| {"event": "start", "run": "spec_A_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": 2.1} |
| {"event": "probe_schedule", "n": 21, "steps": [0, 1, 2, 4, 8, 16, 32, 64, 128, 256, 1000, 2000, 3000, 4000, 5000, 6000, 7000, 8000, 9000, 10000, 11000], "t": 2.1} |
| {"event": "probe", "step": 0, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 8739.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.502155, "cps": 68.273438, "mean_p": 0.50363, "boundary_entropy": 0.687415, "bpb": 8.248358}, "de": {"chunks": 12222.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.595237, "cps": 95.484375, "mean_p": 0.507356, "boundary_entropy": 0.688163, "bpb": 8.204506}, "vi": {"chunks": 6386.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.346406, "cps": 49.890625, "mean_p": 0.491233, "boundary_entropy": 0.688507, "bpb": 8.13873}, "tr": {"chunks": 11403.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.629896, "cps": 89.085938, "mean_p": 0.50659, "boundary_entropy": 0.687851, "bpb": 8.143105}, "fi": {"chunks": 7137.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.379406, "cps": 55.757812, "mean_p": 0.497164, "boundary_entropy": 0.68792, "bpb": 8.234202}, "ru": {"chunks": 10473.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.552665, "cps": 81.820312, "mean_p": 0.496692, "boundary_entropy": 0.690192, "bpb": 8.071624}, "el": {"chunks": 36555.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.781172, "cps": 285.585938, "mean_p": 0.530217, "boundary_entropy": 0.688772, "bpb": 8.142597}, "hi": {"chunks": 42584.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.530996, "cps": 332.6875, "mean_p": 0.521073, "boundary_entropy": 0.69018, "bpb": 7.880816}, "ta": {"chunks": 49758.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.661567, "cps": 388.734375, "mean_p": 0.521091, "boundary_entropy": 0.690409, "bpb": 7.814466}, "th": {"chunks": 45716.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.842151, "cps": 357.15625, "mean_p": 0.525713, "boundary_entropy": 0.689893, "bpb": 8.063302}, "zh": {"chunks": 6218.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.052828, "cps": 48.578125, "mean_p": 0.496822, "boundary_entropy": 0.686552, "bpb": 8.034403}, "my": {"chunks": 53266.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.811167, "cps": 416.140625, "mean_p": 0.521325, "boundary_entropy": 0.690575, "bpb": 8.24483}}, "bpb": {"en": 8.24836, "de": 8.20451, "vi": 8.13873, "tr": 8.1431, "fi": 8.2342, "ru": 8.07162, "el": 8.1426, "hi": 7.88082, "ta": 7.81447, "th": 8.0633, "zh": 8.0344, "my": 8.24483}, "gini_cpc": 0.391191, "gini_cps": 0.409188, "bpic": 1.321301, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1853654183968407, "vi": 0.6898396505624507, "tr": 1.2543852937189204, "fi": 0.755555191784985, "ru": 1.1005867247811874, "el": 3.54705827773647, "hi": 5.040269530092341, "ta": 5.300292378412309, "th": 5.659910185771798, "zh": 2.096619669942964, "my": 5.598208760550946}, "t": 9.0} |
| {"event": "train", "step": 0, "phase": 1, "objective": "global", "lm_loss": 5.63905, "bpb": 8.13544, "chunk_loss": 0.057711, "F": 0.53552, "G": 0.50248, "chunks": 548.38, "bytes_per_chunk": 1.867, "boundary_entropy": 0.69172, "router_noise": 0.0, "grad_norm": 9.284, "lr": 6e-07, "router_trainable": true, "sec_per_step": 0.005, "alpha": 0.03, "t": 9.7} |
| {"event": "probe", "step": 1, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 8723.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.501235, "cps": 68.148438, "mean_p": 0.503564, "boundary_entropy": 0.687414, "bpb": 8.209973}, "de": {"chunks": 12185.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.593435, "cps": 95.195312, "mean_p": 0.507261, "boundary_entropy": 0.688164, "bpb": 8.167568}, "vi": {"chunks": 6365.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.345267, "cps": 49.726562, "mean_p": 0.491183, "boundary_entropy": 0.688503, "bpb": 8.119032}, "tr": {"chunks": 11365.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.627796, "cps": 88.789062, "mean_p": 0.506498, "boundary_entropy": 0.687851, "bpb": 8.116521}, "fi": {"chunks": 7084.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.376588, "cps": 55.34375, "mean_p": 0.497024, "boundary_entropy": 0.687916, "bpb": 8.205207}, "ru": {"chunks": 10505.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.554354, "cps": 82.070312, "mean_p": 0.496728, "boundary_entropy": 0.690193, "bpb": 8.028184}, "el": {"chunks": 36545.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.780685, "cps": 285.507812, "mean_p": 0.53005, "boundary_entropy": 0.688791, "bpb": 8.124604}, "hi": {"chunks": 42564.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.529807, "cps": 332.53125, "mean_p": 0.520975, "boundary_entropy": 0.690188, "bpb": 7.857673}, "ta": {"chunks": 49697.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.658304, "cps": 388.257812, "mean_p": 0.520855, "boundary_entropy": 0.690427, "bpb": 7.780817}, "th": {"chunks": 45715.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.842089, "cps": 357.148438, "mean_p": 0.525585, "boundary_entropy": 0.689906, "bpb": 8.031123}, "zh": {"chunks": 6210.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.051473, "cps": 48.515625, "mean_p": 0.496796, "boundary_entropy": 0.686551, "bpb": 8.029634}, "my": {"chunks": 53258.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.810745, "cps": 416.078125, "mean_p": 0.521208, "boundary_entropy": 0.690584, "bpb": 8.216903}}, "bpb": {"en": 8.20997, "de": 8.16757, "vi": 8.11903, "tr": 8.11652, "fi": 8.20521, "ru": 8.02818, "el": 8.1246, "hi": 7.85767, "ta": 7.78082, "th": 8.03112, "zh": 8.02963, "my": 8.2169}, "gini_cpc": 0.391553, "gini_cps": 0.409488, "bpic": 1.322398, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1839445810310785, "vi": 0.6888323164250548, "tr": 1.2524982748512408, "fi": 0.7513199428602956, "ru": 1.1059744399378468, "el": 3.5525922887443713, "hi": 5.047142995605715, "ta": 5.303504616254146, "th": 5.670167737448628, "zh": 2.0977629233473527, "my": 5.607634835497459}, "t": 15.9} |
| {"event": "probe", "step": 2, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 8646.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.496811, "cps": 67.546875, "mean_p": 0.503388, "boundary_entropy": 0.687412, "bpb": 8.133935}, "de": {"chunks": 12055.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.587104, "cps": 94.179688, "mean_p": 0.507026, "boundary_entropy": 0.688166, "bpb": 8.095658}, "vi": {"chunks": 6329.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.343314, "cps": 49.445312, "mean_p": 0.491077, "boundary_entropy": 0.688497, "bpb": 8.079416}, "tr": {"chunks": 11287.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.623488, "cps": 88.179688, "mean_p": 0.506283, "boundary_entropy": 0.687853, "bpb": 8.0649}, "fi": {"chunks": 6963.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.370156, "cps": 54.398438, "mean_p": 0.496716, "boundary_entropy": 0.687908, "bpb": 8.148121}, "ru": {"chunks": 10548.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.556623, "cps": 82.40625, "mean_p": 0.49676, "boundary_entropy": 0.690194, "bpb": 7.94929}, "el": {"chunks": 36519.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.779418, "cps": 285.304688, "mean_p": 0.529732, "boundary_entropy": 0.688827, "bpb": 8.089757}, "hi": {"chunks": 42522.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.527311, "cps": 332.203125, "mean_p": 0.520726, "boundary_entropy": 0.690207, "bpb": 7.805608}, "ta": {"chunks": 49565.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.651244, "cps": 387.226562, "mean_p": 0.520379, "boundary_entropy": 0.690465, "bpb": 7.714322}, "th": {"chunks": 45709.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.841716, "cps": 357.101562, "mean_p": 0.525309, "boundary_entropy": 0.689933, "bpb": 7.952437}, "zh": {"chunks": 6194.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.048764, "cps": 48.390625, "mean_p": 0.496744, "boundary_entropy": 0.68655, "bpb": 8.018801}, "my": {"chunks": 53244.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.810006, "cps": 415.96875, "mean_p": 0.520992, "boundary_entropy": 0.690601, "bpb": 8.175993}}, "bpb": {"en": 8.13393, "de": 8.09566, "vi": 8.07942, "tr": 8.0649, "fi": 8.14812, "ru": 7.94929, "el": 8.08976, "hi": 7.80561, "ta": 7.71432, "th": 7.95244, "zh": 8.0188, "my": 8.17599}, "gini_cpc": 0.392612, "gini_cps": 0.410351, "bpic": 1.325298, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1817447907099532, "vi": 0.6910362703175081, "tr": 1.2549801672693877, "fi": 0.7450636930397636, "ru": 1.1203914754302478, "el": 3.5816811442540235, "hi": 5.087067487597869, "ta": 5.336524775241704, "th": 5.719914588895693, "zh": 2.110992291211398, "my": 5.656088389466361}, "t": 22.9} |
| {"event": "probe", "step": 4, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 8434.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.484629, "cps": 65.890625, "mean_p": 0.502847, "boundary_entropy": 0.687406, "bpb": 7.880273}, "de": {"chunks": 11665.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.56811, "cps": 91.132812, "mean_p": 0.506292, "boundary_entropy": 0.688173, "bpb": 7.857334}, "vi": {"chunks": 6154.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.333822, "cps": 48.078125, "mean_p": 0.490666, "boundary_entropy": 0.688473, "bpb": 7.948007}, "tr": {"chunks": 11016.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.608518, "cps": 86.0625, "mean_p": 0.505595, "boundary_entropy": 0.687857, "bpb": 7.887226}, "fi": {"chunks": 6503.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.345702, "cps": 50.804688, "mean_p": 0.495723, "boundary_entropy": 0.687877, "bpb": 7.953887}, "ru": {"chunks": 10720.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.565699, "cps": 83.75, "mean_p": 0.496887, "boundary_entropy": 0.690197, "bpb": 7.714127}, "el": {"chunks": 36343.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.770842, "cps": 283.929688, "mean_p": 0.528228, "boundary_entropy": 0.688991, "bpb": 7.945107}, "hi": {"chunks": 42405.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.520357, "cps": 331.289062, "mean_p": 0.519956, "boundary_entropy": 0.690264, "bpb": 7.626577}, "ta": {"chunks": 49031.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.62268, "cps": 383.054688, "mean_p": 0.518698, "boundary_entropy": 0.690591, "bpb": 7.463843}, "th": {"chunks": 45683.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.840099, "cps": 356.898438, "mean_p": 0.524341, "boundary_entropy": 0.690025, "bpb": 7.684564}, "zh": {"chunks": 6136.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.038943, "cps": 47.9375, "mean_p": 0.496565, "boundary_entropy": 0.686548, "bpb": 7.983291}, "my": {"chunks": 53171.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.806154, "cps": 415.398438, "mean_p": 0.520203, "boundary_entropy": 0.690662, "bpb": 8.006445}}, "bpb": {"en": 7.88027, "de": 7.85733, "vi": 7.94801, "tr": 7.88723, "fi": 7.95389, "ru": 7.71413, "el": 7.94511, "hi": 7.62658, "ta": 7.46384, "th": 7.68456, "zh": 7.98329, "my": 8.00644}, "gini_cpc": 0.396288, "gini_cps": 0.413361, "bpic": 1.336001, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1722570667430317, "vi": 0.6888186122542855, "tr": 1.2556364064922887, "fi": 0.7133331287765762, "ru": 1.1672828224494023, "el": 3.6540160653776868, "hi": 5.200588821655501, "ta": 5.411726063514869, "th": 5.860357019624447, "zh": 2.1437909430978617, "my": 5.79031213642179}, "t": 30.1} |
| {"event": "probe", "step": 8, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 7613.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.437453, "cps": 59.476562, "mean_p": 0.500778, "boundary_entropy": 0.68737, "bpb": 7.154011}, "de": {"chunks": 10357.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.504408, "cps": 80.914062, "mean_p": 0.503632, "boundary_entropy": 0.688184, "bpb": 7.186516}, "vi": {"chunks": 5626.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.30518, "cps": 43.953125, "mean_p": 0.489196, "boundary_entropy": 0.688385, "bpb": 7.546672}, "tr": {"chunks": 10041.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.554659, "cps": 78.445312, "mean_p": 0.503186, "boundary_entropy": 0.687852, "bpb": 7.361732}, "fi": {"chunks": 5054.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.268673, "cps": 39.484375, "mean_p": 0.4921, "boundary_entropy": 0.687733, "bpb": 7.360953}, "ru": {"chunks": 12133.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.640264, "cps": 94.789062, "mean_p": 0.498129, "boundary_entropy": 0.690223, "bpb": 6.972911}, "el": {"chunks": 35198.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.715051, "cps": 274.984375, "mean_p": 0.522751, "boundary_entropy": 0.689513, "bpb": 7.476373}, "hi": {"chunks": 41448.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 2.463477, "cps": 323.8125, "mean_p": 0.516811, "boundary_entropy": 0.690472, "bpb": 7.07543}, "ta": {"chunks": 44752.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 2.393795, "cps": 349.625, "mean_p": 0.512875, "boundary_entropy": 0.690936, "bpb": 6.797461}, "th": {"chunks": 45451.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 2.825676, "cps": 355.085938, "mean_p": 0.520858, "boundary_entropy": 0.690323, "bpb": 6.953583}, "zh": {"chunks": 5982.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.012868, "cps": 46.734375, "mean_p": 0.496042, "boundary_entropy": 0.686543, "bpb": 7.877058}, "my": {"chunks": 52938.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.793857, "cps": 413.578125, "mean_p": 0.518188, "boundary_entropy": 0.690804, "bpb": 7.572364}}, "bpb": {"en": 7.15401, "de": 7.18652, "vi": 7.54667, "tr": 7.36173, "fi": 7.36095, "ru": 6.97291, "el": 7.47637, "hi": 7.07543, "ta": 6.79746, "th": 6.95358, "zh": 7.87706, "my": 7.57236}, "gini_cpc": 0.408452, "gini_cps": 0.422691, "bpic": 1.38753, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 1.153054564911188, "vi": 0.6976295632379926, "tr": 1.2679283312626939, "fi": 0.6141743071132484, "ru": 1.4636164219861387, "el": 3.9205358747000534, "hi": 5.631405449620956, "ta": 5.472115682248032, "th": 6.45937752452215, "zh": 2.3153745560343806, "my": 6.386640106873621}, "t": 38.1} |
