| {"event": "start", "run": "jitter_s2", "condition": "A", "params": 22516992, "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": 1.5} |
| {"event": "probe_schedule", "n": 13, "steps": [0, 1, 2, 4, 8, 16, 32, 64, 128, 256, 2000, 4000, 6000], "t": 1.5} |
| {"event": "probe", "step": 0, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 10933.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.628225, "cps": 85.414062, "mean_p": 0.513249, "boundary_entropy": 0.686295, "bpb": 8.12598}, "de": {"chunks": 16287.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.793211, "cps": 127.242188, "mean_p": 0.524631, "boundary_entropy": 0.686391, "bpb": 8.12701}, "vi": {"chunks": 5099.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.276593, "cps": 39.835938, "mean_p": 0.481643, "boundary_entropy": 0.68728, "bpb": 7.869214}, "tr": {"chunks": 8187.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.452245, "cps": 63.960938, "mean_p": 0.498002, "boundary_entropy": 0.686826, "bpb": 8.097536}, "fi": {"chunks": 13540.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.719792, "cps": 105.78125, "mean_p": 0.517169, "boundary_entropy": 0.686661, "bpb": 8.065117}, "ru": {"chunks": 33477.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.766596, "cps": 261.539062, "mean_p": 0.530376, "boundary_entropy": 0.688543, "bpb": 8.240733}, "el": {"chunks": 20696.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.00843, "cps": 161.6875, "mean_p": 0.503648, "boundary_entropy": 0.690112, "bpb": 8.286272}, "hi": {"chunks": 875.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.052006, "cps": 6.835938, "mean_p": 0.461978, "boundary_entropy": 0.68743, "bpb": 8.022688}, "ta": {"chunks": 1261.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.067451, "cps": 9.851562, "mean_p": 0.477518, "boundary_entropy": 0.690053, "bpb": 8.091383}, "th": {"chunks": 3630.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.225676, "cps": 28.359375, "mean_p": 0.484845, "boundary_entropy": 0.690398, "bpb": 7.721507}, "zh": {"chunks": 11722.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.984761, "cps": 91.578125, "mean_p": 0.520316, "boundary_entropy": 0.68562, "bpb": 8.130734}, "my": {"chunks": 1628.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.085919, "cps": 12.71875, "mean_p": 0.474628, "boundary_entropy": 0.689685, "bpb": 7.8029}}, "bpb": {"en": 8.12598, "de": 8.12701, "vi": 7.86921, "tr": 8.09754, "fi": 8.06512, "ru": 8.24073, "el": 8.28627, "hi": 8.02269, "ta": 8.09138, "th": 7.72151, "zh": 8.13073, "my": 7.8029}, "gini_cpc": 0.489538, "gini_cps": 0.463551, "bpic": 3.013947, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.2626223171148012, "vi": 0.440277652395235, "tr": 0.7198781817505053, "fi": 1.1457544508645294, "ru": 2.8120438593717987, "el": 1.6052044042347315, "hi": 0.08278235025738971, "ta": 0.10736788276924242, "th": 0.3592281253258146, "zh": 3.159315851384177, "my": 0.13676526035979295}, "t": 3.3} |
| {"event": "train", "step": 0, "phase": 1, "objective": "global", "lm_loss": 5.64451, "bpb": 8.14331, "chunk_loss": 0.017745, "F": 0.00098, "G": 0.5091, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.69066, "router_noise": 0.0, "grad_norm": 4.2033, "lr": 1e-06, "router_trainable": true, "sec_per_step": 0.001, "alpha": 0.03, "t": 3.8} |
| {"event": "probe", "step": 1, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 10973.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.630523, "cps": 85.726562, "mean_p": 0.513415, "boundary_entropy": 0.686291, "bpb": 8.112407}, "de": {"chunks": 16325.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.795062, "cps": 127.539062, "mean_p": 0.524793, "boundary_entropy": 0.686378, "bpb": 8.114037}, "vi": {"chunks": 5118.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.277624, "cps": 39.984375, "mean_p": 0.481707, "boundary_entropy": 0.687285, "bpb": 7.860188}, "tr": {"chunks": 8241.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.455228, "cps": 64.382812, "mean_p": 0.49822, "boundary_entropy": 0.68683, "bpb": 8.088402}, "fi": {"chunks": 13565.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.721121, "cps": 105.976562, "mean_p": 0.517278, "boundary_entropy": 0.686656, "bpb": 8.053366}, "ru": {"chunks": 33499.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.767757, "cps": 261.710938, "mean_p": 0.530674, "boundary_entropy": 0.688508, "bpb": 8.214279}, "el": {"chunks": 20725.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.009843, "cps": 161.914062, "mean_p": 0.503684, "boundary_entropy": 0.690112, "bpb": 8.285335}, "hi": {"chunks": 878.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.052184, "cps": 6.859375, "mean_p": 0.462055, "boundary_entropy": 0.687442, "bpb": 8.01674}, "ta": {"chunks": 1287.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.068842, "cps": 10.054688, "mean_p": 0.477649, "boundary_entropy": 0.690066, "bpb": 8.087633}, "th": {"chunks": 3656.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.227293, "cps": 28.5625, "mean_p": 0.484903, "boundary_entropy": 0.690401, "bpb": 7.719771}, "zh": {"chunks": 11731.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.986285, "cps": 91.648438, "mean_p": 0.520388, "boundary_entropy": 0.685615, "bpb": 8.128736}, "my": {"chunks": 1618.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.085392, "cps": 12.640625, "mean_p": 0.474603, "boundary_entropy": 0.689682, "bpb": 7.801052}}, "bpb": {"en": 8.11241, "de": 8.11404, "vi": 7.86019, "tr": 8.0884, "fi": 8.05337, "ru": 8.21428, "el": 8.28534, "hi": 8.01674, "ta": 8.08763, "th": 7.71977, "zh": 8.12874, "my": 7.80105}, "gini_cpc": 0.488839, "gini_cps": 0.462872, "bpic": 3.007311, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.2609548133156332, "vi": 0.4403072946933804, "tr": 0.7219848819260126, "fi": 1.1436856070130392, "ru": 2.8036343320027304, "el": 1.6015940070465446, "hi": 0.0827633733970008, "ta": 0.1091821943169754, "th": 0.3604822315015682, "zh": 3.1502160013546776, "my": 0.13542968932840913}, "t": 5.1} |
