| {"event": "start", "run": "exp32_local_onechunk_s0", "condition": "A", "params": 82470528, "langs": ["en", "de", "vi", "tr", "fi", "ru", "el", "hi", "ta", "th", "zh", "my"], "premium": {"en": 1.0, "de": 1.184, "vi": 1.39, "tr": 1.12, "fi": 1.104, "ru": 1.971, "el": 2.174, "hi": 2.552, "ta": 3.192, "th": 2.762, "zh": 0.915, "my": 3.524}, "schedule": {"tau1": 0.4, "tau2": 0.7, "p1_objective": "global", "p2_objective": "global", "p3_objective": "global", "freeze_in_p3": false, "freeze_scope": "router"}, "corpus": "/local/scratch/sas245/beetle-hnet/scratch/data/tier1_corpus", "premium_kind": "flores_per_sent", "doc_filtered": {"my": 37511}, "weights": {"en": 0.2946, "de": 0.0909, "vi": 0.0576, "tr": 0.0612, "fi": 0.0424, "ru": 0.1476, "el": 0.0522, "hi": 0.042, "ta": 0.0313, "th": 0.0542, "zh": 0.1038, "my": 0.0222}, "t": 2.0} |
| {"event": "probe_schedule", "n": 21, "steps": [0, 1, 2, 4, 8, 16, 32, 64, 128, 256, 1000, 2000, 3000, 4000, 5000, 6000, 7000, 8000, 9000, 10000, 11000], "t": 2.0} |
| {"event": "probe", "step": 0, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 8633.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.496064, "cps": 67.445312, "mean_p": 0.503405, "boundary_entropy": 0.687423, "bpb": 8.149224}, "de": {"chunks": 12185.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.593435, "cps": 95.195312, "mean_p": 0.507242, "boundary_entropy": 0.68817, "bpb": 8.118233}, "vi": {"chunks": 6146.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.333388, "cps": 48.015625, "mean_p": 0.490614, "boundary_entropy": 0.688472, "bpb": 8.208651}, "tr": {"chunks": 11394.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.629398, "cps": 89.015625, "mean_p": 0.506619, "boundary_entropy": 0.687865, "bpb": 8.140327}, "fi": {"chunks": 7257.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.385785, "cps": 56.695312, "mean_p": 0.49745, "boundary_entropy": 0.687937, "bpb": 8.133666}, "ru": {"chunks": 10390.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.548285, "cps": 81.171875, "mean_p": 0.49637, "boundary_entropy": 0.69018, "bpb": 8.036225}, "el": {"chunks": 36558.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.781319, "cps": 285.609375, "mean_p": 0.529623, "boundary_entropy": 0.688846, "bpb": 8.175448}, "hi": {"chunks": 42514.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.526835, "cps": 332.140625, "mean_p": 0.520835, "boundary_entropy": 0.690193, "bpb": 7.980936}, "ta": {"chunks": 49746.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.660925, "cps": 388.640625, "mean_p": 0.521454, "boundary_entropy": 0.690378, "bpb": 7.917233}, "th": {"chunks": 45763.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.845073, "cps": 357.523438, "mean_p": 0.525569, "boundary_entropy": 0.689911, "bpb": 8.034201}, "zh": {"chunks": 6153.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.041822, "cps": 48.070312, "mean_p": 0.496515, "boundary_entropy": 0.686535, "bpb": 8.095017}, "my": {"chunks": 53337.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.814915, "cps": 416.695312, "mean_p": 0.521486, "boundary_entropy": 0.69057, "bpb": 8.110061}}, "bpb": {"en": 8.14922, "de": 8.11823, "vi": 8.20865, "tr": 8.14033, "fi": 8.13367, "ru": 8.03622, "el": 8.17545, "hi": 7.98094, "ta": 7.91723, "th": 8.0342, "zh": 8.09502, "my": 8.11006}, "gini_cpc": 0.39279, "gini_cps": 0.41072, "bpic": 1.323036, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1962873370015168, "vi": 0.6720657963081282, "tr": 1.2687850214788603, "fi": 0.7776919775720443, "ru": 1.1052708405220564, "el": 3.5909053880998, "hi": 5.093769334952837, "ta": 5.364077888056945, "th": 5.735295526368824, "zh": 2.1001767762124004, "my": 5.674499837031958}, "t": 5.2} |
| {"event": "train", "step": 0, "phase": 1, "objective": "global", "lm_loss": 5.62332, "bpb": 8.11273, "chunk_loss": 0.018003, "F": 0.00098, "G": 0.50189, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.6917, "router_noise": 0.0, "grad_norm": 4.9212, "lr": 6e-07, "router_trainable": true, "sec_per_step": 0.005, "alpha": 0.03, "t": 5.9} |
| {"event": "probe", "step": 1, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 8666.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.49796, "cps": 67.703125, "mean_p": 0.503511, "boundary_entropy": 0.687422, "bpb": 8.128993}, "de": {"chunks": 12208.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.594555, "cps": 95.375, "mean_p": 0.507302, "boundary_entropy": 0.688169, "bpb": 8.098998}, "vi": {"chunks": 6177.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.335069, "cps": 48.257812, "mean_p": 0.490697, "boundary_entropy": 0.688476, "bpb": 8.194521}, "tr": {"chunks": 11426.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.631166, "cps": 89.265625, "mean_p": 0.506684, "boundary_entropy": 0.687864, "bpb": 8.126327}, "fi": {"chunks": 7304.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.388283, "cps": 57.0625, "mean_p": 0.49754, "boundary_entropy": 0.687939, "bpb": 8.117765}, "ru": {"chunks": 10487.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.553404, "cps": 81.929688, "mean_p": 0.49647, "boundary_entropy": 0.690182, "bpb": 8.014434}, "el": {"chunks": 36555.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.781172, "cps": 285.585938, "mean_p": 0.529585, "boundary_entropy": 0.68885, "bpb": 8.167263}, "hi": {"chunks": 42521.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.527251, "cps": 332.195312, "mean_p": 0.520913, "boundary_entropy": 0.690187, "bpb": 7.971425}, "ta": {"chunks": 49767.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.662049, "cps": 388.804688, "mean_p": 0.521514, "boundary_entropy": 0.690373, "bpb": 7.908567}, "th": {"chunks": 45765.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.845197, "cps": 357.539062, "mean_p": 0.52568, "boundary_entropy": 0.6899, "bpb": 8.027295}, "zh": {"chunks": 6161.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.043176, "cps": 48.132812, "mean_p": 0.496552, "boundary_entropy": 0.686536, "bpb": 8.093398}, "my": {"chunks": 53337.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.814915, "cps": 416.695312, "mean_p": 0.521479, "boundary_entropy": 0.69057, "bpb": 8.092444}}, "bpb": {"en": 8.12899, "de": 8.099, "vi": 8.19452, "tr": 8.12633, "fi": 8.11776, "ru": 8.01443, "el": 8.16726, "hi": 7.97143, "ta": 7.90857, "th": 8.02729, "zh": 8.0934, "my": 8.09244}, "gini_cpc": 0.392027, "gini_cps": 0.410027, "bpic": 1.321678, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1939813659039897, "vi": 0.6728835246824707, "tr": 1.267503314845855, "fi": 0.7797480845545305, "ru": 1.1113413899709355, "el": 3.5769377208791844, "hi": 5.075207840310496, "ta": 5.345907351772137, "th": 5.713705303315577, "zh": 2.0948995412317397, "my": 5.6528914254669855}, "t": 8.6} |
| {"event": "probe", "step": 2, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 8735.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.501925, "cps": 68.242188, "mean_p": 0.50366, "boundary_entropy": 0.687422, "bpb": 8.088893}, "de": {"chunks": 12228.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.595529, "cps": 95.53125, "mean_p": 0.507351, "boundary_entropy": 0.688168, "bpb": 8.061634}, "vi": {"chunks": 6244.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.338704, "cps": 48.78125, "mean_p": 0.490857, "boundary_entropy": 0.688483, "bpb": 8.167455}, "tr": {"chunks": 11458.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.632934, "cps": 89.515625, "mean_p": 0.506785, "boundary_entropy": 0.687862, "bpb": 8.099687}, "fi": {"chunks": 7360.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.39126, "cps": 57.5, "mean_p": 0.497671, "boundary_entropy": 0.687941, "bpb": 8.087088}, "ru": {"chunks": 10661.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.562586, "cps": 83.289062, "mean_p": 0.496635, "boundary_entropy": 0.690186, "bpb": 7.971583}, "el": {"chunks": 36550.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.780929, "cps": 285.546875, "mean_p": 0.529523, "boundary_entropy": 0.688857, "bpb": 8.152596}, "hi": {"chunks": 42536.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.528143, "cps": 332.3125, "mean_p": 0.521052, "boundary_entropy": 0.690176, "bpb": 7.948306}, "ta": {"chunks": 49799.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.66376, "cps": 389.054688, "mean_p": 0.521636, "boundary_entropy": 0.690364, "bpb": 7.881677}, "th": {"chunks": 45768.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.845384, "cps": 357.5625, "mean_p": 0.525908, "boundary_entropy": 0.689878, "bpb": 7.994684}, "zh": {"chunks": 6180.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.046393, "cps": 48.28125, "mean_p": 0.496611, "boundary_entropy": 0.686539, "bpb": 8.089443}, "my": {"chunks": 53336.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.814862, "cps": 416.6875, "mean_p": 0.521428, "boundary_entropy": 0.690575, "bpb": 8.072161}}, "bpb": {"en": 8.08889, "de": 8.06163, "vi": 8.16746, "tr": 8.09969, "fi": 8.08709, "ru": 7.97158, "el": 8.1526, "hi": 7.94831, "ta": 7.88168, "th": 7.99468, "zh": 8.08944, "my": 8.07216}, "gini_cpc": 0.390743, "gini_cps": 0.408855, "bpic": 1.319493, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1864904139819767, "vi": 0.6748091509191859, "tr": 1.2610127489961487, "fi": 0.779519783711786, "ru": 1.1208563070049977, "el": 3.5481971851091108, "hi": 5.036893697623011, "ta": 5.307088907848446, "th": 5.668942870935193, "zh": 2.084760852671631, "my": 5.608132642478031}, "t": 11.9} |
| {"event": "probe", "step": 4, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 8959.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.514796, "cps": 69.992188, "mean_p": 0.504287, "boundary_entropy": 0.68742, "bpb": 7.950784}, "de": {"chunks": 12347.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.601325, "cps": 96.460938, "mean_p": 0.507657, "boundary_entropy": 0.688163, "bpb": 7.932191}, "vi": {"chunks": 6414.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.347925, "cps": 50.109375, "mean_p": 0.491376, "boundary_entropy": 0.688508, "bpb": 8.074547}, "tr": {"chunks": 11627.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.642269, "cps": 90.835938, "mean_p": 0.507218, "boundary_entropy": 0.687853, "bpb": 8.005493}, "fi": {"chunks": 7629.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.405561, "cps": 59.601562, "mean_p": 0.49821, "boundary_entropy": 0.687949, "bpb": 7.976418}, "ru": {"chunks": 11116.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.586596, "cps": 86.84375, "mean_p": 0.497075, "boundary_entropy": 0.690195, "bpb": 7.8426}, "el": {"chunks": 36493.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.778151, "cps": 285.101562, "mean_p": 0.528977, "boundary_entropy": 0.688918, "bpb": 8.077032}, "hi": {"chunks": 42584.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.530996, "cps": 332.6875, "mean_p": 0.521593, "boundary_entropy": 0.690131, "bpb": 7.867616}, "ta": {"chunks": 49876.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.667879, "cps": 389.65625, "mean_p": 0.521952, "boundary_entropy": 0.690339, "bpb": 7.779392}, "th": {"chunks": 45780.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.84613, "cps": 357.65625, "mean_p": 0.526672, "boundary_entropy": 0.689801, "bpb": 7.880721}, "zh": {"chunks": 6260.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.059939, "cps": 48.90625, "mean_p": 0.496855, "boundary_entropy": 0.686548, "bpb": 8.077226}, "my": {"chunks": 53327.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.814387, "cps": 416.617188, "mean_p": 0.521226, "boundary_entropy": 0.690591, "bpb": 8.005448}}, "bpb": {"en": 7.95078, "de": 7.93219, "vi": 8.07455, "tr": 8.00549, "fi": 7.97642, "ru": 7.8426, "el": 8.07703, "hi": 7.86762, "ta": 7.77939, "th": 7.88072, "zh": 8.07723, "my": 8.00545}, "gini_cpc": 0.386509, "gini_cps": 0.404945, "bpic": 1.312467, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.16808278813748, "vi": 0.6758501231221409, "tr": 1.2476182089518795, "fi": 0.7878078701448744, "ru": 1.139472654817711, "el": 3.454087265512159, "hi": 4.916499098501726, "ta": 5.182397609913587, "th": 5.5286526681508334, "zh": 2.058948454159251, "my": 5.466991012678191}, "t": 15.5} |