| {"event": "probe", "step": 16, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 4512.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.259266, "cps": 35.25, "mean_p": 0.492312, "boundary_entropy": 0.687145, "bpb": 6.329284}, "de": {"chunks": 6199.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.301904, "cps": 48.429688, "mean_p": 0.494678, "boundary_entropy": 0.688101, "bpb": 6.4048}, "vi": {"chunks": 3435.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.18633, "cps": 26.835938, "mean_p": 0.481709, "boundary_entropy": 0.687843, "bpb": 7.02421}, "tr": {"chunks": 6314.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.348782, "cps": 49.328125, "mean_p": 0.493769, "boundary_entropy": 0.687686, "bpb": 6.708862}, "fi": {"chunks": 2452.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.130349, "cps": 19.15625, "mean_p": 0.482994, "boundary_entropy": 0.687182, "bpb": 6.607187}, "ru": {"chunks": 14747.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.778206, "cps": 115.210938, "mean_p": 0.500093, "boundary_entropy": 0.690296, "bpb": 5.841541}, "el": {"chunks": 19899.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.969595, "cps": 155.460938, "mean_p": 0.50322, "boundary_entropy": 0.690397, "bpb": 6.520594}, "hi": {"chunks": 27437.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 1.630728, "cps": 214.351562, "mean_p": 0.504113, "boundary_entropy": 0.690918, "bpb": 5.80188}, "ta": {"chunks": 7755.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.414817, "cps": 60.585938, "mean_p": 0.490863, "boundary_entropy": 0.691013, "bpb": 5.849151}, "th": {"chunks": 26106.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 1.623003, "cps": 203.953125, "mean_p": 0.502673, "boundary_entropy": 0.691089, "bpb": 5.339763}, "zh": {"chunks": 5218.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.883508, "cps": 40.765625, "mean_p": 0.493437, "boundary_entropy": 0.686482, "bpb": 7.501171}, "my": {"chunks": 49061.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 2.589244, "cps": 383.289062, "mean_p": 0.510268, "boundary_entropy": 0.691206, "bpb": 6.227632}}, "bpb": {"en": 6.32928, "de": 6.4048, "vi": 7.02421, "tr": 6.70886, "fi": 6.60719, "ru": 5.84154, "el": 6.52059, "hi": 5.80188, "ta": 5.84915, "th": 5.33976, "zh": 7.50117, "my": 6.22763}, "gini_cpc": 0.452263, "gini_cps": 0.471869, "bpic": 2.216658, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1644591516406904, "vi": 0.7186850795491924, "tr": 1.3452687546791937, "fi": 0.5027633503258053, "ru": 3.001577043451412, "el": 3.7397746728890575, "hi": 6.28979628370902, "ta": 1.5999682401711688, "th": 6.2599994874304725, "zh": 3.407733795779265, "my": 9.98683901545776}, "t": 46.7} |
| {"event": "probe", "step": 32, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 5969.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.342987, "cps": 46.632812, "mean_p": 0.498489, "boundary_entropy": 0.687578, "bpb": 5.766197}, "de": {"chunks": 8124.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.395656, "cps": 63.46875, "mean_p": 0.499946, "boundary_entropy": 0.688533, "bpb": 5.889633}, "vi": {"chunks": 3475.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.1885, "cps": 27.148438, "mean_p": 0.484949, "boundary_entropy": 0.688276, "bpb": 6.442789}, "tr": {"chunks": 5515.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.304646, "cps": 43.085938, "mean_p": 0.494083, "boundary_entropy": 0.68797, "bpb": 6.245093}, "fi": {"chunks": 4399.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.233853, "cps": 34.367188, "mean_p": 0.492992, "boundary_entropy": 0.688004, "bpb": 5.942891}, "ru": {"chunks": 2294.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.121055, "cps": 17.921875, "mean_p": 0.488793, "boundary_entropy": 0.690079, "bpb": 5.229596}, "el": {"chunks": 489.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.023827, "cps": 3.820312, "mean_p": 0.466866, "boundary_entropy": 0.688048, "bpb": 6.171031}, "hi": {"chunks": 6119.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.363685, "cps": 47.804688, "mean_p": 0.492925, "boundary_entropy": 0.6908, "bpb": 4.968292}, "ta": {"chunks": 4220.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.225729, "cps": 32.96875, "mean_p": 0.489002, "boundary_entropy": 0.69096, "bpb": 5.234248}, "th": {"chunks": 2580.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.160398, "cps": 20.15625, "mean_p": 0.488678, "boundary_entropy": 0.690771, "bpb": 4.934323}, "zh": {"chunks": 2789.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.472232, "cps": 21.789062, "mean_p": 0.484962, "boundary_entropy": 0.686151, "bpb": 7.083309}, "my": {"chunks": 4965.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.262033, "cps": 38.789062, "mean_p": 0.491066, "boundary_entropy": 0.691176, "bpb": 4.775523}}, "bpb": {"en": 5.7662, "de": 5.88963, "vi": 6.44279, "tr": 6.24509, "fi": 5.94289, "ru": 5.2296, "el": 6.17103, "hi": 4.96829, "ta": 5.23425, "th": 4.93432, "zh": 7.08331, "my": 4.77552}, "gini_cpc": 0.26563, "gini_cps": 0.26243, "bpic": 7.534277, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.1535596295074022, "vi": 0.5495841615092679, "tr": 0.8882137799981449, "fi": 0.681811967399227, "ru": 0.3529447616555369, "el": 0.06946892129826646, "hi": 1.060346779310738, "ta": 0.6581267188130024, "th": 0.4676502620288755, "zh": 1.3768214168054307, "my": 0.7639737174896984}, "t": 55.9} |
| {"event": "probe", "step": 64, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 6270.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.360283, "cps": 48.984375, "mean_p": 0.498953, "boundary_entropy": 0.687616, "bpb": 5.367607}, "de": {"chunks": 6375.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.310476, "cps": 49.804688, "mean_p": 0.496455, "boundary_entropy": 0.68846, "bpb": 5.511131}, "vi": {"chunks": 4534.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.245945, "cps": 35.421875, "mean_p": 0.489237, "boundary_entropy": 0.688653, "bpb": 5.905191}, "tr": {"chunks": 4848.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.267801, "cps": 37.875, "mean_p": 0.493177, "boundary_entropy": 0.688014, "bpb": 5.846633}, "fi": {"chunks": 6366.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.338419, "cps": 49.734375, "mean_p": 0.497393, "boundary_entropy": 0.688184, "bpb": 5.549275}, "ru": {"chunks": 5297.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.279525, "cps": 41.382812, "mean_p": 0.495122, "boundary_entropy": 0.690389, "bpb": 4.832428}, "el": {"chunks": 1414.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.068898, "cps": 11.046875, "mean_p": 0.488616, "boundary_entropy": 0.690309, "bpb": 5.108336}, "hi": {"chunks": 9259.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.550312, "cps": 72.335938, "mean_p": 0.495771, "boundary_entropy": 0.690912, "bpb": 4.444678}, "ta": {"chunks": 13884.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.742658, "cps": 108.46875, "mean_p": 0.494217, "boundary_entropy": 0.691139, "bpb": 4.373962}, "th": {"chunks": 4657.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.289524, "cps": 36.382812, "mean_p": 0.491458, "boundary_entropy": 0.69089, "bpb": 4.443069}, "zh": {"chunks": 1606.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.271927, "cps": 12.546875, "mean_p": 0.484334, "boundary_entropy": 0.686407, "bpb": 6.715008}, "my": {"chunks": 20502.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 1.082014, "cps": 160.171875, "mean_p": 0.499223, "boundary_entropy": 0.6914, "bpb": 4.24448}}, "bpb": {"en": 5.36761, "de": 5.51113, "vi": 5.90519, "tr": 5.84663, "fi": 5.54928, "ru": 4.83243, "el": 5.10834, "hi": 4.44468, "ta": 4.37396, "th": 4.44307, "zh": 6.71501, "my": 4.24448}, "gini_cpc": 0.31601, "gini_cps": 0.360402, "bpic": 4.514433, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 0.8617560901496645, "vi": 0.6826450622450794, "tr": 0.7433077126184113, "fi": 0.9393151568905398, "ru": 0.7758492381108685, "el": 0.19123401519173144, "hi": 1.5274450329525016, "ta": 2.061321413626702, "th": 0.8036033750128281, "zh": 0.7547596556579392, "my": 3.0032358012308684}, "t": 69.9} |
| {"event": "train", "step": 100, "phase": 1, "objective": "global", "lm_loss": 3.29731, "bpb": 4.75701, "chunk_loss": 0.03079, "F": 0.17804, "G": 0.47277, "chunks": 182.31, "bytes_per_chunk": 5.617, "boundary_entropy": 0.6886, "router_noise": 0.0, "grad_norm": 2.2762, "lr": 6.06e-05, "router_trainable": true, "sec_per_step": 0.7, "alpha": 0.03, "t": 79.7} |
| {"event": "probe", "step": 128, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3760.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.216055, "cps": 29.375, "mean_p": 0.459678, "boundary_entropy": 0.662248, "bpb": 4.291402}, "de": {"chunks": 4368.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.212731, "cps": 34.125, "mean_p": 0.46023, "boundary_entropy": 0.667185, "bpb": 4.523153}, "vi": {"chunks": 7303.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.396149, "cps": 57.054688, "mean_p": 0.492307, "boundary_entropy": 0.683186, "bpb": 4.691159}, "tr": {"chunks": 4311.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.238137, "cps": 33.679688, "mean_p": 0.478313, "boundary_entropy": 0.677909, "bpb": 4.833272}, "fi": {"chunks": 4261.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.226516, "cps": 33.289062, "mean_p": 0.472286, "boundary_entropy": 0.677363, "bpb": 4.537254}, "ru": {"chunks": 5165.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.272559, "cps": 40.351562, "mean_p": 0.481868, "boundary_entropy": 0.687925, "bpb": 3.215656}, "el": {"chunks": 6161.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.3002, "cps": 48.132812, "mean_p": 0.490412, "boundary_entropy": 0.690249, "bpb": 4.08754}, "hi": {"chunks": 8762.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.520773, "cps": 68.453125, "mean_p": 0.470488, "boundary_entropy": 0.687884, "bpb": 2.873657}, "ta": {"chunks": 12806.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.684996, "cps": 100.046875, "mean_p": 0.438972, "boundary_entropy": 0.679017, "bpb": 2.948281}, "th": {"chunks": 9520.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.591856, "cps": 74.375, "mean_p": 0.480058, "boundary_entropy": 0.689311, "bpb": 2.734758}, "zh": {"chunks": 2122.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.359296, "cps": 16.578125, "mean_p": 0.486217, "boundary_entropy": 0.686526, "bpb": 6.215631}, "my": {"chunks": 11886.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.627296, "cps": 92.859375, "mean_p": 0.476607, "boundary_entropy": 0.689448, "bpb": 2.983492}}, "bpb": {"en": 4.2914, "de": 4.52315, "vi": 4.69116, "tr": 4.83327, "fi": 4.53725, "ru": 3.21566, "el": 4.08754, "hi": 2.87366, "ta": 2.94828, "th": 2.73476, "zh": 6.21563, "my": 2.98349}, "gini_cpc": 0.238465, "gini_cps": 0.268912, "bpic": 4.771912, "worst_lang": "ta", "cpc_parity_vs_en": {"en": 1.0, "de": 0.984615113605395, "vi": 1.83355707806035, "tr": 1.1022084766719047, "fi": 1.0484215183058352, "ru": 1.2615294307528209, "el": 1.3894618881151506, "hi": 2.410374095033353, "ta": 3.1704747827670383, "th": 2.7393792286986027, "zh": 1.6629845415048525, "my": 2.9034117169498606}, "t": 89.1} |
| {"event": "train", "step": 200, "phase": 1, "objective": "global", "lm_loss": 2.44276, "bpb": 3.52416, "chunk_loss": 0.031663, "F": 0.19553, "G": 0.40702, "chunks": 200.22, "bytes_per_chunk": 5.114, "boundary_entropy": 0.58832, "router_noise": 0.0, "grad_norm": 1.4763, "lr": 0.0001206, "router_trainable": true, "sec_per_step": 0.233, "alpha": 0.03, "t": 103.0} |