| {"event": "probe", "step": 2, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 10974.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.630581, "cps": 85.734375, "mean_p": 0.513418, "boundary_entropy": 0.686297, "bpb": 8.074849}, "de": {"chunks": 16315.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.794575, "cps": 127.460938, "mean_p": 0.524791, "boundary_entropy": 0.686383, "bpb": 8.078929}, "vi": {"chunks": 5118.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.277624, "cps": 39.984375, "mean_p": 0.481666, "boundary_entropy": 0.687281, "bpb": 7.841203}, "tr": {"chunks": 8290.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.457935, "cps": 64.765625, "mean_p": 0.498428, "boundary_entropy": 0.686835, "bpb": 8.062411}, "fi": {"chunks": 13551.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.720376, "cps": 105.867188, "mean_p": 0.517218, "boundary_entropy": 0.686664, "bpb": 8.018631}, "ru": {"chunks": 33514.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.768549, "cps": 261.828125, "mean_p": 0.530885, "boundary_entropy": 0.688481, "bpb": 8.141092}, "el": {"chunks": 20782.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.01262, "cps": 162.359375, "mean_p": 0.50375, "boundary_entropy": 0.690111, "bpb": 8.284844}, "hi": {"chunks": 903.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.05367, "cps": 7.054688, "mean_p": 0.462219, "boundary_entropy": 0.687468, "bpb": 7.986524}, "ta": {"chunks": 1324.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.070821, "cps": 10.34375, "mean_p": 0.477855, "boundary_entropy": 0.690085, "bpb": 8.071872}, "th": {"chunks": 3686.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.229158, "cps": 28.796875, "mean_p": 0.484947, "boundary_entropy": 0.690403, "bpb": 7.703635}, "zh": {"chunks": 11732.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.986454, "cps": 91.65625, "mean_p": 0.52037, "boundary_entropy": 0.685615, "bpb": 8.122673}, "my": {"chunks": 1610.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.084969, "cps": 12.578125, "mean_p": 0.474585, "boundary_entropy": 0.68968, "bpb": 7.800115}}, "bpb": {"en": 8.07485, "de": 8.07893, "vi": 7.8412, "tr": 8.06241, "fi": 8.01863, "ru": 8.14109, "el": 8.28484, "hi": 7.98652, "ta": 8.07187, "th": 7.70364, "zh": 8.12267, "my": 7.80012}, "gini_cpc": 0.488024, "gini_cps": 0.462073, "bpic": 3.003005, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.2600675726178374, "vi": 0.44026717192185744, "tr": 0.7262115359751005, "fi": 1.1424011357081152, "ru": 2.804634134026552, "el": 1.6058525277498072, "hi": 0.08511220496589128, "ta": 0.11231084116104524, "th": 0.3634071183823363, "zh": 3.1501974528224856, "my": 0.1347477941124701}, "t": 6.7} |
| {"event": "probe", "step": 4, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 10958.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.629662, "cps": 85.609375, "mean_p": 0.513328, "boundary_entropy": 0.686323, "bpb": 7.950483}, "de": {"chunks": 16341.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.795841, "cps": 127.664062, "mean_p": 0.524718, "boundary_entropy": 0.686408, "bpb": 7.966392}, "vi": {"chunks": 5092.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.276214, "cps": 39.78125, "mean_p": 0.481544, "boundary_entropy": 0.687268, "bpb": 7.774413}, "tr": {"chunks": 8458.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.467215, "cps": 66.078125, "mean_p": 0.499042, "boundary_entropy": 0.686851, "bpb": 7.975067}, "fi": {"chunks": 13519.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.718675, "cps": 105.617188, "mean_p": 0.516944, "boundary_entropy": 0.686695, "bpb": 7.903791}, "ru": {"chunks": 33540.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.769921, "cps": 262.03125, "mean_p": 0.531365, "boundary_entropy": 0.688418, "bpb": 7.887699}, "el": {"chunks": 20880.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.017395, "cps": 163.125, "mean_p": 0.503844, "boundary_entropy": 0.690112, "bpb": 8.253629}, "hi": {"chunks": 958.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.056939, "cps": 7.484375, "mean_p": 0.462719, "boundary_entropy": 0.687546, "bpb": 7.88726}, "ta": {"chunks": 1398.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.074779, "cps": 10.921875, "mean_p": 0.478275, "boundary_entropy": 0.690123, "bpb": 7.984734}, "th": {"chunks": 3708.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.230525, "cps": 28.96875, "mean_p": 0.484792, "boundary_entropy": 0.690388, "bpb": 7.620259}, "zh": {"chunks": 11691.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.979512, "cps": 91.335938, "mean_p": 0.520229, "boundary_entropy": 0.685623, "bpb": 8.09962}, "my": {"chunks": 1582.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.083492, "cps": 12.359375, "mean_p": 0.474452, "boundary_entropy": 0.689665, "bpb": 7.794292}}, "bpb": {"en": 7.95048, "de": 7.96639, "vi": 7.77441, "tr": 7.97507, "fi": 7.90379, "ru": 7.8877, "el": 8.25363, "hi": 7.88726, "ta": 7.98473, "th": 7.62026, "zh": 8.09962, "my": 7.79429}, "gini_cpc": 0.486186, "gini_cps": 0.460751, "bpic": 2.995364, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.2639184309070375, "vi": 0.43867014390770237, "tr": 0.7420103335401932, "fi": 1.1413675175727382, "ru": 2.8109082363411075, "el": 1.615780911923772, "hi": 0.09042806967681818, "ta": 0.1187611860032478, "th": 0.3661099096939097, "zh": 3.1437720027835248, "my": 0.13259768052143592}, "t": 8.3} |
| {"event": "probe", "step": 8, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 10564.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.607022, "cps": 82.53125, "mean_p": 0.511303, "boundary_entropy": 0.686489, "bpb": 7.511635}, "de": {"chunks": 16085.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.783373, "cps": 125.664062, "mean_p": 0.522715, "boundary_entropy": 0.686663, "bpb": 7.567076}, "vi": {"chunks": 4846.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.26287, "cps": 37.859375, "mean_p": 0.480592, "boundary_entropy": 0.687177, "bpb": 7.546083}, "tr": {"chunks": 8897.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.491466, "cps": 69.507812, "mean_p": 0.500753, "boundary_entropy": 0.686901, "bpb": 7.649824}, "fi": {"chunks": 13006.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.691404, "cps": 101.609375, "mean_p": 0.514458, "boundary_entropy": 0.68689, "bpb": 7.498431}, "ru": {"chunks": 33618.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.774037, "cps": 262.640625, "mean_p": 0.53289, "boundary_entropy": 0.688209, "bpb": 7.323119}, "el": {"chunks": 20734.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 1.010281, "cps": 161.984375, "mean_p": 0.503587, "boundary_entropy": 0.69012, "bpb": 8.046873}, "hi": {"chunks": 1190.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.070728, "cps": 9.296875, "mean_p": 0.46461, "boundary_entropy": 0.687837, "bpb": 7.578698}, "ta": {"chunks": 1773.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.094838, "cps": 13.851562, "mean_p": 0.479794, "boundary_entropy": 0.690254, "bpb": 7.732736}, "th": {"chunks": 3813.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.237053, "cps": 29.789062, "mean_p": 0.484558, "boundary_entropy": 0.69036, "bpb": 7.38321}, "zh": {"chunks": 11553.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.956146, "cps": 90.257812, "mean_p": 0.519491, "boundary_entropy": 0.685665, "bpb": 8.031371}, "my": {"chunks": 1309.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.069084, "cps": 10.226562, "mean_p": 0.473199, "boundary_entropy": 0.689534, "bpb": 7.664766}}, "bpb": {"en": 7.51163, "de": 7.56708, "vi": 7.54608, "tr": 7.64982, "fi": 7.49843, "ru": 7.32312, "el": 8.04687, "hi": 7.5787, "ta": 7.73274, "th": 7.38321, "zh": 8.03137, "my": 7.66477}, "gini_cpc": 0.482797, "gini_cps": 0.457572, "bpic": 3.012694, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.290518948883884, "vi": 0.43304795844708327, "tr": 0.8096340541396518, "fi": 1.1390101492051576, "ru": 2.922526018069033, "el": 1.6643243864351969, "hi": 0.1165165498007475, "ta": 0.156235238157639, "th": 0.3905183697879156, "zh": 3.2225306624431505, "my": 0.11380779213834358}, "t": 10.0} |