| {"event": "probe", "step": 8, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 9874.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.567373, "cps": 77.140625, "mean_p": 0.506549, "boundary_entropy": 0.687399, "bpb": 7.492979}, "de": {"chunks": 12844.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.62553, "cps": 100.34375, "mean_p": 0.508773, "boundary_entropy": 0.688146, "bpb": 7.509399}, "vi": {"chunks": 7010.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.380255, "cps": 54.765625, "mean_p": 0.492942, "boundary_entropy": 0.688578, "bpb": 7.758908}, "tr": {"chunks": 12155.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.671436, "cps": 94.960938, "mean_p": 0.508644, "boundary_entropy": 0.687819, "bpb": 7.683246}, "fi": {"chunks": 8478.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.450694, "cps": 66.234375, "mean_p": 0.500042, "boundary_entropy": 0.687968, "bpb": 7.597206}, "ru": {"chunks": 13427.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.708549, "cps": 104.898438, "mean_p": 0.499181, "boundary_entropy": 0.690232, "bpb": 7.37815}, "el": {"chunks": 36274.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.76748, "cps": 283.390625, "mean_p": 0.527147, "boundary_entropy": 0.689114, "bpb": 7.832756}, "hi": {"chunks": 42727.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.539495, "cps": 333.804688, "mean_p": 0.523175, "boundary_entropy": 0.689992, "bpb": 7.592518}, "ta": {"chunks": 50053.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.677347, "cps": 391.039062, "mean_p": 0.522665, "boundary_entropy": 0.690283, "bpb": 7.469515}, "th": {"chunks": 45806.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.847746, "cps": 357.859375, "mean_p": 0.528656, "boundary_entropy": 0.689593, "bpb": 7.595378}, "zh": {"chunks": 6495.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.099729, "cps": 50.742188, "mean_p": 0.497731, "boundary_entropy": 0.686577, "bpb": 8.047386}, "my": {"chunks": 53313.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.813648, "cps": 416.507812, "mean_p": 0.520768, "boundary_entropy": 0.690626, "bpb": 7.827826}}, "bpb": {"en": 7.49298, "de": 7.5094, "vi": 7.75891, "tr": 7.68325, "fi": 7.59721, "ru": 7.37815, "el": 7.83276, "hi": 7.59252, "ta": 7.46952, "th": 7.59538, "zh": 8.04739, "my": 7.82783}, "gini_cpc": 0.371386, "gini_cps": 0.391148, "bpic": 1.285888, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1025007334167685, "vi": 0.6702022373690436, "tr": 1.1834104748276608, "fi": 0.7943511454146283, "ru": 1.2488226642502926, "el": 3.1151976088600892, "hi": 4.475878873208068, "ta": 4.718844219730791, "th": 5.019174568163146, "zh": 1.9382808726904477, "my": 4.959075757328918}, "t": 19.8} |
| {"event": "probe", "step": 16, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 12426.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.714015, "cps": 97.078125, "mean_p": 0.513152, "boundary_entropy": 0.687254, "bpb": 6.578603}, "de": {"chunks": 15015.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.731262, "cps": 117.304688, "mean_p": 0.513043, "boundary_entropy": 0.68809, "bpb": 6.667572}, "vi": {"chunks": 8286.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.449471, "cps": 64.734375, "mean_p": 0.496166, "boundary_entropy": 0.688732, "bpb": 7.130056}, "tr": {"chunks": 13045.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.720599, "cps": 101.914062, "mean_p": 0.511218, "boundary_entropy": 0.687758, "bpb": 7.005233}, "fi": {"chunks": 10685.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.568019, "cps": 83.476562, "mean_p": 0.504812, "boundary_entropy": 0.687976, "bpb": 6.798942}, "ru": {"chunks": 24015.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.267282, "cps": 187.617188, "mean_p": 0.507546, "boundary_entropy": 0.690208, "bpb": 6.537745}, "el": {"chunks": 34664.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.689032, "cps": 270.8125, "mean_p": 0.520055, "boundary_entropy": 0.689755, "bpb": 7.360614}, "hi": {"chunks": 42922.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.551085, "cps": 335.328125, "mean_p": 0.527615, "boundary_entropy": 0.689545, "bpb": 6.8241}, "ta": {"chunks": 50264.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.688633, "cps": 392.6875, "mean_p": 0.522954, "boundary_entropy": 0.690284, "bpb": 6.753415}, "th": {"chunks": 45879.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.852285, "cps": 358.429688, "mean_p": 0.533563, "boundary_entropy": 0.689012, "bpb": 6.6804}, "zh": {"chunks": 7369.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.247714, "cps": 57.570312, "mean_p": 0.500829, "boundary_entropy": 0.686663, "bpb": 8.010901}, "my": {"chunks": 53172.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.806206, "cps": 415.40625, "mean_p": 0.518972, "boundary_entropy": 0.690759, "bpb": 7.392379}}, "bpb": {"en": 6.5786, "de": 6.66757, "vi": 7.13006, "tr": 7.00523, "fi": 6.79894, "ru": 6.53774, "el": 7.36061, "hi": 6.8241, "ta": 6.75342, "th": 6.6804, "zh": 8.0109, "my": 7.39238}, "gini_cpc": 0.330143, "gini_cps": 0.355196, "bpic": 1.207838, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0241550252172646, "vi": 0.6294982946725658, "tr": 1.0092210560883796, "fi": 0.7955278973815135, "ru": 1.7748683605360622, "el": 2.365541664893268, "hi": 3.5728735678366004, "ta": 3.7655147066515022, "th": 3.9947136072593583, "zh": 1.7474625809172284, "my": 3.9301795444738996}, "t": 25.7} |
| {"event": "probe", "step": 32, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 17201.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.988393, "cps": 134.382812, "mean_p": 0.554044, "boundary_entropy": 0.68246, "bpb": 5.889951}, "de": {"chunks": 20618.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.00414, "cps": 161.078125, "mean_p": 0.55489, "boundary_entropy": 0.683053, "bpb": 6.036471}, "vi": {"chunks": 22035.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.195281, "cps": 172.148438, "mean_p": 0.525467, "boundary_entropy": 0.68781, "bpb": 6.706611}, "tr": {"chunks": 19286.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.065348, "cps": 150.671875, "mean_p": 0.5478, "boundary_entropy": 0.683983, "bpb": 6.428165}, "fi": {"chunks": 19106.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 1.015682, "cps": 149.265625, "mean_p": 0.546597, "boundary_entropy": 0.684245, "bpb": 6.084328}, "ru": {"chunks": 33715.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.779156, "cps": 263.398438, "mean_p": 0.52411, "boundary_entropy": 0.689407, "bpb": 5.639004}, "el": {"chunks": 4737.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.230814, "cps": 37.007812, "mean_p": 0.489256, "boundary_entropy": 0.690178, "bpb": 6.896004}, "hi": {"chunks": 43139.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.563982, "cps": 337.023438, "mean_p": 0.546704, "boundary_entropy": 0.686785, "bpb": 5.816789}, "ta": {"chunks": 50838.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.719337, "cps": 397.171875, "mean_p": 0.530645, "boundary_entropy": 0.689579, "bpb": 5.999262}, "th": {"chunks": 46042.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.862418, "cps": 359.703125, "mean_p": 0.541192, "boundary_entropy": 0.68794, "bpb": 5.411074}, "zh": {"chunks": 12239.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.072299, "cps": 95.617188, "mean_p": 0.519309, "boundary_entropy": 0.686153, "bpb": 7.731869}, "my": {"chunks": 53382.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.817289, "cps": 417.046875, "mean_p": 0.52268, "boundary_entropy": 0.690461, "bpb": 6.195946}}, "bpb": {"en": 5.88995, "de": 6.03647, "vi": 6.70661, "tr": 6.42816, "fi": 6.08433, "ru": 5.639, "el": 6.896, "hi": 5.81679, "ta": 5.99926, "th": 5.41107, "zh": 7.73187, "my": 6.19595}, "gini_cpc": 0.282071, "gini_cps": 0.304096, "bpic": 1.121059, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0159317951942626, "vi": 1.2093174990441122, "tr": 1.0778592058967345, "fi": 1.0276099821252107, "ru": 1.800049193310998, "el": 0.23352477586413242, "hi": 2.5940923024189217, "ta": 2.7512712607642262, "th": 2.8960332221469582, "zh": 2.0966354108874627, "my": 2.8503742712868627}, "t": 34.7} |
| {"event": "probe", "step": 64, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 17390.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.999253, "cps": 135.859375, "mean_p": 0.772944, "boundary_entropy": 0.530571, "bpb": 5.452789}, "de": {"chunks": 20829.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.014416, "cps": 162.726562, "mean_p": 0.774931, "boundary_entropy": 0.528756, "bpb": 5.590548}, "vi": {"chunks": 24126.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.308706, "cps": 188.484375, "mean_p": 0.692939, "boundary_entropy": 0.609086, "bpb": 6.281355}, "tr": {"chunks": 19686.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.087444, "cps": 153.796875, "mean_p": 0.762862, "boundary_entropy": 0.541971, "bpb": 6.027599}, "fi": {"chunks": 19514.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 1.037372, "cps": 152.453125, "mean_p": 0.763566, "boundary_entropy": 0.541391, "bpb": 5.677271}, "ru": {"chunks": 34557.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.823588, "cps": 269.976562, "mean_p": 0.669475, "boundary_entropy": 0.632729, "bpb": 4.950482}, "el": {"chunks": 31189.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.51971, "cps": 243.664062, "mean_p": 0.509714, "boundary_entropy": 0.690441, "bpb": 5.395194}, "hi": {"chunks": 43166.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.565587, "cps": 337.234375, "mean_p": 0.585067, "boundary_entropy": 0.676921, "bpb": 4.902849}, "ta": {"chunks": 50891.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.722172, "cps": 397.585938, "mean_p": 0.579493, "boundary_entropy": 0.679002, "bpb": 4.80841}, "th": {"chunks": 46084.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.86503, "cps": 360.03125, "mean_p": 0.572974, "boundary_entropy": 0.680707, "bpb": 4.667578}, "zh": {"chunks": 16107.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.727227, "cps": 125.835938, "mean_p": 0.620376, "boundary_entropy": 0.657406, "bpb": 6.961089}, "my": {"chunks": 53703.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.834231, "cps": 419.554688, "mean_p": 0.595059, "boundary_entropy": 0.673491, "bpb": 4.831965}}, "bpb": {"en": 5.45279, "de": 5.59055, "vi": 6.28136, "tr": 6.0276, "fi": 5.67727, "ru": 4.95048, "el": 5.39519, "hi": 4.90285, "ta": 4.80841, "th": 4.66758, "zh": 6.96109, "my": 4.83196}, "gini_cpc": 0.225801, "gini_cps": 0.235636, "bpic": 1.017334, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0151741511378463, "vi": 1.3096845965644333, "tr": 1.0882569954554027, "fi": 1.0381472441170423, "ru": 1.8249516250658113, "el": 1.5208456622406306, "hi": 2.5675048442897705, "ta": 2.724206679635768, "th": 2.867171300825492, "zh": 2.729265304024114, "my": 2.8363492718730345}, "t": 49.5} |
| {"event": "train", "step": 100, "phase": 1, "objective": "global", "lm_loss": 3.33227, "bpb": 4.80745, "chunk_loss": 0.007894, "F": 0.00098, "G": 0.78435, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.49437, "router_noise": 0.0, "grad_norm": 1.4346, "lr": 6.06e-05, "router_trainable": true, "sec_per_step": 0.574, "alpha": 0.03, "t": 63.3} |
| {"event": "probe", "step": 128, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 17418.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 1.000862, "cps": 136.078125, "mean_p": 0.842744, "boundary_entropy": 0.427414, "bpb": 4.346018}, "de": {"chunks": 20861.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.015974, "cps": 162.976562, "mean_p": 0.839772, "boundary_entropy": 0.428704, "bpb": 4.569771}, "vi": {"chunks": 23873.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.294982, "cps": 186.507812, "mean_p": 0.678357, "boundary_entropy": 0.615304, "bpb": 4.762373}, "tr": {"chunks": 19717.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.089156, "cps": 154.039062, "mean_p": 0.777791, "boundary_entropy": 0.511596, "bpb": 4.833701}, "fi": {"chunks": 19571.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 1.040402, "cps": 152.898438, "mean_p": 0.82521, "boundary_entropy": 0.446498, "bpb": 4.576677}, "ru": {"chunks": 34560.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.823747, "cps": 270.0, "mean_p": 0.768705, "boundary_entropy": 0.537474, "bpb": 3.203081}, "el": {"chunks": 37259.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.815475, "cps": 291.085938, "mean_p": 0.751003, "boundary_entropy": 0.556606, "bpb": 3.799941}, "hi": {"chunks": 43173.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.566003, "cps": 337.289062, "mean_p": 0.691729, "boundary_entropy": 0.611689, "bpb": 2.789669}, "ta": {"chunks": 50915.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.723455, "cps": 397.773438, "mean_p": 0.703416, "boundary_entropy": 0.60573, "bpb": 2.960236}, "th": {"chunks": 46085.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.865092, "cps": 360.039062, "mean_p": 0.693136, "boundary_entropy": 0.610233, "bpb": 2.738972}, "zh": {"chunks": 16172.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.738232, "cps": 126.34375, "mean_p": 0.867532, "boundary_entropy": 0.384215, "bpb": 6.187697}, "my": {"chunks": 53741.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.836236, "cps": 419.851562, "mean_p": 0.731438, "boundary_entropy": 0.578712, "bpb": 2.835911}}, "bpb": {"en": 4.34602, "de": 4.56977, "vi": 4.76237, "tr": 4.8337, "fi": 4.57668, "ru": 3.20308, "el": 3.79994, "hi": 2.78967, "ta": 2.96024, "th": 2.73897, "zh": 6.1877, "my": 2.83591}, "gini_cpc": 0.222006, "gini_cps": 0.23423, "bpic": 1.001137, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0150993505006827, "vi": 1.293867160044401, "tr": 1.088218535688413, "fi": 1.0395059211933828, "ru": 1.8221761311426539, "el": 1.8139118728004668, "hi": 2.563793186225859, "ta": 2.721110089193083, "th": 2.8626243461390897, "zh": 2.7358742004575296, "my": 2.8337935100368243}, "t": 76.0} |
