1.967429}, "de": {"chunks": 3744.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.182341, "cps": 29.25, "mean_p": 0.191849, "boundary_entropy": 0.092017, "bpb": 2.156103}, "vi": {"chunks": 4254.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.230757, "cps": 33.234375, "mean_p": 0.195555, "boundary_entropy": 0.106783, "bpb": 1.812191}, "tr": {"chunks": 3451.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.190631, "cps": 26.960938, "mean_p": 0.188813, "boundary_entropy": 0.127643, "bpb": 2.130241}, "fi": {"chunks": 3200.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.170113, "cps": 25.0, "mean_p": 0.184062, "boundary_entropy": 0.131524, "bpb": 2.305412}, "ru": {"chunks": 6634.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.350079, "cps": 51.828125, "mean_p": 0.200702, "boundary_entropy": 0.116818, "bpb": 1.269403}, "el": {"chunks": 7435.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.362276, "cps": 58.085938, "mean_p": 0.207225, "boundary_entropy": 0.110019, "bpb": 1.297518}, "hi": {"chunks": 6520.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.387519, "cps": 50.9375, "mean_p": 0.324733, "boundary_entropy": 0.473316, "bpb": 1.124119}, "ta": {"chunks": 9021.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.482535, "cps": 70.476562, "mean_p": 0.335965, "boundary_entropy": 0.520727, "bpb": 0.941703}, "th": {"chunks": 8007.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.497793, "cps": 62.554688, "mean_p": 0.31531, "boundary_entropy": 0.516249, "bpb": 1.079682}, "zh": {"chunks": 3117.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.527768, "cps": 24.351562, "mean_p": 0.216327, "boundary_entropy": 0.132414, "bpb": 2.980135}, "my": {"chunks": 9174.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.484167, "cps": 71.671875, "mean_p": 0.310883, "boundary_entropy": 0.511781, "bpb": 1.089346}}, "bpb": {"en": 1.96743, "de": 2.1561, "vi": 1.81219, "tr": 2.13024, "fi": 2.30541, "ru": 1.2694, "el": 1.29752, "hi": 1.12412, "ta": 0.9417, "th": 1.07968, "zh": 2.98014, "my": 1.08935}, "gini_cpc": 0.220536, "gini_cps": 0.223376, "bpic": 5.647742, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9344151507209199, "vi": 1.182526228626796, "tr": 0.9769016575547578, "fi": 0.8717551736982703, "ru": 1.7940010442208898, "el": 1.8565068941869751, "hi": 1.9858615240025412, "ta": 2.472780982046748, "th": 2.550969122736874, "zh": 2.704579788895284, "my": 2.4811430166697623}, "t": 610.1}
{"event": "eval", "step": 3000, "phase": 1, "bpb": {"en": 1.97466, "de": 2.14845, "vi": 1.81791, "tr": 2.11949, "fi": 2.32089, "ru": 1.26608, "el": 1.29477, "hi": 1.13064, "ta": 0.94237, "th": 1.07584, "zh": 2.98531, "my": 1.09974}, "bytes_per_chunk": {"en": 4.946, "de": 5.617, "vi": 5.693, "tr": 5.673, "fi": 6.014, "ru": 5.343, "el": 5.205, "hi": 6.504, "ta": 5.564, "th": 5.589, "zh": 5.136, "my": 5.859}, "bpb_mean": 1.68135, "t": 615.5}
{"event": "train", "step": 3000, "phase": 1, "objective": "global", "lm_loss": 1.31511, "bpb": 1.89731, "chunk_loss": 0.030225, "F": 0.1906, "G": 0.21505, "chunks": 195.17, "bytes_per_chunk": 5.247, "boundary_entropy": 0.14323, "router_noise": 0.0, "grad_norm": 0.4097, "lr": 0.000264, "router_trainable": true, "sec_per_step": 0.254, "alpha": 0.03, "t": 615.7}
{"event": "train", "step": 3100, "phase": 1, "objective": "global", "lm_loss": 1.31874, "bpb": 1.90255, "chunk_loss": 0.0302, "F": 0.18642, "G": 0.21319, "chunks": 190.89, "bytes_per_chunk": 5.364, "boundary_entropy": 0.14687, "router_noise": 0.0, "grad_norm": 0.3975, "lr": 0.0002612, "router_trainable": true, "sec_per_step": 0.169, "alpha": 0.03, "t": 632.6}
{"event": "train", "step": 3200, "phase": 1, "objective": "global", "lm_loss": 1.19674, "bpb": 1.72652, "chunk_loss": 0.030191, "F": 0.18768, "G": 0.22811, "chunks": 192.19, "bytes_per_chunk": 5.328, "boundary_entropy": 0.17791, "router_noise": 0.0, "grad_norm": 0.3633, "lr": 0.0002583, "router_trainable": true, "sec_per_step": 0.168, "alpha": 0.03, "t": 649.4}
{"event": "train", "step": 3300, "phase": 1, "objective": "global", "lm_loss": 1.27243, "bpb": 1.83573, "chunk_loss": 0.030212, "F": 0.18669, "G": 0.22652, "chunks": 191.17, "bytes_per_chunk": 5.356, "boundary_entropy": 0.18272, "router_noise": 0.0, "grad_norm": 0.4176, "lr": 0.0002553, "router_trainable": true, "sec_per_step": 0.167, "alpha": 0.03, "t": 666.2}
{"event": "train", "step": 3400, "phase": 1, "objective": "global", "lm_loss": 1.22889, "bpb": 1.77291, "chunk_loss": 0.030239, "F": 0.18996, "G": 0.22723, "chunks": 194.52, "bytes_per_chunk": 5.264, "boundary_entropy": 0.16652, "router_noise": 0.0, "grad_norm": 0.3833, "lr": 0.0002523, "router_trainable": true, "sec_per_step": 0.169, "alpha": 0.03, "t": 683.0}
{"event": "train", "step": 3500, "phase": 1, "objective": "global", "lm_loss": 1.30433, "bpb": 1.88175, "chunk_loss": 0.030236, "F": 0.18758, "G": 0.20957, "chunks": 192.08, "bytes_per_chunk": 5.331, "boundary_entropy": 0.13167, "router_noise": 0.0, "grad_norm": 0.3685, "lr": 0.0002492, "router_trainable": true, "sec_per_step": 0.167, "alpha": 0.03, "t": 699.8}
{"event": "train", "step": 3600, "phase": 1, "objective": "global", "lm_loss": 1.22198, "bpb": 1.76295, "chunk_loss": 0.030168, "F": 0.18648, "G": 0.217, "chunks": 190.95, "bytes_per_chunk": 5.363, "boundary_entropy": 0.15507, "router_noise": 0.0, "grad_norm": 0.3786, "lr": 0.000246, "router_trainable": true, "sec_per_step": 0.169, "alpha": 0.03, "t": 716.7}
{"event": "train", "step": 3700, "phase": 1, "objective": "global", "lm_loss": 1.13096, "bpb": 1.63162, "chunk_loss": 0.030064, "F": 0.18021, "G": 0.22944, "chunks": 184.53, "bytes_per_chunk": 5.549, "boundary_entropy": 0.19475, "router_noise": 0.0, "grad_norm": 0.377, "lr": 0.0002427, "router_trainable": true, "sec_per_step": 0.169, "alpha": 0.03, "t": 733.6}
{"event": "train", "step": 3800, "phase": 1, "objective": "global", "lm_loss": 1.22073, "bpb": 1.76114, "chunk_loss": 0.030192, "F": 0.18709, "G": 0.21768, "chunks": 191.58, "bytes_per_chunk": 5.345, "boundary_entropy": 0.15371, "router_noise": 0.0, "grad_norm": 0.3739, "lr": 0.0002394, "router_trainable": true, "sec_per_step": 0.168, "alpha": 0.03, "t": 750.4}
{"event": "train", "step": 3900, "phase": 1, "objective": "global", "lm_loss": 1.2353, "bpb": 1.78217, "chunk_loss": 0.030168, "F": 0.18628, "G": 0.2116, "chunks": 190.75, "bytes_per_chunk": 5.368, "boundary_entropy": 0.13451, "router_noise": 0.0, "grad_norm": 0.377, "lr": 0.000236, "router_trainable": true, "sec_per_step": 0.168, "alpha": 0.03, "t": 767.2}
{"event": "probe", "step": 4000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3373.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.193817, "cps": 26.351562, "mean_p": 0.19353, "boundary_entropy": 0.077396, "bpb": 1.885479}, "de": {"chunks": 3669.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.178688, "cps": 28.664062, "mean_p": 0.185152, "boundary_entropy": 0.078201, "bpb": 2.067095}, "vi": {"chunks": 4203.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.22799, "cps": 32.835938, "mean_p": 0.190089, "boundary_entropy": 0.093461, "bpb": 1.71251}, "tr": {"chunks": 3162.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.174667, "cps": 24.703125, "mean_p": 0.17625, "boundary_entropy": 0.115248, "bpb": 2.041998}, "fi": {"chunks": 3120.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.16586, "cps": 24.375, "mean_p": 0.1793, "boundary_entropy": 0.127302, "bpb": 2.199986}, "ru": {"chunks": 6561.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.346227, "cps": 51.257812, "mean_p": 0.195788, "boundary_entropy": 0.093831, "bpb": 1.212785}, "el": {"chunks": 7188.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.350241, "cps": 56.15625, "mean_p": 0.200456, "boundary_entropy": 0.10184, "bpb": 1.228703}, "hi": {"chunks": 7206.