| {"event": "probe", "step": 256, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2946.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.169281, "cps": 23.015625, "mean_p": 0.336823, "boundary_entropy": 0.461623, "bpb": 3.529895}, "de": {"chunks": 3308.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.161107, "cps": 25.84375, "mean_p": 0.353934, "boundary_entropy": 0.512966, "bpb": 3.742741}, "vi": {"chunks": 6282.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.340765, "cps": 49.078125, "mean_p": 0.342587, "boundary_entropy": 0.439471, "bpb": 3.078041}, "tr": {"chunks": 3594.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.198531, "cps": 28.078125, "mean_p": 0.349837, "boundary_entropy": 0.515656, "bpb": 3.716207}, "fi": {"chunks": 2927.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.1556, "cps": 22.867188, "mean_p": 0.331652, "boundary_entropy": 0.511625, "bpb": 3.784644}, "ru": {"chunks": 6182.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.326227, "cps": 48.296875, "mean_p": 0.347347, "boundary_entropy": 0.595148, "bpb": 2.602784}, "el": {"chunks": 10356.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.504605, "cps": 80.90625, "mean_p": 0.39619, "boundary_entropy": 0.605829, "bpb": 2.685819}, "hi": {"chunks": 10876.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.646419, "cps": 84.96875, "mean_p": 0.439637, "boundary_entropy": 0.665325, "bpb": 2.115026}, "ta": {"chunks": 7954.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.425461, "cps": 62.140625, "mean_p": 0.436208, "boundary_entropy": 0.675131, "bpb": 1.94915}, "th": {"chunks": 5612.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.348896, "cps": 43.84375, "mean_p": 0.44498, "boundary_entropy": 0.674681, "bpb": 2.082021}, "zh": {"chunks": 3881.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.657128, "cps": 30.320312, "mean_p": 0.393163, "boundary_entropy": 0.626965, "bpb": 4.955444}, "my": {"chunks": 10450.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.551509, "cps": 81.640625, "mean_p": 0.301746, "boundary_entropy": 0.556996, "bpb": 2.157735}}, "bpb": {"en": 3.5299, "de": 3.74274, "vi": 3.07804, "tr": 3.71621, "fi": 3.78464, "ru": 2.60278, "el": 2.68582, "hi": 2.11503, "ta": 1.94915, "th": 2.08202, "zh": 4.95544, "my": 2.15774}, "gini_cpc": 0.26746, "gini_cps": 0.26208, "bpic": 5.160566, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9517096466737813, "vi": 2.013011091427634, "tr": 1.172786340394572, "fi": 0.9191834932998969, "ru": 1.927130674032318, "el": 2.9808668293573164, "hi": 3.818611742418511, "ta": 2.5133414567320083, "th": 2.061047376241416, "zh": 3.8818752788941207, "my": 3.257949078773636}, "t": 116.4} |
| {"event": "train", "step": 300, "phase": 1, "objective": "global", "lm_loss": 2.14565, "bpb": 3.09551, "chunk_loss": 0.031036, "F": 0.19417, "G": 0.33926, "chunks": 198.83, "bytes_per_chunk": 5.15, "boundary_entropy": 0.50189, "router_noise": 0.0, "grad_norm": 1.0344, "lr": 0.0001806, "router_trainable": true, "sec_per_step": 0.223, "alpha": 0.03, "t": 125.3} |
| {"event": "train", "step": 400, "phase": 1, "objective": "global", "lm_loss": 2.06232, "bpb": 2.97529, "chunk_loss": 0.030937, "F": 0.19995, "G": 0.30689, "chunks": 204.75, "bytes_per_chunk": 5.001, "boundary_entropy": 0.42034, "router_noise": 0.0, "grad_norm": 1.2051, "lr": 0.0002406, "router_trainable": true, "sec_per_step": 0.183, "alpha": 0.03, "t": 143.6} |
| {"event": "train", "step": 500, "phase": 1, "objective": "global", "lm_loss": 1.97028, "bpb": 2.84251, "chunk_loss": 0.030917, "F": 0.19934, "G": 0.2927, "chunks": 204.12, "bytes_per_chunk": 5.017, "boundary_entropy": 0.37185, "router_noise": 0.0, "grad_norm": 0.9206, "lr": 0.0003, "router_trainable": true, "sec_per_step": 0.173, "alpha": 0.03, "t": 160.8} |
| {"event": "train", "step": 600, "phase": 1, "objective": "global", "lm_loss": 1.83021, "bpb": 2.64043, "chunk_loss": 0.030369, "F": 0.1877, "G": 0.2578, "chunks": 192.2, "bytes_per_chunk": 5.328, "boundary_entropy": 0.30064, "router_noise": 0.0, "grad_norm": 0.842, "lr": 0.0002999, "router_trainable": true, "sec_per_step": 0.172, "alpha": 0.03, "t": 178.0} |
| {"event": "train", "step": 700, "phase": 1, "objective": "global", "lm_loss": 1.716, "bpb": 2.47567, "chunk_loss": 0.030566, "F": 0.19673, "G": 0.25461, "chunks": 201.45, "bytes_per_chunk": 5.083, "boundary_entropy": 0.27726, "router_noise": 0.0, "grad_norm": 0.6722, "lr": 0.0002998, "router_trainable": true, "sec_per_step": 0.168, "alpha": 0.03, "t": 194.8} |
| {"event": "train", "step": 800, "phase": 1, "objective": "global", "lm_loss": 1.6032, "bpb": 2.31293, "chunk_loss": 0.030376, "F": 0.19089, "G": 0.23775, "chunks": 195.47, "bytes_per_chunk": 5.239, "boundary_entropy": 0.23164, "router_noise": 0.0, "grad_norm": 0.6708, "lr": 0.0002995, "router_trainable": true, "sec_per_step": 0.171, "alpha": 0.03, "t": 211.9} |
| {"event": "train", "step": 900, "phase": 1, "objective": "global", "lm_loss": 1.57045, "bpb": 2.26568, "chunk_loss": 0.030515, "F": 0.19669, "G": 0.24411, "chunks": 201.41, "bytes_per_chunk": 5.084, "boundary_entropy": 0.24162, "router_noise": 0.0, "grad_norm": 0.6413, "lr": 0.000299, "router_trainable": true, "sec_per_step": 0.169, "alpha": 0.03, "t": 228.7} |
| {"event": "probe", "step": 1000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3420.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.196518, "cps": 26.71875, "mean_p": 0.219015, "boundary_entropy": 0.150876, "bpb": 2.331789}, "de": {"chunks": 3917.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.190766, "cps": 30.601562, "mean_p": 0.220853, "boundary_entropy": 0.182328, "bpb": 2.614792}, "vi": {"chunks": 4266.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.231408, "cps": 33.328125, "mean_p": 0.227234, "boundary_entropy": 0.206596, "bpb": 2.09889}, "tr": {"chunks": 3797.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.209744, "cps": 29.664062, "mean_p": 0.214389, "boundary_entropy": 0.182758, "bpb": 2.598088}, "fi": {"chunks": 3222.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.171283, "cps": 25.171875, "mean_p": 0.214214, "boundary_entropy": 0.230808, "bpb": 2.852951}, "ru": {"chunks": 7558.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.398839, "cps": 59.046875, "mean_p": 0.229152, "boundary_entropy": 0.130622, "bpb": 1.611162}, "el": {"chunks": 7283.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.35487, "cps": 56.898438, "mean_p": 0.234114, "boundary_entropy": 0.213383, "bpb": 1.738832}, "hi": {"chunks": 7938.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.471798, "cps": 62.015625, "mean_p": 0.272627, "boundary_entropy": 0.41431, "bpb": 1.463074}, "ta": {"chunks": 9159.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.489917, "cps": 71.554688, "mean_p": 0.284386, "boundary_entropy": 0.468971, "bpb": 1.148604}, "th": {"chunks": 6565.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.408144, "cps": 51.289062, "mean_p": 0.288511, "boundary_entropy": 0.491041, "bpb": 1.409392}, "zh": {"chunks": 2876.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.486962, "cps": 22.46875, "mean_p": 0.245984, "boundary_entropy": 0.2693, "bpb": 3.466458}, "my": {"chunks": 9585.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.505858, "cps": 74.882812, "mean_p": 0.31512, "boundary_entropy": 0.517805, "bpb": 1.275724}}, "bpb": {"en": 2.33179, "de": 2.61479, "vi": 2.09889, "tr": 2.59809, "fi": 2.85295, "ru": 1.61116, "el": 1.73883, "hi": 1.46307, "ta": 1.1486, "th": 1.40939, "zh": 3.46646, "my": 1.27572}, "gini_cpc": 0.208333, "gini_cps": 0.227608, "bpic": 5.515204, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9707316250007475, "vi": 1.1775401481735257, "tr": 1.0673038102631047, "fi": 0.8715888519874989, "ru": 2.0295309910660557, "el": 1.8057909782007522, "hi": 2.4007892390709316, "ta": 2.4929903858335325, "th": 2.076881315816741, "zh": 2.4779552165208143, "my": 2.5741079464017864}, "t": 248.4} |
| {"event": "eval", "step": 1000, "phase": 1, "bpb": {"en": 2.33986, "de": 2.60593, "vi": 2.10299, "tr": 2.58872, "fi": 2.86254, "ru": 1.61225, "el": 1.73399, "hi": 1.47121, "ta": 1.14743, "th": 1.40603, "zh": 3.47178, "my": 1.28767}, "bytes_per_chunk": {"en": 4.915, "de": 5.423, "vi": 5.695, "tr": 5.186, "fi": 6.102, "ru": 4.641, "el": 5.213, "hi": 5.291, "ta": 5.582, "th": 6.725, "zh": 5.69, "my": 6.219}, "bpb_mean": 2.05253, "t": 253.8} |
| {"event": "train", "step": 1000, "phase": 1, "objective": "global", "lm_loss": 1.4909, "bpb": 2.15091, "chunk_loss": 0.030289, "F": 0.18941, "G": 0.23174, "chunks": 193.95, "bytes_per_chunk": 5.28, "boundary_entropy": 0.21113, "router_noise": 0.0, "grad_norm": 0.62, "lr": 0.0002985, "router_trainable": true, "sec_per_step": 0.253, "alpha": 0.03, "t": 254.0} |
| {"event": "train", "step": 1100, "phase": 1, "objective": "global", "lm_loss": 1.41284, "bpb": 2.0383, "chunk_loss": 0.03027, "F": 0.18813, "G": 0.23802, "chunks": 192.64, "bytes_per_chunk": 5.316, "boundary_entropy": 0.22854, "router_noise": 0.0, "grad_norm": 0.5533, "lr": 0.0002978, "router_trainable": true, "sec_per_step": 0.169, "alpha": 0.03, "t": 270.9} |
| {"event": "train", "step": 1200, "phase": 1, "objective": "global", "lm_loss": 1.44831, "bpb": 2.08946, "chunk_loss": 0.030394, "F": 0.19136, "G": 0.2321, "chunks": 195.95, "bytes_per_chunk": 5.226, "boundary_entropy": 0.21469, "router_noise": 0.0, "grad_norm": 0.5491, "lr": 0.000297, "router_trainable": true, "sec_per_step": 0.17, "alpha": 0.03, "t": 287.8} |
| {"event": "train", "step": 1300, "phase": 1, "objective": "global", "lm_loss": 1.45797, "bpb": 2.1034, "chunk_loss": 0.030263, "F": 0.18767, "G": 0.22816, "chunks": 192.17, "bytes_per_chunk": 5.329, "boundary_entropy": 0.2043, "router_noise": 0.0, "grad_norm": 0.6151, "lr": 0.0002962, "router_trainable": true, "sec_per_step": 0.168, "alpha": 0.03, "t": 304.6} |
| {"event": "train", "step": 1400, "phase": 1, "objective": "global", "lm_loss": 1.41968, "bpb": 2.04817, "chunk_loss": 0.030328, "F": 0.19185, "G": 0.22786, "chunks": 196.45, "bytes_per_chunk": 5.212, "boundary_entropy": 0.18642, "router_noise": 0.0, "grad_norm": 0.4653, "lr": 0.0002951, "router_trainable": true, "sec_per_step": 0.169, "alpha": 0.03, "t": 321.5} |
| {"event": "train", "step": 1500, "phase": 1, "objective": "global", "lm_loss": 1.4209, "bpb": 2.04993, "chunk_loss": 0.030339, "F": 0.18796, "G": 0.22828, "chunks": 192.47, "bytes_per_chunk": 5.32, "boundary_entropy": 0.20379, "router_noise": 0.0, "grad_norm": 0.498, "lr": 0.000294, "router_trainable": true, "sec_per_step": 0.169, "alpha": 0.03, "t": 338.4} |
| {"event": "train", "step": 1600, "phase": 1, "objective": "global", "lm_loss": 1.40301, "bpb": 2.02412, "chunk_loss": 0.030352, "F": 0.19044, "G": 0.22533, "chunks": 195.02, "bytes_per_chunk": 5.251, "boundary_entropy": 0.18575, "router_noise": 0.0, "grad_norm": 0.4967, "lr": 0.0002928, "router_trainable": true, "sec_per_step": 0.168, "alpha": 0.03, "t": 355.2} |
| {"event": "train", "step": 1700, "phase": 1, "objective": "global", "lm_loss": 1.34787, "bpb": 1.94456, "chunk_loss": 0.030243, "F": 0.18884, "G": 0.2177, "chunks": 193.38, "bytes_per_chunk": 5.295, "boundary_entropy": 0.16426, "router_noise": 0.0, "grad_norm": 0.4446, "lr": 0.0002914, "router_trainable": true, "sec_per_step": 0.168, "alpha": 0.03, "t": 372.0} |
| {"event": "train", "step": 1800, "phase": 1, "objective": "global", "lm_loss": 1.35596, "bpb": 1.95624, "chunk_loss": 0.03019, "F": 0.18494, "G": 0.22791, "chunks": 189.38, "bytes_per_chunk": 5.407, "boundary_entropy": 0.20159, "router_noise": 0.0, "grad_norm": 0.4447, "lr": 0.0002899, "router_trainable": true, "sec_per_step": 0.17, "alpha": 0.03, "t": 389.0} |
| {"event": "train", "step": 1900, "phase": 1, "objective": "global", "lm_loss": 1.35829, "bpb": 1.9596, "chunk_loss": 0.030425, "F": 0.19357, "G": 0.22614, "chunks": 198.22, "bytes_per_chunk": 5.166, "boundary_entropy": 0.1779, "router_noise": 0.0, "grad_norm": 0.4873, "lr": 0.0002883, "router_trainable": true, "sec_per_step": 0.17, "alpha": 0.03, "t": 406.0} |
| {"event": "probe", "step": 2000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3382.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.194334, "cps": 26.421875, "mean_p": 0.20364, "boundary_entropy": 0.102361, "bpb": 2.078448}, "de": {"chunks": 3804.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.185263, "cps": 29.71875, "mean_p": 0.201494, "boundary_entropy": 0.125387, "bpb": 2.288725}, "vi": {"chunks": 4262.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.231191, "cps": 33.296875, "mean_p": 0.201183, "boundary_entropy": 0.12905, "bpb": 1.930903}, "tr": {"chunks": 3339.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.184445, "cps": 26.085938, "mean_p": 0.185624, "boundary_entropy": 0.139047, "bpb": 2.290826}, "fi": {"chunks": 3179.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.168997, "cps": 24.835938, "mean_p": 0.189364, "boundary_entropy": 0.160768, "bpb": 2.481033}, "ru": {"chunks": 7572.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.399578, "cps": 59.15625, "mean_p": 0.222592, "boundary_entropy": 0.109869, "bpb": 1.351432}, "el": {"chunks": 6808.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.331725, "cps": 53.1875, "mean_p": 0.209803, "boundary_entropy": 0.168646, "bpb": 1.425878}, "hi": {"chunks": 6808.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.404636, "cps": 53.1875, "mean_p": 0.263457, "boundary_entropy": 0.409846, "bpb": 1.228226}, "ta": {"chunks": 8282.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.443006, "cps": 64.703125, "mean_p": 0.268315, "boundary_entropy": 0.450787, "bpb": 1.027977}, "th": {"chunks": 6448.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.40087, "cps": 50.375, "mean_p": 0.310815, "boundary_entropy": 0.505847, "bpb": 1.178256}, "zh": {"chunks": 2838.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.480528, "cps": 22.171875, "mean_p": 0.226076, "boundary_entropy": 0.20713, "bpb": 3.174671}, "my": {"chunks": 11895.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.627771, "cps": 92.929688, "mean_p": 0.336582, "boundary_entropy": 0.519398, "bpb": 1.156581}}, "bpb": {"en": 2.07845, "de": 2.28872, "vi": 1.9309, "tr": 2.29083, "fi": 2.48103, "ru": 1.35143, "el": 1.42588, "hi": 1.22823, "ta": 1.02798, "th": 1.17826, "zh": 3.17467, "my": 1.15658}, "gini_cpc": 0.229165, "gini_cps": 0.247161, "bpic": 5.593089, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9533198106073382, "vi": 1.1896544141458225, "tr": 0.9491096401421748, "fi": 0.8696192833055862, "ru": 2.056136335621301, "el": 1.7069830900082303, "hi": 2.082164276745255, "ta": 2.279608531171033, "th": 2.0627874499512586, "zh": 2.4726888210988514, "my": 3.230364933528433}, "t": 425.8} |