| {"event": "train", "step": 200, "phase": 1, "objective": "global", "lm_loss": 2.45862, "bpb": 3.54704, "chunk_loss": 0.003752, "F": 0.00098, "G": 0.90008, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.29174, "router_noise": 0.0, "grad_norm": 1.9108, "lr": 0.0001206, "router_trainable": true, "sec_per_step": 0.399, "alpha": 0.03, "t": 103.2} |
| {"event": "probe", "step": 256, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 17410.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 1.000402, "cps": 136.015625, "mean_p": 0.949323, "boundary_entropy": 0.189186, "bpb": 3.527463}, "de": {"chunks": 20727.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.009448, "cps": 161.929688, "mean_p": 0.935982, "boundary_entropy": 0.212, "bpb": 3.782226}, "vi": {"chunks": 24240.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.31489, "cps": 189.375, "mean_p": 0.876035, "boundary_entropy": 0.346441, "bpb": 3.122798}, "tr": {"chunks": 19370.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.069988, "cps": 151.328125, "mean_p": 0.900142, "boundary_entropy": 0.258507, "bpb": 3.768141}, "fi": {"chunks": 19380.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 1.030248, "cps": 151.40625, "mean_p": 0.926156, "boundary_entropy": 0.220164, "bpb": 3.832356}, "ru": {"chunks": 34548.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.823113, "cps": 269.90625, "mean_p": 0.971002, "boundary_entropy": 0.119497, "bpb": 2.658165}, "el": {"chunks": 37225.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.813819, "cps": 290.820312, "mean_p": 0.94755, "boundary_entropy": 0.198991, "bpb": 2.71668}, "hi": {"chunks": 43172.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.565944, "cps": 337.28125, "mean_p": 0.947746, "boundary_entropy": 0.201929, "bpb": 2.102597}, "ta": {"chunks": 50915.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.723455, "cps": 397.773438, "mean_p": 0.94312, "boundary_entropy": 0.214995, "bpb": 1.9522}, "th": {"chunks": 46079.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.864719, "cps": 359.992188, "mean_p": 0.963036, "boundary_entropy": 0.154544, "bpb": 2.086967}, "zh": {"chunks": 16155.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.735354, "cps": 126.210938, "mean_p": 0.927151, "boundary_entropy": 0.251815, "bpb": 4.525037}, "my": {"chunks": 53742.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.836289, "cps": 419.859375, "mean_p": 0.931697, "boundary_entropy": 0.245613, "bpb": 2.156606}}, "bpb": {"en": 3.52746, "de": 3.78223, "vi": 3.1228, "tr": 3.76814, "fi": 3.83236, "ru": 2.65817, "el": 2.71668, "hi": 2.1026, "ta": 1.9522, "th": 2.08697, "zh": 4.52504, "my": 2.15661}, "gini_cpc": 0.222761, "gini_cps": 0.235168, "bpic": 1.002136, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.009042338731955, "vi": 1.314361479635592, "tr": 1.0695581918553172, "fi": 1.0298340293699981, "ru": 1.8223804412894664, "el": 1.8130893644689363, "hi": 2.564911853190213, "ta": 2.722360455690128, "th": 2.8635668709292625, "zh": 2.7342540797596193, "my": 2.835148410044383}, "t": 125.4} |
| {"event": "train", "step": 300, "phase": 1, "objective": "global", "lm_loss": 2.11136, "bpb": 3.04605, "chunk_loss": 0.001557, "F": 0.00098, "G": 0.96141, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.14729, "router_noise": 0.0, "grad_norm": 1.3497, "lr": 0.0001806, "router_trainable": true, "sec_per_step": 0.38, "alpha": 0.03, "t": 141.2} |
| {"event": "train", "step": 400, "phase": 1, "objective": "global", "lm_loss": 2.02837, "bpb": 2.92631, "chunk_loss": 0.000717, "F": 0.00098, "G": 0.98487, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.07543, "router_noise": 0.0, "grad_norm": 1.2936, "lr": 0.0002406, "router_trainable": true, "sec_per_step": 0.341, "alpha": 0.03, "t": 175.3} |
| {"event": "train", "step": 500, "phase": 1, "objective": "global", "lm_loss": 1.94609, "bpb": 2.80762, "chunk_loss": 0.000329, "F": 0.00098, "G": 0.99572, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.02712, "router_noise": 0.0, "grad_norm": 1.1648, "lr": 0.0003, "router_trainable": true, "sec_per_step": 0.321, "alpha": 0.03, "t": 207.4} |
| {"event": "train", "step": 600, "phase": 1, "objective": "global", "lm_loss": 1.82923, "bpb": 2.63903, "chunk_loss": 0.000239, "F": 0.00098, "G": 0.99823, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.01283, "router_noise": 0.0, "grad_norm": 0.8679, "lr": 0.0002999, "router_trainable": true, "sec_per_step": 0.298, "alpha": 0.03, "t": 237.2} |
| {"event": "train", "step": 700, "phase": 1, "objective": "global", "lm_loss": 1.71723, "bpb": 2.47744, "chunk_loss": 0.000212, "F": 0.00098, "G": 0.99898, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00793, "router_noise": 0.0, "grad_norm": 0.9346, "lr": 0.0002998, "router_trainable": true, "sec_per_step": 0.277, "alpha": 0.03, "t": 264.9} |
| {"event": "train", "step": 800, "phase": 1, "objective": "global", "lm_loss": 1.6051, "bpb": 2.31567, "chunk_loss": 0.000198, "F": 0.00098, "G": 0.99939, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00506, "router_noise": 0.0, "grad_norm": 0.7884, "lr": 0.0002995, "router_trainable": true, "sec_per_step": 0.26, "alpha": 0.03, "t": 290.9} |
| {"event": "train", "step": 900, "phase": 1, "objective": "global", "lm_loss": 1.58583, "bpb": 2.28788, "chunk_loss": 0.00019, "F": 0.00098, "G": 0.99962, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00337, "router_noise": 0.0, "grad_norm": 0.7166, "lr": 0.000299, "router_trainable": true, "sec_per_step": 0.247, "alpha": 0.03, "t": 315.6} |
| {"event": "probe", "step": 1000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 17418.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 1.000862, "cps": 136.078125, "mean_p": 0.999668, "boundary_entropy": 0.002949, "bpb": 2.343393}, "de": {"chunks": 20862.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.016023, "cps": 162.984375, "mean_p": 0.999638, "boundary_entropy": 0.003201, "bpb": 2.621267}, "vi": {"chunks": 24274.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.316734, "cps": 189.640625, "mean_p": 0.999787, "boundary_entropy": 0.001999, "bpb": 2.1194}, "tr": {"chunks": 19742.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.090537, "cps": 154.234375, "mean_p": 0.999666, "boundary_entropy": 0.002987, "bpb": 2.628595}, "fi": {"chunks": 19572.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 1.040455, "cps": 152.90625, "mean_p": 0.999668, "boundary_entropy": 0.00297, "bpb": 2.860241}, "ru": {"chunks": 34560.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.823747, "cps": 270.0, "mean_p": 0.999708, "boundary_entropy": 0.002643, "bpb": 1.702272}, "el": {"chunks": 37264.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.815719, "cps": 291.125, "mean_p": 0.99964, "boundary_entropy": 0.003183, "bpb": 1.788505}, "hi": {"chunks": 43173.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.566003, "cps": 337.289062, "mean_p": 0.999689, "boundary_entropy": 0.002791, "bpb": 1.399524}, "ta": {"chunks": 50915.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.723455, "cps": 397.773438, "mean_p": 0.999753, "boundary_entropy": 0.002281, "bpb": 1.129153}, "th": {"chunks": 46085.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.865092, "cps": 360.039062, "mean_p": 0.999749, "boundary_entropy": 0.002307, "bpb": 1.352441}, "zh": {"chunks": 16173.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.738402, "cps": 126.351562, "mean_p": 0.999646, "boundary_entropy": 0.003143, "bpb": 3.478443}, "my": {"chunks": 53743.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.836342, "cps": 419.867188, "mean_p": 0.999799, "boundary_entropy": 0.001897, "bpb": 1.293504}}, "bpb": {"en": 2.34339, "de": 2.62127, "vi": 2.1194, "tr": 2.6286, "fi": 2.86024, "ru": 1.70227, "el": 1.7885, "hi": 1.39952, "ta": 1.12915, "th": 1.35244, "zh": 3.47844, "my": 1.2935}, "gini_cpc": 0.221516, "gini_cps": 0.233853, "bpic": 1.0, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.01514801064883, "vi": 1.3156005295906583, "tr": 1.0895983329898389, "fi": 1.0395590357977051, "ru": 1.8221761311426539, "el": 1.8141552920914839, "hi": 2.563793186225859, "ta": 2.721110089193083, "th": 2.8626243461390897, "zh": 2.7360433739796948, "my": 2.8338989711748765}, "t": 344.4} |
| {"event": "eval", "step": 1000, "phase": 1, "bpb": {"en": 2.35439, "de": 2.61414, "vi": 2.12392, "tr": 2.61849, "fi": 2.87071, "ru": 1.70357, "el": 1.78394, "hi": 1.40791, "ta": 1.12902, "th": 1.34951, "zh": 3.48699, "my": 1.3052}, "bytes_per_chunk": {"en": 1024.0, "de": 1024.0, "vi": 1024.0, "tr": 1024.0, "fi": 1024.0, "ru": 1024.0, "el": 1024.0, "hi": 1024.0, "ta": 1024.0, "th": 1024.0, "zh": 1024.0, "my": 1024.0}, "bpb_mean": 2.06232, "t": 348.8} |
| {"event": "train", "step": 1000, "phase": 1, "objective": "global", "lm_loss": 1.50444, "bpb": 2.17044, "chunk_loss": 0.000187, "F": 0.00098, "G": 0.99968, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00284, "router_noise": 0.0, "grad_norm": 0.7789, "lr": 0.0002985, "router_trainable": true, "sec_per_step": 0.335, "alpha": 0.03, "t": 349.1} |
| {"event": "train", "step": 1100, "phase": 1, "objective": "global", "lm_loss": 1.43021, "bpb": 2.06336, "chunk_loss": 0.000184, "F": 0.00098, "G": 0.99977, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.0021, "router_noise": 0.0, "grad_norm": 0.673, "lr": 0.0002978, "router_trainable": true, "sec_per_step": 0.249, "alpha": 0.03, "t": 374.1} |
| {"event": "train", "step": 1200, "phase": 1, "objective": "global", "lm_loss": 1.46969, "bpb": 2.12031, "chunk_loss": 0.000182, "F": 0.00098, "G": 0.99982, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00169, "router_noise": 0.0, "grad_norm": 0.6682, "lr": 0.000297, "router_trainable": true, "sec_per_step": 0.249, "alpha": 0.03, "t": 399.0} |
| {"event": "train", "step": 1300, "phase": 1, "objective": "global", "lm_loss": 1.47057, "bpb": 2.12158, "chunk_loss": 0.000181, "F": 0.00098, "G": 0.99985, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00144, "router_noise": 0.0, "grad_norm": 0.6644, "lr": 0.0002962, "router_trainable": true, "sec_per_step": 0.242, "alpha": 0.03, "t": 423.1} |
| {"event": "train", "step": 1400, "phase": 1, "objective": "global", "lm_loss": 1.4439, "bpb": 2.08311, "chunk_loss": 0.00018, "F": 0.00098, "G": 0.99987, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00129, "router_noise": 0.0, "grad_norm": 0.5787, "lr": 0.0002951, "router_trainable": true, "sec_per_step": 0.243, "alpha": 0.03, "t": 447.4} |
| {"event": "train", "step": 1500, "phase": 1, "objective": "global", "lm_loss": 1.44225, "bpb": 2.08073, "chunk_loss": 0.00018, "F": 0.00098, "G": 0.99988, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00118, "router_noise": 0.0, "grad_norm": 0.639, "lr": 0.000294, "router_trainable": true, "sec_per_step": 0.234, "alpha": 0.03, "t": 470.8} |
| {"event": "train", "step": 1600, "phase": 1, "objective": "global", "lm_loss": 1.42855, "bpb": 2.06097, "chunk_loss": 0.00018, "F": 0.00098, "G": 0.99989, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00113, "router_noise": 0.0, "grad_norm": 0.6069, "lr": 0.0002928, "router_trainable": true, "sec_per_step": 0.239, "alpha": 0.03, "t": 494.7} |