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.428291, "cps": 56.296875, "mean_p": 0.343813, "boundary_entropy": 0.487064, "bpb": 1.048547}, "ta": {"chunks": 7731.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.413533, "cps": 60.398438, "mean_p": 0.341166, "boundary_entropy": 0.533359, "bpb": 0.895542}, "th": {"chunks": 7814.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.485794, "cps": 61.046875, "mean_p": 0.324667, "boundary_entropy": 0.529202, "bpb": 1.022608}, "zh": {"chunks": 3035.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.513884, "cps": 23.710938, "mean_p": 0.206047, "boundary_entropy": 0.115493, "bpb": 2.870669}, "my": {"chunks": 7891.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.416456, "cps": 61.648438, "mean_p": 0.319325, "boundary_entropy": 0.531158, "bpb": 1.053043}}, "bpb": {"en": 1.88548, "de": 2.06709, "vi": 1.71251, "tr": 2.042, "fi": 2.19999, "ru": 1.21278, "el": 1.2287, "hi": 1.04855, "ta": 0.89554, "th": 1.02261, "zh": 2.87067, "my": 1.05304}, "gini_cpc": 0.214983, "gini_cps": 0.204924, "bpic": 5.908595, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9219409034438267, "vi": 1.1763160618099457, "tr": 0.9011956633369427, "fi": 0.8557570577030367, "ru": 1.7863584244587039, "el": 1.807070109231789, "hi": 2.20976944347407, "ta": 2.133624466336554, "th": 2.5064562050424932, "zh": 2.6513864457587286, "my": 2.1487032777334143}, "t": 786.9}
{"event": "eval", "step": 4000, "phase": 1, "bpb": {"en": 1.88971, "de": 2.06322, "vi": 1.72223, "tr": 2.03349, "fi": 2.21689, "ru": 1.21009, "el": 1.22442, "hi": 1.05707, "ta": 0.89751, "th": 1.02104, "zh": 2.88086, "my": 1.06074}, "bytes_per_chunk": {"en": 4.957, "de": 5.647, "vi": 5.718, "tr": 6.111, "fi": 6.064, "ru": 5.368, "el": 5.29, "hi": 6.106, "ta": 6.363, "th": 5.871, "zh": 5.224, "my": 6.952}, "bpb_mean": 1.60644, "t": 792.2}
{"event": "train", "step": 4000, "phase": 1, "objective": "global", "lm_loss": 1.30746, "bpb": 1.88627, "chunk_loss": 0.030122, "F": 0.18167, "G": 0.21696, "chunks": 186.03, "bytes_per_chunk": 5.504, "boundary_entropy": 0.1625, "router_noise": 0.0, "grad_norm": 0.3973, "lr": 0.0002325, "router_trainable": true, "sec_per_step": 0.252, "alpha": 0.03, "t": 792.4}
{"event": "train", "step": 4100, "phase": 1, "objective": "global", "lm_loss": 1.2844, "bpb": 1.85299, "chunk_loss": 0.030186, "F": 0.1839, "G": 0.20983, "chunks": 188.31, "bytes_per_chunk": 5.438, "boundary_entropy": 0.14041, "router_noise": 0.0, "grad_norm": 0.429, "lr": 0.000229, "router_trainable": true, "sec_per_step": 0.169, "alpha": 0.03, "t": 809.3}
{"event": "train", "step": 4200, "phase": 1, "objective": "global", "lm_loss": 1.22054, "bpb": 1.76087, "chunk_loss": 0.030143, "F": 0.18626, "G": 0.21149, "chunks": 190.73, "bytes_per_chunk": 5.369, "boundary_entropy": 0.13389, "router_noise": 0.0, "grad_norm": 0.3708, "lr": 0.0002254, "router_trainable": true, "sec_per_step": 0.168, "alpha": 0.03, "t": 826.1}
{"event": "train", "step": 4300, "phase": 1, "objective": "global", "lm_loss": 1.19403, "bpb": 1.72263, "chunk_loss": 0.03009, "F": 0.1814, "G": 0.21715, "chunks": 185.75, "bytes_per_chunk": 5.513, "boundary_entropy": 0.16296, "router_noise": 0.0, "grad_norm": 0.3627, "lr": 0.0002217, "router_trainable": true, "sec_per_step": 0.17, "alpha": 0.03, "t": 843.0}
{"event": "phase_transition", "step": 4400, "phase": 2, "objective": "parity_a", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0, "t": 859.7}
{"event": "train", "step": 4400, "phase": 2, "objective": "parity_a", "lm_loss": 1.14721, "bpb": 1.65508, "chunk_loss": 0.034083, "F": 0.18608, "G": 0.21619, "chunks": 190.55, "bytes_per_chunk": 5.374, "boundary_entropy": 0.15406, "router_noise": 0.0, "grad_norm": 0.3433, "lr": 0.0002181, "router_trainable": true, "sec_per_step": 0.168, "n_target_mean": 9.7474, "alpha": 0.03, "t": 859.8}
{"event": "train", "step": 4500, "phase": 2, "objective": "parity_a", "lm_loss": 1.22846, "bpb": 1.7723, "chunk_loss": 0.030726, "F": 0.16364, "G": 0.18007, "chunks": 167.56, "bytes_per_chunk": 6.111, "boundary_entropy": 0.12974, "router_noise": 0.0, "grad_norm": 0.3776, "lr": 0.0002143, "router_trainable": true, "sec_per_step": 0.169, "n_target_mean": 8.995, "alpha": 0.03, "t": 876.7}
{"event": "train", "step": 4600, "phase": 2, "objective": "parity_a", "lm_loss": 1.17858, "bpb": 1.70033, "chunk_loss": 0.031001, "F": 0.16672, "G": 0.17681, "chunks": 170.72, "bytes_per_chunk": 5.998, "boundary_entropy": 0.11977, "router_noise": 0.0, "grad_norm": 0.3638, "lr": 0.0002105, "router_trainable": true, "sec_per_step": 0.168, "n_target_mean": 9.4029, "alpha": 0.03, "t": 893.5}
{"event": "train", "step": 4700, "phase": 2, "objective": "parity_a", "lm_loss": 1.10376, "bpb": 1.5924, "chunk_loss": 0.030768, "F": 0.15515, "G": 0.17391, "chunks": 158.88, "bytes_per_chunk": 6.445, "boundary_entropy": 0.13956, "router_noise": 0.0, "grad_norm": 0.3482, "lr": 0.0002067, "router_trainable": true, "sec_per_step": 0.168, "n_target_mean": 10.3158, "alpha": 0.03, "t": 910.3}
{"event": "train", "step": 4800, "phase": 2, "objective": "parity_a", "lm_loss": 1.23032, "bpb": 1.77497, "chunk_loss": 0.030831, "F": 0.17372, "G": 0.17899, "chunks": 177.89, "bytes_per_chunk": 5.756, "boundary_entropy": 0.10711, "router_noise": 0.0, "grad_norm": 0.3665, "lr": 0.0002029, "router_trainable": true, "sec_per_step": 0.168, "n_target_mean": 8.5167, "alpha": 0.03, "t": 927.1}
{"event": "train", "step": 4900, "phase": 2, "objective": "parity_a", "lm_loss": 1.21785, "bpb": 1.75699, "chunk_loss": 0.030654, "F": 0.17699, "G": 0.17979, "chunks": 181.23, "bytes_per_chunk": 5.65, "boundary_entropy": 0.09649, "router_noise": 0.0, "grad_norm": 0.3701, "lr": 0.000199, "router_trainable": true, "sec_per_step": 0.168, "n_target_mean": 7.9172, "alpha": 0.03, "t": 943.9}
{"event": "probe", "step": 5000, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 3346.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.192266, "cps": 26.140625, "mean_p": 0.185953, "boundary_entropy": 0.07942, "bpb": 1.784837}, "de": {"chunks": 3622.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.176399, "cps": 28.296875, "mean_p": 0.175194, "boundary_entropy": 0.071646, "bpb": 1.953189}, "vi": {"chunks": 4189.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.227231, "cps": 32.726562, "mean_p": 0.172959, "boundary_entropy": 0.076109, "bpb": 1.621726}, "tr": {"chunks": 2834.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.156549, "cps": 22.140625, "mean_p": 0.153283, "boundary_entropy": 0.091726, "bpb": 1.900403}, "fi": {"chunks": 3075.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.163468, "cps": 24.023438, "mean_p": 0.164381, "boundary_entropy": 0.101122, "bpb": 2.080522}, "ru": {"chunks": 6176.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.32591, "cps": 48.25, "mean_p": 0.158626, "boundary_entropy": 0.111463, "bpb": 1.178087}, "el": {"chunks": 6624.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.32276, "cps": 51.75, "mean_p": 0.1619, "boundary_entropy": 0.106676, "bpb": 1.199728}, "hi": {"chunks": 2845.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.169094, "cps": 22.226562, "mean_p": 0.098586, "boundary_entropy": 0.19918, "bpb": 1.043853}, "ta": {"chunks": 2335.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.1249, "cps": 18.242188, "mean_p": 0.124165, "boundary_entropy": 0.268, "bpb": 0.889755}, "th": {"chunks": 3127.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.194405, "cps": 24.429688, "mean_p": 0.120081, "boundary_entropy": 0.241983, "bpb": 1.005999}, "zh": {"chunks": 3087.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.522689, "cps": 24.117188, "mean_p": 0.212197, "boundary_entropy": 0.116897, "bpb": 2.792465}, "my": {"chunks": 2447.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.129143, "cps": 19.117188, "mean_p": 0.175289, "boundary_entropy": 0.376523, "bpb": 1.021533}}, "bpb": {"en": 1.78484, "de": 1.95319, "vi": 1.62173, "tr": 1.9004, "fi": 2.08052, "ru": 1.17809, "el": 1.19973, "hi": 1.04385, "ta": 0.88975, "th": 1.006, "zh": 2.79246, "my": 1.02153}, "gini_cpc": 0.237869, "gini_cps": 0.18106, "bpic": 8.780767, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9174749646371305, "vi": 1.181858279465614, "tr": 0.8142307092519233, "fi": 0.8502202025159639, "ru": 1.695103640467017, "el": 1.678717671561704, "hi": 0.8794788126071894, "ta": 0.6496203168265485, "th": 1.0111253518878804, "zh": 2.7185753230173697, "my": 0.6716898363798314}, "t": 963.4}