| {"event": "probe", "step": 16, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 6431.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.369534, "cps": 50.242188, "mean_p": 0.495391, "boundary_entropy": 0.686738, "bpb": 6.909054}, "de": {"chunks": 11557.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.56285, "cps": 90.289062, "mean_p": 0.505708, "boundary_entropy": 0.687738, "bpb": 6.988178}, "vi": {"chunks": 3303.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.17917, "cps": 25.804688, "mean_p": 0.474137, "boundary_entropy": 0.686527, "bpb": 7.205538}, "tr": {"chunks": 8697.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.480418, "cps": 67.945312, "mean_p": 0.499746, "boundary_entropy": 0.687046, "bpb": 7.180256}, "fi": {"chunks": 8100.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.430599, "cps": 63.28125, "mean_p": 0.497915, "boundary_entropy": 0.687331, "bpb": 6.96291}, "ru": {"chunks": 32668.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 1.723905, "cps": 255.21875, "mean_p": 0.52748, "boundary_entropy": 0.688774, "bpb": 6.461576}, "el": {"chunks": 17688.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.861862, "cps": 138.1875, "mean_p": 0.499546, "boundary_entropy": 0.690143, "bpb": 7.466983}, "hi": {"chunks": 2156.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.128143, "cps": 16.84375, "mean_p": 0.469561, "boundary_entropy": 0.68855, "bpb": 6.649744}, "ta": {"chunks": 3273.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.175074, "cps": 25.570312, "mean_p": 0.483503, "boundary_entropy": 0.690536, "bpb": 6.839187}, "th": {"chunks": 2779.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.17277, "cps": 21.710938, "mean_p": 0.480699, "boundary_entropy": 0.690044, "bpb": 6.667273}, "zh": {"chunks": 10446.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 1.76871, "cps": 81.609375, "mean_p": 0.514169, "boundary_entropy": 0.685944, "bpb": 7.815019}, "my": {"chunks": 772.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.040743, "cps": 6.03125, "mean_p": 0.469418, "boundary_entropy": 0.689109, "bpb": 7.320966}}, "bpb": {"en": 6.90905, "de": 6.98818, "vi": 7.20554, "tr": 7.18026, "fi": 6.96291, "ru": 6.46158, "el": 7.46698, "hi": 6.64974, "ta": 6.83919, "th": 6.66727, "zh": 7.81502, "my": 7.32097}, "gini_cpc": 0.496787, "gini_cps": 0.468006, "bpic": 3.55781, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 1.5231347154694974, "vi": 0.48485406643142664, "tr": 1.3000633918148687, "fi": 1.1652490192049516, "ru": 4.665078361777489, "el": 2.3322950740787403, "hi": 0.3467682240300107, "ta": 0.4737684612956207, "th": 0.46753388484351743, "zh": 4.786325053211452, "my": 0.11025531513019436}, "t": 12.0} |
| {"event": "probe", "step": 32, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 546.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.031374, "cps": 4.265625, "mean_p": 0.454282, "boundary_entropy": 0.682264, "bpb": 6.444821}, "de": {"chunks": 480.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.023377, "cps": 3.75, "mean_p": 0.457648, "boundary_entropy": 0.684041, "bpb": 6.531547}, "vi": {"chunks": 739.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.040087, "cps": 5.773438, "mean_p": 0.453877, "boundary_entropy": 0.683603, "bpb": 6.918218}, "tr": {"chunks": 1827.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.100922, "cps": 14.273438, "mean_p": 0.470851, "boundary_entropy": 0.685377, "bpb": 6.768646}, "fi": {"chunks": 513.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.027271, "cps": 4.007812, "mean_p": 0.454461, "boundary_entropy": 0.683028, "bpb": 6.550628}, "ru": {"chunks": 10752.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.567388, "cps": 84.0, "mean_p": 0.494644, "boundary_entropy": 0.690088, "bpb": 5.97938}, "el": {"chunks": 15611.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.760659, "cps": 121.960938, "mean_p": 0.497568, "boundary_entropy": 0.690242, "bpb": 6.440687}, "hi": {"chunks": 355.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.0211, "cps": 2.773438, "mean_p": 0.457841, "boundary_entropy": 0.68691, "bpb": 6.323972}, "ta": {"chunks": 296.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.015833, "cps": 2.3125, "mean_p": 0.449685, "boundary_entropy": 0.685801, "bpb": 6.390968}, "th": {"chunks": 244.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.015169, "cps": 1.90625, "mean_p": 0.444814, "boundary_entropy": 0.684494, "bpb": 6.253931}, "zh": {"chunks": 4991.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.845073, "cps": 38.992188, "mean_p": 0.490494, "boundary_entropy": 0.685902, "bpb": 7.462509}, "my": {"chunks": 361.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.019052, "cps": 2.820312, "mean_p": 0.46235, "boundary_entropy": 0.688224, "bpb": 6.498228}}, "bpb": {"en": 6.44482, "de": 6.53155, "vi": 6.91822, "tr": 6.76865, "fi": 6.55063, "ru": 5.97938, "el": 6.44069, "hi": 6.32397, "ta": 6.39097, "th": 6.25393, "zh": 7.46251, "my": 6.49823}, "gini_cpc": 0.679691, "gini_cps": 0.689394, "bpic": 10.452976, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 0.7451098553226835, "vi": 1.2777114125364737, "tr": 3.216766026880373, "fi": 0.8692345079831125, "ru": 18.084708747716665, "el": 24.24495353109591, "hi": 0.6725193083291152, "ta": 0.5046586470496607, "th": 0.4835041862085692, "zh": 26.935534918856966, "my": 0.6072608782393457}, "t": 14.9} |