| {"event": "train", "step": 1700, "phase": 1, "objective": "global", "lm_loss": 1.37297, "bpb": 1.98078, "chunk_loss": 0.00018, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00106, "router_noise": 0.0, "grad_norm": 0.574, "lr": 0.0002914, "router_trainable": true, "sec_per_step": 0.239, "alpha": 0.03, "t": 518.6} |
| {"event": "train", "step": 1800, "phase": 1, "objective": "global", "lm_loss": 1.38153, "bpb": 1.99312, "chunk_loss": 0.00018, "F": 0.00098, "G": 0.99989, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00108, "router_noise": 0.0, "grad_norm": 0.5773, "lr": 0.0002899, "router_trainable": true, "sec_per_step": 0.239, "alpha": 0.03, "t": 542.5} |
| {"event": "train", "step": 1900, "phase": 1, "objective": "global", "lm_loss": 1.37572, "bpb": 1.98474, "chunk_loss": 0.00018, "F": 0.00098, "G": 0.99989, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00107, "router_noise": 0.0, "grad_norm": 0.5642, "lr": 0.0002883, "router_trainable": true, "sec_per_step": 0.238, "alpha": 0.03, "t": 566.3} |
| {"event": "probe", "step": 2000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 17418.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 1.000862, "cps": 136.078125, "mean_p": 0.999898, "boundary_entropy": 0.001038, "bpb": 2.115424}, "de": {"chunks": 20862.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.016023, "cps": 162.984375, "mean_p": 0.999898, "boundary_entropy": 0.00104, "bpb": 2.309067}, "vi": {"chunks": 24274.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.316734, "cps": 189.640625, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 1.979699}, "tr": {"chunks": 19742.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.090537, "cps": 154.234375, "mean_p": 0.999899, "boundary_entropy": 0.001033, "bpb": 2.301262}, "fi": {"chunks": 19572.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 1.040455, "cps": 152.90625, "mean_p": 0.999899, "boundary_entropy": 0.001032, "bpb": 2.514666}, "ru": {"chunks": 34560.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.823747, "cps": 270.0, "mean_p": 0.999895, "boundary_entropy": 0.001064, "bpb": 1.440728}, "el": {"chunks": 37264.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.815719, "cps": 291.125, "mean_p": 0.999894, "boundary_entropy": 0.001072, "bpb": 1.503991}, "hi": {"chunks": 43173.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.566003, "cps": 337.289062, "mean_p": 0.999899, "boundary_entropy": 0.00103, "bpb": 1.219988}, "ta": {"chunks": 50915.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.723455, "cps": 397.773438, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 1.027159}, "th": {"chunks": 46085.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.865092, "cps": 360.039062, "mean_p": 0.999899, "boundary_entropy": 0.001029, "bpb": 1.159326}, "zh": {"chunks": 16173.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.738402, "cps": 126.351562, "mean_p": 0.9999, "boundary_entropy": 0.001023, "bpb": 3.205515}, "my": {"chunks": 53743.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.836342, "cps": 419.867188, "mean_p": 0.999899, "boundary_entropy": 0.001033, "bpb": 1.152068}}, "bpb": {"en": 2.11542, "de": 2.30907, "vi": 1.9797, "tr": 2.30126, "fi": 2.51467, "ru": 1.44073, "el": 1.50399, "hi": 1.21999, "ta": 1.02716, "th": 1.15933, "zh": 3.20552, "my": 1.15207}, "gini_cpc": 0.221516, "gini_cps": 0.233853, "bpic": 1.0, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.01514801064883, "vi": 1.3156005295906583, "tr": 1.0895983329898389, "fi": 1.0395590357977051, "ru": 1.8221761311426539, "el": 1.8141552920914839, "hi": 2.563793186225859, "ta": 2.721110089193083, "th": 2.8626243461390897, "zh": 2.7360433739796948, "my": 2.8338989711748765}, "t": 592.7} |
| {"event": "eval", "step": 2000, "phase": 1, "bpb": {"en": 2.12513, "de": 2.30282, "vi": 1.98721, "tr": 2.29237, "fi": 2.52908, "ru": 1.44082, "el": 1.4991, "hi": 1.2277, "ta": 1.02703, "th": 1.15576, "zh": 3.21214, "my": 1.16239}, "bytes_per_chunk": {"en": 1024.0, "de": 1024.0, "vi": 1024.0, "tr": 1024.0, "fi": 1024.0, "ru": 1024.0, "el": 1024.0, "hi": 1024.0, "ta": 1024.0, "th": 1024.0, "zh": 1024.0, "my": 1024.0}, "bpb_mean": 1.83013, "t": 597.0} |
| {"event": "train", "step": 2000, "phase": 1, "objective": "global", "lm_loss": 1.37427, "bpb": 1.98265, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00105, "router_noise": 0.0, "grad_norm": 0.573, "lr": 0.0002866, "router_trainable": true, "sec_per_step": 0.309, "alpha": 0.03, "t": 597.3} |
| {"event": "train", "step": 2100, "phase": 1, "objective": "global", "lm_loss": 1.31461, "bpb": 1.89658, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00104, "router_noise": 0.0, "grad_norm": 0.5634, "lr": 0.0002848, "router_trainable": true, "sec_per_step": 0.239, "alpha": 0.03, "t": 621.1} |
| {"event": "train", "step": 2200, "phase": 1, "objective": "global", "lm_loss": 1.28059, "bpb": 1.84751, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00104, "router_noise": 0.0, "grad_norm": 0.5217, "lr": 0.0002829, "router_trainable": true, "sec_per_step": 0.234, "alpha": 0.03, "t": 644.5} |
| {"event": "train", "step": 2300, "phase": 1, "objective": "global", "lm_loss": 1.35358, "bpb": 1.9528, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00104, "router_noise": 0.0, "grad_norm": 0.5081, "lr": 0.0002809, "router_trainable": true, "sec_per_step": 0.235, "alpha": 0.03, "t": 668.0} |
| {"event": "train", "step": 2400, "phase": 1, "objective": "global", "lm_loss": 1.36819, "bpb": 1.97388, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00103, "router_noise": 0.0, "grad_norm": 0.5677, "lr": 0.0002788, "router_trainable": true, "sec_per_step": 0.239, "alpha": 0.03, "t": 691.9} |
| {"event": "train", "step": 2500, "phase": 1, "objective": "global", "lm_loss": 1.28511, "bpb": 1.85402, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00103, "router_noise": 0.0, "grad_norm": 0.5312, "lr": 0.0002765, "router_trainable": true, "sec_per_step": 0.235, "alpha": 0.03, "t": 715.5} |
| {"event": "train", "step": 2600, "phase": 1, "objective": "global", "lm_loss": 1.29792, "bpb": 1.8725, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00103, "router_noise": 0.0, "grad_norm": 0.5451, "lr": 0.0002742, "router_trainable": true, "sec_per_step": 0.239, "alpha": 0.03, "t": 739.4} |
| {"event": "train", "step": 2700, "phase": 1, "objective": "global", "lm_loss": 1.29995, "bpb": 1.87543, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00103, "router_noise": 0.0, "grad_norm": 0.4996, "lr": 0.0002718, "router_trainable": true, "sec_per_step": 0.235, "alpha": 0.03, "t": 762.9} |
| {"event": "train", "step": 2800, "phase": 1, "objective": "global", "lm_loss": 1.38568, "bpb": 1.99912, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.5191, "lr": 0.0002693, "router_trainable": true, "sec_per_step": 0.235, "alpha": 0.03, "t": 786.4} |
| {"event": "train", "step": 2900, "phase": 1, "objective": "global", "lm_loss": 1.30117, "bpb": 1.8772, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00103, "router_noise": 0.0, "grad_norm": 0.4732, "lr": 0.0002667, "router_trainable": true, "sec_per_step": 0.239, "alpha": 0.03, "t": 810.3} |
| {"event": "probe", "step": 3000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 17418.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 1.000862, "cps": 136.078125, "mean_p": 0.9999, "boundary_entropy": 0.001015, "bpb": 2.026366}, "de": {"chunks": 20862.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.016023, "cps": 162.984375, "mean_p": 0.9999, "boundary_entropy": 0.001018, "bpb": 2.187737}, "vi": {"chunks": 24274.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.316734, "cps": 189.640625, "mean_p": 0.9999, "boundary_entropy": 0.001017, "bpb": 1.893384}, "tr": {"chunks": 19742.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.090537, "cps": 154.234375, "mean_p": 0.9999, "boundary_entropy": 0.001018, "bpb": 2.157152}, "fi": {"chunks": 19572.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 1.040455, "cps": 152.90625, "mean_p": 0.9999, "boundary_entropy": 0.001016, "bpb": 2.348344}, "ru": {"chunks": 34560.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.823747, "cps": 270.0, "mean_p": 0.999899, "boundary_entropy": 0.001028, "bpb": 1.346939}, "el": {"chunks": 37264.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.815719, "cps": 291.125, "mean_p": 0.9999, "boundary_entropy": 0.001025, "bpb": 1.392886}, "hi": {"chunks": 43173.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.566003, "cps": 337.289062, "mean_p": 0.9999, "boundary_entropy": 0.001022, "bpb": 1.136117}, "ta": {"chunks": 50915.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.723455, "cps": 397.773438, "mean_p": 0.9999, "boundary_entropy": 0.00102, "bpb": 0.952042}, "th": {"chunks": 46085.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.865092, "cps": 360.039062, "mean_p": 0.9999, "boundary_entropy": 0.001027, "bpb": 1.082195}, "zh": {"chunks": 16173.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.738402, "cps": 126.351562, "mean_p": 0.999901, "boundary_entropy": 0.001014, "bpb": 3.025043}, "my": {"chunks": 53743.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.836342, "cps": 419.867188, "mean_p": 0.9999, "boundary_entropy": 0.001022, "bpb": 1.079178}}, "bpb": {"en": 2.02637, "de": 2.18774, "vi": 1.89338, "tr": 2.15715, "fi": 2.34834, "ru": 1.34694, "el": 1.39289, "hi": 1.13612, "ta": 0.95204, "th": 1.08219, "zh": 3.02504, "my": 1.07918}, "gini_cpc": 0.221516, "gini_cps": 0.233853, "bpic": 1.0, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.01514801064883, "vi": 1.3156005295906583, "tr": 1.0895983329898389, "fi": 1.0395590357977051, "ru": 1.8221761311426539, "el": 1.8141552920914839, "hi": 2.563793186225859, "ta": 2.721110089193083, "th": 2.8626243461390897, "zh": 2.7360433739796948, "my": 2.8338989711748765}, "t": 836.4} |
| {"event": "eval", "step": 3000, "phase": 1, "bpb": {"en": 2.03468, "de": 2.18212, "vi": 1.89797, "tr": 2.1472, "fi": 2.36164, "ru": 1.34647, "el": 1.38897, "hi": 1.1445, "ta": 0.95285, "th": 1.07909, "zh": 3.03141, "my": 1.09064}, "bytes_per_chunk": {"en": 1024.0, "de": 1024.0, "vi": 1024.0, "tr": 1024.0, "fi": 1024.0, "ru": 1024.0, "el": 1024.0, "hi": 1024.0, "ta": 1024.0, "th": 1024.0, "zh": 1024.0, "my": 1024.0}, "bpb_mean": 1.72146, "t": 840.8} |
| {"event": "train", "step": 3000, "phase": 1, "objective": "global", "lm_loss": 1.35812, "bpb": 1.95935, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.5448, "lr": 0.000264, "router_trainable": true, "sec_per_step": 0.307, "alpha": 0.03, "t": 841.0} |
| {"event": "train", "step": 3100, "phase": 1, "objective": "global", "lm_loss": 1.35196, "bpb": 1.95046, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.5468, "lr": 0.0002612, "router_trainable": true, "sec_per_step": 0.233, "alpha": 0.03, "t": 864.4} |
| {"event": "train", "step": 3200, "phase": 1, "objective": "global", "lm_loss": 1.23416, "bpb": 1.78052, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.5215, "lr": 0.0002583, "router_trainable": true, "sec_per_step": 0.233, "alpha": 0.03, "t": 887.6} |
| {"event": "train", "step": 3300, "phase": 1, "objective": "global", "lm_loss": 1.30724, "bpb": 1.88594, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4698, "lr": 0.0002553, "router_trainable": true, "sec_per_step": 0.238, "alpha": 0.03, "t": 911.5} |
| {"event": "train", "step": 3400, "phase": 1, "objective": "global", "lm_loss": 1.26275, "bpb": 1.82176, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.481, "lr": 0.0002523, "router_trainable": true, "sec_per_step": 0.233, "alpha": 0.03, "t": 934.8} |
| {"event": "train", "step": 3500, "phase": 1, "objective": "global", "lm_loss": 1.34859, "bpb": 1.94561, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.488, "lr": 0.0002492, "router_trainable": true, "sec_per_step": 0.24, "alpha": 0.03, "t": 958.8} |
| {"event": "train", "step": 3600, "phase": 1, "objective": "global", "lm_loss": 1.26053, "bpb": 1.81856, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4893, "lr": 0.000246, "router_trainable": true, "sec_per_step": 0.232, "alpha": 0.03, "t": 982.0} |