{"event": "eval", "step": 5000, "phase": 2, "bpb": {"en": 1.78608, "de": 1.9546, "vi": 1.63433, "tr": 1.89247, "fi": 2.09406, "ru": 1.17611, "el": 1.19576, "hi": 1.05032, "ta": 0.893, "th": 1.00419, "zh": 2.80266, "my": 1.02968}, "bytes_per_chunk": {"en": 4.996, "de": 5.718, "vi": 5.736, "tr": 6.718, "fi": 6.098, "ru": 5.77, "el": 5.675, "hi": 13.804, "ta": 20.605, "th": 14.509, "zh": 5.208, "my": 20.701}, "bpb_mean": 1.54277, "t": 968.5}
{"event": "train", "step": 5000, "phase": 2, "objective": "parity_a", "lm_loss": 1.1822, "bpb": 1.70556, "chunk_loss": 0.030811, "F": 0.16685, "G": 0.17268, "chunks": 170.86, "bytes_per_chunk": 5.993, "boundary_entropy": 0.11583, "router_noise": 0.0, "grad_norm": 0.3634, "lr": 0.000195, "router_trainable": true, "sec_per_step": 0.248, "n_target_mean": 9.2185, "alpha": 0.03, "t": 968.7}
{"event": "train", "step": 5100, "phase": 2, "objective": "parity_a", "lm_loss": 1.19111, "bpb": 1.71841, "chunk_loss": 0.030731, "F": 0.16774, "G": 0.17296, "chunks": 171.77, "bytes_per_chunk": 5.962, "boundary_entropy": 0.11081, "router_noise": 0.0, "grad_norm": 0.3631, "lr": 0.0001911, "router_trainable": true, "sec_per_step": 0.167, "n_target_mean": 9.0005, "alpha": 0.03, "t": 985.4}
{"event": "train", "step": 5200, "phase": 2, "objective": "parity_a", "lm_loss": 1.1179, "bpb": 1.61278, "chunk_loss": 0.030696, "F": 0.15654, "G": 0.16711, "chunks": 160.3, "bytes_per_chunk": 6.388, "boundary_entropy": 0.1275, "router_noise": 0.0, "grad_norm": 0.3588, "lr": 0.0001871, "router_trainable": true, "sec_per_step": 0.166, "n_target_mean": 9.8194, "alpha": 0.03, "t": 1002.0}
{"event": "train", "step": 5300, "phase": 2, "objective": "parity_a", "lm_loss": 1.14699, "bpb": 1.65475, "chunk_loss": 0.030784, "F": 0.16443, "G": 0.17005, "chunks": 168.38, "bytes_per_chunk": 6.082, "boundary_entropy": 0.11425, "router_noise": 0.0, "grad_norm": 0.3773, "lr": 0.0001831, "router_trainable": true, "sec_per_step": 0.167, "n_target_mean": 9.3411, "alpha": 0.03, "t": 1018.7}
{"event": "train", "step": 5400, "phase": 2, "objective": "parity_a", "lm_loss": 1.20309, "bpb": 1.73569, "chunk_loss": 0.030883, "F": 0.17143, "G": 0.17561, "chunks": 175.55, "bytes_per_chunk": 5.833, "boundary_entropy": 0.11355, "router_noise": 0.0, "grad_norm": 0.3703, "lr": 0.0001791, "router_trainable": true, "sec_per_step": 0.169, "n_target_mean": 9.0989, "alpha": 0.03, "t": 1035.6}
{"event": "train", "step": 5500, "phase": 2, "objective": "parity_a", "lm_loss": 1.15851, "bpb": 1.67138, "chunk_loss": 0.030635, "F": 0.16145, "G": 0.16834, "chunks": 165.33, "bytes_per_chunk": 6.194, "boundary_entropy": 0.11583, "router_noise": 0.0, "grad_norm": 0.3856, "lr": 0.0001751, "router_trainable": true, "sec_per_step": 0.167, "n_target_mean": 9.3685, "alpha": 0.03, "t": 1052.3}
{"event": "train", "step": 5600, "phase": 2, "objective": "parity_a", "lm_loss": 1.10691, "bpb": 1.59694, "chunk_loss": 0.030776, "F": 0.16679, "G": 0.1697, "chunks": 170.8, "bytes_per_chunk": 5.995, "boundary_entropy": 0.10757, "router_noise": 0.0, "grad_norm": 0.3564, "lr": 0.0001711, "router_trainable": true, "sec_per_step": 0.167, "n_target_mean": 9.17, "alpha": 0.03, "t": 1069.0}
{"event": "train", "step": 5700, "phase": 2, "objective": "parity_a", "lm_loss": 1.10523, "bpb": 1.5945, "chunk_loss": 0.030687, "F": 0.15823, "G": 0.1679, "chunks": 162.03, "bytes_per_chunk": 6.32, "boundary_entropy": 0.12469, "router_noise": 0.0, "grad_norm": 0.3468, "lr": 0.000167, "router_trainable": true, "sec_per_step": 0.168, "n_target_mean": 10.0188, "alpha": 0.03, "t": 1085.7}
{"event": "train", "step": 5800, "phase": 2, "objective": "parity_a", "lm_loss": 1.08912, "bpb": 1.57127, "chunk_loss": 0.030508, "F": 0.15762, "G": 0.16506, "chunks": 161.41, "bytes_per_chunk": 6.344, "boundary_entropy": 0.11572, "router_noise": 0.0, "grad_norm": 0.3555, "lr": 0.000163, "router_trainable": true, "sec_per_step": 0.167, "n_target_mean": 9.5687, "alpha": 0.03, "t": 1102.5}
{"event": "train", "step": 5900, "phase": 2, "objective": "parity_a", "lm_loss": 1.14326, "bpb": 1.64938, "chunk_loss": 0.030507, "F": 0.16574, "G": 0.16708, "chunks": 169.72, "bytes_per_chunk": 6.034, "boundary_entropy": 0.10563, "router_noise": 0.0, "grad_norm": 0.3656, "lr": 0.0001589, "router_trainable": true, "sec_per_step": 0.166, "n_target_mean": 8.6873, "alpha": 0.03, "t": 1119.1}
{"event": "probe", "step": 6000, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 3345.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.192208, "cps": 26.132812, "mean_p": 0.18668, "boundary_entropy": 0.071969, "bpb": 1.734989}, "de": {"chunks": 3655.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.178006, "cps": 28.554688, "mean_p": 0.175614, "boundary_entropy": 0.065765, "bpb": 1.887041}, "vi": {"chunks": 4188.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.227177, "cps": 32.71875, "mean_p": 0.167453, "boundary_entropy": 0.079986, "bpb": 1.540275}, "tr": {"chunks": 2899.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.160139, "cps": 22.648438, "mean_p": 0.158552, "boundary_entropy": 0.099573, "bpb": 1.830384}, "fi": {"chunks": 3098.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.164691, "cps": 24.203125, "mean_p": 0.163901, "boundary_entropy": 0.092793, "bpb": 2.015059}, "ru": {"chunks": 6048.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.319156, "cps": 47.25, "mean_p": 0.153019, "boundary_entropy": 0.105238, "bpb": 1.146697}, "el": {"chunks": 6293.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.306632, "cps": 49.164062, "mean_p": 0.148271, "boundary_entropy": 0.106886, "bpb": 1.163766}, "hi": {"chunks": 2584.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.153581, "cps": 20.1875, "mean_p": 0.103005, "boundary_entropy": 0.211222, "bpb": 1.00887}, "ta": {"chunks": 2637.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.141054, "cps": 20.601562, "mean_p": 0.117245, "boundary_entropy": 0.240333, "bpb": 0.859044}, "th": {"chunks": 3203.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.19913, "cps": 25.023438, "mean_p": 0.115926, "boundary_entropy": 0.225198, "bpb": 0.973042}, "zh": {"chunks": 3037.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.514223, "cps": 23.726562, "mean_p": 0.203201, "boundary_entropy": 0.092573, "bpb": 2.724377}, "my": {"chunks": 2154.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.11368, "cps": 16.828125, "mean_p": 0.163091, "boundary_entropy": 0.359921, "bpb": 0.992172}}, "bpb": {"en": 1.73499, "de": 1.88704, "vi": 1.54028, "tr": 1.83038, "fi": 2.01506, "ru": 1.1467, "el": 1.16377, "hi": 1.00887, "ta": 0.85904, "th": 0.97304, "zh": 2.72438, "my": 0.99217}, "gini_cpc": 0.23514, "gini_cps": 0.178428, "bpic": 8.895969, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9261108498864019, "vi": 1.1819293822313521, "tr": 0.8331547257360659, "fi": 0.8568356507367482, "ru": 1.6604682134955098, "el": 1.5953092549122945, "hi": 0.7990343219237868, "ta": 0.7338590568207849, "th": 1.0360098189132565, "zh": 2.6753422452503273, "my": 0.5914394931230245}, "t": 1138.6}
{"event": "eval", "step": 6000, "phase": 2, "bpb": {"en": 1.73487, "de": 1.89052, "vi": 1.55576, "tr": 1.82228, "fi": 2.02855, "ru": 1.1447, "el": 1.1613, "hi": 1.0142, "ta": 0.86256, "th": 0.97017, "zh": 2.73392, "my": 1.00026}, "bytes_per_chunk": {"en": 4.995, "de": 5.692, "vi": 5.741, "tr": 6.582, "fi": 6.131, "ru": 5.913, "el": 5.992, "hi": 15.44, "ta": 17.84, "th": 13.868, "zh": 5.244, "my": 21.285}, "bpb_mean": 1.49326, "t": 1143.6}