| {"event": "eval", "step": 2000, "phase": 1, "bpb": {"en": 2.08667, "de": 2.28074, "vi": 1.93912, "tr": 2.28169, "fi": 2.49799, "ru": 1.35238, "el": 1.42226, "hi": 1.23551, "ta": 1.02848, "th": 1.17335, "zh": 3.18081, "my": 1.16726}, "bytes_per_chunk": {"en": 4.941, "de": 5.466, "vi": 5.719, "tr": 5.845, "fi": 6.024, "ru": 4.64, "el": 5.511, "hi": 5.963, "ta": 5.963, "th": 6.788, "zh": 5.717, "my": 4.884}, "bpb_mean": 1.80385, "t": 431.2} |
| {"event": "train", "step": 2000, "phase": 1, "objective": "global", "lm_loss": 1.34562, "bpb": 1.94132, "chunk_loss": 0.030266, "F": 0.1913, "G": 0.21834, "chunks": 195.89, "bytes_per_chunk": 5.227, "boundary_entropy": 0.16046, "router_noise": 0.0, "grad_norm": 0.4317, "lr": 0.0002866, "router_trainable": true, "sec_per_step": 0.254, "alpha": 0.03, "t": 431.4} |
| {"event": "train", "step": 2100, "phase": 1, "objective": "global", "lm_loss": 1.28984, "bpb": 1.86084, "chunk_loss": 0.030283, "F": 0.19223, "G": 0.22855, "chunks": 196.84, "bytes_per_chunk": 5.202, "boundary_entropy": 0.17929, "router_noise": 0.0, "grad_norm": 0.43, "lr": 0.0002848, "router_trainable": true, "sec_per_step": 0.169, "alpha": 0.03, "t": 448.3} |
| {"event": "train", "step": 2200, "phase": 1, "objective": "global", "lm_loss": 1.25669, "bpb": 1.81303, "chunk_loss": 0.030189, "F": 0.18532, "G": 0.22829, "chunks": 189.77, "bytes_per_chunk": 5.396, "boundary_entropy": 0.1997, "router_noise": 0.0, "grad_norm": 0.459, "lr": 0.0002829, "router_trainable": true, "sec_per_step": 0.168, "alpha": 0.03, "t": 465.1} |
| {"event": "train", "step": 2300, "phase": 1, "objective": "global", "lm_loss": 1.32311, "bpb": 1.90885, "chunk_loss": 0.03021, "F": 0.18854, "G": 0.21922, "chunks": 193.06, "bytes_per_chunk": 5.304, "boundary_entropy": 0.16594, "router_noise": 0.0, "grad_norm": 0.3943, "lr": 0.0002809, "router_trainable": true, "sec_per_step": 0.169, "alpha": 0.03, "t": 481.9} |
| {"event": "train", "step": 2400, "phase": 1, "objective": "global", "lm_loss": 1.34537, "bpb": 1.94096, "chunk_loss": 0.030225, "F": 0.18848, "G": 0.21506, "chunks": 193.0, "bytes_per_chunk": 5.306, "boundary_entropy": 0.15752, "router_noise": 0.0, "grad_norm": 0.4473, "lr": 0.0002788, "router_trainable": true, "sec_per_step": 0.169, "alpha": 0.03, "t": 498.8} |
| {"event": "train", "step": 2500, "phase": 1, "objective": "global", "lm_loss": 1.25067, "bpb": 1.80434, "chunk_loss": 0.030212, "F": 0.18803, "G": 0.22273, "chunks": 192.55, "bytes_per_chunk": 5.318, "boundary_entropy": 0.17086, "router_noise": 0.0, "grad_norm": 0.3775, "lr": 0.0002765, "router_trainable": true, "sec_per_step": 0.169, "alpha": 0.03, "t": 515.7} |
| {"event": "train", "step": 2600, "phase": 1, "objective": "global", "lm_loss": 1.27199, "bpb": 1.8351, "chunk_loss": 0.030234, "F": 0.18857, "G": 0.2253, "chunks": 193.09, "bytes_per_chunk": 5.303, "boundary_entropy": 0.17327, "router_noise": 0.0, "grad_norm": 0.4404, "lr": 0.0002742, "router_trainable": true, "sec_per_step": 0.168, "alpha": 0.03, "t": 532.5} |
| {"event": "train", "step": 2700, "phase": 1, "objective": "global", "lm_loss": 1.27027, "bpb": 1.83261, "chunk_loss": 0.030112, "F": 0.18185, "G": 0.21674, "chunks": 186.22, "bytes_per_chunk": 5.499, "boundary_entropy": 0.17184, "router_noise": 0.0, "grad_norm": 0.3911, "lr": 0.0002718, "router_trainable": true, "sec_per_step": 0.168, "alpha": 0.03, "t": 549.3} |
| {"event": "train", "step": 2800, "phase": 1, "objective": "global", "lm_loss": 1.35898, "bpb": 1.96059, "chunk_loss": 0.030053, "F": 0.1812, "G": 0.22106, "chunks": 185.55, "bytes_per_chunk": 5.519, "boundary_entropy": 0.17984, "router_noise": 0.0, "grad_norm": 0.4084, "lr": 0.0002693, "router_trainable": true, "sec_per_step": 0.168, "alpha": 0.03, "t": 566.1} |
| {"event": "train", "step": 2900, "phase": 1, "objective": "global", "lm_loss": 1.27192, "bpb": 1.83499, "chunk_loss": 0.030104, "F": 0.18251, "G": 0.21983, "chunks": 186.89, "bytes_per_chunk": 5.479, "boundary_entropy": 0.17436, "router_noise": 0.0, "grad_norm": 0.3793, "lr": 0.0002667, "router_trainable": true, "sec_per_step": 0.168, "alpha": 0.03, "t": 582.9} |
| {"event": "probe", "step": 3000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3351.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.192553, "cps": 26.179688, "mean_p": 0.198451, "boundary_entropy": 0.090364, "bpb": 1.970738}, "de": {"chunks": 3840.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.187016, "cps": 30.0, "mean_p": 0.197662, "boundary_entropy": 0.101343, "bpb": 2.151856}, "vi": {"chunks": 4252.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.230648, "cps": 33.21875, "mean_p": 0.195709, "boundary_entropy": 0.114839, "bpb": 1.807461}, "tr": {"chunks": 3500.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.193338, "cps": 27.34375, "mean_p": 0.186959, "boundary_entropy": 0.120871, "bpb": 2.143604}, "fi": {"chunks": 3227.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.171549, "cps": 25.210938, "mean_p": 0.187838, "boundary_entropy": 0.138453, "bpb": 2.309372}, "ru": {"chunks": 7556.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.398734, "cps": 59.03125, "mean_p": 0.216383, "boundary_entropy": 0.09914, "bpb": 1.264797}, "el": {"chunks": 6917.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.337036, "cps": 54.039062, "mean_p": 0.203575, "boundary_entropy": 0.122241, "bpb": 1.305372}, "hi": {"chunks": 7379.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.438574, "cps": 57.648438, "mean_p": 0.251843, "boundary_entropy": 0.365277, "bpb": 1.140879}, "ta": {"chunks": 8454.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.452206, "cps": 66.046875, "mean_p": 0.281321, "boundary_entropy": 0.461409, "bpb": 0.955703}, "th": {"chunks": 7751.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.481878, "cps": 60.554688, "mean_p": 0.36423, "boundary_entropy": 0.541921, "bpb": 1.086697}, "zh": {"chunks": 2823.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.477988, "cps": 22.054688, "mean_p": 0.219268, "boundary_entropy": 0.187903, "bpb": 2.986023}, "my": {"chunks": 8560.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.451763, "cps": 66.875, "mean_p": 0.313707, "boundary_entropy": 0.519256, "bpb": 1.083649}}, "bpb": {"en": 1.97074, "de": 2.15186, "vi": 1.80746, "tr": 2.1436, "fi": 2.30937, "ru": 1.2648, "el": 1.30537, "hi": 1.14088, "ta": 0.9557, "th": 1.0867, "zh": 2.98602, "my": 1.08365}, "gini_cpc": 0.204063, "gini_cps": 0.215096, "bpic": 5.676394, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9712443593104987, "vi": 1.1978427434677577, "tr": 1.0040773845039013, "fi": 0.8909160324344025, "ru": 2.070772683143456, "el": 1.750356948256102, "hi": 2.277676965869894, "ta": 2.3484778387836793, "th": 2.5025707486720754, "zh": 2.4823735084512633, "my": 2.3461732614483326}, "t": 602.6} |
| {"event": "eval", "step": 3000, "phase": 1, "bpb": {"en": 1.98078, "de": 2.14435, "vi": 1.81317, "tr": 2.13171, "fi": 2.32624, "ru": 1.26421, "el": 1.30334, "hi": 1.14876, "ta": 0.95636, "th": 1.08233, "zh": 2.9946, "my": 1.09454}, "bytes_per_chunk": {"en": 4.985, "de": 5.51, "vi": 5.73, "tr": 5.677, "fi": 5.92, "ru": 4.656, "el": 5.436, "hi": 5.582, "ta": 5.977, "th": 5.623, "zh": 5.706, "my": 6.914}, "bpb_mean": 1.6867, "t": 607.9} |
| {"event": "train", "step": 3000, "phase": 1, "objective": "global", "lm_loss": 1.32077, "bpb": 1.90547, "chunk_loss": 0.03022, "F": 0.18988, "G": 0.21638, "chunks": 194.44, "bytes_per_chunk": 5.266, "boundary_entropy": 0.14811, "router_noise": 0.0, "grad_norm": 0.3964, "lr": 0.000264, "router_trainable": true, "sec_per_step": 0.252, "alpha": 0.03, "t": 608.1} |
| {"event": "train", "step": 3100, "phase": 1, "objective": "global", "lm_loss": 1.32538, "bpb": 1.91211, "chunk_loss": 0.030358, "F": 0.19281, "G": 0.21669, "chunks": 197.44, "bytes_per_chunk": 5.186, "boundary_entropy": 0.15169, "router_noise": 0.0, "grad_norm": 0.4308, "lr": 0.0002612, "router_trainable": true, "sec_per_step": 0.168, "alpha": 0.03, "t": 624.9} |
| {"event": "train", "step": 3200, "phase": 1, "objective": "global", "lm_loss": 1.2006, "bpb": 1.7321, "chunk_loss": 0.03024, "F": 0.19092, "G": 0.22653, "chunks": 195.5, "bytes_per_chunk": 5.238, "boundary_entropy": 0.17435, "router_noise": 0.0, "grad_norm": 0.367, "lr": 0.0002583, "router_trainable": true, "sec_per_step": 0.169, "alpha": 0.03, "t": 641.9} |
| {"event": "train", "step": 3300, "phase": 1, "objective": "global", "lm_loss": 1.27888, "bpb": 1.84504, "chunk_loss": 0.030245, "F": 0.18684, "G": 0.22869, "chunks": 191.33, "bytes_per_chunk": 5.352, "boundary_entropy": 0.19, "router_noise": 0.0, "grad_norm": 0.3875, "lr": 0.0002553, "router_trainable": true, "sec_per_step": 0.167, "alpha": 0.03, "t": 658.5} |
| {"event": "train", "step": 3400, "phase": 1, "objective": "global", "lm_loss": 1.23552, "bpb": 1.78248, "chunk_loss": 0.030329, "F": 0.19318, "G": 0.22843, "chunks": 197.81, "bytes_per_chunk": 5.177, "boundary_entropy": 0.16897, "router_noise": 0.0, "grad_norm": 0.3841, "lr": 0.0002523, "router_trainable": true, "sec_per_step": 0.169, "alpha": 0.03, "t": 675.4} |
| {"event": "train", "step": 3500, "phase": 1, "objective": "global", "lm_loss": 1.3105, "bpb": 1.89066, "chunk_loss": 0.030138, "F": 0.18549, "G": 0.21291, "chunks": 189.94, "bytes_per_chunk": 5.391, "boundary_entropy": 0.13895, "router_noise": 0.0, "grad_norm": 0.3668, "lr": 0.0002492, "router_trainable": true, "sec_per_step": 0.168, "alpha": 0.03, "t": 692.2} |
| {"event": "train", "step": 3600, "phase": 1, "objective": "global", "lm_loss": 1.22511, "bpb": 1.76746, "chunk_loss": 0.030196, "F": 0.18771, "G": 0.21915, "chunks": 192.22, "bytes_per_chunk": 5.327, "boundary_entropy": 0.15831, "router_noise": 0.0, "grad_norm": 0.3967, "lr": 0.000246, "router_trainable": true, "sec_per_step": 0.169, "alpha": 0.03, "t": 709.0} |
| {"event": "train", "step": 3700, "phase": 1, "objective": "global", "lm_loss": 1.13512, "bpb": 1.63763, "chunk_loss": 0.030155, "F": 0.18504, "G": 0.22843, "chunks": 189.48, "bytes_per_chunk": 5.404, "boundary_entropy": 0.19786, "router_noise": 0.0, "grad_norm": 0.3657, "lr": 0.0002427, "router_trainable": true, "sec_per_step": 0.168, "alpha": 0.03, "t": 725.9} |
| {"event": "train", "step": 3800, "phase": 1, "objective": "global", "lm_loss": 1.22013, "bpb": 1.76028, "chunk_loss": 0.030226, "F": 0.18945, "G": 0.21754, "chunks": 194.0, "bytes_per_chunk": 5.278, "boundary_entropy": 0.15235, "router_noise": 0.0, "grad_norm": 0.3696, "lr": 0.0002394, "router_trainable": true, "sec_per_step": 0.167, "alpha": 0.03, "t": 742.6} |
| {"event": "train", "step": 3900, "phase": 1, "objective": "global", "lm_loss": 1.23632, "bpb": 1.78363, "chunk_loss": 0.030162, "F": 0.18529, "G": 0.21402, "chunks": 189.73, "bytes_per_chunk": 5.397, "boundary_entropy": 0.14466, "router_noise": 0.0, "grad_norm": 0.3789, "lr": 0.000236, "router_trainable": true, "sec_per_step": 0.168, "alpha": 0.03, "t": 759.5} |
| {"event": "probe", "step": 4000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3337.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.191749, "cps": 26.070312, "mean_p": 0.192977, "boundary_entropy": 0.086368, "bpb": 1.871892}, "de": {"chunks": 3770.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.183607, "cps": 29.453125, "mean_p": 0.191169, "boundary_entropy": 0.092603, "bpb": 2.044758}, "vi": {"chunks": 4240.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.229997, "cps": 33.125, "mean_p": 0.193251, "boundary_entropy": 0.105985, "bpb": 1.700236}, "tr": {"chunks": 3310.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.182843, "cps": 25.859375, "mean_p": 0.177963, "boundary_entropy": 0.115406, "bpb": 2.011333}, "fi": {"chunks": 3241.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.172293, "cps": 25.320312, "mean_p": 0.182991, "boundary_entropy": 0.125613, "bpb": 2.184958}, "ru": {"chunks": 7573.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.399631, "cps": 59.164062, "mean_p": 0.21483, "boundary_entropy": 0.090255, "bpb": 1.213282}, "el": {"chunks": 6813.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.331969, "cps": 53.226562, "mean_p": 0.198919, "boundary_entropy": 0.118141, "bpb": 1.233213}, "hi": {"chunks": 7138.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.42425, "cps": 55.765625, "mean_p": 0.265826, "boundary_entropy": 0.392973, "bpb": 1.063773}, "ta": {"chunks": 7849.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.419845, "cps": 61.320312, "mean_p": 0.288766, "boundary_entropy": 0.482066, "bpb": 0.914769}, "th": {"chunks": 7744.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.481442, "cps": 60.5, "mean_p": 0.375284, "boundary_entropy": 0.559204, "bpb": 1.033386}, "zh": {"chunks": 2835.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.48002, "cps": 22.148438, "mean_p": 0.215499, "boundary_entropy": 0.174367, "bpb": 2.887402}, "my": {"chunks": 8970.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.473401, "cps": 70.078125, "mean_p": 0.336357, "boundary_entropy": 0.540307, "bpb": 1.052711}}, "bpb": {"en": 1.87189, "de": 2.04476, "vi": 1.70024, "tr": 2.01133, "fi": 2.18496, "ru": 1.21328, "el": 1.23321, "hi": 1.06377, "ta": 0.91477, "th": 1.03339, "zh": 2.8874, "my": 1.05271}, "gini_cpc": 0.207828, "gini_cps": 0.216816, "bpic": 5.743505, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9575398489120207, "vi": 1.1994734189462382, "tr": 0.9535541396647303, "fi": 0.8985351295118086, "ru": 2.084138882585357, "el": 1.7312726064185704, "hi": 2.2125311013754114, "ta": 2.189559609596601, "th": 2.510800419817964, "zh": 2.5033843571258885, "my": 2.468862939809908}, "t": 779.2} |