| {"event": "probe", "step": 64, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 207.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.011895, "cps": 1.617188, "mean_p": 0.357804, "boundary_entropy": 0.643436, "bpb": 5.94692}, "de": {"chunks": 159.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.007744, "cps": 1.242188, "mean_p": 0.361607, "boundary_entropy": 0.647364, "bpb": 6.083577}, "vi": {"chunks": 225.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.012205, "cps": 1.757812, "mean_p": 0.372776, "boundary_entropy": 0.653829, "bpb": 6.594595}, "tr": {"chunks": 162.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.008949, "cps": 1.265625, "mean_p": 0.366458, "boundary_entropy": 0.64941, "bpb": 6.436281}, "fi": {"chunks": 174.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.00925, "cps": 1.359375, "mean_p": 0.354408, "boundary_entropy": 0.642381, "bpb": 6.053196}, "ru": {"chunks": 132.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.006966, "cps": 1.03125, "mean_p": 0.410378, "boundary_entropy": 0.673523, "bpb": 5.788062}, "el": {"chunks": 10920.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.532086, "cps": 85.3125, "mean_p": 0.495321, "boundary_entropy": 0.6904, "bpb": 5.750818}, "hi": {"chunks": 160.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.00951, "cps": 1.25, "mean_p": 0.466272, "boundary_entropy": 0.688437, "bpb": 5.804437}, "ta": {"chunks": 192.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.01027, "cps": 1.5, "mean_p": 0.460835, "boundary_entropy": 0.687975, "bpb": 5.786206}, "th": {"chunks": 133.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.008269, "cps": 1.039062, "mean_p": 0.417771, "boundary_entropy": 0.676863, "bpb": 5.474193}, "zh": {"chunks": 149.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.025229, "cps": 1.164062, "mean_p": 0.410025, "boundary_entropy": 0.668755, "bpb": 7.191683}, "my": {"chunks": 180.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.0095, "cps": 1.40625, "mean_p": 0.424851, "boundary_entropy": 0.679574, "bpb": 5.903939}}, "bpb": {"en": 5.94692, "de": 6.08358, "vi": 6.5946, "tr": 6.43628, "fi": 6.0532, "ru": 5.78806, "el": 5.75082, "hi": 5.80444, "ta": 5.78621, "th": 5.47419, "zh": 7.19168, "my": 5.90394}, "gini_cpc": 0.764345, "gini_cps": 0.782583, "bpic": 29.999296, "worst_lang": "el", "cpc_parity_vs_en": {"en": 1.0, "de": 0.6510262377212505, "vi": 1.026108182686525, "tr": 0.7523470767516314, "fi": 0.7776624685371417, "ru": 0.5856234943214409, "el": 44.73377695014501, "hi": 0.7995003983892154, "ta": 0.863434770980383, "th": 0.6951593211787018, "zh": 2.1210289707838257, "my": 0.7986617837376435}, "t": 19.5} |
| {"event": "probe", "step": 128, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 2951.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.169568, "cps": 23.054688, "mean_p": 0.475111, "boundary_entropy": 0.684644, "bpb": 4.546814}, "de": {"chunks": 4472.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.217796, "cps": 34.9375, "mean_p": 0.479347, "boundary_entropy": 0.686126, "bpb": 4.773277}, "vi": {"chunks": 1209.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.065582, "cps": 9.445312, "mean_p": 0.46267, "boundary_entropy": 0.685101, "bpb": 5.290019}, "tr": {"chunks": 4751.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.262443, "cps": 37.117188, "mean_p": 0.480246, "boundary_entropy": 0.685829, "bpb": 5.212214}, "fi": {"chunks": 1847.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.098187, "cps": 14.429688, "mean_p": 0.469739, "boundary_entropy": 0.684909, "bpb": 4.813523}, "ru": {"chunks": 1792.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.094565, "cps": 14.0, "mean_p": 0.467962, "boundary_entropy": 0.687257, "bpb": 3.879847}, "el": {"chunks": 252.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.012279, "cps": 1.96875, "mean_p": 0.454998, "boundary_entropy": 0.686074, "bpb": 4.633605}, "hi": {"chunks": 233.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.013848, "cps": 1.820312, "mean_p": 0.478027, "boundary_entropy": 0.689579, "bpb": 3.718622}, "ta": {"chunks": 13734.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.734635, "cps": 107.296875, "mean_p": 0.468733, "boundary_entropy": 0.687419, "bpb": 3.773613}, "th": {"chunks": 2057.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.127883, "cps": 16.070312, "mean_p": 0.437202, "boundary_entropy": 0.679734, "bpb": 3.607527}, "zh": {"chunks": 335.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.056722, "cps": 2.617188, "mean_p": 0.47456, "boundary_entropy": 0.685611, "bpb": 6.43275}, "my": {"chunks": 1035.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.054623, "cps": 8.085938, "mean_p": 0.48045, "boundary_entropy": 0.690493, "bpb": 4.309861}}, "bpb": {"en": 4.54681, "de": 4.77328, "vi": 5.29002, "tr": 5.21221, "fi": 4.81352, "ru": 3.87985, "el": 4.6336, "hi": 3.71862, "ta": 3.77361, "th": 3.60753, "zh": 6.43275, "my": 4.30986}, "gini_cpc": 0.527577, "gini_cps": 0.5532, "bpic": 11.07018, "worst_lang": "ta", "cpc_parity_vs_en": {"en": 1.0, "de": 1.2844118343073394, "vi": 0.386756899165899, "tr": 1.5477092909725707, "fi": 0.5790418091925529, "ru": 0.5576782433216593, "el": 0.07241267754357077, "hi": 0.08166872186273774, "ta": 4.332379421067414, "th": 0.754168062816912, "zh": 0.334507820074652, "my": 0.3221305279883871}, "t": 28.3} |
| {"event": "train", "step": 200, "phase": 1, "objective": "global", "lm_loss": 2.5608, "bpb": 3.69445, "chunk_loss": 0.022629, "F": 0.00098, "G": 0.37261, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.65348, "router_noise": 0.0, "grad_norm": 1.8307, "lr": 0.000201, "router_trainable": true, "sec_per_step": 0.159, "alpha": 0.03, "t": 35.6} |
| {"event": "probe", "step": 256, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 138.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.00793, "cps": 1.078125, "mean_p": 0.325811, "boundary_entropy": 0.617501, "bpb": 3.688162}, "de": {"chunks": 491.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.023913, "cps": 3.835938, "mean_p": 0.323568, "boundary_entropy": 0.615002, "bpb": 3.901921}, "vi": {"chunks": 246.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.013344, "cps": 1.921875, "mean_p": 0.380783, "boundary_entropy": 0.654541, "bpb": 3.278447}, "tr": {"chunks": 404.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.022317, "cps": 3.15625, "mean_p": 0.339516, "boundary_entropy": 0.622591, "bpb": 3.931836}, "fi": {"chunks": 190.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.0101, "cps": 1.484375, "mean_p": 0.318453, "boundary_entropy": 0.610087, "bpb": 3.947919}, "ru": {"chunks": 129.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.006807, "cps": 1.007812, "mean_p": 0.324843, "boundary_entropy": 0.620993, "bpb": 2.670494}, "el": {"chunks": 131.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.006383, "cps": 1.023438, "mean_p": 0.358232, "boundary_entropy": 0.647862, "bpb": 2.78849}, "hi": {"chunks": 162.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.009629, "cps": 1.265625, "mean_p": 0.37343, "boundary_entropy": 0.653935, "bpb": 2.17733}, "ta": {"chunks": 133.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.007114, "cps": 1.039062, "mean_p": 0.334886, "boundary_entropy": 0.631485, "bpb": 2.044377}, "th": {"chunks": 145.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.009015, "cps": 1.132812, "mean_p": 0.368819, "boundary_entropy": 0.652689, "bpb": 2.125364}, "zh": {"chunks": 183.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.030985, "cps": 1.429688, "mean_p": 0.361058, "boundary_entropy": 0.639788, "bpb": 5.05289}, "my": {"chunks": 130.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.006861, "cps": 1.015625, "mean_p": 0.347609, "boundary_entropy": 0.641585, "bpb": 2.263279}}, "bpb": {"en": 3.68816, "de": 3.90192, "vi": 3.27845, "tr": 3.93184, "fi": 3.94792, "ru": 2.67049, "el": 2.78849, "hi": 2.17733, "ta": 2.04438, "th": 2.12536, "zh": 5.05289, "my": 2.26328}, "gini_cpc": 0.308209, "gini_cps": 0.258192, "bpic": 154.625705, "worst_lang": "zh", "cpc_parity_vs_en": {"en": 1.0, "de": 3.0156026671805085, "vi": 1.6828174196059007, "tr": 2.814335361182028, "fi": 1.2737574915694563, "ru": 0.8584708041757485, "el": 0.8049621951193676, "hi": 1.214241230053621, "ta": 0.8971626917218042, "th": 1.1368206944087793, "zh": 3.9075332381218804, "my": 0.8652169323824471}, "t": 42.2} |