{"event": "train", "step": 6000, "phase": 2, "objective": "parity_a", "lm_loss": 1.12651, "bpb": 1.62521, "chunk_loss": 0.030444, "F": 0.16203, "G": 0.16872, "chunks": 165.92, "bytes_per_chunk": 6.172, "boundary_entropy": 0.11121, "router_noise": 0.0, "grad_norm": 0.3943, "lr": 0.0001549, "router_trainable": true, "sec_per_step": 0.247, "n_target_mean": 9.0576, "alpha": 0.03, "t": 1143.7}
{"event": "train", "step": 6100, "phase": 2, "objective": "parity_a", "lm_loss": 1.05528, "bpb": 1.52244, "chunk_loss": 0.030625, "F": 0.16411, "G": 0.16555, "chunks": 168.05, "bytes_per_chunk": 6.094, "boundary_entropy": 0.11365, "router_noise": 0.0, "grad_norm": 0.3491, "lr": 0.0001509, "router_trainable": true, "sec_per_step": 0.169, "n_target_mean": 9.3786, "alpha": 0.03, "t": 1160.6}
{"event": "train", "step": 6200, "phase": 2, "objective": "parity_a", "lm_loss": 1.16434, "bpb": 1.67979, "chunk_loss": 0.030629, "F": 0.17198, "G": 0.17034, "chunks": 176.11, "bytes_per_chunk": 5.815, "boundary_entropy": 0.09739, "router_noise": 0.0, "grad_norm": 0.3629, "lr": 0.0001469, "router_trainable": true, "sec_per_step": 0.169, "n_target_mean": 8.4082, "alpha": 0.03, "t": 1177.5}
{"event": "train", "step": 6300, "phase": 2, "objective": "parity_a", "lm_loss": 1.09016, "bpb": 1.57277, "chunk_loss": 0.030672, "F": 0.17416, "G": 0.17142, "chunks": 178.34, "bytes_per_chunk": 5.742, "boundary_entropy": 0.09579, "router_noise": 0.0, "grad_norm": 0.3589, "lr": 0.0001429, "router_trainable": true, "sec_per_step": 0.167, "n_target_mean": 8.5209, "alpha": 0.03, "t": 1194.2}
{"event": "train", "step": 6400, "phase": 2, "objective": "parity_a", "lm_loss": 1.00192, "bpb": 1.44547, "chunk_loss": 0.030715, "F": 0.15146, "G": 0.16185, "chunks": 155.09, "bytes_per_chunk": 6.602, "boundary_entropy": 0.12944, "router_noise": 0.0, "grad_norm": 0.3526, "lr": 0.0001389, "router_trainable": true, "sec_per_step": 0.167, "n_target_mean": 10.5328, "alpha": 0.03, "t": 1210.9}
{"event": "train", "step": 6500, "phase": 2, "objective": "parity_a", "lm_loss": 1.06393, "bpb": 1.53493, "chunk_loss": 0.030401, "F": 0.15184, "G": 0.15886, "chunks": 155.48, "bytes_per_chunk": 6.586, "boundary_entropy": 0.11492, "router_noise": 0.0, "grad_norm": 0.3762, "lr": 0.000135, "router_trainable": true, "sec_per_step": 0.168, "n_target_mean": 9.6646, "alpha": 0.03, "t": 1227.7}
{"event": "train", "step": 6600, "phase": 2, "objective": "parity_a", "lm_loss": 1.11616, "bpb": 1.61028, "chunk_loss": 0.030547, "F": 0.17479, "G": 0.17454, "chunks": 178.98, "bytes_per_chunk": 5.721, "boundary_entropy": 0.10041, "router_noise": 0.0, "grad_norm": 0.353, "lr": 0.000131, "router_trainable": true, "sec_per_step": 0.168, "n_target_mean": 8.5192, "alpha": 0.03, "t": 1244.5}
{"event": "train", "step": 6700, "phase": 2, "objective": "parity_a", "lm_loss": 1.10356, "bpb": 1.5921, "chunk_loss": 0.030481, "F": 0.16589, "G": 0.1691, "chunks": 169.88, "bytes_per_chunk": 6.028, "boundary_entropy": 0.10954, "router_noise": 0.0, "grad_norm": 0.3592, "lr": 0.0001271, "router_trainable": true, "sec_per_step": 0.168, "n_target_mean": 8.8379, "alpha": 0.03, "t": 1261.3}
{"event": "train", "step": 6800, "phase": 2, "objective": "parity_a", "lm_loss": 1.11299, "bpb": 1.6057, "chunk_loss": 0.030657, "F": 0.15939, "G": 0.16258, "chunks": 163.22, "bytes_per_chunk": 6.274, "boundary_entropy": 0.11377, "router_noise": 0.0, "grad_norm": 0.3845, "lr": 0.0001233, "router_trainable": true, "sec_per_step": 0.168, "n_target_mean": 9.7731, "alpha": 0.03, "t": 1278.1}
{"event": "train", "step": 6900, "phase": 2, "objective": "parity_a", "lm_loss": 1.09321, "bpb": 1.57717, "chunk_loss": 0.030618, "F": 0.16191, "G": 0.16273, "chunks": 165.8, "bytes_per_chunk": 6.176, "boundary_entropy": 0.10807, "router_noise": 0.0, "grad_norm": 0.3555, "lr": 0.0001195, "router_trainable": true, "sec_per_step": 0.166, "n_target_mean": 9.4099, "alpha": 0.03, "t": 1294.7}
{"event": "probe", "step": 7000, "phase": 2, "objective": "parity_a", "router_trainable": true, "per_lang": {"en": {"chunks": 3325.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.191059, "cps": 25.976562, "mean_p": 0.184749, "boundary_entropy": 0.067371, "bpb": 1.696241}, "de": {"chunks": 3621.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.17635, "cps": 28.289062, "mean_p": 0.173135, "boundary_entropy": 0.061089, "bpb": 1.845986}, "vi": {"chunks": 4183.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.226905, "cps": 32.679688, "mean_p": 0.165105, "boundary_entropy": 0.078931, "bpb": 1.491621}, "tr": {"chunks": 2858.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.157874, "cps": 22.328125, "mean_p": 0.156081, "boundary_entropy": 0.092481, "bpb": 1.778331}, "fi": {"chunks": 3101.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.16485, "cps": 24.226562, "mean_p": 0.161794, "boundary_entropy": 0.082804, "bpb": 1.95324}, "ru": {"chunks": 5994.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.316306, "cps": 46.828125, "mean_p": 0.148562, "boundary_entropy": 0.106889, "bpb": 1.118298}, "el": {"chunks": 6163.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.300297, "cps": 48.148438, "mean_p": 0.145371, "boundary_entropy": 0.103864, "bpb": 1.127361}, "hi": {"chunks": 2748.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.163328, "cps": 21.46875, "mean_p": 0.099947, "boundary_entropy": 0.192963, "bpb": 0.979356}, "ta": {"chunks": 2778.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.148596, "cps": 21.703125, "mean_p": 0.111407, "boundary_entropy": 0.220263, "bpb": 0.838453}, "th": {"chunks": 2857.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.177619, "cps": 22.320312, "mean_p": 0.098814, "boundary_entropy": 0.191248, "bpb": 0.939691}, "zh": {"chunks": 3062.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.518456, "cps": 23.921875, "mean_p": 0.204016, "boundary_entropy": 0.09449, "bpb": 2.65647}, "my": {"chunks": 3026.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.1597, "cps": 23.640625, "mean_p": 0.171174, "boundary_entropy": 0.356638, "bpb": 0.95798}}, "bpb": {"en": 1.69624, "de": 1.84599, "vi": 1.49162, "tr": 1.77833, "fi": 1.95324, "ru": 1.1183, "el": 1.12736, "hi": 0.97936, "ta": 0.83845, "th": 0.93969, "zh": 2.65647, "my": 0.95798}, "gini_cpc": 0.212126, "gini_cps": 0.153533, "bpic": 8.77896, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9230146373580179, "vi": 1.1876191536612075, "tr": 0.8263121594518252, "fi": 0.8628242713529878, "ru": 1.6555412065784514, "el": 1.571751172991681, "hi": 0.8548582868761802, "ta": 0.7777486079467189, "th": 0.9296546495240278, "zh": 2.713589900929092, "my": 0.8358686139033461}, "t": 1314.3}
{"event": "eval", "step": 7000, "phase": 2, "bpb": {"en": 1.69567, "de": 1.84882, "vi": 1.50711, "tr": 1.76998, "fi": 1.96765, "ru": 1.11666, "el": 1.12498, "hi": 0.98623, "ta": 0.84266, "th": 0.93875, "zh": 2.66625, "my": 0.96524}, "bytes_per_chunk": {"en": 5.008, "de": 5.739, "vi": 5.74, "tr": 6.64, "fi": 6.075, "ru": 5.973, "el": 6.097, "hi": 14.476, "ta": 17.304, "th": 15.261, "zh": 5.23, "my": 17.02}, "bpb_mean": 1.4525, "t": 1319.3}
{"event": "train", "step": 7000, "phase": 2, "objective": "parity_a", "lm_loss": 1.08538, "bpb": 1.56588, "chunk_loss": 0.030818, "F": 0.16423, "G": 0.16353, "chunks": 168.17, "bytes_per_chunk": 6.089, "boundary_entropy": 0.09967, "router_noise": 0.0, "grad_norm": 0.3688, "lr": 0.0001157, "router_trainable": true, "sec_per_step": 0.247, "n_target_mean": 9.5778, "alpha": 0.03, "t": 1319.5}
{"event": "train", "step": 7100, "phase": 2, "objective": "parity_a", "lm_loss": 1.06418, "bpb": 1.53529, "chunk_loss": 0.030431, "F": 0.15598, "G": 0.16024, "chunks": 159.72, "bytes_per_chunk": 6.411, "boundary_entropy": 0.10303, "router_noise": 0.0, "grad_norm": 0.3628, "lr": 0.0001119, "router_trainable": true, "sec_per_step": 0.166, "n_target_mean": 9.4019, "alpha": 0.03, "t": 1336.1}