| {"event": "train", "step": 3700, "phase": 1, "objective": "global", "lm_loss": 1.17015, "bpb": 1.68817, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4886, "lr": 0.0002427, "router_trainable": true, "sec_per_step": 0.238, "alpha": 0.03, "t": 1005.7} |
| {"event": "train", "step": 3800, "phase": 1, "objective": "global", "lm_loss": 1.26443, "bpb": 1.82419, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.5058, "lr": 0.0002394, "router_trainable": true, "sec_per_step": 0.226, "alpha": 0.03, "t": 1028.3} |
| {"event": "train", "step": 3900, "phase": 1, "objective": "global", "lm_loss": 1.28467, "bpb": 1.85338, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4706, "lr": 0.000236, "router_trainable": true, "sec_per_step": 0.239, "alpha": 0.03, "t": 1052.3} |
| {"event": "probe", "step": 4000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 17418.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 1.000862, "cps": 136.078125, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 1.961295}, "de": {"chunks": 20862.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.016023, "cps": 162.984375, "mean_p": 0.9999, "boundary_entropy": 0.001017, "bpb": 2.122522}, "vi": {"chunks": 24274.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.316734, "cps": 189.640625, "mean_p": 0.9999, "boundary_entropy": 0.001016, "bpb": 1.817192}, "tr": {"chunks": 19742.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.090537, "cps": 154.234375, "mean_p": 0.9999, "boundary_entropy": 0.001017, "bpb": 2.092641}, "fi": {"chunks": 19572.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 1.040455, "cps": 152.90625, "mean_p": 0.9999, "boundary_entropy": 0.001016, "bpb": 2.245749}, "ru": {"chunks": 34560.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.823747, "cps": 270.0, "mean_p": 0.9999, "boundary_entropy": 0.001021, "bpb": 1.291337}, "el": {"chunks": 37264.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.815719, "cps": 291.125, "mean_p": 0.9999, "boundary_entropy": 0.00102, "bpb": 1.310242}, "hi": {"chunks": 43173.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.566003, "cps": 337.289062, "mean_p": 0.9999, "boundary_entropy": 0.00102, "bpb": 1.079497}, "ta": {"chunks": 50915.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.723455, "cps": 397.773438, "mean_p": 0.9999, "boundary_entropy": 0.00102, "bpb": 0.917372}, "th": {"chunks": 46085.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.865092, "cps": 360.039062, "mean_p": 0.9999, "boundary_entropy": 0.001021, "bpb": 1.030809}, "zh": {"chunks": 16173.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.738402, "cps": 126.351562, "mean_p": 0.999901, "boundary_entropy": 0.001014, "bpb": 2.918217}, "my": {"chunks": 53743.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.836342, "cps": 419.867188, "mean_p": 0.9999, "boundary_entropy": 0.001022, "bpb": 1.043315}}, "bpb": {"en": 1.9613, "de": 2.12252, "vi": 1.81719, "tr": 2.09264, "fi": 2.24575, "ru": 1.29134, "el": 1.31024, "hi": 1.0795, "ta": 0.91737, "th": 1.03081, "zh": 2.91822, "my": 1.04332}, "gini_cpc": 0.221516, "gini_cps": 0.233853, "bpic": 1.0, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.01514801064883, "vi": 1.3156005295906583, "tr": 1.0895983329898389, "fi": 1.0395590357977051, "ru": 1.8221761311426539, "el": 1.8141552920914839, "hi": 2.563793186225859, "ta": 2.721110089193083, "th": 2.8626243461390897, "zh": 2.7360433739796948, "my": 2.8338989711748765}, "t": 1078.2} |
| {"event": "eval", "step": 4000, "phase": 1, "bpb": {"en": 1.96936, "de": 2.11731, "vi": 1.8249, "tr": 2.08428, "fi": 2.26352, "ru": 1.28979, "el": 1.30628, "hi": 1.08818, "ta": 0.91881, "th": 1.02935, "zh": 2.92456, "my": 1.05264}, "bytes_per_chunk": {"en": 1024.0, "de": 1024.0, "vi": 1024.0, "tr": 1024.0, "fi": 1024.0, "ru": 1024.0, "el": 1024.0, "hi": 1024.0, "ta": 1024.0, "th": 1024.0, "zh": 1024.0, "my": 1024.0}, "bpb_mean": 1.65575, "t": 1082.5} |
| {"event": "train", "step": 4000, "phase": 1, "objective": "global", "lm_loss": 1.34729, "bpb": 1.94373, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4965, "lr": 0.0002325, "router_trainable": true, "sec_per_step": 0.305, "alpha": 0.03, "t": 1082.8} |
| {"event": "train", "step": 4100, "phase": 1, "objective": "global", "lm_loss": 1.32923, "bpb": 1.91767, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4889, "lr": 0.000229, "router_trainable": true, "sec_per_step": 0.231, "alpha": 0.03, "t": 1105.9} |
| {"event": "train", "step": 4200, "phase": 1, "objective": "global", "lm_loss": 1.27595, "bpb": 1.8408, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4788, "lr": 0.0002254, "router_trainable": true, "sec_per_step": 0.233, "alpha": 0.03, "t": 1129.1} |
| {"event": "train", "step": 4300, "phase": 1, "objective": "global", "lm_loss": 1.24233, "bpb": 1.79231, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4668, "lr": 0.0002217, "router_trainable": true, "sec_per_step": 0.241, "alpha": 0.03, "t": 1153.2} |
| {"event": "phase_transition", "step": 4400, "phase": 2, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0, "t": 1176.2} |
| {"event": "train", "step": 4400, "phase": 2, "objective": "global", "lm_loss": 1.20537, "bpb": 1.73898, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4631, "lr": 0.0002181, "router_trainable": true, "sec_per_step": 0.232, "alpha": 0.03, "t": 1176.4} |
| {"event": "train", "step": 4500, "phase": 2, "objective": "global", "lm_loss": 1.2852, "bpb": 1.85416, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4746, "lr": 0.0002143, "router_trainable": true, "sec_per_step": 0.24, "alpha": 0.03, "t": 1200.4} |
| {"event": "train", "step": 4600, "phase": 2, "objective": "global", "lm_loss": 1.23425, "bpb": 1.78065, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.457, "lr": 0.0002105, "router_trainable": true, "sec_per_step": 0.233, "alpha": 0.03, "t": 1223.6} |
| {"event": "train", "step": 4700, "phase": 2, "objective": "global", "lm_loss": 1.15698, "bpb": 1.66916, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4299, "lr": 0.0002067, "router_trainable": true, "sec_per_step": 0.239, "alpha": 0.03, "t": 1247.5} |
| {"event": "train", "step": 4800, "phase": 2, "objective": "global", "lm_loss": 1.2914, "bpb": 1.8631, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4705, "lr": 0.0002029, "router_trainable": true, "sec_per_step": 0.236, "alpha": 0.03, "t": 1271.1} |
| {"event": "train", "step": 4900, "phase": 2, "objective": "global", "lm_loss": 1.29234, "bpb": 1.86445, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.467, "lr": 0.000199, "router_trainable": true, "sec_per_step": 0.235, "alpha": 0.03, "t": 1294.7} |
| {"event": "probe", "step": 5000, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 17418.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 1.000862, "cps": 136.078125, "mean_p": 0.999901, "boundary_entropy": 0.001014, "bpb": 1.920744}, "de": {"chunks": 20862.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.016023, "cps": 162.984375, "mean_p": 0.9999, "boundary_entropy": 0.001016, "bpb": 2.072264}, "vi": {"chunks": 24274.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.316734, "cps": 189.640625, "mean_p": 0.9999, "boundary_entropy": 0.001016, "bpb": 1.763379}, "tr": {"chunks": 19742.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.090537, "cps": 154.234375, "mean_p": 0.9999, "boundary_entropy": 0.001015, "bpb": 2.028785}, "fi": {"chunks": 19572.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 1.040455, "cps": 152.90625, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 2.176638}, "ru": {"chunks": 34560.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.823747, "cps": 270.0, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 1.255202}, "el": {"chunks": 37264.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.815719, "cps": 291.125, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 1.263727}, "hi": {"chunks": 43173.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.566003, "cps": 337.289062, "mean_p": 0.9999, "boundary_entropy": 0.00102, "bpb": 1.038555}, "ta": {"chunks": 50915.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.723455, "cps": 397.773438, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 0.877745}, "th": {"chunks": 46085.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.865092, "cps": 360.039062, "mean_p": 0.9999, "boundary_entropy": 0.00102, "bpb": 1.004972}, "zh": {"chunks": 16173.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.738402, "cps": 126.351562, "mean_p": 0.999901, "boundary_entropy": 0.001014, "bpb": 2.843075}, "my": {"chunks": 53743.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.836342, "cps": 419.867188, "mean_p": 0.9999, "boundary_entropy": 0.001021, "bpb": 1.004522}}, "bpb": {"en": 1.92074, "de": 2.07226, "vi": 1.76338, "tr": 2.02879, "fi": 2.17664, "ru": 1.2552, "el": 1.26373, "hi": 1.03855, "ta": 0.87774, "th": 1.00497, "zh": 2.84307, "my": 1.00452}, "gini_cpc": 0.221516, "gini_cps": 0.233853, "bpic": 1.0, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.01514801064883, "vi": 1.3156005295906583, "tr": 1.0895983329898389, "fi": 1.0395590357977051, "ru": 1.8221761311426539, "el": 1.8141552920914839, "hi": 2.563793186225859, "ta": 2.721110089193083, "th": 2.8626243461390897, "zh": 2.7360433739796948, "my": 2.8338989711748765}, "t": 1320.6} |
| {"event": "eval", "step": 5000, "phase": 2, "bpb": {"en": 1.92665, "de": 2.0674, "vi": 1.77298, "tr": 2.02249, "fi": 2.19241, "ru": 1.25453, "el": 1.26012, "hi": 1.04542, "ta": 0.87997, "th": 1.00294, "zh": 2.84986, "my": 1.01238}, "bytes_per_chunk": {"en": 1024.0, "de": 1024.0, "vi": 1024.0, "tr": 1024.0, "fi": 1024.0, "ru": 1024.0, "el": 1024.0, "hi": 1024.0, "ta": 1024.0, "th": 1024.0, "zh": 1024.0, "my": 1024.0}, "bpb_mean": 1.60726, "t": 1325.0} |
| {"event": "train", "step": 5000, "phase": 2, "objective": "global", "lm_loss": 1.25165, "bpb": 1.80575, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4689, "lr": 0.000195, "router_trainable": true, "sec_per_step": 0.306, "alpha": 0.03, "t": 1325.2} |
| {"event": "train", "step": 5100, "phase": 2, "objective": "global", "lm_loss": 1.25342, "bpb": 1.80831, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4688, "lr": 0.0001911, "router_trainable": true, "sec_per_step": 0.242, "alpha": 0.03, "t": 1349.4} |
| {"event": "train", "step": 5200, "phase": 2, "objective": "global", "lm_loss": 1.18315, "bpb": 1.70693, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4777, "lr": 0.0001871, "router_trainable": true, "sec_per_step": 0.238, "alpha": 0.03, "t": 1373.2} |
| {"event": "train", "step": 5300, "phase": 2, "objective": "global", "lm_loss": 1.21961, "bpb": 1.75953, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.5151, "lr": 0.0001831, "router_trainable": true, "sec_per_step": 0.237, "alpha": 0.03, "t": 1396.9} |
| {"event": "train", "step": 5400, "phase": 2, "objective": "global", "lm_loss": 1.27166, "bpb": 1.83462, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.5401, "lr": 0.0001791, "router_trainable": true, "sec_per_step": 0.231, "alpha": 0.03, "t": 1420.0} |
| {"event": "train", "step": 5500, "phase": 2, "objective": "global", "lm_loss": 1.22111, "bpb": 1.76168, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.492, "lr": 0.0001751, "router_trainable": true, "sec_per_step": 0.24, "alpha": 0.03, "t": 1444.0} |
| {"event": "train", "step": 5600, "phase": 2, "objective": "global", "lm_loss": 1.19058, "bpb": 1.71764, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4794, "lr": 0.0001711, "router_trainable": true, "sec_per_step": 0.238, "alpha": 0.03, "t": 1467.8} |
| {"event": "train", "step": 5700, "phase": 2, "objective": "global", "lm_loss": 1.17252, "bpb": 1.69159, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4761, "lr": 0.000167, "router_trainable": true, "sec_per_step": 0.236, "alpha": 0.03, "t": 1491.4} |
| {"event": "train", "step": 5800, "phase": 2, "objective": "global", "lm_loss": 1.16007, "bpb": 1.67362, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4836, "lr": 0.000163, "router_trainable": true, "sec_per_step": 0.236, "alpha": 0.03, "t": 1515.0} |
| {"event": "train", "step": 5900, "phase": 2, "objective": "global", "lm_loss": 1.21907, "bpb": 1.75875, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4931, "lr": 0.0001589, "router_trainable": true, "sec_per_step": 0.237, "alpha": 0.03, "t": 1538.7} |