| {"event": "train", "step": 400, "phase": 1, "objective": "global", "lm_loss": 2.24422, "bpb": 3.23773, "chunk_loss": 0.022311, "F": 0.00177, "G": 0.38255, "chunks": 1.81, "bytes_per_chunk": 564.966, "boundary_entropy": 0.65676, "router_noise": 0.0, "grad_norm": 1.2833, "lr": 0.0002998, "router_trainable": true, "sec_per_step": 0.107, "alpha": 0.03, "t": 57.1} |
| {"event": "train", "step": 600, "phase": 1, "objective": "global", "lm_loss": 1.93344, "bpb": 2.78937, "chunk_loss": 0.025269, "F": 0.04996, "G": 0.35909, "chunks": 51.16, "bytes_per_chunk": 20.017, "boundary_entropy": 0.64475, "router_noise": 0.0, "grad_norm": 1.1866, "lr": 0.0002982, "router_trainable": true, "sec_per_step": 0.095, "alpha": 0.03, "t": 76.2} |
| {"event": "train", "step": 800, "phase": 1, "objective": "global", "lm_loss": 1.87191, "bpb": 2.7006, "chunk_loss": 0.032523, "F": 0.22223, "G": 0.35836, "chunks": 227.56, "bytes_per_chunk": 4.5, "boundary_entropy": 0.6453, "router_noise": 0.0, "grad_norm": 0.8789, "lr": 0.0002949, "router_trainable": true, "sec_per_step": 0.098, "alpha": 0.03, "t": 95.8} |
| {"event": "eval", "step": 1000, "phase": 1, "bpb": {"en": 2.62173, "de": 2.95945, "vi": 2.32517, "tr": 2.95091, "fi": 3.18195, "ru": 1.91645, "el": 2.03022, "hi": 1.53319, "ta": 1.31682, "th": 1.58764, "zh": 3.65126, "my": 1.40758}, "bytes_per_chunk": {"en": 179.06, "de": 148.861, "vi": 166.124, "tr": 35.724, "fi": 213.125, "ru": 63.091, "el": 261.1, "hi": 164.87, "ta": 86.573, "th": 337.814, "zh": 201.649, "my": 47.81}, "bpb_mean": 2.2902, "t": 114.5} |
| {"event": "train", "step": 1000, "phase": 1, "objective": "global", "lm_loss": 1.57618, "bpb": 2.27395, "chunk_loss": 0.023139, "F": 0.0014, "G": 0.35886, "chunks": 1.44, "bytes_per_chunk": 712.348, "boundary_entropy": 0.64447, "router_noise": 0.0, "grad_norm": 0.7905, "lr": 0.0002901, "router_trainable": true, "sec_per_step": 0.094, "alpha": 0.03, "t": 114.6} |
| {"event": "train", "step": 1200, "phase": 1, "objective": "global", "lm_loss": 1.72909, "bpb": 2.49455, "chunk_loss": 0.023445, "F": 0.00098, "G": 0.34983, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.63892, "router_noise": 0.0, "grad_norm": 0.972, "lr": 0.0002837, "router_trainable": true, "sec_per_step": 0.096, "alpha": 0.03, "t": 133.8} |
| {"event": "train", "step": 1400, "phase": 1, "objective": "global", "lm_loss": 1.72227, "bpb": 2.48472, "chunk_loss": 0.022609, "F": 0.00421, "G": 0.37745, "chunks": 4.31, "bytes_per_chunk": 237.449, "boundary_entropy": 0.65458, "router_noise": 0.0, "grad_norm": 0.8507, "lr": 0.0002759, "router_trainable": true, "sec_per_step": 0.096, "alpha": 0.03, "t": 153.1} |
| {"event": "train", "step": 1600, "phase": 1, "objective": "global", "lm_loss": 1.81127, "bpb": 2.61311, "chunk_loss": 0.028345, "F": 0.12985, "G": 0.37895, "chunks": 132.97, "bytes_per_chunk": 7.701, "boundary_entropy": 0.65544, "router_noise": 0.0, "grad_norm": 0.9153, "lr": 0.0002668, "router_trainable": true, "sec_per_step": 0.092, "alpha": 0.03, "t": 171.6} |
| {"event": "train", "step": 1800, "phase": 1, "objective": "global", "lm_loss": 1.46985, "bpb": 2.12055, "chunk_loss": 0.023037, "F": 0.00125, "G": 0.36155, "chunks": 1.28, "bytes_per_chunk": 799.22, "boundary_entropy": 0.64533, "router_noise": 0.0, "grad_norm": 0.7498, "lr": 0.0002564, "router_trainable": true, "sec_per_step": 0.092, "alpha": 0.03, "t": 190.0} |
| {"event": "probe", "step": 2000, "phase": 1, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 712.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.040912, "cps": 5.5625, "mean_p": 0.404796, "boundary_entropy": 0.660596, "bpb": 2.315924}, "de": {"chunks": 888.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.043247, "cps": 6.9375, "mean_p": 0.413895, "boundary_entropy": 0.666974, "bpb": 2.609606}, "vi": {"chunks": 567.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.030757, "cps": 4.429688, "mean_p": 0.396135, "boundary_entropy": 0.659024, "bpb": 2.122418}, "tr": {"chunks": 1251.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.069105, "cps": 9.773438, "mean_p": 0.417649, "boundary_entropy": 0.667817, "bpb": 2.593749}, "fi": {"chunks": 1175.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.062463, "cps": 9.179688, "mean_p": 0.418486, "boundary_entropy": 0.66759, "bpb": 2.862535}, "ru": {"chunks": 1136.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.059947, "cps": 8.875, "mean_p": 0.394438, "boundary_entropy": 0.661856, "bpb": 1.704548}, "el": {"chunks": 1524.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.074258, "cps": 11.90625, "mean_p": 0.417701, "boundary_entropy": 0.672436, "bpb": 1.759558}, "hi": {"chunks": 4278.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.254264, "cps": 33.421875, "mean_p": 0.422668, "boundary_entropy": 0.670879, "bpb": 1.369975}, "ta": {"chunks": 3175.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.169832, "cps": 24.804688, "mean_p": 0.419546, "boundary_entropy": 0.673166, "bpb": 1.112813}, "th": {"chunks": 1965.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.122164, "cps": 15.351562, "mean_p": 0.401491, "boundary_entropy": 0.661667, "bpb": 1.342597}, "zh": {"chunks": 483.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.081781, "cps": 3.773438, "mean_p": 0.397701, "boundary_entropy": 0.658059, "bpb": 3.43383}, "my": {"chunks": 3061.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.161547, "cps": 23.914062, "mean_p": 0.43214, "boundary_entropy": 0.678789, "bpb": 1.260822}}, "bpb": {"en": 2.31592, "de": 2.60961, "vi": 2.12242, "tr": 2.59375, "fi": 2.86254, "ru": 1.70455, "el": 1.75956, "hi": 1.36997, "ta": 1.11281, "th": 1.3426, "zh": 3.43383, "my": 1.26082}, "gini_cpc": 0.343305, "gini_cps": 0.363942, "bpic": 18.984962, "worst_lang": "hi", "cpc_parity_vs_en": {"en": 1.0, "de": 1.0570722821087677, "vi": 0.7517683601356725, "tr": 1.68908258734197, "fi": 1.5267576674895575, "ru": 1.4652551065785182, "el": 1.8150485588380698, "hi": 6.214838305758218, "ta": 4.151092267655534, "th": 2.9859712098298017, "zh": 1.9989310527857787, "my": 3.9486085365506542}, "t": 208.9} |