{"event": "train", "step": 7200, "phase": 2, "objective": "parity_a", "lm_loss": 1.11393, "bpb": 1.60706, "chunk_loss": 0.030571, "F": 0.16553, "G": 0.16352, "chunks": 169.5, "bytes_per_chunk": 6.041, "boundary_entropy": 0.09686, "router_noise": 0.0, "grad_norm": 0.3519, "lr": 0.0001083, "router_trainable": true, "sec_per_step": 0.166, "n_target_mean": 8.9796, "alpha": 0.03, "t": 1352.7}
{"event": "train", "step": 7300, "phase": 2, "objective": "parity_a", "lm_loss": 1.1273, "bpb": 1.62635, "chunk_loss": 0.030462, "F": 0.16238, "G": 0.16724, "chunks": 166.28, "bytes_per_chunk": 6.158, "boundary_entropy": 0.10523, "router_noise": 0.0, "grad_norm": 0.3731, "lr": 0.0001046, "router_trainable": true, "sec_per_step": 0.166, "n_target_mean": 9.1966, "alpha": 0.03, "t": 1369.3}
{"event": "train", "step": 7400, "phase": 2, "objective": "parity_a", "lm_loss": 1.09065, "bpb": 1.57348, "chunk_loss": 0.030512, "F": 0.16385, "G": 0.16513, "chunks": 167.78, "bytes_per_chunk": 6.103, "boundary_entropy": 0.10394, "router_noise": 0.0, "grad_norm": 0.3744, "lr": 0.000101, "router_trainable": true, "sec_per_step": 0.168, "n_target_mean": 9.204, "alpha": 0.03, "t": 1386.1}
{"event": "train", "step": 7500, "phase": 2, "objective": "parity_a", "lm_loss": 1.07525, "bpb": 1.55126, "chunk_loss": 0.030635, "F": 0.16306, "G": 0.16187, "chunks": 166.97, "bytes_per_chunk": 6.133, "boundary_entropy": 0.10448, "router_noise": 0.0, "grad_norm": 0.359, "lr": 9.75e-05, "router_trainable": true, "sec_per_step": 0.167, "n_target_mean": 9.4226, "alpha": 0.03, "t": 1402.8}
{"event": "train", "step": 7600, "phase": 2, "objective": "parity_a", "lm_loss": 1.13476, "bpb": 1.63711, "chunk_loss": 0.030455, "F": 0.16638, "G": 0.16719, "chunks": 170.38, "bytes_per_chunk": 6.01, "boundary_entropy": 0.09906, "router_noise": 0.0, "grad_norm": 0.364, "lr": 9.4e-05, "router_trainable": true, "sec_per_step": 0.167, "n_target_mean": 8.7726, "alpha": 0.03, "t": 1419.5}
{"event": "phase_transition", "step": 7700, "phase": 3, "objective": "none", "router_trainable": false, "encoder_trainable": true, "frozen_router_params": 295680, "t": 1436.0}
{"event": "train", "step": 7700, "phase": 3, "objective": "none", "lm_loss": 1.08322, "bpb": 1.56276, "chunk_loss": 0.0, "F": 0.1647, "G": 0.16534, "chunks": 168.66, "bytes_per_chunk": 6.072, "boundary_entropy": 0.10121, "router_noise": 0.0, "grad_norm": 0.3556, "lr": 9.06e-05, "router_trainable": false, "sec_per_step": 0.166, "alpha": 0.0, "t": 1436.2}
{"event": "train", "step": 7800, "phase": 3, "objective": "none", "lm_loss": 1.11496, "bpb": 1.60855, "chunk_loss": 0.0, "F": 0.1701, "G": 0.1712, "chunks": 174.19, "bytes_per_chunk": 5.879, "boundary_entropy": 0.09033, "router_noise": 0.0, "grad_norm": 0.3671, "lr": 8.73e-05, "router_trainable": false, "sec_per_step": 0.167, "alpha": 0.0, "t": 1452.9}
{"event": "train", "step": 7900, "phase": 3, "objective": "none", "lm_loss": 1.10278, "bpb": 1.59097, "chunk_loss": 0.0, "F": 0.16791, "G": 0.16847, "chunks": 171.94, "bytes_per_chunk": 5.956, "boundary_entropy": 0.08797, "router_noise": 0.0, "grad_norm": 0.3855, "lr": 8.4e-05, "router_trainable": false, "sec_per_step": 0.165, "alpha": 0.0, "t": 1469.4}
{"event": "probe", "step": 8000, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3326.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.191116, "cps": 25.984375, "mean_p": 0.185307, "boundary_entropy": 0.064515, "bpb": 1.670681}, "de": {"chunks": 3632.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.176886, "cps": 28.375, "mean_p": 0.174194, "boundary_entropy": 0.058951, "bpb": 1.81472}, "vi": {"chunks": 4167.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.226037, "cps": 32.554688, "mean_p": 0.165455, "boundary_entropy": 0.077143, "bpb": 1.455411}, "tr": {"chunks": 2807.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.155057, "cps": 21.929688, "mean_p": 0.154366, "boundary_entropy": 0.092366, "bpb": 1.745548}, "fi": {"chunks": 3041.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.161661, "cps": 23.757812, "mean_p": 0.158105, "boundary_entropy": 0.080118, "bpb": 1.91541}, "ru": {"chunks": 5975.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.315303, "cps": 46.679688, "mean_p": 0.15415, "boundary_entropy": 0.100426, "bpb": 1.088725}, "el": {"chunks": 6214.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.302782, "cps": 48.546875, "mean_p": 0.155594, "boundary_entropy": 0.100593, "bpb": 1.097468}, "hi": {"chunks": 2664.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.158336, "cps": 20.8125, "mean_p": 0.099083, "boundary_entropy": 0.190339, "bpb": 0.96114}, "ta": {"chunks": 2726.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.145814, "cps": 21.296875, "mean_p": 0.109405, "boundary_entropy": 0.214276, "bpb": 0.820034}, "th": {"chunks": 2991.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.18595, "cps": 23.367188, "mean_p": 0.109788, "boundary_entropy": 0.20956, "bpb": 0.920726}, "zh": {"chunks": 3046.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.515747, "cps": 23.796875, "mean_p": 0.20447, "boundary_entropy": 0.089399, "bpb": 2.588224}, "my": {"chunks": 2086.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.110091, "cps": 16.296875, "mean_p": 0.156224, "boundary_entropy": 0.346259, "bpb": 0.932807}}, "bpb": {"en": 1.67068, "de": 1.81472, "vi": 1.45541, "tr": 1.74555, "fi": 1.91541, "ru": 1.08872, "el": 1.09747, "hi": 0.96114, "ta": 0.82003, "th": 0.92073, "zh": 2.58822, "my": 0.93281}, "gini_cpc": 0.234979, "gini_cps": 0.176585, "bpic": 8.993111, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.925540245423973, "vi": 1.1827207977974652, "tr": 0.8113229040917028, "fi": 0.8458754334225635, "ru": 1.6497972320106873, "el": 1.5842812380483666, "hi": 0.8284780811991432, "ta": 0.7629608567572917, "th": 0.9729650116797208, "zh": 2.6985988544930914, "my": 0.5760402143740929}, "t": 1488.7}
{"event": "eval", "step": 8000, "phase": 3, "bpb": {"en": 1.66999, "de": 1.81864, "vi": 1.47122, "tr": 1.73635, "fi": 1.92982, "ru": 1.08715, "el": 1.09579, "hi": 0.96737, "ta": 0.82408, "th": 0.92005, "zh": 2.59584, "my": 0.94089}, "bytes_per_chunk": {"en": 5.019, "de": 5.731, "vi": 5.754, "tr": 6.741, "fi": 6.203, "ru": 5.986, "el": 6.043, "hi": 14.908, "ta": 17.841, "th": 14.781, "zh": 5.217, "my": 23.94}, "bpb_mean": 1.42143, "t": 1493.7}
{"event": "train", "step": 8000, "phase": 3, "objective": "none", "lm_loss": 1.09597, "bpb": 1.58116, "chunk_loss": 0.0, "F": 0.16611, "G": 0.17109, "chunks": 170.09, "bytes_per_chunk": 6.02, "boundary_entropy": 0.09866, "router_noise": 0.0, "grad_norm": 0.3757, "lr": 8.08e-05, "router_trainable": false, "sec_per_step": 0.245, "alpha": 0.0, "t": 1493.9}
{"event": "train", "step": 8100, "phase": 3, "objective": "none", "lm_loss": 1.11909, "bpb": 1.61451, "chunk_loss": 0.0, "F": 0.17012, "G": 0.17319, "chunks": 174.2, "bytes_per_chunk": 5.878, "boundary_entropy": 0.09129, "router_noise": 0.0, "grad_norm": 0.3854, "lr": 7.77e-05, "router_trainable": false, "sec_per_step": 0.166, "alpha": 0.0, "t": 1510.4}
{"event": "train", "step": 8200, "phase": 3, "objective": "none", "lm_loss": 1.07639, "bpb": 1.5529, "chunk_loss": 0.0, "F": 0.16649, "G": 0.16849, "chunks": 170.48, "bytes_per_chunk": 6.006, "boundary_entropy": 0.09281, "router_noise": 0.0, "grad_norm": 0.3828, "lr": 7.47e-05, "router_trainable": false, "sec_per_step": 0.167, "alpha": 0.0, "t": 1527.2}
{"event": "train", "step": 8300, "phase": 3, "objective": "none", "lm_loss": 1.01232, "bpb": 1.46047, "chunk_loss": 0.0, "F": 0.16214, "G": 0.1675, "chunks": 166.03, "bytes_per_chunk": 6.168, "boundary_entropy": 0.10373, "router_noise": 0.0, "grad_norm": 0.3615, "lr": 7.17e-05, "router_trainable": false, "sec_per_step": 0.164, "alpha": 0.0, "t": 1543.6}