| {"event": "probe", "step": 6000, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 17418.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 1.000862, "cps": 136.078125, "mean_p": 0.999901, "boundary_entropy": 0.001014, "bpb": 1.884051}, "de": {"chunks": 20862.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.016023, "cps": 162.984375, "mean_p": 0.999901, "boundary_entropy": 0.001016, "bpb": 2.018948}, "vi": {"chunks": 24274.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.316734, "cps": 189.640625, "mean_p": 0.9999, "boundary_entropy": 0.001016, "bpb": 1.706479}, "tr": {"chunks": 19742.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.090537, "cps": 154.234375, "mean_p": 0.9999, "boundary_entropy": 0.001015, "bpb": 1.973675}, "fi": {"chunks": 19572.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 1.040455, "cps": 152.90625, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 2.124111}, "ru": {"chunks": 34560.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.823747, "cps": 270.0, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 1.221707}, "el": {"chunks": 37264.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.815719, "cps": 291.125, "mean_p": 0.9999, "boundary_entropy": 0.001018, "bpb": 1.232517}, "hi": {"chunks": 43173.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.566003, "cps": 337.289062, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 1.014827}, "ta": {"chunks": 50915.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.723455, "cps": 397.773438, "mean_p": 0.999901, "boundary_entropy": 0.001019, "bpb": 0.8649}, "th": {"chunks": 46085.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.865092, "cps": 360.039062, "mean_p": 0.9999, "boundary_entropy": 0.00102, "bpb": 0.981915}, "zh": {"chunks": 16173.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.738402, "cps": 126.351562, "mean_p": 0.999901, "boundary_entropy": 0.001013, "bpb": 2.795642}, "my": {"chunks": 53743.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.836342, "cps": 419.867188, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 0.981378}}, "bpb": {"en": 1.88405, "de": 2.01895, "vi": 1.70648, "tr": 1.97368, "fi": 2.12411, "ru": 1.22171, "el": 1.23252, "hi": 1.01483, "ta": 0.8649, "th": 0.98191, "zh": 2.79564, "my": 0.98138}, "gini_cpc": 0.221516, "gini_cps": 0.233853, "bpic": 1.0, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.01514801064883, "vi": 1.3156005295906583, "tr": 1.0895983329898389, "fi": 1.0395590357977051, "ru": 1.8221761311426539, "el": 1.8141552920914839, "hi": 2.563793186225859, "ta": 2.721110089193083, "th": 2.8626243461390897, "zh": 2.7360433739796948, "my": 2.8338989711748765}, "t": 1564.5} |
| {"event": "eval", "step": 6000, "phase": 2, "bpb": {"en": 1.88727, "de": 2.01459, "vi": 1.71868, "tr": 1.96752, "fi": 2.14037, "ru": 1.22056, "el": 1.2282, "hi": 1.02132, "ta": 0.86689, "th": 0.97916, "zh": 2.80241, "my": 0.99041}, "bytes_per_chunk": {"en": 1024.0, "de": 1024.0, "vi": 1024.0, "tr": 1024.0, "fi": 1024.0, "ru": 1024.0, "el": 1024.0, "hi": 1024.0, "ta": 1024.0, "th": 1024.0, "zh": 1024.0, "my": 1024.0}, "bpb_mean": 1.56978, "t": 1569.5} |
| {"event": "train", "step": 6000, "phase": 2, "objective": "global", "lm_loss": 1.20004, "bpb": 1.73129, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.5616, "lr": 0.0001549, "router_trainable": true, "sec_per_step": 0.311, "alpha": 0.03, "t": 1569.7} |
| {"event": "train", "step": 6100, "phase": 2, "objective": "global", "lm_loss": 1.12673, "bpb": 1.62552, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4752, "lr": 0.0001509, "router_trainable": true, "sec_per_step": 0.237, "alpha": 0.03, "t": 1593.5} |
| {"event": "train", "step": 6200, "phase": 2, "objective": "global", "lm_loss": 1.25757, "bpb": 1.81429, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4796, "lr": 0.0001469, "router_trainable": true, "sec_per_step": 0.236, "alpha": 0.03, "t": 1617.1} |
| {"event": "train", "step": 6300, "phase": 2, "objective": "global", "lm_loss": 1.17824, "bpb": 1.69984, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4845, "lr": 0.0001429, "router_trainable": true, "sec_per_step": 0.236, "alpha": 0.03, "t": 1640.7} |
| {"event": "train", "step": 6400, "phase": 2, "objective": "global", "lm_loss": 1.07137, "bpb": 1.54566, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4545, "lr": 0.0001389, "router_trainable": true, "sec_per_step": 0.24, "alpha": 0.03, "t": 1664.7} |
| {"event": "train", "step": 6500, "phase": 2, "objective": "global", "lm_loss": 1.1377, "bpb": 1.64136, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4744, "lr": 0.000135, "router_trainable": true, "sec_per_step": 0.24, "alpha": 0.03, "t": 1688.7} |
| {"event": "train", "step": 6600, "phase": 2, "objective": "global", "lm_loss": 1.19873, "bpb": 1.7294, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4691, "lr": 0.000131, "router_trainable": true, "sec_per_step": 0.235, "alpha": 0.03, "t": 1712.2} |
| {"event": "train", "step": 6700, "phase": 2, "objective": "global", "lm_loss": 1.17599, "bpb": 1.69659, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4634, "lr": 0.0001271, "router_trainable": true, "sec_per_step": 0.24, "alpha": 0.03, "t": 1736.2} |
| {"event": "train", "step": 6800, "phase": 2, "objective": "global", "lm_loss": 1.19101, "bpb": 1.71827, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4968, "lr": 0.0001233, "router_trainable": true, "sec_per_step": 0.24, "alpha": 0.03, "t": 1760.2} |
| {"event": "train", "step": 6900, "phase": 2, "objective": "global", "lm_loss": 1.166, "bpb": 1.68218, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4833, "lr": 0.0001195, "router_trainable": true, "sec_per_step": 0.239, "alpha": 0.03, "t": 1784.1} |
| {"event": "probe", "step": 7000, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 17418.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 1.000862, "cps": 136.078125, "mean_p": 0.999901, "boundary_entropy": 0.001014, "bpb": 1.845391}, "de": {"chunks": 20862.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.016023, "cps": 162.984375, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 1.981378}, "vi": {"chunks": 24274.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.316734, "cps": 189.640625, "mean_p": 0.9999, "boundary_entropy": 0.001016, "bpb": 1.66989}, "tr": {"chunks": 19742.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.090537, "cps": 154.234375, "mean_p": 0.9999, "boundary_entropy": 0.001015, "bpb": 1.917563}, "fi": {"chunks": 19572.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 1.040455, "cps": 152.90625, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 2.075381}, "ru": {"chunks": 34560.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.823747, "cps": 270.0, "mean_p": 0.999901, "boundary_entropy": 0.001018, "bpb": 1.202951}, "el": {"chunks": 37264.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.815719, "cps": 291.125, "mean_p": 0.9999, "boundary_entropy": 0.001018, "bpb": 1.199903}, "hi": {"chunks": 43173.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.566003, "cps": 337.289062, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 0.991189}, "ta": {"chunks": 50915.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.723455, "cps": 397.773438, "mean_p": 0.999901, "boundary_entropy": 0.001019, "bpb": 0.84611}, "th": {"chunks": 46085.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.865092, "cps": 360.039062, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 0.957466}, "zh": {"chunks": 16173.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.738402, "cps": 126.351562, "mean_p": 0.999901, "boundary_entropy": 0.001013, "bpb": 2.751167}, "my": {"chunks": 53743.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.836342, "cps": 419.867188, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 0.951565}}, "bpb": {"en": 1.84539, "de": 1.98138, "vi": 1.66989, "tr": 1.91756, "fi": 2.07538, "ru": 1.20295, "el": 1.1999, "hi": 0.99119, "ta": 0.84611, "th": 0.95747, "zh": 2.75117, "my": 0.95156}, "gini_cpc": 0.221516, "gini_cps": 0.233853, "bpic": 1.0, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.01514801064883, "vi": 1.3156005295906583, "tr": 1.0895983329898389, "fi": 1.0395590357977051, "ru": 1.8221761311426539, "el": 1.8141552920914839, "hi": 2.563793186225859, "ta": 2.721110089193083, "th": 2.8626243461390897, "zh": 2.7360433739796948, "my": 2.8338989711748765}, "t": 1809.9} |
| {"event": "eval", "step": 7000, "phase": 2, "bpb": {"en": 1.84741, "de": 1.97684, "vi": 1.68107, "tr": 1.91469, "fi": 2.09131, "ru": 1.20124, "el": 1.19662, "hi": 0.99777, "ta": 0.84877, "th": 0.95537, "zh": 2.75869, "my": 0.95935}, "bytes_per_chunk": {"en": 1024.0, "de": 1024.0, "vi": 1024.0, "tr": 1024.0, "fi": 1024.0, "ru": 1024.0, "el": 1024.0, "hi": 1024.0, "ta": 1024.0, "th": 1024.0, "zh": 1024.0, "my": 1024.0}, "bpb_mean": 1.53576, "t": 1814.7} |
| {"event": "train", "step": 7000, "phase": 2, "objective": "global", "lm_loss": 1.16234, "bpb": 1.6769, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4888, "lr": 0.0001157, "router_trainable": true, "sec_per_step": 0.308, "alpha": 0.03, "t": 1814.9} |
| {"event": "train", "step": 7100, "phase": 2, "objective": "global", "lm_loss": 1.15019, "bpb": 1.65937, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.475, "lr": 0.0001119, "router_trainable": true, "sec_per_step": 0.24, "alpha": 0.03, "t": 1838.9} |
| {"event": "train", "step": 7200, "phase": 2, "objective": "global", "lm_loss": 1.19507, "bpb": 1.72412, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4705, "lr": 0.0001083, "router_trainable": true, "sec_per_step": 0.239, "alpha": 0.03, "t": 1862.8} |
| {"event": "train", "step": 7300, "phase": 2, "objective": "global", "lm_loss": 1.2029, "bpb": 1.73542, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4526, "lr": 0.0001046, "router_trainable": true, "sec_per_step": 0.236, "alpha": 0.03, "t": 1886.4} |
| {"event": "train", "step": 7400, "phase": 2, "objective": "global", "lm_loss": 1.17458, "bpb": 1.69456, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4675, "lr": 0.000101, "router_trainable": true, "sec_per_step": 0.242, "alpha": 0.03, "t": 1910.5} |
| {"event": "train", "step": 7500, "phase": 2, "objective": "global", "lm_loss": 1.14885, "bpb": 1.65745, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4556, "lr": 9.75e-05, "router_trainable": true, "sec_per_step": 0.241, "alpha": 0.03, "t": 1934.6} |
| {"event": "train", "step": 7600, "phase": 2, "objective": "global", "lm_loss": 1.20732, "bpb": 1.7418, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4558, "lr": 9.4e-05, "router_trainable": true, "sec_per_step": 0.232, "alpha": 0.03, "t": 1957.9} |
| {"event": "phase_transition", "step": 7700, "phase": 3, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0, "t": 1981.2} |
| {"event": "train", "step": 7700, "phase": 3, "objective": "global", "lm_loss": 1.16647, "bpb": 1.68286, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4714, "lr": 9.06e-05, "router_trainable": true, "sec_per_step": 0.236, "alpha": 0.03, "t": 1981.4} |
| {"event": "train", "step": 7800, "phase": 3, "objective": "global", "lm_loss": 1.20793, "bpb": 1.74267, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4966, "lr": 8.73e-05, "router_trainable": true, "sec_per_step": 0.235, "alpha": 0.03, "t": 2004.9} |
| {"event": "train", "step": 7900, "phase": 3, "objective": "global", "lm_loss": 1.18618, "bpb": 1.71129, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4845, "lr": 8.4e-05, "router_trainable": true, "sec_per_step": 0.229, "alpha": 0.03, "t": 2027.8} |