| {"event": "eval", "step": 4000, "phase": 1, "bpb": {"en": 1.87521, "de": 2.04225, "vi": 1.71047, "tr": 2.00304, "fi": 2.20444, "ru": 1.21084, "el": 1.23012, "hi": 1.0729, "ta": 0.91698, "th": 1.03091, "zh": 2.89544, "my": 1.06002}, "bytes_per_chunk": {"en": 4.999, "de": 5.555, "vi": 5.708, "tr": 5.885, "fi": 5.868, "ru": 4.649, "el": 5.505, "hi": 5.816, "ta": 6.529, "th": 5.874, "zh": 5.729, "my": 6.089}, "bpb_mean": 1.60438, "t": 784.5} |
| {"event": "train", "step": 4000, "phase": 1, "objective": "global", "lm_loss": 1.30372, "bpb": 1.88087, "chunk_loss": 0.030154, "F": 0.18153, "G": 0.21705, "chunks": 185.89, "bytes_per_chunk": 5.509, "boundary_entropy": 0.17268, "router_noise": 0.0, "grad_norm": 0.4003, "lr": 0.0002325, "router_trainable": true, "sec_per_step": 0.252, "alpha": 0.03, "t": 784.7} |
| {"event": "train", "step": 4100, "phase": 1, "objective": "global", "lm_loss": 1.27613, "bpb": 1.84106, "chunk_loss": 0.030164, "F": 0.18317, "G": 0.21157, "chunks": 187.56, "bytes_per_chunk": 5.46, "boundary_entropy": 0.14824, "router_noise": 0.0, "grad_norm": 0.3967, "lr": 0.000229, "router_trainable": true, "sec_per_step": 0.168, "alpha": 0.03, "t": 801.5} |
| {"event": "train", "step": 4200, "phase": 1, "objective": "global", "lm_loss": 1.21395, "bpb": 1.75136, "chunk_loss": 0.030257, "F": 0.19028, "G": 0.21286, "chunks": 194.84, "bytes_per_chunk": 5.255, "boundary_entropy": 0.13501, "router_noise": 0.0, "grad_norm": 0.3621, "lr": 0.0002254, "router_trainable": true, "sec_per_step": 0.168, "alpha": 0.03, "t": 818.3} |
| {"event": "train", "step": 4300, "phase": 1, "objective": "global", "lm_loss": 1.18748, "bpb": 1.71317, "chunk_loss": 0.030173, "F": 0.18442, "G": 0.22217, "chunks": 188.84, "bytes_per_chunk": 5.422, "boundary_entropy": 0.17243, "router_noise": 0.0, "grad_norm": 0.3554, "lr": 0.0002217, "router_trainable": true, "sec_per_step": 0.169, "alpha": 0.03, "t": 835.2} |
| {"event": "phase_transition", "step": 4400, "phase": 2, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0, "t": 851.8} |
| {"event": "train", "step": 4400, "phase": 2, "objective": "global", "lm_loss": 1.14356, "bpb": 1.64982, "chunk_loss": 0.030185, "F": 0.18805, "G": 0.21748, "chunks": 192.56, "bytes_per_chunk": 5.318, "boundary_entropy": 0.15467, "router_noise": 0.0, "grad_norm": 0.3429, "lr": 0.0002181, "router_trainable": true, "sec_per_step": 0.168, "alpha": 0.03, "t": 852.0} |
| {"event": "train", "step": 4500, "phase": 2, "objective": "global", "lm_loss": 1.2247, "bpb": 1.76687, "chunk_loss": 0.030204, "F": 0.18398, "G": 0.22505, "chunks": 188.39, "bytes_per_chunk": 5.436, "boundary_entropy": 0.17622, "router_noise": 0.0, "grad_norm": 0.3851, "lr": 0.0002143, "router_trainable": true, "sec_per_step": 0.169, "alpha": 0.03, "t": 868.9} |
| {"event": "train", "step": 4600, "phase": 2, "objective": "global", "lm_loss": 1.16963, "bpb": 1.68741, "chunk_loss": 0.030232, "F": 0.18922, "G": 0.22033, "chunks": 193.77, "bytes_per_chunk": 5.285, "boundary_entropy": 0.1625, "router_noise": 0.0, "grad_norm": 0.3589, "lr": 0.0002105, "router_trainable": true, "sec_per_step": 0.168, "alpha": 0.03, "t": 885.6} |
| {"event": "train", "step": 4700, "phase": 2, "objective": "global", "lm_loss": 1.09733, "bpb": 1.58311, "chunk_loss": 0.030152, "F": 0.18422, "G": 0.23523, "chunks": 188.64, "bytes_per_chunk": 5.428, "boundary_entropy": 0.20401, "router_noise": 0.0, "grad_norm": 0.3487, "lr": 0.0002067, "router_trainable": true, "sec_per_step": 0.168, "alpha": 0.03, "t": 902.4} |
| {"event": "train", "step": 4800, "phase": 2, "objective": "global", "lm_loss": 1.22778, "bpb": 1.77131, "chunk_loss": 0.030207, "F": 0.18752, "G": 0.21093, "chunks": 192.02, "bytes_per_chunk": 5.333, "boundary_entropy": 0.1349, "router_noise": 0.0, "grad_norm": 0.3715, "lr": 0.0002029, "router_trainable": true, "sec_per_step": 0.167, "alpha": 0.03, "t": 919.2} |
| {"event": "train", "step": 4900, "phase": 2, "objective": "global", "lm_loss": 1.2192, "bpb": 1.75893, "chunk_loss": 0.030187, "F": 0.18788, "G": 0.20913, "chunks": 192.39, "bytes_per_chunk": 5.323, "boundary_entropy": 0.12309, "router_noise": 0.0, "grad_norm": 0.4067, "lr": 0.000199, "router_trainable": true, "sec_per_step": 0.168, "alpha": 0.03, "t": 936.0} |
| {"event": "probe", "step": 5000, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3304.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.189852, "cps": 25.8125, "mean_p": 0.190199, "boundary_entropy": 0.074385, "bpb": 1.781115}, "de": {"chunks": 3797.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.184922, "cps": 29.664062, "mean_p": 0.190445, "boundary_entropy": 0.076831, "bpb": 1.940168}, "vi": {"chunks": 4215.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.228641, "cps": 32.929688, "mean_p": 0.190761, "boundary_entropy": 0.09984, "bpb": 1.604959}, "tr": {"chunks": 3292.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.181848, "cps": 25.71875, "mean_p": 0.177143, "boundary_entropy": 0.107755, "bpb": 1.881742}, "fi": {"chunks": 3182.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.169156, "cps": 24.859375, "mean_p": 0.178239, "boundary_entropy": 0.110116, "bpb": 2.065222}, "ru": {"chunks": 7547.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.398259, "cps": 58.960938, "mean_p": 0.211009, "boundary_entropy": 0.088847, "bpb": 1.16986}, "el": {"chunks": 6891.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.33577, "cps": 53.835938, "mean_p": 0.198803, "boundary_entropy": 0.109303, "bpb": 1.184144}, "hi": {"chunks": 6858.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.407608, "cps": 53.578125, "mean_p": 0.284366, "boundary_entropy": 0.414365, "bpb": 1.025808}, "ta": {"chunks": 8618.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.460979, "cps": 67.328125, "mean_p": 0.303692, "boundary_entropy": 0.48219, "bpb": 0.875591}, "th": {"chunks": 7399.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.459994, "cps": 57.804688, "mean_p": 0.364544, "boundary_entropy": 0.547159, "bpb": 0.999693}, "zh": {"chunks": 2813.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.476295, "cps": 21.976562, "mean_p": 0.213603, "boundary_entropy": 0.167518, "bpb": 2.799863}, "my": {"chunks": 9989.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.52718, "cps": 78.039062, "mean_p": 0.344247, "boundary_entropy": 0.547265, "bpb": 1.010641}}, "bpb": {"en": 1.78111, "de": 1.94017, "vi": 1.60496, "tr": 1.88174, "fi": 2.06522, "ru": 1.16986, "el": 1.18414, "hi": 1.02581, "ta": 0.87559, "th": 0.99969, "zh": 2.79986, "my": 1.01064}, "gini_cpc": 0.217029, "gini_cps": 0.233224, "bpic": 5.651734, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 0.974029861456117, "vi": 1.204310626762048, "tr": 0.9578408539031228, "fi": 0.890989063763243, "ru": 2.09772820344094, "el": 1.7685831634713503, "hi": 2.1469725381274984, "ta": 2.4280917972784737, "th": 2.422903088992271, "zh": 2.50876724658881, "my": 2.776787966531059}, "t": 955.7} |
| {"event": "eval", "step": 5000, "phase": 2, "bpb": {"en": 1.78197, "de": 1.94383, "vi": 1.61891, "tr": 1.87568, "fi": 2.08029, "ru": 1.16797, "el": 1.18261, "hi": 1.03286, "ta": 0.87932, "th": 0.99746, "zh": 2.80983, "my": 1.01818}, "bytes_per_chunk": {"en": 5.047, "de": 5.557, "vi": 5.738, "tr": 5.908, "fi": 5.967, "ru": 4.66, "el": 5.449, "hi": 6.192, "ta": 5.886, "th": 6.043, "zh": 5.764, "my": 5.668}, "bpb_mean": 1.53241, "t": 961.0} |
| {"event": "train", "step": 5000, "phase": 2, "objective": "global", "lm_loss": 1.18004, "bpb": 1.70243, "chunk_loss": 0.03021, "F": 0.18739, "G": 0.21322, "chunks": 191.89, "bytes_per_chunk": 5.336, "boundary_entropy": 0.14772, "router_noise": 0.0, "grad_norm": 0.3755, "lr": 0.000195, "router_trainable": true, "sec_per_step": 0.252, "alpha": 0.03, "t": 961.2} |
| {"event": "train", "step": 5100, "phase": 2, "objective": "global", "lm_loss": 1.18943, "bpb": 1.71598, "chunk_loss": 0.030115, "F": 0.18512, "G": 0.21511, "chunks": 189.56, "bytes_per_chunk": 5.402, "boundary_entropy": 0.15129, "router_noise": 0.0, "grad_norm": 0.3684, "lr": 0.0001911, "router_trainable": true, "sec_per_step": 0.166, "alpha": 0.03, "t": 977.8} |
| {"event": "train", "step": 5200, "phase": 2, "objective": "global", "lm_loss": 1.11227, "bpb": 1.60466, "chunk_loss": 0.03017, "F": 0.18506, "G": 0.22977, "chunks": 189.5, "bytes_per_chunk": 5.404, "boundary_entropy": 0.18265, "router_noise": 0.0, "grad_norm": 0.3546, "lr": 0.0001871, "router_trainable": true, "sec_per_step": 0.166, "alpha": 0.03, "t": 994.4} |
| {"event": "train", "step": 5300, "phase": 2, "objective": "global", "lm_loss": 1.14258, "bpb": 1.64839, "chunk_loss": 0.030198, "F": 0.18776, "G": 0.21702, "chunks": 192.27, "bytes_per_chunk": 5.326, "boundary_entropy": 0.15002, "router_noise": 0.0, "grad_norm": 0.3687, "lr": 0.0001831, "router_trainable": true, "sec_per_step": 0.167, "alpha": 0.03, "t": 1011.1} |
| {"event": "train", "step": 5400, "phase": 2, "objective": "global", "lm_loss": 1.20377, "bpb": 1.73668, "chunk_loss": 0.030274, "F": 0.18829, "G": 0.22032, "chunks": 192.81, "bytes_per_chunk": 5.311, "boundary_entropy": 0.14877, "router_noise": 0.0, "grad_norm": 0.3784, "lr": 0.0001791, "router_trainable": true, "sec_per_step": 0.168, "alpha": 0.03, "t": 1027.9} |
| {"event": "train", "step": 5500, "phase": 2, "objective": "global", "lm_loss": 1.15259, "bpb": 1.66284, "chunk_loss": 0.030099, "F": 0.18257, "G": 0.22342, "chunks": 186.95, "bytes_per_chunk": 5.477, "boundary_entropy": 0.17077, "router_noise": 0.0, "grad_norm": 0.3822, "lr": 0.0001751, "router_trainable": true, "sec_per_step": 0.167, "alpha": 0.03, "t": 1044.6} |
| {"event": "train", "step": 5600, "phase": 2, "objective": "global", "lm_loss": 1.10569, "bpb": 1.59517, "chunk_loss": 0.03017, "F": 0.18617, "G": 0.2131, "chunks": 190.64, "bytes_per_chunk": 5.371, "boundary_entropy": 0.14161, "router_noise": 0.0, "grad_norm": 0.3577, "lr": 0.0001711, "router_trainable": true, "sec_per_step": 0.167, "alpha": 0.03, "t": 1061.3} |
| {"event": "train", "step": 5700, "phase": 2, "objective": "global", "lm_loss": 1.10085, "bpb": 1.5882, "chunk_loss": 0.030129, "F": 0.18596, "G": 0.22364, "chunks": 190.42, "bytes_per_chunk": 5.378, "boundary_entropy": 0.16872, "router_noise": 0.0, "grad_norm": 0.3585, "lr": 0.000167, "router_trainable": true, "sec_per_step": 0.167, "alpha": 0.03, "t": 1078.0} |
| {"event": "train", "step": 5800, "phase": 2, "objective": "global", "lm_loss": 1.0854, "bpb": 1.5659, "chunk_loss": 0.030131, "F": 0.18358, "G": 0.22425, "chunks": 187.98, "bytes_per_chunk": 5.447, "boundary_entropy": 0.17778, "router_noise": 0.0, "grad_norm": 0.3532, "lr": 0.000163, "router_trainable": true, "sec_per_step": 0.167, "alpha": 0.03, "t": 1094.7} |
| {"event": "train", "step": 5900, "phase": 2, "objective": "global", "lm_loss": 1.13615, "bpb": 1.63912, "chunk_loss": 0.030225, "F": 0.18861, "G": 0.21444, "chunks": 193.14, "bytes_per_chunk": 5.302, "boundary_entropy": 0.13531, "router_noise": 0.0, "grad_norm": 0.3705, "lr": 0.0001589, "router_trainable": true, "sec_per_step": 0.166, "alpha": 0.03, "t": 1111.3} |
| {"event": "probe", "step": 6000, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3283.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.188646, "cps": 25.648438, "mean_p": 0.189668, "boundary_entropy": 0.067878, "bpb": 1.737111}, "de": {"chunks": 3783.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.18424, "cps": 29.554688, "mean_p": 0.189082, "boundary_entropy": 0.073776, "bpb": 1.884235}, "vi": {"chunks": 4215.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.228641, "cps": 32.929688, "mean_p": 0.186058, "boundary_entropy": 0.082289, "bpb": 1.531174}, "tr": {"chunks": 3361.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.18566, "cps": 26.257812, "mean_p": 0.179506, "boundary_entropy": 0.104724, "bpb": 1.819679}, "fi": {"chunks": 3181.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.169103, "cps": 24.851562, "mean_p": 0.176849, "boundary_entropy": 0.101627, "bpb": 2.003204}, "ru": {"chunks": 7568.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.399367, "cps": 59.125, "mean_p": 0.211984, "boundary_entropy": 0.080976, "bpb": 1.133151}, "el": {"chunks": 6818.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.332213, "cps": 53.265625, "mean_p": 0.196035, "boundary_entropy": 0.099102, "bpb": 1.145773}, "hi": {"chunks": 7195.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.427637, "cps": 56.210938, "mean_p": 0.295136, "boundary_entropy": 0.424076, "bpb": 0.992834}, "ta": {"chunks": 9015.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.482214, "cps": 70.429688, "mean_p": 0.326657, "boundary_entropy": 0.507828, "bpb": 0.857394}, "th": {"chunks": 8320.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.517252, "cps": 65.0, "mean_p": 0.38503, "boundary_entropy": 0.558343, "bpb": 0.967965}, "zh": {"chunks": 2779.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.470538, "cps": 21.710938, "mean_p": 0.206489, "boundary_entropy": 0.145213, "bpb": 2.714054}, "my": {"chunks": 8635.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.455721, "cps": 67.460938, "mean_p": 0.340468, "boundary_entropy": 0.546138, "bpb": 0.978484}}, "bpb": {"en": 1.73711, "de": 1.88424, "vi": 1.53117, "tr": 1.81968, "fi": 2.0032, "ru": 1.13315, "el": 1.14577, "hi": 0.99283, "ta": 0.85739, "th": 0.96797, "zh": 2.71405, "my": 0.97848}, "gini_cpc": 0.216623, "gini_cps": 0.228426, "bpic": 5.631168, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9766459908026186, "vi": 1.2120141062509309, "tr": 0.9841724383885226, "fi": 0.8964065539127832, "ru": 2.1170209048452207, "el": 1.7610406837738968, "hi": 2.2668822548868297, "ta": 2.556192163069137, "th": 2.741924540196381, "zh": 2.494298017916802, "my": 2.415751196454686}, "t": 1130.9} |