{"event": "train", "step": 8400, "phase": 3, "objective": "none", "lm_loss": 1.05964, "bpb": 1.52874, "chunk_loss": 0.0, "F": 0.16823, "G": 0.17124, "chunks": 172.27, "bytes_per_chunk": 5.944, "boundary_entropy": 0.09559, "router_noise": 0.0, "grad_norm": 0.3823, "lr": 6.88e-05, "router_trainable": false, "sec_per_step": 0.165, "alpha": 0.0, "t": 1560.1}
{"event": "train", "step": 8500, "phase": 3, "objective": "none", "lm_loss": 1.07197, "bpb": 1.54653, "chunk_loss": 0.0, "F": 0.16966, "G": 0.17227, "chunks": 173.73, "bytes_per_chunk": 5.894, "boundary_entropy": 0.09146, "router_noise": 0.0, "grad_norm": 0.361, "lr": 6.6e-05, "router_trainable": false, "sec_per_step": 0.169, "alpha": 0.0, "t": 1577.0}
{"event": "train", "step": 8600, "phase": 3, "objective": "none", "lm_loss": 1.1256, "bpb": 1.6239, "chunk_loss": 0.0, "F": 0.16353, "G": 0.16939, "chunks": 167.45, "bytes_per_chunk": 6.115, "boundary_entropy": 0.09449, "router_noise": 0.0, "grad_norm": 0.3898, "lr": 6.33e-05, "router_trainable": false, "sec_per_step": 0.167, "alpha": 0.0, "t": 1593.7}
{"event": "train", "step": 8700, "phase": 3, "objective": "none", "lm_loss": 0.9967, "bpb": 1.43793, "chunk_loss": 0.0, "F": 0.1608, "G": 0.16935, "chunks": 164.66, "bytes_per_chunk": 6.219, "boundary_entropy": 0.10385, "router_noise": 0.0, "grad_norm": 0.3822, "lr": 6.07e-05, "router_trainable": false, "sec_per_step": 0.165, "alpha": 0.0, "t": 1610.2}
{"event": "train", "step": 8800, "phase": 3, "objective": "none", "lm_loss": 1.01483, "bpb": 1.4641, "chunk_loss": 0.0, "F": 0.16437, "G": 0.171, "chunks": 168.31, "bytes_per_chunk": 6.084, "boundary_entropy": 0.10161, "router_noise": 0.0, "grad_norm": 0.3737, "lr": 5.82e-05, "router_trainable": false, "sec_per_step": 0.167, "alpha": 0.0, "t": 1627.0}
{"event": "train", "step": 8900, "phase": 3, "objective": "none", "lm_loss": 1.07168, "bpb": 1.54611, "chunk_loss": 0.0, "F": 0.17647, "G": 0.17613, "chunks": 180.7, "bytes_per_chunk": 5.667, "boundary_entropy": 0.08167, "router_noise": 0.0, "grad_norm": 0.3612, "lr": 5.58e-05, "router_trainable": false, "sec_per_step": 0.166, "alpha": 0.0, "t": 1643.5}
{"event": "probe", "step": 9000, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3324.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.191002, "cps": 25.96875, "mean_p": 0.185442, "boundary_entropy": 0.062116, "bpb": 1.647991}, "de": {"chunks": 3633.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.176935, "cps": 28.382812, "mean_p": 0.173923, "boundary_entropy": 0.055367, "bpb": 1.788269}, "vi": {"chunks": 4183.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.226905, "cps": 32.679688, "mean_p": 0.167957, "boundary_entropy": 0.073794, "bpb": 1.423316}, "tr": {"chunks": 2858.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.157874, "cps": 22.328125, "mean_p": 0.155655, "boundary_entropy": 0.087873, "bpb": 1.717824}, "fi": {"chunks": 3039.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.161554, "cps": 23.742188, "mean_p": 0.157959, "boundary_entropy": 0.073215, "bpb": 1.885471}, "ru": {"chunks": 6033.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.318364, "cps": 47.132812, "mean_p": 0.159031, "boundary_entropy": 0.096646, "bpb": 1.067539}, "el": {"chunks": 6324.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.308142, "cps": 49.40625, "mean_p": 0.161371, "boundary_entropy": 0.095947, "bpb": 1.071698}, "hi": {"chunks": 2898.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.172244, "cps": 22.640625, "mean_p": 0.104918, "boundary_entropy": 0.195327, "bpb": 0.942192}, "ta": {"chunks": 2993.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.160096, "cps": 23.382812, "mean_p": 0.123298, "boundary_entropy": 0.229835, "bpb": 0.805152}, "th": {"chunks": 3085.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.191794, "cps": 24.101562, "mean_p": 0.109913, "boundary_entropy": 0.204781, "bpb": 0.899985}, "zh": {"chunks": 3039.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.514561, "cps": 23.742188, "mean_p": 0.204049, "boundary_entropy": 0.09002, "bpb": 2.545055}, "my": {"chunks": 3187.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.168197, "cps": 24.898438, "mean_p": 0.172092, "boundary_entropy": 0.356174, "bpb": 0.916297}}, "bpb": {"en": 1.64799, "de": 1.78827, "vi": 1.42332, "tr": 1.71782, "fi": 1.88547, "ru": 1.06754, "el": 1.0717, "hi": 0.94219, "ta": 0.80515, "th": 0.89998, "zh": 2.54506, "my": 0.9163}, "gini_cpc": 0.204426, "gini_cps": 0.14687, "bpic": 8.605727, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9263521115344657, "vi": 1.1879764398085182, "tr": 0.8265607491508179, "fi": 0.8458277350202383, "ru": 1.6668142937427965, "el": 1.613296238384301, "hi": 0.9017920727815, "ta": 0.8381936230753411, "th": 1.0041467991008954, "zh": 2.6940171898858285, "my": 0.8806063088109658}, "t": 1662.9}
{"event": "eval", "step": 9000, "phase": 3, "bpb": {"en": 1.64683, "de": 1.79265, "vi": 1.43987, "tr": 1.71018, "fi": 1.90015, "ru": 1.06555, "el": 1.06934, "hi": 0.94782, "ta": 0.80993, "th": 0.90045, "zh": 2.55257, "my": 0.92456}, "bytes_per_chunk": {"en": 5.009, "de": 5.725, "vi": 5.744, "tr": 6.681, "fi": 6.212, "ru": 5.933, "el": 5.961, "hi": 13.602, "ta": 15.718, "th": 14.144, "zh": 5.24, "my": 16.737}, "bpb_mean": 1.39666, "t": 1668.0}
{"event": "train", "step": 9000, "phase": 3, "objective": "none", "lm_loss": 1.06318, "bpb": 1.53385, "chunk_loss": 0.0, "F": 0.1591, "G": 0.16584, "chunks": 162.92, "bytes_per_chunk": 6.285, "boundary_entropy": 0.09919, "router_noise": 0.0, "grad_norm": 0.394, "lr": 5.35e-05, "router_trainable": false, "sec_per_step": 0.247, "alpha": 0.0, "t": 1668.2}
{"event": "train", "step": 9100, "phase": 3, "objective": "none", "lm_loss": 1.05134, "bpb": 1.51676, "chunk_loss": 0.0, "F": 0.16676, "G": 0.17258, "chunks": 170.77, "bytes_per_chunk": 5.997, "boundary_entropy": 0.09976, "router_noise": 0.0, "grad_norm": 0.3665, "lr": 5.12e-05, "router_trainable": false, "sec_per_step": 0.166, "alpha": 0.0, "t": 1684.8}
{"event": "train", "step": 9200, "phase": 3, "objective": "none", "lm_loss": 0.97345, "bpb": 1.40439, "chunk_loss": 0.0, "F": 0.15833, "G": 0.16378, "chunks": 162.12, "bytes_per_chunk": 6.316, "boundary_entropy": 0.10204, "router_noise": 0.0, "grad_norm": 0.3592, "lr": 4.91e-05, "router_trainable": false, "sec_per_step": 0.167, "alpha": 0.0, "t": 1701.5}
{"event": "train", "step": 9300, "phase": 3, "objective": "none", "lm_loss": 1.04633, "bpb": 1.50954, "chunk_loss": 0.0, "F": 0.17044, "G": 0.17445, "chunks": 174.53, "bytes_per_chunk": 5.867, "boundary_entropy": 0.09248, "router_noise": 0.0, "grad_norm": 0.3769, "lr": 4.71e-05, "router_trainable": false, "sec_per_step": 0.165, "alpha": 0.0, "t": 1718.0}
{"event": "train", "step": 9400, "phase": 3, "objective": "none", "lm_loss": 1.09098, "bpb": 1.57395, "chunk_loss": 0.0, "F": 0.16742, "G": 0.17342, "chunks": 171.44, "bytes_per_chunk": 5.973, "boundary_entropy": 0.08749, "router_noise": 0.0, "grad_norm": 0.3796, "lr": 4.52e-05, "router_trainable": false, "sec_per_step": 0.167, "alpha": 0.0, "t": 1734.6}
{"event": "train", "step": 9500, "phase": 3, "objective": "none", "lm_loss": 1.10185, "bpb": 1.58964, "chunk_loss": 0.0, "F": 0.17146, "G": 0.17432, "chunks": 175.58, "bytes_per_chunk": 5.832, "boundary_entropy": 0.08737, "router_noise": 0.0, "grad_norm": 0.3869, "lr": 4.34e-05, "router_trainable": false, "sec_per_step": 0.167, "alpha": 0.0, "t": 1751.3}
{"event": "train", "step": 9600, "phase": 3, "objective": "none", "lm_loss": 0.96163, "bpb": 1.38734, "chunk_loss": 0.0, "F": 0.15778, "G": 0.16703, "chunks": 161.56, "bytes_per_chunk": 6.338, "boundary_entropy": 0.10667, "router_noise": 0.0, "grad_norm": 0.3835, "lr": 4.17e-05, "router_trainable": false, "sec_per_step": 0.167, "alpha": 0.0, "t": 1768.0}