| {"event": "probe", "step": 8000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 17418.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 1.000862, "cps": 136.078125, "mean_p": 0.999901, "boundary_entropy": 0.001014, "bpb": 1.819822}, "de": {"chunks": 20862.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.016023, "cps": 162.984375, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 1.950174}, "vi": {"chunks": 24274.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.316734, "cps": 189.640625, "mean_p": 0.9999, "boundary_entropy": 0.001016, "bpb": 1.637049}, "tr": {"chunks": 19742.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.090537, "cps": 154.234375, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 1.877595}, "fi": {"chunks": 19572.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 1.040455, "cps": 152.90625, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 2.02913}, "ru": {"chunks": 34560.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.823747, "cps": 270.0, "mean_p": 0.999901, "boundary_entropy": 0.001018, "bpb": 1.18312}, "el": {"chunks": 37264.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.815719, "cps": 291.125, "mean_p": 0.999901, "boundary_entropy": 0.001018, "bpb": 1.179492}, "hi": {"chunks": 43173.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.566003, "cps": 337.289062, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 0.977378}, "ta": {"chunks": 50915.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.723455, "cps": 397.773438, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 0.828535}, "th": {"chunks": 46085.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.865092, "cps": 360.039062, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 0.943273}, "zh": {"chunks": 16173.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.738402, "cps": 126.351562, "mean_p": 0.999901, "boundary_entropy": 0.001013, "bpb": 2.71544}, "my": {"chunks": 53743.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.836342, "cps": 419.867188, "mean_p": 0.999901, "boundary_entropy": 0.001019, "bpb": 0.933339}}, "bpb": {"en": 1.81982, "de": 1.95017, "vi": 1.63705, "tr": 1.87759, "fi": 2.02913, "ru": 1.18312, "el": 1.17949, "hi": 0.97738, "ta": 0.82854, "th": 0.94327, "zh": 2.71544, "my": 0.93334}, "gini_cpc": 0.221516, "gini_cps": 0.233853, "bpic": 1.0, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.01514801064883, "vi": 1.3156005295906583, "tr": 1.0895983329898389, "fi": 1.0395590357977051, "ru": 1.8221761311426539, "el": 1.8141552920914839, "hi": 2.563793186225859, "ta": 2.721110089193083, "th": 2.8626243461390897, "zh": 2.7360433739796948, "my": 2.8338989711748765}, "t": 2053.4} |
| {"event": "eval", "step": 8000, "phase": 3, "bpb": {"en": 1.82141, "de": 1.94602, "vi": 1.64959, "tr": 1.87469, "fi": 2.04417, "ru": 1.18174, "el": 1.17741, "hi": 0.98352, "ta": 0.83205, "th": 0.94166, "zh": 2.72448, "my": 0.94113}, "bytes_per_chunk": {"en": 1024.0, "de": 1024.0, "vi": 1024.0, "tr": 1024.0, "fi": 1024.0, "ru": 1024.0, "el": 1024.0, "hi": 1024.0, "ta": 1024.0, "th": 1024.0, "zh": 1024.0, "my": 1024.0}, "bpb_mean": 1.50982, "t": 2057.9} |
| {"event": "train", "step": 8000, "phase": 3, "objective": "global", "lm_loss": 1.1844, "bpb": 1.70872, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4904, "lr": 8.08e-05, "router_trainable": true, "sec_per_step": 0.303, "alpha": 0.03, "t": 2058.1} |
| {"event": "train", "step": 8100, "phase": 3, "objective": "global", "lm_loss": 1.21157, "bpb": 1.74792, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.5083, "lr": 7.77e-05, "router_trainable": true, "sec_per_step": 0.233, "alpha": 0.03, "t": 2081.4} |
| {"event": "train", "step": 8200, "phase": 3, "objective": "global", "lm_loss": 1.16393, "bpb": 1.67919, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.5, "lr": 7.47e-05, "router_trainable": true, "sec_per_step": 0.237, "alpha": 0.03, "t": 2105.2} |
| {"event": "train", "step": 8300, "phase": 3, "objective": "global", "lm_loss": 1.08994, "bpb": 1.57245, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4984, "lr": 7.17e-05, "router_trainable": true, "sec_per_step": 0.227, "alpha": 0.03, "t": 2127.9} |
| {"event": "train", "step": 8400, "phase": 3, "objective": "global", "lm_loss": 1.15596, "bpb": 1.66769, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4956, "lr": 6.88e-05, "router_trainable": true, "sec_per_step": 0.231, "alpha": 0.03, "t": 2150.9} |
| {"event": "train", "step": 8500, "phase": 3, "objective": "global", "lm_loss": 1.14996, "bpb": 1.65904, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4702, "lr": 6.6e-05, "router_trainable": true, "sec_per_step": 0.233, "alpha": 0.03, "t": 2174.2} |
| {"event": "train", "step": 8600, "phase": 3, "objective": "global", "lm_loss": 1.21204, "bpb": 1.7486, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.507, "lr": 6.33e-05, "router_trainable": true, "sec_per_step": 0.224, "alpha": 0.03, "t": 2196.6} |
| {"event": "train", "step": 8700, "phase": 3, "objective": "global", "lm_loss": 1.08253, "bpb": 1.56176, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4979, "lr": 6.07e-05, "router_trainable": true, "sec_per_step": 0.231, "alpha": 0.03, "t": 2219.7} |
| {"event": "train", "step": 8800, "phase": 3, "objective": "global", "lm_loss": 1.10502, "bpb": 1.5942, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4945, "lr": 5.82e-05, "router_trainable": true, "sec_per_step": 0.242, "alpha": 0.03, "t": 2243.9} |
| {"event": "train", "step": 8900, "phase": 3, "objective": "global", "lm_loss": 1.16653, "bpb": 1.68295, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4844, "lr": 5.58e-05, "router_trainable": true, "sec_per_step": 0.237, "alpha": 0.03, "t": 2267.6} |
| {"event": "probe", "step": 9000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 17418.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 1.000862, "cps": 136.078125, "mean_p": 0.999901, "boundary_entropy": 0.001014, "bpb": 1.803214}, "de": {"chunks": 20862.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.016023, "cps": 162.984375, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 1.926172}, "vi": {"chunks": 24274.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.316734, "cps": 189.640625, "mean_p": 0.9999, "boundary_entropy": 0.001016, "bpb": 1.615381}, "tr": {"chunks": 19742.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.090537, "cps": 154.234375, "mean_p": 0.9999, "boundary_entropy": 0.001015, "bpb": 1.851526}, "fi": {"chunks": 19572.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 1.040455, "cps": 152.90625, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 2.000323}, "ru": {"chunks": 34560.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.823747, "cps": 270.0, "mean_p": 0.999901, "boundary_entropy": 0.001018, "bpb": 1.169801}, "el": {"chunks": 37264.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.815719, "cps": 291.125, "mean_p": 0.9999, "boundary_entropy": 0.001018, "bpb": 1.161052}, "hi": {"chunks": 43173.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.566003, "cps": 337.289062, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 0.962615}, "ta": {"chunks": 50915.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.723455, "cps": 397.773438, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 0.817485}, "th": {"chunks": 46085.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.865092, "cps": 360.039062, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 0.927932}, "zh": {"chunks": 16173.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.738402, "cps": 126.351562, "mean_p": 0.999901, "boundary_entropy": 0.001013, "bpb": 2.691126}, "my": {"chunks": 53743.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.836342, "cps": 419.867188, "mean_p": 0.999901, "boundary_entropy": 0.001019, "bpb": 0.915485}}, "bpb": {"en": 1.80321, "de": 1.92617, "vi": 1.61538, "tr": 1.85153, "fi": 2.00032, "ru": 1.1698, "el": 1.16105, "hi": 0.96261, "ta": 0.81748, "th": 0.92793, "zh": 2.69113, "my": 0.91548}, "gini_cpc": 0.221516, "gini_cps": 0.233853, "bpic": 1.0, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.01514801064883, "vi": 1.3156005295906583, "tr": 1.0895983329898389, "fi": 1.0395590357977051, "ru": 1.8221761311426539, "el": 1.8141552920914839, "hi": 2.563793186225859, "ta": 2.721110089193083, "th": 2.8626243461390897, "zh": 2.7360433739796948, "my": 2.8338989711748765}, "t": 2293.7} |
| {"event": "eval", "step": 9000, "phase": 3, "bpb": {"en": 1.80452, "de": 1.92313, "vi": 1.62763, "tr": 1.84866, "fi": 2.01526, "ru": 1.16795, "el": 1.15862, "hi": 0.96827, "ta": 0.82108, "th": 0.92653, "zh": 2.70015, "my": 0.9243}, "bytes_per_chunk": {"en": 1024.0, "de": 1024.0, "vi": 1024.0, "tr": 1024.0, "fi": 1024.0, "ru": 1024.0, "el": 1024.0, "hi": 1024.0, "ta": 1024.0, "th": 1024.0, "zh": 1024.0, "my": 1024.0}, "bpb_mean": 1.49051, "t": 2298.1} |
| {"event": "train", "step": 9000, "phase": 3, "objective": "global", "lm_loss": 1.16213, "bpb": 1.67659, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.5123, "lr": 5.35e-05, "router_trainable": true, "sec_per_step": 0.308, "alpha": 0.03, "t": 2298.3} |
| {"event": "train", "step": 9100, "phase": 3, "objective": "global", "lm_loss": 1.13667, "bpb": 1.63986, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4946, "lr": 5.12e-05, "router_trainable": true, "sec_per_step": 0.236, "alpha": 0.03, "t": 2321.9} |
| {"event": "train", "step": 9200, "phase": 3, "objective": "global", "lm_loss": 1.0543, "bpb": 1.52103, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4799, "lr": 4.91e-05, "router_trainable": true, "sec_per_step": 0.239, "alpha": 0.03, "t": 2345.8} |
| {"event": "train", "step": 9300, "phase": 3, "objective": "global", "lm_loss": 1.13698, "bpb": 1.64031, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4806, "lr": 4.71e-05, "router_trainable": true, "sec_per_step": 0.234, "alpha": 0.03, "t": 2369.2} |
| {"event": "train", "step": 9400, "phase": 3, "objective": "global", "lm_loss": 1.18973, "bpb": 1.71642, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4954, "lr": 4.52e-05, "router_trainable": true, "sec_per_step": 0.235, "alpha": 0.03, "t": 2392.7} |
| {"event": "train", "step": 9500, "phase": 3, "objective": "global", "lm_loss": 1.20131, "bpb": 1.73313, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.5134, "lr": 4.34e-05, "router_trainable": true, "sec_per_step": 0.237, "alpha": 0.03, "t": 2416.3} |
| {"event": "train", "step": 9600, "phase": 3, "objective": "global", "lm_loss": 1.053, "bpb": 1.51916, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.499, "lr": 4.17e-05, "router_trainable": true, "sec_per_step": 0.247, "alpha": 0.03, "t": 2441.1} |
| {"event": "train", "step": 9700, "phase": 3, "objective": "global", "lm_loss": 1.07267, "bpb": 1.54754, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4835, "lr": 4.01e-05, "router_trainable": true, "sec_per_step": 0.236, "alpha": 0.03, "t": 2464.7} |
| {"event": "train", "step": 9800, "phase": 3, "objective": "global", "lm_loss": 1.08249, "bpb": 1.56171, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.487, "lr": 3.86e-05, "router_trainable": true, "sec_per_step": 0.246, "alpha": 0.03, "t": 2489.3} |
| {"event": "train", "step": 9900, "phase": 3, "objective": "global", "lm_loss": 1.15759, "bpb": 1.67004, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.5092, "lr": 3.72e-05, "router_trainable": true, "sec_per_step": 0.235, "alpha": 0.03, "t": 2512.8} |
| {"event": "probe", "step": 10000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 17418.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 1.000862, "cps": 136.078125, "mean_p": 0.999901, "boundary_entropy": 0.001014, "bpb": 1.791934}, "de": {"chunks": 20862.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.016023, "cps": 162.984375, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 1.910883}, "vi": {"chunks": 24274.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.316734, "cps": 189.640625, "mean_p": 0.9999, "boundary_entropy": 0.001016, "bpb": 1.601951}, "tr": {"chunks": 19742.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.090537, "cps": 154.234375, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 1.833735}, "fi": {"chunks": 19572.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 1.040455, "cps": 152.90625, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 1.981723}, "ru": {"chunks": 34560.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.823747, "cps": 270.0, "mean_p": 0.999901, "boundary_entropy": 0.001018, "bpb": 1.160031}, "el": {"chunks": 37264.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.815719, "cps": 291.125, "mean_p": 0.9999, "boundary_entropy": 0.001018, "bpb": 1.149183}, "hi": {"chunks": 43173.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.566003, "cps": 337.289062, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 0.950307}, "ta": {"chunks": 50915.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.723455, "cps": 397.773438, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 0.807941}, "th": {"chunks": 46085.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.865092, "cps": 360.039062, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 0.920434}, "zh": {"chunks": 16173.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.738402, "cps": 126.351562, "mean_p": 0.999901, "boundary_entropy": 0.001013, "bpb": 2.67441}, "my": {"chunks": 53743.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.836342, "cps": 419.867188, "mean_p": 0.999901, "boundary_entropy": 0.001019, "bpb": 0.907067}}, "bpb": {"en": 1.79193, "de": 1.91088, "vi": 1.60195, "tr": 1.83373, "fi": 1.98172, "ru": 1.16003, "el": 1.14918, "hi": 0.95031, "ta": 0.80794, "th": 0.92043, "zh": 2.67441, "my": 0.90707}, "gini_cpc": 0.221516, "gini_cps": 0.233853, "bpic": 1.0, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.01514801064883, "vi": 1.3156005295906583, "tr": 1.0895983329898389, "fi": 1.0395590357977051, "ru": 1.8221761311426539, "el": 1.8141552920914839, "hi": 2.563793186225859, "ta": 2.721110089193083, "th": 2.8626243461390897, "zh": 2.7360433739796948, "my": 2.8338989711748765}, "t": 2539.5} |