| {"event": "eval", "step": 6000, "phase": 2, "bpb": {"en": 1.73727, "de": 1.88868, "vi": 1.54567, "tr": 1.81162, "fi": 2.01888, "ru": 1.13241, "el": 1.14395, "hi": 0.9983, "ta": 0.86107, "th": 0.96524, "zh": 2.72308, "my": 0.98635}, "bytes_per_chunk": {"en": 5.07, "de": 5.566, "vi": 5.747, "tr": 5.82, "fi": 5.983, "ru": 4.65, "el": 5.5, "hi": 5.781, "ta": 5.639, "th": 5.436, "zh": 5.826, "my": 6.47}, "bpb_mean": 1.48438, "t": 1136.2} |
| {"event": "train", "step": 6000, "phase": 2, "objective": "global", "lm_loss": 1.12045, "bpb": 1.61647, "chunk_loss": 0.030268, "F": 0.18727, "G": 0.22486, "chunks": 191.77, "bytes_per_chunk": 5.34, "boundary_entropy": 0.16632, "router_noise": 0.0, "grad_norm": 0.4161, "lr": 0.0001549, "router_trainable": true, "sec_per_step": 0.251, "alpha": 0.03, "t": 1136.3} |
| {"event": "train", "step": 6100, "phase": 2, "objective": "global", "lm_loss": 1.05641, "bpb": 1.52408, "chunk_loss": 0.030264, "F": 0.19098, "G": 0.23003, "chunks": 195.56, "bytes_per_chunk": 5.236, "boundary_entropy": 0.16679, "router_noise": 0.0, "grad_norm": 0.3559, "lr": 0.0001509, "router_trainable": true, "sec_per_step": 0.168, "alpha": 0.03, "t": 1153.1} |
| {"event": "train", "step": 6200, "phase": 2, "objective": "global", "lm_loss": 1.16284, "bpb": 1.67762, "chunk_loss": 0.030166, "F": 0.18858, "G": 0.21059, "chunks": 193.11, "bytes_per_chunk": 5.303, "boundary_entropy": 0.12486, "router_noise": 0.0, "grad_norm": 0.3642, "lr": 0.0001469, "router_trainable": true, "sec_per_step": 0.168, "alpha": 0.03, "t": 1169.9} |
| {"event": "train", "step": 6300, "phase": 2, "objective": "global", "lm_loss": 1.08684, "bpb": 1.56798, "chunk_loss": 0.030202, "F": 0.18921, "G": 0.20602, "chunks": 193.75, "bytes_per_chunk": 5.285, "boundary_entropy": 0.11328, "router_noise": 0.0, "grad_norm": 0.3593, "lr": 0.0001429, "router_trainable": true, "sec_per_step": 0.167, "alpha": 0.03, "t": 1186.6} |
| {"event": "train", "step": 6400, "phase": 2, "objective": "global", "lm_loss": 0.99837, "bpb": 1.44035, "chunk_loss": 0.030229, "F": 0.18727, "G": 0.23142, "chunks": 191.77, "bytes_per_chunk": 5.34, "boundary_entropy": 0.18764, "router_noise": 0.0, "grad_norm": 0.3502, "lr": 0.0001389, "router_trainable": true, "sec_per_step": 0.167, "alpha": 0.03, "t": 1203.3} |
| {"event": "train", "step": 6500, "phase": 2, "objective": "global", "lm_loss": 1.06207, "bpb": 1.53225, "chunk_loss": 0.030073, "F": 0.17761, "G": 0.21894, "chunks": 181.88, "bytes_per_chunk": 5.63, "boundary_entropy": 0.174, "router_noise": 0.0, "grad_norm": 0.3711, "lr": 0.000135, "router_trainable": true, "sec_per_step": 0.167, "alpha": 0.03, "t": 1220.0} |
| {"event": "train", "step": 6600, "phase": 2, "objective": "global", "lm_loss": 1.1142, "bpb": 1.60745, "chunk_loss": 0.030231, "F": 0.19049, "G": 0.21705, "chunks": 195.06, "bytes_per_chunk": 5.25, "boundary_entropy": 0.1373, "router_noise": 0.0, "grad_norm": 0.3618, "lr": 0.000131, "router_trainable": true, "sec_per_step": 0.167, "alpha": 0.03, "t": 1236.7} |
| {"event": "train", "step": 6700, "phase": 2, "objective": "global", "lm_loss": 1.10026, "bpb": 1.58734, "chunk_loss": 0.030094, "F": 0.1823, "G": 0.21441, "chunks": 186.67, "bytes_per_chunk": 5.486, "boundary_entropy": 0.14955, "router_noise": 0.0, "grad_norm": 0.3692, "lr": 0.0001271, "router_trainable": true, "sec_per_step": 0.167, "alpha": 0.03, "t": 1253.4} |
| {"event": "train", "step": 6800, "phase": 2, "objective": "global", "lm_loss": 1.10897, "bpb": 1.59991, "chunk_loss": 0.030201, "F": 0.18614, "G": 0.22581, "chunks": 190.61, "bytes_per_chunk": 5.372, "boundary_entropy": 0.16757, "router_noise": 0.0, "grad_norm": 0.3761, "lr": 0.0001233, "router_trainable": true, "sec_per_step": 0.168, "alpha": 0.03, "t": 1270.2} |
| {"event": "train", "step": 6900, "phase": 2, "objective": "global", "lm_loss": 1.0897, "bpb": 1.57211, "chunk_loss": 0.030176, "F": 0.18617, "G": 0.21543, "chunks": 190.64, "bytes_per_chunk": 5.371, "boundary_entropy": 0.1435, "router_noise": 0.0, "grad_norm": 0.3594, "lr": 0.0001195, "router_trainable": true, "sec_per_step": 0.167, "alpha": 0.03, "t": 1286.9} |
| {"event": "probe", "step": 7000, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3271.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.187956, "cps": 25.554688, "mean_p": 0.188398, "boundary_entropy": 0.058709, "bpb": 1.701339}, "de": {"chunks": 3733.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.181805, "cps": 29.164062, "mean_p": 0.18572, "boundary_entropy": 0.062294, "bpb": 1.84134}, "vi": {"chunks": 4209.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.228316, "cps": 32.882812, "mean_p": 0.183183, "boundary_entropy": 0.071121, "bpb": 1.484868}, "tr": {"chunks": 3230.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.178423, "cps": 25.234375, "mean_p": 0.17346, "boundary_entropy": 0.0938, "bpb": 1.7738}, "fi": {"chunks": 3189.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.169528, "cps": 24.914062, "mean_p": 0.175968, "boundary_entropy": 0.092518, "bpb": 1.949012}, "ru": {"chunks": 7577.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.399842, "cps": 59.195312, "mean_p": 0.208459, "boundary_entropy": 0.08557, "bpb": 1.107937}, "el": {"chunks": 6864.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.334454, "cps": 53.625, "mean_p": 0.193895, "boundary_entropy": 0.093824, "bpb": 1.108357}, "hi": {"chunks": 6747.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.40101, "cps": 52.710938, "mean_p": 0.276852, "boundary_entropy": 0.403365, "bpb": 0.966785}, "ta": {"chunks": 8186.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.437871, "cps": 63.953125, "mean_p": 0.30228, "boundary_entropy": 0.474983, "bpb": 0.835746}, "th": {"chunks": 7661.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.476282, "cps": 59.851562, "mean_p": 0.373482, "boundary_entropy": 0.548838, "bpb": 0.938659}, "zh": {"chunks": 2810.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.475787, "cps": 21.953125, "mean_p": 0.210036, "boundary_entropy": 0.150879, "bpb": 2.632629}, "my": {"chunks": 9341.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.492981, "cps": 72.976562, "mean_p": 0.333855, "boundary_entropy": 0.531147, "bpb": 0.943829}}, "bpb": {"en": 1.70134, "de": 1.84134, "vi": 1.48487, "tr": 1.7738, "fi": 1.94901, "ru": 1.10794, "el": 1.10836, "hi": 0.96679, "ta": 0.83575, "th": 0.93866, "zh": 2.63263, "my": 0.94383}, "gini_cpc": 0.212937, "gini_cps": 0.226102, "bpic": 5.743677, "worst_lang": "my", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9672732084327481, "vi": 1.2147288883073606, "tr": 0.9492826579007216, "fi": 0.9019577859738201, "ru": 2.127314248747085, "el": 1.7794263024060828, "hi": 2.1335322567676775, "ta": 2.329645537641077, "th": 2.534007950540193, "zh": 2.5313748057175793, "my": 2.622850712374603}, "t": 1306.5} |
| {"event": "eval", "step": 7000, "phase": 2, "bpb": {"en": 1.70081, "de": 1.84686, "vi": 1.49842, "tr": 1.76602, "fi": 1.96461, "ru": 1.10689, "el": 1.10658, "hi": 0.97341, "ta": 0.83961, "th": 0.93653, "zh": 2.64028, "my": 0.95142}, "bytes_per_chunk": {"en": 5.083, "de": 5.58, "vi": 5.731, "tr": 6.004, "fi": 5.921, "ru": 4.647, "el": 5.475, "hi": 6.132, "ta": 6.215, "th": 6.115, "zh": 5.746, "my": 5.894}, "bpb_mean": 1.44429, "t": 1311.8} |
| {"event": "train", "step": 7000, "phase": 2, "objective": "global", "lm_loss": 1.07909, "bpb": 1.55679, "chunk_loss": 0.030172, "F": 0.18677, "G": 0.21751, "chunks": 191.25, "bytes_per_chunk": 5.354, "boundary_entropy": 0.14429, "router_noise": 0.0, "grad_norm": 0.3634, "lr": 0.0001157, "router_trainable": true, "sec_per_step": 0.251, "alpha": 0.03, "t": 1312.0} |
| {"event": "train", "step": 7100, "phase": 2, "objective": "global", "lm_loss": 1.06154, "bpb": 1.53147, "chunk_loss": 0.030114, "F": 0.18227, "G": 0.21941, "chunks": 186.64, "bytes_per_chunk": 5.486, "boundary_entropy": 0.16154, "router_noise": 0.0, "grad_norm": 0.367, "lr": 0.0001119, "router_trainable": true, "sec_per_step": 0.166, "alpha": 0.03, "t": 1328.6} |
| {"event": "train", "step": 7200, "phase": 2, "objective": "global", "lm_loss": 1.11109, "bpb": 1.60296, "chunk_loss": 0.030155, "F": 0.18565, "G": 0.20645, "chunks": 190.11, "bytes_per_chunk": 5.386, "boundary_entropy": 0.12576, "router_noise": 0.0, "grad_norm": 0.3498, "lr": 0.0001083, "router_trainable": true, "sec_per_step": 0.167, "alpha": 0.03, "t": 1345.3} |
| {"event": "train", "step": 7300, "phase": 2, "objective": "global", "lm_loss": 1.12501, "bpb": 1.62305, "chunk_loss": 0.030215, "F": 0.18713, "G": 0.21904, "chunks": 191.62, "bytes_per_chunk": 5.344, "boundary_entropy": 0.15205, "router_noise": 0.0, "grad_norm": 0.362, "lr": 0.0001046, "router_trainable": true, "sec_per_step": 0.166, "alpha": 0.03, "t": 1361.9} |
| {"event": "train", "step": 7400, "phase": 2, "objective": "global", "lm_loss": 1.09056, "bpb": 1.57335, "chunk_loss": 0.030206, "F": 0.18951, "G": 0.2189, "chunks": 194.06, "bytes_per_chunk": 5.277, "boundary_entropy": 0.14802, "router_noise": 0.0, "grad_norm": 0.3727, "lr": 0.000101, "router_trainable": true, "sec_per_step": 0.167, "alpha": 0.03, "t": 1378.6} |
| {"event": "train", "step": 7500, "phase": 2, "objective": "global", "lm_loss": 1.07186, "bpb": 1.54637, "chunk_loss": 0.030115, "F": 0.18616, "G": 0.2171, "chunks": 190.62, "bytes_per_chunk": 5.372, "boundary_entropy": 0.14576, "router_noise": 0.0, "grad_norm": 0.3536, "lr": 9.75e-05, "router_trainable": true, "sec_per_step": 0.167, "alpha": 0.03, "t": 1395.3} |
| {"event": "train", "step": 7600, "phase": 2, "objective": "global", "lm_loss": 1.13271, "bpb": 1.63415, "chunk_loss": 0.030154, "F": 0.18687, "G": 0.21747, "chunks": 191.36, "bytes_per_chunk": 5.351, "boundary_entropy": 0.13926, "router_noise": 0.0, "grad_norm": 0.3826, "lr": 9.4e-05, "router_trainable": true, "sec_per_step": 0.167, "alpha": 0.03, "t": 1412.0} |
| {"event": "phase_transition", "step": 7700, "phase": 3, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0, "t": 1428.6} |
| {"event": "train", "step": 7700, "phase": 3, "objective": "global", "lm_loss": 1.08036, "bpb": 1.55863, "chunk_loss": 0.030172, "F": 0.18755, "G": 0.21101, "chunks": 192.05, "bytes_per_chunk": 5.332, "boundary_entropy": 0.12882, "router_noise": 0.0, "grad_norm": 0.3606, "lr": 9.06e-05, "router_trainable": true, "sec_per_step": 0.167, "alpha": 0.03, "t": 1428.7} |
| {"event": "train", "step": 7800, "phase": 3, "objective": "global", "lm_loss": 1.11356, "bpb": 1.60653, "chunk_loss": 0.030097, "F": 0.18399, "G": 0.20504, "chunks": 188.41, "bytes_per_chunk": 5.435, "boundary_entropy": 0.11583, "router_noise": 0.0, "grad_norm": 0.3671, "lr": 8.73e-05, "router_trainable": true, "sec_per_step": 0.167, "alpha": 0.03, "t": 1445.5} |
| {"event": "train", "step": 7900, "phase": 3, "objective": "global", "lm_loss": 1.10399, "bpb": 1.59272, "chunk_loss": 0.030164, "F": 0.18573, "G": 0.20999, "chunks": 190.19, "bytes_per_chunk": 5.384, "boundary_entropy": 0.12617, "router_noise": 0.0, "grad_norm": 0.3997, "lr": 8.4e-05, "router_trainable": true, "sec_per_step": 0.167, "alpha": 0.03, "t": 1462.1} |
| {"event": "probe", "step": 8000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3255.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.187037, "cps": 25.429688, "mean_p": 0.186699, "boundary_entropy": 0.060507, "bpb": 1.67497}, "de": {"chunks": 3721.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.18122, "cps": 29.070312, "mean_p": 0.185632, "boundary_entropy": 0.063243, "bpb": 1.80855}, "vi": {"chunks": 4196.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.227611, "cps": 32.78125, "mean_p": 0.182728, "boundary_entropy": 0.069279, "bpb": 1.449845}, "tr": {"chunks": 3329.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.183892, "cps": 26.007812, "mean_p": 0.176527, "boundary_entropy": 0.09774, "bpb": 1.740903}, "fi": {"chunks": 3194.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.169794, "cps": 24.953125, "mean_p": 0.17522, "boundary_entropy": 0.091283, "bpb": 1.909283}, "ru": {"chunks": 7568.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.399367, "cps": 59.125, "mean_p": 0.205121, "boundary_entropy": 0.09028, "bpb": 1.081727}, "el": {"chunks": 6875.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.33499, "cps": 53.710938, "mean_p": 0.194552, "boundary_entropy": 0.09301, "bpb": 1.083698}, "hi": {"chunks": 6728.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.399881, "cps": 52.5625, "mean_p": 0.279541, "boundary_entropy": 0.404925, "bpb": 0.946025}, "ta": {"chunks": 8054.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.43081, "cps": 62.921875, "mean_p": 0.306989, "boundary_entropy": 0.482116, "bpb": 0.815779}, "th": {"chunks": 7245.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.45042, "cps": 56.601562, "mean_p": 0.373491, "boundary_entropy": 0.553497, "bpb": 0.919326}, "zh": {"chunks": 2771.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.469184, "cps": 21.648438, "mean_p": 0.206052, "boundary_entropy": 0.1418, "bpb": 2.573069}, "my": {"chunks": 8522.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.449757, "cps": 66.578125, "mean_p": 0.336278, "boundary_entropy": 0.540838, "bpb": 0.917724}}, "bpb": {"en": 1.67497, "de": 1.80855, "vi": 1.44984, "tr": 1.7409, "fi": 1.90928, "ru": 1.08173, "el": 1.0837, "hi": 0.94603, "ta": 0.81578, "th": 0.91933, "zh": 2.57307, "my": 0.91772}, "gini_cpc": 0.201195, "gini_cps": 0.214853, "bpic": 5.863011, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9689031997721551, "vi": 1.2169296281992152, "tr": 0.9831875512142372, "fi": 0.907812494258311, "ru": 2.1352318373600183, "el": 1.7910387603792173, "hi": 2.1379819639690223, "ta": 2.303346541560313, "th": 2.408188354206985, "zh": 2.5085121590010004, "my": 2.404646721709379}, "t": 1481.6} |