{"event": "train", "step": 9700, "phase": 3, "objective": "none", "lm_loss": 0.97403, "bpb": 1.40523, "chunk_loss": 0.0, "F": 0.1655, "G": 0.17126, "chunks": 169.47, "bytes_per_chunk": 6.042, "boundary_entropy": 0.0975, "router_noise": 0.0, "grad_norm": 0.3606, "lr": 4.01e-05, "router_trainable": false, "sec_per_step": 0.166, "alpha": 0.0, "t": 1784.6}
{"event": "train", "step": 9800, "phase": 3, "objective": "none", "lm_loss": 1.00066, "bpb": 1.44365, "chunk_loss": 0.0, "F": 0.15785, "G": 0.16749, "chunks": 161.64, "bytes_per_chunk": 6.335, "boundary_entropy": 0.10581, "router_noise": 0.0, "grad_norm": 0.3859, "lr": 3.86e-05, "router_trainable": false, "sec_per_step": 0.166, "alpha": 0.0, "t": 1801.3}
{"event": "train", "step": 9900, "phase": 3, "objective": "none", "lm_loss": 1.06881, "bpb": 1.54197, "chunk_loss": 0.0, "F": 0.16666, "G": 0.17618, "chunks": 170.66, "bytes_per_chunk": 6.0, "boundary_entropy": 0.0963, "router_noise": 0.0, "grad_norm": 0.375, "lr": 3.72e-05, "router_trainable": false, "sec_per_step": 0.166, "alpha": 0.0, "t": 1817.9}
{"event": "probe", "step": 10000, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3338.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.191806, "cps": 26.078125, "mean_p": 0.187934, "boundary_entropy": 0.060086, "bpb": 1.634477}, "de": {"chunks": 3634.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.176983, "cps": 28.390625, "mean_p": 0.174941, "boundary_entropy": 0.054905, "bpb": 1.770259}, "vi": {"chunks": 4197.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.227665, "cps": 32.789062, "mean_p": 0.168951, "boundary_entropy": 0.069304, "bpb": 1.407913}, "tr": {"chunks": 2868.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.158427, "cps": 22.40625, "mean_p": 0.156412, "boundary_entropy": 0.088266, "bpb": 1.70063}, "fi": {"chunks": 3097.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.164638, "cps": 24.195312, "mean_p": 0.160536, "boundary_entropy": 0.073754, "bpb": 1.862826}, "ru": {"chunks": 6036.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.318522, "cps": 47.15625, "mean_p": 0.160398, "boundary_entropy": 0.093756, "bpb": 1.053137}, "el": {"chunks": 6317.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.307801, "cps": 49.351562, "mean_p": 0.160387, "boundary_entropy": 0.092388, "bpb": 1.056625}, "hi": {"chunks": 3413.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.202853, "cps": 26.664062, "mean_p": 0.116336, "boundary_entropy": 0.201407, "bpb": 0.923819}, "ta": {"chunks": 3397.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.181706, "cps": 26.539062, "mean_p": 0.13276, "boundary_entropy": 0.24054, "bpb": 0.792893}, "th": {"chunks": 3196.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.198694, "cps": 24.96875, "mean_p": 0.11433, "boundary_entropy": 0.20966, "bpb": 0.888834}, "zh": {"chunks": 3039.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.514561, "cps": 23.742188, "mean_p": 0.202533, "boundary_entropy": 0.085454, "bpb": 2.51599}, "my": {"chunks": 2767.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.146031, "cps": 21.617188, "mean_p": 0.170941, "boundary_entropy": 0.361418, "bpb": 0.904016}}, "bpb": {"en": 1.63448, "de": 1.77026, "vi": 1.40791, "tr": 1.70063, "fi": 1.86283, "ru": 1.05314, "el": 1.05662, "hi": 0.92382, "ta": 0.79289, "th": 0.88883, "zh": 2.51599, "my": 0.90402}, "gini_cpc": 0.203744, "gini_cps": 0.145447, "bpic": 8.472174, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9227207852609327, "vi": 1.1869532532404186, "tr": 0.8259740125631551, "fi": 0.858355332994708, "ru": 1.66064883305852, "el": 1.6047516082367013, "hi": 1.0575940607369594, "ta": 0.9473443410837558, "th": 1.0359135010177545, "zh": 2.6827181363632397, "my": 0.7613486305998707}, "t": 1837.2}
{"event": "eval", "step": 10000, "phase": 3, "bpb": {"en": 1.63265, "de": 1.77472, "vi": 1.42444, "tr": 1.69241, "fi": 1.87761, "ru": 1.05105, "el": 1.05442, "hi": 0.92917, "ta": 0.79789, "th": 0.88881, "zh": 2.52443, "my": 0.91191}, "bytes_per_chunk": {"en": 4.993, "de": 5.718, "vi": 5.731, "tr": 6.67, "fi": 6.124, "ru": 5.931, "el": 5.966, "hi": 11.77, "ta": 14.332, "th": 13.618, "zh": 5.25, "my": 18.055}, "bpb_mean": 1.37996, "t": 1842.3}
{"event": "train", "step": 10000, "phase": 3, "objective": "none", "lm_loss": 1.04245, "bpb": 1.50394, "chunk_loss": 0.0, "F": 0.16576, "G": 0.17625, "chunks": 169.73, "bytes_per_chunk": 6.033, "boundary_entropy": 0.10564, "router_noise": 0.0, "grad_norm": 0.4014, "lr": 3.6e-05, "router_trainable": false, "sec_per_step": 0.246, "alpha": 0.0, "t": 1842.5}
{"event": "train", "step": 10100, "phase": 3, "objective": "none", "lm_loss": 0.99371, "bpb": 1.43362, "chunk_loss": 0.0, "F": 0.15062, "G": 0.16422, "chunks": 154.23, "bytes_per_chunk": 6.639, "boundary_entropy": 0.1165, "router_noise": 0.0, "grad_norm": 0.3794, "lr": 3.49e-05, "router_trainable": false, "sec_per_step": 0.169, "alpha": 0.0, "t": 1859.4}
{"event": "train", "step": 10200, "phase": 3, "objective": "none", "lm_loss": 1.08007, "bpb": 1.55821, "chunk_loss": 0.0, "F": 0.17761, "G": 0.17936, "chunks": 181.88, "bytes_per_chunk": 5.63, "boundary_entropy": 0.0846, "router_noise": 0.0, "grad_norm": 0.4055, "lr": 3.38e-05, "router_trainable": false, "sec_per_step": 0.166, "alpha": 0.0, "t": 1876.0}
{"event": "train", "step": 10300, "phase": 3, "objective": "none", "lm_loss": 1.03357, "bpb": 1.49112, "chunk_loss": 0.0, "F": 0.15941, "G": 0.17095, "chunks": 163.23, "bytes_per_chunk": 6.273, "boundary_entropy": 0.10673, "router_noise": 0.0, "grad_norm": 0.4252, "lr": 3.3e-05, "router_trainable": false, "sec_per_step": 0.166, "alpha": 0.0, "t": 1892.6}
{"event": "train", "step": 10400, "phase": 3, "objective": "none", "lm_loss": 1.00896, "bpb": 1.45563, "chunk_loss": 0.0, "F": 0.16478, "G": 0.16966, "chunks": 168.73, "bytes_per_chunk": 6.069, "boundary_entropy": 0.09395, "router_noise": 0.0, "grad_norm": 0.3822, "lr": 3.22e-05, "router_trainable": false, "sec_per_step": 0.167, "alpha": 0.0, "t": 1909.3}
{"event": "train", "step": 10500, "phase": 3, "objective": "none", "lm_loss": 1.03552, "bpb": 1.49394, "chunk_loss": 0.0, "F": 0.16609, "G": 0.17575, "chunks": 170.08, "bytes_per_chunk": 6.021, "boundary_entropy": 0.10038, "router_noise": 0.0, "grad_norm": 0.3817, "lr": 3.15e-05, "router_trainable": false, "sec_per_step": 0.167, "alpha": 0.0, "t": 1926.0}
{"event": "train", "step": 10600, "phase": 3, "objective": "none", "lm_loss": 1.0562, "bpb": 1.52377, "chunk_loss": 0.0, "F": 0.16632, "G": 0.17153, "chunks": 170.31, "bytes_per_chunk": 6.012, "boundary_entropy": 0.09145, "router_noise": 0.0, "grad_norm": 0.3863, "lr": 3.1e-05, "router_trainable": false, "sec_per_step": 0.165, "alpha": 0.0, "t": 1942.5}
{"event": "train", "step": 10700, "phase": 3, "objective": "none", "lm_loss": 1.02499, "bpb": 1.47874, "chunk_loss": 0.0, "F": 0.16222, "G": 0.17126, "chunks": 166.11, "bytes_per_chunk": 6.165, "boundary_entropy": 0.10334, "router_noise": 0.0, "grad_norm": 0.3859, "lr": 3.05e-05, "router_trainable": false, "sec_per_step": 0.166, "alpha": 0.0, "t": 1959.1}
{"event": "train", "step": 10800, "phase": 3, "objective": "none", "lm_loss": 1.04481, "bpb": 1.50734, "chunk_loss": 0.0, "F": 0.16515, "G": 0.17389, "chunks": 169.11, "bytes_per_chunk": 6.055, "boundary_entropy": 0.09981, "router_noise": 0.0, "grad_norm": 0.3844, "lr": 3.02e-05, "router_trainable": false, "sec_per_step": 0.167, "alpha": 0.0, "t": 1975.8}
{"event": "train", "step": 10900, "phase": 3, "objective": "none", "lm_loss": 1.02959, "bpb": 1.48539, "chunk_loss": 0.0, "F": 0.1626, "G": 0.16852, "chunks": 166.5, "bytes_per_chunk": 6.15, "boundary_entropy": 0.09905, "router_noise": 0.0, "grad_norm": 0.3907, "lr": 3.01e-05, "router_trainable": false, "sec_per_step": 0.166, "alpha": 0.0, "t": 1992.4}