| {"event": "eval", "step": 2000, "phase": 1, "bpb": {"en": 2.32214, "de": 2.61804, "vi": 2.12452, "tr": 2.60296, "fi": 2.86484, "ru": 1.68552, "el": 1.75203, "hi": 1.36677, "ta": 1.11994, "th": 1.36287, "zh": 3.43846, "my": 1.26863}, "bytes_per_chunk": {"en": 30.066, "de": 25.372, "vi": 47.064, "tr": 20.845, "fi": 17.901, "ru": 33.816, "el": 26.214, "hi": 9.738, "ta": 16.31, "th": 26.568, "zh": 45.158, "my": 17.82}, "bpb_mean": 2.04389, "t": 210.2} |
| {"event": "train", "step": 2000, "phase": 1, "objective": "global", "lm_loss": 1.44163, "bpb": 2.07983, "chunk_loss": 0.06545, "F": 0.85696, "G": 0.40346, "chunks": 877.53, "bytes_per_chunk": 1.167, "boundary_entropy": 0.6668, "router_noise": 0.0, "grad_norm": 1.1627, "lr": 0.000245, "router_trainable": true, "sec_per_step": 0.101, "alpha": 0.03, "t": 210.3} |
| {"event": "train", "step": 2200, "phase": 1, "objective": "global", "lm_loss": 1.31007, "bpb": 1.89003, "chunk_loss": 0.048572, "F": 0.60114, "G": 0.36269, "chunks": 615.56, "bytes_per_chunk": 1.664, "boundary_entropy": 0.64343, "router_noise": 0.0, "grad_norm": 0.8399, "lr": 0.0002325, "router_trainable": true, "sec_per_step": 0.093, "alpha": 0.03, "t": 228.9} |
| {"event": "phase_transition", "step": 2400, "phase": 2, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0, "t": 247.2} |
| {"event": "train", "step": 2400, "phase": 2, "objective": "global", "lm_loss": 1.34105, "bpb": 1.93472, "chunk_loss": 0.028884, "F": 0.1416, "G": 0.37433, "chunks": 145.0, "bytes_per_chunk": 7.062, "boundary_entropy": 0.65166, "router_noise": 0.0, "grad_norm": 0.8131, "lr": 0.0002192, "router_trainable": true, "sec_per_step": 0.092, "alpha": 0.03, "t": 247.3} |
| {"event": "train", "step": 2600, "phase": 2, "objective": "global", "lm_loss": 1.52519, "bpb": 2.20038, "chunk_loss": 0.028042, "F": 0.12323, "G": 0.37579, "chunks": 126.19, "bytes_per_chunk": 8.115, "boundary_entropy": 0.65264, "router_noise": 0.0, "grad_norm": 0.9188, "lr": 0.0002053, "router_trainable": true, "sec_per_step": 0.091, "alpha": 0.03, "t": 265.5} |
| {"event": "train", "step": 2800, "phase": 2, "objective": "global", "lm_loss": 1.43453, "bpb": 2.06959, "chunk_loss": 0.023349, "F": 0.00125, "G": 0.35281, "chunks": 1.28, "bytes_per_chunk": 799.22, "boundary_entropy": 0.63995, "router_noise": 0.0, "grad_norm": 0.8882, "lr": 0.0001909, "router_trainable": true, "sec_per_step": 0.089, "alpha": 0.03, "t": 283.4} |
| {"event": "eval", "step": 3000, "phase": 2, "bpb": {"en": 2.20921, "de": 2.4469, "vi": 2.04661, "tr": 2.43706, "fi": 2.6653, "ru": 1.55049, "el": 1.63054, "hi": 1.29389, "ta": 1.06619, "th": 1.2542, "zh": 3.3289, "my": 1.20051}, "bytes_per_chunk": {"en": 96.094, "de": 108.954, "vi": 68.445, "tr": 76.988, "fi": 56.424, "ru": 57.425, "el": 40.523, "hi": 68.072, "ta": 91.659, "th": 74.178, "zh": 103.943, "my": 69.222}, "bpb_mean": 1.92748, "t": 302.2} |
| {"event": "train", "step": 3000, "phase": 2, "objective": "global", "lm_loss": 1.53952, "bpb": 2.22105, "chunk_loss": 0.051232, "F": 0.65289, "G": 0.36814, "chunks": 668.56, "bytes_per_chunk": 1.532, "boundary_entropy": 0.64834, "router_noise": 0.0, "grad_norm": 0.8578, "lr": 0.0001761, "router_trainable": true, "sec_per_step": 0.095, "alpha": 0.03, "t": 302.3} |
| {"event": "train", "step": 3200, "phase": 2, "objective": "global", "lm_loss": 1.42466, "bpb": 2.05535, "chunk_loss": 0.0234, "F": 0.00098, "G": 0.35109, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.63906, "router_noise": 0.0, "grad_norm": 0.8193, "lr": 0.0001613, "router_trainable": true, "sec_per_step": 0.091, "alpha": 0.03, "t": 320.6} |
| {"event": "train", "step": 3400, "phase": 2, "objective": "global", "lm_loss": 1.33683, "bpb": 1.92864, "chunk_loss": 0.028723, "F": 0.13754, "G": 0.3715, "chunks": 140.84, "bytes_per_chunk": 7.27, "boundary_entropy": 0.65142, "router_noise": 0.0, "grad_norm": 0.9023, "lr": 0.0001465, "router_trainable": true, "sec_per_step": 0.091, "alpha": 0.03, "t": 338.7} |
| {"event": "train", "step": 3600, "phase": 2, "objective": "global", "lm_loss": 1.3442, "bpb": 1.93927, "chunk_loss": 0.022435, "F": 0.00098, "G": 0.37803, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.65299, "router_noise": 0.0, "grad_norm": 0.8975, "lr": 0.0001319, "router_trainable": true, "sec_per_step": 0.093, "alpha": 0.03, "t": 357.3} |
| {"event": "train", "step": 3800, "phase": 2, "objective": "global", "lm_loss": 1.40932, "bpb": 2.03321, "chunk_loss": 0.022659, "F": 0.00104, "G": 0.37186, "chunks": 1.06, "bytes_per_chunk": 963.765, "boundary_entropy": 0.65173, "router_noise": 0.0, "grad_norm": 0.8772, "lr": 0.0001177, "router_trainable": true, "sec_per_step": 0.094, "alpha": 0.03, "t": 376.2} |
| {"event": "probe", "step": 4000, "phase": 2, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 294.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.016894, "cps": 2.296875, "mean_p": 0.377103, "boundary_entropy": 0.646966, "bpb": 2.132903}, "de": {"chunks": 318.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.015487, "cps": 2.484375, "mean_p": 0.382933, "boundary_entropy": 0.652848, "bpb": 2.346793}, "vi": {"chunks": 469.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.025441, "cps": 3.664062, "mean_p": 0.3835, "boundary_entropy": 0.653029, "bpb": 1.989742}, "tr": {"chunks": 408.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.022538, "cps": 3.1875, "mean_p": 0.380939, "boundary_entropy": 0.651878, "bpb": 2.318921}, "fi": {"chunks": 483.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.025676, "cps": 3.773438, "mean_p": 0.388374, "boundary_entropy": 0.654449, "bpb": 2.534682}, "ru": {"chunks": 785.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.041425, "cps": 6.132812, "mean_p": 0.39287, "boundary_entropy": 0.662042, "bpb": 1.504671}, "el": {"chunks": 665.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.032403, "cps": 5.195312, "mean_p": 0.381358, "boundary_entropy": 0.656416, "bpb": 1.562564}, "hi": {"chunks": 1053.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.062585, "cps": 8.226562, "mean_p": 0.356573, "boundary_entropy": 0.627373, "bpb": 1.260632}, "ta": {"chunks": 1553.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.08307, "cps": 12.132812, "mean_p": 0.391833, "boundary_entropy": 0.662245, "bpb": 1.027909}, "th": {"chunks": 827.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.051414, "cps": 6.460938, "mean_p": 0.372158, "boundary_entropy": 0.646485, "bpb": 1.176101}, "zh": {"chunks": 267.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.045208, "cps": 2.085938, "mean_p": 0.385231, "boundary_entropy": 0.653231, "bpb": 3.214515}, "my": {"chunks": 257.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.013563, "cps": 2.007812, "mean_p": 0.384606, "boundary_entropy": 0.660241, "bpb": 1.159897}}, "bpb": {"en": 2.1329, "de": 2.34679, "vi": 1.98974, "tr": 2.31892, "fi": 2.53468, "ru": 1.50467, "el": 1.56256, "hi": 1.26063, "ta": 1.02791, "th": 1.1761, "zh": 3.21452, "my": 1.1599}, "gini_cpc": 0.305661, "gini_cps": 0.318257, "bpic": 52.009893, "worst_lang": "ta", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9167512327095159, "vi": 1.5059359138285093, "tr": 1.3340939093418949, "fi": 1.519889578286261, "ru": 2.4520946637230088, "el": 1.9180396906606525, "hi": 3.704674773326864, "ta": 4.917255514133976, "th": 3.043415401597381, "zh": 2.6760523715073568, "my": 0.8028724155906821}, "t": 394.7} |