| {"event": "eval", "step": 8000, "phase": 3, "bpb": {"en": 1.67436, "de": 1.81413, "vi": 1.46457, "tr": 1.73328, "fi": 1.92525, "ru": 1.08068, "el": 1.08194, "hi": 0.95198, "ta": 0.81986, "th": 0.91822, "zh": 2.57814, "my": 0.92517}, "bytes_per_chunk": {"en": 5.092, "de": 5.6, "vi": 5.737, "tr": 5.86, "fi": 5.933, "ru": 4.659, "el": 5.455, "hi": 6.119, "ta": 6.127, "th": 6.33, "zh": 5.833, "my": 6.354}, "bpb_mean": 1.41397, "t": 1486.9} |
| {"event": "train", "step": 8000, "phase": 3, "objective": "global", "lm_loss": 1.09726, "bpb": 1.58301, "chunk_loss": 0.030066, "F": 0.18384, "G": 0.2134, "chunks": 188.25, "bytes_per_chunk": 5.44, "boundary_entropy": 0.13755, "router_noise": 0.0, "grad_norm": 0.379, "lr": 8.08e-05, "router_trainable": true, "sec_per_step": 0.249, "alpha": 0.03, "t": 1487.0} |
| {"event": "train", "step": 8100, "phase": 3, "objective": "global", "lm_loss": 1.11911, "bpb": 1.61454, "chunk_loss": 0.030163, "F": 0.18579, "G": 0.20851, "chunks": 190.25, "bytes_per_chunk": 5.382, "boundary_entropy": 0.12629, "router_noise": 0.0, "grad_norm": 0.3874, "lr": 7.77e-05, "router_trainable": true, "sec_per_step": 0.167, "alpha": 0.03, "t": 1503.8} |
| {"event": "train", "step": 8200, "phase": 3, "objective": "global", "lm_loss": 1.07345, "bpb": 1.54866, "chunk_loss": 0.030102, "F": 0.18584, "G": 0.20556, "chunks": 190.3, "bytes_per_chunk": 5.381, "boundary_entropy": 0.12161, "router_noise": 0.0, "grad_norm": 0.3878, "lr": 7.47e-05, "router_trainable": true, "sec_per_step": 0.167, "alpha": 0.03, "t": 1520.5} |
| {"event": "train", "step": 8300, "phase": 3, "objective": "global", "lm_loss": 1.00966, "bpb": 1.45664, "chunk_loss": 0.030129, "F": 0.18858, "G": 0.22209, "chunks": 193.11, "bytes_per_chunk": 5.303, "boundary_entropy": 0.1565, "router_noise": 0.0, "grad_norm": 0.3576, "lr": 7.17e-05, "router_trainable": true, "sec_per_step": 0.166, "alpha": 0.03, "t": 1537.1} |
| {"event": "train", "step": 8400, "phase": 3, "objective": "global", "lm_loss": 1.05767, "bpb": 1.5259, "chunk_loss": 0.030172, "F": 0.18854, "G": 0.21573, "chunks": 193.06, "bytes_per_chunk": 5.304, "boundary_entropy": 0.13693, "router_noise": 0.0, "grad_norm": 0.3778, "lr": 6.88e-05, "router_trainable": true, "sec_per_step": 0.166, "alpha": 0.03, "t": 1553.7} |
| {"event": "train", "step": 8500, "phase": 3, "objective": "global", "lm_loss": 1.07023, "bpb": 1.54401, "chunk_loss": 0.030132, "F": 0.18408, "G": 0.20712, "chunks": 188.5, "bytes_per_chunk": 5.432, "boundary_entropy": 0.12635, "router_noise": 0.0, "grad_norm": 0.3707, "lr": 6.6e-05, "router_trainable": true, "sec_per_step": 0.169, "alpha": 0.03, "t": 1570.6} |
| {"event": "train", "step": 8600, "phase": 3, "objective": "global", "lm_loss": 1.12456, "bpb": 1.6224, "chunk_loss": 0.030099, "F": 0.18066, "G": 0.20825, "chunks": 185.0, "bytes_per_chunk": 5.535, "boundary_entropy": 0.13665, "router_noise": 0.0, "grad_norm": 0.3856, "lr": 6.33e-05, "router_trainable": true, "sec_per_step": 0.167, "alpha": 0.03, "t": 1587.3} |
| {"event": "train", "step": 8700, "phase": 3, "objective": "global", "lm_loss": 0.99409, "bpb": 1.43416, "chunk_loss": 0.030152, "F": 0.18695, "G": 0.22071, "chunks": 191.44, "bytes_per_chunk": 5.349, "boundary_entropy": 0.15918, "router_noise": 0.0, "grad_norm": 0.3783, "lr": 6.07e-05, "router_trainable": true, "sec_per_step": 0.166, "alpha": 0.03, "t": 1603.9} |
| {"event": "train", "step": 8800, "phase": 3, "objective": "global", "lm_loss": 1.01632, "bpb": 1.46624, "chunk_loss": 0.030224, "F": 0.18839, "G": 0.21525, "chunks": 192.91, "bytes_per_chunk": 5.308, "boundary_entropy": 0.1431, "router_noise": 0.0, "grad_norm": 0.3742, "lr": 5.82e-05, "router_trainable": true, "sec_per_step": 0.168, "alpha": 0.03, "t": 1620.7} |
| {"event": "train", "step": 8900, "phase": 3, "objective": "global", "lm_loss": 1.07171, "bpb": 1.54615, "chunk_loss": 0.030181, "F": 0.18774, "G": 0.20238, "chunks": 192.25, "bytes_per_chunk": 5.326, "boundary_entropy": 0.10475, "router_noise": 0.0, "grad_norm": 0.3726, "lr": 5.58e-05, "router_trainable": true, "sec_per_step": 0.167, "alpha": 0.03, "t": 1637.3} |
| {"event": "probe", "step": 9000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3271.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.187956, "cps": 25.554688, "mean_p": 0.186801, "boundary_entropy": 0.055522, "bpb": 1.65275}, "de": {"chunks": 3739.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.182097, "cps": 29.210938, "mean_p": 0.185158, "boundary_entropy": 0.056531, "bpb": 1.784824}, "vi": {"chunks": 4195.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.227556, "cps": 32.773438, "mean_p": 0.181302, "boundary_entropy": 0.063076, "bpb": 1.422078}, "tr": {"chunks": 3211.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.177374, "cps": 25.085938, "mean_p": 0.17125, "boundary_entropy": 0.089427, "bpb": 1.713831}, "fi": {"chunks": 3186.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.169369, "cps": 24.890625, "mean_p": 0.172709, "boundary_entropy": 0.081621, "bpb": 1.878038}, "ru": {"chunks": 7569.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.39942, "cps": 59.132812, "mean_p": 0.205731, "boundary_entropy": 0.086538, "bpb": 1.063334}, "el": {"chunks": 6855.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.334015, "cps": 53.554688, "mean_p": 0.192483, "boundary_entropy": 0.086683, "bpb": 1.062895}, "hi": {"chunks": 6941.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.412541, "cps": 54.226562, "mean_p": 0.284051, "boundary_entropy": 0.406381, "bpb": 0.927486}, "ta": {"chunks": 8075.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.431934, "cps": 63.085938, "mean_p": 0.30378, "boundary_entropy": 0.479358, "bpb": 0.802603}, "th": {"chunks": 7693.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.478272, "cps": 60.101562, "mean_p": 0.379829, "boundary_entropy": 0.55638, "bpb": 0.901192}, "zh": {"chunks": 2781.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.470877, "cps": 21.726562, "mean_p": 0.206009, "boundary_entropy": 0.138749, "bpb": 2.541754}, "my": {"chunks": 8588.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.45324, "cps": 67.09375, "mean_p": 0.326554, "boundary_entropy": 0.524339, "bpb": 0.895038}}, "bpb": {"en": 1.65275, "de": 1.78482, "vi": 1.42208, "tr": 1.71383, "fi": 1.87804, "ru": 1.06333, "el": 1.0629, "hi": 0.92749, "ta": 0.8026, "th": 0.90119, "zh": 2.54175, "my": 0.89504}, "gini_cpc": 0.207613, "gini_cps": 0.218101, "bpic": 5.805715, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9688278934717507, "vi": 1.2106884500948865, "tr": 0.9436986422660115, "fi": 0.9011092838233274, "ru": 2.1250681732567887, "el": 1.7770931385480329, "hi": 2.1948788193603748, "ta": 2.2980561588628996, "th": 2.544592502741901, "zh": 2.5052502970464725, "my": 2.4114165418984146}, "t": 1657.0} |
| {"event": "eval", "step": 9000, "phase": 3, "bpb": {"en": 1.65182, "de": 1.78942, "vi": 1.43673, "tr": 1.7071, "fi": 1.89331, "ru": 1.06155, "el": 1.06158, "hi": 0.9332, "ta": 0.80751, "th": 0.90057, "zh": 2.54751, "my": 0.9039}, "bytes_per_chunk": {"en": 5.084, "de": 5.581, "vi": 5.745, "tr": 6.024, "fi": 5.95, "ru": 4.658, "el": 5.478, "hi": 5.954, "ta": 6.056, "th": 5.95, "zh": 5.803, "my": 6.33}, "bpb_mean": 1.39118, "t": 1662.2} |
| {"event": "train", "step": 9000, "phase": 3, "objective": "global", "lm_loss": 1.0604, "bpb": 1.52984, "chunk_loss": 0.030144, "F": 0.18185, "G": 0.2122, "chunks": 186.22, "bytes_per_chunk": 5.499, "boundary_entropy": 0.14394, "router_noise": 0.0, "grad_norm": 0.3933, "lr": 5.35e-05, "router_trainable": true, "sec_per_step": 0.251, "alpha": 0.03, "t": 1662.4} |
| {"event": "train", "step": 9100, "phase": 3, "objective": "global", "lm_loss": 1.0464, "bpb": 1.50964, "chunk_loss": 0.030159, "F": 0.18748, "G": 0.21393, "chunks": 191.98, "bytes_per_chunk": 5.334, "boundary_entropy": 0.13779, "router_noise": 0.0, "grad_norm": 0.3704, "lr": 5.12e-05, "router_trainable": true, "sec_per_step": 0.167, "alpha": 0.03, "t": 1679.1} |
| {"event": "train", "step": 9200, "phase": 3, "objective": "global", "lm_loss": 0.97228, "bpb": 1.4027, "chunk_loss": 0.030115, "F": 0.18478, "G": 0.21009, "chunks": 189.22, "bytes_per_chunk": 5.412, "boundary_entropy": 0.13817, "router_noise": 0.0, "grad_norm": 0.3639, "lr": 4.91e-05, "router_trainable": true, "sec_per_step": 0.167, "alpha": 0.03, "t": 1695.8} |
| {"event": "train", "step": 9300, "phase": 3, "objective": "global", "lm_loss": 1.04658, "bpb": 1.50989, "chunk_loss": 0.03015, "F": 0.1882, "G": 0.21321, "chunks": 192.72, "bytes_per_chunk": 5.313, "boundary_entropy": 0.12758, "router_noise": 0.0, "grad_norm": 0.3789, "lr": 4.71e-05, "router_trainable": true, "sec_per_step": 0.166, "alpha": 0.03, "t": 1712.4} |
| {"event": "train", "step": 9400, "phase": 3, "objective": "global", "lm_loss": 1.09135, "bpb": 1.57449, "chunk_loss": 0.030138, "F": 0.18335, "G": 0.21331, "chunks": 187.75, "bytes_per_chunk": 5.454, "boundary_entropy": 0.13385, "router_noise": 0.0, "grad_norm": 0.3794, "lr": 4.52e-05, "router_trainable": true, "sec_per_step": 0.167, "alpha": 0.03, "t": 1729.1} |
| {"event": "train", "step": 9500, "phase": 3, "objective": "global", "lm_loss": 1.10241, "bpb": 1.59044, "chunk_loss": 0.030128, "F": 0.18442, "G": 0.20437, "chunks": 188.84, "bytes_per_chunk": 5.422, "boundary_entropy": 0.11903, "router_noise": 0.0, "grad_norm": 0.388, "lr": 4.34e-05, "router_trainable": true, "sec_per_step": 0.166, "alpha": 0.03, "t": 1745.8} |
| {"event": "train", "step": 9600, "phase": 3, "objective": "global", "lm_loss": 0.9623, "bpb": 1.3883, "chunk_loss": 0.030135, "F": 0.18536, "G": 0.22429, "chunks": 189.81, "bytes_per_chunk": 5.395, "boundary_entropy": 0.1668, "router_noise": 0.0, "grad_norm": 0.3869, "lr": 4.17e-05, "router_trainable": true, "sec_per_step": 0.167, "alpha": 0.03, "t": 1762.4} |
| {"event": "train", "step": 9700, "phase": 3, "objective": "global", "lm_loss": 0.97344, "bpb": 1.40438, "chunk_loss": 0.03013, "F": 0.18614, "G": 0.21162, "chunks": 190.61, "bytes_per_chunk": 5.372, "boundary_entropy": 0.13463, "router_noise": 0.0, "grad_norm": 0.3641, "lr": 4.01e-05, "router_trainable": true, "sec_per_step": 0.168, "alpha": 0.03, "t": 1779.2} |
| {"event": "train", "step": 9800, "phase": 3, "objective": "global", "lm_loss": 0.99836, "bpb": 1.44032, "chunk_loss": 0.030244, "F": 0.1869, "G": 0.2271, "chunks": 191.39, "bytes_per_chunk": 5.35, "boundary_entropy": 0.17221, "router_noise": 0.0, "grad_norm": 0.3785, "lr": 3.86e-05, "router_trainable": true, "sec_per_step": 0.168, "alpha": 0.03, "t": 1796.0} |
| {"event": "train", "step": 9900, "phase": 3, "objective": "global", "lm_loss": 1.06844, "bpb": 1.54144, "chunk_loss": 0.030125, "F": 0.18338, "G": 0.21556, "chunks": 187.78, "bytes_per_chunk": 5.453, "boundary_entropy": 0.13751, "router_noise": 0.0, "grad_norm": 0.3766, "lr": 3.72e-05, "router_trainable": true, "sec_per_step": 0.167, "alpha": 0.03, "t": 1812.7} |
| {"event": "probe", "step": 10000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3268.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.187784, "cps": 25.53125, "mean_p": 0.1867, "boundary_entropy": 0.0541, "bpb": 1.639254}, "de": {"chunks": 3729.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.18161, "cps": 29.132812, "mean_p": 0.184207, "boundary_entropy": 0.053234, "bpb": 1.767738}, "vi": {"chunks": 4193.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.227448, "cps": 32.757812, "mean_p": 0.18019, "boundary_entropy": 0.059113, "bpb": 1.406151}, "tr": {"chunks": 3252.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.179639, "cps": 25.40625, "mean_p": 0.172283, "boundary_entropy": 0.086711, "bpb": 1.697912}, "fi": {"chunks": 3197.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.169954, "cps": 24.976562, "mean_p": 0.172761, "boundary_entropy": 0.078296, "bpb": 1.855945}, "ru": {"chunks": 7562.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.39905, "cps": 59.078125, "mean_p": 0.201832, "boundary_entropy": 0.09324, "bpb": 1.04949}, "el": {"chunks": 6842.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.333382, "cps": 53.453125, "mean_p": 0.191613, "boundary_entropy": 0.086456, "bpb": 1.047315}, "hi": {"chunks": 7110.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.422585, "cps": 55.546875, "mean_p": 0.279613, "boundary_entropy": 0.395377, "bpb": 0.914562}, "ta": {"chunks": 8036.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.429848, "cps": 62.78125, "mean_p": 0.307074, "boundary_entropy": 0.477674, "bpb": 0.79101}, "th": {"chunks": 7354.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.457196, "cps": 57.453125, "mean_p": 0.37574, "boundary_entropy": 0.550546, "bpb": 0.892274}, "zh": {"chunks": 2770.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.469015, "cps": 21.640625, "mean_p": 0.204363, "boundary_entropy": 0.13423, "bpb": 2.519625}, "my": {"chunks": 8696.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.45894, "cps": 67.9375, "mean_p": 0.33193, "boundary_entropy": 0.531966, "bpb": 0.882503}}, "bpb": {"en": 1.63925, "de": 1.76774, "vi": 1.40615, "tr": 1.69791, "fi": 1.85594, "ru": 1.04949, "el": 1.04732, "hi": 0.91456, "ta": 0.79101, "th": 0.89227, "zh": 2.51963, "my": 0.8825}, "gini_cpc": 0.204136, "gini_cps": 0.217305, "bpic": 5.814071, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9671237499821912, "vi": 1.2112221178715517, "tr": 0.9566257301903138, "fi": 0.9050505261153807, "ru": 2.1250518500337483, "el": 1.7753512776750875, "hi": 2.250383838236872, "ta": 2.289056597844088, "th": 2.4346953854770157, "zh": 2.4976317053669463, "my": 2.443983262573418}, "t": 1832.3} |