{"event": "probe", "step": 11000, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3349.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.192438, "cps": 26.164062, "mean_p": 0.188635, "boundary_entropy": 0.059373, "bpb": 1.626196}, "de": {"chunks": 3657.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.178104, "cps": 28.570312, "mean_p": 0.175721, "boundary_entropy": 0.054114, "bpb": 1.759922}, "vi": {"chunks": 4198.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.227719, "cps": 32.796875, "mean_p": 0.168707, "boundary_entropy": 0.069131, "bpb": 1.395664}, "tr": {"chunks": 2880.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.15909, "cps": 22.5, "mean_p": 0.157518, "boundary_entropy": 0.089819, "bpb": 1.690516}, "fi": {"chunks": 3128.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.166286, "cps": 24.4375, "mean_p": 0.162646, "boundary_entropy": 0.075142, "bpb": 1.850871}, "ru": {"chunks": 6053.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.31942, "cps": 47.289062, "mean_p": 0.161428, "boundary_entropy": 0.092542, "bpb": 1.043798}, "el": {"chunks": 6355.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.309653, "cps": 49.648438, "mean_p": 0.163461, "boundary_entropy": 0.09559, "bpb": 1.047138}, "hi": {"chunks": 3404.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.202318, "cps": 26.59375, "mean_p": 0.118395, "boundary_entropy": 0.2058, "bpb": 0.915943}, "ta": {"chunks": 3415.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.182669, "cps": 26.679688, "mean_p": 0.134786, "boundary_entropy": 0.238787, "bpb": 0.788742}, "th": {"chunks": 3414.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.212247, "cps": 26.671875, "mean_p": 0.120264, "boundary_entropy": 0.216422, "bpb": 0.880999}, "zh": {"chunks": 3027.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.51253, "cps": 23.648438, "mean_p": 0.201536, "boundary_entropy": 0.083484, "bpb": 2.504361}, "my": {"chunks": 2468.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.130251, "cps": 19.28125, "mean_p": 0.162451, "boundary_entropy": 0.35015, "bpb": 0.894645}}, "bpb": {"en": 1.6262, "de": 1.75992, "vi": 1.39566, "tr": 1.69052, "fi": 1.85087, "ru": 1.0438, "el": 1.04714, "hi": 0.91594, "ta": 0.78874, "th": 0.881, "zh": 2.50436, "my": 0.89465}, "gini_cpc": 0.208825, "gini_cps": 0.151355, "bpic": 8.463019, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9255108742283885, "vi": 1.183336512046757, "tr": 0.8267056561498964, "fi": 0.8640996587211229, "ru": 1.6598560748650208, "el": 1.6091023992526747, "hi": 1.0513406366008438, "ta": 0.9492360231597237, "th": 1.102938823525699, "zh": 2.6633482132790647, "my": 0.6768473796966795}, "t": 2011.8}
{"event": "eval", "step": 11000, "phase": 3, "bpb": {"en": 1.62456, "de": 1.76518, "vi": 1.41204, "tr": 1.6828, "fi": 1.86527, "ru": 1.04237, "el": 1.04505, "hi": 0.92207, "ta": 0.79374, "th": 0.88194, "zh": 2.51185, "my": 0.9026}, "bytes_per_chunk": {"en": 4.985, "de": 5.688, "vi": 5.73, "tr": 6.635, "fi": 6.075, "ru": 5.914, "el": 5.938, "hi": 11.857, "ta": 14.41, "th": 12.95, "zh": 5.255, "my": 20.708}, "bpb_mean": 1.37079, "t": 2016.8}
{"event": "probe", "step": 11000, "phase": 3, "objective": "none", "router_trainable": false, "per_lang": {"en": {"chunks": 3349.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.192438, "cps": 26.164062, "mean_p": 0.188635, "boundary_entropy": 0.059373, "bpb": 1.626196}, "de": {"chunks": 3657.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.178104, "cps": 28.570312, "mean_p": 0.175721, "boundary_entropy": 0.054114, "bpb": 1.759922}, "vi": {"chunks": 4198.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.227719, "cps": 32.796875, "mean_p": 0.168707, "boundary_entropy": 0.069131, "bpb": 1.395664}, "tr": {"chunks": 2880.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.15909, "cps": 22.5, "mean_p": 0.157518, "boundary_entropy": 0.089819, "bpb": 1.690516}, "fi": {"chunks": 3128.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.166286, "cps": 24.4375, "mean_p": 0.162646, "boundary_entropy": 0.075142, "bpb": 1.850871}, "ru": {"chunks": 6053.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.31942, "cps": 47.289062, "mean_p": 0.161428, "boundary_entropy": 0.092542, "bpb": 1.043798}, "el": {"chunks": 6355.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.309653, "cps": 49.648438, "mean_p": 0.163461, "boundary_entropy": 0.09559, "bpb": 1.047138}, "hi": {"chunks": 3404.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.202318, "cps": 26.59375, "mean_p": 0.118395, "boundary_entropy": 0.2058, "bpb": 0.915943}, "ta": {"chunks": 3415.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.182669, "cps": 26.679688, "mean_p": 0.134786, "boundary_entropy": 0.238787, "bpb": 0.788742}, "th": {"chunks": 3414.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.212247, "cps": 26.671875, "mean_p": 0.120264, "boundary_entropy": 0.216422, "bpb": 0.880999}, "zh": {"chunks": 3027.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.51253, "cps": 23.648438, "mean_p": 0.201536, "boundary_entropy": 0.083484, "bpb": 2.504361}, "my": {"chunks": 2468.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.130251, "cps": 19.28125, "mean_p": 0.162451, "boundary_entropy": 0.35015, "bpb": 0.894645}}, "bpb": {"en": 1.6262, "de": 1.75992, "vi": 1.39566, "tr": 1.69052, "fi": 1.85087, "ru": 1.0438, "el": 1.04714, "hi": 0.91594, "ta": 0.78874, "th": 0.881, "zh": 2.50436, "my": 0.89465}, "gini_cpc": 0.208825, "gini_cps": 0.151355, "bpic": 8.463019, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9255108742283885, "vi": 1.183336512046757, "tr": 0.8267056561498964, "fi": 0.8640996587211229, "ru": 1.6598560748650208, "el": 1.6091023992526747, "hi": 1.0513406366008438, "ta": 0.9492360231597237, "th": 1.102938823525699, "zh": 2.6633482132790647, "my": 0.6768473796966795}, "t": 2019.8}
{"event": "reps_dumped", "dirs": ["/local/scratch/sas245/beetle-hnet/scratch/reps/spec_B_s0/th0.30", "/local/scratch/sas245/beetle-hnet/scratch/reps/spec_B_s0/th0.40", "/local/scratch/sas245/beetle-hnet/scratch/reps/spec_B_s0/th0.50", "/local/scratch/sas245/beetle-hnet/scratch/reps/spec_B_s0/th0.60", "/local/scratch/sas245/beetle-hnet/scratch/reps/spec_B_s0/th0.70"], "t": 2022.6}
{"event": "final", "run": "spec_B_s0", "condition": "B", "step": 11000, "params": 82470528, "per_lang": {"en": {"cpc": 0.19243808538757687, "cps": 26.1640625, "bpb": 1.619737152617264, "bytes_per_chunk": 4.99445577000019}, "de": {"cpc": 0.17810354064189354, "cps": 28.5703125, "bpb": 1.7626968712337192, "bytes_per_chunk": 5.662102034645125}, "vi": {"cpc": 0.22771901274749118, "cps": 32.796875, "bpb": 1.4109275665044787, "bytes_per_chunk": 5.73594153428734}, "tr": {"cpc": 0.15908965364856653, "cps": 22.5, "bpb": 1.6820103829149253, "bytes_per_chunk": 6.645515790274231}, "fi": {"cpc": 0.1662856839083515, "cps": 24.4375, "bpb": 1.872060509930391, "bytes_per_chunk": 6.096620211287669}, "ru": {"cpc": 0.31941952506596305, "cps": 47.2890625, "bpb": 1.0428376295307682, "bytes_per_chunk": 5.911766861550085}, "el": {"cpc": 0.309652584904741, "cps": 49.6484375, "bpb": 1.0428060008463829, "bytes_per_chunk": 5.959612837956885}, "hi": {"cpc": 0.2023179791976226, "cps": 26.59375, "bpb": 0.9232596348552572, "bytes_per_chunk": 11.87622887690844}, "ta": {"cpc": 0.1826691628777748, "cps": 26.6796875, "bpb": 0.7916364966357256, "bytes_per_chunk": 14.439217846323327}, "th": {"cpc": 0.21224743549891203, "cps": 26.671875, "bpb": 0.8850218453033855, "bytes_per_chunk": 12.95593940741839}, "zh": {"cpc": 0.512529630883847, "cps": 23.6484375, "bpb": 2.514999914470131, "bytes_per_chunk": 5.272034349955253}, "my": {"cpc": 0.13025121384842728, "cps": 19.28125, "bpb": 0.900903464959239, "bytes_per_chunk": 20.91296370163542}}, "gini_cpc": 0.208825178114471, "gini_cps": 0.15135544382699728, "bpic": 8.463019317279704, "seed": 0, "mean_boundary_entropy": 0.10192666947841644, "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": "none", "router_trainable": false, "encoder_trainable": true, "frozen_router_params": 295680}], "t": 2032.2}