| {"event": "eval", "step": 4000, "phase": 2, "bpb": {"en": 2.13991, "de": 2.35358, "vi": 1.98988, "tr": 2.32957, "fi": 2.53829, "ru": 1.48173, "el": 1.55782, "hi": 1.25631, "ta": 1.03414, "th": 1.19615, "zh": 3.21293, "my": 1.16754}, "bytes_per_chunk": {"en": 82.826, "de": 82.383, "vi": 59.511, "tr": 87.41, "fi": 43.807, "ru": 50.862, "el": 57.513, "hi": 42.604, "ta": 40.605, "th": 61.536, "zh": 96.911, "my": 92.11}, "bpb_mean": 1.85482, "t": 396.0} |
| {"event": "train", "step": 4000, "phase": 2, "objective": "global", "lm_loss": 1.3414, "bpb": 1.93524, "chunk_loss": 0.052572, "F": 0.66241, "G": 0.37645, "chunks": 678.31, "bytes_per_chunk": 1.51, "boundary_entropy": 0.65202, "router_noise": 0.0, "grad_norm": 0.8761, "lr": 0.000104, "router_trainable": true, "sec_per_step": 0.099, "alpha": 0.03, "t": 396.1} |
| {"event": "phase_transition", "step": 4200, "phase": 3, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0, "t": 415.2} |
| {"event": "train", "step": 4200, "phase": 3, "objective": "global", "lm_loss": 1.32287, "bpb": 1.90849, "chunk_loss": 0.022215, "F": 0.0011, "G": 0.38436, "chunks": 1.12, "bytes_per_chunk": 910.222, "boundary_entropy": 0.65843, "router_noise": 0.0, "grad_norm": 0.8063, "lr": 9.12e-05, "router_trainable": true, "sec_per_step": 0.096, "alpha": 0.03, "t": 415.3} |
| {"event": "train", "step": 4400, "phase": 3, "objective": "global", "lm_loss": 1.48374, "bpb": 2.14059, "chunk_loss": 0.022965, "F": 0.00143, "G": 0.36377, "chunks": 1.47, "bytes_per_chunk": 697.191, "boundary_entropy": 0.64664, "router_noise": 0.0, "grad_norm": 0.951, "lr": 7.92e-05, "router_trainable": true, "sec_per_step": 0.089, "alpha": 0.03, "t": 433.1} |
| {"event": "train", "step": 4600, "phase": 3, "objective": "global", "lm_loss": 1.44325, "bpb": 2.08217, "chunk_loss": 0.022639, "F": 0.00098, "G": 0.37234, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.65114, "router_noise": 0.0, "grad_norm": 0.8701, "lr": 6.82e-05, "router_trainable": true, "sec_per_step": 0.091, "alpha": 0.03, "t": 451.4} |
| {"event": "train", "step": 4800, "phase": 3, "objective": "global", "lm_loss": 1.38421, "bpb": 1.99699, "chunk_loss": 0.02217, "F": 0.00098, "G": 0.38544, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.65813, "router_noise": 0.0, "grad_norm": 0.8583, "lr": 5.85e-05, "router_trainable": true, "sec_per_step": 0.094, "alpha": 0.03, "t": 470.1} |
| {"event": "eval", "step": 5000, "phase": 3, "bpb": {"en": 2.0983, "de": 2.30449, "vi": 1.96153, "tr": 2.27631, "fi": 2.46698, "ru": 1.43429, "el": 1.50627, "hi": 1.22834, "ta": 1.01362, "th": 1.16708, "zh": 3.15443, "my": 1.14021}, "bytes_per_chunk": {"en": 64.346, "de": 71.507, "vi": 53.184, "tr": 51.08, "fi": 65.585, "ru": 51.695, "el": 42.876, "hi": 41.849, "ta": 45.189, "th": 54.659, "zh": 65.882, "my": 40.436}, "bpb_mean": 1.81265, "t": 489.7} |
| {"event": "train", "step": 5000, "phase": 3, "objective": "global", "lm_loss": 1.40145, "bpb": 2.02187, "chunk_loss": 0.025775, "F": 0.07776, "G": 0.38607, "chunks": 79.62, "bytes_per_chunk": 12.86, "boundary_entropy": 0.65884, "router_noise": 0.0, "grad_norm": 0.8475, "lr": 5e-05, "router_trainable": true, "sec_per_step": 0.098, "alpha": 0.03, "t": 489.7} |
| {"event": "train", "step": 5200, "phase": 3, "objective": "global", "lm_loss": 1.28923, "bpb": 1.85996, "chunk_loss": 0.022336, "F": 0.00351, "G": 0.38413, "chunks": 3.59, "bytes_per_chunk": 284.939, "boundary_entropy": 0.65839, "router_noise": 0.0, "grad_norm": 0.8719, "lr": 4.29e-05, "router_trainable": true, "sec_per_step": 0.094, "alpha": 0.03, "t": 508.5} |
| {"event": "train", "step": 5400, "phase": 3, "objective": "global", "lm_loss": 1.3869, "bpb": 2.00088, "chunk_loss": 0.02251, "F": 0.00098, "G": 0.37596, "chunks": 1.0, "bytes_per_chunk": 1024.0, "boundary_entropy": 0.6538, "router_noise": 0.0, "grad_norm": 0.954, "lr": 3.73e-05, "router_trainable": true, "sec_per_step": 0.093, "alpha": 0.03, "t": 527.1} |
| {"event": "train", "step": 5600, "phase": 3, "objective": "global", "lm_loss": 1.36892, "bpb": 1.97493, "chunk_loss": 0.023072, "F": 0.01898, "G": 0.38386, "chunks": 19.44, "bytes_per_chunk": 52.682, "boundary_entropy": 0.65772, "router_noise": 0.0, "grad_norm": 0.888, "lr": 3.33e-05, "router_trainable": true, "sec_per_step": 0.095, "alpha": 0.03, "t": 546.1} |
| {"event": "train", "step": 5800, "phase": 3, "objective": "global", "lm_loss": 1.32724, "bpb": 1.9148, "chunk_loss": 0.022554, "F": 0.00406, "G": 0.37868, "chunks": 4.16, "bytes_per_chunk": 246.376, "boundary_entropy": 0.655, "router_noise": 0.0, "grad_norm": 0.9174, "lr": 3.08e-05, "router_trainable": true, "sec_per_step": 0.095, "alpha": 0.03, "t": 565.1} |
| {"event": "probe", "step": 6000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 235.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.013503, "cps": 1.835938, "mean_p": 0.374696, "boundary_entropy": 0.646152, "bpb": 2.072183}, "de": {"chunks": 254.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.01237, "cps": 1.984375, "mean_p": 0.378796, "boundary_entropy": 0.650996, "bpb": 2.273952}, "vi": {"chunks": 353.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.019148, "cps": 2.757812, "mean_p": 0.375519, "boundary_entropy": 0.649338, "bpb": 1.946296}, "tr": {"chunks": 410.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.022648, "cps": 3.203125, "mean_p": 0.381634, "boundary_entropy": 0.652266, "bpb": 2.238207}, "fi": {"chunks": 245.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.013024, "cps": 1.914062, "mean_p": 0.372556, "boundary_entropy": 0.646645, "bpb": 2.439194}, "ru": {"chunks": 591.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.031187, "cps": 4.617188, "mean_p": 0.376535, "boundary_entropy": 0.654067, "bpb": 1.435821}, "el": {"chunks": 566.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.027579, "cps": 4.421875, "mean_p": 0.37797, "boundary_entropy": 0.654882, "bpb": 1.489949}, "hi": {"chunks": 836.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.049688, "cps": 6.53125, "mean_p": 0.361977, "boundary_entropy": 0.633895, "bpb": 1.218163}, "ta": {"chunks": 1009.