| {"event": "eval", "step": 10000, "phase": 3, "bpb": {"en": 1.63794, "de": 1.77287, "vi": 1.42126, "tr": 1.69086, "fi": 1.87273, "ru": 1.0476, "el": 1.04626, "hi": 0.9209, "ta": 0.79586, "th": 0.89057, "zh": 2.52571, "my": 0.89087}, "bytes_per_chunk": {"en": 5.082, "de": 5.591, "vi": 5.746, "tr": 5.914, "fi": 5.938, "ru": 4.662, "el": 5.482, "hi": 5.833, "ta": 6.125, "th": 6.174, "zh": 5.826, "my": 6.255}, "bpb_mean": 1.37612, "t": 1837.5} |
| {"event": "train", "step": 10000, "phase": 3, "objective": "global", "lm_loss": 1.04002, "bpb": 1.50043, "chunk_loss": 0.030165, "F": 0.18626, "G": 0.2167, "chunks": 190.73, "bytes_per_chunk": 5.369, "boundary_entropy": 0.14899, "router_noise": 0.0, "grad_norm": 0.4001, "lr": 3.6e-05, "router_trainable": true, "sec_per_step": 0.25, "alpha": 0.03, "t": 1837.7} |
| {"event": "train", "step": 10100, "phase": 3, "objective": "global", "lm_loss": 0.99145, "bpb": 1.43036, "chunk_loss": 0.0301, "F": 0.18333, "G": 0.23431, "chunks": 187.73, "bytes_per_chunk": 5.455, "boundary_entropy": 0.19565, "router_noise": 0.0, "grad_norm": 0.3809, "lr": 3.49e-05, "router_trainable": true, "sec_per_step": 0.169, "alpha": 0.03, "t": 1854.6} |
| {"event": "train", "step": 10200, "phase": 3, "objective": "global", "lm_loss": 1.08077, "bpb": 1.55922, "chunk_loss": 0.030163, "F": 0.18858, "G": 0.20656, "chunks": 193.11, "bytes_per_chunk": 5.303, "boundary_entropy": 0.10936, "router_noise": 0.0, "grad_norm": 0.4062, "lr": 3.38e-05, "router_trainable": true, "sec_per_step": 0.166, "alpha": 0.03, "t": 1871.3} |
| {"event": "train", "step": 10300, "phase": 3, "objective": "global", "lm_loss": 1.03964, "bpb": 1.49988, "chunk_loss": 0.03046, "F": 0.18994, "G": 0.21659, "chunks": 194.5, "bytes_per_chunk": 5.265, "boundary_entropy": 0.14905, "router_noise": 0.0, "grad_norm": 0.4174, "lr": 3.3e-05, "router_trainable": true, "sec_per_step": 0.168, "alpha": 0.03, "t": 1888.1} |
| {"event": "train", "step": 10400, "phase": 3, "objective": "global", "lm_loss": 1.00832, "bpb": 1.4547, "chunk_loss": 0.030166, "F": 0.18529, "G": 0.20888, "chunks": 189.73, "bytes_per_chunk": 5.397, "boundary_entropy": 0.12976, "router_noise": 0.0, "grad_norm": 0.3961, "lr": 3.22e-05, "router_trainable": true, "sec_per_step": 0.167, "alpha": 0.03, "t": 1904.8} |
| {"event": "train", "step": 10500, "phase": 3, "objective": "global", "lm_loss": 1.0353, "bpb": 1.49362, "chunk_loss": 0.030203, "F": 0.18874, "G": 0.21414, "chunks": 193.27, "bytes_per_chunk": 5.298, "boundary_entropy": 0.13485, "router_noise": 0.0, "grad_norm": 0.399, "lr": 3.15e-05, "router_trainable": true, "sec_per_step": 0.167, "alpha": 0.03, "t": 1921.5} |
| {"event": "train", "step": 10600, "phase": 3, "objective": "global", "lm_loss": 1.05892, "bpb": 1.52769, "chunk_loss": 0.030152, "F": 0.18628, "G": 0.21155, "chunks": 190.75, "bytes_per_chunk": 5.368, "boundary_entropy": 0.13434, "router_noise": 0.0, "grad_norm": 0.3884, "lr": 3.1e-05, "router_trainable": true, "sec_per_step": 0.166, "alpha": 0.03, "t": 1938.1} |
| {"event": "train", "step": 10700, "phase": 3, "objective": "global", "lm_loss": 1.02694, "bpb": 1.48157, "chunk_loss": 0.030173, "F": 0.18689, "G": 0.22266, "chunks": 191.38, "bytes_per_chunk": 5.351, "boundary_entropy": 0.15521, "router_noise": 0.0, "grad_norm": 0.3941, "lr": 3.05e-05, "router_trainable": true, "sec_per_step": 0.165, "alpha": 0.03, "t": 1954.6} |
| {"event": "train", "step": 10800, "phase": 3, "objective": "global", "lm_loss": 1.04347, "bpb": 1.50541, "chunk_loss": 0.030156, "F": 0.18739, "G": 0.20979, "chunks": 191.89, "bytes_per_chunk": 5.336, "boundary_entropy": 0.13046, "router_noise": 0.0, "grad_norm": 0.3883, "lr": 3.02e-05, "router_trainable": true, "sec_per_step": 0.167, "alpha": 0.03, "t": 1971.3} |
| {"event": "train", "step": 10900, "phase": 3, "objective": "global", "lm_loss": 1.02992, "bpb": 1.48586, "chunk_loss": 0.030177, "F": 0.187, "G": 0.21427, "chunks": 191.48, "bytes_per_chunk": 5.348, "boundary_entropy": 0.14505, "router_noise": 0.0, "grad_norm": 0.399, "lr": 3.01e-05, "router_trainable": true, "sec_per_step": 0.166, "alpha": 0.03, "t": 1987.9} |
| {"event": "probe", "step": 11000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3262.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.187439, "cps": 25.484375, "mean_p": 0.186793, "boundary_entropy": 0.052473, "bpb": 1.630515}, "de": {"chunks": 3724.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.181367, "cps": 29.09375, "mean_p": 0.184427, "boundary_entropy": 0.052178, "bpb": 1.758255}, "vi": {"chunks": 4195.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.227556, "cps": 32.773438, "mean_p": 0.180535, "boundary_entropy": 0.058804, "bpb": 1.395474}, "tr": {"chunks": 3260.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.180081, "cps": 25.46875, "mean_p": 0.172576, "boundary_entropy": 0.086863, "bpb": 1.686936}, "fi": {"chunks": 3214.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.170857, "cps": 25.109375, "mean_p": 0.173366, "boundary_entropy": 0.077396, "bpb": 1.84473}, "ru": {"chunks": 7548.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.398311, "cps": 58.96875, "mean_p": 0.200637, "boundary_entropy": 0.095376, "bpb": 1.040529}, "el": {"chunks": 6861.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.334308, "cps": 53.601562, "mean_p": 0.191857, "boundary_entropy": 0.083662, "bpb": 1.039141}, "hi": {"chunks": 7066.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.41997, "cps": 55.203125, "mean_p": 0.27455, "boundary_entropy": 0.387414, "bpb": 0.906757}, "ta": {"chunks": 8248.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.441187, "cps": 64.4375, "mean_p": 0.304616, "boundary_entropy": 0.468604, "bpb": 0.787506}, "th": {"chunks": 7803.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.48511, "cps": 60.960938, "mean_p": 0.377293, "boundary_entropy": 0.550173, "bpb": 0.885507}, "zh": {"chunks": 2800.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.474094, "cps": 21.875, "mean_p": 0.205904, "boundary_entropy": 0.135382, "bpb": 2.508762}, "my": {"chunks": 8702.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.459257, "cps": 67.984375, "mean_p": 0.329114, "boundary_entropy": 0.524518, "bpb": 0.874102}}, "bpb": {"en": 1.63051, "de": 1.75825, "vi": 1.39547, "tr": 1.68694, "fi": 1.84473, "ru": 1.04053, "el": 1.03914, "hi": 0.90676, "ta": 0.78751, "th": 0.88551, "zh": 2.50876, "my": 0.8741}, "gini_cpc": 0.209581, "gini_cps": 0.220135, "bpic": 5.755305, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9676034955976864, "vi": 1.2140287922318742, "tr": 0.9607429629885774, "fi": 0.9115366868094873, "ru": 2.1250191135147025, "el": 1.7835559499827518, "hi": 2.2405710678049635, "ta": 2.353766330961583, "th": 2.588097928558945, "zh": 2.5293256730261944, "my": 2.4501680160630657}, "t": 2007.5} |
| {"event": "eval", "step": 11000, "phase": 3, "bpb": {"en": 1.6299, "de": 1.76366, "vi": 1.41047, "tr": 1.68037, "fi": 1.86128, "ru": 1.0392, "el": 1.03795, "hi": 0.91324, "ta": 0.79234, "th": 0.88395, "zh": 2.51482, "my": 0.88263}, "bytes_per_chunk": {"en": 5.089, "de": 5.583, "vi": 5.742, "tr": 5.932, "fi": 5.931, "ru": 4.668, "el": 5.471, "hi": 5.883, "ta": 5.909, "th": 5.914, "zh": 5.774, "my": 6.254}, "bpb_mean": 1.36748, "t": 2012.9} |
| {"event": "probe", "step": 11000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 3262.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.187439, "cps": 25.484375, "mean_p": 0.186793, "boundary_entropy": 0.052473, "bpb": 1.630515}, "de": {"chunks": 3724.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.181367, "cps": 29.09375, "mean_p": 0.184427, "boundary_entropy": 0.052178, "bpb": 1.758255}, "vi": {"chunks": 4195.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.227556, "cps": 32.773438, "mean_p": 0.180535, "boundary_entropy": 0.058804, "bpb": 1.395474}, "tr": {"chunks": 3260.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.180081, "cps": 25.46875, "mean_p": 0.172576, "boundary_entropy": 0.086863, "bpb": 1.686936}, "fi": {"chunks": 3214.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.170857, "cps": 25.109375, "mean_p": 0.173366, "boundary_entropy": 0.077396, "bpb": 1.84473}, "ru": {"chunks": 7548.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.398311, "cps": 58.96875, "mean_p": 0.200637, "boundary_entropy": 0.095376, "bpb": 1.040529}, "el": {"chunks": 6861.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.334308, "cps": 53.601562, "mean_p": 0.191857, "boundary_entropy": 0.083662, "bpb": 1.039141}, "hi": {"chunks": 7066.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.41997, "cps": 55.203125, "mean_p": 0.27455, "boundary_entropy": 0.387414, "bpb": 0.906757}, "ta": {"chunks": 8248.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.441187, "cps": 64.4375, "mean_p": 0.304616, "boundary_entropy": 0.468604, "bpb": 0.787506}, "th": {"chunks": 7803.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.48511, "cps": 60.960938, "mean_p": 0.377293, "boundary_entropy": 0.550173, "bpb": 0.885507}, "zh": {"chunks": 2800.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.474094, "cps": 21.875, "mean_p": 0.205904, "boundary_entropy": 0.135382, "bpb": 2.508762}, "my": {"chunks": 8702.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.459257, "cps": 67.984375, "mean_p": 0.329114, "boundary_entropy": 0.524518, "bpb": 0.874102}}, "bpb": {"en": 1.63051, "de": 1.75825, "vi": 1.39547, "tr": 1.68694, "fi": 1.84473, "ru": 1.04053, "el": 1.03914, "hi": 0.90676, "ta": 0.78751, "th": 0.88551, "zh": 2.50876, "my": 0.8741}, "gini_cpc": 0.209581, "gini_cps": 0.220135, "bpic": 5.755305, "worst_lang": "th", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9676034955976864, "vi": 1.2140287922318742, "tr": 0.9607429629885774, "fi": 0.9115366868094873, "ru": 2.1250191135147025, "el": 1.7835559499827518, "hi": 2.2405710678049635, "ta": 2.353766330961583, "th": 2.588097928558945, "zh": 2.5293256730261944, "my": 2.4501680160630657}, "t": 2016.0} |
| {"event": "reps_dumped", "dirs": ["/local/scratch/sas245/beetle-hnet/scratch/reps/spec_A_s0/th0.30", "/local/scratch/sas245/beetle-hnet/scratch/reps/spec_A_s0/th0.40", "/local/scratch/sas245/beetle-hnet/scratch/reps/spec_A_s0/th0.50", "/local/scratch/sas245/beetle-hnet/scratch/reps/spec_A_s0/th0.60", "/local/scratch/sas245/beetle-hnet/scratch/reps/spec_A_s0/th0.70"], "t": 2019.9} |
| {"event": "final", "run": "spec_A_s0", "condition": "A", "step": 11000, "params": 82470528, "per_lang": {"en": {"cpc": 0.18743894730793542, "cps": 25.484375, "bpb": 1.624982773953399, "bytes_per_chunk": 5.10015223957548}, "de": {"cpc": 0.18136658062630887, "cps": 29.09375, "bpb": 1.76161577754148, "bytes_per_chunk": 5.566783285464767}, "vi": {"cpc": 0.22755627881746676, "cps": 32.7734375, "bpb": 1.4094545136020074, "bytes_per_chunk": 5.748173161786875}, "tr": {"cpc": 0.18008064961608572, "cps": 25.46875, "bpb": 1.6799797300116097, "bytes_per_chunk": 5.93300742350172}, "fi": {"cpc": 0.17085747700813353, "cps": 25.109375, "bpb": 1.868181290755361, "bytes_per_chunk": 5.955483361636631}, "ru": {"cpc": 0.398311345646438, "cps": 58.96875, "bpb": 1.0388261127015617, "bytes_per_chunk": 4.674444211643136}, "el": {"cpc": 0.3343078497295717, "cps": 53.6015625, "bpb": 1.0367840659476264, "bytes_per_chunk": 5.490760377230051}, "hi": {"cpc": 0.41997028231797917, "cps": 55.203125, "bpb": 0.9135630702680455, "bytes_per_chunk": 5.897270636139185}, "ta": {"cpc": 0.44118748328430063, "cps": 64.4375, "bpb": 0.7907793082315361, "bytes_per_chunk": 5.881030634106945}, "th": {"cpc": 0.4851103512589369, "cps": 60.9609375, "bpb": 0.8860151257289858, "bytes_per_chunk": 5.926769989034716}, "zh": {"cpc": 0.47409414155096513, "cps": 21.875, "bpb": 2.5161213810896603, "bytes_per_chunk": 5.786491990000607}, "my": {"cpc": 0.45925691365843363, "cps": 67.984375, "bpb": 0.8805026195654585, "bytes_per_chunk": 6.253696182450365}}, "gini_cpc": 0.20958090744294489, "gini_cps": 0.2201348169698425, "bpic": 5.755304950287179, "seed": 0, "mean_boundary_entropy": 0.14483880996704102, "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": 2030.7} |
|
|