| {"event": "eval", "step": 10000, "phase": 3, "bpb": {"en": 1.79249, "de": 1.90842, "vi": 1.61462, "tr": 1.83133, "fi": 1.99723, "ru": 1.15819, "el": 1.14677, "hi": 0.95624, "ta": 0.81137, "th": 0.91822, "zh": 2.68515, "my": 0.91651}, "bytes_per_chunk": {"en": 1024.0, "de": 1024.0, "vi": 1024.0, "tr": 1024.0, "fi": 1024.0, "ru": 1024.0, "el": 1024.0, "hi": 1024.0, "ta": 1024.0, "th": 1024.0, "zh": 1024.0, "my": 1024.0}, "bpb_mean": 1.47805, "t": 2544.0} |
| {"event": "train", "step": 10000, "phase": 3, "objective": "global", "lm_loss": 1.14896, "bpb": 1.6576, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.5341, "lr": 3.6e-05, "router_trainable": true, "sec_per_step": 0.315, "alpha": 0.03, "t": 2544.3} |
| {"event": "train", "step": 10100, "phase": 3, "objective": "global", "lm_loss": 1.06735, "bpb": 1.53986, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4846, "lr": 3.49e-05, "router_trainable": true, "sec_per_step": 0.246, "alpha": 0.03, "t": 2568.8} |
| {"event": "train", "step": 10200, "phase": 3, "objective": "global", "lm_loss": 1.17376, "bpb": 1.69337, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.5231, "lr": 3.38e-05, "router_trainable": true, "sec_per_step": 0.241, "alpha": 0.03, "t": 2593.0} |
| {"event": "train", "step": 10300, "phase": 3, "objective": "global", "lm_loss": 1.14329, "bpb": 1.64942, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.5375, "lr": 3.3e-05, "router_trainable": true, "sec_per_step": 0.24, "alpha": 0.03, "t": 2617.0} |
| {"event": "train", "step": 10400, "phase": 3, "objective": "global", "lm_loss": 1.09251, "bpb": 1.57616, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.506, "lr": 3.22e-05, "router_trainable": true, "sec_per_step": 0.247, "alpha": 0.03, "t": 2641.7} |
| {"event": "train", "step": 10500, "phase": 3, "objective": "global", "lm_loss": 1.12237, "bpb": 1.61924, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.5133, "lr": 3.15e-05, "router_trainable": true, "sec_per_step": 0.238, "alpha": 0.03, "t": 2665.4} |
| {"event": "train", "step": 10600, "phase": 3, "objective": "global", "lm_loss": 1.14681, "bpb": 1.6545, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.517, "lr": 3.1e-05, "router_trainable": true, "sec_per_step": 0.246, "alpha": 0.03, "t": 2690.0} |
| {"event": "train", "step": 10700, "phase": 3, "objective": "global", "lm_loss": 1.11016, "bpb": 1.60162, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.4902, "lr": 3.05e-05, "router_trainable": true, "sec_per_step": 0.236, "alpha": 0.03, "t": 2713.6} |
| {"event": "train", "step": 10800, "phase": 3, "objective": "global", "lm_loss": 1.1282, "bpb": 1.62765, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.5041, "lr": 3.02e-05, "router_trainable": true, "sec_per_step": 0.234, "alpha": 0.03, "t": 2737.1} |
| {"event": "train", "step": 10900, "phase": 3, "objective": "global", "lm_loss": 1.11302, "bpb": 1.60575, "chunk_loss": 0.000179, "F": 0.00098, "G": 0.9999, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.00102, "router_noise": 0.0, "grad_norm": 0.5254, "lr": 3.01e-05, "router_trainable": true, "sec_per_step": 0.237, "alpha": 0.03, "t": 2760.7} |
| {"event": "probe", "step": 11000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 17418.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 1.000862, "cps": 136.078125, "mean_p": 0.999901, "boundary_entropy": 0.001014, "bpb": 1.784655}, "de": {"chunks": 20862.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.016023, "cps": 162.984375, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 1.901134}, "vi": {"chunks": 24274.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.316734, "cps": 189.640625, "mean_p": 0.9999, "boundary_entropy": 0.001016, "bpb": 1.589596}, "tr": {"chunks": 19742.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.090537, "cps": 154.234375, "mean_p": 0.9999, "boundary_entropy": 0.001015, "bpb": 1.824606}, "fi": {"chunks": 19572.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 1.040455, "cps": 152.90625, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 1.972199}, "ru": {"chunks": 34560.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.823747, "cps": 270.0, "mean_p": 0.999901, "boundary_entropy": 0.001018, "bpb": 1.153459}, "el": {"chunks": 37264.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.815719, "cps": 291.125, "mean_p": 0.9999, "boundary_entropy": 0.001018, "bpb": 1.142651}, "hi": {"chunks": 43173.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.566003, "cps": 337.289062, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 0.94471}, "ta": {"chunks": 50915.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.723455, "cps": 397.773438, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 0.804868}, "th": {"chunks": 46085.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.865092, "cps": 360.039062, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 0.91481}, "zh": {"chunks": 16173.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.738402, "cps": 126.351562, "mean_p": 0.999901, "boundary_entropy": 0.001013, "bpb": 2.665541}, "my": {"chunks": 53743.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.836342, "cps": 419.867188, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 0.898907}}, "bpb": {"en": 1.78465, "de": 1.90113, "vi": 1.5896, "tr": 1.82461, "fi": 1.9722, "ru": 1.15346, "el": 1.14265, "hi": 0.94471, "ta": 0.80487, "th": 0.91481, "zh": 2.66554, "my": 0.89891}, "gini_cpc": 0.221516, "gini_cps": 0.233853, "bpic": 1.0, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.01514801064883, "vi": 1.3156005295906583, "tr": 1.0895983329898389, "fi": 1.0395590357977051, "ru": 1.8221761311426539, "el": 1.8141552920914839, "hi": 2.563793186225859, "ta": 2.721110089193083, "th": 2.8626243461390897, "zh": 2.7360433739796948, "my": 2.8338989711748765}, "t": 2786.6} |
| {"event": "eval", "step": 11000, "phase": 3, "bpb": {"en": 1.78559, "de": 1.89903, "vi": 1.60366, "tr": 1.82273, "fi": 1.98657, "ru": 1.15196, "el": 1.14023, "hi": 0.95144, "ta": 0.80805, "th": 0.91293, "zh": 2.67545, "my": 0.90803}, "bytes_per_chunk": {"en": 1024.0, "de": 1024.0, "vi": 1024.0, "tr": 1024.0, "fi": 1024.0, "ru": 1024.0, "el": 1024.0, "hi": 1024.0, "ta": 1024.0, "th": 1024.0, "zh": 1024.0, "my": 1024.0}, "bpb_mean": 1.47047, "t": 2791.1} |
| {"event": "probe", "step": 11000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 17418.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 1.000862, "cps": 136.078125, "mean_p": 0.999901, "boundary_entropy": 0.001014, "bpb": 1.784655}, "de": {"chunks": 20862.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 1.016023, "cps": 162.984375, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 1.901134}, "vi": {"chunks": 24274.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 1.316734, "cps": 189.640625, "mean_p": 0.9999, "boundary_entropy": 0.001016, "bpb": 1.589596}, "tr": {"chunks": 19742.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 1.090537, "cps": 154.234375, "mean_p": 0.9999, "boundary_entropy": 0.001015, "bpb": 1.824606}, "fi": {"chunks": 19572.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 1.040455, "cps": 152.90625, "mean_p": 0.999901, "boundary_entropy": 0.001015, "bpb": 1.972199}, "ru": {"chunks": 34560.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.823747, "cps": 270.0, "mean_p": 0.999901, "boundary_entropy": 0.001018, "bpb": 1.153459}, "el": {"chunks": 37264.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.815719, "cps": 291.125, "mean_p": 0.9999, "boundary_entropy": 0.001018, "bpb": 1.142651}, "hi": {"chunks": 43173.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.566003, "cps": 337.289062, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 0.94471}, "ta": {"chunks": 50915.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.723455, "cps": 397.773438, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 0.804868}, "th": {"chunks": 46085.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.865092, "cps": 360.039062, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 0.91481}, "zh": {"chunks": 16173.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 2.738402, "cps": 126.351562, "mean_p": 0.999901, "boundary_entropy": 0.001013, "bpb": 2.665541}, "my": {"chunks": 53743.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.836342, "cps": 419.867188, "mean_p": 0.9999, "boundary_entropy": 0.001019, "bpb": 0.898907}}, "bpb": {"en": 1.78465, "de": 1.90113, "vi": 1.5896, "tr": 1.82461, "fi": 1.9722, "ru": 1.15346, "el": 1.14265, "hi": 0.94471, "ta": 0.80487, "th": 0.91481, "zh": 2.66554, "my": 0.89891}, "gini_cpc": 0.221516, "gini_cps": 0.233853, "bpic": 1.0, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.01514801064883, "vi": 1.3156005295906583, "tr": 1.0895983329898389, "fi": 1.0395590357977051, "ru": 1.8221761311426539, "el": 1.8141552920914839, "hi": 2.563793186225859, "ta": 2.721110089193083, "th": 2.8626243461390897, "zh": 2.7360433739796948, "my": 2.8338989711748765}, "t": 2793.5} |
| {"event": "reps_dumped", "dirs": ["/local/scratch/sas245/beetle-hnet/scratch/reps/exp32_local_onechunk_s0/th0.30", "/local/scratch/sas245/beetle-hnet/scratch/reps/exp32_local_onechunk_s0/th0.40", "/local/scratch/sas245/beetle-hnet/scratch/reps/exp32_local_onechunk_s0/th0.50", "/local/scratch/sas245/beetle-hnet/scratch/reps/exp32_local_onechunk_s0/th0.60", "/local/scratch/sas245/beetle-hnet/scratch/reps/exp32_local_onechunk_s0/th0.70"], "t": 2808.9} |
| {"event": "final", "run": "exp32_local_onechunk_s0", "condition": "A", "step": 11000, "params": 82470528, "per_lang": {"en": {"cpc": 1.000861920358559, "cps": 136.078125, "bpb": 1.7803248457220966, "bytes_per_chunk": 1024.0}, "de": {"cpc": 1.0160229873861588, "cps": 162.984375, "bpb": 1.900838968838543, "bytes_per_chunk": 1024.0}, "vi": {"cpc": 1.3167344724708434, "cps": 189.640625, "bpb": 1.6069871865488026, "bytes_per_chunk": 1024.0}, "tr": {"cpc": 1.0905374799756946, "cps": 154.234375, "bpb": 1.821724971641166, "bytes_per_chunk": 1024.0}, "fi": {"cpc": 1.040455052894583, "cps": 152.90625, "bpb": 1.9921167937324094, "bytes_per_chunk": 1024.0}, "ru": {"cpc": 1.8237467018469657, "cps": 270.0, "bpb": 1.1518317205063986, "bytes_per_chunk": 1024.0}, "el": {"cpc": 1.8157189494713248, "cps": 291.125, "bpb": 1.1372893136294402, "bytes_per_chunk": 1024.0}, "hi": {"cpc": 2.566002971768202, "cps": 337.2890625, "bpb": 0.9531695622489775, "bytes_per_chunk": 1024.0}, "ta": {"cpc": 2.7234554693768387, "cps": 397.7734375, "bpb": 0.8069287371215301, "bytes_per_chunk": 1024.0}, "th": {"cpc": 2.8650917003419334, "cps": 360.0390625, "bpb": 0.9149707267675262, "bytes_per_chunk": 1024.0}, "zh": {"cpc": 2.7384016254656283, "cps": 126.3515625, "bpb": 2.6761373931141095, "bytes_per_chunk": 1024.0}, "my": {"cpc": 2.8363415663922313, "cps": 419.8671875, "bpb": 0.9065022526689209, "bytes_per_chunk": 1024.0}}, "gini_cpc": 0.22151585453568945, "gini_cps": 0.2338532044751218, "bpic": 1.0, "seed": 0, "mean_boundary_entropy": 0.0010205263970419765, "dechunk": "ema", "router_noise": 0.0, "transitions": [{"step": 0, "phase": 1, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0}, {"step": 4400, "phase": 2, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0}, {"step": 7700, "phase": 3, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0}], "t": 2817.2} |
|
|