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.053972, "cps": 7.882812, "mean_p": 0.380425, "boundary_entropy": 0.656419, "bpb": 0.991907}, "th": {"chunks": 638.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.039664, "cps": 4.984375, "mean_p": 0.363322, "boundary_entropy": 0.642576, "bpb": 1.127957}, "zh": {"chunks": 237.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.040129, "cps": 1.851562, "mean_p": 0.379361, "boundary_entropy": 0.650523, "bpb": 3.125459}, "my": {"chunks": 816.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.043065, "cps": 6.375, "mean_p": 0.36954, "boundary_entropy": 0.651657, "bpb": 1.119749}}, "bpb": {"en": 2.07218, "de": 2.27395, "vi": 1.9463, "tr": 2.23821, "fi": 2.43919, "ru": 1.43582, "el": 1.48995, "hi": 1.21816, "ta": 0.99191, "th": 1.12796, "zh": 3.12546, "my": 1.11975}, "gini_cpc": 0.262784, "gini_cps": 0.278568, "bpic": 62.000162, "worst_lang": "ta", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9160887870174738, "vi": 1.4180378442947907, "tr": 1.677218187651982, "fi": 0.9645182707718549, "ru": 2.3095880536686693, "el": 2.042357873522286, "hi": 3.6796580569694286, "ta": 3.9968877786187003, "th": 2.9373511729574933, "zh": 2.9717424040463722, "my": 3.1892094376995943}, "t": 584.2} |
| {"event": "eval", "step": 6000, "phase": 3, "bpb": {"en": 2.08045, "de": 2.28221, "vi": 1.94655, "tr": 2.24572, "fi": 2.44227, "ru": 1.4125, "el": 1.48345, "hi": 1.21199, "ta": 0.99986, "th": 1.14677, "zh": 3.12322, "my": 1.12666}, "bytes_per_chunk": {"en": 123.478, "de": 132.262, "vi": 89.134, "tr": 87.294, "fi": 135.826, "ru": 73.802, "el": 67.079, "hi": 54.613, "ta": 64.22, "th": 81.436, "zh": 119.7, "my": 51.031}, "bpb_mean": 1.7918, "t": 585.5} |
| {"event": "probe", "step": 6000, "phase": 3, "objective": "global", "router_trainable": true, "per_lang": {"en": {"chunks": 235.0, "chars": 17403.0, "bytes": 17418.0, "cpc": 0.013503, "cps": 1.835938, "mean_p": 0.374696, "boundary_entropy": 0.646152, "bpb": 2.072183}, "de": {"chunks": 254.0, "chars": 20533.0, "bytes": 20862.0, "cpc": 0.01237, "cps": 1.984375, "mean_p": 0.378796, "boundary_entropy": 0.650996, "bpb": 2.273952}, "vi": {"chunks": 353.0, "chars": 18435.0, "bytes": 24274.0, "cpc": 0.019148, "cps": 2.757812, "mean_p": 0.375519, "boundary_entropy": 0.649338, "bpb": 1.946296}, "tr": {"chunks": 410.0, "chars": 18103.0, "bytes": 19742.0, "cpc": 0.022648, "cps": 3.203125, "mean_p": 0.381634, "boundary_entropy": 0.652266, "bpb": 2.238207}, "fi": {"chunks": 245.0, "chars": 18811.0, "bytes": 19572.0, "cpc": 0.013024, "cps": 1.914062, "mean_p": 0.372556, "boundary_entropy": 0.646645, "bpb": 2.439194}, "ru": {"chunks": 591.0, "chars": 18950.0, "bytes": 34560.0, "cpc": 0.031187, "cps": 4.617188, "mean_p": 0.376535, "boundary_entropy": 0.654067, "bpb": 1.435821}, "el": {"chunks": 566.0, "chars": 20523.0, "bytes": 37264.0, "cpc": 0.027579, "cps": 4.421875, "mean_p": 0.37797, "boundary_entropy": 0.654882, "bpb": 1.489949}, "hi": {"chunks": 836.0, "chars": 16825.0, "bytes": 43173.0, "cpc": 0.049688, "cps": 6.53125, "mean_p": 0.361977, "boundary_entropy": 0.633895, "bpb": 1.218163}, "ta": {"chunks": 1009.0, "chars": 18695.0, "bytes": 50915.0, "cpc": 0.053972, "cps": 7.882812, "mean_p": 0.380425, "boundary_entropy": 0.656419, "bpb": 0.991907}, "th": {"chunks": 638.0, "chars": 16085.0, "bytes": 46085.0, "cpc": 0.039664, "cps": 4.984375, "mean_p": 0.363322, "boundary_entropy": 0.642576, "bpb": 1.127957}, "zh": {"chunks": 237.0, "chars": 5906.0, "bytes": 16173.0, "cpc": 0.040129, "cps": 1.851562, "mean_p": 0.379361, "boundary_entropy": 0.650523, "bpb": 3.125459}, "my": {"chunks": 816.0, "chars": 18948.0, "bytes": 53743.0, "cpc": 0.043065, "cps": 6.375, "mean_p": 0.36954, "boundary_entropy": 0.651657, "bpb": 1.119749}}, "bpb": {"en": 2.07218, "de": 2.27395, "vi": 1.9463, "tr": 2.23821, "fi": 2.43919, "ru": 1.43582, "el": 1.48995, "hi": 1.21816, "ta": 0.99191, "th": 1.12796, "zh": 3.12546, "my": 1.11975}, "gini_cpc": 0.262784, "gini_cps": 0.278568, "bpic": 62.000162, "worst_lang": "ta", "cpc_parity_vs_en": {"en": 1.0, "de": 0.9160887870174738, "vi": 1.4180378442947907, "tr": 1.677218187651982, "fi": 0.9645182707718549, "ru": 2.3095880536686693, "el": 2.042357873522286, "hi": 3.6796580569694286, "ta": 3.9968877786187003, "th": 2.9373511729574933, "zh": 2.9717424040463722, "my": 3.1892094376995943}, "t": 586.4} |
| {"event": "reps_dumped", "dirs": ["/local/scratch/sas245/beetle-hnet/scratch/reps/jitter_s2/th0.30", "/local/scratch/sas245/beetle-hnet/scratch/reps/jitter_s2/th0.40", "/local/scratch/sas245/beetle-hnet/scratch/reps/jitter_s2/th0.50", "/local/scratch/sas245/beetle-hnet/scratch/reps/jitter_s2/th0.60", "/local/scratch/sas245/beetle-hnet/scratch/reps/jitter_s2/th0.70"], "t": 590.1} |
| {"event": "final", "run": "jitter_s2", "condition": "A", "step": 6000, "params": 22516992, "per_lang": {"en": {"cpc": 0.013503418950755616, "cps": 1.8359375, "bpb": 2.091184529801444, "bytes_per_chunk": 119.15636363636364}, "de": {"cpc": 0.01237033068718648, "cps": 1.984375, "bpb": 2.2734026958836058, "bytes_per_chunk": 135.02137522534125}, "vi": {"cpc": 0.01914835909953892, "cps": 2.7578125, "bpb": 1.9494414809506864, "bytes_per_chunk": 86.68783068783068}, "tr": {"cpc": 0.022648179859691763, "cps": 3.203125, "bpb": 2.240100760041376, "bytes_per_chunk": 89.34696659850034}, "fi": {"cpc": 0.013024294295890703, "cps": 1.9140625, "bpb": 2.4580483003424547, "bytes_per_chunk": 132.8656867714141}, "ru": {"cpc": 0.031187335092348285, "cps": 4.6171875, "bpb": 1.4135231609180474, "bytes_per_chunk": 73.35777249195466}, "el": {"cpc": 0.02757881401354578, "cps": 4.421875, "bpb": 1.4792314436970544, "bytes_per_chunk": 67.52807831014941}, "hi": {"cpc": 0.04968796433878157, "cps": 6.53125, "bpb": 1.2194370256690006, "bytes_per_chunk": 54.50545794781162}, "ta": {"cpc": 0.053971650173843276, "cps": 7.8828125, "bpb": 0.9995220038420514, "bytes_per_chunk": 63.92197025115826}, "th": {"cpc": 0.039664283493938454, "cps": 4.984375, "bpb": 1.1435730867396772, "bytes_per_chunk": 81.6393646838991}, "zh": {"cpc": 0.040128682695563835, "cps": 1.8515625, "bpb": 3.1307407679749546, "bytes_per_chunk": 120.83152800184374}, "my": {"cpc": 0.04306523115896137, "cps": 6.375, "bpb": 1.1354728113731958, "bytes_per_chunk": 50.768664665440106}}, "gini_cpc": 0.262783775060067, "gini_cps": 0.27856758212170174, "bpic": 62.00016155088853, "seed": 2, "mean_boundary_entropy": 0.6519113779067993, "dechunk": "ema", "router_noise": 0.0, "transitions": [{"step": 0, "phase": 1, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0}, {"step": 2400, "phase": 2, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0}, {"step": 4200, "phase": 3, "objective": "global", "router_trainable": true, "encoder_trainable": true, "frozen_router_params": 0}], "t": 592.5} |
|
|
